7325 lines
144 KiB
JSON
7325 lines
144 KiB
JSON
|
|
{
|
||
|
|
"best_metric": null,
|
||
|
|
"best_model_checkpoint": null,
|
||
|
|
"epoch": 2.9972316210396803,
|
||
|
|
"global_step": 1218,
|
||
|
|
"is_hyper_param_search": false,
|
||
|
|
"is_local_process_zero": true,
|
||
|
|
"is_world_process_zero": true,
|
||
|
|
"log_history": [
|
||
|
|
{
|
||
|
|
"epoch": 0.0,
|
||
|
|
"learning_rate": 5.405405405405406e-07,
|
||
|
|
"loss": 1.5052,
|
||
|
|
"step": 1
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.0,
|
||
|
|
"learning_rate": 1.0810810810810812e-06,
|
||
|
|
"loss": 1.572,
|
||
|
|
"step": 2
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.01,
|
||
|
|
"learning_rate": 1.6216216216216219e-06,
|
||
|
|
"loss": 1.4854,
|
||
|
|
"step": 3
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.01,
|
||
|
|
"learning_rate": 2.1621621621621623e-06,
|
||
|
|
"loss": 1.4884,
|
||
|
|
"step": 4
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.01,
|
||
|
|
"learning_rate": 2.1621621621621623e-06,
|
||
|
|
"loss": 1.5765,
|
||
|
|
"step": 5
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.01,
|
||
|
|
"learning_rate": 2.702702702702703e-06,
|
||
|
|
"loss": 1.5366,
|
||
|
|
"step": 6
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.02,
|
||
|
|
"learning_rate": 3.2432432432432437e-06,
|
||
|
|
"loss": 1.5657,
|
||
|
|
"step": 7
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.02,
|
||
|
|
"learning_rate": 3.7837837837837844e-06,
|
||
|
|
"loss": 1.5046,
|
||
|
|
"step": 8
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.02,
|
||
|
|
"learning_rate": 3.7837837837837844e-06,
|
||
|
|
"loss": 1.4414,
|
||
|
|
"step": 9
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.02,
|
||
|
|
"learning_rate": 4.324324324324325e-06,
|
||
|
|
"loss": 1.4408,
|
||
|
|
"step": 10
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.03,
|
||
|
|
"learning_rate": 4.864864864864866e-06,
|
||
|
|
"loss": 1.4576,
|
||
|
|
"step": 11
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.03,
|
||
|
|
"learning_rate": 5.405405405405406e-06,
|
||
|
|
"loss": 1.4792,
|
||
|
|
"step": 12
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.03,
|
||
|
|
"learning_rate": 5.405405405405406e-06,
|
||
|
|
"loss": 1.3751,
|
||
|
|
"step": 13
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.03,
|
||
|
|
"learning_rate": 5.405405405405406e-06,
|
||
|
|
"loss": 1.2943,
|
||
|
|
"step": 14
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.04,
|
||
|
|
"learning_rate": 5.945945945945947e-06,
|
||
|
|
"loss": 1.3096,
|
||
|
|
"step": 15
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.04,
|
||
|
|
"learning_rate": 6.486486486486487e-06,
|
||
|
|
"loss": 1.3018,
|
||
|
|
"step": 16
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.04,
|
||
|
|
"learning_rate": 7.027027027027028e-06,
|
||
|
|
"loss": 1.3502,
|
||
|
|
"step": 17
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.04,
|
||
|
|
"learning_rate": 7.567567567567569e-06,
|
||
|
|
"loss": 1.2168,
|
||
|
|
"step": 18
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.05,
|
||
|
|
"learning_rate": 8.108108108108109e-06,
|
||
|
|
"loss": 1.2869,
|
||
|
|
"step": 19
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.05,
|
||
|
|
"learning_rate": 8.64864864864865e-06,
|
||
|
|
"loss": 1.201,
|
||
|
|
"step": 20
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.05,
|
||
|
|
"learning_rate": 8.64864864864865e-06,
|
||
|
|
"loss": 1.2444,
|
||
|
|
"step": 21
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.05,
|
||
|
|
