{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 1000,
  "global_step": 1972,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 3.0303030303030305e-07,
      "loss": 1.0924,
      "step": 1
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.0303030303030305e-06,
      "loss": 0.8661,
      "step": 10
    },
    {
      "epoch": 0.02,
      "learning_rate": 6.060606060606061e-06,
      "loss": 0.8932,
      "step": 20
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.090909090909091e-06,
      "loss": 0.8119,
      "step": 30
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.2121212121212122e-05,
      "loss": 0.7431,
      "step": 40
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.5151515151515153e-05,
      "loss": 0.6742,
      "step": 50
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.8181818181818182e-05,
      "loss": 0.6329,
      "step": 60
    },
    {
      "epoch": 0.07,
      "learning_rate": 2.121212121212121e-05,
      "loss": 0.5587,
      "step": 70
    },
    {
      "epoch": 0.08,
      "learning_rate": 2.4242424242424244e-05,
      "loss": 0.5006,
      "step": 80
    },
    {
      "epoch": 0.09,
      "learning_rate": 2.7272727272727273e-05,
      "loss": 0.5024,
      "step": 90
    },
    {
      "epoch": 0.1,
      "learning_rate": 2.9999978899859252e-05,
      "loss": 0.4811,
      "step": 100
    },
    {
      "epoch": 0.11,
      "learning_rate": 2.999744695479716e-05,
      "loss": 0.4825,
      "step": 110
    },
    {
      "epoch": 0.12,
      "learning_rate": 2.9990695797777382e-05,
      "loss": 0.4656,
      "step": 120
    },
    {
      "epoch": 0.13,
      "learning_rate": 2.9979727328094035e-05,
      "loss": 0.4804,
      "step": 130
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.9964544631492205e-05,
      "loss": 0.4627,
      "step": 140
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.9945151979299888e-05,
      "loss": 0.4164,
      "step": 150
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.9921554827226334e-05,
      "loss": 0.426,
      "step": 160
    },
    {
      "epoch": 0.17,
      "learning_rate": 2.9893759813827195e-05,
      "loss": 0.4441,
      "step": 170
    },
    {
      "epoch": 0.18,
      "learning_rate": 2.98617747586369e-05,
      "loss": 0.41,
      "step": 180
    },
    {
      "epoch": 0.19,
      "learning_rate": 2.9825608659968814e-05,
      "loss": 0.3916,
      "step": 190
    },
    {
      "epoch": 0.2,
      "learning_rate": 2.9785271692383737e-05,
      "loss": 0.4281,
      "step": 200
    },
    {
      "epoch": 0.21,
      "learning_rate": 2.9740775203827525e-05,
      "loss": 0.4219,
      "step": 210
    },
    {
      "epoch": 0.22,
      "learning_rate": 2.9692131712438576e-05,
      "loss": 0.4138,
      "step": 220
    },
    {
      "epoch": 0.23,
      "learning_rate": 2.963935490302613e-05,
      "loss": 0.392,
      "step": 230
    },
    {
      "epoch": 0.24,
      "learning_rate": 2.9582459623220305e-05,
      "loss": 0.4244,
      "step": 240
    },
    {
      "epoch": 0.25,
      "learning_rate": 2.9521461879295086e-05,
      "loss": 0.3797,
      "step": 250
    },
    {
      "epoch": 0.26,
      "learning_rate": 2.9456378831665264e-05,
      "loss": 0.3956,
      "step": 260
    },
    {
      "epoch": 0.27,
      "learning_rate": 2.938722879005873e-05,
      "loss": 0.4031,
      "step": 270
    },
    {
      "epoch": 0.28,
      "learning_rate": 2.9314031208365425e-05,
      "loss": 0.4382,
      "step": 280
    },
    {
      "epoch": 0.29,
      "learning_rate": 2.923680667916439e-05,
      "loss": 0.4096,
      "step": 290
    },
    {
      "epoch": 0.3,
      "learning_rate": 2.9155576927930516e-05,
      "loss": 0.3965,
      "step": 300
    },
    {
      "epoch": 0.31,
      "learning_rate": 2.9070364806922495e-05,
      "loss": 0.3842,
      "step": 310
    },
    {
      "epoch": 0.32,
      "learning_rate": 2.8981194288753864e-05,
      "loss": 0.4057,
      "step": 320
    },
    {
      "epoch": 0.33,
