{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 49.858356940509914,
  "global_step": 4400,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.11,
      "learning_rate": 1.9999745104274995e-05,
      "loss": 1.1665,
      "step": 10
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.9998980430094333e-05,
      "loss": 1.1142,
      "step": 20
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.9997706016440462e-05,
      "loss": 1.0657,
      "step": 30
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.9995921928281893e-05,
      "loss": 1.0229,
      "step": 40
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.999362825656992e-05,
      "loss": 1.0155,
      "step": 50
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.9990825118233958e-05,
      "loss": 0.9919,
      "step": 60
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.9987512656175612e-05,
      "loss": 0.9623,
      "step": 70
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.9983691039261358e-05,
      "loss": 0.937,
      "step": 80
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.9979360462313965e-05,
      "loss": 0.9527,
      "step": 90
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.9974521146102535e-05,
      "loss": 0.9173,
      "step": 100
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.9969173337331283e-05,
      "loss": 0.91,
      "step": 110
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.9963317308626916e-05,
      "loss": 0.9045,
      "step": 120
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.9956953358524774e-05,
      "loss": 0.94,
      "step": 130
    },
    {
      "epoch": 1.59,
      "learning_rate": 1.9950081811453598e-05,
      "loss": 0.9086,
      "step": 140
    },
    {
      "epoch": 1.7,
      "learning_rate": 1.9942703017718977e-05,
      "loss": 0.8946,
      "step": 150
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.99348173534855e-05,
      "loss": 0.8836,
      "step": 160
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.9926425220757607e-05,
      "loss": 0.9001,
      "step": 170
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.991752704735903e-05,
      "loss": 0.8826,
      "step": 180
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.990812328691105e-05,
      "loss": 0.8549,
      "step": 190
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.989821441880933e-05,
      "loss": 0.869,
      "step": 200
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.9887800948199496e-05,
      "loss": 0.8772,
      "step": 210
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.9876883405951378e-05,
      "loss": 0.8691,
      "step": 220
    },
    {
      "epoch": 2.61,
      "learning_rate": 1.9865462348631945e-05,
      "loss": 0.8978,
      "step": 230
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.9853538358476933e-05,
      "loss": 0.8576,
      "step": 240
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.984111204336116e-05,
      "loss": 0.8506,
      "step": 250
    },
    {
      "epoch": 2.95,
      "learning_rate": 1.9828184036767556e-05,
      "loss": 0.8679,
      "step": 260
    },
    {
      "epoch": 3.06,
      "learning_rate": 1.981475499775484e-05,
      "loss": 0.888,
      "step": 270
    },
    {
      "epoch": 3.17,
      "learning_rate": 1.9800825610923937e-05,
      "loss": 0.8369,
      "step": 280
    },
    {
      "epoch": 3.29,
      "learning_rate": 1.9786396586383078e-05,
      "loss": 0.8244,
      "step": 290
    },
    {
      "epoch": 3.4,
      "learning_rate": 1.9771468659711595e-05,
      "loss": 0.8495,
      "step": 300
    },
    {
      "epoch": 3.51,
      "learning_rate": 1.9756042591922436e-05,
      "loss": 0.85,
      "step": 310
    },
    {
      "epoch": 3.63,
      "learning_rate": 1.9740119169423337e-05,
      "loss": 0.8307,
      "step": 320
    },
    {
      "epoch": 3.74,
      "learning_rate": 1.9723699203976768e-05,
      "loss": 0.8419,
      "step": 330
    },
    {
      "epoch": 3.85,
      "learning_rate": 1.9706783532658528e-05,
      "loss": 0.8483,
      "step": 340
    },
    {
      "epoch": 3.97,
      "learning_rate": 1.9689373017815076e-05,
      "loss": 0.8181,
      "step": 350
    },
    {
      "epoch": 4.08,
      "learning_rate": 1.9671468547019575e-05,
      "loss": 0.8203,
      "step": 360
    },
    {
      "epoch": 4.19,
      "learning_rate": 1.9653071033026635e-05,
      "loss": 0.8132,
      "step": 370
    },
    {
      "epoch": 4.31,
      "learning_rate": 1.963418141372579e-05,
      "loss": 0.8299,
      "step": 380
    },
    {
      "epoch": 4.42,
      "learning_rate": 1.9614800652093685e-05,
      "loss": 0.8228,
      "step": 390
    },
    {
      "epoch": 4.53,
      "learning_rate": 1.9594929736144978e-05,
      "loss": 0.8083,
      "step": 400
    },
    {
      "epoch": 4.65,
      "learning_rate": 1.9574569678881965e-05,
      "loss": 0.8054,
      "step": 410
    },
    {
      "epoch": 4.76,
      "learning_rate": 1.955372151824297e-05,
      "loss": 0.8127,
      "step": 420
    },
    {
      "epoch": 4.87,
      "learning_rate": 1.9532386317049387e-05,
      "loss": 0.8047,
      "step": 430
    },
    {
      "epoch": 4.99,
      "learning_rate": 1.9510565162951538e-05,
      "loss": 0.8113,
      "step": 440
    },
    {
      "epoch": 5.1,
      "learning_rate": 1.9488259168373198e-05,
      "loss": 0.7806,
      "step": 450
    },
    {
      "epoch": 5.21,
      "learning_rate": 1.94654694704549e-05,
      "loss": 0.7973,
      "step": 460
    },
    {
      "epoch": 5.33,
      "learning_rate": 1.944219723099596e-05,
      "loss": 0.7968,
      "step": 470
    },
    {
      "epoch": 5.44,
      "learning_rate": 1.941844363639525e-05,
      "loss": 0.7838,
      "step": 480
    },
    {
      "epoch": 5.55,
      "learning_rate": 1.9394209897590707e-05,
      "loss": 0.7798,
      "step": 490
    },
    {
      "epoch": 5.67,
      "learning_rate": 1.936949724999762e-05,
      "loss": 0.7791,
      "step": 500
    },
    {
      "epoch": 5.78,
      "learning_rate": 1.9344306953445632e-05,
      "loss": 0.7632,
      "step": 510
    },
    {
      "epoch": 5.89,
      "learning_rate": 1.9318640292114526e-05,
      "loss": 0.7984,
      "step": 520
    },
    {
      "epoch": 6.01,
      "learning_rate": 1.929249857446875e-05,
      "loss": 0.7763,
      "step": 530
    },
    {
      "epoch": 6.12,
      "learning_rate": 1.9265883133190715e-05,
      "loss": 0.7489,
      "step": 540
    },
    {
      "epoch": 6.23,
      "learning_rate": 1.9238795325112867e-05,
      "loss": 0.7739,
      "step": 550
    },
    {
      "epoch": 6.35,
      "learning_rate": 1.92112365311485e-05,
