{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0000662220330163,
  "eval_steps": 1322,
  "global_step": 13214,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 7.568232344732983e-05,
      "grad_norm": 19.5,
      "learning_rate": 4.0000000000000003e-07,
      "loss": 1.0007,
      "step": 1
    },
    {
      "epoch": 7.568232344732983e-05,
      "eval_loss": 1.1901112794876099,
      "eval_runtime": 82.5515,
      "eval_samples_per_second": 58.897,
      "eval_steps_per_second": 58.897,
      "step": 1
    },
    {
      "epoch": 0.00015136464689465966,
      "grad_norm": 20.375,
      "learning_rate": 8.000000000000001e-07,
      "loss": 1.0341,
      "step": 2
    },
    {
      "epoch": 0.0002270469703419895,
      "grad_norm": 23.625,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 1.2038,
      "step": 3
    },
    {
      "epoch": 0.0003027292937893193,
      "grad_norm": 25.5,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 1.2451,
      "step": 4
    },
    {
      "epoch": 0.00037841161723664915,
      "grad_norm": 24.25,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 1.0426,
      "step": 5
    },
    {
      "epoch": 0.000454093940683979,
      "grad_norm": 28.5,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 1.0197,
      "step": 6
    },
    {
      "epoch": 0.0005297762641313089,
      "grad_norm": 36.25,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 0.843,
      "step": 7
    },
    {
      "epoch": 0.0006054585875786386,
      "grad_norm": 37.75,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 0.7076,
      "step": 8
    },
    {
      "epoch": 0.0006811409110259685,
      "grad_norm": 36.25,
      "learning_rate": 3.6000000000000003e-06,
      "loss": 1.3896,
      "step": 9
    },
    {
      "epoch": 0.0007568232344732983,
      "grad_norm": 24.25,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.5926,
      "step": 10
    },
    {
      "epoch": 0.0008325055579206282,
      "grad_norm": 19.125,
      "learning_rate": 4.4e-06,
      "loss": 0.5021,
      "step": 11
    },
    {
      "epoch": 0.000908187881367958,
      "grad_norm": 6.40625,
      "learning_rate": 4.800000000000001e-06,
      "loss": 0.4098,
      "step": 12
    },
    {
      "epoch": 0.0009838702048152877,
      "grad_norm": 4.78125,
      "learning_rate": 5.2e-06,
      "loss": 0.4639,
      "step": 13
    },
    {
      "epoch": 0.0010595525282626177,
      "grad_norm": 4.6875,
      "learning_rate": 5.600000000000001e-06,
      "loss": 0.4493,
      "step": 14
    },
    {
      "epoch": 0.0011352348517099475,
      "grad_norm": 2.953125,
      "learning_rate": 6e-06,
      "loss": 0.4178,
      "step": 15
    },
    {
      "epoch": 0.0012109171751572773,
      "grad_norm": 2.75,
      "learning_rate": 6.4000000000000006e-06,
      "loss": 0.3991,
      "step": 16
    },
    {
      "epoch": 0.001286599498604607,
      "grad_norm": 2.015625,
      "learning_rate": 6.800000000000001e-06,
      "loss": 0.4151,
      "step": 17
    },
    {
      "epoch": 0.001362281822051937,
      "grad_norm": 1.796875,
      "learning_rate": 7.2000000000000005e-06,
      "loss": 0.4092,
      "step": 18
    },
    {
      "epoch": 0.0014379641454992668,
      "grad_norm": 1.6953125,
      "learning_rate": 7.600000000000001e-06,
      "loss": 0.3905,
      "step": 19
    },
    {
      "epoch": 0.0015136464689465966,
      "grad_norm": 1.796875,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.4334,
      "step": 20
    },
    {
      "epoch": 0.0015893287923939264,
      "grad_norm": 1.640625,
      "learning_rate": 8.400000000000001e-06,
      "loss": 0.3904,
      "step": 21
    },
    {
      "epoch": 0.0016650111158412564,
      "grad_norm": 1.5546875,
      "learning_rate": 8.8e-06,
      "loss": 0.3629,
      "step": 22
    },
    {
      "epoch": 0.0017406934392885862,
      "grad_norm": 1.5078125,
      "learning_rate": 9.200000000000002e-06,
      "loss": 0.3723,
      "step": 23
    },
    {
      "epoch": 0.001816375762735916,
      "grad_norm": 1.546875,
      "learning_rate": 9.600000000000001e-06,
      "loss": 0.3533,
      "step": 24
    },
    {
      "epoch": 0.0018920580861832457,
      "grad_norm": 1.5390625,
      "learning_rate": 1e-05,
      "loss": 0.3792,
      "step": 25
    },
    {
      "epoch": 0.0019677404096305755,
      "grad_norm": 1.75,
      "learning_rate": 1.04e-05,
      "loss": 0.3982,
      "step": 26
    },
    {
      "epoch": 0.0020434227330779057,
      "grad_norm": 1.5234375,
      "learning_rate": 1.0800000000000002e-05,
      "loss": 0.4199,
      "step": 27
    },
    {
      "epoch": 0.0021191050565252355,
      "grad_norm": 1.328125,
      "learning_rate": 1.1200000000000001e-05,
      "loss": 0.3335,
      "step": 28
    },
    {
      "epoch": 0.0021947873799725653,
      "grad_norm": 1.4765625,
      "learning_rate": 1.16e-05,
      "loss": 0.3999,
      "step": 29
    },
    {
      "epoch": 0.002270469703419895,
      "grad_norm": 1.5703125,
      "learning_rate": 1.2e-05,
      "loss": 0.3985,
      "step": 30
    },
    {
      "epoch": 0.002346152026867225,
      "grad_norm": 1.578125,
      "learning_rate": 1.2400000000000002e-05,
      "loss": 0.4779,
      "step": 31
    },
    {
      "epoch": 0.0024218343503145546,
      "grad_norm": 1.265625,
      "learning_rate": 1.2800000000000001e-05,
      "loss": 0.3573,
      "step": 32
    },
    {
      "epoch": 0.0024975166737618844,
      "grad_norm": 1.4140625,
      "learning_rate": 1.3200000000000002e-05,
      "loss": 0.3693,
      "step": 33
    },
    {
      "epoch": 0.002573198997209214,
      "grad_norm": 1.25,
      "learning_rate": 1.3600000000000002e-05,
      "loss": 0.3716,
      "step": 34
    },
    {
      "epoch": 0.0026488813206565444,
      "grad_norm": 1.4140625,
      "learning_rate": 1.4e-05,
      "loss": 0.3918,
      "step": 35
    },
    {
      "epoch": 0.002724563644103874,
      "grad_norm": 1.375,
      "learning_rate": 1.4400000000000001e-05,
      "loss": 0.4098,
      "step": 36
    },
    {
      "epoch": 0.002800245967551204,
      "grad_norm": 1.359375,
      "learning_rate": 1.48e-05,
      "loss": 0.3875,
      "step": 37
    },
    {
      "epoch": 0.0028759282909985337,
      "grad_norm": 1.4140625,
      "learning_rate": 1.5200000000000002e-05,
      "loss": 0.391,
      "step": 38
    },
    {
      "epoch": 0.0029516106144458635,
      "grad_norm": 1.2109375,
      "learning_rate": 1.5600000000000003e-05,
      "loss": 0.3678,
      "step": 39
    },
    {
      "epoch": 0.0030272929378931932,
      "grad_norm": 1.21875,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.3541,
      "step": 40
    },
    {
      "epoch": 0.003102975261340523,
      "grad_norm": 1.203125,
      "learning_rate": 1.64e-05,
      "loss": 0.3233,
      "step": 41
    },
    {
      "epoch": 0.003178657584787853,
      "grad_norm": 1.234375,
      "learning_rate": 1.6800000000000002e-05,
      "loss": 0.3657,
      "step": 42
    },
    {
      "epoch": 0.003254339908235183,
      "grad_norm": 1.453125,
      "learning_rate": 1.72e-05,
      "loss": 0.3954,
      "step": 43
    },
    {
      "epoch": 0.0033300222316825128,
      "grad_norm": 1.7734375,
      "learning_rate": 1.76e-05,
      "loss": 0.3669,
      "step": 44
    },
    {
      "epoch": 0.0034057045551298426,
      "grad_norm": 1.359375,
      "learning_rate": 1.8e-05,
      "loss": 0.3581,
      "step": 45
    },
    {
      "epoch": 0.0034813868785771723,
      "grad_norm": 1.2890625,
      "learning_rate": 1.8400000000000003e-05,
      "loss": 0.4164,
      "step": 46
    },
    {
      "epoch": 0.003557069202024502,
      "grad_norm": 1.296875,
      "learning_rate": 1.88e-05,
      "loss": 0.3835,
      "step": 47
    },
    {
      "epoch": 0.003632751525471832,
      "grad_norm": 1.140625,
      "learning_rate": 1.9200000000000003e-05,
      "loss": 0.3144,
      "step": 48
    },
    {
      "epoch": 0.0037084338489191617,
      "grad_norm": 1.296875,
      "learning_rate": 1.9600000000000002e-05,
      "loss": 0.4181,
      "step": 49
    },
    {
      "epoch": 0.0037841161723664914,
      "grad_norm": 1.125,
      "learning_rate": 2e-05,
      "loss": 0.3555,
      "step": 50
    },
    {
      "epoch": 0.0038597984958138216,
      "grad_norm": 1.296875,
      "learning_rate": 1.9999999929066423e-05,
      "loss": 0.4083,
      "step": 51
    },
    {
      "epoch": 0.003935480819261151,
      "grad_norm": 1.2578125,
      "learning_rate": 1.9999999716265682e-05,
      "loss": 0.3918,
      "step": 52
    },
    {
      "epoch": 0.004011163142708481,
      "grad_norm": 1.1796875,
      "learning_rate": 1.9999999361597785e-05,
      "loss": 0.3555,
      "step": 53
    },
    {
      "epoch": 0.004086845466155811,
      "grad_norm": 1.296875,
      "learning_rate": 1.9999998865062738e-05,
      "loss": 0.3503,
      "step": 54
    },
    {
      "epoch": 0.004162527789603141,
      "grad_norm": 1.3046875,
      "learning_rate": 1.9999998226660544e-05,
      "loss": 0.3529,
      "step": 55
    },
    {
      "epoch": 0.004238210113050471,
      "grad_norm": 1.21875,
      "learning_rate": 1.999999744639122e-05,
      "loss": 0.3569,
      "step": 56
    },
    {
      "epoch": 0.0043138924364978,
      "grad_norm": 1.25,
      "learning_rate": 1.9999996524254763e-05,
      "loss": 0.449,
      "step": 57
    },
    {
      "epoch": 0.0043895747599451305,
      "grad_norm": 1.21875,
      "learning_rate": 1.9999995460251203e-05,
      "loss": 0.397,
      "step": 58
    },
    {
      "epoch": 0.00446525708339246,
      "grad_norm": 20.75,
      "learning_rate": 1.999999425438054e-05,
      "loss": 0.9439,
      "step": 59
    },
    {
      "epoch": 0.00454093940683979,
      "grad_norm": 1.3515625,
      "learning_rate": 1.9999992906642803e-05,
      "loss": 0.39,
      "step": 60
    },
    {
      "epoch": 0.004616621730287119,
      "grad_norm": 1.25,
      "learning_rate": 1.9999991417038003e-05,
      "loss": 0.3676,
      "step": 61
    },
    {
      "epoch": 0.00469230405373445,
      "grad_norm": 1.140625,
      "learning_rate": 1.9999989785566167e-05,
      "loss": 0.3095,
      "step": 62
    },
    {
      "epoch": 0.00476798637718178,
      "grad_norm": 1.1328125,
      "learning_rate": 1.9999988012227312e-05,
      "loss": 0.3643,
      "step": 63
    },
    {
      "epoch": 0.004843668700629109,
      "grad_norm": 7.96875,
      "learning_rate": 1.9999986097021468e-05,
      "loss": 1.0323,
      "step": 64
    },
    {
      "epoch": 0.004919351024076439,
      "grad_norm": 1.296875,
      "learning_rate": 1.9999984039948664e-05,
      "loss": 0.3486,
      "step": 65
    },
    {
      "epoch": 0.004995033347523769,
      "grad_norm": 1.34375,
      "learning_rate": 1.9999981841008923e-05,
      "loss": 0.3353,
      "step": 66
    },
    {
      "epoch": 0.005070715670971099,
      "grad_norm": 1.3515625,
      "learning_rate": 1.9999979500202277e-05,
      "loss": 0.4026,
      "step": 67
    },
    {
      "epoch": 0.005146397994418428,
      "grad_norm": 1.2734375,
      "learning_rate": 1.9999977017528763e-05,
      "loss": 0.4169,
      "step": 68
    },
    {
      "epoch": 0.0052220803178657585,
      "grad_norm": 1.171875,
      "learning_rate": 1.9999974392988414e-05,
      "loss": 0.3457,
      "step": 69
    },
    {
      "epoch": 0.005297762641313089,
      "grad_norm": 43.0,
      "learning_rate": 1.9999971626581268e-05,
      "loss": 0.9594,
      "step": 70
    },
    {
      "epoch": 0.005373444964760418,
      "grad_norm": 1.3046875,
      "learning_rate": 1.9999968718307362e-05,
      "loss": 0.4245,
      "step": 71
    },
    {
      "epoch": 0.005449127288207748,
      "grad_norm": 1.140625,
      "learning_rate": 1.9999965668166744e-05,
      "loss": 0.3522,
      "step": 72
    },
    {
      "epoch": 0.005524809611655078,
      "grad_norm": 1.125,
      "learning_rate": 1.999996247615945e-05,
      "loss": 0.3849,
      "step": 73
    },
    {
      "epoch": 0.005600491935102408,
      "grad_norm": 1.1328125,
      "learning_rate": 1.999995914228553e-05,
      "loss": 0.3531,
      "step": 74
    },
    {
      "epoch": 0.005676174258549737,
      "grad_norm": 1.421875,
      "learning_rate": 1.9999955666545024e-05,
      "loss": 0.4589,
      "step": 75
    },
    {
      "epoch": 0.005751856581997067,
      "grad_norm": 1.15625,
      "learning_rate": 1.9999952048937986e-05,
      "loss": 0.3464,
      "step": 76
    },
    {
      "epoch": 0.005827538905444398,
      "grad_norm": 1.140625,
      "learning_rate": 1.9999948289464473e-05,
      "loss": 0.3485,
      "step": 77
    },
    {
      "epoch": 0.005903221228891727,
      "grad_norm": 1.21875,
      "learning_rate": 1.999994438812453e-05,
      "loss": 0.3532,
      "step": 78
    },
    {
      "epoch": 0.005978903552339057,
      "grad_norm": 1.140625,
      "learning_rate": 1.9999940344918217e-05,
      "loss": 0.3616,
      "step": 79
    },
    {
      "epoch": 0.0060545858757863865,
      "grad_norm": 1.1875,
      "learning_rate": 1.9999936159845586e-05,
      "loss": 0.3613,
      "step": 80
    },
    {
      "epoch": 0.006130268199233717,
      "grad_norm": 1.09375,
      "learning_rate": 1.9999931832906704e-05,
      "loss": 0.3286,
      "step": 81
    },
    {
      "epoch": 0.006205950522681046,
      "grad_norm": 1.1484375,
      "learning_rate": 1.9999927364101624e-05,
      "loss": 0.4035,
      "step": 82
    },
    {
      "epoch": 0.006281632846128376,
      "grad_norm": 1.109375,
      "learning_rate": 1.999992275343042e-05,
      "loss": 0.3796,
      "step": 83
    },
    {
      "epoch": 0.006357315169575706,
      "grad_norm": 1.1953125,
      "learning_rate": 1.999991800089315e-05,
      "loss": 0.3755,
      "step": 84
    },
    {
      "epoch": 0.006432997493023036,
      "grad_norm": 1.25,
      "learning_rate": 1.999991310648988e-05,
      "loss": 0.4195,
      "step": 85
    },
    {
      "epoch": 0.006508679816470366,
      "grad_norm": 1.234375,
      "learning_rate": 1.9999908070220685e-05,
      "loss": 0.4224,
      "step": 86
    },
    {
      "epoch": 0.006584362139917695,
      "grad_norm": 1.203125,
      "learning_rate": 1.999990289208563e-05,
      "loss": 0.3226,
      "step": 87
    },
    {
      "epoch": 0.0066600444633650255,
      "grad_norm": 1.1015625,
      "learning_rate": 1.9999897572084795e-05,
      "loss": 0.3591,
      "step": 88
    },
    {
      "epoch": 0.006735726786812355,
      "grad_norm": 1.25,
      "learning_rate": 1.9999892110218247e-05,
      "loss": 0.37,
      "step": 89
    },
    {
      "epoch": 0.006811409110259685,
      "grad_norm": 1.40625,
      "learning_rate": 1.9999886506486078e-05,
      "loss": 0.4073,
      "step": 90
    },
    {
      "epoch": 0.0068870914337070144,
      "grad_norm": 1.171875,
      "learning_rate": 1.999988076088835e-05,
      "loss": 0.3899,
      "step": 91
    },
    {
      "epoch": 0.006962773757154345,
      "grad_norm": 1.109375,
      "learning_rate": 1.999987487342516e-05,
      "loss": 0.3489,
      "step": 92
    },
    {
      "epoch": 0.007038456080601675,
      "grad_norm": 1.2578125,
      "learning_rate": 1.999986884409658e-05,
      "loss": 0.37,
      "step": 93
    },
    {
      "epoch": 0.007114138404049004,
      "grad_norm": 1.15625,
      "learning_rate": 1.99998626729027e-05,
      "loss": 0.3446,
      "step": 94
    },
    {
      "epoch": 0.007189820727496334,
      "grad_norm": 1.2890625,
      "learning_rate": 1.999985635984361e-05,
      "loss": 0.3692,
      "step": 95
    },
    {
      "epoch": 0.007265503050943664,
      "grad_norm": 1.171875,
      "learning_rate": 1.9999849904919398e-05,
      "loss": 0.4184,
      "step": 96
    },
    {
      "epoch": 0.007341185374390994,
      "grad_norm": 1.1953125,
      "learning_rate": 1.9999843308130155e-05,
      "loss": 0.4461,
      "step": 97
    },
    {
      "epoch": 0.007416867697838323,
      "grad_norm": 1.1796875,
      "learning_rate": 1.999983656947597e-05,
      "loss": 0.4281,
      "step": 98
    },
    {
      "epoch": 0.0074925500212856535,
      "grad_norm": 1.0859375,
      "learning_rate": 1.9999829688956945e-05,
      "loss": 0.3454,
      "step": 99
    },
    {
      "epoch": 0.007568232344732983,
      "grad_norm": 1.1640625,
      "learning_rate": 1.9999822666573176e-05,
      "loss": 0.314,
      "step": 100
    },
    {
      "epoch": 0.007643914668180313,
      "grad_norm": 1.21875,
      "learning_rate": 1.9999815502324763e-05,
      "loss": 0.3612,
      "step": 101
    },
    {
      "epoch": 0.007719596991627643,
      "grad_norm": 1.296875,
      "learning_rate": 1.9999808196211804e-05,
      "loss": 0.3528,
      "step": 102
    },
    {
      "epoch": 0.007795279315074973,
      "grad_norm": 1.203125,
      "learning_rate": 1.999980074823441e-05,
      "loss": 0.3571,
      "step": 103
    },
    {
      "epoch": 0.007870961638522302,
      "grad_norm": 1.2734375,
      "learning_rate": 1.999979315839268e-05,
      "loss": 0.4314,
      "step": 104
    },
    {
      "epoch": 0.007946643961969632,
      "grad_norm": 1.15625,
      "learning_rate": 1.999978542668672e-05,
      "loss": 0.3485,
      "step": 105
    },
    {
      "epoch": 0.008022326285416962,
      "grad_norm": 1.140625,
      "learning_rate": 1.999977755311665e-05,
      "loss": 0.3732,
      "step": 106
    },
    {
      "epoch": 0.008098008608864293,
      "grad_norm": 1.125,
      "learning_rate": 1.9999769537682572e-05,
      "loss": 0.3563,
      "step": 107
    },
    {
      "epoch": 0.008173690932311623,
      "grad_norm": 1.171875,
      "learning_rate": 1.99997613803846e-05,
      "loss": 0.4039,
      "step": 108
    },
    {
      "epoch": 0.008249373255758951,
      "grad_norm": 1.1640625,
      "learning_rate": 1.9999753081222856e-05,
      "loss": 0.4306,
      "step": 109
    },
    {
      "epoch": 0.008325055579206282,
      "grad_norm": 1.0703125,
      "learning_rate": 1.9999744640197457e-05,
      "loss": 0.3506,
      "step": 110
    },
    {
      "epoch": 0.008400737902653612,
      "grad_norm": 1.1015625,
      "learning_rate": 1.9999736057308517e-05,
      "loss": 0.3868,
      "step": 111
    },
    {
      "epoch": 0.008476420226100942,
      "grad_norm": 1.1796875,
      "learning_rate": 1.999972733255616e-05,
      "loss": 0.3879,
      "step": 112
    },
    {
      "epoch": 0.00855210254954827,
      "grad_norm": 1.171875,
      "learning_rate": 1.9999718465940512e-05,
      "loss": 0.392,
      "step": 113
    },
    {
      "epoch": 0.0086277848729956,
      "grad_norm": 1.1640625,
      "learning_rate": 1.9999709457461697e-05,
      "loss": 0.4244,
      "step": 114
    },
    {
      "epoch": 0.00870346719644293,
      "grad_norm": 1.15625,
      "learning_rate": 1.9999700307119846e-05,
      "loss": 0.4273,
      "step": 115
    },
    {
      "epoch": 0.008779149519890261,
      "grad_norm": 1.1953125,
      "learning_rate": 1.999969101491508e-05,
      "loss": 0.4407,
      "step": 116
    },
    {
      "epoch": 0.008854831843337591,
      "grad_norm": 1.1484375,
      "learning_rate": 1.9999681580847544e-05,
      "loss": 0.3989,
      "step": 117
    },
    {
      "epoch": 0.00893051416678492,
      "grad_norm": 7.34375,
      "learning_rate": 1.999967200491736e-05,
      "loss": 0.9033,
      "step": 118
    },
    {
      "epoch": 0.00900619649023225,
      "grad_norm": 1.125,
      "learning_rate": 1.9999662287124667e-05,
      "loss": 0.3593,
      "step": 119
    },
    {
      "epoch": 0.00908187881367958,
      "grad_norm": 1.0390625,
      "learning_rate": 1.999965242746961e-05,
      "loss": 0.3422,
      "step": 120
    },
    {
      "epoch": 0.00915756113712691,
      "grad_norm": 1.078125,
      "learning_rate": 1.999964242595232e-05,
      "loss": 0.4013,
      "step": 121
    },
    {
      "epoch": 0.009233243460574239,
      "grad_norm": 1.09375,
      "learning_rate": 1.9999632282572944e-05,
      "loss": 0.358,
      "step": 122
    },
    {
      "epoch": 0.009308925784021569,
      "grad_norm": 1.171875,
      "learning_rate": 1.9999621997331624e-05,
      "loss": 0.3844,
      "step": 123
    },
    {
      "epoch": 0.0093846081074689,
      "grad_norm": 1.15625,
      "learning_rate": 1.9999611570228503e-05,
      "loss": 0.3748,
      "step": 124
    },
    {
      "epoch": 0.00946029043091623,
      "grad_norm": 1.0703125,
      "learning_rate": 1.9999601001263733e-05,
      "loss": 0.4099,
      "step": 125
    },
    {
      "epoch": 0.00953597275436356,
      "grad_norm": 1.2109375,
      "learning_rate": 1.9999590290437464e-05,
      "loss": 0.4304,
      "step": 126
    },
    {
      "epoch": 0.009611655077810888,
      "grad_norm": 1.0078125,
      "learning_rate": 1.999957943774985e-05,
      "loss": 0.3123,
      "step": 127
    },
    {
      "epoch": 0.009687337401258218,
      "grad_norm": 1.1328125,
      "learning_rate": 1.9999568443201035e-05,
      "loss": 0.3617,
      "step": 128
    },
    {
      "epoch": 0.009763019724705549,
      "grad_norm": 1.1875,
      "learning_rate": 1.9999557306791183e-05,
      "loss": 0.3849,
      "step": 129
    },
    {
      "epoch": 0.009838702048152879,
      "grad_norm": 1.046875,
      "learning_rate": 1.9999546028520456e-05,
      "loss": 0.3182,
      "step": 130
    },
    {
      "epoch": 0.009914384371600209,
      "grad_norm": 1.125,
      "learning_rate": 1.9999534608389004e-05,
      "loss": 0.3908,
      "step": 131
    },
    {
      "epoch": 0.009990066695047537,
      "grad_norm": 1.09375,
      "learning_rate": 1.9999523046396996e-05,
      "loss": 0.3573,
      "step": 132
    },
    {
      "epoch": 0.010065749018494868,
      "grad_norm": 1.0390625,
      "learning_rate": 1.9999511342544593e-05,
      "loss": 0.3619,
      "step": 133
    },
    {
      "epoch": 0.010141431341942198,
      "grad_norm": 1.0,
      "learning_rate": 1.9999499496831964e-05,
      "loss": 0.3366,
      "step": 134
    },
    {
      "epoch": 0.010217113665389528,
      "grad_norm": 1.2109375,
      "learning_rate": 1.9999487509259273e-05,
      "loss": 0.3631,
      "step": 135
    },
    {
      "epoch": 0.010292795988836857,
      "grad_norm": 1.0625,
      "learning_rate": 1.999947537982669e-05,
      "loss": 0.3707,
      "step": 136
    },
    {
      "epoch": 0.010368478312284187,
      "grad_norm": 1.0859375,
      "learning_rate": 1.9999463108534393e-05,
      "loss": 0.3874,
      "step": 137
    },
    {
      "epoch": 0.010444160635731517,
      "grad_norm": 1.1328125,
      "learning_rate": 1.9999450695382547e-05,
      "loss": 0.3806,
      "step": 138
    },
    {
      "epoch": 0.010519842959178847,
      "grad_norm": 1.0859375,
      "learning_rate": 1.9999438140371338e-05,
      "loss": 0.3772,
      "step": 139
    },
    {
      "epoch": 0.010595525282626177,
      "grad_norm": 1.1171875,
      "learning_rate": 1.9999425443500937e-05,
      "loss": 0.3693,
      "step": 140
    },
    {
      "epoch": 0.010671207606073506,
      "grad_norm": 1.109375,
      "learning_rate": 1.9999412604771526e-05,
      "loss": 0.3593,
      "step": 141
    },
    {
      "epoch": 0.010746889929520836,
      "grad_norm": 1.453125,
      "learning_rate": 1.9999399624183293e-05,
      "loss": 0.4218,
      "step": 142
    },
    {
      "epoch": 0.010822572252968166,
      "grad_norm": 10.5,
      "learning_rate": 1.999938650173641e-05,
      "loss": 1.0997,
      "step": 143
    },
    {
      "epoch": 0.010898254576415497,
      "grad_norm": 1.15625,
      "learning_rate": 1.9999373237431073e-05,
      "loss": 0.3755,
      "step": 144
    },
    {
      "epoch": 0.010973936899862825,
      "grad_norm": 1.1875,
      "learning_rate": 1.9999359831267467e-05,
      "loss": 0.3892,
      "step": 145
    },
    {
      "epoch": 0.011049619223310155,
      "grad_norm": 1.0703125,
      "learning_rate": 1.999934628324578e-05,
      "loss": 0.3264,
      "step": 146
    },
    {
      "epoch": 0.011125301546757485,
      "grad_norm": 1.0859375,
      "learning_rate": 1.999933259336621e-05,
      "loss": 0.3895,
      "step": 147
    },
    {
      "epoch": 0.011200983870204816,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9999318761628946e-05,
      "loss": 0.3766,
      "step": 148
    },
    {
      "epoch": 0.011276666193652146,
      "grad_norm": 1.0859375,
      "learning_rate": 1.999930478803419e-05,
      "loss": 0.3469,
      "step": 149
    },
    {
      "epoch": 0.011352348517099474,
      "grad_norm": 1.109375,
      "learning_rate": 1.9999290672582127e-05,
      "loss": 0.3996,
      "step": 150
    },
    {
      "epoch": 0.011428030840546805,
      "grad_norm": 5.125,
      "learning_rate": 1.9999276415272974e-05,
      "loss": 0.9393,
      "step": 151
    },
    {
      "epoch": 0.011503713163994135,
      "grad_norm": 3.90625,
      "learning_rate": 1.9999262016106924e-05,
      "loss": 0.8319,
      "step": 152
    },
    {
      "epoch": 0.011579395487441465,
      "grad_norm": 1.0703125,
      "learning_rate": 1.9999247475084183e-05,
      "loss": 0.3226,
      "step": 153
    },
    {
      "epoch": 0.011655077810888795,
      "grad_norm": 1.25,
      "learning_rate": 1.9999232792204956e-05,
      "loss": 0.3981,
      "step": 154
    },
    {
      "epoch": 0.011730760134336124,
      "grad_norm": 1.265625,
      "learning_rate": 1.999921796746945e-05,
      "loss": 0.4434,
      "step": 155
    },
    {
      "epoch": 0.011806442457783454,
      "grad_norm": 1.046875,
      "learning_rate": 1.9999203000877883e-05,
      "loss": 0.3647,
      "step": 156
    },
    {
      "epoch": 0.011882124781230784,
      "grad_norm": 1.1171875,
      "learning_rate": 1.9999187892430462e-05,
      "loss": 0.4349,
      "step": 157
    },
    {
      "epoch": 0.011957807104678114,
      "grad_norm": 1.1328125,
      "learning_rate": 1.99991726421274e-05,
      "loss": 0.3511,
      "step": 158
    },
    {
      "epoch": 0.012033489428125443,
      "grad_norm": 1.15625,
      "learning_rate": 1.9999157249968915e-05,
      "loss": 0.4033,
      "step": 159
    },
    {
      "epoch": 0.012109171751572773,
      "grad_norm": 1.0390625,
      "learning_rate": 1.9999141715955224e-05,
      "loss": 0.3764,
      "step": 160
    },
    {
      "epoch": 0.012184854075020103,
      "grad_norm": 1.09375,
      "learning_rate": 1.999912604008655e-05,
      "loss": 0.3644,
      "step": 161
    },
    {
      "epoch": 0.012260536398467433,
      "grad_norm": 1.0859375,
      "learning_rate": 1.9999110222363117e-05,
      "loss": 0.3792,
      "step": 162
    },
    {
      "epoch": 0.012336218721914764,
      "grad_norm": 1.078125,
      "learning_rate": 1.999909426278514e-05,
      "loss": 0.3376,
      "step": 163
    },
    {
      "epoch": 0.012411901045362092,
      "grad_norm": 1.0625,
      "learning_rate": 1.9999078161352857e-05,
      "loss": 0.3844,
      "step": 164
    },
    {
      "epoch": 0.012487583368809422,
      "grad_norm": 1.0703125,
      "learning_rate": 1.999906191806649e-05,
      "loss": 0.3534,
      "step": 165
    },
    {
      "epoch": 0.012563265692256752,
      "grad_norm": 1.09375,
      "learning_rate": 1.999904553292627e-05,
      "loss": 0.36,
      "step": 166
    },
    {
      "epoch": 0.012638948015704083,
      "grad_norm": 1.046875,
      "learning_rate": 1.9999029005932428e-05,
      "loss": 0.3668,
      "step": 167
    },
    {
      "epoch": 0.012714630339151411,
      "grad_norm": 1.078125,
      "learning_rate": 1.9999012337085204e-05,
      "loss": 0.3468,
      "step": 168
    },
    {
      "epoch": 0.012790312662598741,
      "grad_norm": 1.96875,
      "learning_rate": 1.999899552638483e-05,
      "loss": 0.3382,
      "step": 169
    },
    {
      "epoch": 0.012865994986046072,
      "grad_norm": 1.078125,
      "learning_rate": 1.9998978573831546e-05,
      "loss": 0.3992,
      "step": 170
    },
    {
      "epoch": 0.012941677309493402,
      "grad_norm": 0.9921875,
      "learning_rate": 1.999896147942559e-05,
      "loss": 0.3464,
      "step": 171
    },
    {
      "epoch": 0.013017359632940732,
      "grad_norm": 0.99609375,
      "learning_rate": 1.9998944243167207e-05,
      "loss": 0.3193,
      "step": 172
    },
    {
      "epoch": 0.01309304195638806,
      "grad_norm": 0.98046875,
      "learning_rate": 1.9998926865056643e-05,
      "loss": 0.3226,
      "step": 173
    },
    {
      "epoch": 0.01316872427983539,
      "grad_norm": 1.1640625,
      "learning_rate": 1.9998909345094146e-05,
      "loss": 0.3806,
      "step": 174
    },
    {
      "epoch": 0.013244406603282721,
      "grad_norm": 1.0,
      "learning_rate": 1.9998891683279955e-05,
      "loss": 0.3052,
      "step": 175
    },
    {
      "epoch": 0.013320088926730051,
      "grad_norm": 1.0078125,
      "learning_rate": 1.999887387961433e-05,
      "loss": 0.3449,
      "step": 176
    },
    {
      "epoch": 0.013395771250177381,
      "grad_norm": 1.046875,
      "learning_rate": 1.999885593409752e-05,
      "loss": 0.3765,
      "step": 177
    },
    {
      "epoch": 0.01347145357362471,
      "grad_norm": 1.0703125,
      "learning_rate": 1.999883784672978e-05,
      "loss": 0.3538,
      "step": 178
    },
    {
      "epoch": 0.01354713589707204,
      "grad_norm": 1.3125,
      "learning_rate": 1.9998819617511363e-05,
      "loss": 0.3674,
      "step": 179
    },
    {
      "epoch": 0.01362281822051937,
      "grad_norm": 0.984375,
      "learning_rate": 1.9998801246442534e-05,
      "loss": 0.3427,
      "step": 180
    },
    {
      "epoch": 0.0136985005439667,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9998782733523553e-05,
      "loss": 0.3064,
      "step": 181
    },
    {
      "epoch": 0.013774182867414029,
      "grad_norm": 1.109375,
      "learning_rate": 1.9998764078754678e-05,
      "loss": 0.3962,
      "step": 182
    },
    {
      "epoch": 0.013849865190861359,
      "grad_norm": 1.015625,
      "learning_rate": 1.9998745282136177e-05,
      "loss": 0.3672,
      "step": 183
    },
    {
      "epoch": 0.01392554751430869,
      "grad_norm": 1.21875,
      "learning_rate": 1.9998726343668314e-05,
      "loss": 0.3572,
      "step": 184
    },
    {
      "epoch": 0.01400122983775602,
      "grad_norm": 1.0625,
      "learning_rate": 1.999870726335136e-05,
      "loss": 0.3459,
      "step": 185
    },
    {
      "epoch": 0.01407691216120335,
      "grad_norm": 0.99609375,
      "learning_rate": 1.9998688041185584e-05,
      "loss": 0.3594,
      "step": 186
    },
    {
      "epoch": 0.014152594484650678,
      "grad_norm": 1.078125,
      "learning_rate": 1.9998668677171262e-05,
      "loss": 0.3511,
      "step": 187
    },
    {
      "epoch": 0.014228276808098008,
      "grad_norm": 1.171875,
      "learning_rate": 1.9998649171308666e-05,
      "loss": 0.3971,
      "step": 188
    },
    {
      "epoch": 0.014303959131545339,
      "grad_norm": 1.1484375,
      "learning_rate": 1.9998629523598073e-05,
      "loss": 0.392,
      "step": 189
    },
    {
      "epoch": 0.014379641454992669,
      "grad_norm": 12.875,
      "learning_rate": 1.999860973403976e-05,
      "loss": 0.9823,
      "step": 190
    },
    {
      "epoch": 0.014455323778439997,
      "grad_norm": 1.1796875,
      "learning_rate": 1.9998589802634013e-05,
      "loss": 0.3623,
      "step": 191
    },
    {
      "epoch": 0.014531006101887328,
      "grad_norm": 1.0703125,
      "learning_rate": 1.9998569729381115e-05,
      "loss": 0.3461,
      "step": 192
    },
    {
      "epoch": 0.014606688425334658,
      "grad_norm": 1.0859375,
      "learning_rate": 1.9998549514281342e-05,
      "loss": 0.4037,
      "step": 193
    },
    {
      "epoch": 0.014682370748781988,
      "grad_norm": 1.046875,
      "learning_rate": 1.9998529157334985e-05,
      "loss": 0.3782,
      "step": 194
    },
    {
      "epoch": 0.014758053072229318,
      "grad_norm": 1.03125,
      "learning_rate": 1.9998508658542336e-05,
      "loss": 0.3376,
      "step": 195
    },
    {
      "epoch": 0.014833735395676647,
      "grad_norm": 1.140625,
      "learning_rate": 1.999848801790368e-05,
      "loss": 0.4202,
      "step": 196
    },
    {
      "epoch": 0.014909417719123977,
      "grad_norm": 1.0703125,
      "learning_rate": 1.9998467235419318e-05,
      "loss": 0.3973,
      "step": 197
    },
    {
      "epoch": 0.014985100042571307,
      "grad_norm": 1.0546875,
      "learning_rate": 1.999844631108954e-05,
      "loss": 0.3603,
      "step": 198
    },
    {
      "epoch": 0.015060782366018637,
      "grad_norm": 1.03125,
      "learning_rate": 1.9998425244914643e-05,
      "loss": 0.3522,
      "step": 199
    },
    {
      "epoch": 0.015136464689465966,
      "grad_norm": 1.078125,
      "learning_rate": 1.9998404036894925e-05,
      "loss": 0.4143,
      "step": 200
    },
    {
      "epoch": 0.015212147012913296,
      "grad_norm": 1.0625,
      "learning_rate": 1.999838268703069e-05,
      "loss": 0.3269,
      "step": 201
    },
    {
      "epoch": 0.015287829336360626,
      "grad_norm": 3.671875,
      "learning_rate": 1.9998361195322235e-05,
      "loss": 0.8023,
      "step": 202
    },
    {
      "epoch": 0.015363511659807956,
      "grad_norm": 1.0546875,
      "learning_rate": 1.9998339561769867e-05,
      "loss": 0.3638,
      "step": 203
    },
    {
      "epoch": 0.015439193983255287,
      "grad_norm": 0.9765625,
      "learning_rate": 1.99983177863739e-05,
      "loss": 0.367,
      "step": 204
    },
    {
      "epoch": 0.015514876306702615,
      "grad_norm": 1.1015625,
      "learning_rate": 1.9998295869134633e-05,
      "loss": 0.401,
      "step": 205
    },
    {
      "epoch": 0.015590558630149945,
      "grad_norm": 0.953125,
      "learning_rate": 1.9998273810052387e-05,
      "loss": 0.2969,
      "step": 206
    },
    {
      "epoch": 0.015666240953597275,
      "grad_norm": 1.0234375,
      "learning_rate": 1.9998251609127465e-05,
      "loss": 0.3651,
      "step": 207
    },
    {
      "epoch": 0.015741923277044604,
      "grad_norm": 0.9609375,
      "learning_rate": 1.9998229266360187e-05,
      "loss": 0.33,
      "step": 208
    },
    {
      "epoch": 0.015817605600491936,
      "grad_norm": 1.0625,
      "learning_rate": 1.9998206781750872e-05,
      "loss": 0.3632,
      "step": 209
    },
    {
      "epoch": 0.015893287923939264,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9998184155299832e-05,
      "loss": 0.319,
      "step": 210
    },
    {
      "epoch": 0.015968970247386596,
      "grad_norm": 1.109375,
      "learning_rate": 1.9998161387007396e-05,
      "loss": 0.413,
      "step": 211
    },
    {
      "epoch": 0.016044652570833925,
      "grad_norm": 1.1875,
      "learning_rate": 1.999813847687388e-05,
      "loss": 0.3912,
      "step": 212
    },
    {
      "epoch": 0.016120334894281253,
      "grad_norm": 4.5,
      "learning_rate": 1.9998115424899616e-05,
      "loss": 0.7889,
      "step": 213
    },
    {
      "epoch": 0.016196017217728585,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9998092231084926e-05,
      "loss": 0.3191,
      "step": 214
    },
    {
      "epoch": 0.016271699541175914,
      "grad_norm": 1.046875,
      "learning_rate": 1.999806889543014e-05,
      "loss": 0.3763,
      "step": 215
    },
    {
      "epoch": 0.016347381864623246,
      "grad_norm": 1.1484375,
      "learning_rate": 1.9998045417935587e-05,
      "loss": 0.4035,
      "step": 216
    },
    {
      "epoch": 0.016423064188070574,
      "grad_norm": 1.1640625,
      "learning_rate": 1.9998021798601607e-05,
      "loss": 0.389,
      "step": 217
    },
    {
      "epoch": 0.016498746511517903,
      "grad_norm": 0.9921875,
      "learning_rate": 1.9997998037428528e-05,
      "loss": 0.3311,
      "step": 218
    },
    {
      "epoch": 0.016574428834965235,
      "grad_norm": 1.0234375,
      "learning_rate": 1.999797413441669e-05,
      "loss": 0.3615,
      "step": 219
    },
    {
      "epoch": 0.016650111158412563,
      "grad_norm": 1.1015625,
      "learning_rate": 1.999795008956643e-05,
      "loss": 0.4046,
      "step": 220
    },
    {
      "epoch": 0.01672579348185989,
      "grad_norm": 0.95703125,
      "learning_rate": 1.9997925902878093e-05,
      "loss": 0.3066,
      "step": 221
    },
    {
      "epoch": 0.016801475805307223,
      "grad_norm": 1.0078125,
      "learning_rate": 1.999790157435202e-05,
      "loss": 0.3129,
      "step": 222
    },
    {
      "epoch": 0.016877158128754552,
      "grad_norm": 1.078125,
      "learning_rate": 1.9997877103988555e-05,
      "loss": 0.3568,
      "step": 223
    },
    {
      "epoch": 0.016952840452201884,
      "grad_norm": 1.1328125,
      "learning_rate": 1.999785249178805e-05,
      "loss": 0.3858,
      "step": 224
    },
    {
      "epoch": 0.017028522775649212,
      "grad_norm": 1.1484375,
      "learning_rate": 1.9997827737750848e-05,
      "loss": 0.3335,
      "step": 225
    },
    {
      "epoch": 0.01710420509909654,
      "grad_norm": 1.03125,
      "learning_rate": 1.9997802841877304e-05,
      "loss": 0.3605,
      "step": 226
    },
    {
      "epoch": 0.017179887422543873,
      "grad_norm": 1.0390625,
      "learning_rate": 1.9997777804167765e-05,
      "loss": 0.3346,
      "step": 227
    },
    {
      "epoch": 0.0172555697459912,
      "grad_norm": 1.1015625,
      "learning_rate": 1.9997752624622596e-05,
      "loss": 0.4157,
      "step": 228
    },
    {
      "epoch": 0.017331252069438533,
      "grad_norm": 1.0625,
      "learning_rate": 1.9997727303242146e-05,
      "loss": 0.3838,
      "step": 229
    },
    {
      "epoch": 0.01740693439288586,
      "grad_norm": 1.0234375,
      "learning_rate": 1.999770184002678e-05,
      "loss": 0.3644,
      "step": 230
    },
    {
      "epoch": 0.01748261671633319,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9997676234976854e-05,
      "loss": 0.3558,
      "step": 231
    },
    {
      "epoch": 0.017558299039780522,
      "grad_norm": 1.0859375,
      "learning_rate": 1.9997650488092737e-05,
      "loss": 0.4005,
      "step": 232
    },
    {
      "epoch": 0.01763398136322785,
      "grad_norm": 1.0546875,
      "learning_rate": 1.999762459937479e-05,
      "loss": 0.3902,
      "step": 233
    },
    {
      "epoch": 0.017709663686675182,
      "grad_norm": 0.97265625,
      "learning_rate": 1.999759856882338e-05,
      "loss": 0.3187,
      "step": 234
    },
    {
      "epoch": 0.01778534601012251,
      "grad_norm": 1.0546875,
      "learning_rate": 1.9997572396438883e-05,
      "loss": 0.3688,
      "step": 235
    },
    {
      "epoch": 0.01786102833356984,
      "grad_norm": 1.015625,
      "learning_rate": 1.9997546082221663e-05,
      "loss": 0.362,
      "step": 236
    },
    {
      "epoch": 0.01793671065701717,
      "grad_norm": 1.078125,
      "learning_rate": 1.9997519626172093e-05,
      "loss": 0.3781,
      "step": 237
    },
    {
      "epoch": 0.0180123929804645,
      "grad_norm": 1.1640625,
      "learning_rate": 1.9997493028290552e-05,
      "loss": 0.4281,
      "step": 238
    },
    {
      "epoch": 0.018088075303911832,
      "grad_norm": 1.0625,
      "learning_rate": 1.9997466288577417e-05,
      "loss": 0.3922,
      "step": 239
    },
    {
      "epoch": 0.01816375762735916,
      "grad_norm": 1.015625,
      "learning_rate": 1.9997439407033066e-05,
      "loss": 0.3504,
      "step": 240
    },
    {
      "epoch": 0.01823943995080649,
      "grad_norm": 1.1328125,
      "learning_rate": 1.999741238365788e-05,
      "loss": 0.3631,
      "step": 241
    },
    {
      "epoch": 0.01831512227425382,
      "grad_norm": 1.03125,
      "learning_rate": 1.999738521845224e-05,
      "loss": 0.342,
      "step": 242
    },
    {
      "epoch": 0.01839080459770115,
      "grad_norm": 1.1484375,
      "learning_rate": 1.999735791141654e-05,
      "loss": 0.4073,
      "step": 243
    },
    {
      "epoch": 0.018466486921148478,
      "grad_norm": 0.9921875,
      "learning_rate": 1.9997330462551158e-05,
      "loss": 0.3581,
      "step": 244
    },
    {
      "epoch": 0.01854216924459581,
      "grad_norm": 1.09375,
      "learning_rate": 1.9997302871856492e-05,
      "loss": 0.3977,
      "step": 245
    },
    {
      "epoch": 0.018617851568043138,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9997275139332926e-05,
      "loss": 0.32,
      "step": 246
    },
    {
      "epoch": 0.01869353389149047,
      "grad_norm": 1.1171875,
      "learning_rate": 1.9997247264980858e-05,
      "loss": 0.4087,
      "step": 247
    },
    {
      "epoch": 0.0187692162149378,
      "grad_norm": 1.0,
      "learning_rate": 1.999721924880068e-05,
      "loss": 0.3344,
      "step": 248
    },
    {
      "epoch": 0.018844898538385127,
      "grad_norm": 1.1015625,
      "learning_rate": 1.999719109079279e-05,
      "loss": 0.4057,
      "step": 249
    },
    {
      "epoch": 0.01892058086183246,
      "grad_norm": 1.0546875,
      "learning_rate": 1.999716279095759e-05,
      "loss": 0.3431,
      "step": 250
    },
    {
      "epoch": 0.018996263185279787,
      "grad_norm": 1.109375,
      "learning_rate": 1.999713434929548e-05,
      "loss": 0.4121,
      "step": 251
    },
    {
      "epoch": 0.01907194550872712,
      "grad_norm": 1.1015625,
      "learning_rate": 1.9997105765806864e-05,
      "loss": 0.3604,
      "step": 252
    },
    {
      "epoch": 0.019147627832174448,
      "grad_norm": 1.1875,
      "learning_rate": 1.9997077040492145e-05,
      "loss": 0.4072,
      "step": 253
    },
    {
      "epoch": 0.019223310155621776,
      "grad_norm": 8.875,
      "learning_rate": 1.9997048173351733e-05,
      "loss": 0.8261,
      "step": 254
    },
    {
      "epoch": 0.019298992479069108,
      "grad_norm": 1.1171875,
      "learning_rate": 1.9997019164386043e-05,
      "loss": 0.3843,
      "step": 255
    },
    {
      "epoch": 0.019374674802516437,
      "grad_norm": 1.0859375,
      "learning_rate": 1.9996990013595473e-05,
      "loss": 0.3688,
      "step": 256
    },
    {
      "epoch": 0.01945035712596377,
      "grad_norm": 7.625,
      "learning_rate": 1.9996960720980447e-05,
      "loss": 0.7332,
      "step": 257
    },
    {
      "epoch": 0.019526039449411097,
      "grad_norm": 1.1796875,
      "learning_rate": 1.999693128654138e-05,
      "loss": 0.3603,
      "step": 258
    },
    {
      "epoch": 0.019601721772858426,
      "grad_norm": 1.0625,
      "learning_rate": 1.9996901710278686e-05,
      "loss": 0.3348,
      "step": 259
    },
    {
      "epoch": 0.019677404096305758,
      "grad_norm": 1.046875,
      "learning_rate": 1.9996871992192784e-05,
      "loss": 0.3308,
      "step": 260
    },
    {
      "epoch": 0.019753086419753086,
      "grad_norm": 1.0859375,
      "learning_rate": 1.99968421322841e-05,
      "loss": 0.3428,
      "step": 261
    },
    {
      "epoch": 0.019828768743200418,
      "grad_norm": 1.1875,
      "learning_rate": 1.9996812130553054e-05,
      "loss": 0.3764,
      "step": 262
    },
    {
      "epoch": 0.019904451066647746,
      "grad_norm": 1.0703125,
      "learning_rate": 1.9996781987000073e-05,
      "loss": 0.4219,
      "step": 263
    },
    {
      "epoch": 0.019980133390095075,
      "grad_norm": 1.0234375,
      "learning_rate": 1.9996751701625584e-05,
      "loss": 0.3385,
      "step": 264
    },
    {
      "epoch": 0.020055815713542407,
      "grad_norm": 0.9765625,
      "learning_rate": 1.9996721274430017e-05,
      "loss": 0.3173,
      "step": 265
    },
    {
      "epoch": 0.020131498036989735,
      "grad_norm": 1.0703125,
      "learning_rate": 1.9996690705413803e-05,
      "loss": 0.3708,
      "step": 266
    },
    {
      "epoch": 0.020207180360437064,
      "grad_norm": 0.98828125,
      "learning_rate": 1.9996659994577378e-05,
      "loss": 0.3627,
      "step": 267
    },
    {
      "epoch": 0.020282862683884396,
      "grad_norm": 1.03125,
      "learning_rate": 1.9996629141921177e-05,
      "loss": 0.3552,
      "step": 268
    },
    {
      "epoch": 0.020358545007331724,
      "grad_norm": 1.046875,
      "learning_rate": 1.9996598147445636e-05,
      "loss": 0.3997,
      "step": 269
    },
    {
      "epoch": 0.020434227330779056,
      "grad_norm": 1.109375,
      "learning_rate": 1.9996567011151196e-05,
      "loss": 0.4172,
      "step": 270
    },
    {
      "epoch": 0.020509909654226385,
      "grad_norm": 1.109375,
      "learning_rate": 1.9996535733038298e-05,
      "loss": 0.4236,
      "step": 271
    },
    {
      "epoch": 0.020585591977673713,
      "grad_norm": 1.0703125,
      "learning_rate": 1.9996504313107384e-05,
      "loss": 0.3889,
      "step": 272
    },
    {
      "epoch": 0.020661274301121045,
      "grad_norm": 0.9921875,
      "learning_rate": 1.9996472751358903e-05,
      "loss": 0.3753,
      "step": 273
    },
    {
      "epoch": 0.020736956624568374,
      "grad_norm": 1.0078125,
      "learning_rate": 1.99964410477933e-05,
      "loss": 0.3838,
      "step": 274
    },
    {
      "epoch": 0.020812638948015705,
      "grad_norm": 1.1328125,
      "learning_rate": 1.999640920241103e-05,
      "loss": 0.4115,
      "step": 275
    },
    {
      "epoch": 0.020888321271463034,
      "grad_norm": 1.078125,
      "learning_rate": 1.999637721521254e-05,
      "loss": 0.3549,
      "step": 276
    },
    {
      "epoch": 0.020964003594910362,
      "grad_norm": 1.078125,
      "learning_rate": 1.9996345086198284e-05,
      "loss": 0.3832,
      "step": 277
    },
    {
      "epoch": 0.021039685918357694,
      "grad_norm": 1.0703125,
      "learning_rate": 1.9996312815368718e-05,
      "loss": 0.3309,
      "step": 278
    },
    {
      "epoch": 0.021115368241805023,
      "grad_norm": 1.0859375,
      "learning_rate": 1.99962804027243e-05,
      "loss": 0.3958,
      "step": 279
    },
    {
      "epoch": 0.021191050565252355,
      "grad_norm": 0.96484375,
      "learning_rate": 1.999624784826549e-05,
      "loss": 0.3342,
      "step": 280
    },
    {
      "epoch": 0.021266732888699683,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9996215151992752e-05,
      "loss": 0.3387,
      "step": 281
    },
    {
      "epoch": 0.021342415212147012,
      "grad_norm": 1.046875,
      "learning_rate": 1.999618231390655e-05,
      "loss": 0.3561,
      "step": 282
    },
    {
      "epoch": 0.021418097535594344,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9996149334007345e-05,
      "loss": 0.3977,
      "step": 283
    },
    {
      "epoch": 0.021493779859041672,
      "grad_norm": 1.0546875,
      "learning_rate": 1.9996116212295605e-05,
      "loss": 0.3302,
      "step": 284
    },
    {
      "epoch": 0.021569462182489004,
      "grad_norm": 1.0390625,
      "learning_rate": 1.9996082948771807e-05,
      "loss": 0.3461,
      "step": 285
    },
    {
      "epoch": 0.021645144505936333,
      "grad_norm": 1.1171875,
      "learning_rate": 1.9996049543436417e-05,
      "loss": 0.4001,
      "step": 286
    },
    {
      "epoch": 0.02172082682938366,
      "grad_norm": 0.98828125,
      "learning_rate": 1.9996015996289908e-05,
      "loss": 0.358,
      "step": 287
    },
    {
      "epoch": 0.021796509152830993,
      "grad_norm": 1.8125,
      "learning_rate": 1.9995982307332763e-05,
      "loss": 0.3572,
      "step": 288
    },
    {
      "epoch": 0.02187219147627832,
      "grad_norm": 1.078125,
      "learning_rate": 1.9995948476565453e-05,
      "loss": 0.3859,
      "step": 289
    },
    {
      "epoch": 0.02194787379972565,
      "grad_norm": 1.046875,
      "learning_rate": 1.999591450398846e-05,
      "loss": 0.3583,
      "step": 290
    },
    {
      "epoch": 0.022023556123172982,
      "grad_norm": 1.0234375,
      "learning_rate": 1.9995880389602265e-05,
      "loss": 0.352,
      "step": 291
    },
    {
      "epoch": 0.02209923844662031,
      "grad_norm": 0.9921875,
      "learning_rate": 1.9995846133407356e-05,
      "loss": 0.3163,
      "step": 292
    },
    {
      "epoch": 0.022174920770067642,
      "grad_norm": 1.234375,
      "learning_rate": 1.9995811735404213e-05,
      "loss": 0.4019,
      "step": 293
    },
    {
      "epoch": 0.02225060309351497,
      "grad_norm": 1.046875,
      "learning_rate": 1.999577719559333e-05,
      "loss": 0.3729,
      "step": 294
    },
    {
      "epoch": 0.0223262854169623,
      "grad_norm": 0.90625,
      "learning_rate": 1.9995742513975195e-05,
      "loss": 0.3154,
      "step": 295
    },
    {
      "epoch": 0.02240196774040963,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9995707690550293e-05,
      "loss": 0.3759,
      "step": 296
    },
    {
      "epoch": 0.02247765006385696,
      "grad_norm": 1.078125,
      "learning_rate": 1.999567272531913e-05,
      "loss": 0.3599,
      "step": 297
    },
    {
      "epoch": 0.02255333238730429,
      "grad_norm": 1.0234375,
      "learning_rate": 1.9995637618282196e-05,
      "loss": 0.3615,
      "step": 298
    },
    {
      "epoch": 0.02262901471075162,
      "grad_norm": 1.0546875,
      "learning_rate": 1.9995602369439987e-05,
      "loss": 0.3826,
      "step": 299
    },
    {
      "epoch": 0.02270469703419895,
      "grad_norm": 1.0546875,
      "learning_rate": 1.9995566978793004e-05,
      "loss": 0.3432,
      "step": 300
    },
    {
      "epoch": 0.02278037935764628,
      "grad_norm": 5.53125,
      "learning_rate": 1.9995531446341755e-05,
      "loss": 1.1258,
      "step": 301
    },
    {
      "epoch": 0.02285606168109361,
      "grad_norm": 1.1015625,
      "learning_rate": 1.9995495772086735e-05,
      "loss": 0.336,
      "step": 302
    },
    {
      "epoch": 0.02293174400454094,
      "grad_norm": 1.1015625,
      "learning_rate": 1.999545995602846e-05,
      "loss": 0.4079,
      "step": 303
    },
    {
      "epoch": 0.02300742632798827,
      "grad_norm": 1.0859375,
      "learning_rate": 1.9995423998167428e-05,
      "loss": 0.3879,
      "step": 304
    },
    {
      "epoch": 0.023083108651435598,
      "grad_norm": 1.0859375,
      "learning_rate": 1.9995387898504154e-05,
      "loss": 0.4143,
      "step": 305
    },
    {
      "epoch": 0.02315879097488293,
      "grad_norm": 1.0703125,
      "learning_rate": 1.9995351657039152e-05,
      "loss": 0.3996,
      "step": 306
    },
    {
      "epoch": 0.02323447329833026,
      "grad_norm": 1.015625,
      "learning_rate": 1.9995315273772933e-05,
      "loss": 0.3307,
      "step": 307
    },
    {
      "epoch": 0.02331015562177759,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9995278748706017e-05,
      "loss": 0.295,
      "step": 308
    },
    {
      "epoch": 0.02338583794522492,
      "grad_norm": 1.1328125,
      "learning_rate": 1.9995242081838913e-05,
      "loss": 0.3863,
      "step": 309
    },
    {
      "epoch": 0.023461520268672247,
      "grad_norm": 1.0703125,
      "learning_rate": 1.9995205273172152e-05,
      "loss": 0.3738,
      "step": 310
    },
    {
      "epoch": 0.02353720259211958,
      "grad_norm": 1.0703125,
      "learning_rate": 1.9995168322706255e-05,
      "loss": 0.423,
      "step": 311
    },
    {
      "epoch": 0.023612884915566908,
      "grad_norm": 1.1015625,
      "learning_rate": 1.9995131230441736e-05,
      "loss": 0.3942,
      "step": 312
    },
    {
      "epoch": 0.023688567239014236,
      "grad_norm": 0.98828125,
      "learning_rate": 1.9995093996379132e-05,
      "loss": 0.3501,
      "step": 313
    },
    {
      "epoch": 0.023764249562461568,
      "grad_norm": 0.98046875,
      "learning_rate": 1.999505662051897e-05,
      "loss": 0.3541,
      "step": 314
    },
    {
      "epoch": 0.023839931885908897,
      "grad_norm": 0.96875,
      "learning_rate": 1.9995019102861773e-05,
      "loss": 0.3411,
      "step": 315
    },
    {
      "epoch": 0.02391561420935623,
      "grad_norm": 1.046875,
      "learning_rate": 1.999498144340808e-05,
      "loss": 0.3286,
      "step": 316
    },
    {
      "epoch": 0.023991296532803557,
      "grad_norm": 1.046875,
      "learning_rate": 1.9994943642158423e-05,
      "loss": 0.358,
      "step": 317
    },
    {
      "epoch": 0.024066978856250885,
      "grad_norm": 0.921875,
      "learning_rate": 1.9994905699113342e-05,
      "loss": 0.332,
      "step": 318
    },
    {
      "epoch": 0.024142661179698217,
      "grad_norm": 1.03125,
      "learning_rate": 1.9994867614273366e-05,
      "loss": 0.3823,
      "step": 319
    },
    {
      "epoch": 0.024218343503145546,
      "grad_norm": 1.1328125,
      "learning_rate": 1.9994829387639044e-05,
      "loss": 0.4049,
      "step": 320
    },
    {
      "epoch": 0.024294025826592878,
      "grad_norm": 1.09375,
      "learning_rate": 1.9994791019210916e-05,
      "loss": 0.3318,
      "step": 321
    },
    {
      "epoch": 0.024369708150040206,
      "grad_norm": 3.25,
      "learning_rate": 1.9994752508989524e-05,
      "loss": 0.7194,
      "step": 322
    },
    {
      "epoch": 0.024445390473487535,
      "grad_norm": 0.98828125,
      "learning_rate": 1.999471385697542e-05,
      "loss": 0.3353,
      "step": 323
    },
    {
      "epoch": 0.024521072796934867,
      "grad_norm": 0.94921875,
      "learning_rate": 1.9994675063169144e-05,
      "loss": 0.3372,
      "step": 324
    },
    {
      "epoch": 0.024596755120382195,
      "grad_norm": 1.1015625,
      "learning_rate": 1.999463612757125e-05,
      "loss": 0.3642,
      "step": 325
    },
    {
      "epoch": 0.024672437443829527,
      "grad_norm": 1.015625,
      "learning_rate": 1.9994597050182296e-05,
      "loss": 0.3833,
      "step": 326
    },
    {
      "epoch": 0.024748119767276856,
      "grad_norm": 1.1015625,
      "learning_rate": 1.9994557831002827e-05,
      "loss": 0.3944,
      "step": 327
    },
    {
      "epoch": 0.024823802090724184,
      "grad_norm": 0.9375,
      "learning_rate": 1.9994518470033404e-05,
      "loss": 0.3282,
      "step": 328
    },
    {
      "epoch": 0.024899484414171516,
      "grad_norm": 1.0234375,
      "learning_rate": 1.999447896727459e-05,
      "loss": 0.3873,
      "step": 329
    },
    {
      "epoch": 0.024975166737618845,
      "grad_norm": 1.015625,
      "learning_rate": 1.999443932272694e-05,
      "loss": 0.3365,
      "step": 330
    },
    {
      "epoch": 0.025050849061066176,
      "grad_norm": 1.0,
      "learning_rate": 1.9994399536391014e-05,
      "loss": 0.3776,
      "step": 331
    },
    {
      "epoch": 0.025126531384513505,
      "grad_norm": 0.96875,
      "learning_rate": 1.999435960826738e-05,
      "loss": 0.31,
      "step": 332
    },
    {
      "epoch": 0.025202213707960833,
      "grad_norm": 1.0625,
      "learning_rate": 1.9994319538356602e-05,
      "loss": 0.3714,
      "step": 333
    },
    {
      "epoch": 0.025277896031408165,
      "grad_norm": 0.9609375,
      "learning_rate": 1.9994279326659254e-05,
      "loss": 0.3361,
      "step": 334
    },
    {
      "epoch": 0.025353578354855494,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9994238973175902e-05,
      "loss": 0.3584,
      "step": 335
    },
    {
      "epoch": 0.025429260678302822,
      "grad_norm": 6.6875,
      "learning_rate": 1.999419847790712e-05,
      "loss": 0.7323,
      "step": 336
    },
    {
      "epoch": 0.025504943001750154,
      "grad_norm": 0.98828125,
      "learning_rate": 1.9994157840853484e-05,
      "loss": 0.3428,
      "step": 337
    },
    {
      "epoch": 0.025580625325197483,
      "grad_norm": 1.0703125,
      "learning_rate": 1.9994117062015563e-05,
      "loss": 0.3756,
      "step": 338
    },
    {
      "epoch": 0.025656307648644815,
      "grad_norm": 0.953125,
      "learning_rate": 1.9994076141393946e-05,
      "loss": 0.3094,
      "step": 339
    },
    {
      "epoch": 0.025731989972092143,
      "grad_norm": 0.953125,
      "learning_rate": 1.9994035078989205e-05,
      "loss": 0.3173,
      "step": 340
    },
    {
      "epoch": 0.02580767229553947,
      "grad_norm": 1.21875,
      "learning_rate": 1.999399387480193e-05,
      "loss": 0.3591,
      "step": 341
    },
    {
      "epoch": 0.025883354618986804,
      "grad_norm": 1.15625,
      "learning_rate": 1.99939525288327e-05,
      "loss": 0.3846,
      "step": 342
    },
    {
      "epoch": 0.025959036942434132,
      "grad_norm": 1.2265625,
      "learning_rate": 1.99939110410821e-05,
      "loss": 0.381,
      "step": 343
    },
    {
      "epoch": 0.026034719265881464,
      "grad_norm": 1.0546875,
      "learning_rate": 1.9993869411550724e-05,
      "loss": 0.3303,
      "step": 344
    },
    {
      "epoch": 0.026110401589328792,
      "grad_norm": 1.0390625,
      "learning_rate": 1.999382764023916e-05,
      "loss": 0.3957,
      "step": 345
    },
    {
      "epoch": 0.02618608391277612,
      "grad_norm": 0.921875,
      "learning_rate": 1.9993785727148004e-05,
      "loss": 0.311,
      "step": 346
    },
    {
      "epoch": 0.026261766236223453,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9993743672277842e-05,
      "loss": 0.2958,
      "step": 347
    },
    {
      "epoch": 0.02633744855967078,
      "grad_norm": 1.0859375,
      "learning_rate": 1.999370147562928e-05,
      "loss": 0.338,
      "step": 348
    },
    {
      "epoch": 0.026413130883118113,
      "grad_norm": 0.8984375,
      "learning_rate": 1.999365913720291e-05,
      "loss": 0.265,
      "step": 349
    },
    {
      "epoch": 0.026488813206565442,
      "grad_norm": 0.9921875,
      "learning_rate": 1.999361665699934e-05,
      "loss": 0.3467,
      "step": 350
    },
    {
      "epoch": 0.02656449553001277,
      "grad_norm": 0.99609375,
      "learning_rate": 1.9993574035019164e-05,
      "loss": 0.3734,
      "step": 351
    },
    {
      "epoch": 0.026640177853460102,
      "grad_norm": 9.0625,
      "learning_rate": 1.999353127126299e-05,
      "loss": 0.7803,
      "step": 352
    },
    {
      "epoch": 0.02671586017690743,
      "grad_norm": 1.046875,
      "learning_rate": 1.9993488365731427e-05,
      "loss": 0.3987,
      "step": 353
    },
    {
      "epoch": 0.026791542500354763,
      "grad_norm": 1.03125,
      "learning_rate": 1.9993445318425086e-05,
      "loss": 0.3616,
      "step": 354
    },
    {
      "epoch": 0.02686722482380209,
      "grad_norm": 0.98828125,
      "learning_rate": 1.999340212934457e-05,
      "loss": 0.3683,
      "step": 355
    },
    {
      "epoch": 0.02694290714724942,
      "grad_norm": 1.0703125,
      "learning_rate": 1.9993358798490498e-05,
      "loss": 0.3761,
      "step": 356
    },
    {
      "epoch": 0.02701858947069675,
      "grad_norm": 1.0390625,
      "learning_rate": 1.999331532586348e-05,
      "loss": 0.3949,
      "step": 357
    },
    {
      "epoch": 0.02709427179414408,
      "grad_norm": 0.97265625,
      "learning_rate": 1.9993271711464133e-05,
      "loss": 0.3537,
      "step": 358
    },
    {
      "epoch": 0.02716995411759141,
      "grad_norm": 0.91015625,
      "learning_rate": 1.999322795529308e-05,
      "loss": 0.3376,
      "step": 359
    },
    {
      "epoch": 0.02724563644103874,
      "grad_norm": 0.99609375,
      "learning_rate": 1.9993184057350937e-05,
      "loss": 0.3848,
      "step": 360
    },
    {
      "epoch": 0.02732131876448607,
      "grad_norm": 1.1796875,
      "learning_rate": 1.9993140017638332e-05,
      "loss": 0.4399,
      "step": 361
    },
    {
      "epoch": 0.0273970010879334,
      "grad_norm": 1.125,
      "learning_rate": 1.999309583615589e-05,
      "loss": 0.4621,
      "step": 362
    },
    {
      "epoch": 0.02747268341138073,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9993051512904233e-05,
      "loss": 0.3558,
      "step": 363
    },
    {
      "epoch": 0.027548365734828058,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9993007047883988e-05,
      "loss": 0.3733,
      "step": 364
    },
    {
      "epoch": 0.02762404805827539,
      "grad_norm": 0.99609375,
      "learning_rate": 1.999296244109579e-05,
      "loss": 0.3368,
      "step": 365
    },
    {
      "epoch": 0.027699730381722718,
      "grad_norm": 1.0703125,
      "learning_rate": 1.9992917692540273e-05,
      "loss": 0.3937,
      "step": 366
    },
    {
      "epoch": 0.02777541270517005,
      "grad_norm": 0.96484375,
      "learning_rate": 1.999287280221807e-05,
      "loss": 0.3425,
      "step": 367
    },
    {
      "epoch": 0.02785109502861738,
      "grad_norm": 1.1015625,
      "learning_rate": 1.9992827770129816e-05,
      "loss": 0.4042,
      "step": 368
    },
    {
      "epoch": 0.027926777352064707,
      "grad_norm": 1.015625,
      "learning_rate": 1.9992782596276155e-05,
      "loss": 0.3324,
      "step": 369
    },
    {
      "epoch": 0.02800245967551204,
      "grad_norm": 1.0703125,
      "learning_rate": 1.9992737280657723e-05,
      "loss": 0.3725,
      "step": 370
    },
    {
      "epoch": 0.028078141998959368,
      "grad_norm": 1.078125,
      "learning_rate": 1.999269182327517e-05,
      "loss": 0.3757,
      "step": 371
    },
    {
      "epoch": 0.0281538243224067,
      "grad_norm": 0.984375,
      "learning_rate": 1.9992646224129127e-05,
      "loss": 0.362,
      "step": 372
    },
    {
      "epoch": 0.028229506645854028,
      "grad_norm": 0.921875,
      "learning_rate": 1.9992600483220254e-05,
      "loss": 0.3548,
      "step": 373
    },
    {
      "epoch": 0.028305188969301356,
      "grad_norm": 0.9921875,
      "learning_rate": 1.9992554600549192e-05,
      "loss": 0.3442,
      "step": 374
    },
    {
      "epoch": 0.02838087129274869,
      "grad_norm": 1.03125,
      "learning_rate": 1.9992508576116595e-05,
      "loss": 0.3431,
      "step": 375
    },
    {
      "epoch": 0.028456553616196017,
      "grad_norm": 1.03125,
      "learning_rate": 1.9992462409923118e-05,
      "loss": 0.3936,
      "step": 376
    },
    {
      "epoch": 0.02853223593964335,
      "grad_norm": 1.046875,
      "learning_rate": 1.9992416101969417e-05,
      "loss": 0.3671,
      "step": 377
    },
    {
      "epoch": 0.028607918263090677,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9992369652256143e-05,
      "loss": 0.3083,
      "step": 378
    },
    {
      "epoch": 0.028683600586538006,
      "grad_norm": 1.0703125,
      "learning_rate": 1.999232306078396e-05,
      "loss": 0.3725,
      "step": 379
    },
    {
      "epoch": 0.028759282909985338,
      "grad_norm": 1.171875,
      "learning_rate": 1.9992276327553525e-05,
      "loss": 0.3949,
      "step": 380
    },
    {
      "epoch": 0.028834965233432666,
      "grad_norm": 1.21875,
      "learning_rate": 1.99922294525655e-05,
      "loss": 0.4306,
      "step": 381
    },
    {
      "epoch": 0.028910647556879995,
      "grad_norm": 0.96484375,
      "learning_rate": 1.999218243582056e-05,
      "loss": 0.3431,
      "step": 382
    },
    {
      "epoch": 0.028986329880327327,
      "grad_norm": 0.99609375,
      "learning_rate": 1.9992135277319356e-05,
      "loss": 0.343,
      "step": 383
    },
    {
      "epoch": 0.029062012203774655,
      "grad_norm": 1.1171875,
      "learning_rate": 1.9992087977062577e-05,
      "loss": 0.3378,
      "step": 384
    },
    {
      "epoch": 0.029137694527221987,
      "grad_norm": 1.0,
      "learning_rate": 1.9992040535050872e-05,
      "loss": 0.3302,
      "step": 385
    },
    {
      "epoch": 0.029213376850669315,
      "grad_norm": 1.1015625,
      "learning_rate": 1.999199295128493e-05,
      "loss": 0.4263,
      "step": 386
    },
    {
      "epoch": 0.029289059174116644,
      "grad_norm": 1.0234375,
      "learning_rate": 1.9991945225765425e-05,
      "loss": 0.3456,
      "step": 387
    },
    {
      "epoch": 0.029364741497563976,
      "grad_norm": 1.15625,
      "learning_rate": 1.9991897358493023e-05,
      "loss": 0.4506,
      "step": 388
    },
    {
      "epoch": 0.029440423821011304,
      "grad_norm": 4.0625,
      "learning_rate": 1.9991849349468414e-05,
      "loss": 0.7689,
      "step": 389
    },
    {
      "epoch": 0.029516106144458636,
      "grad_norm": 1.0859375,
      "learning_rate": 1.999180119869228e-05,
      "loss": 0.3778,
      "step": 390
    },
    {
      "epoch": 0.029591788467905965,
      "grad_norm": 0.98828125,
      "learning_rate": 1.9991752906165292e-05,
      "loss": 0.3517,
      "step": 391
    },
    {
      "epoch": 0.029667470791353293,
      "grad_norm": 1.015625,
      "learning_rate": 1.9991704471888145e-05,
      "loss": 0.3383,
      "step": 392
    },
    {
      "epoch": 0.029743153114800625,
      "grad_norm": 6.21875,
      "learning_rate": 1.9991655895861528e-05,
      "loss": 0.8147,
      "step": 393
    },
    {
      "epoch": 0.029818835438247954,
      "grad_norm": 0.984375,
      "learning_rate": 1.999160717808612e-05,
      "loss": 0.3261,
      "step": 394
    },
    {
      "epoch": 0.029894517761695286,
      "grad_norm": 1.0546875,
      "learning_rate": 1.999155831856262e-05,
      "loss": 0.3862,
      "step": 395
    },
    {
      "epoch": 0.029970200085142614,
      "grad_norm": 0.99609375,
      "learning_rate": 1.9991509317291716e-05,
      "loss": 0.3471,
      "step": 396
    },
    {
      "epoch": 0.030045882408589943,
      "grad_norm": 1.046875,
      "learning_rate": 1.9991460174274112e-05,
      "loss": 0.4,
      "step": 397
    },
    {
      "epoch": 0.030121564732037275,
      "grad_norm": 1.046875,
      "learning_rate": 1.9991410889510498e-05,
      "loss": 0.4054,
      "step": 398
    },
    {
      "epoch": 0.030197247055484603,
      "grad_norm": 1.09375,
      "learning_rate": 1.999136146300157e-05,
      "loss": 0.3633,
      "step": 399
    },
    {
      "epoch": 0.03027292937893193,
      "grad_norm": 1.015625,
      "learning_rate": 1.9991311894748037e-05,
      "loss": 0.3187,
      "step": 400
    },
    {
      "epoch": 0.030348611702379263,
      "grad_norm": 0.9765625,
      "learning_rate": 1.99912621847506e-05,
      "loss": 0.3414,
      "step": 401
    },
    {
      "epoch": 0.030424294025826592,
      "grad_norm": 1.046875,
      "learning_rate": 1.9991212333009965e-05,
      "loss": 0.3762,
      "step": 402
    },
    {
      "epoch": 0.030499976349273924,
      "grad_norm": 0.9453125,
      "learning_rate": 1.9991162339526837e-05,
      "loss": 0.3128,
      "step": 403
    },
    {
      "epoch": 0.030575658672721252,
      "grad_norm": 0.9921875,
      "learning_rate": 1.9991112204301923e-05,
      "loss": 0.3288,
      "step": 404
    },
    {
      "epoch": 0.03065134099616858,
      "grad_norm": 1.109375,
      "learning_rate": 1.9991061927335935e-05,
      "loss": 0.3776,
      "step": 405
    },
    {
      "epoch": 0.030727023319615913,
      "grad_norm": 0.9453125,
      "learning_rate": 1.9991011508629592e-05,
      "loss": 0.3483,
      "step": 406
    },
    {
      "epoch": 0.03080270564306324,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9990960948183605e-05,
      "loss": 0.375,
      "step": 407
    },
    {
      "epoch": 0.030878387966510573,
      "grad_norm": 1.015625,
      "learning_rate": 1.9990910245998693e-05,
      "loss": 0.3565,
      "step": 408
    },
    {
      "epoch": 0.0309540702899579,
      "grad_norm": 0.94921875,
      "learning_rate": 1.9990859402075572e-05,
      "loss": 0.3278,
      "step": 409
    },
    {
      "epoch": 0.03102975261340523,
      "grad_norm": 32.25,
      "learning_rate": 1.999080841641497e-05,
      "loss": 1.1714,
      "step": 410
    },
    {
      "epoch": 0.031105434936852562,
      "grad_norm": 1.0078125,
      "learning_rate": 1.99907572890176e-05,
      "loss": 0.3316,
      "step": 411
    },
    {
      "epoch": 0.03118111726029989,
      "grad_norm": 1.171875,
      "learning_rate": 1.9990706019884195e-05,
      "loss": 0.4189,
      "step": 412
    },
    {
      "epoch": 0.03125679958374722,
      "grad_norm": 1.0234375,
      "learning_rate": 1.9990654609015482e-05,
      "loss": 0.3271,
      "step": 413
    },
    {
      "epoch": 0.03133248190719455,
      "grad_norm": 0.921875,
      "learning_rate": 1.9990603056412187e-05,
      "loss": 0.3243,
      "step": 414
    },
    {
      "epoch": 0.03140816423064188,
      "grad_norm": 1.015625,
      "learning_rate": 1.9990551362075042e-05,
      "loss": 0.3185,
      "step": 415
    },
    {
      "epoch": 0.03148384655408921,
      "grad_norm": 15.75,
      "learning_rate": 1.9990499526004785e-05,
      "loss": 0.9282,
      "step": 416
    },
    {
      "epoch": 0.03155952887753654,
      "grad_norm": 1.1484375,
      "learning_rate": 1.999044754820215e-05,
      "loss": 0.3902,
      "step": 417
    },
    {
      "epoch": 0.03163521120098387,
      "grad_norm": 1.078125,
      "learning_rate": 1.9990395428667864e-05,
      "loss": 0.3887,
      "step": 418
    },
    {
      "epoch": 0.0317108935244312,
      "grad_norm": 1.140625,
      "learning_rate": 1.999034316740268e-05,
      "loss": 0.3495,
      "step": 419
    },
    {
      "epoch": 0.03178657584787853,
      "grad_norm": 1.09375,
      "learning_rate": 1.9990290764407336e-05,
      "loss": 0.4231,
      "step": 420
    },
    {
      "epoch": 0.03186225817132586,
      "grad_norm": 0.97265625,
      "learning_rate": 1.999023821968257e-05,
      "loss": 0.3373,
      "step": 421
    },
    {
      "epoch": 0.03193794049477319,
      "grad_norm": 12.9375,
      "learning_rate": 1.9990185533229134e-05,
      "loss": 0.6391,
      "step": 422
    },
    {
      "epoch": 0.03201362281822052,
      "grad_norm": 1.0078125,
      "learning_rate": 1.999013270504777e-05,
      "loss": 0.3642,
      "step": 423
    },
    {
      "epoch": 0.03208930514166785,
      "grad_norm": 1.046875,
      "learning_rate": 1.999007973513923e-05,
      "loss": 0.4051,
      "step": 424
    },
    {
      "epoch": 0.03216498746511518,
      "grad_norm": 0.96484375,
      "learning_rate": 1.9990026623504267e-05,
      "loss": 0.3531,
      "step": 425
    },
    {
      "epoch": 0.03224066978856251,
      "grad_norm": 1.0703125,
      "learning_rate": 1.9989973370143633e-05,
      "loss": 0.3922,
      "step": 426
    },
    {
      "epoch": 0.03231635211200984,
      "grad_norm": 1.0390625,
      "learning_rate": 1.998991997505808e-05,
      "loss": 0.3621,
      "step": 427
    },
    {
      "epoch": 0.03239203443545717,
      "grad_norm": 1.09375,
      "learning_rate": 1.9989866438248372e-05,
      "loss": 0.3859,
      "step": 428
    },
    {
      "epoch": 0.032467716758904495,
      "grad_norm": 1.0,
      "learning_rate": 1.9989812759715264e-05,
      "loss": 0.3637,
      "step": 429
    },
    {
      "epoch": 0.03254339908235183,
      "grad_norm": 0.97265625,
      "learning_rate": 1.998975893945952e-05,
      "loss": 0.3548,
      "step": 430
    },
    {
      "epoch": 0.03261908140579916,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9989704977481903e-05,
      "loss": 0.3463,
      "step": 431
    },
    {
      "epoch": 0.03269476372924649,
      "grad_norm": 1.0,
      "learning_rate": 1.998965087378318e-05,
      "loss": 0.3561,
      "step": 432
    },
    {
      "epoch": 0.032770446052693816,
      "grad_norm": 1.1171875,
      "learning_rate": 1.998959662836411e-05,
      "loss": 0.3859,
      "step": 433
    },
    {
      "epoch": 0.03284612837614115,
      "grad_norm": 1.03125,
      "learning_rate": 1.9989542241225473e-05,
      "loss": 0.3837,
      "step": 434
    },
    {
      "epoch": 0.03292181069958848,
      "grad_norm": 1.0,
      "learning_rate": 1.9989487712368034e-05,
      "loss": 0.3426,
      "step": 435
    },
    {
      "epoch": 0.032997493023035805,
      "grad_norm": 1.0703125,
      "learning_rate": 1.998943304179257e-05,
      "loss": 0.3876,
      "step": 436
    },
    {
      "epoch": 0.03307317534648314,
      "grad_norm": 0.9453125,
      "learning_rate": 1.998937822949986e-05,
      "loss": 0.3543,
      "step": 437
    },
    {
      "epoch": 0.03314885766993047,
      "grad_norm": 1.015625,
      "learning_rate": 1.9989323275490674e-05,
      "loss": 0.364,
      "step": 438
    },
    {
      "epoch": 0.033224539993377794,
      "grad_norm": 1.03125,
      "learning_rate": 1.9989268179765797e-05,
      "loss": 0.3972,
      "step": 439
    },
    {
      "epoch": 0.033300222316825126,
      "grad_norm": 0.921875,
      "learning_rate": 1.9989212942326005e-05,
      "loss": 0.3204,
      "step": 440
    },
    {
      "epoch": 0.03337590464027246,
      "grad_norm": 0.88671875,
      "learning_rate": 1.998915756317209e-05,
      "loss": 0.3114,
      "step": 441
    },
    {
      "epoch": 0.03345158696371978,
      "grad_norm": 0.953125,
      "learning_rate": 1.998910204230483e-05,
      "loss": 0.3084,
      "step": 442
    },
    {
      "epoch": 0.033527269287167115,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9989046379725016e-05,
      "loss": 0.337,
      "step": 443
    },
    {
      "epoch": 0.03360295161061445,
      "grad_norm": 1.0,
      "learning_rate": 1.998899057543344e-05,
      "loss": 0.3818,
      "step": 444
    },
    {
      "epoch": 0.03367863393406178,
      "grad_norm": 1.3828125,
      "learning_rate": 1.998893462943089e-05,
      "loss": 0.3819,
      "step": 445
    },
    {
      "epoch": 0.033754316257509104,
      "grad_norm": 0.96875,
      "learning_rate": 1.9988878541718154e-05,
      "loss": 0.3415,
      "step": 446
    },
    {
      "epoch": 0.033829998580956436,
      "grad_norm": 1.0078125,
      "learning_rate": 1.998882231229604e-05,
      "loss": 0.3559,
      "step": 447
    },
    {
      "epoch": 0.03390568090440377,
      "grad_norm": 1.0546875,
      "learning_rate": 1.998876594116534e-05,
      "loss": 0.3875,
      "step": 448
    },
    {
      "epoch": 0.03398136322785109,
      "grad_norm": 1.0703125,
      "learning_rate": 1.9988709428326857e-05,
      "loss": 0.3778,
      "step": 449
    },
    {
      "epoch": 0.034057045551298425,
      "grad_norm": 1.0859375,
      "learning_rate": 1.998865277378138e-05,
      "loss": 0.4017,
      "step": 450
    },
    {
      "epoch": 0.03413272787474576,
      "grad_norm": 1.0390625,
      "learning_rate": 1.9988595977529733e-05,
      "loss": 0.3848,
      "step": 451
    },
    {
      "epoch": 0.03420841019819308,
      "grad_norm": 0.99609375,
      "learning_rate": 1.9988539039572705e-05,
      "loss": 0.3476,
      "step": 452
    },
    {
      "epoch": 0.034284092521640414,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9988481959911114e-05,
      "loss": 0.3858,
      "step": 453
    },
    {
      "epoch": 0.034359774845087745,
      "grad_norm": 1.0625,
      "learning_rate": 1.9988424738545762e-05,
      "loss": 0.3817,
      "step": 454
    },
    {
      "epoch": 0.03443545716853508,
      "grad_norm": 0.953125,
      "learning_rate": 1.9988367375477464e-05,
      "loss": 0.3554,
      "step": 455
    },
    {
      "epoch": 0.0345111394919824,
      "grad_norm": 1.015625,
      "learning_rate": 1.9988309870707035e-05,
      "loss": 0.3682,
      "step": 456
    },
    {
      "epoch": 0.034586821815429734,
      "grad_norm": 1.078125,
      "learning_rate": 1.998825222423529e-05,
      "loss": 0.3893,
      "step": 457
    },
    {
      "epoch": 0.034662504138877066,
      "grad_norm": 0.9375,
      "learning_rate": 1.9988194436063047e-05,
      "loss": 0.3273,
      "step": 458
    },
    {
      "epoch": 0.03473818646232439,
      "grad_norm": 1.0234375,
      "learning_rate": 1.9988136506191126e-05,
      "loss": 0.3448,
      "step": 459
    },
    {
      "epoch": 0.03481386878577172,
      "grad_norm": 1.03125,
      "learning_rate": 1.998807843462035e-05,
      "loss": 0.3604,
      "step": 460
    },
    {
      "epoch": 0.034889551109219055,
      "grad_norm": 1.4375,
      "learning_rate": 1.998802022135154e-05,
      "loss": 0.3509,
      "step": 461
    },
    {
      "epoch": 0.03496523343266638,
      "grad_norm": 1.03125,
      "learning_rate": 1.998796186638552e-05,
      "loss": 0.3841,
      "step": 462
    },
    {
      "epoch": 0.03504091575611371,
      "grad_norm": 0.984375,
      "learning_rate": 1.9987903369723126e-05,
      "loss": 0.3817,
      "step": 463
    },
    {
      "epoch": 0.035116598079561044,
      "grad_norm": 1.015625,
      "learning_rate": 1.998784473136518e-05,
      "loss": 0.362,
      "step": 464
    },
    {
      "epoch": 0.03519228040300837,
      "grad_norm": 1.1796875,
      "learning_rate": 1.9987785951312516e-05,
      "loss": 0.3935,
      "step": 465
    },
    {
      "epoch": 0.0352679627264557,
      "grad_norm": 1.015625,
      "learning_rate": 1.9987727029565972e-05,
      "loss": 0.3549,
      "step": 466
    },
    {
      "epoch": 0.03534364504990303,
      "grad_norm": 0.984375,
      "learning_rate": 1.9987667966126377e-05,
      "loss": 0.3687,
      "step": 467
    },
    {
      "epoch": 0.035419327373350365,
      "grad_norm": 1.015625,
      "learning_rate": 1.9987608760994576e-05,
      "loss": 0.3878,
      "step": 468
    },
    {
      "epoch": 0.03549500969679769,
      "grad_norm": 1.0234375,
      "learning_rate": 1.9987549414171402e-05,
      "loss": 0.3654,
      "step": 469
    },
    {
      "epoch": 0.03557069202024502,
      "grad_norm": 0.99609375,
      "learning_rate": 1.99874899256577e-05,
      "loss": 0.3263,
      "step": 470
    },
    {
      "epoch": 0.035646374343692354,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9987430295454318e-05,
      "loss": 0.3389,
      "step": 471
    },
    {
      "epoch": 0.03572205666713968,
      "grad_norm": 1.1171875,
      "learning_rate": 1.9987370523562095e-05,
      "loss": 0.3571,
      "step": 472
    },
    {
      "epoch": 0.03579773899058701,
      "grad_norm": 1.0390625,
      "learning_rate": 1.998731060998188e-05,
      "loss": 0.3943,
      "step": 473
    },
    {
      "epoch": 0.03587342131403434,
      "grad_norm": 0.9375,
      "learning_rate": 1.9987250554714532e-05,
      "loss": 0.3437,
      "step": 474
    },
    {
      "epoch": 0.03594910363748167,
      "grad_norm": 1.0546875,
      "learning_rate": 1.9987190357760893e-05,
      "loss": 0.4281,
      "step": 475
    },
    {
      "epoch": 0.036024785960929,
      "grad_norm": 1.0625,
      "learning_rate": 1.998713001912182e-05,
      "loss": 0.3613,
      "step": 476
    },
    {
      "epoch": 0.03610046828437633,
      "grad_norm": 1.140625,
      "learning_rate": 1.998706953879817e-05,
      "loss": 0.3518,
      "step": 477
    },
    {
      "epoch": 0.036176150607823664,
      "grad_norm": 1.0390625,
      "learning_rate": 1.9987008916790795e-05,
      "loss": 0.4117,
      "step": 478
    },
    {
      "epoch": 0.03625183293127099,
      "grad_norm": 0.98828125,
      "learning_rate": 1.9986948153100566e-05,
      "loss": 0.3983,
      "step": 479
    },
    {
      "epoch": 0.03632751525471832,
      "grad_norm": 1.0859375,
      "learning_rate": 1.998688724772834e-05,
      "loss": 0.4037,
      "step": 480
    },
    {
      "epoch": 0.03640319757816565,
      "grad_norm": 1.0625,
      "learning_rate": 1.9986826200674974e-05,
      "loss": 0.3783,
      "step": 481
    },
    {
      "epoch": 0.03647887990161298,
      "grad_norm": 1.0625,
      "learning_rate": 1.9986765011941346e-05,
      "loss": 0.3685,
      "step": 482
    },
    {
      "epoch": 0.03655456222506031,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9986703681528318e-05,
      "loss": 0.3238,
      "step": 483
    },
    {
      "epoch": 0.03663024454850764,
      "grad_norm": 1.078125,
      "learning_rate": 1.9986642209436758e-05,
      "loss": 0.3574,
      "step": 484
    },
    {
      "epoch": 0.036705926871954966,
      "grad_norm": 1.09375,
      "learning_rate": 1.9986580595667543e-05,
      "loss": 0.3606,
      "step": 485
    },
    {
      "epoch": 0.0367816091954023,
      "grad_norm": 9.9375,
      "learning_rate": 1.9986518840221544e-05,
      "loss": 0.7674,
      "step": 486
    },
    {
      "epoch": 0.03685729151884963,
      "grad_norm": 1.0,
      "learning_rate": 1.998645694309964e-05,
      "loss": 0.3646,
      "step": 487
    },
    {
      "epoch": 0.036932973842296955,
      "grad_norm": 1.0078125,
      "learning_rate": 1.99863949043027e-05,
      "loss": 0.368,
      "step": 488
    },
    {
      "epoch": 0.03700865616574429,
      "grad_norm": 1.0390625,
      "learning_rate": 1.9986332723831615e-05,
      "loss": 0.3912,
      "step": 489
    },
    {
      "epoch": 0.03708433848919162,
      "grad_norm": 1.1015625,
      "learning_rate": 1.9986270401687266e-05,
      "loss": 0.4062,
      "step": 490
    },
    {
      "epoch": 0.03716002081263895,
      "grad_norm": 1.046875,
      "learning_rate": 1.998620793787053e-05,
      "loss": 0.421,
      "step": 491
    },
    {
      "epoch": 0.037235703136086276,
      "grad_norm": 0.8984375,
      "learning_rate": 1.99861453323823e-05,
      "loss": 0.3463,
      "step": 492
    },
    {
      "epoch": 0.03731138545953361,
      "grad_norm": 0.953125,
      "learning_rate": 1.998608258522346e-05,
      "loss": 0.3751,
      "step": 493
    },
    {
      "epoch": 0.03738706778298094,
      "grad_norm": 1.0234375,
      "learning_rate": 1.9986019696394904e-05,
      "loss": 0.3682,
      "step": 494
    },
    {
      "epoch": 0.037462750106428265,
      "grad_norm": 1.0234375,
      "learning_rate": 1.9985956665897517e-05,
      "loss": 0.3887,
      "step": 495
    },
    {
      "epoch": 0.0375384324298756,
      "grad_norm": 0.98046875,
      "learning_rate": 1.9985893493732204e-05,
      "loss": 0.34,
      "step": 496
    },
    {
      "epoch": 0.03761411475332293,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9985830179899854e-05,
      "loss": 0.3308,
      "step": 497
    },
    {
      "epoch": 0.037689797076770254,
      "grad_norm": 1.125,
      "learning_rate": 1.9985766724401365e-05,
      "loss": 0.3657,
      "step": 498
    },
    {
      "epoch": 0.037765479400217586,
      "grad_norm": 0.98828125,
      "learning_rate": 1.9985703127237643e-05,
      "loss": 0.367,
      "step": 499
    },
    {
      "epoch": 0.03784116172366492,
      "grad_norm": 0.91015625,
      "learning_rate": 1.998563938840958e-05,
      "loss": 0.3113,
      "step": 500
    },
    {
      "epoch": 0.03791684404711225,
      "grad_norm": 1.0703125,
      "learning_rate": 1.998557550791809e-05,
      "loss": 0.3587,
      "step": 501
    },
    {
      "epoch": 0.037992526370559575,
      "grad_norm": 0.9921875,
      "learning_rate": 1.9985511485764075e-05,
      "loss": 0.3151,
      "step": 502
    },
    {
      "epoch": 0.03806820869400691,
      "grad_norm": 1.125,
      "learning_rate": 1.9985447321948443e-05,
      "loss": 0.3864,
      "step": 503
    },
    {
      "epoch": 0.03814389101745424,
      "grad_norm": 1.0,
      "learning_rate": 1.9985383016472108e-05,
      "loss": 0.3827,
      "step": 504
    },
    {
      "epoch": 0.038219573340901564,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9985318569335976e-05,
      "loss": 0.3088,
      "step": 505
    },
    {
      "epoch": 0.038295255664348896,
      "grad_norm": 0.99609375,
      "learning_rate": 1.9985253980540967e-05,
      "loss": 0.3707,
      "step": 506
    },
    {
      "epoch": 0.03837093798779623,
      "grad_norm": 0.98828125,
      "learning_rate": 1.9985189250087992e-05,
      "loss": 0.3577,
      "step": 507
    },
    {
      "epoch": 0.03844662031124355,
      "grad_norm": 1.015625,
      "learning_rate": 1.9985124377977975e-05,
      "loss": 0.3881,
      "step": 508
    },
    {
      "epoch": 0.038522302634690885,
      "grad_norm": 0.98828125,
      "learning_rate": 1.9985059364211832e-05,
      "loss": 0.3535,
      "step": 509
    },
    {
      "epoch": 0.038597984958138216,
      "grad_norm": 0.94140625,
      "learning_rate": 1.998499420879049e-05,
      "loss": 0.3178,
      "step": 510
    },
    {
      "epoch": 0.03867366728158554,
      "grad_norm": 1.0625,
      "learning_rate": 1.9984928911714866e-05,
      "loss": 0.3641,
      "step": 511
    },
    {
      "epoch": 0.03874934960503287,
      "grad_norm": 0.99609375,
      "learning_rate": 1.9984863472985892e-05,
      "loss": 0.3771,
      "step": 512
    },
    {
      "epoch": 0.038825031928480205,
      "grad_norm": 1.0,
      "learning_rate": 1.9984797892604496e-05,
      "loss": 0.4062,
      "step": 513
    },
    {
      "epoch": 0.03890071425192754,
      "grad_norm": 0.96875,
      "learning_rate": 1.998473217057161e-05,
      "loss": 0.3572,
      "step": 514
    },
    {
      "epoch": 0.03897639657537486,
      "grad_norm": 0.94921875,
      "learning_rate": 1.998466630688816e-05,
      "loss": 0.3158,
      "step": 515
    },
    {
      "epoch": 0.039052078898822194,
      "grad_norm": 1.03125,
      "learning_rate": 1.9984600301555085e-05,
      "loss": 0.354,
      "step": 516
    },
    {
      "epoch": 0.039127761222269526,
      "grad_norm": 6.0625,
      "learning_rate": 1.998453415457332e-05,
      "loss": 0.8184,
      "step": 517
    },
    {
      "epoch": 0.03920344354571685,
      "grad_norm": 1.0,
      "learning_rate": 1.9984467865943803e-05,
      "loss": 0.3529,
      "step": 518
    },
    {
      "epoch": 0.03927912586916418,
      "grad_norm": 3.96875,
      "learning_rate": 1.998440143566748e-05,
      "loss": 0.7833,
      "step": 519
    },
    {
      "epoch": 0.039354808192611515,
      "grad_norm": 0.9765625,
      "learning_rate": 1.9984334863745284e-05,
      "loss": 0.3461,
      "step": 520
    },
    {
      "epoch": 0.03943049051605884,
      "grad_norm": 0.98828125,
      "learning_rate": 1.998426815017817e-05,
      "loss": 0.3434,
      "step": 521
    },
    {
      "epoch": 0.03950617283950617,
      "grad_norm": 1.0234375,
      "learning_rate": 1.998420129496707e-05,
      "loss": 0.3631,
      "step": 522
    },
    {
      "epoch": 0.039581855162953504,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9984134298112948e-05,
      "loss": 0.2847,
      "step": 523
    },
    {
      "epoch": 0.039657537486400836,
      "grad_norm": 0.9453125,
      "learning_rate": 1.9984067159616746e-05,
      "loss": 0.3528,
      "step": 524
    },
    {
      "epoch": 0.03973321980984816,
      "grad_norm": 0.98828125,
      "learning_rate": 1.998399987947942e-05,
      "loss": 0.3633,
      "step": 525
    },
    {
      "epoch": 0.03980890213329549,
      "grad_norm": 1.015625,
      "learning_rate": 1.9983932457701918e-05,
      "loss": 0.3944,
      "step": 526
    },
    {
      "epoch": 0.039884584456742825,
      "grad_norm": 1.0,
      "learning_rate": 1.9983864894285204e-05,
      "loss": 0.332,
      "step": 527
    },
    {
      "epoch": 0.03996026678019015,
      "grad_norm": 0.9453125,
      "learning_rate": 1.9983797189230232e-05,
      "loss": 0.3522,
      "step": 528
    },
    {
      "epoch": 0.04003594910363748,
      "grad_norm": 1.03125,
      "learning_rate": 1.9983729342537965e-05,
      "loss": 0.3353,
      "step": 529
    },
    {
      "epoch": 0.040111631427084814,
      "grad_norm": 0.96484375,
      "learning_rate": 1.9983661354209365e-05,
      "loss": 0.3008,
      "step": 530
    },
    {
      "epoch": 0.04018731375053214,
      "grad_norm": 0.9921875,
      "learning_rate": 1.9983593224245393e-05,
      "loss": 0.3544,
      "step": 531
    },
    {
      "epoch": 0.04026299607397947,
      "grad_norm": 0.94921875,
      "learning_rate": 1.9983524952647024e-05,
      "loss": 0.3672,
      "step": 532
    },
    {
      "epoch": 0.0403386783974268,
      "grad_norm": 1.0,
      "learning_rate": 1.9983456539415218e-05,
      "loss": 0.396,
      "step": 533
    },
    {
      "epoch": 0.04041436072087413,
      "grad_norm": 1.0546875,
      "learning_rate": 1.998338798455095e-05,
      "loss": 0.3994,
      "step": 534
    },
    {
      "epoch": 0.04049004304432146,
      "grad_norm": 0.953125,
      "learning_rate": 1.998331928805519e-05,
      "loss": 0.3544,
      "step": 535
    },
    {
      "epoch": 0.04056572536776879,
      "grad_norm": 1.015625,
      "learning_rate": 1.9983250449928915e-05,
      "loss": 0.3955,
      "step": 536
    },
    {
      "epoch": 0.04064140769121612,
      "grad_norm": 0.95703125,
      "learning_rate": 1.99831814701731e-05,
      "loss": 0.3516,
      "step": 537
    },
    {
      "epoch": 0.04071709001466345,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9983112348788725e-05,
      "loss": 0.3386,
      "step": 538
    },
    {
      "epoch": 0.04079277233811078,
      "grad_norm": 1.15625,
      "learning_rate": 1.998304308577677e-05,
      "loss": 0.3939,
      "step": 539
    },
    {
      "epoch": 0.04086845466155811,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9982973681138215e-05,
      "loss": 0.32,
      "step": 540
    },
    {
      "epoch": 0.04094413698500544,
      "grad_norm": 4.46875,
      "learning_rate": 1.9982904134874048e-05,
      "loss": 0.775,
      "step": 541
    },
    {
      "epoch": 0.04101981930845277,
      "grad_norm": 1.03125,
      "learning_rate": 1.9982834446985257e-05,
      "loss": 0.3445,
      "step": 542
    },
    {
      "epoch": 0.0410955016319001,
      "grad_norm": 1.0234375,
      "learning_rate": 1.9982764617472825e-05,
      "loss": 0.3818,
      "step": 543
    },
    {
      "epoch": 0.041171183955347426,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9982694646337748e-05,
      "loss": 0.3759,
      "step": 544
    },
    {
      "epoch": 0.04124686627879476,
      "grad_norm": 0.953125,
      "learning_rate": 1.9982624533581015e-05,
      "loss": 0.3566,
      "step": 545
    },
    {
      "epoch": 0.04132254860224209,
      "grad_norm": 0.95703125,
      "learning_rate": 1.9982554279203623e-05,
      "loss": 0.3193,
      "step": 546
    },
    {
      "epoch": 0.04139823092568942,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9982483883206567e-05,
      "loss": 0.3182,
      "step": 547
    },
    {
      "epoch": 0.04147391324913675,
      "grad_norm": 1.0703125,
      "learning_rate": 1.9982413345590848e-05,
      "loss": 0.3686,
      "step": 548
    },
    {
      "epoch": 0.04154959557258408,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9982342666357467e-05,
      "loss": 0.2526,
      "step": 549
    },
    {
      "epoch": 0.04162527789603141,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9982271845507418e-05,
      "loss": 0.3536,
      "step": 550
    },
    {
      "epoch": 0.041700960219478736,
      "grad_norm": 0.953125,
      "learning_rate": 1.998220088304172e-05,
      "loss": 0.3398,
      "step": 551
    },
    {
      "epoch": 0.04177664254292607,
      "grad_norm": 1.0,
      "learning_rate": 1.9982129778961367e-05,
      "loss": 0.335,
      "step": 552
    },
    {
      "epoch": 0.0418523248663734,
      "grad_norm": 3.78125,
      "learning_rate": 1.9982058533267378e-05,
      "loss": 0.633,
      "step": 553
    },
    {
      "epoch": 0.041928007189820725,
      "grad_norm": 0.953125,
      "learning_rate": 1.998198714596076e-05,
      "loss": 0.3236,
      "step": 554
    },
    {
      "epoch": 0.04200368951326806,
      "grad_norm": 0.96875,
      "learning_rate": 1.998191561704252e-05,
      "loss": 0.3395,
      "step": 555
    },
    {
      "epoch": 0.04207937183671539,
      "grad_norm": 0.96875,
      "learning_rate": 1.9981843946513677e-05,
      "loss": 0.3603,
      "step": 556
    },
    {
      "epoch": 0.042155054160162714,
      "grad_norm": 1.0390625,
      "learning_rate": 1.998177213437525e-05,
      "loss": 0.3889,
      "step": 557
    },
    {
      "epoch": 0.042230736483610046,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9981700180628257e-05,
      "loss": 0.2679,
      "step": 558
    },
    {
      "epoch": 0.04230641880705738,
      "grad_norm": 1.0,
      "learning_rate": 1.9981628085273716e-05,
      "loss": 0.3696,
      "step": 559
    },
    {
      "epoch": 0.04238210113050471,
      "grad_norm": 1.0546875,
      "learning_rate": 1.9981555848312654e-05,
      "loss": 0.3372,
      "step": 560
    },
    {
      "epoch": 0.042457783453952035,
      "grad_norm": 1.0234375,
      "learning_rate": 1.998148346974609e-05,
      "loss": 0.3693,
      "step": 561
    },
    {
      "epoch": 0.04253346577739937,
      "grad_norm": 1.0,
      "learning_rate": 1.9981410949575056e-05,
      "loss": 0.3017,
      "step": 562
    },
    {
      "epoch": 0.0426091481008467,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9981338287800578e-05,
      "loss": 0.331,
      "step": 563
    },
    {
      "epoch": 0.042684830424294024,
      "grad_norm": 1.0078125,
      "learning_rate": 1.998126548442369e-05,
      "loss": 0.3677,
      "step": 564
    },
    {
      "epoch": 0.042760512747741355,
      "grad_norm": 0.9765625,
      "learning_rate": 1.998119253944542e-05,
      "loss": 0.3644,
      "step": 565
    },
    {
      "epoch": 0.04283619507118869,
      "grad_norm": 1.046875,
      "learning_rate": 1.998111945286681e-05,
      "loss": 0.3525,
      "step": 566
    },
    {
      "epoch": 0.04291187739463601,
      "grad_norm": 1.1640625,
      "learning_rate": 1.9981046224688886e-05,
      "loss": 0.3345,
      "step": 567
    },
    {
      "epoch": 0.042987559718083344,
      "grad_norm": 0.984375,
      "learning_rate": 1.99809728549127e-05,
      "loss": 0.3809,
      "step": 568
    },
    {
      "epoch": 0.043063242041530676,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9980899343539278e-05,
      "loss": 0.3317,
      "step": 569
    },
    {
      "epoch": 0.04313892436497801,
      "grad_norm": 1.0703125,
      "learning_rate": 1.9980825690569677e-05,
      "loss": 0.3782,
      "step": 570
    },
    {
      "epoch": 0.04321460668842533,
      "grad_norm": 1.0390625,
      "learning_rate": 1.9980751896004936e-05,
      "loss": 0.366,
      "step": 571
    },
    {
      "epoch": 0.043290289011872665,
      "grad_norm": 0.98046875,
      "learning_rate": 1.99806779598461e-05,
      "loss": 0.3711,
      "step": 572
    },
    {
      "epoch": 0.04336597133532,
      "grad_norm": 0.9921875,
      "learning_rate": 1.9980603882094217e-05,
      "loss": 0.3826,
      "step": 573
    },
    {
      "epoch": 0.04344165365876732,
      "grad_norm": 0.9765625,
      "learning_rate": 1.9980529662750347e-05,
      "loss": 0.3458,
      "step": 574
    },
    {
      "epoch": 0.043517335982214654,
      "grad_norm": 0.9765625,
      "learning_rate": 1.9980455301815532e-05,
      "loss": 0.3253,
      "step": 575
    },
    {
      "epoch": 0.043593018305661986,
      "grad_norm": 0.9921875,
      "learning_rate": 1.9980380799290834e-05,
      "loss": 0.3628,
      "step": 576
    },
    {
      "epoch": 0.04366870062910931,
      "grad_norm": 1.078125,
      "learning_rate": 1.9980306155177306e-05,
      "loss": 0.3591,
      "step": 577
    },
    {
      "epoch": 0.04374438295255664,
      "grad_norm": 1.03125,
      "learning_rate": 1.998023136947601e-05,
      "loss": 0.3509,
      "step": 578
    },
    {
      "epoch": 0.043820065276003975,
      "grad_norm": 1.0234375,
      "learning_rate": 1.9980156442188002e-05,
      "loss": 0.345,
      "step": 579
    },
    {
      "epoch": 0.0438957475994513,
      "grad_norm": 1.0703125,
      "learning_rate": 1.998008137331435e-05,
      "loss": 0.3932,
      "step": 580
    },
    {
      "epoch": 0.04397142992289863,
      "grad_norm": 0.91796875,
      "learning_rate": 1.998000616285612e-05,
      "loss": 0.3204,
      "step": 581
    },
    {
      "epoch": 0.044047112246345964,
      "grad_norm": 0.88671875,
      "learning_rate": 1.997993081081437e-05,
      "loss": 0.297,
      "step": 582
    },
    {
      "epoch": 0.044122794569793296,
      "grad_norm": 1.015625,
      "learning_rate": 1.997985531719018e-05,
      "loss": 0.3705,
      "step": 583
    },
    {
      "epoch": 0.04419847689324062,
      "grad_norm": 1.125,
      "learning_rate": 1.9979779681984617e-05,
      "loss": 0.4036,
      "step": 584
    },
    {
      "epoch": 0.04427415921668795,
      "grad_norm": 0.96484375,
      "learning_rate": 1.997970390519875e-05,
      "loss": 0.3508,
      "step": 585
    },
    {
      "epoch": 0.044349841540135285,
      "grad_norm": 1.0859375,
      "learning_rate": 1.997962798683366e-05,
      "loss": 0.3975,
      "step": 586
    },
    {
      "epoch": 0.04442552386358261,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9979551926890418e-05,
      "loss": 0.3105,
      "step": 587
    },
    {
      "epoch": 0.04450120618702994,
      "grad_norm": 1.015625,
      "learning_rate": 1.997947572537011e-05,
      "loss": 0.3566,
      "step": 588
    },
    {
      "epoch": 0.044576888510477274,
      "grad_norm": 1.046875,
      "learning_rate": 1.9979399382273814e-05,
      "loss": 0.3635,
      "step": 589
    },
    {
      "epoch": 0.0446525708339246,
      "grad_norm": 0.9453125,
      "learning_rate": 1.997932289760261e-05,
      "loss": 0.3219,
      "step": 590
    },
    {
      "epoch": 0.04472825315737193,
      "grad_norm": 1.0546875,
      "learning_rate": 1.9979246271357585e-05,
      "loss": 0.3948,
      "step": 591
    },
    {
      "epoch": 0.04480393548081926,
      "grad_norm": 1.0546875,
      "learning_rate": 1.997916950353983e-05,
      "loss": 0.3301,
      "step": 592
    },
    {
      "epoch": 0.044879617804266594,
      "grad_norm": 0.93359375,
      "learning_rate": 1.997909259415043e-05,
      "loss": 0.3088,
      "step": 593
    },
    {
      "epoch": 0.04495530012771392,
      "grad_norm": 0.96875,
      "learning_rate": 1.9979015543190473e-05,
      "loss": 0.3779,
      "step": 594
    },
    {
      "epoch": 0.04503098245116125,
      "grad_norm": 1.15625,
      "learning_rate": 1.9978938350661062e-05,
      "loss": 0.3463,
      "step": 595
    },
    {
      "epoch": 0.04510666477460858,
      "grad_norm": 0.9921875,
      "learning_rate": 1.997886101656328e-05,
      "loss": 0.352,
      "step": 596
    },
    {
      "epoch": 0.04518234709805591,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9978783540898234e-05,
      "loss": 0.3438,
      "step": 597
    },
    {
      "epoch": 0.04525802942150324,
      "grad_norm": 0.96875,
      "learning_rate": 1.997870592366702e-05,
      "loss": 0.3476,
      "step": 598
    },
    {
      "epoch": 0.04533371174495057,
      "grad_norm": 0.984375,
      "learning_rate": 1.9978628164870733e-05,
      "loss": 0.3411,
      "step": 599
    },
    {
      "epoch": 0.0454093940683979,
      "grad_norm": 1.0,
      "learning_rate": 1.9978550264510485e-05,
      "loss": 0.3547,
      "step": 600
    },
    {
      "epoch": 0.04548507639184523,
      "grad_norm": 0.89453125,
      "learning_rate": 1.997847222258738e-05,
      "loss": 0.3196,
      "step": 601
    },
    {
      "epoch": 0.04556075871529256,
      "grad_norm": 0.98828125,
      "learning_rate": 1.997839403910252e-05,
      "loss": 0.3772,
      "step": 602
    },
    {
      "epoch": 0.045636441038739886,
      "grad_norm": 0.98828125,
      "learning_rate": 1.9978315714057017e-05,
      "loss": 0.3854,
      "step": 603
    },
    {
      "epoch": 0.04571212336218722,
      "grad_norm": 0.98046875,
      "learning_rate": 1.9978237247451984e-05,
      "loss": 0.3589,
      "step": 604
    },
    {
      "epoch": 0.04578780568563455,
      "grad_norm": 3.09375,
      "learning_rate": 1.997815863928853e-05,
      "loss": 0.7429,
      "step": 605
    },
    {
      "epoch": 0.04586348800908188,
      "grad_norm": 0.9921875,
      "learning_rate": 1.9978079889567774e-05,
      "loss": 0.3837,
      "step": 606
    },
    {
      "epoch": 0.04593917033252921,
      "grad_norm": 0.95703125,
      "learning_rate": 1.9978000998290833e-05,
      "loss": 0.3871,
      "step": 607
    },
    {
      "epoch": 0.04601485265597654,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9977921965458824e-05,
      "loss": 0.3267,
      "step": 608
    },
    {
      "epoch": 0.04609053497942387,
      "grad_norm": 0.96484375,
      "learning_rate": 1.997784279107287e-05,
      "loss": 0.335,
      "step": 609
    },
    {
      "epoch": 0.046166217302871196,
      "grad_norm": 0.94921875,
      "learning_rate": 1.997776347513409e-05,
      "loss": 0.3389,
      "step": 610
    },
    {
      "epoch": 0.04624189962631853,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9977684017643618e-05,
      "loss": 0.331,
      "step": 611
    },
    {
      "epoch": 0.04631758194976586,
      "grad_norm": 0.984375,
      "learning_rate": 1.997760441860257e-05,
      "loss": 0.3759,
      "step": 612
    },
    {
      "epoch": 0.046393264273213185,
      "grad_norm": 1.03125,
      "learning_rate": 1.9977524678012085e-05,
      "loss": 0.3465,
      "step": 613
    },
    {
      "epoch": 0.04646894659666052,
      "grad_norm": 0.9921875,
      "learning_rate": 1.9977444795873292e-05,
      "loss": 0.3331,
      "step": 614
    },
    {
      "epoch": 0.04654462892010785,
      "grad_norm": 0.9140625,
      "learning_rate": 1.997736477218732e-05,
      "loss": 0.3442,
      "step": 615
    },
    {
      "epoch": 0.04662031124355518,
      "grad_norm": 0.95703125,
      "learning_rate": 1.9977284606955305e-05,
      "loss": 0.3272,
      "step": 616
    },
    {
      "epoch": 0.046695993567002506,
      "grad_norm": 1.0078125,
      "learning_rate": 1.997720430017839e-05,
      "loss": 0.3522,
      "step": 617
    },
    {
      "epoch": 0.04677167589044984,
      "grad_norm": 1.0625,
      "learning_rate": 1.997712385185771e-05,
      "loss": 0.3632,
      "step": 618
    },
    {
      "epoch": 0.04684735821389717,
      "grad_norm": 0.9375,
      "learning_rate": 1.9977043261994405e-05,
      "loss": 0.3765,
      "step": 619
    },
    {
      "epoch": 0.046923040537344494,
      "grad_norm": 0.89453125,
      "learning_rate": 1.997696253058962e-05,
      "loss": 0.3032,
      "step": 620
    },
    {
      "epoch": 0.046998722860791826,
      "grad_norm": 12.0,
      "learning_rate": 1.99768816576445e-05,
      "loss": 0.7095,
      "step": 621
    },
    {
      "epoch": 0.04707440518423916,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9976800643160196e-05,
      "loss": 0.3323,
      "step": 622
    },
    {
      "epoch": 0.04715008750768648,
      "grad_norm": 0.9765625,
      "learning_rate": 1.9976719487137854e-05,
      "loss": 0.3805,
      "step": 623
    },
    {
      "epoch": 0.047225769831133815,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9976638189578623e-05,
      "loss": 0.3354,
      "step": 624
    },
    {
      "epoch": 0.04730145215458115,
      "grad_norm": 0.96484375,
      "learning_rate": 1.9976556750483657e-05,
      "loss": 0.3668,
      "step": 625
    },
    {
      "epoch": 0.04737713447802847,
      "grad_norm": 1.109375,
      "learning_rate": 1.9976475169854117e-05,
      "loss": 0.4031,
      "step": 626
    },
    {
      "epoch": 0.047452816801475804,
      "grad_norm": 0.9375,
      "learning_rate": 1.9976393447691155e-05,
      "loss": 0.3179,
      "step": 627
    },
    {
      "epoch": 0.047528499124923136,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9976311583995933e-05,
      "loss": 0.325,
      "step": 628
    },
    {
      "epoch": 0.04760418144837047,
      "grad_norm": 0.94140625,
      "learning_rate": 1.997622957876961e-05,
      "loss": 0.3471,
      "step": 629
    },
    {
      "epoch": 0.04767986377181779,
      "grad_norm": 0.91796875,
      "learning_rate": 1.997614743201335e-05,
      "loss": 0.3324,
      "step": 630
    },
    {
      "epoch": 0.047755546095265125,
      "grad_norm": 1.25,
      "learning_rate": 1.997606514372832e-05,
      "loss": 0.3747,
      "step": 631
    },
    {
      "epoch": 0.04783122841871246,
      "grad_norm": 1.1328125,
      "learning_rate": 1.9975982713915686e-05,
      "loss": 0.3654,
      "step": 632
    },
    {
      "epoch": 0.04790691074215978,
      "grad_norm": 0.99609375,
      "learning_rate": 1.9975900142576617e-05,
      "loss": 0.4023,
      "step": 633
    },
    {
      "epoch": 0.047982593065607114,
      "grad_norm": 1.0234375,
      "learning_rate": 1.997581742971229e-05,
      "loss": 0.3849,
      "step": 634
    },
    {
      "epoch": 0.048058275389054446,
      "grad_norm": 0.91796875,
      "learning_rate": 1.997573457532387e-05,
      "loss": 0.3286,
      "step": 635
    },
    {
      "epoch": 0.04813395771250177,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9975651579412534e-05,
      "loss": 0.3205,
      "step": 636
    },
    {
      "epoch": 0.0482096400359491,
      "grad_norm": 1.0546875,
      "learning_rate": 1.9975568441979462e-05,
      "loss": 0.4286,
      "step": 637
    },
    {
      "epoch": 0.048285322359396435,
      "grad_norm": 0.94921875,
      "learning_rate": 1.9975485163025837e-05,
      "loss": 0.3399,
      "step": 638
    },
    {
      "epoch": 0.04836100468284377,
      "grad_norm": 0.95703125,
      "learning_rate": 1.9975401742552834e-05,
      "loss": 0.3474,
      "step": 639
    },
    {
      "epoch": 0.04843668700629109,
      "grad_norm": 1.0234375,
      "learning_rate": 1.9975318180561638e-05,
      "loss": 0.3448,
      "step": 640
    },
    {
      "epoch": 0.048512369329738424,
      "grad_norm": 0.9609375,
      "learning_rate": 1.997523447705344e-05,
      "loss": 0.3819,
      "step": 641
    },
    {
      "epoch": 0.048588051653185756,
      "grad_norm": 1.046875,
      "learning_rate": 1.9975150632029417e-05,
      "loss": 0.3947,
      "step": 642
    },
    {
      "epoch": 0.04866373397663308,
      "grad_norm": 1.015625,
      "learning_rate": 1.9975066645490766e-05,
      "loss": 0.3544,
      "step": 643
    },
    {
      "epoch": 0.04873941630008041,
      "grad_norm": 1.0078125,
      "learning_rate": 1.997498251743868e-05,
      "loss": 0.3741,
      "step": 644
    },
    {
      "epoch": 0.048815098623527745,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9974898247874345e-05,
      "loss": 0.3593,
      "step": 645
    },
    {
      "epoch": 0.04889078094697507,
      "grad_norm": 1.0,
      "learning_rate": 1.997481383679896e-05,
      "loss": 0.3965,
      "step": 646
    },
    {
      "epoch": 0.0489664632704224,
      "grad_norm": 0.9453125,
      "learning_rate": 1.997472928421373e-05,
      "loss": 0.3509,
      "step": 647
    },
    {
      "epoch": 0.04904214559386973,
      "grad_norm": 1.0,
      "learning_rate": 1.9974644590119842e-05,
      "loss": 0.3865,
      "step": 648
    },
    {
      "epoch": 0.04911782791731706,
      "grad_norm": 2.03125,
      "learning_rate": 1.9974559754518506e-05,
      "loss": 0.6595,
      "step": 649
    },
    {
      "epoch": 0.04919351024076439,
      "grad_norm": 2.046875,
      "learning_rate": 1.997447477741092e-05,
      "loss": 0.7222,
      "step": 650
    },
    {
      "epoch": 0.04926919256421172,
      "grad_norm": 1.078125,
      "learning_rate": 1.9974389658798296e-05,
      "loss": 0.3729,
      "step": 651
    },
    {
      "epoch": 0.049344874887659054,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9974304398681837e-05,
      "loss": 0.3121,
      "step": 652
    },
    {
      "epoch": 0.04942055721110638,
      "grad_norm": 0.97265625,
      "learning_rate": 1.9974218997062752e-05,
      "loss": 0.3384,
      "step": 653
    },
    {
      "epoch": 0.04949623953455371,
      "grad_norm": 0.98046875,
      "learning_rate": 1.9974133453942256e-05,
      "loss": 0.3714,
      "step": 654
    },
    {
      "epoch": 0.04957192185800104,
      "grad_norm": 0.98046875,
      "learning_rate": 1.997404776932156e-05,
      "loss": 0.3566,
      "step": 655
    },
    {
      "epoch": 0.04964760418144837,
      "grad_norm": 0.96484375,
      "learning_rate": 1.9973961943201882e-05,
      "loss": 0.3329,
      "step": 656
    },
    {
      "epoch": 0.0497232865048957,
      "grad_norm": 0.94921875,
      "learning_rate": 1.9973875975584434e-05,
      "loss": 0.373,
      "step": 657
    },
    {
      "epoch": 0.04979896882834303,
      "grad_norm": 0.9375,
      "learning_rate": 1.9973789866470444e-05,
      "loss": 0.3513,
      "step": 658
    },
    {
      "epoch": 0.04987465115179036,
      "grad_norm": 1.0546875,
      "learning_rate": 1.997370361586113e-05,
      "loss": 0.379,
      "step": 659
    },
    {
      "epoch": 0.04995033347523769,
      "grad_norm": 0.9375,
      "learning_rate": 1.997361722375771e-05,
      "loss": 0.3525,
      "step": 660
    },
    {
      "epoch": 0.05002601579868502,
      "grad_norm": 0.96484375,
      "learning_rate": 1.9973530690161415e-05,
      "loss": 0.3668,
      "step": 661
    },
    {
      "epoch": 0.05010169812213235,
      "grad_norm": 0.921875,
      "learning_rate": 1.9973444015073474e-05,
      "loss": 0.3448,
      "step": 662
    },
    {
      "epoch": 0.05017738044557968,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9973357198495114e-05,
      "loss": 0.3359,
      "step": 663
    },
    {
      "epoch": 0.05025306276902701,
      "grad_norm": 0.94921875,
      "learning_rate": 1.9973270240427566e-05,
      "loss": 0.3524,
      "step": 664
    },
    {
      "epoch": 0.05032874509247434,
      "grad_norm": 3.5,
      "learning_rate": 1.997318314087207e-05,
      "loss": 0.7076,
      "step": 665
    },
    {
      "epoch": 0.05040442741592167,
      "grad_norm": 1.140625,
      "learning_rate": 1.997309589982985e-05,
      "loss": 0.4001,
      "step": 666
    },
    {
      "epoch": 0.050480109739369,
      "grad_norm": 1.1171875,
      "learning_rate": 1.9973008517302156e-05,
      "loss": 0.3966,
      "step": 667
    },
    {
      "epoch": 0.05055579206281633,
      "grad_norm": 0.890625,
      "learning_rate": 1.9972920993290216e-05,
      "loss": 0.3251,
      "step": 668
    },
    {
      "epoch": 0.050631474386263656,
      "grad_norm": 1.078125,
      "learning_rate": 1.997283332779528e-05,
      "loss": 0.3685,
      "step": 669
    },
    {
      "epoch": 0.05070715670971099,
      "grad_norm": 1.1328125,
      "learning_rate": 1.997274552081859e-05,
      "loss": 0.3826,
      "step": 670
    },
    {
      "epoch": 0.05078283903315832,
      "grad_norm": 1.0078125,
      "learning_rate": 1.997265757236139e-05,
      "loss": 0.3484,
      "step": 671
    },
    {
      "epoch": 0.050858521356605645,
      "grad_norm": 3.046875,
      "learning_rate": 1.997256948242493e-05,
      "loss": 0.5781,
      "step": 672
    },
    {
      "epoch": 0.05093420368005298,
      "grad_norm": 0.96875,
      "learning_rate": 1.997248125101046e-05,
      "loss": 0.3697,
      "step": 673
    },
    {
      "epoch": 0.05100988600350031,
      "grad_norm": 1.0,
      "learning_rate": 1.9972392878119226e-05,
      "loss": 0.3601,
      "step": 674
    },
    {
      "epoch": 0.05108556832694764,
      "grad_norm": 0.94921875,
      "learning_rate": 1.9972304363752488e-05,
      "loss": 0.3388,
      "step": 675
    },
    {
      "epoch": 0.051161250650394965,
      "grad_norm": 0.97265625,
      "learning_rate": 1.99722157079115e-05,
      "loss": 0.3548,
      "step": 676
    },
    {
      "epoch": 0.0512369329738423,
      "grad_norm": 0.984375,
      "learning_rate": 1.9972126910597518e-05,
      "loss": 0.3378,
      "step": 677
    },
    {
      "epoch": 0.05131261529728963,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9972037971811802e-05,
      "loss": 0.3634,
      "step": 678
    },
    {
      "epoch": 0.051388297620736954,
      "grad_norm": 0.96484375,
      "learning_rate": 1.9971948891555617e-05,
      "loss": 0.3471,
      "step": 679
    },
    {
      "epoch": 0.051463979944184286,
      "grad_norm": 0.98828125,
      "learning_rate": 1.997185966983022e-05,
      "loss": 0.3782,
      "step": 680
    },
    {
      "epoch": 0.05153966226763162,
      "grad_norm": 0.98046875,
      "learning_rate": 1.997177030663689e-05,
      "loss": 0.3359,
      "step": 681
    },
    {
      "epoch": 0.05161534459107894,
      "grad_norm": 1.0,
      "learning_rate": 1.9971680801976882e-05,
      "loss": 0.3719,
      "step": 682
    },
    {
      "epoch": 0.051691026914526275,
      "grad_norm": 3.4375,
      "learning_rate": 1.997159115585147e-05,
      "loss": 0.7382,
      "step": 683
    },
    {
      "epoch": 0.05176670923797361,
      "grad_norm": 0.98046875,
      "learning_rate": 1.9971501368261922e-05,
      "loss": 0.386,
      "step": 684
    },
    {
      "epoch": 0.05184239156142094,
      "grad_norm": 1.0703125,
      "learning_rate": 1.9971411439209517e-05,
      "loss": 0.3596,
      "step": 685
    },
    {
      "epoch": 0.051918073884868264,
      "grad_norm": 0.921875,
      "learning_rate": 1.997132136869553e-05,
      "loss": 0.3097,
      "step": 686
    },
    {
      "epoch": 0.051993756208315596,
      "grad_norm": 1.046875,
      "learning_rate": 1.997123115672124e-05,
      "loss": 0.3735,
      "step": 687
    },
    {
      "epoch": 0.05206943853176293,
      "grad_norm": 0.9453125,
      "learning_rate": 1.997114080328792e-05,
      "loss": 0.3549,
      "step": 688
    },
    {
      "epoch": 0.05214512085521025,
      "grad_norm": 0.9375,
      "learning_rate": 1.9971050308396862e-05,
      "loss": 0.3107,
      "step": 689
    },
    {
      "epoch": 0.052220803178657585,
      "grad_norm": 1.015625,
      "learning_rate": 1.997095967204934e-05,
      "loss": 0.3893,
      "step": 690
    },
    {
      "epoch": 0.05229648550210492,
      "grad_norm": 0.97265625,
      "learning_rate": 1.9970868894246646e-05,
      "loss": 0.3742,
      "step": 691
    },
    {
      "epoch": 0.05237216782555224,
      "grad_norm": 1.0234375,
      "learning_rate": 1.997077797499007e-05,
      "loss": 0.3691,
      "step": 692
    },
    {
      "epoch": 0.052447850148999574,
      "grad_norm": 0.97265625,
      "learning_rate": 1.9970686914280892e-05,
      "loss": 0.3697,
      "step": 693
    },
    {
      "epoch": 0.052523532472446906,
      "grad_norm": 3.203125,
      "learning_rate": 1.9970595712120414e-05,
      "loss": 0.6165,
      "step": 694
    },
    {
      "epoch": 0.05259921479589423,
      "grad_norm": 1.0859375,
      "learning_rate": 1.9970504368509923e-05,
      "loss": 0.3582,
      "step": 695
    },
    {
      "epoch": 0.05267489711934156,
      "grad_norm": 0.99609375,
      "learning_rate": 1.997041288345072e-05,
      "loss": 0.3371,
      "step": 696
    },
    {
      "epoch": 0.052750579442788895,
      "grad_norm": 1.0,
      "learning_rate": 1.99703212569441e-05,
      "loss": 0.3712,
      "step": 697
    },
    {
      "epoch": 0.05282626176623623,
      "grad_norm": 0.94140625,
      "learning_rate": 1.997022948899136e-05,
      "loss": 0.3466,
      "step": 698
    },
    {
      "epoch": 0.05290194408968355,
      "grad_norm": 0.98828125,
      "learning_rate": 1.9970137579593805e-05,
      "loss": 0.3577,
      "step": 699
    },
    {
      "epoch": 0.052977626413130884,
      "grad_norm": 0.95703125,
      "learning_rate": 1.9970045528752745e-05,
      "loss": 0.3489,
      "step": 700
    },
    {
      "epoch": 0.053053308736578215,
      "grad_norm": 1.140625,
      "learning_rate": 1.9969953336469478e-05,
      "loss": 0.4045,
      "step": 701
    },
    {
      "epoch": 0.05312899106002554,
      "grad_norm": 0.94921875,
      "learning_rate": 1.9969861002745312e-05,
      "loss": 0.3923,
      "step": 702
    },
    {
      "epoch": 0.05320467338347287,
      "grad_norm": 0.9609375,
      "learning_rate": 1.9969768527581558e-05,
      "loss": 0.3333,
      "step": 703
    },
    {
      "epoch": 0.053280355706920204,
      "grad_norm": 0.91015625,
      "learning_rate": 1.996967591097953e-05,
      "loss": 0.3249,
      "step": 704
    },
    {
      "epoch": 0.05335603803036753,
      "grad_norm": 1.0,
      "learning_rate": 1.9969583152940543e-05,
      "loss": 0.3935,
      "step": 705
    },
    {
      "epoch": 0.05343172035381486,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9969490253465907e-05,
      "loss": 0.3692,
      "step": 706
    },
    {
      "epoch": 0.05350740267726219,
      "grad_norm": 0.9765625,
      "learning_rate": 1.9969397212556942e-05,
      "loss": 0.3336,
      "step": 707
    },
    {
      "epoch": 0.053583085000709525,
      "grad_norm": 0.921875,
      "learning_rate": 1.9969304030214976e-05,
      "loss": 0.3274,
      "step": 708
    },
    {
      "epoch": 0.05365876732415685,
      "grad_norm": 0.921875,
      "learning_rate": 1.996921070644132e-05,
      "loss": 0.3342,
      "step": 709
    },
    {
      "epoch": 0.05373444964760418,
      "grad_norm": 1.015625,
      "learning_rate": 1.9969117241237302e-05,
      "loss": 0.3318,
      "step": 710
    },
    {
      "epoch": 0.053810131971051514,
      "grad_norm": 1.015625,
      "learning_rate": 1.996902363460425e-05,
      "loss": 0.3801,
      "step": 711
    },
    {
      "epoch": 0.05388581429449884,
      "grad_norm": 1.0859375,
      "learning_rate": 1.9968929886543487e-05,
      "loss": 0.3571,
      "step": 712
    },
    {
      "epoch": 0.05396149661794617,
      "grad_norm": 1.046875,
      "learning_rate": 1.996883599705635e-05,
      "loss": 0.3676,
      "step": 713
    },
    {
      "epoch": 0.0540371789413935,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9968741966144167e-05,
      "loss": 0.3179,
      "step": 714
    },
    {
      "epoch": 0.05411286126484083,
      "grad_norm": 0.96875,
      "learning_rate": 1.9968647793808272e-05,
      "loss": 0.3376,
      "step": 715
    },
    {
      "epoch": 0.05418854358828816,
      "grad_norm": 1.0,
      "learning_rate": 1.9968553480050002e-05,
      "loss": 0.3683,
      "step": 716
    },
    {
      "epoch": 0.05426422591173549,
      "grad_norm": 0.984375,
      "learning_rate": 1.996845902487069e-05,
      "loss": 0.3555,
      "step": 717
    },
    {
      "epoch": 0.05433990823518282,
      "grad_norm": 4.25,
      "learning_rate": 1.9968364428271684e-05,
      "loss": 0.7415,
      "step": 718
    },
    {
      "epoch": 0.05441559055863015,
      "grad_norm": 0.890625,
      "learning_rate": 1.9968269690254323e-05,
      "loss": 0.316,
      "step": 719
    },
    {
      "epoch": 0.05449127288207748,
      "grad_norm": 0.9296875,
      "learning_rate": 1.996817481081995e-05,
      "loss": 0.3415,
      "step": 720
    },
    {
      "epoch": 0.05456695520552481,
      "grad_norm": 0.8984375,
      "learning_rate": 1.996807978996991e-05,
      "loss": 0.3331,
      "step": 721
    },
    {
      "epoch": 0.05464263752897214,
      "grad_norm": 0.9375,
      "learning_rate": 1.996798462770555e-05,
      "loss": 0.3407,
      "step": 722
    },
    {
      "epoch": 0.05471831985241947,
      "grad_norm": 1.0625,
      "learning_rate": 1.9967889324028228e-05,
      "loss": 0.3799,
      "step": 723
    },
    {
      "epoch": 0.0547940021758668,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9967793878939286e-05,
      "loss": 0.3156,
      "step": 724
    },
    {
      "epoch": 0.05486968449931413,
      "grad_norm": 0.96484375,
      "learning_rate": 1.9967698292440083e-05,
      "loss": 0.349,
      "step": 725
    },
    {
      "epoch": 0.05494536682276146,
      "grad_norm": 0.953125,
      "learning_rate": 1.9967602564531976e-05,
      "loss": 0.3379,
      "step": 726
    },
    {
      "epoch": 0.05502104914620879,
      "grad_norm": 0.95703125,
      "learning_rate": 1.996750669521632e-05,
      "loss": 0.3445,
      "step": 727
    },
    {
      "epoch": 0.055096731469656116,
      "grad_norm": 0.92578125,
      "learning_rate": 1.996741068449448e-05,
      "loss": 0.3721,
      "step": 728
    },
    {
      "epoch": 0.05517241379310345,
      "grad_norm": 0.95703125,
      "learning_rate": 1.9967314532367813e-05,
      "loss": 0.3768,
      "step": 729
    },
    {
      "epoch": 0.05524809611655078,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9967218238837684e-05,
      "loss": 0.3101,
      "step": 730
    },
    {
      "epoch": 0.05532377843999811,
      "grad_norm": 3.75,
      "learning_rate": 1.996712180390546e-05,
      "loss": 0.7723,
      "step": 731
    },
    {
      "epoch": 0.055399460763445436,
      "grad_norm": 0.94140625,
      "learning_rate": 1.996702522757251e-05,
      "loss": 0.3446,
      "step": 732
    },
    {
      "epoch": 0.05547514308689277,
      "grad_norm": 0.9375,
      "learning_rate": 1.9966928509840204e-05,
      "loss": 0.3165,
      "step": 733
    },
    {
      "epoch": 0.0555508254103401,
      "grad_norm": 0.9609375,
      "learning_rate": 1.996683165070991e-05,
      "loss": 0.3451,
      "step": 734
    },
    {
      "epoch": 0.055626507733787425,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9966734650183008e-05,
      "loss": 0.292,
      "step": 735
    },
    {
      "epoch": 0.05570219005723476,
      "grad_norm": 0.91796875,
      "learning_rate": 1.996663750826087e-05,
      "loss": 0.3305,
      "step": 736
    },
    {
      "epoch": 0.05577787238068209,
      "grad_norm": 0.96875,
      "learning_rate": 1.996654022494488e-05,
      "loss": 0.387,
      "step": 737
    },
    {
      "epoch": 0.055853554704129414,
      "grad_norm": 0.96484375,
      "learning_rate": 1.9966442800236407e-05,
      "loss": 0.3408,
      "step": 738
    },
    {
      "epoch": 0.055929237027576746,
      "grad_norm": 1.015625,
      "learning_rate": 1.9966345234136843e-05,
      "loss": 0.4074,
      "step": 739
    },
    {
      "epoch": 0.05600491935102408,
      "grad_norm": 0.921875,
      "learning_rate": 1.9966247526647567e-05,
      "loss": 0.3149,
      "step": 740
    },
    {
      "epoch": 0.0560806016744714,
      "grad_norm": 0.91015625,
      "learning_rate": 1.996614967776997e-05,
      "loss": 0.3351,
      "step": 741
    },
    {
      "epoch": 0.056156283997918735,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9966051687505436e-05,
      "loss": 0.304,
      "step": 742
    },
    {
      "epoch": 0.05623196632136607,
      "grad_norm": 0.94921875,
      "learning_rate": 1.9965953555855356e-05,
      "loss": 0.3264,
      "step": 743
    },
    {
      "epoch": 0.0563076486448134,
      "grad_norm": 0.9609375,
      "learning_rate": 1.9965855282821125e-05,
      "loss": 0.3736,
      "step": 744
    },
    {
      "epoch": 0.056383330968260724,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9965756868404135e-05,
      "loss": 0.3301,
      "step": 745
    },
    {
      "epoch": 0.056459013291708056,
      "grad_norm": 1.046875,
      "learning_rate": 1.996565831260578e-05,
      "loss": 0.3905,
      "step": 746
    },
    {
      "epoch": 0.05653469561515539,
      "grad_norm": 0.9453125,
      "learning_rate": 1.996555961542746e-05,
      "loss": 0.3321,
      "step": 747
    },
    {
      "epoch": 0.05661037793860271,
      "grad_norm": 7.375,
      "learning_rate": 1.9965460776870577e-05,
      "loss": 0.6833,
      "step": 748
    },
    {
      "epoch": 0.056686060262050045,
      "grad_norm": 0.984375,
      "learning_rate": 1.996536179693653e-05,
      "loss": 0.3706,
      "step": 749
    },
    {
      "epoch": 0.05676174258549738,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9965262675626726e-05,
      "loss": 0.3832,
      "step": 750
    },
    {
      "epoch": 0.0568374249089447,
      "grad_norm": 0.96484375,
      "learning_rate": 1.9965163412942567e-05,
      "loss": 0.3638,
      "step": 751
    },
    {
      "epoch": 0.056913107232392034,
      "grad_norm": 0.97265625,
      "learning_rate": 1.9965064008885465e-05,
      "loss": 0.3278,
      "step": 752
    },
    {
      "epoch": 0.056988789555839366,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9964964463456833e-05,
      "loss": 0.3442,
      "step": 753
    },
    {
      "epoch": 0.0570644718792867,
      "grad_norm": 1.1796875,
      "learning_rate": 1.9964864776658076e-05,
      "loss": 0.3222,
      "step": 754
    },
    {
      "epoch": 0.05714015420273402,
      "grad_norm": 0.99609375,
      "learning_rate": 1.9964764948490615e-05,
      "loss": 0.375,
      "step": 755
    },
    {
      "epoch": 0.057215836526181355,
      "grad_norm": 0.99609375,
      "learning_rate": 1.9964664978955863e-05,
      "loss": 0.359,
      "step": 756
    },
    {
      "epoch": 0.057291518849628686,
      "grad_norm": 0.9609375,
      "learning_rate": 1.9964564868055234e-05,
      "loss": 0.3796,
      "step": 757
    },
    {
      "epoch": 0.05736720117307601,
      "grad_norm": 0.9921875,
      "learning_rate": 1.9964464615790154e-05,
      "loss": 0.3428,
      "step": 758
    },
    {
      "epoch": 0.05744288349652334,
      "grad_norm": 1.078125,
      "learning_rate": 1.9964364222162047e-05,
      "loss": 0.3479,
      "step": 759
    },
    {
      "epoch": 0.057518565819970675,
      "grad_norm": 0.9140625,
      "learning_rate": 1.996426368717233e-05,
      "loss": 0.3721,
      "step": 760
    },
    {
      "epoch": 0.057594248143418,
      "grad_norm": 0.9375,
      "learning_rate": 1.9964163010822438e-05,
      "loss": 0.3827,
      "step": 761
    },
    {
      "epoch": 0.05766993046686533,
      "grad_norm": 1.046875,
      "learning_rate": 1.996406219311379e-05,
      "loss": 0.3407,
      "step": 762
    },
    {
      "epoch": 0.057745612790312664,
      "grad_norm": 1.0625,
      "learning_rate": 1.9963961234047825e-05,
      "loss": 0.3939,
      "step": 763
    },
    {
      "epoch": 0.05782129511375999,
      "grad_norm": 1.0859375,
      "learning_rate": 1.9963860133625967e-05,
      "loss": 0.4425,
      "step": 764
    },
    {
      "epoch": 0.05789697743720732,
      "grad_norm": 0.9609375,
      "learning_rate": 1.996375889184966e-05,
      "loss": 0.3547,
      "step": 765
    },
    {
      "epoch": 0.05797265976065465,
      "grad_norm": 1.0078125,
      "learning_rate": 1.996365750872033e-05,
      "loss": 0.3436,
      "step": 766
    },
    {
      "epoch": 0.058048342084101985,
      "grad_norm": 1.015625,
      "learning_rate": 1.996355598423942e-05,
      "loss": 0.3786,
      "step": 767
    },
    {
      "epoch": 0.05812402440754931,
      "grad_norm": 0.96875,
      "learning_rate": 1.9963454318408372e-05,
      "loss": 0.3625,
      "step": 768
    },
    {
      "epoch": 0.05819970673099664,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9963352511228627e-05,
      "loss": 0.3565,
      "step": 769
    },
    {
      "epoch": 0.058275389054443974,
      "grad_norm": 0.96875,
      "learning_rate": 1.9963250562701624e-05,
      "loss": 0.3537,
      "step": 770
    },
    {
      "epoch": 0.0583510713778913,
      "grad_norm": 1.0703125,
      "learning_rate": 1.996314847282882e-05,
      "loss": 0.3762,
      "step": 771
    },
    {
      "epoch": 0.05842675370133863,
      "grad_norm": 0.95703125,
      "learning_rate": 1.996304624161166e-05,
      "loss": 0.3667,
      "step": 772
    },
    {
      "epoch": 0.05850243602478596,
      "grad_norm": 1.09375,
      "learning_rate": 1.9962943869051585e-05,
      "loss": 0.3633,
      "step": 773
    },
    {
      "epoch": 0.05857811834823329,
      "grad_norm": 0.984375,
      "learning_rate": 1.996284135515006e-05,
      "loss": 0.374,
      "step": 774
    },
    {
      "epoch": 0.05865380067168062,
      "grad_norm": 0.98046875,
      "learning_rate": 1.9962738699908527e-05,
      "loss": 0.3673,
      "step": 775
    },
    {
      "epoch": 0.05872948299512795,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9962635903328454e-05,
      "loss": 0.3383,
      "step": 776
    },
    {
      "epoch": 0.058805165318575284,
      "grad_norm": 1.0,
      "learning_rate": 1.9962532965411292e-05,
      "loss": 0.373,
      "step": 777
    },
    {
      "epoch": 0.05888084764202261,
      "grad_norm": 0.99609375,
      "learning_rate": 1.9962429886158503e-05,
      "loss": 0.3769,
      "step": 778
    },
    {
      "epoch": 0.05895652996546994,
      "grad_norm": 1.1328125,
      "learning_rate": 1.9962326665571556e-05,
      "loss": 0.3535,
      "step": 779
    },
    {
      "epoch": 0.05903221228891727,
      "grad_norm": 0.9765625,
      "learning_rate": 1.99622233036519e-05,
      "loss": 0.353,
      "step": 780
    },
    {
      "epoch": 0.0591078946123646,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9962119800401017e-05,
      "loss": 0.3396,
      "step": 781
    },
    {
      "epoch": 0.05918357693581193,
      "grad_norm": 3.03125,
      "learning_rate": 1.9962016155820367e-05,
      "loss": 0.7116,
      "step": 782
    },
    {
      "epoch": 0.05925925925925926,
      "grad_norm": 1.0390625,
      "learning_rate": 1.9961912369911423e-05,
      "loss": 0.3844,
      "step": 783
    },
    {
      "epoch": 0.059334941582706587,
      "grad_norm": 0.9921875,
      "learning_rate": 1.9961808442675655e-05,
      "loss": 0.4052,
      "step": 784
    },
    {
      "epoch": 0.05941062390615392,
      "grad_norm": 0.9296875,
      "learning_rate": 1.996170437411454e-05,
      "loss": 0.3547,
      "step": 785
    },
    {
      "epoch": 0.05948630622960125,
      "grad_norm": 1.03125,
      "learning_rate": 1.9961600164229557e-05,
      "loss": 0.3654,
      "step": 786
    },
    {
      "epoch": 0.059561988553048575,
      "grad_norm": 0.97265625,
      "learning_rate": 1.9961495813022177e-05,
      "loss": 0.3551,
      "step": 787
    },
    {
      "epoch": 0.05963767087649591,
      "grad_norm": 1.109375,
      "learning_rate": 1.9961391320493886e-05,
      "loss": 0.3956,
      "step": 788
    },
    {
      "epoch": 0.05971335319994324,
      "grad_norm": 1.7734375,
      "learning_rate": 1.9961286686646164e-05,
      "loss": 0.38,
      "step": 789
    },
    {
      "epoch": 0.05978903552339057,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9961181911480495e-05,
      "loss": 0.3607,
      "step": 790
    },
    {
      "epoch": 0.059864717846837896,
      "grad_norm": 1.0078125,
      "learning_rate": 1.996107699499837e-05,
      "loss": 0.4053,
      "step": 791
    },
    {
      "epoch": 0.05994040017028523,
      "grad_norm": 0.98046875,
      "learning_rate": 1.9960971937201273e-05,
      "loss": 0.3192,
      "step": 792
    },
    {
      "epoch": 0.06001608249373256,
      "grad_norm": 0.90625,
      "learning_rate": 1.9960866738090694e-05,
      "loss": 0.3236,
      "step": 793
    },
    {
      "epoch": 0.060091764817179885,
      "grad_norm": 0.92578125,
      "learning_rate": 1.996076139766813e-05,
      "loss": 0.362,
      "step": 794
    },
    {
      "epoch": 0.06016744714062722,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9960655915935066e-05,
      "loss": 0.3447,
      "step": 795
    },
    {
      "epoch": 0.06024312946407455,
      "grad_norm": 2.265625,
      "learning_rate": 1.9960550292893012e-05,
      "loss": 0.6127,
      "step": 796
    },
    {
      "epoch": 0.060318811787521874,
      "grad_norm": 2.796875,
      "learning_rate": 1.9960444528543453e-05,
      "loss": 0.5918,
      "step": 797
    },
    {
      "epoch": 0.060394494110969206,
      "grad_norm": 1.0546875,
      "learning_rate": 1.99603386228879e-05,
      "loss": 0.3917,
      "step": 798
    },
    {
      "epoch": 0.06047017643441654,
      "grad_norm": 0.98046875,
      "learning_rate": 1.996023257592785e-05,
      "loss": 0.3538,
      "step": 799
    },
    {
      "epoch": 0.06054585875786386,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9960126387664808e-05,
      "loss": 0.3274,
      "step": 800
    },
    {
      "epoch": 0.060621541081311195,
      "grad_norm": 0.921875,
      "learning_rate": 1.996002005810028e-05,
      "loss": 0.314,
      "step": 801
    },
    {
      "epoch": 0.06069722340475853,
      "grad_norm": 1.0625,
      "learning_rate": 1.9959913587235776e-05,
      "loss": 0.3665,
      "step": 802
    },
    {
      "epoch": 0.06077290572820586,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9959806975072807e-05,
      "loss": 0.3456,
      "step": 803
    },
    {
      "epoch": 0.060848588051653184,
      "grad_norm": 0.97265625,
      "learning_rate": 1.9959700221612887e-05,
      "loss": 0.3403,
      "step": 804
    },
    {
      "epoch": 0.060924270375100516,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9959593326857524e-05,
      "loss": 0.3457,
      "step": 805
    },
    {
      "epoch": 0.06099995269854785,
      "grad_norm": 1.0234375,
      "learning_rate": 1.995948629080824e-05,
      "loss": 0.3843,
      "step": 806
    },
    {
      "epoch": 0.06107563502199517,
      "grad_norm": 1.0,
      "learning_rate": 1.995937911346655e-05,
      "loss": 0.3499,
      "step": 807
    },
    {
      "epoch": 0.061151317345442505,
      "grad_norm": 0.984375,
      "learning_rate": 1.9959271794833977e-05,
      "loss": 0.3667,
      "step": 808
    },
    {
      "epoch": 0.06122699966888984,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9959164334912045e-05,
      "loss": 0.3389,
      "step": 809
    },
    {
      "epoch": 0.06130268199233716,
      "grad_norm": 0.98046875,
      "learning_rate": 1.9959056733702273e-05,
      "loss": 0.3368,
      "step": 810
    },
    {
      "epoch": 0.061378364315784494,
      "grad_norm": 0.9375,
      "learning_rate": 1.9958948991206192e-05,
      "loss": 0.3538,
      "step": 811
    },
    {
      "epoch": 0.061454046639231825,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9958841107425335e-05,
      "loss": 0.2851,
      "step": 812
    },
    {
      "epoch": 0.06152972896267916,
      "grad_norm": 0.9375,
      "learning_rate": 1.995873308236122e-05,
      "loss": 0.3277,
      "step": 813
    },
    {
      "epoch": 0.06160541128612648,
      "grad_norm": 1.0,
      "learning_rate": 1.995862491601539e-05,
      "loss": 0.377,
      "step": 814
    },
    {
      "epoch": 0.061681093609573814,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9958516608389372e-05,
      "loss": 0.3406,
      "step": 815
    },
    {
      "epoch": 0.061756775933021146,
      "grad_norm": 0.98828125,
      "learning_rate": 1.995840815948471e-05,
      "loss": 0.4069,
      "step": 816
    },
    {
      "epoch": 0.06183245825646847,
      "grad_norm": 0.97265625,
      "learning_rate": 1.995829956930294e-05,
      "loss": 0.3271,
      "step": 817
    },
    {
      "epoch": 0.0619081405799158,
      "grad_norm": 0.96484375,
      "learning_rate": 1.99581908378456e-05,
      "loss": 0.3837,
      "step": 818
    },
    {
      "epoch": 0.061983822903363135,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9958081965114232e-05,
      "loss": 0.4003,
      "step": 819
    },
    {
      "epoch": 0.06205950522681046,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9957972951110384e-05,
      "loss": 0.3105,
      "step": 820
    },
    {
      "epoch": 0.06213518755025779,
      "grad_norm": 1.5078125,
      "learning_rate": 1.9957863795835602e-05,
      "loss": 0.3835,
      "step": 821
    },
    {
      "epoch": 0.062210869873705124,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9957754499291436e-05,
      "loss": 0.3269,
      "step": 822
    },
    {
      "epoch": 0.06228655219715245,
      "grad_norm": 0.9296875,
      "learning_rate": 1.995764506147943e-05,
      "loss": 0.3502,
      "step": 823
    },
    {
      "epoch": 0.06236223452059978,
      "grad_norm": 0.97265625,
      "learning_rate": 1.9957535482401146e-05,
      "loss": 0.3756,
      "step": 824
    },
    {
      "epoch": 0.06243791684404711,
      "grad_norm": 0.95703125,
      "learning_rate": 1.995742576205813e-05,
      "loss": 0.3842,
      "step": 825
    },
    {
      "epoch": 0.06251359916749444,
      "grad_norm": 0.96484375,
      "learning_rate": 1.9957315900451938e-05,
      "loss": 0.3475,
      "step": 826
    },
    {
      "epoch": 0.06258928149094177,
      "grad_norm": 0.8515625,
      "learning_rate": 1.995720589758414e-05,
      "loss": 0.3309,
      "step": 827
    },
    {
      "epoch": 0.0626649638143891,
      "grad_norm": 0.984375,
      "learning_rate": 1.9957095753456287e-05,
      "loss": 0.3971,
      "step": 828
    },
    {
      "epoch": 0.06274064613783643,
      "grad_norm": 0.8671875,
      "learning_rate": 1.995698546806994e-05,
      "loss": 0.3163,
      "step": 829
    },
    {
      "epoch": 0.06281632846128377,
      "grad_norm": 0.9296875,
      "learning_rate": 1.995687504142667e-05,
      "loss": 0.3609,
      "step": 830
    },
    {
      "epoch": 0.0628920107847311,
      "grad_norm": 0.9609375,
      "learning_rate": 1.9956764473528043e-05,
      "loss": 0.3509,
      "step": 831
    },
    {
      "epoch": 0.06296769310817842,
      "grad_norm": 0.984375,
      "learning_rate": 1.9956653764375623e-05,
      "loss": 0.3752,
      "step": 832
    },
    {
      "epoch": 0.06304337543162575,
      "grad_norm": 0.921875,
      "learning_rate": 1.9956542913970986e-05,
      "loss": 0.3339,
      "step": 833
    },
    {
      "epoch": 0.06311905775507308,
      "grad_norm": 1.015625,
      "learning_rate": 1.9956431922315696e-05,
      "loss": 0.4245,
      "step": 834
    },
    {
      "epoch": 0.06319474007852041,
      "grad_norm": 0.953125,
      "learning_rate": 1.9956320789411338e-05,
      "loss": 0.367,
      "step": 835
    },
    {
      "epoch": 0.06327042240196774,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9956209515259485e-05,
      "loss": 0.295,
      "step": 836
    },
    {
      "epoch": 0.06334610472541508,
      "grad_norm": 0.984375,
      "learning_rate": 1.9956098099861713e-05,
      "loss": 0.3758,
      "step": 837
    },
    {
      "epoch": 0.0634217870488624,
      "grad_norm": 0.99609375,
      "learning_rate": 1.9955986543219602e-05,
      "loss": 0.3536,
      "step": 838
    },
    {
      "epoch": 0.06349746937230973,
      "grad_norm": 9.625,
      "learning_rate": 1.995587484533474e-05,
      "loss": 0.697,
      "step": 839
    },
    {
      "epoch": 0.06357315169575706,
      "grad_norm": 0.890625,
      "learning_rate": 1.995576300620871e-05,
      "loss": 0.2943,
      "step": 840
    },
    {
      "epoch": 0.06364883401920439,
      "grad_norm": 0.921875,
      "learning_rate": 1.9955651025843092e-05,
      "loss": 0.3064,
      "step": 841
    },
    {
      "epoch": 0.06372451634265172,
      "grad_norm": 0.95703125,
      "learning_rate": 1.9955538904239482e-05,
      "loss": 0.3422,
      "step": 842
    },
    {
      "epoch": 0.06380019866609905,
      "grad_norm": 1.0234375,
      "learning_rate": 1.9955426641399467e-05,
      "loss": 0.3537,
      "step": 843
    },
    {
      "epoch": 0.06387588098954639,
      "grad_norm": 0.98828125,
      "learning_rate": 1.9955314237324642e-05,
      "loss": 0.3525,
      "step": 844
    },
    {
      "epoch": 0.0639515633129937,
      "grad_norm": 1.0390625,
      "learning_rate": 1.99552016920166e-05,
      "loss": 0.3637,
      "step": 845
    },
    {
      "epoch": 0.06402724563644104,
      "grad_norm": 0.890625,
      "learning_rate": 1.995508900547694e-05,
      "loss": 0.322,
      "step": 846
    },
    {
      "epoch": 0.06410292795988837,
      "grad_norm": 1.03125,
      "learning_rate": 1.9954976177707257e-05,
      "loss": 0.4191,
      "step": 847
    },
    {
      "epoch": 0.0641786102833357,
      "grad_norm": 0.90625,
      "learning_rate": 1.9954863208709155e-05,
      "loss": 0.3535,
      "step": 848
    },
    {
      "epoch": 0.06425429260678303,
      "grad_norm": 2.40625,
      "learning_rate": 1.9954750098484233e-05,
      "loss": 0.669,
      "step": 849
    },
    {
      "epoch": 0.06432997493023036,
      "grad_norm": 0.97265625,
      "learning_rate": 1.9954636847034105e-05,
      "loss": 0.3541,
      "step": 850
    },
    {
      "epoch": 0.06440565725367768,
      "grad_norm": 0.98828125,
      "learning_rate": 1.9954523454360365e-05,
      "loss": 0.3581,
      "step": 851
    },
    {
      "epoch": 0.06448133957712501,
      "grad_norm": 1.125,
      "learning_rate": 1.9954409920464627e-05,
      "loss": 0.3801,
      "step": 852
    },
    {
      "epoch": 0.06455702190057235,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9954296245348503e-05,
      "loss": 0.3088,
      "step": 853
    },
    {
      "epoch": 0.06463270422401968,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9954182429013606e-05,
      "loss": 0.3835,
      "step": 854
    },
    {
      "epoch": 0.06470838654746701,
      "grad_norm": 1.0390625,
      "learning_rate": 1.9954068471461542e-05,
      "loss": 0.3441,
      "step": 855
    },
    {
      "epoch": 0.06478406887091434,
      "grad_norm": 0.953125,
      "learning_rate": 1.9953954372693944e-05,
      "loss": 0.3678,
      "step": 856
    },
    {
      "epoch": 0.06485975119436167,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9953840132712415e-05,
      "loss": 0.3583,
      "step": 857
    },
    {
      "epoch": 0.06493543351780899,
      "grad_norm": 0.94921875,
      "learning_rate": 1.9953725751518588e-05,
      "loss": 0.3647,
      "step": 858
    },
    {
      "epoch": 0.06501111584125632,
      "grad_norm": 1.0390625,
      "learning_rate": 1.995361122911408e-05,
      "loss": 0.418,
      "step": 859
    },
    {
      "epoch": 0.06508679816470365,
      "grad_norm": 0.9296875,
      "learning_rate": 1.995349656550051e-05,
      "loss": 0.3494,
      "step": 860
    },
    {
      "epoch": 0.06516248048815099,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9953381760679513e-05,
      "loss": 0.2929,
      "step": 861
    },
    {
      "epoch": 0.06523816281159832,
      "grad_norm": 0.921875,
      "learning_rate": 1.9953266814652715e-05,
      "loss": 0.2844,
      "step": 862
    },
    {
      "epoch": 0.06531384513504565,
      "grad_norm": 1.0,
      "learning_rate": 1.9953151727421748e-05,
      "loss": 0.4192,
      "step": 863
    },
    {
      "epoch": 0.06538952745849298,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9953036498988246e-05,
      "loss": 0.3211,
      "step": 864
    },
    {
      "epoch": 0.0654652097819403,
      "grad_norm": 0.96875,
      "learning_rate": 1.9952921129353838e-05,
      "loss": 0.3695,
      "step": 865
    },
    {
      "epoch": 0.06554089210538763,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9952805618520165e-05,
      "loss": 0.3541,
      "step": 866
    },
    {
      "epoch": 0.06561657442883496,
      "grad_norm": 0.9921875,
      "learning_rate": 1.995268996648886e-05,
      "loss": 0.3755,
      "step": 867
    },
    {
      "epoch": 0.0656922567522823,
      "grad_norm": 0.921875,
      "learning_rate": 1.9952574173261574e-05,
      "loss": 0.3202,
      "step": 868
    },
    {
      "epoch": 0.06576793907572963,
      "grad_norm": 0.9921875,
      "learning_rate": 1.995245823883994e-05,
      "loss": 0.3728,
      "step": 869
    },
    {
      "epoch": 0.06584362139917696,
      "grad_norm": 3.46875,
      "learning_rate": 1.9952342163225614e-05,
      "loss": 0.6768,
      "step": 870
    },
    {
      "epoch": 0.06591930372262428,
      "grad_norm": 0.890625,
      "learning_rate": 1.995222594642023e-05,
      "loss": 0.305,
      "step": 871
    },
    {
      "epoch": 0.06599498604607161,
      "grad_norm": 0.9765625,
      "learning_rate": 1.9952109588425443e-05,
      "loss": 0.3538,
      "step": 872
    },
    {
      "epoch": 0.06607066836951894,
      "grad_norm": 2.59375,
      "learning_rate": 1.9951993089242906e-05,
      "loss": 0.6232,
      "step": 873
    },
    {
      "epoch": 0.06614635069296627,
      "grad_norm": 0.984375,
      "learning_rate": 1.9951876448874264e-05,
      "loss": 0.3891,
      "step": 874
    },
    {
      "epoch": 0.0662220330164136,
      "grad_norm": 0.8984375,
      "learning_rate": 1.995175966732118e-05,
      "loss": 0.358,
      "step": 875
    },
    {
      "epoch": 0.06629771533986094,
      "grad_norm": 0.98046875,
      "learning_rate": 1.9951642744585307e-05,
      "loss": 0.3655,
      "step": 876
    },
    {
      "epoch": 0.06637339766330827,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9951525680668303e-05,
      "loss": 0.3622,
      "step": 877
    },
    {
      "epoch": 0.06644907998675559,
      "grad_norm": 0.98046875,
      "learning_rate": 1.995140847557183e-05,
      "loss": 0.3467,
      "step": 878
    },
    {
      "epoch": 0.06652476231020292,
      "grad_norm": 1.0,
      "learning_rate": 1.995129112929755e-05,
      "loss": 0.3625,
      "step": 879
    },
    {
      "epoch": 0.06660044463365025,
      "grad_norm": 0.890625,
      "learning_rate": 1.995117364184713e-05,
      "loss": 0.3275,
      "step": 880
    },
    {
      "epoch": 0.06667612695709758,
      "grad_norm": 0.96875,
      "learning_rate": 1.9951056013222234e-05,
      "loss": 0.3637,
      "step": 881
    },
    {
      "epoch": 0.06675180928054492,
      "grad_norm": 1.03125,
      "learning_rate": 1.9950938243424533e-05,
      "loss": 0.3811,
      "step": 882
    },
    {
      "epoch": 0.06682749160399225,
      "grad_norm": 5.34375,
      "learning_rate": 1.9950820332455695e-05,
      "loss": 0.5513,
      "step": 883
    },
    {
      "epoch": 0.06690317392743957,
      "grad_norm": 0.98828125,
      "learning_rate": 1.9950702280317396e-05,
      "loss": 0.3679,
      "step": 884
    },
    {
      "epoch": 0.0669788562508869,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9950584087011305e-05,
      "loss": 0.3307,
      "step": 885
    },
    {
      "epoch": 0.06705453857433423,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9950465752539106e-05,
      "loss": 0.3297,
      "step": 886
    },
    {
      "epoch": 0.06713022089778156,
      "grad_norm": 0.9375,
      "learning_rate": 1.9950347276902477e-05,
      "loss": 0.3548,
      "step": 887
    },
    {
      "epoch": 0.0672059032212289,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9950228660103093e-05,
      "loss": 0.3246,
      "step": 888
    },
    {
      "epoch": 0.06728158554467623,
      "grad_norm": 0.96484375,
      "learning_rate": 1.9950109902142644e-05,
      "loss": 0.3802,
      "step": 889
    },
    {
      "epoch": 0.06735726786812356,
      "grad_norm": 0.94140625,
      "learning_rate": 1.994999100302281e-05,
      "loss": 0.3736,
      "step": 890
    },
    {
      "epoch": 0.06743295019157088,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9949871962745275e-05,
      "loss": 0.3368,
      "step": 891
    },
    {
      "epoch": 0.06750863251501821,
      "grad_norm": 1.0,
      "learning_rate": 1.9949752781311734e-05,
      "loss": 0.3618,
      "step": 892
    },
    {
      "epoch": 0.06758431483846554,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9949633458723876e-05,
      "loss": 0.3114,
      "step": 893
    },
    {
      "epoch": 0.06765999716191287,
      "grad_norm": 0.8828125,
      "learning_rate": 1.994951399498339e-05,
      "loss": 0.2876,
      "step": 894
    },
    {
      "epoch": 0.0677356794853602,
      "grad_norm": 1.1015625,
      "learning_rate": 1.9949394390091977e-05,
      "loss": 0.3384,
      "step": 895
    },
    {
      "epoch": 0.06781136180880754,
      "grad_norm": 0.875,
      "learning_rate": 1.994927464405133e-05,
      "loss": 0.3288,
      "step": 896
    },
    {
      "epoch": 0.06788704413225485,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9949154756863152e-05,
      "loss": 0.343,
      "step": 897
    },
    {
      "epoch": 0.06796272645570219,
      "grad_norm": 1.0625,
      "learning_rate": 1.9949034728529136e-05,
      "loss": 0.3974,
      "step": 898
    },
    {
      "epoch": 0.06803840877914952,
      "grad_norm": 0.9609375,
      "learning_rate": 1.994891455905099e-05,
      "loss": 0.3552,
      "step": 899
    },
    {
      "epoch": 0.06811409110259685,
      "grad_norm": 0.9765625,
      "learning_rate": 1.9948794248430425e-05,
      "loss": 0.3937,
      "step": 900
    },
    {
      "epoch": 0.06818977342604418,
      "grad_norm": 0.9765625,
      "learning_rate": 1.994867379666913e-05,
      "loss": 0.3676,
      "step": 901
    },
    {
      "epoch": 0.06826545574949151,
      "grad_norm": 0.9921875,
      "learning_rate": 1.994855320376883e-05,
      "loss": 0.3359,
      "step": 902
    },
    {
      "epoch": 0.06834113807293885,
      "grad_norm": 0.9609375,
      "learning_rate": 1.9948432469731233e-05,
      "loss": 0.3584,
      "step": 903
    },
    {
      "epoch": 0.06841682039638616,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9948311594558044e-05,
      "loss": 0.3101,
      "step": 904
    },
    {
      "epoch": 0.0684925027198335,
      "grad_norm": 1.0234375,
      "learning_rate": 1.9948190578250988e-05,
      "loss": 0.3242,
      "step": 905
    },
    {
      "epoch": 0.06856818504328083,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9948069420811774e-05,
      "loss": 0.3411,
      "step": 906
    },
    {
      "epoch": 0.06864386736672816,
      "grad_norm": 0.9609375,
      "learning_rate": 1.9947948122242122e-05,
      "loss": 0.3845,
      "step": 907
    },
    {
      "epoch": 0.06871954969017549,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9947826682543757e-05,
      "loss": 0.3438,
      "step": 908
    },
    {
      "epoch": 0.06879523201362282,
      "grad_norm": 0.9765625,
      "learning_rate": 1.99477051017184e-05,
      "loss": 0.3862,
      "step": 909
    },
    {
      "epoch": 0.06887091433707015,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9947583379767773e-05,
      "loss": 0.359,
      "step": 910
    },
    {
      "epoch": 0.06894659666051747,
      "grad_norm": 0.9375,
      "learning_rate": 1.994746151669361e-05,
      "loss": 0.3214,
      "step": 911
    },
    {
      "epoch": 0.0690222789839648,
      "grad_norm": 0.92578125,
      "learning_rate": 1.994733951249763e-05,
      "loss": 0.3485,
      "step": 912
    },
    {
      "epoch": 0.06909796130741214,
      "grad_norm": 0.91015625,
      "learning_rate": 1.994721736718157e-05,
      "loss": 0.339,
      "step": 913
    },
    {
      "epoch": 0.06917364363085947,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9947095080747163e-05,
      "loss": 0.3499,
      "step": 914
    },
    {
      "epoch": 0.0692493259543068,
      "grad_norm": 0.921875,
      "learning_rate": 1.9946972653196143e-05,
      "loss": 0.3234,
      "step": 915
    },
    {
      "epoch": 0.06932500827775413,
      "grad_norm": 0.99609375,
      "learning_rate": 1.9946850084530243e-05,
      "loss": 0.3639,
      "step": 916
    },
    {
      "epoch": 0.06940069060120145,
      "grad_norm": 0.8828125,
      "learning_rate": 1.994672737475121e-05,
      "loss": 0.326,
      "step": 917
    },
    {
      "epoch": 0.06947637292464878,
      "grad_norm": 0.96875,
      "learning_rate": 1.9946604523860775e-05,
      "loss": 0.3545,
      "step": 918
    },
    {
      "epoch": 0.06955205524809611,
      "grad_norm": 4.9375,
      "learning_rate": 1.994648153186069e-05,
      "loss": 0.7124,
      "step": 919
    },
    {
      "epoch": 0.06962773757154345,
      "grad_norm": 1.09375,
      "learning_rate": 1.994635839875269e-05,
      "loss": 0.354,
      "step": 920
    },
    {
      "epoch": 0.06970341989499078,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9946235124538534e-05,
      "loss": 0.3353,
      "step": 921
    },
    {
      "epoch": 0.06977910221843811,
      "grad_norm": 0.98046875,
      "learning_rate": 1.994611170921996e-05,
      "loss": 0.3707,
      "step": 922
    },
    {
      "epoch": 0.06985478454188544,
      "grad_norm": 0.96875,
      "learning_rate": 1.9945988152798725e-05,
      "loss": 0.3184,
      "step": 923
    },
    {
      "epoch": 0.06993046686533276,
      "grad_norm": 0.92578125,
      "learning_rate": 1.994586445527658e-05,
      "loss": 0.3402,
      "step": 924
    },
    {
      "epoch": 0.07000614918878009,
      "grad_norm": 0.94140625,
      "learning_rate": 1.994574061665528e-05,
      "loss": 0.3841,
      "step": 925
    },
    {
      "epoch": 0.07008183151222742,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9945616636936584e-05,
      "loss": 0.2945,
      "step": 926
    },
    {
      "epoch": 0.07015751383567476,
      "grad_norm": 0.96875,
      "learning_rate": 1.9945492516122246e-05,
      "loss": 0.387,
      "step": 927
    },
    {
      "epoch": 0.07023319615912209,
      "grad_norm": 1.0,
      "learning_rate": 1.994536825421403e-05,
      "loss": 0.3854,
      "step": 928
    },
    {
      "epoch": 0.07030887848256942,
      "grad_norm": 0.91015625,
      "learning_rate": 1.99452438512137e-05,
      "loss": 0.3541,
      "step": 929
    },
    {
      "epoch": 0.07038456080601674,
      "grad_norm": 0.95703125,
      "learning_rate": 1.994511930712302e-05,
      "loss": 0.3567,
      "step": 930
    },
    {
      "epoch": 0.07046024312946407,
      "grad_norm": 0.97265625,
      "learning_rate": 1.9944994621943753e-05,
      "loss": 0.4138,
      "step": 931
    },
    {
      "epoch": 0.0705359254529114,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9944869795677673e-05,
      "loss": 0.3298,
      "step": 932
    },
    {
      "epoch": 0.07061160777635873,
      "grad_norm": 0.90625,
      "learning_rate": 1.994474482832655e-05,
      "loss": 0.3151,
      "step": 933
    },
    {
      "epoch": 0.07068729009980607,
      "grad_norm": 0.98046875,
      "learning_rate": 1.9944619719892154e-05,
      "loss": 0.3295,
      "step": 934
    },
    {
      "epoch": 0.0707629724232534,
      "grad_norm": 0.89453125,
      "learning_rate": 1.994449447037626e-05,
      "loss": 0.3373,
      "step": 935
    },
    {
      "epoch": 0.07083865474670073,
      "grad_norm": 0.90625,
      "learning_rate": 1.9944369079780648e-05,
      "loss": 0.3547,
      "step": 936
    },
    {
      "epoch": 0.07091433707014805,
      "grad_norm": 1.046875,
      "learning_rate": 1.9944243548107098e-05,
      "loss": 0.3308,
      "step": 937
    },
    {
      "epoch": 0.07099001939359538,
      "grad_norm": 0.98046875,
      "learning_rate": 1.9944117875357384e-05,
      "loss": 0.3795,
      "step": 938
    },
    {
      "epoch": 0.07106570171704271,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9943992061533297e-05,
      "loss": 0.33,
      "step": 939
    },
    {
      "epoch": 0.07114138404049004,
      "grad_norm": 0.99609375,
      "learning_rate": 1.9943866106636614e-05,
      "loss": 0.3785,
      "step": 940
    },
    {
      "epoch": 0.07121706636393738,
      "grad_norm": 1.0234375,
      "learning_rate": 1.9943740010669132e-05,
      "loss": 0.4174,
      "step": 941
    },
    {
      "epoch": 0.07129274868738471,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9943613773632632e-05,
      "loss": 0.3026,
      "step": 942
    },
    {
      "epoch": 0.07136843101083203,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9943487395528903e-05,
      "loss": 0.3472,
      "step": 943
    },
    {
      "epoch": 0.07144411333427936,
      "grad_norm": 1.1328125,
      "learning_rate": 1.994336087635975e-05,
      "loss": 0.3461,
      "step": 944
    },
    {
      "epoch": 0.07151979565772669,
      "grad_norm": 1.0546875,
      "learning_rate": 1.994323421612695e-05,
      "loss": 0.414,
      "step": 945
    },
    {
      "epoch": 0.07159547798117402,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9943107414832314e-05,
      "loss": 0.3374,
      "step": 946
    },
    {
      "epoch": 0.07167116030462135,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9942980472477635e-05,
      "loss": 0.3116,
      "step": 947
    },
    {
      "epoch": 0.07174684262806869,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9942853389064716e-05,
      "loss": 0.3212,
      "step": 948
    },
    {
      "epoch": 0.07182252495151602,
      "grad_norm": 0.92578125,
      "learning_rate": 1.994272616459536e-05,
      "loss": 0.3534,
      "step": 949
    },
    {
      "epoch": 0.07189820727496334,
      "grad_norm": 0.94140625,
      "learning_rate": 1.994259879907137e-05,
      "loss": 0.3554,
      "step": 950
    },
    {
      "epoch": 0.07197388959841067,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9942471292494553e-05,
      "loss": 0.3476,
      "step": 951
    },
    {
      "epoch": 0.072049571921858,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9942343644866723e-05,
      "loss": 0.3561,
      "step": 952
    },
    {
      "epoch": 0.07212525424530533,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9942215856189683e-05,
      "loss": 0.2968,
      "step": 953
    },
    {
      "epoch": 0.07220093656875266,
      "grad_norm": 0.9453125,
      "learning_rate": 1.994208792646525e-05,
      "loss": 0.3249,
      "step": 954
    },
    {
      "epoch": 0.0722766188922,
      "grad_norm": 0.94921875,
      "learning_rate": 1.994195985569524e-05,
      "loss": 0.3475,
      "step": 955
    },
    {
      "epoch": 0.07235230121564733,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9941831643881464e-05,
      "loss": 0.3324,
      "step": 956
    },
    {
      "epoch": 0.07242798353909465,
      "grad_norm": 0.890625,
      "learning_rate": 1.994170329102575e-05,
      "loss": 0.3268,
      "step": 957
    },
    {
      "epoch": 0.07250366586254198,
      "grad_norm": 0.9140625,
      "learning_rate": 1.994157479712991e-05,
      "loss": 0.348,
      "step": 958
    },
    {
      "epoch": 0.07257934818598931,
      "grad_norm": 0.953125,
      "learning_rate": 1.9941446162195775e-05,
      "loss": 0.3681,
      "step": 959
    },
    {
      "epoch": 0.07265503050943664,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9941317386225165e-05,
      "loss": 0.3651,
      "step": 960
    },
    {
      "epoch": 0.07273071283288397,
      "grad_norm": 0.90234375,
      "learning_rate": 1.994118846921991e-05,
      "loss": 0.3556,
      "step": 961
    },
    {
      "epoch": 0.0728063951563313,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9941059411181837e-05,
      "loss": 0.3084,
      "step": 962
    },
    {
      "epoch": 0.07288207747977862,
      "grad_norm": 0.99609375,
      "learning_rate": 1.9940930212112774e-05,
      "loss": 0.4076,
      "step": 963
    },
    {
      "epoch": 0.07295775980322595,
      "grad_norm": 0.94921875,
      "learning_rate": 1.994080087201456e-05,
      "loss": 0.3301,
      "step": 964
    },
    {
      "epoch": 0.07303344212667329,
      "grad_norm": 0.921875,
      "learning_rate": 1.9940671390889023e-05,
      "loss": 0.3843,
      "step": 965
    },
    {
      "epoch": 0.07310912445012062,
      "grad_norm": 0.96484375,
      "learning_rate": 1.9940541768738004e-05,
      "loss": 0.3571,
      "step": 966
    },
    {
      "epoch": 0.07318480677356795,
      "grad_norm": 0.96875,
      "learning_rate": 1.9940412005563342e-05,
      "loss": 0.3502,
      "step": 967
    },
    {
      "epoch": 0.07326048909701528,
      "grad_norm": 0.921875,
      "learning_rate": 1.9940282101366876e-05,
      "loss": 0.3409,
      "step": 968
    },
    {
      "epoch": 0.07333617142046261,
      "grad_norm": 0.953125,
      "learning_rate": 1.9940152056150452e-05,
      "loss": 0.4069,
      "step": 969
    },
    {
      "epoch": 0.07341185374390993,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9940021869915915e-05,
      "loss": 0.3226,
      "step": 970
    },
    {
      "epoch": 0.07348753606735726,
      "grad_norm": 0.8984375,
      "learning_rate": 1.993989154266511e-05,
      "loss": 0.3521,
      "step": 971
    },
    {
      "epoch": 0.0735632183908046,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9939761074399883e-05,
      "loss": 0.3446,
      "step": 972
    },
    {
      "epoch": 0.07363890071425193,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9939630465122088e-05,
      "loss": 0.3647,
      "step": 973
    },
    {
      "epoch": 0.07371458303769926,
      "grad_norm": 0.96875,
      "learning_rate": 1.993949971483358e-05,
      "loss": 0.3533,
      "step": 974
    },
    {
      "epoch": 0.07379026536114659,
      "grad_norm": 1.015625,
      "learning_rate": 1.9939368823536213e-05,
      "loss": 0.3582,
      "step": 975
    },
    {
      "epoch": 0.07386594768459391,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9939237791231838e-05,
      "loss": 0.358,
      "step": 976
    },
    {
      "epoch": 0.07394163000804124,
      "grad_norm": 0.90625,
      "learning_rate": 1.993910661792232e-05,
      "loss": 0.3443,
      "step": 977
    },
    {
      "epoch": 0.07401731233148857,
      "grad_norm": 0.9453125,
      "learning_rate": 1.993897530360952e-05,
      "loss": 0.354,
      "step": 978
    },
    {
      "epoch": 0.0740929946549359,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9938843848295298e-05,
      "loss": 0.3013,
      "step": 979
    },
    {
      "epoch": 0.07416867697838324,
      "grad_norm": 1.03125,
      "learning_rate": 1.9938712251981523e-05,
      "loss": 0.3587,
      "step": 980
    },
    {
      "epoch": 0.07424435930183057,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9938580514670058e-05,
      "loss": 0.368,
      "step": 981
    },
    {
      "epoch": 0.0743200416252779,
      "grad_norm": 0.96875,
      "learning_rate": 1.9938448636362772e-05,
      "loss": 0.4046,
      "step": 982
    },
    {
      "epoch": 0.07439572394872522,
      "grad_norm": 0.90234375,
      "learning_rate": 1.993831661706154e-05,
      "loss": 0.3271,
      "step": 983
    },
    {
      "epoch": 0.07447140627217255,
      "grad_norm": 0.90625,
      "learning_rate": 1.9938184456768227e-05,
      "loss": 0.3617,
      "step": 984
    },
    {
      "epoch": 0.07454708859561988,
      "grad_norm": 0.99609375,
      "learning_rate": 1.993805215548472e-05,
      "loss": 0.3061,
      "step": 985
    },
    {
      "epoch": 0.07462277091906722,
      "grad_norm": 4.0,
      "learning_rate": 1.9937919713212885e-05,
      "loss": 0.672,
      "step": 986
    },
    {
      "epoch": 0.07469845324251455,
      "grad_norm": 0.94921875,
      "learning_rate": 1.9937787129954605e-05,
      "loss": 0.3752,
      "step": 987
    },
    {
      "epoch": 0.07477413556596188,
      "grad_norm": 1.0234375,
      "learning_rate": 1.9937654405711758e-05,
      "loss": 0.3469,
      "step": 988
    },
    {
      "epoch": 0.0748498178894092,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9937521540486232e-05,
      "loss": 0.3249,
      "step": 989
    },
    {
      "epoch": 0.07492550021285653,
      "grad_norm": 0.890625,
      "learning_rate": 1.993738853427991e-05,
      "loss": 0.3057,
      "step": 990
    },
    {
      "epoch": 0.07500118253630386,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9937255387094677e-05,
      "loss": 0.4027,
      "step": 991
    },
    {
      "epoch": 0.0750768648597512,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9937122098932428e-05,
      "loss": 0.2648,
      "step": 992
    },
    {
      "epoch": 0.07515254718319853,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9936988669795046e-05,
      "loss": 0.3396,
      "step": 993
    },
    {
      "epoch": 0.07522822950664586,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9936855099684427e-05,
      "loss": 0.2744,
      "step": 994
    },
    {
      "epoch": 0.07530391183009319,
      "grad_norm": 1.078125,
      "learning_rate": 1.9936721388602466e-05,
      "loss": 0.4435,
      "step": 995
    },
    {
      "epoch": 0.07537959415354051,
      "grad_norm": 0.96484375,
      "learning_rate": 1.993658753655106e-05,
      "loss": 0.3682,
      "step": 996
    },
    {
      "epoch": 0.07545527647698784,
      "grad_norm": 0.96484375,
      "learning_rate": 1.9936453543532107e-05,
      "loss": 0.3721,
      "step": 997
    },
    {
      "epoch": 0.07553095880043517,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9936319409547514e-05,
      "loss": 0.298,
      "step": 998
    },
    {
      "epoch": 0.0756066411238825,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9936185134599176e-05,
      "loss": 0.3107,
      "step": 999
    },
    {
      "epoch": 0.07568232344732984,
      "grad_norm": 3.375,
      "learning_rate": 1.9936050718689e-05,
      "loss": 0.6367,
      "step": 1000
    },
    {
      "epoch": 0.07575800577077717,
      "grad_norm": 1.1015625,
      "learning_rate": 1.9935916161818892e-05,
      "loss": 0.4079,
      "step": 1001
    },
    {
      "epoch": 0.0758336880942245,
      "grad_norm": 0.98828125,
      "learning_rate": 1.9935781463990765e-05,
      "loss": 0.3226,
      "step": 1002
    },
    {
      "epoch": 0.07590937041767182,
      "grad_norm": 0.94921875,
      "learning_rate": 1.993564662520653e-05,
      "loss": 0.3605,
      "step": 1003
    },
    {
      "epoch": 0.07598505274111915,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9935511645468094e-05,
      "loss": 0.4005,
      "step": 1004
    },
    {
      "epoch": 0.07606073506456648,
      "grad_norm": 0.94921875,
      "learning_rate": 1.9935376524777375e-05,
      "loss": 0.355,
      "step": 1005
    },
    {
      "epoch": 0.07613641738801381,
      "grad_norm": 0.89453125,
      "learning_rate": 1.993524126313629e-05,
      "loss": 0.3335,
      "step": 1006
    },
    {
      "epoch": 0.07621209971146115,
      "grad_norm": 2.171875,
      "learning_rate": 1.993510586054676e-05,
      "loss": 0.6094,
      "step": 1007
    },
    {
      "epoch": 0.07628778203490848,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9934970317010706e-05,
      "loss": 0.3342,
      "step": 1008
    },
    {
      "epoch": 0.0763634643583558,
      "grad_norm": 0.98828125,
      "learning_rate": 1.9934834632530043e-05,
      "loss": 0.3943,
      "step": 1009
    },
    {
      "epoch": 0.07643914668180313,
      "grad_norm": 4.8125,
      "learning_rate": 1.9934698807106706e-05,
      "loss": 0.6409,
      "step": 1010
    },
    {
      "epoch": 0.07651482900525046,
      "grad_norm": 0.9296875,
      "learning_rate": 1.993456284074262e-05,
      "loss": 0.3349,
      "step": 1011
    },
    {
      "epoch": 0.07659051132869779,
      "grad_norm": 0.94921875,
      "learning_rate": 1.9934426733439708e-05,
      "loss": 0.3512,
      "step": 1012
    },
    {
      "epoch": 0.07666619365214512,
      "grad_norm": 0.94140625,
      "learning_rate": 1.993429048519991e-05,
      "loss": 0.2882,
      "step": 1013
    },
    {
      "epoch": 0.07674187597559246,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9934154096025148e-05,
      "loss": 0.368,
      "step": 1014
    },
    {
      "epoch": 0.07681755829903979,
      "grad_norm": 0.9609375,
      "learning_rate": 1.993401756591736e-05,
      "loss": 0.3429,
      "step": 1015
    },
    {
      "epoch": 0.0768932406224871,
      "grad_norm": 0.97265625,
      "learning_rate": 1.9933880894878492e-05,
      "loss": 0.3513,
      "step": 1016
    },
    {
      "epoch": 0.07696892294593444,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9933744082910477e-05,
      "loss": 0.318,
      "step": 1017
    },
    {
      "epoch": 0.07704460526938177,
      "grad_norm": 0.9609375,
      "learning_rate": 1.9933607130015252e-05,
      "loss": 0.3159,
      "step": 1018
    },
    {
      "epoch": 0.0771202875928291,
      "grad_norm": 0.89453125,
      "learning_rate": 1.993347003619476e-05,
      "loss": 0.3614,
      "step": 1019
    },
    {
      "epoch": 0.07719596991627643,
      "grad_norm": 1.0859375,
      "learning_rate": 1.9933332801450955e-05,
      "loss": 0.402,
      "step": 1020
    },
    {
      "epoch": 0.07727165223972376,
      "grad_norm": 1.0859375,
      "learning_rate": 1.9933195425785774e-05,
      "loss": 0.3687,
      "step": 1021
    },
    {
      "epoch": 0.07734733456317108,
      "grad_norm": 0.9453125,
      "learning_rate": 1.9933057909201174e-05,
      "loss": 0.3741,
      "step": 1022
    },
    {
      "epoch": 0.07742301688661841,
      "grad_norm": 1.0,
      "learning_rate": 1.9932920251699098e-05,
      "loss": 0.3556,
      "step": 1023
    },
    {
      "epoch": 0.07749869921006575,
      "grad_norm": 0.9453125,
      "learning_rate": 1.9932782453281504e-05,
      "loss": 0.3586,
      "step": 1024
    },
    {
      "epoch": 0.07757438153351308,
      "grad_norm": 0.87890625,
      "learning_rate": 1.993264451395035e-05,
      "loss": 0.3001,
      "step": 1025
    },
    {
      "epoch": 0.07765006385696041,
      "grad_norm": 0.90234375,
      "learning_rate": 1.993250643370758e-05,
      "loss": 0.337,
      "step": 1026
    },
    {
      "epoch": 0.07772574618040774,
      "grad_norm": 3.4375,
      "learning_rate": 1.9932368212555168e-05,
      "loss": 0.6953,
      "step": 1027
    },
    {
      "epoch": 0.07780142850385507,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9932229850495067e-05,
      "loss": 0.3708,
      "step": 1028
    },
    {
      "epoch": 0.07787711082730239,
      "grad_norm": 0.92578125,
      "learning_rate": 1.993209134752924e-05,
      "loss": 0.3218,
      "step": 1029
    },
    {
      "epoch": 0.07795279315074972,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9931952703659655e-05,
      "loss": 0.3294,
      "step": 1030
    },
    {
      "epoch": 0.07802847547419706,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9931813918888273e-05,
      "loss": 0.3318,
      "step": 1031
    },
    {
      "epoch": 0.07810415779764439,
      "grad_norm": 0.90625,
      "learning_rate": 1.993167499321707e-05,
      "loss": 0.3243,
      "step": 1032
    },
    {
      "epoch": 0.07817984012109172,
      "grad_norm": 0.90625,
      "learning_rate": 1.9931535926648016e-05,
      "loss": 0.3519,
      "step": 1033
    },
    {
      "epoch": 0.07825552244453905,
      "grad_norm": 0.9296875,
      "learning_rate": 1.993139671918308e-05,
      "loss": 0.3414,
      "step": 1034
    },
    {
      "epoch": 0.07833120476798637,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9931257370824237e-05,
      "loss": 0.3233,
      "step": 1035
    },
    {
      "epoch": 0.0784068870914337,
      "grad_norm": 0.98046875,
      "learning_rate": 1.9931117881573466e-05,
      "loss": 0.3536,
      "step": 1036
    },
    {
      "epoch": 0.07848256941488103,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9930978251432748e-05,
      "loss": 0.3469,
      "step": 1037
    },
    {
      "epoch": 0.07855825173832837,
      "grad_norm": 0.85546875,
      "learning_rate": 1.993083848040406e-05,
      "loss": 0.2918,
      "step": 1038
    },
    {
      "epoch": 0.0786339340617757,
      "grad_norm": 0.98828125,
      "learning_rate": 1.9930698568489385e-05,
      "loss": 0.3598,
      "step": 1039
    },
    {
      "epoch": 0.07870961638522303,
      "grad_norm": 1.0390625,
      "learning_rate": 1.993055851569071e-05,
      "loss": 0.3683,
      "step": 1040
    },
    {
      "epoch": 0.07878529870867036,
      "grad_norm": 0.97265625,
      "learning_rate": 1.993041832201002e-05,
      "loss": 0.362,
      "step": 1041
    },
    {
      "epoch": 0.07886098103211768,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9930277987449307e-05,
      "loss": 0.2752,
      "step": 1042
    },
    {
      "epoch": 0.07893666335556501,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9930137512010558e-05,
      "loss": 0.3223,
      "step": 1043
    },
    {
      "epoch": 0.07901234567901234,
      "grad_norm": 3.21875,
      "learning_rate": 1.992999689569577e-05,
      "loss": 0.654,
      "step": 1044
    },
    {
      "epoch": 0.07908802800245968,
      "grad_norm": 0.98046875,
      "learning_rate": 1.9929856138506933e-05,
      "loss": 0.3568,
      "step": 1045
    },
    {
      "epoch": 0.07916371032590701,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9929715240446048e-05,
      "loss": 0.3181,
      "step": 1046
    },
    {
      "epoch": 0.07923939264935434,
      "grad_norm": 0.87890625,
      "learning_rate": 1.992957420151511e-05,
      "loss": 0.3179,
      "step": 1047
    },
    {
      "epoch": 0.07931507497280167,
      "grad_norm": 0.921875,
      "learning_rate": 1.9929433021716125e-05,
      "loss": 0.35,
      "step": 1048
    },
    {
      "epoch": 0.07939075729624899,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9929291701051096e-05,
      "loss": 0.3552,
      "step": 1049
    },
    {
      "epoch": 0.07946643961969632,
      "grad_norm": 0.98046875,
      "learning_rate": 1.9929150239522017e-05,
      "loss": 0.3777,
      "step": 1050
    },
    {
      "epoch": 0.07954212194314365,
      "grad_norm": 0.890625,
      "learning_rate": 1.9929008637130906e-05,
      "loss": 0.3325,
      "step": 1051
    },
    {
      "epoch": 0.07961780426659099,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9928866893879773e-05,
      "loss": 0.3074,
      "step": 1052
    },
    {
      "epoch": 0.07969348659003832,
      "grad_norm": 1.015625,
      "learning_rate": 1.992872500977062e-05,
      "loss": 0.3715,
      "step": 1053
    },
    {
      "epoch": 0.07976916891348565,
      "grad_norm": 0.96484375,
      "learning_rate": 1.9928582984805467e-05,
      "loss": 0.3506,
      "step": 1054
    },
    {
      "epoch": 0.07984485123693297,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9928440818986324e-05,
      "loss": 0.3315,
      "step": 1055
    },
    {
      "epoch": 0.0799205335603803,
      "grad_norm": 0.9296875,
      "learning_rate": 1.992829851231521e-05,
      "loss": 0.3571,
      "step": 1056
    },
    {
      "epoch": 0.07999621588382763,
      "grad_norm": 0.9609375,
      "learning_rate": 1.9928156064794148e-05,
      "loss": 0.4042,
      "step": 1057
    },
    {
      "epoch": 0.08007189820727496,
      "grad_norm": 0.98046875,
      "learning_rate": 1.992801347642515e-05,
      "loss": 0.3798,
      "step": 1058
    },
    {
      "epoch": 0.0801475805307223,
      "grad_norm": 0.9765625,
      "learning_rate": 1.9927870747210247e-05,
      "loss": 0.3656,
      "step": 1059
    },
    {
      "epoch": 0.08022326285416963,
      "grad_norm": 0.97265625,
      "learning_rate": 1.992772787715146e-05,
      "loss": 0.3394,
      "step": 1060
    },
    {
      "epoch": 0.08029894517761696,
      "grad_norm": 0.99609375,
      "learning_rate": 1.9927584866250815e-05,
      "loss": 0.359,
      "step": 1061
    },
    {
      "epoch": 0.08037462750106428,
      "grad_norm": 0.984375,
      "learning_rate": 1.9927441714510344e-05,
      "loss": 0.3533,
      "step": 1062
    },
    {
      "epoch": 0.08045030982451161,
      "grad_norm": 1.0625,
      "learning_rate": 1.9927298421932074e-05,
      "loss": 0.3771,
      "step": 1063
    },
    {
      "epoch": 0.08052599214795894,
      "grad_norm": 1.0546875,
      "learning_rate": 1.9927154988518046e-05,
      "loss": 0.4083,
      "step": 1064
    },
    {
      "epoch": 0.08060167447140627,
      "grad_norm": 4.75,
      "learning_rate": 1.9927011414270283e-05,
      "loss": 0.6568,
      "step": 1065
    },
    {
      "epoch": 0.0806773567948536,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9926867699190828e-05,
      "loss": 0.3587,
      "step": 1066
    },
    {
      "epoch": 0.08075303911830094,
      "grad_norm": 1.0,
      "learning_rate": 1.9926723843281723e-05,
      "loss": 0.3735,
      "step": 1067
    },
    {
      "epoch": 0.08082872144174826,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9926579846545e-05,
      "loss": 0.2796,
      "step": 1068
    },
    {
      "epoch": 0.08090440376519559,
      "grad_norm": 0.9453125,
      "learning_rate": 1.9926435708982713e-05,
      "loss": 0.3424,
      "step": 1069
    },
    {
      "epoch": 0.08098008608864292,
      "grad_norm": 0.9609375,
      "learning_rate": 1.99262914305969e-05,
      "loss": 0.3654,
      "step": 1070
    },
    {
      "epoch": 0.08105576841209025,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9926147011389606e-05,
      "loss": 0.339,
      "step": 1071
    },
    {
      "epoch": 0.08113145073553758,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9926002451362886e-05,
      "loss": 0.3089,
      "step": 1072
    },
    {
      "epoch": 0.08120713305898491,
      "grad_norm": 1.2265625,
      "learning_rate": 1.9925857750518785e-05,
      "loss": 0.3681,
      "step": 1073
    },
    {
      "epoch": 0.08128281538243225,
      "grad_norm": 1.0234375,
      "learning_rate": 1.992571290885936e-05,
      "loss": 0.2929,
      "step": 1074
    },
    {
      "epoch": 0.08135849770587956,
      "grad_norm": 0.90625,
      "learning_rate": 1.9925567926386664e-05,
      "loss": 0.3439,
      "step": 1075
    },
    {
      "epoch": 0.0814341800293269,
      "grad_norm": 0.9921875,
      "learning_rate": 1.9925422803102753e-05,
      "loss": 0.3808,
      "step": 1076
    },
    {
      "epoch": 0.08150986235277423,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9925277539009688e-05,
      "loss": 0.36,
      "step": 1077
    },
    {
      "epoch": 0.08158554467622156,
      "grad_norm": 0.9765625,
      "learning_rate": 1.992513213410953e-05,
      "loss": 0.4312,
      "step": 1078
    },
    {
      "epoch": 0.08166122699966889,
      "grad_norm": 0.89453125,
      "learning_rate": 1.992498658840434e-05,
      "loss": 0.3294,
      "step": 1079
    },
    {
      "epoch": 0.08173690932311622,
      "grad_norm": 0.96484375,
      "learning_rate": 1.992484090189618e-05,
      "loss": 0.3738,
      "step": 1080
    },
    {
      "epoch": 0.08181259164656354,
      "grad_norm": 3.25,
      "learning_rate": 1.9924695074587125e-05,
      "loss": 0.5879,
      "step": 1081
    },
    {
      "epoch": 0.08188827397001087,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9924549106479237e-05,
      "loss": 0.3683,
      "step": 1082
    },
    {
      "epoch": 0.0819639562934582,
      "grad_norm": 0.9296875,
      "learning_rate": 1.992440299757459e-05,
      "loss": 0.3378,
      "step": 1083
    },
    {
      "epoch": 0.08203963861690554,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9924256747875253e-05,
      "loss": 0.3356,
      "step": 1084
    },
    {
      "epoch": 0.08211532094035287,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9924110357383304e-05,
      "loss": 0.3506,
      "step": 1085
    },
    {
      "epoch": 0.0821910032638002,
      "grad_norm": 0.89453125,
      "learning_rate": 1.992396382610082e-05,
      "loss": 0.3318,
      "step": 1086
    },
    {
      "epoch": 0.08226668558724753,
      "grad_norm": 0.98046875,
      "learning_rate": 1.9923817154029882e-05,
      "loss": 0.3563,
      "step": 1087
    },
    {
      "epoch": 0.08234236791069485,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9923670341172564e-05,
      "loss": 0.3335,
      "step": 1088
    },
    {
      "epoch": 0.08241805023414218,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9923523387530953e-05,
      "loss": 0.3402,
      "step": 1089
    },
    {
      "epoch": 0.08249373255758952,
      "grad_norm": 1.046875,
      "learning_rate": 1.9923376293107133e-05,
      "loss": 0.3892,
      "step": 1090
    },
    {
      "epoch": 0.08256941488103685,
      "grad_norm": 0.83203125,
      "learning_rate": 1.992322905790319e-05,
      "loss": 0.2713,
      "step": 1091
    },
    {
      "epoch": 0.08264509720448418,
      "grad_norm": 0.93359375,
      "learning_rate": 1.992308168192122e-05,
      "loss": 0.3563,
      "step": 1092
    },
    {
      "epoch": 0.08272077952793151,
      "grad_norm": 1.03125,
      "learning_rate": 1.99229341651633e-05,
      "loss": 0.3881,
      "step": 1093
    },
    {
      "epoch": 0.08279646185137884,
      "grad_norm": 2.5625,
      "learning_rate": 1.9922786507631535e-05,
      "loss": 0.5227,
      "step": 1094
    },
    {
      "epoch": 0.08287214417482616,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9922638709328015e-05,
      "loss": 0.2893,
      "step": 1095
    },
    {
      "epoch": 0.0829478264982735,
      "grad_norm": 0.9453125,
      "learning_rate": 1.9922490770254835e-05,
      "loss": 0.3714,
      "step": 1096
    },
    {
      "epoch": 0.08302350882172083,
      "grad_norm": 1.1015625,
      "learning_rate": 1.99223426904141e-05,
      "loss": 0.4011,
      "step": 1097
    },
    {
      "epoch": 0.08309919114516816,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9922194469807902e-05,
      "loss": 0.3263,
      "step": 1098
    },
    {
      "epoch": 0.08317487346861549,
      "grad_norm": 0.9609375,
      "learning_rate": 1.9922046108438353e-05,
      "loss": 0.3745,
      "step": 1099
    },
    {
      "epoch": 0.08325055579206282,
      "grad_norm": 0.91796875,
      "learning_rate": 1.992189760630755e-05,
      "loss": 0.3431,
      "step": 1100
    },
    {
      "epoch": 0.08332623811551014,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9921748963417603e-05,
      "loss": 0.3631,
      "step": 1101
    },
    {
      "epoch": 0.08340192043895747,
      "grad_norm": 0.9765625,
      "learning_rate": 1.9921600179770623e-05,
      "loss": 0.4106,
      "step": 1102
    },
    {
      "epoch": 0.0834776027624048,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9921451255368716e-05,
      "loss": 0.3642,
      "step": 1103
    },
    {
      "epoch": 0.08355328508585214,
      "grad_norm": 0.921875,
      "learning_rate": 1.9921302190214e-05,
      "loss": 0.341,
      "step": 1104
    },
    {
      "epoch": 0.08362896740929947,
      "grad_norm": 1.0546875,
      "learning_rate": 1.992115298430858e-05,
      "loss": 0.4104,
      "step": 1105
    },
    {
      "epoch": 0.0837046497327468,
      "grad_norm": 0.84375,
      "learning_rate": 1.9921003637654584e-05,
      "loss": 0.3069,
      "step": 1106
    },
    {
      "epoch": 0.08378033205619413,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9920854150254126e-05,
      "loss": 0.3545,
      "step": 1107
    },
    {
      "epoch": 0.08385601437964145,
      "grad_norm": 1.03125,
      "learning_rate": 1.9920704522109328e-05,
      "loss": 0.3223,
      "step": 1108
    },
    {
      "epoch": 0.08393169670308878,
      "grad_norm": 0.8828125,
      "learning_rate": 1.992055475322231e-05,
      "loss": 0.342,
      "step": 1109
    },
    {
      "epoch": 0.08400737902653611,
      "grad_norm": 0.94921875,
      "learning_rate": 1.9920404843595204e-05,
      "loss": 0.3816,
      "step": 1110
    },
    {
      "epoch": 0.08408306134998345,
      "grad_norm": 0.9453125,
      "learning_rate": 1.9920254793230126e-05,
      "loss": 0.3626,
      "step": 1111
    },
    {
      "epoch": 0.08415874367343078,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9920104602129214e-05,
      "loss": 0.3479,
      "step": 1112
    },
    {
      "epoch": 0.08423442599687811,
      "grad_norm": 0.92578125,
      "learning_rate": 1.991995427029459e-05,
      "loss": 0.3576,
      "step": 1113
    },
    {
      "epoch": 0.08431010832032543,
      "grad_norm": 0.953125,
      "learning_rate": 1.9919803797728393e-05,
      "loss": 0.3229,
      "step": 1114
    },
    {
      "epoch": 0.08438579064377276,
      "grad_norm": 0.8984375,
      "learning_rate": 1.991965318443276e-05,
      "loss": 0.3417,
      "step": 1115
    },
    {
      "epoch": 0.08446147296722009,
      "grad_norm": 0.875,
      "learning_rate": 1.9919502430409818e-05,
      "loss": 0.3444,
      "step": 1116
    },
    {
      "epoch": 0.08453715529066742,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9919351535661712e-05,
      "loss": 0.3481,
      "step": 1117
    },
    {
      "epoch": 0.08461283761411476,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9919200500190588e-05,
      "loss": 0.3171,
      "step": 1118
    },
    {
      "epoch": 0.08468851993756209,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9919049323998577e-05,
      "loss": 0.2945,
      "step": 1119
    },
    {
      "epoch": 0.08476420226100942,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9918898007087834e-05,
      "loss": 0.3139,
      "step": 1120
    },
    {
      "epoch": 0.08483988458445674,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9918746549460497e-05,
      "loss": 0.3494,
      "step": 1121
    },
    {
      "epoch": 0.08491556690790407,
      "grad_norm": 0.99609375,
      "learning_rate": 1.9918594951118724e-05,
      "loss": 0.3829,
      "step": 1122
    },
    {
      "epoch": 0.0849912492313514,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9918443212064656e-05,
      "loss": 0.35,
      "step": 1123
    },
    {
      "epoch": 0.08506693155479873,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9918291332300455e-05,
      "loss": 0.3048,
      "step": 1124
    },
    {
      "epoch": 0.08514261387824607,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9918139311828266e-05,
      "loss": 0.3697,
      "step": 1125
    },
    {
      "epoch": 0.0852182962016934,
      "grad_norm": 0.96484375,
      "learning_rate": 1.9917987150650257e-05,
      "loss": 0.4194,
      "step": 1126
    },
    {
      "epoch": 0.08529397852514072,
      "grad_norm": 0.96875,
      "learning_rate": 1.9917834848768573e-05,
      "loss": 0.362,
      "step": 1127
    },
    {
      "epoch": 0.08536966084858805,
      "grad_norm": 0.90625,
      "learning_rate": 1.9917682406185388e-05,
      "loss": 0.339,
      "step": 1128
    },
    {
      "epoch": 0.08544534317203538,
      "grad_norm": 0.94921875,
      "learning_rate": 1.9917529822902855e-05,
      "loss": 0.3533,
      "step": 1129
    },
    {
      "epoch": 0.08552102549548271,
      "grad_norm": 0.9609375,
      "learning_rate": 1.9917377098923145e-05,
      "loss": 0.3551,
      "step": 1130
    },
    {
      "epoch": 0.08559670781893004,
      "grad_norm": 0.953125,
      "learning_rate": 1.9917224234248424e-05,
      "loss": 0.3815,
      "step": 1131
    },
    {
      "epoch": 0.08567239014237737,
      "grad_norm": 0.890625,
      "learning_rate": 1.9917071228880852e-05,
      "loss": 0.3185,
      "step": 1132
    },
    {
      "epoch": 0.0857480724658247,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9916918082822612e-05,
      "loss": 0.3135,
      "step": 1133
    },
    {
      "epoch": 0.08582375478927202,
      "grad_norm": 0.85546875,
      "learning_rate": 1.991676479607587e-05,
      "loss": 0.3119,
      "step": 1134
    },
    {
      "epoch": 0.08589943711271936,
      "grad_norm": 4.78125,
      "learning_rate": 1.99166113686428e-05,
      "loss": 0.6268,
      "step": 1135
    },
    {
      "epoch": 0.08597511943616669,
      "grad_norm": 0.96875,
      "learning_rate": 1.991645780052558e-05,
      "loss": 0.3447,
      "step": 1136
    },
    {
      "epoch": 0.08605080175961402,
      "grad_norm": 0.91015625,
      "learning_rate": 1.991630409172639e-05,
      "loss": 0.3156,
      "step": 1137
    },
    {
      "epoch": 0.08612648408306135,
      "grad_norm": 1.0078125,
      "learning_rate": 1.991615024224741e-05,
      "loss": 0.3634,
      "step": 1138
    },
    {
      "epoch": 0.08620216640650868,
      "grad_norm": 0.92578125,
      "learning_rate": 1.991599625209082e-05,
      "loss": 0.3391,
      "step": 1139
    },
    {
      "epoch": 0.08627784872995602,
      "grad_norm": 0.953125,
      "learning_rate": 1.991584212125881e-05,
      "loss": 0.3365,
      "step": 1140
    },
    {
      "epoch": 0.08635353105340333,
      "grad_norm": 0.921875,
      "learning_rate": 1.991568784975356e-05,
      "loss": 0.3585,
      "step": 1141
    },
    {
      "epoch": 0.08642921337685067,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9915533437577264e-05,
      "loss": 0.4057,
      "step": 1142
    },
    {
      "epoch": 0.086504895700298,
      "grad_norm": 1.0,
      "learning_rate": 1.991537888473211e-05,
      "loss": 0.3827,
      "step": 1143
    },
    {
      "epoch": 0.08658057802374533,
      "grad_norm": 1.015625,
      "learning_rate": 1.991522419122029e-05,
      "loss": 0.3639,
      "step": 1144
    },
    {
      "epoch": 0.08665626034719266,
      "grad_norm": 1.046875,
      "learning_rate": 1.9915069357044005e-05,
      "loss": 0.3318,
      "step": 1145
    },
    {
      "epoch": 0.08673194267064,
      "grad_norm": 0.9453125,
      "learning_rate": 1.991491438220544e-05,
      "loss": 0.3712,
      "step": 1146
    },
    {
      "epoch": 0.08680762499408731,
      "grad_norm": 0.875,
      "learning_rate": 1.99147592667068e-05,
      "loss": 0.3041,
      "step": 1147
    },
    {
      "epoch": 0.08688330731753464,
      "grad_norm": 0.9609375,
      "learning_rate": 1.991460401055029e-05,
      "loss": 0.3481,
      "step": 1148
    },
    {
      "epoch": 0.08695898964098198,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9914448613738107e-05,
      "loss": 0.3194,
      "step": 1149
    },
    {
      "epoch": 0.08703467196442931,
      "grad_norm": 0.90234375,
      "learning_rate": 1.991429307627245e-05,
      "loss": 0.3523,
      "step": 1150
    },
    {
      "epoch": 0.08711035428787664,
      "grad_norm": 3.359375,
      "learning_rate": 1.9914137398155542e-05,
      "loss": 0.5433,
      "step": 1151
    },
    {
      "epoch": 0.08718603661132397,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9913981579389578e-05,
      "loss": 0.3393,
      "step": 1152
    },
    {
      "epoch": 0.0872617189347713,
      "grad_norm": 2.078125,
      "learning_rate": 1.991382561997677e-05,
      "loss": 0.6192,
      "step": 1153
    },
    {
      "epoch": 0.08733740125821862,
      "grad_norm": 1.0,
      "learning_rate": 1.9913669519919337e-05,
      "loss": 0.3554,
      "step": 1154
    },
    {
      "epoch": 0.08741308358166595,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9913513279219485e-05,
      "loss": 0.3667,
      "step": 1155
    },
    {
      "epoch": 0.08748876590511329,
      "grad_norm": 0.87109375,
      "learning_rate": 1.991335689787944e-05,
      "loss": 0.3198,
      "step": 1156
    },
    {
      "epoch": 0.08756444822856062,
      "grad_norm": 0.921875,
      "learning_rate": 1.991320037590141e-05,
      "loss": 0.363,
      "step": 1157
    },
    {
      "epoch": 0.08764013055200795,
      "grad_norm": 0.953125,
      "learning_rate": 1.9913043713287622e-05,
      "loss": 0.3629,
      "step": 1158
    },
    {
      "epoch": 0.08771581287545528,
      "grad_norm": 0.88671875,
      "learning_rate": 1.99128869100403e-05,
      "loss": 0.3113,
      "step": 1159
    },
    {
      "epoch": 0.0877914951989026,
      "grad_norm": 1.015625,
      "learning_rate": 1.9912729966161666e-05,
      "loss": 0.3086,
      "step": 1160
    },
    {
      "epoch": 0.08786717752234993,
      "grad_norm": 2.765625,
      "learning_rate": 1.9912572881653945e-05,
      "loss": 0.5991,
      "step": 1161
    },
    {
      "epoch": 0.08794285984579726,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9912415656519364e-05,
      "loss": 0.3477,
      "step": 1162
    },
    {
      "epoch": 0.0880185421692446,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9912258290760164e-05,
      "loss": 0.3262,
      "step": 1163
    },
    {
      "epoch": 0.08809422449269193,
      "grad_norm": 2.46875,
      "learning_rate": 1.9912100784378566e-05,
      "loss": 0.605,
      "step": 1164
    },
    {
      "epoch": 0.08816990681613926,
      "grad_norm": 0.8984375,
      "learning_rate": 1.991194313737681e-05,
      "loss": 0.3336,
      "step": 1165
    },
    {
      "epoch": 0.08824558913958659,
      "grad_norm": 0.9609375,
      "learning_rate": 1.991178534975713e-05,
      "loss": 0.3727,
      "step": 1166
    },
    {
      "epoch": 0.08832127146303391,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9911627421521763e-05,
      "loss": 0.3434,
      "step": 1167
    },
    {
      "epoch": 0.08839695378648124,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9911469352672958e-05,
      "loss": 0.3453,
      "step": 1168
    },
    {
      "epoch": 0.08847263610992857,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9911311143212946e-05,
      "loss": 0.3155,
      "step": 1169
    },
    {
      "epoch": 0.0885483184333759,
      "grad_norm": 1.0,
      "learning_rate": 1.991115279314398e-05,
      "loss": 0.3798,
      "step": 1170
    },
    {
      "epoch": 0.08862400075682324,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9910994302468306e-05,
      "loss": 0.3602,
      "step": 1171
    },
    {
      "epoch": 0.08869968308027057,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9910835671188167e-05,
      "loss": 0.2927,
      "step": 1172
    },
    {
      "epoch": 0.08877536540371789,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9910676899305814e-05,
      "loss": 0.3186,
      "step": 1173
    },
    {
      "epoch": 0.08885104772716522,
      "grad_norm": 0.953125,
      "learning_rate": 1.9910517986823506e-05,
      "loss": 0.3651,
      "step": 1174
    },
    {
      "epoch": 0.08892673005061255,
      "grad_norm": 0.9609375,
      "learning_rate": 1.991035893374349e-05,
      "loss": 0.3578,
      "step": 1175
    },
    {
      "epoch": 0.08900241237405988,
      "grad_norm": 0.98828125,
      "learning_rate": 1.991019974006803e-05,
      "loss": 0.3729,
      "step": 1176
    },
    {
      "epoch": 0.08907809469750722,
      "grad_norm": 0.953125,
      "learning_rate": 1.9910040405799377e-05,
      "loss": 0.3555,
      "step": 1177
    },
    {
      "epoch": 0.08915377702095455,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9909880930939796e-05,
      "loss": 0.32,
      "step": 1178
    },
    {
      "epoch": 0.08922945934440188,
      "grad_norm": 0.98046875,
      "learning_rate": 1.9909721315491547e-05,
      "loss": 0.3806,
      "step": 1179
    },
    {
      "epoch": 0.0893051416678492,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9909561559456896e-05,
      "loss": 0.2983,
      "step": 1180
    },
    {
      "epoch": 0.08938082399129653,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9909401662838107e-05,
      "loss": 0.3284,
      "step": 1181
    },
    {
      "epoch": 0.08945650631474386,
      "grad_norm": 0.96484375,
      "learning_rate": 1.9909241625637452e-05,
      "loss": 0.3312,
      "step": 1182
    },
    {
      "epoch": 0.08953218863819119,
      "grad_norm": 0.9453125,
      "learning_rate": 1.9909081447857202e-05,
      "loss": 0.3449,
      "step": 1183
    },
    {
      "epoch": 0.08960787096163852,
      "grad_norm": 0.890625,
      "learning_rate": 1.9908921129499624e-05,
      "loss": 0.3488,
      "step": 1184
    },
    {
      "epoch": 0.08968355328508586,
      "grad_norm": 0.890625,
      "learning_rate": 1.9908760670566994e-05,
      "loss": 0.3126,
      "step": 1185
    },
    {
      "epoch": 0.08975923560853319,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9908600071061593e-05,
      "loss": 0.3421,
      "step": 1186
    },
    {
      "epoch": 0.0898349179319805,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9908439330985697e-05,
      "loss": 0.3024,
      "step": 1187
    },
    {
      "epoch": 0.08991060025542784,
      "grad_norm": 0.96484375,
      "learning_rate": 1.9908278450341584e-05,
      "loss": 0.3767,
      "step": 1188
    },
    {
      "epoch": 0.08998628257887517,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9908117429131536e-05,
      "loss": 0.33,
      "step": 1189
    },
    {
      "epoch": 0.0900619649023225,
      "grad_norm": 0.9140625,
      "learning_rate": 1.990795626735784e-05,
      "loss": 0.3436,
      "step": 1190
    },
    {
      "epoch": 0.09013764722576983,
      "grad_norm": 0.828125,
      "learning_rate": 1.9907794965022783e-05,
      "loss": 0.2992,
      "step": 1191
    },
    {
      "epoch": 0.09021332954921717,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9907633522128655e-05,
      "loss": 0.33,
      "step": 1192
    },
    {
      "epoch": 0.09028901187266448,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9907471938677737e-05,
      "loss": 0.3401,
      "step": 1193
    },
    {
      "epoch": 0.09036469419611182,
      "grad_norm": 0.95703125,
      "learning_rate": 1.9907310214672334e-05,
      "loss": 0.3444,
      "step": 1194
    },
    {
      "epoch": 0.09044037651955915,
      "grad_norm": 1.0703125,
      "learning_rate": 1.9907148350114732e-05,
      "loss": 0.3192,
      "step": 1195
    },
    {
      "epoch": 0.09051605884300648,
      "grad_norm": 0.921875,
      "learning_rate": 1.990698634500723e-05,
      "loss": 0.3068,
      "step": 1196
    },
    {
      "epoch": 0.09059174116645381,
      "grad_norm": 0.9921875,
      "learning_rate": 1.990682419935212e-05,
      "loss": 0.3781,
      "step": 1197
    },
    {
      "epoch": 0.09066742348990114,
      "grad_norm": 0.90625,
      "learning_rate": 1.9906661913151716e-05,
      "loss": 0.3621,
      "step": 1198
    },
    {
      "epoch": 0.09074310581334848,
      "grad_norm": 3.421875,
      "learning_rate": 1.9906499486408308e-05,
      "loss": 0.6082,
      "step": 1199
    },
    {
      "epoch": 0.0908187881367958,
      "grad_norm": 2.609375,
      "learning_rate": 1.9906336919124204e-05,
      "loss": 0.5052,
      "step": 1200
    },
    {
      "epoch": 0.09089447046024313,
      "grad_norm": 0.9609375,
      "learning_rate": 1.9906174211301715e-05,
      "loss": 0.3025,
      "step": 1201
    },
    {
      "epoch": 0.09097015278369046,
      "grad_norm": 0.94921875,
      "learning_rate": 1.990601136294314e-05,
      "loss": 0.3262,
      "step": 1202
    },
    {
      "epoch": 0.09104583510713779,
      "grad_norm": 0.87109375,
      "learning_rate": 1.99058483740508e-05,
      "loss": 0.3044,
      "step": 1203
    },
    {
      "epoch": 0.09112151743058512,
      "grad_norm": 0.921875,
      "learning_rate": 1.9905685244627e-05,
      "loss": 0.3111,
      "step": 1204
    },
    {
      "epoch": 0.09119719975403245,
      "grad_norm": 1.03125,
      "learning_rate": 1.9905521974674055e-05,
      "loss": 0.3582,
      "step": 1205
    },
    {
      "epoch": 0.09127288207747977,
      "grad_norm": 1.0078125,
      "learning_rate": 1.990535856419428e-05,
      "loss": 0.3224,
      "step": 1206
    },
    {
      "epoch": 0.0913485644009271,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9905195013189997e-05,
      "loss": 0.3416,
      "step": 1207
    },
    {
      "epoch": 0.09142424672437444,
      "grad_norm": 0.9453125,
      "learning_rate": 1.9905031321663525e-05,
      "loss": 0.3292,
      "step": 1208
    },
    {
      "epoch": 0.09149992904782177,
      "grad_norm": 0.96875,
      "learning_rate": 1.9904867489617186e-05,
      "loss": 0.3345,
      "step": 1209
    },
    {
      "epoch": 0.0915756113712691,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9904703517053306e-05,
      "loss": 0.3485,
      "step": 1210
    },
    {
      "epoch": 0.09165129369471643,
      "grad_norm": 0.93359375,
      "learning_rate": 1.990453940397421e-05,
      "loss": 0.3735,
      "step": 1211
    },
    {
      "epoch": 0.09172697601816376,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9904375150382223e-05,
      "loss": 0.3538,
      "step": 1212
    },
    {
      "epoch": 0.09180265834161108,
      "grad_norm": 0.89453125,
      "learning_rate": 1.990421075627968e-05,
      "loss": 0.326,
      "step": 1213
    },
    {
      "epoch": 0.09187834066505841,
      "grad_norm": 0.953125,
      "learning_rate": 1.9904046221668912e-05,
      "loss": 0.3567,
      "step": 1214
    },
    {
      "epoch": 0.09195402298850575,
      "grad_norm": 0.9453125,
      "learning_rate": 1.990388154655225e-05,
      "loss": 0.358,
      "step": 1215
    },
    {
      "epoch": 0.09202970531195308,
      "grad_norm": 1.171875,
      "learning_rate": 1.9903716730932036e-05,
      "loss": 0.3086,
      "step": 1216
    },
    {
      "epoch": 0.09210538763540041,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9903551774810605e-05,
      "loss": 0.3333,
      "step": 1217
    },
    {
      "epoch": 0.09218106995884774,
      "grad_norm": 0.88671875,
      "learning_rate": 1.990338667819029e-05,
      "loss": 0.2877,
      "step": 1218
    },
    {
      "epoch": 0.09225675228229506,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9903221441073448e-05,
      "loss": 0.3373,
      "step": 1219
    },
    {
      "epoch": 0.09233243460574239,
      "grad_norm": 0.93359375,
      "learning_rate": 1.990305606346241e-05,
      "loss": 0.371,
      "step": 1220
    },
    {
      "epoch": 0.09240811692918972,
      "grad_norm": 0.92578125,
      "learning_rate": 1.990289054535953e-05,
      "loss": 0.3781,
      "step": 1221
    },
    {
      "epoch": 0.09248379925263706,
      "grad_norm": 0.875,
      "learning_rate": 1.9902724886767152e-05,
      "loss": 0.3388,
      "step": 1222
    },
    {
      "epoch": 0.09255948157608439,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9902559087687628e-05,
      "loss": 0.3448,
      "step": 1223
    },
    {
      "epoch": 0.09263516389953172,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9902393148123313e-05,
      "loss": 0.324,
      "step": 1224
    },
    {
      "epoch": 0.09271084622297905,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9902227068076558e-05,
      "loss": 0.2996,
      "step": 1225
    },
    {
      "epoch": 0.09278652854642637,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9902060847549716e-05,
      "loss": 0.3755,
      "step": 1226
    },
    {
      "epoch": 0.0928622108698737,
      "grad_norm": 0.92578125,
      "learning_rate": 1.990189448654515e-05,
      "loss": 0.3692,
      "step": 1227
    },
    {
      "epoch": 0.09293789319332103,
      "grad_norm": 1.078125,
      "learning_rate": 1.9901727985065216e-05,
      "loss": 0.3504,
      "step": 1228
    },
    {
      "epoch": 0.09301357551676837,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9901561343112284e-05,
      "loss": 0.3251,
      "step": 1229
    },
    {
      "epoch": 0.0930892578402157,
      "grad_norm": 0.86328125,
      "learning_rate": 1.990139456068871e-05,
      "loss": 0.3505,
      "step": 1230
    },
    {
      "epoch": 0.09316494016366303,
      "grad_norm": 1.5859375,
      "learning_rate": 1.9901227637796862e-05,
      "loss": 0.3302,
      "step": 1231
    },
    {
      "epoch": 0.09324062248711036,
      "grad_norm": 1.3984375,
      "learning_rate": 1.990106057443911e-05,
      "loss": 0.3407,
      "step": 1232
    },
    {
      "epoch": 0.09331630481055768,
      "grad_norm": 0.8828125,
      "learning_rate": 1.990089337061782e-05,
      "loss": 0.3539,
      "step": 1233
    },
    {
      "epoch": 0.09339198713400501,
      "grad_norm": 0.8671875,
      "learning_rate": 1.990072602633537e-05,
      "loss": 0.3246,
      "step": 1234
    },
    {
      "epoch": 0.09346766945745234,
      "grad_norm": 0.90625,
      "learning_rate": 1.990055854159413e-05,
      "loss": 0.3173,
      "step": 1235
    },
    {
      "epoch": 0.09354335178089968,
      "grad_norm": 1.1640625,
      "learning_rate": 1.9900390916396478e-05,
      "loss": 0.3823,
      "step": 1236
    },
    {
      "epoch": 0.09361903410434701,
      "grad_norm": 0.9296875,
      "learning_rate": 1.990022315074479e-05,
      "loss": 0.3682,
      "step": 1237
    },
    {
      "epoch": 0.09369471642779434,
      "grad_norm": 0.953125,
      "learning_rate": 1.9900055244641448e-05,
      "loss": 0.3503,
      "step": 1238
    },
    {
      "epoch": 0.09377039875124166,
      "grad_norm": 0.90625,
      "learning_rate": 1.9899887198088833e-05,
      "loss": 0.3366,
      "step": 1239
    },
    {
      "epoch": 0.09384608107468899,
      "grad_norm": 0.96875,
      "learning_rate": 1.989971901108933e-05,
      "loss": 0.3666,
      "step": 1240
    },
    {
      "epoch": 0.09392176339813632,
      "grad_norm": 0.9453125,
      "learning_rate": 1.9899550683645325e-05,
      "loss": 0.3676,
      "step": 1241
    },
    {
      "epoch": 0.09399744572158365,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9899382215759202e-05,
      "loss": 0.3201,
      "step": 1242
    },
    {
      "epoch": 0.09407312804503098,
      "grad_norm": 0.9375,
      "learning_rate": 1.9899213607433357e-05,
      "loss": 0.3366,
      "step": 1243
    },
    {
      "epoch": 0.09414881036847832,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9899044858670178e-05,
      "loss": 0.3319,
      "step": 1244
    },
    {
      "epoch": 0.09422449269192565,
      "grad_norm": 0.85546875,
      "learning_rate": 1.989887596947206e-05,
      "loss": 0.3289,
      "step": 1245
    },
    {
      "epoch": 0.09430017501537297,
      "grad_norm": 0.890625,
      "learning_rate": 1.98987069398414e-05,
      "loss": 0.3422,
      "step": 1246
    },
    {
      "epoch": 0.0943758573388203,
      "grad_norm": 0.99609375,
      "learning_rate": 1.98985377697806e-05,
      "loss": 0.3863,
      "step": 1247
    },
    {
      "epoch": 0.09445153966226763,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9898368459292047e-05,
      "loss": 0.3585,
      "step": 1248
    },
    {
      "epoch": 0.09452722198571496,
      "grad_norm": 0.9375,
      "learning_rate": 1.9898199008378155e-05,
      "loss": 0.3744,
      "step": 1249
    },
    {
      "epoch": 0.0946029043091623,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9898029417041328e-05,
      "loss": 0.3215,
      "step": 1250
    },
    {
      "epoch": 0.09467858663260963,
      "grad_norm": 1.203125,
      "learning_rate": 1.9897859685283962e-05,
      "loss": 0.3393,
      "step": 1251
    },
    {
      "epoch": 0.09475426895605694,
      "grad_norm": 0.890625,
      "learning_rate": 1.9897689813108476e-05,
      "loss": 0.3376,
      "step": 1252
    },
    {
      "epoch": 0.09482995127950428,
      "grad_norm": 0.91796875,
      "learning_rate": 1.989751980051727e-05,
      "loss": 0.3296,
      "step": 1253
    },
    {
      "epoch": 0.09490563360295161,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9897349647512762e-05,
      "loss": 0.2812,
      "step": 1254
    },
    {
      "epoch": 0.09498131592639894,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9897179354097366e-05,
      "loss": 0.3848,
      "step": 1255
    },
    {
      "epoch": 0.09505699824984627,
      "grad_norm": 0.88671875,
      "learning_rate": 1.98970089202735e-05,
      "loss": 0.3526,
      "step": 1256
    },
    {
      "epoch": 0.0951326805732936,
      "grad_norm": 0.9921875,
      "learning_rate": 1.9896838346043574e-05,
      "loss": 0.3867,
      "step": 1257
    },
    {
      "epoch": 0.09520836289674094,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9896667631410013e-05,
      "loss": 0.3833,
      "step": 1258
    },
    {
      "epoch": 0.09528404522018825,
      "grad_norm": 0.875,
      "learning_rate": 1.989649677637524e-05,
      "loss": 0.2948,
      "step": 1259
    },
    {
      "epoch": 0.09535972754363559,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9896325780941677e-05,
      "loss": 0.3575,
      "step": 1260
    },
    {
      "epoch": 0.09543540986708292,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9896154645111753e-05,
      "loss": 0.3494,
      "step": 1261
    },
    {
      "epoch": 0.09551109219053025,
      "grad_norm": 0.953125,
      "learning_rate": 1.9895983368887884e-05,
      "loss": 0.3281,
      "step": 1262
    },
    {
      "epoch": 0.09558677451397758,
      "grad_norm": 1.703125,
      "learning_rate": 1.9895811952272515e-05,
      "loss": 0.3598,
      "step": 1263
    },
    {
      "epoch": 0.09566245683742491,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9895640395268073e-05,
      "loss": 0.297,
      "step": 1264
    },
    {
      "epoch": 0.09573813916087223,
      "grad_norm": 0.87109375,
      "learning_rate": 1.989546869787699e-05,
      "loss": 0.3141,
      "step": 1265
    },
    {
      "epoch": 0.09581382148431956,
      "grad_norm": 0.8515625,
      "learning_rate": 1.98952968601017e-05,
      "loss": 0.3237,
      "step": 1266
    },
    {
      "epoch": 0.0958895038077669,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9895124881944644e-05,
      "loss": 0.3791,
      "step": 1267
    },
    {
      "epoch": 0.09596518613121423,
      "grad_norm": 0.8828125,
      "learning_rate": 1.989495276340826e-05,
      "loss": 0.3047,
      "step": 1268
    },
    {
      "epoch": 0.09604086845466156,
      "grad_norm": 1.046875,
      "learning_rate": 1.9894780504494995e-05,
      "loss": 0.3212,
      "step": 1269
    },
    {
      "epoch": 0.09611655077810889,
      "grad_norm": 0.96875,
      "learning_rate": 1.9894608105207285e-05,
      "loss": 0.3588,
      "step": 1270
    },
    {
      "epoch": 0.09619223310155622,
      "grad_norm": 1.140625,
      "learning_rate": 1.989443556554758e-05,
      "loss": 0.3152,
      "step": 1271
    },
    {
      "epoch": 0.09626791542500354,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9894262885518326e-05,
      "loss": 0.3489,
      "step": 1272
    },
    {
      "epoch": 0.09634359774845087,
      "grad_norm": 0.98046875,
      "learning_rate": 1.9894090065121972e-05,
      "loss": 0.3967,
      "step": 1273
    },
    {
      "epoch": 0.0964192800718982,
      "grad_norm": 0.859375,
      "learning_rate": 1.9893917104360977e-05,
      "loss": 0.3264,
      "step": 1274
    },
    {
      "epoch": 0.09649496239534554,
      "grad_norm": 0.96484375,
      "learning_rate": 1.9893744003237786e-05,
      "loss": 0.3823,
      "step": 1275
    },
    {
      "epoch": 0.09657064471879287,
      "grad_norm": 0.91015625,
      "learning_rate": 1.989357076175486e-05,
      "loss": 0.3194,
      "step": 1276
    },
    {
      "epoch": 0.0966463270422402,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9893397379914653e-05,
      "loss": 0.3245,
      "step": 1277
    },
    {
      "epoch": 0.09672200936568753,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9893223857719627e-05,
      "loss": 0.3,
      "step": 1278
    },
    {
      "epoch": 0.09679769168913485,
      "grad_norm": 1.046875,
      "learning_rate": 1.989305019517224e-05,
      "loss": 0.4236,
      "step": 1279
    },
    {
      "epoch": 0.09687337401258218,
      "grad_norm": 0.98828125,
      "learning_rate": 1.9892876392274966e-05,
      "loss": 0.3957,
      "step": 1280
    },
    {
      "epoch": 0.09694905633602952,
      "grad_norm": 0.859375,
      "learning_rate": 1.989270244903026e-05,
      "loss": 0.3111,
      "step": 1281
    },
    {
      "epoch": 0.09702473865947685,
      "grad_norm": 0.921875,
      "learning_rate": 1.989252836544059e-05,
      "loss": 0.373,
      "step": 1282
    },
    {
      "epoch": 0.09710042098292418,
      "grad_norm": 0.98828125,
      "learning_rate": 1.989235414150843e-05,
      "loss": 0.3485,
      "step": 1283
    },
    {
      "epoch": 0.09717610330637151,
      "grad_norm": 0.94921875,
      "learning_rate": 1.9892179777236255e-05,
      "loss": 0.393,
      "step": 1284
    },
    {
      "epoch": 0.09725178562981883,
      "grad_norm": 0.87890625,
      "learning_rate": 1.989200527262653e-05,
      "loss": 0.3259,
      "step": 1285
    },
    {
      "epoch": 0.09732746795326616,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9891830627681738e-05,
      "loss": 0.3893,
      "step": 1286
    },
    {
      "epoch": 0.0974031502767135,
      "grad_norm": 0.953125,
      "learning_rate": 1.989165584240435e-05,
      "loss": 0.3573,
      "step": 1287
    },
    {
      "epoch": 0.09747883260016083,
      "grad_norm": 1.015625,
      "learning_rate": 1.9891480916796854e-05,
      "loss": 0.4151,
      "step": 1288
    },
    {
      "epoch": 0.09755451492360816,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9891305850861723e-05,
      "loss": 0.3307,
      "step": 1289
    },
    {
      "epoch": 0.09763019724705549,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9891130644601447e-05,
      "loss": 0.3355,
      "step": 1290
    },
    {
      "epoch": 0.09770587957050282,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9890955298018507e-05,
      "loss": 0.3669,
      "step": 1291
    },
    {
      "epoch": 0.09778156189395014,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9890779811115395e-05,
      "loss": 0.3337,
      "step": 1292
    },
    {
      "epoch": 0.09785724421739747,
      "grad_norm": 3.703125,
      "learning_rate": 1.9890604183894597e-05,
      "loss": 0.6069,
      "step": 1293
    },
    {
      "epoch": 0.0979329265408448,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9890428416358603e-05,
      "loss": 0.3069,
      "step": 1294
    },
    {
      "epoch": 0.09800860886429213,
      "grad_norm": 0.9453125,
      "learning_rate": 1.9890252508509915e-05,
      "loss": 0.3752,
      "step": 1295
    },
    {
      "epoch": 0.09808429118773947,
      "grad_norm": 0.91796875,
      "learning_rate": 1.989007646035102e-05,
      "loss": 0.3383,
      "step": 1296
    },
    {
      "epoch": 0.0981599735111868,
      "grad_norm": 0.875,
      "learning_rate": 1.9889900271884417e-05,
      "loss": 0.3412,
      "step": 1297
    },
    {
      "epoch": 0.09823565583463412,
      "grad_norm": 0.921875,
      "learning_rate": 1.988972394311261e-05,
      "loss": 0.3468,
      "step": 1298
    },
    {
      "epoch": 0.09831133815808145,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9889547474038094e-05,
      "loss": 0.3162,
      "step": 1299
    },
    {
      "epoch": 0.09838702048152878,
      "grad_norm": 0.90625,
      "learning_rate": 1.9889370864663377e-05,
      "loss": 0.3562,
      "step": 1300
    },
    {
      "epoch": 0.09846270280497611,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9889194114990963e-05,
      "loss": 0.3937,
      "step": 1301
    },
    {
      "epoch": 0.09853838512842344,
      "grad_norm": 0.98828125,
      "learning_rate": 1.9889017225023364e-05,
      "loss": 0.3661,
      "step": 1302
    },
    {
      "epoch": 0.09861406745187078,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9888840194763084e-05,
      "loss": 0.3365,
      "step": 1303
    },
    {
      "epoch": 0.09868974977531811,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9888663024212634e-05,
      "loss": 0.3522,
      "step": 1304
    },
    {
      "epoch": 0.09876543209876543,
      "grad_norm": 0.9140625,
      "learning_rate": 1.988848571337453e-05,
      "loss": 0.3576,
      "step": 1305
    },
    {
      "epoch": 0.09884111442221276,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9888308262251286e-05,
      "loss": 0.3235,
      "step": 1306
    },
    {
      "epoch": 0.09891679674566009,
      "grad_norm": 0.91015625,
      "learning_rate": 1.988813067084542e-05,
      "loss": 0.3622,
      "step": 1307
    },
    {
      "epoch": 0.09899247906910742,
      "grad_norm": 0.9765625,
      "learning_rate": 1.9887952939159454e-05,
      "loss": 0.3504,
      "step": 1308
    },
    {
      "epoch": 0.09906816139255475,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9887775067195908e-05,
      "loss": 0.3198,
      "step": 1309
    },
    {
      "epoch": 0.09914384371600209,
      "grad_norm": 0.890625,
      "learning_rate": 1.9887597054957304e-05,
      "loss": 0.3141,
      "step": 1310
    },
    {
      "epoch": 0.0992195260394494,
      "grad_norm": 0.91015625,
      "learning_rate": 1.988741890244617e-05,
      "loss": 0.3548,
      "step": 1311
    },
    {
      "epoch": 0.09929520836289674,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9887240609665026e-05,
      "loss": 0.3194,
      "step": 1312
    },
    {
      "epoch": 0.09937089068634407,
      "grad_norm": 0.875,
      "learning_rate": 1.988706217661641e-05,
      "loss": 0.3017,
      "step": 1313
    },
    {
      "epoch": 0.0994465730097914,
      "grad_norm": 0.89453125,
      "learning_rate": 1.988688360330285e-05,
      "loss": 0.3521,
      "step": 1314
    },
    {
      "epoch": 0.09952225533323873,
      "grad_norm": 0.8515625,
      "learning_rate": 1.988670488972688e-05,
      "loss": 0.3294,
      "step": 1315
    },
    {
      "epoch": 0.09959793765668606,
      "grad_norm": 0.90625,
      "learning_rate": 1.9886526035891034e-05,
      "loss": 0.3689,
      "step": 1316
    },
    {
      "epoch": 0.0996736199801334,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9886347041797852e-05,
      "loss": 0.3109,
      "step": 1317
    },
    {
      "epoch": 0.09974930230358071,
      "grad_norm": 0.921875,
      "learning_rate": 1.988616790744987e-05,
      "loss": 0.3409,
      "step": 1318
    },
    {
      "epoch": 0.09982498462702805,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9885988632849634e-05,
      "loss": 0.3369,
      "step": 1319
    },
    {
      "epoch": 0.09990066695047538,
      "grad_norm": 0.93359375,
      "learning_rate": 1.988580921799968e-05,
      "loss": 0.3599,
      "step": 1320
    },
    {
      "epoch": 0.09997634927392271,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9885629662902558e-05,
      "loss": 0.376,
      "step": 1321
    },
    {
      "epoch": 0.10005203159737004,
      "grad_norm": 1.28125,
      "learning_rate": 1.9885449967560818e-05,
      "loss": 0.35,
      "step": 1322
    },
    {
      "epoch": 0.10005203159737004,
      "eval_loss": 0.3635263442993164,
      "eval_runtime": 83.398,
      "eval_samples_per_second": 58.299,
      "eval_steps_per_second": 58.299,
      "step": 1322
    },
    {
      "epoch": 0.10012771392081737,
      "grad_norm": 0.99609375,
      "learning_rate": 1.9885270131977005e-05,
      "loss": 0.3813,
      "step": 1323
    },
    {
      "epoch": 0.1002033962442647,
      "grad_norm": 0.953125,
      "learning_rate": 1.988509015615367e-05,
      "loss": 0.3917,
      "step": 1324
    },
    {
      "epoch": 0.10027907856771202,
      "grad_norm": 0.9140625,
      "learning_rate": 1.988491004009337e-05,
      "loss": 0.3714,
      "step": 1325
    },
    {
      "epoch": 0.10035476089115936,
      "grad_norm": 0.9453125,
      "learning_rate": 1.9884729783798657e-05,
      "loss": 0.3679,
      "step": 1326
    },
    {
      "epoch": 0.10043044321460669,
      "grad_norm": 0.96875,
      "learning_rate": 1.988454938727209e-05,
      "loss": 0.3629,
      "step": 1327
    },
    {
      "epoch": 0.10050612553805402,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9884368850516226e-05,
      "loss": 0.3793,
      "step": 1328
    },
    {
      "epoch": 0.10058180786150135,
      "grad_norm": 0.96484375,
      "learning_rate": 1.9884188173533628e-05,
      "loss": 0.2992,
      "step": 1329
    },
    {
      "epoch": 0.10065749018494868,
      "grad_norm": 1.078125,
      "learning_rate": 1.988400735632686e-05,
      "loss": 0.3598,
      "step": 1330
    },
    {
      "epoch": 0.100733172508396,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9883826398898486e-05,
      "loss": 0.3551,
      "step": 1331
    },
    {
      "epoch": 0.10080885483184333,
      "grad_norm": 0.85546875,
      "learning_rate": 1.988364530125107e-05,
      "loss": 0.3215,
      "step": 1332
    },
    {
      "epoch": 0.10088453715529067,
      "grad_norm": 0.921875,
      "learning_rate": 1.9883464063387187e-05,
      "loss": 0.3456,
      "step": 1333
    },
    {
      "epoch": 0.100960219478738,
      "grad_norm": 0.921875,
      "learning_rate": 1.9883282685309405e-05,
      "loss": 0.3619,
      "step": 1334
    },
    {
      "epoch": 0.10103590180218533,
      "grad_norm": 0.98828125,
      "learning_rate": 1.98831011670203e-05,
      "loss": 0.3638,
      "step": 1335
    },
    {
      "epoch": 0.10111158412563266,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9882919508522445e-05,
      "loss": 0.3056,
      "step": 1336
    },
    {
      "epoch": 0.10118726644908,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9882737709818414e-05,
      "loss": 0.2907,
      "step": 1337
    },
    {
      "epoch": 0.10126294877252731,
      "grad_norm": 1.09375,
      "learning_rate": 1.9882555770910793e-05,
      "loss": 0.3535,
      "step": 1338
    },
    {
      "epoch": 0.10133863109597464,
      "grad_norm": 1.046875,
      "learning_rate": 1.9882373691802156e-05,
      "loss": 0.3805,
      "step": 1339
    },
    {
      "epoch": 0.10141431341942198,
      "grad_norm": 0.9140625,
      "learning_rate": 1.988219147249509e-05,
      "loss": 0.3563,
      "step": 1340
    },
    {
      "epoch": 0.10148999574286931,
      "grad_norm": 0.88671875,
      "learning_rate": 1.988200911299218e-05,
      "loss": 0.2908,
      "step": 1341
    },
    {
      "epoch": 0.10156567806631664,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9881826613296012e-05,
      "loss": 0.3494,
      "step": 1342
    },
    {
      "epoch": 0.10164136038976397,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9881643973409176e-05,
      "loss": 0.3347,
      "step": 1343
    },
    {
      "epoch": 0.10171704271321129,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9881461193334263e-05,
      "loss": 0.3428,
      "step": 1344
    },
    {
      "epoch": 0.10179272503665862,
      "grad_norm": 0.9765625,
      "learning_rate": 1.9881278273073866e-05,
      "loss": 0.4033,
      "step": 1345
    },
    {
      "epoch": 0.10186840736010595,
      "grad_norm": 0.9609375,
      "learning_rate": 1.988109521263058e-05,
      "loss": 0.4084,
      "step": 1346
    },
    {
      "epoch": 0.10194408968355329,
      "grad_norm": 0.9375,
      "learning_rate": 1.9880912012007e-05,
      "loss": 0.4099,
      "step": 1347
    },
    {
      "epoch": 0.10201977200700062,
      "grad_norm": 0.9765625,
      "learning_rate": 1.988072867120573e-05,
      "loss": 0.3799,
      "step": 1348
    },
    {
      "epoch": 0.10209545433044795,
      "grad_norm": 0.953125,
      "learning_rate": 1.9880545190229367e-05,
      "loss": 0.3609,
      "step": 1349
    },
    {
      "epoch": 0.10217113665389528,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9880361569080516e-05,
      "loss": 0.3397,
      "step": 1350
    },
    {
      "epoch": 0.1022468189773426,
      "grad_norm": 0.9453125,
      "learning_rate": 1.988017780776178e-05,
      "loss": 0.3289,
      "step": 1351
    },
    {
      "epoch": 0.10232250130078993,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9879993906275767e-05,
      "loss": 0.3632,
      "step": 1352
    },
    {
      "epoch": 0.10239818362423726,
      "grad_norm": 0.8671875,
      "learning_rate": 1.987980986462508e-05,
      "loss": 0.3141,
      "step": 1353
    },
    {
      "epoch": 0.1024738659476846,
      "grad_norm": 0.83984375,
      "learning_rate": 1.987962568281234e-05,
      "loss": 0.3365,
      "step": 1354
    },
    {
      "epoch": 0.10254954827113193,
      "grad_norm": 0.86328125,
      "learning_rate": 1.987944136084016e-05,
      "loss": 0.335,
      "step": 1355
    },
    {
      "epoch": 0.10262523059457926,
      "grad_norm": 0.875,
      "learning_rate": 1.9879256898711147e-05,
      "loss": 0.3359,
      "step": 1356
    },
    {
      "epoch": 0.10270091291802658,
      "grad_norm": 0.9140625,
      "learning_rate": 1.987907229642792e-05,
      "loss": 0.3863,
      "step": 1357
    },
    {
      "epoch": 0.10277659524147391,
      "grad_norm": 0.828125,
      "learning_rate": 1.98788875539931e-05,
      "loss": 0.3004,
      "step": 1358
    },
    {
      "epoch": 0.10285227756492124,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9878702671409307e-05,
      "loss": 0.3263,
      "step": 1359
    },
    {
      "epoch": 0.10292795988836857,
      "grad_norm": 0.9140625,
      "learning_rate": 1.987851764867917e-05,
      "loss": 0.3652,
      "step": 1360
    },
    {
      "epoch": 0.1030036422118159,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9878332485805298e-05,
      "loss": 0.349,
      "step": 1361
    },
    {
      "epoch": 0.10307932453526324,
      "grad_norm": 0.7890625,
      "learning_rate": 1.9878147182790334e-05,
      "loss": 0.2997,
      "step": 1362
    },
    {
      "epoch": 0.10315500685871057,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9877961739636903e-05,
      "loss": 0.3623,
      "step": 1363
    },
    {
      "epoch": 0.10323068918215789,
      "grad_norm": 1.03125,
      "learning_rate": 1.9877776156347627e-05,
      "loss": 0.3701,
      "step": 1364
    },
    {
      "epoch": 0.10330637150560522,
      "grad_norm": 0.9609375,
      "learning_rate": 1.987759043292515e-05,
      "loss": 0.3864,
      "step": 1365
    },
    {
      "epoch": 0.10338205382905255,
      "grad_norm": 0.84765625,
      "learning_rate": 1.98774045693721e-05,
      "loss": 0.3243,
      "step": 1366
    },
    {
      "epoch": 0.10345773615249988,
      "grad_norm": 0.86328125,
      "learning_rate": 1.987721856569112e-05,
      "loss": 0.3505,
      "step": 1367
    },
    {
      "epoch": 0.10353341847594721,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9877032421884838e-05,
      "loss": 0.3248,
      "step": 1368
    },
    {
      "epoch": 0.10360910079939455,
      "grad_norm": 2.03125,
      "learning_rate": 1.987684613795591e-05,
      "loss": 0.5852,
      "step": 1369
    },
    {
      "epoch": 0.10368478312284188,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9876659713906964e-05,
      "loss": 0.2823,
      "step": 1370
    },
    {
      "epoch": 0.1037604654462892,
      "grad_norm": 0.859375,
      "learning_rate": 1.987647314974065e-05,
      "loss": 0.2943,
      "step": 1371
    },
    {
      "epoch": 0.10383614776973653,
      "grad_norm": 0.8828125,
      "learning_rate": 1.987628644545962e-05,
      "loss": 0.328,
      "step": 1372
    },
    {
      "epoch": 0.10391183009318386,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9876099601066517e-05,
      "loss": 0.3398,
      "step": 1373
    },
    {
      "epoch": 0.10398751241663119,
      "grad_norm": 0.85546875,
      "learning_rate": 1.987591261656399e-05,
      "loss": 0.2904,
      "step": 1374
    },
    {
      "epoch": 0.10406319474007852,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9875725491954698e-05,
      "loss": 0.3556,
      "step": 1375
    },
    {
      "epoch": 0.10413887706352586,
      "grad_norm": 1.0625,
      "learning_rate": 1.9875538227241295e-05,
      "loss": 0.4004,
      "step": 1376
    },
    {
      "epoch": 0.10421455938697317,
      "grad_norm": 0.95703125,
      "learning_rate": 1.9875350822426433e-05,
      "loss": 0.3746,
      "step": 1377
    },
    {
      "epoch": 0.1042902417104205,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9875163277512772e-05,
      "loss": 0.3478,
      "step": 1378
    },
    {
      "epoch": 0.10436592403386784,
      "grad_norm": 2.046875,
      "learning_rate": 1.9874975592502972e-05,
      "loss": 0.6326,
      "step": 1379
    },
    {
      "epoch": 0.10444160635731517,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9874787767399703e-05,
      "loss": 0.366,
      "step": 1380
    },
    {
      "epoch": 0.1045172886807625,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9874599802205617e-05,
      "loss": 0.3144,
      "step": 1381
    },
    {
      "epoch": 0.10459297100420983,
      "grad_norm": 0.92578125,
      "learning_rate": 1.987441169692339e-05,
      "loss": 0.3168,
      "step": 1382
    },
    {
      "epoch": 0.10466865332765717,
      "grad_norm": 0.9140625,
      "learning_rate": 1.987422345155569e-05,
      "loss": 0.3612,
      "step": 1383
    },
    {
      "epoch": 0.10474433565110448,
      "grad_norm": 3.859375,
      "learning_rate": 1.9874035066105184e-05,
      "loss": 0.602,
      "step": 1384
    },
    {
      "epoch": 0.10482001797455182,
      "grad_norm": 0.97265625,
      "learning_rate": 1.9873846540574545e-05,
      "loss": 0.3752,
      "step": 1385
    },
    {
      "epoch": 0.10489570029799915,
      "grad_norm": 0.84765625,
      "learning_rate": 1.987365787496645e-05,
      "loss": 0.2665,
      "step": 1386
    },
    {
      "epoch": 0.10497138262144648,
      "grad_norm": 1.015625,
      "learning_rate": 1.9873469069283576e-05,
      "loss": 0.4033,
      "step": 1387
    },
    {
      "epoch": 0.10504706494489381,
      "grad_norm": 0.890625,
      "learning_rate": 1.98732801235286e-05,
      "loss": 0.3595,
      "step": 1388
    },
    {
      "epoch": 0.10512274726834114,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9873091037704198e-05,
      "loss": 0.3922,
      "step": 1389
    },
    {
      "epoch": 0.10519842959178846,
      "grad_norm": 0.94140625,
      "learning_rate": 1.987290181181306e-05,
      "loss": 0.3881,
      "step": 1390
    },
    {
      "epoch": 0.1052741119152358,
      "grad_norm": 1.015625,
      "learning_rate": 1.9872712445857865e-05,
      "loss": 0.4403,
      "step": 1391
    },
    {
      "epoch": 0.10534979423868313,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9872522939841304e-05,
      "loss": 0.3736,
      "step": 1392
    },
    {
      "epoch": 0.10542547656213046,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9872333293766062e-05,
      "loss": 0.3414,
      "step": 1393
    },
    {
      "epoch": 0.10550115888557779,
      "grad_norm": 0.9140625,
      "learning_rate": 1.987214350763483e-05,
      "loss": 0.3647,
      "step": 1394
    },
    {
      "epoch": 0.10557684120902512,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9871953581450304e-05,
      "loss": 0.3009,
      "step": 1395
    },
    {
      "epoch": 0.10565252353247245,
      "grad_norm": 0.98046875,
      "learning_rate": 1.9871763515215173e-05,
      "loss": 0.3876,
      "step": 1396
    },
    {
      "epoch": 0.10572820585591977,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9871573308932138e-05,
      "loss": 0.3374,
      "step": 1397
    },
    {
      "epoch": 0.1058038881793671,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9871382962603896e-05,
      "loss": 0.311,
      "step": 1398
    },
    {
      "epoch": 0.10587957050281444,
      "grad_norm": 0.94921875,
      "learning_rate": 1.9871192476233143e-05,
      "loss": 0.3521,
      "step": 1399
    },
    {
      "epoch": 0.10595525282626177,
      "grad_norm": 0.9375,
      "learning_rate": 1.987100184982259e-05,
      "loss": 0.3667,
      "step": 1400
    },
    {
      "epoch": 0.1060309351497091,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9870811083374933e-05,
      "loss": 0.3336,
      "step": 1401
    },
    {
      "epoch": 0.10610661747315643,
      "grad_norm": 0.94921875,
      "learning_rate": 1.9870620176892884e-05,
      "loss": 0.4058,
      "step": 1402
    },
    {
      "epoch": 0.10618229979660375,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9870429130379145e-05,
      "loss": 0.3295,
      "step": 1403
    },
    {
      "epoch": 0.10625798212005108,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9870237943836436e-05,
      "loss": 0.3677,
      "step": 1404
    },
    {
      "epoch": 0.10633366444349841,
      "grad_norm": 0.90234375,
      "learning_rate": 1.987004661726746e-05,
      "loss": 0.3149,
      "step": 1405
    },
    {
      "epoch": 0.10640934676694574,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9869855150674938e-05,
      "loss": 0.3735,
      "step": 1406
    },
    {
      "epoch": 0.10648502909039308,
      "grad_norm": 0.93359375,
      "learning_rate": 1.986966354406158e-05,
      "loss": 0.3692,
      "step": 1407
    },
    {
      "epoch": 0.10656071141384041,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9869471797430112e-05,
      "loss": 0.3129,
      "step": 1408
    },
    {
      "epoch": 0.10663639373728774,
      "grad_norm": 0.9375,
      "learning_rate": 1.9869279910783244e-05,
      "loss": 0.3995,
      "step": 1409
    },
    {
      "epoch": 0.10671207606073506,
      "grad_norm": 0.890625,
      "learning_rate": 1.986908788412371e-05,
      "loss": 0.356,
      "step": 1410
    },
    {
      "epoch": 0.10678775838418239,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9868895717454225e-05,
      "loss": 0.3734,
      "step": 1411
    },
    {
      "epoch": 0.10686344070762972,
      "grad_norm": 0.9375,
      "learning_rate": 1.986870341077752e-05,
      "loss": 0.3857,
      "step": 1412
    },
    {
      "epoch": 0.10693912303107705,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9868510964096325e-05,
      "loss": 0.3592,
      "step": 1413
    },
    {
      "epoch": 0.10701480535452439,
      "grad_norm": 0.953125,
      "learning_rate": 1.986831837741336e-05,
      "loss": 0.388,
      "step": 1414
    },
    {
      "epoch": 0.10709048767797172,
      "grad_norm": 0.89453125,
      "learning_rate": 1.986812565073137e-05,
      "loss": 0.3413,
      "step": 1415
    },
    {
      "epoch": 0.10716617000141905,
      "grad_norm": 0.85546875,
      "learning_rate": 1.986793278405308e-05,
      "loss": 0.3112,
      "step": 1416
    },
    {
      "epoch": 0.10724185232486637,
      "grad_norm": 1.015625,
      "learning_rate": 1.986773977738123e-05,
      "loss": 0.4044,
      "step": 1417
    },
    {
      "epoch": 0.1073175346483137,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9867546630718563e-05,
      "loss": 0.3486,
      "step": 1418
    },
    {
      "epoch": 0.10739321697176103,
      "grad_norm": 0.859375,
      "learning_rate": 1.9867353344067806e-05,
      "loss": 0.3373,
      "step": 1419
    },
    {
      "epoch": 0.10746889929520836,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9867159917431712e-05,
      "loss": 0.3105,
      "step": 1420
    },
    {
      "epoch": 0.1075445816186557,
      "grad_norm": 0.97265625,
      "learning_rate": 1.9866966350813025e-05,
      "loss": 0.3509,
      "step": 1421
    },
    {
      "epoch": 0.10762026394210303,
      "grad_norm": 0.890625,
      "learning_rate": 1.9866772644214484e-05,
      "loss": 0.3267,
      "step": 1422
    },
    {
      "epoch": 0.10769594626555035,
      "grad_norm": 0.9296875,
      "learning_rate": 1.986657879763884e-05,
      "loss": 0.358,
      "step": 1423
    },
    {
      "epoch": 0.10777162858899768,
      "grad_norm": 0.828125,
      "learning_rate": 1.986638481108885e-05,
      "loss": 0.3048,
      "step": 1424
    },
    {
      "epoch": 0.10784731091244501,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9866190684567256e-05,
      "loss": 0.347,
      "step": 1425
    },
    {
      "epoch": 0.10792299323589234,
      "grad_norm": 0.953125,
      "learning_rate": 1.9865996418076813e-05,
      "loss": 0.3552,
      "step": 1426
    },
    {
      "epoch": 0.10799867555933967,
      "grad_norm": 0.921875,
      "learning_rate": 1.9865802011620287e-05,
      "loss": 0.3802,
      "step": 1427
    },
    {
      "epoch": 0.108074357882787,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9865607465200426e-05,
      "loss": 0.3392,
      "step": 1428
    },
    {
      "epoch": 0.10815004020623434,
      "grad_norm": 0.875,
      "learning_rate": 1.9865412778819992e-05,
      "loss": 0.3084,
      "step": 1429
    },
    {
      "epoch": 0.10822572252968166,
      "grad_norm": 0.89453125,
      "learning_rate": 1.986521795248175e-05,
      "loss": 0.3494,
      "step": 1430
    },
    {
      "epoch": 0.10830140485312899,
      "grad_norm": 0.88671875,
      "learning_rate": 1.986502298618846e-05,
      "loss": 0.3739,
      "step": 1431
    },
    {
      "epoch": 0.10837708717657632,
      "grad_norm": 0.8671875,
      "learning_rate": 1.986482787994289e-05,
      "loss": 0.3436,
      "step": 1432
    },
    {
      "epoch": 0.10845276950002365,
      "grad_norm": 0.9921875,
      "learning_rate": 1.9864632633747812e-05,
      "loss": 0.4024,
      "step": 1433
    },
    {
      "epoch": 0.10852845182347098,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9864437247605986e-05,
      "loss": 0.3578,
      "step": 1434
    },
    {
      "epoch": 0.10860413414691832,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9864241721520195e-05,
      "loss": 0.3619,
      "step": 1435
    },
    {
      "epoch": 0.10867981647036563,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9864046055493205e-05,
      "loss": 0.3668,
      "step": 1436
    },
    {
      "epoch": 0.10875549879381297,
      "grad_norm": 0.890625,
      "learning_rate": 1.9863850249527793e-05,
      "loss": 0.3245,
      "step": 1437
    },
    {
      "epoch": 0.1088311811172603,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9863654303626744e-05,
      "loss": 0.302,
      "step": 1438
    },
    {
      "epoch": 0.10890686344070763,
      "grad_norm": 0.99609375,
      "learning_rate": 1.9863458217792827e-05,
      "loss": 0.3959,
      "step": 1439
    },
    {
      "epoch": 0.10898254576415496,
      "grad_norm": 0.859375,
      "learning_rate": 1.986326199202883e-05,
      "loss": 0.2925,
      "step": 1440
    },
    {
      "epoch": 0.1090582280876023,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9863065626337534e-05,
      "loss": 0.3375,
      "step": 1441
    },
    {
      "epoch": 0.10913391041104963,
      "grad_norm": 0.859375,
      "learning_rate": 1.986286912072173e-05,
      "loss": 0.3391,
      "step": 1442
    },
    {
      "epoch": 0.10920959273449694,
      "grad_norm": 0.796875,
      "learning_rate": 1.9862672475184198e-05,
      "loss": 0.2817,
      "step": 1443
    },
    {
      "epoch": 0.10928527505794428,
      "grad_norm": 0.96875,
      "learning_rate": 1.9862475689727734e-05,
      "loss": 0.3436,
      "step": 1444
    },
    {
      "epoch": 0.10936095738139161,
      "grad_norm": 0.8359375,
      "learning_rate": 1.986227876435513e-05,
      "loss": 0.3291,
      "step": 1445
    },
    {
      "epoch": 0.10943663970483894,
      "grad_norm": 0.875,
      "learning_rate": 1.9862081699069175e-05,
      "loss": 0.3227,
      "step": 1446
    },
    {
      "epoch": 0.10951232202828627,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9861884493872667e-05,
      "loss": 0.3482,
      "step": 1447
    },
    {
      "epoch": 0.1095880043517336,
      "grad_norm": 0.98046875,
      "learning_rate": 1.9861687148768406e-05,
      "loss": 0.3299,
      "step": 1448
    },
    {
      "epoch": 0.10966368667518092,
      "grad_norm": 0.9609375,
      "learning_rate": 1.9861489663759184e-05,
      "loss": 0.3637,
      "step": 1449
    },
    {
      "epoch": 0.10973936899862825,
      "grad_norm": 0.890625,
      "learning_rate": 1.9861292038847818e-05,
      "loss": 0.353,
      "step": 1450
    },
    {
      "epoch": 0.10981505132207559,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9861094274037095e-05,
      "loss": 0.3278,
      "step": 1451
    },
    {
      "epoch": 0.10989073364552292,
      "grad_norm": 0.89453125,
      "learning_rate": 1.986089636932983e-05,
      "loss": 0.3232,
      "step": 1452
    },
    {
      "epoch": 0.10996641596897025,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9860698324728826e-05,
      "loss": 0.3348,
      "step": 1453
    },
    {
      "epoch": 0.11004209829241758,
      "grad_norm": 0.9453125,
      "learning_rate": 1.9860500140236896e-05,
      "loss": 0.3276,
      "step": 1454
    },
    {
      "epoch": 0.11011778061586491,
      "grad_norm": 0.91796875,
      "learning_rate": 1.986030181585685e-05,
      "loss": 0.3324,
      "step": 1455
    },
    {
      "epoch": 0.11019346293931223,
      "grad_norm": 2.34375,
      "learning_rate": 1.98601033515915e-05,
      "loss": 0.4928,
      "step": 1456
    },
    {
      "epoch": 0.11026914526275956,
      "grad_norm": 2.0,
      "learning_rate": 1.9859904747443665e-05,
      "loss": 0.5517,
      "step": 1457
    },
    {
      "epoch": 0.1103448275862069,
      "grad_norm": 0.94921875,
      "learning_rate": 1.985970600341616e-05,
      "loss": 0.3612,
      "step": 1458
    },
    {
      "epoch": 0.11042050990965423,
      "grad_norm": 1.046875,
      "learning_rate": 1.9859507119511806e-05,
      "loss": 0.3622,
      "step": 1459
    },
    {
      "epoch": 0.11049619223310156,
      "grad_norm": 0.859375,
      "learning_rate": 1.985930809573343e-05,
      "loss": 0.3018,
      "step": 1460
    },
    {
      "epoch": 0.11057187455654889,
      "grad_norm": 0.92578125,
      "learning_rate": 1.985910893208384e-05,
      "loss": 0.3643,
      "step": 1461
    },
    {
      "epoch": 0.11064755687999622,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9858909628565876e-05,
      "loss": 0.3263,
      "step": 1462
    },
    {
      "epoch": 0.11072323920344354,
      "grad_norm": 0.83203125,
      "learning_rate": 1.985871018518236e-05,
      "loss": 0.2866,
      "step": 1463
    },
    {
      "epoch": 0.11079892152689087,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9858510601936124e-05,
      "loss": 0.3302,
      "step": 1464
    },
    {
      "epoch": 0.1108746038503382,
      "grad_norm": 0.96875,
      "learning_rate": 1.9858310878829995e-05,
      "loss": 0.2757,
      "step": 1465
    },
    {
      "epoch": 0.11095028617378554,
      "grad_norm": 0.859375,
      "learning_rate": 1.9858111015866806e-05,
      "loss": 0.3267,
      "step": 1466
    },
    {
      "epoch": 0.11102596849723287,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9857911013049398e-05,
      "loss": 0.27,
      "step": 1467
    },
    {
      "epoch": 0.1111016508206802,
      "grad_norm": 0.859375,
      "learning_rate": 1.9857710870380606e-05,
      "loss": 0.3408,
      "step": 1468
    },
    {
      "epoch": 0.11117733314412752,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9857510587863266e-05,
      "loss": 0.368,
      "step": 1469
    },
    {
      "epoch": 0.11125301546757485,
      "grad_norm": 0.84375,
      "learning_rate": 1.9857310165500227e-05,
      "loss": 0.3107,
      "step": 1470
    },
    {
      "epoch": 0.11132869779102218,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9857109603294326e-05,
      "loss": 0.3482,
      "step": 1471
    },
    {
      "epoch": 0.11140438011446951,
      "grad_norm": 0.76171875,
      "learning_rate": 1.9856908901248407e-05,
      "loss": 0.2425,
      "step": 1472
    },
    {
      "epoch": 0.11148006243791685,
      "grad_norm": 0.9296875,
      "learning_rate": 1.985670805936532e-05,
      "loss": 0.3987,
      "step": 1473
    },
    {
      "epoch": 0.11155574476136418,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9856507077647918e-05,
      "loss": 0.3245,
      "step": 1474
    },
    {
      "epoch": 0.11163142708481151,
      "grad_norm": 0.9453125,
      "learning_rate": 1.985630595609905e-05,
      "loss": 0.3204,
      "step": 1475
    },
    {
      "epoch": 0.11170710940825883,
      "grad_norm": 0.859375,
      "learning_rate": 1.9856104694721562e-05,
      "loss": 0.3536,
      "step": 1476
    },
    {
      "epoch": 0.11178279173170616,
      "grad_norm": 4.15625,
      "learning_rate": 1.9855903293518318e-05,
      "loss": 0.5511,
      "step": 1477
    },
    {
      "epoch": 0.11185847405515349,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9855701752492174e-05,
      "loss": 0.3087,
      "step": 1478
    },
    {
      "epoch": 0.11193415637860082,
      "grad_norm": 1.0,
      "learning_rate": 1.9855500071645987e-05,
      "loss": 0.3836,
      "step": 1479
    },
    {
      "epoch": 0.11200983870204816,
      "grad_norm": 0.921875,
      "learning_rate": 1.985529825098262e-05,
      "loss": 0.3461,
      "step": 1480
    },
    {
      "epoch": 0.11208552102549549,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9855096290504933e-05,
      "loss": 0.3436,
      "step": 1481
    },
    {
      "epoch": 0.1121612033489428,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9854894190215796e-05,
      "loss": 0.3965,
      "step": 1482
    },
    {
      "epoch": 0.11223688567239014,
      "grad_norm": 0.85546875,
      "learning_rate": 1.985469195011807e-05,
      "loss": 0.3402,
      "step": 1483
    },
    {
      "epoch": 0.11231256799583747,
      "grad_norm": 0.8984375,
      "learning_rate": 1.985448957021463e-05,
      "loss": 0.3621,
      "step": 1484
    },
    {
      "epoch": 0.1123882503192848,
      "grad_norm": 4.75,
      "learning_rate": 1.985428705050834e-05,
      "loss": 0.5376,
      "step": 1485
    },
    {
      "epoch": 0.11246393264273213,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9854084391002084e-05,
      "loss": 0.3574,
      "step": 1486
    },
    {
      "epoch": 0.11253961496617947,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9853881591698726e-05,
      "loss": 0.3643,
      "step": 1487
    },
    {
      "epoch": 0.1126152972896268,
      "grad_norm": 0.921875,
      "learning_rate": 1.9853678652601148e-05,
      "loss": 0.3427,
      "step": 1488
    },
    {
      "epoch": 0.11269097961307412,
      "grad_norm": 0.8984375,
      "learning_rate": 1.985347557371223e-05,
      "loss": 0.3273,
      "step": 1489
    },
    {
      "epoch": 0.11276666193652145,
      "grad_norm": 0.9375,
      "learning_rate": 1.9853272355034854e-05,
      "loss": 0.3498,
      "step": 1490
    },
    {
      "epoch": 0.11284234425996878,
      "grad_norm": 0.87890625,
      "learning_rate": 1.98530689965719e-05,
      "loss": 0.362,
      "step": 1491
    },
    {
      "epoch": 0.11291802658341611,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9852865498326253e-05,
      "loss": 0.3051,
      "step": 1492
    },
    {
      "epoch": 0.11299370890686344,
      "grad_norm": 0.8828125,
      "learning_rate": 1.98526618603008e-05,
      "loss": 0.3488,
      "step": 1493
    },
    {
      "epoch": 0.11306939123031078,
      "grad_norm": 0.90625,
      "learning_rate": 1.9852458082498433e-05,
      "loss": 0.3685,
      "step": 1494
    },
    {
      "epoch": 0.1131450735537581,
      "grad_norm": 0.87109375,
      "learning_rate": 1.985225416492204e-05,
      "loss": 0.3257,
      "step": 1495
    },
    {
      "epoch": 0.11322075587720543,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9852050107574508e-05,
      "loss": 0.3101,
      "step": 1496
    },
    {
      "epoch": 0.11329643820065276,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9851845910458746e-05,
      "loss": 0.3293,
      "step": 1497
    },
    {
      "epoch": 0.11337212052410009,
      "grad_norm": 0.87890625,
      "learning_rate": 1.985164157357764e-05,
      "loss": 0.3627,
      "step": 1498
    },
    {
      "epoch": 0.11344780284754742,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9851437096934094e-05,
      "loss": 0.3515,
      "step": 1499
    },
    {
      "epoch": 0.11352348517099475,
      "grad_norm": 0.9609375,
      "learning_rate": 1.9851232480531008e-05,
      "loss": 0.3587,
      "step": 1500
    },
    {
      "epoch": 0.11359916749444209,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9851027724371285e-05,
      "loss": 0.3453,
      "step": 1501
    },
    {
      "epoch": 0.1136748498178894,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9850822828457826e-05,
      "loss": 0.3111,
      "step": 1502
    },
    {
      "epoch": 0.11375053214133674,
      "grad_norm": 0.953125,
      "learning_rate": 1.985061779279354e-05,
      "loss": 0.3055,
      "step": 1503
    },
    {
      "epoch": 0.11382621446478407,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9850412617381338e-05,
      "loss": 0.2994,
      "step": 1504
    },
    {
      "epoch": 0.1139018967882314,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9850207302224128e-05,
      "loss": 0.3196,
      "step": 1505
    },
    {
      "epoch": 0.11397757911167873,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9850001847324825e-05,
      "loss": 0.3329,
      "step": 1506
    },
    {
      "epoch": 0.11405326143512606,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9849796252686344e-05,
      "loss": 0.357,
      "step": 1507
    },
    {
      "epoch": 0.1141289437585734,
      "grad_norm": 0.9375,
      "learning_rate": 1.98495905183116e-05,
      "loss": 0.3836,
      "step": 1508
    },
    {
      "epoch": 0.11420462608202071,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9849384644203508e-05,
      "loss": 0.3042,
      "step": 1509
    },
    {
      "epoch": 0.11428030840546805,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9849178630364995e-05,
      "loss": 0.3442,
      "step": 1510
    },
    {
      "epoch": 0.11435599072891538,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9848972476798982e-05,
      "loss": 0.3534,
      "step": 1511
    },
    {
      "epoch": 0.11443167305236271,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9848766183508393e-05,
      "loss": 0.3663,
      "step": 1512
    },
    {
      "epoch": 0.11450735537581004,
      "grad_norm": 0.953125,
      "learning_rate": 1.9848559750496155e-05,
      "loss": 0.3408,
      "step": 1513
    },
    {
      "epoch": 0.11458303769925737,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9848353177765195e-05,
      "loss": 0.3982,
      "step": 1514
    },
    {
      "epoch": 0.11465872002270469,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9848146465318444e-05,
      "loss": 0.3445,
      "step": 1515
    },
    {
      "epoch": 0.11473440234615202,
      "grad_norm": 1.0390625,
      "learning_rate": 1.9847939613158836e-05,
      "loss": 0.3814,
      "step": 1516
    },
    {
      "epoch": 0.11481008466959935,
      "grad_norm": 0.94140625,
      "learning_rate": 1.984773262128931e-05,
      "loss": 0.3937,
      "step": 1517
    },
    {
      "epoch": 0.11488576699304669,
      "grad_norm": 0.95703125,
      "learning_rate": 1.984752548971279e-05,
      "loss": 0.363,
      "step": 1518
    },
    {
      "epoch": 0.11496144931649402,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9847318218432224e-05,
      "loss": 0.3344,
      "step": 1519
    },
    {
      "epoch": 0.11503713163994135,
      "grad_norm": 0.953125,
      "learning_rate": 1.9847110807450555e-05,
      "loss": 0.3626,
      "step": 1520
    },
    {
      "epoch": 0.11511281396338868,
      "grad_norm": 0.9375,
      "learning_rate": 1.984690325677072e-05,
      "loss": 0.3889,
      "step": 1521
    },
    {
      "epoch": 0.115188496286836,
      "grad_norm": 0.90625,
      "learning_rate": 1.984669556639566e-05,
      "loss": 0.3635,
      "step": 1522
    },
    {
      "epoch": 0.11526417861028333,
      "grad_norm": 0.96875,
      "learning_rate": 1.9846487736328324e-05,
      "loss": 0.3718,
      "step": 1523
    },
    {
      "epoch": 0.11533986093373066,
      "grad_norm": 0.921875,
      "learning_rate": 1.984627976657167e-05,
      "loss": 0.3556,
      "step": 1524
    },
    {
      "epoch": 0.115415543257178,
      "grad_norm": 0.9375,
      "learning_rate": 1.9846071657128635e-05,
      "loss": 0.3681,
      "step": 1525
    },
    {
      "epoch": 0.11549122558062533,
      "grad_norm": 0.8359375,
      "learning_rate": 1.984586340800218e-05,
      "loss": 0.3138,
      "step": 1526
    },
    {
      "epoch": 0.11556690790407266,
      "grad_norm": 0.95703125,
      "learning_rate": 1.9845655019195256e-05,
      "loss": 0.3893,
      "step": 1527
    },
    {
      "epoch": 0.11564259022751998,
      "grad_norm": 0.875,
      "learning_rate": 1.984544649071082e-05,
      "loss": 0.3246,
      "step": 1528
    },
    {
      "epoch": 0.11571827255096731,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9845237822551827e-05,
      "loss": 0.3781,
      "step": 1529
    },
    {
      "epoch": 0.11579395487441464,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9845029014721243e-05,
      "loss": 0.3288,
      "step": 1530
    },
    {
      "epoch": 0.11586963719786197,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9844820067222027e-05,
      "loss": 0.3261,
      "step": 1531
    },
    {
      "epoch": 0.1159453195213093,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9844610980057148e-05,
      "loss": 0.3284,
      "step": 1532
    },
    {
      "epoch": 0.11602100184475664,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9844401753229566e-05,
      "loss": 0.2881,
      "step": 1533
    },
    {
      "epoch": 0.11609668416820397,
      "grad_norm": 0.94921875,
      "learning_rate": 1.984419238674225e-05,
      "loss": 0.353,
      "step": 1534
    },
    {
      "epoch": 0.11617236649165129,
      "grad_norm": 0.97265625,
      "learning_rate": 1.9843982880598172e-05,
      "loss": 0.3588,
      "step": 1535
    },
    {
      "epoch": 0.11624804881509862,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9843773234800302e-05,
      "loss": 0.3184,
      "step": 1536
    },
    {
      "epoch": 0.11632373113854595,
      "grad_norm": 0.90625,
      "learning_rate": 1.984356344935162e-05,
      "loss": 0.3446,
      "step": 1537
    },
    {
      "epoch": 0.11639941346199328,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9843353524255097e-05,
      "loss": 0.372,
      "step": 1538
    },
    {
      "epoch": 0.11647509578544062,
      "grad_norm": 0.90625,
      "learning_rate": 1.984314345951371e-05,
      "loss": 0.3539,
      "step": 1539
    },
    {
      "epoch": 0.11655077810888795,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9842933255130447e-05,
      "loss": 0.4105,
      "step": 1540
    },
    {
      "epoch": 0.11662646043233527,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9842722911108282e-05,
      "loss": 0.3933,
      "step": 1541
    },
    {
      "epoch": 0.1167021427557826,
      "grad_norm": 0.83984375,
      "learning_rate": 1.98425124274502e-05,
      "loss": 0.347,
      "step": 1542
    },
    {
      "epoch": 0.11677782507922993,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9842301804159193e-05,
      "loss": 0.3461,
      "step": 1543
    },
    {
      "epoch": 0.11685350740267726,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9842091041238243e-05,
      "loss": 0.3468,
      "step": 1544
    },
    {
      "epoch": 0.1169291897261246,
      "grad_norm": 0.890625,
      "learning_rate": 1.984188013869034e-05,
      "loss": 0.3323,
      "step": 1545
    },
    {
      "epoch": 0.11700487204957193,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9841669096518478e-05,
      "loss": 0.3144,
      "step": 1546
    },
    {
      "epoch": 0.11708055437301926,
      "grad_norm": 0.94921875,
      "learning_rate": 1.9841457914725654e-05,
      "loss": 0.4085,
      "step": 1547
    },
    {
      "epoch": 0.11715623669646658,
      "grad_norm": 0.90234375,
      "learning_rate": 1.984124659331486e-05,
      "loss": 0.3584,
      "step": 1548
    },
    {
      "epoch": 0.11723191901991391,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9841035132289095e-05,
      "loss": 0.3212,
      "step": 1549
    },
    {
      "epoch": 0.11730760134336124,
      "grad_norm": 0.96484375,
      "learning_rate": 1.9840823531651357e-05,
      "loss": 0.3118,
      "step": 1550
    },
    {
      "epoch": 0.11738328366680857,
      "grad_norm": 0.9609375,
      "learning_rate": 1.984061179140465e-05,
      "loss": 0.3627,
      "step": 1551
    },
    {
      "epoch": 0.1174589659902559,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9840399911551977e-05,
      "loss": 0.3512,
      "step": 1552
    },
    {
      "epoch": 0.11753464831370324,
      "grad_norm": 0.9375,
      "learning_rate": 1.9840187892096348e-05,
      "loss": 0.3664,
      "step": 1553
    },
    {
      "epoch": 0.11761033063715057,
      "grad_norm": 0.90625,
      "learning_rate": 1.9839975733040766e-05,
      "loss": 0.3595,
      "step": 1554
    },
    {
      "epoch": 0.11768601296059789,
      "grad_norm": 1.015625,
      "learning_rate": 1.983976343438824e-05,
      "loss": 0.3219,
      "step": 1555
    },
    {
      "epoch": 0.11776169528404522,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9839550996141787e-05,
      "loss": 0.2983,
      "step": 1556
    },
    {
      "epoch": 0.11783737760749255,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9839338418304418e-05,
      "loss": 0.3325,
      "step": 1557
    },
    {
      "epoch": 0.11791305993093988,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9839125700879146e-05,
      "loss": 0.2813,
      "step": 1558
    },
    {
      "epoch": 0.11798874225438721,
      "grad_norm": 0.890625,
      "learning_rate": 1.9838912843868995e-05,
      "loss": 0.3223,
      "step": 1559
    },
    {
      "epoch": 0.11806442457783455,
      "grad_norm": 0.95703125,
      "learning_rate": 1.9838699847276975e-05,
      "loss": 0.3164,
      "step": 1560
    },
    {
      "epoch": 0.11814010690128186,
      "grad_norm": 0.87109375,
      "learning_rate": 1.983848671110612e-05,
      "loss": 0.3374,
      "step": 1561
    },
    {
      "epoch": 0.1182157892247292,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9838273435359447e-05,
      "loss": 0.3454,
      "step": 1562
    },
    {
      "epoch": 0.11829147154817653,
      "grad_norm": 0.90625,
      "learning_rate": 1.983806002003998e-05,
      "loss": 0.3665,
      "step": 1563
    },
    {
      "epoch": 0.11836715387162386,
      "grad_norm": 0.84765625,
      "learning_rate": 1.983784646515075e-05,
      "loss": 0.3143,
      "step": 1564
    },
    {
      "epoch": 0.11844283619507119,
      "grad_norm": 1.078125,
      "learning_rate": 1.9837632770694783e-05,
      "loss": 0.3065,
      "step": 1565
    },
    {
      "epoch": 0.11851851851851852,
      "grad_norm": 0.99609375,
      "learning_rate": 1.9837418936675115e-05,
      "loss": 0.3636,
      "step": 1566
    },
    {
      "epoch": 0.11859420084196585,
      "grad_norm": 0.8515625,
      "learning_rate": 1.983720496309478e-05,
      "loss": 0.3259,
      "step": 1567
    },
    {
      "epoch": 0.11866988316541317,
      "grad_norm": 0.92578125,
      "learning_rate": 1.983699084995681e-05,
      "loss": 0.3387,
      "step": 1568
    },
    {
      "epoch": 0.1187455654888605,
      "grad_norm": 0.94921875,
      "learning_rate": 1.983677659726424e-05,
      "loss": 0.3499,
      "step": 1569
    },
    {
      "epoch": 0.11882124781230784,
      "grad_norm": 1.0,
      "learning_rate": 1.983656220502012e-05,
      "loss": 0.3879,
      "step": 1570
    },
    {
      "epoch": 0.11889693013575517,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9836347673227482e-05,
      "loss": 0.3252,
      "step": 1571
    },
    {
      "epoch": 0.1189726124592025,
      "grad_norm": 0.859375,
      "learning_rate": 1.983613300188937e-05,
      "loss": 0.3092,
      "step": 1572
    },
    {
      "epoch": 0.11904829478264983,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9835918191008833e-05,
      "loss": 0.3586,
      "step": 1573
    },
    {
      "epoch": 0.11912397710609715,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9835703240588923e-05,
      "loss": 0.2958,
      "step": 1574
    },
    {
      "epoch": 0.11919965942954448,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9835488150632677e-05,
      "loss": 0.3593,
      "step": 1575
    },
    {
      "epoch": 0.11927534175299181,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9835272921143156e-05,
      "loss": 0.3682,
      "step": 1576
    },
    {
      "epoch": 0.11935102407643915,
      "grad_norm": 0.921875,
      "learning_rate": 1.9835057552123412e-05,
      "loss": 0.3767,
      "step": 1577
    },
    {
      "epoch": 0.11942670639988648,
      "grad_norm": 0.890625,
      "learning_rate": 1.9834842043576498e-05,
      "loss": 0.3574,
      "step": 1578
    },
    {
      "epoch": 0.11950238872333381,
      "grad_norm": 0.84375,
      "learning_rate": 1.9834626395505473e-05,
      "loss": 0.3243,
      "step": 1579
    },
    {
      "epoch": 0.11957807104678114,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9834410607913393e-05,
      "loss": 0.3649,
      "step": 1580
    },
    {
      "epoch": 0.11965375337022846,
      "grad_norm": 0.90625,
      "learning_rate": 1.9834194680803325e-05,
      "loss": 0.3955,
      "step": 1581
    },
    {
      "epoch": 0.11972943569367579,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9833978614178328e-05,
      "loss": 0.3566,
      "step": 1582
    },
    {
      "epoch": 0.11980511801712312,
      "grad_norm": 0.875,
      "learning_rate": 1.983376240804147e-05,
      "loss": 0.369,
      "step": 1583
    },
    {
      "epoch": 0.11988080034057046,
      "grad_norm": 0.9453125,
      "learning_rate": 1.9833546062395816e-05,
      "loss": 0.3886,
      "step": 1584
    },
    {
      "epoch": 0.11995648266401779,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9833329577244437e-05,
      "loss": 0.3211,
      "step": 1585
    },
    {
      "epoch": 0.12003216498746512,
      "grad_norm": 1.0546875,
      "learning_rate": 1.98331129525904e-05,
      "loss": 0.3525,
      "step": 1586
    },
    {
      "epoch": 0.12010784731091244,
      "grad_norm": 0.828125,
      "learning_rate": 1.9832896188436784e-05,
      "loss": 0.3155,
      "step": 1587
    },
    {
      "epoch": 0.12018352963435977,
      "grad_norm": 0.96484375,
      "learning_rate": 1.9832679284786663e-05,
      "loss": 0.3352,
      "step": 1588
    },
    {
      "epoch": 0.1202592119578071,
      "grad_norm": 0.97265625,
      "learning_rate": 1.983246224164311e-05,
      "loss": 0.4124,
      "step": 1589
    },
    {
      "epoch": 0.12033489428125443,
      "grad_norm": 0.85546875,
      "learning_rate": 1.983224505900921e-05,
      "loss": 0.3016,
      "step": 1590
    },
    {
      "epoch": 0.12041057660470177,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9832027736888042e-05,
      "loss": 0.3418,
      "step": 1591
    },
    {
      "epoch": 0.1204862589281491,
      "grad_norm": 0.90625,
      "learning_rate": 1.9831810275282685e-05,
      "loss": 0.375,
      "step": 1592
    },
    {
      "epoch": 0.12056194125159643,
      "grad_norm": 2.28125,
      "learning_rate": 1.983159267419623e-05,
      "loss": 0.5103,
      "step": 1593
    },
    {
      "epoch": 0.12063762357504375,
      "grad_norm": 0.8984375,
      "learning_rate": 1.983137493363176e-05,
      "loss": 0.3601,
      "step": 1594
    },
    {
      "epoch": 0.12071330589849108,
      "grad_norm": 2.140625,
      "learning_rate": 1.9831157053592364e-05,
      "loss": 0.4491,
      "step": 1595
    },
    {
      "epoch": 0.12078898822193841,
      "grad_norm": 1.0859375,
      "learning_rate": 1.9830939034081137e-05,
      "loss": 0.4145,
      "step": 1596
    },
    {
      "epoch": 0.12086467054538574,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9830720875101167e-05,
      "loss": 0.336,
      "step": 1597
    },
    {
      "epoch": 0.12094035286883308,
      "grad_norm": 0.98046875,
      "learning_rate": 1.9830502576655553e-05,
      "loss": 0.4333,
      "step": 1598
    },
    {
      "epoch": 0.12101603519228041,
      "grad_norm": 1.984375,
      "learning_rate": 1.9830284138747387e-05,
      "loss": 0.5905,
      "step": 1599
    },
    {
      "epoch": 0.12109171751572773,
      "grad_norm": 0.98828125,
      "learning_rate": 1.9830065561379776e-05,
      "loss": 0.3769,
      "step": 1600
    },
    {
      "epoch": 0.12116739983917506,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9829846844555813e-05,
      "loss": 0.354,
      "step": 1601
    },
    {
      "epoch": 0.12124308216262239,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9829627988278603e-05,
      "loss": 0.2756,
      "step": 1602
    },
    {
      "epoch": 0.12131876448606972,
      "grad_norm": 0.9375,
      "learning_rate": 1.9829408992551255e-05,
      "loss": 0.3457,
      "step": 1603
    },
    {
      "epoch": 0.12139444680951705,
      "grad_norm": 0.8046875,
      "learning_rate": 1.982918985737687e-05,
      "loss": 0.3041,
      "step": 1604
    },
    {
      "epoch": 0.12147012913296439,
      "grad_norm": 0.984375,
      "learning_rate": 1.982897058275856e-05,
      "loss": 0.3712,
      "step": 1605
    },
    {
      "epoch": 0.12154581145641172,
      "grad_norm": 0.859375,
      "learning_rate": 1.9828751168699437e-05,
      "loss": 0.3419,
      "step": 1606
    },
    {
      "epoch": 0.12162149377985904,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9828531615202608e-05,
      "loss": 0.3345,
      "step": 1607
    },
    {
      "epoch": 0.12169717610330637,
      "grad_norm": 0.875,
      "learning_rate": 1.9828311922271192e-05,
      "loss": 0.3385,
      "step": 1608
    },
    {
      "epoch": 0.1217728584267537,
      "grad_norm": 0.890625,
      "learning_rate": 1.982809208990831e-05,
      "loss": 0.3556,
      "step": 1609
    },
    {
      "epoch": 0.12184854075020103,
      "grad_norm": 0.97265625,
      "learning_rate": 1.9827872118117075e-05,
      "loss": 0.3789,
      "step": 1610
    },
    {
      "epoch": 0.12192422307364836,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9827652006900607e-05,
      "loss": 0.3267,
      "step": 1611
    },
    {
      "epoch": 0.1219999053970957,
      "grad_norm": 0.91015625,
      "learning_rate": 1.982743175626203e-05,
      "loss": 0.3507,
      "step": 1612
    },
    {
      "epoch": 0.12207558772054303,
      "grad_norm": 1.03125,
      "learning_rate": 1.9827211366204468e-05,
      "loss": 0.3723,
      "step": 1613
    },
    {
      "epoch": 0.12215127004399035,
      "grad_norm": 0.96875,
      "learning_rate": 1.9826990836731052e-05,
      "loss": 0.3642,
      "step": 1614
    },
    {
      "epoch": 0.12222695236743768,
      "grad_norm": 0.76953125,
      "learning_rate": 1.9826770167844907e-05,
      "loss": 0.2751,
      "step": 1615
    },
    {
      "epoch": 0.12230263469088501,
      "grad_norm": 0.953125,
      "learning_rate": 1.982654935954916e-05,
      "loss": 0.3746,
      "step": 1616
    },
    {
      "epoch": 0.12237831701433234,
      "grad_norm": 0.86328125,
      "learning_rate": 1.982632841184695e-05,
      "loss": 0.3019,
      "step": 1617
    },
    {
      "epoch": 0.12245399933777967,
      "grad_norm": 1.0234375,
      "learning_rate": 1.9826107324741413e-05,
      "loss": 0.4147,
      "step": 1618
    },
    {
      "epoch": 0.122529681661227,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9825886098235677e-05,
      "loss": 0.3196,
      "step": 1619
    },
    {
      "epoch": 0.12260536398467432,
      "grad_norm": 0.921875,
      "learning_rate": 1.9825664732332886e-05,
      "loss": 0.3676,
      "step": 1620
    },
    {
      "epoch": 0.12268104630812166,
      "grad_norm": 0.859375,
      "learning_rate": 1.9825443227036176e-05,
      "loss": 0.304,
      "step": 1621
    },
    {
      "epoch": 0.12275672863156899,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9825221582348696e-05,
      "loss": 0.2833,
      "step": 1622
    },
    {
      "epoch": 0.12283241095501632,
      "grad_norm": 9.4375,
      "learning_rate": 1.982499979827359e-05,
      "loss": 0.5919,
      "step": 1623
    },
    {
      "epoch": 0.12290809327846365,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9824777874813996e-05,
      "loss": 0.3113,
      "step": 1624
    },
    {
      "epoch": 0.12298377560191098,
      "grad_norm": 0.95703125,
      "learning_rate": 1.982455581197307e-05,
      "loss": 0.3498,
      "step": 1625
    },
    {
      "epoch": 0.12305945792535831,
      "grad_norm": 0.98046875,
      "learning_rate": 1.982433360975396e-05,
      "loss": 0.362,
      "step": 1626
    },
    {
      "epoch": 0.12313514024880563,
      "grad_norm": 0.890625,
      "learning_rate": 1.982411126815982e-05,
      "loss": 0.3568,
      "step": 1627
    },
    {
      "epoch": 0.12321082257225296,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9823888787193802e-05,
      "loss": 0.3472,
      "step": 1628
    },
    {
      "epoch": 0.1232865048957003,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9823666166859065e-05,
      "loss": 0.3349,
      "step": 1629
    },
    {
      "epoch": 0.12336218721914763,
      "grad_norm": 0.90234375,
      "learning_rate": 1.982344340715876e-05,
      "loss": 0.3174,
      "step": 1630
    },
    {
      "epoch": 0.12343786954259496,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9823220508096056e-05,
      "loss": 0.3248,
      "step": 1631
    },
    {
      "epoch": 0.12351355186604229,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9822997469674116e-05,
      "loss": 0.3491,
      "step": 1632
    },
    {
      "epoch": 0.12358923418948961,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9822774291896093e-05,
      "loss": 0.3047,
      "step": 1633
    },
    {
      "epoch": 0.12366491651293694,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9822550974765166e-05,
      "loss": 0.3238,
      "step": 1634
    },
    {
      "epoch": 0.12374059883638427,
      "grad_norm": 0.9375,
      "learning_rate": 1.982232751828449e-05,
      "loss": 0.3835,
      "step": 1635
    },
    {
      "epoch": 0.1238162811598316,
      "grad_norm": 1.0234375,
      "learning_rate": 1.9822103922457253e-05,
      "loss": 0.3873,
      "step": 1636
    },
    {
      "epoch": 0.12389196348327894,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9821880187286607e-05,
      "loss": 0.3509,
      "step": 1637
    },
    {
      "epoch": 0.12396764580672627,
      "grad_norm": 0.921875,
      "learning_rate": 1.9821656312775742e-05,
      "loss": 0.3542,
      "step": 1638
    },
    {
      "epoch": 0.1240433281301736,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9821432298927824e-05,
      "loss": 0.3726,
      "step": 1639
    },
    {
      "epoch": 0.12411901045362092,
      "grad_norm": 0.921875,
      "learning_rate": 1.9821208145746036e-05,
      "loss": 0.3381,
      "step": 1640
    },
    {
      "epoch": 0.12419469277706825,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9820983853233556e-05,
      "loss": 0.3236,
      "step": 1641
    },
    {
      "epoch": 0.12427037510051558,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9820759421393568e-05,
      "loss": 0.3656,
      "step": 1642
    },
    {
      "epoch": 0.12434605742396292,
      "grad_norm": 0.90234375,
      "learning_rate": 1.982053485022925e-05,
      "loss": 0.3291,
      "step": 1643
    },
    {
      "epoch": 0.12442173974741025,
      "grad_norm": 0.8984375,
      "learning_rate": 1.98203101397438e-05,
      "loss": 0.3456,
      "step": 1644
    },
    {
      "epoch": 0.12449742207085758,
      "grad_norm": 3.484375,
      "learning_rate": 1.9820085289940393e-05,
      "loss": 0.6152,
      "step": 1645
    },
    {
      "epoch": 0.1245731043943049,
      "grad_norm": 0.8125,
      "learning_rate": 1.9819860300822227e-05,
      "loss": 0.3138,
      "step": 1646
    },
    {
      "epoch": 0.12464878671775223,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9819635172392487e-05,
      "loss": 0.3357,
      "step": 1647
    },
    {
      "epoch": 0.12472446904119956,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9819409904654376e-05,
      "loss": 0.352,
      "step": 1648
    },
    {
      "epoch": 0.1248001513646469,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9819184497611085e-05,
      "loss": 0.3278,
      "step": 1649
    },
    {
      "epoch": 0.12487583368809423,
      "grad_norm": 0.83203125,
      "learning_rate": 1.981895895126581e-05,
      "loss": 0.3372,
      "step": 1650
    },
    {
      "epoch": 0.12495151601154156,
      "grad_norm": 2.21875,
      "learning_rate": 1.981873326562175e-05,
      "loss": 0.5653,
      "step": 1651
    },
    {
      "epoch": 0.12502719833498888,
      "grad_norm": 0.96484375,
      "learning_rate": 1.9818507440682114e-05,
      "loss": 0.3992,
      "step": 1652
    },
    {
      "epoch": 0.12510288065843622,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9818281476450094e-05,
      "loss": 0.374,
      "step": 1653
    },
    {
      "epoch": 0.12517856298188354,
      "grad_norm": 0.921875,
      "learning_rate": 1.9818055372928908e-05,
      "loss": 0.3889,
      "step": 1654
    },
    {
      "epoch": 0.12525424530533089,
      "grad_norm": 0.76171875,
