{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 19.97563946406821,
  "global_step": 8200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 4.999981652304564e-05,
      "loss": 3.5337,
      "step": 10
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.999933765030933e-05,
      "loss": 2.4344,
      "step": 20
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.999845697279959e-05,
      "loss": 1.7436,
      "step": 30
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.999720936402992e-05,
      "loss": 1.9107,
      "step": 40
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.999559484231293e-05,
      "loss": 1.5355,
      "step": 50
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.9993613431346816e-05,
      "loss": 1.3407,
      "step": 60
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.999126516021505e-05,
      "loss": 1.2717,
      "step": 70
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.998855006338591e-05,
      "loss": 1.294,
      "step": 80
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.9985468180712016e-05,
      "loss": 1.0712,
      "step": 90
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.998201955742973e-05,
      "loss": 1.1961,
      "step": 100
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.997820424415846e-05,
      "loss": 1.3744,
      "step": 110
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.99740222969e-05,
      "loss": 1.1906,
      "step": 120
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.996947377703761e-05,
      "loss": 1.026,
      "step": 130
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.9964558751335176e-05,
      "loss": 0.9262,
      "step": 140
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.995982192522523e-05,
      "loss": 1.1451,
      "step": 150
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.9954210741627514e-05,
      "loss": 0.9061,
      "step": 160
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.994823327622299e-05,
      "loss": 0.8371,
      "step": 170
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.99418896167498e-05,
      "loss": 1.1765,
      "step": 180
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.9935179856321204e-05,
      "loss": 1.0249,
      "step": 190
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.992810409342409e-05,
      "loss": 1.2131,
      "step": 200
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.992066243191761e-05,
      "loss": 0.9495,
      "step": 210
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.991285498103165e-05,
      "loss": 1.1412,
      "step": 220
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.99046818553652e-05,
      "loss": 1.0495,
      "step": 230
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.989614317488466e-05,
      "loss": 1.162,
      "step": 240
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.988723906492211e-05,
      "loss": 1.0261,
      "step": 250
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.98789130316581e-05,
      "loss": 0.9507,
      "step": 260
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.986931497017868e-05,
      "loss": 0.9162,
      "step": 270
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.985935187300584e-05,
      "loss": 0.9901,
      "step": 280
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.98490238863795e-05,
      "loss": 0.9501,
      "step": 290
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.983833116189544e-05,
      "loss": 0.9202,
      "step": 300
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.982727385650313e-05,
      "loss": 0.9175,
      "step": 310
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.9815852132503465e-05,
      "loss": 0.8588,
      "step": 320
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.980406615754627e-05,
      "loss": 1.1674,
      "step": 330
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.979191610462793e-05,
      "loss": 0.9042,
      "step": 340
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.977940215208884e-05,
      "loss": 0.8963,
      "step": 350
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.976652448361072e-05,
      "loss": 0.9845,
      "step": 360
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.9753283288214036e-05,
      "loss": 0.9651,
      "step": 370
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.973967876025509e-05,
      "loss": 1.0159,
      "step": 380
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.97257110994233e-05,
      "loss": 0.8109,
      "step": 390
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.971138051073816e-05,
      "loss": 1.0778,
      "step": 400
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.9696687204546286e-05,
      "loss": 0.9938,
      "step": 410
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.968163139651835e-05,
      "loss": 0.9635,
      "step": 420
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.966621330764582e-05,
      "loss": 0.9848,
      "step": 430
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.9650433164237844e-05,
      "loss": 0.9077,
      "step": 440
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.9634291197917816e-05,
      "loss": 0.8833,
      "step": 450
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.961778764562005e-05,
      "loss": 0.8134,
      "step": 460
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.960092274958627e-05,
      "loss": 0.7578,
      "step": 470
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.958369675736205e-05,
      "loss": 0.8654,
      "step": 480
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.95661099217932e-05,
      "loss": 0.8552,
      "step": 490
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.954816250102204e-05,
      "loss": 0.6767,
      "step": 500
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.952985475848362e-05,
      "loss": 0.8525,
      "step": 510
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.9511186962901846e-05,
      "loss": 0.7788,
      "step": 520
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.949215938828554e-05,
      "loss": 0.9851,
      "step": 530
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.9472772313924425e-05,
      "loss": 0.8086,
      "step": 540
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.9453026024385e-05,
      "loss": 0.7749,
      "step": 550
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.943292080950639e-05,
      "loss": 0.9572,
      "step": 560
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.941245696439609e-05,
      "loss": 0.7659,
      "step": 570
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.9391634789425616e-05,
      "loss": 0.7514,
      "step": 580
    },
    {
      "epoch": 1.44,
      "learning_rate": 4.93704545902261e-05,
      "loss": 0.6983,
      "step": 590
    },
    {
      "epoch": 1.46,
      "learning_rate": 4.934891667768383e-05,
      "loss": 0.8481,
      "step": 600
    },
    {
      "epoch": 1.49,
      "learning_rate": 4.9327021367935643e-05,
      "loss": 0.8893,
      "step": 610
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.930476898236432e-05,
      "loss": 0.606,
      "step": 620
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.9282159847593854e-05,
      "loss": 0.9614,
      "step": 630
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.925919429548467e-05,
      "loss": 0.8691,
      "step": 640
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.923587266312873e-05,
      "loss": 0.9466,
      "step": 650
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.921219529284459e-05,
      "loss": 0.8841,
      "step": 660
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.91881625321724e-05,
      "loss": 0.9382,
      "step": 670
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.9163774733868784e-05,
      "loss": 0.7369,
      "step": 680
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.913903225590164e-05,
      "loss": 0.8588,
      "step": 690
    },
    {
      "epoch": 1.71,
      "learning_rate": 4.9113935461444955e-05,
      "loss": 0.9441,
      "step": 700
    },
    {
      "epoch": 1.73,
      "learning_rate": 4.9088484718873385e-05,
      "loss": 0.8136,
      "step": 710
    },
    {
      "epoch": 1.75,
      "learning_rate": 4.90626804017569e-05,
      "loss": 0.838,
      "step": 720
    },
    {
      "epoch": 1.78,
      "learning_rate": 4.9036522888855325e-05,
      "loss": 0.9653,
      "step": 730
    },
    {
      "epoch": 1.8,
      "learning_rate": 4.901001256411271e-05,
      "loss": 0.797,
      "step": 740
    },
    {
      "epoch": 1.83,
      "learning_rate": 4.898314981665174e-05,
      "loss": 0.8807,
      "step": 750
    },
    {
      "epoch": 1.85,
      "learning_rate": 4.8955935040768034e-05,
      "loss": 0.9804,
      "step": 760
    },
    {
      "epoch": 1.88,
      "learning_rate": 4.892836863592432e-05,
      "loss": 0.8705,
      "step": 770
    },
    {
      "epoch": 1.9,
      "learning_rate": 4.890045100674461e-05,
      "loss": 0.8691,
      "step": 780
    },
    {
      "epoch": 1.92,
      "learning_rate": 4.887218256300821e-05,
      "loss": 0.8105,
      "step": 790
    },
    {
      "epoch": 1.95,
      "learning_rate": 4.884356371964377e-05,
      "loss": 0.8711,
      "step": 800
    },
    {
      "epoch": 1.97,
      "learning_rate": 4.8814594896723155e-05,
      "loss": 0.856,
      "step": 810
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.8785276519455264e-05,
      "loss": 0.7324,
      "step": 820
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.875560901817982e-05,
      "loss": 0.7865,
      "step": 830
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.8725592828361066e-05,
      "loss": 0.7155,
      "step": 840
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.86952283905813e-05,
      "loss": 0.6735,
      "step": 850
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.866451615053451e-05,
      "loss": 0.5544,
      "step": 860
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.863345655901975e-05,
      "loss": 0.816,
      "step": 870
    },
    {
      "epoch": 2.14,
      "learning_rate": 4.8602050071934554e-05,
      "loss": 0.5772,
      "step": 880
    },
    {
      "epoch": 2.17,
      "learning_rate": 4.857029715026826e-05,
      "loss": 0.6604,
      "step": 890
    },
    {
      "epoch": 2.19,
      "learning_rate": 4.853819826009521e-05,
      "loss": 0.602,
      "step": 900
    },
    {
      "epoch": 2.22,
      "learning_rate": 4.850575387256794e-05,
      "loss": 0.7244,
      "step": 910
    },
    {
      "epoch": 2.24,
      "learning_rate": 4.8472964463910224e-05,
      "loss": 0.6775,
      "step": 920
    },
    {
      "epoch": 2.27,
      "learning_rate": 4.8439830515410155e-05,
      "loss": 0.6261,
      "step": 930
    },
    {
      "epoch": 2.29,
      "learning_rate": 4.840635251341299e-05,
      "loss": 0.8245,
      "step": 940
    },
    {
      "epoch": 2.31,
      "learning_rate": 4.837253094931409e-05,
      "loss": 0.6961,
      "step": 950
    },
    {
      "epoch": 2.34,
      "learning_rate": 4.833836631955164e-05,
      "loss": 0.9983,
      "step": 960
    },
    {
      "epoch": 2.36,
      "learning_rate": 4.8303859125599446e-05,
      "loss": 0.6283,
      "step": 970
    },
    {
      "epoch": 2.39,
      "learning_rate": 4.8269009873959474e-05,
      "loss": 0.9679,
      "step": 980
    },
    {
      "epoch": 2.41,
      "learning_rate": 4.8233819076154494e-05,
      "loss": 0.7606,
      "step": 990
    },
    {
      "epoch": 2.44,
      "learning_rate": 4.8198287248720546e-05,
      "loss": 0.6978,
      "step": 1000
    },
    {
      "epoch": 2.46,
      "learning_rate": 4.8162414913199336e-05,
      "loss": 0.5871,
      "step": 1010
    },
    {
      "epoch": 2.48,
      "learning_rate": 4.812620259613063e-05,
      "loss": 0.7016,
      "step": 1020
    },
    {
      "epoch": 2.51,
      "learning_rate": 4.808965082904446e-05,
      "loss": 0.8239,
      "step": 1030
    },
    {
      "epoch": 2.53,
      "learning_rate": 4.805276014845339e-05,
      "loss": 0.8266,
      "step": 1040
    },
    {
      "epoch": 2.56,
      "learning_rate": 4.801553109584459e-05,
      "loss": 0.7679,
      "step": 1050
    },
    {
      "epoch": 2.58,
      "learning_rate": 4.7977964217671925e-05,
      "loss": 0.6345,
      "step": 1060
    },
