{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 16.483516483516482,
  "eval_steps": 500,
  "global_step": 1500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.11,
      "learning_rate": 4.996861268047709e-05,
      "loss": 2.5145,
      "step": 10
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.9937225360954175e-05,
      "loss": 2.8458,
      "step": 20
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.9905838041431265e-05,
      "loss": 2.7597,
      "step": 30
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.987445072190835e-05,
      "loss": 2.8058,
      "step": 40
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.984306340238544e-05,
      "loss": 2.6902,
      "step": 50
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.981167608286253e-05,
      "loss": 2.7747,
      "step": 60
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.978028876333961e-05,
      "loss": 2.9923,
      "step": 70
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.97489014438167e-05,
      "loss": 2.7536,
      "step": 80
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.971751412429379e-05,
      "loss": 2.8079,
      "step": 90
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.968612680477088e-05,
      "loss": 2.8927,
      "step": 100
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.965473948524796e-05,
      "loss": 2.815,
      "step": 110
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.962335216572505e-05,
      "loss": 2.9312,
      "step": 120
    },
    {
      "epoch": 1.43,
      "learning_rate": 4.959196484620214e-05,
      "loss": 2.598,
      "step": 130
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.956057752667922e-05,
      "loss": 2.6474,
      "step": 140
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.952919020715631e-05,
      "loss": 2.7857,
      "step": 150
    },
    {
      "epoch": 1.76,
      "learning_rate": 4.9497802887633396e-05,
      "loss": 2.7879,
      "step": 160
    },
    {
      "epoch": 1.87,
      "learning_rate": 4.9466415568110485e-05,
      "loss": 2.6259,
      "step": 170
    },
    {
      "epoch": 1.98,
      "learning_rate": 4.9435028248587575e-05,
      "loss": 2.6434,
      "step": 180
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.940364092906466e-05,
      "loss": 2.6546,
      "step": 190
    },
    {
      "epoch": 2.2,
      "learning_rate": 4.937225360954175e-05,
      "loss": 2.7263,
      "step": 200
    },
    {
      "epoch": 2.31,
      "learning_rate": 4.934086629001883e-05,
      "loss": 2.843,
      "step": 210
    },
    {
      "epoch": 2.42,
      "learning_rate": 4.930947897049592e-05,
      "loss": 2.6683,
      "step": 220
    },
    {
      "epoch": 2.53,
      "learning_rate": 4.927809165097301e-05,
      "loss": 2.441,
      "step": 230
    },
    {
      "epoch": 2.64,
      "learning_rate": 4.924670433145009e-05,
      "loss": 2.8475,
      "step": 240
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.921531701192718e-05,
      "loss": 2.7734,
      "step": 250
    },
    {
      "epoch": 2.86,
      "learning_rate": 4.918392969240427e-05,
      "loss": 2.839,
      "step": 260
    },
    {
      "epoch": 2.97,
      "learning_rate": 4.915254237288136e-05,
      "loss": 2.5971,
      "step": 270
    },
    {
      "epoch": 3.08,
      "learning_rate": 4.9121155053358444e-05,
      "loss": 2.5059,
      "step": 280
    },
    {
      "epoch": 3.19,
      "learning_rate": 4.9089767733835534e-05,
      "loss": 2.6131,
      "step": 290
    },
    {
      "epoch": 3.3,
      "learning_rate": 4.9058380414312623e-05,
      "loss": 2.8308,
      "step": 300
    },
    {