      "learning_rate": 1.9817829130121758e-05,
      "loss": 0.2519,
      "step": 1655
    },
    {
      "epoch": 0.1253299276287782,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9817602748031852e-05,
      "loss": 0.3447,
      "step": 1656
    },
    {
      "epoch": 0.12540560995222552,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9817376226662404e-05,
      "loss": 0.336,
      "step": 1657
    },
    {
      "epoch": 0.12548129227567287,
      "grad_norm": 0.875,
      "learning_rate": 1.9817149566016626e-05,
      "loss": 0.311,
      "step": 1658
    },
    {
      "epoch": 0.12555697459912019,
      "grad_norm": 0.90625,
      "learning_rate": 1.9816922766097736e-05,
      "loss": 0.3164,
      "step": 1659
    },
    {
      "epoch": 0.12563265692256753,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9816695826908948e-05,
      "loss": 0.3953,
      "step": 1660
    },
    {
      "epoch": 0.12570833924601485,
      "grad_norm": 0.890625,
      "learning_rate": 1.9816468748453486e-05,
      "loss": 0.3657,
      "step": 1661
    },
    {
      "epoch": 0.1257840215694622,
      "grad_norm": 0.953125,
      "learning_rate": 1.981624153073457e-05,
      "loss": 0.3665,
      "step": 1662
    },
    {
      "epoch": 0.1258597038929095,
      "grad_norm": 0.89453125,
      "learning_rate": 1.981601417375542e-05,
      "loss": 0.3693,
      "step": 1663
    },
    {
      "epoch": 0.12593538621635683,
      "grad_norm": 0.859375,
      "learning_rate": 1.9815786677519265e-05,
      "loss": 0.2748,
      "step": 1664
    },
    {
      "epoch": 0.12601106853980418,
      "grad_norm": 0.83203125,
      "learning_rate": 1.981555904202933e-05,
      "loss": 0.3128,
      "step": 1665
    },
    {
      "epoch": 0.1260867508632515,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9815331267288847e-05,
      "loss": 0.3931,
      "step": 1666
    },
    {
      "epoch": 0.12616243318669884,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9815103353301048e-05,
      "loss": 0.3138,
      "step": 1667
    },
    {
      "epoch": 0.12623811551014616,
      "grad_norm": 0.828125,
      "learning_rate": 1.9814875300069164e-05,
      "loss": 0.3308,
      "step": 1668
    },
    {
      "epoch": 0.12631379783359348,
      "grad_norm": 0.87890625,
      "learning_rate": 1.981464710759643e-05,
      "loss": 0.3141,
      "step": 1669
    },
    {
      "epoch": 0.12638948015704082,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9814418775886083e-05,
      "loss": 0.3445,
      "step": 1670
    },
    {
      "epoch": 0.12646516248048814,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9814190304941368e-05,
      "loss": 0.3416,
      "step": 1671
    },
    {
      "epoch": 0.1265408448039355,
      "grad_norm": 0.84375,
      "learning_rate": 1.9813961694765518e-05,
      "loss": 0.3251,
      "step": 1672
    },
    {
      "epoch": 0.1266165271273828,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9813732945361784e-05,
      "loss": 0.3613,
      "step": 1673
    },
    {
      "epoch": 0.12669220945083015,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9813504056733403e-05,
      "loss": 0.359,
      "step": 1674
    },
    {
      "epoch": 0.12676789177427747,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9813275028883624e-05,
      "loss": 0.3179,
      "step": 1675
    },
    {
      "epoch": 0.1268435740977248,
      "grad_norm": 0.890625,
      "learning_rate": 1.9813045861815706e-05,
      "loss": 0.3783,
      "step": 1676
    },
    {
      "epoch": 0.12691925642117213,
      "grad_norm": 1.171875,
      "learning_rate": 1.981281655553289e-05,
      "loss": 0.4095,
      "step": 1677
    },
    {
      "epoch": 0.12699493874461945,
      "grad_norm": 0.921875,
      "learning_rate": 1.9812587110038428e-05,
      "loss": 0.3812,
      "step": 1678
    },
    {
      "epoch": 0.1270706210680668,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9812357525335583e-05,
      "loss": 0.3323,
      "step": 1679
    },
    {
      "epoch": 0.12714630339151411,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9812127801427606e-05,
      "loss": 0.3109,
      "step": 1680
    },
    {
      "epoch": 0.12722198571496146,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9811897938317757e-05,
      "loss": 0.384,
      "step": 1681
    },
    {
      "epoch": 0.12729766803840878,
      "grad_norm": 0.87109375,
      "learning_rate": 1.98116679360093e-05,
      "loss": 0.3753,
      "step": 1682
    },
    {
      "epoch": 0.1273733503618561,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9811437794505496e-05,
      "loss": 0.3221,
      "step": 1683
    },
    {
      "epoch": 0.12744903268530344,
      "grad_norm": 0.8515625,
      "learning_rate": 1.981120751380961e-05,
      "loss": 0.3434,
      "step": 1684
    },
    {
      "epoch": 0.12752471500875076,
      "grad_norm": 0.90625,
      "learning_rate": 1.981097709392491e-05,
      "loss": 0.3529,
      "step": 1685
    },
    {
      "epoch": 0.1276003973321981,
      "grad_norm": 0.890625,
      "learning_rate": 1.9810746534854657e-05,
      "loss": 0.3619,
      "step": 1686
    },
    {
      "epoch": 0.12767607965564542,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9810515836602137e-05,
      "loss": 0.3472,
      "step": 1687
    },
    {
      "epoch": 0.12775176197909277,
      "grad_norm": 0.93359375,
      "learning_rate": 1.981028499917061e-05,
      "loss": 0.3384,
      "step": 1688
    },
    {
      "epoch": 0.1278274443025401,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9810054022563358e-05,
      "loss": 0.3341,
      "step": 1689
    },
    {
      "epoch": 0.1279031266259874,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9809822906783652e-05,
      "loss": 0.3169,
      "step": 1690
    },
    {
      "epoch": 0.12797880894943475,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9809591651834773e-05,
      "loss": 0.3683,
      "step": 1691
    },
    {
      "epoch": 0.12805449127288207,
      "grad_norm": 0.94921875,
      "learning_rate": 1.9809360257720004e-05,
      "loss": 0.3781,
      "step": 1692
    },
    {
      "epoch": 0.12813017359632942,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9809128724442627e-05,
      "loss": 0.3578,
      "step": 1693
    },
    {
      "epoch": 0.12820585591977673,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9808897052005927e-05,
      "loss": 0.3362,
      "step": 1694
    },
    {
      "epoch": 0.12828153824322408,
      "grad_norm": 0.91015625,
      "learning_rate": 1.980866524041319e-05,
      "loss": 0.3549,
      "step": 1695
    },
    {
      "epoch": 0.1283572205666714,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9808433289667702e-05,
      "loss": 0.3716,
      "step": 1696
    },
    {
      "epoch": 0.12843290289011872,
      "grad_norm": 0.9375,
      "learning_rate": 1.9808201199772755e-05,
      "loss": 0.3808,
      "step": 1697
    },
    {
      "epoch": 0.12850858521356606,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9807968970731643e-05,
      "loss": 0.3036,
      "step": 1698
    },
    {
      "epoch": 0.12858426753701338,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9807736602547663e-05,
      "loss": 0.3322,
      "step": 1699
    },
    {
      "epoch": 0.12865994986046073,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9807504095224105e-05,
      "loss": 0.329,
      "step": 1700
    },
    {
      "epoch": 0.12873563218390804,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9807271448764273e-05,
      "loss": 0.3587,
      "step": 1701
    },
    {
      "epoch": 0.12881131450735536,
      "grad_norm": 0.82421875,
      "learning_rate": 1.980703866317147e-05,
      "loss": 0.3195,
      "step": 1702
    },
    {
      "epoch": 0.1288869968308027,
      "grad_norm": 0.87109375,
      "learning_rate": 1.980680573844899e-05,
      "loss": 0.3458,
      "step": 1703
    },
    {
      "epoch": 0.12896267915425003,
      "grad_norm": 12.75,
      "learning_rate": 1.9806572674600137e-05,
      "loss": 0.5445,
      "step": 1704
    },
    {
      "epoch": 0.12903836147769737,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9806339471628227e-05,
      "loss": 0.3809,
      "step": 1705
    },
    {
      "epoch": 0.1291140438011447,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9806106129536565e-05,
      "loss": 0.3406,
      "step": 1706
    },
    {
      "epoch": 0.12918972612459204,
      "grad_norm": 0.95703125,
      "learning_rate": 1.9805872648328453e-05,
      "loss": 0.4039,
      "step": 1707
    },
    {
      "epoch": 0.12926540844803935,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9805639028007218e-05,
      "loss": 0.3866,
      "step": 1708
    },
    {
      "epoch": 0.12934109077148667,
      "grad_norm": 0.90625,
      "learning_rate": 1.9805405268576162e-05,
      "loss": 0.3403,
      "step": 1709
    },
    {
      "epoch": 0.12941677309493402,
      "grad_norm": 0.859375,
      "learning_rate": 1.9805171370038607e-05,
      "loss": 0.331,
      "step": 1710
    },
    {
      "epoch": 0.12949245541838134,
      "grad_norm": 0.90625,
      "learning_rate": 1.980493733239787e-05,
      "loss": 0.3589,
      "step": 1711
    },
    {
      "epoch": 0.12956813774182868,
      "grad_norm": 0.87109375,
      "learning_rate": 1.980470315565727e-05,
      "loss": 0.3185,
      "step": 1712
    },
    {
      "epoch": 0.129643820065276,
      "grad_norm": 0.90234375,
      "learning_rate": 1.980446883982013e-05,
      "loss": 0.3755,
      "step": 1713
    },
    {
      "epoch": 0.12971950238872335,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9804234384889774e-05,
      "loss": 0.3508,
      "step": 1714
    },
    {
      "epoch": 0.12979518471217066,
      "grad_norm": 0.90234375,
      "learning_rate": 1.980399979086953e-05,
      "loss": 0.3337,
      "step": 1715
    },
    {
      "epoch": 0.12987086703561798,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9803765057762723e-05,
      "loss": 0.2916,
      "step": 1716
    },
    {
      "epoch": 0.12994654935906533,
      "grad_norm": 0.96484375,
      "learning_rate": 1.9803530185572687e-05,
      "loss": 0.3832,
      "step": 1717
    },
    {
      "epoch": 0.13002223168251265,
      "grad_norm": 0.890625,
      "learning_rate": 1.9803295174302752e-05,
      "loss": 0.3591,
      "step": 1718
    },
    {
      "epoch": 0.13009791400596,
      "grad_norm": 0.859375,
      "learning_rate": 1.980306002395625e-05,
      "loss": 0.3029,
      "step": 1719
    },
    {
      "epoch": 0.1301735963294073,
      "grad_norm": 0.82421875,
      "learning_rate": 1.980282473453652e-05,
      "loss": 0.3067,
      "step": 1720
    },
    {
      "epoch": 0.13024927865285466,
      "grad_norm": 0.91796875,
      "learning_rate": 1.98025893060469e-05,
      "loss": 0.3798,
      "step": 1721
    },
    {
      "epoch": 0.13032496097630197,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9802353738490727e-05,
      "loss": 0.3333,
      "step": 1722
    },
    {
      "epoch": 0.1304006432997493,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9802118031871347e-05,
      "loss": 0.3444,
      "step": 1723
    },
    {
      "epoch": 0.13047632562319664,
      "grad_norm": 0.89453125,
      "learning_rate": 1.98018821861921e-05,
      "loss": 0.3855,
      "step": 1724
    },
    {
      "epoch": 0.13055200794664396,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9801646201456334e-05,
      "loss": 0.4007,
      "step": 1725
    },
    {
      "epoch": 0.1306276902700913,
      "grad_norm": 0.859375,
      "learning_rate": 1.9801410077667396e-05,
      "loss": 0.3413,
      "step": 1726
    },
    {
      "epoch": 0.13070337259353862,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9801173814828635e-05,
      "loss": 0.3305,
      "step": 1727
    },
    {
      "epoch": 0.13077905491698597,
      "grad_norm": 0.859375,
      "learning_rate": 1.980093741294341e-05,
      "loss": 0.3576,
      "step": 1728
    },
    {
      "epoch": 0.13085473724043328,
      "grad_norm": 0.90625,
      "learning_rate": 1.980070087201506e-05,
      "loss": 0.3741,
      "step": 1729
    },
    {
      "epoch": 0.1309304195638806,
      "grad_norm": 0.96875,
      "learning_rate": 1.9800464192046956e-05,
      "loss": 0.3913,
      "step": 1730
    },
    {
      "epoch": 0.13100610188732795,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9800227373042446e-05,
      "loss": 0.3464,
      "step": 1731
    },
    {
      "epoch": 0.13108178421077527,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9799990415004897e-05,
      "loss": 0.3047,
      "step": 1732
    },
    {
      "epoch": 0.1311574665342226,
      "grad_norm": 1.21875,
      "learning_rate": 1.979975331793766e-05,
      "loss": 0.3597,
      "step": 1733
    },
    {
      "epoch": 0.13123314885766993,
      "grad_norm": 0.8984375,
      "learning_rate": 1.979951608184411e-05,
      "loss": 0.3372,
      "step": 1734
    },
    {
      "epoch": 0.13130883118111725,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9799278706727606e-05,
      "loss": 0.3116,
      "step": 1735
    },
    {
      "epoch": 0.1313845135045646,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9799041192591518e-05,
      "loss": 0.3537,
      "step": 1736
    },
    {
      "epoch": 0.1314601958280119,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9798803539439217e-05,
      "loss": 0.3343,
      "step": 1737
    },
    {
      "epoch": 0.13153587815145926,
      "grad_norm": 0.89453125,
      "learning_rate": 1.979856574727407e-05,
      "loss": 0.2967,
      "step": 1738
    },
    {
      "epoch": 0.13161156047490657,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9798327816099453e-05,
      "loss": 0.3363,
      "step": 1739
    },
    {
      "epoch": 0.13168724279835392,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9798089745918744e-05,
      "loss": 0.3334,
      "step": 1740
    },
    {
      "epoch": 0.13176292512180124,
      "grad_norm": 0.953125,
      "learning_rate": 1.9797851536735314e-05,
      "loss": 0.3613,
      "step": 1741
    },
    {
      "epoch": 0.13183860744524856,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9797613188552547e-05,
      "loss": 0.3126,
      "step": 1742
    },
    {
      "epoch": 0.1319142897686959,
      "grad_norm": 0.84375,
      "learning_rate": 1.9797374701373824e-05,
      "loss": 0.3555,
      "step": 1743
    },
    {
      "epoch": 0.13198997209214322,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9797136075202532e-05,
      "loss": 0.3628,
      "step": 1744
    },
    {
      "epoch": 0.13206565441559057,
      "grad_norm": 0.98046875,
      "learning_rate": 1.979689731004205e-05,
      "loss": 0.3862,
      "step": 1745
    },
    {
      "epoch": 0.13214133673903788,
      "grad_norm": 0.984375,
      "learning_rate": 1.9796658405895764e-05,
      "loss": 0.3573,
      "step": 1746
    },
    {
      "epoch": 0.13221701906248523,
      "grad_norm": 3.453125,
      "learning_rate": 1.979641936276707e-05,
      "loss": 0.5546,
      "step": 1747
    },
    {
      "epoch": 0.13229270138593255,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9796180180659355e-05,
      "loss": 0.3304,
      "step": 1748
    },
    {
      "epoch": 0.13236838370937987,
      "grad_norm": 0.9375,
      "learning_rate": 1.9795940859576015e-05,
      "loss": 0.3809,
      "step": 1749
    },
    {
      "epoch": 0.1324440660328272,
      "grad_norm": 0.84375,
      "learning_rate": 1.979570139952044e-05,
      "loss": 0.2976,
      "step": 1750
    },
    {
      "epoch": 0.13251974835627453,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9795461800496033e-05,
      "loss": 0.3484,
      "step": 1751
    },
    {
      "epoch": 0.13259543067972188,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9795222062506193e-05,
      "loss": 0.3579,
      "step": 1752
    },
    {
      "epoch": 0.1326711130031692,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9794982185554317e-05,
      "loss": 0.3073,
      "step": 1753
    },
    {
      "epoch": 0.13274679532661654,
      "grad_norm": 0.84765625,
      "learning_rate": 1.979474216964381e-05,
      "loss": 0.2742,
      "step": 1754
    },
    {
      "epoch": 0.13282247765006386,
      "grad_norm": 0.9375,
      "learning_rate": 1.9794502014778076e-05,
      "loss": 0.3498,
      "step": 1755
    },
    {
      "epoch": 0.13289815997351118,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9794261720960524e-05,
      "loss": 0.3428,
      "step": 1756
    },
    {
      "epoch": 0.13297384229695852,
      "grad_norm": 0.8984375,
      "learning_rate": 1.979402128819456e-05,
      "loss": 0.3307,
      "step": 1757
    },
    {
      "epoch": 0.13304952462040584,
      "grad_norm": 0.83984375,
      "learning_rate": 1.97937807164836e-05,
      "loss": 0.3335,
      "step": 1758
    },
    {
      "epoch": 0.13312520694385319,
      "grad_norm": 0.796875,
      "learning_rate": 1.979354000583105e-05,
      "loss": 0.2751,
      "step": 1759
    },
    {
      "epoch": 0.1332008892673005,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9793299156240333e-05,
      "loss": 0.2743,
      "step": 1760
    },
    {
      "epoch": 0.13327657159074782,
      "grad_norm": 0.96484375,
      "learning_rate": 1.979305816771486e-05,
      "loss": 0.3933,
      "step": 1761
    },
    {
      "epoch": 0.13335225391419517,
      "grad_norm": 0.8515625,
      "learning_rate": 1.979281704025805e-05,
      "loss": 0.3581,
      "step": 1762
    },
    {
      "epoch": 0.13342793623764249,
      "grad_norm": 0.9140625,
      "learning_rate": 1.979257577387333e-05,
      "loss": 0.3616,
      "step": 1763
    },
    {
      "epoch": 0.13350361856108983,
      "grad_norm": 0.81640625,
      "learning_rate": 1.979233436856411e-05,
      "loss": 0.3179,
      "step": 1764
    },
    {
      "epoch": 0.13357930088453715,
      "grad_norm": 0.984375,
      "learning_rate": 1.9792092824333828e-05,
      "loss": 0.3191,
      "step": 1765
    },
    {
      "epoch": 0.1336549832079845,
      "grad_norm": 0.9765625,
      "learning_rate": 1.9791851141185903e-05,
      "loss": 0.3258,
      "step": 1766
    },
    {
      "epoch": 0.1337306655314318,
      "grad_norm": 0.93359375,
      "learning_rate": 1.979160931912377e-05,
      "loss": 0.3863,
      "step": 1767
    },
    {
      "epoch": 0.13380634785487913,
      "grad_norm": 7.03125,
      "learning_rate": 1.979136735815085e-05,
      "loss": 0.5075,
      "step": 1768
    },
    {
      "epoch": 0.13388203017832648,
      "grad_norm": 0.95703125,
      "learning_rate": 1.9791125258270587e-05,
      "loss": 0.4124,
      "step": 1769
    },
    {
      "epoch": 0.1339577125017738,
      "grad_norm": 1.0,
      "learning_rate": 1.979088301948641e-05,
      "loss": 0.3655,
      "step": 1770
    },
    {
      "epoch": 0.13403339482522114,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9790640641801752e-05,
      "loss": 0.3171,
      "step": 1771
    },
    {
      "epoch": 0.13410907714866846,
      "grad_norm": 0.84375,
      "learning_rate": 1.9790398125220056e-05,
      "loss": 0.311,
      "step": 1772
    },
    {
      "epoch": 0.1341847594721158,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9790155469744766e-05,
      "loss": 0.3545,
      "step": 1773
    },
    {
      "epoch": 0.13426044179556312,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9789912675379316e-05,
      "loss": 0.3628,
      "step": 1774
    },
    {
      "epoch": 0.13433612411901044,
      "grad_norm": 0.875,
      "learning_rate": 1.9789669742127154e-05,
      "loss": 0.3249,
      "step": 1775
    },
    {
      "epoch": 0.1344118064424578,
      "grad_norm": 0.84765625,
      "learning_rate": 1.978942666999173e-05,
      "loss": 0.3192,
      "step": 1776
    },
    {
      "epoch": 0.1344874887659051,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9789183458976485e-05,
      "loss": 0.3884,
      "step": 1777
    },
    {
      "epoch": 0.13456317108935245,
      "grad_norm": 2.15625,
      "learning_rate": 1.9788940109084876e-05,
      "loss": 0.5151,
      "step": 1778
    },
    {
      "epoch": 0.13463885341279977,
      "grad_norm": 0.875,
      "learning_rate": 1.9788696620320355e-05,
      "loss": 0.3406,
      "step": 1779
    },
    {
      "epoch": 0.13471453573624712,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9788452992686372e-05,
      "loss": 0.3382,
      "step": 1780
    },
    {
      "epoch": 0.13479021805969443,
      "grad_norm": 0.84765625,
      "learning_rate": 1.978820922618639e-05,
      "loss": 0.3215,
      "step": 1781
    },
    {
      "epoch": 0.13486590038314175,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9787965320823858e-05,
      "loss": 0.2933,
      "step": 1782
    },
    {
      "epoch": 0.1349415827065891,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9787721276602245e-05,
      "loss": 0.3827,
      "step": 1783
    },
    {
      "epoch": 0.13501726503003642,
      "grad_norm": 0.7890625,
      "learning_rate": 1.9787477093525008e-05,
      "loss": 0.2913,
      "step": 1784
    },
    {
      "epoch": 0.13509294735348376,
      "grad_norm": 2.25,
      "learning_rate": 1.978723277159561e-05,
      "loss": 0.5521,
      "step": 1785
    },
    {
      "epoch": 0.13516862967693108,
      "grad_norm": 2.546875,
      "learning_rate": 1.9786988310817523e-05,
      "loss": 0.3724,
      "step": 1786
    },
    {
      "epoch": 0.13524431200037842,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9786743711194208e-05,
      "loss": 0.3436,
      "step": 1787
    },
    {
      "epoch": 0.13531999432382574,
      "grad_norm": 0.875,
      "learning_rate": 1.9786498972729142e-05,
      "loss": 0.3288,
      "step": 1788
    },
    {
      "epoch": 0.13539567664727306,
      "grad_norm": 2.15625,
      "learning_rate": 1.9786254095425795e-05,
      "loss": 0.4652,
      "step": 1789
    },
    {
      "epoch": 0.1354713589707204,
      "grad_norm": 0.95703125,
      "learning_rate": 1.978600907928764e-05,
      "loss": 0.3874,
      "step": 1790
    },
    {
      "epoch": 0.13554704129416772,
      "grad_norm": 0.9453125,
      "learning_rate": 1.978576392431815e-05,
      "loss": 0.313,
      "step": 1791
    },
    {
      "epoch": 0.13562272361761507,
      "grad_norm": 2.609375,
      "learning_rate": 1.9785518630520807e-05,
      "loss": 0.4818,
      "step": 1792
    },
    {
      "epoch": 0.1356984059410624,
      "grad_norm": 0.921875,
      "learning_rate": 1.978527319789909e-05,
      "loss": 0.3114,
      "step": 1793
    },
    {
      "epoch": 0.1357740882645097,
      "grad_norm": 0.90234375,
      "learning_rate": 1.978502762645648e-05,
      "loss": 0.3555,
      "step": 1794
    },
    {
      "epoch": 0.13584977058795705,
      "grad_norm": 0.96484375,
      "learning_rate": 1.978478191619646e-05,
      "loss": 0.3885,
      "step": 1795
    },
    {
      "epoch": 0.13592545291140437,
      "grad_norm": 0.9921875,
      "learning_rate": 1.9784536067122518e-05,
      "loss": 0.3786,
      "step": 1796
    },
    {
      "epoch": 0.13600113523485172,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9784290079238142e-05,
      "loss": 0.349,
      "step": 1797
    },
    {
      "epoch": 0.13607681755829903,
      "grad_norm": 0.890625,
      "learning_rate": 1.978404395254682e-05,
      "loss": 0.3665,
      "step": 1798
    },
    {
      "epoch": 0.13615249988174638,
      "grad_norm": 4.09375,
      "learning_rate": 1.9783797687052046e-05,
      "loss": 0.468,
      "step": 1799
    },
    {
      "epoch": 0.1362281822051937,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9783551282757312e-05,
      "loss": 0.4138,
      "step": 1800
    },
    {
      "epoch": 0.13630386452864102,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9783304739666112e-05,
      "loss": 0.3492,
      "step": 1801
    },
    {
      "epoch": 0.13637954685208836,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9783058057781947e-05,
      "loss": 0.3188,
      "step": 1802
    },
    {
      "epoch": 0.13645522917553568,
      "grad_norm": 1.015625,
      "learning_rate": 1.9782811237108313e-05,
      "loss": 0.3157,
      "step": 1803
    },
    {
      "epoch": 0.13653091149898303,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9782564277648717e-05,
      "loss": 0.3766,
      "step": 1804
    },
    {
      "epoch": 0.13660659382243034,
      "grad_norm": 0.9296875,
      "learning_rate": 1.978231717940666e-05,
      "loss": 0.3742,
      "step": 1805
    },
    {
      "epoch": 0.1366822761458777,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9782069942385642e-05,
      "loss": 0.3613,
      "step": 1806
    },
    {
      "epoch": 0.136757958469325,
      "grad_norm": 0.85546875,
      "learning_rate": 1.978182256658918e-05,
      "loss": 0.3485,
      "step": 1807
    },
    {
      "epoch": 0.13683364079277233,
      "grad_norm": 0.9453125,
      "learning_rate": 1.9781575052020775e-05,
      "loss": 0.3585,
      "step": 1808
    },
    {
      "epoch": 0.13690932311621967,
      "grad_norm": 0.95703125,
      "learning_rate": 1.9781327398683945e-05,
      "loss": 0.3562,
      "step": 1809
    },
    {
      "epoch": 0.136985005439667,
      "grad_norm": 0.9296875,
      "learning_rate": 1.97810796065822e-05,
      "loss": 0.3484,
      "step": 1810
    },
    {
      "epoch": 0.13706068776311434,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9780831675719053e-05,
      "loss": 0.318,
      "step": 1811
    },
    {
      "epoch": 0.13713637008656165,
      "grad_norm": 0.8359375,
      "learning_rate": 1.978058360609803e-05,
      "loss": 0.2911,
      "step": 1812
    },
    {
      "epoch": 0.137212052410009,
      "grad_norm": 0.78515625,
      "learning_rate": 1.978033539772264e-05,
      "loss": 0.2881,
      "step": 1813
    },
    {
      "epoch": 0.13728773473345632,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9780087050596407e-05,
      "loss": 0.3005,
      "step": 1814
    },
    {
      "epoch": 0.13736341705690364,
      "grad_norm": 0.9140625,
      "learning_rate": 1.977983856472286e-05,
      "loss": 0.3685,
      "step": 1815
    },
    {
      "epoch": 0.13743909938035098,
      "grad_norm": 0.82421875,
      "learning_rate": 1.977958994010552e-05,
      "loss": 0.3251,
      "step": 1816
    },
    {
      "epoch": 0.1375147817037983,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9779341176747913e-05,
      "loss": 0.3379,
      "step": 1817
    },
    {
      "epoch": 0.13759046402724565,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9779092274653567e-05,
      "loss": 0.3283,
      "step": 1818
    },
    {
      "epoch": 0.13766614635069296,
      "grad_norm": 0.8515625,
      "learning_rate": 1.977884323382602e-05,
      "loss": 0.3138,
      "step": 1819
    },
    {
      "epoch": 0.1377418286741403,
      "grad_norm": 0.84375,
      "learning_rate": 1.97785940542688e-05,
      "loss": 0.3373,
      "step": 1820
    },
    {
      "epoch": 0.13781751099758763,
      "grad_norm": 0.8046875,
      "learning_rate": 1.977834473598544e-05,
      "loss": 0.2707,
      "step": 1821
    },
    {
      "epoch": 0.13789319332103495,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9778095278979482e-05,
      "loss": 0.364,
      "step": 1822
    },
    {
      "epoch": 0.1379688756444823,
      "grad_norm": 0.7734375,
      "learning_rate": 1.9777845683254464e-05,
      "loss": 0.2823,
      "step": 1823
    },
    {
      "epoch": 0.1380445579679296,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9777595948813922e-05,
      "loss": 0.3409,
      "step": 1824
    },
    {
      "epoch": 0.13812024029137696,
      "grad_norm": 0.99609375,
      "learning_rate": 1.97773460756614e-05,
      "loss": 0.3972,
      "step": 1825
    },
    {
      "epoch": 0.13819592261482427,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9777096063800453e-05,
      "loss": 0.3004,
      "step": 1826
    },
    {
      "epoch": 0.1382716049382716,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9776845913234613e-05,
      "loss": 0.341,
      "step": 1827
    },
    {
      "epoch": 0.13834728726171894,
      "grad_norm": 0.7734375,
      "learning_rate": 1.9776595623967443e-05,
      "loss": 0.2743,
      "step": 1828
    },
    {
      "epoch": 0.13842296958516626,
      "grad_norm": 0.84375,
      "learning_rate": 1.9776345196002484e-05,
      "loss": 0.3161,
      "step": 1829
    },
    {
      "epoch": 0.1384986519086136,
      "grad_norm": 0.8828125,
      "learning_rate": 1.977609462934329e-05,
      "loss": 0.3636,
      "step": 1830
    },
    {
      "epoch": 0.13857433423206092,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9775843923993418e-05,
      "loss": 0.4017,
      "step": 1831
    },
    {
      "epoch": 0.13865001655550827,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9775593079956426e-05,
      "loss": 0.3279,
      "step": 1832
    },
    {
      "epoch": 0.13872569887895558,
      "grad_norm": 0.859375,
      "learning_rate": 1.977534209723587e-05,
      "loss": 0.3212,
      "step": 1833
    },
    {
      "epoch": 0.1388013812024029,
      "grad_norm": 0.88671875,
      "learning_rate": 1.977509097583531e-05,
      "loss": 0.3407,
      "step": 1834
    },
    {
      "epoch": 0.13887706352585025,
      "grad_norm": 0.83203125,
      "learning_rate": 1.977483971575831e-05,
      "loss": 0.312,
      "step": 1835
    },
    {
      "epoch": 0.13895274584929757,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9774588317008432e-05,
      "loss": 0.2955,
      "step": 1836
    },
    {
      "epoch": 0.1390284281727449,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9774336779589253e-05,
      "loss": 0.3311,
      "step": 1837
    },
    {
      "epoch": 0.13910411049619223,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9774085103504326e-05,
      "loss": 0.3896,
      "step": 1838
    },
    {
      "epoch": 0.13917979281963958,
      "grad_norm": 0.796875,
      "learning_rate": 1.9773833288757232e-05,
      "loss": 0.2959,
      "step": 1839
    },
    {
      "epoch": 0.1392554751430869,
      "grad_norm": 0.83984375,
      "learning_rate": 1.977358133535154e-05,
      "loss": 0.3067,
      "step": 1840
    },
    {
      "epoch": 0.1393311574665342,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9773329243290826e-05,
      "loss": 0.3632,
      "step": 1841
    },
    {
      "epoch": 0.13940683978998156,
      "grad_norm": 0.89453125,
      "learning_rate": 1.977307701257866e-05,
      "loss": 0.3301,
      "step": 1842
    },
    {
      "epoch": 0.13948252211342888,
      "grad_norm": 0.859375,
      "learning_rate": 1.977282464321863e-05,
      "loss": 0.346,
      "step": 1843
    },
    {
      "epoch": 0.13955820443687622,
      "grad_norm": 0.89453125,
      "learning_rate": 1.977257213521431e-05,
      "loss": 0.32,
      "step": 1844
    },
    {
      "epoch": 0.13963388676032354,
      "grad_norm": 0.90625,
      "learning_rate": 1.9772319488569284e-05,
      "loss": 0.35,
      "step": 1845
    },
    {
      "epoch": 0.13970956908377088,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9772066703287137e-05,
      "loss": 0.3633,
      "step": 1846
    },
    {
      "epoch": 0.1397852514072182,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9771813779371452e-05,
      "loss": 0.3652,
      "step": 1847
    },
    {
      "epoch": 0.13986093373066552,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9771560716825822e-05,
      "loss": 0.3472,
      "step": 1848
    },
    {
      "epoch": 0.13993661605411287,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9771307515653835e-05,
      "loss": 0.3738,
      "step": 1849
    },
    {
      "epoch": 0.14001229837756018,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9771054175859082e-05,
      "loss": 0.3371,
      "step": 1850
    },
    {
      "epoch": 0.14008798070100753,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9770800697445157e-05,
      "loss": 0.3962,
      "step": 1851
    },
    {
      "epoch": 0.14016366302445485,
      "grad_norm": 0.875,
      "learning_rate": 1.977054708041566e-05,
      "loss": 0.3256,
      "step": 1852
    },
    {
      "epoch": 0.14023934534790217,
      "grad_norm": 0.90234375,
      "learning_rate": 1.977029332477418e-05,
      "loss": 0.3788,
      "step": 1853
    },
    {
      "epoch": 0.1403150276713495,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9770039430524325e-05,
      "loss": 0.3479,
      "step": 1854
    },
    {
      "epoch": 0.14039070999479683,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9769785397669698e-05,
      "loss": 0.3622,
      "step": 1855
    },
    {
      "epoch": 0.14046639231824418,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9769531226213897e-05,
      "loss": 0.3862,
      "step": 1856
    },
    {
      "epoch": 0.1405420746416915,
      "grad_norm": 0.9140625,
      "learning_rate": 1.976927691616053e-05,
      "loss": 0.3798,
      "step": 1857
    },
    {
      "epoch": 0.14061775696513884,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9769022467513208e-05,
      "loss": 0.3406,
      "step": 1858
    },
    {
      "epoch": 0.14069343928858616,
      "grad_norm": 0.984375,
      "learning_rate": 1.9768767880275534e-05,
      "loss": 0.4067,
      "step": 1859
    },
    {
      "epoch": 0.14076912161203348,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9768513154451126e-05,
      "loss": 0.3329,
      "step": 1860
    },
    {
      "epoch": 0.14084480393548082,
      "grad_norm": 0.875,
      "learning_rate": 1.9768258290043597e-05,
      "loss": 0.3484,
      "step": 1861
    },
    {
      "epoch": 0.14092048625892814,
      "grad_norm": 0.91015625,
      "learning_rate": 1.976800328705656e-05,
      "loss": 0.3845,
      "step": 1862
    },
    {
      "epoch": 0.1409961685823755,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9767748145493635e-05,
      "loss": 0.3481,
      "step": 1863
    },
    {
      "epoch": 0.1410718509058228,
      "grad_norm": 0.84375,
      "learning_rate": 1.976749286535844e-05,
      "loss": 0.3467,
      "step": 1864
    },
    {
      "epoch": 0.14114753322927015,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9767237446654595e-05,
      "loss": 0.3658,
      "step": 1865
    },
    {
      "epoch": 0.14122321555271747,
      "grad_norm": 0.859375,
      "learning_rate": 1.9766981889385727e-05,
      "loss": 0.32,
      "step": 1866
    },
    {
      "epoch": 0.1412988978761648,
      "grad_norm": 0.828125,
      "learning_rate": 1.9766726193555458e-05,
      "loss": 0.339,
      "step": 1867
    },
    {
      "epoch": 0.14137458019961213,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9766470359167424e-05,
      "loss": 0.3441,
      "step": 1868
    },
    {
      "epoch": 0.14145026252305945,
      "grad_norm": 0.87890625,
      "learning_rate": 1.976621438622524e-05,
      "loss": 0.3397,
      "step": 1869
    },
    {
      "epoch": 0.1415259448465068,
      "grad_norm": 0.953125,
      "learning_rate": 1.976595827473255e-05,
      "loss": 0.3621,
      "step": 1870
    },
    {
      "epoch": 0.14160162716995411,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9765702024692984e-05,
      "loss": 0.3182,
      "step": 1871
    },
    {
      "epoch": 0.14167730949340146,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9765445636110175e-05,
      "loss": 0.3776,
      "step": 1872
    },
    {
      "epoch": 0.14175299181684878,
      "grad_norm": 1.046875,
      "learning_rate": 1.976518910898776e-05,
      "loss": 0.4067,
      "step": 1873
    },
    {
      "epoch": 0.1418286741402961,
      "grad_norm": 0.890625,
      "learning_rate": 1.9764932443329383e-05,
      "loss": 0.35,
      "step": 1874
    },
    {
      "epoch": 0.14190435646374344,
      "grad_norm": 0.87109375,
      "learning_rate": 1.976467563913868e-05,
      "loss": 0.348,
      "step": 1875
    },
    {
      "epoch": 0.14198003878719076,
      "grad_norm": 0.90625,
      "learning_rate": 1.9764418696419297e-05,
      "loss": 0.3852,
      "step": 1876
    },
    {
      "epoch": 0.1420557211106381,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9764161615174876e-05,
      "loss": 0.3292,
      "step": 1877
    },
    {
      "epoch": 0.14213140343408542,
      "grad_norm": 0.921875,
      "learning_rate": 1.976390439540907e-05,
      "loss": 0.379,
      "step": 1878
    },
    {
      "epoch": 0.14220708575753277,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9763647037125525e-05,
      "loss": 0.3405,
      "step": 1879
    },
    {
      "epoch": 0.1422827680809801,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9763389540327888e-05,
      "loss": 0.3056,
      "step": 1880
    },
    {
      "epoch": 0.1423584504044274,
      "grad_norm": 4.875,
      "learning_rate": 1.976313190501982e-05,
      "loss": 0.556,
      "step": 1881
    },
    {
      "epoch": 0.14243413272787475,
      "grad_norm": 0.91015625,
      "learning_rate": 1.976287413120497e-05,
      "loss": 0.3447,
      "step": 1882
    },
    {
      "epoch": 0.14250981505132207,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9762616218886995e-05,
      "loss": 0.3408,
      "step": 1883
    },
    {
      "epoch": 0.14258549737476942,
      "grad_norm": 0.78125,
      "learning_rate": 1.976235816806956e-05,
      "loss": 0.283,
      "step": 1884
    },
    {
      "epoch": 0.14266117969821673,
      "grad_norm": 2.0,
      "learning_rate": 1.9762099978756314e-05,
      "loss": 0.517,
      "step": 1885
    },
    {
      "epoch": 0.14273686202166405,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9761841650950935e-05,
      "loss": 0.2924,
      "step": 1886
    },
    {
      "epoch": 0.1428125443451114,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9761583184657076e-05,
      "loss": 0.3272,
      "step": 1887
    },
    {
      "epoch": 0.14288822666855872,
      "grad_norm": 0.875,
      "learning_rate": 1.9761324579878412e-05,
      "loss": 0.379,
      "step": 1888
    },
    {
      "epoch": 0.14296390899200606,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9761065836618603e-05,
      "loss": 0.3745,
      "step": 1889
    },
    {
      "epoch": 0.14303959131545338,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9760806954881328e-05,
      "loss": 0.3216,
      "step": 1890
    },
    {
      "epoch": 0.14311527363890073,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9760547934670256e-05,
      "loss": 0.3547,
      "step": 1891
    },
    {
      "epoch": 0.14319095596234804,
      "grad_norm": 0.7890625,
      "learning_rate": 1.976028877598906e-05,
      "loss": 0.2882,
      "step": 1892
    },
    {
      "epoch": 0.14326663828579536,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9760029478841423e-05,
      "loss": 0.3227,
      "step": 1893
    },
    {
      "epoch": 0.1433423206092427,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9759770043231015e-05,
      "loss": 0.3527,
      "step": 1894
    },
    {
      "epoch": 0.14341800293269003,
      "grad_norm": 0.8984375,
      "learning_rate": 1.975951046916152e-05,
      "loss": 0.3298,
      "step": 1895
    },
    {
      "epoch": 0.14349368525613737,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9759250756636624e-05,
      "loss": 0.3353,
      "step": 1896
    },
    {
      "epoch": 0.1435693675795847,
      "grad_norm": 0.875,
      "learning_rate": 1.9758990905660005e-05,
      "loss": 0.342,
      "step": 1897
    },
    {
      "epoch": 0.14364504990303203,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9758730916235356e-05,
      "loss": 0.2974,
      "step": 1898
    },
    {
      "epoch": 0.14372073222647935,
      "grad_norm": 0.94140625,
      "learning_rate": 1.975847078836636e-05,
      "loss": 0.3605,
      "step": 1899
    },
    {
      "epoch": 0.14379641454992667,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9758210522056713e-05,
      "loss": 0.3207,
      "step": 1900
    },
    {
      "epoch": 0.14387209687337402,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9757950117310102e-05,
      "loss": 0.309,
      "step": 1901
    },
    {
      "epoch": 0.14394777919682133,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9757689574130223e-05,
      "loss": 0.3302,
      "step": 1902
    },
    {
      "epoch": 0.14402346152026868,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9757428892520773e-05,
      "loss": 0.3374,
      "step": 1903
    },
    {
      "epoch": 0.144099143843716,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9757168072485448e-05,
      "loss": 0.3885,
      "step": 1904
    },
    {
      "epoch": 0.14417482616716334,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9756907114027954e-05,
      "loss": 0.3287,
      "step": 1905
    },
    {
      "epoch": 0.14425050849061066,
      "grad_norm": 0.84765625,
      "learning_rate": 1.975664601715199e-05,
      "loss": 0.3692,
      "step": 1906
    },
    {
      "epoch": 0.14432619081405798,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9756384781861255e-05,
      "loss": 0.2966,
      "step": 1907
    },
    {
      "epoch": 0.14440187313750533,
      "grad_norm": 0.8203125,
      "learning_rate": 1.975612340815946e-05,
      "loss": 0.3041,
      "step": 1908
    },
    {
      "epoch": 0.14447755546095264,
      "grad_norm": 0.90625,
      "learning_rate": 1.9755861896050312e-05,
      "loss": 0.3628,
      "step": 1909
    },
    {
      "epoch": 0.1445532377844,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9755600245537522e-05,
      "loss": 0.3623,
      "step": 1910
    },
    {
      "epoch": 0.1446289201078473,
      "grad_norm": 0.87890625,
      "learning_rate": 1.97553384566248e-05,
      "loss": 0.3169,
      "step": 1911
    },
    {
      "epoch": 0.14470460243129465,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9755076529315867e-05,
      "loss": 0.3215,
      "step": 1912
    },
    {
      "epoch": 0.14478028475474197,
      "grad_norm": 2.921875,
      "learning_rate": 1.9754814463614428e-05,
      "loss": 0.4627,
      "step": 1913
    },
    {
      "epoch": 0.1448559670781893,
      "grad_norm": 0.9609375,
      "learning_rate": 1.9754552259524208e-05,
      "loss": 0.3599,
      "step": 1914
    },
    {
      "epoch": 0.14493164940163664,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9754289917048926e-05,
      "loss": 0.3296,
      "step": 1915
    },
    {
      "epoch": 0.14500733172508395,
      "grad_norm": 0.8828125,
      "learning_rate": 1.97540274361923e-05,
      "loss": 0.3556,
      "step": 1916
    },
    {
      "epoch": 0.1450830140485313,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9753764816958055e-05,
      "loss": 0.3416,
      "step": 1917
    },
    {
      "epoch": 0.14515869637197862,
      "grad_norm": 0.88671875,
      "learning_rate": 1.975350205934992e-05,
      "loss": 0.3552,
      "step": 1918
    },
    {
      "epoch": 0.14523437869542594,
      "grad_norm": 0.96875,
      "learning_rate": 1.975323916337162e-05,
      "loss": 0.3666,
      "step": 1919
    },
    {
      "epoch": 0.14531006101887328,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9752976129026885e-05,
      "loss": 0.3799,
      "step": 1920
    },
    {
      "epoch": 0.1453857433423206,
      "grad_norm": 0.80078125,
      "learning_rate": 1.975271295631945e-05,
      "loss": 0.3077,
      "step": 1921
    },
    {
      "epoch": 0.14546142566576795,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9752449645253045e-05,
      "loss": 0.3575,
      "step": 1922
    },
    {
      "epoch": 0.14553710798921526,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9752186195831404e-05,
      "loss": 0.3023,
      "step": 1923
    },
    {
      "epoch": 0.1456127903126626,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9751922608058268e-05,
      "loss": 0.3357,
      "step": 1924
    },
    {
      "epoch": 0.14568847263610993,
      "grad_norm": 2.5,
      "learning_rate": 1.9751658881937374e-05,
      "loss": 0.5731,
      "step": 1925
    },
    {
      "epoch": 0.14576415495955725,
      "grad_norm": 0.859375,
      "learning_rate": 1.9751395017472462e-05,
      "loss": 0.3094,
      "step": 1926
    },
    {
      "epoch": 0.1458398372830046,
      "grad_norm": 0.97265625,
      "learning_rate": 1.975113101466728e-05,
      "loss": 0.4003,
      "step": 1927
    },
    {
      "epoch": 0.1459155196064519,
      "grad_norm": 0.91015625,
      "learning_rate": 1.975086687352557e-05,
      "loss": 0.3411,
      "step": 1928
    },
    {
      "epoch": 0.14599120192989926,
      "grad_norm": 0.984375,
      "learning_rate": 1.9750602594051083e-05,
      "loss": 0.3851,
      "step": 1929
    },
    {
      "epoch": 0.14606688425334657,
      "grad_norm": 1.03125,
      "learning_rate": 1.9750338176247563e-05,
      "loss": 0.3925,
      "step": 1930
    },
    {
      "epoch": 0.14614256657679392,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9750073620118765e-05,
      "loss": 0.3717,
      "step": 1931
    },
    {
      "epoch": 0.14621824890024124,
      "grad_norm": 0.98046875,
      "learning_rate": 1.974980892566844e-05,
      "loss": 0.4032,
      "step": 1932
    },
    {
      "epoch": 0.14629393122368856,
      "grad_norm": 0.93359375,
      "learning_rate": 1.974954409290035e-05,
      "loss": 0.371,
      "step": 1933
    },
    {
      "epoch": 0.1463696135471359,
      "grad_norm": 0.875,
      "learning_rate": 1.9749279121818235e-05,
      "loss": 0.3599,
      "step": 1934
    },
    {
      "epoch": 0.14644529587058322,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9749014012425873e-05,
      "loss": 0.3352,
      "step": 1935
    },
    {
      "epoch": 0.14652097819403057,
      "grad_norm": 0.96484375,
      "learning_rate": 1.9748748764727017e-05,
      "loss": 0.3746,
      "step": 1936
    },
    {
      "epoch": 0.14659666051747788,
      "grad_norm": 1.140625,
      "learning_rate": 1.9748483378725428e-05,
      "loss": 0.381,
      "step": 1937
    },
    {
      "epoch": 0.14667234284092523,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9748217854424877e-05,
      "loss": 0.2896,
      "step": 1938
    },
    {
      "epoch": 0.14674802516437255,
      "grad_norm": 0.953125,
      "learning_rate": 1.9747952191829124e-05,
      "loss": 0.3848,
      "step": 1939
    },
    {
      "epoch": 0.14682370748781987,
      "grad_norm": 2.671875,
      "learning_rate": 1.9747686390941944e-05,
      "loss": 0.5197,
      "step": 1940
    },
    {
      "epoch": 0.1468993898112672,
      "grad_norm": 0.94140625,
      "learning_rate": 1.97474204517671e-05,
      "loss": 0.3353,
      "step": 1941
    },
    {
      "epoch": 0.14697507213471453,
      "grad_norm": 0.890625,
      "learning_rate": 1.9747154374308373e-05,
      "loss": 0.3309,
      "step": 1942
    },
    {
      "epoch": 0.14705075445816188,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9746888158569534e-05,
      "loss": 0.326,
      "step": 1943
    },
    {
      "epoch": 0.1471264367816092,
      "grad_norm": 0.82421875,
      "learning_rate": 1.974662180455436e-05,
      "loss": 0.2817,
      "step": 1944
    },
    {
      "epoch": 0.1472021191050565,
      "grad_norm": 3.109375,
      "learning_rate": 1.974635531226663e-05,
      "loss": 0.5003,
      "step": 1945
    },
    {
      "epoch": 0.14727780142850386,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9746088681710125e-05,
      "loss": 0.3044,
      "step": 1946
    },
    {
      "epoch": 0.14735348375195118,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9745821912888625e-05,
      "loss": 0.3407,
      "step": 1947
    },
    {
      "epoch": 0.14742916607539852,
      "grad_norm": 0.86328125,
      "learning_rate": 1.974555500580592e-05,
      "loss": 0.3048,
      "step": 1948
    },
    {
      "epoch": 0.14750484839884584,
      "grad_norm": 2.546875,
      "learning_rate": 1.9745287960465794e-05,
      "loss": 0.4755,
      "step": 1949
    },
    {
      "epoch": 0.14758053072229319,
      "grad_norm": 1.0546875,
      "learning_rate": 1.974502077687203e-05,
      "loss": 0.3642,
      "step": 1950
    },
    {
      "epoch": 0.1476562130457405,
      "grad_norm": 0.90625,
      "learning_rate": 1.9744753455028426e-05,
      "loss": 0.3752,
      "step": 1951
    },
    {
      "epoch": 0.14773189536918782,
      "grad_norm": 0.8515625,
      "learning_rate": 1.974448599493877e-05,
      "loss": 0.3194,
      "step": 1952
    },
    {
      "epoch": 0.14780757769263517,
      "grad_norm": 1.796875,
      "learning_rate": 1.9744218396606863e-05,
      "loss": 0.5015,
      "step": 1953
    },
    {
      "epoch": 0.14788326001608249,
      "grad_norm": 1.5546875,
      "learning_rate": 1.974395066003649e-05,
      "loss": 0.4683,
      "step": 1954
    },
    {
      "epoch": 0.14795894233952983,
      "grad_norm": 0.84375,
      "learning_rate": 1.974368278523146e-05,
      "loss": 0.3117,
      "step": 1955
    },
    {
      "epoch": 0.14803462466297715,
      "grad_norm": 0.953125,
      "learning_rate": 1.9743414772195567e-05,
      "loss": 0.3549,
      "step": 1956
    },
    {
      "epoch": 0.1481103069864245,
      "grad_norm": 0.8515625,
      "learning_rate": 1.974314662093262e-05,
      "loss": 0.3116,
      "step": 1957
    },
    {
      "epoch": 0.1481859893098718,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9742878331446415e-05,
      "loss": 0.3248,
      "step": 1958
    },
    {
      "epoch": 0.14826167163331913,
      "grad_norm": 0.921875,
      "learning_rate": 1.9742609903740757e-05,
      "loss": 0.3002,
      "step": 1959
    },
    {
      "epoch": 0.14833735395676648,
      "grad_norm": 1.953125,
      "learning_rate": 1.974234133781946e-05,
      "loss": 0.3788,
      "step": 1960
    },
    {
      "epoch": 0.1484130362802138,
      "grad_norm": 2.375,
      "learning_rate": 1.974207263368634e-05,
      "loss": 0.4712,
      "step": 1961
    },
    {
      "epoch": 0.14848871860366114,
      "grad_norm": 1.015625,
      "learning_rate": 1.9741803791345196e-05,
      "loss": 0.3651,
      "step": 1962
    },
    {
      "epoch": 0.14856440092710846,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9741534810799846e-05,
      "loss": 0.3313,
      "step": 1963
    },
    {
      "epoch": 0.1486400832505558,
      "grad_norm": 0.84765625,
      "learning_rate": 1.974126569205411e-05,
      "loss": 0.3274,
      "step": 1964
    },
    {
      "epoch": 0.14871576557400312,
      "grad_norm": 0.94921875,
      "learning_rate": 1.9740996435111804e-05,
      "loss": 0.3828,
      "step": 1965
    },
    {
      "epoch": 0.14879144789745044,
      "grad_norm": 1.09375,
      "learning_rate": 1.974072703997675e-05,
      "loss": 0.3713,
      "step": 1966
    },
    {
      "epoch": 0.1488671302208978,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9740457506652765e-05,
      "loss": 0.2969,
      "step": 1967
    },
    {
      "epoch": 0.1489428125443451,
      "grad_norm": 1.0625,
      "learning_rate": 1.9740187835143673e-05,
      "loss": 0.3829,
      "step": 1968
    },
    {
      "epoch": 0.14901849486779245,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9739918025453303e-05,
      "loss": 0.2963,
      "step": 1969
    },
    {
      "epoch": 0.14909417719123977,
      "grad_norm": 0.90234375,
      "learning_rate": 1.973964807758548e-05,
      "loss": 0.3568,
      "step": 1970
    },
    {
      "epoch": 0.14916985951468711,
      "grad_norm": 0.890625,
      "learning_rate": 1.973937799154404e-05,
      "loss": 0.3294,
      "step": 1971
    },
    {
      "epoch": 0.14924554183813443,
      "grad_norm": 8.75,
      "learning_rate": 1.9739107767332806e-05,
      "loss": 0.5585,
      "step": 1972
    },
    {
      "epoch": 0.14932122416158175,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9738837404955616e-05,
      "loss": 0.3064,
      "step": 1973
    },
    {
      "epoch": 0.1493969064850291,
      "grad_norm": 0.890625,
      "learning_rate": 1.9738566904416307e-05,
      "loss": 0.3537,
      "step": 1974
    },
    {
      "epoch": 0.14947258880847641,
      "grad_norm": 5.5,
      "learning_rate": 1.9738296265718714e-05,
      "loss": 0.6174,
      "step": 1975
    },
    {
      "epoch": 0.14954827113192376,
      "grad_norm": 0.875,
      "learning_rate": 1.9738025488866673e-05,
      "loss": 0.302,
      "step": 1976
    },
    {
      "epoch": 0.14962395345537108,
      "grad_norm": 0.95703125,
      "learning_rate": 1.9737754573864032e-05,
      "loss": 0.2877,
      "step": 1977
    },
    {
      "epoch": 0.1496996357788184,
      "grad_norm": 3.421875,
      "learning_rate": 1.9737483520714635e-05,
      "loss": 0.4073,
      "step": 1978
    },
    {
      "epoch": 0.14977531810226574,
      "grad_norm": 1.9609375,
      "learning_rate": 1.973721232942232e-05,
      "loss": 0.4459,
      "step": 1979
    },
    {
      "epoch": 0.14985100042571306,
      "grad_norm": 0.8671875,
      "learning_rate": 1.973694099999094e-05,
      "loss": 0.3462,
      "step": 1980
    },
    {
      "epoch": 0.1499266827491604,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9736669532424343e-05,
      "loss": 0.3418,
      "step": 1981
    },
    {
      "epoch": 0.15000236507260772,
      "grad_norm": 0.84375,
      "learning_rate": 1.973639792672638e-05,
      "loss": 0.3206,
      "step": 1982
    },
    {
      "epoch": 0.15007804739605507,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9736126182900902e-05,
      "loss": 0.3513,
      "step": 1983
    },
    {
      "epoch": 0.1501537297195024,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9735854300951768e-05,
      "loss": 0.3123,
      "step": 1984
    },
    {
      "epoch": 0.1502294120429497,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9735582280882832e-05,
      "loss": 0.3538,
      "step": 1985
    },
    {
      "epoch": 0.15030509436639705,
      "grad_norm": 0.828125,
      "learning_rate": 1.973531012269796e-05,
      "loss": 0.3208,
      "step": 1986
    },
    {
      "epoch": 0.15038077668984437,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9735037826401e-05,
      "loss": 0.3677,
      "step": 1987
    },
    {
      "epoch": 0.15045645901329172,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9734765391995827e-05,
      "loss": 0.3292,
      "step": 1988
    },
    {
      "epoch": 0.15053214133673903,
      "grad_norm": 0.82421875,
      "learning_rate": 1.97344928194863e-05,
      "loss": 0.3364,
      "step": 1989
    },
    {
      "epoch": 0.15060782366018638,
      "grad_norm": 0.83203125,
      "learning_rate": 1.973422010887629e-05,
      "loss": 0.3368,
      "step": 1990
    },
    {
      "epoch": 0.1506835059836337,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9733947260169662e-05,
      "loss": 0.3185,
      "step": 1991
    },
    {
      "epoch": 0.15075918830708102,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9733674273370284e-05,
      "loss": 0.3338,
      "step": 1992
    },
    {
      "epoch": 0.15083487063052836,
      "grad_norm": 0.92578125,
      "learning_rate": 1.973340114848204e-05,
      "loss": 0.3256,
      "step": 1993
    },
    {
      "epoch": 0.15091055295397568,
      "grad_norm": 0.88671875,
      "learning_rate": 1.973312788550879e-05,
      "loss": 0.3547,
      "step": 1994
    },
    {
      "epoch": 0.15098623527742303,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9732854484454426e-05,
      "loss": 0.3423,
      "step": 1995
    },
    {
      "epoch": 0.15106191760087034,
      "grad_norm": 0.859375,
      "learning_rate": 1.9732580945322813e-05,
      "loss": 0.3644,
      "step": 1996
    },
    {
      "epoch": 0.1511375999243177,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9732307268117837e-05,
      "loss": 0.2975,
      "step": 1997
    },
    {
      "epoch": 0.151213282247765,
      "grad_norm": 1.0,
      "learning_rate": 1.9732033452843384e-05,
      "loss": 0.4169,
      "step": 1998
    },
    {
      "epoch": 0.15128896457121233,
      "grad_norm": 0.859375,
      "learning_rate": 1.9731759499503336e-05,
      "loss": 0.3549,
      "step": 1999
    },
    {
      "epoch": 0.15136464689465967,
      "grad_norm": 0.8671875,
      "learning_rate": 1.973148540810158e-05,
      "loss": 0.3217,
      "step": 2000
    },
    {
      "epoch": 0.151440329218107,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9731211178642e-05,
      "loss": 0.3206,
      "step": 2001
    },
    {
      "epoch": 0.15151601154155434,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9730936811128494e-05,
      "loss": 0.3123,
      "step": 2002
    },
    {
      "epoch": 0.15159169386500165,
      "grad_norm": 0.765625,
      "learning_rate": 1.973066230556495e-05,
      "loss": 0.2793,
      "step": 2003
    },
    {
      "epoch": 0.151667376188449,
      "grad_norm": 0.83203125,
      "learning_rate": 1.973038766195526e-05,
      "loss": 0.3297,
      "step": 2004
    },
    {
      "epoch": 0.15174305851189632,
      "grad_norm": 0.98828125,
      "learning_rate": 1.9730112880303326e-05,
      "loss": 0.3736,
      "step": 2005
    },
    {
      "epoch": 0.15181874083534364,
      "grad_norm": 0.890625,
      "learning_rate": 1.9729837960613042e-05,
      "loss": 0.3195,
      "step": 2006
    },
    {
      "epoch": 0.15189442315879098,
      "grad_norm": 0.890625,
      "learning_rate": 1.972956290288831e-05,
      "loss": 0.3615,
      "step": 2007
    },
    {
      "epoch": 0.1519701054822383,
      "grad_norm": 0.81640625,
      "learning_rate": 1.972928770713303e-05,
      "loss": 0.3118,
      "step": 2008
    },
    {
      "epoch": 0.15204578780568564,
      "grad_norm": 0.9140625,
      "learning_rate": 1.972901237335111e-05,
      "loss": 0.3853,
      "step": 2009
    },
    {
      "epoch": 0.15212147012913296,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9728736901546454e-05,
      "loss": 0.4108,
      "step": 2010
    },
    {
      "epoch": 0.15219715245258028,
      "grad_norm": 1.0078125,
      "learning_rate": 1.972846129172297e-05,
      "loss": 0.4034,
      "step": 2011
    },
    {
      "epoch": 0.15227283477602763,
      "grad_norm": 0.90625,
      "learning_rate": 1.972818554388457e-05,
      "loss": 0.3221,
      "step": 2012
    },
    {
      "epoch": 0.15234851709947494,
      "grad_norm": 0.8359375,
      "learning_rate": 1.972790965803516e-05,
      "loss": 0.3273,
      "step": 2013
    },
    {
      "epoch": 0.1524241994229223,
      "grad_norm": 0.875,
      "learning_rate": 1.9727633634178658e-05,
      "loss": 0.3287,
      "step": 2014
    },
    {
      "epoch": 0.1524998817463696,
      "grad_norm": 0.97265625,
      "learning_rate": 1.972735747231898e-05,
      "loss": 0.369,
      "step": 2015
    },
    {
      "epoch": 0.15257556406981695,
      "grad_norm": 0.921875,
      "learning_rate": 1.9727081172460046e-05,
      "loss": 0.3669,
      "step": 2016
    },
    {
      "epoch": 0.15265124639326427,
      "grad_norm": 0.9453125,
      "learning_rate": 1.9726804734605773e-05,
      "loss": 0.3659,
      "step": 2017
    },
    {
      "epoch": 0.1527269287167116,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9726528158760078e-05,
      "loss": 0.4054,
      "step": 2018
    },
    {
      "epoch": 0.15280261104015894,
      "grad_norm": 0.875,
      "learning_rate": 1.9726251444926895e-05,
      "loss": 0.3763,
      "step": 2019
    },
    {
      "epoch": 0.15287829336360625,
      "grad_norm": 0.84765625,
      "learning_rate": 1.972597459311014e-05,
      "loss": 0.3106,
      "step": 2020
    },
    {
      "epoch": 0.1529539756870536,
      "grad_norm": 0.828125,
      "learning_rate": 1.972569760331375e-05,
      "loss": 0.2806,
      "step": 2021
    },
    {
      "epoch": 0.15302965801050092,
      "grad_norm": 0.921875,
      "learning_rate": 1.9725420475541648e-05,
      "loss": 0.3461,
      "step": 2022
    },
    {
      "epoch": 0.15310534033394826,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9725143209797765e-05,
      "loss": 0.3266,
      "step": 2023
    },
    {
      "epoch": 0.15318102265739558,
      "grad_norm": 0.890625,
      "learning_rate": 1.972486580608604e-05,
      "loss": 0.3427,
      "step": 2024
    },
    {
      "epoch": 0.1532567049808429,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9724588264410402e-05,
      "loss": 0.2894,
      "step": 2025
    },
    {
      "epoch": 0.15333238730429025,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9724310584774797e-05,
      "loss": 0.3103,
      "step": 2026
    },
    {
      "epoch": 0.15340806962773756,
      "grad_norm": 3.765625,
      "learning_rate": 1.9724032767183152e-05,
      "loss": 0.5146,
      "step": 2027
    },
    {
      "epoch": 0.1534837519511849,
      "grad_norm": 1.84375,
      "learning_rate": 1.9723754811639418e-05,
      "loss": 0.5251,
      "step": 2028
    },
    {
      "epoch": 0.15355943427463223,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9723476718147536e-05,
      "loss": 0.3209,
      "step": 2029
    },
    {
      "epoch": 0.15363511659807957,
      "grad_norm": 0.96875,
      "learning_rate": 1.972319848671145e-05,
      "loss": 0.3652,
      "step": 2030
    },
    {
      "epoch": 0.1537107989215269,
      "grad_norm": 0.9375,
      "learning_rate": 1.9722920117335105e-05,
      "loss": 0.3824,
      "step": 2031
    },
    {
      "epoch": 0.1537864812449742,
      "grad_norm": 0.921875,
      "learning_rate": 1.9722641610022458e-05,
      "loss": 0.3531,
      "step": 2032
    },
    {
      "epoch": 0.15386216356842156,
      "grad_norm": 0.859375,
      "learning_rate": 1.9722362964777454e-05,
      "loss": 0.3131,
      "step": 2033
    },
    {
      "epoch": 0.15393784589186887,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9722084181604045e-05,
      "loss": 0.3497,
      "step": 2034
    },
    {
      "epoch": 0.15401352821531622,
      "grad_norm": 0.9140625,
      "learning_rate": 1.972180526050619e-05,
      "loss": 0.3318,
      "step": 2035
    },
    {
      "epoch": 0.15408921053876354,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9721526201487842e-05,
      "loss": 0.3525,
      "step": 2036
    },
    {
      "epoch": 0.15416489286221086,
      "grad_norm": 0.828125,
      "learning_rate": 1.9721247004552967e-05,
      "loss": 0.3262,
      "step": 2037
    },
    {
      "epoch": 0.1542405751856582,
      "grad_norm": 0.92578125,
      "learning_rate": 1.972096766970552e-05,
      "loss": 0.3561,
      "step": 2038
    },
    {
      "epoch": 0.15431625750910552,
      "grad_norm": 0.8203125,
      "learning_rate": 1.972068819694946e-05,
      "loss": 0.3091,
      "step": 2039
    },
    {
      "epoch": 0.15439193983255287,
      "grad_norm": 0.87890625,
      "learning_rate": 1.972040858628876e-05,
      "loss": 0.3231,
      "step": 2040
    },
    {
      "epoch": 0.15446762215600018,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9720128837727383e-05,
      "loss": 0.3595,
      "step": 2041
    },
    {
      "epoch": 0.15454330447944753,
      "grad_norm": 0.90625,
      "learning_rate": 1.9719848951269298e-05,
      "loss": 0.3436,
      "step": 2042
    },
    {
      "epoch": 0.15461898680289485,
      "grad_norm": 0.921875,
      "learning_rate": 1.9719568926918477e-05,
      "loss": 0.3661,
      "step": 2043
    },
    {
      "epoch": 0.15469466912634217,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9719288764678888e-05,
      "loss": 0.3507,
      "step": 2044
    },
    {
      "epoch": 0.1547703514497895,
      "grad_norm": 0.890625,
      "learning_rate": 1.9719008464554512e-05,
      "loss": 0.3199,
      "step": 2045
    },
    {
      "epoch": 0.15484603377323683,
      "grad_norm": 0.9765625,
      "learning_rate": 1.9718728026549324e-05,
      "loss": 0.3462,
      "step": 2046
    },
    {
      "epoch": 0.15492171609668418,
      "grad_norm": 0.90625,
      "learning_rate": 1.9718447450667298e-05,
      "loss": 0.3355,
      "step": 2047
    },
    {
      "epoch": 0.1549973984201315,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9718166736912417e-05,
      "loss": 0.3736,
      "step": 2048
    },
    {
      "epoch": 0.15507308074357884,
      "grad_norm": 0.84375,
      "learning_rate": 1.9717885885288664e-05,
      "loss": 0.3032,
      "step": 2049
    },
    {
      "epoch": 0.15514876306702616,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9717604895800026e-05,
      "loss": 0.3164,
      "step": 2050
    },
    {
      "epoch": 0.15522444539047348,
      "grad_norm": 0.875,
      "learning_rate": 1.9717323768450483e-05,
      "loss": 0.3136,
      "step": 2051
    },
    {
      "epoch": 0.15530012771392082,
      "grad_norm": 0.8125,
      "learning_rate": 1.9717042503244032e-05,
      "loss": 0.3072,
      "step": 2052
    },
    {
      "epoch": 0.15537581003736814,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9716761100184652e-05,
      "loss": 0.3419,
      "step": 2053
    },
    {
      "epoch": 0.15545149236081549,
      "grad_norm": 0.90625,
      "learning_rate": 1.9716479559276345e-05,
      "loss": 0.352,
      "step": 2054
    },
    {
      "epoch": 0.1555271746842628,
      "grad_norm": 0.8359375,
      "learning_rate": 1.97161978805231e-05,
      "loss": 0.3414,
      "step": 2055
    },
    {
      "epoch": 0.15560285700771015,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9715916063928916e-05,
      "loss": 0.2886,
      "step": 2056
    },
    {
      "epoch": 0.15567853933115747,
      "grad_norm": 1.0390625,
      "learning_rate": 1.971563410949779e-05,
      "loss": 0.3705,
      "step": 2057
    },
    {
      "epoch": 0.15575422165460479,
      "grad_norm": 0.93359375,
      "learning_rate": 1.971535201723372e-05,
      "loss": 0.3274,
      "step": 2058
    },
    {
      "epoch": 0.15582990397805213,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9715069787140707e-05,
      "loss": 0.3316,
      "step": 2059
    },
    {
      "epoch": 0.15590558630149945,
      "grad_norm": 0.87890625,
      "learning_rate": 1.971478741922276e-05,
      "loss": 0.335,
      "step": 2060
    },
    {
      "epoch": 0.1559812686249468,
      "grad_norm": 0.8671875,
      "learning_rate": 1.971450491348388e-05,
      "loss": 0.3582,
      "step": 2061
    },
    {
      "epoch": 0.1560569509483941,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9714222269928084e-05,
      "loss": 0.3181,
      "step": 2062
    },
    {
      "epoch": 0.15613263327184146,
      "grad_norm": 1.0,
      "learning_rate": 1.971393948855937e-05,
      "loss": 0.3776,
      "step": 2063
    },
    {
      "epoch": 0.15620831559528878,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9713656569381755e-05,
      "loss": 0.3801,
      "step": 2064
    },
    {
      "epoch": 0.1562839979187361,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9713373512399252e-05,
      "loss": 0.2995,
      "step": 2065
    },
    {
      "epoch": 0.15635968024218344,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9713090317615877e-05,
      "loss": 0.3536,
      "step": 2066
    },
    {
      "epoch": 0.15643536256563076,
      "grad_norm": 0.87890625,
      "learning_rate": 1.971280698503565e-05,
      "loss": 0.3529,
      "step": 2067
    },
    {
      "epoch": 0.1565110448890781,
      "grad_norm": 4.1875,
      "learning_rate": 1.9712523514662583e-05,
      "loss": 0.5044,
      "step": 2068
    },
    {
      "epoch": 0.15658672721252542,
      "grad_norm": 0.9609375,
      "learning_rate": 1.971223990650071e-05,
      "loss": 0.3804,
      "step": 2069
    },
    {
      "epoch": 0.15666240953597274,
      "grad_norm": 0.87890625,
      "learning_rate": 1.971195616055404e-05,
      "loss": 0.3396,
      "step": 2070
    },
    {
      "epoch": 0.1567380918594201,
      "grad_norm": 0.89453125,
      "learning_rate": 1.971167227682661e-05,
      "loss": 0.3258,
      "step": 2071
    },
    {
      "epoch": 0.1568137741828674,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9711388255322442e-05,
      "loss": 0.3248,
      "step": 2072
    },
    {
      "epoch": 0.15688945650631475,
      "grad_norm": 2.265625,
      "learning_rate": 1.9711104096045567e-05,
      "loss": 0.4482,
      "step": 2073
    },
    {
      "epoch": 0.15696513882976207,
      "grad_norm": 0.90625,
      "learning_rate": 1.9710819799000012e-05,
      "loss": 0.3394,
      "step": 2074
    },
    {
      "epoch": 0.15704082115320941,
      "grad_norm": 0.859375,
      "learning_rate": 1.9710535364189818e-05,
      "loss": 0.3492,
      "step": 2075
    },
    {
      "epoch": 0.15711650347665673,
      "grad_norm": 0.9453125,
      "learning_rate": 1.971025079161901e-05,
      "loss": 0.3742,
      "step": 2076
    },
    {
      "epoch": 0.15719218580010405,
      "grad_norm": 0.859375,
      "learning_rate": 1.970996608129164e-05,
      "loss": 0.3264,
      "step": 2077
    },
    {
      "epoch": 0.1572678681235514,
      "grad_norm": 0.9296875,
      "learning_rate": 1.970968123321173e-05,
      "loss": 0.3686,
      "step": 2078
    },
    {
      "epoch": 0.15734355044699871,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9709396247383335e-05,
      "loss": 0.3485,
      "step": 2079
    },
    {
      "epoch": 0.15741923277044606,
      "grad_norm": 0.85546875,
      "learning_rate": 1.970911112381049e-05,
      "loss": 0.3449,
      "step": 2080
    },
    {
      "epoch": 0.15749491509389338,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9708825862497243e-05,
      "loss": 0.3096,
      "step": 2081
    },
    {
      "epoch": 0.15757059741734072,
      "grad_norm": 0.94140625,
      "learning_rate": 1.970854046344764e-05,
      "loss": 0.385,
      "step": 2082
    },
    {
      "epoch": 0.15764627974078804,
      "grad_norm": 0.8515625,
      "learning_rate": 1.970825492666573e-05,
      "loss": 0.3313,
      "step": 2083
    },
    {
      "epoch": 0.15772196206423536,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9707969252155563e-05,
      "loss": 0.3448,
      "step": 2084
    },
    {
      "epoch": 0.1577976443876827,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9707683439921196e-05,
      "loss": 0.2934,
      "step": 2085
    },
    {
      "epoch": 0.15787332671113002,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9707397489966675e-05,
      "loss": 0.3799,
      "step": 2086
    },
    {
      "epoch": 0.15794900903457737,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9707111402296065e-05,
      "loss": 0.3742,
      "step": 2087
    },
    {
      "epoch": 0.1580246913580247,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9706825176913423e-05,
      "loss": 0.3609,
      "step": 2088
    },
    {
      "epoch": 0.15810037368147203,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9706538813822807e-05,
      "loss": 0.3756,
      "step": 2089
    },
    {
      "epoch": 0.15817605600491935,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9706252313028283e-05,
      "loss": 0.3015,
      "step": 2090
    },
    {
      "epoch": 0.15825173832836667,
      "grad_norm": 0.88671875,
      "learning_rate": 1.970596567453391e-05,
      "loss": 0.3417,
      "step": 2091
    },
    {
      "epoch": 0.15832742065181402,
      "grad_norm": 0.84765625,
      "learning_rate": 1.970567889834376e-05,
      "loss": 0.3408,
      "step": 2092
    },
    {
      "epoch": 0.15840310297526133,
      "grad_norm": 0.91796875,
      "learning_rate": 1.97053919844619e-05,
      "loss": 0.2891,
      "step": 2093
    },
    {
      "epoch": 0.15847878529870868,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9705104932892398e-05,
      "loss": 0.3897,
      "step": 2094
    },
    {
      "epoch": 0.158554467622156,
      "grad_norm": 2.515625,
      "learning_rate": 1.970481774363933e-05,
      "loss": 0.4627,
      "step": 2095
    },
    {
      "epoch": 0.15863014994560334,
      "grad_norm": 0.9765625,
      "learning_rate": 1.9704530416706765e-05,
      "loss": 0.4061,
      "step": 2096
    },
    {
      "epoch": 0.15870583226905066,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9704242952098784e-05,
      "loss": 0.314,
      "step": 2097
    },
    {
      "epoch": 0.15878151459249798,
      "grad_norm": 0.796875,
      "learning_rate": 1.9703955349819464e-05,
      "loss": 0.2951,
      "step": 2098
    },
    {
      "epoch": 0.15885719691594533,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9703667609872886e-05,
      "loss": 0.3275,
      "step": 2099
    },
    {
      "epoch": 0.15893287923939264,
      "grad_norm": 0.86328125,
      "learning_rate": 1.970337973226313e-05,
      "loss": 0.3656,
      "step": 2100
    },
    {
      "epoch": 0.15900856156284,
      "grad_norm": 0.92578125,
      "learning_rate": 1.970309171699428e-05,
      "loss": 0.4019,
      "step": 2101
    },
    {
      "epoch": 0.1590842438862873,
      "grad_norm": 0.94921875,
      "learning_rate": 1.9702803564070426e-05,
      "loss": 0.3991,
      "step": 2102
    },
    {
      "epoch": 0.15915992620973463,
      "grad_norm": 0.7890625,
      "learning_rate": 1.9702515273495647e-05,
      "loss": 0.2913,
      "step": 2103
    },
    {
      "epoch": 0.15923560853318197,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9702226845274043e-05,
      "loss": 0.3512,
      "step": 2104
    },
    {
      "epoch": 0.1593112908566293,
      "grad_norm": 0.98828125,
      "learning_rate": 1.9701938279409704e-05,
      "loss": 0.4298,
      "step": 2105
    },
    {
      "epoch": 0.15938697318007664,
      "grad_norm": 1.3515625,
      "learning_rate": 1.9701649575906715e-05,
      "loss": 0.3121,
      "step": 2106
    },
    {
      "epoch": 0.15946265550352395,
      "grad_norm": 0.9140625,
      "learning_rate": 1.970136073476918e-05,
      "loss": 0.3635,
      "step": 2107
    },
    {
      "epoch": 0.1595383378269713,
      "grad_norm": 0.95703125,
      "learning_rate": 1.97010717560012e-05,
      "loss": 0.3814,
      "step": 2108
    },
    {
      "epoch": 0.15961402015041862,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9700782639606864e-05,
      "loss": 0.3848,
      "step": 2109
    },
    {
      "epoch": 0.15968970247386594,
      "grad_norm": 2.109375,
      "learning_rate": 1.9700493385590282e-05,
      "loss": 0.4262,
      "step": 2110
    },
    {
      "epoch": 0.15976538479731328,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9700203993955555e-05,
      "loss": 0.3629,
      "step": 2111
    },
    {
      "epoch": 0.1598410671207606,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9699914464706784e-05,
      "loss": 0.3528,
      "step": 2112
    },
    {
      "epoch": 0.15991674944420795,
      "grad_norm": 0.8984375,
      "learning_rate": 1.969962479784808e-05,
      "loss": 0.3398,
      "step": 2113
    },
    {
      "epoch": 0.15999243176765526,
      "grad_norm": 1.7109375,
      "learning_rate": 1.969933499338356e-05,
      "loss": 0.4649,
      "step": 2114
    },
    {
      "epoch": 0.1600681140911026,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9699045051317327e-05,
      "loss": 0.3114,
      "step": 2115
    },
    {
      "epoch": 0.16014379641454993,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9698754971653492e-05,
      "loss": 0.2946,
      "step": 2116
    },
    {
      "epoch": 0.16021947873799725,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9698464754396175e-05,
      "loss": 0.3842,
      "step": 2117
    },
    {
      "epoch": 0.1602951610614446,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9698174399549493e-05,
      "loss": 0.3163,
      "step": 2118
    },
    {
      "epoch": 0.1603708433848919,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9697883907117568e-05,
      "loss": 0.3354,
      "step": 2119
    },
    {
      "epoch": 0.16044652570833925,
      "grad_norm": 0.8984375,
      "learning_rate": 1.969759327710451e-05,
      "loss": 0.3655,
      "step": 2120
    },
    {
      "epoch": 0.16052220803178657,
      "grad_norm": 0.9140625,
      "learning_rate": 1.969730250951446e-05,
      "loss": 0.3418,
      "step": 2121
    },
    {
      "epoch": 0.16059789035523392,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9697011604351524e-05,
      "loss": 0.3543,
      "step": 2122
    },
    {
      "epoch": 0.16067357267868124,
      "grad_norm": 0.92578125,
      "learning_rate": 1.969672056161984e-05,
      "loss": 0.3754,
      "step": 2123
    },
    {
      "epoch": 0.16074925500212855,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9696429381323536e-05,
      "loss": 0.3747,
      "step": 2124
    },
    {
      "epoch": 0.1608249373255759,
      "grad_norm": 0.96875,
      "learning_rate": 1.9696138063466742e-05,
      "loss": 0.3859,
      "step": 2125
    },
    {
      "epoch": 0.16090061964902322,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9695846608053587e-05,
      "loss": 0.4071,
      "step": 2126
    },
    {
      "epoch": 0.16097630197247056,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9695555015088213e-05,
      "loss": 0.3638,
      "step": 2127
    },
    {
      "epoch": 0.16105198429591788,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9695263284574752e-05,
      "loss": 0.3198,
      "step": 2128
    },
    {
      "epoch": 0.1611276666193652,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9694971416517343e-05,
      "loss": 0.3266,
      "step": 2129
    },
    {
      "epoch": 0.16120334894281255,
      "grad_norm": 0.953125,
      "learning_rate": 1.9694679410920127e-05,
      "loss": 0.396,
      "step": 2130
    },
    {
      "epoch": 0.16127903126625986,
      "grad_norm": 0.96875,
      "learning_rate": 1.969438726778725e-05,
      "loss": 0.3907,
      "step": 2131
    },
    {
      "epoch": 0.1613547135897072,
      "grad_norm": 0.8046875,
      "learning_rate": 1.969409498712285e-05,
      "loss": 0.2916,
      "step": 2132
    },
    {
      "epoch": 0.16143039591315453,
      "grad_norm": 0.8671875,
      "learning_rate": 1.969380256893108e-05,
      "loss": 0.3403,
      "step": 2133
    },
    {
      "epoch": 0.16150607823660187,
      "grad_norm": 0.76171875,
      "learning_rate": 1.969351001321608e-05,
      "loss": 0.2686,
      "step": 2134
    },
    {
      "epoch": 0.1615817605600492,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9693217319982013e-05,
      "loss": 0.3249,
      "step": 2135
    },
    {
      "epoch": 0.1616574428834965,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9692924489233015e-05,
      "loss": 0.401,
      "step": 2136
    },
    {
      "epoch": 0.16173312520694386,
      "grad_norm": 0.8671875,
      "learning_rate": 1.969263152097326e-05,
      "loss": 0.3251,
      "step": 2137
    },
    {
      "epoch": 0.16180880753039117,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9692338415206886e-05,
      "loss": 0.3297,
      "step": 2138
    },
    {
      "epoch": 0.16188448985383852,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9692045171938063e-05,
      "loss": 0.3234,
      "step": 2139
    },
    {
      "epoch": 0.16196017217728584,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9691751791170944e-05,
      "loss": 0.3907,
      "step": 2140
    },
    {
      "epoch": 0.16203585450073318,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9691458272909697e-05,
      "loss": 0.3121,
      "step": 2141
    },
    {
      "epoch": 0.1621115368241805,
      "grad_norm": 0.84375,
      "learning_rate": 1.9691164617158482e-05,
      "loss": 0.2817,
      "step": 2142
    },
    {
      "epoch": 0.16218721914762782,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9690870823921466e-05,
      "loss": 0.2948,
      "step": 2143
    },
    {
      "epoch": 0.16226290147107517,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9690576893202818e-05,
      "loss": 0.3499,
      "step": 2144
    },
    {
      "epoch": 0.16233858379452248,
      "grad_norm": 0.921875,
      "learning_rate": 1.9690282825006707e-05,
      "loss": 0.3909,
      "step": 2145
    },
    {
      "epoch": 0.16241426611796983,
      "grad_norm": 0.828125,
      "learning_rate": 1.9689988619337304e-05,
      "loss": 0.3231,
      "step": 2146
    },
    {
      "epoch": 0.16248994844141715,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9689694276198785e-05,
      "loss": 0.283,
      "step": 2147
    },
    {
      "epoch": 0.1625656307648645,
      "grad_norm": 2.328125,
      "learning_rate": 1.9689399795595325e-05,
      "loss": 0.4399,
      "step": 2148
    },
    {
      "epoch": 0.1626413130883118,
      "grad_norm": 0.84375,
      "learning_rate": 1.96891051775311e-05,
      "loss": 0.3319,
      "step": 2149
    },
    {
      "epoch": 0.16271699541175913,
      "grad_norm": 0.890625,
      "learning_rate": 1.968881042201029e-05,
      "loss": 0.3063,
      "step": 2150
    },
    {
      "epoch": 0.16279267773520648,
      "grad_norm": 0.8828125,
      "learning_rate": 1.968851552903708e-05,
      "loss": 0.3579,
      "step": 2151
    },
    {
      "epoch": 0.1628683600586538,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9688220498615652e-05,
      "loss": 0.3369,
      "step": 2152
    },
    {
      "epoch": 0.16294404238210114,
      "grad_norm": 0.83203125,
      "learning_rate": 1.968792533075019e-05,
      "loss": 0.3284,
      "step": 2153
    },
    {
      "epoch": 0.16301972470554846,
      "grad_norm": 0.80859375,
      "learning_rate": 1.968763002544488e-05,
      "loss": 0.328,
      "step": 2154
    },
    {
      "epoch": 0.1630954070289958,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9687334582703916e-05,
      "loss": 0.3527,
      "step": 2155
    },
    {
      "epoch": 0.16317108935244312,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9687039002531486e-05,
      "loss": 0.3255,
      "step": 2156
    },
    {
      "epoch": 0.16324677167589044,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9686743284931783e-05,
      "loss": 0.3231,
      "step": 2157
    },
    {
      "epoch": 0.16332245399933779,
      "grad_norm": 0.94921875,
      "learning_rate": 1.9686447429909005e-05,
      "loss": 0.3713,
      "step": 2158
    },
    {
      "epoch": 0.1633981363227851,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9686151437467347e-05,
      "loss": 0.3514,
      "step": 2159
    },
    {
      "epoch": 0.16347381864623245,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9685855307611007e-05,
      "loss": 0.3551,
      "step": 2160
    },
    {
      "epoch": 0.16354950096967977,
      "grad_norm": 0.83203125,
      "learning_rate": 1.968555904034419e-05,
      "loss": 0.3137,
      "step": 2161
    },
    {
      "epoch": 0.16362518329312709,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9685262635671097e-05,
      "loss": 0.3508,
      "step": 2162
    },
    {
      "epoch": 0.16370086561657443,
      "grad_norm": 0.8125,
      "learning_rate": 1.9684966093595934e-05,
      "loss": 0.3308,
      "step": 2163
    },
    {
      "epoch": 0.16377654794002175,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9684669414122905e-05,
      "loss": 0.3485,
      "step": 2164
    },
    {
      "epoch": 0.1638522302634691,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9684372597256224e-05,
      "loss": 0.3377,
      "step": 2165
    },
    {
      "epoch": 0.1639279125869164,
      "grad_norm": 2.109375,
      "learning_rate": 1.9684075643000097e-05,
      "loss": 0.4312,
      "step": 2166
    },
    {
      "epoch": 0.16400359491036376,
      "grad_norm": 0.890625,
      "learning_rate": 1.9683778551358736e-05,
      "loss": 0.3182,
      "step": 2167
    },
    {
      "epoch": 0.16407927723381108,
      "grad_norm": 0.96484375,
      "learning_rate": 1.968348132233636e-05,
      "loss": 0.3637,
      "step": 2168
    },
    {
      "epoch": 0.1641549595572584,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9683183955937185e-05,
      "loss": 0.3173,
      "step": 2169
    },
    {
      "epoch": 0.16423064188070574,
      "grad_norm": 1.0234375,
      "learning_rate": 1.9682886452165428e-05,
      "loss": 0.4374,
      "step": 2170
    },
    {
      "epoch": 0.16430632420415306,
      "grad_norm": 0.921875,
      "learning_rate": 1.9682588811025313e-05,
      "loss": 0.363,
      "step": 2171
    },
    {
      "epoch": 0.1643820065276004,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9682291032521056e-05,
      "loss": 0.3533,
      "step": 2172
    },
    {
      "epoch": 0.16445768885104772,
      "grad_norm": 0.90625,
      "learning_rate": 1.9681993116656884e-05,
      "loss": 0.3799,
      "step": 2173
    },
    {
      "epoch": 0.16453337117449507,
      "grad_norm": 0.890625,
      "learning_rate": 1.968169506343703e-05,
      "loss": 0.3743,
      "step": 2174
    },
    {
      "epoch": 0.1646090534979424,
      "grad_norm": 0.85546875,
      "learning_rate": 1.968139687286571e-05,
      "loss": 0.3267,
      "step": 2175
    },
    {
      "epoch": 0.1646847358213897,
      "grad_norm": 0.9375,
      "learning_rate": 1.968109854494717e-05,
      "loss": 0.3447,
      "step": 2176
    },
    {
      "epoch": 0.16476041814483705,
      "grad_norm": 0.890625,
      "learning_rate": 1.968080007968563e-05,
      "loss": 0.3561,
      "step": 2177
    },
    {
      "epoch": 0.16483610046828437,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9680501477085326e-05,
      "loss": 0.361,
      "step": 2178
    },
    {
      "epoch": 0.16491178279173171,
      "grad_norm": 0.87890625,
      "learning_rate": 1.96802027371505e-05,
      "loss": 0.3569,
      "step": 2179
    },
    {
      "epoch": 0.16498746511517903,
      "grad_norm": 1.203125,
      "learning_rate": 1.9679903859885384e-05,
      "loss": 0.358,
      "step": 2180
    },
    {
      "epoch": 0.16506314743862638,
      "grad_norm": 0.9375,
      "learning_rate": 1.9679604845294224e-05,
      "loss": 0.3651,
      "step": 2181
    },
    {
      "epoch": 0.1651388297620737,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9679305693381252e-05,
      "loss": 0.3645,
      "step": 2182
    },
    {
      "epoch": 0.16521451208552101,
      "grad_norm": 0.8828125,
      "learning_rate": 1.967900640415072e-05,
      "loss": 0.3285,
      "step": 2183
    },
    {
      "epoch": 0.16529019440896836,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9678706977606876e-05,
      "loss": 0.2957,
      "step": 2184
    },
    {
      "epoch": 0.16536587673241568,
      "grad_norm": 0.98828125,
      "learning_rate": 1.9678407413753965e-05,
      "loss": 0.3269,
      "step": 2185
    },
    {
      "epoch": 0.16544155905586302,
      "grad_norm": 0.87890625,
      "learning_rate": 1.967810771259623e-05,
      "loss": 0.3271,
      "step": 2186
    },
    {
      "epoch": 0.16551724137931034,
      "grad_norm": 0.90625,
      "learning_rate": 1.9677807874137933e-05,
      "loss": 0.3546,
      "step": 2187
    },
    {
      "epoch": 0.1655929237027577,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9677507898383322e-05,
      "loss": 0.3259,
      "step": 2188
    },
    {
      "epoch": 0.165668606026205,
      "grad_norm": 0.828125,
      "learning_rate": 1.9677207785336657e-05,
      "loss": 0.3093,
      "step": 2189
    },
    {
      "epoch": 0.16574428834965232,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9676907535002195e-05,
      "loss": 0.3399,
      "step": 2190
    },
    {
      "epoch": 0.16581997067309967,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9676607147384187e-05,
      "loss": 0.3761,
      "step": 2191
    },
    {
      "epoch": 0.165895652996547,
      "grad_norm": 0.78515625,
      "learning_rate": 1.9676306622486905e-05,
      "loss": 0.2711,
      "step": 2192
    },
    {
      "epoch": 0.16597133531999433,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9676005960314608e-05,
      "loss": 0.3726,
      "step": 2193
    },
    {
      "epoch": 0.16604701764344165,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9675705160871563e-05,
      "loss": 0.3266,
      "step": 2194
    },
    {
      "epoch": 0.16612269996688897,
      "grad_norm": 0.89453125,
      "learning_rate": 1.967540422416203e-05,
      "loss": 0.3343,
      "step": 2195
    },
    {
      "epoch": 0.16619838229033632,
      "grad_norm": 0.86328125,
      "learning_rate": 1.967510315019029e-05,
      "loss": 0.3407,
      "step": 2196
    },
    {
      "epoch": 0.16627406461378363,
      "grad_norm": 0.8984375,
      "learning_rate": 1.967480193896061e-05,
      "loss": 0.3641,
      "step": 2197
    },
    {
      "epoch": 0.16634974693723098,
      "grad_norm": 0.796875,
      "learning_rate": 1.9674500590477258e-05,
      "loss": 0.3034,
      "step": 2198
    },
    {
      "epoch": 0.1664254292606783,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9674199104744517e-05,
      "loss": 0.3595,
      "step": 2199
    },
    {
      "epoch": 0.16650111158412564,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9673897481766655e-05,
      "loss": 0.3777,
      "step": 2200
    },
    {
      "epoch": 0.16657679390757296,
      "grad_norm": 0.81640625,
      "learning_rate": 1.967359572154796e-05,
      "loss": 0.3149,
      "step": 2201
    },
    {
      "epoch": 0.16665247623102028,
      "grad_norm": 0.8125,
      "learning_rate": 1.967329382409271e-05,
      "loss": 0.3206,
      "step": 2202
    },
    {
      "epoch": 0.16672815855446763,
      "grad_norm": 2.453125,
      "learning_rate": 1.967299178940518e-05,
      "loss": 0.5164,
      "step": 2203
    },
    {
      "epoch": 0.16680384087791494,
      "grad_norm": 0.8515625,
      "learning_rate": 1.967268961748967e-05,
      "loss": 0.2984,
      "step": 2204
    },
    {
      "epoch": 0.1668795232013623,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9672387308350458e-05,
      "loss": 0.339,
      "step": 2205
    },
    {
      "epoch": 0.1669552055248096,
      "grad_norm": 0.8984375,
      "learning_rate": 1.967208486199183e-05,
      "loss": 0.3508,
      "step": 2206
    },
    {
      "epoch": 0.16703088784825695,
      "grad_norm": 0.90625,
      "learning_rate": 1.9671782278418084e-05,
      "loss": 0.3612,
      "step": 2207
    },
    {
      "epoch": 0.16710657017170427,
      "grad_norm": 0.828125,
      "learning_rate": 1.9671479557633508e-05,
      "loss": 0.3441,
      "step": 2208
    },
    {
      "epoch": 0.1671822524951516,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9671176699642397e-05,
      "loss": 0.3769,
      "step": 2209
    },
    {
      "epoch": 0.16725793481859894,
      "grad_norm": 1.5234375,
      "learning_rate": 1.967087370444905e-05,
      "loss": 0.4396,
      "step": 2210
    },
    {
      "epoch": 0.16733361714204625,
      "grad_norm": 0.8046875,
      "learning_rate": 1.967057057205776e-05,
      "loss": 0.2914,
      "step": 2211
    },
    {
      "epoch": 0.1674092994654936,
      "grad_norm": 0.921875,
      "learning_rate": 1.9670267302472833e-05,
      "loss": 0.3563,
      "step": 2212
    },
    {
      "epoch": 0.16748498178894092,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9669963895698573e-05,
      "loss": 0.3278,
      "step": 2213
    },
    {
      "epoch": 0.16756066411238826,
      "grad_norm": 0.84375,
      "learning_rate": 1.9669660351739277e-05,
      "loss": 0.3208,
      "step": 2214
    },
    {
      "epoch": 0.16763634643583558,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9669356670599255e-05,
      "loss": 0.3031,
      "step": 2215
    },
    {
      "epoch": 0.1677120287592829,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9669052852282817e-05,
      "loss": 0.3248,
      "step": 2216
    },
    {
      "epoch": 0.16778771108273025,
      "grad_norm": 1.0234375,
      "learning_rate": 1.966874889679427e-05,
      "loss": 0.3596,
      "step": 2217
    },
    {
      "epoch": 0.16786339340617756,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9668444804137928e-05,
      "loss": 0.3752,
      "step": 2218
    },
    {
      "epoch": 0.1679390757296249,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9668140574318108e-05,
      "loss": 0.2902,
      "step": 2219
    },
    {
      "epoch": 0.16801475805307223,
      "grad_norm": 0.81640625,
      "learning_rate": 1.966783620733912e-05,
      "loss": 0.2935,
      "step": 2220
    },
    {
      "epoch": 0.16809044037651955,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9667531703205285e-05,
      "loss": 0.2864,
      "step": 2221
    },
    {
      "epoch": 0.1681661226999669,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9667227061920925e-05,
      "loss": 0.3373,
      "step": 2222
    },
    {
      "epoch": 0.1682418050234142,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9666922283490353e-05,
      "loss": 0.3088,
      "step": 2223
    },
    {
      "epoch": 0.16831748734686156,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9666617367917907e-05,
      "loss": 0.3631,
      "step": 2224
    },
    {
      "epoch": 0.16839316967030887,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9666312315207903e-05,
      "loss": 0.3077,
      "step": 2225
    },
    {
      "epoch": 0.16846885199375622,
      "grad_norm": 0.85546875,
      "learning_rate": 1.966600712536467e-05,
      "loss": 0.3433,
      "step": 2226
    },
    {
      "epoch": 0.16854453431720354,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9665701798392537e-05,
      "loss": 0.3028,
      "step": 2227
    },
    {
      "epoch": 0.16862021664065086,
      "grad_norm": 0.88671875,
      "learning_rate": 1.966539633429584e-05,
      "loss": 0.3663,
      "step": 2228
    },
    {
      "epoch": 0.1686958989640982,
      "grad_norm": 0.890625,
      "learning_rate": 1.9665090733078906e-05,
      "loss": 0.3719,
      "step": 2229
    },
    {
      "epoch": 0.16877158128754552,
      "grad_norm": 0.890625,
      "learning_rate": 1.9664784994746076e-05,
      "loss": 0.327,
      "step": 2230
    },
    {
      "epoch": 0.16884726361099286,
      "grad_norm": 0.9609375,
      "learning_rate": 1.9664479119301687e-05,
      "loss": 0.376,
      "step": 2231
    },
    {
      "epoch": 0.16892294593444018,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9664173106750075e-05,
      "loss": 0.3155,
      "step": 2232
    },
    {
      "epoch": 0.16899862825788753,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9663866957095585e-05,
      "loss": 0.3751,
      "step": 2233
    },
    {
      "epoch": 0.16907431058133485,
      "grad_norm": 0.88671875,
      "learning_rate": 1.966356067034256e-05,
      "loss": 0.3507,
      "step": 2234
    },
    {
      "epoch": 0.16914999290478216,
      "grad_norm": 0.87890625,
      "learning_rate": 1.966325424649534e-05,
      "loss": 0.3214,
      "step": 2235
    },
    {
      "epoch": 0.1692256752282295,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9662947685558275e-05,
      "loss": 0.3455,
      "step": 2236
    },
    {
      "epoch": 0.16930135755167683,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9662640987535716e-05,
      "loss": 0.3714,
      "step": 2237
    },
    {
      "epoch": 0.16937703987512417,
      "grad_norm": 0.98828125,
      "learning_rate": 1.9662334152432015e-05,
      "loss": 0.4089,
      "step": 2238
    },
    {
      "epoch": 0.1694527221985715,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9662027180251518e-05,
      "loss": 0.2972,
      "step": 2239
    },
    {
      "epoch": 0.16952840452201884,
      "grad_norm": 0.8828125,
      "learning_rate": 1.966172007099859e-05,
      "loss": 0.3445,
      "step": 2240
    },
    {
      "epoch": 0.16960408684546616,
      "grad_norm": 0.84375,
      "learning_rate": 1.966141282467758e-05,
      "loss": 0.3463,
      "step": 2241
    },
    {
      "epoch": 0.16967976916891347,
      "grad_norm": 0.875,
      "learning_rate": 1.9661105441292853e-05,
      "loss": 0.3421,
      "step": 2242
    },
    {
      "epoch": 0.16975545149236082,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9660797920848763e-05,
      "loss": 0.3156,
      "step": 2243
    },
    {
      "epoch": 0.16983113381580814,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9660490263349677e-05,
      "loss": 0.3229,
      "step": 2244
    },
    {
      "epoch": 0.16990681613925548,
      "grad_norm": 0.83203125,
      "learning_rate": 1.966018246879996e-05,
      "loss": 0.3456,
      "step": 2245
    },
    {
      "epoch": 0.1699824984627028,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9659874537203976e-05,
      "loss": 0.3163,
      "step": 2246
    },
    {
      "epoch": 0.17005818078615015,
      "grad_norm": 0.84765625,
      "learning_rate": 1.965956646856609e-05,
      "loss": 0.3503,
      "step": 2247
    },
    {
      "epoch": 0.17013386310959747,
      "grad_norm": 0.94921875,
      "learning_rate": 1.9659258262890683e-05,
      "loss": 0.4087,
      "step": 2248
    },
    {
      "epoch": 0.17020954543304478,
      "grad_norm": 2.15625,
      "learning_rate": 1.9658949920182123e-05,
      "loss": 0.4008,
      "step": 2249
    },
    {
      "epoch": 0.17028522775649213,
      "grad_norm": 0.97265625,
      "learning_rate": 1.965864144044478e-05,
      "loss": 0.3971,
      "step": 2250
    },
    {
      "epoch": 0.17036091007993945,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9658332823683033e-05,
      "loss": 0.3744,
      "step": 2251
    },
    {
      "epoch": 0.1704365924033868,
      "grad_norm": 0.90625,
      "learning_rate": 1.965802406990126e-05,
      "loss": 0.3577,
      "step": 2252
    },
    {
      "epoch": 0.1705122747268341,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9657715179103847e-05,
      "loss": 0.2984,
      "step": 2253
    },
    {
      "epoch": 0.17058795705028143,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9657406151295167e-05,
      "loss": 0.3077,
      "step": 2254
    },
    {
      "epoch": 0.17066363937372878,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9657096986479606e-05,
      "loss": 0.311,
      "step": 2255
    },
    {
      "epoch": 0.1707393216971761,
      "grad_norm": 0.921875,
      "learning_rate": 1.9656787684661555e-05,
      "loss": 0.3619,
      "step": 2256
    },
    {
      "epoch": 0.17081500402062344,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9656478245845398e-05,
      "loss": 0.3855,
      "step": 2257
    },
    {
      "epoch": 0.17089068634407076,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9656168670035525e-05,
      "loss": 0.3511,
      "step": 2258
    },
    {
      "epoch": 0.1709663686675181,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9655858957236333e-05,
      "loss": 0.3499,
      "step": 2259
    },
    {
      "epoch": 0.17104205099096542,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9655549107452207e-05,
      "loss": 0.3377,
      "step": 2260
    },
    {
      "epoch": 0.17111773331441274,
      "grad_norm": 0.82421875,
      "learning_rate": 1.965523912068755e-05,
      "loss": 0.3218,
      "step": 2261
    },
    {
      "epoch": 0.17119341563786009,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9654928996946758e-05,
      "loss": 0.3624,
      "step": 2262
    },
    {
      "epoch": 0.1712690979613074,
      "grad_norm": 0.8046875,
      "learning_rate": 1.965461873623423e-05,
      "loss": 0.3111,
      "step": 2263
    },
    {
      "epoch": 0.17134478028475475,
      "grad_norm": 1.15625,
      "learning_rate": 1.9654308338554364e-05,
      "loss": 0.3711,
      "step": 2264
    },
    {
      "epoch": 0.17142046260820207,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9653997803911568e-05,
      "loss": 0.3262,
      "step": 2265
    },
    {
      "epoch": 0.1714961449316494,
      "grad_norm": 0.8203125,
      "learning_rate": 1.965368713231025e-05,
      "loss": 0.3047,
      "step": 2266
    },
    {
      "epoch": 0.17157182725509673,
      "grad_norm": 0.75390625,
      "learning_rate": 1.965337632375481e-05,
      "loss": 0.2777,
      "step": 2267
    },
    {
      "epoch": 0.17164750957854405,
      "grad_norm": 0.82421875,
      "learning_rate": 1.965306537824966e-05,
      "loss": 0.3151,
      "step": 2268
    },
    {
      "epoch": 0.1717231919019914,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9652754295799213e-05,
      "loss": 0.3432,
      "step": 2269
    },
    {
      "epoch": 0.1717988742254387,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9652443076407884e-05,
      "loss": 0.3463,
      "step": 2270
    },
    {
      "epoch": 0.17187455654888606,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9652131720080082e-05,
      "loss": 0.3581,
      "step": 2271
    },
    {
      "epoch": 0.17195023887233338,
      "grad_norm": 0.8515625,
      "learning_rate": 1.965182022682023e-05,
      "loss": 0.2936,
      "step": 2272
    },
    {
      "epoch": 0.17202592119578072,
      "grad_norm": 0.8203125,
      "learning_rate": 1.965150859663275e-05,
      "loss": 0.3099,
      "step": 2273
    },
    {
      "epoch": 0.17210160351922804,
      "grad_norm": 0.99609375,
      "learning_rate": 1.9651196829522054e-05,
      "loss": 0.4214,
      "step": 2274
    },
    {
      "epoch": 0.17217728584267536,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9650884925492567e-05,
      "loss": 0.3404,
      "step": 2275
    },
    {
      "epoch": 0.1722529681661227,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9650572884548715e-05,
      "loss": 0.3453,
      "step": 2276
    },
    {
      "epoch": 0.17232865048957002,
      "grad_norm": 2.453125,
      "learning_rate": 1.965026070669493e-05,
      "loss": 0.4218,
      "step": 2277
    },
    {
      "epoch": 0.17240433281301737,
      "grad_norm": 0.87890625,
      "learning_rate": 1.964994839193563e-05,
      "loss": 0.3505,
      "step": 2278
    },
    {
      "epoch": 0.1724800151364647,
      "grad_norm": 0.7890625,
      "learning_rate": 1.964963594027526e-05,
      "loss": 0.2973,
      "step": 2279
    },
    {
      "epoch": 0.17255569745991203,
      "grad_norm": 0.921875,
      "learning_rate": 1.9649323351718242e-05,
      "loss": 0.3722,
      "step": 2280
    },
    {
      "epoch": 0.17263137978335935,
      "grad_norm": 0.9140625,
      "learning_rate": 1.964901062626901e-05,
      "loss": 0.3871,
      "step": 2281
    },
    {
      "epoch": 0.17270706210680667,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9648697763932007e-05,
      "loss": 0.3732,
      "step": 2282
    },
    {
      "epoch": 0.17278274443025401,
      "grad_norm": 0.82421875,
      "learning_rate": 1.964838476471167e-05,
      "loss": 0.3372,
      "step": 2283
    },
    {
      "epoch": 0.17285842675370133,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9648071628612437e-05,
      "loss": 0.3554,
      "step": 2284
    },
    {
      "epoch": 0.17293410907714868,
      "grad_norm": 0.79296875,
      "learning_rate": 1.964775835563875e-05,
      "loss": 0.2976,
      "step": 2285
    },
    {
      "epoch": 0.173009791400596,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9647444945795056e-05,
      "loss": 0.3499,
      "step": 2286
    },
    {
      "epoch": 0.17308547372404331,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9647131399085796e-05,
      "loss": 0.3288,
      "step": 2287
    },
    {
      "epoch": 0.17316115604749066,
      "grad_norm": 0.921875,
      "learning_rate": 1.9646817715515425e-05,
      "loss": 0.3617,
      "step": 2288
    },
    {
      "epoch": 0.17323683837093798,
      "grad_norm": 0.97265625,
      "learning_rate": 1.964650389508839e-05,
      "loss": 0.4041,
      "step": 2289
    },
    {
      "epoch": 0.17331252069438532,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9646189937809145e-05,
      "loss": 0.3156,
      "step": 2290
    },
    {
      "epoch": 0.17338820301783264,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9645875843682142e-05,
      "loss": 0.3194,
      "step": 2291
    },
    {
      "epoch": 0.17346388534128,
      "grad_norm": 0.890625,
      "learning_rate": 1.9645561612711833e-05,
      "loss": 0.3688,
      "step": 2292
    },
    {
      "epoch": 0.1735395676647273,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9645247244902685e-05,
      "loss": 0.3242,
      "step": 2293
    },
    {
      "epoch": 0.17361524998817462,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9644932740259152e-05,
      "loss": 0.3144,
      "step": 2294
    },
    {
      "epoch": 0.17369093231162197,
      "grad_norm": 0.953125,
      "learning_rate": 1.9644618098785696e-05,
      "loss": 0.3998,
      "step": 2295
    },
    {
      "epoch": 0.1737666146350693,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9644303320486783e-05,
      "loss": 0.3224,
      "step": 2296
    },
    {
      "epoch": 0.17384229695851663,
      "grad_norm": 0.90625,
      "learning_rate": 1.9643988405366872e-05,
      "loss": 0.3487,
      "step": 2297
    },
    {
      "epoch": 0.17391797928196395,
      "grad_norm": 0.796875,
      "learning_rate": 1.9643673353430443e-05,
      "loss": 0.3229,
      "step": 2298
    },
    {
      "epoch": 0.1739936616054113,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9643358164681952e-05,
      "loss": 0.2821,
      "step": 2299
    },
    {
      "epoch": 0.17406934392885862,
      "grad_norm": 0.91015625,
      "learning_rate": 1.964304283912588e-05,
      "loss": 0.403,
      "step": 2300
    },
    {
      "epoch": 0.17414502625230593,
      "grad_norm": 0.86328125,
      "learning_rate": 1.96427273767667e-05,
      "loss": 0.3467,
      "step": 2301
    },
    {
      "epoch": 0.17422070857575328,
      "grad_norm": 0.859375,
      "learning_rate": 1.9642411777608877e-05,
      "loss": 0.3199,
      "step": 2302
    },
    {
      "epoch": 0.1742963908992006,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9642096041656903e-05,
      "loss": 0.3564,
      "step": 2303
    },
    {
      "epoch": 0.17437207322264794,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9641780168915247e-05,
      "loss": 0.3362,
      "step": 2304
    },
    {
      "epoch": 0.17444775554609526,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9641464159388395e-05,
      "loss": 0.3391,
      "step": 2305
    },
    {
      "epoch": 0.1745234378695426,
      "grad_norm": 4.21875,
      "learning_rate": 1.964114801308083e-05,
      "loss": 0.5351,
      "step": 2306
    },
    {
      "epoch": 0.17459912019298993,
      "grad_norm": 0.88671875,
      "learning_rate": 1.964083172999703e-05,
      "loss": 0.3346,
      "step": 2307
    },
    {
      "epoch": 0.17467480251643724,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9640515310141494e-05,
      "loss": 0.2897,
      "step": 2308
    },
    {
      "epoch": 0.1747504848398846,
      "grad_norm": 0.8671875,
      "learning_rate": 1.96401987535187e-05,
      "loss": 0.3374,
      "step": 2309
    },
    {
      "epoch": 0.1748261671633319,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9639882060133148e-05,
      "loss": 0.3554,
      "step": 2310
    },
    {
      "epoch": 0.17490184948677925,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9639565229989325e-05,
      "loss": 0.302,
      "step": 2311
    },
    {
      "epoch": 0.17497753181022657,
      "grad_norm": 0.875,
      "learning_rate": 1.9639248263091725e-05,
      "loss": 0.3376,
      "step": 2312
    },
    {
      "epoch": 0.1750532141336739,
      "grad_norm": 0.8125,
      "learning_rate": 1.963893115944485e-05,
      "loss": 0.3253,
      "step": 2313
    },
    {
      "epoch": 0.17512889645712124,
      "grad_norm": 0.875,
      "learning_rate": 1.963861391905319e-05,
      "loss": 0.3506,
      "step": 2314
    },
    {
      "epoch": 0.17520457878056855,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9638296541921255e-05,
      "loss": 0.3232,
      "step": 2315
    },
    {
      "epoch": 0.1752802611040159,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9637979028053545e-05,
      "loss": 0.3597,
      "step": 2316
    },
    {
      "epoch": 0.17535594342746322,
      "grad_norm": 2.03125,
      "learning_rate": 1.963766137745456e-05,
      "loss": 0.447,
      "step": 2317
    },
    {
      "epoch": 0.17543162575091056,
      "grad_norm": 0.8046875,
      "learning_rate": 1.963734359012881e-05,
      "loss": 0.3016,
      "step": 2318
    },
    {
      "epoch": 0.17550730807435788,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9637025666080808e-05,
      "loss": 0.3372,
      "step": 2319
    },
    {
      "epoch": 0.1755829903978052,
      "grad_norm": 1.0,
      "learning_rate": 1.9636707605315054e-05,
      "loss": 0.364,
      "step": 2320
    },
    {
      "epoch": 0.17565867272125255,
      "grad_norm": 0.828125,
      "learning_rate": 1.9636389407836065e-05,
      "loss": 0.3146,
      "step": 2321
    },
    {
      "epoch": 0.17573435504469986,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9636071073648357e-05,
      "loss": 0.375,
      "step": 2322
    },
    {
      "epoch": 0.1758100373681472,
      "grad_norm": 0.90625,
      "learning_rate": 1.9635752602756443e-05,
      "loss": 0.3656,
      "step": 2323
    },
    {
      "epoch": 0.17588571969159453,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9635433995164846e-05,
      "loss": 0.3053,
      "step": 2324
    },
    {
      "epoch": 0.17596140201504187,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9635115250878083e-05,
      "loss": 0.3877,
      "step": 2325
    },
    {
      "epoch": 0.1760370843384892,
      "grad_norm": 0.90625,
      "learning_rate": 1.9634796369900673e-05,
      "loss": 0.3903,
      "step": 2326
    },
    {
      "epoch": 0.1761127666619365,
      "grad_norm": 0.859375,
      "learning_rate": 1.9634477352237144e-05,
      "loss": 0.3493,
      "step": 2327
    },
    {
      "epoch": 0.17618844898538386,
      "grad_norm": 0.87109375,
      "learning_rate": 1.963415819789202e-05,
      "loss": 0.3753,
      "step": 2328
    },
    {
      "epoch": 0.17626413130883117,
      "grad_norm": 0.83203125,
      "learning_rate": 1.963383890686983e-05,
      "loss": 0.3416,
      "step": 2329
    },
    {
      "epoch": 0.17633981363227852,
      "grad_norm": 0.8125,
      "learning_rate": 1.9633519479175103e-05,
      "loss": 0.3372,
      "step": 2330
    },
    {
      "epoch": 0.17641549595572584,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9633199914812372e-05,
      "loss": 0.4068,
      "step": 2331
    },
    {
      "epoch": 0.17649117827917318,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9632880213786164e-05,
      "loss": 0.3271,
      "step": 2332
    },
    {
      "epoch": 0.1765668606026205,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9632560376101026e-05,
      "loss": 0.393,
      "step": 2333
    },
    {
      "epoch": 0.17664254292606782,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9632240401761484e-05,
      "loss": 0.3458,
      "step": 2334
    },
    {
      "epoch": 0.17671822524951517,
      "grad_norm": 3.765625,
      "learning_rate": 1.963192029077208e-05,
      "loss": 0.5604,
      "step": 2335
    },
    {
      "epoch": 0.17679390757296248,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9631600043137366e-05,
      "loss": 0.3382,
      "step": 2336
    },
    {
      "epoch": 0.17686958989640983,
      "grad_norm": 0.96875,
      "learning_rate": 1.9631279658861873e-05,
      "loss": 0.4015,
      "step": 2337
    },
    {
      "epoch": 0.17694527221985715,
      "grad_norm": 0.82421875,
      "learning_rate": 1.963095913795015e-05,
      "loss": 0.3078,
      "step": 2338
    },
    {
      "epoch": 0.1770209545433045,
      "grad_norm": 0.875,
      "learning_rate": 1.9630638480406742e-05,
      "loss": 0.3938,
      "step": 2339
    },
    {
      "epoch": 0.1770966368667518,
      "grad_norm": 0.8125,
      "learning_rate": 1.9630317686236204e-05,
      "loss": 0.3405,
      "step": 2340
    },
    {
      "epoch": 0.17717231919019913,
      "grad_norm": 0.86328125,
      "learning_rate": 1.962999675544308e-05,
      "loss": 0.3327,
      "step": 2341
    },
    {
      "epoch": 0.17724800151364647,
      "grad_norm": 0.921875,
      "learning_rate": 1.962967568803193e-05,
      "loss": 0.3526,
      "step": 2342
    },
    {
      "epoch": 0.1773236838370938,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9629354484007304e-05,
      "loss": 0.3397,
      "step": 2343
    },
    {
      "epoch": 0.17739936616054114,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9629033143373762e-05,
      "loss": 0.337,
      "step": 2344
    },
    {
      "epoch": 0.17747504848398846,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9628711666135855e-05,
      "loss": 0.3347,
      "step": 2345
    },
    {
      "epoch": 0.17755073080743577,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9628390052298155e-05,
      "loss": 0.3809,
      "step": 2346
    },
    {
      "epoch": 0.17762641313088312,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9628068301865217e-05,
      "loss": 0.3562,
      "step": 2347
    },
    {
      "epoch": 0.17770209545433044,
      "grad_norm": 0.9453125,
      "learning_rate": 1.9627746414841607e-05,
      "loss": 0.4325,
      "step": 2348
    },
    {
      "epoch": 0.17777777777777778,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9627424391231894e-05,
      "loss": 0.3608,
      "step": 2349
    },
    {
      "epoch": 0.1778534601012251,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9627102231040646e-05,
      "loss": 0.3506,
      "step": 2350
    },
    {
      "epoch": 0.17792914242467245,
      "grad_norm": 0.87109375,
      "learning_rate": 1.962677993427243e-05,
      "loss": 0.3256,
      "step": 2351
    },
    {
      "epoch": 0.17800482474811977,
      "grad_norm": 0.8203125,
      "learning_rate": 1.962645750093182e-05,
      "loss": 0.3103,
      "step": 2352
    },
    {
      "epoch": 0.17808050707156708,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9626134931023394e-05,
      "loss": 0.2896,
      "step": 2353
    },
    {
      "epoch": 0.17815618939501443,
      "grad_norm": 1.171875,
      "learning_rate": 1.9625812224551725e-05,
      "loss": 0.36,
      "step": 2354
    },
    {
      "epoch": 0.17823187171846175,
      "grad_norm": 0.828125,
      "learning_rate": 1.9625489381521387e-05,
      "loss": 0.3385,
      "step": 2355
    },
    {
      "epoch": 0.1783075540419091,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9625166401936964e-05,
      "loss": 0.3653,
      "step": 2356
    },
    {
      "epoch": 0.1783832363653564,
      "grad_norm": 0.828125,
      "learning_rate": 1.9624843285803045e-05,
      "loss": 0.3002,
      "step": 2357
    },
    {
      "epoch": 0.17845891868880376,
      "grad_norm": 0.86328125,
      "learning_rate": 1.96245200331242e-05,
      "loss": 0.3508,
      "step": 2358
    },
    {
      "epoch": 0.17853460101225108,
      "grad_norm": 0.90625,
      "learning_rate": 1.9624196643905024e-05,
      "loss": 0.3815,
      "step": 2359
    },
    {
      "epoch": 0.1786102833356984,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9623873118150105e-05,
      "loss": 0.3604,
      "step": 2360
    },
    {
      "epoch": 0.17868596565914574,
      "grad_norm": 0.796875,
      "learning_rate": 1.962354945586403e-05,
      "loss": 0.3155,
      "step": 2361
    },
    {
      "epoch": 0.17876164798259306,
      "grad_norm": 0.90234375,
      "learning_rate": 1.962322565705139e-05,
      "loss": 0.3723,
      "step": 2362
    },
    {
      "epoch": 0.1788373303060404,
      "grad_norm": 0.86328125,
      "learning_rate": 1.962290172171678e-05,
      "loss": 0.347,
      "step": 2363
    },
    {
      "epoch": 0.17891301262948772,
      "grad_norm": 0.859375,
      "learning_rate": 1.9622577649864797e-05,
      "loss": 0.3312,
      "step": 2364
    },
    {
      "epoch": 0.17898869495293507,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9622253441500035e-05,
      "loss": 0.3586,
      "step": 2365
    },
    {
      "epoch": 0.17906437727638239,
      "grad_norm": 0.8125,
      "learning_rate": 1.9621929096627098e-05,
      "loss": 0.3239,
      "step": 2366
    },
    {
      "epoch": 0.1791400595998297,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9621604615250583e-05,
      "loss": 0.3203,
      "step": 2367
    },
    {
      "epoch": 0.17921574192327705,
      "grad_norm": 5.15625,
      "learning_rate": 1.9621279997375096e-05,
      "loss": 0.6106,
      "step": 2368
    },
    {
      "epoch": 0.17929142424672437,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9620955243005243e-05,
      "loss": 0.3502,
      "step": 2369
    },
    {
      "epoch": 0.1793671065701717,
      "grad_norm": 0.921875,
      "learning_rate": 1.962063035214563e-05,
      "loss": 0.3239,
      "step": 2370
    },
    {
      "epoch": 0.17944278889361903,
      "grad_norm": 0.921875,
      "learning_rate": 1.9620305324800865e-05,
      "loss": 0.3943,
      "step": 2371
    },
    {
      "epoch": 0.17951847121706638,
      "grad_norm": 0.7578125,
      "learning_rate": 1.961998016097556e-05,
      "loss": 0.2743,
      "step": 2372
    },
    {
      "epoch": 0.1795941535405137,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9619654860674323e-05,
      "loss": 0.3646,
      "step": 2373
    },
    {
      "epoch": 0.179669835863961,
      "grad_norm": 1.03125,
      "learning_rate": 1.9619329423901782e-05,
      "loss": 0.4032,
      "step": 2374
    },
    {
      "epoch": 0.17974551818740836,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9619003850662542e-05,
      "loss": 0.3288,
      "step": 2375
    },
    {
      "epoch": 0.17982120051085568,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9618678140961226e-05,
      "loss": 0.3078,
      "step": 2376
    },
    {
      "epoch": 0.17989688283430302,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9618352294802455e-05,
      "loss": 0.3339,
      "step": 2377
    },
    {
      "epoch": 0.17997256515775034,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9618026312190847e-05,
      "loss": 0.3239,
      "step": 2378
    },
    {
      "epoch": 0.18004824748119766,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9617700193131034e-05,
      "loss": 0.3338,
      "step": 2379
    },
    {
      "epoch": 0.180123929804645,
      "grad_norm": 0.8515625,
      "learning_rate": 1.961737393762764e-05,
      "loss": 0.3485,
      "step": 2380
    },
    {
      "epoch": 0.18019961212809232,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9617047545685293e-05,
      "loss": 0.3567,
      "step": 2381
    },
    {
      "epoch": 0.18027529445153967,
      "grad_norm": 0.78515625,
      "learning_rate": 1.961672101730862e-05,
      "loss": 0.3012,
      "step": 2382
    },
    {
      "epoch": 0.180350976774987,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9616394352502256e-05,
      "loss": 0.3021,
      "step": 2383
    },
    {
      "epoch": 0.18042665909843433,
      "grad_norm": 0.875,
      "learning_rate": 1.961606755127084e-05,
      "loss": 0.373,
      "step": 2384
    },
    {
      "epoch": 0.18050234142188165,
      "grad_norm": 0.73828125,
      "learning_rate": 1.9615740613619004e-05,
      "loss": 0.27,
      "step": 2385
    },
    {
      "epoch": 0.18057802374532897,
      "grad_norm": 0.9453125,
      "learning_rate": 1.9615413539551382e-05,
      "loss": 0.3783,
      "step": 2386
    },
    {
      "epoch": 0.18065370606877632,
      "grad_norm": 0.82421875,
      "learning_rate": 1.961508632907262e-05,
      "loss": 0.3116,
      "step": 2387
    },
    {
      "epoch": 0.18072938839222363,
      "grad_norm": 1.6875,
      "learning_rate": 1.961475898218736e-05,
      "loss": 0.4357,
      "step": 2388
    },
    {
      "epoch": 0.18080507071567098,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9614431498900246e-05,
      "loss": 0.3629,
      "step": 2389
    },
    {
      "epoch": 0.1808807530391183,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9614103879215917e-05,
      "loss": 0.3554,
      "step": 2390
    },
    {
      "epoch": 0.18095643536256564,
      "grad_norm": 0.8046875,
      "learning_rate": 1.961377612313903e-05,
      "loss": 0.3076,
      "step": 2391
    },
    {
      "epoch": 0.18103211768601296,
      "grad_norm": 1.578125,
      "learning_rate": 1.961344823067423e-05,
      "loss": 0.425,
      "step": 2392
    },
    {
      "epoch": 0.18110780000946028,
      "grad_norm": 0.890625,
      "learning_rate": 1.961312020182617e-05,
      "loss": 0.332,
      "step": 2393
    },
    {
      "epoch": 0.18118348233290762,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9612792036599502e-05,
      "loss": 0.3092,
      "step": 2394
    },
    {
      "epoch": 0.18125916465635494,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9612463734998882e-05,
      "loss": 0.388,
      "step": 2395
    },
    {
      "epoch": 0.1813348469798023,
      "grad_norm": 0.78515625,
      "learning_rate": 1.9612135297028973e-05,
      "loss": 0.2829,
      "step": 2396
    },
    {
      "epoch": 0.1814105293032496,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9611806722694427e-05,
      "loss": 0.3413,
      "step": 2397
    },
    {
      "epoch": 0.18148621162669695,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9611478011999906e-05,
      "loss": 0.3957,
      "step": 2398
    },
    {
      "epoch": 0.18156189395014427,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9611149164950078e-05,
      "loss": 0.293,
      "step": 2399
    },
    {
      "epoch": 0.1816375762735916,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9610820181549606e-05,
      "loss": 0.3421,
      "step": 2400
    },
    {
      "epoch": 0.18171325859703893,
      "grad_norm": 0.84375,
      "learning_rate": 1.9610491061803153e-05,
      "loss": 0.3418,
      "step": 2401
    },
    {
      "epoch": 0.18178894092048625,
      "grad_norm": 0.859375,
      "learning_rate": 1.96101618057154e-05,
      "loss": 0.3002,
      "step": 2402
    },
    {
      "epoch": 0.1818646232439336,
      "grad_norm": 2.78125,
      "learning_rate": 1.9609832413291e-05,
      "loss": 0.5291,
      "step": 2403
    },
    {
      "epoch": 0.18194030556738092,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9609502884534645e-05,
      "loss": 0.3216,
      "step": 2404
    },
    {
      "epoch": 0.18201598789082823,
      "grad_norm": 1.171875,
      "learning_rate": 1.9609173219450998e-05,
      "loss": 0.3753,
      "step": 2405
    },
    {
      "epoch": 0.18209167021427558,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9608843418044738e-05,
      "loss": 0.3055,
      "step": 2406
    },
    {
      "epoch": 0.1821673525377229,
      "grad_norm": 0.7890625,
      "learning_rate": 1.9608513480320545e-05,
      "loss": 0.2999,
      "step": 2407
    },
    {
      "epoch": 0.18224303486117024,
      "grad_norm": 0.8828125,
      "learning_rate": 1.96081834062831e-05,
      "loss": 0.3448,
      "step": 2408
    },
    {
      "epoch": 0.18231871718461756,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9607853195937085e-05,
      "loss": 0.3291,
      "step": 2409
    },
    {
      "epoch": 0.1823943995080649,
      "grad_norm": 0.796875,
      "learning_rate": 1.9607522849287185e-05,
      "loss": 0.2915,
      "step": 2410
    },
    {
      "epoch": 0.18247008183151223,
      "grad_norm": 0.86328125,
      "learning_rate": 1.960719236633809e-05,
      "loss": 0.3271,
      "step": 2411
    },
    {
      "epoch": 0.18254576415495954,
      "grad_norm": 0.78515625,
      "learning_rate": 1.9606861747094478e-05,
      "loss": 0.2954,
      "step": 2412
    },
    {
      "epoch": 0.1826214464784069,
      "grad_norm": 0.8359375,
      "learning_rate": 1.960653099156105e-05,
      "loss": 0.3289,
      "step": 2413
    },
    {
      "epoch": 0.1826971288018542,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9606200099742492e-05,
      "loss": 0.3626,
      "step": 2414
    },
    {
      "epoch": 0.18277281112530155,
      "grad_norm": 0.78125,
      "learning_rate": 1.96058690716435e-05,
      "loss": 0.3002,
      "step": 2415
    },
    {
      "epoch": 0.18284849344874887,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9605537907268776e-05,
      "loss": 0.357,
      "step": 2416
    },
    {
      "epoch": 0.18292417577219622,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9605206606623014e-05,
      "loss": 0.3807,
      "step": 2417
    },
    {
      "epoch": 0.18299985809564354,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9604875169710906e-05,
      "loss": 0.3796,
      "step": 2418
    },
    {
      "epoch": 0.18307554041909085,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9604543596537166e-05,
      "loss": 0.3242,
      "step": 2419
    },
    {
      "epoch": 0.1831512227425382,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9604211887106494e-05,
      "loss": 0.3344,
      "step": 2420
    },
    {
      "epoch": 0.18322690506598552,
      "grad_norm": 0.83203125,
      "learning_rate": 1.960388004142359e-05,
      "loss": 0.3303,
      "step": 2421
    },
    {
      "epoch": 0.18330258738943286,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9603548059493173e-05,
      "loss": 0.2927,
      "step": 2422
    },
    {
      "epoch": 0.18337826971288018,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9603215941319942e-05,
      "loss": 0.3344,
      "step": 2423
    },
    {
      "epoch": 0.18345395203632753,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9602883686908614e-05,
      "loss": 0.3562,
      "step": 2424
    },
    {
      "epoch": 0.18352963435977485,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9602551296263902e-05,
      "loss": 0.3504,
      "step": 2425
    },
    {
      "epoch": 0.18360531668322216,
      "grad_norm": 1.0703125,
      "learning_rate": 1.9602218769390522e-05,
      "loss": 0.3527,
      "step": 2426
    },
    {
      "epoch": 0.1836809990066695,
      "grad_norm": 0.84375,
      "learning_rate": 1.960188610629319e-05,
      "loss": 0.3446,
      "step": 2427
    },
    {
      "epoch": 0.18375668133011683,
      "grad_norm": 0.890625,
      "learning_rate": 1.9601553306976626e-05,
      "loss": 0.3409,
      "step": 2428
    },
    {
      "epoch": 0.18383236365356417,
      "grad_norm": 0.97265625,
      "learning_rate": 1.9601220371445553e-05,
      "loss": 0.3697,
      "step": 2429
    },
    {
      "epoch": 0.1839080459770115,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9600887299704694e-05,
      "loss": 0.3114,
      "step": 2430
    },
    {
      "epoch": 0.18398372830045884,
      "grad_norm": 0.9140625,
      "learning_rate": 1.960055409175877e-05,
      "loss": 0.3658,
      "step": 2431
    },
    {
      "epoch": 0.18405941062390616,
      "grad_norm": 0.81640625,
      "learning_rate": 1.960022074761251e-05,
      "loss": 0.3237,
      "step": 2432
    },
    {
      "epoch": 0.18413509294735347,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9599887267270648e-05,
      "loss": 0.345,
      "step": 2433
    },
    {
      "epoch": 0.18421077527080082,
      "grad_norm": 0.9296875,
      "learning_rate": 1.959955365073791e-05,
      "loss": 0.378,
      "step": 2434
    },
    {
      "epoch": 0.18428645759424814,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9599219898019027e-05,
      "loss": 0.346,
      "step": 2435
    },
    {
      "epoch": 0.18436213991769548,
      "grad_norm": 0.84765625,
      "learning_rate": 1.959888600911874e-05,
      "loss": 0.3133,
      "step": 2436
    },
    {
      "epoch": 0.1844378222411428,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9598551984041782e-05,
      "loss": 0.3588,
      "step": 2437
    },
    {
      "epoch": 0.18451350456459012,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9598217822792892e-05,
      "loss": 0.3259,
      "step": 2438
    },
    {
      "epoch": 0.18458918688803747,
      "grad_norm": 0.796875,
      "learning_rate": 1.959788352537681e-05,
      "loss": 0.3227,
      "step": 2439
    },
    {
      "epoch": 0.18466486921148478,
      "grad_norm": 0.828125,
      "learning_rate": 1.9597549091798284e-05,
      "loss": 0.3518,
      "step": 2440
    },
    {
      "epoch": 0.18474055153493213,
      "grad_norm": 0.828125,
      "learning_rate": 1.9597214522062053e-05,
      "loss": 0.3267,
      "step": 2441
    },
    {
      "epoch": 0.18481623385837945,
      "grad_norm": 0.84375,
      "learning_rate": 1.959687981617286e-05,
      "loss": 0.3105,
      "step": 2442
    },
    {
      "epoch": 0.1848919161818268,
      "grad_norm": 0.8984375,
      "learning_rate": 1.959654497413546e-05,
      "loss": 0.3343,
      "step": 2443
    },
    {
      "epoch": 0.1849675985052741,
      "grad_norm": 1.0390625,
      "learning_rate": 1.9596209995954606e-05,
      "loss": 0.3601,
      "step": 2444
    },
    {
      "epoch": 0.18504328082872143,
      "grad_norm": 0.890625,
      "learning_rate": 1.9595874881635038e-05,
      "loss": 0.3705,
      "step": 2445
    },
    {
      "epoch": 0.18511896315216878,
      "grad_norm": 0.91796875,
      "learning_rate": 1.959553963118152e-05,
      "loss": 0.3879,
      "step": 2446
    },
    {
      "epoch": 0.1851946454756161,
      "grad_norm": 0.8515625,
      "learning_rate": 1.959520424459881e-05,
      "loss": 0.3456,
      "step": 2447
    },
    {
      "epoch": 0.18527032779906344,
      "grad_norm": 0.87109375,
      "learning_rate": 1.959486872189166e-05,
      "loss": 0.339,
      "step": 2448
    },
    {
      "epoch": 0.18534601012251076,
      "grad_norm": 0.8046875,
      "learning_rate": 1.959453306306483e-05,
      "loss": 0.3001,
      "step": 2449
    },
    {
      "epoch": 0.1854216924459581,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9594197268123087e-05,
      "loss": 0.3508,
      "step": 2450
    },
    {
      "epoch": 0.18549737476940542,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9593861337071187e-05,
      "loss": 0.3483,
      "step": 2451
    },
    {
      "epoch": 0.18557305709285274,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9593525269913903e-05,
      "loss": 0.3567,
      "step": 2452
    },
    {
      "epoch": 0.18564873941630008,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9593189066655998e-05,
      "loss": 0.3197,
      "step": 2453
    },
    {
      "epoch": 0.1857244217397474,
      "grad_norm": 1.0390625,
      "learning_rate": 1.9592852727302247e-05,
      "loss": 0.4041,
      "step": 2454
    },
    {
      "epoch": 0.18580010406319475,
      "grad_norm": 0.84375,
      "learning_rate": 1.9592516251857417e-05,
      "loss": 0.3236,
      "step": 2455
    },
    {
      "epoch": 0.18587578638664207,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9592179640326277e-05,
      "loss": 0.3013,
      "step": 2456
    },
    {
      "epoch": 0.1859514687100894,
      "grad_norm": 0.97265625,
      "learning_rate": 1.9591842892713616e-05,
      "loss": 0.3262,
      "step": 2457
    },
    {
      "epoch": 0.18602715103353673,
      "grad_norm": 0.84765625,
      "learning_rate": 1.95915060090242e-05,
      "loss": 0.3442,
      "step": 2458
    },
    {
      "epoch": 0.18610283335698405,
      "grad_norm": 0.859375,
      "learning_rate": 1.9591168989262815e-05,
      "loss": 0.3346,
      "step": 2459
    },
    {
      "epoch": 0.1861785156804314,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9590831833434233e-05,
      "loss": 0.3124,
      "step": 2460
    },
    {
      "epoch": 0.1862541980038787,
      "grad_norm": 0.8671875,
      "learning_rate": 1.959049454154325e-05,
      "loss": 0.3509,
      "step": 2461
    },
    {
      "epoch": 0.18632988032732606,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9590157113594636e-05,
      "loss": 0.355,
      "step": 2462
    },
    {
      "epoch": 0.18640556265077338,
      "grad_norm": 0.8515625,
      "learning_rate": 1.958981954959319e-05,
      "loss": 0.331,
      "step": 2463
    },
    {
      "epoch": 0.18648124497422072,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9589481849543696e-05,
      "loss": 0.3311,
      "step": 2464
    },
    {
      "epoch": 0.18655692729766804,
      "grad_norm": 0.90234375,
      "learning_rate": 1.958914401345095e-05,
      "loss": 0.3557,
      "step": 2465
    },
    {
      "epoch": 0.18663260962111536,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9588806041319735e-05,
      "loss": 0.3739,
      "step": 2466
    },
    {
      "epoch": 0.1867082919445627,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9588467933154856e-05,
      "loss": 0.3212,
      "step": 2467
    },
    {
      "epoch": 0.18678397426801002,
      "grad_norm": 0.921875,
      "learning_rate": 1.95881296889611e-05,
      "loss": 0.3921,
      "step": 2468
    },
    {
      "epoch": 0.18685965659145737,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9587791308743277e-05,
      "loss": 0.3555,
      "step": 2469
    },
    {
      "epoch": 0.1869353389149047,
      "grad_norm": 0.76953125,
      "learning_rate": 1.9587452792506178e-05,
      "loss": 0.2728,
      "step": 2470
    },
    {
      "epoch": 0.187011021238352,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9587114140254605e-05,
      "loss": 0.3262,
      "step": 2471
    },
    {
      "epoch": 0.18708670356179935,
      "grad_norm": 0.953125,
      "learning_rate": 1.958677535199337e-05,
      "loss": 0.3867,
      "step": 2472
    },
    {
      "epoch": 0.18716238588524667,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9586436427727276e-05,
      "loss": 0.339,
      "step": 2473
    },
    {
      "epoch": 0.18723806820869401,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9586097367461128e-05,
      "loss": 0.3259,
      "step": 2474
    },
    {
      "epoch": 0.18731375053214133,
      "grad_norm": 0.921875,
      "learning_rate": 1.9585758171199738e-05,
      "loss": 0.3835,
      "step": 2475
    },
    {
      "epoch": 0.18738943285558868,
      "grad_norm": 0.8203125,
      "learning_rate": 1.958541883894792e-05,
      "loss": 0.3323,
      "step": 2476
    },
    {
      "epoch": 0.187465115179036,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9585079370710484e-05,
      "loss": 0.3144,
      "step": 2477
    },
    {
      "epoch": 0.18754079750248331,
      "grad_norm": 2.453125,
      "learning_rate": 1.9584739766492253e-05,
      "loss": 0.3359,
      "step": 2478
    },
    {
      "epoch": 0.18761647982593066,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9584400026298037e-05,
      "loss": 0.4192,
      "step": 2479
    },
    {
      "epoch": 0.18769216214937798,
      "grad_norm": 0.90625,
      "learning_rate": 1.958406015013266e-05,
      "loss": 0.377,
      "step": 2480
    },
    {
      "epoch": 0.18776784447282532,
      "grad_norm": 0.859375,
      "learning_rate": 1.9583720138000944e-05,
      "loss": 0.323,
      "step": 2481
    },
    {
      "epoch": 0.18784352679627264,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9583379989907712e-05,
      "loss": 0.3636,
      "step": 2482
    },
    {
      "epoch": 0.18791920911972,
      "grad_norm": 0.83203125,
      "learning_rate": 1.958303970585779e-05,
      "loss": 0.325,
      "step": 2483
    },
    {
      "epoch": 0.1879948914431673,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9582699285856e-05,
      "loss": 0.3369,
      "step": 2484
    },
    {
      "epoch": 0.18807057376661462,
      "grad_norm": 0.80859375,
      "learning_rate": 1.958235872990718e-05,
      "loss": 0.325,
      "step": 2485
    },
    {
      "epoch": 0.18814625609006197,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9582018038016156e-05,
      "loss": 0.3765,
      "step": 2486
    },
    {
      "epoch": 0.1882219384135093,
      "grad_norm": 0.859375,
      "learning_rate": 1.9581677210187763e-05,
      "loss": 0.3821,
      "step": 2487
    },
    {
      "epoch": 0.18829762073695663,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9581336246426835e-05,
      "loss": 0.3701,
      "step": 2488
    },
    {
      "epoch": 0.18837330306040395,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9580995146738213e-05,
      "loss": 0.3262,
      "step": 2489
    },
    {
      "epoch": 0.1884489853838513,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9580653911126733e-05,
      "loss": 0.3612,
      "step": 2490
    },
    {
      "epoch": 0.18852466770729862,
      "grad_norm": 0.875,
      "learning_rate": 1.9580312539597236e-05,
      "loss": 0.3505,
      "step": 2491
    },
    {
      "epoch": 0.18860035003074593,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9579971032154563e-05,
      "loss": 0.3341,
      "step": 2492
    },
    {
      "epoch": 0.18867603235419328,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9579629388803563e-05,
      "loss": 0.2972,
      "step": 2493
    },
    {
      "epoch": 0.1887517146776406,
      "grad_norm": 0.93359375,
      "learning_rate": 1.957928760954908e-05,
      "loss": 0.3537,
      "step": 2494
    },
    {
      "epoch": 0.18882739700108794,
      "grad_norm": 4.1875,
      "learning_rate": 1.9578945694395967e-05,
      "loss": 0.5261,
      "step": 2495
    },
    {
      "epoch": 0.18890307932453526,
      "grad_norm": 0.875,
      "learning_rate": 1.9578603643349066e-05,
      "loss": 0.3652,
      "step": 2496
    },
    {
      "epoch": 0.18897876164798258,
      "grad_norm": 0.82421875,
      "learning_rate": 1.957826145641324e-05,
      "loss": 0.3066,
      "step": 2497
    },
    {
      "epoch": 0.18905444397142993,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9577919133593335e-05,
      "loss": 0.3473,
      "step": 2498
    },
    {
      "epoch": 0.18913012629487724,
      "grad_norm": 0.81640625,
      "learning_rate": 1.957757667489421e-05,
      "loss": 0.3608,
      "step": 2499
    },
    {
      "epoch": 0.1892058086183246,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9577234080320725e-05,
      "loss": 0.367,
      "step": 2500
    },
    {
      "epoch": 0.1892814909417719,
      "grad_norm": 0.8828125,
      "learning_rate": 1.957689134987774e-05,
      "loss": 0.3709,
      "step": 2501
    },
    {
      "epoch": 0.18935717326521925,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9576548483570116e-05,
      "loss": 0.336,
      "step": 2502
    },
    {
      "epoch": 0.18943285558866657,
      "grad_norm": 0.82421875,
      "learning_rate": 1.957620548140272e-05,
      "loss": 0.3528,
      "step": 2503
    },
    {
      "epoch": 0.1895085379121139,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9575862343380417e-05,
      "loss": 0.2846,
      "step": 2504
    },
    {
      "epoch": 0.18958422023556123,
      "grad_norm": 0.828125,
      "learning_rate": 1.957551906950807e-05,
      "loss": 0.3203,
      "step": 2505
    },
    {
      "epoch": 0.18965990255900855,
      "grad_norm": 0.875,
      "learning_rate": 1.9575175659790555e-05,
      "loss": 0.3229,
      "step": 2506
    },
    {
      "epoch": 0.1897355848824559,
      "grad_norm": 0.7734375,
      "learning_rate": 1.957483211423274e-05,
      "loss": 0.2881,
      "step": 2507
    },
    {
      "epoch": 0.18981126720590322,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9574488432839502e-05,
      "loss": 0.2668,
      "step": 2508
    },
    {
      "epoch": 0.18988694952935056,
      "grad_norm": 0.94921875,
      "learning_rate": 1.9574144615615713e-05,
      "loss": 0.4168,
      "step": 2509
    },
    {
      "epoch": 0.18996263185279788,
      "grad_norm": 0.875,
      "learning_rate": 1.9573800662566257e-05,
      "loss": 0.3281,
      "step": 2510
    },
    {
      "epoch": 0.1900383141762452,
      "grad_norm": 0.76953125,
      "learning_rate": 1.9573456573696006e-05,
      "loss": 0.3061,
      "step": 2511
    },
    {
      "epoch": 0.19011399649969254,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9573112349009846e-05,
      "loss": 0.322,
      "step": 2512
    },
    {
      "epoch": 0.19018967882313986,
      "grad_norm": 0.81640625,
      "learning_rate": 1.957276798851266e-05,
      "loss": 0.293,
      "step": 2513
    },
    {
      "epoch": 0.1902653611465872,
      "grad_norm": 0.8984375,
      "learning_rate": 1.957242349220933e-05,
      "loss": 0.3513,
      "step": 2514
    },
    {
      "epoch": 0.19034104347003453,
      "grad_norm": 0.765625,
      "learning_rate": 1.9572078860104748e-05,
      "loss": 0.2757,
      "step": 2515
    },
    {
      "epoch": 0.19041672579348187,
      "grad_norm": 0.828125,
      "learning_rate": 1.95717340922038e-05,
      "loss": 0.319,
      "step": 2516
    },
    {
      "epoch": 0.1904924081169292,
      "grad_norm": 0.91015625,
      "learning_rate": 1.957138918851138e-05,
      "loss": 0.3515,
      "step": 2517
    },
    {
      "epoch": 0.1905680904403765,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9571044149032377e-05,
      "loss": 0.3172,
      "step": 2518
    },
    {
      "epoch": 0.19064377276382385,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9570698973771692e-05,
      "loss": 0.3435,
      "step": 2519
    },
    {
      "epoch": 0.19071945508727117,
      "grad_norm": 6.4375,
      "learning_rate": 1.9570353662734217e-05,
      "loss": 0.4546,
      "step": 2520
    },
    {
      "epoch": 0.19079513741071852,
      "grad_norm": 0.859375,
      "learning_rate": 1.957000821592485e-05,
      "loss": 0.3524,
      "step": 2521
    },
    {
      "epoch": 0.19087081973416584,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9569662633348493e-05,
      "loss": 0.3228,
      "step": 2522
    },
    {
      "epoch": 0.19094650205761318,
      "grad_norm": 0.84375,
      "learning_rate": 1.956931691501005e-05,
      "loss": 0.3192,
      "step": 2523
    },
    {
      "epoch": 0.1910221843810605,
      "grad_norm": 0.84375,
      "learning_rate": 1.9568971060914428e-05,
      "loss": 0.3422,
      "step": 2524
    },
    {
      "epoch": 0.19109786670450782,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9568625071066528e-05,
      "loss": 0.293,
      "step": 2525
    },
    {
      "epoch": 0.19117354902795516,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9568278945471264e-05,
      "loss": 0.2948,
      "step": 2526
    },
    {
      "epoch": 0.19124923135140248,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9567932684133543e-05,
      "loss": 0.3321,
      "step": 2527
    },
    {
      "epoch": 0.19132491367484983,
      "grad_norm": 1.0625,
      "learning_rate": 1.9567586287058274e-05,
      "loss": 0.3543,
      "step": 2528
    },
    {
      "epoch": 0.19140059599829715,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9567239754250376e-05,
      "loss": 0.3257,
      "step": 2529
    },
    {
      "epoch": 0.19147627832174446,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9566893085714765e-05,
      "loss": 0.3688,
      "step": 2530
    },
    {
      "epoch": 0.1915519606451918,
      "grad_norm": 0.85546875,
      "learning_rate": 1.956654628145636e-05,
      "loss": 0.3327,
      "step": 2531
    },
    {
      "epoch": 0.19162764296863913,
      "grad_norm": 1.9765625,
      "learning_rate": 1.9566199341480075e-05,
      "loss": 0.4945,
      "step": 2532
    },
    {
      "epoch": 0.19170332529208647,
      "grad_norm": 0.9296875,
      "learning_rate": 1.956585226579084e-05,
      "loss": 0.4076,
      "step": 2533
    },
    {
      "epoch": 0.1917790076155338,
      "grad_norm": 0.88671875,
      "learning_rate": 1.956550505439357e-05,
      "loss": 0.3353,
      "step": 2534
    },
    {
      "epoch": 0.19185468993898114,
      "grad_norm": 0.86328125,
      "learning_rate": 1.95651577072932e-05,
      "loss": 0.2984,
      "step": 2535
    },
    {
      "epoch": 0.19193037226242846,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9564810224494658e-05,
      "loss": 0.3903,
      "step": 2536
    },
    {
      "epoch": 0.19200605458587577,
      "grad_norm": 0.94921875,
      "learning_rate": 1.956446260600286e-05,
      "loss": 0.4078,
      "step": 2537
    },
    {
      "epoch": 0.19208173690932312,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9564114851822752e-05,
      "loss": 0.3382,
      "step": 2538
    },
    {
      "epoch": 0.19215741923277044,
      "grad_norm": 0.890625,
      "learning_rate": 1.956376696195926e-05,
      "loss": 0.3209,
      "step": 2539
    },
    {
      "epoch": 0.19223310155621778,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9563418936417324e-05,
      "loss": 0.343,
      "step": 2540
    },
    {
      "epoch": 0.1923087838796651,
      "grad_norm": 0.796875,
      "learning_rate": 1.956307077520188e-05,
      "loss": 0.305,
      "step": 2541
    },
    {
      "epoch": 0.19238446620311245,
      "grad_norm": 0.8828125,
      "learning_rate": 1.956272247831786e-05,
      "loss": 0.3358,
      "step": 2542
    },
    {
      "epoch": 0.19246014852655977,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9562374045770218e-05,
      "loss": 0.363,
      "step": 2543
    },
    {
      "epoch": 0.19253583085000708,
      "grad_norm": 1.921875,
      "learning_rate": 1.956202547756389e-05,
      "loss": 0.4409,
      "step": 2544
    },
    {
      "epoch": 0.19261151317345443,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9561676773703818e-05,
      "loss": 0.3382,
      "step": 2545
    },
    {
      "epoch": 0.19268719549690175,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9561327934194954e-05,
      "loss": 0.3609,
      "step": 2546
    },
    {
      "epoch": 0.1927628778203491,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9560978959042247e-05,
      "loss": 0.37,
      "step": 2547
    },
    {
      "epoch": 0.1928385601437964,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9560629848250645e-05,
      "loss": 0.3409,
      "step": 2548
    },
    {
      "epoch": 0.19291424246724376,
      "grad_norm": 0.859375,
      "learning_rate": 1.95602806018251e-05,
      "loss": 0.3301,
      "step": 2549
    },
    {
      "epoch": 0.19298992479069108,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9559931219770573e-05,
      "loss": 0.3503,
      "step": 2550
    },
    {
      "epoch": 0.1930656071141384,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9559581702092013e-05,
      "loss": 0.3171,
      "step": 2551
    },
    {
      "epoch": 0.19314128943758574,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9559232048794385e-05,
      "loss": 0.3317,
      "step": 2552
    },
    {
      "epoch": 0.19321697176103306,
      "grad_norm": 0.8984375,
      "learning_rate": 1.955888225988264e-05,
      "loss": 0.3262,
      "step": 2553
    },
    {
      "epoch": 0.1932926540844804,
      "grad_norm": 46.0,
      "learning_rate": 1.9558532335361753e-05,
      "loss": 0.4823,
      "step": 2554
    },
    {
      "epoch": 0.19336833640792772,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9558182275236677e-05,
      "loss": 0.3553,
      "step": 2555
    },
    {
      "epoch": 0.19344401873137507,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9557832079512387e-05,
      "loss": 0.3355,
      "step": 2556
    },
    {
      "epoch": 0.19351970105482239,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9557481748193847e-05,
      "loss": 0.3394,
      "step": 2557
    },
    {
      "epoch": 0.1935953833782697,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9557131281286022e-05,
      "loss": 0.3263,
      "step": 2558
    },
    {
      "epoch": 0.19367106570171705,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9556780678793895e-05,
      "loss": 0.3025,
      "step": 2559
    },
    {
      "epoch": 0.19374674802516437,
      "grad_norm": 0.7890625,
      "learning_rate": 1.9556429940722432e-05,
      "loss": 0.3139,
      "step": 2560
    },
    {
      "epoch": 0.1938224303486117,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9556079067076614e-05,
      "loss": 0.336,
      "step": 2561
    },
    {
      "epoch": 0.19389811267205903,
      "grad_norm": 0.83984375,
      "learning_rate": 1.955572805786141e-05,
      "loss": 0.291,
      "step": 2562
    },
    {
      "epoch": 0.19397379499550635,
      "grad_norm": 0.81640625,
      "learning_rate": 1.955537691308181e-05,
      "loss": 0.3077,
      "step": 2563
    },
    {
      "epoch": 0.1940494773189537,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9555025632742787e-05,
      "loss": 0.3052,
      "step": 2564
    },
    {
      "epoch": 0.194125159642401,
      "grad_norm": 0.9609375,
      "learning_rate": 1.9554674216849328e-05,
      "loss": 0.3395,
      "step": 2565
    },
    {
      "epoch": 0.19420084196584836,
      "grad_norm": 2.1875,
      "learning_rate": 1.9554322665406424e-05,
      "loss": 0.3954,
      "step": 2566
    },
    {
      "epoch": 0.19427652428929568,
      "grad_norm": 1.0,
      "learning_rate": 1.955397097841905e-05,
      "loss": 0.422,
      "step": 2567
    },
    {
      "epoch": 0.19435220661274302,
      "grad_norm": 0.9296875,
      "learning_rate": 1.955361915589221e-05,
      "loss": 0.2688,
      "step": 2568
    },
    {
      "epoch": 0.19442788893619034,
      "grad_norm": 0.84765625,
      "learning_rate": 1.955326719783088e-05,
      "loss": 0.3259,
      "step": 2569
    },
    {
      "epoch": 0.19450357125963766,
      "grad_norm": 0.921875,
      "learning_rate": 1.9552915104240067e-05,
      "loss": 0.3573,
      "step": 2570
    },
    {
      "epoch": 0.194579253583085,
      "grad_norm": 4.34375,
      "learning_rate": 1.9552562875124757e-05,
      "loss": 0.4564,
      "step": 2571
    },
    {
      "epoch": 0.19465493590653232,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9552210510489952e-05,
      "loss": 0.3671,
      "step": 2572
    },
    {
      "epoch": 0.19473061822997967,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9551858010340646e-05,
      "loss": 0.3275,
      "step": 2573
    },
    {
      "epoch": 0.194806300553427,
      "grad_norm": 0.9375,
      "learning_rate": 1.9551505374681844e-05,
      "loss": 0.3791,
      "step": 2574
    },
    {
      "epoch": 0.19488198287687433,
      "grad_norm": 0.8515625,
      "learning_rate": 1.955115260351855e-05,
      "loss": 0.3472,
      "step": 2575
    },
    {
      "epoch": 0.19495766520032165,
      "grad_norm": 1.9375,
      "learning_rate": 1.9550799696855764e-05,
      "loss": 0.4392,
      "step": 2576
    },
    {
      "epoch": 0.19503334752376897,
      "grad_norm": 0.89453125,
      "learning_rate": 1.955044665469849e-05,
      "loss": 0.3341,
      "step": 2577
    },
    {
      "epoch": 0.19510902984721631,
      "grad_norm": 2.015625,
      "learning_rate": 1.955009347705175e-05,
      "loss": 0.4825,
      "step": 2578
    },
    {
      "epoch": 0.19518471217066363,
      "grad_norm": 0.8671875,
      "learning_rate": 1.954974016392054e-05,
      "loss": 0.3481,
      "step": 2579
    },
    {
      "epoch": 0.19526039449411098,
      "grad_norm": 0.84375,
      "learning_rate": 1.9549386715309876e-05,
      "loss": 0.3355,
      "step": 2580
    },
    {
      "epoch": 0.1953360768175583,
      "grad_norm": 1.6171875,
      "learning_rate": 1.9549033131224776e-05,
      "loss": 0.4142,
      "step": 2581
    },
    {
      "epoch": 0.19541175914100564,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9548679411670254e-05,
      "loss": 0.3315,
      "step": 2582
    },
    {
      "epoch": 0.19548744146445296,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9548325556651332e-05,
      "loss": 0.3494,
      "step": 2583
    },
    {
      "epoch": 0.19556312378790028,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9547971566173025e-05,
      "loss": 0.3576,
      "step": 2584
    },
    {
      "epoch": 0.19563880611134762,
      "grad_norm": 0.921875,
      "learning_rate": 1.9547617440240355e-05,
      "loss": 0.3802,
      "step": 2585
    },
    {
      "epoch": 0.19571448843479494,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9547263178858346e-05,
      "loss": 0.329,
      "step": 2586
    },
    {
      "epoch": 0.1957901707582423,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9546908782032024e-05,
      "loss": 0.3255,
      "step": 2587
    },
    {
      "epoch": 0.1958658530816896,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9546554249766422e-05,
      "loss": 0.3391,
      "step": 2588
    },
    {
      "epoch": 0.19594153540513692,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9546199582066564e-05,
      "loss": 0.3091,
      "step": 2589
    },
    {
      "epoch": 0.19601721772858427,
      "grad_norm": 0.765625,
      "learning_rate": 1.9545844778937484e-05,
      "loss": 0.2647,
      "step": 2590
    },
    {
      "epoch": 0.1960929000520316,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9545489840384214e-05,
      "loss": 0.2956,
      "step": 2591
    },
    {
      "epoch": 0.19616858237547893,
      "grad_norm": 0.890625,
      "learning_rate": 1.9545134766411792e-05,
      "loss": 0.3383,
      "step": 2592
    },
    {
      "epoch": 0.19624426469892625,
      "grad_norm": 0.79296875,
      "learning_rate": 1.954477955702525e-05,
      "loss": 0.2926,
      "step": 2593
    },
    {
      "epoch": 0.1963199470223736,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9544424212229632e-05,
      "loss": 0.3104,
      "step": 2594
    },
    {
      "epoch": 0.19639562934582092,
      "grad_norm": 0.859375,
      "learning_rate": 1.9544068732029977e-05,
      "loss": 0.3223,
      "step": 2595
    },
    {
      "epoch": 0.19647131166926823,
      "grad_norm": 0.9296875,
      "learning_rate": 1.954371311643133e-05,
      "loss": 0.3824,
      "step": 2596
    },
    {
      "epoch": 0.19654699399271558,
      "grad_norm": 0.9375,
      "learning_rate": 1.9543357365438734e-05,
      "loss": 0.3471,
      "step": 2597
    },
    {
      "epoch": 0.1966226763161629,
      "grad_norm": 1.0078125,
      "learning_rate": 1.954300147905724e-05,
      "loss": 0.3817,
      "step": 2598
    },
    {
      "epoch": 0.19669835863961024,
      "grad_norm": 0.84375,
      "learning_rate": 1.954264545729189e-05,
      "loss": 0.3025,
      "step": 2599
    },
    {
      "epoch": 0.19677404096305756,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9542289300147735e-05,
      "loss": 0.3107,
      "step": 2600
    },
    {
      "epoch": 0.1968497232865049,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9541933007629836e-05,
      "loss": 0.3077,
      "step": 2601
    },
    {
      "epoch": 0.19692540560995223,
      "grad_norm": 0.90625,
      "learning_rate": 1.954157657974324e-05,
      "loss": 0.3553,
      "step": 2602
    },
    {
      "epoch": 0.19700108793339954,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9541220016493012e-05,
      "loss": 0.3579,
      "step": 2603
    },
    {
      "epoch": 0.1970767702568469,
      "grad_norm": 0.88671875,
      "learning_rate": 1.95408633178842e-05,
      "loss": 0.2861,
      "step": 2604
    },
    {
      "epoch": 0.1971524525802942,
      "grad_norm": 0.86328125,
      "learning_rate": 1.954050648392187e-05,
      "loss": 0.3559,
      "step": 2605
    },
    {
      "epoch": 0.19722813490374155,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9540149514611083e-05,
      "loss": 0.3268,
      "step": 2606
    },
    {
      "epoch": 0.19730381722718887,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9539792409956906e-05,
      "loss": 0.3512,
      "step": 2607
    },
    {
      "epoch": 0.19737949955063622,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9539435169964402e-05,
      "loss": 0.3554,
      "step": 2608
    },
    {
      "epoch": 0.19745518187408354,
      "grad_norm": 0.890625,
      "learning_rate": 1.9539077794638638e-05,
      "loss": 0.332,
      "step": 2609
    },
    {
      "epoch": 0.19753086419753085,
      "grad_norm": 0.890625,
      "learning_rate": 1.9538720283984687e-05,
      "loss": 0.3337,
      "step": 2610
    },
    {
      "epoch": 0.1976065465209782,
      "grad_norm": 0.8984375,
      "learning_rate": 1.953836263800762e-05,
      "loss": 0.4011,
      "step": 2611
    },
    {
      "epoch": 0.19768222884442552,
      "grad_norm": 0.89453125,
      "learning_rate": 1.953800485671251e-05,
      "loss": 0.3559,
      "step": 2612
    },
    {
      "epoch": 0.19775791116787286,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9537646940104433e-05,
      "loss": 0.3957,
      "step": 2613
    },
    {
      "epoch": 0.19783359349132018,
      "grad_norm": 0.78515625,
      "learning_rate": 1.953728888818847e-05,
      "loss": 0.3216,
      "step": 2614
    },
    {
      "epoch": 0.19790927581476753,
      "grad_norm": 0.796875,
      "learning_rate": 1.9536930700969693e-05,
      "loss": 0.3016,
      "step": 2615
    },
    {
      "epoch": 0.19798495813821484,
      "grad_norm": 0.9453125,
      "learning_rate": 1.9536572378453188e-05,
      "loss": 0.3831,
      "step": 2616
    },
    {
      "epoch": 0.19806064046166216,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9536213920644044e-05,
      "loss": 0.3693,
      "step": 2617
    },
    {
      "epoch": 0.1981363227851095,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9535855327547336e-05,
      "loss": 0.3452,
      "step": 2618
    },
    {
      "epoch": 0.19821200510855683,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9535496599168158e-05,
      "loss": 0.3081,
      "step": 2619
    },
    {
      "epoch": 0.19828768743200417,
      "grad_norm": 0.859375,
      "learning_rate": 1.9535137735511597e-05,
      "loss": 0.3555,
      "step": 2620
    },
    {
      "epoch": 0.1983633697554515,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9534778736582746e-05,
      "loss": 0.3538,
      "step": 2621
    },
    {
      "epoch": 0.1984390520788988,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9534419602386694e-05,
      "loss": 0.3272,
      "step": 2622
    },
    {
      "epoch": 0.19851473440234615,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9534060332928537e-05,
      "loss": 0.3652,
      "step": 2623
    },
    {
      "epoch": 0.19859041672579347,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9533700928213373e-05,
      "loss": 0.3381,
      "step": 2624
    },
    {
      "epoch": 0.19866609904924082,
      "grad_norm": 2.5,
      "learning_rate": 1.9533341388246307e-05,
      "loss": 0.4303,
      "step": 2625
    },
    {
      "epoch": 0.19874178137268814,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9532981713032432e-05,
      "loss": 0.3413,
      "step": 2626
    },
    {
      "epoch": 0.19881746369613548,
      "grad_norm": 0.8125,
      "learning_rate": 1.9532621902576846e-05,
      "loss": 0.3184,
      "step": 2627
    },
    {
      "epoch": 0.1988931460195828,
      "grad_norm": 1.671875,
      "learning_rate": 1.9532261956884663e-05,
      "loss": 0.4109,
      "step": 2628
    },
    {
      "epoch": 0.19896882834303012,
      "grad_norm": 0.8203125,
      "learning_rate": 1.953190187596099e-05,
      "loss": 0.3162,
      "step": 2629
    },
    {
      "epoch": 0.19904451066647746,
      "grad_norm": 0.875,
      "learning_rate": 1.9531541659810927e-05,
      "loss": 0.3483,
      "step": 2630
    },
    {
      "epoch": 0.19912019298992478,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9531181308439592e-05,
      "loss": 0.3378,
      "step": 2631
    },
    {
      "epoch": 0.19919587531337213,
      "grad_norm": 0.828125,
      "learning_rate": 1.9530820821852092e-05,
      "loss": 0.3358,
      "step": 2632
    },
    {
      "epoch": 0.19927155763681945,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9530460200053542e-05,
      "loss": 0.3374,
      "step": 2633
    },
    {
      "epoch": 0.1993472399602668,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9530099443049062e-05,
      "loss": 0.2955,
      "step": 2634
    },
    {
      "epoch": 0.1994229222837141,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9529738550843765e-05,
      "loss": 0.3652,
      "step": 2635
    },
    {
      "epoch": 0.19949860460716143,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9529377523442777e-05,
      "loss": 0.3652,
      "step": 2636
    },
    {
      "epoch": 0.19957428693060877,
      "grad_norm": 0.734375,
      "learning_rate": 1.952901636085121e-05,
      "loss": 0.2431,
      "step": 2637
    },
    {
      "epoch": 0.1996499692540561,
      "grad_norm": 0.87109375,
      "learning_rate": 1.95286550630742e-05,
      "loss": 0.3291,
      "step": 2638
    },
    {
      "epoch": 0.19972565157750344,
      "grad_norm": 0.8828125,
      "learning_rate": 1.952829363011686e-05,
      "loss": 0.3656,
      "step": 2639
    },
    {
      "epoch": 0.19980133390095076,
      "grad_norm": 0.875,
      "learning_rate": 1.9527932061984327e-05,
      "loss": 0.3445,
      "step": 2640
    },
    {
      "epoch": 0.1998770162243981,
      "grad_norm": 3.109375,
      "learning_rate": 1.9527570358681726e-05,
      "loss": 0.4393,
      "step": 2641
    },
    {
      "epoch": 0.19995269854784542,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9527208520214192e-05,
      "loss": 0.3503,
      "step": 2642
    },
    {
      "epoch": 0.20002838087129274,
      "grad_norm": 0.84375,
      "learning_rate": 1.9526846546586854e-05,
      "loss": 0.3433,
      "step": 2643
    },
    {
      "epoch": 0.20010406319474008,
      "grad_norm": 0.80078125,
      "learning_rate": 1.952648443780485e-05,
      "loss": 0.3107,
      "step": 2644
    },
    {
      "epoch": 0.20010406319474008,
      "eval_loss": 0.3552929759025574,
      "eval_runtime": 83.7012,
      "eval_samples_per_second": 58.088,
      "eval_steps_per_second": 58.088,
      "step": 2644
    },
    {
      "epoch": 0.2001797455181874,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9526122193873315e-05,
      "loss": 0.3504,
      "step": 2645
    },
    {
      "epoch": 0.20025542784163475,
      "grad_norm": 0.80859375,
      "learning_rate": 1.952575981479739e-05,
      "loss": 0.3325,
      "step": 2646
    },
    {
      "epoch": 0.20033111016508207,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9525397300582214e-05,
      "loss": 0.3324,
      "step": 2647
    },
    {
      "epoch": 0.2004067924885294,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9525034651232933e-05,
      "loss": 0.3457,
      "step": 2648
    },
    {
      "epoch": 0.20048247481197673,
      "grad_norm": 0.87890625,
      "learning_rate": 1.952467186675469e-05,
      "loss": 0.3437,
      "step": 2649
    },
    {
      "epoch": 0.20055815713542405,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9524308947152632e-05,
      "loss": 0.3169,
      "step": 2650
    },
    {
      "epoch": 0.2006338394588714,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9523945892431905e-05,
      "loss": 0.3294,
      "step": 2651
    },
    {
      "epoch": 0.2007095217823187,
      "grad_norm": 0.875,
      "learning_rate": 1.952358270259766e-05,
      "loss": 0.3564,
      "step": 2652
    },
    {
      "epoch": 0.20078520410576606,
      "grad_norm": 0.9375,
      "learning_rate": 1.9523219377655055e-05,
      "loss": 0.3715,
      "step": 2653
    },
    {
      "epoch": 0.20086088642921338,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9522855917609243e-05,
      "loss": 0.3254,
      "step": 2654
    },
    {
      "epoch": 0.2009365687526607,
      "grad_norm": 0.890625,
      "learning_rate": 1.9522492322465375e-05,
      "loss": 0.3669,
      "step": 2655
    },
    {
      "epoch": 0.20101225107610804,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9522128592228612e-05,
      "loss": 0.3727,
      "step": 2656
    },
    {
      "epoch": 0.20108793339955536,
      "grad_norm": 4.8125,
      "learning_rate": 1.9521764726904114e-05,
      "loss": 0.4355,
      "step": 2657
    },
    {
      "epoch": 0.2011636157230027,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9521400726497044e-05,
      "loss": 0.3144,
      "step": 2658
    },
    {
      "epoch": 0.20123929804645002,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9521036591012567e-05,
      "loss": 0.3609,
      "step": 2659
    },
    {
      "epoch": 0.20131498036989737,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9520672320455845e-05,
      "loss": 0.324,
      "step": 2660
    },
    {
      "epoch": 0.20139066269334469,
      "grad_norm": 0.89453125,
      "learning_rate": 1.952030791483205e-05,
      "loss": 0.3724,
      "step": 2661
    },
    {
      "epoch": 0.201466345016792,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9519943374146345e-05,
      "loss": 0.3338,
      "step": 2662
    },
    {
      "epoch": 0.20154202734023935,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9519578698403912e-05,
      "loss": 0.327,
      "step": 2663
    },
    {
      "epoch": 0.20161770966368667,
      "grad_norm": 0.875,
      "learning_rate": 1.9519213887609915e-05,
      "loss": 0.3598,
      "step": 2664
    },
    {
      "epoch": 0.201693391987134,
      "grad_norm": 0.875,
      "learning_rate": 1.951884894176954e-05,
      "loss": 0.3677,
      "step": 2665
    },
    {
      "epoch": 0.20176907431058133,
      "grad_norm": 0.8515625,
      "learning_rate": 1.951848386088795e-05,
      "loss": 0.3459,
      "step": 2666
    },
    {
      "epoch": 0.20184475663402868,
      "grad_norm": 0.7578125,
      "learning_rate": 1.951811864497034e-05,
      "loss": 0.2793,
      "step": 2667
    },
    {
      "epoch": 0.201920438957476,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9517753294021876e-05,
      "loss": 0.3488,
      "step": 2668
    },
    {
      "epoch": 0.2019961212809233,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9517387808047753e-05,
      "loss": 0.3511,
      "step": 2669
    },
    {
      "epoch": 0.20207180360437066,
      "grad_norm": 0.8515625,
      "learning_rate": 1.951702218705315e-05,
      "loss": 0.3237,
      "step": 2670
    },
    {
      "epoch": 0.20214748592781798,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9516656431043254e-05,
      "loss": 0.3373,
      "step": 2671
    },
    {
      "epoch": 0.20222316825126532,
      "grad_norm": 0.91796875,
      "learning_rate": 1.951629054002326e-05,
      "loss": 0.3325,
      "step": 2672
    },
    {
      "epoch": 0.20229885057471264,
      "grad_norm": 0.875,
      "learning_rate": 1.951592451399835e-05,
      "loss": 0.3649,
      "step": 2673
    },
    {
      "epoch": 0.20237453289816,
      "grad_norm": 0.890625,
      "learning_rate": 1.951555835297372e-05,
      "loss": 0.3574,
      "step": 2674
    },
    {
      "epoch": 0.2024502152216073,
      "grad_norm": 1.828125,
      "learning_rate": 1.951519205695457e-05,
      "loss": 0.4106,
      "step": 2675
    },
    {
      "epoch": 0.20252589754505462,
      "grad_norm": 7.3125,
      "learning_rate": 1.951482562594609e-05,
      "loss": 0.4183,
      "step": 2676
    },
    {
      "epoch": 0.20260157986850197,
      "grad_norm": 0.953125,
      "learning_rate": 1.951445905995348e-05,
      "loss": 0.3598,
      "step": 2677
    },
    {
      "epoch": 0.2026772621919493,
      "grad_norm": 0.88671875,
      "learning_rate": 1.951409235898194e-05,
      "loss": 0.3699,
      "step": 2678
    },
    {
      "epoch": 0.20275294451539663,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9513725523036676e-05,
      "loss": 0.3405,
      "step": 2679
    },
    {
      "epoch": 0.20282862683884395,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9513358552122886e-05,
      "loss": 0.3484,
      "step": 2680
    },
    {
      "epoch": 0.20290430916229127,
      "grad_norm": 0.8359375,
      "learning_rate": 1.951299144624578e-05,
      "loss": 0.2988,
      "step": 2681
    },
    {
      "epoch": 0.20297999148573861,
      "grad_norm": 4.53125,
      "learning_rate": 1.9512624205410568e-05,
      "loss": 0.349,
      "step": 2682
    },
    {
      "epoch": 0.20305567380918593,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9512256829622455e-05,
      "loss": 0.3541,
      "step": 2683
    },
    {
      "epoch": 0.20313135613263328,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9511889318886656e-05,
      "loss": 0.2991,
      "step": 2684
    },
    {
      "epoch": 0.2032070384560806,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9511521673208385e-05,
      "loss": 0.3746,
      "step": 2685
    },
    {
      "epoch": 0.20328272077952794,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9511153892592856e-05,
      "loss": 0.3986,
      "step": 2686
    },
    {
      "epoch": 0.20335840310297526,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9510785977045286e-05,
      "loss": 0.3635,
      "step": 2687
    },
    {
      "epoch": 0.20343408542642258,
      "grad_norm": 0.84375,
      "learning_rate": 1.9510417926570898e-05,
      "loss": 0.3508,
      "step": 2688
    },
    {
      "epoch": 0.20350976774986992,
      "grad_norm": 0.859375,
      "learning_rate": 1.9510049741174912e-05,
      "loss": 0.3332,
      "step": 2689
    },
    {
      "epoch": 0.20358545007331724,
      "grad_norm": 0.79296875,
      "learning_rate": 1.950968142086255e-05,
      "loss": 0.3192,
      "step": 2690
    },
    {
      "epoch": 0.2036611323967646,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9509312965639033e-05,
      "loss": 0.3199,
      "step": 2691
    },
    {
      "epoch": 0.2037368147202119,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9508944375509596e-05,
      "loss": 0.2966,
      "step": 2692
    },
    {
      "epoch": 0.20381249704365925,
      "grad_norm": 0.80078125,
      "learning_rate": 1.950857565047947e-05,
      "loss": 0.2825,
      "step": 2693
    },
    {
      "epoch": 0.20388817936710657,
      "grad_norm": 0.890625,
      "learning_rate": 1.9508206790553875e-05,
      "loss": 0.3665,
      "step": 2694
    },
    {
      "epoch": 0.2039638616905539,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9507837795738053e-05,
      "loss": 0.3618,
      "step": 2695
    },
    {
      "epoch": 0.20403954401400123,
      "grad_norm": 23.0,
      "learning_rate": 1.950746866603723e-05,
      "loss": 0.43,
      "step": 2696
    },
    {
      "epoch": 0.20411522633744855,
      "grad_norm": 0.953125,
      "learning_rate": 1.9507099401456658e-05,
      "loss": 0.3388,
      "step": 2697
    },
    {
      "epoch": 0.2041909086608959,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9506730002001562e-05,
      "loss": 0.3597,
      "step": 2698
    },
    {
      "epoch": 0.20426659098434322,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9506360467677184e-05,
      "loss": 0.3031,
      "step": 2699
    },
    {
      "epoch": 0.20434227330779056,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9505990798488773e-05,
      "loss": 0.3436,
      "step": 2700
    },
    {
      "epoch": 0.20441795563123788,
      "grad_norm": 0.8359375,
      "learning_rate": 1.950562099444157e-05,
      "loss": 0.3542,
      "step": 2701
    },
    {
      "epoch": 0.2044936379546852,
      "grad_norm": 1.9921875,
      "learning_rate": 1.9505251055540818e-05,
      "loss": 0.4631,
      "step": 2702
    },
    {
      "epoch": 0.20456932027813254,
      "grad_norm": 0.96875,
      "learning_rate": 1.950488098179177e-05,
      "loss": 0.3537,
      "step": 2703
    },
    {
      "epoch": 0.20464500260157986,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9504510773199677e-05,
      "loss": 0.3728,
      "step": 2704
    },
    {
      "epoch": 0.2047206849250272,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9504140429769784e-05,
      "loss": 0.3366,
      "step": 2705
    },
    {
      "epoch": 0.20479636724847453,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9503769951507353e-05,
      "loss": 0.3234,
      "step": 2706
    },
    {
      "epoch": 0.20487204957192187,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9503399338417632e-05,
      "loss": 0.3801,
      "step": 2707
    },
    {
      "epoch": 0.2049477318953692,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9503028590505885e-05,
      "loss": 0.3373,
      "step": 2708
    },
    {
      "epoch": 0.2050234142188165,
      "grad_norm": 0.796875,
      "learning_rate": 1.950265770777737e-05,
      "loss": 0.2803,
      "step": 2709
    },
    {
      "epoch": 0.20509909654226385,
      "grad_norm": 0.78125,
      "learning_rate": 1.950228669023735e-05,
      "loss": 0.2775,
      "step": 2710
    },
    {
      "epoch": 0.20517477886571117,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9501915537891083e-05,
      "loss": 0.3188,
      "step": 2711
    },
    {
      "epoch": 0.20525046118915852,
      "grad_norm": 1.71875,
      "learning_rate": 1.950154425074384e-05,
      "loss": 0.4247,
      "step": 2712
    },
    {
      "epoch": 0.20532614351260584,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9501172828800887e-05,
      "loss": 0.3294,
      "step": 2713
    },
    {
      "epoch": 0.20540182583605315,
      "grad_norm": 0.83203125,
      "learning_rate": 1.950080127206749e-05,
      "loss": 0.3295,
      "step": 2714
    },
    {
      "epoch": 0.2054775081595005,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9500429580548928e-05,
      "loss": 0.34,
      "step": 2715
    },
    {
      "epoch": 0.20555319048294782,
      "grad_norm": 1.0078125,
      "learning_rate": 1.9500057754250465e-05,
      "loss": 0.3232,
      "step": 2716
    },
    {
      "epoch": 0.20562887280639516,
      "grad_norm": 0.90625,
      "learning_rate": 1.9499685793177384e-05,
      "loss": 0.3572,
      "step": 2717
    },
    {
      "epoch": 0.20570455512984248,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9499313697334955e-05,
      "loss": 0.3392,
      "step": 2718
    },
    {
      "epoch": 0.20578023745328983,
      "grad_norm": 0.90625,
      "learning_rate": 1.9498941466728462e-05,
      "loss": 0.3522,
      "step": 2719
    },
    {
      "epoch": 0.20585591977673715,
      "grad_norm": 0.7890625,
      "learning_rate": 1.949856910136318e-05,
      "loss": 0.2953,
      "step": 2720
    },
    {
      "epoch": 0.20593160210018446,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9498196601244398e-05,
      "loss": 0.3228,
      "step": 2721
    },
    {
      "epoch": 0.2060072844236318,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9497823966377397e-05,
      "loss": 0.3073,
      "step": 2722
    },
    {
      "epoch": 0.20608296674707913,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9497451196767465e-05,
      "loss": 0.3365,
      "step": 2723
    },
    {
      "epoch": 0.20615864907052647,
      "grad_norm": 0.86328125,
      "learning_rate": 1.949707829241989e-05,
      "loss": 0.3315,
      "step": 2724
    },
    {
      "epoch": 0.2062343313939738,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9496705253339958e-05,
      "loss": 0.3384,
      "step": 2725
    },
    {
      "epoch": 0.20631001371742114,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9496332079532968e-05,
      "loss": 0.3452,
      "step": 2726
    },
    {
      "epoch": 0.20638569604086845,
      "grad_norm": 0.7890625,
      "learning_rate": 1.9495958771004215e-05,
      "loss": 0.3139,
      "step": 2727
    },
    {
      "epoch": 0.20646137836431577,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9495585327758985e-05,
      "loss": 0.3321,
      "step": 2728
    },
    {
      "epoch": 0.20653706068776312,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9495211749802583e-05,
      "loss": 0.3779,
      "step": 2729
    },
    {
      "epoch": 0.20661274301121044,
      "grad_norm": 0.89453125,
      "learning_rate": 1.949483803714031e-05,
      "loss": 0.4007,
      "step": 2730
    },
    {
      "epoch": 0.20668842533465778,
      "grad_norm": 2.4375,
      "learning_rate": 1.9494464189777463e-05,
      "loss": 0.4518,
      "step": 2731
    },
    {
      "epoch": 0.2067641076581051,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9494090207719353e-05,
      "loss": 0.3282,
      "step": 2732
    },
    {
      "epoch": 0.20683978998155245,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9493716090971277e-05,
      "loss": 0.3097,
      "step": 2733
    },
    {
      "epoch": 0.20691547230499976,
      "grad_norm": 0.82421875,
      "learning_rate": 1.949334183953855e-05,
      "loss": 0.3476,
      "step": 2734
    },
    {
      "epoch": 0.20699115462844708,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9492967453426475e-05,
      "loss": 0.3658,
      "step": 2735
    },
    {
      "epoch": 0.20706683695189443,
      "grad_norm": 0.90625,
      "learning_rate": 1.9492592932640367e-05,
      "loss": 0.3491,
      "step": 2736
    },
    {
      "epoch": 0.20714251927534175,
      "grad_norm": 0.82421875,
      "learning_rate": 1.949221827718554e-05,
      "loss": 0.2957,
      "step": 2737
    },
    {
      "epoch": 0.2072182015987891,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9491843487067305e-05,
      "loss": 0.3583,
      "step": 2738
    },
    {
      "epoch": 0.2072938839222364,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9491468562290984e-05,
      "loss": 0.3528,
      "step": 2739
    },
    {
      "epoch": 0.20736956624568376,
      "grad_norm": 0.81640625,
      "learning_rate": 1.949109350286189e-05,
      "loss": 0.2679,
      "step": 2740
    },
    {
      "epoch": 0.20744524856913107,
      "grad_norm": 0.91796875,
      "learning_rate": 1.949071830878535e-05,
      "loss": 0.4096,
      "step": 2741
    },
    {
      "epoch": 0.2075209308925784,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9490342980066685e-05,
      "loss": 0.343,
      "step": 2742
    },
    {
      "epoch": 0.20759661321602574,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9489967516711217e-05,
      "loss": 0.3962,
      "step": 2743
    },
    {
      "epoch": 0.20767229553947306,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9489591918724274e-05,
      "loss": 0.3504,
      "step": 2744
    },
    {
      "epoch": 0.2077479778629204,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9489216186111187e-05,
      "loss": 0.3293,
      "step": 2745
    },
    {
      "epoch": 0.20782366018636772,
      "grad_norm": 0.8828125,
      "learning_rate": 1.948884031887728e-05,
      "loss": 0.3751,
      "step": 2746
    },
    {
      "epoch": 0.20789934250981504,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9488464317027894e-05,
      "loss": 0.3276,
      "step": 2747
    },
    {
      "epoch": 0.20797502483326238,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9488088180568357e-05,
      "loss": 0.3288,
      "step": 2748
    },
    {
      "epoch": 0.2080507071567097,
      "grad_norm": 0.84375,
      "learning_rate": 1.948771190950401e-05,
      "loss": 0.3506,
      "step": 2749
    },
    {
      "epoch": 0.20812638948015705,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9487335503840186e-05,
      "loss": 0.3503,
      "step": 2750
    },
    {
      "epoch": 0.20820207180360437,
      "grad_norm": 0.796875,
      "learning_rate": 1.9486958963582228e-05,
      "loss": 0.3139,
      "step": 2751
    },
    {
      "epoch": 0.2082777541270517,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9486582288735475e-05,
      "loss": 0.3209,
      "step": 2752
    },
    {
      "epoch": 0.20835343645049903,
      "grad_norm": 0.78125,
      "learning_rate": 1.9486205479305272e-05,
      "loss": 0.2994,
      "step": 2753
    },
    {
      "epoch": 0.20842911877394635,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9485828535296968e-05,
      "loss": 0.3183,
      "step": 2754
    },
    {
      "epoch": 0.2085048010973937,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9485451456715907e-05,
      "loss": 0.3296,
      "step": 2755
    },
    {
      "epoch": 0.208580483420841,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9485074243567437e-05,
      "loss": 0.3491,
      "step": 2756
    },
    {
      "epoch": 0.20865616574428836,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9484696895856915e-05,
      "loss": 0.2951,
      "step": 2757
    },
    {
      "epoch": 0.20873184806773568,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9484319413589692e-05,
      "loss": 0.3356,
      "step": 2758
    },
    {
      "epoch": 0.20880753039118302,
      "grad_norm": 0.80859375,
      "learning_rate": 1.948394179677112e-05,
      "loss": 0.3359,
      "step": 2759
    },
    {
      "epoch": 0.20888321271463034,
      "grad_norm": 0.8125,
      "learning_rate": 1.9483564045406562e-05,
      "loss": 0.3284,
      "step": 2760
    },
    {
      "epoch": 0.20895889503807766,
      "grad_norm": 0.87890625,
      "learning_rate": 1.948318615950137e-05,
      "loss": 0.3597,
      "step": 2761
    },
    {
      "epoch": 0.209034577361525,
      "grad_norm": 0.8671875,
      "learning_rate": 1.948280813906091e-05,
      "loss": 0.3833,
      "step": 2762
    },
    {
      "epoch": 0.20911025968497232,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9482429984090543e-05,
      "loss": 0.3479,
      "step": 2763
    },
    {
      "epoch": 0.20918594200841967,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9482051694595637e-05,
      "loss": 0.3308,
      "step": 2764
    },
    {
      "epoch": 0.20926162433186699,
      "grad_norm": 0.83984375,
      "learning_rate": 1.948167327058155e-05,
      "loss": 0.3179,
      "step": 2765
    },
    {
      "epoch": 0.20933730665531433,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9481294712053663e-05,
      "loss": 0.3557,
      "step": 2766
    },
    {
      "epoch": 0.20941298897876165,
      "grad_norm": 0.859375,
      "learning_rate": 1.948091601901734e-05,
      "loss": 0.337,
      "step": 2767
    },
    {
      "epoch": 0.20948867130220897,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9480537191477946e-05,
      "loss": 0.3975,
      "step": 2768
    },
    {
      "epoch": 0.2095643536256563,
      "grad_norm": 0.796875,
      "learning_rate": 1.9480158229440868e-05,
      "loss": 0.3276,
      "step": 2769
    },
    {
      "epoch": 0.20964003594910363,
      "grad_norm": 0.77734375,
      "learning_rate": 1.9479779132911476e-05,
      "loss": 0.2988,
      "step": 2770
    },
    {
      "epoch": 0.20971571827255098,
      "grad_norm": 1.9609375,
      "learning_rate": 1.9479399901895152e-05,
      "loss": 0.4358,
      "step": 2771
    },
    {
      "epoch": 0.2097914005959983,
      "grad_norm": 0.84765625,
      "learning_rate": 1.947902053639727e-05,
      "loss": 0.3239,
      "step": 2772
    },
    {
      "epoch": 0.2098670829194456,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9478641036423215e-05,
      "loss": 0.3458,
      "step": 2773
    },
    {
      "epoch": 0.20994276524289296,
      "grad_norm": 0.86328125,
      "learning_rate": 1.947826140197837e-05,
      "loss": 0.3527,
      "step": 2774
    },
    {
      "epoch": 0.21001844756634028,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9477881633068125e-05,
      "loss": 0.3936,
      "step": 2775
    },
    {
      "epoch": 0.21009412988978762,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9477501729697865e-05,
      "loss": 0.3341,
      "step": 2776
    },
    {
      "epoch": 0.21016981221323494,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9477121691872975e-05,
      "loss": 0.3709,
      "step": 2777
    },
    {
      "epoch": 0.2102454945366823,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9476741519598855e-05,
      "loss": 0.3259,
      "step": 2778
    },
    {
      "epoch": 0.2103211768601296,
      "grad_norm": 0.8671875,
      "learning_rate": 1.947636121288089e-05,
      "loss": 0.3516,
      "step": 2779
    },
    {
      "epoch": 0.21039685918357692,
      "grad_norm": 0.8515625,
      "learning_rate": 1.947598077172448e-05,
      "loss": 0.2957,
      "step": 2780
    },
    {
      "epoch": 0.21047254150702427,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9475600196135023e-05,
      "loss": 0.3117,
      "step": 2781
    },
    {
      "epoch": 0.2105482238304716,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9475219486117916e-05,
      "loss": 0.3559,
      "step": 2782
    },
    {
      "epoch": 0.21062390615391893,
      "grad_norm": 0.92578125,
      "learning_rate": 1.947483864167856e-05,
      "loss": 0.3443,
      "step": 2783
    },
    {
      "epoch": 0.21069958847736625,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9474457662822357e-05,
      "loss": 0.3761,
      "step": 2784
    },
    {
      "epoch": 0.2107752708008136,
      "grad_norm": 1.640625,
      "learning_rate": 1.9474076549554717e-05,
      "loss": 0.4228,
      "step": 2785
    },
    {
      "epoch": 0.21085095312426091,
      "grad_norm": 0.796875,
      "learning_rate": 1.947369530188104e-05,
      "loss": 0.2995,
      "step": 2786
    },
    {
      "epoch": 0.21092663544770823,
      "grad_norm": 0.890625,
      "learning_rate": 1.947331391980674e-05,
      "loss": 0.3483,
      "step": 2787
    },
    {
      "epoch": 0.21100231777115558,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9472932403337224e-05,
      "loss": 0.3742,
      "step": 2788
    },
    {
      "epoch": 0.2110780000946029,
      "grad_norm": 0.859375,
      "learning_rate": 1.9472550752477904e-05,
      "loss": 0.3219,
      "step": 2789
    },
    {
      "epoch": 0.21115368241805024,
      "grad_norm": 0.81640625,
      "learning_rate": 1.94721689672342e-05,
      "loss": 0.2972,
      "step": 2790
    },
    {
      "epoch": 0.21122936474149756,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9471787047611522e-05,
      "loss": 0.3038,
      "step": 2791
    },
    {
      "epoch": 0.2113050470649449,
      "grad_norm": 0.83984375,
      "learning_rate": 1.947140499361529e-05,
      "loss": 0.335,
      "step": 2792
    },
    {
      "epoch": 0.21138072938839222,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9471022805250927e-05,
      "loss": 0.3017,
      "step": 2793
    },
    {
      "epoch": 0.21145641171183954,
      "grad_norm": 0.83984375,
      "learning_rate": 1.947064048252385e-05,
      "loss": 0.3293,
      "step": 2794
    },
    {
      "epoch": 0.2115320940352869,
      "grad_norm": 0.859375,
      "learning_rate": 1.9470258025439487e-05,
      "loss": 0.305,
      "step": 2795
    },
    {
      "epoch": 0.2116077763587342,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9469875434003263e-05,
      "loss": 0.3707,
      "step": 2796
    },
    {
      "epoch": 0.21168345868218155,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9469492708220604e-05,
      "loss": 0.312,
      "step": 2797
    },
    {
      "epoch": 0.21175914100562887,
      "grad_norm": 0.84375,
      "learning_rate": 1.946910984809694e-05,
      "loss": 0.3129,
      "step": 2798
    },
    {
      "epoch": 0.21183482332907622,
      "grad_norm": 0.828125,
      "learning_rate": 1.9468726853637704e-05,
      "loss": 0.3124,
      "step": 2799
    },
    {
      "epoch": 0.21191050565252353,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9468343724848328e-05,
      "loss": 0.2974,
      "step": 2800
    },
    {
      "epoch": 0.21198618797597085,
      "grad_norm": 0.890625,
      "learning_rate": 1.9467960461734252e-05,
      "loss": 0.3595,
      "step": 2801
    },
    {
      "epoch": 0.2120618702994182,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9467577064300904e-05,
      "loss": 0.3622,
      "step": 2802
    },
    {
      "epoch": 0.21213755262286552,
      "grad_norm": 0.8828125,
      "learning_rate": 1.946719353255373e-05,
      "loss": 0.3608,
      "step": 2803
    },
    {
      "epoch": 0.21221323494631286,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9466809866498174e-05,
      "loss": 0.3654,
      "step": 2804
    },
    {
      "epoch": 0.21228891726976018,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9466426066139668e-05,
      "loss": 0.351,
      "step": 2805
    },
    {
      "epoch": 0.2123645995932075,
      "grad_norm": 0.875,
      "learning_rate": 1.9466042131483665e-05,
      "loss": 0.3478,
      "step": 2806
    },
    {
      "epoch": 0.21244028191665484,
      "grad_norm": 0.765625,
      "learning_rate": 1.9465658062535613e-05,
      "loss": 0.303,
      "step": 2807
    },
    {
      "epoch": 0.21251596424010216,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9465273859300957e-05,
      "loss": 0.3034,
      "step": 2808
    },
    {
      "epoch": 0.2125916465635495,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9464889521785147e-05,
      "loss": 0.3316,
      "step": 2809
    },
    {
      "epoch": 0.21266732888699683,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9464505049993636e-05,
      "loss": 0.303,
      "step": 2810
    },
    {
      "epoch": 0.21274301121044417,
      "grad_norm": 0.9296875,
      "learning_rate": 1.946412044393188e-05,
      "loss": 0.382,
      "step": 2811
    },
    {
      "epoch": 0.2128186935338915,
      "grad_norm": 0.828125,
      "learning_rate": 1.9463735703605336e-05,
      "loss": 0.3161,
      "step": 2812
    },
    {
      "epoch": 0.2128943758573388,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9463350829019457e-05,
      "loss": 0.3021,
      "step": 2813
    },
    {
      "epoch": 0.21297005818078615,
      "grad_norm": 5.40625,
      "learning_rate": 1.946296582017971e-05,
      "loss": 0.504,
      "step": 2814
    },
    {
      "epoch": 0.21304574050423347,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9462580677091552e-05,
      "loss": 0.3307,
      "step": 2815
    },
    {
      "epoch": 0.21312142282768082,
      "grad_norm": 0.85546875,
      "learning_rate": 1.946219539976045e-05,
      "loss": 0.3378,
      "step": 2816
    },
    {
      "epoch": 0.21319710515112814,
      "grad_norm": 0.80859375,
      "learning_rate": 1.946180998819187e-05,
      "loss": 0.3363,
      "step": 2817
    },
    {
      "epoch": 0.21327278747457548,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9461424442391277e-05,
      "loss": 0.3529,
      "step": 2818
    },
    {
      "epoch": 0.2133484697980228,
      "grad_norm": 0.953125,
      "learning_rate": 1.9461038762364143e-05,
      "loss": 0.4024,
      "step": 2819
    },
    {
      "epoch": 0.21342415212147012,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9460652948115935e-05,
      "loss": 0.3875,
      "step": 2820
    },
    {
      "epoch": 0.21349983444491746,
      "grad_norm": 0.78515625,
      "learning_rate": 1.9460266999652134e-05,
      "loss": 0.2989,
      "step": 2821
    },
    {
      "epoch": 0.21357551676836478,
      "grad_norm": 0.8125,
      "learning_rate": 1.945988091697821e-05,
      "loss": 0.3107,
      "step": 2822
    },
    {
      "epoch": 0.21365119909181213,
      "grad_norm": 0.8671875,
      "learning_rate": 1.945949470009964e-05,
      "loss": 0.3692,
      "step": 2823
    },
    {
      "epoch": 0.21372688141525945,
      "grad_norm": 0.80859375,
      "learning_rate": 1.94591083490219e-05,
      "loss": 0.2946,
      "step": 2824
    },
    {
      "epoch": 0.2138025637387068,
      "grad_norm": 0.921875,
      "learning_rate": 1.9458721863750486e-05,
      "loss": 0.3641,
      "step": 2825
    },
    {
      "epoch": 0.2138782460621541,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9458335244290864e-05,
      "loss": 0.2998,
      "step": 2826
    },
    {
      "epoch": 0.21395392838560143,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9457948490648527e-05,
      "loss": 0.3384,
      "step": 2827
    },
    {
      "epoch": 0.21402961070904877,
      "grad_norm": 0.81640625,
      "learning_rate": 1.945756160282896e-05,
      "loss": 0.3342,
      "step": 2828
    },
    {
      "epoch": 0.2141052930324961,
      "grad_norm": 0.97265625,
      "learning_rate": 1.945717458083765e-05,
      "loss": 0.4023,
      "step": 2829
    },
    {
      "epoch": 0.21418097535594344,
      "grad_norm": 0.84375,
      "learning_rate": 1.9456787424680095e-05,
      "loss": 0.3104,
      "step": 2830
    },
    {
      "epoch": 0.21425665767939076,
      "grad_norm": 0.859375,
      "learning_rate": 1.9456400134361776e-05,
      "loss": 0.3843,
      "step": 2831
    },
    {
      "epoch": 0.2143323400028381,
      "grad_norm": 0.75390625,
      "learning_rate": 1.94560127098882e-05,
      "loss": 0.292,
      "step": 2832
    },
    {
      "epoch": 0.21440802232628542,
      "grad_norm": 1.9453125,
      "learning_rate": 1.945562515126485e-05,
      "loss": 0.4858,
      "step": 2833
    },
    {
      "epoch": 0.21448370464973274,
      "grad_norm": 0.875,
      "learning_rate": 1.9455237458497234e-05,
      "loss": 0.3816,
      "step": 2834
    },
    {
      "epoch": 0.21455938697318008,
      "grad_norm": 0.83203125,
      "learning_rate": 1.945484963159085e-05,
      "loss": 0.3189,
      "step": 2835
    },
    {
      "epoch": 0.2146350692966274,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9454461670551198e-05,
      "loss": 0.3753,
      "step": 2836
    },
    {
      "epoch": 0.21471075162007475,
      "grad_norm": 0.875,
      "learning_rate": 1.945407357538378e-05,
      "loss": 0.3785,
      "step": 2837
    },
    {
      "epoch": 0.21478643394352206,
      "grad_norm": 0.87890625,
      "learning_rate": 1.945368534609411e-05,
      "loss": 0.3805,
      "step": 2838
    },
    {
      "epoch": 0.21486211626696938,
      "grad_norm": 0.8125,
      "learning_rate": 1.9453296982687684e-05,
      "loss": 0.35,
      "step": 2839
    },
    {
      "epoch": 0.21493779859041673,
      "grad_norm": 0.8359375,
      "learning_rate": 1.945290848517002e-05,
      "loss": 0.3572,
      "step": 2840
    },
    {
      "epoch": 0.21501348091386405,
      "grad_norm": 0.87890625,
      "learning_rate": 1.945251985354663e-05,
      "loss": 0.3547,
      "step": 2841
    },
    {
      "epoch": 0.2150891632373114,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9452131087823023e-05,
      "loss": 0.347,
      "step": 2842
    },
    {
      "epoch": 0.2151648455607587,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9451742188004716e-05,
      "loss": 0.2777,
      "step": 2843
    },
    {
      "epoch": 0.21524052788420606,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9451353154097224e-05,
      "loss": 0.3432,
      "step": 2844
    },
    {
      "epoch": 0.21531621020765337,
      "grad_norm": 0.78515625,
      "learning_rate": 1.9450963986106073e-05,
      "loss": 0.2817,
      "step": 2845
    },
    {
      "epoch": 0.2153918925311007,
      "grad_norm": 0.87109375,
      "learning_rate": 1.945057468403678e-05,
      "loss": 0.3614,
      "step": 2846
    },
    {
      "epoch": 0.21546757485454804,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9450185247894862e-05,
      "loss": 0.3192,
      "step": 2847
    },
    {
      "epoch": 0.21554325717799536,
      "grad_norm": 1.765625,
      "learning_rate": 1.944979567768585e-05,
      "loss": 0.4436,
      "step": 2848
    },
    {
      "epoch": 0.2156189395014427,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9449405973415275e-05,
      "loss": 0.3315,
      "step": 2849
    },
    {
      "epoch": 0.21569462182489002,
      "grad_norm": 0.78515625,
      "learning_rate": 1.9449016135088657e-05,
      "loss": 0.3222,
      "step": 2850
    },
    {
      "epoch": 0.21577030414833737,
      "grad_norm": 0.84765625,
      "learning_rate": 1.944862616271153e-05,
      "loss": 0.3548,
      "step": 2851
    },
    {
      "epoch": 0.21584598647178468,
      "grad_norm": 0.90625,
      "learning_rate": 1.9448236056289425e-05,
      "loss": 0.3592,
      "step": 2852
    },
    {
      "epoch": 0.215921668795232,
      "grad_norm": 0.875,
      "learning_rate": 1.944784581582788e-05,
      "loss": 0.3471,
      "step": 2853
    },
    {
      "epoch": 0.21599735111867935,
      "grad_norm": 0.796875,
      "learning_rate": 1.9447455441332433e-05,
      "loss": 0.3217,
      "step": 2854
    },
    {
      "epoch": 0.21607303344212667,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9447064932808613e-05,
      "loss": 0.3613,
      "step": 2855
    },
    {
      "epoch": 0.216148715765574,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9446674290261965e-05,
      "loss": 0.3202,
      "step": 2856
    },
    {
      "epoch": 0.21622439808902133,
      "grad_norm": 0.78515625,
      "learning_rate": 1.9446283513698033e-05,
      "loss": 0.3103,
      "step": 2857
    },
    {
      "epoch": 0.21630008041246868,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9445892603122357e-05,
      "loss": 0.3874,
      "step": 2858
    },
    {
      "epoch": 0.216375762735916,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9445501558540486e-05,
      "loss": 0.3196,
      "step": 2859
    },
    {
      "epoch": 0.2164514450593633,
      "grad_norm": 0.88671875,
      "learning_rate": 1.944511037995797e-05,
      "loss": 0.3536,
      "step": 2860
    },
    {
      "epoch": 0.21652712738281066,
      "grad_norm": 0.796875,
      "learning_rate": 1.944471906738035e-05,
      "loss": 0.3316,
      "step": 2861
    },
    {
      "epoch": 0.21660280970625798,
      "grad_norm": 0.87890625,
      "learning_rate": 1.944432762081318e-05,
      "loss": 0.3852,
      "step": 2862
    },
    {
      "epoch": 0.21667849202970532,
      "grad_norm": 0.8203125,
      "learning_rate": 1.944393604026202e-05,
      "loss": 0.3189,
      "step": 2863
    },
    {
      "epoch": 0.21675417435315264,
      "grad_norm": 0.87890625,
      "learning_rate": 1.944354432573242e-05,
      "loss": 0.3863,
      "step": 2864
    },
    {
      "epoch": 0.21682985667659996,
      "grad_norm": 0.83984375,
      "learning_rate": 1.944315247722994e-05,
      "loss": 0.3255,
      "step": 2865
    },
    {
      "epoch": 0.2169055390000473,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9442760494760137e-05,
      "loss": 0.3024,
      "step": 2866
    },
    {
      "epoch": 0.21698122132349462,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9442368378328568e-05,
      "loss": 0.3417,
      "step": 2867
    },
    {
      "epoch": 0.21705690364694197,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9441976127940802e-05,
      "loss": 0.355,
      "step": 2868
    },
    {
      "epoch": 0.21713258597038929,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9441583743602402e-05,
      "loss": 0.3825,
      "step": 2869
    },
    {
      "epoch": 0.21720826829383663,
      "grad_norm": 0.9453125,
      "learning_rate": 1.9441191225318934e-05,
      "loss": 0.3859,
      "step": 2870
    },
    {
      "epoch": 0.21728395061728395,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9440798573095966e-05,
      "loss": 0.3451,
      "step": 2871
    },
    {
      "epoch": 0.21735963294073127,
      "grad_norm": 0.8203125,
      "learning_rate": 1.944040578693907e-05,
      "loss": 0.2982,
      "step": 2872
    },
    {
      "epoch": 0.2174353152641786,
      "grad_norm": 0.82421875,
      "learning_rate": 1.944001286685382e-05,
      "loss": 0.319,
      "step": 2873
    },
    {
      "epoch": 0.21751099758762593,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9439619812845784e-05,
      "loss": 0.3063,
      "step": 2874
    },
    {
      "epoch": 0.21758667991107328,
      "grad_norm": 0.828125,
      "learning_rate": 1.9439226624920545e-05,
      "loss": 0.3407,
      "step": 2875
    },
    {
      "epoch": 0.2176623622345206,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9438833303083677e-05,
      "loss": 0.3238,
      "step": 2876
    },
    {
      "epoch": 0.21773804455796794,
      "grad_norm": 4.21875,
      "learning_rate": 1.943843984734076e-05,
      "loss": 0.3447,
      "step": 2877
    },
    {
      "epoch": 0.21781372688141526,
      "grad_norm": 0.828125,
      "learning_rate": 1.943804625769738e-05,
      "loss": 0.3335,
      "step": 2878
    },
    {
      "epoch": 0.21788940920486258,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9437652534159116e-05,
      "loss": 0.326,
      "step": 2879
    },
    {
      "epoch": 0.21796509152830992,
      "grad_norm": 0.84375,
      "learning_rate": 1.9437258676731555e-05,
      "loss": 0.3608,
      "step": 2880
    },
    {
      "epoch": 0.21804077385175724,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9436864685420282e-05,
      "loss": 0.3542,
      "step": 2881
    },
    {
      "epoch": 0.2181164561752046,
      "grad_norm": 0.78125,
      "learning_rate": 1.9436470560230892e-05,
      "loss": 0.2975,
      "step": 2882
    },
    {
      "epoch": 0.2181921384986519,
      "grad_norm": 0.7734375,
      "learning_rate": 1.943607630116897e-05,
      "loss": 0.309,
      "step": 2883
    },
    {
      "epoch": 0.21826782082209925,
      "grad_norm": 0.875,
      "learning_rate": 1.943568190824012e-05,
      "loss": 0.3637,
      "step": 2884
    },
    {
      "epoch": 0.21834350314554657,
      "grad_norm": 0.875,
      "learning_rate": 1.9435287381449925e-05,
      "loss": 0.3572,
      "step": 2885
    },
    {
      "epoch": 0.2184191854689939,
      "grad_norm": 0.82421875,
      "learning_rate": 1.943489272080399e-05,
      "loss": 0.3361,
      "step": 2886
    },
    {
      "epoch": 0.21849486779244123,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9434497926307906e-05,
      "loss": 0.3071,
      "step": 2887
    },
    {
      "epoch": 0.21857055011588855,
      "grad_norm": 0.84375,
      "learning_rate": 1.9434102997967283e-05,
      "loss": 0.3674,
      "step": 2888
    },
    {
      "epoch": 0.2186462324393359,
      "grad_norm": 0.84375,
      "learning_rate": 1.9433707935787715e-05,
      "loss": 0.2999,
      "step": 2889
    },
    {
      "epoch": 0.21872191476278321,
      "grad_norm": 0.84375,
      "learning_rate": 1.9433312739774814e-05,
      "loss": 0.3143,
      "step": 2890
    },
    {
      "epoch": 0.21879759708623056,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9432917409934187e-05,
      "loss": 0.3313,
      "step": 2891
    },
    {
      "epoch": 0.21887327940967788,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9432521946271434e-05,
      "loss": 0.341,
      "step": 2892
    },
    {
      "epoch": 0.2189489617331252,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9432126348792173e-05,
      "loss": 0.3025,
      "step": 2893
    },
    {
      "epoch": 0.21902464405657254,
      "grad_norm": 0.8125,
      "learning_rate": 1.9431730617502012e-05,
      "loss": 0.3094,
      "step": 2894
    },
    {
      "epoch": 0.21910032638001986,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9431334752406568e-05,
      "loss": 0.3456,
      "step": 2895
    },
    {
      "epoch": 0.2191760087034672,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9430938753511458e-05,
      "loss": 0.3698,
      "step": 2896
    },
    {
      "epoch": 0.21925169102691452,
      "grad_norm": 0.7578125,
      "learning_rate": 1.9430542620822293e-05,
      "loss": 0.2786,
      "step": 2897
    },
    {
      "epoch": 0.21932737335036184,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9430146354344698e-05,
      "loss": 0.3789,
      "step": 2898
    },
    {
      "epoch": 0.2194030556738092,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9429749954084297e-05,
      "loss": 0.3098,
      "step": 2899
    },
    {
      "epoch": 0.2194787379972565,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9429353420046714e-05,
      "loss": 0.3424,
      "step": 2900
    },
    {
      "epoch": 0.21955442032070385,
      "grad_norm": 0.7890625,
      "learning_rate": 1.9428956752237563e-05,
      "loss": 0.3176,
      "step": 2901
    },
    {
      "epoch": 0.21963010264415117,
      "grad_norm": 2.390625,
      "learning_rate": 1.9428559950662485e-05,
      "loss": 0.5257,
      "step": 2902
    },
    {
      "epoch": 0.21970578496759852,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9428163015327105e-05,
      "loss": 0.3423,
      "step": 2903
    },
    {
      "epoch": 0.21978146729104583,
      "grad_norm": 0.84765625,
      "learning_rate": 1.942776594623705e-05,
      "loss": 0.3526,
      "step": 2904
    },
    {
      "epoch": 0.21985714961449315,
      "grad_norm": 0.87890625,
      "learning_rate": 1.942736874339796e-05,
      "loss": 0.382,
      "step": 2905
    },
    {
      "epoch": 0.2199328319379405,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9426971406815464e-05,
      "loss": 0.3217,
      "step": 2906
    },
    {
      "epoch": 0.22000851426138782,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9426573936495203e-05,
      "loss": 0.3134,
      "step": 2907
    },
    {
      "epoch": 0.22008419658483516,
      "grad_norm": 2.1875,
      "learning_rate": 1.942617633244281e-05,
      "loss": 0.4572,
      "step": 2908
    },
    {
      "epoch": 0.22015987890828248,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9425778594663935e-05,
      "loss": 0.3651,
      "step": 2909
    },
    {
      "epoch": 0.22023556123172983,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9425380723164212e-05,
      "loss": 0.3591,
      "step": 2910
    },
    {
      "epoch": 0.22031124355517714,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9424982717949292e-05,
      "loss": 0.3167,
      "step": 2911
    },
    {
      "epoch": 0.22038692587862446,
      "grad_norm": 1.6015625,
      "learning_rate": 1.9424584579024818e-05,
      "loss": 0.4549,
      "step": 2912
    },
    {
      "epoch": 0.2204626082020718,
      "grad_norm": 0.859375,
      "learning_rate": 1.9424186306396436e-05,
      "loss": 0.3395,
      "step": 2913
    },
    {
      "epoch": 0.22053829052551913,
      "grad_norm": 0.828125,
      "learning_rate": 1.9423787900069803e-05,
      "loss": 0.3348,
      "step": 2914
    },
    {
      "epoch": 0.22061397284896647,
      "grad_norm": 1.375,
      "learning_rate": 1.9423389360050563e-05,
      "loss": 0.3892,
      "step": 2915
    },
    {
      "epoch": 0.2206896551724138,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9422990686344373e-05,
      "loss": 0.3645,
      "step": 2916
    },
    {
      "epoch": 0.22076533749586114,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9422591878956892e-05,
      "loss": 0.3324,
      "step": 2917
    },
    {
      "epoch": 0.22084101981930845,
      "grad_norm": 0.890625,
      "learning_rate": 1.9422192937893775e-05,
      "loss": 0.3557,
      "step": 2918
    },
    {
      "epoch": 0.22091670214275577,
      "grad_norm": 0.84375,
      "learning_rate": 1.942179386316068e-05,
      "loss": 0.3062,
      "step": 2919
    },
    {
      "epoch": 0.22099238446620312,
      "grad_norm": 2.234375,
      "learning_rate": 1.9421394654763275e-05,
      "loss": 0.3846,
      "step": 2920
    },
    {
      "epoch": 0.22106806678965044,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9420995312707216e-05,
      "loss": 0.3159,
      "step": 2921
    },
    {
      "epoch": 0.22114374911309778,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9420595836998173e-05,
      "loss": 0.3628,
      "step": 2922
    },
    {
      "epoch": 0.2212194314365451,
      "grad_norm": 0.79296875,
      "learning_rate": 1.942019622764181e-05,
      "loss": 0.3312,
      "step": 2923
    },
    {
      "epoch": 0.22129511375999245,
      "grad_norm": 0.8671875,
      "learning_rate": 1.94197964846438e-05,
      "loss": 0.3469,
      "step": 2924
    },
    {
      "epoch": 0.22137079608343976,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9419396608009812e-05,
      "loss": 0.3569,
      "step": 2925
    },
    {
      "epoch": 0.22144647840688708,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9418996597745517e-05,
      "loss": 0.3282,
      "step": 2926
    },
    {
      "epoch": 0.22152216073033443,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9418596453856594e-05,
      "loss": 0.331,
      "step": 2927
    },
    {
      "epoch": 0.22159784305378175,
      "grad_norm": 0.859375,
      "learning_rate": 1.9418196176348715e-05,
      "loss": 0.3459,
      "step": 2928
    },
    {
      "epoch": 0.2216735253772291,
      "grad_norm": 0.84375,
      "learning_rate": 1.9417795765227564e-05,
      "loss": 0.3563,
      "step": 2929
    },
    {
      "epoch": 0.2217492077006764,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9417395220498815e-05,
      "loss": 0.3693,
      "step": 2930
    },
    {
      "epoch": 0.22182489002412373,
      "grad_norm": 0.89453125,
      "learning_rate": 1.941699454216816e-05,
      "loss": 0.3288,
      "step": 2931
    },
    {
      "epoch": 0.22190057234757107,
      "grad_norm": 0.796875,
      "learning_rate": 1.941659373024127e-05,
      "loss": 0.3167,
      "step": 2932
    },
    {
      "epoch": 0.2219762546710184,
      "grad_norm": 0.96484375,
      "learning_rate": 1.941619278472384e-05,
      "loss": 0.3902,
      "step": 2933
    },
    {
      "epoch": 0.22205193699446574,
      "grad_norm": 0.875,
      "learning_rate": 1.941579170562156e-05,
      "loss": 0.3536,
      "step": 2934
    },
    {
      "epoch": 0.22212761931791306,
      "grad_norm": 0.76953125,
      "learning_rate": 1.9415390492940114e-05,
      "loss": 0.2858,
      "step": 2935
    },
    {
      "epoch": 0.2222033016413604,
      "grad_norm": 0.77734375,
      "learning_rate": 1.9414989146685194e-05,
      "loss": 0.2908,
      "step": 2936
    },
    {
      "epoch": 0.22227898396480772,
      "grad_norm": 0.83203125,
      "learning_rate": 1.94145876668625e-05,
      "loss": 0.3242,
      "step": 2937
    },
    {
      "epoch": 0.22235466628825504,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9414186053477726e-05,
      "loss": 0.3886,
      "step": 2938
    },
    {
      "epoch": 0.22243034861170238,
      "grad_norm": 0.80859375,
      "learning_rate": 1.941378430653656e-05,
      "loss": 0.3158,
      "step": 2939
    },
    {
      "epoch": 0.2225060309351497,
      "grad_norm": 0.796875,
      "learning_rate": 1.9413382426044718e-05,
      "loss": 0.3173,
      "step": 2940
    },
    {
      "epoch": 0.22258171325859705,
      "grad_norm": 0.75390625,
      "learning_rate": 1.941298041200789e-05,
      "loss": 0.2701,
      "step": 2941
    },
    {
      "epoch": 0.22265739558204437,
      "grad_norm": 0.75,
      "learning_rate": 1.9412578264431778e-05,
      "loss": 0.2459,
      "step": 2942
    },
    {
      "epoch": 0.2227330779054917,
      "grad_norm": 0.765625,
      "learning_rate": 1.9412175983322092e-05,
      "loss": 0.2812,
      "step": 2943
    },
    {
      "epoch": 0.22280876022893903,
      "grad_norm": 0.89453125,
      "learning_rate": 1.941177356868454e-05,
      "loss": 0.3824,
      "step": 2944
    },
    {
      "epoch": 0.22288444255238635,
      "grad_norm": 2.28125,
      "learning_rate": 1.9411371020524827e-05,
      "loss": 0.3879,
      "step": 2945
    },
    {
      "epoch": 0.2229601248758337,
      "grad_norm": 1.6953125,
      "learning_rate": 1.9410968338848666e-05,
      "loss": 0.3164,
      "step": 2946
    },
    {
      "epoch": 0.223035807199281,
      "grad_norm": 1.8125,
      "learning_rate": 1.941056552366177e-05,
      "loss": 0.4709,
      "step": 2947
    },
    {
      "epoch": 0.22311148952272836,
      "grad_norm": 0.828125,
      "learning_rate": 1.941016257496985e-05,
      "loss": 0.3331,
      "step": 2948
    },
    {
      "epoch": 0.22318717184617567,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9409759492778632e-05,
      "loss": 0.3356,
      "step": 2949
    },
    {
      "epoch": 0.22326285416962302,
      "grad_norm": 0.87109375,
      "learning_rate": 1.940935627709382e-05,
      "loss": 0.3667,
      "step": 2950
    },
    {
      "epoch": 0.22333853649307034,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9408952927921147e-05,
      "loss": 0.3412,
      "step": 2951
    },
    {
      "epoch": 0.22341421881651766,
      "grad_norm": 0.77734375,
      "learning_rate": 1.9408549445266328e-05,
      "loss": 0.2948,
      "step": 2952
    },
    {
      "epoch": 0.223489901139965,
      "grad_norm": 0.8359375,
      "learning_rate": 1.940814582913509e-05,
      "loss": 0.3358,
      "step": 2953
    },
    {
      "epoch": 0.22356558346341232,
      "grad_norm": 1.0234375,
      "learning_rate": 1.940774207953316e-05,
      "loss": 0.3356,
      "step": 2954
    },
    {
      "epoch": 0.22364126578685967,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9407338196466267e-05,
      "loss": 0.3091,
      "step": 2955
    },
    {
      "epoch": 0.22371694811030698,
      "grad_norm": 0.76171875,
      "learning_rate": 1.9406934179940134e-05,
      "loss": 0.2805,
      "step": 2956
    },
    {
      "epoch": 0.2237926304337543,
      "grad_norm": 1.015625,
      "learning_rate": 1.9406530029960494e-05,
      "loss": 0.3896,
      "step": 2957
    },
    {
      "epoch": 0.22386831275720165,
      "grad_norm": 0.83984375,
      "learning_rate": 1.940612574653309e-05,
      "loss": 0.3444,
      "step": 2958
    },
    {
      "epoch": 0.22394399508064897,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9405721329663648e-05,
      "loss": 0.3667,
      "step": 2959
    },
    {
      "epoch": 0.2240196774040963,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9405316779357904e-05,
      "loss": 0.3124,
      "step": 2960
    },
    {
      "epoch": 0.22409535972754363,
      "grad_norm": 0.890625,
      "learning_rate": 1.9404912095621608e-05,
      "loss": 0.3399,
      "step": 2961
    },
    {
      "epoch": 0.22417104205099098,
      "grad_norm": 0.921875,
      "learning_rate": 1.9404507278460492e-05,
      "loss": 0.3637,
      "step": 2962
    },
    {
      "epoch": 0.2242467243744383,
      "grad_norm": 0.84765625,
      "learning_rate": 1.94041023278803e-05,
      "loss": 0.3262,
      "step": 2963
    },
    {
      "epoch": 0.2243224066978856,
      "grad_norm": 0.87890625,
      "learning_rate": 1.940369724388678e-05,
      "loss": 0.3711,
      "step": 2964
    },
    {
      "epoch": 0.22439808902133296,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9403292026485676e-05,
      "loss": 0.3407,
      "step": 2965
    },
    {
      "epoch": 0.22447377134478028,
      "grad_norm": 0.91015625,
      "learning_rate": 1.940288667568274e-05,
      "loss": 0.3464,
      "step": 2966
    },
    {
      "epoch": 0.22454945366822762,
      "grad_norm": 0.859375,
      "learning_rate": 1.940248119148372e-05,
      "loss": 0.3579,
      "step": 2967
    },
    {
      "epoch": 0.22462513599167494,
      "grad_norm": 0.80859375,
      "learning_rate": 1.940207557389437e-05,
      "loss": 0.3119,
      "step": 2968
    },
    {
      "epoch": 0.22470081831512229,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9401669822920443e-05,
      "loss": 0.3279,
      "step": 2969
    },
    {
      "epoch": 0.2247765006385696,
      "grad_norm": 0.97265625,
      "learning_rate": 1.9401263938567694e-05,
      "loss": 0.3342,
      "step": 2970
    },
    {
      "epoch": 0.22485218296201692,
      "grad_norm": 0.8125,
      "learning_rate": 1.9400857920841885e-05,
      "loss": 0.3059,
      "step": 2971
    },
    {
      "epoch": 0.22492786528546427,
      "grad_norm": 0.8515625,
      "learning_rate": 1.940045176974877e-05,
      "loss": 0.3551,
      "step": 2972
    },
    {
      "epoch": 0.22500354760891159,
      "grad_norm": 0.84375,
      "learning_rate": 1.9400045485294117e-05,
      "loss": 0.3185,
      "step": 2973
    },
    {
      "epoch": 0.22507922993235893,
      "grad_norm": 0.83203125,
      "learning_rate": 1.939963906748369e-05,
      "loss": 0.3607,
      "step": 2974
    },
    {
      "epoch": 0.22515491225580625,
      "grad_norm": 0.80078125,
      "learning_rate": 1.939923251632325e-05,
      "loss": 0.2988,
      "step": 2975
    },
    {
      "epoch": 0.2252305945792536,
      "grad_norm": 0.73828125,
      "learning_rate": 1.939882583181857e-05,
      "loss": 0.2861,
      "step": 2976
    },
    {
      "epoch": 0.2253062769027009,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9398419013975416e-05,
      "loss": 0.3372,
      "step": 2977
    },
    {
      "epoch": 0.22538195922614823,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9398012062799557e-05,
      "loss": 0.3148,
      "step": 2978
    },
    {
      "epoch": 0.22545764154959558,
      "grad_norm": 0.796875,
      "learning_rate": 1.9397604978296772e-05,
      "loss": 0.3086,
      "step": 2979
    },
    {
      "epoch": 0.2255333238730429,
      "grad_norm": 0.82421875,
      "learning_rate": 1.939719776047283e-05,
      "loss": 0.3142,
      "step": 2980
    },
    {
      "epoch": 0.22560900619649024,
      "grad_norm": 2.078125,
      "learning_rate": 1.939679040933352e-05,
      "loss": 0.395,
      "step": 2981
    },
    {
      "epoch": 0.22568468851993756,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9396382924884606e-05,
      "loss": 0.4044,
      "step": 2982
    },
    {
      "epoch": 0.2257603708433849,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9395975307131873e-05,
      "loss": 0.3515,
      "step": 2983
    },
    {
      "epoch": 0.22583605316683222,
      "grad_norm": 0.828125,
      "learning_rate": 1.9395567556081112e-05,
      "loss": 0.2972,
      "step": 2984
    },
    {
      "epoch": 0.22591173549027954,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9395159671738098e-05,
      "loss": 0.3364,
      "step": 2985
    },
    {
      "epoch": 0.2259874178137269,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9394751654108622e-05,
      "loss": 0.3558,
      "step": 2986
    },
    {
      "epoch": 0.2260631001371742,
      "grad_norm": 0.765625,
      "learning_rate": 1.9394343503198475e-05,
      "loss": 0.2814,
      "step": 2987
    },
    {
      "epoch": 0.22613878246062155,
      "grad_norm": 0.78515625,
      "learning_rate": 1.9393935219013443e-05,
      "loss": 0.3095,
      "step": 2988
    },
    {
      "epoch": 0.22621446478406887,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9393526801559317e-05,
      "loss": 0.3393,
      "step": 2989
    },
    {
      "epoch": 0.2262901471075162,
      "grad_norm": 0.796875,
      "learning_rate": 1.9393118250841897e-05,
      "loss": 0.2999,
      "step": 2990
    },
    {
      "epoch": 0.22636582943096353,
      "grad_norm": 0.875,
      "learning_rate": 1.939270956686697e-05,
      "loss": 0.354,
      "step": 2991
    },
    {
      "epoch": 0.22644151175441085,
      "grad_norm": 1.0625,
      "learning_rate": 1.9392300749640344e-05,
      "loss": 0.3611,
      "step": 2992
    },
    {
      "epoch": 0.2265171940778582,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9391891799167812e-05,
      "loss": 0.3444,
      "step": 2993
    },
    {
      "epoch": 0.22659287640130552,
      "grad_norm": 0.859375,
      "learning_rate": 1.9391482715455178e-05,
      "loss": 0.3554,
      "step": 2994
    },
    {
      "epoch": 0.22666855872475286,
      "grad_norm": 0.7890625,
      "learning_rate": 1.939107349850825e-05,
      "loss": 0.3063,
      "step": 2995
    },
    {
      "epoch": 0.22674424104820018,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9390664148332822e-05,
      "loss": 0.34,
      "step": 2996
    },
    {
      "epoch": 0.2268199233716475,
      "grad_norm": 0.91015625,
      "learning_rate": 1.939025466493471e-05,
      "loss": 0.3533,
      "step": 2997
    },
    {
      "epoch": 0.22689560569509484,
      "grad_norm": 0.8828125,
      "learning_rate": 1.938984504831972e-05,
      "loss": 0.3537,
      "step": 2998
    },
    {
      "epoch": 0.22697128801854216,
      "grad_norm": 0.82421875,
      "learning_rate": 1.938943529849367e-05,
      "loss": 0.3221,
      "step": 2999
    },
    {
      "epoch": 0.2270469703419895,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9389025415462365e-05,
      "loss": 0.384,
      "step": 3000
    },
    {
      "epoch": 0.22712265266543682,
      "grad_norm": 0.9140625,
      "learning_rate": 1.938861539923162e-05,
      "loss": 0.3534,
      "step": 3001
    },
    {
      "epoch": 0.22719833498888417,
      "grad_norm": 0.828125,
      "learning_rate": 1.9388205249807254e-05,
      "loss": 0.3456,
      "step": 3002
    },
    {
      "epoch": 0.2272740173123315,
      "grad_norm": 0.85546875,
      "learning_rate": 1.938779496719509e-05,
      "loss": 0.3623,
      "step": 3003
    },
    {
      "epoch": 0.2273496996357788,
      "grad_norm": 0.84375,
      "learning_rate": 1.9387384551400942e-05,
      "loss": 0.3412,
      "step": 3004
    },
    {
      "epoch": 0.22742538195922615,
      "grad_norm": 0.80078125,
      "learning_rate": 1.938697400243063e-05,
      "loss": 0.3099,
      "step": 3005
    },
    {
      "epoch": 0.22750106428267347,
      "grad_norm": 3.65625,
      "learning_rate": 1.9386563320289992e-05,
      "loss": 0.5281,
      "step": 3006
    },
    {
      "epoch": 0.22757674660612082,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9386152504984837e-05,
      "loss": 0.3221,
      "step": 3007
    },
    {
      "epoch": 0.22765242892956813,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9385741556521008e-05,
      "loss": 0.3335,
      "step": 3008
    },
    {
      "epoch": 0.22772811125301548,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9385330474904327e-05,
      "loss": 0.331,
      "step": 3009
    },
    {
      "epoch": 0.2278037935764628,
      "grad_norm": 0.95703125,
      "learning_rate": 1.9384919260140627e-05,
      "loss": 0.405,
      "step": 3010
    },
    {
      "epoch": 0.22787947589991012,
      "grad_norm": 0.7265625,
      "learning_rate": 1.938450791223574e-05,
      "loss": 0.2806,
      "step": 3011
    },
    {
      "epoch": 0.22795515822335746,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9384096431195507e-05,
      "loss": 0.3312,
      "step": 3012
    },
    {
      "epoch": 0.22803084054680478,
      "grad_norm": 0.9765625,
      "learning_rate": 1.938368481702576e-05,
      "loss": 0.3591,
      "step": 3013
    },
    {
      "epoch": 0.22810652287025213,
      "grad_norm": 0.765625,
      "learning_rate": 1.9383273069732344e-05,
      "loss": 0.2844,
      "step": 3014
    },
    {
      "epoch": 0.22818220519369944,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9382861189321093e-05,
      "loss": 0.3485,
      "step": 3015
    },
    {
      "epoch": 0.2282578875171468,
      "grad_norm": 0.78515625,
      "learning_rate": 1.9382449175797856e-05,
      "loss": 0.3119,
      "step": 3016
    },
    {
      "epoch": 0.2283335698405941,
      "grad_norm": 0.87890625,
      "learning_rate": 1.938203702916848e-05,
      "loss": 0.342,
      "step": 3017
    },
    {
      "epoch": 0.22840925216404143,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9381624749438807e-05,
      "loss": 0.3159,
      "step": 3018
    },
    {
      "epoch": 0.22848493448748877,
      "grad_norm": 1.6171875,
      "learning_rate": 1.9381212336614684e-05,
      "loss": 0.4216,
      "step": 3019
    },
    {
      "epoch": 0.2285606168109361,
      "grad_norm": 0.80859375,
      "learning_rate": 1.938079979070197e-05,
      "loss": 0.3404,
      "step": 3020
    },
    {
      "epoch": 0.22863629913438344,
      "grad_norm": 0.86328125,
      "learning_rate": 1.938038711170651e-05,
      "loss": 0.3744,
      "step": 3021
    },
    {
      "epoch": 0.22871198145783075,
      "grad_norm": 0.796875,
      "learning_rate": 1.9379974299634164e-05,
      "loss": 0.2862,
      "step": 3022
    },
    {
      "epoch": 0.22878766378127807,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9379561354490783e-05,
      "loss": 0.3999,
      "step": 3023
    },
    {
      "epoch": 0.22886334610472542,
      "grad_norm": 0.859375,
      "learning_rate": 1.9379148276282232e-05,
      "loss": 0.3139,
      "step": 3024
    },
    {
      "epoch": 0.22893902842817274,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9378735065014363e-05,
      "loss": 0.3284,
      "step": 3025
    },
    {
      "epoch": 0.22901471075162008,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9378321720693045e-05,
      "loss": 0.3196,
      "step": 3026
    },
    {
      "epoch": 0.2290903930750674,
      "grad_norm": 0.84375,
      "learning_rate": 1.937790824332414e-05,
      "loss": 0.3596,
      "step": 3027
    },
    {
      "epoch": 0.22916607539851475,
      "grad_norm": 0.76171875,
      "learning_rate": 1.9377494632913514e-05,
      "loss": 0.3137,
      "step": 3028
    },
    {
      "epoch": 0.22924175772196206,
      "grad_norm": 0.86328125,
      "learning_rate": 1.937708088946703e-05,
      "loss": 0.3389,
      "step": 3029
    },
    {
      "epoch": 0.22931744004540938,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9376667012990568e-05,
      "loss": 0.3112,
      "step": 3030
    },
    {
      "epoch": 0.22939312236885673,
      "grad_norm": 0.828125,
      "learning_rate": 1.937625300348999e-05,
      "loss": 0.3355,
      "step": 3031
    },
    {
      "epoch": 0.22946880469230405,
      "grad_norm": 0.8984375,
      "learning_rate": 1.937583886097117e-05,
      "loss": 0.3616,
      "step": 3032
    },
    {
      "epoch": 0.2295444870157514,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9375424585439994e-05,
      "loss": 0.361,
      "step": 3033
    },
    {
      "epoch": 0.2296201693391987,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9375010176902327e-05,
      "loss": 0.3297,
      "step": 3034
    },
    {
      "epoch": 0.22969585166264606,
      "grad_norm": 0.78125,
      "learning_rate": 1.9374595635364054e-05,
      "loss": 0.2809,
      "step": 3035
    },
    {
      "epoch": 0.22977153398609337,
      "grad_norm": 0.84375,
      "learning_rate": 1.937418096083105e-05,
      "loss": 0.3509,
      "step": 3036
    },
    {
      "epoch": 0.2298472163095407,
      "grad_norm": 0.83203125,
      "learning_rate": 1.937376615330921e-05,
      "loss": 0.36,
      "step": 3037
    },
    {
      "epoch": 0.22992289863298804,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9373351212804406e-05,
      "loss": 0.3244,
      "step": 3038
    },
    {
      "epoch": 0.22999858095643536,
      "grad_norm": 0.85546875,
      "learning_rate": 1.937293613932253e-05,
      "loss": 0.3138,
      "step": 3039
    },
    {
      "epoch": 0.2300742632798827,
      "grad_norm": 0.7890625,
      "learning_rate": 1.9372520932869473e-05,
      "loss": 0.3,
      "step": 3040
    },
    {
      "epoch": 0.23014994560333002,
      "grad_norm": 0.84765625,
      "learning_rate": 1.937210559345112e-05,
      "loss": 0.3597,
      "step": 3041
    },
    {
      "epoch": 0.23022562792677737,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9371690121073367e-05,
      "loss": 0.3701,
      "step": 3042
    },
    {
      "epoch": 0.23030131025022468,
      "grad_norm": 0.91015625,
      "learning_rate": 1.937127451574211e-05,
      "loss": 0.4122,
      "step": 3043
    },
    {
      "epoch": 0.230376992573672,
      "grad_norm": 0.94140625,
      "learning_rate": 1.937085877746324e-05,
      "loss": 0.3689,
      "step": 3044
    },
    {
      "epoch": 0.23045267489711935,
      "grad_norm": 0.78125,
      "learning_rate": 1.9370442906242656e-05,
      "loss": 0.2947,
      "step": 3045
    },
    {
      "epoch": 0.23052835722056667,
      "grad_norm": 4.28125,
      "learning_rate": 1.937002690208626e-05,
      "loss": 0.4513,
      "step": 3046
    },
    {
      "epoch": 0.230604039544014,
      "grad_norm": 0.74609375,
      "learning_rate": 1.936961076499995e-05,
      "loss": 0.2834,
      "step": 3047
    },
    {
      "epoch": 0.23067972186746133,
      "grad_norm": 0.7890625,
      "learning_rate": 1.9369194494989637e-05,
      "loss": 0.3012,
      "step": 3048
    },
    {
      "epoch": 0.23075540419090865,
      "grad_norm": 0.765625,
      "learning_rate": 1.9368778092061222e-05,
      "loss": 0.295,
      "step": 3049
    },
    {
      "epoch": 0.230831086514356,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9368361556220613e-05,
      "loss": 0.314,
      "step": 3050
    },
    {
      "epoch": 0.2309067688378033,
      "grad_norm": 0.77734375,
      "learning_rate": 1.9367944887473714e-05,
      "loss": 0.277,
      "step": 3051
    },
    {
      "epoch": 0.23098245116125066,
      "grad_norm": 0.84375,
      "learning_rate": 1.9367528085826444e-05,
      "loss": 0.3636,
      "step": 3052
    },
    {
      "epoch": 0.23105813348469798,
      "grad_norm": 0.8125,
      "learning_rate": 1.9367111151284712e-05,
      "loss": 0.3133,
      "step": 3053
    },
    {
      "epoch": 0.23113381580814532,
      "grad_norm": 0.75390625,
      "learning_rate": 1.9366694083854435e-05,
      "loss": 0.2733,
      "step": 3054
    },
    {
      "epoch": 0.23120949813159264,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9366276883541526e-05,
      "loss": 0.3161,
      "step": 3055
    },
    {
      "epoch": 0.23128518045503996,
      "grad_norm": 0.76171875,
      "learning_rate": 1.9365859550351907e-05,
      "loss": 0.2775,
      "step": 3056
    },
    {
      "epoch": 0.2313608627784873,
      "grad_norm": 0.74609375,
      "learning_rate": 1.9365442084291502e-05,
      "loss": 0.245,
      "step": 3057
    },
    {
      "epoch": 0.23143654510193462,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9365024485366227e-05,
      "loss": 0.3872,
      "step": 3058
    },
    {
      "epoch": 0.23151222742538197,
      "grad_norm": 0.83203125,
      "learning_rate": 1.936460675358201e-05,
      "loss": 0.3497,
      "step": 3059
    },
    {
      "epoch": 0.23158790974882928,
      "grad_norm": 0.83203125,
      "learning_rate": 1.936418888894477e-05,
      "loss": 0.3253,
      "step": 3060
    },
    {
      "epoch": 0.23166359207227663,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9363770891460446e-05,
      "loss": 0.3892,
      "step": 3061
    },
    {
      "epoch": 0.23173927439572395,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9363352761134964e-05,
      "loss": 0.3099,
      "step": 3062
    },
    {
      "epoch": 0.23181495671917127,
      "grad_norm": 0.8515625,
      "learning_rate": 1.936293449797425e-05,
      "loss": 0.3306,
      "step": 3063
    },
    {
      "epoch": 0.2318906390426186,
      "grad_norm": 0.921875,
      "learning_rate": 1.9362516101984248e-05,
      "loss": 0.3576,
      "step": 3064
    },
    {
      "epoch": 0.23196632136606593,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9362097573170886e-05,
      "loss": 0.3507,
      "step": 3065
    },
    {
      "epoch": 0.23204200368951328,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9361678911540105e-05,
      "loss": 0.3539,
      "step": 3066
    },
    {
      "epoch": 0.2321176860129606,
      "grad_norm": 0.80859375,
      "learning_rate": 1.936126011709784e-05,
      "loss": 0.3258,
      "step": 3067
    },
    {
      "epoch": 0.23219336833640794,
      "grad_norm": 0.99609375,
      "learning_rate": 1.9360841189850043e-05,
      "loss": 0.3087,
      "step": 3068
    },
    {
      "epoch": 0.23226905065985526,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9360422129802644e-05,
      "loss": 0.3569,
      "step": 3069
    },
    {
      "epoch": 0.23234473298330258,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9360002936961598e-05,
      "loss": 0.3252,
      "step": 3070
    },
    {
      "epoch": 0.23242041530674992,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9359583611332847e-05,
      "loss": 0.3331,
      "step": 3071
    },
    {
      "epoch": 0.23249609763019724,
      "grad_norm": 0.8125,
      "learning_rate": 1.935916415292234e-05,
      "loss": 0.32,
      "step": 3072
    },
    {
      "epoch": 0.2325717799536446,
      "grad_norm": 0.8125,
      "learning_rate": 1.9358744561736027e-05,
      "loss": 0.3128,
      "step": 3073
    },
    {
      "epoch": 0.2326474622770919,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9358324837779864e-05,
      "loss": 0.3622,
      "step": 3074
    },
    {
      "epoch": 0.23272314460053925,
      "grad_norm": 0.7734375,
      "learning_rate": 1.93579049810598e-05,
      "loss": 0.285,
      "step": 3075
    },
    {
      "epoch": 0.23279882692398657,
      "grad_norm": 0.8203125,
      "learning_rate": 1.93574849915818e-05,
      "loss": 0.3189,
      "step": 3076
    },
    {
      "epoch": 0.2328745092474339,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9357064869351814e-05,
      "loss": 0.3676,
      "step": 3077
    },
    {
      "epoch": 0.23295019157088123,
      "grad_norm": 0.8203125,
      "learning_rate": 1.935664461437581e-05,
      "loss": 0.3194,
      "step": 3078
    },
    {
      "epoch": 0.23302587389432855,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9356224226659744e-05,
      "loss": 0.3332,
      "step": 3079
    },
    {
      "epoch": 0.2331015562177759,
      "grad_norm": 0.85546875,
      "learning_rate": 1.935580370620958e-05,
      "loss": 0.3653,
      "step": 3080
    },
    {
      "epoch": 0.23317723854122321,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9355383053031284e-05,
      "loss": 0.3542,
      "step": 3081
    },
    {
      "epoch": 0.23325292086467053,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9354962267130827e-05,
      "loss": 0.3234,
      "step": 3082
    },
    {
      "epoch": 0.23332860318811788,
      "grad_norm": 0.7890625,
      "learning_rate": 1.9354541348514177e-05,
      "loss": 0.3087,
      "step": 3083
    },
    {
      "epoch": 0.2334042855115652,
      "grad_norm": 2.09375,
      "learning_rate": 1.9354120297187303e-05,
      "loss": 0.3045,
      "step": 3084
    },
    {
      "epoch": 0.23347996783501254,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9353699113156183e-05,
      "loss": 0.3309,
      "step": 3085
    },
    {
      "epoch": 0.23355565015845986,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9353277796426788e-05,
      "loss": 0.3709,
      "step": 3086
    },
    {
      "epoch": 0.2336313324819072,
      "grad_norm": 1.1640625,
      "learning_rate": 1.9352856347005097e-05,
      "loss": 0.3563,
      "step": 3087
    },
    {
      "epoch": 0.23370701480535452,
      "grad_norm": 0.8046875,
      "learning_rate": 1.935243476489709e-05,
      "loss": 0.3322,
      "step": 3088
    },
    {
      "epoch": 0.23378269712880184,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9352013050108744e-05,
      "loss": 0.2979,
      "step": 3089
    },
    {
      "epoch": 0.2338583794522492,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9351591202646048e-05,
      "loss": 0.3693,
      "step": 3090
    },
    {
      "epoch": 0.2339340617756965,
      "grad_norm": 0.78515625,
      "learning_rate": 1.935116922251498e-05,
      "loss": 0.3014,
      "step": 3091
    },
    {
      "epoch": 0.23400974409914385,
      "grad_norm": 0.84375,
      "learning_rate": 1.935074710972153e-05,
      "loss": 0.367,
      "step": 3092
    },
    {
      "epoch": 0.23408542642259117,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9350324864271686e-05,
      "loss": 0.3538,
      "step": 3093
    },
    {
      "epoch": 0.23416110874603852,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9349902486171438e-05,
      "loss": 0.4076,
      "step": 3094
    },
    {
      "epoch": 0.23423679106948583,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9349479975426778e-05,
      "loss": 0.3234,
      "step": 3095
    },
    {
      "epoch": 0.23431247339293315,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9349057332043698e-05,
      "loss": 0.3666,
      "step": 3096
    },
    {
      "epoch": 0.2343881557163805,
      "grad_norm": 0.87109375,
      "learning_rate": 1.93486345560282e-05,
      "loss": 0.3329,
      "step": 3097
    },
    {
      "epoch": 0.23446383803982782,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9348211647386278e-05,
      "loss": 0.3115,
      "step": 3098
    },
    {
      "epoch": 0.23453952036327516,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9347788606123928e-05,
      "loss": 0.319,
      "step": 3099
    },
    {
      "epoch": 0.23461520268672248,
      "grad_norm": 0.78515625,
      "learning_rate": 1.9347365432247155e-05,
      "loss": 0.3279,
      "step": 3100
    },
    {
      "epoch": 0.23469088501016983,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9346942125761968e-05,
      "loss": 0.3492,
      "step": 3101
    },
    {
      "epoch": 0.23476656733361714,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9346518686674365e-05,
      "loss": 0.3261,
      "step": 3102
    },
    {
      "epoch": 0.23484224965706446,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9346095114990354e-05,
      "loss": 0.4003,
      "step": 3103
    },
    {
      "epoch": 0.2349179319805118,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9345671410715947e-05,
      "loss": 0.3538,
      "step": 3104
    },
    {
      "epoch": 0.23499361430395913,
      "grad_norm": 0.97265625,
      "learning_rate": 1.9345247573857152e-05,
      "loss": 0.3317,
      "step": 3105
    },
    {
      "epoch": 0.23506929662740647,
      "grad_norm": 0.87890625,
      "learning_rate": 1.934482360441998e-05,
      "loss": 0.3412,
      "step": 3106
    },
    {
      "epoch": 0.2351449789508538,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9344399502410455e-05,
      "loss": 0.3361,
      "step": 3107
    },
    {
      "epoch": 0.23522066127430113,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9343975267834584e-05,
      "loss": 0.3516,
      "step": 3108
    },
    {
      "epoch": 0.23529634359774845,
      "grad_norm": 0.90234375,
      "learning_rate": 1.934355090069839e-05,
      "loss": 0.372,
      "step": 3109
    },
    {
      "epoch": 0.23537202592119577,
      "grad_norm": 0.78515625,
      "learning_rate": 1.9343126401007893e-05,
      "loss": 0.2884,
      "step": 3110
    },
    {
      "epoch": 0.23544770824464312,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9342701768769112e-05,
      "loss": 0.3323,
      "step": 3111
    },
    {
      "epoch": 0.23552339056809043,
      "grad_norm": 1.484375,
      "learning_rate": 1.934227700398808e-05,
      "loss": 0.3914,
      "step": 3112
    },
    {
      "epoch": 0.23559907289153778,
      "grad_norm": 0.79296875,
      "learning_rate": 1.934185210667081e-05,
      "loss": 0.3395,
      "step": 3113
    },
    {
      "epoch": 0.2356747552149851,
      "grad_norm": 1.4296875,
      "learning_rate": 1.934142707682334e-05,
      "loss": 0.4239,
      "step": 3114
    },
    {
      "epoch": 0.23575043753843242,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9341001914451694e-05,
      "loss": 0.2947,
      "step": 3115
    },
    {
      "epoch": 0.23582611986187976,
      "grad_norm": 0.7734375,
      "learning_rate": 1.9340576619561907e-05,
      "loss": 0.3202,
      "step": 3116
    },
    {
      "epoch": 0.23590180218532708,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9340151192160013e-05,
      "loss": 0.3506,
      "step": 3117
    },
    {
      "epoch": 0.23597748450877443,
      "grad_norm": 0.76953125,
      "learning_rate": 1.9339725632252045e-05,
      "loss": 0.2965,
      "step": 3118
    },
    {
      "epoch": 0.23605316683222174,
      "grad_norm": 0.86328125,
      "learning_rate": 1.933929993984404e-05,
      "loss": 0.3581,
      "step": 3119
    },
    {
      "epoch": 0.2361288491556691,
      "grad_norm": 0.8984375,
      "learning_rate": 1.933887411494204e-05,
      "loss": 0.3763,
      "step": 3120
    },
    {
      "epoch": 0.2362045314791164,
      "grad_norm": 0.796875,
      "learning_rate": 1.9338448157552088e-05,
      "loss": 0.317,
      "step": 3121
    },
    {
      "epoch": 0.23628021380256373,
      "grad_norm": 0.87109375,
      "learning_rate": 1.933802206768022e-05,
      "loss": 0.3272,
      "step": 3122
    },
    {
      "epoch": 0.23635589612601107,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9337595845332484e-05,
      "loss": 0.3631,
      "step": 3123
    },
    {
      "epoch": 0.2364315784494584,
      "grad_norm": 0.79296875,
      "learning_rate": 1.933716949051493e-05,
      "loss": 0.3219,
      "step": 3124
    },
    {
      "epoch": 0.23650726077290574,
      "grad_norm": 0.79296875,
      "learning_rate": 1.93367430032336e-05,
      "loss": 0.2724,
      "step": 3125
    },
    {
      "epoch": 0.23658294309635305,
      "grad_norm": 0.85546875,
      "learning_rate": 1.933631638349455e-05,
      "loss": 0.3378,
      "step": 3126
    },
    {
      "epoch": 0.2366586254198004,
      "grad_norm": 0.8203125,
      "learning_rate": 1.933588963130383e-05,
      "loss": 0.3263,
      "step": 3127
    },
    {
      "epoch": 0.23673430774324772,
      "grad_norm": 0.984375,
      "learning_rate": 1.9335462746667496e-05,
      "loss": 0.3293,
      "step": 3128
    },
    {
      "epoch": 0.23680999006669504,
      "grad_norm": 0.828125,
      "learning_rate": 1.93350357295916e-05,
      "loss": 0.3333,
      "step": 3129
    },
    {
      "epoch": 0.23688567239014238,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9334608580082204e-05,
      "loss": 0.3134,
      "step": 3130
    },
    {
      "epoch": 0.2369613547135897,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9334181298145365e-05,
      "loss": 0.3541,
      "step": 3131
    },
    {
      "epoch": 0.23703703703703705,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9333753883787153e-05,
      "loss": 0.3302,
      "step": 3132
    },
    {
      "epoch": 0.23711271936048436,
      "grad_norm": 0.7890625,
      "learning_rate": 1.9333326337013618e-05,
      "loss": 0.3286,
      "step": 3133
    },
    {
      "epoch": 0.2371884016839317,
      "grad_norm": 1.734375,
      "learning_rate": 1.9332898657830836e-05,
      "loss": 0.423,
      "step": 3134
    },
    {
      "epoch": 0.23726408400737903,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9332470846244867e-05,
      "loss": 0.3761,
      "step": 3135
    },
    {
      "epoch": 0.23733976633082635,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9332042902261788e-05,
      "loss": 0.3596,
      "step": 3136
    },
    {
      "epoch": 0.2374154486542737,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9331614825887662e-05,
      "loss": 0.3097,
      "step": 3137
    },
    {
      "epoch": 0.237491130977721,
      "grad_norm": 0.90625,
      "learning_rate": 1.9331186617128572e-05,
      "loss": 0.3717,
      "step": 3138
    },
    {
      "epoch": 0.23756681330116836,
      "grad_norm": 0.86328125,
      "learning_rate": 1.933075827599058e-05,
      "loss": 0.378,
      "step": 3139
    },
    {
      "epoch": 0.23764249562461567,
      "grad_norm": 0.859375,
      "learning_rate": 1.9330329802479776e-05,
      "loss": 0.3609,
      "step": 3140
    },
    {
      "epoch": 0.237718177948063,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9329901196602228e-05,
      "loss": 0.3314,
      "step": 3141
    },
    {
      "epoch": 0.23779386027151034,
      "grad_norm": 2.421875,
      "learning_rate": 1.932947245836402e-05,
      "loss": 0.4572,
      "step": 3142
    },
    {
      "epoch": 0.23786954259495766,
      "grad_norm": 2.59375,
      "learning_rate": 1.9329043587771238e-05,
      "loss": 0.4678,
      "step": 3143
    },
    {
      "epoch": 0.237945224918405,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9328614584829963e-05,
      "loss": 0.3148,
      "step": 3144
    },
    {
      "epoch": 0.23802090724185232,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9328185449546282e-05,
      "loss": 0.3505,
      "step": 3145
    },
    {
      "epoch": 0.23809658956529967,
      "grad_norm": 0.7890625,
      "learning_rate": 1.9327756181926283e-05,
      "loss": 0.2879,
      "step": 3146
    },
    {
      "epoch": 0.23817227188874698,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9327326781976055e-05,
      "loss": 0.3577,
      "step": 3147
    },
    {
      "epoch": 0.2382479542121943,
      "grad_norm": 0.88671875,
      "learning_rate": 1.932689724970169e-05,
      "loss": 0.3653,
      "step": 3148
    },
    {
      "epoch": 0.23832363653564165,
      "grad_norm": 0.96484375,
      "learning_rate": 1.932646758510928e-05,
      "loss": 0.3562,
      "step": 3149
    },
    {
      "epoch": 0.23839931885908897,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9326037788204926e-05,
      "loss": 0.3183,
      "step": 3150
    },
    {
      "epoch": 0.2384750011825363,
      "grad_norm": 0.78515625,
      "learning_rate": 1.9325607858994722e-05,
      "loss": 0.3132,
      "step": 3151
    },
    {
      "epoch": 0.23855068350598363,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9325177797484765e-05,
      "loss": 0.2812,
      "step": 3152
    },
    {
      "epoch": 0.23862636582943098,
      "grad_norm": 1.015625,
      "learning_rate": 1.932474760368116e-05,
      "loss": 0.3254,
      "step": 3153
    },
    {
      "epoch": 0.2387020481528783,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9324317277590007e-05,
      "loss": 0.3753,
      "step": 3154
    },
    {
      "epoch": 0.2387777304763256,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9323886819217413e-05,
      "loss": 0.349,
      "step": 3155
    },
    {
      "epoch": 0.23885341279977296,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9323456228569485e-05,
      "loss": 0.3313,
      "step": 3156
    },
    {
      "epoch": 0.23892909512322028,
      "grad_norm": 0.78125,
      "learning_rate": 1.932302550565233e-05,
      "loss": 0.2944,
      "step": 3157
    },
    {
      "epoch": 0.23900477744666762,
      "grad_norm": 0.90625,
      "learning_rate": 1.932259465047206e-05,
      "loss": 0.3876,
      "step": 3158
    },
    {
      "epoch": 0.23908045977011494,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9322163663034784e-05,
      "loss": 0.3034,
      "step": 3159
    },
    {
      "epoch": 0.23915614209356229,
      "grad_norm": 0.796875,
      "learning_rate": 1.9321732543346624e-05,
      "loss": 0.3044,
      "step": 3160
    },
    {
      "epoch": 0.2392318244170096,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9321301291413686e-05,
      "loss": 0.4053,
      "step": 3161
    },
    {
      "epoch": 0.23930750674045692,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9320869907242097e-05,
      "loss": 0.3403,
      "step": 3162
    },
    {
      "epoch": 0.23938318906390427,
      "grad_norm": 0.890625,
      "learning_rate": 1.9320438390837973e-05,
      "loss": 0.359,
      "step": 3163
    },
    {
      "epoch": 0.23945887138735159,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9320006742207434e-05,
      "loss": 0.3479,
      "step": 3164
    },
    {
      "epoch": 0.23953455371079893,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9319574961356607e-05,
      "loss": 0.3311,
      "step": 3165
    },
    {
      "epoch": 0.23961023603424625,
      "grad_norm": 1.1015625,
      "learning_rate": 1.9319143048291615e-05,
      "loss": 0.3797,
      "step": 3166
    },
    {
      "epoch": 0.2396859183576936,
      "grad_norm": 0.9140625,
      "learning_rate": 1.9318711003018587e-05,
      "loss": 0.3323,
      "step": 3167
    },
    {
      "epoch": 0.2397616006811409,
      "grad_norm": 0.76171875,
      "learning_rate": 1.931827882554365e-05,
      "loss": 0.2719,
      "step": 3168
    },
    {
      "epoch": 0.23983728300458823,
      "grad_norm": 0.828125,
      "learning_rate": 1.9317846515872944e-05,
      "loss": 0.3351,
      "step": 3169
    },
    {
      "epoch": 0.23991296532803558,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9317414074012588e-05,
      "loss": 0.3029,
      "step": 3170
    },
    {
      "epoch": 0.2399886476514829,
      "grad_norm": 0.890625,
      "learning_rate": 1.931698149996873e-05,
      "loss": 0.3246,
      "step": 3171
    },
    {
      "epoch": 0.24006432997493024,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9316548793747495e-05,
      "loss": 0.2911,
      "step": 3172
    },
    {
      "epoch": 0.24014001229837756,
      "grad_norm": 2.125,
      "learning_rate": 1.931611595535503e-05,
      "loss": 0.4411,
      "step": 3173
    },
    {
      "epoch": 0.24021569462182488,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9315682984797475e-05,
      "loss": 0.3408,
      "step": 3174
    },
    {
      "epoch": 0.24029137694527222,
      "grad_norm": 0.89453125,
      "learning_rate": 1.931524988208097e-05,
      "loss": 0.2863,
      "step": 3175
    },
    {
      "epoch": 0.24036705926871954,
      "grad_norm": 0.80859375,
      "learning_rate": 1.931481664721166e-05,
      "loss": 0.3545,
      "step": 3176
    },
    {
      "epoch": 0.2404427415921669,
      "grad_norm": 1.4375,
      "learning_rate": 1.931438328019569e-05,
      "loss": 0.4624,
      "step": 3177
    },
    {
      "epoch": 0.2405184239156142,
      "grad_norm": 0.875,
      "learning_rate": 1.9313949781039207e-05,
      "loss": 0.3485,
      "step": 3178
    },
    {
      "epoch": 0.24059410623906155,
      "grad_norm": 0.8984375,
      "learning_rate": 1.931351614974837e-05,
      "loss": 0.361,
      "step": 3179
    },
    {
      "epoch": 0.24066978856250887,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9313082386329315e-05,
      "loss": 0.3463,
      "step": 3180
    },
    {
      "epoch": 0.2407454708859562,
      "grad_norm": 0.85546875,
      "learning_rate": 1.931264849078821e-05,
      "loss": 0.3407,
      "step": 3181
    },
    {
      "epoch": 0.24082115320940353,
      "grad_norm": 0.75,
      "learning_rate": 1.93122144631312e-05,
      "loss": 0.2695,
      "step": 3182
    },
    {
      "epoch": 0.24089683553285085,
      "grad_norm": 0.828125,
      "learning_rate": 1.9311780303364456e-05,
      "loss": 0.3064,
      "step": 3183
    },
    {
      "epoch": 0.2409725178562982,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9311346011494124e-05,
      "loss": 0.4062,
      "step": 3184
    },
    {
      "epoch": 0.24104820017974551,
      "grad_norm": 0.7890625,
      "learning_rate": 1.9310911587526367e-05,
      "loss": 0.3129,
      "step": 3185
    },
    {
      "epoch": 0.24112388250319286,
      "grad_norm": 1.046875,
      "learning_rate": 1.9310477031467357e-05,
      "loss": 0.3547,
      "step": 3186
    },
    {
      "epoch": 0.24119956482664018,
      "grad_norm": 0.81640625,
      "learning_rate": 1.931004234332325e-05,
      "loss": 0.3097,
      "step": 3187
    },
    {
      "epoch": 0.2412752471500875,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9309607523100218e-05,
      "loss": 0.287,
      "step": 3188
    },
    {
      "epoch": 0.24135092947353484,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9309172570804425e-05,
      "loss": 0.357,
      "step": 3189
    },
    {
      "epoch": 0.24142661179698216,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9308737486442045e-05,
      "loss": 0.3195,
      "step": 3190
    },
    {
      "epoch": 0.2415022941204295,
      "grad_norm": 0.765625,
      "learning_rate": 1.9308302270019248e-05,
      "loss": 0.3079,
      "step": 3191
    },
    {
      "epoch": 0.24157797644387682,
      "grad_norm": 0.80078125,
      "learning_rate": 1.930786692154221e-05,
      "loss": 0.3102,
      "step": 3192
    },
    {
      "epoch": 0.24165365876732417,
      "grad_norm": 0.86328125,
      "learning_rate": 1.930743144101711e-05,
      "loss": 0.3049,
      "step": 3193
    },
    {
      "epoch": 0.2417293410907715,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9306995828450118e-05,
      "loss": 0.3425,
      "step": 3194
    },
    {
      "epoch": 0.2418050234142188,
      "grad_norm": 0.73828125,
      "learning_rate": 1.9306560083847422e-05,
      "loss": 0.2857,
      "step": 3195
    },
    {
      "epoch": 0.24188070573766615,
      "grad_norm": 0.8203125,
      "learning_rate": 1.93061242072152e-05,
      "loss": 0.3294,
      "step": 3196
    },
    {
      "epoch": 0.24195638806111347,
      "grad_norm": 0.83203125,
      "learning_rate": 1.930568819855964e-05,
      "loss": 0.3312,
      "step": 3197
    },
    {
      "epoch": 0.24203207038456082,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9305252057886918e-05,
      "loss": 0.3557,
      "step": 3198
    },
    {
      "epoch": 0.24210775270800813,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9304815785203228e-05,
      "loss": 0.3302,
      "step": 3199
    },
    {
      "epoch": 0.24218343503145545,
      "grad_norm": 0.828125,
      "learning_rate": 1.930437938051476e-05,
      "loss": 0.3396,
      "step": 3200
    },
    {
      "epoch": 0.2422591173549028,
      "grad_norm": 0.84765625,
      "learning_rate": 1.93039428438277e-05,
      "loss": 0.3627,
      "step": 3201
    },
    {
      "epoch": 0.24233479967835012,
      "grad_norm": 0.84765625,
      "learning_rate": 1.930350617514825e-05,
      "loss": 0.3468,
      "step": 3202
    },
    {
      "epoch": 0.24241048200179746,
      "grad_norm": 0.828125,
      "learning_rate": 1.93030693744826e-05,
      "loss": 0.3517,
      "step": 3203
    },
    {
      "epoch": 0.24248616432524478,
      "grad_norm": 0.80859375,
      "learning_rate": 1.930263244183694e-05,
      "loss": 0.3121,
      "step": 3204
    },
    {
      "epoch": 0.24256184664869213,
      "grad_norm": 0.8046875,
      "learning_rate": 1.930219537721748e-05,
      "loss": 0.3228,
      "step": 3205
    },
    {
      "epoch": 0.24263752897213944,
      "grad_norm": 0.7734375,
      "learning_rate": 1.9301758180630414e-05,
      "loss": 0.2701,
      "step": 3206
    },
    {
      "epoch": 0.24271321129558676,
      "grad_norm": 6.09375,
      "learning_rate": 1.9301320852081947e-05,
      "loss": 0.4426,
      "step": 3207
    },
    {
      "epoch": 0.2427888936190341,
      "grad_norm": 0.82421875,
      "learning_rate": 1.930088339157828e-05,
      "loss": 0.3309,
      "step": 3208
    },
    {
      "epoch": 0.24286457594248143,
      "grad_norm": 4.03125,
      "learning_rate": 1.9300445799125623e-05,
      "loss": 0.4246,
      "step": 3209
    },
    {
      "epoch": 0.24294025826592877,
      "grad_norm": 0.828125,
      "learning_rate": 1.930000807473018e-05,
      "loss": 0.3292,
      "step": 3210
    },
    {
      "epoch": 0.2430159405893761,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9299570218398167e-05,
      "loss": 0.3862,
      "step": 3211
    },
    {
      "epoch": 0.24309162291282344,
      "grad_norm": 0.8125,
      "learning_rate": 1.929913223013579e-05,
      "loss": 0.3126,
      "step": 3212
    },
    {
      "epoch": 0.24316730523627075,
      "grad_norm": 0.84375,
      "learning_rate": 1.9298694109949263e-05,
      "loss": 0.3288,
      "step": 3213
    },
    {
      "epoch": 0.24324298755971807,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9298255857844803e-05,
      "loss": 0.315,
      "step": 3214
    },
    {
      "epoch": 0.24331866988316542,
      "grad_norm": 2.234375,
      "learning_rate": 1.9297817473828628e-05,
      "loss": 0.5038,
      "step": 3215
    },
    {
      "epoch": 0.24339435220661274,
      "grad_norm": 0.828125,
      "learning_rate": 1.9297378957906957e-05,
      "loss": 0.3094,
      "step": 3216
    },
    {
      "epoch": 0.24347003453006008,
      "grad_norm": 0.84375,
      "learning_rate": 1.9296940310086013e-05,
      "loss": 0.313,
      "step": 3217
    },
    {
      "epoch": 0.2435457168535074,
      "grad_norm": 0.9453125,
      "learning_rate": 1.929650153037201e-05,
      "loss": 0.3634,
      "step": 3218
    },
    {
      "epoch": 0.24362139917695474,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9296062618771184e-05,
      "loss": 0.3078,
      "step": 3219
    },
    {
      "epoch": 0.24369708150040206,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9295623575289756e-05,
      "loss": 0.3156,
      "step": 3220
    },
    {
      "epoch": 0.24377276382384938,
      "grad_norm": 0.8515625,
      "learning_rate": 1.929518439993396e-05,
      "loss": 0.3375,
      "step": 3221
    },
    {
      "epoch": 0.24384844614729673,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9294745092710016e-05,
      "loss": 0.337,
      "step": 3222
    },
    {
      "epoch": 0.24392412847074404,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9294305653624164e-05,
      "loss": 0.2876,
      "step": 3223
    },
    {
      "epoch": 0.2439998107941914,
      "grad_norm": 0.87890625,
      "learning_rate": 1.929386608268264e-05,
      "loss": 0.3441,
      "step": 3224
    },
    {
      "epoch": 0.2440754931176387,
      "grad_norm": 0.890625,
      "learning_rate": 1.9293426379891677e-05,
      "loss": 0.3204,
      "step": 3225
    },
    {
      "epoch": 0.24415117544108605,
      "grad_norm": 0.83203125,
      "learning_rate": 1.929298654525751e-05,
      "loss": 0.3204,
      "step": 3226
    },
    {
      "epoch": 0.24422685776453337,
      "grad_norm": 2.8125,
      "learning_rate": 1.929254657878638e-05,
      "loss": 0.5313,
      "step": 3227
    },
    {
      "epoch": 0.2443025400879807,
      "grad_norm": 0.8125,
      "learning_rate": 1.9292106480484533e-05,
      "loss": 0.312,
      "step": 3228
    },
    {
      "epoch": 0.24437822241142804,
      "grad_norm": 0.73828125,
      "learning_rate": 1.9291666250358206e-05,
      "loss": 0.2824,
      "step": 3229
    },
    {
      "epoch": 0.24445390473487535,
      "grad_norm": 0.7578125,
      "learning_rate": 1.9291225888413652e-05,
      "loss": 0.3031,
      "step": 3230
    },
    {
      "epoch": 0.2445295870583227,
      "grad_norm": 0.828125,
      "learning_rate": 1.9290785394657113e-05,
      "loss": 0.3182,
      "step": 3231
    },
    {
      "epoch": 0.24460526938177002,
      "grad_norm": 0.81640625,
      "learning_rate": 1.929034476909484e-05,
      "loss": 0.3458,
      "step": 3232
    },
    {
      "epoch": 0.24468095170521734,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9289904011733084e-05,
      "loss": 0.3613,
      "step": 3233
    },
    {
      "epoch": 0.24475663402866468,
      "grad_norm": 0.94140625,
      "learning_rate": 1.92894631225781e-05,
      "loss": 0.3775,
      "step": 3234
    },
    {
      "epoch": 0.244832316352112,
      "grad_norm": 0.828125,
      "learning_rate": 1.9289022101636135e-05,
      "loss": 0.3191,
      "step": 3235
    },
    {
      "epoch": 0.24490799867555935,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9288580948913456e-05,
      "loss": 0.332,
      "step": 3236
    },
    {
      "epoch": 0.24498368099900666,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9288139664416312e-05,
      "loss": 0.3126,
      "step": 3237
    },
    {
      "epoch": 0.245059363322454,
      "grad_norm": 0.890625,
      "learning_rate": 1.9287698248150973e-05,
      "loss": 0.3711,
      "step": 3238
    },
    {
      "epoch": 0.24513504564590133,
      "grad_norm": 0.76953125,
      "learning_rate": 1.9287256700123692e-05,
      "loss": 0.2845,
      "step": 3239
    },
    {
      "epoch": 0.24521072796934865,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9286815020340738e-05,
      "loss": 0.3449,
      "step": 3240
    },
    {
      "epoch": 0.245286410292796,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9286373208808375e-05,
      "loss": 0.2981,
      "step": 3241
    },
    {
      "epoch": 0.2453620926162433,
      "grad_norm": 0.95703125,
      "learning_rate": 1.9285931265532875e-05,
      "loss": 0.3657,
      "step": 3242
    },
    {
      "epoch": 0.24543777493969066,
      "grad_norm": 1.015625,
      "learning_rate": 1.9285489190520502e-05,
      "loss": 0.3399,
      "step": 3243
    },
    {
      "epoch": 0.24551345726313797,
      "grad_norm": 0.953125,
      "learning_rate": 1.9285046983777535e-05,
      "loss": 0.3154,
      "step": 3244
    },
    {
      "epoch": 0.24558913958658532,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9284604645310237e-05,
      "loss": 0.3185,
      "step": 3245
    },
    {
      "epoch": 0.24566482191003264,
      "grad_norm": 0.828125,
      "learning_rate": 1.9284162175124892e-05,
      "loss": 0.351,
      "step": 3246
    },
    {
      "epoch": 0.24574050423347996,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9283719573227773e-05,
      "loss": 0.3645,
      "step": 3247
    },
    {
      "epoch": 0.2458161865569273,
      "grad_norm": 1.78125,
      "learning_rate": 1.9283276839625166e-05,
      "loss": 0.4033,
      "step": 3248
    },
    {
      "epoch": 0.24589186888037462,
      "grad_norm": 0.8125,
      "learning_rate": 1.928283397432334e-05,
      "loss": 0.3177,
      "step": 3249
    },
    {
      "epoch": 0.24596755120382197,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9282390977328586e-05,
      "loss": 0.3395,
      "step": 3250
    },
    {
      "epoch": 0.24604323352726928,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9281947848647186e-05,
      "loss": 0.3099,
      "step": 3251
    },
    {
      "epoch": 0.24611891585071663,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9281504588285428e-05,
      "loss": 0.3364,
      "step": 3252
    },
    {
      "epoch": 0.24619459817416395,
      "grad_norm": 0.75390625,
      "learning_rate": 1.9281061196249603e-05,
      "loss": 0.2914,
      "step": 3253
    },
    {
      "epoch": 0.24627028049761127,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9280617672545993e-05,
      "loss": 0.3126,
      "step": 3254
    },
    {
      "epoch": 0.2463459628210586,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9280174017180895e-05,
      "loss": 0.3373,
      "step": 3255
    },
    {
      "epoch": 0.24642164514450593,
      "grad_norm": 0.83984375,
      "learning_rate": 1.927973023016061e-05,
      "loss": 0.3495,
      "step": 3256
    },
    {
      "epoch": 0.24649732746795328,
      "grad_norm": 0.78125,
      "learning_rate": 1.927928631149142e-05,
      "loss": 0.2766,
      "step": 3257
    },
    {
      "epoch": 0.2465730097914006,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9278842261179635e-05,
      "loss": 0.3417,
      "step": 3258
    },
    {
      "epoch": 0.24664869211484794,
      "grad_norm": 0.8828125,
      "learning_rate": 1.927839807923155e-05,
      "loss": 0.3518,
      "step": 3259
    },
    {
      "epoch": 0.24672437443829526,
      "grad_norm": 0.78125,
      "learning_rate": 1.927795376565346e-05,
      "loss": 0.2888,
      "step": 3260
    },
    {
      "epoch": 0.24680005676174258,
      "grad_norm": 0.8203125,
      "learning_rate": 1.927750932045168e-05,
      "loss": 0.3278,
      "step": 3261
    },
    {
      "epoch": 0.24687573908518992,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9277064743632505e-05,
      "loss": 0.3397,
      "step": 3262
    },
    {
      "epoch": 0.24695142140863724,
      "grad_norm": 1.28125,
      "learning_rate": 1.9276620035202248e-05,
      "loss": 0.3538,
      "step": 3263
    },
    {
      "epoch": 0.24702710373208459,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9276175195167218e-05,
      "loss": 0.3223,
      "step": 3264
    },
    {
      "epoch": 0.2471027860555319,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9275730223533724e-05,
      "loss": 0.3422,
      "step": 3265
    },
    {
      "epoch": 0.24717846837897922,
      "grad_norm": 1.78125,
      "learning_rate": 1.927528512030808e-05,
      "loss": 0.4583,
      "step": 3266
    },
    {
      "epoch": 0.24725415070242657,
      "grad_norm": 0.8125,
      "learning_rate": 1.9274839885496595e-05,
      "loss": 0.3259,
      "step": 3267
    },
    {
      "epoch": 0.24732983302587389,
      "grad_norm": 0.84375,
      "learning_rate": 1.92743945191056e-05,
      "loss": 0.3552,
      "step": 3268
    },
    {
      "epoch": 0.24740551534932123,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9273949021141393e-05,
      "loss": 0.3001,
      "step": 3269
    },
    {
      "epoch": 0.24748119767276855,
      "grad_norm": 0.859375,
      "learning_rate": 1.9273503391610307e-05,
      "loss": 0.3331,
      "step": 3270
    },
    {
      "epoch": 0.2475568799962159,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9273057630518664e-05,
      "loss": 0.3588,
      "step": 3271
    },
    {
      "epoch": 0.2476325623196632,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9272611737872785e-05,
      "loss": 0.3527,
      "step": 3272
    },
    {
      "epoch": 0.24770824464311053,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9272165713678993e-05,
      "loss": 0.3656,
      "step": 3273
    },
    {
      "epoch": 0.24778392696655788,
      "grad_norm": 0.7890625,
      "learning_rate": 1.9271719557943624e-05,
      "loss": 0.3017,
      "step": 3274
    },
    {
      "epoch": 0.2478596092900052,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9271273270673e-05,
      "loss": 0.3133,
      "step": 3275
    },
    {
      "epoch": 0.24793529161345254,
      "grad_norm": 0.75,
      "learning_rate": 1.9270826851873452e-05,
      "loss": 0.265,
      "step": 3276
    },
    {
      "epoch": 0.24801097393689986,
      "grad_norm": 0.828125,
      "learning_rate": 1.9270380301551318e-05,
      "loss": 0.3403,
      "step": 3277
    },
    {
      "epoch": 0.2480866562603472,
      "grad_norm": 0.8046875,
      "learning_rate": 1.926993361971293e-05,
      "loss": 0.3207,
      "step": 3278
    },
    {
      "epoch": 0.24816233858379452,
      "grad_norm": 0.875,
      "learning_rate": 1.9269486806364626e-05,
      "loss": 0.3843,
      "step": 3279
    },
    {
      "epoch": 0.24823802090724184,
      "grad_norm": 0.80859375,
      "learning_rate": 1.926903986151275e-05,
      "loss": 0.2941,
      "step": 3280
    },
    {
      "epoch": 0.2483137032306892,
      "grad_norm": 0.8046875,
      "learning_rate": 1.926859278516363e-05,
      "loss": 0.308,
      "step": 3281
    },
    {
      "epoch": 0.2483893855541365,
      "grad_norm": 0.8203125,
      "learning_rate": 1.926814557732362e-05,
      "loss": 0.2972,
      "step": 3282
    },
    {
      "epoch": 0.24846506787758385,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9267698237999056e-05,
      "loss": 0.3315,
      "step": 3283
    },
    {
      "epoch": 0.24854075020103117,
      "grad_norm": 0.83203125,
      "learning_rate": 1.926725076719629e-05,
      "loss": 0.3184,
      "step": 3284
    },
    {
      "epoch": 0.24861643252447851,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9266803164921673e-05,
      "loss": 0.337,
      "step": 3285
    },
    {
      "epoch": 0.24869211484792583,
      "grad_norm": 0.828125,
      "learning_rate": 1.926635543118155e-05,
      "loss": 0.3504,
      "step": 3286
    },
    {
      "epoch": 0.24876779717137315,
      "grad_norm": 2.484375,
      "learning_rate": 1.926590756598227e-05,
      "loss": 0.5138,
      "step": 3287
    },
    {
      "epoch": 0.2488434794948205,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9265459569330192e-05,
      "loss": 0.4046,
      "step": 3288
    },
    {
      "epoch": 0.24891916181826781,
      "grad_norm": 0.828125,
      "learning_rate": 1.926501144123167e-05,
      "loss": 0.3116,
      "step": 3289
    },
    {
      "epoch": 0.24899484414171516,
      "grad_norm": 0.82421875,
      "learning_rate": 1.926456318169306e-05,
      "loss": 0.3151,
      "step": 3290
    },
    {
      "epoch": 0.24907052646516248,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9264114790720722e-05,
      "loss": 0.3574,
      "step": 3291
    },
    {
      "epoch": 0.2491462087886098,
      "grad_norm": 0.859375,
      "learning_rate": 1.926366626832102e-05,
      "loss": 0.3711,
      "step": 3292
    },
    {
      "epoch": 0.24922189111205714,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9263217614500318e-05,
      "loss": 0.354,
      "step": 3293
    },
    {
      "epoch": 0.24929757343550446,
      "grad_norm": 2.734375,
      "learning_rate": 1.926276882926497e-05,
      "loss": 0.442,
      "step": 3294
    },
    {
      "epoch": 0.2493732557589518,
      "grad_norm": 0.8359375,
      "learning_rate": 1.926231991262136e-05,
      "loss": 0.3335,
      "step": 3295
    },
    {
      "epoch": 0.24944893808239912,
      "grad_norm": 0.859375,
      "learning_rate": 1.926187086457584e-05,
      "loss": 0.3714,
      "step": 3296
    },
    {
      "epoch": 0.24952462040584647,
      "grad_norm": 2.046875,
      "learning_rate": 1.9261421685134792e-05,
      "loss": 0.4568,
      "step": 3297
    },
    {
      "epoch": 0.2496003027292938,
      "grad_norm": 0.97265625,
      "learning_rate": 1.9260972374304582e-05,
      "loss": 0.3947,
      "step": 3298
    },
    {
      "epoch": 0.2496759850527411,
      "grad_norm": 0.81640625,
      "learning_rate": 1.926052293209159e-05,
      "loss": 0.3267,
      "step": 3299
    },
    {
      "epoch": 0.24975166737618845,
      "grad_norm": 0.796875,
      "learning_rate": 1.9260073358502184e-05,
      "loss": 0.3211,
      "step": 3300
    },
    {
      "epoch": 0.24982734969963577,
      "grad_norm": 0.76171875,
      "learning_rate": 1.925962365354275e-05,
      "loss": 0.2975,
      "step": 3301
    },
    {
      "epoch": 0.24990303202308312,
      "grad_norm": 0.80078125,
      "learning_rate": 1.925917381721966e-05,
      "loss": 0.3394,
      "step": 3302
    },
    {
      "epoch": 0.24997871434653043,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9258723849539306e-05,
      "loss": 0.324,
      "step": 3303
    },
    {
      "epoch": 0.25005439666997775,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9258273750508062e-05,
      "loss": 0.3589,
      "step": 3304
    },
    {
      "epoch": 0.2501300789934251,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9257823520132318e-05,
      "loss": 0.3543,
      "step": 3305
    },
    {
      "epoch": 0.25020576131687244,
      "grad_norm": 0.8359375,
      "learning_rate": 1.925737315841846e-05,
      "loss": 0.3502,
      "step": 3306
    },
    {
      "epoch": 0.25028144364031973,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9256922665372876e-05,
      "loss": 0.3706,
      "step": 3307
    },
    {
      "epoch": 0.2503571259637671,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9256472041001962e-05,
      "loss": 0.3771,
      "step": 3308
    },
    {
      "epoch": 0.2504328082872144,
      "grad_norm": 0.78125,
      "learning_rate": 1.9256021285312105e-05,
      "loss": 0.3163,
      "step": 3309
    },
    {
      "epoch": 0.25050849061066177,
      "grad_norm": 1.390625,
      "learning_rate": 1.9255570398309702e-05,
      "loss": 0.4216,
      "step": 3310
    },
    {
      "epoch": 0.25058417293410906,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9255119380001152e-05,
