method1 / trainer_state.json
tingchih's picture
Upload folder using huggingface_hub
046e646
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.0,
"global_step": 5000,
"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": 2.8375,
"step": 10
},
{
"epoch": 0.01,
"learning_rate": 2.0000000000000003e-06,
"loss": 2.6779,
"step": 20
},
{
"epoch": 0.01,
"learning_rate": 3e-06,
"loss": 2.57,
"step": 30
},
{
"epoch": 0.02,
"learning_rate": 4.000000000000001e-06,
"loss": 2.5656,
"step": 40
},
{
"epoch": 0.02,
"learning_rate": 5e-06,
"loss": 2.3681,
"step": 50
},
{
"epoch": 0.02,
"learning_rate": 6e-06,
"loss": 2.3631,
"step": 60
},
{
"epoch": 0.03,
"learning_rate": 7.000000000000001e-06,
"loss": 2.388,
"step": 70
},
{
"epoch": 0.03,
"learning_rate": 8.000000000000001e-06,
"loss": 2.2585,
"step": 80
},
{
"epoch": 0.04,
"learning_rate": 9e-06,
"loss": 2.3164,
"step": 90
},
{
"epoch": 0.04,
"learning_rate": 1e-05,
"loss": 2.3083,
"step": 100
},
{
"epoch": 0.04,
"learning_rate": 1.1000000000000001e-05,
"loss": 2.2434,
"step": 110
},
{
"epoch": 0.05,
"learning_rate": 1.2e-05,
"loss": 2.2638,
"step": 120
},
{
"epoch": 0.05,
"learning_rate": 1.3000000000000001e-05,
"loss": 2.2401,
"step": 130
},
{
"epoch": 0.06,
"learning_rate": 1.4000000000000001e-05,
"loss": 2.2564,
"step": 140
},
{
"epoch": 0.06,
"learning_rate": 1.5e-05,
"loss": 2.2993,
"step": 150
},
{
"epoch": 0.06,
"learning_rate": 1.6000000000000003e-05,
"loss": 2.3755,
"step": 160
},
{
"epoch": 0.07,
"learning_rate": 1.7000000000000003e-05,
"loss": 2.3019,
"step": 170
},
{
"epoch": 0.07,
"learning_rate": 1.8e-05,
"loss": 2.2788,
"step": 180
},
{
"epoch": 0.08,
"learning_rate": 1.9e-05,
"loss": 2.2605,
"step": 190
},
{
"epoch": 0.08,
"learning_rate": 2e-05,
"loss": 2.2447,
"step": 200
},
{
"epoch": 0.08,
"learning_rate": 2.1e-05,
"loss": 2.2964,
"step": 210
},
{
"epoch": 0.09,
"learning_rate": 2.2000000000000003e-05,
"loss": 2.1982,
"step": 220
},
{
"epoch": 0.09,
"learning_rate": 2.3000000000000003e-05,
"loss": 2.2083,
"step": 230
},
{
"epoch": 0.1,
"learning_rate": 2.4e-05,
"loss": 2.2543,
"step": 240
},
{
"epoch": 0.1,
"learning_rate": 2.5e-05,
"loss": 2.2566,
"step": 250
},
{
"epoch": 0.1,
"learning_rate": 2.6000000000000002e-05,
"loss": 2.1604,
"step": 260
},
{
"epoch": 0.11,
"learning_rate": 2.7000000000000002e-05,
"loss": 2.3027,
"step": 270
},
{
"epoch": 0.11,
"learning_rate": 2.8000000000000003e-05,
"loss": 2.2066,
"step": 280
},
{
"epoch": 0.12,
"learning_rate": 2.9e-05,
"loss": 2.2323,
"step": 290
},
{
"epoch": 0.12,
"learning_rate": 3e-05,
"loss": 2.1687,
"step": 300
},
{
"epoch": 0.12,
"learning_rate": 3.1e-05,
"loss": 2.2966,
"step": 310
},
{
"epoch": 0.13,
"learning_rate": 3.2000000000000005e-05,
"loss": 2.2202,
"step": 320
},
{
"epoch": 0.13,
"learning_rate": 3.3e-05,
"loss": 2.2124,
"step": 330
},
{
"epoch": 0.14,
"learning_rate": 3.4000000000000007e-05,
"loss": 2.1902,
"step": 340
},
{
"epoch": 0.14,
"learning_rate": 3.5e-05,
"loss": 2.2963,
"step": 350
},
{
"epoch": 0.14,
"learning_rate": 3.6e-05,
"loss": 2.181,
"step": 360
},
{
"epoch": 0.15,
"learning_rate": 3.7e-05,
"loss": 2.2098,
"step": 370
},
{
"epoch": 0.15,
"learning_rate": 3.8e-05,
"loss": 2.2389,
"step": 380
},
{
"epoch": 0.16,
"learning_rate": 3.9000000000000006e-05,
"loss": 2.2249,
"step": 390
},
{
"epoch": 0.16,
"learning_rate": 4e-05,
"loss": 2.1545,
"step": 400
},
{
"epoch": 0.16,
"learning_rate": 4.1e-05,
"loss": 2.2302,
"step": 410
},
{
"epoch": 0.17,
"learning_rate": 4.2e-05,
"loss": 2.1801,
"step": 420
},
{
"epoch": 0.17,
"learning_rate": 4.3e-05,
"loss": 2.2626,
"step": 430
},
{
"epoch": 0.18,
"learning_rate": 4.4000000000000006e-05,
"loss": 2.2543,
"step": 440
},
{
"epoch": 0.18,
"learning_rate": 4.5e-05,
"loss": 2.1455,
"step": 450
},
{
"epoch": 0.18,
"learning_rate": 4.600000000000001e-05,
"loss": 2.1994,
"step": 460
},
{
"epoch": 0.19,
"learning_rate": 4.7e-05,
"loss": 2.2642,
"step": 470
},
{
"epoch": 0.19,
"learning_rate": 4.8e-05,
"loss": 2.2493,
"step": 480
},
{
"epoch": 0.2,
"learning_rate": 4.9e-05,
"loss": 2.2332,
"step": 490
},
{
"epoch": 0.2,
"learning_rate": 5e-05,
"loss": 2.1859,
"step": 500
},
{
"epoch": 0.2,
"learning_rate": 4.9888888888888894e-05,
"loss": 2.205,
"step": 510
},
{
"epoch": 0.21,
"learning_rate": 4.977777777777778e-05,
"loss": 2.3042,
"step": 520
},
{
"epoch": 0.21,
"learning_rate": 4.966666666666667e-05,
"loss": 2.3318,
"step": 530
},
{
"epoch": 0.22,
"learning_rate": 4.955555555555556e-05,
"loss": 2.249,
"step": 540
},
{
"epoch": 0.22,
"learning_rate": 4.9444444444444446e-05,
"loss": 2.2936,
"step": 550
},
{
"epoch": 0.22,
"learning_rate": 4.933333333333334e-05,
"loss": 2.189,
"step": 560
},
{
"epoch": 0.23,
"learning_rate": 4.922222222222222e-05,
"loss": 2.2387,
"step": 570
},
{
"epoch": 0.23,
"learning_rate": 4.9111111111111114e-05,
"loss": 2.1974,
"step": 580
},
{
"epoch": 0.24,
"learning_rate": 4.9e-05,
"loss": 2.2088,
"step": 590
},
{
"epoch": 0.24,
"learning_rate": 4.888888888888889e-05,
"loss": 2.2264,
"step": 600
},
{
"epoch": 0.24,
"learning_rate": 4.8777777777777775e-05,
"loss": 2.2152,
"step": 610
},
{
"epoch": 0.25,
"learning_rate": 4.866666666666667e-05,
"loss": 2.2981,
"step": 620
},
{
"epoch": 0.25,
"learning_rate": 4.855555555555556e-05,
"loss": 2.1741,
"step": 630
},
{
"epoch": 0.26,
"learning_rate": 4.844444444444445e-05,
"loss": 2.228,
"step": 640
},
{
"epoch": 0.26,
"learning_rate": 4.8333333333333334e-05,
"loss": 2.2239,
"step": 650
},
{
"epoch": 0.26,
"learning_rate": 4.8222222222222225e-05,