      "epoch": 3.41,
      "learning_rate": 4.9026993094789706e-05,
      "loss": 2.7235,
      "step": 310
    },
    {
      "epoch": 3.52,
      "learning_rate": 4.8995605775266796e-05,
      "loss": 2.7609,
      "step": 320
    },
    {
      "epoch": 3.63,
      "learning_rate": 4.896421845574388e-05,
      "loss": 2.7446,
      "step": 330
    },
    {
      "epoch": 3.74,
      "learning_rate": 4.893283113622097e-05,
      "loss": 2.6262,
      "step": 340
    },
    {
      "epoch": 3.85,
      "learning_rate": 4.890144381669806e-05,
      "loss": 2.4824,
      "step": 350
    },
    {
      "epoch": 3.96,
      "learning_rate": 4.887005649717514e-05,
      "loss": 2.8134,
      "step": 360
    },
    {
      "epoch": 4.07,
      "learning_rate": 4.883866917765223e-05,
      "loss": 2.6734,
      "step": 370
    },
    {
      "epoch": 4.18,
      "learning_rate": 4.8807281858129313e-05,
      "loss": 2.6152,
      "step": 380
    },
    {
      "epoch": 4.29,
      "learning_rate": 4.87758945386064e-05,
      "loss": 2.8046,
      "step": 390
    },
    {
      "epoch": 4.4,
      "learning_rate": 4.874450721908349e-05,
      "loss": 2.6603,
      "step": 400
    },
    {
      "epoch": 4.51,
      "learning_rate": 4.8713119899560576e-05,
      "loss": 2.789,
      "step": 410
    },
    {
      "epoch": 4.62,
      "learning_rate": 4.8681732580037665e-05,
      "loss": 2.5235,
      "step": 420
    },
    {
      "epoch": 4.73,
      "learning_rate": 4.8650345260514755e-05,
      "loss": 2.6632,
      "step": 430
    },
    {
      "epoch": 4.84,
      "learning_rate": 4.8618957940991844e-05,
      "loss": 2.6261,
      "step": 440
    },
    {
      "epoch": 4.95,
      "learning_rate": 4.8587570621468934e-05,
      "loss": 2.6481,
      "step": 450
    },
    {
      "epoch": 5.05,
      "learning_rate": 4.855618330194602e-05,
      "loss": 2.6359,
      "step": 460
    },
    {
      "epoch": 5.16,
      "learning_rate": 4.8524795982423107e-05,
      "loss": 2.6283,
      "step": 470
    },
    {
      "epoch": 5.27,
      "learning_rate": 4.849340866290019e-05,
      "loss": 2.6531,
      "step": 480
    },
    {
      "epoch": 5.38,
      "learning_rate": 4.846202134337728e-05,
      "loss": 2.6473,
      "step": 490
    },
    {
      "epoch": 5.49,
      "learning_rate": 4.843063402385437e-05,
      "loss": 2.7426,
      "step": 500
    },
    {
      "epoch": 5.6,
      "learning_rate": 4.839924670433145e-05,
      "loss": 2.7108,
      "step": 510
    },
    {
      "epoch": 5.71,
      "learning_rate": 4.836785938480854e-05,
      "loss": 2.6723,
      "step": 520
    },
    {
      "epoch": 5.82,
      "learning_rate": 4.8336472065285624e-05,
      "loss": 2.6087,
      "step": 530
    },
    {
      "epoch": 5.93,
      "learning_rate": 4.8305084745762714e-05,
      "loss": 2.5313,
      "step": 540
    },
    {
      "epoch": 6.04,
      "learning_rate": 4.82736974262398e-05,
      "loss": 2.6208,
      "step": 550
    },
    {
      "epoch": 6.15,
      "learning_rate": 4.8242310106716886e-05,
      "loss": 2.613,
      "step": 560
    },
    {
      "epoch": 6.26,
      "learning_rate": 4.8210922787193976e-05,
      "loss": 2.645,
      "step": 570
    },
    {
      "epoch": 6.37,
      "learning_rate": 4.817953546767106e-05,
      "loss": 2.521,
      "step": 580
    },
    {
      "epoch": 6.48,
      "learning_rate": 4.814814814814815e-05,
      "loss": 2.7001,
      "step": 590
    },
    {
      "epoch": 6.59,
      "learning_rate": 4.811676082862524e-05,
      "loss": 2.6073,
      "step": 600
    },
    {
      "epoch": 6.7,
      "learning_rate": 4.808537350910233e-05,
