{
  "best_metric": 4.340610504150391,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/existential-there-quantifier/lstm/2/checkpoints/checkpoint-228960",
  "epoch": 0.025000606015738065,
  "eval_steps": 10,
  "global_step": 228960,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 4.999998362119627e-05,
      "loss": 10.8193,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 7.5492,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 7.0645,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 7.002,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 6.9471,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 6.9141,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 6.7538,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994129836742636e-05,
      "loss": 6.6419,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993292879871958e-05,
      "loss": 6.5403,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992454285120906e-05,
      "loss": 6.4622,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.991615690369854e-05,
      "loss": 6.3981,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990777095618801e-05,
      "loss": 6.3393,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989938500867749e-05,
      "loss": 6.268,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989099906116697e-05,
      "loss": 6.2066,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988261311365645e-05,
      "loss": 6.1484,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.987422716614593e-05,
      "loss": 6.0968,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986584121863541e-05,
      "loss": 6.0509,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985745527112489e-05,
      "loss": 6.0172,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984906932361437e-05,
      "loss": 5.967,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984068337610385e-05,
      "loss": 5.9336,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983229742859333e-05,
      "loss": 5.8997,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.982391148108281e-05,
      "loss": 5.8501,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.981552553357229e-05,
      "loss": 5.8196,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.980713958606178e-05,
      "loss": 5.7869,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9798770017354986e-05,
      "loss": 5.7698,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9790384069844466e-05,
      "loss": 5.7335,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9781998122333946e-05,
      "loss": 5.7045,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773612174823426e-05,
      "loss": 5.6859,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9765242606116635e-05,
      "loss": 5.6594,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9756856658606115e-05,
      "loss": 5.6313,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9748487089899324e-05,
      "loss": 5.6066,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9740101142388804e-05,
      "loss": 5.5985,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9731715194878284e-05,
      "loss": 5.5763,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.97233456261715e-05,
      "loss": 5.5659,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.971495967866098e-05,
      "loss": 5.5443,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.970657373115046e-05,
      "loss": 5.525,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.969818778363994e-05,
      "loss": 5.5193,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968980183612942e-05,
      "loss": 5.4807,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.968143226742263e-05,
      "loss": 5.4701,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.967304631991211e-05,
      "loss": 5.4515,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.966466037240159e-05,
      "loss": 5.4446,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965627442489107e-05,
      "loss": 5.4206,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964790485618428e-05,
      "loss": 5.4259,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963951890867376e-05,
      "loss": 5.3917,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963113296116324e-05,
      "loss": 5.391,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9622763392456454e-05,
      "loss": 5.3845,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9614377444945934e-05,
      "loss": 5.3756,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9605991497435414e-05,
      "loss": 5.3678,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.959762192872862e-05,
      "loss": 5.3446,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.95892359812181e-05,
      "loss": 5.3237,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958085003370758e-05,
      "loss": 5.3379,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.957246408619706e-05,
      "loss": 5.3169,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956407813868654e-05,
      "loss": 5.3024,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9555692191176016e-05,
      "loss": 5.2854,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9547306243665496e-05,
      "loss": 5.2922,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9538920296154976e-05,
      "loss": 5.272,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9530534348644456e-05,
      "loss": 5.2853,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9522148401133936e-05,
      "loss": 5.2534,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.951376245362342e-05,
      "loss": 5.2468,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.95053765061129e-05,
      "loss": 5.2473,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.949700693740611e-05,
      "loss": 5.2315,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948862098989559e-05,
      "loss": 5.2092,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948023504238507e-05,
      "loss": 5.2138,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.947184909487455e-05,
      "loss": 5.192,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.946346314736403e-05,