    {
      "epoch": 2.61,
      "learning_rate": 4.79400600653479e-05,
      "loss": 0.77,
      "step": 1070
    },
    {
      "epoch": 2.63,
      "learning_rate": 4.790181919523559e-05,
      "loss": 0.753,
      "step": 1080
    },
    {
      "epoch": 2.66,
      "learning_rate": 4.7863242168640466e-05,
      "loss": 0.6868,
      "step": 1090
    },
    {
      "epoch": 2.68,
      "learning_rate": 4.782432955180216e-05,
      "loss": 0.7269,
      "step": 1100
    },
    {
      "epoch": 2.7,
      "learning_rate": 4.7785081915886134e-05,
      "loss": 0.7442,
      "step": 1110
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.7745499836975335e-05,
      "loss": 0.837,
      "step": 1120
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.77055838960617e-05,
      "loss": 0.563,
      "step": 1130
    },
    {
      "epoch": 2.78,
      "learning_rate": 4.7665334679037656e-05,
      "loss": 0.6673,
      "step": 1140
    },
    {
      "epoch": 2.8,
      "learning_rate": 4.76247527766875e-05,
      "loss": 0.5567,
      "step": 1150
    },
    {
      "epoch": 2.83,
      "learning_rate": 4.758383878467873e-05,
      "loss": 0.6742,
      "step": 1160
    },
    {
      "epoch": 2.85,
      "learning_rate": 4.754259330355334e-05,
      "loss": 0.6214,
      "step": 1170
    },
    {
      "epoch": 2.87,
      "learning_rate": 4.750101693871893e-05,
      "loss": 0.7883,
      "step": 1180
    },
    {
      "epoch": 2.9,
      "learning_rate": 4.7459110300439905e-05,
      "loss": 0.8314,
      "step": 1190
    },
    {
      "epoch": 2.92,
      "learning_rate": 4.7416874003828425e-05,
      "loss": 0.6504,
      "step": 1200
    },
    {
      "epoch": 2.95,
      "learning_rate": 4.7374308668835484e-05,
      "loss": 0.8773,
      "step": 1210
    },
    {
      "epoch": 2.97,
      "learning_rate": 4.7331414920241704e-05,
      "loss": 0.7314,
      "step": 1220
    },
    {
      "epoch": 3.0,
      "learning_rate": 4.728819338764824e-05,
      "loss": 0.7255,
      "step": 1230
    },
    {
      "epoch": 3.02,
      "learning_rate": 4.72446447054675e-05,
      "loss": 0.6126,
      "step": 1240
    },
    {
      "epoch": 3.05,
      "learning_rate": 4.7200769512913855e-05,
      "loss": 0.6329,
      "step": 1250
    },
    {
      "epoch": 3.07,
      "learning_rate": 4.715656845399425e-05,
      "loss": 0.6519,
      "step": 1260
    },
    {
      "epoch": 3.09,
      "learning_rate": 4.711204217749872e-05,
      "loss": 0.7236,
      "step": 1270
    },
    {
      "epoch": 3.12,
      "learning_rate": 4.706719133699093e-05,
      "loss": 0.5436,
      "step": 1280
    },
    {
      "epoch": 3.14,
      "learning_rate": 4.702201659079851e-05,
      "loss": 0.5856,
      "step": 1290
    },
    {
      "epoch": 3.17,
      "learning_rate": 4.697651860200347e-05,
      "loss": 0.6632,
      "step": 1300
    },
    {
      "epoch": 3.19,
      "learning_rate": 4.6930698038432394e-05,
      "loss": 0.4646,
      "step": 1310
    },
    {
      "epoch": 3.22,
      "learning_rate": 4.688455557264667e-05,
      "loss": 0.6182,
      "step": 1320
    },
    {
      "epoch": 3.24,
      "learning_rate": 4.683809188193263e-05,
      "loss": 0.4338,
      "step": 1330
    },
    {
      "epoch": 3.26,
      "learning_rate": 4.679130764829161e-05,
      "loss": 0.757,
      "step": 1340
    },
    {
      "epoch": 3.29,
      "learning_rate": 4.674420355842987e-05,
      "loss": 0.4597,
      "step": 1350
    },
    {
      "epoch": 3.31,
      "learning_rate": 4.6696780303748626e-05,
      "loss": 0.6495,
      "step": 1360
    },
    {
      "epoch": 3.34,
      "learning_rate": 4.6649038580333827e-05,
      "loss": 0.685,
      "step": 1370
    },
    {
      "epoch": 3.36,
      "learning_rate": 4.660097908894595e-05,
      "loss": 0.5783,
      "step": 1380
    },
    {
      "epoch": 3.39,
      "learning_rate": 4.655260253500972e-05,
      "loss": 0.5711,
      "step": 1390
    },
    {
      "epoch": 3.41,
      "learning_rate": 4.650390962860377e-05,
      "loss": 0.5597,
      "step": 1400
    },
    {
      "epoch": 3.43,
      "learning_rate": 4.645490108445018e-05,
      "loss": 0.6217,
      "step": 1410
    },
    {
      "epoch": 3.46,
      "learning_rate": 4.640557762190403e-05,
      "loss": 0.5919,
      "step": 1420
    },
    {
      "epoch": 3.48,
      "learning_rate": 4.635593996494281e-05,
      "loss": 0.6931,
      "step": 1430
    },
    {
      "epoch": 3.51,
      "learning_rate": 4.630598884215582e-05,
      "loss": 0.5832,
      "step": 1440
    },
    {
      "epoch": 3.53,
      "learning_rate": 4.625572498673344e-05,
      "loss": 0.5968,
      "step": 1450
    },
    {
      "epoch": 3.56,
      "learning_rate": 4.620514913645641e-05,
      "loss": 0.7203,
      "step": 1460
    },
    {
      "epoch": 3.58,
      "learning_rate": 4.6154262033684944e-05,
      "loss": 0.5169,
      "step": 1470
    },
    {
      "epoch": 3.61,
      "learning_rate": 4.610306442534792e-05,
      "loss": 0.6657,
      "step": 1480
    },
    {
      "epoch": 3.63,
      "learning_rate": 4.6051557062931824e-05,
      "loss": 0.5104,
      "step": 1490
    },
    {
      "epoch": 3.65,
      "learning_rate": 4.599974070246978e-05,
      "loss": 0.5557,
      "step": 1500
    },
    {
      "epoch": 3.68,
      "learning_rate": 4.594761610453043e-05,
      "loss": 0.6474,
      "step": 1510
    },
    {
      "epoch": 3.7,
      "learning_rate": 4.5895184034206765e-05,
      "loss": 0.5972,
      "step": 1520
    },
    {
      "epoch": 3.73,
      "learning_rate": 4.58424452611049e-05,
      "loss": 0.5479,
      "step": 1530
    },
    {
      "epoch": 3.75,
      "learning_rate": 4.578940055933282e-05,
      "loss": 0.4461,
      "step": 1540
    },
    {
      "epoch": 3.78,
      "learning_rate": 4.573605070748894e-05,
      "loss": 0.7081,
      "step": 1550
    },
    {
      "epoch": 3.8,
      "learning_rate": 4.5682396488650716e-05,
      "loss": 0.7244,
      "step": 1560
    },
    {
      "epoch": 3.82,
      "learning_rate": 4.562843869036317e-05,
      "loss": 0.656,
      "step": 1570
    },
    {
      "epoch": 3.85,
      "learning_rate": 4.557417810462729e-05,
      "loss": 0.5638,
      "step": 1580
    },
    {
      "epoch": 3.87,
      "learning_rate": 4.551961552788847e-05,
      "loss": 0.484,
      "step": 1590
    },
    {
      "epoch": 3.9,
      "learning_rate": 4.54647517610247e-05,
      "loss": 0.4737,
      "step": 1600
    },
    {
      "epoch": 3.92,
      "learning_rate": 4.5409587609334954e-05,
      "loss": 0.7067,
      "step": 1610
    },
    {
      "epoch": 3.95,
      "learning_rate": 4.535412388252727e-05,
      "loss": 0.5712,
      "step": 1620
    },
    {
      "epoch": 3.97,
      "learning_rate": 4.529836139470689e-05,
      "loss": 0.6356,
      "step": 1630
    },
    {
      "epoch": 4.0,
      "learning_rate": 4.524230096436435e-05,
      "loss": 0.6097,
      "step": 1640
    },
    {
      "epoch": 4.02,
      "learning_rate": 4.5185943414363396e-05,
      "loss": 0.4282,
      "step": 1650
    },
    {
      "epoch": 4.04,
      "learning_rate": 4.512928957192896e-05,
      "loss": 0.3885,
      "step": 1660
    },
    {
      "epoch": 4.07,
      "learning_rate": 4.5072340268635e-05,
      "loss": 0.2796,
      "step": 1670
    },
    {
      "epoch": 4.09,
      "learning_rate": 4.5015096340392304e-05,
      "loss": 0.5265,
      "step": 1680
    },
    {
      "epoch": 4.12,
      "learning_rate": 4.495755862743619e-05,
      "loss": 0.6607,
      "step": 1690
    },
    {
      "epoch": 4.14,
      "learning_rate": 4.48997279743142e-05,
      "loss": 0.3981,
      "step": 1700
    },
    {
      "epoch": 4.17,
      "learning_rate": 4.484160522987372e-05,
      "loss": 0.4307,
      "step": 1710
    },
    {
      "epoch": 4.19,
      "learning_rate": 4.4783191247249457e-05,
      "loss": 0.56,
      "step": 1720
    },
    {
      "epoch": 4.21,
      "learning_rate": 4.4724486883851e-05,
      "loss": 0.4518,
      "step": 1730
    },
    {
      "epoch": 4.24,
      "learning_rate": 4.466549300135016e-05,
      "loss": 0.5755,
      "step": 1740
    },
    {
      "epoch": 4.26,
      "learning_rate": 4.4606210465668384e-05,
      "loss": 0.5125,
      "step": 1750
    },
    {
      "epoch": 4.29,
      "learning_rate": 4.454664014696398e-05,
      "loss": 0.5609,
      "step": 1760
    },
    {
      "epoch": 4.31,
      "learning_rate": 4.448678291961942e-05,
      "loss": 0.4354,
      "step": 1770
    },
    {
      "epoch": 4.34,
      "learning_rate": 4.442663966222842e-05,
      "loss": 0.6086,
      "step": 1780
    },
    {
      "epoch": 4.36,
      "learning_rate": 4.4366211257583145e-05,
      "loss": 0.5637,
      "step": 1790
    },
    {
      "epoch": 4.38,
      "learning_rate": 4.4305498592661146e-05,
      "loss": 0.4855,
      "step": 1800
    },
    {
      "epoch": 4.41,
      "learning_rate": 4.424450255861242e-05,
      "loss": 0.361,
      "step": 1810
    },
    {
      "epoch": 4.43,
      "learning_rate": 4.4183224050746286e-05,
      "loss": 0.6417,
      "step": 1820
    },
    {
      "epoch": 4.46,
      "learning_rate": 4.412166396851827e-05,
      "loss": 0.473,
      "step": 1830
    },
    {
      "epoch": 4.48,
      "learning_rate": 4.4059823215516865e-05,
      "loss": 0.587,
      "step": 1840
    },
    {
      "epoch": 4.51,
      "learning_rate": 4.399770269945034e-05,
      "loss": 0.4707,
      "step": 1850
    },
    {
      "epoch": 4.53,
      "learning_rate": 4.393530333213332e-05,
      "loss": 0.4057,
      "step": 1860
    },
    {
      "epoch": 4.56,
      "learning_rate": 4.387262602947349e-05,
      "loss": 0.404,
      "step": 1870
    },
    {
      "epoch": 4.58,
      "learning_rate": 4.380967171145809e-05,
      "loss": 0.478,
      "step": 1880
    },
    {
      "epoch": 4.6,
      "learning_rate": 4.374644130214044e-05,
      "loss": 0.4948,
      "step": 1890
    },
    {
      "epoch": 4.63,
      "learning_rate": 4.3682935729626385e-05,
      "loss": 0.5436,
      "step": 1900
    },
    {
      "epoch": 4.65,
      "learning_rate": 4.361915592606063e-05,
      "loss": 0.4618,
      "step": 1910
    },
    {
      "epoch": 4.68,
      "learning_rate": 4.355510282761312e-05,
      "loss": 0.475,
      "step": 1920
    },
    {
      "epoch": 4.7,
      "learning_rate": 4.349077737446525e-05,
      "loss": 0.4473,
      "step": 1930
    },
    {
      "epoch": 4.73,
      "learning_rate": 4.342618051079606e-05,
      "loss": 0.4606,
      "step": 1940
    },
    {
      "epoch": 4.75,
      "learning_rate": 4.336131318476842e-05,
      "loss": 0.4914,
      "step": 1950
    },
    {
      "epoch": 4.77,
      "learning_rate": 4.32961763485151e-05,
      "loss": 0.5147,
      "step": 1960
    },
    {
      "epoch": 4.8,
      "learning_rate": 4.323077095812476e-05,
      "loss": 0.5582,
      "step": 1970
    },
    {
      "epoch": 4.82,
      "learning_rate": 4.316509797362793e-05,
      "loss": 0.5085,
      "step": 1980
    },
    {
      "epoch": 4.85,
      "learning_rate": 4.309915835898296e-05,
      "loss": 0.5099,
      "step": 1990
    },
    {
      "epoch": 4.87,
      "learning_rate": 4.3032953082061823e-05,
      "loss": 0.4276,
      "step": 2000
    },
    {
      "epoch": 4.9,
      "learning_rate": 4.296648311463591e-05,
      "loss": 0.4463,
      "step": 2010
    },
    {
      "epoch": 4.92,
      "learning_rate": 4.289974943236182e-05,
      "loss": 0.5247,
      "step": 2020
    },
    {
      "epoch": 4.95,
      "learning_rate": 4.2832753014766955e-05,
      "loss": 0.5736,
      "step": 2030
    },
    {
      "epoch": 4.97,
      "learning_rate": 4.276549484523521e-05,
      "loss": 0.4952,
      "step": 2040
    },
    {
      "epoch": 4.99,
      "learning_rate": 4.269797591099253e-05,
      "loss": 0.4736,
      "step": 2050
    },
    {
      "epoch": 5.02,
      "learning_rate": 4.263019720309236e-05,
      "loss": 0.4767,
      "step": 2060
    },
    {
      "epoch": 5.04,
      "learning_rate": 4.2562159716401193e-05,
      "loss": 0.4043,
      "step": 2070
    },
    {
      "epoch": 5.07,
      "learning_rate": 4.2493864449583883e-05,
      "loss": 0.4051,
      "step": 2080
    },
    {
      "epoch": 5.09,
      "learning_rate": 4.2425312405089056e-05,
      "loss": 0.343,
      "step": 2090
    },
    {
      "epoch": 5.12,
