{
  "best_metric": 4.003251552581787,
  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/npi-sent-neg/lstm/3/checkpoints/checkpoint-1373760",
  "epoch": 1.0250006060157382,
  "eval_steps": 10,
  "global_step": 1373760,
  "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": 10.8187,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999161405248948e-05,
      "loss": 7.5487,
      "step": 512
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.998322810497896e-05,
      "loss": 7.0597,
      "step": 1024
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.997484215746844e-05,
      "loss": 6.9956,
      "step": 1536
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.996645620995792e-05,
      "loss": 6.9491,
      "step": 2048
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99580702624474e-05,
      "loss": 6.9236,
      "step": 2560
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994968431493688e-05,
      "loss": 6.7673,
      "step": 3072
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.994129836742636e-05,
      "loss": 6.6605,
      "step": 3584
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.993291241991584e-05,
      "loss": 6.5615,
      "step": 4096
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.992454285120906e-05,
      "loss": 6.4811,
      "step": 4608
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.991615690369854e-05,
      "loss": 6.4166,
      "step": 5120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.990777095618801e-05,
      "loss": 6.3434,
      "step": 5632
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989938500867749e-05,
      "loss": 6.2739,
      "step": 6144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.989099906116697e-05,
      "loss": 6.1955,
      "step": 6656
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.988261311365645e-05,
      "loss": 6.1487,
      "step": 7168
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.987422716614593e-05,
      "loss": 6.0869,
      "step": 7680
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.986584121863541e-05,
      "loss": 6.044,
      "step": 8192
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.985745527112489e-05,
      "loss": 5.9981,
      "step": 8704
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984906932361437e-05,
      "loss": 5.9589,
      "step": 9216
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.984068337610385e-05,
      "loss": 5.917,
      "step": 9728
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.983229742859333e-05,
      "loss": 5.8838,
      "step": 10240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.982391148108281e-05,
      "loss": 5.8415,
      "step": 10752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.981552553357229e-05,
      "loss": 5.8082,
      "step": 11264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9807155964865506e-05,
      "loss": 5.7764,
      "step": 11776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9798770017354986e-05,
      "loss": 5.7561,
      "step": 12288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9790384069844466e-05,
      "loss": 5.7185,
      "step": 12800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9781998122333946e-05,
      "loss": 5.6943,
      "step": 13312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9773612174823426e-05,
      "loss": 5.6664,
      "step": 13824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9765242606116635e-05,
      "loss": 5.6364,
      "step": 14336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9756856658606115e-05,
      "loss": 5.6261,
      "step": 14848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9748470711095595e-05,
      "loss": 5.6005,
      "step": 15360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9740101142388804e-05,
      "loss": 5.5925,
      "step": 15872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9731715194878284e-05,
      "loss": 5.556,
      "step": 16384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9723329247367764e-05,
      "loss": 5.5449,
      "step": 16896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9714943299857244e-05,
      "loss": 5.5334,
      "step": 17408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9706557352346724e-05,
      "loss": 5.5162,
      "step": 17920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9698171404836204e-05,
      "loss": 5.4948,
      "step": 18432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9689785457325684e-05,
      "loss": 5.46,
      "step": 18944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9681399509815164e-05,
      "loss": 5.4632,
      "step": 19456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9673013562304644e-05,
      "loss": 5.429,
      "step": 19968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.966464399359785e-05,
      "loss": 5.4315,
      "step": 20480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.965625804608733e-05,
      "loss": 5.4151,
      "step": 20992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.964787209857681e-05,
      "loss": 5.4045,
      "step": 21504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.963948615106629e-05,
      "loss": 5.3846,
      "step": 22016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96311165823595e-05,
      "loss": 5.3785,
      "step": 22528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.962274701365272e-05,
      "loss": 5.3782,
      "step": 23040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.96143610661422e-05,
      "loss": 5.3608,
      "step": 23552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.960597511863168e-05,
      "loss": 5.3561,
      "step": 24064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.959758917112116e-05,
      "loss": 5.3276,
      "step": 24576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958920322361064e-05,
      "loss": 5.3134,
      "step": 25088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.958081727610012e-05,
      "loss": 5.3225,
      "step": 25600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.957244770739333e-05,
      "loss": 5.3091,
      "step": 26112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.956406175988281e-05,
      "loss": 5.3008,
      "step": 26624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.955567581237229e-05,
      "loss": 5.2703,
      "step": 27136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.954728986486177e-05,
      "loss": 5.286,
      "step": 27648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.953890391735125e-05,
      "loss": 5.2603,
      "step": 28160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9530517969840727e-05,
      "loss": 5.2814,
      "step": 28672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9522132022330207e-05,
      "loss": 5.2324,
      "step": 29184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9513746074819686e-05,
      "loss": 5.2438,
      "step": 29696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9505360127309166e-05,
      "loss": 5.2241,
      "step": 30208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9496974179798646e-05,
      "loss": 5.2122,
      "step": 30720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948858823228813e-05,
      "loss": 5.2144,
      "step": 31232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.948021866358134e-05,
      "loss": 5.2016,
      "step": 31744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.947183271607082e-05,
      "loss": 5.1798,
      "step": 32256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.946346314736403e-05,
      "loss": 5.1807,
      "step": 32768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.945509357865724e-05,
      "loss": 5.2013,
      "step": 33280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.944672400995045e-05,
      "loss": 5.1745,
      "step": 33792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.943833806243993e-05,
      "loss": 5.1634,
      "step": 34304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942995211492941e-05,
      "loss": 5.152,
      "step": 34816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.942156616741889e-05,
      "loss": 5.1378,
      "step": 35328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.941318021990837e-05,
      "loss": 5.1515,
      "step": 35840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9404794272397856e-05,
      "loss": 5.1429,
      "step": 36352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9396408324887336e-05,
      "loss": 5.1357,
      "step": 36864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9388022377376816e-05,
      "loss": 5.1418,
      "step": 37376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9379636429866296e-05,
      "loss": 5.1269,
      "step": 37888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9371250482355776e-05,
      "loss": 5.1211,
      "step": 38400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9362864534845256e-05,
      "loss": 5.0985,
      "step": 38912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.935447858733473e-05,
      "loss": 5.099,
      "step": 39424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.934609263982421e-05,
      "loss": 5.0898,
      "step": 39936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9337723071117425e-05,
      "loss": 5.0805,
      "step": 40448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9329337123606905e-05,
      "loss": 5.0804,
      "step": 40960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.932095117609638e-05,
      "loss": 5.0785,
      "step": 41472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.931256522858586e-05,
      "loss": 5.0806,
      "step": 41984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.930417928107534e-05,
      "loss": 5.0692,
      "step": 42496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9295793333564825e-05,
      "loss": 5.0437,
      "step": 43008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9287423764858034e-05,
      "loss": 5.0544,
      "step": 43520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9279037817347514e-05,
      "loss": 5.0547,
      "step": 44032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9270651869836994e-05,
      "loss": 5.0554,
      "step": 44544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9262265922326474e-05,
      "loss": 5.0337,
      "step": 45056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9253879974815954e-05,
      "loss": 5.037,
      "step": 45568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.924551040610916e-05,
      "loss": 5.0288,
      "step": 46080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.923712445859864e-05,
      "loss": 5.0207,
      "step": 46592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.922873851108812e-05,
      "loss": 5.0154,
      "step": 47104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.92203525635776e-05,
      "loss": 5.0115,
      "step": 47616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.921198299487081e-05,
      "loss": 4.9989,
      "step": 48128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.920361342616403e-05,
      "loss": 5.0067,
      "step": 48640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.919522747865351e-05,
      "loss": 4.9898,
      "step": 49152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.918684153114299e-05,
      "loss": 4.9809,
      "step": 49664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.917845558363247e-05,
      "loss": 4.9775,
      "step": 50176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9170086014925676e-05,
      "loss": 4.9819,
      "step": 50688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9161700067415156e-05,
      "loss": 4.9759,
      "step": 51200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9153314119904636e-05,
      "loss": 4.9611,
      "step": 51712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9144928172394116e-05,
      "loss": 4.9623,
      "step": 52224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9136542224883596e-05,
      "loss": 4.9493,
      "step": 52736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9128156277373076e-05,
      "loss": 4.9535,
      "step": 53248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9119770329862556e-05,
      "loss": 4.9288,
      "step": 53760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9111384382352036e-05,
      "loss": 4.9392,
      "step": 54272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9102998434841516e-05,
      "loss": 4.9317,
      "step": 54784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9094612487330996e-05,
      "loss": 4.9285,
      "step": 55296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.908624291862421e-05,
      "loss": 4.9313,
      "step": 55808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.907787334991742e-05,
      "loss": 4.9255,
      "step": 56320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.90694874024069e-05,
      "loss": 4.904,
      "step": 56832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.906110145489638e-05,
      "loss": 4.913,
      "step": 57344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.905273188618959e-05,
      "loss": 4.9153,
      "step": 57856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.904434593867907e-05,
      "loss": 4.9072,
      "step": 58368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.903595999116855e-05,
      "loss": 4.907,
      "step": 58880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.902757404365803e-05,
      "loss": 4.8983,
      "step": 59392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901918809614751e-05,
      "loss": 4.8945,
      "step": 59904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.901080214863699e-05,
      "loss": 4.8826,
      "step": 60416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.900241620112647e-05,
      "loss": 4.8892,
      "step": 60928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.899403025361595e-05,
      "loss": 4.8845,
      "step": 61440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.898564430610543e-05,
      "loss": 4.8695,
      "step": 61952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.897725835859491e-05,
      "loss": 4.8716,
      "step": 62464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8968872411084396e-05,
      "loss": 4.8665,
      "step": 62976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.896048646357387e-05,
      "loss": 4.8736,
      "step": 63488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.895210051606335e-05,
      "loss": 4.8501,
      "step": 64000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.894371456855283e-05,
      "loss": 4.8564,
      "step": 64512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.893532862104231e-05,
      "loss": 4.8429,
      "step": 65024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.892695905233552e-05,
      "loss": 4.851,
      "step": 65536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8918589483628734e-05,
      "loss": 4.8394,
      "step": 66048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.8910203536118214e-05,
      "loss": 4.8539,
      "step": 66560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.890181758860769e-05,
      "loss": 4.8413,
      "step": 67072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88934480199009e-05,
      "loss": 4.8344,
      "step": 67584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.888506207239038e-05,
      "loss": 4.833,
      "step": 68096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.887667612487986e-05,
      "loss": 4.8279,
      "step": 68608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.886829017736934e-05,
      "loss": 4.8365,
      "step": 69120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.885990422985882e-05,
      "loss": 4.8333,
      "step": 69632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.88515182823483e-05,
      "loss": 4.8244,
      "step": 70144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.884313233483778e-05,
      "loss": 4.8187,
      "step": 70656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.883474638732726e-05,
      "loss": 4.8202,
      "step": 71168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.882637681862047e-05,
      "loss": 4.8096,
      "step": 71680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.881799087110995e-05,
      "loss": 4.7973,
      "step": 72192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880960492359943e-05,
      "loss": 4.8054,
      "step": 72704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.880121897608891e-05,
      "loss": 4.7966,
      "step": 73216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.879284940738212e-05,
      "loss": 4.7888,
      "step": 73728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.87844634598716e-05,
      "loss": 4.79,
      "step": 74240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.877607751236109e-05,
      "loss": 4.7877,
      "step": 74752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.876769156485057e-05,
      "loss": 4.7912,
      "step": 75264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875932199614378e-05,
      "loss": 4.7911,
      "step": 75776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.875093604863326e-05,
      "loss": 4.7787,
      "step": 76288
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.7453789710998535,
      "eval_runtime": 301.8396,
      "eval_samples_per_second": 1264.218,
      "eval_steps_per_second": 39.508,
      "step": 76320
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.874255010112274e-05,
      "loss": 4.769,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.873416415361222e-05,
      "loss": 4.7668,
      "step": 77312
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.87257782061017e-05,
      "loss": 4.7781,
      "step": 77824
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.871739225859118e-05,
      "loss": 4.7697,
      "step": 78336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8709006311080657e-05,
      "loss": 4.7717,
      "step": 78848
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8700620363570137e-05,
      "loss": 4.7534,
      "step": 79360
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8692234416059616e-05,
      "loss": 4.7553,
      "step": 79872
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8683848468549096e-05,
      "loss": 4.7501,
      "step": 80384
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8675478899842306e-05,
      "loss": 4.7553,
      "step": 80896
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8667092952331785e-05,
      "loss": 4.754,
      "step": 81408
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865870700482127e-05,
      "loss": 4.7511,
      "step": 81920
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.865032105731075e-05,
      "loss": 4.7574,
      "step": 82432
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.864195148860396e-05,
      "loss": 4.7257,
      "step": 82944
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.863356554109344e-05,
      "loss": 4.7374,
      "step": 83456
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.862517959358292e-05,
      "loss": 4.7316,
      "step": 83968
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.86167936460724e-05,
      "loss": 4.7222,
      "step": 84480
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8608407698561874e-05,
      "loss": 4.7355,
      "step": 84992
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8600021751051354e-05,
      "loss": 4.7259,
      "step": 85504
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8591635803540834e-05,
      "loss": 4.7222,
      "step": 86016
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.858326623483405e-05,
      "loss": 4.7441,
      "step": 86528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.857489666612726e-05,
      "loss": 4.7242,
      "step": 87040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.856651071861674e-05,
      "loss": 4.7179,
      "step": 87552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8558141149909955e-05,
      "loss": 4.7171,
      "step": 88064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8549755202399435e-05,
      "loss": 4.7316,
      "step": 88576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8541369254888915e-05,
      "loss": 4.7105,
      "step": 89088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8532983307378395e-05,
      "loss": 4.7052,
      "step": 89600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.8524597359867875e-05,
      "loss": 4.7058,
      "step": 90112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.851621141235735e-05,
      "loss": 4.7019,
      "step": 90624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.850782546484683e-05,
      "loss": 4.7012,
      "step": 91136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.849943951733631e-05,
      "loss": 4.6945,
      "step": 91648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8491069948629524e-05,
      "loss": 4.7137,
      "step": 92160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8482684001119e-05,
      "loss": 4.6941,
      "step": 92672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.847429805360848e-05,
      "loss": 4.6929,
      "step": 93184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8465912106097964e-05,
      "loss": 4.6948,
      "step": 93696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.845754253739117e-05,
      "loss": 4.6935,
      "step": 94208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.844915658988065e-05,
      "loss": 4.6874,
      "step": 94720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.844077064237013e-05,
      "loss": 4.6686,
      "step": 95232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.843238469485961e-05,
      "loss": 4.6766,
      "step": 95744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.842399874734909e-05,
      "loss": 4.6618,
      "step": 96256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.841561279983857e-05,
      "loss": 4.6796,
      "step": 96768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.840722685232805e-05,
      "loss": 4.6675,
      "step": 97280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839884090481753e-05,
      "loss": 4.6728,
      "step": 97792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.839047133611074e-05,
      "loss": 4.6631,
      "step": 98304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.838208538860022e-05,
      "loss": 4.6676,
      "step": 98816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.83736994410897e-05,
      "loss": 4.6705,
      "step": 99328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.836531349357918e-05,
      "loss": 4.663,
      "step": 99840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8356960303676126e-05,
      "loss": 4.6663,
      "step": 100352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8348574356165606e-05,
      "loss": 4.6531,
      "step": 100864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.834020478745882e-05,
      "loss": 4.6455,
      "step": 101376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8331818839948295e-05,
      "loss": 4.6571,
      "step": 101888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8323432892437775e-05,
      "loss": 4.6515,
      "step": 102400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8315046944927255e-05,
      "loss": 4.6565,
      "step": 102912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8306660997416735e-05,
      "loss": 4.6289,
      "step": 103424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8298275049906215e-05,
      "loss": 4.6521,
      "step": 103936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8289905481199424e-05,
      "loss": 4.6297,
      "step": 104448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8281519533688904e-05,
      "loss": 4.6618,
      "step": 104960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8273133586178384e-05,
      "loss": 4.6242,
      "step": 105472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.826474763866787e-05,
      "loss": 4.6398,
      "step": 105984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.825636169115735e-05,
      "loss": 4.6291,
      "step": 106496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.824797574364683e-05,
      "loss": 4.6216,
      "step": 107008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823960617494004e-05,
      "loss": 4.6257,
      "step": 107520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.823122022742952e-05,
      "loss": 4.6241,
      "step": 108032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8222834279919e-05,
      "loss": 4.6067,
      "step": 108544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.821444833240848e-05,
      "loss": 4.6132,
      "step": 109056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.820606238489796e-05,
      "loss": 4.6288,
      "step": 109568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.819767643738744e-05,
      "loss": 4.6155,
      "step": 110080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.818929048987692e-05,
      "loss": 4.6117,
      "step": 110592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.81809045423664e-05,
      "loss": 4.6075,
      "step": 111104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.817251859485588e-05,
      "loss": 4.5979,
      "step": 111616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.816413264734535e-05,
      "loss": 4.6164,
      "step": 112128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.815574669983483e-05,
      "loss": 4.6106,
      "step": 112640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.814736075232432e-05,
      "loss": 4.6116,
      "step": 113152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8138991183617535e-05,
      "loss": 4.6166,
      "step": 113664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.813060523610701e-05,
      "loss": 4.6133,
      "step": 114176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.812221928859649e-05,
      "loss": 4.6115,
      "step": 114688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8113849719889704e-05,
      "loss": 4.5901,
      "step": 115200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8105463772379184e-05,
      "loss": 4.6017,
      "step": 115712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.809707782486866e-05,
      "loss": 4.5959,
      "step": 116224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808869187735814e-05,
      "loss": 4.5843,
      "step": 116736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.808032230865135e-05,
      "loss": 4.5946,
      "step": 117248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8071936361140827e-05,
      "loss": 4.5944,
      "step": 117760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8063550413630307e-05,
      "loss": 4.6033,
      "step": 118272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8055164466119786e-05,
      "loss": 4.5956,
      "step": 118784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.8046794897413e-05,
      "loss": 4.5747,
      "step": 119296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803840894990248e-05,
      "loss": 4.5844,
      "step": 119808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.803002300239196e-05,
      "loss": 4.5929,
      "step": 120320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.802163705488144e-05,
      "loss": 4.5948,
      "step": 120832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.801325110737092e-05,
      "loss": 4.5843,
      "step": 121344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.80048651598604e-05,
      "loss": 4.5896,
      "step": 121856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.799647921234988e-05,
      "loss": 4.5851,
      "step": 122368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.798810964364309e-05,
      "loss": 4.5725,
      "step": 122880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797972369613257e-05,
      "loss": 4.5782,
      "step": 123392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.797133774862205e-05,
      "loss": 4.5791,
      "step": 123904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.796295180111153e-05,
      "loss": 4.5686,
      "step": 124416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.795456585360101e-05,
      "loss": 4.5834,
      "step": 124928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.794617990609049e-05,
      "loss": 4.5649,
      "step": 125440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.793781033738371e-05,
      "loss": 4.5621,
      "step": 125952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792942438987319e-05,
      "loss": 4.5651,
      "step": 126464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.792103844236267e-05,
      "loss": 4.5721,
      "step": 126976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.791265249485215e-05,
      "loss": 4.5693,
      "step": 127488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.790426654734163e-05,
      "loss": 4.5579,
      "step": 128000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7895880599831107e-05,
      "loss": 4.5561,
      "step": 128512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7887494652320587e-05,
      "loss": 4.5512,
      "step": 129024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7879108704810067e-05,
      "loss": 4.5619,
      "step": 129536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7870739136103276e-05,
      "loss": 4.5383,
      "step": 130048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7862369567396485e-05,
      "loss": 4.5467,
      "step": 130560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7853983619885965e-05,
      "loss": 4.5543,
      "step": 131072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7845597672375445e-05,
      "loss": 4.5489,
      "step": 131584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7837211724864925e-05,
      "loss": 4.5484,
      "step": 132096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782882577735441e-05,
      "loss": 4.5472,
      "step": 132608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.782045620864762e-05,
      "loss": 4.5365,
      "step": 133120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.78120702611371e-05,
      "loss": 4.5444,
      "step": 133632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.780368431362658e-05,
      "loss": 4.5444,
      "step": 134144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.779529836611606e-05,
      "loss": 4.5434,
      "step": 134656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.778691241860554e-05,
      "loss": 4.5474,
      "step": 135168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777852647109502e-05,
      "loss": 4.5466,
      "step": 135680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.777015690238823e-05,
      "loss": 4.5378,
      "step": 136192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.776177095487771e-05,
      "loss": 4.5318,
      "step": 136704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.775338500736719e-05,
      "loss": 4.5394,
      "step": 137216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.774499905985666e-05,
      "loss": 4.5344,
      "step": 137728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.773662949114988e-05,
      "loss": 4.5235,
      "step": 138240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.772824354363936e-05,
      "loss": 4.5318,
      "step": 138752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7719857596128845e-05,
      "loss": 4.5267,
      "step": 139264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.771147164861832e-05,
      "loss": 4.5355,
      "step": 139776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.77030857011078e-05,
      "loss": 4.518,
      "step": 140288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.769469975359728e-05,
      "loss": 4.5272,
      "step": 140800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.768631380608676e-05,
      "loss": 4.5064,
      "step": 141312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.767792785857624e-05,
      "loss": 4.5284,
      "step": 141824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766955828986945e-05,
      "loss": 4.5149,
      "step": 142336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.766117234235893e-05,
      "loss": 4.532,
      "step": 142848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.765278639484841e-05,
      "loss": 4.5243,
      "step": 143360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.764440044733789e-05,
      "loss": 4.5194,
      "step": 143872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7636030878631096e-05,
      "loss": 4.5112,
      "step": 144384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.762764493112058e-05,
      "loss": 4.513,
      "step": 144896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761925898361006e-05,
      "loss": 4.5269,
      "step": 145408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.761087303609954e-05,
      "loss": 4.5295,
      "step": 145920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.760251984619649e-05,
      "loss": 4.5181,
      "step": 146432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.759413389868596e-05,
      "loss": 4.5144,
      "step": 146944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.758574795117544e-05,
      "loss": 4.5117,
      "step": 147456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.757736200366492e-05,
      "loss": 4.5085,
      "step": 147968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75689760561544e-05,
      "loss": 4.5035,
      "step": 148480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.756059010864388e-05,
      "loss": 4.5093,
      "step": 148992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.755220416113336e-05,
      "loss": 4.5035,
      "step": 149504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.754381821362284e-05,
      "loss": 4.4966,
      "step": 150016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.753543226611232e-05,
      "loss": 4.4994,
      "step": 150528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.75270463186018e-05,
      "loss": 4.5001,
      "step": 151040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.751866037109128e-05,
      "loss": 4.5022,
      "step": 151552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7510290802384496e-05,
      "loss": 4.5016,
      "step": 152064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.7501904854873976e-05,
      "loss": 4.4985,
      "step": 152576
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.4658379554748535,
      "eval_runtime": 298.8499,
      "eval_samples_per_second": 1276.865,
      "eval_steps_per_second": 39.903,
      "step": 152640
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7493518907363456e-05,
      "loss": 4.4838,
      "step": 153088
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7485132959852936e-05,
      "loss": 4.4883,
      "step": 153600
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7476747012342416e-05,
      "loss": 4.4969,
      "step": 154112
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7468361064831896e-05,
      "loss": 4.4946,
      "step": 154624
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7459975117321376e-05,
      "loss": 4.4977,
      "step": 155136
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.745158916981085e-05,
      "loss": 4.4775,
      "step": 155648
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.744320322230033e-05,
      "loss": 4.4889,
      "step": 156160
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.743481727478981e-05,
      "loss": 4.4785,
      "step": 156672
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.742643132727929e-05,
      "loss": 4.4889,
      "step": 157184
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.741804537976877e-05,
      "loss": 4.4876,
      "step": 157696
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.740965943225825e-05,
      "loss": 4.4873,
      "step": 158208
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7401273484747736e-05,
      "loss": 4.4901,
      "step": 158720
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7392887537237216e-05,
      "loss": 4.4648,
      "step": 159232
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7384501589726696e-05,
      "loss": 4.4752,
      "step": 159744
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7376115642216176e-05,
      "loss": 4.4756,
      "step": 160256
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7367729694705656e-05,
      "loss": 4.467,
      "step": 160768
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7359343747195135e-05,
      "loss": 4.4763,
      "step": 161280
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7350957799684615e-05,
      "loss": 4.47,
      "step": 161792
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7342571852174095e-05,
      "loss": 4.4703,
      "step": 162304
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7334202283467305e-05,
      "loss": 4.4981,
      "step": 162816
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7325816335956784e-05,
      "loss": 4.4722,
      "step": 163328
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7317430388446264e-05,
      "loss": 4.4733,
      "step": 163840
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.730904444093574e-05,
      "loss": 4.4692,
      "step": 164352
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7300674872228953e-05,
      "loss": 4.4853,
      "step": 164864
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7292288924718433e-05,
      "loss": 4.4646,
      "step": 165376
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.728390297720791e-05,
      "loss": 4.4623,
      "step": 165888
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.727551702969739e-05,
      "loss": 4.4663,
      "step": 166400
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.726714746099061e-05,
      "loss": 4.4662,
      "step": 166912
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.725876151348009e-05,
      "loss": 4.4581,
      "step": 167424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.725037556596957e-05,
      "loss": 4.4605,
      "step": 167936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.724200599726278e-05,
      "loss": 4.4758,
      "step": 168448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.723362004975226e-05,
      "loss": 4.4616,
      "step": 168960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.722523410224174e-05,
      "loss": 4.4582,
      "step": 169472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.721684815473121e-05,
      "loss": 4.4654,
      "step": 169984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.720846220722069e-05,
      "loss": 4.4608,
      "step": 170496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.720007625971017e-05,
      "loss": 4.457,
      "step": 171008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.719170669100339e-05,
      "loss": 4.4395,
      "step": 171520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.718332074349287e-05,
      "loss": 4.4556,
      "step": 172032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.717493479598235e-05,
      "loss": 4.4386,
      "step": 172544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.716654884847183e-05,
      "loss": 4.4521,
      "step": 173056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.715816290096131e-05,