      "learning_rate": 2.888809045964882e-05,
      "loss": 0.407,
      "step": 330
    },
    {
      "epoch": 0.34,
      "learning_rate": 2.8791079512384716e-05,
      "loss": 0.4048,
      "step": 340
    },
    {
      "epoch": 0.35,
      "learning_rate": 2.869018873892331e-05,
      "loss": 0.4119,
      "step": 350
    },
    {
      "epoch": 0.37,
      "learning_rate": 2.8585446522732738e-05,
      "loss": 0.379,
      "step": 360
    },
    {
      "epoch": 0.38,
      "learning_rate": 2.8476882330802442e-05,
      "loss": 0.4007,
      "step": 370
    },
    {
      "epoch": 0.39,
      "learning_rate": 2.8364526705353253e-05,
      "loss": 0.3841,
      "step": 380
    },
    {
      "epoch": 0.4,
      "learning_rate": 2.8248411255244985e-05,
      "loss": 0.3977,
      "step": 390
    },
    {
      "epoch": 0.41,
      "learning_rate": 2.8128568647083962e-05,
      "loss": 0.3957,
      "step": 400
    },
    {
      "epoch": 0.42,
      "learning_rate": 2.8005032596032966e-05,
      "loss": 0.4228,
      "step": 410
    },
    {
      "epoch": 0.43,
      "learning_rate": 2.7877837856326193e-05,
      "loss": 0.3848,
      "step": 420
    },
    {
      "epoch": 0.44,
      "learning_rate": 2.7747020211491894e-05,
      "loss": 0.3618,
      "step": 430
    },
    {
      "epoch": 0.45,
      "learning_rate": 2.761261646428543e-05,
      "loss": 0.4117,
      "step": 440
    },
    {
      "epoch": 0.46,
      "learning_rate": 2.747466442633561e-05,
      "loss": 0.396,
      "step": 450
    },
    {
      "epoch": 0.47,
      "learning_rate": 2.7333202907507196e-05,
      "loss": 0.4077,
      "step": 460
    },
    {
      "epoch": 0.48,
      "learning_rate": 2.7188271704982577e-05,
      "loss": 0.4131,
      "step": 470
    },
    {
      "epoch": 0.49,
      "learning_rate": 2.7039911592065674e-05,
      "loss": 0.3943,
      "step": 480
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.688816430671124e-05,
      "loss": 0.4165,
      "step": 490
    },
    {
      "epoch": 0.51,
      "learning_rate": 2.6733072539782788e-05,
      "loss": 0.3639,
      "step": 500
    },
    {
      "epoch": 0.52,
      "learning_rate": 2.6574679923042412e-05,
      "loss": 0.4091,
      "step": 510
    },
    {
      "epoch": 0.53,
      "learning_rate": 2.641303101687593e-05,
      "loss": 0.3809,
      "step": 520
    },
    {
      "epoch": 0.54,
      "learning_rate": 2.6248171297756755e-05,
      "loss": 0.3801,
      "step": 530
    },
    {
      "epoch": 0.55,
      "learning_rate": 2.6080147145452088e-05,
      "loss": 0.3986,
      "step": 540
    },
    {
      "epoch": 0.56,
      "learning_rate": 2.590900582997493e-05,
      "loss": 0.4076,
      "step": 550
    },
    {
      "epoch": 0.57,
      "learning_rate": 2.5734795498285684e-05,
      "loss": 0.3971,
      "step": 560
    },
    {
      "epoch": 0.58,
      "learning_rate": 2.555756516074704e-05,
      "loss": 0.3847,
      "step": 570
    },
    {
      "epoch": 0.59,
      "learning_rate": 2.5377364677335944e-05,
      "loss": 0.3839,
      "step": 580
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.5194244743616557e-05,
      "loss": 0.3845,
      "step": 590
    },
    {
      "epoch": 0.61,
      "learning_rate": 2.5008256876478146e-05,
      "loss": 0.3907,
      "step": 600
    },
    {
      "epoch": 0.62,
      "learning_rate": 2.481945339964189e-05,
      "loss": 0.4105,
      "step": 610
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.4627887428940757e-05,
      "loss": 0.3809,
      "step": 620
    },
    {
      "epoch": 0.64,
      "learning_rate": 2.4433612857376438e-05,
      "loss": 0.3887,
      "step": 630
    },
    {
      "epoch": 0.65,
      "learning_rate": 2.4236684339957745e-05,
      "loss": 0.4146,
      "step": 640
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.4037157278324564e-05,
      "loss": 0.3752,
      "step": 650
    },
    {
      "epoch": 0.67,
      "learning_rate": 2.383508780516181e-05,
      "loss": 0.3826,
      "step": 660
