{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.2754030638590854,
  "eval_steps": 500,
  "global_step": 2400,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 2.0833333333333334e-09,
      "loss": 0.6931,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.0416666666666666e-08,
      "loss": 0.6931,
      "step": 5
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.083333333333333e-08,
      "loss": 0.6932,
      "step": 10
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.125e-08,
      "loss": 0.6931,
      "step": 15
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.166666666666666e-08,
      "loss": 0.6931,
      "step": 20
    },
    {
      "epoch": 0.0,
      "learning_rate": 5.208333333333333e-08,
      "loss": 0.6929,
      "step": 25
    },
    {
      "epoch": 0.0,
      "learning_rate": 6.25e-08,
      "loss": 0.6928,
      "step": 30
    },
    {
      "epoch": 0.0,
      "learning_rate": 7.291666666666667e-08,
      "loss": 0.6923,
      "step": 35
    },
    {
      "epoch": 0.0,
      "learning_rate": 8.333333333333333e-08,
      "loss": 0.6919,
      "step": 40
    },
    {
      "epoch": 0.01,
      "learning_rate": 9.375e-08,
      "loss": 0.6914,
      "step": 45
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.0416666666666667e-07,
      "loss": 0.6907,
      "step": 50
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.1458333333333332e-07,
      "loss": 0.6898,
      "step": 55
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.25e-07,
      "loss": 0.6889,
      "step": 60
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.3541666666666666e-07,
      "loss": 0.6878,
      "step": 65
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.4583333333333335e-07,
      "loss": 0.6865,
      "step": 70
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.5624999999999999e-07,
      "loss": 0.685,
      "step": 75
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.6666666666666665e-07,
      "loss": 0.6835,
      "step": 80
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.7708333333333334e-07,
      "loss": 0.6811,
      "step": 85
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.875e-07,
      "loss": 0.6788,
      "step": 90
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.9791666666666664e-07,
      "loss": 0.6767,
      "step": 95
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.0833333333333333e-07,
      "loss": 0.6733,
      "step": 100
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.1875e-07,
      "loss": 0.67,
      "step": 105
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.2916666666666663e-07,
      "loss": 0.6669,
      "step": 110
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.3958333333333335e-07,
      "loss": 0.6629,
      "step": 115
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.5e-07,
      "loss": 0.6596,
      "step": 120
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.604166666666667e-07,
      "loss": 0.6543,
      "step": 125
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.708333333333333e-07,
      "loss": 0.6501,
      "step": 130
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.8125e-07,
      "loss": 0.6456,
      "step": 135
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.916666666666667e-07,
      "loss": 0.6391,
      "step": 140
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.020833333333333e-07,
      "loss": 0.6346,
      "step": 145
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.1249999999999997e-07,
      "loss": 0.6286,
      "step": 150
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.2291666666666666e-07,
      "loss": 0.6219,
      "step": 155
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.333333333333333e-07,
      "loss": 0.6141,
      "step": 160
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.4166666666666664e-07,
      "loss": 0.6126,
      "step": 165
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.5208333333333333e-07,
      "loss": 0.6033,
      "step": 170
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.6249999999999997e-07,
      "loss": 0.5979,
      "step": 175
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.7291666666666666e-07,
      "loss": 0.5881,
      "step": 180
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.8333333333333335e-07,
      "loss": 0.5776,
      "step": 185
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.9375e-07,
      "loss": 0.5714,
      "step": 190
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.041666666666667e-07,
      "loss": 0.562,
      "step": 195
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.145833333333333e-07,
      "loss": 0.5572,
      "step": 200
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.2499999999999995e-07,
      "loss": 0.5453,
      "step": 205
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3541666666666664e-07,
      "loss": 0.5399,
      "step": 210
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.4583333333333334e-07,
      "loss": 0.5318,
      "step": 215
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.5624999999999997e-07,
      "loss": 0.517,
      "step": 220
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.6666666666666666e-07,
      "loss": 0.5073,
      "step": 225
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.770833333333334e-07,
      "loss": 0.4968,
      "step": 230
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.875e-07,
      "loss": 0.4831,
      "step": 235
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.979166666666666e-07,
      "loss": 0.4778,
      "step": 240
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.99995769214436e-07,
      "loss": 0.4682,
      "step": 245
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.999785818935017e-07,
      "loss": 0.4524,
      "step": 250
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.999481745213471e-07,
      "loss": 0.4422,
      "step": 255
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.999045487060579e-07,
      "loss": 0.4382,
      "step": 260
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.998477067547739e-07,
      "loss": 0.4222,
      "step": 265
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.997776516735666e-07,
      "loss": 0.4109,
      "step": 270
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.996943871672807e-07,
      "loss": 0.4046,
      "step": 275
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.995979176393372e-07,
      "loss": 0.384,
      "step": 280
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.994882481915018e-07,
      "loss": 0.3831,
      "step": 285
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993653846236143e-07,
      "loss": 0.3644,
      "step": 290
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.99229333433282e-07,
      "loss": 0.3668,
      "step": 295
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.99080101815536e-07,
      "loss": 0.3467,
      "step": 300
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.989176976624511e-07,
