medtextclassifier / trainer_state.json
devkushal75's picture
Upload trainer_state.json
c9ddb93
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 48.0,
"global_step": 6000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.08,
"learning_rate": 1.0000000000000002e-06,
"loss": 3.6493,
"step": 10
},
{
"epoch": 0.16,
"learning_rate": 2.0000000000000003e-06,
"loss": 3.6538,
"step": 20
},
{
"epoch": 0.24,
"learning_rate": 3e-06,
"loss": 3.6342,
"step": 30
},
{
"epoch": 0.32,
"learning_rate": 4.000000000000001e-06,
"loss": 3.5774,
"step": 40
},
{
"epoch": 0.4,
"learning_rate": 5e-06,
"loss": 3.5288,
"step": 50
},
{
"epoch": 0.48,
"learning_rate": 6e-06,
"loss": 3.452,
"step": 60
},
{
"epoch": 0.56,
"learning_rate": 7.000000000000001e-06,
"loss": 3.3685,
"step": 70
},
{
"epoch": 0.64,
"learning_rate": 8.000000000000001e-06,
"loss": 3.2667,
"step": 80
},
{
"epoch": 0.72,
"learning_rate": 9e-06,
"loss": 3.2796,
"step": 90
},
{
"epoch": 0.8,
"learning_rate": 1e-05,
"loss": 3.1435,
"step": 100
},
{
"epoch": 0.88,
"learning_rate": 1.1000000000000001e-05,
"loss": 3.1206,
"step": 110
},
{
"epoch": 0.96,
"learning_rate": 1.2e-05,
"loss": 3.0387,
"step": 120
},
{
"epoch": 1.04,
"learning_rate": 1.3000000000000001e-05,
"loss": 2.9759,
"step": 130
},
{
"epoch": 1.12,
"learning_rate": 1.4000000000000001e-05,
"loss": 3.0132,
"step": 140
},
{
"epoch": 1.2,
"learning_rate": 1.5e-05,
"loss": 2.9027,
"step": 150
},
{
"epoch": 1.28,
"learning_rate": 1.6000000000000003e-05,
"loss": 2.815,
"step": 160
},
{
"epoch": 1.36,
"learning_rate": 1.7000000000000003e-05,
"loss": 2.8081,
"step": 170
},
{
"epoch": 1.44,
"learning_rate": 1.8e-05,
"loss": 2.8926,
"step": 180
},
{
"epoch": 1.52,
"learning_rate": 1.9e-05,
"loss": 2.8096,
"step": 190
},
{
"epoch": 1.6,
"learning_rate": 2e-05,
"loss": 2.7153,
"step": 200
},
{
"epoch": 1.68,
"learning_rate": 2.1e-05,
"loss": 2.7193,
"step": 210
},
{
"epoch": 1.76,
"learning_rate": 2.2000000000000003e-05,
"loss": 2.5646,
"step": 220
},
{
"epoch": 1.84,
"learning_rate": 2.3000000000000003e-05,
"loss": 2.7471,
"step": 230
},
{
"epoch": 1.92,
"learning_rate": 2.4e-05,
"loss": 2.7368,
"step": 240
},
{
"epoch": 2.0,
"learning_rate": 2.5e-05,
"loss": 2.6167,
"step": 250
},
{
"epoch": 2.08,
"learning_rate": 2.6000000000000002e-05,
"loss": 2.492,
"step": 260
},
{
"epoch": 2.16,
"learning_rate": 2.7000000000000002e-05,
"loss": 2.5191,
"step": 270
},
{
"epoch": 2.24,
"learning_rate": 2.8000000000000003e-05,
"loss": 2.5223,
"step": 280
},
{
"epoch": 2.32,
"learning_rate": 2.9e-05,
"loss": 2.4294,
"step": 290
},
{
"epoch": 2.4,
"learning_rate": 3e-05,
"loss": 2.3719,
"step": 300
},
{
"epoch": 2.48,
"learning_rate": 3.1e-05,
"loss": 2.4148,
"step": 310
},
{
"epoch": 2.56,
"learning_rate": 3.2000000000000005e-05,
"loss": 2.4787,
"step": 320
},
{
"epoch": 2.64,
"learning_rate": 3.3e-05,
"loss": 2.4021,
"step": 330
},
{
"epoch": 2.72,
"learning_rate": 3.4000000000000007e-05,
"loss": 2.3982,
"step": 340
},
{
"epoch": 2.8,
"learning_rate": 3.5e-05,
"loss": 2.28,
"step": 350
},
{
"epoch": 2.88,
"learning_rate": 3.6e-05,
"loss": 2.4302,
"step": 360
},
{
"epoch": 2.96,
"learning_rate": 3.7e-05,
"loss": 2.2597,
"step": 370
},
{
"epoch": 3.04,
"learning_rate": 3.8e-05,
"loss": 2.0781,
"step": 380
},
{
"epoch": 3.12,
"learning_rate": 3.9000000000000006e-05,
"loss": 2.0256,
"step": 390
},
{
"epoch": 3.2,
"learning_rate": 4e-05,
"loss": 2.1566,
"step": 400
},
{
"epoch": 3.28,
"learning_rate": 4.1e-05,
"loss": 2.0487,
"step": 410
},
{
"epoch": 3.36,
"learning_rate": 4.2e-05,
"loss": 2.0661,
"step": 420
},
{
"epoch": 3.44,
"learning_rate": 4.3e-05,
"loss": 2.0816,
"step": 430
},
{
"epoch": 3.52,
"learning_rate": 4.4000000000000006e-05,
"loss": 2.1713,
"step": 440
},
{
"epoch": 3.6,
"learning_rate": 4.5e-05,
"loss": 2.151,
"step": 450
},
{
"epoch": 3.68,
"learning_rate": 4.600000000000001e-05,
"loss": 1.9657,
"step": 460
},
{
"epoch": 3.76,
"learning_rate": 4.7e-05,
"loss": 2.0014,
"step": 470
},
{
"epoch": 3.84,
"learning_rate": 4.8e-05,
"loss": 1.9713,
"step": 480
},
{
"epoch": 3.92,
"learning_rate": 4.9e-05,
"loss": 1.936,
"step": 490
},
{
"epoch": 4.0,
"learning_rate": 5e-05,
"loss": 1.9813,
"step": 500
},
{
"epoch": 4.08,
"learning_rate": 4.9913043478260876e-05,
"loss": 1.7488,
"step": 510
},
{
"epoch": 4.16,
"learning_rate": 4.9826086956521736e-05,
"loss": 1.7985,
"step": 520
},
{
"epoch": 4.24,
"learning_rate": 4.973913043478261e-05,
"loss": 1.7398,
"step": 530
},
{
"epoch": 4.32,
"learning_rate": 4.9652173913043483e-05,
"loss": 1.5946,
"step": 540
},
{
"epoch": 4.4,
"learning_rate": 4.956521739130435e-05,
"loss": 1.879,
"step": 550
},
{
"epoch": 4.48,
"learning_rate": 4.947826086956522e-05,
"loss": 1.8759,
"step": 560
},
{
"epoch": 4.56,
"learning_rate": 4.939130434782609e-05,
"loss": 1.7932,
"step": 570
},
{
"epoch": 4.64,
"learning_rate": 4.930434782608696e-05,
"loss": 1.7335,
"step": 580
},
{
"epoch": 4.72,
"learning_rate": 4.9217391304347824e-05,
"loss": 1.7787,
"step": 590
},
{
"epoch": 4.8,
"learning_rate": 4.91304347826087e-05,
"loss": 1.8258,
"step": 600
},
{
"epoch": 4.88,
"learning_rate": 4.904347826086957e-05,
"loss": 1.7937,
"step": 610
},
{
"epoch": 4.96,
"learning_rate": 4.895652173913044e-05,
"loss": 1.734,
"step": 620
},
{
"epoch": 5.04,
"learning_rate": 4.8869565217391305e-05,
"loss": 1.5941,
"step": 630
},
{
"epoch": 5.12,
"learning_rate": 4.878260869565218e-05,
"loss": 1.6789,
"step": 640
},
{
"epoch": 5.2,
"learning_rate": 4.8695652173913046e-05,
"loss": 1.4186,
"step": 650
},
{
"epoch": 5.28,
"learning_rate": 4.860869565217391e-05,
"loss": 1.5457,
"step": 660
},
{
"epoch": 5.36,
"learning_rate": 4.8521739130434786e-05,
"loss": 1.5282,
"step": 670
},
{
"epoch": 5.44,
"learning_rate": 4.843478260869565e-05,
"loss": 1.5594,
"step": 680
},
{
"epoch": 5.52,
"learning_rate": 4.834782608695652e-05,
"loss": 1.7068,
"step": 690
},
{
"epoch": 5.6,
"learning_rate": 4.8260869565217394e-05,
"loss": 1.5521,
"step": 700
},
{
"epoch": 5.68,
"learning_rate": 4.817391304347826e-05,
"loss": 1.5973,
"step": 710
},
{
"epoch": 5.76,
"learning_rate": 4.8086956521739134e-05,
"loss": 1.5842,
"step": 720
},
{
"epoch": 5.84,
"learning_rate": 4.8e-05,
"loss": 1.5819,
"step": 730
},
{
"epoch": 5.92,
"learning_rate": 4.7913043478260875e-05,
"loss": 1.6313,
"step": 740
},
{
"epoch": 6.0,
"learning_rate": 4.782608695652174e-05,
"loss": 1.5262,
"step": 750
},
{
"epoch": 6.08,
"learning_rate": 4.773913043478261e-05,
"loss": 1.3722,
"step": 760
},
{
"epoch": 6.16,
"learning_rate": 4.765217391304348e-05,
"loss": 1.2905,
"step": 770
},
{
"epoch": 6.24,
"learning_rate": 4.756521739130435e-05,
"loss": 1.4479,
"step": 780
},
{
"epoch": 6.32,
"learning_rate": 4.747826086956522e-05,
"loss": 1.424,