"loss": 2.1791,
"step": 660
},
{
"epoch": 0.27,
"learning_rate": 4.811111111111111e-05,
"loss": 2.1563,
"step": 670
},
{
"epoch": 0.27,
"learning_rate": 4.8e-05,
"loss": 2.1668,
"step": 680
},
{
"epoch": 0.28,
"learning_rate": 4.7888888888888886e-05,
"loss": 2.1757,
"step": 690
},
{
"epoch": 0.28,
"learning_rate": 4.7777777777777784e-05,
"loss": 2.0375,
"step": 700
},
{
"epoch": 0.28,
"learning_rate": 4.766666666666667e-05,
"loss": 2.2324,
"step": 710
},
{
"epoch": 0.29,
"learning_rate": 4.755555555555556e-05,
"loss": 2.2474,
"step": 720
},
{
"epoch": 0.29,
"learning_rate": 4.7444444444444445e-05,
"loss": 2.2136,
"step": 730
},
{
"epoch": 0.3,
"learning_rate": 4.7333333333333336e-05,
"loss": 2.2169,
"step": 740
},
{
"epoch": 0.3,
"learning_rate": 4.722222222222222e-05,
"loss": 2.2344,
"step": 750
},
{
"epoch": 0.3,
"learning_rate": 4.711111111111111e-05,
"loss": 2.2769,
"step": 760
},
{
"epoch": 0.31,
"learning_rate": 4.7e-05,
"loss": 2.1766,
"step": 770
},
{
"epoch": 0.31,
"learning_rate": 4.6888888888888895e-05,
"loss": 2.287,
"step": 780
},
{
"epoch": 0.32,
"learning_rate": 4.677777777777778e-05,
"loss": 2.2396,
"step": 790
},
{
"epoch": 0.32,
"learning_rate": 4.666666666666667e-05,
"loss": 2.2257,
"step": 800
},
{
"epoch": 0.32,
"learning_rate": 4.6555555555555556e-05,
"loss": 2.1978,
"step": 810
},
{
"epoch": 0.33,
"learning_rate": 4.644444444444445e-05,
"loss": 2.1719,
"step": 820
},
{
"epoch": 0.33,
"learning_rate": 4.633333333333333e-05,
"loss": 2.2309,
"step": 830
},
{
"epoch": 0.34,
"learning_rate": 4.6222222222222224e-05,
"loss": 2.0824,
"step": 840
},
{
"epoch": 0.34,
"learning_rate": 4.6111111111111115e-05,
"loss": 2.2039,
"step": 850
},
{
"epoch": 0.34,
"learning_rate": 4.600000000000001e-05,
"loss": 2.1636,
"step": 860
},
{
"epoch": 0.35,
"learning_rate": 4.588888888888889e-05,
"loss": 2.2237,
"step": 870
},
{
"epoch": 0.35,
"learning_rate": 4.577777777777778e-05,
"loss": 2.2943,
"step": 880
},
{
"epoch": 0.36,
"learning_rate": 4.566666666666667e-05,
"loss": 2.2324,
"step": 890
},
{
"epoch": 0.36,
"learning_rate": 4.555555555555556e-05,
"loss": 2.2258,
"step": 900
},
{
"epoch": 0.36,
"learning_rate": 4.5444444444444444e-05,
"loss": 2.1927,
"step": 910
},
{
"epoch": 0.37,
"learning_rate": 4.5333333333333335e-05,
"loss": 2.2067,
"step": 920
},
{
"epoch": 0.37,
"learning_rate": 4.522222222222223e-05,
"loss": 2.2127,
"step": 930
},
{
"epoch": 0.38,
"learning_rate": 4.511111111111112e-05,
"loss": 2.1303,
"step": 940
},
{
"epoch": 0.38,
"learning_rate": 4.5e-05,
"loss": 2.2569,
"step": 950
},
{
"epoch": 0.38,
"learning_rate": 4.4888888888888894e-05,
"loss": 2.0968,
"step": 960
},
{
"epoch": 0.39,
"learning_rate": 4.477777777777778e-05,
"loss": 2.1282,
"step": 970
},
{
"epoch": 0.39,
"learning_rate": 4.466666666666667e-05,
"loss": 2.1771,
"step": 980
},
{
"epoch": 0.4,
"learning_rate": 4.4555555555555555e-05,
"loss": 2.2062,
"step": 990
},
{
"epoch": 0.4,
"learning_rate": 4.445555555555555e-05,
"loss": 2.1468,
"step": 1000
},
{
"epoch": 0.4,
"learning_rate": 4.4344444444444444e-05,
"loss": 2.1912,
"step": 1010
},
{
"epoch": 0.41,
"learning_rate": 4.4233333333333336e-05,
"loss": 2.2167,
"step": 1020
},
{
"epoch": 0.41,
"learning_rate": 4.412222222222223e-05,
"loss": 2.2232,
"step": 1030
},
{
"epoch": 0.42,
"learning_rate": 4.401111111111111e-05,
"loss": 2.1965,
"step": 1040
},
{
"epoch": 0.42,
"learning_rate": 4.39e-05,
"loss": 2.1658,
"step": 1050
},
{
"epoch": 0.42,
"learning_rate": 4.378888888888889e-05,
"loss": 2.3051,
"step": 1060
},
{
"epoch": 0.43,
"learning_rate": 4.367777777777778e-05,
"loss": 2.1915,
"step": 1070
},
{
"epoch": 0.43,
"learning_rate": 4.3566666666666664e-05,
"loss": 2.1081,
"step": 1080
},
{
"epoch": 0.44,
"learning_rate": 4.3455555555555555e-05,
"loss": 2.2553,
"step": 1090
},
{
"epoch": 0.44,
"learning_rate": 4.334444444444445e-05,
"loss": 2.1794,
"step": 1100
},
{
"epoch": 0.44,
"learning_rate": 4.323333333333334e-05,
"loss": 2.1691,
"step": 1110
},
{
"epoch": 0.45,
"learning_rate": 4.312222222222222e-05,
"loss": 2.1764,
"step": 1120
},
{
"epoch": 0.45,
"learning_rate": 4.3011111111111115e-05,
"loss": 2.1462,
"step": 1130
},
{
"epoch": 0.46,
"learning_rate": 4.29e-05,
"loss": 2.1322,
"step": 1140
},
{
"epoch": 0.46,
"learning_rate": 4.278888888888889e-05,
"loss": 2.1283,
"step": 1150
},
{
"epoch": 0.46,
"learning_rate": 4.2677777777777775e-05,
"loss": 2.1366,
"step": 1160
},
{
"epoch": 0.47,
"learning_rate": 4.2566666666666674e-05,
"loss": 2.2162,
"step": 1170
},
{
"epoch": 0.47,
"learning_rate": 4.245555555555556e-05,
"loss": 2.1199,
"step": 1180
},
{
"epoch": 0.48,
"learning_rate": 4.234444444444445e-05,
"loss": 2.206,
"step": 1190
},
{
"epoch": 0.48,
"learning_rate": 4.2233333333333334e-05,
"loss": 2.115,
"step": 1200
},
{
"epoch": 0.48,
"learning_rate": 4.2122222222222226e-05,
"loss": 2.1905,
"step": 1210
},
{
"epoch": 0.49,
"learning_rate": 4.201111111111111e-05,
"loss": 2.1426,
"step": 1220
},
{
"epoch": 0.49,
"learning_rate": 4.19e-05,
"loss": 2.2061,
"step": 1230
},
{
"epoch": 0.5,
"learning_rate": 4.178888888888889e-05,
"loss": 2.2047,
"step": 1240
},
{
"epoch": 0.5,
"learning_rate": 4.1677777777777785e-05,
"loss": 2.2049,
"step": 1250
},
{
"epoch": 0.5,
"learning_rate": 4.156666666666667e-05,
"loss": 2.157,
"step": 1260
},
{
"epoch": 0.51,
"learning_rate": 4.145555555555556e-05,
"loss": 2.1282,
"step": 1270
},
{
"epoch": 0.51,
"learning_rate": 4.1344444444444446e-05,
"loss": 2.1219,
"step": 1280
},
{
"epoch": 0.52,
"learning_rate": 4.123333333333334e-05,
