Cody Jiang
Add files via upload
9d8a58c unverified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.9399674144629653,
"global_step": 7500,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 1.0000000000000002e-06,
"loss": 0.7141,
"step": 10
},
{
"epoch": 0.0,
"learning_rate": 2.0000000000000003e-06,
"loss": 0.7055,
"step": 20
},
{
"epoch": 0.0,
"learning_rate": 3e-06,
"loss": 0.6885,
"step": 30
},
{
"epoch": 0.01,
"learning_rate": 4.000000000000001e-06,
"loss": 0.6619,
"step": 40
},
{
"epoch": 0.01,
"learning_rate": 5e-06,
"loss": 0.6147,
"step": 50
},
{
"epoch": 0.01,
"learning_rate": 6e-06,
"loss": 0.5332,
"step": 60
},
{
"epoch": 0.01,
"learning_rate": 7.000000000000001e-06,
"loss": 0.4607,
"step": 70
},
{
"epoch": 0.01,
"learning_rate": 8.000000000000001e-06,
"loss": 0.3797,
"step": 80
},
{
"epoch": 0.01,
"learning_rate": 9e-06,
"loss": 0.3388,
"step": 90
},
{
"epoch": 0.01,
"learning_rate": 1e-05,
"loss": 0.27,
"step": 100
},
{
"epoch": 0.01,
"learning_rate": 1.1000000000000001e-05,
"loss": 0.2613,
"step": 110
},
{
"epoch": 0.02,
"learning_rate": 1.2e-05,
"loss": 0.2139,
"step": 120
},
{
"epoch": 0.02,
"learning_rate": 1.3000000000000001e-05,
"loss": 0.2052,
"step": 130
},
{
"epoch": 0.02,
"learning_rate": 1.4000000000000001e-05,
"loss": 0.1641,
"step": 140
},
{
"epoch": 0.02,
"learning_rate": 1.5e-05,
"loss": 0.1887,
"step": 150
},
{
"epoch": 0.02,
"learning_rate": 1.6000000000000003e-05,
"loss": 0.1547,
"step": 160
},
{
"epoch": 0.02,
"learning_rate": 1.7000000000000003e-05,
"loss": 0.1562,
"step": 170
},
{
"epoch": 0.02,
"learning_rate": 1.8e-05,
"loss": 0.1266,
"step": 180
},
{
"epoch": 0.02,
"learning_rate": 1.9e-05,
"loss": 0.1668,
"step": 190
},
{
"epoch": 0.03,
"learning_rate": 2e-05,
"loss": 0.15,
"step": 200
},
{
"epoch": 0.03,
"learning_rate": 2.1e-05,
"loss": 0.1001,
"step": 210
},
{
"epoch": 0.03,
"learning_rate": 2.2000000000000003e-05,
"loss": 0.1054,
"step": 220
},
{
"epoch": 0.03,
"learning_rate": 2.3000000000000003e-05,
"loss": 0.0753,
"step": 230
},
{
"epoch": 0.03,
"learning_rate": 2.4e-05,
"loss": 0.0855,
"step": 240
},
{
"epoch": 0.03,
"learning_rate": 2.5e-05,
"loss": 0.0987,
"step": 250
},
{
"epoch": 0.03,
"learning_rate": 2.6000000000000002e-05,
"loss": 0.091,
"step": 260
},
{
"epoch": 0.03,
"learning_rate": 2.7000000000000002e-05,
"loss": 0.0597,
"step": 270
},
{
"epoch": 0.04,
"learning_rate": 2.8000000000000003e-05,
"loss": 0.082,
"step": 280
},
{
"epoch": 0.04,
"learning_rate": 2.9e-05,
"loss": 0.0953,
"step": 290
},
{
"epoch": 0.04,
"learning_rate": 3e-05,
"loss": 0.0678,
"step": 300
},
{
"epoch": 0.04,
"learning_rate": 3.1e-05,
"loss": 0.0692,
"step": 310
},
{
"epoch": 0.04,
"learning_rate": 3.2000000000000005e-05,
"loss": 0.0521,
"step": 320
},
{
"epoch": 0.04,
"learning_rate": 3.3e-05,
"loss": 0.0749,
"step": 330
},
{
"epoch": 0.04,
"learning_rate": 3.4000000000000007e-05,
"loss": 0.0855,
"step": 340
},
{
"epoch": 0.04,
"learning_rate": 3.5e-05,
"loss": 0.0697,
"step": 350
},
{
"epoch": 0.05,
"learning_rate": 3.6e-05,
"loss": 0.0881,
"step": 360
},
{
"epoch": 0.05,
"learning_rate": 3.7e-05,
"loss": 0.0754,
"step": 370
},
{
"epoch": 0.05,
"learning_rate": 3.8e-05,
"loss": 0.0619,
"step": 380
},
{
"epoch": 0.05,
"learning_rate": 3.9000000000000006e-05,
"loss": 0.0509,
"step": 390
},
{
"epoch": 0.05,
"learning_rate": 4e-05,
"loss": 0.0641,
"step": 400
},
{
"epoch": 0.05,
"learning_rate": 4.1e-05,
"loss": 0.0473,
"step": 410
},
{
"epoch": 0.05,
"learning_rate": 4.2e-05,
"loss": 0.0624,
"step": 420
},
{
"epoch": 0.05,
"learning_rate": 4.3e-05,
"loss": 0.0774,
"step": 430
},
{
"epoch": 0.06,
"learning_rate": 4.4000000000000006e-05,
"loss": 0.0608,
"step": 440
},
{
"epoch": 0.06,
"learning_rate": 4.5e-05,
"loss": 0.0658,
"step": 450
},
{
"epoch": 0.06,
"learning_rate": 4.600000000000001e-05,
"loss": 0.0696,
"step": 460
},
{
"epoch": 0.06,
"learning_rate": 4.7e-05,
"loss": 0.0577,
"step": 470
},
{
"epoch": 0.06,
"learning_rate": 4.8e-05,
"loss": 0.0439,
"step": 480
},
{
"epoch": 0.06,
"learning_rate": 4.9e-05,
"loss": 0.0345,
"step": 490
},
{
"epoch": 0.06,
"learning_rate": 5e-05,
"loss": 0.075,
"step": 500
},
{
"epoch": 0.06,
"learning_rate": 4.993314614253243e-05,
"loss": 0.0508,
"step": 510
},
{
"epoch": 0.07,
"learning_rate": 4.986629228506485e-05,
"loss": 0.0911,
"step": 520
},
{
"epoch": 0.07,
"learning_rate": 4.9799438427597275e-05,
"loss": 0.0621,
"step": 530
},
{
"epoch": 0.07,
"learning_rate": 4.97325845701297e-05,
"loss": 0.0669,
"step": 540
},
{
"epoch": 0.07,
"learning_rate": 4.9665730712662124e-05,
"loss": 0.045,
"step": 550
},
{
"epoch": 0.07,
"learning_rate": 4.959887685519455e-05,
"loss": 0.0557,
"step": 560
},
{
"epoch": 0.07,
"learning_rate": 4.953202299772697e-05,
"loss": 0.057,
"step": 570
},
{
"epoch": 0.07,
"learning_rate": 4.946516914025939e-05,
"loss": 0.0718,
"step": 580
},
{
"epoch": 0.07,
"learning_rate": 4.939831528279182e-05,
"loss": 0.0379,
"step": 590
},
{
"epoch": 0.08,
"learning_rate": 4.9331461425324245e-05,
"loss": 0.0516,
"step": 600
},
{
"epoch": 0.08,
"learning_rate": 4.926460756785667e-05,
"loss": 0.0429,
"step": 610
},
{
"epoch": 0.08,
"learning_rate": 4.9197753710389094e-05,
"loss": 0.0652,
"step": 620
},
{
"epoch": 0.08,
"learning_rate": 4.913089985292151e-05,
"loss": 0.0538,
"step": 630
},
{
"epoch": 0.08,
"learning_rate": 4.9064045995453936e-05,
"loss": 0.0708,
"step": 640
},
{
"epoch": 0.08,
"learning_rate": 4.899719213798637e-05,
"loss": 0.0316,
"step": 650
},
{
"epoch": 0.08,
"learning_rate": 4.893033828051879e-05,
"loss": 0.0678,
"step": 660
},
{
"epoch": 0.08,
"learning_rate": 4.8863484423051216e-05,
"loss": 0.0645,
"step": 670
},
{
"epoch": 0.09,
"learning_rate": 4.879663056558363e-05,
"loss": 0.0879,
"step": 680
},
{
"epoch": 0.09,
"learning_rate": 4.872977670811606e-05,
"loss": 0.0391,
"step": 690
},
{
"epoch": 0.09,
"learning_rate": 4.866292285064848e-05,
"loss": 0.0945,
"step": 700
},
{
"epoch": 0.09,
"learning_rate": 4.859606899318091e-05,
"loss": 0.0726,
"step": 710
},
{
"epoch": 0.09,
"learning_rate": 4.852921513571334e-05,
"loss": 0.0898,
"step": 720
},
{
"epoch": 0.09,
"learning_rate": 4.8462361278245755e-05,
"loss": 0.0664,
"step": 730
},
{
"epoch": 0.09,
"learning_rate": 4.839550742077818e-05,
"loss": 0.0519,
"step": 740
},
{
"epoch": 0.09,
"learning_rate": 4.83286535633106e-05,
"loss": 0.0645,
"step": 750
},
{
"epoch": 0.1,
"learning_rate": 4.826179970584303e-05,
"loss": 0.0328,
"step": 760
},
{
"epoch": 0.1,
"learning_rate": 4.819494584837546e-05,
"loss": 0.042,
"step": 770
},
{
"epoch": 0.1,
"learning_rate": 4.8128091990907876e-05,
"loss": 0.0897,
"step": 780
},
{
"epoch": 0.1,
"learning_rate": 4.80612381334403e-05,
"loss": 0.0702,
"step": 790
},
{
"epoch": 0.1,
"learning_rate": 4.7994384275972725e-05,
"loss": 0.0582,
"step": 800
},
{
"epoch": 0.1,
"learning_rate": 4.792753041850515e-05,
"loss": 0.0381,
"step": 810
},
{
"epoch": 0.1,
"learning_rate": 4.786067656103757e-05,
"loss": 0.0623,
"step": 820
},
{
"epoch": 0.1,
"learning_rate": 4.779382270357e-05,
"loss": 0.0628,
"step": 830
},
{
"epoch": 0.11,
"learning_rate": 4.772696884610242e-05,
"loss": 0.0543,
"step": 840
},
{
"epoch": 0.11,
"learning_rate": 4.7660114988634846e-05,
"loss": 0.0455,
"step": 850
},
{
"epoch": 0.11,
"learning_rate": 4.759326113116727e-05,
"loss": 0.0511,
"step": 860
},
{
"epoch": 0.11,
"learning_rate": 4.7526407273699695e-05,
"loss": 0.0421,
"step": 870
},
{
"epoch": 0.11,
"learning_rate": 4.745955341623212e-05,
"loss": 0.0781,
"step": 880
},
{
"epoch": 0.11,
"learning_rate": 4.7392699558764544e-05,
"loss": 0.0389,
"step": 890
},
{
"epoch": 0.11,
"learning_rate": 4.732584570129697e-05,
"loss": 0.0512,
"step": 900
},
{
"epoch": 0.11,
"learning_rate": 4.725899184382939e-05,
"loss": 0.0558,
"step": 910
},
{
"epoch": 0.12,
"learning_rate": 4.7192137986361816e-05,
"loss": 0.0535,
"step": 920
},
{
"epoch": 0.12,
"learning_rate": 4.7125284128894234e-05,
"loss": 0.0811,
"step": 930
},
{
"epoch": 0.12,
"learning_rate": 4.7058430271426665e-05,
"loss": 0.0587,
"step": 940
},
{
"epoch": 0.12,
"learning_rate": 4.699157641395909e-05,
"loss": 0.0865,
"step": 950
},
{
"epoch": 0.12,
"learning_rate": 4.6924722556491514e-05,
"loss": 0.0483,
"step": 960
},
{
"epoch": 0.12,
"learning_rate": 4.685786869902394e-05,
"loss": 0.0479,
"step": 970
},
{
"epoch": 0.12,
"learning_rate": 4.6791014841556355e-05,
"loss": 0.0682,