"learning_rate": 8.64864864864865e-06,
|
||
|
|
"loss": 1.2378,
|
||
|
|
"step": 22
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.06,
|
||
|
|
"learning_rate": 8.64864864864865e-06,
|
||
|
|
"loss": 1.2257,
|
||
|
|
"step": 23
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.06,
|
||
|
|
"learning_rate": 8.64864864864865e-06,
|
||
|
|
"loss": 1.2749,
|
||
|
|
"step": 24
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.06,
|
||
|
|
"learning_rate": 9.189189189189191e-06,
|
||
|
|
"loss": 1.2798,
|
||
|
|
"step": 25
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.06,
|
||
|
|
"learning_rate": 9.729729729729732e-06,
|
||
|
|
"loss": 1.2705,
|
||
|
|
"step": 26
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.07,
|
||
|
|
"learning_rate": 1.027027027027027e-05,
|
||
|
|
"loss": 1.2711,
|
||
|
|
"step": 27
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.07,
|
||
|
|
"learning_rate": 1.0810810810810812e-05,
|
||
|
|
"loss": 1.2435,
|
||
|
|
"step": 28
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.07,
|
||
|
|
"learning_rate": 1.1351351351351352e-05,
|
||
|
|
"loss": 1.2006,
|
||
|
|
"step": 29
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.07,
|
||
|
|
"learning_rate": 1.1351351351351352e-05,
|
||
|
|
"loss": 1.1974,
|
||
|
|
"step": 30
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.08,
|
||
|
|
"learning_rate": 1.1891891891891894e-05,
|
||
|
|
"loss": 1.2233,
|
||
|
|
"step": 31
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.08,
|
||
|
|
"learning_rate": 1.2432432432432433e-05,
|
||
|
|
"loss": 1.2902,
|
||
|
|
"step": 32
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.08,
|
||
|
|
"learning_rate": 1.2972972972972975e-05,
|
||
|
|
"loss": 1.1509,
|
||
|
|
"step": 33
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.08,
|
||
|
|
"learning_rate": 1.3513513513513515e-05,
|
||
|
|
"loss": 1.198,
|
||
|
|
"step": 34
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09,
|
||
|
|
"learning_rate": 1.4054054054054055e-05,
|
||
|
|
"loss": 1.1384,
|
||
|
|
"step": 35
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09,
|
||
|
|
"learning_rate": 1.4594594594594596e-05,
|
||
|
|
"loss": 1.2207,
|
||
|
|
"step": 36
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09,
|
||
|
|
"learning_rate": 1.5135135135135138e-05,
|
||
|
|
"loss": 1.1605,
|
||
|
|
"step": 37
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.09,
|
||
|
|
"learning_rate": 1.5675675675675676e-05,
|
||
|
|
"loss": 1.2356,
|
||
|
|
"step": 38
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1,
|
||
|
|
"learning_rate": 1.6216216216216218e-05,
|
||
|
|
"loss": 1.212,
|
||
|
|
"step": 39
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1,
|
||
|
|
"learning_rate": 1.6756756756756757e-05,
|
||
|
|
"loss": 1.0847,
|
||
|
|
"step": 40
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1,
|
||
|
|
"learning_rate": 1.72972972972973e-05,
|
||
|
|
"loss": 1.1486,
|
||
|
|
"step": 41
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.1,
|
||
|
|
"learning_rate": 1.783783783783784e-05,
|
||
|
|
"loss": 1.1702,
|
||
|
|
"step": 42
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.11,
|
||
|
|
"learning_rate": 1.8378378378378383e-05,
|
||
|
|
"loss": 1.1427,
|
||
|
|
"step": 43
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.11,
|
||
|
|
"learning_rate": 1.891891891891892e-05,
|
||
|
|
"loss": 1.1607,
|
||
|
|
"step": 44
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.11,
|
||
|
|
"learning_rate": 1.9459459459459463e-05,
|
||
|
|
"loss": 1.1498,
|
||
|
|