      "loss": 0.3406,
      "step": 305
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.987421295627278e-07,
      "loss": 0.3281,
      "step": 310
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.98553406801239e-07,
      "loss": 0.3128,
      "step": 315
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.983515393585378e-07,
      "loss": 0.3051,
      "step": 320
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.981365379103305e-07,
      "loss": 0.3054,
      "step": 325
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.97908413826912e-07,
      "loss": 0.2855,
      "step": 330
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.97667179172564e-07,
      "loss": 0.2793,
      "step": 335
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.974128467049176e-07,
      "loss": 0.2664,
      "step": 340
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.971454298742779e-07,
      "loss": 0.2559,
      "step": 345
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.968649428229134e-07,
      "loss": 0.2464,
      "step": 350
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.965714003843078e-07,
      "loss": 0.2435,
      "step": 355
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.962648180823752e-07,
      "loss": 0.2322,
      "step": 360
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.959452121306397e-07,
      "loss": 0.22,
      "step": 365
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.956125994313774e-07,
      "loss": 0.2156,
      "step": 370
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.952669975747232e-07,
      "loss": 0.2027,
      "step": 375
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.949084248377396e-07,
      "loss": 0.197,
      "step": 380
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.945369001834514e-07,
      "loss": 0.1852,
      "step": 385
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.941524432598415e-07,
      "loss": 0.1776,
      "step": 390
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.937550743988127e-07,
      "loss": 0.1731,
      "step": 395
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.933448146151121e-07,
      "loss": 0.1715,
      "step": 400
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9292168560522e-07,
      "loss": 0.1641,
      "step": 405
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.924857097462023e-07,
      "loss": 0.1499,
      "step": 410
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.920369100945269e-07,
      "loss": 0.1555,
      "step": 415
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.915753103848449e-07,
      "loss": 0.139,
      "step": 420
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.911009350287347e-07,
      "loss": 0.132,
      "step": 425
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.906138091134118e-07,
      "loss": 0.1293,
      "step": 430
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.901139584004014e-07,
      "loss": 0.1181,
      "step": 435
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.896014093241763e-07,
      "loss": 0.1132,
      "step": 440
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.890761889907589e-07,
      "loss": 0.1085,
      "step": 445
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.885383251762876e-07,
      "loss": 0.1039,
      "step": 450
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.879878463255482e-07,
      "loss": 0.1046,
      "step": 455
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.874247815504692e-07,
      "loss": 0.101,
      "step": 460
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.868491606285823e-07,
      "loss": 0.0934,
      "step": 465
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.862610140014478e-07,
      "loss": 0.089,
      "step": 470
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.856603727730446e-07,
      "loss": 0.0836,
      "step": 475
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.850472687081252e-07,
      "loss": 0.08,
      "step": 480
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.844217342305363e-07,
      "loss": 0.0765,
      "step": 485
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.837838024215029e-07,
      "loss": 0.0708,
      "step": 490
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.831335070178805e-07,
      "loss": 0.0693,
      "step": 495
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.824708824103693e-07,
      "loss": 0.0651,
      "step": 500
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.817959636416969e-07,
      "loss": 0.0638,
      "step": 505
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.811087864047635e-07,
      "loss": 0.0582,
      "step": 510
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.80409387040756e-07,
      "loss": 0.0535,
      "step": 515
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.796978025372246e-07,
      "loss": 0.0574,
      "step": 520
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.789740705261278e-07,
      "loss": 0.048,
      "step": 525
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.782382292818416e-07,
      "loss": 0.0533,
      "step": 530
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.774903177191357e-07,
      "loss": 0.0474,
      "step": 535
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.767303753911156e-07,
      "loss": 0.0474,
      "step": 540
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.759584424871301e-07,
      "loss": 0.0442,
      "step": 545
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.751745598306469e-07,
      "loss": 0.0461,
      "step": 550
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.7437876887709316e-07,
      "loss": 0.0433,
      "step": 555
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.735711117116629e-07,
      "loss": 0.0392,
      "step": 560
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.7275163104709194e-07,
      "loss": 0.039,
      "step": 565
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.719203702213985e-07,
      "loss": 0.0368,
      "step": 570
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.710773731955917e-07,
      "loss": 0.0355,
      "step": 575
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.702226845513464e-07,
      "loss": 0.0321,
      "step": 580
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.693563494886454e-07,
      "loss": 0.0306,
      "step": 585
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.684784138233898e-07,
      "loss": 0.0345,
      "step": 590
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.675889239849749e-07,
      "loss": 0.0276,
      "step": 595
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.6668792701383576e-07,
      "loss": 0.0264,
      "step": 600
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.6577547055895906e-07,
      "loss": 0.0322,
      "step": 605
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.648516028753632e-07,
      "loss": 0.025,
      "step": 610
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.6391637282154626e-07,