      "loss": 5.1927,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.945507719985351e-05,
      "loss": 5.2033,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.944670763114672e-05,
      "loss": 5.1956,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.94383216836362e-05,
      "loss": 5.177,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942995211492941e-05,
      "loss": 5.1571,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942156616741889e-05,
      "loss": 5.1516,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.941318021990837e-05,
      "loss": 5.1625,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9404794272397856e-05,
      "loss": 5.1509,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9396408324887336e-05,
      "loss": 5.1476,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9388022377376816e-05,
      "loss": 5.1455,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9379636429866296e-05,
      "loss": 5.1444,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9371266861159505e-05,
      "loss": 5.1363,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9362897292452714e-05,
      "loss": 5.1086,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9354511344942194e-05,
      "loss": 5.1124,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9346125397431674e-05,
      "loss": 5.1071,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9337739449921154e-05,
      "loss": 5.093,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9329353502410634e-05,
      "loss": 5.1114,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9320967554900114e-05,
      "loss": 5.0892,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9312581607389594e-05,
      "loss": 5.0963,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9304195659879074e-05,
      "loss": 5.0799,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9295809712368554e-05,
      "loss": 5.0572,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9287423764858034e-05,
      "loss": 5.0618,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9279037817347514e-05,
      "loss": 5.0654,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9270651869836994e-05,
      "loss": 5.0603,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9262265922326474e-05,
      "loss": 5.0547,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.925389635361968e-05,
      "loss": 5.0417,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.924551040610916e-05,
      "loss": 5.0396,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.923712445859864e-05,
      "loss": 5.0283,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922875488989185e-05,
      "loss": 5.0219,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922036894238133e-05,
      "loss": 5.0194,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921198299487081e-05,
      "loss": 5.0117,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920359704736029e-05,
      "loss": 5.02,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919521109984978e-05,
      "loss": 5.0006,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918682515233926e-05,
      "loss": 4.9951,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917843920482874e-05,
      "loss": 4.9888,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917005325731822e-05,
      "loss": 4.9906,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.91616673098077e-05,
      "loss": 4.9893,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.915328136229718e-05,
      "loss": 4.9769,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.914491179359039e-05,
      "loss": 4.9699,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.913652584607987e-05,
      "loss": 4.9543,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.912813989856935e-05,
      "loss": 4.963,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.911975395105883e-05,
      "loss": 4.9493,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9111384382352036e-05,
      "loss": 4.9441,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9102998434841516e-05,
      "loss": 4.9462,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9094612487330996e-05,
      "loss": 4.944,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9086226539820476e-05,
      "loss": 4.9389,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907787334991742e-05,
      "loss": 4.9359,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90694874024069e-05,
      "loss": 4.9245,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906110145489638e-05,
      "loss": 4.9255,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.905271550738586e-05,
      "loss": 4.9159,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.904432955987534e-05,
      "loss": 4.9109,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.903594361236482e-05,
      "loss": 4.9194,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90275576648543e-05,
      "loss": 4.9099,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901917171734378e-05,
      "loss": 4.909,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901080214863699e-05,
      "loss": 4.898,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.900241620112647e-05,
      "loss": 4.9054,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.899403025361595e-05,
      "loss": 4.9012,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8985660684909166e-05,
      "loss": 4.8819,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8977274737398646e-05,
      "loss": 4.8966,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8968888789888125e-05,
      "loss": 4.8841,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8960502842377605e-05,
      "loss": 4.8852,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8952116894867085e-05,
      "loss": 4.864,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8943730947356565e-05,
      "loss": 4.8568,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.893534499984604e-05,
      "loss": 4.8546,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8926975431139254e-05,
      "loss": 4.8574,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8918589483628734e-05,
      "loss": 4.8465,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8910203536118214e-05,
