{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.32533550223668156,
  "global_step": 1000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 4.999967339434433e-05,
      "loss": 1.8952,
      "step": 10
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999869358591101e-05,
      "loss": 1.7279,
      "step": 20
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999706060030093e-05,
      "loss": 1.6717,
      "step": 30
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9994774480181473e-05,
      "loss": 1.64,
      "step": 40
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.999183528528541e-05,
      "loss": 1.5903,
      "step": 50
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.998824309240937e-05,
      "loss": 1.5936,
      "step": 60
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.998399799541179e-05,
      "loss": 1.5633,
      "step": 70
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9979100105210475e-05,
      "loss": 1.5244,
      "step": 80
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.997354954977972e-05,
      "loss": 1.5517,
      "step": 90
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.996734647414696e-05,
      "loss": 1.5226,
      "step": 100
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.996049104038895e-05,
      "loss": 1.5107,
      "step": 110
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9952983427627566e-05,
      "loss": 1.5111,
      "step": 120
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9944823832025115e-05,
      "loss": 1.5107,
      "step": 130
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.99360124667792e-05,
      "loss": 1.5035,
      "step": 140
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9926549562117164e-05,
      "loss": 1.4973,
      "step": 150
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.991643536529006e-05,
      "loss": 1.4968,
      "step": 160
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.99056701405662e-05,
      "loss": 1.479,
      "step": 170
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.989425416922423e-05,
      "loss": 1.4879,
      "step": 180
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.988218774954583e-05,
      "loss": 1.4943,
      "step": 190
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.986947119680788e-05,
      "loss": 1.4952,
      "step": 200
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.985610484327421e-05,
      "loss": 1.4588,
      "step": 210
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9842089038186954e-05,
      "loss": 1.4869,
      "step": 220
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9827424147757405e-05,
      "loss": 1.4762,
      "step": 230
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.981211055515648e-05,
      "loss": 1.4909,
      "step": 240
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9796148660504624e-05,
      "loss": 1.4659,
      "step": 250
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9779538880861455e-05,
      "loss": 1.4683,
      "step": 260
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.976228165021482e-05,
      "loss": 1.4581,
      "step": 270
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9744377419469426e-05,
      "loss": 1.4481,
      "step": 280
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.972582665643514e-05,
      "loss": 1.4571,
      "step": 290
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.970662984581468e-05,
      "loss": 1.4683,
      "step": 300
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.9686787489191e-05,
      "loss": 1.4528,
      "step": 310
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.966630010501417e-05,
      "loss": 1.4548,
      "step": 320
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.964516822858784e-05,
      "loss": 1.4487,
      "step": 330
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.962339241205523e-05,
      "loss": 1.4468,
      "step": 340
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.960097322438474e-05,
      "loss": 1.468,
      "step": 350
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.957791125135503e-05,
      "loss": 1.4601,
      "step": 360
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.955420709553979e-05,
      "loss": 1.4604,
      "step": 370
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.9529861376291895e-05,
      "loss": 1.4528,
      "step": 380
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.9504874729727344e-05,
      "loss": 1.4478,
      "step": 390
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.947924780870853e-05,
      "loss": 1.4697,
      "step": 400
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.945298128282724e-05,
      "loss": 1.45,
      "step": 410
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.942607583838715e-05,
      "loss": 1.4368,
      "step": 420
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.939853217838588e-05,
      "loss": 1.4279,
      "step": 430
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.937035102249665e-05,
      "loss": 1.4384,
      "step": 440
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.934153310704945e-05,
      "loss": 1.4543,
      "step": 450
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.931207918501181e-05,
      "loss": 1.4581,
      "step": 460
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.9281990025969135e-05,
      "loss": 1.4503,
      "step": 470
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.925126641610458e-05,
      "loss": 1.4333,
      "step": 480
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.921990915817852e-05,
      "loss": 1.4506,
      "step": 490
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.91879190715076e-05,
      "loss": 1.4368,
      "step": 500