      "learning_rate": 4.2356504589134316e-05,
      "loss": 0.3777,
      "step": 2100
    },
    {
      "epoch": 5.14,
      "learning_rate": 4.228744201169156e-05,
      "loss": 0.3874,
      "step": 2110
    },
    {
      "epoch": 5.16,
      "learning_rate": 4.2218125686472075e-05,
      "loss": 0.5224,
      "step": 2120
    },
    {
      "epoch": 5.19,
      "learning_rate": 4.2148556630911736e-05,
      "loss": 0.4146,
      "step": 2130
    },
    {
      "epoch": 5.21,
      "learning_rate": 4.207873586615603e-05,
      "loss": 0.4122,
      "step": 2140
    },
    {
      "epoch": 5.24,
      "learning_rate": 4.200866441704504e-05,
      "loss": 0.3635,
      "step": 2150
    },
    {
      "epoch": 5.26,
      "learning_rate": 4.1938343312098446e-05,
      "loss": 0.2448,
      "step": 2160
    },
    {
      "epoch": 5.29,
      "learning_rate": 4.186777358350044e-05,
      "loss": 0.4659,
      "step": 2170
    },
    {
      "epoch": 5.31,
      "learning_rate": 4.179695626708452e-05,
      "loss": 0.3774,
      "step": 2180
    },
    {
      "epoch": 5.33,
      "learning_rate": 4.1725892402318334e-05,
      "loss": 0.2299,
      "step": 2190
    },
    {
      "epoch": 5.36,
      "learning_rate": 4.1654583032288394e-05,
      "loss": 0.371,
      "step": 2200
    },
    {
      "epoch": 5.38,
      "learning_rate": 4.1583029203684784e-05,
      "loss": 0.33,
      "step": 2210
    },
    {
      "epoch": 5.41,
      "learning_rate": 4.1511231966785796e-05,
      "loss": 0.4362,
      "step": 2220
    },
    {
      "epoch": 5.43,
      "learning_rate": 4.1439192375442495e-05,
      "loss": 0.2833,
      "step": 2230
    },
    {
      "epoch": 5.46,
      "learning_rate": 4.136691148706325e-05,
      "loss": 0.3126,
      "step": 2240
    },
    {
      "epoch": 5.48,
      "learning_rate": 4.129439036259826e-05,
      "loss": 0.3969,
      "step": 2250
    },
    {
      "epoch": 5.51,
      "learning_rate": 4.122163006652393e-05,
      "loss": 0.3791,
      "step": 2260
    },
    {
      "epoch": 5.53,
      "learning_rate": 4.114863166682725e-05,
      "loss": 0.4307,
      "step": 2270
    },
    {
      "epoch": 5.55,
      "learning_rate": 4.107539623499014e-05,
      "loss": 0.3985,
      "step": 2280
    },
    {
      "epoch": 5.58,
      "learning_rate": 4.100192484597374e-05,
      "loss": 0.3489,
      "step": 2290
    },
    {
      "epoch": 5.6,
      "learning_rate": 4.0928218578202574e-05,
      "loss": 0.3095,
      "step": 2300
    },
    {
      "epoch": 5.63,
      "learning_rate": 4.0854278513548764e-05,
      "loss": 0.3725,
      "step": 2310
    },
    {
      "epoch": 5.65,
      "learning_rate": 4.078010573731612e-05,
      "loss": 0.428,
      "step": 2320
    },
    {
      "epoch": 5.68,
      "learning_rate": 4.0705701338224276e-05,
      "loss": 0.4104,
      "step": 2330
    },
    {
      "epoch": 5.7,
      "learning_rate": 4.063106640839264e-05,
      "loss": 0.4075,
      "step": 2340
    },
    {
      "epoch": 5.72,
      "learning_rate": 4.055620204332435e-05,
      "loss": 0.3711,
      "step": 2350
    },
    {
      "epoch": 5.75,
      "learning_rate": 4.048110934189028e-05,
      "loss": 0.4147,
      "step": 2360
    },
    {
      "epoch": 5.77,
      "learning_rate": 4.040578940631284e-05,
      "loss": 0.4086,
      "step": 2370
    },
    {
      "epoch": 5.8,
      "learning_rate": 4.0330243342149824e-05,
      "loss": 0.5578,
      "step": 2380
    },
    {
      "epoch": 5.82,
      "learning_rate": 4.0254472258278176e-05,
      "loss": 0.3451,
      "step": 2390
    },
    {
      "epoch": 5.85,
      "learning_rate": 4.01784772668777e-05,
      "loss": 0.325,
      "step": 2400
    },
    {
      "epoch": 5.87,
      "learning_rate": 4.010225948341476e-05,
      "loss": 0.3898,
      "step": 2410
    },
    {
      "epoch": 5.9,
      "learning_rate": 4.002582002662592e-05,
      "loss": 0.3024,
      "step": 2420
    },
    {
      "epoch": 5.92,
      "learning_rate": 3.9949160018501454e-05,
      "loss": 0.4362,
      "step": 2430
    },
    {
      "epoch": 5.94,
      "learning_rate": 3.987228058426896e-05,
      "loss": 0.4197,
      "step": 2440
    },
    {
      "epoch": 5.97,
      "learning_rate": 3.979518285237679e-05,
      "loss": 0.4507,
      "step": 2450
    },
    {
      "epoch": 5.99,
      "learning_rate": 3.971786795447751e-05,
      "loss": 0.4414,
      "step": 2460
    },
    {
      "epoch": 6.02,
      "learning_rate": 3.964033702541127e-05,
      "loss": 0.3511,
      "step": 2470
    },
    {
      "epoch": 6.04,
      "learning_rate": 3.956259120318918e-05,
      "loss": 0.2937,
      "step": 2480
    },
    {
      "epoch": 6.07,
      "learning_rate": 3.948463162897656e-05,
      "loss": 0.2735,
      "step": 2490
    },
    {
      "epoch": 6.09,
      "learning_rate": 3.940645944707625e-05,
      "loss": 0.2736,
      "step": 2500
    },
    {
      "epoch": 6.11,
      "learning_rate": 3.9328075804911746e-05,
      "loss": 0.33,
      "step": 2510
    },
    {
      "epoch": 6.14,
      "learning_rate": 3.9249481853010403e-05,
      "loss": 0.2393,
      "step": 2520
    },
    {
      "epoch": 6.16,
      "learning_rate": 3.917067874498653e-05,
      "loss": 0.3025,
      "step": 2530
    },
    {
      "epoch": 6.19,
      "learning_rate": 3.909166763752448e-05,
      "loss": 0.2683,
      "step": 2540
    },
    {
      "epoch": 6.21,
      "learning_rate": 3.901244969036163e-05,
      "loss": 0.2801,
      "step": 2550
    },
    {
      "epoch": 6.24,
      "learning_rate": 3.8933026066271396e-05,
      "loss": 0.3711,
      "step": 2560
    },
    {
      "epoch": 6.26,
      "learning_rate": 3.8853397931046156e-05,
      "loss": 0.4418,
      "step": 2570
    },
    {
      "epoch": 6.29,
      "learning_rate": 3.877356645348011e-05,
      "loss": 0.2819,
      "step": 2580
    },
    {
      "epoch": 6.31,
      "learning_rate": 3.86935328053522e-05,
      "loss": 0.3661,
      "step": 2590
    },
    {
      "epoch": 6.33,
      "learning_rate": 3.861329816140881e-05,
      "loss": 0.2374,
      "step": 2600
    },
    {
      "epoch": 6.36,
      "learning_rate": 3.853286369934658e-05,
      "loss": 0.2686,
      "step": 2610
    },
    {
      "epoch": 6.38,
      "learning_rate": 3.845223059979512e-05,
      "loss": 0.3097,
      "step": 2620
    },
    {
      "epoch": 6.41,
      "learning_rate": 3.83714000462997e-05,
      "loss": 0.2125,
      "step": 2630
    },
    {
      "epoch": 6.43,
      "learning_rate": 3.8290373225303786e-05,
      "loss": 0.2657,
      "step": 2640
    },
    {
      "epoch": 6.46,
      "learning_rate": 3.820915132613174e-05,
      "loss": 0.284,
      "step": 2650
    },
    {
      "epoch": 6.48,
      "learning_rate": 3.81277355409713e-05,
      "loss": 0.3823,
      "step": 2660
    },
    {
      "epoch": 6.5,
      "learning_rate": 3.804612706485609e-05,
      "loss": 0.2053,
      "step": 2670
    },
    {
      "epoch": 6.53,
      "learning_rate": 3.796432709564807e-05,
      "loss": 0.3304,
      "step": 2680
    },
    {
      "epoch": 6.55,
      "learning_rate": 3.7882336834019994e-05,
      "loss": 0.2427,
      "step": 2690
    },
    {
      "epoch": 6.58,
      "learning_rate": 3.780015748343773e-05,
      "loss": 0.3177,
      "step": 2700
    },
    {
      "epoch": 6.6,
      "learning_rate": 3.771779025014264e-05,
      "loss": 0.2807,
      "step": 2710
    },
    {
      "epoch": 6.63,
      "learning_rate": 3.7635236343133854e-05,
      "loss": 0.3766,
      "step": 2720
    },
    {
      "epoch": 6.65,
      "learning_rate": 3.755249697415051e-05,
      "loss": 0.356,
      "step": 2730
    },
    {
      "epoch": 6.67,
      "learning_rate": 3.746957335765402e-05,
      "loss": 0.2231,
      "step": 2740
    },
    {
      "epoch": 6.7,
      "learning_rate": 3.7386466710810194e-05,
      "loss": 0.3641,
      "step": 2750
    },
    {
      "epoch": 6.72,
      "learning_rate": 3.730317825347137e-05,
      "loss": 0.2509,
      "step": 2760
    },
    {
      "epoch": 6.75,
      "learning_rate": 3.721970920815856e-05,
      "loss": 0.3387,
      "step": 2770
    },
    {
      "epoch": 6.77,
      "learning_rate": 3.713606080004346e-05,
      "loss": 0.3275,
      "step": 2780
    },
    {
      "epoch": 6.8,
      "learning_rate": 3.705223425693049e-05,
      "loss": 0.2544,
      "step": 2790
    },
    {
      "epoch": 6.82,
      "learning_rate": 3.696823080923873e-05,
      "loss": 0.2391,
      "step": 2800
    },
    {
      "epoch": 6.85,
      "learning_rate": 3.688405168998394e-05,
      "loss": 0.3147,
      "step": 2810
    },
    {
      "epoch": 6.87,
      "learning_rate": 3.6799698134760395e-05,
      "loss": 0.2503,
      "step": 2820
    },
    {
      "epoch": 6.89,
      "learning_rate": 3.671517138172277e-05,
      "loss": 0.3549,
      "step": 2830
    },
    {
      "epoch": 6.92,
      "learning_rate": 3.663047267156794e-05,
      "loss": 0.3271,
      "step": 2840
    },
    {
      "epoch": 6.94,
      "learning_rate": 3.654560324751684e-05,
      "loss": 0.2629,
      "step": 2850
    },
    {
      "epoch": 6.97,
      "learning_rate": 3.6460564355296124e-05,
      "loss": 0.2522,
      "step": 2860
    },
    {
      "epoch": 6.99,
      "learning_rate": 3.6375357243119963e-05,
      "loss": 0.3017,
      "step": 2870
    },
    {
      "epoch": 7.02,
      "learning_rate": 3.628998316167167e-05,
      "loss": 0.2657,
      "step": 2880
    },
    {
      "epoch": 7.04,
      "learning_rate": 3.620444336408535e-05,
      "loss": 0.1793,
      "step": 2890
    },
    {
      "epoch": 7.06,
      "learning_rate": 3.611873910592753e-05,
      "loss": 0.1965,
      "step": 2900
    },
    {
      "epoch": 7.09,
      "learning_rate": 3.603287164517872e-05,
      "loss": 0.1778,
      "step": 2910
    },
    {
      "epoch": 7.11,
      "learning_rate": 3.5946842242214934e-05,
      "loss": 0.1364,
      "step": 2920
    },
    {
      "epoch": 7.14,
      "learning_rate": 3.586065215978919e-05,
      "loss": 0.1961,
      "step": 2930
    },
    {
      "epoch": 7.16,
      "learning_rate": 3.577430266301299e-05,
      "loss": 0.1968,
      "step": 2940
    },
    {
      "epoch": 7.19,
      "learning_rate": 3.568779501933777e-05,
      "loss": 0.1688,
      "step": 2950
    },
    {
      "epoch": 7.21,
      "learning_rate": 3.5601130498536226e-05,
      "loss": 0.1694,
      "step": 2960
    },
    {
      "epoch": 7.24,
      "learning_rate": 3.551431037268374e-05,
      "loss": 0.2817,
      "step": 2970
    },
    {
      "epoch": 7.26,
      "learning_rate": 3.5427335916139724e-05,
      "loss": 0.2182,
      "step": 2980
    },
    {
      "epoch": 7.28,
      "learning_rate": 3.534020840552883e-05,
      "loss": 0.1984,
      "step": 2990
    },
    {
      "epoch": 7.31,
      "learning_rate": 3.5252929119722275e-05,
      "loss": 0.1884,
      "step": 3000
    },
    {
      "epoch": 7.33,
      "learning_rate": 3.516549933981907e-05,
      "loss": 0.1926,
      "step": 3010
    },
    {
      "epoch": 7.36,
      "learning_rate": 3.50779203491272e-05,
      "loss": 0.2422,
      "step": 3020
    },
    {
      "epoch": 7.38,
      "learning_rate": 3.499019343314478e-05,
      "loss": 0.2743,
      "step": 3030
    },
    {
      "epoch": 7.41,
      "learning_rate": 3.490231987954119e-05,
      "loss": 0.1692,
      "step": 3040
    },
    {
      "epoch": 7.43,
      "learning_rate": 3.4814300978138194e-05,
      "loss": 0.1952,
      "step": 3050
    },
    {
      "epoch": 7.45,
      "learning_rate": 3.4726138020890994e-05,
      "loss": 0.2851,
      "step": 3060
    },
    {
      "epoch": 7.48,
      "learning_rate": 3.463783230186925e-05,
      "loss": 0.2437,
      "step": 3070
    },
    {
      "epoch": 7.5,
      "learning_rate": 3.454938511723812e-05,
      "loss": 0.2101,
      "step": 3080
    },
    {
      "epoch": 7.53,
      "learning_rate": 3.44607977652392e-05,
      "loss": 0.212,
      "step": 3090
    },
    {
      "epoch": 7.55,
      "learning_rate": 3.4372071546171516e-05,
      "loss": 0.2584,
      "step": 3100
    },
    {
      "epoch": 7.58,
      "learning_rate": 3.428320776237235e-05,
      "loss": 0.2939,
      "step": 3110
    },
    {
      "epoch": 7.6,
      "learning_rate": 3.419420771819824e-05,
      "loss": 0.2372,
      "step": 3120
    },
    {
      "epoch": 7.62,