      "loss": 2.6636,
      "step": 610
    },
    {
      "epoch": 6.81,
      "learning_rate": 4.805398618957942e-05,
      "loss": 2.5819,
      "step": 620
    },
    {
      "epoch": 6.92,
      "learning_rate": 4.80225988700565e-05,
      "loss": 2.6064,
      "step": 630
    },
    {
      "epoch": 7.03,
      "learning_rate": 4.799121155053359e-05,
      "loss": 2.7406,
      "step": 640
    },
    {
      "epoch": 7.14,
      "learning_rate": 4.795982423101067e-05,
      "loss": 2.5997,
      "step": 650
    },
    {
      "epoch": 7.25,
      "learning_rate": 4.792843691148776e-05,
      "loss": 2.5727,
      "step": 660
    },
    {
      "epoch": 7.36,
      "learning_rate": 4.789704959196485e-05,
      "loss": 2.4625,
      "step": 670
    },
    {
      "epoch": 7.47,
      "learning_rate": 4.7865662272441935e-05,
      "loss": 2.5773,
      "step": 680
    },
    {
      "epoch": 7.58,
      "learning_rate": 4.7834274952919024e-05,
      "loss": 2.5806,
      "step": 690
    },
    {
      "epoch": 7.69,
      "learning_rate": 4.780288763339611e-05,
      "loss": 2.5901,
      "step": 700
    },
    {
      "epoch": 7.8,
      "learning_rate": 4.77715003138732e-05,
      "loss": 2.4906,
      "step": 710
    },
    {
      "epoch": 7.91,
      "learning_rate": 4.7740112994350286e-05,
      "loss": 2.6069,
      "step": 720
    },
    {
      "epoch": 8.02,
      "learning_rate": 4.770872567482737e-05,
      "loss": 2.5363,
      "step": 730
    },
    {
      "epoch": 8.13,
      "learning_rate": 4.767733835530446e-05,
      "loss": 2.5511,
      "step": 740
    },
    {
      "epoch": 8.24,
      "learning_rate": 4.764595103578154e-05,
      "loss": 2.5839,
      "step": 750
    },
    {
      "epoch": 8.35,
      "learning_rate": 4.761456371625863e-05,
      "loss": 2.462,
      "step": 760
    },
    {
      "epoch": 8.46,
      "learning_rate": 4.758317639673572e-05,
      "loss": 2.8533,
      "step": 770
    },
    {
      "epoch": 8.57,
      "learning_rate": 4.755178907721281e-05,
      "loss": 2.3477,
      "step": 780
    },
    {
      "epoch": 8.68,
      "learning_rate": 4.75204017576899e-05,
      "loss": 2.553,
      "step": 790
    },
    {
      "epoch": 8.79,
      "learning_rate": 4.748901443816698e-05,
      "loss": 2.614,
      "step": 800
    },
    {
      "epoch": 8.9,
      "learning_rate": 4.745762711864407e-05,
      "loss": 2.6441,
      "step": 810
    },
    {
      "epoch": 9.01,
      "learning_rate": 4.7426239799121155e-05,
      "loss": 2.6618,
      "step": 820
    },
    {
      "epoch": 9.12,
      "learning_rate": 4.7394852479598245e-05,
      "loss": 2.4887,
      "step": 830
    },
    {
      "epoch": 9.23,
      "learning_rate": 4.7363465160075335e-05,
      "loss": 2.3639,
      "step": 840
    },
    {
      "epoch": 9.34,
      "learning_rate": 4.733207784055242e-05,
      "loss": 2.7387,
      "step": 850
    },
    {
      "epoch": 9.45,
      "learning_rate": 4.730069052102951e-05,
      "loss": 2.4262,
      "step": 860
    },
    {
      "epoch": 9.56,
      "learning_rate": 4.726930320150659e-05,
      "loss": 2.6172,
      "step": 870
    },
    {
      "epoch": 9.67,
      "learning_rate": 4.723791588198368e-05,
      "loss": 2.7115,
      "step": 880
    },
    {
      "epoch": 9.78,
      "learning_rate": 4.720652856246077e-05,
      "loss": 2.5397,
      "step": 890
    },
    {
      "epoch": 9.89,
      "learning_rate": 4.717514124293785e-05,
      "loss": 2.4934,
      "step": 900
    },
    {
      "epoch": 10.0,
      "learning_rate": 4.714375392341494e-05,
      "loss": 2.5521,
      "step": 910
    },
    {
      "epoch": 10.11,