"step": 790
},
{
"epoch": 6.4,
"learning_rate": 4.739130434782609e-05,
"loss": 1.3006,
"step": 800
},
{
"epoch": 6.48,
"learning_rate": 4.7304347826086956e-05,
"loss": 1.4776,
"step": 810
},
{
"epoch": 6.56,
"learning_rate": 4.721739130434783e-05,
"loss": 1.4928,
"step": 820
},
{
"epoch": 6.64,
"learning_rate": 4.71304347826087e-05,
"loss": 1.4294,
"step": 830
},
{
"epoch": 6.72,
"learning_rate": 4.7043478260869564e-05,
"loss": 1.386,
"step": 840
},
{
"epoch": 6.8,
"learning_rate": 4.695652173913044e-05,
"loss": 1.4618,
"step": 850
},
{
"epoch": 6.88,
"learning_rate": 4.686956521739131e-05,
"loss": 1.4186,
"step": 860
},
{
"epoch": 6.96,
"learning_rate": 4.678260869565218e-05,
"loss": 1.6092,
"step": 870
},
{
"epoch": 7.04,
"learning_rate": 4.6695652173913045e-05,
"loss": 1.3375,
"step": 880
},
{
"epoch": 7.12,
"learning_rate": 4.660869565217392e-05,
"loss": 1.2338,
"step": 890
},
{
"epoch": 7.2,
"learning_rate": 4.6521739130434785e-05,
"loss": 1.3258,
"step": 900
},
{
"epoch": 7.28,
"learning_rate": 4.643478260869565e-05,
"loss": 1.2257,
"step": 910
},
{
"epoch": 7.36,
"learning_rate": 4.6347826086956526e-05,
"loss": 1.2538,
"step": 920
},
{
"epoch": 7.44,
"learning_rate": 4.62608695652174e-05,
"loss": 1.3203,
"step": 930
},
{
"epoch": 7.52,
"learning_rate": 4.617391304347826e-05,
"loss": 1.2229,
"step": 940
},
{
"epoch": 7.6,
"learning_rate": 4.608695652173913e-05,
"loss": 1.31,
"step": 950
},
{
"epoch": 7.68,
"learning_rate": 4.600000000000001e-05,
"loss": 1.3021,
"step": 960
},
{
"epoch": 7.76,
"learning_rate": 4.591304347826087e-05,
"loss": 1.3159,
"step": 970
},
{
"epoch": 7.84,
"learning_rate": 4.582608695652174e-05,
"loss": 1.3946,
"step": 980
},
{
"epoch": 7.92,
"learning_rate": 4.5739130434782614e-05,
"loss": 1.4731,
"step": 990
},
{
"epoch": 8.0,
"learning_rate": 4.565217391304348e-05,
"loss": 1.4642,
"step": 1000
},
{
"epoch": 8.08,
"learning_rate": 4.556521739130435e-05,
"loss": 1.173,
"step": 1010
},
{
"epoch": 8.16,
"learning_rate": 4.547826086956522e-05,
"loss": 1.1411,
"step": 1020
},
{
"epoch": 8.24,
"learning_rate": 4.539130434782609e-05,
"loss": 1.1576,
"step": 1030
},
{
"epoch": 8.32,
"learning_rate": 4.5304347826086955e-05,
"loss": 1.1805,
"step": 1040
},
{
"epoch": 8.4,
"learning_rate": 4.521739130434783e-05,
"loss": 1.1661,
"step": 1050
},
{
"epoch": 8.48,
"learning_rate": 4.51304347826087e-05,
"loss": 1.1878,
"step": 1060
},
{
"epoch": 8.56,
"learning_rate": 4.504347826086956e-05,
"loss": 1.2674,
"step": 1070
},
{
"epoch": 8.64,
"learning_rate": 4.4956521739130436e-05,
"loss": 1.1968,
"step": 1080
},
{
"epoch": 8.72,
"learning_rate": 4.486956521739131e-05,
"loss": 1.3326,
"step": 1090
},
{
"epoch": 8.8,
"learning_rate": 4.478260869565218e-05,
"loss": 1.1905,
"step": 1100
},
{
"epoch": 8.88,
"learning_rate": 4.4695652173913044e-05,
"loss": 1.3127,
"step": 1110
},
{
"epoch": 8.96,
"learning_rate": 4.460869565217392e-05,
"loss": 1.3091,
"step": 1120
},
{
"epoch": 9.04,
"learning_rate": 4.4521739130434784e-05,
"loss": 1.1955,
"step": 1130
},
{
"epoch": 9.12,
"learning_rate": 4.443478260869565e-05,
"loss": 1.0431,
"step": 1140
},
{
"epoch": 9.2,
"learning_rate": 4.4347826086956525e-05,
"loss": 1.0506,
"step": 1150
},
{
"epoch": 9.28,
"learning_rate": 4.426086956521739e-05,
"loss": 1.0924,
"step": 1160
},
{
"epoch": 9.36,
"learning_rate": 4.4173913043478265e-05,
"loss": 1.1698,
"step": 1170
},
{
"epoch": 9.44,
"learning_rate": 4.408695652173913e-05,
"loss": 1.2168,
"step": 1180
},
{
"epoch": 9.52,
"learning_rate": 4.4000000000000006e-05,
"loss": 1.1931,
"step": 1190
},
{
"epoch": 9.6,
"learning_rate": 4.391304347826087e-05,
"loss": 1.2466,
"step": 1200
},
{
"epoch": 9.68,
"learning_rate": 4.382608695652174e-05,
"loss": 1.1826,
"step": 1210
},
{
"epoch": 9.76,
"learning_rate": 4.373913043478261e-05,
"loss": 1.1994,
"step": 1220
},
{
"epoch": 9.84,
"learning_rate": 4.365217391304348e-05,
"loss": 1.116,
"step": 1230
},
{
"epoch": 9.92,
"learning_rate": 4.3565217391304353e-05,
"loss": 1.1418,
"step": 1240
},
{
"epoch": 10.0,
"learning_rate": 4.347826086956522e-05,
"loss": 1.2596,
"step": 1250
},
{
"epoch": 10.08,
"learning_rate": 4.339130434782609e-05,
"loss": 1.0032,
"step": 1260
},
{
"epoch": 10.16,
"learning_rate": 4.330434782608696e-05,
"loss": 1.0681,
"step": 1270
},
{
"epoch": 10.24,
"learning_rate": 4.321739130434783e-05,
"loss": 1.0991,
"step": 1280
},
{
"epoch": 10.32,
"learning_rate": 4.3130434782608695e-05,
"loss": 0.9904,
"step": 1290
},
{
"epoch": 10.4,
"learning_rate": 4.304347826086957e-05,
"loss": 1.037,
"step": 1300
},
{
"epoch": 10.48,
"learning_rate": 4.2956521739130435e-05,
"loss": 1.1417,
"step": 1310
},
{
"epoch": 10.56,
"learning_rate": 4.286956521739131e-05,
"loss": 1.0919,
"step": 1320
},
{
"epoch": 10.64,
"learning_rate": 4.2782608695652176e-05,
"loss": 1.0887,
"step": 1330
},
{
"epoch": 10.72,
"learning_rate": 4.269565217391305e-05,
"loss": 1.1425,
"step": 1340
},
{
"epoch": 10.8,
"learning_rate": 4.2608695652173916e-05,
"loss": 1.2263,
"step": 1350
},
{
"epoch": 10.88,
"learning_rate": 4.252173913043478e-05,
"loss": 1.2244,
"step": 1360
},
{
"epoch": 10.96,
"learning_rate": 4.2434782608695657e-05,
"loss": 1.0981,
"step": 1370
},
{
"epoch": 11.04,
"learning_rate": 4.2347826086956523e-05,
"loss": 1.1665,
"step": 1380
},
{
"epoch": 11.12,
"learning_rate": 4.226086956521739e-05,
"loss": 0.9729,
"step": 1390
},
{
"epoch": 11.2,
"learning_rate": 4.2173913043478264e-05,
"loss": 1.0235,
"step": 1400
},
{
"epoch": 11.28,
"learning_rate": 4.208695652173914e-05,
"loss": 1.0718,
"step": 1410
},
{
"epoch": 11.36,
"learning_rate": 4.2e-05,
"loss": 1.1057,
"step": 1420
},
{
"epoch": 11.44,
"learning_rate": 4.191304347826087e-05,
"loss": 1.0105,
"step": 1430
},
{
"epoch": 11.52,
"learning_rate": 4.1826086956521745e-05,
"loss": 1.0284,
"step": 1440
},
{
"epoch": 11.6,
"learning_rate": 4.1739130434782605e-05,
"loss": 0.9386,
"step": 1450
},
{
"epoch": 11.68,
"learning_rate": 4.165217391304348e-05,
"loss": 1.0335,
"step": 1460
},
{
"epoch": 11.76,
"learning_rate": 4.156521739130435e-05,
"loss": 1.125,
"step": 1470
},
{
"epoch": 11.84,
"learning_rate": 4.147826086956522e-05,
"loss": 1.1251,
"step": 1480
},
{
"epoch": 11.92,
"learning_rate": 4.1391304347826086e-05,
"loss": 1.1962,
"step": 1490
},
{
"epoch": 12.0,
"learning_rate": 4.130434782608696e-05,
"loss": 1.1286,
"step": 1500
},
{
"epoch": 12.08,
"learning_rate": 4.1217391304347827e-05,
"loss": 0.9152,
"step": 1510
},
{
"epoch": 12.16,
"learning_rate": 4.1130434782608693e-05,
"loss": 0.9163,
"step": 1520
},
{
"epoch": 12.24,
"learning_rate": 4.104347826086957e-05,
"loss": 0.959,
"step": 1530
},
{
"epoch": 12.32,
"learning_rate": 4.095652173913044e-05,
"loss": 1.006,
"step": 1540
},
{
"epoch": 12.4,