      "loss": 4.865,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.890181758860769e-05,
      "loss": 4.8606,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88934480199009e-05,
      "loss": 4.8422,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888506207239038e-05,
      "loss": 4.8469,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.887667612487986e-05,
      "loss": 4.842,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.886829017736934e-05,
      "loss": 4.8419,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885990422985882e-05,
      "loss": 4.8419,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88515182823483e-05,
      "loss": 4.8421,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.884313233483778e-05,
      "loss": 4.8245,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.883474638732726e-05,
      "loss": 4.8258,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.882636043981674e-05,
      "loss": 4.8207,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.881797449230622e-05,
      "loss": 4.8109,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88095885447957e-05,
      "loss": 4.825,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880121897608891e-05,
      "loss": 4.8092,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879283302857839e-05,
      "loss": 4.7991,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.87844634598716e-05,
      "loss": 4.8154,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877607751236109e-05,
      "loss": 4.7976,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.87677079436543e-05,
      "loss": 4.7951,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875932199614378e-05,
      "loss": 4.799,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875093604863326e-05,
      "loss": 4.7913,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.7542266845703125,
      "eval_runtime": 305.3018,
      "eval_samples_per_second": 1249.881,
      "eval_steps_per_second": 39.06,
      "step": 76320
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.874255010112274e-05,
      "loss": 4.7754,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.873416415361222e-05,
      "loss": 4.7866,
      "step": 77312
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.87257782061017e-05,
      "loss": 4.8014,
      "step": 77824
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.871739225859118e-05,
      "loss": 4.7759,
      "step": 78336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8709006311080657e-05,
      "loss": 4.7881,
      "step": 78848
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8700620363570137e-05,
      "loss": 4.7614,
      "step": 79360
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8692234416059616e-05,
      "loss": 4.7783,
      "step": 79872
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8683848468549096e-05,
      "loss": 4.7561,
      "step": 80384
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8675478899842306e-05,
      "loss": 4.7648,
      "step": 80896
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8667092952331785e-05,
      "loss": 4.765,
      "step": 81408
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865870700482127e-05,
      "loss": 4.769,
      "step": 81920
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865032105731075e-05,
      "loss": 4.7632,
      "step": 82432
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8641935109800225e-05,
      "loss": 4.7461,
      "step": 82944
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8633549162289705e-05,
      "loss": 4.7442,
      "step": 83456
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8625163214779185e-05,
      "loss": 4.7382,
      "step": 83968
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8616777267268665e-05,
      "loss": 4.7423,
      "step": 84480
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8608391319758145e-05,
      "loss": 4.7559,
      "step": 84992
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8600021751051354e-05,
      "loss": 4.7366,
      "step": 85504
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8591635803540834e-05,
      "loss": 4.7437,
      "step": 86016
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8583249856030314e-05,
      "loss": 4.7549,
      "step": 86528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8574863908519794e-05,
      "loss": 4.7309,
      "step": 87040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.856649433981301e-05,
      "loss": 4.726,
      "step": 87552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.855810839230249e-05,
      "loss": 4.7266,
      "step": 88064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.854972244479197e-05,
      "loss": 4.7373,
      "step": 88576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.854133649728145e-05,
      "loss": 4.7181,
      "step": 89088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.853295054977093e-05,
      "loss": 4.7208,
      "step": 89600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.852458098106414e-05,
      "loss": 4.718,
      "step": 90112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.851619503355362e-05,
      "loss": 4.717,
      "step": 90624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.85078090860431e-05,
      "loss": 4.6982,
      "step": 91136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849942313853258e-05,
      "loss": 4.7036,
      "step": 91648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849103719102206e-05,
      "loss": 4.7006,
      "step": 92160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.848265124351154e-05,
      "loss": 4.7079,
      "step": 92672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.847426529600102e-05,
      "loss": 4.7124,
      "step": 93184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.846589572729423e-05,
      "loss": 4.6995,
      "step": 93696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.845750977978371e-05,
      "loss": 4.6965,
      "step": 94208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8449123832273194e-05,
      "loss": 4.7056,