      "loss": 0.7664,
      "step": 560
    },
    {
      "epoch": 6.46,
      "learning_rate": 1.918320815622137e-05,
      "loss": 0.76,
      "step": 570
    },
    {
      "epoch": 6.57,
      "learning_rate": 1.9154711629194062e-05,
      "loss": 0.7604,
      "step": 580
    },
    {
      "epoch": 6.69,
      "learning_rate": 1.912574840279516e-05,
      "loss": 0.7598,
      "step": 590
    },
    {
      "epoch": 6.8,
      "learning_rate": 1.9096319953545186e-05,
      "loss": 0.7585,
      "step": 600
    },
    {
      "epoch": 6.91,
      "learning_rate": 1.9066427781681314e-05,
      "loss": 0.7312,
      "step": 610
    },
    {
      "epoch": 7.03,
      "learning_rate": 1.9036073411080917e-05,
      "loss": 0.732,
      "step": 620
    },
    {
      "epoch": 7.14,
      "learning_rate": 1.900525838918385e-05,
      "loss": 0.7392,
      "step": 630
    },
    {
      "epoch": 7.25,
      "learning_rate": 1.8973984286913584e-05,
      "loss": 0.6986,
      "step": 640
    },
    {
      "epoch": 7.37,
      "learning_rate": 1.8942252698597113e-05,
      "loss": 0.7281,
      "step": 650
    },
    {
      "epoch": 7.48,
      "learning_rate": 1.891006524188368e-05,
      "loss": 0.7316,
      "step": 660
    },
    {
      "epoch": 7.59,
      "learning_rate": 1.8877423557662307e-05,
      "loss": 0.7329,
      "step": 670
    },
    {
      "epoch": 7.71,
      "learning_rate": 1.8844329309978146e-05,
      "loss": 0.7409,
      "step": 680
    },
    {
      "epoch": 7.82,
      "learning_rate": 1.8810784185947648e-05,
      "loss": 0.7345,
      "step": 690
    },
    {
      "epoch": 7.93,
      "learning_rate": 1.8776789895672557e-05,
      "loss": 0.7418,
      "step": 700
    },
    {
      "epoch": 8.05,
      "learning_rate": 1.8742348172152728e-05,
      "loss": 0.7426,
      "step": 710
    },
    {
      "epoch": 8.16,
      "learning_rate": 1.8707460771197773e-05,
      "loss": 0.6952,
      "step": 720
    },
    {
      "epoch": 8.27,
      "learning_rate": 1.8672129471337568e-05,
      "loss": 0.7136,
      "step": 730
    },
    {
      "epoch": 8.39,
      "learning_rate": 1.863635607373157e-05,
      "loss": 0.7284,
      "step": 740
    },
    {
      "epoch": 8.5,
      "learning_rate": 1.8600142402077006e-05,
      "loss": 0.6978,
      "step": 750
    },
    {
      "epoch": 8.61,
      "learning_rate": 1.856349030251589e-05,
      "loss": 0.6843,
      "step": 760
    },
    {
      "epoch": 8.73,
      "learning_rate": 1.8526401643540924e-05,
      "loss": 0.6895,
      "step": 770
    },
    {
      "epoch": 8.84,
      "learning_rate": 1.8488878315900228e-05,
      "loss": 0.7084,
      "step": 780
    },
    {
      "epoch": 8.95,
      "learning_rate": 1.8450922232500966e-05,
      "loss": 0.7028,
      "step": 790
    },
    {
      "epoch": 9.07,
      "learning_rate": 1.8412535328311813e-05,
      "loss": 0.6905,
      "step": 800
    },
    {
      "epoch": 9.18,
      "learning_rate": 1.837371956026433e-05,
      "loss": 0.6851,
      "step": 810
    },
    {
      "epoch": 9.29,
      "learning_rate": 1.8334476907153177e-05,
      "loss": 0.6759,
      "step": 820
    },
    {
      "epoch": 9.41,
      "learning_rate": 1.8294809369535265e-05,
      "loss": 0.6613,
      "step": 830
    },
    {
      "epoch": 9.52,
      "learning_rate": 1.825471896962774e-05,
      "loss": 0.6814,
      "step": 840
    },
    {
      "epoch": 9.63,
      "learning_rate": 1.8214207751204917e-05,
      "loss": 0.7044,
      "step": 850
    },
    {
      "epoch": 9.75,
      "learning_rate": 1.817327777949407e-05,
      "loss": 0.6753,
      "step": 860
    },
    {
      "epoch": 9.86,
      "learning_rate": 1.8131931141070166e-05,
      "loss": 0.6748,
      "step": 870
    },
    {
      "epoch": 9.97,
      "learning_rate": 1.8090169943749477e-05,
      "loss": 0.6853,
      "step": 880
    },
    {
      "epoch": 10.08,
      "learning_rate": 1.8047996316482134e-05,
      "loss": 0.6406,
      "step": 890
    },
    {
      "epoch": 10.2,
      "learning_rate": 1.8005412409243604e-05,
      "loss": 0.6391,
      "step": 900
    },
    {
      "epoch": 10.31,
      "learning_rate": 1.7962420392925066e-05,
      "loss": 0.6524,
      "step": 910
    },
    {
      "epoch": 10.42,
      "learning_rate": 1.7919022459222754e-05,
      "loss": 0.6618,
      "step": 920
    },
    {
      "epoch": 10.54,
      "learning_rate": 1.787522082052622e-05,
      "loss": 0.6604,
      "step": 930
    },
    {
      "epoch": 10.65,
      "learning_rate": 1.7831017709805555e-05,
      "loss": 0.6706,
      "step": 940
    },
    {
      "epoch": 10.76,
      "learning_rate": 1.778641538049755e-05,
      "loss": 0.6589,
      "step": 950
    },
    {
      "epoch": 10.88,
      "learning_rate": 1.7741416106390828e-05,
      "loss": 0.6631,
      "step": 960
    },
    {
      "epoch": 10.99,
      "learning_rate": 1.7696022181509892e-05,
      "loss": 0.6408,
      "step": 970
    },
    {
      "epoch": 11.1,
      "learning_rate": 1.7650235919998234e-05,
      "loss": 0.6077,
      "step": 980
    },
    {
      "epoch": 11.22,
      "learning_rate": 1.7604059656000313e-05,
      "loss": 0.6267,
      "step": 990
    },
    {
      "epoch": 11.33,
      "learning_rate": 1.7557495743542586e-05,
      "loss": 0.6324,
      "step": 1000
    },
    {
      "epoch": 11.44,
      "learning_rate": 1.75105465564135e-05,
      "loss": 0.645,
      "step": 1010
    },
    {
      "epoch": 11.56,
      "learning_rate": 1.7463214488042472e-05,
      "loss": 0.6167,
      "step": 1020
    },
    {
      "epoch": 11.67,
      "learning_rate": 1.741550195137788e-05,
      "loss": 0.6311,
      "step": 1030
    },
    {
      "epoch": 11.78,
      "learning_rate": 1.736741137876405e-05,
      "loss": 0.6329,
      "step": 1040
    },
    {
      "epoch": 11.9,
      "learning_rate": 1.7318945221817255e-05,
      "loss": 0.6469,
      "step": 1050
    },
    {
      "epoch": 12.01,
      "learning_rate": 1.727010595130074e-05,
      "loss": 0.6334,
      "step": 1060
    },
    {
      "epoch": 12.12,
      "learning_rate": 1.7220896056998753e-05,
      "loss": 0.5946,
      "step": 1070
    },
    {
      "epoch": 12.24,
      "learning_rate": 1.7171318047589637e-05,
      "loss": 0.5695,
      "step": 1080
    },
    {
      "epoch": 12.35,
      "learning_rate": 1.712137445051792e-05,
      "loss": 0.6101,
      "step": 1090
    },
    {
      "epoch": 12.46,
      "learning_rate": 1.7071067811865477e-05,
      "loss": 0.6203,
      "step": 1100
    },
    {
      "epoch": 12.58,
      "learning_rate": 1.7020400696221737e-05,
      "loss": 0.598,
      "step": 1110
    },
    {