"learning_rate": 4.086956521739131e-05,
"loss": 1.0188,
"step": 1550
},
{
"epoch": 12.48,
"learning_rate": 4.0782608695652174e-05,
"loss": 0.9931,
"step": 1560
},
{
"epoch": 12.56,
"learning_rate": 4.069565217391305e-05,
"loss": 1.0445,
"step": 1570
},
{
"epoch": 12.64,
"learning_rate": 4.0608695652173915e-05,
"loss": 1.0901,
"step": 1580
},
{
"epoch": 12.72,
"learning_rate": 4.052173913043478e-05,
"loss": 0.9693,
"step": 1590
},
{
"epoch": 12.8,
"learning_rate": 4.0434782608695655e-05,
"loss": 1.1191,
"step": 1600
},
{
"epoch": 12.88,
"learning_rate": 4.034782608695652e-05,
"loss": 1.1188,
"step": 1610
},
{
"epoch": 12.96,
"learning_rate": 4.026086956521739e-05,
"loss": 1.1422,
"step": 1620
},
{
"epoch": 13.04,
"learning_rate": 4.017391304347826e-05,
"loss": 1.0245,
"step": 1630
},
{
"epoch": 13.12,
"learning_rate": 4.008695652173913e-05,
"loss": 0.9048,
"step": 1640
},
{
"epoch": 13.2,
"learning_rate": 4e-05,
"loss": 0.925,
"step": 1650
},
{
"epoch": 13.28,
"learning_rate": 3.991304347826087e-05,
"loss": 0.8803,
"step": 1660
},
{
"epoch": 13.36,
"learning_rate": 3.9826086956521744e-05,
"loss": 0.9284,
"step": 1670
},
{
"epoch": 13.44,
"learning_rate": 3.973913043478261e-05,
"loss": 0.9985,
"step": 1680
},
{
"epoch": 13.52,
"learning_rate": 3.965217391304348e-05,
"loss": 1.0461,
"step": 1690
},
{
"epoch": 13.6,
"learning_rate": 3.956521739130435e-05,
"loss": 0.9542,
"step": 1700
},
{
"epoch": 13.68,
"learning_rate": 3.947826086956522e-05,
"loss": 0.9822,
"step": 1710
},
{
"epoch": 13.76,
"learning_rate": 3.939130434782609e-05,
"loss": 1.0781,
"step": 1720
},
{
"epoch": 13.84,
"learning_rate": 3.930434782608696e-05,
"loss": 1.0549,
"step": 1730
},
{
"epoch": 13.92,
"learning_rate": 3.9217391304347825e-05,
"loss": 1.0676,
"step": 1740
},
{
"epoch": 14.0,
"learning_rate": 3.91304347826087e-05,
"loss": 1.1258,
"step": 1750
},
{
"epoch": 14.08,
"learning_rate": 3.9043478260869566e-05,
"loss": 0.8544,
"step": 1760
},
{
"epoch": 14.16,
"learning_rate": 3.895652173913043e-05,
"loss": 0.9229,
"step": 1770
},
{
"epoch": 14.24,
"learning_rate": 3.8869565217391306e-05,
"loss": 0.9222,
"step": 1780
},
{
"epoch": 14.32,
"learning_rate": 3.878260869565218e-05,
"loss": 0.9293,
"step": 1790
},
{
"epoch": 14.4,
"learning_rate": 3.869565217391305e-05,
"loss": 0.9189,
"step": 1800
},
{
"epoch": 14.48,
"learning_rate": 3.8608695652173914e-05,
"loss": 0.9336,
"step": 1810
},
{
"epoch": 14.56,
"learning_rate": 3.852173913043479e-05,
"loss": 0.9787,
"step": 1820
},
{
"epoch": 14.64,
"learning_rate": 3.8434782608695654e-05,
"loss": 1.0282,
"step": 1830
},
{
"epoch": 14.72,
"learning_rate": 3.834782608695652e-05,
"loss": 0.9987,
"step": 1840
},
{
"epoch": 14.8,
"learning_rate": 3.8260869565217395e-05,
"loss": 1.0231,
"step": 1850
},
{
"epoch": 14.88,
"learning_rate": 3.817391304347827e-05,
"loss": 1.0009,
"step": 1860
},
{
"epoch": 14.96,
"learning_rate": 3.808695652173913e-05,
"loss": 1.0475,
"step": 1870
},
{
"epoch": 15.04,
"learning_rate": 3.8e-05,
"loss": 0.9605,
"step": 1880
},
{
"epoch": 15.12,
"learning_rate": 3.7913043478260876e-05,
"loss": 0.8575,
"step": 1890
},
{
"epoch": 15.2,
"learning_rate": 3.7826086956521736e-05,
"loss": 0.8875,
"step": 1900
},
{
"epoch": 15.28,
"learning_rate": 3.773913043478261e-05,
"loss": 0.9066,
"step": 1910
},
{
"epoch": 15.36,
"learning_rate": 3.765217391304348e-05,
"loss": 0.9074,
"step": 1920
},
{
"epoch": 15.44,
"learning_rate": 3.756521739130435e-05,
"loss": 0.9326,
"step": 1930
},
{
"epoch": 15.52,
"learning_rate": 3.747826086956522e-05,
"loss": 0.8979,
"step": 1940
},
{
"epoch": 15.6,
"learning_rate": 3.739130434782609e-05,
"loss": 0.9302,
"step": 1950
},
{
"epoch": 15.68,
"learning_rate": 3.730434782608696e-05,
"loss": 0.9764,
"step": 1960
},
{
"epoch": 15.76,
"learning_rate": 3.7217391304347824e-05,
"loss": 0.9775,
"step": 1970
},
{
"epoch": 15.84,
"learning_rate": 3.71304347826087e-05,
"loss": 0.9807,
"step": 1980
},
{
"epoch": 15.92,
"learning_rate": 3.704347826086957e-05,
"loss": 1.0364,
"step": 1990
},
{
"epoch": 16.0,
"learning_rate": 3.695652173913043e-05,
"loss": 1.0756,
"step": 2000
},
{
"epoch": 16.08,
"learning_rate": 3.6869565217391305e-05,
"loss": 0.8401,
"step": 2010
},
{
"epoch": 16.16,
"learning_rate": 3.678260869565218e-05,
"loss": 0.8194,
"step": 2020
},
{
"epoch": 16.24,
"learning_rate": 3.6695652173913046e-05,
"loss": 0.9139,
"step": 2030
},
{
"epoch": 16.32,
"learning_rate": 3.660869565217391e-05,
"loss": 0.8998,
"step": 2040
},
{
"epoch": 16.4,
"learning_rate": 3.6521739130434786e-05,
"loss": 0.9026,
"step": 2050
},
{
"epoch": 16.48,
"learning_rate": 3.643478260869565e-05,
"loss": 0.9809,
"step": 2060
},
{
"epoch": 16.56,
"learning_rate": 3.634782608695652e-05,
"loss": 0.9175,
"step": 2070
},
{
"epoch": 16.64,
"learning_rate": 3.6260869565217394e-05,
"loss": 0.953,
"step": 2080
},
{
"epoch": 16.72,
"learning_rate": 3.617391304347826e-05,
"loss": 0.9542,
"step": 2090
},
{
"epoch": 16.8,
"learning_rate": 3.6086956521739134e-05,
"loss": 1.0157,
"step": 2100
},
{
"epoch": 16.88,
"learning_rate": 3.6e-05,
"loss": 0.9836,
"step": 2110
},
{
"epoch": 16.96,
"learning_rate": 3.5913043478260875e-05,
"loss": 1.043,
"step": 2120
},
{
"epoch": 17.04,
"learning_rate": 3.582608695652174e-05,
"loss": 0.8193,
"step": 2130
},
{
"epoch": 17.12,
"learning_rate": 3.573913043478261e-05,
"loss": 0.8391,
"step": 2140
},
{
"epoch": 17.2,
"learning_rate": 3.565217391304348e-05,
"loss": 0.8211,
"step": 2150
},
{
"epoch": 17.28,
"learning_rate": 3.556521739130435e-05,
"loss": 0.8447,
"step": 2160
},
{
"epoch": 17.36,
"learning_rate": 3.5478260869565216e-05,
"loss": 0.8472,
"step": 2170
},
{
"epoch": 17.44,
"learning_rate": 3.539130434782609e-05,
"loss": 0.8897,
"step": 2180
},
{
"epoch": 17.52,
"learning_rate": 3.5304347826086956e-05,
"loss": 0.8916,
"step": 2190
},
{
"epoch": 17.6,
"learning_rate": 3.521739130434783e-05,
"loss": 0.9339,
"step": 2200
},
{
"epoch": 17.68,
"learning_rate": 3.51304347826087e-05,
"loss": 0.9641,
"step": 2210
},
{
"epoch": 17.76,
"learning_rate": 3.5043478260869564e-05,
"loss": 0.9489,
"step": 2220
},
{
"epoch": 17.84,
"learning_rate": 3.495652173913044e-05,
"loss": 0.9923,
"step": 2230
},
{
"epoch": 17.92,
"learning_rate": 3.4869565217391304e-05,
"loss": 0.954,
"step": 2240
},
{
"epoch": 18.0,
"learning_rate": 3.478260869565218e-05,
"loss": 0.9964,
"step": 2250
},
{
"epoch": 18.08,
"learning_rate": 3.4695652173913045e-05,
"loss": 0.8709,
"step": 2260
},
{
"epoch": 18.16,
"learning_rate": 3.460869565217392e-05,
"loss": 0.8586,
"step": 2270
},
{
"epoch": 18.24,
"learning_rate": 3.4521739130434785e-05,
"loss": 0.799,
"step": 2280
},
{
"epoch": 18.32,
"learning_rate": 3.443478260869565e-05,
"loss": 0.7715,
"step": 2290