      "step": 94720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8440737884762674e-05,
      "loss": 4.6788,
      "step": 95232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8432351937252154e-05,
      "loss": 4.6874,
      "step": 95744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8423965989741634e-05,
      "loss": 4.6753,
      "step": 96256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8415580042231114e-05,
      "loss": 4.6816,
      "step": 96768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.840719409472059e-05,
      "loss": 4.674,
      "step": 97280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839880814721007e-05,
      "loss": 4.6878,
      "step": 97792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839043857850328e-05,
      "loss": 4.6702,
      "step": 98304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.838205263099276e-05,
      "loss": 4.6765,
      "step": 98816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8373666683482236e-05,
      "loss": 4.6707,
      "step": 99328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8365280735971716e-05,
      "loss": 4.676,
      "step": 99840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8356894788461196e-05,
      "loss": 4.6713,
      "step": 100352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8348508840950676e-05,
      "loss": 4.6614,
      "step": 100864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8340122893440156e-05,
      "loss": 4.649,
      "step": 101376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.833175332473337e-05,
      "loss": 4.673,
      "step": 101888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.832336737722285e-05,
      "loss": 4.6592,
      "step": 102400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.831498142971233e-05,
      "loss": 4.6515,
      "step": 102912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.830659548220181e-05,
      "loss": 4.6462,
      "step": 103424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.829820953469129e-05,
      "loss": 4.6539,
      "step": 103936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.82898399659845e-05,
      "loss": 4.6432,
      "step": 104448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.828145401847398e-05,
      "loss": 4.6612,
      "step": 104960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.827306807096346e-05,
      "loss": 4.6428,
      "step": 105472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.826468212345294e-05,
      "loss": 4.64,
      "step": 105984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.825629617594242e-05,
      "loss": 4.6505,
      "step": 106496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.82479102284319e-05,
      "loss": 4.6292,
      "step": 107008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823952428092138e-05,
      "loss": 4.6244,
      "step": 107520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823113833341086e-05,
      "loss": 4.635,
      "step": 108032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8222768764704077e-05,
      "loss": 4.6161,
      "step": 108544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8214399195997286e-05,
      "loss": 4.6236,
      "step": 109056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8206013248486766e-05,
      "loss": 4.6339,
      "step": 109568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8197627300976246e-05,
      "loss": 4.6395,
      "step": 110080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8189257732269455e-05,
      "loss": 4.6243,
      "step": 110592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8180871784758935e-05,
      "loss": 4.6161,
      "step": 111104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8172502216052144e-05,
      "loss": 4.6073,
      "step": 111616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8164116268541624e-05,
      "loss": 4.6283,
      "step": 112128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8155730321031104e-05,
      "loss": 4.6211,
      "step": 112640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8147344373520584e-05,
      "loss": 4.6226,
      "step": 113152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8138958426010064e-05,
      "loss": 4.6207,
      "step": 113664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813057247849955e-05,
      "loss": 4.629,
      "step": 114176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.812218653098903e-05,
      "loss": 4.6238,
      "step": 114688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.811380058347851e-05,
      "loss": 4.6028,
      "step": 115200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.810541463596799e-05,
      "loss": 4.6123,
      "step": 115712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.809702868845747e-05,
      "loss": 4.6107,
      "step": 116224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808864274094695e-05,
      "loss": 4.5939,
      "step": 116736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808025679343642e-05,
      "loss": 4.6228,
      "step": 117248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.80718708459259e-05,
      "loss": 4.6071,
      "step": 117760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.806348489841538e-05,
      "loss": 4.6119,
      "step": 118272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.80551153297086e-05,
      "loss": 4.6095,
      "step": 118784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.804672938219807e-05,
      "loss": 4.5837,
      "step": 119296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803835981349129e-05,
      "loss": 4.5998,
      "step": 119808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.802997386598077e-05,
      "loss": 4.5992,
      "step": 120320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.802158791847025e-05,
      "loss": 4.5989,
      "step": 120832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.801320197095973e-05,
      "loss": 4.5975,
      "step": 121344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.800481602344921e-05,
      "loss": 4.5933,
      "step": 121856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.799643007593869e-05,