      "learning_rate": 3.4105072720005744e-05,
      "loss": 0.2033,
      "step": 3130
    },
    {
      "epoch": 7.65,
      "learning_rate": 3.401580407613229e-05,
      "loss": 0.2327,
      "step": 3140
    },
    {
      "epoch": 7.67,
      "learning_rate": 3.392640309687701e-05,
      "loss": 0.2603,
      "step": 3150
    },
    {
      "epoch": 7.7,
      "learning_rate": 3.383687109448143e-05,
      "loss": 0.2729,
      "step": 3160
    },
    {
      "epoch": 7.72,
      "learning_rate": 3.37472093831103e-05,
      "loss": 0.2348,
      "step": 3170
    },
    {
      "epoch": 7.75,
      "learning_rate": 3.365741927883223e-05,
      "loss": 0.1297,
      "step": 3180
    },
    {
      "epoch": 7.77,
      "learning_rate": 3.3567502099600414e-05,
      "loss": 0.2293,
      "step": 3190
    },
    {
      "epoch": 7.8,
      "learning_rate": 3.347745916523326e-05,
      "loss": 0.212,
      "step": 3200
    },
    {
      "epoch": 7.82,
      "learning_rate": 3.3387291797395034e-05,
      "loss": 0.2167,
      "step": 3210
    },
    {
      "epoch": 7.84,
      "learning_rate": 3.329700131957648e-05,
      "loss": 0.267,
      "step": 3220
    },
    {
      "epoch": 7.87,
      "learning_rate": 3.3206589057075314e-05,
      "loss": 0.212,
      "step": 3230
    },
    {
      "epoch": 7.89,
      "learning_rate": 3.3116056336976885e-05,
      "loss": 0.1898,
      "step": 3240
    },
    {
      "epoch": 7.92,
      "learning_rate": 3.302540448813462e-05,
      "loss": 0.1769,
      "step": 3250
    },
    {
      "epoch": 7.94,
      "learning_rate": 3.293463484115051e-05,
      "loss": 0.2688,
      "step": 3260
    },
    {
      "epoch": 7.97,
      "learning_rate": 3.284374872835563e-05,
      "loss": 0.2554,
      "step": 3270
    },
    {
      "epoch": 7.99,
      "learning_rate": 3.2752747483790556e-05,
      "loss": 0.1952,
      "step": 3280
    },
    {
      "epoch": 8.01,
      "learning_rate": 3.26616324431858e-05,
      "loss": 0.1999,
      "step": 3290
    },
    {
      "epoch": 8.04,
      "learning_rate": 3.2570404943942136e-05,
      "loss": 0.1176,
      "step": 3300
    },
    {
      "epoch": 8.06,
      "learning_rate": 3.247906632511109e-05,
      "loss": 0.1442,
      "step": 3310
    },
    {
      "epoch": 8.09,
      "learning_rate": 3.238761792737518e-05,
      "loss": 0.187,
      "step": 3320
    },
    {
      "epoch": 8.11,
      "learning_rate": 3.22960610930283e-05,
      "loss": 0.1203,
      "step": 3330
    },
    {
      "epoch": 8.14,
      "learning_rate": 3.220439716595596e-05,
      "loss": 0.1056,
      "step": 3340
    },
    {
      "epoch": 8.16,
      "learning_rate": 3.211262749161562e-05,
      "loss": 0.1151,
      "step": 3350
    },
    {
      "epoch": 8.19,
      "learning_rate": 3.2020753417016914e-05,
      "loss": 0.1765,
      "step": 3360
    },
    {
      "epoch": 8.21,
      "learning_rate": 3.192877629070187e-05,
      "loss": 0.1612,
      "step": 3370
    },
    {
      "epoch": 8.23,
      "learning_rate": 3.183669746272511e-05,
      "loss": 0.1169,
      "step": 3380
    },
    {
      "epoch": 8.26,
      "learning_rate": 3.17445182846341e-05,
      "loss": 0.1377,
      "step": 3390
    },
    {
      "epoch": 8.28,
      "learning_rate": 3.1652240109449204e-05,
      "loss": 0.2031,
      "step": 3400
    },
    {
      "epoch": 8.31,
      "learning_rate": 3.155986429164392e-05,
      "loss": 0.1687,
      "step": 3410
    },
    {
      "epoch": 8.33,
      "learning_rate": 3.1467392187124934e-05,
      "loss": 0.1099,
      "step": 3420
    },
    {
      "epoch": 8.36,
      "learning_rate": 3.137482515321226e-05,
      "loss": 0.1242,
      "step": 3430
    },
    {
      "epoch": 8.38,
      "learning_rate": 3.1282164548619294e-05,
      "loss": 0.1616,
      "step": 3440
    },
    {
      "epoch": 8.4,
      "learning_rate": 3.118941173343287e-05,
      "loss": 0.1509,
      "step": 3450
    },
    {
      "epoch": 8.43,
      "learning_rate": 3.109656806909333e-05,
      "loss": 0.1044,
      "step": 3460
    },
    {
      "epoch": 8.45,
      "learning_rate": 3.100363491837446e-05,
      "loss": 0.1792,
      "step": 3470
    },
    {
      "epoch": 8.48,
      "learning_rate": 3.091061364536361e-05,
      "loss": 0.1662,
      "step": 3480
    },
    {
      "epoch": 8.5,
      "learning_rate": 3.0817505615441564e-05,
      "loss": 0.1553,
      "step": 3490
    },
    {
      "epoch": 8.53,
      "learning_rate": 3.072431219526252e-05,
      "loss": 0.1232,
      "step": 3500
    },
    {
      "epoch": 8.55,
      "learning_rate": 3.06310347527341e-05,
      "loss": 0.1721,
      "step": 3510
    },
    {
      "epoch": 8.57,
      "learning_rate": 3.053767465699717e-05,
      "loss": 0.1511,
      "step": 3520
    },
    {
      "epoch": 8.6,
      "learning_rate": 3.044423327840582e-05,
      "loss": 0.1952,
      "step": 3530
    },
    {
      "epoch": 8.62,
      "learning_rate": 3.0350711988507206e-05,
      "loss": 0.1502,
      "step": 3540
    },
    {
      "epoch": 8.65,
      "learning_rate": 3.025711216002145e-05,
      "loss": 0.1683,
      "step": 3550
    },
    {
      "epoch": 8.67,
      "learning_rate": 3.0163435166821474e-05,
      "loss": 0.1556,
      "step": 3560
    },
    {
      "epoch": 8.7,
      "learning_rate": 3.0069682383912813e-05,
      "loss": 0.1446,
      "step": 3570
    },
    {
      "epoch": 8.72,
      "learning_rate": 2.9975855187413492e-05,
      "loss": 0.132,
      "step": 3580
    },
    {
      "epoch": 8.75,
      "learning_rate": 2.9881954954533758e-05,
      "loss": 0.1375,
      "step": 3590
    },
    {
      "epoch": 8.77,
      "learning_rate": 2.9787983063555918e-05,
      "loss": 0.1546,
      "step": 3600
    },
    {
      "epoch": 8.79,
      "learning_rate": 2.9693940893814072e-05,
      "loss": 0.2073,
      "step": 3610
    },
    {
      "epoch": 8.82,
      "learning_rate": 2.960924399355443e-05,
      "loss": 0.1486,
      "step": 3620
    },
    {
      "epoch": 8.84,
      "learning_rate": 2.9515072097904667e-05,
      "loss": 0.14,
      "step": 3630
    },
    {
      "epoch": 8.87,
      "learning_rate": 2.9420833929320722e-05,
      "loss": 0.214,
      "step": 3640
    },
    {
      "epoch": 8.89,
      "learning_rate": 2.9326530871045176e-05,
      "loss": 0.1181,
      "step": 3650
    },
    {
      "epoch": 8.92,
      "learning_rate": 2.923216430727306e-05,
      "loss": 0.114,
      "step": 3660
    },
    {
      "epoch": 8.94,
      "learning_rate": 2.9137735623131544e-05,
      "loss": 0.1473,
      "step": 3670
    },
    {
      "epoch": 8.96,
      "learning_rate": 2.9043246204659626e-05,
      "loss": 0.1988,
      "step": 3680
    },
    {
      "epoch": 8.99,
      "learning_rate": 2.8948697438787754e-05,
      "loss": 0.1839,
      "step": 3690
    },
    {
      "epoch": 9.01,
      "learning_rate": 2.8854090713317512e-05,
      "loss": 0.1057,
      "step": 3700
    },
    {
      "epoch": 9.04,
      "learning_rate": 2.8759427416901192e-05,
      "loss": 0.0707,
      "step": 3710
    },
    {
      "epoch": 9.06,
      "learning_rate": 2.8664708939021466e-05,
      "loss": 0.0814,
      "step": 3720
    },
    {
      "epoch": 9.09,
      "learning_rate": 2.8569936669970966e-05,
      "loss": 0.0954,
      "step": 3730
    },
    {
      "epoch": 9.11,
      "learning_rate": 2.8475112000831865e-05,
      "loss": 0.0707,
      "step": 3740
    },
    {
      "epoch": 9.14,
      "learning_rate": 2.838023632345549e-05,
      "loss": 0.1007,
      "step": 3750
    },
    {
      "epoch": 9.16,
      "learning_rate": 2.8285311030441858e-05,
      "loss": 0.0725,
      "step": 3760
    },
    {
      "epoch": 9.18,
      "learning_rate": 2.819033751511927e-05,
      "loss": 0.0866,
      "step": 3770
    },
    {
      "epoch": 9.21,
      "learning_rate": 2.809531717152383e-05,
      "loss": 0.1054,
      "step": 3780
    },
    {
      "epoch": 9.23,
      "learning_rate": 2.8000251394379007e-05,
      "loss": 0.0937,
      "step": 3790
    },
    {
      "epoch": 9.26,
      "learning_rate": 2.790514157907512e-05,
      "loss": 0.1217,
      "step": 3800
    },
    {
      "epoch": 9.28,
      "learning_rate": 2.7809989121648927e-05,
      "loss": 0.173,
      "step": 3810
    },
    {
      "epoch": 9.31,
      "learning_rate": 2.7714795418763068e-05,
      "loss": 0.1031,
      "step": 3820
    },
    {
      "epoch": 9.33,
      "learning_rate": 2.7619561867685595e-05,
      "loss": 0.1018,
      "step": 3830
    },
    {
      "epoch": 9.35,
      "learning_rate": 2.7524289866269465e-05,
      "loss": 0.1389,
      "step": 3840
    },
    {
      "epoch": 9.38,
      "learning_rate": 2.7428980812932014e-05,
      "loss": 0.1034,
      "step": 3850
    },
    {
      "epoch": 9.4,
      "learning_rate": 2.7333636106634415e-05,
      "loss": 0.0928,
      "step": 3860
    },
    {
      "epoch": 9.43,
      "learning_rate": 2.723825714686119e-05,
      "loss": 0.0861,
      "step": 3870
    },
    {
      "epoch": 9.45,
      "learning_rate": 2.714284533359961e-05,
      "loss": 0.1036,
      "step": 3880
    },
    {
      "epoch": 9.48,
      "learning_rate": 2.7047402067319194e-05,
      "loss": 0.113,
      "step": 3890
    },
    {
      "epoch": 9.5,
      "learning_rate": 2.695192874895112e-05,
      "loss": 0.1158,
      "step": 3900
    },
    {
      "epoch": 9.52,
      "learning_rate": 2.68564267798677e-05,
      "loss": 0.1194,
      "step": 3910
    },
    {
      "epoch": 9.55,
      "learning_rate": 2.6760897561861743e-05,
      "loss": 0.1169,
      "step": 3920
    },
    {
      "epoch": 9.57,
      "learning_rate": 2.666534249712606e-05,
      "loss": 0.0836,
      "step": 3930
    },
    {
      "epoch": 9.6,
      "learning_rate": 2.656976298823284e-05,
      "loss": 0.0845,
      "step": 3940
    },
    {
      "epoch": 9.62,
      "learning_rate": 2.647416043811304e-05,
      "loss": 0.0784,
      "step": 3950
    },
    {
      "epoch": 9.65,
      "learning_rate": 2.6378536250035852e-05,
      "loss": 0.1129,
      "step": 3960
    },
    {
      "epoch": 9.67,
      "learning_rate": 2.6282891827588046e-05,
      "loss": 0.1254,
      "step": 3970
    },
    {
      "epoch": 9.7,
      "learning_rate": 2.618722857465343e-05,
      "loss": 0.0714,
      "step": 3980
    },
    {
      "epoch": 9.72,
      "learning_rate": 2.609154789539216e-05,
      "loss": 0.0942,
      "step": 3990
    },
    {
      "epoch": 9.74,
      "learning_rate": 2.5995851194220222e-05,
      "loss": 0.0789,
      "step": 4000
    },
    {
      "epoch": 9.77,
      "learning_rate": 2.590013987578876e-05,
      "loss": 0.1355,
      "step": 4010
    },
    {
      "epoch": 9.79,
      "learning_rate": 2.5804415344963462e-05,
      "loss": 0.1063,
      "step": 4020
    },
    {
      "epoch": 9.82,
      "learning_rate": 2.5708679006803964e-05,
      "loss": 0.1322,
      "step": 4030
    },
    {
      "epoch": 9.84,
      "learning_rate": 2.56129322665432e-05,
      "loss": 0.1199,
      "step": 4040
    },
    {
      "epoch": 9.87,
      "learning_rate": 2.5517176529566806e-05,
      "loss": 0.0755,
      "step": 4050
    },
    {
      "epoch": 9.89,
      "learning_rate": 2.542141320139244e-05,
      "loss": 0.1086,
      "step": 4060
    },
    {
      "epoch": 9.91,
      "learning_rate": 2.532564368764922e-05,
      "loss": 0.1059,
      "step": 4070
    },
    {
      "epoch": 9.94,
      "learning_rate": 2.5229869394057036e-05,
      "loss": 0.1158,
      "step": 4080
    },
    {
      "epoch": 9.96,
      "learning_rate": 2.513409172640595e-05,
      "loss": 0.1189,
      "step": 4090
    },
    {
      "epoch": 9.99,
      "learning_rate": 2.5038312090535543e-05,
      "loss": 0.1343,
      "step": 4100
    },
    {
      "epoch": 10.01,
      "learning_rate": 2.4942531892314273e-05,
      "loss": 0.0777,
      "step": 4110
    },
    {
      "epoch": 10.04,
      "learning_rate": 2.4846752537618875e-05,
      "loss": 0.1004,
      "step": 4120
    },
    {
      "epoch": 10.06,
      "learning_rate": 2.4750975432313696e-05,
      "loss": 0.0669,
      "step": 4130
    },
    {
      "epoch": 10.09,
      "learning_rate": 2.4655201982230044e-05,
      "loss": 0.0538,
      "step": 4140
    },
    {
      "epoch": 10.11,
      "learning_rate": 2.4559433593145617e-05,
      "loss": 0.0533,
      "step": 4150
    },