},
{
"epoch": 18.4,
"learning_rate": 3.4347826086956526e-05,
"loss": 0.8854,
"step": 2300
},
{
"epoch": 18.48,
"learning_rate": 3.426086956521739e-05,
"loss": 0.9446,
"step": 2310
},
{
"epoch": 18.56,
"learning_rate": 3.417391304347826e-05,
"loss": 0.892,
"step": 2320
},
{
"epoch": 18.64,
"learning_rate": 3.408695652173913e-05,
"loss": 0.9439,
"step": 2330
},
{
"epoch": 18.72,
"learning_rate": 3.4000000000000007e-05,
"loss": 0.8857,
"step": 2340
},
{
"epoch": 18.8,
"learning_rate": 3.3913043478260867e-05,
"loss": 0.9497,
"step": 2350
},
{
"epoch": 18.88,
"learning_rate": 3.382608695652174e-05,
"loss": 0.9948,
"step": 2360
},
{
"epoch": 18.96,
"learning_rate": 3.3739130434782614e-05,
"loss": 0.9594,
"step": 2370
},
{
"epoch": 19.04,
"learning_rate": 3.365217391304348e-05,
"loss": 0.9142,
"step": 2380
},
{
"epoch": 19.12,
"learning_rate": 3.356521739130435e-05,
"loss": 0.8201,
"step": 2390
},
{
"epoch": 19.2,
"learning_rate": 3.347826086956522e-05,
"loss": 0.8425,
"step": 2400
},
{
"epoch": 19.28,
"learning_rate": 3.339130434782609e-05,
"loss": 0.7843,
"step": 2410
},
{
"epoch": 19.36,
"learning_rate": 3.3304347826086955e-05,
"loss": 0.9149,
"step": 2420
},
{
"epoch": 19.44,
"learning_rate": 3.321739130434783e-05,
"loss": 0.7852,
"step": 2430
},
{
"epoch": 19.52,
"learning_rate": 3.31304347826087e-05,
"loss": 0.8849,
"step": 2440
},
{
"epoch": 19.6,
"learning_rate": 3.304347826086956e-05,
"loss": 0.9418,
"step": 2450
},
{
"epoch": 19.68,
"learning_rate": 3.2956521739130436e-05,
"loss": 0.938,
"step": 2460
},
{
"epoch": 19.76,
"learning_rate": 3.286956521739131e-05,
"loss": 0.9208,
"step": 2470
},
{
"epoch": 19.84,
"learning_rate": 3.278260869565217e-05,
"loss": 0.9258,
"step": 2480
},
{
"epoch": 19.92,
"learning_rate": 3.269565217391304e-05,
"loss": 0.8696,
"step": 2490
},
{
"epoch": 20.0,
"learning_rate": 3.260869565217392e-05,
"loss": 0.9085,
"step": 2500
},
{
"epoch": 20.08,
"learning_rate": 3.2521739130434784e-05,
"loss": 0.8414,
"step": 2510
},
{
"epoch": 20.16,
"learning_rate": 3.243478260869565e-05,
"loss": 0.8509,
"step": 2520
},
{
"epoch": 20.24,
"learning_rate": 3.2347826086956524e-05,
"loss": 0.7594,
"step": 2530
},
{
"epoch": 20.32,
"learning_rate": 3.226086956521739e-05,
"loss": 0.8413,
"step": 2540
},
{
"epoch": 20.4,
"learning_rate": 3.217391304347826e-05,
"loss": 0.8918,
"step": 2550
},
{
"epoch": 20.48,
"learning_rate": 3.208695652173913e-05,
"loss": 0.8074,
"step": 2560
},
{
"epoch": 20.56,
"learning_rate": 3.2000000000000005e-05,
"loss": 0.8728,
"step": 2570
},
{
"epoch": 20.64,
"learning_rate": 3.191304347826087e-05,
"loss": 0.8559,
"step": 2580
},
{
"epoch": 20.72,
"learning_rate": 3.182608695652174e-05,
"loss": 0.8778,
"step": 2590
},
{
"epoch": 20.8,
"learning_rate": 3.173913043478261e-05,
"loss": 0.9155,
"step": 2600
},
{
"epoch": 20.88,
"learning_rate": 3.165217391304348e-05,
"loss": 0.9692,
"step": 2610
},
{
"epoch": 20.96,
"learning_rate": 3.1565217391304346e-05,
"loss": 0.9685,
"step": 2620
},
{
"epoch": 21.04,
"learning_rate": 3.147826086956522e-05,
"loss": 0.9113,
"step": 2630
},
{
"epoch": 21.12,
"learning_rate": 3.139130434782609e-05,
"loss": 0.7723,
"step": 2640
},
{
"epoch": 21.2,
"learning_rate": 3.130434782608696e-05,
"loss": 0.8294,
"step": 2650
},
{
"epoch": 21.28,
"learning_rate": 3.121739130434783e-05,
"loss": 0.7949,
"step": 2660
},
{
"epoch": 21.36,
"learning_rate": 3.1130434782608694e-05,
"loss": 0.8509,
"step": 2670
},
{
"epoch": 21.44,
"learning_rate": 3.104347826086957e-05,
"loss": 0.8727,
"step": 2680
},
{
"epoch": 21.52,
"learning_rate": 3.0956521739130435e-05,
"loss": 0.8094,
"step": 2690
},
{
"epoch": 21.6,
"learning_rate": 3.086956521739131e-05,
"loss": 0.8458,
"step": 2700
},
{
"epoch": 21.68,
"learning_rate": 3.0782608695652175e-05,
"loss": 0.8416,
"step": 2710
},
{
"epoch": 21.76,
"learning_rate": 3.069565217391305e-05,
"loss": 0.8336,
"step": 2720
},
{
"epoch": 21.84,
"learning_rate": 3.0608695652173916e-05,
"loss": 0.9844,
"step": 2730
},
{
"epoch": 21.92,
"learning_rate": 3.052173913043478e-05,
"loss": 0.8814,
"step": 2740
},
{
"epoch": 22.0,
"learning_rate": 3.0434782608695656e-05,
"loss": 0.917,
"step": 2750
},
{
"epoch": 22.08,
"learning_rate": 3.034782608695652e-05,
"loss": 0.8319,
"step": 2760
},
{
"epoch": 22.16,
"learning_rate": 3.0260869565217393e-05,
"loss": 0.7807,
"step": 2770
},
{
"epoch": 22.24,
"learning_rate": 3.0173913043478264e-05,
"loss": 0.8231,
"step": 2780
},
{
"epoch": 22.32,
"learning_rate": 3.008695652173913e-05,
"loss": 0.7917,
"step": 2790
},
{
"epoch": 22.4,
"learning_rate": 3e-05,
"loss": 0.8658,
"step": 2800
},
{
"epoch": 22.48,
"learning_rate": 2.991304347826087e-05,
"loss": 0.8434,
"step": 2810
},
{
"epoch": 22.56,
"learning_rate": 2.982608695652174e-05,
"loss": 0.7944,
"step": 2820
},
{
"epoch": 22.64,
"learning_rate": 2.9739130434782608e-05,
"loss": 0.8425,
"step": 2830
},
{
"epoch": 22.72,
"learning_rate": 2.965217391304348e-05,
"loss": 0.9204,
"step": 2840
},
{
"epoch": 22.8,
"learning_rate": 2.9565217391304352e-05,
"loss": 0.8919,
"step": 2850
},
{
"epoch": 22.88,
"learning_rate": 2.9478260869565215e-05,
"loss": 0.8238,
"step": 2860
},
{
"epoch": 22.96,
"learning_rate": 2.939130434782609e-05,
"loss": 0.8917,
"step": 2870
},
{
"epoch": 23.04,
"learning_rate": 2.930434782608696e-05,
"loss": 0.886,
"step": 2880
},
{
"epoch": 23.12,
"learning_rate": 2.921739130434783e-05,
"loss": 0.8199,
"step": 2890
},
{
"epoch": 23.2,
"learning_rate": 2.9130434782608696e-05,
"loss": 0.7968,
"step": 2900
},
{
"epoch": 23.28,
"learning_rate": 2.9043478260869567e-05,
"loss": 0.8332,
"step": 2910
},
{
"epoch": 23.36,
"learning_rate": 2.8956521739130437e-05,
"loss": 0.8201,
"step": 2920
},
{
"epoch": 23.44,
"learning_rate": 2.8869565217391304e-05,
"loss": 0.8029,
"step": 2930
},
{
"epoch": 23.52,
"learning_rate": 2.8782608695652174e-05,
"loss": 0.7901,
"step": 2940
},
{
"epoch": 23.6,
"learning_rate": 2.8695652173913044e-05,
"loss": 0.8657,
"step": 2950
},
{
"epoch": 23.68,
"learning_rate": 2.8608695652173918e-05,
"loss": 0.8878,
"step": 2960
},
{
"epoch": 23.76,
"learning_rate": 2.852173913043478e-05,
"loss": 0.8956,
"step": 2970
},
{
"epoch": 23.84,
"learning_rate": 2.8434782608695655e-05,
"loss": 0.8472,
"step": 2980
},
{
"epoch": 23.92,
"learning_rate": 2.8347826086956525e-05,
"loss": 0.8308,
"step": 2990
},
{
"epoch": 24.0,
"learning_rate": 2.826086956521739e-05,
"loss": 0.8754,
"step": 3000
},
{
"epoch": 24.08,
"learning_rate": 2.8173913043478262e-05,
"loss": 0.7869,
"step": 3010
},
{
"epoch": 24.16,
"learning_rate": 2.8086956521739133e-05,
"loss": 0.7275,
"step": 3020
},
{
"epoch": 24.24,
"learning_rate": 2.8000000000000003e-05,
"loss": 0.7898,
"step": 3030
},
{
"epoch": 24.32,