      "loss": 0.0277,
      "step": 615
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.6296982985690256e-07,
      "loss": 0.0276,
      "step": 620
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.6201202403910643e-07,
      "loss": 0.0252,
      "step": 625
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.610430060214655e-07,
      "loss": 0.0256,
      "step": 630
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.6006282705024143e-07,
      "loss": 0.0223,
      "step": 635
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.5907153896193986e-07,
      "loss": 0.0252,
      "step": 640
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.5806919418056944e-07,
      "loss": 0.022,
      "step": 645
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.5705584571486887e-07,
      "loss": 0.02,
      "step": 650
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.5603154715550386e-07,
      "loss": 0.0183,
      "step": 655
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.549963526722331e-07,
      "loss": 0.0195,
      "step": 660
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.5395031701104303e-07,
      "loss": 0.0169,
      "step": 665
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.5289349549125306e-07,
      "loss": 0.0185,
      "step": 670
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.5182594400259e-07,
      "loss": 0.0181,
      "step": 675
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.5074771900223195e-07,
      "loss": 0.018,
      "step": 680
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.4965887751182317e-07,
      "loss": 0.017,
      "step": 685
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.48559477114458e-07,
      "loss": 0.017,
      "step": 690
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.474495759516358e-07,
      "loss": 0.0164,
      "step": 695
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.463292327201862e-07,
      "loss": 0.0171,
      "step": 700
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.451985066691648e-07,
      "loss": 0.016,
      "step": 705
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.4405745759671984e-07,
      "loss": 0.0154,
      "step": 710
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.4290614584692996e-07,
      "loss": 0.0135,
      "step": 715
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.4174463230661264e-07,
      "loss": 0.0154,
      "step": 720
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.405729784021045e-07,
      "loss": 0.0137,
      "step": 725
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.393912460960124e-07,
      "loss": 0.0124,
      "step": 730
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.381994978839371e-07,
      "loss": 0.0122,
      "step": 735
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.3699779679116753e-07,
      "loss": 0.0127,
      "step": 740
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.357862063693485e-07,
      "loss": 0.0127,
      "step": 745
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.3456479069311923e-07,
      "loss": 0.0135,
      "step": 750
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.3333361435672465e-07,
      "loss": 0.0133,
      "step": 755
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.320927424706e-07,
      "loss": 0.0138,
      "step": 760
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.3084224065792695e-07,
      "loss": 0.0114,
      "step": 765
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.2958217505116324e-07,
      "loss": 0.0122,
      "step": 770
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.283126122885454e-07,
      "loss": 0.0126,
      "step": 775
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.2703361951056444e-07,
      "loss": 0.0121,
      "step": 780
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.2574526435641546e-07,
      "loss": 0.0099,
      "step": 785
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.2444761496042005e-07,
      "loss": 0.0104,
      "step": 790
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.2314073994842353e-07,
      "loss": 0.0095,
      "step": 795
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.218247084341655e-07,
      "loss": 0.0108,
      "step": 800
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.204995900156246e-07,
      "loss": 0.01,
      "step": 805
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.1916545477133815e-07,
      "loss": 0.0102,
      "step": 810
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.1782237325669587e-07,
      "loss": 0.0121,
      "step": 815
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.1647041650020853e-07,
      "loss": 0.0075,
      "step": 820
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.151096559997519e-07,
      "loss": 0.0089,
      "step": 825
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.137401637187853e-07,
      "loss": 0.0078,
      "step": 830
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.1236201208254586e-07,
      "loss": 0.0095,
      "step": 835
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.109752739742187e-07,
      "loss": 0.0081,
      "step": 840
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.0958002273108204e-07,
      "loss": 0.0082,
      "step": 845
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.0817633214062904e-07,
      "loss": 0.0079,
      "step": 850
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.067642764366654e-07,
      "loss": 0.0079,
      "step": 855
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.053439302953838e-07,
      "loss": 0.0068,
      "step": 860
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.039153688314145e-07,
      "loss": 0.008,
      "step": 865
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.024786675938529e-07,
      "loss": 0.0076,
      "step": 870
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.01033902562264e-07,
      "loss": 0.0072,
      "step": 875
    },
    {
      "epoch": 0.1,
      "learning_rate": 3.995811501426647e-07,
      "loss": 0.0073,
      "step": 880
    },
    {
      "epoch": 0.1,
      "learning_rate": 3.981204871634827e-07,
      "loss": 0.0083,
      "step": 885
    },
    {
      "epoch": 0.1,
      "learning_rate": 3.966519908714933e-07,
      "loss": 0.0067,
      "step": 890
    },
    {
      "epoch": 0.1,
      "learning_rate": 3.9517573892773493e-07,
      "loss": 0.0067,
      "step": 895
    },
    {
      "epoch": 0.1,
      "learning_rate": 3.936918094034013e-07,
      "loss": 0.0066,
      "step": 900
    },
    {
      "epoch": 0.1,
      "learning_rate": 3.922002807757129e-07,
      "loss": 0.0069,
      "step": 905
    },
    {
      "epoch": 0.1,
      "learning_rate": 3.9070123192376713e-07,
      "loss": 0.0072,
      "step": 910
    },
    {
      "epoch": 0.1,
      "learning_rate": 3.891947421243661e-07,
      "loss": 0.0056,
      "step": 915
    },
    {
      "epoch": 0.11,
      "learning_rate": 3.8768089104782464e-07,