      "loss": 4.4452,
      "step": 173568
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.714977695345079e-05,
      "loss": 4.4538,
      "step": 174080
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.714139100594027e-05,
      "loss": 4.4448,
      "step": 174592
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.713300505842975e-05,
      "loss": 4.4455,
      "step": 175104
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.712466824733042e-05,
      "loss": 4.4497,
      "step": 175616
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.71162822998199e-05,
      "loss": 4.4482,
      "step": 176128
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.710789635230938e-05,
      "loss": 4.4512,
      "step": 176640
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.709951040479886e-05,
      "loss": 4.4372,
      "step": 177152
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.709112445728834e-05,
      "loss": 4.4283,
      "step": 177664
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.708273850977782e-05,
      "loss": 4.4476,
      "step": 178176
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.70743525622673e-05,
      "loss": 4.4383,
      "step": 178688
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.706596661475678e-05,
      "loss": 4.4424,
      "step": 179200
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.705758066724626e-05,
      "loss": 4.4218,
      "step": 179712
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.704921109853947e-05,
      "loss": 4.4458,
      "step": 180224
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.704082515102895e-05,
      "loss": 4.4213,
      "step": 180736
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.703243920351843e-05,
      "loss": 4.4509,
      "step": 181248
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.702405325600791e-05,
      "loss": 4.4141,
      "step": 181760
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.701566730849739e-05,
      "loss": 4.4368,
      "step": 182272
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.700728136098687e-05,
      "loss": 4.4252,
      "step": 182784
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.699889541347635e-05,
      "loss": 4.4238,
      "step": 183296
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.699050946596583e-05,
      "loss": 4.4233,
      "step": 183808
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6982139897259045e-05,
      "loss": 4.4242,
      "step": 184320
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6973770328552254e-05,
      "loss": 4.4066,
      "step": 184832
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6965384381041734e-05,
      "loss": 4.4146,
      "step": 185344
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6956998433531214e-05,
      "loss": 4.426,
      "step": 185856
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6948612486020694e-05,
      "loss": 4.4234,
      "step": 186368
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6940226538510174e-05,
      "loss": 4.4131,
      "step": 186880
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6931840590999654e-05,
      "loss": 4.4139,
      "step": 187392
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6923454643489134e-05,
      "loss": 4.4102,
      "step": 187904
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6915068695978614e-05,
      "loss": 4.4169,
      "step": 188416
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.690671550607555e-05,
      "loss": 4.4198,
      "step": 188928
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.689832955856503e-05,
      "loss": 4.419,
      "step": 189440
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.688994361105451e-05,
      "loss": 4.4277,
      "step": 189952
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6881557663544e-05,
      "loss": 4.4275,
      "step": 190464
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.687317171603348e-05,
      "loss": 4.4221,
      "step": 190976
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.686478576852296e-05,
      "loss": 4.4032,
      "step": 191488
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.685641619981617e-05,
      "loss": 4.4131,
      "step": 192000
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.684803025230565e-05,
      "loss": 4.4168,
      "step": 192512
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.683966068359886e-05,
      "loss": 4.3985,
      "step": 193024
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.683127473608834e-05,
      "loss": 4.4122,
      "step": 193536
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.682288878857782e-05,
      "loss": 4.4083,
      "step": 194048
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.68145028410673e-05,
      "loss": 4.4191,
      "step": 194560
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.680611689355678e-05,
      "loss": 4.4124,
      "step": 195072
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.679773094604626e-05,
      "loss": 4.3936,
      "step": 195584
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678934499853574e-05,
      "loss": 4.4071,
      "step": 196096
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.678095905102522e-05,
      "loss": 4.4109,
      "step": 196608
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6772573103514697e-05,
      "loss": 4.4145,
      "step": 197120
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6764187156004177e-05,
      "loss": 4.4071,
      "step": 197632
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.6755801208493657e-05,
      "loss": 4.4117,
      "step": 198144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.674743163978687e-05,
      "loss": 4.4093,
      "step": 198656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6739045692276346e-05,
      "loss": 4.3943,
      "step": 199168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6730659744765826e-05,
      "loss": 4.4063,
      "step": 199680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6722273797255305e-05,
      "loss": 4.4017,
      "step": 200192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6713887849744785e-05,
      "loss": 4.3941,
      "step": 200704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6705501902234265e-05,
      "loss": 4.4133,
      "step": 201216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6697132333527474e-05,
      "loss": 4.3952,
      "step": 201728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6688746386016954e-05,
      "loss": 4.3861,
      "step": 202240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6680360438506434e-05,
      "loss": 4.395,
      "step": 202752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.667197449099592e-05,
      "loss": 4.4033,
      "step": 203264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.66635885434854e-05,
      "loss": 4.3996,
      "step": 203776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.665520259597488e-05,
      "loss": 4.3901,
      "step": 204288
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.664683302726809e-05,
      "loss": 4.3853,
      "step": 204800
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.66384634585613e-05,
      "loss": 4.3837,
      "step": 205312
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.663007751105078e-05,
      "loss": 4.395,
      "step": 205824
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.662169156354026e-05,
      "loss": 4.3729,
      "step": 206336
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.661330561602974e-05,
      "loss": 4.3846,
      "step": 206848
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.660491966851922e-05,
      "loss": 4.3892,
      "step": 207360
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.65965337210087e-05,
      "loss": 4.3859,
      "step": 207872
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.658814777349818e-05,
      "loss": 4.3841,
      "step": 208384
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657976182598766e-05,
      "loss": 4.3872,
      "step": 208896
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.657139225728087e-05,
      "loss": 4.3774,
      "step": 209408
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6563006309770355e-05,
      "loss": 4.383,
      "step": 209920
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6554620362259835e-05,
      "loss": 4.3828,
      "step": 210432
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6546234414749315e-05,
      "loss": 4.381,
      "step": 210944
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6537848467238795e-05,
      "loss": 4.3888,
      "step": 211456
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6529462519728275e-05,
      "loss": 4.3883,
      "step": 211968
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6521092951021484e-05,
      "loss": 4.3857,
      "step": 212480
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6512707003510964e-05,
      "loss": 4.3715,
      "step": 212992
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6504321056000444e-05,
      "loss": 4.3788,
      "step": 213504
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6495935108489924e-05,
      "loss": 4.377,
      "step": 214016
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6487549160979403e-05,
      "loss": 4.3668,
      "step": 214528
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6479163213468883e-05,
      "loss": 4.3815,
      "step": 215040
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.647077726595836e-05,
      "loss": 4.3682,
      "step": 215552
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6462391318447837e-05,
      "loss": 4.3817,
      "step": 216064
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.645402174974105e-05,
      "loss": 4.3651,
      "step": 216576
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.644563580223053e-05,
      "loss": 4.3752,
      "step": 217088
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.643724985472001e-05,
      "loss": 4.3565,
      "step": 217600
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.642888028601323e-05,
      "loss": 4.3757,
      "step": 218112
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.642049433850271e-05,
      "loss": 4.3608,
      "step": 218624
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.641210839099218e-05,
      "loss": 4.3817,
      "step": 219136
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.640372244348166e-05,
      "loss": 4.3793,
      "step": 219648
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.639533649597114e-05,
      "loss": 4.3692,
      "step": 220160
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.638695054846062e-05,
      "loss": 4.3615,
      "step": 220672
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.637858097975383e-05,
      "loss": 4.3657,
      "step": 221184
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.637019503224331e-05,
      "loss": 4.3773,
      "step": 221696
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.636180908473279e-05,
      "loss": 4.3835,
      "step": 222208
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.635342313722228e-05,
      "loss": 4.3683,
      "step": 222720
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.634503718971176e-05,
      "loss": 4.3697,
      "step": 223232
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6336667621004966e-05,
      "loss": 4.3675,
      "step": 223744
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6328281673494446e-05,
      "loss": 4.3656,
      "step": 224256
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6319895725983926e-05,
      "loss": 4.3574,
      "step": 224768
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6311509778473406e-05,
      "loss": 4.3613,
      "step": 225280
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6303123830962886e-05,
      "loss": 4.3629,
      "step": 225792
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6294737883452366e-05,
      "loss": 4.3542,
      "step": 226304
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6286368314745575e-05,
      "loss": 4.3553,
      "step": 226816
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6277982367235055e-05,
      "loss": 4.3608,
      "step": 227328
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6269596419724535e-05,
      "loss": 4.3543,
      "step": 227840
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.6261210472214015e-05,
      "loss": 4.3613,
      "step": 228352
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.625284090350723e-05,
      "loss": 4.3554,
      "step": 228864
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.332054615020752,
      "eval_runtime": 293.744,
      "eval_samples_per_second": 1299.06,
      "eval_steps_per_second": 40.597,
      "step": 228960
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.624445495599671e-05,
      "loss": 4.3516,
      "step": 229376
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.623606900848619e-05,
      "loss": 4.3474,
      "step": 229888
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.622768306097567e-05,
      "loss": 4.3614,
      "step": 230400
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.621929711346515e-05,
      "loss": 4.3507,
      "step": 230912
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.621091116595463e-05,
      "loss": 4.3623,
      "step": 231424
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.620252521844411e-05,
      "loss": 4.3395,
      "step": 231936
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.619415564973732e-05,
      "loss": 4.3517,
      "step": 232448
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.61857697022268e-05,
      "loss": 4.3396,
      "step": 232960
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.617738375471628e-05,
      "loss": 4.3559,
      "step": 233472
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.616899780720576e-05,
      "loss": 4.3505,
      "step": 233984
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.616061185969524e-05,
      "loss": 4.3525,
      "step": 234496
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.615222591218472e-05,
      "loss": 4.3555,
      "step": 235008
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.61438399646742e-05,
      "loss": 4.3321,
      "step": 235520
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.613545401716368e-05,
      "loss": 4.3415,
      "step": 236032
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.612706806965316e-05,
      "loss": 4.3408,
      "step": 236544
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.611868212214264e-05,
      "loss": 4.3385,
      "step": 237056
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.611029617463212e-05,
      "loss": 4.34,
      "step": 237568
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.61019102271216e-05,
      "loss": 4.344,
      "step": 238080
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.609354065841481e-05,
      "loss": 4.3352,
      "step": 238592
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.608515471090429e-05,
      "loss": 4.3694,
      "step": 239104
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.607676876339377e-05,
      "loss": 4.3391,
      "step": 239616
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.606838281588325e-05,
      "loss": 4.3419,
      "step": 240128
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.606001324717646e-05,
      "loss": 4.341,
      "step": 240640
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.605162729966594e-05,
      "loss": 4.3519,
      "step": 241152
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.604324135215542e-05,
      "loss": 4.3335,
      "step": 241664
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.603487178344863e-05,
      "loss": 4.3399,
      "step": 242176
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.602648583593811e-05,
      "loss": 4.3341,
      "step": 242688
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.601809988842759e-05,
      "loss": 4.3372,
      "step": 243200
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.600971394091707e-05,
      "loss": 4.3267,
      "step": 243712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.600132799340655e-05,
      "loss": 4.3366,
      "step": 244224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.599294204589603e-05,
      "loss": 4.3461,
      "step": 244736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.598455609838551e-05,
      "loss": 4.3384,
      "step": 245248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.597617015087499e-05,
      "loss": 4.3365,
      "step": 245760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.59678005821682e-05,
      "loss": 4.3379,
      "step": 246272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.595943101346141e-05,
      "loss": 4.3328,
      "step": 246784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.595104506595089e-05,
      "loss": 4.3369,
      "step": 247296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.594265911844037e-05,
      "loss": 4.3157,
      "step": 247808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.593427317092985e-05,
      "loss": 4.3322,
      "step": 248320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.592588722341933e-05,
      "loss": 4.3127,
      "step": 248832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5917517654712546e-05,
      "loss": 4.3311,
      "step": 249344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5909131707202026e-05,
      "loss": 4.3241,
      "step": 249856
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5900745759691506e-05,
      "loss": 4.3317,
      "step": 250368
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5892359812180986e-05,
      "loss": 4.3218,
      "step": 250880
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5883973864670466e-05,
      "loss": 4.3224,
      "step": 251392
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5875604295963675e-05,
      "loss": 4.3307,
      "step": 251904
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5867218348453155e-05,
      "loss": 4.3271,
      "step": 252416
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5858832400942635e-05,
      "loss": 4.3273,
      "step": 252928
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5850446453432115e-05,
      "loss": 4.3218,
      "step": 253440
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5842060505921595e-05,
      "loss": 4.3066,
      "step": 253952
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5833690937214804e-05,
      "loss": 4.3287,
      "step": 254464
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5825304989704284e-05,
      "loss": 4.3196,
      "step": 254976
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.581691904219377e-05,
      "loss": 4.3233,
      "step": 255488
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.580853309468325e-05,
      "loss": 4.308,
      "step": 256000
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.580014714717273e-05,
      "loss": 4.3311,
      "step": 256512
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5791761199662204e-05,
      "loss": 4.2995,
      "step": 257024
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5783375252151684e-05,
      "loss": 4.3349,
      "step": 257536
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5774989304641164e-05,
      "loss": 4.2988,
      "step": 258048
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.576661973593438e-05,
      "loss": 4.3159,
      "step": 258560
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.575823378842385e-05,
      "loss": 4.3113,
      "step": 259072
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.574984784091333e-05,
      "loss": 4.3091,
      "step": 259584
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.574147827220655e-05,
      "loss": 4.3043,
      "step": 260096
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.573309232469602e-05,
      "loss": 4.3162,
      "step": 260608
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.572470637718551e-05,
      "loss": 4.2906,
      "step": 261120
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.571632042967499e-05,
      "loss": 4.2985,
      "step": 261632
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.570793448216447e-05,
      "loss": 4.314,
      "step": 262144
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.569954853465395e-05,
      "loss": 4.309,
      "step": 262656
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.569116258714343e-05,
      "loss": 4.3022,
      "step": 263168
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.568277663963291e-05,
      "loss": 4.302,
      "step": 263680
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.567440707092612e-05,
      "loss": 4.2917,
      "step": 264192
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.566603750221933e-05,
      "loss": 4.3053,
      "step": 264704
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.565765155470881e-05,
      "loss": 4.3103,
      "step": 265216
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.564926560719829e-05,
      "loss": 4.3085,
      "step": 265728
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5640879659687767e-05,
      "loss": 4.3184,
      "step": 266240
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.5632493712177247e-05,
      "loss": 4.3155,
      "step": 266752
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.562412414347046e-05,
      "loss": 4.3132,
      "step": 267264
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.561573819595994e-05,
      "loss": 4.2934,
      "step": 267776
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.560735224844942e-05,
      "loss": 4.3012,
      "step": 268288
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.55989663009389e-05,
      "loss": 4.308,
      "step": 268800
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.559059673223211e-05,
      "loss": 4.2893,
      "step": 269312
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.558221078472159e-05,
      "loss": 4.3059,
      "step": 269824
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.557382483721107e-05,
      "loss": 4.2988,
      "step": 270336
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.556543888970055e-05,
      "loss": 4.3078,
      "step": 270848
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.555705294219003e-05,
      "loss": 4.305,
      "step": 271360
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.554866699467951e-05,
      "loss": 4.2876,
      "step": 271872
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.554028104716899e-05,
      "loss": 4.3023,
      "step": 272384
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.553189509965847e-05,
      "loss": 4.3007,
      "step": 272896
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.552352553095168e-05,
      "loss": 4.31,
      "step": 273408
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.551513958344116e-05,
      "loss": 4.303,
      "step": 273920
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.550675363593064e-05,
      "loss": 4.3031,
      "step": 274432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.549836768842013e-05,
      "loss": 4.3067,
      "step": 274944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5489998119713336e-05,
      "loss": 4.2881,
      "step": 275456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5481612172202816e-05,
      "loss": 4.3027,
      "step": 275968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5473226224692296e-05,
      "loss": 4.2943,
      "step": 276480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5464840277181776e-05,
      "loss": 4.2949,
      "step": 276992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5456470708474985e-05,
      "loss": 4.3076,
      "step": 277504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5448084760964465e-05,
      "loss": 4.2953,
      "step": 278016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5439698813453945e-05,
      "loss": 4.2811,
      "step": 278528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5431312865943425e-05,
      "loss": 4.2919,
      "step": 279040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5422943297236634e-05,
      "loss": 4.3013,
      "step": 279552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5414557349726114e-05,
      "loss": 4.2982,
      "step": 280064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5406171402215594e-05,
      "loss": 4.2886,
      "step": 280576
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.539778545470508e-05,
      "loss": 4.2836,
      "step": 281088
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.538941588599829e-05,
      "loss": 4.2838,
      "step": 281600
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.538102993848777e-05,
      "loss": 4.2967,
      "step": 282112
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.537264399097725e-05,
      "loss": 4.2692,
      "step": 282624
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.536425804346673e-05,
      "loss": 4.2847,
      "step": 283136
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.535588847475994e-05,
      "loss": 4.2884,
      "step": 283648
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.534750252724942e-05,
      "loss": 4.2878,
      "step": 284160
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.53391165797389e-05,
      "loss": 4.2807,
      "step": 284672
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.533073063222838e-05,
      "loss": 4.2902,
      "step": 285184
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.532237744232532e-05,
      "loss": 4.2808,
      "step": 285696
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.53139914948148e-05,
      "loss": 4.2809,
      "step": 286208
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.530560554730428e-05,
      "loss": 4.2865,
      "step": 286720
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.529721959979376e-05,
      "loss": 4.2815,
      "step": 287232
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.528883365228324e-05,
      "loss": 4.2957,
      "step": 287744
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.528044770477272e-05,
      "loss": 4.2868,
      "step": 288256
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.527207813606593e-05,
      "loss": 4.2882,
      "step": 288768
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.526369218855541e-05,
      "loss": 4.273,
      "step": 289280
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.525530624104489e-05,
      "loss": 4.2827,
      "step": 289792
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.524692029353437e-05,
      "loss": 4.2818,
      "step": 290304
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.523853434602385e-05,
      "loss": 4.2704,
      "step": 290816
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.523014839851333e-05,
      "loss": 4.2836,
      "step": 291328
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5221762451002805e-05,
      "loss": 4.2738,
      "step": 291840
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5213376503492285e-05,
      "loss": 4.2853,
      "step": 292352
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.520499055598177e-05,
      "loss": 4.2703,
      "step": 292864
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.519660460847125e-05,
      "loss": 4.2802,
      "step": 293376
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.518821866096073e-05,
      "loss": 4.256,
      "step": 293888
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.517983271345021e-05,
      "loss": 4.2838,
      "step": 294400
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.517146314474342e-05,
      "loss": 4.2687,
      "step": 294912
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.51630771972329e-05,
      "loss": 4.2836,
      "step": 295424
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.515470762852611e-05,
      "loss": 4.282,
      "step": 295936
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5146338059819326e-05,
      "loss": 4.281,
      "step": 296448
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5137952112308806e-05,
      "loss": 4.2646,
      "step": 296960
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.512956616479828e-05,
      "loss": 4.2762,
      "step": 297472
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.512118021728776e-05,
      "loss": 4.2815,
      "step": 297984
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.511279426977724e-05,
      "loss": 4.2868,
      "step": 298496
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5104408322266726e-05,
      "loss": 4.2777,
      "step": 299008
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5096038753559935e-05,
      "loss": 4.2759,
      "step": 299520
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5087652806049415e-05,
      "loss": 4.2756,
      "step": 300032
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5079266858538895e-05,
      "loss": 4.2789,
      "step": 300544
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5070880911028375e-05,
      "loss": 4.2614,
      "step": 301056
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5062494963517855e-05,
      "loss": 4.2702,
      "step": 301568
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5054109016007334e-05,
      "loss": 4.2695,
      "step": 302080
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5045723068496814e-05,
      "loss": 4.2649,
      "step": 302592
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5037337120986294e-05,
      "loss": 4.2626,
      "step": 303104
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5028951173475774e-05,
      "loss": 4.2722,
      "step": 303616
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.5020581604768983e-05,
      "loss": 4.2609,
      "step": 304128
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.501219565725846e-05,
      "loss": 4.269,
      "step": 304640
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.500380970974794e-05,
      "loss": 4.2729,
      "step": 305152
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.2502336502075195,
      "eval_runtime": 294.515,
      "eval_samples_per_second": 1295.659,
      "eval_steps_per_second": 40.49,
      "step": 305280
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.499542376223742e-05,
      "loss": 4.2551,
      "step": 305664
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.498703781472691e-05,
      "loss": 4.2585,
      "step": 306176
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.497865186721639e-05,
      "loss": 4.27,
      "step": 306688
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.497026591970587e-05,
      "loss": 4.2617,
      "step": 307200
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.496187997219535e-05,
      "loss": 4.2749,
      "step": 307712
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.495349402468482e-05,
      "loss": 4.2494,
      "step": 308224
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.49451080771743e-05,
      "loss": 4.2602,
      "step": 308736
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.493672212966378e-05,
      "loss": 4.2522,
      "step": 309248
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.492833618215326e-05,
      "loss": 4.2662,
      "step": 309760
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.491995023464274e-05,
      "loss": 4.2655,
      "step": 310272
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.491156428713222e-05,
      "loss": 4.2568,
      "step": 310784
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.49031783396217e-05,
      "loss": 4.2727,
      "step": 311296
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.489479239211118e-05,
      "loss": 4.241,
      "step": 311808
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.488640644460066e-05,
      "loss": 4.2544,
      "step": 312320
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.487803687589387e-05,
      "loss": 4.2584,
      "step": 312832
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.486965092838336e-05,
      "loss": 4.248,
      "step": 313344
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.486126498087284e-05,
      "loss": 4.2534,
      "step": 313856
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.485287903336232e-05,
      "loss": 4.2564,
      "step": 314368
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.48444930858518e-05,
      "loss": 4.2515,
      "step": 314880
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.483610713834128e-05,
      "loss": 4.2806,
      "step": 315392
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.482772119083076e-05,
      "loss": 4.255,
      "step": 315904
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.481933524332024e-05,
      "loss": 4.2536,
      "step": 316416
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.481094929580971e-05,
      "loss": 4.2611,
      "step": 316928
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.480257972710293e-05,
      "loss": 4.268,
      "step": 317440
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.479419377959241e-05,
      "loss": 4.2483,
      "step": 317952
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.478580783208188e-05,
      "loss": 4.2584,
      "step": 318464
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.477742188457136e-05,
      "loss": 4.2497,
      "step": 318976
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.476903593706084e-05,
      "loss": 4.2527,
      "step": 319488
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.476064998955033e-05,
      "loss": 4.2447,
      "step": 320000
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.475226404203981e-05,
      "loss": 4.2532,
      "step": 320512
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.474387809452929e-05,
      "loss": 4.2643,
      "step": 321024
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4735508525822496e-05,
      "loss": 4.2538,
      "step": 321536
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4727122578311976e-05,
      "loss": 4.2546,
      "step": 322048
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4718736630801456e-05,
      "loss": 4.2565,
      "step": 322560
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4710367062094665e-05,
      "loss": 4.251,
      "step": 323072
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4701981114584145e-05,
      "loss": 4.2559,
      "step": 323584
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4693595167073625e-05,
      "loss": 4.2345,
      "step": 324096
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4685209219563105e-05,
      "loss": 4.2476,
      "step": 324608
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4676823272052585e-05,
      "loss": 4.2368,
      "step": 325120
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4668437324542065e-05,
      "loss": 4.2433,
      "step": 325632
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.466006775583528e-05,
      "loss": 4.2454,
      "step": 326144
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.465168180832476e-05,
      "loss": 4.2521,
      "step": 326656
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.464329586081424e-05,
      "loss": 4.2408,
      "step": 327168
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.463490991330372e-05,
      "loss": 4.2437,
      "step": 327680
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.46265239657932e-05,
      "loss": 4.248,
      "step": 328192
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.461813801828268e-05,
      "loss": 4.2463,
      "step": 328704
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.460976844957589e-05,
      "loss": 4.247,
      "step": 329216
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.460138250206537e-05,
      "loss": 4.2424,
      "step": 329728
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.459299655455485e-05,
      "loss": 4.2262,
      "step": 330240
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.458461060704433e-05,
      "loss": 4.2501,
      "step": 330752
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.457624103833754e-05,
      "loss": 4.2367,
      "step": 331264
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.456785509082702e-05,
      "loss": 4.2475,
      "step": 331776
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.45594691433165e-05,
      "loss": 4.2329,
      "step": 332288
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.455108319580598e-05,
      "loss": 4.2469,
      "step": 332800
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4542697248295465e-05,
      "loss": 4.2234,
      "step": 333312
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4534311300784945e-05,
      "loss": 4.254,
      "step": 333824
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4525925353274425e-05,
      "loss": 4.221,
      "step": 334336
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.45175394057639e-05,
      "loss": 4.2317,
      "step": 334848
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.450918621586084e-05,
      "loss": 4.2383,
      "step": 335360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.450080026835032e-05,
      "loss": 4.2299,
      "step": 335872
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.44924143208398e-05,
      "loss": 4.2315,
      "step": 336384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.448402837332928e-05,
      "loss": 4.2351,
      "step": 336896
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.447564242581876e-05,
      "loss": 4.2225,
      "step": 337408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.446725647830824e-05,
      "loss": 4.2193,
      "step": 337920
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.445888690960145e-05,
      "loss": 4.2332,
      "step": 338432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.445050096209093e-05,
      "loss": 4.2354,
      "step": 338944
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.444211501458041e-05,
      "loss": 4.2235,
      "step": 339456
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.44337290670699e-05,
      "loss": 4.2319,
      "step": 339968
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.442535949836311e-05,
      "loss": 4.2138,
      "step": 340480
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.441697355085259e-05,
      "loss": 4.23,
      "step": 340992
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.440858760334207e-05,
      "loss": 4.2337,
      "step": 341504
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.440020165583154e-05,
      "loss": 4.2309,
      "step": 342016
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.439181570832102e-05,
      "loss": 4.2407,
      "step": 342528
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.43834297608105e-05,
      "loss": 4.2382,
      "step": 343040
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.437504381329998e-05,
      "loss": 4.2439,
      "step": 343552
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4366690623396926e-05,
      "loss": 4.2186,
      "step": 344064
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4358304675886406e-05,
      "loss": 4.2269,
      "step": 344576
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4349918728375886e-05,
      "loss": 4.2338,
      "step": 345088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4341532780865366e-05,
      "loss": 4.2121,
      "step": 345600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4333146833354846e-05,