"step": 980
},
{
"epoch": 0.12,
"learning_rate": 4.672416098408878e-05,
"loss": 0.0807,
"step": 990
},
{
"epoch": 0.13,
"learning_rate": 4.665730712662121e-05,
"loss": 0.0487,
"step": 1000
},
{
"epoch": 0.13,
"learning_rate": 4.6590453269153635e-05,
"loss": 0.0375,
"step": 1010
},
{
"epoch": 0.13,
"learning_rate": 4.652359941168606e-05,
"loss": 0.0318,
"step": 1020
},
{
"epoch": 0.13,
"learning_rate": 4.645674555421848e-05,
"loss": 0.0575,
"step": 1030
},
{
"epoch": 0.13,
"learning_rate": 4.63898916967509e-05,
"loss": 0.0312,
"step": 1040
},
{
"epoch": 0.13,
"learning_rate": 4.6323037839283326e-05,
"loss": 0.0328,
"step": 1050
},
{
"epoch": 0.13,
"learning_rate": 4.625618398181576e-05,
"loss": 0.0448,
"step": 1060
},
{
"epoch": 0.13,
"learning_rate": 4.618933012434818e-05,
"loss": 0.0525,
"step": 1070
},
{
"epoch": 0.14,
"learning_rate": 4.61224762668806e-05,
"loss": 0.0384,
"step": 1080
},
{
"epoch": 0.14,
"learning_rate": 4.605562240941302e-05,
"loss": 0.0598,
"step": 1090
},
{
"epoch": 0.14,
"learning_rate": 4.598876855194545e-05,
"loss": 0.0279,
"step": 1100
},
{
"epoch": 0.14,
"learning_rate": 4.592191469447787e-05,
"loss": 0.0765,
"step": 1110
},
{
"epoch": 0.14,
"learning_rate": 4.58550608370103e-05,
"loss": 0.0526,
"step": 1120
},
{
"epoch": 0.14,
"learning_rate": 4.578820697954272e-05,
"loss": 0.0595,
"step": 1130
},
{
"epoch": 0.14,
"learning_rate": 4.5721353122075144e-05,
"loss": 0.0373,
"step": 1140
},
{
"epoch": 0.14,
"learning_rate": 4.565449926460757e-05,
"loss": 0.0443,
"step": 1150
},
{
"epoch": 0.15,
"learning_rate": 4.558764540713999e-05,
"loss": 0.041,
"step": 1160
},
{
"epoch": 0.15,
"learning_rate": 4.552079154967242e-05,
"loss": 0.0664,
"step": 1170
},
{
"epoch": 0.15,
"learning_rate": 4.545393769220484e-05,
"loss": 0.0544,
"step": 1180
},
{
"epoch": 0.15,
"learning_rate": 4.5387083834737266e-05,
"loss": 0.0832,
"step": 1190
},
{
"epoch": 0.15,
"learning_rate": 4.532022997726969e-05,
"loss": 0.0731,
"step": 1200
},
{
"epoch": 0.15,
"learning_rate": 4.5253376119802115e-05,
"loss": 0.0496,
"step": 1210
},
{
"epoch": 0.15,
"learning_rate": 4.518652226233454e-05,
"loss": 0.0478,
"step": 1220
},
{
"epoch": 0.15,
"learning_rate": 4.511966840486696e-05,
"loss": 0.0502,
"step": 1230
},
{
"epoch": 0.16,
"learning_rate": 4.505281454739939e-05,
"loss": 0.0392,
"step": 1240
},
{
"epoch": 0.16,
"learning_rate": 4.498596068993181e-05,
"loss": 0.0393,
"step": 1250
},
{
"epoch": 0.16,
"learning_rate": 4.4919106832464236e-05,
"loss": 0.051,
"step": 1260
},
{
"epoch": 0.16,
"learning_rate": 4.485225297499666e-05,
"loss": 0.0487,
"step": 1270
},
{
"epoch": 0.16,
"learning_rate": 4.478539911752908e-05,
"loss": 0.0411,
"step": 1280
},
{
"epoch": 0.16,
"learning_rate": 4.471854526006151e-05,
"loss": 0.0466,
"step": 1290
},
{
"epoch": 0.16,
"learning_rate": 4.465169140259393e-05,
"loss": 0.0604,
"step": 1300
},
{
"epoch": 0.16,
"learning_rate": 4.458483754512636e-05,
"loss": 0.0457,
"step": 1310
},
{
"epoch": 0.17,
"learning_rate": 4.451798368765878e-05,
"loss": 0.037,
"step": 1320
},
{
"epoch": 0.17,
"learning_rate": 4.44511298301912e-05,
"loss": 0.0573,
"step": 1330
},
{
"epoch": 0.17,
"learning_rate": 4.4384275972723624e-05,
"loss": 0.0548,
"step": 1340
},
{
"epoch": 0.17,
"learning_rate": 4.4317422115256055e-05,
"loss": 0.0381,
"step": 1350
},
{
"epoch": 0.17,
"learning_rate": 4.425056825778848e-05,
"loss": 0.1086,
"step": 1360
},
{
"epoch": 0.17,
"learning_rate": 4.4183714400320903e-05,
"loss": 0.039,
"step": 1370
},
{
"epoch": 0.17,
"learning_rate": 4.411686054285332e-05,
"loss": 0.0499,
"step": 1380
},
{
"epoch": 0.17,
"learning_rate": 4.4050006685385745e-05,
"loss": 0.0326,
"step": 1390
},
{
"epoch": 0.18,
"learning_rate": 4.398315282791817e-05,
"loss": 0.0159,
"step": 1400
},
{
"epoch": 0.18,
"learning_rate": 4.39162989704506e-05,
"loss": 0.0553,
"step": 1410
},
{
"epoch": 0.18,
"learning_rate": 4.3849445112983025e-05,
"loss": 0.0654,
"step": 1420
},
{
"epoch": 0.18,
"learning_rate": 4.378259125551544e-05,
"loss": 0.0732,
"step": 1430
},
{
"epoch": 0.18,
"learning_rate": 4.371573739804787e-05,
"loss": 0.0598,
"step": 1440
},
{
"epoch": 0.18,
"learning_rate": 4.364888354058029e-05,
"loss": 0.0495,
"step": 1450
},
{
"epoch": 0.18,
"learning_rate": 4.3582029683112715e-05,
"loss": 0.0773,
"step": 1460
},
{
"epoch": 0.18,
"learning_rate": 4.3515175825645146e-05,
"loss": 0.0368,
"step": 1470
},
{
"epoch": 0.19,
"learning_rate": 4.3448321968177564e-05,
"loss": 0.0349,
"step": 1480
},
{
"epoch": 0.19,
"learning_rate": 4.338146811070999e-05,
"loss": 0.0587,
"step": 1490
},
{
"epoch": 0.19,
"learning_rate": 4.331461425324241e-05,
"loss": 0.0426,
"step": 1500
},
{
"epoch": 0.19,
"learning_rate": 4.324776039577484e-05,
"loss": 0.0483,
"step": 1510
},
{
"epoch": 0.19,
"learning_rate": 4.318090653830726e-05,
"loss": 0.0628,
"step": 1520
},
{
"epoch": 0.19,
"learning_rate": 4.3114052680839686e-05,
"loss": 0.0569,
"step": 1530
},
{
"epoch": 0.19,
"learning_rate": 4.304719882337211e-05,
"loss": 0.0485,
"step": 1540
},
{
"epoch": 0.19,
"learning_rate": 4.2980344965904534e-05,
"loss": 0.0457,
"step": 1550
},
{
"epoch": 0.2,
"learning_rate": 4.291349110843696e-05,
"loss": 0.0529,
"step": 1560
},
{
"epoch": 0.2,
"learning_rate": 4.284663725096938e-05,
"loss": 0.0471,
"step": 1570
},
{
"epoch": 0.2,
"learning_rate": 4.277978339350181e-05,
"loss": 0.0494,
"step": 1580
},
{
"epoch": 0.2,
"learning_rate": 4.271292953603423e-05,
"loss": 0.0402,
"step": 1590
},
{
"epoch": 0.2,
"learning_rate": 4.2646075678566656e-05,
"loss": 0.0277,
"step": 1600
},
{
"epoch": 0.2,
"learning_rate": 4.257922182109908e-05,
"loss": 0.0748,
"step": 1610
},
{
"epoch": 0.2,
"learning_rate": 4.2512367963631504e-05,
"loss": 0.0439,
"step": 1620
},
{
"epoch": 0.2,
"learning_rate": 4.244551410616392e-05,
"loss": 0.0468,
"step": 1630
},
{
"epoch": 0.21,
"learning_rate": 4.237866024869635e-05,
"loss": 0.0555,
"step": 1640
},
{
"epoch": 0.21,
"learning_rate": 4.231180639122878e-05,
"loss": 0.0473,
"step": 1650
},
{
"epoch": 0.21,
"learning_rate": 4.22449525337612e-05,
"loss": 0.0458,
"step": 1660
},
{
"epoch": 0.21,
"learning_rate": 4.2178098676293626e-05,
"loss": 0.0583,
"step": 1670
},
{
"epoch": 0.21,
"learning_rate": 4.211124481882604e-05,
"loss": 0.0405,
"step": 1680
},
{
"epoch": 0.21,
"learning_rate": 4.204439096135847e-05,
"loss": 0.034,
"step": 1690
},
{
"epoch": 0.21,
"learning_rate": 4.19775371038909e-05,
"loss": 0.0857,
"step": 1700
},
{
"epoch": 0.21,
"learning_rate": 4.191068324642332e-05,
"loss": 0.0415,
"step": 1710
},
{
"epoch": 0.22,
"learning_rate": 4.184382938895575e-05,
"loss": 0.0621,
"step": 1720
},
{
"epoch": 0.22,
"learning_rate": 4.1776975531488165e-05,
"loss": 0.055,
"step": 1730
},
{
"epoch": 0.22,
"learning_rate": 4.171012167402059e-05,
"loss": 0.0267,
"step": 1740
},
{
"epoch": 0.22,
"learning_rate": 4.1643267816553013e-05,
"loss": 0.0433,
"step": 1750
},
{
"epoch": 0.22,
"learning_rate": 4.1576413959085445e-05,
"loss": 0.0384,
"step": 1760
},
{
"epoch": 0.22,
"learning_rate": 4.150956010161787e-05,
"loss": 0.0485,
"step": 1770
},
{
"epoch": 0.22,
"learning_rate": 4.1442706244150286e-05,
"loss": 0.0278,
"step": 1780
},
{
"epoch": 0.22,
"learning_rate": 4.137585238668271e-05,
"loss": 0.0669,
"step": 1790
},
{
"epoch": 0.23,
"learning_rate": 4.1308998529215135e-05,
"loss": 0.0554,
"step": 1800
},
{
"epoch": 0.23,
"learning_rate": 4.124214467174756e-05,
"loss": 0.0392,
"step": 1810
},
{
"epoch": 0.23,
"learning_rate": 4.117529081427999e-05,
"loss": 0.0632,
"step": 1820
},
{
"epoch": 0.23,
"learning_rate": 4.1108436956812415e-05,
"loss": 0.0494,
"step": 1830
},
{
"epoch": 0.23,
"learning_rate": 4.104158309934483e-05,
"loss": 0.0334,
"step": 1840
},
{
"epoch": 0.23,
"learning_rate": 4.0974729241877256e-05,
"loss": 0.0658,
"step": 1850
},
{
"epoch": 0.23,
"learning_rate": 4.090787538440968e-05,
"loss": 0.0456,
"step": 1860
},
{
"epoch": 0.23,
"learning_rate": 4.0841021526942105e-05,
"loss": 0.0359,
"step": 1870
},
{
"epoch": 0.24,
"learning_rate": 4.0774167669474536e-05,
"loss": 0.0513,
"step": 1880
},
{
"epoch": 0.24,
"learning_rate": 4.0707313812006954e-05,
"loss": 0.0391,
"step": 1890
},
{
"epoch": 0.24,
"learning_rate": 4.064045995453938e-05,
"loss": 0.0507,
"step": 1900
},
{
"epoch": 0.24,
"learning_rate": 4.05736060970718e-05,
"loss": 0.0458,
"step": 1910
},
{
"epoch": 0.24,