"step": 45
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.11,
|
||
|
|
"learning_rate": 2e-05,
|
||
|
|
"loss": 1.1193,
|
||
|
|
"step": 46
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.12,
|
||
|
|
"learning_rate": 1.999996461903301e-05,
|
||
|
|
"loss": 1.2106,
|
||
|
|
"step": 47
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.12,
|
||
|
|
"learning_rate": 1.9999858476382388e-05,
|
||
|
|
"loss": 1.2074,
|
||
|
|
"step": 48
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.12,
|
||
|
|
"learning_rate": 1.9999681572799226e-05,
|
||
|
|
"loss": 1.1671,
|
||
|
|
"step": 49
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.12,
|
||
|
|
"learning_rate": 1.9999433909535333e-05,
|
||
|
|
"loss": 1.1289,
|
||
|
|
"step": 50
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.13,
|
||
|
|
"learning_rate": 1.9999115488343213e-05,
|
||
|
|
"loss": 1.1084,
|
||
|
|
"step": 51
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.13,
|
||
|
|
"learning_rate": 1.999872631147608e-05,
|
||
|
|
"loss": 1.1514,
|
||
|
|
"step": 52
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.13,
|
||
|
|
"learning_rate": 1.999872631147608e-05,
|
||
|
|
"loss": 1.2061,
|
||
|
|
"step": 53
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.13,
|
||
|
|
"learning_rate": 1.999826638168783e-05,
|
||
|
|
"loss": 1.1689,
|
||
|
|
"step": 54
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.14,
|
||
|
|
"learning_rate": 1.9997735702233006e-05,
|
||
|
|
"loss": 1.1228,
|
||
|
|
"step": 55
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.14,
|
||
|
|
"learning_rate": 1.99971342768668e-05,
|
||
|
|
"loss": 1.1259,
|
||
|
|
"step": 56
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.14,
|
||
|
|
"learning_rate": 1.999646210984502e-05,
|
||
|
|
"loss": 1.1042,
|
||
|
|
"step": 57
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.14,
|
||
|
|
"learning_rate": 1.999571920592405e-05,
|
||
|
|
"loss": 1.1362,
|
||
|
|
"step": 58
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.15,
|
||
|
|
"learning_rate": 1.9994905570360817e-05,
|
||
|
|
"loss": 1.1776,
|
||
|
|
"step": 59
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.15,
|
||
|
|
"learning_rate": 1.999402120891276e-05,
|
||
|
|
"loss": 1.2139,
|
||
|
|
"step": 60
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.15,
|
||
|
|
"learning_rate": 1.99930661278378e-05,
|
||
|
|
"loss": 1.0975,
|
||
|
|
"step": 61
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.15,
|
||
|
|
"learning_rate": 1.9992040333894273e-05,
|
||
|
|
"loss": 1.1779,
|
||
|
|
"step": 62
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.16,
|
||
|
|
"learning_rate": 1.9990943834340893e-05,
|
||
|
|
"loss": 1.1284,
|
||
|
|
"step": 63
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.16,
|
||
|
|
"learning_rate": 1.9989776636936705e-05,
|
||
|
|
"loss": 1.1707,
|
||
|
|
"step": 64
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.16,
|
||
|
|
"learning_rate": 1.9988538749941024e-05,
|
||
|
|
"loss": 1.2277,
|
||
|
|
"step": 65
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.16,
|
||
|
|
"learning_rate": 1.9987230182113374e-05,
|
||
|
|
"loss": 1.2012,
|
||
|
|
"step": 66
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.16,
|
||
|
|
"learning_rate": 1.998585094271344e-05,
|
||
|
|
"loss": 1.1742,
|
||
|
|
"step": 67
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.17,
|
||
|
|
"learning_rate": 1.998440104150098e-05,
|
||
|
|
"loss": 1.1874,
|
||
|
|
"step": 68
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.17,
|
||
|
|