"loss": 2.2332,
"step": 1290
},
{
"epoch": 0.52,
"learning_rate": 4.112222222222222e-05,
"loss": 2.1688,
"step": 1300
},
{
"epoch": 0.52,
"learning_rate": 4.101111111111111e-05,
"loss": 2.0697,
"step": 1310
},
{
"epoch": 0.53,
"learning_rate": 4.09e-05,
"loss": 2.1763,
"step": 1320
},
{
"epoch": 0.53,
"learning_rate": 4.0788888888888896e-05,
"loss": 2.1581,
"step": 1330
},
{
"epoch": 0.54,
"learning_rate": 4.067777777777778e-05,
"loss": 2.1462,
"step": 1340
},
{
"epoch": 0.54,
"learning_rate": 4.056666666666667e-05,
"loss": 2.2394,
"step": 1350
},
{
"epoch": 0.54,
"learning_rate": 4.045555555555556e-05,
"loss": 2.129,
"step": 1360
},
{
"epoch": 0.55,
"learning_rate": 4.034444444444445e-05,
"loss": 2.1997,
"step": 1370
},
{
"epoch": 0.55,
"learning_rate": 4.023333333333333e-05,
"loss": 2.1919,
"step": 1380
},
{
"epoch": 0.56,
"learning_rate": 4.0122222222222225e-05,
"loss": 2.1456,
"step": 1390
},
{
"epoch": 0.56,
"learning_rate": 4.001111111111111e-05,
"loss": 2.1603,
"step": 1400
},
{
"epoch": 0.56,
"learning_rate": 3.99e-05,
"loss": 2.1268,
"step": 1410
},
{
"epoch": 0.57,
"learning_rate": 3.978888888888889e-05,
"loss": 2.1839,
"step": 1420
},
{
"epoch": 0.57,
"learning_rate": 3.9677777777777784e-05,
"loss": 2.1678,
"step": 1430
},
{
"epoch": 0.58,
"learning_rate": 3.956666666666667e-05,
"loss": 2.1552,
"step": 1440
},
{
"epoch": 0.58,
"learning_rate": 3.945555555555556e-05,
"loss": 2.1212,
"step": 1450
},
{
"epoch": 0.58,
"learning_rate": 3.9344444444444445e-05,
"loss": 2.1712,
"step": 1460
},
{
"epoch": 0.59,
"learning_rate": 3.9233333333333336e-05,
"loss": 2.1182,
"step": 1470
},
{
"epoch": 0.59,
"learning_rate": 3.912222222222223e-05,
"loss": 2.0769,
"step": 1480
},
{
"epoch": 0.6,
"learning_rate": 3.901111111111111e-05,
"loss": 2.1975,
"step": 1490
},
{
"epoch": 0.6,
"learning_rate": 3.8900000000000004e-05,
"loss": 2.1386,
"step": 1500
},
{
"epoch": 0.6,
"learning_rate": 3.878888888888889e-05,
"loss": 2.1822,
"step": 1510
},
{
"epoch": 0.61,
"learning_rate": 3.867777777777778e-05,
"loss": 2.1114,
"step": 1520
},
{
"epoch": 0.61,
"learning_rate": 3.8566666666666664e-05,
"loss": 2.0906,
"step": 1530
},
{
"epoch": 0.62,
"learning_rate": 3.8455555555555556e-05,
"loss": 2.2494,
"step": 1540
},
{
"epoch": 0.62,
"learning_rate": 3.834444444444444e-05,
"loss": 2.1981,
"step": 1550
},
{
"epoch": 0.62,
"learning_rate": 3.823333333333334e-05,
"loss": 2.1289,
"step": 1560
},
{
"epoch": 0.63,
"learning_rate": 3.8122222222222224e-05,
"loss": 2.2029,
"step": 1570
},
{
"epoch": 0.63,
"learning_rate": 3.8011111111111115e-05,
"loss": 2.2259,
"step": 1580
},
{
"epoch": 0.64,
"learning_rate": 3.79e-05,
"loss": 2.1545,
"step": 1590
},
{
"epoch": 0.64,
"learning_rate": 3.778888888888889e-05,
"loss": 2.1687,
"step": 1600
},
{
"epoch": 0.64,
"learning_rate": 3.7677777777777776e-05,
"loss": 2.2138,
"step": 1610
},
{
"epoch": 0.65,
"learning_rate": 3.756666666666667e-05,
"loss": 2.1843,
"step": 1620
},
{
"epoch": 0.65,
"learning_rate": 3.745555555555555e-05,
"loss": 2.0983,
"step": 1630
},
{
"epoch": 0.66,
"learning_rate": 3.734444444444445e-05,
"loss": 2.128,
"step": 1640
},
{
"epoch": 0.66,
"learning_rate": 3.7233333333333335e-05,
"loss": 2.0945,
"step": 1650
},
{
"epoch": 0.66,
"learning_rate": 3.7122222222222226e-05,
"loss": 2.2031,
"step": 1660
},
{
"epoch": 0.67,
"learning_rate": 3.701111111111111e-05,
"loss": 2.1492,
"step": 1670
},
{
"epoch": 0.67,
"learning_rate": 3.69e-05,
"loss": 2.1556,
"step": 1680
},
{
"epoch": 0.68,
"learning_rate": 3.678888888888889e-05,
"loss": 2.2145,
"step": 1690
},
{
"epoch": 0.68,
"learning_rate": 3.667777777777778e-05,
"loss": 2.1574,
"step": 1700
},
{
"epoch": 0.68,
"learning_rate": 3.656666666666666e-05,
"loss": 2.1781,
"step": 1710
},
{
"epoch": 0.69,
"learning_rate": 3.645555555555556e-05,
"loss": 2.1396,
"step": 1720
},
{
"epoch": 0.69,
"learning_rate": 3.6344444444444446e-05,
"loss": 2.1376,
"step": 1730
},
{
"epoch": 0.7,
"learning_rate": 3.623333333333334e-05,
"loss": 2.1554,
"step": 1740
},
{
"epoch": 0.7,
"learning_rate": 3.612222222222222e-05,
"loss": 2.162,
"step": 1750
},
{
"epoch": 0.7,
"learning_rate": 3.6011111111111114e-05,
"loss": 2.1683,
"step": 1760
},
{
"epoch": 0.71,
"learning_rate": 3.59e-05,
"loss": 2.0821,
"step": 1770
},
{
"epoch": 0.71,
"learning_rate": 3.578888888888889e-05,
"loss": 2.189,
"step": 1780
},
{
"epoch": 0.72,
"learning_rate": 3.567777777777778e-05,
"loss": 2.193,
"step": 1790
},
{
"epoch": 0.72,
"learning_rate": 3.556666666666667e-05,
"loss": 2.1326,
"step": 1800
},
{
"epoch": 0.72,
"learning_rate": 3.545555555555556e-05,
"loss": 2.0713,
"step": 1810
},
{
"epoch": 0.73,
"learning_rate": 3.534444444444445e-05,
"loss": 2.2754,
"step": 1820
},
{
"epoch": 0.73,
"learning_rate": 3.5233333333333334e-05,
"loss": 2.0774,
"step": 1830
},
{
"epoch": 0.74,
"learning_rate": 3.5122222222222225e-05,
"loss": 2.1025,
"step": 1840
},
{
"epoch": 0.74,
"learning_rate": 3.501111111111111e-05,
"loss": 2.0258,
"step": 1850
},
{
"epoch": 0.74,
"learning_rate": 3.49e-05,
"loss": 2.1155,
"step": 1860
},
{
"epoch": 0.75,
"learning_rate": 3.478888888888889e-05,
"loss": 2.1176,
"step": 1870
},
{
"epoch": 0.75,
"learning_rate": 3.4677777777777784e-05,
"loss": 2.1447,
"step": 1880
},
{
"epoch": 0.76,
"learning_rate": 3.456666666666667e-05,
"loss": 2.1083,
"step": 1890
},
{
"epoch": 0.76,
"learning_rate": 3.445555555555556e-05,
"loss": 2.0821,
"step": 1900
},
{
"epoch": 0.76,
"learning_rate": 3.4344444444444445e-05,
"loss": 2.2,
"step": 1910
},
{
"epoch": 0.77,
