{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 724,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 4.999411623120412e-05,
      "loss": 2.1959,
      "step": 5
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.997646769431532e-05,
      "loss": 1.7787,
      "step": 10
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9947062696526445e-05,
      "loss": 1.167,
      "step": 15
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.990591507881416e-05,
      "loss": 2.0242,
      "step": 20
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9853044209423996e-05,
      "loss": 2.2836,
      "step": 25
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9788474974753686e-05,
      "loss": 1.9629,
      "step": 30
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.971223776763907e-05,
      "loss": 2.5583,
      "step": 35
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.962436847304818e-05,
      "loss": 1.9105,
      "step": 40
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.9524908451190096e-05,
      "loss": 2.1563,
      "step": 45
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.9413904518046674e-05,
      "loss": 2.3299,
      "step": 50
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.929140892333616e-05,
      "loss": 1.4083,
      "step": 55
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.9157479325919156e-05,
      "loss": 1.7837,
      "step": 60
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.901217876665858e-05,
      "loss": 1.4737,
      "step": 65
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.8855575638746135e-05,
      "loss": 2.1072,
      "step": 70
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.868774365550962e-05,
      "loss": 0.8345,
      "step": 75
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.850876181571592e-05,
      "loss": 2.4821,
      "step": 80
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.831871436638613e-05,
      "loss": 1.6394,
      "step": 85
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.811769076314044e-05,
      "loss": 1.6403,
      "step": 90
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.790578562809116e-05,
      "loss": 1.6487,
      "step": 95
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.7683098705304e-05,
      "loss": 1.751,
      "step": 100
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.7449734813848345e-05,
      "loss": 1.7423,
      "step": 105
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.720580379845883e-05,
      "loss": 1.9435,
      "step": 110
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.695142047783118e-05,
      "loss": 1.2112,
      "step": 115
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.668670459057692e-05,
      "loss": 1.5366,
      "step": 120
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.641178073886224e-05,
      "loss": 1.1842,
      "step": 125
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.6126778329757516e-05,
      "loss": 1.2011,
      "step": 130
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.583183151432527e-05,
      "loss": 1.2053,
      "step": 135
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.5527079124475045e-05,
      "loss": 1.4354,
      "step": 140
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.521266460761497e-05,
      "loss": 1.5641,
      "step": 145
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.488873595913091e-05,
      "loss": 1.7298,
      "step": 150
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.4555445652724795e-05,
      "loss": 1.2547,
      "step": 155
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.4212950568645007e-05,
      "loss": 1.6051,
      "step": 160
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.386141191984262e-05,
      "loss": 1.5891,
      "step": 165
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.350099517608823e-05,
      "loss": 1.2693,
      "step": 170
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.313186998608506e-05,
      "loss": 1.5962,
      "step": 175
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.275421009761509e-05,
      "loss": 1.5196,
      "step": 180
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.236819327575571e-05,
      "loss": 1.6221,
      "step": 185
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.197400121920539e-05,
      "loss": 1.1342,
      "step": 190
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.1571819474757894e-05,
      "loss": 2.024,
      "step": 195
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.116183734996509e-05,
      "loss": 1.3701,
      "step": 200
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.074424782402958e-05,
      "loss": 1.7944,
      "step": 205
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.031924745696915e-05,
      "loss": 1.561,
      "step": 210
    },
    {
      "epoch": 0.59,
      "learning_rate": 3.988703629709564e-05,
      "loss": 1.828,
      "step": 215
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.944781778685189e-05,
      "loss": 1.2159,
      "step": 220
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.900179866705112e-05,
      "loss": 1.6311,
      "step": 225
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.854918887956369e-05,
      "loss": 1.8962,
      "step": 230
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.809020146849714e-05,
      "loss": 1.3081,
      "step": 235
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.7625052479916015e-05,
      "loss": 1.6926,
      "step": 240
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.715396086014869e-05,
      "loss": 1.9407,
      "step": 245
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.667714835272895e-05,
