{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.2057062393817929,
  "eval_steps": 2,
  "global_step": 30240,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 4.0625,
      "learning_rate": 4.896848787123102e-07,
      "loss": 1.9129,
      "step": 144
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.609375,
      "learning_rate": 9.793697574246203e-07,
      "loss": 1.9063,
      "step": 288
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.734375,
      "learning_rate": 1.4690546361369305e-06,
      "loss": 1.8962,
      "step": 432
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.71875,
      "learning_rate": 1.9587395148492407e-06,
      "loss": 1.8796,
      "step": 576
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.375,
      "learning_rate": 2.4484243935615507e-06,
      "loss": 1.9003,
      "step": 720
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.3125,
      "learning_rate": 2.938109272273861e-06,
      "loss": 1.8795,
      "step": 864
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.953125,
      "learning_rate": 3.427794150986171e-06,
      "loss": 1.8969,
      "step": 1008
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.375,
      "learning_rate": 3.917479029698481e-06,
      "loss": 1.8597,
      "step": 1152
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.203125,
      "learning_rate": 4.4071639084107905e-06,
      "loss": 1.8685,
      "step": 1296
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.0625,
      "learning_rate": 4.896848787123101e-06,
      "loss": 1.867,
      "step": 1440
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.203125,
      "learning_rate": 5.386533665835411e-06,
      "loss": 1.8622,
      "step": 1584
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.109375,
      "learning_rate": 5.876218544547722e-06,
      "loss": 1.8697,
      "step": 1728
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.453125,
      "learning_rate": 6.365903423260031e-06,
      "loss": 1.873,
      "step": 1872
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.4375,
      "learning_rate": 6.855588301972342e-06,
      "loss": 1.8698,
      "step": 2016
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.125,
      "learning_rate": 7.345273180684652e-06,
      "loss": 1.8587,
      "step": 2160
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.21875,
      "learning_rate": 7.834958059396963e-06,
      "loss": 1.8685,
      "step": 2304
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.921875,
      "learning_rate": 8.324642938109274e-06,
      "loss": 1.8771,
      "step": 2448
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.1875,
      "learning_rate": 8.814327816821581e-06,
      "loss": 1.8423,
      "step": 2592
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.078125,
      "learning_rate": 9.304012695533892e-06,
      "loss": 1.8542,
      "step": 2736
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.015625,
      "learning_rate": 9.793697574246203e-06,
      "loss": 1.8577,
      "step": 2880
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.859375,
      "learning_rate": 1.0283382452958513e-05,
      "loss": 1.8372,
      "step": 3024
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.28125,
      "learning_rate": 1.0773067331670823e-05,
      "loss": 1.8332,
      "step": 3168
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.09375,
      "learning_rate": 1.1262752210383133e-05,
      "loss": 1.8618,
      "step": 3312
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0,
      "learning_rate": 1.1752437089095444e-05,
      "loss": 1.8525,
      "step": 3456
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0625,
      "learning_rate": 1.2242121967807753e-05,
      "loss": 1.8357,
      "step": 3600
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.96875,
      "learning_rate": 1.2731806846520062e-05,
      "loss": 1.8269,
      "step": 3744
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.875,
      "learning_rate": 1.3221491725232373e-05,
      "loss": 1.8274,
      "step": 3888
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.03125,
      "learning_rate": 1.3711176603944684e-05,
      "loss": 1.8343,
      "step": 4032
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.234375,
      "learning_rate": 1.4200861482656995e-05,
      "loss": 1.8468,
      "step": 4176
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.0625,
      "learning_rate": 1.4690546361369304e-05,
      "loss": 1.8347,
      "step": 4320
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.875,
      "learning_rate": 1.4999994886948122e-05,
      "loss": 1.8314,
      "step": 4464
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.640625,
      "learning_rate": 1.499992935844023e-05,
      "loss": 1.832,
      "step": 4608
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.78125,
      "learning_rate": 1.4999788341738499e-05,
      "loss": 1.8244,
      "step": 4752
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.75,
      "learning_rate": 1.4999571838262296e-05,
      "loss": 1.8268,
      "step": 4896
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.75,
      "learning_rate": 1.4999279850190762e-05,
      "loss": 1.8101,