    },
    {
      "epoch": 0.68,
      "learning_rate": 2.3630532768407667e-05,
      "loss": 0.3884,
      "step": 670
    },
    {
      "epoch": 0.69,
      "learning_rate": 2.3423549715260643e-05,
      "loss": 0.3926,
      "step": 680
    },
    {
      "epoch": 0.7,
      "learning_rate": 2.3214196875989905e-05,
      "loss": 0.3777,
      "step": 690
    },
    {
      "epoch": 0.71,
      "learning_rate": 2.300253314755341e-05,
      "loss": 0.3966,
      "step": 700
    },
    {
      "epoch": 0.72,
      "learning_rate": 2.2788618077028512e-05,
      "loss": 0.3884,
      "step": 710
    },
    {
      "epoch": 0.73,
      "learning_rate": 2.2572511844859682e-05,
      "loss": 0.3736,
      "step": 720
    },
    {
      "epoch": 0.74,
      "learning_rate": 2.2354275247927982e-05,
      "loss": 0.3989,
      "step": 730
    },
    {
      "epoch": 0.75,
      "learning_rate": 2.2133969682447195e-05,
      "loss": 0.394,
      "step": 740
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.1911657126691292e-05,
      "loss": 0.3795,
      "step": 750
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.1687400123558168e-05,
      "loss": 0.3941,
      "step": 760
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.146126176297453e-05,
      "loss": 0.3936,
      "step": 770
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.12333056641469e-05,
      "loss": 0.405,
      "step": 780
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.1003595957663693e-05,
      "loss": 0.3903,
      "step": 790
    },
    {
      "epoch": 0.81,
      "learning_rate": 2.077219726745346e-05,
      "loss": 0.3823,
      "step": 800
    },
    {
      "epoch": 0.82,
      "learning_rate": 2.053917469260431e-05,
      "loss": 0.3883,
      "step": 810
    },
    {
      "epoch": 0.83,
      "learning_rate": 2.0304593789049672e-05,
      "loss": 0.3923,
      "step": 820
    },
    {
      "epoch": 0.84,
      "learning_rate": 2.0068520551125535e-05,
      "loss": 0.3741,
      "step": 830
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.9831021393004352e-05,
      "loss": 0.3735,
      "step": 840
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.959216313001081e-05,
      "loss": 0.3828,
      "step": 850
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.9352012959824793e-05,
      "loss": 0.3936,
      "step": 860
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.9110638443576718e-05,
      "loss": 0.3533,
      "step": 870
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.886810748684066e-05,
      "loss": 0.3512,
      "step": 880
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.862448832053059e-05,
      "loss": 0.3703,
      "step": 890
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.837984948170502e-05,
      "loss": 0.3976,
      "step": 900
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.8134259794285595e-05,
      "loss": 0.3846,
      "step": 910
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.7887788349694946e-05,
      "loss": 0.3679,
      "step": 920
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.7640504487419268e-05,
      "loss": 0.3693,
      "step": 930
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.7392477775501178e-05,
      "loss": 0.3825,
      "step": 940
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.714377799096817e-05,
      "loss": 0.3948,
      "step": 950
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.6894475100202392e-05,
      "loss": 0.4137,
      "step": 960
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.6644639239257066e-05,
      "loss": 0.4053,
      "step": 970
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.6394340694125204e-05,
      "loss": 0.3788,
      "step": 980
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.6143649880966164e-05,
      "loss": 0.4028,
      "step": 990
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.5892637326295542e-05,
      "loss": 0.3858,
      "step": 1000
    },
    {
      "epoch": 1.01,