      "loss": 1.3894,
      "step": 250
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.6194839394021e-05,
      "loss": 1.2711,
      "step": 255
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.570726100757693e-05,
      "loss": 2.0264,
      "step": 260
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.5214642697276426e-05,
      "loss": 2.0718,
      "step": 265
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.471721633929885e-05,
      "loss": 1.3508,
      "step": 270
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.421521607297888e-05,
      "loss": 1.4645,
      "step": 275
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.370887819059672e-05,
      "loss": 1.9456,
      "step": 280
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.319844102615497e-05,
      "loss": 1.6282,
      "step": 285
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.268414484319445e-05,
      "loss": 1.7652,
      "step": 290
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.216623172170183e-05,
      "loss": 1.6169,
      "step": 295
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.164494544416215e-05,
      "loss": 1.3647,
      "step": 300
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.11205313808101e-05,
      "loss": 2.0791,
      "step": 305
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.059323637413385e-05,
      "loss": 1.8486,
      "step": 310
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.0063308622685903e-05,
      "loss": 1.0746,
      "step": 315
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.9530997564255725e-05,
      "loss": 0.779,
      "step": 320
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.8996553758458916e-05,
      "loss": 1.7561,
      "step": 325
    },
    {
      "epoch": 0.91,
      "learning_rate": 2.8460228768798506e-05,
      "loss": 0.8527,
      "step": 330
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.792227504425359e-05,
      "loss": 1.211,
      "step": 335
    },
    {
      "epoch": 0.94,
      "learning_rate": 2.738294580045119e-05,
      "loss": 1.4435,
      "step": 340
    },
    {
      "epoch": 0.95,
      "learning_rate": 2.6842494900477365e-05,
      "loss": 1.4192,
      "step": 345
    },
    {
      "epoch": 0.97,
      "learning_rate": 2.6301176735383382e-05,
      "loss": 1.7033,
      "step": 350
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.57592461044435e-05,
      "loss": 1.4952,
      "step": 355
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.521695809522061e-05,
      "loss": 1.3207,
      "step": 360
    },
    {
      "epoch": 1.01,
      "learning_rate": 2.467456796349607e-05,
      "loss": 1.2689,
      "step": 365
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.4132331013120453e-05,
      "loss": 1.5812,
      "step": 370
    },
    {
      "epoch": 1.04,
      "learning_rate": 2.3590502475841642e-05,
      "loss": 1.8583,
      "step": 375
    },
    {
      "epoch": 1.05,
      "learning_rate": 2.304933739116688e-05,
      "loss": 1.7619,
      "step": 380
    },
    {
      "epoch": 1.06,
      "learning_rate": 2.2509090486315246e-05,
      "loss": 1.5837,
      "step": 385
    },
    {
      "epoch": 1.08,
      "learning_rate": 2.1970016056317203e-05,
      "loss": 1.6458,
      "step": 390
    },
    {
      "epoch": 1.09,
      "learning_rate": 2.1432367844317558e-05,
      "loss": 1.4172,
      "step": 395
    },
    {
      "epoch": 1.1,
      "learning_rate": 2.0896398922138122e-05,
      "loss": 1.4327,
      "step": 400
    },
    {
      "epoch": 1.12,
      "learning_rate": 2.0362361571156505e-05,
      "loss": 1.0046,
      "step": 405
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.9830507163556816e-05,
      "loss": 1.2355,
      "step": 410
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.930108604400846e-05,
      "loss": 1.224,
      "step": 415
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.8774347411828472e-05,
      "loss": 1.7486,
      "step": 420
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.825053920368306e-05,
      "loss": 1.6091,
      "step": 425
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.772990797688344e-05,
      "loss": 1.4323,
      "step": 430
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.7212698793330916e-05,
      "loss": 1.0638,
      "step": 435
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.6699155104165904e-05,
      "loss": 1.6245,
      "step": 440
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.61895186351751e-05,
      "loss": 1.1679,
      "step": 445
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.568402927301076e-05,
      "loss": 1.7155,
      "step": 450
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.5182924952275768e-05,
      "loss": 1.2734,
      "step": 455
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.4686441543527374e-05,
      "loss": 1.7568,
      "step": 460
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.4194812742252638e-05,
      "loss": 1.7407,
      "step": 465
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.3708269958867565e-05,
      "loss": 1.5895,
      "step": 470
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.322704220979187e-05,
      "loss": 2.2967,
      "step": 475
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.2751356009650681e-05,
      "loss": 1.4595,
      "step": 480
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.2281435264653665e-05,
      "loss": 1.6341,
      "step": 485
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.1817501167202099e-05,
      "loss": 1.3562,
      "step": 490
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.1359772091773263e-05,
      "loss": 1.6777,
      "step": 495