"learning_rate": 4.050675223960423e-05,
"loss": 0.0572,
"step": 1920
},
{
"epoch": 0.24,
"learning_rate": 4.043989838213665e-05,
"loss": 0.0326,
"step": 1930
},
{
"epoch": 0.24,
"learning_rate": 4.0373044524669075e-05,
"loss": 0.0446,
"step": 1940
},
{
"epoch": 0.24,
"learning_rate": 4.03061906672015e-05,
"loss": 0.0359,
"step": 1950
},
{
"epoch": 0.25,
"learning_rate": 4.0239336809733924e-05,
"loss": 0.0689,
"step": 1960
},
{
"epoch": 0.25,
"learning_rate": 4.017248295226635e-05,
"loss": 0.0165,
"step": 1970
},
{
"epoch": 0.25,
"learning_rate": 4.010562909479877e-05,
"loss": 0.0442,
"step": 1980
},
{
"epoch": 0.25,
"learning_rate": 4.00387752373312e-05,
"loss": 0.0343,
"step": 1990
},
{
"epoch": 0.25,
"learning_rate": 3.997192137986362e-05,
"loss": 0.0684,
"step": 2000
},
{
"epoch": 0.25,
"learning_rate": 3.9905067522396045e-05,
"loss": 0.0565,
"step": 2010
},
{
"epoch": 0.25,
"learning_rate": 3.983821366492847e-05,
"loss": 0.0534,
"step": 2020
},
{
"epoch": 0.25,
"learning_rate": 3.9771359807460894e-05,
"loss": 0.0362,
"step": 2030
},
{
"epoch": 0.26,
"learning_rate": 3.970450594999331e-05,
"loss": 0.0464,
"step": 2040
},
{
"epoch": 0.26,
"learning_rate": 3.963765209252574e-05,
"loss": 0.021,
"step": 2050
},
{
"epoch": 0.26,
"learning_rate": 3.957079823505817e-05,
"loss": 0.0531,
"step": 2060
},
{
"epoch": 0.26,
"learning_rate": 3.950394437759059e-05,
"loss": 0.0669,
"step": 2070
},
{
"epoch": 0.26,
"learning_rate": 3.9437090520123016e-05,
"loss": 0.0509,
"step": 2080
},
{
"epoch": 0.26,
"learning_rate": 3.937023666265543e-05,
"loss": 0.0409,
"step": 2090
},
{
"epoch": 0.26,
"learning_rate": 3.930338280518786e-05,
"loss": 0.0518,
"step": 2100
},
{
"epoch": 0.26,
"learning_rate": 3.923652894772029e-05,
"loss": 0.0316,
"step": 2110
},
{
"epoch": 0.27,
"learning_rate": 3.916967509025271e-05,
"loss": 0.0479,
"step": 2120
},
{
"epoch": 0.27,
"learning_rate": 3.910282123278514e-05,
"loss": 0.0329,
"step": 2130
},
{
"epoch": 0.27,
"learning_rate": 3.9035967375317555e-05,
"loss": 0.032,
"step": 2140
},
{
"epoch": 0.27,
"learning_rate": 3.896911351784998e-05,
"loss": 0.053,
"step": 2150
},
{
"epoch": 0.27,
"learning_rate": 3.89022596603824e-05,
"loss": 0.0545,
"step": 2160
},
{
"epoch": 0.27,
"learning_rate": 3.8835405802914834e-05,
"loss": 0.0269,
"step": 2170
},
{
"epoch": 0.27,
"learning_rate": 3.876855194544726e-05,
"loss": 0.0408,
"step": 2180
},
{
"epoch": 0.27,
"learning_rate": 3.8701698087979676e-05,
"loss": 0.0334,
"step": 2190
},
{
"epoch": 0.28,
"learning_rate": 3.86348442305121e-05,
"loss": 0.041,
"step": 2200
},
{
"epoch": 0.28,
"learning_rate": 3.8567990373044525e-05,
"loss": 0.0353,
"step": 2210
},
{
"epoch": 0.28,
"learning_rate": 3.850113651557695e-05,
"loss": 0.025,
"step": 2220
},
{
"epoch": 0.28,
"learning_rate": 3.843428265810938e-05,
"loss": 0.0365,
"step": 2230
},
{
"epoch": 0.28,
"learning_rate": 3.83674288006418e-05,
"loss": 0.0366,
"step": 2240
},
{
"epoch": 0.28,
"learning_rate": 3.830057494317422e-05,
"loss": 0.0451,
"step": 2250
},
{
"epoch": 0.28,
"learning_rate": 3.8233721085706646e-05,
"loss": 0.0495,
"step": 2260
},
{
"epoch": 0.28,
"learning_rate": 3.816686722823907e-05,
"loss": 0.0451,
"step": 2270
},
{
"epoch": 0.29,
"learning_rate": 3.8100013370771495e-05,
"loss": 0.0441,
"step": 2280
},
{
"epoch": 0.29,
"learning_rate": 3.803315951330392e-05,
"loss": 0.0532,
"step": 2290
},
{
"epoch": 0.29,
"learning_rate": 3.7966305655836343e-05,
"loss": 0.0666,
"step": 2300
},
{
"epoch": 0.29,
"learning_rate": 3.789945179836877e-05,
"loss": 0.0434,
"step": 2310
},
{
"epoch": 0.29,
"learning_rate": 3.783259794090119e-05,
"loss": 0.0522,
"step": 2320
},
{
"epoch": 0.29,
"learning_rate": 3.7765744083433616e-05,
"loss": 0.0368,
"step": 2330
},
{
"epoch": 0.29,
"learning_rate": 3.769889022596604e-05,
"loss": 0.0386,
"step": 2340
},
{
"epoch": 0.29,
"learning_rate": 3.7632036368498465e-05,
"loss": 0.0639,
"step": 2350
},
{
"epoch": 0.3,
"learning_rate": 3.756518251103089e-05,
"loss": 0.0348,
"step": 2360
},
{
"epoch": 0.3,
"learning_rate": 3.7498328653563314e-05,
"loss": 0.0332,
"step": 2370
},
{
"epoch": 0.3,
"learning_rate": 3.743147479609574e-05,
"loss": 0.0439,
"step": 2380
},
{
"epoch": 0.3,
"learning_rate": 3.7364620938628155e-05,
"loss": 0.0433,
"step": 2390
},
{
"epoch": 0.3,
"learning_rate": 3.7297767081160587e-05,
"loss": 0.0506,
"step": 2400
},
{
"epoch": 0.3,
"learning_rate": 3.723091322369301e-05,
"loss": 0.0618,
"step": 2410
},
{
"epoch": 0.3,
"learning_rate": 3.7164059366225435e-05,
"loss": 0.0326,
"step": 2420
},
{
"epoch": 0.3,
"learning_rate": 3.709720550875786e-05,
"loss": 0.0589,
"step": 2430
},
{
"epoch": 0.31,
"learning_rate": 3.703035165129028e-05,
"loss": 0.0586,
"step": 2440
},
{
"epoch": 0.31,
"learning_rate": 3.69634977938227e-05,
"loss": 0.0616,
"step": 2450
},
{
"epoch": 0.31,
"learning_rate": 3.689664393635513e-05,
"loss": 0.0355,
"step": 2460
},
{
"epoch": 0.31,
"learning_rate": 3.682979007888756e-05,
"loss": 0.0399,
"step": 2470
},
{
"epoch": 0.31,
"learning_rate": 3.676293622141998e-05,
"loss": 0.0609,
"step": 2480
},
{
"epoch": 0.31,
"learning_rate": 3.66960823639524e-05,
"loss": 0.0679,
"step": 2490
},
{
"epoch": 0.31,
"learning_rate": 3.662922850648482e-05,
"loss": 0.0427,
"step": 2500
},
{
"epoch": 0.31,
"learning_rate": 3.656237464901725e-05,
"loss": 0.0333,
"step": 2510
},
{
"epoch": 0.32,
"learning_rate": 3.649552079154968e-05,
"loss": 0.0329,
"step": 2520
},
{
"epoch": 0.32,
"learning_rate": 3.64286669340821e-05,
"loss": 0.048,
"step": 2530
},
{
"epoch": 0.32,
"learning_rate": 3.636181307661452e-05,
"loss": 0.0617,
"step": 2540
},
{
"epoch": 0.32,
"learning_rate": 3.6294959219146944e-05,
"loss": 0.061,
"step": 2550
},
{
"epoch": 0.32,
"learning_rate": 3.622810536167937e-05,
"loss": 0.0551,
"step": 2560
},
{
"epoch": 0.32,
"learning_rate": 3.616125150421179e-05,
"loss": 0.05,
"step": 2570
},
{
"epoch": 0.32,
"learning_rate": 3.6094397646744224e-05,
"loss": 0.0198,
"step": 2580
},
{
"epoch": 0.32,
"learning_rate": 3.602754378927664e-05,
"loss": 0.0502,
"step": 2590
},
{
"epoch": 0.33,
"learning_rate": 3.5960689931809066e-05,
"loss": 0.0549,
"step": 2600
},
{
"epoch": 0.33,
"learning_rate": 3.589383607434149e-05,
"loss": 0.0385,
"step": 2610
},
{
"epoch": 0.33,
"learning_rate": 3.5826982216873914e-05,
"loss": 0.0594,
"step": 2620
},
{
"epoch": 0.33,
"learning_rate": 3.576012835940634e-05,
"loss": 0.0599,
"step": 2630
},
{
"epoch": 0.33,
"learning_rate": 3.569327450193876e-05,
"loss": 0.0456,
"step": 2640
},
{
"epoch": 0.33,
"learning_rate": 3.562642064447119e-05,
"loss": 0.0379,
"step": 2650
},
{
"epoch": 0.33,
"learning_rate": 3.555956678700361e-05,
"loss": 0.0476,
"step": 2660
},
{
"epoch": 0.33,
"learning_rate": 3.5492712929536036e-05,
"loss": 0.055,
"step": 2670
},
{
"epoch": 0.34,
"learning_rate": 3.542585907206846e-05,
"loss": 0.0425,
"step": 2680
},
{
"epoch": 0.34,
"learning_rate": 3.5359005214600885e-05,
"loss": 0.0428,
"step": 2690
},
{
"epoch": 0.34,
"learning_rate": 3.529215135713331e-05,
"loss": 0.0264,
"step": 2700
},
{
"epoch": 0.34,
"learning_rate": 3.522529749966573e-05,
"loss": 0.0526,
"step": 2710
},
{
"epoch": 0.34,
"learning_rate": 3.515844364219816e-05,
"loss": 0.0324,
"step": 2720
},
{
"epoch": 0.34,
"learning_rate": 3.509158978473058e-05,
"loss": 0.0385,
"step": 2730
},
{
"epoch": 0.34,
"learning_rate": 3.5024735927263e-05,
"loss": 0.0675,
"step": 2740
},
{
"epoch": 0.34,
"learning_rate": 3.495788206979543e-05,
"loss": 0.0556,
"step": 2750
},
{
"epoch": 0.35,
"learning_rate": 3.4891028212327855e-05,
"loss": 0.0486,
"step": 2760
},
{
"epoch": 0.35,
"learning_rate": 3.482417435486028e-05,
"loss": 0.0347,
"step": 2770
},
{
"epoch": 0.35,
"learning_rate": 3.47573204973927e-05,
"loss": 0.0359,
"step": 2780
},
{
"epoch": 0.35,
"learning_rate": 3.469046663992512e-05,
"loss": 0.0684,
"step": 2790
},
{
"epoch": 0.35,
"learning_rate": 3.4623612782457545e-05,
"loss": 0.0524,
"step": 2800
},
{
"epoch": 0.35,
"learning_rate": 3.4556758924989976e-05,
"loss": 0.0604,
"step": 2810
},
{
"epoch": 0.35,
"learning_rate": 3.44899050675224e-05,
"loss": 0.0377,
"step": 2820
},
{
"epoch": 0.35,
"learning_rate": 3.4423051210054825e-05,
"loss": 0.0592,
"step": 2830
},
{
"epoch": 0.36,
"learning_rate": 3.435619735258724e-05,
"loss": 0.0483,
"step": 2840
},
{
"epoch": 0.36,
"learning_rate": 3.428934349511967e-05,
"loss": 0.0544,