      "loss": 0.0058,
      "step": 920
    },
    {
      "epoch": 0.11,
      "learning_rate": 3.8615975875375676e-07,
      "loss": 0.0075,
      "step": 925
    },
    {
      "epoch": 0.11,
      "learning_rate": 3.846314256868417e-07,
      "loss": 0.0063,
      "step": 930
    },
    {
      "epoch": 0.11,
      "learning_rate": 3.8309597267256966e-07,
      "loss": 0.0061,
      "step": 935
    },
    {
      "epoch": 0.11,
      "learning_rate": 3.8155348091296736e-07,
      "loss": 0.0057,
      "step": 940
    },
    {
      "epoch": 0.11,
      "learning_rate": 3.800040319823038e-07,
      "loss": 0.0058,
      "step": 945
    },
    {
      "epoch": 0.11,
      "learning_rate": 3.784477078227762e-07,
      "loss": 0.0052,
      "step": 950
    },
    {
      "epoch": 0.11,
      "learning_rate": 3.7688459074017603e-07,
      "loss": 0.006,
      "step": 955
    },
    {
      "epoch": 0.11,
      "learning_rate": 3.7531476339953714e-07,
      "loss": 0.0069,
      "step": 960
    },
    {
      "epoch": 0.11,
      "learning_rate": 3.737383088207635e-07,
      "loss": 0.0058,
      "step": 965
    },
    {
      "epoch": 0.11,
      "learning_rate": 3.7215531037423874e-07,
      "loss": 0.0053,
      "step": 970
    },
    {
      "epoch": 0.11,
      "learning_rate": 3.705658517764172e-07,
      "loss": 0.0053,
      "step": 975
    },
    {
      "epoch": 0.11,
      "learning_rate": 3.6897001708539655e-07,
      "loss": 0.0056,
      "step": 980
    },
    {
      "epoch": 0.11,
      "learning_rate": 3.673678906964727e-07,
      "loss": 0.0064,
      "step": 985
    },
    {
      "epoch": 0.11,
      "learning_rate": 3.657595573376761e-07,
      "loss": 0.0053,
      "step": 990
    },
    {
      "epoch": 0.11,
      "learning_rate": 3.6414510206529137e-07,
      "loss": 0.0053,
      "step": 995
    },
    {
      "epoch": 0.11,
      "learning_rate": 3.625246102593588e-07,
      "loss": 0.0058,
      "step": 1000
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.6089816761915904e-07,
      "loss": 0.0049,
      "step": 1005
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.592658601586811e-07,
      "loss": 0.0049,
      "step": 1010
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.5762777420207377e-07,
      "loss": 0.0049,
      "step": 1015
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.559839963790797e-07,
      "loss": 0.0056,
      "step": 1020
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.5433461362045447e-07,
      "loss": 0.0046,
      "step": 1025
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.526797131533693e-07,
      "loss": 0.0042,
      "step": 1030
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.510193824967979e-07,
      "loss": 0.0039,
      "step": 1035
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.493537094568882e-07,
      "loss": 0.0045,
      "step": 1040
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.4768278212231837e-07,
      "loss": 0.0055,
      "step": 1045
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.460066888596391e-07,
      "loss": 0.0046,
      "step": 1050
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.4432551830859926e-07,
      "loss": 0.0047,
      "step": 1055
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.4263935937745906e-07,
      "loss": 0.0035,
      "step": 1060
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.4094830123828786e-07,
      "loss": 0.0043,
      "step": 1065
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.3925243332224835e-07,
      "loss": 0.0036,
      "step": 1070
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.3755184531486684e-07,
      "loss": 0.0042,
      "step": 1075
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.3584662715129065e-07,
      "loss": 0.0044,
      "step": 1080
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.341368690115316e-07,
      "loss": 0.0038,
      "step": 1085
    },
    {
      "epoch": 0.13,
      "learning_rate": 3.324226613156968e-07,
      "loss": 0.0039,
      "step": 1090
    },
    {
      "epoch": 0.13,
      "learning_rate": 3.3070409471920725e-07,
      "loss": 0.0037,
      "step": 1095
    },
    {
      "epoch": 0.13,
      "learning_rate": 3.289812601080029e-07,
      "loss": 0.004,
      "step": 1100
    },
    {
      "epoch": 0.13,
      "learning_rate": 3.272542485937368e-07,
      "loss": 0.0044,
      "step": 1105
    },
    {
      "epoch": 0.13,
      "learning_rate": 3.255231515089565e-07,
      "loss": 0.0033,
      "step": 1110
    },
    {
      "epoch": 0.13,
      "learning_rate": 3.237880604022735e-07,
      "loss": 0.0038,
      "step": 1115
    },
    {
      "epoch": 0.13,
      "learning_rate": 3.220490670335223e-07,
      "loss": 0.0042,
      "step": 1120
    },
    {
      "epoch": 0.13,
      "learning_rate": 3.2030626336890766e-07,
      "loss": 0.0039,
      "step": 1125
    },
    {
      "epoch": 0.13,
      "learning_rate": 3.185597415761405e-07,
      "loss": 0.0033,
      "step": 1130
    },
    {
      "epoch": 0.13,
      "learning_rate": 3.168095940195642e-07,
      "loss": 0.0036,
      "step": 1135
    },
    {
      "epoch": 0.13,
      "learning_rate": 3.150559132552697e-07,
      "loss": 0.0029,
      "step": 1140
    },
    {
      "epoch": 0.13,
      "learning_rate": 3.1329879202620047e-07,
      "loss": 0.0036,
      "step": 1145
    },
    {
      "epoch": 0.13,
      "learning_rate": 3.1153832325724825e-07,
      "loss": 0.004,
      "step": 1150
    },
    {
      "epoch": 0.13,
      "learning_rate": 3.0977460005033854e-07,
      "loss": 0.0037,
      "step": 1155
    },
    {
      "epoch": 0.13,
      "learning_rate": 3.0800771567950697e-07,
      "loss": 0.0027,
      "step": 1160
    },
    {
      "epoch": 0.13,
      "learning_rate": 3.062377635859663e-07,
      "loss": 0.0031,
      "step": 1165
    },
    {
      "epoch": 0.13,
      "learning_rate": 3.04464837373165e-07,
      "loss": 0.0036,
      "step": 1170
    },
    {
      "epoch": 0.13,
      "learning_rate": 3.026890308018374e-07,
      "loss": 0.0028,
      "step": 1175
    },
    {
      "epoch": 0.14,
      "learning_rate": 3.0091043778504433e-07,
      "loss": 0.0034,
      "step": 1180
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.991291523832075e-07,
      "loss": 0.0035,
      "step": 1185
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.9734526879913444e-07,
      "loss": 0.0032,
      "step": 1190
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.955588813730369e-07,
      "loss": 0.0031,
      "step": 1195
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.937700845775416e-07,
      "loss": 0.0032,
      "step": 1200
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.919789730126943e-07,
      "loss": 0.004,
      "step": 1205
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.9018564140095654e-07,
      "loss": 0.003,
      "step": 1210
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.883901845821965e-07,
      "loss": 0.0026,
      "step": 1215