      "loss": 4.5922,
      "step": 122368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.798804412842817e-05,
      "loss": 4.5905,
      "step": 122880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797965818091765e-05,
      "loss": 4.5755,
      "step": 123392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797127223340713e-05,
      "loss": 4.5825,
      "step": 123904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.796288628589661e-05,
      "loss": 4.5849,
      "step": 124416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.795450033838609e-05,
      "loss": 4.5893,
      "step": 124928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.79461307696793e-05,
      "loss": 4.5769,
      "step": 125440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.793774482216878e-05,
      "loss": 4.5694,
      "step": 125952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792935887465826e-05,
      "loss": 4.57,
      "step": 126464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792097292714774e-05,
      "loss": 4.5816,
      "step": 126976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7912586979637217e-05,
      "loss": 4.5785,
      "step": 127488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7904201032126697e-05,
      "loss": 4.5719,
      "step": 128000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.789581508461618e-05,
      "loss": 4.5632,
      "step": 128512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.788742913710566e-05,
      "loss": 4.557,
      "step": 129024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.787905956839887e-05,
      "loss": 4.5668,
      "step": 129536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.787067362088835e-05,
      "loss": 4.5523,
      "step": 130048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.786228767337783e-05,
      "loss": 4.557,
      "step": 130560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.785390172586731e-05,
      "loss": 4.5582,
      "step": 131072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7845515778356785e-05,
      "loss": 4.5585,
      "step": 131584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.783714620965e-05,
      "loss": 4.5573,
      "step": 132096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782876026213948e-05,
      "loss": 4.5602,
      "step": 132608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782037431462896e-05,
      "loss": 4.549,
      "step": 133120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7811988367118434e-05,
      "loss": 4.5513,
      "step": 133632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.780360241960792e-05,
      "loss": 4.5479,
      "step": 134144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.779523285090114e-05,
      "loss": 4.5426,
      "step": 134656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.778684690339061e-05,
      "loss": 4.5576,
      "step": 135168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777846095588009e-05,
      "loss": 4.5527,
      "step": 135680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777007500836957e-05,
      "loss": 4.5517,
      "step": 136192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.776168906085905e-05,
      "loss": 4.5423,
      "step": 136704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.775331949215226e-05,
      "loss": 4.549,
      "step": 137216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.774493354464174e-05,
      "loss": 4.5478,
      "step": 137728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.773654759713122e-05,
      "loss": 4.5354,
      "step": 138240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.77281616496207e-05,
      "loss": 4.5538,
      "step": 138752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771979208091391e-05,
      "loss": 4.5402,
      "step": 139264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771140613340339e-05,
      "loss": 4.5434,
      "step": 139776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7703020185892875e-05,
      "loss": 4.5264,
      "step": 140288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7694634238382355e-05,
      "loss": 4.5276,
      "step": 140800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7686248290871835e-05,
      "loss": 4.5213,
      "step": 141312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7677862343361315e-05,
      "loss": 4.5242,
      "step": 141824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7669476395850795e-05,
      "loss": 4.5212,
      "step": 142336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7661090448340274e-05,
      "loss": 4.5369,
      "step": 142848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7652704500829754e-05,
      "loss": 4.5394,
      "step": 143360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.764435131092669e-05,
      "loss": 4.5277,
      "step": 143872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.763596536341617e-05,
      "loss": 4.5213,
      "step": 144384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.762757941590565e-05,
      "loss": 4.5263,
      "step": 144896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761919346839513e-05,
      "loss": 4.5266,
      "step": 145408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761080752088461e-05,
      "loss": 4.5347,
      "step": 145920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.760242157337409e-05,
      "loss": 4.5289,
      "step": 146432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.759403562586357e-05,
      "loss": 4.5215,
      "step": 146944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.758564967835306e-05,
      "loss": 4.5136,
      "step": 147456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.757726373084254e-05,
      "loss": 4.5216,
      "step": 147968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756887778333202e-05,
      "loss": 4.5135,
      "step": 148480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756050821462523e-05,
      "loss": 4.5187,
      "step": 148992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.755212226711471e-05,
      "loss": 4.5139,