      "eval_loss": 0.4780150055885315,
      "eval_runtime": 11.6787,
      "eval_samples_per_second": 3.853,
      "eval_steps_per_second": 1.028,
      "step": 1000
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.5641373647144043e-05,
      "loss": 0.3697,
      "step": 1010
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.538992953119089e-05,
      "loss": 0.3503,
      "step": 1020
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.5138375716877379e-05,
      "loss": 0.3831,
      "step": 1030
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.4886782973506064e-05,
      "loss": 0.3854,
      "step": 1040
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.463522208133139e-05,
      "loss": 0.3572,
      "step": 1050
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.4383763811647126e-05,
      "loss": 0.3455,
      "step": 1060
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.4132478906876406e-05,
      "loss": 0.3861,
      "step": 1070
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.3881438060669832e-05,
      "loss": 0.3825,
      "step": 1080
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.3630711898017336e-05,
      "loss": 0.3487,
      "step": 1090
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.3380370955379331e-05,
      "loss": 0.3885,
      "step": 1100
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.3130485660842818e-05,
      "loss": 0.3644,
      "step": 1110
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.2881126314307895e-05,
      "loss": 0.3769,
      "step": 1120
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.2632363067710408e-05,
      "loss": 0.3709,
      "step": 1130
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.2384265905286195e-05,
      "loss": 0.3795,
      "step": 1140
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.213690462388252e-05,
      "loss": 0.3525,
      "step": 1150
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.1890348813322198e-05,
      "loss": 0.3664,
      "step": 1160
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.1644667836826004e-05,
      "loss": 0.374,
      "step": 1170
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.1399930811498802e-05,
      "loss": 0.3554,
      "step": 1180
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.115620658888491e-05,
      "loss": 0.3747,
      "step": 1190
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.0913563735598206e-05,
      "loss": 0.3676,
      "step": 1200
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.0672070514032354e-05,
      "loss": 0.3478,
      "step": 1210
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.0431794863156649e-05,
      "loss": 0.3633,
      "step": 1220
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.0192804379402823e-05,
      "loss": 0.3591,
      "step": 1230
    },
    {
      "epoch": 1.26,
      "learning_rate": 9.955166297648241e-06,
      "loss": 0.3727,
      "step": 1240
    },
    {
      "epoch": 1.27,
      "learning_rate": 9.718947472300812e-06,
      "loss": 0.3669,
      "step": 1250
    },
    {
      "epoch": 1.28,
      "learning_rate": 9.484214358490896e-06,
      "loss": 0.3653,
      "step": 1260
    },
    {
      "epoch": 1.29,
      "learning_rate": 9.251032993375591e-06,
      "loss": 0.3713,
      "step": 1270
    },
    {
      "epoch": 1.3,
      "learning_rate": 9.019468977560585e-06,
      "loss": 0.3622,
      "step": 1280
    },
    {
      "epoch": 1.31,
      "learning_rate": 8.789587456644819e-06,
      "loss": 0.3562,
      "step": 1290
    },
    {
      "epoch": 1.32,
      "learning_rate": 8.561453102893177e-06,
      "loss": 0.3619,
      "step": 1300
    },
    {
      "epoch": 1.33,
      "learning_rate": 8.335130097042284e-06,
      "loss": 0.4009,
      "step": 1310
    },
    {
      "epoch": 1.34,
      "learning_rate": 8.110682110244685e-06,
      "loss": 0.347,
      "step": 1320
    },
    {
      "epoch": 1.35,
      "learning_rate": 7.888172286156265e-06,
      "loss": 0.3891,
      "step": 1330