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.8659269750867364e-07,
      "loss": 0.0033,
      "step": 1220
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.8479327524001633e-07,
      "loss": 0.0026,
      "step": 1225
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.8299201293819584e-07,
      "loss": 0.0034,
      "step": 1230
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.811890058624926e-07,
      "loss": 0.0026,
      "step": 1235
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.7938434936445943e-07,
      "loss": 0.003,
      "step": 1240
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.7757813888287795e-07,
      "loss": 0.003,
      "step": 1245
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.75770469938712e-07,
      "loss": 0.0026,
      "step": 1250
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.73961438130056e-07,
      "loss": 0.002,
      "step": 1255
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.721511391270788e-07,
      "loss": 0.0026,
      "step": 1260
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.703396686669646e-07,
      "loss": 0.003,
      "step": 1265
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.6852712254884985e-07,
      "loss": 0.0025,
      "step": 1270
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.667135966287568e-07,
      "loss": 0.0021,
      "step": 1275
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.648991868145244e-07,
      "loss": 0.0019,
      "step": 1280
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.63083989060736e-07,
      "loss": 0.0027,
      "step": 1285
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.6126809936364485e-07,
      "loss": 0.0027,
      "step": 1290
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.5945161375609775e-07,
      "loss": 0.0024,
      "step": 1295
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.576346283024557e-07,
      "loss": 0.0027,
      "step": 1300
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.5581723909351404e-07,
      "loss": 0.0025,
      "step": 1305
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.5399954224142086e-07,
      "loss": 0.0025,
      "step": 1310
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.5218163387459346e-07,
      "loss": 0.0025,
      "step": 1315
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.503636101326354e-07,
      "loss": 0.0024,
      "step": 1320
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.485455671612515e-07,
      "loss": 0.0021,
      "step": 1325
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.4672760110716393e-07,
      "loss": 0.0024,
      "step": 1330
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.4490980811302657e-07,
      "loss": 0.0028,
      "step": 1335
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.4309228431234167e-07,
      "loss": 0.0024,
      "step": 1340
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.412751258243748e-07,
      "loss": 0.0021,
      "step": 1345
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.394584287490721e-07,
      "loss": 0.0021,
      "step": 1350
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.376422891619785e-07,
      "loss": 0.0019,
      "step": 1355
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.3582680310915556e-07,
      "loss": 0.0024,
      "step": 1360
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.340120666021036e-07,
      "loss": 0.0024,
      "step": 1365
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.3219817561268286e-07,
      "loss": 0.0023,
      "step": 1370
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.3038522606803878e-07,
      "loss": 0.002,
      "step": 1375
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.2857331384552887e-07,
      "loss": 0.0025,
      "step": 1380
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.2676253476765194e-07,
      "loss": 0.0018,
      "step": 1385
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.2495298459698094e-07,
      "loss": 0.0024,
      "step": 1390
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.2314475903109824e-07,
      "loss": 0.0021,
      "step": 1395
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.2133795369753476e-07,
      "loss": 0.0019,
      "step": 1400
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.1953266414871316e-07,
      "loss": 0.002,
      "step": 1405
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.1772898585689376e-07,
      "loss": 0.0021,
      "step": 1410
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.159270142091264e-07,
      "loss": 0.0016,
      "step": 1415
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.1412684450220518e-07,
      "loss": 0.0018,
      "step": 1420
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.123285719376292e-07,
      "loss": 0.002,
      "step": 1425
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.1053229161656772e-07,
      "loss": 0.002,
      "step": 1430
    },
    {
      "epoch": 0.16,
      "learning_rate": 2.0873809853483056e-07,
      "loss": 0.0017,
      "step": 1435
    },
    {
      "epoch": 0.17,
      "learning_rate": 2.0694608757784466e-07,
      "loss": 0.0024,
      "step": 1440
    },
    {
      "epoch": 0.17,
      "learning_rate": 2.0515635351563562e-07,
      "loss": 0.002,
      "step": 1445
    },
    {
      "epoch": 0.17,
      "learning_rate": 2.0336899099781632e-07,
      "loss": 0.0021,
      "step": 1450
    },
    {
      "epoch": 0.17,
      "learning_rate": 2.0158409454858103e-07,
      "loss": 0.0018,
      "step": 1455
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9980175856170638e-07,
      "loss": 0.0023,
      "step": 1460
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.980220772955602e-07,
      "loss": 0.0015,
      "step": 1465
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9624514486811548e-07,
      "loss": 0.0017,
      "step": 1470
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.944710552519742e-07,
      "loss": 0.0021,
      "step": 1475
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.926999022693965e-07,
      "loss": 0.002,
      "step": 1480
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9093177958733963e-07,
      "loss": 0.0022,
      "step": 1485
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8916678071250446e-07,
      "loss": 0.0027,
      "step": 1490
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8740499898638958e-07,
      "loss": 0.0019,
      "step": 1495
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8564652758035622e-07,
      "loss": 0.002,
      "step": 1500
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8389145949069951e-07,
      "loss": 0.0015,
      "step": 1505
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8213988753373145e-07,
      "loss": 0.002,
      "step": 1510
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.803919043408721e-07,
      "loss": 0.0019,
      "step": 1515
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.7864760235375035e-07,
      "loss": 0.0022,
      "step": 1520