      "step": 5040
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.625,
      "learning_rate": 1.4998912380462815e-05,
      "loss": 1.8298,
      "step": 5184
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.71875,
      "learning_rate": 1.4998469432777108e-05,
      "loss": 1.8275,
      "step": 5328
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.6875,
      "learning_rate": 1.4997951011591987e-05,
      "loss": 1.8109,
      "step": 5472
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.671875,
      "learning_rate": 1.4997357122125465e-05,
      "loss": 1.825,
      "step": 5616
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.796875,
      "learning_rate": 1.4996687770355145e-05,
      "loss": 1.8322,
      "step": 5760
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.640625,
      "learning_rate": 1.4995942963018183e-05,
      "loss": 1.8085,
      "step": 5904
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.53125,
      "learning_rate": 1.49951227076112e-05,
      "loss": 1.819,
      "step": 6048
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.609375,
      "learning_rate": 1.4994227012390224e-05,
      "loss": 1.7942,
      "step": 6192
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.84375,
      "learning_rate": 1.4993255886370593e-05,
      "loss": 1.7975,
      "step": 6336
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.4375,
      "learning_rate": 1.4992209339326872e-05,
      "loss": 1.8219,
      "step": 6480
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.734375,
      "learning_rate": 1.4991087381792756e-05,
      "loss": 1.8119,
      "step": 6624
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.421875,
      "learning_rate": 1.4989890025060953e-05,
      "loss": 1.815,
      "step": 6768
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.546875,
      "learning_rate": 1.4988617281183087e-05,
      "loss": 1.7912,
      "step": 6912
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.421875,
      "learning_rate": 1.4987269162969559e-05,
      "loss": 1.8018,
      "step": 7056
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.59375,
      "learning_rate": 1.4985845683989435e-05,
      "loss": 1.7973,
      "step": 7200
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.84375,
      "learning_rate": 1.4984346858570297e-05,
      "loss": 1.8109,
      "step": 7344
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.609375,
      "learning_rate": 1.4982772701798105e-05,
      "loss": 1.8162,
      "step": 7488
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.59375,
      "learning_rate": 1.4981123229517039e-05,
      "loss": 1.7892,
      "step": 7632
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.6875,
      "learning_rate": 1.4979398458329347e-05,
      "loss": 1.7908,
      "step": 7776
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.578125,
      "learning_rate": 1.4977598405595177e-05,
      "loss": 1.7722,
      "step": 7920
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.59375,
      "learning_rate": 1.4975723089432394e-05,
      "loss": 1.7826,
      "step": 8064
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.515625,
      "learning_rate": 1.4973772528716404e-05,
      "loss": 1.8005,
      "step": 8208
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.75,
      "learning_rate": 1.4971746743079972e-05,
      "loss": 1.7911,
      "step": 8352
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.75,
      "learning_rate": 1.4969645752913006e-05,
      "loss": 1.7798,
      "step": 8496
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.734375,
      "learning_rate": 1.4967469579362364e-05,
      "loss": 1.7806,
      "step": 8640
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.03125,
      "learning_rate": 1.4965218244331647e-05,
      "loss": 1.7927,
      "step": 8784
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.671875,
      "learning_rate": 1.4962891770480958e-05,
      "loss": 1.7869,
      "step": 8928
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.734375,
      "learning_rate": 1.4960490181226693e-05,
      "loss": 1.8041,
      "step": 9072
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.5,
      "learning_rate": 1.49580135007413e-05,
      "loss": 1.7747,
      "step": 9216
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.75,
      "learning_rate": 1.495546175395303e-05,
      "loss": 1.7883,
      "step": 9360
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.640625,
      "learning_rate": 1.495283496654569e-05,
      "loss": 1.7776,
      "step": 9504
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.671875,
      "learning_rate": 1.4950133164958392e-05,
      "loss": 1.7834,
      "step": 9648
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.8125,
      "learning_rate": 1.4947356376385275e-05,
      "loss": 1.7711,
      "step": 9792
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.640625,
      "learning_rate": 1.4944504628775233e-05,
      "loss": 1.7704,
      "step": 9936
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.546875,
      "learning_rate": 1.4941577950831642e-05,
      "loss": 1.774,
      "step": 10080
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.953125,