"learning_rate": 2.791304347826087e-05,
"loss": 0.781,
"step": 3040
},
{
"epoch": 24.4,
"learning_rate": 2.782608695652174e-05,
"loss": 0.8289,
"step": 3050
},
{
"epoch": 24.48,
"learning_rate": 2.773913043478261e-05,
"loss": 0.8367,
"step": 3060
},
{
"epoch": 24.56,
"learning_rate": 2.7652173913043477e-05,
"loss": 0.8201,
"step": 3070
},
{
"epoch": 24.64,
"learning_rate": 2.7565217391304347e-05,
"loss": 0.8339,
"step": 3080
},
{
"epoch": 24.72,
"learning_rate": 2.747826086956522e-05,
"loss": 0.8857,
"step": 3090
},
{
"epoch": 24.8,
"learning_rate": 2.7391304347826085e-05,
"loss": 0.8688,
"step": 3100
},
{
"epoch": 24.88,
"learning_rate": 2.7304347826086958e-05,
"loss": 0.8792,
"step": 3110
},
{
"epoch": 24.96,
"learning_rate": 2.721739130434783e-05,
"loss": 0.8835,
"step": 3120
},
{
"epoch": 25.04,
"learning_rate": 2.71304347826087e-05,
"loss": 0.8183,
"step": 3130
},
{
"epoch": 25.12,
"learning_rate": 2.7043478260869566e-05,
"loss": 0.772,
"step": 3140
},
{
"epoch": 25.2,
"learning_rate": 2.6956521739130436e-05,
"loss": 0.7913,
"step": 3150
},
{
"epoch": 25.28,
"learning_rate": 2.6869565217391306e-05,
"loss": 0.782,
"step": 3160
},
{
"epoch": 25.36,
"learning_rate": 2.6782608695652173e-05,
"loss": 0.7693,
"step": 3170
},
{
"epoch": 25.44,
"learning_rate": 2.6695652173913043e-05,
"loss": 0.814,
"step": 3180
},
{
"epoch": 25.52,
"learning_rate": 2.6608695652173913e-05,
"loss": 0.771,
"step": 3190
},
{
"epoch": 25.6,
"learning_rate": 2.6521739130434787e-05,
"loss": 0.8702,
"step": 3200
},
{
"epoch": 25.68,
"learning_rate": 2.643478260869565e-05,
"loss": 0.8653,
"step": 3210
},
{
"epoch": 25.76,
"learning_rate": 2.6347826086956524e-05,
"loss": 0.8006,
"step": 3220
},
{
"epoch": 25.84,
"learning_rate": 2.6260869565217394e-05,
"loss": 0.8638,
"step": 3230
},
{
"epoch": 25.92,
"learning_rate": 2.617391304347826e-05,
"loss": 0.8425,
"step": 3240
},
{
"epoch": 26.0,
"learning_rate": 2.608695652173913e-05,
"loss": 0.8959,
"step": 3250
},
{
"epoch": 26.08,
"learning_rate": 2.6000000000000002e-05,
"loss": 0.7395,
"step": 3260
},
{
"epoch": 26.16,
"learning_rate": 2.5913043478260872e-05,
"loss": 0.7737,
"step": 3270
},
{
"epoch": 26.24,
"learning_rate": 2.582608695652174e-05,
"loss": 0.7703,
"step": 3280
},
{
"epoch": 26.32,
"learning_rate": 2.573913043478261e-05,
"loss": 0.7901,
"step": 3290
},
{
"epoch": 26.4,
"learning_rate": 2.5652173913043483e-05,
"loss": 0.8004,
"step": 3300
},
{
"epoch": 26.48,
"learning_rate": 2.5565217391304346e-05,
"loss": 0.8087,
"step": 3310
},
{
"epoch": 26.56,
"learning_rate": 2.5478260869565217e-05,
"loss": 0.8025,
"step": 3320
},
{
"epoch": 26.64,
"learning_rate": 2.539130434782609e-05,
"loss": 0.8545,
"step": 3330
},
{
"epoch": 26.72,
"learning_rate": 2.530434782608696e-05,
"loss": 0.8215,
"step": 3340
},
{
"epoch": 26.8,
"learning_rate": 2.5217391304347827e-05,
"loss": 0.7817,
"step": 3350
},
{
"epoch": 26.88,
"learning_rate": 2.5130434782608698e-05,
"loss": 0.8443,
"step": 3360
},
{
"epoch": 26.96,
"learning_rate": 2.5043478260869568e-05,
"loss": 0.8381,
"step": 3370
},
{
"epoch": 27.04,
"learning_rate": 2.4956521739130438e-05,
"loss": 0.7931,
"step": 3380
},
{
"epoch": 27.12,
"learning_rate": 2.4869565217391305e-05,
"loss": 0.7931,
"step": 3390
},
{
"epoch": 27.2,
"learning_rate": 2.4782608695652175e-05,
"loss": 0.729,
"step": 3400
},
{
"epoch": 27.28,
"learning_rate": 2.4695652173913045e-05,
"loss": 0.8028,
"step": 3410
},
{
"epoch": 27.36,
"learning_rate": 2.4608695652173912e-05,
"loss": 0.7529,
"step": 3420
},
{
"epoch": 27.44,
"learning_rate": 2.4521739130434786e-05,
"loss": 0.825,
"step": 3430
},
{
"epoch": 27.52,
"learning_rate": 2.4434782608695653e-05,
"loss": 0.785,
"step": 3440
},
{
"epoch": 27.6,
"learning_rate": 2.4347826086956523e-05,
"loss": 0.8408,
"step": 3450
},
{
"epoch": 27.68,
"learning_rate": 2.4260869565217393e-05,
"loss": 0.8812,
"step": 3460
},
{
"epoch": 27.76,
"learning_rate": 2.417391304347826e-05,
"loss": 0.8036,
"step": 3470
},
{
"epoch": 27.84,
"learning_rate": 2.408695652173913e-05,
"loss": 0.8844,
"step": 3480
},
{
"epoch": 27.92,
"learning_rate": 2.4e-05,
"loss": 0.8191,
"step": 3490
},
{
"epoch": 28.0,
"learning_rate": 2.391304347826087e-05,
"loss": 0.8419,
"step": 3500
},
{
"epoch": 28.08,
"learning_rate": 2.382608695652174e-05,
"loss": 0.7546,
"step": 3510
},
{
"epoch": 28.16,
"learning_rate": 2.373913043478261e-05,
"loss": 0.74,
"step": 3520
},
{
"epoch": 28.24,
"learning_rate": 2.3652173913043478e-05,
"loss": 0.7936,
"step": 3530
},
{
"epoch": 28.32,
"learning_rate": 2.356521739130435e-05,
"loss": 0.7558,
"step": 3540
},
{
"epoch": 28.4,
"learning_rate": 2.347826086956522e-05,
"loss": 0.8113,
"step": 3550
},
{
"epoch": 28.48,
"learning_rate": 2.339130434782609e-05,
"loss": 0.793,
"step": 3560
},
{
"epoch": 28.56,
"learning_rate": 2.330434782608696e-05,
"loss": 0.7783,
"step": 3570
},
{
"epoch": 28.64,
"learning_rate": 2.3217391304347826e-05,
"loss": 0.825,
"step": 3580
},
{
"epoch": 28.72,
"learning_rate": 2.31304347826087e-05,
"loss": 0.7734,
"step": 3590
},
{
"epoch": 28.8,
"learning_rate": 2.3043478260869567e-05,
"loss": 0.7998,
"step": 3600
},
{
"epoch": 28.88,
"learning_rate": 2.2956521739130433e-05,
"loss": 0.8553,
"step": 3610
},
{
"epoch": 28.96,
"learning_rate": 2.2869565217391307e-05,
"loss": 0.8441,
"step": 3620
},
{
"epoch": 29.04,
"learning_rate": 2.2782608695652174e-05,
"loss": 0.7547,
"step": 3630
},
{
"epoch": 29.12,
"learning_rate": 2.2695652173913044e-05,
"loss": 0.7616,
"step": 3640
},
{
"epoch": 29.2,
"learning_rate": 2.2608695652173914e-05,
"loss": 0.7356,
"step": 3650
},
{
"epoch": 29.28,
"learning_rate": 2.252173913043478e-05,
"loss": 0.739,
"step": 3660
},
{
"epoch": 29.36,
"learning_rate": 2.2434782608695655e-05,
"loss": 0.7591,
"step": 3670
},
{
"epoch": 29.44,
"learning_rate": 2.2347826086956522e-05,
"loss": 0.7885,
"step": 3680
},
{
"epoch": 29.52,
"learning_rate": 2.2260869565217392e-05,
"loss": 0.8358,
"step": 3690
},
{
"epoch": 29.6,
"learning_rate": 2.2173913043478262e-05,
"loss": 0.8253,
"step": 3700
},
{
"epoch": 29.68,
"learning_rate": 2.2086956521739133e-05,
"loss": 0.8004,
"step": 3710
},
{
"epoch": 29.76,
"learning_rate": 2.2000000000000003e-05,
"loss": 0.7941,
"step": 3720
},
{
"epoch": 29.84,
"learning_rate": 2.191304347826087e-05,
"loss": 0.7904,
"step": 3730
},
{
"epoch": 29.92,
"learning_rate": 2.182608695652174e-05,
"loss": 0.8109,
"step": 3740
},
{
"epoch": 30.0,
"learning_rate": 2.173913043478261e-05,
"loss": 0.8528,
"step": 3750
},
{
"epoch": 30.08,
"learning_rate": 2.165217391304348e-05,
"loss": 0.7457,
"step": 3760
},
{
"epoch": 30.16,
"learning_rate": 2.1565217391304347e-05,
"loss": 0.755,
"step": 3770
},
{
"epoch": 30.24,
"learning_rate": 2.1478260869565218e-05,
"loss": 0.7449,