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.7690707381931582e-07,
      "loss": 0.0021,
      "step": 1525
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.7517041078495992e-07,
      "loss": 0.0013,
      "step": 1530
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.73437705093648e-07,
      "loss": 0.0015,
      "step": 1535
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.7170904837906263e-07,
      "loss": 0.0017,
      "step": 1540
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.6998453206075708e-07,
      "loss": 0.0017,
      "step": 1545
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.682642473393211e-07,
      "loss": 0.0015,
      "step": 1550
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.665482851915573e-07,
      "loss": 0.0011,
      "step": 1555
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.6483673636567022e-07,
      "loss": 0.0015,
      "step": 1560
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.6312969137646715e-07,
      "loss": 0.0014,
      "step": 1565
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.61427240500571e-07,
      "loss": 0.0016,
      "step": 1570
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.5972947377164642e-07,
      "loss": 0.0013,
      "step": 1575
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.5803648097563787e-07,
      "loss": 0.0015,
      "step": 1580
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.5634835164602196e-07,
      "loss": 0.0015,
      "step": 1585
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.5466517505907207e-07,
      "loss": 0.0016,
      "step": 1590
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.5298704022913676e-07,
      "loss": 0.0014,
      "step": 1595
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.513140359039332e-07,
      "loss": 0.0017,
      "step": 1600
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.4964625055985264e-07,
      "loss": 0.0015,
      "step": 1605
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.4798377239728236e-07,
      "loss": 0.0016,
      "step": 1610
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.4632668933594028e-07,
      "loss": 0.002,
      "step": 1615
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.44675089010226e-07,
      "loss": 0.0015,
      "step": 1620
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.430290587645865e-07,
      "loss": 0.0012,
      "step": 1625
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.4138868564889573e-07,
      "loss": 0.0014,
      "step": 1630
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.3975405641385253e-07,
      "loss": 0.0016,
      "step": 1635
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.381252575063919e-07,
      "loss": 0.0016,
      "step": 1640
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.365023750651133e-07,
      "loss": 0.0015,
      "step": 1645
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.3488549491572576e-07,
      "loss": 0.0015,
      "step": 1650
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.3327470256650846e-07,
      "loss": 0.0016,
      "step": 1655
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.3167008320378916e-07,
      "loss": 0.0014,
      "step": 1660
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.3007172168743852e-07,
      "loss": 0.002,
      "step": 1665
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.2847970254638263e-07,
      "loss": 0.0012,
      "step": 1670
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.2689410997413325e-07,
      "loss": 0.0014,
      "step": 1675
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.2531502782433416e-07,
      "loss": 0.0018,
      "step": 1680
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.2374253960632754e-07,
      "loss": 0.0012,
      "step": 1685
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.22176728480737e-07,
      "loss": 0.002,
      "step": 1690
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.2061767725507004e-07,
      "loss": 0.0014,
      "step": 1695
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.1906546837933867e-07,
      "loss": 0.0015,
      "step": 1700
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.175201839416988e-07,
      "loss": 0.0019,
      "step": 1705
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.1598190566410946e-07,
      "loss": 0.0015,
      "step": 1710
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.1445071489801073e-07,
      "loss": 0.0015,
      "step": 1715
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.1292669262002158e-07,
      "loss": 0.0016,
      "step": 1720
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.1140991942765713e-07,
      "loss": 0.0014,
      "step": 1725
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.0990047553506676e-07,
      "loss": 0.0017,
      "step": 1730
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.0839844076879185e-07,
      "loss": 0.0012,
      "step": 1735
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.0690389456354367e-07,
      "loss": 0.0012,
      "step": 1740
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.0541691595800336e-07,
      "loss": 0.0015,
      "step": 1745
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.0393758359064144e-07,
      "loss": 0.0014,
      "step": 1750
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.0246597569555892e-07,
      "loss": 0.0017,
      "step": 1755
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.0100217009835039e-07,
      "loss": 0.002,
      "step": 1760
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.95462442119879e-08,
      "loss": 0.0016,
      "step": 1765
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.809827503272713e-08,
      "loss": 0.0011,
      "step": 1770
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.665833913603522e-08,
      "loss": 0.0016,
      "step": 1775
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.522651267254147e-08,
      "loss": 0.0018,
      "step": 1780
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.380287136400999e-08,
      "loss": 0.0014,
      "step": 1785
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.238749049933484e-08,
      "loss": 0.0012,
      "step": 1790
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.098044493055898e-08,
      "loss": 0.0014,
      "step": 1795
    },
    {
      "epoch": 0.21,
      "learning_rate": 8.958180906891547e-08,
      "loss": 0.0014,
      "step": 1800
    },
    {
      "epoch": 0.21,
      "learning_rate": 8.819165688089192e-08,
      "loss": 0.0012,
      "step": 1805
    },
    {
      "epoch": 0.21,
      "learning_rate": 8.681006188431945e-08,
      "loss": 0.0017,
      "step": 1810
    },
    {
      "epoch": 0.21,
      "learning_rate": 8.543709714448402e-08,
      "loss": 0.0012,
      "step": 1815
    },
    {
      "epoch": 0.21,
      "learning_rate": 8.407283527026324e-08,
      "loss": 0.0017,
      "step": 1820
    },
    {
      "epoch": 0.21,
      "learning_rate": 8.271734841028552e-08,