      "loss": 4.2347,
      "step": 346112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4324760885844326e-05,
      "loss": 4.2265,
      "step": 346624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.431639131713754e-05,
      "loss": 4.2331,
      "step": 347136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4308005369627015e-05,
      "loss": 4.2329,
      "step": 347648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4299619422116495e-05,
      "loss": 4.2141,
      "step": 348160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4291233474605975e-05,
      "loss": 4.2292,
      "step": 348672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4282847527095455e-05,
      "loss": 4.2255,
      "step": 349184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4274461579584935e-05,
      "loss": 4.2336,
      "step": 349696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4266075632074415e-05,
      "loss": 4.2281,
      "step": 350208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.4257689684563894e-05,
      "loss": 4.2346,
      "step": 350720
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.4249320115857104e-05,
      "loss": 4.2304,
      "step": 351232
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.424095054715032e-05,
      "loss": 4.2159,
      "step": 351744
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.42325645996398e-05,
      "loss": 4.2286,
      "step": 352256
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.422417865212928e-05,
      "loss": 4.2192,
      "step": 352768
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.421579270461876e-05,
      "loss": 4.2269,
      "step": 353280
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.420740675710824e-05,
      "loss": 4.2348,
      "step": 353792
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.419902080959772e-05,
      "loss": 4.2185,
      "step": 354304
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.41906348620872e-05,
      "loss": 4.214,
      "step": 354816
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.418224891457668e-05,
      "loss": 4.2183,
      "step": 355328
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.417386296706616e-05,
      "loss": 4.2276,
      "step": 355840
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.416549339835937e-05,
      "loss": 4.2316,
      "step": 356352
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.415710745084885e-05,
      "loss": 4.2156,
      "step": 356864
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.414872150333833e-05,
      "loss": 4.215,
      "step": 357376
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.414033555582781e-05,
      "loss": 4.2112,
      "step": 357888
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.4131965987121024e-05,
      "loss": 4.2263,
      "step": 358400
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.4123580039610504e-05,
      "loss": 4.2002,
      "step": 358912
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.4115194092099984e-05,
      "loss": 4.2138,
      "step": 359424
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.4106808144589464e-05,
      "loss": 4.2188,
      "step": 359936
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.4098422197078944e-05,
      "loss": 4.2143,
      "step": 360448
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.4090036249568424e-05,
      "loss": 4.2121,
      "step": 360960
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.4081650302057904e-05,
      "loss": 4.2233,
      "step": 361472
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.407326435454738e-05,
      "loss": 4.2112,
      "step": 361984
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.406487840703686e-05,
      "loss": 4.2076,
      "step": 362496
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.405650883833007e-05,
      "loss": 4.2194,
      "step": 363008
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.404812289081955e-05,
      "loss": 4.2152,
      "step": 363520
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.4039736943309026e-05,
      "loss": 4.2248,
      "step": 364032
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.403135099579851e-05,
      "loss": 4.2198,
      "step": 364544
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.402298142709173e-05,
      "loss": 4.2184,
      "step": 365056
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.40145954795812e-05,
      "loss": 4.2034,
      "step": 365568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.400620953207068e-05,
      "loss": 4.2134,
      "step": 366080
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.399782358456016e-05,
      "loss": 4.213,
      "step": 366592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.398943763704964e-05,
      "loss": 4.2003,
      "step": 367104
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.398105168953912e-05,
      "loss": 4.2201,
      "step": 367616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.397268212083233e-05,
      "loss": 4.2026,
      "step": 368128
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.396429617332181e-05,
      "loss": 4.2169,
      "step": 368640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.395591022581129e-05,
      "loss": 4.2065,
      "step": 369152
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.394752427830077e-05,
      "loss": 4.2118,
      "step": 369664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.393915470959398e-05,
      "loss": 4.191,
      "step": 370176
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3930768762083466e-05,
      "loss": 4.2107,
      "step": 370688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3922382814572946e-05,
      "loss": 4.2067,
      "step": 371200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3913996867062426e-05,
      "loss": 4.2118,
      "step": 371712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3905610919551906e-05,
      "loss": 4.2141,
      "step": 372224
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3897224972041386e-05,
      "loss": 4.2192,
      "step": 372736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3888839024530866e-05,
      "loss": 4.1961,
      "step": 373248
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3880469455824075e-05,
      "loss": 4.2117,
      "step": 373760
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3872083508313555e-05,
      "loss": 4.2176,
      "step": 374272
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3863697560803035e-05,
      "loss": 4.2172,
      "step": 374784
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3855311613292515e-05,
      "loss": 4.2123,
      "step": 375296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3846925665781995e-05,
      "loss": 4.214,
      "step": 375808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3838539718271475e-05,
      "loss": 4.2082,
      "step": 376320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3830153770760955e-05,
      "loss": 4.2125,
      "step": 376832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3821767823250435e-05,
      "loss": 4.1936,
      "step": 377344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3813398254543644e-05,
      "loss": 4.2071,
      "step": 377856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.380501230703313e-05,
      "loss": 4.2013,
      "step": 378368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.379662635952261e-05,
      "loss": 4.2031,
      "step": 378880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.378825679081582e-05,
      "loss": 4.2013,
      "step": 379392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.37798708433053e-05,
      "loss": 4.2026,
      "step": 379904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.377148489579478e-05,
      "loss": 4.2021,
      "step": 380416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.376309894828426e-05,
      "loss": 4.2003,
      "step": 380928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.375471300077374e-05,
      "loss": 4.2103,
      "step": 381440
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.194133281707764,
      "eval_runtime": 312.5876,
      "eval_samples_per_second": 1220.749,
      "eval_steps_per_second": 38.149,
      "step": 381600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.374634343206695e-05,
      "loss": 4.1983,
      "step": 381952
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.373795748455643e-05,
      "loss": 4.1925,
      "step": 382464
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.372957153704591e-05,
      "loss": 4.2049,
      "step": 382976
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.372120196833912e-05,
      "loss": 4.2017,
      "step": 383488
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.37128160208286e-05,
      "loss": 4.2041,
      "step": 384000
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.3704430073318084e-05,
      "loss": 4.1937,
      "step": 384512
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.3696044125807564e-05,
      "loss": 4.1923,
      "step": 385024
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.368765817829704e-05,
      "loss": 4.193,
      "step": 385536
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.367928860959025e-05,
      "loss": 4.1993,
      "step": 386048
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.367090266207973e-05,
      "loss": 4.2076,
      "step": 386560
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.366251671456921e-05,
      "loss": 4.1943,
      "step": 387072
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.3654130767058686e-05,
      "loss": 4.2074,
      "step": 387584
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.3645744819548166e-05,
      "loss": 4.1854,
      "step": 388096
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.363737525084138e-05,
      "loss": 4.1914,
      "step": 388608
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.362898930333086e-05,
      "loss": 4.1985,
      "step": 389120
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.3620603355820335e-05,
      "loss": 4.1824,
      "step": 389632
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.361221740830982e-05,
      "loss": 4.1919,
      "step": 390144
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.36038314607993e-05,
      "loss": 4.1953,
      "step": 390656
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.359544551328878e-05,
      "loss": 4.1868,
      "step": 391168
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.358705956577826e-05,
      "loss": 4.2157,
      "step": 391680
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.357867361826774e-05,
      "loss": 4.1984,
      "step": 392192
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.357030404956095e-05,
      "loss": 4.1943,
      "step": 392704
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.356191810205043e-05,
      "loss": 4.1968,
      "step": 393216
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.355354853334364e-05,
      "loss": 4.2077,
      "step": 393728
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.354516258583312e-05,
      "loss": 4.1861,
      "step": 394240
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.35367766383226e-05,
      "loss": 4.1953,
      "step": 394752
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.352839069081208e-05,
      "loss": 4.193,
      "step": 395264
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.352002112210529e-05,
      "loss": 4.1881,
      "step": 395776
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.3511635174594776e-05,
      "loss": 4.181,
      "step": 396288
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.3503249227084256e-05,
      "loss": 4.196,
      "step": 396800
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3494863279573736e-05,
      "loss": 4.204,
      "step": 397312
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3486477332063216e-05,
      "loss": 4.1944,
      "step": 397824
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3478091384552696e-05,
      "loss": 4.198,
      "step": 398336
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3469705437042175e-05,
      "loss": 4.1927,
      "step": 398848
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3461335868335385e-05,
      "loss": 4.1912,
      "step": 399360
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3452949920824865e-05,
      "loss": 4.1943,
      "step": 399872
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3444580352118074e-05,
      "loss": 4.1811,
      "step": 400384
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3436194404607554e-05,
      "loss": 4.1834,
      "step": 400896
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3427808457097034e-05,
      "loss": 4.1783,
      "step": 401408
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3419422509586514e-05,
      "loss": 4.182,
      "step": 401920
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3411036562075993e-05,
      "loss": 4.1885,
      "step": 402432
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3402650614565473e-05,
      "loss": 4.1908,
      "step": 402944
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.339426466705496e-05,
      "loss": 4.1819,
      "step": 403456
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.338587871954444e-05,
      "loss": 4.1862,
      "step": 403968
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.337749277203392e-05,
      "loss": 4.1914,
      "step": 404480
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.33691068245234e-05,
      "loss": 4.1873,
      "step": 404992
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.336072087701287e-05,
      "loss": 4.1897,
      "step": 405504
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.335233492950235e-05,
      "loss": 4.1842,
      "step": 406016
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.334396536079557e-05,
      "loss": 4.169,
      "step": 406528
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.333559579208878e-05,
      "loss": 4.1936,
      "step": 407040
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.332720984457826e-05,
      "loss": 4.1784,
      "step": 407552
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.331882389706774e-05,
      "loss": 4.1888,
      "step": 408064
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.331043794955722e-05,
      "loss": 4.1751,
      "step": 408576
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.330206838085043e-05,
      "loss": 4.1867,
      "step": 409088
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3293682433339914e-05,
      "loss": 4.1663,
      "step": 409600
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3285296485829394e-05,
      "loss": 4.1907,
      "step": 410112
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3276910538318874e-05,
      "loss": 4.1652,
      "step": 410624
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.326852459080835e-05,
      "loss": 4.1774,
      "step": 411136
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.326013864329783e-05,
      "loss": 4.1834,
      "step": 411648
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.325175269578731e-05,
      "loss": 4.1716,
      "step": 412160
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.324336674827679e-05,
      "loss": 4.1712,
      "step": 412672
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.323498080076627e-05,
      "loss": 4.1831,
      "step": 413184
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3226611232059476e-05,
      "loss": 4.1617,
      "step": 413696
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3218225284548956e-05,
      "loss": 4.1614,
      "step": 414208
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3209839337038436e-05,
      "loss": 4.1763,
      "step": 414720
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3201453389527916e-05,
      "loss": 4.1803,
      "step": 415232
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3193067442017396e-05,
      "loss": 4.1648,
      "step": 415744
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.318469787331061e-05,
      "loss": 4.1787,
      "step": 416256
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.317631192580009e-05,
      "loss": 4.1543,
      "step": 416768
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.316792597828957e-05,
      "loss": 4.1708,
      "step": 417280
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.315954003077905e-05,
      "loss": 4.1787,
      "step": 417792
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.315115408326853e-05,
      "loss": 4.1773,
      "step": 418304
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.314276813575801e-05,
      "loss": 4.1805,
      "step": 418816
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.313438218824749e-05,
      "loss": 4.1828,
      "step": 419328
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.312599624073697e-05,
      "loss": 4.191,
      "step": 419840
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.311764305083391e-05,
      "loss": 4.1606,
      "step": 420352
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.310925710332339e-05,
      "loss": 4.1763,
      "step": 420864
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.310087115581287e-05,
      "loss": 4.1795,
      "step": 421376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.309248520830235e-05,
      "loss": 4.1537,
      "step": 421888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3084115639595565e-05,
      "loss": 4.178,
      "step": 422400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3075729692085045e-05,
      "loss": 4.1722,
      "step": 422912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3067343744574525e-05,
      "loss": 4.1771,
      "step": 423424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3058957797064005e-05,
      "loss": 4.1781,
      "step": 423936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3050571849553485e-05,
      "loss": 4.1627,
      "step": 424448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3042202280846694e-05,
      "loss": 4.1696,
      "step": 424960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3033816333336174e-05,
      "loss": 4.1715,
      "step": 425472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3025430385825654e-05,
      "loss": 4.1815,
      "step": 425984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3017044438315134e-05,
      "loss": 4.1695,
      "step": 426496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.3008658490804614e-05,
      "loss": 4.1805,
      "step": 427008
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.3000272543294094e-05,
      "loss": 4.1805,
      "step": 427520
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2991886595783574e-05,
      "loss": 4.1647,
      "step": 428032
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2983500648273054e-05,
      "loss": 4.1728,
      "step": 428544
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.297513107956627e-05,
      "loss": 4.1632,
      "step": 429056
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.296674513205575e-05,
      "loss": 4.1746,
      "step": 429568
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.295835918454523e-05,
      "loss": 4.184,
      "step": 430080
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.294997323703471e-05,
      "loss": 4.1608,
      "step": 430592
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.294162004713165e-05,
      "loss": 4.1649,
      "step": 431104
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.293323409962113e-05,
      "loss": 4.1597,
      "step": 431616
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.292484815211061e-05,
      "loss": 4.173,
      "step": 432128
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.291646220460009e-05,
      "loss": 4.1803,
      "step": 432640
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.290807625708957e-05,
      "loss": 4.1646,
      "step": 433152
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.289969030957905e-05,
      "loss": 4.16,
      "step": 433664
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.289130436206853e-05,
      "loss": 4.1532,
      "step": 434176
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.288291841455801e-05,
      "loss": 4.1759,
      "step": 434688
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.287454884585122e-05,
      "loss": 4.1461,
      "step": 435200
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.28661628983407e-05,
      "loss": 4.1645,
      "step": 435712
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.285779332963391e-05,
      "loss": 4.1622,
      "step": 436224
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.284940738212339e-05,
      "loss": 4.1654,
      "step": 436736
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.28410378134166e-05,
      "loss": 4.161,
      "step": 437248
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.283265186590608e-05,
      "loss": 4.1673,
      "step": 437760
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.282426591839556e-05,
      "loss": 4.162,
      "step": 438272
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.281587997088504e-05,
      "loss": 4.1559,
      "step": 438784
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.280749402337452e-05,
      "loss": 4.1657,
      "step": 439296
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2799108075864e-05,
      "loss": 4.165,
      "step": 439808
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2790722128353474e-05,
      "loss": 4.1699,
      "step": 440320
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.278233618084296e-05,
      "loss": 4.1716,
      "step": 440832
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.277396661213618e-05,
      "loss": 4.163,
      "step": 441344
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.276558066462566e-05,
      "loss": 4.1526,
      "step": 441856
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.275719471711513e-05,
      "loss": 4.1609,
      "step": 442368
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.274880876960461e-05,
      "loss": 4.1652,
      "step": 442880
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.274042282209409e-05,
      "loss": 4.1518,
      "step": 443392
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.273203687458357e-05,
      "loss": 4.1652,
      "step": 443904
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.272365092707305e-05,
      "loss": 4.1507,
      "step": 444416
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.271526497956253e-05,
      "loss": 4.1654,
      "step": 444928
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.270689541085574e-05,
      "loss": 4.1558,
      "step": 445440
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.269850946334522e-05,
      "loss": 4.1589,
      "step": 445952
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.26901235158347e-05,
      "loss": 4.1463,
      "step": 446464
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.268173756832418e-05,
      "loss": 4.157,
      "step": 446976
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2673367999617395e-05,
      "loss": 4.1589,
      "step": 447488
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2664982052106875e-05,
      "loss": 4.1638,
      "step": 448000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2656596104596355e-05,
      "loss": 4.159,
      "step": 448512
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2648210157085835e-05,
      "loss": 4.1662,
      "step": 449024
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2639824209575315e-05,
      "loss": 4.1483,
      "step": 449536
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2631454640868524e-05,
      "loss": 4.16,
      "step": 450048
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2623068693358004e-05,
      "loss": 4.1675,
      "step": 450560
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2614682745847484e-05,
      "loss": 4.1654,
      "step": 451072
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2606296798336964e-05,
      "loss": 4.1632,
      "step": 451584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2597910850826443e-05,
      "loss": 4.1674,
      "step": 452096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2589524903315923e-05,
      "loss": 4.1557,
      "step": 452608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.2581138955805403e-05,
      "loss": 4.1648,
      "step": 453120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.257275300829488e-05,
      "loss": 4.1454,
      "step": 453632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.256436706078436e-05,
      "loss": 4.1567,
      "step": 454144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.255599749207758e-05,
      "loss": 4.1511,
      "step": 454656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.254762792337079e-05,
      "loss": 4.153,
      "step": 455168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.253924197586027e-05,
      "loss": 4.1541,
      "step": 455680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.253085602834975e-05,
      "loss": 4.1486,
      "step": 456192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.252247008083923e-05,
      "loss": 4.1597,
      "step": 456704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.251408413332871e-05,
      "loss": 4.1447,
      "step": 457216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.250569818581819e-05,
      "loss": 4.1659,
      "step": 457728
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.1535773277282715,
      "eval_runtime": 302.8427,
      "eval_samples_per_second": 1260.03,
      "eval_steps_per_second": 39.377,
      "step": 457920
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.249731223830766e-05,
      "loss": 4.1472,
      "step": 458240
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.248892629079714e-05,
      "loss": 4.1437,
      "step": 458752
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.248054034328662e-05,
      "loss": 4.1562,
      "step": 459264
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.24721543957761e-05,
      "loss": 4.1526,
      "step": 459776
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.246376844826558e-05,
      "loss": 4.1555,
      "step": 460288
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.245538250075506e-05,
      "loss": 4.1467,
      "step": 460800
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.244699655324455e-05,
      "loss": 4.1367,
      "step": 461312
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.243861060573403e-05,
      "loss": 4.1481,
      "step": 461824
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.243022465822351e-05,
      "loss": 4.1502,
      "step": 462336
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.242185508951672e-05,
      "loss": 4.1578,
      "step": 462848
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.24134691420062e-05,
      "loss": 4.1502,
      "step": 463360
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.240508319449568e-05,
      "loss": 4.1543,
      "step": 463872
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.239669724698516e-05,
      "loss": 4.1401,
      "step": 464384
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.2388311299474637e-05,
      "loss": 4.1429,
      "step": 464896
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.2379925351964117e-05,
      "loss": 4.1496,
      "step": 465408
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.2371539404453596e-05,
      "loss": 4.1348,
      "step": 465920
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.2363153456943076e-05,
      "loss": 4.1439,
      "step": 466432
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.235476750943255e-05,
      "loss": 4.1492,
      "step": 466944
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.234638156192203e-05,
      "loss": 4.1417,
      "step": 467456
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.2337995614411516e-05,
      "loss": 4.1656,
      "step": 467968
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.2329609666900996e-05,
      "loss": 4.1554,
      "step": 468480
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.2321223719390476e-05,
      "loss": 4.1436,
      "step": 468992
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.2312837771879956e-05,
      "loss": 4.1498,
      "step": 469504
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.2304468203173165e-05,
      "loss": 4.1629,
      "step": 470016
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.2296082255662645e-05,
      "loss": 4.1391,
      "step": 470528
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.2287696308152125e-05,
      "loss": 4.1507,
      "step": 471040
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.2279310360641605e-05,
      "loss": 4.1426,
      "step": 471552
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.2270924413131085e-05,
      "loss": 4.1423,
      "step": 472064
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.2262538465620565e-05,
      "loss": 4.1371,
      "step": 472576
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.2254152518110045e-05,
      "loss": 4.1478,
      "step": 473088
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.2245766570599525e-05,
      "loss": 4.1553,
      "step": 473600
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.2237397001892734e-05,
      "loss": 4.1466,
      "step": 474112
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.2229011054382214e-05,
      "loss": 4.1534,
      "step": 474624
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.22206251068717e-05,
      "loss": 4.1496,
      "step": 475136
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.221225553816491e-05,
      "loss": 4.1426,
      "step": 475648
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.220386959065439e-05,
      "loss": 4.1475,
      "step": 476160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.219548364314387e-05,
      "loss": 4.1377,
      "step": 476672
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.218711407443708e-05,
      "loss": 4.1358,
      "step": 477184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.217872812692656e-05,
      "loss": 4.1343,
      "step": 477696
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.217034217941604e-05,
      "loss": 4.1343,
      "step": 478208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.216195623190552e-05,
      "loss": 4.1412,
      "step": 478720
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.2153570284395e-05,
      "loss": 4.1468,
      "step": 479232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.214518433688448e-05,
      "loss": 4.1361,
      "step": 479744
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.213679838937396e-05,
      "loss": 4.1372,
      "step": 480256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.212841244186344e-05,
      "loss": 4.148,
      "step": 480768
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.212002649435292e-05,
      "loss": 4.1409,
      "step": 481280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.21116405468424e-05,
      "loss": 4.1465,
      "step": 481792
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.210325459933188e-05,
      "loss": 4.1389,
      "step": 482304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.209486865182136e-05,
      "loss": 4.1201,
      "step": 482816
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.208649908311457e-05,
      "loss": 4.1515,
      "step": 483328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.207812951440778e-05,
      "loss": 4.1314,
      "step": 483840
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.206974356689726e-05,
      "loss": 4.1451,
      "step": 484352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.2061357619386736e-05,
      "loss": 4.134,
      "step": 484864
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.2052971671876216e-05,
      "loss": 4.1357,
      "step": 485376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.204460210316943e-05,
      "loss": 4.1289,
      "step": 485888
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.203621615565891e-05,
      "loss": 4.1424,
      "step": 486400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.2027830208148385e-05,
      "loss": 4.1245,
      "step": 486912
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.201944426063787e-05,
      "loss": 4.1331,
      "step": 487424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.201105831312735e-05,
      "loss": 4.1429,
      "step": 487936
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.200267236561683e-05,
      "loss": 4.1234,
      "step": 488448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.199428641810631e-05,
      "loss": 4.1269,
      "step": 488960
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.198590047059579e-05,
      "loss": 4.1443,
      "step": 489472
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.197751452308527e-05,
      "loss": 4.1143,
      "step": 489984
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.196912857557475e-05,
      "loss": 4.1184,
      "step": 490496
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.196074262806423e-05,
      "loss": 4.1328,
      "step": 491008
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.195235668055371e-05,
      "loss": 4.1364,
      "step": 491520
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.194398711184692e-05,
      "loss": 4.1246,
      "step": 492032
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.193561754314013e-05,
      "loss": 4.1364,
      "step": 492544
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.192723159562961e-05,
      "loss": 4.1147,
      "step": 493056
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.191884564811909e-05,
      "loss": 4.1259,
      "step": 493568
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.191045970060857e-05,
      "loss": 4.1367,
      "step": 494080
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1902073753098057e-05,
      "loss": 4.1336,
      "step": 494592
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1893704184391266e-05,
      "loss": 4.1376,
      "step": 495104
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1885318236880746e-05,
      "loss": 4.1378,
      "step": 495616
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1876932289370226e-05,
      "loss": 4.1461,
      "step": 496128
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1868562720663435e-05,
      "loss": 4.1192,
      "step": 496640
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1860176773152915e-05,
      "loss": 4.1307,
      "step": 497152
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1851790825642395e-05,
      "loss": 4.1363,
      "step": 497664
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1843404878131875e-05,
      "loss": 4.1131,
      "step": 498176
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1835018930621355e-05,
      "loss": 4.1338,
      "step": 498688
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1826632983110834e-05,
      "loss": 4.1305,
      "step": 499200
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1818247035600314e-05,
      "loss": 4.1354,
      "step": 499712
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1809861088089794e-05,
      "loss": 4.1347,
      "step": 500224
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1801475140579274e-05,
      "loss": 4.1227,
      "step": 500736
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1793089193068754e-05,
      "loss": 4.1257,
      "step": 501248
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1784703245558234e-05,
      "loss": 4.1299,
      "step": 501760
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.1776317298047714e-05,
      "loss": 4.1333,
      "step": 502272
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.176794772934093e-05,
      "loss": 4.131,
      "step": 502784
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.17595617818304e-05,
      "loss": 4.138,
      "step": 503296
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.175119221312362e-05,
      "loss": 4.1372,
      "step": 503808
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.17428062656131e-05,
      "loss": 4.1251,
      "step": 504320
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.173442031810257e-05,
      "loss": 4.13,
      "step": 504832
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.172603437059205e-05,
      "loss": 4.1196,
      "step": 505344
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.171764842308153e-05,
      "loss": 4.1384,
      "step": 505856
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.170926247557101e-05,
      "loss": 4.1376,
      "step": 506368
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.170087652806049e-05,
      "loss": 4.1232,
      "step": 506880
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.169249058054998e-05,
      "loss": 4.1217,
      "step": 507392
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.168412101184319e-05,
      "loss": 4.1168,
      "step": 507904
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.167573506433267e-05,
      "loss": 4.1331,
      "step": 508416
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.166734911682215e-05,
      "loss": 4.1385,
      "step": 508928
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.165897954811536e-05,
      "loss": 4.1214,
      "step": 509440
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.165059360060484e-05,
      "loss": 4.1216,
      "step": 509952
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.164220765309432e-05,
      "loss": 4.1165,
      "step": 510464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.16338217055838e-05,
      "loss": 4.1286,
      "step": 510976
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.162543575807328e-05,
      "loss": 4.1096,
      "step": 511488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.161704981056276e-05,
      "loss": 4.1227,
      "step": 512000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.160866386305224e-05,
      "loss": 4.1146,
      "step": 512512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.160027791554172e-05,
      "loss": 4.128,
      "step": 513024
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1591908346834926e-05,
      "loss": 4.1187,
      "step": 513536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.158353877812814e-05,
      "loss": 4.1262,
      "step": 514048
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.157515283061762e-05,
      "loss": 4.1208,
      "step": 514560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.15667668831071e-05,
      "loss": 4.113,
      "step": 515072
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.155838093559658e-05,
      "loss": 4.1269,
      "step": 515584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.154999498808606e-05,
      "loss": 4.1244,
      "step": 516096
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.154162541937927e-05,
      "loss": 4.1295,
      "step": 516608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.153323947186875e-05,
      "loss": 4.1313,
      "step": 517120
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.152485352435823e-05,
      "loss": 4.1185,
      "step": 517632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.151646757684771e-05,
      "loss": 4.1215,
      "step": 518144
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.150808162933719e-05,