      "learning_rate": 1.4938576372012062e-05,
      "loss": 1.7621,
      "step": 10224
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.515625,
      "learning_rate": 1.4935499922527949e-05,
      "loss": 1.7776,
      "step": 10368
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.640625,
      "learning_rate": 1.4932348633344338e-05,
      "loss": 1.7676,
      "step": 10512
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.671875,
      "learning_rate": 1.492912253617955e-05,
      "loss": 1.7807,
      "step": 10656
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.515625,
      "learning_rate": 1.4925821663504856e-05,
      "loss": 1.7738,
      "step": 10800
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.625,
      "learning_rate": 1.4922446048544157e-05,
      "loss": 1.7791,
      "step": 10944
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.515625,
      "learning_rate": 1.4918995725273649e-05,
      "loss": 1.7673,
      "step": 11088
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.796875,
      "learning_rate": 1.4915470728421481e-05,
      "loss": 1.7691,
      "step": 11232
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.953125,
      "learning_rate": 1.4911871093467414e-05,
      "loss": 1.7634,
      "step": 11376
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.578125,
      "learning_rate": 1.4908196856642441e-05,
      "loss": 1.7649,
      "step": 11520
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.765625,
      "learning_rate": 1.4904448054928444e-05,
      "loss": 1.7783,
      "step": 11664
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.71875,
      "learning_rate": 1.4900624726057816e-05,
      "loss": 1.7363,
      "step": 11808
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.609375,
      "learning_rate": 1.489672690851308e-05,
      "loss": 1.7565,
      "step": 11952
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.578125,
      "learning_rate": 1.4892754641526498e-05,
      "loss": 1.7488,
      "step": 12096
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.40625,
      "learning_rate": 1.4888707965079682e-05,
      "loss": 1.7523,
      "step": 12240
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.703125,
      "learning_rate": 1.4884586919903186e-05,
      "loss": 1.7485,
      "step": 12384
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.40625,
      "learning_rate": 1.4880391547476106e-05,
      "loss": 1.7572,
      "step": 12528
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.5625,
      "learning_rate": 1.4876121890025647e-05,
      "loss": 1.7678,
      "step": 12672
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.359375,
      "learning_rate": 1.4871777990526713e-05,
      "loss": 1.7655,
      "step": 12816
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.515625,
      "learning_rate": 1.4867359892701466e-05,
      "loss": 1.7677,
      "step": 12960
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.609375,
      "learning_rate": 1.4862867641018887e-05,
      "loss": 1.7575,
      "step": 13104
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.828125,
      "learning_rate": 1.4858301280694332e-05,
      "loss": 1.7721,
      "step": 13248
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.6875,
      "learning_rate": 1.4853660857689073e-05,
      "loss": 1.7307,
      "step": 13392
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.375,
      "learning_rate": 1.4848946418709838e-05,
      "loss": 1.7483,
      "step": 13536
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.5625,
      "learning_rate": 1.4844158011208336e-05,
      "loss": 1.7484,
      "step": 13680
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.375,
      "learning_rate": 1.4839295683380785e-05,
      "loss": 1.7345,
      "step": 13824
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.515625,
      "learning_rate": 1.4834359484167429e-05,
      "loss": 1.7413,
      "step": 13968
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.6875,
      "learning_rate": 1.4829349463252034e-05,
      "loss": 1.7516,
      "step": 14112
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.96875,
      "learning_rate": 1.48242656710614e-05,
      "loss": 1.7295,
      "step": 14256
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.328125,
      "learning_rate": 1.4819108158764847e-05,
      "loss": 1.7659,
      "step": 14400
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.546875,
      "learning_rate": 1.4813876978273708e-05,
      "loss": 1.7535,
      "step": 14544
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.484375,
      "learning_rate": 1.4808572182240786e-05,
      "loss": 1.7379,
      "step": 14688
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.515625,
      "learning_rate": 1.4803193824059852e-05,
      "loss": 1.7487,
      "step": 14832
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.5625,
      "learning_rate": 1.4797741957865087e-05,
      "loss": 1.7694,
      "step": 14976
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.484375,
      "learning_rate": 1.4792216638530545e-05,
      "loss": 1.7527,
      "step": 15120
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.4375,
      "learning_rate": 1.4786617921669603e-05,