"learning_rate": 1.998440104150098e-05,
|
||
|
|
"loss": 1.1938,
|
||
|
|
"step": 69
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.17,
|
||
|
|
"learning_rate": 1.998288048873578e-05,
|
||
|
|
"loss": 1.1144,
|
||
|
|
"step": 70
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.17,
|
||
|
|
"learning_rate": 1.9981289295177566e-05,
|
||
|
|
"loss": 1.2661,
|
||
|
|
"step": 71
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.18,
|
||
|
|
"learning_rate": 1.9979627472085927e-05,
|
||
|
|
"loss": 1.1696,
|
||
|
|
"step": 72
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.18,
|
||
|
|
"learning_rate": 1.997789503122025e-05,
|
||
|
|
"loss": 1.1022,
|
||
|
|
"step": 73
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.18,
|
||
|
|
"learning_rate": 1.9976091984839616e-05,
|
||
|
|
"loss": 1.1423,
|
||
|
|
"step": 74
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.18,
|
||
|
|
"learning_rate": 1.9974218345702733e-05,
|
||
|
|
"loss": 1.1428,
|
||
|
|
"step": 75
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.19,
|
||
|
|
"learning_rate": 1.9972274127067838e-05,
|
||
|
|
"loss": 1.1857,
|
||
|
|
"step": 76
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.19,
|
||
|
|
"learning_rate": 1.997025934269259e-05,
|
||
|
|
"loss": 1.1107,
|
||
|
|
"step": 77
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.19,
|
||
|
|
"learning_rate": 1.9968174006833996e-05,
|
||
|
|
"loss": 1.1741,
|
||
|
|
"step": 78
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.19,
|
||
|
|
"learning_rate": 1.9966018134248296e-05,
|
||
|
|
"loss": 1.1125,
|
||
|
|
"step": 79
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2,
|
||
|
|
"learning_rate": 1.9963791740190863e-05,
|
||
|
|
"loss": 1.1686,
|
||
|
|
"step": 80
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2,
|
||
|
|
"learning_rate": 1.996149484041609e-05,
|
||
|
|
"loss": 1.2111,
|
||
|
|
"step": 81
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2,
|
||
|
|
"learning_rate": 1.9959127451177287e-05,
|
||
|
|
"loss": 1.176,
|
||
|
|
"step": 82
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.2,
|
||
|
|
"learning_rate": 1.9956689589226555e-05,
|
||
|
|
"loss": 1.141,
|
||
|
|
"step": 83
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.21,
|
||
|
|
"learning_rate": 1.9954181271814673e-05,
|
||
|
|
"loss": 1.2343,
|
||
|
|
"step": 84
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.21,
|
||
|
|
"learning_rate": 1.9951602516690988e-05,
|
||
|
|
"loss": 1.1353,
|
||
|
|
"step": 85
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.21,
|
||
|
|
"learning_rate": 1.9948953342103268e-05,
|
||
|
|
"loss": 1.1547,
|
||
|
|
"step": 86
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.21,
|
||
|
|
"learning_rate": 1.994623376679758e-05,
|
||
|
|
"loss": 1.1914,
|
||
|
|
"step": 87
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.22,
|
||
|
|
"learning_rate": 1.9943443810018174e-05,
|
||
|
|
"loss": 1.1855,
|
||
|
|
"step": 88
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.22,
|
||
|
|
"learning_rate": 1.9940583491507314e-05,
|
||
|
|
"loss": 1.2137,
|
||
|
|
"step": 89
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.22,
|
||
|
|
"learning_rate": 1.993765283150517e-05,
|
||
|
|
"loss": 1.1967,
|
||
|
|
"step": 90
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.22,
|
||
|
|
"learning_rate": 1.9934651850749663e-05,
|
||
|
|
"loss": 1.1978,
|
||
|
|
"step": 91
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.23,
|
||
|
|
"learning_rate": 1.9934651850749663e-05,
|
||
|
|
"loss": 1.1091,
|
||