      "loss": 4.1144,
      "step": 518656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.149969568182667e-05,
      "loss": 4.1259,
      "step": 519168
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.149130973431615e-05,
      "loss": 4.1186,
      "step": 519680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.148292378680563e-05,
      "loss": 4.1206,
      "step": 520192
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1474554218098846e-05,
      "loss": 4.1073,
      "step": 520704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1466184649392055e-05,
      "loss": 4.1238,
      "step": 521216
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1457815080685264e-05,
      "loss": 4.1138,
      "step": 521728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1449429133174744e-05,
      "loss": 4.1206,
      "step": 522240
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1441043185664224e-05,
      "loss": 4.1038,
      "step": 522752
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1432657238153704e-05,
      "loss": 4.1206,
      "step": 523264
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1424271290643184e-05,
      "loss": 4.1154,
      "step": 523776
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1415885343132664e-05,
      "loss": 4.1294,
      "step": 524288
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1407499395622144e-05,
      "loss": 4.1181,
      "step": 524800
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1399113448111624e-05,
      "loss": 4.125,
      "step": 525312
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1390727500601104e-05,
      "loss": 4.1083,
      "step": 525824
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1382341553090584e-05,
      "loss": 4.1235,
      "step": 526336
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1373955605580064e-05,
      "loss": 4.1275,
      "step": 526848
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1365569658069544e-05,
      "loss": 4.1249,
      "step": 527360
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.135720008936276e-05,
      "loss": 4.1238,
      "step": 527872
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.134881414185223e-05,
      "loss": 4.1277,
      "step": 528384
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.134042819434171e-05,
      "loss": 4.1157,
      "step": 528896
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.133204224683119e-05,
      "loss": 4.1267,
      "step": 529408
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.132365629932067e-05,
      "loss": 4.107,
      "step": 529920
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.131528673061388e-05,
      "loss": 4.1181,
      "step": 530432
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.130690078310336e-05,
      "loss": 4.1118,
      "step": 530944
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.129853121439658e-05,
      "loss": 4.1145,
      "step": 531456
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.129014526688606e-05,
      "loss": 4.1105,
      "step": 531968
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.128175931937554e-05,
      "loss": 4.1088,
      "step": 532480
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.127337337186502e-05,
      "loss": 4.1195,
      "step": 532992
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.126500380315823e-05,
      "loss": 4.1104,
      "step": 533504
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.1256617855647707e-05,
      "loss": 4.1289,
      "step": 534016
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.122705459594727,
      "eval_runtime": 295.5193,
      "eval_samples_per_second": 1291.256,
      "eval_steps_per_second": 40.353,
      "step": 534240
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.1248231908137186e-05,
      "loss": 4.1089,
      "step": 534528
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.1239845960626666e-05,
      "loss": 4.1035,
      "step": 535040
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.1231460013116146e-05,
      "loss": 4.1183,
      "step": 535552
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.1223074065605626e-05,
      "loss": 4.1133,
      "step": 536064
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.1214704496898835e-05,
      "loss": 4.1119,
      "step": 536576
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.1206318549388315e-05,
      "loss": 4.113,
      "step": 537088
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.1197932601877795e-05,
      "loss": 4.1042,
      "step": 537600
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.1189546654367275e-05,
      "loss": 4.1071,
      "step": 538112
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.1181160706856755e-05,
      "loss": 4.1134,
      "step": 538624
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.117277475934624e-05,
      "loss": 4.1207,
      "step": 539136
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.116438881183572e-05,
      "loss": 4.1122,
      "step": 539648
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.115601924312893e-05,
      "loss": 4.1114,
      "step": 540160
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.114763329561841e-05,
      "loss": 4.1078,
      "step": 540672
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.113924734810789e-05,
      "loss": 4.1046,
      "step": 541184
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.113086140059737e-05,
      "loss": 4.1086,
      "step": 541696
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.112247545308685e-05,
      "loss": 4.0964,
      "step": 542208
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.111410588438006e-05,
      "loss": 4.1082,
      "step": 542720
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.110571993686954e-05,
      "loss": 4.1121,
      "step": 543232
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.109735036816275e-05,
      "loss": 4.1023,
      "step": 543744
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.108896442065223e-05,
      "loss": 4.1257,
      "step": 544256
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.108057847314171e-05,
      "loss": 4.1214,
      "step": 544768
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.1072192525631196e-05,
      "loss": 4.1064,
      "step": 545280
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.1063806578120676e-05,
      "loss": 4.1121,
      "step": 545792
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.1055420630610156e-05,
      "loss": 4.123,
      "step": 546304
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.1047034683099636e-05,
      "loss": 4.1015,
      "step": 546816
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.1038648735589115e-05,
      "loss": 4.1138,
      "step": 547328
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.1030279166882325e-05,
      "loss": 4.1069,
      "step": 547840
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.1021893219371805e-05,
      "loss": 4.1047,
      "step": 548352
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.1013507271861285e-05,
      "loss": 4.0987,
      "step": 548864
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.1005121324350764e-05,
      "loss": 4.1137,
      "step": 549376
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0996735376840244e-05,
      "loss": 4.1167,
      "step": 549888
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0988365808133454e-05,
      "loss": 4.1138,
      "step": 550400
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0979979860622933e-05,
      "loss": 4.116,
      "step": 550912
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0971593913112413e-05,
      "loss": 4.107,
      "step": 551424
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.096320796560189e-05,
      "loss": 4.1053,
      "step": 551936
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.095483839689511e-05,
      "loss": 4.1126,
      "step": 552448
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.094645244938459e-05,
      "loss": 4.1032,
      "step": 552960
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.09380828806778e-05,
      "loss": 4.0947,
      "step": 553472
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.092969693316728e-05,
      "loss": 4.0977,
      "step": 553984
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.092131098565676e-05,
      "loss": 4.0984,
      "step": 554496
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.091292503814624e-05,
      "loss": 4.1027,
      "step": 555008
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.090453909063572e-05,
      "loss": 4.1103,
      "step": 555520
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.089615314312519e-05,
      "loss": 4.1032,
      "step": 556032
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.088776719561467e-05,
      "loss": 4.0982,
      "step": 556544
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.087938124810415e-05,
      "loss": 4.1121,
      "step": 557056
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.087101167939737e-05,
      "loss": 4.1076,
      "step": 557568
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.086262573188685e-05,
      "loss": 4.108,
      "step": 558080
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.085423978437633e-05,
      "loss": 4.1047,
      "step": 558592
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.084585383686581e-05,
      "loss": 4.0793,
      "step": 559104
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.083746788935529e-05,
      "loss": 4.1193,
      "step": 559616
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.082908194184477e-05,
      "loss": 4.0957,
      "step": 560128
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.082069599433425e-05,
      "loss": 4.1106,
      "step": 560640
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.081231004682373e-05,
      "loss": 4.0963,
      "step": 561152
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.080392409931321e-05,
      "loss": 4.1006,
      "step": 561664
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0795554530606416e-05,
      "loss": 4.0924,
      "step": 562176
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0787168583095896e-05,
      "loss": 4.104,
      "step": 562688
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0778799014389105e-05,
      "loss": 4.0917,
      "step": 563200
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0770413066878585e-05,
      "loss": 4.0959,
      "step": 563712
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0762027119368065e-05,
      "loss": 4.109,
      "step": 564224
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.075364117185755e-05,
      "loss": 4.0896,
      "step": 564736
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.074525522434703e-05,
      "loss": 4.0877,
      "step": 565248
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.073686927683651e-05,
      "loss": 4.1073,
      "step": 565760
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.072848332932599e-05,
      "loss": 4.083,
      "step": 566272
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.072009738181547e-05,
      "loss": 4.0815,
      "step": 566784
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.071172781310868e-05,
      "loss": 4.0983,
      "step": 567296
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.070334186559816e-05,
      "loss": 4.0997,
      "step": 567808
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.069495591808764e-05,
      "loss": 4.0908,
      "step": 568320
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.068656997057712e-05,
      "loss": 4.1007,
      "step": 568832
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.067820040187033e-05,
      "loss": 4.0773,
      "step": 569344
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.066981445435981e-05,
      "loss": 4.0919,
      "step": 569856
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.066142850684929e-05,
      "loss": 4.1013,
      "step": 570368
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.065304255933877e-05,
      "loss": 4.0959,
      "step": 570880
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0644672990631985e-05,
      "loss": 4.1058,
      "step": 571392
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0636287043121465e-05,
      "loss": 4.1024,
      "step": 571904
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0627901095610945e-05,
      "loss": 4.1085,
      "step": 572416
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0619531526904154e-05,
      "loss": 4.0834,
      "step": 572928
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0611145579393634e-05,
      "loss": 4.0989,
      "step": 573440
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0602759631883114e-05,
      "loss": 4.0995,
      "step": 573952
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0594373684372594e-05,
      "loss": 4.0815,
      "step": 574464
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.0585987736862074e-05,
      "loss": 4.0977,
      "step": 574976
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.057761816815528e-05,
      "loss": 4.0974,
      "step": 575488
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.056923222064476e-05,
      "loss": 4.1008,
      "step": 576000
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.056084627313424e-05,
      "loss": 4.1044,
      "step": 576512
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.055246032562372e-05,
      "loss": 4.0833,
      "step": 577024
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.054409075691694e-05,
      "loss": 4.0905,
      "step": 577536
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.053570480940642e-05,
      "loss": 4.0961,
      "step": 578048
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.052733524069963e-05,
      "loss": 4.1003,
      "step": 578560
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.051894929318911e-05,
      "loss": 4.0991,
      "step": 579072
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.051056334567859e-05,
      "loss": 4.0999,
      "step": 579584
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.050217739816807e-05,
      "loss": 4.1023,
      "step": 580096
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.049379145065755e-05,
      "loss": 4.0947,
      "step": 580608
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.048540550314702e-05,
      "loss": 4.0946,
      "step": 581120
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.04770195556365e-05,
      "loss": 4.0884,
      "step": 581632
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.046863360812598e-05,
      "loss": 4.1015,
      "step": 582144
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.046024766061546e-05,
      "loss": 4.1013,
      "step": 582656
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.045186171310494e-05,
      "loss": 4.0929,
      "step": 583168
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0443492144398157e-05,
      "loss": 4.0851,
      "step": 583680
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0435106196887637e-05,
      "loss": 4.0837,
      "step": 584192
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0426720249377116e-05,
      "loss": 4.0975,
      "step": 584704
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0418350680670326e-05,
      "loss": 4.1071,
      "step": 585216
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0409964733159806e-05,
      "loss": 4.0887,
      "step": 585728
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0401578785649285e-05,
      "loss": 4.0883,
      "step": 586240
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0393192838138765e-05,
      "loss": 4.0869,
      "step": 586752
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0384806890628245e-05,
      "loss": 4.0912,
      "step": 587264
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0376420943117725e-05,
      "loss": 4.0786,
      "step": 587776
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0368034995607205e-05,
      "loss": 4.0916,
      "step": 588288
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0359649048096685e-05,
      "loss": 4.0786,
      "step": 588800
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0351263100586165e-05,
      "loss": 4.0926,
      "step": 589312
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0342877153075645e-05,
      "loss": 4.0895,
      "step": 589824
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0334491205565125e-05,
      "loss": 4.0902,
      "step": 590336
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0326105258054605e-05,
      "loss": 4.0863,
      "step": 590848
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.031773568934782e-05,
      "loss": 4.086,
      "step": 591360
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.030936612064103e-05,
      "loss": 4.0903,
      "step": 591872
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.030098017313051e-05,
      "loss": 4.0914,
      "step": 592384
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.029259422561999e-05,
      "loss": 4.0953,
      "step": 592896
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.028420827810947e-05,
      "loss": 4.1023,
      "step": 593408
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.027582233059895e-05,
      "loss": 4.0864,
      "step": 593920
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.026743638308843e-05,
      "loss": 4.0893,
      "step": 594432
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.025905043557791e-05,
      "loss": 4.083,
      "step": 594944
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.025068086687112e-05,
      "loss": 4.0898,
      "step": 595456
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.02422949193606e-05,
      "loss": 4.0879,
      "step": 595968
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.023390897185008e-05,
      "loss": 4.0874,
      "step": 596480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.022552302433956e-05,
      "loss": 4.0759,
      "step": 596992
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.021713707682904e-05,
      "loss": 4.0885,
      "step": 597504
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0208767508122255e-05,
      "loss": 4.0865,
      "step": 598016
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0200381560611735e-05,
      "loss": 4.0859,
      "step": 598528
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0191995613101214e-05,
      "loss": 4.0679,
      "step": 599040
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.018360966559069e-05,
      "loss": 4.0934,
      "step": 599552
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.017522371808017e-05,
      "loss": 4.0811,
      "step": 600064
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.016683777056965e-05,
      "loss": 4.0953,
      "step": 600576
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.015846820186286e-05,
      "loss": 4.0879,
      "step": 601088
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.015008225435234e-05,
      "loss": 4.0923,
      "step": 601600
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0141696306841817e-05,
      "loss": 4.0757,
      "step": 602112
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.0133310359331297e-05,
      "loss": 4.0933,
      "step": 602624
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.012492441182078e-05,
      "loss": 4.095,
      "step": 603136
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.011653846431026e-05,
      "loss": 4.0927,
      "step": 603648
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.010815251679974e-05,
      "loss": 4.0929,
      "step": 604160
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.009976656928922e-05,
      "loss": 4.0931,
      "step": 604672
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.009139700058243e-05,
      "loss": 4.0844,
      "step": 605184
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.008301105307191e-05,
      "loss": 4.0942,
      "step": 605696
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.007462510556139e-05,
      "loss": 4.0774,
      "step": 606208
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.006623915805087e-05,
      "loss": 4.089,
      "step": 606720
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.005786958934408e-05,
      "loss": 4.0804,
      "step": 607232
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.004948364183356e-05,
      "loss": 4.0884,
      "step": 607744
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.004111407312677e-05,
      "loss": 4.0747,
      "step": 608256
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.003272812561625e-05,
      "loss": 4.0786,
      "step": 608768
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.002434217810574e-05,
      "loss": 4.0837,
      "step": 609280
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.001595623059522e-05,
      "loss": 4.0832,
      "step": 609792
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.00075702830847e-05,
      "loss": 4.0969,
      "step": 610304
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.099428653717041,
      "eval_runtime": 294.2439,
      "eval_samples_per_second": 1296.853,
      "eval_steps_per_second": 40.528,
      "step": 610560
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.999918433557418e-05,
      "loss": 4.0751,
      "step": 610816
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.999079838806366e-05,
      "loss": 4.0768,
      "step": 611328
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.998241244055314e-05,
      "loss": 4.08,
      "step": 611840
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.997402649304262e-05,
      "loss": 4.0823,
      "step": 612352
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.99656405455321e-05,
      "loss": 4.0854,
      "step": 612864
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.9957254598021577e-05,
      "loss": 4.078,
      "step": 613376
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.994886865051105e-05,
      "loss": 4.0829,
      "step": 613888
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.994048270300053e-05,
      "loss": 4.0761,
      "step": 614400
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.993209675549001e-05,
      "loss": 4.0778,
      "step": 614912
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.992371080797949e-05,
      "loss": 4.0932,
      "step": 615424
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.991532486046897e-05,
      "loss": 4.0784,
      "step": 615936
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.990693891295845e-05,
      "loss": 4.0814,
      "step": 616448
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.989855296544793e-05,
      "loss": 4.0765,
      "step": 616960
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.9890167017937416e-05,
      "loss": 4.0699,
      "step": 617472
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.9881781070426896e-05,
      "loss": 4.0824,
      "step": 617984
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.9873395122916376e-05,
      "loss": 4.0667,
      "step": 618496
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.9865009175405856e-05,
      "loss": 4.0787,
      "step": 619008
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.9856623227895336e-05,
      "loss": 4.0826,
      "step": 619520
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.9848237280384816e-05,
      "loss": 4.0702,
      "step": 620032
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.983985133287429e-05,
      "loss": 4.0927,
      "step": 620544
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.9831481764167505e-05,
      "loss": 4.0935,
      "step": 621056
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.9823095816656985e-05,
      "loss": 4.0788,
      "step": 621568
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.981470986914646e-05,
      "loss": 4.0788,
      "step": 622080
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.980632392163594e-05,
      "loss": 4.0911,
      "step": 622592
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.9797954352929154e-05,
      "loss": 4.0777,
      "step": 623104
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.9789568405418634e-05,
      "loss": 4.0773,
      "step": 623616
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.9781182457908114e-05,
      "loss": 4.077,
      "step": 624128
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.9772796510397594e-05,
      "loss": 4.0747,
      "step": 624640
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.976442694169081e-05,
      "loss": 4.0762,
      "step": 625152
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.975604099418029e-05,
      "loss": 4.0779,
      "step": 625664
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.974765504666976e-05,
      "loss": 4.082,
      "step": 626176
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.973928547796298e-05,
      "loss": 4.0857,
      "step": 626688
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.973089953045246e-05,
      "loss": 4.0862,
      "step": 627200
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.972251358294193e-05,
      "loss": 4.079,
      "step": 627712
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.971412763543141e-05,
      "loss": 4.0741,
      "step": 628224
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.970574168792089e-05,
      "loss": 4.0844,
      "step": 628736
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.969737211921411e-05,
      "loss": 4.0747,
      "step": 629248
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.968898617170359e-05,
      "loss": 4.066,
      "step": 629760
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.968060022419307e-05,
      "loss": 4.0697,
      "step": 630272
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.967221427668255e-05,
      "loss": 4.0669,
      "step": 630784
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.966382832917203e-05,
      "loss": 4.0732,
      "step": 631296
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.965544238166151e-05,
      "loss": 4.0785,
      "step": 631808
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.964705643415099e-05,
      "loss": 4.0764,
      "step": 632320
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.963867048664047e-05,
      "loss": 4.0735,
      "step": 632832
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.963028453912995e-05,
      "loss": 4.0772,
      "step": 633344
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9621914970423156e-05,
      "loss": 4.0789,
      "step": 633856
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9613545401716366e-05,
      "loss": 4.0744,
      "step": 634368
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9605159454205845e-05,
      "loss": 4.0774,
      "step": 634880
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9596773506695325e-05,
      "loss": 4.0509,
      "step": 635392
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9588387559184805e-05,
      "loss": 4.0891,
      "step": 635904
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.958000161167429e-05,
      "loss": 4.0667,
      "step": 636416
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.957161566416377e-05,
      "loss": 4.084,
      "step": 636928
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.956322971665325e-05,
      "loss": 4.0655,
      "step": 637440
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.955484376914273e-05,
      "loss": 4.0693,
      "step": 637952
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.954647420043594e-05,
      "loss": 4.0665,
      "step": 638464
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.953808825292542e-05,
      "loss": 4.0733,
      "step": 638976
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.95297023054149e-05,
      "loss": 4.0644,
      "step": 639488
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.952131635790438e-05,
      "loss": 4.0671,
      "step": 640000
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.951294678919759e-05,
      "loss": 4.0782,
      "step": 640512
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.95045772204908e-05,
      "loss": 4.0649,
      "step": 641024
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.949619127298028e-05,
      "loss": 4.0563,
      "step": 641536
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.948780532546976e-05,
      "loss": 4.0807,
      "step": 642048
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9479419377959246e-05,
      "loss": 4.054,
      "step": 642560
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9471033430448726e-05,
      "loss": 4.0511,
      "step": 643072
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9462647482938206e-05,
      "loss": 4.0704,
      "step": 643584
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9454261535427686e-05,
      "loss": 4.0723,
      "step": 644096
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9445875587917166e-05,
      "loss": 4.0648,
      "step": 644608
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9437506019210375e-05,
      "loss": 4.069,
      "step": 645120
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9429120071699855e-05,
      "loss": 4.0509,
      "step": 645632
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9420734124189335e-05,
      "loss": 4.0615,
      "step": 646144
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9412348176678815e-05,
      "loss": 4.0736,
      "step": 646656
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9403962229168295e-05,
      "loss": 4.0669,
      "step": 647168
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9395592660461504e-05,
      "loss": 4.0789,
      "step": 647680
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9387206712950984e-05,
      "loss": 4.0693,
      "step": 648192
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.9378820765440464e-05,
      "loss": 4.0823,
      "step": 648704
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.937045119673368e-05,
      "loss": 4.0558,
      "step": 649216
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.936206524922316e-05,
      "loss": 4.0705,
      "step": 649728
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.935367930171264e-05,
      "loss": 4.067,
      "step": 650240
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.934529335420212e-05,
      "loss": 4.0583,
      "step": 650752
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.933690740669159e-05,
      "loss": 4.068,
      "step": 651264
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.932852145918107e-05,
      "loss": 4.0717,
      "step": 651776
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.932013551167055e-05,
      "loss": 4.0699,
      "step": 652288
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.931176594296377e-05,
      "loss": 4.0776,
      "step": 652800
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.930337999545324e-05,
      "loss": 4.0536,
      "step": 653312
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.929499404794272e-05,
      "loss": 4.0591,
      "step": 653824
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.92866081004322e-05,
      "loss": 4.0718,
      "step": 654336
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.927822215292168e-05,
      "loss": 4.0697,
      "step": 654848
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.926983620541116e-05,
      "loss": 4.0707,
      "step": 655360
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.926145025790065e-05,
      "loss": 4.0739,
      "step": 655872
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.925308068919386e-05,
      "loss": 4.0695,
      "step": 656384
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.924469474168334e-05,
      "loss": 4.0691,
      "step": 656896
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.923630879417282e-05,
      "loss": 4.0658,
      "step": 657408
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.92279228466623e-05,
      "loss": 4.0592,
      "step": 657920
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.9219553277955506e-05,
      "loss": 4.0773,
      "step": 658432
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.9211167330444986e-05,
      "loss": 4.0694,
      "step": 658944
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.9202781382934466e-05,
      "loss": 4.0668,
      "step": 659456
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.9194395435423946e-05,
      "loss": 4.0614,
      "step": 659968
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.9186009487913426e-05,
      "loss": 4.0539,
      "step": 660480
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.9177639919206635e-05,
      "loss": 4.0673,
      "step": 660992
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.9169253971696115e-05,
      "loss": 4.0769,
      "step": 661504
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.91608680241856e-05,
      "loss": 4.0621,
      "step": 662016
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.915248207667508e-05,
      "loss": 4.0615,
      "step": 662528
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.914409612916456e-05,
      "loss": 4.0589,
      "step": 663040
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.913572656045777e-05,
      "loss": 4.0606,
      "step": 663552
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.912734061294725e-05,
      "loss": 4.0582,
      "step": 664064
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.911895466543673e-05,
      "loss": 4.0575,
      "step": 664576
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.911056871792621e-05,
      "loss": 4.0505,
      "step": 665088
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.910218277041569e-05,
      "loss": 4.068,
      "step": 665600
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.90938132017089e-05,
      "loss": 4.0642,
      "step": 666112
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.908542725419838e-05,
      "loss": 4.0623,
      "step": 666624
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.907704130668786e-05,
      "loss": 4.0539,
      "step": 667136
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.906865535917734e-05,
      "loss": 4.0629,
      "step": 667648
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.906026941166682e-05,
      "loss": 4.0636,
      "step": 668160
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.90518834641563e-05,
      "loss": 4.063,
      "step": 668672
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.9043497516645786e-05,
      "loss": 4.0687,
      "step": 669184
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.9035127947938995e-05,
      "loss": 4.0728,
      "step": 669696
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.9026758379232204e-05,
      "loss": 4.0574,
      "step": 670208
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.901838881052541e-05,
      "loss": 4.0655,
      "step": 670720
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.901000286301489e-05,
      "loss": 4.0501,
      "step": 671232
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.900161691550437e-05,
      "loss": 4.0635,
      "step": 671744
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.899323096799385e-05,
      "loss": 4.0614,
      "step": 672256
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.898484502048333e-05,
      "loss": 4.0581,
      "step": 672768
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.897645907297281e-05,
      "loss": 4.0532,
      "step": 673280
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.896807312546229e-05,
      "loss": 4.0619,
      "step": 673792
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.895968717795177e-05,
      "loss": 4.0612,
      "step": 674304
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.895131760924499e-05,
      "loss": 4.0583,
      "step": 674816
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.894293166173447e-05,
      "loss": 4.0433,
      "step": 675328
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.893454571422395e-05,
      "loss": 4.0626,
      "step": 675840
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.892615976671343e-05,
      "loss": 4.0543,
      "step": 676352
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.89177738192029e-05,
      "loss": 4.0678,
      "step": 676864
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.890938787169238e-05,
      "loss": 4.0629,
      "step": 677376
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.890100192418186e-05,
      "loss": 4.0607,
      "step": 677888
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.889261597667134e-05,
      "loss": 4.0514,
      "step": 678400
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.888424640796455e-05,
      "loss": 4.0671,
      "step": 678912
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.887587683925777e-05,
      "loss": 4.0676,
      "step": 679424
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.886749089174725e-05,
      "loss": 4.0668,
      "step": 679936
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.885910494423673e-05,
      "loss": 4.0672,
      "step": 680448
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.885071899672621e-05,
      "loss": 4.0655,
      "step": 680960
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.884233304921569e-05,
      "loss": 4.0594,
      "step": 681472
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.8833947101705167e-05,
      "loss": 4.066,
      "step": 681984
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.8825561154194647e-05,
      "loss": 4.0494,
      "step": 682496
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.8817175206684127e-05,
      "loss": 4.0617,
      "step": 683008
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.8808805637977336e-05,
      "loss": 4.0554,
      "step": 683520
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.8800419690466816e-05,
      "loss": 4.0581,
      "step": 684032
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.8792033742956296e-05,
      "loss": 4.0489,
      "step": 684544
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.8783664174249505e-05,
      "loss": 4.0517,
      "step": 685056
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.8775278226738985e-05,
      "loss": 4.0579,
      "step": 685568
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.87669086580322e-05,
      "loss": 4.0589,
      "step": 686080
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.875852271052168e-05,
      "loss": 4.0686,
      "step": 686592
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.08057165145874,
      "eval_runtime": 294.6161,
      "eval_samples_per_second": 1295.214,
      "eval_steps_per_second": 40.476,
      "step": 686880
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.875013676301116e-05,
      "loss": 4.045,
      "step": 687104
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.874175081550064e-05,
      "loss": 4.0497,
      "step": 687616
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.873336486799012e-05,
      "loss": 4.051,
      "step": 688128
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.87249789204796e-05,
      "loss": 4.0617,
      "step": 688640
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.871659297296908e-05,
      "loss": 4.0584,
      "step": 689152
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.870820702545856e-05,
      "loss": 4.0494,