"step": 3780
},
{
"epoch": 30.32,
"learning_rate": 2.1391304347826088e-05,
"loss": 0.7583,
"step": 3790
},
{
"epoch": 30.4,
"learning_rate": 2.1304347826086958e-05,
"loss": 0.7352,
"step": 3800
},
{
"epoch": 30.48,
"learning_rate": 2.1217391304347828e-05,
"loss": 0.7401,
"step": 3810
},
{
"epoch": 30.56,
"learning_rate": 2.1130434782608695e-05,
"loss": 0.7649,
"step": 3820
},
{
"epoch": 30.64,
"learning_rate": 2.104347826086957e-05,
"loss": 0.8294,
"step": 3830
},
{
"epoch": 30.72,
"learning_rate": 2.0956521739130436e-05,
"loss": 0.7539,
"step": 3840
},
{
"epoch": 30.8,
"learning_rate": 2.0869565217391303e-05,
"loss": 0.8142,
"step": 3850
},
{
"epoch": 30.88,
"learning_rate": 2.0782608695652176e-05,
"loss": 0.875,
"step": 3860
},
{
"epoch": 30.96,
"learning_rate": 2.0695652173913043e-05,
"loss": 0.7937,
"step": 3870
},
{
"epoch": 31.04,
"learning_rate": 2.0608695652173913e-05,
"loss": 0.7895,
"step": 3880
},
{
"epoch": 31.12,
"learning_rate": 2.0521739130434784e-05,
"loss": 0.7183,
"step": 3890
},
{
"epoch": 31.2,
"learning_rate": 2.0434782608695654e-05,
"loss": 0.7453,
"step": 3900
},
{
"epoch": 31.28,
"learning_rate": 2.0347826086956524e-05,
"loss": 0.741,
"step": 3910
},
{
"epoch": 31.36,
"learning_rate": 2.026086956521739e-05,
"loss": 0.7427,
"step": 3920
},
{
"epoch": 31.44,
"learning_rate": 2.017391304347826e-05,
"loss": 0.7211,
"step": 3930
},
{
"epoch": 31.52,
"learning_rate": 2.008695652173913e-05,
"loss": 0.7671,
"step": 3940
},
{
"epoch": 31.6,
"learning_rate": 2e-05,
"loss": 0.7455,
"step": 3950
},
{
"epoch": 31.68,
"learning_rate": 1.9913043478260872e-05,
"loss": 0.7925,
"step": 3960
},
{
"epoch": 31.76,
"learning_rate": 1.982608695652174e-05,
"loss": 0.8201,
"step": 3970
},
{
"epoch": 31.84,
"learning_rate": 1.973913043478261e-05,
"loss": 0.8209,
"step": 3980
},
{
"epoch": 31.92,
"learning_rate": 1.965217391304348e-05,
"loss": 0.8644,
"step": 3990
},
{
"epoch": 32.0,
"learning_rate": 1.956521739130435e-05,
"loss": 0.7728,
"step": 4000
},
{
"epoch": 32.08,
"learning_rate": 1.9478260869565216e-05,
"loss": 0.7284,
"step": 4010
},
{
"epoch": 32.16,
"learning_rate": 1.939130434782609e-05,
"loss": 0.7171,
"step": 4020
},
{
"epoch": 32.24,
"learning_rate": 1.9304347826086957e-05,
"loss": 0.7435,
"step": 4030
},
{
"epoch": 32.32,
"learning_rate": 1.9217391304347827e-05,
"loss": 0.736,
"step": 4040
},
{
"epoch": 32.4,
"learning_rate": 1.9130434782608697e-05,
"loss": 0.7511,
"step": 4050
},
{
"epoch": 32.48,
"learning_rate": 1.9043478260869564e-05,
"loss": 0.8181,
"step": 4060
},
{
"epoch": 32.56,
"learning_rate": 1.8956521739130438e-05,
"loss": 0.7707,
"step": 4070
},
{
"epoch": 32.64,
"learning_rate": 1.8869565217391305e-05,
"loss": 0.7859,
"step": 4080
},
{
"epoch": 32.72,
"learning_rate": 1.8782608695652175e-05,
"loss": 0.7545,
"step": 4090
},
{
"epoch": 32.8,
"learning_rate": 1.8695652173913045e-05,
"loss": 0.773,
"step": 4100
},
{
"epoch": 32.88,
"learning_rate": 1.8608695652173912e-05,
"loss": 0.8116,
"step": 4110
},
{
"epoch": 32.96,
"learning_rate": 1.8521739130434786e-05,
"loss": 0.8306,
"step": 4120
},
{
"epoch": 33.04,
"learning_rate": 1.8434782608695653e-05,
"loss": 0.715,
"step": 4130
},
{
"epoch": 33.12,
"learning_rate": 1.8347826086956523e-05,
"loss": 0.7313,
"step": 4140
},
{
"epoch": 33.2,
"learning_rate": 1.8260869565217393e-05,
"loss": 0.7405,
"step": 4150
},
{
"epoch": 33.28,
"learning_rate": 1.817391304347826e-05,
"loss": 0.7378,
"step": 4160
},
{
"epoch": 33.36,
"learning_rate": 1.808695652173913e-05,
"loss": 0.7845,
"step": 4170
},
{
"epoch": 33.44,
"learning_rate": 1.8e-05,
"loss": 0.7204,
"step": 4180
},
{
"epoch": 33.52,
"learning_rate": 1.791304347826087e-05,
"loss": 0.7728,
"step": 4190
},
{
"epoch": 33.6,
"learning_rate": 1.782608695652174e-05,
"loss": 0.8212,
"step": 4200
},
{
"epoch": 33.68,
"learning_rate": 1.7739130434782608e-05,
"loss": 0.7131,
"step": 4210
},
{
"epoch": 33.76,
"learning_rate": 1.7652173913043478e-05,
"loss": 0.7555,
"step": 4220
},
{
"epoch": 33.84,
"learning_rate": 1.756521739130435e-05,
"loss": 0.8008,
"step": 4230
},
{
"epoch": 33.92,
"learning_rate": 1.747826086956522e-05,
"loss": 0.8415,
"step": 4240
},
{
"epoch": 34.0,
"learning_rate": 1.739130434782609e-05,
"loss": 0.7872,
"step": 4250
},
{
"epoch": 34.08,
"learning_rate": 1.730434782608696e-05,
"loss": 0.7312,
"step": 4260
},
{
"epoch": 34.16,
"learning_rate": 1.7217391304347826e-05,
"loss": 0.7106,
"step": 4270
},
{
"epoch": 34.24,
"learning_rate": 1.7130434782608696e-05,
"loss": 0.6981,
"step": 4280
},
{
"epoch": 34.32,
"learning_rate": 1.7043478260869566e-05,
"loss": 0.6903,
"step": 4290
},
{
"epoch": 34.4,
"learning_rate": 1.6956521739130433e-05,
"loss": 0.7726,
"step": 4300
},
{
"epoch": 34.48,
"learning_rate": 1.6869565217391307e-05,
"loss": 0.776,
"step": 4310
},
{
"epoch": 34.56,
"learning_rate": 1.6782608695652174e-05,
"loss": 0.7745,
"step": 4320
},
{
"epoch": 34.64,
"learning_rate": 1.6695652173913044e-05,
"loss": 0.7843,
"step": 4330
},
{
"epoch": 34.72,
"learning_rate": 1.6608695652173914e-05,
"loss": 0.7705,
"step": 4340
},
{
"epoch": 34.8,
"learning_rate": 1.652173913043478e-05,
"loss": 0.8103,
"step": 4350
},
{
"epoch": 34.88,
"learning_rate": 1.6434782608695655e-05,
"loss": 0.793,
"step": 4360
},
{
"epoch": 34.96,
"learning_rate": 1.634782608695652e-05,
"loss": 0.7965,
"step": 4370
},
{
"epoch": 35.04,
"learning_rate": 1.6260869565217392e-05,
"loss": 0.7574,
"step": 4380
},
{
"epoch": 35.12,
"learning_rate": 1.6173913043478262e-05,
"loss": 0.6871,
"step": 4390
},
{
"epoch": 35.2,
"learning_rate": 1.608695652173913e-05,
"loss": 0.7258,
"step": 4400
},
{
"epoch": 35.28,
"learning_rate": 1.6000000000000003e-05,
"loss": 0.7213,
"step": 4410
},
{
"epoch": 35.36,
"learning_rate": 1.591304347826087e-05,
"loss": 0.704,
"step": 4420
},
{
"epoch": 35.44,
"learning_rate": 1.582608695652174e-05,
"loss": 0.7534,
"step": 4430
},
{
"epoch": 35.52,
"learning_rate": 1.573913043478261e-05,
"loss": 0.7666,
"step": 4440
},
{
"epoch": 35.6,
"learning_rate": 1.565217391304348e-05,
"loss": 0.7837,
"step": 4450
},
{
"epoch": 35.68,
"learning_rate": 1.5565217391304347e-05,
"loss": 0.7544,
"step": 4460
},
{
"epoch": 35.76,
"learning_rate": 1.5478260869565217e-05,
"loss": 0.7645,
"step": 4470
},
{
"epoch": 35.84,
"learning_rate": 1.5391304347826088e-05,
"loss": 0.7887,
"step": 4480
},
{
"epoch": 35.92,
"learning_rate": 1.5304347826086958e-05,
"loss": 0.7201,
"step": 4490
},
{
"epoch": 36.0,
"learning_rate": 1.5217391304347828e-05,
"loss": 0.8047,
"step": 4500
},
{
"epoch": 36.08,
"learning_rate": 1.5130434782608697e-05,
"loss": 0.682,
"step": 4510
},
{
"epoch": 36.16,
"learning_rate": 1.5043478260869565e-05,
"loss": 0.7343,
"step": 4520
},
{
"epoch": 36.24,