"step": 2850
},
{
"epoch": 0.36,
"learning_rate": 3.422248963765209e-05,
"loss": 0.0554,
"step": 2860
},
{
"epoch": 0.36,
"learning_rate": 3.415563578018452e-05,
"loss": 0.0427,
"step": 2870
},
{
"epoch": 0.36,
"learning_rate": 3.4088781922716946e-05,
"loss": 0.0514,
"step": 2880
},
{
"epoch": 0.36,
"learning_rate": 3.4021928065249364e-05,
"loss": 0.0445,
"step": 2890
},
{
"epoch": 0.36,
"learning_rate": 3.395507420778179e-05,
"loss": 0.0453,
"step": 2900
},
{
"epoch": 0.36,
"learning_rate": 3.388822035031421e-05,
"loss": 0.0435,
"step": 2910
},
{
"epoch": 0.37,
"learning_rate": 3.382136649284664e-05,
"loss": 0.027,
"step": 2920
},
{
"epoch": 0.37,
"learning_rate": 3.375451263537907e-05,
"loss": 0.0369,
"step": 2930
},
{
"epoch": 0.37,
"learning_rate": 3.3687658777911485e-05,
"loss": 0.0629,
"step": 2940
},
{
"epoch": 0.37,
"learning_rate": 3.362080492044391e-05,
"loss": 0.0589,
"step": 2950
},
{
"epoch": 0.37,
"learning_rate": 3.3553951062976334e-05,
"loss": 0.0449,
"step": 2960
},
{
"epoch": 0.37,
"learning_rate": 3.348709720550876e-05,
"loss": 0.0464,
"step": 2970
},
{
"epoch": 0.37,
"learning_rate": 3.342024334804118e-05,
"loss": 0.0477,
"step": 2980
},
{
"epoch": 0.37,
"learning_rate": 3.335338949057361e-05,
"loss": 0.0263,
"step": 2990
},
{
"epoch": 0.38,
"learning_rate": 3.328653563310603e-05,
"loss": 0.0477,
"step": 3000
},
{
"epoch": 0.38,
"learning_rate": 3.3219681775638456e-05,
"loss": 0.038,
"step": 3010
},
{
"epoch": 0.38,
"learning_rate": 3.315282791817088e-05,
"loss": 0.0581,
"step": 3020
},
{
"epoch": 0.38,
"learning_rate": 3.3085974060703304e-05,
"loss": 0.0237,
"step": 3030
},
{
"epoch": 0.38,
"learning_rate": 3.301912020323573e-05,
"loss": 0.0469,
"step": 3040
},
{
"epoch": 0.38,
"learning_rate": 3.295226634576815e-05,
"loss": 0.0262,
"step": 3050
},
{
"epoch": 0.38,
"learning_rate": 3.288541248830058e-05,
"loss": 0.0294,
"step": 3060
},
{
"epoch": 0.38,
"learning_rate": 3.2818558630833e-05,
"loss": 0.0421,
"step": 3070
},
{
"epoch": 0.39,
"learning_rate": 3.2751704773365426e-05,
"loss": 0.0295,
"step": 3080
},
{
"epoch": 0.39,
"learning_rate": 3.268485091589785e-05,
"loss": 0.0309,
"step": 3090
},
{
"epoch": 0.39,
"learning_rate": 3.2617997058430274e-05,
"loss": 0.0434,
"step": 3100
},
{
"epoch": 0.39,
"learning_rate": 3.25511432009627e-05,
"loss": 0.0609,
"step": 3110
},
{
"epoch": 0.39,
"learning_rate": 3.248428934349512e-05,
"loss": 0.0401,
"step": 3120
},
{
"epoch": 0.39,
"learning_rate": 3.241743548602755e-05,
"loss": 0.0235,
"step": 3130
},
{
"epoch": 0.39,
"learning_rate": 3.2350581628559965e-05,
"loss": 0.0369,
"step": 3140
},
{
"epoch": 0.39,
"learning_rate": 3.228372777109239e-05,
"loss": 0.0446,
"step": 3150
},
{
"epoch": 0.4,
"learning_rate": 3.221687391362482e-05,
"loss": 0.0289,
"step": 3160
},
{
"epoch": 0.4,
"learning_rate": 3.2150020056157244e-05,
"loss": 0.0369,
"step": 3170
},
{
"epoch": 0.4,
"learning_rate": 3.208316619868967e-05,
"loss": 0.0243,
"step": 3180
},
{
"epoch": 0.4,
"learning_rate": 3.2016312341222086e-05,
"loss": 0.0308,
"step": 3190
},
{
"epoch": 0.4,
"learning_rate": 3.194945848375451e-05,
"loss": 0.0564,
"step": 3200
},
{
"epoch": 0.4,
"learning_rate": 3.1882604626286935e-05,
"loss": 0.0855,
"step": 3210
},
{
"epoch": 0.4,
"learning_rate": 3.1815750768819366e-05,
"loss": 0.045,
"step": 3220
},
{
"epoch": 0.4,
"learning_rate": 3.174889691135179e-05,
"loss": 0.0563,
"step": 3230
},
{
"epoch": 0.41,
"learning_rate": 3.168204305388421e-05,
"loss": 0.052,
"step": 3240
},
{
"epoch": 0.41,
"learning_rate": 3.161518919641663e-05,
"loss": 0.0467,
"step": 3250
},
{
"epoch": 0.41,
"learning_rate": 3.1548335338949056e-05,
"loss": 0.0346,
"step": 3260
},
{
"epoch": 0.41,
"learning_rate": 3.148148148148148e-05,
"loss": 0.041,
"step": 3270
},
{
"epoch": 0.41,
"learning_rate": 3.141462762401391e-05,
"loss": 0.0345,
"step": 3280
},
{
"epoch": 0.41,
"learning_rate": 3.134777376654633e-05,
"loss": 0.043,
"step": 3290
},
{
"epoch": 0.41,
"learning_rate": 3.1280919909078754e-05,
"loss": 0.0356,
"step": 3300
},
{
"epoch": 0.41,
"learning_rate": 3.121406605161118e-05,
"loss": 0.052,
"step": 3310
},
{
"epoch": 0.42,
"learning_rate": 3.11472121941436e-05,
"loss": 0.0477,
"step": 3320
},
{
"epoch": 0.42,
"learning_rate": 3.1080358336676027e-05,
"loss": 0.0307,
"step": 3330
},
{
"epoch": 0.42,
"learning_rate": 3.101350447920845e-05,
"loss": 0.0896,
"step": 3340
},
{
"epoch": 0.42,
"learning_rate": 3.0946650621740875e-05,
"loss": 0.0426,
"step": 3350
},
{
"epoch": 0.42,
"learning_rate": 3.08797967642733e-05,
"loss": 0.0581,
"step": 3360
},
{
"epoch": 0.42,
"learning_rate": 3.0812942906805724e-05,
"loss": 0.0531,
"step": 3370
},
{
"epoch": 0.42,
"learning_rate": 3.074608904933815e-05,
"loss": 0.0313,
"step": 3380
},
{
"epoch": 0.42,
"learning_rate": 3.067923519187057e-05,
"loss": 0.0296,
"step": 3390
},
{
"epoch": 0.43,
"learning_rate": 3.0612381334403e-05,
"loss": 0.0459,
"step": 3400
},
{
"epoch": 0.43,
"learning_rate": 3.054552747693542e-05,
"loss": 0.0298,
"step": 3410
},
{
"epoch": 0.43,
"learning_rate": 3.0478673619467845e-05,
"loss": 0.062,
"step": 3420
},
{
"epoch": 0.43,
"learning_rate": 3.041181976200027e-05,
"loss": 0.0329,
"step": 3430
},
{
"epoch": 0.43,
"learning_rate": 3.034496590453269e-05,
"loss": 0.0534,
"step": 3440
},
{
"epoch": 0.43,
"learning_rate": 3.0278112047065115e-05,
"loss": 0.045,
"step": 3450
},
{
"epoch": 0.43,
"learning_rate": 3.0211258189597543e-05,
"loss": 0.0533,
"step": 3460
},
{
"epoch": 0.43,
"learning_rate": 3.0144404332129967e-05,
"loss": 0.0584,
"step": 3470
},
{
"epoch": 0.44,
"learning_rate": 3.007755047466239e-05,
"loss": 0.0691,
"step": 3480
},
{
"epoch": 0.44,
"learning_rate": 3.0010696617194812e-05,
"loss": 0.0473,
"step": 3490
},
{
"epoch": 0.44,
"learning_rate": 2.9943842759727236e-05,
"loss": 0.0391,
"step": 3500
},
{
"epoch": 0.44,
"learning_rate": 2.987698890225966e-05,
"loss": 0.0325,
"step": 3510
},
{
"epoch": 0.44,
"learning_rate": 2.981013504479209e-05,
"loss": 0.0619,
"step": 3520
},
{
"epoch": 0.44,
"learning_rate": 2.9743281187324513e-05,
"loss": 0.0249,
"step": 3530
},
{
"epoch": 0.44,
"learning_rate": 2.967642732985693e-05,
"loss": 0.0431,
"step": 3540
},
{
"epoch": 0.44,
"learning_rate": 2.9609573472389358e-05,
"loss": 0.0391,
"step": 3550
},
{
"epoch": 0.45,
"learning_rate": 2.9542719614921782e-05,
"loss": 0.0374,
"step": 3560
},
{
"epoch": 0.45,
"learning_rate": 2.9475865757454206e-05,
"loss": 0.0354,
"step": 3570
},
{
"epoch": 0.45,
"learning_rate": 2.9409011899986634e-05,
"loss": 0.0467,
"step": 3580
},
{
"epoch": 0.45,
"learning_rate": 2.9342158042519052e-05,
"loss": 0.0186,
"step": 3590
},
{
"epoch": 0.45,
"learning_rate": 2.9275304185051476e-05,
"loss": 0.0408,
"step": 3600
},
{
"epoch": 0.45,
"learning_rate": 2.9208450327583904e-05,
"loss": 0.0468,
"step": 3610
},
{
"epoch": 0.45,
"learning_rate": 2.9141596470116328e-05,
"loss": 0.053,
"step": 3620
},
{
"epoch": 0.45,
"learning_rate": 2.9074742612648752e-05,
"loss": 0.0489,
"step": 3630
},
{
"epoch": 0.46,
"learning_rate": 2.9007888755181173e-05,
"loss": 0.0854,
"step": 3640
},
{
"epoch": 0.46,
"learning_rate": 2.8941034897713598e-05,
"loss": 0.0521,
"step": 3650
},
{
"epoch": 0.46,
"learning_rate": 2.8874181040246022e-05,
"loss": 0.0383,
"step": 3660
},
{
"epoch": 0.46,
"learning_rate": 2.880732718277845e-05,
"loss": 0.0524,
"step": 3670
},
{
"epoch": 0.46,
"learning_rate": 2.8740473325310874e-05,
"loss": 0.0164,
"step": 3680
},
{
"epoch": 0.46,
"learning_rate": 2.8673619467843295e-05,
"loss": 0.0402,
"step": 3690
},
{
"epoch": 0.46,
"learning_rate": 2.860676561037572e-05,
"loss": 0.0559,
"step": 3700
},
{
"epoch": 0.46,
"learning_rate": 2.8539911752908143e-05,
"loss": 0.0442,
"step": 3710
},
{
"epoch": 0.47,
"learning_rate": 2.8473057895440568e-05,
"loss": 0.0563,
"step": 3720
},
{
"epoch": 0.47,
"learning_rate": 2.8406204037972995e-05,
"loss": 0.0336,
"step": 3730
},
{
"epoch": 0.47,
"learning_rate": 2.8339350180505413e-05,
"loss": 0.0225,
"step": 3740
},
{
"epoch": 0.47,
"learning_rate": 2.827249632303784e-05,
"loss": 0.0447,
"step": 3750
},
{
"epoch": 0.47,
"learning_rate": 2.8205642465570265e-05,
"loss": 0.0523,
"step": 3760
},
{
"epoch": 0.47,
"learning_rate": 2.813878860810269e-05,
"loss": 0.0383,
"step": 3770
},
{
"epoch": 0.47,
"learning_rate": 2.8071934750635114e-05,
"loss": 0.0623,
"step": 3780
},
{
"epoch": 0.47,