"learning_rate": 3.4233333333333336e-05,
"loss": 2.1913,
"step": 1920
},
{
"epoch": 0.77,
"learning_rate": 3.412222222222222e-05,
"loss": 2.1596,
"step": 1930
},
{
"epoch": 0.78,
"learning_rate": 3.401111111111111e-05,
"loss": 2.1381,
"step": 1940
},
{
"epoch": 0.78,
"learning_rate": 3.3900000000000004e-05,
"loss": 2.1225,
"step": 1950
},
{
"epoch": 0.78,
"learning_rate": 3.378888888888889e-05,
"loss": 2.1596,
"step": 1960
},
{
"epoch": 0.79,
"learning_rate": 3.367777777777778e-05,
"loss": 2.196,
"step": 1970
},
{
"epoch": 0.79,
"learning_rate": 3.356666666666667e-05,
"loss": 2.1491,
"step": 1980
},
{
"epoch": 0.8,
"learning_rate": 3.3455555555555556e-05,
"loss": 2.143,
"step": 1990
},
{
"epoch": 0.8,
"learning_rate": 3.334444444444445e-05,
"loss": 2.1908,
"step": 2000
},
{
"epoch": 0.8,
"learning_rate": 3.323333333333333e-05,
"loss": 2.224,
"step": 2010
},
{
"epoch": 0.81,
"learning_rate": 3.3122222222222224e-05,
"loss": 2.0906,
"step": 2020
},
{
"epoch": 0.81,
"learning_rate": 3.3011111111111115e-05,
"loss": 2.0705,
"step": 2030
},
{
"epoch": 0.82,
"learning_rate": 3.29e-05,
"loss": 2.1297,
"step": 2040
},
{
"epoch": 0.82,
"learning_rate": 3.278888888888889e-05,
"loss": 2.0697,
"step": 2050
},
{
"epoch": 0.82,
"learning_rate": 3.2677777777777776e-05,
"loss": 2.1413,
"step": 2060
},
{
"epoch": 0.83,
"learning_rate": 3.256666666666667e-05,
"loss": 2.1995,
"step": 2070
},
{
"epoch": 0.83,
"learning_rate": 3.245555555555555e-05,
"loss": 2.2242,
"step": 2080
},
{
"epoch": 0.84,
"learning_rate": 3.2344444444444444e-05,
"loss": 2.1759,
"step": 2090
},
{
"epoch": 0.84,
"learning_rate": 3.2233333333333335e-05,
"loss": 2.1406,
"step": 2100
},
{
"epoch": 0.84,
"learning_rate": 3.212222222222223e-05,
"loss": 2.1722,
"step": 2110
},
{
"epoch": 0.85,
"learning_rate": 3.201111111111111e-05,
"loss": 2.1181,
"step": 2120
},
{
"epoch": 0.85,
"learning_rate": 3.19e-05,
"loss": 2.0824,
"step": 2130
},
{
"epoch": 0.86,
"learning_rate": 3.178888888888889e-05,
"loss": 2.1826,
"step": 2140
},
{
"epoch": 0.86,
"learning_rate": 3.167777777777778e-05,
"loss": 2.0846,
"step": 2150
},
{
"epoch": 0.86,
"learning_rate": 3.1566666666666664e-05,
"loss": 2.1772,
"step": 2160
},
{
"epoch": 0.87,
"learning_rate": 3.1455555555555555e-05,
"loss": 2.1471,
"step": 2170
},
{
"epoch": 0.87,
"learning_rate": 3.134444444444445e-05,
"loss": 2.1125,
"step": 2180
},
{
"epoch": 0.88,
"learning_rate": 3.123333333333334e-05,
"loss": 2.055,
"step": 2190
},
{
"epoch": 0.88,
"learning_rate": 3.112222222222222e-05,
"loss": 2.0922,
"step": 2200
},
{
"epoch": 0.88,
"learning_rate": 3.1011111111111114e-05,
"loss": 2.2003,
"step": 2210
},
{
"epoch": 0.89,
"learning_rate": 3.09e-05,
"loss": 2.1425,
"step": 2220
},
{
"epoch": 0.89,
"learning_rate": 3.078888888888889e-05,
"loss": 2.1195,
"step": 2230
},
{
"epoch": 0.9,
"learning_rate": 3.0677777777777775e-05,
"loss": 2.1251,
"step": 2240
},
{
"epoch": 0.9,
"learning_rate": 3.0566666666666667e-05,
"loss": 2.1828,
"step": 2250
},
{
"epoch": 0.9,
"learning_rate": 3.0455555555555558e-05,
"loss": 2.1316,
"step": 2260
},
{
"epoch": 0.91,
"learning_rate": 3.0344444444444446e-05,
"loss": 2.1339,
"step": 2270
},
{
"epoch": 0.91,
"learning_rate": 3.0233333333333334e-05,
"loss": 2.1252,
"step": 2280
},
{
"epoch": 0.92,
"learning_rate": 3.0122222222222226e-05,
"loss": 2.2308,
"step": 2290
},
{
"epoch": 0.92,
"learning_rate": 3.0011111111111114e-05,
"loss": 2.1462,
"step": 2300
},
{
"epoch": 0.92,
"learning_rate": 2.9900000000000002e-05,
"loss": 2.1068,
"step": 2310
},
{
"epoch": 0.93,
"learning_rate": 2.978888888888889e-05,
"loss": 2.0946,
"step": 2320
},
{
"epoch": 0.93,
"learning_rate": 2.9677777777777778e-05,
"loss": 2.0563,
"step": 2330
},
{
"epoch": 0.94,
"learning_rate": 2.956666666666667e-05,
"loss": 2.1172,
"step": 2340
},
{
"epoch": 0.94,
"learning_rate": 2.9455555555555557e-05,
"loss": 2.0987,
"step": 2350
},
{
"epoch": 0.94,
"learning_rate": 2.9344444444444445e-05,
"loss": 2.1107,
"step": 2360
},
{
"epoch": 0.95,
"learning_rate": 2.9233333333333334e-05,
"loss": 2.061,
"step": 2370
},
{
"epoch": 0.95,
"learning_rate": 2.912222222222222e-05,
"loss": 2.0903,
"step": 2380
},
{
"epoch": 0.96,
"learning_rate": 2.901111111111111e-05,
"loss": 2.1508,
"step": 2390
},
{
"epoch": 0.96,
"learning_rate": 2.8899999999999998e-05,
"loss": 2.1378,
"step": 2400
},
{
"epoch": 0.96,
"learning_rate": 2.8788888888888893e-05,
"loss": 2.117,
"step": 2410
},
{
"epoch": 0.97,
"learning_rate": 2.867777777777778e-05,
"loss": 2.1878,
"step": 2420
},
{
"epoch": 0.97,
"learning_rate": 2.856666666666667e-05,
"loss": 2.11,
"step": 2430
},
{
"epoch": 0.98,
"learning_rate": 2.8455555555555557e-05,
"loss": 2.0339,
"step": 2440
},
{
"epoch": 0.98,
"learning_rate": 2.8344444444444445e-05,
"loss": 2.1127,
"step": 2450
},
{
"epoch": 0.98,
"learning_rate": 2.8233333333333333e-05,
"loss": 2.1542,
"step": 2460
},
{
"epoch": 0.99,
"learning_rate": 2.812222222222222e-05,
"loss": 2.1369,
"step": 2470
},
{
"epoch": 0.99,
"learning_rate": 2.801111111111111e-05,
"loss": 2.1702,
"step": 2480
},
{
"epoch": 1.0,
"learning_rate": 2.7900000000000004e-05,
"loss": 2.0552,
"step": 2490
},
{
"epoch": 1.0,
"learning_rate": 2.7788888888888892e-05,
"loss": 2.2228,
"step": 2500
},
{
"epoch": 1.0,
"learning_rate": 2.767777777777778e-05,
"loss": 1.9165,
"step": 2510
},
{
"epoch": 1.01,
"learning_rate": 2.7566666666666668e-05,
"loss": 1.9713,
"step": 2520
},
{
"epoch": 1.01,
"learning_rate": 2.7455555555555556e-05,
"loss": 1.8864,
"step": 2530
},
{