      "loss": 0.0011,
      "step": 1825
    },
    {
      "epoch": 0.21,
      "learning_rate": 8.137070824911504e-08,
      "loss": 0.0014,
      "step": 1830
    },
    {
      "epoch": 0.21,
      "learning_rate": 8.003298600346085e-08,
      "loss": 0.0011,
      "step": 1835
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.870425241841019e-08,
      "loss": 0.0016,
      "step": 1840
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.738457776368765e-08,
      "loss": 0.0013,
      "step": 1845
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.607403182993821e-08,
      "loss": 0.0018,
      "step": 1850
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.477268392503728e-08,
      "loss": 0.0015,
      "step": 1855
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.348060287042481e-08,
      "loss": 0.0011,
      "step": 1860
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.219785699746572e-08,
      "loss": 0.0016,
      "step": 1865
    },
    {
      "epoch": 0.21,
      "learning_rate": 7.092451414383643e-08,
      "loss": 0.0013,
      "step": 1870
    },
    {
      "epoch": 0.22,
      "learning_rate": 6.966064164993715e-08,
      "loss": 0.0015,
      "step": 1875
    },
    {
      "epoch": 0.22,
      "learning_rate": 6.84063063553307e-08,
      "loss": 0.0011,
      "step": 1880
    },
    {
      "epoch": 0.22,
      "learning_rate": 6.716157459520738e-08,
      "loss": 0.0015,
      "step": 1885
    },
    {
      "epoch": 0.22,
      "learning_rate": 6.592651219687733e-08,
      "loss": 0.0014,
      "step": 1890
    },
    {
      "epoch": 0.22,
      "learning_rate": 6.470118447628911e-08,
      "loss": 0.0015,
      "step": 1895
    },
    {
      "epoch": 0.22,
      "learning_rate": 6.348565623457513e-08,
      "loss": 0.0011,
      "step": 1900
    },
    {
      "epoch": 0.22,
      "learning_rate": 6.22799917546252e-08,
      "loss": 0.0014,
      "step": 1905
    },
    {
      "epoch": 0.22,
      "learning_rate": 6.108425479768668e-08,
      "loss": 0.0013,
      "step": 1910
    },
    {
      "epoch": 0.22,
      "learning_rate": 5.989850859999227e-08,
      "loss": 0.0012,
      "step": 1915
    },
    {
      "epoch": 0.22,
      "learning_rate": 5.872281586941633e-08,
      "loss": 0.0013,
      "step": 1920
    },
    {
      "epoch": 0.22,
      "learning_rate": 5.755723878215801e-08,
      "loss": 0.0015,
      "step": 1925
    },
    {
      "epoch": 0.22,
      "learning_rate": 5.640183897945361e-08,
      "loss": 0.0013,
      "step": 1930
    },
    {
      "epoch": 0.22,
      "learning_rate": 5.525667756431615e-08,
      "loss": 0.0014,
      "step": 1935
    },
    {
      "epoch": 0.22,
      "learning_rate": 5.4121815098304186e-08,
      "loss": 0.0013,
      "step": 1940
    },
    {
      "epoch": 0.22,
      "learning_rate": 5.299731159831952e-08,
      "loss": 0.0014,
      "step": 1945
    },
    {
      "epoch": 0.22,
      "learning_rate": 5.18832265334323e-08,
      "loss": 0.0015,
      "step": 1950
    },
    {
      "epoch": 0.22,
      "learning_rate": 5.077961882173676e-08,
      "loss": 0.0015,
      "step": 1955
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.968654682723486e-08,
      "loss": 0.0014,
      "step": 1960
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.860406835675016e-08,
      "loss": 0.0018,
      "step": 1965
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.753224065687047e-08,
      "loss": 0.0013,
      "step": 1970
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.647112041092022e-08,
      "loss": 0.0013,
      "step": 1975
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.542076373596318e-08,
      "loss": 0.0012,
      "step": 1980
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.438122617983442e-08,
      "loss": 0.0011,
      "step": 1985
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.3352562718202866e-08,
      "loss": 0.0019,
      "step": 1990
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.233482775166364e-08,
      "loss": 0.0013,
      "step": 1995
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.132807510286143e-08,
      "loss": 0.001,
      "step": 2000
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.0332358013644015e-08,
      "loss": 0.0014,
      "step": 2005
    },
    {
      "epoch": 0.23,
      "learning_rate": 3.9347729142246325e-08,
      "loss": 0.0011,
      "step": 2010
    },
    {
      "epoch": 0.23,
      "learning_rate": 3.8374240560505974e-08,
      "loss": 0.0013,
      "step": 2015
    },
    {
      "epoch": 0.23,
      "learning_rate": 3.7411943751109314e-08,
      "loss": 0.0012,
      "step": 2020
    },
    {
      "epoch": 0.23,
      "learning_rate": 3.646088960486862e-08,
      "loss": 0.0011,
      "step": 2025
    },
    {
      "epoch": 0.23,
      "learning_rate": 3.552112841803104e-08,
      "loss": 0.0016,
      "step": 2030
    },
    {
      "epoch": 0.23,
      "learning_rate": 3.459270988961854e-08,
      "loss": 0.0014,
      "step": 2035
    },
    {
      "epoch": 0.23,
      "learning_rate": 3.367568311879959e-08,
      "loss": 0.0013,
      "step": 2040
    },
    {
      "epoch": 0.23,
      "learning_rate": 3.2770096602292464e-08,
      "loss": 0.0013,
      "step": 2045
    },
    {
      "epoch": 0.24,
      "learning_rate": 3.1875998231800704e-08,
      "loss": 0.0012,
      "step": 2050
    },
    {
      "epoch": 0.24,
      "learning_rate": 3.099343529148035e-08,
      "loss": 0.0012,
      "step": 2055
    },
    {
      "epoch": 0.24,
      "learning_rate": 3.0122454455439096e-08,
      "loss": 0.001,
      "step": 2060
    },
    {
      "epoch": 0.24,
      "learning_rate": 2.9263101785268252e-08,
      "loss": 0.001,
      "step": 2065
    },
    {
      "epoch": 0.24,
      "learning_rate": 2.8415422727606638e-08,
      "loss": 0.002,
      "step": 2070
    },
    {
      "epoch": 0.24,
      "learning_rate": 2.7579462111737063e-08,
      "loss": 0.0011,
      "step": 2075
    },
    {
      "epoch": 0.24,
      "learning_rate": 2.6755264147215794e-08,
      "loss": 0.0011,
      "step": 2080
    },
    {
      "epoch": 0.24,
      "learning_rate": 2.5942872421534147e-08,
      "loss": 0.0012,
      "step": 2085
    },
    {
      "epoch": 0.24,
      "learning_rate": 2.5142329897813952e-08,
      "loss": 0.0013,
      "step": 2090
    },
    {
      "epoch": 0.24,
      "learning_rate": 2.4353678912534896e-08,
      "loss": 0.0012,
      "step": 2095
    },
    {
      "epoch": 0.24,
      "learning_rate": 2.3576961173295772e-08,
      "loss": 0.0011,
      "step": 2100
    },
    {
      "epoch": 0.24,
      "learning_rate": 2.2812217756608937e-08,
      "loss": 0.0014,
      "step": 2105
    },
    {
      "epoch": 0.24,
      "learning_rate": 2.2059489105727856e-08,
      "loss": 0.0013,
      "step": 2110
    },
    {
      "epoch": 0.24,
      "learning_rate": 2.131881502850824e-08,
      "loss": 0.0014,
      "step": 2115
    },
    {
      "epoch": 0.24,
      "learning_rate": 2.0590234695302827e-08,
      "loss": 0.0009,