      "learning_rate": 4.7112366603892025e-05,
      "loss": 2.3767,
      "step": 920
    },
    {
      "epoch": 10.22,
      "learning_rate": 4.7080979284369114e-05,
      "loss": 2.5422,
      "step": 930
    },
    {
      "epoch": 10.33,
      "learning_rate": 4.7049591964846204e-05,
      "loss": 2.7068,
      "step": 940
    },
    {
      "epoch": 10.44,
      "learning_rate": 4.7018204645323294e-05,
      "loss": 2.5518,
      "step": 950
    },
    {
      "epoch": 10.55,
      "learning_rate": 4.698681732580038e-05,
      "loss": 2.4129,
      "step": 960
    },
    {
      "epoch": 10.66,
      "learning_rate": 4.6955430006277466e-05,
      "loss": 2.6916,
      "step": 970
    },
    {
      "epoch": 10.77,
      "learning_rate": 4.6924042686754556e-05,
      "loss": 2.5753,
      "step": 980
    },
    {
      "epoch": 10.88,
      "learning_rate": 4.689265536723164e-05,
      "loss": 2.3783,
      "step": 990
    },
    {
      "epoch": 10.99,
      "learning_rate": 4.686126804770873e-05,
      "loss": 2.5766,
      "step": 1000
    },
    {
      "epoch": 11.1,
      "learning_rate": 4.682988072818582e-05,
      "loss": 2.56,
      "step": 1010
    },
    {
      "epoch": 11.21,
      "learning_rate": 4.67984934086629e-05,
      "loss": 2.59,
      "step": 1020
    },
    {
      "epoch": 11.32,
      "learning_rate": 4.676710608913999e-05,
      "loss": 2.5592,
      "step": 1030
    },
    {
      "epoch": 11.43,
      "learning_rate": 4.673571876961707e-05,
      "loss": 2.553,
      "step": 1040
    },
    {
      "epoch": 11.54,
      "learning_rate": 4.670433145009416e-05,
      "loss": 2.564,
      "step": 1050
    },
    {
      "epoch": 11.65,
      "learning_rate": 4.667294413057125e-05,
      "loss": 2.471,
      "step": 1060
    },
    {
      "epoch": 11.76,
      "learning_rate": 4.6641556811048335e-05,
      "loss": 2.4783,
      "step": 1070
    },
    {
      "epoch": 11.87,
      "learning_rate": 4.6610169491525425e-05,
      "loss": 2.5474,
      "step": 1080
    },
    {
      "epoch": 11.98,
      "learning_rate": 4.657878217200251e-05,
      "loss": 2.5478,
      "step": 1090
    },
    {
      "epoch": 12.09,
      "learning_rate": 4.65473948524796e-05,
      "loss": 2.4591,
      "step": 1100
    },
    {
      "epoch": 12.2,
      "learning_rate": 4.651600753295669e-05,
      "loss": 2.4682,
      "step": 1110
    },
    {
      "epoch": 12.31,
      "learning_rate": 4.6484620213433777e-05,
      "loss": 2.5391,
      "step": 1120
    },
    {
      "epoch": 12.42,
      "learning_rate": 4.6453232893910866e-05,
      "loss": 2.6003,
      "step": 1130
    },
    {
      "epoch": 12.53,
      "learning_rate": 4.642184557438795e-05,
      "loss": 2.4073,
      "step": 1140
    },
    {
      "epoch": 12.64,
      "learning_rate": 4.639045825486504e-05,
      "loss": 2.5084,
      "step": 1150
    },
    {
      "epoch": 12.75,
      "learning_rate": 4.635907093534212e-05,
      "loss": 2.4579,
      "step": 1160
    },
    {
      "epoch": 12.86,
      "learning_rate": 4.632768361581921e-05,
      "loss": 2.512,
      "step": 1170
    },
    {
      "epoch": 12.97,
      "learning_rate": 4.62962962962963e-05,
      "loss": 2.6637,
      "step": 1180
    },
    {
      "epoch": 13.08,
      "learning_rate": 4.6264908976773384e-05,
      "loss": 2.3842,
      "step": 1190
    },
    {
      "epoch": 13.19,
      "learning_rate": 4.623352165725047e-05,
      "loss": 2.3314,
      "step": 1200
    },
    {
      "epoch": 13.3,
      "learning_rate": 4.620213433772756e-05,
      "loss": 2.4636,
      "step": 1210
    },
    {
      "epoch": 13.41,