      "loss": 1.7529,
      "step": 15264
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.453125,
      "learning_rate": 1.478094586363439e-05,
      "loss": 1.7451,
      "step": 15408
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.453125,
      "learning_rate": 1.4775200521515237e-05,
      "loss": 1.7626,
      "step": 15552
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.453125,
      "learning_rate": 1.4769381953140084e-05,
      "loss": 1.7435,
      "step": 15696
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.53125,
      "learning_rate": 1.4763490217073917e-05,
      "loss": 1.7327,
      "step": 15840
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.453125,
      "learning_rate": 1.475752537261816e-05,
      "loss": 1.7187,
      "step": 15984
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.65625,
      "learning_rate": 1.475148747981009e-05,
      "loss": 1.7366,
      "step": 16128
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.46875,
      "learning_rate": 1.4745376599422232e-05,
      "loss": 1.7298,
      "step": 16272
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.578125,
      "learning_rate": 1.4739192792961736e-05,
      "loss": 1.7565,
      "step": 16416
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.5,
      "learning_rate": 1.4732936122669777e-05,
      "loss": 1.7442,
      "step": 16560
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.609375,
      "learning_rate": 1.4726606651520911e-05,
      "loss": 1.7295,
      "step": 16704
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.46875,
      "learning_rate": 1.4720204443222452e-05,
      "loss": 1.7239,
      "step": 16848
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.53125,
      "learning_rate": 1.4713729562213825e-05,
      "loss": 1.75,
      "step": 16992
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.59375,
      "learning_rate": 1.4707182073665921e-05,
      "loss": 1.7387,
      "step": 17136
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.375,
      "learning_rate": 1.470056204348044e-05,
      "loss": 1.7465,
      "step": 17280
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.609375,
      "learning_rate": 1.4693869538289221e-05,
      "loss": 1.741,
      "step": 17424
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.484375,
      "learning_rate": 1.4687104625453587e-05,
      "loss": 1.7203,
      "step": 17568
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.734375,
      "learning_rate": 1.4680267373063651e-05,
      "loss": 1.7498,
      "step": 17712
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.46875,
      "learning_rate": 1.467335784993764e-05,
      "loss": 1.7519,
      "step": 17856
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.765625,
      "learning_rate": 1.4666376125621198e-05,
      "loss": 1.7386,
      "step": 18000
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.5625,
      "learning_rate": 1.4659322270386687e-05,
      "loss": 1.7228,
      "step": 18144
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.671875,
      "learning_rate": 1.4652196355232481e-05,
      "loss": 1.723,
      "step": 18288
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.546875,
      "learning_rate": 1.4644998451882254e-05,
      "loss": 1.7334,
      "step": 18432
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.5625,
      "learning_rate": 1.463772863278425e-05,
      "loss": 1.7269,
      "step": 18576
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.515625,
      "learning_rate": 1.4630386971110556e-05,
      "loss": 1.7646,
      "step": 18720
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.46875,
      "learning_rate": 1.4622973540756377e-05,
      "loss": 1.7391,
      "step": 18864
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.125,
      "learning_rate": 1.4615488416339277e-05,
      "loss": 1.7336,
      "step": 19008
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.5,
      "learning_rate": 1.4607931673198435e-05,
      "loss": 1.7431,
      "step": 19152
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.734375,
      "learning_rate": 1.4600303387393886e-05,
      "loss": 1.7331,
      "step": 19296
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.828125,
      "learning_rate": 1.4592603635705753e-05,
      "loss": 1.7113,
      "step": 19440
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.8125,
      "learning_rate": 1.4584832495633486e-05,
      "loss": 1.7154,
      "step": 19584
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.578125,
      "learning_rate": 1.4576990045395056e-05,
      "loss": 1.7258,
      "step": 19728
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.390625,
      "learning_rate": 1.4569076363926202e-05,
      "loss": 1.7353,
      "step": 19872
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.59375,
      "learning_rate": 1.4561091530879606e-05,
      "loss": 1.7143,
      "step": 20016
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.671875,
      "learning_rate": 1.455303562662411e-05,
      "loss": 1.7556,
      "step": 20160
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.578125,
      "learning_rate": 1.4544908732243898e-05,
      "loss": 1.7028,