      "step": 149504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.754373631960419e-05,
      "loss": 4.5094,
      "step": 150016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.753535037209367e-05,
      "loss": 4.5215,
      "step": 150528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.752696442458315e-05,
      "loss": 4.509,
      "step": 151040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751857847707262e-05,
      "loss": 4.5043,
      "step": 151552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75101925295621e-05,
      "loss": 4.5076,
      "step": 152064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.750180658205158e-05,
      "loss": 4.5039,
      "step": 152576
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.4751152992248535,
      "eval_runtime": 300.7316,
      "eval_samples_per_second": 1268.876,
      "eval_steps_per_second": 39.653,
      "step": 152640
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.749342063454106e-05,
      "loss": 4.4902,
      "step": 153088
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.748503468703054e-05,
      "loss": 4.5029,
      "step": 153600
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.747664873952003e-05,
      "loss": 4.5188,
      "step": 154112
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.746826279200951e-05,
      "loss": 4.498,
      "step": 154624
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.745987684449899e-05,
      "loss": 4.5092,
      "step": 155136
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.745149089698847e-05,
      "loss": 4.4903,
      "step": 155648
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.744310494947795e-05,
      "loss": 4.5046,
      "step": 156160
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.743471900196743e-05,
      "loss": 4.48,
      "step": 156672
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.742633305445691e-05,
      "loss": 4.4979,
      "step": 157184
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.741794710694639e-05,
      "loss": 4.4964,
      "step": 157696
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.740956115943586e-05,
      "loss": 4.4971,
      "step": 158208
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.740117521192534e-05,
      "loss": 4.505,
      "step": 158720
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7392805643218556e-05,
      "loss": 4.4806,
      "step": 159232
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7384419695708036e-05,
      "loss": 4.4802,
      "step": 159744
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.737603374819751e-05,
      "loss": 4.4779,
      "step": 160256
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.736764780068699e-05,
      "loss": 4.4864,
      "step": 160768
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7359261853176476e-05,
      "loss": 4.4925,
      "step": 161280
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7350875905665956e-05,
      "loss": 4.4838,
      "step": 161792
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7342489958155436e-05,
      "loss": 4.488,
      "step": 162304
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7334104010644916e-05,
      "loss": 4.5031,
      "step": 162816
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7325718063134396e-05,
      "loss": 4.4811,
      "step": 163328
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7317332115623876e-05,
      "loss": 4.4741,
      "step": 163840
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7308946168113356e-05,
      "loss": 4.4801,
      "step": 164352
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7300576599406565e-05,
      "loss": 4.4903,
      "step": 164864
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7292190651896045e-05,
      "loss": 4.4735,
      "step": 165376
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7283804704385525e-05,
      "loss": 4.4747,
      "step": 165888
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7275418756875005e-05,
      "loss": 4.4739,
      "step": 166400
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7267049188168214e-05,
      "loss": 4.4738,
      "step": 166912
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7258663240657694e-05,
      "loss": 4.4586,
      "step": 167424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7250277293147174e-05,
      "loss": 4.4651,
      "step": 167936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.724190772444039e-05,
      "loss": 4.4669,
      "step": 168448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.72335381557336e-05,
      "loss": 4.4683,
      "step": 168960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.722515220822308e-05,
      "loss": 4.4806,
      "step": 169472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.721676626071256e-05,
      "loss": 4.4614,
      "step": 169984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.720838031320204e-05,
      "loss": 4.4646,
      "step": 170496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.719999436569152e-05,
      "loss": 4.472,
      "step": 171008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7191608418181e-05,
      "loss": 4.4545,
      "step": 171520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.718322247067048e-05,
      "loss": 4.4537,
      "step": 172032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.717483652315996e-05,
      "loss": 4.4503,
      "step": 172544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.716645057564944e-05,
      "loss": 4.4558,
      "step": 173056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.715806462813892e-05,
      "loss": 4.4467,
      "step": 173568
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.714969505943213e-05,
      "loss": 4.4659,
      "step": 174080
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7141309111921614e-05,
      "loss": 4.4436,
      "step": 174592
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7132923164411094e-05,
      "loss": 4.4603,
      "step": 175104
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7124537216900574e-05,
      "loss": 4.451,
      "step": 175616
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.711616764819378e-05,