      "epoch": 12.69,
      "learning_rate": 1.696937568655294e-05,
      "loss": 0.6177,
      "step": 1120
    },
    {
      "epoch": 12.8,
      "learning_rate": 1.691799538407044e-05,
      "loss": 0.6153,
      "step": 1130
    },
    {
      "epoch": 12.92,
      "learning_rate": 1.6866262408098134e-05,
      "loss": 0.6096,
      "step": 1140
    },
    {
      "epoch": 13.03,
      "learning_rate": 1.6814179395938915e-05,
      "loss": 0.612,
      "step": 1150
    },
    {
      "epoch": 13.14,
      "learning_rate": 1.6761749002740195e-05,
      "loss": 0.5858,
      "step": 1160
    },
    {
      "epoch": 13.26,
      "learning_rate": 1.6708973901358603e-05,
      "loss": 0.5715,
      "step": 1170
    },
    {
      "epoch": 13.37,
      "learning_rate": 1.6655856782223682e-05,
      "loss": 0.5834,
      "step": 1180
    },
    {
      "epoch": 13.48,
      "learning_rate": 1.660240035320075e-05,
      "loss": 0.5782,
      "step": 1190
    },
    {
      "epoch": 13.6,
      "learning_rate": 1.6548607339452853e-05,
      "loss": 0.5621,
      "step": 1200
    },
    {
      "epoch": 13.71,
      "learning_rate": 1.6494480483301836e-05,
      "loss": 0.5729,
      "step": 1210
    },
    {
      "epoch": 13.82,
      "learning_rate": 1.6440022544088553e-05,
      "loss": 0.5958,
      "step": 1220
    },
    {
      "epoch": 13.94,
      "learning_rate": 1.6385236298032183e-05,
      "loss": 0.6007,
      "step": 1230
    },
    {
      "epoch": 14.05,
      "learning_rate": 1.6330124538088705e-05,
      "loss": 0.5658,
      "step": 1240
    },
    {
      "epoch": 14.16,
      "learning_rate": 1.627469007380852e-05,
      "loss": 0.5674,
      "step": 1250
    },
    {
      "epoch": 14.28,
      "learning_rate": 1.6218935731193223e-05,
      "loss": 0.5649,
      "step": 1260
    },
    {
      "epoch": 14.39,
      "learning_rate": 1.616286435255153e-05,
      "loss": 0.5497,
      "step": 1270
    },
    {
      "epoch": 14.5,
      "learning_rate": 1.6106478796354382e-05,
      "loss": 0.5865,
      "step": 1280
    },
    {
      "epoch": 14.62,
      "learning_rate": 1.6049781937089227e-05,
      "loss": 0.547,
      "step": 1290
    },
    {
      "epoch": 14.73,
      "learning_rate": 1.599277666511347e-05,
      "loss": 0.5473,
      "step": 1300
    },
    {
      "epoch": 14.84,
      "learning_rate": 1.5935465886507143e-05,
      "loss": 0.5553,
      "step": 1310
    },
    {
      "epoch": 14.96,
      "learning_rate": 1.5877852522924733e-05,
      "loss": 0.5672,
      "step": 1320
    },
    {
      "epoch": 15.07,
      "learning_rate": 1.581993951144626e-05,
      "loss": 0.5392,
      "step": 1330
    },
    {
      "epoch": 15.18,
      "learning_rate": 1.576172980442753e-05,
      "loss": 0.504,
      "step": 1340
    },
    {
      "epoch": 15.3,
      "learning_rate": 1.5703226369349642e-05,
      "loss": 0.5446,
      "step": 1350
    },
    {
      "epoch": 15.41,
      "learning_rate": 1.5644432188667695e-05,
      "loss": 0.5323,
      "step": 1360
    },
    {
      "epoch": 15.52,
      "learning_rate": 1.5585350259658748e-05,
      "loss": 0.54,
      "step": 1370
    },
    {
      "epoch": 15.64,
      "learning_rate": 1.5525983594269026e-05,
      "loss": 0.5387,
      "step": 1380
    },
    {
      "epoch": 15.75,
      "learning_rate": 1.546633521896038e-05,
      "loss": 0.5349,
      "step": 1390
    },
    {
      "epoch": 15.86,
      "learning_rate": 1.5406408174555978e-05,
      "loss": 0.5303,
      "step": 1400
    },
    {
      "epoch": 15.98,
      "learning_rate": 1.5346205516085305e-05,
      "loss": 0.561,
      "step": 1410
    },
    {
      "epoch": 16.09,
      "learning_rate": 1.528573031262842e-05,
      "loss": 0.5295,
      "step": 1420
    },
    {
      "epoch": 16.2,
      "learning_rate": 1.5224985647159489e-05,
      "loss": 0.4995,
      "step": 1430
    },
    {
      "epoch": 16.32,
      "learning_rate": 1.5163974616389621e-05,
      "loss": 0.5151,
      "step": 1440
    },
    {
      "epoch": 16.43,
      "learning_rate": 1.5102700330609e-05,
      "loss": 0.5042,
      "step": 1450
    },
    {
      "epoch": 16.54,
      "learning_rate": 1.504116591352832e-05,
      "loss": 0.502,
      "step": 1460
    },
    {
      "epoch": 16.66,
      "learning_rate": 1.497937450211956e-05,
      "loss": 0.5101,
      "step": 1470
    },
    {
      "epoch": 16.77,
      "learning_rate": 1.491732924645604e-05,
      "loss": 0.5323,
      "step": 1480
    },
    {
      "epoch": 16.88,
      "learning_rate": 1.4855033309551842e-05,
      "loss": 0.509,
      "step": 1490
    },
    {
      "epoch": 17.0,
      "learning_rate": 1.479248986720057e-05,
      "loss": 0.5292,
      "step": 1500
    },
    {
      "epoch": 17.11,
      "learning_rate": 1.4729702107813438e-05,
      "loss": 0.4835,
      "step": 1510
    },
    {
      "epoch": 17.22,
      "learning_rate": 1.4666673232256738e-05,
      "loss": 0.4934,
      "step": 1520
    },
    {
      "epoch": 17.34,
      "learning_rate": 1.4603406453688656e-05,
      "loss": 0.4811,
      "step": 1530
    },
    {
      "epoch": 17.45,
      "learning_rate": 1.4539904997395468e-05,
      "loss": 0.4966,
      "step": 1540
    },
    {
      "epoch": 17.56,
      "learning_rate": 1.4476172100627127e-05,
      "loss": 0.4906,
      "step": 1550
    },
    {
      "epoch": 17.68,
      "learning_rate": 1.4412211012432213e-05,
      "loss": 0.4956,
      "step": 1560
    },
    {
      "epoch": 17.79,
      "learning_rate": 1.4348024993492323e-05,
      "loss": 0.5003,
      "step": 1570
    },
    {
      "epoch": 17.9,
      "learning_rate": 1.4283617315955815e-05,
      "loss": 0.5131,
      "step": 1580
    },
    {
      "epoch": 18.02,
      "learning_rate": 1.4218991263271024e-05,
      "loss": 0.4642,
      "step": 1590
    },
    {
      "epoch": 18.13,
      "learning_rate": 1.4154150130018867e-05,
      "loss": 0.4679,
      "step": 1600
    },
    {
      "epoch": 18.24,
      "learning_rate": 1.408909722174487e-05,
      "loss": 0.4555,
      "step": 1610
    },
    {
      "epoch": 18.36,
      "learning_rate": 1.4023835854790682e-05,
      "loss": 0.4663,
      "step": 1620
    },
    {
      "epoch": 18.47,
      "learning_rate": 1.3958369356124986e-05,
      "loss": 0.4991,
      "step": 1630
    },
    {
      "epoch": 18.58,
      "learning_rate": 1.3892701063173917e-05,
      "loss": 0.4592,
      "step": 1640
    },
    {
      "epoch": 18.7,
      "learning_rate": 1.3826834323650899e-05,
      "loss": 0.4697,
      "step": 1650
    },
    {
      "epoch": 18.81,
      "learning_rate": 1.3760772495385998e-05,
      "loss": 0.4812,
      "step": 1660
    },