"learning_rate": 2.8005080893167534e-05,
"loss": 0.0593,
"step": 3790
},
{
"epoch": 0.48,
"learning_rate": 2.793822703569996e-05,
"loss": 0.0557,
"step": 3800
},
{
"epoch": 0.48,
"learning_rate": 2.7871373178232386e-05,
"loss": 0.0378,
"step": 3810
},
{
"epoch": 0.48,
"learning_rate": 2.780451932076481e-05,
"loss": 0.0453,
"step": 3820
},
{
"epoch": 0.48,
"learning_rate": 2.7737665463297235e-05,
"loss": 0.0486,
"step": 3830
},
{
"epoch": 0.48,
"learning_rate": 2.7670811605829656e-05,
"loss": 0.0371,
"step": 3840
},
{
"epoch": 0.48,
"learning_rate": 2.760395774836208e-05,
"loss": 0.0785,
"step": 3850
},
{
"epoch": 0.48,
"learning_rate": 2.7537103890894505e-05,
"loss": 0.0258,
"step": 3860
},
{
"epoch": 0.49,
"learning_rate": 2.7470250033426932e-05,
"loss": 0.0417,
"step": 3870
},
{
"epoch": 0.49,
"learning_rate": 2.7403396175959357e-05,
"loss": 0.0419,
"step": 3880
},
{
"epoch": 0.49,
"learning_rate": 2.7336542318491774e-05,
"loss": 0.025,
"step": 3890
},
{
"epoch": 0.49,
"learning_rate": 2.7269688461024202e-05,
"loss": 0.0528,
"step": 3900
},
{
"epoch": 0.49,
"learning_rate": 2.7202834603556626e-05,
"loss": 0.0461,
"step": 3910
},
{
"epoch": 0.49,
"learning_rate": 2.713598074608905e-05,
"loss": 0.0362,
"step": 3920
},
{
"epoch": 0.49,
"learning_rate": 2.7069126888621478e-05,
"loss": 0.031,
"step": 3930
},
{
"epoch": 0.49,
"learning_rate": 2.7002273031153896e-05,
"loss": 0.0961,
"step": 3940
},
{
"epoch": 0.5,
"learning_rate": 2.693541917368632e-05,
"loss": 0.0249,
"step": 3950
},
{
"epoch": 0.5,
"learning_rate": 2.6868565316218748e-05,
"loss": 0.0622,
"step": 3960
},
{
"epoch": 0.5,
"learning_rate": 2.6801711458751172e-05,
"loss": 0.0567,
"step": 3970
},
{
"epoch": 0.5,
"learning_rate": 2.6734857601283596e-05,
"loss": 0.0346,
"step": 3980
},
{
"epoch": 0.5,
"learning_rate": 2.6668003743816017e-05,
"loss": 0.0555,
"step": 3990
},
{
"epoch": 0.5,
"learning_rate": 2.660114988634844e-05,
"loss": 0.0486,
"step": 4000
},
{
"epoch": 0.5,
"learning_rate": 2.6534296028880866e-05,
"loss": 0.0431,
"step": 4010
},
{
"epoch": 0.5,
"learning_rate": 2.6467442171413293e-05,
"loss": 0.0649,
"step": 4020
},
{
"epoch": 0.51,
"learning_rate": 2.6400588313945718e-05,
"loss": 0.0697,
"step": 4030
},
{
"epoch": 0.51,
"learning_rate": 2.633373445647814e-05,
"loss": 0.0458,
"step": 4040
},
{
"epoch": 0.51,
"learning_rate": 2.6266880599010563e-05,
"loss": 0.0246,
"step": 4050
},
{
"epoch": 0.51,
"learning_rate": 2.6200026741542987e-05,
"loss": 0.0644,
"step": 4060
},
{
"epoch": 0.51,
"learning_rate": 2.613317288407541e-05,
"loss": 0.0487,
"step": 4070
},
{
"epoch": 0.51,
"learning_rate": 2.606631902660784e-05,
"loss": 0.0341,
"step": 4080
},
{
"epoch": 0.51,
"learning_rate": 2.5999465169140257e-05,
"loss": 0.0605,
"step": 4090
},
{
"epoch": 0.51,
"learning_rate": 2.5932611311672685e-05,
"loss": 0.0392,
"step": 4100
},
{
"epoch": 0.52,
"learning_rate": 2.586575745420511e-05,
"loss": 0.0438,
"step": 4110
},
{
"epoch": 0.52,
"learning_rate": 2.5798903596737533e-05,
"loss": 0.0518,
"step": 4120
},
{
"epoch": 0.52,
"learning_rate": 2.5732049739269957e-05,
"loss": 0.0386,
"step": 4130
},
{
"epoch": 0.52,
"learning_rate": 2.566519588180238e-05,
"loss": 0.0507,
"step": 4140
},
{
"epoch": 0.52,
"learning_rate": 2.5598342024334803e-05,
"loss": 0.0363,
"step": 4150
},
{
"epoch": 0.52,
"learning_rate": 2.553148816686723e-05,
"loss": 0.0262,
"step": 4160
},
{
"epoch": 0.52,
"learning_rate": 2.5464634309399655e-05,
"loss": 0.0399,
"step": 4170
},
{
"epoch": 0.52,
"learning_rate": 2.539778045193208e-05,
"loss": 0.0493,
"step": 4180
},
{
"epoch": 0.53,
"learning_rate": 2.53309265944645e-05,
"loss": 0.042,
"step": 4190
},
{
"epoch": 0.53,
"learning_rate": 2.5264072736996924e-05,
"loss": 0.038,
"step": 4200
},
{
"epoch": 0.53,
"learning_rate": 2.519721887952935e-05,
"loss": 0.0331,
"step": 4210
},
{
"epoch": 0.53,
"learning_rate": 2.5130365022061776e-05,
"loss": 0.0246,
"step": 4220
},
{
"epoch": 0.53,
"learning_rate": 2.50635111645942e-05,
"loss": 0.0578,
"step": 4230
},
{
"epoch": 0.53,
"learning_rate": 2.499665730712662e-05,
"loss": 0.0292,
"step": 4240
},
{
"epoch": 0.53,
"learning_rate": 2.492980344965905e-05,
"loss": 0.0358,
"step": 4250
},
{
"epoch": 0.53,
"learning_rate": 2.486294959219147e-05,
"loss": 0.0508,
"step": 4260
},
{
"epoch": 0.54,
"learning_rate": 2.4796095734723894e-05,
"loss": 0.0461,
"step": 4270
},
{
"epoch": 0.54,
"learning_rate": 2.472924187725632e-05,
"loss": 0.073,
"step": 4280
},
{
"epoch": 0.54,
"learning_rate": 2.4662388019788743e-05,
"loss": 0.0451,
"step": 4290
},
{
"epoch": 0.54,
"learning_rate": 2.4595534162321167e-05,
"loss": 0.0582,
"step": 4300
},
{
"epoch": 0.54,
"learning_rate": 2.452868030485359e-05,
"loss": 0.0356,
"step": 4310
},
{
"epoch": 0.54,
"learning_rate": 2.4461826447386016e-05,
"loss": 0.0308,
"step": 4320
},
{
"epoch": 0.54,
"learning_rate": 2.4394972589918437e-05,
"loss": 0.0531,
"step": 4330
},
{
"epoch": 0.54,
"learning_rate": 2.4328118732450864e-05,
"loss": 0.036,
"step": 4340
},
{
"epoch": 0.55,
"learning_rate": 2.426126487498329e-05,
"loss": 0.0445,
"step": 4350
},
{
"epoch": 0.55,
"learning_rate": 2.419441101751571e-05,
"loss": 0.0286,
"step": 4360
},
{
"epoch": 0.55,
"learning_rate": 2.4127557160048137e-05,
"loss": 0.0394,
"step": 4370
},
{
"epoch": 0.55,
"learning_rate": 2.4060703302580558e-05,
"loss": 0.0366,
"step": 4380
},
{
"epoch": 0.55,
"learning_rate": 2.3993849445112983e-05,
"loss": 0.0482,
"step": 4390
},
{
"epoch": 0.55,
"learning_rate": 2.392699558764541e-05,
"loss": 0.0431,
"step": 4400
},
{
"epoch": 0.55,
"learning_rate": 2.386014173017783e-05,
"loss": 0.0311,
"step": 4410
},
{
"epoch": 0.55,
"learning_rate": 2.3793287872710256e-05,
"loss": 0.053,
"step": 4420
},
{
"epoch": 0.56,
"learning_rate": 2.372643401524268e-05,
"loss": 0.0655,
"step": 4430
},
{
"epoch": 0.56,
"learning_rate": 2.3659580157775104e-05,
"loss": 0.0537,
"step": 4440
},
{
"epoch": 0.56,
"learning_rate": 2.359272630030753e-05,
"loss": 0.0314,
"step": 4450
},
{
"epoch": 0.56,
"learning_rate": 2.3525872442839953e-05,
"loss": 0.0506,
"step": 4460
},
{
"epoch": 0.56,
"learning_rate": 2.3459018585372377e-05,
"loss": 0.0239,
"step": 4470
},
{
"epoch": 0.56,
"learning_rate": 2.33921647279048e-05,
"loss": 0.0306,
"step": 4480
},
{
"epoch": 0.56,
"learning_rate": 2.3325310870437226e-05,
"loss": 0.0385,
"step": 4490
},
{
"epoch": 0.56,
"learning_rate": 2.325845701296965e-05,
"loss": 0.0336,
"step": 4500
},
{
"epoch": 0.57,
"learning_rate": 2.3191603155502074e-05,
"loss": 0.0592,
"step": 4510
},
{
"epoch": 0.57,
"learning_rate": 2.31247492980345e-05,
"loss": 0.04,
"step": 4520
},
{
"epoch": 0.57,
"learning_rate": 2.305789544056692e-05,
"loss": 0.0441,
"step": 4530
},
{
"epoch": 0.57,
"learning_rate": 2.2991041583099347e-05,
"loss": 0.0392,
"step": 4540
},
{
"epoch": 0.57,
"learning_rate": 2.292418772563177e-05,
"loss": 0.0373,
"step": 4550
},
{
"epoch": 0.57,
"learning_rate": 2.2857333868164192e-05,
"loss": 0.0291,
"step": 4560
},
{
"epoch": 0.57,
"learning_rate": 2.279048001069662e-05,
"loss": 0.0489,
"step": 4570
},
{
"epoch": 0.57,
"learning_rate": 2.272362615322904e-05,
"loss": 0.0714,
"step": 4580
},
{
"epoch": 0.58,
"learning_rate": 2.2656772295761465e-05,
"loss": 0.0389,
"step": 4590
},
{
"epoch": 0.58,
"learning_rate": 2.2589918438293893e-05,
"loss": 0.0391,
"step": 4600
},
{
"epoch": 0.58,
"learning_rate": 2.2523064580826314e-05,
"loss": 0.0314,
"step": 4610
},
{
"epoch": 0.58,
"learning_rate": 2.2456210723358738e-05,
"loss": 0.0548,
"step": 4620
},
{
"epoch": 0.58,
"learning_rate": 2.2389356865891163e-05,
"loss": 0.043,
"step": 4630
},
{
"epoch": 0.58,
"learning_rate": 2.2322503008423587e-05,
"loss": 0.039,
"step": 4640
},
{
"epoch": 0.58,
"learning_rate": 2.225564915095601e-05,
"loss": 0.0338,
"step": 4650
},
{
"epoch": 0.58,
"learning_rate": 2.2188795293488435e-05,
"loss": 0.0368,
"step": 4660
},
{
"epoch": 0.59,
"learning_rate": 2.212194143602086e-05,
"loss": 0.0324,
"step": 4670
},
{
"epoch": 0.59,
"learning_rate": 2.205508757855328e-05,
"loss": 0.0423,
"step": 4680
},
{
"epoch": 0.59,
"learning_rate": 2.198823372108571e-05,
"loss": 0.0333,
"step": 4690
},
{
"epoch": 0.59,
"learning_rate": 2.1921379863618133e-05,
"loss": 0.045,
"step": 4700
},
{
"epoch": 0.59,
"learning_rate": 2.1854526006150554e-05,
"loss": 0.0508,
"step": 4710
},
{
"epoch": 0.59,