      "step": 689664
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.869982107794804e-05,
      "loss": 4.0545,
      "step": 690176
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.869143513043752e-05,
      "loss": 4.0485,
      "step": 690688
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.8683049182927e-05,
      "loss": 4.0507,
      "step": 691200
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.867467961422021e-05,
      "loss": 4.0681,
      "step": 691712
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.866629366670969e-05,
      "loss": 4.0518,
      "step": 692224
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.865790771919917e-05,
      "loss": 4.052,
      "step": 692736
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.864952177168865e-05,
      "loss": 4.062,
      "step": 693248
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.864113582417813e-05,
      "loss": 4.0368,
      "step": 693760
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.863274987666761e-05,
      "loss": 4.0539,
      "step": 694272
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.862436392915709e-05,
      "loss": 4.041,
      "step": 694784
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.861597798164657e-05,
      "loss": 4.0533,
      "step": 695296
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.8607608412939785e-05,
      "loss": 4.0552,
      "step": 695808
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.8599222465429265e-05,
      "loss": 4.0441,
      "step": 696320
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.859083651791874e-05,
      "loss": 4.0656,
      "step": 696832
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.858245057040822e-05,
      "loss": 4.0673,
      "step": 697344
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.85740646228977e-05,
      "loss": 4.0559,
      "step": 697856
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.8565695054190914e-05,
      "loss": 4.0572,
      "step": 698368
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.855730910668039e-05,
      "loss": 4.0593,
      "step": 698880
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.854892315916987e-05,
      "loss": 4.0548,
      "step": 699392
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.854053721165935e-05,
      "loss": 4.0494,
      "step": 699904
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.853216764295256e-05,
      "loss": 4.0556,
      "step": 700416
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.852378169544204e-05,
      "loss": 4.0504,
      "step": 700928
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.851539574793152e-05,
      "loss": 4.0489,
      "step": 701440
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.8507009800421e-05,
      "loss": 4.0523,
      "step": 701952
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.849864023171421e-05,
      "loss": 4.0546,
      "step": 702464
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.849027066300743e-05,
      "loss": 4.0636,
      "step": 702976
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8481901094300636e-05,
      "loss": 4.0577,
      "step": 703488
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8473515146790116e-05,
      "loss": 4.0564,
      "step": 704000
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8465129199279596e-05,
      "loss": 4.0484,
      "step": 704512
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8456743251769076e-05,
      "loss": 4.0514,
      "step": 705024
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8448357304258556e-05,
      "loss": 4.0535,
      "step": 705536
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8439971356748036e-05,
      "loss": 4.0431,
      "step": 706048
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8431585409237516e-05,
      "loss": 4.0421,
      "step": 706560
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8423199461726996e-05,
      "loss": 4.0397,
      "step": 707072
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8414813514216476e-05,
      "loss": 4.0484,
      "step": 707584
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8406427566705956e-05,
      "loss": 4.0497,
      "step": 708096
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8398041619195436e-05,
      "loss": 4.0497,
      "step": 708608
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8389655671684916e-05,
      "loss": 4.0491,
      "step": 709120
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8381269724174396e-05,
      "loss": 4.0543,
      "step": 709632
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8372900155467605e-05,
      "loss": 4.0499,
      "step": 710144
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8364514207957085e-05,
      "loss": 4.0492,
      "step": 710656
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8356144639250294e-05,
      "loss": 4.056,
      "step": 711168
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8347758691739774e-05,
      "loss": 4.0243,
      "step": 711680
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8339372744229254e-05,
      "loss": 4.0566,
      "step": 712192
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.833098679671874e-05,
      "loss": 4.0438,
      "step": 712704
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.832260084920822e-05,
      "loss": 4.0602,
      "step": 713216
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.83142149016977e-05,
      "loss": 4.0423,
      "step": 713728
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.830582895418718e-05,
      "loss": 4.0445,
      "step": 714240
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.829744300667666e-05,
      "loss": 4.0365,
      "step": 714752
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.828905705916614e-05,
      "loss": 4.0471,
      "step": 715264
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.828068749045935e-05,
      "loss": 4.0442,
      "step": 715776
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.827230154294883e-05,
      "loss": 4.0383,
      "step": 716288
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.826391559543831e-05,
      "loss": 4.0531,
      "step": 716800
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.825552964792779e-05,
      "loss": 4.043,
      "step": 717312
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.824714370041727e-05,
      "loss": 4.0249,
      "step": 717824
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.823877413171048e-05,
      "loss": 4.0556,
      "step": 718336
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8230404563003694e-05,
      "loss": 4.0331,
      "step": 718848
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8222018615493174e-05,
      "loss": 4.0255,
      "step": 719360
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8213632667982654e-05,
      "loss": 4.0448,
      "step": 719872
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8205246720472134e-05,
      "loss": 4.045,
      "step": 720384
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8196860772961614e-05,
      "loss": 4.034,
      "step": 720896
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8188474825451094e-05,
      "loss": 4.0501,
      "step": 721408
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8180088877940574e-05,
      "loss": 4.0276,
      "step": 721920
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.817170293043005e-05,
      "loss": 4.0347,
      "step": 722432
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.816333336172326e-05,
      "loss": 4.0488,
      "step": 722944
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.815494741421274e-05,
      "loss": 4.0408,
      "step": 723456
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.814657784550595e-05,
      "loss": 4.0564,
      "step": 723968
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.813819189799543e-05,
      "loss": 4.0435,
      "step": 724480
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.812980595048491e-05,
      "loss": 4.0542,
      "step": 724992
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.812143638177813e-05,
      "loss": 4.0352,
      "step": 725504
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.811305043426761e-05,
      "loss": 4.0491,
      "step": 726016
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.810466448675709e-05,
      "loss": 4.0356,
      "step": 726528
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.809627853924657e-05,
      "loss": 4.0376,
      "step": 727040
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.808789259173605e-05,
      "loss": 4.0402,
      "step": 727552
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.807950664422552e-05,
      "loss": 4.0441,
      "step": 728064
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8071120696715e-05,
      "loss": 4.0458,
      "step": 728576
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.806273474920448e-05,
      "loss": 4.0554,
      "step": 729088
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.805436518049769e-05,
      "loss": 4.0283,
      "step": 729600
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.804597923298717e-05,
      "loss": 4.0351,
      "step": 730112
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.803759328547665e-05,
      "loss": 4.0474,
      "step": 730624
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.802920733796613e-05,
      "loss": 4.0466,
      "step": 731136
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.802082139045561e-05,
      "loss": 4.0429,
      "step": 731648
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.8012435442945097e-05,
      "loss": 4.0522,
      "step": 732160
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.8004065874238306e-05,
      "loss": 4.0436,
      "step": 732672
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7995679926727786e-05,
      "loss": 4.0451,
      "step": 733184
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7987293979217266e-05,
      "loss": 4.0422,
      "step": 733696
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7978908031706746e-05,
      "loss": 4.0344,
      "step": 734208
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7970522084196225e-05,
      "loss": 4.0478,
      "step": 734720
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7962136136685705e-05,
      "loss": 4.0478,
      "step": 735232
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7953750189175185e-05,
      "loss": 4.0408,
      "step": 735744
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7945364241664665e-05,
      "loss": 4.039,
      "step": 736256
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7936994672957874e-05,
      "loss": 4.0268,
      "step": 736768
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7928608725447354e-05,
      "loss": 4.0431,
      "step": 737280
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7920239156740564e-05,
      "loss": 4.0529,
      "step": 737792
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.791185320923005e-05,
      "loss": 4.0389,
      "step": 738304
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.790348364052326e-05,
      "loss": 4.0385,
      "step": 738816
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.789509769301274e-05,
      "loss": 4.0332,
      "step": 739328
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.788671174550222e-05,
      "loss": 4.0324,
      "step": 739840
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.78783257979917e-05,
      "loss": 4.0375,
      "step": 740352
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.786993985048118e-05,
      "loss": 4.0317,
      "step": 740864
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.786155390297066e-05,
      "loss": 4.0285,
      "step": 741376
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.785316795546014e-05,
      "loss": 4.0429,
      "step": 741888
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.784478200794962e-05,
      "loss": 4.0435,
      "step": 742400
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.78363960604391e-05,
      "loss": 4.0328,
      "step": 742912
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.782801011292858e-05,
      "loss": 4.033,
      "step": 743424
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.781964054422179e-05,
      "loss": 4.035,
      "step": 743936
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.781125459671127e-05,
      "loss": 4.0436,
      "step": 744448
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.780286864920075e-05,
      "loss": 4.0361,
      "step": 744960
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7794482701690235e-05,
      "loss": 4.0473,
      "step": 745472
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.778609675417971e-05,
      "loss": 4.0472,
      "step": 745984
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7777727185472924e-05,
      "loss": 4.034,
      "step": 746496
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7769341237962404e-05,
      "loss": 4.0416,
      "step": 747008
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.776097166925561e-05,
      "loss": 4.0302,
      "step": 747520
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.775258572174509e-05,
      "loss": 4.039,
      "step": 748032
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.774419977423457e-05,
      "loss": 4.0419,
      "step": 748544
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.773581382672405e-05,
      "loss": 4.0317,
      "step": 749056
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7727427879213526e-05,
      "loss": 4.0269,
      "step": 749568
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7719041931703006e-05,
      "loss": 4.0393,
      "step": 750080
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.771067236299622e-05,
      "loss": 4.0375,
      "step": 750592
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.77022864154857e-05,
      "loss": 4.037,
      "step": 751104
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.769390046797518e-05,
      "loss": 4.0192,
      "step": 751616
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.768551452046466e-05,
      "loss": 4.0356,
      "step": 752128
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.767712857295414e-05,
      "loss": 4.0299,
      "step": 752640
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.766874262544362e-05,
      "loss": 4.0449,
      "step": 753152
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.76603566779331e-05,
      "loss": 4.041,
      "step": 753664
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.765197073042258e-05,
      "loss": 4.0372,
      "step": 754176
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.764360116171579e-05,
      "loss": 4.0249,
      "step": 754688
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.763521521420527e-05,
      "loss": 4.0472,
      "step": 755200
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.762682926669475e-05,
      "loss": 4.0396,
      "step": 755712
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.761844331918423e-05,
      "loss": 4.0444,
      "step": 756224
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.761005737167371e-05,
      "loss": 4.0394,
      "step": 756736
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.760167142416319e-05,
      "loss": 4.0409,
      "step": 757248
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7593301855456406e-05,
      "loss": 4.0362,
      "step": 757760
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7584915907945886e-05,
      "loss": 4.0483,
      "step": 758272
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7576529960435366e-05,
      "loss": 4.0226,
      "step": 758784
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7568144012924846e-05,
      "loss": 4.041,
      "step": 759296
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7559774444218055e-05,
      "loss": 4.0311,
      "step": 759808
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7551388496707535e-05,
      "loss": 4.0339,
      "step": 760320
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7543002549197015e-05,
      "loss": 4.0284,
      "step": 760832
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7534616601686495e-05,
      "loss": 4.0287,
      "step": 761344
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7526230654175975e-05,
      "loss": 4.0335,
      "step": 761856
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7517844706665455e-05,
      "loss": 4.0368,
      "step": 762368
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7509458759154935e-05,
      "loss": 4.042,
      "step": 762880
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.06446647644043,
      "eval_runtime": 295.0472,
      "eval_samples_per_second": 1293.322,
      "eval_steps_per_second": 40.417,
      "step": 763200
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7501072811644415e-05,
      "loss": 4.0223,
      "step": 763392
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.749268686413389e-05,
      "loss": 4.0293,
      "step": 763904
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7484300916623375e-05,
      "loss": 4.0268,
      "step": 764416
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7475914969112855e-05,
      "loss": 4.0393,
      "step": 764928
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7467529021602335e-05,
      "loss": 4.0396,
      "step": 765440
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7459143074091815e-05,
      "loss": 4.0265,
      "step": 765952
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7450757126581294e-05,
      "loss": 4.0305,
      "step": 766464
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7442371179070774e-05,
      "loss": 4.0261,
      "step": 766976
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7433985231560254e-05,
      "loss": 4.0311,
      "step": 767488
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7425599284049734e-05,
      "loss": 4.0443,
      "step": 768000
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7417213336539214e-05,
      "loss": 4.0284,
      "step": 768512
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7408827389028694e-05,
      "loss": 4.0329,
      "step": 769024
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7400441441518174e-05,
      "loss": 4.0386,
      "step": 769536
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7392055494007654e-05,
      "loss": 4.0182,
      "step": 770048
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.738366954649713e-05,
      "loss": 4.0289,
      "step": 770560
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.737528359898661e-05,
      "loss": 4.0177,
      "step": 771072
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.736689765147609e-05,
      "loss": 4.0306,
      "step": 771584
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7358511703965574e-05,
      "loss": 4.0343,
      "step": 772096
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7350125756455054e-05,
      "loss": 4.0204,
      "step": 772608
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.734175618774826e-05,
      "loss": 4.0405,
      "step": 773120
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.733337024023774e-05,
      "loss": 4.0443,
      "step": 773632
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.732498429272722e-05,
      "loss": 4.0327,
      "step": 774144
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.73165983452167e-05,
      "loss": 4.0369,
      "step": 774656
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.730821239770618e-05,
      "loss": 4.0365,
      "step": 775168
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.729982645019566e-05,
      "loss": 4.0358,
      "step": 775680
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.729144050268514e-05,
      "loss": 4.028,
      "step": 776192
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.728305455517462e-05,
      "loss": 4.0293,
      "step": 776704
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.727468498646783e-05,
      "loss": 4.0335,
      "step": 777216
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.726631541776104e-05,
      "loss": 4.0246,
      "step": 777728
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.725792947025053e-05,
      "loss": 4.0288,
      "step": 778240
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.724954352274001e-05,
      "loss": 4.034,
      "step": 778752
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.724115757522949e-05,
      "loss": 4.0338,
      "step": 779264
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.723277162771897e-05,
      "loss": 4.0427,
      "step": 779776
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.722440205901218e-05,
      "loss": 4.0319,
      "step": 780288
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7216016111501657e-05,
      "loss": 4.0266,
      "step": 780800
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7207630163991137e-05,
      "loss": 4.0304,
      "step": 781312
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7199260595284346e-05,
      "loss": 4.0337,
      "step": 781824
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7190874647773826e-05,
      "loss": 4.0189,
      "step": 782336
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7182488700263306e-05,
      "loss": 4.0194,
      "step": 782848
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7174102752752785e-05,
      "loss": 4.0239,
      "step": 783360
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7165716805242265e-05,
      "loss": 4.0236,
      "step": 783872
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7157330857731745e-05,
      "loss": 4.0312,
      "step": 784384
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7148944910221225e-05,
      "loss": 4.0258,
      "step": 784896
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7140558962710705e-05,
      "loss": 4.0306,
      "step": 785408
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.713220577280765e-05,
      "loss": 4.026,
      "step": 785920
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.712381982529713e-05,
      "loss": 4.0337,
      "step": 786432
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.711543387778661e-05,
      "loss": 4.0255,
      "step": 786944
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.710704793027609e-05,
      "loss": 4.0358,
      "step": 787456
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.709866198276557e-05,
      "loss": 4.0041,
      "step": 787968
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.709027603525505e-05,
      "loss": 4.0325,
      "step": 788480
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.708189008774453e-05,
      "loss": 4.0231,
      "step": 788992
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.707350414023401e-05,
      "loss": 4.0387,
      "step": 789504
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.706511819272349e-05,
      "loss": 4.0215,
      "step": 790016
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.705673224521296e-05,
      "loss": 4.0212,
      "step": 790528
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.704834629770244e-05,
      "loss": 4.02,
      "step": 791040
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.703996035019193e-05,
      "loss": 4.025,
      "step": 791552
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7031590781485146e-05,
      "loss": 4.0301,
      "step": 792064
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.702320483397462e-05,
      "loss": 4.0113,
      "step": 792576
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.70148188864641e-05,
      "loss": 4.0359,
      "step": 793088
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.700643293895358e-05,
      "loss": 4.0219,
      "step": 793600
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.699804699144306e-05,
      "loss": 4.0018,
      "step": 794112
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.698967742273627e-05,
      "loss": 4.035,
      "step": 794624
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.698129147522575e-05,
      "loss": 4.0141,
      "step": 795136
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.6972921906518964e-05,
      "loss": 4.0048,
      "step": 795648
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.696453595900844e-05,
      "loss": 4.0274,
      "step": 796160
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.695615001149792e-05,
      "loss": 4.0228,
      "step": 796672
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.69477640639874e-05,
      "loss": 4.0111,
      "step": 797184
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.6939378116476884e-05,
      "loss": 4.0318,
      "step": 797696
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.6930992168966363e-05,
      "loss": 4.0035,
      "step": 798208
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.6922606221455843e-05,
      "loss": 4.0131,
      "step": 798720
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.691422027394532e-05,
      "loss": 4.0263,
      "step": 799232
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.690585070523853e-05,
      "loss": 4.0246,
      "step": 799744
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.689746475772801e-05,
      "loss": 4.0315,
      "step": 800256
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.688907881021749e-05,
      "loss": 4.023,
      "step": 800768
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.688069286270697e-05,
      "loss": 4.0342,
      "step": 801280
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.687232329400018e-05,
      "loss": 4.014,
      "step": 801792
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.686393734648966e-05,
      "loss": 4.0299,
      "step": 802304
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.685555139897914e-05,
      "loss": 4.0134,
      "step": 802816
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.684716545146862e-05,
      "loss": 4.018,
      "step": 803328
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.68387795039581e-05,
      "loss": 4.0216,
      "step": 803840
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.683040993525132e-05,
      "loss": 4.0203,
      "step": 804352
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.68220239877408e-05,
      "loss": 4.0296,
      "step": 804864
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.681363804023028e-05,
      "loss": 4.0327,
      "step": 805376
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.6805268471523486e-05,
      "loss": 4.015,
      "step": 805888
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.6796882524012966e-05,
      "loss": 4.0136,
      "step": 806400
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.6788496576502446e-05,
      "loss": 4.0222,
      "step": 806912
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.6780110628991926e-05,
      "loss": 4.0286,
      "step": 807424
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.6771724681481406e-05,
      "loss": 4.0203,
      "step": 807936
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.6763338733970886e-05,
      "loss": 4.0312,
      "step": 808448
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.6754969165264095e-05,
      "loss": 4.0254,
      "step": 808960
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6746583217753575e-05,
      "loss": 4.0258,
      "step": 809472
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6738197270243055e-05,
      "loss": 4.0192,
      "step": 809984
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6729811322732535e-05,
      "loss": 4.0145,
      "step": 810496
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.672142537522202e-05,
      "loss": 4.0275,
      "step": 811008
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.67130394277115e-05,
      "loss": 4.0246,
      "step": 811520
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.670465348020098e-05,
      "loss": 4.0235,
      "step": 812032
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6696267532690455e-05,
      "loss": 4.0178,
      "step": 812544
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.66879143427874e-05,
      "loss": 4.0065,
      "step": 813056
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.667952839527688e-05,
      "loss": 4.0213,
      "step": 813568
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.667114244776636e-05,
      "loss": 4.0322,
      "step": 814080
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.666275650025584e-05,
      "loss": 4.0198,
      "step": 814592
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.665437055274532e-05,
      "loss": 4.0184,
      "step": 815104
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.66459846052348e-05,
      "loss": 4.0164,
      "step": 815616
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.663759865772427e-05,
      "loss": 4.0113,
      "step": 816128
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.662921271021376e-05,
      "loss": 4.021,
      "step": 816640
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6620843141506975e-05,
      "loss": 4.0088,
      "step": 817152
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6612457193996455e-05,
      "loss": 4.0127,
      "step": 817664
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6604087625289664e-05,
      "loss": 4.0201,
      "step": 818176
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6595701677779144e-05,
      "loss": 4.0224,
      "step": 818688
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6587315730268624e-05,
      "loss": 4.017,
      "step": 819200
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.65789297827581e-05,
      "loss": 4.0107,
      "step": 819712
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.657054383524758e-05,
      "loss": 4.0132,
      "step": 820224
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.656215788773706e-05,
      "loss": 4.0239,
      "step": 820736
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.655377194022654e-05,
      "loss": 4.0161,
      "step": 821248
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.654538599271602e-05,
      "loss": 4.0318,
      "step": 821760
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6537016424009226e-05,
      "loss": 4.0237,
      "step": 822272
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.652864685530244e-05,
      "loss": 4.0175,
      "step": 822784
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.652026090779192e-05,
      "loss": 4.0173,
      "step": 823296
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.65118749602814e-05,
      "loss": 4.0117,
      "step": 823808
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.650350539157462e-05,
      "loss": 4.0197,
      "step": 824320
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.64951194440641e-05,
      "loss": 4.0253,
      "step": 824832
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.648673349655357e-05,
      "loss": 4.0094,
      "step": 825344
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.647834754904305e-05,
      "loss": 4.0083,
      "step": 825856
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.646996160153253e-05,
      "loss": 4.018,
      "step": 826368
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.646157565402201e-05,
      "loss": 4.0185,
      "step": 826880
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.645318970651149e-05,
      "loss": 4.0142,
      "step": 827392
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.644480375900097e-05,
      "loss": 4.0039,
      "step": 827904
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.643641781149045e-05,
      "loss": 4.017,
      "step": 828416
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.642803186397993e-05,
      "loss": 4.0065,
      "step": 828928
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.641964591646941e-05,
      "loss": 4.029,
      "step": 829440
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.641125996895889e-05,
      "loss": 4.0184,
      "step": 829952
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6402890400252107e-05,
      "loss": 4.0189,
      "step": 830464
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6394520831545316e-05,
      "loss": 4.0069,
      "step": 830976
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6386134884034796e-05,
      "loss": 4.0275,
      "step": 831488
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6377748936524276e-05,
      "loss": 4.0179,
      "step": 832000
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6369362989013756e-05,
      "loss": 4.0265,
      "step": 832512
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6360977041503236e-05,
      "loss": 4.0267,
      "step": 833024
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6352607472796445e-05,
      "loss": 4.0205,
      "step": 833536
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6344221525285925e-05,
      "loss": 4.0154,
      "step": 834048
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6335835577775405e-05,
      "loss": 4.03,
      "step": 834560
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6327449630264884e-05,
      "loss": 4.004,
      "step": 835072
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6319063682754364e-05,
      "loss": 4.0202,
      "step": 835584
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6310677735243844e-05,
      "loss": 4.0121,
      "step": 836096
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.630229178773333e-05,
      "loss": 4.0152,
      "step": 836608
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.629390584022281e-05,
      "loss": 4.0088,
      "step": 837120
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.628553627151602e-05,
      "loss": 4.0087,
      "step": 837632
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.62771503240055e-05,
      "loss": 4.0098,
      "step": 838144
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.626876437649498e-05,
      "loss": 4.0218,
      "step": 838656
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.626037842898446e-05,
      "loss": 4.0214,
      "step": 839168
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.052252769470215,
      "eval_runtime": 295.0368,
      "eval_samples_per_second": 1293.368,
      "eval_steps_per_second": 40.419,
      "step": 839520
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.625199248147393e-05,
      "loss": 4.0101,
      "step": 839680
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.624360653396341e-05,
      "loss": 4.0063,
      "step": 840192
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.623522058645289e-05,
      "loss": 4.0061,
      "step": 840704
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.622683463894237e-05,
      "loss": 4.0213,
      "step": 841216
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.621844869143185e-05,
      "loss": 4.018,
      "step": 841728
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.62100955015288e-05,
      "loss": 4.0095,
      "step": 842240
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.6201709554018285e-05,
      "loss": 4.0121,
      "step": 842752
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.619332360650776e-05,
      "loss": 4.0093,
      "step": 843264
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.618493765899724e-05,
      "loss": 4.0094,
      "step": 843776
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.6176568090290454e-05,
      "loss": 4.0192,
      "step": 844288
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.6168182142779934e-05,
      "loss": 4.0152,
      "step": 844800
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.615979619526941e-05,
      "loss": 4.0109,
      "step": 845312
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.615141024775889e-05,
      "loss": 4.0205,
      "step": 845824
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.614302430024837e-05,
      "loss": 3.9956,
      "step": 846336
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.613463835273785e-05,
      "loss": 4.0113,
      "step": 846848
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.612625240522733e-05,
      "loss": 3.9945,
      "step": 847360
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.611786645771681e-05,
      "loss": 4.0159,
      "step": 847872
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.610948051020629e-05,
      "loss": 4.0185,
      "step": 848384
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.610109456269577e-05,
      "loss": 3.9966,
      "step": 848896
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.609272499398898e-05,
      "loss": 4.0196,
      "step": 849408
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.608433904647846e-05,
      "loss": 4.0252,
      "step": 849920
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.607595309896794e-05,
      "loss": 4.0134,
      "step": 850432
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.606756715145742e-05,
      "loss": 4.019,
      "step": 850944
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.605919758275063e-05,
      "loss": 4.0177,
      "step": 851456
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.605081163524011e-05,
      "loss": 4.0158,
      "step": 851968
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.604242568772959e-05,
      "loss": 4.0105,
      "step": 852480
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.603403974021907e-05,
      "loss": 4.0059,
      "step": 852992
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.602565379270855e-05,
      "loss": 4.014,
      "step": 853504
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.601726784519803e-05,
      "loss": 4.0076,
      "step": 854016
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.600888189768751e-05,
      "loss": 4.0113,
      "step": 854528
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.600049595017699e-05,
      "loss": 4.0146,
      "step": 855040
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.599212638147021e-05,
      "loss": 4.0208,
      "step": 855552
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5983756812763416e-05,
      "loss": 4.0139,
      "step": 856064
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5975370865252896e-05,
      "loss": 4.0186,
      "step": 856576
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5966984917742376e-05,
      "loss": 4.004,
      "step": 857088
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5958598970231856e-05,
      "loss": 4.0133,
      "step": 857600
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5950229401525065e-05,
      "loss": 4.0168,
      "step": 858112
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5941843454014545e-05,
      "loss": 3.9954,
      "step": 858624
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5933457506504025e-05,
      "loss": 4.0046,
      "step": 859136
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5925087937797234e-05,
      "loss": 4.0056,
      "step": 859648
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5916701990286714e-05,
      "loss": 4.0042,
      "step": 860160
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5908316042776194e-05,
      "loss": 4.0105,
      "step": 860672
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5899930095265674e-05,
      "loss": 4.0096,
      "step": 861184
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5891544147755154e-05,
      "loss": 4.0088,
      "step": 861696
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.588315820024464e-05,