    {
      "epoch": 18.92,
      "learning_rate": 1.369451894615474e-05,
      "loss": 0.4868,
      "step": 1670
    },
    {
      "epoch": 19.04,
      "learning_rate": 1.362807705350641e-05,
      "loss": 0.4511,
      "step": 1680
    },
    {
      "epoch": 19.15,
      "learning_rate": 1.3561450204591898e-05,
      "loss": 0.4332,
      "step": 1690
    },
    {
      "epoch": 19.26,
      "learning_rate": 1.3494641795990986e-05,
      "loss": 0.4392,
      "step": 1700
    },
    {
      "epoch": 19.38,
      "learning_rate": 1.3427655233539227e-05,
      "loss": 0.435,
      "step": 1710
    },
    {
      "epoch": 19.49,
      "learning_rate": 1.3360493932154301e-05,
      "loss": 0.4377,
      "step": 1720
    },
    {
      "epoch": 19.6,
      "learning_rate": 1.3293161315661934e-05,
      "loss": 0.4509,
      "step": 1730
    },
    {
      "epoch": 19.72,
      "learning_rate": 1.3225660816621342e-05,
      "loss": 0.4564,
      "step": 1740
    },
    {
      "epoch": 19.83,
      "learning_rate": 1.3157995876150252e-05,
      "loss": 0.471,
      "step": 1750
    },
    {
      "epoch": 19.94,
      "learning_rate": 1.3090169943749475e-05,
      "loss": 0.4759,
      "step": 1760
    },
    {
      "epoch": 20.06,
      "learning_rate": 1.302218647712706e-05,
      "loss": 0.454,
      "step": 1770
    },
    {
      "epoch": 20.17,
      "learning_rate": 1.2954048942022002e-05,
      "loss": 0.3999,
      "step": 1780
    },
    {
      "epoch": 20.28,
      "learning_rate": 1.288576081202759e-05,
      "loss": 0.415,
      "step": 1790
    },
    {
      "epoch": 20.4,
      "learning_rate": 1.2817325568414299e-05,
      "loss": 0.4349,
      "step": 1800
    },
    {
      "epoch": 20.51,
      "learning_rate": 1.2748746699952338e-05,
      "loss": 0.4281,
      "step": 1810
    },
    {
      "epoch": 20.62,
      "learning_rate": 1.2680027702733791e-05,
      "loss": 0.4391,
      "step": 1820
    },
    {
      "epoch": 20.74,
      "learning_rate": 1.2611172079994377e-05,
      "loss": 0.4288,
      "step": 1830
    },
    {
      "epoch": 20.85,
      "learning_rate": 1.2542183341934873e-05,
      "loss": 0.4562,
      "step": 1840
    },
    {
      "epoch": 20.96,
      "learning_rate": 1.2473065005542155e-05,
      "loss": 0.4294,
      "step": 1850
    },
    {
      "epoch": 21.08,
      "learning_rate": 1.2403820594409926e-05,
      "loss": 0.4147,
      "step": 1860
    },
    {
      "epoch": 21.19,
      "learning_rate": 1.2334453638559057e-05,
      "loss": 0.4003,
      "step": 1870
    },
    {
      "epoch": 21.3,
      "learning_rate": 1.2264967674257647e-05,
      "loss": 0.402,
      "step": 1880
    },
    {
      "epoch": 21.42,
      "learning_rate": 1.2195366243840745e-05,
      "loss": 0.3808,
      "step": 1890
    },
    {
      "epoch": 21.53,
      "learning_rate": 1.2125652895529766e-05,
      "loss": 0.4199,
      "step": 1900
    },
    {
      "epoch": 21.64,
      "learning_rate": 1.2055831183251608e-05,
      "loss": 0.4153,
      "step": 1910
    },
    {
      "epoch": 21.76,
      "learning_rate": 1.1985904666457455e-05,
      "loss": 0.4206,
      "step": 1920
    },
    {
      "epoch": 21.87,
      "learning_rate": 1.1915876909941356e-05,
      "loss": 0.4326,
      "step": 1930
    },
    {
      "epoch": 21.98,
      "learning_rate": 1.1845751483658454e-05,
      "loss": 0.4263,
      "step": 1940
    },
    {
      "epoch": 22.1,
      "learning_rate": 1.1775531962543036e-05,
      "loss": 0.3938,
      "step": 1950
    },
    {
      "epoch": 22.21,
      "learning_rate": 1.170522192632624e-05,
      "loss": 0.3748,
      "step": 1960
    },
    {
      "epoch": 22.32,
      "learning_rate": 1.1634824959353602e-05,
      "loss": 0.3757,
      "step": 1970
    },
    {
      "epoch": 22.44,
      "learning_rate": 1.156434465040231e-05,
      "loss": 0.3876,
      "step": 1980
    },
    {
      "epoch": 22.55,
      "learning_rate": 1.1493784592498252e-05,
      "loss": 0.395,
      "step": 1990
    },
    {
      "epoch": 22.66,
      "learning_rate": 1.1423148382732854e-05,
      "loss": 0.3938,
      "step": 2000
    },
    {
      "epoch": 22.78,
      "learning_rate": 1.1352439622079689e-05,
      "loss": 0.4146,
      "step": 2010
    },
    {
      "epoch": 22.89,
      "learning_rate": 1.1281661915210931e-05,
      "loss": 0.4206,
      "step": 2020
    },
    {
      "epoch": 23.0,
      "learning_rate": 1.1210818870313548e-05,
      "loss": 0.3975,
      "step": 2030
    },
    {
      "epoch": 23.12,
      "learning_rate": 1.1139914098905406e-05,
      "loss": 0.3613,
      "step": 2040
    },
    {
      "epoch": 23.23,
      "learning_rate": 1.1068951215651132e-05,
      "loss": 0.3572,
      "step": 2050
    },
    {
      "epoch": 23.34,
      "learning_rate": 1.0997933838177828e-05,
      "loss": 0.3773,
      "step": 2060
    },
    {
      "epoch": 23.46,
      "learning_rate": 1.0926865586890689e-05,
      "loss": 0.3846,
      "step": 2070
    },
    {
      "epoch": 23.57,
      "learning_rate": 1.08557500847884e-05,
      "loss": 0.3758,
      "step": 2080
    },
    {
      "epoch": 23.68,
      "learning_rate": 1.0784590957278452e-05,
      "loss": 0.393,
      "step": 2090
    },
    {
      "epoch": 23.8,
      "learning_rate": 1.0713391831992324e-05,
      "loss": 0.3699,
      "step": 2100
    },
    {
      "epoch": 23.91,
      "learning_rate": 1.064215633860055e-05,
      "loss": 0.3639,
      "step": 2110
    },
    {
      "epoch": 24.02,
      "learning_rate": 1.0570888108627682e-05,
      "loss": 0.3869,
      "step": 2120
    },
    {
      "epoch": 24.14,
      "learning_rate": 1.0499590775267142e-05,
      "loss": 0.3387,
      "step": 2130
    },
    {
      "epoch": 24.25,
      "learning_rate": 1.0428267973196027e-05,
      "loss": 0.3507,
      "step": 2140
    },
    {
      "epoch": 24.36,
      "learning_rate": 1.0356923338389807e-05,
      "loss": 0.3424,
      "step": 2150
    },
    {
      "epoch": 24.48,
      "learning_rate": 1.0285560507936962e-05,
      "loss": 0.3615,
      "step": 2160
    },
    {
      "epoch": 24.59,
      "learning_rate": 1.0214183119853583e-05,
      "loss": 0.3585,
      "step": 2170
    },
    {
      "epoch": 24.7,
      "learning_rate": 1.0142794812897874e-05,
      "loss": 0.3519,
      "step": 2180
    },
    {
      "epoch": 24.82,
      "learning_rate": 1.0071399226384695e-05,
      "loss": 0.3734,
      "step": 2190
    },
    {
      "epoch": 24.93,
      "learning_rate": 1e-05,
      "loss": 0.3788,
      "step": 2200
    },
    {
      "epoch": 25.04,
      "learning_rate": 9.928600773615306e-06,
      "loss": 0.3616,
      "step": 2210