    {
      "epoch": 10.13,
      "learning_rate": 2.4463671670763787e-05,
      "loss": 0.0622,
      "step": 4160
    },
    {
      "epoch": 10.16,
      "learning_rate": 2.436791762069303e-05,
      "loss": 0.071,
      "step": 4170
    },
    {
      "epoch": 10.18,
      "learning_rate": 2.4272172848426257e-05,
      "loss": 0.0628,
      "step": 4180
    },
    {
      "epoch": 10.21,
      "learning_rate": 2.417643875932021e-05,
      "loss": 0.0505,
      "step": 4190
    },
    {
      "epoch": 10.23,
      "learning_rate": 2.408071675857482e-05,
      "loss": 0.0769,
      "step": 4200
    },
    {
      "epoch": 10.26,
      "learning_rate": 2.3985008251212574e-05,
      "loss": 0.0689,
      "step": 4210
    },
    {
      "epoch": 10.28,
      "learning_rate": 2.3889314642057913e-05,
      "loss": 0.0869,
      "step": 4220
    },
    {
      "epoch": 10.3,
      "learning_rate": 2.3793637335716586e-05,
      "loss": 0.0823,
      "step": 4230
    },
    {
      "epoch": 10.33,
      "learning_rate": 2.369797773655506e-05,
      "loss": 0.0665,
      "step": 4240
    },
    {
      "epoch": 10.35,
      "learning_rate": 2.3602337248679885e-05,
      "loss": 0.0624,
      "step": 4250
    },
    {
      "epoch": 10.38,
      "learning_rate": 2.3506717275917093e-05,
      "loss": 0.0801,
      "step": 4260
    },
    {
      "epoch": 10.4,
      "learning_rate": 2.3411119221791605e-05,
      "loss": 0.0624,
      "step": 4270
    },
    {
      "epoch": 10.43,
      "learning_rate": 2.3315544489506592e-05,
      "loss": 0.0494,
      "step": 4280
    },
    {
      "epoch": 10.45,
      "learning_rate": 2.3219994481922934e-05,
      "loss": 0.0701,
      "step": 4290
    },
    {
      "epoch": 10.48,
      "learning_rate": 2.312447060153856e-05,
      "loss": 0.0666,
      "step": 4300
    },
    {
      "epoch": 10.5,
      "learning_rate": 2.3028974250467946e-05,
      "loss": 0.0934,
      "step": 4310
    },
    {
      "epoch": 10.52,
      "learning_rate": 2.2933506830421435e-05,
      "loss": 0.0868,
      "step": 4320
    },
    {
      "epoch": 10.55,
      "learning_rate": 2.2838069742684766e-05,
      "loss": 0.0709,
      "step": 4330
    },
    {
      "epoch": 10.57,
      "learning_rate": 2.2742664388098438e-05,
      "loss": 0.0514,
      "step": 4340
    },
    {
      "epoch": 10.6,
      "learning_rate": 2.2647292167037144e-05,
      "loss": 0.0739,
      "step": 4350
    },
    {
      "epoch": 10.62,
      "learning_rate": 2.255195447938927e-05,
      "loss": 0.052,
      "step": 4360
    },
    {
      "epoch": 10.65,
      "learning_rate": 2.2456652724536286e-05,
      "loss": 0.0828,
      "step": 4370
    },
    {
      "epoch": 10.67,
      "learning_rate": 2.2361388301332265e-05,
      "loss": 0.0812,
      "step": 4380
    },
    {
      "epoch": 10.69,
      "learning_rate": 2.2266162608083306e-05,
      "loss": 0.0832,
      "step": 4390
    },
    {
      "epoch": 10.72,
      "learning_rate": 2.217097704252701e-05,
      "loss": 0.0488,
      "step": 4400
    },
    {
      "epoch": 10.74,
      "learning_rate": 2.2075833001812003e-05,
      "loss": 0.0744,
      "step": 4410
    },
    {
      "epoch": 10.77,
      "learning_rate": 2.198073188247738e-05,
      "loss": 0.0664,
      "step": 4420
    },
    {
      "epoch": 10.79,
      "learning_rate": 2.1885675080432247e-05,
      "loss": 0.0743,
      "step": 4430
    },
    {
      "epoch": 10.82,
      "learning_rate": 2.1790663990935202e-05,
      "loss": 0.0809,
      "step": 4440
    },
    {
      "epoch": 10.84,
      "learning_rate": 2.169570000857387e-05,
      "loss": 0.0585,
      "step": 4450
    },
    {
      "epoch": 10.86,
      "learning_rate": 2.1600784527244445e-05,
      "loss": 0.0781,
      "step": 4460
    },
    {
      "epoch": 10.89,
      "learning_rate": 2.150591894013118e-05,
      "loss": 0.0991,
      "step": 4470
    },
    {
      "epoch": 10.91,
      "learning_rate": 2.1411104639686014e-05,
      "loss": 0.0574,
      "step": 4480
    },
    {
      "epoch": 10.94,
      "learning_rate": 2.1316343017608062e-05,
      "loss": 0.05,
      "step": 4490
    },
    {
      "epoch": 10.96,
      "learning_rate": 2.1221635464823237e-05,
      "loss": 0.0812,
      "step": 4500
    },
    {
      "epoch": 10.99,
      "learning_rate": 2.1126983371463795e-05,
      "loss": 0.0496,
      "step": 4510
    },
    {
      "epoch": 11.01,
      "learning_rate": 2.1032388126847967e-05,
      "loss": 0.0417,
      "step": 4520
    },
    {
      "epoch": 11.04,
      "learning_rate": 2.0937851119459548e-05,
      "loss": 0.0426,
      "step": 4530
    },
    {
      "epoch": 11.06,
      "learning_rate": 2.0843373736927504e-05,
      "loss": 0.0482,
      "step": 4540
    },
    {
      "epoch": 11.08,
      "learning_rate": 2.074895736600564e-05,
      "loss": 0.0399,
      "step": 4550
    },
    {
      "epoch": 11.11,
      "learning_rate": 2.0654603392552193e-05,
      "loss": 0.0334,
      "step": 4560
    },
    {
      "epoch": 11.13,
      "learning_rate": 2.0560313201509554e-05,
      "loss": 0.0506,
      "step": 4570
    },
    {
      "epoch": 11.16,
      "learning_rate": 2.0466088176883875e-05,
      "loss": 0.042,
      "step": 4580
    },
    {
      "epoch": 11.18,
      "learning_rate": 2.037192970172481e-05,
      "loss": 0.0464,
      "step": 4590
    },
    {
      "epoch": 11.21,
      "learning_rate": 2.027783915810518e-05,
      "loss": 0.0618,
      "step": 4600
    },
    {
      "epoch": 11.23,
      "learning_rate": 2.0183817927100683e-05,
      "loss": 0.0543,
      "step": 4610
    },
    {
      "epoch": 11.25,
      "learning_rate": 2.0089867388769666e-05,
      "loss": 0.046,
      "step": 4620
    },
    {
      "epoch": 11.28,
      "learning_rate": 1.9995988922132803e-05,
      "loss": 0.0532,
      "step": 4630
    },
    {
      "epoch": 11.3,
      "learning_rate": 1.9902183905152907e-05,
      "loss": 0.0452,
      "step": 4640
    },
    {
      "epoch": 11.33,
      "learning_rate": 1.980845371471469e-05,
      "loss": 0.0493,
      "step": 4650
    },
    {
      "epoch": 11.35,
      "learning_rate": 1.9714799726604537e-05,
      "loss": 0.0422,
      "step": 4660
    },
    {
      "epoch": 11.38,
      "learning_rate": 1.962122331549033e-05,
      "loss": 0.0492,
      "step": 4670
    },
    {
      "epoch": 11.4,
      "learning_rate": 1.9527725854901268e-05,
      "loss": 0.0578,
      "step": 4680
    },
    {
      "epoch": 11.43,
      "learning_rate": 1.9434308717207682e-05,
      "loss": 0.0574,
      "step": 4690
    },
    {
      "epoch": 11.45,
      "learning_rate": 1.9340973273600942e-05,
      "loss": 0.0348,
      "step": 4700
    },
    {
      "epoch": 11.47,
      "learning_rate": 1.9247720894073264e-05,
      "loss": 0.0606,
      "step": 4710
    },
    {
      "epoch": 11.5,
      "learning_rate": 1.9154552947397668e-05,
      "loss": 0.044,
      "step": 4720
    },
    {
      "epoch": 11.52,
      "learning_rate": 1.906147080110784e-05,
      "loss": 0.0497,
      "step": 4730
    },
    {
      "epoch": 11.55,
      "learning_rate": 1.8968475821478066e-05,
      "loss": 0.0604,
      "step": 4740
    },
    {
      "epoch": 11.57,
      "learning_rate": 1.8875569373503215e-05,
      "loss": 0.0497,
      "step": 4750
    },
    {
      "epoch": 11.6,
      "learning_rate": 1.8782752820878634e-05,
      "loss": 0.0315,
      "step": 4760
    },
    {
      "epoch": 11.62,
      "learning_rate": 1.8690027525980212e-05,
      "loss": 0.0475,
      "step": 4770
    },
    {
      "epoch": 11.64,
      "learning_rate": 1.8597394849844317e-05,
      "loss": 0.0521,
      "step": 4780
    },
    {
      "epoch": 11.67,
      "learning_rate": 1.8504856152147855e-05,
      "loss": 0.0616,
      "step": 4790
    },
    {
      "epoch": 11.69,
      "learning_rate": 1.8412412791188305e-05,
      "loss": 0.0409,
      "step": 4800
    },
    {
      "epoch": 11.72,
      "learning_rate": 1.8320066123863754e-05,
      "loss": 0.0435,
      "step": 4810
    },
    {
      "epoch": 11.74,
      "learning_rate": 1.8227817505653043e-05,
      "loss": 0.0452,
      "step": 4820
    },
    {
      "epoch": 11.77,
      "learning_rate": 1.8135668290595804e-05,
      "loss": 0.0313,
      "step": 4830
    },
    {
      "epoch": 11.79,
      "learning_rate": 1.8043619831272625e-05,
      "loss": 0.0592,
      "step": 4840
    },
    {
      "epoch": 11.81,
      "learning_rate": 1.795167347878518e-05,
      "loss": 0.0313,
      "step": 4850
    },
    {
      "epoch": 11.84,
      "learning_rate": 1.7859830582736404e-05,
      "loss": 0.0377,
      "step": 4860
    },
    {
      "epoch": 11.86,
      "learning_rate": 1.77680924912107e-05,
      "loss": 0.0661,
      "step": 4870
    },
    {
      "epoch": 11.89,
      "learning_rate": 1.7676460550754104e-05,
      "loss": 0.0356,
      "step": 4880
    },
    {
      "epoch": 11.91,
      "learning_rate": 1.758493610635457e-05,
      "loss": 0.0513,
      "step": 4890
    },
    {
      "epoch": 11.94,
      "learning_rate": 1.74935205014222e-05,
      "loss": 0.0421,
      "step": 4900
    },
    {
      "epoch": 11.96,
      "learning_rate": 1.740221507776954e-05,
      "loss": 0.054,
      "step": 4910
    },
    {
      "epoch": 11.99,
      "learning_rate": 1.731102117559187e-05,
      "loss": 0.0571,
      "step": 4920
    },
    {
      "epoch": 12.01,
      "learning_rate": 1.7219940133447544e-05,
      "loss": 0.0428,
      "step": 4930
    },
    {
      "epoch": 12.03,
      "learning_rate": 1.7128973288238344e-05,
      "loss": 0.0234,
      "step": 4940
    },
    {
      "epoch": 12.06,
      "learning_rate": 1.7047201869528476e-05,
      "loss": 0.0493,
      "step": 4950
    },
    {
      "epoch": 12.08,
      "learning_rate": 1.69564556756413e-05,
      "loss": 0.0273,
      "step": 4960
    },
    {
      "epoch": 12.11,
      "learning_rate": 1.6865827546155305e-05,
      "loss": 0.0274,
      "step": 4970
    },
    {
      "epoch": 12.13,
      "learning_rate": 1.6775318811324364e-05,
      "loss": 0.0227,
      "step": 4980
    },
    {
      "epoch": 12.16,
      "learning_rate": 1.668493079964982e-05,
      "loss": 0.0322,
      "step": 4990
    },
    {
      "epoch": 12.18,
      "learning_rate": 1.6594664837861045e-05,
      "loss": 0.0249,
      "step": 5000
    },
    {
      "epoch": 12.2,
      "learning_rate": 1.6504522250895954e-05,
      "loss": 0.0484,
      "step": 5010
    },
    {
      "epoch": 12.23,
      "learning_rate": 1.6414504361881512e-05,
      "loss": 0.0411,
      "step": 5020
    },
    {
      "epoch": 12.25,
      "learning_rate": 1.6324612492114378e-05,
      "loss": 0.0371,
      "step": 5030
    },
    {
      "epoch": 12.28,
      "learning_rate": 1.6234847961041465e-05,
      "loss": 0.03,
      "step": 5040
    },
    {
      "epoch": 12.3,
      "learning_rate": 1.6145212086240607e-05,
      "loss": 0.0254,
      "step": 5050
    },
    {
      "epoch": 12.33,
      "learning_rate": 1.6055706183401176e-05,
      "loss": 0.0338,
      "step": 5060
    },
    {
      "epoch": 12.35,
      "learning_rate": 1.596633156630481e-05,
      "loss": 0.0277,
      "step": 5070
    },
    {
      "epoch": 12.38,
      "learning_rate": 1.5877089546806125e-05,
      "loss": 0.0356,
      "step": 5080
    },
    {
      "epoch": 12.4,
      "learning_rate": 1.5787981434813416e-05,
      "loss": 0.0268,
      "step": 5090
    },
    {
      "epoch": 12.42,
      "learning_rate": 1.5699008538269494e-05,
      "loss": 0.0359,
      "step": 5100
    },
    {
      "epoch": 12.45,
      "learning_rate": 1.5610172163132445e-05,
      "loss": 0.0246,
      "step": 5110
    },
    {
      "epoch": 12.47,
      "learning_rate": 1.5521473613356476e-05,
      "loss": 0.0238,
      "step": 5120
    },
    {
      "epoch": 12.5,
      "learning_rate": 1.5432914190872757e-05,
      "loss": 0.028,
      "step": 5130
    },
    {
      "epoch": 12.52,
      "learning_rate": 1.5344495195570348e-05,
      "loss": 0.0356,
      "step": 5140
    },
    {
      "epoch": 12.55,
      "learning_rate": 1.5256217925277087e-05,
      "loss": 0.0509,
      "step": 5150
    },
    {
      "epoch": 12.57,
      "learning_rate": 1.5168083675740539e-05,
      "loss": 0.036,
      "step": 5160
    },
    {
      "epoch": 12.59,
      "learning_rate": 1.5080093740609008e-05,
      "loss": 0.0336,