"learning_rate": 2.178767214868298e-05,
"loss": 0.0314,
"step": 4720
},
{
"epoch": 0.59,
"learning_rate": 2.1720818291215402e-05,
"loss": 0.0305,
"step": 4730
},
{
"epoch": 0.59,
"learning_rate": 2.1653964433747826e-05,
"loss": 0.0338,
"step": 4740
},
{
"epoch": 0.6,
"learning_rate": 2.1587110576280254e-05,
"loss": 0.0562,
"step": 4750
},
{
"epoch": 0.6,
"learning_rate": 2.1520256718812675e-05,
"loss": 0.04,
"step": 4760
},
{
"epoch": 0.6,
"learning_rate": 2.14534028613451e-05,
"loss": 0.0362,
"step": 4770
},
{
"epoch": 0.6,
"learning_rate": 2.1386549003877524e-05,
"loss": 0.0468,
"step": 4780
},
{
"epoch": 0.6,
"learning_rate": 2.1319695146409948e-05,
"loss": 0.039,
"step": 4790
},
{
"epoch": 0.6,
"learning_rate": 2.1252841288942372e-05,
"loss": 0.0533,
"step": 4800
},
{
"epoch": 0.6,
"learning_rate": 2.1185987431474797e-05,
"loss": 0.0505,
"step": 4810
},
{
"epoch": 0.6,
"learning_rate": 2.111913357400722e-05,
"loss": 0.0399,
"step": 4820
},
{
"epoch": 0.61,
"learning_rate": 2.1052279716539645e-05,
"loss": 0.0371,
"step": 4830
},
{
"epoch": 0.61,
"learning_rate": 2.098542585907207e-05,
"loss": 0.0468,
"step": 4840
},
{
"epoch": 0.61,
"learning_rate": 2.0918572001604494e-05,
"loss": 0.0379,
"step": 4850
},
{
"epoch": 0.61,
"learning_rate": 2.0851718144136918e-05,
"loss": 0.0713,
"step": 4860
},
{
"epoch": 0.61,
"learning_rate": 2.0784864286669342e-05,
"loss": 0.0629,
"step": 4870
},
{
"epoch": 0.61,
"learning_rate": 2.0718010429201763e-05,
"loss": 0.0429,
"step": 4880
},
{
"epoch": 0.61,
"learning_rate": 2.065115657173419e-05,
"loss": 0.0408,
"step": 4890
},
{
"epoch": 0.61,
"learning_rate": 2.0584302714266615e-05,
"loss": 0.0369,
"step": 4900
},
{
"epoch": 0.62,
"learning_rate": 2.0517448856799036e-05,
"loss": 0.0362,
"step": 4910
},
{
"epoch": 0.62,
"learning_rate": 2.0450594999331464e-05,
"loss": 0.0354,
"step": 4920
},
{
"epoch": 0.62,
"learning_rate": 2.0383741141863885e-05,
"loss": 0.0308,
"step": 4930
},
{
"epoch": 0.62,
"learning_rate": 2.031688728439631e-05,
"loss": 0.0465,
"step": 4940
},
{
"epoch": 0.62,
"learning_rate": 2.0250033426928737e-05,
"loss": 0.0451,
"step": 4950
},
{
"epoch": 0.62,
"learning_rate": 2.0183179569461158e-05,
"loss": 0.0187,
"step": 4960
},
{
"epoch": 0.62,
"learning_rate": 2.0116325711993582e-05,
"loss": 0.052,
"step": 4970
},
{
"epoch": 0.62,
"learning_rate": 2.0049471854526006e-05,
"loss": 0.0427,
"step": 4980
},
{
"epoch": 0.63,
"learning_rate": 1.998261799705843e-05,
"loss": 0.0458,
"step": 4990
},
{
"epoch": 0.63,
"learning_rate": 1.9915764139590855e-05,
"loss": 0.0323,
"step": 5000
},
{
"epoch": 0.63,
"learning_rate": 1.984891028212328e-05,
"loss": 0.0265,
"step": 5010
},
{
"epoch": 0.63,
"learning_rate": 1.9782056424655704e-05,
"loss": 0.046,
"step": 5020
},
{
"epoch": 0.63,
"learning_rate": 1.9715202567188128e-05,
"loss": 0.0222,
"step": 5030
},
{
"epoch": 0.63,
"learning_rate": 1.9648348709720552e-05,
"loss": 0.0612,
"step": 5040
},
{
"epoch": 0.63,
"learning_rate": 1.9581494852252977e-05,
"loss": 0.0557,
"step": 5050
},
{
"epoch": 0.63,
"learning_rate": 1.95146409947854e-05,
"loss": 0.0448,
"step": 5060
},
{
"epoch": 0.64,
"learning_rate": 1.9447787137317825e-05,
"loss": 0.0312,
"step": 5070
},
{
"epoch": 0.64,
"learning_rate": 1.938093327985025e-05,
"loss": 0.037,
"step": 5080
},
{
"epoch": 0.64,
"learning_rate": 1.9314079422382674e-05,
"loss": 0.0418,
"step": 5090
},
{
"epoch": 0.64,
"learning_rate": 1.9247225564915098e-05,
"loss": 0.0362,
"step": 5100
},
{
"epoch": 0.64,
"learning_rate": 1.918037170744752e-05,
"loss": 0.0473,
"step": 5110
},
{
"epoch": 0.64,
"learning_rate": 1.9113517849979943e-05,
"loss": 0.0587,
"step": 5120
},
{
"epoch": 0.64,
"learning_rate": 1.904666399251237e-05,
"loss": 0.039,
"step": 5130
},
{
"epoch": 0.64,
"learning_rate": 1.8979810135044792e-05,
"loss": 0.0611,
"step": 5140
},
{
"epoch": 0.65,
"learning_rate": 1.8912956277577216e-05,
"loss": 0.0516,
"step": 5150
},
{
"epoch": 0.65,
"learning_rate": 1.884610242010964e-05,
"loss": 0.0362,
"step": 5160
},
{
"epoch": 0.65,
"learning_rate": 1.8779248562642065e-05,
"loss": 0.0436,
"step": 5170
},
{
"epoch": 0.65,
"learning_rate": 1.871239470517449e-05,
"loss": 0.0346,
"step": 5180
},
{
"epoch": 0.65,
"learning_rate": 1.8645540847706913e-05,
"loss": 0.0366,
"step": 5190
},
{
"epoch": 0.65,
"learning_rate": 1.8578686990239338e-05,
"loss": 0.0281,
"step": 5200
},
{
"epoch": 0.65,
"learning_rate": 1.8511833132771762e-05,
"loss": 0.0262,
"step": 5210
},
{
"epoch": 0.65,
"learning_rate": 1.8444979275304186e-05,
"loss": 0.0686,
"step": 5220
},
{
"epoch": 0.66,
"learning_rate": 1.837812541783661e-05,
"loss": 0.0257,
"step": 5230
},
{
"epoch": 0.66,
"learning_rate": 1.8311271560369035e-05,
"loss": 0.0298,
"step": 5240
},
{
"epoch": 0.66,
"learning_rate": 1.824441770290146e-05,
"loss": 0.0454,
"step": 5250
},
{
"epoch": 0.66,
"learning_rate": 1.817756384543388e-05,
"loss": 0.0373,
"step": 5260
},
{
"epoch": 0.66,
"learning_rate": 1.8110709987966308e-05,
"loss": 0.042,
"step": 5270
},
{
"epoch": 0.66,
"learning_rate": 1.8043856130498732e-05,
"loss": 0.0522,
"step": 5280
},
{
"epoch": 0.66,
"learning_rate": 1.7977002273031153e-05,
"loss": 0.0364,
"step": 5290
},
{
"epoch": 0.66,
"learning_rate": 1.791014841556358e-05,
"loss": 0.0548,
"step": 5300
},
{
"epoch": 0.67,
"learning_rate": 1.7843294558096002e-05,
"loss": 0.0292,
"step": 5310
},
{
"epoch": 0.67,
"learning_rate": 1.7776440700628426e-05,
"loss": 0.0331,
"step": 5320
},
{
"epoch": 0.67,
"learning_rate": 1.7709586843160854e-05,
"loss": 0.037,
"step": 5330
},
{
"epoch": 0.67,
"learning_rate": 1.7642732985693275e-05,
"loss": 0.0227,
"step": 5340
},
{
"epoch": 0.67,
"learning_rate": 1.75758791282257e-05,
"loss": 0.0362,
"step": 5350
},
{
"epoch": 0.67,
"learning_rate": 1.7509025270758123e-05,
"loss": 0.044,
"step": 5360
},
{
"epoch": 0.67,
"learning_rate": 1.7442171413290548e-05,
"loss": 0.0458,
"step": 5370
},
{
"epoch": 0.67,
"learning_rate": 1.7375317555822972e-05,
"loss": 0.0301,
"step": 5380
},
{
"epoch": 0.68,
"learning_rate": 1.7308463698355396e-05,
"loss": 0.0474,
"step": 5390
},
{
"epoch": 0.68,
"learning_rate": 1.724160984088782e-05,
"loss": 0.0314,
"step": 5400
},
{
"epoch": 0.68,
"learning_rate": 1.7174755983420245e-05,
"loss": 0.0613,
"step": 5410
},
{
"epoch": 0.68,
"learning_rate": 1.710790212595267e-05,
"loss": 0.0323,
"step": 5420
},
{
"epoch": 0.68,
"learning_rate": 1.7041048268485093e-05,
"loss": 0.05,
"step": 5430
},
{
"epoch": 0.68,
"learning_rate": 1.6974194411017518e-05,
"loss": 0.0334,
"step": 5440
},
{
"epoch": 0.68,
"learning_rate": 1.6907340553549942e-05,
"loss": 0.0488,
"step": 5450
},
{
"epoch": 0.68,
"learning_rate": 1.6840486696082363e-05,
"loss": 0.0313,
"step": 5460
},
{
"epoch": 0.69,
"learning_rate": 1.677363283861479e-05,
"loss": 0.045,
"step": 5470
},
{
"epoch": 0.69,
"learning_rate": 1.6706778981147215e-05,
"loss": 0.068,
"step": 5480
},
{
"epoch": 0.69,
"learning_rate": 1.6639925123679636e-05,
"loss": 0.0409,
"step": 5490
},
{
"epoch": 0.69,
"learning_rate": 1.6573071266212064e-05,
"loss": 0.0402,
"step": 5500
},
{
"epoch": 0.69,
"learning_rate": 1.6506217408744484e-05,
"loss": 0.0411,
"step": 5510
},
{
"epoch": 0.69,
"learning_rate": 1.643936355127691e-05,
"loss": 0.0608,
"step": 5520
},
{
"epoch": 0.69,
"learning_rate": 1.6372509693809336e-05,
"loss": 0.0374,
"step": 5530
},
{
"epoch": 0.69,
"learning_rate": 1.6305655836341757e-05,
"loss": 0.0521,
"step": 5540
},
{
"epoch": 0.7,
"learning_rate": 1.623880197887418e-05,
"loss": 0.0253,
"step": 5550
},
{
"epoch": 0.7,
"learning_rate": 1.6171948121406606e-05,
"loss": 0.0399,
"step": 5560
},
{
"epoch": 0.7,
"learning_rate": 1.610509426393903e-05,
"loss": 0.0196,
"step": 5570
},
{
"epoch": 0.7,
"learning_rate": 1.6038240406471455e-05,
"loss": 0.041,
"step": 5580
},
{
"epoch": 0.7,
"learning_rate": 1.597138654900388e-05,
"loss": 0.0209,
"step": 5590
},
{
"epoch": 0.7,
"learning_rate": 1.5904532691536303e-05,
"loss": 0.0408,
"step": 5600
},
{
"epoch": 0.7,
"learning_rate": 1.5837678834068724e-05,
"loss": 0.0419,
"step": 5610
},
{
"epoch": 0.7,
"learning_rate": 1.5770824976601152e-05,
"loss": 0.024,
"step": 5620
},
{
"epoch": 0.71,
"learning_rate": 1.5703971119133576e-05,
"loss": 0.0261,
"step": 5630
},
{
"epoch": 0.71,
"learning_rate": 1.5637117261665997e-05,
"loss": 0.0511,
"step": 5640
},
{
"epoch": 0.71,
"learning_rate": 1.5570263404198425e-05,