    },
    {
      "epoch": 25.16,
      "learning_rate": 9.85720518710213e-06,
      "loss": 0.3345,
      "step": 2220
    },
    {
      "epoch": 25.27,
      "learning_rate": 9.78581688014642e-06,
      "loss": 0.3483,
      "step": 2230
    },
    {
      "epoch": 25.38,
      "learning_rate": 9.71443949206304e-06,
      "loss": 0.327,
      "step": 2240
    },
    {
      "epoch": 25.5,
      "learning_rate": 9.643076661610197e-06,
      "loss": 0.3422,
      "step": 2250
    },
    {
      "epoch": 25.61,
      "learning_rate": 9.571732026803978e-06,
      "loss": 0.3472,
      "step": 2260
    },
    {
      "epoch": 25.72,
      "learning_rate": 9.500409224732863e-06,
      "loss": 0.3475,
      "step": 2270
    },
    {
      "epoch": 25.84,
      "learning_rate": 9.42911189137232e-06,
      "loss": 0.3507,
      "step": 2280
    },
    {
      "epoch": 25.95,
      "learning_rate": 9.357843661399447e-06,
      "loss": 0.3534,
      "step": 2290
    },
    {
      "epoch": 26.06,
      "learning_rate": 9.286608168007678e-06,
      "loss": 0.3295,
      "step": 2300
    },
    {
      "epoch": 26.18,
      "learning_rate": 9.215409042721553e-06,
      "loss": 0.3187,
      "step": 2310
    },
    {
      "epoch": 26.29,
      "learning_rate": 9.144249915211605e-06,
      "loss": 0.3465,
      "step": 2320
    },
    {
      "epoch": 26.4,
      "learning_rate": 9.073134413109313e-06,
      "loss": 0.3262,
      "step": 2330
    },
    {
      "epoch": 26.52,
      "learning_rate": 9.002066161822174e-06,
      "loss": 0.3154,
      "step": 2340
    },
    {
      "epoch": 26.63,
      "learning_rate": 8.931048784348875e-06,
      "loss": 0.3277,
      "step": 2350
    },
    {
      "epoch": 26.74,
      "learning_rate": 8.860085901094595e-06,
      "loss": 0.3387,
      "step": 2360
    },
    {
      "epoch": 26.86,
      "learning_rate": 8.789181129686452e-06,
      "loss": 0.323,
      "step": 2370
    },
    {
      "epoch": 26.97,
      "learning_rate": 8.718338084789074e-06,
      "loss": 0.3243,
      "step": 2380
    },
    {
      "epoch": 27.08,
      "learning_rate": 8.647560377920311e-06,
      "loss": 0.3112,
      "step": 2390
    },
    {
      "epoch": 27.2,
      "learning_rate": 8.576851617267151e-06,
      "loss": 0.3138,
      "step": 2400
    },
    {
      "epoch": 27.31,
      "learning_rate": 8.50621540750175e-06,
      "loss": 0.294,
      "step": 2410
    },
    {
      "epoch": 27.42,
      "learning_rate": 8.43565534959769e-06,
      "loss": 0.3009,
      "step": 2420
    },
    {
      "epoch": 27.54,
      "learning_rate": 8.365175040646403e-06,
      "loss": 0.3217,
      "step": 2430
    },
    {
      "epoch": 27.65,
      "learning_rate": 8.294778073673762e-06,
      "loss": 0.3083,
      "step": 2440
    },
    {
      "epoch": 27.76,
      "learning_rate": 8.224468037456969e-06,
      "loss": 0.3201,
      "step": 2450
    },
    {
      "epoch": 27.88,
      "learning_rate": 8.154248516341547e-06,
      "loss": 0.3402,
      "step": 2460
    },
    {
      "epoch": 27.99,
      "learning_rate": 8.084123090058646e-06,
      "loss": 0.3128,
      "step": 2470
    },
    {
      "epoch": 28.1,
      "learning_rate": 8.014095333542548e-06,
      "loss": 0.2901,
      "step": 2480
    },
    {
      "epoch": 28.22,
      "learning_rate": 7.944168816748396e-06,
      "loss": 0.2901,
      "step": 2490
    },
    {
      "epoch": 28.33,
      "learning_rate": 7.874347104470234e-06,
      "loss": 0.2886,
      "step": 2500
    },
    {
      "epoch": 28.44,
      "learning_rate": 7.804633756159258e-06,
      "loss": 0.2953,
      "step": 2510
    },
    {
      "epoch": 28.56,
      "learning_rate": 7.735032325742355e-06,
      "loss": 0.3088,
      "step": 2520
    },
    {
      "epoch": 28.67,
      "learning_rate": 7.66554636144095e-06,
      "loss": 0.3004,
      "step": 2530
    },
    {
      "epoch": 28.78,
      "learning_rate": 7.596179405590076e-06,
      "loss": 0.3299,
      "step": 2540
    },
    {
      "epoch": 28.9,
      "learning_rate": 7.5269349944578454e-06,
      "loss": 0.294,
      "step": 2550
    },
    {
      "epoch": 29.01,
      "learning_rate": 7.4578166580651335e-06,
      "loss": 0.2974,
      "step": 2560
    },
    {
      "epoch": 29.12,
      "learning_rate": 7.388827920005628e-06,
      "loss": 0.2773,
      "step": 2570
    },
    {
      "epoch": 29.24,
      "learning_rate": 7.319972297266215e-06,
      "loss": 0.2718,
      "step": 2580
    },
    {
      "epoch": 29.35,
      "learning_rate": 7.2512533000476625e-06,
      "loss": 0.2964,
      "step": 2590
    },
    {
      "epoch": 29.46,
      "learning_rate": 7.182674431585703e-06,
      "loss": 0.2784,
      "step": 2600
    },
    {
      "epoch": 29.58,
      "learning_rate": 7.114239187972416e-06,
      "loss": 0.3001,
      "step": 2610
    },
    {
      "epoch": 29.69,
      "learning_rate": 7.045951057978001e-06,
      "loss": 0.2877,
      "step": 2620
    },
    {
      "epoch": 29.8,
      "learning_rate": 6.977813522872943e-06,
      "loss": 0.2964,
      "step": 2630
    },
    {
      "epoch": 29.92,
      "learning_rate": 6.909830056250527e-06,
      "loss": 0.2923,
      "step": 2640
    },
    {
      "epoch": 30.03,
      "learning_rate": 6.8420041238497525e-06,
      "loss": 0.2839,
      "step": 2650
    },
    {
      "epoch": 30.14,
      "learning_rate": 6.774339183378663e-06,
      "loss": 0.2679,
      "step": 2660
    },
    {
      "epoch": 30.25,
      "learning_rate": 6.7068386843380695e-06,
      "loss": 0.2751,
      "step": 2670
    },
    {
      "epoch": 30.37,
      "learning_rate": 6.639506067845698e-06,
      "loss": 0.2588,
      "step": 2680
    },
    {
      "epoch": 30.48,
      "learning_rate": 6.572344766460776e-06,
      "loss": 0.2828,
      "step": 2690
    },
    {
      "epoch": 30.59,
      "learning_rate": 6.505358204009018e-06,
      "loss": 0.2904,
      "step": 2700
    },
    {
      "epoch": 30.71,
      "learning_rate": 6.438549795408107e-06,
      "loss": 0.2712,
      "step": 2710
    },
    {
      "epoch": 30.82,
      "learning_rate": 6.3719229464935915e-06,
      "loss": 0.2765,
      "step": 2720
    },
    {
      "epoch": 30.93,
      "learning_rate": 6.305481053845262e-06,
      "loss": 0.2732,
      "step": 2730
    },
    {
      "epoch": 31.05,
      "learning_rate": 6.239227504614004e-06,
      "loss": 0.2604,
      "step": 2740
    },
    {
      "epoch": 31.16,
      "learning_rate": 6.173165676349103e-06,
      "loss": 0.2535,
      "step": 2750
    },
    {
      "epoch": 31.27,
      "learning_rate": 6.107298936826086e-06,
      "loss": 0.2607,
      "step": 2760
    },
    {