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.0908463492131227e-05,
      "loss": 1.0831,
      "step": 500
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.0463787799912465e-05,
      "loss": 1.0575,
      "step": 505
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.0025954324633948e-05,
      "loss": 1.3363,
      "step": 510
    },
    {
      "epoch": 1.42,
      "learning_rate": 9.595169155170852e-06,
      "loss": 0.9582,
      "step": 515
    },
    {
      "epoch": 1.44,
      "learning_rate": 9.171635062750189e-06,
      "loss": 1.8537,
      "step": 520
    },
    {
      "epoch": 1.45,
      "learning_rate": 8.755551405506143e-06,
      "loss": 1.372,
      "step": 525
    },
    {
      "epoch": 1.46,
      "learning_rate": 8.347114034641806e-06,
      "loss": 1.6832,
      "step": 530
    },
    {
      "epoch": 1.48,
      "learning_rate": 7.9465152022418e-06,
      "loss": 0.8202,
      "step": 535
    },
    {
      "epoch": 1.49,
      "learning_rate": 7.5539434707789266e-06,
      "loss": 1.4924,
      "step": 540
    },
    {
      "epoch": 1.51,
      "learning_rate": 7.169583624357451e-06,
      "loss": 0.8526,
      "step": 545
    },
    {
      "epoch": 1.52,
      "learning_rate": 6.793616581735062e-06,
      "loss": 1.8087,
      "step": 550
    },
    {
      "epoch": 1.53,
      "learning_rate": 6.42621931116405e-06,
      "loss": 1.5891,
      "step": 555
    },
    {
      "epoch": 1.55,
      "learning_rate": 6.067564747092094e-06,
      "loss": 1.9362,
      "step": 560
    },
    {
      "epoch": 1.56,
      "learning_rate": 5.717821708761822e-06,
      "loss": 1.0487,
      "step": 565
    },
    {
      "epoch": 1.57,
      "learning_rate": 5.377154820747271e-06,
      "loss": 1.6339,
      "step": 570
    },
    {
      "epoch": 1.59,
      "learning_rate": 5.045724435464874e-06,
      "loss": 1.2953,
      "step": 575
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.72368655769535e-06,
      "loss": 1.0077,
      "step": 580
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.411192771152004e-06,
      "loss": 1.1702,
      "step": 585
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.108390167130044e-06,
      "loss": 1.5571,
      "step": 590
    },
    {
      "epoch": 1.64,
      "learning_rate": 3.8154212752704976e-06,
      "loss": 1.3669,
      "step": 595
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.532423996471307e-06,
      "loss": 1.6162,
      "step": 600
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.259531537977123e-06,
      "loss": 0.4967,
      "step": 605
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.9968723506784953e-06,
      "loss": 1.066,
      "step": 610
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.7445700686498545e-06,
      "loss": 1.4617,
      "step": 615
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.502743450954714e-06,
      "loss": 1.3762,
      "step": 620
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.271506325745662e-06,
      "loss": 1.6011,
      "step": 625
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.050967536685233e-06,
      "loss": 1.7583,
      "step": 630
    },
    {
      "epoch": 1.75,
      "learning_rate": 1.8412308917130611e-06,
      "loss": 1.4909,
      "step": 635
    },
    {
      "epoch": 1.77,
      "learning_rate": 1.6423951141833011e-06,
      "loss": 0.8267,
      "step": 640
    },
    {
      "epoch": 1.78,
      "learning_rate": 1.4545537963954247e-06,
      "loss": 1.3815,
      "step": 645
    },
    {
      "epoch": 1.8,
      "learning_rate": 1.2777953555401678e-06,
      "loss": 1.618,
      "step": 650
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.1122029920814236e-06,
      "loss": 1.6904,
      "step": 655
    },
    {
      "epoch": 1.82,
      "learning_rate": 9.578546505936676e-07,
      "loss": 1.1602,
      "step": 660
    },
    {
      "epoch": 1.84,
      "learning_rate": 8.148229830733295e-07,
      "loss": 1.7147,
      "step": 665
    },
    {
      "epoch": 1.85,
      "learning_rate": 6.831753147413827e-07,
      "loss": 1.3837,
      "step": 670
    },
    {
      "epoch": 1.86,
      "learning_rate": 5.629736123532653e-07,
      "loss": 1.5324,
      "step": 675
    },
    {
      "epoch": 1.88,
      "learning_rate": 4.5427445503103684e-07,
      "loss": 0.9563,
      "step": 680
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.571290076314959e-07,
      "loss": 1.2545,
      "step": 685
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.7158299666280864e-07,
      "loss": 1.5162,
      "step": 690
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.9767668876096713e-07,
      "loss": 1.3806,
      "step": 695
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.3544487173623443e-07,
      "loss": 1.4588,
      "step": 700
    },
    {
      "epoch": 1.95,
      "learning_rate": 8.491683819846219e-08,
      "loss": 2.3318,
      "step": 705
    },
    {
      "epoch": 1.96,
      "learning_rate": 4.611637176901162e-08,
      "loss": 1.021,
      "step": 710
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.9061735885772536e-08,
      "loss": 1.5645,
      "step": 715
    },
    {
      "epoch": 1.99,
      "learning_rate": 3.76566520653987e-09,
      "loss": 1.6854,
      "step": 720
    },
    {
      "epoch": 2.0,
      "step": 724,
      "total_flos": 3.95850929775575e+16,
      "train_loss": 1.5255728854658854,
      "train_runtime": 1381.7862,
      "train_samples_per_second": 0.524,
      "train_steps_per_second": 0.524
    }
  ],
  "logging_steps": 5,
  "max_steps": 724,
  "num_train_epochs": 2,
  "save_steps": 5000,
  "total_flos": 3.95850929775575e+16,
  "trial_name": null,
  "trial_params": null
}