      "loss": 4.0114,
      "step": 862208
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.587477225273412e-05,
      "loss": 4.0128,
      "step": 862720
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5866386305223594e-05,
      "loss": 4.0062,
      "step": 863232
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5858000357713074e-05,
      "loss": 4.0164,
      "step": 863744
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5849614410202554e-05,
      "loss": 3.9906,
      "step": 864256
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5841228462692034e-05,
      "loss": 4.013,
      "step": 864768
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5832842515181514e-05,
      "loss": 4.0019,
      "step": 865280
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.582447294647472e-05,
      "loss": 4.0175,
      "step": 865792
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.58160869989642e-05,
      "loss": 4.005,
      "step": 866304
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.580770105145368e-05,
      "loss": 3.9985,
      "step": 866816
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.579931510394316e-05,
      "loss": 4.003,
      "step": 867328
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.579094553523638e-05,
      "loss": 4.006,
      "step": 867840
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.578255958772586e-05,
      "loss": 4.0121,
      "step": 868352
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.577419001901907e-05,
      "loss": 3.993,
      "step": 868864
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.576580407150855e-05,
      "loss": 4.0169,
      "step": 869376
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.575741812399803e-05,
      "loss": 4.0075,
      "step": 869888
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.574903217648751e-05,
      "loss": 3.9814,
      "step": 870400
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.574064622897699e-05,
      "loss": 4.0154,
      "step": 870912
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.573226028146647e-05,
      "loss": 3.9949,
      "step": 871424
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.572387433395595e-05,
      "loss": 3.9882,
      "step": 871936
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.571548838644543e-05,
      "loss": 4.0035,
      "step": 872448
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.570710243893491e-05,
      "loss": 4.0069,
      "step": 872960
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5698732870228116e-05,
      "loss": 3.995,
      "step": 873472
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5690346922717596e-05,
      "loss": 4.0146,
      "step": 873984
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5681960975207076e-05,
      "loss": 3.9827,
      "step": 874496
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.567357502769656e-05,
      "loss": 3.9939,
      "step": 875008
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.566520545898977e-05,
      "loss": 4.0103,
      "step": 875520
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.565681951147925e-05,
      "loss": 4.0101,
      "step": 876032
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.564843356396873e-05,
      "loss": 4.0103,
      "step": 876544
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.564004761645821e-05,
      "loss": 4.007,
      "step": 877056
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.563167804775142e-05,
      "loss": 4.0129,
      "step": 877568
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.56232921002409e-05,
      "loss": 3.9998,
      "step": 878080
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.561492253153411e-05,
      "loss": 4.0127,
      "step": 878592
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.560653658402359e-05,
      "loss": 3.9967,
      "step": 879104
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.559815063651307e-05,
      "loss": 3.998,
      "step": 879616
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.558976468900255e-05,
      "loss": 3.9993,
      "step": 880128
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.558137874149203e-05,
      "loss": 4.0046,
      "step": 880640
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5572992793981517e-05,
      "loss": 4.0117,
      "step": 881152
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5564606846470997e-05,
      "loss": 4.0148,
      "step": 881664
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5556237277764206e-05,
      "loss": 4.0002,
      "step": 882176
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5547851330253686e-05,
      "loss": 3.9924,
      "step": 882688
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5539481761546895e-05,
      "loss": 4.0046,
      "step": 883200
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5531095814036375e-05,
      "loss": 4.0125,
      "step": 883712
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5522709866525855e-05,
      "loss": 4.0004,
      "step": 884224
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5514323919015335e-05,
      "loss": 4.0118,
      "step": 884736
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.5505954350308544e-05,
      "loss": 4.0101,
      "step": 885248
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5497568402798024e-05,
      "loss": 4.0081,
      "step": 885760
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5489182455287504e-05,
      "loss": 4.002,
      "step": 886272
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.548081288658072e-05,
      "loss": 3.997,
      "step": 886784
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.54724269390702e-05,
      "loss": 4.0125,
      "step": 887296
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.546404099155968e-05,
      "loss": 4.0103,
      "step": 887808
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.545565504404916e-05,
      "loss": 4.0036,
      "step": 888320
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.544726909653864e-05,
      "loss": 4.0023,
      "step": 888832
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.543888314902812e-05,
      "loss": 3.9883,
      "step": 889344
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.54304972015176e-05,
      "loss": 4.0024,
      "step": 889856
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.542211125400708e-05,
      "loss": 4.0139,
      "step": 890368
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.541372530649655e-05,
      "loss": 4.0021,
      "step": 890880
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.540533935898603e-05,
      "loss": 3.999,
      "step": 891392
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.539695341147551e-05,
      "loss": 3.9986,
      "step": 891904
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.538856746396499e-05,
      "loss": 3.9985,
      "step": 892416
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.538019789525821e-05,
      "loss": 3.9998,
      "step": 892928
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.537181194774769e-05,
      "loss": 3.9975,
      "step": 893440
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5363442379040904e-05,
      "loss": 3.9938,
      "step": 893952
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.535505643153038e-05,
      "loss": 4.0005,
      "step": 894464
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.534667048401986e-05,
      "loss": 4.0038,
      "step": 894976
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.533828453650934e-05,
      "loss": 3.9992,
      "step": 895488
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.532989858899882e-05,
      "loss": 3.9989,
      "step": 896000
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.53215126414883e-05,
      "loss": 3.9936,
      "step": 896512
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.531312669397778e-05,
      "loss": 4.0054,
      "step": 897024
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5304757125270986e-05,
      "loss": 3.999,
      "step": 897536
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5296371177760466e-05,
      "loss": 4.0105,
      "step": 898048
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5287985230249946e-05,
      "loss": 4.006,
      "step": 898560
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5279599282739426e-05,
      "loss": 3.9996,
      "step": 899072
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.527124609283637e-05,
      "loss": 4.0039,
      "step": 899584
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.526286014532585e-05,
      "loss": 3.9971,
      "step": 900096
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.525447419781533e-05,
      "loss": 3.9999,
      "step": 900608
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.524608825030481e-05,
      "loss": 4.0055,
      "step": 901120
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.523770230279429e-05,
      "loss": 3.9931,
      "step": 901632
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.522931635528377e-05,
      "loss": 3.9917,
      "step": 902144
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.522093040777325e-05,
      "loss": 4.002,
      "step": 902656
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.521254446026273e-05,
      "loss": 3.9999,
      "step": 903168
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.520417489155594e-05,
      "loss": 3.9952,
      "step": 903680
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.519578894404542e-05,
      "loss": 3.9883,
      "step": 904192
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.518741937533863e-05,
      "loss": 3.9969,
      "step": 904704
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.517903342782811e-05,
      "loss": 3.9927,
      "step": 905216
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5170647480317595e-05,
      "loss": 4.0124,
      "step": 905728
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5162261532807075e-05,
      "loss": 3.998,
      "step": 906240
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5153875585296555e-05,
      "loss": 4.0054,
      "step": 906752
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5145489637786035e-05,
      "loss": 3.9913,
      "step": 907264
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5137103690275515e-05,
      "loss": 4.0047,
      "step": 907776
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5128717742764995e-05,
      "loss": 4.0044,
      "step": 908288
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5120331795254475e-05,
      "loss": 4.0057,
      "step": 908800
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5111945847743955e-05,
      "loss": 4.0095,
      "step": 909312
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5103559900233435e-05,
      "loss": 4.0028,
      "step": 909824
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.509517395272291e-05,
      "loss": 4.002,
      "step": 910336
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.5086804384016124e-05,
      "loss": 4.0072,
      "step": 910848
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.507843481530933e-05,
      "loss": 3.9921,
      "step": 911360
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.507004886779881e-05,
      "loss": 3.9999,
      "step": 911872
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.506166292028829e-05,
      "loss": 3.9967,
      "step": 912384
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.505327697277778e-05,
      "loss": 3.9972,
      "step": 912896
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.504489102526726e-05,
      "loss": 3.99,
      "step": 913408
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.503652145656047e-05,
      "loss": 3.9958,
      "step": 913920
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.502813550904995e-05,
      "loss": 3.9928,
      "step": 914432
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.501974956153943e-05,
      "loss": 3.9976,
      "step": 914944
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.501136361402891e-05,
      "loss": 4.0052,
      "step": 915456
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.041686058044434,
      "eval_runtime": 295.6276,
      "eval_samples_per_second": 1290.783,
      "eval_steps_per_second": 40.338,
      "step": 915840
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.500297766651838e-05,
      "loss": 4.006,
      "step": 915968
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.499459171900786e-05,
      "loss": 3.9874,
      "step": 916480
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.498620577149734e-05,
      "loss": 3.9897,
      "step": 916992
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.497781982398682e-05,
      "loss": 4.0031,
      "step": 917504
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.49694338764763e-05,
      "loss": 3.9972,
      "step": 918016
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.496104792896578e-05,
      "loss": 3.995,
      "step": 918528
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.495266198145526e-05,
      "loss": 4.0009,
      "step": 919040
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.494427603394475e-05,
      "loss": 3.9905,
      "step": 919552
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.493589008643423e-05,
      "loss": 3.9925,
      "step": 920064
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.492750413892371e-05,
      "loss": 3.9999,
      "step": 920576
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.491911819141319e-05,
      "loss": 4.003,
      "step": 921088
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.491073224390267e-05,
      "loss": 3.9927,
      "step": 921600
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.490234629639215e-05,
      "loss": 4.0051,
      "step": 922112
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.489396034888163e-05,
      "loss": 3.9759,
      "step": 922624
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.48855744013711e-05,
      "loss": 3.9965,
      "step": 923136
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.487718845386058e-05,
      "loss": 3.9802,
      "step": 923648
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.486880250635006e-05,
      "loss": 3.9963,
      "step": 924160
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.486041655883954e-05,
      "loss": 3.9997,
      "step": 924672
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.485203061132902e-05,
      "loss": 3.983,
      "step": 925184
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.48436446638185e-05,
      "loss": 4.0002,
      "step": 925696
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.483527509511172e-05,
      "loss": 4.0123,
      "step": 926208
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.48268891476012e-05,
      "loss": 3.9949,
      "step": 926720
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.481850320009068e-05,
      "loss": 4.002,
      "step": 927232
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.481011725258016e-05,
      "loss": 4.0028,
      "step": 927744
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.480173130506964e-05,
      "loss": 3.9994,
      "step": 928256
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.4793361736362846e-05,
      "loss": 3.9957,
      "step": 928768
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.4784992167656055e-05,
      "loss": 3.9862,
      "step": 929280
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.4776606220145535e-05,
      "loss": 3.997,
      "step": 929792
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.4768220272635015e-05,
      "loss": 3.9903,
      "step": 930304
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.4759834325124495e-05,
      "loss": 3.9919,
      "step": 930816
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4751448377613975e-05,
      "loss": 4.0015,
      "step": 931328
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4743078808907184e-05,
      "loss": 3.9998,
      "step": 931840
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4734692861396664e-05,
      "loss": 3.9963,
      "step": 932352
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.472630691388615e-05,
      "loss": 4.0063,
      "step": 932864
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.471792096637563e-05,
      "loss": 3.9895,
      "step": 933376
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.470953501886511e-05,
      "loss": 3.9969,
      "step": 933888
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.470114907135459e-05,
      "loss": 3.9954,
      "step": 934400
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.469276312384407e-05,
      "loss": 3.9797,
      "step": 934912
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.468437717633355e-05,
      "loss": 3.9892,
      "step": 935424
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.467600760762676e-05,
      "loss": 3.9885,
      "step": 935936
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.466762166011624e-05,
      "loss": 3.9842,
      "step": 936448
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.465923571260572e-05,
      "loss": 3.9943,
      "step": 936960
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.46508497650952e-05,
      "loss": 3.9956,
      "step": 937472
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.464248019638841e-05,
      "loss": 3.9924,
      "step": 937984
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.463409424887789e-05,
      "loss": 3.9957,
      "step": 938496
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.462570830136737e-05,
      "loss": 3.9994,
      "step": 939008
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4617338732660584e-05,
      "loss": 3.9861,
      "step": 939520
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4608952785150064e-05,
      "loss": 4.0001,
      "step": 940032
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4600566837639544e-05,
      "loss": 3.977,
      "step": 940544
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4592180890129024e-05,
      "loss": 3.9947,
      "step": 941056
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4583794942618504e-05,
      "loss": 3.9866,
      "step": 941568
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4575408995107984e-05,
      "loss": 4.0027,
      "step": 942080
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.456702304759746e-05,
      "loss": 3.9862,
      "step": 942592
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.455863710008694e-05,
      "loss": 3.9827,
      "step": 943104
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.455026753138015e-05,
      "loss": 3.9887,
      "step": 943616
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.454188158386963e-05,
      "loss": 3.9893,
      "step": 944128
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4533495636359106e-05,
      "loss": 3.9985,
      "step": 944640
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.452512606765232e-05,
      "loss": 3.9715,
      "step": 945152
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.45167401201418e-05,
      "loss": 4.0028,
      "step": 945664
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.450835417263129e-05,
      "loss": 3.991,
      "step": 946176
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.449996822512076e-05,
      "loss": 3.9684,
      "step": 946688
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.449158227761024e-05,
      "loss": 3.9994,
      "step": 947200
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.448319633009972e-05,
      "loss": 3.9784,
      "step": 947712
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.44748103825892e-05,
      "loss": 3.973,
      "step": 948224
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.446642443507868e-05,
      "loss": 3.9842,
      "step": 948736
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.445805486637189e-05,
      "loss": 3.9927,
      "step": 949248
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4449685297665107e-05,
      "loss": 3.9848,
      "step": 949760
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.444129935015458e-05,
      "loss": 3.9913,
      "step": 950272
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.443291340264406e-05,
      "loss": 3.9727,
      "step": 950784
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.442452745513354e-05,
      "loss": 3.9764,
      "step": 951296
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4416141507623026e-05,
      "loss": 3.9938,
      "step": 951808
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4407755560112506e-05,
      "loss": 3.9959,
      "step": 952320
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4399369612601986e-05,
      "loss": 3.9935,
      "step": 952832
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4391000043895195e-05,
      "loss": 3.9909,
      "step": 953344
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4382614096384675e-05,
      "loss": 3.9972,
      "step": 953856
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4374228148874155e-05,
      "loss": 3.9838,
      "step": 954368
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4365842201363635e-05,
      "loss": 3.9909,
      "step": 954880
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4357456253853115e-05,
      "loss": 3.9836,
      "step": 955392
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4349070306342595e-05,
      "loss": 3.9832,
      "step": 955904
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4340684358832075e-05,
      "loss": 3.9869,
      "step": 956416
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4332298411321555e-05,
      "loss": 3.9803,
      "step": 956928
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4323928842614764e-05,
      "loss": 3.9975,
      "step": 957440
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4315542895104244e-05,
      "loss": 3.9972,
      "step": 957952
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.4307156947593724e-05,
      "loss": 3.9878,
      "step": 958464
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.429877100008321e-05,
      "loss": 3.9708,
      "step": 958976
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.429040143137642e-05,
      "loss": 3.9892,
      "step": 959488
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.42820154838659e-05,
      "loss": 3.9935,
      "step": 960000
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.427362953635538e-05,
      "loss": 3.9872,
      "step": 960512
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.426524358884486e-05,
      "loss": 3.994,
      "step": 961024
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.425687402013807e-05,
      "loss": 3.9976,
      "step": 961536
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.424848807262755e-05,
      "loss": 3.9882,
      "step": 962048
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.424010212511703e-05,
      "loss": 3.9877,
      "step": 962560
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.423171617760651e-05,
      "loss": 3.9825,
      "step": 963072
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.422333023009599e-05,
      "loss": 3.9958,
      "step": 963584
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.421494428258547e-05,
      "loss": 3.9914,
      "step": 964096
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.420655833507495e-05,
      "loss": 3.9911,
      "step": 964608
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.419817238756443e-05,
      "loss": 3.9852,
      "step": 965120
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.4189802818857644e-05,
      "loss": 3.9738,
      "step": 965632
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.4181416871347124e-05,
      "loss": 3.9862,
      "step": 966144
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.41730309238366e-05,
      "loss": 3.998,
      "step": 966656
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.416464497632608e-05,
      "loss": 3.991,
      "step": 967168
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.415625902881556e-05,
      "loss": 3.977,
      "step": 967680
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.4147889460108767e-05,
      "loss": 3.988,
      "step": 968192
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.4139503512598247e-05,
      "loss": 3.9806,
      "step": 968704
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.4131117565087726e-05,
      "loss": 3.9845,
      "step": 969216
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.4122731617577206e-05,
      "loss": 3.9827,
      "step": 969728
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.4114362048870416e-05,
      "loss": 3.9784,
      "step": 970240
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.4105976101359895e-05,
      "loss": 3.9831,
      "step": 970752
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.409759015384938e-05,
      "loss": 3.9869,
      "step": 971264
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.408920420633886e-05,
      "loss": 3.9847,
      "step": 971776
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.408083463763207e-05,
      "loss": 3.9781,
      "step": 972288
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.407244869012155e-05,
      "loss": 3.9785,
      "step": 972800
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.406406274261103e-05,
      "loss": 3.99,
      "step": 973312
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.405567679510051e-05,
      "loss": 3.9855,
      "step": 973824
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.404729084758999e-05,
      "loss": 3.9924,
      "step": 974336
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.40389212788832e-05,
      "loss": 3.9881,
      "step": 974848
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.403053533137268e-05,
      "loss": 3.9897,
      "step": 975360
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.402216576266589e-05,
      "loss": 3.9889,
      "step": 975872
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.401377981515537e-05,
      "loss": 3.9792,
      "step": 976384
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.4005410246448585e-05,
      "loss": 3.9872,
      "step": 976896
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.3997024298938065e-05,
      "loss": 3.9898,
      "step": 977408
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.3988638351427545e-05,
      "loss": 3.9771,
      "step": 977920
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.3980252403917025e-05,
      "loss": 3.9832,
      "step": 978432
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.3971866456406505e-05,
      "loss": 3.9797,
      "step": 978944
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.3963480508895985e-05,
      "loss": 3.9857,
      "step": 979456
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.3955094561385465e-05,
      "loss": 3.9835,
      "step": 979968
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.3946708613874945e-05,
      "loss": 3.9733,
      "step": 980480
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.3938322666364425e-05,
      "loss": 3.9763,
      "step": 980992
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.3929936718853905e-05,
      "loss": 3.9786,
      "step": 981504
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.3921550771343385e-05,
      "loss": 3.9961,
      "step": 982016
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.3913164823832865e-05,
      "loss": 3.9847,
      "step": 982528
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.3904795255126074e-05,
      "loss": 3.986,
      "step": 983040
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.3896409307615554e-05,
      "loss": 3.9767,
      "step": 983552
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.3888023360105034e-05,
      "loss": 3.9884,
      "step": 984064
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.387963741259452e-05,
      "loss": 3.9894,
      "step": 984576
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.3871251465084e-05,
      "loss": 3.9874,
      "step": 985088
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.386286551757348e-05,
      "loss": 3.9926,
      "step": 985600
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.3854479570062953e-05,
      "loss": 3.9842,
      "step": 986112
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.384611000135617e-05,
      "loss": 3.9904,
      "step": 986624
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.383772405384565e-05,
      "loss": 3.9934,
      "step": 987136
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.382933810633513e-05,
      "loss": 3.975,
      "step": 987648
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.38209521588246e-05,
      "loss": 3.986,
      "step": 988160
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.381256621131408e-05,
      "loss": 3.9806,
      "step": 988672
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.380418026380356e-05,
      "loss": 3.9835,
      "step": 989184
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.379581069509678e-05,
      "loss": 3.9762,
      "step": 989696
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.378742474758626e-05,
      "loss": 3.9795,
      "step": 990208
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.377903880007574e-05,
      "loss": 3.9764,
      "step": 990720
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.377065285256522e-05,
      "loss": 3.9828,
      "step": 991232
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.37622669050547e-05,
      "loss": 3.988,
      "step": 991744
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.032730579376221,
      "eval_runtime": 292.5381,
      "eval_samples_per_second": 1304.415,
      "eval_steps_per_second": 40.764,
      "step": 992160
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.375388095754418e-05,
      "loss": 3.985,
      "step": 992256
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.374549501003366e-05,
      "loss": 3.9715,
      "step": 992768
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.373710906252314e-05,
      "loss": 3.975,
      "step": 993280
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.372877225142381e-05,
      "loss": 3.9867,
      "step": 993792
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.372038630391329e-05,
      "loss": 3.98,
      "step": 994304
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.371200035640277e-05,
      "loss": 3.9819,
      "step": 994816
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.370361440889225e-05,
      "loss": 3.9822,
      "step": 995328
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.3695228461381725e-05,
      "loss": 3.9748,
      "step": 995840
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.368684251387121e-05,
      "loss": 3.976,
      "step": 996352
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.367847294516443e-05,
      "loss": 3.9874,
      "step": 996864
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.36700869976539e-05,
      "loss": 3.9835,
      "step": 997376
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.366170105014338e-05,
      "loss": 3.9803,
      "step": 997888
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.365331510263286e-05,
      "loss": 3.993,
      "step": 998400
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.364492915512234e-05,
      "loss": 3.9588,
      "step": 998912
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.363654320761182e-05,
      "loss": 3.9798,
      "step": 999424
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.36281572601013e-05,
      "loss": 3.9645,
      "step": 999936
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.361977131259078e-05,
      "loss": 3.9847,
      "step": 1000448
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.361138536508026e-05,
      "loss": 3.9822,
      "step": 1000960
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.360299941756974e-05,
      "loss": 3.9721,
      "step": 1001472
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.359461347005922e-05,
      "loss": 3.9772,
      "step": 1001984
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.358624390135243e-05,
      "loss": 4.0013,
      "step": 1002496
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.357785795384191e-05,
      "loss": 3.9816,
      "step": 1003008
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.356947200633139e-05,
      "loss": 3.9862,
      "step": 1003520
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.3561086058820876e-05,
      "loss": 3.9886,
      "step": 1004032
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.3552700111310356e-05,
      "loss": 3.9876,
      "step": 1004544
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.3544330542603565e-05,
      "loss": 3.9771,
      "step": 1005056
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.3535944595093045e-05,
      "loss": 3.9738,
      "step": 1005568
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.3527558647582525e-05,
      "loss": 3.9806,
      "step": 1006080
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.3519172700072005e-05,
      "loss": 3.977,
      "step": 1006592
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.3510803131365214e-05,
      "loss": 3.978,
      "step": 1007104
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3502417183854694e-05,
      "loss": 3.9863,
      "step": 1007616
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3494031236344174e-05,
      "loss": 3.9874,
      "step": 1008128
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3485645288833654e-05,
      "loss": 3.9829,
      "step": 1008640
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.347727572012686e-05,
      "loss": 3.9887,
      "step": 1009152
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.346888977261634e-05,
      "loss": 3.9752,
      "step": 1009664
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.346052020390956e-05,
      "loss": 3.9837,
      "step": 1010176
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.345213425639904e-05,
      "loss": 3.9822,
      "step": 1010688
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.344374830888852e-05,
      "loss": 3.9624,
      "step": 1011200
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3435362361378e-05,
      "loss": 3.9795,
      "step": 1011712
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.342697641386748e-05,
      "loss": 3.9703,
      "step": 1012224
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.341859046635696e-05,
      "loss": 3.9697,
      "step": 1012736
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.341020451884644e-05,
      "loss": 3.9803,
      "step": 1013248
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.340181857133591e-05,
      "loss": 3.9869,
      "step": 1013760
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.339344900262913e-05,
      "loss": 3.9705,
      "step": 1014272
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.338506305511861e-05,
      "loss": 3.9799,
      "step": 1014784
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.337667710760808e-05,
      "loss": 3.9821,
      "step": 1015296
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.336829116009757e-05,
      "loss": 3.9721,
      "step": 1015808
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3359921591390784e-05,
      "loss": 3.9826,
      "step": 1016320
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.335155202268399e-05,
      "loss": 3.9635,
      "step": 1016832
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.334316607517347e-05,
      "loss": 3.9772,
      "step": 1017344
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.333478012766295e-05,
      "loss": 3.975,
      "step": 1017856
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.332639418015243e-05,
      "loss": 3.9871,
      "step": 1018368
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.331800823264191e-05,
      "loss": 3.9742,
      "step": 1018880
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3309622285131386e-05,
      "loss": 3.967,
      "step": 1019392
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3301236337620866e-05,
      "loss": 3.9716,
      "step": 1019904
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.329286676891408e-05,
      "loss": 3.9731,
      "step": 1020416
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3284480821403555e-05,
      "loss": 3.9865,
      "step": 1020928
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3276094873893035e-05,
      "loss": 3.9579,
      "step": 1021440
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.326770892638252e-05,
      "loss": 3.9851,
      "step": 1021952
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3259322978872e-05,
      "loss": 3.9747,
      "step": 1022464
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.325093703136148e-05,
      "loss": 3.9584,
      "step": 1022976
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.324255108385096e-05,
      "loss": 3.9791,
      "step": 1023488
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.323416513634044e-05,
      "loss": 3.966,
      "step": 1024000
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.322579556763365e-05,
      "loss": 3.96,
      "step": 1024512
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.321740962012313e-05,
      "loss": 3.9663,
      "step": 1025024
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.320902367261261e-05,
      "loss": 3.9775,
      "step": 1025536
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.320063772510209e-05,
      "loss": 3.9734,
      "step": 1026048
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.319228453519903e-05,
      "loss": 3.9689,
      "step": 1026560
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.318389858768851e-05,
      "loss": 3.9647,
      "step": 1027072
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.317551264017799e-05,
      "loss": 3.9617,
      "step": 1027584
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3167126692667475e-05,
      "loss": 3.9748,
      "step": 1028096
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3158740745156955e-05,
      "loss": 3.9808,
      "step": 1028608
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3150371176450164e-05,
      "loss": 3.9775,
      "step": 1029120
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3141985228939644e-05,
      "loss": 3.9831,
      "step": 1029632
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3133599281429124e-05,
      "loss": 3.9784,
      "step": 1030144
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3125213333918604e-05,
      "loss": 3.9688,
      "step": 1030656
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3116827386408084e-05,
      "loss": 3.9748,
      "step": 1031168
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3108441438897564e-05,
      "loss": 3.9708,
      "step": 1031680
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3100055491387044e-05,
      "loss": 3.9732,
      "step": 1032192
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.3091669543876524e-05,
      "loss": 3.9654,
      "step": 1032704
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.308329997516973e-05,
      "loss": 3.9697,
      "step": 1033216
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.307491402765921e-05,
      "loss": 3.9854,
      "step": 1033728