      "epoch": 31.39,
      "learning_rate": 6.041630643875018e-06,
      "loss": 0.2769,
      "step": 2770
    },
    {
      "epoch": 31.5,
      "learning_rate": 5.9761641452093225e-06,
      "loss": 0.2747,
      "step": 2780
    },
    {
      "epoch": 31.61,
      "learning_rate": 5.910902778255134e-06,
      "loss": 0.2578,
      "step": 2790
    },
    {
      "epoch": 31.73,
      "learning_rate": 5.845849869981137e-06,
      "loss": 0.2566,
      "step": 2800
    },
    {
      "epoch": 31.84,
      "learning_rate": 5.781008736728975e-06,
      "loss": 0.2658,
      "step": 2810
    },
    {
      "epoch": 31.95,
      "learning_rate": 5.716382684044191e-06,
      "loss": 0.2732,
      "step": 2820
    },
    {
      "epoch": 32.07,
      "learning_rate": 5.6519750065076815e-06,
      "loss": 0.2656,
      "step": 2830
    },
    {
      "epoch": 32.18,
      "learning_rate": 5.587788987567785e-06,
      "loss": 0.2386,
      "step": 2840
    },
    {
      "epoch": 32.29,
      "learning_rate": 5.523827899372876e-06,
      "loss": 0.2554,
      "step": 2850
    },
    {
      "epoch": 32.41,
      "learning_rate": 5.460095002604533e-06,
      "loss": 0.2511,
      "step": 2860
    },
    {
      "epoch": 32.52,
      "learning_rate": 5.396593546311346e-06,
      "loss": 0.2641,
      "step": 2870
    },
    {
      "epoch": 32.63,
      "learning_rate": 5.333326767743263e-06,
      "loss": 0.2553,
      "step": 2880
    },
    {
      "epoch": 32.75,
      "learning_rate": 5.276589982435913e-06,
      "loss": 0.2508,
      "step": 2890
    },
    {
      "epoch": 32.86,
      "learning_rate": 5.213777967188747e-06,
      "loss": 0.2533,
      "step": 2900
    },
    {
      "epoch": 32.97,
      "learning_rate": 5.151209949448599e-06,
      "loss": 0.2621,
      "step": 2910
    },
    {
      "epoch": 33.09,
      "learning_rate": 5.088889118879516e-06,
      "loss": 0.2381,
      "step": 2920
    },
    {
      "epoch": 33.2,
      "learning_rate": 5.02681865254417e-06,
      "loss": 0.2405,
      "step": 2930
    },
    {
      "epoch": 33.31,
      "learning_rate": 4.965001714741851e-06,
      "loss": 0.2473,
      "step": 2940
    },
    {
      "epoch": 33.43,
      "learning_rate": 4.903441456847198e-06,
      "loss": 0.2345,
      "step": 2950
    },
    {
      "epoch": 33.54,
      "learning_rate": 4.8421410171495265e-06,
      "loss": 0.2349,
      "step": 2960
    },
    {
      "epoch": 33.65,
      "learning_rate": 4.781103520692839e-06,
      "loss": 0.2466,
      "step": 2970
    },
    {
      "epoch": 33.77,
      "learning_rate": 4.720332079116523e-06,
      "loss": 0.2512,
      "step": 2980
    },
    {
      "epoch": 33.88,
      "learning_rate": 4.659829790496699e-06,
      "loss": 0.2633,
      "step": 2990
    },
    {
      "epoch": 33.99,
      "learning_rate": 4.599599739188322e-06,
      "loss": 0.247,
      "step": 3000
    },
    {
      "epoch": 34.11,
      "learning_rate": 4.539644995667911e-06,
      "loss": 0.2231,
      "step": 3010
    },
    {
      "epoch": 34.22,
      "learning_rate": 4.479968616377024e-06,
      "loss": 0.2174,
      "step": 3020
    },
    {
      "epoch": 34.33,
      "learning_rate": 4.420573643566455e-06,
      "loss": 0.231,
      "step": 3030
    },
    {
      "epoch": 34.45,
      "learning_rate": 4.361463105141137e-06,
      "loss": 0.2545,
      "step": 3040
    },
    {
      "epoch": 34.56,
      "learning_rate": 4.302640014505779e-06,
      "loss": 0.2502,
      "step": 3050
    },
    {
      "epoch": 34.67,
      "learning_rate": 4.244107370411248e-06,
      "loss": 0.2505,
      "step": 3060
    },
    {
      "epoch": 34.79,
      "learning_rate": 4.185868156801695e-06,
      "loss": 0.225,
      "step": 3070
    },
    {
      "epoch": 34.9,
      "learning_rate": 4.1279253426624345e-06,
      "loss": 0.2436,
      "step": 3080
    },
    {
      "epoch": 35.01,
      "learning_rate": 4.07028188186859e-06,
      "loss": 0.2274,
      "step": 3090
    },
    {
      "epoch": 35.13,
      "learning_rate": 4.0129407130345114e-06,
      "loss": 0.214,
      "step": 3100
    },
    {
      "epoch": 35.24,
      "learning_rate": 3.955904759363958e-06,
      "loss": 0.2088,
      "step": 3110
    },
    {
      "epoch": 35.35,
      "learning_rate": 3.899176928501078e-06,
      "loss": 0.244,
      "step": 3120
    },
    {
      "epoch": 35.47,
      "learning_rate": 3.842760112382183e-06,
      "loss": 0.2255,
      "step": 3130
    },
    {
      "epoch": 35.58,
      "learning_rate": 3.7866571870883382e-06,
      "loss": 0.2282,
      "step": 3140
    },
    {
      "epoch": 35.69,
      "learning_rate": 3.7308710126986934e-06,
      "loss": 0.2415,
      "step": 3150
    },
    {
      "epoch": 35.81,
      "learning_rate": 3.675404433144727e-06,
      "loss": 0.2415,
      "step": 3160
    },
    {
      "epoch": 35.92,
      "learning_rate": 3.6202602760652395e-06,
      "loss": 0.2301,
      "step": 3170
    },
    {
      "epoch": 36.03,
      "learning_rate": 3.565441352662211e-06,
      "loss": 0.2301,
      "step": 3180
    },
    {
      "epoch": 36.15,
      "learning_rate": 3.510950457557487e-06,
      "loss": 0.2213,
      "step": 3190
    },
    {
      "epoch": 36.26,
      "learning_rate": 3.4567903686503103e-06,
      "loss": 0.2174,
      "step": 3200
    },
    {
      "epoch": 36.37,
      "learning_rate": 3.4029638469757055e-06,
      "loss": 0.2236,
      "step": 3210
    },
    {
      "epoch": 36.49,
      "learning_rate": 3.3494736365637304e-06,
      "loss": 0.2235,
      "step": 3220
    },
    {
      "epoch": 36.6,
      "learning_rate": 3.2963224642995675e-06,
      "loss": 0.222,
      "step": 3230
    },
    {
      "epoch": 36.71,
      "learning_rate": 3.2435130397845472e-06,
      "loss": 0.2237,
      "step": 3240
    },
    {
      "epoch": 36.83,
      "learning_rate": 3.1910480551979706e-06,
      "loss": 0.2239,
      "step": 3250
    },
    {
      "epoch": 36.94,
      "learning_rate": 3.1389301851598976e-06,
      "loss": 0.2149,
      "step": 3260
    },
    {
      "epoch": 37.05,
      "learning_rate": 3.0871620865947816e-06,
      "loss": 0.2279,
      "step": 3270
    },
    {
      "epoch": 37.17,
      "learning_rate": 3.0357463985960257e-06,
      "loss": 0.2007,
      "step": 3280
    },
    {
      "epoch": 37.28,
      "learning_rate": 2.9846857422914434e-06,
      "loss": 0.2132,
      "step": 3290
    },
    {
      "epoch": 37.39,
      "learning_rate": 2.933982720709637e-06,
      "loss": 0.2147,
      "step": 3300
    },
    {
      "epoch": 37.51,
      "learning_rate": 2.883639918647296e-06,
      "loss": 0.2121,
      "step": 3310
    },
    {
      "epoch": 37.62,