    },
    {
      "epoch": 1.36,
      "learning_rate": 7.667663223172156e-06,
      "loss": 0.3655,
      "step": 1340
    },
    {
      "epoch": 1.37,
      "learning_rate": 7.449216956816015e-06,
      "loss": 0.3552,
      "step": 1350
    },
    {
      "epoch": 1.38,
      "learning_rate": 7.232894942287638e-06,
      "loss": 0.3632,
      "step": 1360
    },
    {
      "epoch": 1.39,
      "learning_rate": 7.018758037173876e-06,
      "loss": 0.3519,
      "step": 1370
    },
    {
      "epoch": 1.4,
      "learning_rate": 6.806866484327612e-06,
      "loss": 0.375,
      "step": 1380
    },
    {
      "epoch": 1.41,
      "learning_rate": 6.597279894919741e-06,
      "loss": 0.3745,
      "step": 1390
    },
    {
      "epoch": 1.42,
      "learning_rate": 6.390057231668853e-06,
      "loss": 0.3651,
      "step": 1400
    },
    {
      "epoch": 1.43,
      "learning_rate": 6.185256792253298e-06,
      "loss": 0.3639,
      "step": 1410
    },
    {
      "epoch": 1.44,
      "learning_rate": 5.982936192910392e-06,
      "loss": 0.3725,
      "step": 1420
    },
    {
      "epoch": 1.45,
      "learning_rate": 5.78315235222732e-06,
      "loss": 0.3568,
      "step": 1430
    },
    {
      "epoch": 1.46,
      "learning_rate": 5.585961475128293e-06,
      "loss": 0.3373,
      "step": 1440
    },
    {
      "epoch": 1.47,
      "learning_rate": 5.39141903706252e-06,
      "loss": 0.3593,
      "step": 1450
    },
    {
      "epoch": 1.48,
      "learning_rate": 5.199579768397331e-06,
      "loss": 0.3451,
      "step": 1460
    },
    {
      "epoch": 1.49,
      "learning_rate": 5.010497639020985e-06,
      "loss": 0.3503,
      "step": 1470
    },
    {
      "epoch": 1.5,
      "learning_rate": 4.824225843159382e-06,
      "loss": 0.3575,
      "step": 1480
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.640816784411005e-06,
      "loss": 0.3542,
      "step": 1490
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.460322061004282e-06,
      "loss": 0.3447,
      "step": 1500
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.28279245128152e-06,
      "loss": 0.3791,
      "step": 1510
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.108277899413525e-06,
      "loss": 0.38,
      "step": 1520
    },
    {
      "epoch": 1.55,
      "learning_rate": 3.936827501348838e-06,
      "loss": 0.3568,
      "step": 1530
    },
    {
      "epoch": 1.56,
      "learning_rate": 3.76848949100166e-06,
      "loss": 0.384,
      "step": 1540
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.6033112266822636e-06,
      "loss": 0.3476,
      "step": 1550
    },
    {
      "epoch": 1.58,
      "learning_rate": 3.441339177773745e-06,
      "loss": 0.3807,
      "step": 1560
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.2826189116588445e-06,
      "loss": 0.3548,
      "step": 1570
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.1271950809005424e-06,
      "loss": 0.371,
      "step": 1580
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.9751114106800186e-06,
      "loss": 0.373,
      "step": 1590
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.82641068649549e-06,
      "loss": 0.3704,
      "step": 1600
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.6811347421254266e-06,
      "loss": 0.3719,
      "step": 1610
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.5393244478595223e-06,
      "loss": 0.3669,
      "step": 1620
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.401019699000693e-06,
      "loss": 0.3544,
      "step": 1630
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.266259404641401e-06,
      "loss": 0.3862,
      "step": 1640
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.135081476717396e-06,
      "loss": 0.3763,
      "step": 1650
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.007522819342005e-06,
      "loss": 0.3751,
      "step": 1660
    },
    {
      "epoch": 1.69,
      "learning_rate": 1.8836193184239536e-06,
      "loss": 0.3443,
      "step": 1670
    },