      "step": 5170
    },
    {
      "epoch": 12.62,
      "learning_rate": 1.4992249411412513e-05,
      "loss": 0.0289,
      "step": 5180
    },
    {
      "epoch": 12.64,
      "learning_rate": 1.4904551977543859e-05,
      "loss": 0.0364,
      "step": 5190
    },
    {
      "epoch": 12.67,
      "learning_rate": 1.4817002726239682e-05,
      "loss": 0.0348,
      "step": 5200
    },
    {
      "epoch": 12.69,
      "learning_rate": 1.4729602942561588e-05,
      "loss": 0.0381,
      "step": 5210
    },
    {
      "epoch": 12.72,
      "learning_rate": 1.4642353909377268e-05,
      "loss": 0.039,
      "step": 5220
    },
    {
      "epoch": 12.74,
      "learning_rate": 1.4555256907341667e-05,
      "loss": 0.027,
      "step": 5230
    },
    {
      "epoch": 12.76,
      "learning_rate": 1.4468313214878199e-05,
      "loss": 0.029,
      "step": 5240
    },
    {
      "epoch": 12.79,
      "learning_rate": 1.4381524108159989e-05,
      "loss": 0.0477,
      "step": 5250
    },
    {
      "epoch": 12.81,
      "learning_rate": 1.4294890861091087e-05,
      "loss": 0.0416,
      "step": 5260
    },
    {
      "epoch": 12.84,
      "learning_rate": 1.4208414745287862e-05,
      "loss": 0.048,
      "step": 5270
    },
    {
      "epoch": 12.86,
      "learning_rate": 1.4122097030060249e-05,
      "loss": 0.03,
      "step": 5280
    },
    {
      "epoch": 12.89,
      "learning_rate": 1.4035938982393176e-05,
      "loss": 0.0434,
      "step": 5290
    },
    {
      "epoch": 12.91,
      "learning_rate": 1.3949941866927946e-05,
      "loss": 0.0295,
      "step": 5300
    },
    {
      "epoch": 12.94,
      "learning_rate": 1.3864106945943642e-05,
      "loss": 0.0311,
      "step": 5310
    },
    {
      "epoch": 12.96,
      "learning_rate": 1.3778435479338666e-05,
      "loss": 0.0353,
      "step": 5320
    },
    {
      "epoch": 12.98,
      "learning_rate": 1.3692928724612203e-05,
      "loss": 0.028,
      "step": 5330
    },
    {
      "epoch": 13.01,
      "learning_rate": 1.3607587936845761e-05,
      "loss": 0.0362,
      "step": 5340
    },
    {
      "epoch": 13.03,
      "learning_rate": 1.3522414368684783e-05,
      "loss": 0.0285,
      "step": 5350
    },
    {
      "epoch": 13.06,
      "learning_rate": 1.3445902163436381e-05,
      "loss": 0.0253,
      "step": 5360
    },
    {
      "epoch": 13.08,
      "learning_rate": 1.3361049754758403e-05,
      "loss": 0.0245,
      "step": 5370
    },
    {
      "epoch": 13.11,
      "learning_rate": 1.3276368184411857e-05,
      "loss": 0.0239,
      "step": 5380
    },
    {
      "epoch": 13.13,
      "learning_rate": 1.3191858695366082e-05,
      "loss": 0.0212,
      "step": 5390
    },
    {
      "epoch": 13.15,
      "learning_rate": 1.3107522528064552e-05,
      "loss": 0.0249,
      "step": 5400
    },
    {
      "epoch": 13.18,
      "learning_rate": 1.302336092040673e-05,
      "loss": 0.027,
      "step": 5410
    },
    {
      "epoch": 13.2,
      "learning_rate": 1.2939375107729852e-05,
      "loss": 0.0196,
      "step": 5420
    },
    {
      "epoch": 13.23,
      "learning_rate": 1.2855566322790797e-05,
      "loss": 0.019,
      "step": 5430
    },
    {
      "epoch": 13.25,
      "learning_rate": 1.2771935795748036e-05,
      "loss": 0.024,
      "step": 5440
    },
    {
      "epoch": 13.28,
      "learning_rate": 1.2688484754143493e-05,
      "loss": 0.0243,
      "step": 5450
    },
    {
      "epoch": 13.3,
      "learning_rate": 1.2605214422884615e-05,
      "loss": 0.0232,
      "step": 5460
    },
    {
      "epoch": 13.33,
      "learning_rate": 1.2522126024226346e-05,
      "loss": 0.0226,
      "step": 5470
    },
    {
      "epoch": 13.35,
      "learning_rate": 1.2439220777753193e-05,
      "loss": 0.0311,
      "step": 5480
    },
    {
      "epoch": 13.37,
      "learning_rate": 1.2356499900361332e-05,
      "loss": 0.0239,
      "step": 5490
    },
    {
      "epoch": 13.4,
      "learning_rate": 1.2273964606240718e-05,
      "loss": 0.0221,
      "step": 5500
    },
    {
      "epoch": 13.42,
      "learning_rate": 1.2191616106857312e-05,
      "loss": 0.0203,
      "step": 5510
    },
    {
      "epoch": 13.45,
      "learning_rate": 1.2109455610935261e-05,
      "loss": 0.0257,
      "step": 5520
    },
    {
      "epoch": 13.47,
      "learning_rate": 1.2027484324439178e-05,
      "loss": 0.0298,
      "step": 5530
    },
    {
      "epoch": 13.5,
      "learning_rate": 1.194570345055641e-05,
      "loss": 0.0242,
      "step": 5540
    },
    {
      "epoch": 13.52,
      "learning_rate": 1.1864114189679412e-05,
      "loss": 0.0285,
      "step": 5550
    },
    {
      "epoch": 13.54,
      "learning_rate": 1.1782717739388116e-05,
      "loss": 0.025,
      "step": 5560
    },
    {
      "epoch": 13.57,
      "learning_rate": 1.1701515294432346e-05,
      "loss": 0.02,
      "step": 5570
    },
    {
      "epoch": 13.59,
      "learning_rate": 1.162050804671429e-05,
      "loss": 0.0232,
      "step": 5580
    },
    {
      "epoch": 13.62,
      "learning_rate": 1.1539697185270982e-05,
      "loss": 0.0245,
      "step": 5590
    },
    {
      "epoch": 13.64,
      "learning_rate": 1.1459083896256887e-05,
      "loss": 0.0226,
      "step": 5600
    },
    {
      "epoch": 13.67,
      "learning_rate": 1.1378669362926467e-05,
      "loss": 0.0344,
      "step": 5610
    },
    {
      "epoch": 13.69,
      "learning_rate": 1.1298454765616812e-05,
      "loss": 0.031,
      "step": 5620
    },
    {
      "epoch": 13.71,
      "learning_rate": 1.1218441281730332e-05,
      "loss": 0.029,
      "step": 5630
    },
    {
      "epoch": 13.74,
      "learning_rate": 1.1138630085717438e-05,
      "loss": 0.0288,
      "step": 5640
    },
    {
      "epoch": 13.76,
      "learning_rate": 1.1059022349059361e-05,
      "loss": 0.0358,
      "step": 5650
    },
    {
      "epoch": 13.79,
      "learning_rate": 1.0979619240250888e-05,
      "loss": 0.0236,
      "step": 5660
    },
    {
      "epoch": 13.81,
      "learning_rate": 1.0900421924783271e-05,
      "loss": 0.0383,
      "step": 5670
    },
    {
      "epoch": 13.84,
      "learning_rate": 1.0821431565127102e-05,
      "loss": 0.0238,
      "step": 5680
    },
    {
      "epoch": 13.86,
      "learning_rate": 1.0742649320715209e-05,
      "loss": 0.0224,
      "step": 5690
    },
    {
      "epoch": 13.89,
      "learning_rate": 1.0664076347925703e-05,
      "loss": 0.0243,
      "step": 5700
    },
    {
      "epoch": 13.91,
      "learning_rate": 1.0585713800064963e-05,
      "loss": 0.0267,
      "step": 5710
    },
    {
      "epoch": 13.93,
      "learning_rate": 1.0507562827350719e-05,
      "loss": 0.0278,
      "step": 5720
    },
    {
      "epoch": 13.96,
      "learning_rate": 1.0429624576895179e-05,
      "loss": 0.0252,
      "step": 5730
    },
    {
      "epoch": 13.98,
      "learning_rate": 1.0351900192688144e-05,
      "loss": 0.0214,
      "step": 5740
    },
    {
      "epoch": 14.01,
      "learning_rate": 1.027439081558029e-05,
      "loss": 0.0271,
      "step": 5750
    },
    {
      "epoch": 14.03,
      "learning_rate": 1.0197097583266374e-05,
      "loss": 0.0194,
      "step": 5760
    },
    {
      "epoch": 14.06,
      "learning_rate": 1.0120021630268541e-05,
      "loss": 0.0247,
      "step": 5770
    },
    {
      "epoch": 14.08,
      "learning_rate": 1.0043164087919696e-05,
      "loss": 0.0166,
      "step": 5780
    },
    {
      "epoch": 14.1,
      "learning_rate": 9.966526084346836e-06,
      "loss": 0.025,
      "step": 5790
    },
    {
      "epoch": 14.13,
      "learning_rate": 9.890108744454573e-06,
      "loss": 0.0262,
      "step": 5800
    },
    {
      "epoch": 14.15,
      "learning_rate": 9.813913189908571e-06,
      "loss": 0.0163,
      "step": 5810
    },
    {
      "epoch": 14.18,
      "learning_rate": 9.73794053911909e-06,
      "loss": 0.0217,
      "step": 5820
    },
    {
      "epoch": 14.2,
      "learning_rate": 9.662191907224583e-06,
      "loss": 0.0171,
      "step": 5830
    },
    {
      "epoch": 14.23,
      "learning_rate": 9.5866684060753e-06,
      "loss": 0.0161,
      "step": 5840
    },
    {
      "epoch": 14.25,
      "learning_rate": 9.511371144217004e-06,
      "loss": 0.0265,
      "step": 5850
    },
    {
      "epoch": 14.28,
      "learning_rate": 9.436301226874677e-06,
      "loss": 0.0172,
      "step": 5860
    },
    {
      "epoch": 14.3,
      "learning_rate": 9.361459755936317e-06,
      "loss": 0.0179,
      "step": 5870
    },
    {
      "epoch": 14.32,
      "learning_rate": 9.28684782993671e-06,
      "loss": 0.0211,
      "step": 5880
    },
    {
      "epoch": 14.35,
      "learning_rate": 9.212466544041384e-06,
      "loss": 0.0221,
      "step": 5890
    },
    {
      "epoch": 14.37,
      "learning_rate": 9.138316990030485e-06,
      "loss": 0.0175,
      "step": 5900
    },
    {
      "epoch": 14.4,
      "learning_rate": 9.064400256282757e-06,
      "loss": 0.0177,
      "step": 5910
    },
    {
      "epoch": 14.42,
      "learning_rate": 8.990717427759585e-06,
      "loss": 0.0168,
      "step": 5920
    },
    {
      "epoch": 14.45,
      "learning_rate": 8.917269585989026e-06,
      "loss": 0.0242,
      "step": 5930
    },
    {
      "epoch": 14.47,
      "learning_rate": 8.844057809049999e-06,
      "loss": 0.0249,
      "step": 5940
    },
    {
      "epoch": 14.49,
      "learning_rate": 8.771083171556408e-06,
      "loss": 0.0158,
      "step": 5950
    },
    {
      "epoch": 14.52,
      "learning_rate": 8.698346744641392e-06,
      "loss": 0.0204,
      "step": 5960
    },
    {
      "epoch": 14.54,
      "learning_rate": 8.625849595941608e-06,
      "loss": 0.0225,
      "step": 5970
    },
    {
      "epoch": 14.57,
      "learning_rate": 8.553592789581518e-06,
      "loss": 0.0158,
      "step": 5980
    },
    {
      "epoch": 14.59,
      "learning_rate": 8.48157738615784e-06,
      "loss": 0.029,
      "step": 5990
    },
    {
      "epoch": 14.62,
      "learning_rate": 8.409804442723918e-06,
      "loss": 0.0187,
      "step": 6000
    },
    {
      "epoch": 14.64,
      "learning_rate": 8.338275012774247e-06,
      "loss": 0.0211,
      "step": 6010
    },
    {
      "epoch": 14.67,
      "learning_rate": 8.266990146228964e-06,
      "loss": 0.0184,
      "step": 6020
    },
    {
      "epoch": 14.69,
      "learning_rate": 8.195950889418502e-06,
      "loss": 0.0167,
      "step": 6030
    },
    {
      "epoch": 14.71,
      "learning_rate": 8.12515828506817e-06,
      "loss": 0.0199,
      "step": 6040
    },
    {
      "epoch": 14.74,
      "learning_rate": 8.054613372282891e-06,
      "loss": 0.0184,
      "step": 6050
    },
    {
      "epoch": 14.76,
      "learning_rate": 7.984317186531928e-06,
      "loss": 0.0196,
      "step": 6060
    },
    {
      "epoch": 14.79,
      "learning_rate": 7.914270759633668e-06,
      "loss": 0.0185,
      "step": 6070
    },
    {
      "epoch": 14.81,
      "learning_rate": 7.84447511974053e-06,
      "loss": 0.0178,
      "step": 6080
    },
    {
      "epoch": 14.84,
      "learning_rate": 7.774931291323825e-06,
      "loss": 0.022,
      "step": 6090
    },
    {
      "epoch": 14.86,
      "learning_rate": 7.705640295158741e-06,
      "loss": 0.0201,
      "step": 6100
    },
    {
      "epoch": 14.88,
      "learning_rate": 7.636603148309363e-06,
      "loss": 0.0221,
      "step": 6110
    },
    {
      "epoch": 14.91,
      "learning_rate": 7.567820864113706e-06,
      "loss": 0.0212,
      "step": 6120
    },
    {
      "epoch": 14.93,
      "learning_rate": 7.499294452168904e-06,
      "loss": 0.0174,
      "step": 6130
    },
    {
      "epoch": 14.96,
      "learning_rate": 7.431024918316343e-06,
      "loss": 0.0275,
      "step": 6140
    },
    {
      "epoch": 14.98,
      "learning_rate": 7.363013264626914e-06,
      "loss": 0.022,
      "step": 6150
    },
    {
      "epoch": 15.01,
      "learning_rate": 7.295260489386313e-06,
      "loss": 0.0215,
      "step": 6160
    },
    {
      "epoch": 15.03,
      "learning_rate": 7.22776758708035e-06,
      "loss": 0.0165,
      "step": 6170
    },
    {
      "epoch": 15.05,
      "learning_rate": 7.160535548380409e-06,
      "loss": 0.0145,
      "step": 6180
    },
    {
      "epoch": 15.08,
      "learning_rate": 7.093565360128862e-06,
      "loss": 0.0144,