      "learning_rate": 2.833659902537429e-06,
      "loss": 0.2147,
      "step": 3320
    },
    {
      "epoch": 37.73,
      "learning_rate": 2.7840452203185154e-06,
      "loss": 0.2127,
      "step": 3330
    },
    {
      "epoch": 37.85,
      "learning_rate": 2.7347984013046435e-06,
      "loss": 0.2308,
      "step": 3340
    },
    {
      "epoch": 37.96,
      "learning_rate": 2.6859219560565407e-06,
      "loss": 0.2152,
      "step": 3350
    },
    {
      "epoch": 38.07,
      "learning_rate": 2.637418376253591e-06,
      "loss": 0.2161,
      "step": 3360
    },
    {
      "epoch": 38.19,
      "learning_rate": 2.589290134566824e-06,
      "loss": 0.2206,
      "step": 3370
    },
    {
      "epoch": 38.3,
      "learning_rate": 2.541539684532852e-06,
      "loss": 0.2256,
      "step": 3380
    },
    {
      "epoch": 38.41,
      "learning_rate": 2.4941694604287913e-06,
      "loss": 0.2096,
      "step": 3390
    },
    {
      "epoch": 38.53,
      "learning_rate": 2.447181877148165e-06,
      "loss": 0.2058,
      "step": 3400
    },
    {
      "epoch": 38.64,
      "learning_rate": 2.4005793300777933e-06,
      "loss": 0.2102,
      "step": 3410
    },
    {
      "epoch": 38.75,
      "learning_rate": 2.3543641949756835e-06,
      "loss": 0.2139,
      "step": 3420
    },
    {
      "epoch": 38.87,
      "learning_rate": 2.30853882784991e-06,
      "loss": 0.2048,
      "step": 3430
    },
    {
      "epoch": 38.98,
      "learning_rate": 2.2631055648385e-06,
      "loss": 0.2071,
      "step": 3440
    },
    {
      "epoch": 39.09,
      "learning_rate": 2.2180667220903697e-06,
      "loss": 0.206,
      "step": 3450
    },
    {
      "epoch": 39.21,
      "learning_rate": 2.1734245956472024e-06,
      "loss": 0.2049,
      "step": 3460
    },
    {
      "epoch": 39.32,
      "learning_rate": 2.1291814613264383e-06,
      "loss": 0.2096,
      "step": 3470
    },
    {
      "epoch": 39.43,
      "learning_rate": 2.085339574605243e-06,
      "loss": 0.1986,
      "step": 3480
    },
    {
      "epoch": 39.55,
      "learning_rate": 2.0419011705055115e-06,
      "loss": 0.2036,
      "step": 3490
    },
    {
      "epoch": 39.66,
      "learning_rate": 1.998868463479945e-06,
      "loss": 0.21,
      "step": 3500
    },
    {
      "epoch": 39.77,
      "learning_rate": 1.956243647299155e-06,
      "loss": 0.2155,
      "step": 3510
    },
    {
      "epoch": 39.89,
      "learning_rate": 1.914028894939829e-06,
      "loss": 0.2086,
      "step": 3520
    },
    {
      "epoch": 40.0,
      "learning_rate": 1.8722263584739486e-06,
      "loss": 0.1997,
      "step": 3530
    },
    {
      "epoch": 40.11,
      "learning_rate": 1.830838168959075e-06,
      "loss": 0.1932,
      "step": 3540
    },
    {
      "epoch": 40.23,
      "learning_rate": 1.7898664363297302e-06,
      "loss": 0.1997,
      "step": 3550
    },
    {
      "epoch": 40.34,
      "learning_rate": 1.7493132492898134e-06,
      "loss": 0.1975,
      "step": 3560
    },
    {
      "epoch": 40.45,
      "learning_rate": 1.7091806752061212e-06,
      "loss": 0.2072,
      "step": 3570
    },
    {
      "epoch": 40.57,
      "learning_rate": 1.6694707600029702e-06,
      "loss": 0.1786,
      "step": 3580
    },
    {
      "epoch": 40.68,
      "learning_rate": 1.630185528057886e-06,
      "loss": 0.2044,
      "step": 3590
    },
    {
      "epoch": 40.79,
      "learning_rate": 1.5913269820984023e-06,
      "loss": 0.2129,
      "step": 3600
    },
    {
      "epoch": 40.91,
      "learning_rate": 1.5528971030999706e-06,
      "loss": 0.2219,
      "step": 3610
    },
    {
      "epoch": 41.02,
      "learning_rate": 1.5148978501849642e-06,
      "loss": 0.2023,
      "step": 3620
    },
    {
      "epoch": 41.13,
      "learning_rate": 1.4773311605228059e-06,
      "loss": 0.1975,
      "step": 3630
    },
    {
      "epoch": 41.25,
      "learning_rate": 1.4401989492312164e-06,
      "loss": 0.1922,
      "step": 3640
    },
    {
      "epoch": 41.36,
      "learning_rate": 1.403503109278579e-06,
      "loss": 0.1868,
      "step": 3650
    },
    {
      "epoch": 41.47,
      "learning_rate": 1.3672455113874429e-06,
      "loss": 0.2009,
      "step": 3660
    },
    {
      "epoch": 41.59,
      "learning_rate": 1.3314280039391426e-06,
      "loss": 0.2064,
      "step": 3670
    },
    {
      "epoch": 41.7,
      "learning_rate": 1.2960524128795837e-06,
      "loss": 0.2036,
      "step": 3680
    },
    {
      "epoch": 41.81,
      "learning_rate": 1.2611205416261595e-06,
      "loss": 0.1896,
      "step": 3690
    },
    {
      "epoch": 41.93,
      "learning_rate": 1.2266341709757946e-06,
      "loss": 0.2078,
      "step": 3700
    },
    {
      "epoch": 42.04,
      "learning_rate": 1.192595059014179e-06,
      "loss": 0.1931,
      "step": 3710
    },
    {
      "epoch": 42.15,
      "learning_rate": 1.1590049410261384e-06,
      "loss": 0.203,
      "step": 3720
    },
    {
      "epoch": 42.27,
      "learning_rate": 1.1258655294071686e-06,
      "loss": 0.1928,
      "step": 3730
    },
    {
      "epoch": 42.38,
      "learning_rate": 1.0931785135761375e-06,
      "loss": 0.1898,
      "step": 3740
    },
    {
      "epoch": 42.49,
      "learning_rate": 1.0609455598891682e-06,
      "loss": 0.2001,
      "step": 3750
    },
    {
      "epoch": 42.61,
      "learning_rate": 1.029168311554678e-06,
      "loss": 0.2011,
      "step": 3760
    },
    {
      "epoch": 42.72,
      "learning_rate": 9.978483885496204e-07,
      "loss": 0.1871,
      "step": 3770
    },
    {
      "epoch": 42.83,
      "learning_rate": 9.669873875368852e-07,
      "loss": 0.2027,
      "step": 3780
    },
    {
      "epoch": 42.95,
      "learning_rate": 9.365868817839264e-07,
      "loss": 0.1941,
      "step": 3790
    },
    {
      "epoch": 43.06,
      "learning_rate": 9.066484210825288e-07,
      "loss": 0.1917,
      "step": 3800
    },
    {
      "epoch": 43.17,
      "learning_rate": 8.771735316698249e-07,
      "loss": 0.192,
      "step": 3810
    },
    {
      "epoch": 43.29,
      "learning_rate": 8.481637161504741e-07,
      "loss": 0.1959,
      "step": 3820
    },
    {
      "epoch": 43.4,
      "learning_rate": 8.19620453420068e-07,
      "loss": 0.192,
      "step": 3830
    },
    {
      "epoch": 43.51,
      "learning_rate": 7.915451985897382e-07,
      "loss": 0.2027,
      "step": 3840
    },
    {
      "epoch": 43.63,
      "learning_rate": 7.639393829119701e-07,
      "loss": 0.19,
      "step": 3850
    },
    {
      "epoch": 43.74,
      "learning_rate": 7.368044137076435e-07,
      "loss": 0.1919,
      "step": 3860
    },
    {