      "loss": 0.337,
      "step": 3311
    },
    {
      "epoch": 0.2506598552575564,
      "grad_norm": 0.80859375,
      "learning_rate": 1.925466823039285e-05,
      "loss": 0.3233,
      "step": 3312
    },
    {
      "epoch": 0.25073553758100375,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9254216949491198e-05,
      "loss": 0.3342,
      "step": 3313
    },
    {
      "epoch": 0.25081121990445104,
      "grad_norm": 1.4296875,
      "learning_rate": 1.9253765537302597e-05,
      "loss": 0.4322,
      "step": 3314
    },
    {
      "epoch": 0.2508869022278984,
      "grad_norm": 0.828125,
      "learning_rate": 1.925331399383345e-05,
      "loss": 0.3559,
      "step": 3315
    },
    {
      "epoch": 0.25096258455134574,
      "grad_norm": 0.796875,
      "learning_rate": 1.925286231909017e-05,
      "loss": 0.3407,
      "step": 3316
    },
    {
      "epoch": 0.2510382668747931,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9252410513079155e-05,
      "loss": 0.3866,
      "step": 3317
    },
    {
      "epoch": 0.25111394919824037,
      "grad_norm": 0.8203125,
      "learning_rate": 1.925195857580682e-05,
      "loss": 0.3151,
      "step": 3318
    },
    {
      "epoch": 0.2511896315216877,
      "grad_norm": 0.8359375,
      "learning_rate": 1.925150650727958e-05,
      "loss": 0.3251,
      "step": 3319
    },
    {
      "epoch": 0.25126531384513506,
      "grad_norm": 0.77734375,
      "learning_rate": 1.925105430750384e-05,
      "loss": 0.288,
      "step": 3320
    },
    {
      "epoch": 0.25134099616858235,
      "grad_norm": 0.80078125,
      "learning_rate": 1.925060197648602e-05,
      "loss": 0.3385,
      "step": 3321
    },
    {
      "epoch": 0.2514166784920297,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9250149514232535e-05,
      "loss": 0.4011,
      "step": 3322
    },
    {
      "epoch": 0.25149236081547705,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9249696920749807e-05,
      "loss": 0.3623,
      "step": 3323
    },
    {
      "epoch": 0.2515680431389244,
      "grad_norm": 0.7578125,
      "learning_rate": 1.9249244196044253e-05,
      "loss": 0.2909,
      "step": 3324
    },
    {
      "epoch": 0.2516437254623717,
      "grad_norm": 1.2421875,
      "learning_rate": 1.9248791340122302e-05,
      "loss": 0.3876,
      "step": 3325
    },
    {
      "epoch": 0.251719407785819,
      "grad_norm": 0.8828125,
      "learning_rate": 1.924833835299037e-05,
      "loss": 0.3853,
      "step": 3326
    },
    {
      "epoch": 0.2517950901092664,
      "grad_norm": 0.875,
      "learning_rate": 1.9247885234654887e-05,
      "loss": 0.3663,
      "step": 3327
    },
    {
      "epoch": 0.25187077243271366,
      "grad_norm": 0.8046875,
      "learning_rate": 1.924743198512229e-05,
      "loss": 0.3345,
      "step": 3328
    },
    {
      "epoch": 0.251946454756161,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9246978604398994e-05,
      "loss": 0.3516,
      "step": 3329
    },
    {
      "epoch": 0.25202213707960835,
      "grad_norm": 0.83984375,
      "learning_rate": 1.924652509249144e-05,
      "loss": 0.368,
      "step": 3330
    },
    {
      "epoch": 0.2520978194030557,
      "grad_norm": 0.84765625,
      "learning_rate": 1.924607144940606e-05,
      "loss": 0.3667,
      "step": 3331
    },
    {
      "epoch": 0.252173501726503,
      "grad_norm": 0.83984375,
      "learning_rate": 1.924561767514929e-05,
      "loss": 0.3879,
      "step": 3332
    },
    {
      "epoch": 0.25224918404995034,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9245163769727566e-05,
      "loss": 0.328,
      "step": 3333
    },
    {
      "epoch": 0.2523248663733977,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9244709733147328e-05,
      "loss": 0.3377,
      "step": 3334
    },
    {
      "epoch": 0.252400548696845,
      "grad_norm": 0.828125,
      "learning_rate": 1.9244255565415023e-05,
      "loss": 0.3444,
      "step": 3335
    },
    {
      "epoch": 0.2524762310202923,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9243801266537084e-05,
      "loss": 0.34,
      "step": 3336
    },
    {
      "epoch": 0.25255191334373966,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9243346836519962e-05,
      "loss": 0.3455,
      "step": 3337
    },
    {
      "epoch": 0.25262759566718695,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9242892275370103e-05,
      "loss": 0.3538,
      "step": 3338
    },
    {
      "epoch": 0.2527032779906343,
      "grad_norm": 0.8984375,
      "learning_rate": 1.924243758309396e-05,
      "loss": 0.3795,
      "step": 3339
    },
    {
      "epoch": 0.25277896031408165,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9241982759697976e-05,
      "loss": 0.317,
      "step": 3340
    },
    {
      "epoch": 0.252854642637529,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9241527805188604e-05,
      "loss": 0.3531,
      "step": 3341
    },
    {
      "epoch": 0.2529303249609763,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9241072719572305e-05,
      "loss": 0.342,
      "step": 3342
    },
    {
      "epoch": 0.25300600728442363,
      "grad_norm": 0.984375,
      "learning_rate": 1.924061750285553e-05,
      "loss": 0.3056,
      "step": 3343
    },
    {
      "epoch": 0.253081689607871,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9240162155044738e-05,
      "loss": 0.3345,
      "step": 3344
    },
    {
      "epoch": 0.25315737193131826,
      "grad_norm": 0.8828125,
      "learning_rate": 1.923970667614639e-05,
      "loss": 0.366,
      "step": 3345
    },
    {
      "epoch": 0.2532330542547656,
      "grad_norm": 0.97265625,
      "learning_rate": 1.9239251066166944e-05,
      "loss": 0.3753,
      "step": 3346
    },
    {
      "epoch": 0.25330873657821296,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9238795325112867e-05,
      "loss": 0.3443,
      "step": 3347
    },
    {
      "epoch": 0.2533844189016603,
      "grad_norm": 0.7578125,
      "learning_rate": 1.923833945299063e-05,
      "loss": 0.2874,
      "step": 3348
    },
    {
      "epoch": 0.2534601012251076,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9237883449806687e-05,
      "loss": 0.3196,
      "step": 3349
    },
    {
      "epoch": 0.25353578354855494,
      "grad_norm": 0.82421875,
      "learning_rate": 1.923742731556752e-05,
      "loss": 0.3408,
      "step": 3350
    },
    {
      "epoch": 0.2536114658720023,
      "grad_norm": 0.80078125,
      "learning_rate": 1.923697105027959e-05,
      "loss": 0.3347,
      "step": 3351
    },
    {
      "epoch": 0.2536871481954496,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9236514653949377e-05,
      "loss": 0.3345,
      "step": 3352
    },
    {
      "epoch": 0.2537628305188969,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9236058126583354e-05,
      "loss": 0.3365,
      "step": 3353
    },
    {
      "epoch": 0.25383851284234427,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9235601468187997e-05,
      "loss": 0.3394,
      "step": 3354
    },
    {
      "epoch": 0.2539141951657916,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9235144678769783e-05,
      "loss": 0.3513,
      "step": 3355
    },
    {
      "epoch": 0.2539898774892389,
      "grad_norm": 0.859375,
      "learning_rate": 1.9234687758335193e-05,
      "loss": 0.4039,
      "step": 3356
    },
    {
      "epoch": 0.25406555981268625,
      "grad_norm": 0.84375,
      "learning_rate": 1.9234230706890707e-05,
      "loss": 0.3398,
      "step": 3357
    },
    {
      "epoch": 0.2541412421361336,
      "grad_norm": 0.859375,
      "learning_rate": 1.923377352444282e-05,
      "loss": 0.3395,
      "step": 3358
    },
    {
      "epoch": 0.2542169244595809,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9233316210998002e-05,
      "loss": 0.3648,
      "step": 3359
    },
    {
      "epoch": 0.25429260678302823,
      "grad_norm": 0.796875,
      "learning_rate": 1.9232858766562756e-05,
      "loss": 0.3132,
      "step": 3360
    },
    {
      "epoch": 0.2543682891064756,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9232401191143557e-05,
      "loss": 0.3008,
      "step": 3361
    },
    {
      "epoch": 0.2544439714299229,
      "grad_norm": 0.77734375,
      "learning_rate": 1.9231943484746905e-05,
      "loss": 0.2835,
      "step": 3362
    },
    {
      "epoch": 0.2545196537533702,
      "grad_norm": 0.85546875,
      "learning_rate": 1.92314856473793e-05,
      "loss": 0.339,
      "step": 3363
    },
    {
      "epoch": 0.25459533607681756,
      "grad_norm": 0.90234375,
      "learning_rate": 1.923102767904722e-05,
      "loss": 0.3725,
      "step": 3364
    },
    {
      "epoch": 0.2546710184002649,
      "grad_norm": 0.796875,
      "learning_rate": 1.9230569579757173e-05,
      "loss": 0.3264,
      "step": 3365
    },
    {
      "epoch": 0.2547467007237122,
      "grad_norm": 0.8203125,
      "learning_rate": 1.923011134951566e-05,
      "loss": 0.338,
      "step": 3366
    },
    {
      "epoch": 0.25482238304715954,
      "grad_norm": 0.859375,
      "learning_rate": 1.9229652988329178e-05,
      "loss": 0.3641,
      "step": 3367
    },
    {
      "epoch": 0.2548980653706069,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9229194496204226e-05,
      "loss": 0.3548,
      "step": 3368
    },
    {
      "epoch": 0.25497374769405423,
      "grad_norm": 1.2421875,
      "learning_rate": 1.9228735873147315e-05,
      "loss": 0.3667,
      "step": 3369
    },
    {
      "epoch": 0.2550494300175015,
      "grad_norm": 0.87890625,
      "learning_rate": 1.922827711916495e-05,
      "loss": 0.3812,
      "step": 3370
    },
    {
      "epoch": 0.25512511234094887,
      "grad_norm": 0.84375,
      "learning_rate": 1.9227818234263633e-05,
      "loss": 0.3493,
      "step": 3371
    },
    {
      "epoch": 0.2552007946643962,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9227359218449885e-05,
      "loss": 0.3566,
      "step": 3372
    },
    {
      "epoch": 0.2552764769878435,
      "grad_norm": 0.73828125,
      "learning_rate": 1.9226900071730206e-05,
      "loss": 0.2638,
      "step": 3373
    },
    {
      "epoch": 0.25535215931129085,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9226440794111117e-05,
      "loss": 0.3216,
      "step": 3374
    },
    {
      "epoch": 0.2554278416347382,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9225981385599132e-05,
      "loss": 0.3178,
      "step": 3375
    },
    {
      "epoch": 0.25550352395818554,
      "grad_norm": 0.8125,
      "learning_rate": 1.9225521846200773e-05,
      "loss": 0.3607,
      "step": 3376
    },
    {
      "epoch": 0.25557920628163283,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9225062175922552e-05,
      "loss": 0.314,
      "step": 3377
    },
    {
      "epoch": 0.2556548886050802,
      "grad_norm": 0.84375,
      "learning_rate": 1.9224602374770992e-05,
      "loss": 0.3079,
      "step": 3378
    },
    {
      "epoch": 0.2557305709285275,
      "grad_norm": 0.87890625,
      "learning_rate": 1.922414244275262e-05,
      "loss": 0.3796,
      "step": 3379
    },
    {
      "epoch": 0.2558062532519748,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9223682379873952e-05,
      "loss": 0.2865,
      "step": 3380
    },
    {
      "epoch": 0.25588193557542216,
      "grad_norm": 0.84375,
      "learning_rate": 1.922322218614153e-05,
      "loss": 0.3261,
      "step": 3381
    },
    {
      "epoch": 0.2559576178988695,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9222761861561868e-05,
      "loss": 0.3161,
      "step": 3382
    },
    {
      "epoch": 0.25603330022231685,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9222301406141504e-05,
      "loss": 0.3364,
      "step": 3383
    },
    {
      "epoch": 0.25610898254576414,
      "grad_norm": 0.921875,
      "learning_rate": 1.922184081988697e-05,
      "loss": 0.3503,
      "step": 3384
    },
    {
      "epoch": 0.2561846648692115,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9221380102804796e-05,
      "loss": 0.2888,
      "step": 3385
    },
    {
      "epoch": 0.25626034719265883,
      "grad_norm": 0.91796875,
      "learning_rate": 1.922091925490152e-05,
      "loss": 0.3955,
      "step": 3386
    },
    {
      "epoch": 0.2563360295161061,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9220458276183683e-05,
      "loss": 0.2966,
      "step": 3387
    },
    {
      "epoch": 0.25641171183955347,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9219997166657826e-05,
      "loss": 0.326,
      "step": 3388
    },
    {
      "epoch": 0.2564873941630008,
      "grad_norm": 0.84375,
      "learning_rate": 1.9219535926330478e-05,
      "loss": 0.3003,
      "step": 3389
    },
    {
      "epoch": 0.25656307648644816,
      "grad_norm": 0.859375,
      "learning_rate": 1.92190745552082e-05,
      "loss": 0.3815,
      "step": 3390
    },
    {
      "epoch": 0.25663875880989545,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9218613053297527e-05,
      "loss": 0.3373,
      "step": 3391
    },
    {
      "epoch": 0.2567144411333428,
      "grad_norm": 0.875,
      "learning_rate": 1.9218151420605006e-05,
      "loss": 0.3553,
      "step": 3392
    },
    {
      "epoch": 0.25679012345679014,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9217689657137187e-05,
      "loss": 0.2974,
      "step": 3393
    },
    {
      "epoch": 0.25686580578023743,
      "grad_norm": 0.78125,
      "learning_rate": 1.9217227762900624e-05,
      "loss": 0.2921,
      "step": 3394
    },
    {
      "epoch": 0.2569414881036848,
      "grad_norm": 0.94140625,
      "learning_rate": 1.921676573790187e-05,
      "loss": 0.377,
      "step": 3395
    },
    {
      "epoch": 0.2570171704271321,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9216303582147475e-05,
      "loss": 0.3409,
      "step": 3396
    },
    {
      "epoch": 0.25709285275057947,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9215841295644e-05,
      "loss": 0.3405,
      "step": 3397
    },
    {
      "epoch": 0.25716853507402676,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9215378878398e-05,
      "loss": 0.3358,
      "step": 3398
    },
    {
      "epoch": 0.2572442173974741,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9214916330416035e-05,
      "loss": 0.3123,
      "step": 3399
    },
    {
      "epoch": 0.25731989972092145,
      "grad_norm": 0.85546875,
      "learning_rate": 1.921445365170467e-05,
      "loss": 0.3178,
      "step": 3400
    },
    {
      "epoch": 0.25739558204436874,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9213990842270466e-05,
      "loss": 0.354,
      "step": 3401
    },
    {
      "epoch": 0.2574712643678161,
      "grad_norm": 0.81640625,
      "learning_rate": 1.921352790211999e-05,
      "loss": 0.3314,
      "step": 3402
    },
    {
      "epoch": 0.25754694669126343,
      "grad_norm": 2.0,
      "learning_rate": 1.9213064831259814e-05,
      "loss": 0.4013,
      "step": 3403
    },
    {
      "epoch": 0.2576226290147107,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9212601629696503e-05,
      "loss": 0.3125,
      "step": 3404
    },
    {
      "epoch": 0.25769831133815807,
      "grad_norm": 0.84375,
      "learning_rate": 1.921213829743662e-05,
      "loss": 0.3424,
      "step": 3405
    },
    {
      "epoch": 0.2577739936616054,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9211674834486756e-05,
      "loss": 0.3455,
      "step": 3406
    },
    {
      "epoch": 0.25784967598505276,
      "grad_norm": 0.7734375,
      "learning_rate": 1.9211211240853473e-05,
      "loss": 0.2869,
      "step": 3407
    },
    {
      "epoch": 0.25792535830850005,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9210747516543352e-05,
      "loss": 0.3251,
      "step": 3408
    },
    {
      "epoch": 0.2580010406319474,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9210283661562972e-05,
      "loss": 0.3019,
      "step": 3409
    },
    {
      "epoch": 0.25807672295539474,
      "grad_norm": 0.8984375,
      "learning_rate": 1.920981967591891e-05,
      "loss": 0.3625,
      "step": 3410
    },
    {
      "epoch": 0.25815240527884203,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9209355559617754e-05,
      "loss": 0.3231,
      "step": 3411
    },
    {
      "epoch": 0.2582280876022894,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9208891312666085e-05,
      "loss": 0.3114,
      "step": 3412
    },
    {
      "epoch": 0.2583037699257367,
      "grad_norm": 0.890625,
      "learning_rate": 1.920842693507049e-05,
      "loss": 0.3324,
      "step": 3413
    },
    {
      "epoch": 0.25837945224918407,
      "grad_norm": 0.76171875,
      "learning_rate": 1.9207962426837557e-05,
      "loss": 0.3201,
      "step": 3414
    },
    {
      "epoch": 0.25845513457263136,
      "grad_norm": 0.78125,
      "learning_rate": 1.9207497787973874e-05,
      "loss": 0.298,
      "step": 3415
    },
    {
      "epoch": 0.2585308168960787,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9207033018486038e-05,
      "loss": 0.3256,
      "step": 3416
    },
    {
      "epoch": 0.25860649921952605,
      "grad_norm": 0.828125,
      "learning_rate": 1.9206568118380633e-05,
      "loss": 0.3397,
      "step": 3417
    },
    {
      "epoch": 0.25868218154297334,
      "grad_norm": 0.84375,
      "learning_rate": 1.920610308766426e-05,
      "loss": 0.3618,
      "step": 3418
    },
    {
      "epoch": 0.2587578638664207,
      "grad_norm": 0.7578125,
      "learning_rate": 1.9205637926343524e-05,
      "loss": 0.2647,
      "step": 3419
    },
    {
      "epoch": 0.25883354618986804,
      "grad_norm": 0.79296875,
      "learning_rate": 1.920517263442501e-05,
      "loss": 0.3087,
      "step": 3420
    },
    {
      "epoch": 0.2589092285133154,
      "grad_norm": 0.90234375,
      "learning_rate": 1.920470721191533e-05,
      "loss": 0.3513,
      "step": 3421
    },
    {
      "epoch": 0.25898491083676267,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9204241658821078e-05,
      "loss": 0.3379,
      "step": 3422
    },
    {
      "epoch": 0.25906059316021,
      "grad_norm": 0.78125,
      "learning_rate": 1.9203775975148868e-05,
      "loss": 0.3136,
      "step": 3423
    },
    {
      "epoch": 0.25913627548365736,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9203310160905296e-05,
      "loss": 0.3133,
      "step": 3424
    },
    {
      "epoch": 0.25921195780710465,
      "grad_norm": 0.75390625,
      "learning_rate": 1.9202844216096982e-05,
      "loss": 0.3051,
      "step": 3425
    },
    {
      "epoch": 0.259287640130552,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9202378140730524e-05,
      "loss": 0.3493,
      "step": 3426
    },
    {
      "epoch": 0.25936332245399935,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9201911934812544e-05,
      "loss": 0.3361,
      "step": 3427
    },
    {
      "epoch": 0.2594390047774467,
      "grad_norm": 1.0234375,
      "learning_rate": 1.9201445598349653e-05,
      "loss": 0.386,
      "step": 3428
    },
    {
      "epoch": 0.259514687100894,
      "grad_norm": 0.8515625,
      "learning_rate": 1.920097913134846e-05,
      "loss": 0.3646,
      "step": 3429
    },
    {
      "epoch": 0.2595903694243413,
      "grad_norm": 0.8359375,
      "learning_rate": 1.92005125338156e-05,
      "loss": 0.3542,
      "step": 3430
    },
    {
      "epoch": 0.2596660517477887,
      "grad_norm": 0.8203125,
      "learning_rate": 1.920004580575767e-05,
      "loss": 0.3592,
      "step": 3431
    },
    {
      "epoch": 0.25974173407123596,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9199578947181306e-05,
      "loss": 0.3596,
      "step": 3432
    },
    {
      "epoch": 0.2598174163946833,
      "grad_norm": 0.84765625,
      "learning_rate": 1.919911195809313e-05,
      "loss": 0.3434,
      "step": 3433
    },
    {
      "epoch": 0.25989309871813066,
      "grad_norm": 0.82421875,
      "learning_rate": 1.919864483849976e-05,
      "loss": 0.3006,
      "step": 3434
    },
    {
      "epoch": 0.259968781041578,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9198177588407834e-05,
      "loss": 0.3921,
      "step": 3435
    },
    {
      "epoch": 0.2600444633650253,
      "grad_norm": 1.0078125,
      "learning_rate": 1.919771020782397e-05,
      "loss": 0.3751,
      "step": 3436
    },
    {
      "epoch": 0.26012014568847264,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9197242696754803e-05,
      "loss": 0.3162,
      "step": 3437
    },
    {
      "epoch": 0.26019582801192,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9196775055206965e-05,
      "loss": 0.3375,
      "step": 3438
    },
    {
      "epoch": 0.2602715103353673,
      "grad_norm": 2.5,
      "learning_rate": 1.9196307283187092e-05,
      "loss": 0.4579,
      "step": 3439
    },
    {
      "epoch": 0.2603471926588146,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9195839380701818e-05,
      "loss": 0.3193,
      "step": 3440
    },
    {
      "epoch": 0.26042287498226196,
      "grad_norm": 9.375,
      "learning_rate": 1.9195371347757782e-05,
      "loss": 0.3793,
      "step": 3441
    },
    {
      "epoch": 0.2604985573057093,
      "grad_norm": 0.84375,
      "learning_rate": 1.9194903184361623e-05,
      "loss": 0.3094,
      "step": 3442
    },
    {
      "epoch": 0.2605742396291566,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9194434890519985e-05,
      "loss": 0.331,
      "step": 3443
    },
    {
      "epoch": 0.26064992195260395,
      "grad_norm": 0.859375,
      "learning_rate": 1.9193966466239508e-05,
      "loss": 0.3463,
      "step": 3444
    },
    {
      "epoch": 0.2607256042760513,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9193497911526842e-05,
      "loss": 0.3468,
      "step": 3445
    },
    {
      "epoch": 0.2608012865994986,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9193029226388627e-05,
      "loss": 0.3494,
      "step": 3446
    },
    {
      "epoch": 0.26087696892294593,
      "grad_norm": 0.8359375,
      "learning_rate": 1.919256041083152e-05,
      "loss": 0.3368,
      "step": 3447
    },
    {
      "epoch": 0.2609526512463933,
      "grad_norm": 1.0625,
      "learning_rate": 1.9192091464862172e-05,
      "loss": 0.3364,
      "step": 3448
    },
    {
      "epoch": 0.2610283335698406,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9191622388487224e-05,
      "loss": 0.3117,
      "step": 3449
    },
    {
      "epoch": 0.2611040158932879,
      "grad_norm": 0.7890625,
      "learning_rate": 1.9191153181713343e-05,
      "loss": 0.3225,
      "step": 3450
    },
    {
      "epoch": 0.26117969821673526,
      "grad_norm": 0.8125,
      "learning_rate": 1.9190683844547186e-05,
      "loss": 0.3564,
      "step": 3451
    },
    {
      "epoch": 0.2612553805401826,
      "grad_norm": 0.83203125,
      "learning_rate": 1.91902143769954e-05,
      "loss": 0.346,
      "step": 3452
    },
    {
      "epoch": 0.2613310628636299,
      "grad_norm": 0.73828125,
      "learning_rate": 1.9189744779064656e-05,
      "loss": 0.2855,
      "step": 3453
    },
    {
      "epoch": 0.26140674518707724,
      "grad_norm": 0.91015625,
      "learning_rate": 1.918927505076161e-05,
      "loss": 0.3534,
      "step": 3454
    },
    {
      "epoch": 0.2614824275105246,
      "grad_norm": 0.828125,
      "learning_rate": 1.918880519209293e-05,
      "loss": 0.3406,
      "step": 3455
    },
    {
      "epoch": 0.26155810983397193,
      "grad_norm": 0.90625,
      "learning_rate": 1.918833520306528e-05,
      "loss": 0.396,
      "step": 3456
    },
    {
      "epoch": 0.2616337921574192,
      "grad_norm": 1.8515625,
      "learning_rate": 1.9187865083685328e-05,
      "loss": 0.3696,
      "step": 3457
    },
    {
      "epoch": 0.26170947448086657,
      "grad_norm": 0.8828125,
      "learning_rate": 1.918739483395974e-05,
      "loss": 0.3008,
      "step": 3458
    },
    {
      "epoch": 0.2617851568043139,
      "grad_norm": 0.828125,
      "learning_rate": 1.918692445389519e-05,
      "loss": 0.3058,
      "step": 3459
    },
    {
      "epoch": 0.2618608391277612,
      "grad_norm": 0.828125,
      "learning_rate": 1.9186453943498353e-05,
      "loss": 0.3281,
      "step": 3460
    },
    {
      "epoch": 0.26193652145120855,
      "grad_norm": 0.8046875,
      "learning_rate": 1.91859833027759e-05,
      "loss": 0.3408,
      "step": 3461
    },
    {
      "epoch": 0.2620122037746559,
      "grad_norm": 1.15625,
      "learning_rate": 1.9185512531734515e-05,
      "loss": 0.3718,
      "step": 3462
    },
    {
      "epoch": 0.2620878860981032,
      "grad_norm": 0.84375,
      "learning_rate": 1.9185041630380868e-05,
      "loss": 0.324,
      "step": 3463
    },
    {
      "epoch": 0.26216356842155053,
      "grad_norm": 0.859375,
      "learning_rate": 1.9184570598721644e-05,
      "loss": 0.309,
      "step": 3464
    },
    {
      "epoch": 0.2622392507449979,
      "grad_norm": 0.734375,
      "learning_rate": 1.918409943676353e-05,
      "loss": 0.2826,
      "step": 3465
    },
    {
      "epoch": 0.2623149330684452,
      "grad_norm": 0.875,
      "learning_rate": 1.9183628144513196e-05,
      "loss": 0.3595,
      "step": 3466
    },
    {
      "epoch": 0.2623906153918925,
      "grad_norm": 0.828125,
      "learning_rate": 1.9183156721977344e-05,
      "loss": 0.3282,
      "step": 3467
    },
    {
      "epoch": 0.26246629771533986,
      "grad_norm": 0.90625,
      "learning_rate": 1.918268516916265e-05,
      "loss": 0.4316,
      "step": 3468
    },
    {
      "epoch": 0.2625419800387872,
      "grad_norm": 1.1328125,
      "learning_rate": 1.9182213486075813e-05,
      "loss": 0.3609,
      "step": 3469
    },
    {
      "epoch": 0.2626176623622345,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9181741672723516e-05,
      "loss": 0.3769,
      "step": 3470
    },
    {
      "epoch": 0.26269334468568184,
      "grad_norm": 0.77734375,
      "learning_rate": 1.9181269729112463e-05,
      "loss": 0.3299,
      "step": 3471
    },
    {
      "epoch": 0.2627690270091292,
      "grad_norm": 0.8359375,
      "learning_rate": 1.918079765524934e-05,
      "loss": 0.3445,
      "step": 3472
    },
    {
      "epoch": 0.26284470933257653,
      "grad_norm": 0.77734375,
      "learning_rate": 1.918032545114085e-05,
      "loss": 0.2859,
      "step": 3473
    },
    {
      "epoch": 0.2629203916560238,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9179853116793684e-05,
      "loss": 0.346,
      "step": 3474
    },
    {
      "epoch": 0.26299607397947117,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9179380652214553e-05,
      "loss": 0.3517,
      "step": 3475
    },
    {
      "epoch": 0.2630717563029185,
      "grad_norm": 0.8125,
      "learning_rate": 1.9178908057410154e-05,
      "loss": 0.3123,
      "step": 3476
    },
    {
      "epoch": 0.2631474386263658,
      "grad_norm": 1.5546875,
      "learning_rate": 1.9178435332387192e-05,
      "loss": 0.4312,
      "step": 3477
    },
    {
      "epoch": 0.26322312094981315,
      "grad_norm": 0.8125,
      "learning_rate": 1.9177962477152373e-05,
      "loss": 0.3158,
      "step": 3478
    },
    {
      "epoch": 0.2632988032732605,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9177489491712408e-05,
      "loss": 0.3793,
      "step": 3479
    },
    {
      "epoch": 0.26337448559670784,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9177016376074006e-05,
      "loss": 0.3622,
      "step": 3480
    },
    {
      "epoch": 0.26345016792015513,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9176543130243875e-05,
      "loss": 0.3216,
      "step": 3481
    },
    {
      "epoch": 0.2635258502436025,
      "grad_norm": 0.78125,
      "learning_rate": 1.9176069754228734e-05,
      "loss": 0.3028,
      "step": 3482
    },
    {
      "epoch": 0.2636015325670498,
      "grad_norm": 0.796875,
      "learning_rate": 1.9175596248035298e-05,
      "loss": 0.3392,
      "step": 3483
    },
    {
      "epoch": 0.2636772148904971,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9175122611670286e-05,
      "loss": 0.3582,
      "step": 3484
    },
    {
      "epoch": 0.26375289721394446,
      "grad_norm": 0.88671875,
      "learning_rate": 1.917464884514041e-05,
      "loss": 0.3859,
      "step": 3485
    },
    {
      "epoch": 0.2638285795373918,
      "grad_norm": 0.78125,
      "learning_rate": 1.91741749484524e-05,
      "loss": 0.3057,
      "step": 3486
    },
    {
      "epoch": 0.26390426186083915,
      "grad_norm": 0.8984375,
      "learning_rate": 1.917370092161297e-05,
      "loss": 0.3691,
      "step": 3487
    },
    {
      "epoch": 0.26397994418428644,
      "grad_norm": 1.8359375,
      "learning_rate": 1.917322676462885e-05,
      "loss": 0.4142,
      "step": 3488
    },
    {
      "epoch": 0.2640556265077338,
      "grad_norm": 0.8828125,
      "learning_rate": 1.917275247750677e-05,
      "loss": 0.3499,
      "step": 3489
    },
    {
      "epoch": 0.26413130883118113,
      "grad_norm": 0.91015625,
      "learning_rate": 1.9172278060253454e-05,
      "loss": 0.3897,
      "step": 3490
    },
    {
      "epoch": 0.2642069911546284,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9171803512875634e-05,
      "loss": 0.3466,
      "step": 3491
    },
    {
      "epoch": 0.26428267347807577,
      "grad_norm": 0.8515625,
      "learning_rate": 1.917132883538004e-05,
      "loss": 0.336,
      "step": 3492
    },
    {
      "epoch": 0.2643583558015231,
      "grad_norm": 0.82421875,
      "learning_rate": 1.917085402777341e-05,
      "loss": 0.3743,
      "step": 3493
    },
    {
      "epoch": 0.26443403812497046,
      "grad_norm": 1.1171875,
      "learning_rate": 1.917037909006248e-05,
      "loss": 0.3712,
      "step": 3494
    },
    {
      "epoch": 0.26450972044841775,
      "grad_norm": 0.9609375,
      "learning_rate": 1.916990402225398e-05,
      "loss": 0.4192,
      "step": 3495
    },
    {
      "epoch": 0.2645854027718651,
      "grad_norm": 0.859375,
      "learning_rate": 1.916942882435466e-05,
      "loss": 0.3431,
      "step": 3496
    },
    {
      "epoch": 0.26466108509531244,
      "grad_norm": 0.890625,
      "learning_rate": 1.9168953496371255e-05,
      "loss": 0.3728,
      "step": 3497
    },
    {
      "epoch": 0.26473676741875973,
      "grad_norm": 0.8046875,
      "learning_rate": 1.916847803831051e-05,
      "loss": 0.3147,
      "step": 3498
    },
    {
      "epoch": 0.2648124497422071,
      "grad_norm": 1.0078125,
      "learning_rate": 1.916800245017917e-05,
      "loss": 0.352,
      "step": 3499
    },
    {
      "epoch": 0.2648881320656544,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9167526731983982e-05,
      "loss": 0.3598,
      "step": 3500
    },
    {
      "epoch": 0.26496381438910177,
      "grad_norm": 0.98046875,
      "learning_rate": 1.9167050883731698e-05,
      "loss": 0.2833,
      "step": 3501
    },
    {
      "epoch": 0.26503949671254906,
      "grad_norm": 0.76171875,
      "learning_rate": 1.916657490542906e-05,
      "loss": 0.2701,
      "step": 3502
    },
    {
      "epoch": 0.2651151790359964,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9166098797082836e-05,
      "loss": 0.3313,
      "step": 3503
    },
    {
      "epoch": 0.26519086135944375,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9165622558699763e-05,
      "loss": 0.3343,
      "step": 3504
    },
    {
      "epoch": 0.26526654368289104,
      "grad_norm": 0.7890625,
      "learning_rate": 1.916514619028661e-05,
      "loss": 0.3079,
      "step": 3505
    },
    {
      "epoch": 0.2653422260063384,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9164669691850126e-05,
      "loss": 0.3342,
      "step": 3506
    },
    {
      "epoch": 0.26541790832978573,
      "grad_norm": 0.90625,
      "learning_rate": 1.9164193063397078e-05,
      "loss": 0.3205,
      "step": 3507
    },
    {
      "epoch": 0.2654935906532331,
      "grad_norm": 0.8125,
      "learning_rate": 1.9163716304934227e-05,
      "loss": 0.3251,
      "step": 3508
    },
    {
      "epoch": 0.26556927297668037,
      "grad_norm": 0.8125,
      "learning_rate": 1.9163239416468333e-05,
      "loss": 0.3232,
      "step": 3509
    },
    {
      "epoch": 0.2656449553001277,
      "grad_norm": 0.7734375,
      "learning_rate": 1.9162762398006164e-05,
      "loss": 0.3025,
      "step": 3510
    },
    {
      "epoch": 0.26572063762357506,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9162285249554484e-05,
      "loss": 0.3103,
      "step": 3511
    },
    {
      "epoch": 0.26579631994702235,
      "grad_norm": 0.8828125,
      "learning_rate": 1.9161807971120065e-05,
      "loss": 0.3533,
      "step": 3512
    },
    {
      "epoch": 0.2658720022704697,
      "grad_norm": 0.859375,
      "learning_rate": 1.916133056270968e-05,
      "loss": 0.3486,
      "step": 3513
    },
    {
      "epoch": 0.26594768459391704,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9160853024330096e-05,
      "loss": 0.3093,
      "step": 3514
    },
    {
      "epoch": 0.2660233669173644,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9160375355988095e-05,
      "loss": 0.3586,
      "step": 3515
    },
    {
      "epoch": 0.2660990492408117,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9159897557690448e-05,
      "loss": 0.3416,
      "step": 3516
    },
    {
      "epoch": 0.266174731564259,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9159419629443936e-05,
      "loss": 0.3433,
      "step": 3517
    },
    {
      "epoch": 0.26625041388770637,
      "grad_norm": 0.83984375,
      "learning_rate": 1.915894157125534e-05,
      "loss": 0.3405,
      "step": 3518
    },
    {
      "epoch": 0.26632609621115366,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9158463383131435e-05,
      "loss": 0.3213,
      "step": 3519
    },
    {
      "epoch": 0.266401778534601,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9157985065079016e-05,
      "loss": 0.2926,
      "step": 3520
    },
    {
      "epoch": 0.26647746085804835,
      "grad_norm": 0.8828125,
      "learning_rate": 1.915750661710486e-05,
      "loss": 0.3691,
      "step": 3521
    },
    {
      "epoch": 0.26655314318149564,
      "grad_norm": 0.859375,
      "learning_rate": 1.915702803921576e-05,
      "loss": 0.3595,
      "step": 3522
    },
    {
      "epoch": 0.266628825504943,
      "grad_norm": 0.828125,
      "learning_rate": 1.9156549331418503e-05,
      "loss": 0.2666,
      "step": 3523
    },
    {
      "epoch": 0.26670450782839034,
      "grad_norm": 0.81640625,
      "learning_rate": 1.915607049371988e-05,
      "loss": 0.3272,
      "step": 3524
    },
    {
      "epoch": 0.2667801901518377,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9155591526126684e-05,
      "loss": 0.3617,
      "step": 3525
    },
    {
      "epoch": 0.26685587247528497,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9155112428645714e-05,
      "loss": 0.3157,
      "step": 3526
    },
    {
      "epoch": 0.2669315547987323,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9154633201283758e-05,
      "loss": 0.32,
      "step": 3527
    },
    {
      "epoch": 0.26700723712217966,
      "grad_norm": 0.7890625,
      "learning_rate": 1.9154153844047622e-05,
      "loss": 0.3178,
      "step": 3528
    },
    {
      "epoch": 0.26708291944562695,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9153674356944107e-05,
      "loss": 0.3362,
      "step": 3529
    },
    {
      "epoch": 0.2671586017690743,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9153194739980012e-05,
      "loss": 0.326,
      "step": 3530
    },
    {
      "epoch": 0.26723428409252165,
      "grad_norm": 0.8125,
      "learning_rate": 1.915271499316214e-05,
      "loss": 0.3442,
      "step": 3531
    },
    {
      "epoch": 0.267309966415969,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9152235116497305e-05,
      "loss": 0.3268,
      "step": 3532
    },
    {
      "epoch": 0.2673856487394163,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9151755109992302e-05,
      "loss": 0.3175,
      "step": 3533
    },
    {
      "epoch": 0.2674613310628636,
      "grad_norm": 0.796875,
      "learning_rate": 1.9151274973653954e-05,
      "loss": 0.276,
      "step": 3534
    },
    {
      "epoch": 0.267537013386311,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9150794707489063e-05,
      "loss": 0.3315,
      "step": 3535
    },
    {
      "epoch": 0.26761269570975826,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9150314311504444e-05,
      "loss": 0.3197,
      "step": 3536
    },
    {
      "epoch": 0.2676883780332056,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9149833785706916e-05,
      "loss": 0.3195,
      "step": 3537
    },
    {
      "epoch": 0.26776406035665296,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9149353130103297e-05,
      "loss": 0.3279,
      "step": 3538
    },
    {
      "epoch": 0.2678397426801003,
      "grad_norm": 0.8671875,
      "learning_rate": 1.91488723447004e-05,
      "loss": 0.3863,
      "step": 3539
    },
    {
      "epoch": 0.2679154250035476,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9148391429505047e-05,
      "loss": 0.3184,
      "step": 3540
    },
    {
      "epoch": 0.26799110732699494,
      "grad_norm": 0.875,
      "learning_rate": 1.9147910384524065e-05,
      "loss": 0.3725,
      "step": 3541
    },
    {
      "epoch": 0.2680667896504423,
      "grad_norm": 0.796875,
      "learning_rate": 1.9147429209764276e-05,
      "loss": 0.3231,
      "step": 3542
    },
    {
      "epoch": 0.2681424719738896,
      "grad_norm": 1.7578125,
      "learning_rate": 1.9146947905232506e-05,
      "loss": 0.3941,
      "step": 3543
    },
    {
      "epoch": 0.2682181542973369,
      "grad_norm": 0.8515625,
      "learning_rate": 1.914646647093558e-05,
      "loss": 0.3385,
      "step": 3544
    },
    {
      "epoch": 0.26829383662078427,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9145984906880333e-05,
      "loss": 0.3311,
      "step": 3545
    },
    {
      "epoch": 0.2683695189442316,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9145503213073597e-05,
      "loss": 0.3333,
      "step": 3546
    },
    {
      "epoch": 0.2684452012676789,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9145021389522203e-05,
      "loss": 0.3481,
      "step": 3547
    },
    {
      "epoch": 0.26852088359112625,
      "grad_norm": 0.80859375,
      "learning_rate": 1.914453943623299e-05,
      "loss": 0.2982,
      "step": 3548
    },
    {
      "epoch": 0.2685965659145736,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9144057353212787e-05,
      "loss": 0.3111,
      "step": 3549
    },
    {
      "epoch": 0.2686722482380209,
      "grad_norm": 0.84375,
      "learning_rate": 1.914357514046844e-05,
      "loss": 0.3548,
      "step": 3550
    },
    {
      "epoch": 0.26874793056146823,
      "grad_norm": 0.84375,
      "learning_rate": 1.914309279800679e-05,
      "loss": 0.3059,
      "step": 3551
    },
    {
      "epoch": 0.2688236128849156,
      "grad_norm": 0.8203125,
      "learning_rate": 1.914261032583468e-05,
      "loss": 0.3286,
      "step": 3552
    },
    {
      "epoch": 0.2688992952083629,
      "grad_norm": 0.84375,
      "learning_rate": 1.914212772395895e-05,
      "loss": 0.3495,
      "step": 3553
    },
    {
      "epoch": 0.2689749775318102,
      "grad_norm": 0.85546875,
      "learning_rate": 1.914164499238645e-05,
      "loss": 0.3657,
      "step": 3554
    },
    {
      "epoch": 0.26905065985525756,
      "grad_norm": 1.5,
      "learning_rate": 1.9141162131124032e-05,
      "loss": 0.4043,
      "step": 3555
    },
    {
      "epoch": 0.2691263421787049,
      "grad_norm": 0.75,
      "learning_rate": 1.914067914017854e-05,
      "loss": 0.2963,
      "step": 3556
    },
    {
      "epoch": 0.2692020245021522,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9140196019556828e-05,
      "loss": 0.3415,
      "step": 3557
    },
    {
      "epoch": 0.26927770682559954,
      "grad_norm": 0.82421875,
      "learning_rate": 1.913971276926575e-05,
      "loss": 0.3306,
      "step": 3558
    },
    {
      "epoch": 0.2693533891490469,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9139229389312162e-05,
      "loss": 0.3612,
      "step": 3559
    },
    {
      "epoch": 0.26942907147249423,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9138745879702923e-05,
      "loss": 0.3415,
      "step": 3560
    },
    {
      "epoch": 0.2695047537959415,
      "grad_norm": 0.79296875,
      "learning_rate": 1.913826224044489e-05,
      "loss": 0.3244,
      "step": 3561
    },
    {
      "epoch": 0.26958043611938887,
      "grad_norm": 0.765625,
      "learning_rate": 1.9137778471544927e-05,
      "loss": 0.2896,
      "step": 3562
    },
    {
      "epoch": 0.2696561184428362,
      "grad_norm": 0.76171875,
      "learning_rate": 1.9137294573009895e-05,
      "loss": 0.2842,
      "step": 3563
    },
    {
      "epoch": 0.2697318007662835,
      "grad_norm": 0.8125,
      "learning_rate": 1.913681054484666e-05,
      "loss": 0.3486,
      "step": 3564
    },
    {
      "epoch": 0.26980748308973085,
      "grad_norm": 1.1796875,
      "learning_rate": 1.9136326387062086e-05,
      "loss": 0.3821,
      "step": 3565
    },
    {
      "epoch": 0.2698831654131782,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9135842099663044e-05,
      "loss": 0.2991,
      "step": 3566
    },
    {
      "epoch": 0.26995884773662554,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9135357682656404e-05,
      "loss": 0.3036,
      "step": 3567
    },
    {
      "epoch": 0.27003453006007283,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9134873136049038e-05,
      "loss": 0.3444,
      "step": 3568
    },
    {
      "epoch": 0.2701102123835202,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9134388459847822e-05,
      "loss": 0.3732,
      "step": 3569
    },
    {
      "epoch": 0.2701858947069675,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9133903654059633e-05,
      "loss": 0.3581,
      "step": 3570
    },
    {
      "epoch": 0.2702615770304148,
      "grad_norm": 0.77734375,
      "learning_rate": 1.9133418718691343e-05,
      "loss": 0.3176,
      "step": 3571
    },
    {
      "epoch": 0.27033725935386216,
      "grad_norm": 0.78515625,
      "learning_rate": 1.9132933653749835e-05,
      "loss": 0.3054,
      "step": 3572
    },
    {
      "epoch": 0.2704129416773095,
      "grad_norm": 0.8125,
      "learning_rate": 1.913244845924199e-05,
      "loss": 0.3327,
      "step": 3573
    },
    {
      "epoch": 0.27048862400075685,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9131963135174693e-05,
      "loss": 0.3758,
      "step": 3574
    },
    {
      "epoch": 0.27056430632420414,
      "grad_norm": 0.765625,
      "learning_rate": 1.9131477681554827e-05,
      "loss": 0.3331,
      "step": 3575
    },
    {
      "epoch": 0.2706399886476515,
      "grad_norm": 0.7890625,
      "learning_rate": 1.9130992098389282e-05,
      "loss": 0.3256,
      "step": 3576
    },
    {
      "epoch": 0.27071567097109883,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9130506385684945e-05,
      "loss": 0.3184,
      "step": 3577
    },
    {
      "epoch": 0.2707913532945461,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9130020543448705e-05,
      "loss": 0.3714,
      "step": 3578
    },
    {
      "epoch": 0.27086703561799347,
      "grad_norm": 0.828125,
      "learning_rate": 1.9129534571687456e-05,
      "loss": 0.3463,
      "step": 3579
    },
    {
      "epoch": 0.2709427179414408,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9129048470408093e-05,
      "loss": 0.3141,
      "step": 3580
    },
    {
      "epoch": 0.27101840026488816,
      "grad_norm": 0.78515625,
      "learning_rate": 1.9128562239617507e-05,
      "loss": 0.3135,
      "step": 3581
    },
    {
      "epoch": 0.27109408258833545,
      "grad_norm": 0.78515625,
      "learning_rate": 1.9128075879322608e-05,
      "loss": 0.3036,
      "step": 3582
    },
    {
      "epoch": 0.2711697649117828,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9127589389530282e-05,
      "loss": 0.3237,
      "step": 3583
    },
    {
      "epoch": 0.27124544723523014,
      "grad_norm": 0.78125,
      "learning_rate": 1.912710277024744e-05,
      "loss": 0.3048,
      "step": 3584
    },
    {
      "epoch": 0.27132112955867743,
      "grad_norm": 0.86328125,
      "learning_rate": 1.912661602148098e-05,
      "loss": 0.3779,
      "step": 3585
    },
    {
      "epoch": 0.2713968118821248,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9126129143237814e-05,
      "loss": 0.3675,
      "step": 3586
    },
    {
      "epoch": 0.2714724942055721,
      "grad_norm": 1.21875,
      "learning_rate": 1.9125642135524845e-05,
      "loss": 0.3285,
      "step": 3587
    },
    {
      "epoch": 0.2715481765290194,
      "grad_norm": 0.84375,
      "learning_rate": 1.912515499834898e-05,
      "loss": 0.3454,
      "step": 3588
    },
    {
      "epoch": 0.27162385885246676,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9124667731717133e-05,
      "loss": 0.3721,
      "step": 3589
    },
    {
      "epoch": 0.2716995411759141,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9124180335636213e-05,
      "loss": 0.3174,
      "step": 3590
    },
    {
      "epoch": 0.27177522349936145,
      "grad_norm": 0.83984375,
      "learning_rate": 1.912369281011314e-05,
      "loss": 0.3793,
      "step": 3591
    },
    {
      "epoch": 0.27185090582280874,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9123205155154827e-05,
      "loss": 0.3543,
      "step": 3592
    },
    {
      "epoch": 0.2719265881462561,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9122717370768194e-05,
      "loss": 0.3707,
      "step": 3593
    },
    {
      "epoch": 0.27200227046970343,
      "grad_norm": 0.7890625,
      "learning_rate": 1.9122229456960157e-05,
      "loss": 0.3113,
      "step": 3594
    },
    {
      "epoch": 0.2720779527931507,
      "grad_norm": 0.84375,
      "learning_rate": 1.9121741413737643e-05,
      "loss": 0.3273,
      "step": 3595
    },
    {
      "epoch": 0.27215363511659807,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9121253241107573e-05,
      "loss": 0.3023,
      "step": 3596
    },
    {
      "epoch": 0.2722293174400454,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9120764939076873e-05,
      "loss": 0.2716,
      "step": 3597
    },
    {
      "epoch": 0.27230499976349276,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9120276507652475e-05,
      "loss": 0.3442,
      "step": 3598
    },
    {
      "epoch": 0.27238068208694005,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9119787946841296e-05,
      "loss": 0.3484,
      "step": 3599
    },
    {
      "epoch": 0.2724563644103874,
      "grad_norm": 0.78515625,
      "learning_rate": 1.911929925665028e-05,
      "loss": 0.2981,
      "step": 3600
    },
    {
      "epoch": 0.27253204673383474,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9118810437086353e-05,
      "loss": 0.3322,
      "step": 3601
    },
    {
      "epoch": 0.27260772905728203,
      "grad_norm": 0.83984375,
      "learning_rate": 1.911832148815645e-05,
      "loss": 0.3451,
      "step": 3602
    },
    {
      "epoch": 0.2726834113807294,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9117832409867513e-05,
      "loss": 0.3847,
      "step": 3603
    },
    {
      "epoch": 0.2727590937041767,
      "grad_norm": 0.8515625,
      "learning_rate": 1.911734320222648e-05,
      "loss": 0.3487,
      "step": 3604
    },
    {
      "epoch": 0.27283477602762407,
      "grad_norm": 0.78515625,
      "learning_rate": 1.9116853865240278e-05,
      "loss": 0.3212,
      "step": 3605
    },
    {
      "epoch": 0.27291045835107136,
      "grad_norm": 0.84375,
      "learning_rate": 1.9116364398915866e-05,
      "loss": 0.346,
      "step": 3606
    },
    {
      "epoch": 0.2729861406745187,
      "grad_norm": 0.8125,
      "learning_rate": 1.911587480326018e-05,
      "loss": 0.3305,
      "step": 3607
    },
    {
      "epoch": 0.27306182299796605,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9115385078280158e-05,
      "loss": 0.3175,
      "step": 3608
    },
    {
      "epoch": 0.27313750532141334,
      "grad_norm": 1.109375,
      "learning_rate": 1.9114895223982767e-05,
      "loss": 0.3206,
      "step": 3609
    },
    {
      "epoch": 0.2732131876448607,
      "grad_norm": 0.76953125,
      "learning_rate": 1.911440524037494e-05,
      "loss": 0.2825,
      "step": 3610
    },
    {
      "epoch": 0.27328886996830803,
      "grad_norm": 0.8125,
      "learning_rate": 1.9113915127463637e-05,
      "loss": 0.3174,
      "step": 3611
    },
    {
      "epoch": 0.2733645522917554,
      "grad_norm": 3.484375,
      "learning_rate": 1.9113424885255805e-05,
      "loss": 0.4233,
      "step": 3612
    },
    {
      "epoch": 0.27344023461520267,
      "grad_norm": 2.28125,
      "learning_rate": 1.9112934513758403e-05,
      "loss": 0.4382,
      "step": 3613
    },
    {
      "epoch": 0.27351591693865,
      "grad_norm": 1.28125,
      "learning_rate": 1.9112444012978385e-05,
      "loss": 0.4053,
      "step": 3614
    },
    {
      "epoch": 0.27359159926209736,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9111953382922715e-05,
      "loss": 0.3109,
      "step": 3615
    },
    {
      "epoch": 0.27366728158554465,
      "grad_norm": 0.84375,
      "learning_rate": 1.9111462623598345e-05,
      "loss": 0.3295,
      "step": 3616
    },
    {
      "epoch": 0.273742963908992,
      "grad_norm": 0.859375,
      "learning_rate": 1.9110971735012245e-05,
      "loss": 0.3547,
      "step": 3617
    },
    {
      "epoch": 0.27381864623243934,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9110480717171376e-05,
      "loss": 0.3309,
      "step": 3618
    },
    {
      "epoch": 0.2738943285558867,
      "grad_norm": 0.77734375,
      "learning_rate": 1.9109989570082703e-05,
      "loss": 0.3028,
      "step": 3619
    },
    {
      "epoch": 0.273970010879334,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9109498293753196e-05,
      "loss": 0.3596,
      "step": 3620
    },
    {
      "epoch": 0.2740456932027813,
      "grad_norm": 0.79296875,
      "learning_rate": 1.910900688818982e-05,
      "loss": 0.3166,
      "step": 3621
    },
    {
      "epoch": 0.27412137552622867,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9108515353399553e-05,
      "loss": 0.3585,
      "step": 3622
    },
    {
      "epoch": 0.27419705784967596,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9108023689389362e-05,
      "loss": 0.3946,
      "step": 3623
    },
    {
      "epoch": 0.2742727401731233,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9107531896166226e-05,
      "loss": 0.3514,
      "step": 3624
    },
    {
      "epoch": 0.27434842249657065,
      "grad_norm": 0.796875,
      "learning_rate": 1.9107039973737125e-05,
      "loss": 0.3177,
      "step": 3625
    },
    {
      "epoch": 0.274424104820018,
      "grad_norm": 0.796875,
      "learning_rate": 1.9106547922109027e-05,
      "loss": 0.3048,
      "step": 3626
    },
    {
      "epoch": 0.2744997871434653,
      "grad_norm": 0.875,
      "learning_rate": 1.9106055741288924e-05,
      "loss": 0.3512,
      "step": 3627
    },
    {
      "epoch": 0.27457546946691264,
      "grad_norm": 0.78515625,
      "learning_rate": 1.9105563431283794e-05,
      "loss": 0.3288,
      "step": 3628
    },
    {
      "epoch": 0.27465115179036,
      "grad_norm": 0.81640625,
      "learning_rate": 1.910507099210062e-05,
      "loss": 0.3147,
      "step": 3629
    },
    {
      "epoch": 0.2747268341138073,
      "grad_norm": 0.7578125,
      "learning_rate": 1.910457842374639e-05,
      "loss": 0.2823,
      "step": 3630
    },
    {
      "epoch": 0.2748025164372546,
      "grad_norm": 0.7734375,
      "learning_rate": 1.9104085726228086e-05,
      "loss": 0.264,
      "step": 3631
    },
    {
      "epoch": 0.27487819876070196,
      "grad_norm": 0.7890625,
      "learning_rate": 1.910359289955271e-05,
      "loss": 0.3082,
      "step": 3632
    },
    {
      "epoch": 0.2749538810841493,
      "grad_norm": 0.84375,
      "learning_rate": 1.910309994372724e-05,
      "loss": 0.3465,
      "step": 3633
    },
    {
      "epoch": 0.2750295634075966,
      "grad_norm": 0.7734375,
      "learning_rate": 1.910260685875868e-05,
      "loss": 0.302,
      "step": 3634
    },
    {
      "epoch": 0.27510524573104395,
      "grad_norm": 0.828125,
      "learning_rate": 1.910211364465402e-05,
      "loss": 0.3529,
      "step": 3635
    },
    {
      "epoch": 0.2751809280544913,
      "grad_norm": 0.81640625,
      "learning_rate": 1.910162030142026e-05,
      "loss": 0.3137,
      "step": 3636
    },
    {
      "epoch": 0.2752566103779386,
      "grad_norm": 0.796875,
      "learning_rate": 1.9101126829064394e-05,
      "loss": 0.3418,
      "step": 3637
    },
    {
      "epoch": 0.27533229270138593,
      "grad_norm": 0.83203125,
      "learning_rate": 1.910063322759343e-05,
      "loss": 0.3604,
      "step": 3638
    },
    {
      "epoch": 0.2754079750248333,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9100139497014363e-05,
      "loss": 0.3658,
      "step": 3639
    },
    {
      "epoch": 0.2754836573482806,
      "grad_norm": 0.828125,
      "learning_rate": 1.9099645637334202e-05,
      "loss": 0.3461,
      "step": 3640
    },
    {
      "epoch": 0.2755593396717279,
      "grad_norm": 0.8515625,
      "learning_rate": 1.909915164855995e-05,
      "loss": 0.3721,
      "step": 3641
    },
    {
      "epoch": 0.27563502199517526,
      "grad_norm": 0.7890625,
      "learning_rate": 1.909865753069862e-05,
      "loss": 0.3123,
      "step": 3642
    },
    {
      "epoch": 0.2757107043186226,
      "grad_norm": 0.828125,
      "learning_rate": 1.9098163283757215e-05,
      "loss": 0.3064,
      "step": 3643
    },
    {
      "epoch": 0.2757863866420699,
      "grad_norm": 0.875,
      "learning_rate": 1.9097668907742757e-05,
      "loss": 0.3687,
      "step": 3644
    },
    {
      "epoch": 0.27586206896551724,
      "grad_norm": 0.8359375,
      "learning_rate": 1.909717440266225e-05,
      "loss": 0.3376,
      "step": 3645
    },
    {
      "epoch": 0.2759377512889646,
      "grad_norm": 0.890625,
      "learning_rate": 1.9096679768522712e-05,
      "loss": 0.3476,
      "step": 3646
    },
    {
      "epoch": 0.2760134336124119,
      "grad_norm": 0.828125,
      "learning_rate": 1.9096185005331162e-05,
      "loss": 0.3267,
      "step": 3647
    },
    {
      "epoch": 0.2760891159358592,
      "grad_norm": 0.91015625,
      "learning_rate": 1.909569011309462e-05,
      "loss": 0.3568,
      "step": 3648
    },
    {
      "epoch": 0.27616479825930657,
      "grad_norm": 0.80078125,
      "learning_rate": 1.90951950918201e-05,
      "loss": 0.3055,
      "step": 3649
    },
    {
      "epoch": 0.2762404805827539,
      "grad_norm": 0.84375,
      "learning_rate": 1.9094699941514634e-05,
      "loss": 0.3628,
      "step": 3650
    },
    {
      "epoch": 0.2763161629062012,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9094204662185244e-05,
      "loss": 0.3769,
      "step": 3651
    },
    {
      "epoch": 0.27639184522964855,
      "grad_norm": 0.87890625,
      "learning_rate": 1.909370925383895e-05,
      "loss": 0.4016,
      "step": 3652
    },
    {
      "epoch": 0.2764675275530959,
      "grad_norm": 0.859375,
      "learning_rate": 1.9093213716482786e-05,
      "loss": 0.345,
      "step": 3653
    },
    {
      "epoch": 0.2765432098765432,
      "grad_norm": 0.765625,
      "learning_rate": 1.909271805012378e-05,
      "loss": 0.2836,
      "step": 3654
    },
    {
      "epoch": 0.27661889219999053,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9092222254768967e-05,
      "loss": 0.3487,
      "step": 3655
    },
    {
      "epoch": 0.2766945745234379,
      "grad_norm": 0.79296875,
      "learning_rate": 1.909172633042538e-05,
      "loss": 0.3232,
      "step": 3656
    },
    {
      "epoch": 0.2767702568468852,
      "grad_norm": 0.93359375,
      "learning_rate": 1.909123027710005e-05,
      "loss": 0.3472,
      "step": 3657
    },
    {
      "epoch": 0.2768459391703325,
      "grad_norm": 3.015625,
      "learning_rate": 1.909073409480002e-05,
      "loss": 0.4214,
      "step": 3658
    },
    {
      "epoch": 0.27692162149377986,
      "grad_norm": 0.765625,
      "learning_rate": 1.9090237783532324e-05,
      "loss": 0.2957,
      "step": 3659
    },
    {
      "epoch": 0.2769973038172272,
      "grad_norm": 0.8125,
      "learning_rate": 1.9089741343304005e-05,
      "loss": 0.3482,
      "step": 3660
    },
    {
      "epoch": 0.2770729861406745,
      "grad_norm": 0.9140625,
      "learning_rate": 1.908924477412211e-05,
      "loss": 0.3926,
      "step": 3661
    },
    {
      "epoch": 0.27714866846412184,
      "grad_norm": 0.84375,
      "learning_rate": 1.908874807599368e-05,
      "loss": 0.3203,
      "step": 3662
    },
    {
      "epoch": 0.2772243507875692,
      "grad_norm": 0.8828125,
      "learning_rate": 1.908825124892576e-05,
      "loss": 0.3824,
      "step": 3663
    },
    {
      "epoch": 0.27730003311101653,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9087754292925397e-05,
      "loss": 0.35,
      "step": 3664
    },
    {
      "epoch": 0.2773757154344638,
      "grad_norm": 0.85546875,
      "learning_rate": 1.908725720799965e-05,
      "loss": 0.3777,
      "step": 3665
    },
    {
      "epoch": 0.27745139775791117,
      "grad_norm": 0.828125,
      "learning_rate": 1.9086759994155563e-05,
      "loss": 0.3445,
      "step": 3666
    },
    {
      "epoch": 0.2775270800813585,
      "grad_norm": 0.8671875,
      "learning_rate": 1.908626265140019e-05,
      "loss": 0.3742,
      "step": 3667
    },
    {
      "epoch": 0.2776027624048058,
      "grad_norm": 0.8203125,
      "learning_rate": 1.908576517974059e-05,
      "loss": 0.3379,
      "step": 3668
    },
    {
      "epoch": 0.27767844472825315,
      "grad_norm": 0.7890625,
      "learning_rate": 1.908526757918382e-05,
      "loss": 0.3541,
      "step": 3669
    },
    {
      "epoch": 0.2777541270517005,
      "grad_norm": 0.765625,
      "learning_rate": 1.908476984973694e-05,
      "loss": 0.3103,
      "step": 3670
    },
    {
      "epoch": 0.27782980937514784,
      "grad_norm": 0.7734375,
      "learning_rate": 1.9084271991407006e-05,
      "loss": 0.3136,
      "step": 3671
    },
    {
      "epoch": 0.27790549169859513,
      "grad_norm": 0.77734375,
      "learning_rate": 1.908377400420109e-05,
      "loss": 0.2886,
      "step": 3672
    },
    {
      "epoch": 0.2779811740220425,
      "grad_norm": 0.8125,
      "learning_rate": 1.9083275888126248e-05,
      "loss": 0.3603,
      "step": 3673
    },
    {
      "epoch": 0.2780568563454898,
      "grad_norm": 0.80859375,
      "learning_rate": 1.908277764318955e-05,
      "loss": 0.3138,
      "step": 3674
    },
    {
      "epoch": 0.2781325386689371,
      "grad_norm": 0.8125,
      "learning_rate": 1.9082279269398064e-05,
      "loss": 0.3531,
      "step": 3675
    },
    {
      "epoch": 0.27820822099238446,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9081780766758862e-05,
      "loss": 0.3638,
      "step": 3676
    },
    {
      "epoch": 0.2782839033158318,
      "grad_norm": 0.7734375,
      "learning_rate": 1.9081282135279012e-05,
      "loss": 0.3151,
      "step": 3677
    },
    {
      "epoch": 0.27835958563927915,
      "grad_norm": 1.890625,
      "learning_rate": 1.9080783374965594e-05,
      "loss": 0.4538,
      "step": 3678
    },
    {
      "epoch": 0.27843526796272644,
      "grad_norm": 0.81640625,
      "learning_rate": 1.908028448582568e-05,
      "loss": 0.303,
      "step": 3679
    },
    {
      "epoch": 0.2785109502861738,
      "grad_norm": 0.83984375,
      "learning_rate": 1.907978546786635e-05,
      "loss": 0.3483,
      "step": 3680
    },
    {
      "epoch": 0.27858663260962113,
      "grad_norm": 0.81640625,
      "learning_rate": 1.907928632109468e-05,
      "loss": 0.3338,
      "step": 3681
    },
    {
      "epoch": 0.2786623149330684,
      "grad_norm": 1.21875,
      "learning_rate": 1.9078787045517754e-05,
      "loss": 0.3715,
      "step": 3682
    },
    {
      "epoch": 0.27873799725651577,
      "grad_norm": 0.76953125,
      "learning_rate": 1.907828764114265e-05,
      "loss": 0.2833,
      "step": 3683
    },
    {
      "epoch": 0.2788136795799631,
      "grad_norm": 0.7734375,
      "learning_rate": 1.907778810797646e-05,
      "loss": 0.3088,
      "step": 3684
    },
    {
      "epoch": 0.27888936190341046,
      "grad_norm": 0.84375,
      "learning_rate": 1.9077288446026267e-05,
      "loss": 0.33,
      "step": 3685
    },
    {
      "epoch": 0.27896504422685775,
      "grad_norm": 0.93359375,
      "learning_rate": 1.9076788655299163e-05,
      "loss": 0.4103,
      "step": 3686
    },
    {
      "epoch": 0.2790407265503051,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9076288735802233e-05,
      "loss": 0.3283,
      "step": 3687
    },
    {
      "epoch": 0.27911640887375244,
      "grad_norm": 0.7578125,
      "learning_rate": 1.907578868754257e-05,
      "loss": 0.307,
      "step": 3688
    },
    {
      "epoch": 0.27919209119719973,
      "grad_norm": 0.78125,
      "learning_rate": 1.9075288510527273e-05,
      "loss": 0.3091,
      "step": 3689
    },
    {
      "epoch": 0.2792677735206471,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9074788204763438e-05,
      "loss": 0.3467,
      "step": 3690
    },
    {
      "epoch": 0.2793434558440944,
      "grad_norm": 0.89453125,
      "learning_rate": 1.9074287770258154e-05,
      "loss": 0.3474,
      "step": 3691
    },
    {
      "epoch": 0.27941913816754177,
      "grad_norm": 0.734375,
      "learning_rate": 1.9073787207018526e-05,
      "loss": 0.2828,
      "step": 3692
    },
    {
      "epoch": 0.27949482049098906,
      "grad_norm": 3.015625,
      "learning_rate": 1.9073286515051656e-05,
      "loss": 0.3257,
      "step": 3693
    },
    {
      "epoch": 0.2795705028144364,
      "grad_norm": 0.88671875,
      "learning_rate": 1.9072785694364647e-05,
      "loss": 0.3294,
      "step": 3694
    },
    {
      "epoch": 0.27964618513788375,
      "grad_norm": 0.828125,
      "learning_rate": 1.9072284744964603e-05,
      "loss": 0.3422,
      "step": 3695
    },
    {
      "epoch": 0.27972186746133104,
      "grad_norm": 0.72265625,
      "learning_rate": 1.907178366685863e-05,
      "loss": 0.2668,
      "step": 3696
    },
    {
      "epoch": 0.2797975497847784,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9071282460053845e-05,
      "loss": 0.3563,
      "step": 3697
    },
    {
      "epoch": 0.27987323210822573,
      "grad_norm": 8.5625,
      "learning_rate": 1.9070781124557345e-05,
      "loss": 0.3177,
      "step": 3698
    },
    {
      "epoch": 0.2799489144316731,
      "grad_norm": 0.7890625,
      "learning_rate": 1.907027966037625e-05,
      "loss": 0.3262,
      "step": 3699
    },
    {
      "epoch": 0.28002459675512037,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9069778067517672e-05,
      "loss": 0.3018,
      "step": 3700
    },
    {
      "epoch": 0.2801002790785677,
      "grad_norm": 0.82421875,
      "learning_rate": 1.906927634598873e-05,
      "loss": 0.3164,
      "step": 3701
    },
    {
      "epoch": 0.28017596140201506,
      "grad_norm": 0.765625,
      "learning_rate": 1.9068774495796538e-05,
      "loss": 0.3074,
      "step": 3702
    },
    {
      "epoch": 0.28025164372546235,
      "grad_norm": 0.8125,
      "learning_rate": 1.9068272516948218e-05,
      "loss": 0.3161,
      "step": 3703
    },
    {
      "epoch": 0.2803273260489097,
      "grad_norm": 0.81640625,
      "learning_rate": 1.906777040945089e-05,
      "loss": 0.3432,
      "step": 3704
    },
    {
      "epoch": 0.28040300837235704,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9067268173311678e-05,
      "loss": 0.3341,
      "step": 3705
    },
    {
      "epoch": 0.28047869069580433,
      "grad_norm": 1.875,
      "learning_rate": 1.9066765808537708e-05,
      "loss": 0.4098,
      "step": 3706
    },
    {
      "epoch": 0.2805543730192517,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9066263315136107e-05,
      "loss": 0.3384,
      "step": 3707
    },
    {
      "epoch": 0.280630055342699,
      "grad_norm": 0.9296875,
      "learning_rate": 1.9065760693114e-05,
      "loss": 0.3901,
      "step": 3708
    },
    {
      "epoch": 0.28070573766614637,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9065257942478518e-05,
      "loss": 0.3617,
      "step": 3709
    },
    {
      "epoch": 0.28078141998959366,
      "grad_norm": 0.74609375,
      "learning_rate": 1.90647550632368e-05,
      "loss": 0.2876,
      "step": 3710
    },
    {
      "epoch": 0.280857102313041,
      "grad_norm": 1.3125,
      "learning_rate": 1.9064252055395977e-05,
      "loss": 0.4475,
      "step": 3711
    },
    {
      "epoch": 0.28093278463648835,
      "grad_norm": 0.875,
      "learning_rate": 1.9063748918963178e-05,
      "loss": 0.3593,
      "step": 3712
    },
    {
      "epoch": 0.28100846695993564,
      "grad_norm": 0.83984375,
      "learning_rate": 1.906324565394555e-05,
      "loss": 0.326,
      "step": 3713
    },
    {
      "epoch": 0.281084149283383,
      "grad_norm": 0.87109375,
      "learning_rate": 1.906274226035023e-05,
      "loss": 0.3844,
      "step": 3714
    },
    {
      "epoch": 0.28115983160683033,
      "grad_norm": 0.828125,
      "learning_rate": 1.9062238738184357e-05,
      "loss": 0.3403,
      "step": 3715
    },
    {
      "epoch": 0.2812355139302777,
      "grad_norm": 0.75,
      "learning_rate": 1.9061735087455078e-05,
      "loss": 0.3011,
      "step": 3716
    },
    {
      "epoch": 0.28131119625372497,
      "grad_norm": 0.7890625,
      "learning_rate": 1.9061231308169534e-05,
      "loss": 0.2973,
      "step": 3717
    },
    {
      "epoch": 0.2813868785771723,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9060727400334875e-05,
      "loss": 0.3327,
      "step": 3718
    },
    {
      "epoch": 0.28146256090061966,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9060223363958254e-05,
      "loss": 0.2923,
      "step": 3719
    },
    {
      "epoch": 0.28153824322406695,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9059719199046812e-05,
      "loss": 0.3662,
      "step": 3720
    },
    {
      "epoch": 0.2816139255475143,
      "grad_norm": 0.76953125,
      "learning_rate": 1.9059214905607705e-05,
      "loss": 0.2771,
      "step": 3721
    },
    {
      "epoch": 0.28168960787096164,
      "grad_norm": 1.390625,
      "learning_rate": 1.905871048364809e-05,
      "loss": 0.4283,
      "step": 3722
    },
    {
      "epoch": 0.281765290194409,
      "grad_norm": 0.8359375,
      "learning_rate": 1.905820593317512e-05,
      "loss": 0.343,
      "step": 3723
    },
    {
      "epoch": 0.2818409725178563,
      "grad_norm": 0.859375,
      "learning_rate": 1.9057701254195958e-05,
      "loss": 0.3499,
      "step": 3724
    },
    {
      "epoch": 0.2819166548413036,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9057196446717757e-05,
      "loss": 0.3383,
      "step": 3725
    },
    {
      "epoch": 0.281992337164751,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9056691510747686e-05,
      "loss": 0.3737,
      "step": 3726
    },
    {
      "epoch": 0.28206801948819826,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9056186446292895e-05,
      "loss": 0.3673,
      "step": 3727
    },
    {
      "epoch": 0.2821437018116456,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9055681253360568e-05,
      "loss": 0.3292,
      "step": 3728
    },
    {
      "epoch": 0.28221938413509295,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9055175931957858e-05,
      "loss": 0.3187,
      "step": 3729
    },
    {
      "epoch": 0.2822950664585403,
      "grad_norm": 0.828125,
      "learning_rate": 1.905467048209194e-05,
      "loss": 0.343,
      "step": 3730
    },
    {
      "epoch": 0.2823707487819876,
      "grad_norm": 0.81640625,
      "learning_rate": 1.905416490376998e-05,
      "loss": 0.3212,
      "step": 3731
    },
    {
      "epoch": 0.28244643110543494,
      "grad_norm": 0.94140625,
      "learning_rate": 1.9053659196999157e-05,
      "loss": 0.3907,
      "step": 3732
    },
    {
      "epoch": 0.2825221134288823,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9053153361786637e-05,
      "loss": 0.2943,
      "step": 3733
    },
    {
      "epoch": 0.2825977957523296,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9052647398139607e-05,
      "loss": 0.3513,
      "step": 3734
    },
    {
      "epoch": 0.2826734780757769,
      "grad_norm": 0.87109375,
      "learning_rate": 1.9052141306065234e-05,
      "loss": 0.3466,
      "step": 3735
    },
    {
      "epoch": 0.28274916039922426,
      "grad_norm": 0.92578125,
      "learning_rate": 1.9051635085570702e-05,
      "loss": 0.3675,
      "step": 3736
    },
    {
      "epoch": 0.2828248427226716,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9051128736663197e-05,
      "loss": 0.358,
      "step": 3737
    },
    {
      "epoch": 0.2829005250461189,
      "grad_norm": 0.7578125,
      "learning_rate": 1.9050622259349895e-05,
      "loss": 0.3153,
      "step": 3738
    },
    {
      "epoch": 0.28297620736956625,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9050115653637985e-05,
      "loss": 0.3476,
      "step": 3739
    },
    {
      "epoch": 0.2830518896930136,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9049608919534658e-05,
      "loss": 0.3237,
      "step": 3740
    },
    {
      "epoch": 0.2831275720164609,
      "grad_norm": 1.484375,
      "learning_rate": 1.9049102057047095e-05,
      "loss": 0.4218,
      "step": 3741
    },
    {
      "epoch": 0.28320325433990823,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9048595066182495e-05,
      "loss": 0.3511,
      "step": 3742
    },
    {
      "epoch": 0.2832789366633556,
      "grad_norm": 0.859375,
      "learning_rate": 1.9048087946948038e-05,
      "loss": 0.3652,
      "step": 3743
    },
    {
      "epoch": 0.2833546189868029,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9047580699350932e-05,
      "loss": 0.3509,
      "step": 3744
    },
    {
      "epoch": 0.2834303013102502,
      "grad_norm": 0.84375,
      "learning_rate": 1.9047073323398367e-05,
      "loss": 0.3385,
      "step": 3745
    },
    {
      "epoch": 0.28350598363369756,
      "grad_norm": 0.85546875,
      "learning_rate": 1.9046565819097546e-05,
      "loss": 0.3508,
      "step": 3746
    },
    {
      "epoch": 0.2835816659571449,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9046058186455657e-05,
      "loss": 0.3514,
      "step": 3747
    },
    {
      "epoch": 0.2836573482805922,
      "grad_norm": 0.8125,
      "learning_rate": 1.9045550425479912e-05,
      "loss": 0.2954,
      "step": 3748
    },
    {
      "epoch": 0.28373303060403954,
      "grad_norm": 0.80078125,
      "learning_rate": 1.904504253617751e-05,
      "loss": 0.3062,
      "step": 3749
    },
    {
      "epoch": 0.2838087129274869,
      "grad_norm": 0.75390625,
      "learning_rate": 1.904453451855566e-05,
      "loss": 0.2739,
      "step": 3750
    },
    {
      "epoch": 0.28388439525093423,
      "grad_norm": 0.859375,
      "learning_rate": 1.9044026372621568e-05,
      "loss": 0.3714,
      "step": 3751
    },
    {
      "epoch": 0.2839600775743815,
      "grad_norm": 0.91796875,
      "learning_rate": 1.9043518098382437e-05,
      "loss": 0.3837,
      "step": 3752
    },
    {
      "epoch": 0.28403575989782887,
      "grad_norm": 0.984375,
      "learning_rate": 1.9043009695845486e-05,
      "loss": 0.3569,
      "step": 3753
    },
    {
      "epoch": 0.2841114422212762,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9042501165017928e-05,
      "loss": 0.3142,
      "step": 3754
    },
    {
      "epoch": 0.2841871245447235,
      "grad_norm": 0.875,
      "learning_rate": 1.9041992505906965e-05,
      "loss": 0.3707,
      "step": 3755
    },
    {
      "epoch": 0.28426280686817085,
      "grad_norm": 0.78125,
      "learning_rate": 1.9041483718519825e-05,
      "loss": 0.2967,
      "step": 3756
    },
    {
      "epoch": 0.2843384891916182,
      "grad_norm": 0.8125,
      "learning_rate": 1.9040974802863725e-05,
      "loss": 0.3353,
      "step": 3757
    },
    {
      "epoch": 0.28441417151506554,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9040465758945884e-05,
      "loss": 0.3212,
      "step": 3758
    },
    {
      "epoch": 0.28448985383851283,
      "grad_norm": 0.7734375,
      "learning_rate": 1.9039956586773516e-05,
      "loss": 0.3115,
      "step": 3759
    },
    {
      "epoch": 0.2845655361619602,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9039447286353855e-05,
      "loss": 0.3807,
      "step": 3760
    },
    {
      "epoch": 0.2846412184854075,
      "grad_norm": 0.80078125,
      "learning_rate": 1.903893785769412e-05,
      "loss": 0.3337,
      "step": 3761
    },
    {
      "epoch": 0.2847169008088548,
      "grad_norm": 0.84765625,
      "learning_rate": 1.9038428300801542e-05,
      "loss": 0.3775,
      "step": 3762
    },
    {
      "epoch": 0.28479258313230216,
      "grad_norm": 0.7890625,
      "learning_rate": 1.9037918615683345e-05,
      "loss": 0.3268,
      "step": 3763
    },
    {
      "epoch": 0.2848682654557495,
      "grad_norm": 0.75390625,
      "learning_rate": 1.903740880234677e-05,
      "loss": 0.2918,
      "step": 3764
    },
    {
      "epoch": 0.28494394777919685,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9036898860799037e-05,
      "loss": 0.3574,
      "step": 3765
    },
    {
      "epoch": 0.28501963010264414,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9036388791047383e-05,
      "loss": 0.3642,
      "step": 3766
    },
    {
      "epoch": 0.2850953124260915,
      "grad_norm": 0.8359375,
      "learning_rate": 1.903587859309905e-05,
      "loss": 0.3289,
      "step": 3767
    },
    {
      "epoch": 0.28517099474953883,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9035368266961275e-05,
      "loss": 0.3217,
      "step": 3768
    },
    {
      "epoch": 0.2852466770729861,
      "grad_norm": 0.87890625,
      "learning_rate": 1.9034857812641294e-05,
      "loss": 0.3627,
      "step": 3769
    },
    {
      "epoch": 0.28532235939643347,
      "grad_norm": 0.75390625,
      "learning_rate": 1.9034347230146353e-05,
      "loss": 0.292,
      "step": 3770
    },
    {
      "epoch": 0.2853980417198808,
      "grad_norm": 0.859375,
      "learning_rate": 1.903383651948369e-05,
      "loss": 0.3647,
      "step": 3771
    },
    {
      "epoch": 0.2854737240433281,
      "grad_norm": 0.99609375,
      "learning_rate": 1.9033325680660556e-05,
      "loss": 0.3515,
      "step": 3772
    },
    {
      "epoch": 0.28554940636677545,
      "grad_norm": 0.8984375,
      "learning_rate": 1.9032814713684195e-05,
      "loss": 0.3847,
      "step": 3773
    },
    {
      "epoch": 0.2856250886902228,
      "grad_norm": 0.7734375,
      "learning_rate": 1.9032303618561855e-05,
      "loss": 0.2958,
      "step": 3774
    },
    {
      "epoch": 0.28570077101367014,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9031792395300787e-05,
      "loss": 0.3382,
      "step": 3775
    },
    {
      "epoch": 0.28577645333711743,
      "grad_norm": 0.81640625,
      "learning_rate": 1.9031281043908246e-05,
      "loss": 0.3355,
      "step": 3776
    },
    {
      "epoch": 0.2858521356605648,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9030769564391486e-05,
      "loss": 0.3395,
      "step": 3777
    },
    {
      "epoch": 0.2859278179840121,
      "grad_norm": 0.85546875,
      "learning_rate": 1.903025795675776e-05,
      "loss": 0.3265,
      "step": 3778
    },
    {
      "epoch": 0.2860035003074594,
      "grad_norm": 0.79296875,
      "learning_rate": 1.9029746221014333e-05,
      "loss": 0.3432,
      "step": 3779
    },
    {
      "epoch": 0.28607918263090676,
      "grad_norm": 0.97265625,
      "learning_rate": 1.9029234357168454e-05,
      "loss": 0.3859,
      "step": 3780
    },
    {
      "epoch": 0.2861548649543541,
      "grad_norm": 0.76953125,
      "learning_rate": 1.9028722365227396e-05,
      "loss": 0.3002,
      "step": 3781
    },
    {
      "epoch": 0.28623054727780145,
      "grad_norm": 1.546875,
      "learning_rate": 1.9028210245198416e-05,
      "loss": 0.4838,
      "step": 3782
    },
    {
      "epoch": 0.28630622960124874,
      "grad_norm": 0.82421875,
      "learning_rate": 1.902769799708878e-05,
      "loss": 0.3655,
      "step": 3783
    },
    {
      "epoch": 0.2863819119246961,
      "grad_norm": 0.78125,
      "learning_rate": 1.9027185620905757e-05,
      "loss": 0.3074,
      "step": 3784
    },
    {
      "epoch": 0.28645759424814343,
      "grad_norm": 0.8515625,
      "learning_rate": 1.9026673116656615e-05,
      "loss": 0.3526,
      "step": 3785
    },
    {
      "epoch": 0.2865332765715907,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9026160484348626e-05,
      "loss": 0.3953,
      "step": 3786
    },
    {
      "epoch": 0.28660895889503807,
      "grad_norm": 0.8671875,
      "learning_rate": 1.9025647723989057e-05,
      "loss": 0.3692,
      "step": 3787
    },
    {
      "epoch": 0.2866846412184854,
      "grad_norm": 0.76171875,
      "learning_rate": 1.9025134835585187e-05,
      "loss": 0.2989,
      "step": 3788
    },
    {
      "epoch": 0.28676032354193276,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9024621819144294e-05,
      "loss": 0.3727,
      "step": 3789
    },
    {
      "epoch": 0.28683600586538005,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9024108674673653e-05,
      "loss": 0.3451,
      "step": 3790
    },
    {
      "epoch": 0.2869116881888274,
      "grad_norm": 0.828125,
      "learning_rate": 1.9023595402180545e-05,
      "loss": 0.3462,
      "step": 3791
    },
    {
      "epoch": 0.28698737051227474,
      "grad_norm": 0.703125,
      "learning_rate": 1.902308200167225e-05,
      "loss": 0.2529,
      "step": 3792
    },
    {
      "epoch": 0.28706305283572203,
      "grad_norm": 0.83984375,
      "learning_rate": 1.9022568473156052e-05,
      "loss": 0.3541,
      "step": 3793
    },
    {
      "epoch": 0.2871387351591694,
      "grad_norm": 0.8671875,
      "learning_rate": 1.902205481663924e-05,
      "loss": 0.3227,
      "step": 3794
    },
    {
      "epoch": 0.2872144174826167,
      "grad_norm": 0.7578125,
      "learning_rate": 1.9021541032129098e-05,
      "loss": 0.301,
      "step": 3795
    },
    {
      "epoch": 0.28729009980606407,
      "grad_norm": 1.6015625,
      "learning_rate": 1.9021027119632914e-05,
      "loss": 0.4558,
      "step": 3796
    },
    {
      "epoch": 0.28736578212951136,
      "grad_norm": 0.828125,
      "learning_rate": 1.9020513079157978e-05,
      "loss": 0.3212,
      "step": 3797
    },
    {
      "epoch": 0.2874414644529587,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9019998910711582e-05,
      "loss": 0.3674,
      "step": 3798
    },
    {
      "epoch": 0.28751714677640605,
      "grad_norm": 0.83203125,
      "learning_rate": 1.9019484614301025e-05,
      "loss": 0.3452,
      "step": 3799
    },
    {
      "epoch": 0.28759282909985334,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9018970189933603e-05,
      "loss": 0.3149,
      "step": 3800
    },
    {
      "epoch": 0.2876685114233007,
      "grad_norm": 0.90234375,
      "learning_rate": 1.9018455637616612e-05,
      "loss": 0.3621,
      "step": 3801
    },
    {
      "epoch": 0.28774419374674803,
      "grad_norm": 0.796875,
      "learning_rate": 1.901794095735735e-05,
      "loss": 0.333,
      "step": 3802
    },
    {
      "epoch": 0.2878198760701954,
      "grad_norm": 0.828125,
      "learning_rate": 1.901742614916312e-05,
      "loss": 0.3253,
      "step": 3803
    },
    {
      "epoch": 0.28789555839364267,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9016911213041225e-05,
      "loss": 0.3613,
      "step": 3804
    },
    {
      "epoch": 0.28797124071709,
      "grad_norm": 0.8359375,
      "learning_rate": 1.9016396148998973e-05,
      "loss": 0.3205,
      "step": 3805
    },
    {
      "epoch": 0.28804692304053736,
      "grad_norm": 0.7890625,
      "learning_rate": 1.9015880957043666e-05,
      "loss": 0.3149,
      "step": 3806
    },
    {
      "epoch": 0.28812260536398465,
      "grad_norm": 0.92578125,
      "learning_rate": 1.901536563718262e-05,
      "loss": 0.3456,
      "step": 3807
    },
    {
      "epoch": 0.288198287687432,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9014850189423136e-05,
      "loss": 0.375,
      "step": 3808
    },
    {
      "epoch": 0.28827397001087934,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9014334613772537e-05,
      "loss": 0.2958,
      "step": 3809
    },
    {
      "epoch": 0.2883496523343267,
      "grad_norm": 0.76171875,
      "learning_rate": 1.901381891023813e-05,
      "loss": 0.2713,
      "step": 3810
    },
    {
      "epoch": 0.288425334657774,
      "grad_norm": 0.734375,
      "learning_rate": 1.9013303078827235e-05,
      "loss": 0.2814,
      "step": 3811
    },
    {
      "epoch": 0.2885010169812213,
      "grad_norm": 0.86328125,
      "learning_rate": 1.9012787119547167e-05,
      "loss": 0.3473,
      "step": 3812
    },
    {
      "epoch": 0.28857669930466867,
      "grad_norm": 0.7421875,
      "learning_rate": 1.9012271032405246e-05,
      "loss": 0.2853,
      "step": 3813
    },
    {
      "epoch": 0.28865238162811596,
      "grad_norm": 0.828125,
      "learning_rate": 1.9011754817408797e-05,
      "loss": 0.3331,
      "step": 3814
    },
    {
      "epoch": 0.2887280639515633,
      "grad_norm": 0.796875,
      "learning_rate": 1.9011238474565143e-05,
      "loss": 0.2986,
      "step": 3815
    },
    {
      "epoch": 0.28880374627501065,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9010722003881604e-05,
      "loss": 0.3545,
      "step": 3816
    },
    {
      "epoch": 0.288879428598458,
      "grad_norm": 0.77734375,
      "learning_rate": 1.9010205405365513e-05,
      "loss": 0.3241,
      "step": 3817
    },
    {
      "epoch": 0.2889551109219053,
      "grad_norm": 0.984375,
      "learning_rate": 1.900968867902419e-05,
      "loss": 0.3654,
      "step": 3818
    },
    {
      "epoch": 0.28903079324535264,
      "grad_norm": 0.8046875,
      "learning_rate": 1.900917182486498e-05,
      "loss": 0.3097,
      "step": 3819
    },
    {
      "epoch": 0.2891064755688,
      "grad_norm": 0.8203125,
      "learning_rate": 1.9008654842895206e-05,
      "loss": 0.3317,
      "step": 3820
    },
    {
      "epoch": 0.28918215789224727,
      "grad_norm": 0.7578125,
      "learning_rate": 1.90081377331222e-05,
      "loss": 0.2953,
      "step": 3821
    },
    {
      "epoch": 0.2892578402156946,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9007620495553304e-05,
      "loss": 0.3184,
      "step": 3822
    },
    {
      "epoch": 0.28933352253914196,
      "grad_norm": 0.7578125,
      "learning_rate": 1.9007103130195853e-05,
      "loss": 0.2878,
      "step": 3823
    },
    {
      "epoch": 0.2894092048625893,
      "grad_norm": 0.765625,
      "learning_rate": 1.9006585637057187e-05,
      "loss": 0.3156,
      "step": 3824
    },
    {
      "epoch": 0.2894848871860366,
      "grad_norm": 0.85546875,
      "learning_rate": 1.900606801614465e-05,
      "loss": 0.3435,
      "step": 3825
    },
    {
      "epoch": 0.28956056950948394,
      "grad_norm": 0.8125,
      "learning_rate": 1.900555026746558e-05,
      "loss": 0.3357,
      "step": 3826
    },
    {
      "epoch": 0.2896362518329313,
      "grad_norm": 1.4375,
      "learning_rate": 1.9005032391027328e-05,
      "loss": 0.4226,
      "step": 3827
    },
    {
      "epoch": 0.2897119341563786,
      "grad_norm": 0.80859375,
      "learning_rate": 1.9004514386837236e-05,
      "loss": 0.3375,
      "step": 3828
    },
    {
      "epoch": 0.2897876164798259,
      "grad_norm": 0.82421875,
      "learning_rate": 1.9003996254902658e-05,
      "loss": 0.359,
      "step": 3829
    },
    {
      "epoch": 0.2898632988032733,
      "grad_norm": 0.80078125,
      "learning_rate": 1.9003477995230942e-05,
      "loss": 0.3321,
      "step": 3830
    },
    {
      "epoch": 0.28993898112672056,
      "grad_norm": 0.74609375,
      "learning_rate": 1.9002959607829436e-05,
      "loss": 0.2809,
      "step": 3831
    },
    {
      "epoch": 0.2900146634501679,
      "grad_norm": 0.859375,
      "learning_rate": 1.90024410927055e-05,
      "loss": 0.3693,
      "step": 3832
    },
    {
      "epoch": 0.29009034577361525,
      "grad_norm": 0.828125,
      "learning_rate": 1.900192244986649e-05,
      "loss": 0.3555,
      "step": 3833
    },
    {
      "epoch": 0.2901660280970626,
      "grad_norm": 0.875,
      "learning_rate": 1.9001403679319762e-05,
      "loss": 0.36,
      "step": 3834
    },
    {
      "epoch": 0.2902417104205099,
      "grad_norm": 0.8046875,
      "learning_rate": 1.9000884781072675e-05,
      "loss": 0.3533,
      "step": 3835
    },
    {
      "epoch": 0.29031739274395724,
      "grad_norm": 0.77734375,
      "learning_rate": 1.9000365755132593e-05,
      "loss": 0.3129,
      "step": 3836
    },
    {
      "epoch": 0.2903930750674046,
      "grad_norm": 0.76171875,
      "learning_rate": 1.8999846601506874e-05,
      "loss": 0.2837,
      "step": 3837
    },
    {
      "epoch": 0.2904687573908519,
      "grad_norm": 0.83984375,
      "learning_rate": 1.8999327320202893e-05,
      "loss": 0.3282,
      "step": 3838
    },
    {
      "epoch": 0.2905444397142992,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8998807911228003e-05,
      "loss": 0.3444,
      "step": 3839
    },
    {
      "epoch": 0.29062012203774656,
      "grad_norm": 0.859375,
      "learning_rate": 1.8998288374589585e-05,
      "loss": 0.3464,
      "step": 3840
    },
    {
      "epoch": 0.2906958043611939,
      "grad_norm": 1.421875,
      "learning_rate": 1.8997768710295004e-05,
      "loss": 0.4511,
      "step": 3841
    },
    {
      "epoch": 0.2907714866846412,
      "grad_norm": 0.79296875,
      "learning_rate": 1.899724891835163e-05,
      "loss": 0.319,
      "step": 3842
    },
    {
      "epoch": 0.29084716900808855,
      "grad_norm": 0.859375,
      "learning_rate": 1.8996728998766845e-05,
      "loss": 0.3889,
      "step": 3843
    },
    {
      "epoch": 0.2909228513315359,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8996208951548022e-05,
      "loss": 0.3158,
      "step": 3844
    },
    {
      "epoch": 0.2909985336549832,
      "grad_norm": 0.8125,
      "learning_rate": 1.899568877670253e-05,
      "loss": 0.3389,
      "step": 3845
    },
    {
      "epoch": 0.29107421597843053,
      "grad_norm": 0.90234375,
      "learning_rate": 1.899516847423776e-05,
      "loss": 0.3751,
      "step": 3846
    },
    {
      "epoch": 0.2911498983018779,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8994648044161087e-05,
      "loss": 0.3429,
      "step": 3847
    },
    {
      "epoch": 0.2912255806253252,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8994127486479897e-05,
      "loss": 0.2964,
      "step": 3848
    },
    {
      "epoch": 0.2913012629487725,
      "grad_norm": 0.83203125,
      "learning_rate": 1.8993606801201574e-05,
      "loss": 0.3018,
      "step": 3849
    },
    {
      "epoch": 0.29137694527221986,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8993085988333505e-05,
      "loss": 0.3415,
      "step": 3850
    },
    {
      "epoch": 0.2914526275956672,
      "grad_norm": 0.78125,
      "learning_rate": 1.899256504788308e-05,
      "loss": 0.3169,
      "step": 3851
    },
    {
      "epoch": 0.2915283099191145,
      "grad_norm": 0.76953125,
      "learning_rate": 1.8992043979857686e-05,
      "loss": 0.2794,
      "step": 3852
    },
    {
      "epoch": 0.29160399224256184,
      "grad_norm": 0.8203125,
      "learning_rate": 1.899152278426472e-05,
      "loss": 0.3622,
      "step": 3853
    },
    {
      "epoch": 0.2916796745660092,
      "grad_norm": 0.83984375,
      "learning_rate": 1.8991001461111568e-05,
      "loss": 0.367,
      "step": 3854
    },
    {
      "epoch": 0.29175535688945653,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8990480010405636e-05,
      "loss": 0.3163,
      "step": 3855
    },
    {
      "epoch": 0.2918310392129038,
      "grad_norm": 0.828125,
      "learning_rate": 1.8989958432154317e-05,
      "loss": 0.3653,
      "step": 3856
    },
    {
      "epoch": 0.29190672153635117,
      "grad_norm": 1.0234375,
      "learning_rate": 1.8989436726365006e-05,
      "loss": 0.3423,
      "step": 3857
    },
    {
      "epoch": 0.2919824038597985,
      "grad_norm": 0.796875,
      "learning_rate": 1.898891489304511e-05,
      "loss": 0.3128,
      "step": 3858
    },
    {
      "epoch": 0.2920580861832458,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8988392932202036e-05,
      "loss": 0.3362,
      "step": 3859
    },
    {
      "epoch": 0.29213376850669315,
      "grad_norm": 0.75,
      "learning_rate": 1.898787084384318e-05,
      "loss": 0.2924,
      "step": 3860
    },
    {
      "epoch": 0.2922094508301405,
      "grad_norm": 0.859375,
      "learning_rate": 1.898734862797595e-05,
      "loss": 0.3688,
      "step": 3861
    },
    {
      "epoch": 0.29228513315358784,
      "grad_norm": 0.875,
      "learning_rate": 1.898682628460776e-05,
      "loss": 0.3697,
      "step": 3862
    },
    {
      "epoch": 0.29236081547703513,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8986303813746016e-05,
      "loss": 0.3288,
      "step": 3863
    },
    {
      "epoch": 0.2924364978004825,
      "grad_norm": 0.83203125,
      "learning_rate": 1.8985781215398132e-05,
      "loss": 0.3543,
      "step": 3864
    },
    {
      "epoch": 0.2925121801239298,
      "grad_norm": 0.7578125,
      "learning_rate": 1.898525848957152e-05,
      "loss": 0.3066,
      "step": 3865
    },
    {
      "epoch": 0.2925878624473771,
      "grad_norm": 0.8671875,
      "learning_rate": 1.89847356362736e-05,
      "loss": 0.3757,
      "step": 3866
    },
    {
      "epoch": 0.29266354477082446,
      "grad_norm": 0.89453125,
      "learning_rate": 1.8984212655511784e-05,
      "loss": 0.3726,
      "step": 3867
    },
    {
      "epoch": 0.2927392270942718,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8983689547293496e-05,
      "loss": 0.3078,
      "step": 3868
    },
    {
      "epoch": 0.29281490941771915,
      "grad_norm": 0.83203125,
      "learning_rate": 1.898316631162615e-05,
      "loss": 0.3376,
      "step": 3869
    },
    {
      "epoch": 0.29289059174116644,
      "grad_norm": 0.90234375,
      "learning_rate": 1.898264294851718e-05,
      "loss": 0.3798,
      "step": 3870
    },
    {
      "epoch": 0.2929662740646138,
      "grad_norm": 0.87109375,
      "learning_rate": 1.8982119457974005e-05,
      "loss": 0.3507,
      "step": 3871
    },
    {
      "epoch": 0.29304195638806113,
      "grad_norm": 1.65625,
      "learning_rate": 1.898159584000405e-05,
      "loss": 0.4661,
      "step": 3872
    },
    {
      "epoch": 0.2931176387115084,
      "grad_norm": 0.875,
      "learning_rate": 1.8981072094614743e-05,
      "loss": 0.3378,
      "step": 3873
    },
    {
      "epoch": 0.29319332103495577,
      "grad_norm": 5.25,
      "learning_rate": 1.8980548221813518e-05,
      "loss": 0.482,
      "step": 3874
    },
    {
      "epoch": 0.2932690033584031,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8980024221607807e-05,
      "loss": 0.3407,
      "step": 3875
    },
    {
      "epoch": 0.29334468568185046,
      "grad_norm": 0.84375,
      "learning_rate": 1.8979500094005038e-05,
      "loss": 0.3347,
      "step": 3876
    },
    {
      "epoch": 0.29342036800529775,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8978975839012657e-05,
      "loss": 0.3455,
      "step": 3877
    },
    {
      "epoch": 0.2934960503287451,
      "grad_norm": 0.88671875,
      "learning_rate": 1.8978451456638088e-05,
      "loss": 0.3677,
      "step": 3878
    },
    {
      "epoch": 0.29357173265219244,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8977926946888785e-05,
      "loss": 0.3323,
      "step": 3879
    },
    {
      "epoch": 0.29364741497563973,
      "grad_norm": 0.88671875,
      "learning_rate": 1.8977402309772178e-05,
      "loss": 0.372,
      "step": 3880
    },
    {
      "epoch": 0.2937230972990871,
      "grad_norm": 0.890625,
      "learning_rate": 1.8976877545295713e-05,
      "loss": 0.3552,
      "step": 3881
    },
    {
      "epoch": 0.2937987796225344,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8976352653466836e-05,
      "loss": 0.3012,
      "step": 3882
    },
    {
      "epoch": 0.29387446194598177,
      "grad_norm": 0.84375,
      "learning_rate": 1.897582763429299e-05,
      "loss": 0.308,
      "step": 3883
    },
    {
      "epoch": 0.29395014426942906,
      "grad_norm": 0.7890625,
      "learning_rate": 1.897530248778163e-05,
      "loss": 0.2992,
      "step": 3884
    },
    {
      "epoch": 0.2940258265928764,
      "grad_norm": 0.859375,
      "learning_rate": 1.8974777213940197e-05,
      "loss": 0.3271,
      "step": 3885
    },
    {
      "epoch": 0.29410150891632375,
      "grad_norm": 0.84375,
      "learning_rate": 1.8974251812776155e-05,
      "loss": 0.312,
      "step": 3886
    },
    {
      "epoch": 0.29417719123977104,
      "grad_norm": 0.8828125,
      "learning_rate": 1.8973726284296945e-05,
      "loss": 0.3641,
      "step": 3887
    },
    {
      "epoch": 0.2942528735632184,
      "grad_norm": 0.94140625,
      "learning_rate": 1.8973200628510027e-05,
      "loss": 0.3437,
      "step": 3888
    },
    {
      "epoch": 0.29432855588666573,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8972674845422867e-05,
      "loss": 0.3247,
      "step": 3889
    },
    {
      "epoch": 0.294404238210113,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8972148935042912e-05,
      "loss": 0.3497,
      "step": 3890
    },
    {
      "epoch": 0.29447992053356037,
      "grad_norm": 3.5625,
      "learning_rate": 1.8971622897377628e-05,
      "loss": 0.4467,
      "step": 3891
    },
    {
      "epoch": 0.2945556028570077,
      "grad_norm": 0.87109375,
      "learning_rate": 1.8971096732434476e-05,
      "loss": 0.3806,
      "step": 3892
    },
    {
      "epoch": 0.29463128518045506,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8970570440220924e-05,
      "loss": 0.3135,
      "step": 3893
    },
    {
      "epoch": 0.29470696750390235,
      "grad_norm": 0.86328125,
      "learning_rate": 1.8970044020744437e-05,
      "loss": 0.3614,
      "step": 3894
    },
    {
      "epoch": 0.2947826498273497,
      "grad_norm": 0.9140625,
      "learning_rate": 1.896951747401248e-05,
      "loss": 0.3419,
      "step": 3895
    },
    {
      "epoch": 0.29485833215079704,
      "grad_norm": 0.83984375,
      "learning_rate": 1.8968990800032526e-05,
      "loss": 0.3189,
      "step": 3896
    },
    {
      "epoch": 0.29493401447424433,
      "grad_norm": 1.203125,
      "learning_rate": 1.8968463998812048e-05,
      "loss": 0.3439,
      "step": 3897
    },
    {
      "epoch": 0.2950096967976917,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8967937070358514e-05,
      "loss": 0.3043,
      "step": 3898
    },
    {
      "epoch": 0.295085379121139,
      "grad_norm": 0.83203125,
      "learning_rate": 1.8967410014679408e-05,
      "loss": 0.3421,
      "step": 3899
    },
    {
      "epoch": 0.29516106144458637,
      "grad_norm": 0.78125,
      "learning_rate": 1.8966882831782197e-05,
      "loss": 0.3112,
      "step": 3900
    },
    {
      "epoch": 0.29523674376803366,
      "grad_norm": 0.82421875,
      "learning_rate": 1.896635552167437e-05,
      "loss": 0.3391,
      "step": 3901
    },
    {
      "epoch": 0.295312426091481,
      "grad_norm": 0.89453125,
      "learning_rate": 1.89658280843634e-05,
      "loss": 0.3474,
      "step": 3902
    },
    {
      "epoch": 0.29538810841492835,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8965300519856774e-05,
      "loss": 0.3405,
      "step": 3903
    },
    {
      "epoch": 0.29546379073837564,
      "grad_norm": 0.828125,
      "learning_rate": 1.8964772828161977e-05,
      "loss": 0.3511,
      "step": 3904
    },
    {
      "epoch": 0.295539473061823,
      "grad_norm": 0.796875,
      "learning_rate": 1.896424500928649e-05,
      "loss": 0.3128,
      "step": 3905
    },
    {
      "epoch": 0.29561515538527033,
      "grad_norm": 0.83203125,
      "learning_rate": 1.8963717063237805e-05,
      "loss": 0.3401,
      "step": 3906
    },
    {
      "epoch": 0.2956908377087177,
      "grad_norm": 0.875,
      "learning_rate": 1.8963188990023414e-05,
      "loss": 0.3108,
      "step": 3907
    },
    {
      "epoch": 0.29576652003216497,
      "grad_norm": 1.7265625,
      "learning_rate": 1.8962660789650803e-05,
      "loss": 0.3899,
      "step": 3908
    },
    {
      "epoch": 0.2958422023556123,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8962132462127468e-05,
      "loss": 0.339,
      "step": 3909
    },
    {
      "epoch": 0.29591788467905966,
      "grad_norm": 0.7578125,
      "learning_rate": 1.896160400746091e-05,
      "loss": 0.2808,
      "step": 3910
    },
    {
      "epoch": 0.29599356700250695,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8961075425658612e-05,
      "loss": 0.3194,
      "step": 3911
    },
    {
      "epoch": 0.2960692493259543,
      "grad_norm": 0.85546875,
      "learning_rate": 1.8960546716728087e-05,
      "loss": 0.3645,
      "step": 3912
    },
    {
      "epoch": 0.29614493164940164,
      "grad_norm": 0.75,
      "learning_rate": 1.896001788067683e-05,
      "loss": 0.304,
      "step": 3913
    },
    {
      "epoch": 0.296220613972849,
      "grad_norm": 0.8125,
      "learning_rate": 1.895948891751234e-05,
      "loss": 0.3393,
      "step": 3914
    },
    {
      "epoch": 0.2962962962962963,
      "grad_norm": 0.875,
      "learning_rate": 1.8958959827242128e-05,
      "loss": 0.4112,
      "step": 3915
    },
    {
      "epoch": 0.2963719786197436,
      "grad_norm": 0.8125,
      "learning_rate": 1.8958430609873693e-05,
      "loss": 0.3345,
      "step": 3916
    },
    {
      "epoch": 0.29644766094319097,
      "grad_norm": 0.75390625,
      "learning_rate": 1.895790126541455e-05,
      "loss": 0.2701,
      "step": 3917
    },
    {
      "epoch": 0.29652334326663826,
      "grad_norm": 0.76171875,
      "learning_rate": 1.8957371793872204e-05,
      "loss": 0.2916,
      "step": 3918
    },
    {
      "epoch": 0.2965990255900856,
      "grad_norm": 0.7734375,
      "learning_rate": 1.895684219525417e-05,
      "loss": 0.3117,
      "step": 3919
    },
    {
      "epoch": 0.29667470791353295,
      "grad_norm": 0.76171875,
      "learning_rate": 1.8956312469567954e-05,
      "loss": 0.3064,
      "step": 3920
    },
    {
      "epoch": 0.2967503902369803,
      "grad_norm": 0.8359375,
      "learning_rate": 1.895578261682108e-05,
      "loss": 0.3477,
      "step": 3921
    },
    {
      "epoch": 0.2968260725604276,
      "grad_norm": 0.77734375,
      "learning_rate": 1.895525263702106e-05,
      "loss": 0.3289,
      "step": 3922
    },
    {
      "epoch": 0.29690175488387494,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8954722530175415e-05,
      "loss": 0.3188,
      "step": 3923
    },
    {
      "epoch": 0.2969774372073223,
      "grad_norm": 1.8046875,
      "learning_rate": 1.895419229629166e-05,
      "loss": 0.4155,
      "step": 3924
    },
    {
      "epoch": 0.29705311953076957,
      "grad_norm": 0.890625,
      "learning_rate": 1.8953661935377326e-05,
      "loss": 0.3308,
      "step": 3925
    },
    {
      "epoch": 0.2971288018542169,
      "grad_norm": 0.82421875,
      "learning_rate": 1.895313144743993e-05,
      "loss": 0.3288,
      "step": 3926
    },
    {
      "epoch": 0.29720448417766426,
      "grad_norm": 1.2890625,
      "learning_rate": 1.8952600832487e-05,
      "loss": 0.3987,
      "step": 3927
    },
    {
      "epoch": 0.2972801665011116,
      "grad_norm": 0.8828125,
      "learning_rate": 1.8952070090526066e-05,
      "loss": 0.3222,
      "step": 3928
    },
    {
      "epoch": 0.2973558488245589,
      "grad_norm": 0.859375,
      "learning_rate": 1.8951539221564652e-05,
      "loss": 0.3758,
      "step": 3929
    },
    {
      "epoch": 0.29743153114800625,
      "grad_norm": 0.8828125,
      "learning_rate": 1.8951008225610296e-05,
      "loss": 0.3917,
      "step": 3930
    },
    {
      "epoch": 0.2975072134714536,
      "grad_norm": 0.9453125,
      "learning_rate": 1.895047710267053e-05,
      "loss": 0.3801,
      "step": 3931
    },
    {
      "epoch": 0.2975828957949009,
      "grad_norm": 0.796875,
      "learning_rate": 1.894994585275288e-05,
      "loss": 0.3057,
      "step": 3932
    },
    {
      "epoch": 0.2976585781183482,
      "grad_norm": 0.796875,
      "learning_rate": 1.8949414475864892e-05,
      "loss": 0.2711,
      "step": 3933
    },
    {
      "epoch": 0.2977342604417956,
      "grad_norm": 0.859375,
      "learning_rate": 1.8948882972014104e-05,
      "loss": 0.3606,
      "step": 3934
    },
    {
      "epoch": 0.2978099427652429,
      "grad_norm": 0.86328125,
      "learning_rate": 1.8948351341208052e-05,
      "loss": 0.3894,
      "step": 3935
    },
    {
      "epoch": 0.2978856250886902,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8947819583454282e-05,
      "loss": 0.3105,
      "step": 3936
    },
    {
      "epoch": 0.29796130741213755,
      "grad_norm": 0.796875,
      "learning_rate": 1.8947287698760336e-05,
      "loss": 0.3093,
      "step": 3937
    },
    {
      "epoch": 0.2980369897355849,
      "grad_norm": 0.83203125,
      "learning_rate": 1.894675568713376e-05,
      "loss": 0.3428,
      "step": 3938
    },
    {
      "epoch": 0.2981126720590322,
      "grad_norm": 0.828125,
      "learning_rate": 1.8946223548582102e-05,
      "loss": 0.3457,
      "step": 3939
    },
    {
      "epoch": 0.29818835438247954,
      "grad_norm": 0.9375,
      "learning_rate": 1.8945691283112908e-05,
      "loss": 0.3852,
      "step": 3940
    },
    {
      "epoch": 0.2982640367059269,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8945158890733733e-05,
      "loss": 0.2854,
      "step": 3941
    },
    {
      "epoch": 0.29833971902937423,
      "grad_norm": 2.109375,
      "learning_rate": 1.894462637145213e-05,
      "loss": 0.4656,
      "step": 3942
    },
    {
      "epoch": 0.2984154013528215,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8944093725275655e-05,
      "loss": 0.3449,
      "step": 3943
    },
    {
      "epoch": 0.29849108367626886,
      "grad_norm": 0.83203125,
      "learning_rate": 1.8943560952211857e-05,
      "loss": 0.326,
      "step": 3944
    },
    {
      "epoch": 0.2985667659997162,
      "grad_norm": 0.8125,
      "learning_rate": 1.89430280522683e-05,
      "loss": 0.3351,
      "step": 3945
    },
    {
      "epoch": 0.2986424483231635,
      "grad_norm": 0.8515625,
      "learning_rate": 1.8942495025452544e-05,
      "loss": 0.3681,
      "step": 3946
    },
    {
      "epoch": 0.29871813064661085,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8941961871772153e-05,
      "loss": 0.3319,
      "step": 3947
    },
    {
      "epoch": 0.2987938129700582,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8941428591234684e-05,
      "loss": 0.3126,
      "step": 3948
    },
    {
      "epoch": 0.29886949529350554,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8940895183847708e-05,
      "loss": 0.3447,
      "step": 3949
    },
    {
      "epoch": 0.29894517761695283,
      "grad_norm": 0.87109375,
      "learning_rate": 1.894036164961879e-05,
      "loss": 0.3521,
      "step": 3950
    },
    {
      "epoch": 0.2990208599404002,
      "grad_norm": 0.87109375,
      "learning_rate": 1.8939827988555502e-05,
      "loss": 0.3735,
      "step": 3951
    },
    {
      "epoch": 0.2990965422638475,
      "grad_norm": 0.83984375,
      "learning_rate": 1.893929420066541e-05,
      "loss": 0.3443,
      "step": 3952
    },
    {
      "epoch": 0.2991722245872948,
      "grad_norm": 0.859375,
      "learning_rate": 1.893876028595609e-05,
      "loss": 0.3736,
      "step": 3953
    },
    {
      "epoch": 0.29924790691074216,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8938226244435117e-05,
      "loss": 0.3441,
      "step": 3954
    },
    {
      "epoch": 0.2993235892341895,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8937692076110066e-05,
      "loss": 0.3536,
      "step": 3955
    },
    {
      "epoch": 0.2993992715576368,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8937157780988512e-05,
      "loss": 0.3488,
      "step": 3956
    },
    {
      "epoch": 0.29947495388108414,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8936623359078043e-05,
      "loss": 0.3194,
      "step": 3957
    },
    {
      "epoch": 0.2995506362045315,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8936088810386233e-05,
      "loss": 0.3457,
      "step": 3958
    },
    {
      "epoch": 0.29962631852797883,
      "grad_norm": 0.8828125,
      "learning_rate": 1.893555413492067e-05,
      "loss": 0.3797,
      "step": 3959
    },
    {
      "epoch": 0.2997020008514261,
      "grad_norm": 0.765625,
      "learning_rate": 1.8935019332688937e-05,
      "loss": 0.3103,
      "step": 3960
    },
    {
      "epoch": 0.29977768317487347,
      "grad_norm": 0.8125,
      "learning_rate": 1.893448440369862e-05,
      "loss": 0.335,
      "step": 3961
    },
    {
      "epoch": 0.2998533654983208,
      "grad_norm": 0.8046875,
      "learning_rate": 1.893394934795731e-05,
      "loss": 0.3591,
      "step": 3962
    },
    {
      "epoch": 0.2999290478217681,
      "grad_norm": 0.84765625,
      "learning_rate": 1.8933414165472598e-05,
      "loss": 0.3119,
      "step": 3963
    },
    {
      "epoch": 0.30000473014521545,
      "grad_norm": 0.8125,
      "learning_rate": 1.893287885625208e-05,
      "loss": 0.3603,
      "step": 3964
    },
    {
      "epoch": 0.3000804124686628,
      "grad_norm": 1.546875,
      "learning_rate": 1.893234342030334e-05,
      "loss": 0.4595,
      "step": 3965
    },
    {
      "epoch": 0.30015609479211014,
      "grad_norm": 0.859375,
      "learning_rate": 1.8931807857633984e-05,
      "loss": 0.352,
      "step": 3966
    },
    {
      "epoch": 0.30015609479211014,
      "eval_loss": 0.3503101170063019,
      "eval_runtime": 83.3888,
      "eval_samples_per_second": 58.305,
      "eval_steps_per_second": 58.305,
      "step": 3966
    },
    {
      "epoch": 0.30023177711555743,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8931272168251605e-05,
      "loss": 0.3716,
      "step": 3967
    },
    {
      "epoch": 0.3003074594390048,
      "grad_norm": 0.73046875,
      "learning_rate": 1.8930736352163804e-05,
      "loss": 0.279,
      "step": 3968
    },
    {
      "epoch": 0.3003831417624521,
      "grad_norm": 1.203125,
      "learning_rate": 1.893020040937818e-05,
      "loss": 0.3315,
      "step": 3969
    },
    {
      "epoch": 0.3004588240858994,
      "grad_norm": 0.74609375,
      "learning_rate": 1.8929664339902342e-05,
      "loss": 0.2718,
      "step": 3970
    },
    {
      "epoch": 0.30053450640934676,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8929128143743888e-05,
      "loss": 0.3162,
      "step": 3971
    },
    {
      "epoch": 0.3006101887327941,
      "grad_norm": 0.8046875,
      "learning_rate": 1.892859182091043e-05,
      "loss": 0.3402,
      "step": 3972
    },
    {
      "epoch": 0.30068587105624145,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8928055371409574e-05,
      "loss": 0.3395,
      "step": 3973
    },
    {
      "epoch": 0.30076155337968874,
      "grad_norm": 0.80078125,
      "learning_rate": 1.892751879524893e-05,
      "loss": 0.3372,
      "step": 3974
    },
    {
      "epoch": 0.3008372357031361,
      "grad_norm": 0.859375,
      "learning_rate": 1.8926982092436117e-05,
      "loss": 0.3343,
      "step": 3975
    },
    {
      "epoch": 0.30091291802658343,
      "grad_norm": 0.76953125,
      "learning_rate": 1.892644526297874e-05,
      "loss": 0.3057,
      "step": 3976
    },
    {
      "epoch": 0.3009886003500307,
      "grad_norm": 1.4765625,
      "learning_rate": 1.892590830688442e-05,
      "loss": 0.4119,
      "step": 3977
    },
    {
      "epoch": 0.30106428267347807,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8925371224160772e-05,
      "loss": 0.3231,
      "step": 3978
    },
    {
      "epoch": 0.3011399649969254,
      "grad_norm": 0.8125,
      "learning_rate": 1.8924834014815418e-05,
      "loss": 0.3355,
      "step": 3979
    },
    {
      "epoch": 0.30121564732037276,
      "grad_norm": 0.87109375,
      "learning_rate": 1.892429667885598e-05,
      "loss": 0.3751,
      "step": 3980
    },
    {
      "epoch": 0.30129132964382005,
      "grad_norm": 0.828125,
      "learning_rate": 1.8923759216290074e-05,
      "loss": 0.3326,
      "step": 3981
    },
    {
      "epoch": 0.3013670119672674,
      "grad_norm": 1.1796875,
      "learning_rate": 1.8923221627125335e-05,
      "loss": 0.3752,
      "step": 3982
    },
    {
      "epoch": 0.30144269429071474,
      "grad_norm": 0.765625,
      "learning_rate": 1.892268391136938e-05,
      "loss": 0.3193,
      "step": 3983
    },
    {
      "epoch": 0.30151837661416203,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8922146069029843e-05,
      "loss": 0.3263,
      "step": 3984
    },
    {
      "epoch": 0.3015940589376094,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8921608100114355e-05,
      "loss": 0.3586,
      "step": 3985
    },
    {
      "epoch": 0.3016697412610567,
      "grad_norm": 0.84765625,
      "learning_rate": 1.8921070004630545e-05,
      "loss": 0.3396,
      "step": 3986
    },
    {
      "epoch": 0.30174542358450407,
      "grad_norm": 0.84765625,
      "learning_rate": 1.8920531782586047e-05,
      "loss": 0.3364,
      "step": 3987
    },
    {
      "epoch": 0.30182110590795136,
      "grad_norm": 0.83984375,
      "learning_rate": 1.8919993433988497e-05,
      "loss": 0.3498,
      "step": 3988
    },
    {
      "epoch": 0.3018967882313987,
      "grad_norm": 0.8671875,
      "learning_rate": 1.8919454958845533e-05,
      "loss": 0.3536,
      "step": 3989
    },
    {
      "epoch": 0.30197247055484605,
      "grad_norm": 1.3046875,
      "learning_rate": 1.8918916357164795e-05,
      "loss": 0.3473,
      "step": 3990
    },
    {
      "epoch": 0.30204815287829334,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8918377628953926e-05,
      "loss": 0.3301,
      "step": 3991
    },
    {
      "epoch": 0.3021238352017407,
      "grad_norm": 0.8828125,
      "learning_rate": 1.8917838774220563e-05,
      "loss": 0.3736,
      "step": 3992
    },
    {
      "epoch": 0.30219951752518803,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8917299792972353e-05,
      "loss": 0.2892,
      "step": 3993
    },
    {
      "epoch": 0.3022751998486354,
      "grad_norm": 0.80859375,
      "learning_rate": 1.891676068521694e-05,
      "loss": 0.3484,
      "step": 3994
    },
    {
      "epoch": 0.30235088217208267,
      "grad_norm": 0.75,
      "learning_rate": 1.8916221450961982e-05,
      "loss": 0.2921,
      "step": 3995
    },
    {
      "epoch": 0.30242656449553,
      "grad_norm": 1.6953125,
      "learning_rate": 1.8915682090215118e-05,
      "loss": 0.2767,
      "step": 3996
    },
    {
      "epoch": 0.30250224681897736,
      "grad_norm": 0.8671875,
      "learning_rate": 1.8915142602984e-05,
      "loss": 0.3153,
      "step": 3997
    },
    {
      "epoch": 0.30257792914242465,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8914602989276294e-05,
      "loss": 0.3272,
      "step": 3998
    },
    {
      "epoch": 0.302653611465872,
      "grad_norm": 0.73046875,
      "learning_rate": 1.891406324909964e-05,
      "loss": 0.2781,
      "step": 3999
    },
    {
      "epoch": 0.30272929378931934,
      "grad_norm": 0.9140625,
      "learning_rate": 1.8913523382461706e-05,
      "loss": 0.3302,
      "step": 4000
    },
    {
      "epoch": 0.3028049761127667,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8912983389370143e-05,
      "loss": 0.2961,
      "step": 4001
    },
    {
      "epoch": 0.302880658436214,
      "grad_norm": 0.75,
      "learning_rate": 1.8912443269832618e-05,
      "loss": 0.2739,
      "step": 4002
    },
    {
      "epoch": 0.3029563407596613,
      "grad_norm": 0.84765625,
      "learning_rate": 1.8911903023856787e-05,
      "loss": 0.3339,
      "step": 4003
    },
    {
      "epoch": 0.30303202308310867,
      "grad_norm": 0.76171875,
      "learning_rate": 1.8911362651450322e-05,
      "loss": 0.2885,
      "step": 4004
    },
    {
      "epoch": 0.30310770540655596,
      "grad_norm": 0.75390625,
      "learning_rate": 1.8910822152620885e-05,
      "loss": 0.2829,
      "step": 4005
    },
    {
      "epoch": 0.3031833877300033,
      "grad_norm": 0.8671875,
      "learning_rate": 1.8910281527376146e-05,
      "loss": 0.3739,
      "step": 4006
    },
    {
      "epoch": 0.30325907005345065,
      "grad_norm": 0.8125,
      "learning_rate": 1.890974077572377e-05,
      "loss": 0.3079,
      "step": 4007
    },
    {
      "epoch": 0.303334752376898,
      "grad_norm": 0.78125,
      "learning_rate": 1.8909199897671432e-05,
      "loss": 0.2969,
      "step": 4008
    },
    {
      "epoch": 0.3034104347003453,
      "grad_norm": 12.75,
      "learning_rate": 1.8908658893226803e-05,
      "loss": 0.3578,
      "step": 4009
    },
    {
      "epoch": 0.30348611702379263,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8908117762397564e-05,
      "loss": 0.3435,
      "step": 4010
    },
    {
      "epoch": 0.30356179934724,
      "grad_norm": 0.94921875,
      "learning_rate": 1.8907576505191384e-05,
      "loss": 0.3608,
      "step": 4011
    },
    {
      "epoch": 0.30363748167068727,
      "grad_norm": 0.8125,
      "learning_rate": 1.8907035121615946e-05,
      "loss": 0.3297,
      "step": 4012
    },
    {
      "epoch": 0.3037131639941346,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8906493611678927e-05,
      "loss": 0.3158,
      "step": 4013
    },
    {
      "epoch": 0.30378884631758196,
      "grad_norm": 0.8125,
      "learning_rate": 1.8905951975388018e-05,
      "loss": 0.3156,
      "step": 4014
    },
    {
      "epoch": 0.30386452864102925,
      "grad_norm": 0.87109375,
      "learning_rate": 1.8905410212750896e-05,
      "loss": 0.3932,
      "step": 4015
    },
    {
      "epoch": 0.3039402109644766,
      "grad_norm": 1.2890625,
      "learning_rate": 1.8904868323775246e-05,
      "loss": 0.3782,
      "step": 4016
    },
    {
      "epoch": 0.30401589328792394,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8904326308468758e-05,
      "loss": 0.3099,
      "step": 4017
    },
    {
      "epoch": 0.3040915756113713,
      "grad_norm": 0.80078125,
      "learning_rate": 1.890378416683912e-05,
      "loss": 0.3005,
      "step": 4018
    },
    {
      "epoch": 0.3041672579348186,
      "grad_norm": 0.90234375,
      "learning_rate": 1.8903241898894028e-05,
      "loss": 0.3806,
      "step": 4019
    },
    {
      "epoch": 0.3042429402582659,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8902699504641168e-05,
      "loss": 0.3498,
      "step": 4020
    },
    {
      "epoch": 0.30431862258171327,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8902156984088238e-05,
      "loss": 0.3508,
      "step": 4021
    },
    {
      "epoch": 0.30439430490516056,
      "grad_norm": 0.78125,
      "learning_rate": 1.8901614337242936e-05,
      "loss": 0.2962,
      "step": 4022
    },
    {
      "epoch": 0.3044699872286079,
      "grad_norm": 0.8515625,
      "learning_rate": 1.8901071564112957e-05,
      "loss": 0.3563,
      "step": 4023
    },
    {
      "epoch": 0.30454566955205525,
      "grad_norm": 0.85546875,
      "learning_rate": 1.8900528664706e-05,
      "loss": 0.3727,
      "step": 4024
    },
    {
      "epoch": 0.3046213518755026,
      "grad_norm": 0.7421875,
      "learning_rate": 1.8899985639029778e-05,
      "loss": 0.3,
      "step": 4025
    },
    {
      "epoch": 0.3046970341989499,
      "grad_norm": 0.7890625,
      "learning_rate": 1.889944248709198e-05,
      "loss": 0.328,
      "step": 4026
    },
    {
      "epoch": 0.30477271652239724,
      "grad_norm": 0.8828125,
      "learning_rate": 1.8898899208900324e-05,
      "loss": 0.3683,
      "step": 4027
    },
    {
      "epoch": 0.3048483988458446,
      "grad_norm": 0.88671875,
      "learning_rate": 1.8898355804462507e-05,
      "loss": 0.4048,
      "step": 4028
    },
    {
      "epoch": 0.30492408116929187,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8897812273786246e-05,
      "loss": 0.3273,
      "step": 4029
    },
    {
      "epoch": 0.3049997634927392,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8897268616879247e-05,
      "loss": 0.3257,
      "step": 4030
    },
    {
      "epoch": 0.30507544581618656,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8896724833749225e-05,
      "loss": 0.3126,
      "step": 4031
    },
    {
      "epoch": 0.3051511281396339,
      "grad_norm": 0.76953125,
      "learning_rate": 1.8896180924403894e-05,
      "loss": 0.3209,
      "step": 4032
    },
    {
      "epoch": 0.3052268104630812,
      "grad_norm": 0.87109375,
      "learning_rate": 1.889563688885097e-05,
      "loss": 0.3687,
      "step": 4033
    },
    {
      "epoch": 0.30530249278652855,
      "grad_norm": 0.80078125,
      "learning_rate": 1.889509272709817e-05,
      "loss": 0.2909,
      "step": 4034
    },
    {
      "epoch": 0.3053781751099759,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8894548439153218e-05,
      "loss": 0.3546,
      "step": 4035
    },
    {
      "epoch": 0.3054538574334232,
      "grad_norm": 1.140625,
      "learning_rate": 1.889400402502383e-05,
      "loss": 0.3132,
      "step": 4036
    },
    {
      "epoch": 0.3055295397568705,
      "grad_norm": 0.953125,
      "learning_rate": 1.8893459484717734e-05,
      "loss": 0.3803,
      "step": 4037
    },
    {
      "epoch": 0.3056052220803179,
      "grad_norm": 0.8125,
      "learning_rate": 1.889291481824265e-05,
      "loss": 0.3485,
      "step": 4038
    },
    {
      "epoch": 0.3056809044037652,
      "grad_norm": 0.84765625,
      "learning_rate": 1.8892370025606314e-05,
      "loss": 0.3755,
      "step": 4039
    },
    {
      "epoch": 0.3057565867272125,
      "grad_norm": 1.0390625,
      "learning_rate": 1.8891825106816444e-05,
      "loss": 0.3062,
      "step": 4040
    },
    {
      "epoch": 0.30583226905065986,
      "grad_norm": 0.83203125,
      "learning_rate": 1.889128006188078e-05,
      "loss": 0.3274,
      "step": 4041
    },
    {
      "epoch": 0.3059079513741072,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8890734890807053e-05,
      "loss": 0.3279,
      "step": 4042
    },
    {
      "epoch": 0.3059836336975545,
      "grad_norm": 0.80859375,
      "learning_rate": 1.889018959360299e-05,
      "loss": 0.3575,
      "step": 4043
    },
    {
      "epoch": 0.30605931602100184,
      "grad_norm": 0.796875,
      "learning_rate": 1.888964417027633e-05,
      "loss": 0.3244,
      "step": 4044
    },
    {
      "epoch": 0.3061349983444492,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8889098620834814e-05,
      "loss": 0.3316,
      "step": 4045
    },
    {
      "epoch": 0.30621068066789653,
      "grad_norm": 0.80859375,
      "learning_rate": 1.888855294528618e-05,
      "loss": 0.3278,
      "step": 4046
    },
    {
      "epoch": 0.3062863629913438,
      "grad_norm": 0.984375,
      "learning_rate": 1.8888007143638172e-05,
      "loss": 0.3334,
      "step": 4047
    },
    {
      "epoch": 0.30636204531479116,
      "grad_norm": 0.83203125,
      "learning_rate": 1.888746121589853e-05,
      "loss": 0.3028,
      "step": 4048
    },
    {
      "epoch": 0.3064377276382385,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8886915162074996e-05,
      "loss": 0.3293,
      "step": 4049
    },
    {
      "epoch": 0.3065134099616858,
      "grad_norm": 1.71875,
      "learning_rate": 1.888636898217532e-05,
      "loss": 0.4204,
      "step": 4050
    },
    {
      "epoch": 0.30658909228513315,
      "grad_norm": 0.8515625,
      "learning_rate": 1.8885822676207254e-05,
      "loss": 0.3201,
      "step": 4051
    },
    {
      "epoch": 0.3066647746085805,
      "grad_norm": 0.921875,
      "learning_rate": 1.888527624417854e-05,
      "loss": 0.3789,
      "step": 4052
    },
    {
      "epoch": 0.30674045693202784,
      "grad_norm": 0.86328125,
      "learning_rate": 1.8884729686096936e-05,
      "loss": 0.3459,
      "step": 4053
    },
    {
      "epoch": 0.30681613925547513,
      "grad_norm": 0.83984375,
      "learning_rate": 1.8884183001970197e-05,
      "loss": 0.349,
      "step": 4054
    },
    {
      "epoch": 0.3068918215789225,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8883636191806073e-05,
      "loss": 0.3415,
      "step": 4055
    },
    {
      "epoch": 0.3069675039023698,
      "grad_norm": 0.81640625,
      "learning_rate": 1.888308925561233e-05,
      "loss": 0.3315,
      "step": 4056
    },
    {
      "epoch": 0.3070431862258171,
      "grad_norm": 0.87890625,
      "learning_rate": 1.8882542193396716e-05,
      "loss": 0.3599,
      "step": 4057
    },
    {
      "epoch": 0.30711886854926446,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8881995005167005e-05,
      "loss": 0.3387,
      "step": 4058
    },
    {
      "epoch": 0.3071945508727118,
      "grad_norm": 0.90234375,
      "learning_rate": 1.888144769093095e-05,
      "loss": 0.3891,
      "step": 4059
    },
    {
      "epoch": 0.30727023319615915,
      "grad_norm": 0.875,
      "learning_rate": 1.8880900250696315e-05,
      "loss": 0.3216,
      "step": 4060
    },
    {
      "epoch": 0.30734591551960644,
      "grad_norm": 0.84765625,
      "learning_rate": 1.8880352684470874e-05,
      "loss": 0.3178,
      "step": 4061
    },
    {
      "epoch": 0.3074215978430538,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8879804992262387e-05,
      "loss": 0.2919,
      "step": 4062
    },
    {
      "epoch": 0.30749728016650113,
      "grad_norm": 0.8046875,
      "learning_rate": 1.887925717407863e-05,
      "loss": 0.3504,
      "step": 4063
    },
    {
      "epoch": 0.3075729624899484,
      "grad_norm": 0.73828125,
      "learning_rate": 1.8878709229927372e-05,
      "loss": 0.2839,
      "step": 4064
    },
    {
      "epoch": 0.30764864481339577,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8878161159816386e-05,
      "loss": 0.3249,
      "step": 4065
    },
    {
      "epoch": 0.3077243271368431,
      "grad_norm": 0.859375,
      "learning_rate": 1.887761296375345e-05,
      "loss": 0.3532,
      "step": 4066
    },
    {
      "epoch": 0.30780000946029046,
      "grad_norm": 0.79296875,
      "learning_rate": 1.887706464174634e-05,
      "loss": 0.3295,
      "step": 4067
    },
    {
      "epoch": 0.30787569178373775,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8876516193802838e-05,
      "loss": 0.3646,
      "step": 4068
    },
    {
      "epoch": 0.3079513741071851,
      "grad_norm": 0.76953125,
      "learning_rate": 1.8875967619930715e-05,
      "loss": 0.302,
      "step": 4069
    },
    {
      "epoch": 0.30802705643063244,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8875418920137764e-05,
      "loss": 0.3273,
      "step": 4070
    },
    {
      "epoch": 0.30810273875407973,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8874870094431762e-05,
      "loss": 0.3174,
      "step": 4071
    },
    {
      "epoch": 0.3081784210775271,
      "grad_norm": 0.84375,
      "learning_rate": 1.8874321142820496e-05,
      "loss": 0.3375,
      "step": 4072
    },
    {
      "epoch": 0.3082541034009744,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8873772065311762e-05,
      "loss": 0.2792,
      "step": 4073
    },
    {
      "epoch": 0.3083297857244217,
      "grad_norm": 0.8125,
      "learning_rate": 1.887322286191334e-05,
      "loss": 0.335,
      "step": 4074
    },
    {
      "epoch": 0.30840546804786906,
      "grad_norm": 0.73828125,
      "learning_rate": 1.8872673532633022e-05,
      "loss": 0.2883,
      "step": 4075
    },
    {
      "epoch": 0.3084811503713164,
      "grad_norm": 0.8671875,
      "learning_rate": 1.887212407747861e-05,
      "loss": 0.3874,
      "step": 4076
    },
    {
      "epoch": 0.30855683269476375,
      "grad_norm": 0.86328125,
      "learning_rate": 1.887157449645789e-05,
      "loss": 0.3916,
      "step": 4077
    },
    {
      "epoch": 0.30863251501821104,
      "grad_norm": 0.78515625,
      "learning_rate": 1.887102478957866e-05,
      "loss": 0.3258,
      "step": 4078
    },
    {
      "epoch": 0.3087081973416584,
      "grad_norm": 0.7734375,
      "learning_rate": 1.887047495684872e-05,
      "loss": 0.3141,
      "step": 4079
    },
    {
      "epoch": 0.30878387966510573,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8869924998275873e-05,
      "loss": 0.312,
      "step": 4080
    },
    {
      "epoch": 0.308859561988553,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8869374913867914e-05,
      "loss": 0.3178,
      "step": 4081
    },
    {
      "epoch": 0.30893524431200037,
      "grad_norm": 0.87109375,
      "learning_rate": 1.8868824703632658e-05,
      "loss": 0.3436,
      "step": 4082
    },
    {
      "epoch": 0.3090109266354477,
      "grad_norm": 0.8203125,
      "learning_rate": 1.88682743675779e-05,
      "loss": 0.3439,
      "step": 4083
    },
    {
      "epoch": 0.30908660895889506,
      "grad_norm": 0.8203125,
      "learning_rate": 1.886772390571145e-05,
      "loss": 0.317,
      "step": 4084
    },
    {
      "epoch": 0.30916229128234235,
      "grad_norm": 0.76953125,
      "learning_rate": 1.8867173318041123e-05,
      "loss": 0.3096,
      "step": 4085
    },
    {
      "epoch": 0.3092379736057897,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8866622604574723e-05,
      "loss": 0.2912,
      "step": 4086
    },
    {
      "epoch": 0.30931365592923704,
      "grad_norm": 0.7734375,
      "learning_rate": 1.886607176532007e-05,
      "loss": 0.3249,
      "step": 4087
    },
    {
      "epoch": 0.30938933825268433,
      "grad_norm": 0.77734375,
      "learning_rate": 1.886552080028497e-05,
      "loss": 0.2988,
      "step": 4088
    },
    {
      "epoch": 0.3094650205761317,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8864969709477248e-05,
      "loss": 0.3284,
      "step": 4089
    },
    {
      "epoch": 0.309540702899579,
      "grad_norm": 0.84765625,
      "learning_rate": 1.8864418492904712e-05,
      "loss": 0.3808,
      "step": 4090
    },
    {
      "epoch": 0.30961638522302637,
      "grad_norm": 0.7578125,
      "learning_rate": 1.886386715057519e-05,
      "loss": 0.307,
      "step": 4091
    },
    {
      "epoch": 0.30969206754647366,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8863315682496503e-05,
      "loss": 0.3351,
      "step": 4092
    },
    {
      "epoch": 0.309767749869921,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8862764088676472e-05,
      "loss": 0.3428,
      "step": 4093
    },
    {
      "epoch": 0.30984343219336835,
      "grad_norm": 0.859375,
      "learning_rate": 1.8862212369122925e-05,
      "loss": 0.336,
      "step": 4094
    },
    {
      "epoch": 0.30991911451681564,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8861660523843684e-05,
      "loss": 0.3301,
      "step": 4095
    },
    {
      "epoch": 0.309994796840263,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8861108552846587e-05,
      "loss": 0.3227,
      "step": 4096
    },
    {
      "epoch": 0.31007047916371033,
      "grad_norm": 0.83984375,
      "learning_rate": 1.8860556456139453e-05,
      "loss": 0.3054,
      "step": 4097
    },
    {
      "epoch": 0.3101461614871577,
      "grad_norm": 0.8515625,
      "learning_rate": 1.8860004233730126e-05,
      "loss": 0.3796,
      "step": 4098
    },
    {
      "epoch": 0.31022184381060497,
      "grad_norm": 0.74609375,
      "learning_rate": 1.885945188562643e-05,
      "loss": 0.2525,
      "step": 4099
    },
    {
      "epoch": 0.3102975261340523,
      "grad_norm": 0.77734375,
      "learning_rate": 1.885889941183621e-05,
      "loss": 0.2972,
      "step": 4100
    },
    {
      "epoch": 0.31037320845749966,
      "grad_norm": 11.25,
      "learning_rate": 1.8858346812367295e-05,
      "loss": 0.4335,
      "step": 4101
    },
    {
      "epoch": 0.31044889078094695,
      "grad_norm": 0.75,
      "learning_rate": 1.885779408722753e-05,
      "loss": 0.2992,
      "step": 4102
    },
    {
      "epoch": 0.3105245731043943,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8857241236424756e-05,
      "loss": 0.3506,
      "step": 4103
    },
    {
      "epoch": 0.31060025542784164,
      "grad_norm": 1.6953125,
      "learning_rate": 1.8856688259966815e-05,
      "loss": 0.4637,
      "step": 4104
    },
    {
      "epoch": 0.310675937751289,
      "grad_norm": 0.75,
      "learning_rate": 1.8856135157861554e-05,
      "loss": 0.2974,
      "step": 4105
    },
    {
      "epoch": 0.3107516200747363,
      "grad_norm": 0.90234375,
      "learning_rate": 1.8855581930116814e-05,
      "loss": 0.3962,
      "step": 4106
    },
    {
      "epoch": 0.3108273023981836,
      "grad_norm": 0.84375,
      "learning_rate": 1.8855028576740453e-05,
      "loss": 0.3568,
      "step": 4107
    },
    {
      "epoch": 0.31090298472163097,
      "grad_norm": 0.890625,
      "learning_rate": 1.8854475097740313e-05,
      "loss": 0.3384,
      "step": 4108
    },
    {
      "epoch": 0.31097866704507826,
      "grad_norm": 0.8125,
      "learning_rate": 1.885392149312425e-05,
      "loss": 0.3361,
      "step": 4109
    },
    {
      "epoch": 0.3110543493685256,
      "grad_norm": 0.796875,
      "learning_rate": 1.8853367762900117e-05,
      "loss": 0.3362,
      "step": 4110
    },
    {
      "epoch": 0.31113003169197295,
      "grad_norm": 0.84765625,
      "learning_rate": 1.8852813907075768e-05,
      "loss": 0.3621,
      "step": 4111
    },
    {
      "epoch": 0.3112057140154203,
      "grad_norm": 0.8671875,
      "learning_rate": 1.8852259925659062e-05,
      "loss": 0.3909,
      "step": 4112
    },
    {
      "epoch": 0.3112813963388676,
      "grad_norm": 0.7890625,
      "learning_rate": 1.885170581865786e-05,
      "loss": 0.3238,
      "step": 4113
    },
    {
      "epoch": 0.31135707866231493,
      "grad_norm": 0.85546875,
      "learning_rate": 1.885115158608002e-05,
      "loss": 0.3834,
      "step": 4114
    },
    {
      "epoch": 0.3114327609857623,
      "grad_norm": 0.796875,
      "learning_rate": 1.8850597227933406e-05,
      "loss": 0.3231,
      "step": 4115
    },
    {
      "epoch": 0.31150844330920957,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8850042744225877e-05,
      "loss": 0.3128,
      "step": 4116
    },
    {
      "epoch": 0.3115841256326569,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8849488134965312e-05,
      "loss": 0.3249,
      "step": 4117
    },
    {
      "epoch": 0.31165980795610426,
      "grad_norm": 0.81640625,
      "learning_rate": 1.884893340015957e-05,
      "loss": 0.3251,
      "step": 4118
    },
    {
      "epoch": 0.3117354902795516,
      "grad_norm": 0.87109375,
      "learning_rate": 1.884837853981652e-05,
      "loss": 0.3472,
      "step": 4119
    },
    {
      "epoch": 0.3118111726029989,
      "grad_norm": 3.9375,
      "learning_rate": 1.8847823553944032e-05,
      "loss": 0.4659,
      "step": 4120
    },
    {
      "epoch": 0.31188685492644624,
      "grad_norm": 0.84765625,
      "learning_rate": 1.884726844254999e-05,
      "loss": 0.3426,
      "step": 4121
    },
    {
      "epoch": 0.3119625372498936,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8846713205642262e-05,
      "loss": 0.3022,
      "step": 4122
    },
    {
      "epoch": 0.3120382195733409,
      "grad_norm": 0.8671875,
      "learning_rate": 1.8846157843228724e-05,
      "loss": 0.3718,
      "step": 4123
    },
    {
      "epoch": 0.3121139018967882,
      "grad_norm": 0.84765625,
      "learning_rate": 1.8845602355317257e-05,
      "loss": 0.3092,
      "step": 4124
    },
    {
      "epoch": 0.31218958422023557,
      "grad_norm": 0.87109375,
      "learning_rate": 1.884504674191574e-05,
      "loss": 0.4013,
      "step": 4125
    },
    {
      "epoch": 0.3122652665436829,
      "grad_norm": 1.375,
      "learning_rate": 1.884449100303206e-05,
      "loss": 0.3971,
      "step": 4126
    },
    {
      "epoch": 0.3123409488671302,
      "grad_norm": 0.828125,
      "learning_rate": 1.8843935138674093e-05,
      "loss": 0.3542,
      "step": 4127
    },
    {
      "epoch": 0.31241663119057755,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8843379148849732e-05,
      "loss": 0.31,
      "step": 4128
    },
    {
      "epoch": 0.3124923135140249,
      "grad_norm": 0.83984375,
      "learning_rate": 1.884282303356686e-05,
      "loss": 0.3792,
      "step": 4129
    },
    {
      "epoch": 0.3125679958374722,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8842266792833374e-05,
      "loss": 0.29,
      "step": 4130
    },
    {
      "epoch": 0.31264367816091954,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8841710426657153e-05,
      "loss": 0.3188,
      "step": 4131
    },
    {
      "epoch": 0.3127193604843669,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8841153935046098e-05,
      "loss": 0.3348,
      "step": 4132
    },
    {
      "epoch": 0.3127950428078142,
      "grad_norm": 0.859375,
      "learning_rate": 1.8840597318008104e-05,
      "loss": 0.3305,
      "step": 4133
    },
    {
      "epoch": 0.3128707251312615,
      "grad_norm": 0.83203125,
      "learning_rate": 1.8840040575551066e-05,
      "loss": 0.3655,
      "step": 4134
    },
    {
      "epoch": 0.31294640745470886,
      "grad_norm": 0.85546875,
      "learning_rate": 1.8839483707682883e-05,
      "loss": 0.2813,
      "step": 4135
    },
    {
      "epoch": 0.3130220897781562,
      "grad_norm": 0.85546875,
      "learning_rate": 1.8838926714411455e-05,
      "loss": 0.3639,
      "step": 4136
    },
    {
      "epoch": 0.3130977721016035,
      "grad_norm": 0.89453125,
      "learning_rate": 1.883836959574468e-05,
      "loss": 0.3377,
      "step": 4137
    },
    {
      "epoch": 0.31317345442505085,
      "grad_norm": 0.77734375,
      "learning_rate": 1.883781235169047e-05,
      "loss": 0.3135,
      "step": 4138
    },
    {
      "epoch": 0.3132491367484982,
      "grad_norm": 0.78125,
      "learning_rate": 1.883725498225672e-05,
      "loss": 0.3166,
      "step": 4139
    },
    {
      "epoch": 0.3133248190719455,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8836697487451348e-05,
      "loss": 0.3013,
      "step": 4140
    },
    {
      "epoch": 0.3134005013953928,
      "grad_norm": 2.0,
      "learning_rate": 1.8836139867282253e-05,
      "loss": 0.3797,
      "step": 4141
    },
    {
      "epoch": 0.3134761837188402,
      "grad_norm": 2.484375,
      "learning_rate": 1.8835582121757355e-05,
      "loss": 0.3877,
      "step": 4142
    },
    {
      "epoch": 0.3135518660422875,
      "grad_norm": 1.21875,
      "learning_rate": 1.883502425088456e-05,
      "loss": 0.3912,
      "step": 4143
    },
    {
      "epoch": 0.3136275483657348,
      "grad_norm": 0.875,
      "learning_rate": 1.8834466254671783e-05,
      "loss": 0.3353,
      "step": 4144
    },
    {
      "epoch": 0.31370323068918216,
      "grad_norm": 0.8515625,
      "learning_rate": 1.8833908133126946e-05,
      "loss": 0.3519,
      "step": 4145
    },
    {
      "epoch": 0.3137789130126295,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8833349886257958e-05,
      "loss": 0.3072,
      "step": 4146
    },
    {
      "epoch": 0.3138545953360768,
      "grad_norm": 0.8125,
      "learning_rate": 1.8832791514072748e-05,
      "loss": 0.3371,
      "step": 4147
    },
    {
      "epoch": 0.31393027765952414,
      "grad_norm": 0.859375,
      "learning_rate": 1.883223301657923e-05,
      "loss": 0.3487,
      "step": 4148
    },
    {
      "epoch": 0.3140059599829715,
      "grad_norm": 0.88671875,
      "learning_rate": 1.883167439378533e-05,
      "loss": 0.3195,
      "step": 4149
    },
    {
      "epoch": 0.31408164230641883,
      "grad_norm": 0.83984375,
      "learning_rate": 1.8831115645698973e-05,
      "loss": 0.31,
      "step": 4150
    },
    {
      "epoch": 0.3141573246298661,
      "grad_norm": 2.140625,
      "learning_rate": 1.8830556772328086e-05,
      "loss": 0.3845,
      "step": 4151
    },
    {
      "epoch": 0.31423300695331347,
      "grad_norm": 0.87890625,
      "learning_rate": 1.8829997773680598e-05,
      "loss": 0.3153,
      "step": 4152
    },
    {
      "epoch": 0.3143086892767608,
      "grad_norm": 0.8125,
      "learning_rate": 1.882943864976444e-05,
      "loss": 0.3379,
      "step": 4153
    },
    {
      "epoch": 0.3143843716002081,
      "grad_norm": 0.92578125,
      "learning_rate": 1.882887940058754e-05,
      "loss": 0.377,
      "step": 4154
    },
    {
      "epoch": 0.31446005392365545,
      "grad_norm": 0.890625,
      "learning_rate": 1.8828320026157836e-05,
      "loss": 0.3662,
      "step": 4155
    },
    {
      "epoch": 0.3145357362471028,
      "grad_norm": 0.96875,
      "learning_rate": 1.8827760526483263e-05,
      "loss": 0.3782,
      "step": 4156
    },
    {
      "epoch": 0.31461141857055014,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8827200901571757e-05,
      "loss": 0.3109,
      "step": 4157
    },
    {
      "epoch": 0.31468710089399743,
      "grad_norm": 0.7578125,
      "learning_rate": 1.882664115143126e-05,
      "loss": 0.3031,
      "step": 4158
    },
    {
      "epoch": 0.3147627832174448,
      "grad_norm": 0.8984375,
      "learning_rate": 1.882608127606971e-05,
      "loss": 0.3487,
      "step": 4159
    },
    {
      "epoch": 0.3148384655408921,
      "grad_norm": 0.8125,
      "learning_rate": 1.8825521275495052e-05,
      "loss": 0.3345,
      "step": 4160
    },
    {
      "epoch": 0.3149141478643394,
      "grad_norm": 0.8359375,
      "learning_rate": 1.882496114971523e-05,
      "loss": 0.3317,
      "step": 4161
    },
    {
      "epoch": 0.31498983018778676,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8824400898738185e-05,
      "loss": 0.3355,
      "step": 4162
    },
    {
      "epoch": 0.3150655125112341,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8823840522571875e-05,
      "loss": 0.307,
      "step": 4163
    },
    {
      "epoch": 0.31514119483468145,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8823280021224245e-05,
      "loss": 0.3413,
      "step": 4164
    },
    {
      "epoch": 0.31521687715812874,
      "grad_norm": 2.640625,
      "learning_rate": 1.8822719394703242e-05,
      "loss": 0.4683,
      "step": 4165
    },
    {
      "epoch": 0.3152925594815761,
      "grad_norm": 0.8203125,
      "learning_rate": 1.882215864301683e-05,
      "loss": 0.3238,
      "step": 4166
    },
    {
      "epoch": 0.31536824180502343,
      "grad_norm": 0.76171875,
      "learning_rate": 1.8821597766172957e-05,
      "loss": 0.2995,
      "step": 4167
    },
    {
      "epoch": 0.3154439241284707,
      "grad_norm": 1.0625,
      "learning_rate": 1.882103676417958e-05,
      "loss": 0.3594,
      "step": 4168
    },
    {
      "epoch": 0.31551960645191807,
      "grad_norm": 0.78125,
      "learning_rate": 1.8820475637044656e-05,
      "loss": 0.321,
      "step": 4169
    },
    {
      "epoch": 0.3155952887753654,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8819914384776155e-05,
      "loss": 0.2969,
      "step": 4170
    },
    {
      "epoch": 0.31567097109881276,
      "grad_norm": 0.828125,
      "learning_rate": 1.8819353007382027e-05,
      "loss": 0.3239,
      "step": 4171
    },
    {
      "epoch": 0.31574665342226005,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8818791504870247e-05,
      "loss": 0.3041,
      "step": 4172
    },
    {
      "epoch": 0.3158223357457074,
      "grad_norm": 0.85546875,
      "learning_rate": 1.8818229877248772e-05,
      "loss": 0.3529,
      "step": 4173
    },
    {
      "epoch": 0.31589801806915474,
      "grad_norm": 1.1640625,
      "learning_rate": 1.8817668124525577e-05,
      "loss": 0.3524,
      "step": 4174
    },
    {
      "epoch": 0.31597370039260203,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8817106246708625e-05,
      "loss": 0.2878,
      "step": 4175
    },
    {
      "epoch": 0.3160493827160494,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8816544243805893e-05,
      "loss": 0.3314,
      "step": 4176
    },
    {
      "epoch": 0.3161250650394967,
      "grad_norm": 0.8828125,
      "learning_rate": 1.881598211582535e-05,
      "loss": 0.3593,
      "step": 4177
    },
    {
      "epoch": 0.31620074736294407,
      "grad_norm": 0.80078125,
      "learning_rate": 1.881541986277497e-05,
      "loss": 0.3182,
      "step": 4178
    },
    {
      "epoch": 0.31627642968639136,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8814857484662733e-05,
      "loss": 0.2886,
      "step": 4179
    },
    {
      "epoch": 0.3163521120098387,
      "grad_norm": 0.8671875,
      "learning_rate": 1.8814294981496616e-05,
      "loss": 0.3825,
      "step": 4180
    },
    {
      "epoch": 0.31642779433328605,
      "grad_norm": 0.83984375,
      "learning_rate": 1.8813732353284598e-05,
      "loss": 0.3528,
      "step": 4181
    },
    {
      "epoch": 0.31650347665673334,
      "grad_norm": 0.84765625,
      "learning_rate": 1.8813169600034664e-05,
      "loss": 0.3452,
      "step": 4182
    },
    {
      "epoch": 0.3165791589801807,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8812606721754792e-05,
      "loss": 0.3268,
      "step": 4183
    },
    {
      "epoch": 0.31665484130362803,
      "grad_norm": 0.96484375,
      "learning_rate": 1.8812043718452973e-05,
      "loss": 0.4419,
      "step": 4184
    },
    {
      "epoch": 0.3167305236270754,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8811480590137192e-05,
      "loss": 0.3193,
      "step": 4185
    },
    {
      "epoch": 0.31680620595052267,
      "grad_norm": 0.859375,
      "learning_rate": 1.8810917336815438e-05,
      "loss": 0.3798,
      "step": 4186
    },
    {
      "epoch": 0.31688188827397,
      "grad_norm": 0.74609375,
      "learning_rate": 1.88103539584957e-05,
      "loss": 0.2795,
      "step": 4187
    },
    {
      "epoch": 0.31695757059741736,
      "grad_norm": 0.81640625,
      "learning_rate": 1.880979045518597e-05,
      "loss": 0.3386,
      "step": 4188
    },
    {
      "epoch": 0.31703325292086465,
      "grad_norm": 0.765625,
      "learning_rate": 1.880922682689425e-05,
      "loss": 0.2999,
      "step": 4189
    },
    {
      "epoch": 0.317108935244312,
      "grad_norm": 0.79296875,
      "learning_rate": 1.880866307362853e-05,
      "loss": 0.314,
      "step": 4190
    },
    {
      "epoch": 0.31718461756775934,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8808099195396804e-05,
      "loss": 0.3222,
      "step": 4191
    },
    {
      "epoch": 0.3172602998912067,
      "grad_norm": 1.3828125,
      "learning_rate": 1.8807535192207075e-05,
      "loss": 0.4347,
      "step": 4192
    },
    {
      "epoch": 0.317335982214654,
      "grad_norm": 0.83203125,
      "learning_rate": 1.8806971064067348e-05,
      "loss": 0.3596,
      "step": 4193
    },
    {
      "epoch": 0.3174116645381013,
      "grad_norm": 0.83984375,
      "learning_rate": 1.8806406810985624e-05,
      "loss": 0.2854,
      "step": 4194
    },
    {
      "epoch": 0.31748734686154867,
      "grad_norm": 0.85546875,
      "learning_rate": 1.8805842432969904e-05,
      "loss": 0.3364,
      "step": 4195
    },
    {
      "epoch": 0.31756302918499596,
      "grad_norm": 0.83203125,
      "learning_rate": 1.8805277930028198e-05,
      "loss": 0.347,
      "step": 4196
    },
    {
      "epoch": 0.3176387115084433,
      "grad_norm": 0.84765625,
      "learning_rate": 1.8804713302168513e-05,
      "loss": 0.3824,
      "step": 4197
    },
    {
      "epoch": 0.31771439383189065,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8804148549398864e-05,
      "loss": 0.3334,
      "step": 4198
    },
    {
      "epoch": 0.31779007615533794,
      "grad_norm": 0.84375,
      "learning_rate": 1.880358367172726e-05,
      "loss": 0.3216,
      "step": 4199
    },
    {
      "epoch": 0.3178657584787853,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8803018669161712e-05,
      "loss": 0.3707,
      "step": 4200
    },
    {
      "epoch": 0.31794144080223263,
      "grad_norm": 0.8671875,
      "learning_rate": 1.8802453541710238e-05,
      "loss": 0.4154,
      "step": 4201
    },
    {
      "epoch": 0.31801712312568,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8801888289380855e-05,
      "loss": 0.3449,
      "step": 4202
    },
    {
      "epoch": 0.31809280544912727,
      "grad_norm": 0.83203125,
      "learning_rate": 1.880132291218158e-05,
      "loss": 0.3333,
      "step": 4203
    },
    {
      "epoch": 0.3181684877725746,
      "grad_norm": 0.84765625,
      "learning_rate": 1.8800757410120438e-05,
      "loss": 0.3211,
      "step": 4204
    },
    {
      "epoch": 0.31824417009602196,
      "grad_norm": 0.875,
      "learning_rate": 1.880019178320545e-05,
      "loss": 0.3671,
      "step": 4205
    },
    {
      "epoch": 0.31831985241946925,
      "grad_norm": 0.83984375,
      "learning_rate": 1.8799626031444638e-05,
      "loss": 0.3528,
      "step": 4206
    },
    {
      "epoch": 0.3183955347429166,
      "grad_norm": 0.80078125,
      "learning_rate": 1.879906015484603e-05,
      "loss": 0.3133,
      "step": 4207
    },
    {
      "epoch": 0.31847121706636394,
      "grad_norm": 0.94921875,
      "learning_rate": 1.8798494153417657e-05,
      "loss": 0.2837,
      "step": 4208
    },
    {
      "epoch": 0.3185468993898113,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8797928027167543e-05,
      "loss": 0.3078,
      "step": 4209
    },
    {
      "epoch": 0.3186225817132586,
      "grad_norm": 0.828125,
      "learning_rate": 1.8797361776103727e-05,
      "loss": 0.3153,
      "step": 4210
    },
    {
      "epoch": 0.3186982640367059,
      "grad_norm": 0.87109375,
      "learning_rate": 1.879679540023423e-05,
      "loss": 0.3697,
      "step": 4211
    },
    {
      "epoch": 0.31877394636015327,
      "grad_norm": 0.828125,
      "learning_rate": 1.87962288995671e-05,
      "loss": 0.319,
      "step": 4212
    },
    {
      "epoch": 0.31884962868360056,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8795662274110365e-05,
      "loss": 0.344,
      "step": 4213
    },
    {
      "epoch": 0.3189253110070479,
      "grad_norm": 0.80859375,
      "learning_rate": 1.879509552387207e-05,
      "loss": 0.3286,
      "step": 4214
    },
    {
      "epoch": 0.31900099333049525,
      "grad_norm": 0.7421875,
      "learning_rate": 1.879452864886025e-05,
      "loss": 0.2607,
      "step": 4215
    },
    {
      "epoch": 0.3190766756539426,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8793961649082947e-05,
      "loss": 0.344,
      "step": 4216
    },
    {
      "epoch": 0.3191523579773899,
      "grad_norm": 0.8046875,
      "learning_rate": 1.879339452454821e-05,
      "loss": 0.3138,
      "step": 4217
    },
    {
      "epoch": 0.31922804030083723,
      "grad_norm": 0.86328125,
      "learning_rate": 1.879282727526408e-05,
      "loss": 0.3944,
      "step": 4218
    },
    {
      "epoch": 0.3193037226242846,
      "grad_norm": 1.4921875,
      "learning_rate": 1.8792259901238608e-05,
      "loss": 0.4251,
      "step": 4219
    },
    {
      "epoch": 0.31937940494773187,
      "grad_norm": 0.77734375,
      "learning_rate": 1.879169240247984e-05,
      "loss": 0.318,
      "step": 4220
    },
    {
      "epoch": 0.3194550872711792,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8791124778995828e-05,
      "loss": 0.3004,
      "step": 4221
    },
    {
      "epoch": 0.31953076959462656,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8790557030794627e-05,
      "loss": 0.3188,
      "step": 4222
    },
    {
      "epoch": 0.3196064519180739,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8789989157884286e-05,
      "loss": 0.3524,
      "step": 4223
    },
    {
      "epoch": 0.3196821342415212,
      "grad_norm": 0.85546875,
      "learning_rate": 1.8789421160272868e-05,
      "loss": 0.3784,
      "step": 4224
    },
    {
      "epoch": 0.31975781656496854,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8788853037968425e-05,
      "loss": 0.3344,
      "step": 4225
    },
    {
      "epoch": 0.3198334988884159,
      "grad_norm": 1.09375,
      "learning_rate": 1.8788284790979025e-05,
      "loss": 0.376,
      "step": 4226
    },
    {
      "epoch": 0.3199091812118632,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8787716419312717e-05,
      "loss": 0.3117,
      "step": 4227
    },
    {
      "epoch": 0.3199848635353105,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8787147922977573e-05,
      "loss": 0.3078,
      "step": 4228
    },
    {
      "epoch": 0.32006054585875787,
      "grad_norm": 0.79296875,
      "learning_rate": 1.878657930198166e-05,
      "loss": 0.3386,
      "step": 4229
    },
    {
      "epoch": 0.3201362281822052,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8786010556333036e-05,
      "loss": 0.3549,
      "step": 4230
    },
    {
      "epoch": 0.3202119105056525,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8785441686039777e-05,
      "loss": 0.3366,
      "step": 4231
    },
    {
      "epoch": 0.32028759282909985,
      "grad_norm": 0.83203125,
      "learning_rate": 1.8784872691109954e-05,
      "loss": 0.3522,
      "step": 4232
    },
    {
      "epoch": 0.3203632751525472,
      "grad_norm": 0.89453125,
      "learning_rate": 1.8784303571551636e-05,
      "loss": 0.4243,
      "step": 4233
    },
    {
      "epoch": 0.3204389574759945,
      "grad_norm": 0.76171875,
      "learning_rate": 1.8783734327372895e-05,
      "loss": 0.3072,
      "step": 4234
    },
    {
      "epoch": 0.32051463979944184,
      "grad_norm": 0.80078125,
      "learning_rate": 1.878316495858181e-05,
      "loss": 0.3321,
      "step": 4235
    },
    {
      "epoch": 0.3205903221228892,
      "grad_norm": 0.828125,
      "learning_rate": 1.878259546518646e-05,
      "loss": 0.3273,
      "step": 4236
    },
    {
      "epoch": 0.32066600444633653,
      "grad_norm": 0.86328125,
      "learning_rate": 1.878202584719492e-05,
      "loss": 0.3541,
      "step": 4237
    },
    {
      "epoch": 0.3207416867697838,
      "grad_norm": 0.88671875,
      "learning_rate": 1.8781456104615273e-05,
      "loss": 0.3701,
      "step": 4238
    },
    {
      "epoch": 0.32081736909323116,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8780886237455597e-05,
      "loss": 0.3454,
      "step": 4239
    },
    {
      "epoch": 0.3208930514166785,
      "grad_norm": 0.8984375,
      "learning_rate": 1.8780316245723988e-05,
      "loss": 0.4031,
      "step": 4240
    },
    {
      "epoch": 0.3209687337401258,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8779746129428523e-05,
      "loss": 0.3673,
      "step": 4241
    },
    {
      "epoch": 0.32104441606357315,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8779175888577294e-05,
      "loss": 0.3226,
      "step": 4242
    },
    {
      "epoch": 0.3211200983870205,
      "grad_norm": 0.84765625,
      "learning_rate": 1.8778605523178388e-05,
      "loss": 0.3072,
      "step": 4243
    },
    {
      "epoch": 0.32119578071046784,
      "grad_norm": 1.0,
      "learning_rate": 1.8778035033239895e-05,
      "loss": 0.3452,
      "step": 4244
    },
    {
      "epoch": 0.32127146303391513,
      "grad_norm": 0.98828125,
      "learning_rate": 1.8777464418769917e-05,
      "loss": 0.3611,
      "step": 4245
    },
    {
      "epoch": 0.3213471453573625,
      "grad_norm": 1.328125,
      "learning_rate": 1.877689367977654e-05,
      "loss": 0.407,
      "step": 4246
    },
    {
      "epoch": 0.3214228276808098,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8776322816267866e-05,
      "loss": 0.3275,
      "step": 4247
    },
    {
      "epoch": 0.3214985100042571,
      "grad_norm": 0.83984375,
      "learning_rate": 1.877575182825199e-05,
      "loss": 0.3166,
      "step": 4248
    },
    {
      "epoch": 0.32157419232770446,
      "grad_norm": 0.78125,
      "learning_rate": 1.8775180715737018e-05,
      "loss": 0.3157,
      "step": 4249
    },
    {
      "epoch": 0.3216498746511518,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8774609478731048e-05,
      "loss": 0.3243,
      "step": 4250
    },
    {
      "epoch": 0.32172555697459915,
      "grad_norm": 0.8125,
      "learning_rate": 1.8774038117242185e-05,
      "loss": 0.3036,
      "step": 4251
    },
    {
      "epoch": 0.32180123929804644,
      "grad_norm": 0.82421875,
      "learning_rate": 1.877346663127853e-05,
      "loss": 0.3068,
      "step": 4252
    },
    {
      "epoch": 0.3218769216214938,
      "grad_norm": 0.84375,
      "learning_rate": 1.87728950208482e-05,
      "loss": 0.3647,
      "step": 4253
    },
    {
      "epoch": 0.32195260394494113,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8772323285959298e-05,
      "loss": 0.3488,
      "step": 4254
    },
    {
      "epoch": 0.3220282862683884,
      "grad_norm": 0.859375,
      "learning_rate": 1.8771751426619936e-05,
      "loss": 0.3452,
      "step": 4255
    },
    {
      "epoch": 0.32210396859183577,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8771179442838227e-05,
      "loss": 0.3344,
      "step": 4256
    },
    {
      "epoch": 0.3221796509152831,
      "grad_norm": 0.73828125,
      "learning_rate": 1.8770607334622285e-05,
      "loss": 0.2938,
      "step": 4257
    },
    {
      "epoch": 0.3222553332387304,
      "grad_norm": 0.94140625,
      "learning_rate": 1.877003510198023e-05,
      "loss": 0.3574,
      "step": 4258
    },
    {
      "epoch": 0.32233101556217775,
      "grad_norm": 0.87890625,
      "learning_rate": 1.8769462744920174e-05,
      "loss": 0.343,
      "step": 4259
    },
    {
      "epoch": 0.3224066978856251,
      "grad_norm": 0.84375,
      "learning_rate": 1.8768890263450244e-05,
      "loss": 0.3439,
      "step": 4260
    },
    {
      "epoch": 0.32248238020907244,
      "grad_norm": 0.8515625,
      "learning_rate": 1.8768317657578553e-05,
      "loss": 0.3291,
      "step": 4261
    },
    {
      "epoch": 0.32255806253251973,
      "grad_norm": 0.83984375,
      "learning_rate": 1.8767744927313233e-05,
      "loss": 0.3195,
      "step": 4262
    },
    {
      "epoch": 0.3226337448559671,
      "grad_norm": 0.95703125,
      "learning_rate": 1.8767172072662402e-05,
      "loss": 0.4226,
      "step": 4263
    },
    {
      "epoch": 0.3227094271794144,
      "grad_norm": 0.87109375,
      "learning_rate": 1.876659909363419e-05,
      "loss": 0.3948,
      "step": 4264
    },
    {
      "epoch": 0.3227851095028617,
      "grad_norm": 0.87109375,
      "learning_rate": 1.8766025990236733e-05,
      "loss": 0.3432,
      "step": 4265
    },
    {
      "epoch": 0.32286079182630906,
      "grad_norm": 0.890625,
      "learning_rate": 1.8765452762478148e-05,
      "loss": 0.3691,
      "step": 4266
    },
    {
      "epoch": 0.3229364741497564,
      "grad_norm": 0.7890625,
      "learning_rate": 1.876487941036657e-05,
      "loss": 0.2959,
      "step": 4267
    },
    {
      "epoch": 0.32301215647320375,
      "grad_norm": 0.83203125,
      "learning_rate": 1.876430593391014e-05,
      "loss": 0.2932,
      "step": 4268
    },
    {
      "epoch": 0.32308783879665104,
      "grad_norm": 0.83203125,
      "learning_rate": 1.8763732333116996e-05,
      "loss": 0.3408,
      "step": 4269
    },
    {
      "epoch": 0.3231635211200984,
      "grad_norm": 0.8515625,
      "learning_rate": 1.8763158607995264e-05,
      "loss": 0.3464,
      "step": 4270
    },
    {
      "epoch": 0.32323920344354573,
      "grad_norm": 0.7578125,
      "learning_rate": 1.876258475855309e-05,
      "loss": 0.3135,
      "step": 4271
    },
    {
      "epoch": 0.323314885766993,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8762010784798616e-05,
      "loss": 0.3275,
      "step": 4272
    },
    {
      "epoch": 0.32339056809044037,
      "grad_norm": 1.078125,
      "learning_rate": 1.8761436686739976e-05,
      "loss": 0.3915,
      "step": 4273
    },
    {
      "epoch": 0.3234662504138877,
      "grad_norm": 1.0,
      "learning_rate": 1.8760862464385327e-05,
      "loss": 0.285,
      "step": 4274
    },
    {
      "epoch": 0.32354193273733506,
      "grad_norm": 0.796875,
      "learning_rate": 1.8760288117742812e-05,
      "loss": 0.3377,
      "step": 4275
    },
    {
      "epoch": 0.32361761506078235,
      "grad_norm": 0.828125,
      "learning_rate": 1.875971364682057e-05,
      "loss": 0.3568,
      "step": 4276
    },
    {
      "epoch": 0.3236932973842297,
      "grad_norm": 0.859375,
      "learning_rate": 1.8759139051626765e-05,
      "loss": 0.3795,
      "step": 4277
    },
    {
      "epoch": 0.32376897970767704,
      "grad_norm": 0.82421875,
      "learning_rate": 1.875856433216954e-05,
      "loss": 0.3493,
      "step": 4278
    },
    {
      "epoch": 0.32384466203112433,
      "grad_norm": 0.875,
      "learning_rate": 1.8757989488457045e-05,
      "loss": 0.3549,
      "step": 4279
    },
    {
      "epoch": 0.3239203443545717,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8757414520497442e-05,
      "loss": 0.3394,
      "step": 4280
    },
    {
      "epoch": 0.323996026678019,
      "grad_norm": 0.76171875,
      "learning_rate": 1.875683942829889e-05,
      "loss": 0.2899,
      "step": 4281
    },
    {
      "epoch": 0.32407170900146637,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8756264211869537e-05,
      "loss": 0.3294,
      "step": 4282
    },
    {
      "epoch": 0.32414739132491366,
      "grad_norm": 0.828125,
      "learning_rate": 1.8755688871217552e-05,
      "loss": 0.3454,
      "step": 4283
    },
    {
      "epoch": 0.324223073648361,
      "grad_norm": 0.78515625,
      "learning_rate": 1.87551134063511e-05,
      "loss": 0.3305,
      "step": 4284
    },
    {
      "epoch": 0.32429875597180835,
      "grad_norm": 0.83984375,
      "learning_rate": 1.8754537817278333e-05,
      "loss": 0.3064,
      "step": 4285
    },
    {
      "epoch": 0.32437443829525564,
      "grad_norm": 0.83984375,
      "learning_rate": 1.8753962104007426e-05,
      "loss": 0.3549,
      "step": 4286
    },
    {
      "epoch": 0.324450120618703,
      "grad_norm": 0.87109375,
      "learning_rate": 1.8753386266546548e-05,
      "loss": 0.3635,
      "step": 4287
    },
    {
      "epoch": 0.32452580294215033,
      "grad_norm": 0.82421875,
      "learning_rate": 1.875281030490386e-05,
      "loss": 0.3321,
      "step": 4288
    },
    {
      "epoch": 0.3246014852655977,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8752234219087538e-05,
      "loss": 0.3515,
      "step": 4289
    },
    {
      "epoch": 0.32467716758904497,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8751658009105754e-05,
      "loss": 0.3875,
      "step": 4290
    },
    {
      "epoch": 0.3247528499124923,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8751081674966687e-05,
      "loss": 0.3312,
      "step": 4291
    },
    {
      "epoch": 0.32482853223593966,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8750505216678505e-05,
      "loss": 0.3067,
      "step": 4292
    },
    {
      "epoch": 0.32490421455938695,
      "grad_norm": 0.80859375,
      "learning_rate": 1.874992863424939e-05,
      "loss": 0.3316,
      "step": 4293
    },
    {
      "epoch": 0.3249798968828343,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8749351927687526e-05,
      "loss": 0.3162,
      "step": 4294
    },
    {
      "epoch": 0.32505557920628164,
      "grad_norm": 0.8125,
      "learning_rate": 1.8748775097001086e-05,
      "loss": 0.3124,
      "step": 4295
    },
    {
      "epoch": 0.325131261529729,
      "grad_norm": 0.80859375,
      "learning_rate": 1.874819814219826e-05,
      "loss": 0.3278,
      "step": 4296
    },
    {
      "epoch": 0.3252069438531763,
      "grad_norm": 0.75390625,
      "learning_rate": 1.874762106328723e-05,
      "loss": 0.3062,
      "step": 4297
    },
    {
      "epoch": 0.3252826261766236,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8747043860276185e-05,
      "loss": 0.2856,
      "step": 4298
    },
    {
      "epoch": 0.32535830850007097,
      "grad_norm": 0.90234375,
      "learning_rate": 1.874646653317331e-05,
      "loss": 0.3497,
      "step": 4299
    },
    {
      "epoch": 0.32543399082351826,
      "grad_norm": 1.078125,
      "learning_rate": 1.8745889081986802e-05,
      "loss": 0.3696,
      "step": 4300
    },
    {
      "epoch": 0.3255096731469656,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8745311506724845e-05,
      "loss": 0.3281,
      "step": 4301
    },
    {
      "epoch": 0.32558535547041295,
      "grad_norm": 0.74609375,
      "learning_rate": 1.8744733807395638e-05,
      "loss": 0.3005,
      "step": 4302
    },
    {
      "epoch": 0.3256610377938603,
      "grad_norm": 0.8984375,
      "learning_rate": 1.8744155984007376e-05,
      "loss": 0.3552,
      "step": 4303
    },
    {
      "epoch": 0.3257367201173076,
      "grad_norm": 0.9375,
      "learning_rate": 1.8743578036568257e-05,
      "loss": 0.3316,
      "step": 4304
    },
    {
      "epoch": 0.32581240244075493,
      "grad_norm": 0.91015625,
      "learning_rate": 1.8742999965086474e-05,
      "loss": 0.3856,
      "step": 4305
    },
    {
      "epoch": 0.3258880847642023,
      "grad_norm": 0.828125,
      "learning_rate": 1.8742421769570235e-05,
      "loss": 0.3433,
      "step": 4306
    },
    {
      "epoch": 0.32596376708764957,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8741843450027743e-05,
      "loss": 0.3375,
      "step": 4307
    },
    {
      "epoch": 0.3260394494110969,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8741265006467202e-05,
      "loss": 0.3411,
      "step": 4308
    },
    {
      "epoch": 0.32611513173454426,
      "grad_norm": 0.84765625,
      "learning_rate": 1.874068643889681e-05,
      "loss": 0.3554,
      "step": 4309
    },
    {
      "epoch": 0.3261908140579916,
      "grad_norm": 0.86328125,
      "learning_rate": 1.8740107747324785e-05,
      "loss": 0.34,
      "step": 4310
    },
    {
      "epoch": 0.3262664963814389,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8739528931759334e-05,
      "loss": 0.3085,
      "step": 4311
    },
    {
      "epoch": 0.32634217870488624,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8738949992208667e-05,
      "loss": 0.3209,
      "step": 4312
    },
    {
      "epoch": 0.3264178610283336,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8738370928681e-05,
      "loss": 0.3172,
      "step": 4313
    },
    {
      "epoch": 0.3264935433517809,
      "grad_norm": 0.8359375,
      "learning_rate": 1.873779174118454e-05,
      "loss": 0.352,
      "step": 4314
    },
    {
      "epoch": 0.3265692256752282,
      "grad_norm": 0.73046875,
      "learning_rate": 1.8737212429727512e-05,
      "loss": 0.2782,
      "step": 4315
    },
    {
      "epoch": 0.32664490799867557,
      "grad_norm": 5.0625,
      "learning_rate": 1.8736632994318133e-05,
      "loss": 0.4819,
      "step": 4316
    },
    {
      "epoch": 0.3267205903221229,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8736053434964625e-05,
      "loss": 0.3472,
      "step": 4317
    },
    {
      "epoch": 0.3267962726455702,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8735473751675204e-05,
      "loss": 0.3198,
      "step": 4318
    },
    {
      "epoch": 0.32687195496901755,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8734893944458097e-05,
      "loss": 0.3164,
      "step": 4319
    },
    {
      "epoch": 0.3269476372924649,
      "grad_norm": 0.76171875,
      "learning_rate": 1.873431401332153e-05,
      "loss": 0.2858,
      "step": 4320
    },
    {
      "epoch": 0.3270233196159122,
      "grad_norm": 0.84765625,
      "learning_rate": 1.8733733958273734e-05,
      "loss": 0.3472,
      "step": 4321
    },
    {
      "epoch": 0.32709900193935953,
      "grad_norm": 0.8359375,
      "learning_rate": 1.873315377932293e-05,
      "loss": 0.3421,
      "step": 4322
    },
    {
      "epoch": 0.3271746842628069,
      "grad_norm": 1.390625,
      "learning_rate": 1.8732573476477355e-05,
      "loss": 0.3893,
      "step": 4323
    },
    {
      "epoch": 0.32725036658625417,
      "grad_norm": 1.1015625,
      "learning_rate": 1.8731993049745236e-05,
      "loss": 0.3748,
      "step": 4324
    },
    {
      "epoch": 0.3273260489097015,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8731412499134816e-05,
      "loss": 0.3011,
      "step": 4325
    },
    {
      "epoch": 0.32740173123314886,
      "grad_norm": 1.0703125,
      "learning_rate": 1.873083182465432e-05,
      "loss": 0.3398,
      "step": 4326
    },
    {
      "epoch": 0.3274774135565962,
      "grad_norm": 0.8125,
      "learning_rate": 1.8730251026312e-05,
      "loss": 0.3008,
      "step": 4327
    },
    {
      "epoch": 0.3275530958800435,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8729670104116083e-05,
      "loss": 0.2873,
      "step": 4328
    },
    {
      "epoch": 0.32762877820349084,
      "grad_norm": 0.8515625,
      "learning_rate": 1.8729089058074815e-05,
      "loss": 0.3568,
      "step": 4329
    },
    {
      "epoch": 0.3277044605269382,
      "grad_norm": 0.80078125,
      "learning_rate": 1.872850788819644e-05,
      "loss": 0.3286,
      "step": 4330
    },
    {
      "epoch": 0.3277801428503855,
      "grad_norm": 0.765625,
      "learning_rate": 1.8727926594489202e-05,
      "loss": 0.3136,
      "step": 4331
    },
    {
      "epoch": 0.3278558251738328,
      "grad_norm": 0.8671875,
      "learning_rate": 1.8727345176961343e-05,
      "loss": 0.3871,
      "step": 4332
    },
    {
      "epoch": 0.3279315074972802,
      "grad_norm": 0.83203125,
      "learning_rate": 1.8726763635621124e-05,
      "loss": 0.3421,
      "step": 4333
    },
    {
      "epoch": 0.3280071898207275,
      "grad_norm": 0.7890625,
      "learning_rate": 1.872618197047678e-05,
      "loss": 0.3138,
      "step": 4334
    },
    {
      "epoch": 0.3280828721441748,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8725600181536575e-05,
      "loss": 0.291,
      "step": 4335
    },
    {
      "epoch": 0.32815855446762215,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8725018268808756e-05,
      "loss": 0.3215,
      "step": 4336
    },
    {
      "epoch": 0.3282342367910695,
      "grad_norm": 0.8125,
      "learning_rate": 1.872443623230158e-05,
      "loss": 0.3308,
      "step": 4337
    },
    {
      "epoch": 0.3283099191145168,
      "grad_norm": 1.140625,
      "learning_rate": 1.8723854072023303e-05,
      "loss": 0.3909,
      "step": 4338
    },
    {
      "epoch": 0.32838560143796414,
      "grad_norm": 0.77734375,
      "learning_rate": 1.872327178798219e-05,
      "loss": 0.2746,
      "step": 4339
    },
    {
      "epoch": 0.3284612837614115,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8722689380186492e-05,
      "loss": 0.3475,
      "step": 4340
    },
    {
      "epoch": 0.32853696608485883,
      "grad_norm": 0.875,
      "learning_rate": 1.872210684864448e-05,
      "loss": 0.3392,
      "step": 4341
    },
    {
      "epoch": 0.3286126484083061,
      "grad_norm": 0.78515625,
      "learning_rate": 1.872152419336441e-05,
      "loss": 0.306,
      "step": 4342
    },
    {
      "epoch": 0.32868833073175346,
      "grad_norm": 0.8359375,
      "learning_rate": 1.872094141435456e-05,
      "loss": 0.341,
      "step": 4343
    },
    {
      "epoch": 0.3287640130552008,
      "grad_norm": 0.734375,
      "learning_rate": 1.8720358511623184e-05,
      "loss": 0.2854,
      "step": 4344
    },
    {
      "epoch": 0.3288396953786481,
      "grad_norm": 0.8515625,
      "learning_rate": 1.871977548517856e-05,
      "loss": 0.3794,
      "step": 4345
    },
    {
      "epoch": 0.32891537770209545,
      "grad_norm": 0.8671875,
      "learning_rate": 1.871919233502896e-05,
      "loss": 0.3235,
      "step": 4346
    },
    {
      "epoch": 0.3289910600255428,
      "grad_norm": 0.8671875,
      "learning_rate": 1.871860906118265e-05,
      "loss": 0.3899,
      "step": 4347
    },
    {
      "epoch": 0.32906674234899014,
      "grad_norm": 0.75390625,
      "learning_rate": 1.871802566364791e-05,
      "loss": 0.3299,
      "step": 4348
    },
    {
      "epoch": 0.32914242467243743,
      "grad_norm": 0.828125,
      "learning_rate": 1.8717442142433015e-05,
      "loss": 0.3593,
      "step": 4349
    },
    {
      "epoch": 0.3292181069958848,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8716858497546244e-05,
      "loss": 0.3093,
      "step": 4350
    },
    {
      "epoch": 0.3292937893193321,
      "grad_norm": 0.8515625,
      "learning_rate": 1.8716274728995875e-05,
      "loss": 0.3518,
      "step": 4351
    },
    {
      "epoch": 0.3293694716427794,
      "grad_norm": 0.78125,
      "learning_rate": 1.8715690836790195e-05,
      "loss": 0.3185,
      "step": 4352
    },
    {
      "epoch": 0.32944515396622676,
      "grad_norm": 0.8203125,
      "learning_rate": 1.871510682093748e-05,
      "loss": 0.3244,
      "step": 4353
    },
    {
      "epoch": 0.3295208362896741,
      "grad_norm": 0.828125,
      "learning_rate": 1.8714522681446024e-05,
      "loss": 0.3459,
      "step": 4354
    },
    {
      "epoch": 0.32959651861312145,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8713938418324105e-05,
      "loss": 0.3631,
      "step": 4355
    },
    {
      "epoch": 0.32967220093656874,
      "grad_norm": 0.859375,
      "learning_rate": 1.8713354031580014e-05,
      "loss": 0.3748,
      "step": 4356
    },
    {
      "epoch": 0.3297478832600161,
      "grad_norm": 0.71875,
      "learning_rate": 1.8712769521222046e-05,
      "loss": 0.2868,
      "step": 4357
    },
    {
      "epoch": 0.32982356558346343,
      "grad_norm": 1.390625,
      "learning_rate": 1.8712184887258494e-05,
      "loss": 0.4212,
      "step": 4358
    },
    {
      "epoch": 0.3298992479069107,
      "grad_norm": 0.734375,
      "learning_rate": 1.8711600129697643e-05,
      "loss": 0.2955,
      "step": 4359
    },
    {
      "epoch": 0.32997493023035807,
      "grad_norm": 0.75390625,
      "learning_rate": 1.87110152485478e-05,
      "loss": 0.2912,
      "step": 4360
    },
    {
      "epoch": 0.3300506125538054,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8710430243817253e-05,
      "loss": 0.3313,
      "step": 4361
    },
    {
      "epoch": 0.33012629487725276,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8709845115514306e-05,
      "loss": 0.3785,
      "step": 4362
    },
    {
      "epoch": 0.33020197720070005,
      "grad_norm": 0.84375,
      "learning_rate": 1.8709259863647263e-05,
      "loss": 0.3308,
      "step": 4363
    },
    {
      "epoch": 0.3302776595241474,
      "grad_norm": 0.765625,
      "learning_rate": 1.8708674488224422e-05,
      "loss": 0.3165,
      "step": 4364
    },
    {
      "epoch": 0.33035334184759474,
      "grad_norm": 0.74609375,
      "learning_rate": 1.8708088989254084e-05,
      "loss": 0.2986,
      "step": 4365
    },
    {
      "epoch": 0.33042902417104203,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8707503366744563e-05,
      "loss": 0.29,
      "step": 4366
    },
    {
      "epoch": 0.3305047064944894,
      "grad_norm": 0.765625,
      "learning_rate": 1.8706917620704167e-05,
      "loss": 0.2929,
      "step": 4367
    },
    {
      "epoch": 0.3305803888179367,
      "grad_norm": 0.8828125,
      "learning_rate": 1.87063317511412e-05,
      "loss": 0.3548,
      "step": 4368
    },
    {
      "epoch": 0.33065607114138407,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8705745758063977e-05,
      "loss": 0.3287,
      "step": 4369
    },
    {
      "epoch": 0.33073175346483136,
      "grad_norm": 0.78515625,
      "learning_rate": 1.870515964148081e-05,
      "loss": 0.3427,
      "step": 4370
    },
    {
      "epoch": 0.3308074357882787,
      "grad_norm": 0.828125,
      "learning_rate": 1.8704573401400016e-05,
      "loss": 0.3503,
      "step": 4371
    },
    {
      "epoch": 0.33088311811172605,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8703987037829908e-05,
      "loss": 0.3125,
      "step": 4372
    },
    {
      "epoch": 0.33095880043517334,
      "grad_norm": 0.7890625,
      "learning_rate": 1.870340055077881e-05,
      "loss": 0.296,
      "step": 4373
    },
    {
      "epoch": 0.3310344827586207,
      "grad_norm": 1.6484375,
      "learning_rate": 1.870281394025504e-05,
      "loss": 0.3248,
      "step": 4374
    },
    {
      "epoch": 0.33111016508206803,
      "grad_norm": 3.109375,
      "learning_rate": 1.870222720626692e-05,
      "loss": 0.5142,
      "step": 4375
    },
    {
      "epoch": 0.3311858474055154,
      "grad_norm": 0.75390625,
      "learning_rate": 1.870164034882277e-05,
      "loss": 0.3061,
      "step": 4376
    },
    {
      "epoch": 0.33126152972896267,
      "grad_norm": 0.84375,
      "learning_rate": 1.8701053367930923e-05,
      "loss": 0.3678,
      "step": 4377
    },
    {
      "epoch": 0.33133721205241,
      "grad_norm": 0.91015625,
      "learning_rate": 1.8700466263599704e-05,
      "loss": 0.3642,
      "step": 4378
    },
    {
      "epoch": 0.33141289437585736,
      "grad_norm": 1.25,
      "learning_rate": 1.8699879035837435e-05,
      "loss": 0.3755,
      "step": 4379
    },
    {
      "epoch": 0.33148857669930465,
      "grad_norm": 0.85546875,
      "learning_rate": 1.869929168465246e-05,
      "loss": 0.3555,
      "step": 4380
    },
    {
      "epoch": 0.331564259022752,
      "grad_norm": 0.703125,
      "learning_rate": 1.8698704210053097e-05,
      "loss": 0.2655,
      "step": 4381
    },
    {
      "epoch": 0.33163994134619934,
      "grad_norm": 0.84765625,
      "learning_rate": 1.8698116612047688e-05,
      "loss": 0.3597,
      "step": 4382
    },
    {
      "epoch": 0.33171562366964663,
      "grad_norm": 0.859375,
      "learning_rate": 1.8697528890644568e-05,
      "loss": 0.3534,
      "step": 4383
    },
    {
      "epoch": 0.331791305993094,
      "grad_norm": 0.87890625,
      "learning_rate": 1.8696941045852078e-05,
      "loss": 0.3615,
      "step": 4384
    },
    {
      "epoch": 0.3318669883165413,
      "grad_norm": 0.95703125,
      "learning_rate": 1.8696353077678557e-05,
      "loss": 0.36,
      "step": 4385
    },
    {
      "epoch": 0.33194267063998867,
      "grad_norm": 0.83203125,
      "learning_rate": 1.869576498613234e-05,
      "loss": 0.3426,
      "step": 4386
    },
    {
      "epoch": 0.33201835296343596,
      "grad_norm": 0.76171875,
      "learning_rate": 1.8695176771221775e-05,
      "loss": 0.2982,
      "step": 4387
    },
    {
      "epoch": 0.3320940352868833,
      "grad_norm": 0.765625,
      "learning_rate": 1.8694588432955205e-05,
      "loss": 0.2692,
      "step": 4388
    },
    {
      "epoch": 0.33216971761033065,
      "grad_norm": 0.89453125,
      "learning_rate": 1.869399997134098e-05,
      "loss": 0.3527,
      "step": 4389
    },
    {
      "epoch": 0.33224539993377794,
      "grad_norm": 0.875,
      "learning_rate": 1.8693411386387445e-05,
      "loss": 0.3769,
      "step": 4390
    },
    {
      "epoch": 0.3323210822572253,
      "grad_norm": 0.9140625,
      "learning_rate": 1.869282267810295e-05,
      "loss": 0.405,
      "step": 4391
    },
    {
      "epoch": 0.33239676458067263,
      "grad_norm": 0.81640625,
      "learning_rate": 1.869223384649585e-05,
      "loss": 0.3447,
      "step": 4392
    },
    {
      "epoch": 0.33247244690412,
      "grad_norm": 0.828125,
      "learning_rate": 1.8691644891574493e-05,
      "loss": 0.3565,
      "step": 4393
    },
    {
      "epoch": 0.33254812922756727,
      "grad_norm": 0.83984375,
      "learning_rate": 1.8691055813347245e-05,
      "loss": 0.3807,
      "step": 4394
    },
    {
      "epoch": 0.3326238115510146,
      "grad_norm": 0.7734375,
      "learning_rate": 1.869046661182245e-05,
      "loss": 0.3191,
      "step": 4395
    },
    {
      "epoch": 0.33269949387446196,
      "grad_norm": 0.9609375,
      "learning_rate": 1.868987728700847e-05,
      "loss": 0.3922,
      "step": 4396
    },
    {
      "epoch": 0.33277517619790925,
      "grad_norm": 0.76953125,
      "learning_rate": 1.8689287838913674e-05,
      "loss": 0.3295,
      "step": 4397
    },
    {
      "epoch": 0.3328508585213566,
      "grad_norm": 0.79296875,
      "learning_rate": 1.868869826754642e-05,
      "loss": 0.3267,
      "step": 4398
    },
    {
      "epoch": 0.33292654084480394,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8688108572915068e-05,
      "loss": 0.345,
      "step": 4399
    },
    {
      "epoch": 0.3330022231682513,
      "grad_norm": 0.765625,
      "learning_rate": 1.8687518755027984e-05,
      "loss": 0.3313,
      "step": 4400
    },
    {
      "epoch": 0.3330779054916986,
      "grad_norm": 1.40625,
      "learning_rate": 1.8686928813893538e-05,
      "loss": 0.3832,
      "step": 4401
    },
    {
      "epoch": 0.3331535878151459,
      "grad_norm": 1.28125,
      "learning_rate": 1.8686338749520107e-05,
      "loss": 0.4082,
      "step": 4402
    },
    {
      "epoch": 0.33322927013859327,
      "grad_norm": 1.0234375,
      "learning_rate": 1.8685748561916047e-05,
      "loss": 0.3448,
      "step": 4403
    },
    {
      "epoch": 0.33330495246204056,
      "grad_norm": 0.7890625,
      "learning_rate": 1.868515825108974e-05,
      "loss": 0.3167,
      "step": 4404
    },
    {
      "epoch": 0.3333806347854879,
      "grad_norm": 0.8203125,
      "learning_rate": 1.868456781704956e-05,
      "loss": 0.3161,
      "step": 4405
    },
    {
      "epoch": 0.33345631710893525,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8683977259803883e-05,
      "loss": 0.3539,
      "step": 4406
    },
    {
      "epoch": 0.3335319994323826,
      "grad_norm": 1.234375,
      "learning_rate": 1.8683386579361086e-05,
      "loss": 0.3318,
      "step": 4407
    },
    {
      "epoch": 0.3336076817558299,
      "grad_norm": 0.8125,
      "learning_rate": 1.8682795775729547e-05,
      "loss": 0.3236,
      "step": 4408
    },
    {
      "epoch": 0.33368336407927723,
      "grad_norm": 0.89453125,
      "learning_rate": 1.868220484891765e-05,
      "loss": 0.415,
      "step": 4409
    },
    {
      "epoch": 0.3337590464027246,
      "grad_norm": 0.8671875,
      "learning_rate": 1.868161379893378e-05,
      "loss": 0.3148,
      "step": 4410
    },
    {
      "epoch": 0.33383472872617187,
      "grad_norm": 0.828125,
      "learning_rate": 1.868102262578632e-05,
      "loss": 0.3273,
      "step": 4411
    },
    {
      "epoch": 0.3339104110496192,
      "grad_norm": 1.3984375,
      "learning_rate": 1.868043132948366e-05,
      "loss": 0.3939,
      "step": 4412
    },
    {
      "epoch": 0.33398609337306656,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8679839910034177e-05,
      "loss": 0.3271,
      "step": 4413
    },
    {
      "epoch": 0.3340617756965139,
      "grad_norm": 0.78125,
      "learning_rate": 1.8679248367446278e-05,
      "loss": 0.3221,
      "step": 4414
    },
    {
      "epoch": 0.3341374580199612,
      "grad_norm": 0.89453125,
      "learning_rate": 1.8678656701728342e-05,
      "loss": 0.3609,
      "step": 4415
    },
    {
      "epoch": 0.33421314034340854,
      "grad_norm": 0.76953125,
      "learning_rate": 1.8678064912888768e-05,
      "loss": 0.3208,
      "step": 4416
    },
    {
      "epoch": 0.3342888226668559,
      "grad_norm": 0.890625,
      "learning_rate": 1.8677473000935956e-05,
      "loss": 0.3573,
      "step": 4417
    },
    {
      "epoch": 0.3343645049903032,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8676880965878293e-05,
      "loss": 0.2952,
      "step": 4418
    },
    {
      "epoch": 0.3344401873137505,
      "grad_norm": 0.8515625,
      "learning_rate": 1.8676288807724184e-05,
      "loss": 0.3531,
      "step": 4419
    },
    {
      "epoch": 0.33451586963719787,
      "grad_norm": 0.84375,
      "learning_rate": 1.8675696526482033e-05,
      "loss": 0.3482,
      "step": 4420
    },
    {
      "epoch": 0.3345915519606452,
      "grad_norm": 0.9375,
      "learning_rate": 1.8675104122160235e-05,
      "loss": 0.3982,
      "step": 4421
    },
    {
      "epoch": 0.3346672342840925,
      "grad_norm": 0.79296875,
      "learning_rate": 1.86745115947672e-05,
      "loss": 0.3455,
      "step": 4422
    },
    {
      "epoch": 0.33474291660753985,
      "grad_norm": 0.7890625,
      "learning_rate": 1.867391894431133e-05,
      "loss": 0.3144,
      "step": 4423
    },
    {
      "epoch": 0.3348185989309872,
      "grad_norm": 0.765625,
      "learning_rate": 1.8673326170801033e-05,
      "loss": 0.328,
      "step": 4424
    },
    {
      "epoch": 0.3348942812544345,
      "grad_norm": 1.0078125,
      "learning_rate": 1.8672733274244727e-05,
      "loss": 0.3465,
      "step": 4425
    },
    {
      "epoch": 0.33496996357788184,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8672140254650813e-05,
      "loss": 0.344,
      "step": 4426
    },
    {
      "epoch": 0.3350456459013292,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8671547112027708e-05,
      "loss": 0.3243,
      "step": 4427
    },
    {
      "epoch": 0.3351213282247765,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8670953846383825e-05,
      "loss": 0.3037,
      "step": 4428
    },
    {
      "epoch": 0.3351970105482238,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8670360457727583e-05,
      "loss": 0.3397,
      "step": 4429
    },
    {
      "epoch": 0.33527269287167116,
      "grad_norm": 0.92578125,
      "learning_rate": 1.8669766946067398e-05,
      "loss": 0.375,
      "step": 4430
    },
    {
      "epoch": 0.3353483751951185,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8669173311411694e-05,
      "loss": 0.3301,
      "step": 4431
    },
    {
      "epoch": 0.3354240575185658,
      "grad_norm": 1.2890625,
      "learning_rate": 1.8668579553768882e-05,
      "loss": 0.397,
      "step": 4432
    },
    {
      "epoch": 0.33549973984201314,
      "grad_norm": 0.89453125,
      "learning_rate": 1.86679856731474e-05,
      "loss": 0.3633,
      "step": 4433
    },
    {
      "epoch": 0.3355754221654605,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8667391669555668e-05,
      "loss": 0.3258,
      "step": 4434
    },
    {
      "epoch": 0.33565110448890784,
      "grad_norm": 0.8203125,
      "learning_rate": 1.866679754300211e-05,
      "loss": 0.3028,
      "step": 4435
    },
    {
      "epoch": 0.3357267868123551,
      "grad_norm": 0.77734375,
      "learning_rate": 1.866620329349515e-05,
      "loss": 0.3312,
      "step": 4436
    },
    {
      "epoch": 0.3358024691358025,
      "grad_norm": 0.734375,
      "learning_rate": 1.8665608921043232e-05,
      "loss": 0.3036,
      "step": 4437
    },
    {
      "epoch": 0.3358781514592498,
      "grad_norm": 0.8125,
      "learning_rate": 1.8665014425654778e-05,
      "loss": 0.348,
      "step": 4438
    },
    {
      "epoch": 0.3359538337826971,
      "grad_norm": 0.8984375,
      "learning_rate": 1.8664419807338224e-05,
      "loss": 0.3984,
      "step": 4439
    },
    {
      "epoch": 0.33602951610614445,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8663825066102008e-05,
      "loss": 0.3375,
      "step": 4440
    },
    {
      "epoch": 0.3361051984295918,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8663230201954566e-05,
      "loss": 0.3355,
      "step": 4441
    },
    {
      "epoch": 0.3361808807530391,
      "grad_norm": 1.90625,
      "learning_rate": 1.8662635214904333e-05,
      "loss": 0.4336,
      "step": 4442
    },
    {
      "epoch": 0.33625656307648644,
      "grad_norm": 0.8984375,
      "learning_rate": 1.8662040104959757e-05,
      "loss": 0.3604,
      "step": 4443
    },
    {
      "epoch": 0.3363322453999338,
      "grad_norm": 0.95703125,
      "learning_rate": 1.8661444872129277e-05,
      "loss": 0.365,
      "step": 4444
    },
    {
      "epoch": 0.33640792772338113,
      "grad_norm": 0.78125,
      "learning_rate": 1.866084951642134e-05,
      "loss": 0.3031,
      "step": 4445
    },
    {
      "epoch": 0.3364836100468284,
      "grad_norm": 0.83203125,
      "learning_rate": 1.866025403784439e-05,
      "loss": 0.3385,
      "step": 4446
    },
    {
      "epoch": 0.33655929237027576,
      "grad_norm": 0.82421875,
      "learning_rate": 1.865965843640687e-05,
      "loss": 0.3427,
      "step": 4447
    },
    {
      "epoch": 0.3366349746937231,
      "grad_norm": 1.453125,
      "learning_rate": 1.865906271211724e-05,
      "loss": 0.444,
      "step": 4448
    },
    {
      "epoch": 0.3367106570171704,
      "grad_norm": 0.7890625,
      "learning_rate": 1.865846686498394e-05,
      "loss": 0.3115,
      "step": 4449
    },
    {
      "epoch": 0.33678633934061775,
      "grad_norm": 0.796875,
      "learning_rate": 1.8657870895015437e-05,
      "loss": 0.3108,
      "step": 4450
    },
    {
      "epoch": 0.3368620216640651,
      "grad_norm": 0.86328125,
      "learning_rate": 1.8657274802220172e-05,
      "loss": 0.373,
      "step": 4451
    },
    {
      "epoch": 0.33693770398751244,
      "grad_norm": 0.8203125,
      "learning_rate": 1.865667858660661e-05,
      "loss": 0.3503,
      "step": 4452
    },
    {
      "epoch": 0.33701338631095973,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8656082248183208e-05,
      "loss": 0.3226,
      "step": 4453
    },
    {
      "epoch": 0.3370890686344071,
      "grad_norm": 0.83984375,
      "learning_rate": 1.8655485786958423e-05,
      "loss": 0.2886,
      "step": 4454
    },
    {
      "epoch": 0.3371647509578544,
      "grad_norm": 0.875,
      "learning_rate": 1.865488920294072e-05,
      "loss": 0.3613,
      "step": 4455
    },
    {
      "epoch": 0.3372404332813017,
      "grad_norm": 0.76953125,
      "learning_rate": 1.8654292496138558e-05,
      "loss": 0.2952,
      "step": 4456
    },
    {
      "epoch": 0.33731611560474906,
      "grad_norm": 0.8125,
      "learning_rate": 1.8653695666560408e-05,
      "loss": 0.3204,
      "step": 4457
    },
    {
      "epoch": 0.3373917979281964,
      "grad_norm": 0.734375,
      "learning_rate": 1.8653098714214736e-05,
      "loss": 0.2726,
      "step": 4458
    },
    {
      "epoch": 0.33746748025164375,
      "grad_norm": 1.8203125,
      "learning_rate": 1.865250163911001e-05,
      "loss": 0.4156,
      "step": 4459
    },
    {
      "epoch": 0.33754316257509104,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8651904441254696e-05,
      "loss": 0.3484,
      "step": 4460
    },
    {
      "epoch": 0.3376188448985384,
      "grad_norm": 0.875,
      "learning_rate": 1.8651307120657273e-05,
      "loss": 0.3445,
      "step": 4461
    },
    {
      "epoch": 0.33769452722198573,
      "grad_norm": 0.83203125,
      "learning_rate": 1.8650709677326212e-05,
      "loss": 0.3475,
      "step": 4462
    },
    {
      "epoch": 0.337770209545433,
      "grad_norm": 0.8984375,
      "learning_rate": 1.8650112111269987e-05,
      "loss": 0.361,
      "step": 4463
    },
    {
      "epoch": 0.33784589186888037,
      "grad_norm": 0.796875,
      "learning_rate": 1.8649514422497083e-05,
      "loss": 0.2952,
      "step": 4464
    },
    {
      "epoch": 0.3379215741923277,
      "grad_norm": 0.81640625,
      "learning_rate": 1.864891661101597e-05,
      "loss": 0.3368,
      "step": 4465
    },
    {
      "epoch": 0.33799725651577506,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8648318676835132e-05,
      "loss": 0.3414,
      "step": 4466
    },
    {
      "epoch": 0.33807293883922235,
      "grad_norm": 0.89453125,
      "learning_rate": 1.8647720619963054e-05,
      "loss": 0.3885,
      "step": 4467
    },
    {
      "epoch": 0.3381486211626697,
      "grad_norm": 0.83984375,
      "learning_rate": 1.8647122440408218e-05,
      "loss": 0.3524,
      "step": 4468
    },
    {
      "epoch": 0.33822430348611704,
      "grad_norm": 0.8671875,
      "learning_rate": 1.8646524138179115e-05,
      "loss": 0.3955,
      "step": 4469
    },
    {
      "epoch": 0.33829998580956433,
      "grad_norm": 0.83203125,
      "learning_rate": 1.8645925713284225e-05,
      "loss": 0.3428,
      "step": 4470
    },
    {
      "epoch": 0.3383756681330117,
      "grad_norm": 0.73828125,
      "learning_rate": 1.8645327165732045e-05,
      "loss": 0.3099,
      "step": 4471
    },
    {
      "epoch": 0.338451350456459,
      "grad_norm": 0.828125,
      "learning_rate": 1.864472849553106e-05,
      "loss": 0.3227,
      "step": 4472
    },
    {
      "epoch": 0.33852703277990637,
      "grad_norm": 1.15625,
      "learning_rate": 1.8644129702689767e-05,
      "loss": 0.3765,
      "step": 4473
    },
    {
      "epoch": 0.33860271510335366,
      "grad_norm": 0.84765625,
      "learning_rate": 1.864353078721666e-05,
      "loss": 0.3188,
      "step": 4474
    },
    {
      "epoch": 0.338678397426801,
      "grad_norm": 1.1171875,
      "learning_rate": 1.864293174912024e-05,
      "loss": 0.359,
      "step": 4475
    },
    {
      "epoch": 0.33875407975024835,
      "grad_norm": 0.796875,
      "learning_rate": 1.8642332588408998e-05,
      "loss": 0.3295,
      "step": 4476
    },
    {
      "epoch": 0.33882976207369564,
      "grad_norm": 0.828125,
      "learning_rate": 1.8641733305091437e-05,
      "loss": 0.3214,
      "step": 4477
    },
    {
      "epoch": 0.338905444397143,
      "grad_norm": 0.74609375,
      "learning_rate": 1.8641133899176062e-05,
      "loss": 0.2766,
      "step": 4478
    },
    {
      "epoch": 0.33898112672059033,
      "grad_norm": 0.85546875,
      "learning_rate": 1.864053437067137e-05,
      "loss": 0.3383,
      "step": 4479
    },
    {
      "epoch": 0.3390568090440377,
      "grad_norm": 0.74609375,
      "learning_rate": 1.8639934719585874e-05,
      "loss": 0.277,
      "step": 4480
    },
    {
      "epoch": 0.33913249136748497,
      "grad_norm": 0.7421875,
      "learning_rate": 1.8639334945928075e-05,
      "loss": 0.2955,
      "step": 4481
    },
    {
      "epoch": 0.3392081736909323,
      "grad_norm": 0.78125,
      "learning_rate": 1.8638735049706486e-05,
      "loss": 0.3127,
      "step": 4482
    },
    {
      "epoch": 0.33928385601437966,
      "grad_norm": 2.203125,
      "learning_rate": 1.8638135030929618e-05,
      "loss": 0.3112,
      "step": 4483
    },
    {
      "epoch": 0.33935953833782695,
      "grad_norm": 0.89453125,
      "learning_rate": 1.8637534889605974e-05,
      "loss": 0.3479,
      "step": 4484
    },
    {
      "epoch": 0.3394352206612743,
      "grad_norm": 1.546875,
      "learning_rate": 1.863693462574408e-05,
      "loss": 0.417,
      "step": 4485
    },
    {
      "epoch": 0.33951090298472164,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8636334239352445e-05,
      "loss": 0.3121,
      "step": 4486
    },
    {
      "epoch": 0.339586585308169,
      "grad_norm": 0.7890625,
      "learning_rate": 1.863573373043959e-05,
      "loss": 0.3149,
      "step": 4487
    },
    {
      "epoch": 0.3396622676316163,
      "grad_norm": 1.890625,
      "learning_rate": 1.8635133099014032e-05,
      "loss": 0.3383,
      "step": 4488
    },
    {
      "epoch": 0.3397379499550636,
      "grad_norm": 5.0,
      "learning_rate": 1.8634532345084292e-05,
      "loss": 0.2517,
      "step": 4489
    },
    {
      "epoch": 0.33981363227851097,
      "grad_norm": 0.84765625,
      "learning_rate": 1.8633931468658894e-05,
      "loss": 0.3093,
      "step": 4490
    },
    {
      "epoch": 0.33988931460195826,
      "grad_norm": 0.80859375,
      "learning_rate": 1.863333046974636e-05,
      "loss": 0.3364,
      "step": 4491
    },
    {
      "epoch": 0.3399649969254056,
      "grad_norm": 0.90625,
      "learning_rate": 1.863272934835522e-05,
      "loss": 0.3774,
      "step": 4492
    },
    {
      "epoch": 0.34004067924885295,
      "grad_norm": 4.40625,
      "learning_rate": 1.8632128104494e-05,
      "loss": 0.3095,
      "step": 4493
    },
    {
      "epoch": 0.3401163615723003,
      "grad_norm": 1.1484375,
      "learning_rate": 1.8631526738171226e-05,
      "loss": 0.329,
      "step": 4494
    },
    {
      "epoch": 0.3401920438957476,
      "grad_norm": 0.8671875,
      "learning_rate": 1.863092524939544e-05,
      "loss": 0.3255,
      "step": 4495
    },
    {
      "epoch": 0.34026772621919493,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8630323638175164e-05,
      "loss": 0.3119,
      "step": 4496
    },
    {
      "epoch": 0.3403434085426423,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8629721904518935e-05,
      "loss": 0.3392,
      "step": 4497
    },
    {
      "epoch": 0.34041909086608957,
      "grad_norm": 0.73046875,
      "learning_rate": 1.8629120048435294e-05,
      "loss": 0.2968,
      "step": 4498
    },
    {
      "epoch": 0.3404947731895369,
      "grad_norm": 0.99609375,
      "learning_rate": 1.8628518069932777e-05,
      "loss": 0.3341,
      "step": 4499
    },
    {
      "epoch": 0.34057045551298426,
      "grad_norm": 1.5078125,
      "learning_rate": 1.8627915969019925e-05,
      "loss": 0.4053,
      "step": 4500
    },
    {
      "epoch": 0.3406461378364316,
      "grad_norm": 0.8046875,
      "learning_rate": 1.862731374570528e-05,
      "loss": 0.3064,
      "step": 4501
    },
    {
      "epoch": 0.3407218201598789,
      "grad_norm": 0.875,
      "learning_rate": 1.862671139999738e-05,
      "loss": 0.36,
      "step": 4502
    },
    {
      "epoch": 0.34079750248332624,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8626108931904777e-05,
      "loss": 0.3297,
      "step": 4503
    },
    {
      "epoch": 0.3408731848067736,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8625506341436016e-05,
      "loss": 0.3223,
      "step": 4504
    },
    {
      "epoch": 0.3409488671302209,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8624903628599647e-05,
      "loss": 0.3725,
      "step": 4505
    },
    {
      "epoch": 0.3410245494536682,
      "grad_norm": 0.82421875,
      "learning_rate": 1.862430079340422e-05,
      "loss": 0.3484,
      "step": 4506
    },
    {
      "epoch": 0.34110023177711557,
      "grad_norm": 0.80078125,
      "learning_rate": 1.862369783585828e-05,
      "loss": 0.2912,
      "step": 4507
    },
    {
      "epoch": 0.34117591410056286,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8623094755970395e-05,
      "loss": 0.3353,
      "step": 4508
    },
    {
      "epoch": 0.3412515964240102,
      "grad_norm": 0.74609375,
      "learning_rate": 1.8622491553749106e-05,
      "loss": 0.2826,
      "step": 4509
    },
    {
      "epoch": 0.34132727874745755,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8621888229202985e-05,
      "loss": 0.3361,
      "step": 4510
    },
    {
      "epoch": 0.3414029610709049,
      "grad_norm": 0.80859375,
      "learning_rate": 1.862128478234058e-05,
      "loss": 0.3458,
      "step": 4511
    },
    {
      "epoch": 0.3414786433943522,
      "grad_norm": 0.8125,
      "learning_rate": 1.862068121317046e-05,
      "loss": 0.3139,
      "step": 4512
    },
    {
      "epoch": 0.34155432571779953,
      "grad_norm": 0.796875,
      "learning_rate": 1.8620077521701177e-05,
      "loss": 0.3137,
      "step": 4513
    },
    {
      "epoch": 0.3416300080412469,
      "grad_norm": 0.8515625,
      "learning_rate": 1.8619473707941304e-05,
      "loss": 0.3396,
      "step": 4514
    },
    {
      "epoch": 0.34170569036469417,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8618869771899404e-05,
      "loss": 0.3314,
      "step": 4515
    },
    {
      "epoch": 0.3417813726881415,
      "grad_norm": 0.87109375,
      "learning_rate": 1.8618265713584047e-05,
      "loss": 0.3517,
      "step": 4516
    },
    {
      "epoch": 0.34185705501158886,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8617661533003803e-05,
      "loss": 0.3125,
      "step": 4517
    },
    {
      "epoch": 0.3419327373350362,
      "grad_norm": 0.86328125,
      "learning_rate": 1.8617057230167238e-05,
      "loss": 0.3912,
      "step": 4518
    },
    {
      "epoch": 0.3420084196584835,
      "grad_norm": 0.79296875,
      "learning_rate": 1.861645280508293e-05,
      "loss": 0.326,
      "step": 4519
    },
    {
      "epoch": 0.34208410198193084,
      "grad_norm": 0.859375,
      "learning_rate": 1.8615848257759458e-05,
      "loss": 0.3618,
      "step": 4520
    },
    {
      "epoch": 0.3421597843053782,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8615243588205386e-05,
      "loss": 0.3176,
      "step": 4521
    },
    {
      "epoch": 0.3422354666288255,
      "grad_norm": 0.76953125,
      "learning_rate": 1.8614638796429302e-05,
      "loss": 0.3103,
      "step": 4522
    },
    {
      "epoch": 0.3423111489522728,
      "grad_norm": 0.74609375,
      "learning_rate": 1.8614033882439787e-05,
      "loss": 0.3001,
      "step": 4523
    },
    {
      "epoch": 0.34238683127572017,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8613428846245414e-05,
      "loss": 0.302,
      "step": 4524
    },
    {
      "epoch": 0.3424625135991675,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8612823687854774e-05,
      "loss": 0.3282,
      "step": 4525
    },
    {
      "epoch": 0.3425381959226148,
      "grad_norm": 0.82421875,
      "learning_rate": 1.861221840727645e-05,
      "loss": 0.3647,
      "step": 4526
    },
    {
      "epoch": 0.34261387824606215,
      "grad_norm": 1.3984375,
      "learning_rate": 1.8611613004519028e-05,
      "loss": 0.4393,
      "step": 4527
    },
    {
      "epoch": 0.3426895605695095,
      "grad_norm": 0.796875,
      "learning_rate": 1.86110074795911e-05,
      "loss": 0.3046,
      "step": 4528
    },
    {
      "epoch": 0.3427652428929568,
      "grad_norm": 0.79296875,
      "learning_rate": 1.861040183250125e-05,
      "loss": 0.3488,
      "step": 4529
    },
    {
      "epoch": 0.34284092521640414,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8609796063258076e-05,
      "loss": 0.3397,
      "step": 4530
    },
    {
      "epoch": 0.3429166075398515,
      "grad_norm": 0.80078125,
      "learning_rate": 1.860919017187017e-05,
      "loss": 0.3162,
      "step": 4531
    },
    {
      "epoch": 0.3429922898632988,
      "grad_norm": 0.85546875,
      "learning_rate": 1.8608584158346124e-05,
      "loss": 0.3503,
      "step": 4532
    },
    {
      "epoch": 0.3430679721867461,
      "grad_norm": 0.86328125,
      "learning_rate": 1.8607978022694545e-05,
      "loss": 0.3637,
      "step": 4533
    },
    {
      "epoch": 0.34314365451019346,
      "grad_norm": 0.8515625,
      "learning_rate": 1.860737176492402e-05,
      "loss": 0.3308,
      "step": 4534
    },
    {
      "epoch": 0.3432193368336408,
      "grad_norm": 0.7890625,
      "learning_rate": 1.860676538504316e-05,
      "loss": 0.3191,
      "step": 4535
    },
    {
      "epoch": 0.3432950191570881,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8606158883060562e-05,
      "loss": 0.3084,
      "step": 4536
    },
    {
      "epoch": 0.34337070148053545,
      "grad_norm": 0.84765625,
      "learning_rate": 1.8605552258984832e-05,
      "loss": 0.3673,
      "step": 4537
    },
    {
      "epoch": 0.3434463838039828,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8604945512824577e-05,
      "loss": 0.2909,
      "step": 4538
    },
    {
      "epoch": 0.34352206612743014,
      "grad_norm": 1.171875,
      "learning_rate": 1.86043386445884e-05,
      "loss": 0.3333,
      "step": 4539
    },
    {
      "epoch": 0.3435977484508774,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8603731654284916e-05,
      "loss": 0.3326,
      "step": 4540
    },
    {
      "epoch": 0.3436734307743248,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8603124541922732e-05,
      "loss": 0.3182,
      "step": 4541
    },
    {
      "epoch": 0.3437491130977721,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8602517307510463e-05,
      "loss": 0.3316,
      "step": 4542
    },
    {
      "epoch": 0.3438247954212194,
      "grad_norm": 0.87109375,
      "learning_rate": 1.8601909951056728e-05,
      "loss": 0.3712,
      "step": 4543
    },
    {
      "epoch": 0.34390047774466675,
      "grad_norm": 0.8671875,
      "learning_rate": 1.8601302472570137e-05,
      "loss": 0.3416,
      "step": 4544
    },
    {
      "epoch": 0.3439761600681141,
      "grad_norm": 0.74609375,
      "learning_rate": 1.8600694872059305e-05,
      "loss": 0.2971,
      "step": 4545
    },
    {
      "epoch": 0.34405184239156145,
      "grad_norm": 0.79296875,
      "learning_rate": 1.860008714953286e-05,
      "loss": 0.2874,
      "step": 4546
    },
    {
      "epoch": 0.34412752471500874,
      "grad_norm": 0.78125,
      "learning_rate": 1.859947930499942e-05,
      "loss": 0.2965,
      "step": 4547
    },
    {
      "epoch": 0.3442032070384561,
      "grad_norm": 0.83984375,
      "learning_rate": 1.8598871338467613e-05,
      "loss": 0.3573,
      "step": 4548
    },
    {
      "epoch": 0.34427888936190343,
      "grad_norm": 1.359375,
      "learning_rate": 1.8598263249946056e-05,
      "loss": 0.4494,
      "step": 4549
    },
    {
      "epoch": 0.3443545716853507,
      "grad_norm": 0.796875,
      "learning_rate": 1.8597655039443384e-05,
      "loss": 0.3344,
      "step": 4550
    },
    {
      "epoch": 0.34443025400879806,
      "grad_norm": 0.84765625,
      "learning_rate": 1.8597046706968217e-05,
      "loss": 0.3722,
      "step": 4551
    },
    {
      "epoch": 0.3445059363322454,
      "grad_norm": 0.85546875,
      "learning_rate": 1.8596438252529186e-05,
      "loss": 0.3461,
      "step": 4552
    },
    {
      "epoch": 0.34458161865569276,
      "grad_norm": 0.875,
      "learning_rate": 1.8595829676134934e-05,
      "loss": 0.3565,
      "step": 4553
    },
    {
      "epoch": 0.34465730097914005,
      "grad_norm": 0.828125,
      "learning_rate": 1.8595220977794085e-05,
      "loss": 0.3488,
      "step": 4554
    },
    {
      "epoch": 0.3447329833025874,
      "grad_norm": 0.765625,
      "learning_rate": 1.8594612157515276e-05,
      "loss": 0.3108,
      "step": 4555
    },
    {
      "epoch": 0.34480866562603474,
      "grad_norm": 0.828125,
      "learning_rate": 1.8594003215307142e-05,
      "loss": 0.3505,
      "step": 4556
    },
    {
      "epoch": 0.34488434794948203,
      "grad_norm": 0.875,
      "learning_rate": 1.8593394151178326e-05,
      "loss": 0.3759,
      "step": 4557
    },
    {
      "epoch": 0.3449600302729294,
      "grad_norm": 0.76171875,
      "learning_rate": 1.859278496513747e-05,
      "loss": 0.3155,
      "step": 4558
    },
    {
      "epoch": 0.3450357125963767,
      "grad_norm": 0.77734375,
      "learning_rate": 1.859217565719321e-05,
      "loss": 0.3387,
      "step": 4559
    },
    {
      "epoch": 0.34511139491982407,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8591566227354194e-05,
      "loss": 0.3139,
      "step": 4560
    },
    {
      "epoch": 0.34518707724327136,
      "grad_norm": 0.74609375,
      "learning_rate": 1.859095667562907e-05,
      "loss": 0.2989,
      "step": 4561
    },
    {
      "epoch": 0.3452627595667187,
      "grad_norm": 0.76953125,
      "learning_rate": 1.8590347002026478e-05,
      "loss": 0.3006,
      "step": 4562
    },
    {
      "epoch": 0.34533844189016605,
      "grad_norm": 0.8671875,
      "learning_rate": 1.8589737206555077e-05,
      "loss": 0.3632,
      "step": 4563
    },
    {
      "epoch": 0.34541412421361334,
      "grad_norm": 0.79296875,
      "learning_rate": 1.858912728922351e-05,
      "loss": 0.3292,
      "step": 4564
    },
    {
      "epoch": 0.3454898065370607,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8588517250040436e-05,
      "loss": 0.3347,
      "step": 4565
    },
    {
      "epoch": 0.34556548886050803,
      "grad_norm": 1.0546875,
      "learning_rate": 1.8587907089014504e-05,
      "loss": 0.3189,
      "step": 4566
    },
    {
      "epoch": 0.3456411711839553,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8587296806154375e-05,
      "loss": 0.2993,
      "step": 4567
    },
    {
      "epoch": 0.34571685350740267,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8586686401468705e-05,
      "loss": 0.3194,
      "step": 4568
    },
    {
      "epoch": 0.34579253583085,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8586075874966152e-05,
      "loss": 0.3061,
      "step": 4569
    },
    {
      "epoch": 0.34586821815429736,
      "grad_norm": 0.8125,
      "learning_rate": 1.8585465226655376e-05,
      "loss": 0.3436,
      "step": 4570
    },
    {
      "epoch": 0.34594390047774465,
      "grad_norm": 0.8828125,
      "learning_rate": 1.8584854456545046e-05,
      "loss": 0.2886,
      "step": 4571
    },
    {
      "epoch": 0.346019582801192,
      "grad_norm": 0.76953125,
      "learning_rate": 1.858424356464382e-05,
      "loss": 0.3195,
      "step": 4572
    },
    {
      "epoch": 0.34609526512463934,
      "grad_norm": 0.91015625,
      "learning_rate": 1.858363255096037e-05,
      "loss": 0.378,
      "step": 4573
    },
    {
      "epoch": 0.34617094744808663,
      "grad_norm": 0.81640625,
      "learning_rate": 1.858302141550336e-05,
      "loss": 0.3453,
      "step": 4574
    },
    {
      "epoch": 0.346246629771534,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8582410158281466e-05,
      "loss": 0.3367,
      "step": 4575
    },
    {
      "epoch": 0.3463223120949813,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8581798779303356e-05,
      "loss": 0.3528,
      "step": 4576
    },
    {
      "epoch": 0.34639799441842867,
      "grad_norm": 0.86328125,
      "learning_rate": 1.85811872785777e-05,
      "loss": 0.3793,
      "step": 4577
    },
    {
      "epoch": 0.34647367674187596,
      "grad_norm": 0.796875,
      "learning_rate": 1.858057565611318e-05,
      "loss": 0.3039,
      "step": 4578
    },
    {
      "epoch": 0.3465493590653233,
      "grad_norm": 4.40625,
      "learning_rate": 1.8579963911918467e-05,
      "loss": 0.4365,
      "step": 4579
    },
    {
      "epoch": 0.34662504138877065,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8579352046002243e-05,
      "loss": 0.3438,
      "step": 4580
    },
    {
      "epoch": 0.34670072371221794,
      "grad_norm": 0.8046875,
      "learning_rate": 1.857874005837319e-05,
      "loss": 0.341,
      "step": 4581
    },
    {
      "epoch": 0.3467764060356653,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8578127949039985e-05,
      "loss": 0.3486,
      "step": 4582
    },
    {
      "epoch": 0.34685208835911263,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8577515718011314e-05,
      "loss": 0.354,
      "step": 4583
    },
    {
      "epoch": 0.34692777068256,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8576903365295864e-05,
      "loss": 0.3128,
      "step": 4584
    },
    {
      "epoch": 0.34700345300600727,
      "grad_norm": 0.87890625,
      "learning_rate": 1.857629089090232e-05,
      "loss": 0.3519,
      "step": 4585
    },
    {
      "epoch": 0.3470791353294546,
      "grad_norm": 0.8515625,
      "learning_rate": 1.857567829483937e-05,
      "loss": 0.3905,
      "step": 4586
    },
    {
      "epoch": 0.34715481765290196,
      "grad_norm": 0.8125,
      "learning_rate": 1.8575065577115713e-05,
      "loss": 0.323,
      "step": 4587
    },
    {
      "epoch": 0.34723049997634925,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8574452737740034e-05,
      "loss": 0.3161,
      "step": 4588
    },
    {
      "epoch": 0.3473061822997966,
      "grad_norm": 0.80078125,
      "learning_rate": 1.857383977672103e-05,
      "loss": 0.3182,
      "step": 4589
    },
    {
      "epoch": 0.34738186462324394,
      "grad_norm": 0.73046875,
      "learning_rate": 1.857322669406739e-05,
      "loss": 0.2838,
      "step": 4590
    },
    {
      "epoch": 0.3474575469466913,
      "grad_norm": 0.83203125,
      "learning_rate": 1.857261348978782e-05,
      "loss": 0.3635,
      "step": 4591
    },
    {
      "epoch": 0.3475332292701386,
      "grad_norm": 0.828125,
      "learning_rate": 1.8572000163891017e-05,
      "loss": 0.348,
      "step": 4592
    },
    {
      "epoch": 0.3476089115935859,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8571386716385682e-05,
      "loss": 0.3112,
      "step": 4593
    },
    {
      "epoch": 0.34768459391703327,
      "grad_norm": 0.8671875,
      "learning_rate": 1.8570773147280518e-05,
      "loss": 0.3332,
      "step": 4594
    },
    {
      "epoch": 0.34776027624048056,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8570159456584225e-05,
      "loss": 0.3294,
      "step": 4595
    },
    {
      "epoch": 0.3478359585639279,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8569545644305517e-05,
      "loss": 0.3314,
      "step": 4596
    },
    {
      "epoch": 0.34791164088737525,
      "grad_norm": 0.8671875,
      "learning_rate": 1.85689317104531e-05,
      "loss": 0.326,
      "step": 4597
    },
    {
      "epoch": 0.3479873232108226,
      "grad_norm": 1.4921875,
      "learning_rate": 1.8568317655035677e-05,
      "loss": 0.429,
      "step": 4598
    },
    {
      "epoch": 0.3480630055342699,
      "grad_norm": 0.8671875,
      "learning_rate": 1.8567703478061963e-05,
      "loss": 0.3751,
      "step": 4599
    },
    {
      "epoch": 0.34813868785771723,
      "grad_norm": 0.9140625,
      "learning_rate": 1.856708917954068e-05,
      "loss": 0.4014,
      "step": 4600
    },
    {
      "epoch": 0.3482143701811646,
      "grad_norm": 0.87109375,
      "learning_rate": 1.856647475948053e-05,
      "loss": 0.3518,
      "step": 4601
    },
    {
      "epoch": 0.34829005250461187,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8565860217890235e-05,
      "loss": 0.3563,
      "step": 4602
    },
    {
      "epoch": 0.3483657348280592,
      "grad_norm": 0.8515625,
      "learning_rate": 1.8565245554778516e-05,
      "loss": 0.3364,
      "step": 4603
    },
    {
      "epoch": 0.34844141715150656,
      "grad_norm": 0.83203125,
      "learning_rate": 1.856463077015409e-05,
      "loss": 0.3331,
      "step": 4604
    },
    {
      "epoch": 0.3485170994749539,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8564015864025677e-05,
      "loss": 0.3477,
      "step": 4605
    },
    {
      "epoch": 0.3485927817984012,
      "grad_norm": 0.90234375,
      "learning_rate": 1.8563400836402002e-05,
      "loss": 0.3679,
      "step": 4606
    },
    {
      "epoch": 0.34866846412184854,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8562785687291797e-05,
      "loss": 0.3024,
      "step": 4607
    },
    {
      "epoch": 0.3487441464452959,
      "grad_norm": 0.875,
      "learning_rate": 1.8562170416703775e-05,
      "loss": 0.337,
      "step": 4608
    },
    {
      "epoch": 0.3488198287687432,
      "grad_norm": 0.76953125,
      "learning_rate": 1.8561555024646676e-05,
      "loss": 0.3377,
      "step": 4609
    },
    {
      "epoch": 0.3488955110921905,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8560939511129225e-05,
      "loss": 0.348,
      "step": 4610
    },
    {
      "epoch": 0.34897119341563787,
      "grad_norm": 0.7421875,
      "learning_rate": 1.856032387616016e-05,
      "loss": 0.2983,
      "step": 4611
    },
    {
      "epoch": 0.3490468757390852,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8559708119748205e-05,
      "loss": 0.3257,
      "step": 4612
    },
    {
      "epoch": 0.3491225580625325,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8559092241902105e-05,
      "loss": 0.2958,
      "step": 4613
    },
    {
      "epoch": 0.34919824038597985,
      "grad_norm": 0.859375,
      "learning_rate": 1.8558476242630595e-05,
      "loss": 0.3617,
      "step": 4614
    },
    {
      "epoch": 0.3492739227094272,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8557860121942412e-05,
      "loss": 0.3225,
      "step": 4615
    },
    {
      "epoch": 0.3493496050328745,
      "grad_norm": 1.171875,
      "learning_rate": 1.8557243879846296e-05,
      "loss": 0.3616,
      "step": 4616
    },
    {
      "epoch": 0.34942528735632183,
      "grad_norm": 1.1796875,
      "learning_rate": 1.8556627516350988e-05,
      "loss": 0.3756,
      "step": 4617
    },
    {
      "epoch": 0.3495009696797692,
      "grad_norm": 1.171875,
      "learning_rate": 1.855601103146524e-05,
      "loss": 0.4531,
      "step": 4618
    },
    {
      "epoch": 0.3495766520032165,
      "grad_norm": 0.84765625,
      "learning_rate": 1.855539442519779e-05,
      "loss": 0.3898,
      "step": 4619
    },
    {
      "epoch": 0.3496523343266638,
      "grad_norm": 0.81640625,
      "learning_rate": 1.855477769755739e-05,
      "loss": 0.3149,
      "step": 4620
    },
    {
      "epoch": 0.34972801665011116,
      "grad_norm": 0.91796875,
      "learning_rate": 1.855416084855279e-05,
      "loss": 0.3508,
      "step": 4621
    },
    {
      "epoch": 0.3498036989735585,
      "grad_norm": 0.83203125,
      "learning_rate": 1.8553543878192733e-05,
      "loss": 0.3534,
      "step": 4622
    },
    {
      "epoch": 0.3498793812970058,
      "grad_norm": 0.87109375,
      "learning_rate": 1.8552926786485983e-05,
      "loss": 0.3762,
      "step": 4623
    },
    {
      "epoch": 0.34995506362045314,
      "grad_norm": 0.859375,
      "learning_rate": 1.8552309573441288e-05,
      "loss": 0.3138,
      "step": 4624
    },
    {
      "epoch": 0.3500307459439005,
      "grad_norm": 0.82421875,
      "learning_rate": 1.85516922390674e-05,
      "loss": 0.3064,
      "step": 4625
    },
    {
      "epoch": 0.3501064282673478,
      "grad_norm": 0.85546875,
      "learning_rate": 1.855107478337309e-05,
      "loss": 0.3638,
      "step": 4626
    },
    {
      "epoch": 0.3501821105907951,
      "grad_norm": 0.796875,
      "learning_rate": 1.8550457206367106e-05,
      "loss": 0.3321,
      "step": 4627
    },
    {
      "epoch": 0.35025779291424247,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8549839508058215e-05,
      "loss": 0.3415,
      "step": 4628
    },
    {
      "epoch": 0.3503334752376898,
      "grad_norm": 0.85546875,
      "learning_rate": 1.854922168845518e-05,
      "loss": 0.3664,
      "step": 4629
    },
    {
      "epoch": 0.3504091575611371,
      "grad_norm": 0.8515625,
      "learning_rate": 1.854860374756676e-05,
      "loss": 0.3409,
      "step": 4630
    },
    {
      "epoch": 0.35048483988458445,
      "grad_norm": 0.83984375,
      "learning_rate": 1.854798568540173e-05,
      "loss": 0.3329,
      "step": 4631
    },
    {
      "epoch": 0.3505605222080318,
      "grad_norm": 0.84765625,
      "learning_rate": 1.8547367501968853e-05,
      "loss": 0.3684,
      "step": 4632
    },
    {
      "epoch": 0.3506362045314791,
      "grad_norm": 0.765625,
      "learning_rate": 1.8546749197276898e-05,
      "loss": 0.3365,
      "step": 4633
    },
    {
      "epoch": 0.35071188685492644,
      "grad_norm": 0.78125,
      "learning_rate": 1.854613077133464e-05,
      "loss": 0.3179,
      "step": 4634
    },
    {
      "epoch": 0.3507875691783738,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8545512224150854e-05,
      "loss": 0.3363,
      "step": 4635
    },
    {
      "epoch": 0.3508632515018211,
      "grad_norm": 0.87109375,
      "learning_rate": 1.854489355573431e-05,
      "loss": 0.3808,
      "step": 4636
    },
    {
      "epoch": 0.3509389338252684,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8544274766093786e-05,
      "loss": 0.3524,
      "step": 4637
    },
    {
      "epoch": 0.35101461614871576,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8543655855238064e-05,
      "loss": 0.3335,
      "step": 4638
    },
    {
      "epoch": 0.3510902984721631,
      "grad_norm": 0.796875,
      "learning_rate": 1.8543036823175922e-05,
      "loss": 0.333,
      "step": 4639
    },
    {
      "epoch": 0.3511659807956104,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8542417669916143e-05,
      "loss": 0.3373,
      "step": 4640
    },
    {
      "epoch": 0.35124166311905775,
      "grad_norm": 0.828125,
      "learning_rate": 1.854179839546751e-05,
      "loss": 0.356,
      "step": 4641
    },
    {
      "epoch": 0.3513173454425051,
      "grad_norm": 0.71875,
      "learning_rate": 1.854117899983881e-05,
      "loss": 0.2777,
      "step": 4642
    },
    {
      "epoch": 0.35139302776595244,
      "grad_norm": 1.9140625,
      "learning_rate": 1.8540559483038825e-05,
      "loss": 0.41,
      "step": 4643
    },
    {
      "epoch": 0.3514687100893997,
      "grad_norm": 0.8359375,
      "learning_rate": 1.853993984507635e-05,
      "loss": 0.3603,
      "step": 4644
    },
    {
      "epoch": 0.3515443924128471,
      "grad_norm": 0.83203125,
      "learning_rate": 1.853932008596017e-05,
      "loss": 0.3338,
      "step": 4645
    },
    {
      "epoch": 0.3516200747362944,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8538700205699084e-05,
      "loss": 0.3422,
      "step": 4646
    },
    {
      "epoch": 0.3516957570597417,
      "grad_norm": 0.96484375,
      "learning_rate": 1.8538080204301884e-05,
      "loss": 0.2888,
      "step": 4647
    },
    {
      "epoch": 0.35177143938318906,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8537460081777362e-05,
      "loss": 0.34,
      "step": 4648
    },
    {
      "epoch": 0.3518471217066364,
      "grad_norm": 0.8125,
      "learning_rate": 1.8536839838134316e-05,
      "loss": 0.3148,
      "step": 4649
    },
    {
      "epoch": 0.35192280403008375,
      "grad_norm": 0.8828125,
      "learning_rate": 1.8536219473381553e-05,
      "loss": 0.3402,
      "step": 4650
    },
    {
      "epoch": 0.35199848635353104,
      "grad_norm": 0.83203125,
      "learning_rate": 1.8535598987527863e-05,
      "loss": 0.3414,
      "step": 4651
    },
    {
      "epoch": 0.3520741686769784,
      "grad_norm": 0.87890625,
      "learning_rate": 1.8534978380582055e-05,
      "loss": 0.4047,
      "step": 4652
    },
    {
      "epoch": 0.35214985100042573,
      "grad_norm": 0.765625,
      "learning_rate": 1.853435765255293e-05,
      "loss": 0.3116,
      "step": 4653
    },
    {
      "epoch": 0.352225533323873,
      "grad_norm": 0.7265625,
      "learning_rate": 1.8533736803449304e-05,
      "loss": 0.2721,
      "step": 4654
    },
    {
      "epoch": 0.35230121564732036,
      "grad_norm": 0.80078125,
      "learning_rate": 1.853311583327997e-05,
      "loss": 0.3633,
      "step": 4655
    },
    {
      "epoch": 0.3523768979707677,
      "grad_norm": 0.75,
      "learning_rate": 1.8532494742053747e-05,
      "loss": 0.2961,
      "step": 4656
    },
    {
      "epoch": 0.35245258029421506,
      "grad_norm": 0.828125,
      "learning_rate": 1.853187352977944e-05,
      "loss": 0.3362,
      "step": 4657
    },
    {
      "epoch": 0.35252826261766235,
      "grad_norm": 0.8125,
      "learning_rate": 1.853125219646587e-05,
      "loss": 0.3415,
      "step": 4658
    },
    {
      "epoch": 0.3526039449411097,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8530630742121846e-05,
      "loss": 0.3049,
      "step": 4659
    },
    {
      "epoch": 0.35267962726455704,
      "grad_norm": 0.94140625,
      "learning_rate": 1.8530009166756188e-05,
      "loss": 0.3202,
      "step": 4660
    },
    {
      "epoch": 0.35275530958800433,
      "grad_norm": 0.8515625,
      "learning_rate": 1.8529387470377708e-05,
      "loss": 0.3732,
      "step": 4661
    },
    {
      "epoch": 0.3528309919114517,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8528765652995235e-05,
      "loss": 0.3073,
      "step": 4662
    },
    {
      "epoch": 0.352906674234899,
      "grad_norm": 0.77734375,
      "learning_rate": 1.852814371461758e-05,
      "loss": 0.3201,
      "step": 4663
    },
    {
      "epoch": 0.35298235655834637,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8527521655253574e-05,
      "loss": 0.3423,
      "step": 4664
    },
    {
      "epoch": 0.35305803888179366,
      "grad_norm": 0.8125,
      "learning_rate": 1.8526899474912042e-05,
      "loss": 0.312,
      "step": 4665
    },
    {
      "epoch": 0.353133721205241,
      "grad_norm": 0.87109375,
      "learning_rate": 1.8526277173601804e-05,
      "loss": 0.3776,
      "step": 4666
    },
    {
      "epoch": 0.35320940352868835,
      "grad_norm": 1.15625,
      "learning_rate": 1.8525654751331694e-05,
      "loss": 0.3919,
      "step": 4667
    },
    {
      "epoch": 0.35328508585213564,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8525032208110544e-05,
      "loss": 0.311,
      "step": 4668
    },
    {
      "epoch": 0.353360768175583,
      "grad_norm": 0.765625,
      "learning_rate": 1.8524409543947177e-05,
      "loss": 0.3309,
      "step": 4669
    },
    {
      "epoch": 0.35343645049903033,
      "grad_norm": 0.78125,
      "learning_rate": 1.8523786758850436e-05,
      "loss": 0.2959,
      "step": 4670
    },
    {
      "epoch": 0.3535121328224777,
      "grad_norm": 0.7734375,
      "learning_rate": 1.852316385282915e-05,
      "loss": 0.292,
      "step": 4671
    },
    {
      "epoch": 0.35358781514592497,
      "grad_norm": 0.83984375,
      "learning_rate": 1.8522540825892164e-05,
      "loss": 0.3238,
      "step": 4672
    },
    {
      "epoch": 0.3536634974693723,
      "grad_norm": 0.71875,
      "learning_rate": 1.852191767804831e-05,
      "loss": 0.2738,
      "step": 4673
    },
    {
      "epoch": 0.35373917979281966,
      "grad_norm": 0.828125,
      "learning_rate": 1.8521294409306424e-05,
      "loss": 0.3531,
      "step": 4674
    },
    {
      "epoch": 0.35381486211626695,
      "grad_norm": 0.828125,
      "learning_rate": 1.8520671019675357e-05,
      "loss": 0.3228,
      "step": 4675
    },
    {
      "epoch": 0.3538905444397143,
      "grad_norm": 0.83203125,
      "learning_rate": 1.852004750916395e-05,
      "loss": 0.3434,
      "step": 4676
    },
    {
      "epoch": 0.35396622676316164,
      "grad_norm": 0.765625,
      "learning_rate": 1.8519423877781047e-05,
      "loss": 0.2983,
      "step": 4677
    },
    {
      "epoch": 0.354041909086609,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8518800125535497e-05,
      "loss": 0.316,
      "step": 4678
    },
    {
      "epoch": 0.3541175914100563,
      "grad_norm": 0.84765625,
      "learning_rate": 1.851817625243615e-05,
      "loss": 0.3435,
      "step": 4679
    },
    {
      "epoch": 0.3541932737335036,
      "grad_norm": 1.0625,
      "learning_rate": 1.8517552258491852e-05,
      "loss": 0.3656,
      "step": 4680
    },
    {
      "epoch": 0.35426895605695097,
      "grad_norm": 0.8203125,
      "learning_rate": 1.851692814371146e-05,
      "loss": 0.3447,
      "step": 4681
    },
    {
      "epoch": 0.35434463838039826,
      "grad_norm": 0.75,
      "learning_rate": 1.851630390810383e-05,
      "loss": 0.2777,
      "step": 4682
    },
    {
      "epoch": 0.3544203207038456,
      "grad_norm": 0.83203125,
      "learning_rate": 1.851567955167781e-05,
      "loss": 0.3264,
      "step": 4683
    },
    {
      "epoch": 0.35449600302729295,
      "grad_norm": 0.84765625,
      "learning_rate": 1.851505507444226e-05,
      "loss": 0.3799,
      "step": 4684
    },
    {
      "epoch": 0.3545716853507403,
      "grad_norm": 0.89453125,
      "learning_rate": 1.8514430476406047e-05,
      "loss": 0.3835,
      "step": 4685
    },
    {
      "epoch": 0.3546473676741876,
      "grad_norm": 0.76953125,
      "learning_rate": 1.8513805757578027e-05,
      "loss": 0.3118,
      "step": 4686
    },
    {
      "epoch": 0.35472304999763493,
      "grad_norm": 0.87109375,
      "learning_rate": 1.851318091796706e-05,
      "loss": 0.3133,
      "step": 4687
    },
    {
      "epoch": 0.3547987323210823,
      "grad_norm": 0.7265625,
      "learning_rate": 1.8512555957582012e-05,
      "loss": 0.2797,
      "step": 4688
    },
    {
      "epoch": 0.35487441464452957,
      "grad_norm": 0.83203125,
      "learning_rate": 1.851193087643175e-05,
      "loss": 0.3319,
      "step": 4689
    },
    {
      "epoch": 0.3549500969679769,
      "grad_norm": 0.7421875,
      "learning_rate": 1.851130567452514e-05,
      "loss": 0.2963,
      "step": 4690
    },
    {
      "epoch": 0.35502577929142426,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8510680351871057e-05,
      "loss": 0.3507,
      "step": 4691
    },
    {
      "epoch": 0.35510146161487155,
      "grad_norm": 0.796875,
      "learning_rate": 1.8510054908478366e-05,
      "loss": 0.3129,
      "step": 4692
    },
    {
      "epoch": 0.3551771439383189,
      "grad_norm": 1.2421875,
      "learning_rate": 1.850942934435594e-05,
      "loss": 0.4334,
      "step": 4693
    },
    {
      "epoch": 0.35525282626176624,
      "grad_norm": 0.6640625,
      "learning_rate": 1.850880365951266e-05,
      "loss": 0.2371,
      "step": 4694
    },
    {
      "epoch": 0.3553285085852136,
      "grad_norm": 1.2421875,
      "learning_rate": 1.8508177853957398e-05,
      "loss": 0.468,
      "step": 4695
    },
    {
      "epoch": 0.3554041909086609,
      "grad_norm": 0.86328125,
      "learning_rate": 1.850755192769903e-05,
      "loss": 0.3231,
      "step": 4696
    },
    {
      "epoch": 0.3554798732321082,
      "grad_norm": 0.79296875,
      "learning_rate": 1.850692588074644e-05,
      "loss": 0.314,
      "step": 4697
    },
    {
      "epoch": 0.35555555555555557,
      "grad_norm": 0.76171875,
      "learning_rate": 1.850629971310851e-05,
      "loss": 0.312,
      "step": 4698
    },
    {
      "epoch": 0.35563123787900286,
      "grad_norm": 0.85546875,
      "learning_rate": 1.850567342479412e-05,
      "loss": 0.36,
      "step": 4699
    },
    {
      "epoch": 0.3557069202024502,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8505047015812155e-05,
      "loss": 0.332,
      "step": 4700
    },
    {
      "epoch": 0.35578260252589755,
      "grad_norm": 0.921875,
      "learning_rate": 1.85044204861715e-05,
      "loss": 0.4041,
      "step": 4701
    },
    {
      "epoch": 0.3558582848493449,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8503793835881052e-05,
      "loss": 0.295,
      "step": 4702
    },
    {
      "epoch": 0.3559339671727922,
      "grad_norm": 0.828125,
      "learning_rate": 1.8503167064949693e-05,
      "loss": 0.3247,
      "step": 4703
    },
    {
      "epoch": 0.35600964949623953,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8502540173386317e-05,
      "loss": 0.3291,
      "step": 4704
    },
    {
      "epoch": 0.3560853318196869,
      "grad_norm": 0.796875,
      "learning_rate": 1.8501913161199818e-05,
      "loss": 0.3163,
      "step": 4705
    },
    {
      "epoch": 0.35616101414313417,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8501286028399094e-05,
      "loss": 0.3244,
      "step": 4706
    },
    {
      "epoch": 0.3562366964665815,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8500658774993035e-05,
      "loss": 0.3295,
      "step": 4707
    },
    {
      "epoch": 0.35631237879002886,
      "grad_norm": 0.8984375,
      "learning_rate": 1.8500031400990544e-05,
      "loss": 0.3839,
      "step": 4708
    },
    {
      "epoch": 0.3563880611134762,
      "grad_norm": 0.875,
      "learning_rate": 1.849940390640052e-05,
      "loss": 0.3844,
      "step": 4709
    },
    {
      "epoch": 0.3564637434369235,
      "grad_norm": 0.8984375,
      "learning_rate": 1.849877629123187e-05,
      "loss": 0.3851,
      "step": 4710
    },
    {
      "epoch": 0.35653942576037084,
      "grad_norm": 0.8125,
      "learning_rate": 1.849814855549349e-05,
      "loss": 0.3212,
      "step": 4711
    },
    {
      "epoch": 0.3566151080838182,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8497520699194295e-05,
      "loss": 0.3411,
      "step": 4712
    },
    {
      "epoch": 0.3566907904072655,
      "grad_norm": 0.88671875,
      "learning_rate": 1.849689272234318e-05,
      "loss": 0.4062,
      "step": 4713
    },
    {
      "epoch": 0.3567664727307128,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8496264624949063e-05,
      "loss": 0.3292,
      "step": 4714
    },
    {
      "epoch": 0.35684215505416017,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8495636407020854e-05,
      "loss": 0.342,
      "step": 4715
    },
    {
      "epoch": 0.3569178373776075,
      "grad_norm": 0.73828125,
      "learning_rate": 1.8495008068567463e-05,
      "loss": 0.3128,
      "step": 4716
    },
    {
      "epoch": 0.3569935197010548,
      "grad_norm": 0.83984375,
      "learning_rate": 1.8494379609597804e-05,
      "loss": 0.3281,
      "step": 4717
    },
    {
      "epoch": 0.35706920202450215,
      "grad_norm": 0.73828125,
      "learning_rate": 1.849375103012079e-05,
      "loss": 0.3108,
      "step": 4718
    },
    {
      "epoch": 0.3571448843479495,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8493122330145346e-05,
      "loss": 0.3004,
      "step": 4719
    },
    {
      "epoch": 0.3572205666713968,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8492493509680386e-05,
      "loss": 0.3316,
      "step": 4720
    },
    {
      "epoch": 0.35729624899484413,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8491864568734834e-05,
      "loss": 0.3491,
      "step": 4721
    },
    {
      "epoch": 0.3573719313182915,
      "grad_norm": 0.78515625,
      "learning_rate": 1.849123550731761e-05,
      "loss": 0.3381,
      "step": 4722
    },
    {
      "epoch": 0.3574476136417388,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8490606325437636e-05,
      "loss": 0.3496,
      "step": 4723
    },
    {
      "epoch": 0.3575232959651861,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8489977023103845e-05,
      "loss": 0.2832,
      "step": 4724
    },
    {
      "epoch": 0.35759897828863346,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8489347600325156e-05,
      "loss": 0.3479,
      "step": 4725
    },
    {
      "epoch": 0.3576746606120808,
      "grad_norm": 0.73828125,
      "learning_rate": 1.8488718057110503e-05,
      "loss": 0.2828,
      "step": 4726
    },
    {
      "epoch": 0.3577503429355281,
      "grad_norm": 0.85546875,
      "learning_rate": 1.848808839346882e-05,
      "loss": 0.3094,
      "step": 4727
    },
    {
      "epoch": 0.35782602525897544,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8487458609409035e-05,
      "loss": 0.3568,
      "step": 4728
    },
    {
      "epoch": 0.3579017075824228,
      "grad_norm": 0.78125,
      "learning_rate": 1.848682870494009e-05,
      "loss": 0.3111,
      "step": 4729
    },
    {
      "epoch": 0.35797738990587014,
      "grad_norm": 0.796875,
      "learning_rate": 1.848619868007091e-05,
      "loss": 0.3255,
      "step": 4730
    },
    {
      "epoch": 0.3580530722293174,
      "grad_norm": 0.7890625,
      "learning_rate": 1.848556853481044e-05,
      "loss": 0.3324,
      "step": 4731
    },
    {
      "epoch": 0.35812875455276477,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8484938269167623e-05,
      "loss": 0.3157,
      "step": 4732
    },
    {
      "epoch": 0.3582044368762121,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8484307883151392e-05,
      "loss": 0.3223,
      "step": 4733
    },
    {
      "epoch": 0.3582801191996594,
      "grad_norm": 0.71875,
      "learning_rate": 1.8483677376770695e-05,
      "loss": 0.289,
      "step": 4734
    },
    {
      "epoch": 0.35835580152310675,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8483046750034476e-05,
      "loss": 0.3411,
      "step": 4735
    },
    {
      "epoch": 0.3584314838465541,
      "grad_norm": 0.875,
      "learning_rate": 1.848241600295168e-05,
      "loss": 0.356,
      "step": 4736
    },
    {
      "epoch": 0.35850716617000145,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8481785135531257e-05,
      "loss": 0.3461,
      "step": 4737
    },
    {
      "epoch": 0.35858284849344874,
      "grad_norm": 0.74609375,
      "learning_rate": 1.848115414778216e-05,
      "loss": 0.2793,
      "step": 4738
    },
    {
      "epoch": 0.3586585308168961,
      "grad_norm": 0.765625,
      "learning_rate": 1.8480523039713333e-05,
      "loss": 0.31,
      "step": 4739
    },
    {
      "epoch": 0.3587342131403434,
      "grad_norm": 0.859375,
      "learning_rate": 1.8479891811333735e-05,
      "loss": 0.3385,
      "step": 4740
    },
    {
      "epoch": 0.3588098954637907,
      "grad_norm": 0.80859375,
      "learning_rate": 1.847926046265232e-05,
      "loss": 0.3083,
      "step": 4741
    },
    {
      "epoch": 0.35888557778723806,
      "grad_norm": 0.83203125,
      "learning_rate": 1.8478628993678043e-05,
      "loss": 0.3315,
      "step": 4742
    },
    {
      "epoch": 0.3589612601106854,
      "grad_norm": 0.86328125,
      "learning_rate": 1.8477997404419865e-05,
      "loss": 0.3282,
      "step": 4743
    },
    {
      "epoch": 0.35903694243413276,
      "grad_norm": 0.90234375,
      "learning_rate": 1.8477365694886742e-05,
      "loss": 0.3505,
      "step": 4744
    },
    {
      "epoch": 0.35911262475758005,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8476733865087647e-05,
      "loss": 0.3289,
      "step": 4745
    },
    {
      "epoch": 0.3591883070810274,
      "grad_norm": 0.7890625,
      "learning_rate": 1.847610191503153e-05,
      "loss": 0.3271,
      "step": 4746
    },
    {
      "epoch": 0.35926398940447474,
      "grad_norm": 0.8515625,
      "learning_rate": 1.8475469844727362e-05,
      "loss": 0.3282,
      "step": 4747
    },
    {
      "epoch": 0.359339671727922,
      "grad_norm": 1.5390625,
      "learning_rate": 1.847483765418411e-05,
      "loss": 0.4488,
      "step": 4748
    },
    {
      "epoch": 0.3594153540513694,
      "grad_norm": 0.76953125,
      "learning_rate": 1.8474205343410742e-05,
      "loss": 0.3247,
      "step": 4749
    },
    {
      "epoch": 0.3594910363748167,
      "grad_norm": 0.796875,
      "learning_rate": 1.8473572912416232e-05,
      "loss": 0.2889,
      "step": 4750
    },
    {
      "epoch": 0.359566718698264,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8472940361209547e-05,
      "loss": 0.3414,
      "step": 4751
    },
    {
      "epoch": 0.35964240102171136,
      "grad_norm": 0.88671875,
      "learning_rate": 1.8472307689799663e-05,
      "loss": 0.3728,
      "step": 4752
    },
    {
      "epoch": 0.3597180833451587,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8471674898195554e-05,
      "loss": 0.3277,
      "step": 4753
    },
    {
      "epoch": 0.35979376566860605,
      "grad_norm": 0.703125,
      "learning_rate": 1.84710419864062e-05,
      "loss": 0.2761,
      "step": 4754
    },
    {
      "epoch": 0.35986944799205334,
      "grad_norm": 2.03125,
      "learning_rate": 1.847040895444058e-05,
      "loss": 0.4875,
      "step": 4755
    },
    {
      "epoch": 0.3599451303155007,
      "grad_norm": 0.93359375,
      "learning_rate": 1.846977580230767e-05,
      "loss": 0.3828,
      "step": 4756
    },
    {
      "epoch": 0.36002081263894803,
      "grad_norm": 0.8984375,
      "learning_rate": 1.846914253001646e-05,
      "loss": 0.3751,
      "step": 4757
    },
    {
      "epoch": 0.3600964949623953,
      "grad_norm": 0.76171875,
      "learning_rate": 1.8468509137575924e-05,
      "loss": 0.3005,
      "step": 4758
    },
    {
      "epoch": 0.36017217728584267,
      "grad_norm": 0.84765625,
      "learning_rate": 1.846787562499506e-05,
      "loss": 0.3363,
      "step": 4759
    },
    {
      "epoch": 0.36024785960929,
      "grad_norm": 0.87109375,
      "learning_rate": 1.8467241992282842e-05,
      "loss": 0.3636,
      "step": 4760
    },
    {
      "epoch": 0.36032354193273736,
      "grad_norm": 0.90625,
      "learning_rate": 1.846660823944827e-05,
      "loss": 0.3929,
      "step": 4761
    },
    {
      "epoch": 0.36039922425618465,
      "grad_norm": 0.73828125,
      "learning_rate": 1.8465974366500332e-05,
      "loss": 0.3014,
      "step": 4762
    },
    {
      "epoch": 0.360474906579632,
      "grad_norm": 0.796875,
      "learning_rate": 1.8465340373448018e-05,
      "loss": 0.3363,
      "step": 4763
    },
    {
      "epoch": 0.36055058890307934,
      "grad_norm": 0.8515625,
      "learning_rate": 1.8464706260300322e-05,
      "loss": 0.3358,
      "step": 4764
    },
    {
      "epoch": 0.36062627122652663,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8464072027066245e-05,
      "loss": 0.317,
      "step": 4765
    },
    {
      "epoch": 0.360701953549974,
      "grad_norm": 0.83984375,
      "learning_rate": 1.846343767375478e-05,
      "loss": 0.3625,
      "step": 4766
    },
    {
      "epoch": 0.3607776358734213,
      "grad_norm": 1.6328125,
      "learning_rate": 1.8462803200374928e-05,
      "loss": 0.3942,
      "step": 4767
    },
    {
      "epoch": 0.36085331819686867,
      "grad_norm": 0.75,
      "learning_rate": 1.846216860693569e-05,
      "loss": 0.2813,
      "step": 4768
    },
    {
      "epoch": 0.36092900052031596,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8461533893446066e-05,
      "loss": 0.3019,
      "step": 4769
    },
    {
      "epoch": 0.3610046828437633,
      "grad_norm": 0.98046875,
      "learning_rate": 1.8460899059915067e-05,
      "loss": 0.3776,
      "step": 4770
    },
    {
      "epoch": 0.36108036516721065,
      "grad_norm": 0.8828125,
      "learning_rate": 1.8460264106351693e-05,
      "loss": 0.3427,
      "step": 4771
    },
    {
      "epoch": 0.36115604749065794,
      "grad_norm": 0.75390625,
      "learning_rate": 1.8459629032764957e-05,
      "loss": 0.3103,
      "step": 4772
    },
    {
      "epoch": 0.3612317298141053,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8458993839163867e-05,
      "loss": 0.3568,
      "step": 4773
    },
    {
      "epoch": 0.36130741213755263,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8458358525557426e-05,
      "loss": 0.3087,
      "step": 4774
    },
    {
      "epoch": 0.361383094461,
      "grad_norm": 0.92578125,
      "learning_rate": 1.845772309195466e-05,
      "loss": 0.3834,
      "step": 4775
    },
    {
      "epoch": 0.36145877678444727,
      "grad_norm": 0.84765625,
      "learning_rate": 1.8457087538364578e-05,
      "loss": 0.3571,
      "step": 4776
    },
    {
      "epoch": 0.3615344591078946,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8456451864796195e-05,
      "loss": 0.3266,
      "step": 4777
    },
    {
      "epoch": 0.36161014143134196,
      "grad_norm": 0.79296875,
      "learning_rate": 1.845581607125853e-05,
      "loss": 0.3148,
      "step": 4778
    },
    {
      "epoch": 0.36168582375478925,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8455180157760602e-05,
      "loss": 0.3302,
      "step": 4779
    },
    {
      "epoch": 0.3617615060782366,
      "grad_norm": 0.84375,
      "learning_rate": 1.8454544124311438e-05,
      "loss": 0.3726,
      "step": 4780
    },
    {
      "epoch": 0.36183718840168394,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8453907970920056e-05,
      "loss": 0.3192,
      "step": 4781
    },
    {
      "epoch": 0.3619128707251313,
      "grad_norm": 0.76171875,
      "learning_rate": 1.845327169759548e-05,
      "loss": 0.3327,
      "step": 4782
    },
    {
      "epoch": 0.3619885530485786,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8452635304346743e-05,
      "loss": 0.37,
      "step": 4783
    },
    {
      "epoch": 0.3620642353720259,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8451998791182864e-05,
      "loss": 0.3255,
      "step": 4784
    },
    {
      "epoch": 0.36213991769547327,
      "grad_norm": 0.83984375,
      "learning_rate": 1.845136215811288e-05,
      "loss": 0.3346,
      "step": 4785
    },
    {
      "epoch": 0.36221560001892056,
      "grad_norm": 0.828125,
      "learning_rate": 1.845072540514582e-05,
      "loss": 0.3264,
      "step": 4786
    },
    {
      "epoch": 0.3622912823423679,
      "grad_norm": 0.8828125,
      "learning_rate": 1.845008853229072e-05,
      "loss": 0.3857,
      "step": 4787
    },
    {
      "epoch": 0.36236696466581525,
      "grad_norm": 1.4296875,
      "learning_rate": 1.8449451539556612e-05,
      "loss": 0.4404,
      "step": 4788
    },
    {
      "epoch": 0.3624426469892626,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8448814426952534e-05,
      "loss": 0.3053,
      "step": 4789
    },
    {
      "epoch": 0.3625183293127099,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8448177194487524e-05,
      "loss": 0.3445,
      "step": 4790
    },
    {
      "epoch": 0.36259401163615723,
      "grad_norm": 0.74609375,
      "learning_rate": 1.8447539842170625e-05,
      "loss": 0.2807,
      "step": 4791
    },
    {
      "epoch": 0.3626696939596046,
      "grad_norm": 1.1328125,
      "learning_rate": 1.8446902370010876e-05,
      "loss": 0.3736,
      "step": 4792
    },
    {
      "epoch": 0.36274537628305187,
      "grad_norm": 0.80078125,
      "learning_rate": 1.844626477801732e-05,
      "loss": 0.3003,
      "step": 4793
    },
    {
      "epoch": 0.3628210586064992,
      "grad_norm": 1.0859375,
      "learning_rate": 1.8445627066199005e-05,
      "loss": 0.3907,
      "step": 4794
    },
    {
      "epoch": 0.36289674092994656,
      "grad_norm": 0.87890625,
      "learning_rate": 1.844498923456498e-05,
      "loss": 0.3722,
      "step": 4795
    },
    {
      "epoch": 0.3629724232533939,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8444351283124288e-05,
      "loss": 0.2967,
      "step": 4796
    },
    {
      "epoch": 0.3630481055768412,
      "grad_norm": 0.83203125,
      "learning_rate": 1.844371321188598e-05,
      "loss": 0.342,
      "step": 4797
    },
    {
      "epoch": 0.36312378790028854,
      "grad_norm": 0.83984375,
      "learning_rate": 1.844307502085911e-05,
      "loss": 0.3536,
      "step": 4798
    },
    {
      "epoch": 0.3631994702237359,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8442436710052734e-05,
      "loss": 0.344,
      "step": 4799
    },
    {
      "epoch": 0.3632751525471832,
      "grad_norm": 0.75,
      "learning_rate": 1.8441798279475907e-05,
      "loss": 0.2777,
      "step": 4800
    },
    {
      "epoch": 0.3633508348706305,
      "grad_norm": 0.85546875,
      "learning_rate": 1.8441159729137684e-05,
      "loss": 0.3295,
      "step": 4801
    },
    {
      "epoch": 0.36342651719407787,
      "grad_norm": 1.1953125,
      "learning_rate": 1.8440521059047122e-05,
      "loss": 0.3383,
      "step": 4802
    },
    {
      "epoch": 0.3635021995175252,
      "grad_norm": 0.70703125,
      "learning_rate": 1.8439882269213285e-05,
      "loss": 0.2634,
      "step": 4803
    },
    {
      "epoch": 0.3635778818409725,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8439243359645238e-05,
      "loss": 0.3403,
      "step": 4804
    },
    {
      "epoch": 0.36365356416441985,
      "grad_norm": 0.84375,
      "learning_rate": 1.8438604330352036e-05,
      "loss": 0.3333,
      "step": 4805
    },
    {
      "epoch": 0.3637292464878672,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8437965181342757e-05,
      "loss": 0.359,
      "step": 4806
    },
    {
      "epoch": 0.3638049288113145,
      "grad_norm": 0.8515625,
      "learning_rate": 1.843732591262646e-05,
      "loss": 0.3645,
      "step": 4807
    },
    {
      "epoch": 0.36388061113476183,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8436686524212212e-05,
      "loss": 0.3371,
      "step": 4808
    },
    {
      "epoch": 0.3639562934582092,
      "grad_norm": 0.90234375,
      "learning_rate": 1.8436047016109092e-05,
      "loss": 0.3737,
      "step": 4809
    },
    {
      "epoch": 0.36403197578165647,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8435407388326167e-05,
      "loss": 0.348,
      "step": 4810
    },
    {
      "epoch": 0.3641076581051038,
      "grad_norm": 0.85546875,
      "learning_rate": 1.8434767640872512e-05,
      "loss": 0.3777,
      "step": 4811
    },
    {
      "epoch": 0.36418334042855116,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8434127773757205e-05,
      "loss": 0.3262,
      "step": 4812
    },
    {
      "epoch": 0.3642590227519985,
      "grad_norm": 0.859375,
      "learning_rate": 1.843348778698932e-05,
      "loss": 0.3584,
      "step": 4813
    },
    {
      "epoch": 0.3643347050754458,
      "grad_norm": 0.7734375,
      "learning_rate": 1.843284768057794e-05,
      "loss": 0.3394,
      "step": 4814
    },
    {
      "epoch": 0.36441038739889314,
      "grad_norm": 1.171875,
      "learning_rate": 1.8432207454532146e-05,
      "loss": 0.368,
      "step": 4815
    },
    {
      "epoch": 0.3644860697223405,
      "grad_norm": 0.76953125,
      "learning_rate": 1.8431567108861014e-05,
      "loss": 0.3233,
      "step": 4816
    },
    {
      "epoch": 0.3645617520457878,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8430926643573635e-05,
      "loss": 0.3053,
      "step": 4817
    },
    {
      "epoch": 0.3646374343692351,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8430286058679096e-05,
      "loss": 0.3316,
      "step": 4818
    },
    {
      "epoch": 0.36471311669268247,
      "grad_norm": 0.81640625,
      "learning_rate": 1.842964535418648e-05,
      "loss": 0.3076,
      "step": 4819
    },
    {
      "epoch": 0.3647887990161298,
      "grad_norm": 1.046875,
      "learning_rate": 1.842900453010488e-05,
      "loss": 0.3722,
      "step": 4820
    },
    {
      "epoch": 0.3648644813395771,
      "grad_norm": 0.79296875,
      "learning_rate": 1.842836358644339e-05,
      "loss": 0.323,
      "step": 4821
    },
    {
      "epoch": 0.36494016366302445,
      "grad_norm": 0.76171875,
      "learning_rate": 1.8427722523211096e-05,
      "loss": 0.2938,
      "step": 4822
    },
    {
      "epoch": 0.3650158459864718,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8427081340417093e-05,
      "loss": 0.3449,
      "step": 4823
    },
    {
      "epoch": 0.3650915283099191,
      "grad_norm": 0.81640625,
      "learning_rate": 1.842644003807048e-05,
      "loss": 0.3582,
      "step": 4824
    },
    {
      "epoch": 0.36516721063336643,
      "grad_norm": 0.72265625,
      "learning_rate": 1.842579861618036e-05,
      "loss": 0.2902,
      "step": 4825
    },
    {
      "epoch": 0.3652428929568138,
      "grad_norm": 1.125,
      "learning_rate": 1.842515707475582e-05,
      "loss": 0.3633,
      "step": 4826
    },
    {
      "epoch": 0.3653185752802611,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8424515413805973e-05,
      "loss": 0.3037,
      "step": 4827
    },
    {
      "epoch": 0.3653942576037084,
      "grad_norm": 1.15625,
      "learning_rate": 1.842387363333992e-05,
      "loss": 0.3487,
      "step": 4828
    },
    {
      "epoch": 0.36546993992715576,
      "grad_norm": 0.78125,
      "learning_rate": 1.8423231733366762e-05,
      "loss": 0.3054,
      "step": 4829
    },
    {
      "epoch": 0.3655456222506031,
      "grad_norm": 0.73046875,
      "learning_rate": 1.8422589713895606e-05,
      "loss": 0.3005,
      "step": 4830
    },
    {
      "epoch": 0.3656213045740504,
      "grad_norm": 0.828125,
      "learning_rate": 1.8421947574935562e-05,
      "loss": 0.3265,
      "step": 4831
    },
    {
      "epoch": 0.36569698689749774,
      "grad_norm": 0.765625,
      "learning_rate": 1.8421305316495737e-05,
      "loss": 0.3004,
      "step": 4832
    },
    {
      "epoch": 0.3657726692209451,
      "grad_norm": 1.21875,
      "learning_rate": 1.8420662938585246e-05,
      "loss": 0.3896,
      "step": 4833
    },
    {
      "epoch": 0.36584835154439244,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8420020441213202e-05,
      "loss": 0.3219,
      "step": 4834
    },
    {
      "epoch": 0.3659240338678397,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8419377824388716e-05,
      "loss": 0.3596,
      "step": 4835
    },
    {
      "epoch": 0.36599971619128707,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8418735088120913e-05,
      "loss": 0.3593,
      "step": 4836
    },
    {
      "epoch": 0.3660753985147344,
      "grad_norm": 0.8515625,
      "learning_rate": 1.84180922324189e-05,
      "loss": 0.3751,
      "step": 4837
    },
    {
      "epoch": 0.3661510808381817,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8417449257291802e-05,
      "loss": 0.3507,
      "step": 4838
    },
    {
      "epoch": 0.36622676316162905,
      "grad_norm": 0.83984375,
      "learning_rate": 1.8416806162748747e-05,
      "loss": 0.362,
      "step": 4839
    },
    {
      "epoch": 0.3663024454850764,
      "grad_norm": 0.7734375,
      "learning_rate": 1.841616294879885e-05,
      "loss": 0.318,
      "step": 4840
    },
    {
      "epoch": 0.36637812780852375,
      "grad_norm": 0.87890625,
      "learning_rate": 1.841551961545124e-05,
      "loss": 0.4275,
      "step": 4841
    },
    {
      "epoch": 0.36645381013197104,
      "grad_norm": 0.90625,
      "learning_rate": 1.841487616271504e-05,
      "loss": 0.3838,
      "step": 4842
    },
    {
      "epoch": 0.3665294924554184,
      "grad_norm": 0.8125,
      "learning_rate": 1.8414232590599382e-05,
      "loss": 0.3343,
      "step": 4843
    },
    {
      "epoch": 0.36660517477886573,
      "grad_norm": 0.859375,
      "learning_rate": 1.8413588899113398e-05,
      "loss": 0.3721,
      "step": 4844
    },
    {
      "epoch": 0.366680857102313,
      "grad_norm": 0.8515625,
      "learning_rate": 1.8412945088266218e-05,
      "loss": 0.3699,
      "step": 4845
    },
    {
      "epoch": 0.36675653942576036,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8412301158066972e-05,
      "loss": 0.259,
      "step": 4846
    },
    {
      "epoch": 0.3668322217492077,
      "grad_norm": 0.85546875,
      "learning_rate": 1.84116571085248e-05,
      "loss": 0.359,
      "step": 4847
    },
    {
      "epoch": 0.36690790407265506,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8411012939648834e-05,
      "loss": 0.3338,
      "step": 4848
    },
    {
      "epoch": 0.36698358639610235,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8410368651448217e-05,
      "loss": 0.3825,
      "step": 4849
    },
    {
      "epoch": 0.3670592687195497,
      "grad_norm": 0.83203125,
      "learning_rate": 1.840972424393209e-05,
      "loss": 0.3344,
      "step": 4850
    },
    {
      "epoch": 0.36713495104299704,
      "grad_norm": 0.75390625,
      "learning_rate": 1.8409079717109588e-05,
      "loss": 0.3071,
      "step": 4851
    },
    {
      "epoch": 0.36721063336644433,
      "grad_norm": 0.890625,
      "learning_rate": 1.8408435070989866e-05,
      "loss": 0.4057,
      "step": 4852
    },
    {
      "epoch": 0.3672863156898917,
      "grad_norm": 0.80078125,
      "learning_rate": 1.840779030558206e-05,
      "loss": 0.3369,
      "step": 4853
    },
    {
      "epoch": 0.367361998013339,
      "grad_norm": 0.85546875,
      "learning_rate": 1.840714542089532e-05,
      "loss": 0.3395,
      "step": 4854
    },
    {
      "epoch": 0.36743768033678637,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8406500416938793e-05,
      "loss": 0.2888,
      "step": 4855
    },
    {
      "epoch": 0.36751336266023366,
      "grad_norm": 0.796875,
      "learning_rate": 1.8405855293721634e-05,
      "loss": 0.3195,
      "step": 4856
    },
    {
      "epoch": 0.367589044983681,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8405210051252993e-05,
      "loss": 0.3219,
      "step": 4857
    },
    {
      "epoch": 0.36766472730712835,
      "grad_norm": 0.859375,
      "learning_rate": 1.8404564689542022e-05,
      "loss": 0.397,
      "step": 4858
    },
    {
      "epoch": 0.36774040963057564,
      "grad_norm": 0.7734375,
      "learning_rate": 1.840391920859788e-05,
      "loss": 0.2756,
      "step": 4859
    },
    {
      "epoch": 0.367816091954023,
      "grad_norm": 0.734375,
      "learning_rate": 1.8403273608429722e-05,
      "loss": 0.2844,
      "step": 4860
    },
    {
      "epoch": 0.36789177427747033,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8402627889046706e-05,
      "loss": 0.3439,
      "step": 4861
    },
    {
      "epoch": 0.3679674566009177,
      "grad_norm": 0.8515625,
      "learning_rate": 1.8401982050457995e-05,
      "loss": 0.3787,
      "step": 4862
    },
    {
      "epoch": 0.36804313892436497,
      "grad_norm": 0.83984375,
      "learning_rate": 1.840133609267275e-05,
      "loss": 0.3785,
      "step": 4863
    },
    {
      "epoch": 0.3681188212478123,
      "grad_norm": 0.88671875,
      "learning_rate": 1.840069001570014e-05,
      "loss": 0.3466,
      "step": 4864
    },
    {
      "epoch": 0.36819450357125966,
      "grad_norm": 0.8203125,
      "learning_rate": 1.840004381954932e-05,
      "loss": 0.3269,
      "step": 4865
    },
    {
      "epoch": 0.36827018589470695,
      "grad_norm": 0.7265625,
      "learning_rate": 1.8399397504229464e-05,
      "loss": 0.2617,
      "step": 4866
    },
    {
      "epoch": 0.3683458682181543,
      "grad_norm": 0.796875,
      "learning_rate": 1.839875106974974e-05,
      "loss": 0.3129,
      "step": 4867
    },
    {
      "epoch": 0.36842155054160164,
      "grad_norm": 2.28125,
      "learning_rate": 1.8398104516119323e-05,
      "loss": 0.3996,
      "step": 4868
    },
    {
      "epoch": 0.36849723286504893,
      "grad_norm": 0.8515625,
      "learning_rate": 1.8397457843347382e-05,
      "loss": 0.3413,
      "step": 4869
    },
    {
      "epoch": 0.3685729151884963,
      "grad_norm": 0.796875,
      "learning_rate": 1.8396811051443086e-05,
      "loss": 0.3541,
      "step": 4870
    },
    {
      "epoch": 0.3686485975119436,
      "grad_norm": 0.78515625,
      "learning_rate": 1.839616414041562e-05,
      "loss": 0.3128,
      "step": 4871
    },
    {
      "epoch": 0.36872427983539097,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8395517110274156e-05,
      "loss": 0.3036,
      "step": 4872
    },
    {
      "epoch": 0.36879996215883826,
      "grad_norm": 0.86328125,
      "learning_rate": 1.8394869961027875e-05,
      "loss": 0.3324,
      "step": 4873
    },
    {
      "epoch": 0.3688756444822856,
      "grad_norm": 0.765625,
      "learning_rate": 1.8394222692685958e-05,
      "loss": 0.3098,
      "step": 4874
    },
    {
      "epoch": 0.36895132680573295,
      "grad_norm": 0.796875,
      "learning_rate": 1.8393575305257587e-05,
      "loss": 0.3485,
      "step": 4875
    },
    {
      "epoch": 0.36902700912918024,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8392927798751946e-05,
      "loss": 0.3065,
      "step": 4876
    },
    {
      "epoch": 0.3691026914526276,
      "grad_norm": 0.80078125,
      "learning_rate": 1.839228017317822e-05,
      "loss": 0.3319,
      "step": 4877
    },
    {
      "epoch": 0.36917837377607493,
      "grad_norm": 0.8515625,
      "learning_rate": 1.83916324285456e-05,
      "loss": 0.3631,
      "step": 4878
    },
    {
      "epoch": 0.3692540560995223,
      "grad_norm": 0.859375,
      "learning_rate": 1.839098456486327e-05,
      "loss": 0.3808,
      "step": 4879
    },
    {
      "epoch": 0.36932973842296957,
      "grad_norm": 0.80859375,
      "learning_rate": 1.839033658214043e-05,
      "loss": 0.3216,
      "step": 4880
    },
    {
      "epoch": 0.3694054207464169,
      "grad_norm": 0.796875,
      "learning_rate": 1.8389688480386264e-05,
      "loss": 0.3348,
      "step": 4881
    },
    {
      "epoch": 0.36948110306986426,
      "grad_norm": 0.86328125,
      "learning_rate": 1.838904025960997e-05,
      "loss": 0.3708,
      "step": 4882
    },
    {
      "epoch": 0.36955678539331155,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8388391919820745e-05,
      "loss": 0.341,
      "step": 4883
    },
    {
      "epoch": 0.3696324677167589,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8387743461027786e-05,
      "loss": 0.3364,
      "step": 4884
    },
    {
      "epoch": 0.36970815004020624,
      "grad_norm": 1.296875,
      "learning_rate": 1.8387094883240288e-05,
      "loss": 0.4074,
      "step": 4885
    },
    {
      "epoch": 0.3697838323636536,
      "grad_norm": 0.8046875,
      "learning_rate": 1.838644618646746e-05,
      "loss": 0.3479,
      "step": 4886
    },
    {
      "epoch": 0.3698595146871009,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8385797370718502e-05,
      "loss": 0.345,
      "step": 4887
    },
    {
      "epoch": 0.3699351970105482,
      "grad_norm": 0.8828125,
      "learning_rate": 1.8385148436002613e-05,
      "loss": 0.3521,
      "step": 4888
    },
    {
      "epoch": 0.37001087933399557,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8384499382329006e-05,
      "loss": 0.3131,
      "step": 4889
    },
    {
      "epoch": 0.37008656165744286,
      "grad_norm": 0.8203125,
      "learning_rate": 1.838385020970689e-05,
      "loss": 0.3066,
      "step": 4890
    },
    {
      "epoch": 0.3701622439808902,
      "grad_norm": 0.7890625,
      "learning_rate": 1.838320091814547e-05,
      "loss": 0.2936,
      "step": 4891
    },
    {
      "epoch": 0.37023792630433755,
      "grad_norm": 0.73046875,
      "learning_rate": 1.838255150765396e-05,
      "loss": 0.2916,
      "step": 4892
    },
    {
      "epoch": 0.3703136086277849,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8381901978241566e-05,
      "loss": 0.3267,
      "step": 4893
    },
    {
      "epoch": 0.3703892909512322,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8381252329917515e-05,
      "loss": 0.3137,
      "step": 4894
    },
    {
      "epoch": 0.37046497327467953,
      "grad_norm": 0.8125,
      "learning_rate": 1.8380602562691018e-05,
      "loss": 0.3369,
      "step": 4895
    },
    {
      "epoch": 0.3705406555981269,
      "grad_norm": 0.8359375,
      "learning_rate": 1.837995267657129e-05,
      "loss": 0.3192,
      "step": 4896
    },
    {
      "epoch": 0.37061633792157417,
      "grad_norm": 0.80859375,
      "learning_rate": 1.837930267156755e-05,
      "loss": 0.3636,
      "step": 4897
    },
    {
      "epoch": 0.3706920202450215,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8378652547689023e-05,
      "loss": 0.326,
      "step": 4898
    },
    {
      "epoch": 0.37076770256846886,
      "grad_norm": 0.86328125,
      "learning_rate": 1.8378002304944934e-05,
      "loss": 0.3613,
      "step": 4899
    },
    {
      "epoch": 0.3708433848919162,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8377351943344504e-05,
      "loss": 0.3264,
      "step": 4900
    },
    {
      "epoch": 0.3709190672153635,
      "grad_norm": 0.78125,
      "learning_rate": 1.837670146289696e-05,
      "loss": 0.3026,
      "step": 4901
    },
    {
      "epoch": 0.37099474953881084,
      "grad_norm": 0.9453125,
      "learning_rate": 1.837605086361153e-05,
      "loss": 0.3424,
      "step": 4902
    },
    {
      "epoch": 0.3710704318622582,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8375400145497445e-05,
      "loss": 0.3124,
      "step": 4903
    },
    {
      "epoch": 0.3711461141857055,
      "grad_norm": 0.796875,
      "learning_rate": 1.8374749308563938e-05,
      "loss": 0.3276,
      "step": 4904
    },
    {
      "epoch": 0.3712217965091528,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8374098352820236e-05,
      "loss": 0.3278,
      "step": 4905
    },
    {
      "epoch": 0.37129747883260017,
      "grad_norm": 0.7421875,
      "learning_rate": 1.8373447278275586e-05,
      "loss": 0.3021,
      "step": 4906
    },
    {
      "epoch": 0.3713731611560475,
      "grad_norm": 0.84375,
      "learning_rate": 1.8372796084939214e-05,
      "loss": 0.3711,
      "step": 4907
    },
    {
      "epoch": 0.3714488434794948,
      "grad_norm": 0.87890625,
      "learning_rate": 1.837214477282036e-05,
      "loss": 0.3799,
      "step": 4908
    },
    {
      "epoch": 0.37152452580294215,
      "grad_norm": 0.78125,
      "learning_rate": 1.8371493341928267e-05,
      "loss": 0.3283,
      "step": 4909
    },
    {
      "epoch": 0.3716002081263895,
      "grad_norm": 0.84765625,
      "learning_rate": 1.837084179227217e-05,
      "loss": 0.374,
      "step": 4910
    },
    {
      "epoch": 0.3716758904498368,
      "grad_norm": 0.765625,
      "learning_rate": 1.837019012386132e-05,
      "loss": 0.2898,
      "step": 4911
    },
    {
      "epoch": 0.37175157277328413,
      "grad_norm": 0.78125,
      "learning_rate": 1.8369538336704963e-05,
      "loss": 0.3146,
      "step": 4912
    },
    {
      "epoch": 0.3718272550967315,
      "grad_norm": 0.8046875,
      "learning_rate": 1.836888643081234e-05,
      "loss": 0.3253,
      "step": 4913
    },
    {
      "epoch": 0.3719029374201788,
      "grad_norm": 0.80078125,
      "learning_rate": 1.83682344061927e-05,
      "loss": 0.3356,
      "step": 4914
    },
    {
      "epoch": 0.3719786197436261,
      "grad_norm": 0.8125,
      "learning_rate": 1.8367582262855297e-05,
      "loss": 0.3465,
      "step": 4915
    },
    {
      "epoch": 0.37205430206707346,
      "grad_norm": 0.85546875,
      "learning_rate": 1.8366930000809378e-05,
      "loss": 0.3606,
      "step": 4916
    },
    {
      "epoch": 0.3721299843905208,
      "grad_norm": 0.77734375,
      "learning_rate": 1.83662776200642e-05,
      "loss": 0.3271,
      "step": 4917
    },
    {
      "epoch": 0.3722056667139681,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8365625120629016e-05,
      "loss": 0.3187,
      "step": 4918
    },
    {
      "epoch": 0.37228134903741544,
      "grad_norm": 0.87890625,
      "learning_rate": 1.836497250251308e-05,
      "loss": 0.3288,
      "step": 4919
    },
    {
      "epoch": 0.3723570313608628,
      "grad_norm": 1.0859375,
      "learning_rate": 1.8364319765725663e-05,
      "loss": 0.3359,
      "step": 4920
    },
    {
      "epoch": 0.37243271368431013,
      "grad_norm": 0.703125,
      "learning_rate": 1.836366691027601e-05,
      "loss": 0.2824,
      "step": 4921
    },
    {
      "epoch": 0.3725083960077574,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8363013936173393e-05,
      "loss": 0.328,
      "step": 4922
    },
    {
      "epoch": 0.37258407833120477,
      "grad_norm": 1.0625,
      "learning_rate": 1.836236084342707e-05,
      "loss": 0.4022,
      "step": 4923
    },
    {
      "epoch": 0.3726597606546521,
      "grad_norm": 0.75390625,
      "learning_rate": 1.836170763204631e-05,
      "loss": 0.3114,
      "step": 4924
    },
    {
      "epoch": 0.3727354429780994,
      "grad_norm": 0.8671875,
      "learning_rate": 1.8361054302040372e-05,
      "loss": 0.3228,
      "step": 4925
    },
    {
      "epoch": 0.37281112530154675,
      "grad_norm": 0.78125,
      "learning_rate": 1.836040085341854e-05,
      "loss": 0.3152,
      "step": 4926
    },
    {
      "epoch": 0.3728868076249941,
      "grad_norm": 0.7421875,
      "learning_rate": 1.835974728619007e-05,
      "loss": 0.2994,
      "step": 4927
    },
    {
      "epoch": 0.37296248994844144,
      "grad_norm": 0.796875,
      "learning_rate": 1.8359093600364237e-05,
      "loss": 0.3216,
      "step": 4928
    },
    {
      "epoch": 0.37303817227188873,
      "grad_norm": 0.7890625,
      "learning_rate": 1.835843979595032e-05,
      "loss": 0.3041,
      "step": 4929
    },
    {
      "epoch": 0.3731138545953361,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8357785872957588e-05,
      "loss": 0.3239,
      "step": 4930
    },
    {
      "epoch": 0.3731895369187834,
      "grad_norm": 0.84765625,
      "learning_rate": 1.8357131831395322e-05,
      "loss": 0.3919,
      "step": 4931
    },
    {
      "epoch": 0.3732652192422307,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8356477671272797e-05,
      "loss": 0.2948,
      "step": 4932
    },
    {
      "epoch": 0.37334090156567806,
      "grad_norm": 0.7890625,
      "learning_rate": 1.83558233925993e-05,
      "loss": 0.3174,
      "step": 4933
    },
    {
      "epoch": 0.3734165838891254,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8355168995384113e-05,
      "loss": 0.3327,
      "step": 4934
    },
    {
      "epoch": 0.3734922662125727,
      "grad_norm": 1.1796875,
      "learning_rate": 1.8354514479636506e-05,
      "loss": 0.3823,
      "step": 4935
    },
    {
      "epoch": 0.37356794853602004,
      "grad_norm": 1.1171875,
      "learning_rate": 1.835385984536578e-05,
      "loss": 0.3702,
      "step": 4936
    },
    {
      "epoch": 0.3736436308594674,
      "grad_norm": 0.88671875,
      "learning_rate": 1.8353205092581216e-05,
      "loss": 0.3783,
      "step": 4937
    },
    {
      "epoch": 0.37371931318291474,
      "grad_norm": 0.7421875,
      "learning_rate": 1.8352550221292102e-05,
      "loss": 0.2857,
      "step": 4938
    },
    {
      "epoch": 0.373794995506362,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8351895231507734e-05,
      "loss": 0.3359,
      "step": 4939
    },
    {
      "epoch": 0.3738706778298094,
      "grad_norm": 0.828125,
      "learning_rate": 1.8351240123237396e-05,
      "loss": 0.3296,
      "step": 4940
    },
    {
      "epoch": 0.3739463601532567,
      "grad_norm": 0.76171875,
      "learning_rate": 1.8350584896490384e-05,
      "loss": 0.3086,
      "step": 4941
    },
    {
      "epoch": 0.374022042476704,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8349929551275996e-05,
      "loss": 0.3239,
      "step": 4942
    },
    {
      "epoch": 0.37409772480015135,
      "grad_norm": 0.73828125,
      "learning_rate": 1.834927408760353e-05,
      "loss": 0.277,
      "step": 4943
    },
    {
      "epoch": 0.3741734071235987,
      "grad_norm": 0.81640625,
      "learning_rate": 1.834861850548228e-05,
      "loss": 0.3457,
      "step": 4944
    },
    {
      "epoch": 0.37424908944704605,
      "grad_norm": 0.8671875,
      "learning_rate": 1.8347962804921556e-05,
      "loss": 0.3621,
      "step": 4945
    },
    {
      "epoch": 0.37432477177049334,
      "grad_norm": 0.79296875,
      "learning_rate": 1.834730698593065e-05,
      "loss": 0.3256,
      "step": 4946
    },
    {
      "epoch": 0.3744004540939407,
      "grad_norm": 0.79296875,
      "learning_rate": 1.834665104851887e-05,
      "loss": 0.3411,
      "step": 4947
    },
    {
      "epoch": 0.37447613641738803,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8345994992695523e-05,
      "loss": 0.3416,
      "step": 4948
    },
    {
      "epoch": 0.3745518187408353,
      "grad_norm": 0.84765625,
      "learning_rate": 1.834533881846991e-05,
      "loss": 0.3549,
      "step": 4949
    },
    {
      "epoch": 0.37462750106428266,
      "grad_norm": 0.8203125,
      "learning_rate": 1.834468252585135e-05,
      "loss": 0.3447,
      "step": 4950
    },
    {
      "epoch": 0.37470318338773,
      "grad_norm": 0.8671875,
      "learning_rate": 1.8344026114849147e-05,
      "loss": 0.3688,
      "step": 4951
    },
    {
      "epoch": 0.37477886571117736,
      "grad_norm": 0.83203125,
      "learning_rate": 1.8343369585472615e-05,
      "loss": 0.3316,
      "step": 4952
    },
    {
      "epoch": 0.37485454803462465,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8342712937731066e-05,
      "loss": 0.308,
      "step": 4953
    },
    {
      "epoch": 0.374930230358072,
      "grad_norm": 0.765625,
      "learning_rate": 1.834205617163382e-05,
      "loss": 0.3031,
      "step": 4954
    },
    {
      "epoch": 0.37500591268151934,
      "grad_norm": 0.79296875,
      "learning_rate": 1.834139928719019e-05,
      "loss": 0.3379,
      "step": 4955
    },
    {
      "epoch": 0.37508159500496663,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8340742284409496e-05,
      "loss": 0.3298,
      "step": 4956
    },
    {
      "epoch": 0.375157277328414,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8340085163301064e-05,
      "loss": 0.3096,
      "step": 4957
    },
    {
      "epoch": 0.3752329596518613,
      "grad_norm": 0.796875,
      "learning_rate": 1.8339427923874207e-05,
      "loss": 0.3335,
      "step": 4958
    },
    {
      "epoch": 0.37530864197530867,
      "grad_norm": 0.796875,
      "learning_rate": 1.8338770566138254e-05,
      "loss": 0.3161,
      "step": 4959
    },
    {
      "epoch": 0.37538432429875596,
      "grad_norm": 1.015625,
      "learning_rate": 1.8338113090102536e-05,
      "loss": 0.3071,
      "step": 4960
    },
    {
      "epoch": 0.3754600066222033,
      "grad_norm": 0.8515625,
      "learning_rate": 1.833745549577637e-05,
      "loss": 0.3913,
      "step": 4961
    },
    {
      "epoch": 0.37553568894565065,
      "grad_norm": 0.78125,
      "learning_rate": 1.8336797783169088e-05,
      "loss": 0.3267,
      "step": 4962
    },
    {
      "epoch": 0.37561137126909794,
      "grad_norm": 0.75,
      "learning_rate": 1.8336139952290027e-05,
      "loss": 0.3067,
      "step": 4963
    },
    {
      "epoch": 0.3756870535925453,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8335482003148518e-05,
      "loss": 0.3263,
      "step": 4964
    },
    {
      "epoch": 0.37576273591599263,
      "grad_norm": 0.88671875,
      "learning_rate": 1.833482393575389e-05,
      "loss": 0.3413,
      "step": 4965