      "step": 20304
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.75,
      "learning_rate": 1.4536710929537685e-05,
      "loss": 1.7214,
      "step": 20448
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.8125,
      "learning_rate": 1.4528442301017889e-05,
      "loss": 1.7206,
      "step": 20592
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.4375,
      "learning_rate": 1.4520102929909807e-05,
      "loss": 1.7248,
      "step": 20736
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.671875,
      "learning_rate": 1.4511692900150771e-05,
      "loss": 1.7353,
      "step": 20880
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.453125,
      "learning_rate": 1.4503212296389307e-05,
      "loss": 1.7482,
      "step": 21024
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.59375,
      "learning_rate": 1.4494661203984277e-05,
      "loss": 1.7368,
      "step": 21168
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.703125,
      "learning_rate": 1.4486039709004033e-05,
      "loss": 1.7314,
      "step": 21312
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.0,
      "learning_rate": 1.447734789822553e-05,
      "loss": 1.727,
      "step": 21456
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.796875,
      "learning_rate": 1.4468585859133475e-05,
      "loss": 1.7388,
      "step": 21600
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.484375,
      "learning_rate": 1.4459753679919435e-05,
      "loss": 1.7324,
      "step": 21744
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.453125,
      "learning_rate": 1.4450851449480947e-05,
      "loss": 1.725,
      "step": 21888
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.578125,
      "learning_rate": 1.4441879257420625e-05,
      "loss": 1.7084,
      "step": 22032
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.484375,
      "learning_rate": 1.4432837194045265e-05,
      "loss": 1.7264,
      "step": 22176
    },
    {
      "epoch": 0.15,
      "grad_norm": 3.546875,
      "learning_rate": 1.4423725350364928e-05,
      "loss": 1.7102,
      "step": 22320
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.5,
      "learning_rate": 1.441454381809203e-05,
      "loss": 1.7565,
      "step": 22464
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.625,
      "learning_rate": 1.440529268964041e-05,
      "loss": 1.7304,
      "step": 22608
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.640625,
      "learning_rate": 1.439597205812441e-05,
      "loss": 1.7348,
      "step": 22752
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.5625,
      "learning_rate": 1.438658201735793e-05,
      "loss": 1.7227,
      "step": 22896
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.578125,
      "learning_rate": 1.4377122661853493e-05,
      "loss": 1.6986,
      "step": 23040
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.375,
      "learning_rate": 1.436759408682128e-05,
      "loss": 1.7251,
      "step": 23184
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.53125,
      "learning_rate": 1.4357996388168187e-05,
      "loss": 1.7317,
      "step": 23328
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.484375,
      "learning_rate": 1.4348329662496846e-05,
      "loss": 1.7185,
      "step": 23472
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.546875,
      "learning_rate": 1.4338594007104664e-05,
      "loss": 1.7359,
      "step": 23616
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.34375,
      "learning_rate": 1.4328789519982834e-05,
      "loss": 1.7192,
      "step": 23760
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.609375,
      "learning_rate": 1.4318916299815355e-05,
      "loss": 1.7156,
      "step": 23904
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.421875,
      "learning_rate": 1.4308974445978041e-05,
      "loss": 1.7172,
      "step": 24048
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.359375,
      "learning_rate": 1.4298964058537512e-05,
      "loss": 1.7061,
      "step": 24192
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.484375,
      "learning_rate": 1.4288885238250189e-05,
      "loss": 1.7198,
      "step": 24336
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.421875,
      "learning_rate": 1.4278738086561289e-05,
      "loss": 1.7154,
      "step": 24480
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.515625,
      "learning_rate": 1.4268522705603794e-05,
      "loss": 1.7287,
      "step": 24624
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.890625,
      "learning_rate": 1.4258239198197428e-05,
      "loss": 1.6973,
      "step": 24768
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.609375,
      "learning_rate": 1.4247887667847619e-05,
      "loss": 1.7118,
      "step": 24912
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.484375,
      "learning_rate": 1.4237468218744457e-05,
      "loss": 1.732,
      "step": 25056
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.65625,
      "learning_rate": 1.4226980955761649e-05,
      "loss": 1.7271,
      "step": 25200
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.484375,
      "learning_rate": 1.4216425984455464e-05,
      "loss": 1.7145,
      "step": 25344