      "epoch": 43.85,
      "learning_rate": 7.101416742942913e-07,
      "loss": 0.1897,
      "step": 3870
    },
    {
      "epoch": 43.97,
      "learning_rate": 6.839525239155675e-07,
      "loss": 0.199,
      "step": 3880
    },
    {
      "epoch": 44.08,
      "learning_rate": 6.582382976719703e-07,
      "loss": 0.2,
      "step": 3890
    },
    {
      "epoch": 44.19,
      "learning_rate": 6.330003064527679e-07,
      "loss": 0.1917,
      "step": 3900
    },
    {
      "epoch": 44.31,
      "learning_rate": 6.082398368691711e-07,
      "loss": 0.1804,
      "step": 3910
    },
    {
      "epoch": 44.42,
      "learning_rate": 5.839581511887515e-07,
      "loss": 0.1971,
      "step": 3920
    },
    {
      "epoch": 44.53,
      "learning_rate": 5.601564872710851e-07,
      "loss": 0.1895,
      "step": 3930
    },
    {
      "epoch": 44.65,
      "learning_rate": 5.36836058504645e-07,
      "loss": 0.2008,
      "step": 3940
    },
    {
      "epoch": 44.76,
      "learning_rate": 5.13998053744954e-07,
      "loss": 0.1871,
      "step": 3950
    },
    {
      "epoch": 44.87,
      "learning_rate": 4.916436372539668e-07,
      "loss": 0.1899,
      "step": 3960
    },
    {
      "epoch": 44.99,
      "learning_rate": 4.697739486407227e-07,
      "loss": 0.1917,
      "step": 3970
    },
    {
      "epoch": 45.1,
      "learning_rate": 4.4839010280325003e-07,
      "loss": 0.1976,
      "step": 3980
    },
    {
      "epoch": 45.21,
      "learning_rate": 4.2749318987172385e-07,
      "loss": 0.1915,
      "step": 3990
    },
    {
      "epoch": 45.33,
      "learning_rate": 4.070842751529025e-07,
      "loss": 0.2004,
      "step": 4000
    },
    {
      "epoch": 45.44,
      "learning_rate": 3.871643990758056e-07,
      "loss": 0.1947,
      "step": 4010
    },
    {
      "epoch": 45.55,
      "learning_rate": 3.6773457713868423e-07,
      "loss": 0.1793,
      "step": 4020
    },
    {
      "epoch": 45.67,
      "learning_rate": 3.487957998572511e-07,
      "loss": 0.1882,
      "step": 4030
    },
    {
      "epoch": 45.78,
      "learning_rate": 3.3034903271417564e-07,
      "loss": 0.1862,
      "step": 4040
    },
    {
      "epoch": 45.89,
      "learning_rate": 3.1239521610987757e-07,
      "loss": 0.1862,
      "step": 4050
    },
    {
      "epoch": 46.01,
      "learning_rate": 2.949352653145754e-07,
      "loss": 0.1987,
      "step": 4060
    },
    {
      "epoch": 46.12,
      "learning_rate": 2.779700704216337e-07,
      "loss": 0.1837,
      "step": 4070
    },
    {
      "epoch": 46.23,
      "learning_rate": 2.6150049630218277e-07,
      "loss": 0.1949,
      "step": 4080
    },
    {
      "epoch": 46.35,
      "learning_rate": 2.4552738256102717e-07,
      "loss": 0.1892,
      "step": 4090
    },
    {
      "epoch": 46.46,
      "learning_rate": 2.3005154349385106e-07,
      "loss": 0.1959,
      "step": 4100
    },
    {
      "epoch": 46.57,
      "learning_rate": 2.1507376804569935e-07,
      "loss": 0.1925,
      "step": 4110
    },
    {
      "epoch": 46.69,
      "learning_rate": 2.0059481977075523e-07,
      "loss": 0.1933,
      "step": 4120
    },
    {
      "epoch": 46.8,
      "learning_rate": 1.8661543679342365e-07,
      "loss": 0.2049,
      "step": 4130
    },
    {
      "epoch": 46.91,
      "learning_rate": 1.731363317706969e-07,
      "loss": 0.1756,
      "step": 4140
    },
    {
      "epoch": 47.03,
      "learning_rate": 1.601581918558237e-07,
      "loss": 0.1746,
      "step": 4150
    },
    {
      "epoch": 47.14,
      "learning_rate": 1.4768167866328176e-07,
      "loss": 0.1963,
      "step": 4160
    },
    {
      "epoch": 47.25,
      "learning_rate": 1.357074282350457e-07,
      "loss": 0.1903,
      "step": 4170
    },
    {
      "epoch": 47.37,
      "learning_rate": 1.2423605100816304e-07,
      "loss": 0.1821,
      "step": 4180
    },
    {
      "epoch": 47.48,
      "learning_rate": 1.1326813178363927e-07,
      "loss": 0.1946,
      "step": 4190
    },
    {
      "epoch": 47.59,
      "learning_rate": 1.0280422969661696e-07,
      "loss": 0.1832,
      "step": 4200
    },
    {
      "epoch": 47.71,
      "learning_rate": 9.284487818787879e-08,
      "loss": 0.1892,
      "step": 4210
    },
    {
      "epoch": 47.82,
      "learning_rate": 8.33905849766481e-08,
      "loss": 0.1883,
      "step": 4220
    },
    {
      "epoch": 47.93,
      "learning_rate": 7.444183203471067e-08,
      "loss": 0.1921,
      "step": 4230
    },
    {
      "epoch": 48.05,
      "learning_rate": 6.599907556184115e-08,
      "loss": 0.2003,
      "step": 4240
    },
    {
      "epoch": 48.16,
      "learning_rate": 5.806274596254491e-08,
      "loss": 0.1789,
      "step": 4250
    },
    {
      "epoch": 48.27,
      "learning_rate": 5.0633247824118936e-08,
      "loss": 0.1853,
      "step": 4260
    },
    {
      "epoch": 48.39,
      "learning_rate": 4.371095989602614e-08,
      "loss": 0.1968,
      "step": 4270
    },
    {
      "epoch": 48.5,
      "learning_rate": 3.729623507058744e-08,
      "loss": 0.2019,
      "step": 4280
    },
    {
      "epoch": 48.61,
      "learning_rate": 3.13894003649895e-08,
      "loss": 0.1808,
      "step": 4290
    },
    {
      "epoch": 48.73,
      "learning_rate": 2.5990756904614723e-08,
      "loss": 0.1959,
      "step": 4300
    },
    {
      "epoch": 48.84,
      "learning_rate": 2.1100579907691322e-08,
      "loss": 0.1781,
      "step": 4310
    },
    {
      "epoch": 48.95,
      "learning_rate": 1.6719118671262302e-08,
      "loss": 0.1868,
      "step": 4320
    },
    {
      "epoch": 49.07,
      "learning_rate": 1.2846596558473424e-08,
      "loss": 0.1887,
      "step": 4330
    },
    {
      "epoch": 49.18,
      "learning_rate": 9.483210987193404e-09,
      "loss": 0.1982,
      "step": 4340
    },
    {
      "epoch": 49.29,
      "learning_rate": 6.62913341994087e-09,
      "loss": 0.1957,
      "step": 4350
    },
    {
      "epoch": 49.41,
      "learning_rate": 4.284509355151345e-09,
      "loss": 0.1925,
      "step": 4360
    },
    {
      "epoch": 49.52,
      "learning_rate": 2.449458319754294e-09,
      "loss": 0.1859,
      "step": 4370
    },
    {
      "epoch": 49.63,
      "learning_rate": 1.124073863082442e-09,
      "loss": 0.187,
      "step": 4380
    },
    {
      "epoch": 49.75,
      "learning_rate": 3.0842355210336515e-10,
      "loss": 0.1792,
      "step": 4390
    },
    {
      "epoch": 49.86,
      "learning_rate": 2.548967970028571e-12,
      "loss": 0.1895,
      "step": 4400
    }
  ],
  "max_steps": 4400,
  "num_train_epochs": 50,
  "total_flos": 6.496431653421466e+17,
  "trial_name": null,
  "trial_params": null
}