      "loss": 4.4524,
      "step": 176128
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.710778170068326e-05,
      "loss": 4.4563,
      "step": 176640
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.709939575317274e-05,
      "loss": 4.4415,
      "step": 177152
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.709100980566222e-05,
      "loss": 4.4333,
      "step": 177664
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7082623858151696e-05,
      "loss": 4.4575,
      "step": 178176
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7074237910641176e-05,
      "loss": 4.4479,
      "step": 178688
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7065851963130656e-05,
      "loss": 4.4375,
      "step": 179200
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7057466015620136e-05,
      "loss": 4.4397,
      "step": 179712
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.704909644691335e-05,
      "loss": 4.4458,
      "step": 180224
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.704071049940283e-05,
      "loss": 4.4294,
      "step": 180736
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.703232455189231e-05,
      "loss": 4.4499,
      "step": 181248
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.702393860438179e-05,
      "loss": 4.4353,
      "step": 181760
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.7015569035675e-05,
      "loss": 4.433,
      "step": 182272
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.700718308816448e-05,
      "loss": 4.446,
      "step": 182784
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.699879714065396e-05,
      "loss": 4.4254,
      "step": 183296
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.699041119314344e-05,
      "loss": 4.4227,
      "step": 183808
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.698204162443665e-05,
      "loss": 4.4353,
      "step": 184320
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.697365567692613e-05,
      "loss": 4.418,
      "step": 184832
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.696526972941561e-05,
      "loss": 4.4217,
      "step": 185344
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.695688378190509e-05,
      "loss": 4.4349,
      "step": 185856
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.694849783439457e-05,
      "loss": 4.4392,
      "step": 186368
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.694011188688405e-05,
      "loss": 4.4227,
      "step": 186880
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.693172593937353e-05,
      "loss": 4.4215,
      "step": 187392
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6923356370666746e-05,
      "loss": 4.41,
      "step": 187904
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6914970423156226e-05,
      "loss": 4.4345,
      "step": 188416
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6906584475645706e-05,
      "loss": 4.4252,
      "step": 188928
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6898198528135186e-05,
      "loss": 4.4321,
      "step": 189440
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6889812580624665e-05,
      "loss": 4.4262,
      "step": 189952
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6881426633114145e-05,
      "loss": 4.4357,
      "step": 190464
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6873040685603625e-05,
      "loss": 4.436,
      "step": 190976
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6864671116896834e-05,
      "loss": 4.41,
      "step": 191488
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6856285169386314e-05,
      "loss": 4.4224,
      "step": 192000
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6847899221875794e-05,
      "loss": 4.427,
      "step": 192512
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6839513274365274e-05,
      "loss": 4.4073,
      "step": 193024
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6831143705658483e-05,
      "loss": 4.4333,
      "step": 193536
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.682275775814797e-05,
      "loss": 4.4229,
      "step": 194048
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.681437181063745e-05,
      "loss": 4.4277,
      "step": 194560
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.680598586312693e-05,
      "loss": 4.4274,
      "step": 195072
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.679759991561641e-05,
      "loss": 4.4019,
      "step": 195584
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678921396810588e-05,
      "loss": 4.4177,
      "step": 196096
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678082802059536e-05,
      "loss": 4.4166,
      "step": 196608
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.677244207308484e-05,
      "loss": 4.4209,
      "step": 197120
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.676407250437806e-05,
      "loss": 4.4153,
      "step": 197632
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.675570293567127e-05,
      "loss": 4.4163,
      "step": 198144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.674731698816075e-05,
      "loss": 4.413,
      "step": 198656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.673893104065023e-05,
      "loss": 4.4119,
      "step": 199168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.673054509313971e-05,
      "loss": 4.3993,
      "step": 199680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6722175524432924e-05,
      "loss": 4.4061,
      "step": 200192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6713789576922404e-05,
      "loss": 4.4115,
      "step": 200704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6705403629411884e-05,
      "loss": 4.414,
      "step": 201216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.669701768190136e-05,
      "loss": 4.405,
      "step": 201728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.668863173439084e-05,
      "loss": 4.3936,
      "step": 202240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.668024578688032e-05,
      "loss": 4.3984,
      "step": 202752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.66718598393698e-05,
      "loss": 4.4094,