      "learning_rate": 4.6170747018204646e-05,
      "loss": 2.4403,
      "step": 1220
    },
    {
      "epoch": 13.52,
      "learning_rate": 4.6139359698681735e-05,
      "loss": 2.7347,
      "step": 1230
    },
    {
      "epoch": 13.63,
      "learning_rate": 4.610797237915882e-05,
      "loss": 2.4503,
      "step": 1240
    },
    {
      "epoch": 13.74,
      "learning_rate": 4.607658505963591e-05,
      "loss": 2.566,
      "step": 1250
    },
    {
      "epoch": 13.85,
      "learning_rate": 4.6045197740113e-05,
      "loss": 2.3861,
      "step": 1260
    },
    {
      "epoch": 13.96,
      "learning_rate": 4.601381042059008e-05,
      "loss": 2.4167,
      "step": 1270
    },
    {
      "epoch": 14.07,
      "learning_rate": 4.598242310106718e-05,
      "loss": 2.751,
      "step": 1280
    },
    {
      "epoch": 14.18,
      "learning_rate": 4.595103578154426e-05,
      "loss": 2.4427,
      "step": 1290
    },
    {
      "epoch": 14.29,
      "learning_rate": 4.591964846202135e-05,
      "loss": 2.5363,
      "step": 1300
    },
    {
      "epoch": 14.4,
      "learning_rate": 4.588826114249843e-05,
      "loss": 2.3152,
      "step": 1310
    },
    {
      "epoch": 14.51,
      "learning_rate": 4.585687382297552e-05,
      "loss": 2.6868,
      "step": 1320
    },
    {
      "epoch": 14.62,
      "learning_rate": 4.582548650345261e-05,
      "loss": 2.4857,
      "step": 1330
    },
    {
      "epoch": 14.73,
      "learning_rate": 4.5794099183929694e-05,
      "loss": 2.5417,
      "step": 1340
    },
    {
      "epoch": 14.84,
      "learning_rate": 4.5762711864406784e-05,
      "loss": 2.4462,
      "step": 1350
    },
    {
      "epoch": 14.95,
      "learning_rate": 4.573132454488387e-05,
      "loss": 2.3253,
      "step": 1360
    },
    {
      "epoch": 15.05,
      "learning_rate": 4.5699937225360956e-05,
      "loss": 2.674,
      "step": 1370
    },
    {
      "epoch": 15.16,
      "learning_rate": 4.5668549905838046e-05,
      "loss": 2.607,
      "step": 1380
    },
    {
      "epoch": 15.27,
      "learning_rate": 4.563716258631513e-05,
      "loss": 2.3933,
      "step": 1390
    },
    {
      "epoch": 15.38,
      "learning_rate": 4.560577526679222e-05,
      "loss": 2.4335,
      "step": 1400
    },
    {
      "epoch": 15.49,
      "learning_rate": 4.55743879472693e-05,
      "loss": 2.2217,
      "step": 1410
    },
    {
      "epoch": 15.6,
      "learning_rate": 4.554300062774639e-05,
      "loss": 2.4138,
      "step": 1420
    },
    {
      "epoch": 15.71,
      "learning_rate": 4.551161330822348e-05,
      "loss": 2.5214,
      "step": 1430
    },
    {
      "epoch": 15.82,
      "learning_rate": 4.548022598870056e-05,
      "loss": 2.6247,
      "step": 1440
    },
    {
      "epoch": 15.93,
      "learning_rate": 4.544883866917766e-05,
      "loss": 2.495,
      "step": 1450
    },
    {
      "epoch": 16.04,
      "learning_rate": 4.541745134965474e-05,
      "loss": 2.4676,
      "step": 1460
    },
    {
      "epoch": 16.15,
      "learning_rate": 4.538606403013183e-05,
      "loss": 2.5501,
      "step": 1470
    },
    {
      "epoch": 16.26,
      "learning_rate": 4.5354676710608915e-05,
      "loss": 2.612,
      "step": 1480
    },
    {
      "epoch": 16.37,
      "learning_rate": 4.5323289391086005e-05,
      "loss": 2.6939,
      "step": 1490
    },
    {
      "epoch": 16.48,
      "learning_rate": 4.5291902071563094e-05,
      "loss": 2.3641,
      "step": 1500
    }
  ],
  "logging_steps": 10,
  "max_steps": 15930,
  "num_train_epochs": 176,
  "save_steps": 500,
  "total_flos": 7.73986862813184e+16,
  "trial_name": null,
  "trial_params": null
}