"loss": 0.0469,
"step": 5650
},
{
"epoch": 0.71,
"learning_rate": 1.5503409546730846e-05,
"loss": 0.029,
"step": 5660
},
{
"epoch": 0.71,
"learning_rate": 1.543655568926327e-05,
"loss": 0.0364,
"step": 5670
},
{
"epoch": 0.71,
"learning_rate": 1.5369701831795698e-05,
"loss": 0.05,
"step": 5680
},
{
"epoch": 0.71,
"learning_rate": 1.530284797432812e-05,
"loss": 0.0304,
"step": 5690
},
{
"epoch": 0.71,
"learning_rate": 1.5235994116860545e-05,
"loss": 0.0303,
"step": 5700
},
{
"epoch": 0.72,
"learning_rate": 1.5169140259392967e-05,
"loss": 0.0626,
"step": 5710
},
{
"epoch": 0.72,
"learning_rate": 1.5102286401925391e-05,
"loss": 0.0532,
"step": 5720
},
{
"epoch": 0.72,
"learning_rate": 1.5035432544457817e-05,
"loss": 0.0292,
"step": 5730
},
{
"epoch": 0.72,
"learning_rate": 1.496857868699024e-05,
"loss": 0.0319,
"step": 5740
},
{
"epoch": 0.72,
"learning_rate": 1.4901724829522664e-05,
"loss": 0.0505,
"step": 5750
},
{
"epoch": 0.72,
"learning_rate": 1.4834870972055087e-05,
"loss": 0.0361,
"step": 5760
},
{
"epoch": 0.72,
"learning_rate": 1.4768017114587513e-05,
"loss": 0.0574,
"step": 5770
},
{
"epoch": 0.72,
"learning_rate": 1.4701163257119937e-05,
"loss": 0.0465,
"step": 5780
},
{
"epoch": 0.73,
"learning_rate": 1.463430939965236e-05,
"loss": 0.0368,
"step": 5790
},
{
"epoch": 0.73,
"learning_rate": 1.4567455542184786e-05,
"loss": 0.0273,
"step": 5800
},
{
"epoch": 0.73,
"learning_rate": 1.4500601684717209e-05,
"loss": 0.0281,
"step": 5810
},
{
"epoch": 0.73,
"learning_rate": 1.4433747827249633e-05,
"loss": 0.0449,
"step": 5820
},
{
"epoch": 0.73,
"learning_rate": 1.4366893969782059e-05,
"loss": 0.0325,
"step": 5830
},
{
"epoch": 0.73,
"learning_rate": 1.4300040112314481e-05,
"loss": 0.0382,
"step": 5840
},
{
"epoch": 0.73,
"learning_rate": 1.4233186254846906e-05,
"loss": 0.0376,
"step": 5850
},
{
"epoch": 0.73,
"learning_rate": 1.4166332397379328e-05,
"loss": 0.0325,
"step": 5860
},
{
"epoch": 0.74,
"learning_rate": 1.4099478539911753e-05,
"loss": 0.0423,
"step": 5870
},
{
"epoch": 0.74,
"learning_rate": 1.4032624682444179e-05,
"loss": 0.0333,
"step": 5880
},
{
"epoch": 0.74,
"learning_rate": 1.3965770824976601e-05,
"loss": 0.0266,
"step": 5890
},
{
"epoch": 0.74,
"learning_rate": 1.3898916967509026e-05,
"loss": 0.0388,
"step": 5900
},
{
"epoch": 0.74,
"learning_rate": 1.3832063110041448e-05,
"loss": 0.0321,
"step": 5910
},
{
"epoch": 0.74,
"learning_rate": 1.3765209252573874e-05,
"loss": 0.0298,
"step": 5920
},
{
"epoch": 0.74,
"learning_rate": 1.3698355395106298e-05,
"loss": 0.0443,
"step": 5930
},
{
"epoch": 0.74,
"learning_rate": 1.3631501537638721e-05,
"loss": 0.026,
"step": 5940
},
{
"epoch": 0.75,
"learning_rate": 1.3564647680171147e-05,
"loss": 0.0296,
"step": 5950
},
{
"epoch": 0.75,
"learning_rate": 1.349779382270357e-05,
"loss": 0.0472,
"step": 5960
},
{
"epoch": 0.75,
"learning_rate": 1.3430939965235994e-05,
"loss": 0.031,
"step": 5970
},
{
"epoch": 0.75,
"learning_rate": 1.336408610776842e-05,
"loss": 0.0501,
"step": 5980
},
{
"epoch": 0.75,
"learning_rate": 1.3297232250300843e-05,
"loss": 0.0452,
"step": 5990
},
{
"epoch": 0.75,
"learning_rate": 1.3230378392833267e-05,
"loss": 0.0374,
"step": 6000
},
{
"epoch": 0.75,
"learning_rate": 1.316352453536569e-05,
"loss": 0.0368,
"step": 6010
},
{
"epoch": 0.75,
"learning_rate": 1.3096670677898116e-05,
"loss": 0.0329,
"step": 6020
},
{
"epoch": 0.76,
"learning_rate": 1.302981682043054e-05,
"loss": 0.024,
"step": 6030
},
{
"epoch": 0.76,
"learning_rate": 1.2962962962962962e-05,
"loss": 0.0291,
"step": 6040
},
{
"epoch": 0.76,
"learning_rate": 1.2896109105495388e-05,
"loss": 0.0576,
"step": 6050
},
{
"epoch": 0.76,
"learning_rate": 1.2829255248027811e-05,
"loss": 0.0353,
"step": 6060
},
{
"epoch": 0.76,
"learning_rate": 1.2762401390560235e-05,
"loss": 0.0228,
"step": 6070
},
{
"epoch": 0.76,
"learning_rate": 1.2695547533092661e-05,
"loss": 0.0425,
"step": 6080
},
{
"epoch": 0.76,
"learning_rate": 1.2628693675625084e-05,
"loss": 0.0274,
"step": 6090
},
{
"epoch": 0.76,
"learning_rate": 1.2561839818157508e-05,
"loss": 0.0352,
"step": 6100
},
{
"epoch": 0.77,
"learning_rate": 1.2494985960689933e-05,
"loss": 0.0261,
"step": 6110
},
{
"epoch": 0.77,
"learning_rate": 1.2428132103222357e-05,
"loss": 0.0299,
"step": 6120
},
{
"epoch": 0.77,
"learning_rate": 1.236127824575478e-05,
"loss": 0.0283,
"step": 6130
},
{
"epoch": 0.77,
"learning_rate": 1.2294424388287206e-05,
"loss": 0.0364,
"step": 6140
},
{
"epoch": 0.77,
"learning_rate": 1.222757053081963e-05,
"loss": 0.0488,
"step": 6150
},
{
"epoch": 0.77,
"learning_rate": 1.2160716673352052e-05,
"loss": 0.0447,
"step": 6160
},
{
"epoch": 0.77,
"learning_rate": 1.2093862815884477e-05,
"loss": 0.0486,
"step": 6170
},
{
"epoch": 0.77,
"learning_rate": 1.2027008958416901e-05,
"loss": 0.0478,
"step": 6180
},
{
"epoch": 0.78,
"learning_rate": 1.1960155100949325e-05,
"loss": 0.0663,
"step": 6190
},
{
"epoch": 0.78,
"learning_rate": 1.189330124348175e-05,
"loss": 0.0213,
"step": 6200
},
{
"epoch": 0.78,
"learning_rate": 1.1826447386014174e-05,
"loss": 0.0611,
"step": 6210
},
{
"epoch": 0.78,
"learning_rate": 1.1759593528546598e-05,
"loss": 0.0648,
"step": 6220
},
{
"epoch": 0.78,
"learning_rate": 1.1692739671079021e-05,
"loss": 0.0485,
"step": 6230
},
{
"epoch": 0.78,
"learning_rate": 1.1625885813611447e-05,
"loss": 0.029,
"step": 6240
},
{
"epoch": 0.78,
"learning_rate": 1.1559031956143871e-05,
"loss": 0.0242,
"step": 6250
},
{
"epoch": 0.78,
"learning_rate": 1.1492178098676294e-05,
"loss": 0.0401,
"step": 6260
},
{
"epoch": 0.79,
"learning_rate": 1.1425324241208718e-05,
"loss": 0.0677,
"step": 6270
},
{
"epoch": 0.79,
"learning_rate": 1.1358470383741142e-05,
"loss": 0.0189,
"step": 6280
},
{
"epoch": 0.79,
"learning_rate": 1.1291616526273567e-05,
"loss": 0.0302,
"step": 6290
},
{
"epoch": 0.79,
"learning_rate": 1.1224762668805991e-05,
"loss": 0.0405,
"step": 6300
},
{
"epoch": 0.79,
"learning_rate": 1.1157908811338415e-05,
"loss": 0.059,
"step": 6310
},
{
"epoch": 0.79,
"learning_rate": 1.1091054953870838e-05,
"loss": 0.0259,
"step": 6320
},
{
"epoch": 0.79,
"learning_rate": 1.1024201096403262e-05,
"loss": 0.0365,
"step": 6330
},
{
"epoch": 0.79,
"learning_rate": 1.0957347238935688e-05,
"loss": 0.0304,
"step": 6340
},
{
"epoch": 0.8,
"learning_rate": 1.089049338146811e-05,
"loss": 0.0412,
"step": 6350
},
{
"epoch": 0.8,
"learning_rate": 1.0823639524000535e-05,
"loss": 0.0424,
"step": 6360
},
{
"epoch": 0.8,
"learning_rate": 1.075678566653296e-05,
"loss": 0.0343,
"step": 6370
},
{
"epoch": 0.8,
"learning_rate": 1.0689931809065384e-05,
"loss": 0.0444,
"step": 6380
},
{
"epoch": 0.8,
"learning_rate": 1.0623077951597808e-05,
"loss": 0.0412,
"step": 6390
},
{
"epoch": 0.8,
"learning_rate": 1.0556224094130232e-05,
"loss": 0.0346,
"step": 6400
},
{
"epoch": 0.8,
"learning_rate": 1.0489370236662657e-05,
"loss": 0.0568,
"step": 6410
},
{
"epoch": 0.8,
"learning_rate": 1.042251637919508e-05,
"loss": 0.0407,
"step": 6420
},
{
"epoch": 0.81,
"learning_rate": 1.0355662521727504e-05,
"loss": 0.0439,
"step": 6430
},
{
"epoch": 0.81,
"learning_rate": 1.028880866425993e-05,
"loss": 0.0605,
"step": 6440
},
{
"epoch": 0.81,
"learning_rate": 1.0221954806792352e-05,
"loss": 0.0544,
"step": 6450
},
{
"epoch": 0.81,
"learning_rate": 1.0155100949324776e-05,
"loss": 0.0437,
"step": 6460
},
{
"epoch": 0.81,
"learning_rate": 1.00882470918572e-05,
"loss": 0.0369,
"step": 6470
},
{
"epoch": 0.81,
"learning_rate": 1.0021393234389623e-05,
"loss": 0.0342,
"step": 6480
},
{
"epoch": 0.81,
"learning_rate": 9.95453937692205e-06,
"loss": 0.0448,
"step": 6490
},
{
"epoch": 0.81,
"learning_rate": 9.887685519454474e-06,
"loss": 0.0538,
"step": 6500
},
{
"epoch": 0.82,
"learning_rate": 9.820831661986896e-06,
"loss": 0.0244,
"step": 6510
},
{
"epoch": 0.82,
"learning_rate": 9.75397780451932e-06,
"loss": 0.0824,
"step": 6520
},
{
"epoch": 0.82,
"learning_rate": 9.687123947051745e-06,
"loss": 0.0502,
"step": 6530
},
{
"epoch": 0.82,
"learning_rate": 9.62027008958417e-06,
"loss": 0.0338,
"step": 6540
},
{
"epoch": 0.82,
"learning_rate": 9.553416232116594e-06,
"loss": 0.0419,
"step": 6550
},
{
"epoch": 0.82,
"learning_rate": 9.486562374649018e-06,
"loss": 0.0576,
"step": 6560
},
{
"epoch": 0.82,
"learning_rate": 9.419708517181442e-06,
"loss": 0.039,
"step": 6570
},
{
"epoch": 0.82,
"learning_rate": 9.352854659713865e-06,