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.306654445895243e-05,
      "loss": 3.9802,
      "step": 1034240
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.305815851144191e-05,
      "loss": 3.9769,
      "step": 1034752
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.304977256393139e-05,
      "loss": 3.9564,
      "step": 1035264
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.304138661642087e-05,
      "loss": 3.9736,
      "step": 1035776
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.303300066891035e-05,
      "loss": 3.9781,
      "step": 1036288
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.302461472139983e-05,
      "loss": 3.9758,
      "step": 1036800
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.301622877388931e-05,
      "loss": 3.9765,
      "step": 1037312
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.300784282637879e-05,
      "loss": 3.9862,
      "step": 1037824
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.299945687886827e-05,
      "loss": 3.9733,
      "step": 1038336
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.299107093135775e-05,
      "loss": 3.9717,
      "step": 1038848
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.298268498384722e-05,
      "loss": 3.9685,
      "step": 1039360
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.297431541514044e-05,
      "loss": 3.9785,
      "step": 1039872
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.296592946762992e-05,
      "loss": 3.9768,
      "step": 1040384
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.29575435201194e-05,
      "loss": 3.9807,
      "step": 1040896
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.294915757260888e-05,
      "loss": 3.9672,
      "step": 1041408
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.294077162509836e-05,
      "loss": 3.9629,
      "step": 1041920
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.293240205639157e-05,
      "loss": 3.9692,
      "step": 1042432
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2924016108881046e-05,
      "loss": 3.9822,
      "step": 1042944
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2915630161370526e-05,
      "loss": 3.9764,
      "step": 1043456
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2907244213860006e-05,
      "loss": 3.9647,
      "step": 1043968
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.289887464515322e-05,
      "loss": 3.9724,
      "step": 1044480
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2890488697642695e-05,
      "loss": 3.964,
      "step": 1044992
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2882102750132175e-05,
      "loss": 3.971,
      "step": 1045504
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2873716802621655e-05,
      "loss": 3.9662,
      "step": 1046016
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2865347233914864e-05,
      "loss": 3.9671,
      "step": 1046528
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2856961286404344e-05,
      "loss": 3.9718,
      "step": 1047040
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.284857533889383e-05,
      "loss": 3.9732,
      "step": 1047552
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.284018939138331e-05,
      "loss": 3.967,
      "step": 1048064
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.283181982267652e-05,
      "loss": 3.9661,
      "step": 1048576
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2823433875166e-05,
      "loss": 3.9668,
      "step": 1049088
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.281504792765548e-05,
      "loss": 3.9761,
      "step": 1049600
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.280666198014496e-05,
      "loss": 3.9725,
      "step": 1050112
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.279827603263444e-05,
      "loss": 3.98,
      "step": 1050624
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.278990646392765e-05,
      "loss": 3.9714,
      "step": 1051136
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2781536895220865e-05,
      "loss": 3.9765,
      "step": 1051648
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.277315094771034e-05,
      "loss": 3.9714,
      "step": 1052160
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.276476500019982e-05,
      "loss": 3.9636,
      "step": 1052672
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.27563790526893e-05,
      "loss": 3.9764,
      "step": 1053184
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2747993105178784e-05,
      "loss": 3.9731,
      "step": 1053696
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2739607157668264e-05,
      "loss": 3.9637,
      "step": 1054208
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2731237588961474e-05,
      "loss": 3.9679,
      "step": 1054720
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2722851641450954e-05,
      "loss": 3.9692,
      "step": 1055232
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2714465693940433e-05,
      "loss": 3.9711,
      "step": 1055744
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.2706079746429913e-05,
      "loss": 3.9629,
      "step": 1056256
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.269769379891939e-05,
      "loss": 3.962,
      "step": 1056768
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.268930785140887e-05,
      "loss": 3.964,
      "step": 1057280
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.268093828270208e-05,
      "loss": 3.9662,
      "step": 1057792
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.267255233519156e-05,
      "loss": 3.9812,
      "step": 1058304
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.266416638768104e-05,
      "loss": 3.9743,
      "step": 1058816
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.265578044017052e-05,
      "loss": 3.9744,
      "step": 1059328
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.264739449266e-05,
      "loss": 3.963,
      "step": 1059840
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.263902492395322e-05,
      "loss": 3.9765,
      "step": 1060352
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.26306389764427e-05,
      "loss": 3.9753,
      "step": 1060864
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.262225302893218e-05,
      "loss": 3.9767,
      "step": 1061376
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.261386708142166e-05,
      "loss": 3.9751,
      "step": 1061888
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.260548113391114e-05,
      "loss": 3.9714,
      "step": 1062400
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.259711156520435e-05,
      "loss": 3.978,
      "step": 1062912
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.258872561769383e-05,
      "loss": 3.9789,
      "step": 1063424
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.258033967018331e-05,
      "loss": 3.9628,
      "step": 1063936
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.257195372267279e-05,
      "loss": 3.9687,
      "step": 1064448
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.256356777516227e-05,
      "loss": 3.9706,
      "step": 1064960
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.255518182765175e-05,
      "loss": 3.9685,
      "step": 1065472
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.254679588014123e-05,
      "loss": 3.962,
      "step": 1065984
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.253840993263071e-05,
      "loss": 3.9681,
      "step": 1066496
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.253004036392392e-05,
      "loss": 3.9593,
      "step": 1067008
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.25216544164134e-05,
      "loss": 3.9767,
      "step": 1067520
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.251326846890288e-05,
      "loss": 3.9675,
      "step": 1068032
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.024904727935791,
      "eval_runtime": 295.9734,
      "eval_samples_per_second": 1289.274,
      "eval_steps_per_second": 40.291,
      "step": 1068480
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2504882521392356e-05,
      "loss": 3.98,
      "step": 1068544
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2496496573881836e-05,
      "loss": 3.9586,
      "step": 1069056
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2488110626371316e-05,
      "loss": 3.9633,
      "step": 1069568
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2479724678860796e-05,
      "loss": 3.9753,
      "step": 1070080
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2471338731350276e-05,
      "loss": 3.9688,
      "step": 1070592
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2462952783839755e-05,
      "loss": 3.9672,
      "step": 1071104
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2454566836329235e-05,
      "loss": 3.9652,
      "step": 1071616
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2446180888818715e-05,
      "loss": 3.962,
      "step": 1072128
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2437794941308195e-05,
      "loss": 3.9648,
      "step": 1072640
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2429408993797675e-05,
      "loss": 3.976,
      "step": 1073152
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2421023046287155e-05,
      "loss": 3.9711,
      "step": 1073664
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2412637098776635e-05,
      "loss": 3.9681,
      "step": 1074176
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2404251151266115e-05,
      "loss": 3.9755,
      "step": 1074688
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2395865203755595e-05,
      "loss": 3.9496,
      "step": 1075200
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2387479256245075e-05,
      "loss": 3.9634,
      "step": 1075712
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2379093308734555e-05,
      "loss": 3.9564,
      "step": 1076224
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2370707361224035e-05,
      "loss": 3.9686,
      "step": 1076736
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2362321413713515e-05,
      "loss": 3.965,
      "step": 1077248
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2353935466202995e-05,
      "loss": 3.9626,
      "step": 1077760
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2345549518692475e-05,
      "loss": 3.9656,
      "step": 1078272
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2337179949985684e-05,
      "loss": 3.9856,
      "step": 1078784
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2328794002475164e-05,
      "loss": 3.97,
      "step": 1079296
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2320408054964644e-05,
      "loss": 3.9699,
      "step": 1079808
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2312022107454124e-05,
      "loss": 3.9732,
      "step": 1080320
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.230365253874734e-05,
      "loss": 3.9791,
      "step": 1080832
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.229526659123682e-05,
      "loss": 3.9586,
      "step": 1081344
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.22868806437263e-05,
      "loss": 3.9617,
      "step": 1081856
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.227849469621578e-05,
      "loss": 3.9722,
      "step": 1082368
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.227012512750899e-05,
      "loss": 3.9579,
      "step": 1082880
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.226173917999847e-05,
      "loss": 3.9654,
      "step": 1083392
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.225335323248795e-05,
      "loss": 3.9723,
      "step": 1083904
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.224498366378116e-05,
      "loss": 3.973,
      "step": 1084416
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.223659771627064e-05,
      "loss": 3.9716,
      "step": 1084928
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.222822814756385e-05,
      "loss": 3.9749,
      "step": 1085440
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.221984220005333e-05,
      "loss": 3.9627,
      "step": 1085952
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.221145625254281e-05,
      "loss": 3.9726,
      "step": 1086464
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.220307030503229e-05,
      "loss": 3.9666,
      "step": 1086976
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.219468435752177e-05,
      "loss": 3.9505,
      "step": 1087488
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.218629841001125e-05,
      "loss": 3.9688,
      "step": 1088000
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.217791246250073e-05,
      "loss": 3.9553,
      "step": 1088512
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.216952651499021e-05,
      "loss": 3.9561,
      "step": 1089024
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.216114056747969e-05,
      "loss": 3.9661,
      "step": 1089536
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.21527709987729e-05,
      "loss": 3.9722,
      "step": 1090048
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.214438505126238e-05,
      "loss": 3.9541,
      "step": 1090560
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.213599910375186e-05,
      "loss": 3.9735,
      "step": 1091072
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.212761315624134e-05,
      "loss": 3.9661,
      "step": 1091584
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.211922720873082e-05,
      "loss": 3.9649,
      "step": 1092096
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.21108412612203e-05,
      "loss": 3.9693,
      "step": 1092608
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.2102455313709775e-05,
      "loss": 3.9491,
      "step": 1093120
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.209406936619926e-05,
      "loss": 3.9635,
      "step": 1093632
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.208569979749248e-05,
      "loss": 3.9625,
      "step": 1094144
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.207731384998196e-05,
      "loss": 3.9691,
      "step": 1094656
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.206892790247143e-05,
      "loss": 3.9648,
      "step": 1095168
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.206054195496091e-05,
      "loss": 3.9504,
      "step": 1095680
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.205217238625413e-05,
      "loss": 3.9585,
      "step": 1096192
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.20437864387436e-05,
      "loss": 3.9592,
      "step": 1096704
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.2035416870036816e-05,
      "loss": 3.9787,
      "step": 1097216
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.2027030922526296e-05,
      "loss": 3.9445,
      "step": 1097728
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.2018644975015776e-05,
      "loss": 3.9709,
      "step": 1098240
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.201025902750525e-05,
      "loss": 3.964,
      "step": 1098752
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.200187307999473e-05,
      "loss": 3.9468,
      "step": 1099264
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.1993487132484216e-05,
      "loss": 3.9631,
      "step": 1099776
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.198511756377743e-05,
      "loss": 3.9586,
      "step": 1100288
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.1976731616266905e-05,
      "loss": 3.9472,
      "step": 1100800
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.1968345668756385e-05,
      "loss": 3.9489,
      "step": 1101312
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.1959959721245865e-05,
      "loss": 3.9708,
      "step": 1101824
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.1951573773735344e-05,
      "loss": 3.9596,
      "step": 1102336
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.1943187826224824e-05,
      "loss": 3.9553,
      "step": 1102848
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.1934818257518034e-05,
      "loss": 3.9484,
      "step": 1103360
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.1926432310007514e-05,
      "loss": 3.9512,
      "step": 1103872
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.1918046362496993e-05,
      "loss": 3.9627,
      "step": 1104384
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.1909660414986473e-05,
      "loss": 3.9679,
      "step": 1104896
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.190127446747595e-05,
      "loss": 3.963,
      "step": 1105408
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.189288851996543e-05,
      "loss": 3.9695,
      "step": 1105920
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.188450257245491e-05,
      "loss": 3.966,
      "step": 1106432
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.187611662494439e-05,
      "loss": 3.9641,
      "step": 1106944
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.186774705623761e-05,
      "loss": 3.959,
      "step": 1107456
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.185936110872709e-05,
      "loss": 3.9543,
      "step": 1107968
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.185097516121657e-05,
      "loss": 3.9576,
      "step": 1108480
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.184258921370605e-05,
      "loss": 3.9582,
      "step": 1108992
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.183420326619553e-05,
      "loss": 3.9569,
      "step": 1109504
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.182581731868501e-05,
      "loss": 3.9675,
      "step": 1110016
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.181744774997822e-05,
      "loss": 3.9734,
      "step": 1110528
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.18090618024677e-05,
      "loss": 3.9627,
      "step": 1111040
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.180067585495718e-05,
      "loss": 3.9443,
      "step": 1111552
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.179228990744666e-05,
      "loss": 3.9617,
      "step": 1112064
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.178390395993614e-05,
      "loss": 3.9659,
      "step": 1112576
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.177551801242562e-05,
      "loss": 3.966,
      "step": 1113088
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.17671320649151e-05,
      "loss": 3.959,
      "step": 1113600
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.1758762496208314e-05,
      "loss": 3.9739,
      "step": 1114112
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1750376548697794e-05,
      "loss": 3.962,
      "step": 1114624
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.174199060118727e-05,
      "loss": 3.9584,
      "step": 1115136
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.173360465367675e-05,
      "loss": 3.9581,
      "step": 1115648
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.172521870616623e-05,
      "loss": 3.9682,
      "step": 1116160
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1716832758655707e-05,
      "loss": 3.9612,
      "step": 1116672
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1708446811145187e-05,
      "loss": 3.9678,
      "step": 1117184
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1700060863634667e-05,
      "loss": 3.9555,
      "step": 1117696
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1691691294927876e-05,
      "loss": 3.9517,
      "step": 1118208
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1683305347417356e-05,
      "loss": 3.9581,
      "step": 1118720
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1674919399906836e-05,
      "loss": 3.9705,
      "step": 1119232
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1666533452396315e-05,
      "loss": 3.9646,
      "step": 1119744
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.165816388368953e-05,
      "loss": 3.953,
      "step": 1120256
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.164977793617901e-05,
      "loss": 3.9581,
      "step": 1120768
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.164139198866849e-05,
      "loss": 3.9516,
      "step": 1121280
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.163300604115797e-05,
      "loss": 3.9631,
      "step": 1121792
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.162462009364745e-05,
      "loss": 3.9534,
      "step": 1122304
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.161625052494066e-05,
      "loss": 3.9523,
      "step": 1122816
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.160786457743014e-05,
      "loss": 3.9574,
      "step": 1123328
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.159947862991962e-05,
      "loss": 3.96,
      "step": 1123840
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.15910926824091e-05,
      "loss": 3.959,
      "step": 1124352
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.158272311370231e-05,
      "loss": 3.9548,
      "step": 1124864
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1574353544995525e-05,
      "loss": 3.951,
      "step": 1125376
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1565967597485005e-05,
      "loss": 3.9614,
      "step": 1125888
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1557581649974485e-05,
      "loss": 3.9621,
      "step": 1126400
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1549195702463965e-05,
      "loss": 3.9592,
      "step": 1126912
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1540809754953445e-05,
      "loss": 3.9636,
      "step": 1127424
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1532423807442925e-05,
      "loss": 3.9648,
      "step": 1127936
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1524054238736134e-05,
      "loss": 3.958,
      "step": 1128448
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1515668291225614e-05,
      "loss": 3.9517,
      "step": 1128960
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1507282343715094e-05,
      "loss": 3.9628,
      "step": 1129472
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1498896396204574e-05,
      "loss": 3.958,
      "step": 1129984
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1490510448694054e-05,
      "loss": 3.9517,
      "step": 1130496
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1482124501183534e-05,
      "loss": 3.9576,
      "step": 1131008
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1473738553673014e-05,
      "loss": 3.9556,
      "step": 1131520
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.146536898496622e-05,
      "loss": 3.962,
      "step": 1132032
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.145698303745571e-05,
      "loss": 3.9471,
      "step": 1132544
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.144859708994519e-05,
      "loss": 3.9559,
      "step": 1133056
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.144021114243467e-05,
      "loss": 3.9528,
      "step": 1133568
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.143182519492415e-05,
      "loss": 3.9528,
      "step": 1134080
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.142343924741362e-05,
      "loss": 3.9627,
      "step": 1134592
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.14150532999031e-05,
      "loss": 3.9652,
      "step": 1135104
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.140668373119632e-05,
      "loss": 3.9568,
      "step": 1135616
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.13982977836858e-05,
      "loss": 3.9507,
      "step": 1136128
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.138991183617527e-05,
      "loss": 3.966,
      "step": 1136640
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.138152588866475e-05,
      "loss": 3.9617,
      "step": 1137152
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.137313994115423e-05,
      "loss": 3.9616,
      "step": 1137664
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.136477037244745e-05,
      "loss": 3.9658,
      "step": 1138176
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.135638442493693e-05,
      "loss": 3.9599,
      "step": 1138688
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.134799847742641e-05,
      "loss": 3.9694,
      "step": 1139200
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.133961252991589e-05,
      "loss": 3.9654,
      "step": 1139712
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1331242961209096e-05,
      "loss": 3.9511,
      "step": 1140224
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1322857013698576e-05,
      "loss": 3.9564,
      "step": 1140736
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1314471066188056e-05,
      "loss": 3.956,
      "step": 1141248
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1306085118677536e-05,
      "loss": 3.957,
      "step": 1141760
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1297699171167016e-05,
      "loss": 3.9534,
      "step": 1142272
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1289313223656496e-05,
      "loss": 3.9563,
      "step": 1142784
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1280927276145976e-05,
      "loss": 3.9496,
      "step": 1143296
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1272541328635456e-05,
      "loss": 3.9648,
      "step": 1143808
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1264171759928665e-05,
      "loss": 3.9565,
      "step": 1144320
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.018359184265137,
      "eval_runtime": 300.9569,
      "eval_samples_per_second": 1267.926,
      "eval_steps_per_second": 39.624,
      "step": 1144800
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.1255785812418145e-05,
      "loss": 3.9572,
      "step": 1144832
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.1247399864907625e-05,
      "loss": 3.9533,
      "step": 1145344
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.123901391739711e-05,
      "loss": 3.947,
      "step": 1145856
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.123064434869032e-05,
      "loss": 3.9624,
      "step": 1146368
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.12222584011798e-05,
      "loss": 3.9577,
      "step": 1146880
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.121387245366928e-05,
      "loss": 3.9573,
      "step": 1147392
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.120548650615876e-05,
      "loss": 3.9514,
      "step": 1147904
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.119711693745197e-05,
      "loss": 3.9472,
      "step": 1148416
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.118873098994145e-05,
      "loss": 3.9578,
      "step": 1148928
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.118034504243093e-05,
      "loss": 3.958,
      "step": 1149440
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.117195909492041e-05,
      "loss": 3.9604,
      "step": 1149952
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.116358952621362e-05,
      "loss": 3.9542,
      "step": 1150464
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.11552035787031e-05,
      "loss": 3.9687,
      "step": 1150976
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.114681763119258e-05,
      "loss": 3.9369,
      "step": 1151488
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.1138431683682065e-05,
      "loss": 3.952,
      "step": 1152000
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.1130062114975274e-05,
      "loss": 3.9446,
      "step": 1152512
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.1121676167464754e-05,
      "loss": 3.9553,
      "step": 1153024
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.1113290219954234e-05,
      "loss": 3.9561,
      "step": 1153536
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.1104904272443714e-05,
      "loss": 3.951,
      "step": 1154048
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.1096534703736923e-05,
      "loss": 3.9527,
      "step": 1154560
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.108816513503013e-05,
      "loss": 3.9752,
      "step": 1155072
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.107977918751961e-05,
      "loss": 3.9581,
      "step": 1155584
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.107139324000909e-05,
      "loss": 3.9554,
      "step": 1156096
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.106300729249857e-05,
      "loss": 3.9645,
      "step": 1156608
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.105462134498805e-05,
      "loss": 3.9654,
      "step": 1157120
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.104623539747753e-05,
      "loss": 3.9515,
      "step": 1157632
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.103784944996702e-05,
      "loss": 3.9476,
      "step": 1158144
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.102947988126023e-05,
      "loss": 3.9606,
      "step": 1158656
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.102109393374971e-05,
      "loss": 3.946,
      "step": 1159168
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.101270798623919e-05,
      "loss": 3.9554,
      "step": 1159680
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.100432203872867e-05,
      "loss": 3.9549,
      "step": 1160192
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.099595247002188e-05,
      "loss": 3.9649,
      "step": 1160704
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.098756652251136e-05,
      "loss": 3.9607,
      "step": 1161216
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.097918057500084e-05,
      "loss": 3.9577,
      "step": 1161728
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.097079462749032e-05,
      "loss": 3.9546,
      "step": 1162240
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0962425058783526e-05,
      "loss": 3.9626,
      "step": 1162752
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0954039111273006e-05,
      "loss": 3.9562,
      "step": 1163264
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0945653163762486e-05,
      "loss": 3.9389,
      "step": 1163776
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.09372835950557e-05,
      "loss": 3.9577,
      "step": 1164288
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.092889764754518e-05,
      "loss": 3.9443,
      "step": 1164800
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.092051170003466e-05,
      "loss": 3.9493,
      "step": 1165312
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.091212575252414e-05,
      "loss": 3.9515,
      "step": 1165824
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.090373980501362e-05,
      "loss": 3.96,
      "step": 1166336
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.08953538575031e-05,
      "loss": 3.9485,
      "step": 1166848
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.088696790999258e-05,
      "loss": 3.953,
      "step": 1167360
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0878581962482055e-05,
      "loss": 3.9609,
      "step": 1167872
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0870196014971535e-05,
      "loss": 3.9476,
      "step": 1168384
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0861810067461015e-05,
      "loss": 3.9631,
      "step": 1168896
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0853424119950495e-05,
      "loss": 3.9399,
      "step": 1169408
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0845038172439975e-05,
      "loss": 3.9479,
      "step": 1169920
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.083666860373319e-05,
      "loss": 3.9549,
      "step": 1170432
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.082828265622267e-05,
      "loss": 3.9564,
      "step": 1170944
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.081989670871215e-05,
      "loss": 3.9554,
      "step": 1171456
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.081151076120163e-05,
      "loss": 3.9389,
      "step": 1171968
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.080312481369111e-05,
      "loss": 3.9526,
      "step": 1172480
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.079475524498432e-05,
      "loss": 3.9441,
      "step": 1172992
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.07863692974738e-05,
      "loss": 3.967,
      "step": 1173504
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.077798334996328e-05,
      "loss": 3.9306,
      "step": 1174016
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.076959740245276e-05,
      "loss": 3.9602,
      "step": 1174528
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.076121145494224e-05,
      "loss": 3.9523,
      "step": 1175040
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.075282550743172e-05,
      "loss": 3.9355,
      "step": 1175552
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.07444395599212e-05,
      "loss": 3.9529,
      "step": 1176064
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.073606999121441e-05,
      "loss": 3.9486,
      "step": 1176576
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.072768404370389e-05,
      "loss": 3.9322,
      "step": 1177088
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0719298096193375e-05,
      "loss": 3.9379,
      "step": 1177600
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0710912148682855e-05,
      "loss": 3.9605,
      "step": 1178112
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0702526201172335e-05,
      "loss": 3.9481,
      "step": 1178624
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0694140253661815e-05,
      "loss": 3.9457,
      "step": 1179136
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0685754306151295e-05,
      "loss": 3.9379,
      "step": 1179648
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.067736835864077e-05,
      "loss": 3.9384,
      "step": 1180160
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.066898241113025e-05,
      "loss": 3.9526,
      "step": 1180672
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0660612842423464e-05,
      "loss": 3.9558,
      "step": 1181184
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0652226894912944e-05,
      "loss": 3.9498,
      "step": 1181696
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.064384094740242e-05,
      "loss": 3.9592,
      "step": 1182208
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.06354549998919e-05,
      "loss": 3.9562,
      "step": 1182720
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.062708543118511e-05,
      "loss": 3.9492,
      "step": 1183232
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.061871586247833e-05,
      "loss": 3.9487,
      "step": 1183744
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.061032991496781e-05,
      "loss": 3.9455,
      "step": 1184256
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.060194396745729e-05,
      "loss": 3.949,
      "step": 1184768
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.059355801994677e-05,
      "loss": 3.9492,
      "step": 1185280
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.058517207243624e-05,
      "loss": 3.944,
      "step": 1185792
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.057680250372946e-05,
      "loss": 3.9571,
      "step": 1186304
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.056841655621894e-05,
      "loss": 3.956,
      "step": 1186816
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.056003060870842e-05,
      "loss": 3.9521,
      "step": 1187328
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.055164466119789e-05,
      "loss": 3.9356,
      "step": 1187840
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0543275092491106e-05,
      "loss": 3.9453,
      "step": 1188352
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0534889144980586e-05,
      "loss": 3.9598,
      "step": 1188864
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0526503197470066e-05,
      "loss": 3.9545,
      "step": 1189376
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0518117249959546e-05,
      "loss": 3.9484,
      "step": 1189888
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.0509731302449023e-05,
      "loss": 3.9605,
      "step": 1190400
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.050136173374224e-05,
      "loss": 3.9513,
      "step": 1190912
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0492975786231715e-05,
      "loss": 3.9467,
      "step": 1191424
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0484589838721195e-05,
      "loss": 3.9477,
      "step": 1191936
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.047622027001441e-05,
      "loss": 3.956,
      "step": 1192448
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.046783432250389e-05,
      "loss": 3.9482,
      "step": 1192960
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0459448374993364e-05,
      "loss": 3.9615,
      "step": 1193472
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0451062427482848e-05,
      "loss": 3.9396,
      "step": 1193984
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0442676479972328e-05,
      "loss": 3.9477,
      "step": 1194496
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0434290532461808e-05,
      "loss": 3.9452,
      "step": 1195008
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0425904584951288e-05,
      "loss": 3.956,
      "step": 1195520
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0417518637440767e-05,
      "loss": 3.9596,
      "step": 1196032
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0409132689930247e-05,
      "loss": 3.9388,
      "step": 1196544
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0400746742419727e-05,
      "loss": 3.9488,
      "step": 1197056
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0392360794909207e-05,
      "loss": 3.9398,
      "step": 1197568
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0383974847398687e-05,
      "loss": 3.9508,
      "step": 1198080
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.03756052786919e-05,
      "loss": 3.935,
      "step": 1198592
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.036721933118138e-05,
      "loss": 3.9473,
      "step": 1199104
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.035883338367086e-05,
      "loss": 3.9492,
      "step": 1199616
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.035044743616034e-05,
      "loss": 3.9447,
      "step": 1200128
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.034207786745355e-05,
      "loss": 3.9505,
      "step": 1200640
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.033369191994303e-05,
      "loss": 3.9476,
      "step": 1201152
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0325305972432512e-05,
      "loss": 3.9355,
      "step": 1201664
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.031693640372572e-05,
      "loss": 3.9539,
      "step": 1202176
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.03085504562152e-05,
      "loss": 3.9518,
      "step": 1202688
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.030016450870468e-05,
      "loss": 3.9487,
      "step": 1203200
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.029177856119416e-05,
      "loss": 3.9533,
      "step": 1203712
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.028339261368364e-05,
      "loss": 3.9548,
      "step": 1204224
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0275023044976853e-05,
      "loss": 3.9493,