      "step": 2120
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.9873786636889904e-08,
      "loss": 0.0011,
      "step": 2125
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.916950874243575e-08,
      "loss": 0.0011,
      "step": 2130
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.847743825749054e-08,
      "loss": 0.0013,
      "step": 2135
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.779761178201894e-08,
      "loss": 0.0013,
      "step": 2140
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.713006526846439e-08,
      "loss": 0.0013,
      "step": 2145
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.6474834019847865e-08,
      "loss": 0.0015,
      "step": 2150
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.5831952687900606e-08,
      "loss": 0.0012,
      "step": 2155
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.5201455271231955e-08,
      "loss": 0.0012,
      "step": 2160
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.4583375113531194e-08,
      "loss": 0.0012,
      "step": 2165
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.3977744901803951e-08,
      "loss": 0.0011,
      "step": 2170
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.3384596664643921e-08,
      "loss": 0.001,
      "step": 2175
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.2803961770538885e-08,
      "loss": 0.0012,
      "step": 2180
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.2235870926211616e-08,
      "loss": 0.0011,
      "step": 2185
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.1680354174996299e-08,
      "loss": 0.0015,
      "step": 2190
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.1137440895249456e-08,
      "loss": 0.0012,
      "step": 2195
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.0607159798796394e-08,
      "loss": 0.0016,
      "step": 2200
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.0089538929412723e-08,
      "loss": 0.0014,
      "step": 2205
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.584605661341144e-09,
      "loss": 0.001,
      "step": 2210
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.092386697844262e-09,
      "loss": 0.0011,
      "step": 2215
    },
    {
      "epoch": 0.25,
      "learning_rate": 8.612908069791703e-09,
      "loss": 0.0014,
      "step": 2220
    },
    {
      "epoch": 0.26,
      "learning_rate": 8.14619513428405e-09,
      "loss": 0.0013,
      "step": 2225
    },
    {
      "epoch": 0.26,
      "learning_rate": 7.692272573311426e-09,
      "loss": 0.0014,
      "step": 2230
    },
    {
      "epoch": 0.26,
      "learning_rate": 7.251164392448495e-09,
      "loss": 0.0011,
      "step": 2235
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.822893919584877e-09,
      "loss": 0.0011,
      "step": 2240
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.407483803691216e-09,
      "loss": 0.0012,
      "step": 2245
    },
    {
      "epoch": 0.26,
      "learning_rate": 6.0049560136216926e-09,
      "loss": 0.0012,
      "step": 2250
    },
    {
      "epoch": 0.26,
      "learning_rate": 5.615331836952119e-09,
      "loss": 0.001,
      "step": 2255
    },
    {
      "epoch": 0.26,
      "learning_rate": 5.238631878854038e-09,
      "loss": 0.0013,
      "step": 2260
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.874876061005173e-09,
      "loss": 0.0011,
      "step": 2265
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.5240836205357735e-09,
      "loss": 0.001,
      "step": 2270
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.186273109011373e-09,
      "loss": 0.0012,
      "step": 2275
    },
    {
      "epoch": 0.26,
      "learning_rate": 3.861462391451492e-09,
      "loss": 0.0013,
      "step": 2280
    },
    {
      "epoch": 0.26,
      "learning_rate": 3.5496686453850843e-09,
      "loss": 0.0011,
      "step": 2285
    },
    {
      "epoch": 0.26,
      "learning_rate": 3.250908359942045e-09,
      "loss": 0.001,
      "step": 2290
    },
    {
      "epoch": 0.26,
      "learning_rate": 2.9651973349810177e-09,
      "loss": 0.0012,
      "step": 2295
    },
    {
      "epoch": 0.26,
      "learning_rate": 2.6925506802540353e-09,
      "loss": 0.0012,
      "step": 2300
    },
    {
      "epoch": 0.26,
      "learning_rate": 2.4329828146074096e-09,
      "loss": 0.0011,
      "step": 2305
    },
    {
      "epoch": 0.27,
      "learning_rate": 2.1865074652190894e-09,
      "loss": 0.0013,
      "step": 2310
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.9531376668727174e-09,
      "loss": 0.0021,
      "step": 2315
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.7328857612684266e-09,
      "loss": 0.0013,
      "step": 2320
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.5257633963700055e-09,
      "loss": 0.0011,
      "step": 2325
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.3317815257889998e-09,
      "loss": 0.0016,
      "step": 2330
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.1509504082052867e-09,
      "loss": 0.0016,
      "step": 2335
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.832796068247883e-10,
      "loss": 0.0015,
      "step": 2340
    },
    {
      "epoch": 0.27,
      "learning_rate": 8.287779888734858e-10,
      "loss": 0.0015,
      "step": 2345
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.874537251286005e-10,
      "loss": 0.001,
      "step": 2350
    },
    {
      "epoch": 0.27,
      "learning_rate": 5.593142894864122e-10,
      "loss": 0.0011,
      "step": 2355
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.443664585671858e-10,
      "loss": 0.0015,
      "step": 2360
    },
    {
      "epoch": 0.27,
      "learning_rate": 3.4261631135654167e-10,
      "loss": 0.0012,
      "step": 2365
    },
    {
      "epoch": 0.27,
      "learning_rate": 2.54069228884074e-10,
      "loss": 0.0012,
      "step": 2370
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.7872989393888372e-10,
      "loss": 0.0014,
      "step": 2375
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.1660229082177675e-10,
      "loss": 0.0012,
      "step": 2380
    },
    {
      "epoch": 0.27,
      "learning_rate": 6.768970513457151e-11,
      "loss": 0.001,
      "step": 2385
    },
    {
      "epoch": 0.27,
      "learning_rate": 3.19947236064877e-11,
      "loss": 0.0011,
      "step": 2390
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.519233957172579e-12,
      "loss": 0.0011,
      "step": 2395
    },
    {
      "epoch": 0.28,
      "learning_rate": 2.644247969474378e-13,
      "loss": 0.0013,
      "step": 2400
    },
    {
      "epoch": 0.28,
      "step": 2400,
      "total_flos": 3.3948129315510026e+18,
      "train_loss": 0.09296217595246466,
      "train_runtime": 26220.1068,
      "train_samples_per_second": 1.465,
      "train_steps_per_second": 0.092
    }
  ],
  "logging_steps": 5,
  "max_steps": 2400,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "total_flos": 3.3948129315510026e+18,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}