"learning_rate": 1.4956521739130436e-05,
"loss": 0.7002,
"step": 4530
},
{
"epoch": 36.32,
"learning_rate": 1.4869565217391304e-05,
"loss": 0.7418,
"step": 4540
},
{
"epoch": 36.4,
"learning_rate": 1.4782608695652176e-05,
"loss": 0.7528,
"step": 4550
},
{
"epoch": 36.48,
"learning_rate": 1.4695652173913045e-05,
"loss": 0.7683,
"step": 4560
},
{
"epoch": 36.56,
"learning_rate": 1.4608695652173915e-05,
"loss": 0.7687,
"step": 4570
},
{
"epoch": 36.64,
"learning_rate": 1.4521739130434783e-05,
"loss": 0.7404,
"step": 4580
},
{
"epoch": 36.72,
"learning_rate": 1.4434782608695652e-05,
"loss": 0.7402,
"step": 4590
},
{
"epoch": 36.8,
"learning_rate": 1.4347826086956522e-05,
"loss": 0.7542,
"step": 4600
},
{
"epoch": 36.88,
"learning_rate": 1.426086956521739e-05,
"loss": 0.8203,
"step": 4610
},
{
"epoch": 36.96,
"learning_rate": 1.4173913043478263e-05,
"loss": 0.7896,
"step": 4620
},
{
"epoch": 37.04,
"learning_rate": 1.4086956521739131e-05,
"loss": 0.7888,
"step": 4630
},
{
"epoch": 37.12,
"learning_rate": 1.4000000000000001e-05,
"loss": 0.691,
"step": 4640
},
{
"epoch": 37.2,
"learning_rate": 1.391304347826087e-05,
"loss": 0.7158,
"step": 4650
},
{
"epoch": 37.28,
"learning_rate": 1.3826086956521739e-05,
"loss": 0.6983,
"step": 4660
},
{
"epoch": 37.36,
"learning_rate": 1.373913043478261e-05,
"loss": 0.7424,
"step": 4670
},
{
"epoch": 37.44,
"learning_rate": 1.3652173913043479e-05,
"loss": 0.7273,
"step": 4680
},
{
"epoch": 37.52,
"learning_rate": 1.356521739130435e-05,
"loss": 0.7105,
"step": 4690
},
{
"epoch": 37.6,
"learning_rate": 1.3478260869565218e-05,
"loss": 0.765,
"step": 4700
},
{
"epoch": 37.68,
"learning_rate": 1.3391304347826086e-05,
"loss": 0.7455,
"step": 4710
},
{
"epoch": 37.76,
"learning_rate": 1.3304347826086957e-05,
"loss": 0.7378,
"step": 4720
},
{
"epoch": 37.84,
"learning_rate": 1.3217391304347825e-05,
"loss": 0.7797,
"step": 4730
},
{
"epoch": 37.92,
"learning_rate": 1.3130434782608697e-05,
"loss": 0.7881,
"step": 4740
},
{
"epoch": 38.0,
"learning_rate": 1.3043478260869566e-05,
"loss": 0.7675,
"step": 4750
},
{
"epoch": 38.08,
"learning_rate": 1.2956521739130436e-05,
"loss": 0.7251,
"step": 4760
},
{
"epoch": 38.16,
"learning_rate": 1.2869565217391305e-05,
"loss": 0.736,
"step": 4770
},
{
"epoch": 38.24,
"learning_rate": 1.2782608695652173e-05,
"loss": 0.6723,
"step": 4780
},
{
"epoch": 38.32,
"learning_rate": 1.2695652173913045e-05,
"loss": 0.736,
"step": 4790
},
{
"epoch": 38.4,
"learning_rate": 1.2608695652173914e-05,
"loss": 0.7543,
"step": 4800
},
{
"epoch": 38.48,
"learning_rate": 1.2521739130434784e-05,
"loss": 0.708,
"step": 4810
},
{
"epoch": 38.56,
"learning_rate": 1.2434782608695652e-05,
"loss": 0.7561,
"step": 4820
},
{
"epoch": 38.64,
"learning_rate": 1.2347826086956523e-05,
"loss": 0.7343,
"step": 4830
},
{
"epoch": 38.72,
"learning_rate": 1.2260869565217393e-05,
"loss": 0.748,
"step": 4840
},
{
"epoch": 38.8,
"learning_rate": 1.2173913043478261e-05,
"loss": 0.7882,
"step": 4850
},
{
"epoch": 38.88,
"learning_rate": 1.208695652173913e-05,
"loss": 0.7107,
"step": 4860
},
{
"epoch": 38.96,
"learning_rate": 1.2e-05,
"loss": 0.7572,
"step": 4870
},
{
"epoch": 39.04,
"learning_rate": 1.191304347826087e-05,
"loss": 0.7749,
"step": 4880
},
{
"epoch": 39.12,
"learning_rate": 1.1826086956521739e-05,
"loss": 0.7238,
"step": 4890
},
{
"epoch": 39.2,
"learning_rate": 1.173913043478261e-05,
"loss": 0.7204,
"step": 4900
},
{
"epoch": 39.28,
"learning_rate": 1.165217391304348e-05,
"loss": 0.7088,
"step": 4910
},
{
"epoch": 39.36,
"learning_rate": 1.156521739130435e-05,
"loss": 0.7424,
"step": 4920
},
{
"epoch": 39.44,
"learning_rate": 1.1478260869565217e-05,
"loss": 0.7483,
"step": 4930
},
{
"epoch": 39.52,
"learning_rate": 1.1391304347826087e-05,
"loss": 0.7048,
"step": 4940
},
{
"epoch": 39.6,
"learning_rate": 1.1304347826086957e-05,
"loss": 0.7583,
"step": 4950
},
{
"epoch": 39.68,
"learning_rate": 1.1217391304347827e-05,
"loss": 0.6905,
"step": 4960
},
{
"epoch": 39.76,
"learning_rate": 1.1130434782608696e-05,
"loss": 0.6934,
"step": 4970
},
{
"epoch": 39.84,
"learning_rate": 1.1043478260869566e-05,
"loss": 0.7766,
"step": 4980
},
{
"epoch": 39.92,
"learning_rate": 1.0956521739130435e-05,
"loss": 0.7716,
"step": 4990
},
{
"epoch": 40.0,
"learning_rate": 1.0869565217391305e-05,
"loss": 0.7482,
"step": 5000
},
{
"epoch": 40.08,
"learning_rate": 1.0782608695652174e-05,
"loss": 0.6505,
"step": 5010
},
{
"epoch": 40.16,
"learning_rate": 1.0695652173913044e-05,
"loss": 0.6988,
"step": 5020
},
{
"epoch": 40.24,
"learning_rate": 1.0608695652173914e-05,
"loss": 0.7083,
"step": 5030
},
{
"epoch": 40.32,
"learning_rate": 1.0521739130434784e-05,
"loss": 0.7353,
"step": 5040
},
{
"epoch": 40.4,
"learning_rate": 1.0434782608695651e-05,
"loss": 0.7568,
"step": 5050
},
{
"epoch": 40.48,
"learning_rate": 1.0347826086956522e-05,
"loss": 0.7481,
"step": 5060
},
{
"epoch": 40.56,
"learning_rate": 1.0260869565217392e-05,
"loss": 0.7494,
"step": 5070
},
{
"epoch": 40.64,
"learning_rate": 1.0173913043478262e-05,
"loss": 0.7494,
"step": 5080
},
{
"epoch": 40.72,
"learning_rate": 1.008695652173913e-05,
"loss": 0.7494,
"step": 5090
},
{
"epoch": 40.8,
"learning_rate": 1e-05,
"loss": 0.7479,
"step": 5100
},
{
"epoch": 40.88,
"learning_rate": 9.91304347826087e-06,
"loss": 0.7476,
"step": 5110
},
{
"epoch": 40.96,
"learning_rate": 9.82608695652174e-06,
"loss": 0.7544,
"step": 5120
},
{
"epoch": 41.04,
"learning_rate": 9.739130434782608e-06,
"loss": 0.6957,
"step": 5130
},
{
"epoch": 41.12,
"learning_rate": 9.652173913043478e-06,
"loss": 0.7071,
"step": 5140
},
{
"epoch": 41.2,
"learning_rate": 9.565217391304349e-06,
"loss": 0.7036,
"step": 5150
},
{
"epoch": 41.28,
"learning_rate": 9.478260869565219e-06,
"loss": 0.6897,
"step": 5160
},
{
"epoch": 41.36,
"learning_rate": 9.391304347826087e-06,
"loss": 0.7168,
"step": 5170
},
{
"epoch": 41.44,
"learning_rate": 9.304347826086956e-06,
"loss": 0.6823,
"step": 5180
},
{
"epoch": 41.52,
"learning_rate": 9.217391304347826e-06,
"loss": 0.7238,
"step": 5190
},
{
"epoch": 41.6,
"learning_rate": 9.130434782608697e-06,
"loss": 0.7366,
"step": 5200
},
{
"epoch": 41.68,
"learning_rate": 9.043478260869565e-06,
"loss": 0.7688,
"step": 5210
},
{
"epoch": 41.76,
"learning_rate": 8.956521739130435e-06,
"loss": 0.7606,
"step": 5220
},
{
"epoch": 41.84,
"learning_rate": 8.869565217391304e-06,
"loss": 0.7199,
"step": 5230
},
{
"epoch": 41.92,
"learning_rate": 8.782608695652174e-06,
"loss": 0.7336,
"step": 5240
},
{
"epoch": 42.0,
"learning_rate": 8.695652173913044e-06,
"loss": 0.7061,
"step": 5250
},
{
"epoch": 42.08,
"learning_rate": 8.608695652173913e-06,
"loss": 0.7108,
"step": 5260
},
{
"epoch": 42.16,
"learning_rate": 8.521739130434783e-06,