    },
    {
      "epoch": 0.37583841823944,
      "grad_norm": 0.8125,
      "learning_rate": 1.833416575011548e-05,
      "loss": 0.3649,
      "step": 4966
    },
    {
      "epoch": 0.37591410056288727,
      "grad_norm": 0.921875,
      "learning_rate": 1.8333507446242628e-05,
      "loss": 0.3841,
      "step": 4967
    },
    {
      "epoch": 0.3759897828863346,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8332849024144673e-05,
      "loss": 0.3473,
      "step": 4968
    },
    {
      "epoch": 0.37606546520978196,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8332190483830952e-05,
      "loss": 0.3538,
      "step": 4969
    },
    {
      "epoch": 0.37614114753322925,
      "grad_norm": 0.76953125,
      "learning_rate": 1.8331531825310815e-05,
      "loss": 0.2941,
      "step": 4970
    },
    {
      "epoch": 0.3762168298566766,
      "grad_norm": 0.76953125,
      "learning_rate": 1.8330873048593596e-05,
      "loss": 0.3052,
      "step": 4971
    },
    {
      "epoch": 0.37629251218012394,
      "grad_norm": 0.9140625,
      "learning_rate": 1.833021415368865e-05,
      "loss": 0.3791,
      "step": 4972
    },
    {
      "epoch": 0.3763681945035713,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8329555140605323e-05,
      "loss": 0.3378,
      "step": 4973
    },
    {
      "epoch": 0.3764438768270186,
      "grad_norm": 0.76171875,
      "learning_rate": 1.832889600935296e-05,
      "loss": 0.3077,
      "step": 4974
    },
    {
      "epoch": 0.3765195591504659,
      "grad_norm": 0.84375,
      "learning_rate": 1.8328236759940915e-05,
      "loss": 0.3421,
      "step": 4975
    },
    {
      "epoch": 0.37659524147391327,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8327577392378537e-05,
      "loss": 0.3521,
      "step": 4976
    },
    {
      "epoch": 0.37667092379736056,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8326917906675182e-05,
      "loss": 0.2818,
      "step": 4977
    },
    {
      "epoch": 0.3767466061208079,
      "grad_norm": 0.84375,
      "learning_rate": 1.8326258302840214e-05,
      "loss": 0.3729,
      "step": 4978
    },
    {
      "epoch": 0.37682228844425525,
      "grad_norm": 0.8515625,
      "learning_rate": 1.832559858088298e-05,
      "loss": 0.3521,
      "step": 4979
    },
    {
      "epoch": 0.3768979707677026,
      "grad_norm": 0.875,
      "learning_rate": 1.8324938740812842e-05,
      "loss": 0.3576,
      "step": 4980
    },
    {
      "epoch": 0.3769736530911499,
      "grad_norm": 0.8046875,
      "learning_rate": 1.832427878263916e-05,
      "loss": 0.3162,
      "step": 4981
    },
    {
      "epoch": 0.37704933541459723,
      "grad_norm": 0.78125,
      "learning_rate": 1.8323618706371303e-05,
      "loss": 0.3038,
      "step": 4982
    },
    {
      "epoch": 0.3771250177380446,
      "grad_norm": 0.796875,
      "learning_rate": 1.8322958512018625e-05,
      "loss": 0.3215,
      "step": 4983
    },
    {
      "epoch": 0.37720070006149187,
      "grad_norm": 0.8671875,
      "learning_rate": 1.8322298199590504e-05,
      "loss": 0.3576,
      "step": 4984
    },
    {
      "epoch": 0.3772763823849392,
      "grad_norm": 0.8828125,
      "learning_rate": 1.83216377690963e-05,
      "loss": 0.3935,
      "step": 4985
    },
    {
      "epoch": 0.37735206470838656,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8320977220545384e-05,
      "loss": 0.3288,
      "step": 4986
    },
    {
      "epoch": 0.3774277470318339,
      "grad_norm": 0.9453125,
      "learning_rate": 1.8320316553947125e-05,
      "loss": 0.343,
      "step": 4987
    },
    {
      "epoch": 0.3775034293552812,
      "grad_norm": 0.87109375,
      "learning_rate": 1.83196557693109e-05,
      "loss": 0.388,
      "step": 4988
    },
    {
      "epoch": 0.37757911167872854,
      "grad_norm": 0.765625,
      "learning_rate": 1.831899486664608e-05,
      "loss": 0.3098,
      "step": 4989
    },
    {
      "epoch": 0.3776547940021759,
      "grad_norm": 0.81640625,
      "learning_rate": 1.831833384596204e-05,
      "loss": 0.3492,
      "step": 4990
    },
    {
      "epoch": 0.3777304763256232,
      "grad_norm": 0.82421875,
      "learning_rate": 1.831767270726816e-05,
      "loss": 0.3435,
      "step": 4991
    },
    {
      "epoch": 0.3778061586490705,
      "grad_norm": 0.83984375,
      "learning_rate": 1.831701145057382e-05,
      "loss": 0.3918,
      "step": 4992
    },
    {
      "epoch": 0.37788184097251787,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8316350075888402e-05,
      "loss": 0.2975,
      "step": 4993
    },
    {
      "epoch": 0.37795752329596516,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8315688583221282e-05,
      "loss": 0.2912,
      "step": 4994
    },
    {
      "epoch": 0.3780332056194125,
      "grad_norm": 0.83984375,
      "learning_rate": 1.8315026972581852e-05,
      "loss": 0.3448,
      "step": 4995
    },
    {
      "epoch": 0.37810888794285985,
      "grad_norm": 0.76171875,
      "learning_rate": 1.83143652439795e-05,
      "loss": 0.3286,
      "step": 4996
    },
    {
      "epoch": 0.3781845702663072,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8313703397423602e-05,
      "loss": 0.2973,
      "step": 4997
    },
    {
      "epoch": 0.3782602525897545,
      "grad_norm": 0.8203125,
      "learning_rate": 1.831304143292356e-05,
      "loss": 0.3593,
      "step": 4998
    },
    {
      "epoch": 0.37833593491320183,
      "grad_norm": 0.859375,
      "learning_rate": 1.8312379350488752e-05,
      "loss": 0.3582,
      "step": 4999
    },
    {
      "epoch": 0.3784116172366492,
      "grad_norm": 0.8671875,
      "learning_rate": 1.8311717150128583e-05,
      "loss": 0.3225,
      "step": 5000
    },
    {
      "epoch": 0.37848729956009647,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8311054831852446e-05,
      "loss": 0.3246,
      "step": 5001
    },
    {
      "epoch": 0.3785629818835438,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8310392395669728e-05,
      "loss": 0.3231,
      "step": 5002
    },
    {
      "epoch": 0.37863866420699116,
      "grad_norm": 0.828125,
      "learning_rate": 1.8309729841589833e-05,
      "loss": 0.3309,
      "step": 5003
    },
    {
      "epoch": 0.3787143465304385,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8309067169622166e-05,
      "loss": 0.3085,
      "step": 5004
    },
    {
      "epoch": 0.3787900288538858,
      "grad_norm": 0.7578125,
      "learning_rate": 1.830840437977612e-05,
      "loss": 0.2807,
      "step": 5005
    },
    {
      "epoch": 0.37886571117733314,
      "grad_norm": 0.75390625,
      "learning_rate": 1.8307741472061097e-05,
      "loss": 0.3,
      "step": 5006
    },
    {
      "epoch": 0.3789413935007805,
      "grad_norm": 0.87890625,
      "learning_rate": 1.8307078446486506e-05,
      "loss": 0.2915,
      "step": 5007
    },
    {
      "epoch": 0.3790170758242278,
      "grad_norm": 0.83984375,
      "learning_rate": 1.8306415303061754e-05,
      "loss": 0.3445,
      "step": 5008
    },
    {
      "epoch": 0.3790927581476751,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8305752041796247e-05,
      "loss": 0.3446,
      "step": 5009
    },
    {
      "epoch": 0.37916844047112247,
      "grad_norm": 1.6171875,
      "learning_rate": 1.830508866269939e-05,
      "loss": 0.3658,
      "step": 5010
    },
    {
      "epoch": 0.3792441227945698,
      "grad_norm": 0.765625,
      "learning_rate": 1.8304425165780603e-05,
      "loss": 0.3066,
      "step": 5011
    },
    {
      "epoch": 0.3793198051180171,
      "grad_norm": 0.86328125,
      "learning_rate": 1.8303761551049287e-05,
      "loss": 0.3461,
      "step": 5012
    },
    {
      "epoch": 0.37939548744146445,
      "grad_norm": 0.8515625,
      "learning_rate": 1.830309781851487e-05,
      "loss": 0.3385,
      "step": 5013
    },
    {
      "epoch": 0.3794711697649118,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8302433968186763e-05,
      "loss": 0.3348,
      "step": 5014
    },
    {
      "epoch": 0.3795468520883591,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8301770000074376e-05,
      "loss": 0.3137,
      "step": 5015
    },
    {
      "epoch": 0.37962253441180643,
      "grad_norm": 0.85546875,
      "learning_rate": 1.830110591418714e-05,
      "loss": 0.3778,
      "step": 5016
    },
    {
      "epoch": 0.3796982167352538,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8300441710534473e-05,
      "loss": 0.3177,
      "step": 5017
    },
    {
      "epoch": 0.3797738990587011,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8299777389125792e-05,
      "loss": 0.3637,
      "step": 5018
    },
    {
      "epoch": 0.3798495813821484,
      "grad_norm": 0.79296875,
      "learning_rate": 1.829911294997053e-05,
      "loss": 0.3139,
      "step": 5019
    },
    {
      "epoch": 0.37992526370559576,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8298448393078106e-05,
      "loss": 0.3751,
      "step": 5020
    },
    {
      "epoch": 0.3800009460290431,
      "grad_norm": 0.8125,
      "learning_rate": 1.8297783718457953e-05,
      "loss": 0.3406,
      "step": 5021
    },
    {
      "epoch": 0.3800766283524904,
      "grad_norm": 0.75390625,
      "learning_rate": 1.8297118926119496e-05,
      "loss": 0.2945,
      "step": 5022
    },
    {
      "epoch": 0.38015231067593774,
      "grad_norm": 0.875,
      "learning_rate": 1.829645401607217e-05,
      "loss": 0.4039,
      "step": 5023
    },
    {
      "epoch": 0.3802279929993851,
      "grad_norm": 0.78125,
      "learning_rate": 1.8295788988325408e-05,
      "loss": 0.3344,
      "step": 5024
    },
    {
      "epoch": 0.38030367532283244,
      "grad_norm": 0.796875,
      "learning_rate": 1.829512384288864e-05,
      "loss": 0.3335,
      "step": 5025
    },
    {
      "epoch": 0.3803793576462797,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8294458579771305e-05,
      "loss": 0.3509,
      "step": 5026
    },
    {
      "epoch": 0.38045503996972707,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8293793198982844e-05,
      "loss": 0.3223,
      "step": 5027
    },
    {
      "epoch": 0.3805307222931744,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8293127700532693e-05,
      "loss": 0.3318,
      "step": 5028
    },
    {
      "epoch": 0.3806064046166217,
      "grad_norm": 0.8125,
      "learning_rate": 1.8292462084430297e-05,
      "loss": 0.3185,
      "step": 5029
    },
    {
      "epoch": 0.38068208694006905,
      "grad_norm": 0.765625,
      "learning_rate": 1.829179635068509e-05,
      "loss": 0.3055,
      "step": 5030
    },
    {
      "epoch": 0.3807577692635164,
      "grad_norm": 0.75,
      "learning_rate": 1.8291130499306526e-05,
      "loss": 0.2807,
      "step": 5031
    },
    {
      "epoch": 0.38083345158696374,
      "grad_norm": 0.86328125,
      "learning_rate": 1.8290464530304047e-05,
      "loss": 0.3565,
      "step": 5032
    },
    {
      "epoch": 0.38090913391041104,
      "grad_norm": 0.7421875,
      "learning_rate": 1.8289798443687103e-05,
      "loss": 0.2944,
      "step": 5033
    },
    {
      "epoch": 0.3809848162338584,
      "grad_norm": 0.85546875,
      "learning_rate": 1.828913223946514e-05,
      "loss": 0.2991,
      "step": 5034
    },
    {
      "epoch": 0.3810604985573057,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8288465917647615e-05,
      "loss": 0.3369,
      "step": 5035
    },
    {
      "epoch": 0.381136180880753,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8287799478243976e-05,
      "loss": 0.3239,
      "step": 5036
    },
    {
      "epoch": 0.38121186320420036,
      "grad_norm": 0.73046875,
      "learning_rate": 1.8287132921263676e-05,
      "loss": 0.3068,
      "step": 5037
    },
    {
      "epoch": 0.3812875455276477,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8286466246716176e-05,
      "loss": 0.3465,
      "step": 5038
    },
    {
      "epoch": 0.38136322785109505,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8285799454610935e-05,
      "loss": 0.3188,
      "step": 5039
    },
    {
      "epoch": 0.38143891017454234,
      "grad_norm": 0.7734375,
      "learning_rate": 1.828513254495741e-05,
      "loss": 0.2741,
      "step": 5040
    },
    {
      "epoch": 0.3815145924979897,
      "grad_norm": 0.90625,
      "learning_rate": 1.828446551776506e-05,
      "loss": 0.3641,
      "step": 5041
    },
    {
      "epoch": 0.38159027482143704,
      "grad_norm": 0.859375,
      "learning_rate": 1.828379837304335e-05,
      "loss": 0.3919,
      "step": 5042
    },
    {
      "epoch": 0.3816659571448843,
      "grad_norm": 1.296875,
      "learning_rate": 1.828313111080174e-05,
      "loss": 0.4029,
      "step": 5043
    },
    {
      "epoch": 0.3817416394683317,
      "grad_norm": 0.78515625,
      "learning_rate": 1.828246373104971e-05,
      "loss": 0.3331,
      "step": 5044
    },
    {
      "epoch": 0.381817321791779,
      "grad_norm": 0.79296875,
      "learning_rate": 1.828179623379671e-05,
      "loss": 0.3269,
      "step": 5045
    },
    {
      "epoch": 0.38189300411522636,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8281128619052224e-05,
      "loss": 0.3309,
      "step": 5046
    },
    {
      "epoch": 0.38196868643867365,
      "grad_norm": 0.828125,
      "learning_rate": 1.8280460886825718e-05,
      "loss": 0.3392,
      "step": 5047
    },
    {
      "epoch": 0.382044368762121,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8279793037126662e-05,
      "loss": 0.3567,
      "step": 5048
    },
    {
      "epoch": 0.38212005108556835,
      "grad_norm": 1.3515625,
      "learning_rate": 1.8279125069964533e-05,
      "loss": 0.3818,
      "step": 5049
    },
    {
      "epoch": 0.38219573340901564,
      "grad_norm": 0.7890625,
      "learning_rate": 1.827845698534881e-05,
      "loss": 0.3127,
      "step": 5050
    },
    {
      "epoch": 0.382271415732463,
      "grad_norm": 0.84375,
      "learning_rate": 1.8277788783288965e-05,
      "loss": 0.3335,
      "step": 5051
    },
    {
      "epoch": 0.38234709805591033,
      "grad_norm": 0.84375,
      "learning_rate": 1.8277120463794486e-05,
      "loss": 0.3774,
      "step": 5052
    },
    {
      "epoch": 0.3824227803793576,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8276452026874843e-05,
      "loss": 0.3323,
      "step": 5053
    },
    {
      "epoch": 0.38249846270280496,
      "grad_norm": 0.78125,
      "learning_rate": 1.8275783472539528e-05,
      "loss": 0.3321,
      "step": 5054
    },
    {
      "epoch": 0.3825741450262523,
      "grad_norm": 0.8984375,
      "learning_rate": 1.8275114800798022e-05,
      "loss": 0.3492,
      "step": 5055
    },
    {
      "epoch": 0.38264982734969966,
      "grad_norm": 0.828125,
      "learning_rate": 1.8274446011659813e-05,
      "loss": 0.3509,
      "step": 5056
    },
    {
      "epoch": 0.38272550967314695,
      "grad_norm": 1.03125,
      "learning_rate": 1.8273777105134388e-05,
      "loss": 0.3425,
      "step": 5057
    },
    {
      "epoch": 0.3828011919965943,
      "grad_norm": 0.78125,
      "learning_rate": 1.8273108081231234e-05,
      "loss": 0.3349,
      "step": 5058
    },
    {
      "epoch": 0.38287687432004164,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8272438939959846e-05,
      "loss": 0.3316,
      "step": 5059
    },
    {
      "epoch": 0.38295255664348893,
      "grad_norm": 0.93359375,
      "learning_rate": 1.827176968132972e-05,
      "loss": 0.3607,
      "step": 5060
    },
    {
      "epoch": 0.3830282389669363,
      "grad_norm": 0.74609375,
      "learning_rate": 1.827110030535034e-05,
      "loss": 0.28,
      "step": 5061
    },
    {
      "epoch": 0.3831039212903836,
      "grad_norm": 0.83203125,
      "learning_rate": 1.827043081203121e-05,
      "loss": 0.3658,
      "step": 5062
    },
    {
      "epoch": 0.38317960361383097,
      "grad_norm": 0.84375,
      "learning_rate": 1.8269761201381828e-05,
      "loss": 0.3471,
      "step": 5063
    },
    {
      "epoch": 0.38325528593727826,
      "grad_norm": 0.76171875,
      "learning_rate": 1.8269091473411694e-05,
      "loss": 0.3228,
      "step": 5064
    },
    {
      "epoch": 0.3833309682607256,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8268421628130303e-05,
      "loss": 0.3265,
      "step": 5065
    },
    {
      "epoch": 0.38340665058417295,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8267751665547164e-05,
      "loss": 0.3693,
      "step": 5066
    },
    {
      "epoch": 0.38348233290762024,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8267081585671778e-05,
      "loss": 0.3518,
      "step": 5067
    },
    {
      "epoch": 0.3835580152310676,
      "grad_norm": 0.75,
      "learning_rate": 1.8266411388513656e-05,
      "loss": 0.2958,
      "step": 5068
    },
    {
      "epoch": 0.38363369755451493,
      "grad_norm": 0.8046875,
      "learning_rate": 1.82657410740823e-05,
      "loss": 0.3337,
      "step": 5069
    },
    {
      "epoch": 0.3837093798779623,
      "grad_norm": 0.76171875,
      "learning_rate": 1.8265070642387223e-05,
      "loss": 0.3189,
      "step": 5070
    },
    {
      "epoch": 0.38378506220140957,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8264400093437937e-05,
      "loss": 0.3085,
      "step": 5071
    },
    {
      "epoch": 0.3838607445248569,
      "grad_norm": 0.8828125,
      "learning_rate": 1.8263729427243953e-05,
      "loss": 0.3757,
      "step": 5072
    },
    {
      "epoch": 0.38393642684830426,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8263058643814784e-05,
      "loss": 0.3109,
      "step": 5073
    },
    {
      "epoch": 0.38401210917175155,
      "grad_norm": 0.77734375,
      "learning_rate": 1.826238774315995e-05,
      "loss": 0.3137,
      "step": 5074
    },
    {
      "epoch": 0.3840877914951989,
      "grad_norm": 0.69140625,
      "learning_rate": 1.8261716725288968e-05,
      "loss": 0.2301,
      "step": 5075
    },
    {
      "epoch": 0.38416347381864624,
      "grad_norm": 0.81640625,
      "learning_rate": 1.826104559021135e-05,
      "loss": 0.2957,
      "step": 5076
    },
    {
      "epoch": 0.3842391561420936,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8260374337936633e-05,
      "loss": 0.2994,
      "step": 5077
    },
    {
      "epoch": 0.3843148384655409,
      "grad_norm": 0.76171875,
      "learning_rate": 1.825970296847433e-05,
      "loss": 0.3211,
      "step": 5078
    },
    {
      "epoch": 0.3843905207889882,
      "grad_norm": 0.71484375,
      "learning_rate": 1.825903148183396e-05,
      "loss": 0.2929,
      "step": 5079
    },
    {
      "epoch": 0.38446620311243557,
      "grad_norm": 0.76953125,
      "learning_rate": 1.8258359878025062e-05,
      "loss": 0.3201,
      "step": 5080
    },
    {
      "epoch": 0.38454188543588286,
      "grad_norm": 0.76953125,
      "learning_rate": 1.825768815705715e-05,
      "loss": 0.274,
      "step": 5081
    },
    {
      "epoch": 0.3846175677593302,
      "grad_norm": 0.7734375,
      "learning_rate": 1.825701631893977e-05,
      "loss": 0.3189,
      "step": 5082
    },
    {
      "epoch": 0.38469325008277755,
      "grad_norm": 2.734375,
      "learning_rate": 1.8256344363682436e-05,
      "loss": 0.4166,
      "step": 5083
    },
    {
      "epoch": 0.3847689324062249,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8255672291294694e-05,
      "loss": 0.3064,
      "step": 5084
    },
    {
      "epoch": 0.3848446147296722,
      "grad_norm": 0.79296875,
      "learning_rate": 1.825500010178607e-05,
      "loss": 0.3262,
      "step": 5085
    },
    {
      "epoch": 0.38492029705311953,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8254327795166104e-05,
      "loss": 0.3153,
      "step": 5086
    },
    {
      "epoch": 0.3849959793765669,
      "grad_norm": 0.74609375,
      "learning_rate": 1.8253655371444332e-05,
      "loss": 0.3118,
      "step": 5087
    },
    {
      "epoch": 0.38507166170001417,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8252982830630297e-05,
      "loss": 0.3213,
      "step": 5088
    },
    {
      "epoch": 0.3851473440234615,
      "grad_norm": 0.6953125,
      "learning_rate": 1.8252310172733537e-05,
      "loss": 0.2889,
      "step": 5089
    },
    {
      "epoch": 0.38522302634690886,
      "grad_norm": 0.73828125,
      "learning_rate": 1.8251637397763597e-05,
      "loss": 0.3095,
      "step": 5090
    },
    {
      "epoch": 0.3852987086703562,
      "grad_norm": 0.7265625,
      "learning_rate": 1.825096450573002e-05,
      "loss": 0.3063,
      "step": 5091
    },
    {
      "epoch": 0.3853743909938035,
      "grad_norm": 0.75,
      "learning_rate": 1.8250291496642352e-05,
      "loss": 0.2811,
      "step": 5092
    },
    {
      "epoch": 0.38545007331725084,
      "grad_norm": 1.21875,
      "learning_rate": 1.824961837051014e-05,
      "loss": 0.3975,
      "step": 5093
    },
    {
      "epoch": 0.3855257556406982,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8248945127342937e-05,
      "loss": 0.2928,
      "step": 5094
    },
    {
      "epoch": 0.3856014379641455,
      "grad_norm": 0.75,
      "learning_rate": 1.824827176715029e-05,
      "loss": 0.3025,
      "step": 5095
    },
    {
      "epoch": 0.3856771202875928,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8247598289941755e-05,
      "loss": 0.3578,
      "step": 5096
    },
    {
      "epoch": 0.38575280261104017,
      "grad_norm": 0.75390625,
      "learning_rate": 1.824692469572688e-05,
      "loss": 0.3098,
      "step": 5097
    },
    {
      "epoch": 0.3858284849344875,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8246250984515233e-05,
      "loss": 0.3138,
      "step": 5098
    },
    {
      "epoch": 0.3859041672579348,
      "grad_norm": 0.8125,
      "learning_rate": 1.824557715631636e-05,
      "loss": 0.3586,
      "step": 5099
    },
    {
      "epoch": 0.38597984958138215,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8244903211139825e-05,
      "loss": 0.3438,
      "step": 5100
    },
    {
      "epoch": 0.3860555319048295,
      "grad_norm": 0.75390625,
      "learning_rate": 1.8244229148995193e-05,
      "loss": 0.32,
      "step": 5101
    },
    {
      "epoch": 0.3861312142282768,
      "grad_norm": 0.78125,
      "learning_rate": 1.824355496989202e-05,
      "loss": 0.302,
      "step": 5102
    },
    {
      "epoch": 0.38620689655172413,
      "grad_norm": 0.84375,
      "learning_rate": 1.8242880673839876e-05,
      "loss": 0.3841,
      "step": 5103
    },
    {
      "epoch": 0.3862825788751715,
      "grad_norm": 0.765625,
      "learning_rate": 1.8242206260848322e-05,
      "loss": 0.3184,
      "step": 5104
    },
    {
      "epoch": 0.3863582611986188,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8241531730926925e-05,
      "loss": 0.3427,
      "step": 5105
    },
    {
      "epoch": 0.3864339435220661,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8240857084085265e-05,
      "loss": 0.2943,
      "step": 5106
    },
    {
      "epoch": 0.38650962584551346,
      "grad_norm": 0.80859375,
      "learning_rate": 1.82401823203329e-05,
      "loss": 0.3758,
      "step": 5107
    },
    {
      "epoch": 0.3865853081689608,
      "grad_norm": 0.83203125,
      "learning_rate": 1.823950743967941e-05,
      "loss": 0.3729,
      "step": 5108
    },
    {
      "epoch": 0.3866609904924081,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8238832442134366e-05,
      "loss": 0.2985,
      "step": 5109
    },
    {
      "epoch": 0.38673667281585544,
      "grad_norm": 0.74609375,
      "learning_rate": 1.8238157327707347e-05,
      "loss": 0.2804,
      "step": 5110
    },
    {
      "epoch": 0.3868123551393028,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8237482096407928e-05,
      "loss": 0.341,
      "step": 5111
    },
    {
      "epoch": 0.38688803746275013,
      "grad_norm": 0.74609375,
      "learning_rate": 1.823680674824569e-05,
      "loss": 0.2843,
      "step": 5112
    },
    {
      "epoch": 0.3869637197861974,
      "grad_norm": 0.7578125,
      "learning_rate": 1.823613128323021e-05,
      "loss": 0.3065,
      "step": 5113
    },
    {
      "epoch": 0.38703940210964477,
      "grad_norm": 0.83203125,
      "learning_rate": 1.8235455701371082e-05,
      "loss": 0.369,
      "step": 5114
    },
    {
      "epoch": 0.3871150844330921,
      "grad_norm": 0.78515625,
      "learning_rate": 1.823478000267788e-05,
      "loss": 0.3135,
      "step": 5115
    },
    {
      "epoch": 0.3871907667565394,
      "grad_norm": 0.76953125,
      "learning_rate": 1.823410418716019e-05,
      "loss": 0.3276,
      "step": 5116
    },
    {
      "epoch": 0.38726644907998675,
      "grad_norm": 0.71875,
      "learning_rate": 1.82334282548276e-05,
      "loss": 0.28,
      "step": 5117
    },
    {
      "epoch": 0.3873421314034341,
      "grad_norm": 0.8125,
      "learning_rate": 1.82327522056897e-05,
      "loss": 0.3647,
      "step": 5118
    },
    {
      "epoch": 0.3874178137268814,
      "grad_norm": 0.765625,
      "learning_rate": 1.8232076039756086e-05,
      "loss": 0.3377,
      "step": 5119
    },
    {
      "epoch": 0.38749349605032873,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8231399757036347e-05,
      "loss": 0.3424,
      "step": 5120
    },
    {
      "epoch": 0.3875691783737761,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8230723357540076e-05,
      "loss": 0.3164,
      "step": 5121
    },
    {
      "epoch": 0.3876448606972234,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8230046841276873e-05,
      "loss": 0.3356,
      "step": 5122
    },
    {
      "epoch": 0.3877205430206707,
      "grad_norm": 0.78125,
      "learning_rate": 1.8229370208256325e-05,
      "loss": 0.3481,
      "step": 5123
    },
    {
      "epoch": 0.38779622534411806,
      "grad_norm": 0.76953125,
      "learning_rate": 1.8228693458488043e-05,
      "loss": 0.278,
      "step": 5124
    },
    {
      "epoch": 0.3878719076675654,
      "grad_norm": 0.89453125,
      "learning_rate": 1.8228016591981623e-05,
      "loss": 0.3844,
      "step": 5125
    },
    {
      "epoch": 0.3879475899910127,
      "grad_norm": 0.76953125,
      "learning_rate": 1.822733960874667e-05,
      "loss": 0.313,
      "step": 5126
    },
    {
      "epoch": 0.38802327231446004,
      "grad_norm": 0.72265625,
      "learning_rate": 1.822666250879278e-05,
      "loss": 0.268,
      "step": 5127
    },
    {
      "epoch": 0.3880989546379074,
      "grad_norm": 0.78515625,
      "learning_rate": 1.822598529212957e-05,
      "loss": 0.3125,
      "step": 5128
    },
    {
      "epoch": 0.38817463696135474,
      "grad_norm": 0.79296875,
      "learning_rate": 1.822530795876664e-05,
      "loss": 0.346,
      "step": 5129
    },
    {
      "epoch": 0.388250319284802,
      "grad_norm": 0.859375,
      "learning_rate": 1.8224630508713606e-05,
      "loss": 0.3313,
      "step": 5130
    },
    {
      "epoch": 0.38832600160824937,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8223952941980068e-05,
      "loss": 0.312,
      "step": 5131
    },
    {
      "epoch": 0.3884016839316967,
      "grad_norm": 0.7421875,
      "learning_rate": 1.822327525857565e-05,
      "loss": 0.2972,
      "step": 5132
    },
    {
      "epoch": 0.388477366255144,
      "grad_norm": 0.76953125,
      "learning_rate": 1.8222597458509955e-05,
      "loss": 0.3145,
      "step": 5133
    },
    {
      "epoch": 0.38855304857859135,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8221919541792608e-05,
      "loss": 0.3488,
      "step": 5134
    },
    {
      "epoch": 0.3886287309020387,
      "grad_norm": 0.74609375,
      "learning_rate": 1.8221241508433224e-05,
      "loss": 0.3006,
      "step": 5135
    },
    {
      "epoch": 0.38870441322548605,
      "grad_norm": 0.765625,
      "learning_rate": 1.822056335844142e-05,
      "loss": 0.3105,
      "step": 5136
    },
    {
      "epoch": 0.38878009554893334,
      "grad_norm": 0.71484375,
      "learning_rate": 1.821988509182682e-05,
      "loss": 0.273,
      "step": 5137
    },
    {
      "epoch": 0.3888557778723807,
      "grad_norm": 0.77734375,
      "learning_rate": 1.821920670859904e-05,
      "loss": 0.3387,
      "step": 5138
    },
    {
      "epoch": 0.388931460195828,
      "grad_norm": 0.76171875,
      "learning_rate": 1.821852820876771e-05,
      "loss": 0.3164,
      "step": 5139
    },
    {
      "epoch": 0.3890071425192753,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8217849592342457e-05,
      "loss": 0.3566,
      "step": 5140
    },
    {
      "epoch": 0.38908282484272266,
      "grad_norm": 0.8125,
      "learning_rate": 1.82171708593329e-05,
      "loss": 0.3678,
      "step": 5141
    },
    {
      "epoch": 0.38915850716617,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8216492009748674e-05,
      "loss": 0.3473,
      "step": 5142
    },
    {
      "epoch": 0.38923418948961735,
      "grad_norm": 0.77734375,
      "learning_rate": 1.821581304359941e-05,
      "loss": 0.3127,
      "step": 5143
    },
    {
      "epoch": 0.38930987181306465,
      "grad_norm": 0.76171875,
      "learning_rate": 1.8215133960894743e-05,
      "loss": 0.3226,
      "step": 5144
    },
    {
      "epoch": 0.389385554136512,
      "grad_norm": 0.7265625,
      "learning_rate": 1.8214454761644295e-05,
      "loss": 0.2936,
      "step": 5145
    },
    {
      "epoch": 0.38946123645995934,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8213775445857716e-05,
      "loss": 0.303,
      "step": 5146
    },
    {
      "epoch": 0.3895369187834066,
      "grad_norm": 0.84765625,
      "learning_rate": 1.8213096013544635e-05,
      "loss": 0.3491,
      "step": 5147
    },
    {
      "epoch": 0.389612601106854,
      "grad_norm": 0.89453125,
      "learning_rate": 1.8212416464714695e-05,
      "loss": 0.3316,
      "step": 5148
    },
    {
      "epoch": 0.3896882834303013,
      "grad_norm": 0.83984375,
      "learning_rate": 1.8211736799377534e-05,
      "loss": 0.3035,
      "step": 5149
    },
    {
      "epoch": 0.38976396575374866,
      "grad_norm": 0.7890625,
      "learning_rate": 1.821105701754279e-05,
      "loss": 0.3025,
      "step": 5150
    },
    {
      "epoch": 0.38983964807719595,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8210377119220116e-05,
      "loss": 0.3114,
      "step": 5151
    },
    {
      "epoch": 0.3899153304006433,
      "grad_norm": 0.8125,
      "learning_rate": 1.820969710441915e-05,
      "loss": 0.3432,
      "step": 5152
    },
    {
      "epoch": 0.38999101272409065,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8209016973149545e-05,
      "loss": 0.3223,
      "step": 5153
    },
    {
      "epoch": 0.39006669504753794,
      "grad_norm": 0.8125,
      "learning_rate": 1.8208336725420946e-05,
      "loss": 0.365,
      "step": 5154
    },
    {
      "epoch": 0.3901423773709853,
      "grad_norm": 0.765625,
      "learning_rate": 1.8207656361243006e-05,
      "loss": 0.3377,
      "step": 5155
    },
    {
      "epoch": 0.39021805969443263,
      "grad_norm": 0.76171875,
      "learning_rate": 1.8206975880625375e-05,
      "loss": 0.3201,
      "step": 5156
    },
    {
      "epoch": 0.39029374201788,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8206295283577705e-05,
      "loss": 0.3399,
      "step": 5157
    },
    {
      "epoch": 0.39036942434132726,
      "grad_norm": 0.828125,
      "learning_rate": 1.8205614570109656e-05,
      "loss": 0.3531,
      "step": 5158
    },
    {
      "epoch": 0.3904451066647746,
      "grad_norm": 0.76171875,
      "learning_rate": 1.8204933740230884e-05,
      "loss": 0.3177,
      "step": 5159
    },
    {
      "epoch": 0.39052078898822196,
      "grad_norm": 0.74609375,
      "learning_rate": 1.8204252793951046e-05,
      "loss": 0.2998,
      "step": 5160
    },
    {
      "epoch": 0.39059647131166925,
      "grad_norm": 0.71484375,
      "learning_rate": 1.82035717312798e-05,
      "loss": 0.285,
      "step": 5161
    },
    {
      "epoch": 0.3906721536351166,
      "grad_norm": 0.796875,
      "learning_rate": 1.8202890552226814e-05,
      "loss": 0.3231,
      "step": 5162
    },
    {
      "epoch": 0.39074783595856394,
      "grad_norm": 0.83984375,
      "learning_rate": 1.820220925680175e-05,
      "loss": 0.3062,
      "step": 5163
    },
    {
      "epoch": 0.3908235182820113,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8201527845014274e-05,
      "loss": 0.3367,
      "step": 5164
    },
    {
      "epoch": 0.3908992006054586,
      "grad_norm": 0.75390625,
      "learning_rate": 1.8200846316874046e-05,
      "loss": 0.2957,
      "step": 5165
    },
    {
      "epoch": 0.3909748829289059,
      "grad_norm": 0.72265625,
      "learning_rate": 1.8200164672390743e-05,
      "loss": 0.2931,
      "step": 5166
    },
    {
      "epoch": 0.39105056525235327,
      "grad_norm": 0.78125,
      "learning_rate": 1.819948291157403e-05,
      "loss": 0.3146,
      "step": 5167
    },
    {
      "epoch": 0.39112624757580056,
      "grad_norm": 0.86328125,
      "learning_rate": 1.8198801034433585e-05,
      "loss": 0.329,
      "step": 5168
    },
    {
      "epoch": 0.3912019298992479,
      "grad_norm": 0.78515625,
      "learning_rate": 1.819811904097907e-05,
      "loss": 0.33,
      "step": 5169
    },
    {
      "epoch": 0.39127761222269525,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8197436931220177e-05,
      "loss": 0.3645,
      "step": 5170
    },
    {
      "epoch": 0.3913532945461426,
      "grad_norm": 0.7734375,
      "learning_rate": 1.819675470516657e-05,
      "loss": 0.3163,
      "step": 5171
    },
    {
      "epoch": 0.3914289768695899,
      "grad_norm": 0.75390625,
      "learning_rate": 1.819607236282793e-05,
      "loss": 0.2847,
      "step": 5172
    },
    {
      "epoch": 0.39150465919303723,
      "grad_norm": 0.765625,
      "learning_rate": 1.819538990421394e-05,
      "loss": 0.3063,
      "step": 5173
    },
    {
      "epoch": 0.3915803415164846,
      "grad_norm": 0.734375,
      "learning_rate": 1.8194707329334277e-05,
      "loss": 0.289,
      "step": 5174
    },
    {
      "epoch": 0.39165602383993187,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8194024638198632e-05,
      "loss": 0.3192,
      "step": 5175
    },
    {
      "epoch": 0.3917317061633792,
      "grad_norm": 0.828125,
      "learning_rate": 1.8193341830816687e-05,
      "loss": 0.3522,
      "step": 5176
    },
    {
      "epoch": 0.39180738848682656,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8192658907198126e-05,
      "loss": 0.3336,
      "step": 5177
    },
    {
      "epoch": 0.39188307081027385,
      "grad_norm": 0.8359375,
      "learning_rate": 1.819197586735264e-05,
      "loss": 0.3146,
      "step": 5178
    },
    {
      "epoch": 0.3919587531337212,
      "grad_norm": 0.84375,
      "learning_rate": 1.8191292711289915e-05,
      "loss": 0.372,
      "step": 5179
    },
    {
      "epoch": 0.39203443545716854,
      "grad_norm": 0.75,
      "learning_rate": 1.819060943901965e-05,
      "loss": 0.2993,
      "step": 5180
    },
    {
      "epoch": 0.3921101177806159,
      "grad_norm": 0.8125,
      "learning_rate": 1.8189926050551534e-05,
      "loss": 0.3199,
      "step": 5181
    },
    {
      "epoch": 0.3921858001040632,
      "grad_norm": 0.78125,
      "learning_rate": 1.8189242545895262e-05,
      "loss": 0.3362,
      "step": 5182
    },
    {
      "epoch": 0.3922614824275105,
      "grad_norm": 0.83984375,
      "learning_rate": 1.8188558925060534e-05,
      "loss": 0.3475,
      "step": 5183
    },
    {
      "epoch": 0.39233716475095787,
      "grad_norm": 0.78125,
      "learning_rate": 1.8187875188057043e-05,
      "loss": 0.308,
      "step": 5184
    },
    {
      "epoch": 0.39241284707440516,
      "grad_norm": 1.2578125,
      "learning_rate": 1.818719133489449e-05,
      "loss": 0.3763,
      "step": 5185
    },
    {
      "epoch": 0.3924885293978525,
      "grad_norm": 0.79296875,
      "learning_rate": 1.818650736558258e-05,
      "loss": 0.3054,
      "step": 5186
    },
    {
      "epoch": 0.39256421172129985,
      "grad_norm": 0.84765625,
      "learning_rate": 1.8185823280131015e-05,
      "loss": 0.3496,
      "step": 5187
    },
    {
      "epoch": 0.3926398940447472,
      "grad_norm": 0.84375,
      "learning_rate": 1.81851390785495e-05,
      "loss": 0.399,
      "step": 5188
    },
    {
      "epoch": 0.3927155763681945,
      "grad_norm": 0.95703125,
      "learning_rate": 1.8184454760847737e-05,
      "loss": 0.3029,
      "step": 5189
    },
    {
      "epoch": 0.39279125869164183,
      "grad_norm": 0.76953125,
      "learning_rate": 1.818377032703544e-05,
      "loss": 0.3128,
      "step": 5190
    },
    {
      "epoch": 0.3928669410150892,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8183085777122322e-05,
      "loss": 0.3605,
      "step": 5191
    },
    {
      "epoch": 0.39294262333853647,
      "grad_norm": 0.78125,
      "learning_rate": 1.818240111111808e-05,
      "loss": 0.3238,
      "step": 5192
    },
    {
      "epoch": 0.3930183056619838,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8181716329032445e-05,
      "loss": 0.3473,
      "step": 5193
    },
    {
      "epoch": 0.39309398798543116,
      "grad_norm": 0.91015625,
      "learning_rate": 1.818103143087512e-05,
      "loss": 0.3908,
      "step": 5194
    },
    {
      "epoch": 0.3931696703088785,
      "grad_norm": 0.76953125,
      "learning_rate": 1.8180346416655827e-05,
      "loss": 0.3123,
      "step": 5195
    },
    {
      "epoch": 0.3932453526323258,
      "grad_norm": 2.734375,
      "learning_rate": 1.8179661286384284e-05,
      "loss": 0.4499,
      "step": 5196
    },
    {
      "epoch": 0.39332103495577314,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8178976040070206e-05,
      "loss": 0.3389,
      "step": 5197
    },
    {
      "epoch": 0.3933967172792205,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8178290677723314e-05,
      "loss": 0.3613,
      "step": 5198
    },
    {
      "epoch": 0.3934723996026678,
      "grad_norm": 0.83984375,
      "learning_rate": 1.8177605199353338e-05,
      "loss": 0.3734,
      "step": 5199
    },
    {
      "epoch": 0.3935480819261151,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8176919604969997e-05,
      "loss": 0.3159,
      "step": 5200
    },
    {
      "epoch": 0.39362376424956247,
      "grad_norm": 0.875,
      "learning_rate": 1.8176233894583023e-05,
      "loss": 0.3585,
      "step": 5201
    },
    {
      "epoch": 0.3936994465730098,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8175548068202138e-05,
      "loss": 0.3054,
      "step": 5202
    },
    {
      "epoch": 0.3937751288964571,
      "grad_norm": 0.828125,
      "learning_rate": 1.8174862125837075e-05,
      "loss": 0.3192,
      "step": 5203
    },
    {
      "epoch": 0.39385081121990445,
      "grad_norm": 0.765625,
      "learning_rate": 1.8174176067497563e-05,
      "loss": 0.3159,
      "step": 5204
    },
    {
      "epoch": 0.3939264935433518,
      "grad_norm": 1.1796875,
      "learning_rate": 1.8173489893193334e-05,
      "loss": 0.3637,
      "step": 5205
    },
    {
      "epoch": 0.3940021758667991,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8172803602934127e-05,
      "loss": 0.3613,
      "step": 5206
    },
    {
      "epoch": 0.39407785819024643,
      "grad_norm": 0.78125,
      "learning_rate": 1.8172117196729675e-05,
      "loss": 0.3235,
      "step": 5207
    },
    {
      "epoch": 0.3941535405136938,
      "grad_norm": 0.8359375,
      "learning_rate": 1.817143067458972e-05,
      "loss": 0.3694,
      "step": 5208
    },
    {
      "epoch": 0.3942292228371411,
      "grad_norm": 0.8515625,
      "learning_rate": 1.8170744036523995e-05,
      "loss": 0.371,
      "step": 5209
    },
    {
      "epoch": 0.3943049051605884,
      "grad_norm": 0.859375,
      "learning_rate": 1.8170057282542246e-05,
      "loss": 0.3526,
      "step": 5210
    },
    {
      "epoch": 0.39438058748403576,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8169370412654212e-05,
      "loss": 0.3347,
      "step": 5211
    },
    {
      "epoch": 0.3944562698074831,
      "grad_norm": 0.77734375,
      "learning_rate": 1.816868342686964e-05,
      "loss": 0.3356,
      "step": 5212
    },
    {
      "epoch": 0.3945319521309304,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8167996325198278e-05,
      "loss": 0.3524,
      "step": 5213
    },
    {
      "epoch": 0.39460763445437774,
      "grad_norm": 0.74609375,
      "learning_rate": 1.816730910764987e-05,
      "loss": 0.2936,
      "step": 5214
    },
    {
      "epoch": 0.3946833167778251,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8166621774234166e-05,
      "loss": 0.313,
      "step": 5215
    },
    {
      "epoch": 0.39475899910127243,
      "grad_norm": 0.7890625,
      "learning_rate": 1.816593432496092e-05,
      "loss": 0.2986,
      "step": 5216
    },
    {
      "epoch": 0.3948346814247197,
      "grad_norm": 0.7734375,
      "learning_rate": 1.816524675983988e-05,
      "loss": 0.2761,
      "step": 5217
    },
    {
      "epoch": 0.39491036374816707,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8164559078880804e-05,
      "loss": 0.3206,
      "step": 5218
    },
    {
      "epoch": 0.3949860460716144,
      "grad_norm": 0.69140625,
      "learning_rate": 1.8163871282093445e-05,
      "loss": 0.2684,
      "step": 5219
    },
    {
      "epoch": 0.3950617283950617,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8163183369487565e-05,
      "loss": 0.3594,
      "step": 5220
    },
    {
      "epoch": 0.39513741071850905,
      "grad_norm": 0.84375,
      "learning_rate": 1.8162495341072918e-05,
      "loss": 0.367,
      "step": 5221
    },
    {
      "epoch": 0.3952130930419564,
      "grad_norm": 0.87890625,
      "learning_rate": 1.8161807196859268e-05,
      "loss": 0.3882,
      "step": 5222
    },
    {
      "epoch": 0.39528877536540374,
      "grad_norm": 0.85546875,
      "learning_rate": 1.8161118936856374e-05,
      "loss": 0.4031,
      "step": 5223
    },
    {
      "epoch": 0.39536445768885103,
      "grad_norm": 0.7421875,
      "learning_rate": 1.8160430561074006e-05,
      "loss": 0.3128,
      "step": 5224
    },
    {
      "epoch": 0.3954401400122984,
      "grad_norm": 0.78125,
      "learning_rate": 1.8159742069521926e-05,
      "loss": 0.3259,
      "step": 5225
    },
    {
      "epoch": 0.3955158223357457,
      "grad_norm": 0.78125,
      "learning_rate": 1.8159053462209903e-05,
      "loss": 0.3465,
      "step": 5226
    },
    {
      "epoch": 0.395591504659193,
      "grad_norm": 0.90625,
      "learning_rate": 1.8158364739147706e-05,
      "loss": 0.3782,
      "step": 5227
    },
    {
      "epoch": 0.39566718698264036,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8157675900345105e-05,
      "loss": 0.3045,
      "step": 5228
    },
    {
      "epoch": 0.3957428693060877,
      "grad_norm": 0.8125,
      "learning_rate": 1.815698694581187e-05,
      "loss": 0.3513,
      "step": 5229
    },
    {
      "epoch": 0.39581855162953505,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8156297875557777e-05,
      "loss": 0.3106,
      "step": 5230
    },
    {
      "epoch": 0.39589423395298234,
      "grad_norm": 1.0703125,
      "learning_rate": 1.8155608689592604e-05,
      "loss": 0.3497,
      "step": 5231
    },
    {
      "epoch": 0.3959699162764297,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8154919387926124e-05,
      "loss": 0.3515,
      "step": 5232
    },
    {
      "epoch": 0.39604559859987704,
      "grad_norm": 0.81640625,
      "learning_rate": 1.815422997056812e-05,
      "loss": 0.3423,
      "step": 5233
    },
    {
      "epoch": 0.3961212809233243,
      "grad_norm": 0.75,
      "learning_rate": 1.815354043752837e-05,
      "loss": 0.2813,
      "step": 5234
    },
    {
      "epoch": 0.39619696324677167,
      "grad_norm": 0.796875,
      "learning_rate": 1.8152850788816658e-05,
      "loss": 0.3384,
      "step": 5235
    },
    {
      "epoch": 0.396272645570219,
      "grad_norm": 0.7109375,
      "learning_rate": 1.8152161024442768e-05,
      "loss": 0.2772,
      "step": 5236
    },
    {
      "epoch": 0.3963483278936663,
      "grad_norm": 0.78125,
      "learning_rate": 1.815147114441648e-05,
      "loss": 0.3246,
      "step": 5237
    },
    {
      "epoch": 0.39642401021711365,
      "grad_norm": 0.890625,
      "learning_rate": 1.815078114874759e-05,
      "loss": 0.3126,
      "step": 5238
    },
    {
      "epoch": 0.396499692540561,
      "grad_norm": 0.86328125,
      "learning_rate": 1.8150091037445876e-05,
      "loss": 0.3247,
      "step": 5239
    },
    {
      "epoch": 0.39657537486400835,
      "grad_norm": 0.76171875,
      "learning_rate": 1.814940081052114e-05,
      "loss": 0.2831,
      "step": 5240
    },
    {
      "epoch": 0.39665105718745564,
      "grad_norm": 0.828125,
      "learning_rate": 1.8148710467983168e-05,
      "loss": 0.3377,
      "step": 5241
    },
    {
      "epoch": 0.396726739510903,
      "grad_norm": 1.1875,
      "learning_rate": 1.8148020009841755e-05,
      "loss": 0.3389,
      "step": 5242
    },
    {
      "epoch": 0.3968024218343503,
      "grad_norm": 0.84375,
      "learning_rate": 1.814732943610669e-05,
      "loss": 0.3174,
      "step": 5243
    },
    {
      "epoch": 0.3968781041577976,
      "grad_norm": 0.84765625,
      "learning_rate": 1.814663874678778e-05,
      "loss": 0.3715,
      "step": 5244
    },
    {
      "epoch": 0.39695378648124496,
      "grad_norm": 0.86328125,
      "learning_rate": 1.814594794189482e-05,
      "loss": 0.3417,
      "step": 5245
    },
    {
      "epoch": 0.3970294688046923,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8145257021437607e-05,
      "loss": 0.2988,
      "step": 5246
    },
    {
      "epoch": 0.39710515112813966,
      "grad_norm": 0.83203125,
      "learning_rate": 1.814456598542595e-05,
      "loss": 0.3652,
      "step": 5247
    },
    {
      "epoch": 0.39718083345158695,
      "grad_norm": 0.78125,
      "learning_rate": 1.8143874833869645e-05,
      "loss": 0.3062,
      "step": 5248
    },
    {
      "epoch": 0.3972565157750343,
      "grad_norm": 0.8203125,
      "learning_rate": 1.81431835667785e-05,
      "loss": 0.3762,
      "step": 5249
    },
    {
      "epoch": 0.39733219809848164,
      "grad_norm": 0.8515625,
      "learning_rate": 1.8142492184162323e-05,
      "loss": 0.3763,
      "step": 5250
    },
    {
      "epoch": 0.3974078804219289,
      "grad_norm": 0.828125,
      "learning_rate": 1.8141800686030917e-05,
      "loss": 0.368,
      "step": 5251
    },
    {
      "epoch": 0.3974835627453763,
      "grad_norm": 0.87890625,
      "learning_rate": 1.81411090723941e-05,
      "loss": 0.365,
      "step": 5252
    },
    {
      "epoch": 0.3975592450688236,
      "grad_norm": 0.8671875,
      "learning_rate": 1.814041734326168e-05,
      "loss": 0.3644,
      "step": 5253
    },
    {
      "epoch": 0.39763492739227096,
      "grad_norm": 0.75390625,
      "learning_rate": 1.813972549864347e-05,
      "loss": 0.2971,
      "step": 5254
    },
    {
      "epoch": 0.39771060971571826,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8139033538549288e-05,
      "loss": 0.3497,
      "step": 5255
    },
    {
      "epoch": 0.3977862920391656,
      "grad_norm": 0.86328125,
      "learning_rate": 1.8138341462988945e-05,
      "loss": 0.3587,
      "step": 5256
    },
    {
      "epoch": 0.39786197436261295,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8137649271972265e-05,
      "loss": 0.3277,
      "step": 5257
    },
    {
      "epoch": 0.39793765668606024,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8136956965509064e-05,
      "loss": 0.3236,
      "step": 5258
    },
    {
      "epoch": 0.3980133390095076,
      "grad_norm": 0.828125,
      "learning_rate": 1.8136264543609163e-05,
      "loss": 0.3759,
      "step": 5259
    },
    {
      "epoch": 0.39808902133295493,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8135572006282392e-05,
      "loss": 0.3388,
      "step": 5260
    },
    {
      "epoch": 0.3981647036564023,
      "grad_norm": 0.9296875,
      "learning_rate": 1.8134879353538566e-05,
      "loss": 0.3192,
      "step": 5261
    },
    {
      "epoch": 0.39824038597984956,
      "grad_norm": 0.78515625,
      "learning_rate": 1.813418658538752e-05,
      "loss": 0.3418,
      "step": 5262
    },
    {
      "epoch": 0.3983160683032969,
      "grad_norm": 0.765625,
      "learning_rate": 1.813349370183908e-05,
      "loss": 0.318,
      "step": 5263
    },
    {
      "epoch": 0.39839175062674426,
      "grad_norm": 0.875,
      "learning_rate": 1.8132800702903068e-05,
      "loss": 0.3532,
      "step": 5264
    },
    {
      "epoch": 0.39846743295019155,
      "grad_norm": 1.1875,
      "learning_rate": 1.8132107588589325e-05,
      "loss": 0.3508,
      "step": 5265
    },
    {
      "epoch": 0.3985431152736389,
      "grad_norm": 0.7109375,
      "learning_rate": 1.813141435890768e-05,
      "loss": 0.2748,
      "step": 5266
    },
    {
      "epoch": 0.39861879759708624,
      "grad_norm": 0.85546875,
      "learning_rate": 1.813072101386797e-05,
      "loss": 0.3673,
      "step": 5267
    },
    {
      "epoch": 0.3986944799205336,
      "grad_norm": 0.796875,
      "learning_rate": 1.813002755348003e-05,
      "loss": 0.3489,
      "step": 5268
    },
    {
      "epoch": 0.3987701622439809,
      "grad_norm": 0.75390625,
      "learning_rate": 1.8129333977753695e-05,
      "loss": 0.3198,
      "step": 5269
    },
    {
      "epoch": 0.3988458445674282,
      "grad_norm": 0.8359375,
      "learning_rate": 1.812864028669881e-05,
      "loss": 0.3535,
      "step": 5270
    },
    {
      "epoch": 0.39892152689087557,
      "grad_norm": 0.75390625,
      "learning_rate": 1.812794648032521e-05,
      "loss": 0.2865,
      "step": 5271
    },
    {
      "epoch": 0.39899720921432286,
      "grad_norm": 0.78125,
      "learning_rate": 1.8127252558642742e-05,
      "loss": 0.3376,
      "step": 5272
    },
    {
      "epoch": 0.3990728915377702,
      "grad_norm": 0.796875,
      "learning_rate": 1.812655852166125e-05,
      "loss": 0.3323,
      "step": 5273
    },
    {
      "epoch": 0.39914857386121755,
      "grad_norm": 0.84375,
      "learning_rate": 1.812586436939058e-05,
      "loss": 0.4025,
      "step": 5274
    },
    {
      "epoch": 0.3992242561846649,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8125170101840582e-05,
      "loss": 0.3248,
      "step": 5275
    },
    {
      "epoch": 0.3992999385081122,
      "grad_norm": 0.8515625,
      "learning_rate": 1.81244757190211e-05,
      "loss": 0.3579,
      "step": 5276
    },
    {
      "epoch": 0.39937562083155953,
      "grad_norm": 0.83203125,
      "learning_rate": 1.8123781220941987e-05,
      "loss": 0.3558,
      "step": 5277
    },
    {
      "epoch": 0.3994513031550069,
      "grad_norm": 0.79296875,
      "learning_rate": 1.81230866076131e-05,
      "loss": 0.3231,
      "step": 5278
    },
    {
      "epoch": 0.39952698547845417,
      "grad_norm": 0.89453125,
      "learning_rate": 1.8122391879044284e-05,
      "loss": 0.3889,
      "step": 5279
    },
    {
      "epoch": 0.3996026678019015,
      "grad_norm": 0.79296875,
      "learning_rate": 1.812169703524541e-05,
      "loss": 0.3378,
      "step": 5280
    },
    {
      "epoch": 0.39967835012534886,
      "grad_norm": 0.78125,
      "learning_rate": 1.8121002076226318e-05,
      "loss": 0.3206,
      "step": 5281
    },
    {
      "epoch": 0.3997540324487962,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8120307001996876e-05,
      "loss": 0.315,
      "step": 5282
    },
    {
      "epoch": 0.3998297147722435,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8119611812566948e-05,
      "loss": 0.3658,
      "step": 5283
    },
    {
      "epoch": 0.39990539709569084,
      "grad_norm": 0.75390625,
      "learning_rate": 1.8118916507946387e-05,
      "loss": 0.3256,
      "step": 5284
    },
    {
      "epoch": 0.3999810794191382,
      "grad_norm": 0.78125,
      "learning_rate": 1.811822108814507e-05,
      "loss": 0.3228,
      "step": 5285
    },
    {
      "epoch": 0.4000567617425855,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8117525553172853e-05,
      "loss": 0.3476,
      "step": 5286
    },
    {
      "epoch": 0.4001324440660328,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8116829903039604e-05,
      "loss": 0.3263,
      "step": 5287
    },
    {
      "epoch": 0.40020812638948017,
      "grad_norm": 0.765625,
      "learning_rate": 1.8116134137755198e-05,
      "loss": 0.3392,
      "step": 5288
    },
    {
      "epoch": 0.40020812638948017,
      "eval_loss": 0.34545648097991943,
      "eval_runtime": 83.3962,
      "eval_samples_per_second": 58.3,
      "eval_steps_per_second": 58.3,
      "step": 5288
    },
    {
      "epoch": 0.4002838087129275,
      "grad_norm": 1.3203125,
      "learning_rate": 1.8115438257329497e-05,
      "loss": 0.4362,
      "step": 5289
    },
    {
      "epoch": 0.4003594910363748,
      "grad_norm": 0.8125,
      "learning_rate": 1.8114742261772377e-05,
      "loss": 0.3421,
      "step": 5290
    },
    {
      "epoch": 0.40043517335982215,
      "grad_norm": 0.6875,
      "learning_rate": 1.8114046151093716e-05,
      "loss": 0.2691,
      "step": 5291
    },
    {
      "epoch": 0.4005108556832695,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8113349925303385e-05,
      "loss": 0.3077,
      "step": 5292
    },
    {
      "epoch": 0.4005865380067168,
      "grad_norm": 0.84765625,
      "learning_rate": 1.811265358441126e-05,
      "loss": 0.3154,
      "step": 5293
    },
    {
      "epoch": 0.40066222033016413,
      "grad_norm": 0.72265625,
      "learning_rate": 1.8111957128427228e-05,
      "loss": 0.2972,
      "step": 5294
    },
    {
      "epoch": 0.4007379026536115,
      "grad_norm": 0.72265625,
      "learning_rate": 1.811126055736116e-05,
      "loss": 0.2866,
      "step": 5295
    },
    {
      "epoch": 0.4008135849770588,
      "grad_norm": 0.75390625,
      "learning_rate": 1.8110563871222942e-05,
      "loss": 0.321,
      "step": 5296
    },
    {
      "epoch": 0.4008892673005061,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8109867070022456e-05,
      "loss": 0.3366,
      "step": 5297
    },
    {
      "epoch": 0.40096494962395346,
      "grad_norm": 0.82421875,
      "learning_rate": 1.810917015376959e-05,
      "loss": 0.3448,
      "step": 5298
    },
    {
      "epoch": 0.4010406319474008,
      "grad_norm": 0.9765625,
      "learning_rate": 1.8108473122474226e-05,
      "loss": 0.3328,
      "step": 5299
    },
    {
      "epoch": 0.4011163142708481,
      "grad_norm": 0.75390625,
      "learning_rate": 1.810777597614626e-05,
      "loss": 0.3017,
      "step": 5300
    },
    {
      "epoch": 0.40119199659429544,
      "grad_norm": 0.74609375,
      "learning_rate": 1.8107078714795578e-05,
      "loss": 0.3126,
      "step": 5301
    },
    {
      "epoch": 0.4012676789177428,
      "grad_norm": 0.765625,
      "learning_rate": 1.8106381338432073e-05,
      "loss": 0.3254,
      "step": 5302
    },
    {
      "epoch": 0.4013433612411901,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8105683847065635e-05,
      "loss": 0.3242,
      "step": 5303
    },
    {
      "epoch": 0.4014190435646374,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8104986240706162e-05,
      "loss": 0.3458,
      "step": 5304
    },
    {
      "epoch": 0.40149472588808477,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8104288519363553e-05,
      "loss": 0.3106,
      "step": 5305
    },
    {
      "epoch": 0.4015704082115321,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8103590683047704e-05,
      "loss": 0.3195,
      "step": 5306
    },
    {
      "epoch": 0.4016460905349794,
      "grad_norm": 0.7421875,
      "learning_rate": 1.810289273176851e-05,
      "loss": 0.2983,
      "step": 5307
    },
    {
      "epoch": 0.40172177285842675,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8102194665535885e-05,
      "loss": 0.3447,
      "step": 5308
    },
    {
      "epoch": 0.4017974551818741,
      "grad_norm": 0.78125,
      "learning_rate": 1.8101496484359716e-05,
      "loss": 0.325,
      "step": 5309
    },
    {
      "epoch": 0.4018731375053214,
      "grad_norm": 0.91796875,
      "learning_rate": 1.8100798188249924e-05,
      "loss": 0.3986,
      "step": 5310
    },
    {
      "epoch": 0.40194881982876873,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8100099777216406e-05,
      "loss": 0.312,
      "step": 5311
    },
    {
      "epoch": 0.4020245021522161,
      "grad_norm": 0.83203125,
      "learning_rate": 1.8099401251269072e-05,
      "loss": 0.3644,
      "step": 5312
    },
    {
      "epoch": 0.4021001844756634,
      "grad_norm": 0.796875,
      "learning_rate": 1.8098702610417833e-05,
      "loss": 0.3126,
      "step": 5313
    },
    {
      "epoch": 0.4021758667991107,
      "grad_norm": 0.796875,
      "learning_rate": 1.8098003854672602e-05,
      "loss": 0.3381,
      "step": 5314
    },
    {
      "epoch": 0.40225154912255806,
      "grad_norm": 0.78125,
      "learning_rate": 1.8097304984043286e-05,
      "loss": 0.3102,
      "step": 5315
    },
    {
      "epoch": 0.4023272314460054,
      "grad_norm": 0.6953125,
      "learning_rate": 1.8096605998539805e-05,
      "loss": 0.2756,
      "step": 5316
    },
    {
      "epoch": 0.4024029137694527,
      "grad_norm": 0.7578125,
      "learning_rate": 1.809590689817207e-05,
      "loss": 0.3159,
      "step": 5317
    },
    {
      "epoch": 0.40247859609290004,
      "grad_norm": 0.75390625,
      "learning_rate": 1.8095207682950006e-05,
      "loss": 0.3213,
      "step": 5318
    },
    {
      "epoch": 0.4025542784163474,
      "grad_norm": 0.796875,
      "learning_rate": 1.8094508352883527e-05,
      "loss": 0.3322,
      "step": 5319
    },
    {
      "epoch": 0.40262996073979473,
      "grad_norm": 0.76953125,
      "learning_rate": 1.809380890798256e-05,
      "loss": 0.3229,
      "step": 5320
    },
    {
      "epoch": 0.402705643063242,
      "grad_norm": 0.75,
      "learning_rate": 1.8093109348257023e-05,
      "loss": 0.2803,
      "step": 5321
    },
    {
      "epoch": 0.40278132538668937,
      "grad_norm": 0.75390625,
      "learning_rate": 1.809240967371684e-05,
      "loss": 0.2976,
      "step": 5322
    },
    {
      "epoch": 0.4028570077101367,
      "grad_norm": 0.85546875,
      "learning_rate": 1.809170988437194e-05,
      "loss": 0.3758,
      "step": 5323
    },
    {
      "epoch": 0.402932690033584,
      "grad_norm": 0.76953125,
      "learning_rate": 1.8091009980232247e-05,
      "loss": 0.3266,
      "step": 5324
    },
    {
      "epoch": 0.40300837235703135,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8090309961307693e-05,
      "loss": 0.327,
      "step": 5325
    },
    {
      "epoch": 0.4030840546804787,
      "grad_norm": 0.78125,
      "learning_rate": 1.8089609827608213e-05,
      "loss": 0.3513,
      "step": 5326
    },
    {
      "epoch": 0.40315973700392604,
      "grad_norm": 0.76171875,
      "learning_rate": 1.808890957914373e-05,
      "loss": 0.2972,
      "step": 5327
    },
    {
      "epoch": 0.40323541932737333,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8088209215924187e-05,
      "loss": 0.3394,
      "step": 5328
    },
    {
      "epoch": 0.4033111016508207,
      "grad_norm": 0.8046875,
      "learning_rate": 1.808750873795952e-05,
      "loss": 0.3381,
      "step": 5329
    },
    {
      "epoch": 0.403386783974268,
      "grad_norm": 0.79296875,
      "learning_rate": 1.808680814525966e-05,
      "loss": 0.342,
      "step": 5330
    },
    {
      "epoch": 0.4034624662977153,
      "grad_norm": 0.84375,
      "learning_rate": 1.8086107437834544e-05,
      "loss": 0.3257,
      "step": 5331
    },
    {
      "epoch": 0.40353814862116266,
      "grad_norm": 0.8671875,
      "learning_rate": 1.8085406615694122e-05,
      "loss": 0.3878,
      "step": 5332
    },
    {
      "epoch": 0.40361383094461,
      "grad_norm": 0.796875,
      "learning_rate": 1.8084705678848332e-05,
      "loss": 0.3236,
      "step": 5333
    },
    {
      "epoch": 0.40368951326805735,
      "grad_norm": 0.875,
      "learning_rate": 1.8084004627307118e-05,
      "loss": 0.3448,
      "step": 5334
    },
    {
      "epoch": 0.40376519559150464,
      "grad_norm": 0.828125,
      "learning_rate": 1.8083303461080425e-05,
      "loss": 0.3189,
      "step": 5335
    },
    {
      "epoch": 0.403840877914952,
      "grad_norm": 1.5078125,
      "learning_rate": 1.80826021801782e-05,
      "loss": 0.418,
      "step": 5336
    },
    {
      "epoch": 0.40391656023839934,
      "grad_norm": 0.74609375,
      "learning_rate": 1.8081900784610394e-05,
      "loss": 0.3023,
      "step": 5337
    },
    {
      "epoch": 0.4039922425618466,
      "grad_norm": 0.8828125,
      "learning_rate": 1.8081199274386958e-05,
      "loss": 0.3383,
      "step": 5338
    },
    {
      "epoch": 0.40406792488529397,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8080497649517843e-05,
      "loss": 0.3271,
      "step": 5339
    },
    {
      "epoch": 0.4041436072087413,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8079795910013e-05,
      "loss": 0.3014,
      "step": 5340
    },
    {
      "epoch": 0.40421928953218866,
      "grad_norm": 0.85546875,
      "learning_rate": 1.8079094055882387e-05,
      "loss": 0.3583,
      "step": 5341
    },
    {
      "epoch": 0.40429497185563595,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8078392087135957e-05,
      "loss": 0.3141,
      "step": 5342
    },
    {
      "epoch": 0.4043706541790833,
      "grad_norm": 1.8359375,
      "learning_rate": 1.807769000378368e-05,
      "loss": 0.3711,
      "step": 5343
    },
    {
      "epoch": 0.40444633650253065,
      "grad_norm": 0.765625,
      "learning_rate": 1.8076987805835502e-05,
      "loss": 0.2613,
      "step": 5344
    },
    {
      "epoch": 0.40452201882597794,
      "grad_norm": 0.796875,
      "learning_rate": 1.8076285493301394e-05,
      "loss": 0.3196,
      "step": 5345
    },
    {
      "epoch": 0.4045977011494253,
      "grad_norm": 0.78515625,
      "learning_rate": 1.807558306619132e-05,
      "loss": 0.313,
      "step": 5346
    },
    {
      "epoch": 0.4046733834728726,
      "grad_norm": 0.83984375,
      "learning_rate": 1.8074880524515236e-05,
      "loss": 0.3545,
      "step": 5347
    },
    {
      "epoch": 0.40474906579632,
      "grad_norm": 0.84375,
      "learning_rate": 1.8074177868283114e-05,
      "loss": 0.3482,
      "step": 5348
    },
    {
      "epoch": 0.40482474811976726,
      "grad_norm": 0.7890625,
      "learning_rate": 1.807347509750493e-05,
      "loss": 0.3455,
      "step": 5349
    },
    {
      "epoch": 0.4049004304432146,
      "grad_norm": 0.765625,
      "learning_rate": 1.8072772212190645e-05,
      "loss": 0.3178,
      "step": 5350
    },
    {
      "epoch": 0.40497611276666196,
      "grad_norm": 0.78515625,
      "learning_rate": 1.807206921235023e-05,
      "loss": 0.3212,
      "step": 5351
    },
    {
      "epoch": 0.40505179509010925,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8071366097993663e-05,
      "loss": 0.3045,
      "step": 5352
    },
    {
      "epoch": 0.4051274774135566,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8070662869130914e-05,
      "loss": 0.3288,
      "step": 5353
    },
    {
      "epoch": 0.40520315973700394,
      "grad_norm": 0.7734375,
      "learning_rate": 1.806995952577197e-05,
      "loss": 0.3117,
      "step": 5354
    },
    {
      "epoch": 0.4052788420604513,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8069256067926794e-05,
      "loss": 0.3391,
      "step": 5355
    },
    {
      "epoch": 0.4053545243838986,
      "grad_norm": 0.87109375,
      "learning_rate": 1.8068552495605375e-05,
      "loss": 0.3316,
      "step": 5356
    },
    {
      "epoch": 0.4054302067073459,
      "grad_norm": 0.8359375,
      "learning_rate": 1.8067848808817695e-05,
      "loss": 0.3179,
      "step": 5357
    },
    {
      "epoch": 0.40550588903079327,
      "grad_norm": 0.84765625,
      "learning_rate": 1.8067145007573733e-05,
      "loss": 0.3404,
      "step": 5358
    },
    {
      "epoch": 0.40558157135424056,
      "grad_norm": 0.87109375,
      "learning_rate": 1.8066441091883477e-05,
      "loss": 0.3429,
      "step": 5359
    },
    {
      "epoch": 0.4056572536776879,
      "grad_norm": 0.85546875,
      "learning_rate": 1.806573706175691e-05,
      "loss": 0.3292,
      "step": 5360
    },
    {
      "epoch": 0.40573293600113525,
      "grad_norm": 0.83984375,
      "learning_rate": 1.8065032917204025e-05,
      "loss": 0.3563,
      "step": 5361
    },
    {
      "epoch": 0.40580861832458254,
      "grad_norm": 0.79296875,
      "learning_rate": 1.80643286582348e-05,
      "loss": 0.3068,
      "step": 5362
    },
    {
      "epoch": 0.4058843006480299,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8063624284859244e-05,
      "loss": 0.3402,
      "step": 5363
    },
    {
      "epoch": 0.40595998297147723,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8062919797087333e-05,
      "loss": 0.365,
      "step": 5364
    },
    {
      "epoch": 0.4060356652949246,
      "grad_norm": 0.78515625,
      "learning_rate": 1.806221519492907e-05,
      "loss": 0.3334,
      "step": 5365
    },
    {
      "epoch": 0.40611134761837187,
      "grad_norm": 0.78125,
      "learning_rate": 1.8061510478394448e-05,
      "loss": 0.3126,
      "step": 5366
    },
    {
      "epoch": 0.4061870299418192,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8060805647493468e-05,
      "loss": 0.3367,
      "step": 5367
    },
    {
      "epoch": 0.40626271226526656,
      "grad_norm": 0.796875,
      "learning_rate": 1.8060100702236128e-05,
      "loss": 0.3064,
      "step": 5368
    },
    {
      "epoch": 0.40633839458871385,
      "grad_norm": 0.87109375,
      "learning_rate": 1.8059395642632423e-05,
      "loss": 0.3535,
      "step": 5369
    },
    {
      "epoch": 0.4064140769121612,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8058690468692366e-05,
      "loss": 0.3159,
      "step": 5370
    },
    {
      "epoch": 0.40648975923560854,
      "grad_norm": 1.0859375,
      "learning_rate": 1.8057985180425954e-05,
      "loss": 0.3974,
      "step": 5371
    },
    {
      "epoch": 0.4065654415590559,
      "grad_norm": 0.78125,
      "learning_rate": 1.8057279777843193e-05,
      "loss": 0.3345,
      "step": 5372
    },
    {
      "epoch": 0.4066411238825032,
      "grad_norm": 0.8515625,
      "learning_rate": 1.8056574260954095e-05,
      "loss": 0.3478,
      "step": 5373
    },
    {
      "epoch": 0.4067168062059505,
      "grad_norm": 0.86328125,
      "learning_rate": 1.805586862976866e-05,
      "loss": 0.3462,
      "step": 5374
    },
    {
      "epoch": 0.40679248852939787,
      "grad_norm": 0.76953125,
      "learning_rate": 1.8055162884296907e-05,
      "loss": 0.3,
      "step": 5375
    },
    {
      "epoch": 0.40686817085284516,
      "grad_norm": 0.82421875,
      "learning_rate": 1.8054457024548845e-05,
      "loss": 0.3308,
      "step": 5376
    },
    {
      "epoch": 0.4069438531762925,
      "grad_norm": 0.80078125,
      "learning_rate": 1.805375105053449e-05,
      "loss": 0.3398,
      "step": 5377
    },
    {
      "epoch": 0.40701953549973985,
      "grad_norm": 0.69140625,
      "learning_rate": 1.805304496226385e-05,
      "loss": 0.2717,
      "step": 5378
    },
    {
      "epoch": 0.4070952178231872,
      "grad_norm": 0.7578125,
      "learning_rate": 1.8052338759746955e-05,
      "loss": 0.314,
      "step": 5379
    },
    {
      "epoch": 0.4071709001466345,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8051632442993813e-05,
      "loss": 0.3284,
      "step": 5380
    },
    {
      "epoch": 0.40724658247008183,
      "grad_norm": 0.796875,
      "learning_rate": 1.8050926012014445e-05,
      "loss": 0.2884,
      "step": 5381
    },
    {
      "epoch": 0.4073222647935292,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8050219466818882e-05,
      "loss": 0.3745,
      "step": 5382
    },
    {
      "epoch": 0.40739794711697647,
      "grad_norm": 0.8515625,
      "learning_rate": 1.8049512807417135e-05,
      "loss": 0.3659,
      "step": 5383
    },
    {
      "epoch": 0.4074736294404238,
      "grad_norm": 0.8671875,
      "learning_rate": 1.804880603381924e-05,
      "loss": 0.407,
      "step": 5384
    },
    {
      "epoch": 0.40754931176387116,
      "grad_norm": 0.921875,
      "learning_rate": 1.8048099146035217e-05,
      "loss": 0.3937,
      "step": 5385
    },
    {
      "epoch": 0.4076249940873185,
      "grad_norm": 0.76171875,
      "learning_rate": 1.8047392144075094e-05,
      "loss": 0.2999,
      "step": 5386
    },
    {
      "epoch": 0.4077006764107658,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8046685027948906e-05,
      "loss": 0.3448,
      "step": 5387
    },
    {
      "epoch": 0.40777635873421314,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8045977797666685e-05,
      "loss": 0.3007,
      "step": 5388
    },
    {
      "epoch": 0.4078520410576605,
      "grad_norm": 0.8125,
      "learning_rate": 1.804527045323846e-05,
      "loss": 0.3686,
      "step": 5389
    },
    {
      "epoch": 0.4079277233811078,
      "grad_norm": 0.84765625,
      "learning_rate": 1.8044562994674266e-05,
      "loss": 0.3498,
      "step": 5390
    },
    {
      "epoch": 0.4080034057045551,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8043855421984142e-05,
      "loss": 0.3,
      "step": 5391
    },
    {
      "epoch": 0.40807908802800247,
      "grad_norm": 0.71484375,
      "learning_rate": 1.8043147735178125e-05,
      "loss": 0.2931,
      "step": 5392
    },
    {
      "epoch": 0.4081547703514498,
      "grad_norm": 0.8515625,
      "learning_rate": 1.8042439934266253e-05,
      "loss": 0.3526,
      "step": 5393
    },
    {
      "epoch": 0.4082304526748971,
      "grad_norm": 0.83984375,
      "learning_rate": 1.8041732019258573e-05,
      "loss": 0.3445,
      "step": 5394
    },
    {
      "epoch": 0.40830613499834445,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8041023990165122e-05,
      "loss": 0.3116,
      "step": 5395
    },
    {
      "epoch": 0.4083818173217918,
      "grad_norm": 0.7421875,
      "learning_rate": 1.8040315846995947e-05,
      "loss": 0.2911,
      "step": 5396
    },
    {
      "epoch": 0.4084574996452391,
      "grad_norm": 0.7421875,
      "learning_rate": 1.803960758976109e-05,
      "loss": 0.2896,
      "step": 5397
    },
    {
      "epoch": 0.40853318196868643,
      "grad_norm": 0.77734375,
      "learning_rate": 1.803889921847061e-05,
      "loss": 0.3198,
      "step": 5398
    },
    {
      "epoch": 0.4086088642921338,
      "grad_norm": 0.81640625,
      "learning_rate": 1.803819073313455e-05,
      "loss": 0.3453,
      "step": 5399
    },
    {
      "epoch": 0.4086845466155811,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8037482133762954e-05,
      "loss": 0.3503,
      "step": 5400
    },
    {
      "epoch": 0.4087602289390284,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8036773420365886e-05,
      "loss": 0.3345,
      "step": 5401
    },
    {
      "epoch": 0.40883591126247576,
      "grad_norm": 0.76171875,
      "learning_rate": 1.803606459295339e-05,
      "loss": 0.2659,
      "step": 5402
    },
    {
      "epoch": 0.4089115935859231,
      "grad_norm": 0.7734375,
      "learning_rate": 1.803535565153553e-05,
      "loss": 0.3236,
      "step": 5403
    },
    {
      "epoch": 0.4089872759093704,
      "grad_norm": 1.765625,
      "learning_rate": 1.8034646596122365e-05,
      "loss": 0.34,
      "step": 5404
    },
    {
      "epoch": 0.40906295823281774,
      "grad_norm": 0.83203125,
      "learning_rate": 1.8033937426723945e-05,
      "loss": 0.3493,
      "step": 5405
    },
    {
      "epoch": 0.4091386405562651,
      "grad_norm": 0.7421875,
      "learning_rate": 1.803322814335034e-05,
      "loss": 0.3176,
      "step": 5406
    },
    {
      "epoch": 0.40921432287971243,
      "grad_norm": 0.75390625,
      "learning_rate": 1.8032518746011603e-05,
      "loss": 0.304,
      "step": 5407
    },
    {
      "epoch": 0.4092900052031597,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8031809234717807e-05,
      "loss": 0.3098,
      "step": 5408
    },
    {
      "epoch": 0.40936568752660707,
      "grad_norm": 0.8046875,
      "learning_rate": 1.803109960947901e-05,
      "loss": 0.3783,
      "step": 5409
    },
    {
      "epoch": 0.4094413698500544,
      "grad_norm": 0.734375,
      "learning_rate": 1.803038987030529e-05,
      "loss": 0.2981,
      "step": 5410
    },
    {
      "epoch": 0.4095170521735017,
      "grad_norm": 1.0390625,
      "learning_rate": 1.8029680017206705e-05,
      "loss": 0.3451,
      "step": 5411
    },
    {
      "epoch": 0.40959273449694905,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8028970050193327e-05,
      "loss": 0.3185,
      "step": 5412
    },
    {
      "epoch": 0.4096684168203964,
      "grad_norm": 0.81640625,
      "learning_rate": 1.8028259969275237e-05,
      "loss": 0.3413,
      "step": 5413
    },
    {
      "epoch": 0.40974409914384374,
      "grad_norm": 0.8203125,
      "learning_rate": 1.80275497744625e-05,
      "loss": 0.3418,
      "step": 5414
    },
    {
      "epoch": 0.40981978146729103,
      "grad_norm": 0.89453125,
      "learning_rate": 1.802683946576519e-05,
      "loss": 0.2789,
      "step": 5415
    },
    {
      "epoch": 0.4098954637907384,
      "grad_norm": 0.78515625,
      "learning_rate": 1.802612904319339e-05,
      "loss": 0.2973,
      "step": 5416
    },
    {
      "epoch": 0.4099711461141857,
      "grad_norm": 0.78125,
      "learning_rate": 1.802541850675718e-05,
      "loss": 0.3281,
      "step": 5417
    },
    {
      "epoch": 0.410046828437633,
      "grad_norm": 0.78125,
      "learning_rate": 1.802470785646663e-05,
      "loss": 0.3316,
      "step": 5418
    },
    {
      "epoch": 0.41012251076108036,
      "grad_norm": 0.86328125,
      "learning_rate": 1.8023997092331833e-05,
      "loss": 0.3595,
      "step": 5419
    },
    {
      "epoch": 0.4101981930845277,
      "grad_norm": 0.76171875,
      "learning_rate": 1.802328621436287e-05,
      "loss": 0.3112,
      "step": 5420
    },
    {
      "epoch": 0.410273875407975,
      "grad_norm": 0.76171875,
      "learning_rate": 1.8022575222569814e-05,
      "loss": 0.2942,
      "step": 5421
    },
    {
      "epoch": 0.41034955773142234,
      "grad_norm": 0.7265625,
      "learning_rate": 1.8021864116962767e-05,
      "loss": 0.2726,
      "step": 5422
    },
    {
      "epoch": 0.4104252400548697,
      "grad_norm": 0.7421875,
      "learning_rate": 1.8021152897551813e-05,
      "loss": 0.2647,
      "step": 5423
    },
    {
      "epoch": 0.41050092237831703,
      "grad_norm": 0.7734375,
      "learning_rate": 1.8020441564347038e-05,
      "loss": 0.3196,
      "step": 5424
    },
    {
      "epoch": 0.4105766047017643,
      "grad_norm": 0.890625,
      "learning_rate": 1.8019730117358538e-05,
      "loss": 0.3856,
      "step": 5425
    },
    {
      "epoch": 0.41065228702521167,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8019018556596402e-05,
      "loss": 0.3341,
      "step": 5426
    },
    {
      "epoch": 0.410727969348659,
      "grad_norm": 0.796875,
      "learning_rate": 1.8018306882070726e-05,
      "loss": 0.338,
      "step": 5427
    },
    {
      "epoch": 0.4108036516721063,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8017595093791607e-05,
      "loss": 0.3304,
      "step": 5428
    },
    {
      "epoch": 0.41087933399555365,
      "grad_norm": 0.80859375,
      "learning_rate": 1.8016883191769144e-05,
      "loss": 0.3292,
      "step": 5429
    },
    {
      "epoch": 0.410955016319001,
      "grad_norm": 0.8046875,
      "learning_rate": 1.8016171176013436e-05,
      "loss": 0.346,
      "step": 5430
    },
    {
      "epoch": 0.41103069864244834,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8015459046534582e-05,
      "loss": 0.3334,
      "step": 5431
    },
    {
      "epoch": 0.41110638096589563,
      "grad_norm": 0.796875,
      "learning_rate": 1.8014746803342688e-05,
      "loss": 0.3435,
      "step": 5432
    },
    {
      "epoch": 0.411182063289343,
      "grad_norm": 0.73046875,
      "learning_rate": 1.8014034446447858e-05,
      "loss": 0.254,
      "step": 5433
    },
    {
      "epoch": 0.4112577456127903,
      "grad_norm": 0.99609375,
      "learning_rate": 1.8013321975860194e-05,
      "loss": 0.3834,
      "step": 5434
    },
    {
      "epoch": 0.4113334279362376,
      "grad_norm": 0.78125,
      "learning_rate": 1.8012609391589805e-05,
      "loss": 0.3217,
      "step": 5435
    },
    {
      "epoch": 0.41140911025968496,
      "grad_norm": 1.3828125,
      "learning_rate": 1.8011896693646805e-05,
      "loss": 0.4139,
      "step": 5436
    },
    {
      "epoch": 0.4114847925831323,
      "grad_norm": 0.7890625,
      "learning_rate": 1.80111838820413e-05,
      "loss": 0.3251,
      "step": 5437
    },
    {
      "epoch": 0.41156047490657965,
      "grad_norm": 0.79296875,
      "learning_rate": 1.8010470956783407e-05,
      "loss": 0.3143,
      "step": 5438
    },
    {
      "epoch": 0.41163615723002694,
      "grad_norm": 0.734375,
      "learning_rate": 1.8009757917883236e-05,
      "loss": 0.2666,
      "step": 5439
    },
    {
      "epoch": 0.4117118395534743,
      "grad_norm": 0.88671875,
      "learning_rate": 1.8009044765350904e-05,
      "loss": 0.3552,
      "step": 5440
    },
    {
      "epoch": 0.41178752187692164,
      "grad_norm": 0.80078125,
      "learning_rate": 1.8008331499196523e-05,
      "loss": 0.3371,
      "step": 5441
    },
    {
      "epoch": 0.4118632042003689,
      "grad_norm": 0.77734375,
      "learning_rate": 1.8007618119430222e-05,
      "loss": 0.3365,
      "step": 5442
    },
    {
      "epoch": 0.4119388865238163,
      "grad_norm": 0.7890625,
      "learning_rate": 1.8006904626062115e-05,
      "loss": 0.3274,
      "step": 5443
    },
    {
      "epoch": 0.4120145688472636,
      "grad_norm": 0.90625,
      "learning_rate": 1.8006191019102327e-05,
      "loss": 0.408,
      "step": 5444
    },
    {
      "epoch": 0.41209025117071096,
      "grad_norm": 0.78515625,
      "learning_rate": 1.8005477298560977e-05,
      "loss": 0.3031,
      "step": 5445
    },
    {
      "epoch": 0.41216593349415825,
      "grad_norm": 1.1953125,
      "learning_rate": 1.8004763464448195e-05,
      "loss": 0.2946,
      "step": 5446
    },
    {
      "epoch": 0.4122416158176056,
      "grad_norm": 0.8828125,
      "learning_rate": 1.8004049516774106e-05,
      "loss": 0.3707,
      "step": 5447
    },
    {
      "epoch": 0.41231729814105295,
      "grad_norm": 0.79296875,
      "learning_rate": 1.800333545554884e-05,
      "loss": 0.3518,
      "step": 5448
    },
    {
      "epoch": 0.41239298046450024,
      "grad_norm": 0.83203125,
      "learning_rate": 1.8002621280782525e-05,
      "loss": 0.3511,
      "step": 5449
    },
    {
      "epoch": 0.4124686627879476,
      "grad_norm": 0.71484375,
      "learning_rate": 1.8001906992485297e-05,
      "loss": 0.2846,
      "step": 5450
    },
    {
      "epoch": 0.41254434511139493,
      "grad_norm": 0.8203125,
      "learning_rate": 1.8001192590667287e-05,
      "loss": 0.3456,
      "step": 5451
    },
    {
      "epoch": 0.4126200274348423,
      "grad_norm": 0.7890625,
      "learning_rate": 1.800047807533863e-05,
      "loss": 0.3457,
      "step": 5452
    },
    {
      "epoch": 0.41269570975828956,
      "grad_norm": 0.84765625,
      "learning_rate": 1.7999763446509456e-05,
      "loss": 0.3639,
      "step": 5453
    },
    {
      "epoch": 0.4127713920817369,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7999048704189914e-05,
      "loss": 0.3469,
      "step": 5454
    },
    {
      "epoch": 0.41284707440518426,
      "grad_norm": 1.171875,
      "learning_rate": 1.799833384839014e-05,
      "loss": 0.3742,
      "step": 5455
    },
    {
      "epoch": 0.41292275672863155,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7997618879120274e-05,
      "loss": 0.3368,
      "step": 5456
    },
    {
      "epoch": 0.4129984390520789,
      "grad_norm": 0.75,
      "learning_rate": 1.7996903796390455e-05,
      "loss": 0.3103,
      "step": 5457
    },
    {
      "epoch": 0.41307412137552624,
      "grad_norm": 0.69140625,
      "learning_rate": 1.799618860021084e-05,
      "loss": 0.2652,
      "step": 5458
    },
    {
      "epoch": 0.4131498036989736,
      "grad_norm": 0.75390625,
      "learning_rate": 1.7995473290591566e-05,
      "loss": 0.3042,
      "step": 5459
    },
    {
      "epoch": 0.4132254860224209,
      "grad_norm": 0.765625,
      "learning_rate": 1.799475786754278e-05,
      "loss": 0.3165,
      "step": 5460
    },
    {
      "epoch": 0.4133011683458682,
      "grad_norm": 0.83984375,
      "learning_rate": 1.7994042331074634e-05,
      "loss": 0.319,
      "step": 5461
    },
    {
      "epoch": 0.41337685066931557,
      "grad_norm": 0.8359375,
      "learning_rate": 1.7993326681197283e-05,
      "loss": 0.3547,
      "step": 5462
    },
    {
      "epoch": 0.41345253299276286,
      "grad_norm": 0.8046875,
      "learning_rate": 1.7992610917920876e-05,
      "loss": 0.3363,
      "step": 5463
    },
    {
      "epoch": 0.4135282153162102,
      "grad_norm": 0.86328125,
      "learning_rate": 1.7991895041255562e-05,
      "loss": 0.3573,
      "step": 5464
    },
    {
      "epoch": 0.41360389763965755,
      "grad_norm": 0.796875,
      "learning_rate": 1.7991179051211504e-05,
      "loss": 0.3611,
      "step": 5465
    },
    {
      "epoch": 0.4136795799631049,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7990462947798863e-05,
      "loss": 0.3361,
      "step": 5466
    },
    {
      "epoch": 0.4137552622865522,
      "grad_norm": 0.8203125,
      "learning_rate": 1.7989746731027788e-05,
      "loss": 0.3714,
      "step": 5467
    },
    {
      "epoch": 0.41383094460999953,
      "grad_norm": 0.80859375,
      "learning_rate": 1.798903040090844e-05,
      "loss": 0.3227,
      "step": 5468
    },
    {
      "epoch": 0.4139066269334469,
      "grad_norm": 0.9296875,
      "learning_rate": 1.7988313957450998e-05,
      "loss": 0.3081,
      "step": 5469
    },
    {
      "epoch": 0.41398230925689417,
      "grad_norm": 0.80078125,
      "learning_rate": 1.7987597400665607e-05,
      "loss": 0.3264,
      "step": 5470
    },
    {
      "epoch": 0.4140579915803415,
      "grad_norm": 0.82421875,
      "learning_rate": 1.798688073056244e-05,
      "loss": 0.3207,
      "step": 5471
    },
    {
      "epoch": 0.41413367390378886,
      "grad_norm": 0.828125,
      "learning_rate": 1.7986163947151665e-05,
      "loss": 0.3235,
      "step": 5472
    },
    {
      "epoch": 0.4142093562272362,
      "grad_norm": 0.71875,
      "learning_rate": 1.798544705044345e-05,
      "loss": 0.295,
      "step": 5473
    },
    {
      "epoch": 0.4142850385506835,
      "grad_norm": 0.7421875,
      "learning_rate": 1.7984730040447964e-05,
      "loss": 0.2929,
      "step": 5474
    },
    {
      "epoch": 0.41436072087413084,
      "grad_norm": 0.703125,
      "learning_rate": 1.798401291717538e-05,
      "loss": 0.2703,
      "step": 5475
    },
    {
      "epoch": 0.4144364031975782,
      "grad_norm": 0.796875,
      "learning_rate": 1.7983295680635873e-05,
      "loss": 0.3351,
      "step": 5476
    },
    {
      "epoch": 0.4145120855210255,
      "grad_norm": 0.859375,
      "learning_rate": 1.7982578330839617e-05,
      "loss": 0.386,
      "step": 5477
    },
    {
      "epoch": 0.4145877678444728,
      "grad_norm": 0.75,
      "learning_rate": 1.7981860867796786e-05,
      "loss": 0.3209,
      "step": 5478
    },
    {
      "epoch": 0.41466345016792017,
      "grad_norm": 1.6953125,
      "learning_rate": 1.7981143291517564e-05,
      "loss": 0.3573,
      "step": 5479
    },
    {
      "epoch": 0.4147391324913675,
      "grad_norm": 0.84765625,
      "learning_rate": 1.798042560201213e-05,
      "loss": 0.3475,
      "step": 5480
    },
    {
      "epoch": 0.4148148148148148,
      "grad_norm": 0.796875,
      "learning_rate": 1.797970779929066e-05,
      "loss": 0.3282,
      "step": 5481
    },
    {
      "epoch": 0.41489049713826215,
      "grad_norm": 0.8359375,
      "learning_rate": 1.7978989883363344e-05,
      "loss": 0.3507,
      "step": 5482
    },
    {
      "epoch": 0.4149661794617095,
      "grad_norm": 0.75,
      "learning_rate": 1.7978271854240362e-05,
      "loss": 0.3175,
      "step": 5483
    },
    {
      "epoch": 0.4150418617851568,
      "grad_norm": 1.125,
      "learning_rate": 1.79775537119319e-05,
      "loss": 0.2984,
      "step": 5484
    },
    {
      "epoch": 0.41511754410860413,
      "grad_norm": 0.765625,
      "learning_rate": 1.7976835456448155e-05,
      "loss": 0.3084,
      "step": 5485
    },
    {
      "epoch": 0.4151932264320515,
      "grad_norm": 0.90234375,
      "learning_rate": 1.7976117087799307e-05,
      "loss": 0.3955,
      "step": 5486
    },
    {
      "epoch": 0.41526890875549877,
      "grad_norm": 0.82421875,
      "learning_rate": 1.7975398605995548e-05,
      "loss": 0.3581,
      "step": 5487
    },
    {
      "epoch": 0.4153445910789461,
      "grad_norm": 0.765625,
      "learning_rate": 1.797468001104708e-05,
      "loss": 0.3257,
      "step": 5488
    },
    {
      "epoch": 0.41542027340239346,
      "grad_norm": 0.8671875,
      "learning_rate": 1.7973961302964086e-05,
      "loss": 0.3912,
      "step": 5489
    },
    {
      "epoch": 0.4154959557258408,
      "grad_norm": 0.7890625,
      "learning_rate": 1.797324248175677e-05,
      "loss": 0.326,
      "step": 5490
    },
    {
      "epoch": 0.4155716380492881,
      "grad_norm": 0.80078125,
      "learning_rate": 1.7972523547435324e-05,
      "loss": 0.3145,
      "step": 5491
    },
    {
      "epoch": 0.41564732037273544,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7971804500009953e-05,
      "loss": 0.3106,
      "step": 5492
    },
    {
      "epoch": 0.4157230026961828,
      "grad_norm": 0.75390625,
      "learning_rate": 1.7971085339490853e-05,
      "loss": 0.3033,
      "step": 5493
    },
    {
      "epoch": 0.4157986850196301,
      "grad_norm": 0.76953125,
      "learning_rate": 1.797036606588823e-05,
      "loss": 0.3081,
      "step": 5494
    },
    {
      "epoch": 0.4158743673430774,
      "grad_norm": 0.80859375,
      "learning_rate": 1.7969646679212285e-05,
      "loss": 0.3314,
      "step": 5495
    },
    {
      "epoch": 0.41595004966652477,
      "grad_norm": 0.796875,
      "learning_rate": 1.7968927179473228e-05,
      "loss": 0.3291,
      "step": 5496
    },
    {
      "epoch": 0.4160257319899721,
      "grad_norm": 0.8203125,
      "learning_rate": 1.7968207566681265e-05,
      "loss": 0.3448,
      "step": 5497
    },
    {
      "epoch": 0.4161014143134194,
      "grad_norm": 0.8046875,
      "learning_rate": 1.79674878408466e-05,
      "loss": 0.3174,
      "step": 5498
    },
    {
      "epoch": 0.41617709663686675,
      "grad_norm": 0.74609375,
      "learning_rate": 1.7966768001979446e-05,
      "loss": 0.3217,
      "step": 5499
    },
    {
      "epoch": 0.4162527789603141,
      "grad_norm": 0.7421875,
      "learning_rate": 1.796604805009002e-05,
      "loss": 0.3059,
      "step": 5500
    },
    {
      "epoch": 0.4163284612837614,
      "grad_norm": 0.80859375,
      "learning_rate": 1.796532798518853e-05,
      "loss": 0.3464,
      "step": 5501
    },
    {
      "epoch": 0.41640414360720873,
      "grad_norm": 0.7578125,
      "learning_rate": 1.7964607807285194e-05,
      "loss": 0.3218,
      "step": 5502
    },
    {
      "epoch": 0.4164798259306561,
      "grad_norm": 0.7578125,
      "learning_rate": 1.796388751639023e-05,
      "loss": 0.3308,
      "step": 5503
    },
    {
      "epoch": 0.4165555082541034,
      "grad_norm": 0.8046875,
      "learning_rate": 1.7963167112513854e-05,
      "loss": 0.3477,
      "step": 5504
    },
    {
      "epoch": 0.4166311905775507,
      "grad_norm": 0.80078125,
      "learning_rate": 1.7962446595666286e-05,
      "loss": 0.3176,
      "step": 5505
    },
    {
      "epoch": 0.41670687290099806,
      "grad_norm": 0.765625,
      "learning_rate": 1.796172596585775e-05,
      "loss": 0.3083,
      "step": 5506
    },
    {
      "epoch": 0.4167825552244454,
      "grad_norm": 0.8671875,
      "learning_rate": 1.796100522309847e-05,
      "loss": 0.3561,
      "step": 5507
    },
    {
      "epoch": 0.4168582375478927,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7960284367398668e-05,
      "loss": 0.3239,
      "step": 5508
    },
    {
      "epoch": 0.41693391987134004,
      "grad_norm": 7.3125,
      "learning_rate": 1.7959563398768575e-05,
      "loss": 0.663,
      "step": 5509
    },
    {
      "epoch": 0.4170096021947874,
      "grad_norm": 0.828125,
      "learning_rate": 1.7958842317218413e-05,
      "loss": 0.3366,
      "step": 5510
    },
    {
      "epoch": 0.41708528451823473,
      "grad_norm": 0.7265625,
      "learning_rate": 1.7958121122758412e-05,
      "loss": 0.2906,
      "step": 5511
    },
    {
      "epoch": 0.417160966841682,
      "grad_norm": 0.796875,
      "learning_rate": 1.7957399815398814e-05,
      "loss": 0.3525,
      "step": 5512
    },
    {
      "epoch": 0.41723664916512937,
      "grad_norm": 0.765625,
      "learning_rate": 1.795667839514984e-05,
      "loss": 0.3086,
      "step": 5513
    },
    {
      "epoch": 0.4173123314885767,
      "grad_norm": 0.76171875,
      "learning_rate": 1.7955956862021726e-05,
      "loss": 0.2909,
      "step": 5514
    },
    {
      "epoch": 0.417388013812024,
      "grad_norm": 0.8125,
      "learning_rate": 1.7955235216024714e-05,
      "loss": 0.3136,
      "step": 5515
    },
    {
      "epoch": 0.41746369613547135,
      "grad_norm": 0.82421875,
      "learning_rate": 1.795451345716904e-05,
      "loss": 0.3325,
      "step": 5516
    },
    {
      "epoch": 0.4175393784589187,
      "grad_norm": 0.79296875,
      "learning_rate": 1.795379158546494e-05,
      "loss": 0.3671,
      "step": 5517
    },
    {
      "epoch": 0.41761506078236604,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7953069600922653e-05,
      "loss": 0.3084,
      "step": 5518
    },
    {
      "epoch": 0.41769074310581333,
      "grad_norm": 1.640625,
      "learning_rate": 1.7952347503552433e-05,
      "loss": 0.4064,
      "step": 5519
    },
    {
      "epoch": 0.4177664254292607,
      "grad_norm": 0.85546875,
      "learning_rate": 1.7951625293364516e-05,
      "loss": 0.3599,
      "step": 5520
    },
    {
      "epoch": 0.417842107752708,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7950902970369147e-05,
      "loss": 0.3138,
      "step": 5521
    },
    {
      "epoch": 0.4179177900761553,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7950180534576572e-05,
      "loss": 0.308,
      "step": 5522
    },
    {
      "epoch": 0.41799347239960266,
      "grad_norm": 0.8125,
      "learning_rate": 1.794945798599705e-05,
      "loss": 0.3662,
      "step": 5523
    },
    {
      "epoch": 0.41806915472305,
      "grad_norm": 0.75390625,
      "learning_rate": 1.794873532464082e-05,
      "loss": 0.3163,
      "step": 5524
    },
    {
      "epoch": 0.41814483704649735,
      "grad_norm": 0.76171875,
      "learning_rate": 1.7948012550518142e-05,
      "loss": 0.3113,
      "step": 5525
    },
    {
      "epoch": 0.41822051936994464,
      "grad_norm": 0.77734375,
      "learning_rate": 1.7947289663639265e-05,
      "loss": 0.3192,
      "step": 5526
    },
    {
      "epoch": 0.418296201693392,
      "grad_norm": 0.8359375,
      "learning_rate": 1.7946566664014446e-05,
      "loss": 0.3105,
      "step": 5527
    },
    {
      "epoch": 0.41837188401683933,
      "grad_norm": 0.8203125,
      "learning_rate": 1.7945843551653942e-05,
      "loss": 0.3742,
      "step": 5528
    },
    {
      "epoch": 0.4184475663402866,
      "grad_norm": 0.859375,
      "learning_rate": 1.794512032656801e-05,
      "loss": 0.3542,
      "step": 5529
    },
    {
      "epoch": 0.41852324866373397,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7944396988766916e-05,
      "loss": 0.3375,
      "step": 5530
    },
    {
      "epoch": 0.4185989309871813,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7943673538260912e-05,
      "loss": 0.3587,
      "step": 5531
    },
    {
      "epoch": 0.41867461331062866,
      "grad_norm": 0.7890625,
      "learning_rate": 1.794294997506027e-05,
      "loss": 0.3161,
      "step": 5532
    },
    {
      "epoch": 0.41875029563407595,
      "grad_norm": 0.90625,
      "learning_rate": 1.7942226299175253e-05,
      "loss": 0.3505,
      "step": 5533
    },
    {
      "epoch": 0.4188259779575233,
      "grad_norm": 0.7578125,
      "learning_rate": 1.7941502510616126e-05,
      "loss": 0.313,
      "step": 5534
    },
    {
      "epoch": 0.41890166028097064,
      "grad_norm": 0.78515625,
      "learning_rate": 1.794077860939316e-05,
      "loss": 0.3237,
      "step": 5535
    },
    {
      "epoch": 0.41897734260441793,
      "grad_norm": 0.72265625,
      "learning_rate": 1.794005459551662e-05,
      "loss": 0.2929,
      "step": 5536
    },
    {
      "epoch": 0.4190530249278653,
      "grad_norm": 0.8359375,
      "learning_rate": 1.793933046899678e-05,
      "loss": 0.3806,
      "step": 5537
    },
    {
      "epoch": 0.4191287072513126,
      "grad_norm": 0.81640625,
      "learning_rate": 1.7938606229843913e-05,
      "loss": 0.3788,
      "step": 5538
    },
    {
      "epoch": 0.41920438957476,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7937881878068297e-05,
      "loss": 0.2863,
      "step": 5539
    },
    {
      "epoch": 0.41928007189820726,
      "grad_norm": 0.80078125,
      "learning_rate": 1.7937157413680196e-05,
      "loss": 0.3701,
      "step": 5540
    },
    {
      "epoch": 0.4193557542216546,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7936432836689907e-05,
      "loss": 0.3183,
      "step": 5541
    },
    {
      "epoch": 0.41943143654510195,
      "grad_norm": 0.72265625,
      "learning_rate": 1.7935708147107692e-05,
      "loss": 0.2968,
      "step": 5542
    },
    {
      "epoch": 0.41950711886854924,
      "grad_norm": 0.81640625,
      "learning_rate": 1.7934983344943838e-05,
      "loss": 0.3203,
      "step": 5543
    },
    {
      "epoch": 0.4195828011919966,
      "grad_norm": 0.70703125,
      "learning_rate": 1.793425843020863e-05,
      "loss": 0.2537,
      "step": 5544
    },
    {
      "epoch": 0.41965848351544394,
      "grad_norm": 0.83203125,
      "learning_rate": 1.7933533402912354e-05,
      "loss": 0.3326,
      "step": 5545
    },
    {
      "epoch": 0.4197341658388912,
      "grad_norm": 0.80078125,
      "learning_rate": 1.793280826306529e-05,
      "loss": 0.3398,
      "step": 5546
    },
    {
      "epoch": 0.4198098481623386,
      "grad_norm": 0.73828125,
      "learning_rate": 1.7932083010677725e-05,
      "loss": 0.3072,
      "step": 5547
    },
    {
      "epoch": 0.4198855304857859,
      "grad_norm": 0.8359375,
      "learning_rate": 1.7931357645759953e-05,
      "loss": 0.3618,
      "step": 5548
    },
    {
      "epoch": 0.41996121280923326,
      "grad_norm": 0.76953125,
      "learning_rate": 1.793063216832226e-05,
      "loss": 0.2928,
      "step": 5549
    },
    {
      "epoch": 0.42003689513268055,
      "grad_norm": 0.796875,
      "learning_rate": 1.7929906578374946e-05,
      "loss": 0.3148,
      "step": 5550
    },
    {
      "epoch": 0.4201125774561279,
      "grad_norm": 0.75,
      "learning_rate": 1.7929180875928293e-05,
      "loss": 0.312,
      "step": 5551
    },
    {
      "epoch": 0.42018825977957525,
      "grad_norm": 0.75,
      "learning_rate": 1.7928455060992606e-05,
      "loss": 0.3247,
      "step": 5552
    },
    {
      "epoch": 0.42026394210302254,
      "grad_norm": 0.76171875,
      "learning_rate": 1.7927729133578177e-05,
      "loss": 0.3127,
      "step": 5553
    },
    {
      "epoch": 0.4203396244264699,
      "grad_norm": 0.75390625,
      "learning_rate": 1.7927003093695306e-05,
      "loss": 0.3147,
      "step": 5554
    },
    {
      "epoch": 0.42041530674991723,
      "grad_norm": 0.76171875,
      "learning_rate": 1.7926276941354293e-05,
      "loss": 0.277,
      "step": 5555
    },
    {
      "epoch": 0.4204909890733646,
      "grad_norm": 0.78125,
      "learning_rate": 1.792555067656544e-05,
      "loss": 0.3249,
      "step": 5556
    },
    {
      "epoch": 0.42056667139681186,
      "grad_norm": 0.81640625,
      "learning_rate": 1.792482429933905e-05,
      "loss": 0.3412,
      "step": 5557
    },
    {
      "epoch": 0.4206423537202592,
      "grad_norm": 0.84765625,
      "learning_rate": 1.7924097809685427e-05,
      "loss": 0.3648,
      "step": 5558
    },
    {
      "epoch": 0.42071803604370656,
      "grad_norm": 0.82421875,
      "learning_rate": 1.7923371207614876e-05,
      "loss": 0.3369,
      "step": 5559
    },
    {
      "epoch": 0.42079371836715385,
      "grad_norm": 0.765625,
      "learning_rate": 1.792264449313771e-05,
      "loss": 0.3353,
      "step": 5560
    },
    {
      "epoch": 0.4208694006906012,
      "grad_norm": 0.76171875,
      "learning_rate": 1.7921917666264236e-05,
      "loss": 0.2906,
      "step": 5561
    },
    {
      "epoch": 0.42094508301404854,
      "grad_norm": 0.77734375,
      "learning_rate": 1.7921190727004763e-05,
      "loss": 0.3288,
      "step": 5562
    },
    {
      "epoch": 0.4210207653374959,
      "grad_norm": 0.765625,
      "learning_rate": 1.792046367536961e-05,
      "loss": 0.2989,
      "step": 5563
    },
    {
      "epoch": 0.4210964476609432,
      "grad_norm": 0.859375,
      "learning_rate": 1.7919736511369087e-05,
      "loss": 0.3683,
      "step": 5564
    },
    {
      "epoch": 0.4211721299843905,
      "grad_norm": 0.9921875,
      "learning_rate": 1.791900923501351e-05,
      "loss": 0.3412,
      "step": 5565
    },
    {
      "epoch": 0.42124781230783787,
      "grad_norm": 0.703125,
      "learning_rate": 1.79182818463132e-05,
      "loss": 0.2783,
      "step": 5566
    },
    {
      "epoch": 0.42132349463128516,
      "grad_norm": 0.73828125,
      "learning_rate": 1.791755434527847e-05,
      "loss": 0.3241,
      "step": 5567
    },
    {
      "epoch": 0.4213991769547325,
      "grad_norm": 0.81640625,
      "learning_rate": 1.7916826731919644e-05,
      "loss": 0.3261,
      "step": 5568
    },
    {
      "epoch": 0.42147485927817985,
      "grad_norm": 0.7578125,
      "learning_rate": 1.791609900624705e-05,
      "loss": 0.2798,
      "step": 5569
    },
    {
      "epoch": 0.4215505416016272,
      "grad_norm": 0.859375,
      "learning_rate": 1.7915371168271e-05,
      "loss": 0.3747,
      "step": 5570
    },
    {
      "epoch": 0.4216262239250745,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7914643218001836e-05,
      "loss": 0.3161,
      "step": 5571
    },
    {
      "epoch": 0.42170190624852183,
      "grad_norm": 0.69140625,
      "learning_rate": 1.7913915155449867e-05,
      "loss": 0.285,
      "step": 5572
    },
    {
      "epoch": 0.4217775885719692,
      "grad_norm": 0.91796875,
      "learning_rate": 1.7913186980625437e-05,
      "loss": 0.3263,
      "step": 5573
    },
    {
      "epoch": 0.42185327089541647,
      "grad_norm": 0.828125,
      "learning_rate": 1.7912458693538868e-05,
      "loss": 0.3663,
      "step": 5574
    },
    {
      "epoch": 0.4219289532188638,
      "grad_norm": 0.80078125,
      "learning_rate": 1.7911730294200494e-05,
      "loss": 0.3783,
      "step": 5575
    },
    {
      "epoch": 0.42200463554231116,
      "grad_norm": 0.74609375,
      "learning_rate": 1.791100178262065e-05,
      "loss": 0.3099,
      "step": 5576
    },
    {
      "epoch": 0.4220803178657585,
      "grad_norm": 0.76953125,
      "learning_rate": 1.791027315880967e-05,
      "loss": 0.3337,
      "step": 5577
    },
    {
      "epoch": 0.4221560001892058,
      "grad_norm": 0.73828125,
      "learning_rate": 1.7909544422777887e-05,
      "loss": 0.3047,
      "step": 5578
    },
    {
      "epoch": 0.42223168251265314,
      "grad_norm": 0.8125,
      "learning_rate": 1.7908815574535646e-05,
      "loss": 0.3229,
      "step": 5579
    },
    {
      "epoch": 0.4223073648361005,
      "grad_norm": 0.7109375,
      "learning_rate": 1.7908086614093287e-05,
      "loss": 0.2626,
      "step": 5580
    },
    {
      "epoch": 0.4223830471595478,
      "grad_norm": 0.73828125,
      "learning_rate": 1.7907357541461142e-05,
      "loss": 0.2869,
      "step": 5581
    },
    {
      "epoch": 0.4224587294829951,
      "grad_norm": 0.89453125,
      "learning_rate": 1.7906628356649564e-05,
      "loss": 0.3828,
      "step": 5582
    },
    {
      "epoch": 0.42253441180644247,
      "grad_norm": 0.8125,
      "learning_rate": 1.7905899059668895e-05,
      "loss": 0.3399,
      "step": 5583
    },
    {
      "epoch": 0.4226100941298898,
      "grad_norm": 1.015625,
      "learning_rate": 1.790516965052948e-05,
      "loss": 0.3373,
      "step": 5584
    },
    {
      "epoch": 0.4226857764533371,
      "grad_norm": 0.8671875,
      "learning_rate": 1.790444012924167e-05,
      "loss": 0.3728,
      "step": 5585
    },
    {
      "epoch": 0.42276145877678445,
      "grad_norm": 0.89453125,
      "learning_rate": 1.790371049581581e-05,
      "loss": 0.3613,
      "step": 5586
    },
    {
      "epoch": 0.4228371411002318,
      "grad_norm": 0.7421875,
      "learning_rate": 1.7902980750262252e-05,
      "loss": 0.3105,
      "step": 5587
    },
    {
      "epoch": 0.4229128234236791,
      "grad_norm": 1.203125,
      "learning_rate": 1.7902250892591356e-05,
      "loss": 0.3366,
      "step": 5588
    },
    {
      "epoch": 0.42298850574712643,
      "grad_norm": 0.84765625,
      "learning_rate": 1.7901520922813467e-05,
      "loss": 0.3592,
      "step": 5589
    },
    {
      "epoch": 0.4230641880705738,
      "grad_norm": 0.9140625,
      "learning_rate": 1.790079084093894e-05,
      "loss": 0.3292,
      "step": 5590
    },
    {
      "epoch": 0.4231398703940211,
      "grad_norm": 1.1171875,
      "learning_rate": 1.7900060646978143e-05,
      "loss": 0.385,
      "step": 5591
    },
    {
      "epoch": 0.4232155527174684,
      "grad_norm": 0.91796875,
      "learning_rate": 1.7899330340941426e-05,
      "loss": 0.4322,
      "step": 5592
    },
    {
      "epoch": 0.42329123504091576,
      "grad_norm": 0.796875,
      "learning_rate": 1.7898599922839153e-05,
      "loss": 0.3274,
      "step": 5593
    },
    {
      "epoch": 0.4233669173643631,
      "grad_norm": 0.75,
      "learning_rate": 1.7897869392681685e-05,
      "loss": 0.2951,
      "step": 5594
    },
    {
      "epoch": 0.4234425996878104,
      "grad_norm": 0.77734375,
      "learning_rate": 1.7897138750479386e-05,
      "loss": 0.306,
      "step": 5595
    },
    {
      "epoch": 0.42351828201125774,
      "grad_norm": 0.75,
      "learning_rate": 1.7896407996242625e-05,
      "loss": 0.3175,
      "step": 5596
    },
    {
      "epoch": 0.4235939643347051,
      "grad_norm": 0.765625,
      "learning_rate": 1.7895677129981762e-05,
      "loss": 0.3287,
      "step": 5597
    },
    {
      "epoch": 0.42366964665815243,
      "grad_norm": 0.796875,
      "learning_rate": 1.789494615170717e-05,
      "loss": 0.3198,
      "step": 5598
    },
    {
      "epoch": 0.4237453289815997,
      "grad_norm": 0.97265625,
      "learning_rate": 1.7894215061429218e-05,
      "loss": 0.37,
      "step": 5599
    },
    {
      "epoch": 0.42382101130504707,
      "grad_norm": 0.73046875,
      "learning_rate": 1.7893483859158282e-05,
      "loss": 0.3084,
      "step": 5600
    },
    {
      "epoch": 0.4238966936284944,
      "grad_norm": 0.7890625,
      "learning_rate": 1.789275254490473e-05,
      "loss": 0.3545,
      "step": 5601
    },
    {
      "epoch": 0.4239723759519417,
      "grad_norm": 0.765625,
      "learning_rate": 1.7892021118678936e-05,
      "loss": 0.3148,
      "step": 5602
    },
    {
      "epoch": 0.42404805827538905,
      "grad_norm": 0.765625,
      "learning_rate": 1.7891289580491285e-05,
      "loss": 0.3233,
      "step": 5603
    },
    {
      "epoch": 0.4241237405988364,
      "grad_norm": 0.8359375,
      "learning_rate": 1.7890557930352143e-05,
      "loss": 0.3585,
      "step": 5604
    },
    {
      "epoch": 0.4241994229222837,
      "grad_norm": 0.84375,
      "learning_rate": 1.78898261682719e-05,
      "loss": 0.3686,
      "step": 5605
    },
    {
      "epoch": 0.42427510524573103,
      "grad_norm": 0.765625,
      "learning_rate": 1.7889094294260935e-05,
      "loss": 0.3138,
      "step": 5606
    },
    {
      "epoch": 0.4243507875691784,
      "grad_norm": 0.71875,
      "learning_rate": 1.7888362308329624e-05,
      "loss": 0.2969,
      "step": 5607
    },
    {
      "epoch": 0.4244264698926257,
      "grad_norm": 0.79296875,
      "learning_rate": 1.788763021048836e-05,
      "loss": 0.3429,
      "step": 5608
    },
    {
      "epoch": 0.424502152216073,
      "grad_norm": 0.80859375,
      "learning_rate": 1.7886898000747525e-05,
      "loss": 0.3171,
      "step": 5609
    },
    {
      "epoch": 0.42457783453952036,
      "grad_norm": 0.75390625,
      "learning_rate": 1.7886165679117508e-05,
      "loss": 0.3142,
      "step": 5610
    },
    {
      "epoch": 0.4246535168629677,
      "grad_norm": 0.75390625,
      "learning_rate": 1.78854332456087e-05,
      "loss": 0.3028,
      "step": 5611
    },
    {
      "epoch": 0.424729199186415,
      "grad_norm": 0.796875,
      "learning_rate": 1.7884700700231487e-05,
      "loss": 0.2881,
      "step": 5612
    },
    {
      "epoch": 0.42480488150986234,
      "grad_norm": 0.7421875,
      "learning_rate": 1.7883968042996262e-05,
      "loss": 0.3116,
      "step": 5613
    },
    {
      "epoch": 0.4248805638333097,
      "grad_norm": 1.9140625,
      "learning_rate": 1.7883235273913424e-05,
      "loss": 0.4047,
      "step": 5614
    },
    {
      "epoch": 0.42495624615675703,
      "grad_norm": 0.73046875,
      "learning_rate": 1.7882502392993366e-05,
      "loss": 0.2846,
      "step": 5615
    },
    {
      "epoch": 0.4250319284802043,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7881769400246484e-05,
      "loss": 0.3443,
      "step": 5616
    },
    {
      "epoch": 0.42510761080365167,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7881036295683175e-05,
      "loss": 0.3131,
      "step": 5617
    },
    {
      "epoch": 0.425183293127099,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7880303079313845e-05,
      "loss": 0.3248,
      "step": 5618
    },
    {
      "epoch": 0.4252589754505463,
      "grad_norm": 0.828125,
      "learning_rate": 1.787956975114889e-05,
      "loss": 0.3122,
      "step": 5619
    },
    {
      "epoch": 0.42533465777399365,
      "grad_norm": 0.765625,
      "learning_rate": 1.7878836311198716e-05,
      "loss": 0.3234,
      "step": 5620
    },
    {
      "epoch": 0.425410340097441,
      "grad_norm": 1.0,
      "learning_rate": 1.787810275947373e-05,
      "loss": 0.3453,
      "step": 5621
    },
    {
      "epoch": 0.42548602242088834,
      "grad_norm": 0.765625,
      "learning_rate": 1.7877369095984338e-05,
      "loss": 0.309,
      "step": 5622
    },
    {
      "epoch": 0.42556170474433563,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7876635320740948e-05,
      "loss": 0.3575,
      "step": 5623
    },
    {
      "epoch": 0.425637387067783,
      "grad_norm": 0.8125,
      "learning_rate": 1.7875901433753967e-05,
      "loss": 0.3179,
      "step": 5624
    },
    {
      "epoch": 0.4257130693912303,
      "grad_norm": 0.81640625,
      "learning_rate": 1.787516743503381e-05,
      "loss": 0.3156,
      "step": 5625
    },
    {
      "epoch": 0.4257887517146776,
      "grad_norm": 0.8515625,
      "learning_rate": 1.787443332459089e-05,
      "loss": 0.3608,
      "step": 5626
    },
    {
      "epoch": 0.42586443403812496,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7873699102435616e-05,
      "loss": 0.3465,
      "step": 5627
    },
    {
      "epoch": 0.4259401163615723,
      "grad_norm": 0.72265625,
      "learning_rate": 1.7872964768578412e-05,
      "loss": 0.2794,
      "step": 5628
    },
    {
      "epoch": 0.42601579868501965,
      "grad_norm": 0.85546875,
      "learning_rate": 1.787223032302969e-05,
      "loss": 0.3643,
      "step": 5629
    },
    {
      "epoch": 0.42609148100846694,
      "grad_norm": 0.76171875,
      "learning_rate": 1.7871495765799875e-05,
      "loss": 0.2769,
      "step": 5630
    },
    {
      "epoch": 0.4261671633319143,
      "grad_norm": 0.80859375,
      "learning_rate": 1.787076109689938e-05,
      "loss": 0.3304,
      "step": 5631
    },
    {
      "epoch": 0.42624284565536164,
      "grad_norm": 0.8125,
      "learning_rate": 1.7870026316338635e-05,
      "loss": 0.3416,
      "step": 5632
    },
    {
      "epoch": 0.4263185279788089,
      "grad_norm": 0.78125,
      "learning_rate": 1.786929142412806e-05,
      "loss": 0.3318,
      "step": 5633
    },
    {
      "epoch": 0.42639421030225627,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7868556420278084e-05,
      "loss": 0.2971,
      "step": 5634
    },
    {
      "epoch": 0.4264698926257036,
      "grad_norm": 0.7265625,
      "learning_rate": 1.786782130479913e-05,
      "loss": 0.2975,
      "step": 5635
    },
    {
      "epoch": 0.42654557494915096,
      "grad_norm": 0.796875,
      "learning_rate": 1.786708607770163e-05,
      "loss": 0.3347,
      "step": 5636
    },
    {
      "epoch": 0.42662125727259825,
      "grad_norm": 0.796875,
      "learning_rate": 1.7866350738996015e-05,
      "loss": 0.3439,
      "step": 5637
    },
    {
      "epoch": 0.4266969395960456,
      "grad_norm": 1.1328125,
      "learning_rate": 1.7865615288692716e-05,
      "loss": 0.388,
      "step": 5638
    },
    {
      "epoch": 0.42677262191949294,
      "grad_norm": 0.8046875,
      "learning_rate": 1.786487972680216e-05,
      "loss": 0.33,
      "step": 5639
    },
    {
      "epoch": 0.42684830424294024,
      "grad_norm": 0.78125,
      "learning_rate": 1.7864144053334796e-05,
      "loss": 0.3221,
      "step": 5640
    },
    {
      "epoch": 0.4269239865663876,
      "grad_norm": 0.8046875,
      "learning_rate": 1.786340826830105e-05,
      "loss": 0.3291,
      "step": 5641
    },
    {
      "epoch": 0.4269996688898349,
      "grad_norm": 0.83203125,
      "learning_rate": 1.7862672371711364e-05,
      "loss": 0.331,
      "step": 5642
    },
    {
      "epoch": 0.4270753512132823,
      "grad_norm": 0.71875,
      "learning_rate": 1.7861936363576178e-05,
      "loss": 0.2826,
      "step": 5643
    },
    {
      "epoch": 0.42715103353672956,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7861200243905933e-05,
      "loss": 0.3281,
      "step": 5644
    },
    {
      "epoch": 0.4272267158601769,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7860464012711072e-05,
      "loss": 0.3245,
      "step": 5645
    },
    {
      "epoch": 0.42730239818362425,
      "grad_norm": 0.7890625,
      "learning_rate": 1.785972767000204e-05,
      "loss": 0.306,
      "step": 5646
    },
    {
      "epoch": 0.42737808050707154,
      "grad_norm": 0.77734375,
      "learning_rate": 1.7858991215789282e-05,
      "loss": 0.3229,
      "step": 5647
    },
    {
      "epoch": 0.4274537628305189,
      "grad_norm": 0.85546875,
      "learning_rate": 1.7858254650083247e-05,
      "loss": 0.3868,
      "step": 5648
    },
    {
      "epoch": 0.42752944515396624,
      "grad_norm": 0.8203125,
      "learning_rate": 1.785751797289439e-05,
      "loss": 0.3419,
      "step": 5649
    },
    {
      "epoch": 0.4276051274774136,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7856781184233152e-05,
      "loss": 0.3388,
      "step": 5650
    },
    {
      "epoch": 0.4276808098008609,
      "grad_norm": 0.8125,
      "learning_rate": 1.785604428410999e-05,
      "loss": 0.3529,
      "step": 5651
    },
    {
      "epoch": 0.4277564921243082,
      "grad_norm": 0.74609375,
      "learning_rate": 1.785530727253536e-05,
      "loss": 0.3069,
      "step": 5652
    },
    {
      "epoch": 0.42783217444775556,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7854570149519714e-05,
      "loss": 0.3322,
      "step": 5653
    },
    {
      "epoch": 0.42790785677120285,
      "grad_norm": 0.875,
      "learning_rate": 1.785383291507351e-05,
      "loss": 0.3356,
      "step": 5654
    },
    {
      "epoch": 0.4279835390946502,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7853095569207214e-05,
      "loss": 0.2978,
      "step": 5655
    },
    {
      "epoch": 0.42805922141809755,
      "grad_norm": 0.77734375,
      "learning_rate": 1.785235811193128e-05,
      "loss": 0.2985,
      "step": 5656
    },
    {
      "epoch": 0.4281349037415449,
      "grad_norm": 0.83203125,
      "learning_rate": 1.7851620543256166e-05,
      "loss": 0.3266,
      "step": 5657
    },
    {
      "epoch": 0.4282105860649922,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7850882863192347e-05,
      "loss": 0.3215,
      "step": 5658
    },
    {
      "epoch": 0.42828626838843953,
      "grad_norm": 0.859375,
      "learning_rate": 1.7850145071750277e-05,
      "loss": 0.3715,
      "step": 5659
    },
    {
      "epoch": 0.4283619507118869,
      "grad_norm": 0.8203125,
      "learning_rate": 1.784940716894043e-05,
      "loss": 0.3456,
      "step": 5660
    },
    {
      "epoch": 0.42843763303533416,
      "grad_norm": 0.77734375,
      "learning_rate": 1.7848669154773275e-05,
      "loss": 0.3222,
      "step": 5661
    },
    {
      "epoch": 0.4285133153587815,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7847931029259276e-05,
      "loss": 0.3531,
      "step": 5662
    },
    {
      "epoch": 0.42858899768222886,
      "grad_norm": 0.75390625,
      "learning_rate": 1.7847192792408906e-05,
      "loss": 0.3094,
      "step": 5663
    },
    {
      "epoch": 0.4286646800056762,
      "grad_norm": 0.734375,
      "learning_rate": 1.7846454444232643e-05,
      "loss": 0.3166,
      "step": 5664
    },
    {
      "epoch": 0.4287403623291235,
      "grad_norm": 0.796875,
      "learning_rate": 1.784571598474096e-05,
      "loss": 0.3404,
      "step": 5665
    },
    {
      "epoch": 0.42881604465257084,
      "grad_norm": 0.765625,
      "learning_rate": 1.784497741394433e-05,
      "loss": 0.3237,
      "step": 5666
    },
    {
      "epoch": 0.4288917269760182,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7844238731853233e-05,
      "loss": 0.3154,
      "step": 5667
    },
    {
      "epoch": 0.4289674092994655,
      "grad_norm": 0.80859375,
      "learning_rate": 1.7843499938478147e-05,
      "loss": 0.3617,
      "step": 5668
    },
    {
      "epoch": 0.4290430916229128,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7842761033829555e-05,
      "loss": 0.3279,
      "step": 5669
    },
    {
      "epoch": 0.42911877394636017,
      "grad_norm": 0.88671875,
      "learning_rate": 1.784202201791794e-05,
      "loss": 0.4103,
      "step": 5670
    },
    {
      "epoch": 0.42919445626980746,
      "grad_norm": 0.7734375,
      "learning_rate": 1.784128289075379e-05,
      "loss": 0.3552,
      "step": 5671
    },
    {
      "epoch": 0.4292701385932548,
      "grad_norm": 0.765625,
      "learning_rate": 1.784054365234758e-05,
      "loss": 0.2941,
      "step": 5672
    },
    {
      "epoch": 0.42934582091670215,
      "grad_norm": 0.9140625,
      "learning_rate": 1.7839804302709805e-05,
      "loss": 0.3633,
      "step": 5673
    },
    {
      "epoch": 0.4294215032401495,
      "grad_norm": 1.2265625,
      "learning_rate": 1.783906484185095e-05,
      "loss": 0.3784,
      "step": 5674
    },
    {
      "epoch": 0.4294971855635968,
      "grad_norm": 0.8359375,
      "learning_rate": 1.7838325269781508e-05,
      "loss": 0.3619,
      "step": 5675
    },
    {
      "epoch": 0.42957286788704413,
      "grad_norm": 0.83203125,
      "learning_rate": 1.7837585586511975e-05,
      "loss": 0.347,
      "step": 5676
    },
    {
      "epoch": 0.4296485502104915,
      "grad_norm": 0.80859375,
      "learning_rate": 1.783684579205284e-05,
      "loss": 0.3405,
      "step": 5677
    },
    {
      "epoch": 0.42972423253393877,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7836105886414596e-05,
      "loss": 0.3324,
      "step": 5678
    },
    {
      "epoch": 0.4297999148573861,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7835365869607747e-05,
      "loss": 0.3118,
      "step": 5679
    },
    {
      "epoch": 0.42987559718083346,
      "grad_norm": 0.83203125,
      "learning_rate": 1.7834625741642782e-05,
      "loss": 0.3462,
      "step": 5680
    },
    {
      "epoch": 0.4299512795042808,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7833885502530212e-05,
      "loss": 0.3216,
      "step": 5681
    },
    {
      "epoch": 0.4300269618277281,
      "grad_norm": 0.8671875,
      "learning_rate": 1.783314515228053e-05,
      "loss": 0.3386,
      "step": 5682
    },
    {
      "epoch": 0.43010264415117544,
      "grad_norm": 0.79296875,
      "learning_rate": 1.783240469090424e-05,
      "loss": 0.335,
      "step": 5683
    },
    {
      "epoch": 0.4301783264746228,
      "grad_norm": 0.8203125,
      "learning_rate": 1.783166411841185e-05,
      "loss": 0.3603,
      "step": 5684
    },
    {
      "epoch": 0.4302540087980701,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7830923434813866e-05,
      "loss": 0.2968,
      "step": 5685
    },
    {
      "epoch": 0.4303296911215174,
      "grad_norm": 0.828125,
      "learning_rate": 1.783018264012079e-05,
      "loss": 0.3481,
      "step": 5686
    },
    {
      "epoch": 0.43040537344496477,
      "grad_norm": 0.8984375,
      "learning_rate": 1.7829441734343143e-05,
      "loss": 0.3977,
      "step": 5687
    },
    {
      "epoch": 0.4304810557684121,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7828700717491426e-05,
      "loss": 0.2961,
      "step": 5688
    },
    {
      "epoch": 0.4305567380918594,
      "grad_norm": 0.84765625,
      "learning_rate": 1.7827959589576155e-05,
      "loss": 0.332,
      "step": 5689
    },
    {
      "epoch": 0.43063242041530675,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7827218350607845e-05,
      "loss": 0.3133,
      "step": 5690
    },
    {
      "epoch": 0.4307081027387541,
      "grad_norm": 0.80859375,
      "learning_rate": 1.782647700059701e-05,
      "loss": 0.3416,
      "step": 5691
    },
    {
      "epoch": 0.4307837850622014,
      "grad_norm": 1.2734375,
      "learning_rate": 1.7825735539554167e-05,
      "loss": 0.3738,
      "step": 5692
    },
    {
      "epoch": 0.43085946738564873,
      "grad_norm": 0.8359375,
      "learning_rate": 1.7824993967489835e-05,
      "loss": 0.3388,
      "step": 5693
    },
    {
      "epoch": 0.4309351497090961,
      "grad_norm": 0.7421875,
      "learning_rate": 1.7824252284414538e-05,
      "loss": 0.2863,
      "step": 5694
    },
    {
      "epoch": 0.4310108320325434,
      "grad_norm": 0.828125,
      "learning_rate": 1.7823510490338795e-05,
      "loss": 0.3349,
      "step": 5695
    },
    {
      "epoch": 0.4310865143559907,
      "grad_norm": 0.78515625,
      "learning_rate": 1.782276858527313e-05,
      "loss": 0.3413,
      "step": 5696
    },
    {
      "epoch": 0.43116219667943806,
      "grad_norm": 0.77734375,
      "learning_rate": 1.782202656922807e-05,
      "loss": 0.3172,
      "step": 5697
    },
    {
      "epoch": 0.4312378790028854,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7821284442214138e-05,
      "loss": 0.3173,
      "step": 5698
    },
    {
      "epoch": 0.4313135613263327,
      "grad_norm": 0.78125,
      "learning_rate": 1.782054220424186e-05,
      "loss": 0.3121,
      "step": 5699
    },
    {
      "epoch": 0.43138924364978004,
      "grad_norm": 0.84765625,
      "learning_rate": 1.7819799855321777e-05,
      "loss": 0.3624,
      "step": 5700
    },
    {
      "epoch": 0.4314649259732274,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7819057395464412e-05,
      "loss": 0.308,
      "step": 5701
    },
    {
      "epoch": 0.43154060829667473,
      "grad_norm": 0.7578125,
      "learning_rate": 1.78183148246803e-05,
      "loss": 0.2844,
      "step": 5702
    },
    {
      "epoch": 0.431616290620122,
      "grad_norm": 0.80078125,
      "learning_rate": 1.7817572142979975e-05,
      "loss": 0.3221,
      "step": 5703
    },
    {
      "epoch": 0.43169197294356937,
      "grad_norm": 0.83203125,
      "learning_rate": 1.7816829350373974e-05,
      "loss": 0.3561,
      "step": 5704
    },
    {
      "epoch": 0.4317676552670167,
      "grad_norm": 0.8203125,
      "learning_rate": 1.7816086446872833e-05,
      "loss": 0.3145,
      "step": 5705
    },
    {
      "epoch": 0.431843337590464,
      "grad_norm": 0.83203125,
      "learning_rate": 1.7815343432487094e-05,
      "loss": 0.3596,
      "step": 5706
    },
    {
      "epoch": 0.43191901991391135,
      "grad_norm": 1.171875,
      "learning_rate": 1.78146003072273e-05,
      "loss": 0.2981,
      "step": 5707
    },
    {
      "epoch": 0.4319947022373587,
      "grad_norm": 0.7578125,
      "learning_rate": 1.7813857071103986e-05,
      "loss": 0.2815,
      "step": 5708
    },
    {
      "epoch": 0.43207038456080604,
      "grad_norm": 0.7734375,
      "learning_rate": 1.78131137241277e-05,
      "loss": 0.3021,
      "step": 5709
    },
    {
      "epoch": 0.43214606688425333,
      "grad_norm": 0.8203125,
      "learning_rate": 1.781237026630899e-05,
      "loss": 0.3191,
      "step": 5710
    },
    {
      "epoch": 0.4322217492077007,
      "grad_norm": 0.765625,
      "learning_rate": 1.78116266976584e-05,
      "loss": 0.3261,
      "step": 5711
    },
    {
      "epoch": 0.432297431531148,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7810883018186485e-05,
      "loss": 0.2913,
      "step": 5712
    },
    {
      "epoch": 0.4323731138545953,
      "grad_norm": 1.0390625,
      "learning_rate": 1.7810139227903783e-05,
      "loss": 0.3642,
      "step": 5713
    },
    {
      "epoch": 0.43244879617804266,
      "grad_norm": 0.9765625,
      "learning_rate": 1.7809395326820858e-05,
      "loss": 0.3406,
      "step": 5714
    },
    {
      "epoch": 0.43252447850149,
      "grad_norm": 0.8046875,
      "learning_rate": 1.7808651314948256e-05,
      "loss": 0.3109,
      "step": 5715
    },
    {
      "epoch": 0.43260016082493735,
      "grad_norm": 0.75,
      "learning_rate": 1.780790719229654e-05,
      "loss": 0.2755,
      "step": 5716
    },
    {
      "epoch": 0.43267584314838464,
      "grad_norm": 0.77734375,
      "learning_rate": 1.7807162958876256e-05,
      "loss": 0.3019,
      "step": 5717
    },
    {
      "epoch": 0.432751525471832,
      "grad_norm": 0.82421875,
      "learning_rate": 1.780641861469797e-05,
      "loss": 0.3551,
      "step": 5718
    },
    {
      "epoch": 0.43282720779527933,
      "grad_norm": 0.82421875,
      "learning_rate": 1.7805674159772244e-05,
      "loss": 0.3735,
      "step": 5719
    },
    {
      "epoch": 0.4329028901187266,
      "grad_norm": 0.80078125,
      "learning_rate": 1.780492959410963e-05,
      "loss": 0.3229,
      "step": 5720
    },
    {
      "epoch": 0.43297857244217397,
      "grad_norm": 0.734375,
      "learning_rate": 1.7804184917720694e-05,
      "loss": 0.2734,
      "step": 5721
    },
    {
      "epoch": 0.4330542547656213,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7803440130616007e-05,
      "loss": 0.321,
      "step": 5722
    },
    {
      "epoch": 0.43312993708906866,
      "grad_norm": 0.890625,
      "learning_rate": 1.780269523280613e-05,
      "loss": 0.3483,
      "step": 5723
    },
    {
      "epoch": 0.43320561941251595,
      "grad_norm": 0.796875,
      "learning_rate": 1.7801950224301633e-05,
      "loss": 0.3245,
      "step": 5724
    },
    {
      "epoch": 0.4332813017359633,
      "grad_norm": 1.1484375,
      "learning_rate": 1.7801205105113077e-05,
      "loss": 0.3835,
      "step": 5725
    },
    {
      "epoch": 0.43335698405941064,
      "grad_norm": 0.73828125,
      "learning_rate": 1.7800459875251045e-05,
      "loss": 0.3052,
      "step": 5726
    },
    {
      "epoch": 0.43343266638285793,
      "grad_norm": 0.78125,
      "learning_rate": 1.77997145347261e-05,
      "loss": 0.3049,
      "step": 5727
    },
    {
      "epoch": 0.4335083487063053,
      "grad_norm": 0.7734375,
      "learning_rate": 1.779896908354882e-05,
      "loss": 0.3086,
      "step": 5728
    },
    {
      "epoch": 0.4335840310297526,
      "grad_norm": 0.796875,
      "learning_rate": 1.7798223521729783e-05,
      "loss": 0.3363,
      "step": 5729
    },
    {
      "epoch": 0.4336597133531999,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7797477849279564e-05,
      "loss": 0.324,
      "step": 5730
    },
    {
      "epoch": 0.43373539567664726,
      "grad_norm": 0.78125,
      "learning_rate": 1.7796732066208736e-05,
      "loss": 0.2897,
      "step": 5731
    },
    {
      "epoch": 0.4338110780000946,
      "grad_norm": 0.80859375,
      "learning_rate": 1.7795986172527888e-05,
      "loss": 0.3324,
      "step": 5732
    },
    {
      "epoch": 0.43388676032354195,
      "grad_norm": 0.8125,
      "learning_rate": 1.7795240168247592e-05,
      "loss": 0.3363,
      "step": 5733
    },
    {
      "epoch": 0.43396244264698924,
      "grad_norm": 0.8046875,
      "learning_rate": 1.7794494053378445e-05,
      "loss": 0.364,
      "step": 5734
    },
    {
      "epoch": 0.4340381249704366,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7793747827931017e-05,
      "loss": 0.3348,
      "step": 5735
    },
    {
      "epoch": 0.43411380729388394,
      "grad_norm": 0.84375,
      "learning_rate": 1.7793001491915905e-05,
      "loss": 0.4092,
      "step": 5736
    },
    {
      "epoch": 0.4341894896173312,
      "grad_norm": 0.7421875,
      "learning_rate": 1.7792255045343696e-05,
      "loss": 0.3015,
      "step": 5737
    },
    {
      "epoch": 0.43426517194077857,
      "grad_norm": 0.765625,
      "learning_rate": 1.7791508488224972e-05,
      "loss": 0.2965,
      "step": 5738
    },
    {
      "epoch": 0.4343408542642259,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7790761820570336e-05,
      "loss": 0.3166,
      "step": 5739
    },
    {
      "epoch": 0.43441653658767326,
      "grad_norm": 0.796875,
      "learning_rate": 1.779001504239037e-05,
      "loss": 0.2946,
      "step": 5740
    },
    {
      "epoch": 0.43449221891112055,
      "grad_norm": 0.8125,
      "learning_rate": 1.778926815369567e-05,
      "loss": 0.3475,
      "step": 5741
    },
    {
      "epoch": 0.4345679012345679,
      "grad_norm": 0.77734375,
      "learning_rate": 1.778852115449684e-05,
      "loss": 0.295,
      "step": 5742
    },
    {
      "epoch": 0.43464358355801525,
      "grad_norm": 0.74609375,
      "learning_rate": 1.7787774044804467e-05,
      "loss": 0.2986,
      "step": 5743
    },
    {
      "epoch": 0.43471926588146254,
      "grad_norm": 0.8046875,
      "learning_rate": 1.7787026824629158e-05,
      "loss": 0.3198,
      "step": 5744
    },
    {
      "epoch": 0.4347949482049099,
      "grad_norm": 0.796875,
      "learning_rate": 1.7786279493981506e-05,
      "loss": 0.3306,
      "step": 5745
    },
    {
      "epoch": 0.4348706305283572,
      "grad_norm": 0.7421875,
      "learning_rate": 1.7785532052872124e-05,
      "loss": 0.3047,
      "step": 5746
    },
    {
      "epoch": 0.4349463128518046,
      "grad_norm": 0.80078125,
      "learning_rate": 1.7784784501311604e-05,
      "loss": 0.3463,
      "step": 5747
    },
    {
      "epoch": 0.43502199517525186,
      "grad_norm": 0.84375,
      "learning_rate": 1.7784036839310558e-05,
      "loss": 0.3806,
      "step": 5748
    },
    {
      "epoch": 0.4350976774986992,
      "grad_norm": 0.80859375,
      "learning_rate": 1.7783289066879593e-05,
      "loss": 0.3234,
      "step": 5749
    },
    {
      "epoch": 0.43517335982214655,
      "grad_norm": 0.73828125,
      "learning_rate": 1.7782541184029316e-05,
      "loss": 0.3024,
      "step": 5750
    },
    {
      "epoch": 0.43524904214559385,
      "grad_norm": 0.671875,
      "learning_rate": 1.7781793190770336e-05,
      "loss": 0.2563,
      "step": 5751
    },
    {
      "epoch": 0.4353247244690412,
      "grad_norm": 0.79296875,
      "learning_rate": 1.778104508711327e-05,
      "loss": 0.3268,
      "step": 5752
    },
    {
      "epoch": 0.43540040679248854,
      "grad_norm": 0.7578125,
      "learning_rate": 1.778029687306872e-05,
      "loss": 0.3007,
      "step": 5753
    },
    {
      "epoch": 0.4354760891159359,
      "grad_norm": 0.71484375,
      "learning_rate": 1.7779548548647312e-05,
      "loss": 0.3001,
      "step": 5754
    },
    {
      "epoch": 0.4355517714393832,
      "grad_norm": 0.8515625,
      "learning_rate": 1.7778800113859656e-05,
      "loss": 0.3628,
      "step": 5755
    },
    {
      "epoch": 0.4356274537628305,
      "grad_norm": 0.7265625,
      "learning_rate": 1.777805156871637e-05,
      "loss": 0.298,
      "step": 5756
    },
    {
      "epoch": 0.43570313608627786,
      "grad_norm": 0.83984375,
      "learning_rate": 1.777730291322808e-05,
      "loss": 0.3479,
      "step": 5757
    },
    {
      "epoch": 0.43577881840972515,
      "grad_norm": 0.9609375,
      "learning_rate": 1.7776554147405396e-05,
      "loss": 0.3466,
      "step": 5758
    },
    {
      "epoch": 0.4358545007331725,
      "grad_norm": 0.83203125,
      "learning_rate": 1.7775805271258948e-05,
      "loss": 0.3238,
      "step": 5759
    },
    {
      "epoch": 0.43593018305661985,
      "grad_norm": 0.83203125,
      "learning_rate": 1.777505628479936e-05,
      "loss": 0.3596,
      "step": 5760
    },
    {
      "epoch": 0.4360058653800672,
      "grad_norm": 0.76171875,
      "learning_rate": 1.7774307188037255e-05,
      "loss": 0.3293,
      "step": 5761
    },
    {
      "epoch": 0.4360815477035145,
      "grad_norm": 1.2734375,
      "learning_rate": 1.7773557980983264e-05,
      "loss": 0.3983,
      "step": 5762
    },
    {
      "epoch": 0.43615723002696183,
      "grad_norm": 0.8515625,
      "learning_rate": 1.7772808663648006e-05,
      "loss": 0.3469,
      "step": 5763
    },
    {
      "epoch": 0.4362329123504092,
      "grad_norm": 0.90234375,
      "learning_rate": 1.7772059236042126e-05,
      "loss": 0.3259,
      "step": 5764
    },
    {
      "epoch": 0.43630859467385646,
      "grad_norm": 0.76171875,
      "learning_rate": 1.7771309698176246e-05,
      "loss": 0.2818,
      "step": 5765
    },
    {
      "epoch": 0.4363842769973038,
      "grad_norm": 0.72265625,
      "learning_rate": 1.7770560050061002e-05,
      "loss": 0.3098,
      "step": 5766
    },
    {
      "epoch": 0.43645995932075116,
      "grad_norm": 0.83984375,
      "learning_rate": 1.7769810291707025e-05,
      "loss": 0.3442,
      "step": 5767
    },
    {
      "epoch": 0.4365356416441985,
      "grad_norm": 0.7421875,
      "learning_rate": 1.776906042312496e-05,
      "loss": 0.3021,
      "step": 5768
    },
    {
      "epoch": 0.4366113239676458,
      "grad_norm": 0.85546875,
      "learning_rate": 1.7768310444325442e-05,
      "loss": 0.3653,
      "step": 5769
    },
    {
      "epoch": 0.43668700629109314,
      "grad_norm": 0.80078125,
      "learning_rate": 1.7767560355319106e-05,
      "loss": 0.3453,
      "step": 5770
    },
    {
      "epoch": 0.4367626886145405,
      "grad_norm": 0.82421875,
      "learning_rate": 1.77668101561166e-05,
      "loss": 0.3592,
      "step": 5771
    },
    {
      "epoch": 0.4368383709379878,
      "grad_norm": 0.7265625,
      "learning_rate": 1.776605984672856e-05,
      "loss": 0.2862,
      "step": 5772
    },
    {
      "epoch": 0.4369140532614351,
      "grad_norm": 0.86328125,
      "learning_rate": 1.7765309427165636e-05,
      "loss": 0.3577,
      "step": 5773
    },
    {
      "epoch": 0.43698973558488247,
      "grad_norm": 0.76171875,
      "learning_rate": 1.776455889743847e-05,
      "loss": 0.3005,
      "step": 5774
    },
    {
      "epoch": 0.4370654179083298,
      "grad_norm": 0.97265625,
      "learning_rate": 1.7763808257557714e-05,
      "loss": 0.3,
      "step": 5775
    },
    {
      "epoch": 0.4371411002317771,
      "grad_norm": 0.74609375,
      "learning_rate": 1.7763057507534016e-05,
      "loss": 0.3084,
      "step": 5776
    },
    {
      "epoch": 0.43721678255522445,
      "grad_norm": 0.83984375,
      "learning_rate": 1.7762306647378023e-05,
      "loss": 0.3815,
      "step": 5777
    },
    {
      "epoch": 0.4372924648786718,
      "grad_norm": 1.1640625,
      "learning_rate": 1.776155567710039e-05,
      "loss": 0.3964,
      "step": 5778
    },
    {
      "epoch": 0.4373681472021191,
      "grad_norm": 0.74609375,
      "learning_rate": 1.7760804596711772e-05,
      "loss": 0.3245,
      "step": 5779
    },
    {
      "epoch": 0.43744382952556643,
      "grad_norm": 0.72265625,
      "learning_rate": 1.776005340622282e-05,
      "loss": 0.296,
      "step": 5780
    },
    {
      "epoch": 0.4375195118490138,
      "grad_norm": 0.87109375,
      "learning_rate": 1.7759302105644196e-05,
      "loss": 0.3496,
      "step": 5781
    },
    {
      "epoch": 0.4375951941724611,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7758550694986555e-05,
      "loss": 0.3055,
      "step": 5782
    },
    {
      "epoch": 0.4376708764959084,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7757799174260558e-05,
      "loss": 0.314,
      "step": 5783
    },
    {
      "epoch": 0.43774655881935576,
      "grad_norm": 0.83984375,
      "learning_rate": 1.775704754347687e-05,
      "loss": 0.3214,
      "step": 5784
    },
    {
      "epoch": 0.4378222411428031,
      "grad_norm": 0.86328125,
      "learning_rate": 1.7756295802646145e-05,
      "loss": 0.3524,
      "step": 5785
    },
    {
      "epoch": 0.4378979234662504,
      "grad_norm": 0.7578125,
      "learning_rate": 1.7755543951779058e-05,
      "loss": 0.3245,
      "step": 5786
    },
    {
      "epoch": 0.43797360578969774,
      "grad_norm": 0.7890625,
      "learning_rate": 1.775479199088627e-05,
      "loss": 0.339,
      "step": 5787
    },
    {
      "epoch": 0.4380492881131451,
      "grad_norm": 0.796875,
      "learning_rate": 1.7754039919978452e-05,
      "loss": 0.3271,
      "step": 5788
    },
    {
      "epoch": 0.4381249704365924,
      "grad_norm": 0.8203125,
      "learning_rate": 1.7753287739066268e-05,
      "loss": 0.3178,
      "step": 5789
    },
    {
      "epoch": 0.4382006527600397,
      "grad_norm": 0.82421875,
      "learning_rate": 1.7752535448160395e-05,
      "loss": 0.3594,
      "step": 5790
    },
    {
      "epoch": 0.43827633508348707,
      "grad_norm": 0.78125,
      "learning_rate": 1.77517830472715e-05,
      "loss": 0.2886,
      "step": 5791
    },
    {
      "epoch": 0.4383520174069344,
      "grad_norm": 0.83984375,
      "learning_rate": 1.7751030536410264e-05,
      "loss": 0.3632,
      "step": 5792
    },
    {
      "epoch": 0.4384276997303817,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7750277915587358e-05,
      "loss": 0.3324,
      "step": 5793
    },
    {
      "epoch": 0.43850338205382905,
      "grad_norm": 0.69921875,
      "learning_rate": 1.7749525184813456e-05,
      "loss": 0.2586,
      "step": 5794
    },
    {
      "epoch": 0.4385790643772764,
      "grad_norm": 0.83984375,
      "learning_rate": 1.7748772344099242e-05,
      "loss": 0.3647,
      "step": 5795
    },
    {
      "epoch": 0.4386547467007237,
      "grad_norm": 0.8046875,
      "learning_rate": 1.77480193934554e-05,
      "loss": 0.3131,
      "step": 5796
    },
    {
      "epoch": 0.43873042902417103,
      "grad_norm": 0.8359375,
      "learning_rate": 1.77472663328926e-05,
      "loss": 0.3705,
      "step": 5797
    },
    {
      "epoch": 0.4388061113476184,
      "grad_norm": 0.734375,
      "learning_rate": 1.7746513162421535e-05,
      "loss": 0.2732,
      "step": 5798
    },
    {
      "epoch": 0.4388817936710657,
      "grad_norm": 0.74609375,
      "learning_rate": 1.7745759882052887e-05,
      "loss": 0.302,
      "step": 5799
    },
    {
      "epoch": 0.438957475994513,
      "grad_norm": 0.80859375,
      "learning_rate": 1.7745006491797345e-05,
      "loss": 0.3347,
      "step": 5800
    },
    {
      "epoch": 0.43903315831796036,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7744252991665592e-05,
      "loss": 0.3201,
      "step": 5801
    },
    {
      "epoch": 0.4391088406414077,
      "grad_norm": 0.7265625,
      "learning_rate": 1.774349938166832e-05,
      "loss": 0.2878,
      "step": 5802
    },
    {
      "epoch": 0.439184522964855,
      "grad_norm": 0.98828125,
      "learning_rate": 1.7742745661816222e-05,
      "loss": 0.3563,
      "step": 5803
    },
    {
      "epoch": 0.43926020528830234,
      "grad_norm": 0.71484375,
      "learning_rate": 1.7741991832119993e-05,
      "loss": 0.2853,
      "step": 5804
    },
    {
      "epoch": 0.4393358876117497,
      "grad_norm": 0.765625,
      "learning_rate": 1.774123789259032e-05,
      "loss": 0.3137,
      "step": 5805
    },
    {
      "epoch": 0.43941156993519703,
      "grad_norm": 0.71484375,
      "learning_rate": 1.7740483843237905e-05,
      "loss": 0.2924,
      "step": 5806
    },
    {
      "epoch": 0.4394872522586443,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7739729684073442e-05,
      "loss": 0.3541,
      "step": 5807
    },
    {
      "epoch": 0.43956293458209167,
      "grad_norm": 0.73046875,
      "learning_rate": 1.7738975415107634e-05,
      "loss": 0.2791,
      "step": 5808
    },
    {
      "epoch": 0.439638616905539,
      "grad_norm": 0.80078125,
      "learning_rate": 1.773822103635118e-05,
      "loss": 0.3505,
      "step": 5809
    },
    {
      "epoch": 0.4397142992289863,
      "grad_norm": 2.25,
      "learning_rate": 1.773746654781478e-05,
      "loss": 0.4116,
      "step": 5810
    },
    {
      "epoch": 0.43978998155243365,
      "grad_norm": 0.78125,
      "learning_rate": 1.7736711949509137e-05,
      "loss": 0.3585,
      "step": 5811
    },
    {
      "epoch": 0.439865663875881,
      "grad_norm": 0.75390625,
      "learning_rate": 1.773595724144496e-05,
      "loss": 0.325,
      "step": 5812
    },
    {
      "epoch": 0.43994134619932834,
      "grad_norm": 0.8046875,
      "learning_rate": 1.7735202423632952e-05,
      "loss": 0.3452,
      "step": 5813
    },
    {
      "epoch": 0.44001702852277563,
      "grad_norm": 0.8359375,
      "learning_rate": 1.773444749608383e-05,
      "loss": 0.3539,
      "step": 5814
    },
    {
      "epoch": 0.440092710846223,
      "grad_norm": 0.828125,
      "learning_rate": 1.7733692458808295e-05,
      "loss": 0.3548,
      "step": 5815
    },
    {
      "epoch": 0.4401683931696703,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7732937311817055e-05,
      "loss": 0.3285,
      "step": 5816
    },
    {
      "epoch": 0.4402440754931176,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7732182055120835e-05,
      "loss": 0.3504,
      "step": 5817
    },
    {
      "epoch": 0.44031975781656496,
      "grad_norm": 0.76171875,
      "learning_rate": 1.773142668873034e-05,
      "loss": 0.3204,
      "step": 5818
    },
    {
      "epoch": 0.4403954401400123,
      "grad_norm": 0.75390625,
      "learning_rate": 1.7730671212656295e-05,
      "loss": 0.3338,
      "step": 5819
    },
    {
      "epoch": 0.44047112246345965,
      "grad_norm": 0.76953125,
      "learning_rate": 1.772991562690941e-05,
      "loss": 0.3323,
      "step": 5820
    },
    {
      "epoch": 0.44054680478690694,
      "grad_norm": 1.03125,
      "learning_rate": 1.7729159931500405e-05,
      "loss": 0.3896,
      "step": 5821
    },
    {
      "epoch": 0.4406224871103543,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7728404126440005e-05,
      "loss": 0.3095,
      "step": 5822
    },
    {
      "epoch": 0.44069816943380163,
      "grad_norm": 0.8984375,
      "learning_rate": 1.7727648211738925e-05,
      "loss": 0.4083,
      "step": 5823
    },
    {
      "epoch": 0.4407738517572489,
      "grad_norm": 0.80859375,
      "learning_rate": 1.77268921874079e-05,
      "loss": 0.3674,
      "step": 5824
    },
    {
      "epoch": 0.44084953408069627,
      "grad_norm": 0.921875,
      "learning_rate": 1.7726136053457647e-05,
      "loss": 0.4043,
      "step": 5825
    },
    {
      "epoch": 0.4409252164041436,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7725379809898898e-05,
      "loss": 0.3175,
      "step": 5826
    },
    {
      "epoch": 0.44100089872759096,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7724623456742377e-05,
      "loss": 0.3302,
      "step": 5827
    },
    {
      "epoch": 0.44107658105103825,
      "grad_norm": 1.21875,
      "learning_rate": 1.7723866993998813e-05,
      "loss": 0.3642,
      "step": 5828
    },
    {
      "epoch": 0.4411522633744856,
      "grad_norm": 0.8125,
      "learning_rate": 1.7723110421678948e-05,
      "loss": 0.315,
      "step": 5829
    },
    {
      "epoch": 0.44122794569793294,
      "grad_norm": 0.82421875,
      "learning_rate": 1.7722353739793504e-05,
      "loss": 0.3409,
      "step": 5830
    },
    {
      "epoch": 0.44130362802138023,
      "grad_norm": 0.765625,
      "learning_rate": 1.772159694835322e-05,
      "loss": 0.3092,
      "step": 5831
    },
    {
      "epoch": 0.4413793103448276,
      "grad_norm": 0.74609375,
      "learning_rate": 1.7720840047368834e-05,
      "loss": 0.3209,
      "step": 5832
    },
    {
      "epoch": 0.4414549926682749,
      "grad_norm": 0.80859375,
      "learning_rate": 1.7720083036851083e-05,
      "loss": 0.3469,
      "step": 5833
    },
    {
      "epoch": 0.44153067499172227,
      "grad_norm": 0.73828125,
      "learning_rate": 1.7719325916810705e-05,
      "loss": 0.2983,
      "step": 5834
    },
    {
      "epoch": 0.44160635731516956,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7718568687258443e-05,
      "loss": 0.3524,
      "step": 5835
    },
    {
      "epoch": 0.4416820396386169,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7717811348205035e-05,
      "loss": 0.3118,
      "step": 5836
    },
    {
      "epoch": 0.44175772196206425,
      "grad_norm": 0.8125,
      "learning_rate": 1.7717053899661234e-05,
      "loss": 0.3383,
      "step": 5837
    },
    {
      "epoch": 0.44183340428551154,
      "grad_norm": 0.82421875,
      "learning_rate": 1.771629634163778e-05,
      "loss": 0.362,
      "step": 5838
    },
    {
      "epoch": 0.4419090866089589,
      "grad_norm": 1.21875,
      "learning_rate": 1.7715538674145416e-05,
      "loss": 0.3547,
      "step": 5839
    },
    {
      "epoch": 0.44198476893240624,
      "grad_norm": 0.80078125,
      "learning_rate": 1.77147808971949e-05,
      "loss": 0.3556,
      "step": 5840
    },
    {
      "epoch": 0.4420604512558536,
      "grad_norm": 0.8671875,
      "learning_rate": 1.7714023010796976e-05,
      "loss": 0.3722,
      "step": 5841
    },
    {
      "epoch": 0.44213613357930087,
      "grad_norm": 0.78515625,
      "learning_rate": 1.77132650149624e-05,
      "loss": 0.3262,
      "step": 5842
    },
    {
      "epoch": 0.4422118159027482,
      "grad_norm": 0.796875,
      "learning_rate": 1.771250690970192e-05,
      "loss": 0.3021,
      "step": 5843
    },
    {
      "epoch": 0.44228749822619556,
      "grad_norm": 0.80859375,
      "learning_rate": 1.7711748695026298e-05,
      "loss": 0.3651,
      "step": 5844
    },
    {
      "epoch": 0.44236318054964285,
      "grad_norm": 0.73046875,
      "learning_rate": 1.7710990370946285e-05,
      "loss": 0.319,
      "step": 5845
    },
    {
      "epoch": 0.4424388628730902,
      "grad_norm": 0.7734375,
      "learning_rate": 1.771023193747264e-05,
      "loss": 0.2739,
      "step": 5846
    },
    {
      "epoch": 0.44251454519653755,
      "grad_norm": 0.875,
      "learning_rate": 1.7709473394616126e-05,
      "loss": 0.3245,
      "step": 5847
    },
    {
      "epoch": 0.4425902275199849,
      "grad_norm": 0.82421875,
      "learning_rate": 1.7708714742387502e-05,
      "loss": 0.3729,
      "step": 5848
    },
    {
      "epoch": 0.4426659098434322,
      "grad_norm": 0.78125,
      "learning_rate": 1.770795598079753e-05,
      "loss": 0.3373,
      "step": 5849
    },
    {
      "epoch": 0.4427415921668795,
      "grad_norm": 0.7265625,
      "learning_rate": 1.7707197109856972e-05,
      "loss": 0.2986,
      "step": 5850
    },
    {
      "epoch": 0.4428172744903269,
      "grad_norm": 0.734375,
      "learning_rate": 1.77064381295766e-05,
      "loss": 0.2966,
      "step": 5851
    },
    {
      "epoch": 0.44289295681377416,
      "grad_norm": 0.76171875,
      "learning_rate": 1.770567903996718e-05,
      "loss": 0.3235,
      "step": 5852
    },
    {
      "epoch": 0.4429686391372215,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7704919841039477e-05,
      "loss": 0.3277,
      "step": 5853
    },
    {
      "epoch": 0.44304432146066886,
      "grad_norm": 0.828125,
      "learning_rate": 1.7704160532804266e-05,
      "loss": 0.3474,
      "step": 5854
    },
    {
      "epoch": 0.44312000378411615,
      "grad_norm": 0.8359375,
      "learning_rate": 1.7703401115272317e-05,
      "loss": 0.3197,
      "step": 5855
    },
    {
      "epoch": 0.4431956861075635,
      "grad_norm": 0.79296875,
      "learning_rate": 1.77026415884544e-05,
      "loss": 0.3152,
      "step": 5856
    },
    {
      "epoch": 0.44327136843101084,
      "grad_norm": 0.828125,
      "learning_rate": 1.7701881952361296e-05,
      "loss": 0.343,
      "step": 5857
    },
    {
      "epoch": 0.4433470507544582,
      "grad_norm": 0.75,
      "learning_rate": 1.770112220700378e-05,
      "loss": 0.2861,
      "step": 5858
    },
    {
      "epoch": 0.4434227330779055,
      "grad_norm": 0.74609375,
      "learning_rate": 1.7700362352392632e-05,
      "loss": 0.2982,
      "step": 5859
    },
    {
      "epoch": 0.4434984154013528,
      "grad_norm": 0.7578125,
      "learning_rate": 1.7699602388538622e-05,
      "loss": 0.3397,
      "step": 5860
    },
    {
      "epoch": 0.44357409772480016,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7698842315452548e-05,
      "loss": 0.2991,
      "step": 5861
    },
    {
      "epoch": 0.44364978004824746,
      "grad_norm": 0.8203125,
      "learning_rate": 1.769808213314518e-05,
      "loss": 0.3139,
      "step": 5862
    },
    {
      "epoch": 0.4437254623716948,
      "grad_norm": 0.78515625,
      "learning_rate": 1.769732184162731e-05,
      "loss": 0.3192,
      "step": 5863
    },
    {
      "epoch": 0.44380114469514215,
      "grad_norm": 0.78125,
      "learning_rate": 1.7696561440909715e-05,
      "loss": 0.3142,
      "step": 5864
    },
    {
      "epoch": 0.4438768270185895,
      "grad_norm": 1.1171875,
      "learning_rate": 1.769580093100319e-05,
      "loss": 0.3681,
      "step": 5865
    },
    {
      "epoch": 0.4439525093420368,
      "grad_norm": 0.828125,
      "learning_rate": 1.7695040311918526e-05,
      "loss": 0.3441,
      "step": 5866
    },
    {
      "epoch": 0.44402819166548413,
      "grad_norm": 0.80859375,
      "learning_rate": 1.7694279583666508e-05,
      "loss": 0.3457,
      "step": 5867
    },
    {
      "epoch": 0.4441038739889315,
      "grad_norm": 0.77734375,
      "learning_rate": 1.769351874625793e-05,
      "loss": 0.3345,
      "step": 5868
    },
    {
      "epoch": 0.44417955631237876,
      "grad_norm": 0.796875,
      "learning_rate": 1.7692757799703586e-05,
      "loss": 0.3283,
      "step": 5869
    },
    {
      "epoch": 0.4442552386358261,
      "grad_norm": 0.79296875,
      "learning_rate": 1.769199674401427e-05,
      "loss": 0.336,
      "step": 5870
    },
    {
      "epoch": 0.44433092095927346,
      "grad_norm": 0.81640625,
      "learning_rate": 1.7691235579200784e-05,
      "loss": 0.3754,
      "step": 5871
    },
    {
      "epoch": 0.4444066032827208,
      "grad_norm": 0.80859375,
      "learning_rate": 1.769047430527392e-05,
      "loss": 0.3256,
      "step": 5872
    },
    {
      "epoch": 0.4444822856061681,
      "grad_norm": 0.8359375,
      "learning_rate": 1.768971292224448e-05,
      "loss": 0.3697,
      "step": 5873
    },
    {
      "epoch": 0.44455796792961544,
      "grad_norm": 0.7578125,
      "learning_rate": 1.768895143012327e-05,
      "loss": 0.3097,
      "step": 5874
    },
    {
      "epoch": 0.4446336502530628,
      "grad_norm": 0.8125,
      "learning_rate": 1.7688189828921083e-05,
      "loss": 0.3638,
      "step": 5875
    },
    {
      "epoch": 0.4447093325765101,
      "grad_norm": 0.7578125,
      "learning_rate": 1.7687428118648732e-05,
      "loss": 0.3353,
      "step": 5876
    },
    {
      "epoch": 0.4447850148999574,
      "grad_norm": 0.71875,
      "learning_rate": 1.7686666299317024e-05,
      "loss": 0.2962,
      "step": 5877
    },
    {
      "epoch": 0.44486069722340477,
      "grad_norm": 0.74609375,
      "learning_rate": 1.768590437093676e-05,
      "loss": 0.3173,
      "step": 5878
    },
    {
      "epoch": 0.4449363795468521,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7685142333518755e-05,
      "loss": 0.3409,
      "step": 5879
    },
    {
      "epoch": 0.4450120618702994,
      "grad_norm": 0.8671875,
      "learning_rate": 1.768438018707382e-05,
      "loss": 0.3659,
      "step": 5880
    },
    {
      "epoch": 0.44508774419374675,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7683617931612763e-05,
      "loss": 0.3236,
      "step": 5881
    },
    {
      "epoch": 0.4451634265171941,
      "grad_norm": 0.78125,
      "learning_rate": 1.76828555671464e-05,
      "loss": 0.3015,
      "step": 5882
    },
    {
      "epoch": 0.4452391088406414,
      "grad_norm": 0.75,
      "learning_rate": 1.7682093093685546e-05,
      "loss": 0.3122,
      "step": 5883
    },
    {
      "epoch": 0.44531479116408873,
      "grad_norm": 0.8359375,
      "learning_rate": 1.768133051124102e-05,
      "loss": 0.3953,
      "step": 5884
    },
    {
      "epoch": 0.4453904734875361,
      "grad_norm": 0.74609375,
      "learning_rate": 1.768056781982364e-05,
      "loss": 0.3,
      "step": 5885
    },
    {
      "epoch": 0.4454661558109834,
      "grad_norm": 0.78125,
      "learning_rate": 1.7679805019444224e-05,
      "loss": 0.3252,
      "step": 5886
    },
    {
      "epoch": 0.4455418381344307,
      "grad_norm": 0.75,
      "learning_rate": 1.7679042110113595e-05,
      "loss": 0.3333,
      "step": 5887
    },
    {
      "epoch": 0.44561752045787806,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7678279091842575e-05,
      "loss": 0.3126,
      "step": 5888
    },
    {
      "epoch": 0.4456932027813254,
      "grad_norm": 0.8515625,
      "learning_rate": 1.7677515964641987e-05,
      "loss": 0.364,
      "step": 5889
    },
    {
      "epoch": 0.4457688851047727,
      "grad_norm": 0.890625,
      "learning_rate": 1.7676752728522662e-05,
      "loss": 0.3293,
      "step": 5890
    },
    {
      "epoch": 0.44584456742822004,
      "grad_norm": 0.8203125,
      "learning_rate": 1.767598938349543e-05,
      "loss": 0.3449,
      "step": 5891
    },
    {
      "epoch": 0.4459202497516674,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7675225929571112e-05,
      "loss": 0.3523,
      "step": 5892
    },
    {
      "epoch": 0.44599593207511473,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7674462366760544e-05,
      "loss": 0.2997,
      "step": 5893
    },
    {
      "epoch": 0.446071614398562,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7673698695074557e-05,
      "loss": 0.338,
      "step": 5894
    },
    {
      "epoch": 0.44614729672200937,
      "grad_norm": 0.84765625,
      "learning_rate": 1.7672934914523985e-05,
      "loss": 0.3803,
      "step": 5895
    },
    {
      "epoch": 0.4462229790454567,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7672171025119667e-05,
      "loss": 0.3598,
      "step": 5896
    },
    {
      "epoch": 0.446298661368904,
      "grad_norm": 0.6953125,
      "learning_rate": 1.7671407026872436e-05,
      "loss": 0.2599,
      "step": 5897
    },
    {
      "epoch": 0.44637434369235135,
      "grad_norm": 0.81640625,
      "learning_rate": 1.767064291979313e-05,
      "loss": 0.3433,
      "step": 5898
    },
    {
      "epoch": 0.4464500260157987,
      "grad_norm": 0.77734375,
      "learning_rate": 1.766987870389259e-05,
      "loss": 0.3503,
      "step": 5899
    },
    {
      "epoch": 0.44652570833924604,
      "grad_norm": 0.75,
      "learning_rate": 1.7669114379181663e-05,
      "loss": 0.2969,
      "step": 5900
    },
    {
      "epoch": 0.44660139066269333,
      "grad_norm": 0.70703125,
      "learning_rate": 1.7668349945671183e-05,
      "loss": 0.2948,
      "step": 5901
    },
    {
      "epoch": 0.4466770729861407,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7667585403372003e-05,
      "loss": 0.3166,
      "step": 5902
    },
    {
      "epoch": 0.446752755309588,
      "grad_norm": 0.85546875,
      "learning_rate": 1.7666820752294965e-05,
      "loss": 0.3633,
      "step": 5903
    },
    {
      "epoch": 0.4468284376330353,
      "grad_norm": 0.73828125,
      "learning_rate": 1.7666055992450917e-05,
      "loss": 0.2968,
      "step": 5904
    },
    {
      "epoch": 0.44690411995648266,
      "grad_norm": 1.359375,
      "learning_rate": 1.7665291123850712e-05,
      "loss": 0.393,
      "step": 5905
    },
    {
      "epoch": 0.44697980227993,
      "grad_norm": 1.109375,
      "learning_rate": 1.7664526146505195e-05,
      "loss": 0.3306,
      "step": 5906
    },
    {
      "epoch": 0.44705548460337735,
      "grad_norm": 0.78125,
      "learning_rate": 1.7663761060425225e-05,
      "loss": 0.3245,
      "step": 5907
    },
    {
      "epoch": 0.44713116692682464,
      "grad_norm": 1.1015625,
      "learning_rate": 1.7662995865621654e-05,
      "loss": 0.407,
      "step": 5908
    },
    {
      "epoch": 0.447206849250272,
      "grad_norm": 0.77734375,
      "learning_rate": 1.7662230562105328e-05,
      "loss": 0.3328,
      "step": 5909
    },
    {
      "epoch": 0.44728253157371933,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7661465149887123e-05,
      "loss": 0.3514,
      "step": 5910
    },
    {
      "epoch": 0.4473582138971666,
      "grad_norm": 0.8359375,
      "learning_rate": 1.7660699628977884e-05,
      "loss": 0.2934,
      "step": 5911
    },
    {
      "epoch": 0.44743389622061397,
      "grad_norm": 0.765625,
      "learning_rate": 1.765993399938847e-05,
      "loss": 0.3276,
      "step": 5912
    },
    {
      "epoch": 0.4475095785440613,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7659168261129752e-05,
      "loss": 0.3662,
      "step": 5913
    },
    {
      "epoch": 0.4475852608675086,
      "grad_norm": 1.734375,
      "learning_rate": 1.7658402414212585e-05,
      "loss": 0.3726,
      "step": 5914
    },
    {
      "epoch": 0.44766094319095595,
      "grad_norm": 0.765625,
      "learning_rate": 1.7657636458647843e-05,
      "loss": 0.3108,
      "step": 5915
    },
    {
      "epoch": 0.4477366255144033,
      "grad_norm": 0.84375,
      "learning_rate": 1.7656870394446382e-05,
      "loss": 0.3344,
      "step": 5916
    },
    {
      "epoch": 0.44781230783785064,
      "grad_norm": 0.87109375,
      "learning_rate": 1.7656104221619075e-05,
      "loss": 0.3046,
      "step": 5917
    },
    {
      "epoch": 0.44788799016129793,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7655337940176795e-05,
      "loss": 0.3124,
      "step": 5918
    },
    {
      "epoch": 0.4479636724847453,
      "grad_norm": 0.8359375,
      "learning_rate": 1.7654571550130406e-05,
      "loss": 0.3476,
      "step": 5919
    },
    {
      "epoch": 0.4480393548081926,
      "grad_norm": 0.8125,
      "learning_rate": 1.7653805051490785e-05,
      "loss": 0.3447,
      "step": 5920
    },
    {
      "epoch": 0.4481150371316399,
      "grad_norm": 0.8359375,
      "learning_rate": 1.7653038444268804e-05,
      "loss": 0.3547,
      "step": 5921
    },
    {
      "epoch": 0.44819071945508726,
      "grad_norm": 0.8515625,
      "learning_rate": 1.7652271728475342e-05,
      "loss": 0.3521,
      "step": 5922
    },
    {
      "epoch": 0.4482664017785346,
      "grad_norm": 0.80078125,
      "learning_rate": 1.765150490412127e-05,
      "loss": 0.3451,
      "step": 5923
    },
    {
      "epoch": 0.44834208410198195,
      "grad_norm": 0.90625,
      "learning_rate": 1.7650737971217476e-05,
      "loss": 0.3414,
      "step": 5924
    },
    {
      "epoch": 0.44841776642542924,
      "grad_norm": 0.796875,
      "learning_rate": 1.764997092977483e-05,
      "loss": 0.3362,
      "step": 5925
    },
    {
      "epoch": 0.4484934487488766,
      "grad_norm": 1.0390625,
      "learning_rate": 1.764920377980422e-05,
      "loss": 0.3545,
      "step": 5926
    },
    {
      "epoch": 0.44856913107232393,
      "grad_norm": 0.8046875,
      "learning_rate": 1.7648436521316528e-05,
      "loss": 0.3521,
      "step": 5927
    },
    {
      "epoch": 0.4486448133957712,
      "grad_norm": 0.69140625,
      "learning_rate": 1.7647669154322636e-05,
      "loss": 0.2588,
      "step": 5928
    },
    {
      "epoch": 0.44872049571921857,
      "grad_norm": 0.71875,
      "learning_rate": 1.7646901678833435e-05,
      "loss": 0.2724,
      "step": 5929
    },
    {
      "epoch": 0.4487961780426659,
      "grad_norm": 0.72265625,
      "learning_rate": 1.7646134094859816e-05,
      "loss": 0.2657,
      "step": 5930
    },
    {
      "epoch": 0.44887186036611326,
      "grad_norm": 0.828125,
      "learning_rate": 1.764536640241266e-05,
      "loss": 0.341,
      "step": 5931
    },
    {
      "epoch": 0.44894754268956055,
      "grad_norm": 0.7578125,
      "learning_rate": 1.764459860150286e-05,
      "loss": 0.2932,
      "step": 5932
    },
    {
      "epoch": 0.4490232250130079,
      "grad_norm": 0.87109375,
      "learning_rate": 1.764383069214131e-05,
      "loss": 0.3515,
      "step": 5933
    },
    {
      "epoch": 0.44909890733645524,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7643062674338908e-05,
      "loss": 0.3051,
      "step": 5934
    },
    {
      "epoch": 0.44917458965990253,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7642294548106545e-05,
      "loss": 0.3154,
      "step": 5935
    },
    {
      "epoch": 0.4492502719833499,
      "grad_norm": 0.78125,
      "learning_rate": 1.7641526313455118e-05,
      "loss": 0.3151,
      "step": 5936
    },
    {
      "epoch": 0.4493259543067972,
      "grad_norm": 0.796875,
      "learning_rate": 1.764075797039553e-05,
      "loss": 0.3484,
      "step": 5937
    },
    {
      "epoch": 0.44940163663024457,
      "grad_norm": 0.82421875,
      "learning_rate": 1.7639989518938673e-05,
      "loss": 0.3309,
      "step": 5938
    },
    {
      "epoch": 0.44947731895369186,
      "grad_norm": 0.76171875,
      "learning_rate": 1.7639220959095457e-05,
      "loss": 0.2958,
      "step": 5939
    },
    {
      "epoch": 0.4495530012771392,
      "grad_norm": 0.8046875,
      "learning_rate": 1.763845229087678e-05,
      "loss": 0.3246,
      "step": 5940
    },
    {
      "epoch": 0.44962868360058655,
      "grad_norm": 0.796875,
      "learning_rate": 1.7637683514293553e-05,
      "loss": 0.324,
      "step": 5941
    },
    {
      "epoch": 0.44970436592403384,
      "grad_norm": 0.8515625,
      "learning_rate": 1.7636914629356673e-05,
      "loss": 0.3607,
      "step": 5942
    },
    {
      "epoch": 0.4497800482474812,
      "grad_norm": 0.828125,
      "learning_rate": 1.763614563607706e-05,
      "loss": 0.3356,
      "step": 5943
    },
    {
      "epoch": 0.44985573057092854,
      "grad_norm": 0.796875,
      "learning_rate": 1.7635376534465614e-05,
      "loss": 0.3231,
      "step": 5944
    },
    {
      "epoch": 0.4499314128943759,
      "grad_norm": 0.8359375,
      "learning_rate": 1.763460732453325e-05,
      "loss": 0.3524,
      "step": 5945
    },
    {
      "epoch": 0.45000709521782317,
      "grad_norm": 0.8125,
      "learning_rate": 1.7633838006290878e-05,
      "loss": 0.3467,
      "step": 5946
    },
    {
      "epoch": 0.4500827775412705,
      "grad_norm": 0.84375,
      "learning_rate": 1.7633068579749415e-05,
      "loss": 0.3218,
      "step": 5947
    },
    {
      "epoch": 0.45015845986471786,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7632299044919775e-05,
      "loss": 0.3478,
      "step": 5948
    },
    {
      "epoch": 0.45023414218816515,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7631529401812875e-05,
      "loss": 0.3224,
      "step": 5949
    },
    {
      "epoch": 0.4503098245116125,
      "grad_norm": 0.76171875,
      "learning_rate": 1.7630759650439638e-05,
      "loss": 0.287,
      "step": 5950
    },
    {
      "epoch": 0.45038550683505985,
      "grad_norm": 0.75390625,
      "learning_rate": 1.762998979081097e-05,
      "loss": 0.3077,
      "step": 5951
    },
    {
      "epoch": 0.4504611891585072,
      "grad_norm": 0.85546875,
      "learning_rate": 1.7629219822937817e-05,
      "loss": 0.3411,
      "step": 5952
    },
    {
      "epoch": 0.4505368714819545,
      "grad_norm": 0.78515625,
      "learning_rate": 1.762844974683108e-05,
      "loss": 0.2966,
      "step": 5953
    },
    {
      "epoch": 0.4506125538054018,
      "grad_norm": 0.796875,
      "learning_rate": 1.7627679562501695e-05,
      "loss": 0.3303,
      "step": 5954
    },
    {
      "epoch": 0.4506882361288492,
      "grad_norm": 0.7421875,
      "learning_rate": 1.7626909269960588e-05,
      "loss": 0.309,
      "step": 5955
    },
    {
      "epoch": 0.45076391845229646,
      "grad_norm": 0.80078125,
      "learning_rate": 1.762613886921868e-05,
      "loss": 0.2958,
      "step": 5956
    },
    {
      "epoch": 0.4508396007757438,
      "grad_norm": 0.80078125,
      "learning_rate": 1.762536836028691e-05,
      "loss": 0.3321,
      "step": 5957
    },
    {
      "epoch": 0.45091528309919116,
      "grad_norm": 0.6953125,
      "learning_rate": 1.76245977431762e-05,
      "loss": 0.2534,
      "step": 5958
    },
    {
      "epoch": 0.4509909654226385,
      "grad_norm": 0.78125,
      "learning_rate": 1.7623827017897494e-05,
      "loss": 0.3273,
      "step": 5959
    },
    {
      "epoch": 0.4510666477460858,
      "grad_norm": 0.78515625,
      "learning_rate": 1.762305618446171e-05,
      "loss": 0.2873,
      "step": 5960
    },
    {
      "epoch": 0.45114233006953314,
      "grad_norm": 0.8125,
      "learning_rate": 1.7622285242879797e-05,
      "loss": 0.3538,
      "step": 5961
    },
    {
      "epoch": 0.4512180123929805,
      "grad_norm": 0.765625,
      "learning_rate": 1.7621514193162688e-05,
      "loss": 0.3234,
      "step": 5962
    },
    {
      "epoch": 0.4512936947164278,
      "grad_norm": 1.1171875,
      "learning_rate": 1.7620743035321318e-05,
      "loss": 0.4017,
      "step": 5963
    },
    {
      "epoch": 0.4513693770398751,
      "grad_norm": 0.8046875,
      "learning_rate": 1.7619971769366634e-05,
      "loss": 0.3522,
      "step": 5964
    },
    {
      "epoch": 0.45144505936332247,
      "grad_norm": 0.76953125,
      "learning_rate": 1.761920039530957e-05,
      "loss": 0.2934,
      "step": 5965
    },
    {
      "epoch": 0.4515207416867698,
      "grad_norm": 0.80078125,
      "learning_rate": 1.7618428913161078e-05,
      "loss": 0.2924,
      "step": 5966
    },
    {
      "epoch": 0.4515964240102171,
      "grad_norm": 0.75390625,
      "learning_rate": 1.7617657322932092e-05,
      "loss": 0.3031,
      "step": 5967
    },
    {
      "epoch": 0.45167210633366445,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7616885624633567e-05,
      "loss": 0.3042,
      "step": 5968
    },
    {
      "epoch": 0.4517477886571118,
      "grad_norm": 0.8203125,
      "learning_rate": 1.7616113818276454e-05,
      "loss": 0.3243,
      "step": 5969
    },
    {
      "epoch": 0.4518234709805591,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7615341903871693e-05,
      "loss": 0.3169,
      "step": 5970
    },
    {
      "epoch": 0.45189915330400643,
      "grad_norm": 0.7109375,
      "learning_rate": 1.7614569881430238e-05,
      "loss": 0.2701,
      "step": 5971
    },
    {
      "epoch": 0.4519748356274538,
      "grad_norm": 0.7578125,
      "learning_rate": 1.761379775096304e-05,
      "loss": 0.3199,
      "step": 5972
    },
    {
      "epoch": 0.45205051795090107,
      "grad_norm": 0.78515625,
      "learning_rate": 1.761302551248106e-05,
      "loss": 0.3294,
      "step": 5973
    },
    {
      "epoch": 0.4521262002743484,
      "grad_norm": 0.84375,
      "learning_rate": 1.7612253165995244e-05,
      "loss": 0.3645,
      "step": 5974
    },
    {
      "epoch": 0.45220188259779576,
      "grad_norm": 0.828125,
      "learning_rate": 1.7611480711516557e-05,
      "loss": 0.379,
      "step": 5975
    },
    {
      "epoch": 0.4522775649212431,
      "grad_norm": 0.7734375,
      "learning_rate": 1.761070814905595e-05,
      "loss": 0.3001,
      "step": 5976
    },
    {
      "epoch": 0.4523532472446904,
      "grad_norm": 0.734375,
      "learning_rate": 1.7609935478624394e-05,
      "loss": 0.2964,
      "step": 5977
    },
    {
      "epoch": 0.45242892956813774,
      "grad_norm": 0.8203125,
      "learning_rate": 1.7609162700232844e-05,
      "loss": 0.3943,
      "step": 5978
    },
    {
      "epoch": 0.4525046118915851,
      "grad_norm": 0.74609375,
      "learning_rate": 1.7608389813892257e-05,
      "loss": 0.3135,
      "step": 5979
    },
    {
      "epoch": 0.4525802942150324,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7607616819613608e-05,
      "loss": 0.34,
      "step": 5980
    },
    {
      "epoch": 0.4526559765384797,
      "grad_norm": 0.828125,
      "learning_rate": 1.760684371740786e-05,
      "loss": 0.3469,
      "step": 5981
    },
    {
      "epoch": 0.45273165886192707,
      "grad_norm": 0.80859375,
      "learning_rate": 1.7606070507285978e-05,
      "loss": 0.3274,
      "step": 5982
    },
    {
      "epoch": 0.4528073411853744,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7605297189258935e-05,
      "loss": 0.3174,
      "step": 5983
    },
    {
      "epoch": 0.4528830235088217,
      "grad_norm": 1.1875,
      "learning_rate": 1.76045237633377e-05,
      "loss": 0.339,
      "step": 5984
    },
    {
      "epoch": 0.45295870583226905,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7603750229533246e-05,
      "loss": 0.3232,
      "step": 5985
    },
    {
      "epoch": 0.4530343881557164,
      "grad_norm": 0.8125,
      "learning_rate": 1.7602976587856547e-05,
      "loss": 0.3088,
      "step": 5986
    },
    {
      "epoch": 0.4531100704791637,
      "grad_norm": 0.87109375,
      "learning_rate": 1.7602202838318575e-05,
      "loss": 0.3819,
      "step": 5987
    },
    {
      "epoch": 0.45318575280261103,
      "grad_norm": 0.76171875,
      "learning_rate": 1.760142898093031e-05,
      "loss": 0.3113,
      "step": 5988
    },
    {
      "epoch": 0.4532614351260584,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7600655015702733e-05,
      "loss": 0.3017,
      "step": 5989
    },
    {
      "epoch": 0.4533371174495057,
      "grad_norm": 0.77734375,
      "learning_rate": 1.759988094264682e-05,
      "loss": 0.3172,
      "step": 5990
    },
    {
      "epoch": 0.453412799772953,
      "grad_norm": 0.87890625,
      "learning_rate": 1.7599106761773554e-05,
      "loss": 0.387,
      "step": 5991
    },
    {
      "epoch": 0.45348848209640036,
      "grad_norm": 0.72265625,
      "learning_rate": 1.7598332473093917e-05,
      "loss": 0.2795,
      "step": 5992
    },
    {
      "epoch": 0.4535641644198477,
      "grad_norm": 0.8359375,
      "learning_rate": 1.7597558076618898e-05,
      "loss": 0.3325,
      "step": 5993
    },
    {
      "epoch": 0.453639846743295,
      "grad_norm": 0.77734375,
      "learning_rate": 1.7596783572359476e-05,
      "loss": 0.3306,
      "step": 5994
    },
    {
      "epoch": 0.45371552906674234,
      "grad_norm": 1.1171875,
      "learning_rate": 1.7596008960326644e-05,
      "loss": 0.3263,
      "step": 5995
    },
    {
      "epoch": 0.4537912113901897,
      "grad_norm": 0.8046875,
      "learning_rate": 1.759523424053139e-05,
      "loss": 0.3298,
      "step": 5996
    },
    {
      "epoch": 0.45386689371363703,
      "grad_norm": 0.75390625,
      "learning_rate": 1.7594459412984707e-05,
      "loss": 0.3067,
      "step": 5997
    },
    {
      "epoch": 0.4539425760370843,
      "grad_norm": 0.67578125,
      "learning_rate": 1.759368447769758e-05,
      "loss": 0.2615,
      "step": 5998
    },
    {
      "epoch": 0.45401825836053167,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7592909434681012e-05,
      "loss": 0.3247,
      "step": 5999
    },
    {
      "epoch": 0.454093940683979,
      "grad_norm": 0.7265625,
      "learning_rate": 1.759213428394599e-05,
      "loss": 0.2862,
      "step": 6000
    },
    {
      "epoch": 0.4541696230074263,
      "grad_norm": 0.82421875,
      "learning_rate": 1.7591359025503515e-05,
      "loss": 0.3837,
      "step": 6001
    },
    {
      "epoch": 0.45424530533087365,
      "grad_norm": 0.765625,
      "learning_rate": 1.7590583659364586e-05,
      "loss": 0.3112,
      "step": 6002
    },
    {
      "epoch": 0.454320987654321,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7589808185540206e-05,
      "loss": 0.3294,
      "step": 6003
    },
    {
      "epoch": 0.45439666997776834,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7589032604041368e-05,
      "loss": 0.3251,
      "step": 6004
    },
    {
      "epoch": 0.45447235230121563,
      "grad_norm": 0.73046875,
      "learning_rate": 1.758825691487908e-05,
      "loss": 0.2769,
      "step": 6005
    },
    {
      "epoch": 0.454548034624663,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7587481118064343e-05,
      "loss": 0.3154,
      "step": 6006
    },
    {
      "epoch": 0.4546237169481103,
      "grad_norm": 0.8125,
      "learning_rate": 1.758670521360817e-05,
      "loss": 0.3491,
      "step": 6007
    },
    {
      "epoch": 0.4546993992715576,
      "grad_norm": 0.84765625,
      "learning_rate": 1.7585929201521562e-05,
      "loss": 0.3316,
      "step": 6008
    },
    {
      "epoch": 0.45477508159500496,
      "grad_norm": 0.7890625,
      "learning_rate": 1.758515308181553e-05,
      "loss": 0.3464,
      "step": 6009
    },
    {
      "epoch": 0.4548507639184523,
      "grad_norm": 0.734375,
      "learning_rate": 1.7584376854501083e-05,
      "loss": 0.2831,
      "step": 6010
    },
    {
      "epoch": 0.45492644624189965,
      "grad_norm": 0.75,
      "learning_rate": 1.758360051958924e-05,
      "loss": 0.2897,
      "step": 6011
    },
    {
      "epoch": 0.45500212856534694,
      "grad_norm": 0.8125,
      "learning_rate": 1.758282407709101e-05,
      "loss": 0.334,
      "step": 6012
    },
    {
      "epoch": 0.4550778108887943,
      "grad_norm": 0.71484375,
      "learning_rate": 1.7582047527017402e-05,
      "loss": 0.3048,
      "step": 6013
    },
    {
      "epoch": 0.45515349321224163,
      "grad_norm": 0.80078125,
      "learning_rate": 1.7581270869379443e-05,
      "loss": 0.3141,
      "step": 6014
    },
    {
      "epoch": 0.4552291755356889,
      "grad_norm": 0.80078125,
      "learning_rate": 1.7580494104188145e-05,
      "loss": 0.3224,
      "step": 6015
    },
    {
      "epoch": 0.45530485785913627,
      "grad_norm": 0.8515625,
      "learning_rate": 1.757971723145453e-05,
      "loss": 0.3686,
      "step": 6016
    },
    {
      "epoch": 0.4553805401825836,
      "grad_norm": 0.80859375,
      "learning_rate": 1.757894025118962e-05,
      "loss": 0.3454,
      "step": 6017
    },
    {
      "epoch": 0.45545622250603096,
      "grad_norm": 0.78515625,
      "learning_rate": 1.757816316340444e-05,
      "loss": 0.3279,
      "step": 6018
    },
    {
      "epoch": 0.45553190482947825,
      "grad_norm": 0.8359375,
      "learning_rate": 1.7577385968110003e-05,
      "loss": 0.3791,
      "step": 6019
    },
    {
      "epoch": 0.4556075871529256,
      "grad_norm": 0.72265625,
      "learning_rate": 1.757660866531735e-05,
      "loss": 0.2899,
      "step": 6020
    },
    {
      "epoch": 0.45568326947637294,
      "grad_norm": 0.90234375,
      "learning_rate": 1.7575831255037494e-05,
      "loss": 0.3871,
      "step": 6021
    },
    {
      "epoch": 0.45575895179982023,
      "grad_norm": 0.796875,
      "learning_rate": 1.7575053737281474e-05,
      "loss": 0.3624,
      "step": 6022
    },
    {
      "epoch": 0.4558346341232676,
      "grad_norm": 0.77734375,
      "learning_rate": 1.757427611206032e-05,
      "loss": 0.3132,
      "step": 6023
    },
    {
      "epoch": 0.4559103164467149,
      "grad_norm": 0.86328125,
      "learning_rate": 1.757349837938506e-05,
      "loss": 0.3822,
      "step": 6024
    },
    {
      "epoch": 0.45598599877016227,
      "grad_norm": 1.2734375,
      "learning_rate": 1.757272053926673e-05,
      "loss": 0.364,
      "step": 6025
    },
    {
      "epoch": 0.45606168109360956,
      "grad_norm": 0.7578125,
      "learning_rate": 1.757194259171636e-05,
      "loss": 0.3006,
      "step": 6026
    },
    {
      "epoch": 0.4561373634170569,
      "grad_norm": 0.796875,
      "learning_rate": 1.7571164536744992e-05,
      "loss": 0.343,
      "step": 6027
    },
    {
      "epoch": 0.45621304574050425,
      "grad_norm": 0.76171875,
      "learning_rate": 1.7570386374363665e-05,
      "loss": 0.3012,
      "step": 6028
    },
    {
      "epoch": 0.45628872806395154,
      "grad_norm": 0.8203125,
      "learning_rate": 1.7569608104583416e-05,
      "loss": 0.3741,
      "step": 6029
    },
    {
      "epoch": 0.4563644103873989,
      "grad_norm": 0.94921875,
      "learning_rate": 1.7568829727415283e-05,
      "loss": 0.3083,
      "step": 6030
    },
    {
      "epoch": 0.45644009271084623,
      "grad_norm": 0.765625,
      "learning_rate": 1.7568051242870314e-05,
      "loss": 0.3148,
      "step": 6031
    },
    {
      "epoch": 0.4565157750342936,
      "grad_norm": 0.8203125,
      "learning_rate": 1.7567272650959552e-05,
      "loss": 0.3579,
      "step": 6032
    },
    {
      "epoch": 0.45659145735774087,
      "grad_norm": 0.94921875,
      "learning_rate": 1.756649395169404e-05,
      "loss": 0.3893,
      "step": 6033
    },
    {
      "epoch": 0.4566671396811882,
      "grad_norm": 0.83203125,
      "learning_rate": 1.7565715145084826e-05,
      "loss": 0.3761,
      "step": 6034
    },
    {
      "epoch": 0.45674282200463556,
      "grad_norm": 0.8515625,
      "learning_rate": 1.7564936231142962e-05,
      "loss": 0.3664,
      "step": 6035
    },
    {
      "epoch": 0.45681850432808285,
      "grad_norm": 0.82421875,
      "learning_rate": 1.7564157209879493e-05,
      "loss": 0.3712,
      "step": 6036
    },
    {
      "epoch": 0.4568941866515302,
      "grad_norm": 0.78125,
      "learning_rate": 1.7563378081305475e-05,
      "loss": 0.3245,
      "step": 6037
    },
    {
      "epoch": 0.45696986897497754,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7562598845431958e-05,
      "loss": 0.3054,
      "step": 6038
    },
    {
      "epoch": 0.45704555129842483,
      "grad_norm": 0.75,
      "learning_rate": 1.7561819502270003e-05,
      "loss": 0.2846,
      "step": 6039
    },
    {
      "epoch": 0.4571212336218722,
      "grad_norm": 0.765625,
      "learning_rate": 1.7561040051830654e-05,
      "loss": 0.3014,
      "step": 6040
    },
    {
      "epoch": 0.4571969159453195,
      "grad_norm": 0.83984375,
      "learning_rate": 1.756026049412498e-05,
      "loss": 0.3643,
      "step": 6041
    },
    {
      "epoch": 0.45727259826876687,
      "grad_norm": 0.8984375,
      "learning_rate": 1.755948082916404e-05,
      "loss": 0.3549,
      "step": 6042
    },
    {
      "epoch": 0.45734828059221416,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7558701056958888e-05,
      "loss": 0.3278,
      "step": 6043
    },
    {
      "epoch": 0.4574239629156615,
      "grad_norm": 0.75,
      "learning_rate": 1.7557921177520593e-05,
      "loss": 0.2781,
      "step": 6044
    },
    {
      "epoch": 0.45749964523910885,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7557141190860217e-05,
      "loss": 0.3095,
      "step": 6045
    },
    {
      "epoch": 0.45757532756255614,
      "grad_norm": 0.77734375,
      "learning_rate": 1.7556361096988823e-05,
      "loss": 0.2715,
      "step": 6046
    },
    {
      "epoch": 0.4576510098860035,
      "grad_norm": 0.78125,
      "learning_rate": 1.7555580895917482e-05,
      "loss": 0.3309,
      "step": 6047
    },
    {
      "epoch": 0.45772669220945084,
      "grad_norm": 0.75390625,
      "learning_rate": 1.7554800587657256e-05,
      "loss": 0.3164,
      "step": 6048
    },
    {
      "epoch": 0.4578023745328982,
      "grad_norm": 0.7421875,
      "learning_rate": 1.7554020172219223e-05,
      "loss": 0.292,
      "step": 6049
    },
    {
      "epoch": 0.4578780568563455,
      "grad_norm": 0.7890625,
      "learning_rate": 1.755323964961445e-05,
      "loss": 0.3245,
      "step": 6050
    },
    {
      "epoch": 0.4579537391797928,
      "grad_norm": 0.84765625,
      "learning_rate": 1.7552459019854008e-05,
      "loss": 0.3876,
      "step": 6051
    },
    {
      "epoch": 0.45802942150324016,
      "grad_norm": 0.80859375,
      "learning_rate": 1.7551678282948974e-05,
      "loss": 0.3125,
      "step": 6052
    },
    {
      "epoch": 0.45810510382668745,
      "grad_norm": 0.8359375,
      "learning_rate": 1.7550897438910427e-05,
      "loss": 0.3541,
      "step": 6053
    },
    {
      "epoch": 0.4581807861501348,
      "grad_norm": 0.77734375,
      "learning_rate": 1.7550116487749442e-05,
      "loss": 0.3115,
      "step": 6054
    },
    {
      "epoch": 0.45825646847358215,
      "grad_norm": 0.85546875,
      "learning_rate": 1.75493354294771e-05,
      "loss": 0.3426,
      "step": 6055
    },
    {
      "epoch": 0.4583321507970295,
      "grad_norm": 0.796875,
      "learning_rate": 1.7548554264104477e-05,
      "loss": 0.3084,
      "step": 6056
    },
    {
      "epoch": 0.4584078331204768,
      "grad_norm": 0.6796875,
      "learning_rate": 1.754777299164266e-05,
      "loss": 0.2495,
      "step": 6057
    },
    {
      "epoch": 0.45848351544392413,
      "grad_norm": 0.78125,
      "learning_rate": 1.7546991612102727e-05,
      "loss": 0.2968,
      "step": 6058
    },
    {
      "epoch": 0.4585591977673715,
      "grad_norm": 0.796875,
      "learning_rate": 1.754621012549577e-05,
      "loss": 0.321,
      "step": 6059
    },
    {
      "epoch": 0.45863488009081876,
      "grad_norm": 0.73046875,
      "learning_rate": 1.754542853183287e-05,
      "loss": 0.2714,
      "step": 6060
    },
    {
      "epoch": 0.4587105624142661,
      "grad_norm": 0.80859375,
      "learning_rate": 1.754464683112512e-05,
      "loss": 0.3208,
      "step": 6061
    },
    {
      "epoch": 0.45878624473771346,
      "grad_norm": 0.76953125,
      "learning_rate": 1.754386502338361e-05,
      "loss": 0.3168,
      "step": 6062
    },
    {
      "epoch": 0.4588619270611608,
      "grad_norm": 0.7421875,
      "learning_rate": 1.7543083108619424e-05,
      "loss": 0.3035,
      "step": 6063
    },
    {
      "epoch": 0.4589376093846081,
      "grad_norm": 0.91015625,
      "learning_rate": 1.754230108684366e-05,
      "loss": 0.3984,
      "step": 6064
    },
    {
      "epoch": 0.45901329170805544,
      "grad_norm": 0.73046875,
      "learning_rate": 1.7541518958067417e-05,
      "loss": 0.2803,
      "step": 6065
    },
    {
      "epoch": 0.4590889740315028,
      "grad_norm": 0.80859375,
      "learning_rate": 1.7540736722301784e-05,
      "loss": 0.2949,
      "step": 6066
    },
    {
      "epoch": 0.4591646563549501,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7539954379557858e-05,
      "loss": 0.2929,
      "step": 6067
    },
    {
      "epoch": 0.4592403386783974,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7539171929846742e-05,
      "loss": 0.3393,
      "step": 6068
    },
    {
      "epoch": 0.45931602100184477,
      "grad_norm": 0.75,
      "learning_rate": 1.7538389373179533e-05,
      "loss": 0.3063,
      "step": 6069
    },
    {
      "epoch": 0.4593917033252921,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7537606709567336e-05,
      "loss": 0.3469,
      "step": 6070
    },
    {
      "epoch": 0.4594673856487394,
      "grad_norm": 0.81640625,
      "learning_rate": 1.753682393902125e-05,
      "loss": 0.3409,
      "step": 6071
    },
    {
      "epoch": 0.45954306797218675,
      "grad_norm": 0.921875,
      "learning_rate": 1.753604106155239e-05,
      "loss": 0.4333,
      "step": 6072
    },
    {
      "epoch": 0.4596187502956341,
      "grad_norm": 0.734375,
      "learning_rate": 1.7535258077171847e-05,
      "loss": 0.2874,
      "step": 6073
    },
    {
      "epoch": 0.4596944326190814,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7534474985890743e-05,
      "loss": 0.3483,
      "step": 6074
    },
    {
      "epoch": 0.45977011494252873,
      "grad_norm": 0.88671875,
      "learning_rate": 1.7533691787720177e-05,
      "loss": 0.3643,
      "step": 6075
    },
    {
      "epoch": 0.4598457972659761,
      "grad_norm": 0.68359375,
      "learning_rate": 1.7532908482671267e-05,
      "loss": 0.2662,
      "step": 6076
    },
    {
      "epoch": 0.4599214795894234,
      "grad_norm": 0.84375,
      "learning_rate": 1.753212507075512e-05,
      "loss": 0.3723,
      "step": 6077
    },
    {
      "epoch": 0.4599971619128707,
      "grad_norm": 0.8515625,
      "learning_rate": 1.753134155198286e-05,
      "loss": 0.368,
      "step": 6078
    },
    {
      "epoch": 0.46007284423631806,
      "grad_norm": 0.80859375,
      "learning_rate": 1.753055792636559e-05,
      "loss": 0.329,
      "step": 6079
    },
    {
      "epoch": 0.4601485265597654,
      "grad_norm": 0.75,
      "learning_rate": 1.7529774193914436e-05,
      "loss": 0.3062,
      "step": 6080
    },
    {
      "epoch": 0.4602242088832127,
      "grad_norm": 0.828125,
      "learning_rate": 1.752899035464051e-05,
      "loss": 0.3861,
      "step": 6081
    },
    {
      "epoch": 0.46029989120666004,
      "grad_norm": 0.765625,
      "learning_rate": 1.752820640855494e-05,
      "loss": 0.3211,
      "step": 6082
    },
    {
      "epoch": 0.4603755735301074,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7527422355668844e-05,
      "loss": 0.3404,
      "step": 6083
    },
    {
      "epoch": 0.46045125585355473,
      "grad_norm": 0.76171875,
      "learning_rate": 1.752663819599334e-05,
      "loss": 0.3183,
      "step": 6084
    },
    {
      "epoch": 0.460526938177002,
      "grad_norm": 0.78515625,
      "learning_rate": 1.752585392953956e-05,
      "loss": 0.341,
      "step": 6085
    },
    {
      "epoch": 0.46060262050044937,
      "grad_norm": 0.84765625,
      "learning_rate": 1.7525069556318627e-05,
      "loss": 0.3961,
      "step": 6086
    },
    {
      "epoch": 0.4606783028238967,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7524285076341667e-05,
      "loss": 0.3365,
      "step": 6087
    },
    {
      "epoch": 0.460753985147344,
      "grad_norm": 0.796875,
      "learning_rate": 1.7523500489619812e-05,
      "loss": 0.3295,
      "step": 6088
    },
    {
      "epoch": 0.46082966747079135,
      "grad_norm": 0.734375,
      "learning_rate": 1.7522715796164192e-05,
      "loss": 0.2646,
      "step": 6089
    },
    {
      "epoch": 0.4609053497942387,
      "grad_norm": 0.7890625,
      "learning_rate": 1.752193099598594e-05,
      "loss": 0.3383,
      "step": 6090
    },
    {
      "epoch": 0.46098103211768604,
      "grad_norm": 0.8125,
      "learning_rate": 1.7521146089096187e-05,
      "loss": 0.3659,
      "step": 6091
    },
    {
      "epoch": 0.46105671444113333,
      "grad_norm": 0.8515625,
      "learning_rate": 1.752036107550607e-05,
      "loss": 0.3722,
      "step": 6092
    },
    {
      "epoch": 0.4611323967645807,
      "grad_norm": 0.7578125,
      "learning_rate": 1.7519575955226728e-05,
      "loss": 0.3281,
      "step": 6093
    },
    {
      "epoch": 0.461208079088028,
      "grad_norm": 0.734375,
      "learning_rate": 1.7518790728269294e-05,
      "loss": 0.2915,
      "step": 6094
    },
    {
      "epoch": 0.4612837614114753,
      "grad_norm": 0.75,
      "learning_rate": 1.7518005394644913e-05,
      "loss": 0.3065,
      "step": 6095
    },
    {
      "epoch": 0.46135944373492266,
      "grad_norm": 0.77734375,
      "learning_rate": 1.7517219954364725e-05,
      "loss": 0.3161,
      "step": 6096
    },
    {
      "epoch": 0.46143512605837,
      "grad_norm": 0.75390625,
      "learning_rate": 1.751643440743987e-05,
      "loss": 0.3241,
      "step": 6097
    },
    {
      "epoch": 0.4615108083818173,
      "grad_norm": 0.8515625,
      "learning_rate": 1.7515648753881495e-05,
      "loss": 0.3293,
      "step": 6098
    },
    {
      "epoch": 0.46158649070526464,
      "grad_norm": 0.828125,
      "learning_rate": 1.7514862993700744e-05,
      "loss": 0.3605,
      "step": 6099
    },
    {
      "epoch": 0.461662173028712,
      "grad_norm": 0.83984375,
      "learning_rate": 1.7514077126908767e-05,
      "loss": 0.3645,
      "step": 6100
    },
    {
      "epoch": 0.46173785535215933,
      "grad_norm": 0.83203125,
      "learning_rate": 1.7513291153516708e-05,
      "loss": 0.3329,
      "step": 6101
    },
    {
      "epoch": 0.4618135376756066,
      "grad_norm": 0.74609375,
      "learning_rate": 1.7512505073535723e-05,
      "loss": 0.3128,
      "step": 6102
    },
    {
      "epoch": 0.46188921999905397,
      "grad_norm": 0.796875,
      "learning_rate": 1.7511718886976964e-05,
      "loss": 0.3639,
      "step": 6103
    },
    {
      "epoch": 0.4619649023225013,
      "grad_norm": 0.71484375,
      "learning_rate": 1.751093259385158e-05,
      "loss": 0.2652,
      "step": 6104
    },
    {
      "epoch": 0.4620405846459486,
      "grad_norm": 0.69140625,
      "learning_rate": 1.751014619417073e-05,
      "loss": 0.2872,
      "step": 6105
    },
    {
      "epoch": 0.46211626696939595,
      "grad_norm": 1.171875,
      "learning_rate": 1.7509359687945564e-05,
      "loss": 0.3547,
      "step": 6106
    },
    {
      "epoch": 0.4621919492928433,
      "grad_norm": 0.8671875,
      "learning_rate": 1.750857307518725e-05,
      "loss": 0.3762,
      "step": 6107
    },
    {
      "epoch": 0.46226763161629064,
      "grad_norm": 0.78125,
      "learning_rate": 1.7507786355906934e-05,
      "loss": 0.3237,
      "step": 6108
    },
    {
      "epoch": 0.46234331393973793,
      "grad_norm": 0.7734375,
      "learning_rate": 1.750699953011579e-05,
      "loss": 0.3015,
      "step": 6109
    },
    {
      "epoch": 0.4624189962631853,
      "grad_norm": 0.85546875,
      "learning_rate": 1.7506212597824976e-05,
      "loss": 0.3582,
      "step": 6110
    },
    {
      "epoch": 0.4624946785866326,
      "grad_norm": 0.7109375,
      "learning_rate": 1.7505425559045654e-05,
      "loss": 0.2898,
      "step": 6111
    },
    {
      "epoch": 0.4625703609100799,
      "grad_norm": 0.734375,
      "learning_rate": 1.7504638413788987e-05,
      "loss": 0.287,
      "step": 6112
    },
    {
      "epoch": 0.46264604323352726,
      "grad_norm": 0.8671875,
      "learning_rate": 1.750385116206615e-05,
      "loss": 0.3893,
      "step": 6113
    },
    {
      "epoch": 0.4627217255569746,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7503063803888308e-05,
      "loss": 0.3413,
      "step": 6114
    },
    {
      "epoch": 0.46279740788042195,
      "grad_norm": 0.76171875,
      "learning_rate": 1.7502276339266627e-05,
      "loss": 0.2996,
      "step": 6115
    },
    {
      "epoch": 0.46287309020386924,
      "grad_norm": 0.85546875,
      "learning_rate": 1.7501488768212285e-05,
      "loss": 0.3409,
      "step": 6116
    },
    {
      "epoch": 0.4629487725273166,
      "grad_norm": 0.83203125,
      "learning_rate": 1.750070109073645e-05,
      "loss": 0.3541,
      "step": 6117
    },
    {
      "epoch": 0.46302445485076393,
      "grad_norm": 0.80859375,
      "learning_rate": 1.74999133068503e-05,
      "loss": 0.3148,
      "step": 6118
    },
    {
      "epoch": 0.4631001371742112,
      "grad_norm": 0.76171875,
      "learning_rate": 1.7499125416565005e-05,
      "loss": 0.2768,
      "step": 6119
    },
    {
      "epoch": 0.46317581949765857,
      "grad_norm": 0.84375,
      "learning_rate": 1.749833741989175e-05,
      "loss": 0.3455,
      "step": 6120
    },
    {
      "epoch": 0.4632515018211059,
      "grad_norm": 0.7421875,
      "learning_rate": 1.749754931684171e-05,
      "loss": 0.2916,
      "step": 6121
    },
    {
      "epoch": 0.46332718414455326,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7496761107426066e-05,
      "loss": 0.323,
      "step": 6122
    },
    {
      "epoch": 0.46340286646800055,
      "grad_norm": 0.8046875,
      "learning_rate": 1.7495972791656007e-05,
      "loss": 0.3372,
      "step": 6123
    },
    {
      "epoch": 0.4634785487914479,
      "grad_norm": 1.1640625,
      "learning_rate": 1.7495184369542705e-05,
      "loss": 0.3597,
      "step": 6124
    },
    {
      "epoch": 0.46355423111489524,
      "grad_norm": 0.79296875,
      "learning_rate": 1.749439584109735e-05,
      "loss": 0.3108,
      "step": 6125
    },
    {
      "epoch": 0.46362991343834253,
      "grad_norm": 0.796875,
      "learning_rate": 1.749360720633113e-05,
      "loss": 0.34,
      "step": 6126
    },
    {
      "epoch": 0.4637055957617899,
      "grad_norm": 0.83203125,
      "learning_rate": 1.7492818465255234e-05,
      "loss": 0.3621,
      "step": 6127
    },
    {
      "epoch": 0.4637812780852372,
      "grad_norm": 0.984375,
      "learning_rate": 1.749202961788085e-05,
      "loss": 0.3079,
      "step": 6128
    },
    {
      "epoch": 0.46385696040868457,
      "grad_norm": 0.8203125,
      "learning_rate": 1.7491240664219167e-05,
      "loss": 0.3457,
      "step": 6129
    },
    {
      "epoch": 0.46393264273213186,
      "grad_norm": 0.73046875,
      "learning_rate": 1.7490451604281383e-05,
      "loss": 0.3004,
      "step": 6130
    },
    {
      "epoch": 0.4640083250555792,
      "grad_norm": 0.77734375,
      "learning_rate": 1.7489662438078683e-05,
      "loss": 0.3142,
      "step": 6131
    },
    {
      "epoch": 0.46408400737902655,
      "grad_norm": 0.84765625,
      "learning_rate": 1.7488873165622274e-05,
      "loss": 0.3449,
      "step": 6132
    },
    {
      "epoch": 0.46415968970247384,
      "grad_norm": 0.80859375,
      "learning_rate": 1.7488083786923345e-05,
      "loss": 0.3358,
      "step": 6133
    },
    {
      "epoch": 0.4642353720259212,
      "grad_norm": 0.78125,
      "learning_rate": 1.7487294301993102e-05,
      "loss": 0.3325,
      "step": 6134
    },
    {
      "epoch": 0.46431105434936853,
      "grad_norm": 0.84765625,
      "learning_rate": 1.7486504710842735e-05,
      "loss": 0.3643,
      "step": 6135
    },
    {
      "epoch": 0.4643867366728159,
      "grad_norm": 0.78125,
      "learning_rate": 1.7485715013483452e-05,
      "loss": 0.3233,
      "step": 6136
    },
    {
      "epoch": 0.46446241899626317,
      "grad_norm": 0.7421875,
      "learning_rate": 1.748492520992646e-05,
      "loss": 0.3201,
      "step": 6137
    },
    {
      "epoch": 0.4645381013197105,
      "grad_norm": 0.80078125,
      "learning_rate": 1.7484135300182957e-05,
      "loss": 0.3382,
      "step": 6138
    },
    {
      "epoch": 0.46461378364315786,
      "grad_norm": 0.80078125,
      "learning_rate": 1.748334528426415e-05,
      "loss": 0.3533,
      "step": 6139
    },
    {
      "epoch": 0.46468946596660515,
      "grad_norm": 0.78125,
      "learning_rate": 1.748255516218125e-05,
      "loss": 0.3167,
      "step": 6140
    },
    {
      "epoch": 0.4647651482900525,
      "grad_norm": 0.69140625,
      "learning_rate": 1.7481764933945464e-05,
      "loss": 0.2485,
      "step": 6141
    },
    {
      "epoch": 0.46484083061349984,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7480974599568002e-05,
      "loss": 0.3135,
      "step": 6142
    },
    {
      "epoch": 0.4649165129369472,
      "grad_norm": 0.734375,
      "learning_rate": 1.748018415906008e-05,
      "loss": 0.2722,
      "step": 6143
    },
    {
      "epoch": 0.4649921952603945,
      "grad_norm": 0.765625,
      "learning_rate": 1.747939361243291e-05,
      "loss": 0.3062,
      "step": 6144
    },
    {
      "epoch": 0.4650678775838418,
      "grad_norm": 0.8203125,
      "learning_rate": 1.7478602959697702e-05,
      "loss": 0.3635,
      "step": 6145
    },
    {
      "epoch": 0.4651435599072892,
      "grad_norm": 1.203125,
      "learning_rate": 1.747781220086568e-05,
      "loss": 0.3475,
      "step": 6146
    },
    {
      "epoch": 0.46521924223073646,
      "grad_norm": 0.7578125,
      "learning_rate": 1.747702133594806e-05,
      "loss": 0.3144,
      "step": 6147
    },
    {
      "epoch": 0.4652949245541838,
      "grad_norm": 0.7890625,
      "learning_rate": 1.747623036495606e-05,
      "loss": 0.3454,
      "step": 6148
    },
    {
      "epoch": 0.46537060687763115,
      "grad_norm": 0.76171875,
      "learning_rate": 1.7475439287900907e-05,
      "loss": 0.306,
      "step": 6149
    },
    {
      "epoch": 0.4654462892010785,
      "grad_norm": 0.796875,
      "learning_rate": 1.7474648104793813e-05,
      "loss": 0.3427,
      "step": 6150
    },
    {
      "epoch": 0.4655219715245258,
      "grad_norm": 0.82421875,
      "learning_rate": 1.7473856815646015e-05,
      "loss": 0.3298,
      "step": 6151
    },
    {
      "epoch": 0.46559765384797314,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7473065420468725e-05,
      "loss": 0.3591,
      "step": 6152
    },
    {
      "epoch": 0.4656733361714205,
      "grad_norm": 0.7265625,
      "learning_rate": 1.7472273919273184e-05,
      "loss": 0.2925,
      "step": 6153
    },
    {
      "epoch": 0.4657490184948678,
      "grad_norm": 0.75390625,
      "learning_rate": 1.7471482312070612e-05,
      "loss": 0.3324,
      "step": 6154
    },
    {
      "epoch": 0.4658247008183151,
      "grad_norm": 0.9765625,
      "learning_rate": 1.747069059887224e-05,
      "loss": 0.3743,
      "step": 6155
    },
    {
      "epoch": 0.46590038314176246,
      "grad_norm": 0.83203125,
      "learning_rate": 1.7469898779689306e-05,
      "loss": 0.3397,
      "step": 6156
    },
    {
      "epoch": 0.46597606546520975,
      "grad_norm": 1.28125,
      "learning_rate": 1.746910685453304e-05,
      "loss": 0.393,
      "step": 6157
    },
    {
      "epoch": 0.4660517477886571,
      "grad_norm": 0.7421875,
      "learning_rate": 1.746831482341467e-05,
      "loss": 0.2998,
      "step": 6158
    },
    {
      "epoch": 0.46612743011210445,
      "grad_norm": 0.8125,
      "learning_rate": 1.746752268634544e-05,
      "loss": 0.3673,
      "step": 6159
    },
    {
      "epoch": 0.4662031124355518,
      "grad_norm": 0.703125,
      "learning_rate": 1.7466730443336586e-05,
      "loss": 0.278,
      "step": 6160
    },
    {
      "epoch": 0.4662787947589991,
      "grad_norm": 0.7734375,
      "learning_rate": 1.746593809439935e-05,
      "loss": 0.3222,
      "step": 6161
    },
    {
      "epoch": 0.46635447708244643,
      "grad_norm": 0.8125,
      "learning_rate": 1.746514563954497e-05,
      "loss": 0.3855,
      "step": 6162
    },
    {
      "epoch": 0.4664301594058938,
      "grad_norm": 0.76171875,
      "learning_rate": 1.7464353078784686e-05,
      "loss": 0.3082,
      "step": 6163
    },
    {
      "epoch": 0.46650584172934106,
      "grad_norm": 0.85546875,
      "learning_rate": 1.7463560412129746e-05,
      "loss": 0.3542,
      "step": 6164
    },
    {
      "epoch": 0.4665815240527884,
      "grad_norm": 0.734375,
      "learning_rate": 1.746276763959139e-05,
      "loss": 0.3126,
      "step": 6165
    },
    {
      "epoch": 0.46665720637623576,
      "grad_norm": 1.0546875,
      "learning_rate": 1.7461974761180875e-05,
      "loss": 0.3646,
      "step": 6166
    },
    {
      "epoch": 0.4667328886996831,
      "grad_norm": 0.8203125,
      "learning_rate": 1.7461181776909437e-05,
      "loss": 0.3608,
      "step": 6167
    },
    {
      "epoch": 0.4668085710231304,
      "grad_norm": 0.8046875,
      "learning_rate": 1.7460388686788334e-05,
      "loss": 0.3303,
      "step": 6168
    },
    {
      "epoch": 0.46688425334657774,
      "grad_norm": 0.78125,
      "learning_rate": 1.7459595490828814e-05,
      "loss": 0.3232,
      "step": 6169
    },
    {
      "epoch": 0.4669599356700251,
      "grad_norm": 0.75,
      "learning_rate": 1.7458802189042133e-05,
      "loss": 0.2964,
      "step": 6170
    },
    {
      "epoch": 0.4670356179934724,
      "grad_norm": 0.734375,
      "learning_rate": 1.7458008781439542e-05,
      "loss": 0.2548,
      "step": 6171
    },
    {
      "epoch": 0.4671113003169197,
      "grad_norm": 0.74609375,
      "learning_rate": 1.7457215268032298e-05,
      "loss": 0.3133,
      "step": 6172
    },
    {
      "epoch": 0.46718698264036707,
      "grad_norm": 0.80859375,
      "learning_rate": 1.7456421648831658e-05,
      "loss": 0.3076,
      "step": 6173
    },
    {
      "epoch": 0.4672626649638144,
      "grad_norm": 0.859375,
      "learning_rate": 1.745562792384888e-05,
      "loss": 0.3947,
      "step": 6174
    },
    {
      "epoch": 0.4673383472872617,
      "grad_norm": 0.8125,
      "learning_rate": 1.745483409309523e-05,
      "loss": 0.346,
      "step": 6175
    },
    {
      "epoch": 0.46741402961070905,
      "grad_norm": 0.84765625,
      "learning_rate": 1.745404015658196e-05,
      "loss": 0.3602,
      "step": 6176
    },
    {
      "epoch": 0.4674897119341564,
      "grad_norm": 0.75,
      "learning_rate": 1.745324611432034e-05,
      "loss": 0.2902,
      "step": 6177
    },
    {
      "epoch": 0.4675653942576037,
      "grad_norm": 0.78125,
      "learning_rate": 1.7452451966321636e-05,
      "loss": 0.2547,
      "step": 6178
    },
    {
      "epoch": 0.46764107658105103,
      "grad_norm": 0.8359375,
      "learning_rate": 1.7451657712597116e-05,
      "loss": 0.3545,
      "step": 6179
    },
    {
      "epoch": 0.4677167589044984,
      "grad_norm": 0.77734375,
      "learning_rate": 1.7450863353158036e-05,
      "loss": 0.2996,
      "step": 6180
    },
    {
      "epoch": 0.4677924412279457,
      "grad_norm": 1.203125,
      "learning_rate": 1.7450068888015678e-05,
      "loss": 0.3504,
      "step": 6181
    },
    {
      "epoch": 0.467868123551393,
      "grad_norm": 0.84765625,
      "learning_rate": 1.7449274317181304e-05,
      "loss": 0.3429,
      "step": 6182
    },
    {
      "epoch": 0.46794380587484036,
      "grad_norm": 0.8203125,
      "learning_rate": 1.7448479640666194e-05,
      "loss": 0.3534,
      "step": 6183
    },
    {
      "epoch": 0.4680194881982877,
      "grad_norm": 0.7578125,
      "learning_rate": 1.7447684858481616e-05,
      "loss": 0.2951,
      "step": 6184
    },
    {
      "epoch": 0.468095170521735,
      "grad_norm": 0.7421875,
      "learning_rate": 1.744688997063885e-05,
      "loss": 0.3105,
      "step": 6185
    },
    {
      "epoch": 0.46817085284518234,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7446094977149167e-05,
      "loss": 0.3199,
      "step": 6186
    },
    {
      "epoch": 0.4682465351686297,
      "grad_norm": 0.703125,
      "learning_rate": 1.744529987802385e-05,
      "loss": 0.2705,
      "step": 6187
    },
    {
      "epoch": 0.46832221749207703,
      "grad_norm": 0.8203125,
      "learning_rate": 1.744450467327418e-05,
      "loss": 0.3616,
      "step": 6188
    },
    {
      "epoch": 0.4683978998155243,
      "grad_norm": 0.8125,
      "learning_rate": 1.7443709362911434e-05,
      "loss": 0.3333,
      "step": 6189
    },
    {
      "epoch": 0.46847358213897167,
      "grad_norm": 0.83984375,
      "learning_rate": 1.74429139469469e-05,
      "loss": 0.3638,
      "step": 6190
    },
    {
      "epoch": 0.468549264462419,
      "grad_norm": 0.75,
      "learning_rate": 1.7442118425391855e-05,
      "loss": 0.3182,
      "step": 6191
    },
    {
      "epoch": 0.4686249467858663,
      "grad_norm": 0.78125,
      "learning_rate": 1.744132279825759e-05,
      "loss": 0.3371,
      "step": 6192
    },
    {
      "epoch": 0.46870062910931365,
      "grad_norm": 1.0234375,
      "learning_rate": 1.7440527065555394e-05,
      "loss": 0.3646,
      "step": 6193
    },
    {
      "epoch": 0.468776311432761,
      "grad_norm": 0.6875,
      "learning_rate": 1.743973122729655e-05,
      "loss": 0.2756,
      "step": 6194
    },
    {
      "epoch": 0.46885199375620834,
      "grad_norm": 0.75390625,
      "learning_rate": 1.743893528349235e-05,
      "loss": 0.2935,
      "step": 6195
    },
    {
      "epoch": 0.46892767607965563,
      "grad_norm": 0.80859375,
      "learning_rate": 1.743813923415409e-05,
      "loss": 0.3255,
      "step": 6196
    },
    {
      "epoch": 0.469003358403103,
      "grad_norm": 0.7265625,
      "learning_rate": 1.7437343079293062e-05,
      "loss": 0.2829,
      "step": 6197
    },
    {
      "epoch": 0.4690790407265503,
      "grad_norm": 0.7890625,
      "learning_rate": 1.743654681892056e-05,
      "loss": 0.3204,
      "step": 6198
    },
    {
      "epoch": 0.4691547230499976,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7435750453047877e-05,
      "loss": 0.3025,
      "step": 6199
    },
    {
      "epoch": 0.46923040537344496,
      "grad_norm": 0.83203125,
      "learning_rate": 1.7434953981686313e-05,
      "loss": 0.3552,
      "step": 6200
    },
    {
      "epoch": 0.4693060876968923,
      "grad_norm": 0.734375,
      "learning_rate": 1.7434157404847172e-05,
      "loss": 0.2843,
      "step": 6201
    },
    {
      "epoch": 0.46938177002033965,
      "grad_norm": 0.80078125,
      "learning_rate": 1.743336072254175e-05,
      "loss": 0.346,
      "step": 6202
    },
    {
      "epoch": 0.46945745234378694,
      "grad_norm": 0.74609375,
      "learning_rate": 1.7432563934781344e-05,
      "loss": 0.3197,
      "step": 6203
    },
    {
      "epoch": 0.4695331346672343,
      "grad_norm": 0.71875,
      "learning_rate": 1.743176704157727e-05,
      "loss": 0.2838,
      "step": 6204
    },
    {
      "epoch": 0.46960881699068163,
      "grad_norm": 0.8359375,
      "learning_rate": 1.7430970042940827e-05,
      "loss": 0.3344,
      "step": 6205
    },
    {
      "epoch": 0.4696844993141289,
      "grad_norm": 0.74609375,
      "learning_rate": 1.7430172938883318e-05,
      "loss": 0.2867,
      "step": 6206
    },
    {
      "epoch": 0.46976018163757627,
      "grad_norm": 0.77734375,
      "learning_rate": 1.7429375729416057e-05,
      "loss": 0.3131,
      "step": 6207
    },
    {
      "epoch": 0.4698358639610236,
      "grad_norm": 3.953125,
      "learning_rate": 1.7428578414550353e-05,
      "loss": 0.4277,
      "step": 6208
    },
    {
      "epoch": 0.46991154628447096,
      "grad_norm": 0.84375,
      "learning_rate": 1.7427780994297517e-05,
      "loss": 0.356,
      "step": 6209
    },
    {
      "epoch": 0.46998722860791825,
      "grad_norm": 0.703125,
      "learning_rate": 1.742698346866886e-05,
      "loss": 0.2687,
      "step": 6210
    },
    {
      "epoch": 0.4700629109313656,
      "grad_norm": 0.81640625,
      "learning_rate": 1.7426185837675698e-05,
      "loss": 0.3004,
      "step": 6211
    },
    {
      "epoch": 0.47013859325481294,
      "grad_norm": 0.80859375,
      "learning_rate": 1.7425388101329347e-05,
      "loss": 0.3428,
      "step": 6212
    },
    {
      "epoch": 0.47021427557826023,
      "grad_norm": 0.83203125,
      "learning_rate": 1.742459025964112e-05,
      "loss": 0.3507,
      "step": 6213
    },
    {
      "epoch": 0.4702899579017076,
      "grad_norm": 1.0546875,
      "learning_rate": 1.742379231262234e-05,
      "loss": 0.3749,
      "step": 6214
    },
    {
      "epoch": 0.4703656402251549,
      "grad_norm": 0.80859375,
      "learning_rate": 1.742299426028433e-05,
      "loss": 0.3578,
      "step": 6215
    },
    {
      "epoch": 0.47044132254860227,
      "grad_norm": 0.69921875,
      "learning_rate": 1.7422196102638404e-05,
      "loss": 0.2806,
      "step": 6216
    },
    {
      "epoch": 0.47051700487204956,
      "grad_norm": 0.77734375,
      "learning_rate": 1.7421397839695894e-05,
      "loss": 0.322,
      "step": 6217
    },
    {
      "epoch": 0.4705926871954969,
      "grad_norm": 0.8046875,
      "learning_rate": 1.7420599471468114e-05,
      "loss": 0.304,
      "step": 6218
    },
    {
      "epoch": 0.47066836951894425,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7419800997966397e-05,
      "loss": 0.3334,
      "step": 6219
    },
    {
      "epoch": 0.47074405184239154,
      "grad_norm": 0.85546875,
      "learning_rate": 1.7419002419202077e-05,
      "loss": 0.3888,
      "step": 6220
    },
    {
      "epoch": 0.4708197341658389,
      "grad_norm": 0.80859375,
      "learning_rate": 1.741820373518647e-05,
      "loss": 0.3518,
      "step": 6221
    },
    {
      "epoch": 0.47089541648928623,
      "grad_norm": 0.875,
      "learning_rate": 1.7417404945930916e-05,
      "loss": 0.3531,
      "step": 6222
    },
    {
      "epoch": 0.4709710988127335,
      "grad_norm": 0.8125,
      "learning_rate": 1.741660605144674e-05,
      "loss": 0.3534,
      "step": 6223
    },
    {
      "epoch": 0.47104678113618087,
      "grad_norm": 0.875,
      "learning_rate": 1.7415807051745283e-05,
      "loss": 0.3412,
      "step": 6224
    },
    {
      "epoch": 0.4711224634596282,
      "grad_norm": 0.8203125,
      "learning_rate": 1.7415007946837875e-05,
      "loss": 0.368,
      "step": 6225
    },
    {
      "epoch": 0.47119814578307556,
      "grad_norm": 0.72265625,
      "learning_rate": 1.741420873673586e-05,
      "loss": 0.2836,
      "step": 6226
    },
    {
      "epoch": 0.47127382810652285,
      "grad_norm": 0.81640625,
      "learning_rate": 1.7413409421450563e-05,
      "loss": 0.3846,
      "step": 6227
    },
    {
      "epoch": 0.4713495104299702,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7412610000993337e-05,
      "loss": 0.3407,
      "step": 6228
    },
    {
      "epoch": 0.47142519275341754,
      "grad_norm": 0.828125,
      "learning_rate": 1.741181047537552e-05,
      "loss": 0.3213,
      "step": 6229
    },
    {
      "epoch": 0.47150087507686483,
      "grad_norm": 0.8515625,
      "learning_rate": 1.7411010844608448e-05,
      "loss": 0.32,
      "step": 6230
    },
    {
      "epoch": 0.4715765574003122,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7410211108703467e-05,
      "loss": 0.3185,
      "step": 6231
    },
    {
      "epoch": 0.4716522397237595,
      "grad_norm": 0.84765625,
      "learning_rate": 1.740941126767193e-05,
      "loss": 0.3676,
      "step": 6232
    },
    {
      "epoch": 0.47172792204720687,
      "grad_norm": 0.80078125,
      "learning_rate": 1.7408611321525177e-05,
      "loss": 0.3401,
      "step": 6233
    },
    {
      "epoch": 0.47180360437065416,
      "grad_norm": 0.8125,
      "learning_rate": 1.740781127027456e-05,
      "loss": 0.3746,
      "step": 6234
    },
    {
      "epoch": 0.4718792866941015,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7407011113931425e-05,
      "loss": 0.3183,
      "step": 6235
    },
    {
      "epoch": 0.47195496901754885,
      "grad_norm": 0.734375,
      "learning_rate": 1.7406210852507128e-05,
      "loss": 0.2754,
      "step": 6236
    },
    {
      "epoch": 0.47203065134099614,
      "grad_norm": 0.77734375,
      "learning_rate": 1.7405410486013023e-05,
      "loss": 0.3155,
      "step": 6237
    },
    {
      "epoch": 0.4721063336644435,
      "grad_norm": 0.8125,
      "learning_rate": 1.740461001446046e-05,
      "loss": 0.3554,
      "step": 6238
    },
    {
      "epoch": 0.47218201598789084,
      "grad_norm": 0.8125,
      "learning_rate": 1.7403809437860796e-05,
      "loss": 0.3355,
      "step": 6239
    },
    {
      "epoch": 0.4722576983113382,
      "grad_norm": 0.81640625,
      "learning_rate": 1.740300875622539e-05,
      "loss": 0.361,
      "step": 6240
    },
    {
      "epoch": 0.47233338063478547,
      "grad_norm": 0.84375,
      "learning_rate": 1.7402207969565604e-05,
      "loss": 0.3751,
      "step": 6241
    },
    {
      "epoch": 0.4724090629582328,
      "grad_norm": 0.8359375,
      "learning_rate": 1.7401407077892792e-05,
      "loss": 0.3641,
      "step": 6242
    },
    {
      "epoch": 0.47248474528168016,
      "grad_norm": 0.90625,
      "learning_rate": 1.7400606081218317e-05,
      "loss": 0.3562,
      "step": 6243
    },
    {
      "epoch": 0.47256042760512745,
      "grad_norm": 0.8125,
      "learning_rate": 1.7399804979553552e-05,
      "loss": 0.3476,
      "step": 6244
    },
    {
      "epoch": 0.4726361099285748,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7399003772909852e-05,
      "loss": 0.3325,
      "step": 6245
    },
    {
      "epoch": 0.47271179225202214,
      "grad_norm": 0.71875,
      "learning_rate": 1.7398202461298583e-05,
      "loss": 0.2857,
      "step": 6246
    },
    {
      "epoch": 0.4727874745754695,
      "grad_norm": 0.8828125,
      "learning_rate": 1.739740104473112e-05,
      "loss": 0.3491,
      "step": 6247
    },
    {
      "epoch": 0.4728631568989168,
      "grad_norm": 0.8515625,
      "learning_rate": 1.739659952321883e-05,
      "loss": 0.3583,
      "step": 6248
    },
    {
      "epoch": 0.4729388392223641,
      "grad_norm": 1.28125,
      "learning_rate": 1.739579789677308e-05,
      "loss": 0.3719,
      "step": 6249
    },
    {
      "epoch": 0.4730145215458115,
      "grad_norm": 0.796875,
      "learning_rate": 1.7394996165405244e-05,
      "loss": 0.3567,
      "step": 6250
    },
    {
      "epoch": 0.47309020386925876,
      "grad_norm": 0.9296875,
      "learning_rate": 1.73941943291267e-05,
      "loss": 0.3571,
      "step": 6251
    },
    {
      "epoch": 0.4731658861927061,
      "grad_norm": 0.73046875,
      "learning_rate": 1.7393392387948818e-05,
      "loss": 0.3025,
      "step": 6252
    },
    {
      "epoch": 0.47324156851615345,
      "grad_norm": 0.74609375,
      "learning_rate": 1.7392590341882978e-05,
      "loss": 0.2937,
      "step": 6253
    },
    {
      "epoch": 0.4733172508396008,
      "grad_norm": 0.8203125,
      "learning_rate": 1.739178819094056e-05,
      "loss": 0.3742,
      "step": 6254
    },
    {
      "epoch": 0.4733929331630481,
      "grad_norm": 0.82421875,
      "learning_rate": 1.739098593513294e-05,
      "loss": 0.3742,
      "step": 6255
    },
    {
      "epoch": 0.47346861548649544,
      "grad_norm": 0.84765625,
      "learning_rate": 1.7390183574471502e-05,
      "loss": 0.3508,
      "step": 6256
    },
    {
      "epoch": 0.4735442978099428,
      "grad_norm": 0.7265625,
      "learning_rate": 1.7389381108967625e-05,
      "loss": 0.2857,
      "step": 6257
    },
    {
      "epoch": 0.4736199801333901,
      "grad_norm": 0.73046875,
      "learning_rate": 1.7388578538632705e-05,
      "loss": 0.308,
      "step": 6258
    },
    {
      "epoch": 0.4736956624568374,
      "grad_norm": 0.77734375,
      "learning_rate": 1.738777586347811e-05,
      "loss": 0.3381,
      "step": 6259
    },
    {
      "epoch": 0.47377134478028476,
      "grad_norm": 0.78515625,
      "learning_rate": 1.738697308351524e-05,
      "loss": 0.3419,
      "step": 6260
    },
    {
      "epoch": 0.4738470271037321,
      "grad_norm": 0.734375,
      "learning_rate": 1.7386170198755482e-05,
      "loss": 0.2977,
      "step": 6261
    },
    {
      "epoch": 0.4739227094271794,
      "grad_norm": 0.7734375,
      "learning_rate": 1.738536720921022e-05,
      "loss": 0.334,
      "step": 6262
    },
    {
      "epoch": 0.47399839175062675,
      "grad_norm": 0.78125,
      "learning_rate": 1.7384564114890855e-05,
      "loss": 0.3188,
      "step": 6263
    },
    {
      "epoch": 0.4740740740740741,
      "grad_norm": 0.74609375,
      "learning_rate": 1.7383760915808772e-05,
      "loss": 0.3,
      "step": 6264
    },
    {
      "epoch": 0.4741497563975214,
      "grad_norm": 0.8203125,
      "learning_rate": 1.738295761197537e-05,
      "loss": 0.34,
      "step": 6265
    },
    {
      "epoch": 0.47422543872096873,
      "grad_norm": 0.76953125,
      "learning_rate": 1.738215420340205e-05,
      "loss": 0.2875,
      "step": 6266
    },
    {
      "epoch": 0.4743011210444161,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7381350690100198e-05,
      "loss": 0.318,
      "step": 6267
    },
    {
      "epoch": 0.4743768033678634,
      "grad_norm": 0.8203125,
      "learning_rate": 1.7380547072081222e-05,
      "loss": 0.3317,
      "step": 6268
    },
    {
      "epoch": 0.4744524856913107,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7379743349356523e-05,
      "loss": 0.3766,
      "step": 6269
    },
    {
      "epoch": 0.47452816801475806,
      "grad_norm": 0.80078125,
      "learning_rate": 1.7378939521937495e-05,
      "loss": 0.3139,
      "step": 6270
    },
    {
      "epoch": 0.4746038503382054,
      "grad_norm": 0.796875,
      "learning_rate": 1.737813558983555e-05,
      "loss": 0.3501,
      "step": 6271
    },
    {
      "epoch": 0.4746795326616527,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7377331553062088e-05,
      "loss": 0.3192,
      "step": 6272
    },
    {
      "epoch": 0.47475521498510004,
      "grad_norm": 0.8515625,
      "learning_rate": 1.7376527411628525e-05,
      "loss": 0.3859,
      "step": 6273
    },
    {
      "epoch": 0.4748308973085474,
      "grad_norm": 0.8203125,
      "learning_rate": 1.7375723165546254e-05,
      "loss": 0.331,
      "step": 6274
    },
    {
      "epoch": 0.47490657963199473,
      "grad_norm": 0.75,
      "learning_rate": 1.7374918814826696e-05,
      "loss": 0.2927,
      "step": 6275
    },
    {
      "epoch": 0.474982261955442,
      "grad_norm": 0.71875,
      "learning_rate": 1.737411435948126e-05,
      "loss": 0.2855,
      "step": 6276
    },
    {
      "epoch": 0.47505794427888937,
      "grad_norm": 0.73046875,
      "learning_rate": 1.7373309799521355e-05,
      "loss": 0.2861,
      "step": 6277
    },
    {
      "epoch": 0.4751336266023367,
      "grad_norm": 0.78515625,
      "learning_rate": 1.73725051349584e-05,
      "loss": 0.3142,
      "step": 6278
    },
    {
      "epoch": 0.475209308925784,
      "grad_norm": 0.77734375,
      "learning_rate": 1.7371700365803808e-05,
      "loss": 0.3292,
      "step": 6279
    },
    {
      "epoch": 0.47528499124923135,
      "grad_norm": 0.7265625,
      "learning_rate": 1.7370895492068994e-05,
      "loss": 0.3074,
      "step": 6280
    },
    {
      "epoch": 0.4753606735726787,
      "grad_norm": 0.73046875,
      "learning_rate": 1.737009051376538e-05,
      "loss": 0.281,
      "step": 6281
    },
    {
      "epoch": 0.475436355896126,
      "grad_norm": 0.82421875,
      "learning_rate": 1.7369285430904382e-05,
      "loss": 0.3611,
      "step": 6282
    },
    {
      "epoch": 0.47551203821957333,
      "grad_norm": 0.76171875,
      "learning_rate": 1.7368480243497425e-05,
      "loss": 0.3395,
      "step": 6283
    },
    {
      "epoch": 0.4755877205430207,
      "grad_norm": 0.85546875,
      "learning_rate": 1.7367674951555934e-05,
      "loss": 0.3464,
      "step": 6284
    },
    {
      "epoch": 0.475663402866468,
      "grad_norm": 0.86328125,
      "learning_rate": 1.736686955509133e-05,
      "loss": 0.3591,
      "step": 6285
    },
    {
      "epoch": 0.4757390851899153,
      "grad_norm": 0.78125,
      "learning_rate": 1.736606405411504e-05,
      "loss": 0.3131,
      "step": 6286
    },
    {
      "epoch": 0.47581476751336266,
      "grad_norm": 0.85546875,
      "learning_rate": 1.7365258448638485e-05,
      "loss": 0.3532,
      "step": 6287
    },
    {
      "epoch": 0.47589044983681,
      "grad_norm": 1.1328125,
      "learning_rate": 1.7364452738673105e-05,
      "loss": 0.3581,
      "step": 6288
    },
    {
      "epoch": 0.4759661321602573,
      "grad_norm": 0.77734375,
      "learning_rate": 1.736364692423032e-05,
      "loss": 0.3374,
      "step": 6289
    },
    {
      "epoch": 0.47604181448370464,
      "grad_norm": 0.80078125,
      "learning_rate": 1.736284100532157e-05,
      "loss": 0.3402,
      "step": 6290
    },
    {
      "epoch": 0.476117496807152,
      "grad_norm": 0.7265625,
      "learning_rate": 1.736203498195829e-05,
      "loss": 0.2877,
      "step": 6291
    },
    {
      "epoch": 0.47619317913059933,
      "grad_norm": 0.734375,
      "learning_rate": 1.7361228854151902e-05,
      "loss": 0.2877,
      "step": 6292
    },
    {
      "epoch": 0.4762688614540466,
      "grad_norm": 0.8203125,
      "learning_rate": 1.7360422621913854e-05,
      "loss": 0.368,
      "step": 6293
    },
    {
      "epoch": 0.47634454377749397,
      "grad_norm": 0.82421875,
      "learning_rate": 1.735961628525558e-05,
      "loss": 0.3705,
      "step": 6294
    },
    {
      "epoch": 0.4764202261009413,
      "grad_norm": 0.76953125,
      "learning_rate": 1.735880984418852e-05,
      "loss": 0.3086,
      "step": 6295
    },
    {
      "epoch": 0.4764959084243886,
      "grad_norm": 0.75,
      "learning_rate": 1.7358003298724116e-05,
      "loss": 0.304,
      "step": 6296
    },
    {
      "epoch": 0.47657159074783595,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7357196648873804e-05,
      "loss": 0.3238,
      "step": 6297
    },
    {
      "epoch": 0.4766472730712833,
      "grad_norm": 0.80859375,
      "learning_rate": 1.735638989464904e-05,
      "loss": 0.3586,
      "step": 6298
    },
    {
      "epoch": 0.47672295539473064,
      "grad_norm": 0.76171875,
      "learning_rate": 1.735558303606125e-05,
      "loss": 0.3439,
      "step": 6299
    },
    {
      "epoch": 0.47679863771817793,
      "grad_norm": 0.78125,
      "learning_rate": 1.73547760731219e-05,
      "loss": 0.325,
      "step": 6300
    },
    {
      "epoch": 0.4768743200416253,
      "grad_norm": 0.73828125,
      "learning_rate": 1.7353969005842425e-05,
      "loss": 0.3079,
      "step": 6301
    },
    {
      "epoch": 0.4769500023650726,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7353161834234288e-05,
      "loss": 0.3272,
      "step": 6302
    },
    {
      "epoch": 0.4770256846885199,
      "grad_norm": 0.7265625,
      "learning_rate": 1.7352354558308924e-05,
      "loss": 0.2839,
      "step": 6303
    },
    {
      "epoch": 0.47710136701196726,
      "grad_norm": 0.8125,
      "learning_rate": 1.7351547178077798e-05,
      "loss": 0.3564,
      "step": 6304
    },
    {
      "epoch": 0.4771770493354146,
      "grad_norm": 0.765625,
      "learning_rate": 1.7350739693552356e-05,
      "loss": 0.3028,
      "step": 6305
    },
    {
      "epoch": 0.47725273165886195,
      "grad_norm": 0.81640625,
      "learning_rate": 1.734993210474406e-05,
      "loss": 0.3254,
      "step": 6306
    },
    {
      "epoch": 0.47732841398230924,
      "grad_norm": 0.796875,
      "learning_rate": 1.734912441166436e-05,
      "loss": 0.3297,
      "step": 6307
    },
    {
      "epoch": 0.4774040963057566,
      "grad_norm": 0.8046875,
      "learning_rate": 1.7348316614324724e-05,
      "loss": 0.3379,
      "step": 6308
    },
    {
      "epoch": 0.47747977862920393,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7347508712736603e-05,
      "loss": 0.3378,
      "step": 6309
    },
    {
      "epoch": 0.4775554609526512,
      "grad_norm": 0.81640625,
      "learning_rate": 1.7346700706911466e-05,
      "loss": 0.3449,
      "step": 6310
    },
    {
      "epoch": 0.47763114327609857,
      "grad_norm": 0.703125,
      "learning_rate": 1.734589259686077e-05,
      "loss": 0.2494,
      "step": 6311
    },
    {
      "epoch": 0.4777068255995459,
      "grad_norm": 0.80078125,
      "learning_rate": 1.734508438259598e-05,
      "loss": 0.3303,
      "step": 6312
    },
    {
      "epoch": 0.47778250792299326,
      "grad_norm": 1.015625,
      "learning_rate": 1.7344276064128567e-05,
      "loss": 0.3643,
      "step": 6313
    },
    {
      "epoch": 0.47785819024644055,
      "grad_norm": 0.84765625,
      "learning_rate": 1.734346764146999e-05,
      "loss": 0.3282,
      "step": 6314
    },
    {
      "epoch": 0.4779338725698879,
      "grad_norm": 0.75,
      "learning_rate": 1.7342659114631727e-05,
      "loss": 0.3102,
      "step": 6315
    },
    {
      "epoch": 0.47800955489333524,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7341850483625242e-05,
      "loss": 0.3277,
      "step": 6316
    },
    {
      "epoch": 0.47808523721678253,
      "grad_norm": 0.81640625,
      "learning_rate": 1.734104174846201e-05,
      "loss": 0.3569,
      "step": 6317
    },
    {
      "epoch": 0.4781609195402299,
      "grad_norm": 0.88671875,
      "learning_rate": 1.73402329091535e-05,
      "loss": 0.3122,
      "step": 6318
    },
    {
      "epoch": 0.4782366018636772,
      "grad_norm": 0.84375,
      "learning_rate": 1.7339423965711193e-05,
      "loss": 0.3601,
      "step": 6319
    },
    {
      "epoch": 0.47831228418712457,
      "grad_norm": 0.859375,
      "learning_rate": 1.7338614918146563e-05,
      "loss": 0.3696,
      "step": 6320
    },
    {
      "epoch": 0.47838796651057186,
      "grad_norm": 0.83984375,
      "learning_rate": 1.7337805766471084e-05,
      "loss": 0.3645,
      "step": 6321
    },
    {
      "epoch": 0.4784636488340192,
      "grad_norm": 0.81640625,
      "learning_rate": 1.7336996510696238e-05,
      "loss": 0.3518,
      "step": 6322
    },
    {
      "epoch": 0.47853933115746655,
      "grad_norm": 0.78125,
      "learning_rate": 1.7336187150833508e-05,
      "loss": 0.3292,
      "step": 6323
    },
    {
      "epoch": 0.47861501348091384,
      "grad_norm": 0.8046875,
      "learning_rate": 1.7335377686894373e-05,
      "loss": 0.3331,
      "step": 6324
    },
    {
      "epoch": 0.4786906958043612,
      "grad_norm": 0.89453125,
      "learning_rate": 1.7334568118890318e-05,
      "loss": 0.3443,
      "step": 6325
    },
    {
      "epoch": 0.47876637812780853,
      "grad_norm": 0.75390625,
      "learning_rate": 1.7333758446832826e-05,
      "loss": 0.3142,
      "step": 6326
    },
    {
      "epoch": 0.4788420604512559,
      "grad_norm": 0.70703125,
      "learning_rate": 1.7332948670733386e-05,
      "loss": 0.2825,
      "step": 6327
    },
    {
      "epoch": 0.47891774277470317,
      "grad_norm": 0.73046875,
      "learning_rate": 1.7332138790603487e-05,
      "loss": 0.2824,
      "step": 6328
    },
    {
      "epoch": 0.4789934250981505,
      "grad_norm": 0.78125,
      "learning_rate": 1.7331328806454615e-05,
      "loss": 0.2842,
      "step": 6329
    },
    {
      "epoch": 0.47906910742159786,
      "grad_norm": 0.77734375,
      "learning_rate": 1.7330518718298263e-05,
      "loss": 0.3489,
      "step": 6330
    },
    {
      "epoch": 0.47914478974504515,
      "grad_norm": 0.76171875,
      "learning_rate": 1.7329708526145926e-05,
      "loss": 0.3182,
      "step": 6331
    },
    {
      "epoch": 0.4792204720684925,
      "grad_norm": 0.6796875,
      "learning_rate": 1.7328898230009094e-05,
      "loss": 0.2346,
      "step": 6332
    },
    {
      "epoch": 0.47929615439193984,
      "grad_norm": 0.87109375,
      "learning_rate": 1.7328087829899262e-05,
      "loss": 0.3257,
      "step": 6333
    },
    {
      "epoch": 0.4793718367153872,
      "grad_norm": 0.7578125,
      "learning_rate": 1.732727732582793e-05,
      "loss": 0.277,
      "step": 6334
    },
    {
      "epoch": 0.4794475190388345,
      "grad_norm": 0.8203125,
      "learning_rate": 1.7326466717806597e-05,
      "loss": 0.3603,
      "step": 6335
    },
    {
      "epoch": 0.4795232013622818,
      "grad_norm": 0.75390625,
      "learning_rate": 1.7325656005846755e-05,
      "loss": 0.3087,
      "step": 6336
    },
    {
      "epoch": 0.47959888368572917,
      "grad_norm": 0.7109375,
      "learning_rate": 1.732484518995992e-05,
      "loss": 0.2599,
      "step": 6337
    },
    {
      "epoch": 0.47967456600917646,
      "grad_norm": 0.80859375,
      "learning_rate": 1.732403427015758e-05,
      "loss": 0.3508,
      "step": 6338
    },
    {
      "epoch": 0.4797502483326238,
      "grad_norm": 0.78125,
      "learning_rate": 1.7323223246451246e-05,
      "loss": 0.3269,
      "step": 6339
    },
    {
      "epoch": 0.47982593065607115,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7322412118852422e-05,
      "loss": 0.3357,
      "step": 6340
    },
    {
      "epoch": 0.47990161297951844,
      "grad_norm": 0.7578125,
      "learning_rate": 1.732160088737262e-05,
      "loss": 0.3241,
      "step": 6341
    },
    {
      "epoch": 0.4799772953029658,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7320789552023344e-05,
      "loss": 0.3266,
      "step": 6342
    },
    {
      "epoch": 0.48005297762641314,
      "grad_norm": 0.78515625,
      "learning_rate": 1.73199781128161e-05,
      "loss": 0.324,
      "step": 6343
    },
    {
      "epoch": 0.4801286599498605,
      "grad_norm": 0.69921875,
      "learning_rate": 1.7319166569762416e-05,
      "loss": 0.2712,
      "step": 6344
    },
    {
      "epoch": 0.48020434227330777,
      "grad_norm": 0.8046875,
      "learning_rate": 1.7318354922873785e-05,
      "loss": 0.3333,
      "step": 6345
    },
    {
      "epoch": 0.4802800245967551,
      "grad_norm": 0.86328125,
      "learning_rate": 1.7317543172161736e-05,
      "loss": 0.3139,
      "step": 6346
    },
    {
      "epoch": 0.48035570692020246,
      "grad_norm": 0.78125,
      "learning_rate": 1.7316731317637777e-05,
      "loss": 0.327,
      "step": 6347
    },
    {
      "epoch": 0.48043138924364975,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7315919359313426e-05,
      "loss": 0.3198,
      "step": 6348
    },
    {
      "epoch": 0.4805070715670971,
      "grad_norm": 0.80078125,
      "learning_rate": 1.7315107297200207e-05,
      "loss": 0.3449,
      "step": 6349
    },
    {
      "epoch": 0.48058275389054445,
      "grad_norm": 0.8125,
      "learning_rate": 1.731429513130964e-05,
      "loss": 0.3664,
      "step": 6350
    },
    {
      "epoch": 0.4806584362139918,
      "grad_norm": 0.73828125,
      "learning_rate": 1.731348286165324e-05,
      "loss": 0.2962,
      "step": 6351
    },
    {
      "epoch": 0.4807341185374391,
      "grad_norm": 0.7734375,
      "learning_rate": 1.731267048824254e-05,
      "loss": 0.3142,
      "step": 6352
    },
    {
      "epoch": 0.4808098008608864,
      "grad_norm": 1.25,
      "learning_rate": 1.7311858011089055e-05,
      "loss": 0.3991,
      "step": 6353
    },
    {
      "epoch": 0.4808854831843338,
      "grad_norm": 0.85546875,
      "learning_rate": 1.731104543020432e-05,
      "loss": 0.3885,
      "step": 6354
    },
    {
      "epoch": 0.48096116550778106,
      "grad_norm": 0.75,
      "learning_rate": 1.731023274559986e-05,
      "loss": 0.2913,
      "step": 6355
    },
    {
      "epoch": 0.4810368478312284,
      "grad_norm": 0.79296875,
      "learning_rate": 1.73094199572872e-05,
      "loss": 0.3452,
      "step": 6356
    },
    {
      "epoch": 0.48111253015467575,
      "grad_norm": 0.71484375,
      "learning_rate": 1.730860706527788e-05,
      "loss": 0.2783,
      "step": 6357
    },
    {
      "epoch": 0.4811882124781231,
      "grad_norm": 0.80078125,
      "learning_rate": 1.7307794069583422e-05,
      "loss": 0.3199,
      "step": 6358
    },
    {
      "epoch": 0.4812638948015704,
      "grad_norm": 0.9296875,
      "learning_rate": 1.7306980970215367e-05,
      "loss": 0.2932,
      "step": 6359
    },
    {
      "epoch": 0.48133957712501774,
      "grad_norm": 0.84765625,
      "learning_rate": 1.7306167767185245e-05,
      "loss": 0.3714,
      "step": 6360
    },
    {
      "epoch": 0.4814152594484651,
      "grad_norm": 0.76953125,
      "learning_rate": 1.73053544605046e-05,
      "loss": 0.326,
      "step": 6361
    },
    {
      "epoch": 0.4814909417719124,
      "grad_norm": 0.80078125,
      "learning_rate": 1.7304541050184963e-05,
      "loss": 0.3593,
      "step": 6362
    },
    {
      "epoch": 0.4815666240953597,
      "grad_norm": 0.828125,
      "learning_rate": 1.7303727536237876e-05,
      "loss": 0.3613,
      "step": 6363
    },
    {
      "epoch": 0.48164230641880706,
      "grad_norm": 0.921875,
      "learning_rate": 1.730291391867488e-05,
      "loss": 0.4119,
      "step": 6364
    },
    {
      "epoch": 0.4817179887422544,
      "grad_norm": 0.78515625,
      "learning_rate": 1.730210019750752e-05,
      "loss": 0.3382,
      "step": 6365
    },
    {
      "epoch": 0.4817936710657017,
      "grad_norm": 0.8203125,
      "learning_rate": 1.7301286372747336e-05,
      "loss": 0.3296,
      "step": 6366
    },
    {
      "epoch": 0.48186935338914905,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7300472444405873e-05,
      "loss": 0.3789,
      "step": 6367
    },
    {
      "epoch": 0.4819450357125964,
      "grad_norm": 0.75,
      "learning_rate": 1.7299658412494686e-05,
      "loss": 0.2982,
      "step": 6368
    },
    {
      "epoch": 0.4820207180360437,
      "grad_norm": 0.80078125,
      "learning_rate": 1.7298844277025315e-05,
      "loss": 0.3277,
      "step": 6369
    },
    {
      "epoch": 0.48209640035949103,
      "grad_norm": 0.90625,
      "learning_rate": 1.7298030038009314e-05,
      "loss": 0.4042,
      "step": 6370
    },
    {
      "epoch": 0.4821720826829384,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7297215695458226e-05,
      "loss": 0.3459,
      "step": 6371
    },
    {
      "epoch": 0.4822477650063857,
      "grad_norm": 0.79296875,
      "learning_rate": 1.729640124938362e-05,
      "loss": 0.3256,
      "step": 6372
    },
    {
      "epoch": 0.482323447329833,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7295586699797034e-05,
      "loss": 0.3005,
      "step": 6373
    },
    {
      "epoch": 0.48239912965328036,
      "grad_norm": 0.68359375,
      "learning_rate": 1.7294772046710034e-05,
      "loss": 0.2658,
      "step": 6374
    },
    {
      "epoch": 0.4824748119767277,
      "grad_norm": 0.78125,
      "learning_rate": 1.7293957290134178e-05,
      "loss": 0.3621,
      "step": 6375
    },
    {
      "epoch": 0.482550494300175,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7293142430081017e-05,
      "loss": 0.293,
      "step": 6376
    },
    {
      "epoch": 0.48262617662362234,
      "grad_norm": 0.765625,
      "learning_rate": 1.7292327466562116e-05,
      "loss": 0.3463,
      "step": 6377
    },
    {
      "epoch": 0.4827018589470697,
      "grad_norm": 0.7421875,
      "learning_rate": 1.7291512399589038e-05,
      "loss": 0.3094,
      "step": 6378
    },
    {
      "epoch": 0.48277754127051703,
      "grad_norm": 0.796875,
      "learning_rate": 1.729069722917334e-05,
      "loss": 0.3471,
      "step": 6379
    },
    {
      "epoch": 0.4828532235939643,
      "grad_norm": 1.140625,
      "learning_rate": 1.7289881955326594e-05,
      "loss": 0.3847,
      "step": 6380
    },
    {
      "epoch": 0.48292890591741167,
      "grad_norm": 0.75390625,
      "learning_rate": 1.728906657806036e-05,
      "loss": 0.3255,
      "step": 6381
    },
    {
      "epoch": 0.483004588240859,
      "grad_norm": 0.796875,
      "learning_rate": 1.7288251097386208e-05,
      "loss": 0.3437,
      "step": 6382
    },
    {
      "epoch": 0.4830802705643063,
      "grad_norm": 0.875,
      "learning_rate": 1.728743551331571e-05,
      "loss": 0.3766,
      "step": 6383
    },
    {
      "epoch": 0.48315595288775365,
      "grad_norm": 0.80078125,
      "learning_rate": 1.728661982586043e-05,
      "loss": 0.3477,
      "step": 6384
    },
    {
      "epoch": 0.483231635211201,
      "grad_norm": 0.82421875,
      "learning_rate": 1.7285804035031945e-05,
      "loss": 0.3543,
      "step": 6385
    },
    {
      "epoch": 0.48330731753464834,
      "grad_norm": 1.109375,
      "learning_rate": 1.7284988140841825e-05,
      "loss": 0.3357,
      "step": 6386
    },
    {
      "epoch": 0.48338299985809563,
      "grad_norm": 0.71484375,
      "learning_rate": 1.728417214330165e-05,
      "loss": 0.2885,
      "step": 6387
    },
    {
      "epoch": 0.483458682181543,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7283356042422992e-05,
      "loss": 0.346,
      "step": 6388
    },
    {
      "epoch": 0.4835343645049903,
      "grad_norm": 0.81640625,
      "learning_rate": 1.728253983821743e-05,
      "loss": 0.3823,
      "step": 6389
    },
    {
      "epoch": 0.4836100468284376,
      "grad_norm": 0.875,
      "learning_rate": 1.7281723530696544e-05,
      "loss": 0.3518,
      "step": 6390
    },
    {
      "epoch": 0.48368572915188496,
      "grad_norm": 0.78125,
      "learning_rate": 1.728090711987191e-05,
      "loss": 0.3197,
      "step": 6391
    },
    {
      "epoch": 0.4837614114753323,
      "grad_norm": 0.72265625,
      "learning_rate": 1.728009060575512e-05,
      "loss": 0.3181,
      "step": 6392
    },
    {
      "epoch": 0.48383709379877965,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7279273988357747e-05,
      "loss": 0.3809,
      "step": 6393
    },
    {
      "epoch": 0.48391277612222694,
      "grad_norm": 0.9296875,
      "learning_rate": 1.7278457267691383e-05,
      "loss": 0.3006,
      "step": 6394
    },
    {
      "epoch": 0.4839884584456743,
      "grad_norm": 0.79296875,
      "learning_rate": 1.727764044376761e-05,
      "loss": 0.3212,
      "step": 6395
    },
    {
      "epoch": 0.48406414076912163,
      "grad_norm": 0.6875,
      "learning_rate": 1.7276823516598022e-05,
      "loss": 0.2864,
      "step": 6396
    },
    {
      "epoch": 0.4841398230925689,
      "grad_norm": 0.8203125,
      "learning_rate": 1.7276006486194203e-05,
      "loss": 0.3616,
      "step": 6397
    },
    {
      "epoch": 0.48421550541601627,
      "grad_norm": 0.8515625,
      "learning_rate": 1.7275189352567744e-05,
      "loss": 0.3566,
      "step": 6398
    },
    {
      "epoch": 0.4842911877394636,
      "grad_norm": 0.765625,
      "learning_rate": 1.7274372115730242e-05,
      "loss": 0.3185,
      "step": 6399
    },
    {
      "epoch": 0.4843668700629109,
      "grad_norm": 0.78515625,
      "learning_rate": 1.727355477569329e-05,
      "loss": 0.3423,
      "step": 6400
    },
    {
      "epoch": 0.48444255238635825,
      "grad_norm": 0.72265625,
      "learning_rate": 1.7272737332468478e-05,
      "loss": 0.2943,
      "step": 6401
    },
    {
      "epoch": 0.4845182347098056,
      "grad_norm": 0.84375,
      "learning_rate": 1.7271919786067407e-05,
      "loss": 0.3404,
      "step": 6402
    },
    {
      "epoch": 0.48459391703325294,
      "grad_norm": 0.7109375,
      "learning_rate": 1.7271102136501677e-05,
      "loss": 0.2731,
      "step": 6403
    },
    {
      "epoch": 0.48466959935670023,
      "grad_norm": 0.7421875,
      "learning_rate": 1.7270284383782882e-05,
      "loss": 0.2855,
      "step": 6404
    },
    {
      "epoch": 0.4847452816801476,
      "grad_norm": 0.7734375,
      "learning_rate": 1.726946652792263e-05,
      "loss": 0.3458,
      "step": 6405
    },
    {
      "epoch": 0.4848209640035949,
      "grad_norm": 0.76953125,
      "learning_rate": 1.726864856893252e-05,
      "loss": 0.3205,
      "step": 6406
    },
    {
      "epoch": 0.4848966463270422,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7267830506824156e-05,
      "loss": 0.2986,
      "step": 6407
    },
    {
      "epoch": 0.48497232865048956,
      "grad_norm": 0.7421875,
      "learning_rate": 1.7267012341609145e-05,
      "loss": 0.3188,
      "step": 6408
    },
    {
      "epoch": 0.4850480109739369,
      "grad_norm": 0.82421875,
      "learning_rate": 1.726619407329909e-05,
      "loss": 0.3732,
      "step": 6409
    },
    {
      "epoch": 0.48512369329738425,
      "grad_norm": 0.73828125,
      "learning_rate": 1.7265375701905606e-05,
      "loss": 0.2896,
      "step": 6410
    },
    {
      "epoch": 0.48519937562083154,
      "grad_norm": 0.7890625,
      "learning_rate": 1.72645572274403e-05,
      "loss": 0.3452,
      "step": 6411
    },
    {
      "epoch": 0.4852750579442789,
      "grad_norm": 0.78125,
      "learning_rate": 1.7263738649914783e-05,
      "loss": 0.3348,
      "step": 6412
    },
    {
      "epoch": 0.48535074026772623,
      "grad_norm": 0.75,
      "learning_rate": 1.726291996934067e-05,
      "loss": 0.3047,
      "step": 6413
    },
    {
      "epoch": 0.4854264225911735,
      "grad_norm": 0.83203125,
      "learning_rate": 1.726210118572957e-05,
      "loss": 0.3604,
      "step": 6414
    },
    {
      "epoch": 0.48550210491462087,
      "grad_norm": 0.80078125,
      "learning_rate": 1.72612822990931e-05,
      "loss": 0.3444,
      "step": 6415
    },
    {
      "epoch": 0.4855777872380682,
      "grad_norm": 0.8046875,
      "learning_rate": 1.7260463309442888e-05,
      "loss": 0.3402,
      "step": 6416
    },
    {
      "epoch": 0.48565346956151556,
      "grad_norm": 0.76171875,
      "learning_rate": 1.7259644216790538e-05,
      "loss": 0.3267,
      "step": 6417
    },
    {
      "epoch": 0.48572915188496285,
      "grad_norm": 0.828125,
      "learning_rate": 1.7258825021147682e-05,
      "loss": 0.3426,
      "step": 6418
    },
    {
      "epoch": 0.4858048342084102,
      "grad_norm": 0.71875,
      "learning_rate": 1.725800572252593e-05,
      "loss": 0.3027,
      "step": 6419
    },
    {
      "epoch": 0.48588051653185754,
      "grad_norm": 0.8203125,
      "learning_rate": 1.7257186320936918e-05,
      "loss": 0.3508,
      "step": 6420
    },
    {
      "epoch": 0.48595619885530483,
      "grad_norm": 0.8359375,
      "learning_rate": 1.725636681639226e-05,
      "loss": 0.3637,
      "step": 6421
    },
    {
      "epoch": 0.4860318811787522,
      "grad_norm": 0.71484375,
      "learning_rate": 1.725554720890359e-05,
      "loss": 0.2762,
      "step": 6422
    },
    {
      "epoch": 0.4861075635021995,
      "grad_norm": 0.82421875,
      "learning_rate": 1.7254727498482528e-05,
      "loss": 0.336,
      "step": 6423
    },
    {
      "epoch": 0.48618324582564687,
      "grad_norm": 0.765625,
      "learning_rate": 1.725390768514071e-05,
      "loss": 0.3244,
      "step": 6424
    },
    {
      "epoch": 0.48625892814909416,
      "grad_norm": 0.74609375,
      "learning_rate": 1.725308776888976e-05,
      "loss": 0.2706,
      "step": 6425
    },
    {
      "epoch": 0.4863346104725415,
      "grad_norm": 0.8203125,
      "learning_rate": 1.7252267749741317e-05,
      "loss": 0.3167,
      "step": 6426
    },
    {
      "epoch": 0.48641029279598885,
      "grad_norm": 0.765625,
      "learning_rate": 1.7251447627707007e-05,
      "loss": 0.3487,
      "step": 6427
    },
    {
      "epoch": 0.48648597511943614,
      "grad_norm": 0.8046875,
      "learning_rate": 1.7250627402798472e-05,
      "loss": 0.3343,
      "step": 6428
    },
    {
      "epoch": 0.4865616574428835,
      "grad_norm": 0.7578125,
      "learning_rate": 1.7249807075027344e-05,
      "loss": 0.3056,
      "step": 6429
    },
    {
      "epoch": 0.48663733976633083,
      "grad_norm": 0.73046875,
      "learning_rate": 1.724898664440526e-05,
      "loss": 0.3047,
      "step": 6430
    },
    {
      "epoch": 0.4867130220897782,
      "grad_norm": 1.265625,
      "learning_rate": 1.7248166110943863e-05,
      "loss": 0.3427,
      "step": 6431
    },
    {
      "epoch": 0.48678870441322547,
      "grad_norm": 0.84765625,
      "learning_rate": 1.724734547465479e-05,
      "loss": 0.3589,
      "step": 6432
    },
    {
      "epoch": 0.4868643867366728,
      "grad_norm": 0.76171875,
      "learning_rate": 1.7246524735549687e-05,
      "loss": 0.3119,
      "step": 6433
    },
    {
      "epoch": 0.48694006906012016,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7245703893640188e-05,
      "loss": 0.3163,
      "step": 6434
    },
    {
      "epoch": 0.48701575138356745,
      "grad_norm": 0.83203125,
      "learning_rate": 1.7244882948937952e-05,
      "loss": 0.3931,
      "step": 6435
    },
    {
      "epoch": 0.4870914337070148,
      "grad_norm": 0.859375,
      "learning_rate": 1.7244061901454617e-05,
      "loss": 0.3473,
      "step": 6436
    },
    {
      "epoch": 0.48716711603046214,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7243240751201832e-05,
      "loss": 0.3296,
      "step": 6437
    },
    {
      "epoch": 0.4872427983539095,
      "grad_norm": 0.76953125,
      "learning_rate": 1.724241949819125e-05,
      "loss": 0.3204,
      "step": 6438
    },
    {
      "epoch": 0.4873184806773568,
      "grad_norm": 0.765625,
      "learning_rate": 1.724159814243451e-05,
      "loss": 0.3439,
      "step": 6439
    },
    {
      "epoch": 0.4873941630008041,
      "grad_norm": 0.8359375,
      "learning_rate": 1.7240776683943283e-05,
      "loss": 0.3431,
      "step": 6440
    },
    {
      "epoch": 0.48746984532425147,
      "grad_norm": 0.75390625,
      "learning_rate": 1.723995512272921e-05,
      "loss": 0.2976,
      "step": 6441
    },
    {
      "epoch": 0.48754552764769876,
      "grad_norm": 0.765625,
      "learning_rate": 1.723913345880395e-05,
      "loss": 0.2937,
      "step": 6442
    },
    {
      "epoch": 0.4876212099711461,
      "grad_norm": 0.80078125,
      "learning_rate": 1.723831169217916e-05,
      "loss": 0.3336,
      "step": 6443
    },
    {
      "epoch": 0.48769689229459345,
      "grad_norm": 0.89453125,
      "learning_rate": 1.7237489822866496e-05,
      "loss": 0.3601,
      "step": 6444
    },
    {
      "epoch": 0.4877725746180408,
      "grad_norm": 0.859375,
      "learning_rate": 1.7236667850877616e-05,
      "loss": 0.3721,
      "step": 6445
    },
    {
      "epoch": 0.4878482569414881,
      "grad_norm": 0.73828125,
      "learning_rate": 1.7235845776224186e-05,
      "loss": 0.2905,
      "step": 6446
    },
    {
      "epoch": 0.48792393926493544,
      "grad_norm": 0.7734375,
      "learning_rate": 1.723502359891787e-05,
      "loss": 0.3179,
      "step": 6447
    },
    {
      "epoch": 0.4879996215883828,
      "grad_norm": 0.82421875,
      "learning_rate": 1.7234201318970325e-05,
      "loss": 0.3628,
      "step": 6448
    },
    {
      "epoch": 0.48807530391183007,
      "grad_norm": 0.83203125,
      "learning_rate": 1.723337893639322e-05,
      "loss": 0.3359,
      "step": 6449
    },
    {
      "epoch": 0.4881509862352774,
      "grad_norm": 0.8203125,
      "learning_rate": 1.7232556451198223e-05,
      "loss": 0.3446,
      "step": 6450
    },
    {
      "epoch": 0.48822666855872476,
      "grad_norm": 1.015625,
      "learning_rate": 1.7231733863397e-05,
      "loss": 0.2921,
      "step": 6451
    },
    {
      "epoch": 0.4883023508821721,
      "grad_norm": 0.77734375,
      "learning_rate": 1.7230911173001223e-05,
      "loss": 0.3165,
      "step": 6452
    },
    {
      "epoch": 0.4883780332056194,
      "grad_norm": 0.81640625,
      "learning_rate": 1.7230088380022562e-05,
      "loss": 0.3615,
      "step": 6453
    },
    {
      "epoch": 0.48845371552906675,
      "grad_norm": 0.8359375,
      "learning_rate": 1.7229265484472694e-05,
      "loss": 0.3515,
      "step": 6454
    },
    {
      "epoch": 0.4885293978525141,
      "grad_norm": 0.8046875,
      "learning_rate": 1.7228442486363283e-05,
      "loss": 0.3143,
      "step": 6455
    },
    {
      "epoch": 0.4886050801759614,
      "grad_norm": 0.7578125,
      "learning_rate": 1.722761938570601e-05,
      "loss": 0.305,
      "step": 6456
    },
    {
      "epoch": 0.4886807624994087,
      "grad_norm": 0.8203125,
      "learning_rate": 1.722679618251256e-05,
      "loss": 0.3477,
      "step": 6457
    },
    {
      "epoch": 0.4887564448228561,
      "grad_norm": 0.8125,
      "learning_rate": 1.72259728767946e-05,
      "loss": 0.3423,
      "step": 6458
    },
    {
      "epoch": 0.4888321271463034,
      "grad_norm": 0.75390625,
      "learning_rate": 1.7225149468563816e-05,
      "loss": 0.3392,
      "step": 6459
    },
    {
      "epoch": 0.4889078094697507,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7224325957831886e-05,
      "loss": 0.3416,
      "step": 6460
    },
    {
      "epoch": 0.48898349179319806,
      "grad_norm": 0.75390625,
      "learning_rate": 1.72235023446105e-05,
      "loss": 0.3058,
      "step": 6461
    },
    {
      "epoch": 0.4890591741166454,
      "grad_norm": 0.78125,
      "learning_rate": 1.7222678628911335e-05,
      "loss": 0.3377,
      "step": 6462
    },
    {
      "epoch": 0.4891348564400927,
      "grad_norm": 0.76171875,
      "learning_rate": 1.7221854810746076e-05,
      "loss": 0.3166,
      "step": 6463
    },
    {
      "epoch": 0.48921053876354004,
      "grad_norm": 0.70703125,
      "learning_rate": 1.7221030890126415e-05,
      "loss": 0.2713,
      "step": 6464
    },
    {
      "epoch": 0.4892862210869874,
      "grad_norm": 0.7890625,
      "learning_rate": 1.722020686706404e-05,
      "loss": 0.3479,
      "step": 6465
    },
    {
      "epoch": 0.4893619034104347,
      "grad_norm": 0.82421875,
      "learning_rate": 1.7219382741570637e-05,
      "loss": 0.3343,
      "step": 6466
    },
    {
      "epoch": 0.489437585733882,
      "grad_norm": 0.78125,
      "learning_rate": 1.7218558513657903e-05,
      "loss": 0.323,
      "step": 6467
    },
    {
      "epoch": 0.48951326805732936,
      "grad_norm": 0.7578125,
      "learning_rate": 1.721773418333753e-05,
      "loss": 0.3034,
      "step": 6468
    },
    {
      "epoch": 0.4895889503807767,
      "grad_norm": 0.7421875,
      "learning_rate": 1.721690975062121e-05,
      "loss": 0.2874,
      "step": 6469
    },
    {
      "epoch": 0.489664632704224,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7216085215520644e-05,
      "loss": 0.3277,
      "step": 6470
    },
    {
      "epoch": 0.48974031502767135,
      "grad_norm": 0.8515625,
      "learning_rate": 1.7215260578047523e-05,
      "loss": 0.3659,
      "step": 6471
    },
    {
      "epoch": 0.4898159973511187,
      "grad_norm": 0.78515625,
      "learning_rate": 1.721443583821355e-05,
      "loss": 0.3227,
      "step": 6472
    },
    {
      "epoch": 0.489891679674566,
      "grad_norm": 0.84375,
      "learning_rate": 1.721361099603042e-05,
      "loss": 0.3813,
      "step": 6473
    },
    {
      "epoch": 0.48996736199801333,
      "grad_norm": 0.73046875,
      "learning_rate": 1.7212786051509843e-05,
      "loss": 0.3071,
      "step": 6474
    },
    {
      "epoch": 0.4900430443214607,
      "grad_norm": 0.74609375,
      "learning_rate": 1.7211961004663523e-05,
      "loss": 0.3098,
      "step": 6475
    },
    {
      "epoch": 0.490118726644908,
      "grad_norm": 0.72265625,
      "learning_rate": 1.7211135855503152e-05,
      "loss": 0.3115,
      "step": 6476
    },
    {
      "epoch": 0.4901944089683553,
      "grad_norm": 0.68359375,
      "learning_rate": 1.721031060404045e-05,
      "loss": 0.2642,
      "step": 6477
    },
    {
      "epoch": 0.49027009129180266,
      "grad_norm": 0.75390625,
      "learning_rate": 1.7209485250287117e-05,
      "loss": 0.3231,
      "step": 6478
    },
    {
      "epoch": 0.49034577361525,
      "grad_norm": 0.765625,
      "learning_rate": 1.720865979425486e-05,
      "loss": 0.3471,
      "step": 6479
    },
    {
      "epoch": 0.4904214559386973,
      "grad_norm": 0.75390625,
      "learning_rate": 1.7207834235955398e-05,
      "loss": 0.3217,
      "step": 6480
    },
    {
      "epoch": 0.49049713826214464,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7207008575400433e-05,
      "loss": 0.3342,
      "step": 6481
    },
    {
      "epoch": 0.490572820585592,
      "grad_norm": 0.78515625,
      "learning_rate": 1.720618281260169e-05,
      "loss": 0.3221,
      "step": 6482
    },
    {
      "epoch": 0.49064850290903933,
      "grad_norm": 0.796875,
      "learning_rate": 1.7205356947570875e-05,
      "loss": 0.3251,
      "step": 6483
    },
    {
      "epoch": 0.4907241852324866,
      "grad_norm": 0.77734375,
      "learning_rate": 1.7204530980319707e-05,
      "loss": 0.3345,
      "step": 6484
    },
    {
      "epoch": 0.49079986755593397,
      "grad_norm": 0.71875,
      "learning_rate": 1.7203704910859902e-05,
      "loss": 0.2906,
      "step": 6485
    },
    {
      "epoch": 0.4908755498793813,
      "grad_norm": 0.76953125,
      "learning_rate": 1.720287873920318e-05,
      "loss": 0.2978,
      "step": 6486
    },
    {
      "epoch": 0.4909512322028286,
      "grad_norm": 0.83984375,
      "learning_rate": 1.7202052465361268e-05,
      "loss": 0.3756,
      "step": 6487
    },
    {
      "epoch": 0.49102691452627595,
      "grad_norm": 0.859375,
      "learning_rate": 1.7201226089345877e-05,
      "loss": 0.2814,
      "step": 6488
    },
    {
      "epoch": 0.4911025968497233,
      "grad_norm": 0.93359375,
      "learning_rate": 1.7200399611168742e-05,
      "loss": 0.3237,
      "step": 6489
    },
    {
      "epoch": 0.49117827917317064,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7199573030841577e-05,
      "loss": 0.3078,
      "step": 6490
    },
    {
      "epoch": 0.49125396149661793,
      "grad_norm": 0.71484375,
      "learning_rate": 1.7198746348376114e-05,
      "loss": 0.2799,
      "step": 6491
    },
    {
      "epoch": 0.4913296438200653,
      "grad_norm": 0.79296875,
      "learning_rate": 1.719791956378408e-05,
      "loss": 0.3376,
      "step": 6492
    },
    {
      "epoch": 0.4914053261435126,
      "grad_norm": 0.703125,
      "learning_rate": 1.719709267707721e-05,
      "loss": 0.2616,
      "step": 6493
    },
    {
      "epoch": 0.4914810084669599,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7196265688267227e-05,
      "loss": 0.3316,
      "step": 6494
    },
    {
      "epoch": 0.49155669079040726,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7195438597365865e-05,
      "loss": 0.3412,
      "step": 6495
    },
    {
      "epoch": 0.4916323731138546,
      "grad_norm": 0.78515625,
      "learning_rate": 1.719461140438486e-05,
      "loss": 0.326,
      "step": 6496
    },
    {
      "epoch": 0.49170805543730195,
      "grad_norm": 0.77734375,
      "learning_rate": 1.7193784109335945e-05,
      "loss": 0.3117,
      "step": 6497
    },
    {
      "epoch": 0.49178373776074924,
      "grad_norm": 0.80859375,
      "learning_rate": 1.7192956712230855e-05,
      "loss": 0.3557,
      "step": 6498
    },
    {
      "epoch": 0.4918594200841966,
      "grad_norm": 0.69140625,
      "learning_rate": 1.7192129213081333e-05,
      "loss": 0.2688,
      "step": 6499
    },
    {
      "epoch": 0.49193510240764393,
      "grad_norm": 0.8359375,
      "learning_rate": 1.7191301611899117e-05,
      "loss": 0.387,
      "step": 6500
    },
    {
      "epoch": 0.4920107847310912,
      "grad_norm": 0.75390625,
      "learning_rate": 1.7190473908695947e-05,
      "loss": 0.2973,
      "step": 6501
    },
    {
      "epoch": 0.49208646705453857,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7189646103483565e-05,
      "loss": 0.3466,
      "step": 6502
    },
    {
      "epoch": 0.4921621493779859,
      "grad_norm": 0.72265625,
      "learning_rate": 1.7188818196273715e-05,
      "loss": 0.2862,
      "step": 6503
    },
    {
      "epoch": 0.49223783170143326,
      "grad_norm": 0.7421875,
      "learning_rate": 1.718799018707814e-05,
      "loss": 0.3155,
      "step": 6504
    },
    {
      "epoch": 0.49231351402488055,
      "grad_norm": 0.75390625,
      "learning_rate": 1.718716207590859e-05,
      "loss": 0.2969,
      "step": 6505
    },
    {
      "epoch": 0.4923891963483279,
      "grad_norm": 0.796875,
      "learning_rate": 1.7186333862776815e-05,
      "loss": 0.3195,
      "step": 6506
    },
    {
      "epoch": 0.49246487867177524,
      "grad_norm": 0.7890625,
      "learning_rate": 1.718550554769456e-05,
      "loss": 0.3206,
      "step": 6507
    },
    {
      "epoch": 0.49254056099522253,
      "grad_norm": 0.703125,
      "learning_rate": 1.718467713067358e-05,
      "loss": 0.2735,
      "step": 6508
    },
    {
      "epoch": 0.4926162433186699,
      "grad_norm": 0.83203125,
      "learning_rate": 1.7183848611725623e-05,
      "loss": 0.3461,
      "step": 6509
    },
    {
      "epoch": 0.4926919256421172,
      "grad_norm": 0.70703125,
      "learning_rate": 1.718301999086245e-05,
      "loss": 0.2683,
      "step": 6510
    },
    {
      "epoch": 0.49276760796556457,
      "grad_norm": 0.73828125,
      "learning_rate": 1.7182191268095806e-05,
      "loss": 0.2842,
      "step": 6511
    },
    {
      "epoch": 0.49284329028901186,
      "grad_norm": 1.921875,
      "learning_rate": 1.7181362443437456e-05,
      "loss": 0.4517,
      "step": 6512
    },
    {
      "epoch": 0.4929189726124592,
      "grad_norm": 0.80078125,
      "learning_rate": 1.718053351689916e-05,
      "loss": 0.334,
      "step": 6513
    },
    {
      "epoch": 0.49299465493590655,
      "grad_norm": 0.8046875,
      "learning_rate": 1.7179704488492664e-05,
      "loss": 0.3454,
      "step": 6514
    },
    {
      "epoch": 0.49307033725935384,
      "grad_norm": 0.83984375,
      "learning_rate": 1.7178875358229746e-05,
      "loss": 0.3605,
      "step": 6515
    },
    {
      "epoch": 0.4931460195828012,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7178046126122162e-05,
      "loss": 0.3582,
      "step": 6516
    },
    {
      "epoch": 0.49322170190624853,
      "grad_norm": 0.77734375,
      "learning_rate": 1.7177216792181678e-05,
      "loss": 0.3555,
      "step": 6517
    },
    {
      "epoch": 0.4932973842296959,
      "grad_norm": 0.765625,
      "learning_rate": 1.717638735642005e-05,
      "loss": 0.3348,
      "step": 6518
    },
    {
      "epoch": 0.49337306655314317,
      "grad_norm": 0.78515625,
      "learning_rate": 1.717555781884906e-05,
      "loss": 0.3425,
      "step": 6519
    },
    {
      "epoch": 0.4934487488765905,
      "grad_norm": 0.74609375,
      "learning_rate": 1.717472817948046e-05,
      "loss": 0.3072,
      "step": 6520
    },
    {
      "epoch": 0.49352443120003786,
      "grad_norm": 3.46875,
      "learning_rate": 1.7173898438326038e-05,
      "loss": 0.3997,
      "step": 6521
    },
    {
      "epoch": 0.49360011352348515,
      "grad_norm": 0.78125,
      "learning_rate": 1.717306859539755e-05,
      "loss": 0.3322,
      "step": 6522
    },
    {
      "epoch": 0.4936757958469325,
      "grad_norm": 0.73828125,
      "learning_rate": 1.7172238650706775e-05,
      "loss": 0.293,
      "step": 6523
    },
    {
      "epoch": 0.49375147817037984,
      "grad_norm": 0.734375,
      "learning_rate": 1.717140860426549e-05,
      "loss": 0.3107,
      "step": 6524
    },
    {
      "epoch": 0.49382716049382713,
      "grad_norm": 0.796875,
      "learning_rate": 1.7170578456085464e-05,
      "loss": 0.3667,
      "step": 6525
    },
    {
      "epoch": 0.4939028428172745,
      "grad_norm": 0.78125,
      "learning_rate": 1.716974820617848e-05,
      "loss": 0.3233,
      "step": 6526
    },
    {
      "epoch": 0.4939785251407218,
      "grad_norm": 0.73046875,
      "learning_rate": 1.716891785455631e-05,
      "loss": 0.2776,
      "step": 6527
    },
    {
      "epoch": 0.49405420746416917,
      "grad_norm": 0.74609375,
      "learning_rate": 1.716808740123074e-05,
      "loss": 0.2775,
      "step": 6528
    },
    {
      "epoch": 0.49412988978761646,
      "grad_norm": 0.765625,
      "learning_rate": 1.7167256846213552e-05,
      "loss": 0.3203,
      "step": 6529
    },
    {
      "epoch": 0.4942055721110638,
      "grad_norm": 0.80078125,
      "learning_rate": 1.7166426189516524e-05,
      "loss": 0.3394,
      "step": 6530
    },
    {
      "epoch": 0.49428125443451115,
      "grad_norm": 0.74609375,
      "learning_rate": 1.7165595431151442e-05,
      "loss": 0.3034,
      "step": 6531
    },
    {
      "epoch": 0.49435693675795844,
      "grad_norm": 0.7421875,
      "learning_rate": 1.716476457113009e-05,
      "loss": 0.2921,
      "step": 6532
    },
    {
      "epoch": 0.4944326190814058,
      "grad_norm": 0.73828125,
      "learning_rate": 1.716393360946426e-05,
      "loss": 0.2776,
      "step": 6533
    },
    {
      "epoch": 0.49450830140485313,
      "grad_norm": 0.84375,
      "learning_rate": 1.716310254616574e-05,
      "loss": 0.3532,
      "step": 6534
    },
    {
      "epoch": 0.4945839837283005,
      "grad_norm": 0.8125,
      "learning_rate": 1.7162271381246315e-05,
      "loss": 0.3397,
      "step": 6535
    },
    {
      "epoch": 0.49465966605174777,
      "grad_norm": 0.80078125,
      "learning_rate": 1.7161440114717783e-05,
      "loss": 0.3091,
      "step": 6536
    },
    {
      "epoch": 0.4947353483751951,
      "grad_norm": 0.78515625,
      "learning_rate": 1.716060874659193e-05,
      "loss": 0.3088,
      "step": 6537
    },
    {
      "epoch": 0.49481103069864246,
      "grad_norm": 0.78125,
      "learning_rate": 1.7159777276880557e-05,
      "loss": 0.305,
      "step": 6538
    },
    {
      "epoch": 0.49488671302208975,
      "grad_norm": 0.82421875,
      "learning_rate": 1.7158945705595453e-05,
      "loss": 0.3918,
      "step": 6539
    },
    {
      "epoch": 0.4949623953455371,
      "grad_norm": 0.80859375,
      "learning_rate": 1.715811403274842e-05,
      "loss": 0.3301,
      "step": 6540
    },
    {
      "epoch": 0.49503807766898444,
      "grad_norm": 0.7734375,
      "learning_rate": 1.715728225835126e-05,
      "loss": 0.2849,
      "step": 6541
    },
    {
      "epoch": 0.4951137599924318,
      "grad_norm": 0.83203125,
      "learning_rate": 1.7156450382415763e-05,
      "loss": 0.3706,
      "step": 6542
    },
    {
      "epoch": 0.4951894423158791,
      "grad_norm": 0.8125,
      "learning_rate": 1.715561840495374e-05,
      "loss": 0.3279,
      "step": 6543
    },
    {
      "epoch": 0.4952651246393264,
      "grad_norm": 0.69140625,
      "learning_rate": 1.715478632597699e-05,
      "loss": 0.2711,
      "step": 6544
    },
    {
      "epoch": 0.49534080696277377,
      "grad_norm": 0.7578125,
      "learning_rate": 1.7153954145497315e-05,
      "loss": 0.3066,
      "step": 6545
    },
    {
      "epoch": 0.49541648928622106,
      "grad_norm": 1.46875,
      "learning_rate": 1.7153121863526526e-05,
      "loss": 0.4334,
      "step": 6546
    },
    {
      "epoch": 0.4954921716096684,
      "grad_norm": 0.7734375,
      "learning_rate": 1.715228948007643e-05,
      "loss": 0.3232,
      "step": 6547
    },
    {
      "epoch": 0.49556785393311575,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7151456995158833e-05,
      "loss": 0.352,
      "step": 6548
    },
    {
      "epoch": 0.4956435362565631,
      "grad_norm": 0.85546875,
      "learning_rate": 1.7150624408785544e-05,
      "loss": 0.3285,
      "step": 6549
    },
    {
      "epoch": 0.4957192185800104,
      "grad_norm": 0.859375,
      "learning_rate": 1.7149791720968378e-05,
      "loss": 0.3696,
      "step": 6550
    },
    {
      "epoch": 0.49579490090345774,
      "grad_norm": 0.828125,
      "learning_rate": 1.7148958931719145e-05,
      "loss": 0.3538,
      "step": 6551
    },
    {
      "epoch": 0.4958705832269051,
      "grad_norm": 0.734375,
      "learning_rate": 1.7148126041049667e-05,
      "loss": 0.2878,
      "step": 6552
    },
    {
      "epoch": 0.49594626555035237,
      "grad_norm": 0.80859375,
      "learning_rate": 1.714729304897175e-05,
      "loss": 0.3551,
      "step": 6553
    },
    {
      "epoch": 0.4960219478737997,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7146459955497217e-05,
      "loss": 0.2963,
      "step": 6554
    },
    {
      "epoch": 0.49609763019724706,
      "grad_norm": 0.84375,
      "learning_rate": 1.7145626760637886e-05,
      "loss": 0.3272,
      "step": 6555
    },
    {
      "epoch": 0.4961733125206944,
      "grad_norm": 0.7734375,
      "learning_rate": 1.714479346440558e-05,
      "loss": 0.3324,
      "step": 6556
    },
    {
      "epoch": 0.4962489948441417,
      "grad_norm": 0.8359375,
      "learning_rate": 1.7143960066812117e-05,
      "loss": 0.3386,
      "step": 6557
    },
    {
      "epoch": 0.49632467716758905,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7143126567869316e-05,
      "loss": 0.3268,
      "step": 6558
    },
    {
      "epoch": 0.4964003594910364,
      "grad_norm": 0.73046875,
      "learning_rate": 1.714229296758901e-05,
      "loss": 0.2837,
      "step": 6559
    },
    {
      "epoch": 0.4964760418144837,
      "grad_norm": 0.8203125,
      "learning_rate": 1.714145926598302e-05,
      "loss": 0.34,
      "step": 6560
    },
    {
      "epoch": 0.496551724137931,
      "grad_norm": 0.74609375,
      "learning_rate": 1.7140625463063176e-05,
      "loss": 0.3007,
      "step": 6561
    },
    {
      "epoch": 0.4966274064613784,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7139791558841307e-05,
      "loss": 0.3277,
      "step": 6562
    },
    {
      "epoch": 0.4967030887848257,
      "grad_norm": 0.7265625,
      "learning_rate": 1.7138957553329242e-05,
      "loss": 0.2716,
      "step": 6563
    },
    {
      "epoch": 0.496778771108273,
      "grad_norm": 0.8359375,
      "learning_rate": 1.7138123446538813e-05,
      "loss": 0.3462,
      "step": 6564
    },
    {
      "epoch": 0.49685445343172036,
      "grad_norm": 0.7265625,
      "learning_rate": 1.7137289238481855e-05,
      "loss": 0.3081,
      "step": 6565
    },
    {
      "epoch": 0.4969301357551677,
      "grad_norm": 0.8125,
      "learning_rate": 1.7136454929170198e-05,
      "loss": 0.331,
      "step": 6566
    },
    {
      "epoch": 0.497005818078615,
      "grad_norm": 0.859375,
      "learning_rate": 1.7135620518615682e-05,
      "loss": 0.2513,
      "step": 6567
    },
    {
      "epoch": 0.49708150040206234,
      "grad_norm": 0.88671875,
      "learning_rate": 1.7134786006830147e-05,
      "loss": 0.3582,
      "step": 6568
    },
    {
      "epoch": 0.4971571827255097,
      "grad_norm": 0.7421875,
      "learning_rate": 1.7133951393825426e-05,
      "loss": 0.2875,
      "step": 6569
    },
    {
      "epoch": 0.49723286504895703,
      "grad_norm": 0.83984375,
      "learning_rate": 1.7133116679613362e-05,
      "loss": 0.3597,
      "step": 6570
    },
    {
      "epoch": 0.4973085473724043,
      "grad_norm": 0.8359375,
      "learning_rate": 1.7132281864205798e-05,
      "loss": 0.3702,
      "step": 6571
    },
    {
      "epoch": 0.49738422969585167,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7131446947614575e-05,
      "loss": 0.3283,
      "step": 6572
    },
    {
      "epoch": 0.497459912019299,
      "grad_norm": 0.76171875,
      "learning_rate": 1.713061192985154e-05,
      "loss": 0.3111,
      "step": 6573
    },
    {
      "epoch": 0.4975355943427463,
      "grad_norm": 0.71484375,
      "learning_rate": 1.712977681092854e-05,
      "loss": 0.2902,
      "step": 6574
    },
    {
      "epoch": 0.49761127666619365,
      "grad_norm": 0.7265625,
      "learning_rate": 1.712894159085742e-05,
      "loss": 0.2925,
      "step": 6575
    },
    {
      "epoch": 0.497686958989641,
      "grad_norm": 0.796875,
      "learning_rate": 1.7128106269650028e-05,
      "loss": 0.341,
      "step": 6576
    },
    {
      "epoch": 0.49776264131308834,
      "grad_norm": 0.83203125,
      "learning_rate": 1.7127270847318217e-05,
      "loss": 0.3502,
      "step": 6577
    },
    {
      "epoch": 0.49783832363653563,
      "grad_norm": 0.72265625,
      "learning_rate": 1.712643532387384e-05,
      "loss": 0.2936,
      "step": 6578
    },
    {
      "epoch": 0.497914005959983,
      "grad_norm": 0.8046875,
      "learning_rate": 1.7125599699328746e-05,
      "loss": 0.3691,
      "step": 6579
    },
    {
      "epoch": 0.4979896882834303,
      "grad_norm": 0.87109375,
      "learning_rate": 1.7124763973694793e-05,
      "loss": 0.3992,
      "step": 6580
    },
    {
      "epoch": 0.4980653706068776,
      "grad_norm": 0.72265625,
      "learning_rate": 1.712392814698384e-05,
      "loss": 0.2982,
      "step": 6581
    },
    {
      "epoch": 0.49814105293032496,
      "grad_norm": 0.76171875,
      "learning_rate": 1.7123092219207736e-05,
      "loss": 0.3284,
      "step": 6582
    },
    {
      "epoch": 0.4982167352537723,
      "grad_norm": 0.73046875,
      "learning_rate": 1.7122256190378346e-05,
      "loss": 0.2936,
      "step": 6583
    },
    {
      "epoch": 0.4982924175772196,
      "grad_norm": 0.7109375,
      "learning_rate": 1.712142006050753e-05,
      "loss": 0.2827,
      "step": 6584
    },
    {
      "epoch": 0.49836809990066694,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7120583829607152e-05,
      "loss": 0.3437,
      "step": 6585
    },
    {
      "epoch": 0.4984437822241143,
      "grad_norm": 0.83984375,
      "learning_rate": 1.7119747497689072e-05,
      "loss": 0.3579,
      "step": 6586
    },
    {
      "epoch": 0.49851946454756163,
      "grad_norm": 1.1796875,
      "learning_rate": 1.711891106476516e-05,
      "loss": 0.3969,
      "step": 6587
    },
    {
      "epoch": 0.4985951468710089,
      "grad_norm": 0.73828125,
      "learning_rate": 1.711807453084727e-05,
      "loss": 0.3198,
      "step": 6588
    },
    {
      "epoch": 0.49867082919445627,
      "grad_norm": 0.7578125,
      "learning_rate": 1.711723789594728e-05,
      "loss": 0.3161,
      "step": 6589
    },
    {
      "epoch": 0.4987465115179036,
      "grad_norm": 0.8515625,
      "learning_rate": 1.711640116007706e-05,
      "loss": 0.3911,
      "step": 6590
    },
    {
      "epoch": 0.4988221938413509,
      "grad_norm": 0.8359375,
      "learning_rate": 1.7115564323248475e-05,
      "loss": 0.3756,
      "step": 6591
    },
    {
      "epoch": 0.49889787616479825,
      "grad_norm": 0.82421875,
      "learning_rate": 1.71147273854734e-05,
      "loss": 0.3678,
      "step": 6592
    },
    {
      "epoch": 0.4989735584882456,
      "grad_norm": 0.76953125,
      "learning_rate": 1.711389034676371e-05,
      "loss": 0.304,
      "step": 6593
    },
    {
      "epoch": 0.49904924081169294,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7113053207131274e-05,
      "loss": 0.3014,
      "step": 6594
    },
    {
      "epoch": 0.49912492313514023,
      "grad_norm": 0.78515625,
      "learning_rate": 1.711221596658797e-05,
      "loss": 0.346,
      "step": 6595
    },
    {
      "epoch": 0.4992006054585876,
      "grad_norm": 0.8203125,
      "learning_rate": 1.7111378625145684e-05,
      "loss": 0.3651,
      "step": 6596
    },
    {
      "epoch": 0.4992762877820349,
      "grad_norm": 0.8359375,
      "learning_rate": 1.7110541182816284e-05,
      "loss": 0.3332,
      "step": 6597
    },
    {
      "epoch": 0.4993519701054822,
      "grad_norm": 0.83203125,
      "learning_rate": 1.710970363961166e-05,
      "loss": 0.3792,
      "step": 6598
    },
    {
      "epoch": 0.49942765242892956,
      "grad_norm": 0.921875,
      "learning_rate": 1.7108865995543684e-05,
      "loss": 0.3715,
      "step": 6599
    },
    {
      "epoch": 0.4995033347523769,
      "grad_norm": 0.7421875,
      "learning_rate": 1.7108028250624245e-05,
      "loss": 0.3119,
      "step": 6600
    },
    {
      "epoch": 0.49957901707582425,
      "grad_norm": 0.8046875,
      "learning_rate": 1.7107190404865232e-05,
      "loss": 0.3639,
      "step": 6601
    },
    {
      "epoch": 0.49965469939927154,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7106352458278524e-05,
      "loss": 0.3145,
      "step": 6602
    },
    {
      "epoch": 0.4997303817227189,
      "grad_norm": 0.78125,
      "learning_rate": 1.710551441087601e-05,
      "loss": 0.3276,
      "step": 6603
    },
    {
      "epoch": 0.49980606404616623,
      "grad_norm": 0.7890625,
      "learning_rate": 1.710467626266958e-05,
      "loss": 0.3532,
      "step": 6604
    },
    {
      "epoch": 0.4998817463696135,
      "grad_norm": 0.765625,
      "learning_rate": 1.7103838013671126e-05,
      "loss": 0.3113,
      "step": 6605
    },
    {
      "epoch": 0.49995742869306087,
      "grad_norm": 0.8046875,
      "learning_rate": 1.710299966389254e-05,
      "loss": 0.355,
      "step": 6606
    },
    {
      "epoch": 0.5000331110165082,
      "grad_norm": 1.046875,
      "learning_rate": 1.7102161213345716e-05,
      "loss": 0.3393,
      "step": 6607
    },
    {
      "epoch": 0.5001087933399555,
      "grad_norm": 0.80859375,
      "learning_rate": 1.7101322662042546e-05,
      "loss": 0.3443,
      "step": 6608
    },
    {
      "epoch": 0.5001844756634028,
      "grad_norm": 0.8203125,
      "learning_rate": 1.7100484009994927e-05,
      "loss": 0.3615,
      "step": 6609
    },
    {
      "epoch": 0.5002601579868502,
      "grad_norm": 0.7421875,
      "learning_rate": 1.709964525721476e-05,
      "loss": 0.2891,
      "step": 6610
    },
    {
      "epoch": 0.5002601579868502,
      "eval_loss": 0.3412069082260132,
      "eval_runtime": 83.9577,
      "eval_samples_per_second": 57.91,
      "eval_steps_per_second": 57.91,
      "step": 6610
    },
    {
      "epoch": 0.5003358403102975,
      "grad_norm": 0.828125,
      "learning_rate": 1.709880640371394e-05,
      "loss": 0.3405,
      "step": 6611
    },
    {
      "epoch": 0.5004115226337449,
      "grad_norm": 0.8984375,
      "learning_rate": 1.709796744950437e-05,
      "loss": 0.3942,
      "step": 6612
    },
    {
      "epoch": 0.5004872049571922,
      "grad_norm": 0.7109375,
      "learning_rate": 1.7097128394597948e-05,
      "loss": 0.2746,
      "step": 6613
    },
    {
      "epoch": 0.5005628872806395,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7096289239006583e-05,