"epoch": 1.02,
"learning_rate": 2.7344444444444444e-05,
"loss": 1.8848,
"step": 2540
},
{
"epoch": 1.02,
"learning_rate": 2.7233333333333332e-05,
"loss": 1.962,
"step": 2550
},
{
"epoch": 1.02,
"learning_rate": 2.712222222222222e-05,
"loss": 1.9398,
"step": 2560
},
{
"epoch": 1.03,
"learning_rate": 2.7011111111111115e-05,
"loss": 1.9051,
"step": 2570
},
{
"epoch": 1.03,
"learning_rate": 2.6900000000000003e-05,
"loss": 2.0315,
"step": 2580
},
{
"epoch": 1.04,
"learning_rate": 2.678888888888889e-05,
"loss": 1.9723,
"step": 2590
},
{
"epoch": 1.04,
"learning_rate": 2.667777777777778e-05,
"loss": 1.9358,
"step": 2600
},
{
"epoch": 1.04,
"learning_rate": 2.6566666666666668e-05,
"loss": 1.9036,
"step": 2610
},
{
"epoch": 1.05,
"learning_rate": 2.6455555555555556e-05,
"loss": 1.902,
"step": 2620
},
{
"epoch": 1.05,
"learning_rate": 2.6344444444444444e-05,
"loss": 1.911,
"step": 2630
},
{
"epoch": 1.06,
"learning_rate": 2.6233333333333332e-05,
"loss": 1.9689,
"step": 2640
},
{
"epoch": 1.06,
"learning_rate": 2.6122222222222227e-05,
"loss": 1.9162,
"step": 2650
},
{
"epoch": 1.06,
"learning_rate": 2.6011111111111115e-05,
"loss": 1.8957,
"step": 2660
},
{
"epoch": 1.07,
"learning_rate": 2.5900000000000003e-05,
"loss": 1.9308,
"step": 2670
},
{
"epoch": 1.07,
"learning_rate": 2.578888888888889e-05,
"loss": 1.9296,
"step": 2680
},
{
"epoch": 1.08,
"learning_rate": 2.567777777777778e-05,
"loss": 1.8809,
"step": 2690
},
{
"epoch": 1.08,
"learning_rate": 2.5566666666666667e-05,
"loss": 1.8844,
"step": 2700
},
{
"epoch": 1.08,
"learning_rate": 2.5455555555555555e-05,
"loss": 1.9367,
"step": 2710
},
{
"epoch": 1.09,
"learning_rate": 2.534444444444445e-05,
"loss": 1.9344,
"step": 2720
},
{
"epoch": 1.09,
"learning_rate": 2.5233333333333338e-05,
"loss": 1.897,
"step": 2730
},
{
"epoch": 1.1,
"learning_rate": 2.5122222222222226e-05,
"loss": 1.8671,
"step": 2740
},
{
"epoch": 1.1,
"learning_rate": 2.5011111111111114e-05,
"loss": 1.9589,
"step": 2750
},
{
"epoch": 1.1,
"learning_rate": 2.4900000000000002e-05,
"loss": 2.0053,
"step": 2760
},
{
"epoch": 1.11,
"learning_rate": 2.478888888888889e-05,
"loss": 1.8778,
"step": 2770
},
{
"epoch": 1.11,
"learning_rate": 2.467777777777778e-05,
"loss": 1.9399,
"step": 2780
},
{
"epoch": 1.12,
"learning_rate": 2.456666666666667e-05,
"loss": 1.9006,
"step": 2790
},
{
"epoch": 1.12,
"learning_rate": 2.4455555555555558e-05,
"loss": 1.9952,
"step": 2800
},
{
"epoch": 1.12,
"learning_rate": 2.4344444444444446e-05,
"loss": 1.97,
"step": 2810
},
{
"epoch": 1.13,
"learning_rate": 2.4233333333333337e-05,
"loss": 1.9419,
"step": 2820
},
{
"epoch": 1.13,
"learning_rate": 2.4122222222222225e-05,
"loss": 1.947,
"step": 2830
},
{
"epoch": 1.14,
"learning_rate": 2.4011111111111113e-05,
"loss": 1.852,
"step": 2840
},
{
"epoch": 1.14,
"learning_rate": 2.39e-05,
"loss": 1.9214,
"step": 2850
},
{
"epoch": 1.14,
"learning_rate": 2.378888888888889e-05,
"loss": 1.9193,
"step": 2860
},
{
"epoch": 1.15,
"learning_rate": 2.3677777777777778e-05,
"loss": 1.9713,
"step": 2870
},
{
"epoch": 1.15,
"learning_rate": 2.3566666666666666e-05,
"loss": 1.9001,
"step": 2880
},
{
"epoch": 1.16,
"learning_rate": 2.3455555555555557e-05,
"loss": 1.9345,
"step": 2890
},
{
"epoch": 1.16,
"learning_rate": 2.3344444444444445e-05,
"loss": 1.9964,
"step": 2900
},
{
"epoch": 1.16,
"learning_rate": 2.3233333333333333e-05,
"loss": 1.9469,
"step": 2910
},
{
"epoch": 1.17,
"learning_rate": 2.312222222222222e-05,
"loss": 1.9957,
"step": 2920
},
{
"epoch": 1.17,
"learning_rate": 2.301111111111111e-05,
"loss": 1.8821,
"step": 2930
},
{
"epoch": 1.18,
"learning_rate": 2.29e-05,
"loss": 1.9346,
"step": 2940
},
{
"epoch": 1.18,
"learning_rate": 2.278888888888889e-05,
"loss": 1.9556,
"step": 2950
},
{
"epoch": 1.18,
"learning_rate": 2.2677777777777777e-05,
"loss": 1.9467,
"step": 2960
},
{
"epoch": 1.19,
"learning_rate": 2.2566666666666665e-05,
"loss": 1.8833,
"step": 2970
},
{
"epoch": 1.19,
"learning_rate": 2.2455555555555557e-05,
"loss": 2.0343,
"step": 2980
},
{
"epoch": 1.2,
"learning_rate": 2.2344444444444445e-05,
"loss": 1.9408,
"step": 2990
},
{
"epoch": 1.2,
"learning_rate": 2.2233333333333333e-05,
"loss": 1.9366,
"step": 3000
},
{
"epoch": 1.2,
"learning_rate": 2.212222222222222e-05,
"loss": 1.9199,
"step": 3010
},
{
"epoch": 1.21,
"learning_rate": 2.2011111111111112e-05,
"loss": 1.8901,
"step": 3020
},
{
"epoch": 1.21,
"learning_rate": 2.19e-05,
"loss": 2.0184,
"step": 3030
},
{
"epoch": 1.22,
"learning_rate": 2.178888888888889e-05,
"loss": 1.9105,
"step": 3040
},
{
"epoch": 1.22,
"learning_rate": 2.167777777777778e-05,
"loss": 1.9003,
"step": 3050
},
{
"epoch": 1.22,
"learning_rate": 2.1566666666666668e-05,
"loss": 1.8537,
"step": 3060
},
{
"epoch": 1.23,
"learning_rate": 2.1455555555555556e-05,
"loss": 1.9618,
"step": 3070
},
{
"epoch": 1.23,
"learning_rate": 2.1344444444444444e-05,
"loss": 1.9464,
"step": 3080
},
{
"epoch": 1.24,
"learning_rate": 2.1233333333333336e-05,
"loss": 1.9511,
"step": 3090
},
{
"epoch": 1.24,
"learning_rate": 2.1122222222222224e-05,
"loss": 1.9676,
"step": 3100
},
{
"epoch": 1.24,
"learning_rate": 2.1011111111111112e-05,
"loss": 1.8459,
"step": 3110
},
{
"epoch": 1.25,
"learning_rate": 2.09e-05,
"loss": 1.8821,
"step": 3120
},
{
"epoch": 1.25,
"learning_rate": 2.078888888888889e-05,
"loss": 1.9443,
"step": 3130
},
{
"epoch": 1.26,
"learning_rate": 2.067777777777778e-05,
"loss": 2.0075,
"step": 3140
},
{
"epoch": 1.26,
"learning_rate": 2.0566666666666667e-05,
"loss": 1.9962,
"step": 3150
},
{