    {
      "epoch": 1.7,
      "learning_rate": 1.7634058315716257e-06,
      "loss": 0.3694,
      "step": 1680
    },
    {
      "epoch": 1.71,
      "learning_rate": 1.6469161782866115e-06,
      "loss": 0.3459,
      "step": 1690
    },
    {
      "epoch": 1.72,
      "learning_rate": 1.5341831304493258e-06,
      "loss": 0.3807,
      "step": 1700
    },
    {
      "epoch": 1.73,
      "learning_rate": 1.425238403099323e-06,
      "loss": 0.3747,
      "step": 1710
    },
    {
      "epoch": 1.74,
      "learning_rate": 1.3201126455129669e-06,
      "loss": 0.353,
      "step": 1720
    },
    {
      "epoch": 1.75,
      "learning_rate": 1.2188354325808853e-06,
      "loss": 0.3695,
      "step": 1730
    },
    {
      "epoch": 1.76,
      "learning_rate": 1.1214352564877266e-06,
      "loss": 0.3623,
      "step": 1740
    },
    {
      "epoch": 1.77,
      "learning_rate": 1.027939518696478e-06,
      "loss": 0.3616,
      "step": 1750
    },
    {
      "epoch": 1.78,
      "learning_rate": 9.383745222396545e-07,
      "loss": 0.366,
      "step": 1760
    },
    {
      "epoch": 1.8,
      "learning_rate": 8.527654643194959e-07,
      "loss": 0.3463,
      "step": 1770
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.711364292192768e-07,
      "loss": 0.3657,
      "step": 1780
    },
    {
      "epoch": 1.82,
      "learning_rate": 6.935103815277094e-07,
      "loss": 0.3636,
      "step": 1790
    },
    {
      "epoch": 1.83,
      "learning_rate": 6.199091596783352e-07,
      "loss": 0.3644,
      "step": 1800
    },
    {
      "epoch": 1.84,
      "learning_rate": 5.503534698057561e-07,
      "loss": 0.3631,
      "step": 1810
    },
    {
      "epoch": 1.85,
      "learning_rate": 4.848628799204024e-07,
      "loss": 0.3544,
      "step": 1820
    },
    {
      "epoch": 1.86,
      "learning_rate": 4.2345581440348834e-07,
      "loss": 0.354,
      "step": 1830
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.66149548823711e-07,
      "loss": 0.3841,
      "step": 1840
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.129602050771285e-07,
      "loss": 0.3435,
      "step": 1850
    },
    {
      "epoch": 1.89,
      "learning_rate": 2.639027468516181e-07,
      "loss": 0.3456,
      "step": 1860
    },
    {
      "epoch": 1.9,
      "learning_rate": 2.1899097541715485e-07,
      "loss": 0.3673,
      "step": 1870
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.782375257431196e-07,
      "loss": 0.3566,
      "step": 1880
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.4165386294372028e-07,
      "loss": 0.3729,
      "step": 1890
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.0925027905253072e-07,
      "loss": 0.3607,
      "step": 1900
    },
    {
      "epoch": 1.94,
      "learning_rate": 8.103589012703805e-08,
      "loss": 0.3696,
      "step": 1910
    },
    {
      "epoch": 1.95,
      "learning_rate": 5.701863368403948e-08,
      "loss": 0.3594,
      "step": 1920
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.720526646659339e-08,
      "loss": 0.3631,
      "step": 1930
    },
    {
      "epoch": 1.97,
      "learning_rate": 2.1601362543148263e-08,
      "loss": 0.3805,
      "step": 1940
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.0211311739400376e-08,
      "loss": 0.3676,
      "step": 1950
    },
    {
      "epoch": 1.99,
      "learning_rate": 3.038318403308371e-09,
      "loss": 0.3749,
      "step": 1960
    },
    {
      "epoch": 2.0,
      "learning_rate": 8.440050362101825e-11,
      "loss": 0.3555,
      "step": 1970
    },
    {
      "epoch": 2.0,
      "step": 1972,
      "total_flos": 3.329734212625367e+17,
      "train_loss": 0.3953790905872175,
      "train_runtime": 6822.6023,
      "train_samples_per_second": 1.155,
      "train_steps_per_second": 0.289
    }
  ],
  "logging_steps": 10,
  "max_steps": 1972,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 1000,
  "total_flos": 3.329734212625367e+17,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}