|
|
"step": 92
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.23,
|
||
|
|
"learning_rate": 1.9934651850749663e-05,
|
||
|
|
"loss": 1.1659,
|
||
|
|
"step": 93
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.23,
|
||
|
|
"learning_rate": 1.9931580570476306e-05,
|
||
|
|
"loss": 1.2263,
|
||
|
|
"step": 94
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.23,
|
||
|
|
"learning_rate": 1.9931580570476306e-05,
|
||
|
|
"loss": 1.2255,
|
||
|
|
"step": 95
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.24,
|
||
|
|
"learning_rate": 1.9928439012418076e-05,
|
||
|
|
"loss": 1.1399,
|
||
|
|
"step": 96
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.24,
|
||
|
|
"learning_rate": 1.9925227198805247e-05,
|
||
|
|
"loss": 1.1647,
|
||
|
|
"step": 97
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.24,
|
||
|
|
"learning_rate": 1.9921945152365235e-05,
|
||
|
|
"loss": 1.1642,
|
||
|
|
"step": 98
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.24,
|
||
|
|
"learning_rate": 1.9918592896322432e-05,
|
||
|
|
"loss": 1.2057,
|
||
|
|
"step": 99
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.25,
|
||
|
|
"learning_rate": 1.9918592896322432e-05,
|
||
|
|
"loss": 1.139,
|
||
|
|
"step": 100
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.25,
|
||
|
|
"learning_rate": 1.9915170454398045e-05,
|
||
|
|
"loss": 1.2032,
|
||
|
|
"step": 101
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.25,
|
||
|
|
"learning_rate": 1.9911677850809943e-05,
|
||
|
|
"loss": 1.1182,
|
||
|
|
"step": 102
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.25,
|
||
|
|
"learning_rate": 1.9908115110272463e-05,
|
||
|
|
"loss": 1.1721,
|
||
|
|
"step": 103
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.26,
|
||
|
|
"learning_rate": 1.9904482257996244e-05,
|
||
|
|
"loss": 1.1924,
|
||
|
|
"step": 104
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.26,
|
||
|
|
"learning_rate": 1.990077931968805e-05,
|
||
|
|
"loss": 1.0885,
|
||
|
|
"step": 105
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.26,
|
||
|
|
"learning_rate": 1.9897006321550592e-05,
|
||
|
|
"loss": 1.1357,
|
||
|
|
"step": 106
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.26,
|
||
|
|
"learning_rate": 1.9893163290282335e-05,
|
||
|
|
"loss": 1.17,
|
||
|
|
"step": 107
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.27,
|
||
|
|
"learning_rate": 1.9893163290282335e-05,
|
||
|
|
"loss": 1.1068,
|
||
|
|
"step": 108
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.27,
|
||
|
|
"learning_rate": 1.9889250253077306e-05,
|
||
|
|
"loss": 1.0895,
|
||
|
|
"step": 109
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.27,
|
||
|
|
"learning_rate": 1.9889250253077306e-05,
|
||
|
|
"loss": 1.1376,
|
||
|
|
"step": 110
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.27,
|
||
|
|
"learning_rate": 1.9885267237624923e-05,
|
||
|
|
"loss": 1.2073,
|
||
|
|
"step": 111
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.28,
|
||
|
|
"learning_rate": 1.988121427210976e-05,
|
||
|
|
"loss": 1.2018,
|
||
|
|
"step": 112
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.28,
|
||
|
|
"learning_rate": 1.98770913852114e-05,
|
||
|
|
"loss": 1.0813,
|
||
|
|
"step": 113
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.28,
|
||
|
|
"learning_rate": 1.9872898606104175e-05,
|
||
|
|
"loss": 1.145,
|
||
|
|
"step": 114
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.28,
|
||
|
|
"learning_rate": 1.9868635964457007e-05,
|
||
|
|
"loss": 1.1122,
|
||
|
|
"step": 115