"loss": 0.0662,
"step": 6580
},
{
"epoch": 0.83,
"learning_rate": 9.28600080224629e-06,
"loss": 0.0398,
"step": 6590
},
{
"epoch": 0.83,
"learning_rate": 9.219146944778715e-06,
"loss": 0.0349,
"step": 6600
},
{
"epoch": 0.83,
"learning_rate": 9.152293087311138e-06,
"loss": 0.0314,
"step": 6610
},
{
"epoch": 0.83,
"learning_rate": 9.085439229843562e-06,
"loss": 0.0388,
"step": 6620
},
{
"epoch": 0.83,
"learning_rate": 9.018585372375988e-06,
"loss": 0.0285,
"step": 6630
},
{
"epoch": 0.83,
"learning_rate": 8.95173151490841e-06,
"loss": 0.0473,
"step": 6640
},
{
"epoch": 0.83,
"learning_rate": 8.884877657440835e-06,
"loss": 0.0462,
"step": 6650
},
{
"epoch": 0.83,
"learning_rate": 8.81802379997326e-06,
"loss": 0.043,
"step": 6660
},
{
"epoch": 0.84,
"learning_rate": 8.751169942505682e-06,
"loss": 0.0387,
"step": 6670
},
{
"epoch": 0.84,
"learning_rate": 8.684316085038108e-06,
"loss": 0.0545,
"step": 6680
},
{
"epoch": 0.84,
"learning_rate": 8.617462227570532e-06,
"loss": 0.0785,
"step": 6690
},
{
"epoch": 0.84,
"learning_rate": 8.550608370102955e-06,
"loss": 0.0413,
"step": 6700
},
{
"epoch": 0.84,
"learning_rate": 8.483754512635379e-06,
"loss": 0.0433,
"step": 6710
},
{
"epoch": 0.84,
"learning_rate": 8.416900655167803e-06,
"loss": 0.0467,
"step": 6720
},
{
"epoch": 0.84,
"learning_rate": 8.350046797700228e-06,
"loss": 0.0201,
"step": 6730
},
{
"epoch": 0.84,
"learning_rate": 8.283192940232652e-06,
"loss": 0.0233,
"step": 6740
},
{
"epoch": 0.85,
"learning_rate": 8.216339082765076e-06,
"loss": 0.0421,
"step": 6750
},
{
"epoch": 0.85,
"learning_rate": 8.1494852252975e-06,
"loss": 0.0279,
"step": 6760
},
{
"epoch": 0.85,
"learning_rate": 8.082631367829923e-06,
"loss": 0.0358,
"step": 6770
},
{
"epoch": 0.85,
"learning_rate": 8.01577751036235e-06,
"loss": 0.0391,
"step": 6780
},
{
"epoch": 0.85,
"learning_rate": 7.948923652894773e-06,
"loss": 0.0345,
"step": 6790
},
{
"epoch": 0.85,
"learning_rate": 7.882069795427196e-06,
"loss": 0.0546,
"step": 6800
},
{
"epoch": 0.85,
"learning_rate": 7.81521593795962e-06,
"loss": 0.0252,
"step": 6810
},
{
"epoch": 0.85,
"learning_rate": 7.748362080492045e-06,
"loss": 0.0382,
"step": 6820
},
{
"epoch": 0.86,
"learning_rate": 7.681508223024469e-06,
"loss": 0.0392,
"step": 6830
},
{
"epoch": 0.86,
"learning_rate": 7.614654365556893e-06,
"loss": 0.0318,
"step": 6840
},
{
"epoch": 0.86,
"learning_rate": 7.547800508089317e-06,
"loss": 0.0471,
"step": 6850
},
{
"epoch": 0.86,
"learning_rate": 7.480946650621741e-06,
"loss": 0.0433,
"step": 6860
},
{
"epoch": 0.86,
"learning_rate": 7.4140927931541645e-06,
"loss": 0.0531,
"step": 6870
},
{
"epoch": 0.86,
"learning_rate": 7.34723893568659e-06,
"loss": 0.0278,
"step": 6880
},
{
"epoch": 0.86,
"learning_rate": 7.280385078219014e-06,
"loss": 0.0209,
"step": 6890
},
{
"epoch": 0.86,
"learning_rate": 7.2135312207514375e-06,
"loss": 0.0532,
"step": 6900
},
{
"epoch": 0.87,
"learning_rate": 7.146677363283862e-06,
"loss": 0.0439,
"step": 6910
},
{
"epoch": 0.87,
"learning_rate": 7.079823505816285e-06,
"loss": 0.04,
"step": 6920
},
{
"epoch": 0.87,
"learning_rate": 7.01296964834871e-06,
"loss": 0.0464,
"step": 6930
},
{
"epoch": 0.87,
"learning_rate": 6.946115790881135e-06,
"loss": 0.032,
"step": 6940
},
{
"epoch": 0.87,
"learning_rate": 6.879261933413558e-06,
"loss": 0.0284,
"step": 6950
},
{
"epoch": 0.87,
"learning_rate": 6.8124080759459824e-06,
"loss": 0.0267,
"step": 6960
},
{
"epoch": 0.87,
"learning_rate": 6.745554218478406e-06,
"loss": 0.0363,
"step": 6970
},
{
"epoch": 0.87,
"learning_rate": 6.678700361010831e-06,
"loss": 0.0317,
"step": 6980
},
{
"epoch": 0.88,
"learning_rate": 6.611846503543255e-06,
"loss": 0.0336,
"step": 6990
},
{
"epoch": 0.88,
"learning_rate": 6.544992646075679e-06,
"loss": 0.028,
"step": 7000
},
{
"epoch": 0.88,
"learning_rate": 6.478138788608102e-06,
"loss": 0.0347,
"step": 7010
},
{
"epoch": 0.88,
"learning_rate": 6.411284931140527e-06,
"loss": 0.0359,
"step": 7020
},
{
"epoch": 0.88,
"learning_rate": 6.344431073672952e-06,
"loss": 0.0381,
"step": 7030
},
{
"epoch": 0.88,
"learning_rate": 6.277577216205375e-06,
"loss": 0.0175,
"step": 7040
},
{
"epoch": 0.88,
"learning_rate": 6.2107233587377995e-06,
"loss": 0.0453,
"step": 7050
},
{
"epoch": 0.88,
"learning_rate": 6.143869501270223e-06,
"loss": 0.038,
"step": 7060
},
{
"epoch": 0.89,
"learning_rate": 6.077015643802648e-06,
"loss": 0.029,
"step": 7070
},
{
"epoch": 0.89,
"learning_rate": 6.0101617863350716e-06,
"loss": 0.0458,
"step": 7080
},
{
"epoch": 0.89,
"learning_rate": 5.943307928867496e-06,
"loss": 0.039,
"step": 7090
},
{
"epoch": 0.89,
"learning_rate": 5.87645407139992e-06,
"loss": 0.0384,
"step": 7100
},
{
"epoch": 0.89,
"learning_rate": 5.8096002139323445e-06,
"loss": 0.0621,
"step": 7110
},
{
"epoch": 0.89,
"learning_rate": 5.742746356464768e-06,
"loss": 0.0291,
"step": 7120
},
{
"epoch": 0.89,
"learning_rate": 5.675892498997192e-06,
"loss": 0.054,
"step": 7130
},
{
"epoch": 0.89,
"learning_rate": 5.6090386415296165e-06,
"loss": 0.0325,
"step": 7140
},
{
"epoch": 0.9,
"learning_rate": 5.542184784062041e-06,
"loss": 0.0417,
"step": 7150
},
{
"epoch": 0.9,
"learning_rate": 5.475330926594465e-06,
"loss": 0.0361,
"step": 7160
},
{
"epoch": 0.9,
"learning_rate": 5.408477069126889e-06,
"loss": 0.0531,
"step": 7170
},
{
"epoch": 0.9,
"learning_rate": 5.341623211659313e-06,
"loss": 0.0265,
"step": 7180
},
{
"epoch": 0.9,
"learning_rate": 5.274769354191737e-06,
"loss": 0.0435,
"step": 7190
},
{
"epoch": 0.9,
"learning_rate": 5.2079154967241615e-06,
"loss": 0.0276,
"step": 7200
},
{
"epoch": 0.9,
"learning_rate": 5.141061639256586e-06,
"loss": 0.0264,
"step": 7210
},
{
"epoch": 0.9,
"learning_rate": 5.074207781789009e-06,
"loss": 0.0234,
"step": 7220
},
{
"epoch": 0.91,
"learning_rate": 5.007353924321434e-06,
"loss": 0.0362,
"step": 7230
},
{
"epoch": 0.91,
"learning_rate": 4.940500066853858e-06,
"loss": 0.0304,
"step": 7240
},
{
"epoch": 0.91,
"learning_rate": 4.873646209386281e-06,
"loss": 0.0339,
"step": 7250
},
{
"epoch": 0.91,
"learning_rate": 4.8067923519187065e-06,
"loss": 0.0381,
"step": 7260
},
{
"epoch": 0.91,
"learning_rate": 4.73993849445113e-06,
"loss": 0.0472,
"step": 7270
},
{
"epoch": 0.91,
"learning_rate": 4.673084636983554e-06,
"loss": 0.029,
"step": 7280
},
{
"epoch": 0.91,
"learning_rate": 4.606230779515979e-06,
"loss": 0.0503,
"step": 7290
},
{
"epoch": 0.91,
"learning_rate": 4.539376922048402e-06,
"loss": 0.0446,
"step": 7300
},
{
"epoch": 0.92,
"learning_rate": 4.472523064580827e-06,
"loss": 0.0506,
"step": 7310
},
{
"epoch": 0.92,
"learning_rate": 4.405669207113251e-06,
"loss": 0.0322,
"step": 7320
},
{
"epoch": 0.92,
"learning_rate": 4.338815349645674e-06,
"loss": 0.0477,
"step": 7330
},
{
"epoch": 0.92,
"learning_rate": 4.271961492178099e-06,
"loss": 0.0276,
"step": 7340
},
{
"epoch": 0.92,
"learning_rate": 4.205107634710523e-06,
"loss": 0.0483,
"step": 7350
},
{
"epoch": 0.92,
"learning_rate": 4.138253777242947e-06,
"loss": 0.049,
"step": 7360
},
{
"epoch": 0.92,
"learning_rate": 4.071399919775371e-06,
"loss": 0.0388,
"step": 7370
},
{
"epoch": 0.92,
"learning_rate": 4.004546062307795e-06,
"loss": 0.034,
"step": 7380
},
{
"epoch": 0.93,
"learning_rate": 3.93769220484022e-06,
"loss": 0.0408,
"step": 7390
},
{
"epoch": 0.93,
"learning_rate": 3.870838347372643e-06,
"loss": 0.0448,
"step": 7400
},
{
"epoch": 0.93,
"learning_rate": 3.803984489905068e-06,
"loss": 0.0377,
"step": 7410
},
{
"epoch": 0.93,
"learning_rate": 3.7371306324374916e-06,
"loss": 0.0175,
"step": 7420
},
{
"epoch": 0.93,
"learning_rate": 3.6702767749699155e-06,
"loss": 0.0432,
"step": 7430
},
{
"epoch": 0.93,
"learning_rate": 3.60342291750234e-06,
"loss": 0.0381,
"step": 7440
},
{
"epoch": 0.93,
"learning_rate": 3.536569060034764e-06,
"loss": 0.0425,
"step": 7450
},
{
"epoch": 0.93,
"learning_rate": 3.4697152025671884e-06,
"loss": 0.0432,
"step": 7460
},
{
"epoch": 0.94,
"learning_rate": 3.4028613450996123e-06,
"loss": 0.0355,
"step": 7470
},
{
"epoch": 0.94,
"learning_rate": 3.336007487632037e-06,
"loss": 0.0802,
"step": 7480
},
{
"epoch": 0.94,
"learning_rate": 3.269153630164461e-06,
"loss": 0.0484,
"step": 7490
},
{
"epoch": 0.94,
"learning_rate": 3.2022997726968843e-06,
"loss": 0.0535,
"step": 7500
}
],
"max_steps": 7979,
"num_train_epochs": 1,
"total_flos": 3974305443840000.0,
"trial_name": null,
"trial_params": null
}