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.359375,
      "learning_rate": 1.4205803411063663e-05,
      "loss": 1.7272,
      "step": 25488
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.0,
      "learning_rate": 1.4195113342504438e-05,
      "loss": 1.6899,
      "step": 25632
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.59375,
      "learning_rate": 1.4184355886375332e-05,
      "loss": 1.7089,
      "step": 25776
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.4375,
      "learning_rate": 1.4173531150952153e-05,
      "loss": 1.7156,
      "step": 25920
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.5625,
      "learning_rate": 1.4162639245187894e-05,
      "loss": 1.7069,
      "step": 26064
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.578125,
      "learning_rate": 1.4151680278711626e-05,
      "loss": 1.7139,
      "step": 26208
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.546875,
      "learning_rate": 1.4140654361827399e-05,
      "loss": 1.7375,
      "step": 26352
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.8125,
      "learning_rate": 1.412956160551313e-05,
      "loss": 1.7372,
      "step": 26496
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.859375,
      "learning_rate": 1.411840212141949e-05,
      "loss": 1.7045,
      "step": 26640
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.3125,
      "learning_rate": 1.410717602186878e-05,
      "loss": 1.7077,
      "step": 26784
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.546875,
      "learning_rate": 1.4095883419853792e-05,
      "loss": 1.7075,
      "step": 26928
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.34375,
      "learning_rate": 1.408452442903668e-05,
      "loss": 1.7091,
      "step": 27072
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.5,
      "learning_rate": 1.4073099163747816e-05,
      "loss": 1.7108,
      "step": 27216
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.90625,
      "learning_rate": 1.406160773898463e-05,
      "loss": 1.7036,
      "step": 27360
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.125,
      "learning_rate": 1.4050050270410474e-05,
      "loss": 1.7362,
      "step": 27504
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.578125,
      "learning_rate": 1.4038426874353428e-05,
      "loss": 1.6984,
      "step": 27648
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.515625,
      "learning_rate": 1.402673766780515e-05,
      "loss": 1.7191,
      "step": 27792
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.140625,
      "learning_rate": 1.4014982768419697e-05,
      "loss": 1.7401,
      "step": 27936
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.90625,
      "learning_rate": 1.400316229451233e-05,
      "loss": 1.7152,
      "step": 28080
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.671875,
      "learning_rate": 1.3991276365058336e-05,
      "loss": 1.7208,
      "step": 28224
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.421875,
      "learning_rate": 1.3979325099691819e-05,
      "loss": 1.707,
      "step": 28368
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.734375,
      "learning_rate": 1.3967308618704503e-05,
      "loss": 1.7274,
      "step": 28512
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.0,
      "learning_rate": 1.3955227043044524e-05,
      "loss": 1.7213,
      "step": 28656
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.484375,
      "learning_rate": 1.3943080494315198e-05,
      "loss": 1.702,
      "step": 28800
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.625,
      "learning_rate": 1.393086909477382e-05,
      "loss": 1.706,
      "step": 28944
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.46875,
      "learning_rate": 1.3918592967330407e-05,
      "loss": 1.7095,
      "step": 29088
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.328125,
      "learning_rate": 1.3906252235546486e-05,
      "loss": 1.6814,
      "step": 29232
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.40625,
      "learning_rate": 1.3893847023633831e-05,
      "loss": 1.7225,
      "step": 29376
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.671875,
      "learning_rate": 1.3881377456453227e-05,
      "loss": 1.7103,
      "step": 29520
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.78125,
      "learning_rate": 1.3868843659513198e-05,
      "loss": 1.6974,
      "step": 29664
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.75,
      "learning_rate": 1.3856245758968766e-05,
      "loss": 1.7212,
      "step": 29808
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.484375,
      "learning_rate": 1.3843583881620151e-05,
      "loss": 1.727,
      "step": 29952
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.609375,
      "learning_rate": 1.3830858154911528e-05,
      "loss": 1.7035,
      "step": 30096
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.5625,
      "learning_rate": 1.3818068706929713e-05,
      "loss": 1.7041,
      "step": 30240
    }
  ],
  "logging_steps": 144,
  "max_steps": 147005,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1440,
  "total_flos": 1.2669053109629092e+20,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}