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.29,
|
||
|
|
"learning_rate": 1.986430349043317e-05,
|
||
|
|
"loss": 1.1345,
|
||
|
|
"step": 116
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.29,
|
||
|
|
"learning_rate": 1.9859901214690094e-05,
|
||
|
|
"loss": 1.0742,
|
||
|
|
"step": 117
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.29,
|
||
|
|
"learning_rate": 1.9855429168379127e-05,
|
||
|
|
"loss": 1.0388,
|
||
|
|
"step": 118
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.29,
|
||
|
|
"learning_rate": 1.9855429168379127e-05,
|
||
|
|
"loss": 1.1202,
|
||
|
|
"step": 119
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3,
|
||
|
|
"learning_rate": 1.9855429168379127e-05,
|
||
|
|
"loss": 1.17,
|
||
|
|
"step": 120
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3,
|
||
|
|
"learning_rate": 1.9855429168379127e-05,
|
||
|
|
"loss": 1.2388,
|
||
|
|
"step": 121
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3,
|
||
|
|
"learning_rate": 1.985088738314534e-05,
|
||
|
|
"loss": 1.0845,
|
||
|
|
"step": 122
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.3,
|
||
|
|
"learning_rate": 1.9846275891127275e-05,
|
||
|
|
"loss": 1.1517,
|
||
|
|
"step": 123
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.31,
|
||
|
|
"learning_rate": 1.9841594724956746e-05,
|
||
|
|
"loss": 1.1195,
|
||
|
|
"step": 124
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.31,
|
||
|
|
"learning_rate": 1.9836843917758593e-05,
|
||
|
|
"loss": 1.1406,
|
||
|
|
"step": 125
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.31,
|
||
|
|
"learning_rate": 1.983202350315044e-05,
|
||
|
|
"loss": 1.1288,
|
||
|
|
"step": 126
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.31,
|
||
|
|
"learning_rate": 1.982713351524248e-05,
|
||
|
|
"loss": 1.2129,
|
||
|
|
"step": 127
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.31,
|
||
|
|
"learning_rate": 1.982713351524248e-05,
|
||
|
|
"loss": 1.1557,
|
||
|
|
"step": 128
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.32,
|
||
|
|
"learning_rate": 1.982217398863721e-05,
|
||
|
|
"loss": 1.1713,
|
||
|
|
"step": 129
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.32,
|
||
|
|
"learning_rate": 1.98171449584292e-05,
|
||
|
|
"loss": 1.0706,
|
||
|
|
"step": 130
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.32,
|
||
|
|
"learning_rate": 1.9812046460204837e-05,
|
||
|
|
"loss": 1.1652,
|
||
|
|
"step": 131
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.32,
|
||
|
|
"learning_rate": 1.9806878530042083e-05,
|
||
|
|
"loss": 1.1527,
|
||
|
|
"step": 132
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.33,
|
||
|
|
"learning_rate": 1.9801641204510216e-05,
|
||
|
|
"loss": 1.0598,
|
||
|
|
"step": 133
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.33,
|
||
|
|
"learning_rate": 1.9796334520669555e-05,
|
||
|
|
"loss": 1.1816,
|
||
|
|
"step": 134
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.33,
|
||
|
|
"learning_rate": 1.9790958516071228e-05,
|
||
|
|
"loss": 1.1124,
|
||
|
|
"step": 135
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.33,
|
||
|
|
"learning_rate": 1.978551322875688e-05,
|
||
|
|
"loss": 1.0941,
|
||
|
|
"step": 136
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.34,
|
||
|
|
"learning_rate": 1.977999869725842e-05,
|
||
|
|
"loss": 1.1864,
|
||
|
|
"step": 137
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.34,
|
||
|
|
"learning_rate": 1.977441496059774e-05,
|
||
|
|
"loss": 1.1492,
|
||
|
|
"step": 138
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.34,
|
||
|
|