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.915529699194326e-05,
      "loss": 1.4428,
      "step": 510
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.9122043771849957e-05,
      "loss": 1.4521,
      "step": 520
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.9088160280082875e-05,
      "loss": 1.4296,
      "step": 530
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.905364740196521e-05,
      "loss": 1.4456,
      "step": 540
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.901850603926507e-05,
      "loss": 1.425,
      "step": 550
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.898273711017187e-05,
      "loss": 1.4604,
      "step": 560
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.894634154927238e-05,
      "loss": 1.4328,
      "step": 570
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.890932030752627e-05,
      "loss": 1.4232,
      "step": 580
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.887167435224131e-05,
      "loss": 1.4344,
      "step": 590
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.8833404667048045e-05,
      "loss": 1.4459,
      "step": 600
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.8794512251874125e-05,
      "loss": 1.4308,
      "step": 610
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.875499812291817e-05,
      "loss": 1.442,
      "step": 620
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.871486331262322e-05,
      "loss": 1.4297,
      "step": 630
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.867410886964976e-05,
      "loss": 1.442,
      "step": 640
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.863273585884831e-05,
      "loss": 1.436,
      "step": 650
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.8590745361231626e-05,
      "loss": 1.4153,
      "step": 660
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.8548138473946424e-05,
      "loss": 1.4344,
      "step": 670
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.850491631024473e-05,
      "loss": 1.4287,
      "step": 680
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.846107999945479e-05,
      "loss": 1.4492,
      "step": 690
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.841663068695157e-05,
      "loss": 1.4636,
      "step": 700
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.837156953412682e-05,
      "loss": 1.4465,
      "step": 710
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.8325897718358725e-05,
      "loss": 1.4127,
      "step": 720
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.8279616432981154e-05,
      "loss": 1.4162,
      "step": 730
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.823272688725248e-05,
      "loss": 1.4313,
      "step": 740
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.818523030632396e-05,
      "loss": 1.4191,
      "step": 750
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.8137127931207745e-05,
      "loss": 1.4212,
      "step": 760
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.808842101874447e-05,
      "loss": 1.4179,
      "step": 770
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.803911084157037e-05,
      "loss": 1.4353,
      "step": 780
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.798919868808406e-05,
      "loss": 1.4294,
      "step": 790
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.793868586241289e-05,
      "loss": 1.437,
      "step": 800
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.7887573684378815e-05,
      "loss": 1.4265,
      "step": 810
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.783586348946393e-05,
      "loss": 1.4121,
      "step": 820
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.778355662877564e-05,
      "loss": 1.414,
      "step": 830
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.773065446901123e-05,
      "loss": 1.4205,
      "step": 840
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.76771583924223e-05,
      "loss": 1.4289,
      "step": 850
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.7623069796778506e-05,
      "loss": 1.4088,
      "step": 860
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.756839009533118e-05,
      "loss": 1.4206,
      "step": 870
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.7513120716776294e-05,
      "loss": 1.4196,
      "step": 880
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.745726310521717e-05,
      "loss": 1.4409,
      "step": 890
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.740081872012676e-05,
      "loss": 1.4221,
      "step": 900
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.734378903630949e-05,
      "loss": 1.4097,
      "step": 910
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.728617554386276e-05,
      "loss": 1.4256,
      "step": 920
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.722797974813795e-05,
      "loss": 1.4139,
      "step": 930
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.716920316970115e-05,
      "loss": 1.4258,
      "step": 940
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.710984734429339e-05,
      "loss": 1.4362,
      "step": 950
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.7049913822790526e-05,
      "loss": 1.4329,
      "step": 960
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.6989404171162745e-05,
      "loss": 1.4058,
      "step": 970
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.692831997043359e-05,
      "loss": 1.4412,
      "step": 980
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.6866662816638694e-05,
      "loss": 1.4148,
      "step": 990
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.680443432078407e-05,
      "loss": 1.4253,
      "step": 1000
    }
  ],
  "max_steps": 6146,
  "num_train_epochs": 2,
  "total_flos": 1.7543484519040942e+18,
  "trial_name": null,
  "trial_params": null
}