"epoch": 1.26,
"learning_rate": 2.0455555555555555e-05,
"loss": 1.936,
"step": 3160
},
{
"epoch": 1.27,
"learning_rate": 2.0344444444444447e-05,
"loss": 1.9004,
"step": 3170
},
{
"epoch": 1.27,
"learning_rate": 2.0233333333333335e-05,
"loss": 1.9157,
"step": 3180
},
{
"epoch": 1.28,
"learning_rate": 2.0122222222222223e-05,
"loss": 2.0501,
"step": 3190
},
{
"epoch": 1.28,
"learning_rate": 2.001111111111111e-05,
"loss": 1.9606,
"step": 3200
},
{
"epoch": 1.28,
"learning_rate": 1.9900000000000003e-05,
"loss": 1.9151,
"step": 3210
},
{
"epoch": 1.29,
"learning_rate": 1.978888888888889e-05,
"loss": 1.9383,
"step": 3220
},
{
"epoch": 1.29,
"learning_rate": 1.967777777777778e-05,
"loss": 1.9378,
"step": 3230
},
{
"epoch": 1.3,
"learning_rate": 1.9566666666666667e-05,
"loss": 1.8672,
"step": 3240
},
{
"epoch": 1.3,
"learning_rate": 1.9455555555555558e-05,
"loss": 1.905,
"step": 3250
},
{
"epoch": 1.3,
"learning_rate": 1.9344444444444446e-05,
"loss": 1.9974,
"step": 3260
},
{
"epoch": 1.31,
"learning_rate": 1.9233333333333334e-05,
"loss": 1.9442,
"step": 3270
},
{
"epoch": 1.31,
"learning_rate": 1.9122222222222222e-05,
"loss": 1.9567,
"step": 3280
},
{
"epoch": 1.32,
"learning_rate": 1.9011111111111114e-05,
"loss": 1.9023,
"step": 3290
},
{
"epoch": 1.32,
"learning_rate": 1.8900000000000002e-05,
"loss": 1.9873,
"step": 3300
},
{
"epoch": 1.32,
"learning_rate": 1.878888888888889e-05,
"loss": 1.9145,
"step": 3310
},
{
"epoch": 1.33,
"learning_rate": 1.8677777777777778e-05,
"loss": 1.8977,
"step": 3320
},
{
"epoch": 1.33,
"learning_rate": 1.856666666666667e-05,
"loss": 1.966,
"step": 3330
},
{
"epoch": 1.34,
"learning_rate": 1.8455555555555558e-05,
"loss": 1.827,
"step": 3340
},
{
"epoch": 1.34,
"learning_rate": 1.8344444444444446e-05,
"loss": 1.8427,
"step": 3350
},
{
"epoch": 1.34,
"learning_rate": 1.8233333333333334e-05,
"loss": 1.909,
"step": 3360
},
{
"epoch": 1.35,
"learning_rate": 1.8122222222222225e-05,
"loss": 1.8733,
"step": 3370
},
{
"epoch": 1.35,
"learning_rate": 1.8011111111111113e-05,
"loss": 1.8588,
"step": 3380
},
{
"epoch": 1.36,
"learning_rate": 1.79e-05,
"loss": 1.9429,
"step": 3390
},
{
"epoch": 1.36,
"learning_rate": 1.778888888888889e-05,
"loss": 1.9621,
"step": 3400
},
{
"epoch": 1.36,
"learning_rate": 1.7677777777777778e-05,
"loss": 1.8835,
"step": 3410
},
{
"epoch": 1.37,
"learning_rate": 1.756666666666667e-05,
"loss": 1.8782,
"step": 3420
},
{
"epoch": 1.37,
"learning_rate": 1.7455555555555557e-05,
"loss": 1.9371,
"step": 3430
},
{
"epoch": 1.38,
"learning_rate": 1.7344444444444445e-05,
"loss": 1.9396,
"step": 3440
},
{
"epoch": 1.38,
"learning_rate": 1.7233333333333333e-05,
"loss": 1.9371,
"step": 3450
},
{
"epoch": 1.38,
"learning_rate": 1.712222222222222e-05,
"loss": 1.9918,
"step": 3460
},
{
"epoch": 1.39,
"learning_rate": 1.701111111111111e-05,
"loss": 1.9009,
"step": 3470
},
{
"epoch": 1.39,
"learning_rate": 1.69e-05,
"loss": 1.9499,
"step": 3480
},
{
"epoch": 1.4,
"learning_rate": 1.678888888888889e-05,
"loss": 1.9332,
"step": 3490
},
{
"epoch": 1.4,
"learning_rate": 1.6677777777777777e-05,
"loss": 2.0289,
"step": 3500
},
{
"epoch": 1.4,
"learning_rate": 1.6566666666666665e-05,
"loss": 1.9174,
"step": 3510
},
{
"epoch": 1.41,
"learning_rate": 1.6455555555555556e-05,
"loss": 1.9847,
"step": 3520
},
{
"epoch": 1.41,
"learning_rate": 1.6344444444444445e-05,
"loss": 1.8906,
"step": 3530
},
{
"epoch": 1.42,
"learning_rate": 1.6233333333333333e-05,
"loss": 1.8828,
"step": 3540
},
{
"epoch": 1.42,
"learning_rate": 1.612222222222222e-05,
"loss": 1.9662,
"step": 3550
},
{
"epoch": 1.42,
"learning_rate": 1.6011111111111112e-05,
"loss": 1.9673,
"step": 3560
},
{
"epoch": 1.43,
"learning_rate": 1.59e-05,
"loss": 1.9257,
"step": 3570
},
{
"epoch": 1.43,
"learning_rate": 1.5788888888888888e-05,
"loss": 1.8924,
"step": 3580
},
{
"epoch": 1.44,
"learning_rate": 1.5677777777777776e-05,
"loss": 1.8919,
"step": 3590
},
{
"epoch": 1.44,
"learning_rate": 1.5566666666666668e-05,
"loss": 2.0188,
"step": 3600
},
{
"epoch": 1.44,
"learning_rate": 1.5455555555555556e-05,
"loss": 1.9547,
"step": 3610
},
{
"epoch": 1.45,
"learning_rate": 1.5344444444444444e-05,
"loss": 1.9254,
"step": 3620
},
{
"epoch": 1.45,
"learning_rate": 1.5233333333333332e-05,
"loss": 1.8957,
"step": 3630
},
{
"epoch": 1.46,
"learning_rate": 1.5122222222222224e-05,
"loss": 2.0042,
"step": 3640
},
{
"epoch": 1.46,
"learning_rate": 1.5011111111111112e-05,
"loss": 1.9171,
"step": 3650
},
{
"epoch": 1.46,
"learning_rate": 1.49e-05,
"loss": 1.9193,
"step": 3660
},
{
"epoch": 1.47,
"learning_rate": 1.4788888888888888e-05,
"loss": 1.9595,
"step": 3670
},
{
"epoch": 1.47,
"learning_rate": 1.467777777777778e-05,
"loss": 1.9526,
"step": 3680
},
{
"epoch": 1.48,
"learning_rate": 1.4566666666666667e-05,
"loss": 1.8556,
"step": 3690
},
{
"epoch": 1.48,
"learning_rate": 1.4455555555555555e-05,
"loss": 1.9649,
"step": 3700
},
{
"epoch": 1.48,
"learning_rate": 1.4344444444444447e-05,
"loss": 1.9939,
"step": 3710
},
{
"epoch": 1.49,
"learning_rate": 1.4233333333333335e-05,
"loss": 1.9012,
"step": 3720
},
{
"epoch": 1.49,
"learning_rate": 1.4122222222222223e-05,
"loss": 1.9954,
"step": 3730
},
{
"epoch": 1.5,
"learning_rate": 1.4011111111111111e-05,
"loss": 1.8724,
"step": 3740
},
{
"epoch": 1.5,
"learning_rate": 1.3900000000000002e-05,
"loss": 1.8799,
"step": 3750
},
{
"epoch": 1.5,
"learning_rate": 1.378888888888889e-05,
"loss": 1.9449,
"step": 3760
},
{
"epoch": 1.51,
"learning_rate": 1.3677777777777779e-05,
"loss": 1.9504,
"step": 3770
},