      "step": 1204736
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0266637097466333e-05,
      "loss": 3.9395,
      "step": 1205248
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0258251149955813e-05,
      "loss": 3.9503,
      "step": 1205760
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0249865202445293e-05,
      "loss": 3.9454,
      "step": 1206272
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0241479254934773e-05,
      "loss": 3.9419,
      "step": 1206784
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.023309330742425e-05,
      "loss": 3.9442,
      "step": 1207296
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.022470735991373e-05,
      "loss": 3.9464,
      "step": 1207808
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.021632141240321e-05,
      "loss": 3.9495,
      "step": 1208320
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0207951843696426e-05,
      "loss": 3.9393,
      "step": 1208832
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.01995658961859e-05,
      "loss": 3.9439,
      "step": 1209344
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0191196327479115e-05,
      "loss": 3.9415,
      "step": 1209856
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0182810379968595e-05,
      "loss": 3.9474,
      "step": 1210368
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0174424432458075e-05,
      "loss": 3.9472,
      "step": 1210880
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.016603848494755e-05,
      "loss": 3.9529,
      "step": 1211392
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0157668916240767e-05,
      "loss": 3.9475,
      "step": 1211904
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0149282968730247e-05,
      "loss": 3.9423,
      "step": 1212416
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.014089702121972e-05,
      "loss": 3.9513,
      "step": 1212928
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0132511073709203e-05,
      "loss": 3.9521,
      "step": 1213440
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0124125126198683e-05,
      "loss": 3.9543,
      "step": 1213952
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0115739178688163e-05,
      "loss": 3.9551,
      "step": 1214464
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0107353231177643e-05,
      "loss": 3.9471,
      "step": 1214976
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0098983662470852e-05,
      "loss": 3.9601,
      "step": 1215488
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0090597714960332e-05,
      "loss": 3.9521,
      "step": 1216000
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0082228146253548e-05,
      "loss": 3.9428,
      "step": 1216512
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0073842198743025e-05,
      "loss": 3.9458,
      "step": 1217024
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0065456251232505e-05,
      "loss": 3.9427,
      "step": 1217536
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0057070303721985e-05,
      "loss": 3.9443,
      "step": 1218048
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0048684356211465e-05,
      "loss": 3.9394,
      "step": 1218560
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0040298408700945e-05,
      "loss": 3.9497,
      "step": 1219072
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0031912461190425e-05,
      "loss": 3.9378,
      "step": 1219584
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0023526513679905e-05,
      "loss": 3.9512,
      "step": 1220096
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.0015140566169388e-05,
      "loss": 3.949,
      "step": 1220608
    },
    {
      "epoch": 1.03,
      "learning_rate": 3.0006754618658868e-05,
      "loss": 3.947,
      "step": 1221120
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.013096809387207,
      "eval_runtime": 303.1106,
      "eval_samples_per_second": 1258.917,
      "eval_steps_per_second": 39.342,
      "step": 1221120
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9998368671148348e-05,
      "loss": 3.9394,
      "step": 1221632
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9989982723637828e-05,
      "loss": 3.9374,
      "step": 1222144
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9981596776127308e-05,
      "loss": 3.9489,
      "step": 1222656
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9973210828616788e-05,
      "loss": 3.9498,
      "step": 1223168
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9964824881106264e-05,
      "loss": 3.947,
      "step": 1223680
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9956438933595744e-05,
      "loss": 3.9414,
      "step": 1224192
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9948052986085224e-05,
      "loss": 3.937,
      "step": 1224704
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9939667038574704e-05,
      "loss": 3.94,
      "step": 1225216
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9931281091064184e-05,
      "loss": 3.9539,
      "step": 1225728
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9922895143553664e-05,
      "loss": 3.9443,
      "step": 1226240
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9914509196043144e-05,
      "loss": 3.9498,
      "step": 1226752
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9906123248532624e-05,
      "loss": 3.9574,
      "step": 1227264
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9897737301022104e-05,
      "loss": 3.9289,
      "step": 1227776
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9889351353511584e-05,
      "loss": 3.937,
      "step": 1228288
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9880965406001067e-05,
      "loss": 3.9345,
      "step": 1228800
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9872579458490547e-05,
      "loss": 3.9427,
      "step": 1229312
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9864193510980027e-05,
      "loss": 3.9459,
      "step": 1229824
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.98558075634695e-05,
      "loss": 3.9424,
      "step": 1230336
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.984742161595898e-05,
      "loss": 3.9375,
      "step": 1230848
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9839052047252196e-05,
      "loss": 3.9627,
      "step": 1231360
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9830666099741676e-05,
      "loss": 3.9451,
      "step": 1231872
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9822280152231153e-05,
      "loss": 3.9481,
      "step": 1232384
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9813894204720633e-05,
      "loss": 3.9463,
      "step": 1232896
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.980552463601385e-05,
      "loss": 3.9601,
      "step": 1233408
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.979713868850332e-05,
      "loss": 3.9398,
      "step": 1233920
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9788752740992805e-05,
      "loss": 3.9398,
      "step": 1234432
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9780366793482285e-05,
      "loss": 3.9468,
      "step": 1234944
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.97719972247755e-05,
      "loss": 3.9367,
      "step": 1235456
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9763611277264974e-05,
      "loss": 3.9423,
      "step": 1235968
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9755225329754454e-05,
      "loss": 3.9431,
      "step": 1236480
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9746839382243934e-05,
      "loss": 3.956,
      "step": 1236992
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9738453434733414e-05,
      "loss": 3.9455,
      "step": 1237504
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9730067487222897e-05,
      "loss": 3.9495,
      "step": 1238016
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9721681539712377e-05,
      "loss": 3.9446,
      "step": 1238528
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9713311971005586e-05,
      "loss": 3.9474,
      "step": 1239040
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9704942402298795e-05,
      "loss": 3.9458,
      "step": 1239552
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9696556454788275e-05,
      "loss": 3.9305,
      "step": 1240064
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.968817050727776e-05,
      "loss": 3.9417,
      "step": 1240576
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.967978455976724e-05,
      "loss": 3.9319,
      "step": 1241088
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.967139861225672e-05,
      "loss": 3.9419,
      "step": 1241600
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.96630126647462e-05,
      "loss": 3.9418,
      "step": 1242112
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.965462671723568e-05,
      "loss": 3.9496,
      "step": 1242624
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.964624076972516e-05,
      "loss": 3.9388,
      "step": 1243136
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.963785482221464e-05,
      "loss": 3.9433,
      "step": 1243648
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.962946887470412e-05,
      "loss": 3.9491,
      "step": 1244160
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.96210829271936e-05,
      "loss": 3.9381,
      "step": 1244672
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.961269697968308e-05,
      "loss": 3.9503,
      "step": 1245184
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.960432741097629e-05,
      "loss": 3.9272,
      "step": 1245696
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.959594146346577e-05,
      "loss": 3.9375,
      "step": 1246208
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.958755551595525e-05,
      "loss": 3.9473,
      "step": 1246720
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.957916956844473e-05,
      "loss": 3.9443,
      "step": 1247232
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.957078362093421e-05,
      "loss": 3.9427,
      "step": 1247744
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9562414052227423e-05,
      "loss": 3.9245,
      "step": 1248256
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9554028104716903e-05,
      "loss": 3.9452,
      "step": 1248768
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9545658536010112e-05,
      "loss": 3.93,
      "step": 1249280
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9537272588499592e-05,
      "loss": 3.9572,
      "step": 1249792
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9528886640989072e-05,
      "loss": 3.9245,
      "step": 1250304
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9520500693478552e-05,
      "loss": 3.9481,
      "step": 1250816
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9512114745968032e-05,
      "loss": 3.9377,
      "step": 1251328
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9503728798457515e-05,
      "loss": 3.9271,
      "step": 1251840
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.949534285094699e-05,
      "loss": 3.9432,
      "step": 1252352
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9486973282240204e-05,
      "loss": 3.9368,
      "step": 1252864
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9478587334729684e-05,
      "loss": 3.9195,
      "step": 1253376
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.947020138721916e-05,
      "loss": 3.9295,
      "step": 1253888
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.946181543970864e-05,
      "loss": 3.9508,
      "step": 1254400
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.945342949219812e-05,
      "loss": 3.9354,
      "step": 1254912
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.94450435446876e-05,
      "loss": 3.9333,
      "step": 1255424
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.943667397598081e-05,
      "loss": 3.9315,
      "step": 1255936
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.942828802847029e-05,
      "loss": 3.9275,
      "step": 1256448
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.941990208095977e-05,
      "loss": 3.9401,
      "step": 1256960
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9411516133449253e-05,
      "loss": 3.9473,
      "step": 1257472
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9403130185938733e-05,
      "loss": 3.9405,
      "step": 1257984
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9394744238428213e-05,
      "loss": 3.951,
      "step": 1258496
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9386358290917693e-05,
      "loss": 3.9427,
      "step": 1259008
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9377972343407173e-05,
      "loss": 3.9426,
      "step": 1259520
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9369602774700382e-05,
      "loss": 3.9329,
      "step": 1260032
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9361216827189862e-05,
      "loss": 3.9397,
      "step": 1260544
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9352830879679345e-05,
      "loss": 3.941,
      "step": 1261056
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9344444932168825e-05,
      "loss": 3.933,
      "step": 1261568
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9336058984658305e-05,
      "loss": 3.9335,
      "step": 1262080
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9327673037147785e-05,
      "loss": 3.9449,
      "step": 1262592
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9319303468440994e-05,
      "loss": 3.9483,
      "step": 1263104
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9310917520930474e-05,
      "loss": 3.9384,
      "step": 1263616
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9302531573419954e-05,
      "loss": 3.9237,
      "step": 1264128
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9294145625909438e-05,
      "loss": 3.9397,
      "step": 1264640
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9285759678398917e-05,
      "loss": 3.9478,
      "step": 1265152
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.9277373730888397e-05,
      "loss": 3.9468,
      "step": 1265664
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.926898778337787e-05,
      "loss": 3.9365,
      "step": 1266176
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.926060183586735e-05,
      "loss": 3.953,
      "step": 1266688
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9252232267160566e-05,
      "loss": 3.9421,
      "step": 1267200
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9243846319650046e-05,
      "loss": 3.9318,
      "step": 1267712
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.923547675094326e-05,
      "loss": 3.9391,
      "step": 1268224
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.922709080343274e-05,
      "loss": 3.9428,
      "step": 1268736
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.921870485592222e-05,
      "loss": 3.9358,
      "step": 1269248
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.92103189084117e-05,
      "loss": 3.953,
      "step": 1269760
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9201932960901175e-05,
      "loss": 3.9298,
      "step": 1270272
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9193547013390655e-05,
      "loss": 3.9319,
      "step": 1270784
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.918517744468387e-05,
      "loss": 3.9352,
      "step": 1271296
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9176791497173344e-05,
      "loss": 3.9475,
      "step": 1271808
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9168405549662824e-05,
      "loss": 3.9453,
      "step": 1272320
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9160019602152304e-05,
      "loss": 3.9309,
      "step": 1272832
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9151633654641784e-05,
      "loss": 3.9356,
      "step": 1273344
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9143247707131268e-05,
      "loss": 3.9313,
      "step": 1273856
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9134861759620747e-05,
      "loss": 3.945,
      "step": 1274368
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9126475812110227e-05,
      "loss": 3.9225,
      "step": 1274880
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9118106243403437e-05,
      "loss": 3.9368,
      "step": 1275392
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9109736674696646e-05,
      "loss": 3.941,
      "step": 1275904
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.910135072718613e-05,
      "loss": 3.9358,
      "step": 1276416
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.909296477967561e-05,
      "loss": 3.9399,
      "step": 1276928
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.908457883216509e-05,
      "loss": 3.9368,
      "step": 1277440
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.907619288465457e-05,
      "loss": 3.9264,
      "step": 1277952
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9067823315947778e-05,
      "loss": 3.9395,
      "step": 1278464
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9059437368437258e-05,
      "loss": 3.9406,
      "step": 1278976
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9051051420926738e-05,
      "loss": 3.9388,
      "step": 1279488
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.904266547341622e-05,
      "loss": 3.9439,
      "step": 1280000
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.90342795259057e-05,
      "loss": 3.9444,
      "step": 1280512
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.902590995719891e-05,
      "loss": 3.9413,
      "step": 1281024
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.901752400968839e-05,
      "loss": 3.9311,
      "step": 1281536
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.900913806217787e-05,
      "loss": 3.9374,
      "step": 1282048
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.900075211466735e-05,
      "loss": 3.9365,
      "step": 1282560
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.899236616715683e-05,
      "loss": 3.9285,
      "step": 1283072
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8983980219646313e-05,
      "loss": 3.9346,
      "step": 1283584
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8975594272135793e-05,
      "loss": 3.9332,
      "step": 1284096
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8967208324625273e-05,
      "loss": 3.9444,
      "step": 1284608
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8958838755918482e-05,
      "loss": 3.9286,
      "step": 1285120
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8950452808407962e-05,
      "loss": 3.9325,
      "step": 1285632
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8942066860897442e-05,
      "loss": 3.9284,
      "step": 1286144
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8933697292190655e-05,
      "loss": 3.9363,
      "step": 1286656
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8925311344680135e-05,
      "loss": 3.931,
      "step": 1287168
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8916925397169615e-05,
      "loss": 3.9502,
      "step": 1287680
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8908539449659095e-05,
      "loss": 3.9384,
      "step": 1288192
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8900153502148575e-05,
      "loss": 3.9333,
      "step": 1288704
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8891767554638055e-05,
      "loss": 3.9373,
      "step": 1289216
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8883381607127535e-05,
      "loss": 3.9386,
      "step": 1289728
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.887499565961701e-05,
      "loss": 3.944,
      "step": 1290240
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8866626090910227e-05,
      "loss": 3.946,
      "step": 1290752
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8858240143399707e-05,
      "loss": 3.9407,
      "step": 1291264
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8849870574692916e-05,
      "loss": 3.9476,
      "step": 1291776
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8841484627182396e-05,
      "loss": 3.9402,
      "step": 1292288
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8833098679671876e-05,
      "loss": 3.9357,
      "step": 1292800
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8824712732161356e-05,
      "loss": 3.932,
      "step": 1293312
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8816326784650832e-05,
      "loss": 3.9385,
      "step": 1293824
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8807940837140312e-05,
      "loss": 3.9349,
      "step": 1294336
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8799554889629792e-05,
      "loss": 3.9278,
      "step": 1294848
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8791168942119272e-05,
      "loss": 3.9354,
      "step": 1295360
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8782815752216217e-05,
      "loss": 3.9307,
      "step": 1295872
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.87744298047057e-05,
      "loss": 3.9396,
      "step": 1296384
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.876606023599891e-05,
      "loss": 3.9375,
      "step": 1296896
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.875767428848839e-05,
      "loss": 3.9356,
      "step": 1297408
    },
    {
      "epoch": 0.03,
      "eval_loss": 4.008285045623779,
      "eval_runtime": 289.0949,
      "eval_samples_per_second": 1319.95,
      "eval_steps_per_second": 41.249,
      "step": 1297440
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.874928834097787e-05,
      "loss": 3.9289,
      "step": 1297920
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.874090239346735e-05,
      "loss": 3.929,
      "step": 1298432
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.873251644595683e-05,
      "loss": 3.9409,
      "step": 1298944
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8724130498446306e-05,
      "loss": 3.9359,
      "step": 1299456
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8715744550935786e-05,
      "loss": 3.938,
      "step": 1299968
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8707358603425266e-05,
      "loss": 3.9316,
      "step": 1300480
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8698972655914746e-05,
      "loss": 3.9255,
      "step": 1300992
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8690586708404226e-05,
      "loss": 3.9305,
      "step": 1301504
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8682200760893706e-05,
      "loss": 3.9423,
      "step": 1302016
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8673814813383186e-05,
      "loss": 3.9384,
      "step": 1302528
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.86654452446764e-05,
      "loss": 3.9353,
      "step": 1303040
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.865705929716588e-05,
      "loss": 3.9449,
      "step": 1303552
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.864867334965536e-05,
      "loss": 3.9203,
      "step": 1304064
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8640287402144838e-05,
      "loss": 3.9304,
      "step": 1304576
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8631917833438047e-05,
      "loss": 3.9307,
      "step": 1305088
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.862353188592753e-05,
      "loss": 3.9224,
      "step": 1305600
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.861514593841701e-05,
      "loss": 3.938,
      "step": 1306112
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.860675999090649e-05,
      "loss": 3.9298,
      "step": 1306624
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.85983904221997e-05,
      "loss": 3.929,
      "step": 1307136
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.859000447468918e-05,
      "loss": 3.9593,
      "step": 1307648
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.858161852717866e-05,
      "loss": 3.9345,
      "step": 1308160
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.857323257966814e-05,
      "loss": 3.9375,
      "step": 1308672
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8564846632157623e-05,
      "loss": 3.9381,
      "step": 1309184
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8556460684647103e-05,
      "loss": 3.9489,
      "step": 1309696
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8548074737136583e-05,
      "loss": 3.9303,
      "step": 1310208
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8539688789626063e-05,
      "loss": 3.9294,
      "step": 1310720
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8531319220919272e-05,
      "loss": 3.9335,
      "step": 1311232
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8522949652212484e-05,
      "loss": 3.9324,
      "step": 1311744
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.8514563704701964e-05,
      "loss": 3.9318,
      "step": 1312256
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8506177757191444e-05,
      "loss": 3.9282,
      "step": 1312768
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8497791809680924e-05,
      "loss": 3.9497,
      "step": 1313280
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8489405862170404e-05,
      "loss": 3.9381,
      "step": 1313792
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8481019914659884e-05,
      "loss": 3.9364,
      "step": 1314304
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8472633967149364e-05,
      "loss": 3.9365,
      "step": 1314816
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8464264398442577e-05,
      "loss": 3.9384,
      "step": 1315328
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8455878450932057e-05,
      "loss": 3.936,
      "step": 1315840
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8447492503421537e-05,
      "loss": 3.9213,
      "step": 1316352
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8439106555911016e-05,
      "loss": 3.9318,
      "step": 1316864
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.843072060840049e-05,
      "loss": 3.9201,
      "step": 1317376
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.842233466088997e-05,
      "loss": 3.9335,
      "step": 1317888
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8413948713379453e-05,
      "loss": 3.9299,
      "step": 1318400
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8405562765868933e-05,
      "loss": 3.9381,
      "step": 1318912
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8397176818358413e-05,
      "loss": 3.9287,
      "step": 1319424
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8388823628455358e-05,
      "loss": 3.9327,
      "step": 1319936
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8380437680944838e-05,
      "loss": 3.9402,
      "step": 1320448
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8372051733434314e-05,
      "loss": 3.9295,
      "step": 1320960
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8363665785923794e-05,
      "loss": 3.937,
      "step": 1321472
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8355279838413274e-05,
      "loss": 3.9227,
      "step": 1321984
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8346893890902754e-05,
      "loss": 3.9282,
      "step": 1322496
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8338507943392234e-05,
      "loss": 3.9364,
      "step": 1323008
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8330138374685443e-05,
      "loss": 3.9327,
      "step": 1323520
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8321752427174923e-05,
      "loss": 3.9373,
      "step": 1324032
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.831338285846814e-05,
      "loss": 3.9108,
      "step": 1324544
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8304996910957616e-05,
      "loss": 3.936,
      "step": 1325056
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8296610963447096e-05,
      "loss": 3.9174,
      "step": 1325568
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8288225015936576e-05,
      "loss": 3.946,
      "step": 1326080
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8279839068426056e-05,
      "loss": 3.9177,
      "step": 1326592
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8271453120915536e-05,
      "loss": 3.9369,
      "step": 1327104
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8263067173405015e-05,
      "loss": 3.9307,
      "step": 1327616
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8254681225894495e-05,
      "loss": 3.9192,
      "step": 1328128
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.824629527838398e-05,
      "loss": 3.9276,
      "step": 1328640
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.823790933087346e-05,
      "loss": 3.9298,
      "step": 1329152
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.822952338336294e-05,
      "loss": 3.9119,
      "step": 1329664
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.822113743585242e-05,
      "loss": 3.9202,
      "step": 1330176
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8212767867145628e-05,
      "loss": 3.9364,
      "step": 1330688
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.820439829843884e-05,
      "loss": 3.9278,
      "step": 1331200
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.819601235092832e-05,
      "loss": 3.924,
      "step": 1331712
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.81876264034178e-05,
      "loss": 3.9215,
      "step": 1332224
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.817924045590728e-05,
      "loss": 3.9159,
      "step": 1332736
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.817085450839676e-05,
      "loss": 3.9306,
      "step": 1333248
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.816246856088624e-05,
      "loss": 3.9332,
      "step": 1333760
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.815408261337572e-05,
      "loss": 3.9336,
      "step": 1334272
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8145713044668932e-05,
      "loss": 3.9403,
      "step": 1334784
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8137327097158412e-05,
      "loss": 3.9338,
      "step": 1335296
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8128941149647892e-05,
      "loss": 3.9353,
      "step": 1335808
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8120555202137372e-05,
      "loss": 3.9203,
      "step": 1336320
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8112169254626852e-05,
      "loss": 3.9316,
      "step": 1336832
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8103783307116325e-05,
      "loss": 3.9278,
      "step": 1337344
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.809539735960581e-05,
      "loss": 3.9215,
      "step": 1337856
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.808701141209529e-05,
      "loss": 3.9258,
      "step": 1338368
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8078641843388498e-05,
      "loss": 3.9311,
      "step": 1338880
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8070272274681714e-05,
      "loss": 3.9412,
      "step": 1339392
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8061886327171194e-05,
      "loss": 3.9327,
      "step": 1339904
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8053516758464403e-05,
      "loss": 3.9167,
      "step": 1340416
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8045130810953886e-05,
      "loss": 3.9235,
      "step": 1340928
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8036744863443366e-05,
      "loss": 3.9386,
      "step": 1341440
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8028358915932846e-05,
      "loss": 3.9343,
      "step": 1341952
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.8019972968422326e-05,
      "loss": 3.9305,
      "step": 1342464
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.80115870209118e-05,
      "loss": 3.9396,
      "step": 1342976
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.800320107340128e-05,
      "loss": 3.9351,
      "step": 1343488
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7994815125890762e-05,
      "loss": 3.9199,
      "step": 1344000
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7986429178380242e-05,
      "loss": 3.9294,
      "step": 1344512
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7978043230869722e-05,
      "loss": 3.937,
      "step": 1345024
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7969657283359202e-05,
      "loss": 3.9256,
      "step": 1345536
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7961271335848682e-05,
      "loss": 3.9432,
      "step": 1346048
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.795290176714189e-05,
      "loss": 3.9218,
      "step": 1346560
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.794451581963137e-05,
      "loss": 3.923,
      "step": 1347072
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7936129872120855e-05,
      "loss": 3.9278,
      "step": 1347584
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7927743924610335e-05,
      "loss": 3.9356,
      "step": 1348096
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7919357977099815e-05,
      "loss": 3.9351,
      "step": 1348608
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7910972029589295e-05,
      "loss": 3.9233,
      "step": 1349120
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7902586082078774e-05,
      "loss": 3.9235,
      "step": 1349632
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7894200134568254e-05,
      "loss": 3.9214,
      "step": 1350144
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7885830565861464e-05,
      "loss": 3.9368,
      "step": 1350656
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7877444618350947e-05,
      "loss": 3.9113,
      "step": 1351168
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7869058670840427e-05,
      "loss": 3.9208,
      "step": 1351680
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7860689102133636e-05,
      "loss": 3.934,
      "step": 1352192
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7852303154623116e-05,
      "loss": 3.9272,
      "step": 1352704
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7843917207112596e-05,
      "loss": 3.9241,
      "step": 1353216
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7835531259602076e-05,
      "loss": 3.9271,
      "step": 1353728
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7827145312091556e-05,
      "loss": 3.9201,
      "step": 1354240
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7818759364581036e-05,
      "loss": 3.9303,
      "step": 1354752
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7810389795874248e-05,
      "loss": 3.9268,
      "step": 1355264
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7802003848363728e-05,
      "loss": 3.9289,
      "step": 1355776
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7793617900853208e-05,
      "loss": 3.9322,
      "step": 1356288
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7785231953342688e-05,
      "loss": 3.9381,
      "step": 1356800
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7776846005832165e-05,
      "loss": 3.9282,
      "step": 1357312
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7768460058321645e-05,
      "loss": 3.9212,
      "step": 1357824
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7760074110811125e-05,
      "loss": 3.9295,
      "step": 1358336
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7751704542104334e-05,
      "loss": 3.9271,
      "step": 1358848
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7743318594593814e-05,
      "loss": 3.917,
      "step": 1359360
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7734932647083294e-05,
      "loss": 3.9245,
      "step": 1359872
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7726546699572774e-05,
      "loss": 3.9236,
      "step": 1360384
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7718160752062257e-05,
      "loss": 3.9347,
      "step": 1360896
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7709774804551737e-05,
      "loss": 3.9183,
      "step": 1361408
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7701388857041217e-05,
      "loss": 3.9264,
      "step": 1361920
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7693002909530697e-05,
      "loss": 3.912,
      "step": 1362432
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7684633340823906e-05,
      "loss": 3.929,
      "step": 1362944
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7676247393313386e-05,
      "loss": 3.9226,
      "step": 1363456
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7667861445802866e-05,
      "loss": 3.9387,
      "step": 1363968
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.765947549829235e-05,
      "loss": 3.9297,
      "step": 1364480
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7651105929585558e-05,
      "loss": 3.9289,
      "step": 1364992
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7642719982075038e-05,
      "loss": 3.9219,
      "step": 1365504
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7634334034564518e-05,
      "loss": 3.9269,
      "step": 1366016
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7625948087053998e-05,
      "loss": 3.9368,
      "step": 1366528
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.761757851834721e-05,
      "loss": 3.9395,
      "step": 1367040
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.760919257083669e-05,
      "loss": 3.9316,
      "step": 1367552
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.760080662332617e-05,
      "loss": 3.937,
      "step": 1368064
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.759242067581565e-05,
      "loss": 3.9276,
      "step": 1368576
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.758405110710886e-05,
      "loss": 3.9266,
      "step": 1369088
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.757566515959834e-05,
      "loss": 3.9239,
      "step": 1369600
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.756727921208782e-05,
      "loss": 3.9281,
      "step": 1370112
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7558893264577303e-05,
      "loss": 3.9274,
      "step": 1370624
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7550523695870512e-05,
      "loss": 3.9175,
      "step": 1371136
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7542137748359992e-05,
      "loss": 3.9273,
      "step": 1371648
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.75337681796532e-05,
      "loss": 3.9227,
      "step": 1372160
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.752538223214268e-05,
      "loss": 3.9308,
      "step": 1372672
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7516996284632164e-05,
      "loss": 3.9263,
      "step": 1373184
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.7508610337121644e-05,
      "loss": 3.9268,
      "step": 1373696
    },
    {
      "epoch": 1.03,
      "eval_loss": 4.003251552581787,
      "eval_runtime": 314.9348,
      "eval_samples_per_second": 1211.651,
      "eval_steps_per_second": 37.865,
      "step": 1373760
    }
  ],
  "logging_steps": 512,
  "max_steps": 3052726,
  "num_train_epochs": 9223372036854775807,
  "save_steps": 10,
  "total_flos": 5.6389591165700275e+17,
  "trial_name": null,
  "trial_params": null
}