      "loss": 0.3085,
      "step": 6614
    },
    {
      "epoch": 0.5006385696040868,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7095449982742177e-05,
      "loss": 0.3306,
      "step": 6615
    },
    {
      "epoch": 0.5007142519275342,
      "grad_norm": 0.76171875,
      "learning_rate": 1.7094610625816637e-05,
      "loss": 0.323,
      "step": 6616
    },
    {
      "epoch": 0.5007899342509815,
      "grad_norm": 0.83203125,
      "learning_rate": 1.709377116824187e-05,
      "loss": 0.371,
      "step": 6617
    },
    {
      "epoch": 0.5008656165744289,
      "grad_norm": 0.71484375,
      "learning_rate": 1.709293161002979e-05,
      "loss": 0.2974,
      "step": 6618
    },
    {
      "epoch": 0.5009412988978762,
      "grad_norm": 0.75,
      "learning_rate": 1.7092091951192298e-05,
      "loss": 0.3231,
      "step": 6619
    },
    {
      "epoch": 0.5010169812213235,
      "grad_norm": 0.75,
      "learning_rate": 1.7091252191741312e-05,
      "loss": 0.2759,
      "step": 6620
    },
    {
      "epoch": 0.5010926635447708,
      "grad_norm": 0.74609375,
      "learning_rate": 1.7090412331688745e-05,
      "loss": 0.3117,
      "step": 6621
    },
    {
      "epoch": 0.5011683458682181,
      "grad_norm": 0.7265625,
      "learning_rate": 1.7089572371046513e-05,
      "loss": 0.3055,
      "step": 6622
    },
    {
      "epoch": 0.5012440281916655,
      "grad_norm": 0.74609375,
      "learning_rate": 1.708873230982653e-05,
      "loss": 0.3032,
      "step": 6623
    },
    {
      "epoch": 0.5013197105151128,
      "grad_norm": 0.81640625,
      "learning_rate": 1.7087892148040713e-05,
      "loss": 0.3519,
      "step": 6624
    },
    {
      "epoch": 0.5013953928385602,
      "grad_norm": 0.69140625,
      "learning_rate": 1.7087051885700984e-05,
      "loss": 0.2684,
      "step": 6625
    },
    {
      "epoch": 0.5014710751620075,
      "grad_norm": 0.7421875,
      "learning_rate": 1.708621152281926e-05,
      "loss": 0.3071,
      "step": 6626
    },
    {
      "epoch": 0.5015467574854549,
      "grad_norm": 0.84375,
      "learning_rate": 1.708537105940747e-05,
      "loss": 0.3731,
      "step": 6627
    },
    {
      "epoch": 0.5016224398089021,
      "grad_norm": 0.79296875,
      "learning_rate": 1.708453049547753e-05,
      "loss": 0.354,
      "step": 6628
    },
    {
      "epoch": 0.5016981221323494,
      "grad_norm": 0.79296875,
      "learning_rate": 1.708368983104137e-05,
      "loss": 0.3624,
      "step": 6629
    },
    {
      "epoch": 0.5017738044557968,
      "grad_norm": 0.9921875,
      "learning_rate": 1.708284906611091e-05,
      "loss": 0.295,
      "step": 6630
    },
    {
      "epoch": 0.5018494867792441,
      "grad_norm": 0.765625,
      "learning_rate": 1.7082008200698082e-05,
      "loss": 0.3231,
      "step": 6631
    },
    {
      "epoch": 0.5019251691026915,
      "grad_norm": 0.84375,
      "learning_rate": 1.7081167234814814e-05,
      "loss": 0.3682,
      "step": 6632
    },
    {
      "epoch": 0.5020008514261388,
      "grad_norm": 0.80859375,
      "learning_rate": 1.7080326168473038e-05,
      "loss": 0.3327,
      "step": 6633
    },
    {
      "epoch": 0.5020765337495862,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7079485001684685e-05,
      "loss": 0.327,
      "step": 6634
    },
    {
      "epoch": 0.5021522160730334,
      "grad_norm": 0.72265625,
      "learning_rate": 1.707864373446169e-05,
      "loss": 0.2963,
      "step": 6635
    },
    {
      "epoch": 0.5022278983964807,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7077802366815986e-05,
      "loss": 0.3355,
      "step": 6636
    },
    {
      "epoch": 0.5023035807199281,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7076960898759508e-05,
      "loss": 0.326,
      "step": 6637
    },
    {
      "epoch": 0.5023792630433754,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7076119330304194e-05,
      "loss": 0.3266,
      "step": 6638
    },
    {
      "epoch": 0.5024549453668228,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7075277661461987e-05,
      "loss": 0.3108,
      "step": 6639
    },
    {
      "epoch": 0.5025306276902701,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7074435892244824e-05,
      "loss": 0.3375,
      "step": 6640
    },
    {
      "epoch": 0.5026063100137175,
      "grad_norm": 0.78125,
      "learning_rate": 1.7073594022664646e-05,
      "loss": 0.3356,
      "step": 6641
    },
    {
      "epoch": 0.5026819923371647,
      "grad_norm": 0.77734375,
      "learning_rate": 1.70727520527334e-05,
      "loss": 0.3529,
      "step": 6642
    },
    {
      "epoch": 0.502757674660612,
      "grad_norm": 0.765625,
      "learning_rate": 1.707190998246303e-05,
      "loss": 0.272,
      "step": 6643
    },
    {
      "epoch": 0.5028333569840594,
      "grad_norm": 0.81640625,
      "learning_rate": 1.7071067811865477e-05,
      "loss": 0.3553,
      "step": 6644
    },
    {
      "epoch": 0.5029090393075067,
      "grad_norm": 0.8046875,
      "learning_rate": 1.7070225540952696e-05,
      "loss": 0.341,
      "step": 6645
    },
    {
      "epoch": 0.5029847216309541,
      "grad_norm": 0.7265625,
      "learning_rate": 1.706938316973663e-05,
      "loss": 0.2772,
      "step": 6646
    },
    {
      "epoch": 0.5030604039544014,
      "grad_norm": 0.82421875,
      "learning_rate": 1.7068540698229235e-05,
      "loss": 0.3206,
      "step": 6647
    },
    {
      "epoch": 0.5031360862778488,
      "grad_norm": 0.77734375,
      "learning_rate": 1.7067698126442456e-05,
      "loss": 0.3203,
      "step": 6648
    },
    {
      "epoch": 0.503211768601296,
      "grad_norm": 0.79296875,
      "learning_rate": 1.7066855454388257e-05,
      "loss": 0.3247,
      "step": 6649
    },
    {
      "epoch": 0.5032874509247434,
      "grad_norm": 0.79296875,
      "learning_rate": 1.706601268207858e-05,
      "loss": 0.3172,
      "step": 6650
    },
    {
      "epoch": 0.5033631332481907,
      "grad_norm": 0.734375,
      "learning_rate": 1.7065169809525392e-05,
      "loss": 0.3106,
      "step": 6651
    },
    {
      "epoch": 0.503438815571638,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7064326836740646e-05,
      "loss": 0.3287,
      "step": 6652
    },
    {
      "epoch": 0.5035144978950854,
      "grad_norm": 0.82421875,
      "learning_rate": 1.70634837637363e-05,
      "loss": 0.3078,
      "step": 6653
    },
    {
      "epoch": 0.5035901802185327,
      "grad_norm": 0.78125,
      "learning_rate": 1.7062640590524314e-05,
      "loss": 0.3509,
      "step": 6654
    },
    {
      "epoch": 0.5036658625419801,
      "grad_norm": 0.74609375,
      "learning_rate": 1.7061797317116652e-05,
      "loss": 0.2799,
      "step": 6655
    },
    {
      "epoch": 0.5037415448654273,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7060953943525277e-05,
      "loss": 0.3064,
      "step": 6656
    },
    {
      "epoch": 0.5038172271888747,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7060110469762152e-05,
      "loss": 0.3389,
      "step": 6657
    },
    {
      "epoch": 0.503892909512322,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7059266895839245e-05,
      "loss": 0.3282,
      "step": 6658
    },
    {
      "epoch": 0.5039685918357694,
      "grad_norm": 0.7421875,
      "learning_rate": 1.7058423221768525e-05,
      "loss": 0.2913,
      "step": 6659
    },
    {
      "epoch": 0.5040442741592167,
      "grad_norm": 0.75390625,
      "learning_rate": 1.7057579447561957e-05,
      "loss": 0.3142,
      "step": 6660
    },
    {
      "epoch": 0.5041199564826641,
      "grad_norm": 0.765625,
      "learning_rate": 1.7056735573231513e-05,
      "loss": 0.3027,
      "step": 6661
    },
    {
      "epoch": 0.5041956388061114,
      "grad_norm": 0.7421875,
      "learning_rate": 1.705589159878917e-05,
      "loss": 0.2945,
      "step": 6662
    },
    {
      "epoch": 0.5042713211295586,
      "grad_norm": 0.73828125,
      "learning_rate": 1.705504752424689e-05,
      "loss": 0.2943,
      "step": 6663
    },
    {
      "epoch": 0.504347003453006,
      "grad_norm": 0.97265625,
      "learning_rate": 1.7054203349616652e-05,
      "loss": 0.3494,
      "step": 6664
    },
    {
      "epoch": 0.5044226857764533,
      "grad_norm": 0.80859375,
      "learning_rate": 1.705335907491044e-05,
      "loss": 0.3204,
      "step": 6665
    },
    {
      "epoch": 0.5044983680999007,
      "grad_norm": 0.76171875,
      "learning_rate": 1.7052514700140227e-05,
      "loss": 0.3238,
      "step": 6666
    },
    {
      "epoch": 0.504574050423348,
      "grad_norm": 0.80078125,
      "learning_rate": 1.7051670225317983e-05,
      "loss": 0.3559,
      "step": 6667
    },
    {
      "epoch": 0.5046497327467954,
      "grad_norm": 0.78125,
      "learning_rate": 1.7050825650455704e-05,
      "loss": 0.3407,
      "step": 6668
    },
    {
      "epoch": 0.5047254150702427,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7049980975565355e-05,
      "loss": 0.3373,
      "step": 6669
    },
    {
      "epoch": 0.50480109739369,
      "grad_norm": 0.75,
      "learning_rate": 1.7049136200658935e-05,
      "loss": 0.2984,
      "step": 6670
    },
    {
      "epoch": 0.5048767797171373,
      "grad_norm": 1.0078125,
      "learning_rate": 1.7048291325748417e-05,
      "loss": 0.3569,
      "step": 6671
    },
    {
      "epoch": 0.5049524620405846,
      "grad_norm": 0.75390625,
      "learning_rate": 1.7047446350845795e-05,
      "loss": 0.3078,
      "step": 6672
    },
    {
      "epoch": 0.505028144364032,
      "grad_norm": 1.1328125,
      "learning_rate": 1.704660127596305e-05,
      "loss": 0.3759,
      "step": 6673
    },
    {
      "epoch": 0.5051038266874793,
      "grad_norm": 0.765625,
      "learning_rate": 1.7045756101112175e-05,
      "loss": 0.2937,
      "step": 6674
    },
    {
      "epoch": 0.5051795090109267,
      "grad_norm": 0.8046875,
      "learning_rate": 1.7044910826305157e-05,
      "loss": 0.3387,
      "step": 6675
    },
    {
      "epoch": 0.5052551913343739,
      "grad_norm": 0.7421875,
      "learning_rate": 1.704406545155399e-05,
      "loss": 0.273,
      "step": 6676
    },
    {
      "epoch": 0.5053308736578213,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7043219976870665e-05,
      "loss": 0.3392,
      "step": 6677
    },
    {
      "epoch": 0.5054065559812686,
      "grad_norm": 0.8984375,
      "learning_rate": 1.704237440226718e-05,
      "loss": 0.3333,
      "step": 6678
    },
    {
      "epoch": 0.505482238304716,
      "grad_norm": 0.7421875,
      "learning_rate": 1.704152872775553e-05,
      "loss": 0.3093,
      "step": 6679
    },
    {
      "epoch": 0.5055579206281633,
      "grad_norm": 0.8125,
      "learning_rate": 1.704068295334771e-05,
      "loss": 0.3104,
      "step": 6680
    },
    {
      "epoch": 0.5056336029516106,
      "grad_norm": 0.7890625,
      "learning_rate": 1.703983707905572e-05,
      "loss": 0.2974,
      "step": 6681
    },
    {
      "epoch": 0.505709285275058,
      "grad_norm": 0.8984375,
      "learning_rate": 1.703899110489156e-05,
      "loss": 0.3434,
      "step": 6682
    },
    {
      "epoch": 0.5057849675985052,
      "grad_norm": 0.83203125,
      "learning_rate": 1.703814503086723e-05,
      "loss": 0.3653,
      "step": 6683
    },
    {
      "epoch": 0.5058606499219526,
      "grad_norm": 3.34375,
      "learning_rate": 1.703729885699474e-05,
      "loss": 0.3513,
      "step": 6684
    },
    {
      "epoch": 0.5059363322453999,
      "grad_norm": 0.7734375,
      "learning_rate": 1.7036452583286085e-05,
      "loss": 0.3073,
      "step": 6685
    },
    {
      "epoch": 0.5060120145688473,
      "grad_norm": 0.77734375,
      "learning_rate": 1.7035606209753276e-05,
      "loss": 0.3119,
      "step": 6686
    },
    {
      "epoch": 0.5060876968922946,
      "grad_norm": 0.76171875,
      "learning_rate": 1.703475973640832e-05,
      "loss": 0.3032,
      "step": 6687
    },
    {
      "epoch": 0.506163379215742,
      "grad_norm": 0.74609375,
      "learning_rate": 1.7033913163263224e-05,
      "loss": 0.3188,
      "step": 6688
    },
    {
      "epoch": 0.5062390615391893,
      "grad_norm": 0.828125,
      "learning_rate": 1.703306649033e-05,
      "loss": 0.3817,
      "step": 6689
    },
    {
      "epoch": 0.5063147438626365,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7032219717620663e-05,
      "loss": 0.3303,
      "step": 6690
    },
    {
      "epoch": 0.5063904261860839,
      "grad_norm": 0.76953125,
      "learning_rate": 1.7031372845147212e-05,
      "loss": 0.3222,
      "step": 6691
    },
    {
      "epoch": 0.5064661085095312,
      "grad_norm": 0.73046875,
      "learning_rate": 1.703052587292168e-05,
      "loss": 0.264,
      "step": 6692
    },
    {
      "epoch": 0.5065417908329786,
      "grad_norm": 0.76171875,
      "learning_rate": 1.702967880095607e-05,
      "loss": 0.3297,
      "step": 6693
    },
    {
      "epoch": 0.5066174731564259,
      "grad_norm": 1.2890625,
      "learning_rate": 1.70288316292624e-05,
      "loss": 0.3611,
      "step": 6694
    },
    {
      "epoch": 0.5066931554798733,
      "grad_norm": 0.78515625,
      "learning_rate": 1.7027984357852697e-05,
      "loss": 0.3472,
      "step": 6695
    },
    {
      "epoch": 0.5067688378033206,
      "grad_norm": 0.77734375,
      "learning_rate": 1.7027136986738974e-05,
      "loss": 0.3124,
      "step": 6696
    },
    {
      "epoch": 0.5068445201267678,
      "grad_norm": 0.78125,
      "learning_rate": 1.702628951593325e-05,
      "loss": 0.3248,
      "step": 6697
    },
    {
      "epoch": 0.5069202024502152,
      "grad_norm": 0.74609375,
      "learning_rate": 1.7025441945447557e-05,
      "loss": 0.3065,
      "step": 6698
    },
    {
      "epoch": 0.5069958847736625,
      "grad_norm": 0.828125,
      "learning_rate": 1.7024594275293907e-05,
      "loss": 0.349,
      "step": 6699
    },
    {
      "epoch": 0.5070715670971099,
      "grad_norm": 0.8359375,
      "learning_rate": 1.7023746505484337e-05,
      "loss": 0.3405,
      "step": 6700
    },
    {
      "epoch": 0.5071472494205572,
      "grad_norm": 1.21875,
      "learning_rate": 1.7022898636030867e-05,
      "loss": 0.4052,
      "step": 6701
    },
    {
      "epoch": 0.5072229317440046,
      "grad_norm": 0.71484375,
      "learning_rate": 1.7022050666945534e-05,
      "loss": 0.2768,
      "step": 6702
    },
    {
      "epoch": 0.5072986140674519,
      "grad_norm": 0.90625,
      "learning_rate": 1.702120259824036e-05,
      "loss": 0.3702,
      "step": 6703
    },
    {
      "epoch": 0.5073742963908991,
      "grad_norm": 0.83203125,
      "learning_rate": 1.7020354429927376e-05,
      "loss": 0.369,
      "step": 6704
    },
    {
      "epoch": 0.5074499787143465,
      "grad_norm": 0.796875,
      "learning_rate": 1.7019506162018617e-05,
      "loss": 0.3317,
      "step": 6705
    },
    {
      "epoch": 0.5075256610377938,
      "grad_norm": 0.74609375,
      "learning_rate": 1.701865779452612e-05,
      "loss": 0.3226,
      "step": 6706
    },
    {
      "epoch": 0.5076013433612412,
      "grad_norm": 0.796875,
      "learning_rate": 1.7017809327461914e-05,
      "loss": 0.3527,
      "step": 6707
    },
    {
      "epoch": 0.5076770256846885,
      "grad_norm": 0.8359375,
      "learning_rate": 1.7016960760838045e-05,
      "loss": 0.3396,
      "step": 6708
    },
    {
      "epoch": 0.5077527080081359,
      "grad_norm": 0.79296875,
      "learning_rate": 1.701611209466654e-05,
      "loss": 0.3121,
      "step": 6709
    },
    {
      "epoch": 0.5078283903315832,
      "grad_norm": 0.7890625,
      "learning_rate": 1.701526332895945e-05,
      "loss": 0.3402,
      "step": 6710
    },
    {
      "epoch": 0.5079040726550305,
      "grad_norm": 0.73828125,
      "learning_rate": 1.7014414463728806e-05,
      "loss": 0.3145,
      "step": 6711
    },
    {
      "epoch": 0.5079797549784778,
      "grad_norm": 0.73828125,
      "learning_rate": 1.7013565498986662e-05,
      "loss": 0.3138,
      "step": 6712
    },
    {
      "epoch": 0.5080554373019252,
      "grad_norm": 0.76171875,
      "learning_rate": 1.7012716434745052e-05,
      "loss": 0.3522,
      "step": 6713
    },
    {
      "epoch": 0.5081311196253725,
      "grad_norm": 0.85546875,
      "learning_rate": 1.7011867271016023e-05,
      "loss": 0.4048,
      "step": 6714
    },
    {
      "epoch": 0.5082068019488198,
      "grad_norm": 0.8125,
      "learning_rate": 1.701101800781163e-05,
      "loss": 0.3423,
      "step": 6715
    },
    {
      "epoch": 0.5082824842722672,
      "grad_norm": 0.76953125,
      "learning_rate": 1.701016864514391e-05,
      "loss": 0.335,
      "step": 6716
    },
    {
      "epoch": 0.5083581665957145,
      "grad_norm": 36.0,
      "learning_rate": 1.7009319183024916e-05,
      "loss": 0.4992,
      "step": 6717
    },
    {
      "epoch": 0.5084338489191618,
      "grad_norm": 0.7578125,
      "learning_rate": 1.7008469621466708e-05,
      "loss": 0.3183,
      "step": 6718
    },
    {
      "epoch": 0.5085095312426091,
      "grad_norm": 0.8203125,
      "learning_rate": 1.7007619960481326e-05,
      "loss": 0.354,
      "step": 6719
    },
    {
      "epoch": 0.5085852135660565,
      "grad_norm": 0.79296875,
      "learning_rate": 1.700677020008083e-05,
      "loss": 0.3285,
      "step": 6720
    },
    {
      "epoch": 0.5086608958895038,
      "grad_norm": 0.7265625,
      "learning_rate": 1.7005920340277276e-05,
      "loss": 0.2658,
      "step": 6721
    },
    {
      "epoch": 0.5087365782129512,
      "grad_norm": 0.71875,
      "learning_rate": 1.7005070381082724e-05,
      "loss": 0.2755,
      "step": 6722
    },
    {
      "epoch": 0.5088122605363985,
      "grad_norm": 0.83203125,
      "learning_rate": 1.700422032250922e-05,
      "loss": 0.3215,
      "step": 6723
    },
    {
      "epoch": 0.5088879428598458,
      "grad_norm": 0.796875,
      "learning_rate": 1.7003370164568834e-05,
      "loss": 0.3231,
      "step": 6724
    },
    {
      "epoch": 0.5089636251832931,
      "grad_norm": 0.7890625,
      "learning_rate": 1.7002519907273623e-05,
      "loss": 0.3272,
      "step": 6725
    },
    {
      "epoch": 0.5090393075067404,
      "grad_norm": 0.8046875,
      "learning_rate": 1.7001669550635653e-05,
      "loss": 0.333,
      "step": 6726
    },
    {
      "epoch": 0.5091149898301878,
      "grad_norm": 0.81640625,
      "learning_rate": 1.7000819094666984e-05,
      "loss": 0.3644,
      "step": 6727
    },
    {
      "epoch": 0.5091906721536351,
      "grad_norm": 0.80078125,
      "learning_rate": 1.699996853937968e-05,
      "loss": 0.3472,
      "step": 6728
    },
    {
      "epoch": 0.5092663544770825,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6999117884785813e-05,
      "loss": 0.2749,
      "step": 6729
    },
    {
      "epoch": 0.5093420368005298,
      "grad_norm": 0.90234375,
      "learning_rate": 1.6998267130897445e-05,
      "loss": 0.385,
      "step": 6730
    },
    {
      "epoch": 0.5094177191239772,
      "grad_norm": 0.84375,
      "learning_rate": 1.699741627772665e-05,
      "loss": 0.3757,
      "step": 6731
    },
    {
      "epoch": 0.5094934014474244,
      "grad_norm": 0.70703125,
      "learning_rate": 1.69965653252855e-05,
      "loss": 0.2822,
      "step": 6732
    },
    {
      "epoch": 0.5095690837708717,
      "grad_norm": 0.94921875,
      "learning_rate": 1.699571427358606e-05,
      "loss": 0.312,
      "step": 6733
    },
    {
      "epoch": 0.5096447660943191,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6994863122640407e-05,
      "loss": 0.3312,
      "step": 6734
    },
    {
      "epoch": 0.5097204484177664,
      "grad_norm": 0.7890625,
      "learning_rate": 1.699401187246062e-05,
      "loss": 0.3557,
      "step": 6735
    },
    {
      "epoch": 0.5097961307412138,
      "grad_norm": 0.83984375,
      "learning_rate": 1.6993160523058775e-05,
      "loss": 0.33,
      "step": 6736
    },
    {
      "epoch": 0.5098718130646611,
      "grad_norm": 0.85546875,
      "learning_rate": 1.6992309074446943e-05,
      "loss": 0.3425,
      "step": 6737
    },
    {
      "epoch": 0.5099474953881085,
      "grad_norm": 40.75,
      "learning_rate": 1.699145752663721e-05,
      "loss": 0.3391,
      "step": 6738
    },
    {
      "epoch": 0.5100231777115557,
      "grad_norm": 0.8046875,
      "learning_rate": 1.6990605879641652e-05,
      "loss": 0.3045,
      "step": 6739
    },
    {
      "epoch": 0.510098860035003,
      "grad_norm": 0.703125,
      "learning_rate": 1.6989754133472357e-05,
      "loss": 0.2727,
      "step": 6740
    },
    {
      "epoch": 0.5101745423584504,
      "grad_norm": 0.79296875,
      "learning_rate": 1.69889022881414e-05,
      "loss": 0.3148,
      "step": 6741
    },
    {
      "epoch": 0.5102502246818977,
      "grad_norm": 0.79296875,
      "learning_rate": 1.698805034366087e-05,
      "loss": 0.2984,
      "step": 6742
    },
    {
      "epoch": 0.5103259070053451,
      "grad_norm": 0.73828125,
      "learning_rate": 1.6987198300042858e-05,
      "loss": 0.2968,
      "step": 6743
    },
    {
      "epoch": 0.5104015893287924,
      "grad_norm": 0.8046875,
      "learning_rate": 1.6986346157299446e-05,
      "loss": 0.3172,
      "step": 6744
    },
    {
      "epoch": 0.5104772716522398,
      "grad_norm": 0.7109375,
      "learning_rate": 1.6985493915442726e-05,
      "loss": 0.2746,
      "step": 6745
    },
    {
      "epoch": 0.510552953975687,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6984641574484788e-05,
      "loss": 0.3161,
      "step": 6746
    },
    {
      "epoch": 0.5106286362991344,
      "grad_norm": 0.73828125,
      "learning_rate": 1.698378913443772e-05,
      "loss": 0.3143,
      "step": 6747
    },
    {
      "epoch": 0.5107043186225817,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6982936595313622e-05,
      "loss": 0.2908,
      "step": 6748
    },
    {
      "epoch": 0.510780000946029,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6982083957124582e-05,
      "loss": 0.3341,
      "step": 6749
    },
    {
      "epoch": 0.5108556832694764,
      "grad_norm": 0.87890625,
      "learning_rate": 1.69812312198827e-05,
      "loss": 0.3625,
      "step": 6750
    },
    {
      "epoch": 0.5109313655929237,
      "grad_norm": 0.7421875,
      "learning_rate": 1.6980378383600077e-05,
      "loss": 0.3292,
      "step": 6751
    },
    {
      "epoch": 0.5110070479163711,
      "grad_norm": 0.8046875,
      "learning_rate": 1.6979525448288803e-05,
      "loss": 0.3173,
      "step": 6752
    },
    {
      "epoch": 0.5110827302398183,
      "grad_norm": 0.71875,
      "learning_rate": 1.6978672413960984e-05,
      "loss": 0.3017,
      "step": 6753
    },
    {
      "epoch": 0.5111584125632657,
      "grad_norm": 0.79296875,
      "learning_rate": 1.697781928062872e-05,
      "loss": 0.3001,
      "step": 6754
    },
    {
      "epoch": 0.511234094886713,
      "grad_norm": 0.796875,
      "learning_rate": 1.697696604830412e-05,
      "loss": 0.3318,
      "step": 6755
    },
    {
      "epoch": 0.5113097772101604,
      "grad_norm": 0.828125,
      "learning_rate": 1.697611271699928e-05,
      "loss": 0.3377,
      "step": 6756
    },
    {
      "epoch": 0.5113854595336077,
      "grad_norm": 0.73828125,
      "learning_rate": 1.6975259286726312e-05,
      "loss": 0.3208,
      "step": 6757
    },
    {
      "epoch": 0.511461141857055,
      "grad_norm": 0.77734375,
      "learning_rate": 1.6974405757497318e-05,
      "loss": 0.3226,
      "step": 6758
    },
    {
      "epoch": 0.5115368241805024,
      "grad_norm": 0.76171875,
      "learning_rate": 1.6973552129324413e-05,
      "loss": 0.3206,
      "step": 6759
    },
    {
      "epoch": 0.5116125065039496,
      "grad_norm": 0.7734375,
      "learning_rate": 1.69726984022197e-05,
      "loss": 0.3205,
      "step": 6760
    },
    {
      "epoch": 0.511688188827397,
      "grad_norm": 0.8046875,
      "learning_rate": 1.6971844576195297e-05,
      "loss": 0.301,
      "step": 6761
    },
    {
      "epoch": 0.5117638711508443,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6970990651263314e-05,
      "loss": 0.295,
      "step": 6762
    },
    {
      "epoch": 0.5118395534742917,
      "grad_norm": 0.78125,
      "learning_rate": 1.6970136627435868e-05,
      "loss": 0.339,
      "step": 6763
    },
    {
      "epoch": 0.511915235797739,
      "grad_norm": 0.78515625,
      "learning_rate": 1.696928250472507e-05,
      "loss": 0.3037,
      "step": 6764
    },
    {
      "epoch": 0.5119909181211864,
      "grad_norm": 0.8984375,
      "learning_rate": 1.696842828314304e-05,
      "loss": 0.3799,
      "step": 6765
    },
    {
      "epoch": 0.5120666004446337,
      "grad_norm": 0.7421875,
      "learning_rate": 1.69675739627019e-05,
      "loss": 0.2802,
      "step": 6766
    },
    {
      "epoch": 0.5121422827680809,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6966719543413767e-05,
      "loss": 0.3288,
      "step": 6767
    },
    {
      "epoch": 0.5122179650915283,
      "grad_norm": 1.5625,
      "learning_rate": 1.6965865025290758e-05,
      "loss": 0.4399,
      "step": 6768
    },
    {
      "epoch": 0.5122936474149756,
      "grad_norm": 0.875,
      "learning_rate": 1.6965010408345e-05,
      "loss": 0.3997,
      "step": 6769
    },
    {
      "epoch": 0.512369329738423,
      "grad_norm": 0.734375,
      "learning_rate": 1.696415569258862e-05,
      "loss": 0.3058,
      "step": 6770
    },
    {
      "epoch": 0.5124450120618703,
      "grad_norm": 0.77734375,
      "learning_rate": 1.6963300878033737e-05,
      "loss": 0.3306,
      "step": 6771
    },
    {
      "epoch": 0.5125206943853177,
      "grad_norm": 0.74609375,
      "learning_rate": 1.6962445964692483e-05,
      "loss": 0.3056,
      "step": 6772
    },
    {
      "epoch": 0.512596376708765,
      "grad_norm": 0.8203125,
      "learning_rate": 1.6961590952576986e-05,
      "loss": 0.3247,
      "step": 6773
    },
    {
      "epoch": 0.5126720590322122,
      "grad_norm": 0.8203125,
      "learning_rate": 1.6960735841699375e-05,
      "loss": 0.3535,
      "step": 6774
    },
    {
      "epoch": 0.5127477413556596,
      "grad_norm": 5.125,
      "learning_rate": 1.695988063207178e-05,
      "loss": 0.447,
      "step": 6775
    },
    {
      "epoch": 0.5128234236791069,
      "grad_norm": 0.76953125,
      "learning_rate": 1.6959025323706333e-05,
      "loss": 0.3166,
      "step": 6776
    },
    {
      "epoch": 0.5128991060025543,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6958169916615174e-05,
      "loss": 0.3134,
      "step": 6777
    },
    {
      "epoch": 0.5129747883260016,
      "grad_norm": 0.8125,
      "learning_rate": 1.6957314410810433e-05,
      "loss": 0.3131,
      "step": 6778
    },
    {
      "epoch": 0.513050470649449,
      "grad_norm": 0.81640625,
      "learning_rate": 1.6956458806304247e-05,
      "loss": 0.364,
      "step": 6779
    },
    {
      "epoch": 0.5131261529728963,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6955603103108754e-05,
      "loss": 0.3423,
      "step": 6780
    },
    {
      "epoch": 0.5132018352963436,
      "grad_norm": 0.7109375,
      "learning_rate": 1.695474730123609e-05,
      "loss": 0.2872,
      "step": 6781
    },
    {
      "epoch": 0.5132775176197909,
      "grad_norm": 0.765625,
      "learning_rate": 1.695389140069841e-05,
      "loss": 0.3201,
      "step": 6782
    },
    {
      "epoch": 0.5133531999432382,
      "grad_norm": 0.72265625,
      "learning_rate": 1.6953035401507845e-05,
      "loss": 0.2808,
      "step": 6783
    },
    {
      "epoch": 0.5134288822666856,
      "grad_norm": 0.80078125,
      "learning_rate": 1.6952179303676537e-05,
      "loss": 0.345,
      "step": 6784
    },
    {
      "epoch": 0.5135045645901329,
      "grad_norm": 0.73046875,
      "learning_rate": 1.695132310721664e-05,
      "loss": 0.2985,
      "step": 6785
    },
    {
      "epoch": 0.5135802469135803,
      "grad_norm": 0.765625,
      "learning_rate": 1.6950466812140292e-05,
      "loss": 0.324,
      "step": 6786
    },
    {
      "epoch": 0.5136559292370276,
      "grad_norm": 0.703125,
      "learning_rate": 1.6949610418459648e-05,
      "loss": 0.2703,
      "step": 6787
    },
    {
      "epoch": 0.5137316115604749,
      "grad_norm": 0.7109375,
      "learning_rate": 1.6948753926186852e-05,
      "loss": 0.2919,
      "step": 6788
    },
    {
      "epoch": 0.5138072938839222,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6947897335334058e-05,
      "loss": 0.31,
      "step": 6789
    },
    {
      "epoch": 0.5138829762073696,
      "grad_norm": 0.8203125,
      "learning_rate": 1.694704064591342e-05,
      "loss": 0.3675,
      "step": 6790
    },
    {
      "epoch": 0.5139586585308169,
      "grad_norm": 0.8046875,
      "learning_rate": 1.694618385793708e-05,
      "loss": 0.3479,
      "step": 6791
    },
    {
      "epoch": 0.5140343408542642,
      "grad_norm": 0.83984375,
      "learning_rate": 1.694532697141721e-05,
      "loss": 0.3654,
      "step": 6792
    },
    {
      "epoch": 0.5141100231777116,
      "grad_norm": 0.74609375,
      "learning_rate": 1.694446998636596e-05,
      "loss": 0.3091,
      "step": 6793
    },
    {
      "epoch": 0.5141857055011589,
      "grad_norm": 0.88671875,
      "learning_rate": 1.694361290279548e-05,
      "loss": 0.3221,
      "step": 6794
    },
    {
      "epoch": 0.5142613878246062,
      "grad_norm": 0.8203125,
      "learning_rate": 1.694275572071794e-05,
      "loss": 0.3538,
      "step": 6795
    },
    {
      "epoch": 0.5143370701480535,
      "grad_norm": 0.7265625,
      "learning_rate": 1.694189844014549e-05,
      "loss": 0.275,
      "step": 6796
    },
    {
      "epoch": 0.5144127524715009,
      "grad_norm": 0.72265625,
      "learning_rate": 1.6941041061090305e-05,
      "loss": 0.2879,
      "step": 6797
    },
    {
      "epoch": 0.5144884347949482,
      "grad_norm": 0.81640625,
      "learning_rate": 1.6940183583564536e-05,
      "loss": 0.3507,
      "step": 6798
    },
    {
      "epoch": 0.5145641171183956,
      "grad_norm": 0.765625,
      "learning_rate": 1.6939326007580355e-05,
      "loss": 0.3244,
      "step": 6799
    },
    {
      "epoch": 0.5146397994418429,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6938468333149923e-05,
      "loss": 0.3303,
      "step": 6800
    },
    {
      "epoch": 0.5147154817652901,
      "grad_norm": 0.796875,
      "learning_rate": 1.693761056028542e-05,
      "loss": 0.3519,
      "step": 6801
    },
    {
      "epoch": 0.5147911640887375,
      "grad_norm": 0.83984375,
      "learning_rate": 1.6936752688998997e-05,
      "loss": 0.3598,
      "step": 6802
    },
    {
      "epoch": 0.5148668464121848,
      "grad_norm": 0.83203125,
      "learning_rate": 1.6935894719302837e-05,
      "loss": 0.3923,
      "step": 6803
    },
    {
      "epoch": 0.5149425287356322,
      "grad_norm": 0.83203125,
      "learning_rate": 1.6935036651209106e-05,
      "loss": 0.3287,
      "step": 6804
    },
    {
      "epoch": 0.5150182110590795,
      "grad_norm": 0.8203125,
      "learning_rate": 1.693417848472998e-05,
      "loss": 0.3614,
      "step": 6805
    },
    {
      "epoch": 0.5150938933825269,
      "grad_norm": 0.75,
      "learning_rate": 1.6933320219877634e-05,
      "loss": 0.3029,
      "step": 6806
    },
    {
      "epoch": 0.5151695757059742,
      "grad_norm": 0.80859375,
      "learning_rate": 1.693246185666424e-05,
      "loss": 0.3642,
      "step": 6807
    },
    {
      "epoch": 0.5152452580294214,
      "grad_norm": 0.7734375,
      "learning_rate": 1.693160339510198e-05,
      "loss": 0.3387,
      "step": 6808
    },
    {
      "epoch": 0.5153209403528688,
      "grad_norm": 0.78125,
      "learning_rate": 1.6930744835203032e-05,
      "loss": 0.3366,
      "step": 6809
    },
    {
      "epoch": 0.5153966226763161,
      "grad_norm": 0.76171875,
      "learning_rate": 1.6929886176979576e-05,
      "loss": 0.2829,
      "step": 6810
    },
    {
      "epoch": 0.5154723049997635,
      "grad_norm": 0.74609375,
      "learning_rate": 1.6929027420443788e-05,
      "loss": 0.32,
      "step": 6811
    },
    {
      "epoch": 0.5155479873232108,
      "grad_norm": 0.7265625,
      "learning_rate": 1.6928168565607857e-05,
      "loss": 0.2742,
      "step": 6812
    },
    {
      "epoch": 0.5156236696466582,
      "grad_norm": 1.3984375,
      "learning_rate": 1.6927309612483965e-05,
      "loss": 0.3801,
      "step": 6813
    },
    {
      "epoch": 0.5156993519701055,
      "grad_norm": 0.73046875,
      "learning_rate": 1.6926450561084302e-05,
      "loss": 0.2909,
      "step": 6814
    },
    {
      "epoch": 0.5157750342935528,
      "grad_norm": 0.70703125,
      "learning_rate": 1.6925591411421048e-05,
      "loss": 0.2753,
      "step": 6815
    },
    {
      "epoch": 0.5158507166170001,
      "grad_norm": 0.8359375,
      "learning_rate": 1.69247321635064e-05,
      "loss": 0.3279,
      "step": 6816
    },
    {
      "epoch": 0.5159263989404475,
      "grad_norm": 0.83203125,
      "learning_rate": 1.692387281735254e-05,
      "loss": 0.3548,
      "step": 6817
    },
    {
      "epoch": 0.5160020812638948,
      "grad_norm": 0.74609375,
      "learning_rate": 1.692301337297166e-05,
      "loss": 0.3038,
      "step": 6818
    },
    {
      "epoch": 0.5160777635873421,
      "grad_norm": 0.79296875,
      "learning_rate": 1.6922153830375956e-05,
      "loss": 0.3332,
      "step": 6819
    },
    {
      "epoch": 0.5161534459107895,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6921294189577622e-05,
      "loss": 0.3461,
      "step": 6820
    },
    {
      "epoch": 0.5162291282342368,
      "grad_norm": 0.8046875,
      "learning_rate": 1.6920434450588854e-05,
      "loss": 0.3845,
      "step": 6821
    },
    {
      "epoch": 0.5163048105576841,
      "grad_norm": 0.79296875,
      "learning_rate": 1.6919574613421845e-05,
      "loss": 0.3094,
      "step": 6822
    },
    {
      "epoch": 0.5163804928811314,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6918714678088797e-05,
      "loss": 0.3071,
      "step": 6823
    },
    {
      "epoch": 0.5164561752045788,
      "grad_norm": 0.7890625,
      "learning_rate": 1.691785464460191e-05,
      "loss": 0.3144,
      "step": 6824
    },
    {
      "epoch": 0.5165318575280261,
      "grad_norm": 0.77734375,
      "learning_rate": 1.691699451297338e-05,
      "loss": 0.3301,
      "step": 6825
    },
    {
      "epoch": 0.5166075398514735,
      "grad_norm": 0.71875,
      "learning_rate": 1.6916134283215412e-05,
      "loss": 0.2758,
      "step": 6826
    },
    {
      "epoch": 0.5166832221749208,
      "grad_norm": 0.76953125,
      "learning_rate": 1.6915273955340214e-05,
      "loss": 0.3169,
      "step": 6827
    },
    {
      "epoch": 0.5167589044983681,
      "grad_norm": 0.734375,
      "learning_rate": 1.6914413529359987e-05,
      "loss": 0.2957,
      "step": 6828
    },
    {
      "epoch": 0.5168345868218154,
      "grad_norm": 0.765625,
      "learning_rate": 1.6913553005286937e-05,
      "loss": 0.3384,
      "step": 6829
    },
    {
      "epoch": 0.5169102691452627,
      "grad_norm": 0.81640625,
      "learning_rate": 1.6912692383133276e-05,
      "loss": 0.3265,
      "step": 6830
    },
    {
      "epoch": 0.5169859514687101,
      "grad_norm": 0.734375,
      "learning_rate": 1.691183166291121e-05,
      "loss": 0.2836,
      "step": 6831
    },
    {
      "epoch": 0.5170616337921574,
      "grad_norm": 0.7578125,
      "learning_rate": 1.691097084463295e-05,
      "loss": 0.2907,
      "step": 6832
    },
    {
      "epoch": 0.5171373161156048,
      "grad_norm": 0.84765625,
      "learning_rate": 1.691010992831071e-05,
      "loss": 0.3721,
      "step": 6833
    },
    {
      "epoch": 0.5172129984390521,
      "grad_norm": 0.76171875,
      "learning_rate": 1.69092489139567e-05,
      "loss": 0.301,
      "step": 6834
    },
    {
      "epoch": 0.5172886807624995,
      "grad_norm": 0.76171875,
      "learning_rate": 1.690838780158314e-05,
      "loss": 0.2917,
      "step": 6835
    },
    {
      "epoch": 0.5173643630859467,
      "grad_norm": 0.828125,
      "learning_rate": 1.6907526591202246e-05,
      "loss": 0.2912,
      "step": 6836
    },
    {
      "epoch": 0.517440045409394,
      "grad_norm": 0.7109375,
      "learning_rate": 1.690666528282623e-05,
      "loss": 0.2777,
      "step": 6837
    },
    {
      "epoch": 0.5175157277328414,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6905803876467317e-05,
      "loss": 0.3051,
      "step": 6838
    },
    {
      "epoch": 0.5175914100562887,
      "grad_norm": 0.7421875,
      "learning_rate": 1.6904942372137725e-05,
      "loss": 0.3078,
      "step": 6839
    },
    {
      "epoch": 0.5176670923797361,
      "grad_norm": 0.6953125,
      "learning_rate": 1.6904080769849674e-05,
      "loss": 0.2989,
      "step": 6840
    },
    {
      "epoch": 0.5177427747031834,
      "grad_norm": 0.734375,
      "learning_rate": 1.6903219069615388e-05,
      "loss": 0.2814,
      "step": 6841
    },
    {
      "epoch": 0.5178184570266308,
      "grad_norm": 0.7734375,
      "learning_rate": 1.69023572714471e-05,
      "loss": 0.3152,
      "step": 6842
    },
    {
      "epoch": 0.517894139350078,
      "grad_norm": 0.8203125,
      "learning_rate": 1.6901495375357023e-05,
      "loss": 0.3488,
      "step": 6843
    },
    {
      "epoch": 0.5179698216735253,
      "grad_norm": 0.71875,
      "learning_rate": 1.6900633381357393e-05,
      "loss": 0.2875,
      "step": 6844
    },
    {
      "epoch": 0.5180455039969727,
      "grad_norm": 0.80859375,
      "learning_rate": 1.6899771289460436e-05,
      "loss": 0.3534,
      "step": 6845
    },
    {
      "epoch": 0.51812118632042,
      "grad_norm": 0.765625,
      "learning_rate": 1.6898909099678383e-05,
      "loss": 0.3212,
      "step": 6846
    },
    {
      "epoch": 0.5181968686438674,
      "grad_norm": 1.265625,
      "learning_rate": 1.6898046812023465e-05,
      "loss": 0.3683,
      "step": 6847
    },
    {
      "epoch": 0.5182725509673147,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6897184426507917e-05,
      "loss": 0.3088,
      "step": 6848
    },
    {
      "epoch": 0.5183482332907621,
      "grad_norm": 0.78125,
      "learning_rate": 1.689632194314397e-05,
      "loss": 0.3285,
      "step": 6849
    },
    {
      "epoch": 0.5184239156142093,
      "grad_norm": 0.8125,
      "learning_rate": 1.6895459361943865e-05,
      "loss": 0.3716,
      "step": 6850
    },
    {
      "epoch": 0.5184995979376567,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6894596682919834e-05,
      "loss": 0.3155,
      "step": 6851
    },
    {
      "epoch": 0.518575280261104,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6893733906084118e-05,
      "loss": 0.3417,
      "step": 6852
    },
    {
      "epoch": 0.5186509625845513,
      "grad_norm": 0.84765625,
      "learning_rate": 1.6892871031448957e-05,
      "loss": 0.3564,
      "step": 6853
    },
    {
      "epoch": 0.5187266449079987,
      "grad_norm": 0.81640625,
      "learning_rate": 1.6892008059026587e-05,
      "loss": 0.3638,
      "step": 6854
    },
    {
      "epoch": 0.518802327231446,
      "grad_norm": 0.7734375,
      "learning_rate": 1.689114498882926e-05,
      "loss": 0.3638,
      "step": 6855
    },
    {
      "epoch": 0.5188780095548934,
      "grad_norm": 0.71875,
      "learning_rate": 1.6890281820869215e-05,
      "loss": 0.2957,
      "step": 6856
    },
    {
      "epoch": 0.5189536918783406,
      "grad_norm": 0.71875,
      "learning_rate": 1.68894185551587e-05,
      "loss": 0.3024,
      "step": 6857
    },
    {
      "epoch": 0.519029374201788,
      "grad_norm": 0.76953125,
      "learning_rate": 1.688855519170996e-05,
      "loss": 0.349,
      "step": 6858
    },
    {
      "epoch": 0.5191050565252353,
      "grad_norm": 0.74609375,
      "learning_rate": 1.6887691730535237e-05,
      "loss": 0.3213,
      "step": 6859
    },
    {
      "epoch": 0.5191807388486827,
      "grad_norm": 0.828125,
      "learning_rate": 1.688682817164679e-05,
      "loss": 0.3621,
      "step": 6860
    },
    {
      "epoch": 0.51925642117213,
      "grad_norm": 0.73828125,
      "learning_rate": 1.688596451505687e-05,
      "loss": 0.3153,
      "step": 6861
    },
    {
      "epoch": 0.5193321034955773,
      "grad_norm": 0.70703125,
      "learning_rate": 1.688510076077773e-05,
      "loss": 0.2621,
      "step": 6862
    },
    {
      "epoch": 0.5194077858190247,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6884236908821617e-05,
      "loss": 0.3207,
      "step": 6863
    },
    {
      "epoch": 0.5194834681424719,
      "grad_norm": 0.81640625,
      "learning_rate": 1.6883372959200786e-05,
      "loss": 0.3628,
      "step": 6864
    },
    {
      "epoch": 0.5195591504659193,
      "grad_norm": 0.7890625,
      "learning_rate": 1.68825089119275e-05,
      "loss": 0.3414,
      "step": 6865
    },
    {
      "epoch": 0.5196348327893666,
      "grad_norm": 0.6875,
      "learning_rate": 1.6881644767014017e-05,
      "loss": 0.2913,
      "step": 6866
    },
    {
      "epoch": 0.519710515112814,
      "grad_norm": 0.87890625,
      "learning_rate": 1.688078052447259e-05,
      "loss": 0.406,
      "step": 6867
    },
    {
      "epoch": 0.5197861974362613,
      "grad_norm": 0.7421875,
      "learning_rate": 1.6879916184315485e-05,
      "loss": 0.3074,
      "step": 6868
    },
    {
      "epoch": 0.5198618797597087,
      "grad_norm": 0.77734375,
      "learning_rate": 1.6879051746554963e-05,
      "loss": 0.3153,
      "step": 6869
    },
    {
      "epoch": 0.519937562083156,
      "grad_norm": 0.68359375,
      "learning_rate": 1.6878187211203286e-05,
      "loss": 0.2873,
      "step": 6870
    },
    {
      "epoch": 0.5200132444066032,
      "grad_norm": 0.80078125,
      "learning_rate": 1.6877322578272722e-05,
      "loss": 0.3491,
      "step": 6871
    },
    {
      "epoch": 0.5200889267300506,
      "grad_norm": 0.875,
      "learning_rate": 1.687645784777553e-05,
      "loss": 0.3768,
      "step": 6872
    },
    {
      "epoch": 0.5201646090534979,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6875593019723988e-05,
      "loss": 0.3297,
      "step": 6873
    },
    {
      "epoch": 0.5202402913769453,
      "grad_norm": 0.8046875,
      "learning_rate": 1.687472809413036e-05,
      "loss": 0.3599,
      "step": 6874
    },
    {
      "epoch": 0.5203159737003926,
      "grad_norm": 0.8828125,
      "learning_rate": 1.6873863071006914e-05,
      "loss": 0.3552,
      "step": 6875
    },
    {
      "epoch": 0.52039165602384,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6872997950365926e-05,
      "loss": 0.3167,
      "step": 6876
    },
    {
      "epoch": 0.5204673383472873,
      "grad_norm": 0.78125,
      "learning_rate": 1.687213273221967e-05,
      "loss": 0.3285,
      "step": 6877
    },
    {
      "epoch": 0.5205430206707345,
      "grad_norm": 0.82421875,
      "learning_rate": 1.687126741658041e-05,
      "loss": 0.3593,
      "step": 6878
    },
    {
      "epoch": 0.5206187029941819,
      "grad_norm": 0.7265625,
      "learning_rate": 1.687040200346044e-05,
      "loss": 0.2754,
      "step": 6879
    },
    {
      "epoch": 0.5206943853176292,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6869536492872023e-05,
      "loss": 0.3092,
      "step": 6880
    },
    {
      "epoch": 0.5207700676410766,
      "grad_norm": 0.83203125,
      "learning_rate": 1.6868670884827443e-05,
      "loss": 0.387,
      "step": 6881
    },
    {
      "epoch": 0.5208457499645239,
      "grad_norm": 0.86328125,
      "learning_rate": 1.6867805179338977e-05,
      "loss": 0.3467,
      "step": 6882
    },
    {
      "epoch": 0.5209214322879713,
      "grad_norm": 0.8359375,
      "learning_rate": 1.686693937641891e-05,
      "loss": 0.3685,
      "step": 6883
    },
    {
      "epoch": 0.5209971146114186,
      "grad_norm": 0.859375,
      "learning_rate": 1.6866073476079524e-05,
      "loss": 0.355,
      "step": 6884
    },
    {
      "epoch": 0.5210727969348659,
      "grad_norm": 0.76953125,
      "learning_rate": 1.6865207478333107e-05,
      "loss": 0.3289,
      "step": 6885
    },
    {
      "epoch": 0.5211484792583132,
      "grad_norm": 0.77734375,
      "learning_rate": 1.6864341383191938e-05,
      "loss": 0.3113,
      "step": 6886
    },
    {
      "epoch": 0.5212241615817605,
      "grad_norm": 0.83984375,
      "learning_rate": 1.6863475190668306e-05,
      "loss": 0.3818,
      "step": 6887
    },
    {
      "epoch": 0.5212998439052079,
      "grad_norm": 0.8203125,
      "learning_rate": 1.68626089007745e-05,
      "loss": 0.3702,
      "step": 6888
    },
    {
      "epoch": 0.5213755262286552,
      "grad_norm": 0.80078125,
      "learning_rate": 1.686174251352281e-05,
      "loss": 0.3403,
      "step": 6889
    },
    {
      "epoch": 0.5214512085521026,
      "grad_norm": 0.765625,
      "learning_rate": 1.686087602892553e-05,
      "loss": 0.3081,
      "step": 6890
    },
    {
      "epoch": 0.5215268908755499,
      "grad_norm": 0.71875,
      "learning_rate": 1.686000944699495e-05,
      "loss": 0.2793,
      "step": 6891
    },
    {
      "epoch": 0.5216025731989972,
      "grad_norm": 0.8125,
      "learning_rate": 1.6859142767743363e-05,
      "loss": 0.329,
      "step": 6892
    },
    {
      "epoch": 0.5216782555224445,
      "grad_norm": 0.765625,
      "learning_rate": 1.6858275991183063e-05,
      "loss": 0.3283,
      "step": 6893
    },
    {
      "epoch": 0.5217539378458919,
      "grad_norm": 0.77734375,
      "learning_rate": 1.685740911732635e-05,
      "loss": 0.336,
      "step": 6894
    },
    {
      "epoch": 0.5218296201693392,
      "grad_norm": 0.74609375,
      "learning_rate": 1.6856542146185522e-05,
      "loss": 0.3314,
      "step": 6895
    },
    {
      "epoch": 0.5219053024927865,
      "grad_norm": 0.8203125,
      "learning_rate": 1.6855675077772875e-05,
      "loss": 0.3381,
      "step": 6896
    },
    {
      "epoch": 0.5219809848162339,
      "grad_norm": 0.79296875,
      "learning_rate": 1.6854807912100716e-05,
      "loss": 0.3464,
      "step": 6897
    },
    {
      "epoch": 0.5220566671396812,
      "grad_norm": 0.80859375,
      "learning_rate": 1.685394064918134e-05,
      "loss": 0.367,
      "step": 6898
    },
    {
      "epoch": 0.5221323494631285,
      "grad_norm": 0.7421875,
      "learning_rate": 1.6853073289027055e-05,
      "loss": 0.3097,
      "step": 6899
    },
    {
      "epoch": 0.5222080317865758,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6852205831650166e-05,
      "loss": 0.3428,
      "step": 6900
    },
    {
      "epoch": 0.5222837141100232,
      "grad_norm": 0.84375,
      "learning_rate": 1.685133827706298e-05,
      "loss": 0.3719,
      "step": 6901
    },
    {
      "epoch": 0.5223593964334705,
      "grad_norm": 0.74609375,
      "learning_rate": 1.68504706252778e-05,
      "loss": 0.3151,
      "step": 6902
    },
    {
      "epoch": 0.5224350787569179,
      "grad_norm": 0.7890625,
      "learning_rate": 1.684960287630694e-05,
      "loss": 0.325,
      "step": 6903
    },
    {
      "epoch": 0.5225107610803652,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6848735030162707e-05,
      "loss": 0.366,
      "step": 6904
    },
    {
      "epoch": 0.5225864434038126,
      "grad_norm": 0.75,
      "learning_rate": 1.6847867086857416e-05,
      "loss": 0.2924,
      "step": 6905
    },
    {
      "epoch": 0.5226621257272598,
      "grad_norm": 0.81640625,
      "learning_rate": 1.684699904640338e-05,
      "loss": 0.3516,
      "step": 6906
    },
    {
      "epoch": 0.5227378080507071,
      "grad_norm": 0.75390625,
      "learning_rate": 1.684613090881291e-05,
      "loss": 0.2948,
      "step": 6907
    },
    {
      "epoch": 0.5228134903741545,
      "grad_norm": 0.74609375,
      "learning_rate": 1.684526267409833e-05,
      "loss": 0.3247,
      "step": 6908
    },
    {
      "epoch": 0.5228891726976018,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6844394342271944e-05,
      "loss": 0.3141,
      "step": 6909
    },
    {
      "epoch": 0.5229648550210492,
      "grad_norm": 0.82421875,
      "learning_rate": 1.6843525913346087e-05,
      "loss": 0.3762,
      "step": 6910
    },
    {
      "epoch": 0.5230405373444965,
      "grad_norm": 0.78125,
      "learning_rate": 1.6842657387333065e-05,
      "loss": 0.2964,
      "step": 6911
    },
    {
      "epoch": 0.5231162196679439,
      "grad_norm": 0.80859375,
      "learning_rate": 1.684178876424521e-05,
      "loss": 0.3395,
      "step": 6912
    },
    {
      "epoch": 0.5231919019913911,
      "grad_norm": 0.73828125,
      "learning_rate": 1.684092004409484e-05,
      "loss": 0.3036,
      "step": 6913
    },
    {
      "epoch": 0.5232675843148384,
      "grad_norm": 0.8125,
      "learning_rate": 1.684005122689428e-05,
      "loss": 0.3463,
      "step": 6914
    },
    {
      "epoch": 0.5233432666382858,
      "grad_norm": 0.8046875,
      "learning_rate": 1.6839182312655853e-05,
      "loss": 0.3422,
      "step": 6915
    },
    {
      "epoch": 0.5234189489617331,
      "grad_norm": 0.73046875,
      "learning_rate": 1.6838313301391887e-05,
      "loss": 0.3038,
      "step": 6916
    },
    {
      "epoch": 0.5234946312851805,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6837444193114712e-05,
      "loss": 0.3269,
      "step": 6917
    },
    {
      "epoch": 0.5235703136086278,
      "grad_norm": 0.7421875,
      "learning_rate": 1.6836574987836663e-05,
      "loss": 0.2799,
      "step": 6918
    },
    {
      "epoch": 0.5236459959320752,
      "grad_norm": 0.84765625,
      "learning_rate": 1.6835705685570062e-05,
      "loss": 0.3708,
      "step": 6919
    },
    {
      "epoch": 0.5237216782555224,
      "grad_norm": 0.8359375,
      "learning_rate": 1.6834836286327245e-05,
      "loss": 0.3436,
      "step": 6920
    },
    {
      "epoch": 0.5237973605789698,
      "grad_norm": 0.83203125,
      "learning_rate": 1.6833966790120548e-05,
      "loss": 0.3864,
      "step": 6921
    },
    {
      "epoch": 0.5238730429024171,
      "grad_norm": 0.88671875,
      "learning_rate": 1.6833097196962304e-05,
      "loss": 0.4003,
      "step": 6922
    },
    {
      "epoch": 0.5239487252258644,
      "grad_norm": 0.77734375,
      "learning_rate": 1.6832227506864853e-05,
      "loss": 0.3463,
      "step": 6923
    },
    {
      "epoch": 0.5240244075493118,
      "grad_norm": 0.6875,
      "learning_rate": 1.6831357719840523e-05,
      "loss": 0.2448,
      "step": 6924
    },
    {
      "epoch": 0.5241000898727591,
      "grad_norm": 0.796875,
      "learning_rate": 1.683048783590167e-05,
      "loss": 0.3582,
      "step": 6925
    },
    {
      "epoch": 0.5241757721962064,
      "grad_norm": 0.75,
      "learning_rate": 1.682961785506062e-05,
      "loss": 0.2914,
      "step": 6926
    },
    {
      "epoch": 0.5242514545196537,
      "grad_norm": 0.77734375,
      "learning_rate": 1.682874777732972e-05,
      "loss": 0.3321,
      "step": 6927
    },
    {
      "epoch": 0.5243271368431011,
      "grad_norm": 0.7421875,
      "learning_rate": 1.6827877602721316e-05,
      "loss": 0.2925,
      "step": 6928
    },
    {
      "epoch": 0.5244028191665484,
      "grad_norm": 0.96484375,
      "learning_rate": 1.682700733124775e-05,
      "loss": 0.299,
      "step": 6929
    },
    {
      "epoch": 0.5244785014899958,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6826136962921373e-05,
      "loss": 0.3325,
      "step": 6930
    },
    {
      "epoch": 0.5245541838134431,
      "grad_norm": 0.85546875,
      "learning_rate": 1.6825266497754524e-05,
      "loss": 0.3733,
      "step": 6931
    },
    {
      "epoch": 0.5246298661368904,
      "grad_norm": 0.71484375,
      "learning_rate": 1.682439593575956e-05,
      "loss": 0.2644,
      "step": 6932
    },
    {
      "epoch": 0.5247055484603377,
      "grad_norm": 0.890625,
      "learning_rate": 1.6823525276948827e-05,
      "loss": 0.3683,
      "step": 6933
    },
    {
      "epoch": 0.524781230783785,
      "grad_norm": 0.83203125,
      "learning_rate": 1.682265452133468e-05,
      "loss": 0.3687,
      "step": 6934
    },
    {
      "epoch": 0.5248569131072324,
      "grad_norm": 0.8359375,
      "learning_rate": 1.6821783668929473e-05,
      "loss": 0.3729,
      "step": 6935
    },
    {
      "epoch": 0.5249325954306797,
      "grad_norm": 0.81640625,
      "learning_rate": 1.6820912719745555e-05,
      "loss": 0.2999,
      "step": 6936
    },
    {
      "epoch": 0.5250082777541271,
      "grad_norm": 0.7265625,
      "learning_rate": 1.6820041673795282e-05,
      "loss": 0.2859,
      "step": 6937
    },
    {
      "epoch": 0.5250839600775744,
      "grad_norm": 0.79296875,
      "learning_rate": 1.6819170531091018e-05,
      "loss": 0.3228,
      "step": 6938
    },
    {
      "epoch": 0.5251596424010218,
      "grad_norm": 0.765625,
      "learning_rate": 1.6818299291645118e-05,
      "loss": 0.2904,
      "step": 6939
    },
    {
      "epoch": 0.525235324724469,
      "grad_norm": 0.83984375,
      "learning_rate": 1.681742795546994e-05,
      "loss": 0.4091,
      "step": 6940
    },
    {
      "epoch": 0.5253110070479163,
      "grad_norm": 0.80859375,
      "learning_rate": 1.6816556522577848e-05,
      "loss": 0.3488,
      "step": 6941
    },
    {
      "epoch": 0.5253866893713637,
      "grad_norm": 0.8046875,
      "learning_rate": 1.681568499298121e-05,
      "loss": 0.3388,
      "step": 6942
    },
    {
      "epoch": 0.525462371694811,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6814813366692377e-05,
      "loss": 0.3437,
      "step": 6943
    },
    {
      "epoch": 0.5255380540182584,
      "grad_norm": 0.8046875,
      "learning_rate": 1.6813941643723722e-05,
      "loss": 0.3406,
      "step": 6944
    },
    {
      "epoch": 0.5256137363417057,
      "grad_norm": 0.84765625,
      "learning_rate": 1.6813069824087613e-05,
      "loss": 0.4133,
      "step": 6945
    },
    {
      "epoch": 0.5256894186651531,
      "grad_norm": 0.7265625,
      "learning_rate": 1.681219790779642e-05,
      "loss": 0.3135,
      "step": 6946
    },
    {
      "epoch": 0.5257651009886003,
      "grad_norm": 0.7265625,
      "learning_rate": 1.6811325894862504e-05,
      "loss": 0.2962,
      "step": 6947
    },
    {
      "epoch": 0.5258407833120476,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6810453785298248e-05,
      "loss": 0.3253,
      "step": 6948
    },
    {
      "epoch": 0.525916465635495,
      "grad_norm": 0.76171875,
      "learning_rate": 1.6809581579116013e-05,
      "loss": 0.3105,
      "step": 6949
    },
    {
      "epoch": 0.5259921479589423,
      "grad_norm": 0.81640625,
      "learning_rate": 1.680870927632818e-05,
      "loss": 0.3606,
      "step": 6950
    },
    {
      "epoch": 0.5260678302823897,
      "grad_norm": 0.7890625,
      "learning_rate": 1.680783687694712e-05,
      "loss": 0.3378,
      "step": 6951
    },
    {
      "epoch": 0.526143512605837,
      "grad_norm": 0.76171875,
      "learning_rate": 1.680696438098521e-05,
      "loss": 0.3292,
      "step": 6952
    },
    {
      "epoch": 0.5262191949292844,
      "grad_norm": 0.79296875,
      "learning_rate": 1.6806091788454832e-05,
      "loss": 0.3207,
      "step": 6953
    },
    {
      "epoch": 0.5262948772527316,
      "grad_norm": 0.76171875,
      "learning_rate": 1.680521909936836e-05,
      "loss": 0.3187,
      "step": 6954
    },
    {
      "epoch": 0.526370559576179,
      "grad_norm": 0.703125,
      "learning_rate": 1.680434631373818e-05,
      "loss": 0.2701,
      "step": 6955
    },
    {
      "epoch": 0.5264462418996263,
      "grad_norm": 0.70703125,
      "learning_rate": 1.6803473431576672e-05,
      "loss": 0.277,
      "step": 6956
    },
    {
      "epoch": 0.5265219242230736,
      "grad_norm": 0.74609375,
      "learning_rate": 1.6802600452896215e-05,
      "loss": 0.3142,
      "step": 6957
    },
    {
      "epoch": 0.526597606546521,
      "grad_norm": 0.828125,
      "learning_rate": 1.6801727377709195e-05,
      "loss": 0.3487,
      "step": 6958
    },
    {
      "epoch": 0.5266732888699683,
      "grad_norm": 0.73046875,
      "learning_rate": 1.6800854206028003e-05,
      "loss": 0.2875,
      "step": 6959
    },
    {
      "epoch": 0.5267489711934157,
      "grad_norm": 0.80078125,
      "learning_rate": 1.679998093786502e-05,
      "loss": 0.3273,
      "step": 6960
    },
    {
      "epoch": 0.5268246535168629,
      "grad_norm": 0.79296875,
      "learning_rate": 1.6799107573232643e-05,
      "loss": 0.331,
      "step": 6961
    },
    {
      "epoch": 0.5269003358403103,
      "grad_norm": 0.765625,
      "learning_rate": 1.6798234112143257e-05,
      "loss": 0.3197,
      "step": 6962
    },
    {
      "epoch": 0.5269760181637576,
      "grad_norm": 0.796875,
      "learning_rate": 1.679736055460925e-05,
      "loss": 0.3604,
      "step": 6963
    },
    {
      "epoch": 0.527051700487205,
      "grad_norm": 0.7421875,
      "learning_rate": 1.6796486900643024e-05,
      "loss": 0.3219,
      "step": 6964
    },
    {
      "epoch": 0.5271273828106523,
      "grad_norm": 0.78125,
      "learning_rate": 1.6795613150256965e-05,
      "loss": 0.3308,
      "step": 6965
    },
    {
      "epoch": 0.5272030651340996,
      "grad_norm": 1.015625,
      "learning_rate": 1.679473930346347e-05,
      "loss": 0.3454,
      "step": 6966
    },
    {
      "epoch": 0.527278747457547,
      "grad_norm": 0.7421875,
      "learning_rate": 1.679386536027494e-05,
      "loss": 0.2977,
      "step": 6967
    },
    {
      "epoch": 0.5273544297809942,
      "grad_norm": 0.83984375,
      "learning_rate": 1.6792991320703772e-05,
      "loss": 0.3312,
      "step": 6968
    },
    {
      "epoch": 0.5274301121044416,
      "grad_norm": 1.21875,
      "learning_rate": 1.6792117184762363e-05,
      "loss": 0.3879,
      "step": 6969
    },
    {
      "epoch": 0.5275057944278889,
      "grad_norm": 0.734375,
      "learning_rate": 1.679124295246312e-05,
      "loss": 0.2854,
      "step": 6970
    },
    {
      "epoch": 0.5275814767513363,
      "grad_norm": 0.78125,
      "learning_rate": 1.6790368623818438e-05,
      "loss": 0.325,
      "step": 6971
    },
    {
      "epoch": 0.5276571590747836,
      "grad_norm": 0.77734375,
      "learning_rate": 1.6789494198840723e-05,
      "loss": 0.3202,
      "step": 6972
    },
    {
      "epoch": 0.527732841398231,
      "grad_norm": 0.7265625,
      "learning_rate": 1.6788619677542385e-05,
      "loss": 0.2869,
      "step": 6973
    },
    {
      "epoch": 0.5278085237216783,
      "grad_norm": 0.703125,
      "learning_rate": 1.6787745059935826e-05,
      "loss": 0.2707,
      "step": 6974
    },
    {
      "epoch": 0.5278842060451255,
      "grad_norm": 0.82421875,
      "learning_rate": 1.678687034603346e-05,
      "loss": 0.3409,
      "step": 6975
    },
    {
      "epoch": 0.5279598883685729,
      "grad_norm": 0.80078125,
      "learning_rate": 1.678599553584768e-05,
      "loss": 0.3727,
      "step": 6976
    },
    {
      "epoch": 0.5280355706920202,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6785120629390916e-05,
      "loss": 0.3307,
      "step": 6977
    },
    {
      "epoch": 0.5281112530154676,
      "grad_norm": 0.8125,
      "learning_rate": 1.6784245626675572e-05,
      "loss": 0.3176,
      "step": 6978
    },
    {
      "epoch": 0.5281869353389149,
      "grad_norm": 0.82421875,
      "learning_rate": 1.678337052771406e-05,
      "loss": 0.3625,
      "step": 6979
    },
    {
      "epoch": 0.5282626176623623,
      "grad_norm": 0.765625,
      "learning_rate": 1.67824953325188e-05,
      "loss": 0.3163,
      "step": 6980
    },
    {
      "epoch": 0.5283382999858096,
      "grad_norm": 0.76171875,
      "learning_rate": 1.67816200411022e-05,
      "loss": 0.2785,
      "step": 6981
    },
    {
      "epoch": 0.5284139823092568,
      "grad_norm": 0.76953125,
      "learning_rate": 1.6780744653476683e-05,
      "loss": 0.3131,
      "step": 6982
    },
    {
      "epoch": 0.5284896646327042,
      "grad_norm": 0.73046875,
      "learning_rate": 1.677986916965467e-05,
      "loss": 0.3045,
      "step": 6983
    },
    {
      "epoch": 0.5285653469561515,
      "grad_norm": 0.80078125,
      "learning_rate": 1.677899358964857e-05,
      "loss": 0.3649,
      "step": 6984
    },
    {
      "epoch": 0.5286410292795989,
      "grad_norm": 0.6953125,
      "learning_rate": 1.677811791347082e-05,
      "loss": 0.2837,
      "step": 6985
    },
    {
      "epoch": 0.5287167116030462,
      "grad_norm": 0.69140625,
      "learning_rate": 1.6777242141133837e-05,
      "loss": 0.2717,
      "step": 6986
    },
    {
      "epoch": 0.5287923939264936,
      "grad_norm": 0.765625,
      "learning_rate": 1.6776366272650037e-05,
      "loss": 0.3349,
      "step": 6987
    },
    {
      "epoch": 0.5288680762499409,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6775490308031857e-05,
      "loss": 0.3594,
      "step": 6988
    },
    {
      "epoch": 0.5289437585733882,
      "grad_norm": 0.80078125,
      "learning_rate": 1.677461424729172e-05,
      "loss": 0.3297,
      "step": 6989
    },
    {
      "epoch": 0.5290194408968355,
      "grad_norm": 0.75,
      "learning_rate": 1.6773738090442057e-05,
      "loss": 0.3213,
      "step": 6990
    },
    {
      "epoch": 0.5290951232202828,
      "grad_norm": 0.78125,
      "learning_rate": 1.6772861837495292e-05,
      "loss": 0.3409,
      "step": 6991
    },
    {
      "epoch": 0.5291708055437302,
      "grad_norm": 0.734375,
      "learning_rate": 1.677198548846386e-05,
      "loss": 0.3168,
      "step": 6992
    },
    {
      "epoch": 0.5292464878671775,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6771109043360193e-05,
      "loss": 0.3299,
      "step": 6993
    },
    {
      "epoch": 0.5293221701906249,
      "grad_norm": 0.8125,
      "learning_rate": 1.6770232502196722e-05,
      "loss": 0.3525,
      "step": 6994
    },
    {
      "epoch": 0.5293978525140722,
      "grad_norm": 0.75,
      "learning_rate": 1.6769355864985888e-05,
      "loss": 0.3195,
      "step": 6995
    },
    {
      "epoch": 0.5294735348375195,
      "grad_norm": 0.73046875,
      "learning_rate": 1.676847913174012e-05,
      "loss": 0.2938,
      "step": 6996
    },
    {
      "epoch": 0.5295492171609668,
      "grad_norm": 0.77734375,
      "learning_rate": 1.6767602302471865e-05,
      "loss": 0.2696,
      "step": 6997
    },
    {
      "epoch": 0.5296248994844142,
      "grad_norm": 0.6796875,
      "learning_rate": 1.6766725377193558e-05,
      "loss": 0.2648,
      "step": 6998
    },
    {
      "epoch": 0.5297005818078615,
      "grad_norm": 0.81640625,
      "learning_rate": 1.6765848355917636e-05,
      "loss": 0.3308,
      "step": 6999
    },
    {
      "epoch": 0.5297762641313088,
      "grad_norm": 0.73046875,
      "learning_rate": 1.6764971238656548e-05,
      "loss": 0.2939,
      "step": 7000
    },
    {
      "epoch": 0.5298519464547562,
      "grad_norm": 0.765625,
      "learning_rate": 1.676409402542273e-05,
      "loss": 0.3226,
      "step": 7001
    },
    {
      "epoch": 0.5299276287782035,
      "grad_norm": 0.765625,
      "learning_rate": 1.6763216716228634e-05,
      "loss": 0.3345,
      "step": 7002
    },
    {
      "epoch": 0.5300033111016508,
      "grad_norm": 0.82421875,
      "learning_rate": 1.67623393110867e-05,
      "loss": 0.3512,
      "step": 7003
    },
    {
      "epoch": 0.5300789934250981,
      "grad_norm": 0.7890625,
      "learning_rate": 1.676146181000938e-05,
      "loss": 0.3377,
      "step": 7004
    },
    {
      "epoch": 0.5301546757485455,
      "grad_norm": 0.77734375,
      "learning_rate": 1.6760584213009118e-05,
      "loss": 0.3365,
      "step": 7005
    },
    {
      "epoch": 0.5302303580719928,
      "grad_norm": 1.0703125,
      "learning_rate": 1.6759706520098373e-05,
      "loss": 0.3373,
      "step": 7006
    },
    {
      "epoch": 0.5303060403954402,
      "grad_norm": 0.71484375,
      "learning_rate": 1.6758828731289588e-05,
      "loss": 0.2955,
      "step": 7007
    },
    {
      "epoch": 0.5303817227188875,
      "grad_norm": 0.828125,
      "learning_rate": 1.675795084659522e-05,
      "loss": 0.3369,
      "step": 7008
    },
    {
      "epoch": 0.5304574050423349,
      "grad_norm": 0.796875,
      "learning_rate": 1.6757072866027718e-05,
      "loss": 0.2983,
      "step": 7009
    },
    {
      "epoch": 0.5305330873657821,
      "grad_norm": 0.77734375,
      "learning_rate": 1.6756194789599547e-05,
      "loss": 0.3483,
      "step": 7010
    },
    {
      "epoch": 0.5306087696892294,
      "grad_norm": 0.76953125,
      "learning_rate": 1.6755316617323157e-05,
      "loss": 0.3188,
      "step": 7011
    },
    {
      "epoch": 0.5306844520126768,
      "grad_norm": 0.8203125,
      "learning_rate": 1.6754438349211007e-05,
      "loss": 0.3124,
      "step": 7012
    },
    {
      "epoch": 0.5307601343361241,
      "grad_norm": 0.80078125,
      "learning_rate": 1.675355998527556e-05,
      "loss": 0.3438,
      "step": 7013
    },
    {
      "epoch": 0.5308358166595715,
      "grad_norm": 0.73046875,
      "learning_rate": 1.6752681525529274e-05,
      "loss": 0.2879,
      "step": 7014
    },
    {
      "epoch": 0.5309114989830188,
      "grad_norm": 0.703125,
      "learning_rate": 1.675180296998461e-05,
      "loss": 0.2896,
      "step": 7015
    },
    {
      "epoch": 0.5309871813064662,
      "grad_norm": 0.78125,
      "learning_rate": 1.6750924318654044e-05,
      "loss": 0.3537,
      "step": 7016
    },
    {
      "epoch": 0.5310628636299134,
      "grad_norm": 0.765625,
      "learning_rate": 1.6750045571550023e-05,
      "loss": 0.34,
      "step": 7017
    },
    {
      "epoch": 0.5311385459533607,
      "grad_norm": 0.8046875,
      "learning_rate": 1.6749166728685025e-05,
      "loss": 0.34,
      "step": 7018
    },
    {
      "epoch": 0.5312142282768081,
      "grad_norm": 1.171875,
      "learning_rate": 1.674828779007151e-05,
      "loss": 0.3593,
      "step": 7019
    },
    {
      "epoch": 0.5312899106002554,
      "grad_norm": 0.8125,
      "learning_rate": 1.6747408755721962e-05,
      "loss": 0.3519,
      "step": 7020
    },
    {
      "epoch": 0.5313655929237028,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6746529625648834e-05,
      "loss": 0.3229,
      "step": 7021
    },
    {
      "epoch": 0.5314412752471501,
      "grad_norm": 0.7265625,
      "learning_rate": 1.6745650399864608e-05,
      "loss": 0.3088,
      "step": 7022
    },
    {
      "epoch": 0.5315169575705975,
      "grad_norm": 0.75,
      "learning_rate": 1.6744771078381756e-05,
      "loss": 0.311,
      "step": 7023
    },
    {
      "epoch": 0.5315926398940447,
      "grad_norm": 0.82421875,
      "learning_rate": 1.674389166121275e-05,
      "loss": 0.3258,
      "step": 7024
    },
    {
      "epoch": 0.531668322217492,
      "grad_norm": 0.71484375,
      "learning_rate": 1.6743012148370067e-05,
      "loss": 0.2806,
      "step": 7025
    },
    {
      "epoch": 0.5317440045409394,
      "grad_norm": 0.7421875,
      "learning_rate": 1.674213253986619e-05,
      "loss": 0.3015,
      "step": 7026
    },
    {
      "epoch": 0.5318196868643867,
      "grad_norm": 0.80078125,
      "learning_rate": 1.6741252835713586e-05,
      "loss": 0.3197,
      "step": 7027
    },
    {
      "epoch": 0.5318953691878341,
      "grad_norm": 0.8046875,
      "learning_rate": 1.6740373035924747e-05,
      "loss": 0.3281,
      "step": 7028
    },
    {
      "epoch": 0.5319710515112814,
      "grad_norm": 0.859375,
      "learning_rate": 1.673949314051215e-05,
      "loss": 0.3562,
      "step": 7029
    },
    {
      "epoch": 0.5320467338347288,
      "grad_norm": 0.8046875,
      "learning_rate": 1.6738613149488276e-05,
      "loss": 0.3358,
      "step": 7030
    },
    {
      "epoch": 0.532122416158176,
      "grad_norm": 0.72265625,
      "learning_rate": 1.6737733062865607e-05,
      "loss": 0.3018,
      "step": 7031
    },
    {
      "epoch": 0.5321980984816234,
      "grad_norm": 1.0078125,
      "learning_rate": 1.673685288065664e-05,
      "loss": 0.3441,
      "step": 7032
    },
    {
      "epoch": 0.5322737808050707,
      "grad_norm": 0.81640625,
      "learning_rate": 1.6735972602873847e-05,
      "loss": 0.3687,
      "step": 7033
    },
    {
      "epoch": 0.532349463128518,
      "grad_norm": 0.77734375,
      "learning_rate": 1.6735092229529727e-05,
      "loss": 0.3475,
      "step": 7034
    },
    {
      "epoch": 0.5324251454519654,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6734211760636766e-05,
      "loss": 0.3563,
      "step": 7035
    },
    {
      "epoch": 0.5325008277754127,
      "grad_norm": 0.8203125,
      "learning_rate": 1.6733331196207453e-05,
      "loss": 0.319,
      "step": 7036
    },
    {
      "epoch": 0.5325765100988601,
      "grad_norm": 0.84765625,
      "learning_rate": 1.6732450536254285e-05,
      "loss": 0.3693,
      "step": 7037
    },
    {
      "epoch": 0.5326521924223073,
      "grad_norm": 1.171875,
      "learning_rate": 1.6731569780789752e-05,
      "loss": 0.3644,
      "step": 7038
    },
    {
      "epoch": 0.5327278747457547,
      "grad_norm": 1.0625,
      "learning_rate": 1.6730688929826348e-05,
      "loss": 0.3949,
      "step": 7039
    },
    {
      "epoch": 0.532803557069202,
      "grad_norm": 0.8046875,
      "learning_rate": 1.6729807983376574e-05,
      "loss": 0.3297,
      "step": 7040
    },
    {
      "epoch": 0.5328792393926494,
      "grad_norm": 0.73046875,
      "learning_rate": 1.6728926941452922e-05,
      "loss": 0.283,
      "step": 7041
    },
    {
      "epoch": 0.5329549217160967,
      "grad_norm": 0.6953125,
      "learning_rate": 1.6728045804067895e-05,
      "loss": 0.2868,
      "step": 7042
    },
    {
      "epoch": 0.533030604039544,
      "grad_norm": 0.81640625,
      "learning_rate": 1.672716457123399e-05,
      "loss": 0.3352,
      "step": 7043
    },
    {
      "epoch": 0.5331062863629913,
      "grad_norm": 0.80078125,
      "learning_rate": 1.6726283242963716e-05,
      "loss": 0.3255,
      "step": 7044
    },
    {
      "epoch": 0.5331819686864386,
      "grad_norm": 0.80859375,
      "learning_rate": 1.6725401819269572e-05,
      "loss": 0.3437,
      "step": 7045
    },
    {
      "epoch": 0.533257651009886,
      "grad_norm": 0.8125,
      "learning_rate": 1.672452030016406e-05,
      "loss": 0.333,
      "step": 7046
    },
    {
      "epoch": 0.5333333333333333,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6723638685659687e-05,
      "loss": 0.3444,
      "step": 7047
    },
    {
      "epoch": 0.5334090156567807,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6722756975768963e-05,
      "loss": 0.3026,
      "step": 7048
    },
    {
      "epoch": 0.533484697980228,
      "grad_norm": 0.74609375,
      "learning_rate": 1.672187517050439e-05,
      "loss": 0.298,
      "step": 7049
    },
    {
      "epoch": 0.5335603803036754,
      "grad_norm": 0.79296875,
      "learning_rate": 1.6720993269878486e-05,
      "loss": 0.3527,
      "step": 7050
    },
    {
      "epoch": 0.5336360626271226,
      "grad_norm": 0.7734375,
      "learning_rate": 1.672011127390376e-05,
      "loss": 0.3238,
      "step": 7051
    },
    {
      "epoch": 0.5337117449505699,
      "grad_norm": 0.83203125,
      "learning_rate": 1.671922918259272e-05,
      "loss": 0.3658,
      "step": 7052
    },
    {
      "epoch": 0.5337874272740173,
      "grad_norm": 0.76171875,
      "learning_rate": 1.6718346995957892e-05,
      "loss": 0.3056,
      "step": 7053
    },
    {
      "epoch": 0.5338631095974646,
      "grad_norm": 0.80078125,
      "learning_rate": 1.6717464714011775e-05,
      "loss": 0.3063,
      "step": 7054
    },
    {
      "epoch": 0.533938791920912,
      "grad_norm": 0.83203125,
      "learning_rate": 1.6716582336766892e-05,
      "loss": 0.3799,
      "step": 7055
    },
    {
      "epoch": 0.5340144742443593,
      "grad_norm": 0.8046875,
      "learning_rate": 1.6715699864235767e-05,
      "loss": 0.3325,
      "step": 7056
    },
    {
      "epoch": 0.5340901565678067,
      "grad_norm": 0.76171875,
      "learning_rate": 1.6714817296430914e-05,
      "loss": 0.3369,
      "step": 7057
    },
    {
      "epoch": 0.5341658388912539,
      "grad_norm": 0.765625,
      "learning_rate": 1.6713934633364853e-05,
      "loss": 0.3071,
      "step": 7058
    },
    {
      "epoch": 0.5342415212147013,
      "grad_norm": 1.125,
      "learning_rate": 1.671305187505011e-05,
      "loss": 0.3813,
      "step": 7059
    },
    {
      "epoch": 0.5343172035381486,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6712169021499205e-05,
      "loss": 0.2991,
      "step": 7060
    },
    {
      "epoch": 0.534392885861596,
      "grad_norm": 0.82421875,
      "learning_rate": 1.6711286072724662e-05,
      "loss": 0.3654,
      "step": 7061
    },
    {
      "epoch": 0.5344685681850433,
      "grad_norm": 2.46875,
      "learning_rate": 1.671040302873901e-05,
      "loss": 0.3332,
      "step": 7062
    },
    {
      "epoch": 0.5345442505084906,
      "grad_norm": 0.8046875,
      "learning_rate": 1.6709519889554776e-05,
      "loss": 0.3436,
      "step": 7063
    },
    {
      "epoch": 0.534619932831938,
      "grad_norm": 0.73828125,
      "learning_rate": 1.670863665518449e-05,
      "loss": 0.2754,
      "step": 7064
    },
    {
      "epoch": 0.5346956151553852,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6707753325640683e-05,
      "loss": 0.3423,
      "step": 7065
    },
    {
      "epoch": 0.5347712974788326,
      "grad_norm": 0.6953125,
      "learning_rate": 1.670686990093588e-05,
      "loss": 0.2748,
      "step": 7066
    },
    {
      "epoch": 0.5348469798022799,
      "grad_norm": 0.76953125,
      "learning_rate": 1.670598638108262e-05,
      "loss": 0.3059,
      "step": 7067
    },
    {
      "epoch": 0.5349226621257273,
      "grad_norm": 0.85546875,
      "learning_rate": 1.6705102766093432e-05,
      "loss": 0.325,
      "step": 7068
    },
    {
      "epoch": 0.5349983444491746,
      "grad_norm": 0.75,
      "learning_rate": 1.670421905598086e-05,
      "loss": 0.2858,
      "step": 7069
    },
    {
      "epoch": 0.535074026772622,
      "grad_norm": 0.953125,
      "learning_rate": 1.6703335250757428e-05,
      "loss": 0.3817,
      "step": 7070
    },
    {
      "epoch": 0.5351497090960693,
      "grad_norm": 1.2109375,
      "learning_rate": 1.6702451350435688e-05,
      "loss": 0.4103,
      "step": 7071
    },
    {
      "epoch": 0.5352253914195165,
      "grad_norm": 0.73046875,
      "learning_rate": 1.6701567355028174e-05,
      "loss": 0.2858,
      "step": 7072
    },
    {
      "epoch": 0.5353010737429639,
      "grad_norm": 0.8359375,
      "learning_rate": 1.6700683264547425e-05,
      "loss": 0.3167,
      "step": 7073
    },
    {
      "epoch": 0.5353767560664112,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6699799079005985e-05,
      "loss": 0.3196,
      "step": 7074
    },
    {
      "epoch": 0.5354524383898586,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6698914798416402e-05,
      "loss": 0.3289,
      "step": 7075
    },
    {
      "epoch": 0.5355281207133059,
      "grad_norm": 0.77734375,
      "learning_rate": 1.669803042279121e-05,
      "loss": 0.2881,
      "step": 7076
    },
    {
      "epoch": 0.5356038030367533,
      "grad_norm": 0.75390625,
      "learning_rate": 1.669714595214297e-05,
      "loss": 0.308,
      "step": 7077
    },
    {
      "epoch": 0.5356794853602006,
      "grad_norm": 0.79296875,
      "learning_rate": 1.6696261386484215e-05,
      "loss": 0.3565,
      "step": 7078
    },
    {
      "epoch": 0.5357551676836478,
      "grad_norm": 0.71875,
      "learning_rate": 1.6695376725827503e-05,
      "loss": 0.2933,
      "step": 7079
    },
    {
      "epoch": 0.5358308500070952,
      "grad_norm": 0.734375,
      "learning_rate": 1.669449197018538e-05,
      "loss": 0.2813,
      "step": 7080
    },
    {
      "epoch": 0.5359065323305425,
      "grad_norm": 0.71875,
      "learning_rate": 1.6693607119570408e-05,
      "loss": 0.272,
      "step": 7081
    },
    {
      "epoch": 0.5359822146539899,
      "grad_norm": 0.83203125,
      "learning_rate": 1.6692722173995127e-05,
      "loss": 0.3746,
      "step": 7082
    },
    {
      "epoch": 0.5360578969774372,
      "grad_norm": 0.80859375,
      "learning_rate": 1.6691837133472097e-05,
      "loss": 0.3481,
      "step": 7083
    },
    {
      "epoch": 0.5361335793008846,
      "grad_norm": 0.81640625,
      "learning_rate": 1.6690951998013876e-05,
      "loss": 0.3579,
      "step": 7084
    },
    {
      "epoch": 0.5362092616243319,
      "grad_norm": 0.76953125,
      "learning_rate": 1.6690066767633016e-05,
      "loss": 0.2966,
      "step": 7085
    },
    {
      "epoch": 0.5362849439477791,
      "grad_norm": 1.0546875,
      "learning_rate": 1.668918144234208e-05,
      "loss": 0.339,
      "step": 7086
    },
    {
      "epoch": 0.5363606262712265,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6688296022153628e-05,
      "loss": 0.3149,
      "step": 7087
    },
    {
      "epoch": 0.5364363085946738,
      "grad_norm": 0.84765625,
      "learning_rate": 1.6687410507080214e-05,
      "loss": 0.3453,
      "step": 7088
    },
    {
      "epoch": 0.5365119909181212,
      "grad_norm": 0.7734375,
      "learning_rate": 1.668652489713441e-05,
      "loss": 0.3252,
      "step": 7089
    },
    {
      "epoch": 0.5365876732415685,
      "grad_norm": 0.78125,
      "learning_rate": 1.6685639192328777e-05,
      "loss": 0.2966,
      "step": 7090
    },
    {
      "epoch": 0.5366633555650159,
      "grad_norm": 0.796875,
      "learning_rate": 1.6684753392675873e-05,
      "loss": 0.3425,
      "step": 7091
    },
    {
      "epoch": 0.5367390378884632,
      "grad_norm": 0.80078125,
      "learning_rate": 1.6683867498188277e-05,
      "loss": 0.3176,
      "step": 7092
    },
    {
      "epoch": 0.5368147202119105,
      "grad_norm": 0.765625,
      "learning_rate": 1.6682981508878547e-05,
      "loss": 0.3206,
      "step": 7093
    },
    {
      "epoch": 0.5368904025353578,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6682095424759258e-05,
      "loss": 0.3164,
      "step": 7094
    },
    {
      "epoch": 0.5369660848588051,
      "grad_norm": 0.7421875,
      "learning_rate": 1.6681209245842978e-05,
      "loss": 0.3142,
      "step": 7095
    },
    {
      "epoch": 0.5370417671822525,
      "grad_norm": 0.7578125,
      "learning_rate": 1.668032297214228e-05,
      "loss": 0.2992,
      "step": 7096
    },
    {
      "epoch": 0.5371174495056998,
      "grad_norm": 0.78125,
      "learning_rate": 1.6679436603669736e-05,
      "loss": 0.2998,
      "step": 7097
    },
    {
      "epoch": 0.5371931318291472,
      "grad_norm": 0.73828125,
      "learning_rate": 1.667855014043792e-05,
      "loss": 0.2959,
      "step": 7098
    },
    {
      "epoch": 0.5372688141525945,
      "grad_norm": 0.87109375,
      "learning_rate": 1.6677663582459414e-05,
      "loss": 0.3348,
      "step": 7099
    },
    {
      "epoch": 0.5373444964760418,
      "grad_norm": 0.75390625,
      "learning_rate": 1.667677692974679e-05,
      "loss": 0.3007,
      "step": 7100
    },
    {
      "epoch": 0.5374201787994891,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6675890182312624e-05,
      "loss": 0.2882,
      "step": 7101
    },
    {
      "epoch": 0.5374958611229365,
      "grad_norm": 0.76171875,
      "learning_rate": 1.66750033401695e-05,
      "loss": 0.3207,
      "step": 7102
    },
    {
      "epoch": 0.5375715434463838,
      "grad_norm": 0.828125,
      "learning_rate": 1.6674116403330002e-05,
      "loss": 0.3656,
      "step": 7103
    },
    {
      "epoch": 0.5376472257698311,
      "grad_norm": 0.796875,
      "learning_rate": 1.6673229371806706e-05,
      "loss": 0.3546,
      "step": 7104
    },
    {
      "epoch": 0.5377229080932785,
      "grad_norm": 0.73828125,
      "learning_rate": 1.66723422456122e-05,
      "loss": 0.3263,
      "step": 7105
    },
    {
      "epoch": 0.5377985904167258,
      "grad_norm": 0.7734375,
      "learning_rate": 1.667145502475907e-05,
      "loss": 0.3162,
      "step": 7106
    },
    {
      "epoch": 0.5378742727401731,
      "grad_norm": 0.75,
      "learning_rate": 1.6670567709259903e-05,
      "loss": 0.3193,
      "step": 7107
    },
    {
      "epoch": 0.5379499550636204,
      "grad_norm": 0.86328125,
      "learning_rate": 1.6669680299127286e-05,
      "loss": 0.3855,
      "step": 7108
    },
    {
      "epoch": 0.5380256373870678,
      "grad_norm": 0.76953125,
      "learning_rate": 1.6668792794373807e-05,
      "loss": 0.3297,
      "step": 7109
    },
    {
      "epoch": 0.5381013197105151,
      "grad_norm": 0.78125,
      "learning_rate": 1.666790519501206e-05,
      "loss": 0.3666,
      "step": 7110
    },
    {
      "epoch": 0.5381770020339625,
      "grad_norm": 0.79296875,
      "learning_rate": 1.6667017501054633e-05,
      "loss": 0.3605,
      "step": 7111
    },
    {
      "epoch": 0.5382526843574098,
      "grad_norm": 0.84765625,
      "learning_rate": 1.6666129712514123e-05,
      "loss": 0.3765,
      "step": 7112
    },
    {
      "epoch": 0.5383283666808572,
      "grad_norm": 0.734375,
      "learning_rate": 1.6665241829403122e-05,
      "loss": 0.3076,
      "step": 7113
    },
    {
      "epoch": 0.5384040490043044,
      "grad_norm": 0.8203125,
      "learning_rate": 1.6664353851734228e-05,
      "loss": 0.3632,
      "step": 7114
    },
    {
      "epoch": 0.5384797313277517,
      "grad_norm": 0.79296875,
      "learning_rate": 1.6663465779520042e-05,
      "loss": 0.3488,
      "step": 7115
    },
    {
      "epoch": 0.5385554136511991,
      "grad_norm": 0.765625,
      "learning_rate": 1.6662577612773154e-05,
      "loss": 0.3246,
      "step": 7116
    },
    {
      "epoch": 0.5386310959746464,
      "grad_norm": 0.8203125,
      "learning_rate": 1.6661689351506174e-05,
      "loss": 0.3593,
      "step": 7117
    },
    {
      "epoch": 0.5387067782980938,
      "grad_norm": 0.82421875,
      "learning_rate": 1.6660800995731693e-05,
      "loss": 0.3326,
      "step": 7118
    },
    {
      "epoch": 0.5387824606215411,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6659912545462325e-05,
      "loss": 0.3365,
      "step": 7119
    },
    {
      "epoch": 0.5388581429449885,
      "grad_norm": 0.71484375,
      "learning_rate": 1.6659024000710667e-05,
      "loss": 0.2896,
      "step": 7120
    },
    {
      "epoch": 0.5389338252684357,
      "grad_norm": 0.76953125,
      "learning_rate": 1.6658135361489326e-05,
      "loss": 0.346,
      "step": 7121
    },
    {
      "epoch": 0.539009507591883,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6657246627810906e-05,
      "loss": 0.3211,
      "step": 7122
    },
    {
      "epoch": 0.5390851899153304,
      "grad_norm": 0.70703125,
      "learning_rate": 1.6656357799688025e-05,
      "loss": 0.2724,
      "step": 7123
    },
    {
      "epoch": 0.5391608722387777,
      "grad_norm": 0.78125,
      "learning_rate": 1.6655468877133282e-05,
      "loss": 0.3377,
      "step": 7124
    },
    {
      "epoch": 0.5392365545622251,
      "grad_norm": 0.71484375,
      "learning_rate": 1.665457986015929e-05,
      "loss": 0.2872,
      "step": 7125
    },
    {
      "epoch": 0.5393122368856724,
      "grad_norm": 0.81640625,
      "learning_rate": 1.6653690748778662e-05,
      "loss": 0.3712,
      "step": 7126
    },
    {
      "epoch": 0.5393879192091198,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6652801543004017e-05,
      "loss": 0.3314,
      "step": 7127
    },
    {
      "epoch": 0.539463601532567,
      "grad_norm": 0.79296875,
      "learning_rate": 1.6651912242847966e-05,
      "loss": 0.3539,
      "step": 7128
    },
    {
      "epoch": 0.5395392838560144,
      "grad_norm": 0.8125,
      "learning_rate": 1.665102284832312e-05,
      "loss": 0.3368,
      "step": 7129
    },
    {
      "epoch": 0.5396149661794617,
      "grad_norm": 4.03125,
      "learning_rate": 1.6650133359442106e-05,
      "loss": 0.3956,
      "step": 7130
    },
    {
      "epoch": 0.539690648502909,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6649243776217538e-05,
      "loss": 0.3221,
      "step": 7131
    },
    {
      "epoch": 0.5397663308263564,
      "grad_norm": 0.80859375,
      "learning_rate": 1.6648354098662034e-05,
      "loss": 0.3564,
      "step": 7132
    },
    {
      "epoch": 0.5398420131498037,
      "grad_norm": 0.73828125,
      "learning_rate": 1.664746432678822e-05,
      "loss": 0.3329,
      "step": 7133
    },
    {
      "epoch": 0.5399176954732511,
      "grad_norm": 0.75,
      "learning_rate": 1.6646574460608717e-05,
      "loss": 0.2971,
      "step": 7134
    },
    {
      "epoch": 0.5399933777966983,
      "grad_norm": 0.80078125,
      "learning_rate": 1.6645684500136145e-05,
      "loss": 0.3469,
      "step": 7135
    },
    {
      "epoch": 0.5400690601201457,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6644794445383138e-05,
      "loss": 0.326,
      "step": 7136
    },
    {
      "epoch": 0.540144742443593,
      "grad_norm": 0.78515625,
      "learning_rate": 1.664390429636232e-05,
      "loss": 0.2968,
      "step": 7137
    },
    {
      "epoch": 0.5402204247670404,
      "grad_norm": 0.77734375,
      "learning_rate": 1.6643014053086318e-05,
      "loss": 0.3616,
      "step": 7138
    },
    {
      "epoch": 0.5402961070904877,
      "grad_norm": 0.77734375,
      "learning_rate": 1.664212371556776e-05,
      "loss": 0.354,
      "step": 7139
    },
    {
      "epoch": 0.540371789413935,
      "grad_norm": 0.76953125,
      "learning_rate": 1.664123328381928e-05,
      "loss": 0.3483,
      "step": 7140
    },
    {
      "epoch": 0.5404474717373824,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6640342757853506e-05,
      "loss": 0.316,
      "step": 7141
    },
    {
      "epoch": 0.5405231540608296,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6639452137683082e-05,
      "loss": 0.2964,
      "step": 7142
    },
    {
      "epoch": 0.540598836384277,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6638561423320626e-05,
      "loss": 0.339,
      "step": 7143
    },
    {
      "epoch": 0.5406745187077243,
      "grad_norm": 0.9375,
      "learning_rate": 1.6637670614778793e-05,
      "loss": 0.2547,
      "step": 7144
    },
    {
      "epoch": 0.5407502010311717,
      "grad_norm": 0.72265625,
      "learning_rate": 1.6636779712070206e-05,
      "loss": 0.2769,
      "step": 7145
    },
    {
      "epoch": 0.540825883354619,
      "grad_norm": 0.88671875,
      "learning_rate": 1.663588871520751e-05,
      "loss": 0.3425,
      "step": 7146
    },
    {
      "epoch": 0.5409015656780664,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6634997624203347e-05,
      "loss": 0.3429,
      "step": 7147
    },
    {
      "epoch": 0.5409772480015137,
      "grad_norm": 0.8046875,
      "learning_rate": 1.6634106439070357e-05,
      "loss": 0.379,
      "step": 7148
    },
    {
      "epoch": 0.5410529303249609,
      "grad_norm": 0.78515625,
      "learning_rate": 1.663321515982118e-05,
      "loss": 0.3227,
      "step": 7149
    },
    {
      "epoch": 0.5411286126484083,
      "grad_norm": 0.73828125,
      "learning_rate": 1.6632323786468468e-05,
      "loss": 0.309,
      "step": 7150
    },
    {
      "epoch": 0.5412042949718556,
      "grad_norm": 0.82421875,
      "learning_rate": 1.6631432319024855e-05,
      "loss": 0.306,
      "step": 7151
    },
    {
      "epoch": 0.541279977295303,
      "grad_norm": 0.8125,
      "learning_rate": 1.6630540757503e-05,
      "loss": 0.3486,
      "step": 7152
    },
    {
      "epoch": 0.5413556596187503,
      "grad_norm": 0.81640625,
      "learning_rate": 1.6629649101915544e-05,
      "loss": 0.3651,
      "step": 7153
    },
    {
      "epoch": 0.5414313419421977,
      "grad_norm": 1.1640625,
      "learning_rate": 1.662875735227514e-05,
      "loss": 0.3591,
      "step": 7154
    },
    {
      "epoch": 0.541507024265645,
      "grad_norm": 0.74609375,
      "learning_rate": 1.6627865508594433e-05,
      "loss": 0.2609,
      "step": 7155
    },
    {
      "epoch": 0.5415827065890922,
      "grad_norm": 0.72265625,
      "learning_rate": 1.6626973570886086e-05,
      "loss": 0.2712,
      "step": 7156
    },
    {
      "epoch": 0.5416583889125396,
      "grad_norm": 0.82421875,
      "learning_rate": 1.662608153916274e-05,
      "loss": 0.3372,
      "step": 7157
    },
    {
      "epoch": 0.5417340712359869,
      "grad_norm": 0.7109375,
      "learning_rate": 1.662518941343706e-05,
      "loss": 0.3032,
      "step": 7158
    },
    {
      "epoch": 0.5418097535594343,
      "grad_norm": 1.1640625,
      "learning_rate": 1.66242971937217e-05,
      "loss": 0.3611,
      "step": 7159
    },
    {
      "epoch": 0.5418854358828816,
      "grad_norm": 0.7421875,
      "learning_rate": 1.6623404880029317e-05,
      "loss": 0.2772,
      "step": 7160
    },
    {
      "epoch": 0.541961118206329,
      "grad_norm": 0.8046875,
      "learning_rate": 1.662251247237257e-05,
      "loss": 0.2896,
      "step": 7161
    },
    {
      "epoch": 0.5420368005297763,
      "grad_norm": 0.796875,
      "learning_rate": 1.6621619970764117e-05,
      "loss": 0.3512,
      "step": 7162
    },
    {
      "epoch": 0.5421124828532236,
      "grad_norm": 0.859375,
      "learning_rate": 1.662072737521662e-05,
      "loss": 0.426,
      "step": 7163
    },
    {
      "epoch": 0.5421881651766709,
      "grad_norm": 0.828125,
      "learning_rate": 1.6619834685742745e-05,
      "loss": 0.3682,
      "step": 7164
    },
    {
      "epoch": 0.5422638475001182,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6618941902355156e-05,
      "loss": 0.3321,
      "step": 7165
    },
    {
      "epoch": 0.5423395298235656,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6618049025066517e-05,
      "loss": 0.3121,
      "step": 7166
    },
    {
      "epoch": 0.5424152121470129,
      "grad_norm": 0.734375,
      "learning_rate": 1.6617156053889497e-05,
      "loss": 0.3071,
      "step": 7167
    },
    {
      "epoch": 0.5424908944704603,
      "grad_norm": 1.0078125,
      "learning_rate": 1.661626298883676e-05,
      "loss": 0.3578,
      "step": 7168
    },
    {
      "epoch": 0.5425665767939075,
      "grad_norm": 0.74609375,
      "learning_rate": 1.6615369829920983e-05,
      "loss": 0.3267,
      "step": 7169
    },
    {
      "epoch": 0.5426422591173549,
      "grad_norm": 0.75390625,
      "learning_rate": 1.661447657715483e-05,
      "loss": 0.3139,
      "step": 7170
    },
    {
      "epoch": 0.5427179414408022,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6613583230550977e-05,
      "loss": 0.3452,
      "step": 7171
    },
    {
      "epoch": 0.5427936237642496,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6612689790122095e-05,
      "loss": 0.3256,
      "step": 7172
    },
    {
      "epoch": 0.5428693060876969,
      "grad_norm": 0.8046875,
      "learning_rate": 1.6611796255880862e-05,
      "loss": 0.3636,
      "step": 7173
    },
    {
      "epoch": 0.5429449884111442,
      "grad_norm": 0.73046875,
      "learning_rate": 1.6610902627839953e-05,
      "loss": 0.2928,
      "step": 7174
    },
    {
      "epoch": 0.5430206707345916,
      "grad_norm": 0.7734375,
      "learning_rate": 1.661000890601204e-05,
      "loss": 0.3187,
      "step": 7175
    },
    {
      "epoch": 0.5430963530580388,
      "grad_norm": 0.74609375,
      "learning_rate": 1.6609115090409814e-05,
      "loss": 0.3098,
      "step": 7176
    },
    {
      "epoch": 0.5431720353814862,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6608221181045946e-05,
      "loss": 0.325,
      "step": 7177
    },
    {
      "epoch": 0.5432477177049335,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6607327177933125e-05,
      "loss": 0.3346,
      "step": 7178
    },
    {
      "epoch": 0.5433234000283809,
      "grad_norm": 0.734375,
      "learning_rate": 1.6606433081084022e-05,
      "loss": 0.3065,
      "step": 7179
    },
    {
      "epoch": 0.5433990823518282,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6605538890511334e-05,
      "loss": 0.3181,
      "step": 7180
    },
    {
      "epoch": 0.5434747646752756,
      "grad_norm": 0.78515625,
      "learning_rate": 1.660464460622774e-05,
      "loss": 0.3258,
      "step": 7181
    },
    {
      "epoch": 0.5435504469987229,
      "grad_norm": 0.79296875,
      "learning_rate": 1.6603750228245928e-05,
      "loss": 0.2732,
      "step": 7182
    },
    {
      "epoch": 0.5436261293221701,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6602855756578587e-05,
      "loss": 0.3464,
      "step": 7183
    },
    {
      "epoch": 0.5437018116456175,
      "grad_norm": 0.76171875,
      "learning_rate": 1.6601961191238408e-05,
      "loss": 0.32,
      "step": 7184
    },
    {
      "epoch": 0.5437774939690648,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6601066532238075e-05,
      "loss": 0.3179,
      "step": 7185
    },
    {
      "epoch": 0.5438531762925122,
      "grad_norm": 0.765625,
      "learning_rate": 1.6600171779590288e-05,
      "loss": 0.3134,
      "step": 7186
    },
    {
      "epoch": 0.5439288586159595,
      "grad_norm": 0.75390625,
      "learning_rate": 1.659927693330774e-05,
      "loss": 0.302,
      "step": 7187
    },
    {
      "epoch": 0.5440045409394069,
      "grad_norm": 0.85546875,
      "learning_rate": 1.6598381993403122e-05,
      "loss": 0.3561,
      "step": 7188
    },
    {
      "epoch": 0.5440802232628542,
      "grad_norm": 0.69140625,
      "learning_rate": 1.6597486959889134e-05,
      "loss": 0.2862,
      "step": 7189
    },
    {
      "epoch": 0.5441559055863014,
      "grad_norm": 0.76953125,
      "learning_rate": 1.659659183277847e-05,
      "loss": 0.3547,
      "step": 7190
    },
    {
      "epoch": 0.5442315879097488,
      "grad_norm": 0.7265625,
      "learning_rate": 1.6595696612083833e-05,
      "loss": 0.2708,
      "step": 7191
    },
    {
      "epoch": 0.5443072702331961,
      "grad_norm": 0.7265625,
      "learning_rate": 1.6594801297817914e-05,
      "loss": 0.2844,
      "step": 7192
    },
    {
      "epoch": 0.5443829525566435,
      "grad_norm": 0.734375,
      "learning_rate": 1.659390588999343e-05,
      "loss": 0.2835,
      "step": 7193
    },
    {
      "epoch": 0.5444586348800908,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6593010388623073e-05,
      "loss": 0.3292,
      "step": 7194
    },
    {
      "epoch": 0.5445343172035382,
      "grad_norm": 0.76953125,
      "learning_rate": 1.6592114793719547e-05,
      "loss": 0.3255,
      "step": 7195
    },
    {
      "epoch": 0.5446099995269855,
      "grad_norm": 0.7890625,
      "learning_rate": 1.659121910529556e-05,
      "loss": 0.3509,
      "step": 7196
    },
    {
      "epoch": 0.5446856818504328,
      "grad_norm": 0.76953125,
      "learning_rate": 1.6590323323363825e-05,
      "loss": 0.3228,
      "step": 7197
    },
    {
      "epoch": 0.5447613641738801,
      "grad_norm": 0.7421875,
      "learning_rate": 1.6589427447937038e-05,
      "loss": 0.2977,
      "step": 7198
    },
    {
      "epoch": 0.5448370464973274,
      "grad_norm": 0.74609375,
      "learning_rate": 1.6588531479027917e-05,
      "loss": 0.3177,
      "step": 7199
    },
    {
      "epoch": 0.5449127288207748,
      "grad_norm": 0.7890625,
      "learning_rate": 1.658763541664917e-05,
      "loss": 0.3162,
      "step": 7200
    },
    {
      "epoch": 0.5449884111442221,
      "grad_norm": 0.76171875,
      "learning_rate": 1.658673926081351e-05,
      "loss": 0.303,
      "step": 7201
    },
    {
      "epoch": 0.5450640934676695,
      "grad_norm": 0.7421875,
      "learning_rate": 1.6585843011533652e-05,
      "loss": 0.3106,
      "step": 7202
    },
    {
      "epoch": 0.5451397757911168,
      "grad_norm": 0.71484375,
      "learning_rate": 1.6584946668822308e-05,
      "loss": 0.2625,
      "step": 7203
    },
    {
      "epoch": 0.5452154581145641,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6584050232692198e-05,
      "loss": 0.3076,
      "step": 7204
    },
    {
      "epoch": 0.5452911404380114,
      "grad_norm": 0.7265625,
      "learning_rate": 1.658315370315603e-05,
      "loss": 0.3047,
      "step": 7205
    },
    {
      "epoch": 0.5453668227614588,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6582257080226537e-05,
      "loss": 0.2993,
      "step": 7206
    },
    {
      "epoch": 0.5454425050849061,
      "grad_norm": 0.78515625,
      "learning_rate": 1.658136036391643e-05,
      "loss": 0.3386,
      "step": 7207
    },
    {
      "epoch": 0.5455181874083534,
      "grad_norm": 0.70703125,
      "learning_rate": 1.658046355423843e-05,
      "loss": 0.2995,
      "step": 7208
    },
    {
      "epoch": 0.5455938697318008,
      "grad_norm": 0.76953125,
      "learning_rate": 1.6579566651205262e-05,
      "loss": 0.3136,
      "step": 7209
    },
    {
      "epoch": 0.5456695520552481,
      "grad_norm": 0.7734375,
      "learning_rate": 1.657866965482965e-05,
      "loss": 0.306,
      "step": 7210
    },
    {
      "epoch": 0.5457452343786954,
      "grad_norm": 0.8046875,
      "learning_rate": 1.6577772565124322e-05,
      "loss": 0.3404,
      "step": 7211
    },
    {
      "epoch": 0.5458209167021427,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6576875382102e-05,
      "loss": 0.3283,
      "step": 7212
    },
    {
      "epoch": 0.5458965990255901,
      "grad_norm": 0.7421875,
      "learning_rate": 1.6575978105775416e-05,
      "loss": 0.3099,
      "step": 7213
    },
    {
      "epoch": 0.5459722813490374,
      "grad_norm": 0.76953125,
      "learning_rate": 1.6575080736157295e-05,
      "loss": 0.3115,
      "step": 7214
    },
    {
      "epoch": 0.5460479636724848,
      "grad_norm": 0.72265625,
      "learning_rate": 1.6574183273260372e-05,
      "loss": 0.3043,
      "step": 7215
    },
    {
      "epoch": 0.5461236459959321,
      "grad_norm": 0.73046875,
      "learning_rate": 1.657328571709738e-05,
      "loss": 0.3311,
      "step": 7216
    },
    {
      "epoch": 0.5461993283193795,
      "grad_norm": 0.74609375,
      "learning_rate": 1.6572388067681047e-05,
      "loss": 0.3359,
      "step": 7217
    },
    {
      "epoch": 0.5462750106428267,
      "grad_norm": 0.73046875,
      "learning_rate": 1.6571490325024114e-05,
      "loss": 0.3069,
      "step": 7218
    },
    {
      "epoch": 0.546350692966274,
      "grad_norm": 0.80078125,
      "learning_rate": 1.6570592489139313e-05,
      "loss": 0.3505,
      "step": 7219
    },
    {
      "epoch": 0.5464263752897214,
      "grad_norm": 0.76953125,
      "learning_rate": 1.656969456003938e-05,
      "loss": 0.3229,
      "step": 7220
    },
    {
      "epoch": 0.5465020576131687,
      "grad_norm": 0.74609375,
      "learning_rate": 1.6568796537737056e-05,
      "loss": 0.293,
      "step": 7221
    },
    {
      "epoch": 0.5465777399366161,
      "grad_norm": 0.703125,
      "learning_rate": 1.6567898422245083e-05,
      "loss": 0.2669,
      "step": 7222
    },
    {
      "epoch": 0.5466534222600634,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6567000213576196e-05,
      "loss": 0.3364,
      "step": 7223
    },
    {
      "epoch": 0.5467291045835108,
      "grad_norm": 0.78515625,
      "learning_rate": 1.656610191174314e-05,
      "loss": 0.3231,
      "step": 7224
    },
    {
      "epoch": 0.546804786906958,
      "grad_norm": 0.77734375,
      "learning_rate": 1.6565203516758667e-05,
      "loss": 0.351,
      "step": 7225
    },
    {
      "epoch": 0.5468804692304053,
      "grad_norm": 0.7734375,
      "learning_rate": 1.656430502863551e-05,
      "loss": 0.3082,
      "step": 7226
    },
    {
      "epoch": 0.5469561515538527,
      "grad_norm": 0.72265625,
      "learning_rate": 1.656340644738643e-05,
      "loss": 0.2897,
      "step": 7227
    },
    {
      "epoch": 0.5470318338773,
      "grad_norm": 0.78515625,
      "learning_rate": 1.656250777302416e-05,
      "loss": 0.3418,
      "step": 7228
    },
    {
      "epoch": 0.5471075162007474,
      "grad_norm": 0.7734375,
      "learning_rate": 1.656160900556145e-05,
      "loss": 0.3535,
      "step": 7229
    },
    {
      "epoch": 0.5471831985241947,
      "grad_norm": 0.7109375,
      "learning_rate": 1.6560710145011064e-05,
      "loss": 0.2943,
      "step": 7230
    },
    {
      "epoch": 0.5472588808476421,
      "grad_norm": 0.9609375,
      "learning_rate": 1.6559811191385746e-05,
      "loss": 0.385,
      "step": 7231
    },
    {
      "epoch": 0.5473345631710893,
      "grad_norm": 0.73828125,
      "learning_rate": 1.6558912144698247e-05,
      "loss": 0.3355,
      "step": 7232
    },
    {
      "epoch": 0.5474102454945367,
      "grad_norm": 0.81640625,
      "learning_rate": 1.6558013004961325e-05,
      "loss": 0.3659,
      "step": 7233
    },
    {
      "epoch": 0.547485927817984,
      "grad_norm": 0.78125,
      "learning_rate": 1.6557113772187734e-05,
      "loss": 0.364,
      "step": 7234
    },
    {
      "epoch": 0.5475616101414313,
      "grad_norm": 0.8046875,
      "learning_rate": 1.6556214446390233e-05,
      "loss": 0.3341,
      "step": 7235
    },
    {
      "epoch": 0.5476372924648787,
      "grad_norm": 0.734375,
      "learning_rate": 1.6555315027581576e-05,
      "loss": 0.3008,
      "step": 7236
    },
    {
      "epoch": 0.547712974788326,
      "grad_norm": 0.83203125,
      "learning_rate": 1.655441551577453e-05,
      "loss": 0.3464,
      "step": 7237
    },
    {
      "epoch": 0.5477886571117734,
      "grad_norm": 0.7890625,
      "learning_rate": 1.655351591098185e-05,
      "loss": 0.3363,
      "step": 7238
    },
    {
      "epoch": 0.5478643394352206,
      "grad_norm": 0.7109375,
      "learning_rate": 1.65526162132163e-05,
      "loss": 0.2932,
      "step": 7239
    },
    {
      "epoch": 0.547940021758668,
      "grad_norm": 0.69921875,
      "learning_rate": 1.6551716422490645e-05,
      "loss": 0.2834,
      "step": 7240
    },
    {
      "epoch": 0.5480157040821153,
      "grad_norm": 0.8828125,
      "learning_rate": 1.655081653881765e-05,
      "loss": 0.3428,
      "step": 7241
    },
    {
      "epoch": 0.5480913864055627,
      "grad_norm": 0.703125,
      "learning_rate": 1.654991656221008e-05,
      "loss": 0.2877,
      "step": 7242
    },
    {
      "epoch": 0.54816706872901,
      "grad_norm": 0.8828125,
      "learning_rate": 1.6549016492680702e-05,
      "loss": 0.266,
      "step": 7243
    },
    {
      "epoch": 0.5482427510524573,
      "grad_norm": 0.8046875,
      "learning_rate": 1.654811633024229e-05,
      "loss": 0.2952,
      "step": 7244
    },
    {
      "epoch": 0.5483184333759047,
      "grad_norm": 0.76953125,
      "learning_rate": 1.654721607490761e-05,
      "loss": 0.3422,
      "step": 7245
    },
    {
      "epoch": 0.5483941156993519,
      "grad_norm": 0.72265625,
      "learning_rate": 1.6546315726689432e-05,
      "loss": 0.301,
      "step": 7246
    },
    {
      "epoch": 0.5484697980227993,
      "grad_norm": 0.79296875,
      "learning_rate": 1.6545415285600533e-05,
      "loss": 0.3189,
      "step": 7247
    },
    {
      "epoch": 0.5485454803462466,
      "grad_norm": 0.734375,
      "learning_rate": 1.654451475165369e-05,
      "loss": 0.2912,
      "step": 7248
    },
    {
      "epoch": 0.548621162669694,
      "grad_norm": 0.78125,
      "learning_rate": 1.654361412486167e-05,
      "loss": 0.3653,
      "step": 7249
    },
    {
      "epoch": 0.5486968449931413,
      "grad_norm": 0.80078125,
      "learning_rate": 1.6542713405237254e-05,
      "loss": 0.3627,
      "step": 7250
    },
    {
      "epoch": 0.5487725273165887,
      "grad_norm": 0.76953125,
      "learning_rate": 1.654181259279322e-05,
      "loss": 0.3247,
      "step": 7251
    },
    {
      "epoch": 0.548848209640036,
      "grad_norm": 0.78125,
      "learning_rate": 1.6540911687542352e-05,
      "loss": 0.3549,
      "step": 7252
    },
    {
      "epoch": 0.5489238919634832,
      "grad_norm": 1.5625,
      "learning_rate": 1.6540010689497424e-05,
      "loss": 0.3284,
      "step": 7253
    },
    {
      "epoch": 0.5489995742869306,
      "grad_norm": 0.8515625,
      "learning_rate": 1.6539109598671225e-05,
      "loss": 0.3032,
      "step": 7254
    },
    {
      "epoch": 0.5490752566103779,
      "grad_norm": 0.8046875,
      "learning_rate": 1.653820841507653e-05,
      "loss": 0.3544,
      "step": 7255
    },
    {
      "epoch": 0.5491509389338253,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6537307138726132e-05,
      "loss": 0.3504,
      "step": 7256
    },
    {
      "epoch": 0.5492266212572726,
      "grad_norm": 0.76171875,
      "learning_rate": 1.653640576963281e-05,
      "loss": 0.336,
      "step": 7257
    },
    {
      "epoch": 0.54930230358072,
      "grad_norm": 0.765625,
      "learning_rate": 1.653550430780936e-05,
      "loss": 0.3063,
      "step": 7258
    },
    {
      "epoch": 0.5493779859041673,
      "grad_norm": 0.7421875,
      "learning_rate": 1.6534602753268566e-05,
      "loss": 0.298,
      "step": 7259
    },
    {
      "epoch": 0.5494536682276145,
      "grad_norm": 0.80078125,
      "learning_rate": 1.6533701106023216e-05,
      "loss": 0.3502,
      "step": 7260
    },
    {
      "epoch": 0.5495293505510619,
      "grad_norm": 0.77734375,
      "learning_rate": 1.6532799366086104e-05,
      "loss": 0.306,
      "step": 7261
    },
    {
      "epoch": 0.5496050328745092,
      "grad_norm": 0.7734375,
      "learning_rate": 1.653189753347002e-05,
      "loss": 0.3068,
      "step": 7262
    },
    {
      "epoch": 0.5496807151979566,
      "grad_norm": 0.71875,
      "learning_rate": 1.6530995608187767e-05,
      "loss": 0.282,
      "step": 7263
    },
    {
      "epoch": 0.5497563975214039,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6530093590252128e-05,
      "loss": 0.2978,
      "step": 7264
    },
    {
      "epoch": 0.5498320798448513,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6529191479675906e-05,
      "loss": 0.3236,
      "step": 7265
    },
    {
      "epoch": 0.5499077621682986,
      "grad_norm": 0.8125,
      "learning_rate": 1.65282892764719e-05,
      "loss": 0.3756,
      "step": 7266
    },
    {
      "epoch": 0.5499834444917459,
      "grad_norm": 0.80078125,
      "learning_rate": 1.6527386980652905e-05,
      "loss": 0.3275,
      "step": 7267
    },
    {
      "epoch": 0.5500591268151932,
      "grad_norm": 0.70703125,
      "learning_rate": 1.6526484592231726e-05,
      "loss": 0.2912,
      "step": 7268
    },
    {
      "epoch": 0.5501348091386405,
      "grad_norm": 0.75,
      "learning_rate": 1.6525582111221163e-05,
      "loss": 0.2912,
      "step": 7269
    },
    {
      "epoch": 0.5502104914620879,
      "grad_norm": 0.77734375,
      "learning_rate": 1.652467953763402e-05,
      "loss": 0.3235,
      "step": 7270
    },
    {
      "epoch": 0.5502861737855352,
      "grad_norm": 0.75,
      "learning_rate": 1.6523776871483102e-05,
      "loss": 0.2935,
      "step": 7271
    },
    {
      "epoch": 0.5503618561089826,
      "grad_norm": 0.796875,
      "learning_rate": 1.6522874112781213e-05,
      "loss": 0.3504,
      "step": 7272
    },
    {
      "epoch": 0.5504375384324299,
      "grad_norm": 0.84375,
      "learning_rate": 1.6521971261541164e-05,
      "loss": 0.3547,
      "step": 7273
    },
    {
      "epoch": 0.5505132207558772,
      "grad_norm": 0.79296875,
      "learning_rate": 1.6521068317775756e-05,
      "loss": 0.3376,
      "step": 7274
    },
    {
      "epoch": 0.5505889030793245,
      "grad_norm": 0.71875,
      "learning_rate": 1.6520165281497803e-05,
      "loss": 0.3064,
      "step": 7275
    },
    {
      "epoch": 0.5506645854027719,
      "grad_norm": 0.73828125,
      "learning_rate": 1.651926215272012e-05,
      "loss": 0.2569,
      "step": 7276
    },
    {
      "epoch": 0.5507402677262192,
      "grad_norm": 0.828125,
      "learning_rate": 1.6518358931455514e-05,
      "loss": 0.3644,
      "step": 7277
    },
    {
      "epoch": 0.5508159500496665,
      "grad_norm": 0.78125,
      "learning_rate": 1.65174556177168e-05,
      "loss": 0.3193,
      "step": 7278
    },
    {
      "epoch": 0.5508916323731139,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6516552211516794e-05,
      "loss": 0.3085,
      "step": 7279
    },
    {
      "epoch": 0.5509673146965612,
      "grad_norm": 0.82421875,
      "learning_rate": 1.651564871286831e-05,
      "loss": 0.3545,
      "step": 7280
    },
    {
      "epoch": 0.5510429970200085,
      "grad_norm": 2.609375,
      "learning_rate": 1.6514745121784174e-05,
      "loss": 0.3849,
      "step": 7281
    },
    {
      "epoch": 0.5511186793434558,
      "grad_norm": 0.828125,
      "learning_rate": 1.6513841438277195e-05,
      "loss": 0.3935,
      "step": 7282
    },
    {
      "epoch": 0.5511943616669032,
      "grad_norm": 0.73828125,
      "learning_rate": 1.6512937662360195e-05,
      "loss": 0.2998,
      "step": 7283
    },
    {
      "epoch": 0.5512700439903505,
      "grad_norm": 1.3125,
      "learning_rate": 1.6512033794046002e-05,
      "loss": 0.4143,
      "step": 7284
    },
    {
      "epoch": 0.5513457263137979,
      "grad_norm": 0.8125,
      "learning_rate": 1.6511129833347434e-05,
      "loss": 0.337,
      "step": 7285
    },
    {
      "epoch": 0.5514214086372452,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6510225780277313e-05,
      "loss": 0.3291,
      "step": 7286
    },
    {
      "epoch": 0.5514970909606925,
      "grad_norm": 0.7890625,
      "learning_rate": 1.650932163484847e-05,
      "loss": 0.3709,
      "step": 7287
    },
    {
      "epoch": 0.5515727732841398,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6508417397073727e-05,
      "loss": 0.3357,
      "step": 7288
    },
    {
      "epoch": 0.5516484556075871,
      "grad_norm": 0.70703125,
      "learning_rate": 1.6507513066965913e-05,
      "loss": 0.2584,
      "step": 7289
    },
    {
      "epoch": 0.5517241379310345,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6506608644537863e-05,
      "loss": 0.286,
      "step": 7290
    },
    {
      "epoch": 0.5517998202544818,
      "grad_norm": 0.82421875,
      "learning_rate": 1.6505704129802403e-05,
      "loss": 0.3459,
      "step": 7291
    },
    {
      "epoch": 0.5518755025779292,
      "grad_norm": 0.75,
      "learning_rate": 1.6504799522772366e-05,
      "loss": 0.2978,
      "step": 7292
    },
    {
      "epoch": 0.5519511849013765,
      "grad_norm": 0.82421875,
      "learning_rate": 1.6503894823460583e-05,
      "loss": 0.3748,
      "step": 7293
    },
    {
      "epoch": 0.5520268672248237,
      "grad_norm": 0.72265625,
      "learning_rate": 1.650299003187989e-05,
      "loss": 0.2929,
      "step": 7294
    },
    {
      "epoch": 0.5521025495482711,
      "grad_norm": 0.77734375,
      "learning_rate": 1.6502085148043126e-05,
      "loss": 0.3391,
      "step": 7295
    },
    {
      "epoch": 0.5521782318717184,
      "grad_norm": 0.80078125,
      "learning_rate": 1.6501180171963124e-05,
      "loss": 0.307,
      "step": 7296
    },
    {
      "epoch": 0.5522539141951658,
      "grad_norm": 0.8203125,
      "learning_rate": 1.650027510365273e-05,
      "loss": 0.3034,
      "step": 7297
    },
    {
      "epoch": 0.5523295965186131,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6499369943124775e-05,
      "loss": 0.333,
      "step": 7298
    },
    {
      "epoch": 0.5524052788420605,
      "grad_norm": 0.76171875,
      "learning_rate": 1.6498464690392103e-05,
      "loss": 0.3063,
      "step": 7299
    },
    {
      "epoch": 0.5524809611655078,
      "grad_norm": 0.765625,
      "learning_rate": 1.649755934546756e-05,
      "loss": 0.3156,
      "step": 7300
    },
    {
      "epoch": 0.5525566434889551,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6496653908363986e-05,
      "loss": 0.3113,
      "step": 7301
    },
    {
      "epoch": 0.5526323258124024,
      "grad_norm": 0.80859375,
      "learning_rate": 1.6495748379094227e-05,
      "loss": 0.3234,
      "step": 7302
    },
    {
      "epoch": 0.5527080081358497,
      "grad_norm": 0.984375,
      "learning_rate": 1.6494842757671136e-05,
      "loss": 0.3494,
      "step": 7303
    },
    {
      "epoch": 0.5527836904592971,
      "grad_norm": 0.765625,
      "learning_rate": 1.6493937044107546e-05,
      "loss": 0.3192,
      "step": 7304
    },
    {
      "epoch": 0.5528593727827444,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6493031238416323e-05,
      "loss": 0.3101,
      "step": 7305
    },
    {
      "epoch": 0.5529350551061918,
      "grad_norm": 0.88671875,
      "learning_rate": 1.6492125340610304e-05,
      "loss": 0.3882,
      "step": 7306
    },
    {
      "epoch": 0.5530107374296391,
      "grad_norm": 0.76953125,
      "learning_rate": 1.6491219350702348e-05,
      "loss": 0.3327,
      "step": 7307
    },
    {
      "epoch": 0.5530864197530864,
      "grad_norm": 0.734375,
      "learning_rate": 1.6490313268705308e-05,
      "loss": 0.2964,
      "step": 7308
    },
    {
      "epoch": 0.5531621020765337,
      "grad_norm": 0.75,
      "learning_rate": 1.6489407094632033e-05,
      "loss": 0.2998,
      "step": 7309
    },
    {
      "epoch": 0.5532377843999811,
      "grad_norm": 0.75,
      "learning_rate": 1.6488500828495384e-05,
      "loss": 0.2964,
      "step": 7310
    },
    {
      "epoch": 0.5533134667234284,
      "grad_norm": 0.91015625,
      "learning_rate": 1.6487594470308215e-05,
      "loss": 0.3786,
      "step": 7311
    },
    {
      "epoch": 0.5533891490468758,
      "grad_norm": 0.7265625,
      "learning_rate": 1.6486688020083388e-05,
      "loss": 0.2687,
      "step": 7312
    },
    {
      "epoch": 0.5534648313703231,
      "grad_norm": 0.74609375,
      "learning_rate": 1.6485781477833758e-05,
      "loss": 0.2955,
      "step": 7313
    },
    {
      "epoch": 0.5535405136937704,
      "grad_norm": 0.8359375,
      "learning_rate": 1.648487484357219e-05,
      "loss": 0.3691,
      "step": 7314
    },
    {
      "epoch": 0.5536161960172177,
      "grad_norm": 0.78125,
      "learning_rate": 1.6483968117311538e-05,
      "loss": 0.3252,
      "step": 7315
    },
    {
      "epoch": 0.553691878340665,
      "grad_norm": 0.7578125,
      "learning_rate": 1.648306129906468e-05,
      "loss": 0.2601,
      "step": 7316
    },
    {
      "epoch": 0.5537675606641124,
      "grad_norm": 0.73828125,
      "learning_rate": 1.648215438884446e-05,
      "loss": 0.3134,
      "step": 7317
    },
    {
      "epoch": 0.5538432429875597,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6481247386663767e-05,
      "loss": 0.3388,
      "step": 7318
    },
    {
      "epoch": 0.5539189253110071,
      "grad_norm": 0.81640625,
      "learning_rate": 1.6480340292535454e-05,
      "loss": 0.3603,
      "step": 7319
    },
    {
      "epoch": 0.5539946076344544,
      "grad_norm": 0.8203125,
      "learning_rate": 1.6479433106472394e-05,
      "loss": 0.352,
      "step": 7320
    },
    {
      "epoch": 0.5540702899579018,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6478525828487456e-05,
      "loss": 0.3169,
      "step": 7321
    },
    {
      "epoch": 0.554145972281349,
      "grad_norm": 0.73046875,
      "learning_rate": 1.6477618458593513e-05,
      "loss": 0.2911,
      "step": 7322
    },
    {
      "epoch": 0.5542216546047963,
      "grad_norm": 0.75,
      "learning_rate": 1.6476710996803434e-05,
      "loss": 0.2916,
      "step": 7323
    },
    {
      "epoch": 0.5542973369282437,
      "grad_norm": 0.8125,
      "learning_rate": 1.6475803443130095e-05,
      "loss": 0.3594,
      "step": 7324
    },
    {
      "epoch": 0.554373019251691,
      "grad_norm": 0.734375,
      "learning_rate": 1.6474895797586376e-05,
      "loss": 0.3037,
      "step": 7325
    },
    {
      "epoch": 0.5544487015751384,
      "grad_norm": 0.765625,
      "learning_rate": 1.6473988060185142e-05,
      "loss": 0.3278,
      "step": 7326
    },
    {
      "epoch": 0.5545243838985857,
      "grad_norm": 0.7734375,
      "learning_rate": 1.647308023093928e-05,
      "loss": 0.3255,
      "step": 7327
    },
    {
      "epoch": 0.5546000662220331,
      "grad_norm": 0.83203125,
      "learning_rate": 1.647217230986167e-05,
      "loss": 0.3423,
      "step": 7328
    },
    {
      "epoch": 0.5546757485454803,
      "grad_norm": 0.796875,
      "learning_rate": 1.6471264296965185e-05,
      "loss": 0.3415,
      "step": 7329
    },
    {
      "epoch": 0.5547514308689276,
      "grad_norm": 0.78515625,
      "learning_rate": 1.647035619226271e-05,
      "loss": 0.3491,
      "step": 7330
    },
    {
      "epoch": 0.554827113192375,
      "grad_norm": 0.80859375,
      "learning_rate": 1.6469447995767132e-05,
      "loss": 0.3458,
      "step": 7331
    },
    {
      "epoch": 0.5549027955158223,
      "grad_norm": 0.6953125,
      "learning_rate": 1.646853970749133e-05,
      "loss": 0.2909,
      "step": 7332
    },
    {
      "epoch": 0.5549784778392697,
      "grad_norm": 0.78515625,
      "learning_rate": 1.64676313274482e-05,
      "loss": 0.3357,
      "step": 7333
    },
    {
      "epoch": 0.555054160162717,
      "grad_norm": 0.82421875,
      "learning_rate": 1.646672285565061e-05,
      "loss": 0.3631,
      "step": 7334
    },
    {
      "epoch": 0.5551298424861644,
      "grad_norm": 0.7109375,
      "learning_rate": 1.646581429211146e-05,
      "loss": 0.2796,
      "step": 7335
    },
    {
      "epoch": 0.5552055248096116,
      "grad_norm": 0.81640625,
      "learning_rate": 1.646490563684364e-05,
      "loss": 0.3655,
      "step": 7336
    },
    {
      "epoch": 0.555281207133059,
      "grad_norm": 0.79296875,
      "learning_rate": 1.6463996889860047e-05,
      "loss": 0.3165,
      "step": 7337
    },
    {
      "epoch": 0.5553568894565063,
      "grad_norm": 0.796875,
      "learning_rate": 1.646308805117356e-05,
      "loss": 0.3503,
      "step": 7338
    },
    {
      "epoch": 0.5554325717799536,
      "grad_norm": 0.71484375,
      "learning_rate": 1.6462179120797075e-05,
      "loss": 0.3026,
      "step": 7339
    },
    {
      "epoch": 0.555508254103401,
      "grad_norm": 0.77734375,
      "learning_rate": 1.6461270098743495e-05,
      "loss": 0.3201,
      "step": 7340
    },
    {
      "epoch": 0.5555839364268483,
      "grad_norm": 0.7421875,
      "learning_rate": 1.6460360985025703e-05,
      "loss": 0.3,
      "step": 7341
    },
    {
      "epoch": 0.5556596187502957,
      "grad_norm": 0.765625,
      "learning_rate": 1.645945177965661e-05,
      "loss": 0.3187,
      "step": 7342
    },
    {
      "epoch": 0.5557353010737429,
      "grad_norm": 0.71875,
      "learning_rate": 1.645854248264911e-05,
      "loss": 0.2847,
      "step": 7343
    },
    {
      "epoch": 0.5558109833971903,
      "grad_norm": 0.71875,
      "learning_rate": 1.64576330940161e-05,
      "loss": 0.2731,
      "step": 7344
    },
    {
      "epoch": 0.5558866657206376,
      "grad_norm": 0.7890625,
      "learning_rate": 1.645672361377048e-05,
      "loss": 0.3225,
      "step": 7345
    },
    {
      "epoch": 0.555962348044085,
      "grad_norm": 0.828125,
      "learning_rate": 1.6455814041925157e-05,
      "loss": 0.3826,
      "step": 7346
    },
    {
      "epoch": 0.5560380303675323,
      "grad_norm": 0.765625,
      "learning_rate": 1.645490437849304e-05,
      "loss": 0.3297,
      "step": 7347
    },
    {
      "epoch": 0.5561137126909796,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6453994623487016e-05,
      "loss": 0.3027,
      "step": 7348
    },
    {
      "epoch": 0.556189395014427,
      "grad_norm": 0.76953125,
      "learning_rate": 1.6453084776920008e-05,
      "loss": 0.331,
      "step": 7349
    },
    {
      "epoch": 0.5562650773378742,
      "grad_norm": 0.80078125,
      "learning_rate": 1.645217483880492e-05,
      "loss": 0.3556,
      "step": 7350
    },
    {
      "epoch": 0.5563407596613216,
      "grad_norm": 0.73046875,
      "learning_rate": 1.6451264809154658e-05,
      "loss": 0.2843,
      "step": 7351
    },
    {
      "epoch": 0.5564164419847689,
      "grad_norm": 0.68359375,
      "learning_rate": 1.6450354687982134e-05,
      "loss": 0.2424,
      "step": 7352
    },
    {
      "epoch": 0.5564921243082163,
      "grad_norm": 0.765625,
      "learning_rate": 1.644944447530026e-05,
      "loss": 0.3146,
      "step": 7353
    },
    {
      "epoch": 0.5565678066316636,
      "grad_norm": 0.7109375,
      "learning_rate": 1.6448534171121946e-05,
      "loss": 0.2806,
      "step": 7354
    },
    {
      "epoch": 0.556643488955111,
      "grad_norm": 0.7890625,
      "learning_rate": 1.644762377546011e-05,
      "loss": 0.3402,
      "step": 7355
    },
    {
      "epoch": 0.5567191712785583,
      "grad_norm": 0.74609375,
      "learning_rate": 1.644671328832767e-05,
      "loss": 0.3159,
      "step": 7356
    },
    {
      "epoch": 0.5567948536020055,
      "grad_norm": 0.79296875,
      "learning_rate": 1.6445802709737536e-05,
      "loss": 0.3319,
      "step": 7357
    },
    {
      "epoch": 0.5568705359254529,
      "grad_norm": 0.80078125,
      "learning_rate": 1.644489203970263e-05,
      "loss": 0.3699,
      "step": 7358
    },
    {
      "epoch": 0.5569462182489002,
      "grad_norm": 0.7890625,
      "learning_rate": 1.644398127823587e-05,
      "loss": 0.3502,
      "step": 7359
    },
    {
      "epoch": 0.5570219005723476,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6443070425350178e-05,
      "loss": 0.2922,
      "step": 7360
    },
    {
      "epoch": 0.5570975828957949,
      "grad_norm": 0.8984375,
      "learning_rate": 1.6442159481058472e-05,
      "loss": 0.392,
      "step": 7361
    },
    {
      "epoch": 0.5571732652192423,
      "grad_norm": 0.8125,
      "learning_rate": 1.6441248445373684e-05,
      "loss": 0.3359,
      "step": 7362
    },
    {
      "epoch": 0.5572489475426896,
      "grad_norm": 0.75,
      "learning_rate": 1.644033731830873e-05,
      "loss": 0.3052,
      "step": 7363
    },
    {
      "epoch": 0.5573246298661368,
      "grad_norm": 0.79296875,
      "learning_rate": 1.643942609987654e-05,
      "loss": 0.3256,
      "step": 7364
    },
    {
      "epoch": 0.5574003121895842,
      "grad_norm": 0.8046875,
      "learning_rate": 1.643851479009004e-05,
      "loss": 0.302,
      "step": 7365
    },
    {
      "epoch": 0.5574759945130315,
      "grad_norm": 0.765625,
      "learning_rate": 1.643760338896216e-05,
      "loss": 0.3271,
      "step": 7366
    },
    {
      "epoch": 0.5575516768364789,
      "grad_norm": 0.7265625,
      "learning_rate": 1.6436691896505824e-05,
      "loss": 0.3058,
      "step": 7367
    },
    {
      "epoch": 0.5576273591599262,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6435780312733974e-05,
      "loss": 0.3271,
      "step": 7368
    },
    {
      "epoch": 0.5577030414833736,
      "grad_norm": 0.71484375,
      "learning_rate": 1.6434868637659535e-05,
      "loss": 0.2743,
      "step": 7369
    },
    {
      "epoch": 0.5577787238068209,
      "grad_norm": 0.78125,
      "learning_rate": 1.6433956871295435e-05,
      "loss": 0.3338,
      "step": 7370
    },
    {
      "epoch": 0.5578544061302682,
      "grad_norm": 0.76171875,
      "learning_rate": 1.643304501365462e-05,
      "loss": 0.3147,
      "step": 7371
    },
    {
      "epoch": 0.5579300884537155,
      "grad_norm": 0.73828125,
      "learning_rate": 1.6432133064750026e-05,
      "loss": 0.2988,
      "step": 7372
    },
    {
      "epoch": 0.5580057707771628,
      "grad_norm": 0.734375,
      "learning_rate": 1.6431221024594584e-05,
      "loss": 0.2923,
      "step": 7373
    },
    {
      "epoch": 0.5580814531006102,
      "grad_norm": 0.79296875,
      "learning_rate": 1.6430308893201233e-05,
      "loss": 0.2919,
      "step": 7374
    },
    {
      "epoch": 0.5581571354240575,
      "grad_norm": 0.83984375,
      "learning_rate": 1.642939667058292e-05,
      "loss": 0.3736,
      "step": 7375
    },
    {
      "epoch": 0.5582328177475049,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6428484356752578e-05,
      "loss": 0.3308,
      "step": 7376
    },
    {
      "epoch": 0.5583085000709522,
      "grad_norm": 0.734375,
      "learning_rate": 1.6427571951723152e-05,
      "loss": 0.294,
      "step": 7377
    },
    {
      "epoch": 0.5583841823943995,
      "grad_norm": 1.1640625,
      "learning_rate": 1.642665945550759e-05,
      "loss": 0.4286,
      "step": 7378
    },
    {
      "epoch": 0.5584598647178468,
      "grad_norm": 0.8515625,
      "learning_rate": 1.642574686811884e-05,
      "loss": 0.3582,
      "step": 7379
    },
    {
      "epoch": 0.5585355470412942,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6424834189569838e-05,
      "loss": 0.3028,
      "step": 7380
    },
    {
      "epoch": 0.5586112293647415,
      "grad_norm": 0.75,
      "learning_rate": 1.6423921419873542e-05,
      "loss": 0.3157,
      "step": 7381
    },
    {
      "epoch": 0.5586869116881888,
      "grad_norm": 0.8671875,
      "learning_rate": 1.642300855904289e-05,
      "loss": 0.4058,
      "step": 7382
    },
    {
      "epoch": 0.5587625940116362,
      "grad_norm": 0.76171875,
      "learning_rate": 1.6422095607090844e-05,
      "loss": 0.289,
      "step": 7383
    },
    {
      "epoch": 0.5588382763350835,
      "grad_norm": 0.73046875,
      "learning_rate": 1.6421182564030355e-05,
      "loss": 0.3134,
      "step": 7384
    },
    {
      "epoch": 0.5589139586585308,
      "grad_norm": 0.796875,
      "learning_rate": 1.642026942987437e-05,
      "loss": 0.3265,
      "step": 7385
    },
    {
      "epoch": 0.5589896409819781,
      "grad_norm": 0.78125,
      "learning_rate": 1.641935620463584e-05,
      "loss": 0.3371,
      "step": 7386
    },
    {
      "epoch": 0.5590653233054255,
      "grad_norm": 0.80078125,
      "learning_rate": 1.6418442888327728e-05,
      "loss": 0.3681,
      "step": 7387
    },
    {
      "epoch": 0.5591410056288728,
      "grad_norm": 0.734375,
      "learning_rate": 1.641752948096299e-05,
      "loss": 0.2786,
      "step": 7388
    },
    {
      "epoch": 0.5592166879523202,
      "grad_norm": 0.765625,
      "learning_rate": 1.6416615982554586e-05,
      "loss": 0.3085,
      "step": 7389
    },
    {
      "epoch": 0.5592923702757675,
      "grad_norm": 0.7265625,
      "learning_rate": 1.641570239311547e-05,
      "loss": 0.3006,
      "step": 7390
    },
    {
      "epoch": 0.5593680525992148,
      "grad_norm": 0.80859375,
      "learning_rate": 1.641478871265861e-05,
      "loss": 0.3227,
      "step": 7391
    },
    {
      "epoch": 0.5594437349226621,
      "grad_norm": 0.8203125,
      "learning_rate": 1.6413874941196957e-05,
      "loss": 0.3317,
      "step": 7392
    },
    {
      "epoch": 0.5595194172461094,
      "grad_norm": 0.73828125,
      "learning_rate": 1.6412961078743486e-05,
      "loss": 0.3131,
      "step": 7393
    },
    {
      "epoch": 0.5595950995695568,
      "grad_norm": 0.76171875,
      "learning_rate": 1.6412047125311155e-05,
      "loss": 0.3329,
      "step": 7394
    },
    {
      "epoch": 0.5596707818930041,
      "grad_norm": 0.78125,
      "learning_rate": 1.6411133080912937e-05,
      "loss": 0.3373,
      "step": 7395
    },
    {
      "epoch": 0.5597464642164515,
      "grad_norm": 1.5078125,
      "learning_rate": 1.6410218945561788e-05,
      "loss": 0.4338,
      "step": 7396
    },
    {
      "epoch": 0.5598221465398988,
      "grad_norm": 0.77734375,
      "learning_rate": 1.6409304719270685e-05,
      "loss": 0.2903,
      "step": 7397
    },
    {
      "epoch": 0.5598978288633462,
      "grad_norm": 0.77734375,
      "learning_rate": 1.6408390402052597e-05,
      "loss": 0.3011,
      "step": 7398
    },
    {
      "epoch": 0.5599735111867934,
      "grad_norm": 1.0703125,
      "learning_rate": 1.6407475993920495e-05,
      "loss": 0.36,
      "step": 7399
    },
    {
      "epoch": 0.5600491935102407,
      "grad_norm": 0.75,
      "learning_rate": 1.640656149488735e-05,
      "loss": 0.3405,
      "step": 7400
    },
    {
      "epoch": 0.5601248758336881,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6405646904966135e-05,
      "loss": 0.3092,
      "step": 7401
    },
    {
      "epoch": 0.5602005581571354,
      "grad_norm": 0.7109375,
      "learning_rate": 1.6404732224169827e-05,
      "loss": 0.2832,
      "step": 7402
    },
    {
      "epoch": 0.5602762404805828,
      "grad_norm": 0.7421875,
      "learning_rate": 1.64038174525114e-05,
      "loss": 0.2892,
      "step": 7403
    },
    {
      "epoch": 0.5603519228040301,
      "grad_norm": 0.7421875,
      "learning_rate": 1.6402902590003836e-05,
      "loss": 0.2998,
      "step": 7404
    },
    {
      "epoch": 0.5604276051274775,
      "grad_norm": 0.79296875,
      "learning_rate": 1.640198763666011e-05,
      "loss": 0.3145,
      "step": 7405
    },
    {
      "epoch": 0.5605032874509247,
      "grad_norm": 1.296875,
      "learning_rate": 1.64010725924932e-05,
      "loss": 0.4183,
      "step": 7406
    },
    {
      "epoch": 0.560578969774372,
      "grad_norm": 0.8203125,
      "learning_rate": 1.640015745751609e-05,
      "loss": 0.3502,
      "step": 7407
    },
    {
      "epoch": 0.5606546520978194,
      "grad_norm": 0.7734375,
      "learning_rate": 1.639924223174177e-05,
      "loss": 0.3471,
      "step": 7408
    },
    {
      "epoch": 0.5607303344212667,
      "grad_norm": 0.80078125,
      "learning_rate": 1.6398326915183212e-05,
      "loss": 0.3596,
      "step": 7409
    },
    {
      "epoch": 0.5608060167447141,
      "grad_norm": 0.83203125,
      "learning_rate": 1.639741150785341e-05,
      "loss": 0.3042,
      "step": 7410
    },
    {
      "epoch": 0.5608816990681614,
      "grad_norm": 0.7421875,
      "learning_rate": 1.6396496009765345e-05,
      "loss": 0.3217,
      "step": 7411
    },
    {
      "epoch": 0.5609573813916087,
      "grad_norm": 0.76953125,
      "learning_rate": 1.6395580420932008e-05,
      "loss": 0.2988,
      "step": 7412
    },
    {
      "epoch": 0.561033063715056,
      "grad_norm": 0.84765625,
      "learning_rate": 1.639466474136639e-05,
      "loss": 0.3322,
      "step": 7413
    },
    {
      "epoch": 0.5611087460385034,
      "grad_norm": 0.765625,
      "learning_rate": 1.6393748971081475e-05,
      "loss": 0.3479,
      "step": 7414
    },
    {
      "epoch": 0.5611844283619507,
      "grad_norm": 0.81640625,
      "learning_rate": 1.639283311009026e-05,
      "loss": 0.3453,
      "step": 7415
    },
    {
      "epoch": 0.561260110685398,
      "grad_norm": 0.76953125,
      "learning_rate": 1.639191715840574e-05,
      "loss": 0.3294,
      "step": 7416
    },
    {
      "epoch": 0.5613357930088454,
      "grad_norm": 0.74609375,
      "learning_rate": 1.63910011160409e-05,
      "loss": 0.3013,
      "step": 7417
    },
    {
      "epoch": 0.5614114753322927,
      "grad_norm": 0.8125,
      "learning_rate": 1.6390084983008746e-05,
      "loss": 0.361,
      "step": 7418
    },
    {
      "epoch": 0.56148715765574,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6389168759322274e-05,
      "loss": 0.3427,
      "step": 7419
    },
    {
      "epoch": 0.5615628399791873,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6388252444994474e-05,
      "loss": 0.2863,
      "step": 7420
    },
    {
      "epoch": 0.5616385223026347,
      "grad_norm": 0.78125,
      "learning_rate": 1.638733604003835e-05,
      "loss": 0.3573,
      "step": 7421
    },
    {
      "epoch": 0.561714204626082,
      "grad_norm": 0.78515625,
      "learning_rate": 1.638641954446691e-05,
      "loss": 0.3393,
      "step": 7422
    },
    {
      "epoch": 0.5617898869495294,
      "grad_norm": 0.7734375,
      "learning_rate": 1.638550295829314e-05,
      "loss": 0.3395,
      "step": 7423
    },
    {
      "epoch": 0.5618655692729767,
      "grad_norm": 0.73828125,
      "learning_rate": 1.6384586281530056e-05,
      "loss": 0.2972,
      "step": 7424
    },
    {
      "epoch": 0.561941251596424,
      "grad_norm": 0.703125,
      "learning_rate": 1.6383669514190662e-05,
      "loss": 0.2793,
      "step": 7425
    },
    {
      "epoch": 0.5620169339198713,
      "grad_norm": 0.73828125,
      "learning_rate": 1.638275265628796e-05,
      "loss": 0.3098,
      "step": 7426
    },
    {
      "epoch": 0.5620926162433186,
      "grad_norm": 0.8125,
      "learning_rate": 1.6381835707834956e-05,
      "loss": 0.334,
      "step": 7427
    },
    {
      "epoch": 0.562168298566766,
      "grad_norm": 0.76953125,
      "learning_rate": 1.6380918668844663e-05,
      "loss": 0.3398,
      "step": 7428
    },
    {
      "epoch": 0.5622439808902133,
      "grad_norm": 0.72265625,
      "learning_rate": 1.6380001539330088e-05,
      "loss": 0.2624,
      "step": 7429
    },
    {
      "epoch": 0.5623196632136607,
      "grad_norm": 0.78125,
      "learning_rate": 1.6379084319304245e-05,
      "loss": 0.3545,
      "step": 7430
    },
    {
      "epoch": 0.562395345537108,
      "grad_norm": 0.69921875,
      "learning_rate": 1.637816700878014e-05,
      "loss": 0.2439,
      "step": 7431
    },
    {
      "epoch": 0.5624710278605554,
      "grad_norm": 0.78125,
      "learning_rate": 1.6377249607770792e-05,
      "loss": 0.3217,
      "step": 7432
    },
    {
      "epoch": 0.5625467101840026,
      "grad_norm": 0.7421875,
      "learning_rate": 1.6376332116289216e-05,
      "loss": 0.3267,
      "step": 7433
    },
    {
      "epoch": 0.5626223925074499,
      "grad_norm": 0.71484375,
      "learning_rate": 1.6375414534348425e-05,
      "loss": 0.2926,
      "step": 7434
    },
    {
      "epoch": 0.5626980748308973,
      "grad_norm": 0.765625,
      "learning_rate": 1.637449686196144e-05,
      "loss": 0.3171,
      "step": 7435
    },
    {
      "epoch": 0.5627737571543446,
      "grad_norm": 0.84375,
      "learning_rate": 1.6373579099141277e-05,
      "loss": 0.3216,
      "step": 7436
    },
    {
      "epoch": 0.562849439477792,
      "grad_norm": 0.81640625,
      "learning_rate": 1.637266124590096e-05,
      "loss": 0.3801,
      "step": 7437
    },
    {
      "epoch": 0.5629251218012393,
      "grad_norm": 0.78125,
      "learning_rate": 1.63717433022535e-05,
      "loss": 0.3517,
      "step": 7438
    },
    {
      "epoch": 0.5630008041246867,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6370825268211933e-05,
      "loss": 0.3544,
      "step": 7439
    },
    {
      "epoch": 0.5630764864481339,
      "grad_norm": 0.81640625,
      "learning_rate": 1.6369907143789277e-05,
      "loss": 0.3193,
      "step": 7440
    },
    {
      "epoch": 0.5631521687715813,
      "grad_norm": 0.796875,
      "learning_rate": 1.6368988928998556e-05,
      "loss": 0.317,
      "step": 7441
    },
    {
      "epoch": 0.5632278510950286,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6368070623852794e-05,
      "loss": 0.3124,
      "step": 7442
    },
    {
      "epoch": 0.5633035334184759,
      "grad_norm": 0.734375,
      "learning_rate": 1.636715222836503e-05,
      "loss": 0.2875,
      "step": 7443
    },
    {
      "epoch": 0.5633792157419233,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6366233742548275e-05,
      "loss": 0.3268,
      "step": 7444
    },
    {
      "epoch": 0.5634548980653706,
      "grad_norm": 0.76953125,
      "learning_rate": 1.636531516641558e-05,
      "loss": 0.3155,
      "step": 7445
    },
    {
      "epoch": 0.563530580388818,
      "grad_norm": 0.78125,
      "learning_rate": 1.636439649997996e-05,
      "loss": 0.3302,
      "step": 7446
    },
    {
      "epoch": 0.5636062627122652,
      "grad_norm": 0.77734375,
      "learning_rate": 1.6363477743254454e-05,
      "loss": 0.335,
      "step": 7447
    },
    {
      "epoch": 0.5636819450357126,
      "grad_norm": 0.84375,
      "learning_rate": 1.6362558896252097e-05,
      "loss": 0.2904,
      "step": 7448
    },
    {
      "epoch": 0.5637576273591599,
      "grad_norm": 0.73828125,
      "learning_rate": 1.6361639958985925e-05,
      "loss": 0.3301,
      "step": 7449
    },
    {
      "epoch": 0.5638333096826073,
      "grad_norm": 1.1484375,
      "learning_rate": 1.636072093146897e-05,
      "loss": 0.3652,
      "step": 7450
    },
    {
      "epoch": 0.5639089920060546,
      "grad_norm": 0.85546875,
      "learning_rate": 1.6359801813714274e-05,
      "loss": 0.3468,
      "step": 7451
    },
    {
      "epoch": 0.563984674329502,
      "grad_norm": 0.80078125,
      "learning_rate": 1.6358882605734878e-05,
      "loss": 0.3267,
      "step": 7452
    },
    {
      "epoch": 0.5640603566529493,
      "grad_norm": 0.8203125,
      "learning_rate": 1.635796330754382e-05,
      "loss": 0.3707,
      "step": 7453
    },
    {
      "epoch": 0.5641360389763965,
      "grad_norm": 0.7421875,
      "learning_rate": 1.6357043919154137e-05,
      "loss": 0.3218,
      "step": 7454
    },
    {
      "epoch": 0.5642117212998439,
      "grad_norm": 0.78125,
      "learning_rate": 1.635612444057888e-05,
      "loss": 0.3078,
      "step": 7455
    },
    {
      "epoch": 0.5642874036232912,
      "grad_norm": 0.83984375,
      "learning_rate": 1.6355204871831094e-05,
      "loss": 0.3796,
      "step": 7456
    },
    {
      "epoch": 0.5643630859467386,
      "grad_norm": 0.828125,
      "learning_rate": 1.6354285212923818e-05,
      "loss": 0.3348,
      "step": 7457
    },
    {
      "epoch": 0.5644387682701859,
      "grad_norm": 0.78125,
      "learning_rate": 1.63533654638701e-05,
      "loss": 0.2973,
      "step": 7458
    },
    {
      "epoch": 0.5645144505936333,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6352445624682993e-05,
      "loss": 0.3006,
      "step": 7459
    },
    {
      "epoch": 0.5645901329170806,
      "grad_norm": 0.80078125,
      "learning_rate": 1.6351525695375543e-05,
      "loss": 0.3328,
      "step": 7460
    },
    {
      "epoch": 0.5646658152405278,
      "grad_norm": 0.83203125,
      "learning_rate": 1.63506056759608e-05,
      "loss": 0.3592,
      "step": 7461
    },
    {
      "epoch": 0.5647414975639752,
      "grad_norm": 0.74609375,
      "learning_rate": 1.634968556645182e-05,
      "loss": 0.295,
      "step": 7462
    },
    {
      "epoch": 0.5648171798874225,
      "grad_norm": 0.890625,
      "learning_rate": 1.634876536686165e-05,
      "loss": 0.3919,
      "step": 7463
    },
    {
      "epoch": 0.5648928622108699,
      "grad_norm": 0.828125,
      "learning_rate": 1.634784507720335e-05,
      "loss": 0.3372,
      "step": 7464
    },
    {
      "epoch": 0.5649685445343172,
      "grad_norm": 0.8046875,
      "learning_rate": 1.6346924697489975e-05,
      "loss": 0.3386,
      "step": 7465
    },
    {
      "epoch": 0.5650442268577646,
      "grad_norm": 0.78125,
      "learning_rate": 1.6346004227734583e-05,
      "loss": 0.3205,
      "step": 7466
    },
    {
      "epoch": 0.5651199091812119,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6345083667950227e-05,
      "loss": 0.3091,
      "step": 7467
    },
    {
      "epoch": 0.5651955915046591,
      "grad_norm": 1.125,
      "learning_rate": 1.6344163018149975e-05,
      "loss": 0.3631,
      "step": 7468
    },
    {
      "epoch": 0.5652712738281065,
      "grad_norm": 0.796875,
      "learning_rate": 1.634324227834688e-05,
      "loss": 0.3373,
      "step": 7469
    },
    {
      "epoch": 0.5653469561515538,
      "grad_norm": 0.7890625,
      "learning_rate": 1.634232144855401e-05,
      "loss": 0.3275,
      "step": 7470
    },
    {
      "epoch": 0.5654226384750012,
      "grad_norm": 0.80859375,
      "learning_rate": 1.6341400528784426e-05,
      "loss": 0.3417,
      "step": 7471
    },
    {
      "epoch": 0.5654983207984485,
      "grad_norm": 0.8046875,
      "learning_rate": 1.6340479519051192e-05,
      "loss": 0.3367,
      "step": 7472
    },
    {
      "epoch": 0.5655740031218959,
      "grad_norm": 0.80078125,
      "learning_rate": 1.6339558419367378e-05,
      "loss": 0.3411,
      "step": 7473
    },
    {
      "epoch": 0.5656496854453432,
      "grad_norm": 0.77734375,
      "learning_rate": 1.6338637229746046e-05,
      "loss": 0.3034,
      "step": 7474
    },
    {
      "epoch": 0.5657253677687905,
      "grad_norm": 0.80078125,
      "learning_rate": 1.633771595020027e-05,
      "loss": 0.33,
      "step": 7475
    },
    {
      "epoch": 0.5658010500922378,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6336794580743115e-05,
      "loss": 0.326,
      "step": 7476
    },
    {
      "epoch": 0.5658767324156851,
      "grad_norm": 0.74609375,
      "learning_rate": 1.6335873121387655e-05,
      "loss": 0.3199,
      "step": 7477
    },
    {
      "epoch": 0.5659524147391325,
      "grad_norm": 0.8203125,
      "learning_rate": 1.6334951572146966e-05,
      "loss": 0.3556,
      "step": 7478
    },
    {
      "epoch": 0.5660280970625798,
      "grad_norm": 0.734375,
      "learning_rate": 1.6334029933034112e-05,
      "loss": 0.3003,
      "step": 7479
    },
    {
      "epoch": 0.5661037793860272,
      "grad_norm": 0.796875,
      "learning_rate": 1.6333108204062177e-05,
      "loss": 0.3668,
      "step": 7480
    },
    {
      "epoch": 0.5661794617094745,
      "grad_norm": 1.3671875,
      "learning_rate": 1.6332186385244233e-05,
      "loss": 0.3967,
      "step": 7481
    },
    {
      "epoch": 0.5662551440329218,
      "grad_norm": 0.71875,
      "learning_rate": 1.633126447659336e-05,
      "loss": 0.3084,
      "step": 7482
    },
    {
      "epoch": 0.5663308263563691,
      "grad_norm": 0.734375,
      "learning_rate": 1.6330342478122635e-05,
      "loss": 0.2928,
      "step": 7483
    },
    {
      "epoch": 0.5664065086798165,
      "grad_norm": 0.7109375,
      "learning_rate": 1.632942038984514e-05,
      "loss": 0.2878,
      "step": 7484
    },
    {
      "epoch": 0.5664821910032638,
      "grad_norm": 0.71875,
      "learning_rate": 1.632849821177395e-05,
      "loss": 0.3046,
      "step": 7485
    },
    {
      "epoch": 0.5665578733267111,
      "grad_norm": 0.734375,
      "learning_rate": 1.6327575943922157e-05,
      "loss": 0.3038,
      "step": 7486
    },
    {
      "epoch": 0.5666335556501585,
      "grad_norm": 0.79296875,
      "learning_rate": 1.6326653586302842e-05,
      "loss": 0.3249,
      "step": 7487
    },
    {
      "epoch": 0.5667092379736058,
      "grad_norm": 0.80078125,
      "learning_rate": 1.6325731138929084e-05,
      "loss": 0.3135,
      "step": 7488
    },
    {
      "epoch": 0.5667849202970531,
      "grad_norm": 0.8125,
      "learning_rate": 1.632480860181398e-05,
      "loss": 0.3308,
      "step": 7489
    },
    {
      "epoch": 0.5668606026205004,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6323885974970606e-05,
      "loss": 0.2997,
      "step": 7490
    },
    {
      "epoch": 0.5669362849439478,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6322963258412064e-05,
      "loss": 0.3041,
      "step": 7491
    },
    {
      "epoch": 0.5670119672673951,
      "grad_norm": 0.796875,
      "learning_rate": 1.6322040452151435e-05,
      "loss": 0.3415,
      "step": 7492
    },
    {
      "epoch": 0.5670876495908425,
      "grad_norm": 0.8515625,
      "learning_rate": 1.632111755620181e-05,
      "loss": 0.3803,
      "step": 7493
    },
    {
      "epoch": 0.5671633319142898,
      "grad_norm": 0.734375,
      "learning_rate": 1.6320194570576288e-05,
      "loss": 0.2892,
      "step": 7494
    },
    {
      "epoch": 0.5672390142377371,
      "grad_norm": 0.72265625,
      "learning_rate": 1.6319271495287963e-05,
      "loss": 0.3026,
      "step": 7495
    },
    {
      "epoch": 0.5673146965611844,
      "grad_norm": 0.78125,
      "learning_rate": 1.631834833034992e-05,
      "loss": 0.316,
      "step": 7496
    },
    {
      "epoch": 0.5673903788846317,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6317425075775272e-05,
      "loss": 0.3126,
      "step": 7497
    },
    {
      "epoch": 0.5674660612080791,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6316501731577103e-05,
      "loss": 0.3031,
      "step": 7498
    },
    {
      "epoch": 0.5675417435315264,
      "grad_norm": 0.7734375,
      "learning_rate": 1.631557829776852e-05,
      "loss": 0.2841,
      "step": 7499
    },
    {
      "epoch": 0.5676174258549738,
      "grad_norm": 0.76953125,
      "learning_rate": 1.631465477436262e-05,
      "loss": 0.3001,
      "step": 7500
    },
    {
      "epoch": 0.5676931081784211,
      "grad_norm": 0.79296875,
      "learning_rate": 1.6313731161372506e-05,
      "loss": 0.3662,
      "step": 7501
    },
    {
      "epoch": 0.5677687905018685,
      "grad_norm": 1.203125,
      "learning_rate": 1.631280745881128e-05,
      "loss": 0.3863,
      "step": 7502
    },
    {
      "epoch": 0.5678444728253157,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6311883666692046e-05,
      "loss": 0.2935,
      "step": 7503
    },
    {
      "epoch": 0.567920155148763,
      "grad_norm": 0.82421875,
      "learning_rate": 1.631095978502791e-05,
      "loss": 0.3506,
      "step": 7504
    },
    {
      "epoch": 0.5679958374722104,
      "grad_norm": 0.734375,
      "learning_rate": 1.631003581383198e-05,
      "loss": 0.2941,
      "step": 7505
    },
    {
      "epoch": 0.5680715197956577,
      "grad_norm": 0.828125,
      "learning_rate": 1.630911175311737e-05,
      "loss": 0.3729,
      "step": 7506
    },
    {
      "epoch": 0.5681472021191051,
      "grad_norm": 0.80078125,
      "learning_rate": 1.6308187602897173e-05,
      "loss": 0.3259,
      "step": 7507
    },
    {
      "epoch": 0.5682228844425524,
      "grad_norm": 0.953125,
      "learning_rate": 1.6307263363184517e-05,
      "loss": 0.3466,
      "step": 7508
    },
    {
      "epoch": 0.5682985667659998,
      "grad_norm": 0.73828125,
      "learning_rate": 1.6306339033992503e-05,
      "loss": 0.2895,
      "step": 7509
    },
    {
      "epoch": 0.568374249089447,
      "grad_norm": 0.7421875,
      "learning_rate": 1.6305414615334247e-05,
      "loss": 0.3217,
      "step": 7510
    },
    {
      "epoch": 0.5684499314128943,
      "grad_norm": 0.70703125,
      "learning_rate": 1.6304490107222867e-05,
      "loss": 0.2894,
      "step": 7511
    },
    {
      "epoch": 0.5685256137363417,
      "grad_norm": 0.76171875,
      "learning_rate": 1.6303565509671476e-05,
      "loss": 0.3215,
      "step": 7512
    },
    {
      "epoch": 0.568601296059789,
      "grad_norm": 0.80859375,
      "learning_rate": 1.630264082269319e-05,
      "loss": 0.3394,
      "step": 7513
    },
    {
      "epoch": 0.5686769783832364,
      "grad_norm": 0.71875,
      "learning_rate": 1.6301716046301125e-05,
      "loss": 0.2603,
      "step": 7514
    },
    {
      "epoch": 0.5687526607066837,
      "grad_norm": 0.75,
      "learning_rate": 1.630079118050841e-05,
      "loss": 0.3115,
      "step": 7515
    },
    {
      "epoch": 0.5688283430301311,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6299866225328155e-05,
      "loss": 0.3128,
      "step": 7516
    },
    {
      "epoch": 0.5689040253535783,
      "grad_norm": 0.7421875,
      "learning_rate": 1.6298941180773487e-05,
      "loss": 0.3077,
      "step": 7517
    },
    {
      "epoch": 0.5689797076770257,
      "grad_norm": 0.84375,
      "learning_rate": 1.6298016046857533e-05,
      "loss": 0.3382,
      "step": 7518
    },
    {
      "epoch": 0.569055390000473,
      "grad_norm": 0.78125,
      "learning_rate": 1.6297090823593407e-05,
      "loss": 0.2909,
      "step": 7519
    },
    {
      "epoch": 0.5691310723239204,
      "grad_norm": 0.83203125,
      "learning_rate": 1.6296165510994248e-05,
      "loss": 0.3018,
      "step": 7520
    },
    {
      "epoch": 0.5692067546473677,
      "grad_norm": 0.76953125,
      "learning_rate": 1.6295240109073173e-05,
      "loss": 0.3242,
      "step": 7521
    },
    {
      "epoch": 0.569282436970815,
      "grad_norm": 0.828125,
      "learning_rate": 1.6294314617843314e-05,
      "loss": 0.3682,
      "step": 7522
    },
    {
      "epoch": 0.5693581192942624,
      "grad_norm": 0.76171875,
      "learning_rate": 1.62933890373178e-05,
      "loss": 0.2974,
      "step": 7523
    },
    {
      "epoch": 0.5694338016177096,
      "grad_norm": 0.74609375,
      "learning_rate": 1.6292463367509767e-05,
      "loss": 0.2984,
      "step": 7524
    },
    {
      "epoch": 0.569509483941157,
      "grad_norm": 0.7890625,
      "learning_rate": 1.629153760843234e-05,
      "loss": 0.3385,
      "step": 7525
    },
    {
      "epoch": 0.5695851662646043,
      "grad_norm": 0.76171875,
      "learning_rate": 1.6290611760098655e-05,
      "loss": 0.3246,
      "step": 7526
    },
    {
      "epoch": 0.5696608485880517,
      "grad_norm": 0.765625,
      "learning_rate": 1.6289685822521847e-05,
      "loss": 0.2987,
      "step": 7527
    },
    {
      "epoch": 0.569736530911499,
      "grad_norm": 0.76953125,
      "learning_rate": 1.6288759795715053e-05,
      "loss": 0.3397,
      "step": 7528
    },
    {
      "epoch": 0.5698122132349464,
      "grad_norm": 0.77734375,
      "learning_rate": 1.628783367969141e-05,
      "loss": 0.3269,
      "step": 7529
    },
    {
      "epoch": 0.5698878955583937,
      "grad_norm": 0.8125,
      "learning_rate": 1.6286907474464056e-05,
      "loss": 0.3851,
      "step": 7530
    },
    {
      "epoch": 0.5699635778818409,
      "grad_norm": 0.81640625,
      "learning_rate": 1.6285981180046127e-05,
      "loss": 0.3231,
      "step": 7531
    },
    {
      "epoch": 0.5700392602052883,
      "grad_norm": 0.79296875,
      "learning_rate": 1.628505479645077e-05,
      "loss": 0.3378,
      "step": 7532
    },
    {
      "epoch": 0.5701149425287356,
      "grad_norm": 0.7734375,
      "learning_rate": 1.628412832369113e-05,
      "loss": 0.3279,
      "step": 7533
    },
    {
      "epoch": 0.570190624852183,
      "grad_norm": 0.828125,
      "learning_rate": 1.628320176178034e-05,
      "loss": 0.3378,
      "step": 7534
    },
    {
      "epoch": 0.5702663071756303,
      "grad_norm": 0.7421875,
      "learning_rate": 1.628227511073155e-05,
      "loss": 0.3064,
      "step": 7535
    },
    {
      "epoch": 0.5703419894990777,
      "grad_norm": 0.765625,
      "learning_rate": 1.628134837055791e-05,
      "loss": 0.3347,
      "step": 7536
    },
    {
      "epoch": 0.5704176718225249,
      "grad_norm": 0.703125,
      "learning_rate": 1.628042154127256e-05,
      "loss": 0.2786,
      "step": 7537
    },
    {
      "epoch": 0.5704933541459722,
      "grad_norm": 0.79296875,
      "learning_rate": 1.6279494622888662e-05,
      "loss": 0.3325,
      "step": 7538
    },
    {
      "epoch": 0.5705690364694196,
      "grad_norm": 0.765625,
      "learning_rate": 1.627856761541935e-05,
      "loss": 0.2939,
      "step": 7539
    },
    {
      "epoch": 0.5706447187928669,
      "grad_norm": 0.6875,
      "learning_rate": 1.627764051887778e-05,
      "loss": 0.2827,
      "step": 7540
    },
    {
      "epoch": 0.5707204011163143,
      "grad_norm": 0.82421875,
      "learning_rate": 1.6276713333277112e-05,
      "loss": 0.3593,
      "step": 7541
    },
    {
      "epoch": 0.5707960834397616,
      "grad_norm": 0.73046875,
      "learning_rate": 1.6275786058630488e-05,
      "loss": 0.3018,
      "step": 7542
    },
    {
      "epoch": 0.570871765763209,
      "grad_norm": 0.78125,
      "learning_rate": 1.6274858694951076e-05,
      "loss": 0.3475,
      "step": 7543
    },
    {
      "epoch": 0.5709474480866562,
      "grad_norm": 0.79296875,
      "learning_rate": 1.6273931242252024e-05,
      "loss": 0.3579,
      "step": 7544
    },
    {
      "epoch": 0.5710231304101036,
      "grad_norm": 0.765625,
      "learning_rate": 1.6273003700546486e-05,
      "loss": 0.3327,
      "step": 7545
    },
    {
      "epoch": 0.5710988127335509,
      "grad_norm": 0.79296875,
      "learning_rate": 1.627207606984763e-05,
      "loss": 0.3304,
      "step": 7546
    },
    {
      "epoch": 0.5711744950569982,
      "grad_norm": 0.76953125,
      "learning_rate": 1.627114835016861e-05,
      "loss": 0.3181,
      "step": 7547
    },
    {
      "epoch": 0.5712501773804456,
      "grad_norm": 0.828125,
      "learning_rate": 1.627022054152259e-05,
      "loss": 0.3346,
      "step": 7548
    },
    {
      "epoch": 0.5713258597038929,
      "grad_norm": 0.74609375,
      "learning_rate": 1.6269292643922732e-05,
      "loss": 0.3031,
      "step": 7549
    },
    {
      "epoch": 0.5714015420273403,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6268364657382196e-05,
      "loss": 0.3225,
      "step": 7550
    },
    {
      "epoch": 0.5714772243507875,
      "grad_norm": 0.78515625,
      "learning_rate": 1.626743658191415e-05,
      "loss": 0.3338,
      "step": 7551
    },
    {
      "epoch": 0.5715529066742349,
      "grad_norm": 0.82421875,
      "learning_rate": 1.626650841753177e-05,
      "loss": 0.3659,
      "step": 7552
    },
    {
      "epoch": 0.5716285889976822,
      "grad_norm": 0.75,
      "learning_rate": 1.6265580164248204e-05,
      "loss": 0.3196,
      "step": 7553
    },
    {
      "epoch": 0.5717042713211296,
      "grad_norm": 0.8046875,
      "learning_rate": 1.6264651822076636e-05,
      "loss": 0.3713,
      "step": 7554
    },
    {
      "epoch": 0.5717799536445769,
      "grad_norm": 0.82421875,
      "learning_rate": 1.626372339103023e-05,
      "loss": 0.3517,
      "step": 7555
    },
    {
      "epoch": 0.5718556359680242,
      "grad_norm": 0.71484375,
      "learning_rate": 1.6262794871122157e-05,
      "loss": 0.2839,
      "step": 7556
    },
    {
      "epoch": 0.5719313182914716,
      "grad_norm": 0.796875,
      "learning_rate": 1.6261866262365597e-05,
      "loss": 0.356,
      "step": 7557
    },
    {
      "epoch": 0.5720070006149188,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6260937564773715e-05,
      "loss": 0.3285,
      "step": 7558
    },
    {
      "epoch": 0.5720826829383662,
      "grad_norm": 0.81640625,
      "learning_rate": 1.626000877835969e-05,
      "loss": 0.3638,
      "step": 7559
    },
    {
      "epoch": 0.5721583652618135,
      "grad_norm": 0.76171875,
      "learning_rate": 1.6259079903136697e-05,
      "loss": 0.2975,
      "step": 7560
    },
    {
      "epoch": 0.5722340475852609,
      "grad_norm": 0.76171875,
      "learning_rate": 1.6258150939117917e-05,
      "loss": 0.3146,
      "step": 7561
    },
    {
      "epoch": 0.5723097299087082,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6257221886316524e-05,
      "loss": 0.3067,
      "step": 7562
    },
    {
      "epoch": 0.5723854122321556,
      "grad_norm": 0.7578125,
      "learning_rate": 1.62562927447457e-05,
      "loss": 0.3399,
      "step": 7563
    },
    {
      "epoch": 0.5724610945556029,
      "grad_norm": 0.79296875,
      "learning_rate": 1.625536351441863e-05,
      "loss": 0.3172,
      "step": 7564
    },
    {
      "epoch": 0.5725367768790501,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6254434195348492e-05,
      "loss": 0.3181,
      "step": 7565
    },
    {
      "epoch": 0.5726124592024975,
      "grad_norm": 0.8203125,
      "learning_rate": 1.625350478754847e-05,
      "loss": 0.3209,
      "step": 7566
    },
    {
      "epoch": 0.5726881415259448,
      "grad_norm": 0.796875,
      "learning_rate": 1.6252575291031755e-05,
      "loss": 0.3264,
      "step": 7567
    },
    {
      "epoch": 0.5727638238493922,
      "grad_norm": 1.078125,
      "learning_rate": 1.625164570581153e-05,
      "loss": 0.4296,
      "step": 7568
    },
    {
      "epoch": 0.5728395061728395,
      "grad_norm": 0.73046875,
      "learning_rate": 1.6250716031900983e-05,
      "loss": 0.3067,
      "step": 7569
    },
    {
      "epoch": 0.5729151884962869,
      "grad_norm": 0.796875,
      "learning_rate": 1.62497862693133e-05,
      "loss": 0.3527,
      "step": 7570
    },
    {
      "epoch": 0.5729908708197342,
      "grad_norm": 0.9140625,
      "learning_rate": 1.6248856418061672e-05,
      "loss": 0.3872,
      "step": 7571
    },
    {
      "epoch": 0.5730665531431814,
      "grad_norm": 0.80078125,
      "learning_rate": 1.6247926478159296e-05,
      "loss": 0.3753,
      "step": 7572
    },
    {
      "epoch": 0.5731422354666288,
      "grad_norm": 0.984375,
      "learning_rate": 1.624699644961936e-05,
      "loss": 0.3505,
      "step": 7573
    },
    {
      "epoch": 0.5732179177900761,
      "grad_norm": 0.79296875,
      "learning_rate": 1.624606633245506e-05,
      "loss": 0.333,
      "step": 7574
    },
    {
      "epoch": 0.5732936001135235,
      "grad_norm": 0.8046875,
      "learning_rate": 1.6245136126679584e-05,
      "loss": 0.3612,
      "step": 7575
    },
    {
      "epoch": 0.5733692824369708,
      "grad_norm": 0.8359375,
      "learning_rate": 1.6244205832306144e-05,
      "loss": 0.3409,
      "step": 7576
    },
    {
      "epoch": 0.5734449647604182,
      "grad_norm": 0.82421875,
      "learning_rate": 1.6243275449347925e-05,
      "loss": 0.3697,
      "step": 7577
    },
    {
      "epoch": 0.5735206470838655,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6242344977818133e-05,
      "loss": 0.2931,
      "step": 7578
    },
    {
      "epoch": 0.5735963294073128,
      "grad_norm": 0.7421875,
      "learning_rate": 1.624141441772996e-05,
      "loss": 0.325,
      "step": 7579
    },
    {
      "epoch": 0.5736720117307601,
      "grad_norm": 0.8125,
      "learning_rate": 1.6240483769096614e-05,
      "loss": 0.334,
      "step": 7580
    },
    {
      "epoch": 0.5737476940542074,
      "grad_norm": 0.74609375,
      "learning_rate": 1.62395530319313e-05,
      "loss": 0.2986,
      "step": 7581
    },
    {
      "epoch": 0.5738233763776548,
      "grad_norm": 0.734375,
      "learning_rate": 1.6238622206247218e-05,
      "loss": 0.2719,
      "step": 7582
    },
    {
      "epoch": 0.5738990587011021,
      "grad_norm": 0.71484375,
      "learning_rate": 1.623769129205757e-05,
      "loss": 0.2957,
      "step": 7583
    },
    {
      "epoch": 0.5739747410245495,
      "grad_norm": 0.7109375,
      "learning_rate": 1.623676028937557e-05,
      "loss": 0.2799,
      "step": 7584
    },
    {
      "epoch": 0.5740504233479968,
      "grad_norm": 0.80078125,
      "learning_rate": 1.6235829198214425e-05,
      "loss": 0.3245,
      "step": 7585
    },
    {
      "epoch": 0.5741261056714441,
      "grad_norm": 0.7265625,
      "learning_rate": 1.6234898018587336e-05,
      "loss": 0.2708,
      "step": 7586
    },
    {
      "epoch": 0.5742017879948914,
      "grad_norm": 0.71484375,
      "learning_rate": 1.6233966750507523e-05,
      "loss": 0.2749,
      "step": 7587
    },
    {
      "epoch": 0.5742774703183388,
      "grad_norm": 0.93359375,
      "learning_rate": 1.623303539398819e-05,
      "loss": 0.3281,
      "step": 7588
    },
    {
      "epoch": 0.5743531526417861,
      "grad_norm": 0.7265625,
      "learning_rate": 1.623210394904256e-05,
      "loss": 0.2977,
      "step": 7589
    },
    {
      "epoch": 0.5744288349652334,
      "grad_norm": 0.7890625,
      "learning_rate": 1.623117241568384e-05,
      "loss": 0.2938,
      "step": 7590
    },
    {
      "epoch": 0.5745045172886808,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6230240793925244e-05,
      "loss": 0.2953,
      "step": 7591
    },
    {
      "epoch": 0.5745801996121281,
      "grad_norm": 0.734375,
      "learning_rate": 1.6229309083779987e-05,
      "loss": 0.2783,
      "step": 7592
    },
    {
      "epoch": 0.5746558819355754,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6228377285261295e-05,
      "loss": 0.3131,
      "step": 7593
    },
    {
      "epoch": 0.5747315642590227,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6227445398382383e-05,
      "loss": 0.3179,
      "step": 7594
    },
    {
      "epoch": 0.5748072465824701,
      "grad_norm": 0.67578125,
      "learning_rate": 1.6226513423156468e-05,
      "loss": 0.2506,
      "step": 7595
    },
    {
      "epoch": 0.5748829289059174,
      "grad_norm": 0.859375,
      "learning_rate": 1.622558135959678e-05,
      "loss": 0.3553,
      "step": 7596
    },
    {
      "epoch": 0.5749586112293648,
      "grad_norm": 0.76953125,
      "learning_rate": 1.6224649207716536e-05,
      "loss": 0.322,
      "step": 7597
    },
    {
      "epoch": 0.5750342935528121,
      "grad_norm": 0.7421875,
      "learning_rate": 1.6223716967528956e-05,
      "loss": 0.281,
      "step": 7598
    },
    {
      "epoch": 0.5751099758762594,
      "grad_norm": 1.078125,
      "learning_rate": 1.6222784639047276e-05,
      "loss": 0.3287,
      "step": 7599
    },
    {
      "epoch": 0.5751856581997067,
      "grad_norm": 0.77734375,
      "learning_rate": 1.622185222228472e-05,
      "loss": 0.3434,
      "step": 7600
    },
    {
      "epoch": 0.575261340523154,
      "grad_norm": 0.72265625,
      "learning_rate": 1.6220919717254507e-05,
      "loss": 0.2923,
      "step": 7601
    },
    {
      "epoch": 0.5753370228466014,
      "grad_norm": 0.8203125,
      "learning_rate": 1.6219987123969874e-05,
      "loss": 0.3695,
      "step": 7602
    },
    {
      "epoch": 0.5754127051700487,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6219054442444048e-05,
      "loss": 0.3406,
      "step": 7603
    },
    {
      "epoch": 0.5754883874934961,
      "grad_norm": 1.109375,
      "learning_rate": 1.6218121672690267e-05,
      "loss": 0.406,
      "step": 7604
    },
    {
      "epoch": 0.5755640698169434,
      "grad_norm": 0.765625,
      "learning_rate": 1.6217188814721756e-05,
      "loss": 0.3299,
      "step": 7605
    },
    {
      "epoch": 0.5756397521403908,
      "grad_norm": 0.734375,
      "learning_rate": 1.621625586855175e-05,
      "loss": 0.3194,
      "step": 7606
    },
    {
      "epoch": 0.575715434463838,
      "grad_norm": 0.92578125,
      "learning_rate": 1.6215322834193494e-05,
      "loss": 0.3233,
      "step": 7607
    },
    {
      "epoch": 0.5757911167872853,
      "grad_norm": 0.73828125,
      "learning_rate": 1.621438971166021e-05,
      "loss": 0.2992,
      "step": 7608
    },
    {
      "epoch": 0.5758667991107327,
      "grad_norm": 0.74609375,
      "learning_rate": 1.621345650096515e-05,
      "loss": 0.3122,
      "step": 7609
    },
    {
      "epoch": 0.57594248143418,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6212523202121547e-05,
      "loss": 0.3506,
      "step": 7610
    },
    {
      "epoch": 0.5760181637576274,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6211589815142637e-05,
      "loss": 0.3342,
      "step": 7611
    },
    {
      "epoch": 0.5760938460810747,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6210656340041668e-05,
      "loss": 0.347,
      "step": 7612
    },
    {
      "epoch": 0.5761695284045221,
      "grad_norm": 0.75,
      "learning_rate": 1.620972277683188e-05,
      "loss": 0.295,
      "step": 7613
    },
    {
      "epoch": 0.5762452107279693,
      "grad_norm": 1.265625,
      "learning_rate": 1.620878912552652e-05,
      "loss": 0.3795,
      "step": 7614
    },
    {
      "epoch": 0.5763208930514166,
      "grad_norm": 0.73828125,
      "learning_rate": 1.6207855386138832e-05,
      "loss": 0.2755,
      "step": 7615
    },
    {
      "epoch": 0.576396575374864,
      "grad_norm": 0.71875,
      "learning_rate": 1.620692155868206e-05,
      "loss": 0.3011,
      "step": 7616
    },
    {
      "epoch": 0.5764722576983113,
      "grad_norm": 0.796875,
      "learning_rate": 1.6205987643169455e-05,
      "loss": 0.3041,
      "step": 7617
    },
    {
      "epoch": 0.5765479400217587,
      "grad_norm": 0.76171875,
      "learning_rate": 1.6205053639614267e-05,
      "loss": 0.3166,
      "step": 7618
    },
    {
      "epoch": 0.576623622345206,
      "grad_norm": 0.88671875,
      "learning_rate": 1.6204119548029746e-05,
      "loss": 0.3575,
      "step": 7619
    },
    {
      "epoch": 0.5766993046686534,
      "grad_norm": 0.76171875,
      "learning_rate": 1.6203185368429137e-05,
      "loss": 0.3189,
      "step": 7620
    },
    {
      "epoch": 0.5767749869921006,
      "grad_norm": 0.76953125,
      "learning_rate": 1.6202251100825705e-05,
      "loss": 0.3137,
      "step": 7621
    },
    {
      "epoch": 0.576850669315548,
      "grad_norm": 0.6796875,
      "learning_rate": 1.6201316745232693e-05,
      "loss": 0.2801,
      "step": 7622
    },
    {
      "epoch": 0.5769263516389953,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6200382301663363e-05,
      "loss": 0.2968,
      "step": 7623
    },
    {
      "epoch": 0.5770020339624427,
      "grad_norm": 0.78515625,
      "learning_rate": 1.619944777013097e-05,
      "loss": 0.3494,
      "step": 7624
    },
    {
      "epoch": 0.57707771628589,
      "grad_norm": 0.71875,
      "learning_rate": 1.619851315064877e-05,
      "loss": 0.2908,
      "step": 7625
    },
    {
      "epoch": 0.5771533986093373,
      "grad_norm": 0.78125,
      "learning_rate": 1.6197578443230026e-05,
      "loss": 0.3542,
      "step": 7626
    },
    {
      "epoch": 0.5772290809327847,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6196643647887994e-05,
      "loss": 0.2969,
      "step": 7627
    },
    {
      "epoch": 0.5773047632562319,
      "grad_norm": 0.71875,
      "learning_rate": 1.619570876463594e-05,
      "loss": 0.2855,
      "step": 7628
    },
    {
      "epoch": 0.5773804455796793,
      "grad_norm": 0.734375,
      "learning_rate": 1.6194773793487125e-05,
      "loss": 0.2893,
      "step": 7629
    },
    {
      "epoch": 0.5774561279031266,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6193838734454813e-05,
      "loss": 0.3128,
      "step": 7630
    },
    {
      "epoch": 0.577531810226574,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6192903587552268e-05,
      "loss": 0.3021,
      "step": 7631
    },
    {
      "epoch": 0.5776074925500213,
      "grad_norm": 1.0625,
      "learning_rate": 1.6191968352792762e-05,
      "loss": 0.376,
      "step": 7632
    },
    {
      "epoch": 0.5776831748734687,
      "grad_norm": 0.80078125,
      "learning_rate": 1.6191033030189554e-05,
      "loss": 0.3402,
      "step": 7633
    },
    {
      "epoch": 0.577758857196916,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6190097619755923e-05,
      "loss": 0.35,
      "step": 7634
    },
    {
      "epoch": 0.5778345395203632,
      "grad_norm": 0.828125,
      "learning_rate": 1.618916212150513e-05,
      "loss": 0.3743,
      "step": 7635
    },
    {
      "epoch": 0.5779102218438106,
      "grad_norm": 0.703125,
      "learning_rate": 1.6188226535450454e-05,
      "loss": 0.27,
      "step": 7636
    },
    {
      "epoch": 0.5779859041672579,
      "grad_norm": 0.7421875,
      "learning_rate": 1.6187290861605166e-05,
      "loss": 0.2945,
      "step": 7637
    },
    {
      "epoch": 0.5780615864907053,
      "grad_norm": 0.72265625,
      "learning_rate": 1.6186355099982537e-05,
      "loss": 0.2735,
      "step": 7638
    },
    {
      "epoch": 0.5781372688141526,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6185419250595847e-05,
      "loss": 0.3255,
      "step": 7639
    },
    {
      "epoch": 0.5782129511376,
      "grad_norm": 0.78125,
      "learning_rate": 1.618448331345837e-05,
      "loss": 0.3267,
      "step": 7640
    },
    {
      "epoch": 0.5782886334610473,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6183547288583384e-05,
      "loss": 0.3073,
      "step": 7641
    },
    {
      "epoch": 0.5783643157844945,
      "grad_norm": 0.76171875,
      "learning_rate": 1.618261117598417e-05,
      "loss": 0.3324,
      "step": 7642
    },
    {
      "epoch": 0.5784399981079419,
      "grad_norm": 0.796875,
      "learning_rate": 1.6181674975674003e-05,
      "loss": 0.3257,
      "step": 7643
    },
    {
      "epoch": 0.5785156804313892,
      "grad_norm": 0.71484375,
      "learning_rate": 1.618073868766617e-05,
      "loss": 0.2701,
      "step": 7644
    },
    {
      "epoch": 0.5785913627548366,
      "grad_norm": 0.73046875,
      "learning_rate": 1.6179802311973953e-05,
      "loss": 0.269,
      "step": 7645
    },
    {
      "epoch": 0.5786670450782839,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6178865848610637e-05,
      "loss": 0.3681,
      "step": 7646
    },
    {
      "epoch": 0.5787427274017313,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6177929297589504e-05,
      "loss": 0.3318,
      "step": 7647
    },
    {
      "epoch": 0.5788184097251786,
      "grad_norm": 0.76953125,
      "learning_rate": 1.617699265892384e-05,
      "loss": 0.275,
      "step": 7648
    },
    {
      "epoch": 0.5788940920486259,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6176055932626937e-05,
      "loss": 0.3498,
      "step": 7649
    },
    {
      "epoch": 0.5789697743720732,
      "grad_norm": 0.7734375,
      "learning_rate": 1.617511911871208e-05,
      "loss": 0.3306,
      "step": 7650
    },
    {
      "epoch": 0.5790454566955205,
      "grad_norm": 1.453125,
      "learning_rate": 1.6174182217192566e-05,
      "loss": 0.34,
      "step": 7651
    },
    {
      "epoch": 0.5791211390189679,
      "grad_norm": 0.78125,
      "learning_rate": 1.617324522808168e-05,
      "loss": 0.3322,
      "step": 7652
    },
    {
      "epoch": 0.5791968213424152,
      "grad_norm": 0.765625,
      "learning_rate": 1.6172308151392716e-05,
      "loss": 0.2752,
      "step": 7653
    },
    {
      "epoch": 0.5792725036658626,
      "grad_norm": 0.796875,
      "learning_rate": 1.6171370987138967e-05,
      "loss": 0.3338,
      "step": 7654
    },
    {
      "epoch": 0.5793481859893099,
      "grad_norm": 0.80859375,
      "learning_rate": 1.6170433735333734e-05,
      "loss": 0.3355,
      "step": 7655
    },
    {
      "epoch": 0.5794238683127572,
      "grad_norm": 0.78125,
      "learning_rate": 1.6169496395990306e-05,
      "loss": 0.3184,
      "step": 7656
    },
    {
      "epoch": 0.5794995506362045,
      "grad_norm": 0.77734375,
      "learning_rate": 1.6168558969121987e-05,
      "loss": 0.3351,
      "step": 7657
    },
    {
      "epoch": 0.5795752329596519,
      "grad_norm": 0.83203125,
      "learning_rate": 1.6167621454742074e-05,
      "loss": 0.3551,
      "step": 7658
    },
    {
      "epoch": 0.5796509152830992,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6166683852863864e-05,
      "loss": 0.3407,
      "step": 7659
    },
    {
      "epoch": 0.5797265976065465,
      "grad_norm": 0.78515625,
      "learning_rate": 1.616574616350066e-05,
      "loss": 0.2968,
      "step": 7660
    },
    {
      "epoch": 0.5798022799299939,
      "grad_norm": 0.78125,
      "learning_rate": 1.616480838666577e-05,
      "loss": 0.325,
      "step": 7661
    },
    {
      "epoch": 0.5798779622534411,
      "grad_norm": 0.765625,
      "learning_rate": 1.6163870522372494e-05,
      "loss": 0.2977,
      "step": 7662
    },
    {
      "epoch": 0.5799536445768885,
      "grad_norm": 0.734375,
      "learning_rate": 1.6162932570634135e-05,
      "loss": 0.2947,
      "step": 7663
    },
    {
      "epoch": 0.5800293269003358,
      "grad_norm": 0.71875,
      "learning_rate": 1.6161994531464e-05,
      "loss": 0.3007,
      "step": 7664
    },
    {
      "epoch": 0.5801050092237832,
      "grad_norm": 0.76953125,
      "learning_rate": 1.6161056404875403e-05,
      "loss": 0.3184,
      "step": 7665
    },
    {
      "epoch": 0.5801806915472305,
      "grad_norm": 0.78125,
      "learning_rate": 1.6160118190881643e-05,
      "loss": 0.3177,
      "step": 7666
    },
    {
      "epoch": 0.5802563738706779,
      "grad_norm": 0.828125,
      "learning_rate": 1.6159179889496036e-05,
      "loss": 0.3551,
      "step": 7667
    },
    {
      "epoch": 0.5803320561941252,
      "grad_norm": 0.765625,
      "learning_rate": 1.6158241500731895e-05,
      "loss": 0.3018,
      "step": 7668
    },
    {
      "epoch": 0.5804077385175724,
      "grad_norm": 0.80859375,
      "learning_rate": 1.615730302460253e-05,
      "loss": 0.3581,
      "step": 7669
    },
    {
      "epoch": 0.5804834208410198,
      "grad_norm": 0.6875,
      "learning_rate": 1.6156364461121255e-05,
      "loss": 0.2997,
      "step": 7670
    },
    {
      "epoch": 0.5805591031644671,
      "grad_norm": 0.765625,
      "learning_rate": 1.6155425810301383e-05,
      "loss": 0.2886,
      "step": 7671
    },
    {
      "epoch": 0.5806347854879145,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6154487072156237e-05,
      "loss": 0.2964,
      "step": 7672
    },
    {
      "epoch": 0.5807104678113618,
      "grad_norm": 0.75,
      "learning_rate": 1.615354824669913e-05,
      "loss": 0.3314,
      "step": 7673
    },
    {
      "epoch": 0.5807861501348092,
      "grad_norm": 0.82421875,
      "learning_rate": 1.615260933394338e-05,
      "loss": 0.289,
      "step": 7674
    },
    {
      "epoch": 0.5808618324582565,
      "grad_norm": 0.8359375,
      "learning_rate": 1.6151670333902307e-05,
      "loss": 0.3783,
      "step": 7675
    },
    {
      "epoch": 0.5809375147817037,
      "grad_norm": 0.77734375,
      "learning_rate": 1.6150731246589234e-05,
      "loss": 0.3488,
      "step": 7676
    },
    {
      "epoch": 0.5810131971051511,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6149792072017483e-05,
      "loss": 0.2878,
      "step": 7677
    },
    {
      "epoch": 0.5810888794285984,
      "grad_norm": 0.76171875,
      "learning_rate": 1.614885281020038e-05,
      "loss": 0.3206,
      "step": 7678
    },
    {
      "epoch": 0.5811645617520458,
      "grad_norm": 0.796875,
      "learning_rate": 1.614791346115125e-05,
      "loss": 0.3236,
      "step": 7679
    },
    {
      "epoch": 0.5812402440754931,
      "grad_norm": 0.91015625,
      "learning_rate": 1.6146974024883414e-05,
      "loss": 0.3892,
      "step": 7680
    },
    {
      "epoch": 0.5813159263989405,
      "grad_norm": 0.70703125,
      "learning_rate": 1.61460345014102e-05,
      "loss": 0.2732,
      "step": 7681
    },
    {
      "epoch": 0.5813916087223878,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6145094890744942e-05,
      "loss": 0.3339,
      "step": 7682
    },
    {
      "epoch": 0.581467291045835,
      "grad_norm": 0.7421875,
      "learning_rate": 1.614415519290097e-05,
      "loss": 0.3204,
      "step": 7683
    },
    {
      "epoch": 0.5815429733692824,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6143215407891615e-05,
      "loss": 0.3215,
      "step": 7684
    },
    {
      "epoch": 0.5816186556927297,
      "grad_norm": 0.7890625,
      "learning_rate": 1.61422755357302e-05,
      "loss": 0.3262,
      "step": 7685
    },
    {
      "epoch": 0.5816943380161771,
      "grad_norm": 0.8203125,
      "learning_rate": 1.6141335576430074e-05,
      "loss": 0.314,
      "step": 7686
    },
    {
      "epoch": 0.5817700203396244,
      "grad_norm": 0.7265625,
      "learning_rate": 1.6140395530004556e-05,
      "loss": 0.3032,
      "step": 7687
    },
    {
      "epoch": 0.5818457026630718,
      "grad_norm": 0.7265625,
      "learning_rate": 1.6139455396466995e-05,
      "loss": 0.2778,
      "step": 7688
    },
    {
      "epoch": 0.5819213849865191,
      "grad_norm": 0.74609375,
      "learning_rate": 1.6138515175830724e-05,
      "loss": 0.3383,
      "step": 7689
    },
    {
      "epoch": 0.5819970673099664,
      "grad_norm": 0.83984375,
      "learning_rate": 1.6137574868109077e-05,
      "loss": 0.3777,
      "step": 7690
    },
    {
      "epoch": 0.5820727496334137,
      "grad_norm": 1.09375,
      "learning_rate": 1.61366344733154e-05,
      "loss": 0.3788,
      "step": 7691
    },
    {
      "epoch": 0.5821484319568611,
      "grad_norm": 1.1640625,
      "learning_rate": 1.613569399146304e-05,
      "loss": 0.4059,
      "step": 7692
    },
    {
      "epoch": 0.5822241142803084,
      "grad_norm": 0.7421875,
      "learning_rate": 1.613475342256532e-05,
      "loss": 0.2922,
      "step": 7693
    },
    {
      "epoch": 0.5822997966037557,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6133812766635603e-05,
      "loss": 0.3319,
      "step": 7694
    },
    {
      "epoch": 0.5823754789272031,
      "grad_norm": 0.7109375,
      "learning_rate": 1.613287202368722e-05,
      "loss": 0.3009,
      "step": 7695
    },
    {
      "epoch": 0.5824511612506504,
      "grad_norm": 0.734375,
      "learning_rate": 1.6131931193733527e-05,
      "loss": 0.3043,
      "step": 7696
    },
    {
      "epoch": 0.5825268435740977,
      "grad_norm": 0.72265625,
      "learning_rate": 1.6130990276787864e-05,
      "loss": 0.3147,
      "step": 7697
    },
    {
      "epoch": 0.582602525897545,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6130049272863587e-05,
      "loss": 0.3034,
      "step": 7698
    },
    {
      "epoch": 0.5826782082209924,
      "grad_norm": 0.75,
      "learning_rate": 1.6129108181974037e-05,
      "loss": 0.318,
      "step": 7699
    },
    {
      "epoch": 0.5827538905444397,
      "grad_norm": 0.78125,
      "learning_rate": 1.612816700413257e-05,
      "loss": 0.3113,
      "step": 7700
    },
    {
      "epoch": 0.5828295728678871,
      "grad_norm": 0.80859375,
      "learning_rate": 1.612722573935254e-05,
      "loss": 0.3031,
      "step": 7701
    },
    {
      "epoch": 0.5829052551913344,
      "grad_norm": 0.7890625,
      "learning_rate": 1.61262843876473e-05,
      "loss": 0.3447,
      "step": 7702
    },
    {
      "epoch": 0.5829809375147817,
      "grad_norm": 0.765625,
      "learning_rate": 1.61253429490302e-05,
      "loss": 0.3141,
      "step": 7703
    },
    {
      "epoch": 0.583056619838229,
      "grad_norm": 0.76171875,
      "learning_rate": 1.61244014235146e-05,
      "loss": 0.3113,
      "step": 7704
    },
    {
      "epoch": 0.5831323021616763,
      "grad_norm": 0.75,
      "learning_rate": 1.6123459811113855e-05,
      "loss": 0.3229,
      "step": 7705
    },
    {
      "epoch": 0.5832079844851237,
      "grad_norm": 0.7265625,
      "learning_rate": 1.6122518111841322e-05,
      "loss": 0.2767,
      "step": 7706
    },
    {
      "epoch": 0.583283666808571,
      "grad_norm": 0.77734375,
      "learning_rate": 1.6121576325710368e-05,
      "loss": 0.3155,
      "step": 7707
    },
    {
      "epoch": 0.5833593491320184,
      "grad_norm": 0.8046875,
      "learning_rate": 1.6120634452734346e-05,
      "loss": 0.3326,
      "step": 7708
    },
    {
      "epoch": 0.5834350314554657,
      "grad_norm": 0.73828125,
      "learning_rate": 1.611969249292662e-05,
      "loss": 0.2819,
      "step": 7709
    },
    {
      "epoch": 0.5835107137789131,
      "grad_norm": 0.7109375,
      "learning_rate": 1.6118750446300558e-05,
      "loss": 0.2724,
      "step": 7710
    },
    {
      "epoch": 0.5835863961023603,
      "grad_norm": 0.7890625,
      "learning_rate": 1.6117808312869517e-05,
      "loss": 0.354,
      "step": 7711
    },
    {
      "epoch": 0.5836620784258076,
      "grad_norm": 0.7421875,
      "learning_rate": 1.6116866092646866e-05,
      "loss": 0.2959,
      "step": 7712
    },
    {
      "epoch": 0.583737760749255,
      "grad_norm": 0.73828125,
      "learning_rate": 1.6115923785645975e-05,
      "loss": 0.3118,
      "step": 7713
    },
    {
      "epoch": 0.5838134430727023,
      "grad_norm": 0.75,
      "learning_rate": 1.6114981391880212e-05,
      "loss": 0.319,
      "step": 7714
    },
    {
      "epoch": 0.5838891253961497,
      "grad_norm": 0.76171875,
      "learning_rate": 1.611403891136294e-05,
      "loss": 0.332,
      "step": 7715
    },
    {
      "epoch": 0.583964807719597,
      "grad_norm": 0.765625,
      "learning_rate": 1.6113096344107537e-05,
      "loss": 0.3199,
      "step": 7716
    },
    {
      "epoch": 0.5840404900430444,
      "grad_norm": 0.84375,
      "learning_rate": 1.611215369012737e-05,
      "loss": 0.3912,
      "step": 7717
    },
    {
      "epoch": 0.5841161723664916,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6111210949435815e-05,
      "loss": 0.3347,
      "step": 7718
    },
    {
      "epoch": 0.584191854689939,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6110268122046243e-05,
      "loss": 0.2954,
      "step": 7719
    },
    {
      "epoch": 0.5842675370133863,
      "grad_norm": 0.76953125,
      "learning_rate": 1.6109325207972037e-05,
      "loss": 0.321,
      "step": 7720
    },
    {
      "epoch": 0.5843432193368336,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6108382207226564e-05,
      "loss": 0.2867,
      "step": 7721
    },
    {
      "epoch": 0.584418901660281,
      "grad_norm": 0.76171875,
      "learning_rate": 1.610743911982321e-05,
      "loss": 0.3432,
      "step": 7722
    },
    {
      "epoch": 0.5844945839837283,
      "grad_norm": 0.7265625,
      "learning_rate": 1.610649594577535e-05,
      "loss": 0.272,
      "step": 7723
    },
    {
      "epoch": 0.5845702663071757,
      "grad_norm": 0.71875,
      "learning_rate": 1.6105552685096368e-05,
      "loss": 0.2909,
      "step": 7724
    },
    {
      "epoch": 0.5846459486306229,
      "grad_norm": 0.765625,
      "learning_rate": 1.610460933779964e-05,
      "loss": 0.3117,
      "step": 7725
    },
    {
      "epoch": 0.5847216309540703,
      "grad_norm": 0.80859375,
      "learning_rate": 1.6103665903898556e-05,
      "loss": 0.3215,
      "step": 7726
    },
    {
      "epoch": 0.5847973132775176,
      "grad_norm": 0.8046875,
      "learning_rate": 1.6102722383406497e-05,
      "loss": 0.3625,
      "step": 7727
    },
    {
      "epoch": 0.584872995600965,
      "grad_norm": 0.796875,
      "learning_rate": 1.6101778776336845e-05,
      "loss": 0.3497,
      "step": 7728
    },
    {
      "epoch": 0.5849486779244123,
      "grad_norm": 0.75390625,
      "learning_rate": 1.610083508270299e-05,
      "loss": 0.2695,
      "step": 7729
    },
    {
      "epoch": 0.5850243602478596,
      "grad_norm": 0.79296875,
      "learning_rate": 1.6099891302518326e-05,
      "loss": 0.3326,
      "step": 7730
    },
    {
      "epoch": 0.585100042571307,
      "grad_norm": 0.74609375,
      "learning_rate": 1.609894743579623e-05,
      "loss": 0.3134,
      "step": 7731
    },
    {
      "epoch": 0.5851757248947542,
      "grad_norm": 0.83203125,
      "learning_rate": 1.6098003482550102e-05,
      "loss": 0.3658,
      "step": 7732
    },
    {
      "epoch": 0.5852514072182016,
      "grad_norm": 0.7265625,
      "learning_rate": 1.6097059442793327e-05,
      "loss": 0.293,
      "step": 7733
    },
    {
      "epoch": 0.5853270895416489,
      "grad_norm": 0.81640625,
      "learning_rate": 1.6096115316539303e-05,
      "loss": 0.3336,
      "step": 7734
    },
    {
      "epoch": 0.5854027718650963,
      "grad_norm": 1.0859375,
      "learning_rate": 1.6095171103801423e-05,
      "loss": 0.3166,
      "step": 7735
    },
    {
      "epoch": 0.5854784541885436,
      "grad_norm": 0.81640625,
      "learning_rate": 1.609422680459308e-05,
      "loss": 0.3125,
      "step": 7736
    },
    {
      "epoch": 0.585554136511991,
      "grad_norm": 0.75390625,
      "learning_rate": 1.609328241892767e-05,
      "loss": 0.3291,
      "step": 7737
    },
    {
      "epoch": 0.5856298188354383,
      "grad_norm": 0.796875,
      "learning_rate": 1.6092337946818595e-05,
      "loss": 0.3379,
      "step": 7738
    },
    {
      "epoch": 0.5857055011588855,
      "grad_norm": 0.765625,
      "learning_rate": 1.609139338827925e-05,
      "loss": 0.3123,
      "step": 7739
    },
    {
      "epoch": 0.5857811834823329,
      "grad_norm": 0.78515625,
      "learning_rate": 1.609044874332304e-05,
      "loss": 0.3396,
      "step": 7740
    },
    {
      "epoch": 0.5858568658057802,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6089504011963358e-05,
      "loss": 0.3376,
      "step": 7741
    },
    {
      "epoch": 0.5859325481292276,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6088559194213615e-05,
      "loss": 0.3367,
      "step": 7742
    },
    {
      "epoch": 0.5860082304526749,
      "grad_norm": 0.80859375,
      "learning_rate": 1.608761429008721e-05,
      "loss": 0.3423,
      "step": 7743
    },
    {
      "epoch": 0.5860839127761223,
      "grad_norm": 0.703125,
      "learning_rate": 1.6086669299597547e-05,
      "loss": 0.2759,
      "step": 7744
    },
    {
      "epoch": 0.5861595950995696,
      "grad_norm": 0.73828125,
      "learning_rate": 1.608572422275804e-05,
      "loss": 0.3232,
      "step": 7745
    },
    {
      "epoch": 0.5862352774230168,
      "grad_norm": 0.7265625,
      "learning_rate": 1.6084779059582088e-05,
      "loss": 0.2811,
      "step": 7746
    },
    {
      "epoch": 0.5863109597464642,
      "grad_norm": 0.76953125,
      "learning_rate": 1.6083833810083104e-05,
      "loss": 0.3351,
      "step": 7747
    },
    {
      "epoch": 0.5863866420699115,
      "grad_norm": 0.81640625,
      "learning_rate": 1.6082888474274495e-05,
      "loss": 0.3289,
      "step": 7748
    },
    {
      "epoch": 0.5864623243933589,
      "grad_norm": 0.7578125,
      "learning_rate": 1.608194305216968e-05,
      "loss": 0.3155,
      "step": 7749
    },
    {
      "epoch": 0.5865380067168062,
      "grad_norm": 1.1875,
      "learning_rate": 1.6080997543782063e-05,
      "loss": 0.346,
      "step": 7750
    },
    {
      "epoch": 0.5866136890402536,
      "grad_norm": 0.7734375,
      "learning_rate": 1.608005194912506e-05,
      "loss": 0.332,
      "step": 7751
    },
    {
      "epoch": 0.5866893713637009,
      "grad_norm": 0.76953125,
      "learning_rate": 1.6079106268212085e-05,
      "loss": 0.3293,
      "step": 7752
    },
    {
      "epoch": 0.5867650536871482,
      "grad_norm": 0.77734375,
      "learning_rate": 1.607816050105656e-05,
      "loss": 0.3653,
      "step": 7753
    },
    {
      "epoch": 0.5868407360105955,
      "grad_norm": 0.8203125,
      "learning_rate": 1.6077214647671892e-05,
      "loss": 0.3601,
      "step": 7754
    },
    {
      "epoch": 0.5869164183340428,
      "grad_norm": 0.7421875,
      "learning_rate": 1.6076268708071507e-05,
      "loss": 0.308,
      "step": 7755
    },
    {
      "epoch": 0.5869921006574902,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6075322682268825e-05,
      "loss": 0.338,
      "step": 7756
    },
    {
      "epoch": 0.5870677829809375,
      "grad_norm": 0.73828125,
      "learning_rate": 1.6074376570277265e-05,
      "loss": 0.307,
      "step": 7757
    },
    {
      "epoch": 0.5871434653043849,
      "grad_norm": 0.80859375,
      "learning_rate": 1.607343037211025e-05,
      "loss": 0.3613,
      "step": 7758
    },
    {
      "epoch": 0.5872191476278322,
      "grad_norm": 0.72265625,
      "learning_rate": 1.60724840877812e-05,
      "loss": 0.2904,
      "step": 7759
    },
    {
      "epoch": 0.5872948299512795,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6071537717303544e-05,
      "loss": 0.3496,
      "step": 7760
    },
    {
      "epoch": 0.5873705122747268,
      "grad_norm": 0.73046875,
      "learning_rate": 1.6070591260690707e-05,
      "loss": 0.2785,
      "step": 7761
    },
    {
      "epoch": 0.5874461945981742,
      "grad_norm": 0.765625,
      "learning_rate": 1.6069644717956114e-05,
      "loss": 0.3218,
      "step": 7762
    },
    {
      "epoch": 0.5875218769216215,
      "grad_norm": 0.82421875,
      "learning_rate": 1.6068698089113198e-05,
      "loss": 0.3462,
      "step": 7763
    },
    {
      "epoch": 0.5875975592450688,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6067751374175383e-05,
      "loss": 0.278,
      "step": 7764
    },
    {
      "epoch": 0.5876732415685162,
      "grad_norm": 0.82421875,
      "learning_rate": 1.6066804573156104e-05,
      "loss": 0.3673,
      "step": 7765
    },
    {
      "epoch": 0.5877489238919635,
      "grad_norm": 0.69140625,
      "learning_rate": 1.6065857686068787e-05,
      "loss": 0.2554,
      "step": 7766
    },
    {
      "epoch": 0.5878246062154108,
      "grad_norm": 0.73046875,
      "learning_rate": 1.6064910712926873e-05,
      "loss": 0.3134,
      "step": 7767
    },
    {
      "epoch": 0.5879002885388581,
      "grad_norm": 0.828125,
      "learning_rate": 1.606396365374379e-05,
      "loss": 0.3324,
      "step": 7768
    },
    {
      "epoch": 0.5879759708623055,
      "grad_norm": 0.78125,
      "learning_rate": 1.606301650853298e-05,
      "loss": 0.2941,
      "step": 7769
    },
    {
      "epoch": 0.5880516531857528,
      "grad_norm": 0.74609375,
      "learning_rate": 1.6062069277307872e-05,
      "loss": 0.3052,
      "step": 7770
    },
    {
      "epoch": 0.5881273355092002,
      "grad_norm": 0.8203125,
      "learning_rate": 1.606112196008191e-05,
      "loss": 0.366,
      "step": 7771
    },
    {
      "epoch": 0.5882030178326475,
      "grad_norm": 0.78125,
      "learning_rate": 1.6060174556868534e-05,
      "loss": 0.3413,
      "step": 7772
    },
    {
      "epoch": 0.5882787001560948,
      "grad_norm": 0.7890625,
      "learning_rate": 1.605922706768118e-05,
      "loss": 0.3704,
      "step": 7773
    },
    {
      "epoch": 0.5883543824795421,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6058279492533292e-05,
      "loss": 0.3298,
      "step": 7774
    },
    {
      "epoch": 0.5884300648029894,
      "grad_norm": 0.7109375,
      "learning_rate": 1.6057331831438313e-05,
      "loss": 0.2948,
      "step": 7775
    },
    {
      "epoch": 0.5885057471264368,
      "grad_norm": 0.98046875,
      "learning_rate": 1.605638408440969e-05,
      "loss": 0.3429,
      "step": 7776
    },
    {
      "epoch": 0.5885814294498841,
      "grad_norm": 0.79296875,
      "learning_rate": 1.6055436251460864e-05,
      "loss": 0.347,
      "step": 7777
    },
    {
      "epoch": 0.5886571117733315,
      "grad_norm": 0.76953125,
      "learning_rate": 1.6054488332605282e-05,
      "loss": 0.3253,
      "step": 7778
    },
    {
      "epoch": 0.5887327940967788,
      "grad_norm": 1.265625,
      "learning_rate": 1.6053540327856395e-05,
      "loss": 0.3066,
      "step": 7779
    },
    {
      "epoch": 0.588808476420226,
      "grad_norm": 0.7890625,
      "learning_rate": 1.605259223722765e-05,
      "loss": 0.3391,
      "step": 7780
    },
    {
      "epoch": 0.5888841587436734,
      "grad_norm": 0.77734375,
      "learning_rate": 1.6051644060732497e-05,
      "loss": 0.2869,
      "step": 7781
    },
    {
      "epoch": 0.5889598410671207,
      "grad_norm": 0.84375,
      "learning_rate": 1.6050695798384387e-05,
      "loss": 0.3389,
      "step": 7782
    },
    {
      "epoch": 0.5890355233905681,
      "grad_norm": 0.7734375,
      "learning_rate": 1.6049747450196777e-05,
      "loss": 0.3148,
      "step": 7783
    },
    {
      "epoch": 0.5891112057140154,
      "grad_norm": 0.765625,
      "learning_rate": 1.6048799016183112e-05,
      "loss": 0.3114,
      "step": 7784
    },
    {
      "epoch": 0.5891868880374628,
      "grad_norm": 0.81640625,
      "learning_rate": 1.6047850496356858e-05,
      "loss": 0.3729,
      "step": 7785
    },
    {
      "epoch": 0.5892625703609101,
      "grad_norm": 0.74609375,
      "learning_rate": 1.6046901890731465e-05,
      "loss": 0.3053,
      "step": 7786
    },
    {
      "epoch": 0.5893382526843574,
      "grad_norm": 0.70703125,
      "learning_rate": 1.604595319932039e-05,
      "loss": 0.2894,
      "step": 7787
    },
    {
      "epoch": 0.5894139350078047,
      "grad_norm": 0.8359375,
      "learning_rate": 1.6045004422137095e-05,
      "loss": 0.3435,
      "step": 7788
    },
    {
      "epoch": 0.589489617331252,
      "grad_norm": 0.76953125,
      "learning_rate": 1.604405555919504e-05,
      "loss": 0.3372,
      "step": 7789
    },
    {
      "epoch": 0.5895652996546994,
      "grad_norm": 0.734375,
      "learning_rate": 1.6043106610507683e-05,
      "loss": 0.2987,
      "step": 7790
    },
    {
      "epoch": 0.5896409819781467,
      "grad_norm": 0.80859375,
      "learning_rate": 1.604215757608849e-05,
      "loss": 0.3196,
      "step": 7791
    },
    {
      "epoch": 0.5897166643015941,
      "grad_norm": 0.69921875,
      "learning_rate": 1.6041208455950924e-05,
      "loss": 0.2624,
      "step": 7792
    },
    {
      "epoch": 0.5897923466250414,
      "grad_norm": 0.765625,
      "learning_rate": 1.6040259250108444e-05,
      "loss": 0.3198,
      "step": 7793
    },
    {
      "epoch": 0.5898680289484887,
      "grad_norm": 0.796875,
      "learning_rate": 1.6039309958574524e-05,
      "loss": 0.3364,
      "step": 7794
    },
    {
      "epoch": 0.589943711271936,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6038360581362628e-05,
      "loss": 0.3146,
      "step": 7795
    },
    {
      "epoch": 0.5900193935953834,
      "grad_norm": 0.765625,
      "learning_rate": 1.6037411118486228e-05,
      "loss": 0.3284,
      "step": 7796
    },
    {
      "epoch": 0.5900950759188307,
      "grad_norm": 0.76171875,
      "learning_rate": 1.6036461569958785e-05,
      "loss": 0.3024,
      "step": 7797
    },
    {
      "epoch": 0.590170758242278,
      "grad_norm": 0.703125,
      "learning_rate": 1.6035511935793783e-05,
      "loss": 0.2613,
      "step": 7798
    },
    {
      "epoch": 0.5902464405657254,
      "grad_norm": 0.7265625,
      "learning_rate": 1.603456221600468e-05,
      "loss": 0.2747,
      "step": 7799
    },
    {
      "epoch": 0.5903221228891727,
      "grad_norm": 0.74609375,
      "learning_rate": 1.6033612410604962e-05,
      "loss": 0.3049,
      "step": 7800
    },
    {
      "epoch": 0.59039780521262,
      "grad_norm": 0.80859375,
      "learning_rate": 1.603266251960809e-05,
      "loss": 0.3579,
      "step": 7801
    },
    {
      "epoch": 0.5904734875360673,
      "grad_norm": 0.77734375,
      "learning_rate": 1.6031712543027554e-05,
      "loss": 0.3437,
      "step": 7802
    },
    {
      "epoch": 0.5905491698595147,
      "grad_norm": 1.0078125,
      "learning_rate": 1.603076248087683e-05,
      "loss": 0.3362,
      "step": 7803
    },
    {
      "epoch": 0.590624852182962,
      "grad_norm": 0.74609375,
      "learning_rate": 1.602981233316938e-05,
      "loss": 0.3138,
      "step": 7804
    },
    {
      "epoch": 0.5907005345064094,
      "grad_norm": 0.72265625,
      "learning_rate": 1.60288620999187e-05,
      "loss": 0.3038,
      "step": 7805
    },
    {
      "epoch": 0.5907762168298567,
      "grad_norm": 0.703125,
      "learning_rate": 1.6027911781138263e-05,
      "loss": 0.2841,
      "step": 7806
    },
    {
      "epoch": 0.590851899153304,
      "grad_norm": 0.76171875,
      "learning_rate": 1.6026961376841557e-05,
      "loss": 0.331,
      "step": 7807
    },
    {
      "epoch": 0.5909275814767513,
      "grad_norm": 0.76171875,
      "learning_rate": 1.6026010887042057e-05,
      "loss": 0.3334,
      "step": 7808
    },
    {
      "epoch": 0.5910032638001986,
      "grad_norm": 0.71484375,
      "learning_rate": 1.6025060311753252e-05,
      "loss": 0.2822,
      "step": 7809
    },
    {
      "epoch": 0.591078946123646,
      "grad_norm": 0.69921875,
      "learning_rate": 1.602410965098863e-05,
      "loss": 0.2793,
      "step": 7810
    },
    {
      "epoch": 0.5911546284470933,
      "grad_norm": 0.76171875,
      "learning_rate": 1.6023158904761676e-05,
      "loss": 0.3239,
      "step": 7811
    },
    {
      "epoch": 0.5912303107705407,
      "grad_norm": 0.84375,
      "learning_rate": 1.602220807308587e-05,
      "loss": 0.3728,
      "step": 7812
    },
    {
      "epoch": 0.591305993093988,
      "grad_norm": 0.765625,
      "learning_rate": 1.6021257155974718e-05,
      "loss": 0.3225,
      "step": 7813
    },
    {
      "epoch": 0.5913816754174354,
      "grad_norm": 0.78515625,
      "learning_rate": 1.6020306153441695e-05,
      "loss": 0.3172,
      "step": 7814
    },
    {
      "epoch": 0.5914573577408826,
      "grad_norm": 0.7421875,
      "learning_rate": 1.6019355065500293e-05,
      "loss": 0.3017,
      "step": 7815
    },
    {
      "epoch": 0.5915330400643299,
      "grad_norm": 0.75390625,
      "learning_rate": 1.6018403892164015e-05,
      "loss": 0.3364,
      "step": 7816
    },
    {
      "epoch": 0.5916087223877773,
      "grad_norm": 0.80078125,
      "learning_rate": 1.601745263344635e-05,
      "loss": 0.3535,
      "step": 7817
    },
    {
      "epoch": 0.5916844047112246,
      "grad_norm": 0.6640625,
      "learning_rate": 1.6016501289360794e-05,
      "loss": 0.2521,
      "step": 7818
    },
    {
      "epoch": 0.591760087034672,
      "grad_norm": 0.73046875,
      "learning_rate": 1.601554985992084e-05,
      "loss": 0.3011,
      "step": 7819
    },
    {
      "epoch": 0.5918357693581193,
      "grad_norm": 0.84375,
      "learning_rate": 1.6014598345139992e-05,
      "loss": 0.3248,
      "step": 7820
    },
    {
      "epoch": 0.5919114516815667,
      "grad_norm": 0.76953125,
      "learning_rate": 1.601364674503174e-05,
      "loss": 0.3102,
      "step": 7821
    },
    {
      "epoch": 0.5919871340050139,
      "grad_norm": 0.78515625,
      "learning_rate": 1.601269505960959e-05,
      "loss": 0.359,
      "step": 7822
    },
    {
      "epoch": 0.5920628163284612,
      "grad_norm": 0.71875,
      "learning_rate": 1.6011743288887047e-05,
      "loss": 0.3075,
      "step": 7823
    },
    {
      "epoch": 0.5921384986519086,
      "grad_norm": 0.71875,
      "learning_rate": 1.6010791432877605e-05,
      "loss": 0.2893,
      "step": 7824
    },
    {
      "epoch": 0.5922141809753559,
      "grad_norm": 0.8125,
      "learning_rate": 1.6009839491594772e-05,
      "loss": 0.3375,
      "step": 7825
    },
    {
      "epoch": 0.5922898632988033,
      "grad_norm": 0.734375,
      "learning_rate": 1.6008887465052052e-05,
      "loss": 0.314,
      "step": 7826
    },
    {
      "epoch": 0.5923655456222506,
      "grad_norm": 0.75,
      "learning_rate": 1.6007935353262953e-05,
      "loss": 0.3138,
      "step": 7827
    },
    {
      "epoch": 0.592441227945698,
      "grad_norm": 1.0234375,
      "learning_rate": 1.600698315624098e-05,
      "loss": 0.3241,
      "step": 7828
    },
    {
      "epoch": 0.5925169102691452,
      "grad_norm": 0.73828125,
      "learning_rate": 1.6006030873999648e-05,
      "loss": 0.3297,
      "step": 7829
    },
    {
      "epoch": 0.5925925925925926,
      "grad_norm": 0.79296875,
      "learning_rate": 1.6005078506552455e-05,
      "loss": 0.3545,
      "step": 7830
    },
    {
      "epoch": 0.5926682749160399,
      "grad_norm": 0.71875,
      "learning_rate": 1.6004126053912918e-05,
      "loss": 0.297,
      "step": 7831
    },
    {
      "epoch": 0.5927439572394873,
      "grad_norm": 0.703125,
      "learning_rate": 1.600317351609455e-05,
      "loss": 0.273,
      "step": 7832
    },
    {
      "epoch": 0.5928196395629346,
      "grad_norm": 0.7265625,
      "learning_rate": 1.6002220893110867e-05,
      "loss": 0.2884,
      "step": 7833
    },
    {
      "epoch": 0.5928953218863819,
      "grad_norm": 1.2421875,
      "learning_rate": 1.600126818497538e-05,
      "loss": 0.3645,
      "step": 7834
    },
    {
      "epoch": 0.5929710042098293,
      "grad_norm": 0.7578125,
      "learning_rate": 1.6000315391701605e-05,
      "loss": 0.2869,
      "step": 7835
    },
    {
      "epoch": 0.5930466865332765,
      "grad_norm": 0.84375,
      "learning_rate": 1.5999362513303058e-05,
      "loss": 0.389,
      "step": 7836
    },
    {
      "epoch": 0.5931223688567239,
      "grad_norm": 0.75390625,
      "learning_rate": 1.599840954979326e-05,
      "loss": 0.2997,
      "step": 7837
    },
    {
      "epoch": 0.5931980511801712,
      "grad_norm": 0.85546875,
      "learning_rate": 1.599745650118573e-05,
      "loss": 0.3547,
      "step": 7838
    },
    {
      "epoch": 0.5932737335036186,
      "grad_norm": 0.7578125,
      "learning_rate": 1.5996503367493982e-05,
      "loss": 0.3322,
      "step": 7839
    },
    {
      "epoch": 0.5933494158270659,
      "grad_norm": 1.046875,
      "learning_rate": 1.5995550148731548e-05,
      "loss": 0.3467,
      "step": 7840
    },
    {
      "epoch": 0.5934250981505133,
      "grad_norm": 0.7109375,
      "learning_rate": 1.5994596844911946e-05,
      "loss": 0.2903,
      "step": 7841
    },
    {
      "epoch": 0.5935007804739606,
      "grad_norm": 0.8046875,
      "learning_rate": 1.5993643456048698e-05,
      "loss": 0.3494,
      "step": 7842
    },
    {
      "epoch": 0.5935764627974078,
      "grad_norm": 0.81640625,
      "learning_rate": 1.5992689982155333e-05,
      "loss": 0.341,
      "step": 7843
    },
    {
      "epoch": 0.5936521451208552,
      "grad_norm": 0.85546875,
      "learning_rate": 1.599173642324538e-05,
      "loss": 0.3812,
      "step": 7844
    },
    {
      "epoch": 0.5937278274443025,
      "grad_norm": 0.75,
      "learning_rate": 1.5990782779332362e-05,
      "loss": 0.3014,
      "step": 7845
    },
    {
      "epoch": 0.5938035097677499,
      "grad_norm": 0.7265625,
      "learning_rate": 1.5989829050429807e-05,
      "loss": 0.2993,
      "step": 7846
    },
    {
      "epoch": 0.5938791920911972,
      "grad_norm": 1.0234375,
      "learning_rate": 1.598887523655125e-05,
      "loss": 0.3447,
      "step": 7847
    },
    {
      "epoch": 0.5939548744146446,
      "grad_norm": 0.734375,
      "learning_rate": 1.5987921337710224e-05,
      "loss": 0.2805,
      "step": 7848
    },
    {
      "epoch": 0.5940305567380919,
      "grad_norm": 0.8046875,
      "learning_rate": 1.5986967353920253e-05,
      "loss": 0.3453,
      "step": 7849
    },
    {
      "epoch": 0.5941062390615391,
      "grad_norm": 0.83203125,
      "learning_rate": 1.5986013285194877e-05,
      "loss": 0.3693,
      "step": 7850
    },
    {
      "epoch": 0.5941819213849865,
      "grad_norm": 0.796875,
      "learning_rate": 1.598505913154763e-05,
      "loss": 0.3463,
      "step": 7851
    },
    {
      "epoch": 0.5942576037084338,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5984104892992053e-05,
      "loss": 0.3285,
      "step": 7852
    },
    {
      "epoch": 0.5943332860318812,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5983150569541674e-05,
      "loss": 0.318,
      "step": 7853
    },
    {
      "epoch": 0.5944089683553285,
      "grad_norm": 0.734375,
      "learning_rate": 1.5982196161210038e-05,
      "loss": 0.3109,
      "step": 7854
    },
    {
      "epoch": 0.5944846506787759,
      "grad_norm": 0.83203125,
      "learning_rate": 1.5981241668010685e-05,
      "loss": 0.3591,
      "step": 7855
    },
    {
      "epoch": 0.5945603330022232,
      "grad_norm": 0.74609375,
      "learning_rate": 1.5980287089957153e-05,
      "loss": 0.3049,
      "step": 7856
    },
    {
      "epoch": 0.5946360153256705,
      "grad_norm": 0.79296875,
      "learning_rate": 1.597933242706299e-05,
      "loss": 0.33,
      "step": 7857
    },
    {
      "epoch": 0.5947116976491178,
      "grad_norm": 0.7734375,
      "learning_rate": 1.597837767934173e-05,
      "loss": 0.3264,
      "step": 7858
    },
    {
      "epoch": 0.5947873799725651,
      "grad_norm": 0.796875,
      "learning_rate": 1.5977422846806926e-05,
      "loss": 0.3265,
      "step": 7859
    },
    {
      "epoch": 0.5948630622960125,
      "grad_norm": 0.82421875,
      "learning_rate": 1.597646792947212e-05,
      "loss": 0.3706,
      "step": 7860
    },
    {
      "epoch": 0.5949387446194598,
      "grad_norm": 0.79296875,
      "learning_rate": 1.5975512927350864e-05,
      "loss": 0.3351,
      "step": 7861
    },
    {
      "epoch": 0.5950144269429072,
      "grad_norm": 1.3046875,
      "learning_rate": 1.59745578404567e-05,
      "loss": 0.388,
      "step": 7862
    },
    {
      "epoch": 0.5950901092663545,
      "grad_norm": 0.7578125,
      "learning_rate": 1.597360266880318e-05,
      "loss": 0.2993,
      "step": 7863
    },
    {
      "epoch": 0.5951657915898018,
      "grad_norm": 0.72265625,
      "learning_rate": 1.5972647412403857e-05,
      "loss": 0.2986,
      "step": 7864
    },
    {
      "epoch": 0.5952414739132491,
      "grad_norm": 0.78125,
      "learning_rate": 1.597169207127228e-05,
      "loss": 0.34,
      "step": 7865
    },
    {
      "epoch": 0.5953171562366965,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5970736645422004e-05,
      "loss": 0.3178,
      "step": 7866
    },
    {
      "epoch": 0.5953928385601438,
      "grad_norm": 0.73828125,
      "learning_rate": 1.596978113486658e-05,
      "loss": 0.3087,
      "step": 7867
    },
    {
      "epoch": 0.5954685208835911,
      "grad_norm": 0.72265625,
      "learning_rate": 1.5968825539619568e-05,
      "loss": 0.2908,
      "step": 7868
    },
    {
      "epoch": 0.5955442032070385,
      "grad_norm": 0.80859375,
      "learning_rate": 1.5967869859694523e-05,
      "loss": 0.3721,
      "step": 7869
    },
    {
      "epoch": 0.5956198855304858,
      "grad_norm": 0.80859375,
      "learning_rate": 1.5966914095105005e-05,
      "loss": 0.3469,
      "step": 7870
    },
    {
      "epoch": 0.5956955678539331,
      "grad_norm": 0.765625,
      "learning_rate": 1.596595824586457e-05,
      "loss": 0.3352,
      "step": 7871
    },
    {
      "epoch": 0.5957712501773804,
      "grad_norm": 0.81640625,
      "learning_rate": 1.5965002311986774e-05,
      "loss": 0.3525,
      "step": 7872
    },
    {
      "epoch": 0.5958469325008278,
      "grad_norm": 0.75390625,
      "learning_rate": 1.596404629348519e-05,
      "loss": 0.317,
      "step": 7873
    },
    {
      "epoch": 0.5959226148242751,
      "grad_norm": 0.78125,
      "learning_rate": 1.5963090190373368e-05,
      "loss": 0.3383,
      "step": 7874
    },
    {
      "epoch": 0.5959982971477225,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5962134002664885e-05,
      "loss": 0.3152,
      "step": 7875
    },
    {
      "epoch": 0.5960739794711698,
      "grad_norm": 1.0625,
      "learning_rate": 1.5961177730373298e-05,
      "loss": 0.3683,
      "step": 7876
    },
    {
      "epoch": 0.5961496617946171,
      "grad_norm": 0.7890625,
      "learning_rate": 1.5960221373512176e-05,
      "loss": 0.3402,
      "step": 7877
    },
    {
      "epoch": 0.5962253441180644,
      "grad_norm": 0.83203125,
      "learning_rate": 1.5959264932095085e-05,
      "loss": 0.3928,
      "step": 7878
    },
    {
      "epoch": 0.5963010264415117,
      "grad_norm": 0.72265625,
      "learning_rate": 1.5958308406135593e-05,
      "loss": 0.2616,
      "step": 7879
    },
    {
      "epoch": 0.5963767087649591,
      "grad_norm": 0.80078125,
      "learning_rate": 1.5957351795647277e-05,
      "loss": 0.3143,
      "step": 7880
    },
    {
      "epoch": 0.5964523910884064,
      "grad_norm": 0.828125,
      "learning_rate": 1.5956395100643694e-05,
      "loss": 0.3579,
      "step": 7881
    },
    {
      "epoch": 0.5965280734118538,
      "grad_norm": 0.73828125,
      "learning_rate": 1.595543832113843e-05,
      "loss": 0.3238,
      "step": 7882
    },
    {
      "epoch": 0.5966037557353011,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5954481457145053e-05,
      "loss": 0.3085,
      "step": 7883
    },
    {
      "epoch": 0.5966794380587485,
      "grad_norm": 0.87109375,
      "learning_rate": 1.595352450867714e-05,
      "loss": 0.3829,
      "step": 7884
    },
    {
      "epoch": 0.5967551203821957,
      "grad_norm": 0.98828125,
      "learning_rate": 1.5952567475748262e-05,
      "loss": 0.3489,
      "step": 7885
    },
    {
      "epoch": 0.596830802705643,
      "grad_norm": 1.40625,
      "learning_rate": 1.5951610358372002e-05,
      "loss": 0.3756,
      "step": 7886
    },
    {
      "epoch": 0.5969064850290904,
      "grad_norm": 0.765625,
      "learning_rate": 1.5950653156561932e-05,
      "loss": 0.3151,
      "step": 7887
    },
    {
      "epoch": 0.5969821673525377,
      "grad_norm": 0.78125,
      "learning_rate": 1.594969587033164e-05,
      "loss": 0.3001,
      "step": 7888
    },
    {
      "epoch": 0.5970578496759851,
      "grad_norm": 0.765625,
      "learning_rate": 1.59487384996947e-05,
      "loss": 0.2883,
      "step": 7889
    },
    {
      "epoch": 0.5971335319994324,
      "grad_norm": 0.7890625,
      "learning_rate": 1.5947781044664696e-05,
      "loss": 0.3593,
      "step": 7890
    },
    {
      "epoch": 0.5972092143228798,
      "grad_norm": 0.7734375,
      "learning_rate": 1.594682350525521e-05,
      "loss": 0.3382,
      "step": 7891
    },
    {
      "epoch": 0.597284896646327,
      "grad_norm": 0.75,
      "learning_rate": 1.594586588147983e-05,
      "loss": 0.3021,
      "step": 7892
    },
    {
      "epoch": 0.5973605789697743,
      "grad_norm": 0.8359375,
      "learning_rate": 1.5944908173352132e-05,
      "loss": 0.3795,
      "step": 7893
    },
    {
      "epoch": 0.5974362612932217,
      "grad_norm": 0.8125,
      "learning_rate": 1.594395038088572e-05,
      "loss": 0.3429,
      "step": 7894
    },
    {
      "epoch": 0.597511943616669,
      "grad_norm": 0.734375,
      "learning_rate": 1.5942992504094167e-05,
      "loss": 0.3227,
      "step": 7895
    },
    {
      "epoch": 0.5975876259401164,
      "grad_norm": 0.71484375,
      "learning_rate": 1.5942034542991068e-05,
      "loss": 0.2759,
      "step": 7896
    },
    {
      "epoch": 0.5976633082635637,
      "grad_norm": 0.7578125,
      "learning_rate": 1.594107649759001e-05,
      "loss": 0.3235,
      "step": 7897
    },
    {
      "epoch": 0.5977389905870111,
      "grad_norm": 0.84375,
      "learning_rate": 1.594011836790459e-05,
      "loss": 0.3336,
      "step": 7898
    },
    {
      "epoch": 0.5978146729104583,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5939160153948393e-05,
      "loss": 0.3385,
      "step": 7899
    },
    {
      "epoch": 0.5978903552339057,
      "grad_norm": 0.8046875,
      "learning_rate": 1.5938201855735017e-05,
      "loss": 0.3691,
      "step": 7900
    },
    {
      "epoch": 0.597966037557353,
      "grad_norm": 0.80078125,
      "learning_rate": 1.593724347327806e-05,
      "loss": 0.3453,
      "step": 7901
    },
    {
      "epoch": 0.5980417198808003,
      "grad_norm": 0.7890625,
      "learning_rate": 1.5936285006591113e-05,
      "loss": 0.3527,
      "step": 7902
    },
    {
      "epoch": 0.5981174022042477,
      "grad_norm": 0.74609375,
      "learning_rate": 1.5935326455687782e-05,
      "loss": 0.3142,
      "step": 7903
    },
    {
      "epoch": 0.598193084527695,
      "grad_norm": 0.74609375,
      "learning_rate": 1.5934367820581654e-05,
      "loss": 0.3218,
      "step": 7904
    },
    {
      "epoch": 0.5982687668511423,
      "grad_norm": 0.765625,
      "learning_rate": 1.593340910128634e-05,
      "loss": 0.337,
      "step": 7905
    },
    {
      "epoch": 0.5983444491745896,
      "grad_norm": 0.85546875,
      "learning_rate": 1.5932450297815433e-05,
      "loss": 0.3638,
      "step": 7906
    },
    {
      "epoch": 0.598420131498037,
      "grad_norm": 0.75,
      "learning_rate": 1.5931491410182535e-05,
      "loss": 0.3254,
      "step": 7907
    },
    {
      "epoch": 0.5984958138214843,
      "grad_norm": 0.7578125,
      "learning_rate": 1.5930532438401255e-05,
      "loss": 0.2852,
      "step": 7908
    },
    {
      "epoch": 0.5985714961449317,
      "grad_norm": 0.78515625,
      "learning_rate": 1.59295733824852e-05,
      "loss": 0.3312,
      "step": 7909
    },
    {
      "epoch": 0.598647178468379,
      "grad_norm": 0.78125,
      "learning_rate": 1.5928614242447965e-05,
      "loss": 0.3293,
      "step": 7910
    },
    {
      "epoch": 0.5987228607918264,
      "grad_norm": 0.75,
      "learning_rate": 1.5927655018303162e-05,
      "loss": 0.3138,
      "step": 7911
    },
    {
      "epoch": 0.5987985431152736,
      "grad_norm": 0.87890625,
      "learning_rate": 1.5926695710064404e-05,
      "loss": 0.3564,
      "step": 7912
    },
    {
      "epoch": 0.5988742254387209,
      "grad_norm": 0.78515625,
      "learning_rate": 1.5925736317745295e-05,
      "loss": 0.3221,
      "step": 7913
    },
    {
      "epoch": 0.5989499077621683,
      "grad_norm": 0.83203125,
      "learning_rate": 1.592477684135945e-05,
      "loss": 0.3907,
      "step": 7914
    },
    {
      "epoch": 0.5990255900856156,
      "grad_norm": 0.74609375,
      "learning_rate": 1.5923817280920477e-05,
      "loss": 0.288,
      "step": 7915
    },
    {
      "epoch": 0.599101272409063,
      "grad_norm": 0.7421875,
      "learning_rate": 1.592285763644199e-05,
      "loss": 0.3029,
      "step": 7916
    },
    {
      "epoch": 0.5991769547325103,
      "grad_norm": 0.80859375,
      "learning_rate": 1.5921897907937604e-05,
      "loss": 0.3684,
      "step": 7917
    },
    {
      "epoch": 0.5992526370559577,
      "grad_norm": 1.8828125,
      "learning_rate": 1.5920938095420934e-05,
      "loss": 0.4115,
      "step": 7918
    },
    {
      "epoch": 0.5993283193794049,
      "grad_norm": 0.796875,
      "learning_rate": 1.5919978198905593e-05,
      "loss": 0.3312,
      "step": 7919
    },
    {
      "epoch": 0.5994040017028522,
      "grad_norm": 0.75,
      "learning_rate": 1.5919018218405206e-05,
      "loss": 0.2818,
      "step": 7920
    },
    {
      "epoch": 0.5994796840262996,
      "grad_norm": 0.74609375,
      "learning_rate": 1.5918058153933386e-05,
      "loss": 0.2959,
      "step": 7921
    },
    {
      "epoch": 0.5995553663497469,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5917098005503757e-05,
      "loss": 0.3384,
      "step": 7922
    },
    {
      "epoch": 0.5996310486731943,
      "grad_norm": 0.7890625,
      "learning_rate": 1.591613777312994e-05,
      "loss": 0.3446,
      "step": 7923
    },
    {
      "epoch": 0.5997067309966416,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5915177456825553e-05,
      "loss": 0.3138,
      "step": 7924
    },
    {
      "epoch": 0.599782413320089,
      "grad_norm": 0.81640625,
      "learning_rate": 1.591421705660422e-05,
      "loss": 0.3223,
      "step": 7925
    },
    {
      "epoch": 0.5998580956435362,
      "grad_norm": 0.7890625,
      "learning_rate": 1.591325657247958e-05,
      "loss": 0.3192,
      "step": 7926
    },
    {
      "epoch": 0.5999337779669836,
      "grad_norm": 0.7421875,
      "learning_rate": 1.591229600446524e-05,
      "loss": 0.311,
      "step": 7927
    },
    {
      "epoch": 0.6000094602904309,
      "grad_norm": 0.72265625,
      "learning_rate": 1.5911335352574835e-05,
      "loss": 0.257,
      "step": 7928
    },
    {
      "epoch": 0.6000851426138782,
      "grad_norm": 0.75,
      "learning_rate": 1.5910374616821994e-05,
      "loss": 0.3007,
      "step": 7929
    },
    {
      "epoch": 0.6001608249373256,
      "grad_norm": 0.77734375,
      "learning_rate": 1.590941379722035e-05,
      "loss": 0.3066,
      "step": 7930
    },
    {
      "epoch": 0.6002365072607729,
      "grad_norm": 0.7265625,
      "learning_rate": 1.590845289378353e-05,
      "loss": 0.2896,
      "step": 7931
    },
    {
      "epoch": 0.6003121895842203,
      "grad_norm": 0.81640625,
      "learning_rate": 1.5907491906525162e-05,
      "loss": 0.3203,
      "step": 7932
    },
    {
      "epoch": 0.6003121895842203,
      "eval_loss": 0.33722907304763794,
      "eval_runtime": 83.3824,
      "eval_samples_per_second": 58.31,
      "eval_steps_per_second": 58.31,
      "step": 7932
    },
    {
      "epoch": 0.6003878719076675,
      "grad_norm": 1.8125,
      "learning_rate": 1.5906530835458885e-05,
      "loss": 0.3822,
      "step": 7933
    },
    {
      "epoch": 0.6004635542311149,
      "grad_norm": 0.7265625,
      "learning_rate": 1.5905569680598333e-05,
      "loss": 0.2781,
      "step": 7934
    },
    {
      "epoch": 0.6005392365545622,
      "grad_norm": 0.8046875,
      "learning_rate": 1.5904608441957142e-05,
      "loss": 0.3457,
      "step": 7935
    },
    {
      "epoch": 0.6006149188780096,
      "grad_norm": 0.84375,
      "learning_rate": 1.590364711954895e-05,
      "loss": 0.3088,
      "step": 7936
    },
    {
      "epoch": 0.6006906012014569,
      "grad_norm": 0.7890625,
      "learning_rate": 1.5902685713387386e-05,
      "loss": 0.3631,
      "step": 7937
    },
    {
      "epoch": 0.6007662835249042,
      "grad_norm": 0.7890625,
      "learning_rate": 1.5901724223486102e-05,
      "loss": 0.3528,
      "step": 7938
    },
    {
      "epoch": 0.6008419658483516,
      "grad_norm": 0.765625,
      "learning_rate": 1.590076264985873e-05,
      "loss": 0.3206,
      "step": 7939
    },
    {
      "epoch": 0.6009176481717988,
      "grad_norm": 0.8125,
      "learning_rate": 1.5899800992518913e-05,
      "loss": 0.3716,
      "step": 7940
    },
    {
      "epoch": 0.6009933304952462,
      "grad_norm": 0.79296875,
      "learning_rate": 1.5898839251480295e-05,
      "loss": 0.3778,
      "step": 7941
    },
    {
      "epoch": 0.6010690128186935,
      "grad_norm": 0.8359375,
      "learning_rate": 1.589787742675652e-05,
      "loss": 0.3754,
      "step": 7942
    },
    {
      "epoch": 0.6011446951421409,
      "grad_norm": 0.72265625,
      "learning_rate": 1.5896915518361227e-05,
      "loss": 0.288,
      "step": 7943
    },
    {
      "epoch": 0.6012203774655882,
      "grad_norm": 0.7890625,
      "learning_rate": 1.5895953526308075e-05,
      "loss": 0.3416,
      "step": 7944
    },
    {
      "epoch": 0.6012960597890356,
      "grad_norm": 0.7109375,
      "learning_rate": 1.5894991450610704e-05,
      "loss": 0.2946,
      "step": 7945
    },
    {
      "epoch": 0.6013717421124829,
      "grad_norm": 0.7109375,
      "learning_rate": 1.589402929128276e-05,
      "loss": 0.2909,
      "step": 7946
    },
    {
      "epoch": 0.6014474244359301,
      "grad_norm": 0.85546875,
      "learning_rate": 1.58930670483379e-05,
      "loss": 0.3363,
      "step": 7947
    },
    {
      "epoch": 0.6015231067593775,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5892104721789765e-05,
      "loss": 0.3052,
      "step": 7948
    },
    {
      "epoch": 0.6015987890828248,
      "grad_norm": 0.8203125,
      "learning_rate": 1.5891142311652016e-05,
      "loss": 0.3183,
      "step": 7949
    },
    {
      "epoch": 0.6016744714062722,
      "grad_norm": 1.3203125,
      "learning_rate": 1.5890179817938306e-05,
      "loss": 0.3539,
      "step": 7950
    },
    {
      "epoch": 0.6017501537297195,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5889217240662288e-05,
      "loss": 0.3054,
      "step": 7951
    },
    {
      "epoch": 0.6018258360531669,
      "grad_norm": 0.7109375,
      "learning_rate": 1.5888254579837613e-05,
      "loss": 0.2805,
      "step": 7952
    },
    {
      "epoch": 0.6019015183766142,
      "grad_norm": 0.8828125,
      "learning_rate": 1.5887291835477946e-05,
      "loss": 0.3041,
      "step": 7953
    },
    {
      "epoch": 0.6019772007000614,
      "grad_norm": 0.79296875,
      "learning_rate": 1.588632900759694e-05,
      "loss": 0.3727,
      "step": 7954
    },
    {
      "epoch": 0.6020528830235088,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5885366096208254e-05,
      "loss": 0.3318,
      "step": 7955
    },
    {
      "epoch": 0.6021285653469561,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5884403101325552e-05,
      "loss": 0.265,
      "step": 7956
    },
    {
      "epoch": 0.6022042476704035,
      "grad_norm": 0.81640625,
      "learning_rate": 1.5883440022962495e-05,
      "loss": 0.3778,
      "step": 7957
    },
    {
      "epoch": 0.6022799299938508,
      "grad_norm": 0.75390625,
      "learning_rate": 1.588247686113274e-05,
      "loss": 0.3059,
      "step": 7958
    },
    {
      "epoch": 0.6023556123172982,
      "grad_norm": 0.74609375,
      "learning_rate": 1.588151361584996e-05,
      "loss": 0.3177,
      "step": 7959
    },
    {
      "epoch": 0.6024312946407455,
      "grad_norm": 0.734375,
      "learning_rate": 1.588055028712782e-05,
      "loss": 0.2972,
      "step": 7960
    },
    {
      "epoch": 0.6025069769641928,
      "grad_norm": 0.75,
      "learning_rate": 1.5879586874979977e-05,
      "loss": 0.3133,
      "step": 7961
    },
    {
      "epoch": 0.6025826592876401,
      "grad_norm": 0.71875,
      "learning_rate": 1.5878623379420105e-05,
      "loss": 0.2864,
      "step": 7962
    },
    {
      "epoch": 0.6026583416110874,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5877659800461873e-05,
      "loss": 0.2989,
      "step": 7963
    },
    {
      "epoch": 0.6027340239345348,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5876696138118953e-05,
      "loss": 0.3218,
      "step": 7964
    },
    {
      "epoch": 0.6028097062579821,
      "grad_norm": 0.734375,
      "learning_rate": 1.587573239240501e-05,
      "loss": 0.3056,
      "step": 7965
    },
    {
      "epoch": 0.6028853885814295,
      "grad_norm": 0.78515625,
      "learning_rate": 1.5874768563333724e-05,
      "loss": 0.3481,
      "step": 7966
    },
    {
      "epoch": 0.6029610709048768,
      "grad_norm": 0.703125,
      "learning_rate": 1.587380465091876e-05,
      "loss": 0.2934,
      "step": 7967
    },
    {
      "epoch": 0.6030367532283241,
      "grad_norm": 0.77734375,
      "learning_rate": 1.58728406551738e-05,
      "loss": 0.3327,
      "step": 7968
    },
    {
      "epoch": 0.6031124355517714,
      "grad_norm": 0.82421875,
      "learning_rate": 1.5871876576112516e-05,
      "loss": 0.3375,
      "step": 7969
    },
    {
      "epoch": 0.6031881178752188,
      "grad_norm": 0.7578125,
      "learning_rate": 1.5870912413748585e-05,
      "loss": 0.3063,
      "step": 7970
    },
    {
      "epoch": 0.6032638001986661,
      "grad_norm": 0.77734375,
      "learning_rate": 1.5869948168095687e-05,
      "loss": 0.3108,
      "step": 7971
    },
    {
      "epoch": 0.6033394825221134,
      "grad_norm": 0.734375,
      "learning_rate": 1.58689838391675e-05,
      "loss": 0.2903,
      "step": 7972
    },
    {
      "epoch": 0.6034151648455608,
      "grad_norm": 1.140625,
      "learning_rate": 1.586801942697771e-05,
      "loss": 0.4094,
      "step": 7973
    },
    {
      "epoch": 0.6034908471690081,
      "grad_norm": 0.78515625,
      "learning_rate": 1.5867054931539995e-05,
      "loss": 0.2985,
      "step": 7974
    },
    {
      "epoch": 0.6035665294924554,
      "grad_norm": 0.8046875,
      "learning_rate": 1.5866090352868037e-05,
      "loss": 0.3565,
      "step": 7975
    },
    {
      "epoch": 0.6036422118159027,
      "grad_norm": 0.73046875,
      "learning_rate": 1.5865125690975512e-05,
      "loss": 0.3001,
      "step": 7976
    },
    {
      "epoch": 0.6037178941393501,
      "grad_norm": 0.765625,
      "learning_rate": 1.5864160945876122e-05,
      "loss": 0.3197,
      "step": 7977
    },
    {
      "epoch": 0.6037935764627974,
      "grad_norm": 0.87890625,
      "learning_rate": 1.586319611758355e-05,
      "loss": 0.3329,
      "step": 7978
    },
    {
      "epoch": 0.6038692587862448,
      "grad_norm": 0.76953125,
      "learning_rate": 1.586223120611147e-05,
      "loss": 0.3164,
      "step": 7979
    },
    {
      "epoch": 0.6039449411096921,
      "grad_norm": 0.7265625,
      "learning_rate": 1.5861266211473587e-05,
      "loss": 0.2804,
      "step": 7980
    },
    {
      "epoch": 0.6040206234331394,
      "grad_norm": 0.96875,
      "learning_rate": 1.586030113368358e-05,
      "loss": 0.3928,
      "step": 7981
    },
    {
      "epoch": 0.6040963057565867,
      "grad_norm": 0.75390625,
      "learning_rate": 1.585933597275515e-05,
      "loss": 0.3201,
      "step": 7982
    },
    {
      "epoch": 0.604171988080034,
      "grad_norm": 0.765625,
      "learning_rate": 1.585837072870198e-05,
      "loss": 0.3032,
      "step": 7983
    },
    {
      "epoch": 0.6042476704034814,
      "grad_norm": 0.70703125,
      "learning_rate": 1.585740540153777e-05,
      "loss": 0.2915,
      "step": 7984
    },
    {
      "epoch": 0.6043233527269287,
      "grad_norm": 1.3984375,
      "learning_rate": 1.585643999127621e-05,
      "loss": 0.3277,
      "step": 7985
    },
    {
      "epoch": 0.6043990350503761,
      "grad_norm": 0.69921875,
      "learning_rate": 1.5855474497930998e-05,
      "loss": 0.2739,
      "step": 7986
    },
    {
      "epoch": 0.6044747173738234,
      "grad_norm": 0.79296875,
      "learning_rate": 1.5854508921515838e-05,
      "loss": 0.3264,
      "step": 7987
    },
    {
      "epoch": 0.6045503996972708,
      "grad_norm": 0.6953125,
      "learning_rate": 1.585354326204442e-05,
      "loss": 0.2985,
      "step": 7988
    },
    {
      "epoch": 0.604626082020718,
      "grad_norm": 0.84765625,
      "learning_rate": 1.585257751953044e-05,
      "loss": 0.3854,
      "step": 7989
    },
    {
      "epoch": 0.6047017643441653,
      "grad_norm": 0.765625,
      "learning_rate": 1.5851611693987608e-05,
      "loss": 0.3134,
      "step": 7990
    },
    {
      "epoch": 0.6047774466676127,
      "grad_norm": 0.77734375,
      "learning_rate": 1.5850645785429627e-05,
      "loss": 0.2991,
      "step": 7991
    },
    {
      "epoch": 0.60485312899106,
      "grad_norm": 0.765625,
      "learning_rate": 1.5849679793870192e-05,
      "loss": 0.2965,
      "step": 7992
    },
    {
      "epoch": 0.6049288113145074,
      "grad_norm": 0.73046875,
      "learning_rate": 1.584871371932301e-05,
      "loss": 0.3307,
      "step": 7993
    },
    {
      "epoch": 0.6050044936379547,
      "grad_norm": 0.734375,
      "learning_rate": 1.5847747561801788e-05,
      "loss": 0.3202,
      "step": 7994
    },
    {
      "epoch": 0.6050801759614021,
      "grad_norm": 0.74609375,
      "learning_rate": 1.584678132132023e-05,
      "loss": 0.3159,
      "step": 7995
    },
    {
      "epoch": 0.6051558582848493,
      "grad_norm": 0.73046875,
      "learning_rate": 1.5845814997892048e-05,
      "loss": 0.312,
      "step": 7996
    },
    {
      "epoch": 0.6052315406082966,
      "grad_norm": 0.84765625,
      "learning_rate": 1.5844848591530947e-05,
      "loss": 0.332,
      "step": 7997
    },
    {
      "epoch": 0.605307222931744,
      "grad_norm": 0.8359375,
      "learning_rate": 1.584388210225064e-05,
      "loss": 0.3811,
      "step": 7998
    },
    {
      "epoch": 0.6053829052551913,
      "grad_norm": 0.78125,
      "learning_rate": 1.5842915530064834e-05,
      "loss": 0.2958,
      "step": 7999
    },
    {
      "epoch": 0.6054585875786387,
      "grad_norm": 0.765625,
      "learning_rate": 1.584194887498725e-05,
      "loss": 0.356,
      "step": 8000
    },
    {
      "epoch": 0.605534269902086,
      "grad_norm": 0.74609375,
      "learning_rate": 1.5840982137031595e-05,
      "loss": 0.3154,
      "step": 8001
    },
    {
      "epoch": 0.6056099522255334,
      "grad_norm": 0.75390625,
      "learning_rate": 1.584001531621158e-05,
      "loss": 0.3149,
      "step": 8002
    },
    {
      "epoch": 0.6056856345489806,
      "grad_norm": 0.7265625,
      "learning_rate": 1.5839048412540926e-05,
      "loss": 0.2808,
      "step": 8003
    },
    {
      "epoch": 0.605761316872428,
      "grad_norm": 0.73828125,
      "learning_rate": 1.5838081426033352e-05,
      "loss": 0.3064,
      "step": 8004
    },
    {
      "epoch": 0.6058369991958753,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5837114356702575e-05,
      "loss": 0.3052,
      "step": 8005
    },
    {
      "epoch": 0.6059126815193226,
      "grad_norm": 0.83984375,
      "learning_rate": 1.5836147204562317e-05,
      "loss": 0.2982,
      "step": 8006
    },
    {
      "epoch": 0.60598836384277,
      "grad_norm": 0.765625,
      "learning_rate": 1.5835179969626292e-05,
      "loss": 0.3163,
      "step": 8007
    },
    {
      "epoch": 0.6060640461662173,
      "grad_norm": 0.69921875,
      "learning_rate": 1.5834212651908226e-05,
      "loss": 0.279,
      "step": 8008
    },
    {
      "epoch": 0.6061397284896647,
      "grad_norm": 0.9921875,
      "learning_rate": 1.583324525142184e-05,
      "loss": 0.383,
      "step": 8009
    },
    {
      "epoch": 0.6062154108131119,
      "grad_norm": 0.83203125,
      "learning_rate": 1.5832277768180863e-05,
      "loss": 0.3459,
      "step": 8010
    },
    {
      "epoch": 0.6062910931365593,
      "grad_norm": 0.70703125,
      "learning_rate": 1.583131020219902e-05,
      "loss": 0.2911,
      "step": 8011
    },
    {
      "epoch": 0.6063667754600066,
      "grad_norm": 0.796875,
      "learning_rate": 1.583034255349003e-05,
      "loss": 0.3334,
      "step": 8012
    },
    {
      "epoch": 0.606442457783454,
      "grad_norm": 0.765625,
      "learning_rate": 1.5829374822067624e-05,
      "loss": 0.3191,
      "step": 8013
    },
    {
      "epoch": 0.6065181401069013,
      "grad_norm": 0.7890625,
      "learning_rate": 1.5828407007945537e-05,
      "loss": 0.3235,
      "step": 8014
    },
    {
      "epoch": 0.6065938224303487,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5827439111137492e-05,
      "loss": 0.3175,
      "step": 8015
    },
    {
      "epoch": 0.606669504753796,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5826471131657227e-05,
      "loss": 0.3295,
      "step": 8016
    },
    {
      "epoch": 0.6067451870772432,
      "grad_norm": 0.75,
      "learning_rate": 1.5825503069518465e-05,
      "loss": 0.3212,
      "step": 8017
    },
    {
      "epoch": 0.6068208694006906,
      "grad_norm": 0.859375,
      "learning_rate": 1.5824534924734946e-05,
      "loss": 0.3843,
      "step": 8018
    },
    {
      "epoch": 0.6068965517241379,
      "grad_norm": 0.83984375,
      "learning_rate": 1.5823566697320408e-05,
      "loss": 0.3452,
      "step": 8019
    },
    {
      "epoch": 0.6069722340475853,
      "grad_norm": 0.7421875,
      "learning_rate": 1.582259838728858e-05,
      "loss": 0.3043,
      "step": 8020
    },
    {
      "epoch": 0.6070479163710326,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5821629994653203e-05,
      "loss": 0.3222,
      "step": 8021
    },
    {
      "epoch": 0.60712359869448,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5820661519428012e-05,
      "loss": 0.3201,
      "step": 8022
    },
    {
      "epoch": 0.6071992810179272,
      "grad_norm": 0.703125,
      "learning_rate": 1.5819692961626752e-05,
      "loss": 0.3,
      "step": 8023
    },
    {
      "epoch": 0.6072749633413745,
      "grad_norm": 0.8125,
      "learning_rate": 1.5818724321263158e-05,
      "loss": 0.3597,
      "step": 8024
    },
    {
      "epoch": 0.6073506456648219,
      "grad_norm": 0.734375,
      "learning_rate": 1.5817755598350975e-05,
      "loss": 0.315,
      "step": 8025
    },
    {
      "epoch": 0.6074263279882692,
      "grad_norm": 0.734375,
      "learning_rate": 1.5816786792903943e-05,
      "loss": 0.3009,
      "step": 8026
    },
    {
      "epoch": 0.6075020103117166,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5815817904935813e-05,
      "loss": 0.3377,
      "step": 8027
    },
    {
      "epoch": 0.6075776926351639,
      "grad_norm": 0.77734375,
      "learning_rate": 1.581484893446032e-05,
      "loss": 0.2703,
      "step": 8028
    },
    {
      "epoch": 0.6076533749586113,
      "grad_norm": 0.80859375,
      "learning_rate": 1.5813879881491224e-05,
      "loss": 0.3426,
      "step": 8029
    },
    {
      "epoch": 0.6077290572820585,
      "grad_norm": 0.75390625,
      "learning_rate": 1.581291074604226e-05,
      "loss": 0.2977,
      "step": 8030
    },
    {
      "epoch": 0.6078047396055059,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5811941528127183e-05,
      "loss": 0.3269,
      "step": 8031
    },
    {
      "epoch": 0.6078804219289532,
      "grad_norm": 0.75,
      "learning_rate": 1.581097222775974e-05,
      "loss": 0.3126,
      "step": 8032
    },
    {
      "epoch": 0.6079561042524005,
      "grad_norm": 0.796875,
      "learning_rate": 1.5810002844953683e-05,
      "loss": 0.3269,
      "step": 8033
    },
    {
      "epoch": 0.6080317865758479,
      "grad_norm": 0.7421875,
      "learning_rate": 1.580903337972277e-05,
      "loss": 0.2931,
      "step": 8034
    },
    {
      "epoch": 0.6081074688992952,
      "grad_norm": 0.81640625,
      "learning_rate": 1.5808063832080746e-05,
      "loss": 0.3631,
      "step": 8035
    },
    {
      "epoch": 0.6081831512227426,
      "grad_norm": 0.74609375,
      "learning_rate": 1.580709420204137e-05,
      "loss": 0.3111,
      "step": 8036
    },
    {
      "epoch": 0.6082588335461898,
      "grad_norm": 0.75390625,
      "learning_rate": 1.58061244896184e-05,
      "loss": 0.2729,
      "step": 8037
    },
    {
      "epoch": 0.6083345158696372,
      "grad_norm": 0.765625,
      "learning_rate": 1.580515469482559e-05,
      "loss": 0.3544,
      "step": 8038
    },
    {
      "epoch": 0.6084101981930845,
      "grad_norm": 0.94921875,
      "learning_rate": 1.5804184817676696e-05,
      "loss": 0.3601,
      "step": 8039
    },
    {
      "epoch": 0.6084858805165319,
      "grad_norm": 0.765625,
      "learning_rate": 1.580321485818548e-05,
      "loss": 0.3287,
      "step": 8040
    },
    {
      "epoch": 0.6085615628399792,
      "grad_norm": 0.7890625,
      "learning_rate": 1.5802244816365702e-05,
      "loss": 0.307,
      "step": 8041
    },
    {
      "epoch": 0.6086372451634265,
      "grad_norm": 0.80078125,
      "learning_rate": 1.5801274692231126e-05,
      "loss": 0.343,
      "step": 8042
    },
    {
      "epoch": 0.6087129274868739,
      "grad_norm": 0.8046875,
      "learning_rate": 1.5800304485795516e-05,
      "loss": 0.3151,
      "step": 8043
    },
    {
      "epoch": 0.6087886098103211,
      "grad_norm": 0.7265625,
      "learning_rate": 1.579933419707263e-05,
      "loss": 0.2924,
      "step": 8044
    },
    {
      "epoch": 0.6088642921337685,
      "grad_norm": 0.8359375,
      "learning_rate": 1.5798363826076237e-05,
      "loss": 0.4142,
      "step": 8045
    },
    {
      "epoch": 0.6089399744572158,
      "grad_norm": 0.77734375,
      "learning_rate": 1.57973933728201e-05,
      "loss": 0.3061,
      "step": 8046
    },
    {
      "epoch": 0.6090156567806632,
      "grad_norm": 0.796875,
      "learning_rate": 1.5796422837317995e-05,
      "loss": 0.3382,
      "step": 8047
    },
    {
      "epoch": 0.6090913391041105,
      "grad_norm": 0.78125,
      "learning_rate": 1.5795452219583683e-05,
      "loss": 0.3271,
      "step": 8048
    },
    {
      "epoch": 0.6091670214275579,
      "grad_norm": 0.78515625,
      "learning_rate": 1.5794481519630936e-05,
      "loss": 0.3576,
      "step": 8049
    },
    {
      "epoch": 0.6092427037510052,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5793510737473523e-05,
      "loss": 0.319,
      "step": 8050
    },
    {
      "epoch": 0.6093183860744524,
      "grad_norm": 0.75390625,
      "learning_rate": 1.579253987312522e-05,
      "loss": 0.3277,
      "step": 8051
    },
    {
      "epoch": 0.6093940683978998,
      "grad_norm": 0.71875,
      "learning_rate": 1.5791568926599796e-05,
      "loss": 0.3052,
      "step": 8052
    },
    {
      "epoch": 0.6094697507213471,
      "grad_norm": 0.765625,
      "learning_rate": 1.579059789791103e-05,
      "loss": 0.3244,
      "step": 8053
    },
    {
      "epoch": 0.6095454330447945,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5789626787072698e-05,
      "loss": 0.3187,
      "step": 8054
    },
    {
      "epoch": 0.6096211153682418,
      "grad_norm": 0.83203125,
      "learning_rate": 1.5788655594098572e-05,
      "loss": 0.3781,
      "step": 8055
    },
    {
      "epoch": 0.6096967976916892,
      "grad_norm": 0.80859375,
      "learning_rate": 1.578768431900243e-05,
      "loss": 0.3013,
      "step": 8056
    },
    {
      "epoch": 0.6097724800151365,
      "grad_norm": 0.75390625,
      "learning_rate": 1.578671296179806e-05,
      "loss": 0.3477,
      "step": 8057
    },
    {
      "epoch": 0.6098481623385837,
      "grad_norm": 0.765625,
      "learning_rate": 1.5785741522499232e-05,
      "loss": 0.3345,
      "step": 8058
    },
    {
      "epoch": 0.6099238446620311,
      "grad_norm": 0.7578125,
      "learning_rate": 1.5784770001119733e-05,
      "loss": 0.3184,
      "step": 8059
    },
    {
      "epoch": 0.6099995269854784,
      "grad_norm": 0.82421875,
      "learning_rate": 1.5783798397673343e-05,
      "loss": 0.3571,
      "step": 8060
    },
    {
      "epoch": 0.6100752093089258,
      "grad_norm": 0.77734375,
      "learning_rate": 1.5782826712173852e-05,
      "loss": 0.3088,
      "step": 8061
    },
    {
      "epoch": 0.6101508916323731,
      "grad_norm": 0.8125,
      "learning_rate": 1.5781854944635036e-05,
      "loss": 0.3693,
      "step": 8062
    },
    {
      "epoch": 0.6102265739558205,
      "grad_norm": 0.73828125,
      "learning_rate": 1.5780883095070687e-05,
      "loss": 0.3086,
      "step": 8063
    },
    {
      "epoch": 0.6103022562792678,
      "grad_norm": 0.70703125,
      "learning_rate": 1.5779911163494592e-05,
      "loss": 0.2717,
      "step": 8064
    },
    {
      "epoch": 0.610377938602715,
      "grad_norm": 0.69921875,
      "learning_rate": 1.5778939149920538e-05,
      "loss": 0.2753,
      "step": 8065
    },
    {
      "epoch": 0.6104536209261624,
      "grad_norm": 0.90234375,
      "learning_rate": 1.5777967054362314e-05,
      "loss": 0.2985,
      "step": 8066
    },
    {
      "epoch": 0.6105293032496097,
      "grad_norm": 0.71484375,
      "learning_rate": 1.5776994876833713e-05,
      "loss": 0.3138,
      "step": 8067
    },
    {
      "epoch": 0.6106049855730571,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5776022617348525e-05,
      "loss": 0.2939,
      "step": 8068
    },
    {
      "epoch": 0.6106806678965044,
      "grad_norm": 1.015625,
      "learning_rate": 1.5775050275920546e-05,
      "loss": 0.3847,
      "step": 8069
    },
    {
      "epoch": 0.6107563502199518,
      "grad_norm": 0.75,
      "learning_rate": 1.5774077852563567e-05,
      "loss": 0.3199,
      "step": 8070
    },
    {
      "epoch": 0.6108320325433991,
      "grad_norm": 0.796875,
      "learning_rate": 1.5773105347291386e-05,
      "loss": 0.3179,
      "step": 8071
    },
    {
      "epoch": 0.6109077148668464,
      "grad_norm": 0.77734375,
      "learning_rate": 1.5772132760117797e-05,
      "loss": 0.2868,
      "step": 8072
    },
    {
      "epoch": 0.6109833971902937,
      "grad_norm": 0.75390625,
      "learning_rate": 1.57711600910566e-05,
      "loss": 0.3002,
      "step": 8073
    },
    {
      "epoch": 0.611059079513741,
      "grad_norm": 1.28125,
      "learning_rate": 1.5770187340121595e-05,
      "loss": 0.383,
      "step": 8074
    },
    {
      "epoch": 0.6111347618371884,
      "grad_norm": 0.7265625,
      "learning_rate": 1.576921450732658e-05,
      "loss": 0.2753,
      "step": 8075
    },
    {
      "epoch": 0.6112104441606357,
      "grad_norm": 0.7890625,
      "learning_rate": 1.5768241592685357e-05,
      "loss": 0.3512,
      "step": 8076
    },
    {
      "epoch": 0.6112861264840831,
      "grad_norm": 1.109375,
      "learning_rate": 1.576726859621173e-05,
      "loss": 0.3429,
      "step": 8077
    },
    {
      "epoch": 0.6113618088075304,
      "grad_norm": 0.75,
      "learning_rate": 1.5766295517919496e-05,
      "loss": 0.3218,
      "step": 8078
    },
    {
      "epoch": 0.6114374911309777,
      "grad_norm": 0.7265625,
      "learning_rate": 1.5765322357822465e-05,
      "loss": 0.2815,
      "step": 8079
    },
    {
      "epoch": 0.611513173454425,
      "grad_norm": 0.859375,
      "learning_rate": 1.5764349115934448e-05,
      "loss": 0.382,
      "step": 8080
    },
    {
      "epoch": 0.6115888557778724,
      "grad_norm": 0.73046875,
      "learning_rate": 1.5763375792269246e-05,
      "loss": 0.3087,
      "step": 8081
    },
    {
      "epoch": 0.6116645381013197,
      "grad_norm": 0.7890625,
      "learning_rate": 1.5762402386840664e-05,
      "loss": 0.3692,
      "step": 8082
    },
    {
      "epoch": 0.6117402204247671,
      "grad_norm": 0.77734375,
      "learning_rate": 1.5761428899662517e-05,
      "loss": 0.3311,
      "step": 8083
    },
    {
      "epoch": 0.6118159027482144,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5760455330748615e-05,
      "loss": 0.3217,
      "step": 8084
    },
    {
      "epoch": 0.6118915850716617,
      "grad_norm": 0.796875,
      "learning_rate": 1.575948168011277e-05,
      "loss": 0.3252,
      "step": 8085
    },
    {
      "epoch": 0.611967267395109,
      "grad_norm": 0.79296875,
      "learning_rate": 1.5758507947768794e-05,
      "loss": 0.34,
      "step": 8086
    },
    {
      "epoch": 0.6120429497185563,
      "grad_norm": 0.73828125,
      "learning_rate": 1.57575341337305e-05,
      "loss": 0.3113,
      "step": 8087
    },
    {
      "epoch": 0.6121186320420037,
      "grad_norm": 0.76171875,
      "learning_rate": 1.57565602380117e-05,
      "loss": 0.3358,
      "step": 8088
    },
    {
      "epoch": 0.612194314365451,
      "grad_norm": 0.73046875,
      "learning_rate": 1.5755586260626217e-05,
      "loss": 0.2981,
      "step": 8089
    },
    {
      "epoch": 0.6122699966888984,
      "grad_norm": 0.80078125,
      "learning_rate": 1.5754612201587868e-05,
      "loss": 0.3385,
      "step": 8090
    },
    {
      "epoch": 0.6123456790123457,
      "grad_norm": 0.7421875,
      "learning_rate": 1.575363806091047e-05,
      "loss": 0.3217,
      "step": 8091
    },
    {
      "epoch": 0.6124213613357931,
      "grad_norm": 0.7578125,
      "learning_rate": 1.575266383860784e-05,
      "loss": 0.3007,
      "step": 8092
    },
    {
      "epoch": 0.6124970436592403,
      "grad_norm": 0.8046875,
      "learning_rate": 1.57516895346938e-05,
      "loss": 0.3483,
      "step": 8093
    },
    {
      "epoch": 0.6125727259826876,
      "grad_norm": 0.75390625,
      "learning_rate": 1.575071514918218e-05,
      "loss": 0.3139,
      "step": 8094
    },
    {
      "epoch": 0.612648408306135,
      "grad_norm": 0.77734375,
      "learning_rate": 1.5749740682086795e-05,
      "loss": 0.3281,
      "step": 8095
    },
    {
      "epoch": 0.6127240906295823,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5748766133421466e-05,
      "loss": 0.309,
      "step": 8096
    },
    {
      "epoch": 0.6127997729530297,
      "grad_norm": 0.7265625,
      "learning_rate": 1.5747791503200028e-05,
      "loss": 0.2613,
      "step": 8097
    },
    {
      "epoch": 0.612875455276477,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5746816791436303e-05,
      "loss": 0.2917,
      "step": 8098
    },
    {
      "epoch": 0.6129511375999244,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5745841998144126e-05,
      "loss": 0.3109,
      "step": 8099
    },
    {
      "epoch": 0.6130268199233716,
      "grad_norm": 0.80078125,
      "learning_rate": 1.5744867123337313e-05,
      "loss": 0.3568,
      "step": 8100
    },
    {
      "epoch": 0.613102502246819,
      "grad_norm": 0.73828125,
      "learning_rate": 1.57438921670297e-05,
      "loss": 0.3233,
      "step": 8101
    },
    {
      "epoch": 0.6131781845702663,
      "grad_norm": 0.79296875,
      "learning_rate": 1.5742917129235125e-05,
      "loss": 0.3296,
      "step": 8102
    },
    {
      "epoch": 0.6132538668937136,
      "grad_norm": 0.8125,
      "learning_rate": 1.5741942009967414e-05,
      "loss": 0.3692,
      "step": 8103
    },
    {
      "epoch": 0.613329549217161,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5740966809240405e-05,
      "loss": 0.3219,
      "step": 8104
    },
    {
      "epoch": 0.6134052315406083,
      "grad_norm": 0.80859375,
      "learning_rate": 1.5739991527067922e-05,
      "loss": 0.3595,
      "step": 8105
    },
    {
      "epoch": 0.6134809138640557,
      "grad_norm": 0.6953125,
      "learning_rate": 1.5739016163463813e-05,
      "loss": 0.2839,
      "step": 8106
    },
    {
      "epoch": 0.6135565961875029,
      "grad_norm": 0.70703125,
      "learning_rate": 1.573804071844191e-05,
      "loss": 0.2999,
      "step": 8107
    },
    {
      "epoch": 0.6136322785109503,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5737065192016053e-05,
      "loss": 0.3402,
      "step": 8108
    },
    {
      "epoch": 0.6137079608343976,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5736089584200084e-05,
      "loss": 0.3043,
      "step": 8109
    },
    {
      "epoch": 0.613783643157845,
      "grad_norm": 0.78125,
      "learning_rate": 1.573511389500784e-05,
      "loss": 0.3518,
      "step": 8110
    },
    {
      "epoch": 0.6138593254812923,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5734138124453165e-05,
      "loss": 0.3142,
      "step": 8111
    },
    {
      "epoch": 0.6139350078047396,
      "grad_norm": 0.77734375,
      "learning_rate": 1.5733162272549896e-05,
      "loss": 0.3614,
      "step": 8112
    },
    {
      "epoch": 0.614010690128187,
      "grad_norm": 0.78125,
      "learning_rate": 1.5732186339311888e-05,
      "loss": 0.3297,
      "step": 8113
    },
    {
      "epoch": 0.6140863724516342,
      "grad_norm": 0.75,
      "learning_rate": 1.573121032475297e-05,
      "loss": 0.3287,
      "step": 8114
    },
    {
      "epoch": 0.6141620547750816,
      "grad_norm": 0.7578125,
      "learning_rate": 1.573023422888701e-05,
      "loss": 0.2913,
      "step": 8115
    },
    {
      "epoch": 0.6142377370985289,
      "grad_norm": 0.69921875,
      "learning_rate": 1.5729258051727835e-05,
      "loss": 0.2681,
      "step": 8116
    },
    {
      "epoch": 0.6143134194219763,
      "grad_norm": 1.15625,
      "learning_rate": 1.5728281793289308e-05,
      "loss": 0.3858,
      "step": 8117
    },
    {
      "epoch": 0.6143891017454236,
      "grad_norm": 0.8359375,
      "learning_rate": 1.5727305453585274e-05,
      "loss": 0.3986,
      "step": 8118
    },
    {
      "epoch": 0.614464784068871,
      "grad_norm": 0.80078125,
      "learning_rate": 1.5726329032629578e-05,
      "loss": 0.3181,
      "step": 8119
    },
    {
      "epoch": 0.6145404663923183,
      "grad_norm": 0.72265625,
      "learning_rate": 1.5725352530436083e-05,
      "loss": 0.2895,
      "step": 8120
    },
    {
      "epoch": 0.6146161487157655,
      "grad_norm": 0.765625,
      "learning_rate": 1.5724375947018637e-05,
      "loss": 0.3202,
      "step": 8121
    },
    {
      "epoch": 0.6146918310392129,
      "grad_norm": 0.7578125,
      "learning_rate": 1.572339928239109e-05,
      "loss": 0.3257,
      "step": 8122
    },
    {
      "epoch": 0.6147675133626602,
      "grad_norm": 0.71484375,
      "learning_rate": 1.572242253656731e-05,
      "loss": 0.3013,
      "step": 8123
    },
    {
      "epoch": 0.6148431956861076,
      "grad_norm": 0.73046875,
      "learning_rate": 1.572144570956114e-05,
      "loss": 0.2761,
      "step": 8124
    },
    {
      "epoch": 0.6149188780095549,
      "grad_norm": 0.71875,
      "learning_rate": 1.572046880138645e-05,
      "loss": 0.2872,
      "step": 8125
    },
    {
      "epoch": 0.6149945603330023,
      "grad_norm": 0.82421875,
      "learning_rate": 1.571949181205709e-05,
      "loss": 0.3552,
      "step": 8126
    },
    {
      "epoch": 0.6150702426564496,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5718514741586924e-05,
      "loss": 0.3085,
      "step": 8127
    },
    {
      "epoch": 0.6151459249798968,
      "grad_norm": 0.75,
      "learning_rate": 1.5717537589989813e-05,
      "loss": 0.3036,
      "step": 8128
    },
    {
      "epoch": 0.6152216073033442,
      "grad_norm": 1.1015625,
      "learning_rate": 1.571656035727962e-05,
      "loss": 0.3768,
      "step": 8129
    },
    {
      "epoch": 0.6152972896267915,
      "grad_norm": 0.765625,
      "learning_rate": 1.5715583043470204e-05,
      "loss": 0.2995,
      "step": 8130
    },
    {
      "epoch": 0.6153729719502389,
      "grad_norm": 0.77734375,
      "learning_rate": 1.571460564857544e-05,
      "loss": 0.3155,
      "step": 8131
    },
    {
      "epoch": 0.6154486542736862,
      "grad_norm": 0.80859375,
      "learning_rate": 1.571362817260919e-05,
      "loss": 0.3378,
      "step": 8132
    },
    {
      "epoch": 0.6155243365971336,
      "grad_norm": 0.83984375,
      "learning_rate": 1.5712650615585314e-05,
      "loss": 0.3128,
      "step": 8133
    },
    {
      "epoch": 0.6156000189205809,
      "grad_norm": 0.76171875,
      "learning_rate": 1.571167297751769e-05,
      "loss": 0.3248,
      "step": 8134
    },
    {
      "epoch": 0.6156757012440282,
      "grad_norm": 0.73046875,
      "learning_rate": 1.571069525842018e-05,
      "loss": 0.283,
      "step": 8135
    },
    {
      "epoch": 0.6157513835674755,
      "grad_norm": 0.875,
      "learning_rate": 1.5709717458306657e-05,
      "loss": 0.3628,
      "step": 8136
    },
    {
      "epoch": 0.6158270658909228,
      "grad_norm": 0.78515625,
      "learning_rate": 1.5708739577190996e-05,
      "loss": 0.308,
      "step": 8137
    },
    {
      "epoch": 0.6159027482143702,
      "grad_norm": 0.9296875,
      "learning_rate": 1.5707761615087068e-05,
      "loss": 0.3543,
      "step": 8138
    },
    {
      "epoch": 0.6159784305378175,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5706783572008746e-05,
      "loss": 0.3022,
      "step": 8139
    },
    {
      "epoch": 0.6160541128612649,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5705805447969903e-05,
      "loss": 0.3238,
      "step": 8140
    },
    {
      "epoch": 0.6161297951847122,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5704827242984425e-05,
      "loss": 0.3262,
      "step": 8141
    },
    {
      "epoch": 0.6162054775081595,
      "grad_norm": 0.765625,
      "learning_rate": 1.5703848957066178e-05,
      "loss": 0.3418,
      "step": 8142
    },
    {
      "epoch": 0.6162811598316068,
      "grad_norm": 1.0234375,
      "learning_rate": 1.5702870590229044e-05,
      "loss": 0.3551,
      "step": 8143
    },
    {
      "epoch": 0.6163568421550542,
      "grad_norm": 0.8125,
      "learning_rate": 1.5701892142486908e-05,
      "loss": 0.3383,
      "step": 8144
    },
    {
      "epoch": 0.6164325244785015,
      "grad_norm": 0.69921875,
      "learning_rate": 1.5700913613853647e-05,
      "loss": 0.2935,
      "step": 8145
    },
    {
      "epoch": 0.6165082068019488,
      "grad_norm": 0.7890625,
      "learning_rate": 1.569993500434314e-05,
      "loss": 0.3549,
      "step": 8146
    },
    {
      "epoch": 0.6165838891253962,
      "grad_norm": 0.80078125,
      "learning_rate": 1.5698956313969275e-05,
      "loss": 0.3072,
      "step": 8147
    },
    {
      "epoch": 0.6166595714488434,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5697977542745934e-05,
      "loss": 0.3051,
      "step": 8148
    },
    {
      "epoch": 0.6167352537722908,
      "grad_norm": 0.6640625,
      "learning_rate": 1.5696998690687006e-05,
      "loss": 0.2607,
      "step": 8149
    },
    {
      "epoch": 0.6168109360957381,
      "grad_norm": 0.796875,
      "learning_rate": 1.5696019757806373e-05,
      "loss": 0.2882,
      "step": 8150
    },
    {
      "epoch": 0.6168866184191855,
      "grad_norm": 0.78125,
      "learning_rate": 1.5695040744117927e-05,
      "loss": 0.3094,
      "step": 8151
    },
    {
      "epoch": 0.6169623007426328,
      "grad_norm": 0.7109375,
      "learning_rate": 1.5694061649635554e-05,
      "loss": 0.2969,
      "step": 8152
    },
    {
      "epoch": 0.6170379830660802,
      "grad_norm": 0.73046875,
      "learning_rate": 1.5693082474373146e-05,
      "loss": 0.3005,
      "step": 8153
    },
    {
      "epoch": 0.6171136653895275,
      "grad_norm": 0.796875,
      "learning_rate": 1.569210321834459e-05,
      "loss": 0.3524,
      "step": 8154
    },
    {
      "epoch": 0.6171893477129747,
      "grad_norm": 0.765625,
      "learning_rate": 1.5691123881563786e-05,
      "loss": 0.3524,
      "step": 8155
    },
    {
      "epoch": 0.6172650300364221,
      "grad_norm": 0.73828125,
      "learning_rate": 1.569014446404462e-05,
      "loss": 0.2983,
      "step": 8156
    },
    {
      "epoch": 0.6173407123598694,
      "grad_norm": 0.78515625,
      "learning_rate": 1.5689164965800994e-05,
      "loss": 0.3557,
      "step": 8157
    },
    {
      "epoch": 0.6174163946833168,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5688185386846793e-05,
      "loss": 0.3345,
      "step": 8158
    },
    {
      "epoch": 0.6174920770067641,
      "grad_norm": 0.80859375,
      "learning_rate": 1.568720572719593e-05,
      "loss": 0.3433,
      "step": 8159
    },
    {
      "epoch": 0.6175677593302115,
      "grad_norm": 0.79296875,
      "learning_rate": 1.5686225986862292e-05,
      "loss": 0.3159,
      "step": 8160
    },
    {
      "epoch": 0.6176434416536588,
      "grad_norm": 0.8125,
      "learning_rate": 1.5685246165859774e-05,
      "loss": 0.3477,
      "step": 8161
    },
    {
      "epoch": 0.617719123977106,
      "grad_norm": 0.734375,
      "learning_rate": 1.568426626420229e-05,
      "loss": 0.2963,
      "step": 8162
    },
    {
      "epoch": 0.6177948063005534,
      "grad_norm": 0.69921875,
      "learning_rate": 1.5683286281903733e-05,
      "loss": 0.2967,
      "step": 8163
    },
    {
      "epoch": 0.6178704886240007,
      "grad_norm": 0.73828125,
      "learning_rate": 1.5682306218978004e-05,
      "loss": 0.3198,
      "step": 8164
    },
    {
      "epoch": 0.6179461709474481,
      "grad_norm": 0.765625,
      "learning_rate": 1.5681326075439014e-05,
      "loss": 0.332,
      "step": 8165
    },
    {
      "epoch": 0.6180218532708954,
      "grad_norm": 0.734375,
      "learning_rate": 1.568034585130066e-05,
      "loss": 0.3077,
      "step": 8166
    },
    {
      "epoch": 0.6180975355943428,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5679365546576858e-05,
      "loss": 0.3273,
      "step": 8167
    },
    {
      "epoch": 0.6181732179177901,
      "grad_norm": 0.72265625,
      "learning_rate": 1.5678385161281505e-05,
      "loss": 0.2723,
      "step": 8168
    },
    {
      "epoch": 0.6182489002412374,
      "grad_norm": 0.8046875,
      "learning_rate": 1.5677404695428518e-05,
      "loss": 0.3334,
      "step": 8169
    },
    {
      "epoch": 0.6183245825646847,
      "grad_norm": 0.83203125,
      "learning_rate": 1.5676424149031798e-05,
      "loss": 0.3638,
      "step": 8170
    },
    {
      "epoch": 0.618400264888132,
      "grad_norm": 0.73046875,
      "learning_rate": 1.5675443522105264e-05,
      "loss": 0.3164,
      "step": 8171
    },
    {
      "epoch": 0.6184759472115794,
      "grad_norm": 0.671875,
      "learning_rate": 1.5674462814662824e-05,
      "loss": 0.2566,
      "step": 8172
    },
    {
      "epoch": 0.6185516295350267,
      "grad_norm": 0.78125,
      "learning_rate": 1.5673482026718386e-05,
      "loss": 0.3411,
      "step": 8173
    },
    {
      "epoch": 0.6186273118584741,
      "grad_norm": 0.8359375,
      "learning_rate": 1.5672501158285876e-05,
      "loss": 0.3576,
      "step": 8174
    },
    {
      "epoch": 0.6187029941819214,
      "grad_norm": 0.83984375,
      "learning_rate": 1.56715202093792e-05,
      "loss": 0.3874,
      "step": 8175
    },
    {
      "epoch": 0.6187786765053687,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5670539180012277e-05,
      "loss": 0.3006,
      "step": 8176
    },
    {
      "epoch": 0.618854358828816,
      "grad_norm": 0.796875,
      "learning_rate": 1.5669558070199026e-05,
      "loss": 0.3266,
      "step": 8177
    },
    {
      "epoch": 0.6189300411522634,
      "grad_norm": 0.765625,
      "learning_rate": 1.5668576879953363e-05,
      "loss": 0.275,
      "step": 8178
    },
    {
      "epoch": 0.6190057234757107,
      "grad_norm": 0.7734375,
      "learning_rate": 1.566759560928921e-05,
      "loss": 0.3012,
      "step": 8179
    },
    {
      "epoch": 0.619081405799158,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5666614258220486e-05,
      "loss": 0.3306,
      "step": 8180
    },
    {
      "epoch": 0.6191570881226054,
      "grad_norm": 0.80859375,
      "learning_rate": 1.5665632826761113e-05,
      "loss": 0.3297,
      "step": 8181
    },
    {
      "epoch": 0.6192327704460527,
      "grad_norm": 0.765625,
      "learning_rate": 1.5664651314925022e-05,
      "loss": 0.328,
      "step": 8182
    },
    {
      "epoch": 0.6193084527695,
      "grad_norm": 0.78125,
      "learning_rate": 1.5663669722726128e-05,
      "loss": 0.3146,
      "step": 8183
    },
    {
      "epoch": 0.6193841350929473,
      "grad_norm": 0.828125,
      "learning_rate": 1.566268805017836e-05,
      "loss": 0.3762,
      "step": 8184
    },
    {
      "epoch": 0.6194598174163947,
      "grad_norm": 0.7421875,
      "learning_rate": 1.566170629729564e-05,
      "loss": 0.3052,
      "step": 8185
    },
    {
      "epoch": 0.619535499739842,
      "grad_norm": 0.68359375,
      "learning_rate": 1.5660724464091906e-05,
      "loss": 0.2529,
      "step": 8186
    },
    {
      "epoch": 0.6196111820632894,
      "grad_norm": 0.7265625,
      "learning_rate": 1.5659742550581082e-05,
      "loss": 0.295,
      "step": 8187
    },
    {
      "epoch": 0.6196868643867367,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5658760556777095e-05,
      "loss": 0.3382,
      "step": 8188
    },
    {
      "epoch": 0.619762546710184,
      "grad_norm": 0.75,
      "learning_rate": 1.565777848269388e-05,
      "loss": 0.2976,
      "step": 8189
    },
    {
      "epoch": 0.6198382290336313,
      "grad_norm": 0.70703125,
      "learning_rate": 1.5656796328345368e-05,
      "loss": 0.2657,
      "step": 8190
    },
    {
      "epoch": 0.6199139113570786,
      "grad_norm": 0.73828125,
      "learning_rate": 1.565581409374549e-05,
      "loss": 0.3033,
      "step": 8191
    },
    {
      "epoch": 0.619989593680526,
      "grad_norm": 0.734375,
      "learning_rate": 1.565483177890818e-05,
      "loss": 0.3048,
      "step": 8192
    },
    {
      "epoch": 0.6200652760039733,
      "grad_norm": 0.765625,
      "learning_rate": 1.565384938384739e-05,
      "loss": 0.3337,
      "step": 8193
    },
    {
      "epoch": 0.6201409583274207,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5652866908577034e-05,
      "loss": 0.3446,
      "step": 8194
    },
    {
      "epoch": 0.620216640650868,
      "grad_norm": 0.796875,
      "learning_rate": 1.565188435311106e-05,
      "loss": 0.3657,
      "step": 8195
    },
    {
      "epoch": 0.6202923229743154,
      "grad_norm": 0.78515625,
      "learning_rate": 1.565090171746341e-05,
      "loss": 0.3161,
      "step": 8196
    },
    {
      "epoch": 0.6203680052977626,
      "grad_norm": 0.84375,
      "learning_rate": 1.5649919001648024e-05,
      "loss": 0.3658,
      "step": 8197
    },
    {
      "epoch": 0.6204436876212099,
      "grad_norm": 0.78515625,
      "learning_rate": 1.564893620567884e-05,
      "loss": 0.3418,
      "step": 8198
    },
    {
      "epoch": 0.6205193699446573,
      "grad_norm": 0.75390625,
      "learning_rate": 1.56479533295698e-05,
      "loss": 0.2622,
      "step": 8199
    },
    {
      "epoch": 0.6205950522681046,
      "grad_norm": 0.73046875,
      "learning_rate": 1.564697037333485e-05,
      "loss": 0.3031,
      "step": 8200
    },
    {
      "epoch": 0.620670734591552,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5645987336987938e-05,
      "loss": 0.2941,
      "step": 8201
    },
    {
      "epoch": 0.6207464169149993,
      "grad_norm": 0.7890625,
      "learning_rate": 1.5645004220543006e-05,
      "loss": 0.3525,
      "step": 8202
    },
    {
      "epoch": 0.6208220992384467,
      "grad_norm": 0.74609375,
      "learning_rate": 1.5644021024014e-05,
      "loss": 0.3229,
      "step": 8203
    },
    {
      "epoch": 0.6208977815618939,
      "grad_norm": 0.78515625,
      "learning_rate": 1.564303774741487e-05,
      "loss": 0.3295,
      "step": 8204
    },
    {
      "epoch": 0.6209734638853412,
      "grad_norm": 0.7734375,
      "learning_rate": 1.564205439075957e-05,
      "loss": 0.3279,
      "step": 8205
    },
    {
      "epoch": 0.6210491462087886,
      "grad_norm": 0.79296875,
      "learning_rate": 1.5641070954062038e-05,
      "loss": 0.3398,
      "step": 8206
    },
    {
      "epoch": 0.6211248285322359,
      "grad_norm": 0.84375,
      "learning_rate": 1.5640087437336244e-05,
      "loss": 0.3922,
      "step": 8207
    },
    {
      "epoch": 0.6212005108556833,
      "grad_norm": 0.7578125,
      "learning_rate": 1.5639103840596123e-05,
      "loss": 0.2885,
      "step": 8208
    },
    {
      "epoch": 0.6212761931791306,
      "grad_norm": 0.74609375,
      "learning_rate": 1.563812016385564e-05,
      "loss": 0.3065,
      "step": 8209
    },
    {
      "epoch": 0.621351875502578,
      "grad_norm": 0.796875,
      "learning_rate": 1.563713640712875e-05,
      "loss": 0.3499,
      "step": 8210
    },
    {
      "epoch": 0.6214275578260252,
      "grad_norm": 0.74609375,
      "learning_rate": 1.5636152570429405e-05,
      "loss": 0.2935,
      "step": 8211
    },
    {
      "epoch": 0.6215032401494726,
      "grad_norm": 0.859375,
      "learning_rate": 1.5635168653771564e-05,
      "loss": 0.3636,
      "step": 8212
    },
    {
      "epoch": 0.6215789224729199,
      "grad_norm": 0.73046875,
      "learning_rate": 1.563418465716918e-05,
      "loss": 0.295,
      "step": 8213
    },
    {
      "epoch": 0.6216546047963672,
      "grad_norm": 0.75390625,
      "learning_rate": 1.563320058063622e-05,
      "loss": 0.3386,
      "step": 8214
    },
    {
      "epoch": 0.6217302871198146,
      "grad_norm": 0.80859375,
      "learning_rate": 1.5632216424186646e-05,
      "loss": 0.3605,
      "step": 8215
    },
    {
      "epoch": 0.6218059694432619,
      "grad_norm": 0.7890625,
      "learning_rate": 1.5631232187834416e-05,
      "loss": 0.3226,
      "step": 8216
    },
    {
      "epoch": 0.6218816517667093,
      "grad_norm": 0.734375,
      "learning_rate": 1.5630247871593494e-05,
      "loss": 0.3294,
      "step": 8217
    },
    {
      "epoch": 0.6219573340901565,
      "grad_norm": 1.1328125,
      "learning_rate": 1.5629263475477844e-05,
      "loss": 0.2937,
      "step": 8218
    },
    {
      "epoch": 0.6220330164136039,
      "grad_norm": 0.71875,
      "learning_rate": 1.562827899950143e-05,
      "loss": 0.3,
      "step": 8219
    },
    {
      "epoch": 0.6221086987370512,
      "grad_norm": 0.8125,
      "learning_rate": 1.562729444367822e-05,
      "loss": 0.3279,
      "step": 8220
    },
    {
      "epoch": 0.6221843810604986,
      "grad_norm": 0.78515625,
      "learning_rate": 1.5626309808022185e-05,
      "loss": 0.3242,
      "step": 8221
    },
    {
      "epoch": 0.6222600633839459,
      "grad_norm": 0.72265625,
      "learning_rate": 1.5625325092547287e-05,
      "loss": 0.246,
      "step": 8222
    },
    {
      "epoch": 0.6223357457073933,
      "grad_norm": 0.78125,
      "learning_rate": 1.56243402972675e-05,
      "loss": 0.3591,
      "step": 8223
    },
    {
      "epoch": 0.6224114280308406,
      "grad_norm": 0.78515625,
      "learning_rate": 1.5623355422196797e-05,
      "loss": 0.288,
      "step": 8224
    },
    {
      "epoch": 0.6224871103542878,
      "grad_norm": 0.79296875,
      "learning_rate": 1.5622370467349144e-05,
      "loss": 0.2996,
      "step": 8225
    },
    {
      "epoch": 0.6225627926777352,
      "grad_norm": 1.0859375,
      "learning_rate": 1.5621385432738516e-05,
      "loss": 0.3212,
      "step": 8226
    },
    {
      "epoch": 0.6226384750011825,
      "grad_norm": 0.76953125,
      "learning_rate": 1.562040031837889e-05,
      "loss": 0.3182,
      "step": 8227
    },
    {
      "epoch": 0.6227141573246299,
      "grad_norm": 0.81640625,
      "learning_rate": 1.5619415124284242e-05,
      "loss": 0.3489,
      "step": 8228
    },
    {
      "epoch": 0.6227898396480772,
      "grad_norm": 0.75390625,
      "learning_rate": 1.561842985046855e-05,
      "loss": 0.2976,
      "step": 8229
    },
    {
      "epoch": 0.6228655219715246,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5617444496945785e-05,
      "loss": 0.3105,
      "step": 8230
    },
    {
      "epoch": 0.6229412042949719,
      "grad_norm": 0.82421875,
      "learning_rate": 1.5616459063729932e-05,
      "loss": 0.3693,
      "step": 8231
    },
    {
      "epoch": 0.6230168866184191,
      "grad_norm": 0.81640625,
      "learning_rate": 1.5615473550834964e-05,
      "loss": 0.2889,
      "step": 8232
    },
    {
      "epoch": 0.6230925689418665,
      "grad_norm": 1.5703125,
      "learning_rate": 1.561448795827487e-05,
      "loss": 0.3866,
      "step": 8233
    },
    {
      "epoch": 0.6231682512653138,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5613502286063634e-05,
      "loss": 0.3408,
      "step": 8234
    },
    {
      "epoch": 0.6232439335887612,
      "grad_norm": 0.84765625,
      "learning_rate": 1.561251653421523e-05,
      "loss": 0.3713,
      "step": 8235
    },
    {
      "epoch": 0.6233196159122085,
      "grad_norm": 0.78515625,
      "learning_rate": 1.5611530702743653e-05,
      "loss": 0.3308,
      "step": 8236
    },
    {
      "epoch": 0.6233952982356559,
      "grad_norm": 0.828125,
      "learning_rate": 1.5610544791662877e-05,
      "loss": 0.3291,
      "step": 8237
    },
    {
      "epoch": 0.6234709805591032,
      "grad_norm": 0.7890625,
      "learning_rate": 1.5609558800986893e-05,
      "loss": 0.3366,
      "step": 8238
    },
    {
      "epoch": 0.6235466628825505,
      "grad_norm": 0.71484375,
      "learning_rate": 1.56085727307297e-05,
      "loss": 0.2936,
      "step": 8239
    },
    {
      "epoch": 0.6236223452059978,
      "grad_norm": 0.828125,
      "learning_rate": 1.5607586580905273e-05,
      "loss": 0.338,
      "step": 8240
    },
    {
      "epoch": 0.6236980275294451,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5606600351527608e-05,
      "loss": 0.3164,
      "step": 8241
    },
    {
      "epoch": 0.6237737098528925,
      "grad_norm": 0.765625,
      "learning_rate": 1.5605614042610694e-05,
      "loss": 0.3118,
      "step": 8242
    },
    {
      "epoch": 0.6238493921763398,
      "grad_norm": 0.76953125,
      "learning_rate": 1.560462765416853e-05,
      "loss": 0.3149,
      "step": 8243
    },
    {
      "epoch": 0.6239250744997872,
      "grad_norm": 0.80859375,
      "learning_rate": 1.5603641186215104e-05,
      "loss": 0.3296,
      "step": 8244
    },
    {
      "epoch": 0.6240007568232345,
      "grad_norm": 0.796875,
      "learning_rate": 1.5602654638764408e-05,
      "loss": 0.3548,
      "step": 8245
    },
    {
      "epoch": 0.6240764391466818,
      "grad_norm": 0.78125,
      "learning_rate": 1.560166801183044e-05,
      "loss": 0.3159,
      "step": 8246
    },
    {
      "epoch": 0.6241521214701291,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5600681305427202e-05,
      "loss": 0.3279,
      "step": 8247
    },
    {
      "epoch": 0.6242278037935765,
      "grad_norm": 1.2265625,
      "learning_rate": 1.5599694519568688e-05,
      "loss": 0.3516,
      "step": 8248
    },
    {
      "epoch": 0.6243034861170238,
      "grad_norm": 0.765625,
      "learning_rate": 1.5598707654268902e-05,
      "loss": 0.2953,
      "step": 8249
    },
    {
      "epoch": 0.6243791684404711,
      "grad_norm": 0.87109375,
      "learning_rate": 1.5597720709541834e-05,
      "loss": 0.3678,
      "step": 8250
    },
    {
      "epoch": 0.6244548507639185,
      "grad_norm": 0.828125,
      "learning_rate": 1.5596733685401493e-05,
      "loss": 0.3401,
      "step": 8251
    },
    {
      "epoch": 0.6245305330873658,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5595746581861883e-05,
      "loss": 0.3187,
      "step": 8252
    },
    {
      "epoch": 0.6246062154108131,
      "grad_norm": 0.7890625,
      "learning_rate": 1.5594759398937004e-05,
      "loss": 0.3181,
      "step": 8253
    },
    {
      "epoch": 0.6246818977342604,
      "grad_norm": 0.7890625,
      "learning_rate": 1.559377213664086e-05,
      "loss": 0.3266,
      "step": 8254
    },
    {
      "epoch": 0.6247575800577078,
      "grad_norm": 0.8828125,
      "learning_rate": 1.559278479498746e-05,
      "loss": 0.3698,
      "step": 8255
    },
    {
      "epoch": 0.6248332623811551,
      "grad_norm": 0.71875,
      "learning_rate": 1.559179737399081e-05,
      "loss": 0.2833,
      "step": 8256
    },
    {
      "epoch": 0.6249089447046025,
      "grad_norm": 0.79296875,
      "learning_rate": 1.5590809873664918e-05,
      "loss": 0.3372,
      "step": 8257
    },
    {
      "epoch": 0.6249846270280498,
      "grad_norm": 0.85546875,
      "learning_rate": 1.55898222940238e-05,
      "loss": 0.3842,
      "step": 8258
    },
    {
      "epoch": 0.6250603093514971,
      "grad_norm": 0.79296875,
      "learning_rate": 1.558883463508145e-05,
      "loss": 0.3402,
      "step": 8259
    },
    {
      "epoch": 0.6251359916749444,
      "grad_norm": 0.67578125,
      "learning_rate": 1.5587846896851894e-05,
      "loss": 0.2715,
      "step": 8260
    },
    {
      "epoch": 0.6252116739983917,
      "grad_norm": 0.734375,
      "learning_rate": 1.558685907934914e-05,
      "loss": 0.3133,
      "step": 8261
    },
    {
      "epoch": 0.6252873563218391,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5585871182587208e-05,
      "loss": 0.3023,
      "step": 8262
    },
    {
      "epoch": 0.6253630386452864,
      "grad_norm": 0.78515625,
      "learning_rate": 1.5584883206580105e-05,
      "loss": 0.3638,
      "step": 8263
    },
    {
      "epoch": 0.6254387209687338,
      "grad_norm": 0.734375,
      "learning_rate": 1.5583895151341845e-05,
      "loss": 0.305,
      "step": 8264
    },
    {
      "epoch": 0.6255144032921811,
      "grad_norm": 0.890625,
      "learning_rate": 1.5582907016886454e-05,
      "loss": 0.3739,
      "step": 8265
    },
    {
      "epoch": 0.6255900856156285,
      "grad_norm": 0.71875,
      "learning_rate": 1.5581918803227947e-05,
      "loss": 0.2932,
      "step": 8266
    },
    {
      "epoch": 0.6256657679390757,
      "grad_norm": 0.79296875,
      "learning_rate": 1.5580930510380343e-05,
      "loss": 0.3487,
      "step": 8267
    },
    {
      "epoch": 0.625741450262523,