{
"epoch": 1.51,
"learning_rate": 1.3566666666666667e-05,
"loss": 1.8637,
"step": 3780
},
{
"epoch": 1.52,
"learning_rate": 1.3455555555555558e-05,
"loss": 1.9266,
"step": 3790
},
{
"epoch": 1.52,
"learning_rate": 1.3344444444444446e-05,
"loss": 1.8858,
"step": 3800
},
{
"epoch": 1.52,
"learning_rate": 1.3233333333333334e-05,
"loss": 1.9453,
"step": 3810
},
{
"epoch": 1.53,
"learning_rate": 1.3122222222222222e-05,
"loss": 1.911,
"step": 3820
},
{
"epoch": 1.53,
"learning_rate": 1.3011111111111112e-05,
"loss": 1.9267,
"step": 3830
},
{
"epoch": 1.54,
"learning_rate": 1.29e-05,
"loss": 1.9917,
"step": 3840
},
{
"epoch": 1.54,
"learning_rate": 1.278888888888889e-05,
"loss": 1.8827,
"step": 3850
},
{
"epoch": 1.54,
"learning_rate": 1.2677777777777778e-05,
"loss": 1.9137,
"step": 3860
},
{
"epoch": 1.55,
"learning_rate": 1.2566666666666668e-05,
"loss": 1.9071,
"step": 3870
},
{
"epoch": 1.55,
"learning_rate": 1.2455555555555556e-05,
"loss": 1.9658,
"step": 3880
},
{
"epoch": 1.56,
"learning_rate": 1.2344444444444444e-05,
"loss": 1.8949,
"step": 3890
},
{
"epoch": 1.56,
"learning_rate": 1.2233333333333334e-05,
"loss": 1.9165,
"step": 3900
},
{
"epoch": 1.56,
"learning_rate": 1.2122222222222222e-05,
"loss": 1.8986,
"step": 3910
},
{
"epoch": 1.57,
"learning_rate": 1.2011111111111111e-05,
"loss": 1.9468,
"step": 3920
},
{
"epoch": 1.57,
"learning_rate": 1.19e-05,
"loss": 1.9018,
"step": 3930
},
{
"epoch": 1.58,
"learning_rate": 1.178888888888889e-05,
"loss": 1.9663,
"step": 3940
},
{
"epoch": 1.58,
"learning_rate": 1.1677777777777777e-05,
"loss": 1.9203,
"step": 3950
},
{
"epoch": 1.58,
"learning_rate": 1.1566666666666667e-05,
"loss": 1.9288,
"step": 3960
},
{
"epoch": 1.59,
"learning_rate": 1.1455555555555555e-05,
"loss": 1.8868,
"step": 3970
},
{
"epoch": 1.59,
"learning_rate": 1.1344444444444445e-05,
"loss": 1.9763,
"step": 3980
},
{
"epoch": 1.6,
"learning_rate": 1.1233333333333333e-05,
"loss": 1.854,
"step": 3990
},
{
"epoch": 1.6,
"learning_rate": 1.1122222222222223e-05,
"loss": 1.8515,
"step": 4000
},
{
"epoch": 1.6,
"learning_rate": 1.1011111111111113e-05,
"loss": 1.8175,
"step": 4010
},
{
"epoch": 1.61,
"learning_rate": 1.09e-05,
"loss": 1.9691,
"step": 4020
},
{
"epoch": 1.61,
"learning_rate": 1.078888888888889e-05,
"loss": 1.9804,
"step": 4030
},
{
"epoch": 1.62,
"learning_rate": 1.0677777777777779e-05,
"loss": 1.9356,
"step": 4040
},
{
"epoch": 1.62,
"learning_rate": 1.0566666666666668e-05,
"loss": 1.9108,
"step": 4050
},
{
"epoch": 1.62,
"learning_rate": 1.0466666666666668e-05,
"loss": 1.8432,
"step": 4060
},
{
"epoch": 1.63,
"learning_rate": 1.0355555555555556e-05,
"loss": 1.9151,
"step": 4070
},
{
"epoch": 1.63,
"learning_rate": 1.0244444444444445e-05,
"loss": 1.8866,
"step": 4080
},
{
"epoch": 1.64,
"learning_rate": 1.0133333333333333e-05,
"loss": 1.943,
"step": 4090
},
{
"epoch": 1.64,
"learning_rate": 1.0022222222222223e-05,
"loss": 1.8325,
"step": 4100
},
{
"epoch": 1.64,
"learning_rate": 9.911111111111111e-06,
"loss": 1.8784,
"step": 4110
},
{
"epoch": 1.65,
"learning_rate": 9.800000000000001e-06,
"loss": 1.8412,
"step": 4120
},
{
"epoch": 1.65,
"learning_rate": 9.688888888888889e-06,
"loss": 1.8959,
"step": 4130
},
{
"epoch": 1.66,
"learning_rate": 9.577777777777779e-06,
"loss": 1.8586,
"step": 4140
},
{
"epoch": 1.66,
"learning_rate": 9.466666666666667e-06,
"loss": 1.9707,
"step": 4150
},
{
"epoch": 1.66,
"learning_rate": 9.355555555555557e-06,
"loss": 1.9178,
"step": 4160
},
{
"epoch": 1.67,
"learning_rate": 9.244444444444445e-06,
"loss": 1.9418,
"step": 4170
},
{
"epoch": 1.67,
"learning_rate": 9.133333333333335e-06,
"loss": 1.9177,
"step": 4180
},
{
"epoch": 1.68,
"learning_rate": 9.022222222222223e-06,
"loss": 1.9082,
"step": 4190
},
{
"epoch": 1.68,
"learning_rate": 8.911111111111112e-06,
"loss": 1.9093,
"step": 4200
},
{
"epoch": 1.68,
"learning_rate": 8.8e-06,
"loss": 1.863,
"step": 4210
},
{
"epoch": 1.69,
"learning_rate": 8.68888888888889e-06,
"loss": 1.9436,
"step": 4220
},
{
"epoch": 1.69,
"learning_rate": 8.577777777777778e-06,
"loss": 1.852,
"step": 4230
},
{
"epoch": 1.7,
"learning_rate": 8.466666666666666e-06,
"loss": 1.9117,
"step": 4240
},
{
"epoch": 1.7,
"learning_rate": 8.355555555555556e-06,
"loss": 1.8644,
"step": 4250
},
{
"epoch": 1.7,
"learning_rate": 8.244444444444444e-06,
"loss": 1.9091,
"step": 4260
},
{
"epoch": 1.71,
"learning_rate": 8.133333333333332e-06,
"loss": 1.9222,
"step": 4270
},
{
"epoch": 1.71,
"learning_rate": 8.022222222222222e-06,
"loss": 1.9054,
"step": 4280
},
{
"epoch": 1.72,
"learning_rate": 7.91111111111111e-06,
"loss": 1.9646,
"step": 4290
},
{
"epoch": 1.72,
"learning_rate": 7.8e-06,
"loss": 2.0081,
"step": 4300
},
{
"epoch": 1.72,
"learning_rate": 7.68888888888889e-06,
"loss": 1.9646,
"step": 4310
},
{
"epoch": 1.73,
"learning_rate": 7.577777777777778e-06,
"loss": 1.993,
"step": 4320
},
{
"epoch": 1.73,
"learning_rate": 7.4666666666666675e-06,
"loss": 1.9562,
"step": 4330
},
{
"epoch": 1.74,
"learning_rate": 7.3555555555555555e-06,
"loss": 1.8728,
"step": 4340
},
{
"epoch": 1.74,
"learning_rate": 7.244444444444445e-06,
"loss": 1.9075,
"step": 4350
},
{
"epoch": 1.74,
"learning_rate": 7.133333333333333e-06,
"loss": 1.9145,
"step": 4360
},
{
"epoch": 1.75,
"learning_rate": 7.022222222222223e-06,
"loss": 1.9148,
"step": 4370
},
{
"epoch": 1.75,
"learning_rate": 6.911111111111111e-06,
"loss": 1.8758,
"step": 4380
},
{
"epoch": 1.76,
"learning_rate": 6.800000000000001e-06,
"loss": 1.9491,
"step": 4390
},
{
"epoch": 1.76,