"loss": 0.6611,
"step": 5270
},
{
"epoch": 42.24,
"learning_rate": 8.434782608695653e-06,
"loss": 0.7001,
"step": 5280
},
{
"epoch": 42.32,
"learning_rate": 8.347826086956522e-06,
"loss": 0.7304,
"step": 5290
},
{
"epoch": 42.4,
"learning_rate": 8.26086956521739e-06,
"loss": 0.7143,
"step": 5300
},
{
"epoch": 42.48,
"learning_rate": 8.17391304347826e-06,
"loss": 0.6987,
"step": 5310
},
{
"epoch": 42.56,
"learning_rate": 8.086956521739131e-06,
"loss": 0.7078,
"step": 5320
},
{
"epoch": 42.64,
"learning_rate": 8.000000000000001e-06,
"loss": 0.7553,
"step": 5330
},
{
"epoch": 42.72,
"learning_rate": 7.91304347826087e-06,
"loss": 0.7014,
"step": 5340
},
{
"epoch": 42.8,
"learning_rate": 7.82608695652174e-06,
"loss": 0.7483,
"step": 5350
},
{
"epoch": 42.88,
"learning_rate": 7.739130434782609e-06,
"loss": 0.7629,
"step": 5360
},
{
"epoch": 42.96,
"learning_rate": 7.652173913043479e-06,
"loss": 0.7117,
"step": 5370
},
{
"epoch": 43.04,
"learning_rate": 7.565217391304348e-06,
"loss": 0.7447,
"step": 5380
},
{
"epoch": 43.12,
"learning_rate": 7.478260869565218e-06,
"loss": 0.6614,
"step": 5390
},
{
"epoch": 43.2,
"learning_rate": 7.391304347826088e-06,
"loss": 0.7417,
"step": 5400
},
{
"epoch": 43.28,
"learning_rate": 7.304347826086957e-06,
"loss": 0.7217,
"step": 5410
},
{
"epoch": 43.36,
"learning_rate": 7.217391304347826e-06,
"loss": 0.704,
"step": 5420
},
{
"epoch": 43.44,
"learning_rate": 7.130434782608695e-06,
"loss": 0.704,
"step": 5430
},
{
"epoch": 43.52,
"learning_rate": 7.043478260869566e-06,
"loss": 0.7896,
"step": 5440
},
{
"epoch": 43.6,
"learning_rate": 6.956521739130435e-06,
"loss": 0.7481,
"step": 5450
},
{
"epoch": 43.68,
"learning_rate": 6.869565217391305e-06,
"loss": 0.7105,
"step": 5460
},
{
"epoch": 43.76,
"learning_rate": 6.782608695652175e-06,
"loss": 0.6717,
"step": 5470
},
{
"epoch": 43.84,
"learning_rate": 6.695652173913043e-06,
"loss": 0.6871,
"step": 5480
},
{
"epoch": 43.92,
"learning_rate": 6.608695652173913e-06,
"loss": 0.7229,
"step": 5490
},
{
"epoch": 44.0,
"learning_rate": 6.521739130434783e-06,
"loss": 0.7324,
"step": 5500
},
{
"epoch": 44.08,
"learning_rate": 6.434782608695652e-06,
"loss": 0.6597,
"step": 5510
},
{
"epoch": 44.16,
"learning_rate": 6.3478260869565225e-06,
"loss": 0.6771,
"step": 5520
},
{
"epoch": 44.24,
"learning_rate": 6.260869565217392e-06,
"loss": 0.7184,
"step": 5530
},
{
"epoch": 44.32,
"learning_rate": 6.173913043478261e-06,
"loss": 0.6816,
"step": 5540
},
{
"epoch": 44.4,
"learning_rate": 6.086956521739131e-06,
"loss": 0.7171,
"step": 5550
},
{
"epoch": 44.48,
"learning_rate": 6e-06,
"loss": 0.6723,
"step": 5560
},
{
"epoch": 44.56,
"learning_rate": 5.9130434782608696e-06,
"loss": 0.6991,
"step": 5570
},
{
"epoch": 44.64,
"learning_rate": 5.82608695652174e-06,
"loss": 0.7538,
"step": 5580
},
{
"epoch": 44.72,
"learning_rate": 5.739130434782608e-06,
"loss": 0.7405,
"step": 5590
},
{
"epoch": 44.8,
"learning_rate": 5.652173913043479e-06,
"loss": 0.6698,
"step": 5600
},
{
"epoch": 44.88,
"learning_rate": 5.565217391304348e-06,
"loss": 0.7276,
"step": 5610
},
{
"epoch": 44.96,
"learning_rate": 5.478260869565217e-06,
"loss": 0.7446,
"step": 5620
},
{
"epoch": 45.04,
"learning_rate": 5.391304347826087e-06,
"loss": 0.7269,
"step": 5630
},
{
"epoch": 45.12,
"learning_rate": 5.304347826086957e-06,
"loss": 0.6781,
"step": 5640
},
{
"epoch": 45.2,
"learning_rate": 5.217391304347826e-06,
"loss": 0.7352,
"step": 5650
},
{
"epoch": 45.28,
"learning_rate": 5.130434782608696e-06,
"loss": 0.7599,
"step": 5660
},
{
"epoch": 45.36,
"learning_rate": 5.043478260869565e-06,
"loss": 0.7081,
"step": 5670
},
{
"epoch": 45.44,
"learning_rate": 4.956521739130435e-06,
"loss": 0.7287,
"step": 5680
},
{
"epoch": 45.52,
"learning_rate": 4.869565217391304e-06,
"loss": 0.6988,
"step": 5690
},
{
"epoch": 45.6,
"learning_rate": 4.782608695652174e-06,
"loss": 0.716,
"step": 5700
},
{
"epoch": 45.68,
"learning_rate": 4.695652173913044e-06,
"loss": 0.6744,
"step": 5710
},
{
"epoch": 45.76,
"learning_rate": 4.608695652173913e-06,
"loss": 0.6928,
"step": 5720
},
{
"epoch": 45.84,
"learning_rate": 4.5217391304347826e-06,
"loss": 0.721,
"step": 5730
},
{
"epoch": 45.92,
"learning_rate": 4.434782608695652e-06,
"loss": 0.7276,
"step": 5740
},
{
"epoch": 46.0,
"learning_rate": 4.347826086956522e-06,
"loss": 0.7224,
"step": 5750
},
{
"epoch": 46.08,
"learning_rate": 4.260869565217392e-06,
"loss": 0.7101,
"step": 5760
},
{
"epoch": 46.16,
"learning_rate": 4.173913043478261e-06,
"loss": 0.6956,
"step": 5770
},
{
"epoch": 46.24,
"learning_rate": 4.08695652173913e-06,
"loss": 0.734,
"step": 5780
},
{
"epoch": 46.32,
"learning_rate": 4.000000000000001e-06,
"loss": 0.6655,
"step": 5790
},
{
"epoch": 46.4,
"learning_rate": 3.91304347826087e-06,
"loss": 0.6528,
"step": 5800
},
{
"epoch": 46.48,
"learning_rate": 3.8260869565217395e-06,
"loss": 0.6876,
"step": 5810
},
{
"epoch": 46.56,
"learning_rate": 3.739130434782609e-06,
"loss": 0.6792,
"step": 5820
},
{
"epoch": 46.64,
"learning_rate": 3.6521739130434787e-06,
"loss": 0.7622,
"step": 5830
},
{
"epoch": 46.72,
"learning_rate": 3.5652173913043477e-06,
"loss": 0.6852,
"step": 5840
},
{
"epoch": 46.8,
"learning_rate": 3.4782608695652175e-06,
"loss": 0.6863,
"step": 5850
},
{
"epoch": 46.88,
"learning_rate": 3.3913043478260873e-06,
"loss": 0.7497,
"step": 5860
},
{
"epoch": 46.96,
"learning_rate": 3.3043478260869563e-06,
"loss": 0.7246,
"step": 5870
},
{
"epoch": 47.04,
"learning_rate": 3.217391304347826e-06,
"loss": 0.7235,
"step": 5880
},
{
"epoch": 47.12,
"learning_rate": 3.130434782608696e-06,
"loss": 0.7068,
"step": 5890
},
{
"epoch": 47.2,
"learning_rate": 3.0434782608695654e-06,
"loss": 0.7105,
"step": 5900
},
{
"epoch": 47.28,
"learning_rate": 2.9565217391304348e-06,
"loss": 0.6939,
"step": 5910
},
{
"epoch": 47.36,
"learning_rate": 2.869565217391304e-06,
"loss": 0.6867,
"step": 5920
},
{
"epoch": 47.44,
"learning_rate": 2.782608695652174e-06,
"loss": 0.6587,
"step": 5930
},
{
"epoch": 47.52,
"learning_rate": 2.6956521739130434e-06,
"loss": 0.6914,
"step": 5940
},
{
"epoch": 47.6,
"learning_rate": 2.608695652173913e-06,
"loss": 0.6501,
"step": 5950
},
{
"epoch": 47.68,
"learning_rate": 2.5217391304347826e-06,
"loss": 0.7491,
"step": 5960
},
{
"epoch": 47.76,
"learning_rate": 2.434782608695652e-06,
"loss": 0.7428,
"step": 5970
},
{
"epoch": 47.84,
"learning_rate": 2.347826086956522e-06,
"loss": 0.716,
"step": 5980
},
{
"epoch": 47.92,
"learning_rate": 2.2608695652173913e-06,
"loss": 0.6706,
"step": 5990
},
{
"epoch": 48.0,
"learning_rate": 2.173913043478261e-06,
"loss": 0.6691,
"step": 6000
}
],
"max_steps": 6250,
"num_train_epochs": 50,
"total_flos": 8948999081401344.0,
"trial_name": null,
"trial_params": null
}