      "step": 6190
    },
    {
      "epoch": 15.1,
      "learning_rate": 7.026858005324616e-06,
      "loss": 0.0173,
      "step": 6200
    },
    {
      "epoch": 15.13,
      "learning_rate": 6.96041446310863e-06,
      "loss": 0.0141,
      "step": 6210
    },
    {
      "epoch": 15.15,
      "learning_rate": 6.894235708749622e-06,
      "loss": 0.013,
      "step": 6220
    },
    {
      "epoch": 15.18,
      "learning_rate": 6.828322713629689e-06,
      "loss": 0.0174,
      "step": 6230
    },
    {
      "epoch": 15.2,
      "learning_rate": 6.762676445230084e-06,
      "loss": 0.0159,
      "step": 6240
    },
    {
      "epoch": 15.23,
      "learning_rate": 6.697297867117e-06,
      "loss": 0.016,
      "step": 6250
    },
    {
      "epoch": 15.25,
      "learning_rate": 6.63218793892742e-06,
      "loss": 0.0182,
      "step": 6260
    },
    {
      "epoch": 15.27,
      "learning_rate": 6.567347616355049e-06,
      "loss": 0.0157,
      "step": 6270
    },
    {
      "epoch": 15.3,
      "learning_rate": 6.502777851136288e-06,
      "loss": 0.0158,
      "step": 6280
    },
    {
      "epoch": 15.32,
      "learning_rate": 6.438479591036239e-06,
      "loss": 0.0147,
      "step": 6290
    },
    {
      "epoch": 15.35,
      "learning_rate": 6.37445377983483e-06,
      "loss": 0.0171,
      "step": 6300
    },
    {
      "epoch": 15.37,
      "learning_rate": 6.310701357312909e-06,
      "loss": 0.0189,
      "step": 6310
    },
    {
      "epoch": 15.4,
      "learning_rate": 6.247223259238511e-06,
      "loss": 0.0153,
      "step": 6320
    },
    {
      "epoch": 15.42,
      "learning_rate": 6.184020417353084e-06,
      "loss": 0.0179,
      "step": 6330
    },
    {
      "epoch": 15.44,
      "learning_rate": 6.121093759357824e-06,
      "loss": 0.0166,
      "step": 6340
    },
    {
      "epoch": 15.47,
      "learning_rate": 6.05844420890006e-06,
      "loss": 0.0226,
      "step": 6350
    },
    {
      "epoch": 15.49,
      "learning_rate": 5.996072685559675e-06,
      "loss": 0.0184,
      "step": 6360
    },
    {
      "epoch": 15.52,
      "learning_rate": 5.9339801048356515e-06,
      "loss": 0.0162,
      "step": 6370
    },
    {
      "epoch": 15.54,
      "learning_rate": 5.87216737813259e-06,
      "loss": 0.0177,
      "step": 6380
    },
    {
      "epoch": 15.57,
      "learning_rate": 5.810635412747373e-06,
      "loss": 0.0162,
      "step": 6390
    },
    {
      "epoch": 15.59,
      "learning_rate": 5.749385111855787e-06,
      "loss": 0.017,
      "step": 6400
    },
    {
      "epoch": 15.62,
      "learning_rate": 5.688417374499335e-06,
      "loss": 0.0165,
      "step": 6410
    },
    {
      "epoch": 15.64,
      "learning_rate": 5.627733095572002e-06,
      "loss": 0.0225,
      "step": 6420
    },
    {
      "epoch": 15.66,
      "learning_rate": 5.567333165807115e-06,
      "loss": 0.018,
      "step": 6430
    },
    {
      "epoch": 15.69,
      "learning_rate": 5.5072184717643e-06,
      "loss": 0.0156,
      "step": 6440
    },
    {
      "epoch": 15.71,
      "learning_rate": 5.447389895816415e-06,
      "loss": 0.017,
      "step": 6450
    },
    {
      "epoch": 15.74,
      "learning_rate": 5.387848316136665e-06,
      "loss": 0.0167,
      "step": 6460
    },
    {
      "epoch": 15.76,
      "learning_rate": 5.328594606685661e-06,
      "loss": 0.0149,
      "step": 6470
    },
    {
      "epoch": 15.79,
      "learning_rate": 5.269629637198617e-06,
      "loss": 0.0172,
      "step": 6480
    },
    {
      "epoch": 15.81,
      "learning_rate": 5.210954273172577e-06,
      "loss": 0.025,
      "step": 6490
    },
    {
      "epoch": 15.83,
      "learning_rate": 5.1525693758537025e-06,
      "loss": 0.0209,
      "step": 6500
    },
    {
      "epoch": 15.86,
      "learning_rate": 5.0944758022246435e-06,
      "loss": 0.0237,
      "step": 6510
    },
    {
      "epoch": 15.88,
      "learning_rate": 5.0366744049919614e-06,
      "loss": 0.0244,
      "step": 6520
    },
    {
      "epoch": 15.91,
      "learning_rate": 4.979166032573607e-06,
      "loss": 0.0161,
      "step": 6530
    },
    {
      "epoch": 15.93,
      "learning_rate": 4.921951529086447e-06,
      "loss": 0.0197,
      "step": 6540
    },
    {
      "epoch": 15.96,
      "learning_rate": 4.865031734333919e-06,
      "loss": 0.0206,
      "step": 6550
    },
    {
      "epoch": 15.98,
      "learning_rate": 4.808407483793667e-06,
      "loss": 0.0169,
      "step": 6560
    },
    {
      "epoch": 16.0,
      "learning_rate": 4.752079608605295e-06,
      "loss": 0.0167,
      "step": 6570
    },
    {
      "epoch": 16.03,
      "learning_rate": 4.696048935558167e-06,
      "loss": 0.0167,
      "step": 6580
    },
    {
      "epoch": 16.05,
      "learning_rate": 4.640316287079252e-06,
      "loss": 0.0116,
      "step": 6590
    },
    {
      "epoch": 16.08,
      "learning_rate": 4.584882481221081e-06,
      "loss": 0.0141,
      "step": 6600
    },
    {
      "epoch": 16.1,
      "learning_rate": 4.529748331649728e-06,
      "loss": 0.014,
      "step": 6610
    },
    {
      "epoch": 16.13,
      "learning_rate": 4.474914647632855e-06,
      "loss": 0.0145,
      "step": 6620
    },
    {
      "epoch": 16.15,
      "learning_rate": 4.420382234027859e-06,
      "loss": 0.0246,
      "step": 6630
    },
    {
      "epoch": 16.18,
      "learning_rate": 4.366151891270018e-06,
      "loss": 0.0137,
      "step": 6640
    },
    {
      "epoch": 16.2,
      "learning_rate": 4.312224415360791e-06,
      "loss": 0.0146,
      "step": 6650
    },
    {
      "epoch": 16.22,
      "learning_rate": 4.258600597856099e-06,
      "loss": 0.0143,
      "step": 6660
    },
    {
      "epoch": 16.25,
      "learning_rate": 4.205281225854727e-06,
      "loss": 0.0139,
      "step": 6670
    },
    {
      "epoch": 16.27,
      "learning_rate": 4.152267081986741e-06,
      "loss": 0.014,
      "step": 6680
    },
    {
      "epoch": 16.3,
      "learning_rate": 4.099558944402043e-06,
      "loss": 0.0135,
      "step": 6690
    },
    {
      "epoch": 16.32,
      "learning_rate": 4.047157586758907e-06,
      "loss": 0.0153,
      "step": 6700
    },
    {
      "epoch": 16.35,
      "learning_rate": 3.99506377821266e-06,
      "loss": 0.0145,
      "step": 6710
    },
    {
      "epoch": 16.37,
      "learning_rate": 3.943278283404369e-06,
      "loss": 0.0184,
      "step": 6720
    },
    {
      "epoch": 16.39,
      "learning_rate": 3.891801862449629e-06,
      "loss": 0.0153,
      "step": 6730
    },
    {
      "epoch": 16.42,
      "learning_rate": 3.840635270927381e-06,
      "loss": 0.0151,
      "step": 6740
    },
    {
      "epoch": 16.44,
      "learning_rate": 3.7897792598688643e-06,
      "loss": 0.0127,
      "step": 6750
    },
    {
      "epoch": 16.47,
      "learning_rate": 3.7392345757465586e-06,
      "loss": 0.0199,
      "step": 6760
    },
    {
      "epoch": 16.49,
      "learning_rate": 3.6890019604632424e-06,
      "loss": 0.0148,
      "step": 6770
    },
    {
      "epoch": 16.52,
      "learning_rate": 3.6390821513410974e-06,
      "loss": 0.0133,
      "step": 6780
    },
    {
      "epoch": 16.54,
      "learning_rate": 3.5894758811108793e-06,
      "loss": 0.014,
      "step": 6790
    },
    {
      "epoch": 16.57,
      "learning_rate": 3.540183877901182e-06,
      "loss": 0.0166,
      "step": 6800
    },
    {
      "epoch": 16.59,
      "learning_rate": 3.491206865227739e-06,
      "loss": 0.0166,
      "step": 6810
    },
    {
      "epoch": 16.61,
      "learning_rate": 3.4425455619827963e-06,
      "loss": 0.0163,
      "step": 6820
    },
    {
      "epoch": 16.64,
      "learning_rate": 3.394200682424578e-06,
      "loss": 0.0185,
      "step": 6830
    },
    {
      "epoch": 16.66,
      "learning_rate": 3.3461729361667722e-06,
      "loss": 0.015,
      "step": 6840
    },
    {
      "epoch": 16.69,
      "learning_rate": 3.2984630281681555e-06,
      "loss": 0.0171,
      "step": 6850
    },
    {
      "epoch": 16.71,
      "learning_rate": 3.251071658722213e-06,
      "loss": 0.016,
      "step": 6860
    },
    {
      "epoch": 16.74,
      "learning_rate": 3.203999523446885e-06,
      "loss": 0.02,
      "step": 6870
    },
    {
      "epoch": 16.76,
      "learning_rate": 3.1572473132743222e-06,
      "loss": 0.0132,
      "step": 6880
    },
    {
      "epoch": 16.78,
      "learning_rate": 3.1108157144407764e-06,
      "loss": 0.0157,
      "step": 6890
    },
    {
      "epoch": 16.81,
      "learning_rate": 3.0647054084765176e-06,
      "loss": 0.0149,
      "step": 6900
    },
    {
      "epoch": 16.83,
      "learning_rate": 3.018917072195823e-06,
      "loss": 0.0138,
      "step": 6910
    },
    {
      "epoch": 16.86,
      "learning_rate": 2.9734513776870616e-06,
      "loss": 0.0146,
      "step": 6920
    },
    {
      "epoch": 16.88,
      "learning_rate": 2.928308992302792e-06,
      "loss": 0.0188,
      "step": 6930
    },
    {
      "epoch": 16.91,
      "learning_rate": 2.8834905786500083e-06,
      "loss": 0.023,
      "step": 6940
    },
    {
      "epoch": 16.93,
      "learning_rate": 2.838996794580398e-06,
      "loss": 0.0158,
      "step": 6950
    },
    {
      "epoch": 16.95,
      "learning_rate": 2.7948282931806793e-06,
      "loss": 0.0142,
      "step": 6960
    },
    {
      "epoch": 16.98,
      "learning_rate": 2.750985722763022e-06,
      "loss": 0.0168,
      "step": 6970
    },
    {
      "epoch": 17.0,
      "learning_rate": 2.7074697268555244e-06,
      "loss": 0.0161,
      "step": 6980
    },
    {
      "epoch": 17.03,
      "learning_rate": 2.6642809441927817e-06,
      "loss": 0.0132,
      "step": 6990
    },
    {
      "epoch": 17.05,
      "learning_rate": 2.621420008706499e-06,
      "loss": 0.0131,
      "step": 7000
    },
    {
      "epoch": 17.08,
      "learning_rate": 2.5788875495161845e-06,
      "loss": 0.0122,
      "step": 7010
    },
    {
      "epoch": 17.1,
      "learning_rate": 2.5366841909199356e-06,
      "loss": 0.0142,
      "step": 7020
    },
    {
      "epoch": 17.13,
      "learning_rate": 2.4948105523852323e-06,
      "loss": 0.0234,
      "step": 7030
    },
    {
      "epoch": 17.15,
      "learning_rate": 2.4532672485398976e-06,
      "loss": 0.014,
      "step": 7040
    },
    {
      "epoch": 17.17,
      "learning_rate": 2.4120548891630347e-06,
      "loss": 0.0154,
      "step": 7050
    },
    {
      "epoch": 17.2,
      "learning_rate": 2.3711740791761083e-06,
      "loss": 0.0132,
      "step": 7060
    },
    {
      "epoch": 17.22,
      "learning_rate": 2.3306254186340303e-06,
      "loss": 0.013,
      "step": 7070
    },
    {
      "epoch": 17.25,
      "learning_rate": 2.2904095027163786e-06,
      "loss": 0.0242,
      "step": 7080
    },
    {
      "epoch": 17.27,
      "learning_rate": 2.2505269217186596e-06,
      "loss": 0.0143,
      "step": 7090
    },
    {
      "epoch": 17.3,
      "learning_rate": 2.210978261043631e-06,
      "loss": 0.0121,
      "step": 7100
    },
    {
      "epoch": 17.32,
      "learning_rate": 2.171764101192722e-06,
      "loss": 0.0138,
      "step": 7110
    },
    {
      "epoch": 17.34,
      "learning_rate": 2.1328850177574936e-06,
      "loss": 0.0133,
      "step": 7120
    },
    {
      "epoch": 17.37,
      "learning_rate": 2.0943415814112162e-06,
      "loss": 0.0125,
      "step": 7130
    },
    {
      "epoch": 17.39,
      "learning_rate": 2.0561343579004715e-06,
      "loss": 0.0145,
      "step": 7140
    },
    {
      "epoch": 17.42,
      "learning_rate": 2.018263908036863e-06,
      "loss": 0.0148,
      "step": 7150
    },
    {
      "epoch": 17.44,
      "learning_rate": 1.980730787688781e-06,
      "loss": 0.0133,
      "step": 7160
    },
    {
      "epoch": 17.47,
      "learning_rate": 1.9435355477732204e-06,
      "loss": 0.0129,
      "step": 7170
    },
    {
      "epoch": 17.49,
      "learning_rate": 1.9066787342477354e-06,
      "loss": 0.0147,
      "step": 7180
    },
    {
      "epoch": 17.52,
      "learning_rate": 1.8701608881023958e-06,
      "loss": 0.0154,
      "step": 7190
    },
    {
      "epoch": 17.54,
      "learning_rate": 1.8339825453518645e-06,
      "loss": 0.0134,
      "step": 7200
    },
    {
      "epoch": 17.56,
      "learning_rate": 1.7981442370274992e-06,
      "loss": 0.013,
      "step": 7210
    },
    {