      "grad_norm": 0.79296875,
      "learning_rate": 1.557994213835766e-05,
      "loss": 0.3343,
      "step": 8268
    },
    {
      "epoch": 0.6258171325859704,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5578953687173923e-05,
      "loss": 0.3301,
      "step": 8269
    },
    {
      "epoch": 0.6258928149094177,
      "grad_norm": 0.78125,
      "learning_rate": 1.5577965156843153e-05,
      "loss": 0.3337,
      "step": 8270
    },
    {
      "epoch": 0.6259684972328651,
      "grad_norm": 0.75390625,
      "learning_rate": 1.557697654737938e-05,
      "loss": 0.3198,
      "step": 8271
    },
    {
      "epoch": 0.6260441795563124,
      "grad_norm": 0.7734375,
      "learning_rate": 1.557598785879662e-05,
      "loss": 0.3272,
      "step": 8272
    },
    {
      "epoch": 0.6261198618797597,
      "grad_norm": 0.84375,
      "learning_rate": 1.5574999091108903e-05,
      "loss": 0.384,
      "step": 8273
    },
    {
      "epoch": 0.626195544203207,
      "grad_norm": 0.6796875,
      "learning_rate": 1.557401024433026e-05,
      "loss": 0.2747,
      "step": 8274
    },
    {
      "epoch": 0.6262712265266543,
      "grad_norm": 0.80859375,
      "learning_rate": 1.5573021318474714e-05,
      "loss": 0.3603,
      "step": 8275
    },
    {
      "epoch": 0.6263469088501017,
      "grad_norm": 0.828125,
      "learning_rate": 1.5572032313556296e-05,
      "loss": 0.3819,
      "step": 8276
    },
    {
      "epoch": 0.626422591173549,
      "grad_norm": 0.70703125,
      "learning_rate": 1.557104322958904e-05,
      "loss": 0.2913,
      "step": 8277
    },
    {
      "epoch": 0.6264982734969964,
      "grad_norm": 0.84375,
      "learning_rate": 1.5570054066586974e-05,
      "loss": 0.3358,
      "step": 8278
    },
    {
      "epoch": 0.6265739558204437,
      "grad_norm": 0.77734375,
      "learning_rate": 1.5569064824564132e-05,
      "loss": 0.3232,
      "step": 8279
    },
    {
      "epoch": 0.626649638143891,
      "grad_norm": 0.76953125,
      "learning_rate": 1.556807550353455e-05,
      "loss": 0.3091,
      "step": 8280
    },
    {
      "epoch": 0.6267253204673383,
      "grad_norm": 0.77734375,
      "learning_rate": 1.5567086103512265e-05,
      "loss": 0.2939,
      "step": 8281
    },
    {
      "epoch": 0.6268010027907857,
      "grad_norm": 0.8046875,
      "learning_rate": 1.5566096624511306e-05,
      "loss": 0.3652,
      "step": 8282
    },
    {
      "epoch": 0.626876685114233,
      "grad_norm": 0.78125,
      "learning_rate": 1.5565107066545717e-05,
      "loss": 0.3446,
      "step": 8283
    },
    {
      "epoch": 0.6269523674376803,
      "grad_norm": 0.6796875,
      "learning_rate": 1.5564117429629533e-05,
      "loss": 0.277,
      "step": 8284
    },
    {
      "epoch": 0.6270280497611277,
      "grad_norm": 1.046875,
      "learning_rate": 1.5563127713776793e-05,
      "loss": 0.3093,
      "step": 8285
    },
    {
      "epoch": 0.627103732084575,
      "grad_norm": 0.796875,
      "learning_rate": 1.556213791900154e-05,
      "loss": 0.3469,
      "step": 8286
    },
    {
      "epoch": 0.6271794144080223,
      "grad_norm": 0.7265625,
      "learning_rate": 1.5561148045317818e-05,
      "loss": 0.2803,
      "step": 8287
    },
    {
      "epoch": 0.6272550967314696,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5560158092739668e-05,
      "loss": 0.3469,
      "step": 8288
    },
    {
      "epoch": 0.627330779054917,
      "grad_norm": 0.765625,
      "learning_rate": 1.555916806128113e-05,
      "loss": 0.3345,
      "step": 8289
    },
    {
      "epoch": 0.6274064613783643,
      "grad_norm": 0.75,
      "learning_rate": 1.5558177950956256e-05,
      "loss": 0.3287,
      "step": 8290
    },
    {
      "epoch": 0.6274821437018117,
      "grad_norm": 0.7265625,
      "learning_rate": 1.5557187761779093e-05,
      "loss": 0.2852,
      "step": 8291
    },
    {
      "epoch": 0.627557826025259,
      "grad_norm": 0.78125,
      "learning_rate": 1.555619749376368e-05,
      "loss": 0.3483,
      "step": 8292
    },
    {
      "epoch": 0.6276335083487063,
      "grad_norm": 0.82421875,
      "learning_rate": 1.5555207146924074e-05,
      "loss": 0.3937,
      "step": 8293
    },
    {
      "epoch": 0.6277091906721536,
      "grad_norm": 0.7421875,
      "learning_rate": 1.555421672127432e-05,
      "loss": 0.3053,
      "step": 8294
    },
    {
      "epoch": 0.6277848729956009,
      "grad_norm": 0.74609375,
      "learning_rate": 1.555322621682847e-05,
      "loss": 0.3199,
      "step": 8295
    },
    {
      "epoch": 0.6278605553190483,
      "grad_norm": 0.82421875,
      "learning_rate": 1.5552235633600576e-05,
      "loss": 0.3839,
      "step": 8296
    },
    {
      "epoch": 0.6279362376424956,
      "grad_norm": 0.77734375,
      "learning_rate": 1.5551244971604695e-05,
      "loss": 0.3189,
      "step": 8297
    },
    {
      "epoch": 0.628011919965943,
      "grad_norm": 0.89453125,
      "learning_rate": 1.5550254230854872e-05,
      "loss": 0.3898,
      "step": 8298
    },
    {
      "epoch": 0.6280876022893903,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5549263411365175e-05,
      "loss": 0.3011,
      "step": 8299
    },
    {
      "epoch": 0.6281632846128377,
      "grad_norm": 0.72265625,
      "learning_rate": 1.554827251314965e-05,
      "loss": 0.284,
      "step": 8300
    },
    {
      "epoch": 0.6282389669362849,
      "grad_norm": 0.75,
      "learning_rate": 1.554728153622236e-05,
      "loss": 0.304,
      "step": 8301
    },
    {
      "epoch": 0.6283146492597322,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5546290480597356e-05,
      "loss": 0.3267,
      "step": 8302
    },
    {
      "epoch": 0.6283903315831796,
      "grad_norm": 0.77734375,
      "learning_rate": 1.5545299346288708e-05,
      "loss": 0.336,
      "step": 8303
    },
    {
      "epoch": 0.6284660139066269,
      "grad_norm": 0.796875,
      "learning_rate": 1.554430813331047e-05,
      "loss": 0.3491,
      "step": 8304
    },
    {
      "epoch": 0.6285416962300743,
      "grad_norm": 0.73828125,
      "learning_rate": 1.5543316841676715e-05,
      "loss": 0.314,
      "step": 8305
    },
    {
      "epoch": 0.6286173785535216,
      "grad_norm": 0.6875,
      "learning_rate": 1.5542325471401493e-05,
      "loss": 0.2607,
      "step": 8306
    },
    {
      "epoch": 0.628693060876969,
      "grad_norm": 0.82421875,
      "learning_rate": 1.554133402249887e-05,
      "loss": 0.3185,
      "step": 8307
    },
    {
      "epoch": 0.6287687432004162,
      "grad_norm": 0.78125,
      "learning_rate": 1.554034249498292e-05,
      "loss": 0.3291,
      "step": 8308
    },
    {
      "epoch": 0.6288444255238635,
      "grad_norm": 0.77734375,
      "learning_rate": 1.55393508888677e-05,
      "loss": 0.3505,
      "step": 8309
    },
    {
      "epoch": 0.6289201078473109,
      "grad_norm": 0.79296875,
      "learning_rate": 1.5538359204167285e-05,
      "loss": 0.3653,
      "step": 8310
    },
    {
      "epoch": 0.6289957901707582,
      "grad_norm": 0.75,
      "learning_rate": 1.5537367440895737e-05,
      "loss": 0.2883,
      "step": 8311
    },
    {
      "epoch": 0.6290714724942056,
      "grad_norm": 0.71484375,
      "learning_rate": 1.5536375599067133e-05,
      "loss": 0.2897,
      "step": 8312
    },
    {
      "epoch": 0.6291471548176529,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5535383678695538e-05,
      "loss": 0.3153,
      "step": 8313
    },
    {
      "epoch": 0.6292228371411003,
      "grad_norm": 0.765625,
      "learning_rate": 1.5534391679795027e-05,
      "loss": 0.3464,
      "step": 8314
    },
    {
      "epoch": 0.6292985194645475,
      "grad_norm": 0.7578125,
      "learning_rate": 1.5533399602379674e-05,
      "loss": 0.3358,
      "step": 8315
    },
    {
      "epoch": 0.6293742017879949,
      "grad_norm": 0.80859375,
      "learning_rate": 1.5532407446463548e-05,
      "loss": 0.3363,
      "step": 8316
    },
    {
      "epoch": 0.6294498841114422,
      "grad_norm": 0.78125,
      "learning_rate": 1.5531415212060732e-05,
      "loss": 0.3164,
      "step": 8317
    },
    {
      "epoch": 0.6295255664348895,
      "grad_norm": 0.6953125,
      "learning_rate": 1.5530422899185298e-05,
      "loss": 0.2826,
      "step": 8318
    },
    {
      "epoch": 0.6296012487583369,
      "grad_norm": 0.78515625,
      "learning_rate": 1.5529430507851325e-05,
      "loss": 0.3548,
      "step": 8319
    },
    {
      "epoch": 0.6296769310817842,
      "grad_norm": 0.72265625,
      "learning_rate": 1.552843803807289e-05,
      "loss": 0.3027,
      "step": 8320
    },
    {
      "epoch": 0.6297526134052316,
      "grad_norm": 0.81640625,
      "learning_rate": 1.5527445489864076e-05,
      "loss": 0.3672,
      "step": 8321
    },
    {
      "epoch": 0.6298282957286788,
      "grad_norm": 0.8046875,
      "learning_rate": 1.5526452863238964e-05,
      "loss": 0.3535,
      "step": 8322
    },
    {
      "epoch": 0.6299039780521262,
      "grad_norm": 0.734375,
      "learning_rate": 1.5525460158211632e-05,
      "loss": 0.281,
      "step": 8323
    },
    {
      "epoch": 0.6299796603755735,
      "grad_norm": 0.765625,
      "learning_rate": 1.5524467374796165e-05,
      "loss": 0.3116,
      "step": 8324
    },
    {
      "epoch": 0.6300553426990209,
      "grad_norm": 0.79296875,
      "learning_rate": 1.552347451300665e-05,
      "loss": 0.3279,
      "step": 8325
    },
    {
      "epoch": 0.6301310250224682,
      "grad_norm": 0.73046875,
      "learning_rate": 1.5522481572857168e-05,
      "loss": 0.3102,
      "step": 8326
    },
    {
      "epoch": 0.6302067073459156,
      "grad_norm": 0.796875,
      "learning_rate": 1.552148855436181e-05,
      "loss": 0.3253,
      "step": 8327
    },
    {
      "epoch": 0.6302823896693629,
      "grad_norm": 0.77734375,
      "learning_rate": 1.5520495457534663e-05,
      "loss": 0.3189,
      "step": 8328
    },
    {
      "epoch": 0.6303580719928101,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5519502282389812e-05,
      "loss": 0.3181,
      "step": 8329
    },
    {
      "epoch": 0.6304337543162575,
      "grad_norm": 0.81640625,
      "learning_rate": 1.551850902894135e-05,
      "loss": 0.3594,
      "step": 8330
    },
    {
      "epoch": 0.6305094366397048,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5517515697203366e-05,
      "loss": 0.3499,
      "step": 8331
    },
    {
      "epoch": 0.6305851189631522,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5516522287189953e-05,
      "loss": 0.3097,
      "step": 8332
    },
    {
      "epoch": 0.6306608012865995,
      "grad_norm": 0.81640625,
      "learning_rate": 1.551552879891521e-05,
      "loss": 0.3422,
      "step": 8333
    },
    {
      "epoch": 0.6307364836100469,
      "grad_norm": 0.765625,
      "learning_rate": 1.551453523239322e-05,
      "loss": 0.284,
      "step": 8334
    },
    {
      "epoch": 0.6308121659334942,
      "grad_norm": 0.75,
      "learning_rate": 1.5513541587638084e-05,
      "loss": 0.3147,
      "step": 8335
    },
    {
      "epoch": 0.6308878482569414,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5512547864663904e-05,
      "loss": 0.3176,
      "step": 8336
    },
    {
      "epoch": 0.6309635305803888,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5511554063484774e-05,
      "loss": 0.2795,
      "step": 8337
    },
    {
      "epoch": 0.6310392129038361,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5510560184114785e-05,
      "loss": 0.3186,
      "step": 8338
    },
    {
      "epoch": 0.6311148952272835,
      "grad_norm": 0.79296875,
      "learning_rate": 1.5509566226568044e-05,
      "loss": 0.3794,
      "step": 8339
    },
    {
      "epoch": 0.6311905775507308,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5508572190858654e-05,
      "loss": 0.3235,
      "step": 8340
    },
    {
      "epoch": 0.6312662598741782,
      "grad_norm": 0.73046875,
      "learning_rate": 1.5507578077000713e-05,
      "loss": 0.3134,
      "step": 8341
    },
    {
      "epoch": 0.6313419421976255,
      "grad_norm": 0.78125,
      "learning_rate": 1.550658388500833e-05,
      "loss": 0.3083,
      "step": 8342
    },
    {
      "epoch": 0.6314176245210728,
      "grad_norm": 0.75,
      "learning_rate": 1.5505589614895602e-05,
      "loss": 0.3311,
      "step": 8343
    },
    {
      "epoch": 0.6314933068445201,
      "grad_norm": 0.734375,
      "learning_rate": 1.5504595266676635e-05,
      "loss": 0.2913,
      "step": 8344
    },
    {
      "epoch": 0.6315689891679674,
      "grad_norm": 0.73046875,
      "learning_rate": 1.550360084036554e-05,
      "loss": 0.3109,
      "step": 8345
    },
    {
      "epoch": 0.6316446714914148,
      "grad_norm": 0.734375,
      "learning_rate": 1.550260633597642e-05,
      "loss": 0.3018,
      "step": 8346
    },
    {
      "epoch": 0.6317203538148621,
      "grad_norm": 0.71875,
      "learning_rate": 1.550161175352339e-05,
      "loss": 0.3077,
      "step": 8347
    },
    {
      "epoch": 0.6317960361383095,
      "grad_norm": 0.765625,
      "learning_rate": 1.5500617093020557e-05,
      "loss": 0.3082,
      "step": 8348
    },
    {
      "epoch": 0.6318717184617568,
      "grad_norm": 0.73046875,
      "learning_rate": 1.5499622354482028e-05,
      "loss": 0.3097,
      "step": 8349
    },
    {
      "epoch": 0.6319474007852041,
      "grad_norm": 0.77734375,
      "learning_rate": 1.549862753792192e-05,
      "loss": 0.337,
      "step": 8350
    },
    {
      "epoch": 0.6320230831086514,
      "grad_norm": 0.75,
      "learning_rate": 1.5497632643354344e-05,
      "loss": 0.3323,
      "step": 8351
    },
    {
      "epoch": 0.6320987654320988,
      "grad_norm": 0.8046875,
      "learning_rate": 1.5496637670793415e-05,
      "loss": 0.3418,
      "step": 8352
    },
    {
      "epoch": 0.6321744477555461,
      "grad_norm": 1.0078125,
      "learning_rate": 1.5495642620253248e-05,
      "loss": 0.3725,
      "step": 8353
    },
    {
      "epoch": 0.6322501300789934,
      "grad_norm": 0.77734375,
      "learning_rate": 1.5494647491747957e-05,
      "loss": 0.2894,
      "step": 8354
    },
    {
      "epoch": 0.6323258124024408,
      "grad_norm": 0.765625,
      "learning_rate": 1.5493652285291666e-05,
      "loss": 0.2996,
      "step": 8355
    },
    {
      "epoch": 0.6324014947258881,
      "grad_norm": 0.765625,
      "learning_rate": 1.549265700089849e-05,
      "loss": 0.3256,
      "step": 8356
    },
    {
      "epoch": 0.6324771770493354,
      "grad_norm": 0.72265625,
      "learning_rate": 1.5491661638582546e-05,
      "loss": 0.2727,
      "step": 8357
    },
    {
      "epoch": 0.6325528593727827,
      "grad_norm": 0.73828125,
      "learning_rate": 1.5490666198357956e-05,
      "loss": 0.2916,
      "step": 8358
    },
    {
      "epoch": 0.6326285416962301,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5489670680238848e-05,
      "loss": 0.3159,
      "step": 8359
    },
    {
      "epoch": 0.6327042240196774,
      "grad_norm": 0.7734375,
      "learning_rate": 1.548867508423934e-05,
      "loss": 0.3186,
      "step": 8360
    },
    {
      "epoch": 0.6327799063431248,
      "grad_norm": 0.74609375,
      "learning_rate": 1.5487679410373555e-05,
      "loss": 0.317,
      "step": 8361
    },
    {
      "epoch": 0.6328555886665721,
      "grad_norm": 0.7578125,
      "learning_rate": 1.548668365865562e-05,
      "loss": 0.2982,
      "step": 8362
    },
    {
      "epoch": 0.6329312709900194,
      "grad_norm": 0.765625,
      "learning_rate": 1.5485687829099665e-05,
      "loss": 0.3287,
      "step": 8363
    },
    {
      "epoch": 0.6330069533134667,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5484691921719813e-05,
      "loss": 0.3091,
      "step": 8364
    },
    {
      "epoch": 0.633082635636914,
      "grad_norm": 0.80859375,
      "learning_rate": 1.5483695936530193e-05,
      "loss": 0.3334,
      "step": 8365
    },
    {
      "epoch": 0.6331583179603614,
      "grad_norm": 0.7890625,
      "learning_rate": 1.5482699873544937e-05,
      "loss": 0.3327,
      "step": 8366
    },
    {
      "epoch": 0.6332340002838087,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5481703732778173e-05,
      "loss": 0.3589,
      "step": 8367
    },
    {
      "epoch": 0.6333096826072561,
      "grad_norm": 0.80859375,
      "learning_rate": 1.5480707514244038e-05,
      "loss": 0.294,
      "step": 8368
    },
    {
      "epoch": 0.6333853649307034,
      "grad_norm": 0.79296875,
      "learning_rate": 1.5479711217956658e-05,
      "loss": 0.3519,
      "step": 8369
    },
    {
      "epoch": 0.6334610472541508,
      "grad_norm": 0.7578125,
      "learning_rate": 1.547871484393017e-05,
      "loss": 0.3091,
      "step": 8370
    },
    {
      "epoch": 0.633536729577598,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5477718392178716e-05,
      "loss": 0.2901,
      "step": 8371
    },
    {
      "epoch": 0.6336124119010453,
      "grad_norm": 0.92578125,
      "learning_rate": 1.547672186271642e-05,
      "loss": 0.309,
      "step": 8372
    },
    {
      "epoch": 0.6336880942244927,
      "grad_norm": 0.83203125,
      "learning_rate": 1.547572525555743e-05,
      "loss": 0.3395,
      "step": 8373
    },
    {
      "epoch": 0.63376377654794,
      "grad_norm": 0.73828125,
      "learning_rate": 1.547472857071588e-05,
      "loss": 0.2912,
      "step": 8374
    },
    {
      "epoch": 0.6338394588713874,
      "grad_norm": 0.7109375,
      "learning_rate": 1.547373180820591e-05,
      "loss": 0.2585,
      "step": 8375
    },
    {
      "epoch": 0.6339151411948347,
      "grad_norm": 0.77734375,
      "learning_rate": 1.547273496804166e-05,
      "loss": 0.3394,
      "step": 8376
    },
    {
      "epoch": 0.6339908235182821,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5471738050237277e-05,
      "loss": 0.3211,
      "step": 8377
    },
    {
      "epoch": 0.6340665058417293,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5470741054806894e-05,
      "loss": 0.3093,
      "step": 8378
    },
    {
      "epoch": 0.6341421881651766,
      "grad_norm": 0.71484375,
      "learning_rate": 1.5469743981764667e-05,
      "loss": 0.2902,
      "step": 8379
    },
    {
      "epoch": 0.634217870488624,
      "grad_norm": 0.7734375,
      "learning_rate": 1.546874683112473e-05,
      "loss": 0.3479,
      "step": 8380
    },
    {
      "epoch": 0.6342935528120713,
      "grad_norm": 0.828125,
      "learning_rate": 1.5467749602901235e-05,
      "loss": 0.3731,
      "step": 8381
    },
    {
      "epoch": 0.6343692351355187,
      "grad_norm": 0.7890625,
      "learning_rate": 1.546675229710833e-05,
      "loss": 0.3367,
      "step": 8382
    },
    {
      "epoch": 0.634444917458966,
      "grad_norm": 0.765625,
      "learning_rate": 1.5465754913760164e-05,
      "loss": 0.3263,
      "step": 8383
    },
    {
      "epoch": 0.6345205997824134,
      "grad_norm": 0.71875,
      "learning_rate": 1.5464757452870885e-05,
      "loss": 0.2863,
      "step": 8384
    },
    {
      "epoch": 0.6345962821058606,
      "grad_norm": 0.71875,
      "learning_rate": 1.5463759914454643e-05,
      "loss": 0.2933,
      "step": 8385
    },
    {
      "epoch": 0.634671964429308,
      "grad_norm": 0.890625,
      "learning_rate": 1.546276229852559e-05,
      "loss": 0.3545,
      "step": 8386
    },
    {
      "epoch": 0.6347476467527553,
      "grad_norm": 0.671875,
      "learning_rate": 1.5461764605097877e-05,
      "loss": 0.2559,
      "step": 8387
    },
    {
      "epoch": 0.6348233290762026,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5460766834185666e-05,
      "loss": 0.3089,
      "step": 8388
    },
    {
      "epoch": 0.63489901139965,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5459768985803104e-05,
      "loss": 0.3,
      "step": 8389
    },
    {
      "epoch": 0.6349746937230973,
      "grad_norm": 0.7890625,
      "learning_rate": 1.5458771059964348e-05,
      "loss": 0.3371,
      "step": 8390
    },
    {
      "epoch": 0.6350503760465446,
      "grad_norm": 0.796875,
      "learning_rate": 1.545777305668356e-05,
      "loss": 0.3444,
      "step": 8391
    },
    {
      "epoch": 0.6351260583699919,
      "grad_norm": 0.796875,
      "learning_rate": 1.5456774975974892e-05,
      "loss": 0.3523,
      "step": 8392
    },
    {
      "epoch": 0.6352017406934393,
      "grad_norm": 0.74609375,
      "learning_rate": 1.545577681785251e-05,
      "loss": 0.306,
      "step": 8393
    },
    {
      "epoch": 0.6352774230168866,
      "grad_norm": 0.78125,
      "learning_rate": 1.5454778582330567e-05,
      "loss": 0.3354,
      "step": 8394
    },
    {
      "epoch": 0.635353105340334,
      "grad_norm": 0.77734375,
      "learning_rate": 1.545378026942323e-05,
      "loss": 0.309,
      "step": 8395
    },
    {
      "epoch": 0.6354287876637813,
      "grad_norm": 0.7734375,
      "learning_rate": 1.545278187914466e-05,
      "loss": 0.352,
      "step": 8396
    },
    {
      "epoch": 0.6355044699872286,
      "grad_norm": 0.765625,
      "learning_rate": 1.5451783411509026e-05,
      "loss": 0.3224,
      "step": 8397
    },
    {
      "epoch": 0.6355801523106759,
      "grad_norm": 0.78515625,
      "learning_rate": 1.5450784866530488e-05,
      "loss": 0.3527,
      "step": 8398
    },
    {
      "epoch": 0.6356558346341232,
      "grad_norm": 1.5234375,
      "learning_rate": 1.544978624422321e-05,
      "loss": 0.4038,
      "step": 8399
    },
    {
      "epoch": 0.6357315169575706,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5448787544601366e-05,
      "loss": 0.3431,
      "step": 8400
    },
    {
      "epoch": 0.6358071992810179,
      "grad_norm": 0.79296875,
      "learning_rate": 1.5447788767679115e-05,
      "loss": 0.3469,
      "step": 8401
    },
    {
      "epoch": 0.6358828816044653,
      "grad_norm": 0.80859375,
      "learning_rate": 1.5446789913470633e-05,
      "loss": 0.3787,
      "step": 8402
    },
    {
      "epoch": 0.6359585639279126,
      "grad_norm": 0.8828125,
      "learning_rate": 1.5445790981990093e-05,
      "loss": 0.3405,
      "step": 8403
    },
    {
      "epoch": 0.63603424625136,
      "grad_norm": 0.73046875,
      "learning_rate": 1.544479197325166e-05,
      "loss": 0.2994,
      "step": 8404
    },
    {
      "epoch": 0.6361099285748072,
      "grad_norm": 0.72265625,
      "learning_rate": 1.544379288726951e-05,
      "loss": 0.3122,
      "step": 8405
    },
    {
      "epoch": 0.6361856108982545,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5442793724057815e-05,
      "loss": 0.3118,
      "step": 8406
    },
    {
      "epoch": 0.6362612932217019,
      "grad_norm": 0.73828125,
      "learning_rate": 1.5441794483630753e-05,
      "loss": 0.3202,
      "step": 8407
    },
    {
      "epoch": 0.6363369755451492,
      "grad_norm": 0.7109375,
      "learning_rate": 1.5440795166002492e-05,
      "loss": 0.3046,
      "step": 8408
    },
    {
      "epoch": 0.6364126578685966,
      "grad_norm": 0.78125,
      "learning_rate": 1.543979577118722e-05,
      "loss": 0.2938,
      "step": 8409
    },
    {
      "epoch": 0.6364883401920439,
      "grad_norm": 0.7890625,
      "learning_rate": 1.543879629919911e-05,
      "loss": 0.3064,
      "step": 8410
    },
    {
      "epoch": 0.6365640225154913,
      "grad_norm": 0.75390625,
      "learning_rate": 1.543779675005234e-05,
      "loss": 0.3317,
      "step": 8411
    },
    {
      "epoch": 0.6366397048389385,
      "grad_norm": 0.73828125,
      "learning_rate": 1.5436797123761095e-05,
      "loss": 0.3107,
      "step": 8412
    },
    {
      "epoch": 0.6367153871623858,
      "grad_norm": 0.83203125,
      "learning_rate": 1.5435797420339545e-05,
      "loss": 0.3133,
      "step": 8413
    },
    {
      "epoch": 0.6367910694858332,
      "grad_norm": 0.71484375,
      "learning_rate": 1.5434797639801887e-05,
      "loss": 0.2811,
      "step": 8414
    },
    {
      "epoch": 0.6368667518092805,
      "grad_norm": 0.80859375,
      "learning_rate": 1.5433797782162295e-05,
      "loss": 0.3224,
      "step": 8415
    },
    {
      "epoch": 0.6369424341327279,
      "grad_norm": 0.77734375,
      "learning_rate": 1.5432797847434957e-05,
      "loss": 0.2974,
      "step": 8416
    },
    {
      "epoch": 0.6370181164561752,
      "grad_norm": 0.71484375,
      "learning_rate": 1.543179783563406e-05,
      "loss": 0.2809,
      "step": 8417
    },
    {
      "epoch": 0.6370937987796226,
      "grad_norm": 0.765625,
      "learning_rate": 1.543079774677379e-05,
      "loss": 0.3025,
      "step": 8418
    },
    {
      "epoch": 0.6371694811030698,
      "grad_norm": 0.73046875,
      "learning_rate": 1.5429797580868332e-05,
      "loss": 0.2958,
      "step": 8419
    },
    {
      "epoch": 0.6372451634265172,
      "grad_norm": 0.78515625,
      "learning_rate": 1.5428797337931877e-05,
      "loss": 0.3471,
      "step": 8420
    },
    {
      "epoch": 0.6373208457499645,
      "grad_norm": 0.79296875,
      "learning_rate": 1.542779701797862e-05,
      "loss": 0.3608,
      "step": 8421
    },
    {
      "epoch": 0.6373965280734118,
      "grad_norm": 0.7109375,
      "learning_rate": 1.5426796621022746e-05,
      "loss": 0.2837,
      "step": 8422
    },
    {
      "epoch": 0.6374722103968592,
      "grad_norm": 0.75,
      "learning_rate": 1.5425796147078448e-05,
      "loss": 0.3209,
      "step": 8423
    },
    {
      "epoch": 0.6375478927203065,
      "grad_norm": 0.77734375,
      "learning_rate": 1.542479559615992e-05,
      "loss": 0.3118,
      "step": 8424
    },
    {
      "epoch": 0.6376235750437539,
      "grad_norm": 0.734375,
      "learning_rate": 1.5423794968281356e-05,
      "loss": 0.2859,
      "step": 8425
    },
    {
      "epoch": 0.6376992573672011,
      "grad_norm": 0.75,
      "learning_rate": 1.5422794263456958e-05,
      "loss": 0.2992,
      "step": 8426
    },
    {
      "epoch": 0.6377749396906485,
      "grad_norm": 0.734375,
      "learning_rate": 1.5421793481700915e-05,
      "loss": 0.3072,
      "step": 8427
    },
    {
      "epoch": 0.6378506220140958,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5420792623027425e-05,
      "loss": 0.3343,
      "step": 8428
    },
    {
      "epoch": 0.6379263043375432,
      "grad_norm": 0.75,
      "learning_rate": 1.5419791687450693e-05,
      "loss": 0.2774,
      "step": 8429
    },
    {
      "epoch": 0.6380019866609905,
      "grad_norm": 0.69140625,
      "learning_rate": 1.541879067498491e-05,
      "loss": 0.2709,
      "step": 8430
    },
    {
      "epoch": 0.6380776689844379,
      "grad_norm": 0.734375,
      "learning_rate": 1.541778958564429e-05,
      "loss": 0.3107,
      "step": 8431
    },
    {
      "epoch": 0.6381533513078852,
      "grad_norm": 0.78125,
      "learning_rate": 1.541678841944302e-05,
      "loss": 0.3463,
      "step": 8432
    },
    {
      "epoch": 0.6382290336313324,
      "grad_norm": 0.84765625,
      "learning_rate": 1.5415787176395313e-05,
      "loss": 0.3962,
      "step": 8433
    },
    {
      "epoch": 0.6383047159547798,
      "grad_norm": 0.85546875,
      "learning_rate": 1.5414785856515372e-05,
      "loss": 0.3609,
      "step": 8434
    },
    {
      "epoch": 0.6383803982782271,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5413784459817403e-05,
      "loss": 0.3214,
      "step": 8435
    },
    {
      "epoch": 0.6384560806016745,
      "grad_norm": 0.734375,
      "learning_rate": 1.541278298631561e-05,
      "loss": 0.3273,
      "step": 8436
    },
    {
      "epoch": 0.6385317629251218,
      "grad_norm": 0.7734375,
      "learning_rate": 1.54117814360242e-05,
      "loss": 0.3633,
      "step": 8437
    },
    {
      "epoch": 0.6386074452485692,
      "grad_norm": 0.734375,
      "learning_rate": 1.5410779808957387e-05,
      "loss": 0.3195,
      "step": 8438
    },
    {
      "epoch": 0.6386831275720165,
      "grad_norm": 0.79296875,
      "learning_rate": 1.5409778105129375e-05,
      "loss": 0.3091,
      "step": 8439
    },
    {
      "epoch": 0.6387588098954637,
      "grad_norm": 0.8046875,
      "learning_rate": 1.5408776324554377e-05,
      "loss": 0.3228,
      "step": 8440
    },
    {
      "epoch": 0.6388344922189111,
      "grad_norm": 0.78125,
      "learning_rate": 1.5407774467246603e-05,
      "loss": 0.3393,
      "step": 8441
    },
    {
      "epoch": 0.6389101745423584,
      "grad_norm": 0.765625,
      "learning_rate": 1.540677253322027e-05,
      "loss": 0.2996,
      "step": 8442
    },
    {
      "epoch": 0.6389858568658058,
      "grad_norm": 0.72265625,
      "learning_rate": 1.5405770522489592e-05,
      "loss": 0.3094,
      "step": 8443
    },
    {
      "epoch": 0.6390615391892531,
      "grad_norm": 0.75,
      "learning_rate": 1.5404768435068778e-05,
      "loss": 0.3056,
      "step": 8444
    },
    {
      "epoch": 0.6391372215127005,
      "grad_norm": 0.7265625,
      "learning_rate": 1.5403766270972054e-05,
      "loss": 0.2835,
      "step": 8445
    },
    {
      "epoch": 0.6392129038361478,
      "grad_norm": 0.78515625,
      "learning_rate": 1.540276403021363e-05,
      "loss": 0.34,
      "step": 8446
    },
    {
      "epoch": 0.639288586159595,
      "grad_norm": 0.83984375,
      "learning_rate": 1.5401761712807724e-05,
      "loss": 0.3996,
      "step": 8447
    },
    {
      "epoch": 0.6393642684830424,
      "grad_norm": 0.78515625,
      "learning_rate": 1.5400759318768566e-05,
      "loss": 0.3622,
      "step": 8448
    },
    {
      "epoch": 0.6394399508064897,
      "grad_norm": 0.78125,
      "learning_rate": 1.539975684811036e-05,
      "loss": 0.3486,
      "step": 8449
    },
    {
      "epoch": 0.6395156331299371,
      "grad_norm": 1.21875,
      "learning_rate": 1.5398754300847346e-05,
      "loss": 0.3889,
      "step": 8450
    },
    {
      "epoch": 0.6395913154533844,
      "grad_norm": 0.75,
      "learning_rate": 1.5397751676993732e-05,
      "loss": 0.2958,
      "step": 8451
    },
    {
      "epoch": 0.6396669977768318,
      "grad_norm": 0.79296875,
      "learning_rate": 1.5396748976563745e-05,
      "loss": 0.3556,
      "step": 8452
    },
    {
      "epoch": 0.6397426801002791,
      "grad_norm": 0.7109375,
      "learning_rate": 1.5395746199571616e-05,
      "loss": 0.2707,
      "step": 8453
    },
    {
      "epoch": 0.6398183624237264,
      "grad_norm": 0.734375,
      "learning_rate": 1.539474334603157e-05,
      "loss": 0.3273,
      "step": 8454
    },
    {
      "epoch": 0.6398940447471737,
      "grad_norm": 0.82421875,
      "learning_rate": 1.539374041595783e-05,
      "loss": 0.3845,
      "step": 8455
    },
    {
      "epoch": 0.639969727070621,
      "grad_norm": 0.80078125,
      "learning_rate": 1.5392737409364624e-05,
      "loss": 0.3156,
      "step": 8456
    },
    {
      "epoch": 0.6400454093940684,
      "grad_norm": 0.83984375,
      "learning_rate": 1.5391734326266185e-05,
      "loss": 0.358,
      "step": 8457
    },
    {
      "epoch": 0.6401210917175157,
      "grad_norm": 0.80078125,
      "learning_rate": 1.539073116667674e-05,
      "loss": 0.315,
      "step": 8458
    },
    {
      "epoch": 0.6401967740409631,
      "grad_norm": 0.7578125,
      "learning_rate": 1.538972793061053e-05,
      "loss": 0.3156,
      "step": 8459
    },
    {
      "epoch": 0.6402724563644104,
      "grad_norm": 0.6953125,
      "learning_rate": 1.5388724618081774e-05,
      "loss": 0.281,
      "step": 8460
    },
    {
      "epoch": 0.6403481386878577,
      "grad_norm": 0.7890625,
      "learning_rate": 1.5387721229104712e-05,
      "loss": 0.3366,
      "step": 8461
    },
    {
      "epoch": 0.640423821011305,
      "grad_norm": 0.6953125,
      "learning_rate": 1.5386717763693578e-05,
      "loss": 0.2615,
      "step": 8462
    },
    {
      "epoch": 0.6404995033347524,
      "grad_norm": 0.73828125,
      "learning_rate": 1.5385714221862614e-05,
      "loss": 0.3184,
      "step": 8463
    },
    {
      "epoch": 0.6405751856581997,
      "grad_norm": 0.71875,
      "learning_rate": 1.538471060362605e-05,
      "loss": 0.2749,
      "step": 8464
    },
    {
      "epoch": 0.640650867981647,
      "grad_norm": 0.7265625,
      "learning_rate": 1.5383706908998123e-05,
      "loss": 0.2815,
      "step": 8465
    },
    {
      "epoch": 0.6407265503050944,
      "grad_norm": 0.8203125,
      "learning_rate": 1.5382703137993074e-05,
      "loss": 0.3579,
      "step": 8466
    },
    {
      "epoch": 0.6408022326285417,
      "grad_norm": 0.7578125,
      "learning_rate": 1.5381699290625142e-05,
      "loss": 0.3164,
      "step": 8467
    },
    {
      "epoch": 0.640877914951989,
      "grad_norm": 0.79296875,
      "learning_rate": 1.5380695366908577e-05,
      "loss": 0.3227,
      "step": 8468
    },
    {
      "epoch": 0.6409535972754363,
      "grad_norm": 0.83984375,
      "learning_rate": 1.537969136685761e-05,
      "loss": 0.3713,
      "step": 8469
    },
    {
      "epoch": 0.6410292795988837,
      "grad_norm": 0.73046875,
      "learning_rate": 1.537868729048649e-05,
      "loss": 0.2861,
      "step": 8470
    },
    {
      "epoch": 0.641104961922331,
      "grad_norm": 0.78515625,
      "learning_rate": 1.537768313780946e-05,
      "loss": 0.318,
      "step": 8471
    },
    {
      "epoch": 0.6411806442457784,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5376678908840766e-05,
      "loss": 0.3353,
      "step": 8472
    },
    {
      "epoch": 0.6412563265692257,
      "grad_norm": 0.96484375,
      "learning_rate": 1.537567460359466e-05,
      "loss": 0.3307,
      "step": 8473
    },
    {
      "epoch": 0.6413320088926731,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5374670222085383e-05,
      "loss": 0.2968,
      "step": 8474
    },
    {
      "epoch": 0.6414076912161203,
      "grad_norm": 0.74609375,
      "learning_rate": 1.5373665764327182e-05,
      "loss": 0.3062,
      "step": 8475
    },
    {
      "epoch": 0.6414833735395676,
      "grad_norm": 0.78515625,
      "learning_rate": 1.537266123033431e-05,
      "loss": 0.3513,
      "step": 8476
    },
    {
      "epoch": 0.641559055863015,
      "grad_norm": 0.73828125,
      "learning_rate": 1.5371656620121024e-05,
      "loss": 0.337,
      "step": 8477
    },
    {
      "epoch": 0.6416347381864623,
      "grad_norm": 0.73828125,
      "learning_rate": 1.5370651933701566e-05,
      "loss": 0.2761,
      "step": 8478
    },
    {
      "epoch": 0.6417104205099097,
      "grad_norm": 0.7265625,
      "learning_rate": 1.5369647171090197e-05,
      "loss": 0.2974,
      "step": 8479
    },
    {
      "epoch": 0.641786102833357,
      "grad_norm": 0.8046875,
      "learning_rate": 1.5368642332301165e-05,
      "loss": 0.3534,
      "step": 8480
    },
    {
      "epoch": 0.6418617851568044,
      "grad_norm": 0.9453125,
      "learning_rate": 1.536763741734873e-05,
      "loss": 0.371,
      "step": 8481
    },
    {
      "epoch": 0.6419374674802516,
      "grad_norm": 0.80078125,
      "learning_rate": 1.536663242624715e-05,
      "loss": 0.3155,
      "step": 8482
    },
    {
      "epoch": 0.642013149803699,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5365627359010674e-05,
      "loss": 0.2965,
      "step": 8483
    },
    {
      "epoch": 0.6420888321271463,
      "grad_norm": 0.74609375,
      "learning_rate": 1.536462221565357e-05,
      "loss": 0.3246,
      "step": 8484
    },
    {
      "epoch": 0.6421645144505936,
      "grad_norm": 0.69140625,
      "learning_rate": 1.5363616996190093e-05,
      "loss": 0.2667,
      "step": 8485
    },
    {
      "epoch": 0.642240196774041,
      "grad_norm": 0.796875,
      "learning_rate": 1.5362611700634505e-05,
      "loss": 0.3603,
      "step": 8486
    },
    {
      "epoch": 0.6423158790974883,
      "grad_norm": 0.75,
      "learning_rate": 1.5361606329001066e-05,
      "loss": 0.3067,
      "step": 8487
    },
    {
      "epoch": 0.6423915614209357,
      "grad_norm": 0.87109375,
      "learning_rate": 1.5360600881304043e-05,
      "loss": 0.3822,
      "step": 8488
    },
    {
      "epoch": 0.6424672437443829,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5359595357557697e-05,
      "loss": 0.2886,
      "step": 8489
    },
    {
      "epoch": 0.6425429260678303,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5358589757776294e-05,
      "loss": 0.2989,
      "step": 8490
    },
    {
      "epoch": 0.6426186083912776,
      "grad_norm": 0.75,
      "learning_rate": 1.5357584081974097e-05,
      "loss": 0.3319,
      "step": 8491
    },
    {
      "epoch": 0.642694290714725,
      "grad_norm": 0.78125,
      "learning_rate": 1.5356578330165377e-05,
      "loss": 0.3107,
      "step": 8492
    },
    {
      "epoch": 0.6427699730381723,
      "grad_norm": 0.71875,
      "learning_rate": 1.5355572502364398e-05,
      "loss": 0.2961,
      "step": 8493
    },
    {
      "epoch": 0.6428456553616196,
      "grad_norm": 0.8203125,
      "learning_rate": 1.5354566598585435e-05,
      "loss": 0.3368,
      "step": 8494
    },
    {
      "epoch": 0.642921337685067,
      "grad_norm": 0.734375,
      "learning_rate": 1.5353560618842758e-05,
      "loss": 0.301,
      "step": 8495
    },
    {
      "epoch": 0.6429970200085142,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5352554563150638e-05,
      "loss": 0.3111,
      "step": 8496
    },
    {
      "epoch": 0.6430727023319616,
      "grad_norm": 0.8046875,
      "learning_rate": 1.5351548431523342e-05,
      "loss": 0.3297,
      "step": 8497
    },
    {
      "epoch": 0.6431483846554089,
      "grad_norm": 1.1796875,
      "learning_rate": 1.535054222397515e-05,
      "loss": 0.3846,
      "step": 8498
    },
    {
      "epoch": 0.6432240669788563,
      "grad_norm": 0.7734375,
      "learning_rate": 1.534953594052033e-05,
      "loss": 0.3384,
      "step": 8499
    },
    {
      "epoch": 0.6432997493023036,
      "grad_norm": 0.70703125,
      "learning_rate": 1.534852958117317e-05,
      "loss": 0.2845,
      "step": 8500
    },
    {
      "epoch": 0.643375431625751,
      "grad_norm": 0.72265625,
      "learning_rate": 1.534752314594794e-05,
      "loss": 0.3035,
      "step": 8501
    },
    {
      "epoch": 0.6434511139491983,
      "grad_norm": 0.75,
      "learning_rate": 1.534651663485891e-05,
      "loss": 0.3031,
      "step": 8502
    },
    {
      "epoch": 0.6435267962726455,
      "grad_norm": 1.09375,
      "learning_rate": 1.5345510047920372e-05,
      "loss": 0.309,
      "step": 8503
    },
    {
      "epoch": 0.6436024785960929,
      "grad_norm": 0.73828125,
      "learning_rate": 1.53445033851466e-05,
      "loss": 0.2588,
      "step": 8504
    },
    {
      "epoch": 0.6436781609195402,
      "grad_norm": 0.828125,
      "learning_rate": 1.5343496646551875e-05,
      "loss": 0.3641,
      "step": 8505
    },
    {
      "epoch": 0.6437538432429876,
      "grad_norm": 0.84375,
      "learning_rate": 1.534248983215048e-05,
      "loss": 0.369,
      "step": 8506
    },
    {
      "epoch": 0.6438295255664349,
      "grad_norm": 0.78125,
      "learning_rate": 1.5341482941956697e-05,
      "loss": 0.3284,
      "step": 8507
    },
    {
      "epoch": 0.6439052078898823,
      "grad_norm": 0.7109375,
      "learning_rate": 1.5340475975984814e-05,
      "loss": 0.2879,
      "step": 8508
    },
    {
      "epoch": 0.6439808902133296,
      "grad_norm": 0.77734375,
      "learning_rate": 1.5339468934249115e-05,
      "loss": 0.3094,
      "step": 8509
    },
    {
      "epoch": 0.6440565725367768,
      "grad_norm": 0.79296875,
      "learning_rate": 1.533846181676389e-05,
      "loss": 0.3159,
      "step": 8510
    },
    {
      "epoch": 0.6441322548602242,
      "grad_norm": 1.0390625,
      "learning_rate": 1.5337454623543417e-05,
      "loss": 0.368,
      "step": 8511
    },
    {
      "epoch": 0.6442079371836715,
      "grad_norm": 1.3203125,
      "learning_rate": 1.5336447354601995e-05,
      "loss": 0.4046,
      "step": 8512
    },
    {
      "epoch": 0.6442836195071189,
      "grad_norm": 0.75,
      "learning_rate": 1.5335440009953906e-05,
      "loss": 0.323,
      "step": 8513
    },
    {
      "epoch": 0.6443593018305662,
      "grad_norm": 0.8046875,
      "learning_rate": 1.5334432589613445e-05,
      "loss": 0.3514,
      "step": 8514
    },
    {
      "epoch": 0.6444349841540136,
      "grad_norm": 0.82421875,
      "learning_rate": 1.5333425093594908e-05,
      "loss": 0.3863,
      "step": 8515
    },
    {
      "epoch": 0.6445106664774608,
      "grad_norm": 0.8359375,
      "learning_rate": 1.533241752191258e-05,
      "loss": 0.3463,
      "step": 8516
    },
    {
      "epoch": 0.6445863488009081,
      "grad_norm": 0.7578125,
      "learning_rate": 1.5331409874580757e-05,
      "loss": 0.3023,
      "step": 8517
    },
    {
      "epoch": 0.6446620311243555,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5330402151613738e-05,
      "loss": 0.3052,
      "step": 8518
    },
    {
      "epoch": 0.6447377134478028,
      "grad_norm": 0.78125,
      "learning_rate": 1.532939435302582e-05,
      "loss": 0.3338,
      "step": 8519
    },
    {
      "epoch": 0.6448133957712502,
      "grad_norm": 0.72265625,
      "learning_rate": 1.5328386478831295e-05,
      "loss": 0.2742,
      "step": 8520
    },
    {
      "epoch": 0.6448890780946975,
      "grad_norm": 0.7578125,
      "learning_rate": 1.5327378529044464e-05,
      "loss": 0.3042,
      "step": 8521
    },
    {
      "epoch": 0.6449647604181449,
      "grad_norm": 0.80078125,
      "learning_rate": 1.5326370503679625e-05,
      "loss": 0.3242,
      "step": 8522
    },
    {
      "epoch": 0.6450404427415921,
      "grad_norm": 0.859375,
      "learning_rate": 1.5325362402751078e-05,
      "loss": 0.2636,
      "step": 8523
    },
    {
      "epoch": 0.6451161250650395,
      "grad_norm": 0.8203125,
      "learning_rate": 1.532435422627313e-05,
      "loss": 0.3718,
      "step": 8524
    },
    {
      "epoch": 0.6451918073884868,
      "grad_norm": 0.765625,
      "learning_rate": 1.5323345974260084e-05,
      "loss": 0.2868,
      "step": 8525
    },
    {
      "epoch": 0.6452674897119342,
      "grad_norm": 0.7890625,
      "learning_rate": 1.5322337646726232e-05,
      "loss": 0.2942,
      "step": 8526
    },
    {
      "epoch": 0.6453431720353815,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5321329243685895e-05,
      "loss": 0.3045,
      "step": 8527
    },
    {
      "epoch": 0.6454188543588288,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5320320765153367e-05,
      "loss": 0.2814,
      "step": 8528
    },
    {
      "epoch": 0.6454945366822762,
      "grad_norm": 0.9375,
      "learning_rate": 1.531931221114296e-05,
      "loss": 0.3494,
      "step": 8529
    },
    {
      "epoch": 0.6455702190057234,
      "grad_norm": 0.70703125,
      "learning_rate": 1.531830358166898e-05,
      "loss": 0.2773,
      "step": 8530
    },
    {
      "epoch": 0.6456459013291708,
      "grad_norm": 0.7265625,
      "learning_rate": 1.531729487674574e-05,
      "loss": 0.3075,
      "step": 8531
    },
    {
      "epoch": 0.6457215836526181,
      "grad_norm": 0.69921875,
      "learning_rate": 1.531628609638754e-05,
      "loss": 0.2766,
      "step": 8532
    },
    {
      "epoch": 0.6457972659760655,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5315277240608707e-05,
      "loss": 0.3186,
      "step": 8533
    },
    {
      "epoch": 0.6458729482995128,
      "grad_norm": 0.89453125,
      "learning_rate": 1.531426830942354e-05,
      "loss": 0.3237,
      "step": 8534
    },
    {
      "epoch": 0.6459486306229602,
      "grad_norm": 0.6953125,
      "learning_rate": 1.531325930284636e-05,
      "loss": 0.2676,
      "step": 8535
    },
    {
      "epoch": 0.6460243129464075,
      "grad_norm": 0.7890625,
      "learning_rate": 1.531225022089148e-05,
      "loss": 0.3311,
      "step": 8536
    },
    {
      "epoch": 0.6460999952698547,
      "grad_norm": 0.75390625,
      "learning_rate": 1.531124106357321e-05,
      "loss": 0.3119,
      "step": 8537
    },
    {
      "epoch": 0.6461756775933021,
      "grad_norm": 0.72265625,
      "learning_rate": 1.5310231830905876e-05,
      "loss": 0.29,
      "step": 8538
    },
    {
      "epoch": 0.6462513599167494,
      "grad_norm": 0.81640625,
      "learning_rate": 1.530922252290379e-05,
      "loss": 0.381,
      "step": 8539
    },
    {
      "epoch": 0.6463270422401968,
      "grad_norm": 0.7265625,
      "learning_rate": 1.5308213139581268e-05,
      "loss": 0.3152,
      "step": 8540
    },
    {
      "epoch": 0.6464027245636441,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5307203680952632e-05,
      "loss": 0.3304,
      "step": 8541
    },
    {
      "epoch": 0.6464784068870915,
      "grad_norm": 0.75390625,
      "learning_rate": 1.530619414703221e-05,
      "loss": 0.2726,
      "step": 8542
    },
    {
      "epoch": 0.6465540892105388,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5305184537834317e-05,
      "loss": 0.3025,
      "step": 8543
    },
    {
      "epoch": 0.646629771533986,
      "grad_norm": 0.78125,
      "learning_rate": 1.5304174853373277e-05,
      "loss": 0.3098,
      "step": 8544
    },
    {
      "epoch": 0.6467054538574334,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5303165093663413e-05,
      "loss": 0.3413,
      "step": 8545
    },
    {
      "epoch": 0.6467811361808807,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5302155258719052e-05,
      "loss": 0.315,
      "step": 8546
    },
    {
      "epoch": 0.6468568185043281,
      "grad_norm": 0.71484375,
      "learning_rate": 1.5301145348554517e-05,
      "loss": 0.314,
      "step": 8547
    },
    {
      "epoch": 0.6469325008277754,
      "grad_norm": 0.765625,
      "learning_rate": 1.5300135363184145e-05,
      "loss": 0.2948,
      "step": 8548
    },
    {
      "epoch": 0.6470081831512228,
      "grad_norm": 0.734375,
      "learning_rate": 1.5299125302622252e-05,
      "loss": 0.282,
      "step": 8549
    },
    {
      "epoch": 0.6470838654746701,
      "grad_norm": 0.97265625,
      "learning_rate": 1.5298115166883176e-05,
      "loss": 0.3286,
      "step": 8550
    },
    {
      "epoch": 0.6471595477981174,
      "grad_norm": 0.72265625,
      "learning_rate": 1.5297104955981246e-05,
      "loss": 0.2958,
      "step": 8551
    },
    {
      "epoch": 0.6472352301215647,
      "grad_norm": 0.74609375,
      "learning_rate": 1.5296094669930787e-05,
      "loss": 0.3367,
      "step": 8552
    },
    {
      "epoch": 0.647310912445012,
      "grad_norm": 0.734375,
      "learning_rate": 1.529508430874614e-05,
      "loss": 0.2948,
      "step": 8553
    },
    {
      "epoch": 0.6473865947684594,
      "grad_norm": 0.796875,
      "learning_rate": 1.5294073872441634e-05,
      "loss": 0.3498,
      "step": 8554
    },
    {
      "epoch": 0.6474622770919067,
      "grad_norm": 0.74609375,
      "learning_rate": 1.5293063361031605e-05,
      "loss": 0.2865,
      "step": 8555
    },
    {
      "epoch": 0.6475379594153541,
      "grad_norm": 0.7578125,
      "learning_rate": 1.529205277453039e-05,
      "loss": 0.3286,
      "step": 8556
    },
    {
      "epoch": 0.6476136417388014,
      "grad_norm": 0.7265625,
      "learning_rate": 1.529104211295233e-05,
      "loss": 0.2873,
      "step": 8557
    },
    {
      "epoch": 0.6476893240622487,
      "grad_norm": 0.703125,
      "learning_rate": 1.529003137631175e-05,
      "loss": 0.2793,
      "step": 8558
    },
    {
      "epoch": 0.647765006385696,
      "grad_norm": 0.7265625,
      "learning_rate": 1.5289020564622998e-05,
      "loss": 0.3256,
      "step": 8559
    },
    {
      "epoch": 0.6478406887091434,
      "grad_norm": 0.79296875,
      "learning_rate": 1.5288009677900415e-05,
      "loss": 0.2851,
      "step": 8560
    },
    {
      "epoch": 0.6479163710325907,
      "grad_norm": 0.76953125,
      "learning_rate": 1.528699871615834e-05,
      "loss": 0.3123,
      "step": 8561
    },
    {
      "epoch": 0.647992053356038,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5285987679411114e-05,
      "loss": 0.3107,
      "step": 8562
    },
    {
      "epoch": 0.6480677356794854,
      "grad_norm": 0.75,
      "learning_rate": 1.5284976567673083e-05,
      "loss": 0.3199,
      "step": 8563
    },
    {
      "epoch": 0.6481434180029327,
      "grad_norm": 0.74609375,
      "learning_rate": 1.528396538095859e-05,
      "loss": 0.3047,
      "step": 8564
    },
    {
      "epoch": 0.64821910032638,
      "grad_norm": 0.78515625,
      "learning_rate": 1.5282954119281978e-05,
      "loss": 0.3715,
      "step": 8565
    },
    {
      "epoch": 0.6482947826498273,
      "grad_norm": 0.75,
      "learning_rate": 1.52819427826576e-05,
      "loss": 0.3172,
      "step": 8566
    },
    {
      "epoch": 0.6483704649732747,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5280931371099797e-05,
      "loss": 0.2939,
      "step": 8567
    },
    {
      "epoch": 0.648446147296722,
      "grad_norm": 0.78515625,
      "learning_rate": 1.5279919884622922e-05,
      "loss": 0.3241,
      "step": 8568
    },
    {
      "epoch": 0.6485218296201694,
      "grad_norm": 0.80078125,
      "learning_rate": 1.527890832324132e-05,
      "loss": 0.3227,
      "step": 8569
    },
    {
      "epoch": 0.6485975119436167,
      "grad_norm": 0.78515625,
      "learning_rate": 1.5277896686969345e-05,
      "loss": 0.3353,
      "step": 8570
    },
    {
      "epoch": 0.648673194267064,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5276884975821355e-05,
      "loss": 0.3177,
      "step": 8571
    },
    {
      "epoch": 0.6487488765905113,
      "grad_norm": 0.78125,
      "learning_rate": 1.5275873189811688e-05,
      "loss": 0.3193,
      "step": 8572
    },
    {
      "epoch": 0.6488245589139586,
      "grad_norm": 0.75390625,
      "learning_rate": 1.527486132895471e-05,
      "loss": 0.2876,
      "step": 8573
    },
    {
      "epoch": 0.648900241237406,
      "grad_norm": 0.7265625,
      "learning_rate": 1.5273849393264767e-05,
      "loss": 0.3017,
      "step": 8574
    },
    {
      "epoch": 0.6489759235608533,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5272837382756227e-05,
      "loss": 0.3235,
      "step": 8575
    },
    {
      "epoch": 0.6490516058843007,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5271825297443436e-05,
      "loss": 0.3152,
      "step": 8576
    },
    {
      "epoch": 0.649127288207748,
      "grad_norm": 0.74609375,
      "learning_rate": 1.527081313734076e-05,
      "loss": 0.3253,
      "step": 8577
    },
    {
      "epoch": 0.6492029705311954,
      "grad_norm": 0.76953125,
      "learning_rate": 1.526980090246255e-05,
      "loss": 0.3419,
      "step": 8578
    },
    {
      "epoch": 0.6492786528546426,
      "grad_norm": 1.1171875,
      "learning_rate": 1.5268788592823173e-05,
      "loss": 0.35,
      "step": 8579
    },
    {
      "epoch": 0.6493543351780899,
      "grad_norm": 0.8203125,
      "learning_rate": 1.5267776208436988e-05,
      "loss": 0.3945,
      "step": 8580
    },
    {
      "epoch": 0.6494300175015373,
      "grad_norm": 0.71875,
      "learning_rate": 1.5266763749318362e-05,
      "loss": 0.2731,
      "step": 8581
    },
    {
      "epoch": 0.6495056998249846,
      "grad_norm": 0.796875,
      "learning_rate": 1.526575121548165e-05,
      "loss": 0.3233,
      "step": 8582
    },
    {
      "epoch": 0.649581382148432,
      "grad_norm": 0.765625,
      "learning_rate": 1.5264738606941215e-05,
      "loss": 0.3213,
      "step": 8583
    },
    {
      "epoch": 0.6496570644718793,
      "grad_norm": 0.72265625,
      "learning_rate": 1.5263725923711438e-05,
      "loss": 0.3062,
      "step": 8584
    },
    {
      "epoch": 0.6497327467953267,
      "grad_norm": 0.75390625,
      "learning_rate": 1.526271316580667e-05,
      "loss": 0.3073,
      "step": 8585
    },
    {
      "epoch": 0.6498084291187739,
      "grad_norm": 0.83203125,
      "learning_rate": 1.526170033324129e-05,
      "loss": 0.3274,
      "step": 8586
    },
    {
      "epoch": 0.6498841114422212,
      "grad_norm": 0.8125,
      "learning_rate": 1.5260687426029657e-05,
      "loss": 0.3448,
      "step": 8587
    },
    {
      "epoch": 0.6499597937656686,
      "grad_norm": 0.98046875,
      "learning_rate": 1.5259674444186145e-05,
      "loss": 0.3673,
      "step": 8588
    },
    {
      "epoch": 0.6500354760891159,
      "grad_norm": 0.8046875,
      "learning_rate": 1.5258661387725124e-05,
      "loss": 0.3332,
      "step": 8589
    },
    {
      "epoch": 0.6501111584125633,
      "grad_norm": 0.75390625,
      "learning_rate": 1.525764825666097e-05,
      "loss": 0.3263,
      "step": 8590
    },
    {
      "epoch": 0.6501868407360106,
      "grad_norm": 0.73828125,
      "learning_rate": 1.5256635051008054e-05,
      "loss": 0.3022,
      "step": 8591
    },
    {
      "epoch": 0.650262523059458,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5255621770780748e-05,
      "loss": 0.3051,
      "step": 8592
    },
    {
      "epoch": 0.6503382053829052,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5254608415993427e-05,
      "loss": 0.3347,
      "step": 8593
    },
    {
      "epoch": 0.6504138877063526,
      "grad_norm": 0.8203125,
      "learning_rate": 1.525359498666047e-05,
      "loss": 0.3553,
      "step": 8594
    },
    {
      "epoch": 0.6504895700297999,
      "grad_norm": 0.6796875,
      "learning_rate": 1.5252581482796252e-05,
      "loss": 0.2859,
      "step": 8595
    },
    {
      "epoch": 0.6505652523532472,
      "grad_norm": 0.7109375,
      "learning_rate": 1.525156790441515e-05,
      "loss": 0.2891,
      "step": 8596
    },
    {
      "epoch": 0.6506409346766946,
      "grad_norm": 0.73828125,
      "learning_rate": 1.5250554251531547e-05,
      "loss": 0.3139,
      "step": 8597
    },
    {
      "epoch": 0.6507166170001419,
      "grad_norm": 0.73046875,
      "learning_rate": 1.524954052415982e-05,
      "loss": 0.3017,
      "step": 8598
    },
    {
      "epoch": 0.6507922993235893,
      "grad_norm": 0.74609375,
      "learning_rate": 1.5248526722314357e-05,
      "loss": 0.3068,
      "step": 8599
    },
    {
      "epoch": 0.6508679816470365,
      "grad_norm": 0.73046875,
      "learning_rate": 1.5247512846009532e-05,
      "loss": 0.3091,
      "step": 8600
    },
    {
      "epoch": 0.6509436639704839,
      "grad_norm": 1.0,
      "learning_rate": 1.5246498895259732e-05,
      "loss": 0.353,
      "step": 8601
    },
    {
      "epoch": 0.6510193462939312,
      "grad_norm": 0.79296875,
      "learning_rate": 1.5245484870079343e-05,
      "loss": 0.3491,
      "step": 8602
    },
    {
      "epoch": 0.6510950286173786,
      "grad_norm": 0.69921875,
      "learning_rate": 1.5244470770482749e-05,
      "loss": 0.2797,
      "step": 8603
    },
    {
      "epoch": 0.6511707109408259,
      "grad_norm": 0.78515625,
      "learning_rate": 1.5243456596484339e-05,
      "loss": 0.3423,
      "step": 8604
    },
    {
      "epoch": 0.6512463932642732,
      "grad_norm": 0.75,
      "learning_rate": 1.5242442348098495e-05,
      "loss": 0.3285,
      "step": 8605
    },
    {
      "epoch": 0.6513220755877206,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5241428025339613e-05,
      "loss": 0.3343,
      "step": 8606
    },
    {
      "epoch": 0.6513977579111678,
      "grad_norm": 0.7265625,
      "learning_rate": 1.524041362822208e-05,
      "loss": 0.2832,
      "step": 8607
    },
    {
      "epoch": 0.6514734402346152,
      "grad_norm": 0.796875,
      "learning_rate": 1.5239399156760287e-05,
      "loss": 0.3497,
      "step": 8608
    },
    {
      "epoch": 0.6515491225580625,
      "grad_norm": 0.7890625,
      "learning_rate": 1.5238384610968623e-05,
      "loss": 0.3508,
      "step": 8609
    },
    {
      "epoch": 0.6516248048815099,
      "grad_norm": 0.8046875,
      "learning_rate": 1.5237369990861486e-05,
      "loss": 0.3629,
      "step": 8610
    },
    {
      "epoch": 0.6517004872049572,
      "grad_norm": 0.8046875,
      "learning_rate": 1.5236355296453268e-05,
      "loss": 0.3374,
      "step": 8611
    },
    {
      "epoch": 0.6517761695284046,
      "grad_norm": 0.75,
      "learning_rate": 1.5235340527758364e-05,
      "loss": 0.2794,
      "step": 8612
    },
    {
      "epoch": 0.6518518518518519,
      "grad_norm": 0.765625,
      "learning_rate": 1.5234325684791173e-05,
      "loss": 0.2804,
      "step": 8613
    },
    {
      "epoch": 0.6519275341752991,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5233310767566086e-05,
      "loss": 0.3067,
      "step": 8614
    },
    {
      "epoch": 0.6520032164987465,
      "grad_norm": 0.71484375,
      "learning_rate": 1.5232295776097505e-05,
      "loss": 0.2752,
      "step": 8615
    },
    {
      "epoch": 0.6520788988221938,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5231280710399832e-05,
      "loss": 0.3227,
      "step": 8616
    },
    {
      "epoch": 0.6521545811456412,
      "grad_norm": 0.76171875,
      "learning_rate": 1.523026557048746e-05,
      "loss": 0.2915,
      "step": 8617
    },
    {
      "epoch": 0.6522302634690885,
      "grad_norm": 0.82421875,
      "learning_rate": 1.5229250356374804e-05,
      "loss": 0.3339,
      "step": 8618
    },
    {
      "epoch": 0.6523059457925359,
      "grad_norm": 0.765625,
      "learning_rate": 1.5228235068076253e-05,
      "loss": 0.3381,
      "step": 8619
    },
    {
      "epoch": 0.6523816281159832,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5227219705606215e-05,
      "loss": 0.3289,
      "step": 8620
    },
    {
      "epoch": 0.6524573104394304,
      "grad_norm": 0.72265625,
      "learning_rate": 1.5226204268979096e-05,
      "loss": 0.3022,
      "step": 8621
    },
    {
      "epoch": 0.6525329927628778,
      "grad_norm": 0.75,
      "learning_rate": 1.5225188758209303e-05,
      "loss": 0.319,
      "step": 8622
    },
    {
      "epoch": 0.6526086750863251,
      "grad_norm": 0.77734375,
      "learning_rate": 1.5224173173311239e-05,
      "loss": 0.3477,
      "step": 8623
    },
    {
      "epoch": 0.6526843574097725,
      "grad_norm": 0.7578125,
      "learning_rate": 1.5223157514299311e-05,
      "loss": 0.344,
      "step": 8624
    },
    {
      "epoch": 0.6527600397332198,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5222141781187936e-05,
      "loss": 0.3271,
      "step": 8625
    },
    {
      "epoch": 0.6528357220566672,
      "grad_norm": 0.8359375,
      "learning_rate": 1.5221125973991513e-05,
      "loss": 0.3646,
      "step": 8626
    },
    {
      "epoch": 0.6529114043801145,
      "grad_norm": 0.79296875,
      "learning_rate": 1.5220110092724462e-05,
      "loss": 0.3291,
      "step": 8627
    },
    {
      "epoch": 0.6529870867035618,
      "grad_norm": 1.015625,
      "learning_rate": 1.5219094137401192e-05,
      "loss": 0.3551,
      "step": 8628
    },
    {
      "epoch": 0.6530627690270091,
      "grad_norm": 0.7578125,
      "learning_rate": 1.5218078108036109e-05,
      "loss": 0.3413,
      "step": 8629
    },
    {
      "epoch": 0.6531384513504565,
      "grad_norm": 0.83203125,
      "learning_rate": 1.521706200464364e-05,
      "loss": 0.356,
      "step": 8630
    },
    {
      "epoch": 0.6532141336739038,
      "grad_norm": 0.7890625,
      "learning_rate": 1.521604582723819e-05,
      "loss": 0.3353,
      "step": 8631
    },
    {
      "epoch": 0.6532898159973511,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5215029575834182e-05,
      "loss": 0.3198,
      "step": 8632
    },
    {
      "epoch": 0.6533654983207985,
      "grad_norm": 0.83203125,
      "learning_rate": 1.5214013250446029e-05,
      "loss": 0.3576,
      "step": 8633
    },
    {
      "epoch": 0.6534411806442458,
      "grad_norm": 0.71484375,
      "learning_rate": 1.5212996851088149e-05,
      "loss": 0.2675,
      "step": 8634
    },
    {
      "epoch": 0.6535168629676931,
      "grad_norm": 0.7890625,
      "learning_rate": 1.5211980377774961e-05,
      "loss": 0.33,
      "step": 8635
    },
    {
      "epoch": 0.6535925452911404,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5210963830520893e-05,
      "loss": 0.3266,
      "step": 8636
    },
    {
      "epoch": 0.6536682276145878,
      "grad_norm": 0.875,
      "learning_rate": 1.5209947209340356e-05,
      "loss": 0.3746,
      "step": 8637
    },
    {
      "epoch": 0.6537439099380351,
      "grad_norm": 0.82421875,
      "learning_rate": 1.520893051424778e-05,
      "loss": 0.3701,
      "step": 8638
    },
    {
      "epoch": 0.6538195922614825,
      "grad_norm": 0.71875,
      "learning_rate": 1.5207913745257586e-05,
      "loss": 0.2951,
      "step": 8639
    },
    {
      "epoch": 0.6538952745849298,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5206896902384196e-05,
      "loss": 0.2875,
      "step": 8640
    },
    {
      "epoch": 0.653970956908377,
      "grad_norm": 0.74609375,
      "learning_rate": 1.5205879985642041e-05,
      "loss": 0.2936,
      "step": 8641
    },
    {
      "epoch": 0.6540466392318244,
      "grad_norm": 0.8515625,
      "learning_rate": 1.5204862995045541e-05,
      "loss": 0.4014,
      "step": 8642
    },
    {
      "epoch": 0.6541223215552717,
      "grad_norm": 0.75390625,
      "learning_rate": 1.520384593060913e-05,
      "loss": 0.3136,
      "step": 8643
    },
    {
      "epoch": 0.6541980038787191,
      "grad_norm": 0.78515625,
      "learning_rate": 1.5202828792347232e-05,
      "loss": 0.3495,
      "step": 8644
    },
    {
      "epoch": 0.6542736862021664,
      "grad_norm": 0.8125,
      "learning_rate": 1.5201811580274284e-05,
      "loss": 0.3361,
      "step": 8645
    },
    {
      "epoch": 0.6543493685256138,
      "grad_norm": 0.80078125,
      "learning_rate": 1.5200794294404709e-05,
      "loss": 0.3442,
      "step": 8646
    },
    {
      "epoch": 0.6544250508490611,
      "grad_norm": 0.7265625,
      "learning_rate": 1.5199776934752944e-05,
      "loss": 0.2882,
      "step": 8647
    },
    {
      "epoch": 0.6545007331725083,
      "grad_norm": 0.69140625,
      "learning_rate": 1.5198759501333418e-05,
      "loss": 0.2784,
      "step": 8648
    },
    {
      "epoch": 0.6545764154959557,
      "grad_norm": 0.72265625,
      "learning_rate": 1.5197741994160569e-05,
      "loss": 0.3189,
      "step": 8649
    },
    {
      "epoch": 0.654652097819403,
      "grad_norm": 0.80078125,
      "learning_rate": 1.5196724413248829e-05,
      "loss": 0.3631,
      "step": 8650
    },
    {
      "epoch": 0.6547277801428504,
      "grad_norm": 0.7109375,
      "learning_rate": 1.5195706758612634e-05,
      "loss": 0.3053,
      "step": 8651
    },
    {
      "epoch": 0.6548034624662977,
      "grad_norm": 0.69921875,
      "learning_rate": 1.5194689030266422e-05,
      "loss": 0.2893,
      "step": 8652
    },
    {
      "epoch": 0.6548791447897451,
      "grad_norm": 0.83203125,
      "learning_rate": 1.5193671228224634e-05,
      "loss": 0.3768,
      "step": 8653
    },
    {
      "epoch": 0.6549548271131924,
      "grad_norm": 0.828125,
      "learning_rate": 1.5192653352501707e-05,
      "loss": 0.3397,
      "step": 8654
    },
    {
      "epoch": 0.6550305094366397,
      "grad_norm": 0.7578125,
      "learning_rate": 1.5191635403112081e-05,
      "loss": 0.3478,
      "step": 8655
    },
    {
      "epoch": 0.655106191760087,
      "grad_norm": 0.8359375,
      "learning_rate": 1.5190617380070199e-05,
      "loss": 0.3399,
      "step": 8656
    },
    {
      "epoch": 0.6551818740835343,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5189599283390499e-05,
      "loss": 0.3113,
      "step": 8657
    },
    {
      "epoch": 0.6552575564069817,
      "grad_norm": 0.75,
      "learning_rate": 1.518858111308743e-05,
      "loss": 0.3121,
      "step": 8658
    },
    {
      "epoch": 0.655333238730429,
      "grad_norm": 0.7578125,
      "learning_rate": 1.518756286917543e-05,
      "loss": 0.3368,
      "step": 8659
    },
    {
      "epoch": 0.6554089210538764,
      "grad_norm": 0.79296875,
      "learning_rate": 1.5186544551668955e-05,
      "loss": 0.3514,
      "step": 8660
    },
    {
      "epoch": 0.6554846033773237,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5185526160582437e-05,
      "loss": 0.2902,
      "step": 8661
    },
    {
      "epoch": 0.655560285700771,
      "grad_norm": 0.78515625,
      "learning_rate": 1.518450769593034e-05,
      "loss": 0.3281,
      "step": 8662
    },
    {
      "epoch": 0.6556359680242183,
      "grad_norm": 0.80078125,
      "learning_rate": 1.5183489157727099e-05,
      "loss": 0.3365,
      "step": 8663
    },
    {
      "epoch": 0.6557116503476657,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5182470545987174e-05,
      "loss": 0.3203,
      "step": 8664
    },
    {
      "epoch": 0.655787332671113,
      "grad_norm": 0.703125,
      "learning_rate": 1.5181451860725007e-05,
      "loss": 0.282,
      "step": 8665
    },
    {
      "epoch": 0.6558630149945603,
      "grad_norm": 0.75,
      "learning_rate": 1.5180433101955055e-05,
      "loss": 0.2983,
      "step": 8666
    },
    {
      "epoch": 0.6559386973180077,
      "grad_norm": 0.81640625,
      "learning_rate": 1.5179414269691766e-05,
      "loss": 0.3716,
      "step": 8667
    },
    {
      "epoch": 0.656014379641455,
      "grad_norm": 0.70703125,
      "learning_rate": 1.5178395363949601e-05,
      "loss": 0.3059,
      "step": 8668
    },
    {
      "epoch": 0.6560900619649023,
      "grad_norm": 0.74609375,
      "learning_rate": 1.5177376384743013e-05,
      "loss": 0.3382,
      "step": 8669
    },
    {
      "epoch": 0.6561657442883496,
      "grad_norm": 0.8203125,
      "learning_rate": 1.5176357332086451e-05,
      "loss": 0.3403,
      "step": 8670
    },
    {
      "epoch": 0.656241426611797,
      "grad_norm": 0.7578125,
      "learning_rate": 1.5175338205994382e-05,
      "loss": 0.3135,
      "step": 8671
    },
    {
      "epoch": 0.6563171089352443,
      "grad_norm": 0.78515625,
      "learning_rate": 1.5174319006481257e-05,
      "loss": 0.3502,
      "step": 8672
    },
    {
      "epoch": 0.6563927912586917,
      "grad_norm": 0.7578125,
      "learning_rate": 1.5173299733561537e-05,
      "loss": 0.3235,
      "step": 8673
    },
    {
      "epoch": 0.656468473582139,
      "grad_norm": 0.703125,
      "learning_rate": 1.5172280387249685e-05,
      "loss": 0.2852,
      "step": 8674
    },
    {
      "epoch": 0.6565441559055863,
      "grad_norm": 0.76953125,
      "learning_rate": 1.517126096756016e-05,
      "loss": 0.33,
      "step": 8675
    },
    {
      "epoch": 0.6566198382290336,
      "grad_norm": 0.78125,
      "learning_rate": 1.5170241474507421e-05,
      "loss": 0.3349,
      "step": 8676
    },
    {
      "epoch": 0.6566955205524809,
      "grad_norm": 0.80078125,
      "learning_rate": 1.5169221908105933e-05,
      "loss": 0.3764,
      "step": 8677
    },
    {
      "epoch": 0.6567712028759283,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5168202268370169e-05,
      "loss": 0.3227,
      "step": 8678
    },
    {
      "epoch": 0.6568468851993756,
      "grad_norm": 0.859375,
      "learning_rate": 1.5167182555314578e-05,
      "loss": 0.405,
      "step": 8679
    },
    {
      "epoch": 0.656922567522823,
      "grad_norm": 0.73828125,
      "learning_rate": 1.516616276895364e-05,
      "loss": 0.2942,
      "step": 8680
    },
    {
      "epoch": 0.6569982498462703,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5165142909301817e-05,
      "loss": 0.294,
      "step": 8681
    },
    {
      "epoch": 0.6570739321697177,
      "grad_norm": 0.7578125,
      "learning_rate": 1.5164122976373579e-05,
      "loss": 0.3076,
      "step": 8682
    },
    {
      "epoch": 0.6571496144931649,
      "grad_norm": 0.80859375,
      "learning_rate": 1.5163102970183397e-05,
      "loss": 0.3506,
      "step": 8683
    },
    {
      "epoch": 0.6572252968166122,
      "grad_norm": 0.828125,
      "learning_rate": 1.5162082890745735e-05,
      "loss": 0.3619,
      "step": 8684
    },
    {
      "epoch": 0.6573009791400596,
      "grad_norm": 0.765625,
      "learning_rate": 1.5161062738075068e-05,
      "loss": 0.316,
      "step": 8685
    },
    {
      "epoch": 0.6573766614635069,
      "grad_norm": 0.78125,
      "learning_rate": 1.516004251218587e-05,
      "loss": 0.335,
      "step": 8686
    },
    {
      "epoch": 0.6574523437869543,
      "grad_norm": 0.78125,
      "learning_rate": 1.5159022213092615e-05,
      "loss": 0.3528,
      "step": 8687
    },
    {
      "epoch": 0.6575280261104016,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5158001840809776e-05,
      "loss": 0.3279,
      "step": 8688
    },
    {
      "epoch": 0.657603708433849,
      "grad_norm": 0.71484375,
      "learning_rate": 1.5156981395351832e-05,
      "loss": 0.2753,
      "step": 8689
    },
    {
      "epoch": 0.6576793907572962,
      "grad_norm": 0.74609375,
      "learning_rate": 1.5155960876733255e-05,
      "loss": 0.2701,
      "step": 8690
    },
    {
      "epoch": 0.6577550730807435,
      "grad_norm": 0.828125,
      "learning_rate": 1.5154940284968525e-05,
      "loss": 0.3818,
      "step": 8691
    },
    {
      "epoch": 0.6578307554041909,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5153919620072123e-05,
      "loss": 0.3242,
      "step": 8692
    },
    {
      "epoch": 0.6579064377276382,
      "grad_norm": 1.125,
      "learning_rate": 1.5152898882058527e-05,
      "loss": 0.3773,
      "step": 8693
    },
    {
      "epoch": 0.6579821200510856,
      "grad_norm": 0.7265625,
      "learning_rate": 1.5151878070942214e-05,
      "loss": 0.3019,
      "step": 8694
    },
    {
      "epoch": 0.6580578023745329,
      "grad_norm": 0.71875,
      "learning_rate": 1.5150857186737673e-05,
      "loss": 0.2904,
      "step": 8695
    },
    {
      "epoch": 0.6581334846979803,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5149836229459381e-05,
      "loss": 0.3245,
      "step": 8696
    },
    {
      "epoch": 0.6582091670214275,
      "grad_norm": 0.734375,
      "learning_rate": 1.5148815199121827e-05,
      "loss": 0.3332,
      "step": 8697
    },
    {
      "epoch": 0.6582848493448749,
      "grad_norm": 0.68359375,
      "learning_rate": 1.5147794095739491e-05,
      "loss": 0.2875,
      "step": 8698
    },
    {
      "epoch": 0.6583605316683222,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5146772919326864e-05,
      "loss": 0.2946,
      "step": 8699
    },
    {
      "epoch": 0.6584362139917695,
      "grad_norm": 0.75,
      "learning_rate": 1.514575166989843e-05,
      "loss": 0.3079,
      "step": 8700
    },
    {
      "epoch": 0.6585118963152169,
      "grad_norm": 0.7421875,
      "learning_rate": 1.514473034746868e-05,
      "loss": 0.3216,
      "step": 8701
    },
    {
      "epoch": 0.6585875786386642,
      "grad_norm": 0.78125,
      "learning_rate": 1.5143708952052099e-05,
      "loss": 0.3358,
      "step": 8702
    },
    {
      "epoch": 0.6586632609621116,
      "grad_norm": 0.765625,
      "learning_rate": 1.5142687483663178e-05,
      "loss": 0.3059,
      "step": 8703
    },
    {
      "epoch": 0.6587389432855588,
      "grad_norm": 0.74609375,
      "learning_rate": 1.5141665942316414e-05,
      "loss": 0.3083,
      "step": 8704
    },
    {
      "epoch": 0.6588146256090062,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5140644328026295e-05,
      "loss": 0.3384,
      "step": 8705
    },
    {
      "epoch": 0.6588903079324535,
      "grad_norm": 0.7890625,
      "learning_rate": 1.5139622640807315e-05,
      "loss": 0.3267,
      "step": 8706
    },
    {
      "epoch": 0.6589659902559009,
      "grad_norm": 0.77734375,
      "learning_rate": 1.5138600880673963e-05,
      "loss": 0.2958,
      "step": 8707
    },
    {
      "epoch": 0.6590416725793482,
      "grad_norm": 0.765625,
      "learning_rate": 1.5137579047640743e-05,
      "loss": 0.3097,
      "step": 8708
    },
    {
      "epoch": 0.6591173549027955,
      "grad_norm": 0.765625,
      "learning_rate": 1.5136557141722148e-05,
      "loss": 0.3365,
      "step": 8709
    },
    {
      "epoch": 0.6591930372262429,
      "grad_norm": 0.80078125,
      "learning_rate": 1.5135535162932673e-05,
      "loss": 0.3321,
      "step": 8710
    },
    {
      "epoch": 0.6592687195496901,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5134513111286822e-05,
      "loss": 0.314,
      "step": 8711
    },
    {
      "epoch": 0.6593444018731375,
      "grad_norm": 0.78515625,
      "learning_rate": 1.513349098679909e-05,
      "loss": 0.3334,
      "step": 8712
    },
    {
      "epoch": 0.6594200841965848,
      "grad_norm": 0.765625,
      "learning_rate": 1.5132468789483979e-05,
      "loss": 0.3292,
      "step": 8713
    },
    {
      "epoch": 0.6594957665200322,
      "grad_norm": 0.8125,
      "learning_rate": 1.5131446519355988e-05,
      "loss": 0.3062,
      "step": 8714
    },
    {
      "epoch": 0.6595714488434795,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5130424176429625e-05,
      "loss": 0.3464,
      "step": 8715
    },
    {
      "epoch": 0.6596471311669269,
      "grad_norm": 0.80859375,
      "learning_rate": 1.5129401760719393e-05,
      "loss": 0.3325,
      "step": 8716
    },
    {
      "epoch": 0.6597228134903742,
      "grad_norm": 0.7890625,
      "learning_rate": 1.5128379272239791e-05,
      "loss": 0.3521,
      "step": 8717
    },
    {
      "epoch": 0.6597984958138214,
      "grad_norm": 0.8046875,
      "learning_rate": 1.5127356711005327e-05,
      "loss": 0.3433,
      "step": 8718
    },
    {
      "epoch": 0.6598741781372688,
      "grad_norm": 0.890625,
      "learning_rate": 1.5126334077030515e-05,
      "loss": 0.3261,
      "step": 8719
    },
    {
      "epoch": 0.6599498604607161,
      "grad_norm": 0.72265625,
      "learning_rate": 1.5125311370329854e-05,
      "loss": 0.3002,
      "step": 8720
    },
    {
      "epoch": 0.6600255427841635,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5124288590917857e-05,
      "loss": 0.274,
      "step": 8721
    },
    {
      "epoch": 0.6601012251076108,
      "grad_norm": 0.77734375,
      "learning_rate": 1.5123265738809032e-05,
      "loss": 0.3118,
      "step": 8722
    },
    {
      "epoch": 0.6601769074310582,
      "grad_norm": 0.74609375,
      "learning_rate": 1.5122242814017887e-05,
      "loss": 0.3059,
      "step": 8723
    },
    {
      "epoch": 0.6602525897545055,
      "grad_norm": 0.7578125,
      "learning_rate": 1.512121981655894e-05,
      "loss": 0.3246,
      "step": 8724
    },
    {
      "epoch": 0.6603282720779527,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5120196746446704e-05,
      "loss": 0.3061,
      "step": 8725
    },
    {
      "epoch": 0.6604039544014001,
      "grad_norm": 0.765625,
      "learning_rate": 1.511917360369569e-05,
      "loss": 0.3292,
      "step": 8726
    },
    {
      "epoch": 0.6604796367248474,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5118150388320412e-05,
      "loss": 0.2917,
      "step": 8727
    },
    {
      "epoch": 0.6605553190482948,
      "grad_norm": 0.8125,
      "learning_rate": 1.511712710033539e-05,
      "loss": 0.3323,
      "step": 8728
    },
    {
      "epoch": 0.6606310013717421,
      "grad_norm": 1.0546875,
      "learning_rate": 1.5116103739755134e-05,
      "loss": 0.3264,
      "step": 8729
    },
    {
      "epoch": 0.6607066836951895,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5115080306594172e-05,
      "loss": 0.3178,
      "step": 8730
    },
    {
      "epoch": 0.6607823660186368,
      "grad_norm": 0.703125,
      "learning_rate": 1.5114056800867017e-05,
      "loss": 0.281,
      "step": 8731
    },
    {
      "epoch": 0.6608580483420841,
      "grad_norm": 0.69921875,
      "learning_rate": 1.5113033222588188e-05,
      "loss": 0.2846,
      "step": 8732
    },
    {
      "epoch": 0.6609337306655314,
      "grad_norm": 0.78515625,
      "learning_rate": 1.5112009571772208e-05,
      "loss": 0.3669,
      "step": 8733
    },
    {
      "epoch": 0.6610094129889788,
      "grad_norm": 0.8515625,
      "learning_rate": 1.5110985848433606e-05,
      "loss": 0.3583,
      "step": 8734
    },
    {
      "epoch": 0.6610850953124261,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5109962052586892e-05,
      "loss": 0.3153,
      "step": 8735
    },
    {
      "epoch": 0.6611607776358734,
      "grad_norm": 0.84375,
      "learning_rate": 1.5108938184246599e-05,
      "loss": 0.3768,
      "step": 8736
    },
    {
      "epoch": 0.6612364599593208,
      "grad_norm": 0.7890625,
      "learning_rate": 1.5107914243427253e-05,
      "loss": 0.343,
      "step": 8737
    },
    {
      "epoch": 0.6613121422827681,
      "grad_norm": 0.765625,
      "learning_rate": 1.5106890230143374e-05,
      "loss": 0.3253,
      "step": 8738
    },
    {
      "epoch": 0.6613878246062154,
      "grad_norm": 1.2734375,
      "learning_rate": 1.5105866144409498e-05,
      "loss": 0.3839,
      "step": 8739
    },
    {
      "epoch": 0.6614635069296627,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5104841986240146e-05,
      "loss": 0.31,
      "step": 8740
    },
    {
      "epoch": 0.6615391892531101,
      "grad_norm": 0.7109375,
      "learning_rate": 1.510381775564985e-05,
      "loss": 0.2873,
      "step": 8741
    },
    {
      "epoch": 0.6616148715765574,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5102793452653141e-05,
      "loss": 0.3344,
      "step": 8742
    },
    {
      "epoch": 0.6616905539000048,
      "grad_norm": 0.94921875,
      "learning_rate": 1.5101769077264552e-05,
      "loss": 0.3168,
      "step": 8743
    },
    {
      "epoch": 0.6617662362234521,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5100744629498611e-05,
      "loss": 0.3063,
      "step": 8744
    },
    {
      "epoch": 0.6618419185468994,
      "grad_norm": 0.73828125,
      "learning_rate": 1.5099720109369858e-05,
      "loss": 0.3212,
      "step": 8745
    },
    {
      "epoch": 0.6619176008703467,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5098695516892821e-05,
      "loss": 0.3061,
      "step": 8746
    },
    {
      "epoch": 0.661993283193794,
      "grad_norm": 1.09375,
      "learning_rate": 1.5097670852082039e-05,
      "loss": 0.3684,
      "step": 8747
    },
    {
      "epoch": 0.6620689655172414,
      "grad_norm": 0.81640625,
      "learning_rate": 1.5096646114952047e-05,
      "loss": 0.33,
      "step": 8748
    },
    {
      "epoch": 0.6621446478406887,
      "grad_norm": 0.671875,
      "learning_rate": 1.5095621305517387e-05,
      "loss": 0.273,
      "step": 8749
    },
    {
      "epoch": 0.6622203301641361,
      "grad_norm": 0.80859375,
      "learning_rate": 1.509459642379259e-05,
      "loss": 0.3253,
      "step": 8750
    },
    {
      "epoch": 0.6622960124875834,
      "grad_norm": 0.83984375,
      "learning_rate": 1.50935714697922e-05,
      "loss": 0.3624,
      "step": 8751
    },
    {
      "epoch": 0.6623716948110308,
      "grad_norm": 0.71875,
      "learning_rate": 1.5092546443530762e-05,
      "loss": 0.2687,
      "step": 8752
    },
    {
      "epoch": 0.662447377134478,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5091521345022813e-05,
      "loss": 0.3394,
      "step": 8753
    },
    {
      "epoch": 0.6625230594579253,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5090496174282893e-05,
      "loss": 0.3197,
      "step": 8754
    },
    {
      "epoch": 0.6625987417813727,
      "grad_norm": 0.80859375,
      "learning_rate": 1.5089470931325551e-05,
      "loss": 0.3518,
      "step": 8755
    },
    {
      "epoch": 0.66267442410482,
      "grad_norm": 0.78515625,
      "learning_rate": 1.508844561616533e-05,
      "loss": 0.31,
      "step": 8756
    },
    {
      "epoch": 0.6627501064282674,
      "grad_norm": 0.75,
      "learning_rate": 1.5087420228816776e-05,
      "loss": 0.2984,
      "step": 8757
    },
    {
      "epoch": 0.6628257887517147,
      "grad_norm": 0.765625,
      "learning_rate": 1.5086394769294437e-05,
      "loss": 0.3256,
      "step": 8758
    },
    {
      "epoch": 0.662901471075162,
      "grad_norm": 0.69921875,
      "learning_rate": 1.5085369237612857e-05,
      "loss": 0.2869,
      "step": 8759
    },
    {
      "epoch": 0.6629771533986093,
      "grad_norm": 0.8203125,
      "learning_rate": 1.5084343633786592e-05,
      "loss": 0.3492,
      "step": 8760
    },
    {
      "epoch": 0.6630528357220566,
      "grad_norm": 0.734375,
      "learning_rate": 1.5083317957830183e-05,
      "loss": 0.2951,
      "step": 8761
    },
    {
      "epoch": 0.663128518045504,
      "grad_norm": 0.75,
      "learning_rate": 1.5082292209758186e-05,
      "loss": 0.33,
      "step": 8762
    },
    {
      "epoch": 0.6632042003689513,
      "grad_norm": 0.75,
      "learning_rate": 1.5081266389585155e-05,
      "loss": 0.2982,
      "step": 8763
    },
    {
      "epoch": 0.6632798826923987,
      "grad_norm": 1.0390625,
      "learning_rate": 1.5080240497325638e-05,
      "loss": 0.3924,
      "step": 8764
    },
    {
      "epoch": 0.663355565015846,
      "grad_norm": 0.765625,
      "learning_rate": 1.507921453299419e-05,
      "loss": 0.2954,
      "step": 8765
    },
    {
      "epoch": 0.6634312473392933,
      "grad_norm": 0.80078125,
      "learning_rate": 1.507818849660537e-05,
      "loss": 0.3249,
      "step": 8766
    },
    {
      "epoch": 0.6635069296627406,
      "grad_norm": 0.80859375,
      "learning_rate": 1.5077162388173734e-05,
      "loss": 0.3522,
      "step": 8767
    },
    {
      "epoch": 0.663582611986188,
      "grad_norm": 0.84375,
      "learning_rate": 1.5076136207713834e-05,
      "loss": 0.3411,
      "step": 8768
    },
    {
      "epoch": 0.6636582943096353,
      "grad_norm": 0.74609375,
      "learning_rate": 1.5075109955240233e-05,
      "loss": 0.3046,
      "step": 8769
    },
    {
      "epoch": 0.6637339766330826,
      "grad_norm": 0.7109375,
      "learning_rate": 1.5074083630767484e-05,
      "loss": 0.3137,
      "step": 8770
    },
    {
      "epoch": 0.66380965895653,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5073057234310154e-05,
      "loss": 0.3302,
      "step": 8771
    },
    {
      "epoch": 0.6638853412799773,
      "grad_norm": 0.78515625,
      "learning_rate": 1.5072030765882804e-05,
      "loss": 0.337,
      "step": 8772
    },
    {
      "epoch": 0.6639610236034246,
      "grad_norm": 0.80859375,
      "learning_rate": 1.507100422549999e-05,
      "loss": 0.325,
      "step": 8773
    },
    {
      "epoch": 0.6640367059268719,
      "grad_norm": 0.6953125,
      "learning_rate": 1.5069977613176278e-05,
      "loss": 0.27,
      "step": 8774
    },
    {
      "epoch": 0.6641123882503193,
      "grad_norm": 0.7578125,
      "learning_rate": 1.5068950928926236e-05,
      "loss": 0.331,
      "step": 8775
    },
    {
      "epoch": 0.6641880705737666,
      "grad_norm": 0.828125,
      "learning_rate": 1.5067924172764427e-05,
      "loss": 0.3698,
      "step": 8776
    },
    {
      "epoch": 0.664263752897214,
      "grad_norm": 0.7578125,
      "learning_rate": 1.5066897344705416e-05,
      "loss": 0.3131,
      "step": 8777
    },
    {
      "epoch": 0.6643394352206613,
      "grad_norm": 0.828125,
      "learning_rate": 1.506587044476377e-05,
      "loss": 0.3604,
      "step": 8778
    },
    {
      "epoch": 0.6644151175441086,
      "grad_norm": 0.80859375,
      "learning_rate": 1.5064843472954058e-05,
      "loss": 0.3765,
      "step": 8779
    },
    {
      "epoch": 0.6644907998675559,
      "grad_norm": 0.76171875,
      "learning_rate": 1.506381642929085e-05,
      "loss": 0.3473,
      "step": 8780
    },
    {
      "epoch": 0.6645664821910032,
      "grad_norm": 0.7265625,
      "learning_rate": 1.506278931378872e-05,
      "loss": 0.283,
      "step": 8781
    },
    {
      "epoch": 0.6646421645144506,
      "grad_norm": 0.73828125,
      "learning_rate": 1.5061762126462229e-05,
      "loss": 0.2904,
      "step": 8782
    },
    {
      "epoch": 0.6647178468378979,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5060734867325959e-05,
      "loss": 0.3004,
      "step": 8783
    },
    {
      "epoch": 0.6647935291613453,
      "grad_norm": 0.7578125,
      "learning_rate": 1.5059707536394483e-05,
      "loss": 0.3254,
      "step": 8784
    },
    {
      "epoch": 0.6648692114847926,
      "grad_norm": 0.796875,
      "learning_rate": 1.5058680133682369e-05,
      "loss": 0.3472,
      "step": 8785
    },
    {
      "epoch": 0.66494489380824,
      "grad_norm": 0.78515625,
      "learning_rate": 1.5057652659204198e-05,
      "loss": 0.3572,
      "step": 8786
    },
    {
      "epoch": 0.6650205761316872,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5056625112974545e-05,
      "loss": 0.3213,
      "step": 8787
    },
    {
      "epoch": 0.6650962584551345,
      "grad_norm": 0.69140625,
      "learning_rate": 1.5055597495007988e-05,
      "loss": 0.253,
      "step": 8788
    },
    {
      "epoch": 0.6651719407785819,
      "grad_norm": 0.72265625,
      "learning_rate": 1.5054569805319102e-05,
      "loss": 0.2993,
      "step": 8789
    },
    {
      "epoch": 0.6652476231020292,
      "grad_norm": 0.7265625,
      "learning_rate": 1.5053542043922474e-05,
      "loss": 0.2983,
      "step": 8790
    },
    {
      "epoch": 0.6653233054254766,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5052514210832674e-05,
      "loss": 0.3395,
      "step": 8791
    },
    {
      "epoch": 0.6653989877489239,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5051486306064293e-05,
      "loss": 0.3351,
      "step": 8792
    },
    {
      "epoch": 0.6654746700723713,
      "grad_norm": 0.83984375,
      "learning_rate": 1.505045832963191e-05,
      "loss": 0.3768,
      "step": 8793
    },
    {
      "epoch": 0.6655503523958185,
      "grad_norm": 0.828125,
      "learning_rate": 1.5049430281550108e-05,
      "loss": 0.3791,
      "step": 8794
    },
    {
      "epoch": 0.6656260347192658,
      "grad_norm": 0.73046875,
      "learning_rate": 1.5048402161833475e-05,
      "loss": 0.2861,
      "step": 8795
    },
    {
      "epoch": 0.6657017170427132,
      "grad_norm": 0.69921875,
      "learning_rate": 1.504737397049659e-05,
      "loss": 0.2589,
      "step": 8796
    },
    {
      "epoch": 0.6657773993661605,
      "grad_norm": 0.77734375,
      "learning_rate": 1.5046345707554044e-05,
      "loss": 0.3327,
      "step": 8797
    },
    {
      "epoch": 0.6658530816896079,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5045317373020428e-05,
      "loss": 0.3547,
      "step": 8798
    },
    {
      "epoch": 0.6659287640130552,
      "grad_norm": 0.78125,
      "learning_rate": 1.5044288966910323e-05,
      "loss": 0.3293,
      "step": 8799
    },
    {
      "epoch": 0.6660044463365026,
      "grad_norm": 0.734375,
      "learning_rate": 1.5043260489238325e-05,
      "loss": 0.2974,
      "step": 8800
    },
    {
      "epoch": 0.6660801286599498,
      "grad_norm": 0.77734375,
      "learning_rate": 1.5042231940019023e-05,
      "loss": 0.3408,
      "step": 8801
    },
    {
      "epoch": 0.6661558109833972,
      "grad_norm": 0.734375,
      "learning_rate": 1.5041203319267007e-05,
      "loss": 0.3112,
      "step": 8802
    },
    {
      "epoch": 0.6662314933068445,
      "grad_norm": 0.80078125,
      "learning_rate": 1.5040174626996868e-05,
      "loss": 0.3467,
      "step": 8803
    },
    {
      "epoch": 0.6663071756302918,
      "grad_norm": 0.99609375,
      "learning_rate": 1.5039145863223207e-05,
      "loss": 0.3562,
      "step": 8804
    },
    {
      "epoch": 0.6663828579537392,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5038117027960613e-05,
      "loss": 0.3134,
      "step": 8805
    },
    {
      "epoch": 0.6664585402771865,
      "grad_norm": 0.78515625,
      "learning_rate": 1.5037088121223682e-05,
      "loss": 0.3165,
      "step": 8806
    },
    {
      "epoch": 0.6665342226006339,
      "grad_norm": 0.75390625,
      "learning_rate": 1.5036059143027015e-05,
      "loss": 0.3325,
      "step": 8807
    },
    {
      "epoch": 0.6666099049240811,
      "grad_norm": 0.68359375,
      "learning_rate": 1.5035030093385206e-05,
      "loss": 0.2381,
      "step": 8808
    },
    {
      "epoch": 0.6666855872475285,
      "grad_norm": 0.7421875,
      "learning_rate": 1.5034000972312854e-05,
      "loss": 0.3142,
      "step": 8809
    },
    {
      "epoch": 0.6667612695709758,
      "grad_norm": 0.7265625,
      "learning_rate": 1.503297177982456e-05,
      "loss": 0.2879,
      "step": 8810
    },
    {
      "epoch": 0.6668369518944232,
      "grad_norm": 0.78125,
      "learning_rate": 1.5031942515934923e-05,
      "loss": 0.3352,
      "step": 8811
    },
    {
      "epoch": 0.6669126342178705,
      "grad_norm": 0.78515625,
      "learning_rate": 1.5030913180658546e-05,
      "loss": 0.3299,
      "step": 8812
    },
    {
      "epoch": 0.6669883165413178,
      "grad_norm": 0.88671875,
      "learning_rate": 1.5029883774010032e-05,
      "loss": 0.3803,
      "step": 8813
    },
    {
      "epoch": 0.6670639988647652,
      "grad_norm": 0.77734375,
      "learning_rate": 1.5028854296003987e-05,
      "loss": 0.3444,
      "step": 8814
    },
    {
      "epoch": 0.6671396811882124,
      "grad_norm": 0.796875,
      "learning_rate": 1.5027824746655016e-05,
      "loss": 0.3343,
      "step": 8815
    },
    {
      "epoch": 0.6672153635116598,
      "grad_norm": 0.8046875,
      "learning_rate": 1.502679512597772e-05,
      "loss": 0.3587,
      "step": 8816
    },
    {
      "epoch": 0.6672910458351071,
      "grad_norm": 0.77734375,
      "learning_rate": 1.502576543398671e-05,
      "loss": 0.3556,
      "step": 8817
    },
    {
      "epoch": 0.6673667281585545,
      "grad_norm": 0.78125,
      "learning_rate": 1.5024735670696591e-05,
      "loss": 0.3096,
      "step": 8818
    },
    {
      "epoch": 0.6674424104820018,
      "grad_norm": 0.77734375,
      "learning_rate": 1.5023705836121974e-05,
      "loss": 0.3013,
      "step": 8819
    },
    {
      "epoch": 0.6675180928054492,
      "grad_norm": 0.76953125,
      "learning_rate": 1.5022675930277468e-05,
      "loss": 0.3192,
      "step": 8820
    },
    {
      "epoch": 0.6675937751288965,
      "grad_norm": 0.7734375,
      "learning_rate": 1.5021645953177685e-05,
      "loss": 0.3097,
      "step": 8821
    },
    {
      "epoch": 0.6676694574523437,
      "grad_norm": 0.765625,
      "learning_rate": 1.5020615904837241e-05,
      "loss": 0.3207,
      "step": 8822
    },
    {
      "epoch": 0.6677451397757911,
      "grad_norm": 0.7109375,
      "learning_rate": 1.501958578527074e-05,
      "loss": 0.2821,
      "step": 8823
    },
    {
      "epoch": 0.6678208220992384,
      "grad_norm": 0.7265625,
      "learning_rate": 1.5018555594492802e-05,
      "loss": 0.3093,
      "step": 8824
    },
    {
      "epoch": 0.6678965044226858,
      "grad_norm": 0.77734375,
      "learning_rate": 1.5017525332518043e-05,
      "loss": 0.3315,
      "step": 8825
    },
    {
      "epoch": 0.6679721867461331,
      "grad_norm": 0.74609375,
      "learning_rate": 1.5016494999361074e-05,
      "loss": 0.3114,
      "step": 8826
    },
    {
      "epoch": 0.6680478690695805,
      "grad_norm": 0.78125,
      "learning_rate": 1.5015464595036516e-05,
      "loss": 0.3149,
      "step": 8827
    },
    {
      "epoch": 0.6681235513930278,
      "grad_norm": 0.80078125,
      "learning_rate": 1.5014434119558988e-05,
      "loss": 0.3441,
      "step": 8828
    },
    {
      "epoch": 0.668199233716475,
      "grad_norm": 0.7578125,
      "learning_rate": 1.5013403572943102e-05,
      "loss": 0.3334,
      "step": 8829
    },
    {
      "epoch": 0.6682749160399224,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5012372955203487e-05,
      "loss": 0.3126,
      "step": 8830
    },
    {
      "epoch": 0.6683505983633697,
      "grad_norm": 0.78125,
      "learning_rate": 1.5011342266354758e-05,
      "loss": 0.352,
      "step": 8831
    },
    {
      "epoch": 0.6684262806868171,
      "grad_norm": 0.73046875,
      "learning_rate": 1.501031150641154e-05,
      "loss": 0.3077,
      "step": 8832
    },
    {
      "epoch": 0.6685019630102644,
      "grad_norm": 0.734375,
      "learning_rate": 1.5009280675388455e-05,
      "loss": 0.3036,
      "step": 8833
    },
    {
      "epoch": 0.6685776453337118,
      "grad_norm": 0.76171875,
      "learning_rate": 1.5008249773300127e-05,
      "loss": 0.3134,
      "step": 8834
    },
    {
      "epoch": 0.6686533276571591,
      "grad_norm": 0.78125,
      "learning_rate": 1.5007218800161183e-05,
      "loss": 0.2865,
      "step": 8835
    },
    {
      "epoch": 0.6687290099806064,
      "grad_norm": 0.796875,
      "learning_rate": 1.5006187755986248e-05,
      "loss": 0.3204,
      "step": 8836
    },
    {
      "epoch": 0.6688046923040537,
      "grad_norm": 0.8125,
      "learning_rate": 1.5005156640789948e-05,
      "loss": 0.3503,
      "step": 8837
    },
    {
      "epoch": 0.668880374627501,
      "grad_norm": 0.7890625,
      "learning_rate": 1.500412545458691e-05,
      "loss": 0.3468,
      "step": 8838
    },
    {
      "epoch": 0.6689560569509484,
      "grad_norm": 0.78515625,
      "learning_rate": 1.5003094197391766e-05,
      "loss": 0.3654,
      "step": 8839
    },
    {
      "epoch": 0.6690317392743957,
      "grad_norm": 0.71875,
      "learning_rate": 1.5002062869219147e-05,
      "loss": 0.2717,
      "step": 8840
    },
    {
      "epoch": 0.6691074215978431,
      "grad_norm": 0.828125,
      "learning_rate": 1.500103147008368e-05,
      "loss": 0.3263,
      "step": 8841
    },
    {
      "epoch": 0.6691831039212904,
      "grad_norm": 0.75,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.3162,
      "step": 8842
    },
    {
      "epoch": 0.6692587862447377,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4998968458982743e-05,
      "loss": 0.298,
      "step": 8843
    },
    {
      "epoch": 0.669334468568185,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4997936847046535e-05,
      "loss": 0.3153,
      "step": 8844
    },
    {
      "epoch": 0.6694101508916324,
      "grad_norm": 0.8359375,
      "learning_rate": 1.499690516420602e-05,
      "loss": 0.3661,
      "step": 8845
    },
    {
      "epoch": 0.6694858332150797,
      "grad_norm": 0.86328125,
      "learning_rate": 1.499587341047583e-05,
      "loss": 0.3615,
      "step": 8846
    },
    {
      "epoch": 0.669561515538527,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4994841585870599e-05,
      "loss": 0.3312,
      "step": 8847
    },
    {
      "epoch": 0.6696371978619744,
      "grad_norm": 0.73046875,
      "learning_rate": 1.499380969040497e-05,
      "loss": 0.3069,
      "step": 8848
    },
    {
      "epoch": 0.6697128801854217,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4992777724093583e-05,
      "loss": 0.3107,
      "step": 8849
    },
    {
      "epoch": 0.669788562508869,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4991745686951071e-05,
      "loss": 0.3136,
      "step": 8850
    },
    {
      "epoch": 0.6698642448323163,
      "grad_norm": 0.7421875,
      "learning_rate": 1.499071357899209e-05,
      "loss": 0.2779,
      "step": 8851
    },
    {
      "epoch": 0.6699399271557637,
      "grad_norm": 0.8203125,
      "learning_rate": 1.4989681400231265e-05,
      "loss": 0.359,
      "step": 8852
    },
    {
      "epoch": 0.670015609479211,
      "grad_norm": 0.7109375,
      "learning_rate": 1.4988649150683246e-05,
      "loss": 0.2905,
      "step": 8853
    },
    {
      "epoch": 0.6700912918026584,
      "grad_norm": 0.76953125,
      "learning_rate": 1.498761683036268e-05,
      "loss": 0.3269,
      "step": 8854
    },
    {
      "epoch": 0.6701669741261057,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4986584439284212e-05,
      "loss": 0.3009,
      "step": 8855
    },
    {
      "epoch": 0.670242656449553,
      "grad_norm": 0.73046875,
      "learning_rate": 1.4985551977462484e-05,
      "loss": 0.2988,
      "step": 8856
    },
    {
      "epoch": 0.6703183387730003,
      "grad_norm": 0.65625,
      "learning_rate": 1.4984519444912147e-05,
      "loss": 0.2453,
      "step": 8857
    },
    {
      "epoch": 0.6703940210964476,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4983486841647847e-05,
      "loss": 0.3273,
      "step": 8858
    },
    {
      "epoch": 0.670469703419895,
      "grad_norm": 0.78515625,
      "learning_rate": 1.4982454167684233e-05,
      "loss": 0.3419,
      "step": 8859
    },
    {
      "epoch": 0.6705453857433423,
      "grad_norm": 0.8125,
      "learning_rate": 1.4981421423035963e-05,
      "loss": 0.3785,
      "step": 8860
    },
    {
      "epoch": 0.6706210680667897,
      "grad_norm": 0.8359375,
      "learning_rate": 1.4980388607717673e-05,
      "loss": 0.3793,
      "step": 8861
    },
    {
      "epoch": 0.670696750390237,
      "grad_norm": 0.83984375,
      "learning_rate": 1.497935572174403e-05,
      "loss": 0.3771,
      "step": 8862
    },
    {
      "epoch": 0.6707724327136844,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4978322765129678e-05,
      "loss": 0.3142,
      "step": 8863
    },
    {
      "epoch": 0.6708481150371316,
      "grad_norm": 0.73046875,
      "learning_rate": 1.4977289737889275e-05,
      "loss": 0.2922,
      "step": 8864
    },
    {
      "epoch": 0.6709237973605789,
      "grad_norm": 0.828125,
      "learning_rate": 1.4976256640037477e-05,
      "loss": 0.3774,
      "step": 8865
    },
    {
      "epoch": 0.6709994796840263,
      "grad_norm": 0.8203125,
      "learning_rate": 1.4975223471588937e-05,
      "loss": 0.3199,
      "step": 8866
    },
    {
      "epoch": 0.6710751620074736,
      "grad_norm": 0.765625,
      "learning_rate": 1.4974190232558317e-05,
      "loss": 0.3293,
      "step": 8867
    },
    {
      "epoch": 0.671150844330921,
      "grad_norm": 0.71875,
      "learning_rate": 1.497315692296027e-05,
      "loss": 0.2964,
      "step": 8868
    },
    {
      "epoch": 0.6712265266543683,
      "grad_norm": 0.82421875,
      "learning_rate": 1.4972123542809458e-05,
      "loss": 0.3577,
      "step": 8869
    },
    {
      "epoch": 0.6713022089778157,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4971090092120544e-05,
      "loss": 0.279,
      "step": 8870
    },
    {
      "epoch": 0.6713778913012629,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4970056570908184e-05,
      "loss": 0.307,
      "step": 8871
    },
    {
      "epoch": 0.6714535736247103,
      "grad_norm": 0.765625,
      "learning_rate": 1.4969022979187042e-05,
      "loss": 0.3323,
      "step": 8872
    },
    {
      "epoch": 0.6715292559481576,
      "grad_norm": 0.6796875,
      "learning_rate": 1.4967989316971783e-05,
      "loss": 0.2521,
      "step": 8873
    },
    {
      "epoch": 0.671604938271605,
      "grad_norm": 0.70703125,
      "learning_rate": 1.4966955584277072e-05,
      "loss": 0.3012,
      "step": 8874
    },
    {
      "epoch": 0.6716806205950523,
      "grad_norm": 0.69140625,
      "learning_rate": 1.496592178111757e-05,
      "loss": 0.2755,
      "step": 8875
    },
    {
      "epoch": 0.6717563029184996,
      "grad_norm": 0.796875,
      "learning_rate": 1.4964887907507946e-05,
      "loss": 0.383,
      "step": 8876
    },
    {
      "epoch": 0.671831985241947,
      "grad_norm": 0.78515625,
      "learning_rate": 1.4963853963462866e-05,
      "loss": 0.3373,
      "step": 8877
    },
    {
      "epoch": 0.6719076675653942,
      "grad_norm": 0.8671875,
      "learning_rate": 1.4962819948997e-05,
      "loss": 0.3832,
      "step": 8878
    },
    {
      "epoch": 0.6719833498888416,
      "grad_norm": 1.234375,
      "learning_rate": 1.4961785864125019e-05,
      "loss": 0.3807,
      "step": 8879
    },
    {
      "epoch": 0.6720590322122889,
      "grad_norm": 0.828125,
      "learning_rate": 1.4960751708861587e-05,
      "loss": 0.3716,
      "step": 8880
    },
    {
      "epoch": 0.6721347145357363,
      "grad_norm": 0.6953125,
      "learning_rate": 1.4959717483221381e-05,
      "loss": 0.2553,
      "step": 8881
    },
    {
      "epoch": 0.6722103968591836,
      "grad_norm": 0.75390625,
      "learning_rate": 1.495868318721907e-05,
      "loss": 0.3193,
      "step": 8882
    },
    {
      "epoch": 0.672286079182631,
      "grad_norm": 0.75390625,
      "learning_rate": 1.495764882086933e-05,
      "loss": 0.3057,
      "step": 8883
    },
    {
      "epoch": 0.6723617615060782,
      "grad_norm": 0.75390625,
      "learning_rate": 1.4956614384186837e-05,
      "loss": 0.3248,
      "step": 8884
    },
    {
      "epoch": 0.6724374438295255,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4955579877186257e-05,
      "loss": 0.3419,
      "step": 8885
    },
    {
      "epoch": 0.6725131261529729,
      "grad_norm": 0.7109375,
      "learning_rate": 1.4954545299882278e-05,
      "loss": 0.2945,
      "step": 8886
    },
    {
      "epoch": 0.6725888084764202,
      "grad_norm": 0.78515625,
      "learning_rate": 1.4953510652289567e-05,
      "loss": 0.3386,
      "step": 8887
    },
    {
      "epoch": 0.6726644907998676,
      "grad_norm": 0.828125,
      "learning_rate": 1.495247593442281e-05,
      "loss": 0.3447,
      "step": 8888
    },
    {
      "epoch": 0.6727401731233149,
      "grad_norm": 0.77734375,
      "learning_rate": 1.495144114629668e-05,
      "loss": 0.2899,
      "step": 8889
    },
    {
      "epoch": 0.6728158554467623,
      "grad_norm": 0.71875,
      "learning_rate": 1.4950406287925865e-05,
      "loss": 0.3039,
      "step": 8890
    },
    {
      "epoch": 0.6728915377702095,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4949371359325036e-05,
      "loss": 0.3077,
      "step": 8891
    },
    {
      "epoch": 0.6729672200936568,
      "grad_norm": 0.80078125,
      "learning_rate": 1.4948336360508883e-05,
      "loss": 0.3408,
      "step": 8892
    },
    {
      "epoch": 0.6730429024171042,
      "grad_norm": 0.67578125,
      "learning_rate": 1.4947301291492092e-05,
      "loss": 0.26,
      "step": 8893
    },
    {
      "epoch": 0.6731185847405515,
      "grad_norm": 0.796875,
      "learning_rate": 1.4946266152289334e-05,
      "loss": 0.3714,
      "step": 8894
    },
    {
      "epoch": 0.6731942670639989,
      "grad_norm": 0.79296875,
      "learning_rate": 1.4945230942915307e-05,
      "loss": 0.3739,
      "step": 8895
    },
    {
      "epoch": 0.6732699493874462,
      "grad_norm": 1.0234375,
      "learning_rate": 1.4944195663384693e-05,
      "loss": 0.3936,
      "step": 8896
    },
    {
      "epoch": 0.6733456317108936,
      "grad_norm": 0.78515625,
      "learning_rate": 1.4943160313712178e-05,
      "loss": 0.3417,
      "step": 8897
    },
    {
      "epoch": 0.6734213140343408,
      "grad_norm": 0.703125,
      "learning_rate": 1.4942124893912453e-05,
      "loss": 0.2766,
      "step": 8898
    },
    {
      "epoch": 0.6734969963577881,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4941089404000204e-05,
      "loss": 0.3214,
      "step": 8899
    },
    {
      "epoch": 0.6735726786812355,
      "grad_norm": 0.80078125,
      "learning_rate": 1.4940053843990121e-05,
      "loss": 0.3609,
      "step": 8900
    },
    {
      "epoch": 0.6736483610046828,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4939018213896898e-05,
      "loss": 0.325,
      "step": 8901
    },
    {
      "epoch": 0.6737240433281302,
      "grad_norm": 0.7265625,
      "learning_rate": 1.493798251373523e-05,
      "loss": 0.2752,
      "step": 8902
    },
    {
      "epoch": 0.6737997256515775,
      "grad_norm": 0.68359375,
      "learning_rate": 1.49369467435198e-05,
      "loss": 0.2754,
      "step": 8903
    },
    {
      "epoch": 0.6738754079750249,
      "grad_norm": 0.79296875,
      "learning_rate": 1.4935910903265311e-05,
      "loss": 0.3359,
      "step": 8904
    },
    {
      "epoch": 0.6739510902984721,
      "grad_norm": 0.79296875,
      "learning_rate": 1.4934874992986457e-05,
      "loss": 0.3608,
      "step": 8905
    },
    {
      "epoch": 0.6740267726219195,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4933839012697928e-05,
      "loss": 0.3291,
      "step": 8906
    },
    {
      "epoch": 0.6741024549453668,
      "grad_norm": 0.76171875,
      "learning_rate": 1.493280296241443e-05,
      "loss": 0.3326,
      "step": 8907
    },
    {
      "epoch": 0.6741781372688141,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4931766842150656e-05,
      "loss": 0.3228,
      "step": 8908
    },
    {
      "epoch": 0.6742538195922615,
      "grad_norm": 0.71484375,
      "learning_rate": 1.4930730651921306e-05,
      "loss": 0.2677,
      "step": 8909
    },
    {
      "epoch": 0.6743295019157088,
      "grad_norm": 0.77734375,
      "learning_rate": 1.492969439174108e-05,
      "loss": 0.3592,
      "step": 8910
    },
    {
      "epoch": 0.6744051842391562,
      "grad_norm": 0.734375,
      "learning_rate": 1.4928658061624681e-05,
      "loss": 0.2867,
      "step": 8911
    },
    {
      "epoch": 0.6744808665626034,
      "grad_norm": 0.78125,
      "learning_rate": 1.4927621661586806e-05,
      "loss": 0.3494,
      "step": 8912
    },
    {
      "epoch": 0.6745565488860508,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4926585191642166e-05,
      "loss": 0.3309,
      "step": 8913
    },
    {
      "epoch": 0.6746322312094981,
      "grad_norm": 0.71484375,
      "learning_rate": 1.4925548651805459e-05,
      "loss": 0.2806,
      "step": 8914
    },
    {
      "epoch": 0.6747079135329455,
      "grad_norm": 0.73046875,
      "learning_rate": 1.492451204209139e-05,
      "loss": 0.3131,
      "step": 8915
    },
    {
      "epoch": 0.6747835958563928,
      "grad_norm": 0.703125,
      "learning_rate": 1.492347536251467e-05,
      "loss": 0.2662,
      "step": 8916
    },
    {
      "epoch": 0.6748592781798401,
      "grad_norm": 0.73046875,
      "learning_rate": 1.492243861309e-05,
      "loss": 0.313,
      "step": 8917
    },
    {
      "epoch": 0.6749349605032875,
      "grad_norm": 0.71875,
      "learning_rate": 1.4921401793832094e-05,
      "loss": 0.2841,
      "step": 8918
    },
    {
      "epoch": 0.6750106428267347,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4920364904755656e-05,
      "loss": 0.3048,
      "step": 8919
    },
    {
      "epoch": 0.6750863251501821,
      "grad_norm": 0.8046875,
      "learning_rate": 1.4919327945875399e-05,
      "loss": 0.3217,
      "step": 8920
    },
    {
      "epoch": 0.6751620074736294,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4918290917206032e-05,
      "loss": 0.3144,
      "step": 8921
    },
    {
      "epoch": 0.6752376897970768,
      "grad_norm": 0.6953125,
      "learning_rate": 1.4917253818762269e-05,
      "loss": 0.2819,
      "step": 8922
    },
    {
      "epoch": 0.6753133721205241,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4916216650558824e-05,
      "loss": 0.3116,
      "step": 8923
    },
    {
      "epoch": 0.6753890544439715,
      "grad_norm": 1.09375,
      "learning_rate": 1.4915179412610408e-05,
      "loss": 0.3921,
      "step": 8924
    },
    {
      "epoch": 0.6754647367674188,
      "grad_norm": 0.75,
      "learning_rate": 1.4914142104931737e-05,
      "loss": 0.2928,
      "step": 8925
    },
    {
      "epoch": 0.675540419090866,
      "grad_norm": 0.765625,
      "learning_rate": 1.4913104727537526e-05,
      "loss": 0.3256,
      "step": 8926
    },
    {
      "epoch": 0.6756161014143134,
      "grad_norm": 0.75,
      "learning_rate": 1.4912067280442494e-05,
      "loss": 0.3194,
      "step": 8927
    },
    {
      "epoch": 0.6756917837377607,
      "grad_norm": 0.69921875,
      "learning_rate": 1.4911029763661362e-05,
      "loss": 0.2761,
      "step": 8928
    },
    {
      "epoch": 0.6757674660612081,
      "grad_norm": 0.859375,
      "learning_rate": 1.4909992177208841e-05,
      "loss": 0.3541,
      "step": 8929
    },
    {
      "epoch": 0.6758431483846554,
      "grad_norm": 0.70703125,
      "learning_rate": 1.4908954521099656e-05,
      "loss": 0.2839,
      "step": 8930
    },
    {
      "epoch": 0.6759188307081028,
      "grad_norm": 0.890625,
      "learning_rate": 1.4907916795348529e-05,
      "loss": 0.3158,
      "step": 8931
    },
    {
      "epoch": 0.6759945130315501,
      "grad_norm": 0.765625,
      "learning_rate": 1.4906878999970176e-05,
      "loss": 0.2994,
      "step": 8932
    },
    {
      "epoch": 0.6760701953549973,
      "grad_norm": 0.80859375,
      "learning_rate": 1.4905841134979326e-05,
      "loss": 0.3804,
      "step": 8933
    },
    {
      "epoch": 0.6761458776784447,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4904803200390702e-05,
      "loss": 0.3251,
      "step": 8934
    },
    {
      "epoch": 0.676221560001892,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4903765196219027e-05,
      "loss": 0.2955,
      "step": 8935
    },
    {
      "epoch": 0.6762972423253394,
      "grad_norm": 0.72265625,
      "learning_rate": 1.4902727122479029e-05,
      "loss": 0.3209,
      "step": 8936
    },
    {
      "epoch": 0.6763729246487867,
      "grad_norm": 0.74609375,
      "learning_rate": 1.490168897918543e-05,
      "loss": 0.3354,
      "step": 8937
    },
    {
      "epoch": 0.6764486069722341,
      "grad_norm": 0.8046875,
      "learning_rate": 1.4900650766352962e-05,
      "loss": 0.3298,
      "step": 8938
    },
    {
      "epoch": 0.6765242892956814,
      "grad_norm": 0.80078125,
      "learning_rate": 1.4899612483996352e-05,
      "loss": 0.337,
      "step": 8939
    },
    {
      "epoch": 0.6765999716191287,
      "grad_norm": 0.70703125,
      "learning_rate": 1.4898574132130337e-05,
      "loss": 0.2917,
      "step": 8940
    },
    {
      "epoch": 0.676675653942576,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4897535710769636e-05,
      "loss": 0.3348,
      "step": 8941
    },
    {
      "epoch": 0.6767513362660234,
      "grad_norm": 0.73828125,
      "learning_rate": 1.489649721992899e-05,
      "loss": 0.3058,
      "step": 8942
    },
    {
      "epoch": 0.6768270185894707,
      "grad_norm": 0.796875,
      "learning_rate": 1.4895458659623125e-05,
      "loss": 0.3483,
      "step": 8943
    },
    {
      "epoch": 0.676902700912918,
      "grad_norm": 0.7109375,
      "learning_rate": 1.489442002986678e-05,
      "loss": 0.2992,
      "step": 8944
    },
    {
      "epoch": 0.6769783832363654,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4893381330674688e-05,
      "loss": 0.3145,
      "step": 8945
    },
    {
      "epoch": 0.6770540655598127,
      "grad_norm": 0.78125,
      "learning_rate": 1.4892342562061583e-05,
      "loss": 0.3278,
      "step": 8946
    },
    {
      "epoch": 0.67712974788326,
      "grad_norm": 0.74609375,
      "learning_rate": 1.48913037240422e-05,
      "loss": 0.2792,
      "step": 8947
    },
    {
      "epoch": 0.6772054302067073,
      "grad_norm": 0.75390625,
      "learning_rate": 1.4890264816631285e-05,
      "loss": 0.3125,
      "step": 8948
    },
    {
      "epoch": 0.6772811125301547,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4889225839843574e-05,
      "loss": 0.294,
      "step": 8949
    },
    {
      "epoch": 0.677356794853602,
      "grad_norm": 0.75,
      "learning_rate": 1.4888186793693798e-05,
      "loss": 0.2999,
      "step": 8950
    },
    {
      "epoch": 0.6774324771770494,
      "grad_norm": 0.79296875,
      "learning_rate": 1.4887147678196708e-05,
      "loss": 0.337,
      "step": 8951
    },
    {
      "epoch": 0.6775081595004967,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4886108493367037e-05,
      "loss": 0.3168,
      "step": 8952
    },
    {
      "epoch": 0.677583841823944,
      "grad_norm": 0.796875,
      "learning_rate": 1.4885069239219534e-05,
      "loss": 0.3317,
      "step": 8953
    },
    {
      "epoch": 0.6776595241473913,
      "grad_norm": 0.796875,
      "learning_rate": 1.4884029915768945e-05,
      "loss": 0.3382,
      "step": 8954
    },
    {
      "epoch": 0.6777352064708386,
      "grad_norm": 0.78515625,
      "learning_rate": 1.4882990523030008e-05,
      "loss": 0.3327,
      "step": 8955
    },
    {
      "epoch": 0.677810888794286,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4881951061017469e-05,
      "loss": 0.3371,
      "step": 8956
    },
    {
      "epoch": 0.6778865711177333,
      "grad_norm": 0.734375,
      "learning_rate": 1.4880911529746076e-05,
      "loss": 0.2895,
      "step": 8957
    },
    {
      "epoch": 0.6779622534411807,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4879871929230583e-05,
      "loss": 0.3272,
      "step": 8958
    },
    {
      "epoch": 0.678037935764628,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4878832259485724e-05,
      "loss": 0.2817,
      "step": 8959
    },
    {
      "epoch": 0.6781136180880754,
      "grad_norm": 0.75390625,
      "learning_rate": 1.4877792520526264e-05,
      "loss": 0.3191,
      "step": 8960
    },
    {
      "epoch": 0.6781893004115226,
      "grad_norm": 0.6953125,
      "learning_rate": 1.4876752712366945e-05,
      "loss": 0.2634,
      "step": 8961
    },
    {
      "epoch": 0.6782649827349699,
      "grad_norm": 0.68359375,
      "learning_rate": 1.4875712835022515e-05,
      "loss": 0.2573,
      "step": 8962
    },
    {
      "epoch": 0.6783406650584173,
      "grad_norm": 0.8203125,
      "learning_rate": 1.4874672888507738e-05,
      "loss": 0.3586,
      "step": 8963
    },
    {
      "epoch": 0.6784163473818646,
      "grad_norm": 0.8125,
      "learning_rate": 1.4873632872837357e-05,
      "loss": 0.3558,
      "step": 8964
    },
    {
      "epoch": 0.678492029705312,
      "grad_norm": 0.7734375,
      "learning_rate": 1.487259278802613e-05,
      "loss": 0.3529,
      "step": 8965
    },
    {
      "epoch": 0.6785677120287593,
      "grad_norm": 0.75,
      "learning_rate": 1.4871552634088816e-05,
      "loss": 0.315,
      "step": 8966
    },
    {
      "epoch": 0.6786433943522067,
      "grad_norm": 0.9765625,
      "learning_rate": 1.4870512411040165e-05,
      "loss": 0.3777,
      "step": 8967
    },
    {
      "epoch": 0.6787190766756539,
      "grad_norm": 0.8203125,
      "learning_rate": 1.4869472118894939e-05,
      "loss": 0.332,
      "step": 8968
    },
    {
      "epoch": 0.6787947589991012,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4868431757667895e-05,
      "loss": 0.3542,
      "step": 8969
    },
    {
      "epoch": 0.6788704413225486,
      "grad_norm": 0.7734375,
      "learning_rate": 1.486739132737379e-05,
      "loss": 0.3041,
      "step": 8970
    },
    {
      "epoch": 0.6789461236459959,
      "grad_norm": 0.75,
      "learning_rate": 1.4866350828027387e-05,
      "loss": 0.3154,
      "step": 8971
    },
    {
      "epoch": 0.6790218059694433,
      "grad_norm": 0.765625,
      "learning_rate": 1.486531025964345e-05,
      "loss": 0.3408,
      "step": 8972
    },
    {
      "epoch": 0.6790974882928906,
      "grad_norm": 1.0390625,
      "learning_rate": 1.4864269622236732e-05,
      "loss": 0.3177,
      "step": 8973
    },
    {
      "epoch": 0.679173170616338,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4863228915822006e-05,
      "loss": 0.3116,
      "step": 8974
    },
    {
      "epoch": 0.6792488529397852,
      "grad_norm": 0.8046875,
      "learning_rate": 1.486218814041403e-05,
      "loss": 0.3615,
      "step": 8975
    },
    {
      "epoch": 0.6793245352632326,
      "grad_norm": 0.91015625,
      "learning_rate": 1.4861147296027572e-05,
      "loss": 0.3345,
      "step": 8976
    },
    {
      "epoch": 0.6794002175866799,
      "grad_norm": 0.79296875,
      "learning_rate": 1.48601063826774e-05,
      "loss": 0.3482,
      "step": 8977
    },
    {
      "epoch": 0.6794758999101272,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4859065400378277e-05,
      "loss": 0.3161,
      "step": 8978
    },
    {
      "epoch": 0.6795515822335746,
      "grad_norm": 0.78515625,
      "learning_rate": 1.485802434914497e-05,
      "loss": 0.3329,
      "step": 8979
    },
    {
      "epoch": 0.6796272645570219,
      "grad_norm": 0.80859375,
      "learning_rate": 1.4856983228992254e-05,
      "loss": 0.3462,
      "step": 8980
    },
    {
      "epoch": 0.6797029468804693,
      "grad_norm": 0.765625,
      "learning_rate": 1.4855942039934896e-05,
      "loss": 0.3429,
      "step": 8981
    },
    {
      "epoch": 0.6797786292039165,
      "grad_norm": 0.8515625,
      "learning_rate": 1.4854900781987667e-05,
      "loss": 0.3004,
      "step": 8982
    },
    {
      "epoch": 0.6798543115273639,
      "grad_norm": 0.671875,
      "learning_rate": 1.4853859455165338e-05,
      "loss": 0.264,
      "step": 8983
    },
    {
      "epoch": 0.6799299938508112,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4852818059482682e-05,
      "loss": 0.2759,
      "step": 8984
    },
    {
      "epoch": 0.6800056761742586,
      "grad_norm": 0.78515625,
      "learning_rate": 1.4851776594954478e-05,
      "loss": 0.3234,
      "step": 8985
    },
    {
      "epoch": 0.6800813584977059,
      "grad_norm": 0.83203125,
      "learning_rate": 1.4850735061595496e-05,
      "loss": 0.3216,
      "step": 8986
    },
    {
      "epoch": 0.6801570408211532,
      "grad_norm": 0.73046875,
      "learning_rate": 1.4849693459420512e-05,
      "loss": 0.3126,
      "step": 8987
    },
    {
      "epoch": 0.6802327231446006,
      "grad_norm": 0.78125,
      "learning_rate": 1.4848651788444302e-05,
      "loss": 0.3483,
      "step": 8988
    },
    {
      "epoch": 0.6803084054680478,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4847610048681649e-05,
      "loss": 0.3409,
      "step": 8989
    },
    {
      "epoch": 0.6803840877914952,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4846568240147327e-05,
      "loss": 0.287,
      "step": 8990
    },
    {
      "epoch": 0.6804597701149425,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4845526362856117e-05,
      "loss": 0.3285,
      "step": 8991
    },
    {
      "epoch": 0.6805354524383899,
      "grad_norm": 0.67578125,
      "learning_rate": 1.4844484416822802e-05,
      "loss": 0.269,
      "step": 8992
    },
    {
      "epoch": 0.6806111347618372,
      "grad_norm": 0.82421875,
      "learning_rate": 1.4843442402062163e-05,
      "loss": 0.3367,
      "step": 8993
    },
    {
      "epoch": 0.6806868170852846,
      "grad_norm": 0.75390625,
      "learning_rate": 1.4842400318588979e-05,
      "loss": 0.3179,
      "step": 8994
    },
    {
      "epoch": 0.6807624994087319,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4841358166418039e-05,
      "loss": 0.322,
      "step": 8995
    },
    {
      "epoch": 0.6808381817321791,
      "grad_norm": 0.83203125,
      "learning_rate": 1.4840315945564126e-05,
      "loss": 0.3721,
      "step": 8996
    },
    {
      "epoch": 0.6809138640556265,
      "grad_norm": 0.80859375,
      "learning_rate": 1.4839273656042022e-05,
      "loss": 0.336,
      "step": 8997
    },
    {
      "epoch": 0.6809895463790738,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4838231297866516e-05,
      "loss": 0.3041,
      "step": 8998
    },
    {
      "epoch": 0.6810652287025212,
      "grad_norm": 0.765625,
      "learning_rate": 1.4837188871052399e-05,
      "loss": 0.3026,
      "step": 8999
    },
    {
      "epoch": 0.6811409110259685,
      "grad_norm": 0.75,
      "learning_rate": 1.4836146375614456e-05,
      "loss": 0.2963,
      "step": 9000
    },
    {
      "epoch": 0.6812165933494159,
      "grad_norm": 0.73046875,
      "learning_rate": 1.4835103811567479e-05,
      "loss": 0.3086,
      "step": 9001
    },
    {
      "epoch": 0.6812922756728632,
      "grad_norm": 0.78515625,
      "learning_rate": 1.4834061178926256e-05,
      "loss": 0.3263,
      "step": 9002
    },
    {
      "epoch": 0.6813679579963104,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4833018477705579e-05,
      "loss": 0.3181,
      "step": 9003
    },
    {
      "epoch": 0.6814436403197578,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4831975707920242e-05,
      "loss": 0.3097,
      "step": 9004
    },
    {
      "epoch": 0.6815193226432051,
      "grad_norm": 0.77734375,
      "learning_rate": 1.483093286958504e-05,
      "loss": 0.3516,
      "step": 9005
    },
    {
      "epoch": 0.6815950049666525,
      "grad_norm": 0.65625,
      "learning_rate": 1.4829889962714761e-05,
      "loss": 0.2538,
      "step": 9006
    },
    {
      "epoch": 0.6816706872900998,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4828846987324206e-05,
      "loss": 0.3185,
      "step": 9007
    },
    {
      "epoch": 0.6817463696135472,
      "grad_norm": 0.734375,
      "learning_rate": 1.482780394342817e-05,
      "loss": 0.312,
      "step": 9008
    },
    {
      "epoch": 0.6818220519369944,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4826760831041447e-05,
      "loss": 0.3223,
      "step": 9009
    },
    {
      "epoch": 0.6818977342604418,
      "grad_norm": 0.89453125,
      "learning_rate": 1.4825717650178846e-05,
      "loss": 0.334,
      "step": 9010
    },
    {
      "epoch": 0.6819734165838891,
      "grad_norm": 0.765625,
      "learning_rate": 1.4824674400855155e-05,
      "loss": 0.3009,
      "step": 9011
    },
    {
      "epoch": 0.6820490989073364,
      "grad_norm": 0.734375,
      "learning_rate": 1.4823631083085175e-05,
      "loss": 0.2974,
      "step": 9012
    },
    {
      "epoch": 0.6821247812307838,
      "grad_norm": 1.1328125,
      "learning_rate": 1.4822587696883715e-05,
      "loss": 0.3784,
      "step": 9013
    },
    {
      "epoch": 0.6822004635542311,
      "grad_norm": 0.79296875,
      "learning_rate": 1.482154424226557e-05,
      "loss": 0.3392,
      "step": 9014
    },
    {
      "epoch": 0.6822761458776785,
      "grad_norm": 0.78125,
      "learning_rate": 1.4820500719245547e-05,
      "loss": 0.3021,
      "step": 9015
    },
    {
      "epoch": 0.6823518282011257,
      "grad_norm": 0.734375,
      "learning_rate": 1.4819457127838451e-05,
      "loss": 0.2722,
      "step": 9016
    },
    {
      "epoch": 0.6824275105245731,
      "grad_norm": 0.79296875,
      "learning_rate": 1.4818413468059083e-05,
      "loss": 0.3405,
      "step": 9017
    },
    {
      "epoch": 0.6825031928480204,
      "grad_norm": 0.79296875,
      "learning_rate": 1.481736973992225e-05,
      "loss": 0.2971,
      "step": 9018
    },
    {
      "epoch": 0.6825788751714678,
      "grad_norm": 0.83203125,
      "learning_rate": 1.4816325943442762e-05,
      "loss": 0.3845,
      "step": 9019
    },
    {
      "epoch": 0.6826545574949151,
      "grad_norm": 0.6953125,
      "learning_rate": 1.4815282078635426e-05,
      "loss": 0.2453,
      "step": 9020
    },
    {
      "epoch": 0.6827302398183624,
      "grad_norm": 0.83203125,
      "learning_rate": 1.4814238145515048e-05,
      "loss": 0.3379,
      "step": 9021
    },
    {
      "epoch": 0.6828059221418098,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4813194144096441e-05,
      "loss": 0.33,
      "step": 9022
    },
    {
      "epoch": 0.682881604465257,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4812150074394418e-05,
      "loss": 0.2789,
      "step": 9023
    },
    {
      "epoch": 0.6829572867887044,
      "grad_norm": 0.734375,
      "learning_rate": 1.4811105936423787e-05,
      "loss": 0.3068,
      "step": 9024
    },
    {
      "epoch": 0.6830329691121517,
      "grad_norm": 0.75,
      "learning_rate": 1.481006173019936e-05,
      "loss": 0.3022,
      "step": 9025
    },
    {
      "epoch": 0.6831086514355991,
      "grad_norm": 0.75390625,
      "learning_rate": 1.4809017455735957e-05,
      "loss": 0.3297,
      "step": 9026
    },
    {
      "epoch": 0.6831843337590464,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4807973113048384e-05,
      "loss": 0.3069,
      "step": 9027
    },
    {
      "epoch": 0.6832600160824938,
      "grad_norm": 0.703125,
      "learning_rate": 1.4806928702151464e-05,
      "loss": 0.2958,
      "step": 9028
    },
    {
      "epoch": 0.6833356984059411,
      "grad_norm": 0.734375,
      "learning_rate": 1.4805884223060012e-05,
      "loss": 0.2968,
      "step": 9029
    },
    {
      "epoch": 0.6834113807293883,
      "grad_norm": 0.765625,
      "learning_rate": 1.4804839675788844e-05,
      "loss": 0.3227,
      "step": 9030
    },
    {
      "epoch": 0.6834870630528357,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4803795060352782e-05,
      "loss": 0.3378,
      "step": 9031
    },
    {
      "epoch": 0.683562745376283,
      "grad_norm": 0.828125,
      "learning_rate": 1.4802750376766641e-05,
      "loss": 0.3493,
      "step": 9032
    },
    {
      "epoch": 0.6836384276997304,
      "grad_norm": 0.69921875,
      "learning_rate": 1.4801705625045246e-05,
      "loss": 0.2735,
      "step": 9033
    },
    {
      "epoch": 0.6837141100231777,
      "grad_norm": 0.75390625,
      "learning_rate": 1.4800660805203417e-05,
      "loss": 0.3399,
      "step": 9034
    },
    {
      "epoch": 0.6837897923466251,
      "grad_norm": 0.8125,
      "learning_rate": 1.4799615917255974e-05,
      "loss": 0.3415,
      "step": 9035
    },
    {
      "epoch": 0.6838654746700724,
      "grad_norm": 0.734375,
      "learning_rate": 1.4798570961217743e-05,
      "loss": 0.3056,
      "step": 9036
    },
    {
      "epoch": 0.6839411569935196,
      "grad_norm": 0.76953125,
      "learning_rate": 1.479752593710355e-05,
      "loss": 0.3395,
      "step": 9037
    },
    {
      "epoch": 0.684016839316967,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4796480844928218e-05,
      "loss": 0.33,
      "step": 9038
    },
    {
      "epoch": 0.6840925216404143,
      "grad_norm": 0.75,
      "learning_rate": 1.4795435684706574e-05,
      "loss": 0.3226,
      "step": 9039
    },
    {
      "epoch": 0.6841682039638617,
      "grad_norm": 0.75,
      "learning_rate": 1.4794390456453444e-05,
      "loss": 0.3099,
      "step": 9040
    },
    {
      "epoch": 0.684243886287309,
      "grad_norm": 0.765625,
      "learning_rate": 1.479334516018366e-05,
      "loss": 0.3325,
      "step": 9041
    },
    {
      "epoch": 0.6843195686107564,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4792299795912049e-05,
      "loss": 0.3133,
      "step": 9042
    },
    {
      "epoch": 0.6843952509342037,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4791254363653443e-05,
      "loss": 0.3109,
      "step": 9043
    },
    {
      "epoch": 0.684470933257651,
      "grad_norm": 0.8359375,
      "learning_rate": 1.4790208863422669e-05,
      "loss": 0.3444,
      "step": 9044
    },
    {
      "epoch": 0.6845466155810983,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4789163295234561e-05,
      "loss": 0.3217,
      "step": 9045
    },
    {
      "epoch": 0.6846222979045457,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4788117659103956e-05,
      "loss": 0.3092,
      "step": 9046
    },
    {
      "epoch": 0.684697980227993,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4787071955045686e-05,
      "loss": 0.3059,
      "step": 9047
    },
    {
      "epoch": 0.6847736625514403,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4786026183074582e-05,
      "loss": 0.3164,
      "step": 9048
    },
    {
      "epoch": 0.6848493448748877,
      "grad_norm": 0.796875,
      "learning_rate": 1.4784980343205489e-05,
      "loss": 0.3478,
      "step": 9049
    },
    {
      "epoch": 0.684925027198335,
      "grad_norm": 0.72265625,
      "learning_rate": 1.4783934435453232e-05,
      "loss": 0.2925,
      "step": 9050
    },
    {
      "epoch": 0.6850007095217823,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4782888459832658e-05,
      "loss": 0.341,
      "step": 9051
    },
    {
      "epoch": 0.6850763918452296,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4781842416358604e-05,
      "loss": 0.3144,
      "step": 9052
    },
    {
      "epoch": 0.685152074168677,
      "grad_norm": 0.73828125,
      "learning_rate": 1.478079630504591e-05,
      "loss": 0.3293,
      "step": 9053
    },
    {
      "epoch": 0.6852277564921243,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4779750125909415e-05,
      "loss": 0.3422,
      "step": 9054
    },
    {
      "epoch": 0.6853034388155717,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4778703878963963e-05,
      "loss": 0.2945,
      "step": 9055
    },
    {
      "epoch": 0.685379121139019,
      "grad_norm": 0.765625,
      "learning_rate": 1.4777657564224394e-05,
      "loss": 0.3136,
      "step": 9056
    },
    {
      "epoch": 0.6854548034624663,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4776611181705557e-05,
      "loss": 0.3161,
      "step": 9057
    },
    {
      "epoch": 0.6855304857859136,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4775564731422294e-05,
      "loss": 0.3207,
      "step": 9058
    },
    {
      "epoch": 0.6856061681093609,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4774518213389446e-05,
      "loss": 0.3263,
      "step": 9059
    },
    {
      "epoch": 0.6856818504328083,
      "grad_norm": 0.796875,
      "learning_rate": 1.4773471627621865e-05,
      "loss": 0.3212,
      "step": 9060
    },
    {
      "epoch": 0.6857575327562556,
      "grad_norm": 0.76953125,
      "learning_rate": 1.47724249741344e-05,
      "loss": 0.3279,
      "step": 9061
    },
    {
      "epoch": 0.685833215079703,
      "grad_norm": 0.828125,
      "learning_rate": 1.4771378252941896e-05,
      "loss": 0.3422,
      "step": 9062
    },
    {
      "epoch": 0.6859088974031503,
      "grad_norm": 0.74609375,
      "learning_rate": 1.47703314640592e-05,
      "loss": 0.2994,
      "step": 9063
    },
    {
      "epoch": 0.6859845797265977,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4769284607501172e-05,
      "loss": 0.3092,
      "step": 9064
    },
    {
      "epoch": 0.6860602620500449,
      "grad_norm": 0.69140625,
      "learning_rate": 1.476823768328265e-05,
      "loss": 0.2565,
      "step": 9065
    },
    {
      "epoch": 0.6861359443734922,
      "grad_norm": 0.796875,
      "learning_rate": 1.47671906914185e-05,
      "loss": 0.3055,
      "step": 9066
    },
    {
      "epoch": 0.6862116266969396,
      "grad_norm": 0.80078125,
      "learning_rate": 1.4766143631923569e-05,
      "loss": 0.3306,
      "step": 9067
    },
    {
      "epoch": 0.6862873090203869,
      "grad_norm": 0.765625,
      "learning_rate": 1.4765096504812708e-05,
      "loss": 0.3196,
      "step": 9068
    },
    {
      "epoch": 0.6863629913438343,
      "grad_norm": 0.765625,
      "learning_rate": 1.4764049310100777e-05,
      "loss": 0.3112,
      "step": 9069
    },
    {
      "epoch": 0.6864386736672816,
      "grad_norm": 0.80078125,
      "learning_rate": 1.4763002047802633e-05,
      "loss": 0.3244,
      "step": 9070
    },
    {
      "epoch": 0.686514355990729,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4761954717933132e-05,
      "loss": 0.3414,
      "step": 9071
    },
    {
      "epoch": 0.6865900383141762,
      "grad_norm": 0.73046875,
      "learning_rate": 1.4760907320507127e-05,
      "loss": 0.3093,
      "step": 9072
    },
    {
      "epoch": 0.6866657206376235,
      "grad_norm": 0.82421875,
      "learning_rate": 1.4759859855539484e-05,
      "loss": 0.3357,
      "step": 9073
    },
    {
      "epoch": 0.6867414029610709,
      "grad_norm": 0.79296875,
      "learning_rate": 1.4758812323045062e-05,
      "loss": 0.3283,
      "step": 9074
    },
    {
      "epoch": 0.6868170852845182,
      "grad_norm": 0.75,
      "learning_rate": 1.4757764723038721e-05,
      "loss": 0.3239,
      "step": 9075
    },
    {
      "epoch": 0.6868927676079656,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4756717055535324e-05,
      "loss": 0.3039,
      "step": 9076
    },
    {
      "epoch": 0.6869684499314129,
      "grad_norm": 0.81640625,
      "learning_rate": 1.4755669320549727e-05,
      "loss": 0.3031,
      "step": 9077
    },
    {
      "epoch": 0.6870441322548603,
      "grad_norm": 0.72265625,
      "learning_rate": 1.4754621518096803e-05,
      "loss": 0.2887,
      "step": 9078
    },
    {
      "epoch": 0.6871198145783075,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4753573648191416e-05,
      "loss": 0.3406,
      "step": 9079
    },
    {
      "epoch": 0.6871954969017549,
      "grad_norm": 0.75390625,
      "learning_rate": 1.4752525710848426e-05,
      "loss": 0.3194,
      "step": 9080
    },
    {
      "epoch": 0.6872711792252022,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4751477706082704e-05,
      "loss": 0.3433,
      "step": 9081
    },
    {
      "epoch": 0.6873468615486495,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4750429633909118e-05,
      "loss": 0.2885,
      "step": 9082
    },
    {
      "epoch": 0.6874225438720969,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4749381494342534e-05,
      "loss": 0.3134,
      "step": 9083
    },
    {
      "epoch": 0.6874982261955442,
      "grad_norm": 0.78515625,
      "learning_rate": 1.4748333287397824e-05,
      "loss": 0.3176,
      "step": 9084
    },
    {
      "epoch": 0.6875739085189916,
      "grad_norm": 1.0390625,
      "learning_rate": 1.4747285013089858e-05,
      "loss": 0.3539,
      "step": 9085
    },
    {
      "epoch": 0.6876495908424388,
      "grad_norm": 0.73828125,
      "learning_rate": 1.474623667143351e-05,
      "loss": 0.2871,
      "step": 9086
    },
    {
      "epoch": 0.6877252731658862,
      "grad_norm": 0.8046875,
      "learning_rate": 1.4745188262443646e-05,
      "loss": 0.3397,
      "step": 9087
    },
    {
      "epoch": 0.6878009554893335,
      "grad_norm": 0.80078125,
      "learning_rate": 1.4744139786135146e-05,
      "loss": 0.3661,
      "step": 9088
    },
    {
      "epoch": 0.6878766378127809,
      "grad_norm": 0.72265625,
      "learning_rate": 1.474309124252288e-05,
      "loss": 0.271,
      "step": 9089
    },
    {
      "epoch": 0.6879523201362282,
      "grad_norm": 0.6875,
      "learning_rate": 1.4742042631621727e-05,
      "loss": 0.2512,
      "step": 9090
    },
    {
      "epoch": 0.6880280024596755,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4740993953446561e-05,
      "loss": 0.294,
      "step": 9091
    },
    {
      "epoch": 0.6881036847831229,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4739945208012258e-05,
      "loss": 0.3591,
      "step": 9092
    },
    {
      "epoch": 0.6881793671065701,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4738896395333702e-05,
      "loss": 0.2783,
      "step": 9093
    },
    {
      "epoch": 0.6882550494300175,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4737847515425766e-05,
      "loss": 0.3135,
      "step": 9094
    },
    {
      "epoch": 0.6883307317534648,
      "grad_norm": 1.0390625,
      "learning_rate": 1.4736798568303333e-05,
      "loss": 0.3536,
      "step": 9095
    },
    {
      "epoch": 0.6884064140769122,
      "grad_norm": 0.796875,
      "learning_rate": 1.4735749553981285e-05,
      "loss": 0.3537,
      "step": 9096
    },
    {
      "epoch": 0.6884820964003595,
      "grad_norm": 0.85546875,
      "learning_rate": 1.4734700472474502e-05,
      "loss": 0.3911,
      "step": 9097
    },
    {
      "epoch": 0.6885577787238069,
      "grad_norm": 0.81640625,
      "learning_rate": 1.4733651323797863e-05,
      "loss": 0.3574,
      "step": 9098
    },
    {
      "epoch": 0.6886334610472542,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4732602107966262e-05,
      "loss": 0.3031,
      "step": 9099
    },
    {
      "epoch": 0.6887091433707014,
      "grad_norm": 0.7265625,
      "learning_rate": 1.473155282499458e-05,
      "loss": 0.3134,
      "step": 9100
    },
    {
      "epoch": 0.6887848256941488,
      "grad_norm": 0.796875,
      "learning_rate": 1.4730503474897698e-05,
      "loss": 0.3269,
      "step": 9101
    },
    {
      "epoch": 0.6888605080175961,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4729454057690507e-05,
      "loss": 0.3102,
      "step": 9102
    },
    {
      "epoch": 0.6889361903410435,
      "grad_norm": 0.765625,
      "learning_rate": 1.4728404573387895e-05,
      "loss": 0.311,
      "step": 9103
    },
    {
      "epoch": 0.6890118726644908,
      "grad_norm": 0.78125,
      "learning_rate": 1.472735502200475e-05,
      "loss": 0.3394,
      "step": 9104
    },
    {
      "epoch": 0.6890875549879382,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4726305403555961e-05,
      "loss": 0.2973,
      "step": 9105
    },
    {
      "epoch": 0.6891632373113855,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4725255718056419e-05,
      "loss": 0.3109,
      "step": 9106
    },
    {
      "epoch": 0.6892389196348327,
      "grad_norm": 0.73046875,
      "learning_rate": 1.4724205965521018e-05,
      "loss": 0.2715,
      "step": 9107
    },
    {
      "epoch": 0.6893146019582801,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4723156145964647e-05,
      "loss": 0.3342,
      "step": 9108
    },
    {
      "epoch": 0.6893902842817274,
      "grad_norm": 0.84765625,
      "learning_rate": 1.4722106259402197e-05,
      "loss": 0.3811,
      "step": 9109
    },
    {
      "epoch": 0.6894659666051748,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4721056305848571e-05,
      "loss": 0.3272,
      "step": 9110
    },
    {
      "epoch": 0.6895416489286221,
      "grad_norm": 0.73046875,
      "learning_rate": 1.472000628531866e-05,
      "loss": 0.2898,
      "step": 9111
    },
    {
      "epoch": 0.6896173312520695,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4718956197827356e-05,
      "loss": 0.3371,
      "step": 9112
    },
    {
      "epoch": 0.6896930135755168,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4717906043389564e-05,
      "loss": 0.3036,
      "step": 9113
    },
    {
      "epoch": 0.6897686958989641,
      "grad_norm": 0.80078125,
      "learning_rate": 1.471685582202018e-05,
      "loss": 0.3379,
      "step": 9114
    },
    {
      "epoch": 0.6898443782224114,
      "grad_norm": 0.8046875,
      "learning_rate": 1.4715805533734097e-05,
      "loss": 0.3292,
      "step": 9115
    },
    {
      "epoch": 0.6899200605458587,
      "grad_norm": 0.734375,
      "learning_rate": 1.4714755178546223e-05,
      "loss": 0.3007,
      "step": 9116
    },
    {
      "epoch": 0.6899957428693061,
      "grad_norm": 0.79296875,
      "learning_rate": 1.4713704756471456e-05,
      "loss": 0.3157,
      "step": 9117
    },
    {
      "epoch": 0.6900714251927534,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4712654267524695e-05,
      "loss": 0.3028,
      "step": 9118
    },
    {
      "epoch": 0.6901471075162008,
      "grad_norm": 0.734375,
      "learning_rate": 1.471160371172085e-05,
      "loss": 0.2737,
      "step": 9119
    },
    {
      "epoch": 0.6902227898396481,
      "grad_norm": 0.765625,
      "learning_rate": 1.4710553089074817e-05,
      "loss": 0.3131,
      "step": 9120
    },
    {
      "epoch": 0.6902984721630954,
      "grad_norm": 0.71875,
      "learning_rate": 1.4709502399601507e-05,
      "loss": 0.3028,
      "step": 9121
    },
    {
      "epoch": 0.6903741544865427,
      "grad_norm": 0.765625,
      "learning_rate": 1.4708451643315827e-05,
      "loss": 0.3288,
      "step": 9122
    },
    {
      "epoch": 0.6904498368099901,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4707400820232675e-05,
      "loss": 0.298,
      "step": 9123
    },
    {
      "epoch": 0.6905255191334374,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4706349930366965e-05,
      "loss": 0.328,
      "step": 9124
    },
    {
      "epoch": 0.6906012014568848,
      "grad_norm": 1.609375,
      "learning_rate": 1.4705298973733607e-05,
      "loss": 0.3488,
      "step": 9125
    },
    {
      "epoch": 0.6906768837803321,
      "grad_norm": 0.9296875,
      "learning_rate": 1.4704247950347508e-05,
      "loss": 0.3325,
      "step": 9126
    },
    {
      "epoch": 0.6907525661037793,
      "grad_norm": 0.8125,
      "learning_rate": 1.4703196860223579e-05,
      "loss": 0.3362,
      "step": 9127
    },
    {
      "epoch": 0.6908282484272267,
      "grad_norm": 0.78125,
      "learning_rate": 1.4702145703376731e-05,
      "loss": 0.3023,
      "step": 9128
    },
    {
      "epoch": 0.690903930750674,
      "grad_norm": 0.75390625,
      "learning_rate": 1.4701094479821878e-05,
      "loss": 0.3316,
      "step": 9129
    },
    {
      "epoch": 0.6909796130741214,
      "grad_norm": 0.7421875,
      "learning_rate": 1.470004318957393e-05,
      "loss": 0.3138,
      "step": 9130
    },
    {
      "epoch": 0.6910552953975687,
      "grad_norm": 0.8046875,
      "learning_rate": 1.4698991832647807e-05,
      "loss": 0.3418,
      "step": 9131
    },
    {
      "epoch": 0.6911309777210161,
      "grad_norm": 0.72265625,
      "learning_rate": 1.469794040905842e-05,
      "loss": 0.2723,
      "step": 9132
    },
    {
      "epoch": 0.6912066600444634,
      "grad_norm": 0.79296875,
      "learning_rate": 1.4696888918820684e-05,
      "loss": 0.3405,
      "step": 9133
    },
    {
      "epoch": 0.6912823423679106,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4695837361949522e-05,
      "loss": 0.3202,
      "step": 9134
    },
    {
      "epoch": 0.691358024691358,
      "grad_norm": 0.70703125,
      "learning_rate": 1.4694785738459847e-05,
      "loss": 0.2841,
      "step": 9135
    },
    {
      "epoch": 0.6914337070148053,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4693734048366582e-05,
      "loss": 0.3142,
      "step": 9136
    },
    {
      "epoch": 0.6915093893382527,
      "grad_norm": 1.0,
      "learning_rate": 1.4692682291684642e-05,
      "loss": 0.3409,
      "step": 9137
    },
    {
      "epoch": 0.6915850716617,
      "grad_norm": 0.84765625,
      "learning_rate": 1.469163046842895e-05,
      "loss": 0.3273,
      "step": 9138
    },
    {
      "epoch": 0.6916607539851474,
      "grad_norm": 0.734375,
      "learning_rate": 1.469057857861443e-05,
      "loss": 0.3057,
      "step": 9139
    },
    {
      "epoch": 0.6917364363085947,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4689526622256005e-05,
      "loss": 0.3203,
      "step": 9140
    },
    {
      "epoch": 0.691812118632042,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4688474599368596e-05,
      "loss": 0.3128,
      "step": 9141
    },
    {
      "epoch": 0.6918878009554893,
      "grad_norm": 0.6796875,
      "learning_rate": 1.4687422509967129e-05,
      "loss": 0.2675,
      "step": 9142
    },
    {
      "epoch": 0.6919634832789366,
      "grad_norm": 0.71484375,
      "learning_rate": 1.468637035406653e-05,
      "loss": 0.2734,
      "step": 9143
    },
    {
      "epoch": 0.692039165602384,
      "grad_norm": 0.80078125,
      "learning_rate": 1.4685318131681725e-05,
      "loss": 0.328,
      "step": 9144
    },
    {
      "epoch": 0.6921148479258313,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4684265842827646e-05,
      "loss": 0.3539,
      "step": 9145
    },
    {
      "epoch": 0.6921905302492787,
      "grad_norm": 0.73046875,
      "learning_rate": 1.4683213487519212e-05,
      "loss": 0.3156,
      "step": 9146
    },
    {
      "epoch": 0.692266212572726,
      "grad_norm": 0.765625,
      "learning_rate": 1.468216106577136e-05,
      "loss": 0.3163,
      "step": 9147
    },
    {
      "epoch": 0.6923418948961733,
      "grad_norm": 0.6875,
      "learning_rate": 1.4681108577599017e-05,
      "loss": 0.2593,
      "step": 9148
    },
    {
      "epoch": 0.6924175772196206,
      "grad_norm": 0.83203125,
      "learning_rate": 1.4680056023017117e-05,
      "loss": 0.371,
      "step": 9149
    },
    {
      "epoch": 0.692493259543068,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4679003402040593e-05,
      "loss": 0.3394,
      "step": 9150
    },
    {
      "epoch": 0.6925689418665153,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4677950714684375e-05,
      "loss": 0.3177,
      "step": 9151
    },
    {
      "epoch": 0.6926446241899626,
      "grad_norm": 0.71875,
      "learning_rate": 1.4676897960963399e-05,
      "loss": 0.315,
      "step": 9152
    },
    {
      "epoch": 0.69272030651341,
      "grad_norm": 0.71875,
      "learning_rate": 1.4675845140892597e-05,
      "loss": 0.2601,
      "step": 9153
    },
    {
      "epoch": 0.6927959888368573,
      "grad_norm": 0.73046875,
      "learning_rate": 1.4674792254486909e-05,
      "loss": 0.3133,
      "step": 9154
    },
    {
      "epoch": 0.6928716711603046,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4673739301761273e-05,
      "loss": 0.3193,
      "step": 9155
    },
    {
      "epoch": 0.6929473534837519,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4672686282730622e-05,
      "loss": 0.3085,
      "step": 9156
    },
    {
      "epoch": 0.6930230358071993,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4671633197409896e-05,
      "loss": 0.324,
      "step": 9157
    },
    {
      "epoch": 0.6930987181306466,
      "grad_norm": 0.765625,
      "learning_rate": 1.467058004581404e-05,
      "loss": 0.3346,
      "step": 9158
    },
    {
      "epoch": 0.693174400454094,
      "grad_norm": 0.8046875,
      "learning_rate": 1.4669526827957989e-05,
      "loss": 0.3332,
      "step": 9159
    },
    {
      "epoch": 0.6932500827775413,
      "grad_norm": 0.8203125,
      "learning_rate": 1.4668473543856687e-05,
      "loss": 0.3877,
      "step": 9160
    },
    {
      "epoch": 0.6933257651009886,
      "grad_norm": 0.80859375,
      "learning_rate": 1.4667420193525078e-05,
      "loss": 0.3768,
      "step": 9161
    },
    {
      "epoch": 0.6934014474244359,
      "grad_norm": 0.875,
      "learning_rate": 1.46663667769781e-05,
      "loss": 0.356,
      "step": 9162
    },
    {
      "epoch": 0.6934771297478832,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4665313294230705e-05,
      "loss": 0.3078,
      "step": 9163
    },
    {
      "epoch": 0.6935528120713306,
      "grad_norm": 0.796875,
      "learning_rate": 1.4664259745297838e-05,
      "loss": 0.3264,
      "step": 9164
    },
    {
      "epoch": 0.6936284943947779,
      "grad_norm": 0.80078125,
      "learning_rate": 1.4663206130194434e-05,
      "loss": 0.3563,
      "step": 9165
    },
    {
      "epoch": 0.6937041767182253,
      "grad_norm": 0.81640625,
      "learning_rate": 1.4662152448935453e-05,
      "loss": 0.3336,
      "step": 9166
    },
    {
      "epoch": 0.6937798590416726,
      "grad_norm": 0.85546875,
      "learning_rate": 1.466109870153584e-05,
      "loss": 0.3297,
      "step": 9167
    },
    {
      "epoch": 0.69385554136512,
      "grad_norm": 1.0234375,
      "learning_rate": 1.4660044888010543e-05,
      "loss": 0.3733,
      "step": 9168
    },
    {
      "epoch": 0.6939312236885672,
      "grad_norm": 0.796875,
      "learning_rate": 1.4658991008374512e-05,
      "loss": 0.3595,
      "step": 9169
    },
    {
      "epoch": 0.6940069060120145,
      "grad_norm": 0.78515625,
      "learning_rate": 1.46579370626427e-05,
      "loss": 0.3116,
      "step": 9170
    },
    {
      "epoch": 0.6940825883354619,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4656883050830053e-05,
      "loss": 0.3483,
      "step": 9171
    },
    {
      "epoch": 0.6941582706589092,
      "grad_norm": 0.765625,
      "learning_rate": 1.4655828972951531e-05,
      "loss": 0.3279,
      "step": 9172
    },
    {
      "epoch": 0.6942339529823566,
      "grad_norm": 0.8203125,
      "learning_rate": 1.4654774829022087e-05,
      "loss": 0.3426,
      "step": 9173
    },
    {
      "epoch": 0.6943096353058039,
      "grad_norm": 0.78125,
      "learning_rate": 1.4653720619056672e-05,
      "loss": 0.3402,
      "step": 9174
    },
    {
      "epoch": 0.6943853176292513,
      "grad_norm": 0.765625,
      "learning_rate": 1.4652666343070246e-05,
      "loss": 0.3101,
      "step": 9175
    },
    {
      "epoch": 0.6944609999526985,
      "grad_norm": 0.77734375,
      "learning_rate": 1.465161200107776e-05,
      "loss": 0.3531,
      "step": 9176
    },
    {
      "epoch": 0.6945366822761458,
      "grad_norm": 0.734375,
      "learning_rate": 1.4650557593094176e-05,
      "loss": 0.2907,
      "step": 9177
    },
    {
      "epoch": 0.6946123645995932,
      "grad_norm": 0.80859375,
      "learning_rate": 1.4649503119134456e-05,
      "loss": 0.3603,
      "step": 9178
    },
    {
      "epoch": 0.6946880469230405,
      "grad_norm": 0.78515625,
      "learning_rate": 1.4648448579213552e-05,
      "loss": 0.3333,
      "step": 9179
    },
    {
      "epoch": 0.6947637292464879,
      "grad_norm": 0.7890625,
      "learning_rate": 1.464739397334643e-05,
      "loss": 0.3528,
      "step": 9180
    },
    {
      "epoch": 0.6948394115699352,
      "grad_norm": 0.82421875,
      "learning_rate": 1.4646339301548046e-05,
      "loss": 0.3472,
      "step": 9181
    },
    {
      "epoch": 0.6949150938933826,
      "grad_norm": 0.8203125,
      "learning_rate": 1.4645284563833368e-05,
      "loss": 0.3433,
      "step": 9182
    },
    {
      "epoch": 0.6949907762168298,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4644229760217356e-05,
      "loss": 0.305,
      "step": 9183
    },
    {
      "epoch": 0.6950664585402772,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4643174890714978e-05,
      "loss": 0.2848,
      "step": 9184
    },
    {
      "epoch": 0.6951421408637245,
      "grad_norm": 0.94140625,
      "learning_rate": 1.4642119955341194e-05,
      "loss": 0.3256,
      "step": 9185
    },
    {
      "epoch": 0.6952178231871718,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4641064954110969e-05,
      "loss": 0.3162,
      "step": 9186
    },
    {
      "epoch": 0.6952935055106192,
      "grad_norm": 0.796875,
      "learning_rate": 1.4640009887039279e-05,
      "loss": 0.358,
      "step": 9187
    },
    {
      "epoch": 0.6953691878340665,
      "grad_norm": 0.8203125,
      "learning_rate": 1.4638954754141084e-05,
      "loss": 0.353,
      "step": 9188
    },
    {
      "epoch": 0.6954448701575139,
      "grad_norm": 1.078125,
      "learning_rate": 1.4637899555431352e-05,
      "loss": 0.3647,
      "step": 9189
    },
    {
      "epoch": 0.6955205524809611,
      "grad_norm": 0.75390625,
      "learning_rate": 1.463684429092506e-05,
      "loss": 0.3207,
      "step": 9190
    },
    {
      "epoch": 0.6955962348044085,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4635788960637174e-05,
      "loss": 0.29,
      "step": 9191
    },
    {
      "epoch": 0.6956719171278558,
      "grad_norm": 0.8125,
      "learning_rate": 1.4634733564582666e-05,
      "loss": 0.3391,
      "step": 9192
    },
    {
      "epoch": 0.6957475994513032,
      "grad_norm": 0.79296875,
      "learning_rate": 1.4633678102776508e-05,
      "loss": 0.3282,
      "step": 9193
    },
    {
      "epoch": 0.6958232817747505,
      "grad_norm": 0.83984375,
      "learning_rate": 1.4632622575233674e-05,
      "loss": 0.3348,
      "step": 9194
    },
    {
      "epoch": 0.6958989640981978,
      "grad_norm": 1.125,
      "learning_rate": 1.4631566981969139e-05,
      "loss": 0.3091,
      "step": 9195
    },
    {
      "epoch": 0.6959746464216452,
      "grad_norm": 0.76953125,
      "learning_rate": 1.463051132299788e-05,
      "loss": 0.3056,
      "step": 9196
    },
    {
      "epoch": 0.6960503287450924,
      "grad_norm": 0.78125,
      "learning_rate": 1.4629455598334871e-05,
      "loss": 0.3227,
      "step": 9197
    },
    {
      "epoch": 0.6961260110685398,
      "grad_norm": 1.453125,
      "learning_rate": 1.462839980799509e-05,
      "loss": 0.3846,
      "step": 9198
    },
    {
      "epoch": 0.6962016933919871,
      "grad_norm": 0.78515625,
      "learning_rate": 1.4627343951993513e-05,
      "loss": 0.3255,
      "step": 9199
    },
    {
      "epoch": 0.6962773757154345,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4626288030345124e-05,
      "loss": 0.3094,
      "step": 9200
    },
    {
      "epoch": 0.6963530580388818,
      "grad_norm": 0.84765625,
      "learning_rate": 1.4625232043064901e-05,
      "loss": 0.3794,
      "step": 9201
    },
    {
      "epoch": 0.6964287403623292,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4624175990167824e-05,
      "loss": 0.3384,
      "step": 9202
    },
    {
      "epoch": 0.6965044226857765,
      "grad_norm": 0.8125,
      "learning_rate": 1.4623119871668871e-05,
      "loss": 0.3483,
      "step": 9203
    },
    {
      "epoch": 0.6965801050092237,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4622063687583033e-05,
      "loss": 0.2987,
      "step": 9204
    },
    {
      "epoch": 0.6966557873326711,
      "grad_norm": 0.71875,
      "learning_rate": 1.4621007437925287e-05,
      "loss": 0.2768,
      "step": 9205
    },
    {
      "epoch": 0.6967314696561184,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4619951122710624e-05,
      "loss": 0.3454,
      "step": 9206
    },
    {
      "epoch": 0.6968071519795658,
      "grad_norm": 0.828125,
      "learning_rate": 1.4618894741954024e-05,
      "loss": 0.3351,
      "step": 9207
    },
    {
      "epoch": 0.6968828343030131,
      "grad_norm": 0.9375,
      "learning_rate": 1.4617838295670477e-05,
      "loss": 0.322,
      "step": 9208
    },
    {
      "epoch": 0.6969585166264605,
      "grad_norm": 0.71875,
      "learning_rate": 1.4616781783874967e-05,
      "loss": 0.3006,
      "step": 9209
    },
    {
      "epoch": 0.6970341989499078,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4615725206582487e-05,
      "loss": 0.3237,
      "step": 9210
    },
    {
      "epoch": 0.697109881273355,
      "grad_norm": 1.078125,
      "learning_rate": 1.4614668563808025e-05,
      "loss": 0.3684,
      "step": 9211
    },
    {
      "epoch": 0.6971855635968024,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4613611855566567e-05,
      "loss": 0.3274,
      "step": 9212
    },
    {
      "epoch": 0.6972612459202497,
      "grad_norm": 0.74609375,
      "learning_rate": 1.461255508187311e-05,
      "loss": 0.3078,
      "step": 9213
    },
    {
      "epoch": 0.6973369282436971,
      "grad_norm": 0.96484375,
      "learning_rate": 1.4611498242742643e-05,
      "loss": 0.3336,
      "step": 9214
    },
    {
      "epoch": 0.6974126105671444,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4610441338190163e-05,
      "loss": 0.3045,
      "step": 9215
    },
    {
      "epoch": 0.6974882928905918,
      "grad_norm": 0.75,
      "learning_rate": 1.4609384368230658e-05,
      "loss": 0.328,
      "step": 9216
    },
    {
      "epoch": 0.6975639752140391,
      "grad_norm": 0.796875,
      "learning_rate": 1.4608327332879124e-05,
      "loss": 0.3659,
      "step": 9217
    },
    {
      "epoch": 0.6976396575374864,
      "grad_norm": 0.75,
      "learning_rate": 1.460727023215056e-05,
      "loss": 0.3228,
      "step": 9218
    },
    {
      "epoch": 0.6977153398609337,
      "grad_norm": 0.76953125,
      "learning_rate": 1.460621306605996e-05,
      "loss": 0.3182,
      "step": 9219
    },
    {
      "epoch": 0.697791022184381,
      "grad_norm": 0.765625,
      "learning_rate": 1.460515583462233e-05,
      "loss": 0.3141,
      "step": 9220
    },
    {
      "epoch": 0.6978667045078284,
      "grad_norm": 0.8203125,
      "learning_rate": 1.4604098537852653e-05,
      "loss": 0.3462,
      "step": 9221
    },
    {
      "epoch": 0.6979423868312757,
      "grad_norm": 0.75,
      "learning_rate": 1.4603041175765941e-05,
      "loss": 0.2978,
      "step": 9222
    },
    {
      "epoch": 0.6980180691547231,
      "grad_norm": 0.7265625,
      "learning_rate": 1.460198374837719e-05,
      "loss": 0.2814,
      "step": 9223
    },
    {
      "epoch": 0.6980937514781704,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4600926255701402e-05,
      "loss": 0.2952,
      "step": 9224
    },
    {
      "epoch": 0.6981694338016177,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4599868697753583e-05,
      "loss": 0.3185,
      "step": 9225
    },
    {
      "epoch": 0.698245116125065,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4598811074548732e-05,
      "loss": 0.2977,
      "step": 9226
    },
    {
      "epoch": 0.6983207984485124,
      "grad_norm": 0.8203125,
      "learning_rate": 1.459775338610185e-05,
      "loss": 0.3568,
      "step": 9227
    },
    {
      "epoch": 0.6983964807719597,
      "grad_norm": 0.80859375,
      "learning_rate": 1.4596695632427948e-05,
      "loss": 0.3565,
      "step": 9228
    },
    {
      "epoch": 0.698472163095407,
      "grad_norm": 1.1484375,
      "learning_rate": 1.4595637813542035e-05,
      "loss": 0.3844,
      "step": 9229
    },
    {
      "epoch": 0.6985478454188544,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4594579929459107e-05,
      "loss": 0.3171,
      "step": 9230
    },
    {
      "epoch": 0.6986235277423017,
      "grad_norm": 0.78515625,
      "learning_rate": 1.4593521980194181e-05,
      "loss": 0.302,
      "step": 9231
    },
    {
      "epoch": 0.698699210065749,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4592463965762262e-05,
      "loss": 0.3061,
      "step": 9232
    },
    {
      "epoch": 0.6987748923891963,
      "grad_norm": 0.84765625,
      "learning_rate": 1.459140588617836e-05,
      "loss": 0.3662,
      "step": 9233
    },
    {
      "epoch": 0.6988505747126437,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4590347741457489e-05,
      "loss": 0.3269,
      "step": 9234
    },
    {
      "epoch": 0.698926257036091,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4589289531614657e-05,
      "loss": 0.3003,
      "step": 9235
    },
    {
      "epoch": 0.6990019393595384,
      "grad_norm": 0.734375,
      "learning_rate": 1.4588231256664876e-05,
      "loss": 0.3049,
      "step": 9236
    },
    {
      "epoch": 0.6990776216829857,
      "grad_norm": 0.78515625,
      "learning_rate": 1.458717291662316e-05,
      "loss": 0.3272,
      "step": 9237
    },
    {
      "epoch": 0.699153304006433,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4586114511504527e-05,
      "loss": 0.3205,
      "step": 9238
    },
    {
      "epoch": 0.6992289863298803,
      "grad_norm": 0.65625,
      "learning_rate": 1.4585056041323991e-05,
      "loss": 0.2552,
      "step": 9239
    },
    {
      "epoch": 0.6993046686533276,
      "grad_norm": 0.765625,
      "learning_rate": 1.4583997506096563e-05,
      "loss": 0.3199,
      "step": 9240
    },
    {
      "epoch": 0.699380350976775,
      "grad_norm": 0.80078125,
      "learning_rate": 1.4582938905837267e-05,
      "loss": 0.3501,
      "step": 9241
    },
    {
      "epoch": 0.6994560333002223,
      "grad_norm": 0.80078125,
      "learning_rate": 1.4581880240561113e-05,
      "loss": 0.3725,
      "step": 9242
    },
    {
      "epoch": 0.6995317156236697,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4580821510283132e-05,
      "loss": 0.3165,
      "step": 9243
    },
    {
      "epoch": 0.699607397947117,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4579762715018332e-05,
      "loss": 0.3064,
      "step": 9244
    },
    {
      "epoch": 0.6996830802705644,
      "grad_norm": 0.7578125,
      "learning_rate": 1.457870385478174e-05,
      "loss": 0.2957,
      "step": 9245
    },
    {
      "epoch": 0.6997587625940116,
      "grad_norm": 0.81640625,
      "learning_rate": 1.4577644929588378e-05,
      "loss": 0.3337,
      "step": 9246
    },
    {
      "epoch": 0.6998344449174589,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4576585939453265e-05,
      "loss": 0.3029,
      "step": 9247
    },
    {
      "epoch": 0.6999101272409063,
      "grad_norm": 0.83984375,
      "learning_rate": 1.457552688439143e-05,
      "loss": 0.3007,
      "step": 9248
    },
    {
      "epoch": 0.6999858095643536,
      "grad_norm": 0.76953125,
      "learning_rate": 1.457446776441789e-05,
      "loss": 0.327,
      "step": 9249
    },
    {
      "epoch": 0.700061491887801,
      "grad_norm": 0.6796875,
      "learning_rate": 1.4573408579547676e-05,
      "loss": 0.2446,
      "step": 9250
    },
    {
      "epoch": 0.7001371742112483,
      "grad_norm": 0.82421875,
      "learning_rate": 1.4572349329795814e-05,
      "loss": 0.3611,
      "step": 9251
    },
    {
      "epoch": 0.7002128565346956,
      "grad_norm": 0.703125,
      "learning_rate": 1.4571290015177328e-05,
      "loss": 0.2884,
      "step": 9252
    },
    {
      "epoch": 0.7002885388581429,
      "grad_norm": 0.71875,
      "learning_rate": 1.4570230635707251e-05,
      "loss": 0.2919,
      "step": 9253
    },
    {
      "epoch": 0.7003642211815903,
      "grad_norm": 0.78125,
      "learning_rate": 1.4569171191400611e-05,
      "loss": 0.3352,
      "step": 9254
    },
    {
      "epoch": 0.7003642211815903,
      "eval_loss": 0.33362916111946106,
      "eval_runtime": 83.3144,
      "eval_samples_per_second": 58.357,
      "eval_steps_per_second": 58.357,
      "step": 9254
    },
    {
      "epoch": 0.7004399035050376,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4568111682272434e-05,
      "loss": 0.3506,
      "step": 9255
    },
    {
      "epoch": 0.7005155858284849,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4567052108337755e-05,
      "loss": 0.3164,
      "step": 9256
    },
    {
      "epoch": 0.7005912681519323,
      "grad_norm": 0.78515625,
      "learning_rate": 1.4565992469611606e-05,
      "loss": 0.3276,
      "step": 9257
    },
    {
      "epoch": 0.7006669504753796,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4564932766109015e-05,
      "loss": 0.3345,
      "step": 9258
    },
    {
      "epoch": 0.7007426327988269,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4563872997845018e-05,
      "loss": 0.2941,
      "step": 9259
    },
    {
      "epoch": 0.7008183151222742,
      "grad_norm": 0.796875,
      "learning_rate": 1.4562813164834656e-05,
      "loss": 0.3316,
      "step": 9260
    },
    {
      "epoch": 0.7008939974457216,
      "grad_norm": 0.8203125,
      "learning_rate": 1.4561753267092954e-05,
      "loss": 0.3901,
      "step": 9261
    },
    {
      "epoch": 0.7009696797691689,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4560693304634956e-05,
      "loss": 0.3239,
      "step": 9262
    },
    {
      "epoch": 0.7010453620926163,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4559633277475697e-05,
      "loss": 0.318,
      "step": 9263
    },
    {
      "epoch": 0.7011210444160636,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4558573185630217e-05,
      "loss": 0.3008,
      "step": 9264
    },
    {
      "epoch": 0.701196726739511,
      "grad_norm": 0.80859375,
      "learning_rate": 1.455751302911355e-05,
      "loss": 0.3633,
      "step": 9265
    },
    {
      "epoch": 0.7012724090629582,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4556452807940743e-05,
      "loss": 0.3349,
      "step": 9266
    },
    {
      "epoch": 0.7013480913864055,
      "grad_norm": 0.6875,
      "learning_rate": 1.4555392522126834e-05,
      "loss": 0.2496,
      "step": 9267
    },
    {
      "epoch": 0.7014237737098529,
      "grad_norm": 0.76171875,
      "learning_rate": 1.455433217168686e-05,
      "loss": 0.3172,
      "step": 9268
    },
    {
      "epoch": 0.7014994560333002,
      "grad_norm": 0.734375,
      "learning_rate": 1.4553271756635874e-05,
      "loss": 0.3085,
      "step": 9269
    },
    {
      "epoch": 0.7015751383567476,
      "grad_norm": 0.734375,
      "learning_rate": 1.4552211276988913e-05,
      "loss": 0.3133,
      "step": 9270
    },
    {
      "epoch": 0.7016508206801949,
      "grad_norm": 0.65234375,
      "learning_rate": 1.4551150732761022e-05,
      "loss": 0.2469,
      "step": 9271
    },
    {
      "epoch": 0.7017265030036423,
      "grad_norm": 0.85546875,
      "learning_rate": 1.4550090123967249e-05,
      "loss": 0.3936,
      "step": 9272
    },
    {
      "epoch": 0.7018021853270895,
      "grad_norm": 0.85546875,
      "learning_rate": 1.4549029450622638e-05,
      "loss": 0.3796,
      "step": 9273
    },
    {
      "epoch": 0.7018778676505368,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4547968712742236e-05,
      "loss": 0.3411,
      "step": 9274
    },
    {
      "epoch": 0.7019535499739842,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4546907910341096e-05,
      "loss": 0.3586,
      "step": 9275
    },
    {
      "epoch": 0.7020292322974315,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4545847043434266e-05,
      "loss": 0.3123,
      "step": 9276
    },
    {
      "epoch": 0.7021049146208789,
      "grad_norm": 0.72265625,
      "learning_rate": 1.4544786112036792e-05,
      "loss": 0.2934,
      "step": 9277
    },
    {
      "epoch": 0.7021805969443262,
      "grad_norm": 0.80078125,
      "learning_rate": 1.454372511616373e-05,
      "loss": 0.361,
      "step": 9278
    },
    {
      "epoch": 0.7022562792677736,
      "grad_norm": 0.7109375,
      "learning_rate": 1.4542664055830129e-05,
      "loss": 0.2675,
      "step": 9279
    },
    {
      "epoch": 0.7023319615912208,
      "grad_norm": 0.796875,
      "learning_rate": 1.454160293105104e-05,
      "loss": 0.3468,
      "step": 9280
    },
    {
      "epoch": 0.7024076439146681,
      "grad_norm": 0.84765625,
      "learning_rate": 1.4540541741841524e-05,
      "loss": 0.3279,
      "step": 9281
    },
    {
      "epoch": 0.7024833262381155,
      "grad_norm": 0.78125,
      "learning_rate": 1.453948048821663e-05,
      "loss": 0.3077,
      "step": 9282
    },
    {
      "epoch": 0.7025590085615628,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4538419170191413e-05,
      "loss": 0.2959,
      "step": 9283
    },
    {
      "epoch": 0.7026346908850102,
      "grad_norm": 0.70703125,
      "learning_rate": 1.4537357787780937e-05,
      "loss": 0.3062,
      "step": 9284
    },
    {
      "epoch": 0.7027103732084575,
      "grad_norm": 0.71484375,
      "learning_rate": 1.4536296341000253e-05,
      "loss": 0.2817,
      "step": 9285
    },
    {
      "epoch": 0.7027860555319049,
      "grad_norm": 0.75390625,
      "learning_rate": 1.4535234829864417e-05,
      "loss": 0.3094,
      "step": 9286
    },
    {
      "epoch": 0.7028617378553521,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4534173254388495e-05,
      "loss": 0.3309,
      "step": 9287
    },
    {
      "epoch": 0.7029374201787995,
      "grad_norm": 0.71484375,
      "learning_rate": 1.4533111614587545e-05,
      "loss": 0.27,
      "step": 9288
    },
    {
      "epoch": 0.7030131025022468,
      "grad_norm": 0.71875,
      "learning_rate": 1.4532049910476625e-05,
      "loss": 0.2844,
      "step": 9289
    },
    {
      "epoch": 0.7030887848256941,
      "grad_norm": 0.78125,
      "learning_rate": 1.4530988142070802e-05,
      "loss": 0.3485,
      "step": 9290
    },
    {
      "epoch": 0.7031644671491415,
      "grad_norm": 1.0546875,
      "learning_rate": 1.4529926309385139e-05,
      "loss": 0.3307,
      "step": 9291
    },
    {
      "epoch": 0.7032401494725888,
      "grad_norm": 0.734375,
      "learning_rate": 1.4528864412434691e-05,
      "loss": 0.282,
      "step": 9292
    },
    {
      "epoch": 0.7033158317960362,
      "grad_norm": 0.75390625,
      "learning_rate": 1.4527802451234535e-05,
      "loss": 0.2918,
      "step": 9293
    },
    {
      "epoch": 0.7033915141194834,
      "grad_norm": 0.84375,
      "learning_rate": 1.4526740425799731e-05,
      "loss": 0.3615,
      "step": 9294
    },
    {
      "epoch": 0.7034671964429308,
      "grad_norm": 0.75,
      "learning_rate": 1.4525678336145344e-05,
      "loss": 0.3068,
      "step": 9295
    },
    {
      "epoch": 0.7035428787663781,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4524616182286444e-05,
      "loss": 0.312,
      "step": 9296
    },
    {
      "epoch": 0.7036185610898255,
      "grad_norm": 0.8046875,
      "learning_rate": 1.4523553964238098e-05,
      "loss": 0.3211,
      "step": 9297
    },
    {
      "epoch": 0.7036942434132728,
      "grad_norm": 0.7109375,
      "learning_rate": 1.4522491682015377e-05,
      "loss": 0.2636,
      "step": 9298
    },
    {
      "epoch": 0.7037699257367201,
      "grad_norm": 0.828125,
      "learning_rate": 1.4521429335633352e-05,
      "loss": 0.3486,
      "step": 9299
    },
    {
      "epoch": 0.7038456080601675,
      "grad_norm": 0.78125,
      "learning_rate": 1.452036692510709e-05,
      "loss": 0.3276,
      "step": 9300
    },
    {
      "epoch": 0.7039212903836147,
      "grad_norm": 0.796875,
      "learning_rate": 1.4519304450451668e-05,
      "loss": 0.351,
      "step": 9301
    },
    {
      "epoch": 0.7039969727070621,
      "grad_norm": 0.734375,
      "learning_rate": 1.4518241911682157e-05,
      "loss": 0.3052,
      "step": 9302
    },
    {
      "epoch": 0.7040726550305094,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4517179308813631e-05,
      "loss": 0.331,
      "step": 9303
    },
    {
      "epoch": 0.7041483373539568,
      "grad_norm": 0.94140625,
      "learning_rate": 1.4516116641861164e-05,
      "loss": 0.3361,
      "step": 9304
    },
    {
      "epoch": 0.7042240196774041,
      "grad_norm": 0.8203125,
      "learning_rate": 1.4515053910839833e-05,
      "loss": 0.3415,
      "step": 9305
    },
    {
      "epoch": 0.7042997020008515,
      "grad_norm": 0.828125,
      "learning_rate": 1.4513991115764717e-05,
      "loss": 0.36,
      "step": 9306
    },
    {
      "epoch": 0.7043753843242988,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4512928256650884e-05,
      "loss": 0.3308,
      "step": 9307
    },
    {
      "epoch": 0.704451066647746,
      "grad_norm": 0.72265625,
      "learning_rate": 1.4511865333513424e-05,
      "loss": 0.2899,
      "step": 9308
    },
    {
      "epoch": 0.7045267489711934,
      "grad_norm": 1.1796875,
      "learning_rate": 1.4510802346367409e-05,
      "loss": 0.3904,
      "step": 9309
    },
    {
      "epoch": 0.7046024312946407,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4509739295227924e-05,
      "loss": 0.2988,
      "step": 9310
    },
    {
      "epoch": 0.7046781136180881,
      "grad_norm": 0.8203125,
      "learning_rate": 1.4508676180110048e-05,
      "loss": 0.3621,
      "step": 9311
    },
    {
      "epoch": 0.7047537959415354,
      "grad_norm": 0.70703125,
      "learning_rate": 1.4507613001028864e-05,
      "loss": 0.2726,
      "step": 9312
    },
    {
      "epoch": 0.7048294782649828,
      "grad_norm": 0.796875,
      "learning_rate": 1.4506549757999456e-05,
      "loss": 0.3405,
      "step": 9313
    },
    {
      "epoch": 0.7049051605884301,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4505486451036901e-05,
      "loss": 0.3336,
      "step": 9314
    },
    {
      "epoch": 0.7049808429118773,
      "grad_norm": 0.80859375,
      "learning_rate": 1.450442308015629e-05,
      "loss": 0.3795,
      "step": 9315
    },
    {
      "epoch": 0.7050565252353247,
      "grad_norm": 0.74609375,
      "learning_rate": 1.450335964537271e-05,
      "loss": 0.3085,
      "step": 9316
    },
    {
      "epoch": 0.705132207558772,
      "grad_norm": 0.67578125,
      "learning_rate": 1.4502296146701245e-05,
      "loss": 0.2713,
      "step": 9317
    },
    {
      "epoch": 0.7052078898822194,
      "grad_norm": 0.85546875,
      "learning_rate": 1.4501232584156984e-05,
      "loss": 0.3499,
      "step": 9318
    },
    {
      "epoch": 0.7052835722056667,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4500168957755012e-05,
      "loss": 0.33,
      "step": 9319
    },
    {
      "epoch": 0.7053592545291141,
      "grad_norm": 0.80859375,
      "learning_rate": 1.449910526751042e-05,
      "loss": 0.3205,
      "step": 9320
    },
    {
      "epoch": 0.7054349368525614,
      "grad_norm": 0.81640625,
      "learning_rate": 1.44980415134383e-05,
      "loss": 0.2836,
      "step": 9321
    },
    {
      "epoch": 0.7055106191760087,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4496977695553743e-05,
      "loss": 0.3448,
      "step": 9322
    },
    {
      "epoch": 0.705586301499456,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4495913813871843e-05,
      "loss": 0.289,
      "step": 9323
    },
    {
      "epoch": 0.7056619838229033,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4494849868407686e-05,
      "loss": 0.3353,
      "step": 9324
    },
    {
      "epoch": 0.7057376661463507,
      "grad_norm": 0.79296875,
      "learning_rate": 1.4493785859176373e-05,
      "loss": 0.3521,
      "step": 9325
    },
    {
      "epoch": 0.705813348469798,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4492721786192993e-05,
      "loss": 0.3343,
      "step": 9326
    },
    {
      "epoch": 0.7058890307932454,
      "grad_norm": 0.734375,
      "learning_rate": 1.4491657649472647e-05,
      "loss": 0.3072,
      "step": 9327
    },
    {
      "epoch": 0.7059647131166927,
      "grad_norm": 0.90234375,
      "learning_rate": 1.449059344903043e-05,
      "loss": 0.3825,
      "step": 9328
    },
    {
      "epoch": 0.70604039544014,
      "grad_norm": 0.78515625,
      "learning_rate": 1.4489529184881436e-05,
      "loss": 0.3367,
      "step": 9329
    },
    {
      "epoch": 0.7061160777635873,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4488464857040767e-05,
      "loss": 0.2835,
      "step": 9330
    },
    {
      "epoch": 0.7061917600870347,
      "grad_norm": 0.6796875,
      "learning_rate": 1.4487400465523522e-05,
      "loss": 0.2558,
      "step": 9331
    },
    {
      "epoch": 0.706267442410482,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4486336010344804e-05,
      "loss": 0.3501,
      "step": 9332
    },
    {
      "epoch": 0.7063431247339294,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4485271491519706e-05,
      "loss": 0.304,
      "step": 9333
    },
    {
      "epoch": 0.7064188070573767,
      "grad_norm": 1.125,
      "learning_rate": 1.4484206909063336e-05,
      "loss": 0.3262,
      "step": 9334
    },
    {
      "epoch": 0.706494489380824,
      "grad_norm": 0.7421875,
      "learning_rate": 1.44831422629908e-05,
      "loss": 0.3109,
      "step": 9335
    },
    {
      "epoch": 0.7065701717042713,
      "grad_norm": 0.76953125,
      "learning_rate": 1.448207755331719e-05,
      "loss": 0.3192,
      "step": 9336
    },
    {
      "epoch": 0.7066458540277186,
      "grad_norm": 0.84375,
      "learning_rate": 1.4481012780057625e-05,
      "loss": 0.3583,
      "step": 9337
    },
    {
      "epoch": 0.706721536351166,
      "grad_norm": 0.78515625,
      "learning_rate": 1.4479947943227201e-05,
      "loss": 0.3233,
      "step": 9338
    },
    {
      "epoch": 0.7067972186746133,
      "grad_norm": 1.8515625,
      "learning_rate": 1.4478883042841027e-05,
      "loss": 0.3682,
      "step": 9339
    },
    {
      "epoch": 0.7068729009980607,
      "grad_norm": 0.75390625,
      "learning_rate": 1.4477818078914214e-05,
      "loss": 0.2958,
      "step": 9340
    },
    {
      "epoch": 0.706948583321508,
      "grad_norm": 0.734375,
      "learning_rate": 1.4476753051461868e-05,
      "loss": 0.309,
      "step": 9341
    },
    {
      "epoch": 0.7070242656449554,
      "grad_norm": 0.7109375,
      "learning_rate": 1.4475687960499093e-05,
      "loss": 0.2734,
      "step": 9342
    },
    {
      "epoch": 0.7070999479684026,
      "grad_norm": 0.75390625,
      "learning_rate": 1.4474622806041005e-05,
      "loss": 0.3137,
      "step": 9343
    },
    {
      "epoch": 0.7071756302918499,
      "grad_norm": 0.90625,
      "learning_rate": 1.4473557588102716e-05,
      "loss": 0.3624,
      "step": 9344
    },
    {
      "epoch": 0.7072513126152973,
      "grad_norm": 0.75390625,
      "learning_rate": 1.4472492306699333e-05,
      "loss": 0.3026,
      "step": 9345
    },
    {
      "epoch": 0.7073269949387446,
      "grad_norm": 0.71484375,
      "learning_rate": 1.4471426961845976e-05,
      "loss": 0.2851,
      "step": 9346
    },
    {
      "epoch": 0.707402677262192,
      "grad_norm": 0.73828125,
      "learning_rate": 1.447036155355775e-05,
      "loss": 0.3325,
      "step": 9347
    },
    {
      "epoch": 0.7074783595856393,
      "grad_norm": 0.72265625,
      "learning_rate": 1.4469296081849776e-05,
      "loss": 0.3054,
      "step": 9348
    },
    {
      "epoch": 0.7075540419090867,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4468230546737165e-05,
      "loss": 0.3421,
      "step": 9349
    },
    {
      "epoch": 0.7076297242325339,
      "grad_norm": 0.84765625,
      "learning_rate": 1.446716494823504e-05,
      "loss": 0.3746,
      "step": 9350
    },
    {
      "epoch": 0.7077054065559812,
      "grad_norm": 0.75390625,
      "learning_rate": 1.446609928635851e-05,
      "loss": 0.3192,
      "step": 9351
    },
    {
      "epoch": 0.7077810888794286,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4465033561122702e-05,
      "loss": 0.3139,
      "step": 9352
    },
    {
      "epoch": 0.7078567712028759,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4463967772542726e-05,
      "loss": 0.3016,
      "step": 9353
    },
    {
      "epoch": 0.7079324535263233,
      "grad_norm": 1.0390625,
      "learning_rate": 1.446290192063371e-05,
      "loss": 0.3441,
      "step": 9354
    },
    {
      "epoch": 0.7080081358497706,
      "grad_norm": 0.80859375,
      "learning_rate": 1.4461836005410771e-05,
      "loss": 0.3578,
      "step": 9355
    },
    {
      "epoch": 0.708083818173218,
      "grad_norm": 0.74609375,
      "learning_rate": 1.446077002688903e-05,
      "loss": 0.3342,
      "step": 9356
    },
    {
      "epoch": 0.7081595004966652,
      "grad_norm": 0.828125,
      "learning_rate": 1.4459703985083613e-05,
      "loss": 0.3562,
      "step": 9357
    },
    {
      "epoch": 0.7082351828201126,
      "grad_norm": 0.78125,
      "learning_rate": 1.4458637880009641e-05,
      "loss": 0.3415,
      "step": 9358
    },
    {
      "epoch": 0.7083108651435599,
      "grad_norm": 0.75390625,
      "learning_rate": 1.4457571711682238e-05,
      "loss": 0.324,
      "step": 9359
    },
    {
      "epoch": 0.7083865474670072,
      "grad_norm": 0.80859375,
      "learning_rate": 1.4456505480116533e-05,
      "loss": 0.3209,
      "step": 9360
    },
    {
      "epoch": 0.7084622297904546,
      "grad_norm": 0.765625,
      "learning_rate": 1.4455439185327652e-05,
      "loss": 0.3243,
      "step": 9361
    },
    {
      "epoch": 0.7085379121139019,
      "grad_norm": 0.75,
      "learning_rate": 1.4454372827330719e-05,
      "loss": 0.3006,
      "step": 9362
    },
    {
      "epoch": 0.7086135944373493,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4453306406140859e-05,
      "loss": 0.3325,
      "step": 9363
    },
    {
      "epoch": 0.7086892767607965,
      "grad_norm": 0.95703125,
      "learning_rate": 1.4452239921773209e-05,
      "loss": 0.3001,
      "step": 9364
    },
    {
      "epoch": 0.7087649590842439,
      "grad_norm": 0.80078125,
      "learning_rate": 1.4451173374242894e-05,
      "loss": 0.3638,
      "step": 9365
    },
    {
      "epoch": 0.7088406414076912,
      "grad_norm": 0.7890625,
      "learning_rate": 1.445010676356505e-05,
      "loss": 0.3537,
      "step": 9366
    },
    {
      "epoch": 0.7089163237311386,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4449040089754802e-05,
      "loss": 0.3322,
      "step": 9367
    },
    {
      "epoch": 0.7089920060545859,
      "grad_norm": 0.75390625,
      "learning_rate": 1.4447973352827287e-05,
      "loss": 0.3398,
      "step": 9368
    },
    {
      "epoch": 0.7090676883780332,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4446906552797635e-05,
      "loss": 0.3139,
      "step": 9369
    },
    {
      "epoch": 0.7091433707014806,
      "grad_norm": 0.81640625,
      "learning_rate": 1.4445839689680985e-05,
      "loss": 0.3407,
      "step": 9370
    },
    {
      "epoch": 0.7092190530249278,
      "grad_norm": 0.8046875,
      "learning_rate": 1.4444772763492466e-05,
      "loss": 0.3539,
      "step": 9371
    },
    {
      "epoch": 0.7092947353483752,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4443705774247219e-05,
      "loss": 0.326,
      "step": 9372
    },
    {
      "epoch": 0.7093704176718225,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4442638721960381e-05,
      "loss": 0.34,
      "step": 9373
    },
    {
      "epoch": 0.7094460999952699,
      "grad_norm": 0.78515625,
      "learning_rate": 1.4441571606647089e-05,
      "loss": 0.3397,
      "step": 9374
    },
    {
      "epoch": 0.7095217823187172,
      "grad_norm": 0.80859375,
      "learning_rate": 1.4440504428322482e-05,
      "loss": 0.3248,
      "step": 9375
    },
    {
      "epoch": 0.7095974646421646,
      "grad_norm": 0.72265625,
      "learning_rate": 1.4439437187001699e-05,
      "loss": 0.2781,
      "step": 9376
    },
    {
      "epoch": 0.7096731469656118,
      "grad_norm": 0.765625,
      "learning_rate": 1.4438369882699879e-05,
      "loss": 0.3304,
      "step": 9377
    },
    {
      "epoch": 0.7097488292890591,
      "grad_norm": 0.75390625,
      "learning_rate": 1.4437302515432168e-05,
      "loss": 0.2893,
      "step": 9378
    },
    {
      "epoch": 0.7098245116125065,
      "grad_norm": 0.77734375,
      "learning_rate": 1.443623508521371e-05,
      "loss": 0.3357,
      "step": 9379
    },
    {
      "epoch": 0.7099001939359538,
      "grad_norm": 0.75,
      "learning_rate": 1.4435167592059638e-05,
      "loss": 0.326,
      "step": 9380
    },
    {
      "epoch": 0.7099758762594012,
      "grad_norm": 0.6953125,
      "learning_rate": 1.4434100035985105e-05,
      "loss": 0.2847,
      "step": 9381
    },
    {
      "epoch": 0.7100515585828485,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4433032417005255e-05,
      "loss": 0.3063,
      "step": 9382
    },
    {
      "epoch": 0.7101272409062959,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4431964735135231e-05,
      "loss": 0.3186,
      "step": 9383
    },
    {
      "epoch": 0.7102029232297431,
      "grad_norm": 0.75,
      "learning_rate": 1.4430896990390184e-05,
      "loss": 0.3045,
      "step": 9384
    },
    {
      "epoch": 0.7102786055531904,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4429829182785258e-05,
      "loss": 0.3328,
      "step": 9385
    },
    {
      "epoch": 0.7103542878766378,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4428761312335601e-05,
      "loss": 0.2961,
      "step": 9386
    },
    {
      "epoch": 0.7104299702000851,
      "grad_norm": 0.80859375,
      "learning_rate": 1.4427693379056366e-05,
      "loss": 0.3227,
      "step": 9387
    },
    {
      "epoch": 0.7105056525235325,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4426625382962706e-05,
      "loss": 0.3313,
      "step": 9388
    },
    {
      "epoch": 0.7105813348469798,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4425557324069766e-05,
      "loss": 0.3278,
      "step": 9389
    },
    {
      "epoch": 0.7106570171704272,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4424489202392699e-05,
      "loss": 0.3102,
      "step": 9390
    },
    {
      "epoch": 0.7107326994938744,
      "grad_norm": 0.765625,
      "learning_rate": 1.4423421017946662e-05,
      "loss": 0.3245,
      "step": 9391
    },
    {
      "epoch": 0.7108083818173218,
      "grad_norm": 0.72265625,
      "learning_rate": 1.4422352770746807e-05,
      "loss": 0.2737,
      "step": 9392
    },
    {
      "epoch": 0.7108840641407691,
      "grad_norm": 0.8125,
      "learning_rate": 1.442128446080829e-05,
      "loss": 0.3373,
      "step": 9393
    },
    {
      "epoch": 0.7109597464642164,
      "grad_norm": 1.0546875,
      "learning_rate": 1.4420216088146263e-05,
      "loss": 0.3658,
      "step": 9394
    },
    {
      "epoch": 0.7110354287876638,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4419147652775887e-05,
      "loss": 0.3055,
      "step": 9395
    },
    {
      "epoch": 0.7111111111111111,
      "grad_norm": 0.84375,
      "learning_rate": 1.4418079154712317e-05,
      "loss": 0.3502,
      "step": 9396
    },
    {
      "epoch": 0.7111867934345585,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4417010593970716e-05,
      "loss": 0.3305,
      "step": 9397
    },
    {
      "epoch": 0.7112624757580057,
      "grad_norm": 0.8046875,
      "learning_rate": 1.4415941970566234e-05,
      "loss": 0.323,
      "step": 9398
    },
    {
      "epoch": 0.7113381580814531,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4414873284514041e-05,
      "loss": 0.3398,
      "step": 9399
    },
    {
      "epoch": 0.7114138404049004,
      "grad_norm": 0.80078125,
      "learning_rate": 1.4413804535829295e-05,
      "loss": 0.346,
      "step": 9400
    },
    {
      "epoch": 0.7114895227283478,
      "grad_norm": 0.6328125,
      "learning_rate": 1.4412735724527154e-05,
      "loss": 0.2317,
      "step": 9401
    },
    {
      "epoch": 0.7115652050517951,
      "grad_norm": 0.81640625,
      "learning_rate": 1.4411666850622788e-05,
      "loss": 0.3365,
      "step": 9402
    },
    {
      "epoch": 0.7116408873752424,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4410597914131354e-05,
      "loss": 0.3088,
      "step": 9403
    },
    {
      "epoch": 0.7117165696986898,
      "grad_norm": 0.78125,
      "learning_rate": 1.440952891506802e-05,
      "loss": 0.318,
      "step": 9404
    },
    {
      "epoch": 0.711792252022137,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4408459853447952e-05,
      "loss": 0.2996,
      "step": 9405
    },
    {
      "epoch": 0.7118679343455844,
      "grad_norm": 0.828125,
      "learning_rate": 1.4407390729286318e-05,
      "loss": 0.3319,
      "step": 9406
    },
    {
      "epoch": 0.7119436166690317,
      "grad_norm": 0.7890625,
      "learning_rate": 1.440632154259828e-05,
      "loss": 0.3537,
      "step": 9407
    },
    {
      "epoch": 0.7120192989924791,
      "grad_norm": 0.76953125,
      "learning_rate": 1.440525229339901e-05,
      "loss": 0.2996,
      "step": 9408
    },
    {
      "epoch": 0.7120949813159264,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4404182981703679e-05,
      "loss": 0.3389,
      "step": 9409
    },
    {
      "epoch": 0.7121706636393738,
      "grad_norm": 0.734375,
      "learning_rate": 1.4403113607527451e-05,
      "loss": 0.3242,
      "step": 9410
    },
    {
      "epoch": 0.7122463459628211,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4402044170885503e-05,
      "loss": 0.3175,
      "step": 9411
    },
    {
      "epoch": 0.7123220282862683,
      "grad_norm": 0.83203125,
      "learning_rate": 1.4400974671793002e-05,
      "loss": 0.3808,
      "step": 9412
    },
    {
      "epoch": 0.7123977106097157,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4399905110265124e-05,
      "loss": 0.3254,
      "step": 9413
    },
    {
      "epoch": 0.712473392933163,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4398835486317042e-05,
      "loss": 0.2928,
      "step": 9414
    },
    {
      "epoch": 0.7125490752566104,
      "grad_norm": 0.7578125,
      "learning_rate": 1.439776579996393e-05,
      "loss": 0.3197,
      "step": 9415
    },
    {
      "epoch": 0.7126247575800577,
      "grad_norm": 0.8125,
      "learning_rate": 1.4396696051220965e-05,
      "loss": 0.3718,
      "step": 9416
    },
    {
      "epoch": 0.7127004399035051,
      "grad_norm": 0.73046875,
      "learning_rate": 1.439562624010332e-05,
      "loss": 0.2885,
      "step": 9417
    },
    {
      "epoch": 0.7127761222269524,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4394556366626174e-05,
      "loss": 0.3199,
      "step": 9418
    },
    {
      "epoch": 0.7128518045503996,
      "grad_norm": 0.79296875,
      "learning_rate": 1.4393486430804701e-05,
      "loss": 0.3339,
      "step": 9419
    },
    {
      "epoch": 0.712927486873847,
      "grad_norm": 0.69921875,
      "learning_rate": 1.4392416432654086e-05,
      "loss": 0.2858,
      "step": 9420
    },
    {
      "epoch": 0.7130031691972943,
      "grad_norm": 0.73046875,
      "learning_rate": 1.4391346372189507e-05,
      "loss": 0.2819,
      "step": 9421
    },
    {
      "epoch": 0.7130788515207417,
      "grad_norm": 0.78125,
      "learning_rate": 1.4390276249426145e-05,
      "loss": 0.3188,
      "step": 9422
    },
    {
      "epoch": 0.713154533844189,
      "grad_norm": 0.78125,
      "learning_rate": 1.438920606437918e-05,
      "loss": 0.2939,
      "step": 9423
    },
    {
      "epoch": 0.7132302161676364,
      "grad_norm": 0.73046875,
      "learning_rate": 1.4388135817063793e-05,
      "loss": 0.2726,
      "step": 9424
    },
    {
      "epoch": 0.7133058984910837,
      "grad_norm": 0.73828125,
      "learning_rate": 1.438706550749517e-05,
      "loss": 0.3124,
      "step": 9425
    },
    {
      "epoch": 0.713381580814531,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4385995135688495e-05,
      "loss": 0.31,
      "step": 9426
    },
    {
      "epoch": 0.7134572631379783,
      "grad_norm": 0.78515625,
      "learning_rate": 1.4384924701658948e-05,
      "loss": 0.3547,
      "step": 9427
    },
    {
      "epoch": 0.7135329454614256,
      "grad_norm": 0.78125,
      "learning_rate": 1.4383854205421726e-05,
      "loss": 0.3468,
      "step": 9428
    },
    {
      "epoch": 0.713608627784873,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4382783646992006e-05,
      "loss": 0.3132,
      "step": 9429
    },
    {
      "epoch": 0.7136843101083203,
      "grad_norm": 0.77734375,
      "learning_rate": 1.438171302638498e-05,
      "loss": 0.3131,
      "step": 9430
    },
    {
      "epoch": 0.7137599924317677,
      "grad_norm": 0.734375,
      "learning_rate": 1.4380642343615834e-05,
      "loss": 0.2923,
      "step": 9431
    },
    {
      "epoch": 0.713835674755215,
      "grad_norm": 0.69921875,
      "learning_rate": 1.437957159869976e-05,
      "loss": 0.2918,
      "step": 9432
    },
    {
      "epoch": 0.7139113570786623,
      "grad_norm": 0.71484375,
      "learning_rate": 1.4378500791651944e-05,
      "loss": 0.2975,
      "step": 9433
    },
    {
      "epoch": 0.7139870394021096,
      "grad_norm": 1.0546875,
      "learning_rate": 1.4377429922487584e-05,
      "loss": 0.3315,
      "step": 9434
    },
    {
      "epoch": 0.714062721725557,
      "grad_norm": 0.8203125,
      "learning_rate": 1.4376358991221869e-05,
      "loss": 0.3436,
      "step": 9435
    },
    {
      "epoch": 0.7141384040490043,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4375287997869988e-05,
      "loss": 0.317,
      "step": 9436
    },
    {
      "epoch": 0.7142140863724517,
      "grad_norm": 0.69921875,
      "learning_rate": 1.4374216942447144e-05,
      "loss": 0.2791,
      "step": 9437
    },
    {
      "epoch": 0.714289768695899,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4373145824968521e-05,
      "loss": 0.3259,
      "step": 9438
    },
    {
      "epoch": 0.7143654510193463,
      "grad_norm": 1.1484375,
      "learning_rate": 1.4372074645449322e-05,
      "loss": 0.353,
      "step": 9439
    },
    {
      "epoch": 0.7144411333427936,
      "grad_norm": 0.91015625,
      "learning_rate": 1.4371003403904742e-05,
      "loss": 0.3557,
      "step": 9440
    },
    {
      "epoch": 0.7145168156662409,
      "grad_norm": 0.78515625,
      "learning_rate": 1.4369932100349977e-05,
      "loss": 0.3499,
      "step": 9441
    },
    {
      "epoch": 0.7145924979896883,
      "grad_norm": 0.80078125,
      "learning_rate": 1.4368860734800226e-05,
      "loss": 0.3251,
      "step": 9442
    },
    {
      "epoch": 0.7146681803131356,
      "grad_norm": 0.71875,
      "learning_rate": 1.4367789307270688e-05,
      "loss": 0.3152,
      "step": 9443
    },
    {
      "epoch": 0.714743862636583,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4366717817776567e-05,
      "loss": 0.3595,
      "step": 9444
    },
    {
      "epoch": 0.7148195449600303,
      "grad_norm": 0.765625,
      "learning_rate": 1.4365646266333055e-05,
      "loss": 0.3254,
      "step": 9445
    },
    {
      "epoch": 0.7148952272834777,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4364574652955361e-05,
      "loss": 0.3353,
      "step": 9446
    },
    {
      "epoch": 0.7149709096069249,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4363502977658685e-05,
      "loss": 0.3085,
      "step": 9447
    },
    {
      "epoch": 0.7150465919303722,
      "grad_norm": 0.734375,
      "learning_rate": 1.4362431240458232e-05,
      "loss": 0.2974,
      "step": 9448
    },
    {
      "epoch": 0.7151222742538196,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4361359441369206e-05,
      "loss": 0.3319,
      "step": 9449
    },
    {
      "epoch": 0.7151979565772669,
      "grad_norm": 0.80859375,
      "learning_rate": 1.4360287580406813e-05,
      "loss": 0.3619,
      "step": 9450
    },
    {
      "epoch": 0.7152736389007143,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4359215657586257e-05,
      "loss": 0.3162,
      "step": 9451
    },
    {
      "epoch": 0.7153493212241616,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4358143672922747e-05,
      "loss": 0.3028,
      "step": 9452
    },
    {
      "epoch": 0.715425003547609,
      "grad_norm": 0.7109375,
      "learning_rate": 1.4357071626431489e-05,
      "loss": 0.2855,
      "step": 9453
    },
    {
      "epoch": 0.7155006858710562,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4355999518127692e-05,
      "loss": 0.3339,
      "step": 9454
    },
    {
      "epoch": 0.7155763681945035,
      "grad_norm": 0.75,
      "learning_rate": 1.4354927348026569e-05,
      "loss": 0.2984,
      "step": 9455
    },
    {
      "epoch": 0.7156520505179509,
      "grad_norm": 0.79296875,
      "learning_rate": 1.4353855116143327e-05,
      "loss": 0.3374,
      "step": 9456
    },
    {
      "epoch": 0.7157277328413982,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4352782822493176e-05,
      "loss": 0.3621,
      "step": 9457
    },
    {
      "epoch": 0.7158034151648456,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4351710467091337e-05,
      "loss": 0.3221,
      "step": 9458
    },
    {
      "epoch": 0.7158790974882929,
      "grad_norm": 0.765625,
      "learning_rate": 1.4350638049953014e-05,
      "loss": 0.3483,
      "step": 9459
    },
    {
      "epoch": 0.7159547798117403,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4349565571093423e-05,
      "loss": 0.3013,
      "step": 9460
    },
    {
      "epoch": 0.7160304621351875,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4348493030527781e-05,
      "loss": 0.3217,
      "step": 9461
    },
    {
      "epoch": 0.7161061444586349,
      "grad_norm": 0.765625,
      "learning_rate": 1.4347420428271304e-05,
      "loss": 0.3301,
      "step": 9462
    },
    {
      "epoch": 0.7161818267820822,
      "grad_norm": 1.1953125,
      "learning_rate": 1.4346347764339208e-05,
      "loss": 0.3712,
      "step": 9463
    },
    {
      "epoch": 0.7162575091055295,
      "grad_norm": 0.75,
      "learning_rate": 1.434527503874671e-05,
      "loss": 0.2814,
      "step": 9464
    },
    {
      "epoch": 0.7163331914289769,
      "grad_norm": 0.765625,
      "learning_rate": 1.434420225150903e-05,
      "loss": 0.3286,
      "step": 9465
    },
    {
      "epoch": 0.7164088737524242,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4343129402641383e-05,
      "loss": 0.3107,
      "step": 9466
    },
    {
      "epoch": 0.7164845560758716,
      "grad_norm": 0.703125,
      "learning_rate": 1.4342056492158998e-05,
      "loss": 0.2827,
      "step": 9467
    },
    {
      "epoch": 0.7165602383993188,
      "grad_norm": 0.71875,
      "learning_rate": 1.4340983520077083e-05,
      "loss": 0.3014,
      "step": 9468
    },
    {
      "epoch": 0.7166359207227662,
      "grad_norm": 0.796875,
      "learning_rate": 1.4339910486410873e-05,
      "loss": 0.3097,
      "step": 9469
    },
    {
      "epoch": 0.7167116030462135,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4338837391175582e-05,
      "loss": 0.3198,
      "step": 9470
    },
    {
      "epoch": 0.7167872853696609,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4337764234386439e-05,
      "loss": 0.3137,
      "step": 9471
    },
    {
      "epoch": 0.7168629676931082,
      "grad_norm": 0.765625,
      "learning_rate": 1.4336691016058665e-05,
      "loss": 0.308,
      "step": 9472
    },
    {
      "epoch": 0.7169386500165555,
      "grad_norm": 1.0234375,
      "learning_rate": 1.433561773620749e-05,
      "loss": 0.3335,
      "step": 9473
    },
    {
      "epoch": 0.7170143323400029,
      "grad_norm": 0.6953125,
      "learning_rate": 1.4334544394848134e-05,
      "loss": 0.2748,
      "step": 9474
    },
    {
      "epoch": 0.7170900146634501,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4333470991995827e-05,
      "loss": 0.3069,
      "step": 9475
    },
    {
      "epoch": 0.7171656969868975,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4332397527665799e-05,
      "loss": 0.3326,
      "step": 9476
    },
    {
      "epoch": 0.7172413793103448,
      "grad_norm": 0.8828125,
      "learning_rate": 1.4331324001873276e-05,
      "loss": 0.3977,
      "step": 9477
    },
    {
      "epoch": 0.7173170616337922,
      "grad_norm": 0.75,
      "learning_rate": 1.433025041463349e-05,
      "loss": 0.3234,
      "step": 9478
    },
    {
      "epoch": 0.7173927439572395,
      "grad_norm": 0.734375,
      "learning_rate": 1.4329176765961675e-05,
      "loss": 0.2785,
      "step": 9479
    },
    {
      "epoch": 0.7174684262806869,
      "grad_norm": 0.765625,
      "learning_rate": 1.4328103055873052e-05,
      "loss": 0.3109,
      "step": 9480
    },
    {
      "epoch": 0.7175441086041342,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4327029284382863e-05,
      "loss": 0.3066,
      "step": 9481
    },
    {
      "epoch": 0.7176197909275814,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4325955451506339e-05,
      "loss": 0.3252,
      "step": 9482
    },
    {
      "epoch": 0.7176954732510288,
      "grad_norm": 1.28125,
      "learning_rate": 1.4324881557258712e-05,
      "loss": 0.3777,
      "step": 9483
    },
    {
      "epoch": 0.7177711555744761,
      "grad_norm": 0.8203125,
      "learning_rate": 1.4323807601655218e-05,
      "loss": 0.3494,
      "step": 9484
    },
    {
      "epoch": 0.7178468378979235,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4322733584711093e-05,
      "loss": 0.3055,
      "step": 9485
    },
    {
      "epoch": 0.7179225202213708,
      "grad_norm": 0.71875,
      "learning_rate": 1.4321659506441577e-05,
      "loss": 0.2921,
      "step": 9486
    },
    {
      "epoch": 0.7179982025448182,
      "grad_norm": 0.71875,
      "learning_rate": 1.4320585366861901e-05,
      "loss": 0.2825,
      "step": 9487
    },
    {
      "epoch": 0.7180738848682655,
      "grad_norm": 0.703125,
      "learning_rate": 1.431951116598731e-05,
      "loss": 0.2632,
      "step": 9488
    },
    {
      "epoch": 0.7181495671917127,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4318436903833037e-05,
      "loss": 0.3335,
      "step": 9489
    },
    {
      "epoch": 0.7182252495151601,
      "grad_norm": 0.78515625,
      "learning_rate": 1.4317362580414327e-05,
      "loss": 0.3087,
      "step": 9490
    },
    {
      "epoch": 0.7183009318386074,
      "grad_norm": 0.8046875,
      "learning_rate": 1.4316288195746423e-05,
      "loss": 0.3271,
      "step": 9491
    },
    {
      "epoch": 0.7183766141620548,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4315213749844562e-05,
      "loss": 0.3183,
      "step": 9492
    },
    {
      "epoch": 0.7184522964855021,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4314139242723988e-05,
      "loss": 0.2982,
      "step": 9493
    },
    {
      "epoch": 0.7185279788089495,
      "grad_norm": 0.796875,
      "learning_rate": 1.4313064674399945e-05,
      "loss": 0.3392,
      "step": 9494
    },
    {
      "epoch": 0.7186036611323967,
      "grad_norm": 0.6796875,
      "learning_rate": 1.431199004488768e-05,
      "loss": 0.2586,
      "step": 9495
    },
    {
      "epoch": 0.718679343455844,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4310915354202438e-05,
      "loss": 0.3064,
      "step": 9496
    },
    {
      "epoch": 0.7187550257792914,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4309840602359462e-05,
      "loss": 0.3147,
      "step": 9497
    },
    {
      "epoch": 0.7188307081027387,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4308765789374e-05,
      "loss": 0.3008,
      "step": 9498
    },
    {
      "epoch": 0.7189063904261861,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4307690915261302e-05,
      "loss": 0.3112,
      "step": 9499
    },
    {
      "epoch": 0.7189820727496334,
      "grad_norm": 0.79296875,
      "learning_rate": 1.4306615980036618e-05,
      "loss": 0.3586,
      "step": 9500
    },
    {
      "epoch": 0.7190577550730808,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4305540983715193e-05,
      "loss": 0.3426,
      "step": 9501
    },
    {
      "epoch": 0.719133437396528,
      "grad_norm": 0.73046875,
      "learning_rate": 1.4304465926312281e-05,
      "loss": 0.3338,
      "step": 9502
    },
    {
      "epoch": 0.7192091197199754,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4303390807843135e-05,
      "loss": 0.3041,
      "step": 9503
    },
    {
      "epoch": 0.7192848020434227,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4302315628323002e-05,
      "loss": 0.3025,
      "step": 9504
    },
    {
      "epoch": 0.7193604843668701,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4301240387767144e-05,
      "loss": 0.3448,
      "step": 9505
    },
    {
      "epoch": 0.7194361666903174,
      "grad_norm": 0.765625,
      "learning_rate": 1.4300165086190807e-05,
      "loss": 0.3397,
      "step": 9506
    },
    {
      "epoch": 0.7195118490137647,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4299089723609247e-05,
      "loss": 0.3017,
      "step": 9507
    },
    {
      "epoch": 0.7195875313372121,
      "grad_norm": 0.73046875,
      "learning_rate": 1.4298014300037726e-05,
      "loss": 0.3152,
      "step": 9508
    },
    {
      "epoch": 0.7196632136606593,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4296938815491493e-05,
      "loss": 0.3021,
      "step": 9509
    },
    {
      "epoch": 0.7197388959841067,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4295863269985813e-05,
      "loss": 0.2793,
      "step": 9510
    },
    {
      "epoch": 0.719814578307554,
      "grad_norm": 0.70703125,
      "learning_rate": 1.4294787663535936e-05,
      "loss": 0.2744,
      "step": 9511
    },
    {
      "epoch": 0.7198902606310014,
      "grad_norm": 0.8125,
      "learning_rate": 1.429371199615713e-05,
      "loss": 0.35,
      "step": 9512
    },
    {
      "epoch": 0.7199659429544487,
      "grad_norm": 0.72265625,
      "learning_rate": 1.4292636267864645e-05,
      "loss": 0.302,
      "step": 9513
    },
    {
      "epoch": 0.7200416252778961,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4291560478673756e-05,
      "loss": 0.312,
      "step": 9514
    },
    {
      "epoch": 0.7201173076013434,
      "grad_norm": 0.69921875,
      "learning_rate": 1.4290484628599712e-05,
      "loss": 0.2946,
      "step": 9515
    },
    {
      "epoch": 0.7201929899247906,
      "grad_norm": 0.75,
      "learning_rate": 1.428940871765778e-05,
      "loss": 0.3065,
      "step": 9516
    },
    {
      "epoch": 0.720268672248238,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4288332745863226e-05,
      "loss": 0.329,
      "step": 9517
    },
    {
      "epoch": 0.7203443545716853,
      "grad_norm": 0.80859375,
      "learning_rate": 1.4287256713231314e-05,
      "loss": 0.337,
      "step": 9518
    },
    {
      "epoch": 0.7204200368951327,
      "grad_norm": 0.80859375,
      "learning_rate": 1.4286180619777308e-05,
      "loss": 0.3242,
      "step": 9519
    },
    {
      "epoch": 0.72049571921858,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4285104465516474e-05,
      "loss": 0.3181,
      "step": 9520
    },
    {
      "epoch": 0.7205714015420274,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4284028250464082e-05,
      "loss": 0.3458,
      "step": 9521
    },
    {
      "epoch": 0.7206470838654747,
      "grad_norm": 0.7109375,
      "learning_rate": 1.4282951974635392e-05,
      "loss": 0.2676,
      "step": 9522
    },
    {
      "epoch": 0.720722766188922,
      "grad_norm": 0.71875,
      "learning_rate": 1.4281875638045683e-05,
      "loss": 0.2973,
      "step": 9523
    },
    {
      "epoch": 0.7207984485123693,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4280799240710219e-05,
      "loss": 0.3411,
      "step": 9524
    },
    {
      "epoch": 0.7208741308358166,
      "grad_norm": 0.703125,
      "learning_rate": 1.4279722782644274e-05,
      "loss": 0.2784,
      "step": 9525
    },
    {
      "epoch": 0.720949813159264,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4278646263863114e-05,
      "loss": 0.3126,
      "step": 9526
    },
    {
      "epoch": 0.7210254954827113,
      "grad_norm": 0.78515625,
      "learning_rate": 1.4277569684382015e-05,
      "loss": 0.3103,
      "step": 9527
    },
    {
      "epoch": 0.7211011778061587,
      "grad_norm": 0.890625,
      "learning_rate": 1.427649304421625e-05,
      "loss": 0.2772,
      "step": 9528
    },
    {
      "epoch": 0.721176860129606,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4275416343381095e-05,
      "loss": 0.3243,
      "step": 9529
    },
    {
      "epoch": 0.7212525424530533,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4274339581891816e-05,
      "loss": 0.3016,
      "step": 9530
    },
    {
      "epoch": 0.7213282247765006,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4273262759763697e-05,
      "loss": 0.3304,
      "step": 9531
    },
    {
      "epoch": 0.721403907099948,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4272185877012014e-05,
      "loss": 0.2933,
      "step": 9532
    },
    {
      "epoch": 0.7214795894233953,
      "grad_norm": 0.81640625,
      "learning_rate": 1.4271108933652044e-05,
      "loss": 0.3486,
      "step": 9533
    },
    {
      "epoch": 0.7215552717468426,
      "grad_norm": 0.75390625,
      "learning_rate": 1.4270031929699064e-05,
      "loss": 0.3067,
      "step": 9534
    },
    {
      "epoch": 0.72163095407029,
      "grad_norm": 1.15625,
      "learning_rate": 1.4268954865168354e-05,
      "loss": 0.4152,
      "step": 9535
    },
    {
      "epoch": 0.7217066363937373,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4267877740075191e-05,
      "loss": 0.2947,
      "step": 9536
    },
    {
      "epoch": 0.7217823187171846,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4266800554434859e-05,
      "loss": 0.2696,
      "step": 9537
    },
    {
      "epoch": 0.7218580010406319,
      "grad_norm": 0.81640625,
      "learning_rate": 1.4265723308262643e-05,
      "loss": 0.3252,
      "step": 9538
    },
    {
      "epoch": 0.7219336833640793,
      "grad_norm": 0.86328125,
      "learning_rate": 1.4264646001573814e-05,
      "loss": 0.3768,
      "step": 9539
    },
    {
      "epoch": 0.7220093656875266,
      "grad_norm": 0.82421875,
      "learning_rate": 1.4263568634383668e-05,
      "loss": 0.3176,
      "step": 9540
    },
    {
      "epoch": 0.722085048010974,
      "grad_norm": 0.7109375,
      "learning_rate": 1.4262491206707484e-05,
      "loss": 0.2997,
      "step": 9541
    },
    {
      "epoch": 0.7221607303344213,
      "grad_norm": 0.84375,
      "learning_rate": 1.4261413718560549e-05,
      "loss": 0.3489,
      "step": 9542
    },
    {
      "epoch": 0.7222364126578686,
      "grad_norm": 3.21875,
      "learning_rate": 1.4260336169958145e-05,
      "loss": 0.3674,
      "step": 9543
    },
    {
      "epoch": 0.7223120949813159,
      "grad_norm": 1.1171875,
      "learning_rate": 1.4259258560915563e-05,
      "loss": 0.3884,
      "step": 9544
    },
    {
      "epoch": 0.7223877773047632,
      "grad_norm": 0.66015625,
      "learning_rate": 1.4258180891448086e-05,
      "loss": 0.2616,
      "step": 9545
    },
    {
      "epoch": 0.7224634596282106,
      "grad_norm": 0.78515625,
      "learning_rate": 1.4257103161571009e-05,
      "loss": 0.3032,
      "step": 9546
    },
    {
      "epoch": 0.7225391419516579,
      "grad_norm": 0.71875,
      "learning_rate": 1.425602537129962e-05,
      "loss": 0.2923,
      "step": 9547
    },
    {
      "epoch": 0.7226148242751053,
      "grad_norm": 0.6796875,
      "learning_rate": 1.4254947520649203e-05,
      "loss": 0.2803,
      "step": 9548
    },
    {
      "epoch": 0.7226905065985526,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4253869609635056e-05,
      "loss": 0.3296,
      "step": 9549
    },
    {
      "epoch": 0.722766188922,
      "grad_norm": 0.734375,
      "learning_rate": 1.4252791638272469e-05,
      "loss": 0.3003,
      "step": 9550
    },
    {
      "epoch": 0.7228418712454472,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4251713606576731e-05,
      "loss": 0.3214,
      "step": 9551
    },
    {
      "epoch": 0.7229175535688945,
      "grad_norm": 0.734375,
      "learning_rate": 1.4250635514563145e-05,
      "loss": 0.3083,
      "step": 9552
    },
    {
      "epoch": 0.7229932358923419,
      "grad_norm": 0.8125,
      "learning_rate": 1.4249557362246997e-05,
      "loss": 0.3281,
      "step": 9553
    },
    {
      "epoch": 0.7230689182157892,
      "grad_norm": 1.234375,
      "learning_rate": 1.4248479149643583e-05,
      "loss": 0.395,
      "step": 9554
    },
    {
      "epoch": 0.7231446005392366,
      "grad_norm": 0.70703125,
      "learning_rate": 1.4247400876768205e-05,
      "loss": 0.2673,
      "step": 9555
    },
    {
      "epoch": 0.7232202828626839,
      "grad_norm": 0.828125,
      "learning_rate": 1.424632254363616e-05,
      "loss": 0.36,
      "step": 9556
    },
    {
      "epoch": 0.7232959651861313,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4245244150262736e-05,
      "loss": 0.3034,
      "step": 9557
    },
    {
      "epoch": 0.7233716475095785,
      "grad_norm": 0.765625,
      "learning_rate": 1.4244165696663241e-05,
      "loss": 0.3231,
      "step": 9558
    },
    {
      "epoch": 0.7234473298330258,
      "grad_norm": 0.71875,
      "learning_rate": 1.4243087182852972e-05,
      "loss": 0.2905,
      "step": 9559
    },
    {
      "epoch": 0.7235230121564732,
      "grad_norm": 1.1328125,
      "learning_rate": 1.424200860884723e-05,
      "loss": 0.3246,
      "step": 9560
    },
    {
      "epoch": 0.7235986944799205,
      "grad_norm": 0.76953125,
      "learning_rate": 1.424092997466132e-05,
      "loss": 0.3345,
      "step": 9561
    },
    {
      "epoch": 0.7236743768033679,
      "grad_norm": 0.68359375,
      "learning_rate": 1.4239851280310538e-05,
      "loss": 0.2505,
      "step": 9562
    },
    {
      "epoch": 0.7237500591268152,
      "grad_norm": 0.90234375,
      "learning_rate": 1.4238772525810185e-05,
      "loss": 0.3473,
      "step": 9563
    },
    {
      "epoch": 0.7238257414502626,
      "grad_norm": 1.078125,
      "learning_rate": 1.4237693711175574e-05,
      "loss": 0.3415,
      "step": 9564
    },
    {
      "epoch": 0.7239014237737098,
      "grad_norm": 0.78515625,
      "learning_rate": 1.4236614836422006e-05,
      "loss": 0.3127,
      "step": 9565
    },
    {
      "epoch": 0.7239771060971572,
      "grad_norm": 0.8125,
      "learning_rate": 1.4235535901564788e-05,
      "loss": 0.3788,
      "step": 9566
    },
    {
      "epoch": 0.7240527884206045,
      "grad_norm": 0.6796875,
      "learning_rate": 1.423445690661922e-05,
      "loss": 0.2797,
      "step": 9567
    },
    {
      "epoch": 0.7241284707440518,
      "grad_norm": 0.73046875,
      "learning_rate": 1.4233377851600619e-05,
      "loss": 0.2998,
      "step": 9568
    },
    {
      "epoch": 0.7242041530674992,
      "grad_norm": 0.78125,
      "learning_rate": 1.4232298736524284e-05,
      "loss": 0.3298,
      "step": 9569
    },
    {
      "epoch": 0.7242798353909465,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4231219561405533e-05,
      "loss": 0.3771,
      "step": 9570
    },
    {
      "epoch": 0.7243555177143939,
      "grad_norm": 0.78515625,
      "learning_rate": 1.4230140326259672e-05,
      "loss": 0.3433,
      "step": 9571
    },
    {
      "epoch": 0.7244312000378411,
      "grad_norm": 0.75390625,
      "learning_rate": 1.4229061031102005e-05,
      "loss": 0.3025,
      "step": 9572
    },
    {
      "epoch": 0.7245068823612885,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4227981675947855e-05,
      "loss": 0.3174,
      "step": 9573
    },
    {
      "epoch": 0.7245825646847358,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4226902260812527e-05,
      "loss": 0.3368,
      "step": 9574
    },
    {
      "epoch": 0.7246582470081832,
      "grad_norm": 0.796875,
      "learning_rate": 1.4225822785711338e-05,
      "loss": 0.3418,
      "step": 9575
    },
    {
      "epoch": 0.7247339293316305,
      "grad_norm": 0.6953125,
      "learning_rate": 1.4224743250659601e-05,
      "loss": 0.2896,
      "step": 9576
    },
    {
      "epoch": 0.7248096116550778,
      "grad_norm": 0.73046875,
      "learning_rate": 1.422366365567263e-05,
      "loss": 0.2882,
      "step": 9577
    },
    {
      "epoch": 0.7248852939785252,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4222584000765743e-05,
      "loss": 0.2945,
      "step": 9578
    },
    {
      "epoch": 0.7249609763019724,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4221504285954256e-05,
      "loss": 0.2714,
      "step": 9579
    },
    {
      "epoch": 0.7250366586254198,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4220424511253486e-05,
      "loss": 0.3393,
      "step": 9580
    },
    {
      "epoch": 0.7251123409488671,
      "grad_norm": 0.734375,
      "learning_rate": 1.4219344676678753e-05,
      "loss": 0.2773,
      "step": 9581
    },
    {
      "epoch": 0.7251880232723145,
      "grad_norm": 0.8515625,
      "learning_rate": 1.4218264782245373e-05,
      "loss": 0.3354,
      "step": 9582
    },
    {
      "epoch": 0.7252637055957618,
      "grad_norm": 0.85546875,
      "learning_rate": 1.4217184827968668e-05,
      "loss": 0.3656,
      "step": 9583
    },
    {
      "epoch": 0.7253393879192092,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4216104813863962e-05,
      "loss": 0.297,
      "step": 9584
    },
    {
      "epoch": 0.7254150702426565,
      "grad_norm": 0.83984375,
      "learning_rate": 1.4215024739946572e-05,
      "loss": 0.3563,
      "step": 9585
    },
    {
      "epoch": 0.7254907525661037,
      "grad_norm": 0.8203125,
      "learning_rate": 1.4213944606231823e-05,
      "loss": 0.3193,
      "step": 9586
    },
    {
      "epoch": 0.7255664348895511,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4212864412735039e-05,
      "loss": 0.3247,
      "step": 9587
    },
    {
      "epoch": 0.7256421172129984,
      "grad_norm": 0.80859375,
      "learning_rate": 1.4211784159471545e-05,
      "loss": 0.3388,
      "step": 9588
    },
    {
      "epoch": 0.7257177995364458,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4210703846456664e-05,
      "loss": 0.3451,
      "step": 9589
    },
    {
      "epoch": 0.7257934818598931,
      "grad_norm": 0.75,
      "learning_rate": 1.4209623473705722e-05,
      "loss": 0.3165,
      "step": 9590
    },
    {
      "epoch": 0.7258691641833405,
      "grad_norm": 0.765625,
      "learning_rate": 1.4208543041234049e-05,
      "loss": 0.314,
      "step": 9591
    },
    {
      "epoch": 0.7259448465067878,
      "grad_norm": 0.765625,
      "learning_rate": 1.420746254905697e-05,
      "loss": 0.3119,
      "step": 9592
    },
    {
      "epoch": 0.726020528830235,
      "grad_norm": 0.72265625,
      "learning_rate": 1.4206381997189811e-05,
      "loss": 0.2735,
      "step": 9593
    },
    {
      "epoch": 0.7260962111536824,
      "grad_norm": 0.796875,
      "learning_rate": 1.4205301385647911e-05,
      "loss": 0.315,
      "step": 9594
    },
    {
      "epoch": 0.7261718934771297,
      "grad_norm": 0.7421875,
      "learning_rate": 1.420422071444659e-05,
      "loss": 0.3237,
      "step": 9595
    },
    {
      "epoch": 0.7262475758005771,
      "grad_norm": 0.6875,
      "learning_rate": 1.4203139983601184e-05,
      "loss": 0.3036,
      "step": 9596
    },
    {
      "epoch": 0.7263232581240244,
      "grad_norm": 1.1015625,
      "learning_rate": 1.4202059193127027e-05,
      "loss": 0.3802,
      "step": 9597
    },
    {
      "epoch": 0.7263989404474718,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4200978343039448e-05,
      "loss": 0.329,
      "step": 9598
    },
    {
      "epoch": 0.7264746227709191,
      "grad_norm": 0.78125,
      "learning_rate": 1.4199897433353782e-05,
      "loss": 0.345,
      "step": 9599
    },
    {
      "epoch": 0.7265503050943664,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4198816464085363e-05,
      "loss": 0.3262,
      "step": 9600
    },
    {
      "epoch": 0.7266259874178137,
      "grad_norm": 0.703125,
      "learning_rate": 1.4197735435249527e-05,
      "loss": 0.2925,
      "step": 9601
    },
    {
      "epoch": 0.726701669741261,
      "grad_norm": 0.79296875,
      "learning_rate": 1.4196654346861611e-05,
      "loss": 0.2697,
      "step": 9602
    },
    {
      "epoch": 0.7267773520647084,
      "grad_norm": 0.8203125,
      "learning_rate": 1.4195573198936954e-05,
      "loss": 0.3649,
      "step": 9603
    },
    {
      "epoch": 0.7268530343881557,
      "grad_norm": 0.765625,
      "learning_rate": 1.4194491991490886e-05,
      "loss": 0.3236,
      "step": 9604
    },
    {
      "epoch": 0.7269287167116031,
      "grad_norm": 0.8046875,
      "learning_rate": 1.4193410724538753e-05,
      "loss": 0.359,
      "step": 9605
    },
    {
      "epoch": 0.7270043990350504,
      "grad_norm": 0.765625,
      "learning_rate": 1.4192329398095894e-05,
      "loss": 0.3439,
      "step": 9606
    },
    {
      "epoch": 0.7270800813584977,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4191248012177647e-05,
      "loss": 0.3024,
      "step": 9607
    },
    {
      "epoch": 0.727155763681945,
      "grad_norm": 0.78125,
      "learning_rate": 1.4190166566799358e-05,
      "loss": 0.3209,
      "step": 9608
    },
    {
      "epoch": 0.7272314460053924,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4189085061976362e-05,
      "loss": 0.2992,
      "step": 9609
    },
    {
      "epoch": 0.7273071283288397,
      "grad_norm": 0.93359375,
      "learning_rate": 1.4188003497724005e-05,
      "loss": 0.3837,
      "step": 9610
    },
    {
      "epoch": 0.727382810652287,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4186921874057634e-05,
      "loss": 0.3177,
      "step": 9611
    },
    {
      "epoch": 0.7274584929757344,
      "grad_norm": 0.734375,
      "learning_rate": 1.418584019099259e-05,
      "loss": 0.3061,
      "step": 9612
    },
    {
      "epoch": 0.7275341752991817,
      "grad_norm": 0.75390625,
      "learning_rate": 1.418475844854422e-05,
      "loss": 0.2844,
      "step": 9613
    },
    {
      "epoch": 0.727609857622629,
      "grad_norm": 0.7265625,
      "learning_rate": 1.418367664672787e-05,
      "loss": 0.2845,
      "step": 9614
    },
    {
      "epoch": 0.7276855399460763,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4182594785558888e-05,
      "loss": 0.3185,
      "step": 9615
    },
    {
      "epoch": 0.7277612222695237,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4181512865052619e-05,
      "loss": 0.3234,
      "step": 9616
    },
    {
      "epoch": 0.727836904592971,
      "grad_norm": 0.796875,
      "learning_rate": 1.4180430885224419e-05,
      "loss": 0.3379,
      "step": 9617
    },
    {
      "epoch": 0.7279125869164184,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4179348846089628e-05,
      "loss": 0.2907,
      "step": 9618
    },
    {
      "epoch": 0.7279882692398657,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4178266747663602e-05,
      "loss": 0.35,
      "step": 9619
    },
    {
      "epoch": 0.7280639515633129,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4177184589961695e-05,
      "loss": 0.3286,
      "step": 9620
    },
    {
      "epoch": 0.7281396338867603,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4176102372999254e-05,
      "loss": 0.3177,
      "step": 9621
    },
    {
      "epoch": 0.7282153162102076,
      "grad_norm": 0.9296875,
      "learning_rate": 1.4175020096791636e-05,
      "loss": 0.3947,
      "step": 9622
    },
    {
      "epoch": 0.728290998533655,
      "grad_norm": 0.734375,
      "learning_rate": 1.4173937761354194e-05,
      "loss": 0.3302,
      "step": 9623
    },
    {
      "epoch": 0.7283666808571023,
      "grad_norm": 0.734375,
      "learning_rate": 1.417285536670228e-05,
      "loss": 0.3002,
      "step": 9624
    },
    {
      "epoch": 0.7284423631805497,
      "grad_norm": 0.80078125,
      "learning_rate": 1.4171772912851252e-05,
      "loss": 0.3306,
      "step": 9625
    },
    {
      "epoch": 0.728518045503997,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4170690399816469e-05,
      "loss": 0.2949,
      "step": 9626
    },
    {
      "epoch": 0.7285937278274442,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4169607827613284e-05,
      "loss": 0.3361,
      "step": 9627
    },
    {
      "epoch": 0.7286694101508916,
      "grad_norm": 0.72265625,
      "learning_rate": 1.4168525196257054e-05,
      "loss": 0.3065,
      "step": 9628
    },
    {
      "epoch": 0.7287450924743389,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4167442505763143e-05,
      "loss": 0.3047,
      "step": 9629
    },
    {
      "epoch": 0.7288207747977863,
      "grad_norm": 0.8203125,
      "learning_rate": 1.4166359756146908e-05,
      "loss": 0.3543,
      "step": 9630
    },
    {
      "epoch": 0.7288964571212336,
      "grad_norm": 0.703125,
      "learning_rate": 1.4165276947423711e-05,
      "loss": 0.2683,
      "step": 9631
    },
    {
      "epoch": 0.728972139444681,
      "grad_norm": 0.81640625,
      "learning_rate": 1.4164194079608913e-05,
      "loss": 0.3535,
      "step": 9632
    },
    {
      "epoch": 0.7290478217681283,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4163111152717874e-05,
      "loss": 0.3134,
      "step": 9633
    },
    {
      "epoch": 0.7291235040915756,
      "grad_norm": 0.7734375,
      "learning_rate": 1.416202816676596e-05,
      "loss": 0.3282,
      "step": 9634
    },
    {
      "epoch": 0.7291991864150229,
      "grad_norm": 0.875,
      "learning_rate": 1.4160945121768532e-05,
      "loss": 0.2954,
      "step": 9635
    },
    {
      "epoch": 0.7292748687384702,
      "grad_norm": 0.8359375,
      "learning_rate": 1.4159862017740962e-05,
      "loss": 0.3477,
      "step": 9636
    },
    {
      "epoch": 0.7293505510619176,
      "grad_norm": 0.8203125,
      "learning_rate": 1.4158778854698608e-05,
      "loss": 0.372,
      "step": 9637
    },
    {
      "epoch": 0.7294262333853649,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4157695632656838e-05,
      "loss": 0.3215,
      "step": 9638
    },
    {
      "epoch": 0.7295019157088123,
      "grad_norm": 0.8125,
      "learning_rate": 1.4156612351631022e-05,
      "loss": 0.3248,
      "step": 9639
    },
    {
      "epoch": 0.7295775980322596,
      "grad_norm": 0.80859375,
      "learning_rate": 1.4155529011636528e-05,
      "loss": 0.3202,
      "step": 9640
    },
    {
      "epoch": 0.7296532803557069,
      "grad_norm": 0.8203125,
      "learning_rate": 1.4154445612688724e-05,
      "loss": 0.3262,
      "step": 9641
    },
    {
      "epoch": 0.7297289626791542,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4153362154802976e-05,
      "loss": 0.3258,
      "step": 9642
    },
    {
      "epoch": 0.7298046450026016,
      "grad_norm": 0.80859375,
      "learning_rate": 1.415227863799466e-05,
      "loss": 0.3375,
      "step": 9643
    },
    {
      "epoch": 0.7298803273260489,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4151195062279147e-05,
      "loss": 0.2939,
      "step": 9644
    },
    {
      "epoch": 0.7299560096494963,
      "grad_norm": 0.8046875,
      "learning_rate": 1.415011142767181e-05,
      "loss": 0.3443,
      "step": 9645
    },
    {
      "epoch": 0.7300316919729436,
      "grad_norm": 0.71484375,
      "learning_rate": 1.4149027734188016e-05,
      "loss": 0.305,
      "step": 9646
    },
    {
      "epoch": 0.7301073742963909,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4147943981843147e-05,
      "loss": 0.341,
      "step": 9647
    },
    {
      "epoch": 0.7301830566198382,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4146860170652572e-05,
      "loss": 0.3048,
      "step": 9648
    },
    {
      "epoch": 0.7302587389432855,
      "grad_norm": 0.75390625,
      "learning_rate": 1.414577630063167e-05,
      "loss": 0.3316,
      "step": 9649
    },
    {
      "epoch": 0.7303344212667329,
      "grad_norm": 0.703125,
      "learning_rate": 1.414469237179582e-05,
      "loss": 0.2508,
      "step": 9650
    },
    {
      "epoch": 0.7304101035901802,
      "grad_norm": 0.80078125,
      "learning_rate": 1.4143608384160392e-05,
      "loss": 0.3729,
      "step": 9651
    },
    {
      "epoch": 0.7304857859136276,
      "grad_norm": 0.7421875,
      "learning_rate": 1.414252433774077e-05,
      "loss": 0.3207,
      "step": 9652
    },
    {
      "epoch": 0.7305614682370749,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4141440232552332e-05,
      "loss": 0.3163,
      "step": 9653
    },
    {
      "epoch": 0.7306371505605223,
      "grad_norm": 0.75,
      "learning_rate": 1.4140356068610459e-05,
      "loss": 0.3002,
      "step": 9654
    },
    {
      "epoch": 0.7307128328839695,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4139271845930526e-05,
      "loss": 0.3208,
      "step": 9655
    },
    {
      "epoch": 0.7307885152074168,
      "grad_norm": 0.8046875,
      "learning_rate": 1.4138187564527922e-05,
      "loss": 0.3603,
      "step": 9656
    },
    {
      "epoch": 0.7308641975308642,
      "grad_norm": 0.78125,
      "learning_rate": 1.4137103224418022e-05,
      "loss": 0.3282,
      "step": 9657
    },
    {
      "epoch": 0.7309398798543115,
      "grad_norm": 0.69921875,
      "learning_rate": 1.4136018825616218e-05,
      "loss": 0.2819,
      "step": 9658
    },
    {
      "epoch": 0.7310155621777589,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4134934368137889e-05,
      "loss": 0.2855,
      "step": 9659
    },
    {
      "epoch": 0.7310912445012062,
      "grad_norm": 0.75,
      "learning_rate": 1.4133849851998418e-05,
      "loss": 0.284,
      "step": 9660
    },
    {
      "epoch": 0.7311669268246536,
      "grad_norm": 0.6484375,
      "learning_rate": 1.4132765277213195e-05,
      "loss": 0.2371,
      "step": 9661
    },
    {
      "epoch": 0.7312426091481008,
      "grad_norm": 0.80078125,
      "learning_rate": 1.4131680643797601e-05,
      "loss": 0.3349,
      "step": 9662
    },
    {
      "epoch": 0.7313182914715481,
      "grad_norm": 0.8046875,
      "learning_rate": 1.4130595951767027e-05,
      "loss": 0.314,
      "step": 9663
    },
    {
      "epoch": 0.7313939737949955,
      "grad_norm": 0.765625,
      "learning_rate": 1.4129511201136865e-05,
      "loss": 0.333,
      "step": 9664
    },
    {
      "epoch": 0.7314696561184428,
      "grad_norm": 0.75390625,
      "learning_rate": 1.4128426391922498e-05,
      "loss": 0.3027,
      "step": 9665
    },
    {
      "epoch": 0.7315453384418902,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4127341524139317e-05,
      "loss": 0.3693,
      "step": 9666
    },
    {
      "epoch": 0.7316210207653375,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4126256597802715e-05,
      "loss": 0.3283,
      "step": 9667
    },
    {
      "epoch": 0.7316967030887849,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4125171612928083e-05,
      "loss": 0.3454,
      "step": 9668
    },
    {
      "epoch": 0.7317723854122321,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4124086569530811e-05,
      "loss": 0.3264,
      "step": 9669
    },
    {
      "epoch": 0.7318480677356795,
      "grad_norm": 0.80859375,
      "learning_rate": 1.4123001467626293e-05,
      "loss": 0.3894,
      "step": 9670
    },
    {
      "epoch": 0.7319237500591268,
      "grad_norm": 0.71875,
      "learning_rate": 1.4121916307229926e-05,
      "loss": 0.2895,
      "step": 9671
    },
    {
      "epoch": 0.7319994323825741,
      "grad_norm": 0.80859375,
      "learning_rate": 1.41208310883571e-05,
      "loss": 0.3731,
      "step": 9672
    },
    {
      "epoch": 0.7320751147060215,
      "grad_norm": 0.734375,
      "learning_rate": 1.4119745811023218e-05,
      "loss": 0.3125,
      "step": 9673
    },
    {
      "epoch": 0.7321507970294688,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4118660475243671e-05,
      "loss": 0.35,
      "step": 9674
    },
    {
      "epoch": 0.7322264793529162,
      "grad_norm": 0.72265625,
      "learning_rate": 1.4117575081033855e-05,
      "loss": 0.2988,
      "step": 9675
    },
    {
      "epoch": 0.7323021616763634,
      "grad_norm": 0.79296875,
      "learning_rate": 1.4116489628409169e-05,
      "loss": 0.3117,
      "step": 9676
    },
    {
      "epoch": 0.7323778439998108,
      "grad_norm": 0.7734375,
      "learning_rate": 1.4115404117385017e-05,
      "loss": 0.3038,
      "step": 9677
    },
    {
      "epoch": 0.7324535263232581,
      "grad_norm": 0.78515625,
      "learning_rate": 1.4114318547976793e-05,
      "loss": 0.3125,
      "step": 9678
    },
    {
      "epoch": 0.7325292086467055,
      "grad_norm": 0.75390625,
      "learning_rate": 1.4113232920199902e-05,
      "loss": 0.31,
      "step": 9679
    },
    {
      "epoch": 0.7326048909701528,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4112147234069742e-05,
      "loss": 0.3521,
      "step": 9680
    },
    {
      "epoch": 0.7326805732936001,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4111061489601716e-05,
      "loss": 0.3134,
      "step": 9681
    },
    {
      "epoch": 0.7327562556170475,
      "grad_norm": 0.765625,
      "learning_rate": 1.4109975686811231e-05,
      "loss": 0.3299,
      "step": 9682
    },
    {
      "epoch": 0.7328319379404947,
      "grad_norm": 0.78125,
      "learning_rate": 1.4108889825713688e-05,
      "loss": 0.3335,
      "step": 9683
    },
    {
      "epoch": 0.7329076202639421,
      "grad_norm": 0.9609375,
      "learning_rate": 1.4107803906324487e-05,
      "loss": 0.3272,
      "step": 9684
    },
    {
      "epoch": 0.7329833025873894,
      "grad_norm": 0.75390625,
      "learning_rate": 1.4106717928659039e-05,
      "loss": 0.3072,
      "step": 9685
    },
    {
      "epoch": 0.7330589849108368,
      "grad_norm": 0.734375,
      "learning_rate": 1.4105631892732754e-05,
      "loss": 0.2777,
      "step": 9686
    },
    {
      "epoch": 0.7331346672342841,
      "grad_norm": 0.8203125,
      "learning_rate": 1.4104545798561033e-05,
      "loss": 0.3576,
      "step": 9687
    },
    {
      "epoch": 0.7332103495577315,
      "grad_norm": 1.09375,
      "learning_rate": 1.4103459646159285e-05,
      "loss": 0.4036,
      "step": 9688
    },
    {
      "epoch": 0.7332860318811788,
      "grad_norm": 0.8046875,
      "learning_rate": 1.4102373435542922e-05,
      "loss": 0.3718,
      "step": 9689
    },
    {
      "epoch": 0.733361714204626,
      "grad_norm": 0.6953125,
      "learning_rate": 1.410128716672735e-05,
      "loss": 0.2736,
      "step": 9690
    },
    {
      "epoch": 0.7334373965280734,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4100200839727984e-05,
      "loss": 0.3348,
      "step": 9691
    },
    {
      "epoch": 0.7335130788515207,
      "grad_norm": 0.79296875,
      "learning_rate": 1.4099114454560231e-05,
      "loss": 0.3427,
      "step": 9692
    },
    {
      "epoch": 0.7335887611749681,
      "grad_norm": 1.0703125,
      "learning_rate": 1.4098028011239507e-05,
      "loss": 0.3827,
      "step": 9693
    },
    {
      "epoch": 0.7336644434984154,
      "grad_norm": 0.75390625,
      "learning_rate": 1.409694150978122e-05,
      "loss": 0.3041,
      "step": 9694
    },
    {
      "epoch": 0.7337401258218628,
      "grad_norm": 0.75,
      "learning_rate": 1.4095854950200792e-05,
      "loss": 0.3056,
      "step": 9695
    },
    {
      "epoch": 0.7338158081453101,
      "grad_norm": 0.71875,
      "learning_rate": 1.4094768332513629e-05,
      "loss": 0.301,
      "step": 9696
    },
    {
      "epoch": 0.7338914904687573,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4093681656735152e-05,
      "loss": 0.2975,
      "step": 9697
    },
    {
      "epoch": 0.7339671727922047,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4092594922880773e-05,
      "loss": 0.3165,
      "step": 9698
    },
    {
      "epoch": 0.734042855115652,
      "grad_norm": 0.72265625,
      "learning_rate": 1.4091508130965914e-05,
      "loss": 0.291,
      "step": 9699
    },
    {
      "epoch": 0.7341185374390994,
      "grad_norm": 0.7265625,
      "learning_rate": 1.409042128100599e-05,
      "loss": 0.3236,
      "step": 9700
    },
    {
      "epoch": 0.7341942197625467,
      "grad_norm": 0.73046875,
      "learning_rate": 1.4089334373016424e-05,
      "loss": 0.2931,
      "step": 9701
    },
    {
      "epoch": 0.7342699020859941,
      "grad_norm": 0.83203125,
      "learning_rate": 1.408824740701263e-05,
      "loss": 0.3637,
      "step": 9702
    },
    {
      "epoch": 0.7343455844094414,
      "grad_norm": 0.78515625,
      "learning_rate": 1.408716038301003e-05,
      "loss": 0.3117,
      "step": 9703
    },
    {
      "epoch": 0.7344212667328887,
      "grad_norm": 0.796875,
      "learning_rate": 1.4086073301024046e-05,
      "loss": 0.3373,
      "step": 9704
    },
    {
      "epoch": 0.734496949056336,
      "grad_norm": 0.765625,
      "learning_rate": 1.40849861610701e-05,
      "loss": 0.3167,
      "step": 9705
    },
    {
      "epoch": 0.7345726313797833,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4083898963163618e-05,
      "loss": 0.3109,
      "step": 9706
    },
    {
      "epoch": 0.7346483137032307,
      "grad_norm": 0.8203125,
      "learning_rate": 1.408281170732002e-05,
      "loss": 0.3375,
      "step": 9707
    },
    {
      "epoch": 0.734723996026678,
      "grad_norm": 0.80859375,
      "learning_rate": 1.408172439355473e-05,
      "loss": 0.3486,
      "step": 9708
    },
    {
      "epoch": 0.7347996783501254,
      "grad_norm": 0.78515625,
      "learning_rate": 1.4080637021883177e-05,
      "loss": 0.3488,
      "step": 9709
    },
    {
      "epoch": 0.7348753606735727,
      "grad_norm": 0.82421875,
      "learning_rate": 1.4079549592320782e-05,
      "loss": 0.3671,
      "step": 9710
    },
    {
      "epoch": 0.73495104299702,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4078462104882979e-05,
      "loss": 0.3044,
      "step": 9711
    },
    {
      "epoch": 0.7350267253204673,
      "grad_norm": 0.7109375,
      "learning_rate": 1.407737455958519e-05,
      "loss": 0.2772,
      "step": 9712
    },
    {
      "epoch": 0.7351024076439147,
      "grad_norm": 0.78515625,
      "learning_rate": 1.4076286956442846e-05,
      "loss": 0.3502,
      "step": 9713
    },
    {
      "epoch": 0.735178089967362,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4075199295471378e-05,
      "loss": 0.2847,
      "step": 9714
    },
    {
      "epoch": 0.7352537722908093,
      "grad_norm": 0.72265625,
      "learning_rate": 1.4074111576686215e-05,
      "loss": 0.3004,
      "step": 9715
    },
    {
      "epoch": 0.7353294546142567,
      "grad_norm": 0.76953125,
      "learning_rate": 1.4073023800102785e-05,
      "loss": 0.3285,
      "step": 9716
    },
    {
      "epoch": 0.735405136937704,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4071935965736523e-05,
      "loss": 0.298,
      "step": 9717
    },
    {
      "epoch": 0.7354808192611513,
      "grad_norm": 0.78125,
      "learning_rate": 1.4070848073602862e-05,
      "loss": 0.335,
      "step": 9718
    },
    {
      "epoch": 0.7355565015845986,
      "grad_norm": 0.73046875,
      "learning_rate": 1.4069760123717235e-05,
      "loss": 0.293,
      "step": 9719
    },
    {
      "epoch": 0.735632183908046,
      "grad_norm": 0.7109375,
      "learning_rate": 1.4068672116095081e-05,
      "loss": 0.2923,
      "step": 9720
    },
    {
      "epoch": 0.7357078662314933,
      "grad_norm": 0.671875,
      "learning_rate": 1.4067584050751827e-05,
      "loss": 0.2821,
      "step": 9721
    },
    {
      "epoch": 0.7357835485549407,
      "grad_norm": 0.78515625,
      "learning_rate": 1.4066495927702915e-05,
      "loss": 0.3492,
      "step": 9722
    },
    {
      "epoch": 0.735859230878388,
      "grad_norm": 0.80078125,
      "learning_rate": 1.4065407746963777e-05,
      "loss": 0.3478,
      "step": 9723
    },
    {
      "epoch": 0.7359349132018354,
      "grad_norm": 0.734375,
      "learning_rate": 1.4064319508549858e-05,
      "loss": 0.2865,
      "step": 9724
    },
    {
      "epoch": 0.7360105955252826,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4063231212476586e-05,
      "loss": 0.3181,
      "step": 9725
    },
    {
      "epoch": 0.7360862778487299,
      "grad_norm": 0.7734375,
      "learning_rate": 1.406214285875941e-05,
      "loss": 0.3289,
      "step": 9726
    },
    {
      "epoch": 0.7361619601721773,
      "grad_norm": 0.76171875,
      "learning_rate": 1.406105444741377e-05,
      "loss": 0.3098,
      "step": 9727
    },
    {
      "epoch": 0.7362376424956246,
      "grad_norm": 0.71484375,
      "learning_rate": 1.4059965978455097e-05,
      "loss": 0.2859,
      "step": 9728
    },
    {
      "epoch": 0.736313324819072,
      "grad_norm": 0.78515625,
      "learning_rate": 1.4058877451898844e-05,
      "loss": 0.3282,
      "step": 9729
    },
    {
      "epoch": 0.7363890071425193,
      "grad_norm": 0.765625,
      "learning_rate": 1.4057788867760448e-05,
      "loss": 0.3208,
      "step": 9730
    },
    {
      "epoch": 0.7364646894659667,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4056700226055354e-05,
      "loss": 0.3116,
      "step": 9731
    },
    {
      "epoch": 0.7365403717894139,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4055611526799004e-05,
      "loss": 0.311,
      "step": 9732
    },
    {
      "epoch": 0.7366160541128612,
      "grad_norm": 0.796875,
      "learning_rate": 1.4054522770006847e-05,
      "loss": 0.337,
      "step": 9733
    },
    {
      "epoch": 0.7366917364363086,
      "grad_norm": 0.71484375,
      "learning_rate": 1.4053433955694327e-05,
      "loss": 0.2925,
      "step": 9734
    },
    {
      "epoch": 0.7367674187597559,
      "grad_norm": 0.76171875,
      "learning_rate": 1.4052345083876888e-05,
      "loss": 0.3288,
      "step": 9735
    },
    {
      "epoch": 0.7368431010832033,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4051256154569982e-05,
      "loss": 0.2964,
      "step": 9736
    },
    {
      "epoch": 0.7369187834066506,
      "grad_norm": 0.78125,
      "learning_rate": 1.4050167167789052e-05,
      "loss": 0.2989,
      "step": 9737
    },
    {
      "epoch": 0.7369944657300979,
      "grad_norm": 0.78125,
      "learning_rate": 1.4049078123549556e-05,
      "loss": 0.3248,
      "step": 9738
    },
    {
      "epoch": 0.7370701480535452,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4047989021866935e-05,
      "loss": 0.2807,
      "step": 9739
    },
    {
      "epoch": 0.7371458303769926,
      "grad_norm": 0.78125,
      "learning_rate": 1.4046899862756642e-05,
      "loss": 0.3304,
      "step": 9740
    },
    {
      "epoch": 0.7372215127004399,
      "grad_norm": 0.80859375,
      "learning_rate": 1.404581064623413e-05,
      "loss": 0.3625,
      "step": 9741
    },
    {
      "epoch": 0.7372971950238872,
      "grad_norm": 0.74609375,
      "learning_rate": 1.4044721372314854e-05,
      "loss": 0.3147,
      "step": 9742
    },
    {
      "epoch": 0.7373728773473346,
      "grad_norm": 0.7109375,
      "learning_rate": 1.4043632041014262e-05,
      "loss": 0.3075,
      "step": 9743
    },
    {
      "epoch": 0.7374485596707819,
      "grad_norm": 0.72265625,
      "learning_rate": 1.4042542652347811e-05,
      "loss": 0.2756,
      "step": 9744
    },
    {
      "epoch": 0.7375242419942292,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4041453206330955e-05,
      "loss": 0.3143,
      "step": 9745
    },
    {
      "epoch": 0.7375999243176765,
      "grad_norm": 0.90625,
      "learning_rate": 1.4040363702979147e-05,
      "loss": 0.3382,
      "step": 9746
    },
    {
      "epoch": 0.7376756066411239,
      "grad_norm": 0.79296875,
      "learning_rate": 1.403927414230785e-05,
      "loss": 0.3339,
      "step": 9747
    },
    {
      "epoch": 0.7377512889645712,
      "grad_norm": 0.73046875,
      "learning_rate": 1.4038184524332517e-05,
      "loss": 0.3025,
      "step": 9748
    },
    {
      "epoch": 0.7378269712880186,
      "grad_norm": 0.72265625,
      "learning_rate": 1.4037094849068606e-05,
      "loss": 0.3218,
      "step": 9749
    },
    {
      "epoch": 0.7379026536114659,
      "grad_norm": 0.78125,
      "learning_rate": 1.4036005116531579e-05,
      "loss": 0.3356,
      "step": 9750
    },
    {
      "epoch": 0.7379783359349132,
      "grad_norm": 0.765625,
      "learning_rate": 1.403491532673689e-05,
      "loss": 0.3228,
      "step": 9751
    },
    {
      "epoch": 0.7380540182583605,
      "grad_norm": 0.703125,
      "learning_rate": 1.4033825479700008e-05,
      "loss": 0.249,
      "step": 9752
    },
    {
      "epoch": 0.7381297005818078,
      "grad_norm": 0.75390625,
      "learning_rate": 1.4032735575436384e-05,
      "loss": 0.256,
      "step": 9753
    },
    {
      "epoch": 0.7382053829052552,
      "grad_norm": 0.765625,
      "learning_rate": 1.4031645613961487e-05,
      "loss": 0.3154,
      "step": 9754
    },
    {
      "epoch": 0.7382810652287025,
      "grad_norm": 0.82421875,
      "learning_rate": 1.403055559529078e-05,
      "loss": 0.3764,
      "step": 9755
    },
    {
      "epoch": 0.7383567475521499,
      "grad_norm": 0.73828125,
      "learning_rate": 1.4029465519439724e-05,
      "loss": 0.2895,
      "step": 9756
    },
    {
      "epoch": 0.7384324298755972,
      "grad_norm": 0.73046875,
      "learning_rate": 1.4028375386423785e-05,
      "loss": 0.2734,
      "step": 9757
    },
    {
      "epoch": 0.7385081121990446,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4027285196258426e-05,
      "loss": 0.2793,
      "step": 9758
    },
    {
      "epoch": 0.7385837945224918,
      "grad_norm": 0.77734375,
      "learning_rate": 1.402619494895912e-05,
      "loss": 0.3283,
      "step": 9759
    },
    {
      "epoch": 0.7386594768459391,
      "grad_norm": 0.8046875,
      "learning_rate": 1.4025104644541325e-05,
      "loss": 0.3407,
      "step": 9760
    },
    {
      "epoch": 0.7387351591693865,
      "grad_norm": 0.73046875,
      "learning_rate": 1.4024014283020514e-05,
      "loss": 0.3057,
      "step": 9761
    },
    {
      "epoch": 0.7388108414928338,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4022923864412161e-05,
      "loss": 0.3153,
      "step": 9762
    },
    {
      "epoch": 0.7388865238162812,
      "grad_norm": 0.7265625,
      "learning_rate": 1.4021833388731723e-05,
      "loss": 0.2902,
      "step": 9763
    },
    {
      "epoch": 0.7389622061397285,
      "grad_norm": 0.79296875,
      "learning_rate": 1.4020742855994677e-05,
      "loss": 0.3313,
      "step": 9764
    },
    {
      "epoch": 0.7390378884631759,
      "grad_norm": 0.72265625,
      "learning_rate": 1.4019652266216497e-05,
      "loss": 0.3058,
      "step": 9765
    },
    {
      "epoch": 0.7391135707866231,
      "grad_norm": 0.8125,
      "learning_rate": 1.401856161941265e-05,
      "loss": 0.3272,
      "step": 9766
    },
    {
      "epoch": 0.7391892531100704,
      "grad_norm": 0.71875,
      "learning_rate": 1.401747091559861e-05,
      "loss": 0.2673,
      "step": 9767
    },
    {
      "epoch": 0.7392649354335178,
      "grad_norm": 0.73046875,
      "learning_rate": 1.401638015478985e-05,
      "loss": 0.3094,
      "step": 9768
    },
    {
      "epoch": 0.7393406177569651,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4015289337001846e-05,
      "loss": 0.3153,
      "step": 9769
    },
    {
      "epoch": 0.7394163000804125,
      "grad_norm": 0.765625,
      "learning_rate": 1.4014198462250072e-05,
      "loss": 0.337,
      "step": 9770
    },
    {
      "epoch": 0.7394919824038598,
      "grad_norm": 0.765625,
      "learning_rate": 1.4013107530550007e-05,
      "loss": 0.318,
      "step": 9771
    },
    {
      "epoch": 0.7395676647273072,
      "grad_norm": 0.8359375,
      "learning_rate": 1.401201654191712e-05,
      "loss": 0.3643,
      "step": 9772
    },
    {
      "epoch": 0.7396433470507544,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4010925496366896e-05,
      "loss": 0.2939,
      "step": 9773
    },
    {
      "epoch": 0.7397190293742018,
      "grad_norm": 0.78515625,
      "learning_rate": 1.4009834393914812e-05,
      "loss": 0.3471,
      "step": 9774
    },
    {
      "epoch": 0.7397947116976491,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4008743234576343e-05,
      "loss": 0.3396,
      "step": 9775
    },
    {
      "epoch": 0.7398703940210964,
      "grad_norm": 0.7421875,
      "learning_rate": 1.4007652018366974e-05,
      "loss": 0.2875,
      "step": 9776
    },
    {
      "epoch": 0.7399460763445438,
      "grad_norm": 0.7578125,
      "learning_rate": 1.4006560745302188e-05,
      "loss": 0.3195,
      "step": 9777
    },
    {
      "epoch": 0.7400217586679911,
      "grad_norm": 0.7890625,
      "learning_rate": 1.4005469415397453e-05,
      "loss": 0.3294,
      "step": 9778
    },
    {
      "epoch": 0.7400974409914385,
      "grad_norm": 1.0859375,
      "learning_rate": 1.4004378028668266e-05,
      "loss": 0.3805,
      "step": 9779
    },
    {
      "epoch": 0.7401731233148857,
      "grad_norm": 0.75,
      "learning_rate": 1.4003286585130106e-05,
      "loss": 0.3139,
      "step": 9780
    },
    {
      "epoch": 0.7402488056383331,
      "grad_norm": 0.78125,
      "learning_rate": 1.4002195084798453e-05,
      "loss": 0.3448,
      "step": 9781
    },
    {
      "epoch": 0.7403244879617804,
      "grad_norm": 0.77734375,
      "learning_rate": 1.4001103527688797e-05,
      "loss": 0.3219,
      "step": 9782
    },
    {
      "epoch": 0.7404001702852278,
      "grad_norm": 0.703125,
      "learning_rate": 1.4000011913816623e-05,
      "loss": 0.2817,
      "step": 9783
    },
    {
      "epoch": 0.7404758526086751,
      "grad_norm": 0.85546875,
      "learning_rate": 1.3998920243197408e-05,
      "loss": 0.3565,
      "step": 9784
    },
    {
      "epoch": 0.7405515349321224,
      "grad_norm": 0.796875,
      "learning_rate": 1.3997828515846656e-05,
      "loss": 0.3408,
      "step": 9785
    },
    {
      "epoch": 0.7406272172555698,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3996736731779843e-05,
      "loss": 0.317,
      "step": 9786
    },
    {
      "epoch": 0.740702899579017,
      "grad_norm": 0.9140625,
      "learning_rate": 1.3995644891012459e-05,
      "loss": 0.3498,
      "step": 9787
    },
    {
      "epoch": 0.7407785819024644,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3994552993559996e-05,
      "loss": 0.3053,
      "step": 9788
    },
    {
      "epoch": 0.7408542642259117,
      "grad_norm": 1.015625,
      "learning_rate": 1.3993461039437946e-05,
      "loss": 0.3651,
      "step": 9789
    },
    {
      "epoch": 0.7409299465493591,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3992369028661796e-05,
      "loss": 0.33,
      "step": 9790
    },
    {
      "epoch": 0.7410056288728064,
      "grad_norm": 0.7890625,
      "learning_rate": 1.399127696124704e-05,
      "loss": 0.3688,
      "step": 9791
    },
    {
      "epoch": 0.7410813111962538,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3990184837209173e-05,
      "loss": 0.2895,
      "step": 9792
    },
    {
      "epoch": 0.7411569935197011,
      "grad_norm": 0.734375,
      "learning_rate": 1.3989092656563684e-05,
      "loss": 0.2794,
      "step": 9793
    },
    {
      "epoch": 0.7412326758431483,
      "grad_norm": 0.75,
      "learning_rate": 1.3988000419326073e-05,
      "loss": 0.2881,
      "step": 9794
    },
    {
      "epoch": 0.7413083581665957,
      "grad_norm": 0.82421875,
      "learning_rate": 1.398690812551183e-05,
      "loss": 0.3453,
      "step": 9795
    },
    {
      "epoch": 0.741384040490043,
      "grad_norm": 0.75,
      "learning_rate": 1.3985815775136455e-05,
      "loss": 0.2979,
      "step": 9796
    },
    {
      "epoch": 0.7414597228134904,
      "grad_norm": 0.7890625,
      "learning_rate": 1.3984723368215442e-05,
      "loss": 0.349,
      "step": 9797
    },
    {
      "epoch": 0.7415354051369377,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3983630904764292e-05,
      "loss": 0.2823,
      "step": 9798
    },
    {
      "epoch": 0.7416110874603851,
      "grad_norm": 0.81640625,
      "learning_rate": 1.39825383847985e-05,
      "loss": 0.3643,
      "step": 9799
    },
    {
      "epoch": 0.7416867697838324,
      "grad_norm": 0.69921875,
      "learning_rate": 1.3981445808333566e-05,
      "loss": 0.282,
      "step": 9800
    },
    {
      "epoch": 0.7417624521072796,
      "grad_norm": 0.765625,
      "learning_rate": 1.3980353175384993e-05,
      "loss": 0.3153,
      "step": 9801
    },
    {
      "epoch": 0.741838134430727,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3979260485968278e-05,
      "loss": 0.3148,
      "step": 9802
    },
    {
      "epoch": 0.7419138167541743,
      "grad_norm": 0.703125,
      "learning_rate": 1.3978167740098925e-05,
      "loss": 0.2798,
      "step": 9803
    },
    {
      "epoch": 0.7419894990776217,
      "grad_norm": 0.6953125,
      "learning_rate": 1.3977074937792434e-05,
      "loss": 0.3028,
      "step": 9804
    },
    {
      "epoch": 0.742065181401069,
      "grad_norm": 0.80078125,
      "learning_rate": 1.3975982079064314e-05,
      "loss": 0.3498,
      "step": 9805
    },
    {
      "epoch": 0.7421408637245164,
      "grad_norm": 0.80859375,
      "learning_rate": 1.3974889163930063e-05,
      "loss": 0.3603,
      "step": 9806
    },
    {
      "epoch": 0.7422165460479637,
      "grad_norm": 0.8203125,
      "learning_rate": 1.397379619240519e-05,
      "loss": 0.3621,
      "step": 9807
    },
    {
      "epoch": 0.742292228371411,
      "grad_norm": 0.6875,
      "learning_rate": 1.3972703164505195e-05,
      "loss": 0.2772,
      "step": 9808
    },
    {
      "epoch": 0.7423679106948583,
      "grad_norm": 0.671875,
      "learning_rate": 1.397161008024559e-05,
      "loss": 0.2724,
      "step": 9809
    },
    {
      "epoch": 0.7424435930183056,
      "grad_norm": 0.80859375,
      "learning_rate": 1.3970516939641878e-05,
      "loss": 0.3823,
      "step": 9810
    },
    {
      "epoch": 0.742519275341753,
      "grad_norm": 0.71875,
      "learning_rate": 1.3969423742709574e-05,
      "loss": 0.3004,
      "step": 9811
    },
    {
      "epoch": 0.7425949576652003,
      "grad_norm": 0.80859375,
      "learning_rate": 1.396833048946418e-05,
      "loss": 0.3614,
      "step": 9812
    },
    {
      "epoch": 0.7426706399886477,
      "grad_norm": 0.7890625,
      "learning_rate": 1.396723717992121e-05,
      "loss": 0.3768,
      "step": 9813
    },
    {
      "epoch": 0.742746322312095,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3966143814096172e-05,
      "loss": 0.3239,
      "step": 9814
    },
    {
      "epoch": 0.7428220046355423,
      "grad_norm": 0.9140625,
      "learning_rate": 1.3965050392004576e-05,
      "loss": 0.2983,
      "step": 9815
    },
    {
      "epoch": 0.7428976869589896,
      "grad_norm": 0.765625,
      "learning_rate": 1.3963956913661936e-05,
      "loss": 0.3226,
      "step": 9816
    },
    {
      "epoch": 0.742973369282437,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3962863379083766e-05,
      "loss": 0.338,
      "step": 9817
    },
    {
      "epoch": 0.7430490516058843,
      "grad_norm": 0.81640625,
      "learning_rate": 1.3961769788285579e-05,
      "loss": 0.3616,
      "step": 9818
    },
    {
      "epoch": 0.7431247339293316,
      "grad_norm": 0.80859375,
      "learning_rate": 1.3960676141282888e-05,
      "loss": 0.3475,
      "step": 9819
    },
    {
      "epoch": 0.743200416252779,
      "grad_norm": 0.8125,
      "learning_rate": 1.395958243809121e-05,
      "loss": 0.3262,
      "step": 9820
    },
    {
      "epoch": 0.7432760985762263,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3958488678726058e-05,
      "loss": 0.3052,
      "step": 9821
    },
    {
      "epoch": 0.7433517808996736,
      "grad_norm": 0.75,
      "learning_rate": 1.3957394863202955e-05,
      "loss": 0.2887,
      "step": 9822
    },
    {
      "epoch": 0.7434274632231209,
      "grad_norm": 0.8046875,
      "learning_rate": 1.3956300991537412e-05,
      "loss": 0.3227,
      "step": 9823
    },
    {
      "epoch": 0.7435031455465683,
      "grad_norm": 0.73828125,
      "learning_rate": 1.395520706374495e-05,
      "loss": 0.3209,
      "step": 9824
    },
    {
      "epoch": 0.7435788278700156,
      "grad_norm": 0.75390625,
      "learning_rate": 1.395411307984109e-05,
      "loss": 0.3193,
      "step": 9825
    },
    {
      "epoch": 0.743654510193463,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3953019039841349e-05,
      "loss": 0.2829,
      "step": 9826
    },
    {
      "epoch": 0.7437301925169103,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3951924943761252e-05,
      "loss": 0.3052,
      "step": 9827
    },
    {
      "epoch": 0.7438058748403577,
      "grad_norm": 0.77734375,
      "learning_rate": 1.3950830791616314e-05,
      "loss": 0.3477,
      "step": 9828
    },
    {
      "epoch": 0.7438815571638049,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3949736583422064e-05,
      "loss": 0.2879,
      "step": 9829
    },
    {
      "epoch": 0.7439572394872522,
      "grad_norm": 0.78515625,
      "learning_rate": 1.3948642319194021e-05,
      "loss": 0.3423,
      "step": 9830
    },
    {
      "epoch": 0.7440329218106996,
      "grad_norm": 0.78125,
      "learning_rate": 1.394754799894771e-05,
      "loss": 0.3111,
      "step": 9831
    },
    {
      "epoch": 0.7441086041341469,
      "grad_norm": 0.83203125,
      "learning_rate": 1.394645362269866e-05,
      "loss": 0.3482,
      "step": 9832
    },
    {
      "epoch": 0.7441842864575943,
      "grad_norm": 0.7734375,
      "learning_rate": 1.394535919046239e-05,
      "loss": 0.2913,
      "step": 9833
    },
    {
      "epoch": 0.7442599687810416,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3944264702254431e-05,
      "loss": 0.3364,
      "step": 9834
    },
    {
      "epoch": 0.744335651104489,
      "grad_norm": 0.72265625,
      "learning_rate": 1.3943170158090307e-05,
      "loss": 0.3073,
      "step": 9835
    },
    {
      "epoch": 0.7444113334279362,
      "grad_norm": 0.76953125,
      "learning_rate": 1.394207555798555e-05,
      "loss": 0.326,
      "step": 9836
    },
    {
      "epoch": 0.7444870157513835,
      "grad_norm": 0.72265625,
      "learning_rate": 1.3940980901955684e-05,
      "loss": 0.3096,
      "step": 9837
    },
    {
      "epoch": 0.7445626980748309,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3939886190016241e-05,
      "loss": 0.3218,
      "step": 9838
    },
    {
      "epoch": 0.7446383803982782,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3938791422182754e-05,
      "loss": 0.2889,
      "step": 9839
    },
    {
      "epoch": 0.7447140627217256,
      "grad_norm": 1.09375,
      "learning_rate": 1.3937696598470747e-05,
      "loss": 0.3807,
      "step": 9840
    },
    {
      "epoch": 0.7447897450451729,
      "grad_norm": 0.80078125,
      "learning_rate": 1.3936601718895762e-05,
      "loss": 0.3318,
      "step": 9841
    },
    {
      "epoch": 0.7448654273686203,
      "grad_norm": 1.0234375,
      "learning_rate": 1.3935506783473323e-05,
      "loss": 0.3683,
      "step": 9842
    },
    {
      "epoch": 0.7449411096920675,
      "grad_norm": 0.859375,
      "learning_rate": 1.3934411792218965e-05,
      "loss": 0.3248,
      "step": 9843
    },
    {
      "epoch": 0.7450167920155149,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3933316745148226e-05,
      "loss": 0.292,
      "step": 9844
    },
    {
      "epoch": 0.7450924743389622,
      "grad_norm": 0.7421875,
      "learning_rate": 1.393222164227664e-05,
      "loss": 0.2967,
      "step": 9845
    },
    {
      "epoch": 0.7451681566624095,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3931126483619742e-05,
      "loss": 0.3067,
      "step": 9846
    },
    {
      "epoch": 0.7452438389858569,
      "grad_norm": 0.85546875,
      "learning_rate": 1.3930031269193068e-05,
      "loss": 0.3549,
      "step": 9847
    },
    {
      "epoch": 0.7453195213093042,
      "grad_norm": 0.70703125,
      "learning_rate": 1.3928935999012156e-05,
      "loss": 0.2883,
      "step": 9848
    },
    {
      "epoch": 0.7453952036327516,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3927840673092543e-05,
      "loss": 0.2957,
      "step": 9849
    },
    {
      "epoch": 0.7454708859561988,
      "grad_norm": 0.8359375,
      "learning_rate": 1.3926745291449773e-05,
      "loss": 0.3718,
      "step": 9850
    },
    {
      "epoch": 0.7455465682796462,
      "grad_norm": 0.765625,
      "learning_rate": 1.3925649854099379e-05,
      "loss": 0.3438,
      "step": 9851
    },
    {
      "epoch": 0.7456222506030935,
      "grad_norm": 0.71875,
      "learning_rate": 1.3924554361056907e-05,
      "loss": 0.3084,
      "step": 9852
    },
    {
      "epoch": 0.7456979329265409,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3923458812337898e-05,
      "loss": 0.2976,
      "step": 9853
    },
    {
      "epoch": 0.7457736152499882,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3922363207957893e-05,
      "loss": 0.3352,
      "step": 9854
    },
    {
      "epoch": 0.7458492975734355,
      "grad_norm": 0.78515625,
      "learning_rate": 1.3921267547932433e-05,
      "loss": 0.3381,
      "step": 9855
    },
    {
      "epoch": 0.7459249798968829,
      "grad_norm": 0.765625,
      "learning_rate": 1.3920171832277066e-05,
      "loss": 0.3282,
      "step": 9856
    },
    {
      "epoch": 0.7460006622203301,
      "grad_norm": 0.80078125,
      "learning_rate": 1.391907606100733e-05,
      "loss": 0.3167,
      "step": 9857
    },
    {
      "epoch": 0.7460763445437775,
      "grad_norm": 0.8125,
      "learning_rate": 1.3917980234138777e-05,
      "loss": 0.3965,
      "step": 9858
    },
    {
      "epoch": 0.7461520268672248,
      "grad_norm": 0.76171875,
      "learning_rate": 1.391688435168695e-05,
      "loss": 0.323,
      "step": 9859
    },
    {
      "epoch": 0.7462277091906722,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3915788413667398e-05,
      "loss": 0.3098,
      "step": 9860
    },
    {
      "epoch": 0.7463033915141195,
      "grad_norm": 0.7734375,
      "learning_rate": 1.391469242009567e-05,
      "loss": 0.3199,
      "step": 9861
    },
    {
      "epoch": 0.7463790738375669,
      "grad_norm": 1.4296875,
      "learning_rate": 1.3913596370987308e-05,
      "loss": 0.4255,
      "step": 9862
    },
    {
      "epoch": 0.7464547561610141,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3912500266357868e-05,
      "loss": 0.2937,
      "step": 9863
    },
    {
      "epoch": 0.7465304384844614,
      "grad_norm": 0.78515625,
      "learning_rate": 1.3911404106222896e-05,
      "loss": 0.3185,
      "step": 9864
    },
    {
      "epoch": 0.7466061208079088,
      "grad_norm": 0.82421875,
      "learning_rate": 1.3910307890597948e-05,
      "loss": 0.3422,
      "step": 9865
    },
    {
      "epoch": 0.7466818031313561,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3909211619498568e-05,
      "loss": 0.3209,
      "step": 9866
    },
    {
      "epoch": 0.7467574854548035,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3908115292940315e-05,
      "loss": 0.3203,
      "step": 9867
    },
    {
      "epoch": 0.7468331677782508,
      "grad_norm": 0.79296875,
      "learning_rate": 1.3907018910938739e-05,
      "loss": 0.3319,
      "step": 9868
    },
    {
      "epoch": 0.7469088501016982,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3905922473509397e-05,
      "loss": 0.3151,
      "step": 9869
    },
    {
      "epoch": 0.7469845324251454,
      "grad_norm": 0.7265625,
      "learning_rate": 1.390482598066784e-05,
      "loss": 0.2849,
      "step": 9870
    },
    {
      "epoch": 0.7470602147485927,
      "grad_norm": 0.765625,
      "learning_rate": 1.3903729432429627e-05,
      "loss": 0.327,
      "step": 9871
    },
    {
      "epoch": 0.7471358970720401,
      "grad_norm": 0.765625,
      "learning_rate": 1.390263282881031e-05,
      "loss": 0.281,
      "step": 9872
    },
    {
      "epoch": 0.7472115793954874,
      "grad_norm": 0.80078125,
      "learning_rate": 1.3901536169825451e-05,
      "loss": 0.3457,
      "step": 9873
    },
    {
      "epoch": 0.7472872617189348,
      "grad_norm": 0.70703125,
      "learning_rate": 1.3900439455490609e-05,
      "loss": 0.2945,
      "step": 9874
    },
    {
      "epoch": 0.7473629440423821,
      "grad_norm": 0.7109375,
      "learning_rate": 1.3899342685821335e-05,
      "loss": 0.288,
      "step": 9875
    },
    {
      "epoch": 0.7474386263658295,
      "grad_norm": 0.765625,
      "learning_rate": 1.3898245860833198e-05,
      "loss": 0.3454,
      "step": 9876
    },
    {
      "epoch": 0.7475143086892767,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3897148980541751e-05,
      "loss": 0.3405,
      "step": 9877
    },
    {
      "epoch": 0.747589991012724,
      "grad_norm": 0.75,
      "learning_rate": 1.3896052044962558e-05,
      "loss": 0.322,
      "step": 9878
    },
    {
      "epoch": 0.7476656733361714,
      "grad_norm": 0.72265625,
      "learning_rate": 1.389495505411118e-05,
      "loss": 0.2908,
      "step": 9879
    },
    {
      "epoch": 0.7477413556596187,
      "grad_norm": 0.6953125,
      "learning_rate": 1.3893858008003185e-05,
      "loss": 0.2769,
      "step": 9880
    },
    {
      "epoch": 0.7478170379830661,
      "grad_norm": 0.79296875,
      "learning_rate": 1.3892760906654125e-05,
      "loss": 0.3269,
      "step": 9881
    },
    {
      "epoch": 0.7478927203065134,
      "grad_norm": 0.80078125,
      "learning_rate": 1.3891663750079577e-05,
      "loss": 0.3025,
      "step": 9882
    },
    {
      "epoch": 0.7479684026299608,
      "grad_norm": 0.71484375,
      "learning_rate": 1.3890566538295102e-05,
      "loss": 0.3009,
      "step": 9883
    },
    {
      "epoch": 0.748044084953408,
      "grad_norm": 0.765625,
      "learning_rate": 1.3889469271316259e-05,
      "loss": 0.3185,
      "step": 9884
    },
    {
      "epoch": 0.7481197672768554,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3888371949158624e-05,
      "loss": 0.3157,
      "step": 9885
    },
    {
      "epoch": 0.7481954496003027,
      "grad_norm": 0.77734375,
      "learning_rate": 1.3887274571837757e-05,
      "loss": 0.319,
      "step": 9886
    },
    {
      "epoch": 0.74827113192375,
      "grad_norm": 0.734375,
      "learning_rate": 1.3886177139369229e-05,
      "loss": 0.3057,
      "step": 9887
    },
    {
      "epoch": 0.7483468142471974,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3885079651768613e-05,
      "loss": 0.2685,
      "step": 9888
    },
    {
      "epoch": 0.7484224965706447,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3883982109051472e-05,
      "loss": 0.2939,
      "step": 9889
    },
    {
      "epoch": 0.7484981788940921,
      "grad_norm": 0.71484375,
      "learning_rate": 1.3882884511233381e-05,
      "loss": 0.2896,
      "step": 9890
    },
    {
      "epoch": 0.7485738612175393,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3881786858329909e-05,
      "loss": 0.2951,
      "step": 9891
    },
    {
      "epoch": 0.7486495435409867,
      "grad_norm": 0.72265625,
      "learning_rate": 1.3880689150356633e-05,
      "loss": 0.3136,
      "step": 9892
    },
    {
      "epoch": 0.748725225864434,
      "grad_norm": 0.796875,
      "learning_rate": 1.3879591387329114e-05,
      "loss": 0.3602,
      "step": 9893
    },
    {
      "epoch": 0.7488009081878814,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3878493569262941e-05,
      "loss": 0.3332,
      "step": 9894
    },
    {
      "epoch": 0.7488765905113287,
      "grad_norm": 0.7890625,
      "learning_rate": 1.3877395696173677e-05,
      "loss": 0.3179,
      "step": 9895
    },
    {
      "epoch": 0.7489522728347761,
      "grad_norm": 0.734375,
      "learning_rate": 1.3876297768076901e-05,
      "loss": 0.3302,
      "step": 9896
    },
    {
      "epoch": 0.7490279551582234,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3875199784988192e-05,
      "loss": 0.3354,
      "step": 9897
    },
    {
      "epoch": 0.7491036374816706,
      "grad_norm": 0.82421875,
      "learning_rate": 1.3874101746923124e-05,
      "loss": 0.3748,
      "step": 9898
    },
    {
      "epoch": 0.749179319805118,
      "grad_norm": 0.77734375,
      "learning_rate": 1.3873003653897271e-05,
      "loss": 0.3334,
      "step": 9899
    },
    {
      "epoch": 0.7492550021285653,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3871905505926217e-05,
      "loss": 0.2979,
      "step": 9900
    },
    {
      "epoch": 0.7493306844520127,
      "grad_norm": 0.68359375,
      "learning_rate": 1.3870807303025539e-05,
      "loss": 0.2593,
      "step": 9901
    },
    {
      "epoch": 0.74940636677546,
      "grad_norm": 0.7890625,
      "learning_rate": 1.3869709045210817e-05,
      "loss": 0.3461,
      "step": 9902
    },
    {
      "epoch": 0.7494820490989074,
      "grad_norm": 0.71875,
      "learning_rate": 1.3868610732497632e-05,
      "loss": 0.2668,
      "step": 9903
    },
    {
      "epoch": 0.7495577314223547,
      "grad_norm": 0.828125,
      "learning_rate": 1.3867512364901563e-05,
      "loss": 0.3403,
      "step": 9904
    },
    {
      "epoch": 0.749633413745802,
      "grad_norm": 0.796875,
      "learning_rate": 1.3866413942438195e-05,
      "loss": 0.3199,
      "step": 9905
    },
    {
      "epoch": 0.7497090960692493,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3865315465123112e-05,
      "loss": 0.3158,
      "step": 9906
    },
    {
      "epoch": 0.7497847783926966,
      "grad_norm": 0.82421875,
      "learning_rate": 1.3864216932971894e-05,
      "loss": 0.3071,
      "step": 9907
    },
    {
      "epoch": 0.749860460716144,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3863118346000128e-05,
      "loss": 0.3086,
      "step": 9908
    },
    {
      "epoch": 0.7499361430395913,
      "grad_norm": 0.8359375,
      "learning_rate": 1.3862019704223401e-05,
      "loss": 0.3803,
      "step": 9909
    },
    {
      "epoch": 0.7500118253630387,
      "grad_norm": 0.77734375,
      "learning_rate": 1.3860921007657295e-05,
      "loss": 0.356,
      "step": 9910
    },
    {
      "epoch": 0.750087507686486,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3859822256317399e-05,
      "loss": 0.3118,
      "step": 9911
    },
    {
      "epoch": 0.7501631900099333,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3858723450219302e-05,
      "loss": 0.3436,
      "step": 9912
    },
    {
      "epoch": 0.7502388723333806,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3857624589378591e-05,
      "loss": 0.3059,
      "step": 9913
    },
    {
      "epoch": 0.750314554656828,
      "grad_norm": 0.71484375,
      "learning_rate": 1.3856525673810852e-05,
      "loss": 0.2882,
      "step": 9914
    },
    {
      "epoch": 0.7503902369802753,
      "grad_norm": 0.80078125,
      "learning_rate": 1.385542670353168e-05,
      "loss": 0.3616,
      "step": 9915
    },
    {
      "epoch": 0.7504659193037226,
      "grad_norm": 0.859375,
      "learning_rate": 1.3854327678556665e-05,
      "loss": 0.3729,
      "step": 9916
    },
    {
      "epoch": 0.75054160162717,
      "grad_norm": 0.79296875,
      "learning_rate": 1.3853228598901398e-05,
      "loss": 0.3166,
      "step": 9917
    },
    {
      "epoch": 0.7506172839506173,
      "grad_norm": 0.8046875,
      "learning_rate": 1.385212946458147e-05,
      "loss": 0.2882,
      "step": 9918
    },
    {
      "epoch": 0.7506929662740646,
      "grad_norm": 0.78125,
      "learning_rate": 1.3851030275612477e-05,
      "loss": 0.3203,
      "step": 9919
    },
    {
      "epoch": 0.7507686485975119,
      "grad_norm": 0.8046875,
      "learning_rate": 1.384993103201001e-05,
      "loss": 0.365,
      "step": 9920
    },
    {
      "epoch": 0.7508443309209593,
      "grad_norm": 0.78125,
      "learning_rate": 1.3848831733789664e-05,
      "loss": 0.3232,
      "step": 9921
    },
    {
      "epoch": 0.7509200132444066,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3847732380967031e-05,
      "loss": 0.3095,
      "step": 9922
    },
    {
      "epoch": 0.750995695567854,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3846632973557714e-05,
      "loss": 0.3183,
      "step": 9923
    },
    {
      "epoch": 0.7510713778913013,
      "grad_norm": 0.71875,
      "learning_rate": 1.3845533511577309e-05,