      "step": 203264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6663490270663006e-05,
      "loss": 4.4033,
      "step": 203776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6655104323152486e-05,
      "loss": 4.4018,
      "step": 204288
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6646718375641966e-05,
      "loss": 4.3973,
      "step": 204800
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6638332428131446e-05,
      "loss": 4.3856,
      "step": 205312
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6629946480620926e-05,
      "loss": 4.4029,
      "step": 205824
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6621560533110406e-05,
      "loss": 4.3806,
      "step": 206336
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.661317458559989e-05,
      "loss": 4.3945,
      "step": 206848
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.66048050168931e-05,
      "loss": 4.3852,
      "step": 207360
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.659641906938258e-05,
      "loss": 4.4008,
      "step": 207872
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.658803312187206e-05,
      "loss": 4.3877,
      "step": 208384
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657964717436154e-05,
      "loss": 4.3982,
      "step": 208896
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657127760565475e-05,
      "loss": 4.3862,
      "step": 209408
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.656290803694796e-05,
      "loss": 4.3842,
      "step": 209920
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.655452208943744e-05,
      "loss": 4.392,
      "step": 210432
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.654613614192692e-05,
      "loss": 4.3783,
      "step": 210944
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.65377501944164e-05,
      "loss": 4.3983,
      "step": 211456
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6529380625709615e-05,
      "loss": 4.3929,
      "step": 211968
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6520994678199095e-05,
      "loss": 4.3932,
      "step": 212480
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6512608730688575e-05,
      "loss": 4.3831,
      "step": 212992
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6504222783178055e-05,
      "loss": 4.389,
      "step": 213504
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6495836835667535e-05,
      "loss": 4.3933,
      "step": 214016
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6487467266960744e-05,
      "loss": 4.3733,
      "step": 214528
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6479081319450224e-05,
      "loss": 4.3967,
      "step": 215040
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6470695371939704e-05,
      "loss": 4.3854,
      "step": 215552
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6462309424429184e-05,
      "loss": 4.3857,
      "step": 216064
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6453923476918664e-05,
      "loss": 4.3757,
      "step": 216576
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6445537529408144e-05,
      "loss": 4.3776,
      "step": 217088
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.643716796070135e-05,
      "loss": 4.3632,
      "step": 217600
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.642878201319083e-05,
      "loss": 4.3743,
      "step": 218112
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.642039606568031e-05,
      "loss": 4.37,
      "step": 218624
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.641202649697353e-05,
      "loss": 4.3845,
      "step": 219136
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.640364054946301e-05,
      "loss": 4.387,
      "step": 219648
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.639525460195249e-05,
      "loss": 4.3773,
      "step": 220160
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.638686865444197e-05,
      "loss": 4.3721,
      "step": 220672
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.637848270693145e-05,
      "loss": 4.3741,
      "step": 221184
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.637009675942093e-05,
      "loss": 4.3784,
      "step": 221696
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.636171081191041e-05,
      "loss": 4.3838,
      "step": 222208
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.635332486439989e-05,
      "loss": 4.3785,
      "step": 222720
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.634493891688936e-05,
      "loss": 4.3695,
      "step": 223232
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.633655296937884e-05,
      "loss": 4.3736,
      "step": 223744
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.632816702186832e-05,
      "loss": 4.3762,
      "step": 224256
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.63197810743578e-05,
      "loss": 4.3651,
      "step": 224768
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.631141150565102e-05,
      "loss": 4.3739,
      "step": 225280
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.63030255581405e-05,
      "loss": 4.3682,
      "step": 225792
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.629463961062998e-05,
      "loss": 4.3651,
      "step": 226304
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.628625366311946e-05,
      "loss": 4.3786,
      "step": 226816
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6277884094412666e-05,
      "loss": 4.366,
      "step": 227328
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6269498146902146e-05,
      "loss": 4.3607,
      "step": 227840
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.626112857819536e-05,
      "loss": 4.3642,
      "step": 228352
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6252742630684835e-05,
      "loss": 4.357,
      "step": 228864
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.340610504150391,
      "eval_runtime": 301.3171,
      "eval_samples_per_second": 1266.41,
      "eval_steps_per_second": 39.576,
      "step": 228960
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 9.420222446476171e+16,
  "trial_name": null,
  "trial_params": null
}