"learning_rate": 6.688888888888889e-06,
"loss": 1.8994,
"step": 4400
},
{
"epoch": 1.76,
"learning_rate": 6.577777777777779e-06,
"loss": 1.8458,
"step": 4410
},
{
"epoch": 1.77,
"learning_rate": 6.466666666666667e-06,
"loss": 1.9138,
"step": 4420
},
{
"epoch": 1.77,
"learning_rate": 6.355555555555557e-06,
"loss": 1.9492,
"step": 4430
},
{
"epoch": 1.78,
"learning_rate": 6.244444444444445e-06,
"loss": 1.9269,
"step": 4440
},
{
"epoch": 1.78,
"learning_rate": 6.133333333333334e-06,
"loss": 1.9292,
"step": 4450
},
{
"epoch": 1.78,
"learning_rate": 6.0222222222222225e-06,
"loss": 1.9449,
"step": 4460
},
{
"epoch": 1.79,
"learning_rate": 5.9111111111111115e-06,
"loss": 1.8984,
"step": 4470
},
{
"epoch": 1.79,
"learning_rate": 5.8e-06,
"loss": 1.9091,
"step": 4480
},
{
"epoch": 1.8,
"learning_rate": 5.688888888888889e-06,
"loss": 1.8666,
"step": 4490
},
{
"epoch": 1.8,
"learning_rate": 5.577777777777778e-06,
"loss": 1.8918,
"step": 4500
},
{
"epoch": 1.8,
"learning_rate": 5.466666666666667e-06,
"loss": 1.9444,
"step": 4510
},
{
"epoch": 1.81,
"learning_rate": 5.355555555555556e-06,
"loss": 1.9556,
"step": 4520
},
{
"epoch": 1.81,
"learning_rate": 5.244444444444445e-06,
"loss": 1.9307,
"step": 4530
},
{
"epoch": 1.82,
"learning_rate": 5.133333333333334e-06,
"loss": 1.9893,
"step": 4540
},
{
"epoch": 1.82,
"learning_rate": 5.022222222222223e-06,
"loss": 1.9046,
"step": 4550
},
{
"epoch": 1.82,
"learning_rate": 4.911111111111112e-06,
"loss": 1.909,
"step": 4560
},
{
"epoch": 1.83,
"learning_rate": 4.800000000000001e-06,
"loss": 1.9228,
"step": 4570
},
{
"epoch": 1.83,
"learning_rate": 4.6888888888888895e-06,
"loss": 1.911,
"step": 4580
},
{
"epoch": 1.84,
"learning_rate": 4.5777777777777785e-06,
"loss": 1.8263,
"step": 4590
},
{
"epoch": 1.84,
"learning_rate": 4.4666666666666665e-06,
"loss": 1.8926,
"step": 4600
},
{
"epoch": 1.84,
"learning_rate": 4.3555555555555555e-06,
"loss": 1.9381,
"step": 4610
},
{
"epoch": 1.85,
"learning_rate": 4.244444444444444e-06,
"loss": 1.9395,
"step": 4620
},
{
"epoch": 1.85,
"learning_rate": 4.133333333333333e-06,
"loss": 1.9294,
"step": 4630
},
{
"epoch": 1.86,
"learning_rate": 4.022222222222222e-06,
"loss": 1.9132,
"step": 4640
},
{
"epoch": 1.86,
"learning_rate": 3.911111111111111e-06,
"loss": 1.9346,
"step": 4650
},
{
"epoch": 1.86,
"learning_rate": 3.8e-06,
"loss": 1.9006,
"step": 4660
},
{
"epoch": 1.87,
"learning_rate": 3.688888888888889e-06,
"loss": 1.911,
"step": 4670
},
{
"epoch": 1.87,
"learning_rate": 3.577777777777778e-06,
"loss": 1.8575,
"step": 4680
},
{
"epoch": 1.88,
"learning_rate": 3.466666666666667e-06,
"loss": 1.9116,
"step": 4690
},
{
"epoch": 1.88,
"learning_rate": 3.3555555555555557e-06,
"loss": 1.941,
"step": 4700
},
{
"epoch": 1.88,
"learning_rate": 3.244444444444444e-06,
"loss": 1.9793,
"step": 4710
},
{
"epoch": 1.89,
"learning_rate": 3.133333333333333e-06,
"loss": 1.8679,
"step": 4720
},
{
"epoch": 1.89,
"learning_rate": 3.0222222222222225e-06,
"loss": 1.953,
"step": 4730
},
{
"epoch": 1.9,
"learning_rate": 2.9111111111111114e-06,
"loss": 1.8952,
"step": 4740
},
{
"epoch": 1.9,
"learning_rate": 2.8000000000000003e-06,
"loss": 1.8668,
"step": 4750
},
{
"epoch": 1.9,
"learning_rate": 2.6888888888888892e-06,
"loss": 1.9272,
"step": 4760
},
{
"epoch": 1.91,
"learning_rate": 2.5777777777777777e-06,
"loss": 1.9028,
"step": 4770
},
{
"epoch": 1.91,
"learning_rate": 2.4666666666666666e-06,
"loss": 1.9104,
"step": 4780
},
{
"epoch": 1.92,
"learning_rate": 2.3555555555555555e-06,
"loss": 1.93,
"step": 4790
},
{
"epoch": 1.92,
"learning_rate": 2.2444444444444445e-06,
"loss": 1.8212,
"step": 4800
},
{
"epoch": 1.92,
"learning_rate": 2.1333333333333334e-06,
"loss": 1.9506,
"step": 4810
},
{
"epoch": 1.93,
"learning_rate": 2.0222222222222223e-06,
"loss": 1.9462,
"step": 4820
},
{
"epoch": 1.93,
"learning_rate": 1.9111111111111112e-06,
"loss": 1.9526,
"step": 4830
},
{
"epoch": 1.94,
"learning_rate": 1.8e-06,
"loss": 1.8744,
"step": 4840
},
{
"epoch": 1.94,
"learning_rate": 1.6888888888888888e-06,
"loss": 1.882,
"step": 4850
},
{
"epoch": 1.94,
"learning_rate": 1.577777777777778e-06,
"loss": 1.8525,
"step": 4860
},
{
"epoch": 1.95,
"learning_rate": 1.4666666666666667e-06,
"loss": 1.9561,
"step": 4870
},
{
"epoch": 1.95,
"learning_rate": 1.3555555555555556e-06,
"loss": 1.9858,
"step": 4880
},
{
"epoch": 1.96,
"learning_rate": 1.2444444444444445e-06,
"loss": 1.8857,
"step": 4890
},
{
"epoch": 1.96,
"learning_rate": 1.1333333333333334e-06,
"loss": 1.8852,
"step": 4900
},
{
"epoch": 1.96,
"learning_rate": 1.0222222222222223e-06,
"loss": 1.9136,
"step": 4910
},
{
"epoch": 1.97,
"learning_rate": 9.111111111111112e-07,
"loss": 1.8647,
"step": 4920
},
{
"epoch": 1.97,
"learning_rate": 8.000000000000001e-07,
"loss": 1.9747,
"step": 4930
},
{
"epoch": 1.98,
"learning_rate": 6.888888888888889e-07,
"loss": 1.844,
"step": 4940
},
{
"epoch": 1.98,
"learning_rate": 5.777777777777778e-07,
"loss": 1.9402,
"step": 4950
},
{
"epoch": 1.98,
"learning_rate": 4.666666666666667e-07,
"loss": 1.8838,
"step": 4960
},
{
"epoch": 1.99,
"learning_rate": 3.555555555555556e-07,
"loss": 1.8683,
"step": 4970
},
{
"epoch": 1.99,
"learning_rate": 2.4444444444444445e-07,
"loss": 1.9806,
"step": 4980
},
{
"epoch": 2.0,
"learning_rate": 1.3333333333333334e-07,
"loss": 1.8673,
"step": 4990
},
{
"epoch": 2.0,
"learning_rate": 2.2222222222222224e-08,
"loss": 1.9773,
"step": 5000
}
],
"max_steps": 5000,
"num_train_epochs": 2,
"total_flos": 1.7336836816896e+17,
"trial_name": null,
"trial_params": null
}