      "loss": 0.2984,
      "step": 9924
    },
    {
      "epoch": 0.7511470602147486,
      "grad_norm": 0.734375,
      "learning_rate": 1.3844433995041414e-05,
      "loss": 0.3262,
      "step": 9925
    },
    {
      "epoch": 0.7512227425381959,
      "grad_norm": 0.734375,
      "learning_rate": 1.3843334423965623e-05,
      "loss": 0.2927,
      "step": 9926
    },
    {
      "epoch": 0.7512984248616432,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3842234798365537e-05,
      "loss": 0.3079,
      "step": 9927
    },
    {
      "epoch": 0.7513741071850906,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3841135118256754e-05,
      "loss": 0.3375,
      "step": 9928
    },
    {
      "epoch": 0.7514497895085379,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3840035383654882e-05,
      "loss": 0.3275,
      "step": 9929
    },
    {
      "epoch": 0.7515254718319853,
      "grad_norm": 0.76953125,
      "learning_rate": 1.383893559457552e-05,
      "loss": 0.3264,
      "step": 9930
    },
    {
      "epoch": 0.7516011541554326,
      "grad_norm": 0.828125,
      "learning_rate": 1.3837835751034266e-05,
      "loss": 0.341,
      "step": 9931
    },
    {
      "epoch": 0.75167683647888,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3836735853046725e-05,
      "loss": 0.302,
      "step": 9932
    },
    {
      "epoch": 0.7517525188023272,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3835635900628504e-05,
      "loss": 0.2848,
      "step": 9933
    },
    {
      "epoch": 0.7518282011257745,
      "grad_norm": 0.765625,
      "learning_rate": 1.3834535893795204e-05,
      "loss": 0.3231,
      "step": 9934
    },
    {
      "epoch": 0.7519038834492219,
      "grad_norm": 0.77734375,
      "learning_rate": 1.3833435832562434e-05,
      "loss": 0.3462,
      "step": 9935
    },
    {
      "epoch": 0.7519795657726692,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3832335716945797e-05,
      "loss": 0.3109,
      "step": 9936
    },
    {
      "epoch": 0.7520552480961166,
      "grad_norm": 0.81640625,
      "learning_rate": 1.38312355469609e-05,
      "loss": 0.3325,
      "step": 9937
    },
    {
      "epoch": 0.7521309304195639,
      "grad_norm": 0.8515625,
      "learning_rate": 1.3830135322623353e-05,
      "loss": 0.3003,
      "step": 9938
    },
    {
      "epoch": 0.7522066127430113,
      "grad_norm": 0.70703125,
      "learning_rate": 1.3829035043948767e-05,
      "loss": 0.2869,
      "step": 9939
    },
    {
      "epoch": 0.7522822950664585,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3827934710952744e-05,
      "loss": 0.3333,
      "step": 9940
    },
    {
      "epoch": 0.7523579773899058,
      "grad_norm": 0.79296875,
      "learning_rate": 1.3826834323650899e-05,
      "loss": 0.3302,
      "step": 9941
    },
    {
      "epoch": 0.7524336597133532,
      "grad_norm": 0.79296875,
      "learning_rate": 1.3825733882058843e-05,
      "loss": 0.3627,
      "step": 9942
    },
    {
      "epoch": 0.7525093420368005,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3824633386192185e-05,
      "loss": 0.3263,
      "step": 9943
    },
    {
      "epoch": 0.7525850243602479,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3823532836066542e-05,
      "loss": 0.298,
      "step": 9944
    },
    {
      "epoch": 0.7526607066836952,
      "grad_norm": 0.89453125,
      "learning_rate": 1.3822432231697522e-05,
      "loss": 0.3182,
      "step": 9945
    },
    {
      "epoch": 0.7527363890071426,
      "grad_norm": 0.75390625,
      "learning_rate": 1.382133157310074e-05,
      "loss": 0.3038,
      "step": 9946
    },
    {
      "epoch": 0.7528120713305898,
      "grad_norm": 0.75,
      "learning_rate": 1.3820230860291814e-05,
      "loss": 0.2869,
      "step": 9947
    },
    {
      "epoch": 0.7528877536540372,
      "grad_norm": 0.80078125,
      "learning_rate": 1.3819130093286359e-05,
      "loss": 0.362,
      "step": 9948
    },
    {
      "epoch": 0.7529634359774845,
      "grad_norm": 0.8125,
      "learning_rate": 1.3818029272099988e-05,
      "loss": 0.3272,
      "step": 9949
    },
    {
      "epoch": 0.7530391183009318,
      "grad_norm": 0.75,
      "learning_rate": 1.3816928396748321e-05,
      "loss": 0.3039,
      "step": 9950
    },
    {
      "epoch": 0.7531148006243792,
      "grad_norm": 0.671875,
      "learning_rate": 1.3815827467246973e-05,
      "loss": 0.2564,
      "step": 9951
    },
    {
      "epoch": 0.7531904829478265,
      "grad_norm": 1.0078125,
      "learning_rate": 1.3814726483611565e-05,
      "loss": 0.3847,
      "step": 9952
    },
    {
      "epoch": 0.7532661652712739,
      "grad_norm": 0.765625,
      "learning_rate": 1.3813625445857717e-05,
      "loss": 0.3212,
      "step": 9953
    },
    {
      "epoch": 0.7533418475947211,
      "grad_norm": 0.75,
      "learning_rate": 1.3812524354001045e-05,
      "loss": 0.3383,
      "step": 9954
    },
    {
      "epoch": 0.7534175299181685,
      "grad_norm": 0.8046875,
      "learning_rate": 1.3811423208057173e-05,
      "loss": 0.3302,
      "step": 9955
    },
    {
      "epoch": 0.7534932122416158,
      "grad_norm": 0.78515625,
      "learning_rate": 1.3810322008041723e-05,
      "loss": 0.3286,
      "step": 9956
    },
    {
      "epoch": 0.7535688945650632,
      "grad_norm": 0.765625,
      "learning_rate": 1.3809220753970316e-05,
      "loss": 0.3403,
      "step": 9957
    },
    {
      "epoch": 0.7536445768885105,
      "grad_norm": 0.78515625,
      "learning_rate": 1.3808119445858576e-05,
      "loss": 0.3276,
      "step": 9958
    },
    {
      "epoch": 0.7537202592119578,
      "grad_norm": 0.71484375,
      "learning_rate": 1.3807018083722125e-05,
      "loss": 0.2986,
      "step": 9959
    },
    {
      "epoch": 0.7537959415354052,
      "grad_norm": 0.78125,
      "learning_rate": 1.3805916667576593e-05,
      "loss": 0.3024,
      "step": 9960
    },
    {
      "epoch": 0.7538716238588524,
      "grad_norm": 0.765625,
      "learning_rate": 1.3804815197437598e-05,
      "loss": 0.3055,
      "step": 9961
    },
    {
      "epoch": 0.7539473061822998,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3803713673320773e-05,
      "loss": 0.3052,
      "step": 9962
    },
    {
      "epoch": 0.7540229885057471,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3802612095241742e-05,
      "loss": 0.3123,
      "step": 9963
    },
    {
      "epoch": 0.7540986708291945,
      "grad_norm": 0.78125,
      "learning_rate": 1.3801510463216133e-05,
      "loss": 0.3202,
      "step": 9964
    },
    {
      "epoch": 0.7541743531526418,
      "grad_norm": 0.78125,
      "learning_rate": 1.3800408777259573e-05,
      "loss": 0.3217,
      "step": 9965
    },
    {
      "epoch": 0.7542500354760892,
      "grad_norm": 0.83203125,
      "learning_rate": 1.3799307037387695e-05,
      "loss": 0.3217,
      "step": 9966
    },
    {
      "epoch": 0.7543257177995365,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3798205243616125e-05,
      "loss": 0.3263,
      "step": 9967
    },
    {
      "epoch": 0.7544014001229837,
      "grad_norm": 0.69921875,
      "learning_rate": 1.3797103395960493e-05,
      "loss": 0.2773,
      "step": 9968
    },
    {
      "epoch": 0.7544770824464311,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3796001494436437e-05,
      "loss": 0.3376,
      "step": 9969
    },
    {
      "epoch": 0.7545527647698784,
      "grad_norm": 0.85546875,
      "learning_rate": 1.3794899539059585e-05,
      "loss": 0.374,
      "step": 9970
    },
    {
      "epoch": 0.7546284470933258,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3793797529845572e-05,
      "loss": 0.2838,
      "step": 9971
    },
    {
      "epoch": 0.7547041294167731,
      "grad_norm": 0.71484375,
      "learning_rate": 1.3792695466810029e-05,
      "loss": 0.2869,
      "step": 9972
    },
    {
      "epoch": 0.7547798117402205,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3791593349968592e-05,
      "loss": 0.317,
      "step": 9973
    },
    {
      "epoch": 0.7548554940636678,
      "grad_norm": 0.75,
      "learning_rate": 1.3790491179336899e-05,
      "loss": 0.3034,
      "step": 9974
    },
    {
      "epoch": 0.754931176387115,
      "grad_norm": 0.7890625,
      "learning_rate": 1.3789388954930583e-05,
      "loss": 0.3128,
      "step": 9975
    },
    {
      "epoch": 0.7550068587105624,
      "grad_norm": 0.75,
      "learning_rate": 1.378828667676528e-05,
      "loss": 0.3101,
      "step": 9976
    },
    {
      "epoch": 0.7550825410340097,
      "grad_norm": 0.72265625,
      "learning_rate": 1.378718434485663e-05,
      "loss": 0.3205,
      "step": 9977
    },
    {
      "epoch": 0.7551582233574571,
      "grad_norm": 0.74609375,
      "learning_rate": 1.378608195922027e-05,
      "loss": 0.3245,
      "step": 9978
    },
    {
      "epoch": 0.7552339056809044,
      "grad_norm": 0.8046875,
      "learning_rate": 1.3784979519871842e-05,
      "loss": 0.3626,
      "step": 9979
    },
    {
      "epoch": 0.7553095880043518,
      "grad_norm": 0.796875,
      "learning_rate": 1.3783877026826986e-05,
      "loss": 0.3439,
      "step": 9980
    },
    {
      "epoch": 0.7553852703277991,
      "grad_norm": 0.80078125,
      "learning_rate": 1.378277448010134e-05,
      "loss": 0.318,
      "step": 9981
    },
    {
      "epoch": 0.7554609526512464,
      "grad_norm": 0.765625,
      "learning_rate": 1.3781671879710546e-05,
      "loss": 0.2841,
      "step": 9982
    },
    {
      "epoch": 0.7555366349746937,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3780569225670246e-05,
      "loss": 0.3163,
      "step": 9983
    },
    {
      "epoch": 0.755612317298141,
      "grad_norm": 0.87109375,
      "learning_rate": 1.3779466517996083e-05,
      "loss": 0.3598,
      "step": 9984
    },
    {
      "epoch": 0.7556879996215884,
      "grad_norm": 0.703125,
      "learning_rate": 1.3778363756703703e-05,
      "loss": 0.3001,
      "step": 9985
    },
    {
      "epoch": 0.7557636819450357,
      "grad_norm": 0.703125,
      "learning_rate": 1.3777260941808753e-05,
      "loss": 0.2673,
      "step": 9986
    },
    {
      "epoch": 0.7558393642684831,
      "grad_norm": 0.72265625,
      "learning_rate": 1.3776158073326869e-05,
      "loss": 0.3047,
      "step": 9987
    },
    {
      "epoch": 0.7559150465919303,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3775055151273704e-05,
      "loss": 0.3014,
      "step": 9988
    },
    {
      "epoch": 0.7559907289153777,
      "grad_norm": 0.80078125,
      "learning_rate": 1.3773952175664904e-05,
      "loss": 0.3485,
      "step": 9989
    },
    {
      "epoch": 0.756066411238825,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3772849146516114e-05,
      "loss": 0.2936,
      "step": 9990
    },
    {
      "epoch": 0.7561420935622724,
      "grad_norm": 0.73046875,
      "learning_rate": 1.377174606384299e-05,
      "loss": 0.3126,
      "step": 9991
    },
    {
      "epoch": 0.7562177758857197,
      "grad_norm": 0.75,
      "learning_rate": 1.3770642927661171e-05,
      "loss": 0.3345,
      "step": 9992
    },
    {
      "epoch": 0.756293458209167,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3769539737986311e-05,
      "loss": 0.2995,
      "step": 9993
    },
    {
      "epoch": 0.7563691405326144,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3768436494834063e-05,
      "loss": 0.3302,
      "step": 9994
    },
    {
      "epoch": 0.7564448228560616,
      "grad_norm": 0.671875,
      "learning_rate": 1.3767333198220078e-05,
      "loss": 0.2863,
      "step": 9995
    },
    {
      "epoch": 0.756520505179509,
      "grad_norm": 0.8203125,
      "learning_rate": 1.3766229848160002e-05,
      "loss": 0.3795,
      "step": 9996
    },
    {
      "epoch": 0.7565961875029563,
      "grad_norm": 0.8515625,
      "learning_rate": 1.3765126444669496e-05,
      "loss": 0.37,
      "step": 9997
    },
    {
      "epoch": 0.7566718698264037,
      "grad_norm": 0.80078125,
      "learning_rate": 1.376402298776421e-05,
      "loss": 0.3615,
      "step": 9998
    },
    {
      "epoch": 0.756747552149851,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3762919477459796e-05,
      "loss": 0.312,
      "step": 9999
    },
    {
      "epoch": 0.7568232344732984,
      "grad_norm": 0.78515625,
      "learning_rate": 1.3761815913771917e-05,
      "loss": 0.3526,
      "step": 10000
    },
    {
      "epoch": 0.7568989167967457,
      "grad_norm": 0.7109375,
      "learning_rate": 1.3760712296716219e-05,
      "loss": 0.2906,
      "step": 10001
    },
    {
      "epoch": 0.7569745991201929,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3759608626308363e-05,
      "loss": 0.2873,
      "step": 10002
    },
    {
      "epoch": 0.7570502814436403,
      "grad_norm": 0.69921875,
      "learning_rate": 1.3758504902564009e-05,
      "loss": 0.2774,
      "step": 10003
    },
    {
      "epoch": 0.7571259637670876,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3757401125498815e-05,
      "loss": 0.3024,
      "step": 10004
    },
    {
      "epoch": 0.757201646090535,
      "grad_norm": 0.85546875,
      "learning_rate": 1.3756297295128436e-05,
      "loss": 0.3276,
      "step": 10005
    },
    {
      "epoch": 0.7572773284139823,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3755193411468532e-05,
      "loss": 0.3045,
      "step": 10006
    },
    {
      "epoch": 0.7573530107374297,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3754089474534768e-05,
      "loss": 0.2975,
      "step": 10007
    },
    {
      "epoch": 0.757428693060877,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3752985484342802e-05,
      "loss": 0.2864,
      "step": 10008
    },
    {
      "epoch": 0.7575043753843242,
      "grad_norm": 0.70703125,
      "learning_rate": 1.3751881440908299e-05,
      "loss": 0.2797,
      "step": 10009
    },
    {
      "epoch": 0.7575800577077716,
      "grad_norm": 0.77734375,
      "learning_rate": 1.3750777344246917e-05,
      "loss": 0.3509,
      "step": 10010
    },
    {
      "epoch": 0.7576557400312189,
      "grad_norm": 0.921875,
      "learning_rate": 1.374967319437432e-05,
      "loss": 0.2722,
      "step": 10011
    },
    {
      "epoch": 0.7577314223546663,
      "grad_norm": 0.8203125,
      "learning_rate": 1.3748568991306175e-05,
      "loss": 0.3244,
      "step": 10012
    },
    {
      "epoch": 0.7578071046781136,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3747464735058148e-05,
      "loss": 0.3181,
      "step": 10013
    },
    {
      "epoch": 0.757882787001561,
      "grad_norm": 0.69921875,
      "learning_rate": 1.3746360425645901e-05,
      "loss": 0.2661,
      "step": 10014
    },
    {
      "epoch": 0.7579584693250083,
      "grad_norm": 0.71875,
      "learning_rate": 1.3745256063085105e-05,
      "loss": 0.3092,
      "step": 10015
    },
    {
      "epoch": 0.7580341516484556,
      "grad_norm": 0.7578125,
      "learning_rate": 1.374415164739142e-05,
      "loss": 0.3326,
      "step": 10016
    },
    {
      "epoch": 0.7581098339719029,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3743047178580523e-05,
      "loss": 0.3072,
      "step": 10017
    },
    {
      "epoch": 0.7581855162953502,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3741942656668075e-05,
      "loss": 0.3097,
      "step": 10018
    },
    {
      "epoch": 0.7582611986187976,
      "grad_norm": 0.75,
      "learning_rate": 1.3740838081669753e-05,
      "loss": 0.3092,
      "step": 10019
    },
    {
      "epoch": 0.7583368809422449,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3739733453601219e-05,
      "loss": 0.3057,
      "step": 10020
    },
    {
      "epoch": 0.7584125632656923,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3738628772478152e-05,
      "loss": 0.3358,
      "step": 10021
    },
    {
      "epoch": 0.7584882455891396,
      "grad_norm": 0.78515625,
      "learning_rate": 1.3737524038316219e-05,
      "loss": 0.3313,
      "step": 10022
    },
    {
      "epoch": 0.7585639279125869,
      "grad_norm": 0.7890625,
      "learning_rate": 1.3736419251131094e-05,
      "loss": 0.3492,
      "step": 10023
    },
    {
      "epoch": 0.7586396102360342,
      "grad_norm": 0.8984375,
      "learning_rate": 1.373531441093845e-05,
      "loss": 0.3779,
      "step": 10024
    },
    {
      "epoch": 0.7587152925594816,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3734209517753959e-05,
      "loss": 0.294,
      "step": 10025
    },
    {
      "epoch": 0.7587909748829289,
      "grad_norm": 0.71484375,
      "learning_rate": 1.3733104571593298e-05,
      "loss": 0.2895,
      "step": 10026
    },
    {
      "epoch": 0.7588666572063762,
      "grad_norm": 0.8125,
      "learning_rate": 1.3731999572472146e-05,
      "loss": 0.3402,
      "step": 10027
    },
    {
      "epoch": 0.7589423395298236,
      "grad_norm": 0.8125,
      "learning_rate": 1.3730894520406173e-05,
      "loss": 0.3322,
      "step": 10028
    },
    {
      "epoch": 0.7590180218532709,
      "grad_norm": 0.68359375,
      "learning_rate": 1.372978941541106e-05,
      "loss": 0.2463,
      "step": 10029
    },
    {
      "epoch": 0.7590937041767182,
      "grad_norm": 0.7421875,
      "learning_rate": 1.372868425750248e-05,
      "loss": 0.3332,
      "step": 10030
    },
    {
      "epoch": 0.7591693865001655,
      "grad_norm": 0.76171875,
      "learning_rate": 1.372757904669612e-05,
      "loss": 0.3121,
      "step": 10031
    },
    {
      "epoch": 0.7592450688236129,
      "grad_norm": 0.74609375,
      "learning_rate": 1.372647378300765e-05,
      "loss": 0.2842,
      "step": 10032
    },
    {
      "epoch": 0.7593207511470602,
      "grad_norm": 0.75,
      "learning_rate": 1.372536846645276e-05,
      "loss": 0.3283,
      "step": 10033
    },
    {
      "epoch": 0.7593964334705076,
      "grad_norm": 0.68359375,
      "learning_rate": 1.3724263097047122e-05,
      "loss": 0.2858,
      "step": 10034
    },
    {
      "epoch": 0.7594721157939549,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3723157674806419e-05,
      "loss": 0.3026,
      "step": 10035
    },
    {
      "epoch": 0.7595477981174023,
      "grad_norm": 0.8046875,
      "learning_rate": 1.3722052199746338e-05,
      "loss": 0.3338,
      "step": 10036
    },
    {
      "epoch": 0.7596234804408495,
      "grad_norm": 0.78515625,
      "learning_rate": 1.372094667188256e-05,
      "loss": 0.3287,
      "step": 10037
    },
    {
      "epoch": 0.7596991627642968,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3719841091230769e-05,
      "loss": 0.3049,
      "step": 10038
    },
    {
      "epoch": 0.7597748450877442,
      "grad_norm": 0.67578125,
      "learning_rate": 1.3718735457806646e-05,
      "loss": 0.2652,
      "step": 10039
    },
    {
      "epoch": 0.7598505274111915,
      "grad_norm": 0.81640625,
      "learning_rate": 1.3717629771625878e-05,
      "loss": 0.347,
      "step": 10040
    },
    {
      "epoch": 0.7599262097346389,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3716524032704153e-05,
      "loss": 0.2965,
      "step": 10041
    },
    {
      "epoch": 0.7600018920580862,
      "grad_norm": 0.796875,
      "learning_rate": 1.3715418241057159e-05,
      "loss": 0.3476,
      "step": 10042
    },
    {
      "epoch": 0.7600775743815336,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3714312396700579e-05,
      "loss": 0.331,
      "step": 10043
    },
    {
      "epoch": 0.7601532567049808,
      "grad_norm": 0.8046875,
      "learning_rate": 1.3713206499650105e-05,
      "loss": 0.309,
      "step": 10044
    },
    {
      "epoch": 0.7602289390284281,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3712100549921423e-05,
      "loss": 0.3146,
      "step": 10045
    },
    {
      "epoch": 0.7603046213518755,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3710994547530225e-05,
      "loss": 0.3462,
      "step": 10046
    },
    {
      "epoch": 0.7603803036753228,
      "grad_norm": 0.78515625,
      "learning_rate": 1.3709888492492204e-05,
      "loss": 0.341,
      "step": 10047
    },
    {
      "epoch": 0.7604559859987702,
      "grad_norm": 0.734375,
      "learning_rate": 1.3708782384823045e-05,
      "loss": 0.3116,
      "step": 10048
    },
    {
      "epoch": 0.7605316683222175,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3707676224538445e-05,
      "loss": 0.3218,
      "step": 10049
    },
    {
      "epoch": 0.7606073506456649,
      "grad_norm": 0.71875,
      "learning_rate": 1.3706570011654094e-05,
      "loss": 0.2948,
      "step": 10050
    },
    {
      "epoch": 0.7606830329691121,
      "grad_norm": 0.6953125,
      "learning_rate": 1.370546374618569e-05,
      "loss": 0.2933,
      "step": 10051
    },
    {
      "epoch": 0.7607587152925595,
      "grad_norm": 0.72265625,
      "learning_rate": 1.370435742814892e-05,
      "loss": 0.2811,
      "step": 10052
    },
    {
      "epoch": 0.7608343976160068,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3703251057559484e-05,
      "loss": 0.3482,
      "step": 10053
    },
    {
      "epoch": 0.7609100799394541,
      "grad_norm": 0.734375,
      "learning_rate": 1.3702144634433077e-05,
      "loss": 0.3045,
      "step": 10054
    },
    {
      "epoch": 0.7609857622629015,
      "grad_norm": 0.85546875,
      "learning_rate": 1.3701038158785393e-05,
      "loss": 0.3503,
      "step": 10055
    },
    {
      "epoch": 0.7610614445863488,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3699931630632137e-05,
      "loss": 0.3534,
      "step": 10056
    },
    {
      "epoch": 0.7611371269097962,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3698825049988998e-05,
      "loss": 0.323,
      "step": 10057
    },
    {
      "epoch": 0.7612128092332434,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3697718416871677e-05,
      "loss": 0.328,
      "step": 10058
    },
    {
      "epoch": 0.7612884915566908,
      "grad_norm": 0.75,
      "learning_rate": 1.3696611731295877e-05,
      "loss": 0.3019,
      "step": 10059
    },
    {
      "epoch": 0.7613641738801381,
      "grad_norm": 0.78515625,
      "learning_rate": 1.3695504993277296e-05,
      "loss": 0.3412,
      "step": 10060
    },
    {
      "epoch": 0.7614398562035855,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3694398202831633e-05,
      "loss": 0.3155,
      "step": 10061
    },
    {
      "epoch": 0.7615155385270328,
      "grad_norm": 0.8125,
      "learning_rate": 1.3693291359974595e-05,
      "loss": 0.3506,
      "step": 10062
    },
    {
      "epoch": 0.7615912208504801,
      "grad_norm": 1.40625,
      "learning_rate": 1.3692184464721877e-05,
      "loss": 0.3705,
      "step": 10063
    },
    {
      "epoch": 0.7616669031739275,
      "grad_norm": 0.69140625,
      "learning_rate": 1.3691077517089188e-05,
      "loss": 0.285,
      "step": 10064
    },
    {
      "epoch": 0.7617425854973747,
      "grad_norm": 0.72265625,
      "learning_rate": 1.3689970517092232e-05,
      "loss": 0.2967,
      "step": 10065
    },
    {
      "epoch": 0.7618182678208221,
      "grad_norm": 0.77734375,
      "learning_rate": 1.3688863464746711e-05,
      "loss": 0.3455,
      "step": 10066
    },
    {
      "epoch": 0.7618939501442694,
      "grad_norm": 0.94140625,
      "learning_rate": 1.3687756360068332e-05,
      "loss": 0.3893,
      "step": 10067
    },
    {
      "epoch": 0.7619696324677168,
      "grad_norm": 0.7578125,
      "learning_rate": 1.36866492030728e-05,
      "loss": 0.3373,
      "step": 10068
    },
    {
      "epoch": 0.7620453147911641,
      "grad_norm": 0.75,
      "learning_rate": 1.3685541993775822e-05,
      "loss": 0.2972,
      "step": 10069
    },
    {
      "epoch": 0.7621209971146115,
      "grad_norm": 0.8125,
      "learning_rate": 1.3684434732193107e-05,
      "loss": 0.3221,
      "step": 10070
    },
    {
      "epoch": 0.7621966794380588,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3683327418340362e-05,
      "loss": 0.2951,
      "step": 10071
    },
    {
      "epoch": 0.762272361761506,
      "grad_norm": 0.765625,
      "learning_rate": 1.3682220052233297e-05,
      "loss": 0.3394,
      "step": 10072
    },
    {
      "epoch": 0.7623480440849534,
      "grad_norm": 0.75,
      "learning_rate": 1.368111263388762e-05,
      "loss": 0.3322,
      "step": 10073
    },
    {
      "epoch": 0.7624237264084007,
      "grad_norm": 0.80078125,
      "learning_rate": 1.3680005163319045e-05,
      "loss": 0.3088,
      "step": 10074
    },
    {
      "epoch": 0.7624994087318481,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3678897640543283e-05,
      "loss": 0.2795,
      "step": 10075
    },
    {
      "epoch": 0.7625750910552954,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3677790065576045e-05,
      "loss": 0.3068,
      "step": 10076
    },
    {
      "epoch": 0.7626507733787428,
      "grad_norm": 0.734375,
      "learning_rate": 1.367668243843304e-05,
      "loss": 0.3098,
      "step": 10077
    },
    {
      "epoch": 0.7627264557021901,
      "grad_norm": 0.72265625,
      "learning_rate": 1.3675574759129988e-05,
      "loss": 0.3055,
      "step": 10078
    },
    {
      "epoch": 0.7628021380256373,
      "grad_norm": 0.73828125,
      "learning_rate": 1.36744670276826e-05,
      "loss": 0.2893,
      "step": 10079
    },
    {
      "epoch": 0.7628778203490847,
      "grad_norm": 0.80859375,
      "learning_rate": 1.3673359244106593e-05,
      "loss": 0.329,
      "step": 10080
    },
    {
      "epoch": 0.762953502672532,
      "grad_norm": 0.70703125,
      "learning_rate": 1.3672251408417676e-05,
      "loss": 0.2628,
      "step": 10081
    },
    {
      "epoch": 0.7630291849959794,
      "grad_norm": 0.7890625,
      "learning_rate": 1.3671143520631575e-05,
      "loss": 0.3616,
      "step": 10082
    },
    {
      "epoch": 0.7631048673194267,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3670035580764002e-05,
      "loss": 0.3347,
      "step": 10083
    },
    {
      "epoch": 0.7631805496428741,
      "grad_norm": 0.69140625,
      "learning_rate": 1.366892758883068e-05,
      "loss": 0.2872,
      "step": 10084
    },
    {
      "epoch": 0.7632562319663214,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3667819544847321e-05,
      "loss": 0.3394,
      "step": 10085
    },
    {
      "epoch": 0.7633319142897687,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3666711448829648e-05,
      "loss": 0.3094,
      "step": 10086
    },
    {
      "epoch": 0.763407596613216,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3665603300793382e-05,
      "loss": 0.3419,
      "step": 10087
    },
    {
      "epoch": 0.7634832789366633,
      "grad_norm": 0.68359375,
      "learning_rate": 1.3664495100754238e-05,
      "loss": 0.2729,
      "step": 10088
    },
    {
      "epoch": 0.7635589612601107,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3663386848727949e-05,
      "loss": 0.3237,
      "step": 10089
    },
    {
      "epoch": 0.763634643583558,
      "grad_norm": 0.71484375,
      "learning_rate": 1.3662278544730225e-05,
      "loss": 0.3037,
      "step": 10090
    },
    {
      "epoch": 0.7637103259070054,
      "grad_norm": 0.71484375,
      "learning_rate": 1.36611701887768e-05,
      "loss": 0.2967,
      "step": 10091
    },
    {
      "epoch": 0.7637860082304527,
      "grad_norm": 0.77734375,
      "learning_rate": 1.366006178088339e-05,
      "loss": 0.3201,
      "step": 10092
    },
    {
      "epoch": 0.7638616905539,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3658953321065727e-05,
      "loss": 0.3101,
      "step": 10093
    },
    {
      "epoch": 0.7639373728773473,
      "grad_norm": 0.75,
      "learning_rate": 1.3657844809339529e-05,
      "loss": 0.3288,
      "step": 10094
    },
    {
      "epoch": 0.7640130552007947,
      "grad_norm": 0.7890625,
      "learning_rate": 1.3656736245720525e-05,
      "loss": 0.3478,
      "step": 10095
    },
    {
      "epoch": 0.764088737524242,
      "grad_norm": 0.8359375,
      "learning_rate": 1.3655627630224442e-05,
      "loss": 0.3482,
      "step": 10096
    },
    {
      "epoch": 0.7641644198476893,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3654518962867007e-05,
      "loss": 0.309,
      "step": 10097
    },
    {
      "epoch": 0.7642401021711367,
      "grad_norm": 0.83984375,
      "learning_rate": 1.3653410243663953e-05,
      "loss": 0.3756,
      "step": 10098
    },
    {
      "epoch": 0.764315784494584,
      "grad_norm": 0.77734375,
      "learning_rate": 1.3652301472631e-05,
      "loss": 0.299,
      "step": 10099
    },
    {
      "epoch": 0.7643914668180313,
      "grad_norm": 0.6484375,
      "learning_rate": 1.3651192649783887e-05,
      "loss": 0.2253,
      "step": 10100
    },
    {
      "epoch": 0.7644671491414786,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3650083775138341e-05,
      "loss": 0.2931,
      "step": 10101
    },
    {
      "epoch": 0.764542831464926,
      "grad_norm": 0.8203125,
      "learning_rate": 1.3648974848710088e-05,
      "loss": 0.3634,
      "step": 10102
    },
    {
      "epoch": 0.7646185137883733,
      "grad_norm": 0.77734375,
      "learning_rate": 1.3647865870514872e-05,
      "loss": 0.3367,
      "step": 10103
    },
    {
      "epoch": 0.7646941961118207,
      "grad_norm": 1.109375,
      "learning_rate": 1.3646756840568413e-05,
      "loss": 0.3345,
      "step": 10104
    },
    {
      "epoch": 0.764769878435268,
      "grad_norm": 0.71875,
      "learning_rate": 1.364564775888645e-05,
      "loss": 0.2788,
      "step": 10105
    },
    {
      "epoch": 0.7648455607587152,
      "grad_norm": 0.77734375,
      "learning_rate": 1.364453862548472e-05,
      "loss": 0.3419,
      "step": 10106
    },
    {
      "epoch": 0.7649212430821626,
      "grad_norm": 0.84375,
      "learning_rate": 1.364342944037896e-05,
      "loss": 0.3239,
      "step": 10107
    },
    {
      "epoch": 0.7649969254056099,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3642320203584892e-05,
      "loss": 0.2981,
      "step": 10108
    },
    {
      "epoch": 0.7650726077290573,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3641210915118265e-05,
      "loss": 0.317,
      "step": 10109
    },
    {
      "epoch": 0.7651482900525046,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3640101574994816e-05,
      "loss": 0.3201,
      "step": 10110
    },
    {
      "epoch": 0.765223972375952,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3638992183230273e-05,
      "loss": 0.3349,
      "step": 10111
    },
    {
      "epoch": 0.7652996546993993,
      "grad_norm": 0.81640625,
      "learning_rate": 1.3637882739840387e-05,
      "loss": 0.3322,
      "step": 10112
    },
    {
      "epoch": 0.7653753370228465,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3636773244840891e-05,
      "loss": 0.3279,
      "step": 10113
    },
    {
      "epoch": 0.7654510193462939,
      "grad_norm": 0.8515625,
      "learning_rate": 1.3635663698247522e-05,
      "loss": 0.3697,
      "step": 10114
    },
    {
      "epoch": 0.7655267016697412,
      "grad_norm": 0.79296875,
      "learning_rate": 1.3634554100076025e-05,
      "loss": 0.3312,
      "step": 10115
    },
    {
      "epoch": 0.7656023839931886,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3633444450342142e-05,
      "loss": 0.2634,
      "step": 10116
    },
    {
      "epoch": 0.7656780663166359,
      "grad_norm": 0.75,
      "learning_rate": 1.3632334749061615e-05,
      "loss": 0.3034,
      "step": 10117
    },
    {
      "epoch": 0.7657537486400833,
      "grad_norm": 0.77734375,
      "learning_rate": 1.3631224996250186e-05,
      "loss": 0.3541,
      "step": 10118
    },
    {
      "epoch": 0.7658294309635306,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3630115191923598e-05,
      "loss": 0.3403,
      "step": 10119
    },
    {
      "epoch": 0.7659051132869779,
      "grad_norm": 0.72265625,
      "learning_rate": 1.3629005336097596e-05,
      "loss": 0.3082,
      "step": 10120
    },
    {
      "epoch": 0.7659807956104252,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3627895428787929e-05,
      "loss": 0.2752,
      "step": 10121
    },
    {
      "epoch": 0.7660564779338725,
      "grad_norm": 0.765625,
      "learning_rate": 1.362678547001034e-05,
      "loss": 0.3231,
      "step": 10122
    },
    {
      "epoch": 0.7661321602573199,
      "grad_norm": 0.77734375,
      "learning_rate": 1.362567545978057e-05,
      "loss": 0.342,
      "step": 10123
    },
    {
      "epoch": 0.7662078425807672,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3624565398114375e-05,
      "loss": 0.3169,
      "step": 10124
    },
    {
      "epoch": 0.7662835249042146,
      "grad_norm": 0.78125,
      "learning_rate": 1.3623455285027501e-05,
      "loss": 0.3655,
      "step": 10125
    },
    {
      "epoch": 0.7663592072276619,
      "grad_norm": 0.80078125,
      "learning_rate": 1.3622345120535693e-05,
      "loss": 0.3254,
      "step": 10126
    },
    {
      "epoch": 0.7664348895511092,
      "grad_norm": 0.77734375,
      "learning_rate": 1.3621234904654705e-05,
      "loss": 0.3166,
      "step": 10127
    },
    {
      "epoch": 0.7665105718745565,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3620124637400286e-05,
      "loss": 0.2992,
      "step": 10128
    },
    {
      "epoch": 0.7665862541980039,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3619014318788185e-05,
      "loss": 0.3311,
      "step": 10129
    },
    {
      "epoch": 0.7666619365214512,
      "grad_norm": 0.78125,
      "learning_rate": 1.3617903948834155e-05,
      "loss": 0.3473,
      "step": 10130
    },
    {
      "epoch": 0.7667376188448985,
      "grad_norm": 0.84375,
      "learning_rate": 1.3616793527553952e-05,
      "loss": 0.3937,
      "step": 10131
    },
    {
      "epoch": 0.7668133011683459,
      "grad_norm": 0.703125,
      "learning_rate": 1.3615683054963325e-05,
      "loss": 0.2598,
      "step": 10132
    },
    {
      "epoch": 0.7668889834917932,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3614572531078028e-05,
      "loss": 0.3125,
      "step": 10133
    },
    {
      "epoch": 0.7669646658152405,
      "grad_norm": 0.70703125,
      "learning_rate": 1.3613461955913818e-05,
      "loss": 0.2911,
      "step": 10134
    },
    {
      "epoch": 0.7670403481386878,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3612351329486448e-05,
      "loss": 0.2832,
      "step": 10135
    },
    {
      "epoch": 0.7671160304621352,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3611240651811677e-05,
      "loss": 0.3425,
      "step": 10136
    },
    {
      "epoch": 0.7671917127855825,
      "grad_norm": 0.71484375,
      "learning_rate": 1.3610129922905257e-05,
      "loss": 0.2837,
      "step": 10137
    },
    {
      "epoch": 0.7672673951090299,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3609019142782951e-05,
      "loss": 0.3116,
      "step": 10138
    },
    {
      "epoch": 0.7673430774324772,
      "grad_norm": 0.80078125,
      "learning_rate": 1.3607908311460516e-05,
      "loss": 0.3565,
      "step": 10139
    },
    {
      "epoch": 0.7674187597559246,
      "grad_norm": 0.67578125,
      "learning_rate": 1.360679742895371e-05,
      "loss": 0.2551,
      "step": 10140
    },
    {
      "epoch": 0.7674944420793718,
      "grad_norm": 0.734375,
      "learning_rate": 1.3605686495278293e-05,
      "loss": 0.3249,
      "step": 10141
    },
    {
      "epoch": 0.7675701244028191,
      "grad_norm": 0.94921875,
      "learning_rate": 1.3604575510450025e-05,
      "loss": 0.3525,
      "step": 10142
    },
    {
      "epoch": 0.7676458067262665,
      "grad_norm": 0.82421875,
      "learning_rate": 1.3603464474484666e-05,
      "loss": 0.3355,
      "step": 10143
    },
    {
      "epoch": 0.7677214890497138,
      "grad_norm": 0.71875,
      "learning_rate": 1.3602353387397983e-05,
      "loss": 0.2993,
      "step": 10144
    },
    {
      "epoch": 0.7677971713731612,
      "grad_norm": 0.734375,
      "learning_rate": 1.3601242249205736e-05,
      "loss": 0.315,
      "step": 10145
    },
    {
      "epoch": 0.7678728536966085,
      "grad_norm": 0.71875,
      "learning_rate": 1.3600131059923683e-05,
      "loss": 0.2836,
      "step": 10146
    },
    {
      "epoch": 0.7679485360200559,
      "grad_norm": 0.7109375,
      "learning_rate": 1.3599019819567596e-05,
      "loss": 0.2779,
      "step": 10147
    },
    {
      "epoch": 0.7680242183435031,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3597908528153236e-05,
      "loss": 0.3163,
      "step": 10148
    },
    {
      "epoch": 0.7680999006669504,
      "grad_norm": 0.72265625,
      "learning_rate": 1.3596797185696371e-05,
      "loss": 0.2985,
      "step": 10149
    },
    {
      "epoch": 0.7681755829903978,
      "grad_norm": 0.7109375,
      "learning_rate": 1.3595685792212763e-05,
      "loss": 0.2762,
      "step": 10150
    },
    {
      "epoch": 0.7682512653138451,
      "grad_norm": 0.796875,
      "learning_rate": 1.3594574347718183e-05,
      "loss": 0.3254,
      "step": 10151
    },
    {
      "epoch": 0.7683269476372925,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3593462852228396e-05,
      "loss": 0.3091,
      "step": 10152
    },
    {
      "epoch": 0.7684026299607398,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3592351305759173e-05,
      "loss": 0.3099,
      "step": 10153
    },
    {
      "epoch": 0.7684783122841872,
      "grad_norm": 0.97265625,
      "learning_rate": 1.3591239708326286e-05,
      "loss": 0.3451,
      "step": 10154
    },
    {
      "epoch": 0.7685539946076344,
      "grad_norm": 0.765625,
      "learning_rate": 1.3590128059945493e-05,
      "loss": 0.3442,
      "step": 10155
    },
    {
      "epoch": 0.7686296769310818,
      "grad_norm": 0.71875,
      "learning_rate": 1.3589016360632577e-05,
      "loss": 0.2834,
      "step": 10156
    },
    {
      "epoch": 0.7687053592545291,
      "grad_norm": 0.69140625,
      "learning_rate": 1.3587904610403304e-05,
      "loss": 0.2728,
      "step": 10157
    },
    {
      "epoch": 0.7687810415779764,
      "grad_norm": 0.7109375,
      "learning_rate": 1.3586792809273447e-05,
      "loss": 0.2796,
      "step": 10158
    },
    {
      "epoch": 0.7688567239014238,
      "grad_norm": 0.76171875,
      "learning_rate": 1.358568095725878e-05,
      "loss": 0.3081,
      "step": 10159
    },
    {
      "epoch": 0.7689324062248711,
      "grad_norm": 0.80859375,
      "learning_rate": 1.3584569054375076e-05,
      "loss": 0.3391,
      "step": 10160
    },
    {
      "epoch": 0.7690080885483185,
      "grad_norm": 0.78125,
      "learning_rate": 1.3583457100638104e-05,
      "loss": 0.3122,
      "step": 10161
    },
    {
      "epoch": 0.7690837708717657,
      "grad_norm": 0.7890625,
      "learning_rate": 1.358234509606365e-05,
      "loss": 0.3518,
      "step": 10162
    },
    {
      "epoch": 0.7691594531952131,
      "grad_norm": 0.71484375,
      "learning_rate": 1.3581233040667482e-05,
      "loss": 0.2936,
      "step": 10163
    },
    {
      "epoch": 0.7692351355186604,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3580120934465372e-05,
      "loss": 0.3386,
      "step": 10164
    },
    {
      "epoch": 0.7693108178421078,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3579008777473108e-05,
      "loss": 0.3178,
      "step": 10165
    },
    {
      "epoch": 0.7693865001655551,
      "grad_norm": 0.75,
      "learning_rate": 1.3577896569706464e-05,
      "loss": 0.3029,
      "step": 10166
    },
    {
      "epoch": 0.7694621824890024,
      "grad_norm": 0.69921875,
      "learning_rate": 1.3576784311181212e-05,
      "loss": 0.2636,
      "step": 10167
    },
    {
      "epoch": 0.7695378648124498,
      "grad_norm": 0.72265625,
      "learning_rate": 1.3575672001913142e-05,
      "loss": 0.248,
      "step": 10168
    },
    {
      "epoch": 0.769613547135897,
      "grad_norm": 0.71875,
      "learning_rate": 1.3574559641918025e-05,
      "loss": 0.291,
      "step": 10169
    },
    {
      "epoch": 0.7696892294593444,
      "grad_norm": 0.95703125,
      "learning_rate": 1.3573447231211644e-05,
      "loss": 0.3213,
      "step": 10170
    },
    {
      "epoch": 0.7697649117827917,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3572334769809785e-05,
      "loss": 0.3343,
      "step": 10171
    },
    {
      "epoch": 0.7698405941062391,
      "grad_norm": 1.015625,
      "learning_rate": 1.3571222257728228e-05,
      "loss": 0.3235,
      "step": 10172
    },
    {
      "epoch": 0.7699162764296864,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3570109694982751e-05,
      "loss": 0.2945,
      "step": 10173
    },
    {
      "epoch": 0.7699919587531338,
      "grad_norm": 0.78125,
      "learning_rate": 1.3568997081589146e-05,
      "loss": 0.3738,
      "step": 10174
    },
    {
      "epoch": 0.7700676410765811,
      "grad_norm": 0.765625,
      "learning_rate": 1.3567884417563188e-05,
      "loss": 0.3176,
      "step": 10175
    },
    {
      "epoch": 0.7701433234000283,
      "grad_norm": 0.765625,
      "learning_rate": 1.3566771702920667e-05,
      "loss": 0.3217,
      "step": 10176
    },
    {
      "epoch": 0.7702190057234757,
      "grad_norm": 0.71484375,
      "learning_rate": 1.3565658937677375e-05,
      "loss": 0.3217,
      "step": 10177
    },
    {
      "epoch": 0.770294688046923,
      "grad_norm": 0.7890625,
      "learning_rate": 1.3564546121849087e-05,
      "loss": 0.3519,
      "step": 10178
    },
    {
      "epoch": 0.7703703703703704,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3563433255451594e-05,
      "loss": 0.3094,
      "step": 10179
    },
    {
      "epoch": 0.7704460526938177,
      "grad_norm": 0.70703125,
      "learning_rate": 1.3562320338500687e-05,
      "loss": 0.277,
      "step": 10180
    },
    {
      "epoch": 0.7705217350172651,
      "grad_norm": 0.69921875,
      "learning_rate": 1.3561207371012155e-05,
      "loss": 0.2694,
      "step": 10181
    },
    {
      "epoch": 0.7705974173407124,
      "grad_norm": 0.734375,
      "learning_rate": 1.3560094353001783e-05,
      "loss": 0.2804,
      "step": 10182
    },
    {
      "epoch": 0.7706730996641596,
      "grad_norm": 0.80078125,
      "learning_rate": 1.3558981284485361e-05,
      "loss": 0.3235,
      "step": 10183
    },
    {
      "epoch": 0.770748781987607,
      "grad_norm": 0.8125,
      "learning_rate": 1.3557868165478688e-05,
      "loss": 0.3529,
      "step": 10184
    },
    {
      "epoch": 0.7708244643110543,
      "grad_norm": 0.8125,
      "learning_rate": 1.3556754995997543e-05,
      "loss": 0.3469,
      "step": 10185
    },
    {
      "epoch": 0.7709001466345017,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3555641776057729e-05,
      "loss": 0.3094,
      "step": 10186
    },
    {
      "epoch": 0.770975828957949,
      "grad_norm": 0.79296875,
      "learning_rate": 1.3554528505675033e-05,
      "loss": 0.3238,
      "step": 10187
    },
    {
      "epoch": 0.7710515112813964,
      "grad_norm": 0.80078125,
      "learning_rate": 1.3553415184865251e-05,
      "loss": 0.3093,
      "step": 10188
    },
    {
      "epoch": 0.7711271936048437,
      "grad_norm": 0.78125,
      "learning_rate": 1.3552301813644177e-05,
      "loss": 0.3095,
      "step": 10189
    },
    {
      "epoch": 0.771202875928291,
      "grad_norm": 0.734375,
      "learning_rate": 1.3551188392027606e-05,
      "loss": 0.3148,
      "step": 10190
    },
    {
      "epoch": 0.7712785582517383,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3550074920031332e-05,
      "loss": 0.3346,
      "step": 10191
    },
    {
      "epoch": 0.7713542405751856,
      "grad_norm": 0.8828125,
      "learning_rate": 1.3548961397671156e-05,
      "loss": 0.3623,
      "step": 10192
    },
    {
      "epoch": 0.771429922898633,
      "grad_norm": 0.74609375,
      "learning_rate": 1.354784782496287e-05,
      "loss": 0.3263,
      "step": 10193
    },
    {
      "epoch": 0.7715056052220803,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3546734201922273e-05,
      "loss": 0.3269,
      "step": 10194
    },
    {
      "epoch": 0.7715812875455277,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3545620528565169e-05,
      "loss": 0.3221,
      "step": 10195
    },
    {
      "epoch": 0.771656969868975,
      "grad_norm": 0.71875,
      "learning_rate": 1.3544506804907351e-05,
      "loss": 0.2917,
      "step": 10196
    },
    {
      "epoch": 0.7717326521924223,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3543393030964622e-05,
      "loss": 0.3285,
      "step": 10197
    },
    {
      "epoch": 0.7718083345158696,
      "grad_norm": 0.9921875,
      "learning_rate": 1.3542279206752782e-05,
      "loss": 0.3506,
      "step": 10198
    },
    {
      "epoch": 0.771884016839317,
      "grad_norm": 0.7890625,
      "learning_rate": 1.354116533228763e-05,
      "loss": 0.3132,
      "step": 10199
    },
    {
      "epoch": 0.7719596991627643,
      "grad_norm": 0.8203125,
      "learning_rate": 1.3540051407584973e-05,
      "loss": 0.3631,
      "step": 10200
    },
    {
      "epoch": 0.7720353814862116,
      "grad_norm": 0.68359375,
      "learning_rate": 1.3538937432660614e-05,
      "loss": 0.2723,
      "step": 10201
    },
    {
      "epoch": 0.772111063809659,
      "grad_norm": 0.68359375,
      "learning_rate": 1.353782340753035e-05,
      "loss": 0.2789,
      "step": 10202
    },
    {
      "epoch": 0.7721867461331063,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3536709332209995e-05,
      "loss": 0.2841,
      "step": 10203
    },
    {
      "epoch": 0.7722624284565536,
      "grad_norm": 0.71875,
      "learning_rate": 1.3535595206715346e-05,
      "loss": 0.3088,
      "step": 10204
    },
    {
      "epoch": 0.7723381107800009,
      "grad_norm": 0.77734375,
      "learning_rate": 1.3534481031062213e-05,
      "loss": 0.3177,
      "step": 10205
    },
    {
      "epoch": 0.7724137931034483,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3533366805266402e-05,
      "loss": 0.3305,
      "step": 10206
    },
    {
      "epoch": 0.7724894754268956,
      "grad_norm": 0.82421875,
      "learning_rate": 1.3532252529343718e-05,
      "loss": 0.3494,
      "step": 10207
    },
    {
      "epoch": 0.772565157750343,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3531138203309966e-05,
      "loss": 0.3245,
      "step": 10208
    },
    {
      "epoch": 0.7726408400737903,
      "grad_norm": 0.69140625,
      "learning_rate": 1.3530023827180965e-05,
      "loss": 0.2643,
      "step": 10209
    },
    {
      "epoch": 0.7727165223972376,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3528909400972517e-05,
      "loss": 0.2875,
      "step": 10210
    },
    {
      "epoch": 0.7727922047206849,
      "grad_norm": 0.7578125,
      "learning_rate": 1.352779492470043e-05,
      "loss": 0.3322,
      "step": 10211
    },
    {
      "epoch": 0.7728678870441322,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3526680398380523e-05,
      "loss": 0.2988,
      "step": 10212
    },
    {
      "epoch": 0.7729435693675796,
      "grad_norm": 0.703125,
      "learning_rate": 1.35255658220286e-05,
      "loss": 0.2914,
      "step": 10213
    },
    {
      "epoch": 0.7730192516910269,
      "grad_norm": 0.83984375,
      "learning_rate": 1.3524451195660472e-05,
      "loss": 0.343,
      "step": 10214
    },
    {
      "epoch": 0.7730949340144743,
      "grad_norm": 0.7890625,
      "learning_rate": 1.3523336519291963e-05,
      "loss": 0.3164,
      "step": 10215
    },
    {
      "epoch": 0.7731706163379216,
      "grad_norm": 0.78515625,
      "learning_rate": 1.3522221792938877e-05,
      "loss": 0.3327,
      "step": 10216
    },
    {
      "epoch": 0.773246298661369,
      "grad_norm": 0.72265625,
      "learning_rate": 1.3521107016617027e-05,
      "loss": 0.2972,
      "step": 10217
    },
    {
      "epoch": 0.7733219809848162,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3519992190342234e-05,
      "loss": 0.3025,
      "step": 10218
    },
    {
      "epoch": 0.7733976633082635,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3518877314130314e-05,
      "loss": 0.3355,
      "step": 10219
    },
    {
      "epoch": 0.7734733456317109,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3517762387997076e-05,
      "loss": 0.332,
      "step": 10220
    },
    {
      "epoch": 0.7735490279551582,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3516647411958344e-05,
      "loss": 0.3313,
      "step": 10221
    },
    {
      "epoch": 0.7736247102786056,
      "grad_norm": 0.70703125,
      "learning_rate": 1.3515532386029935e-05,
      "loss": 0.3055,
      "step": 10222
    },
    {
      "epoch": 0.7737003926020529,
      "grad_norm": 0.69140625,
      "learning_rate": 1.351441731022766e-05,
      "loss": 0.281,
      "step": 10223
    },
    {
      "epoch": 0.7737760749255003,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3513302184567355e-05,
      "loss": 0.2879,
      "step": 10224
    },
    {
      "epoch": 0.7738517572489475,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3512187009064822e-05,
      "loss": 0.3405,
      "step": 10225
    },
    {
      "epoch": 0.7739274395723948,
      "grad_norm": 0.984375,
      "learning_rate": 1.3511071783735887e-05,
      "loss": 0.3381,
      "step": 10226
    },
    {
      "epoch": 0.7740031218958422,
      "grad_norm": 0.73046875,
      "learning_rate": 1.350995650859638e-05,
      "loss": 0.2777,
      "step": 10227
    },
    {
      "epoch": 0.7740788042192895,
      "grad_norm": 0.9296875,
      "learning_rate": 1.3508841183662114e-05,
      "loss": 0.3151,
      "step": 10228
    },
    {
      "epoch": 0.7741544865427369,
      "grad_norm": 0.79296875,
      "learning_rate": 1.3507725808948916e-05,
      "loss": 0.341,
      "step": 10229
    },
    {
      "epoch": 0.7742301688661842,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3506610384472605e-05,
      "loss": 0.2846,
      "step": 10230
    },
    {
      "epoch": 0.7743058511896315,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3505494910249008e-05,
      "loss": 0.2932,
      "step": 10231
    },
    {
      "epoch": 0.7743815335130788,
      "grad_norm": 0.703125,
      "learning_rate": 1.3504379386293951e-05,
      "loss": 0.282,
      "step": 10232
    },
    {
      "epoch": 0.7744572158365262,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3503263812623262e-05,
      "loss": 0.2855,
      "step": 10233
    },
    {
      "epoch": 0.7745328981599735,
      "grad_norm": 0.80078125,
      "learning_rate": 1.3502148189252761e-05,
      "loss": 0.3527,
      "step": 10234
    },
    {
      "epoch": 0.7746085804834208,
      "grad_norm": 0.96875,
      "learning_rate": 1.3501032516198276e-05,
      "loss": 0.3386,
      "step": 10235
    },
    {
      "epoch": 0.7746842628068682,
      "grad_norm": 0.74609375,
      "learning_rate": 1.349991679347564e-05,
      "loss": 0.3064,
      "step": 10236
    },
    {
      "epoch": 0.7747599451303155,
      "grad_norm": 0.71484375,
      "learning_rate": 1.3498801021100678e-05,
      "loss": 0.2906,
      "step": 10237
    },
    {
      "epoch": 0.7748356274537628,
      "grad_norm": 0.765625,
      "learning_rate": 1.349768519908922e-05,
      "loss": 0.3219,
      "step": 10238
    },
    {
      "epoch": 0.7749113097772101,
      "grad_norm": 1.1640625,
      "learning_rate": 1.3496569327457092e-05,
      "loss": 0.3275,
      "step": 10239
    },
    {
      "epoch": 0.7749869921006575,
      "grad_norm": 0.6875,
      "learning_rate": 1.349545340622013e-05,
      "loss": 0.2514,
      "step": 10240
    },
    {
      "epoch": 0.7750626744241048,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3494337435394161e-05,
      "loss": 0.3103,
      "step": 10241
    },
    {
      "epoch": 0.7751383567475522,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3493221414995021e-05,
      "loss": 0.3292,
      "step": 10242
    },
    {
      "epoch": 0.7752140390709995,
      "grad_norm": 0.734375,
      "learning_rate": 1.349210534503854e-05,
      "loss": 0.3095,
      "step": 10243
    },
    {
      "epoch": 0.7752897213944469,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3490989225540552e-05,
      "loss": 0.3678,
      "step": 10244
    },
    {
      "epoch": 0.7753654037178941,
      "grad_norm": 0.70703125,
      "learning_rate": 1.3489873056516892e-05,
      "loss": 0.2756,
      "step": 10245
    },
    {
      "epoch": 0.7754410860413414,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3488756837983395e-05,
      "loss": 0.3126,
      "step": 10246
    },
    {
      "epoch": 0.7755167683647888,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3487640569955894e-05,
      "loss": 0.3244,
      "step": 10247
    },
    {
      "epoch": 0.7755924506882361,
      "grad_norm": 0.8984375,
      "learning_rate": 1.3486524252450226e-05,
      "loss": 0.4057,
      "step": 10248
    },
    {
      "epoch": 0.7756681330116835,
      "grad_norm": 0.78125,
      "learning_rate": 1.3485407885482226e-05,
      "loss": 0.3123,
      "step": 10249
    },
    {
      "epoch": 0.7757438153351308,
      "grad_norm": 0.79296875,
      "learning_rate": 1.3484291469067736e-05,
      "loss": 0.3224,
      "step": 10250
    },
    {
      "epoch": 0.7758194976585782,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3483175003222594e-05,
      "loss": 0.3437,
      "step": 10251
    },
    {
      "epoch": 0.7758951799820254,
      "grad_norm": 0.71484375,
      "learning_rate": 1.3482058487962638e-05,
      "loss": 0.2663,
      "step": 10252
    },
    {
      "epoch": 0.7759708623054727,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3480941923303702e-05,
      "loss": 0.3087,
      "step": 10253
    },
    {
      "epoch": 0.7760465446289201,
      "grad_norm": 0.765625,
      "learning_rate": 1.3479825309261635e-05,
      "loss": 0.3411,
      "step": 10254
    },
    {
      "epoch": 0.7761222269523674,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3478708645852272e-05,
      "loss": 0.3246,
      "step": 10255
    },
    {
      "epoch": 0.7761979092758148,
      "grad_norm": 0.75,
      "learning_rate": 1.347759193309146e-05,
      "loss": 0.3167,
      "step": 10256
    },
    {
      "epoch": 0.7762735915992621,
      "grad_norm": 0.70703125,
      "learning_rate": 1.3476475170995041e-05,
      "loss": 0.2941,
      "step": 10257
    },
    {
      "epoch": 0.7763492739227095,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3475358359578849e-05,
      "loss": 0.3349,
      "step": 10258
    },
    {
      "epoch": 0.7764249562461567,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3474241498858737e-05,
      "loss": 0.3055,
      "step": 10259
    },
    {
      "epoch": 0.776500638569604,
      "grad_norm": 0.95703125,
      "learning_rate": 1.347312458885055e-05,
      "loss": 0.3539,
      "step": 10260
    },
    {
      "epoch": 0.7765763208930514,
      "grad_norm": 1.0234375,
      "learning_rate": 1.347200762957013e-05,
      "loss": 0.3186,
      "step": 10261
    },
    {
      "epoch": 0.7766520032164987,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3470890621033323e-05,
      "loss": 0.3165,
      "step": 10262
    },
    {
      "epoch": 0.7767276855399461,
      "grad_norm": 0.734375,
      "learning_rate": 1.3469773563255975e-05,
      "loss": 0.315,
      "step": 10263
    },
    {
      "epoch": 0.7768033678633934,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3468656456253934e-05,
      "loss": 0.3276,
      "step": 10264
    },
    {
      "epoch": 0.7768790501868408,
      "grad_norm": 0.77734375,
      "learning_rate": 1.346753930004305e-05,
      "loss": 0.3353,
      "step": 10265
    },
    {
      "epoch": 0.776954732510288,
      "grad_norm": 0.765625,
      "learning_rate": 1.3466422094639175e-05,
      "loss": 0.3181,
      "step": 10266
    },
    {
      "epoch": 0.7770304148337354,
      "grad_norm": 0.78125,
      "learning_rate": 1.3465304840058147e-05,
      "loss": 0.3406,
      "step": 10267
    },
    {
      "epoch": 0.7771060971571827,
      "grad_norm": 0.80078125,
      "learning_rate": 1.3464187536315828e-05,
      "loss": 0.3193,
      "step": 10268
    },
    {
      "epoch": 0.77718177948063,
      "grad_norm": 0.71875,
      "learning_rate": 1.3463070183428062e-05,
      "loss": 0.3049,
      "step": 10269
    },
    {
      "epoch": 0.7772574618040774,
      "grad_norm": 0.6953125,
      "learning_rate": 1.34619527814107e-05,
      "loss": 0.2826,
      "step": 10270
    },
    {
      "epoch": 0.7773331441275247,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3460835330279602e-05,
      "loss": 0.3128,
      "step": 10271
    },
    {
      "epoch": 0.7774088264509721,
      "grad_norm": 0.765625,
      "learning_rate": 1.3459717830050613e-05,
      "loss": 0.3529,
      "step": 10272
    },
    {
      "epoch": 0.7774845087744193,
      "grad_norm": 0.80078125,
      "learning_rate": 1.3458600280739589e-05,
      "loss": 0.3384,
      "step": 10273
    },
    {
      "epoch": 0.7775601910978667,
      "grad_norm": 0.83984375,
      "learning_rate": 1.3457482682362387e-05,
      "loss": 0.3765,
      "step": 10274
    },
    {
      "epoch": 0.777635873421314,
      "grad_norm": 0.765625,
      "learning_rate": 1.3456365034934859e-05,
      "loss": 0.3142,
      "step": 10275
    },
    {
      "epoch": 0.7777115557447614,
      "grad_norm": 0.79296875,
      "learning_rate": 1.3455247338472861e-05,
      "loss": 0.3555,
      "step": 10276
    },
    {
      "epoch": 0.7777872380682087,
      "grad_norm": 0.7421875,
      "learning_rate": 1.345412959299225e-05,
      "loss": 0.3088,
      "step": 10277
    },
    {
      "epoch": 0.777862920391656,
      "grad_norm": 0.94140625,
      "learning_rate": 1.3453011798508886e-05,
      "loss": 0.3256,
      "step": 10278
    },
    {
      "epoch": 0.7779386027151034,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3451893955038618e-05,
      "loss": 0.3076,
      "step": 10279
    },
    {
      "epoch": 0.7780142850385506,
      "grad_norm": 0.859375,
      "learning_rate": 1.3450776062597316e-05,
      "loss": 0.369,
      "step": 10280
    },
    {
      "epoch": 0.778089967361998,
      "grad_norm": 1.140625,
      "learning_rate": 1.3449658121200834e-05,
      "loss": 0.3798,
      "step": 10281
    },
    {
      "epoch": 0.7781656496854453,
      "grad_norm": 0.69921875,
      "learning_rate": 1.3448540130865028e-05,
      "loss": 0.26,
      "step": 10282
    },
    {
      "epoch": 0.7782413320088927,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3447422091605764e-05,
      "loss": 0.3369,
      "step": 10283
    },
    {
      "epoch": 0.77831701433234,
      "grad_norm": 0.80859375,
      "learning_rate": 1.3446304003438904e-05,
      "loss": 0.3401,
      "step": 10284
    },
    {
      "epoch": 0.7783926966557874,
      "grad_norm": 0.87109375,
      "learning_rate": 1.3445185866380307e-05,
      "loss": 0.278,
      "step": 10285
    },
    {
      "epoch": 0.7784683789792347,
      "grad_norm": 0.80859375,
      "learning_rate": 1.3444067680445834e-05,
      "loss": 0.3473,
      "step": 10286
    },
    {
      "epoch": 0.778544061302682,
      "grad_norm": 0.78125,
      "learning_rate": 1.3442949445651352e-05,
      "loss": 0.3051,
      "step": 10287
    },
    {
      "epoch": 0.7786197436261293,
      "grad_norm": 0.79296875,
      "learning_rate": 1.3441831162012719e-05,
      "loss": 0.3279,
      "step": 10288
    },
    {
      "epoch": 0.7786954259495766,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3440712829545813e-05,
      "loss": 0.3271,
      "step": 10289
    },
    {
      "epoch": 0.778771108273024,
      "grad_norm": 0.82421875,
      "learning_rate": 1.3439594448266484e-05,
      "loss": 0.3429,
      "step": 10290
    },
    {
      "epoch": 0.7788467905964713,
      "grad_norm": 0.6875,
      "learning_rate": 1.343847601819061e-05,
      "loss": 0.2722,
      "step": 10291
    },
    {
      "epoch": 0.7789224729199187,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3437357539334053e-05,
      "loss": 0.3328,
      "step": 10292
    },
    {
      "epoch": 0.778998155243366,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3436239011712677e-05,
      "loss": 0.3117,
      "step": 10293
    },
    {
      "epoch": 0.7790738375668133,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3435120435342357e-05,
      "loss": 0.3082,
      "step": 10294
    },
    {
      "epoch": 0.7791495198902606,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3434001810238957e-05,
      "loss": 0.3198,
      "step": 10295
    },
    {
      "epoch": 0.779225202213708,
      "grad_norm": 0.83203125,
      "learning_rate": 1.343288313641835e-05,
      "loss": 0.3436,
      "step": 10296
    },
    {
      "epoch": 0.7793008845371553,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3431764413896403e-05,
      "loss": 0.2896,
      "step": 10297
    },
    {
      "epoch": 0.7793765668606026,
      "grad_norm": 0.75390625,
      "learning_rate": 1.343064564268899e-05,
      "loss": 0.329,
      "step": 10298
    },
    {
      "epoch": 0.77945224918405,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3429526822811978e-05,
      "loss": 0.3216,
      "step": 10299
    },
    {
      "epoch": 0.7795279315074973,
      "grad_norm": 0.78515625,
      "learning_rate": 1.3428407954281246e-05,
      "loss": 0.3279,
      "step": 10300
    },
    {
      "epoch": 0.7796036138309446,
      "grad_norm": 0.9921875,
      "learning_rate": 1.3427289037112665e-05,
      "loss": 0.3085,
      "step": 10301
    },
    {
      "epoch": 0.7796792961543919,
      "grad_norm": 0.72265625,
      "learning_rate": 1.3426170071322104e-05,
      "loss": 0.2971,
      "step": 10302
    },
    {
      "epoch": 0.7797549784778393,
      "grad_norm": 0.77734375,
      "learning_rate": 1.3425051056925444e-05,
      "loss": 0.3411,
      "step": 10303
    },
    {
      "epoch": 0.7798306608012866,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3423931993938557e-05,
      "loss": 0.3267,
      "step": 10304
    },
    {
      "epoch": 0.779906343124734,
      "grad_norm": 1.0234375,
      "learning_rate": 1.3422812882377315e-05,
      "loss": 0.3105,
      "step": 10305
    },
    {
      "epoch": 0.7799820254481813,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3421693722257603e-05,
      "loss": 0.2925,
      "step": 10306
    },
    {
      "epoch": 0.7800577077716286,
      "grad_norm": 0.7109375,
      "learning_rate": 1.3420574513595292e-05,
      "loss": 0.29,
      "step": 10307
    },
    {
      "epoch": 0.7801333900950759,
      "grad_norm": 0.72265625,
      "learning_rate": 1.3419455256406263e-05,
      "loss": 0.3176,
      "step": 10308
    },
    {
      "epoch": 0.7802090724185232,
      "grad_norm": 0.734375,
      "learning_rate": 1.3418335950706391e-05,
      "loss": 0.3229,
      "step": 10309
    },
    {
      "epoch": 0.7802847547419706,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3417216596511557e-05,
      "loss": 0.3503,
      "step": 10310
    },
    {
      "epoch": 0.7803604370654179,
      "grad_norm": 0.77734375,
      "learning_rate": 1.3416097193837642e-05,
      "loss": 0.2937,
      "step": 10311
    },
    {
      "epoch": 0.7804361193888653,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3414977742700528e-05,
      "loss": 0.3139,
      "step": 10312
    },
    {
      "epoch": 0.7805118017123126,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3413858243116095e-05,
      "loss": 0.3207,
      "step": 10313
    },
    {
      "epoch": 0.78058748403576,
      "grad_norm": 0.92578125,
      "learning_rate": 1.3412738695100222e-05,
      "loss": 0.315,
      "step": 10314
    },
    {
      "epoch": 0.7806631663592072,
      "grad_norm": 0.796875,
      "learning_rate": 1.3411619098668795e-05,
      "loss": 0.3291,
      "step": 10315
    },
    {
      "epoch": 0.7807388486826545,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3410499453837697e-05,
      "loss": 0.2816,
      "step": 10316
    },
    {
      "epoch": 0.7808145310061019,
      "grad_norm": 0.734375,
      "learning_rate": 1.3409379760622812e-05,
      "loss": 0.2842,
      "step": 10317
    },
    {
      "epoch": 0.7808902133295492,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3408260019040022e-05,
      "loss": 0.2973,
      "step": 10318
    },
    {
      "epoch": 0.7809658956529966,
      "grad_norm": 0.80859375,
      "learning_rate": 1.3407140229105218e-05,
      "loss": 0.3316,
      "step": 10319
    },
    {
      "epoch": 0.7810415779764439,
      "grad_norm": 0.70703125,
      "learning_rate": 1.340602039083428e-05,
      "loss": 0.2724,
      "step": 10320
    },
    {
      "epoch": 0.7811172602998913,
      "grad_norm": 0.75,
      "learning_rate": 1.3404900504243099e-05,
      "loss": 0.3046,
      "step": 10321
    },
    {
      "epoch": 0.7811929426233385,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3403780569347563e-05,
      "loss": 0.3319,
      "step": 10322
    },
    {
      "epoch": 0.7812686249467858,
      "grad_norm": 0.78125,
      "learning_rate": 1.3402660586163556e-05,
      "loss": 0.3311,
      "step": 10323
    },
    {
      "epoch": 0.7813443072702332,
      "grad_norm": 0.76953125,
      "learning_rate": 1.340154055470697e-05,
      "loss": 0.3298,
      "step": 10324
    },
    {
      "epoch": 0.7814199895936805,
      "grad_norm": 0.828125,
      "learning_rate": 1.3400420474993696e-05,
      "loss": 0.3591,
      "step": 10325
    },
    {
      "epoch": 0.7814956719171279,
      "grad_norm": 0.7109375,
      "learning_rate": 1.339930034703962e-05,
      "loss": 0.2645,
      "step": 10326
    },
    {
      "epoch": 0.7815713542405752,
      "grad_norm": 0.83203125,
      "learning_rate": 1.3398180170860637e-05,
      "loss": 0.3536,
      "step": 10327
    },
    {
      "epoch": 0.7816470365640226,
      "grad_norm": 0.75,
      "learning_rate": 1.3397059946472637e-05,
      "loss": 0.3104,
      "step": 10328
    },
    {
      "epoch": 0.7817227188874698,
      "grad_norm": 0.734375,
      "learning_rate": 1.339593967389151e-05,
      "loss": 0.3127,
      "step": 10329
    },
    {
      "epoch": 0.7817984012109171,
      "grad_norm": 0.80859375,
      "learning_rate": 1.3394819353133151e-05,
      "loss": 0.3309,
      "step": 10330
    },
    {
      "epoch": 0.7818740835343645,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3393698984213456e-05,
      "loss": 0.3125,
      "step": 10331
    },
    {
      "epoch": 0.7819497658578118,
      "grad_norm": 0.71875,
      "learning_rate": 1.3392578567148316e-05,
      "loss": 0.2936,
      "step": 10332
    },
    {
      "epoch": 0.7820254481812592,
      "grad_norm": 0.75,
      "learning_rate": 1.3391458101953628e-05,
      "loss": 0.3452,
      "step": 10333
    },
    {
      "epoch": 0.7821011305047065,
      "grad_norm": 0.7109375,
      "learning_rate": 1.3390337588645283e-05,
      "loss": 0.3029,
      "step": 10334
    },
    {
      "epoch": 0.7821768128281539,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3389217027239182e-05,
      "loss": 0.3026,
      "step": 10335
    },
    {
      "epoch": 0.7822524951516011,
      "grad_norm": 0.72265625,
      "learning_rate": 1.3388096417751229e-05,
      "loss": 0.2928,
      "step": 10336
    },
    {
      "epoch": 0.7823281774750485,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3386975760197307e-05,
      "loss": 0.299,
      "step": 10337
    },
    {
      "epoch": 0.7824038597984958,
      "grad_norm": 0.87890625,
      "learning_rate": 1.3385855054593322e-05,
      "loss": 0.3601,
      "step": 10338
    },
    {
      "epoch": 0.7824795421219432,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3384734300955176e-05,
      "loss": 0.338,
      "step": 10339
    },
    {
      "epoch": 0.7825552244453905,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3383613499298762e-05,
      "loss": 0.2981,
      "step": 10340
    },
    {
      "epoch": 0.7826309067688378,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3382492649639987e-05,
      "loss": 0.2937,
      "step": 10341
    },
    {
      "epoch": 0.7827065890922852,
      "grad_norm": 0.77734375,
      "learning_rate": 1.3381371751994747e-05,
      "loss": 0.3079,
      "step": 10342
    },
    {
      "epoch": 0.7827822714157324,
      "grad_norm": 0.8046875,
      "learning_rate": 1.3380250806378947e-05,
      "loss": 0.3222,
      "step": 10343
    },
    {
      "epoch": 0.7828579537391798,
      "grad_norm": 0.78125,
      "learning_rate": 1.3379129812808488e-05,
      "loss": 0.3256,
      "step": 10344
    },
    {
      "epoch": 0.7829336360626271,
      "grad_norm": 0.75,
      "learning_rate": 1.3378008771299276e-05,
      "loss": 0.3059,
      "step": 10345
    },
    {
      "epoch": 0.7830093183860745,
      "grad_norm": 0.70703125,
      "learning_rate": 1.337688768186721e-05,
      "loss": 0.2729,
      "step": 10346
    },
    {
      "epoch": 0.7830850007095218,
      "grad_norm": 0.72265625,
      "learning_rate": 1.3375766544528197e-05,
      "loss": 0.3232,
      "step": 10347
    },
    {
      "epoch": 0.7831606830329692,
      "grad_norm": 0.79296875,
      "learning_rate": 1.3374645359298144e-05,
      "loss": 0.3489,
      "step": 10348
    },
    {
      "epoch": 0.7832363653564165,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3373524126192956e-05,
      "loss": 0.3038,
      "step": 10349
    },
    {
      "epoch": 0.7833120476798637,
      "grad_norm": 0.72265625,
      "learning_rate": 1.3372402845228538e-05,
      "loss": 0.2917,
      "step": 10350
    },
    {
      "epoch": 0.7833877300033111,
      "grad_norm": 0.73828125,
      "learning_rate": 1.33712815164208e-05,
      "loss": 0.3002,
      "step": 10351
    },
    {
      "epoch": 0.7834634123267584,
      "grad_norm": 0.76171875,
      "learning_rate": 1.337016013978565e-05,
      "loss": 0.3234,
      "step": 10352
    },
    {
      "epoch": 0.7835390946502058,
      "grad_norm": 0.70703125,
      "learning_rate": 1.336903871533899e-05,
      "loss": 0.266,
      "step": 10353
    },
    {
      "epoch": 0.7836147769736531,
      "grad_norm": 0.83203125,
      "learning_rate": 1.3367917243096736e-05,
      "loss": 0.312,
      "step": 10354
    },
    {
      "epoch": 0.7836904592971005,
      "grad_norm": 0.703125,
      "learning_rate": 1.3366795723074795e-05,
      "loss": 0.306,
      "step": 10355
    },
    {
      "epoch": 0.7837661416205477,
      "grad_norm": 0.7109375,
      "learning_rate": 1.3365674155289081e-05,
      "loss": 0.3108,
      "step": 10356
    },
    {
      "epoch": 0.783841823943995,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3364552539755504e-05,
      "loss": 0.3446,
      "step": 10357
    },
    {
      "epoch": 0.7839175062674424,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3363430876489975e-05,
      "loss": 0.3246,
      "step": 10358
    },
    {
      "epoch": 0.7839931885908897,
      "grad_norm": 0.78125,
      "learning_rate": 1.3362309165508407e-05,
      "loss": 0.3152,
      "step": 10359
    },
    {
      "epoch": 0.7840688709143371,
      "grad_norm": 0.734375,
      "learning_rate": 1.3361187406826714e-05,
      "loss": 0.3023,
      "step": 10360
    },
    {
      "epoch": 0.7841445532377844,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3360065600460808e-05,
      "loss": 0.3161,
      "step": 10361
    },
    {
      "epoch": 0.7842202355612318,
      "grad_norm": 0.79296875,
      "learning_rate": 1.3358943746426607e-05,
      "loss": 0.3501,
      "step": 10362
    },
    {
      "epoch": 0.784295917884679,
      "grad_norm": 0.8359375,
      "learning_rate": 1.3357821844740024e-05,
      "loss": 0.3712,
      "step": 10363
    },
    {
      "epoch": 0.7843716002081264,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3356699895416978e-05,
      "loss": 0.3299,
      "step": 10364
    },
    {
      "epoch": 0.7844472825315737,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3355577898473382e-05,
      "loss": 0.3194,
      "step": 10365
    },
    {
      "epoch": 0.784522964855021,
      "grad_norm": 0.72265625,
      "learning_rate": 1.3354455853925155e-05,
      "loss": 0.3024,
      "step": 10366
    },
    {
      "epoch": 0.7845986471784684,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3353333761788214e-05,
      "loss": 0.3322,
      "step": 10367
    },
    {
      "epoch": 0.7846743295019157,
      "grad_norm": 0.7421875,
      "learning_rate": 1.335221162207848e-05,
      "loss": 0.3131,
      "step": 10368
    },
    {
      "epoch": 0.7847500118253631,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3351089434811875e-05,
      "loss": 0.3283,
      "step": 10369
    },
    {
      "epoch": 0.7848256941488103,
      "grad_norm": 0.796875,
      "learning_rate": 1.334996720000431e-05,
      "loss": 0.3273,
      "step": 10370
    },
    {
      "epoch": 0.7849013764722577,
      "grad_norm": 0.72265625,
      "learning_rate": 1.3348844917671714e-05,
      "loss": 0.2957,
      "step": 10371
    },
    {
      "epoch": 0.784977058795705,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3347722587830008e-05,
      "loss": 0.3191,
      "step": 10372
    },
    {
      "epoch": 0.7850527411191524,
      "grad_norm": 0.734375,
      "learning_rate": 1.3346600210495111e-05,
      "loss": 0.3085,
      "step": 10373
    },
    {
      "epoch": 0.7851284234425997,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3345477785682945e-05,
      "loss": 0.3489,
      "step": 10374
    },
    {
      "epoch": 0.785204105766047,
      "grad_norm": 0.703125,
      "learning_rate": 1.3344355313409438e-05,
      "loss": 0.2837,
      "step": 10375
    },
    {
      "epoch": 0.7852797880894944,
      "grad_norm": 0.734375,
      "learning_rate": 1.334323279369051e-05,
      "loss": 0.3165,
      "step": 10376
    },
    {
      "epoch": 0.7853554704129416,
      "grad_norm": 0.8046875,
      "learning_rate": 1.3342110226542089e-05,
      "loss": 0.3588,
      "step": 10377
    },
    {
      "epoch": 0.785431152736389,
      "grad_norm": 0.734375,
      "learning_rate": 1.3340987611980101e-05,
      "loss": 0.3013,
      "step": 10378
    },
    {
      "epoch": 0.7855068350598363,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3339864950020465e-05,
      "loss": 0.3078,
      "step": 10379
    },
    {
      "epoch": 0.7855825173832837,
      "grad_norm": 0.78125,
      "learning_rate": 1.3338742240679116e-05,
      "loss": 0.3421,
      "step": 10380
    },
    {
      "epoch": 0.785658199706731,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3337619483971983e-05,
      "loss": 0.3398,
      "step": 10381
    },
    {
      "epoch": 0.7857338820301784,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3336496679914982e-05,
      "loss": 0.3152,
      "step": 10382
    },
    {
      "epoch": 0.7858095643536257,
      "grad_norm": 0.6875,
      "learning_rate": 1.3335373828524057e-05,
      "loss": 0.2719,
      "step": 10383
    },
    {
      "epoch": 0.7858852466770729,
      "grad_norm": 0.75,
      "learning_rate": 1.333425092981513e-05,
      "loss": 0.312,
      "step": 10384
    },
    {
      "epoch": 0.7859609290005203,
      "grad_norm": 0.79296875,
      "learning_rate": 1.3333127983804126e-05,
      "loss": 0.3432,
      "step": 10385
    },
    {
      "epoch": 0.7860366113239676,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3332004990506986e-05,
      "loss": 0.3204,
      "step": 10386
    },
    {
      "epoch": 0.786112293647415,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3330881949939638e-05,
      "loss": 0.2894,
      "step": 10387
    },
    {
      "epoch": 0.7861879759708623,
      "grad_norm": 0.84765625,
      "learning_rate": 1.3329758862118012e-05,
      "loss": 0.3827,
      "step": 10388
    },
    {
      "epoch": 0.7862636582943097,
      "grad_norm": 0.74609375,
      "learning_rate": 1.332863572705804e-05,
      "loss": 0.3357,
      "step": 10389
    },
    {
      "epoch": 0.786339340617757,
      "grad_norm": 0.765625,
      "learning_rate": 1.3327512544775662e-05,
      "loss": 0.3124,
      "step": 10390
    },
    {
      "epoch": 0.7864150229412042,
      "grad_norm": 0.77734375,
      "learning_rate": 1.3326389315286807e-05,
      "loss": 0.3204,
      "step": 10391
    },
    {
      "epoch": 0.7864907052646516,
      "grad_norm": 0.78125,
      "learning_rate": 1.3325266038607412e-05,
      "loss": 0.3142,
      "step": 10392
    },
    {
      "epoch": 0.7865663875880989,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3324142714753413e-05,
      "loss": 0.3193,
      "step": 10393
    },
    {
      "epoch": 0.7866420699115463,
      "grad_norm": 0.77734375,
      "learning_rate": 1.332301934374074e-05,
      "loss": 0.3185,
      "step": 10394
    },
    {
      "epoch": 0.7867177522349936,
      "grad_norm": 0.734375,
      "learning_rate": 1.332189592558534e-05,
      "loss": 0.3013,
      "step": 10395
    },
    {
      "epoch": 0.786793434558441,
      "grad_norm": 0.75,
      "learning_rate": 1.3320772460303145e-05,
      "loss": 0.3174,
      "step": 10396
    },
    {
      "epoch": 0.7868691168818883,
      "grad_norm": 0.71484375,
      "learning_rate": 1.3319648947910094e-05,
      "loss": 0.2681,
      "step": 10397
    },
    {
      "epoch": 0.7869447992053356,
      "grad_norm": 0.71875,
      "learning_rate": 1.3318525388422123e-05,
      "loss": 0.286,
      "step": 10398
    },
    {
      "epoch": 0.7870204815287829,
      "grad_norm": 0.7734375,
      "learning_rate": 1.331740178185518e-05,
      "loss": 0.3175,
      "step": 10399
    },
    {
      "epoch": 0.7870961638522302,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3316278128225192e-05,
      "loss": 0.3005,
      "step": 10400
    },
    {
      "epoch": 0.7871718461756776,
      "grad_norm": 0.8125,
      "learning_rate": 1.3315154427548117e-05,
      "loss": 0.3374,
      "step": 10401
    },
    {
      "epoch": 0.7872475284991249,
      "grad_norm": 0.765625,
      "learning_rate": 1.331403067983988e-05,
      "loss": 0.3065,
      "step": 10402
    },
    {
      "epoch": 0.7873232108225723,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3312906885116435e-05,
      "loss": 0.3069,
      "step": 10403
    },
    {
      "epoch": 0.7873988931460196,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3311783043393718e-05,
      "loss": 0.2975,
      "step": 10404
    },
    {
      "epoch": 0.7874745754694669,
      "grad_norm": 0.7109375,
      "learning_rate": 1.3310659154687677e-05,
      "loss": 0.3056,
      "step": 10405
    },
    {
      "epoch": 0.7875502577929142,
      "grad_norm": 0.78515625,
      "learning_rate": 1.3309535219014254e-05,
      "loss": 0.3651,
      "step": 10406
    },
    {
      "epoch": 0.7876259401163616,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3308411236389395e-05,
      "loss": 0.3032,
      "step": 10407
    },
    {
      "epoch": 0.7877016224398089,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3307287206829047e-05,
      "loss": 0.2805,
      "step": 10408
    },
    {
      "epoch": 0.7877773047632562,
      "grad_norm": 0.78515625,
      "learning_rate": 1.330616313034915e-05,
      "loss": 0.3388,
      "step": 10409
    },
    {
      "epoch": 0.7878529870867036,
      "grad_norm": 0.80078125,
      "learning_rate": 1.3305039006965657e-05,
      "loss": 0.3683,
      "step": 10410
    },
    {
      "epoch": 0.7879286694101509,
      "grad_norm": 0.72265625,
      "learning_rate": 1.3303914836694515e-05,
      "loss": 0.3213,
      "step": 10411
    },
    {
      "epoch": 0.7880043517335982,
      "grad_norm": 0.83203125,
      "learning_rate": 1.3302790619551673e-05,
      "loss": 0.3445,
      "step": 10412
    },
    {
      "epoch": 0.7880800340570455,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3301666355553076e-05,
      "loss": 0.3178,
      "step": 10413
    },
    {
      "epoch": 0.7881557163804929,
      "grad_norm": 0.9140625,
      "learning_rate": 1.3300542044714677e-05,
      "loss": 0.3323,
      "step": 10414
    },
    {
      "epoch": 0.7882313987039402,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3299417687052423e-05,
      "loss": 0.3437,
      "step": 10415
    },
    {
      "epoch": 0.7883070810273876,
      "grad_norm": 0.78125,
      "learning_rate": 1.329829328258227e-05,
      "loss": 0.315,
      "step": 10416
    },
    {
      "epoch": 0.7883827633508349,
      "grad_norm": 0.796875,
      "learning_rate": 1.3297168831320164e-05,
      "loss": 0.3475,
      "step": 10417
    },
    {
      "epoch": 0.7884584456742822,
      "grad_norm": 0.78125,
      "learning_rate": 1.329604433328206e-05,
      "loss": 0.333,
      "step": 10418
    },
    {
      "epoch": 0.7885341279977295,
      "grad_norm": 0.7109375,
      "learning_rate": 1.3294919788483915e-05,
      "loss": 0.2977,
      "step": 10419
    },
    {
      "epoch": 0.7886098103211768,
      "grad_norm": 0.796875,
      "learning_rate": 1.3293795196941677e-05,
      "loss": 0.3675,
      "step": 10420
    },
    {
      "epoch": 0.7886854926446242,
      "grad_norm": 0.671875,
      "learning_rate": 1.3292670558671299e-05,
      "loss": 0.2587,
      "step": 10421
    },
    {
      "epoch": 0.7887611749680715,
      "grad_norm": 0.8515625,
      "learning_rate": 1.329154587368874e-05,
      "loss": 0.3612,
      "step": 10422
    },
    {
      "epoch": 0.7888368572915189,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3290421142009956e-05,
      "loss": 0.3279,
      "step": 10423
    },
    {
      "epoch": 0.7889125396149662,
      "grad_norm": 0.73046875,
      "learning_rate": 1.32892963636509e-05,
      "loss": 0.2922,
      "step": 10424
    },
    {
      "epoch": 0.7889882219384136,
      "grad_norm": 0.765625,
      "learning_rate": 1.3288171538627533e-05,
      "loss": 0.3232,
      "step": 10425
    },
    {
      "epoch": 0.7890639042618608,
      "grad_norm": 0.71484375,
      "learning_rate": 1.3287046666955805e-05,
      "loss": 0.2965,
      "step": 10426
    },
    {
      "epoch": 0.7891395865853081,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3285921748651682e-05,
      "loss": 0.306,
      "step": 10427
    },
    {
      "epoch": 0.7892152689087555,
      "grad_norm": 0.69140625,
      "learning_rate": 1.328479678373112e-05,
      "loss": 0.2897,
      "step": 10428
    },
    {
      "epoch": 0.7892909512322028,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3283671772210079e-05,
      "loss": 0.3171,
      "step": 10429
    },
    {
      "epoch": 0.7893666335556502,
      "grad_norm": 0.7265625,
      "learning_rate": 1.328254671410452e-05,
      "loss": 0.288,
      "step": 10430
    },
    {
      "epoch": 0.7894423158790975,
      "grad_norm": 0.72265625,
      "learning_rate": 1.32814216094304e-05,
      "loss": 0.2936,
      "step": 10431
    },
    {
      "epoch": 0.7895179982025449,
      "grad_norm": 0.796875,
      "learning_rate": 1.3280296458203684e-05,
      "loss": 0.3673,
      "step": 10432
    },
    {
      "epoch": 0.7895936805259921,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3279171260440336e-05,
      "loss": 0.2988,
      "step": 10433
    },
    {
      "epoch": 0.7896693628494394,
      "grad_norm": 0.6953125,
      "learning_rate": 1.3278046016156316e-05,
      "loss": 0.2845,
      "step": 10434
    },
    {
      "epoch": 0.7897450451728868,
      "grad_norm": 0.94140625,
      "learning_rate": 1.3276920725367586e-05,
      "loss": 0.3247,
      "step": 10435
    },
    {
      "epoch": 0.7898207274963341,
      "grad_norm": 0.7109375,
      "learning_rate": 1.3275795388090114e-05,
      "loss": 0.2835,
      "step": 10436
    },
    {
      "epoch": 0.7898964098197815,
      "grad_norm": 0.74609375,
      "learning_rate": 1.327467000433986e-05,
      "loss": 0.325,
      "step": 10437
    },
    {
      "epoch": 0.7899720921432288,
      "grad_norm": 0.75,
      "learning_rate": 1.3273544574132792e-05,
      "loss": 0.2881,
      "step": 10438
    },
    {
      "epoch": 0.7900477744666762,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3272419097484883e-05,
      "loss": 0.3073,
      "step": 10439
    },
    {
      "epoch": 0.7901234567901234,
      "grad_norm": 0.7265625,
      "learning_rate": 1.327129357441209e-05,
      "loss": 0.2973,
      "step": 10440
    },
    {
      "epoch": 0.7901991391135708,
      "grad_norm": 0.80078125,
      "learning_rate": 1.3270168004930382e-05,
      "loss": 0.3099,
      "step": 10441
    },
    {
      "epoch": 0.7902748214370181,
      "grad_norm": 0.88671875,
      "learning_rate": 1.326904238905573e-05,
      "loss": 0.3438,
      "step": 10442
    },
    {
      "epoch": 0.7903505037604655,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3267916726804102e-05,
      "loss": 0.3219,
      "step": 10443
    },
    {
      "epoch": 0.7904261860839128,
      "grad_norm": 0.73828125,
      "learning_rate": 1.326679101819147e-05,
      "loss": 0.3101,
      "step": 10444
    },
    {
      "epoch": 0.7905018684073601,
      "grad_norm": 0.77734375,
      "learning_rate": 1.3265665263233798e-05,
      "loss": 0.2841,
      "step": 10445
    },
    {
      "epoch": 0.7905775507308075,
      "grad_norm": 0.77734375,
      "learning_rate": 1.3264539461947061e-05,
      "loss": 0.2983,
      "step": 10446
    },
    {
      "epoch": 0.7906532330542547,
      "grad_norm": 0.78125,
      "learning_rate": 1.3263413614347226e-05,
      "loss": 0.329,
      "step": 10447
    },
    {
      "epoch": 0.7907289153777021,
      "grad_norm": 0.6875,
      "learning_rate": 1.3262287720450275e-05,
      "loss": 0.2704,
      "step": 10448
    },
    {
      "epoch": 0.7908045977011494,
      "grad_norm": 0.69921875,
      "learning_rate": 1.3261161780272173e-05,
      "loss": 0.27,
      "step": 10449
    },
    {
      "epoch": 0.7908802800245968,
      "grad_norm": 0.77734375,
      "learning_rate": 1.3260035793828891e-05,
      "loss": 0.328,
      "step": 10450
    },
    {
      "epoch": 0.7909559623480441,
      "grad_norm": 0.80859375,
      "learning_rate": 1.325890976113641e-05,
      "loss": 0.3243,
      "step": 10451
    },
    {
      "epoch": 0.7910316446714915,
      "grad_norm": 0.66796875,
      "learning_rate": 1.3257783682210704e-05,
      "loss": 0.2561,
      "step": 10452
    },
    {
      "epoch": 0.7911073269949388,
      "grad_norm": 1.1484375,
      "learning_rate": 1.325665755706774e-05,
      "loss": 0.3951,
      "step": 10453
    },
    {
      "epoch": 0.791183009318386,
      "grad_norm": 0.6796875,
      "learning_rate": 1.3255531385723504e-05,
      "loss": 0.2579,
      "step": 10454
    },
    {
      "epoch": 0.7912586916418334,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3254405168193969e-05,
      "loss": 0.2703,
      "step": 10455
    },
    {
      "epoch": 0.7913343739652807,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3253278904495109e-05,
      "loss": 0.3552,
      "step": 10456
    },
    {
      "epoch": 0.7914100562887281,
      "grad_norm": 0.703125,
      "learning_rate": 1.3252152594642906e-05,
      "loss": 0.2577,
      "step": 10457
    },
    {
      "epoch": 0.7914857386121754,
      "grad_norm": 0.765625,
      "learning_rate": 1.3251026238653337e-05,
      "loss": 0.2802,
      "step": 10458
    },
    {
      "epoch": 0.7915614209356228,
      "grad_norm": 0.78125,
      "learning_rate": 1.3249899836542384e-05,
      "loss": 0.3168,
      "step": 10459
    },
    {
      "epoch": 0.7916371032590701,
      "grad_norm": 0.77734375,
      "learning_rate": 1.3248773388326024e-05,
      "loss": 0.3194,
      "step": 10460
    },
    {
      "epoch": 0.7917127855825173,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3247646894020236e-05,
      "loss": 0.3294,
      "step": 10461
    },
    {
      "epoch": 0.7917884679059647,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3246520353641008e-05,
      "loss": 0.2913,
      "step": 10462
    },
    {
      "epoch": 0.791864150229412,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3245393767204315e-05,
      "loss": 0.3286,
      "step": 10463
    },
    {
      "epoch": 0.7919398325528594,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3244267134726141e-05,
      "loss": 0.2746,
      "step": 10464
    },
    {
      "epoch": 0.7920155148763067,
      "grad_norm": 0.71484375,
      "learning_rate": 1.3243140456222472e-05,
      "loss": 0.2852,
      "step": 10465
    },
    {
      "epoch": 0.7920911971997541,
      "grad_norm": 0.7265625,
      "learning_rate": 1.324201373170929e-05,
      "loss": 0.3056,
      "step": 10466
    },
    {
      "epoch": 0.7921668795232014,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3240886961202579e-05,
      "loss": 0.3282,
      "step": 10467
    },
    {
      "epoch": 0.7922425618466487,
      "grad_norm": 0.69140625,
      "learning_rate": 1.3239760144718326e-05,
      "loss": 0.2578,
      "step": 10468
    },
    {
      "epoch": 0.792318244170096,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3238633282272514e-05,
      "loss": 0.2856,
      "step": 10469
    },
    {
      "epoch": 0.7923939264935433,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3237506373881135e-05,
      "loss": 0.2548,
      "step": 10470
    },
    {
      "epoch": 0.7924696088169907,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3236379419560166e-05,
      "loss": 0.3324,
      "step": 10471
    },
    {
      "epoch": 0.792545291140438,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3235252419325608e-05,
      "loss": 0.3274,
      "step": 10472
    },
    {
      "epoch": 0.7926209734638854,
      "grad_norm": 0.69921875,
      "learning_rate": 1.3234125373193437e-05,
      "loss": 0.2998,
      "step": 10473
    },
    {
      "epoch": 0.7926966557873326,
      "grad_norm": 0.80859375,
      "learning_rate": 1.3232998281179649e-05,
      "loss": 0.3461,
      "step": 10474
    },
    {
      "epoch": 0.79277233811078,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3231871143300236e-05,
      "loss": 0.3108,
      "step": 10475
    },
    {
      "epoch": 0.7928480204342273,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3230743959571182e-05,
      "loss": 0.2959,
      "step": 10476
    },
    {
      "epoch": 0.7929237027576747,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3229616730008481e-05,
      "loss": 0.2682,
      "step": 10477
    },
    {
      "epoch": 0.792999385081122,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3228489454628126e-05,
      "loss": 0.3277,
      "step": 10478
    },
    {
      "epoch": 0.7930750674045693,
      "grad_norm": 0.734375,
      "learning_rate": 1.3227362133446105e-05,
      "loss": 0.299,
      "step": 10479
    },
    {
      "epoch": 0.7931507497280167,
      "grad_norm": 0.71484375,
      "learning_rate": 1.3226234766478413e-05,
      "loss": 0.2964,
      "step": 10480
    },
    {
      "epoch": 0.7932264320514639,
      "grad_norm": 0.73046875,
      "learning_rate": 1.322510735374105e-05,
      "loss": 0.3215,
      "step": 10481
    },
    {
      "epoch": 0.7933021143749113,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3223979895249998e-05,
      "loss": 0.2989,
      "step": 10482
    },
    {
      "epoch": 0.7933777966983586,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3222852391021261e-05,
      "loss": 0.3318,
      "step": 10483
    },
    {
      "epoch": 0.793453479021806,
      "grad_norm": 0.734375,
      "learning_rate": 1.3221724841070835e-05,
      "loss": 0.3098,
      "step": 10484
    },
    {
      "epoch": 0.7935291613452533,
      "grad_norm": 0.7734375,
      "learning_rate": 1.322059724541471e-05,
      "loss": 0.3277,
      "step": 10485
    },
    {
      "epoch": 0.7936048436687007,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3219469604068888e-05,
      "loss": 0.325,
      "step": 10486
    },
    {
      "epoch": 0.793680525992148,
      "grad_norm": 0.81640625,
      "learning_rate": 1.3218341917049367e-05,
      "loss": 0.3423,
      "step": 10487
    },
    {
      "epoch": 0.7937562083155952,
      "grad_norm": 0.78515625,
      "learning_rate": 1.3217214184372138e-05,
      "loss": 0.3408,
      "step": 10488
    },
    {
      "epoch": 0.7938318906390426,
      "grad_norm": 0.8125,
      "learning_rate": 1.3216086406053208e-05,
      "loss": 0.2876,
      "step": 10489
    },
    {
      "epoch": 0.7939075729624899,
      "grad_norm": 0.6875,
      "learning_rate": 1.3214958582108576e-05,
      "loss": 0.2642,
      "step": 10490
    },
    {
      "epoch": 0.7939832552859373,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3213830712554235e-05,
      "loss": 0.3041,
      "step": 10491
    },
    {
      "epoch": 0.7940589376093846,
      "grad_norm": 0.7890625,
      "learning_rate": 1.3212702797406192e-05,
      "loss": 0.3501,
      "step": 10492
    },
    {
      "epoch": 0.794134619932832,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3211574836680446e-05,
      "loss": 0.3002,
      "step": 10493
    },
    {
      "epoch": 0.7942103022562793,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3210446830392997e-05,
      "loss": 0.3126,
      "step": 10494
    },
    {
      "epoch": 0.7942859845797265,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3209318778559856e-05,
      "loss": 0.3076,
      "step": 10495
    },
    {
      "epoch": 0.7943616669031739,
      "grad_norm": 0.76953125,
      "learning_rate": 1.320819068119702e-05,
      "loss": 0.3034,
      "step": 10496
    },
    {
      "epoch": 0.7944373492266212,
      "grad_norm": 0.78125,
      "learning_rate": 1.3207062538320487e-05,
      "loss": 0.2931,
      "step": 10497
    },
    {
      "epoch": 0.7945130315500686,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3205934349946275e-05,
      "loss": 0.3134,
      "step": 10498
    },
    {
      "epoch": 0.7945887138735159,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3204806116090381e-05,
      "loss": 0.3332,
      "step": 10499
    },
    {
      "epoch": 0.7946643961969633,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3203677836768813e-05,
      "loss": 0.3301,
      "step": 10500
    },
    {
      "epoch": 0.7947400785204106,
      "grad_norm": 0.75,
      "learning_rate": 1.3202549511997578e-05,
      "loss": 0.329,
      "step": 10501
    },
    {
      "epoch": 0.7948157608438579,
      "grad_norm": 0.71875,
      "learning_rate": 1.3201421141792681e-05,
      "loss": 0.3128,
      "step": 10502
    },
    {
      "epoch": 0.7948914431673052,
      "grad_norm": 0.71875,
      "learning_rate": 1.320029272617013e-05,
      "loss": 0.2793,
      "step": 10503
    },
    {
      "epoch": 0.7949671254907525,
      "grad_norm": 0.83984375,
      "learning_rate": 1.3199164265145936e-05,
      "loss": 0.3658,
      "step": 10504
    },
    {
      "epoch": 0.7950428078141999,
      "grad_norm": 0.75,
      "learning_rate": 1.3198035758736109e-05,
      "loss": 0.3304,
      "step": 10505
    },
    {
      "epoch": 0.7951184901376472,
      "grad_norm": 0.80078125,
      "learning_rate": 1.3196907206956652e-05,
      "loss": 0.3187,
      "step": 10506
    },
    {
      "epoch": 0.7951941724610946,
      "grad_norm": 0.71484375,
      "learning_rate": 1.3195778609823584e-05,
      "loss": 0.2678,
      "step": 10507
    },
    {
      "epoch": 0.7952698547845419,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3194649967352913e-05,
      "loss": 0.3036,
      "step": 10508
    },
    {
      "epoch": 0.7953455371079892,
      "grad_norm": 0.734375,
      "learning_rate": 1.3193521279560648e-05,
      "loss": 0.3061,
      "step": 10509
    },
    {
      "epoch": 0.7954212194314365,
      "grad_norm": 0.72265625,
      "learning_rate": 1.3192392546462802e-05,
      "loss": 0.288,
      "step": 10510
    },
    {
      "epoch": 0.7954969017548839,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3191263768075393e-05,
      "loss": 0.2935,
      "step": 10511
    },
    {
      "epoch": 0.7955725840783312,
      "grad_norm": 0.72265625,
      "learning_rate": 1.3190134944414425e-05,
      "loss": 0.2982,
      "step": 10512
    },
    {
      "epoch": 0.7956482664017785,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3189006075495925e-05,
      "loss": 0.3314,
      "step": 10513
    },
    {
      "epoch": 0.7957239487252259,
      "grad_norm": 0.734375,
      "learning_rate": 1.3187877161335899e-05,
      "loss": 0.3144,
      "step": 10514
    },
    {
      "epoch": 0.7957996310486732,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3186748201950363e-05,
      "loss": 0.3402,
      "step": 10515
    },
    {
      "epoch": 0.7958753133721205,
      "grad_norm": 0.765625,
      "learning_rate": 1.318561919735534e-05,
      "loss": 0.3135,
      "step": 10516
    },
    {
      "epoch": 0.7959509956955678,
      "grad_norm": 0.8125,
      "learning_rate": 1.3184490147566837e-05,
      "loss": 0.352,
      "step": 10517
    },
    {
      "epoch": 0.7960266780190152,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3183361052600878e-05,
      "loss": 0.3055,
      "step": 10518
    },
    {
      "epoch": 0.7961023603424625,
      "grad_norm": 0.75390625,
      "learning_rate": 1.318223191247348e-05,
      "loss": 0.2992,
      "step": 10519
    },
    {
      "epoch": 0.7961780426659099,
      "grad_norm": 0.79296875,
      "learning_rate": 1.318110272720066e-05,
      "loss": 0.3439,
      "step": 10520
    },
    {
      "epoch": 0.7962537249893572,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3179973496798439e-05,
      "loss": 0.3016,
      "step": 10521
    },
    {
      "epoch": 0.7963294073128045,
      "grad_norm": 0.78515625,
      "learning_rate": 1.317884422128284e-05,
      "loss": 0.3452,
      "step": 10522
    },
    {
      "epoch": 0.7964050896362518,
      "grad_norm": 0.7265625,
      "learning_rate": 1.317771490066988e-05,
      "loss": 0.3184,
      "step": 10523
    },
    {
      "epoch": 0.7964807719596991,
      "grad_norm": 0.75,
      "learning_rate": 1.3176585534975578e-05,
      "loss": 0.2899,
      "step": 10524
    },
    {
      "epoch": 0.7965564542831465,
      "grad_norm": 0.79296875,
      "learning_rate": 1.3175456124215961e-05,
      "loss": 0.3061,
      "step": 10525
    },
    {
      "epoch": 0.7966321366065938,
      "grad_norm": 0.80078125,
      "learning_rate": 1.3174326668407051e-05,
      "loss": 0.3194,
      "step": 10526
    },
    {
      "epoch": 0.7967078189300412,
      "grad_norm": 0.78125,
      "learning_rate": 1.3173197167564867e-05,
      "loss": 0.3278,
      "step": 10527
    },
    {
      "epoch": 0.7967835012534885,
      "grad_norm": 0.796875,
      "learning_rate": 1.3172067621705438e-05,
      "loss": 0.3533,
      "step": 10528
    },
    {
      "epoch": 0.7968591835769359,
      "grad_norm": 0.71875,
      "learning_rate": 1.3170938030844785e-05,
      "loss": 0.2844,
      "step": 10529
    },
    {
      "epoch": 0.7969348659003831,
      "grad_norm": 0.7109375,
      "learning_rate": 1.3169808394998933e-05,
      "loss": 0.2957,
      "step": 10530
    },
    {
      "epoch": 0.7970105482238304,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3168678714183911e-05,
      "loss": 0.2999,
      "step": 10531
    },
    {
      "epoch": 0.7970862305472778,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3167548988415747e-05,
      "loss": 0.3274,
      "step": 10532
    },
    {
      "epoch": 0.7971619128707251,
      "grad_norm": 0.9921875,
      "learning_rate": 1.316641921771046e-05,
      "loss": 0.3363,
      "step": 10533
    },
    {
      "epoch": 0.7972375951941725,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3165289402084084e-05,
      "loss": 0.3208,
      "step": 10534
    },
    {
      "epoch": 0.7973132775176198,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3164159541552646e-05,
      "loss": 0.3042,
      "step": 10535
    },
    {
      "epoch": 0.7973889598410672,
      "grad_norm": 0.734375,
      "learning_rate": 1.3163029636132176e-05,
      "loss": 0.2759,
      "step": 10536
    },
    {
      "epoch": 0.7974646421645144,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3161899685838704e-05,
      "loss": 0.292,
      "step": 10537
    },
    {
      "epoch": 0.7975403244879617,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3160769690688256e-05,
      "loss": 0.2867,
      "step": 10538
    },
    {
      "epoch": 0.7976160068114091,
      "grad_norm": 0.91796875,
      "learning_rate": 1.3159639650696865e-05,
      "loss": 0.3022,
      "step": 10539
    },
    {
      "epoch": 0.7976916891348564,
      "grad_norm": 0.70703125,
      "learning_rate": 1.3158509565880566e-05,
      "loss": 0.2772,
      "step": 10540
    },
    {
      "epoch": 0.7977673714583038,
      "grad_norm": 0.72265625,
      "learning_rate": 1.3157379436255386e-05,
      "loss": 0.272,
      "step": 10541
    },
    {
      "epoch": 0.7978430537817511,
      "grad_norm": 0.83203125,
      "learning_rate": 1.3156249261837366e-05,
      "loss": 0.311,
      "step": 10542
    },
    {
      "epoch": 0.7979187361051985,
      "grad_norm": 0.75,
      "learning_rate": 1.315511904264253e-05,
      "loss": 0.3048,
      "step": 10543
    },
    {
      "epoch": 0.7979944184286457,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3153988778686914e-05,
      "loss": 0.317,
      "step": 10544
    },
    {
      "epoch": 0.7980701007520931,
      "grad_norm": 0.6953125,
      "learning_rate": 1.3152858469986559e-05,
      "loss": 0.243,
      "step": 10545
    },
    {
      "epoch": 0.7981457830755404,
      "grad_norm": 0.72265625,
      "learning_rate": 1.3151728116557496e-05,
      "loss": 0.3054,
      "step": 10546
    },
    {
      "epoch": 0.7982214653989878,
      "grad_norm": 0.8125,
      "learning_rate": 1.3150597718415759e-05,
      "loss": 0.3351,
      "step": 10547
    },
    {
      "epoch": 0.7982971477224351,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3149467275577385e-05,
      "loss": 0.3011,
      "step": 10548
    },
    {
      "epoch": 0.7983728300458824,
      "grad_norm": 0.86328125,
      "learning_rate": 1.3148336788058415e-05,
      "loss": 0.3548,
      "step": 10549
    },
    {
      "epoch": 0.7984485123693298,
      "grad_norm": 0.8125,
      "learning_rate": 1.3147206255874886e-05,
      "loss": 0.3516,
      "step": 10550
    },
    {
      "epoch": 0.798524194692777,
      "grad_norm": 0.7890625,
      "learning_rate": 1.3146075679042835e-05,
      "loss": 0.3615,
      "step": 10551
    },
    {
      "epoch": 0.7985998770162244,
      "grad_norm": 0.79296875,
      "learning_rate": 1.3144945057578303e-05,
      "loss": 0.3351,
      "step": 10552
    },
    {
      "epoch": 0.7986755593396717,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3143814391497323e-05,
      "loss": 0.3059,
      "step": 10553
    },
    {
      "epoch": 0.7987512416631191,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3142683680815944e-05,
      "loss": 0.2827,
      "step": 10554
    },
    {
      "epoch": 0.7988269239865664,
      "grad_norm": 0.78515625,
      "learning_rate": 1.3141552925550205e-05,
      "loss": 0.3541,
      "step": 10555
    },
    {
      "epoch": 0.7989026063100138,
      "grad_norm": 0.8046875,
      "learning_rate": 1.3140422125716148e-05,
      "loss": 0.3405,
      "step": 10556
    },
    {
      "epoch": 0.7989782886334611,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3139291281329812e-05,
      "loss": 0.3306,
      "step": 10557
    },
    {
      "epoch": 0.7990539709569083,
      "grad_norm": 0.7109375,
      "learning_rate": 1.3138160392407242e-05,
      "loss": 0.2854,
      "step": 10558
    },
    {
      "epoch": 0.7991296532803557,
      "grad_norm": 0.71875,
      "learning_rate": 1.3137029458964481e-05,
      "loss": 0.2952,
      "step": 10559
    },
    {
      "epoch": 0.799205335603803,
      "grad_norm": 0.859375,
      "learning_rate": 1.3135898481017575e-05,
      "loss": 0.4094,
      "step": 10560
    },
    {
      "epoch": 0.7992810179272504,
      "grad_norm": 0.8046875,
      "learning_rate": 1.3134767458582567e-05,
      "loss": 0.3478,
      "step": 10561
    },
    {
      "epoch": 0.7993567002506977,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3133636391675501e-05,
      "loss": 0.3164,
      "step": 10562
    },
    {
      "epoch": 0.7994323825741451,
      "grad_norm": 0.765625,
      "learning_rate": 1.3132505280312429e-05,
      "loss": 0.3102,
      "step": 10563
    },
    {
      "epoch": 0.7995080648975924,
      "grad_norm": 0.8671875,
      "learning_rate": 1.3131374124509393e-05,
      "loss": 0.3689,
      "step": 10564
    },
    {
      "epoch": 0.7995837472210396,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3130242924282442e-05,
      "loss": 0.3247,
      "step": 10565
    },
    {
      "epoch": 0.799659429544487,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3129111679647622e-05,
      "loss": 0.3192,
      "step": 10566
    },
    {
      "epoch": 0.7997351118679343,
      "grad_norm": 0.80859375,
      "learning_rate": 1.3127980390620984e-05,
      "loss": 0.3767,
      "step": 10567
    },
    {
      "epoch": 0.7998107941913817,
      "grad_norm": 0.7890625,
      "learning_rate": 1.3126849057218573e-05,
      "loss": 0.3212,
      "step": 10568
    },
    {
      "epoch": 0.799886476514829,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3125717679456447e-05,
      "loss": 0.311,
      "step": 10569
    },
    {
      "epoch": 0.7999621588382764,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3124586257350649e-05,
      "loss": 0.3137,
      "step": 10570
    },
    {
      "epoch": 0.8000378411617237,
      "grad_norm": 0.765625,
      "learning_rate": 1.3123454790917235e-05,
      "loss": 0.3128,
      "step": 10571
    },
    {
      "epoch": 0.800113523485171,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3122323280172254e-05,
      "loss": 0.3081,
      "step": 10572
    },
    {
      "epoch": 0.8001892058086183,
      "grad_norm": 0.80859375,
      "learning_rate": 1.312119172513176e-05,
      "loss": 0.3598,
      "step": 10573
    },
    {
      "epoch": 0.8002648881320656,
      "grad_norm": 0.90625,
      "learning_rate": 1.3120060125811802e-05,
      "loss": 0.3386,
      "step": 10574
    },
    {
      "epoch": 0.800340570455513,
      "grad_norm": 0.68359375,
      "learning_rate": 1.311892848222844e-05,
      "loss": 0.2798,
      "step": 10575
    },
    {
      "epoch": 0.8004162527789603,
      "grad_norm": 0.703125,
      "learning_rate": 1.3117796794397721e-05,
      "loss": 0.2808,
      "step": 10576
    },
    {
      "epoch": 0.8004162527789603,
      "eval_loss": 0.33043158054351807,
      "eval_runtime": 83.4562,
      "eval_samples_per_second": 58.258,
      "eval_steps_per_second": 58.258,
      "step": 10576
    },
    {
      "epoch": 0.8004919351024077,
      "grad_norm": 0.7265625,
      "learning_rate": 1.311666506233571e-05,
      "loss": 0.3104,
      "step": 10577
    },
    {
      "epoch": 0.800567617425855,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3115533286058453e-05,
      "loss": 0.324,
      "step": 10578
    },
    {
      "epoch": 0.8006432997493023,
      "grad_norm": 0.8203125,
      "learning_rate": 1.3114401465582012e-05,
      "loss": 0.3713,
      "step": 10579
    },
    {
      "epoch": 0.8007189820727496,
      "grad_norm": 0.7890625,
      "learning_rate": 1.3113269600922441e-05,
      "loss": 0.3304,
      "step": 10580
    },
    {
      "epoch": 0.800794664396197,
      "grad_norm": 1.0546875,
      "learning_rate": 1.3112137692095797e-05,
      "loss": 0.3934,
      "step": 10581
    },
    {
      "epoch": 0.8008703467196443,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3111005739118142e-05,
      "loss": 0.3232,
      "step": 10582
    },
    {
      "epoch": 0.8009460290430916,
      "grad_norm": 0.703125,
      "learning_rate": 1.3109873742005528e-05,
      "loss": 0.2872,
      "step": 10583
    },
    {
      "epoch": 0.801021711366539,
      "grad_norm": 0.765625,
      "learning_rate": 1.3108741700774021e-05,
      "loss": 0.3194,
      "step": 10584
    },
    {
      "epoch": 0.8010973936899863,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3107609615439677e-05,
      "loss": 0.3012,
      "step": 10585
    },
    {
      "epoch": 0.8011730760134336,
      "grad_norm": 0.88671875,
      "learning_rate": 1.3106477486018556e-05,
      "loss": 0.3104,
      "step": 10586
    },
    {
      "epoch": 0.8012487583368809,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3105345312526725e-05,
      "loss": 0.3093,
      "step": 10587
    },
    {
      "epoch": 0.8013244406603283,
      "grad_norm": 0.796875,
      "learning_rate": 1.310421309498024e-05,
      "loss": 0.3335,
      "step": 10588
    },
    {
      "epoch": 0.8014001229837756,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3103080833395165e-05,
      "loss": 0.3158,
      "step": 10589
    },
    {
      "epoch": 0.801475805307223,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3101948527787565e-05,
      "loss": 0.3132,
      "step": 10590
    },
    {
      "epoch": 0.8015514876306703,
      "grad_norm": 0.8203125,
      "learning_rate": 1.3100816178173496e-05,
      "loss": 0.2856,
      "step": 10591
    },
    {
      "epoch": 0.8016271699541176,
      "grad_norm": 0.734375,
      "learning_rate": 1.3099683784569036e-05,
      "loss": 0.2727,
      "step": 10592
    },
    {
      "epoch": 0.8017028522775649,
      "grad_norm": 0.734375,
      "learning_rate": 1.309855134699024e-05,
      "loss": 0.3016,
      "step": 10593
    },
    {
      "epoch": 0.8017785346010122,
      "grad_norm": 0.8203125,
      "learning_rate": 1.3097418865453173e-05,
      "loss": 0.2891,
      "step": 10594
    },
    {
      "epoch": 0.8018542169244596,
      "grad_norm": 0.84765625,
      "learning_rate": 1.3096286339973907e-05,
      "loss": 0.3656,
      "step": 10595
    },
    {
      "epoch": 0.8019298992479069,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3095153770568504e-05,
      "loss": 0.329,
      "step": 10596
    },
    {
      "epoch": 0.8020055815713543,
      "grad_norm": 0.75,
      "learning_rate": 1.3094021157253032e-05,
      "loss": 0.3012,
      "step": 10597
    },
    {
      "epoch": 0.8020812638948016,
      "grad_norm": 0.84375,
      "learning_rate": 1.3092888500043566e-05,
      "loss": 0.3796,
      "step": 10598
    },
    {
      "epoch": 0.8021569462182488,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3091755798956164e-05,
      "loss": 0.314,
      "step": 10599
    },
    {
      "epoch": 0.8022326285416962,
      "grad_norm": 1.1015625,
      "learning_rate": 1.3090623054006901e-05,
      "loss": 0.3432,
      "step": 10600
    },
    {
      "epoch": 0.8023083108651435,
      "grad_norm": 0.68359375,
      "learning_rate": 1.3089490265211848e-05,
      "loss": 0.2847,
      "step": 10601
    },
    {
      "epoch": 0.8023839931885909,
      "grad_norm": 0.78125,
      "learning_rate": 1.3088357432587076e-05,
      "loss": 0.324,
      "step": 10602
    },
    {
      "epoch": 0.8024596755120382,
      "grad_norm": 0.8984375,
      "learning_rate": 1.3087224556148648e-05,
      "loss": 0.3243,
      "step": 10603
    },
    {
      "epoch": 0.8025353578354856,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3086091635912645e-05,
      "loss": 0.2865,
      "step": 10604
    },
    {
      "epoch": 0.8026110401589329,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3084958671895136e-05,
      "loss": 0.3168,
      "step": 10605
    },
    {
      "epoch": 0.8026867224823802,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3083825664112193e-05,
      "loss": 0.306,
      "step": 10606
    },
    {
      "epoch": 0.8027624048058275,
      "grad_norm": 0.71484375,
      "learning_rate": 1.3082692612579895e-05,
      "loss": 0.2768,
      "step": 10607
    },
    {
      "epoch": 0.8028380871292748,
      "grad_norm": 0.78125,
      "learning_rate": 1.3081559517314308e-05,
      "loss": 0.3306,
      "step": 10608
    },
    {
      "epoch": 0.8029137694527222,
      "grad_norm": 0.7265625,
      "learning_rate": 1.3080426378331512e-05,
      "loss": 0.2788,
      "step": 10609
    },
    {
      "epoch": 0.8029894517761695,
      "grad_norm": 0.79296875,
      "learning_rate": 1.3079293195647582e-05,
      "loss": 0.3605,
      "step": 10610
    },
    {
      "epoch": 0.8030651340996169,
      "grad_norm": 0.77734375,
      "learning_rate": 1.3078159969278595e-05,
      "loss": 0.2779,
      "step": 10611
    },
    {
      "epoch": 0.8031408164230642,
      "grad_norm": 3.25,
      "learning_rate": 1.3077026699240624e-05,
      "loss": 0.3627,
      "step": 10612
    },
    {
      "epoch": 0.8032164987465115,
      "grad_norm": 0.7890625,
      "learning_rate": 1.3075893385549752e-05,
      "loss": 0.3761,
      "step": 10613
    },
    {
      "epoch": 0.8032921810699588,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3074760028222053e-05,
      "loss": 0.3134,
      "step": 10614
    },
    {
      "epoch": 0.8033678633934062,
      "grad_norm": 0.828125,
      "learning_rate": 1.3073626627273604e-05,
      "loss": 0.3655,
      "step": 10615
    },
    {
      "epoch": 0.8034435457168535,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3072493182720492e-05,
      "loss": 0.3056,
      "step": 10616
    },
    {
      "epoch": 0.8035192280403008,
      "grad_norm": 0.8671875,
      "learning_rate": 1.3071359694578787e-05,
      "loss": 0.3313,
      "step": 10617
    },
    {
      "epoch": 0.8035949103637482,
      "grad_norm": 0.671875,
      "learning_rate": 1.3070226162864575e-05,
      "loss": 0.2484,
      "step": 10618
    },
    {
      "epoch": 0.8036705926871955,
      "grad_norm": 0.6953125,
      "learning_rate": 1.3069092587593936e-05,
      "loss": 0.2841,
      "step": 10619
    },
    {
      "epoch": 0.8037462750106428,
      "grad_norm": 0.81640625,
      "learning_rate": 1.3067958968782952e-05,
      "loss": 0.3448,
      "step": 10620
    },
    {
      "epoch": 0.8038219573340901,
      "grad_norm": 0.8125,
      "learning_rate": 1.3066825306447706e-05,
      "loss": 0.3343,
      "step": 10621
    },
    {
      "epoch": 0.8038976396575375,
      "grad_norm": 0.76171875,
      "learning_rate": 1.306569160060428e-05,
      "loss": 0.332,
      "step": 10622
    },
    {
      "epoch": 0.8039733219809848,
      "grad_norm": 0.765625,
      "learning_rate": 1.3064557851268758e-05,
      "loss": 0.3316,
      "step": 10623
    },
    {
      "epoch": 0.8040490043044322,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3063424058457222e-05,
      "loss": 0.2999,
      "step": 10624
    },
    {
      "epoch": 0.8041246866278795,
      "grad_norm": 0.734375,
      "learning_rate": 1.3062290222185761e-05,
      "loss": 0.3178,
      "step": 10625
    },
    {
      "epoch": 0.8042003689513268,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3061156342470458e-05,
      "loss": 0.3188,
      "step": 10626
    },
    {
      "epoch": 0.8042760512747741,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3060022419327398e-05,
      "loss": 0.3159,
      "step": 10627
    },
    {
      "epoch": 0.8043517335982214,
      "grad_norm": 0.69921875,
      "learning_rate": 1.3058888452772668e-05,
      "loss": 0.2919,
      "step": 10628
    },
    {
      "epoch": 0.8044274159216688,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3057754442822358e-05,
      "loss": 0.2918,
      "step": 10629
    },
    {
      "epoch": 0.8045030982451161,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3056620389492554e-05,
      "loss": 0.3411,
      "step": 10630
    },
    {
      "epoch": 0.8045787805685635,
      "grad_norm": 0.765625,
      "learning_rate": 1.3055486292799345e-05,
      "loss": 0.2734,
      "step": 10631
    },
    {
      "epoch": 0.8046544628920108,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3054352152758816e-05,
      "loss": 0.2881,
      "step": 10632
    },
    {
      "epoch": 0.8047301452154582,
      "grad_norm": 0.75,
      "learning_rate": 1.3053217969387064e-05,
      "loss": 0.3155,
      "step": 10633
    },
    {
      "epoch": 0.8048058275389054,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3052083742700172e-05,
      "loss": 0.3241,
      "step": 10634
    },
    {
      "epoch": 0.8048815098623527,
      "grad_norm": 0.75,
      "learning_rate": 1.3050949472714239e-05,
      "loss": 0.3052,
      "step": 10635
    },
    {
      "epoch": 0.8049571921858001,
      "grad_norm": 0.7265625,
      "learning_rate": 1.304981515944535e-05,
      "loss": 0.3028,
      "step": 10636
    },
    {
      "epoch": 0.8050328745092474,
      "grad_norm": 0.734375,
      "learning_rate": 1.30486808029096e-05,
      "loss": 0.2938,
      "step": 10637
    },
    {
      "epoch": 0.8051085568326948,
      "grad_norm": 0.72265625,
      "learning_rate": 1.3047546403123077e-05,
      "loss": 0.2914,
      "step": 10638
    },
    {
      "epoch": 0.8051842391561421,
      "grad_norm": 0.7578125,
      "learning_rate": 1.304641196010188e-05,
      "loss": 0.293,
      "step": 10639
    },
    {
      "epoch": 0.8052599214795895,
      "grad_norm": 0.78515625,
      "learning_rate": 1.3045277473862109e-05,
      "loss": 0.3356,
      "step": 10640
    },
    {
      "epoch": 0.8053356038030367,
      "grad_norm": 0.7578125,
      "learning_rate": 1.3044142944419841e-05,
      "loss": 0.3313,
      "step": 10641
    },
    {
      "epoch": 0.805411286126484,
      "grad_norm": 1.1328125,
      "learning_rate": 1.3043008371791186e-05,
      "loss": 0.3356,
      "step": 10642
    },
    {
      "epoch": 0.8054869684499314,
      "grad_norm": 0.91015625,
      "learning_rate": 1.3041873755992236e-05,
      "loss": 0.3529,
      "step": 10643
    },
    {
      "epoch": 0.8055626507733787,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3040739097039087e-05,
      "loss": 0.2817,
      "step": 10644
    },
    {
      "epoch": 0.8056383330968261,
      "grad_norm": 0.8203125,
      "learning_rate": 1.3039604394947833e-05,
      "loss": 0.3738,
      "step": 10645
    },
    {
      "epoch": 0.8057140154202734,
      "grad_norm": 0.80859375,
      "learning_rate": 1.3038469649734577e-05,
      "loss": 0.3362,
      "step": 10646
    },
    {
      "epoch": 0.8057896977437208,
      "grad_norm": 0.734375,
      "learning_rate": 1.3037334861415414e-05,
      "loss": 0.3166,
      "step": 10647
    },
    {
      "epoch": 0.805865380067168,
      "grad_norm": 0.75390625,
      "learning_rate": 1.3036200030006443e-05,
      "loss": 0.3475,
      "step": 10648
    },
    {
      "epoch": 0.8059410623906154,
      "grad_norm": 0.71875,
      "learning_rate": 1.3035065155523768e-05,
      "loss": 0.2669,
      "step": 10649
    },
    {
      "epoch": 0.8060167447140627,
      "grad_norm": 0.75390625,
      "learning_rate": 1.303393023798348e-05,
      "loss": 0.3549,
      "step": 10650
    },
    {
      "epoch": 0.80609242703751,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3032795277401689e-05,
      "loss": 0.333,
      "step": 10651
    },
    {
      "epoch": 0.8061681093609574,
      "grad_norm": 0.72265625,
      "learning_rate": 1.3031660273794491e-05,
      "loss": 0.2821,
      "step": 10652
    },
    {
      "epoch": 0.8062437916844047,
      "grad_norm": 0.78515625,
      "learning_rate": 1.3030525227177985e-05,
      "loss": 0.3405,
      "step": 10653
    },
    {
      "epoch": 0.8063194740078521,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3029390137568288e-05,
      "loss": 0.2855,
      "step": 10654
    },
    {
      "epoch": 0.8063951563312993,
      "grad_norm": 0.80078125,
      "learning_rate": 1.3028255004981488e-05,
      "loss": 0.3484,
      "step": 10655
    },
    {
      "epoch": 0.8064708386547467,
      "grad_norm": 0.78125,
      "learning_rate": 1.3027119829433693e-05,
      "loss": 0.3036,
      "step": 10656
    },
    {
      "epoch": 0.806546520978194,
      "grad_norm": 0.7734375,
      "learning_rate": 1.3025984610941009e-05,
      "loss": 0.3264,
      "step": 10657
    },
    {
      "epoch": 0.8066222033016414,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3024849349519543e-05,
      "loss": 0.2975,
      "step": 10658
    },
    {
      "epoch": 0.8066978856250887,
      "grad_norm": 0.8203125,
      "learning_rate": 1.3023714045185393e-05,
      "loss": 0.367,
      "step": 10659
    },
    {
      "epoch": 0.806773567948536,
      "grad_norm": 0.80078125,
      "learning_rate": 1.3022578697954675e-05,
      "loss": 0.341,
      "step": 10660
    },
    {
      "epoch": 0.8068492502719834,
      "grad_norm": 0.71875,
      "learning_rate": 1.302144330784349e-05,
      "loss": 0.2874,
      "step": 10661
    },
    {
      "epoch": 0.8069249325954306,
      "grad_norm": 0.76171875,
      "learning_rate": 1.3020307874867946e-05,
      "loss": 0.2967,
      "step": 10662
    },
    {
      "epoch": 0.807000614918878,
      "grad_norm": 0.71484375,
      "learning_rate": 1.3019172399044156e-05,
      "loss": 0.2962,
      "step": 10663
    },
    {
      "epoch": 0.8070762972423253,
      "grad_norm": 0.84765625,
      "learning_rate": 1.3018036880388219e-05,
      "loss": 0.3734,
      "step": 10664
    },
    {
      "epoch": 0.8071519795657727,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3016901318916251e-05,
      "loss": 0.3157,
      "step": 10665
    },
    {
      "epoch": 0.80722766188922,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3015765714644362e-05,
      "loss": 0.3068,
      "step": 10666
    },
    {
      "epoch": 0.8073033442126674,
      "grad_norm": 0.7421875,
      "learning_rate": 1.3014630067588658e-05,
      "loss": 0.3029,
      "step": 10667
    },
    {
      "epoch": 0.8073790265361147,
      "grad_norm": 0.734375,
      "learning_rate": 1.3013494377765256e-05,
      "loss": 0.3129,
      "step": 10668
    },
    {
      "epoch": 0.8074547088595619,
      "grad_norm": 0.74609375,
      "learning_rate": 1.3012358645190264e-05,
      "loss": 0.3001,
      "step": 10669
    },
    {
      "epoch": 0.8075303911830093,
      "grad_norm": 0.703125,
      "learning_rate": 1.3011222869879796e-05,
      "loss": 0.2714,
      "step": 10670
    },
    {
      "epoch": 0.8076060735064566,
      "grad_norm": 0.7890625,
      "learning_rate": 1.3010087051849962e-05,
      "loss": 0.3367,
      "step": 10671
    },
    {
      "epoch": 0.807681755829904,
      "grad_norm": 0.6796875,
      "learning_rate": 1.3008951191116884e-05,
      "loss": 0.2595,
      "step": 10672
    },
    {
      "epoch": 0.8077574381533513,
      "grad_norm": 0.73828125,
      "learning_rate": 1.3007815287696665e-05,
      "loss": 0.288,
      "step": 10673
    },
    {
      "epoch": 0.8078331204767987,
      "grad_norm": 0.73046875,
      "learning_rate": 1.3006679341605421e-05,
      "loss": 0.2896,
      "step": 10674
    },
    {
      "epoch": 0.807908802800246,
      "grad_norm": 0.70703125,
      "learning_rate": 1.3005543352859276e-05,
      "loss": 0.2853,
      "step": 10675
    },
    {
      "epoch": 0.8079844851236933,
      "grad_norm": 0.77734375,
      "learning_rate": 1.300440732147434e-05,
      "loss": 0.3177,
      "step": 10676
    },
    {
      "epoch": 0.8080601674471406,
      "grad_norm": 0.8203125,
      "learning_rate": 1.300327124746673e-05,
      "loss": 0.3435,
      "step": 10677
    },
    {
      "epoch": 0.8081358497705879,
      "grad_norm": 0.76953125,
      "learning_rate": 1.3002135130852565e-05,
      "loss": 0.3372,
      "step": 10678
    },
    {
      "epoch": 0.8082115320940353,
      "grad_norm": 0.75,
      "learning_rate": 1.300099897164796e-05,
      "loss": 0.3087,
      "step": 10679
    },
    {
      "epoch": 0.8082872144174826,
      "grad_norm": 0.796875,
      "learning_rate": 1.2999862769869032e-05,
      "loss": 0.3301,
      "step": 10680
    },
    {
      "epoch": 0.80836289674093,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2998726525531906e-05,
      "loss": 0.2935,
      "step": 10681
    },
    {
      "epoch": 0.8084385790643773,
      "grad_norm": 0.71875,
      "learning_rate": 1.2997590238652697e-05,
      "loss": 0.298,
      "step": 10682
    },
    {
      "epoch": 0.8085142613878246,
      "grad_norm": 0.78125,
      "learning_rate": 1.2996453909247529e-05,
      "loss": 0.3547,
      "step": 10683
    },
    {
      "epoch": 0.8085899437112719,
      "grad_norm": 0.71875,
      "learning_rate": 1.2995317537332517e-05,
      "loss": 0.2798,
      "step": 10684
    },
    {
      "epoch": 0.8086656260347193,
      "grad_norm": 0.73828125,
      "learning_rate": 1.2994181122923788e-05,
      "loss": 0.3243,
      "step": 10685
    },
    {
      "epoch": 0.8087413083581666,
      "grad_norm": 0.75390625,
      "learning_rate": 1.299304466603746e-05,
      "loss": 0.3183,
      "step": 10686
    },
    {
      "epoch": 0.808816990681614,
      "grad_norm": 0.80078125,
      "learning_rate": 1.2991908166689657e-05,
      "loss": 0.3497,
      "step": 10687
    },
    {
      "epoch": 0.8088926730050613,
      "grad_norm": 0.70703125,
      "learning_rate": 1.29907716248965e-05,
      "loss": 0.2989,
      "step": 10688
    },
    {
      "epoch": 0.8089683553285086,
      "grad_norm": 0.75390625,
      "learning_rate": 1.298963504067412e-05,
      "loss": 0.3172,
      "step": 10689
    },
    {
      "epoch": 0.8090440376519559,
      "grad_norm": 0.671875,
      "learning_rate": 1.2988498414038635e-05,
      "loss": 0.258,
      "step": 10690
    },
    {
      "epoch": 0.8091197199754032,
      "grad_norm": 0.76171875,
      "learning_rate": 1.298736174500617e-05,
      "loss": 0.2859,
      "step": 10691
    },
    {
      "epoch": 0.8091954022988506,
      "grad_norm": 0.78515625,
      "learning_rate": 1.2986225033592854e-05,
      "loss": 0.3361,
      "step": 10692
    },
    {
      "epoch": 0.8092710846222979,
      "grad_norm": 0.80078125,
      "learning_rate": 1.2985088279814813e-05,
      "loss": 0.3337,
      "step": 10693
    },
    {
      "epoch": 0.8093467669457453,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2983951483688167e-05,
      "loss": 0.2881,
      "step": 10694
    },
    {
      "epoch": 0.8094224492691926,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2982814645229052e-05,
      "loss": 0.3387,
      "step": 10695
    },
    {
      "epoch": 0.80949813159264,
      "grad_norm": 0.70703125,
      "learning_rate": 1.2981677764453593e-05,
      "loss": 0.2838,
      "step": 10696
    },
    {
      "epoch": 0.8095738139160872,
      "grad_norm": 0.80078125,
      "learning_rate": 1.2980540841377915e-05,
      "loss": 0.2897,
      "step": 10697
    },
    {
      "epoch": 0.8096494962395345,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2979403876018153e-05,
      "loss": 0.2974,
      "step": 10698
    },
    {
      "epoch": 0.8097251785629819,
      "grad_norm": 0.765625,
      "learning_rate": 1.2978266868390435e-05,
      "loss": 0.3307,
      "step": 10699
    },
    {
      "epoch": 0.8098008608864292,
      "grad_norm": 0.8671875,
      "learning_rate": 1.2977129818510887e-05,
      "loss": 0.3538,
      "step": 10700
    },
    {
      "epoch": 0.8098765432098766,
      "grad_norm": 0.79296875,
      "learning_rate": 1.2975992726395647e-05,
      "loss": 0.353,
      "step": 10701
    },
    {
      "epoch": 0.8099522255333239,
      "grad_norm": 0.71484375,
      "learning_rate": 1.2974855592060841e-05,
      "loss": 0.3084,
      "step": 10702
    },
    {
      "epoch": 0.8100279078567713,
      "grad_norm": 0.734375,
      "learning_rate": 1.2973718415522605e-05,
      "loss": 0.2927,
      "step": 10703
    },
    {
      "epoch": 0.8101035901802185,
      "grad_norm": 0.75,
      "learning_rate": 1.297258119679707e-05,
      "loss": 0.296,
      "step": 10704
    },
    {
      "epoch": 0.8101792725036658,
      "grad_norm": 0.78125,
      "learning_rate": 1.2971443935900371e-05,
      "loss": 0.3243,
      "step": 10705
    },
    {
      "epoch": 0.8102549548271132,
      "grad_norm": 0.671875,
      "learning_rate": 1.2970306632848637e-05,
      "loss": 0.2782,
      "step": 10706
    },
    {
      "epoch": 0.8103306371505605,
      "grad_norm": 0.75,
      "learning_rate": 1.2969169287658008e-05,
      "loss": 0.304,
      "step": 10707
    },
    {
      "epoch": 0.8104063194740079,
      "grad_norm": 1.140625,
      "learning_rate": 1.2968031900344618e-05,
      "loss": 0.3618,
      "step": 10708
    },
    {
      "epoch": 0.8104820017974552,
      "grad_norm": 0.71875,
      "learning_rate": 1.29668944709246e-05,
      "loss": 0.2969,
      "step": 10709
    },
    {
      "epoch": 0.8105576841209026,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2965756999414097e-05,
      "loss": 0.309,
      "step": 10710
    },
    {
      "epoch": 0.8106333664443498,
      "grad_norm": 0.78125,
      "learning_rate": 1.2964619485829237e-05,
      "loss": 0.3406,
      "step": 10711
    },
    {
      "epoch": 0.8107090487677971,
      "grad_norm": 1.203125,
      "learning_rate": 1.2963481930186163e-05,
      "loss": 0.3431,
      "step": 10712
    },
    {
      "epoch": 0.8107847310912445,
      "grad_norm": 0.73828125,
      "learning_rate": 1.2962344332501013e-05,
      "loss": 0.3051,
      "step": 10713
    },
    {
      "epoch": 0.8108604134146918,
      "grad_norm": 0.70703125,
      "learning_rate": 1.296120669278993e-05,
      "loss": 0.3031,
      "step": 10714
    },
    {
      "epoch": 0.8109360957381392,
      "grad_norm": 0.70703125,
      "learning_rate": 1.296006901106904e-05,
      "loss": 0.285,
      "step": 10715
    },
    {
      "epoch": 0.8110117780615865,
      "grad_norm": 0.734375,
      "learning_rate": 1.2958931287354495e-05,
      "loss": 0.3019,
      "step": 10716
    },
    {
      "epoch": 0.8110874603850339,
      "grad_norm": 0.73046875,
      "learning_rate": 1.2957793521662434e-05,
      "loss": 0.3266,
      "step": 10717
    },
    {
      "epoch": 0.8111631427084811,
      "grad_norm": 0.734375,
      "learning_rate": 1.2956655714008992e-05,
      "loss": 0.3136,
      "step": 10718
    },
    {
      "epoch": 0.8112388250319285,
      "grad_norm": 0.78125,
      "learning_rate": 1.295551786441032e-05,
      "loss": 0.3211,
      "step": 10719
    },
    {
      "epoch": 0.8113145073553758,
      "grad_norm": 0.73046875,
      "learning_rate": 1.2954379972882551e-05,
      "loss": 0.3018,
      "step": 10720
    },
    {
      "epoch": 0.8113901896788231,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2953242039441833e-05,
      "loss": 0.3063,
      "step": 10721
    },
    {
      "epoch": 0.8114658720022705,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2952104064104309e-05,
      "loss": 0.3035,
      "step": 10722
    },
    {
      "epoch": 0.8115415543257178,
      "grad_norm": 0.796875,
      "learning_rate": 1.2950966046886124e-05,
      "loss": 0.3491,
      "step": 10723
    },
    {
      "epoch": 0.8116172366491651,
      "grad_norm": 0.7890625,
      "learning_rate": 1.2949827987803421e-05,
      "loss": 0.3316,
      "step": 10724
    },
    {
      "epoch": 0.8116929189726124,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2948689886872346e-05,
      "loss": 0.3272,
      "step": 10725
    },
    {
      "epoch": 0.8117686012960598,
      "grad_norm": 0.7265625,
      "learning_rate": 1.2947551744109044e-05,
      "loss": 0.3193,
      "step": 10726
    },
    {
      "epoch": 0.8118442836195071,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2946413559529662e-05,
      "loss": 0.3296,
      "step": 10727
    },
    {
      "epoch": 0.8119199659429545,
      "grad_norm": 0.76171875,
      "learning_rate": 1.2945275333150348e-05,
      "loss": 0.3198,
      "step": 10728
    },
    {
      "epoch": 0.8119956482664018,
      "grad_norm": 0.77734375,
      "learning_rate": 1.2944137064987249e-05,
      "loss": 0.3166,
      "step": 10729
    },
    {
      "epoch": 0.8120713305898491,
      "grad_norm": 0.8046875,
      "learning_rate": 1.2942998755056515e-05,
      "loss": 0.3155,
      "step": 10730
    },
    {
      "epoch": 0.8121470129132964,
      "grad_norm": 0.75,
      "learning_rate": 1.2941860403374294e-05,
      "loss": 0.3078,
      "step": 10731
    },
    {
      "epoch": 0.8122226952367437,
      "grad_norm": 0.7109375,
      "learning_rate": 1.2940722009956731e-05,
      "loss": 0.2635,
      "step": 10732
    },
    {
      "epoch": 0.8122983775601911,
      "grad_norm": 0.78515625,
      "learning_rate": 1.2939583574819983e-05,
      "loss": 0.3071,
      "step": 10733
    },
    {
      "epoch": 0.8123740598836384,
      "grad_norm": 1.1796875,
      "learning_rate": 1.2938445097980197e-05,
      "loss": 0.3631,
      "step": 10734
    },
    {
      "epoch": 0.8124497422070858,
      "grad_norm": 0.81640625,
      "learning_rate": 1.2937306579453524e-05,
      "loss": 0.3538,
      "step": 10735
    },
    {
      "epoch": 0.8125254245305331,
      "grad_norm": 0.8046875,
      "learning_rate": 1.2936168019256114e-05,
      "loss": 0.3442,
      "step": 10736
    },
    {
      "epoch": 0.8126011068539805,
      "grad_norm": 0.75,
      "learning_rate": 1.2935029417404126e-05,
      "loss": 0.3055,
      "step": 10737
    },
    {
      "epoch": 0.8126767891774277,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2933890773913707e-05,
      "loss": 0.3135,
      "step": 10738
    },
    {
      "epoch": 0.812752471500875,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2932752088801013e-05,
      "loss": 0.3402,
      "step": 10739
    },
    {
      "epoch": 0.8128281538243224,
      "grad_norm": 0.7734375,
      "learning_rate": 1.2931613362082198e-05,
      "loss": 0.3002,
      "step": 10740
    },
    {
      "epoch": 0.8129038361477697,
      "grad_norm": 0.80078125,
      "learning_rate": 1.2930474593773416e-05,
      "loss": 0.3174,
      "step": 10741
    },
    {
      "epoch": 0.8129795184712171,
      "grad_norm": 0.71875,
      "learning_rate": 1.2929335783890826e-05,
      "loss": 0.2893,
      "step": 10742
    },
    {
      "epoch": 0.8130552007946644,
      "grad_norm": 0.7734375,
      "learning_rate": 1.2928196932450579e-05,
      "loss": 0.3354,
      "step": 10743
    },
    {
      "epoch": 0.8131308831181118,
      "grad_norm": 0.82421875,
      "learning_rate": 1.2927058039468832e-05,
      "loss": 0.3428,
      "step": 10744
    },
    {
      "epoch": 0.813206565441559,
      "grad_norm": 0.71484375,
      "learning_rate": 1.2925919104961744e-05,
      "loss": 0.287,
      "step": 10745
    },
    {
      "epoch": 0.8132822477650063,
      "grad_norm": 0.67578125,
      "learning_rate": 1.2924780128945473e-05,
      "loss": 0.2731,
      "step": 10746
    },
    {
      "epoch": 0.8133579300884537,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2923641111436178e-05,
      "loss": 0.3007,
      "step": 10747
    },
    {
      "epoch": 0.813433612411901,
      "grad_norm": 0.80859375,
      "learning_rate": 1.2922502052450017e-05,
      "loss": 0.3322,
      "step": 10748
    },
    {
      "epoch": 0.8135092947353484,
      "grad_norm": 0.703125,
      "learning_rate": 1.292136295200315e-05,
      "loss": 0.2895,
      "step": 10749
    },
    {
      "epoch": 0.8135849770587957,
      "grad_norm": 0.76171875,
      "learning_rate": 1.2920223810111731e-05,
      "loss": 0.3075,
      "step": 10750
    },
    {
      "epoch": 0.8136606593822431,
      "grad_norm": 0.703125,
      "learning_rate": 1.2919084626791929e-05,
      "loss": 0.2759,
      "step": 10751
    },
    {
      "epoch": 0.8137363417056903,
      "grad_norm": 0.76171875,
      "learning_rate": 1.2917945402059906e-05,
      "loss": 0.3078,
      "step": 10752
    },
    {
      "epoch": 0.8138120240291377,
      "grad_norm": 0.7109375,
      "learning_rate": 1.2916806135931815e-05,
      "loss": 0.284,
      "step": 10753
    },
    {
      "epoch": 0.813887706352585,
      "grad_norm": 0.79296875,
      "learning_rate": 1.2915666828423826e-05,
      "loss": 0.3369,
      "step": 10754
    },
    {
      "epoch": 0.8139633886760324,
      "grad_norm": 0.73828125,
      "learning_rate": 1.29145274795521e-05,
      "loss": 0.3081,
      "step": 10755
    },
    {
      "epoch": 0.8140390709994797,
      "grad_norm": 0.77734375,
      "learning_rate": 1.2913388089332797e-05,
      "loss": 0.3175,
      "step": 10756
    },
    {
      "epoch": 0.814114753322927,
      "grad_norm": 0.78515625,
      "learning_rate": 1.2912248657782086e-05,
      "loss": 0.3059,
      "step": 10757
    },
    {
      "epoch": 0.8141904356463744,
      "grad_norm": 0.80078125,
      "learning_rate": 1.2911109184916132e-05,
      "loss": 0.3113,
      "step": 10758
    },
    {
      "epoch": 0.8142661179698216,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2909969670751097e-05,
      "loss": 0.3113,
      "step": 10759
    },
    {
      "epoch": 0.814341800293269,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2908830115303149e-05,
      "loss": 0.3281,
      "step": 10760
    },
    {
      "epoch": 0.8144174826167163,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2907690518588456e-05,
      "loss": 0.3193,
      "step": 10761
    },
    {
      "epoch": 0.8144931649401637,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2906550880623178e-05,
      "loss": 0.3121,
      "step": 10762
    },
    {
      "epoch": 0.814568847263611,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2905411201423491e-05,
      "loss": 0.2978,
      "step": 10763
    },
    {
      "epoch": 0.8146445295870584,
      "grad_norm": 0.75,
      "learning_rate": 1.2904271481005562e-05,
      "loss": 0.3181,
      "step": 10764
    },
    {
      "epoch": 0.8147202119105057,
      "grad_norm": 1.1015625,
      "learning_rate": 1.2903131719385553e-05,
      "loss": 0.3747,
      "step": 10765
    },
    {
      "epoch": 0.8147958942339529,
      "grad_norm": 0.734375,
      "learning_rate": 1.2901991916579646e-05,
      "loss": 0.3079,
      "step": 10766
    },
    {
      "epoch": 0.8148715765574003,
      "grad_norm": 0.734375,
      "learning_rate": 1.2900852072603999e-05,
      "loss": 0.2925,
      "step": 10767
    },
    {
      "epoch": 0.8149472588808476,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2899712187474787e-05,
      "loss": 0.3342,
      "step": 10768
    },
    {
      "epoch": 0.815022941204295,
      "grad_norm": 0.8046875,
      "learning_rate": 1.2898572261208182e-05,
      "loss": 0.3637,
      "step": 10769
    },
    {
      "epoch": 0.8150986235277423,
      "grad_norm": 0.734375,
      "learning_rate": 1.289743229382036e-05,
      "loss": 0.2805,
      "step": 10770
    },
    {
      "epoch": 0.8151743058511897,
      "grad_norm": 0.70703125,
      "learning_rate": 1.289629228532748e-05,
      "loss": 0.2894,
      "step": 10771
    },
    {
      "epoch": 0.815249988174637,
      "grad_norm": 0.8046875,
      "learning_rate": 1.2895152235745732e-05,
      "loss": 0.3633,
      "step": 10772
    },
    {
      "epoch": 0.8153256704980842,
      "grad_norm": 0.78515625,
      "learning_rate": 1.2894012145091277e-05,
      "loss": 0.324,
      "step": 10773
    },
    {
      "epoch": 0.8154013528215316,
      "grad_norm": 0.98046875,
      "learning_rate": 1.2892872013380293e-05,
      "loss": 0.3843,
      "step": 10774
    },
    {
      "epoch": 0.8154770351449789,
      "grad_norm": 0.7734375,
      "learning_rate": 1.2891731840628958e-05,
      "loss": 0.3095,
      "step": 10775
    },
    {
      "epoch": 0.8155527174684263,
      "grad_norm": 0.734375,
      "learning_rate": 1.2890591626853443e-05,
      "loss": 0.2913,
      "step": 10776
    },
    {
      "epoch": 0.8156283997918736,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2889451372069924e-05,
      "loss": 0.3255,
      "step": 10777
    },
    {
      "epoch": 0.815704082115321,
      "grad_norm": 0.796875,
      "learning_rate": 1.288831107629458e-05,
      "loss": 0.3362,
      "step": 10778
    },
    {
      "epoch": 0.8157797644387683,
      "grad_norm": 0.796875,
      "learning_rate": 1.2887170739543586e-05,
      "loss": 0.3453,
      "step": 10779
    },
    {
      "epoch": 0.8158554467622156,
      "grad_norm": 0.78515625,
      "learning_rate": 1.2886030361833122e-05,
      "loss": 0.3421,
      "step": 10780
    },
    {
      "epoch": 0.8159311290856629,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2884889943179363e-05,
      "loss": 0.3393,
      "step": 10781
    },
    {
      "epoch": 0.8160068114091102,
      "grad_norm": 0.7734375,
      "learning_rate": 1.2883749483598491e-05,
      "loss": 0.3206,
      "step": 10782
    },
    {
      "epoch": 0.8160824937325576,
      "grad_norm": 0.7734375,
      "learning_rate": 1.2882608983106684e-05,
      "loss": 0.3342,
      "step": 10783
    },
    {
      "epoch": 0.8161581760560049,
      "grad_norm": 0.80859375,
      "learning_rate": 1.2881468441720122e-05,
      "loss": 0.369,
      "step": 10784
    },
    {
      "epoch": 0.8162338583794523,
      "grad_norm": 0.67578125,
      "learning_rate": 1.2880327859454983e-05,
      "loss": 0.2277,
      "step": 10785
    },
    {
      "epoch": 0.8163095407028996,
      "grad_norm": 0.71875,
      "learning_rate": 1.2879187236327452e-05,
      "loss": 0.2834,
      "step": 10786
    },
    {
      "epoch": 0.8163852230263469,
      "grad_norm": 0.76171875,
      "learning_rate": 1.287804657235371e-05,
      "loss": 0.33,
      "step": 10787
    },
    {
      "epoch": 0.8164609053497942,
      "grad_norm": 0.73828125,
      "learning_rate": 1.2876905867549938e-05,
      "loss": 0.3147,
      "step": 10788
    },
    {
      "epoch": 0.8165365876732416,
      "grad_norm": 0.8125,
      "learning_rate": 1.287576512193232e-05,
      "loss": 0.3401,
      "step": 10789
    },
    {
      "epoch": 0.8166122699966889,
      "grad_norm": 0.859375,
      "learning_rate": 1.287462433551704e-05,
      "loss": 0.3448,
      "step": 10790
    },
    {
      "epoch": 0.8166879523201362,
      "grad_norm": 0.76171875,
      "learning_rate": 1.2873483508320277e-05,
      "loss": 0.324,
      "step": 10791
    },
    {
      "epoch": 0.8167636346435836,
      "grad_norm": 0.69140625,
      "learning_rate": 1.287234264035822e-05,
      "loss": 0.2698,
      "step": 10792
    },
    {
      "epoch": 0.8168393169670309,
      "grad_norm": 0.6796875,
      "learning_rate": 1.2871201731647054e-05,
      "loss": 0.2727,
      "step": 10793
    },
    {
      "epoch": 0.8169149992904782,
      "grad_norm": 0.79296875,
      "learning_rate": 1.2870060782202964e-05,
      "loss": 0.3283,
      "step": 10794
    },
    {
      "epoch": 0.8169906816139255,
      "grad_norm": 0.69921875,
      "learning_rate": 1.286891979204214e-05,
      "loss": 0.2969,
      "step": 10795
    },
    {
      "epoch": 0.8170663639373729,
      "grad_norm": 0.72265625,
      "learning_rate": 1.2867778761180763e-05,
      "loss": 0.2826,
      "step": 10796
    },
    {
      "epoch": 0.8171420462608202,
      "grad_norm": 0.8046875,
      "learning_rate": 1.2866637689635023e-05,
      "loss": 0.341,
      "step": 10797
    },
    {
      "epoch": 0.8172177285842676,
      "grad_norm": 0.76171875,
      "learning_rate": 1.2865496577421108e-05,
      "loss": 0.3466,
      "step": 10798
    },
    {
      "epoch": 0.8172934109077149,
      "grad_norm": 0.67578125,
      "learning_rate": 1.2864355424555206e-05,
      "loss": 0.2571,
      "step": 10799
    },
    {
      "epoch": 0.8173690932311622,
      "grad_norm": 0.84375,
      "learning_rate": 1.2863214231053507e-05,
      "loss": 0.3378,
      "step": 10800
    },
    {
      "epoch": 0.8174447755546095,
      "grad_norm": 0.75,
      "learning_rate": 1.2862072996932202e-05,
      "loss": 0.3061,
      "step": 10801
    },
    {
      "epoch": 0.8175204578780568,
      "grad_norm": 0.75390625,
      "learning_rate": 1.286093172220748e-05,
      "loss": 0.3327,
      "step": 10802
    },
    {
      "epoch": 0.8175961402015042,
      "grad_norm": 0.77734375,
      "learning_rate": 1.2859790406895532e-05,
      "loss": 0.3339,
      "step": 10803
    },
    {
      "epoch": 0.8176718225249515,
      "grad_norm": 0.7890625,
      "learning_rate": 1.2858649051012549e-05,
      "loss": 0.3491,
      "step": 10804
    },
    {
      "epoch": 0.8177475048483989,
      "grad_norm": 0.7265625,
      "learning_rate": 1.2857507654574723e-05,
      "loss": 0.3038,
      "step": 10805
    },
    {
      "epoch": 0.8178231871718462,
      "grad_norm": 0.78515625,
      "learning_rate": 1.2856366217598246e-05,
      "loss": 0.3448,
      "step": 10806
    },
    {
      "epoch": 0.8178988694952936,
      "grad_norm": 0.78515625,
      "learning_rate": 1.2855224740099314e-05,
      "loss": 0.3291,
      "step": 10807
    },
    {
      "epoch": 0.8179745518187408,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2854083222094122e-05,
      "loss": 0.2936,
      "step": 10808
    },
    {
      "epoch": 0.8180502341421881,
      "grad_norm": 0.74609375,
      "learning_rate": 1.285294166359886e-05,
      "loss": 0.2894,
      "step": 10809
    },
    {
      "epoch": 0.8181259164656355,
      "grad_norm": 0.76171875,
      "learning_rate": 1.2851800064629723e-05,
      "loss": 0.3179,
      "step": 10810
    },
    {
      "epoch": 0.8182015987890828,
      "grad_norm": 0.80078125,
      "learning_rate": 1.2850658425202908e-05,
      "loss": 0.341,
      "step": 10811
    },
    {
      "epoch": 0.8182772811125302,
      "grad_norm": 0.7890625,
      "learning_rate": 1.2849516745334614e-05,
      "loss": 0.3113,
      "step": 10812
    },
    {
      "epoch": 0.8183529634359775,
      "grad_norm": 0.734375,
      "learning_rate": 1.2848375025041034e-05,
      "loss": 0.3097,
      "step": 10813
    },
    {
      "epoch": 0.8184286457594249,
      "grad_norm": 0.75,
      "learning_rate": 1.2847233264338366e-05,
      "loss": 0.3053,
      "step": 10814
    },
    {
      "epoch": 0.8185043280828721,
      "grad_norm": 0.73828125,
      "learning_rate": 1.2846091463242804e-05,
      "loss": 0.3234,
      "step": 10815
    },
    {
      "epoch": 0.8185800104063194,
      "grad_norm": 0.8046875,
      "learning_rate": 1.2844949621770558e-05,
      "loss": 0.347,
      "step": 10816
    },
    {
      "epoch": 0.8186556927297668,
      "grad_norm": 0.73046875,
      "learning_rate": 1.2843807739937816e-05,
      "loss": 0.3139,
      "step": 10817
    },
    {
      "epoch": 0.8187313750532141,
      "grad_norm": 0.69140625,
      "learning_rate": 1.284266581776078e-05,
      "loss": 0.2748,
      "step": 10818
    },
    {
      "epoch": 0.8188070573766615,
      "grad_norm": 0.77734375,
      "learning_rate": 1.2841523855255654e-05,
      "loss": 0.3392,
      "step": 10819
    },
    {
      "epoch": 0.8188827397001088,
      "grad_norm": 0.765625,
      "learning_rate": 1.2840381852438633e-05,
      "loss": 0.3142,
      "step": 10820
    },
    {
      "epoch": 0.8189584220235562,
      "grad_norm": 1.0390625,
      "learning_rate": 1.2839239809325921e-05,
      "loss": 0.308,
      "step": 10821
    },
    {
      "epoch": 0.8190341043470034,
      "grad_norm": 0.734375,
      "learning_rate": 1.2838097725933723e-05,
      "loss": 0.3134,
      "step": 10822
    },
    {
      "epoch": 0.8191097866704508,
      "grad_norm": 0.796875,
      "learning_rate": 1.2836955602278238e-05,
      "loss": 0.3234,
      "step": 10823
    },
    {
      "epoch": 0.8191854689938981,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2835813438375667e-05,
      "loss": 0.2919,
      "step": 10824
    },
    {
      "epoch": 0.8192611513173454,
      "grad_norm": 0.6875,
      "learning_rate": 1.2834671234242215e-05,
      "loss": 0.2517,
      "step": 10825
    },
    {
      "epoch": 0.8193368336407928,
      "grad_norm": 0.98046875,
      "learning_rate": 1.2833528989894093e-05,
      "loss": 0.3597,
      "step": 10826
    },
    {
      "epoch": 0.8194125159642401,
      "grad_norm": 0.8125,
      "learning_rate": 1.2832386705347494e-05,
      "loss": 0.354,
      "step": 10827
    },
    {
      "epoch": 0.8194881982876875,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2831244380618633e-05,
      "loss": 0.314,
      "step": 10828
    },
    {
      "epoch": 0.8195638806111347,
      "grad_norm": 0.765625,
      "learning_rate": 1.283010201572371e-05,
      "loss": 0.3503,
      "step": 10829
    },
    {
      "epoch": 0.8196395629345821,
      "grad_norm": 0.72265625,
      "learning_rate": 1.282895961067893e-05,
      "loss": 0.2774,
      "step": 10830
    },
    {
      "epoch": 0.8197152452580294,
      "grad_norm": 0.8046875,
      "learning_rate": 1.282781716550051e-05,
      "loss": 0.3236,
      "step": 10831
    },
    {
      "epoch": 0.8197909275814768,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2826674680204647e-05,
      "loss": 0.3169,
      "step": 10832
    },
    {
      "epoch": 0.8198666099049241,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2825532154807552e-05,
      "loss": 0.2916,
      "step": 10833
    },
    {
      "epoch": 0.8199422922283714,
      "grad_norm": 0.7734375,
      "learning_rate": 1.2824389589325435e-05,
      "loss": 0.3448,
      "step": 10834
    },
    {
      "epoch": 0.8200179745518188,
      "grad_norm": 0.79296875,
      "learning_rate": 1.2823246983774506e-05,
      "loss": 0.2739,
      "step": 10835
    },
    {
      "epoch": 0.820093656875266,
      "grad_norm": 0.73828125,
      "learning_rate": 1.2822104338170973e-05,
      "loss": 0.2955,
      "step": 10836
    },
    {
      "epoch": 0.8201693391987134,
      "grad_norm": 0.796875,
      "learning_rate": 1.2820961652531048e-05,
      "loss": 0.3524,
      "step": 10837
    },
    {
      "epoch": 0.8202450215221607,
      "grad_norm": 0.734375,
      "learning_rate": 1.2819818926870942e-05,
      "loss": 0.2826,
      "step": 10838
    },
    {
      "epoch": 0.8203207038456081,
      "grad_norm": 0.75,
      "learning_rate": 1.2818676161206862e-05,
      "loss": 0.329,
      "step": 10839
    },
    {
      "epoch": 0.8203963861690554,
      "grad_norm": 0.765625,
      "learning_rate": 1.2817533355555028e-05,
      "loss": 0.3425,
      "step": 10840
    },
    {
      "epoch": 0.8204720684925028,
      "grad_norm": 0.83203125,
      "learning_rate": 1.2816390509931641e-05,
      "loss": 0.3701,
      "step": 10841
    },
    {
      "epoch": 0.82054775081595,
      "grad_norm": 0.82421875,
      "learning_rate": 1.281524762435293e-05,
      "loss": 0.3331,
      "step": 10842
    },
    {
      "epoch": 0.8206234331393973,
      "grad_norm": 0.7265625,
      "learning_rate": 1.2814104698835096e-05,
      "loss": 0.322,
      "step": 10843
    },
    {
      "epoch": 0.8206991154628447,
      "grad_norm": 0.90234375,
      "learning_rate": 1.2812961733394357e-05,
      "loss": 0.3177,
      "step": 10844
    },
    {
      "epoch": 0.820774797786292,
      "grad_norm": 0.7734375,
      "learning_rate": 1.2811818728046932e-05,
      "loss": 0.3429,
      "step": 10845
    },
    {
      "epoch": 0.8208504801097394,
      "grad_norm": 1.015625,
      "learning_rate": 1.2810675682809031e-05,
      "loss": 0.2918,
      "step": 10846
    },
    {
      "epoch": 0.8209261624331867,
      "grad_norm": 0.7890625,
      "learning_rate": 1.2809532597696872e-05,
      "loss": 0.3453,
      "step": 10847
    },
    {
      "epoch": 0.8210018447566341,
      "grad_norm": 0.71875,
      "learning_rate": 1.280838947272667e-05,
      "loss": 0.2965,
      "step": 10848
    },
    {
      "epoch": 0.8210775270800813,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2807246307914648e-05,
      "loss": 0.3352,
      "step": 10849
    },
    {
      "epoch": 0.8211532094035286,
      "grad_norm": 0.68359375,
      "learning_rate": 1.2806103103277017e-05,
      "loss": 0.2768,
      "step": 10850
    },
    {
      "epoch": 0.821228891726976,
      "grad_norm": 0.80859375,
      "learning_rate": 1.280495985883e-05,
      "loss": 0.3339,
      "step": 10851
    },
    {
      "epoch": 0.8213045740504233,
      "grad_norm": 0.78125,
      "learning_rate": 1.2803816574589814e-05,
      "loss": 0.2909,
      "step": 10852
    },
    {
      "epoch": 0.8213802563738707,
      "grad_norm": 0.83203125,
      "learning_rate": 1.2802673250572674e-05,
      "loss": 0.3436,
      "step": 10853
    },
    {
      "epoch": 0.821455938697318,
      "grad_norm": 0.83984375,
      "learning_rate": 1.280152988679481e-05,
      "loss": 0.3771,
      "step": 10854
    },
    {
      "epoch": 0.8215316210207654,
      "grad_norm": 0.76171875,
      "learning_rate": 1.2800386483272433e-05,
      "loss": 0.3188,
      "step": 10855
    },
    {
      "epoch": 0.8216073033442126,
      "grad_norm": 0.75390625,
      "learning_rate": 1.279924304002177e-05,
      "loss": 0.2892,
      "step": 10856
    },
    {
      "epoch": 0.82168298566766,
      "grad_norm": 0.8203125,
      "learning_rate": 1.279809955705904e-05,
      "loss": 0.3326,
      "step": 10857
    },
    {
      "epoch": 0.8217586679911073,
      "grad_norm": 0.80078125,
      "learning_rate": 1.2796956034400467e-05,
      "loss": 0.3105,
      "step": 10858
    },
    {
      "epoch": 0.8218343503145547,
      "grad_norm": 0.75,
      "learning_rate": 1.2795812472062273e-05,
      "loss": 0.3055,
      "step": 10859
    },
    {
      "epoch": 0.821910032638002,
      "grad_norm": 0.7265625,
      "learning_rate": 1.279466887006068e-05,
      "loss": 0.28,
      "step": 10860
    },
    {
      "epoch": 0.8219857149614493,
      "grad_norm": 0.703125,
      "learning_rate": 1.2793525228411913e-05,
      "loss": 0.28,
      "step": 10861
    },
    {
      "epoch": 0.8220613972848967,
      "grad_norm": 0.8828125,
      "learning_rate": 1.2792381547132195e-05,
      "loss": 0.2976,
      "step": 10862
    },
    {
      "epoch": 0.8221370796083439,
      "grad_norm": 0.80859375,
      "learning_rate": 1.2791237826237756e-05,
      "loss": 0.3334,
      "step": 10863
    },
    {
      "epoch": 0.8222127619317913,
      "grad_norm": 0.73046875,
      "learning_rate": 1.2790094065744822e-05,
      "loss": 0.2707,
      "step": 10864
    },
    {
      "epoch": 0.8222884442552386,
      "grad_norm": 0.73828125,
      "learning_rate": 1.2788950265669611e-05,
      "loss": 0.3064,
      "step": 10865
    },
    {
      "epoch": 0.822364126578686,
      "grad_norm": 0.75,
      "learning_rate": 1.2787806426028354e-05,
      "loss": 0.2878,
      "step": 10866
    },
    {
      "epoch": 0.8224398089021333,
      "grad_norm": 0.69921875,
      "learning_rate": 1.2786662546837281e-05,
      "loss": 0.2716,
      "step": 10867
    },
    {
      "epoch": 0.8225154912255807,
      "grad_norm": 0.734375,
      "learning_rate": 1.2785518628112616e-05,
      "loss": 0.2989,
      "step": 10868
    },
    {
      "epoch": 0.822591173549028,
      "grad_norm": 0.7890625,
      "learning_rate": 1.2784374669870592e-05,
      "loss": 0.3179,
      "step": 10869
    },
    {
      "epoch": 0.8226668558724752,
      "grad_norm": 0.8203125,
      "learning_rate": 1.2783230672127434e-05,
      "loss": 0.3567,
      "step": 10870
    },
    {
      "epoch": 0.8227425381959226,
      "grad_norm": 0.73046875,
      "learning_rate": 1.2782086634899369e-05,
      "loss": 0.3105,
      "step": 10871
    },
    {
      "epoch": 0.8228182205193699,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2780942558202633e-05,
      "loss": 0.3011,
      "step": 10872
    },
    {
      "epoch": 0.8228939028428173,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2779798442053456e-05,
      "loss": 0.3041,
      "step": 10873
    },
    {
      "epoch": 0.8229695851662646,
      "grad_norm": 0.83984375,
      "learning_rate": 1.2778654286468067e-05,
      "loss": 0.3566,
      "step": 10874
    },
    {
      "epoch": 0.823045267489712,
      "grad_norm": 0.7109375,
      "learning_rate": 1.2777510091462699e-05,
      "loss": 0.299,
      "step": 10875
    },
    {
      "epoch": 0.8231209498131593,
      "grad_norm": 0.8515625,
      "learning_rate": 1.2776365857053584e-05,
      "loss": 0.3563,
      "step": 10876
    },
    {
      "epoch": 0.8231966321366065,
      "grad_norm": 0.71875,
      "learning_rate": 1.2775221583256952e-05,
      "loss": 0.2785,
      "step": 10877
    },
    {
      "epoch": 0.8232723144600539,
      "grad_norm": 0.73046875,
      "learning_rate": 1.2774077270089044e-05,
      "loss": 0.2959,
      "step": 10878
    },
    {
      "epoch": 0.8233479967835012,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2772932917566087e-05,
      "loss": 0.3075,
      "step": 10879
    },
    {
      "epoch": 0.8234236791069486,
      "grad_norm": 0.87890625,
      "learning_rate": 1.2771788525704318e-05,
      "loss": 0.3776,
      "step": 10880
    },
    {
      "epoch": 0.8234993614303959,
      "grad_norm": 0.7734375,
      "learning_rate": 1.2770644094519973e-05,
      "loss": 0.3213,
      "step": 10881
    },
    {
      "epoch": 0.8235750437538433,
      "grad_norm": 0.671875,
      "learning_rate": 1.2769499624029288e-05,
      "loss": 0.2651,
      "step": 10882
    },
    {
      "epoch": 0.8236507260772906,
      "grad_norm": 0.796875,
      "learning_rate": 1.2768355114248493e-05,
      "loss": 0.3141,
      "step": 10883
    },
    {
      "epoch": 0.8237264084007379,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2767210565193838e-05,
      "loss": 0.2994,
      "step": 10884
    },
    {
      "epoch": 0.8238020907241852,
      "grad_norm": 0.75,
      "learning_rate": 1.2766065976881547e-05,
      "loss": 0.299,
      "step": 10885
    },
    {
      "epoch": 0.8238777730476325,
      "grad_norm": 0.73046875,
      "learning_rate": 1.2764921349327864e-05,
      "loss": 0.2884,
      "step": 10886
    },
    {
      "epoch": 0.8239534553710799,
      "grad_norm": 0.94921875,
      "learning_rate": 1.2763776682549028e-05,
      "loss": 0.3389,
      "step": 10887
    },
    {
      "epoch": 0.8240291376945272,
      "grad_norm": 0.71875,
      "learning_rate": 1.2762631976561277e-05,
      "loss": 0.289,
      "step": 10888
    },
    {
      "epoch": 0.8241048200179746,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2761487231380849e-05,
      "loss": 0.3033,
      "step": 10889
    },
    {
      "epoch": 0.8241805023414219,
      "grad_norm": 0.73046875,
      "learning_rate": 1.2760342447023986e-05,
      "loss": 0.3054,
      "step": 10890
    },
    {
      "epoch": 0.8242561846648692,
      "grad_norm": 0.6953125,
      "learning_rate": 1.2759197623506929e-05,
      "loss": 0.2757,
      "step": 10891
    },
    {
      "epoch": 0.8243318669883165,
      "grad_norm": 0.75390625,
      "learning_rate": 1.275805276084592e-05,
      "loss": 0.3311,
      "step": 10892
    },
    {
      "epoch": 0.8244075493117639,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2756907859057199e-05,
      "loss": 0.3279,
      "step": 10893
    },
    {
      "epoch": 0.8244832316352112,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2755762918157008e-05,
      "loss": 0.3508,
      "step": 10894
    },
    {
      "epoch": 0.8245589139586585,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2754617938161591e-05,
      "loss": 0.3229,
      "step": 10895
    },
    {
      "epoch": 0.8246345962821059,
      "grad_norm": 0.7109375,
      "learning_rate": 1.2753472919087192e-05,
      "loss": 0.2622,
      "step": 10896
    },
    {
      "epoch": 0.8247102786055532,
      "grad_norm": 0.69140625,
      "learning_rate": 1.2752327860950056e-05,
      "loss": 0.2583,
      "step": 10897
    },
    {
      "epoch": 0.8247859609290005,
      "grad_norm": 0.8046875,
      "learning_rate": 1.2751182763766425e-05,
      "loss": 0.3206,
      "step": 10898
    },
    {
      "epoch": 0.8248616432524478,
      "grad_norm": 0.71875,
      "learning_rate": 1.2750037627552544e-05,
      "loss": 0.2646,
      "step": 10899
    },
    {
      "epoch": 0.8249373255758952,
      "grad_norm": 0.7734375,
      "learning_rate": 1.2748892452324661e-05,
      "loss": 0.3338,
      "step": 10900
    },
    {
      "epoch": 0.8250130078993425,
      "grad_norm": 0.6953125,
      "learning_rate": 1.2747747238099022e-05,
      "loss": 0.2929,
      "step": 10901
    },
    {
      "epoch": 0.8250886902227899,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2746601984891873e-05,
      "loss": 0.3307,
      "step": 10902
    },
    {
      "epoch": 0.8251643725462372,
      "grad_norm": 0.79296875,
      "learning_rate": 1.2745456692719458e-05,
      "loss": 0.3204,
      "step": 10903
    },
    {
      "epoch": 0.8252400548696845,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2744311361598034e-05,
      "loss": 0.3469,
      "step": 10904
    },
    {
      "epoch": 0.8253157371931318,
      "grad_norm": 0.76171875,
      "learning_rate": 1.274316599154384e-05,
      "loss": 0.3128,
      "step": 10905
    },
    {
      "epoch": 0.8253914195165791,
      "grad_norm": 0.84375,
      "learning_rate": 1.2742020582573131e-05,
      "loss": 0.3833,
      "step": 10906
    },
    {
      "epoch": 0.8254671018400265,
      "grad_norm": 0.76171875,
      "learning_rate": 1.2740875134702156e-05,
      "loss": 0.3134,
      "step": 10907
    },
    {
      "epoch": 0.8255427841634738,
      "grad_norm": 0.671875,
      "learning_rate": 1.2739729647947159e-05,
      "loss": 0.2493,
      "step": 10908
    },
    {
      "epoch": 0.8256184664869212,
      "grad_norm": 0.74609375,
      "learning_rate": 1.27385841223244e-05,
      "loss": 0.3137,
      "step": 10909
    },
    {
      "epoch": 0.8256941488103685,
      "grad_norm": 0.80859375,
      "learning_rate": 1.273743855785012e-05,
      "loss": 0.3644,
      "step": 10910
    },
    {
      "epoch": 0.8257698311338159,
      "grad_norm": 0.765625,
      "learning_rate": 1.2736292954540583e-05,
      "loss": 0.3163,
      "step": 10911
    },
    {
      "epoch": 0.8258455134572631,
      "grad_norm": 0.71875,
      "learning_rate": 1.273514731241203e-05,
      "loss": 0.2711,
      "step": 10912
    },
    {
      "epoch": 0.8259211957807104,
      "grad_norm": 0.734375,
      "learning_rate": 1.2734001631480721e-05,
      "loss": 0.3059,
      "step": 10913
    },
    {
      "epoch": 0.8259968781041578,
      "grad_norm": 0.73828125,
      "learning_rate": 1.2732855911762907e-05,
      "loss": 0.3318,
      "step": 10914
    },
    {
      "epoch": 0.8260725604276051,
      "grad_norm": 0.796875,
      "learning_rate": 1.2731710153274843e-05,
      "loss": 0.3295,
      "step": 10915
    },
    {
      "epoch": 0.8261482427510525,
      "grad_norm": 0.7109375,
      "learning_rate": 1.2730564356032782e-05,
      "loss": 0.2801,
      "step": 10916
    },
    {
      "epoch": 0.8262239250744998,
      "grad_norm": 0.78515625,
      "learning_rate": 1.272941852005298e-05,
      "loss": 0.3482,
      "step": 10917
    },
    {
      "epoch": 0.8262996073979472,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2728272645351688e-05,
      "loss": 0.3255,
      "step": 10918
    },
    {
      "epoch": 0.8263752897213944,
      "grad_norm": 0.734375,
      "learning_rate": 1.2727126731945171e-05,
      "loss": 0.3031,
      "step": 10919
    },
    {
      "epoch": 0.8264509720448417,
      "grad_norm": 0.7734375,
      "learning_rate": 1.2725980779849685e-05,
      "loss": 0.305,
      "step": 10920
    },
    {
      "epoch": 0.8265266543682891,
      "grad_norm": 0.69140625,
      "learning_rate": 1.2724834789081477e-05,
      "loss": 0.2673,
      "step": 10921
    },
    {
      "epoch": 0.8266023366917364,
      "grad_norm": 0.71484375,
      "learning_rate": 1.2723688759656817e-05,
      "loss": 0.3052,
      "step": 10922
    },
    {
      "epoch": 0.8266780190151838,
      "grad_norm": 0.765625,
      "learning_rate": 1.2722542691591953e-05,
      "loss": 0.3261,
      "step": 10923
    },
    {
      "epoch": 0.8267537013386311,
      "grad_norm": 0.71484375,
      "learning_rate": 1.2721396584903149e-05,
      "loss": 0.3354,
      "step": 10924
    },
    {
      "epoch": 0.8268293836620785,
      "grad_norm": 0.8125,
      "learning_rate": 1.272025043960667e-05,
      "loss": 0.331,
      "step": 10925
    },
    {
      "epoch": 0.8269050659855257,
      "grad_norm": 0.79296875,
      "learning_rate": 1.2719104255718767e-05,
      "loss": 0.3358,
      "step": 10926
    },
    {
      "epoch": 0.8269807483089731,
      "grad_norm": 0.796875,
      "learning_rate": 1.2717958033255702e-05,
      "loss": 0.311,
      "step": 10927
    },
    {
      "epoch": 0.8270564306324204,
      "grad_norm": 0.76953125,
      "learning_rate": 1.271681177223374e-05,
      "loss": 0.3503,
      "step": 10928
    },
    {
      "epoch": 0.8271321129558677,
      "grad_norm": 0.80859375,
      "learning_rate": 1.271566547266914e-05,
      "loss": 0.3336,
      "step": 10929
    },
    {
      "epoch": 0.8272077952793151,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2714519134578165e-05,
      "loss": 0.3129,
      "step": 10930
    },
    {
      "epoch": 0.8272834776027624,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2713372757977078e-05,
      "loss": 0.3071,
      "step": 10931
    },
    {
      "epoch": 0.8273591599262098,
      "grad_norm": 0.80078125,
      "learning_rate": 1.2712226342882143e-05,
      "loss": 0.3394,
      "step": 10932
    },
    {
      "epoch": 0.827434842249657,
      "grad_norm": 0.765625,
      "learning_rate": 1.2711079889309624e-05,
      "loss": 0.2937,
      "step": 10933
    },
    {
      "epoch": 0.8275105245731044,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2709933397275785e-05,
      "loss": 0.3237,
      "step": 10934
    },
    {
      "epoch": 0.8275862068965517,
      "grad_norm": 0.7109375,
      "learning_rate": 1.2708786866796887e-05,
      "loss": 0.2952,
      "step": 10935
    },
    {
      "epoch": 0.8276618892199991,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2707640297889198e-05,
      "loss": 0.3418,
      "step": 10936
    },
    {
      "epoch": 0.8277375715434464,
      "grad_norm": 0.72265625,
      "learning_rate": 1.2706493690568988e-05,
      "loss": 0.3087,
      "step": 10937
    },
    {
      "epoch": 0.8278132538668938,
      "grad_norm": 0.7890625,
      "learning_rate": 1.2705347044852521e-05,
      "loss": 0.3704,
      "step": 10938
    },
    {
      "epoch": 0.8278889361903411,
      "grad_norm": 0.7109375,
      "learning_rate": 1.270420036075606e-05,
      "loss": 0.2831,
      "step": 10939
    },
    {
      "epoch": 0.8279646185137883,
      "grad_norm": 0.703125,
      "learning_rate": 1.2703053638295879e-05,
      "loss": 0.2531,
      "step": 10940
    },
    {
      "epoch": 0.8280403008372357,
      "grad_norm": 0.80859375,
      "learning_rate": 1.2701906877488243e-05,
      "loss": 0.3572,
      "step": 10941
    },
    {
      "epoch": 0.828115983160683,
      "grad_norm": 0.7734375,
      "learning_rate": 1.270076007834942e-05,
      "loss": 0.3074,
      "step": 10942
    },
    {
      "epoch": 0.8281916654841304,
      "grad_norm": 0.7265625,
      "learning_rate": 1.2699613240895686e-05,
      "loss": 0.3177,
      "step": 10943
    },
    {
      "epoch": 0.8282673478075777,
      "grad_norm": 0.76171875,
      "learning_rate": 1.26984663651433e-05,
      "loss": 0.2972,
      "step": 10944
    },
    {
      "epoch": 0.8283430301310251,
      "grad_norm": 0.83984375,
      "learning_rate": 1.2697319451108538e-05,
      "loss": 0.3791,
      "step": 10945
    },
    {
      "epoch": 0.8284187124544724,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2696172498807673e-05,
      "loss": 0.3263,
      "step": 10946
    },
    {
      "epoch": 0.8284943947779196,
      "grad_norm": 0.7265625,
      "learning_rate": 1.2695025508256973e-05,
      "loss": 0.2942,
      "step": 10947
    },
    {
      "epoch": 0.828570077101367,
      "grad_norm": 0.80859375,
      "learning_rate": 1.2693878479472712e-05,
      "loss": 0.356,
      "step": 10948
    },
    {
      "epoch": 0.8286457594248143,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2692731412471163e-05,
      "loss": 0.3022,
      "step": 10949
    },
    {
      "epoch": 0.8287214417482617,
      "grad_norm": 0.703125,
      "learning_rate": 1.2691584307268596e-05,
      "loss": 0.3038,
      "step": 10950
    },
    {
      "epoch": 0.828797124071709,
      "grad_norm": 0.84375,
      "learning_rate": 1.2690437163881286e-05,
      "loss": 0.3326,
      "step": 10951
    },
    {
      "epoch": 0.8288728063951564,
      "grad_norm": 0.75,
      "learning_rate": 1.268928998232551e-05,
      "loss": 0.3065,
      "step": 10952
    },
    {
      "epoch": 0.8289484887186037,
      "grad_norm": 0.71875,
      "learning_rate": 1.2688142762617542e-05,
      "loss": 0.2905,
      "step": 10953
    },
    {
      "epoch": 0.829024171042051,
      "grad_norm": 0.7109375,
      "learning_rate": 1.2686995504773654e-05,
      "loss": 0.2851,
      "step": 10954
    },
    {
      "epoch": 0.8290998533654983,
      "grad_norm": 0.73046875,
      "learning_rate": 1.2685848208810123e-05,
      "loss": 0.2672,
      "step": 10955
    },
    {
      "epoch": 0.8291755356889456,
      "grad_norm": 0.8203125,
      "learning_rate": 1.2684700874743228e-05,
      "loss": 0.3556,
      "step": 10956
    },
    {
      "epoch": 0.829251218012393,
      "grad_norm": 0.73046875,
      "learning_rate": 1.268355350258924e-05,
      "loss": 0.3058,
      "step": 10957
    },
    {
      "epoch": 0.8293269003358403,
      "grad_norm": 0.79296875,
      "learning_rate": 1.2682406092364446e-05,
      "loss": 0.3408,
      "step": 10958
    },
    {
      "epoch": 0.8294025826592877,
      "grad_norm": 0.78515625,
      "learning_rate": 1.2681258644085112e-05,
      "loss": 0.3341,
      "step": 10959
    },
    {
      "epoch": 0.829478264982735,
      "grad_norm": 0.7109375,
      "learning_rate": 1.2680111157767528e-05,
      "loss": 0.2773,
      "step": 10960
    },
    {
      "epoch": 0.8295539473061823,
      "grad_norm": 0.8125,
      "learning_rate": 1.2678963633427966e-05,
      "loss": 0.3601,
      "step": 10961
    },
    {
      "epoch": 0.8296296296296296,
      "grad_norm": 0.75390625,
      "learning_rate": 1.267781607108271e-05,
      "loss": 0.29,
      "step": 10962
    },
    {
      "epoch": 0.829705311953077,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2676668470748035e-05,
      "loss": 0.3076,
      "step": 10963
    },
    {
      "epoch": 0.8297809942765243,
      "grad_norm": 0.77734375,
      "learning_rate": 1.2675520832440225e-05,
      "loss": 0.3381,
      "step": 10964
    },
    {
      "epoch": 0.8298566765999716,
      "grad_norm": 0.7890625,
      "learning_rate": 1.2674373156175562e-05,
      "loss": 0.3339,
      "step": 10965
    },
    {
      "epoch": 0.829932358923419,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2673225441970322e-05,
      "loss": 0.3344,
      "step": 10966
    },
    {
      "epoch": 0.8300080412468662,
      "grad_norm": 0.68359375,
      "learning_rate": 1.2672077689840799e-05,
      "loss": 0.2753,
      "step": 10967
    },
    {
      "epoch": 0.8300837235703136,
      "grad_norm": 0.76171875,
      "learning_rate": 1.2670929899803261e-05,
      "loss": 0.3212,
      "step": 10968
    },
    {
      "epoch": 0.8301594058937609,
      "grad_norm": 0.7890625,
      "learning_rate": 1.2669782071874004e-05,
      "loss": 0.3166,
      "step": 10969
    },
    {
      "epoch": 0.8302350882172083,
      "grad_norm": 0.83984375,
      "learning_rate": 1.2668634206069305e-05,
      "loss": 0.3597,
      "step": 10970
    },
    {
      "epoch": 0.8303107705406556,
      "grad_norm": 0.7421875,
      "learning_rate": 1.266748630240545e-05,
      "loss": 0.2888,
      "step": 10971
    },
    {
      "epoch": 0.830386452864103,
      "grad_norm": 0.76171875,
      "learning_rate": 1.2666338360898724e-05,
      "loss": 0.3285,
      "step": 10972
    },
    {
      "epoch": 0.8304621351875503,
      "grad_norm": 0.78125,
      "learning_rate": 1.2665190381565415e-05,
      "loss": 0.3154,
      "step": 10973
    },
    {
      "epoch": 0.8305378175109975,
      "grad_norm": 0.75,
      "learning_rate": 1.2664042364421803e-05,
      "loss": 0.331,
      "step": 10974
    },
    {
      "epoch": 0.8306134998344449,
      "grad_norm": 0.765625,
      "learning_rate": 1.2662894309484181e-05,
      "loss": 0.3018,
      "step": 10975
    },
    {
      "epoch": 0.8306891821578922,
      "grad_norm": 0.73046875,
      "learning_rate": 1.2661746216768835e-05,
      "loss": 0.2972,
      "step": 10976
    },
    {
      "epoch": 0.8307648644813396,
      "grad_norm": 0.8046875,
      "learning_rate": 1.2660598086292049e-05,
      "loss": 0.3014,
      "step": 10977
    },
    {
      "epoch": 0.8308405468047869,
      "grad_norm": 0.9375,
      "learning_rate": 1.2659449918070114e-05,
      "loss": 0.3331,
      "step": 10978
    },
    {
      "epoch": 0.8309162291282343,
      "grad_norm": 0.7265625,
      "learning_rate": 1.2658301712119316e-05,
      "loss": 0.2878,
      "step": 10979
    },
    {
      "epoch": 0.8309919114516816,
      "grad_norm": 0.73046875,
      "learning_rate": 1.2657153468455947e-05,
      "loss": 0.3216,
      "step": 10980
    },
    {
      "epoch": 0.8310675937751288,
      "grad_norm": 0.72265625,
      "learning_rate": 1.26560051870963e-05,
      "loss": 0.3075,
      "step": 10981
    },
    {
      "epoch": 0.8311432760985762,
      "grad_norm": 0.78515625,
      "learning_rate": 1.2654856868056662e-05,
      "loss": 0.3285,
      "step": 10982
    },
    {
      "epoch": 0.8312189584220235,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2653708511353318e-05,
      "loss": 0.3026,
      "step": 10983
    },
    {
      "epoch": 0.8312946407454709,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2652560117002567e-05,
      "loss": 0.3188,
      "step": 10984
    },
    {
      "epoch": 0.8313703230689182,
      "grad_norm": 0.79296875,
      "learning_rate": 1.26514116850207e-05,
      "loss": 0.3329,
      "step": 10985
    },
    {
      "epoch": 0.8314460053923656,
      "grad_norm": 0.84765625,
      "learning_rate": 1.2650263215424008e-05,
      "loss": 0.285,
      "step": 10986
    },
    {
      "epoch": 0.8315216877158129,
      "grad_norm": 0.73046875,
      "learning_rate": 1.2649114708228783e-05,
      "loss": 0.2734,
      "step": 10987
    },
    {
      "epoch": 0.8315973700392602,
      "grad_norm": 1.0546875,
      "learning_rate": 1.2647966163451321e-05,
      "loss": 0.3843,
      "step": 10988
    },
    {
      "epoch": 0.8316730523627075,
      "grad_norm": 0.71875,
      "learning_rate": 1.2646817581107913e-05,
      "loss": 0.3214,
      "step": 10989
    },
    {
      "epoch": 0.8317487346861548,
      "grad_norm": 0.73046875,
      "learning_rate": 1.2645668961214858e-05,
      "loss": 0.282,
      "step": 10990
    },
    {
      "epoch": 0.8318244170096022,
      "grad_norm": 0.765625,
      "learning_rate": 1.2644520303788448e-05,
      "loss": 0.323,
      "step": 10991
    },
    {
      "epoch": 0.8319000993330495,
      "grad_norm": 0.78125,
      "learning_rate": 1.264337160884498e-05,
      "loss": 0.3102,
      "step": 10992
    },
    {
      "epoch": 0.8319757816564969,
      "grad_norm": 1.125,
      "learning_rate": 1.2642222876400748e-05,
      "loss": 0.3456,
      "step": 10993
    },
    {
      "epoch": 0.8320514639799442,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2641074106472049e-05,
      "loss": 0.3231,
      "step": 10994
    },
    {
      "epoch": 0.8321271463033915,
      "grad_norm": 0.82421875,
      "learning_rate": 1.2639925299075185e-05,
      "loss": 0.3734,
      "step": 10995
    },
    {
      "epoch": 0.8322028286268388,
      "grad_norm": 0.83203125,
      "learning_rate": 1.2638776454226449e-05,
      "loss": 0.3897,
      "step": 10996
    },
    {
      "epoch": 0.8322785109502862,
      "grad_norm": 0.76171875,
      "learning_rate": 1.263762757194214e-05,
      "loss": 0.3219,
      "step": 10997
    },
    {
      "epoch": 0.8323541932737335,
      "grad_norm": 0.8515625,
      "learning_rate": 1.2636478652238556e-05,
      "loss": 0.349,
      "step": 10998
    },
    {
      "epoch": 0.8324298755971808,
      "grad_norm": 0.75,
      "learning_rate": 1.2635329695132003e-05,
      "loss": 0.338,
      "step": 10999
    },
    {
      "epoch": 0.8325055579206282,
      "grad_norm": 0.84765625,
      "learning_rate": 1.2634180700638771e-05,
      "loss": 0.3104,
      "step": 11000
    },
    {
      "epoch": 0.8325812402440755,
      "grad_norm": 0.671875,
      "learning_rate": 1.2633031668775169e-05,
      "loss": 0.2478,
      "step": 11001
    },
    {
      "epoch": 0.8326569225675228,
      "grad_norm": 0.75390625,
      "learning_rate": 1.263188259955749e-05,
      "loss": 0.3185,
      "step": 11002
    },
    {
      "epoch": 0.8327326048909701,
      "grad_norm": 0.75,
      "learning_rate": 1.2630733493002043e-05,
      "loss": 0.3199,
      "step": 11003
    },
    {
      "epoch": 0.8328082872144175,
      "grad_norm": 0.7109375,
      "learning_rate": 1.2629584349125129e-05,
      "loss": 0.3107,
      "step": 11004
    },
    {
      "epoch": 0.8328839695378648,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2628435167943045e-05,
      "loss": 0.3175,
      "step": 11005
    },
    {
      "epoch": 0.8329596518613122,
      "grad_norm": 0.79296875,
      "learning_rate": 1.26272859494721e-05,
      "loss": 0.3809,
      "step": 11006
    },
    {
      "epoch": 0.8330353341847595,
      "grad_norm": 0.796875,
      "learning_rate": 1.262613669372859e-05,
      "loss": 0.3378,
      "step": 11007
    },
    {
      "epoch": 0.8331110165082068,
      "grad_norm": 0.73046875,
      "learning_rate": 1.2624987400728829e-05,
      "loss": 0.3084,
      "step": 11008
    },
    {
      "epoch": 0.8331866988316541,
      "grad_norm": 0.72265625,
      "learning_rate": 1.2623838070489116e-05,
      "loss": 0.3139,
      "step": 11009
    },
    {
      "epoch": 0.8332623811551014,
      "grad_norm": 0.73046875,
      "learning_rate": 1.2622688703025758e-05,
      "loss": 0.297,
      "step": 11010
    },
    {
      "epoch": 0.8333380634785488,
      "grad_norm": 0.73046875,
      "learning_rate": 1.262153929835506e-05,
      "loss": 0.3079,
      "step": 11011
    },
    {
      "epoch": 0.8334137458019961,
      "grad_norm": 0.8046875,
      "learning_rate": 1.2620389856493326e-05,
      "loss": 0.3295,
      "step": 11012
    },
    {
      "epoch": 0.8334894281254435,
      "grad_norm": 0.7890625,
      "learning_rate": 1.2619240377456869e-05,
      "loss": 0.3259,
      "step": 11013
    },
    {
      "epoch": 0.8335651104488908,
      "grad_norm": 0.80859375,
      "learning_rate": 1.261809086126199e-05,
      "loss": 0.3726,
      "step": 11014
    },
    {
      "epoch": 0.8336407927723382,
      "grad_norm": 1.078125,
      "learning_rate": 1.2616941307925e-05,
      "loss": 0.3871,
      "step": 11015
    },
    {
      "epoch": 0.8337164750957854,
      "grad_norm": 0.71484375,
      "learning_rate": 1.2615791717462205e-05,
      "loss": 0.2643,
      "step": 11016
    },
    {
      "epoch": 0.8337921574192327,
      "grad_norm": 0.78515625,
      "learning_rate": 1.2614642089889917e-05,
      "loss": 0.3491,
      "step": 11017
    },
    {
      "epoch": 0.8338678397426801,
      "grad_norm": 0.71484375,
      "learning_rate": 1.2613492425224445e-05,
      "loss": 0.3049,
      "step": 11018
    },
    {
      "epoch": 0.8339435220661274,
      "grad_norm": 0.77734375,
      "learning_rate": 1.2612342723482099e-05,
      "loss": 0.3485,
      "step": 11019
    },
    {
      "epoch": 0.8340192043895748,
      "grad_norm": 0.765625,
      "learning_rate": 1.2611192984679186e-05,
      "loss": 0.3209,
      "step": 11020
    },
    {
      "epoch": 0.8340948867130221,
      "grad_norm": 0.79296875,
      "learning_rate": 1.2610043208832021e-05,
      "loss": 0.3193,
      "step": 11021
    },
    {
      "epoch": 0.8341705690364695,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2608893395956911e-05,
      "loss": 0.3104,
      "step": 11022
    },
    {
      "epoch": 0.8342462513599167,
      "grad_norm": 0.72265625,
      "learning_rate": 1.2607743546070176e-05,
      "loss": 0.3074,
      "step": 11023
    },
    {
      "epoch": 0.834321933683364,
      "grad_norm": 0.7265625,
      "learning_rate": 1.260659365918812e-05,
      "loss": 0.285,
      "step": 11024
    },
    {
      "epoch": 0.8343976160068114,
      "grad_norm": 0.75,
      "learning_rate": 1.2605443735327061e-05,
      "loss": 0.337,
      "step": 11025
    },
    {
      "epoch": 0.8344732983302587,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2604293774503311e-05,
      "loss": 0.322,
      "step": 11026
    },
    {
      "epoch": 0.8345489806537061,
      "grad_norm": 0.8203125,
      "learning_rate": 1.2603143776733186e-05,
      "loss": 0.3388,
      "step": 11027
    },
    {
      "epoch": 0.8346246629771534,
      "grad_norm": 0.828125,
      "learning_rate": 1.2601993742033001e-05,
      "loss": 0.3635,
      "step": 11028
    },
    {
      "epoch": 0.8347003453006008,
      "grad_norm": 0.78515625,
      "learning_rate": 1.2600843670419067e-05,
      "loss": 0.3269,
      "step": 11029
    },
    {
      "epoch": 0.834776027624048,
      "grad_norm": 0.91015625,
      "learning_rate": 1.25996935619077e-05,
      "loss": 0.3114,
      "step": 11030
    },
    {
      "epoch": 0.8348517099474954,
      "grad_norm": 0.75,
      "learning_rate": 1.259854341651522e-05,
      "loss": 0.2922,
      "step": 11031
    },
    {
      "epoch": 0.8349273922709427,
      "grad_norm": 0.71875,
      "learning_rate": 1.259739323425795e-05,
      "loss": 0.291,
      "step": 11032
    },
    {
      "epoch": 0.83500307459439,
      "grad_norm": 0.765625,
      "learning_rate": 1.259624301515219e-05,
      "loss": 0.3285,
      "step": 11033
    },
    {
      "epoch": 0.8350787569178374,
      "grad_norm": 0.78125,
      "learning_rate": 1.2595092759214271e-05,
      "loss": 0.321,
      "step": 11034
    },
    {
      "epoch": 0.8351544392412847,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2593942466460507e-05,
      "loss": 0.3156,
      "step": 11035
    },
    {
      "epoch": 0.8352301215647321,
      "grad_norm": 0.83984375,
      "learning_rate": 1.2592792136907217e-05,
      "loss": 0.3776,
      "step": 11036
    },
    {
      "epoch": 0.8353058038881793,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2591641770570726e-05,
      "loss": 0.3243,
      "step": 11037
    },
    {
      "epoch": 0.8353814862116267,
      "grad_norm": 0.859375,
      "learning_rate": 1.2590491367467346e-05,
      "loss": 0.3897,
      "step": 11038
    },
    {
      "epoch": 0.835457168535074,
      "grad_norm": 0.79296875,
      "learning_rate": 1.2589340927613398e-05,
      "loss": 0.3634,
      "step": 11039
    },
    {
      "epoch": 0.8355328508585214,
      "grad_norm": 0.859375,
      "learning_rate": 1.2588190451025209e-05,
      "loss": 0.3687,
      "step": 11040
    },
    {
      "epoch": 0.8356085331819687,
      "grad_norm": 0.73828125,
      "learning_rate": 1.2587039937719098e-05,
      "loss": 0.3191,
      "step": 11041
    },
    {
      "epoch": 0.835684215505416,
      "grad_norm": 0.7265625,
      "learning_rate": 1.258588938771138e-05,
      "loss": 0.2961,
      "step": 11042
    },
    {
      "epoch": 0.8357598978288634,
      "grad_norm": 0.69140625,
      "learning_rate": 1.258473880101839e-05,
      "loss": 0.2849,
      "step": 11043
    },
    {
      "epoch": 0.8358355801523106,
      "grad_norm": 0.8203125,
      "learning_rate": 1.2583588177656441e-05,
      "loss": 0.3674,
      "step": 11044
    },
    {
      "epoch": 0.835911262475758,
      "grad_norm": 0.76953125,
      "learning_rate": 1.258243751764186e-05,
      "loss": 0.3207,
      "step": 11045
    },
    {
      "epoch": 0.8359869447992053,
      "grad_norm": 0.8046875,
      "learning_rate": 1.2581286820990975e-05,
      "loss": 0.3377,
      "step": 11046
    },
    {
      "epoch": 0.8360626271226527,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2580136087720103e-05,
      "loss": 0.3214,
      "step": 11047
    },
    {
      "epoch": 0.8361383094461,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2578985317845571e-05,
      "loss": 0.3587,
      "step": 11048
    },
    {
      "epoch": 0.8362139917695474,
      "grad_norm": 0.76953125,
      "learning_rate": 1.257783451138371e-05,
      "loss": 0.3206,
      "step": 11049
    },
    {
      "epoch": 0.8362896740929947,
      "grad_norm": 0.71484375,
      "learning_rate": 1.2576683668350842e-05,
      "loss": 0.3076,
      "step": 11050
    },
    {
      "epoch": 0.8363653564164419,
      "grad_norm": 0.78125,
      "learning_rate": 1.2575532788763295e-05,
      "loss": 0.3663,
      "step": 11051
    },
    {
      "epoch": 0.8364410387398893,
      "grad_norm": 0.73046875,
      "learning_rate": 1.2574381872637395e-05,
      "loss": 0.2816,
      "step": 11052
    },
    {
      "epoch": 0.8365167210633366,
      "grad_norm": 0.734375,
      "learning_rate": 1.2573230919989473e-05,
      "loss": 0.2988,
      "step": 11053
    },
    {
      "epoch": 0.836592403386784,
      "grad_norm": 0.7890625,
      "learning_rate": 1.257207993083585e-05,
      "loss": 0.3194,
      "step": 11054
    },
    {
      "epoch": 0.8366680857102313,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2570928905192862e-05,
      "loss": 0.2936,
      "step": 11055
    },
    {
      "epoch": 0.8367437680336787,
      "grad_norm": 0.81640625,
      "learning_rate": 1.2569777843076836e-05,
      "loss": 0.3398,
      "step": 11056
    },
    {
      "epoch": 0.836819450357126,
      "grad_norm": 0.796875,
      "learning_rate": 1.2568626744504098e-05,
      "loss": 0.3389,
      "step": 11057
    },
    {
      "epoch": 0.8368951326805733,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2567475609490986e-05,
      "loss": 0.2842,
      "step": 11058
    },
    {
      "epoch": 0.8369708150040206,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2566324438053825e-05,
      "loss": 0.304,
      "step": 11059
    },
    {
      "epoch": 0.8370464973274679,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2565173230208946e-05,
      "loss": 0.2981,
      "step": 11060
    },
    {
      "epoch": 0.8371221796509153,
      "grad_norm": 0.78125,
      "learning_rate": 1.2564021985972686e-05,
      "loss": 0.3409,
      "step": 11061
    },
    {
      "epoch": 0.8371978619743626,
      "grad_norm": 0.671875,
      "learning_rate": 1.2562870705361372e-05,
      "loss": 0.242,
      "step": 11062
    },
    {
      "epoch": 0.83727354429781,
      "grad_norm": 0.71875,
      "learning_rate": 1.2561719388391338e-05,
      "loss": 0.316,
      "step": 11063
    },
    {
      "epoch": 0.8373492266212573,
      "grad_norm": 0.78515625,
      "learning_rate": 1.256056803507892e-05,
      "loss": 0.3518,
      "step": 11064
    },
    {
      "epoch": 0.8374249089447046,
      "grad_norm": 0.69921875,
      "learning_rate": 1.2559416645440452e-05,
      "loss": 0.292,
      "step": 11065
    },
    {
      "epoch": 0.8375005912681519,
      "grad_norm": 0.77734375,
      "learning_rate": 1.2558265219492265e-05,
      "loss": 0.3576,
      "step": 11066
    },
    {
      "epoch": 0.8375762735915993,
      "grad_norm": 0.765625,
      "learning_rate": 1.2557113757250697e-05,
      "loss": 0.3362,
      "step": 11067
    },
    {
      "epoch": 0.8376519559150466,
      "grad_norm": 0.73828125,
      "learning_rate": 1.255596225873208e-05,
      "loss": 0.3179,
      "step": 11068
    },
    {
      "epoch": 0.8377276382384939,
      "grad_norm": 0.70703125,
      "learning_rate": 1.2554810723952752e-05,
      "loss": 0.3084,
      "step": 11069
    },
    {
      "epoch": 0.8378033205619413,
      "grad_norm": 0.82421875,
      "learning_rate": 1.2553659152929052e-05,
      "loss": 0.3561,
      "step": 11070
    },
    {
      "epoch": 0.8378790028853886,
      "grad_norm": 0.85546875,
      "learning_rate": 1.2552507545677313e-05,
      "loss": 0.3872,
      "step": 11071
    },
    {
      "epoch": 0.8379546852088359,
      "grad_norm": 0.77734375,
      "learning_rate": 1.2551355902213873e-05,
      "loss": 0.3453,
      "step": 11072
    },
    {
      "epoch": 0.8380303675322832,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2550204222555074e-05,
      "loss": 0.3039,
      "step": 11073
    },
    {
      "epoch": 0.8381060498557306,
      "grad_norm": 0.83203125,
      "learning_rate": 1.2549052506717249e-05,
      "loss": 0.3744,
      "step": 11074
    },
    {
      "epoch": 0.8381817321791779,
      "grad_norm": 0.75,
      "learning_rate": 1.2547900754716745e-05,
      "loss": 0.3328,
      "step": 11075
    },
    {
      "epoch": 0.8382574145026253,
      "grad_norm": 0.78515625,
      "learning_rate": 1.2546748966569894e-05,
      "loss": 0.317,
      "step": 11076
    },
    {
      "epoch": 0.8383330968260726,
      "grad_norm": 0.75,
      "learning_rate": 1.2545597142293033e-05,
      "loss": 0.2962,
      "step": 11077
    },
    {
      "epoch": 0.83840877914952,
      "grad_norm": 0.76171875,
      "learning_rate": 1.2544445281902514e-05,
      "loss": 0.3388,
      "step": 11078
    },
    {
      "epoch": 0.8384844614729672,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2543293385414672e-05,
      "loss": 0.3106,
      "step": 11079
    },
    {
      "epoch": 0.8385601437964145,
      "grad_norm": 0.7734375,
      "learning_rate": 1.2542141452845847e-05,
      "loss": 0.3231,
      "step": 11080
    },
    {
      "epoch": 0.8386358261198619,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2540989484212385e-05,
      "loss": 0.3493,
      "step": 11081
    },
    {
      "epoch": 0.8387115084433092,
      "grad_norm": 0.65234375,
      "learning_rate": 1.2539837479530626e-05,
      "loss": 0.2255,
      "step": 11082
    },
    {
      "epoch": 0.8387871907667566,
      "grad_norm": 0.77734375,
      "learning_rate": 1.2538685438816914e-05,
      "loss": 0.3089,
      "step": 11083
    },
    {
      "epoch": 0.8388628730902039,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2537533362087591e-05,
      "loss": 0.3265,
      "step": 11084
    },
    {
      "epoch": 0.8389385554136511,
      "grad_norm": 0.82421875,
      "learning_rate": 1.2536381249359004e-05,
      "loss": 0.3584,
      "step": 11085
    },
    {
      "epoch": 0.8390142377370985,
      "grad_norm": 0.8125,
      "learning_rate": 1.2535229100647493e-05,
      "loss": 0.3838,
      "step": 11086
    },
    {
      "epoch": 0.8390899200605458,
      "grad_norm": 0.72265625,
      "learning_rate": 1.2534076915969412e-05,
      "loss": 0.2831,
      "step": 11087
    },
    {
      "epoch": 0.8391656023839932,
      "grad_norm": 0.73828125,
      "learning_rate": 1.25329246953411e-05,
      "loss": 0.2943,
      "step": 11088
    },
    {
      "epoch": 0.8392412847074405,
      "grad_norm": 0.9296875,
      "learning_rate": 1.2531772438778902e-05,
      "loss": 0.2902,
      "step": 11089
    },
    {
      "epoch": 0.8393169670308879,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2530620146299168e-05,
      "loss": 0.3018,
      "step": 11090
    },
    {
      "epoch": 0.8393926493543352,
      "grad_norm": 0.6875,
      "learning_rate": 1.2529467817918245e-05,
      "loss": 0.2444,
      "step": 11091
    },
    {
      "epoch": 0.8394683316777825,
      "grad_norm": 0.7578125,
      "learning_rate": 1.252831545365248e-05,
      "loss": 0.3341,
      "step": 11092
    },
    {
      "epoch": 0.8395440140012298,
      "grad_norm": 0.77734375,
      "learning_rate": 1.2527163053518225e-05,
      "loss": 0.2922,
      "step": 11093
    },
    {
      "epoch": 0.8396196963246771,
      "grad_norm": 0.8125,
      "learning_rate": 1.2526010617531822e-05,
      "loss": 0.3447,
      "step": 11094
    },
    {
      "epoch": 0.8396953786481245,
      "grad_norm": 0.80859375,
      "learning_rate": 1.2524858145709622e-05,
      "loss": 0.3351,
      "step": 11095
    },
    {
      "epoch": 0.8397710609715718,
      "grad_norm": 0.7578125,
      "learning_rate": 1.252370563806798e-05,
      "loss": 0.3268,
      "step": 11096
    },
    {
      "epoch": 0.8398467432950192,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2522553094623242e-05,
      "loss": 0.3455,
      "step": 11097
    },
    {
      "epoch": 0.8399224256184665,
      "grad_norm": 0.82421875,
      "learning_rate": 1.2521400515391757e-05,
      "loss": 0.3299,
      "step": 11098
    },
    {
      "epoch": 0.8399981079419138,
      "grad_norm": 0.79296875,
      "learning_rate": 1.2520247900389881e-05,
      "loss": 0.323,
      "step": 11099
    },
    {
      "epoch": 0.8400737902653611,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2519095249633964e-05,
      "loss": 0.3437,
      "step": 11100
    },
    {
      "epoch": 0.8401494725888085,
      "grad_norm": 0.78515625,
      "learning_rate": 1.2517942563140355e-05,
      "loss": 0.3316,
      "step": 11101
    },
    {
      "epoch": 0.8402251549122558,
      "grad_norm": 0.7109375,
      "learning_rate": 1.2516789840925415e-05,
      "loss": 0.2603,
      "step": 11102
    },
    {
      "epoch": 0.8403008372357031,
      "grad_norm": 0.7734375,
      "learning_rate": 1.2515637083005489e-05,
      "loss": 0.3279,
      "step": 11103
    },
    {
      "epoch": 0.8403765195591505,
      "grad_norm": 0.83984375,
      "learning_rate": 1.2514484289396931e-05,
      "loss": 0.3428,
      "step": 11104
    },
    {
      "epoch": 0.8404522018825978,
      "grad_norm": 0.796875,
      "learning_rate": 1.2513331460116102e-05,
      "loss": 0.3651,
      "step": 11105
    },
    {
      "epoch": 0.8405278842060451,
      "grad_norm": 0.703125,
      "learning_rate": 1.2512178595179354e-05,
      "loss": 0.2753,
      "step": 11106
    },
    {
      "epoch": 0.8406035665294924,
      "grad_norm": 0.75,
      "learning_rate": 1.251102569460304e-05,
      "loss": 0.2931,
      "step": 11107
    },
    {
      "epoch": 0.8406792488529398,
      "grad_norm": 0.73828125,
      "learning_rate": 1.2509872758403518e-05,
      "loss": 0.2854,
      "step": 11108
    },
    {
      "epoch": 0.8407549311763871,
      "grad_norm": 0.80859375,
      "learning_rate": 1.2508719786597143e-05,
      "loss": 0.3389,
      "step": 11109
    },
    {
      "epoch": 0.8408306134998345,
      "grad_norm": 0.71875,
      "learning_rate": 1.2507566779200273e-05,
      "loss": 0.2842,
      "step": 11110
    },
    {
      "epoch": 0.8409062958232818,
      "grad_norm": 0.79296875,
      "learning_rate": 1.2506413736229267e-05,
      "loss": 0.3324,
      "step": 11111
    },
    {
      "epoch": 0.8409819781467291,
      "grad_norm": 0.72265625,
      "learning_rate": 1.250526065770048e-05,
      "loss": 0.2917,
      "step": 11112
    },
    {
      "epoch": 0.8410576604701764,
      "grad_norm": 0.765625,
      "learning_rate": 1.2504107543630269e-05,
      "loss": 0.321,
      "step": 11113
    },
    {
      "epoch": 0.8411333427936237,
      "grad_norm": 0.69140625,
      "learning_rate": 1.2502954394034997e-05,
      "loss": 0.2673,
      "step": 11114
    },
    {
      "epoch": 0.8412090251170711,
      "grad_norm": 0.73828125,
      "learning_rate": 1.2501801208931021e-05,
      "loss": 0.3101,
      "step": 11115
    },
    {
      "epoch": 0.8412847074405184,
      "grad_norm": 0.8515625,
      "learning_rate": 1.2500647988334703e-05,
      "loss": 0.3623,
      "step": 11116
    },
    {
      "epoch": 0.8413603897639658,
      "grad_norm": 0.796875,
      "learning_rate": 1.24994947322624e-05,
      "loss": 0.3275,
      "step": 11117
    },
    {
      "epoch": 0.8414360720874131,
      "grad_norm": 0.7265625,
      "learning_rate": 1.2498341440730478e-05,
      "loss": 0.2923,
      "step": 11118
    },
    {
      "epoch": 0.8415117544108605,
      "grad_norm": 0.9375,
      "learning_rate": 1.2497188113755292e-05,
      "loss": 0.3132,
      "step": 11119
    },
    {
      "epoch": 0.8415874367343077,
      "grad_norm": 0.76953125,
      "learning_rate": 1.249603475135321e-05,
      "loss": 0.3279,
      "step": 11120
    },
    {
      "epoch": 0.841663119057755,
      "grad_norm": 0.73046875,
      "learning_rate": 1.249488135354059e-05,
      "loss": 0.2861,
      "step": 11121
    },
    {
      "epoch": 0.8417388013812024,
      "grad_norm": 0.71875,
      "learning_rate": 1.2493727920333799e-05,
      "loss": 0.2852,
      "step": 11122
    },
    {
      "epoch": 0.8418144837046497,
      "grad_norm": 0.8359375,
      "learning_rate": 1.2492574451749195e-05,
      "loss": 0.3019,
      "step": 11123
    },
    {
      "epoch": 0.8418901660280971,
      "grad_norm": 0.71484375,
      "learning_rate": 1.2491420947803148e-05,
      "loss": 0.2909,
      "step": 11124
    },
    {
      "epoch": 0.8419658483515444,
      "grad_norm": 0.68359375,
      "learning_rate": 1.2490267408512016e-05,
      "loss": 0.2611,
      "step": 11125
    },
    {
      "epoch": 0.8420415306749918,
      "grad_norm": 0.68359375,
      "learning_rate": 1.248911383389217e-05,
      "loss": 0.2705,
      "step": 11126
    },
    {
      "epoch": 0.842117212998439,
      "grad_norm": 0.765625,
      "learning_rate": 1.248796022395997e-05,
      "loss": 0.313,
      "step": 11127
    },
    {
      "epoch": 0.8421928953218863,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2486806578731787e-05,
      "loss": 0.3181,
      "step": 11128
    },
    {
      "epoch": 0.8422685776453337,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2485652898223985e-05,
      "loss": 0.3067,
      "step": 11129
    },
    {
      "epoch": 0.842344259968781,
      "grad_norm": 0.8046875,
      "learning_rate": 1.248449918245293e-05,
      "loss": 0.3289,
      "step": 11130
    },
    {
      "epoch": 0.8424199422922284,
      "grad_norm": 0.70703125,
      "learning_rate": 1.2483345431434994e-05,
      "loss": 0.2801,
      "step": 11131
    },
    {
      "epoch": 0.8424956246156757,
      "grad_norm": 0.80859375,
      "learning_rate": 1.2482191645186538e-05,
      "loss": 0.3422,
      "step": 11132
    },
    {
      "epoch": 0.8425713069391231,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2481037823723935e-05,
      "loss": 0.3416,
      "step": 11133
    },
    {
      "epoch": 0.8426469892625703,
      "grad_norm": 0.66796875,
      "learning_rate": 1.2479883967063553e-05,
      "loss": 0.2344,
      "step": 11134
    },
    {
      "epoch": 0.8427226715860177,
      "grad_norm": 1.0,
      "learning_rate": 1.247873007522176e-05,
      "loss": 0.3637,
      "step": 11135
    },
    {
      "epoch": 0.842798353909465,
      "grad_norm": 0.79296875,
      "learning_rate": 1.247757614821493e-05,
      "loss": 0.3386,
      "step": 11136
    },
    {
      "epoch": 0.8428740362329123,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2476422186059426e-05,
      "loss": 0.3093,
      "step": 11137
    },
    {
      "epoch": 0.8429497185563597,
      "grad_norm": 0.80078125,
      "learning_rate": 1.2475268188771628e-05,
      "loss": 0.2759,
      "step": 11138
    },
    {
      "epoch": 0.843025400879807,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2474114156367903e-05,
      "loss": 0.3268,
      "step": 11139
    },
    {
      "epoch": 0.8431010832032544,
      "grad_norm": 0.77734375,
      "learning_rate": 1.2472960088864621e-05,
      "loss": 0.3311,
      "step": 11140
    },
    {
      "epoch": 0.8431767655267016,
      "grad_norm": 0.78125,
      "learning_rate": 1.2471805986278159e-05,
      "loss": 0.3347,
      "step": 11141
    },
    {
      "epoch": 0.843252447850149,
      "grad_norm": 0.77734375,
      "learning_rate": 1.2470651848624881e-05,
      "loss": 0.3034,
      "step": 11142
    },
    {
      "epoch": 0.8433281301735963,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2469497675921172e-05,
      "loss": 0.3176,
      "step": 11143
    },
    {
      "epoch": 0.8434038124970437,
      "grad_norm": 0.8515625,
      "learning_rate": 1.2468343468183404e-05,
      "loss": 0.3644,
      "step": 11144
    },
    {
      "epoch": 0.843479494820491,
      "grad_norm": 0.73046875,
      "learning_rate": 1.246718922542794e-05,
      "loss": 0.2941,
      "step": 11145
    },
    {
      "epoch": 0.8435551771439384,
      "grad_norm": 0.75,
      "learning_rate": 1.2466034947671168e-05,
      "loss": 0.3096,
      "step": 11146
    },
    {
      "epoch": 0.8436308594673857,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2464880634929457e-05,
      "loss": 0.3154,
      "step": 11147
    },
    {
      "epoch": 0.8437065417908329,
      "grad_norm": 0.72265625,
      "learning_rate": 1.2463726287219182e-05,
      "loss": 0.3004,
      "step": 11148
    },
    {
      "epoch": 0.8437822241142803,
      "grad_norm": 0.69921875,
      "learning_rate": 1.2462571904556729e-05,
      "loss": 0.2823,
      "step": 11149
    },
    {
      "epoch": 0.8438579064377276,
      "grad_norm": 0.734375,
      "learning_rate": 1.2461417486958463e-05,
      "loss": 0.3003,
      "step": 11150
    },
    {
      "epoch": 0.843933588761175,
      "grad_norm": 0.77734375,
      "learning_rate": 1.2460263034440763e-05,
      "loss": 0.3353,
      "step": 11151
    },
    {
      "epoch": 0.8440092710846223,
      "grad_norm": 0.78515625,
      "learning_rate": 1.2459108547020014e-05,
      "loss": 0.3318,
      "step": 11152
    },
    {
      "epoch": 0.8440849534080697,
      "grad_norm": 0.875,
      "learning_rate": 1.245795402471259e-05,
      "loss": 0.3591,
      "step": 11153
    },
    {
      "epoch": 0.844160635731517,
      "grad_norm": 0.81640625,
      "learning_rate": 1.2456799467534868e-05,
      "loss": 0.3228,
      "step": 11154
    },
    {
      "epoch": 0.8442363180549642,
      "grad_norm": 0.72265625,
      "learning_rate": 1.245564487550323e-05,
      "loss": 0.2875,
      "step": 11155
    },
    {
      "epoch": 0.8443120003784116,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2454490248634057e-05,
      "loss": 0.3005,
      "step": 11156
    },
    {
      "epoch": 0.8443876827018589,
      "grad_norm": 0.78125,
      "learning_rate": 1.2453335586943724e-05,
      "loss": 0.3233,
      "step": 11157
    },
    {
      "epoch": 0.8444633650253063,
      "grad_norm": 0.73828125,
      "learning_rate": 1.2452180890448622e-05,
      "loss": 0.2802,
      "step": 11158
    },
    {
      "epoch": 0.8445390473487536,
      "grad_norm": 0.83984375,
      "learning_rate": 1.2451026159165122e-05,
      "loss": 0.3481,
      "step": 11159
    },
    {
      "epoch": 0.844614729672201,
      "grad_norm": 0.74609375,
      "learning_rate": 1.244987139310961e-05,
      "loss": 0.2936,
      "step": 11160
    },
    {
      "epoch": 0.8446904119956483,
      "grad_norm": 0.7734375,
      "learning_rate": 1.2448716592298468e-05,
      "loss": 0.3254,
      "step": 11161
    },
    {
      "epoch": 0.8447660943190956,
      "grad_norm": 0.859375,
      "learning_rate": 1.2447561756748079e-05,
      "loss": 0.3706,
      "step": 11162
    },
    {
      "epoch": 0.8448417766425429,
      "grad_norm": 0.72265625,
      "learning_rate": 1.244640688647483e-05,
      "loss": 0.2907,
      "step": 11163
    },
    {
      "epoch": 0.8449174589659902,
      "grad_norm": 0.79296875,
      "learning_rate": 1.2445251981495096e-05,
      "loss": 0.3399,
      "step": 11164
    },
    {
      "epoch": 0.8449931412894376,
      "grad_norm": 0.75,
      "learning_rate": 1.2444097041825269e-05,
      "loss": 0.3008,
      "step": 11165
    },
    {
      "epoch": 0.8450688236128849,
      "grad_norm": 0.75,
      "learning_rate": 1.244294206748173e-05,
      "loss": 0.3038,
      "step": 11166
    },
    {
      "epoch": 0.8451445059363323,
      "grad_norm": 0.75,
      "learning_rate": 1.2441787058480867e-05,
      "loss": 0.3175,
      "step": 11167
    },
    {
      "epoch": 0.8452201882597796,
      "grad_norm": 0.77734375,
      "learning_rate": 1.2440632014839065e-05,
      "loss": 0.3634,
      "step": 11168
    },
    {
      "epoch": 0.8452958705832269,
      "grad_norm": 0.796875,
      "learning_rate": 1.2439476936572705e-05,
      "loss": 0.346,
      "step": 11169
    },
    {
      "epoch": 0.8453715529066742,
      "grad_norm": 0.72265625,
      "learning_rate": 1.2438321823698182e-05,
      "loss": 0.3084,
      "step": 11170
    },
    {
      "epoch": 0.8454472352301216,
      "grad_norm": 0.9765625,
      "learning_rate": 1.243716667623188e-05,
      "loss": 0.3252,
      "step": 11171
    },
    {
      "epoch": 0.8455229175535689,
      "grad_norm": 0.85546875,
      "learning_rate": 1.2436011494190185e-05,
      "loss": 0.3241,
      "step": 11172
    },
    {
      "epoch": 0.8455985998770162,
      "grad_norm": 1.1328125,
      "learning_rate": 1.2434856277589486e-05,
      "loss": 0.3401,
      "step": 11173
    },
    {
      "epoch": 0.8456742822004636,
      "grad_norm": 0.7890625,
      "learning_rate": 1.2433701026446175e-05,
      "loss": 0.3428,
      "step": 11174
    },
    {
      "epoch": 0.8457499645239109,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2432545740776636e-05,
      "loss": 0.3115,
      "step": 11175
    },
    {
      "epoch": 0.8458256468473582,
      "grad_norm": 0.7890625,
      "learning_rate": 1.2431390420597263e-05,
      "loss": 0.365,
      "step": 11176
    },
    {
      "epoch": 0.8459013291708055,
      "grad_norm": 0.71484375,
      "learning_rate": 1.2430235065924443e-05,
      "loss": 0.3042,
      "step": 11177
    },
    {
      "epoch": 0.8459770114942529,
      "grad_norm": 0.7578125,
      "learning_rate": 1.242907967677457e-05,
      "loss": 0.3008,
      "step": 11178
    },
    {
      "epoch": 0.8460526938177002,
      "grad_norm": 0.7265625,
      "learning_rate": 1.2427924253164032e-05,
      "loss": 0.3197,
      "step": 11179
    },
    {
      "epoch": 0.8461283761411476,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2426768795109223e-05,
      "loss": 0.2936,
      "step": 11180
    },
    {
      "epoch": 0.8462040584645949,
      "grad_norm": 0.71484375,
      "learning_rate": 1.2425613302626533e-05,
      "loss": 0.3085,
      "step": 11181
    },
    {
      "epoch": 0.8462797407880422,
      "grad_norm": 0.7265625,
      "learning_rate": 1.2424457775732357e-05,
      "loss": 0.2895,
      "step": 11182
    },
    {
      "epoch": 0.8463554231114895,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2423302214443085e-05,
      "loss": 0.3314,
      "step": 11183
    },
    {
      "epoch": 0.8464311054349368,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2422146618775113e-05,
      "loss": 0.3635,
      "step": 11184
    },
    {
      "epoch": 0.8465067877583842,
      "grad_norm": 0.75,
      "learning_rate": 1.2420990988744837e-05,
      "loss": 0.3259,
      "step": 11185
    },
    {
      "epoch": 0.8465824700818315,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2419835324368648e-05,
      "loss": 0.3221,
      "step": 11186
    },
    {
      "epoch": 0.8466581524052789,
      "grad_norm": 0.859375,
      "learning_rate": 1.241867962566294e-05,
      "loss": 0.3886,
      "step": 11187
    },
    {
      "epoch": 0.8467338347287262,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2417523892644115e-05,
      "loss": 0.304,
      "step": 11188
    },
    {
      "epoch": 0.8468095170521736,
      "grad_norm": 0.73046875,
      "learning_rate": 1.241636812532856e-05,
      "loss": 0.2951,
      "step": 11189
    },
    {
      "epoch": 0.8468851993756208,
      "grad_norm": 0.75,
      "learning_rate": 1.241521232373268e-05,
      "loss": 0.2954,
      "step": 11190
    },
    {
      "epoch": 0.8469608816990681,
      "grad_norm": 0.71484375,
      "learning_rate": 1.241405648787287e-05,
      "loss": 0.2795,
      "step": 11191
    },
    {
      "epoch": 0.8470365640225155,
      "grad_norm": 0.71484375,
      "learning_rate": 1.2412900617765522e-05,
      "loss": 0.2604,
      "step": 11192
    },
    {
      "epoch": 0.8471122463459628,
      "grad_norm": 0.77734375,
      "learning_rate": 1.2411744713427041e-05,
      "loss": 0.3327,
      "step": 11193
    },
    {
      "epoch": 0.8471879286694102,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2410588774873822e-05,
      "loss": 0.2934,
      "step": 11194
    },
    {
      "epoch": 0.8472636109928575,
      "grad_norm": 0.7734375,
      "learning_rate": 1.2409432802122265e-05,
      "loss": 0.3058,
      "step": 11195
    },
    {
      "epoch": 0.8473392933163049,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2408276795188766e-05,
      "loss": 0.2795,
      "step": 11196
    },
    {
      "epoch": 0.8474149756397521,
      "grad_norm": 0.69921875,
      "learning_rate": 1.2407120754089733e-05,
      "loss": 0.2973,
      "step": 11197
    },
    {
      "epoch": 0.8474906579631994,
      "grad_norm": 0.78125,
      "learning_rate": 1.2405964678841556e-05,
      "loss": 0.3436,
      "step": 11198
    },
    {
      "epoch": 0.8475663402866468,
      "grad_norm": 0.71875,
      "learning_rate": 1.2404808569460644e-05,
      "loss": 0.274,
      "step": 11199
    },
    {
      "epoch": 0.8476420226100941,
      "grad_norm": 0.765625,
      "learning_rate": 1.2403652425963398e-05,
      "loss": 0.3469,
      "step": 11200
    },
    {
      "epoch": 0.8477177049335415,
      "grad_norm": 0.75,
      "learning_rate": 1.240249624836621e-05,
      "loss": 0.2866,
      "step": 11201
    },
    {
      "epoch": 0.8477933872569888,
      "grad_norm": 0.78515625,
      "learning_rate": 1.2401340036685497e-05,
      "loss": 0.3191,
      "step": 11202
    },
    {
      "epoch": 0.8478690695804362,
      "grad_norm": 0.7734375,
      "learning_rate": 1.2400183790937652e-05,
      "loss": 0.3256,
      "step": 11203
    },
    {
      "epoch": 0.8479447519038834,
      "grad_norm": 0.76171875,
      "learning_rate": 1.2399027511139081e-05,
      "loss": 0.3231,
      "step": 11204
    },
    {
      "epoch": 0.8480204342273308,
      "grad_norm": 0.80859375,
      "learning_rate": 1.239787119730619e-05,
      "loss": 0.36,
      "step": 11205
    },
    {
      "epoch": 0.8480961165507781,
      "grad_norm": 0.71484375,
      "learning_rate": 1.239671484945538e-05,
      "loss": 0.3217,
      "step": 11206
    },
    {
      "epoch": 0.8481717988742254,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2395558467603055e-05,
      "loss": 0.3085,
      "step": 11207
    },
    {
      "epoch": 0.8482474811976728,
      "grad_norm": 0.69140625,
      "learning_rate": 1.2394402051765624e-05,
      "loss": 0.2994,
      "step": 11208
    },
    {
      "epoch": 0.8483231635211201,
      "grad_norm": 0.8046875,
      "learning_rate": 1.2393245601959493e-05,
      "loss": 0.348,
      "step": 11209
    },
    {
      "epoch": 0.8483988458445674,
      "grad_norm": 0.8046875,
      "learning_rate": 1.2392089118201062e-05,
      "loss": 0.3351,
      "step": 11210
    },
    {
      "epoch": 0.8484745281680147,
      "grad_norm": 0.69921875,
      "learning_rate": 1.2390932600506745e-05,
      "loss": 0.2811,
      "step": 11211
    },
    {
      "epoch": 0.8485502104914621,
      "grad_norm": 0.7734375,
      "learning_rate": 1.2389776048892945e-05,
      "loss": 0.2816,
      "step": 11212
    },
    {
      "epoch": 0.8486258928149094,
      "grad_norm": 0.65234375,
      "learning_rate": 1.2388619463376068e-05,
      "loss": 0.2452,
      "step": 11213
    },
    {
      "epoch": 0.8487015751383568,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2387462843972531e-05,
      "loss": 0.3371,
      "step": 11214
    },
    {
      "epoch": 0.8487772574618041,
      "grad_norm": 0.765625,
      "learning_rate": 1.2386306190698731e-05,
      "loss": 0.3259,
      "step": 11215
    },
    {
      "epoch": 0.8488529397852514,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2385149503571084e-05,
      "loss": 0.3219,
      "step": 11216
    },
    {
      "epoch": 0.8489286221086987,
      "grad_norm": 0.69921875,
      "learning_rate": 1.2383992782605999e-05,
      "loss": 0.2937,
      "step": 11217
    },
    {
      "epoch": 0.849004304432146,
      "grad_norm": 0.70703125,
      "learning_rate": 1.2382836027819887e-05,
      "loss": 0.2919,
      "step": 11218
    },
    {
      "epoch": 0.8490799867555934,
      "grad_norm": 0.75,
      "learning_rate": 1.2381679239229155e-05,
      "loss": 0.3118,
      "step": 11219
    },
    {
      "epoch": 0.8491556690790407,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2380522416850216e-05,
      "loss": 0.2992,
      "step": 11220
    },
    {
      "epoch": 0.8492313514024881,
      "grad_norm": 0.734375,
      "learning_rate": 1.2379365560699482e-05,
      "loss": 0.3138,
      "step": 11221
    },
    {
      "epoch": 0.8493070337259354,
      "grad_norm": 0.8671875,
      "learning_rate": 1.2378208670793361e-05,
      "loss": 0.3026,
      "step": 11222
    },
    {
      "epoch": 0.8493827160493828,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2377051747148274e-05,
      "loss": 0.2852,
      "step": 11223
    },
    {
      "epoch": 0.84945839837283,
      "grad_norm": 0.7734375,
      "learning_rate": 1.2375894789780625e-05,
      "loss": 0.3026,
      "step": 11224
    },
    {
      "epoch": 0.8495340806962773,
      "grad_norm": 0.7265625,
      "learning_rate": 1.2374737798706832e-05,
      "loss": 0.2989,
      "step": 11225
    },
    {
      "epoch": 0.8496097630197247,
      "grad_norm": 0.69921875,
      "learning_rate": 1.237358077394331e-05,
      "loss": 0.2879,
      "step": 11226
    },
    {
      "epoch": 0.849685445343172,
      "grad_norm": 0.7890625,
      "learning_rate": 1.2372423715506469e-05,
      "loss": 0.3267,
      "step": 11227
    },
    {
      "epoch": 0.8497611276666194,
      "grad_norm": 0.73046875,
      "learning_rate": 1.237126662341273e-05,
      "loss": 0.3165,
      "step": 11228
    },
    {
      "epoch": 0.8498368099900667,
      "grad_norm": 0.71484375,
      "learning_rate": 1.2370109497678503e-05,
      "loss": 0.3003,
      "step": 11229
    },
    {
      "epoch": 0.8499124923135141,
      "grad_norm": 0.95703125,
      "learning_rate": 1.2368952338320204e-05,
      "loss": 0.3222,
      "step": 11230
    },
    {
      "epoch": 0.8499881746369613,
      "grad_norm": 0.7890625,
      "learning_rate": 1.2367795145354249e-05,
      "loss": 0.3382,
      "step": 11231
    },
    {
      "epoch": 0.8500638569604086,
      "grad_norm": 0.74609375,
      "learning_rate": 1.236663791879706e-05,
      "loss": 0.3154,
      "step": 11232
    },
    {
      "epoch": 0.850139539283856,
      "grad_norm": 0.7578125,
      "learning_rate": 1.236548065866505e-05,
      "loss": 0.3325,
      "step": 11233
    },
    {
      "epoch": 0.8502152216073033,
      "grad_norm": 0.734375,
      "learning_rate": 1.236432336497464e-05,
      "loss": 0.3237,
      "step": 11234
    },
    {
      "epoch": 0.8502909039307507,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2363166037742242e-05,
      "loss": 0.343,
      "step": 11235
    },
    {
      "epoch": 0.850366586254198,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2362008676984281e-05,
      "loss": 0.3287,
      "step": 11236
    },
    {
      "epoch": 0.8504422685776454,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2360851282717172e-05,
      "loss": 0.333,
      "step": 11237
    },
    {
      "epoch": 0.8505179509010926,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2359693854957337e-05,
      "loss": 0.3251,
      "step": 11238
    },
    {
      "epoch": 0.85059363322454,
      "grad_norm": 0.68359375,
      "learning_rate": 1.2358536393721194e-05,
      "loss": 0.2599,
      "step": 11239
    },
    {
      "epoch": 0.8506693155479873,
      "grad_norm": 0.7734375,
      "learning_rate": 1.2357378899025166e-05,
      "loss": 0.3384,
      "step": 11240
    },
    {
      "epoch": 0.8507449978714346,
      "grad_norm": 0.75,
      "learning_rate": 1.2356221370885673e-05,
      "loss": 0.2897,
      "step": 11241
    },
    {
      "epoch": 0.850820680194882,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2355063809319137e-05,
      "loss": 0.3198,
      "step": 11242
    },
    {
      "epoch": 0.8508963625183293,
      "grad_norm": 0.671875,
      "learning_rate": 1.2353906214341977e-05,
      "loss": 0.2741,
      "step": 11243
    },
    {
      "epoch": 0.8509720448417767,
      "grad_norm": 0.78125,
      "learning_rate": 1.2352748585970619e-05,
      "loss": 0.3281,
      "step": 11244
    },
    {
      "epoch": 0.8510477271652239,
      "grad_norm": 0.8203125,
      "learning_rate": 1.2351590924221486e-05,
      "loss": 0.3703,
      "step": 11245
    },
    {
      "epoch": 0.8511234094886713,
      "grad_norm": 0.7734375,
      "learning_rate": 1.2350433229110999e-05,
      "loss": 0.3242,
      "step": 11246
    },
    {
      "epoch": 0.8511990918121186,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2349275500655587e-05,
      "loss": 0.3192,
      "step": 11247
    },
    {
      "epoch": 0.851274774135566,
      "grad_norm": 0.68359375,
      "learning_rate": 1.2348117738871666e-05,
      "loss": 0.276,
      "step": 11248
    },
    {
      "epoch": 0.8513504564590133,
      "grad_norm": 0.68359375,
      "learning_rate": 1.2346959943775665e-05,
      "loss": 0.2418,
      "step": 11249
    },
    {
      "epoch": 0.8514261387824607,
      "grad_norm": 0.7109375,
      "learning_rate": 1.2345802115384014e-05,
      "loss": 0.2864,
      "step": 11250
    },
    {
      "epoch": 0.851501821105908,
      "grad_norm": 0.76953125,
      "learning_rate": 1.234464425371313e-05,
      "loss": 0.3159,
      "step": 11251
    },
    {
      "epoch": 0.8515775034293552,
      "grad_norm": 0.80078125,
      "learning_rate": 1.2343486358779446e-05,
      "loss": 0.3366,
      "step": 11252
    },
    {
      "epoch": 0.8516531857528026,
      "grad_norm": 0.73828125,
      "learning_rate": 1.2342328430599385e-05,
      "loss": 0.3214,
      "step": 11253
    },
    {
      "epoch": 0.8517288680762499,
      "grad_norm": 1.3671875,
      "learning_rate": 1.2341170469189377e-05,
      "loss": 0.3637,
      "step": 11254
    },
    {
      "epoch": 0.8518045503996973,
      "grad_norm": 0.71875,
      "learning_rate": 1.2340012474565848e-05,
      "loss": 0.3013,
      "step": 11255
    },
    {
      "epoch": 0.8518802327231446,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2338854446745228e-05,
      "loss": 0.3096,
      "step": 11256
    },
    {
      "epoch": 0.851955915046592,
      "grad_norm": 0.74609375,
      "learning_rate": 1.233769638574394e-05,
      "loss": 0.3086,
      "step": 11257
    },
    {
      "epoch": 0.8520315973700393,
      "grad_norm": 0.75,
      "learning_rate": 1.2336538291578422e-05,
      "loss": 0.3191,
      "step": 11258
    },
    {
      "epoch": 0.8521072796934865,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2335380164265097e-05,
      "loss": 0.3171,
      "step": 11259
    },
    {
      "epoch": 0.8521829620169339,
      "grad_norm": 0.68359375,
      "learning_rate": 1.2334222003820393e-05,
      "loss": 0.2844,
      "step": 11260
    },
    {
      "epoch": 0.8522586443403812,
      "grad_norm": 0.75,
      "learning_rate": 1.2333063810260751e-05,
      "loss": 0.3135,
      "step": 11261
    },
    {
      "epoch": 0.8523343266638286,
      "grad_norm": 0.72265625,
      "learning_rate": 1.233190558360259e-05,
      "loss": 0.264,
      "step": 11262
    },
    {
      "epoch": 0.8524100089872759,
      "grad_norm": 0.72265625,
      "learning_rate": 1.2330747323862347e-05,
      "loss": 0.2961,
      "step": 11263
    },
    {
      "epoch": 0.8524856913107233,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2329589031056454e-05,
      "loss": 0.3044,
      "step": 11264
    },
    {
      "epoch": 0.8525613736341706,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2328430705201347e-05,
      "loss": 0.3354,
      "step": 11265
    },
    {
      "epoch": 0.8526370559576179,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2327272346313449e-05,
      "loss": 0.3253,
      "step": 11266
    },
    {
      "epoch": 0.8527127382810652,
      "grad_norm": 0.8203125,
      "learning_rate": 1.23261139544092e-05,
      "loss": 0.3607,
      "step": 11267
    },
    {
      "epoch": 0.8527884206045125,
      "grad_norm": 0.734375,
      "learning_rate": 1.2324955529505034e-05,
      "loss": 0.3145,
      "step": 11268
    },
    {
      "epoch": 0.8528641029279599,
      "grad_norm": 0.7265625,
      "learning_rate": 1.2323797071617383e-05,
      "loss": 0.3043,
      "step": 11269
    },
    {
      "epoch": 0.8529397852514072,
      "grad_norm": 0.82421875,
      "learning_rate": 1.2322638580762684e-05,
      "loss": 0.3661,
      "step": 11270
    },
    {
      "epoch": 0.8530154675748546,
      "grad_norm": 0.79296875,
      "learning_rate": 1.2321480056957373e-05,
      "loss": 0.3543,
      "step": 11271
    },
    {
      "epoch": 0.8530911498983019,
      "grad_norm": 0.765625,
      "learning_rate": 1.232032150021788e-05,
      "loss": 0.3251,
      "step": 11272
    },
    {
      "epoch": 0.8531668322217492,
      "grad_norm": 0.6875,
      "learning_rate": 1.2319162910560645e-05,
      "loss": 0.263,
      "step": 11273
    },
    {
      "epoch": 0.8532425145451965,
      "grad_norm": 0.80078125,
      "learning_rate": 1.2318004288002106e-05,
      "loss": 0.3761,
      "step": 11274
    },
    {
      "epoch": 0.8533181968686439,
      "grad_norm": 0.7265625,
      "learning_rate": 1.2316845632558698e-05,
      "loss": 0.3108,
      "step": 11275
    },
    {
      "epoch": 0.8533938791920912,
      "grad_norm": 0.7109375,
      "learning_rate": 1.2315686944246859e-05,
      "loss": 0.2948,
      "step": 11276
    },
    {
      "epoch": 0.8534695615155385,
      "grad_norm": 0.73046875,
      "learning_rate": 1.2314528223083027e-05,
      "loss": 0.3005,
      "step": 11277
    },
    {
      "epoch": 0.8535452438389859,
      "grad_norm": 0.69921875,
      "learning_rate": 1.2313369469083636e-05,
      "loss": 0.2729,
      "step": 11278
    },
    {
      "epoch": 0.8536209261624332,
      "grad_norm": 0.734375,
      "learning_rate": 1.2312210682265135e-05,
      "loss": 0.3067,
      "step": 11279
    },
    {
      "epoch": 0.8536966084858805,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2311051862643955e-05,
      "loss": 0.3333,
      "step": 11280
    },
    {
      "epoch": 0.8537722908093278,
      "grad_norm": 0.80078125,
      "learning_rate": 1.230989301023654e-05,
      "loss": 0.3032,
      "step": 11281
    },
    {
      "epoch": 0.8538479731327752,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2308734125059327e-05,
      "loss": 0.3234,
      "step": 11282
    },
    {
      "epoch": 0.8539236554562225,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2307575207128758e-05,
      "loss": 0.3154,
      "step": 11283
    },
    {
      "epoch": 0.8539993377796699,
      "grad_norm": 0.75,
      "learning_rate": 1.2306416256461278e-05,
      "loss": 0.3294,
      "step": 11284
    },
    {
      "epoch": 0.8540750201031172,
      "grad_norm": 1.0078125,
      "learning_rate": 1.2305257273073323e-05,
      "loss": 0.376,
      "step": 11285
    },
    {
      "epoch": 0.8541507024265645,
      "grad_norm": 0.765625,
      "learning_rate": 1.2304098256981339e-05,
      "loss": 0.305,
      "step": 11286
    },
    {
      "epoch": 0.8542263847500118,
      "grad_norm": 0.8359375,
      "learning_rate": 1.2302939208201763e-05,
      "loss": 0.3888,
      "step": 11287
    },
    {
      "epoch": 0.8543020670734591,
      "grad_norm": 0.796875,
      "learning_rate": 1.2301780126751047e-05,
      "loss": 0.3441,
      "step": 11288
    },
    {
      "epoch": 0.8543777493969065,
      "grad_norm": 0.78125,
      "learning_rate": 1.2300621012645629e-05,
      "loss": 0.3499,
      "step": 11289
    },
    {
      "epoch": 0.8544534317203538,
      "grad_norm": 0.734375,
      "learning_rate": 1.2299461865901954e-05,
      "loss": 0.3147,
      "step": 11290
    },
    {
      "epoch": 0.8545291140438012,
      "grad_norm": 0.76171875,
      "learning_rate": 1.2298302686536464e-05,
      "loss": 0.2942,
      "step": 11291
    },
    {
      "epoch": 0.8546047963672485,
      "grad_norm": 0.72265625,
      "learning_rate": 1.229714347456561e-05,
      "loss": 0.2973,
      "step": 11292
    },
    {
      "epoch": 0.8546804786906959,
      "grad_norm": 1.046875,
      "learning_rate": 1.229598423000583e-05,
      "loss": 0.3324,
      "step": 11293
    },
    {
      "epoch": 0.8547561610141431,
      "grad_norm": 0.76171875,
      "learning_rate": 1.2294824952873578e-05,
      "loss": 0.3295,
      "step": 11294
    },
    {
      "epoch": 0.8548318433375904,
      "grad_norm": 0.77734375,
      "learning_rate": 1.2293665643185292e-05,
      "loss": 0.3431,
      "step": 11295
    },
    {
      "epoch": 0.8549075256610378,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2292506300957423e-05,
      "loss": 0.2856,
      "step": 11296
    },
    {
      "epoch": 0.8549832079844851,
      "grad_norm": 0.70703125,
      "learning_rate": 1.229134692620642e-05,
      "loss": 0.2809,
      "step": 11297
    },
    {
      "epoch": 0.8550588903079325,
      "grad_norm": 0.80859375,
      "learning_rate": 1.2290187518948729e-05,
      "loss": 0.3539,
      "step": 11298
    },
    {
      "epoch": 0.8551345726313798,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2289028079200798e-05,
      "loss": 0.3393,
      "step": 11299
    },
    {
      "epoch": 0.8552102549548272,
      "grad_norm": 0.7734375,
      "learning_rate": 1.2287868606979074e-05,
      "loss": 0.3485,
      "step": 11300
    },
    {
      "epoch": 0.8552859372782744,
      "grad_norm": 0.7734375,
      "learning_rate": 1.2286709102300008e-05,
      "loss": 0.2972,
      "step": 11301
    },
    {
      "epoch": 0.8553616196017217,
      "grad_norm": 0.85546875,
      "learning_rate": 1.2285549565180048e-05,
      "loss": 0.3234,
      "step": 11302
    },
    {
      "epoch": 0.8554373019251691,
      "grad_norm": 0.67578125,
      "learning_rate": 1.228438999563565e-05,
      "loss": 0.2736,
      "step": 11303
    },
    {
      "epoch": 0.8555129842486164,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2283230393683257e-05,
      "loss": 0.2895,
      "step": 11304
    },
    {
      "epoch": 0.8555886665720638,
      "grad_norm": 0.78515625,
      "learning_rate": 1.2282070759339321e-05,
      "loss": 0.3261,
      "step": 11305
    },
    {
      "epoch": 0.8556643488955111,
      "grad_norm": 0.77734375,
      "learning_rate": 1.2280911092620298e-05,
      "loss": 0.3067,
      "step": 11306
    },
    {
      "epoch": 0.8557400312189585,
      "grad_norm": 0.75,
      "learning_rate": 1.2279751393542634e-05,
      "loss": 0.3002,
      "step": 11307
    },
    {
      "epoch": 0.8558157135424057,
      "grad_norm": 0.734375,
      "learning_rate": 1.2278591662122787e-05,
      "loss": 0.3197,
      "step": 11308
    },
    {
      "epoch": 0.855891395865853,
      "grad_norm": 0.71875,
      "learning_rate": 1.2277431898377205e-05,
      "loss": 0.2838,
      "step": 11309
    },
    {
      "epoch": 0.8559670781893004,
      "grad_norm": 0.828125,
      "learning_rate": 1.2276272102322344e-05,
      "loss": 0.3413,
      "step": 11310
    },
    {
      "epoch": 0.8560427605127477,
      "grad_norm": 0.859375,
      "learning_rate": 1.2275112273974657e-05,
      "loss": 0.3657,
      "step": 11311
    },
    {
      "epoch": 0.8561184428361951,
      "grad_norm": 0.77734375,
      "learning_rate": 1.2273952413350601e-05,
      "loss": 0.3377,
      "step": 11312
    },
    {
      "epoch": 0.8561941251596424,
      "grad_norm": 0.69921875,
      "learning_rate": 1.2272792520466623e-05,
      "loss": 0.2537,
      "step": 11313
    },
    {
      "epoch": 0.8562698074830898,
      "grad_norm": 0.72265625,
      "learning_rate": 1.2271632595339184e-05,
      "loss": 0.31,
      "step": 11314
    },
    {
      "epoch": 0.856345489806537,
      "grad_norm": 0.78515625,
      "learning_rate": 1.227047263798474e-05,
      "loss": 0.3318,
      "step": 11315
    },
    {
      "epoch": 0.8564211721299844,
      "grad_norm": 0.79296875,
      "learning_rate": 1.2269312648419744e-05,
      "loss": 0.3536,
      "step": 11316
    },
    {
      "epoch": 0.8564968544534317,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2268152626660657e-05,
      "loss": 0.3101,
      "step": 11317
    },
    {
      "epoch": 0.8565725367768791,
      "grad_norm": 0.7734375,
      "learning_rate": 1.226699257272393e-05,
      "loss": 0.3462,
      "step": 11318
    },
    {
      "epoch": 0.8566482191003264,
      "grad_norm": 0.77734375,
      "learning_rate": 1.2265832486626022e-05,
      "loss": 0.3165,
      "step": 11319
    },
    {
      "epoch": 0.8567239014237737,
      "grad_norm": 0.765625,
      "learning_rate": 1.2264672368383392e-05,
      "loss": 0.318,
      "step": 11320
    },
    {
      "epoch": 0.8567995837472211,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2263512218012503e-05,
      "loss": 0.3098,
      "step": 11321
    },
    {
      "epoch": 0.8568752660706683,
      "grad_norm": 0.7109375,
      "learning_rate": 1.2262352035529803e-05,
      "loss": 0.297,
      "step": 11322
    },
    {
      "epoch": 0.8569509483941157,
      "grad_norm": 0.703125,
      "learning_rate": 1.2261191820951758e-05,
      "loss": 0.2856,
      "step": 11323
    },
    {
      "epoch": 0.857026630717563,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2260031574294831e-05,
      "loss": 0.3417,
      "step": 11324
    },
    {
      "epoch": 0.8571023130410104,
      "grad_norm": 0.69921875,
      "learning_rate": 1.2258871295575472e-05,
      "loss": 0.2527,
      "step": 11325
    },
    {
      "epoch": 0.8571779953644577,
      "grad_norm": 0.7109375,
      "learning_rate": 1.2257710984810151e-05,
      "loss": 0.28,
      "step": 11326
    },
    {
      "epoch": 0.8572536776879051,
      "grad_norm": 0.70703125,
      "learning_rate": 1.2256550642015326e-05,
      "loss": 0.289,
      "step": 11327
    },
    {
      "epoch": 0.8573293600113524,
      "grad_norm": 0.76171875,
      "learning_rate": 1.2255390267207454e-05,
      "loss": 0.2989,
      "step": 11328
    },
    {
      "epoch": 0.8574050423347996,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2254229860403005e-05,
      "loss": 0.3299,
      "step": 11329
    },
    {
      "epoch": 0.857480724658247,
      "grad_norm": 0.765625,
      "learning_rate": 1.2253069421618434e-05,
      "loss": 0.3308,
      "step": 11330
    },
    {
      "epoch": 0.8575564069816943,
      "grad_norm": 0.77734375,
      "learning_rate": 1.2251908950870209e-05,
      "loss": 0.3337,
      "step": 11331
    },
    {
      "epoch": 0.8576320893051417,
      "grad_norm": 0.70703125,
      "learning_rate": 1.225074844817479e-05,
      "loss": 0.2931,
      "step": 11332
    },
    {
      "epoch": 0.857707771628589,
      "grad_norm": 0.82421875,
      "learning_rate": 1.224958791354864e-05,
      "loss": 0.3697,
      "step": 11333
    },
    {
      "epoch": 0.8577834539520364,
      "grad_norm": 0.80078125,
      "learning_rate": 1.2248427347008224e-05,
      "loss": 0.3704,
      "step": 11334
    },
    {
      "epoch": 0.8578591362754836,
      "grad_norm": 0.71875,
      "learning_rate": 1.2247266748570015e-05,
      "loss": 0.3091,
      "step": 11335
    },
    {
      "epoch": 0.857934818598931,
      "grad_norm": 0.7890625,
      "learning_rate": 1.2246106118250462e-05,
      "loss": 0.3693,
      "step": 11336
    },
    {
      "epoch": 0.8580105009223783,
      "grad_norm": 0.75,
      "learning_rate": 1.2244945456066044e-05,
      "loss": 0.3097,
      "step": 11337
    },
    {
      "epoch": 0.8580861832458256,
      "grad_norm": 0.77734375,
      "learning_rate": 1.2243784762033222e-05,
      "loss": 0.3106,
      "step": 11338
    },
    {
      "epoch": 0.858161865569273,
      "grad_norm": 0.734375,
      "learning_rate": 1.2242624036168462e-05,
      "loss": 0.3134,
      "step": 11339
    },
    {
      "epoch": 0.8582375478927203,
      "grad_norm": 0.6640625,
      "learning_rate": 1.2241463278488233e-05,
      "loss": 0.2537,
      "step": 11340
    },
    {
      "epoch": 0.8583132302161677,
      "grad_norm": 0.94921875,
      "learning_rate": 1.2240302489009e-05,
      "loss": 0.3519,
      "step": 11341
    },
    {
      "epoch": 0.8583889125396149,
      "grad_norm": 0.69140625,
      "learning_rate": 1.2239141667747231e-05,
      "loss": 0.2891,
      "step": 11342
    },
    {
      "epoch": 0.8584645948630623,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2237980814719395e-05,
      "loss": 0.3105,
      "step": 11343
    },
    {
      "epoch": 0.8585402771865096,
      "grad_norm": 0.6953125,
      "learning_rate": 1.2236819929941962e-05,
      "loss": 0.2897,
      "step": 11344
    },
    {
      "epoch": 0.858615959509957,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2235659013431398e-05,
      "loss": 0.3051,
      "step": 11345
    },
    {
      "epoch": 0.8586916418334043,
      "grad_norm": 0.72265625,
      "learning_rate": 1.2234498065204175e-05,
      "loss": 0.2969,
      "step": 11346
    },
    {
      "epoch": 0.8587673241568516,
      "grad_norm": 0.71875,
      "learning_rate": 1.2233337085276765e-05,
      "loss": 0.2707,
      "step": 11347
    },
    {
      "epoch": 0.858843006480299,
      "grad_norm": 0.78125,
      "learning_rate": 1.2232176073665631e-05,
      "loss": 0.3188,
      "step": 11348
    },
    {
      "epoch": 0.8589186888037462,
      "grad_norm": 0.73828125,
      "learning_rate": 1.2231015030387252e-05,
      "loss": 0.313,
      "step": 11349
    },
    {
      "epoch": 0.8589943711271936,
      "grad_norm": 0.77734375,
      "learning_rate": 1.2229853955458095e-05,
      "loss": 0.3339,
      "step": 11350
    },
    {
      "epoch": 0.8590700534506409,
      "grad_norm": 0.6875,
      "learning_rate": 1.2228692848894633e-05,
      "loss": 0.2849,
      "step": 11351
    },
    {
      "epoch": 0.8591457357740883,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2227531710713341e-05,
      "loss": 0.3225,
      "step": 11352
    },
    {
      "epoch": 0.8592214180975356,
      "grad_norm": 0.73046875,
      "learning_rate": 1.2226370540930685e-05,
      "loss": 0.3086,
      "step": 11353
    },
    {
      "epoch": 0.859297100420983,
      "grad_norm": 0.8359375,
      "learning_rate": 1.2225209339563144e-05,
      "loss": 0.3087,
      "step": 11354
    },
    {
      "epoch": 0.8593727827444303,
      "grad_norm": 0.734375,
      "learning_rate": 1.2224048106627194e-05,
      "loss": 0.3131,
      "step": 11355
    },
    {
      "epoch": 0.8594484650678775,
      "grad_norm": 0.90234375,
      "learning_rate": 1.2222886842139299e-05,
      "loss": 0.3631,
      "step": 11356
    },
    {
      "epoch": 0.8595241473913249,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2221725546115941e-05,
      "loss": 0.3196,
      "step": 11357
    },
    {
      "epoch": 0.8595998297147722,
      "grad_norm": 0.80859375,
      "learning_rate": 1.2220564218573594e-05,
      "loss": 0.3303,
      "step": 11358
    },
    {
      "epoch": 0.8596755120382196,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2219402859528737e-05,
      "loss": 0.3092,
      "step": 11359
    },
    {
      "epoch": 0.8597511943616669,
      "grad_norm": 0.7734375,
      "learning_rate": 1.2218241468997836e-05,
      "loss": 0.2954,
      "step": 11360
    },
    {
      "epoch": 0.8598268766851143,
      "grad_norm": 0.80859375,
      "learning_rate": 1.2217080046997376e-05,
      "loss": 0.3383,
      "step": 11361
    },
    {
      "epoch": 0.8599025590085616,
      "grad_norm": 0.69921875,
      "learning_rate": 1.221591859354383e-05,
      "loss": 0.3006,
      "step": 11362
    },
    {
      "epoch": 0.8599782413320088,
      "grad_norm": 0.96484375,
      "learning_rate": 1.2214757108653674e-05,
      "loss": 0.2985,
      "step": 11363
    },
    {
      "epoch": 0.8600539236554562,
      "grad_norm": 0.91015625,
      "learning_rate": 1.2213595592343392e-05,
      "loss": 0.317,
      "step": 11364
    },
    {
      "epoch": 0.8601296059789035,
      "grad_norm": 0.7109375,
      "learning_rate": 1.2212434044629458e-05,
      "loss": 0.2655,
      "step": 11365
    },
    {
      "epoch": 0.8602052883023509,
      "grad_norm": 0.7265625,
      "learning_rate": 1.2211272465528345e-05,
      "loss": 0.2936,
      "step": 11366
    },
    {
      "epoch": 0.8602809706257982,
      "grad_norm": 0.81640625,
      "learning_rate": 1.221011085505654e-05,
      "loss": 0.3349,
      "step": 11367
    },
    {
      "epoch": 0.8603566529492456,
      "grad_norm": 0.7109375,
      "learning_rate": 1.2208949213230526e-05,
      "loss": 0.2517,
      "step": 11368
    },
    {
      "epoch": 0.8604323352726929,
      "grad_norm": 0.7109375,
      "learning_rate": 1.2207787540066769e-05,
      "loss": 0.2896,
      "step": 11369
    },
    {
      "epoch": 0.8605080175961402,
      "grad_norm": 0.78125,
      "learning_rate": 1.2206625835581757e-05,
      "loss": 0.3295,
      "step": 11370
    },
    {
      "epoch": 0.8605836999195875,
      "grad_norm": 0.8046875,
      "learning_rate": 1.2205464099791976e-05,
      "loss": 0.3749,
      "step": 11371
    },
    {
      "epoch": 0.8606593822430348,
      "grad_norm": 0.76171875,
      "learning_rate": 1.2204302332713896e-05,
      "loss": 0.3183,
      "step": 11372
    },
    {
      "epoch": 0.8607350645664822,
      "grad_norm": 0.75390625,
      "learning_rate": 1.220314053436401e-05,
      "loss": 0.3125,
      "step": 11373
    },
    {
      "epoch": 0.8608107468899295,
      "grad_norm": 0.79296875,
      "learning_rate": 1.2201978704758793e-05,
      "loss": 0.331,
      "step": 11374
    },
    {
      "epoch": 0.8608864292133769,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2200816843914726e-05,
      "loss": 0.3051,
      "step": 11375
    },
    {
      "epoch": 0.8609621115368242,
      "grad_norm": 0.765625,
      "learning_rate": 1.2199654951848301e-05,
      "loss": 0.3234,
      "step": 11376
    },
    {
      "epoch": 0.8610377938602715,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2198493028575992e-05,
      "loss": 0.2983,
      "step": 11377
    },
    {
      "epoch": 0.8611134761837188,
      "grad_norm": 0.74609375,
      "learning_rate": 1.219733107411429e-05,
      "loss": 0.2925,
      "step": 11378
    },
    {
      "epoch": 0.8611891585071662,
      "grad_norm": 0.73828125,
      "learning_rate": 1.2196169088479673e-05,
      "loss": 0.3003,
      "step": 11379
    },
    {
      "epoch": 0.8612648408306135,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2195007071688629e-05,
      "loss": 0.3187,
      "step": 11380
    },
    {
      "epoch": 0.8613405231540608,
      "grad_norm": 0.77734375,
      "learning_rate": 1.2193845023757643e-05,
      "loss": 0.3519,
      "step": 11381
    },
    {
      "epoch": 0.8614162054775082,
      "grad_norm": 0.7734375,
      "learning_rate": 1.2192682944703202e-05,
      "loss": 0.3294,
      "step": 11382
    },
    {
      "epoch": 0.8614918878009555,
      "grad_norm": 0.73828125,
      "learning_rate": 1.219152083454179e-05,
      "loss": 0.312,
      "step": 11383
    },
    {
      "epoch": 0.8615675701244028,
      "grad_norm": 0.8203125,
      "learning_rate": 1.2190358693289892e-05,
      "loss": 0.3803,
      "step": 11384
    },
    {
      "epoch": 0.8616432524478501,
      "grad_norm": 0.78515625,
      "learning_rate": 1.2189196520964e-05,
      "loss": 0.3205,
      "step": 11385
    },
    {
      "epoch": 0.8617189347712975,
      "grad_norm": 0.765625,
      "learning_rate": 1.2188034317580598e-05,
      "loss": 0.3161,
      "step": 11386
    },
    {
      "epoch": 0.8617946170947448,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2186872083156174e-05,
      "loss": 0.3147,
      "step": 11387
    },
    {
      "epoch": 0.8618702994181922,
      "grad_norm": 0.71875,
      "learning_rate": 1.2185709817707217e-05,
      "loss": 0.304,
      "step": 11388
    },
    {
      "epoch": 0.8619459817416395,
      "grad_norm": 0.8125,
      "learning_rate": 1.2184547521250213e-05,
      "loss": 0.3927,
      "step": 11389
    },
    {
      "epoch": 0.8620216640650868,
      "grad_norm": 0.75,
      "learning_rate": 1.2183385193801655e-05,
      "loss": 0.308,
      "step": 11390
    },
    {
      "epoch": 0.8620973463885341,
      "grad_norm": 0.765625,
      "learning_rate": 1.2182222835378033e-05,
      "loss": 0.2769,
      "step": 11391
    },
    {
      "epoch": 0.8621730287119814,
      "grad_norm": 1.015625,
      "learning_rate": 1.2181060445995833e-05,
      "loss": 0.3336,
      "step": 11392
    },
    {
      "epoch": 0.8622487110354288,
      "grad_norm": 0.74609375,
      "learning_rate": 1.217989802567155e-05,
      "loss": 0.2921,
      "step": 11393
    },
    {
      "epoch": 0.8623243933588761,
      "grad_norm": 0.73046875,
      "learning_rate": 1.2178735574421671e-05,
      "loss": 0.2999,
      "step": 11394
    },
    {
      "epoch": 0.8624000756823235,
      "grad_norm": 0.79296875,
      "learning_rate": 1.217757309226269e-05,
      "loss": 0.3432,
      "step": 11395
    },
    {
      "epoch": 0.8624757580057708,
      "grad_norm": 0.82421875,
      "learning_rate": 1.21764105792111e-05,
      "loss": 0.2616,
      "step": 11396
    },
    {
      "epoch": 0.8625514403292182,
      "grad_norm": 0.71484375,
      "learning_rate": 1.2175248035283387e-05,
      "loss": 0.2784,
      "step": 11397
    },
    {
      "epoch": 0.8626271226526654,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2174085460496048e-05,
      "loss": 0.3108,
      "step": 11398
    },
    {
      "epoch": 0.8627028049761127,
      "grad_norm": 0.64453125,
      "learning_rate": 1.2172922854865578e-05,
      "loss": 0.2242,
      "step": 11399
    },
    {
      "epoch": 0.8627784872995601,
      "grad_norm": 0.765625,
      "learning_rate": 1.2171760218408468e-05,
      "loss": 0.3249,
      "step": 11400
    },
    {
      "epoch": 0.8628541696230074,
      "grad_norm": 0.84375,
      "learning_rate": 1.2170597551141212e-05,
      "loss": 0.3676,
      "step": 11401
    },
    {
      "epoch": 0.8629298519464548,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2169434853080304e-05,
      "loss": 0.2908,
      "step": 11402
    },
    {
      "epoch": 0.8630055342699021,
      "grad_norm": 0.73046875,
      "learning_rate": 1.2168272124242242e-05,
      "loss": 0.2833,
      "step": 11403
    },
    {
      "epoch": 0.8630812165933495,
      "grad_norm": 0.76171875,
      "learning_rate": 1.2167109364643518e-05,
      "loss": 0.3003,
      "step": 11404
    },
    {
      "epoch": 0.8631568989167967,
      "grad_norm": 0.79296875,
      "learning_rate": 1.2165946574300626e-05,
      "loss": 0.343,
      "step": 11405
    },
    {
      "epoch": 0.863232581240244,
      "grad_norm": 0.71875,
      "learning_rate": 1.216478375323007e-05,
      "loss": 0.2987,
      "step": 11406
    },
    {
      "epoch": 0.8633082635636914,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2163620901448339e-05,
      "loss": 0.3032,
      "step": 11407
    },
    {
      "epoch": 0.8633839458871387,
      "grad_norm": 0.7109375,
      "learning_rate": 1.2162458018971931e-05,
      "loss": 0.3152,
      "step": 11408
    },
    {
      "epoch": 0.8634596282105861,
      "grad_norm": 0.7109375,
      "learning_rate": 1.2161295105817347e-05,
      "loss": 0.291,
      "step": 11409
    },
    {
      "epoch": 0.8635353105340334,
      "grad_norm": 0.95703125,
      "learning_rate": 1.2160132162001084e-05,
      "loss": 0.3933,
      "step": 11410
    },
    {
      "epoch": 0.8636109928574808,
      "grad_norm": 0.734375,
      "learning_rate": 1.2158969187539637e-05,
      "loss": 0.2971,
      "step": 11411
    },
    {
      "epoch": 0.863686675180928,
      "grad_norm": 0.73046875,
      "learning_rate": 1.215780618244951e-05,
      "loss": 0.2938,
      "step": 11412
    },
    {
      "epoch": 0.8637623575043754,
      "grad_norm": 0.8515625,
      "learning_rate": 1.2156643146747195e-05,
      "loss": 0.3912,
      "step": 11413
    },
    {
      "epoch": 0.8638380398278227,
      "grad_norm": 0.734375,
      "learning_rate": 1.21554800804492e-05,
      "loss": 0.3051,
      "step": 11414
    },
    {
      "epoch": 0.86391372215127,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2154316983572021e-05,
      "loss": 0.3217,
      "step": 11415
    },
    {
      "epoch": 0.8639894044747174,
      "grad_norm": 0.75,
      "learning_rate": 1.2153153856132155e-05,
      "loss": 0.2957,
      "step": 11416
    },
    {
      "epoch": 0.8640650867981647,
      "grad_norm": 0.7109375,
      "learning_rate": 1.215199069814611e-05,
      "loss": 0.28,
      "step": 11417
    },
    {
      "epoch": 0.8641407691216121,
      "grad_norm": 0.7734375,
      "learning_rate": 1.2150827509630383e-05,
      "loss": 0.3421,
      "step": 11418
    },
    {
      "epoch": 0.8642164514450593,
      "grad_norm": 0.78125,
      "learning_rate": 1.2149664290601474e-05,
      "loss": 0.359,
      "step": 11419
    },
    {
      "epoch": 0.8642921337685067,
      "grad_norm": 0.69921875,
      "learning_rate": 1.2148501041075896e-05,
      "loss": 0.2603,
      "step": 11420
    },
    {
      "epoch": 0.864367816091954,
      "grad_norm": 0.82421875,
      "learning_rate": 1.2147337761070138e-05,
      "loss": 0.3617,
      "step": 11421
    },
    {
      "epoch": 0.8644434984154014,
      "grad_norm": 0.78515625,
      "learning_rate": 1.2146174450600707e-05,
      "loss": 0.3224,
      "step": 11422
    },
    {
      "epoch": 0.8645191807388487,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2145011109684112e-05,
      "loss": 0.3082,
      "step": 11423
    },
    {
      "epoch": 0.864594863062296,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2143847738336856e-05,
      "loss": 0.3191,
      "step": 11424
    },
    {
      "epoch": 0.8646705453857434,
      "grad_norm": 0.75,
      "learning_rate": 1.2142684336575434e-05,
      "loss": 0.3187,
      "step": 11425
    },
    {
      "epoch": 0.8647462277091906,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2141520904416362e-05,
      "loss": 0.3108,
      "step": 11426
    },
    {
      "epoch": 0.864821910032638,
      "grad_norm": 0.78515625,
      "learning_rate": 1.214035744187614e-05,
      "loss": 0.3263,
      "step": 11427
    },
    {
      "epoch": 0.8648975923560853,
      "grad_norm": 0.84765625,
      "learning_rate": 1.2139193948971273e-05,
      "loss": 0.3503,
      "step": 11428
    },
    {
      "epoch": 0.8649732746795327,
      "grad_norm": 0.7265625,
      "learning_rate": 1.2138030425718273e-05,
      "loss": 0.2968,
      "step": 11429
    },
    {
      "epoch": 0.86504895700298,
      "grad_norm": 0.78125,
      "learning_rate": 1.2136866872133641e-05,
      "loss": 0.3161,
      "step": 11430
    },
    {
      "epoch": 0.8651246393264274,
      "grad_norm": 0.73046875,
      "learning_rate": 1.2135703288233882e-05,
      "loss": 0.3171,
      "step": 11431
    },
    {
      "epoch": 0.8652003216498747,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2134539674035511e-05,
      "loss": 0.3165,
      "step": 11432
    },
    {
      "epoch": 0.8652760039733219,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2133376029555028e-05,
      "loss": 0.3105,
      "step": 11433
    },
    {
      "epoch": 0.8653516862967693,
      "grad_norm": 0.71875,
      "learning_rate": 1.2132212354808947e-05,
      "loss": 0.2956,
      "step": 11434
    },
    {
      "epoch": 0.8654273686202166,
      "grad_norm": 0.7734375,
      "learning_rate": 1.2131048649813778e-05,
      "loss": 0.3399,
      "step": 11435
    },
    {
      "epoch": 0.865503050943664,
      "grad_norm": 0.77734375,
      "learning_rate": 1.2129884914586024e-05,
      "loss": 0.3209,
      "step": 11436
    },
    {
      "epoch": 0.8655787332671113,
      "grad_norm": 0.75,
      "learning_rate": 1.2128721149142192e-05,
      "loss": 0.3068,
      "step": 11437
    },
    {
      "epoch": 0.8656544155905587,
      "grad_norm": 0.73828125,
      "learning_rate": 1.2127557353498807e-05,
      "loss": 0.3227,
      "step": 11438
    },
    {
      "epoch": 0.865730097914006,
      "grad_norm": 0.72265625,
      "learning_rate": 1.2126393527672367e-05,
      "loss": 0.298,
      "step": 11439
    },
    {
      "epoch": 0.8658057802374532,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2125229671679382e-05,
      "loss": 0.3161,
      "step": 11440
    },
    {
      "epoch": 0.8658814625609006,
      "grad_norm": 0.703125,
      "learning_rate": 1.212406578553637e-05,
      "loss": 0.2725,
      "step": 11441
    },
    {
      "epoch": 0.8659571448843479,
      "grad_norm": 0.7578125,
      "learning_rate": 1.212290186925984e-05,
      "loss": 0.3391,
      "step": 11442
    },
    {
      "epoch": 0.8660328272077953,
      "grad_norm": 0.79296875,
      "learning_rate": 1.2121737922866303e-05,
      "loss": 0.3559,
      "step": 11443
    },
    {
      "epoch": 0.8661085095312426,
      "grad_norm": 0.75,
      "learning_rate": 1.2120573946372272e-05,
      "loss": 0.3428,
      "step": 11444
    },
    {
      "epoch": 0.86618419185469,
      "grad_norm": 0.66796875,
      "learning_rate": 1.2119409939794262e-05,
      "loss": 0.2655,
      "step": 11445
    },
    {
      "epoch": 0.8662598741781373,
      "grad_norm": 0.77734375,
      "learning_rate": 1.2118245903148782e-05,
      "loss": 0.3328,
      "step": 11446
    },
    {
      "epoch": 0.8663355565015846,
      "grad_norm": 0.72265625,
      "learning_rate": 1.2117081836452353e-05,
      "loss": 0.2724,
      "step": 11447
    },
    {
      "epoch": 0.8664112388250319,
      "grad_norm": 0.73046875,
      "learning_rate": 1.2115917739721483e-05,
      "loss": 0.3146,
      "step": 11448
    },
    {
      "epoch": 0.8664869211484792,
      "grad_norm": 0.75,
      "learning_rate": 1.211475361297269e-05,
      "loss": 0.3174,
      "step": 11449
    },
    {
      "epoch": 0.8665626034719266,
      "grad_norm": 0.79296875,
      "learning_rate": 1.2113589456222488e-05,
      "loss": 0.3419,
      "step": 11450
    },
    {
      "epoch": 0.8666382857953739,
      "grad_norm": 0.7265625,
      "learning_rate": 1.211242526948739e-05,
      "loss": 0.2907,
      "step": 11451
    },
    {
      "epoch": 0.8667139681188213,
      "grad_norm": 0.71484375,
      "learning_rate": 1.2111261052783916e-05,
      "loss": 0.2827,
      "step": 11452
    },
    {
      "epoch": 0.8667896504422685,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2110096806128581e-05,
      "loss": 0.2879,
      "step": 11453
    },
    {
      "epoch": 0.8668653327657159,
      "grad_norm": 0.73046875,
      "learning_rate": 1.21089325295379e-05,
      "loss": 0.2617,
      "step": 11454
    },
    {
      "epoch": 0.8669410150891632,
      "grad_norm": 0.70703125,
      "learning_rate": 1.2107768223028395e-05,
      "loss": 0.2757,
      "step": 11455
    },
    {
      "epoch": 0.8670166974126106,
      "grad_norm": 0.75390625,
      "learning_rate": 1.210660388661658e-05,
      "loss": 0.3199,
      "step": 11456
    },
    {
      "epoch": 0.8670923797360579,
      "grad_norm": 0.765625,
      "learning_rate": 1.2105439520318973e-05,
      "loss": 0.2985,
      "step": 11457
    },
    {
      "epoch": 0.8671680620595053,
      "grad_norm": 0.703125,
      "learning_rate": 1.2104275124152094e-05,
      "loss": 0.261,
      "step": 11458
    },
    {
      "epoch": 0.8672437443829526,
      "grad_norm": 0.90625,
      "learning_rate": 1.2103110698132461e-05,
      "loss": 0.3939,
      "step": 11459
    },
    {
      "epoch": 0.8673194267063998,
      "grad_norm": 0.69921875,
      "learning_rate": 1.2101946242276594e-05,
      "loss": 0.2777,
      "step": 11460
    },
    {
      "epoch": 0.8673951090298472,
      "grad_norm": 0.75,
      "learning_rate": 1.210078175660101e-05,
      "loss": 0.2964,
      "step": 11461
    },
    {
      "epoch": 0.8674707913532945,
      "grad_norm": 1.0625,
      "learning_rate": 1.2099617241122237e-05,
      "loss": 0.3682,
      "step": 11462
    },
    {
      "epoch": 0.8675464736767419,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2098452695856787e-05,
      "loss": 0.3353,
      "step": 11463
    },
    {
      "epoch": 0.8676221560001892,
      "grad_norm": 0.8203125,
      "learning_rate": 1.2097288120821182e-05,
      "loss": 0.3043,
      "step": 11464
    },
    {
      "epoch": 0.8676978383236366,
      "grad_norm": 0.73828125,
      "learning_rate": 1.209612351603195e-05,
      "loss": 0.3124,
      "step": 11465
    },
    {
      "epoch": 0.8677735206470839,
      "grad_norm": 0.73828125,
      "learning_rate": 1.2094958881505608e-05,
      "loss": 0.3173,
      "step": 11466
    },
    {
      "epoch": 0.8678492029705311,
      "grad_norm": 0.734375,
      "learning_rate": 1.2093794217258679e-05,
      "loss": 0.2964,
      "step": 11467
    },
    {
      "epoch": 0.8679248852939785,
      "grad_norm": 0.7265625,
      "learning_rate": 1.2092629523307686e-05,
      "loss": 0.2816,
      "step": 11468
    },
    {
      "epoch": 0.8680005676174258,
      "grad_norm": 0.8203125,
      "learning_rate": 1.2091464799669151e-05,
      "loss": 0.3738,
      "step": 11469
    },
    {
      "epoch": 0.8680762499408732,
      "grad_norm": 0.79296875,
      "learning_rate": 1.2090300046359597e-05,
      "loss": 0.3428,
      "step": 11470
    },
    {
      "epoch": 0.8681519322643205,
      "grad_norm": 0.70703125,
      "learning_rate": 1.2089135263395556e-05,
      "loss": 0.2888,
      "step": 11471
    },
    {
      "epoch": 0.8682276145877679,
      "grad_norm": 0.921875,
      "learning_rate": 1.208797045079354e-05,
      "loss": 0.3555,
      "step": 11472
    },
    {
      "epoch": 0.8683032969112152,
      "grad_norm": 0.828125,
      "learning_rate": 1.2086805608570082e-05,
      "loss": 0.3795,
      "step": 11473
    },
    {
      "epoch": 0.8683789792346625,
      "grad_norm": 0.71875,
      "learning_rate": 1.2085640736741708e-05,
      "loss": 0.2741,
      "step": 11474
    },
    {
      "epoch": 0.8684546615581098,
      "grad_norm": 0.79296875,
      "learning_rate": 1.2084475835324935e-05,
      "loss": 0.3335,
      "step": 11475
    },
    {
      "epoch": 0.8685303438815571,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2083310904336302e-05,
      "loss": 0.2865,
      "step": 11476
    },
    {
      "epoch": 0.8686060262050045,
      "grad_norm": 0.70703125,
      "learning_rate": 1.2082145943792327e-05,
      "loss": 0.2755,
      "step": 11477
    },
    {
      "epoch": 0.8686817085284518,
      "grad_norm": 0.75,
      "learning_rate": 1.2080980953709536e-05,
      "loss": 0.3211,
      "step": 11478
    },
    {
      "epoch": 0.8687573908518992,
      "grad_norm": 0.71875,
      "learning_rate": 1.2079815934104462e-05,
      "loss": 0.2817,
      "step": 11479
    },
    {
      "epoch": 0.8688330731753465,
      "grad_norm": 0.72265625,
      "learning_rate": 1.207865088499363e-05,
      "loss": 0.2869,
      "step": 11480
    },
    {
      "epoch": 0.8689087554987938,
      "grad_norm": 0.69921875,
      "learning_rate": 1.2077485806393565e-05,
      "loss": 0.2755,
      "step": 11481
    },
    {
      "epoch": 0.8689844378222411,
      "grad_norm": 0.78515625,
      "learning_rate": 1.20763206983208e-05,
      "loss": 0.3138,
      "step": 11482
    },
    {
      "epoch": 0.8690601201456885,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2075155560791865e-05,
      "loss": 0.3244,
      "step": 11483
    },
    {
      "epoch": 0.8691358024691358,
      "grad_norm": 0.73046875,
      "learning_rate": 1.2073990393823282e-05,
      "loss": 0.3053,
      "step": 11484
    },
    {
      "epoch": 0.8692114847925831,
      "grad_norm": 0.71484375,
      "learning_rate": 1.2072825197431592e-05,
      "loss": 0.2911,
      "step": 11485
    },
    {
      "epoch": 0.8692871671160305,
      "grad_norm": 0.80078125,
      "learning_rate": 1.2071659971633318e-05,
      "loss": 0.3259,
      "step": 11486
    },
    {
      "epoch": 0.8693628494394778,
      "grad_norm": 0.76953125,
      "learning_rate": 1.207049471644499e-05,
      "loss": 0.3248,
      "step": 11487
    },
    {
      "epoch": 0.8694385317629251,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2069329431883144e-05,
      "loss": 0.292,
      "step": 11488
    },
    {
      "epoch": 0.8695142140863724,
      "grad_norm": 0.828125,
      "learning_rate": 1.2068164117964307e-05,
      "loss": 0.3883,
      "step": 11489
    },
    {
      "epoch": 0.8695898964098198,
      "grad_norm": 0.734375,
      "learning_rate": 1.2066998774705012e-05,
      "loss": 0.2996,
      "step": 11490
    },
    {
      "epoch": 0.8696655787332671,
      "grad_norm": 0.86328125,
      "learning_rate": 1.2065833402121796e-05,
      "loss": 0.319,
      "step": 11491
    },
    {
      "epoch": 0.8697412610567145,
      "grad_norm": 0.6953125,
      "learning_rate": 1.2064668000231185e-05,
      "loss": 0.2696,
      "step": 11492
    },
    {
      "epoch": 0.8698169433801618,
      "grad_norm": 0.75,
      "learning_rate": 1.2063502569049715e-05,
      "loss": 0.3164,
      "step": 11493
    },
    {
      "epoch": 0.8698926257036091,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2062337108593924e-05,
      "loss": 0.2956,
      "step": 11494
    },
    {
      "epoch": 0.8699683080270564,
      "grad_norm": 0.79296875,
      "learning_rate": 1.2061171618880338e-05,
      "loss": 0.3068,
      "step": 11495
    },
    {
      "epoch": 0.8700439903505037,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2060006099925496e-05,
      "loss": 0.3138,
      "step": 11496
    },
    {
      "epoch": 0.8701196726739511,
      "grad_norm": 0.80078125,
      "learning_rate": 1.2058840551745934e-05,
      "loss": 0.3368,
      "step": 11497
    },
    {
      "epoch": 0.8701953549973984,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2057674974358185e-05,
      "loss": 0.2995,
      "step": 11498
    },
    {
      "epoch": 0.8702710373208458,
      "grad_norm": 0.734375,
      "learning_rate": 1.2056509367778784e-05,
      "loss": 0.2736,
      "step": 11499
    },
    {
      "epoch": 0.8703467196442931,
      "grad_norm": 0.78515625,
      "learning_rate": 1.2055343732024271e-05,
      "loss": 0.3285,
      "step": 11500
    },
    {
      "epoch": 0.8704224019677405,
      "grad_norm": 0.7109375,
      "learning_rate": 1.205417806711118e-05,
      "loss": 0.2782,
      "step": 11501
    },
    {
      "epoch": 0.8704980842911877,
      "grad_norm": 0.73046875,
      "learning_rate": 1.2053012373056043e-05,
      "loss": 0.2844,
      "step": 11502
    },
    {
      "epoch": 0.870573766614635,
      "grad_norm": 0.71484375,
      "learning_rate": 1.2051846649875407e-05,
      "loss": 0.2978,
      "step": 11503
    },
    {
      "epoch": 0.8706494489380824,
      "grad_norm": 0.7734375,
      "learning_rate": 1.2050680897585805e-05,
      "loss": 0.3539,
      "step": 11504
    },
    {
      "epoch": 0.8707251312615297,
      "grad_norm": 0.76953125,
      "learning_rate": 1.2049515116203776e-05,
      "loss": 0.308,
      "step": 11505
    },
    {
      "epoch": 0.8708008135849771,
      "grad_norm": 0.69921875,
      "learning_rate": 1.2048349305745856e-05,
      "loss": 0.2737,
      "step": 11506
    },
    {
      "epoch": 0.8708764959084244,
      "grad_norm": 0.73828125,
      "learning_rate": 1.2047183466228587e-05,
      "loss": 0.3233,
      "step": 11507
    },
    {
      "epoch": 0.8709521782318718,
      "grad_norm": 0.765625,
      "learning_rate": 1.2046017597668507e-05,
      "loss": 0.3225,
      "step": 11508
    },
    {
      "epoch": 0.871027860555319,
      "grad_norm": 0.78125,
      "learning_rate": 1.2044851700082156e-05,
      "loss": 0.3112,
      "step": 11509
    },
    {
      "epoch": 0.8711035428787663,
      "grad_norm": 0.796875,
      "learning_rate": 1.2043685773486073e-05,
      "loss": 0.3386,
      "step": 11510
    },
    {
      "epoch": 0.8711792252022137,
      "grad_norm": 0.7890625,
      "learning_rate": 1.2042519817896805e-05,
      "loss": 0.3152,
      "step": 11511
    },
    {
      "epoch": 0.871254907525661,
      "grad_norm": 0.75390625,
      "learning_rate": 1.2041353833330884e-05,
      "loss": 0.3264,
      "step": 11512
    },
    {
      "epoch": 0.8713305898491084,
      "grad_norm": 0.87109375,
      "learning_rate": 1.2040187819804858e-05,
      "loss": 0.2754,
      "step": 11513
    },
    {
      "epoch": 0.8714062721725557,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2039021777335267e-05,
      "loss": 0.3258,
      "step": 11514
    },
    {
      "epoch": 0.8714819544960031,
      "grad_norm": 0.703125,
      "learning_rate": 1.2037855705938653e-05,
      "loss": 0.2894,
      "step": 11515
    },
    {
      "epoch": 0.8715576368194503,
      "grad_norm": 0.9296875,
      "learning_rate": 1.2036689605631557e-05,
      "loss": 0.3066,
      "step": 11516
    },
    {
      "epoch": 0.8716333191428977,
      "grad_norm": 0.82421875,
      "learning_rate": 1.2035523476430523e-05,
      "loss": 0.3082,
      "step": 11517
    },
    {
      "epoch": 0.871709001466345,
      "grad_norm": 0.7265625,
      "learning_rate": 1.2034357318352099e-05,
      "loss": 0.2822,
      "step": 11518
    },
    {
      "epoch": 0.8717846837897923,
      "grad_norm": 0.81640625,
      "learning_rate": 1.2033191131412822e-05,
      "loss": 0.3629,
      "step": 11519
    },
    {
      "epoch": 0.8718603661132397,
      "grad_norm": 0.70703125,
      "learning_rate": 1.2032024915629241e-05,
      "loss": 0.2766,
      "step": 11520
    },
    {
      "epoch": 0.871936048436687,
      "grad_norm": 0.7890625,
      "learning_rate": 1.20308586710179e-05,
      "loss": 0.3164,
      "step": 11521
    },
    {
      "epoch": 0.8720117307601344,
      "grad_norm": 0.7421875,
      "learning_rate": 1.2029692397595345e-05,
      "loss": 0.3038,
      "step": 11522
    },
    {
      "epoch": 0.8720874130835816,
      "grad_norm": 0.703125,
      "learning_rate": 1.2028526095378118e-05,
      "loss": 0.2831,
      "step": 11523
    },
    {
      "epoch": 0.872163095407029,
      "grad_norm": 0.69921875,
      "learning_rate": 1.202735976438277e-05,
      "loss": 0.2945,
      "step": 11524
    },
    {
      "epoch": 0.8722387777304763,
      "grad_norm": 0.72265625,
      "learning_rate": 1.2026193404625839e-05,
      "loss": 0.2884,
      "step": 11525
    },
    {
      "epoch": 0.8723144600539237,
      "grad_norm": 0.75390625,
      "learning_rate": 1.202502701612388e-05,
      "loss": 0.3209,
      "step": 11526
    },
    {
      "epoch": 0.872390142377371,
      "grad_norm": 0.74609375,
      "learning_rate": 1.2023860598893442e-05,
      "loss": 0.2949,
      "step": 11527
    },
    {
      "epoch": 0.8724658247008183,
      "grad_norm": 0.8203125,
      "learning_rate": 1.2022694152951065e-05,
      "loss": 0.3448,
      "step": 11528
    },
    {
      "epoch": 0.8725415070242657,
      "grad_norm": 0.73046875,
      "learning_rate": 1.2021527678313299e-05,
      "loss": 0.2927,
      "step": 11529
    },
    {
      "epoch": 0.8726171893477129,
      "grad_norm": 0.78515625,
      "learning_rate": 1.2020361174996694e-05,
      "loss": 0.3485,
      "step": 11530
    },
    {
      "epoch": 0.8726928716711603,
      "grad_norm": 0.75390625,
      "learning_rate": 1.20191946430178e-05,
      "loss": 0.2883,
      "step": 11531
    },
    {
      "epoch": 0.8727685539946076,
      "grad_norm": 0.765625,
      "learning_rate": 1.2018028082393167e-05,
      "loss": 0.3265,
      "step": 11532
    },
    {
      "epoch": 0.872844236318055,
      "grad_norm": 0.73046875,
      "learning_rate": 1.201686149313934e-05,
      "loss": 0.2869,
      "step": 11533
    },
    {
      "epoch": 0.8729199186415023,
      "grad_norm": 0.796875,
      "learning_rate": 1.2015694875272873e-05,
      "loss": 0.3117,
      "step": 11534
    },
    {
      "epoch": 0.8729956009649497,
      "grad_norm": 0.76171875,
      "learning_rate": 1.2014528228810311e-05,
      "loss": 0.334,
      "step": 11535
    },
    {
      "epoch": 0.873071283288397,
      "grad_norm": 0.81640625,
      "learning_rate": 1.2013361553768217e-05,
      "loss": 0.3825,
      "step": 11536
    },
    {
      "epoch": 0.8731469656118442,
      "grad_norm": 0.765625,
      "learning_rate": 1.2012194850163127e-05,
      "loss": 0.3346,
      "step": 11537
    },
    {
      "epoch": 0.8732226479352916,
      "grad_norm": 0.8203125,
      "learning_rate": 1.2011028118011602e-05,
      "loss": 0.3106,
      "step": 11538
    },
    {
      "epoch": 0.8732983302587389,
      "grad_norm": 0.8046875,
      "learning_rate": 1.2009861357330195e-05,
      "loss": 0.3524,
      "step": 11539
    },
    {
      "epoch": 0.8733740125821863,
      "grad_norm": 0.7265625,
      "learning_rate": 1.2008694568135451e-05,
      "loss": 0.2882,
      "step": 11540
    },
    {
      "epoch": 0.8734496949056336,
      "grad_norm": 0.7578125,
      "learning_rate": 1.2007527750443932e-05,
      "loss": 0.3054,
      "step": 11541
    },
    {
      "epoch": 0.873525377229081,
      "grad_norm": 0.71875,
      "learning_rate": 1.2006360904272186e-05,
      "loss": 0.3018,
      "step": 11542
    },
    {
      "epoch": 0.8736010595525283,
      "grad_norm": 0.703125,
      "learning_rate": 1.2005194029636761e-05,
      "loss": 0.2661,
      "step": 11543
    },
    {
      "epoch": 0.8736767418759755,
      "grad_norm": 0.80078125,
      "learning_rate": 1.2004027126554225e-05,
      "loss": 0.3354,
      "step": 11544
    },
    {
      "epoch": 0.8737524241994229,
      "grad_norm": 0.76171875,
      "learning_rate": 1.2002860195041124e-05,
      "loss": 0.3189,
      "step": 11545
    },
    {
      "epoch": 0.8738281065228702,
      "grad_norm": 0.73828125,
      "learning_rate": 1.2001693235114013e-05,
      "loss": 0.2836,
      "step": 11546
    },
    {
      "epoch": 0.8739037888463176,
      "grad_norm": 0.78515625,
      "learning_rate": 1.2000526246789449e-05,
      "loss": 0.321,
      "step": 11547
    },
    {
      "epoch": 0.8739794711697649,
      "grad_norm": 0.71875,
      "learning_rate": 1.1999359230083986e-05,
      "loss": 0.3046,
      "step": 11548
    },
    {
      "epoch": 0.8740551534932123,
      "grad_norm": 0.76953125,
      "learning_rate": 1.199819218501418e-05,
      "loss": 0.3125,
      "step": 11549
    },
    {
      "epoch": 0.8741308358166596,
      "grad_norm": 1.296875,
      "learning_rate": 1.1997025111596595e-05,
      "loss": 0.361,
      "step": 11550
    },
    {
      "epoch": 0.8742065181401069,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1995858009847776e-05,
      "loss": 0.3187,
      "step": 11551
    },
    {
      "epoch": 0.8742822004635542,
      "grad_norm": 0.78125,
      "learning_rate": 1.1994690879784287e-05,
      "loss": 0.3446,
      "step": 11552
    },
    {
      "epoch": 0.8743578827870016,
      "grad_norm": 0.69140625,
      "learning_rate": 1.1993523721422687e-05,
      "loss": 0.2672,
      "step": 11553
    },
    {
      "epoch": 0.8744335651104489,
      "grad_norm": 0.7109375,
      "learning_rate": 1.199235653477953e-05,
      "loss": 0.2735,
      "step": 11554
    },
    {
      "epoch": 0.8745092474338962,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1991189319871377e-05,
      "loss": 0.3278,
      "step": 11555
    },
    {
      "epoch": 0.8745849297573436,
      "grad_norm": 1.125,
      "learning_rate": 1.1990022076714787e-05,
      "loss": 0.3892,
      "step": 11556
    },
    {
      "epoch": 0.8746606120807909,
      "grad_norm": 0.77734375,
      "learning_rate": 1.198885480532632e-05,
      "loss": 0.3528,
      "step": 11557
    },
    {
      "epoch": 0.8747362944042382,
      "grad_norm": 0.7109375,
      "learning_rate": 1.1987687505722532e-05,
      "loss": 0.2785,
      "step": 11558
    },
    {
      "epoch": 0.8748119767276855,
      "grad_norm": 0.7734375,
      "learning_rate": 1.1986520177919986e-05,
      "loss": 0.3449,
      "step": 11559
    },
    {
      "epoch": 0.8748876590511329,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1985352821935243e-05,
      "loss": 0.3199,
      "step": 11560
    },
    {
      "epoch": 0.8749633413745802,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1984185437784862e-05,
      "loss": 0.2964,
      "step": 11561
    },
    {
      "epoch": 0.8750390236980276,
      "grad_norm": 0.7265625,
      "learning_rate": 1.1983018025485408e-05,
      "loss": 0.3088,
      "step": 11562
    },
    {
      "epoch": 0.8751147060214749,
      "grad_norm": 0.765625,
      "learning_rate": 1.198185058505344e-05,
      "loss": 0.3264,
      "step": 11563
    },
    {
      "epoch": 0.8751903883449222,
      "grad_norm": 0.74609375,
      "learning_rate": 1.198068311650552e-05,
      "loss": 0.3209,
      "step": 11564
    },
    {
      "epoch": 0.8752660706683695,
      "grad_norm": 0.89453125,
      "learning_rate": 1.1979515619858213e-05,
      "loss": 0.3115,
      "step": 11565
    },
    {
      "epoch": 0.8753417529918168,
      "grad_norm": 0.734375,
      "learning_rate": 1.1978348095128076e-05,
      "loss": 0.3148,
      "step": 11566
    },
    {
      "epoch": 0.8754174353152642,
      "grad_norm": 0.7734375,
      "learning_rate": 1.1977180542331677e-05,
      "loss": 0.2999,
      "step": 11567
    },
    {
      "epoch": 0.8754931176387115,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1976012961485582e-05,
      "loss": 0.3173,
      "step": 11568
    },
    {
      "epoch": 0.8755687999621589,
      "grad_norm": 0.73828125,
      "learning_rate": 1.197484535260635e-05,
      "loss": 0.2989,
      "step": 11569
    },
    {
      "epoch": 0.8756444822856062,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1973677715710547e-05,