      "epoch": 17.59,
      "learning_rate": 1.7626464891696031e-06,
      "loss": 0.0166,
      "step": 7220
    },
    {
      "epoch": 17.61,
      "learning_rate": 1.727489822819664e-06,
      "loss": 0.0122,
      "step": 7230
    },
    {
      "epoch": 17.64,
      "learning_rate": 1.6926747540127297e-06,
      "loss": 0.0139,
      "step": 7240
    },
    {
      "epoch": 17.66,
      "learning_rate": 1.6582017937698285e-06,
      "loss": 0.0141,
      "step": 7250
    },
    {
      "epoch": 17.69,
      "learning_rate": 1.62407144809045e-06,
      "loss": 0.0141,
      "step": 7260
    },
    {
      "epoch": 17.71,
      "learning_rate": 1.5902842179451482e-06,
      "loss": 0.0133,
      "step": 7270
    },
    {
      "epoch": 17.73,
      "learning_rate": 1.5568405992681722e-06,
      "loss": 0.015,
      "step": 7280
    },
    {
      "epoch": 17.76,
      "learning_rate": 1.5237410829501864e-06,
      "loss": 0.015,
      "step": 7290
    },
    {
      "epoch": 17.78,
      "learning_rate": 1.4909861548310672e-06,
      "loss": 0.0195,
      "step": 7300
    },
    {
      "epoch": 17.81,
      "learning_rate": 1.4585762956927624e-06,
      "loss": 0.0165,
      "step": 7310
    },
    {
      "epoch": 17.83,
      "learning_rate": 1.4265119812522597e-06,
      "loss": 0.0145,
      "step": 7320
    },
    {
      "epoch": 17.86,
      "learning_rate": 1.394793682154577e-06,
      "loss": 0.0125,
      "step": 7330
    },
    {
      "epoch": 17.88,
      "learning_rate": 1.363421863965869e-06,
      "loss": 0.0139,
      "step": 7340
    },
    {
      "epoch": 17.9,
      "learning_rate": 1.3323969871665897e-06,
      "loss": 0.015,
      "step": 7350
    },
    {
      "epoch": 17.93,
      "learning_rate": 1.3017195071447246e-06,
      "loss": 0.0149,
      "step": 7360
    },
    {
      "epoch": 17.95,
      "learning_rate": 1.2713898741891244e-06,
      "loss": 0.0184,
      "step": 7370
    },
    {
      "epoch": 17.98,
      "learning_rate": 1.2414085334828878e-06,
      "loss": 0.0132,
      "step": 7380
    },
    {
      "epoch": 18.0,
      "learning_rate": 1.2117759250968224e-06,
      "loss": 0.0149,
      "step": 7390
    },
    {
      "epoch": 18.03,
      "learning_rate": 1.1824924839829776e-06,
      "loss": 0.0134,
      "step": 7400
    },
    {
      "epoch": 18.05,
      "learning_rate": 1.1535586399682885e-06,
      "loss": 0.0126,
      "step": 7410
    },
    {
      "epoch": 18.08,
      "learning_rate": 1.1249748177482366e-06,
      "loss": 0.0137,
      "step": 7420
    },
    {
      "epoch": 18.1,
      "learning_rate": 1.0967414368806383e-06,
      "loss": 0.0134,
      "step": 7430
    },
    {
      "epoch": 18.12,
      "learning_rate": 1.0688589117794717e-06,
      "loss": 0.0129,
      "step": 7440
    },
    {
      "epoch": 18.15,
      "learning_rate": 1.0413276517087956e-06,
      "loss": 0.0121,
      "step": 7450
    },
    {
      "epoch": 18.17,
      "learning_rate": 1.0141480607767513e-06,
      "loss": 0.0159,
      "step": 7460
    },
    {
      "epoch": 18.2,
      "learning_rate": 9.87320537929623e-07,
      "loss": 0.0141,
      "step": 7470
    },
    {
      "epoch": 18.22,
      "learning_rate": 9.6084547694599e-07,
      "loss": 0.0122,
      "step": 7480
    },
    {
      "epoch": 18.25,
      "learning_rate": 9.347232664309368e-07,
      "loss": 0.013,
      "step": 7490
    },
    {
      "epoch": 18.27,
      "learning_rate": 9.089542898103459e-07,
      "loss": 0.0117,
      "step": 7500
    },
    {
      "epoch": 18.29,
      "learning_rate": 8.835389253252918e-07,
      "loss": 0.0143,
      "step": 7510
    },
    {
      "epoch": 18.32,
      "learning_rate": 8.584775460264621e-07,
      "loss": 0.0122,
      "step": 7520
    },
    {
      "epoch": 18.34,
      "learning_rate": 8.33770519768709e-07,
      "loss": 0.0122,
      "step": 7530
    },
    {
      "epoch": 18.37,
      "learning_rate": 8.09418209205623e-07,
      "loss": 0.0135,
      "step": 7540
    },
    {
      "epoch": 18.39,
      "learning_rate": 7.854209717842231e-07,
      "loss": 0.0141,
      "step": 7550
    },
    {
      "epoch": 18.42,
      "learning_rate": 7.617791597397117e-07,
      "loss": 0.0109,
      "step": 7560
    },
    {
      "epoch": 18.44,
      "learning_rate": 7.384931200903084e-07,
      "loss": 0.0124,
      "step": 7570
    },
    {
      "epoch": 18.47,
      "learning_rate": 7.155631946321406e-07,
      "loss": 0.012,
      "step": 7580
    },
    {
      "epoch": 18.49,
      "learning_rate": 6.929897199342395e-07,
      "loss": 0.0136,
      "step": 7590
    },
    {
      "epoch": 18.51,
      "learning_rate": 6.707730273335932e-07,
      "loss": 0.0159,
      "step": 7600
    },
    {
      "epoch": 18.54,
      "learning_rate": 6.489134429302907e-07,
      "loss": 0.016,
      "step": 7610
    },
    {
      "epoch": 18.56,
      "learning_rate": 6.274112875827326e-07,
      "loss": 0.012,
      "step": 7620
    },
    {
      "epoch": 18.59,
      "learning_rate": 6.062668769029167e-07,
      "loss": 0.0124,
      "step": 7630
    },
    {
      "epoch": 18.61,
      "learning_rate": 5.854805212518022e-07,
      "loss": 0.0205,
      "step": 7640
    },
    {
      "epoch": 18.64,
      "learning_rate": 5.650525257347744e-07,
      "loss": 0.0136,
      "step": 7650
    },
    {
      "epoch": 18.66,
      "learning_rate": 5.449831901971431e-07,
      "loss": 0.0133,
      "step": 7660
    },
    {
      "epoch": 18.68,
      "learning_rate": 5.252728092197539e-07,
      "loss": 0.015,
      "step": 7670
    },
    {
      "epoch": 18.71,
      "learning_rate": 5.059216721146592e-07,
      "loss": 0.0134,
      "step": 7680
    },
    {
      "epoch": 18.73,
      "learning_rate": 4.869300629208762e-07,
      "loss": 0.0125,
      "step": 7690
    },
    {
      "epoch": 18.76,
      "learning_rate": 4.6829826040021287e-07,
      "loss": 0.0141,
      "step": 7700
    },
    {
      "epoch": 18.78,
      "learning_rate": 4.500265380331797e-07,
      "loss": 0.0158,
      "step": 7710
    },
    {
      "epoch": 18.81,
      "learning_rate": 4.3211516401497344e-07,
      "loss": 0.013,
      "step": 7720
    },
    {
      "epoch": 18.83,
      "learning_rate": 4.145644012515465e-07,
      "loss": 0.0131,
      "step": 7730
    },
    {
      "epoch": 18.86,
      "learning_rate": 3.973745073557328e-07,
      "loss": 0.0152,
      "step": 7740
    },
    {
      "epoch": 18.88,
      "learning_rate": 3.805457346434865e-07,
      "loss": 0.0138,
      "step": 7750
    },
    {
      "epoch": 18.9,
      "learning_rate": 3.640783301301631e-07,
      "loss": 0.0147,
      "step": 7760
    },
    {
      "epoch": 18.93,
      "learning_rate": 3.479725355268998e-07,
      "loss": 0.0154,
      "step": 7770
    },
    {
      "epoch": 18.95,
      "learning_rate": 3.3222858723707395e-07,
      "loss": 0.0125,
      "step": 7780
    },
    {
      "epoch": 18.98,
      "learning_rate": 3.168467163528116e-07,
      "loss": 0.0236,
      "step": 7790
    },
    {
      "epoch": 19.0,
      "learning_rate": 3.018271486516233e-07,
      "loss": 0.0128,
      "step": 7800
    },
    {
      "epoch": 19.03,
      "learning_rate": 2.871701045930708e-07,
      "loss": 0.0168,
      "step": 7810
    },
    {
      "epoch": 19.05,
      "learning_rate": 2.72875799315539e-07,
      "loss": 0.0201,
      "step": 7820
    },
    {
      "epoch": 19.07,
      "learning_rate": 2.589444426330773e-07,
      "loss": 0.0121,
      "step": 7830
    },
    {
      "epoch": 19.1,
      "learning_rate": 2.453762390323133e-07,
      "loss": 0.0128,
      "step": 7840
    },
    {
      "epoch": 19.12,
      "learning_rate": 2.3217138766946366e-07,
      "loss": 0.0146,
      "step": 7850
    },
    {
      "epoch": 19.15,
      "learning_rate": 2.1933008236739993e-07,
      "loss": 0.0132,
      "step": 7860
    },
    {
      "epoch": 19.17,
      "learning_rate": 2.068525116128095e-07,
      "loss": 0.012,
      "step": 7870
    },
    {
      "epoch": 19.2,
      "learning_rate": 1.9473885855342822e-07,
      "loss": 0.012,
      "step": 7880
    },
    {
      "epoch": 19.22,
      "learning_rate": 1.8298930099534818e-07,
      "loss": 0.0162,
      "step": 7890
    },
    {
      "epoch": 19.24,
      "learning_rate": 1.716040114004114e-07,
      "loss": 0.0117,
      "step": 7900
    },
    {
      "epoch": 19.27,
      "learning_rate": 1.6058315688367854e-07,
      "loss": 0.0135,
      "step": 7910
    },
    {
      "epoch": 19.29,
      "learning_rate": 1.4992689921097814e-07,
      "loss": 0.0135,
      "step": 7920
    },
    {
      "epoch": 19.32,
      "learning_rate": 1.396353947965251e-07,
      "loss": 0.0144,
      "step": 7930
    },
    {
      "epoch": 19.34,
      "learning_rate": 1.2970879470062813e-07,
      "loss": 0.0128,
      "step": 7940
    },
    {
      "epoch": 19.37,
      "learning_rate": 1.2014724462747762e-07,
      "loss": 0.0121,
      "step": 7950
    },
    {
      "epoch": 19.39,
      "learning_rate": 1.109508849230001e-07,
      "loss": 0.0132,
      "step": 7960
    },
    {
      "epoch": 19.42,
      "learning_rate": 1.021198505728016e-07,
      "loss": 0.0124,
      "step": 7970
    },
    {
      "epoch": 19.44,
      "learning_rate": 9.365427120018311e-08,
      "loss": 0.0141,
      "step": 7980
    },
    {
      "epoch": 19.46,
      "learning_rate": 8.555427106424485e-08,
      "loss": 0.0154,
      "step": 7990
    },
    {
      "epoch": 19.49,
      "learning_rate": 7.781996905805167e-08,
      "loss": 0.0124,
      "step": 8000
    },
    {
      "epoch": 19.51,
      "learning_rate": 7.045147870690105e-08,
      "loss": 0.0118,
      "step": 8010
    },
    {
      "epoch": 19.54,
      "learning_rate": 6.344890816664673e-08,
      "loss": 0.0137,
      "step": 8020
    },
    {
      "epoch": 19.56,
      "learning_rate": 5.681236022211378e-08,
      "loss": 0.0134,
      "step": 8030
    },
    {
      "epoch": 19.59,
      "learning_rate": 5.0541932285586014e-08,
      "loss": 0.0123,
      "step": 8040
    },
    {
      "epoch": 19.61,
      "learning_rate": 4.463771639539038e-08,
      "loss": 0.0129,
      "step": 8050
    },
    {
      "epoch": 19.63,
      "learning_rate": 3.909979921452867e-08,
      "loss": 0.0132,
      "step": 8060
    },
    {
      "epoch": 19.66,
      "learning_rate": 3.392826202941179e-08,
      "loss": 0.0116,
      "step": 8070
    },
    {
      "epoch": 19.68,
      "learning_rate": 2.912318074867748e-08,
      "loss": 0.0134,
      "step": 8080
    },
    {
      "epoch": 19.71,
      "learning_rate": 2.4684625902057778e-08,
      "loss": 0.0115,
      "step": 8090
    },
    {
      "epoch": 19.73,
      "learning_rate": 2.061266263935213e-08,
      "loss": 0.014,
      "step": 8100
    },
    {
      "epoch": 19.76,
      "learning_rate": 1.6907350729478134e-08,
      "loss": 0.0131,
      "step": 8110
    },
    {
      "epoch": 19.78,
      "learning_rate": 1.356874455958057e-08,
      "loss": 0.0145,
      "step": 8120
    },
    {
      "epoch": 19.81,
      "learning_rate": 1.0596893134240393e-08,
      "loss": 0.0145,
      "step": 8130
    },
    {
      "epoch": 19.83,
      "learning_rate": 7.991840074764167e-09,
      "loss": 0.0131,
      "step": 8140
    },
    {
      "epoch": 19.85,
      "learning_rate": 5.75362361852072e-09,
      "loss": 0.0126,
      "step": 8150
    },
    {
      "epoch": 19.88,
      "learning_rate": 3.882276618405456e-09,
      "loss": 0.0131,
      "step": 8160
    },
    {
      "epoch": 19.9,
      "learning_rate": 2.377826542343531e-09,
      "loss": 0.0149,
      "step": 8170
    },
    {
      "epoch": 19.93,
      "learning_rate": 1.2402954728846228e-09,
      "loss": 0.013,
      "step": 8180
    },
    {
      "epoch": 19.95,
      "learning_rate": 4.697001068892925e-10,
      "loss": 0.0127,
      "step": 8190
    },
    {
      "epoch": 19.98,
      "learning_rate": 6.605175527640839e-11,
      "loss": 0.0116,
      "step": 8200
    },
    {
      "epoch": 19.98,
      "step": 8200,
      "total_flos": 3.0532626578313216e+17,
      "train_loss": 0.2642190715797791,
      "train_runtime": 4480.4571,
      "train_samples_per_second": 29.305,
      "train_steps_per_second": 1.83
    }
  ],
  "max_steps": 8200,
  "num_train_epochs": 20,
  "total_flos": 3.0532626578313216e+17,
  "trial_name": null,
  "trial_params": null
}