      "loss": 0.3194,
      "step": 11570
    },
    {
      "epoch": 0.8757201646090536,
      "grad_norm": 0.78515625,
      "learning_rate": 1.1972510050814741e-05,
      "loss": 0.3282,
      "step": 11571
    },
    {
      "epoch": 0.8757958469325008,
      "grad_norm": 0.734375,
      "learning_rate": 1.1971342357935491e-05,
      "loss": 0.3035,
      "step": 11572
    },
    {
      "epoch": 0.8758715292559481,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1970174637089372e-05,
      "loss": 0.2786,
      "step": 11573
    },
    {
      "epoch": 0.8759472115793955,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1969006888292945e-05,
      "loss": 0.3602,
      "step": 11574
    },
    {
      "epoch": 0.8760228939028428,
      "grad_norm": 0.82421875,
      "learning_rate": 1.1967839111562771e-05,
      "loss": 0.366,
      "step": 11575
    },
    {
      "epoch": 0.8760985762262902,
      "grad_norm": 0.79296875,
      "learning_rate": 1.1966671306915425e-05,
      "loss": 0.3214,
      "step": 11576
    },
    {
      "epoch": 0.8761742585497375,
      "grad_norm": 0.828125,
      "learning_rate": 1.196550347436747e-05,
      "loss": 0.3593,
      "step": 11577
    },
    {
      "epoch": 0.8762499408731848,
      "grad_norm": 0.71875,
      "learning_rate": 1.1964335613935478e-05,
      "loss": 0.2804,
      "step": 11578
    },
    {
      "epoch": 0.8763256231966321,
      "grad_norm": 0.734375,
      "learning_rate": 1.1963167725636013e-05,
      "loss": 0.2964,
      "step": 11579
    },
    {
      "epoch": 0.8764013055200794,
      "grad_norm": 0.75,
      "learning_rate": 1.1961999809485648e-05,
      "loss": 0.3307,
      "step": 11580
    },
    {
      "epoch": 0.8764769878435268,
      "grad_norm": 1.0234375,
      "learning_rate": 1.1960831865500941e-05,
      "loss": 0.3556,
      "step": 11581
    },
    {
      "epoch": 0.8765526701669741,
      "grad_norm": 0.79296875,
      "learning_rate": 1.1959663893698475e-05,
      "loss": 0.3621,
      "step": 11582
    },
    {
      "epoch": 0.8766283524904215,
      "grad_norm": 0.70703125,
      "learning_rate": 1.1958495894094815e-05,
      "loss": 0.2739,
      "step": 11583
    },
    {
      "epoch": 0.8767040348138688,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1957327866706528e-05,
      "loss": 0.3176,
      "step": 11584
    },
    {
      "epoch": 0.8767797171373161,
      "grad_norm": 0.80078125,
      "learning_rate": 1.1956159811550185e-05,
      "loss": 0.3248,
      "step": 11585
    },
    {
      "epoch": 0.8768553994607634,
      "grad_norm": 0.765625,
      "learning_rate": 1.1954991728642358e-05,
      "loss": 0.3249,
      "step": 11586
    },
    {
      "epoch": 0.8769310817842108,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1953823617999618e-05,
      "loss": 0.3291,
      "step": 11587
    },
    {
      "epoch": 0.8770067641076581,
      "grad_norm": 0.81640625,
      "learning_rate": 1.1952655479638542e-05,
      "loss": 0.343,
      "step": 11588
    },
    {
      "epoch": 0.8770824464311054,
      "grad_norm": 0.796875,
      "learning_rate": 1.1951487313575693e-05,
      "loss": 0.3515,
      "step": 11589
    },
    {
      "epoch": 0.8771581287545528,
      "grad_norm": 0.9765625,
      "learning_rate": 1.1950319119827647e-05,
      "loss": 0.3089,
      "step": 11590
    },
    {
      "epoch": 0.8772338110780001,
      "grad_norm": 0.8046875,
      "learning_rate": 1.194915089841098e-05,
      "loss": 0.3385,
      "step": 11591
    },
    {
      "epoch": 0.8773094934014474,
      "grad_norm": 0.78125,
      "learning_rate": 1.1947982649342263e-05,
      "loss": 0.3275,
      "step": 11592
    },
    {
      "epoch": 0.8773851757248947,
      "grad_norm": 0.8046875,
      "learning_rate": 1.1946814372638066e-05,
      "loss": 0.3608,
      "step": 11593
    },
    {
      "epoch": 0.8774608580483421,
      "grad_norm": 0.81640625,
      "learning_rate": 1.194564606831497e-05,
      "loss": 0.3698,
      "step": 11594
    },
    {
      "epoch": 0.8775365403717894,
      "grad_norm": 0.66015625,
      "learning_rate": 1.1944477736389542e-05,
      "loss": 0.253,
      "step": 11595
    },
    {
      "epoch": 0.8776122226952368,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1943309376878361e-05,
      "loss": 0.2883,
      "step": 11596
    },
    {
      "epoch": 0.8776879050186841,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1942140989798006e-05,
      "loss": 0.307,
      "step": 11597
    },
    {
      "epoch": 0.8777635873421314,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1940972575165044e-05,
      "loss": 0.3118,
      "step": 11598
    },
    {
      "epoch": 0.8778392696655787,
      "grad_norm": 0.79296875,
      "learning_rate": 1.1939804132996056e-05,
      "loss": 0.3284,
      "step": 11599
    },
    {
      "epoch": 0.877914951989026,
      "grad_norm": 0.703125,
      "learning_rate": 1.1938635663307617e-05,
      "loss": 0.3078,
      "step": 11600
    },
    {
      "epoch": 0.8779906343124734,
      "grad_norm": 0.81640625,
      "learning_rate": 1.1937467166116305e-05,
      "loss": 0.3446,
      "step": 11601
    },
    {
      "epoch": 0.8780663166359207,
      "grad_norm": 0.8046875,
      "learning_rate": 1.1936298641438697e-05,
      "loss": 0.3359,
      "step": 11602
    },
    {
      "epoch": 0.8781419989593681,
      "grad_norm": 0.76953125,
      "learning_rate": 1.193513008929137e-05,
      "loss": 0.3488,
      "step": 11603
    },
    {
      "epoch": 0.8782176812828154,
      "grad_norm": 1.0390625,
      "learning_rate": 1.1933961509690902e-05,
      "loss": 0.329,
      "step": 11604
    },
    {
      "epoch": 0.8782933636062628,
      "grad_norm": 1.0625,
      "learning_rate": 1.1932792902653867e-05,
      "loss": 0.366,
      "step": 11605
    },
    {
      "epoch": 0.87836904592971,
      "grad_norm": 0.79296875,
      "learning_rate": 1.1931624268196855e-05,
      "loss": 0.3306,
      "step": 11606
    },
    {
      "epoch": 0.8784447282531573,
      "grad_norm": 0.828125,
      "learning_rate": 1.1930455606336435e-05,
      "loss": 0.319,
      "step": 11607
    },
    {
      "epoch": 0.8785204105766047,
      "grad_norm": 0.80078125,
      "learning_rate": 1.1929286917089187e-05,
      "loss": 0.3472,
      "step": 11608
    },
    {
      "epoch": 0.878596092900052,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1928118200471695e-05,
      "loss": 0.3088,
      "step": 11609
    },
    {
      "epoch": 0.8786717752234994,
      "grad_norm": 0.765625,
      "learning_rate": 1.1926949456500538e-05,
      "loss": 0.33,
      "step": 11610
    },
    {
      "epoch": 0.8787474575469467,
      "grad_norm": 1.0390625,
      "learning_rate": 1.1925780685192297e-05,
      "loss": 0.346,
      "step": 11611
    },
    {
      "epoch": 0.8788231398703941,
      "grad_norm": 0.74609375,
      "learning_rate": 1.192461188656355e-05,
      "loss": 0.3103,
      "step": 11612
    },
    {
      "epoch": 0.8788988221938413,
      "grad_norm": 0.86328125,
      "learning_rate": 1.1923443060630884e-05,
      "loss": 0.356,
      "step": 11613
    },
    {
      "epoch": 0.8789745045172886,
      "grad_norm": 0.734375,
      "learning_rate": 1.1922274207410874e-05,
      "loss": 0.2776,
      "step": 11614
    },
    {
      "epoch": 0.879050186840736,
      "grad_norm": 0.734375,
      "learning_rate": 1.1921105326920107e-05,
      "loss": 0.2688,
      "step": 11615
    },
    {
      "epoch": 0.8791258691641833,
      "grad_norm": 0.7734375,
      "learning_rate": 1.1919936419175161e-05,
      "loss": 0.3204,
      "step": 11616
    },
    {
      "epoch": 0.8792015514876307,
      "grad_norm": 0.828125,
      "learning_rate": 1.1918767484192628e-05,
      "loss": 0.3588,
      "step": 11617
    },
    {
      "epoch": 0.879277233811078,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1917598521989082e-05,
      "loss": 0.2896,
      "step": 11618
    },
    {
      "epoch": 0.8793529161345254,
      "grad_norm": 0.77734375,
      "learning_rate": 1.1916429532581111e-05,
      "loss": 0.3258,
      "step": 11619
    },
    {
      "epoch": 0.8794285984579726,
      "grad_norm": 1.0390625,
      "learning_rate": 1.19152605159853e-05,
      "loss": 0.4109,
      "step": 11620
    },
    {
      "epoch": 0.87950428078142,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1914091472218229e-05,
      "loss": 0.3105,
      "step": 11621
    },
    {
      "epoch": 0.8795799631048673,
      "grad_norm": 0.6953125,
      "learning_rate": 1.1912922401296488e-05,
      "loss": 0.2712,
      "step": 11622
    },
    {
      "epoch": 0.8796556454283146,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1911753303236656e-05,
      "loss": 0.3519,
      "step": 11623
    },
    {
      "epoch": 0.879731327751762,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1910584178055328e-05,
      "loss": 0.33,
      "step": 11624
    },
    {
      "epoch": 0.8798070100752093,
      "grad_norm": 0.71875,
      "learning_rate": 1.1909415025769083e-05,
      "loss": 0.2916,
      "step": 11625
    },
    {
      "epoch": 0.8798826923986567,
      "grad_norm": 0.81640625,
      "learning_rate": 1.1908245846394508e-05,
      "loss": 0.3328,
      "step": 11626
    },
    {
      "epoch": 0.8799583747221039,
      "grad_norm": 0.95703125,
      "learning_rate": 1.1907076639948194e-05,
      "loss": 0.3165,
      "step": 11627
    },
    {
      "epoch": 0.8800340570455513,
      "grad_norm": 0.7109375,
      "learning_rate": 1.190590740644672e-05,
      "loss": 0.2911,
      "step": 11628
    },
    {
      "epoch": 0.8801097393689986,
      "grad_norm": 0.81640625,
      "learning_rate": 1.1904738145906682e-05,
      "loss": 0.3639,
      "step": 11629
    },
    {
      "epoch": 0.880185421692446,
      "grad_norm": 0.8046875,
      "learning_rate": 1.1903568858344667e-05,
      "loss": 0.3035,
      "step": 11630
    },
    {
      "epoch": 0.8802611040158933,
      "grad_norm": 0.734375,
      "learning_rate": 1.1902399543777256e-05,
      "loss": 0.309,
      "step": 11631
    },
    {
      "epoch": 0.8803367863393406,
      "grad_norm": 0.765625,
      "learning_rate": 1.1901230202221048e-05,
      "loss": 0.3292,
      "step": 11632
    },
    {
      "epoch": 0.880412468662788,
      "grad_norm": 0.87890625,
      "learning_rate": 1.1900060833692624e-05,
      "loss": 0.3489,
      "step": 11633
    },
    {
      "epoch": 0.8804881509862352,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1898891438208578e-05,
      "loss": 0.3217,
      "step": 11634
    },
    {
      "epoch": 0.8805638333096826,
      "grad_norm": 0.71875,
      "learning_rate": 1.1897722015785496e-05,
      "loss": 0.2784,
      "step": 11635
    },
    {
      "epoch": 0.8806395156331299,
      "grad_norm": 0.78125,
      "learning_rate": 1.1896552566439972e-05,
      "loss": 0.3214,
      "step": 11636
    },
    {
      "epoch": 0.8807151979565773,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1895383090188592e-05,
      "loss": 0.3588,
      "step": 11637
    },
    {
      "epoch": 0.8807908802800246,
      "grad_norm": 0.78125,
      "learning_rate": 1.1894213587047955e-05,
      "loss": 0.3266,
      "step": 11638
    },
    {
      "epoch": 0.880866562603472,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1893044057034648e-05,
      "loss": 0.3033,
      "step": 11639
    },
    {
      "epoch": 0.8809422449269193,
      "grad_norm": 1.0546875,
      "learning_rate": 1.1891874500165259e-05,
      "loss": 0.3381,
      "step": 11640
    },
    {
      "epoch": 0.8810179272503665,
      "grad_norm": 0.7109375,
      "learning_rate": 1.1890704916456386e-05,
      "loss": 0.2914,
      "step": 11641
    },
    {
      "epoch": 0.8810936095738139,
      "grad_norm": 0.79296875,
      "learning_rate": 1.1889535305924619e-05,
      "loss": 0.3535,
      "step": 11642
    },
    {
      "epoch": 0.8811692918972612,
      "grad_norm": 0.796875,
      "learning_rate": 1.1888365668586548e-05,
      "loss": 0.3098,
      "step": 11643
    },
    {
      "epoch": 0.8812449742207086,
      "grad_norm": 0.78125,
      "learning_rate": 1.1887196004458774e-05,
      "loss": 0.3177,
      "step": 11644
    },
    {
      "epoch": 0.8813206565441559,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1886026313557886e-05,
      "loss": 0.3237,
      "step": 11645
    },
    {
      "epoch": 0.8813963388676033,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1884856595900474e-05,
      "loss": 0.31,
      "step": 11646
    },
    {
      "epoch": 0.8814720211910506,
      "grad_norm": 0.765625,
      "learning_rate": 1.188368685150314e-05,
      "loss": 0.3159,
      "step": 11647
    },
    {
      "epoch": 0.8815477035144978,
      "grad_norm": 0.6875,
      "learning_rate": 1.1882517080382476e-05,
      "loss": 0.2586,
      "step": 11648
    },
    {
      "epoch": 0.8816233858379452,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1881347282555073e-05,
      "loss": 0.3506,
      "step": 11649
    },
    {
      "epoch": 0.8816990681613925,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1880177458037534e-05,
      "loss": 0.3075,
      "step": 11650
    },
    {
      "epoch": 0.8817747504848399,
      "grad_norm": 0.7734375,
      "learning_rate": 1.1879007606846448e-05,
      "loss": 0.3289,
      "step": 11651
    },
    {
      "epoch": 0.8818504328082872,
      "grad_norm": 0.69921875,
      "learning_rate": 1.1877837728998416e-05,
      "loss": 0.2622,
      "step": 11652
    },
    {
      "epoch": 0.8819261151317346,
      "grad_norm": 0.73046875,
      "learning_rate": 1.1876667824510035e-05,
      "loss": 0.2909,
      "step": 11653
    },
    {
      "epoch": 0.8820017974551819,
      "grad_norm": 0.78515625,
      "learning_rate": 1.18754978933979e-05,
      "loss": 0.3288,
      "step": 11654
    },
    {
      "epoch": 0.8820774797786292,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1874327935678607e-05,
      "loss": 0.3255,
      "step": 11655
    },
    {
      "epoch": 0.8821531621020765,
      "grad_norm": 0.6796875,
      "learning_rate": 1.1873157951368757e-05,
      "loss": 0.2301,
      "step": 11656
    },
    {
      "epoch": 0.8822288444255239,
      "grad_norm": 0.9453125,
      "learning_rate": 1.1871987940484946e-05,
      "loss": 0.3478,
      "step": 11657
    },
    {
      "epoch": 0.8823045267489712,
      "grad_norm": 0.71875,
      "learning_rate": 1.1870817903043775e-05,
      "loss": 0.2792,
      "step": 11658
    },
    {
      "epoch": 0.8823802090724185,
      "grad_norm": 0.73046875,
      "learning_rate": 1.1869647839061841e-05,
      "loss": 0.2794,
      "step": 11659
    },
    {
      "epoch": 0.8824558913958659,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1868477748555745e-05,
      "loss": 0.2848,
      "step": 11660
    },
    {
      "epoch": 0.8825315737193132,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1867307631542082e-05,
      "loss": 0.3338,
      "step": 11661
    },
    {
      "epoch": 0.8826072560427605,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1866137488037461e-05,
      "loss": 0.307,
      "step": 11662
    },
    {
      "epoch": 0.8826829383662078,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1864967318058473e-05,
      "loss": 0.3151,
      "step": 11663
    },
    {
      "epoch": 0.8827586206896552,
      "grad_norm": 0.98828125,
      "learning_rate": 1.1863797121621725e-05,
      "loss": 0.3644,
      "step": 11664
    },
    {
      "epoch": 0.8828343030131025,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1862626898743819e-05,
      "loss": 0.307,
      "step": 11665
    },
    {
      "epoch": 0.8829099853365499,
      "grad_norm": 0.78125,
      "learning_rate": 1.1861456649441351e-05,
      "loss": 0.3185,
      "step": 11666
    },
    {
      "epoch": 0.8829856676599972,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1860286373730927e-05,
      "loss": 0.2911,
      "step": 11667
    },
    {
      "epoch": 0.8830613499834445,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1859116071629148e-05,
      "loss": 0.341,
      "step": 11668
    },
    {
      "epoch": 0.8831370323068918,
      "grad_norm": 0.7109375,
      "learning_rate": 1.1857945743152617e-05,
      "loss": 0.2748,
      "step": 11669
    },
    {
      "epoch": 0.8832127146303391,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1856775388317936e-05,
      "loss": 0.3063,
      "step": 11670
    },
    {
      "epoch": 0.8832883969537865,
      "grad_norm": 0.82421875,
      "learning_rate": 1.1855605007141713e-05,
      "loss": 0.3798,
      "step": 11671
    },
    {
      "epoch": 0.8833640792772338,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1854434599640547e-05,
      "loss": 0.3258,
      "step": 11672
    },
    {
      "epoch": 0.8834397616006812,
      "grad_norm": 0.71484375,
      "learning_rate": 1.1853264165831044e-05,
      "loss": 0.2915,
      "step": 11673
    },
    {
      "epoch": 0.8835154439241285,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1852093705729809e-05,
      "loss": 0.3276,
      "step": 11674
    },
    {
      "epoch": 0.8835911262475759,
      "grad_norm": 0.765625,
      "learning_rate": 1.1850923219353446e-05,
      "loss": 0.3444,
      "step": 11675
    },
    {
      "epoch": 0.8836668085710231,
      "grad_norm": 0.7421875,
      "learning_rate": 1.184975270671856e-05,
      "loss": 0.2929,
      "step": 11676
    },
    {
      "epoch": 0.8837424908944704,
      "grad_norm": 0.7109375,
      "learning_rate": 1.1848582167841758e-05,
      "loss": 0.2967,
      "step": 11677
    },
    {
      "epoch": 0.8838181732179178,
      "grad_norm": 0.69921875,
      "learning_rate": 1.1847411602739644e-05,
      "loss": 0.2745,
      "step": 11678
    },
    {
      "epoch": 0.8838938555413651,
      "grad_norm": 0.75,
      "learning_rate": 1.184624101142883e-05,
      "loss": 0.2864,
      "step": 11679
    },
    {
      "epoch": 0.8839695378648125,
      "grad_norm": 0.90234375,
      "learning_rate": 1.1845070393925917e-05,
      "loss": 0.3895,
      "step": 11680
    },
    {
      "epoch": 0.8840452201882598,
      "grad_norm": 0.796875,
      "learning_rate": 1.1843899750247514e-05,
      "loss": 0.327,
      "step": 11681
    },
    {
      "epoch": 0.8841209025117072,
      "grad_norm": 0.8359375,
      "learning_rate": 1.1842729080410229e-05,
      "loss": 0.3541,
      "step": 11682
    },
    {
      "epoch": 0.8841965848351544,
      "grad_norm": 0.85546875,
      "learning_rate": 1.184155838443067e-05,
      "loss": 0.3619,
      "step": 11683
    },
    {
      "epoch": 0.8842722671586017,
      "grad_norm": 0.7421875,
      "learning_rate": 1.1840387662325443e-05,
      "loss": 0.3165,
      "step": 11684
    },
    {
      "epoch": 0.8843479494820491,
      "grad_norm": 0.79296875,
      "learning_rate": 1.1839216914111162e-05,
      "loss": 0.359,
      "step": 11685
    },
    {
      "epoch": 0.8844236318054964,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1838046139804433e-05,
      "loss": 0.3356,
      "step": 11686
    },
    {
      "epoch": 0.8844993141289438,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1836875339421862e-05,
      "loss": 0.2935,
      "step": 11687
    },
    {
      "epoch": 0.8845749964523911,
      "grad_norm": 0.6953125,
      "learning_rate": 1.1835704512980067e-05,
      "loss": 0.2882,
      "step": 11688
    },
    {
      "epoch": 0.8846506787758385,
      "grad_norm": 0.734375,
      "learning_rate": 1.1834533660495649e-05,
      "loss": 0.3083,
      "step": 11689
    },
    {
      "epoch": 0.8847263610992857,
      "grad_norm": 0.78515625,
      "learning_rate": 1.1833362781985225e-05,
      "loss": 0.3205,
      "step": 11690
    },
    {
      "epoch": 0.884802043422733,
      "grad_norm": 0.796875,
      "learning_rate": 1.1832191877465401e-05,
      "loss": 0.3357,
      "step": 11691
    },
    {
      "epoch": 0.8848777257461804,
      "grad_norm": 0.8203125,
      "learning_rate": 1.1831020946952796e-05,
      "loss": 0.3363,
      "step": 11692
    },
    {
      "epoch": 0.8849534080696277,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1829849990464012e-05,
      "loss": 0.274,
      "step": 11693
    },
    {
      "epoch": 0.8850290903930751,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1828679008015666e-05,
      "loss": 0.3013,
      "step": 11694
    },
    {
      "epoch": 0.8851047727165224,
      "grad_norm": 0.77734375,
      "learning_rate": 1.1827507999624375e-05,
      "loss": 0.3299,
      "step": 11695
    },
    {
      "epoch": 0.8851804550399698,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1826336965306742e-05,
      "loss": 0.3198,
      "step": 11696
    },
    {
      "epoch": 0.885256137363417,
      "grad_norm": 0.81640625,
      "learning_rate": 1.1825165905079387e-05,
      "loss": 0.3513,
      "step": 11697
    },
    {
      "epoch": 0.8853318196868644,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1823994818958922e-05,
      "loss": 0.3138,
      "step": 11698
    },
    {
      "epoch": 0.8854075020103117,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1822823706961957e-05,
      "loss": 0.2867,
      "step": 11699
    },
    {
      "epoch": 0.885483184333759,
      "grad_norm": 0.75,
      "learning_rate": 1.1821652569105116e-05,
      "loss": 0.2937,
      "step": 11700
    },
    {
      "epoch": 0.8855588666572064,
      "grad_norm": 0.68359375,
      "learning_rate": 1.1820481405405003e-05,
      "loss": 0.2588,
      "step": 11701
    },
    {
      "epoch": 0.8856345489806537,
      "grad_norm": 0.77734375,
      "learning_rate": 1.1819310215878235e-05,
      "loss": 0.3261,
      "step": 11702
    },
    {
      "epoch": 0.885710231304101,
      "grad_norm": 0.75,
      "learning_rate": 1.1818139000541433e-05,
      "loss": 0.3133,
      "step": 11703
    },
    {
      "epoch": 0.8857859136275483,
      "grad_norm": 0.796875,
      "learning_rate": 1.1816967759411211e-05,
      "loss": 0.3383,
      "step": 11704
    },
    {
      "epoch": 0.8858615959509957,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1815796492504179e-05,
      "loss": 0.3097,
      "step": 11705
    },
    {
      "epoch": 0.885937278274443,
      "grad_norm": 0.71875,
      "learning_rate": 1.181462519983696e-05,
      "loss": 0.2951,
      "step": 11706
    },
    {
      "epoch": 0.8860129605978904,
      "grad_norm": 1.2421875,
      "learning_rate": 1.1813453881426167e-05,
      "loss": 0.3691,
      "step": 11707
    },
    {
      "epoch": 0.8860886429213377,
      "grad_norm": 0.71875,
      "learning_rate": 1.1812282537288419e-05,
      "loss": 0.2811,
      "step": 11708
    },
    {
      "epoch": 0.8861643252447851,
      "grad_norm": 0.7734375,
      "learning_rate": 1.1811111167440338e-05,
      "loss": 0.2424,
      "step": 11709
    },
    {
      "epoch": 0.8862400075682323,
      "grad_norm": 1.1328125,
      "learning_rate": 1.1809939771898531e-05,
      "loss": 0.3255,
      "step": 11710
    },
    {
      "epoch": 0.8863156898916796,
      "grad_norm": 0.859375,
      "learning_rate": 1.1808768350679623e-05,
      "loss": 0.3366,
      "step": 11711
    },
    {
      "epoch": 0.886391372215127,
      "grad_norm": 0.8828125,
      "learning_rate": 1.1807596903800234e-05,
      "loss": 0.336,
      "step": 11712
    },
    {
      "epoch": 0.8864670545385743,
      "grad_norm": 0.890625,
      "learning_rate": 1.180642543127698e-05,
      "loss": 0.3248,
      "step": 11713
    },
    {
      "epoch": 0.8865427368620217,
      "grad_norm": 0.6875,
      "learning_rate": 1.1805253933126482e-05,
      "loss": 0.2347,
      "step": 11714
    },
    {
      "epoch": 0.886618419185469,
      "grad_norm": 0.765625,
      "learning_rate": 1.180408240936536e-05,
      "loss": 0.303,
      "step": 11715
    },
    {
      "epoch": 0.8866941015089164,
      "grad_norm": 0.80078125,
      "learning_rate": 1.1802910860010233e-05,
      "loss": 0.3214,
      "step": 11716
    },
    {
      "epoch": 0.8867697838323636,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1801739285077718e-05,
      "loss": 0.2993,
      "step": 11717
    },
    {
      "epoch": 0.886845466155811,
      "grad_norm": 0.9140625,
      "learning_rate": 1.1800567684584444e-05,
      "loss": 0.315,
      "step": 11718
    },
    {
      "epoch": 0.8869211484792583,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1799396058547024e-05,
      "loss": 0.2956,
      "step": 11719
    },
    {
      "epoch": 0.8869968308027056,
      "grad_norm": 0.77734375,
      "learning_rate": 1.1798224406982084e-05,
      "loss": 0.3165,
      "step": 11720
    },
    {
      "epoch": 0.887072513126153,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1797052729906246e-05,
      "loss": 0.2895,
      "step": 11721
    },
    {
      "epoch": 0.8871481954496003,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1795881027336131e-05,
      "loss": 0.3207,
      "step": 11722
    },
    {
      "epoch": 0.8872238777730477,
      "grad_norm": 0.75390625,
      "learning_rate": 1.179470929928836e-05,
      "loss": 0.309,
      "step": 11723
    },
    {
      "epoch": 0.8872995600964949,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1793537545779562e-05,
      "loss": 0.3175,
      "step": 11724
    },
    {
      "epoch": 0.8873752424199423,
      "grad_norm": 0.7421875,
      "learning_rate": 1.1792365766826352e-05,
      "loss": 0.2991,
      "step": 11725
    },
    {
      "epoch": 0.8874509247433896,
      "grad_norm": 0.7109375,
      "learning_rate": 1.1791193962445358e-05,
      "loss": 0.2845,
      "step": 11726
    },
    {
      "epoch": 0.887526607066837,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1790022132653202e-05,
      "loss": 0.2955,
      "step": 11727
    },
    {
      "epoch": 0.8876022893902843,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1788850277466514e-05,
      "loss": 0.3258,
      "step": 11728
    },
    {
      "epoch": 0.8876779717137316,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1787678396901916e-05,
      "loss": 0.3362,
      "step": 11729
    },
    {
      "epoch": 0.887753654037179,
      "grad_norm": 0.70703125,
      "learning_rate": 1.178650649097603e-05,
      "loss": 0.2537,
      "step": 11730
    },
    {
      "epoch": 0.8878293363606262,
      "grad_norm": 0.984375,
      "learning_rate": 1.1785334559705484e-05,
      "loss": 0.3548,
      "step": 11731
    },
    {
      "epoch": 0.8879050186840736,
      "grad_norm": 0.79296875,
      "learning_rate": 1.1784162603106904e-05,
      "loss": 0.3615,
      "step": 11732
    },
    {
      "epoch": 0.8879807010075209,
      "grad_norm": 0.71484375,
      "learning_rate": 1.1782990621196918e-05,
      "loss": 0.2779,
      "step": 11733
    },
    {
      "epoch": 0.8880563833309683,
      "grad_norm": 0.73046875,
      "learning_rate": 1.1781818613992144e-05,
      "loss": 0.2849,
      "step": 11734
    },
    {
      "epoch": 0.8881320656544156,
      "grad_norm": 0.7265625,
      "learning_rate": 1.1780646581509219e-05,
      "loss": 0.3058,
      "step": 11735
    },
    {
      "epoch": 0.888207747977863,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1779474523764766e-05,
      "loss": 0.2969,
      "step": 11736
    },
    {
      "epoch": 0.8882834303013103,
      "grad_norm": 0.6953125,
      "learning_rate": 1.1778302440775413e-05,
      "loss": 0.2713,
      "step": 11737
    },
    {
      "epoch": 0.8883591126247575,
      "grad_norm": 0.78125,
      "learning_rate": 1.177713033255779e-05,
      "loss": 0.3211,
      "step": 11738
    },
    {
      "epoch": 0.8884347949482049,
      "grad_norm": 0.75,
      "learning_rate": 1.1775958199128524e-05,
      "loss": 0.3191,
      "step": 11739
    },
    {
      "epoch": 0.8885104772716522,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1774786040504238e-05,
      "loss": 0.3025,
      "step": 11740
    },
    {
      "epoch": 0.8885861595950996,
      "grad_norm": 0.859375,
      "learning_rate": 1.1773613856701572e-05,
      "loss": 0.4146,
      "step": 11741
    },
    {
      "epoch": 0.8886618419185469,
      "grad_norm": 0.80859375,
      "learning_rate": 1.177244164773715e-05,
      "loss": 0.3545,
      "step": 11742
    },
    {
      "epoch": 0.8887375242419943,
      "grad_norm": 0.765625,
      "learning_rate": 1.1771269413627597e-05,
      "loss": 0.3191,
      "step": 11743
    },
    {
      "epoch": 0.8888132065654416,
      "grad_norm": 0.78125,
      "learning_rate": 1.177009715438955e-05,
      "loss": 0.3366,
      "step": 11744
    },
    {
      "epoch": 0.8888888888888888,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1768924870039637e-05,
      "loss": 0.3115,
      "step": 11745
    },
    {
      "epoch": 0.8889645712123362,
      "grad_norm": 0.79296875,
      "learning_rate": 1.176775256059449e-05,
      "loss": 0.3476,
      "step": 11746
    },
    {
      "epoch": 0.8890402535357835,
      "grad_norm": 0.77734375,
      "learning_rate": 1.1766580226070741e-05,
      "loss": 0.3316,
      "step": 11747
    },
    {
      "epoch": 0.8891159358592309,
      "grad_norm": 0.734375,
      "learning_rate": 1.1765407866485017e-05,
      "loss": 0.3042,
      "step": 11748
    },
    {
      "epoch": 0.8891916181826782,
      "grad_norm": 0.8125,
      "learning_rate": 1.1764235481853951e-05,
      "loss": 0.3371,
      "step": 11749
    },
    {
      "epoch": 0.8892673005061256,
      "grad_norm": 0.765625,
      "learning_rate": 1.1763063072194181e-05,
      "loss": 0.3234,
      "step": 11750
    },
    {
      "epoch": 0.8893429828295729,
      "grad_norm": 0.78125,
      "learning_rate": 1.1761890637522337e-05,
      "loss": 0.3287,
      "step": 11751
    },
    {
      "epoch": 0.8894186651530201,
      "grad_norm": 0.73046875,
      "learning_rate": 1.1760718177855047e-05,
      "loss": 0.2982,
      "step": 11752
    },
    {
      "epoch": 0.8894943474764675,
      "grad_norm": 0.80078125,
      "learning_rate": 1.175954569320895e-05,
      "loss": 0.3548,
      "step": 11753
    },
    {
      "epoch": 0.8895700297999148,
      "grad_norm": 0.7421875,
      "learning_rate": 1.1758373183600678e-05,
      "loss": 0.3113,
      "step": 11754
    },
    {
      "epoch": 0.8896457121233622,
      "grad_norm": 0.78125,
      "learning_rate": 1.1757200649046863e-05,
      "loss": 0.2971,
      "step": 11755
    },
    {
      "epoch": 0.8897213944468095,
      "grad_norm": 0.7421875,
      "learning_rate": 1.1756028089564145e-05,
      "loss": 0.3092,
      "step": 11756
    },
    {
      "epoch": 0.8897970767702569,
      "grad_norm": 0.8046875,
      "learning_rate": 1.1754855505169154e-05,
      "loss": 0.3407,
      "step": 11757
    },
    {
      "epoch": 0.8898727590937042,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1753682895878522e-05,
      "loss": 0.3175,
      "step": 11758
    },
    {
      "epoch": 0.8899484414171515,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1752510261708892e-05,
      "loss": 0.3279,
      "step": 11759
    },
    {
      "epoch": 0.8900241237405988,
      "grad_norm": 0.68359375,
      "learning_rate": 1.17513376026769e-05,
      "loss": 0.2789,
      "step": 11760
    },
    {
      "epoch": 0.8900998060640462,
      "grad_norm": 0.7734375,
      "learning_rate": 1.1750164918799172e-05,
      "loss": 0.3214,
      "step": 11761
    },
    {
      "epoch": 0.8901754883874935,
      "grad_norm": 0.79296875,
      "learning_rate": 1.1748992210092354e-05,
      "loss": 0.3624,
      "step": 11762
    },
    {
      "epoch": 0.8902511707109408,
      "grad_norm": 0.72265625,
      "learning_rate": 1.174781947657308e-05,
      "loss": 0.3041,
      "step": 11763
    },
    {
      "epoch": 0.8903268530343882,
      "grad_norm": 0.765625,
      "learning_rate": 1.1746646718257987e-05,
      "loss": 0.3133,
      "step": 11764
    },
    {
      "epoch": 0.8904025353578355,
      "grad_norm": 0.80078125,
      "learning_rate": 1.1745473935163716e-05,
      "loss": 0.3278,
      "step": 11765
    },
    {
      "epoch": 0.8904782176812828,
      "grad_norm": 0.72265625,
      "learning_rate": 1.17443011273069e-05,
      "loss": 0.3069,
      "step": 11766
    },
    {
      "epoch": 0.8905539000047301,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1743128294704178e-05,
      "loss": 0.3136,
      "step": 11767
    },
    {
      "epoch": 0.8906295823281775,
      "grad_norm": 0.8203125,
      "learning_rate": 1.174195543737219e-05,
      "loss": 0.3099,
      "step": 11768
    },
    {
      "epoch": 0.8907052646516248,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1740782555327577e-05,
      "loss": 0.3043,
      "step": 11769
    },
    {
      "epoch": 0.8907809469750722,
      "grad_norm": 0.99609375,
      "learning_rate": 1.1739609648586973e-05,
      "loss": 0.3837,
      "step": 11770
    },
    {
      "epoch": 0.8908566292985195,
      "grad_norm": 0.77734375,
      "learning_rate": 1.1738436717167022e-05,
      "loss": 0.3378,
      "step": 11771
    },
    {
      "epoch": 0.8909323116219668,
      "grad_norm": 0.8984375,
      "learning_rate": 1.1737263761084366e-05,
      "loss": 0.3079,
      "step": 11772
    },
    {
      "epoch": 0.8910079939454141,
      "grad_norm": 0.75,
      "learning_rate": 1.1736090780355637e-05,
      "loss": 0.3318,
      "step": 11773
    },
    {
      "epoch": 0.8910836762688614,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1734917774997486e-05,
      "loss": 0.307,
      "step": 11774
    },
    {
      "epoch": 0.8911593585923088,
      "grad_norm": 0.71484375,
      "learning_rate": 1.1733744745026544e-05,
      "loss": 0.2966,
      "step": 11775
    },
    {
      "epoch": 0.8912350409157561,
      "grad_norm": 0.828125,
      "learning_rate": 1.1732571690459462e-05,
      "loss": 0.3687,
      "step": 11776
    },
    {
      "epoch": 0.8913107232392035,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1731398611312877e-05,
      "loss": 0.2989,
      "step": 11777
    },
    {
      "epoch": 0.8913864055626508,
      "grad_norm": 0.7421875,
      "learning_rate": 1.173022550760343e-05,
      "loss": 0.3019,
      "step": 11778
    },
    {
      "epoch": 0.8914620878860982,
      "grad_norm": 0.71484375,
      "learning_rate": 1.1729052379347766e-05,
      "loss": 0.3106,
      "step": 11779
    },
    {
      "epoch": 0.8915377702095454,
      "grad_norm": 0.73046875,
      "learning_rate": 1.1727879226562527e-05,
      "loss": 0.2943,
      "step": 11780
    },
    {
      "epoch": 0.8916134525329927,
      "grad_norm": 0.80078125,
      "learning_rate": 1.1726706049264353e-05,
      "loss": 0.3518,
      "step": 11781
    },
    {
      "epoch": 0.8916891348564401,
      "grad_norm": 0.77734375,
      "learning_rate": 1.1725532847469892e-05,
      "loss": 0.3316,
      "step": 11782
    },
    {
      "epoch": 0.8917648171798874,
      "grad_norm": 0.97265625,
      "learning_rate": 1.1724359621195788e-05,
      "loss": 0.3369,
      "step": 11783
    },
    {
      "epoch": 0.8918404995033348,
      "grad_norm": 0.73046875,
      "learning_rate": 1.1723186370458684e-05,
      "loss": 0.3144,
      "step": 11784
    },
    {
      "epoch": 0.8919161818267821,
      "grad_norm": 0.65234375,
      "learning_rate": 1.1722013095275223e-05,
      "loss": 0.2582,
      "step": 11785
    },
    {
      "epoch": 0.8919918641502295,
      "grad_norm": 0.7890625,
      "learning_rate": 1.172083979566205e-05,
      "loss": 0.3375,
      "step": 11786
    },
    {
      "epoch": 0.8920675464736767,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1719666471635813e-05,
      "loss": 0.3359,
      "step": 11787
    },
    {
      "epoch": 0.892143228797124,
      "grad_norm": 1.140625,
      "learning_rate": 1.1718493123213157e-05,
      "loss": 0.3896,
      "step": 11788
    },
    {
      "epoch": 0.8922189111205714,
      "grad_norm": 0.75,
      "learning_rate": 1.1717319750410727e-05,
      "loss": 0.3065,
      "step": 11789
    },
    {
      "epoch": 0.8922945934440187,
      "grad_norm": 0.7265625,
      "learning_rate": 1.1716146353245165e-05,
      "loss": 0.3168,
      "step": 11790
    },
    {
      "epoch": 0.8923702757674661,
      "grad_norm": 0.70703125,
      "learning_rate": 1.1714972931733125e-05,
      "loss": 0.291,
      "step": 11791
    },
    {
      "epoch": 0.8924459580909134,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1713799485891253e-05,
      "loss": 0.318,
      "step": 11792
    },
    {
      "epoch": 0.8925216404143608,
      "grad_norm": 0.75,
      "learning_rate": 1.1712626015736193e-05,
      "loss": 0.3312,
      "step": 11793
    },
    {
      "epoch": 0.892597322737808,
      "grad_norm": 0.75,
      "learning_rate": 1.1711452521284593e-05,
      "loss": 0.3062,
      "step": 11794
    },
    {
      "epoch": 0.8926730050612554,
      "grad_norm": 0.69140625,
      "learning_rate": 1.1710279002553103e-05,
      "loss": 0.2817,
      "step": 11795
    },
    {
      "epoch": 0.8927486873847027,
      "grad_norm": 0.78125,
      "learning_rate": 1.1709105459558367e-05,
      "loss": 0.3284,
      "step": 11796
    },
    {
      "epoch": 0.89282436970815,
      "grad_norm": 0.81640625,
      "learning_rate": 1.1707931892317041e-05,
      "loss": 0.3662,
      "step": 11797
    },
    {
      "epoch": 0.8929000520315974,
      "grad_norm": 0.71875,
      "learning_rate": 1.170675830084577e-05,
      "loss": 0.2811,
      "step": 11798
    },
    {
      "epoch": 0.8929757343550447,
      "grad_norm": 0.953125,
      "learning_rate": 1.1705584685161202e-05,
      "loss": 0.3497,
      "step": 11799
    },
    {
      "epoch": 0.8930514166784921,
      "grad_norm": 0.75,
      "learning_rate": 1.1704411045279989e-05,
      "loss": 0.2957,
      "step": 11800
    },
    {
      "epoch": 0.8931270990019393,
      "grad_norm": 0.8359375,
      "learning_rate": 1.1703237381218782e-05,
      "loss": 0.3534,
      "step": 11801
    },
    {
      "epoch": 0.8932027813253867,
      "grad_norm": 0.765625,
      "learning_rate": 1.1702063692994227e-05,
      "loss": 0.2993,
      "step": 11802
    },
    {
      "epoch": 0.893278463648834,
      "grad_norm": 0.7890625,
      "learning_rate": 1.170088998062298e-05,
      "loss": 0.3532,
      "step": 11803
    },
    {
      "epoch": 0.8933541459722814,
      "grad_norm": 0.78515625,
      "learning_rate": 1.169971624412169e-05,
      "loss": 0.3254,
      "step": 11804
    },
    {
      "epoch": 0.8934298282957287,
      "grad_norm": 0.78125,
      "learning_rate": 1.1698542483507004e-05,
      "loss": 0.3194,
      "step": 11805
    },
    {
      "epoch": 0.893505510619176,
      "grad_norm": 0.82421875,
      "learning_rate": 1.1697368698795581e-05,
      "loss": 0.3031,
      "step": 11806
    },
    {
      "epoch": 0.8935811929426234,
      "grad_norm": 0.77734375,
      "learning_rate": 1.1696194890004076e-05,
      "loss": 0.2701,
      "step": 11807
    },
    {
      "epoch": 0.8936568752660706,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1695021057149128e-05,
      "loss": 0.3234,
      "step": 11808
    },
    {
      "epoch": 0.893732557589518,
      "grad_norm": 0.78515625,
      "learning_rate": 1.16938472002474e-05,
      "loss": 0.3195,
      "step": 11809
    },
    {
      "epoch": 0.8938082399129653,
      "grad_norm": 0.8125,
      "learning_rate": 1.1692673319315541e-05,
      "loss": 0.344,
      "step": 11810
    },
    {
      "epoch": 0.8938839222364127,
      "grad_norm": 0.8125,
      "learning_rate": 1.1691499414370207e-05,
      "loss": 0.356,
      "step": 11811
    },
    {
      "epoch": 0.89395960455986,
      "grad_norm": 0.78515625,
      "learning_rate": 1.1690325485428054e-05,
      "loss": 0.3548,
      "step": 11812
    },
    {
      "epoch": 0.8940352868833074,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1689151532505732e-05,
      "loss": 0.3001,
      "step": 11813
    },
    {
      "epoch": 0.8941109692067547,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1687977555619894e-05,
      "loss": 0.2952,
      "step": 11814
    },
    {
      "epoch": 0.8941866515302019,
      "grad_norm": 0.671875,
      "learning_rate": 1.1686803554787198e-05,
      "loss": 0.2438,
      "step": 11815
    },
    {
      "epoch": 0.8942623338536493,
      "grad_norm": 0.80078125,
      "learning_rate": 1.1685629530024305e-05,
      "loss": 0.3546,
      "step": 11816
    },
    {
      "epoch": 0.8943380161770966,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1684455481347858e-05,
      "loss": 0.3434,
      "step": 11817
    },
    {
      "epoch": 0.894413698500544,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1683281408774521e-05,
      "loss": 0.3092,
      "step": 11818
    },
    {
      "epoch": 0.8944893808239913,
      "grad_norm": 0.80078125,
      "learning_rate": 1.168210731232095e-05,
      "loss": 0.3384,
      "step": 11819
    },
    {
      "epoch": 0.8945650631474387,
      "grad_norm": 0.78125,
      "learning_rate": 1.1680933192003796e-05,
      "loss": 0.3234,
      "step": 11820
    },
    {
      "epoch": 0.8946407454708859,
      "grad_norm": 1.015625,
      "learning_rate": 1.1679759047839727e-05,
      "loss": 0.3472,
      "step": 11821
    },
    {
      "epoch": 0.8947164277943332,
      "grad_norm": 0.765625,
      "learning_rate": 1.1678584879845389e-05,
      "loss": 0.2808,
      "step": 11822
    },
    {
      "epoch": 0.8947921101177806,
      "grad_norm": 0.765625,
      "learning_rate": 1.1677410688037443e-05,
      "loss": 0.3256,
      "step": 11823
    },
    {
      "epoch": 0.8948677924412279,
      "grad_norm": 0.6953125,
      "learning_rate": 1.1676236472432549e-05,
      "loss": 0.2897,
      "step": 11824
    },
    {
      "epoch": 0.8949434747646753,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1675062233047365e-05,
      "loss": 0.3018,
      "step": 11825
    },
    {
      "epoch": 0.8950191570881226,
      "grad_norm": 0.69921875,
      "learning_rate": 1.167388796989855e-05,
      "loss": 0.2665,
      "step": 11826
    },
    {
      "epoch": 0.89509483941157,
      "grad_norm": 0.7421875,
      "learning_rate": 1.1672713683002759e-05,
      "loss": 0.299,
      "step": 11827
    },
    {
      "epoch": 0.8951705217350172,
      "grad_norm": 0.75,
      "learning_rate": 1.1671539372376654e-05,
      "loss": 0.2846,
      "step": 11828
    },
    {
      "epoch": 0.8952462040584646,
      "grad_norm": 0.8203125,
      "learning_rate": 1.1670365038036896e-05,
      "loss": 0.3509,
      "step": 11829
    },
    {
      "epoch": 0.8953218863819119,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1669190680000142e-05,
      "loss": 0.3302,
      "step": 11830
    },
    {
      "epoch": 0.8953975687053592,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1668016298283055e-05,
      "loss": 0.3052,
      "step": 11831
    },
    {
      "epoch": 0.8954732510288066,
      "grad_norm": 0.78125,
      "learning_rate": 1.1666841892902294e-05,
      "loss": 0.3512,
      "step": 11832
    },
    {
      "epoch": 0.8955489333522539,
      "grad_norm": 0.7265625,
      "learning_rate": 1.1665667463874521e-05,
      "loss": 0.2933,
      "step": 11833
    },
    {
      "epoch": 0.8956246156757013,
      "grad_norm": 0.80078125,
      "learning_rate": 1.1664493011216396e-05,
      "loss": 0.335,
      "step": 11834
    },
    {
      "epoch": 0.8957002979991485,
      "grad_norm": 0.78515625,
      "learning_rate": 1.1663318534944581e-05,
      "loss": 0.331,
      "step": 11835
    },
    {
      "epoch": 0.8957759803225959,
      "grad_norm": 0.7578125,
      "learning_rate": 1.166214403507574e-05,
      "loss": 0.3151,
      "step": 11836
    },
    {
      "epoch": 0.8958516626460432,
      "grad_norm": 0.734375,
      "learning_rate": 1.1660969511626529e-05,
      "loss": 0.3034,
      "step": 11837
    },
    {
      "epoch": 0.8959273449694906,
      "grad_norm": 0.67578125,
      "learning_rate": 1.1659794964613619e-05,
      "loss": 0.2583,
      "step": 11838
    },
    {
      "epoch": 0.8960030272929379,
      "grad_norm": 0.8359375,
      "learning_rate": 1.1658620394053668e-05,
      "loss": 0.3803,
      "step": 11839
    },
    {
      "epoch": 0.8960787096163852,
      "grad_norm": 0.75,
      "learning_rate": 1.1657445799963343e-05,
      "loss": 0.3151,
      "step": 11840
    },
    {
      "epoch": 0.8961543919398326,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1656271182359302e-05,
      "loss": 0.2925,
      "step": 11841
    },
    {
      "epoch": 0.8962300742632798,
      "grad_norm": 0.765625,
      "learning_rate": 1.1655096541258213e-05,
      "loss": 0.3071,
      "step": 11842
    },
    {
      "epoch": 0.8963057565867272,
      "grad_norm": 0.76171875,
      "learning_rate": 1.165392187667674e-05,
      "loss": 0.3188,
      "step": 11843
    },
    {
      "epoch": 0.8963814389101745,
      "grad_norm": 0.8515625,
      "learning_rate": 1.1652747188631543e-05,
      "loss": 0.3658,
      "step": 11844
    },
    {
      "epoch": 0.8964571212336219,
      "grad_norm": 0.828125,
      "learning_rate": 1.1651572477139298e-05,
      "loss": 0.3509,
      "step": 11845
    },
    {
      "epoch": 0.8965328035570692,
      "grad_norm": 0.78125,
      "learning_rate": 1.1650397742216657e-05,
      "loss": 0.3471,
      "step": 11846
    },
    {
      "epoch": 0.8966084858805166,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1649222983880294e-05,
      "loss": 0.3357,
      "step": 11847
    },
    {
      "epoch": 0.8966841682039639,
      "grad_norm": 0.6875,
      "learning_rate": 1.1648048202146873e-05,
      "loss": 0.2446,
      "step": 11848
    },
    {
      "epoch": 0.8967598505274111,
      "grad_norm": 0.73828125,
      "learning_rate": 1.164687339703306e-05,
      "loss": 0.2751,
      "step": 11849
    },
    {
      "epoch": 0.8968355328508585,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1645698568555522e-05,
      "loss": 0.3381,
      "step": 11850
    },
    {
      "epoch": 0.8969112151743058,
      "grad_norm": 0.69921875,
      "learning_rate": 1.1644523716730926e-05,
      "loss": 0.2668,
      "step": 11851
    },
    {
      "epoch": 0.8969868974977532,
      "grad_norm": 0.78515625,
      "learning_rate": 1.1643348841575936e-05,
      "loss": 0.3419,
      "step": 11852
    },
    {
      "epoch": 0.8970625798212005,
      "grad_norm": 0.796875,
      "learning_rate": 1.1642173943107227e-05,
      "loss": 0.3358,
      "step": 11853
    },
    {
      "epoch": 0.8971382621446479,
      "grad_norm": 0.72265625,
      "learning_rate": 1.164099902134146e-05,
      "loss": 0.2916,
      "step": 11854
    },
    {
      "epoch": 0.8972139444680952,
      "grad_norm": 0.77734375,
      "learning_rate": 1.1639824076295306e-05,
      "loss": 0.3546,
      "step": 11855
    },
    {
      "epoch": 0.8972896267915424,
      "grad_norm": 0.79296875,
      "learning_rate": 1.1638649107985433e-05,
      "loss": 0.3377,
      "step": 11856
    },
    {
      "epoch": 0.8973653091149898,
      "grad_norm": 0.7734375,
      "learning_rate": 1.1637474116428513e-05,
      "loss": 0.3239,
      "step": 11857
    },
    {
      "epoch": 0.8974409914384371,
      "grad_norm": 0.71875,
      "learning_rate": 1.1636299101641209e-05,
      "loss": 0.3107,
      "step": 11858
    },
    {
      "epoch": 0.8975166737618845,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1635124063640198e-05,
      "loss": 0.2976,
      "step": 11859
    },
    {
      "epoch": 0.8975923560853318,
      "grad_norm": 0.6953125,
      "learning_rate": 1.1633949002442145e-05,
      "loss": 0.2635,
      "step": 11860
    },
    {
      "epoch": 0.8976680384087792,
      "grad_norm": 0.7109375,
      "learning_rate": 1.1632773918063718e-05,
      "loss": 0.2982,
      "step": 11861
    },
    {
      "epoch": 0.8977437207322265,
      "grad_norm": 0.78515625,
      "learning_rate": 1.1631598810521596e-05,
      "loss": 0.339,
      "step": 11862
    },
    {
      "epoch": 0.8978194030556738,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1630423679832442e-05,
      "loss": 0.3104,
      "step": 11863
    },
    {
      "epoch": 0.8978950853791211,
      "grad_norm": 0.73828125,
      "learning_rate": 1.162924852601293e-05,
      "loss": 0.3137,
      "step": 11864
    },
    {
      "epoch": 0.8979707677025685,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1628073349079734e-05,
      "loss": 0.3234,
      "step": 11865
    },
    {
      "epoch": 0.8980464500260158,
      "grad_norm": 0.71875,
      "learning_rate": 1.1626898149049523e-05,
      "loss": 0.2804,
      "step": 11866
    },
    {
      "epoch": 0.8981221323494631,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1625722925938967e-05,
      "loss": 0.2851,
      "step": 11867
    },
    {
      "epoch": 0.8981978146729105,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1624547679764745e-05,
      "loss": 0.3275,
      "step": 11868
    },
    {
      "epoch": 0.8982734969963578,
      "grad_norm": 0.7734375,
      "learning_rate": 1.1623372410543527e-05,
      "loss": 0.3216,
      "step": 11869
    },
    {
      "epoch": 0.8983491793198051,
      "grad_norm": 0.734375,
      "learning_rate": 1.1622197118291982e-05,
      "loss": 0.2972,
      "step": 11870
    },
    {
      "epoch": 0.8984248616432524,
      "grad_norm": 0.75,
      "learning_rate": 1.1621021803026789e-05,
      "loss": 0.3069,
      "step": 11871
    },
    {
      "epoch": 0.8985005439666998,
      "grad_norm": 0.7421875,
      "learning_rate": 1.161984646476462e-05,
      "loss": 0.2903,
      "step": 11872
    },
    {
      "epoch": 0.8985762262901471,
      "grad_norm": 0.84765625,
      "learning_rate": 1.1618671103522148e-05,
      "loss": 0.3561,
      "step": 11873
    },
    {
      "epoch": 0.8986519086135945,
      "grad_norm": 0.73828125,
      "learning_rate": 1.161749571931605e-05,
      "loss": 0.3122,
      "step": 11874
    },
    {
      "epoch": 0.8987275909370418,
      "grad_norm": 1.046875,
      "learning_rate": 1.1616320312162999e-05,
      "loss": 0.3642,
      "step": 11875
    },
    {
      "epoch": 0.8988032732604891,
      "grad_norm": 0.765625,
      "learning_rate": 1.161514488207967e-05,
      "loss": 0.3193,
      "step": 11876
    },
    {
      "epoch": 0.8988789555839364,
      "grad_norm": 0.71484375,
      "learning_rate": 1.1613969429082742e-05,
      "loss": 0.2709,
      "step": 11877
    },
    {
      "epoch": 0.8989546379073837,
      "grad_norm": 0.6953125,
      "learning_rate": 1.1612793953188885e-05,
      "loss": 0.2878,
      "step": 11878
    },
    {
      "epoch": 0.8990303202308311,
      "grad_norm": 0.7265625,
      "learning_rate": 1.1611618454414777e-05,
      "loss": 0.2661,
      "step": 11879
    },
    {
      "epoch": 0.8991060025542784,
      "grad_norm": 0.75,
      "learning_rate": 1.1610442932777098e-05,
      "loss": 0.3247,
      "step": 11880
    },
    {
      "epoch": 0.8991816848777258,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1609267388292523e-05,
      "loss": 0.3287,
      "step": 11881
    },
    {
      "epoch": 0.8992573672011731,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1608091820977728e-05,
      "loss": 0.2798,
      "step": 11882
    },
    {
      "epoch": 0.8993330495246205,
      "grad_norm": 0.76171875,
      "learning_rate": 1.160691623084939e-05,
      "loss": 0.3133,
      "step": 11883
    },
    {
      "epoch": 0.8994087318480677,
      "grad_norm": 0.9609375,
      "learning_rate": 1.1605740617924188e-05,
      "loss": 0.3349,
      "step": 11884
    },
    {
      "epoch": 0.899484414171515,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1604564982218798e-05,
      "loss": 0.31,
      "step": 11885
    },
    {
      "epoch": 0.8995600964949624,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1603389323749902e-05,
      "loss": 0.3279,
      "step": 11886
    },
    {
      "epoch": 0.8996357788184097,
      "grad_norm": 0.8046875,
      "learning_rate": 1.1602213642534177e-05,
      "loss": 0.3377,
      "step": 11887
    },
    {
      "epoch": 0.8997114611418571,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1601037938588302e-05,
      "loss": 0.3014,
      "step": 11888
    },
    {
      "epoch": 0.8997871434653044,
      "grad_norm": 0.7734375,
      "learning_rate": 1.1599862211928958e-05,
      "loss": 0.315,
      "step": 11889
    },
    {
      "epoch": 0.8998628257887518,
      "grad_norm": 0.734375,
      "learning_rate": 1.1598686462572822e-05,
      "loss": 0.3135,
      "step": 11890
    },
    {
      "epoch": 0.899938508112199,
      "grad_norm": 0.80859375,
      "learning_rate": 1.1597510690536576e-05,
      "loss": 0.3523,
      "step": 11891
    },
    {
      "epoch": 0.9000141904356463,
      "grad_norm": 0.70703125,
      "learning_rate": 1.1596334895836896e-05,
      "loss": 0.2731,
      "step": 11892
    },
    {
      "epoch": 0.9000898727590937,
      "grad_norm": 0.78125,
      "learning_rate": 1.1595159078490465e-05,
      "loss": 0.308,
      "step": 11893
    },
    {
      "epoch": 0.900165555082541,
      "grad_norm": 0.71484375,
      "learning_rate": 1.1593983238513971e-05,
      "loss": 0.2638,
      "step": 11894
    },
    {
      "epoch": 0.9002412374059884,
      "grad_norm": 0.7421875,
      "learning_rate": 1.1592807375924084e-05,
      "loss": 0.2833,
      "step": 11895
    },
    {
      "epoch": 0.9003169197294357,
      "grad_norm": 0.8046875,
      "learning_rate": 1.1591631490737495e-05,
      "loss": 0.3632,
      "step": 11896
    },
    {
      "epoch": 0.9003926020528831,
      "grad_norm": 0.75,
      "learning_rate": 1.159045558297088e-05,
      "loss": 0.3026,
      "step": 11897
    },
    {
      "epoch": 0.9004682843763303,
      "grad_norm": 0.79296875,
      "learning_rate": 1.1589279652640926e-05,
      "loss": 0.3138,
      "step": 11898
    },
    {
      "epoch": 0.9004682843763303,
      "eval_loss": 0.32722821831703186,
      "eval_runtime": 83.5552,
      "eval_samples_per_second": 58.189,
      "eval_steps_per_second": 58.189,
      "step": 11898
    },
    {
      "epoch": 0.9005439666997777,
      "grad_norm": 0.7421875,
      "learning_rate": 1.1588103699764311e-05,
      "loss": 0.3179,
      "step": 11899
    },
    {
      "epoch": 0.900619649023225,
      "grad_norm": 0.7421875,
      "learning_rate": 1.1586927724357717e-05,
      "loss": 0.2809,
      "step": 11900
    },
    {
      "epoch": 0.9006953313466723,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1585751726437835e-05,
      "loss": 0.295,
      "step": 11901
    },
    {
      "epoch": 0.9007710136701197,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1584575706021337e-05,
      "loss": 0.3259,
      "step": 11902
    },
    {
      "epoch": 0.900846695993567,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1583399663124919e-05,
      "loss": 0.3287,
      "step": 11903
    },
    {
      "epoch": 0.9009223783170144,
      "grad_norm": 0.734375,
      "learning_rate": 1.158222359776526e-05,
      "loss": 0.3282,
      "step": 11904
    },
    {
      "epoch": 0.9009980606404616,
      "grad_norm": 0.80078125,
      "learning_rate": 1.158104750995904e-05,
      "loss": 0.3412,
      "step": 11905
    },
    {
      "epoch": 0.901073742963909,
      "grad_norm": 0.79296875,
      "learning_rate": 1.157987139972295e-05,
      "loss": 0.3641,
      "step": 11906
    },
    {
      "epoch": 0.9011494252873563,
      "grad_norm": 0.75,
      "learning_rate": 1.1578695267073672e-05,
      "loss": 0.3342,
      "step": 11907
    },
    {
      "epoch": 0.9012251076108037,
      "grad_norm": 0.76171875,
      "learning_rate": 1.157751911202789e-05,
      "loss": 0.3236,
      "step": 11908
    },
    {
      "epoch": 0.901300789934251,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1576342934602298e-05,
      "loss": 0.3435,
      "step": 11909
    },
    {
      "epoch": 0.9013764722576983,
      "grad_norm": 0.734375,
      "learning_rate": 1.1575166734813576e-05,
      "loss": 0.3145,
      "step": 11910
    },
    {
      "epoch": 0.9014521545811457,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1573990512678406e-05,
      "loss": 0.3099,
      "step": 11911
    },
    {
      "epoch": 0.9015278369045929,
      "grad_norm": 0.75,
      "learning_rate": 1.1572814268213483e-05,
      "loss": 0.3026,
      "step": 11912
    },
    {
      "epoch": 0.9016035192280403,
      "grad_norm": 0.72265625,
      "learning_rate": 1.157163800143549e-05,
      "loss": 0.2905,
      "step": 11913
    },
    {
      "epoch": 0.9016792015514876,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1570461712361114e-05,
      "loss": 0.3041,
      "step": 11914
    },
    {
      "epoch": 0.901754883874935,
      "grad_norm": 0.84765625,
      "learning_rate": 1.1569285401007046e-05,
      "loss": 0.3781,
      "step": 11915
    },
    {
      "epoch": 0.9018305661983823,
      "grad_norm": 0.80859375,
      "learning_rate": 1.156810906738997e-05,
      "loss": 0.3234,
      "step": 11916
    },
    {
      "epoch": 0.9019062485218297,
      "grad_norm": 0.90625,
      "learning_rate": 1.1566932711526575e-05,
      "loss": 0.264,
      "step": 11917
    },
    {
      "epoch": 0.901981930845277,
      "grad_norm": 0.7734375,
      "learning_rate": 1.156575633343355e-05,
      "loss": 0.3021,
      "step": 11918
    },
    {
      "epoch": 0.9020576131687242,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1564579933127589e-05,
      "loss": 0.3086,
      "step": 11919
    },
    {
      "epoch": 0.9021332954921716,
      "grad_norm": 0.73828125,
      "learning_rate": 1.156340351062537e-05,
      "loss": 0.2888,
      "step": 11920
    },
    {
      "epoch": 0.9022089778156189,
      "grad_norm": 0.80078125,
      "learning_rate": 1.1562227065943595e-05,
      "loss": 0.3669,
      "step": 11921
    },
    {
      "epoch": 0.9022846601390663,
      "grad_norm": 0.75,
      "learning_rate": 1.1561050599098946e-05,
      "loss": 0.2954,
      "step": 11922
    },
    {
      "epoch": 0.9023603424625136,
      "grad_norm": 0.71484375,
      "learning_rate": 1.1559874110108113e-05,
      "loss": 0.2994,
      "step": 11923
    },
    {
      "epoch": 0.902436024785961,
      "grad_norm": 0.8046875,
      "learning_rate": 1.1558697598987794e-05,
      "loss": 0.3599,
      "step": 11924
    },
    {
      "epoch": 0.9025117071094083,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1557521065754673e-05,
      "loss": 0.2936,
      "step": 11925
    },
    {
      "epoch": 0.9025873894328555,
      "grad_norm": 0.69921875,
      "learning_rate": 1.1556344510425439e-05,
      "loss": 0.2916,
      "step": 11926
    },
    {
      "epoch": 0.9026630717563029,
      "grad_norm": 0.7421875,
      "learning_rate": 1.155516793301679e-05,
      "loss": 0.3111,
      "step": 11927
    },
    {
      "epoch": 0.9027387540797502,
      "grad_norm": 1.0078125,
      "learning_rate": 1.1553991333545419e-05,
      "loss": 0.3302,
      "step": 11928
    },
    {
      "epoch": 0.9028144364031976,
      "grad_norm": 0.7265625,
      "learning_rate": 1.1552814712028009e-05,
      "loss": 0.3049,
      "step": 11929
    },
    {
      "epoch": 0.9028901187266449,
      "grad_norm": 0.8046875,
      "learning_rate": 1.1551638068481259e-05,
      "loss": 0.3356,
      "step": 11930
    },
    {
      "epoch": 0.9029658010500923,
      "grad_norm": 0.7734375,
      "learning_rate": 1.1550461402921863e-05,
      "loss": 0.3144,
      "step": 11931
    },
    {
      "epoch": 0.9030414833735396,
      "grad_norm": 0.7109375,
      "learning_rate": 1.1549284715366506e-05,
      "loss": 0.3028,
      "step": 11932
    },
    {
      "epoch": 0.9031171656969869,
      "grad_norm": 0.765625,
      "learning_rate": 1.154810800583189e-05,
      "loss": 0.3215,
      "step": 11933
    },
    {
      "epoch": 0.9031928480204342,
      "grad_norm": 0.6875,
      "learning_rate": 1.1546931274334708e-05,
      "loss": 0.2895,
      "step": 11934
    },
    {
      "epoch": 0.9032685303438815,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1545754520891645e-05,
      "loss": 0.2933,
      "step": 11935
    },
    {
      "epoch": 0.9033442126673289,
      "grad_norm": 0.703125,
      "learning_rate": 1.1544577745519404e-05,
      "loss": 0.2903,
      "step": 11936
    },
    {
      "epoch": 0.9034198949907762,
      "grad_norm": 1.078125,
      "learning_rate": 1.1543400948234677e-05,
      "loss": 0.3696,
      "step": 11937
    },
    {
      "epoch": 0.9034955773142236,
      "grad_norm": 0.76953125,
      "learning_rate": 1.154222412905416e-05,
      "loss": 0.3188,
      "step": 11938
    },
    {
      "epoch": 0.9035712596376709,
      "grad_norm": 0.78515625,
      "learning_rate": 1.1541047287994548e-05,
      "loss": 0.3462,
      "step": 11939
    },
    {
      "epoch": 0.9036469419611182,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1539870425072535e-05,
      "loss": 0.354,
      "step": 11940
    },
    {
      "epoch": 0.9037226242845655,
      "grad_norm": 0.79296875,
      "learning_rate": 1.1538693540304815e-05,
      "loss": 0.3287,
      "step": 11941
    },
    {
      "epoch": 0.9037983066080129,
      "grad_norm": 0.7578125,
      "learning_rate": 1.153751663370809e-05,
      "loss": 0.3345,
      "step": 11942
    },
    {
      "epoch": 0.9038739889314602,
      "grad_norm": 0.8046875,
      "learning_rate": 1.1536339705299051e-05,
      "loss": 0.3342,
      "step": 11943
    },
    {
      "epoch": 0.9039496712549075,
      "grad_norm": 0.7421875,
      "learning_rate": 1.1535162755094397e-05,
      "loss": 0.3119,
      "step": 11944
    },
    {
      "epoch": 0.9040253535783549,
      "grad_norm": 0.71484375,
      "learning_rate": 1.1533985783110829e-05,
      "loss": 0.3122,
      "step": 11945
    },
    {
      "epoch": 0.9041010359018021,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1532808789365036e-05,
      "loss": 0.3159,
      "step": 11946
    },
    {
      "epoch": 0.9041767182252495,
      "grad_norm": 0.8515625,
      "learning_rate": 1.1531631773873719e-05,
      "loss": 0.3109,
      "step": 11947
    },
    {
      "epoch": 0.9042524005486968,
      "grad_norm": 0.76953125,
      "learning_rate": 1.153045473665358e-05,
      "loss": 0.2646,
      "step": 11948
    },
    {
      "epoch": 0.9043280828721442,
      "grad_norm": 0.76171875,
      "learning_rate": 1.152927767772131e-05,
      "loss": 0.3146,
      "step": 11949
    },
    {
      "epoch": 0.9044037651955915,
      "grad_norm": 0.9453125,
      "learning_rate": 1.1528100597093617e-05,
      "loss": 0.3738,
      "step": 11950
    },
    {
      "epoch": 0.9044794475190389,
      "grad_norm": 0.73046875,
      "learning_rate": 1.152692349478719e-05,
      "loss": 0.3084,
      "step": 11951
    },
    {
      "epoch": 0.9045551298424862,
      "grad_norm": 0.765625,
      "learning_rate": 1.1525746370818735e-05,
      "loss": 0.3333,
      "step": 11952
    },
    {
      "epoch": 0.9046308121659334,
      "grad_norm": 0.78515625,
      "learning_rate": 1.1524569225204951e-05,
      "loss": 0.3001,
      "step": 11953
    },
    {
      "epoch": 0.9047064944893808,
      "grad_norm": 0.7734375,
      "learning_rate": 1.1523392057962538e-05,
      "loss": 0.321,
      "step": 11954
    },
    {
      "epoch": 0.9047821768128281,
      "grad_norm": 0.8828125,
      "learning_rate": 1.152221486910819e-05,
      "loss": 0.3187,
      "step": 11955
    },
    {
      "epoch": 0.9048578591362755,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1521037658658612e-05,
      "loss": 0.3459,
      "step": 11956
    },
    {
      "epoch": 0.9049335414597228,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1519860426630507e-05,
      "loss": 0.2758,
      "step": 11957
    },
    {
      "epoch": 0.9050092237831702,
      "grad_norm": 0.94921875,
      "learning_rate": 1.1518683173040573e-05,
      "loss": 0.3112,
      "step": 11958
    },
    {
      "epoch": 0.9050849061066175,
      "grad_norm": 0.78515625,
      "learning_rate": 1.151750589790551e-05,
      "loss": 0.3226,
      "step": 11959
    },
    {
      "epoch": 0.9051605884300647,
      "grad_norm": 0.765625,
      "learning_rate": 1.1516328601242024e-05,
      "loss": 0.3199,
      "step": 11960
    },
    {
      "epoch": 0.9052362707535121,
      "grad_norm": 0.80078125,
      "learning_rate": 1.1515151283066814e-05,
      "loss": 0.332,
      "step": 11961
    },
    {
      "epoch": 0.9053119530769594,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1513973943396584e-05,
      "loss": 0.3142,
      "step": 11962
    },
    {
      "epoch": 0.9053876354004068,
      "grad_norm": 0.79296875,
      "learning_rate": 1.1512796582248034e-05,
      "loss": 0.3689,
      "step": 11963
    },
    {
      "epoch": 0.9054633177238541,
      "grad_norm": 0.70703125,
      "learning_rate": 1.1511619199637867e-05,
      "loss": 0.2724,
      "step": 11964
    },
    {
      "epoch": 0.9055390000473015,
      "grad_norm": 0.70703125,
      "learning_rate": 1.151044179558279e-05,
      "loss": 0.3063,
      "step": 11965
    },
    {
      "epoch": 0.9056146823707488,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1509264370099506e-05,
      "loss": 0.306,
      "step": 11966
    },
    {
      "epoch": 0.9056903646941961,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1508086923204714e-05,
      "loss": 0.33,
      "step": 11967
    },
    {
      "epoch": 0.9057660470176434,
      "grad_norm": 0.69921875,
      "learning_rate": 1.150690945491512e-05,
      "loss": 0.2931,
      "step": 11968
    },
    {
      "epoch": 0.9058417293410908,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1505731965247432e-05,
      "loss": 0.3194,
      "step": 11969
    },
    {
      "epoch": 0.9059174116645381,
      "grad_norm": 0.7890625,
      "learning_rate": 1.150455445421835e-05,
      "loss": 0.3356,
      "step": 11970
    },
    {
      "epoch": 0.9059930939879854,
      "grad_norm": 0.734375,
      "learning_rate": 1.1503376921844584e-05,
      "loss": 0.3082,
      "step": 11971
    },
    {
      "epoch": 0.9060687763114328,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1502199368142835e-05,
      "loss": 0.3097,
      "step": 11972
    },
    {
      "epoch": 0.9061444586348801,
      "grad_norm": 0.76953125,
      "learning_rate": 1.150102179312981e-05,
      "loss": 0.3099,
      "step": 11973
    },
    {
      "epoch": 0.9062201409583274,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1499844196822213e-05,
      "loss": 0.3168,
      "step": 11974
    },
    {
      "epoch": 0.9062958232817747,
      "grad_norm": 0.703125,
      "learning_rate": 1.1498666579236756e-05,
      "loss": 0.311,
      "step": 11975
    },
    {
      "epoch": 0.9063715056052221,
      "grad_norm": 0.75,
      "learning_rate": 1.149748894039014e-05,
      "loss": 0.3048,
      "step": 11976
    },
    {
      "epoch": 0.9064471879286694,
      "grad_norm": 0.7109375,
      "learning_rate": 1.1496311280299073e-05,
      "loss": 0.2794,
      "step": 11977
    },
    {
      "epoch": 0.9065228702521168,
      "grad_norm": 0.83203125,
      "learning_rate": 1.1495133598980263e-05,
      "loss": 0.377,
      "step": 11978
    },
    {
      "epoch": 0.9065985525755641,
      "grad_norm": 0.78515625,
      "learning_rate": 1.1493955896450417e-05,
      "loss": 0.3269,
      "step": 11979
    },
    {
      "epoch": 0.9066742348990114,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1492778172726246e-05,
      "loss": 0.3209,
      "step": 11980
    },
    {
      "epoch": 0.9067499172224587,
      "grad_norm": 0.79296875,
      "learning_rate": 1.1491600427824451e-05,
      "loss": 0.3352,
      "step": 11981
    },
    {
      "epoch": 0.906825599545906,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1490422661761744e-05,
      "loss": 0.2996,
      "step": 11982
    },
    {
      "epoch": 0.9069012818693534,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1489244874554837e-05,
      "loss": 0.3068,
      "step": 11983
    },
    {
      "epoch": 0.9069769641928007,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1488067066220433e-05,
      "loss": 0.2907,
      "step": 11984
    },
    {
      "epoch": 0.9070526465162481,
      "grad_norm": 0.78125,
      "learning_rate": 1.1486889236775246e-05,
      "loss": 0.2889,
      "step": 11985
    },
    {
      "epoch": 0.9071283288396954,
      "grad_norm": 0.98046875,
      "learning_rate": 1.1485711386235985e-05,
      "loss": 0.3293,
      "step": 11986
    },
    {
      "epoch": 0.9072040111631428,
      "grad_norm": 0.9140625,
      "learning_rate": 1.1484533514619357e-05,
      "loss": 0.309,
      "step": 11987
    },
    {
      "epoch": 0.90727969348659,
      "grad_norm": 0.7734375,
      "learning_rate": 1.1483355621942069e-05,
      "loss": 0.3016,
      "step": 11988
    },
    {
      "epoch": 0.9073553758100373,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1482177708220843e-05,
      "loss": 0.3078,
      "step": 11989
    },
    {
      "epoch": 0.9074310581334847,
      "grad_norm": 0.67578125,
      "learning_rate": 1.148099977347238e-05,
      "loss": 0.2543,
      "step": 11990
    },
    {
      "epoch": 0.907506740456932,
      "grad_norm": 0.921875,
      "learning_rate": 1.1479821817713392e-05,
      "loss": 0.3178,
      "step": 11991
    },
    {
      "epoch": 0.9075824227803794,
      "grad_norm": 0.78125,
      "learning_rate": 1.1478643840960593e-05,
      "loss": 0.3113,
      "step": 11992
    },
    {
      "epoch": 0.9076581051038267,
      "grad_norm": 0.7734375,
      "learning_rate": 1.1477465843230695e-05,
      "loss": 0.3162,
      "step": 11993
    },
    {
      "epoch": 0.9077337874272741,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1476287824540409e-05,
      "loss": 0.3115,
      "step": 11994
    },
    {
      "epoch": 0.9078094697507213,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1475109784906444e-05,
      "loss": 0.3332,
      "step": 11995
    },
    {
      "epoch": 0.9078851520741686,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1473931724345517e-05,
      "loss": 0.327,
      "step": 11996
    },
    {
      "epoch": 0.907960834397616,
      "grad_norm": 0.88671875,
      "learning_rate": 1.1472753642874337e-05,
      "loss": 0.3785,
      "step": 11997
    },
    {
      "epoch": 0.9080365167210633,
      "grad_norm": 0.7109375,
      "learning_rate": 1.1471575540509624e-05,
      "loss": 0.2919,
      "step": 11998
    },
    {
      "epoch": 0.9081121990445107,
      "grad_norm": 0.796875,
      "learning_rate": 1.1470397417268082e-05,
      "loss": 0.31,
      "step": 11999
    },
    {
      "epoch": 0.908187881367958,
      "grad_norm": 0.8203125,
      "learning_rate": 1.1469219273166432e-05,
      "loss": 0.3588,
      "step": 12000
    },
    {
      "epoch": 0.9082635636914054,
      "grad_norm": 0.78125,
      "learning_rate": 1.1468041108221385e-05,
      "loss": 0.3423,
      "step": 12001
    },
    {
      "epoch": 0.9083392460148526,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1466862922449655e-05,
      "loss": 0.3165,
      "step": 12002
    },
    {
      "epoch": 0.9084149283383,
      "grad_norm": 0.828125,
      "learning_rate": 1.1465684715867958e-05,
      "loss": 0.348,
      "step": 12003
    },
    {
      "epoch": 0.9084906106617473,
      "grad_norm": 0.69921875,
      "learning_rate": 1.1464506488493008e-05,
      "loss": 0.2864,
      "step": 12004
    },
    {
      "epoch": 0.9085662929851946,
      "grad_norm": 0.71484375,
      "learning_rate": 1.1463328240341519e-05,
      "loss": 0.2852,
      "step": 12005
    },
    {
      "epoch": 0.908641975308642,
      "grad_norm": 0.765625,
      "learning_rate": 1.1462149971430207e-05,
      "loss": 0.3133,
      "step": 12006
    },
    {
      "epoch": 0.9087176576320893,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1460971681775789e-05,
      "loss": 0.295,
      "step": 12007
    },
    {
      "epoch": 0.9087933399555367,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1459793371394983e-05,
      "loss": 0.2945,
      "step": 12008
    },
    {
      "epoch": 0.9088690222789839,
      "grad_norm": 0.71875,
      "learning_rate": 1.1458615040304501e-05,
      "loss": 0.2928,
      "step": 12009
    },
    {
      "epoch": 0.9089447046024313,
      "grad_norm": 0.69921875,
      "learning_rate": 1.1457436688521065e-05,
      "loss": 0.2634,
      "step": 12010
    },
    {
      "epoch": 0.9090203869258786,
      "grad_norm": 0.73046875,
      "learning_rate": 1.1456258316061382e-05,
      "loss": 0.2882,
      "step": 12011
    },
    {
      "epoch": 0.909096069249326,
      "grad_norm": 0.72265625,
      "learning_rate": 1.145507992294218e-05,
      "loss": 0.2943,
      "step": 12012
    },
    {
      "epoch": 0.9091717515727733,
      "grad_norm": 0.73046875,
      "learning_rate": 1.1453901509180175e-05,
      "loss": 0.2826,
      "step": 12013
    },
    {
      "epoch": 0.9092474338962206,
      "grad_norm": 0.734375,
      "learning_rate": 1.1452723074792077e-05,
      "loss": 0.316,
      "step": 12014
    },
    {
      "epoch": 0.909323116219668,
      "grad_norm": 0.7890625,
      "learning_rate": 1.145154461979461e-05,
      "loss": 0.326,
      "step": 12015
    },
    {
      "epoch": 0.9093987985431152,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1450366144204493e-05,
      "loss": 0.3285,
      "step": 12016
    },
    {
      "epoch": 0.9094744808665626,
      "grad_norm": 0.73046875,
      "learning_rate": 1.1449187648038445e-05,
      "loss": 0.3037,
      "step": 12017
    },
    {
      "epoch": 0.9095501631900099,
      "grad_norm": 0.71484375,
      "learning_rate": 1.144800913131318e-05,
      "loss": 0.3065,
      "step": 12018
    },
    {
      "epoch": 0.9096258455134573,
      "grad_norm": 0.83984375,
      "learning_rate": 1.1446830594045421e-05,
      "loss": 0.3898,
      "step": 12019
    },
    {
      "epoch": 0.9097015278369046,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1445652036251889e-05,
      "loss": 0.2994,
      "step": 12020
    },
    {
      "epoch": 0.909777210160352,
      "grad_norm": 0.66796875,
      "learning_rate": 1.14444734579493e-05,
      "loss": 0.2567,
      "step": 12021
    },
    {
      "epoch": 0.9098528924837993,
      "grad_norm": 0.7734375,
      "learning_rate": 1.144329485915438e-05,
      "loss": 0.3287,
      "step": 12022
    },
    {
      "epoch": 0.9099285748072465,
      "grad_norm": 0.71875,
      "learning_rate": 1.1442116239883841e-05,
      "loss": 0.3116,
      "step": 12023
    },
    {
      "epoch": 0.9100042571306939,
      "grad_norm": 0.734375,
      "learning_rate": 1.144093760015441e-05,
      "loss": 0.2813,
      "step": 12024
    },
    {
      "epoch": 0.9100799394541412,
      "grad_norm": 0.70703125,
      "learning_rate": 1.1439758939982808e-05,
      "loss": 0.3009,
      "step": 12025
    },
    {
      "epoch": 0.9101556217775886,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1438580259385752e-05,
      "loss": 0.301,
      "step": 12026
    },
    {
      "epoch": 0.9102313041010359,
      "grad_norm": 0.81640625,
      "learning_rate": 1.143740155837997e-05,
      "loss": 0.3795,
      "step": 12027
    },
    {
      "epoch": 0.9103069864244833,
      "grad_norm": 0.81640625,
      "learning_rate": 1.1436222836982179e-05,
      "loss": 0.3333,
      "step": 12028
    },
    {
      "epoch": 0.9103826687479306,
      "grad_norm": 1.0,
      "learning_rate": 1.1435044095209099e-05,
      "loss": 0.3306,
      "step": 12029
    },
    {
      "epoch": 0.9104583510713778,
      "grad_norm": 0.87109375,
      "learning_rate": 1.1433865333077458e-05,
      "loss": 0.3205,
      "step": 12030
    },
    {
      "epoch": 0.9105340333948252,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1432686550603979e-05,
      "loss": 0.3474,
      "step": 12031
    },
    {
      "epoch": 0.9106097157182725,
      "grad_norm": 0.6953125,
      "learning_rate": 1.143150774780538e-05,
      "loss": 0.2746,
      "step": 12032
    },
    {
      "epoch": 0.9106853980417199,
      "grad_norm": 0.8046875,
      "learning_rate": 1.1430328924698389e-05,
      "loss": 0.3403,
      "step": 12033
    },
    {
      "epoch": 0.9107610803651672,
      "grad_norm": 0.875,
      "learning_rate": 1.1429150081299725e-05,
      "loss": 0.3828,
      "step": 12034
    },
    {
      "epoch": 0.9108367626886146,
      "grad_norm": 0.73046875,
      "learning_rate": 1.1427971217626116e-05,
      "loss": 0.2983,
      "step": 12035
    },
    {
      "epoch": 0.9109124450120619,
      "grad_norm": 0.7421875,
      "learning_rate": 1.1426792333694285e-05,
      "loss": 0.3008,
      "step": 12036
    },
    {
      "epoch": 0.9109881273355092,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1425613429520956e-05,
      "loss": 0.3038,
      "step": 12037
    },
    {
      "epoch": 0.9110638096589565,
      "grad_norm": 0.71484375,
      "learning_rate": 1.1424434505122851e-05,
      "loss": 0.2863,
      "step": 12038
    },
    {
      "epoch": 0.9111394919824038,
      "grad_norm": 0.765625,
      "learning_rate": 1.1423255560516704e-05,
      "loss": 0.3239,
      "step": 12039
    },
    {
      "epoch": 0.9112151743058512,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1422076595719232e-05,
      "loss": 0.2921,
      "step": 12040
    },
    {
      "epoch": 0.9112908566292985,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1420897610747163e-05,
      "loss": 0.3266,
      "step": 12041
    },
    {
      "epoch": 0.9113665389527459,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1419718605617223e-05,
      "loss": 0.2995,
      "step": 12042
    },
    {
      "epoch": 0.9114422212761932,
      "grad_norm": 0.71484375,
      "learning_rate": 1.141853958034614e-05,
      "loss": 0.2909,
      "step": 12043
    },
    {
      "epoch": 0.9115179035996405,
      "grad_norm": 0.7109375,
      "learning_rate": 1.1417360534950635e-05,
      "loss": 0.2972,
      "step": 12044
    },
    {
      "epoch": 0.9115935859230878,
      "grad_norm": 0.8046875,
      "learning_rate": 1.1416181469447442e-05,
      "loss": 0.3827,
      "step": 12045
    },
    {
      "epoch": 0.9116692682465352,
      "grad_norm": 0.7265625,
      "learning_rate": 1.1415002383853282e-05,
      "loss": 0.2903,
      "step": 12046
    },
    {
      "epoch": 0.9117449505699825,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1413823278184885e-05,
      "loss": 0.3538,
      "step": 12047
    },
    {
      "epoch": 0.9118206328934298,
      "grad_norm": 0.73828125,
      "learning_rate": 1.141264415245898e-05,
      "loss": 0.2779,
      "step": 12048
    },
    {
      "epoch": 0.9118963152168772,
      "grad_norm": 0.734375,
      "learning_rate": 1.1411465006692292e-05,
      "loss": 0.2967,
      "step": 12049
    },
    {
      "epoch": 0.9119719975403245,
      "grad_norm": 0.7265625,
      "learning_rate": 1.1410285840901554e-05,
      "loss": 0.2925,
      "step": 12050
    },
    {
      "epoch": 0.9120476798637718,
      "grad_norm": 0.71484375,
      "learning_rate": 1.140910665510349e-05,
      "loss": 0.2832,
      "step": 12051
    },
    {
      "epoch": 0.9121233621872191,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1407927449314829e-05,
      "loss": 0.3465,
      "step": 12052
    },
    {
      "epoch": 0.9121990445106665,
      "grad_norm": 0.734375,
      "learning_rate": 1.14067482235523e-05,
      "loss": 0.2889,
      "step": 12053
    },
    {
      "epoch": 0.9122747268341138,
      "grad_norm": 0.71484375,
      "learning_rate": 1.1405568977832633e-05,
      "loss": 0.2914,
      "step": 12054
    },
    {
      "epoch": 0.9123504091575612,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1404389712172561e-05,
      "loss": 0.3028,
      "step": 12055
    },
    {
      "epoch": 0.9124260914810085,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1403210426588808e-05,
      "loss": 0.3099,
      "step": 12056
    },
    {
      "epoch": 0.9125017738044559,
      "grad_norm": 0.66796875,
      "learning_rate": 1.140203112109811e-05,
      "loss": 0.2388,
      "step": 12057
    },
    {
      "epoch": 0.9125774561279031,
      "grad_norm": 0.83203125,
      "learning_rate": 1.1400851795717193e-05,
      "loss": 0.3804,
      "step": 12058
    },
    {
      "epoch": 0.9126531384513504,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1399672450462791e-05,
      "loss": 0.3527,
      "step": 12059
    },
    {
      "epoch": 0.9127288207747978,
      "grad_norm": 0.8046875,
      "learning_rate": 1.1398493085351633e-05,
      "loss": 0.3614,
      "step": 12060
    },
    {
      "epoch": 0.9128045030982451,
      "grad_norm": 0.765625,
      "learning_rate": 1.1397313700400447e-05,
      "loss": 0.306,
      "step": 12061
    },
    {
      "epoch": 0.9128801854216925,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1396134295625971e-05,
      "loss": 0.336,
      "step": 12062
    },
    {
      "epoch": 0.9129558677451398,
      "grad_norm": 0.7109375,
      "learning_rate": 1.1394954871044935e-05,
      "loss": 0.3044,
      "step": 12063
    },
    {
      "epoch": 0.9130315500685872,
      "grad_norm": 0.73046875,
      "learning_rate": 1.1393775426674068e-05,
      "loss": 0.3007,
      "step": 12064
    },
    {
      "epoch": 0.9131072323920344,
      "grad_norm": 0.671875,
      "learning_rate": 1.1392595962530106e-05,
      "loss": 0.2425,
      "step": 12065
    },
    {
      "epoch": 0.9131829147154817,
      "grad_norm": 0.7265625,
      "learning_rate": 1.1391416478629781e-05,
      "loss": 0.3026,
      "step": 12066
    },
    {
      "epoch": 0.9132585970389291,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1390236974989824e-05,
      "loss": 0.3166,
      "step": 12067
    },
    {
      "epoch": 0.9133342793623764,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1389057451626971e-05,
      "loss": 0.2841,
      "step": 12068
    },
    {
      "epoch": 0.9134099616858238,
      "grad_norm": 0.765625,
      "learning_rate": 1.1387877908557956e-05,
      "loss": 0.2966,
      "step": 12069
    },
    {
      "epoch": 0.9134856440092711,
      "grad_norm": 0.7421875,
      "learning_rate": 1.1386698345799507e-05,
      "loss": 0.2943,
      "step": 12070
    },
    {
      "epoch": 0.9135613263327184,
      "grad_norm": 0.68359375,
      "learning_rate": 1.1385518763368366e-05,
      "loss": 0.251,
      "step": 12071
    },
    {
      "epoch": 0.9136370086561657,
      "grad_norm": 0.7421875,
      "learning_rate": 1.1384339161281264e-05,
      "loss": 0.2834,
      "step": 12072
    },
    {
      "epoch": 0.913712690979613,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1383159539554933e-05,
      "loss": 0.3191,
      "step": 12073
    },
    {
      "epoch": 0.9137883733030604,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1381979898206112e-05,
      "loss": 0.3259,
      "step": 12074
    },
    {
      "epoch": 0.9138640556265077,
      "grad_norm": 0.66796875,
      "learning_rate": 1.1380800237251533e-05,
      "loss": 0.2464,
      "step": 12075
    },
    {
      "epoch": 0.9139397379499551,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1379620556707934e-05,
      "loss": 0.3303,
      "step": 12076
    },
    {
      "epoch": 0.9140154202734024,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1378440856592048e-05,
      "loss": 0.334,
      "step": 12077
    },
    {
      "epoch": 0.9140911025968497,
      "grad_norm": 0.7734375,
      "learning_rate": 1.1377261136920618e-05,
      "loss": 0.3569,
      "step": 12078
    },
    {
      "epoch": 0.914166784920297,
      "grad_norm": 0.765625,
      "learning_rate": 1.137608139771037e-05,
      "loss": 0.3428,
      "step": 12079
    },
    {
      "epoch": 0.9142424672437444,
      "grad_norm": 0.7421875,
      "learning_rate": 1.1374901638978048e-05,
      "loss": 0.3108,
      "step": 12080
    },
    {
      "epoch": 0.9143181495671917,
      "grad_norm": 0.77734375,
      "learning_rate": 1.1373721860740388e-05,
      "loss": 0.3391,
      "step": 12081
    },
    {
      "epoch": 0.914393831890639,
      "grad_norm": 0.796875,
      "learning_rate": 1.1372542063014123e-05,
      "loss": 0.3711,
      "step": 12082
    },
    {
      "epoch": 0.9144695142140864,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1371362245815995e-05,
      "loss": 0.3024,
      "step": 12083
    },
    {
      "epoch": 0.9145451965375337,
      "grad_norm": 0.640625,
      "learning_rate": 1.1370182409162742e-05,
      "loss": 0.2202,
      "step": 12084
    },
    {
      "epoch": 0.914620878860981,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1369002553071098e-05,
      "loss": 0.3466,
      "step": 12085
    },
    {
      "epoch": 0.9146965611844283,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1367822677557804e-05,
      "loss": 0.2995,
      "step": 12086
    },
    {
      "epoch": 0.9147722435078757,
      "grad_norm": 0.74609375,
      "learning_rate": 1.13666427826396e-05,
      "loss": 0.2954,
      "step": 12087
    },
    {
      "epoch": 0.914847925831323,
      "grad_norm": 0.7890625,
      "learning_rate": 1.136546286833322e-05,
      "loss": 0.3356,
      "step": 12088
    },
    {
      "epoch": 0.9149236081547704,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1364282934655407e-05,
      "loss": 0.3407,
      "step": 12089
    },
    {
      "epoch": 0.9149992904782177,
      "grad_norm": 0.703125,
      "learning_rate": 1.13631029816229e-05,
      "loss": 0.2782,
      "step": 12090
    },
    {
      "epoch": 0.915074972801665,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1361923009252436e-05,
      "loss": 0.3138,
      "step": 12091
    },
    {
      "epoch": 0.9151506551251123,
      "grad_norm": 0.76953125,
      "learning_rate": 1.136074301756076e-05,
      "loss": 0.2988,
      "step": 12092
    },
    {
      "epoch": 0.9152263374485596,
      "grad_norm": 0.76953125,
      "learning_rate": 1.135956300656461e-05,
      "loss": 0.2754,
      "step": 12093
    },
    {
      "epoch": 0.915302019772007,
      "grad_norm": 0.79296875,
      "learning_rate": 1.1358382976280719e-05,
      "loss": 0.3527,
      "step": 12094
    },
    {
      "epoch": 0.9153777020954543,
      "grad_norm": 0.78125,
      "learning_rate": 1.1357202926725841e-05,
      "loss": 0.2975,
      "step": 12095
    },
    {
      "epoch": 0.9154533844189017,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1356022857916709e-05,
      "loss": 0.3229,
      "step": 12096
    },
    {
      "epoch": 0.915529066742349,
      "grad_norm": 0.734375,
      "learning_rate": 1.1354842769870065e-05,
      "loss": 0.285,
      "step": 12097
    },
    {
      "epoch": 0.9156047490657964,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1353662662602652e-05,
      "loss": 0.3056,
      "step": 12098
    },
    {
      "epoch": 0.9156804313892436,
      "grad_norm": 0.75,
      "learning_rate": 1.1352482536131209e-05,
      "loss": 0.3102,
      "step": 12099
    },
    {
      "epoch": 0.915756113712691,
      "grad_norm": 0.78515625,
      "learning_rate": 1.1351302390472481e-05,
      "loss": 0.3301,
      "step": 12100
    },
    {
      "epoch": 0.9158317960361383,
      "grad_norm": 0.78515625,
      "learning_rate": 1.1350122225643213e-05,
      "loss": 0.3321,
      "step": 12101
    },
    {
      "epoch": 0.9159074783595856,
      "grad_norm": 1.7265625,
      "learning_rate": 1.134894204166014e-05,
      "loss": 0.4015,
      "step": 12102
    },
    {
      "epoch": 0.915983160683033,
      "grad_norm": 0.75390625,
      "learning_rate": 1.134776183854001e-05,
      "loss": 0.315,
      "step": 12103
    },
    {
      "epoch": 0.9160588430064803,
      "grad_norm": 0.78515625,
      "learning_rate": 1.1346581616299565e-05,
      "loss": 0.3383,
      "step": 12104
    },
    {
      "epoch": 0.9161345253299277,
      "grad_norm": 0.7734375,
      "learning_rate": 1.1345401374955554e-05,
      "loss": 0.3249,
      "step": 12105
    },
    {
      "epoch": 0.9162102076533749,
      "grad_norm": 0.77734375,
      "learning_rate": 1.1344221114524712e-05,
      "loss": 0.3333,
      "step": 12106
    },
    {
      "epoch": 0.9162858899768223,
      "grad_norm": 0.91796875,
      "learning_rate": 1.1343040835023786e-05,
      "loss": 0.2849,
      "step": 12107
    },
    {
      "epoch": 0.9163615723002696,
      "grad_norm": 0.70703125,
      "learning_rate": 1.1341860536469523e-05,
      "loss": 0.2888,
      "step": 12108
    },
    {
      "epoch": 0.916437254623717,
      "grad_norm": 0.734375,
      "learning_rate": 1.1340680218878662e-05,
      "loss": 0.2746,
      "step": 12109
    },
    {
      "epoch": 0.9165129369471643,
      "grad_norm": 0.78515625,
      "learning_rate": 1.1339499882267955e-05,
      "loss": 0.3639,
      "step": 12110
    },
    {
      "epoch": 0.9165886192706116,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1338319526654143e-05,
      "loss": 0.3315,
      "step": 12111
    },
    {
      "epoch": 0.916664301594059,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1337139152053971e-05,
      "loss": 0.344,
      "step": 12112
    },
    {
      "epoch": 0.9167399839175062,
      "grad_norm": 0.75,
      "learning_rate": 1.1335958758484185e-05,
      "loss": 0.3191,
      "step": 12113
    },
    {
      "epoch": 0.9168156662409536,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1334778345961532e-05,
      "loss": 0.3161,
      "step": 12114
    },
    {
      "epoch": 0.9168913485644009,
      "grad_norm": 1.0078125,
      "learning_rate": 1.1333597914502758e-05,
      "loss": 0.3512,
      "step": 12115
    },
    {
      "epoch": 0.9169670308878483,
      "grad_norm": 0.80859375,
      "learning_rate": 1.133241746412461e-05,
      "loss": 0.3511,
      "step": 12116
    },
    {
      "epoch": 0.9170427132112956,
      "grad_norm": 0.82421875,
      "learning_rate": 1.1331236994843831e-05,
      "loss": 0.3412,
      "step": 12117
    },
    {
      "epoch": 0.917118395534743,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1330056506677174e-05,
      "loss": 0.3216,
      "step": 12118
    },
    {
      "epoch": 0.9171940778581903,
      "grad_norm": 0.78125,
      "learning_rate": 1.132887599964138e-05,
      "loss": 0.3487,
      "step": 12119
    },
    {
      "epoch": 0.9172697601816375,
      "grad_norm": 0.8046875,
      "learning_rate": 1.1327695473753201e-05,
      "loss": 0.3372,
      "step": 12120
    },
    {
      "epoch": 0.9173454425050849,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1326514929029385e-05,
      "loss": 0.2848,
      "step": 12121
    },
    {
      "epoch": 0.9174211248285322,
      "grad_norm": 0.7265625,
      "learning_rate": 1.1325334365486676e-05,
      "loss": 0.3333,
      "step": 12122
    },
    {
      "epoch": 0.9174968071519796,
      "grad_norm": 0.65234375,
      "learning_rate": 1.1324153783141825e-05,
      "loss": 0.2714,
      "step": 12123
    },
    {
      "epoch": 0.9175724894754269,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1322973182011579e-05,
      "loss": 0.309,
      "step": 12124
    },
    {
      "epoch": 0.9176481717988743,
      "grad_norm": 0.7734375,
      "learning_rate": 1.1321792562112692e-05,
      "loss": 0.3258,
      "step": 12125
    },
    {
      "epoch": 0.9177238541223216,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1320611923461906e-05,
      "loss": 0.3371,
      "step": 12126
    },
    {
      "epoch": 0.9177995364457688,
      "grad_norm": 0.79296875,
      "learning_rate": 1.1319431266075975e-05,
      "loss": 0.3405,
      "step": 12127
    },
    {
      "epoch": 0.9178752187692162,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1318250589971646e-05,
      "loss": 0.3406,
      "step": 12128
    },
    {
      "epoch": 0.9179509010926635,
      "grad_norm": 0.734375,
      "learning_rate": 1.1317069895165671e-05,
      "loss": 0.3295,
      "step": 12129
    },
    {
      "epoch": 0.9180265834161109,
      "grad_norm": 0.66796875,
      "learning_rate": 1.1315889181674799e-05,
      "loss": 0.2483,
      "step": 12130
    },
    {
      "epoch": 0.9181022657395582,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1314708449515783e-05,
      "loss": 0.2733,
      "step": 12131
    },
    {
      "epoch": 0.9181779480630056,
      "grad_norm": 0.7734375,
      "learning_rate": 1.1313527698705367e-05,
      "loss": 0.3395,
      "step": 12132
    },
    {
      "epoch": 0.9182536303864529,
      "grad_norm": 0.78125,
      "learning_rate": 1.131234692926031e-05,
      "loss": 0.3302,
      "step": 12133
    },
    {
      "epoch": 0.9183293127099001,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1311166141197361e-05,
      "loss": 0.3346,
      "step": 12134
    },
    {
      "epoch": 0.9184049950333475,
      "grad_norm": 0.7421875,
      "learning_rate": 1.1309985334533266e-05,
      "loss": 0.3232,
      "step": 12135
    },
    {
      "epoch": 0.9184806773567948,
      "grad_norm": 0.69921875,
      "learning_rate": 1.1308804509284784e-05,
      "loss": 0.249,
      "step": 12136
    },
    {
      "epoch": 0.9185563596802422,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1307623665468663e-05,
      "loss": 0.3157,
      "step": 12137
    },
    {
      "epoch": 0.9186320420036895,
      "grad_norm": 0.73046875,
      "learning_rate": 1.1306442803101655e-05,
      "loss": 0.3106,
      "step": 12138
    },
    {
      "epoch": 0.9187077243271369,
      "grad_norm": 0.828125,
      "learning_rate": 1.130526192220052e-05,
      "loss": 0.351,
      "step": 12139
    },
    {
      "epoch": 0.9187834066505842,
      "grad_norm": 0.75,
      "learning_rate": 1.1304081022781998e-05,
      "loss": 0.3099,
      "step": 12140
    },
    {
      "epoch": 0.9188590889740315,
      "grad_norm": 0.78515625,
      "learning_rate": 1.130290010486285e-05,
      "loss": 0.3377,
      "step": 12141
    },
    {
      "epoch": 0.9189347712974788,
      "grad_norm": 0.94140625,
      "learning_rate": 1.1301719168459828e-05,
      "loss": 0.2887,
      "step": 12142
    },
    {
      "epoch": 0.9190104536209261,
      "grad_norm": 0.703125,
      "learning_rate": 1.1300538213589692e-05,
      "loss": 0.2874,
      "step": 12143
    },
    {
      "epoch": 0.9190861359443735,
      "grad_norm": 0.7421875,
      "learning_rate": 1.1299357240269182e-05,
      "loss": 0.3217,
      "step": 12144
    },
    {
      "epoch": 0.9191618182678208,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1298176248515063e-05,
      "loss": 0.3407,
      "step": 12145
    },
    {
      "epoch": 0.9192375005912682,
      "grad_norm": 0.77734375,
      "learning_rate": 1.1296995238344084e-05,
      "loss": 0.3418,
      "step": 12146
    },
    {
      "epoch": 0.9193131829147155,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1295814209773001e-05,
      "loss": 0.3513,
      "step": 12147
    },
    {
      "epoch": 0.9193888652381628,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1294633162818576e-05,
      "loss": 0.3095,
      "step": 12148
    },
    {
      "epoch": 0.9194645475616101,
      "grad_norm": 0.828125,
      "learning_rate": 1.1293452097497554e-05,
      "loss": 0.3309,
      "step": 12149
    },
    {
      "epoch": 0.9195402298850575,
      "grad_norm": 1.03125,
      "learning_rate": 1.1292271013826692e-05,
      "loss": 0.3666,
      "step": 12150
    },
    {
      "epoch": 0.9196159122085048,
      "grad_norm": 0.7734375,
      "learning_rate": 1.129108991182275e-05,
      "loss": 0.3309,
      "step": 12151
    },
    {
      "epoch": 0.9196915945319522,
      "grad_norm": 0.78515625,
      "learning_rate": 1.1289908791502484e-05,
      "loss": 0.3364,
      "step": 12152
    },
    {
      "epoch": 0.9197672768553995,
      "grad_norm": 0.75,
      "learning_rate": 1.1288727652882648e-05,
      "loss": 0.3229,
      "step": 12153
    },
    {
      "epoch": 0.9198429591788468,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1287546495979997e-05,
      "loss": 0.2982,
      "step": 12154
    },
    {
      "epoch": 0.9199186415022941,
      "grad_norm": 0.71484375,
      "learning_rate": 1.128636532081129e-05,
      "loss": 0.296,
      "step": 12155
    },
    {
      "epoch": 0.9199943238257414,
      "grad_norm": 0.81640625,
      "learning_rate": 1.128518412739328e-05,
      "loss": 0.3298,
      "step": 12156
    },
    {
      "epoch": 0.9200700061491888,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1284002915742733e-05,
      "loss": 0.3263,
      "step": 12157
    },
    {
      "epoch": 0.9201456884726361,
      "grad_norm": 0.72265625,
      "learning_rate": 1.12828216858764e-05,
      "loss": 0.269,
      "step": 12158
    },
    {
      "epoch": 0.9202213707960835,
      "grad_norm": 0.71875,
      "learning_rate": 1.1281640437811039e-05,
      "loss": 0.2924,
      "step": 12159
    },
    {
      "epoch": 0.9202970531195308,
      "grad_norm": 0.6796875,
      "learning_rate": 1.1280459171563409e-05,
      "loss": 0.268,
      "step": 12160
    },
    {
      "epoch": 0.9203727354429782,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1279277887150268e-05,
      "loss": 0.3317,
      "step": 12161
    },
    {
      "epoch": 0.9204484177664254,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1278096584588376e-05,
      "loss": 0.2939,
      "step": 12162
    },
    {
      "epoch": 0.9205241000898727,
      "grad_norm": 0.76953125,
      "learning_rate": 1.127691526389449e-05,
      "loss": 0.341,
      "step": 12163
    },
    {
      "epoch": 0.9205997824133201,
      "grad_norm": 0.7890625,
      "learning_rate": 1.127573392508537e-05,
      "loss": 0.3376,
      "step": 12164
    },
    {
      "epoch": 0.9206754647367674,
      "grad_norm": 0.8359375,
      "learning_rate": 1.1274552568177772e-05,
      "loss": 0.3271,
      "step": 12165
    },
    {
      "epoch": 0.9207511470602148,
      "grad_norm": 0.77734375,
      "learning_rate": 1.1273371193188463e-05,
      "loss": 0.3069,
      "step": 12166
    },
    {
      "epoch": 0.9208268293836621,
      "grad_norm": 0.81640625,
      "learning_rate": 1.1272189800134197e-05,
      "loss": 0.354,
      "step": 12167
    },
    {
      "epoch": 0.9209025117071095,
      "grad_norm": 0.7734375,
      "learning_rate": 1.1271008389031734e-05,
      "loss": 0.3183,
      "step": 12168
    },
    {
      "epoch": 0.9209781940305567,
      "grad_norm": 0.7734375,
      "learning_rate": 1.1269826959897839e-05,
      "loss": 0.3251,
      "step": 12169
    },
    {
      "epoch": 0.921053876354004,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1268645512749267e-05,
      "loss": 0.2981,
      "step": 12170
    },
    {
      "epoch": 0.9211295586774514,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1267464047602782e-05,
      "loss": 0.3242,
      "step": 12171
    },
    {
      "epoch": 0.9212052410008987,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1266282564475146e-05,
      "loss": 0.28,
      "step": 12172
    },
    {
      "epoch": 0.9212809233243461,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1265101063383116e-05,
      "loss": 0.3355,
      "step": 12173
    },
    {
      "epoch": 0.9213566056477934,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1263919544343458e-05,
      "loss": 0.3149,
      "step": 12174
    },
    {
      "epoch": 0.9214322879712408,
      "grad_norm": 0.75,
      "learning_rate": 1.1262738007372931e-05,
      "loss": 0.3312,
      "step": 12175
    },
    {
      "epoch": 0.921507970294688,
      "grad_norm": 0.74609375,
      "learning_rate": 1.12615564524883e-05,
      "loss": 0.3235,
      "step": 12176
    },
    {
      "epoch": 0.9215836526181354,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1260374879706325e-05,
      "loss": 0.3395,
      "step": 12177
    },
    {
      "epoch": 0.9216593349415827,
      "grad_norm": 0.7578125,
      "learning_rate": 1.125919328904377e-05,
      "loss": 0.2909,
      "step": 12178
    },
    {
      "epoch": 0.92173501726503,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1258011680517396e-05,
      "loss": 0.3228,
      "step": 12179
    },
    {
      "epoch": 0.9218106995884774,
      "grad_norm": 0.79296875,
      "learning_rate": 1.125683005414397e-05,
      "loss": 0.3465,
      "step": 12180
    },
    {
      "epoch": 0.9218863819119247,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1255648409940254e-05,
      "loss": 0.3248,
      "step": 12181
    },
    {
      "epoch": 0.9219620642353721,
      "grad_norm": 0.77734375,
      "learning_rate": 1.1254466747923005e-05,
      "loss": 0.3449,
      "step": 12182
    },
    {
      "epoch": 0.9220377465588193,
      "grad_norm": 0.7421875,
      "learning_rate": 1.1253285068108996e-05,
      "loss": 0.3238,
      "step": 12183
    },
    {
      "epoch": 0.9221134288822667,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1252103370514986e-05,
      "loss": 0.3054,
      "step": 12184
    },
    {
      "epoch": 0.922189111205714,
      "grad_norm": 0.78125,
      "learning_rate": 1.1250921655157743e-05,
      "loss": 0.3622,
      "step": 12185
    },
    {
      "epoch": 0.9222647935291614,
      "grad_norm": 0.7265625,
      "learning_rate": 1.1249739922054029e-05,
      "loss": 0.3161,
      "step": 12186
    },
    {
      "epoch": 0.9223404758526087,
      "grad_norm": 1.0546875,
      "learning_rate": 1.124855817122061e-05,
      "loss": 0.3907,
      "step": 12187
    },
    {
      "epoch": 0.922416158176056,
      "grad_norm": 0.671875,
      "learning_rate": 1.1247376402674248e-05,
      "loss": 0.2559,
      "step": 12188
    },
    {
      "epoch": 0.9224918404995033,
      "grad_norm": 0.6875,
      "learning_rate": 1.1246194616431713e-05,
      "loss": 0.2654,
      "step": 12189
    },
    {
      "epoch": 0.9225675228229506,
      "grad_norm": 0.8046875,
      "learning_rate": 1.1245012812509773e-05,
      "loss": 0.3583,
      "step": 12190
    },
    {
      "epoch": 0.922643205146398,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1243830990925183e-05,
      "loss": 0.3197,
      "step": 12191
    },
    {
      "epoch": 0.9227188874698453,
      "grad_norm": 0.78125,
      "learning_rate": 1.1242649151694718e-05,
      "loss": 0.3355,
      "step": 12192
    },
    {
      "epoch": 0.9227945697932927,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1241467294835144e-05,
      "loss": 0.321,
      "step": 12193
    },
    {
      "epoch": 0.92287025211674,
      "grad_norm": 0.7734375,
      "learning_rate": 1.1240285420363221e-05,
      "loss": 0.3181,
      "step": 12194
    },
    {
      "epoch": 0.9229459344401874,
      "grad_norm": 0.90234375,
      "learning_rate": 1.123910352829573e-05,
      "loss": 0.3425,
      "step": 12195
    },
    {
      "epoch": 0.9230216167636346,
      "grad_norm": 1.015625,
      "learning_rate": 1.1237921618649423e-05,
      "loss": 0.3654,
      "step": 12196
    },
    {
      "epoch": 0.9230972990870819,
      "grad_norm": 0.765625,
      "learning_rate": 1.1236739691441073e-05,
      "loss": 0.335,
      "step": 12197
    },
    {
      "epoch": 0.9231729814105293,
      "grad_norm": 0.7109375,
      "learning_rate": 1.1235557746687448e-05,
      "loss": 0.283,
      "step": 12198
    },
    {
      "epoch": 0.9232486637339766,
      "grad_norm": 0.765625,
      "learning_rate": 1.1234375784405322e-05,
      "loss": 0.317,
      "step": 12199
    },
    {
      "epoch": 0.923324346057424,
      "grad_norm": 0.74609375,
      "learning_rate": 1.123319380461145e-05,
      "loss": 0.3326,
      "step": 12200
    },
    {
      "epoch": 0.9234000283808713,
      "grad_norm": 0.75,
      "learning_rate": 1.1232011807322611e-05,
      "loss": 0.3328,
      "step": 12201
    },
    {
      "epoch": 0.9234757107043187,
      "grad_norm": 0.70703125,
      "learning_rate": 1.1230829792555571e-05,
      "loss": 0.2753,
      "step": 12202
    },
    {
      "epoch": 0.9235513930277659,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1229647760327096e-05,
      "loss": 0.3079,
      "step": 12203
    },
    {
      "epoch": 0.9236270753512132,
      "grad_norm": 0.7109375,
      "learning_rate": 1.122846571065396e-05,
      "loss": 0.2961,
      "step": 12204
    },
    {
      "epoch": 0.9237027576746606,
      "grad_norm": 0.765625,
      "learning_rate": 1.1227283643552928e-05,
      "loss": 0.3439,
      "step": 12205
    },
    {
      "epoch": 0.9237784399981079,
      "grad_norm": 1.046875,
      "learning_rate": 1.122610155904077e-05,
      "loss": 0.3687,
      "step": 12206
    },
    {
      "epoch": 0.9238541223215553,
      "grad_norm": 0.69140625,
      "learning_rate": 1.122491945713426e-05,
      "loss": 0.2651,
      "step": 12207
    },
    {
      "epoch": 0.9239298046450026,
      "grad_norm": 0.6953125,
      "learning_rate": 1.1223737337850165e-05,
      "loss": 0.2674,
      "step": 12208
    },
    {
      "epoch": 0.92400548696845,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1222555201205253e-05,
      "loss": 0.2657,
      "step": 12209
    },
    {
      "epoch": 0.9240811692918972,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1221373047216302e-05,
      "loss": 0.3114,
      "step": 12210
    },
    {
      "epoch": 0.9241568516153446,
      "grad_norm": 0.7109375,
      "learning_rate": 1.1220190875900076e-05,
      "loss": 0.2774,
      "step": 12211
    },
    {
      "epoch": 0.9242325339387919,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1219008687273347e-05,
      "loss": 0.3158,
      "step": 12212
    },
    {
      "epoch": 0.9243082162622392,
      "grad_norm": 0.68359375,
      "learning_rate": 1.1217826481352892e-05,
      "loss": 0.2637,
      "step": 12213
    },
    {
      "epoch": 0.9243838985856866,
      "grad_norm": 0.70703125,
      "learning_rate": 1.1216644258155473e-05,
      "loss": 0.2947,
      "step": 12214
    },
    {
      "epoch": 0.9244595809091339,
      "grad_norm": 0.77734375,
      "learning_rate": 1.1215462017697871e-05,
      "loss": 0.328,
      "step": 12215
    },
    {
      "epoch": 0.9245352632325813,
      "grad_norm": 0.765625,
      "learning_rate": 1.1214279759996856e-05,
      "loss": 0.3417,
      "step": 12216
    },
    {
      "epoch": 0.9246109455560285,
      "grad_norm": 0.75,
      "learning_rate": 1.1213097485069195e-05,
      "loss": 0.3039,
      "step": 12217
    },
    {
      "epoch": 0.9246866278794759,
      "grad_norm": 0.7109375,
      "learning_rate": 1.1211915192931669e-05,
      "loss": 0.2925,
      "step": 12218
    },
    {
      "epoch": 0.9247623102029232,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1210732883601044e-05,
      "loss": 0.3174,
      "step": 12219
    },
    {
      "epoch": 0.9248379925263706,
      "grad_norm": 0.765625,
      "learning_rate": 1.1209550557094093e-05,
      "loss": 0.3406,
      "step": 12220
    },
    {
      "epoch": 0.9249136748498179,
      "grad_norm": 0.83203125,
      "learning_rate": 1.1208368213427591e-05,
      "loss": 0.3049,
      "step": 12221
    },
    {
      "epoch": 0.9249893571732652,
      "grad_norm": 0.71484375,
      "learning_rate": 1.1207185852618317e-05,
      "loss": 0.2741,
      "step": 12222
    },
    {
      "epoch": 0.9250650394967126,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1206003474683036e-05,
      "loss": 0.2905,
      "step": 12223
    },
    {
      "epoch": 0.9251407218201598,
      "grad_norm": 0.71875,
      "learning_rate": 1.1204821079638528e-05,
      "loss": 0.3016,
      "step": 12224
    },
    {
      "epoch": 0.9252164041436072,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1203638667501566e-05,
      "loss": 0.3092,
      "step": 12225
    },
    {
      "epoch": 0.9252920864670545,
      "grad_norm": 0.6796875,
      "learning_rate": 1.1202456238288923e-05,
      "loss": 0.2461,
      "step": 12226
    },
    {
      "epoch": 0.9253677687905019,
      "grad_norm": 0.7265625,
      "learning_rate": 1.1201273792017378e-05,
      "loss": 0.2805,
      "step": 12227
    },
    {
      "epoch": 0.9254434511139492,
      "grad_norm": 0.75390625,
      "learning_rate": 1.12000913287037e-05,
      "loss": 0.3073,
      "step": 12228
    },
    {
      "epoch": 0.9255191334373966,
      "grad_norm": 0.77734375,
      "learning_rate": 1.1198908848364665e-05,
      "loss": 0.3379,
      "step": 12229
    },
    {
      "epoch": 0.9255948157608439,
      "grad_norm": 0.80078125,
      "learning_rate": 1.1197726351017052e-05,
      "loss": 0.2983,
      "step": 12230
    },
    {
      "epoch": 0.9256704980842911,
      "grad_norm": 0.63671875,
      "learning_rate": 1.1196543836677635e-05,
      "loss": 0.2183,
      "step": 12231
    },
    {
      "epoch": 0.9257461804077385,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1195361305363193e-05,
      "loss": 0.2999,
      "step": 12232
    },
    {
      "epoch": 0.9258218627311858,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1194178757090496e-05,
      "loss": 0.305,
      "step": 12233
    },
    {
      "epoch": 0.9258975450546332,
      "grad_norm": 0.69921875,
      "learning_rate": 1.1192996191876329e-05,
      "loss": 0.2846,
      "step": 12234
    },
    {
      "epoch": 0.9259732273780805,
      "grad_norm": 0.78515625,
      "learning_rate": 1.1191813609737459e-05,
      "loss": 0.3287,
      "step": 12235
    },
    {
      "epoch": 0.9260489097015279,
      "grad_norm": 0.69921875,
      "learning_rate": 1.119063101069067e-05,
      "loss": 0.2807,
      "step": 12236
    },
    {
      "epoch": 0.9261245920249752,
      "grad_norm": 0.94921875,
      "learning_rate": 1.1189448394752737e-05,
      "loss": 0.3308,
      "step": 12237
    },
    {
      "epoch": 0.9262002743484224,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1188265761940436e-05,
      "loss": 0.2924,
      "step": 12238
    },
    {
      "epoch": 0.9262759566718698,
      "grad_norm": 0.80078125,
      "learning_rate": 1.1187083112270546e-05,
      "loss": 0.3567,
      "step": 12239
    },
    {
      "epoch": 0.9263516389953171,
      "grad_norm": 0.734375,
      "learning_rate": 1.1185900445759846e-05,
      "loss": 0.272,
      "step": 12240
    },
    {
      "epoch": 0.9264273213187645,
      "grad_norm": 0.7734375,
      "learning_rate": 1.1184717762425113e-05,
      "loss": 0.3151,
      "step": 12241
    },
    {
      "epoch": 0.9265030036422118,
      "grad_norm": 1.015625,
      "learning_rate": 1.1183535062283127e-05,
      "loss": 0.3467,
      "step": 12242
    },
    {
      "epoch": 0.9265786859656592,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1182352345350662e-05,
      "loss": 0.3151,
      "step": 12243
    },
    {
      "epoch": 0.9266543682891065,
      "grad_norm": 0.76171875,
      "learning_rate": 1.11811696116445e-05,
      "loss": 0.3031,
      "step": 12244
    },
    {
      "epoch": 0.9267300506125538,
      "grad_norm": 0.84375,
      "learning_rate": 1.1179986861181423e-05,
      "loss": 0.3346,
      "step": 12245
    },
    {
      "epoch": 0.9268057329360011,
      "grad_norm": 0.78125,
      "learning_rate": 1.1178804093978205e-05,
      "loss": 0.3371,
      "step": 12246
    },
    {
      "epoch": 0.9268814152594484,
      "grad_norm": 0.796875,
      "learning_rate": 1.117762131005163e-05,
      "loss": 0.3349,
      "step": 12247
    },
    {
      "epoch": 0.9269570975828958,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1176438509418475e-05,
      "loss": 0.329,
      "step": 12248
    },
    {
      "epoch": 0.9270327799063431,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1175255692095518e-05,
      "loss": 0.3051,
      "step": 12249
    },
    {
      "epoch": 0.9271084622297905,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1174072858099545e-05,
      "loss": 0.2868,
      "step": 12250
    },
    {
      "epoch": 0.9271841445532378,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1172890007447336e-05,
      "loss": 0.2829,
      "step": 12251
    },
    {
      "epoch": 0.9272598268766851,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1171707140155668e-05,
      "loss": 0.3327,
      "step": 12252
    },
    {
      "epoch": 0.9273355092001324,
      "grad_norm": 0.7109375,
      "learning_rate": 1.117052425624132e-05,
      "loss": 0.2862,
      "step": 12253
    },
    {
      "epoch": 0.9274111915235798,
      "grad_norm": 0.78125,
      "learning_rate": 1.116934135572108e-05,
      "loss": 0.3097,
      "step": 12254
    },
    {
      "epoch": 0.9274868738470271,
      "grad_norm": 0.765625,
      "learning_rate": 1.1168158438611727e-05,
      "loss": 0.3461,
      "step": 12255
    },
    {
      "epoch": 0.9275625561704745,
      "grad_norm": 0.80078125,
      "learning_rate": 1.116697550493004e-05,
      "loss": 0.3446,
      "step": 12256
    },
    {
      "epoch": 0.9276382384939218,
      "grad_norm": 0.78515625,
      "learning_rate": 1.1165792554692802e-05,
      "loss": 0.3471,
      "step": 12257
    },
    {
      "epoch": 0.9277139208173691,
      "grad_norm": 0.7421875,
      "learning_rate": 1.11646095879168e-05,
      "loss": 0.2857,
      "step": 12258
    },
    {
      "epoch": 0.9277896031408164,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1163426604618808e-05,
      "loss": 0.3278,
      "step": 12259
    },
    {
      "epoch": 0.9278652854642637,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1162243604815616e-05,
      "loss": 0.319,
      "step": 12260
    },
    {
      "epoch": 0.9279409677877111,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1161060588524004e-05,
      "loss": 0.3229,
      "step": 12261
    },
    {
      "epoch": 0.9280166501111584,
      "grad_norm": 0.78125,
      "learning_rate": 1.1159877555760751e-05,
      "loss": 0.3325,
      "step": 12262
    },
    {
      "epoch": 0.9280923324346058,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1158694506542649e-05,
      "loss": 0.2965,
      "step": 12263
    },
    {
      "epoch": 0.9281680147580531,
      "grad_norm": 0.79296875,
      "learning_rate": 1.1157511440886475e-05,
      "loss": 0.3254,
      "step": 12264
    },
    {
      "epoch": 0.9282436970815005,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1156328358809016e-05,
      "loss": 0.3083,
      "step": 12265
    },
    {
      "epoch": 0.9283193794049477,
      "grad_norm": 0.9453125,
      "learning_rate": 1.1155145260327054e-05,
      "loss": 0.2921,
      "step": 12266
    },
    {
      "epoch": 0.928395061728395,
      "grad_norm": 0.828125,
      "learning_rate": 1.1153962145457376e-05,
      "loss": 0.3684,
      "step": 12267
    },
    {
      "epoch": 0.9284707440518424,
      "grad_norm": 0.7734375,
      "learning_rate": 1.1152779014216762e-05,
      "loss": 0.3141,
      "step": 12268
    },
    {
      "epoch": 0.9285464263752897,
      "grad_norm": 0.734375,
      "learning_rate": 1.1151595866622002e-05,
      "loss": 0.3146,
      "step": 12269
    },
    {
      "epoch": 0.9286221086987371,
      "grad_norm": 0.71875,
      "learning_rate": 1.1150412702689875e-05,
      "loss": 0.2816,
      "step": 12270
    },
    {
      "epoch": 0.9286977910221844,
      "grad_norm": 0.765625,
      "learning_rate": 1.1149229522437173e-05,
      "loss": 0.326,
      "step": 12271
    },
    {
      "epoch": 0.9287734733456318,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1148046325880675e-05,
      "loss": 0.3016,
      "step": 12272
    },
    {
      "epoch": 0.928849155669079,
      "grad_norm": 0.734375,
      "learning_rate": 1.1146863113037172e-05,
      "loss": 0.3258,
      "step": 12273
    },
    {
      "epoch": 0.9289248379925263,
      "grad_norm": 0.7265625,
      "learning_rate": 1.1145679883923448e-05,
      "loss": 0.2705,
      "step": 12274
    },
    {
      "epoch": 0.9290005203159737,
      "grad_norm": 0.734375,
      "learning_rate": 1.1144496638556288e-05,
      "loss": 0.311,
      "step": 12275
    },
    {
      "epoch": 0.929076202639421,
      "grad_norm": 0.75390625,
      "learning_rate": 1.114331337695248e-05,
      "loss": 0.3276,
      "step": 12276
    },
    {
      "epoch": 0.9291518849628684,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1142130099128808e-05,
      "loss": 0.3243,
      "step": 12277
    },
    {
      "epoch": 0.9292275672863157,
      "grad_norm": 0.8984375,
      "learning_rate": 1.114094680510206e-05,
      "loss": 0.2966,
      "step": 12278
    },
    {
      "epoch": 0.9293032496097631,
      "grad_norm": 0.69921875,
      "learning_rate": 1.1139763494889027e-05,
      "loss": 0.2746,
      "step": 12279
    },
    {
      "epoch": 0.9293789319332103,
      "grad_norm": 0.78125,
      "learning_rate": 1.113858016850649e-05,
      "loss": 0.3216,
      "step": 12280
    },
    {
      "epoch": 0.9294546142566577,
      "grad_norm": 0.73046875,
      "learning_rate": 1.1137396825971241e-05,
      "loss": 0.2968,
      "step": 12281
    },
    {
      "epoch": 0.929530296580105,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1136213467300065e-05,
      "loss": 0.2762,
      "step": 12282
    },
    {
      "epoch": 0.9296059789035523,
      "grad_norm": 0.765625,
      "learning_rate": 1.1135030092509753e-05,
      "loss": 0.307,
      "step": 12283
    },
    {
      "epoch": 0.9296816612269997,
      "grad_norm": 0.77734375,
      "learning_rate": 1.113384670161709e-05,
      "loss": 0.3369,
      "step": 12284
    },
    {
      "epoch": 0.929757343550447,
      "grad_norm": 0.8359375,
      "learning_rate": 1.1132663294638864e-05,
      "loss": 0.3074,
      "step": 12285
    },
    {
      "epoch": 0.9298330258738944,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1131479871591869e-05,
      "loss": 0.3059,
      "step": 12286
    },
    {
      "epoch": 0.9299087081973416,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1130296432492888e-05,
      "loss": 0.3262,
      "step": 12287
    },
    {
      "epoch": 0.929984390520789,
      "grad_norm": 0.71484375,
      "learning_rate": 1.1129112977358713e-05,
      "loss": 0.2614,
      "step": 12288
    },
    {
      "epoch": 0.9300600728442363,
      "grad_norm": 0.765625,
      "learning_rate": 1.1127929506206134e-05,
      "loss": 0.2901,
      "step": 12289
    },
    {
      "epoch": 0.9301357551676837,
      "grad_norm": 0.78515625,
      "learning_rate": 1.112674601905194e-05,
      "loss": 0.3321,
      "step": 12290
    },
    {
      "epoch": 0.930211437491131,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1125562515912915e-05,
      "loss": 0.3307,
      "step": 12291
    },
    {
      "epoch": 0.9302871198145783,
      "grad_norm": 0.82421875,
      "learning_rate": 1.1124378996805858e-05,
      "loss": 0.3516,
      "step": 12292
    },
    {
      "epoch": 0.9303628021380257,
      "grad_norm": 0.77734375,
      "learning_rate": 1.1123195461747558e-05,
      "loss": 0.3147,
      "step": 12293
    },
    {
      "epoch": 0.9304384844614729,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1122011910754797e-05,
      "loss": 0.2973,
      "step": 12294
    },
    {
      "epoch": 0.9305141667849203,
      "grad_norm": 0.65234375,
      "learning_rate": 1.1120828343844377e-05,
      "loss": 0.2612,
      "step": 12295
    },
    {
      "epoch": 0.9305898491083676,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1119644761033079e-05,
      "loss": 0.3467,
      "step": 12296
    },
    {
      "epoch": 0.930665531431815,
      "grad_norm": 1.0078125,
      "learning_rate": 1.11184611623377e-05,
      "loss": 0.347,
      "step": 12297
    },
    {
      "epoch": 0.9307412137552623,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1117277547775035e-05,
      "loss": 0.333,
      "step": 12298
    },
    {
      "epoch": 0.9308168960787097,
      "grad_norm": 0.78515625,
      "learning_rate": 1.1116093917361865e-05,
      "loss": 0.3111,
      "step": 12299
    },
    {
      "epoch": 0.930892578402157,
      "grad_norm": 0.70703125,
      "learning_rate": 1.1114910271114987e-05,
      "loss": 0.2776,
      "step": 12300
    },
    {
      "epoch": 0.9309682607256042,
      "grad_norm": 0.73046875,
      "learning_rate": 1.1113726609051197e-05,
      "loss": 0.3169,
      "step": 12301
    },
    {
      "epoch": 0.9310439430490516,
      "grad_norm": 0.79296875,
      "learning_rate": 1.1112542931187284e-05,
      "loss": 0.3208,
      "step": 12302
    },
    {
      "epoch": 0.9311196253724989,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1111359237540039e-05,
      "loss": 0.2713,
      "step": 12303
    },
    {
      "epoch": 0.9311953076959463,
      "grad_norm": 0.78125,
      "learning_rate": 1.1110175528126255e-05,
      "loss": 0.3105,
      "step": 12304
    },
    {
      "epoch": 0.9312709900193936,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1108991802962724e-05,
      "loss": 0.3085,
      "step": 12305
    },
    {
      "epoch": 0.931346672342841,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1107808062066245e-05,
      "loss": 0.2964,
      "step": 12306
    },
    {
      "epoch": 0.9314223546662883,
      "grad_norm": 0.7109375,
      "learning_rate": 1.1106624305453608e-05,
      "loss": 0.2895,
      "step": 12307
    },
    {
      "epoch": 0.9314980369897355,
      "grad_norm": 0.7265625,
      "learning_rate": 1.1105440533141605e-05,
      "loss": 0.2866,
      "step": 12308
    },
    {
      "epoch": 0.9315737193131829,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1104256745147031e-05,
      "loss": 0.2799,
      "step": 12309
    },
    {
      "epoch": 0.9316494016366302,
      "grad_norm": 0.80078125,
      "learning_rate": 1.110307294148668e-05,
      "loss": 0.3637,
      "step": 12310
    },
    {
      "epoch": 0.9317250839600776,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1101889122177345e-05,
      "loss": 0.2844,
      "step": 12311
    },
    {
      "epoch": 0.9318007662835249,
      "grad_norm": 0.734375,
      "learning_rate": 1.1100705287235824e-05,
      "loss": 0.2953,
      "step": 12312
    },
    {
      "epoch": 0.9318764486069723,
      "grad_norm": 0.765625,
      "learning_rate": 1.1099521436678908e-05,
      "loss": 0.3299,
      "step": 12313
    },
    {
      "epoch": 0.9319521309304195,
      "grad_norm": 0.796875,
      "learning_rate": 1.1098337570523397e-05,
      "loss": 0.3633,
      "step": 12314
    },
    {
      "epoch": 0.9320278132538669,
      "grad_norm": 0.75,
      "learning_rate": 1.1097153688786077e-05,
      "loss": 0.323,
      "step": 12315
    },
    {
      "epoch": 0.9321034955773142,
      "grad_norm": 0.7421875,
      "learning_rate": 1.1095969791483756e-05,
      "loss": 0.3271,
      "step": 12316
    },
    {
      "epoch": 0.9321791779007615,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1094785878633219e-05,
      "loss": 0.3063,
      "step": 12317
    },
    {
      "epoch": 0.9322548602242089,
      "grad_norm": 0.78515625,
      "learning_rate": 1.1093601950251262e-05,
      "loss": 0.3507,
      "step": 12318
    },
    {
      "epoch": 0.9323305425476562,
      "grad_norm": 0.765625,
      "learning_rate": 1.109241800635469e-05,
      "loss": 0.3295,
      "step": 12319
    },
    {
      "epoch": 0.9324062248711036,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1091234046960294e-05,
      "loss": 0.2953,
      "step": 12320
    },
    {
      "epoch": 0.9324819071945508,
      "grad_norm": 0.78125,
      "learning_rate": 1.1090050072084868e-05,
      "loss": 0.3307,
      "step": 12321
    },
    {
      "epoch": 0.9325575895179982,
      "grad_norm": 0.69140625,
      "learning_rate": 1.108886608174521e-05,
      "loss": 0.2655,
      "step": 12322
    },
    {
      "epoch": 0.9326332718414455,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1087682075958121e-05,
      "loss": 0.3266,
      "step": 12323
    },
    {
      "epoch": 0.9327089541648929,
      "grad_norm": 0.734375,
      "learning_rate": 1.1086498054740393e-05,
      "loss": 0.3067,
      "step": 12324
    },
    {
      "epoch": 0.9327846364883402,
      "grad_norm": 0.953125,
      "learning_rate": 1.1085314018108826e-05,
      "loss": 0.3323,
      "step": 12325
    },
    {
      "epoch": 0.9328603188117875,
      "grad_norm": 0.765625,
      "learning_rate": 1.108412996608022e-05,
      "loss": 0.3088,
      "step": 12326
    },
    {
      "epoch": 0.9329360011352349,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1082945898671367e-05,
      "loss": 0.316,
      "step": 12327
    },
    {
      "epoch": 0.9330116834586821,
      "grad_norm": 0.703125,
      "learning_rate": 1.108176181589907e-05,
      "loss": 0.2805,
      "step": 12328
    },
    {
      "epoch": 0.9330873657821295,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1080577717780123e-05,
      "loss": 0.3047,
      "step": 12329
    },
    {
      "epoch": 0.9331630481055768,
      "grad_norm": 0.7890625,
      "learning_rate": 1.107939360433133e-05,
      "loss": 0.3704,
      "step": 12330
    },
    {
      "epoch": 0.9332387304290242,
      "grad_norm": 0.6953125,
      "learning_rate": 1.1078209475569484e-05,
      "loss": 0.2651,
      "step": 12331
    },
    {
      "epoch": 0.9333144127524715,
      "grad_norm": 0.7734375,
      "learning_rate": 1.1077025331511385e-05,
      "loss": 0.3126,
      "step": 12332
    },
    {
      "epoch": 0.9333900950759189,
      "grad_norm": 0.85546875,
      "learning_rate": 1.1075841172173836e-05,
      "loss": 0.3148,
      "step": 12333
    },
    {
      "epoch": 0.9334657773993662,
      "grad_norm": 0.7265625,
      "learning_rate": 1.1074656997573632e-05,
      "loss": 0.2878,
      "step": 12334
    },
    {
      "epoch": 0.9335414597228134,
      "grad_norm": 0.75,
      "learning_rate": 1.1073472807727574e-05,
      "loss": 0.3182,
      "step": 12335
    },
    {
      "epoch": 0.9336171420462608,
      "grad_norm": 0.734375,
      "learning_rate": 1.1072288602652466e-05,
      "loss": 0.3137,
      "step": 12336
    },
    {
      "epoch": 0.9336928243697081,
      "grad_norm": 0.6953125,
      "learning_rate": 1.1071104382365101e-05,
      "loss": 0.2827,
      "step": 12337
    },
    {
      "epoch": 0.9337685066931555,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1069920146882283e-05,
      "loss": 0.3338,
      "step": 12338
    },
    {
      "epoch": 0.9338441890166028,
      "grad_norm": 0.71484375,
      "learning_rate": 1.1068735896220808e-05,
      "loss": 0.2819,
      "step": 12339
    },
    {
      "epoch": 0.9339198713400502,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1067551630397486e-05,
      "loss": 0.2992,
      "step": 12340
    },
    {
      "epoch": 0.9339955536634975,
      "grad_norm": 0.7734375,
      "learning_rate": 1.1066367349429108e-05,
      "loss": 0.298,
      "step": 12341
    },
    {
      "epoch": 0.9340712359869447,
      "grad_norm": 0.66796875,
      "learning_rate": 1.1065183053332481e-05,
      "loss": 0.2551,
      "step": 12342
    },
    {
      "epoch": 0.9341469183103921,
      "grad_norm": 0.765625,
      "learning_rate": 1.1063998742124404e-05,
      "loss": 0.2994,
      "step": 12343
    },
    {
      "epoch": 0.9342226006338394,
      "grad_norm": 0.71875,
      "learning_rate": 1.1062814415821681e-05,
      "loss": 0.2909,
      "step": 12344
    },
    {
      "epoch": 0.9342982829572868,
      "grad_norm": 0.73046875,
      "learning_rate": 1.1061630074441109e-05,
      "loss": 0.2863,
      "step": 12345
    },
    {
      "epoch": 0.9343739652807341,
      "grad_norm": 0.65234375,
      "learning_rate": 1.1060445717999495e-05,
      "loss": 0.2455,
      "step": 12346
    },
    {
      "epoch": 0.9344496476041815,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1059261346513636e-05,
      "loss": 0.3126,
      "step": 12347
    },
    {
      "epoch": 0.9345253299276288,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1058076960000338e-05,
      "loss": 0.3152,
      "step": 12348
    },
    {
      "epoch": 0.9346010122510761,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1056892558476407e-05,
      "loss": 0.3124,
      "step": 12349
    },
    {
      "epoch": 0.9346766945745234,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1055708141958634e-05,
      "loss": 0.2994,
      "step": 12350
    },
    {
      "epoch": 0.9347523768979707,
      "grad_norm": 0.7421875,
      "learning_rate": 1.1054523710463833e-05,
      "loss": 0.3018,
      "step": 12351
    },
    {
      "epoch": 0.9348280592214181,
      "grad_norm": 0.796875,
      "learning_rate": 1.1053339264008804e-05,
      "loss": 0.2684,
      "step": 12352
    },
    {
      "epoch": 0.9349037415448654,
      "grad_norm": 0.73828125,
      "learning_rate": 1.1052154802610347e-05,
      "loss": 0.3151,
      "step": 12353
    },
    {
      "epoch": 0.9349794238683128,
      "grad_norm": 0.81640625,
      "learning_rate": 1.1050970326285273e-05,
      "loss": 0.3357,
      "step": 12354
    },
    {
      "epoch": 0.9350551061917601,
      "grad_norm": 0.73046875,
      "learning_rate": 1.104978583505038e-05,
      "loss": 0.304,
      "step": 12355
    },
    {
      "epoch": 0.9351307885152074,
      "grad_norm": 0.81640625,
      "learning_rate": 1.104860132892247e-05,
      "loss": 0.3681,
      "step": 12356
    },
    {
      "epoch": 0.9352064708386547,
      "grad_norm": 0.68359375,
      "learning_rate": 1.1047416807918354e-05,
      "loss": 0.2619,
      "step": 12357
    },
    {
      "epoch": 0.9352821531621021,
      "grad_norm": 0.875,
      "learning_rate": 1.1046232272054833e-05,
      "loss": 0.3459,
      "step": 12358
    },
    {
      "epoch": 0.9353578354855494,
      "grad_norm": 0.78125,
      "learning_rate": 1.104504772134871e-05,
      "loss": 0.3345,
      "step": 12359
    },
    {
      "epoch": 0.9354335178089968,
      "grad_norm": 0.74609375,
      "learning_rate": 1.1043863155816794e-05,
      "loss": 0.3164,
      "step": 12360
    },
    {
      "epoch": 0.9355092001324441,
      "grad_norm": 0.76171875,
      "learning_rate": 1.1042678575475887e-05,
      "loss": 0.3189,
      "step": 12361
    },
    {
      "epoch": 0.9355848824558914,
      "grad_norm": 0.8046875,
      "learning_rate": 1.1041493980342792e-05,
      "loss": 0.33,
      "step": 12362
    },
    {
      "epoch": 0.9356605647793387,
      "grad_norm": 0.6875,
      "learning_rate": 1.1040309370434323e-05,
      "loss": 0.2872,
      "step": 12363
    },
    {
      "epoch": 0.935736247102786,
      "grad_norm": 0.75,
      "learning_rate": 1.1039124745767278e-05,
      "loss": 0.3146,
      "step": 12364
    },
    {
      "epoch": 0.9358119294262334,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1037940106358462e-05,
      "loss": 0.3343,
      "step": 12365
    },
    {
      "epoch": 0.9358876117496807,
      "grad_norm": 0.7734375,
      "learning_rate": 1.103675545222469e-05,
      "loss": 0.3076,
      "step": 12366
    },
    {
      "epoch": 0.9359632940731281,
      "grad_norm": 0.7578125,
      "learning_rate": 1.103557078338276e-05,
      "loss": 0.3119,
      "step": 12367
    },
    {
      "epoch": 0.9360389763965754,
      "grad_norm": 0.7265625,
      "learning_rate": 1.1034386099849481e-05,
      "loss": 0.276,
      "step": 12368
    },
    {
      "epoch": 0.9361146587200228,
      "grad_norm": 0.6953125,
      "learning_rate": 1.1033201401641662e-05,
      "loss": 0.2742,
      "step": 12369
    },
    {
      "epoch": 0.93619034104347,
      "grad_norm": 0.75,
      "learning_rate": 1.1032016688776106e-05,
      "loss": 0.3144,
      "step": 12370
    },
    {
      "epoch": 0.9362660233669173,
      "grad_norm": 0.75390625,
      "learning_rate": 1.1030831961269622e-05,
      "loss": 0.2998,
      "step": 12371
    },
    {
      "epoch": 0.9363417056903647,
      "grad_norm": 0.79296875,
      "learning_rate": 1.1029647219139022e-05,
      "loss": 0.3437,
      "step": 12372
    },
    {
      "epoch": 0.936417388013812,
      "grad_norm": 0.7578125,
      "learning_rate": 1.1028462462401106e-05,
      "loss": 0.3161,
      "step": 12373
    },
    {
      "epoch": 0.9364930703372594,
      "grad_norm": 0.82421875,
      "learning_rate": 1.1027277691072681e-05,
      "loss": 0.3614,
      "step": 12374
    },
    {
      "epoch": 0.9365687526607067,
      "grad_norm": 0.73046875,
      "learning_rate": 1.1026092905170564e-05,
      "loss": 0.2906,
      "step": 12375
    },
    {
      "epoch": 0.9366444349841541,
      "grad_norm": 0.69921875,
      "learning_rate": 1.1024908104711558e-05,
      "loss": 0.2766,
      "step": 12376
    },
    {
      "epoch": 0.9367201173076013,
      "grad_norm": 0.75,
      "learning_rate": 1.1023723289712472e-05,
      "loss": 0.3212,
      "step": 12377
    },
    {
      "epoch": 0.9367957996310486,
      "grad_norm": 1.03125,
      "learning_rate": 1.1022538460190112e-05,
      "loss": 0.3173,
      "step": 12378
    },
    {
      "epoch": 0.936871481954496,
      "grad_norm": 0.76953125,
      "learning_rate": 1.1021353616161292e-05,
      "loss": 0.3322,
      "step": 12379
    },
    {
      "epoch": 0.9369471642779433,
      "grad_norm": 0.75,
      "learning_rate": 1.1020168757642815e-05,
      "loss": 0.3209,
      "step": 12380
    },
    {
      "epoch": 0.9370228466013907,
      "grad_norm": 0.82421875,
      "learning_rate": 1.1018983884651497e-05,
      "loss": 0.3782,
      "step": 12381
    },
    {
      "epoch": 0.937098528924838,
      "grad_norm": 0.7265625,
      "learning_rate": 1.1017798997204143e-05,
      "loss": 0.2978,
      "step": 12382
    },
    {
      "epoch": 0.9371742112482854,
      "grad_norm": 0.78515625,
      "learning_rate": 1.1016614095317562e-05,
      "loss": 0.3308,
      "step": 12383
    },
    {
      "epoch": 0.9372498935717326,
      "grad_norm": 0.6796875,
      "learning_rate": 1.1015429179008567e-05,
      "loss": 0.2442,
      "step": 12384
    },
    {
      "epoch": 0.93732557589518,
      "grad_norm": 0.6953125,
      "learning_rate": 1.1014244248293965e-05,
      "loss": 0.2804,
      "step": 12385
    },
    {
      "epoch": 0.9374012582186273,
      "grad_norm": 0.72265625,
      "learning_rate": 1.1013059303190572e-05,
      "loss": 0.2929,
      "step": 12386
    },
    {
      "epoch": 0.9374769405420746,
      "grad_norm": 0.70703125,
      "learning_rate": 1.101187434371519e-05,
      "loss": 0.2789,
      "step": 12387
    },
    {
      "epoch": 0.937552622865522,
      "grad_norm": 0.97265625,
      "learning_rate": 1.1010689369884632e-05,
      "loss": 0.3916,
      "step": 12388
    },
    {
      "epoch": 0.9376283051889693,
      "grad_norm": 0.7734375,
      "learning_rate": 1.1009504381715717e-05,
      "loss": 0.3146,
      "step": 12389
    },
    {
      "epoch": 0.9377039875124167,
      "grad_norm": 0.7890625,
      "learning_rate": 1.1008319379225247e-05,
      "loss": 0.3477,
      "step": 12390
    },
    {
      "epoch": 0.9377796698358639,
      "grad_norm": 0.77734375,
      "learning_rate": 1.1007134362430036e-05,
      "loss": 0.3365,
      "step": 12391
    },
    {
      "epoch": 0.9378553521593113,
      "grad_norm": 0.70703125,
      "learning_rate": 1.1005949331346897e-05,
      "loss": 0.2553,
      "step": 12392
    },
    {
      "epoch": 0.9379310344827586,
      "grad_norm": 0.83984375,
      "learning_rate": 1.100476428599264e-05,
      "loss": 0.2847,
      "step": 12393
    },
    {
      "epoch": 0.938006716806206,
      "grad_norm": 0.75,
      "learning_rate": 1.1003579226384078e-05,
      "loss": 0.3139,
      "step": 12394
    },
    {
      "epoch": 0.9380823991296533,
      "grad_norm": 0.8125,
      "learning_rate": 1.1002394152538018e-05,
      "loss": 0.3564,
      "step": 12395
    },
    {
      "epoch": 0.9381580814531006,
      "grad_norm": 0.86328125,
      "learning_rate": 1.1001209064471283e-05,
      "loss": 0.3195,
      "step": 12396
    },
    {
      "epoch": 0.938233763776548,
      "grad_norm": 0.78125,
      "learning_rate": 1.1000023962200674e-05,
      "loss": 0.3475,
      "step": 12397
    },
    {
      "epoch": 0.9383094460999952,
      "grad_norm": 0.80859375,
      "learning_rate": 1.0998838845743012e-05,
      "loss": 0.3562,
      "step": 12398
    },
    {
      "epoch": 0.9383851284234426,
      "grad_norm": 0.734375,
      "learning_rate": 1.0997653715115107e-05,
      "loss": 0.3073,
      "step": 12399
    },
    {
      "epoch": 0.9384608107468899,
      "grad_norm": 0.7109375,
      "learning_rate": 1.0996468570333771e-05,
      "loss": 0.2856,
      "step": 12400
    },
    {
      "epoch": 0.9385364930703373,
      "grad_norm": 0.77734375,
      "learning_rate": 1.099528341141582e-05,
      "loss": 0.3618,
      "step": 12401
    },
    {
      "epoch": 0.9386121753937846,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0994098238378066e-05,
      "loss": 0.3002,
      "step": 12402
    },
    {
      "epoch": 0.938687857717232,
      "grad_norm": 0.7265625,
      "learning_rate": 1.099291305123732e-05,
      "loss": 0.2461,
      "step": 12403
    },
    {
      "epoch": 0.9387635400406793,
      "grad_norm": 0.6953125,
      "learning_rate": 1.09917278500104e-05,
      "loss": 0.2727,
      "step": 12404
    },
    {
      "epoch": 0.9388392223641265,
      "grad_norm": 0.75,
      "learning_rate": 1.099054263471412e-05,
      "loss": 0.308,
      "step": 12405
    },
    {
      "epoch": 0.9389149046875739,
      "grad_norm": 0.7578125,
      "learning_rate": 1.0989357405365289e-05,
      "loss": 0.3099,
      "step": 12406
    },
    {
      "epoch": 0.9389905870110212,
      "grad_norm": 0.6953125,
      "learning_rate": 1.0988172161980727e-05,
      "loss": 0.2793,
      "step": 12407
    },
    {
      "epoch": 0.9390662693344686,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0986986904577249e-05,
      "loss": 0.3285,
      "step": 12408
    },
    {
      "epoch": 0.9391419516579159,
      "grad_norm": 0.73046875,
      "learning_rate": 1.0985801633171665e-05,
      "loss": 0.2851,
      "step": 12409
    },
    {
      "epoch": 0.9392176339813633,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0984616347780797e-05,
      "loss": 0.3398,
      "step": 12410
    },
    {
      "epoch": 0.9392933163048106,
      "grad_norm": 0.83203125,
      "learning_rate": 1.0983431048421454e-05,
      "loss": 0.3483,
      "step": 12411
    },
    {
      "epoch": 0.9393689986282578,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0982245735110453e-05,
      "loss": 0.3062,
      "step": 12412
    },
    {
      "epoch": 0.9394446809517052,
      "grad_norm": 0.79296875,
      "learning_rate": 1.0981060407864612e-05,
      "loss": 0.3337,
      "step": 12413
    },
    {
      "epoch": 0.9395203632751525,
      "grad_norm": 0.7734375,
      "learning_rate": 1.0979875066700747e-05,
      "loss": 0.3225,
      "step": 12414
    },
    {
      "epoch": 0.9395960455985999,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0978689711635671e-05,
      "loss": 0.3274,
      "step": 12415
    },
    {
      "epoch": 0.9396717279220472,
      "grad_norm": 0.78515625,
      "learning_rate": 1.09775043426862e-05,
      "loss": 0.3468,
      "step": 12416
    },
    {
      "epoch": 0.9397474102454946,
      "grad_norm": 1.1015625,
      "learning_rate": 1.0976318959869157e-05,
      "loss": 0.3722,
      "step": 12417
    },
    {
      "epoch": 0.9398230925689419,
      "grad_norm": 0.73046875,
      "learning_rate": 1.0975133563201347e-05,
      "loss": 0.3036,
      "step": 12418
    },
    {
      "epoch": 0.9398987748923892,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0973948152699603e-05,
      "loss": 0.3042,
      "step": 12419
    },
    {
      "epoch": 0.9399744572158365,
      "grad_norm": 0.765625,
      "learning_rate": 1.0972762728380728e-05,
      "loss": 0.283,
      "step": 12420
    },
    {
      "epoch": 0.9400501395392838,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0971577290261544e-05,
      "loss": 0.2908,
      "step": 12421
    },
    {
      "epoch": 0.9401258218627312,
      "grad_norm": 0.7109375,
      "learning_rate": 1.0970391838358869e-05,
      "loss": 0.2861,
      "step": 12422
    },
    {
      "epoch": 0.9402015041861785,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0969206372689523e-05,
      "loss": 0.3169,
      "step": 12423
    },
    {
      "epoch": 0.9402771865096259,
      "grad_norm": 0.71484375,
      "learning_rate": 1.096802089327032e-05,
      "loss": 0.303,
      "step": 12424
    },
    {
      "epoch": 0.9403528688330732,
      "grad_norm": 0.7734375,
      "learning_rate": 1.0966835400118078e-05,
      "loss": 0.3252,
      "step": 12425
    },
    {
      "epoch": 0.9404285511565205,
      "grad_norm": 0.90234375,
      "learning_rate": 1.0965649893249619e-05,
      "loss": 0.302,
      "step": 12426
    },
    {
      "epoch": 0.9405042334799678,
      "grad_norm": 0.78125,
      "learning_rate": 1.0964464372681757e-05,
      "loss": 0.3543,
      "step": 12427
    },
    {
      "epoch": 0.9405799158034152,
      "grad_norm": 0.77734375,
      "learning_rate": 1.0963278838431318e-05,
      "loss": 0.3253,
      "step": 12428
    },
    {
      "epoch": 0.9406555981268625,
      "grad_norm": 0.75,
      "learning_rate": 1.096209329051511e-05,
      "loss": 0.3009,
      "step": 12429
    },
    {
      "epoch": 0.9407312804503098,
      "grad_norm": 0.78515625,
      "learning_rate": 1.0960907728949959e-05,
      "loss": 0.3418,
      "step": 12430
    },
    {
      "epoch": 0.9408069627737572,
      "grad_norm": 0.7890625,
      "learning_rate": 1.0959722153752683e-05,
      "loss": 0.3453,
      "step": 12431
    },
    {
      "epoch": 0.9408826450972045,
      "grad_norm": 0.83203125,
      "learning_rate": 1.0958536564940102e-05,
      "loss": 0.3479,
      "step": 12432
    },
    {
      "epoch": 0.9409583274206518,
      "grad_norm": 0.796875,
      "learning_rate": 1.0957350962529035e-05,
      "loss": 0.3489,
      "step": 12433
    },
    {
      "epoch": 0.9410340097440991,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0956165346536301e-05,
      "loss": 0.3113,
      "step": 12434
    },
    {
      "epoch": 0.9411096920675465,
      "grad_norm": 0.79296875,
      "learning_rate": 1.095497971697872e-05,
      "loss": 0.3742,
      "step": 12435
    },
    {
      "epoch": 0.9411853743909938,
      "grad_norm": 0.7890625,
      "learning_rate": 1.0953794073873111e-05,
      "loss": 0.3121,
      "step": 12436
    },
    {
      "epoch": 0.9412610567144412,
      "grad_norm": 0.796875,
      "learning_rate": 1.0952608417236301e-05,
      "loss": 0.3339,
      "step": 12437
    },
    {
      "epoch": 0.9413367390378885,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0951422747085103e-05,
      "loss": 0.2893,
      "step": 12438
    },
    {
      "epoch": 0.9414124213613357,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0950237063436342e-05,
      "loss": 0.2838,
      "step": 12439
    },
    {
      "epoch": 0.9414881036847831,
      "grad_norm": 0.80859375,
      "learning_rate": 1.0949051366306838e-05,
      "loss": 0.3627,
      "step": 12440
    },
    {
      "epoch": 0.9415637860082304,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0947865655713412e-05,
      "loss": 0.2881,
      "step": 12441
    },
    {
      "epoch": 0.9416394683316778,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0946679931672883e-05,
      "loss": 0.3135,
      "step": 12442
    },
    {
      "epoch": 0.9417151506551251,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0945494194202077e-05,
      "loss": 0.2812,
      "step": 12443
    },
    {
      "epoch": 0.9417908329785725,
      "grad_norm": 0.76171875,
      "learning_rate": 1.094430844331781e-05,
      "loss": 0.3209,
      "step": 12444
    },
    {
      "epoch": 0.9418665153020198,
      "grad_norm": 0.83203125,
      "learning_rate": 1.0943122679036909e-05,
      "loss": 0.3309,
      "step": 12445
    },
    {
      "epoch": 0.941942197625467,
      "grad_norm": 0.765625,
      "learning_rate": 1.0941936901376195e-05,
      "loss": 0.3058,
      "step": 12446
    },
    {
      "epoch": 0.9420178799489144,
      "grad_norm": 0.71875,
      "learning_rate": 1.094075111035249e-05,
      "loss": 0.3028,
      "step": 12447
    },
    {
      "epoch": 0.9420935622723617,
      "grad_norm": 0.7734375,
      "learning_rate": 1.0939565305982616e-05,
      "loss": 0.3351,
      "step": 12448
    },
    {
      "epoch": 0.9421692445958091,
      "grad_norm": 0.734375,
      "learning_rate": 1.0938379488283396e-05,
      "loss": 0.3238,
      "step": 12449
    },
    {
      "epoch": 0.9422449269192564,
      "grad_norm": 3.390625,
      "learning_rate": 1.093719365727165e-05,
      "loss": 0.3376,
      "step": 12450
    },
    {
      "epoch": 0.9423206092427038,
      "grad_norm": 0.765625,
      "learning_rate": 1.0936007812964204e-05,
      "loss": 0.3194,
      "step": 12451
    },
    {
      "epoch": 0.9423962915661511,
      "grad_norm": 0.7734375,
      "learning_rate": 1.0934821955377885e-05,
      "loss": 0.3005,
      "step": 12452
    },
    {
      "epoch": 0.9424719738895984,
      "grad_norm": 0.796875,
      "learning_rate": 1.0933636084529507e-05,
      "loss": 0.3521,
      "step": 12453
    },
    {
      "epoch": 0.9425476562130457,
      "grad_norm": 0.671875,
      "learning_rate": 1.0932450200435902e-05,
      "loss": 0.2742,
      "step": 12454
    },
    {
      "epoch": 0.942623338536493,
      "grad_norm": 0.7578125,
      "learning_rate": 1.093126430311389e-05,
      "loss": 0.3163,
      "step": 12455
    },
    {
      "epoch": 0.9426990208599404,
      "grad_norm": 0.77734375,
      "learning_rate": 1.0930078392580296e-05,
      "loss": 0.3276,
      "step": 12456
    },
    {
      "epoch": 0.9427747031833877,
      "grad_norm": 0.73046875,
      "learning_rate": 1.0928892468851946e-05,
      "loss": 0.3064,
      "step": 12457
    },
    {
      "epoch": 0.9428503855068351,
      "grad_norm": 0.703125,
      "learning_rate": 1.0927706531945661e-05,
      "loss": 0.2883,
      "step": 12458
    },
    {
      "epoch": 0.9429260678302824,
      "grad_norm": 0.79296875,
      "learning_rate": 1.0926520581878264e-05,
      "loss": 0.3238,
      "step": 12459
    },
    {
      "epoch": 0.9430017501537297,
      "grad_norm": 0.7109375,
      "learning_rate": 1.0925334618666585e-05,
      "loss": 0.2889,
      "step": 12460
    },
    {
      "epoch": 0.943077432477177,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0924148642327448e-05,
      "loss": 0.3171,
      "step": 12461
    },
    {
      "epoch": 0.9431531148006244,
      "grad_norm": 0.69921875,
      "learning_rate": 1.0922962652877673e-05,
      "loss": 0.2822,
      "step": 12462
    },
    {
      "epoch": 0.9432287971240717,
      "grad_norm": 0.73828125,
      "learning_rate": 1.0921776650334093e-05,
      "loss": 0.2862,
      "step": 12463
    },
    {
      "epoch": 0.943304479447519,
      "grad_norm": 0.75,
      "learning_rate": 1.0920590634713527e-05,
      "loss": 0.2932,
      "step": 12464
    },
    {
      "epoch": 0.9433801617709664,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0919404606032802e-05,
      "loss": 0.3045,
      "step": 12465
    },
    {
      "epoch": 0.9434558440944137,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0918218564308748e-05,
      "loss": 0.3016,
      "step": 12466
    },
    {
      "epoch": 0.943531526417861,
      "grad_norm": 1.2890625,
      "learning_rate": 1.0917032509558185e-05,
      "loss": 0.3579,
      "step": 12467
    },
    {
      "epoch": 0.9436072087413083,
      "grad_norm": 0.7890625,
      "learning_rate": 1.0915846441797942e-05,
      "loss": 0.3383,
      "step": 12468
    },
    {
      "epoch": 0.9436828910647557,
      "grad_norm": 0.82421875,
      "learning_rate": 1.0914660361044849e-05,
      "loss": 0.3613,
      "step": 12469
    },
    {
      "epoch": 0.943758573388203,
      "grad_norm": 0.69140625,
      "learning_rate": 1.091347426731573e-05,
      "loss": 0.2818,
      "step": 12470
    },
    {
      "epoch": 0.9438342557116504,
      "grad_norm": 0.78125,
      "learning_rate": 1.0912288160627404e-05,
      "loss": 0.351,
      "step": 12471
    },
    {
      "epoch": 0.9439099380350977,
      "grad_norm": 0.734375,
      "learning_rate": 1.091110204099671e-05,
      "loss": 0.3208,
      "step": 12472
    },
    {
      "epoch": 0.943985620358545,
      "grad_norm": 0.68359375,
      "learning_rate": 1.0909915908440469e-05,
      "loss": 0.259,
      "step": 12473
    },
    {
      "epoch": 0.9440613026819923,
      "grad_norm": 0.7890625,
      "learning_rate": 1.0908729762975507e-05,
      "loss": 0.3412,
      "step": 12474
    },
    {
      "epoch": 0.9441369850054396,
      "grad_norm": 1.140625,
      "learning_rate": 1.0907543604618658e-05,
      "loss": 0.4005,
      "step": 12475
    },
    {
      "epoch": 0.944212667328887,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0906357433386745e-05,
      "loss": 0.3161,
      "step": 12476
    },
    {
      "epoch": 0.9442883496523343,
      "grad_norm": 0.71875,
      "learning_rate": 1.0905171249296593e-05,
      "loss": 0.3042,
      "step": 12477
    },
    {
      "epoch": 0.9443640319757817,
      "grad_norm": 0.71875,
      "learning_rate": 1.0903985052365035e-05,
      "loss": 0.2895,
      "step": 12478
    },
    {
      "epoch": 0.944439714299229,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0902798842608899e-05,
      "loss": 0.3202,
      "step": 12479
    },
    {
      "epoch": 0.9445153966226764,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0901612620045013e-05,
      "loss": 0.3119,
      "step": 12480
    },
    {
      "epoch": 0.9445910789461236,
      "grad_norm": 0.7890625,
      "learning_rate": 1.0900426384690204e-05,
      "loss": 0.3356,
      "step": 12481
    },
    {
      "epoch": 0.9446667612695709,
      "grad_norm": 0.734375,
      "learning_rate": 1.08992401365613e-05,
      "loss": 0.2866,
      "step": 12482
    },
    {
      "epoch": 0.9447424435930183,
      "grad_norm": 0.78515625,
      "learning_rate": 1.0898053875675132e-05,
      "loss": 0.3379,
      "step": 12483
    },
    {
      "epoch": 0.9448181259164656,
      "grad_norm": 0.75,
      "learning_rate": 1.089686760204853e-05,
      "loss": 0.3136,
      "step": 12484
    },
    {
      "epoch": 0.944893808239913,
      "grad_norm": 0.796875,
      "learning_rate": 1.0895681315698323e-05,
      "loss": 0.3584,
      "step": 12485
    },
    {
      "epoch": 0.9449694905633603,
      "grad_norm": 0.734375,
      "learning_rate": 1.0894495016641336e-05,
      "loss": 0.2835,
      "step": 12486
    },
    {
      "epoch": 0.9450451728868077,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0893308704894403e-05,
      "loss": 0.3012,
      "step": 12487
    },
    {
      "epoch": 0.9451208552102549,
      "grad_norm": 0.83984375,
      "learning_rate": 1.0892122380474354e-05,
      "loss": 0.3673,
      "step": 12488
    },
    {
      "epoch": 0.9451965375337023,
      "grad_norm": 0.7734375,
      "learning_rate": 1.0890936043398019e-05,
      "loss": 0.3154,
      "step": 12489
    },
    {
      "epoch": 0.9452722198571496,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0889749693682226e-05,
      "loss": 0.3332,
      "step": 12490
    },
    {
      "epoch": 0.9453479021805969,
      "grad_norm": 0.78125,
      "learning_rate": 1.0888563331343807e-05,
      "loss": 0.3522,
      "step": 12491
    },
    {
      "epoch": 0.9454235845040443,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0887376956399591e-05,
      "loss": 0.3442,
      "step": 12492
    },
    {
      "epoch": 0.9454992668274916,
      "grad_norm": 0.78515625,
      "learning_rate": 1.0886190568866411e-05,
      "loss": 0.354,
      "step": 12493
    },
    {
      "epoch": 0.945574949150939,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0885004168761097e-05,
      "loss": 0.3194,
      "step": 12494
    },
    {
      "epoch": 0.9456506314743862,
      "grad_norm": 0.85546875,
      "learning_rate": 1.0883817756100481e-05,
      "loss": 0.3369,
      "step": 12495
    },
    {
      "epoch": 0.9457263137978336,
      "grad_norm": 0.73828125,
      "learning_rate": 1.0882631330901394e-05,
      "loss": 0.2659,
      "step": 12496
    },
    {
      "epoch": 0.9458019961212809,
      "grad_norm": 0.80078125,
      "learning_rate": 1.0881444893180665e-05,
      "loss": 0.3813,
      "step": 12497
    },
    {
      "epoch": 0.9458776784447283,
      "grad_norm": 0.74609375,
      "learning_rate": 1.088025844295513e-05,
      "loss": 0.3007,
      "step": 12498
    },
    {
      "epoch": 0.9459533607681756,
      "grad_norm": 0.765625,
      "learning_rate": 1.0879071980241617e-05,
      "loss": 0.3175,
      "step": 12499
    },
    {
      "epoch": 0.946029043091623,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0877885505056958e-05,
      "loss": 0.3094,
      "step": 12500
    },
    {
      "epoch": 0.9461047254150703,
      "grad_norm": 0.7734375,
      "learning_rate": 1.0876699017417988e-05,
      "loss": 0.3089,
      "step": 12501
    },
    {
      "epoch": 0.9461804077385175,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0875512517341538e-05,
      "loss": 0.2862,
      "step": 12502
    },
    {
      "epoch": 0.9462560900619649,
      "grad_norm": 0.79296875,
      "learning_rate": 1.087432600484444e-05,
      "loss": 0.3541,
      "step": 12503
    },
    {
      "epoch": 0.9463317723854122,
      "grad_norm": 0.7421875,
      "learning_rate": 1.087313947994353e-05,
      "loss": 0.3073,
      "step": 12504
    },
    {
      "epoch": 0.9464074547088596,
      "grad_norm": 0.69921875,
      "learning_rate": 1.0871952942655634e-05,
      "loss": 0.2767,
      "step": 12505
    },
    {
      "epoch": 0.9464831370323069,
      "grad_norm": 0.66015625,
      "learning_rate": 1.087076639299759e-05,
      "loss": 0.2618,
      "step": 12506
    },
    {
      "epoch": 0.9465588193557543,
      "grad_norm": 0.6953125,
      "learning_rate": 1.0869579830986233e-05,
      "loss": 0.2751,
      "step": 12507
    },
    {
      "epoch": 0.9466345016792016,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0868393256638394e-05,
      "loss": 0.3111,
      "step": 12508
    },
    {
      "epoch": 0.9467101840026488,
      "grad_norm": 0.828125,
      "learning_rate": 1.0867206669970902e-05,
      "loss": 0.3296,
      "step": 12509
    },
    {
      "epoch": 0.9467858663260962,
      "grad_norm": 0.8125,
      "learning_rate": 1.0866020071000597e-05,
      "loss": 0.3756,
      "step": 12510
    },
    {
      "epoch": 0.9468615486495435,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0864833459744313e-05,
      "loss": 0.3086,
      "step": 12511
    },
    {
      "epoch": 0.9469372309729909,
      "grad_norm": 0.828125,
      "learning_rate": 1.0863646836218882e-05,
      "loss": 0.3303,
      "step": 12512
    },
    {
      "epoch": 0.9470129132964382,
      "grad_norm": 0.75,
      "learning_rate": 1.0862460200441138e-05,
      "loss": 0.3026,
      "step": 12513
    },
    {
      "epoch": 0.9470885956198856,
      "grad_norm": 0.93359375,
      "learning_rate": 1.0861273552427915e-05,
      "loss": 0.3155,
      "step": 12514
    },
    {
      "epoch": 0.9471642779433329,
      "grad_norm": 0.8125,
      "learning_rate": 1.0860086892196049e-05,
      "loss": 0.3607,
      "step": 12515
    },
    {
      "epoch": 0.9472399602667801,
      "grad_norm": 0.68359375,
      "learning_rate": 1.0858900219762377e-05,
      "loss": 0.2804,
      "step": 12516
    },
    {
      "epoch": 0.9473156425902275,
      "grad_norm": 0.796875,
      "learning_rate": 1.085771353514373e-05,
      "loss": 0.3529,
      "step": 12517
    },
    {
      "epoch": 0.9473913249136748,
      "grad_norm": 0.875,
      "learning_rate": 1.0856526838356943e-05,
      "loss": 0.3056,
      "step": 12518
    },
    {
      "epoch": 0.9474670072371222,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0855340129418854e-05,
      "loss": 0.2943,
      "step": 12519
    },
    {
      "epoch": 0.9475426895605695,
      "grad_norm": 0.7109375,
      "learning_rate": 1.0854153408346298e-05,
      "loss": 0.29,
      "step": 12520
    },
    {
      "epoch": 0.9476183718840169,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0852966675156106e-05,
      "loss": 0.3354,
      "step": 12521
    },
    {
      "epoch": 0.9476940542074642,
      "grad_norm": 0.7109375,
      "learning_rate": 1.0851779929865125e-05,
      "loss": 0.2862,
      "step": 12522
    },
    {
      "epoch": 0.9477697365309115,
      "grad_norm": 0.8203125,
      "learning_rate": 1.085059317249018e-05,
      "loss": 0.3565,
      "step": 12523
    },
    {
      "epoch": 0.9478454188543588,
      "grad_norm": 0.98828125,
      "learning_rate": 1.084940640304811e-05,
      "loss": 0.3361,
      "step": 12524
    },
    {
      "epoch": 0.9479211011778061,
      "grad_norm": 0.97265625,
      "learning_rate": 1.0848219621555756e-05,
      "loss": 0.344,
      "step": 12525
    },
    {
      "epoch": 0.9479967835012535,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0847032828029952e-05,
      "loss": 0.2929,
      "step": 12526
    },
    {
      "epoch": 0.9480724658247008,
      "grad_norm": 0.71875,
      "learning_rate": 1.084584602248753e-05,
      "loss": 0.276,
      "step": 12527
    },
    {
      "epoch": 0.9481481481481482,
      "grad_norm": 0.83984375,
      "learning_rate": 1.0844659204945331e-05,
      "loss": 0.3338,
      "step": 12528
    },
    {
      "epoch": 0.9482238304715955,
      "grad_norm": 0.7734375,
      "learning_rate": 1.0843472375420192e-05,
      "loss": 0.3216,
      "step": 12529
    },
    {
      "epoch": 0.9482995127950428,
      "grad_norm": 0.77734375,
      "learning_rate": 1.084228553392895e-05,
      "loss": 0.3448,
      "step": 12530
    },
    {
      "epoch": 0.9483751951184901,
      "grad_norm": 0.7734375,
      "learning_rate": 1.0841098680488445e-05,
      "loss": 0.3323,
      "step": 12531
    },
    {
      "epoch": 0.9484508774419375,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0839911815115509e-05,
      "loss": 0.3162,
      "step": 12532
    },
    {
      "epoch": 0.9485265597653848,
      "grad_norm": 0.77734375,
      "learning_rate": 1.083872493782698e-05,
      "loss": 0.3223,
      "step": 12533
    },
    {
      "epoch": 0.9486022420888321,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0837538048639705e-05,
      "loss": 0.2835,
      "step": 12534
    },
    {
      "epoch": 0.9486779244122795,
      "grad_norm": 0.72265625,
      "learning_rate": 1.083635114757051e-05,
      "loss": 0.2876,
      "step": 12535
    },
    {
      "epoch": 0.9487536067357268,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0835164234636242e-05,
      "loss": 0.3168,
      "step": 12536
    },
    {
      "epoch": 0.9488292890591741,
      "grad_norm": 0.77734375,
      "learning_rate": 1.0833977309853736e-05,
      "loss": 0.3606,
      "step": 12537
    },
    {
      "epoch": 0.9489049713826214,
      "grad_norm": 0.78515625,
      "learning_rate": 1.083279037323983e-05,
      "loss": 0.355,
      "step": 12538
    },
    {
      "epoch": 0.9489806537060688,
      "grad_norm": 0.7734375,
      "learning_rate": 1.0831603424811361e-05,
      "loss": 0.3302,
      "step": 12539
    },
    {
      "epoch": 0.9490563360295161,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0830416464585174e-05,
      "loss": 0.2979,
      "step": 12540
    },
    {
      "epoch": 0.9491320183529635,
      "grad_norm": 0.7890625,
      "learning_rate": 1.0829229492578104e-05,
      "loss": 0.3225,
      "step": 12541
    },
    {
      "epoch": 0.9492077006764108,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0828042508806986e-05,
      "loss": 0.2932,
      "step": 12542
    },
    {
      "epoch": 0.9492833829998581,
      "grad_norm": 0.66796875,
      "learning_rate": 1.0826855513288667e-05,
      "loss": 0.2672,
      "step": 12543
    },
    {
      "epoch": 0.9493590653233054,
      "grad_norm": 0.78515625,
      "learning_rate": 1.0825668506039986e-05,
      "loss": 0.3475,
      "step": 12544
    },
    {
      "epoch": 0.9494347476467527,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0824481487077778e-05,
      "loss": 0.3077,
      "step": 12545
    },
    {
      "epoch": 0.9495104299702001,
      "grad_norm": 1.9140625,
      "learning_rate": 1.0823294456418886e-05,
      "loss": 0.3992,
      "step": 12546
    },
    {
      "epoch": 0.9495861122936474,
      "grad_norm": 0.703125,
      "learning_rate": 1.0822107414080149e-05,
      "loss": 0.2888,
      "step": 12547
    },
    {
      "epoch": 0.9496617946170948,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0820920360078407e-05,
      "loss": 0.3328,
      "step": 12548
    },
    {
      "epoch": 0.9497374769405421,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0819733294430502e-05,
      "loss": 0.3035,
      "step": 12549
    },
    {
      "epoch": 0.9498131592639895,
      "grad_norm": 0.73046875,
      "learning_rate": 1.0818546217153272e-05,
      "loss": 0.2993,
      "step": 12550
    },
    {
      "epoch": 0.9498888415874367,
      "grad_norm": 0.82421875,
      "learning_rate": 1.081735912826356e-05,
      "loss": 0.3828,
      "step": 12551
    },
    {
      "epoch": 0.949964523910884,
      "grad_norm": 0.7890625,
      "learning_rate": 1.0816172027778208e-05,
      "loss": 0.3184,
      "step": 12552
    },
    {
      "epoch": 0.9500402062343314,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0814984915714053e-05,
      "loss": 0.2934,
      "step": 12553
    },
    {
      "epoch": 0.9501158885577787,
      "grad_norm": 0.7109375,
      "learning_rate": 1.081379779208794e-05,
      "loss": 0.2816,
      "step": 12554
    },
    {
      "epoch": 0.9501915708812261,
      "grad_norm": 0.734375,
      "learning_rate": 1.0812610656916708e-05,
      "loss": 0.3045,
      "step": 12555
    },
    {
      "epoch": 0.9502672532046734,
      "grad_norm": 0.79296875,
      "learning_rate": 1.0811423510217198e-05,
      "loss": 0.3191,
      "step": 12556
    },
    {
      "epoch": 0.9503429355281207,
      "grad_norm": 0.7109375,
      "learning_rate": 1.0810236352006254e-05,
      "loss": 0.2762,
      "step": 12557
    },
    {
      "epoch": 0.950418617851568,
      "grad_norm": 1.1796875,
      "learning_rate": 1.0809049182300716e-05,
      "loss": 0.4041,
      "step": 12558
    },
    {
      "epoch": 0.9504943001750153,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0807862001117431e-05,
      "loss": 0.3224,
      "step": 12559
    },
    {
      "epoch": 0.9505699824984627,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0806674808473234e-05,
      "loss": 0.3007,
      "step": 12560
    },
    {
      "epoch": 0.95064566482191,
      "grad_norm": 0.73828125,
      "learning_rate": 1.080548760438497e-05,
      "loss": 0.2946,
      "step": 12561
    },
    {
      "epoch": 0.9507213471453574,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0804300388869481e-05,
      "loss": 0.3197,
      "step": 12562
    },
    {
      "epoch": 0.9507970294688047,
      "grad_norm": 0.69921875,
      "learning_rate": 1.0803113161943613e-05,
      "loss": 0.2622,
      "step": 12563
    },
    {
      "epoch": 0.950872711792252,
      "grad_norm": 0.7421875,
      "learning_rate": 1.080192592362421e-05,
      "loss": 0.2934,
      "step": 12564
    },
    {
      "epoch": 0.9509483941156993,
      "grad_norm": 0.734375,
      "learning_rate": 1.0800738673928106e-05,
      "loss": 0.3054,
      "step": 12565
    },
    {
      "epoch": 0.9510240764391467,
      "grad_norm": 0.70703125,
      "learning_rate": 1.0799551412872151e-05,
      "loss": 0.2784,
      "step": 12566
    },
    {
      "epoch": 0.951099758762594,
      "grad_norm": 0.703125,
      "learning_rate": 1.0798364140473187e-05,
      "loss": 0.2455,
      "step": 12567
    },
    {
      "epoch": 0.9511754410860414,
      "grad_norm": 0.765625,
      "learning_rate": 1.079717685674806e-05,
      "loss": 0.3301,
      "step": 12568
    },
    {
      "epoch": 0.9512511234094887,
      "grad_norm": 0.7578125,
      "learning_rate": 1.0795989561713609e-05,
      "loss": 0.3465,
      "step": 12569
    },
    {
      "epoch": 0.951326805732936,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0794802255386682e-05,
      "loss": 0.3377,
      "step": 12570
    },
    {
      "epoch": 0.9514024880563833,
      "grad_norm": 0.8359375,
      "learning_rate": 1.0793614937784118e-05,
      "loss": 0.3629,
      "step": 12571
    },
    {
      "epoch": 0.9514781703798306,
      "grad_norm": 0.765625,
      "learning_rate": 1.0792427608922766e-05,
      "loss": 0.3433,
      "step": 12572
    },
    {
      "epoch": 0.951553852703278,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0791240268819472e-05,
      "loss": 0.3302,
      "step": 12573
    },
    {
      "epoch": 0.9516295350267253,
      "grad_norm": 0.80859375,
      "learning_rate": 1.0790052917491071e-05,
      "loss": 0.3691,
      "step": 12574
    },
    {
      "epoch": 0.9517052173501727,
      "grad_norm": 0.7578125,
      "learning_rate": 1.0788865554954417e-05,
      "loss": 0.3304,
      "step": 12575
    },
    {
      "epoch": 0.95178089967362,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0787678181226351e-05,
      "loss": 0.3133,
      "step": 12576
    },
    {
      "epoch": 0.9518565819970674,
      "grad_norm": 0.79296875,
      "learning_rate": 1.0786490796323716e-05,
      "loss": 0.3457,
      "step": 12577
    },
    {
      "epoch": 0.9519322643205146,
      "grad_norm": 0.73828125,
      "learning_rate": 1.0785303400263362e-05,
      "loss": 0.3123,
      "step": 12578
    },
    {
      "epoch": 0.9520079466439619,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0784115993062132e-05,
      "loss": 0.2734,
      "step": 12579
    },
    {
      "epoch": 0.9520836289674093,
      "grad_norm": 0.78125,
      "learning_rate": 1.0782928574736869e-05,
      "loss": 0.3533,
      "step": 12580
    },
    {
      "epoch": 0.9521593112908566,
      "grad_norm": 0.796875,
      "learning_rate": 1.078174114530442e-05,
      "loss": 0.3008,
      "step": 12581
    },
    {
      "epoch": 0.952234993614304,
      "grad_norm": 1.1796875,
      "learning_rate": 1.0780553704781637e-05,
      "loss": 0.32,
      "step": 12582
    },
    {
      "epoch": 0.9523106759377513,
      "grad_norm": 0.71875,
      "learning_rate": 1.0779366253185355e-05,
      "loss": 0.2903,
      "step": 12583
    },
    {
      "epoch": 0.9523863582611987,
      "grad_norm": 0.79296875,
      "learning_rate": 1.0778178790532425e-05,
      "loss": 0.3454,
      "step": 12584
    },
    {
      "epoch": 0.9524620405846459,
      "grad_norm": 0.734375,
      "learning_rate": 1.0776991316839695e-05,
      "loss": 0.3107,
      "step": 12585
    },
    {
      "epoch": 0.9525377229080932,
      "grad_norm": 0.7890625,
      "learning_rate": 1.0775803832124008e-05,
      "loss": 0.3433,
      "step": 12586
    },
    {
      "epoch": 0.9526134052315406,
      "grad_norm": 0.80859375,
      "learning_rate": 1.0774616336402217e-05,
      "loss": 0.3684,
      "step": 12587
    },
    {
      "epoch": 0.9526890875549879,
      "grad_norm": 0.83984375,
      "learning_rate": 1.0773428829691161e-05,
      "loss": 0.3727,
      "step": 12588
    },
    {
      "epoch": 0.9527647698784353,
      "grad_norm": 0.73046875,
      "learning_rate": 1.0772241312007687e-05,
      "loss": 0.2842,
      "step": 12589
    },
    {
      "epoch": 0.9528404522018826,
      "grad_norm": 0.7109375,
      "learning_rate": 1.0771053783368647e-05,
      "loss": 0.2988,
      "step": 12590
    },
    {
      "epoch": 0.95291613452533,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0769866243790887e-05,
      "loss": 0.3005,
      "step": 12591
    },
    {
      "epoch": 0.9529918168487772,
      "grad_norm": 0.7109375,
      "learning_rate": 1.0768678693291255e-05,
      "loss": 0.2565,
      "step": 12592
    },
    {
      "epoch": 0.9530674991722246,
      "grad_norm": 0.79296875,
      "learning_rate": 1.0767491131886594e-05,
      "loss": 0.3192,
      "step": 12593
    },
    {
      "epoch": 0.9531431814956719,
      "grad_norm": 0.78515625,
      "learning_rate": 1.0766303559593754e-05,
      "loss": 0.3148,
      "step": 12594
    },
    {
      "epoch": 0.9532188638191192,
      "grad_norm": 0.7734375,
      "learning_rate": 1.076511597642958e-05,
      "loss": 0.3317,
      "step": 12595
    },
    {
      "epoch": 0.9532945461425666,
      "grad_norm": 0.7734375,
      "learning_rate": 1.076392838241093e-05,
      "loss": 0.3261,
      "step": 12596
    },
    {
      "epoch": 0.9533702284660139,
      "grad_norm": 0.84765625,
      "learning_rate": 1.0762740777554642e-05,
      "loss": 0.3539,
      "step": 12597
    },
    {
      "epoch": 0.9534459107894613,
      "grad_norm": 0.71875,
      "learning_rate": 1.0761553161877565e-05,
      "loss": 0.2746,
      "step": 12598
    },
    {
      "epoch": 0.9535215931129085,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0760365535396552e-05,
      "loss": 0.3204,
      "step": 12599
    },
    {
      "epoch": 0.9535972754363559,
      "grad_norm": 0.91015625,
      "learning_rate": 1.0759177898128449e-05,
      "loss": 0.3205,
      "step": 12600
    },
    {
      "epoch": 0.9536729577598032,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0757990250090102e-05,
      "loss": 0.3045,
      "step": 12601
    },
    {
      "epoch": 0.9537486400832506,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0756802591298365e-05,
      "loss": 0.3136,
      "step": 12602
    },
    {
      "epoch": 0.9538243224066979,
      "grad_norm": 0.734375,
      "learning_rate": 1.0755614921770084e-05,
      "loss": 0.3228,
      "step": 12603
    },
    {
      "epoch": 0.9539000047301452,
      "grad_norm": 0.6953125,
      "learning_rate": 1.0754427241522108e-05,
      "loss": 0.2576,
      "step": 12604
    },
    {
      "epoch": 0.9539756870535926,
      "grad_norm": 0.73828125,
      "learning_rate": 1.0753239550571286e-05,
      "loss": 0.2681,
      "step": 12605
    },
    {
      "epoch": 0.9540513693770398,
      "grad_norm": 0.765625,
      "learning_rate": 1.0752051848934471e-05,
      "loss": 0.3145,
      "step": 12606
    },
    {
      "epoch": 0.9541270517004872,
      "grad_norm": 0.703125,
      "learning_rate": 1.0750864136628509e-05,
      "loss": 0.2955,
      "step": 12607
    },
    {
      "epoch": 0.9542027340239345,
      "grad_norm": 0.87890625,
      "learning_rate": 1.0749676413670252e-05,
      "loss": 0.316,
      "step": 12608
    },
    {
      "epoch": 0.9542784163473819,
      "grad_norm": 0.765625,
      "learning_rate": 1.0748488680076546e-05,
      "loss": 0.2904,
      "step": 12609
    },
    {
      "epoch": 0.9543540986708292,
      "grad_norm": 0.73828125,
      "learning_rate": 1.0747300935864245e-05,
      "loss": 0.3218,
      "step": 12610
    },
    {
      "epoch": 0.9544297809942766,
      "grad_norm": 0.765625,
      "learning_rate": 1.0746113181050197e-05,
      "loss": 0.3043,
      "step": 12611
    },
    {
      "epoch": 0.9545054633177239,
      "grad_norm": 0.78515625,
      "learning_rate": 1.0744925415651252e-05,
      "loss": 0.3444,
      "step": 12612
    },
    {
      "epoch": 0.9545811456411711,
      "grad_norm": 0.77734375,
      "learning_rate": 1.0743737639684264e-05,
      "loss": 0.3233,
      "step": 12613
    },
    {
      "epoch": 0.9546568279646185,
      "grad_norm": 0.7734375,
      "learning_rate": 1.074254985316608e-05,
      "loss": 0.3145,
      "step": 12614
    },
    {
      "epoch": 0.9547325102880658,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0741362056113554e-05,
      "loss": 0.2719,
      "step": 12615
    },
    {
      "epoch": 0.9548081926115132,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0740174248543534e-05,
      "loss": 0.3026,
      "step": 12616
    },
    {
      "epoch": 0.9548838749349605,
      "grad_norm": 0.8515625,
      "learning_rate": 1.073898643047287e-05,
      "loss": 0.3291,
      "step": 12617
    },
    {
      "epoch": 0.9549595572584079,
      "grad_norm": 0.7578125,
      "learning_rate": 1.0737798601918416e-05,
      "loss": 0.3012,
      "step": 12618
    },
    {
      "epoch": 0.9550352395818552,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0736610762897025e-05,
      "loss": 0.2977,
      "step": 12619
    },
    {
      "epoch": 0.9551109219053024,
      "grad_norm": 0.73828125,
      "learning_rate": 1.0735422913425549e-05,
      "loss": 0.2733,
      "step": 12620
    },
    {
      "epoch": 0.9551866042287498,
      "grad_norm": 0.79296875,
      "learning_rate": 1.073423505352083e-05,
      "loss": 0.3312,
      "step": 12621
    },
    {
      "epoch": 0.9552622865521971,
      "grad_norm": 0.75,
      "learning_rate": 1.0733047183199729e-05,
      "loss": 0.2985,
      "step": 12622
    },
    {
      "epoch": 0.9553379688756445,
      "grad_norm": 0.77734375,
      "learning_rate": 1.0731859302479097e-05,
      "loss": 0.3455,
      "step": 12623
    },
    {
      "epoch": 0.9554136511990918,
      "grad_norm": 0.71875,
      "learning_rate": 1.0730671411375786e-05,
      "loss": 0.2735,
      "step": 12624
    },
    {
      "epoch": 0.9554893335225392,
      "grad_norm": 1.03125,
      "learning_rate": 1.0729483509906644e-05,
      "loss": 0.3626,
      "step": 12625
    },
    {
      "epoch": 0.9555650158459865,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0728295598088529e-05,
      "loss": 0.3157,
      "step": 12626
    },
    {
      "epoch": 0.9556406981694338,
      "grad_norm": 0.81640625,
      "learning_rate": 1.0727107675938289e-05,
      "loss": 0.3422,
      "step": 12627
    },
    {
      "epoch": 0.9557163804928811,
      "grad_norm": 0.8125,
      "learning_rate": 1.0725919743472778e-05,
      "loss": 0.3626,
      "step": 12628
    },
    {
      "epoch": 0.9557920628163284,
      "grad_norm": 0.70703125,
      "learning_rate": 1.0724731800708857e-05,
      "loss": 0.2946,
      "step": 12629
    },
    {
      "epoch": 0.9558677451397758,
      "grad_norm": 0.8828125,
      "learning_rate": 1.0723543847663364e-05,
      "loss": 0.3855,
      "step": 12630
    },
    {
      "epoch": 0.9559434274632231,
      "grad_norm": 1.078125,
      "learning_rate": 1.0722355884353162e-05,
      "loss": 0.2852,
      "step": 12631
    },
    {
      "epoch": 0.9560191097866705,
      "grad_norm": 0.8046875,
      "learning_rate": 1.0721167910795102e-05,
      "loss": 0.2714,
      "step": 12632
    },
    {
      "epoch": 0.9560947921101178,
      "grad_norm": 0.8046875,
      "learning_rate": 1.0719979927006039e-05,
      "loss": 0.3239,
      "step": 12633
    },
    {
      "epoch": 0.9561704744335651,
      "grad_norm": 0.8046875,
      "learning_rate": 1.0718791933002826e-05,
      "loss": 0.3752,
      "step": 12634
    },
    {
      "epoch": 0.9562461567570124,
      "grad_norm": 0.71875,
      "learning_rate": 1.0717603928802313e-05,
      "loss": 0.2978,
      "step": 12635
    },
    {
      "epoch": 0.9563218390804598,
      "grad_norm": 0.74609375,
      "learning_rate": 1.071641591442136e-05,
      "loss": 0.3153,
      "step": 12636
    },
    {
      "epoch": 0.9563975214039071,
      "grad_norm": 0.73828125,
      "learning_rate": 1.0715227889876813e-05,
      "loss": 0.3161,
      "step": 12637
    },
    {
      "epoch": 0.9564732037273544,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0714039855185539e-05,
      "loss": 0.3399,
      "step": 12638
    },
    {
      "epoch": 0.9565488860508018,
      "grad_norm": 0.6875,
      "learning_rate": 1.0712851810364376e-05,
      "loss": 0.2534,
      "step": 12639
    },
    {
      "epoch": 0.9566245683742491,
      "grad_norm": 0.77734375,
      "learning_rate": 1.0711663755430191e-05,
      "loss": 0.3316,
      "step": 12640
    },
    {
      "epoch": 0.9567002506976964,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0710475690399833e-05,
      "loss": 0.3185,
      "step": 12641
    },
    {
      "epoch": 0.9567759330211437,
      "grad_norm": 0.66796875,
      "learning_rate": 1.0709287615290158e-05,
      "loss": 0.2641,
      "step": 12642
    },
    {
      "epoch": 0.9568516153445911,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0708099530118024e-05,
      "loss": 0.3358,
      "step": 12643
    },
    {
      "epoch": 0.9569272976680384,
      "grad_norm": 0.6953125,
      "learning_rate": 1.070691143490028e-05,
      "loss": 0.275,
      "step": 12644
    },
    {
      "epoch": 0.9570029799914858,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0705723329653782e-05,
      "loss": 0.3006,
      "step": 12645
    },
    {
      "epoch": 0.9570786623149331,
      "grad_norm": 0.7890625,
      "learning_rate": 1.0704535214395392e-05,
      "loss": 0.3627,
      "step": 12646
    },
    {
      "epoch": 0.9571543446383804,
      "grad_norm": 0.7421875,
      "learning_rate": 1.070334708914196e-05,
      "loss": 0.2954,
      "step": 12647
    },
    {
      "epoch": 0.9572300269618277,
      "grad_norm": 0.7578125,
      "learning_rate": 1.0702158953910342e-05,
      "loss": 0.342,
      "step": 12648
    },
    {
      "epoch": 0.957305709285275,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0700970808717394e-05,
      "loss": 0.272,
      "step": 12649
    },
    {
      "epoch": 0.9573813916087224,
      "grad_norm": 0.69921875,
      "learning_rate": 1.0699782653579973e-05,
      "loss": 0.2678,
      "step": 12650
    },
    {
      "epoch": 0.9574570739321697,
      "grad_norm": 0.6875,
      "learning_rate": 1.069859448851493e-05,
      "loss": 0.2857,
      "step": 12651
    },
    {
      "epoch": 0.9575327562556171,
      "grad_norm": 0.7578125,
      "learning_rate": 1.0697406313539134e-05,
      "loss": 0.2967,
      "step": 12652
    },
    {
      "epoch": 0.9576084385790644,
      "grad_norm": 0.70703125,
      "learning_rate": 1.0696218128669424e-05,
      "loss": 0.272,
      "step": 12653
    },
    {
      "epoch": 0.9576841209025118,
      "grad_norm": 0.734375,
      "learning_rate": 1.0695029933922669e-05,
      "loss": 0.2963,
      "step": 12654
    },
    {
      "epoch": 0.957759803225959,
      "grad_norm": 0.78125,
      "learning_rate": 1.0693841729315721e-05,
      "loss": 0.3175,
      "step": 12655
    },
    {
      "epoch": 0.9578354855494063,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0692653514865438e-05,
      "loss": 0.3158,
      "step": 12656
    },
    {
      "epoch": 0.9579111678728537,
      "grad_norm": 0.80078125,
      "learning_rate": 1.0691465290588678e-05,
      "loss": 0.3538,
      "step": 12657
    },
    {
      "epoch": 0.957986850196301,
      "grad_norm": 0.7890625,
      "learning_rate": 1.0690277056502294e-05,
      "loss": 0.3312,
      "step": 12658
    },
    {
      "epoch": 0.9580625325197484,
      "grad_norm": 0.8203125,
      "learning_rate": 1.0689088812623145e-05,
      "loss": 0.3889,
      "step": 12659
    },
    {
      "epoch": 0.9581382148431957,
      "grad_norm": 0.7578125,
      "learning_rate": 1.068790055896809e-05,
      "loss": 0.321,
      "step": 12660
    },
    {
      "epoch": 0.9582138971666431,
      "grad_norm": 0.79296875,
      "learning_rate": 1.0686712295553983e-05,
      "loss": 0.3332,
      "step": 12661
    },
    {
      "epoch": 0.9582895794900903,
      "grad_norm": 0.8203125,
      "learning_rate": 1.0685524022397686e-05,
      "loss": 0.3437,
      "step": 12662
    },
    {
      "epoch": 0.9583652618135376,
      "grad_norm": 0.80859375,
      "learning_rate": 1.0684335739516054e-05,
      "loss": 0.3232,
      "step": 12663
    },
    {
      "epoch": 0.958440944136985,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0683147446925943e-05,
      "loss": 0.2979,
      "step": 12664
    },
    {
      "epoch": 0.9585166264604323,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0681959144644216e-05,
      "loss": 0.3243,
      "step": 12665
    },
    {
      "epoch": 0.9585923087838797,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0680770832687728e-05,
      "loss": 0.2758,
      "step": 12666
    },
    {
      "epoch": 0.958667991107327,
      "grad_norm": 0.7890625,
      "learning_rate": 1.0679582511073337e-05,
      "loss": 0.345,
      "step": 12667
    },
    {
      "epoch": 0.9587436734307744,
      "grad_norm": 0.74609375,
      "learning_rate": 1.06783941798179e-05,
      "loss": 0.3129,
      "step": 12668
    },
    {
      "epoch": 0.9588193557542216,
      "grad_norm": 0.75,
      "learning_rate": 1.0677205838938279e-05,
      "loss": 0.2952,
      "step": 12669
    },
    {
      "epoch": 0.958895038077669,
      "grad_norm": 0.75390625,
      "learning_rate": 1.067601748845133e-05,
      "loss": 0.325,
      "step": 12670
    },
    {
      "epoch": 0.9589707204011163,
      "grad_norm": 0.7734375,
      "learning_rate": 1.0674829128373914e-05,
      "loss": 0.3233,
      "step": 12671
    },
    {
      "epoch": 0.9590464027245637,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0673640758722888e-05,
      "loss": 0.2998,
      "step": 12672
    },
    {
      "epoch": 0.959122085048011,
      "grad_norm": 0.80078125,
      "learning_rate": 1.0672452379515113e-05,
      "loss": 0.3625,
      "step": 12673
    },
    {
      "epoch": 0.9591977673714583,
      "grad_norm": 0.86328125,
      "learning_rate": 1.0671263990767445e-05,
      "loss": 0.3594,
      "step": 12674
    },
    {
      "epoch": 0.9592734496949057,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0670075592496744e-05,
      "loss": 0.337,
      "step": 12675
    },
    {
      "epoch": 0.9593491320183529,
      "grad_norm": 0.7890625,
      "learning_rate": 1.0668887184719878e-05,
      "loss": 0.3235,
      "step": 12676
    },
    {
      "epoch": 0.9594248143418003,
      "grad_norm": 0.67578125,
      "learning_rate": 1.0667698767453694e-05,
      "loss": 0.2386,
      "step": 12677
    },
    {
      "epoch": 0.9595004966652476,
      "grad_norm": 0.75,
      "learning_rate": 1.0666510340715057e-05,
      "loss": 0.2932,
      "step": 12678
    },
    {
      "epoch": 0.959576178988695,
      "grad_norm": 0.62109375,
      "learning_rate": 1.066532190452083e-05,
      "loss": 0.2183,
      "step": 12679
    },
    {
      "epoch": 0.9596518613121423,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0664133458887867e-05,
      "loss": 0.3059,
      "step": 12680
    },
    {
      "epoch": 0.9597275436355897,
      "grad_norm": 0.69921875,
      "learning_rate": 1.0662945003833031e-05,
      "loss": 0.2664,
      "step": 12681
    },
    {
      "epoch": 0.9598032259590369,
      "grad_norm": 0.7890625,
      "learning_rate": 1.0661756539373184e-05,
      "loss": 0.3378,
      "step": 12682
    },
    {
      "epoch": 0.9598789082824842,
      "grad_norm": 0.78125,
      "learning_rate": 1.0660568065525185e-05,
      "loss": 0.3315,
      "step": 12683
    },
    {
      "epoch": 0.9599545906059316,
      "grad_norm": 0.7734375,
      "learning_rate": 1.0659379582305894e-05,
      "loss": 0.3042,
      "step": 12684
    },
    {
      "epoch": 0.9600302729293789,
      "grad_norm": 0.7578125,
      "learning_rate": 1.0658191089732174e-05,
      "loss": 0.306,
      "step": 12685
    },
    {
      "epoch": 0.9601059552528263,
      "grad_norm": 0.69921875,
      "learning_rate": 1.0657002587820879e-05,
      "loss": 0.2707,
      "step": 12686
    },
    {
      "epoch": 0.9601816375762736,
      "grad_norm": 0.77734375,
      "learning_rate": 1.0655814076588879e-05,
      "loss": 0.3185,
      "step": 12687
    },
    {
      "epoch": 0.960257319899721,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0654625556053031e-05,
      "loss": 0.3042,
      "step": 12688
    },
    {
      "epoch": 0.9603330022231682,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0653437026230193e-05,
      "loss": 0.322,
      "step": 12689
    },
    {
      "epoch": 0.9604086845466155,
      "grad_norm": 0.703125,
      "learning_rate": 1.0652248487137233e-05,
      "loss": 0.2588,
      "step": 12690
    },
    {
      "epoch": 0.9604843668700629,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0651059938791008e-05,
      "loss": 0.3177,
      "step": 12691
    },
    {
      "epoch": 0.9605600491935102,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0649871381208379e-05,
      "loss": 0.3125,
      "step": 12692
    },
    {
      "epoch": 0.9606357315169576,
      "grad_norm": 0.76171875,
      "learning_rate": 1.064868281440621e-05,
      "loss": 0.294,
      "step": 12693
    },
    {
      "epoch": 0.9607114138404049,
      "grad_norm": 1.0234375,
      "learning_rate": 1.0647494238401363e-05,
      "loss": 0.2925,
      "step": 12694
    },
    {
      "epoch": 0.9607870961638523,
      "grad_norm": 0.7578125,
      "learning_rate": 1.0646305653210698e-05,
      "loss": 0.3127,
      "step": 12695
    },
    {
      "epoch": 0.9608627784872995,
      "grad_norm": 0.79296875,
      "learning_rate": 1.0645117058851077e-05,
      "loss": 0.3322,
      "step": 12696
    },
    {
      "epoch": 0.9609384608107469,
      "grad_norm": 0.80859375,
      "learning_rate": 1.0643928455339367e-05,
      "loss": 0.3366,
      "step": 12697
    },
    {
      "epoch": 0.9610141431341942,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0642739842692422e-05,
      "loss": 0.311,
      "step": 12698
    },
    {
      "epoch": 0.9610898254576415,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0641551220927115e-05,
      "loss": 0.2968,
      "step": 12699
    },
    {
      "epoch": 0.9611655077810889,
      "grad_norm": 0.8046875,
      "learning_rate": 1.06403625900603e-05,
      "loss": 0.3466,
      "step": 12700
    },
    {
      "epoch": 0.9612411901045362,
      "grad_norm": 0.703125,
      "learning_rate": 1.0639173950108841e-05,
      "loss": 0.2871,
      "step": 12701
    },
    {
      "epoch": 0.9613168724279836,
      "grad_norm": 0.73046875,
      "learning_rate": 1.0637985301089605e-05,
      "loss": 0.2969,
      "step": 12702
    },
    {
      "epoch": 0.9613925547514308,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0636796643019453e-05,
      "loss": 0.2963,
      "step": 12703
    },
    {
      "epoch": 0.9614682370748782,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0635607975915247e-05,
      "loss": 0.276,
      "step": 12704
    },
    {
      "epoch": 0.9615439193983255,
      "grad_norm": 0.6796875,
      "learning_rate": 1.0634419299793852e-05,
      "loss": 0.2638,
      "step": 12705
    },
    {
      "epoch": 0.9616196017217729,
      "grad_norm": 0.80859375,
      "learning_rate": 1.063323061467213e-05,
      "loss": 0.3514,
      "step": 12706
    },
    {
      "epoch": 0.9616952840452202,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0632041920566942e-05,
      "loss": 0.2618,
      "step": 12707
    },
    {
      "epoch": 0.9617709663686675,
      "grad_norm": 0.69140625,
      "learning_rate": 1.063085321749516e-05,
      "loss": 0.2726,
      "step": 12708
    },
    {
      "epoch": 0.9618466486921149,
      "grad_norm": 0.79296875,
      "learning_rate": 1.0629664505473637e-05,
      "loss": 0.3494,
      "step": 12709
    },
    {
      "epoch": 0.9619223310155621,
      "grad_norm": 0.7109375,
      "learning_rate": 1.0628475784519248e-05,
      "loss": 0.2992,
      "step": 12710
    },
    {
      "epoch": 0.9619980133390095,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0627287054648846e-05,
      "loss": 0.3271,
      "step": 12711
    },
    {
      "epoch": 0.9620736956624568,
      "grad_norm": 0.75,
      "learning_rate": 1.0626098315879305e-05,
      "loss": 0.3117,
      "step": 12712
    },
    {
      "epoch": 0.9621493779859042,
      "grad_norm": 0.81640625,
      "learning_rate": 1.0624909568227486e-05,
      "loss": 0.3207,
      "step": 12713
    },
    {
      "epoch": 0.9622250603093515,
      "grad_norm": 0.79296875,
      "learning_rate": 1.062372081171025e-05,
      "loss": 0.3449,
      "step": 12714
    },
    {
      "epoch": 0.9623007426327989,
      "grad_norm": 0.7578125,
      "learning_rate": 1.0622532046344462e-05,
      "loss": 0.3021,
      "step": 12715
    },
    {
      "epoch": 0.9623764249562462,
      "grad_norm": 0.84765625,
      "learning_rate": 1.062134327214699e-05,
      "loss": 0.3555,
      "step": 12716
    },
    {
      "epoch": 0.9624521072796934,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0620154489134698e-05,
      "loss": 0.2706,
      "step": 12717
    },
    {
      "epoch": 0.9625277896031408,
      "grad_norm": 0.75390625,
      "learning_rate": 1.061896569732445e-05,
      "loss": 0.3268,
      "step": 12718
    },
    {
      "epoch": 0.9626034719265881,
      "grad_norm": 0.77734375,
      "learning_rate": 1.0617776896733115e-05,
      "loss": 0.3194,
      "step": 12719
    },
    {
      "epoch": 0.9626791542500355,
      "grad_norm": 0.78515625,
      "learning_rate": 1.061658808737755e-05,
      "loss": 0.3373,
      "step": 12720
    },
    {
      "epoch": 0.9627548365734828,
      "grad_norm": 0.80859375,
      "learning_rate": 1.0615399269274627e-05,
      "loss": 0.3654,
      "step": 12721
    },
    {
      "epoch": 0.9628305188969302,
      "grad_norm": 3.625,
      "learning_rate": 1.0614210442441209e-05,
      "loss": 0.2968,
      "step": 12722
    },
    {
      "epoch": 0.9629062012203775,
      "grad_norm": 0.78125,
      "learning_rate": 1.061302160689416e-05,
      "loss": 0.3203,
      "step": 12723
    },
    {
      "epoch": 0.9629818835438247,
      "grad_norm": 0.765625,
      "learning_rate": 1.0611832762650347e-05,
      "loss": 0.2915,
      "step": 12724
    },
    {
      "epoch": 0.9630575658672721,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0610643909726639e-05,
      "loss": 0.2905,
      "step": 12725
    },
    {
      "epoch": 0.9631332481907194,
      "grad_norm": 0.77734375,
      "learning_rate": 1.0609455048139899e-05,
      "loss": 0.2946,
      "step": 12726
    },
    {
      "epoch": 0.9632089305141668,
      "grad_norm": 0.734375,
      "learning_rate": 1.0608266177906992e-05,
      "loss": 0.2892,
      "step": 12727
    },
    {
      "epoch": 0.9632846128376141,
      "grad_norm": 0.73828125,
      "learning_rate": 1.0607077299044788e-05,
      "loss": 0.3018,
      "step": 12728
    },
    {
      "epoch": 0.9633602951610615,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0605888411570148e-05,
      "loss": 0.3212,
      "step": 12729
    },
    {
      "epoch": 0.9634359774845088,
      "grad_norm": 0.703125,
      "learning_rate": 1.0604699515499941e-05,
      "loss": 0.2967,
      "step": 12730
    },
    {
      "epoch": 0.9635116598079561,
      "grad_norm": 0.6796875,
      "learning_rate": 1.0603510610851036e-05,
      "loss": 0.2677,
      "step": 12731
    },
    {
      "epoch": 0.9635873421314034,
      "grad_norm": 0.7890625,
      "learning_rate": 1.0602321697640297e-05,
      "loss": 0.3203,
      "step": 12732
    },
    {
      "epoch": 0.9636630244548507,
      "grad_norm": 0.7890625,
      "learning_rate": 1.0601132775884589e-05,
      "loss": 0.3123,
      "step": 12733
    },
    {
      "epoch": 0.9637387067782981,
      "grad_norm": 0.73828125,
      "learning_rate": 1.0599943845600781e-05,
      "loss": 0.2687,
      "step": 12734
    },
    {
      "epoch": 0.9638143891017454,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0598754906805742e-05,
      "loss": 0.3259,
      "step": 12735
    },
    {
      "epoch": 0.9638900714251928,
      "grad_norm": 0.703125,
      "learning_rate": 1.0597565959516337e-05,
      "loss": 0.2876,
      "step": 12736
    },
    {
      "epoch": 0.9639657537486401,
      "grad_norm": 0.80859375,
      "learning_rate": 1.0596377003749431e-05,
      "loss": 0.3565,
      "step": 12737
    },
    {
      "epoch": 0.9640414360720874,
      "grad_norm": 0.80859375,
      "learning_rate": 1.0595188039521897e-05,
      "loss": 0.3636,
      "step": 12738
    },
    {
      "epoch": 0.9641171183955347,
      "grad_norm": 0.765625,
      "learning_rate": 1.0593999066850596e-05,
      "loss": 0.2826,
      "step": 12739
    },
    {
      "epoch": 0.9641928007189821,
      "grad_norm": 0.77734375,
      "learning_rate": 1.0592810085752401e-05,
      "loss": 0.3433,
      "step": 12740
    },
    {
      "epoch": 0.9642684830424294,
      "grad_norm": 0.7109375,
      "learning_rate": 1.059162109624418e-05,
      "loss": 0.2947,
      "step": 12741
    },
    {
      "epoch": 0.9643441653658767,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0590432098342794e-05,
      "loss": 0.3207,
      "step": 12742
    },
    {
      "epoch": 0.9644198476893241,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0589243092065118e-05,
      "loss": 0.3247,
      "step": 12743
    },
    {
      "epoch": 0.9644955300127714,
      "grad_norm": 0.8203125,
      "learning_rate": 1.0588054077428016e-05,
      "loss": 0.3457,
      "step": 12744
    },
    {
      "epoch": 0.9645712123362187,
      "grad_norm": 0.7578125,
      "learning_rate": 1.0586865054448356e-05,
      "loss": 0.2294,
      "step": 12745
    },
    {
      "epoch": 0.964646894659666,
      "grad_norm": 0.73828125,
      "learning_rate": 1.0585676023143014e-05,
      "loss": 0.305,
      "step": 12746
    },
    {
      "epoch": 0.9647225769831134,
      "grad_norm": 0.734375,
      "learning_rate": 1.058448698352885e-05,
      "loss": 0.3322,
      "step": 12747
    },
    {
      "epoch": 0.9647982593065607,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0583297935622732e-05,
      "loss": 0.3181,
      "step": 12748
    },
    {
      "epoch": 0.9648739416300081,
      "grad_norm": 0.6953125,
      "learning_rate": 1.0582108879441534e-05,
      "loss": 0.2807,
      "step": 12749
    },
    {
      "epoch": 0.9649496239534554,
      "grad_norm": 0.6953125,
      "learning_rate": 1.0580919815002126e-05,
      "loss": 0.2664,
      "step": 12750
    },
    {
      "epoch": 0.9650253062769028,
      "grad_norm": 0.703125,
      "learning_rate": 1.0579730742321367e-05,
      "loss": 0.2831,
      "step": 12751
    },
    {
      "epoch": 0.96510098860035,
      "grad_norm": 1.2421875,
      "learning_rate": 1.0578541661416137e-05,
      "loss": 0.3193,
      "step": 12752
    },
    {
      "epoch": 0.9651766709237973,
      "grad_norm": 0.7109375,
      "learning_rate": 1.0577352572303301e-05,
      "loss": 0.2955,
      "step": 12753
    },
    {
      "epoch": 0.9652523532472447,
      "grad_norm": 0.765625,
      "learning_rate": 1.0576163474999725e-05,
      "loss": 0.3182,
      "step": 12754
    },
    {
      "epoch": 0.965328035570692,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0574974369522284e-05,
      "loss": 0.304,
      "step": 12755
    },
    {
      "epoch": 0.9654037178941394,
      "grad_norm": 0.67578125,
      "learning_rate": 1.0573785255887845e-05,
      "loss": 0.2545,
      "step": 12756
    },
    {
      "epoch": 0.9654794002175867,
      "grad_norm": 0.77734375,
      "learning_rate": 1.0572596134113276e-05,
      "loss": 0.326,
      "step": 12757
    },
    {
      "epoch": 0.9655550825410341,
      "grad_norm": 0.703125,
      "learning_rate": 1.0571407004215448e-05,
      "loss": 0.278,
      "step": 12758
    },
    {
      "epoch": 0.9656307648644813,
      "grad_norm": 1.109375,
      "learning_rate": 1.0570217866211232e-05,
      "loss": 0.3624,
      "step": 12759
    },
    {
      "epoch": 0.9657064471879286,
      "grad_norm": 0.7734375,
      "learning_rate": 1.0569028720117496e-05,
      "loss": 0.3149,
      "step": 12760
    },
    {
      "epoch": 0.965782129511376,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0567839565951112e-05,
      "loss": 0.3216,
      "step": 12761
    },
    {
      "epoch": 0.9658578118348233,
      "grad_norm": 0.8046875,
      "learning_rate": 1.056665040372895e-05,
      "loss": 0.3299,
      "step": 12762
    },
    {
      "epoch": 0.9659334941582707,
      "grad_norm": 0.828125,
      "learning_rate": 1.0565461233467877e-05,
      "loss": 0.3495,
      "step": 12763
    },
    {
      "epoch": 0.966009176481718,
      "grad_norm": 0.75390625,
      "learning_rate": 1.056427205518477e-05,
      "loss": 0.3092,
      "step": 12764
    },
    {
      "epoch": 0.9660848588051654,
      "grad_norm": 0.71875,
      "learning_rate": 1.0563082868896492e-05,
      "loss": 0.2901,
      "step": 12765
    },
    {
      "epoch": 0.9661605411286126,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0561893674619918e-05,
      "loss": 0.2825,
      "step": 12766
    },
    {
      "epoch": 0.96623622345206,
      "grad_norm": 0.73046875,
      "learning_rate": 1.0560704472371919e-05,
      "loss": 0.2968,
      "step": 12767
    },
    {
      "epoch": 0.9663119057755073,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0559515262169364e-05,
      "loss": 0.3092,
      "step": 12768
    },
    {
      "epoch": 0.9663875880989546,
      "grad_norm": 0.73828125,
      "learning_rate": 1.0558326044029126e-05,
      "loss": 0.3106,
      "step": 12769
    },
    {
      "epoch": 0.966463270422402,
      "grad_norm": 0.80078125,
      "learning_rate": 1.0557136817968075e-05,
      "loss": 0.3524,
      "step": 12770
    },
    {
      "epoch": 0.9665389527458493,
      "grad_norm": 0.75390625,
      "learning_rate": 1.055594758400308e-05,
      "loss": 0.3119,
      "step": 12771
    },
    {
      "epoch": 0.9666146350692967,
      "grad_norm": 0.8984375,
      "learning_rate": 1.0554758342151015e-05,
      "loss": 0.2884,
      "step": 12772
    },
    {
      "epoch": 0.9666903173927439,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0553569092428752e-05,
      "loss": 0.3311,
      "step": 12773
    },
    {
      "epoch": 0.9667659997161913,
      "grad_norm": 0.68359375,
      "learning_rate": 1.055237983485316e-05,
      "loss": 0.2672,
      "step": 12774
    },
    {
      "epoch": 0.9668416820396386,
      "grad_norm": 0.78515625,
      "learning_rate": 1.0551190569441117e-05,
      "loss": 0.3251,
      "step": 12775
    },
    {
      "epoch": 0.966917364363086,
      "grad_norm": 0.7890625,
      "learning_rate": 1.0550001296209487e-05,
      "loss": 0.3431,
      "step": 12776
    },
    {
      "epoch": 0.9669930466865333,
      "grad_norm": 0.7578125,
      "learning_rate": 1.0548812015175143e-05,
      "loss": 0.3012,
      "step": 12777
    },
    {
      "epoch": 0.9670687290099806,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0547622726354958e-05,
      "loss": 0.2986,
      "step": 12778
    },
    {
      "epoch": 0.967144411333428,
      "grad_norm": 0.7578125,
      "learning_rate": 1.054643342976581e-05,
      "loss": 0.2952,
      "step": 12779
    },
    {
      "epoch": 0.9672200936568752,
      "grad_norm": 0.7578125,
      "learning_rate": 1.054524412542456e-05,
      "loss": 0.303,
      "step": 12780
    },
    {
      "epoch": 0.9672957759803226,
      "grad_norm": 0.83203125,
      "learning_rate": 1.054405481334809e-05,
      "loss": 0.3438,
      "step": 12781
    },
    {
      "epoch": 0.9673714583037699,
      "grad_norm": 0.71875,
      "learning_rate": 1.0542865493553267e-05,
      "loss": 0.2874,
      "step": 12782
    },
    {
      "epoch": 0.9674471406272173,
      "grad_norm": 0.68359375,
      "learning_rate": 1.0541676166056966e-05,
      "loss": 0.2679,
      "step": 12783
    },
    {
      "epoch": 0.9675228229506646,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0540486830876062e-05,
      "loss": 0.2887,
      "step": 12784
    },
    {
      "epoch": 0.967598505274112,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0539297488027422e-05,
      "loss": 0.288,
      "step": 12785
    },
    {
      "epoch": 0.9676741875975593,
      "grad_norm": 0.76171875,
      "learning_rate": 1.053810813752792e-05,
      "loss": 0.3167,
      "step": 12786
    },
    {
      "epoch": 0.9677498699210065,
      "grad_norm": 0.7734375,
      "learning_rate": 1.0536918779394432e-05,
      "loss": 0.3389,
      "step": 12787
    },
    {
      "epoch": 0.9678255522444539,
      "grad_norm": 0.6875,
      "learning_rate": 1.0535729413643834e-05,
      "loss": 0.2537,
      "step": 12788
    },
    {
      "epoch": 0.9679012345679012,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0534540040292988e-05,
      "loss": 0.3176,
      "step": 12789
    },
    {
      "epoch": 0.9679769168913486,
      "grad_norm": 0.75,
      "learning_rate": 1.0533350659358779e-05,
      "loss": 0.3024,
      "step": 12790
    },
    {
      "epoch": 0.9680525992147959,
      "grad_norm": 0.765625,
      "learning_rate": 1.0532161270858074e-05,
      "loss": 0.3302,
      "step": 12791
    },
    {
      "epoch": 0.9681282815382433,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0530971874807746e-05,
      "loss": 0.2915,
      "step": 12792
    },
    {
      "epoch": 0.9682039638616906,
      "grad_norm": 0.8046875,
      "learning_rate": 1.0529782471224677e-05,
      "loss": 0.3352,
      "step": 12793
    },
    {
      "epoch": 0.9682796461851378,
      "grad_norm": 0.8046875,
      "learning_rate": 1.052859306012573e-05,
      "loss": 0.3526,
      "step": 12794
    },
    {
      "epoch": 0.9683553285085852,
      "grad_norm": 0.734375,
      "learning_rate": 1.052740364152778e-05,
      "loss": 0.2947,
      "step": 12795
    },
    {
      "epoch": 0.9684310108320325,
      "grad_norm": 0.75,
      "learning_rate": 1.052621421544771e-05,
      "loss": 0.2756,
      "step": 12796
    },
    {
      "epoch": 0.9685066931554799,
      "grad_norm": 0.75,
      "learning_rate": 1.0525024781902389e-05,
      "loss": 0.3227,
      "step": 12797
    },
    {
      "epoch": 0.9685823754789272,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0523835340908686e-05,
      "loss": 0.3085,
      "step": 12798
    },
    {
      "epoch": 0.9686580578023746,
      "grad_norm": 0.828125,
      "learning_rate": 1.0522645892483483e-05,
      "loss": 0.3366,
      "step": 12799
    },
    {
      "epoch": 0.9687337401258218,
      "grad_norm": 0.79296875,
      "learning_rate": 1.0521456436643648e-05,
      "loss": 0.3238,
      "step": 12800
    },
    {
      "epoch": 0.9688094224492692,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0520266973406059e-05,
      "loss": 0.3069,
      "step": 12801
    },
    {
      "epoch": 0.9688851047727165,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0519077502787596e-05,
      "loss": 0.2814,
      "step": 12802
    },
    {
      "epoch": 0.9689607870961638,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0517888024805123e-05,
      "loss": 0.3317,
      "step": 12803
    },
    {
      "epoch": 0.9690364694196112,
      "grad_norm": 0.6875,
      "learning_rate": 1.0516698539475518e-05,
      "loss": 0.254,
      "step": 12804
    },
    {
      "epoch": 0.9691121517430585,
      "grad_norm": 0.859375,
      "learning_rate": 1.051550904681566e-05,
      "loss": 0.3308,
      "step": 12805
    },
    {
      "epoch": 0.9691878340665059,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0514319546842422e-05,
      "loss": 0.3027,
      "step": 12806
    },
    {
      "epoch": 0.9692635163899531,
      "grad_norm": 0.78515625,
      "learning_rate": 1.0513130039572676e-05,
      "loss": 0.3156,
      "step": 12807
    },
    {
      "epoch": 0.9693391987134005,
      "grad_norm": 0.7734375,
      "learning_rate": 1.0511940525023302e-05,
      "loss": 0.3316,
      "step": 12808
    },
    {
      "epoch": 0.9694148810368478,
      "grad_norm": 0.734375,
      "learning_rate": 1.0510751003211175e-05,
      "loss": 0.3229,
      "step": 12809
    },
    {
      "epoch": 0.9694905633602952,
      "grad_norm": 0.71875,
      "learning_rate": 1.0509561474153162e-05,
      "loss": 0.2972,
      "step": 12810
    },
    {
      "epoch": 0.9695662456837425,
      "grad_norm": 0.734375,
      "learning_rate": 1.0508371937866153e-05,
      "loss": 0.3042,
      "step": 12811
    },
    {
      "epoch": 0.9696419280071898,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0507182394367013e-05,
      "loss": 0.2618,
      "step": 12812
    },
    {
      "epoch": 0.9697176103306372,
      "grad_norm": 0.79296875,
      "learning_rate": 1.0505992843672617e-05,
      "loss": 0.3349,
      "step": 12813
    },
    {
      "epoch": 0.9697932926540844,
      "grad_norm": 0.78515625,
      "learning_rate": 1.0504803285799848e-05,
      "loss": 0.3218,
      "step": 12814
    },
    {
      "epoch": 0.9698689749775318,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0503613720765576e-05,
      "loss": 0.2883,
      "step": 12815
    },
    {
      "epoch": 0.9699446573009791,
      "grad_norm": 0.79296875,
      "learning_rate": 1.050242414858668e-05,
      "loss": 0.3627,
      "step": 12816
    },
    {
      "epoch": 0.9700203396244265,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0501234569280037e-05,
      "loss": 0.2987,
      "step": 12817
    },
    {
      "epoch": 0.9700960219478738,
      "grad_norm": 0.82421875,
      "learning_rate": 1.0500044982862519e-05,
      "loss": 0.3774,
      "step": 12818
    },
    {
      "epoch": 0.9701717042713212,
      "grad_norm": 0.75,
      "learning_rate": 1.0498855389351004e-05,
      "loss": 0.308,
      "step": 12819
    },
    {
      "epoch": 0.9702473865947685,
      "grad_norm": 0.703125,
      "learning_rate": 1.049766578876237e-05,
      "loss": 0.2784,
      "step": 12820
    },
    {
      "epoch": 0.9703230689182157,
      "grad_norm": 0.77734375,
      "learning_rate": 1.0496476181113492e-05,
      "loss": 0.328,
      "step": 12821
    },
    {
      "epoch": 0.9703987512416631,
      "grad_norm": 0.82421875,
      "learning_rate": 1.049528656642125e-05,
      "loss": 0.3385,
      "step": 12822
    },
    {
      "epoch": 0.9704744335651104,
      "grad_norm": 0.734375,
      "learning_rate": 1.0494096944702518e-05,
      "loss": 0.2964,
      "step": 12823
    },
    {
      "epoch": 0.9705501158885578,
      "grad_norm": 0.80078125,
      "learning_rate": 1.0492907315974172e-05,
      "loss": 0.3458,
      "step": 12824
    },
    {
      "epoch": 0.9706257982120051,
      "grad_norm": 0.7890625,
      "learning_rate": 1.0491717680253091e-05,
      "loss": 0.3264,
      "step": 12825
    },
    {
      "epoch": 0.9707014805354525,
      "grad_norm": 0.73828125,
      "learning_rate": 1.0490528037556149e-05,
      "loss": 0.2922,
      "step": 12826
    },
    {
      "epoch": 0.9707771628588998,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0489338387900222e-05,
      "loss": 0.3046,
      "step": 12827
    },
    {
      "epoch": 0.970852845182347,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0488148731302197e-05,
      "loss": 0.3006,
      "step": 12828
    },
    {
      "epoch": 0.9709285275057944,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0486959067778942e-05,
      "loss": 0.2989,
      "step": 12829
    },
    {
      "epoch": 0.9710042098292417,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0485769397347335e-05,
      "loss": 0.3063,
      "step": 12830
    },
    {
      "epoch": 0.9710798921526891,
      "grad_norm": 0.71875,
      "learning_rate": 1.0484579720024257e-05,
      "loss": 0.2592,
      "step": 12831
    },
    {
      "epoch": 0.9711555744761364,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0483390035826586e-05,
      "loss": 0.2663,
      "step": 12832
    },
    {
      "epoch": 0.9712312567995838,
      "grad_norm": 0.734375,
      "learning_rate": 1.0482200344771196e-05,
      "loss": 0.2835,
      "step": 12833
    },
    {
      "epoch": 0.9713069391230311,
      "grad_norm": 0.80078125,
      "learning_rate": 1.0481010646874963e-05,
      "loss": 0.349,
      "step": 12834
    },
    {
      "epoch": 0.9713826214464784,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0479820942154775e-05,
      "loss": 0.3024,
      "step": 12835
    },
    {
      "epoch": 0.9714583037699257,
      "grad_norm": 0.6953125,
      "learning_rate": 1.0478631230627498e-05,
      "loss": 0.2633,
      "step": 12836
    },
    {
      "epoch": 0.971533986093373,
      "grad_norm": 0.70703125,
      "learning_rate": 1.0477441512310019e-05,
      "loss": 0.299,
      "step": 12837
    },
    {
      "epoch": 0.9716096684168204,
      "grad_norm": 0.6640625,
      "learning_rate": 1.0476251787219212e-05,
      "loss": 0.2612,
      "step": 12838
    },
    {
      "epoch": 0.9716853507402677,
      "grad_norm": 0.71875,
      "learning_rate": 1.0475062055371957e-05,
      "loss": 0.2713,
      "step": 12839
    },
    {
      "epoch": 0.9717610330637151,
      "grad_norm": 0.76171875,
      "learning_rate": 1.047387231678513e-05,
      "loss": 0.283,
      "step": 12840
    },
    {
      "epoch": 0.9718367153871624,
      "grad_norm": 0.7734375,
      "learning_rate": 1.0472682571475612e-05,
      "loss": 0.3366,
      "step": 12841
    },
    {
      "epoch": 0.9719123977106097,
      "grad_norm": 0.8359375,
      "learning_rate": 1.0471492819460278e-05,
      "loss": 0.3825,
      "step": 12842
    },
    {
      "epoch": 0.971988080034057,
      "grad_norm": 0.7109375,
      "learning_rate": 1.0470303060756013e-05,
      "loss": 0.2799,
      "step": 12843
    },
    {
      "epoch": 0.9720637623575044,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0469113295379691e-05,
      "loss": 0.3072,
      "step": 12844
    },
    {
      "epoch": 0.9721394446809517,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0467923523348192e-05,
      "loss": 0.2984,
      "step": 12845
    },
    {
      "epoch": 0.972215127004399,
      "grad_norm": 0.78125,
      "learning_rate": 1.0466733744678392e-05,
      "loss": 0.3461,
      "step": 12846
    },
    {
      "epoch": 0.9722908093278464,
      "grad_norm": 0.796875,
      "learning_rate": 1.0465543959387176e-05,
      "loss": 0.3444,
      "step": 12847
    },
    {
      "epoch": 0.9723664916512937,
      "grad_norm": 0.75,
      "learning_rate": 1.0464354167491418e-05,
      "loss": 0.2955,
      "step": 12848
    },
    {
      "epoch": 0.972442173974741,
      "grad_norm": 0.7734375,
      "learning_rate": 1.0463164369008003e-05,
      "loss": 0.3429,
      "step": 12849
    },
    {
      "epoch": 0.9725178562981883,
      "grad_norm": 0.765625,
      "learning_rate": 1.0461974563953802e-05,
      "loss": 0.3107,
      "step": 12850
    },
    {
      "epoch": 0.9725935386216357,
      "grad_norm": 0.7578125,
      "learning_rate": 1.04607847523457e-05,
      "loss": 0.3098,
      "step": 12851
    },
    {
      "epoch": 0.972669220945083,
      "grad_norm": 0.765625,
      "learning_rate": 1.0459594934200578e-05,
      "loss": 0.3086,
      "step": 12852
    },
    {
      "epoch": 0.9727449032685304,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0458405109535314e-05,
      "loss": 0.3056,
      "step": 12853
    },
    {
      "epoch": 0.9728205855919777,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0457215278366783e-05,
      "loss": 0.3202,
      "step": 12854
    },
    {
      "epoch": 0.972896267915425,
      "grad_norm": 0.7421875,
      "learning_rate": 1.045602544071187e-05,
      "loss": 0.304,
      "step": 12855
    },
    {
      "epoch": 0.9729719502388723,
      "grad_norm": 0.75,
      "learning_rate": 1.0454835596587454e-05,
      "loss": 0.3252,
      "step": 12856
    },
    {
      "epoch": 0.9730476325623196,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0453645746010412e-05,
      "loss": 0.2597,
      "step": 12857
    },
    {
      "epoch": 0.973123314885767,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0452455888997633e-05,
      "loss": 0.2862,
      "step": 12858
    },
    {
      "epoch": 0.9731989972092143,
      "grad_norm": 0.80078125,
      "learning_rate": 1.0451266025565987e-05,
      "loss": 0.3646,
      "step": 12859
    },
    {
      "epoch": 0.9732746795326617,
      "grad_norm": 0.73828125,
      "learning_rate": 1.0450076155732357e-05,
      "loss": 0.2974,
      "step": 12860
    },
    {
      "epoch": 0.973350361856109,
      "grad_norm": 0.8046875,
      "learning_rate": 1.0448886279513623e-05,
      "loss": 0.3376,
      "step": 12861
    },
    {
      "epoch": 0.9734260441795564,
      "grad_norm": 0.7421875,
      "learning_rate": 1.044769639692667e-05,
      "loss": 0.3044,
      "step": 12862
    },
    {
      "epoch": 0.9735017265030036,
      "grad_norm": 0.73046875,
      "learning_rate": 1.0446506507988375e-05,
      "loss": 0.2987,
      "step": 12863
    },
    {
      "epoch": 0.9735774088264509,
      "grad_norm": 0.828125,
      "learning_rate": 1.0445316612715616e-05,
      "loss": 0.3817,
      "step": 12864
    },
    {
      "epoch": 0.9736530911498983,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0444126711125283e-05,
      "loss": 0.2774,
      "step": 12865
    },
    {
      "epoch": 0.9737287734733456,
      "grad_norm": 1.0703125,
      "learning_rate": 1.0442936803234242e-05,
      "loss": 0.3147,
      "step": 12866
    },
    {
      "epoch": 0.973804455796793,
      "grad_norm": 0.70703125,
      "learning_rate": 1.044174688905939e-05,
      "loss": 0.2769,
      "step": 12867
    },
    {
      "epoch": 0.9738801381202403,
      "grad_norm": 0.796875,
      "learning_rate": 1.0440556968617597e-05,
      "loss": 0.3578,
      "step": 12868
    },
    {
      "epoch": 0.9739558204436877,
      "grad_norm": 0.765625,
      "learning_rate": 1.0439367041925746e-05,
      "loss": 0.2992,
      "step": 12869
    },
    {
      "epoch": 0.9740315027671349,
      "grad_norm": 0.77734375,
      "learning_rate": 1.043817710900072e-05,
      "loss": 0.3426,
      "step": 12870
    },
    {
      "epoch": 0.9741071850905823,
      "grad_norm": 0.765625,
      "learning_rate": 1.0436987169859401e-05,
      "loss": 0.3117,
      "step": 12871
    },
    {
      "epoch": 0.9741828674140296,
      "grad_norm": 0.78125,
      "learning_rate": 1.0435797224518669e-05,
      "loss": 0.3442,
      "step": 12872
    },
    {
      "epoch": 0.9742585497374769,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0434607272995404e-05,
      "loss": 0.2998,
      "step": 12873
    },
    {
      "epoch": 0.9743342320609243,
      "grad_norm": 0.828125,
      "learning_rate": 1.0433417315306494e-05,
      "loss": 0.3622,
      "step": 12874
    },
    {
      "epoch": 0.9744099143843716,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0432227351468809e-05,
      "loss": 0.3134,
      "step": 12875
    },
    {
      "epoch": 0.974485596707819,
      "grad_norm": 0.7734375,
      "learning_rate": 1.043103738149924e-05,
      "loss": 0.3327,
      "step": 12876
    },
    {
      "epoch": 0.9745612790312662,
      "grad_norm": 0.8046875,
      "learning_rate": 1.0429847405414666e-05,
      "loss": 0.3316,
      "step": 12877
    },
    {
      "epoch": 0.9746369613547136,
      "grad_norm": 0.78515625,
      "learning_rate": 1.0428657423231968e-05,
      "loss": 0.3057,
      "step": 12878
    },
    {
      "epoch": 0.9747126436781609,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0427467434968033e-05,
      "loss": 0.285,
      "step": 12879
    },
    {
      "epoch": 0.9747883260016083,
      "grad_norm": 0.734375,
      "learning_rate": 1.0426277440639737e-05,
      "loss": 0.3015,
      "step": 12880
    },
    {
      "epoch": 0.9748640083250556,
      "grad_norm": 0.73046875,
      "learning_rate": 1.042508744026396e-05,
      "loss": 0.2754,
      "step": 12881
    },
    {
      "epoch": 0.9749396906485029,
      "grad_norm": 0.78515625,
      "learning_rate": 1.0423897433857593e-05,
      "loss": 0.3247,
      "step": 12882
    },
    {
      "epoch": 0.9750153729719503,
      "grad_norm": 0.7578125,
      "learning_rate": 1.042270742143751e-05,
      "loss": 0.3285,
      "step": 12883
    },
    {
      "epoch": 0.9750910552953975,
      "grad_norm": 0.7578125,
      "learning_rate": 1.0421517403020597e-05,
      "loss": 0.3169,
      "step": 12884
    },
    {
      "epoch": 0.9751667376188449,
      "grad_norm": 0.7421875,
      "learning_rate": 1.042032737862374e-05,
      "loss": 0.3124,
      "step": 12885
    },
    {
      "epoch": 0.9752424199422922,
      "grad_norm": 0.8359375,
      "learning_rate": 1.0419137348263815e-05,
      "loss": 0.3428,
      "step": 12886
    },
    {
      "epoch": 0.9753181022657396,
      "grad_norm": 0.81640625,
      "learning_rate": 1.0417947311957707e-05,
      "loss": 0.3613,
      "step": 12887
    },
    {
      "epoch": 0.9753937845891869,
      "grad_norm": 0.734375,
      "learning_rate": 1.0416757269722301e-05,
      "loss": 0.3009,
      "step": 12888
    },
    {
      "epoch": 0.9754694669126343,
      "grad_norm": 0.875,
      "learning_rate": 1.0415567221574478e-05,
      "loss": 0.4172,
      "step": 12889
    },
    {
      "epoch": 0.9755451492360816,
      "grad_norm": 0.79296875,
      "learning_rate": 1.0414377167531118e-05,
      "loss": 0.3362,
      "step": 12890
    },
    {
      "epoch": 0.9756208315595288,
      "grad_norm": 0.7421875,
      "learning_rate": 1.041318710760911e-05,
      "loss": 0.3093,
      "step": 12891
    },
    {
      "epoch": 0.9756965138829762,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0411997041825333e-05,
      "loss": 0.3018,
      "step": 12892
    },
    {
      "epoch": 0.9757721962064235,
      "grad_norm": 0.765625,
      "learning_rate": 1.041080697019667e-05,
      "loss": 0.318,
      "step": 12893
    },
    {
      "epoch": 0.9758478785298709,
      "grad_norm": 0.8046875,
      "learning_rate": 1.0409616892740007e-05,
      "loss": 0.3261,
      "step": 12894
    },
    {
      "epoch": 0.9759235608533182,
      "grad_norm": 0.77734375,
      "learning_rate": 1.0408426809472223e-05,
      "loss": 0.3063,
      "step": 12895
    },
    {
      "epoch": 0.9759992431767656,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0407236720410205e-05,
      "loss": 0.3209,
      "step": 12896
    },
    {
      "epoch": 0.9760749255002129,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0406046625570836e-05,
      "loss": 0.2897,
      "step": 12897
    },
    {
      "epoch": 0.9761506078236601,
      "grad_norm": 0.75,
      "learning_rate": 1.0404856524970996e-05,
      "loss": 0.3239,
      "step": 12898
    },
    {
      "epoch": 0.9762262901471075,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0403666418627574e-05,
      "loss": 0.3233,
      "step": 12899
    },
    {
      "epoch": 0.9763019724705548,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0402476306557451e-05,
      "loss": 0.3491,
      "step": 12900
    },
    {
      "epoch": 0.9763776547940022,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0401286188777511e-05,
      "loss": 0.2697,
      "step": 12901
    },
    {
      "epoch": 0.9764533371174495,
      "grad_norm": 0.80859375,
      "learning_rate": 1.0400096065304637e-05,
      "loss": 0.3761,
      "step": 12902
    },
    {
      "epoch": 0.9765290194408969,
      "grad_norm": 0.75,
      "learning_rate": 1.0398905936155714e-05,
      "loss": 0.3137,
      "step": 12903
    },
    {
      "epoch": 0.9766047017643442,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0397715801347624e-05,
      "loss": 0.3154,
      "step": 12904
    },
    {
      "epoch": 0.9766803840877915,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0396525660897257e-05,
      "loss": 0.31,
      "step": 12905
    },
    {
      "epoch": 0.9767560664112388,
      "grad_norm": 0.765625,
      "learning_rate": 1.039533551482149e-05,
      "loss": 0.311,
      "step": 12906
    },
    {
      "epoch": 0.9768317487346861,
      "grad_norm": 0.78125,
      "learning_rate": 1.039414536313721e-05,
      "loss": 0.3413,
      "step": 12907
    },
    {
      "epoch": 0.9769074310581335,
      "grad_norm": 0.765625,
      "learning_rate": 1.03929552058613e-05,
      "loss": 0.3316,
      "step": 12908
    },
    {
      "epoch": 0.9769831133815808,
      "grad_norm": 0.68359375,
      "learning_rate": 1.0391765043010653e-05,
      "loss": 0.2555,
      "step": 12909
    },
    {
      "epoch": 0.9770587957050282,
      "grad_norm": 0.73046875,
      "learning_rate": 1.0390574874602142e-05,
      "loss": 0.3098,
      "step": 12910
    },
    {
      "epoch": 0.9771344780284755,
      "grad_norm": 0.796875,
      "learning_rate": 1.0389384700652656e-05,
      "loss": 0.3237,
      "step": 12911
    },
    {
      "epoch": 0.9772101603519228,
      "grad_norm": 0.78125,
      "learning_rate": 1.038819452117908e-05,
      "loss": 0.3383,
      "step": 12912
    },
    {
      "epoch": 0.9772858426753701,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0387004336198297e-05,
      "loss": 0.2952,
      "step": 12913
    },
    {
      "epoch": 0.9773615249988175,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0385814145727197e-05,
      "loss": 0.3011,
      "step": 12914
    },
    {
      "epoch": 0.9774372073222648,
      "grad_norm": 0.76171875,
      "learning_rate": 1.038462394978266e-05,
      "loss": 0.3133,
      "step": 12915
    },
    {
      "epoch": 0.9775128896457121,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0383433748381571e-05,
      "loss": 0.3422,
      "step": 12916
    },
    {
      "epoch": 0.9775885719691595,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0382243541540816e-05,
      "loss": 0.3046,
      "step": 12917
    },
    {
      "epoch": 0.9776642542926068,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0381053329277281e-05,
      "loss": 0.2712,
      "step": 12918
    },
    {
      "epoch": 0.9777399366160541,
      "grad_norm": 0.7421875,
      "learning_rate": 1.037986311160785e-05,
      "loss": 0.3119,
      "step": 12919
    },
    {
      "epoch": 0.9778156189395014,
      "grad_norm": 1.1328125,
      "learning_rate": 1.0378672888549413e-05,
      "loss": 0.3272,
      "step": 12920
    },
    {
      "epoch": 0.9778913012629488,
      "grad_norm": 0.8125,
      "learning_rate": 1.0377482660118847e-05,
      "loss": 0.393,
      "step": 12921
    },
    {
      "epoch": 0.9779669835863961,
      "grad_norm": 0.7734375,
      "learning_rate": 1.0376292426333042e-05,
      "loss": 0.3266,
      "step": 12922
    },
    {
      "epoch": 0.9780426659098435,
      "grad_norm": 0.83984375,
      "learning_rate": 1.0375102187208885e-05,
      "loss": 0.3601,
      "step": 12923
    },
    {
      "epoch": 0.9781183482332908,
      "grad_norm": 0.71484375,
      "learning_rate": 1.037391194276326e-05,
      "loss": 0.2796,
      "step": 12924
    },
    {
      "epoch": 0.978194030556738,
      "grad_norm": 0.91015625,
      "learning_rate": 1.0372721693013047e-05,
      "loss": 0.356,
      "step": 12925
    },
    {
      "epoch": 0.9782697128801854,
      "grad_norm": 0.70703125,
      "learning_rate": 1.0371531437975142e-05,
      "loss": 0.2941,
      "step": 12926
    },
    {
      "epoch": 0.9783453952036327,
      "grad_norm": 0.73828125,
      "learning_rate": 1.0370341177666428e-05,
      "loss": 0.2957,
      "step": 12927
    },
    {
      "epoch": 0.9784210775270801,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0369150912103784e-05,
      "loss": 0.3042,
      "step": 12928
    },
    {
      "epoch": 0.9784967598505274,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0367960641304105e-05,
      "loss": 0.3262,
      "step": 12929
    },
    {
      "epoch": 0.9785724421739748,
      "grad_norm": 0.70703125,
      "learning_rate": 1.0366770365284271e-05,
      "loss": 0.2793,
      "step": 12930
    },
    {
      "epoch": 0.9786481244974221,
      "grad_norm": 0.78515625,
      "learning_rate": 1.0365580084061166e-05,
      "loss": 0.3262,
      "step": 12931
    },
    {
      "epoch": 0.9787238068208693,
      "grad_norm": 0.77734375,
      "learning_rate": 1.0364389797651686e-05,
      "loss": 0.3107,
      "step": 12932
    },
    {
      "epoch": 0.9787994891443167,
      "grad_norm": 0.6953125,
      "learning_rate": 1.0363199506072709e-05,
      "loss": 0.2696,
      "step": 12933
    },
    {
      "epoch": 0.978875171467764,
      "grad_norm": 0.8125,
      "learning_rate": 1.0362009209341125e-05,
      "loss": 0.3487,
      "step": 12934
    },
    {
      "epoch": 0.9789508537912114,
      "grad_norm": 1.0390625,
      "learning_rate": 1.0360818907473817e-05,
      "loss": 0.3875,
      "step": 12935
    },
    {
      "epoch": 0.9790265361146587,
      "grad_norm": 0.65625,
      "learning_rate": 1.0359628600487677e-05,
      "loss": 0.2513,
      "step": 12936
    },
    {
      "epoch": 0.9791022184381061,
      "grad_norm": 0.7109375,
      "learning_rate": 1.0358438288399587e-05,
      "loss": 0.2553,
      "step": 12937
    },
    {
      "epoch": 0.9791779007615534,
      "grad_norm": 0.703125,
      "learning_rate": 1.0357247971226432e-05,
      "loss": 0.2601,
      "step": 12938
    },
    {
      "epoch": 0.9792535830850007,
      "grad_norm": 0.71875,
      "learning_rate": 1.0356057648985102e-05,
      "loss": 0.3096,
      "step": 12939
    },
    {
      "epoch": 0.979329265408448,
      "grad_norm": 0.6953125,
      "learning_rate": 1.0354867321692486e-05,
      "loss": 0.2446,
      "step": 12940
    },
    {
      "epoch": 0.9794049477318953,
      "grad_norm": 0.7578125,
      "learning_rate": 1.0353676989365467e-05,
      "loss": 0.3112,
      "step": 12941
    },
    {
      "epoch": 0.9794806300553427,
      "grad_norm": 0.78515625,
      "learning_rate": 1.0352486652020933e-05,
      "loss": 0.3431,
      "step": 12942
    },
    {
      "epoch": 0.97955631237879,
      "grad_norm": 0.84375,
      "learning_rate": 1.0351296309675768e-05,
      "loss": 0.3722,
      "step": 12943
    },
    {
      "epoch": 0.9796319947022374,
      "grad_norm": 0.9296875,
      "learning_rate": 1.0350105962346866e-05,
      "loss": 0.3156,
      "step": 12944
    },
    {
      "epoch": 0.9797076770256847,
      "grad_norm": 0.73828125,
      "learning_rate": 1.0348915610051107e-05,
      "loss": 0.3041,
      "step": 12945
    },
    {
      "epoch": 0.979783359349132,
      "grad_norm": 0.73828125,
      "learning_rate": 1.0347725252805383e-05,
      "loss": 0.2989,
      "step": 12946
    },
    {
      "epoch": 0.9798590416725793,
      "grad_norm": 0.7109375,
      "learning_rate": 1.0346534890626583e-05,
      "loss": 0.3258,
      "step": 12947
    },
    {
      "epoch": 0.9799347239960267,
      "grad_norm": 0.71875,
      "learning_rate": 1.0345344523531584e-05,
      "loss": 0.2859,
      "step": 12948
    },
    {
      "epoch": 0.980010406319474,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0344154151537285e-05,
      "loss": 0.3094,
      "step": 12949
    },
    {
      "epoch": 0.9800860886429213,
      "grad_norm": 0.75,
      "learning_rate": 1.0342963774660566e-05,
      "loss": 0.2891,
      "step": 12950
    },
    {
      "epoch": 0.9801617709663687,
      "grad_norm": 0.80078125,
      "learning_rate": 1.0341773392918319e-05,
      "loss": 0.3483,
      "step": 12951
    },
    {
      "epoch": 0.980237453289816,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0340583006327432e-05,
      "loss": 0.3259,
      "step": 12952
    },
    {
      "epoch": 0.9803131356132633,
      "grad_norm": 1.0234375,
      "learning_rate": 1.0339392614904789e-05,
      "loss": 0.3634,
      "step": 12953
    },
    {
      "epoch": 0.9803888179367106,
      "grad_norm": 0.8046875,
      "learning_rate": 1.0338202218667275e-05,
      "loss": 0.3297,
      "step": 12954
    },
    {
      "epoch": 0.980464500260158,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0337011817631786e-05,
      "loss": 0.2881,
      "step": 12955
    },
    {
      "epoch": 0.9805401825836053,
      "grad_norm": 0.703125,
      "learning_rate": 1.0335821411815207e-05,
      "loss": 0.2962,
      "step": 12956
    },
    {
      "epoch": 0.9806158649070527,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0334631001234423e-05,
      "loss": 0.3017,
      "step": 12957
    },
    {
      "epoch": 0.9806915472305,
      "grad_norm": 0.71875,
      "learning_rate": 1.0333440585906326e-05,
      "loss": 0.313,
      "step": 12958
    },
    {
      "epoch": 0.9807672295539474,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0332250165847802e-05,
      "loss": 0.3047,
      "step": 12959
    },
    {
      "epoch": 0.9808429118773946,
      "grad_norm": 0.6953125,
      "learning_rate": 1.0331059741075735e-05,
      "loss": 0.2728,
      "step": 12960
    },
    {
      "epoch": 0.9809185942008419,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0329869311607024e-05,
      "loss": 0.2944,
      "step": 12961
    },
    {
      "epoch": 0.9809942765242893,
      "grad_norm": 0.84375,
      "learning_rate": 1.0328678877458547e-05,
      "loss": 0.3402,
      "step": 12962
    },
    {
      "epoch": 0.9810699588477366,
      "grad_norm": 0.77734375,
      "learning_rate": 1.0327488438647195e-05,
      "loss": 0.3062,
      "step": 12963
    },
    {
      "epoch": 0.981145641171184,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0326297995189858e-05,
      "loss": 0.3226,
      "step": 12964
    },
    {
      "epoch": 0.9812213234946313,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0325107547103429e-05,
      "loss": 0.3139,
      "step": 12965
    },
    {
      "epoch": 0.9812970058180787,
      "grad_norm": 0.77734375,
      "learning_rate": 1.0323917094404784e-05,
      "loss": 0.3133,
      "step": 12966
    },
    {
      "epoch": 0.9813726881415259,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0322726637110822e-05,
      "loss": 0.3473,
      "step": 12967
    },
    {
      "epoch": 0.9814483704649732,
      "grad_norm": 0.765625,
      "learning_rate": 1.0321536175238431e-05,
      "loss": 0.3293,
      "step": 12968
    },
    {
      "epoch": 0.9815240527884206,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0320345708804494e-05,
      "loss": 0.3132,
      "step": 12969
    },
    {
      "epoch": 0.9815997351118679,
      "grad_norm": 0.81640625,
      "learning_rate": 1.031915523782591e-05,
      "loss": 0.3531,
      "step": 12970
    },
    {
      "epoch": 0.9816754174353153,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0317964762319554e-05,
      "loss": 0.3126,
      "step": 12971
    },
    {
      "epoch": 0.9817510997587626,
      "grad_norm": 0.68359375,
      "learning_rate": 1.0316774282302324e-05,
      "loss": 0.2806,
      "step": 12972
    },
    {
      "epoch": 0.98182678208221,
      "grad_norm": 0.70703125,
      "learning_rate": 1.0315583797791109e-05,
      "loss": 0.2953,
      "step": 12973
    },
    {
      "epoch": 0.9819024644056572,
      "grad_norm": 0.734375,
      "learning_rate": 1.0314393308802795e-05,
      "loss": 0.3079,
      "step": 12974
    },
    {
      "epoch": 0.9819781467291046,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0313202815354273e-05,
      "loss": 0.2894,
      "step": 12975
    },
    {
      "epoch": 0.9820538290525519,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0312012317462433e-05,
      "loss": 0.2717,
      "step": 12976
    },
    {
      "epoch": 0.9821295113759992,
      "grad_norm": 0.7578125,
      "learning_rate": 1.0310821815144161e-05,
      "loss": 0.32,
      "step": 12977
    },
    {
      "epoch": 0.9822051936994466,
      "grad_norm": 0.65234375,
      "learning_rate": 1.0309631308416346e-05,
      "loss": 0.2574,
      "step": 12978
    },
    {
      "epoch": 0.9822808760228939,
      "grad_norm": 0.703125,
      "learning_rate": 1.0308440797295885e-05,
      "loss": 0.2698,
      "step": 12979
    },
    {
      "epoch": 0.9823565583463413,
      "grad_norm": 0.703125,
      "learning_rate": 1.030725028179966e-05,
      "loss": 0.2702,
      "step": 12980
    },
    {
      "epoch": 0.9824322406697885,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0306059761944558e-05,
      "loss": 0.315,
      "step": 12981
    },
    {
      "epoch": 0.9825079229932359,
      "grad_norm": 0.78125,
      "learning_rate": 1.0304869237747476e-05,
      "loss": 0.3288,
      "step": 12982
    },
    {
      "epoch": 0.9825836053166832,
      "grad_norm": 0.79296875,
      "learning_rate": 1.0303678709225302e-05,
      "loss": 0.3392,
      "step": 12983
    },
    {
      "epoch": 0.9826592876401306,
      "grad_norm": 0.69921875,
      "learning_rate": 1.0302488176394925e-05,
      "loss": 0.2967,
      "step": 12984
    },
    {
      "epoch": 0.9827349699635779,
      "grad_norm": 0.75390625,
      "learning_rate": 1.030129763927323e-05,
      "loss": 0.2618,
      "step": 12985
    },
    {
      "epoch": 0.9828106522870252,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0300107097877114e-05,
      "loss": 0.2904,
      "step": 12986
    },
    {
      "epoch": 0.9828863346104726,
      "grad_norm": 0.73046875,
      "learning_rate": 1.0298916552223463e-05,
      "loss": 0.2872,
      "step": 12987
    },
    {
      "epoch": 0.9829620169339198,
      "grad_norm": 0.77734375,
      "learning_rate": 1.0297726002329169e-05,
      "loss": 0.3187,
      "step": 12988
    },
    {
      "epoch": 0.9830376992573672,
      "grad_norm": 0.75,
      "learning_rate": 1.029653544821112e-05,
      "loss": 0.2949,
      "step": 12989
    },
    {
      "epoch": 0.9831133815808145,
      "grad_norm": 0.75,
      "learning_rate": 1.0295344889886207e-05,
      "loss": 0.3114,
      "step": 12990
    },
    {
      "epoch": 0.9831890639042619,
      "grad_norm": 0.796875,
      "learning_rate": 1.029415432737132e-05,
      "loss": 0.3435,
      "step": 12991
    },
    {
      "epoch": 0.9832647462277092,
      "grad_norm": 0.78515625,
      "learning_rate": 1.0292963760683347e-05,
      "loss": 0.3375,
      "step": 12992
    },
    {
      "epoch": 0.9833404285511566,
      "grad_norm": 0.75,
      "learning_rate": 1.0291773189839182e-05,
      "loss": 0.3272,
      "step": 12993
    },
    {
      "epoch": 0.9834161108746039,
      "grad_norm": 0.69140625,
      "learning_rate": 1.0290582614855716e-05,
      "loss": 0.2548,
      "step": 12994
    },
    {
      "epoch": 0.9834917931980511,
      "grad_norm": 0.75,
      "learning_rate": 1.0289392035749832e-05,
      "loss": 0.309,
      "step": 12995
    },
    {
      "epoch": 0.9835674755214985,
      "grad_norm": 0.7734375,
      "learning_rate": 1.0288201452538428e-05,
      "loss": 0.3447,
      "step": 12996
    },
    {
      "epoch": 0.9836431578449458,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0287010865238391e-05,
      "loss": 0.3151,
      "step": 12997
    },
    {
      "epoch": 0.9837188401683932,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0285820273866613e-05,
      "loss": 0.2857,
      "step": 12998
    },
    {
      "epoch": 0.9837945224918405,
      "grad_norm": 0.83203125,
      "learning_rate": 1.0284629678439986e-05,
      "loss": 0.3544,
      "step": 12999
    },
    {
      "epoch": 0.9838702048152879,
      "grad_norm": 0.81640625,
      "learning_rate": 1.0283439078975398e-05,
      "loss": 0.3308,
      "step": 13000
    },
    {
      "epoch": 0.9839458871387352,
      "grad_norm": 0.765625,
      "learning_rate": 1.0282248475489735e-05,
      "loss": 0.3194,
      "step": 13001
    },
    {
      "epoch": 0.9840215694621824,
      "grad_norm": 0.73828125,
      "learning_rate": 1.0281057867999897e-05,
      "loss": 0.2865,
      "step": 13002
    },
    {
      "epoch": 0.9840972517856298,
      "grad_norm": 0.765625,
      "learning_rate": 1.0279867256522774e-05,
      "loss": 0.3162,
      "step": 13003
    },
    {
      "epoch": 0.9841729341090771,
      "grad_norm": 0.828125,
      "learning_rate": 1.027867664107525e-05,
      "loss": 0.3318,
      "step": 13004
    },
    {
      "epoch": 0.9842486164325245,
      "grad_norm": 0.7578125,
      "learning_rate": 1.0277486021674219e-05,
      "loss": 0.3297,
      "step": 13005
    },
    {
      "epoch": 0.9843242987559718,
      "grad_norm": 0.69921875,
      "learning_rate": 1.0276295398336573e-05,
      "loss": 0.2588,
      "step": 13006
    },
    {
      "epoch": 0.9843999810794192,
      "grad_norm": 0.7890625,
      "learning_rate": 1.0275104771079203e-05,
      "loss": 0.3316,
      "step": 13007
    },
    {
      "epoch": 0.9844756634028665,
      "grad_norm": 0.7578125,
      "learning_rate": 1.0273914139919e-05,
      "loss": 0.3402,
      "step": 13008
    },
    {
      "epoch": 0.9845513457263138,
      "grad_norm": 0.78125,
      "learning_rate": 1.0272723504872857e-05,
      "loss": 0.3523,
      "step": 13009
    },
    {
      "epoch": 0.9846270280497611,
      "grad_norm": 0.69921875,
      "learning_rate": 1.0271532865957658e-05,
      "loss": 0.3003,
      "step": 13010
    },
    {
      "epoch": 0.9847027103732084,
      "grad_norm": 0.7578125,
      "learning_rate": 1.0270342223190304e-05,
      "loss": 0.33,
      "step": 13011
    },
    {
      "epoch": 0.9847783926966558,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0269151576587681e-05,
      "loss": 0.2866,
      "step": 13012
    },
    {
      "epoch": 0.9848540750201031,
      "grad_norm": 0.7109375,
      "learning_rate": 1.0267960926166678e-05,
      "loss": 0.3058,
      "step": 13013
    },
    {
      "epoch": 0.9849297573435505,
      "grad_norm": 0.734375,
      "learning_rate": 1.0266770271944191e-05,
      "loss": 0.3096,
      "step": 13014
    },
    {
      "epoch": 0.9850054396669978,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0265579613937112e-05,
      "loss": 0.3085,
      "step": 13015
    },
    {
      "epoch": 0.9850811219904451,
      "grad_norm": 0.8125,
      "learning_rate": 1.0264388952162325e-05,
      "loss": 0.3361,
      "step": 13016
    },
    {
      "epoch": 0.9851568043138924,
      "grad_norm": 0.75,
      "learning_rate": 1.0263198286636733e-05,
      "loss": 0.3331,
      "step": 13017
    },
    {
      "epoch": 0.9852324866373398,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0262007617377221e-05,
      "loss": 0.282,
      "step": 13018
    },
    {
      "epoch": 0.9853081689607871,
      "grad_norm": 0.73046875,
      "learning_rate": 1.0260816944400678e-05,
      "loss": 0.3233,
      "step": 13019
    },
    {
      "epoch": 0.9853838512842344,
      "grad_norm": 0.77734375,
      "learning_rate": 1.0259626267724001e-05,
      "loss": 0.3376,
      "step": 13020
    },
    {
      "epoch": 0.9854595336076818,
      "grad_norm": 0.75,
      "learning_rate": 1.0258435587364081e-05,
      "loss": 0.3295,
      "step": 13021
    },
    {
      "epoch": 0.9855352159311291,
      "grad_norm": 0.7109375,
      "learning_rate": 1.0257244903337807e-05,
      "loss": 0.3077,
      "step": 13022
    },
    {
      "epoch": 0.9856108982545764,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0256054215662072e-05,
      "loss": 0.3195,
      "step": 13023
    },
    {
      "epoch": 0.9856865805780237,
      "grad_norm": 0.81640625,
      "learning_rate": 1.0254863524353768e-05,
      "loss": 0.3767,
      "step": 13024
    },
    {
      "epoch": 0.9857622629014711,
      "grad_norm": 0.78515625,
      "learning_rate": 1.0253672829429788e-05,
      "loss": 0.3498,
      "step": 13025
    },
    {
      "epoch": 0.9858379452249184,
      "grad_norm": 0.73046875,
      "learning_rate": 1.0252482130907027e-05,
      "loss": 0.291,
      "step": 13026
    },
    {
      "epoch": 0.9859136275483658,
      "grad_norm": 1.3046875,
      "learning_rate": 1.0251291428802373e-05,
      "loss": 0.3428,
      "step": 13027
    },
    {
      "epoch": 0.9859893098718131,
      "grad_norm": 0.80078125,
      "learning_rate": 1.0250100723132714e-05,
      "loss": 0.2736,
      "step": 13028
    },
    {
      "epoch": 0.9860649921952604,
      "grad_norm": 0.69140625,
      "learning_rate": 1.024891001391495e-05,
      "loss": 0.2779,
      "step": 13029
    },
    {
      "epoch": 0.9861406745187077,
      "grad_norm": 0.71484375,
      "learning_rate": 1.024771930116597e-05,
      "loss": 0.2909,
      "step": 13030
    },
    {
      "epoch": 0.986216356842155,
      "grad_norm": 0.78515625,
      "learning_rate": 1.0246528584902668e-05,
      "loss": 0.3405,
      "step": 13031
    },
    {
      "epoch": 0.9862920391656024,
      "grad_norm": 0.69140625,
      "learning_rate": 1.0245337865141935e-05,
      "loss": 0.2689,
      "step": 13032
    },
    {
      "epoch": 0.9863677214890497,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0244147141900663e-05,
      "loss": 0.3189,
      "step": 13033
    },
    {
      "epoch": 0.9864434038124971,
      "grad_norm": 0.7734375,
      "learning_rate": 1.0242956415195742e-05,
      "loss": 0.3229,
      "step": 13034
    },
    {
      "epoch": 0.9865190861359444,
      "grad_norm": 1.0390625,
      "learning_rate": 1.0241765685044071e-05,
      "loss": 0.3152,
      "step": 13035
    },
    {
      "epoch": 0.9865947684593918,
      "grad_norm": 0.70703125,
      "learning_rate": 1.0240574951462539e-05,
      "loss": 0.2643,
      "step": 13036
    },
    {
      "epoch": 0.986670450782839,
      "grad_norm": 0.75,
      "learning_rate": 1.0239384214468033e-05,
      "loss": 0.3248,
      "step": 13037
    },
    {
      "epoch": 0.9867461331062863,
      "grad_norm": 0.8046875,
      "learning_rate": 1.0238193474077456e-05,
      "loss": 0.3657,
      "step": 13038
    },
    {
      "epoch": 0.9868218154297337,
      "grad_norm": 0.6953125,
      "learning_rate": 1.0237002730307695e-05,
      "loss": 0.2689,
      "step": 13039
    },
    {
      "epoch": 0.986897497753181,
      "grad_norm": 0.765625,
      "learning_rate": 1.0235811983175646e-05,
      "loss": 0.3568,
      "step": 13040
    },
    {
      "epoch": 0.9869731800766284,
      "grad_norm": 0.703125,
      "learning_rate": 1.0234621232698197e-05,
      "loss": 0.2796,
      "step": 13041
    },
    {
      "epoch": 0.9870488624000757,
      "grad_norm": 0.765625,
      "learning_rate": 1.0233430478892242e-05,
      "loss": 0.2838,
      "step": 13042
    },
    {
      "epoch": 0.9871245447235231,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0232239721774672e-05,
      "loss": 0.308,
      "step": 13043
    },
    {
      "epoch": 0.9872002270469703,
      "grad_norm": 0.75,
      "learning_rate": 1.0231048961362389e-05,
      "loss": 0.2959,
      "step": 13044
    },
    {
      "epoch": 0.9872759093704176,
      "grad_norm": 0.765625,
      "learning_rate": 1.0229858197672277e-05,
      "loss": 0.3363,
      "step": 13045
    },
    {
      "epoch": 0.987351591693865,
      "grad_norm": 0.69921875,
      "learning_rate": 1.0228667430721235e-05,
      "loss": 0.2629,
      "step": 13046
    },
    {
      "epoch": 0.9874272740173123,
      "grad_norm": 0.75,
      "learning_rate": 1.022747666052615e-05,
      "loss": 0.3049,
      "step": 13047
    },
    {
      "epoch": 0.9875029563407597,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0226285887103919e-05,
      "loss": 0.3094,
      "step": 13048
    },
    {
      "epoch": 0.987578638664207,
      "grad_norm": 0.7734375,
      "learning_rate": 1.0225095110471437e-05,
      "loss": 0.3009,
      "step": 13049
    },
    {
      "epoch": 0.9876543209876543,
      "grad_norm": 0.7734375,
      "learning_rate": 1.022390433064559e-05,
      "loss": 0.3118,
      "step": 13050
    },
    {
      "epoch": 0.9877300033111016,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0222713547643277e-05,
      "loss": 0.288,
      "step": 13051
    },
    {
      "epoch": 0.987805685634549,
      "grad_norm": 0.73828125,
      "learning_rate": 1.0221522761481394e-05,
      "loss": 0.2987,
      "step": 13052
    },
    {
      "epoch": 0.9878813679579963,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0220331972176826e-05,
      "loss": 0.3206,
      "step": 13053
    },
    {
      "epoch": 0.9879570502814436,
      "grad_norm": 0.79296875,
      "learning_rate": 1.0219141179746472e-05,
      "loss": 0.3149,
      "step": 13054
    },
    {
      "epoch": 0.988032732604891,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0217950384207227e-05,
      "loss": 0.3151,
      "step": 13055
    },
    {
      "epoch": 0.9881084149283383,
      "grad_norm": 0.73828125,
      "learning_rate": 1.0216759585575979e-05,
      "loss": 0.3197,
      "step": 13056
    },
    {
      "epoch": 0.9881840972517856,
      "grad_norm": 0.89453125,
      "learning_rate": 1.0215568783869622e-05,
      "loss": 0.2767,
      "step": 13057
    },
    {
      "epoch": 0.9882597795752329,
      "grad_norm": 0.77734375,
      "learning_rate": 1.0214377979105057e-05,
      "loss": 0.3385,
      "step": 13058
    },
    {
      "epoch": 0.9883354618986803,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0213187171299173e-05,
      "loss": 0.2803,
      "step": 13059
    },
    {
      "epoch": 0.9884111442221276,
      "grad_norm": 1.0390625,
      "learning_rate": 1.0211996360468859e-05,
      "loss": 0.401,
      "step": 13060
    },
    {
      "epoch": 0.988486826545575,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0210805546631013e-05,
      "loss": 0.3157,
      "step": 13061
    },
    {
      "epoch": 0.9885625088690223,
      "grad_norm": 0.66796875,
      "learning_rate": 1.020961472980253e-05,
      "loss": 0.2659,
      "step": 13062
    },
    {
      "epoch": 0.9886381911924697,
      "grad_norm": 0.75,
      "learning_rate": 1.0208423910000302e-05,
      "loss": 0.3115,
      "step": 13063
    },
    {
      "epoch": 0.9887138735159169,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0207233087241224e-05,
      "loss": 0.2896,
      "step": 13064
    },
    {
      "epoch": 0.9887895558393642,
      "grad_norm": 0.8125,
      "learning_rate": 1.020604226154219e-05,
      "loss": 0.3529,
      "step": 13065
    },
    {
      "epoch": 0.9888652381628116,
      "grad_norm": 0.734375,
      "learning_rate": 1.0204851432920088e-05,
      "loss": 0.2979,
      "step": 13066
    },
    {
      "epoch": 0.9889409204862589,
      "grad_norm": 0.78125,
      "learning_rate": 1.0203660601391821e-05,
      "loss": 0.3455,
      "step": 13067
    },
    {
      "epoch": 0.9890166028097063,
      "grad_norm": 0.96484375,
      "learning_rate": 1.0202469766974281e-05,
      "loss": 0.3677,
      "step": 13068
    },
    {
      "epoch": 0.9890922851331536,
      "grad_norm": 0.8359375,
      "learning_rate": 1.0201278929684355e-05,
      "loss": 0.3287,
      "step": 13069
    },
    {
      "epoch": 0.989167967456601,
      "grad_norm": 0.82421875,
      "learning_rate": 1.0200088089538944e-05,
      "loss": 0.3572,
      "step": 13070
    },
    {
      "epoch": 0.9892436497800482,
      "grad_norm": 0.73828125,
      "learning_rate": 1.0198897246554942e-05,
      "loss": 0.2641,
      "step": 13071
    },
    {
      "epoch": 0.9893193321034955,
      "grad_norm": 0.703125,
      "learning_rate": 1.0197706400749236e-05,
      "loss": 0.2964,
      "step": 13072
    },
    {
      "epoch": 0.9893950144269429,
      "grad_norm": 0.74609375,
      "learning_rate": 1.019651555213873e-05,
      "loss": 0.3077,
      "step": 13073
    },
    {
      "epoch": 0.9894706967503902,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0195324700740314e-05,
      "loss": 0.3076,
      "step": 13074
    },
    {
      "epoch": 0.9895463790738376,
      "grad_norm": 0.7578125,
      "learning_rate": 1.0194133846570879e-05,
      "loss": 0.3098,
      "step": 13075
    },
    {
      "epoch": 0.9896220613972849,
      "grad_norm": 0.734375,
      "learning_rate": 1.0192942989647322e-05,
      "loss": 0.2678,
      "step": 13076
    },
    {
      "epoch": 0.9896977437207323,
      "grad_norm": 0.671875,
      "learning_rate": 1.0191752129986543e-05,
      "loss": 0.2622,
      "step": 13077
    },
    {
      "epoch": 0.9897734260441795,
      "grad_norm": 0.79296875,
      "learning_rate": 1.0190561267605425e-05,
      "loss": 0.3331,
      "step": 13078
    },
    {
      "epoch": 0.9898491083676269,
      "grad_norm": 0.73046875,
      "learning_rate": 1.018937040252087e-05,
      "loss": 0.2783,
      "step": 13079
    },
    {
      "epoch": 0.9899247906910742,
      "grad_norm": 1.15625,
      "learning_rate": 1.0188179534749773e-05,
      "loss": 0.3553,
      "step": 13080
    },
    {
      "epoch": 0.9900004730145215,
      "grad_norm": 0.765625,
      "learning_rate": 1.0186988664309023e-05,
      "loss": 0.3207,
      "step": 13081
    },
    {
      "epoch": 0.9900761553379689,
      "grad_norm": 0.7890625,
      "learning_rate": 1.0185797791215521e-05,
      "loss": 0.3414,
      "step": 13082
    },
    {
      "epoch": 0.9901518376614162,
      "grad_norm": 0.71484375,
      "learning_rate": 1.018460691548616e-05,
      "loss": 0.3028,
      "step": 13083
    },
    {
      "epoch": 0.9902275199848636,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0183416037137828e-05,
      "loss": 0.3459,
      "step": 13084
    },
    {
      "epoch": 0.9903032023083108,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0182225156187426e-05,
      "loss": 0.2796,
      "step": 13085
    },
    {
      "epoch": 0.9903788846317582,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0181034272651848e-05,
      "loss": 0.3177,
      "step": 13086
    },
    {
      "epoch": 0.9904545669552055,
      "grad_norm": 0.87109375,
      "learning_rate": 1.017984338654799e-05,
      "loss": 0.3698,
      "step": 13087
    },
    {
      "epoch": 0.9905302492786529,
      "grad_norm": 0.796875,
      "learning_rate": 1.0178652497892744e-05,
      "loss": 0.3288,
      "step": 13088
    },
    {
      "epoch": 0.9906059316021002,
      "grad_norm": 0.77734375,
      "learning_rate": 1.0177461606703005e-05,
      "loss": 0.3247,
      "step": 13089
    },
    {
      "epoch": 0.9906816139255475,
      "grad_norm": 0.7734375,
      "learning_rate": 1.0176270712995668e-05,
      "loss": 0.3076,
      "step": 13090
    },
    {
      "epoch": 0.9907572962489949,
      "grad_norm": 0.6953125,
      "learning_rate": 1.0175079816787631e-05,
      "loss": 0.2568,
      "step": 13091
    },
    {
      "epoch": 0.9908329785724421,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0173888918095782e-05,
      "loss": 0.2973,
      "step": 13092
    },
    {
      "epoch": 0.9909086608958895,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0172698016937023e-05,
      "loss": 0.2969,
      "step": 13093
    },
    {
      "epoch": 0.9909843432193368,
      "grad_norm": 0.77734375,
      "learning_rate": 1.0171507113328248e-05,
      "loss": 0.3467,
      "step": 13094
    },
    {
      "epoch": 0.9910600255427842,
      "grad_norm": 0.67578125,
      "learning_rate": 1.0170316207286349e-05,
      "loss": 0.2565,
      "step": 13095
    },
    {
      "epoch": 0.9911357078662315,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0169125298828222e-05,
      "loss": 0.3115,
      "step": 13096
    },
    {
      "epoch": 0.9912113901896789,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0167934387970764e-05,
      "loss": 0.3294,
      "step": 13097
    },
    {
      "epoch": 0.9912870725131262,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0166743474730865e-05,
      "loss": 0.3032,
      "step": 13098
    },
    {
      "epoch": 0.9913627548365734,
      "grad_norm": 0.73046875,
      "learning_rate": 1.0165552559125424e-05,
      "loss": 0.3344,
      "step": 13099
    },
    {
      "epoch": 0.9914384371600208,
      "grad_norm": 0.6796875,
      "learning_rate": 1.016436164117134e-05,
      "loss": 0.2513,
      "step": 13100
    },
    {
      "epoch": 0.9915141194834681,
      "grad_norm": 0.83203125,
      "learning_rate": 1.01631707208855e-05,
      "loss": 0.3315,
      "step": 13101
    },
    {
      "epoch": 0.9915898018069155,
      "grad_norm": 0.7109375,
      "learning_rate": 1.0161979798284805e-05,
      "loss": 0.2759,
      "step": 13102
    },
    {
      "epoch": 0.9916654841303628,
      "grad_norm": 0.8203125,
      "learning_rate": 1.016078887338615e-05,
      "loss": 0.3433,
      "step": 13103
    },
    {
      "epoch": 0.9917411664538102,
      "grad_norm": 0.70703125,
      "learning_rate": 1.0159597946206425e-05,
      "loss": 0.2904,
      "step": 13104
    },
    {
      "epoch": 0.9918168487772575,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0158407016762533e-05,
      "loss": 0.3152,
      "step": 13105
    },
    {
      "epoch": 0.9918925311007047,
      "grad_norm": 0.73046875,
      "learning_rate": 1.0157216085071365e-05,
      "loss": 0.2872,
      "step": 13106
    },
    {
      "epoch": 0.9919682134241521,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0156025151149813e-05,
      "loss": 0.2691,
      "step": 13107
    },
    {
      "epoch": 0.9920438957475994,
      "grad_norm": 0.75390625,
      "learning_rate": 1.015483421501478e-05,
      "loss": 0.323,
      "step": 13108
    },
    {
      "epoch": 0.9921195780710468,
      "grad_norm": 0.78515625,
      "learning_rate": 1.0153643276683158e-05,
      "loss": 0.3241,
      "step": 13109
    },
    {
      "epoch": 0.9921952603944941,
      "grad_norm": 0.78125,
      "learning_rate": 1.0152452336171842e-05,
      "loss": 0.3133,
      "step": 13110
    },
    {
      "epoch": 0.9922709427179415,
      "grad_norm": 0.75,
      "learning_rate": 1.0151261393497728e-05,
      "loss": 0.2865,
      "step": 13111
    },
    {
      "epoch": 0.9923466250413888,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0150070448677714e-05,
      "loss": 0.3271,
      "step": 13112
    },
    {
      "epoch": 0.992422307364836,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0148879501728688e-05,
      "loss": 0.3295,
      "step": 13113
    },
    {
      "epoch": 0.9924979896882834,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0147688552667555e-05,
      "loss": 0.2885,
      "step": 13114
    },
    {
      "epoch": 0.9925736720117307,
      "grad_norm": 0.8203125,
      "learning_rate": 1.0146497601511205e-05,
      "loss": 0.3448,
      "step": 13115
    },
    {
      "epoch": 0.9926493543351781,
      "grad_norm": 0.80859375,
      "learning_rate": 1.0145306648276535e-05,
      "loss": 0.3526,
      "step": 13116
    },
    {
      "epoch": 0.9927250366586254,
      "grad_norm": 0.79296875,
      "learning_rate": 1.014411569298044e-05,
      "loss": 0.3507,
      "step": 13117
    },
    {
      "epoch": 0.9928007189820728,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0142924735639819e-05,
      "loss": 0.3037,
      "step": 13118
    },
    {
      "epoch": 0.9928764013055201,
      "grad_norm": 0.70703125,
      "learning_rate": 1.0141733776271563e-05,
      "loss": 0.2773,
      "step": 13119
    },
    {
      "epoch": 0.9929520836289674,
      "grad_norm": 0.70703125,
      "learning_rate": 1.014054281489257e-05,
      "loss": 0.251,
      "step": 13120
    },
    {
      "epoch": 0.9930277659524147,
      "grad_norm": 0.76171875,
      "learning_rate": 1.013935185151974e-05,
      "loss": 0.2742,
      "step": 13121
    },
    {
      "epoch": 0.993103448275862,
      "grad_norm": 0.69921875,
      "learning_rate": 1.0138160886169958e-05,
      "loss": 0.2863,
      "step": 13122
    },
    {
      "epoch": 0.9931791305993094,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0136969918860131e-05,
      "loss": 0.3204,
      "step": 13123
    },
    {
      "epoch": 0.9932548129227567,
      "grad_norm": 0.8046875,
      "learning_rate": 1.0135778949607153e-05,
      "loss": 0.3343,
      "step": 13124
    },
    {
      "epoch": 0.9933304952462041,
      "grad_norm": 0.77734375,
      "learning_rate": 1.0134587978427914e-05,
      "loss": 0.3278,
      "step": 13125
    },
    {
      "epoch": 0.9934061775696514,
      "grad_norm": 0.734375,
      "learning_rate": 1.0133397005339313e-05,
      "loss": 0.3047,
      "step": 13126
    },
    {
      "epoch": 0.9934818598930987,
      "grad_norm": 0.765625,
      "learning_rate": 1.013220603035825e-05,
      "loss": 0.3153,
      "step": 13127
    },
    {
      "epoch": 0.993557542216546,
      "grad_norm": 0.8203125,
      "learning_rate": 1.0131015053501614e-05,
      "loss": 0.35,
      "step": 13128
    },
    {
      "epoch": 0.9936332245399934,
      "grad_norm": 0.7578125,
      "learning_rate": 1.0129824074786307e-05,
      "loss": 0.3139,
      "step": 13129
    },
    {
      "epoch": 0.9937089068634407,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0128633094229224e-05,
      "loss": 0.2959,
      "step": 13130
    },
    {
      "epoch": 0.9937845891868881,
      "grad_norm": 0.703125,
      "learning_rate": 1.0127442111847254e-05,
      "loss": 0.2899,
      "step": 13131
    },
    {
      "epoch": 0.9938602715103354,
      "grad_norm": 0.6875,
      "learning_rate": 1.0126251127657305e-05,
      "loss": 0.2592,
      "step": 13132
    },
    {
      "epoch": 0.9939359538337827,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0125060141676267e-05,
      "loss": 0.3211,
      "step": 13133
    },
    {
      "epoch": 0.99401163615723,
      "grad_norm": 0.84375,
      "learning_rate": 1.0123869153921031e-05,
      "loss": 0.3649,
      "step": 13134
    },
    {
      "epoch": 0.9940873184806773,
      "grad_norm": 0.7109375,
      "learning_rate": 1.01226781644085e-05,
      "loss": 0.2733,
      "step": 13135
    },
    {
      "epoch": 0.9941630008041247,
      "grad_norm": 0.77734375,
      "learning_rate": 1.0121487173155573e-05,
      "loss": 0.3187,
      "step": 13136
    },
    {
      "epoch": 0.994238683127572,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0120296180179137e-05,
      "loss": 0.2731,
      "step": 13137
    },
    {
      "epoch": 0.9943143654510194,
      "grad_norm": 0.7890625,
      "learning_rate": 1.0119105185496099e-05,
      "loss": 0.3425,
      "step": 13138
    },
    {
      "epoch": 0.9943900477744667,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0117914189123346e-05,
      "loss": 0.3083,
      "step": 13139
    },
    {
      "epoch": 0.9944657300979141,
      "grad_norm": 0.73046875,
      "learning_rate": 1.0116723191077775e-05,
      "loss": 0.295,
      "step": 13140
    },
    {
      "epoch": 0.9945414124213613,
      "grad_norm": 0.7578125,
      "learning_rate": 1.0115532191376289e-05,
      "loss": 0.311,
      "step": 13141
    },
    {
      "epoch": 0.9946170947448086,
      "grad_norm": 0.74609375,
      "learning_rate": 1.0114341190035782e-05,
      "loss": 0.3313,
      "step": 13142
    },
    {
      "epoch": 0.994692777068256,
      "grad_norm": 0.73828125,
      "learning_rate": 1.0113150187073147e-05,
      "loss": 0.3043,
      "step": 13143
    },
    {
      "epoch": 0.9947684593917033,
      "grad_norm": 0.65234375,
      "learning_rate": 1.011195918250528e-05,
      "loss": 0.2359,
      "step": 13144
    },
    {
      "epoch": 0.9948441417151507,
      "grad_norm": 0.73828125,
      "learning_rate": 1.0110768176349084e-05,
      "loss": 0.3198,
      "step": 13145
    },
    {
      "epoch": 0.994919824038598,
      "grad_norm": 0.69140625,
      "learning_rate": 1.0109577168621447e-05,
      "loss": 0.2581,
      "step": 13146
    },
    {
      "epoch": 0.9949955063620454,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0108386159339274e-05,
      "loss": 0.288,
      "step": 13147
    },
    {
      "epoch": 0.9950711886854926,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0107195148519453e-05,
      "loss": 0.295,
      "step": 13148
    },
    {
      "epoch": 0.99514687100894,
      "grad_norm": 0.82421875,
      "learning_rate": 1.0106004136178888e-05,
      "loss": 0.3316,
      "step": 13149
    },
    {
      "epoch": 0.9952225533323873,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0104813122334474e-05,
      "loss": 0.3184,
      "step": 13150
    },
    {
      "epoch": 0.9952982356558346,
      "grad_norm": 0.71875,
      "learning_rate": 1.0103622107003103e-05,
      "loss": 0.2958,
      "step": 13151
    },
    {
      "epoch": 0.995373917979282,
      "grad_norm": 0.75,
      "learning_rate": 1.0102431090201677e-05,
      "loss": 0.3153,
      "step": 13152
    },
    {
      "epoch": 0.9954496003027293,
      "grad_norm": 0.73046875,
      "learning_rate": 1.0101240071947087e-05,
      "loss": 0.2838,
      "step": 13153
    },
    {
      "epoch": 0.9955252826261767,
      "grad_norm": 0.79296875,
      "learning_rate": 1.0100049052256236e-05,
      "loss": 0.3121,
      "step": 13154
    },
    {
      "epoch": 0.9956009649496239,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0098858031146012e-05,
      "loss": 0.2859,
      "step": 13155
    },
    {
      "epoch": 0.9956766472730713,
      "grad_norm": 0.7265625,
      "learning_rate": 1.0097667008633324e-05,
      "loss": 0.2706,
      "step": 13156
    },
    {
      "epoch": 0.9957523295965186,
      "grad_norm": 0.7109375,
      "learning_rate": 1.0096475984735059e-05,
      "loss": 0.2734,
      "step": 13157
    },
    {
      "epoch": 0.995828011919966,
      "grad_norm": 0.69140625,
      "learning_rate": 1.0095284959468117e-05,
      "loss": 0.274,
      "step": 13158
    },
    {
      "epoch": 0.9959036942434133,
      "grad_norm": 0.796875,
      "learning_rate": 1.0094093932849394e-05,
      "loss": 0.3686,
      "step": 13159
    },
    {
      "epoch": 0.9959793765668606,
      "grad_norm": 0.7890625,
      "learning_rate": 1.0092902904895787e-05,
      "loss": 0.3449,
      "step": 13160
    },
    {
      "epoch": 0.996055058890308,
      "grad_norm": 0.79296875,
      "learning_rate": 1.0091711875624192e-05,
      "loss": 0.3066,
      "step": 13161
    },
    {
      "epoch": 0.9961307412137552,
      "grad_norm": 0.8203125,
      "learning_rate": 1.0090520845051508e-05,
      "loss": 0.333,
      "step": 13162
    },
    {
      "epoch": 0.9962064235372026,
      "grad_norm": 0.68359375,
      "learning_rate": 1.0089329813194628e-05,
      "loss": 0.2549,
      "step": 13163
    },
    {
      "epoch": 0.9962821058606499,
      "grad_norm": 0.78515625,
      "learning_rate": 1.0088138780070456e-05,
      "loss": 0.3208,
      "step": 13164
    },
    {
      "epoch": 0.9963577881840973,
      "grad_norm": 0.796875,
      "learning_rate": 1.008694774569588e-05,
      "loss": 0.3509,
      "step": 13165
    },
    {
      "epoch": 0.9964334705075446,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0085756710087803e-05,
      "loss": 0.3137,
      "step": 13166
    },
    {
      "epoch": 0.996509152830992,
      "grad_norm": 0.6875,
      "learning_rate": 1.0084565673263121e-05,
      "loss": 0.2831,
      "step": 13167
    },
    {
      "epoch": 0.9965848351544392,
      "grad_norm": 0.69140625,
      "learning_rate": 1.0083374635238728e-05,
      "loss": 0.2637,
      "step": 13168
    },
    {
      "epoch": 0.9966605174778865,
      "grad_norm": 0.74609375,
      "learning_rate": 1.008218359603152e-05,
      "loss": 0.283,
      "step": 13169
    },
    {
      "epoch": 0.9967361998013339,
      "grad_norm": 0.75,
      "learning_rate": 1.00809925556584e-05,
      "loss": 0.3031,
      "step": 13170
    },
    {
      "epoch": 0.9968118821247812,
      "grad_norm": 0.7421875,
      "learning_rate": 1.0079801514136265e-05,
      "loss": 0.321,
      "step": 13171
    },
    {
      "epoch": 0.9968875644482286,
      "grad_norm": 0.80859375,
      "learning_rate": 1.0078610471482002e-05,
      "loss": 0.355,
      "step": 13172
    },
    {
      "epoch": 0.9969632467716759,
      "grad_norm": 0.75,
      "learning_rate": 1.0077419427712519e-05,
      "loss": 0.3279,
      "step": 13173
    },
    {
      "epoch": 0.9970389290951233,
      "grad_norm": 0.70703125,
      "learning_rate": 1.0076228382844706e-05,
      "loss": 0.3004,
      "step": 13174
    },
    {
      "epoch": 0.9971146114185705,
      "grad_norm": 0.69921875,
      "learning_rate": 1.0075037336895463e-05,
      "loss": 0.244,
      "step": 13175
    },
    {
      "epoch": 0.9971902937420178,
      "grad_norm": 0.69921875,
      "learning_rate": 1.0073846289881687e-05,
      "loss": 0.2801,
      "step": 13176
    },
    {
      "epoch": 0.9972659760654652,
      "grad_norm": 0.80078125,
      "learning_rate": 1.0072655241820278e-05,
      "loss": 0.356,
      "step": 13177
    },
    {
      "epoch": 0.9973416583889125,
      "grad_norm": 0.7109375,
      "learning_rate": 1.0071464192728122e-05,
      "loss": 0.293,
      "step": 13178
    },
    {
      "epoch": 0.9974173407123599,
      "grad_norm": 0.7890625,
      "learning_rate": 1.0070273142622129e-05,
      "loss": 0.3525,
      "step": 13179
    },
    {
      "epoch": 0.9974930230358072,
      "grad_norm": 0.7734375,
      "learning_rate": 1.0069082091519193e-05,
      "loss": 0.3498,
      "step": 13180
    },
    {
      "epoch": 0.9975687053592546,
      "grad_norm": 0.78515625,
      "learning_rate": 1.0067891039436203e-05,
      "loss": 0.3377,
      "step": 13181
    },
    {
      "epoch": 0.9976443876827018,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0066699986390067e-05,
      "loss": 0.3151,
      "step": 13182
    },
    {
      "epoch": 0.9977200700061492,
      "grad_norm": 0.73828125,
      "learning_rate": 1.0065508932397675e-05,
      "loss": 0.309,
      "step": 13183
    },
    {
      "epoch": 0.9977957523295965,
      "grad_norm": 0.8203125,
      "learning_rate": 1.0064317877475926e-05,
      "loss": 0.3464,
      "step": 13184
    },
    {
      "epoch": 0.9978714346530438,
      "grad_norm": 0.69921875,
      "learning_rate": 1.006312682164172e-05,
      "loss": 0.2533,
      "step": 13185
    },
    {
      "epoch": 0.9979471169764912,
      "grad_norm": 0.75390625,
      "learning_rate": 1.0061935764911953e-05,
      "loss": 0.3411,
      "step": 13186
    },
    {
      "epoch": 0.9980227992999385,
      "grad_norm": 0.82421875,
      "learning_rate": 1.0060744707303513e-05,
      "loss": 0.3017,
      "step": 13187
    },
    {
      "epoch": 0.9980984816233859,
      "grad_norm": 0.69140625,
      "learning_rate": 1.0059553648833312e-05,
      "loss": 0.2428,
      "step": 13188
    },
    {
      "epoch": 0.9981741639468331,
      "grad_norm": 0.66796875,
      "learning_rate": 1.0058362589518241e-05,
      "loss": 0.2564,
      "step": 13189
    },
    {
      "epoch": 0.9982498462702805,
      "grad_norm": 0.78515625,
      "learning_rate": 1.0057171529375192e-05,
      "loss": 0.3088,
      "step": 13190
    },
    {
      "epoch": 0.9983255285937278,
      "grad_norm": 0.7890625,
      "learning_rate": 1.005598046842107e-05,
      "loss": 0.3417,
      "step": 13191
    },
    {
      "epoch": 0.9984012109171752,
      "grad_norm": 0.72265625,
      "learning_rate": 1.0054789406672767e-05,
      "loss": 0.27,
      "step": 13192
    },
    {
      "epoch": 0.9984768932406225,
      "grad_norm": 0.73046875,
      "learning_rate": 1.005359834414718e-05,
      "loss": 0.2642,
      "step": 13193
    },
    {
      "epoch": 0.9985525755640698,
      "grad_norm": 0.76171875,
      "learning_rate": 1.0052407280861214e-05,
      "loss": 0.3164,
      "step": 13194
    },
    {
      "epoch": 0.9986282578875172,
      "grad_norm": 0.76171875,
      "learning_rate": 1.005121621683176e-05,
      "loss": 0.3242,
      "step": 13195
    },
    {
      "epoch": 0.9987039402109644,
      "grad_norm": 0.76953125,
      "learning_rate": 1.0050025152075711e-05,
      "loss": 0.3214,
      "step": 13196
    },
    {
      "epoch": 0.9987796225344118,
      "grad_norm": 0.80859375,
      "learning_rate": 1.0048834086609973e-05,
      "loss": 0.3524,
      "step": 13197
    },
    {
      "epoch": 0.9988553048578591,
      "grad_norm": 0.84765625,
      "learning_rate": 1.0047643020451439e-05,
      "loss": 0.3819,
      "step": 13198
    },
    {
      "epoch": 0.9989309871813065,
      "grad_norm": 0.80078125,
      "learning_rate": 1.0046451953617008e-05,
      "loss": 0.361,
      "step": 13199
    },
    {
      "epoch": 0.9990066695047538,
      "grad_norm": 0.765625,
      "learning_rate": 1.0045260886123576e-05,
      "loss": 0.3095,
      "step": 13200
    },
    {
      "epoch": 0.9990823518282012,
      "grad_norm": 0.703125,
      "learning_rate": 1.004406981798804e-05,
      "loss": 0.2856,
      "step": 13201
    },
    {
      "epoch": 0.9991580341516485,
      "grad_norm": 0.7734375,
      "learning_rate": 1.0042878749227296e-05,
      "loss": 0.2885,
      "step": 13202
    },
    {
      "epoch": 0.9992337164750957,
      "grad_norm": 0.82421875,
      "learning_rate": 1.004168767985825e-05,
      "loss": 0.351,
      "step": 13203
    },
    {
      "epoch": 0.9993093987985431,
      "grad_norm": 0.69921875,
      "learning_rate": 1.0040496609897785e-05,
      "loss": 0.2948,
      "step": 13204
    },
    {
      "epoch": 0.9993850811219904,
      "grad_norm": 0.7421875,
      "learning_rate": 1.003930553936281e-05,
      "loss": 0.3158,
      "step": 13205
    },
    {
      "epoch": 0.9994607634454378,
      "grad_norm": 0.734375,
      "learning_rate": 1.003811446827022e-05,
      "loss": 0.2831,
      "step": 13206
    },
    {
      "epoch": 0.9995364457688851,
      "grad_norm": 0.73046875,
      "learning_rate": 1.0036923396636906e-05,
      "loss": 0.3073,
      "step": 13207
    },
    {
      "epoch": 0.9996121280923325,
      "grad_norm": 0.765625,
      "learning_rate": 1.0035732324479774e-05,
      "loss": 0.3321,
      "step": 13208
    },
    {
      "epoch": 0.9996878104157798,
      "grad_norm": 0.8125,
      "learning_rate": 1.0034541251815716e-05,
      "loss": 0.303,
      "step": 13209
    },
    {
      "epoch": 0.999763492739227,
      "grad_norm": 0.80078125,
      "learning_rate": 1.0033350178661633e-05,
      "loss": 0.3667,
      "step": 13210
    },
    {
      "epoch": 0.9998391750626744,
      "grad_norm": 0.7734375,
      "learning_rate": 1.0032159105034418e-05,
      "loss": 0.3296,
      "step": 13211
    },
    {
      "epoch": 0.9999148573861217,
      "grad_norm": 0.734375,
      "learning_rate": 1.0030968030950972e-05,
      "loss": 0.3005,
      "step": 13212
    },
    {
      "epoch": 0.9999905397095691,
      "grad_norm": 0.73828125,
      "learning_rate": 1.0029776956428192e-05,
      "loss": 0.2956,
      "step": 13213
    },
    {
      "epoch": 1.0000662220330163,
      "grad_norm": 0.71484375,
      "learning_rate": 1.0028585881482975e-05,
      "loss": 0.2994,
      "step": 13214
    }
  ],
  "logging_steps": 1,
  "max_steps": 26426,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 6607,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.8995302625402946e+19,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}