"learning_rate": 1.9768762058286433e-05,
|
||
|
|
"loss": 1.1783,
|
||
|
|
"step": 139
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.34,
|
||
|
|
"learning_rate": 1.976304003032554e-05,
|
||
|
|
"loss": 1.0771,
|
||
|
|
"step": 140
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.35,
|
||
|
|
"learning_rate": 1.9757248917205228e-05,
|
||
|
|
"loss": 1.1648,
|
||
|
|
"step": 141
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.35,
|
||
|
|
"learning_rate": 1.975138875990454e-05,
|
||
|
|
"loss": 1.1304,
|
||
|
|
"step": 142
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.35,
|
||
|
|
"learning_rate": 1.974545959989108e-05,
|
||
|
|
"loss": 1.153,
|
||
|
|
"step": 143
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.35,
|
||
|
|
"learning_rate": 1.9739461479120727e-05,
|
||
|
|
"loss": 1.1888,
|
||
|
|
"step": 144
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.36,
|
||
|
|
"learning_rate": 1.9739461479120727e-05,
|
||
|
|
"loss": 1.109,
|
||
|
|
"step": 145
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.36,
|
||
|
|
"learning_rate": 1.973339444003735e-05,
|
||
|
|
"loss": 1.1372,
|
||
|
|
"step": 146
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.36,
|
||
|
|
"learning_rate": 1.9727258525572487e-05,
|
||
|
|
"loss": 1.1119,
|
||
|
|
"step": 147
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.36,
|
||
|
|
"learning_rate": 1.9721053779145057e-05,
|
||
|
|
"loss": 1.1979,
|
||
|
|
"step": 148
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.37,
|
||
|
|
"learning_rate": 1.9714780244661044e-05,
|
||
|
|
"loss": 1.1624,
|
||
|
|
"step": 149
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.37,
|
||
|
|
"learning_rate": 1.9708437966513196e-05,
|
||
|
|
"loss": 1.1357,
|
||
|
|
"step": 150
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.37,
|
||
|
|
"learning_rate": 1.9702026989580694e-05,
|
||
|
|
"loss": 1.0822,
|
||
|
|
"step": 151
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.37,
|
||
|
|
"learning_rate": 1.969554735922885e-05,
|
||
|
|
"loss": 1.1286,
|
||
|
|
"step": 152
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.38,
|
||
|
|
"learning_rate": 1.968899912130879e-05,
|
||
|
|
"loss": 1.1906,
|
||
|
|
"step": 153
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.38,
|
||
|
|
"learning_rate": 1.968899912130879e-05,
|
||
|
|
"loss": 1.1154,
|
||
|
|
"step": 154
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.38,
|
||
|
|
"learning_rate": 1.9682382322157103e-05,
|
||
|
|
"loss": 1.1316,
|
||
|
|
"step": 155
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.38,
|
||
|
|
"learning_rate": 1.9675697008595545e-05,
|
||
|
|
"loss": 1.1555,
|
||
|
|
"step": 156
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.39,
|
||
|
|
"learning_rate": 1.9668943227930686e-05,
|
||
|
|
"loss": 1.1378,
|
||
|
|
"step": 157
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.39,
|
||
|
|
"learning_rate": 1.966212102795358e-05,
|
||
|
|
"loss": 1.0963,
|
||
|
|
"step": 158
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.39,
|
||
|
|
"learning_rate": 1.966212102795358e-05,
|
||
|
|
"loss": 1.1602,
|
||
|
|
"step": 159
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.39,
|
||
|
|
"learning_rate": 1.965523045693944e-05,
|
||
|
|
"loss": 1.1456,
|
||
|
|
"step": 160
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4,
|
||
|
|
"learning_rate": 1.964827156364728e-05,
|
||
|
|
"loss": 1.0217,
|
||
|
|
"step": 161
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4,
|
||
|
|
"learning_rate": 1.964124439731957e-05,
|
||
|
|
"loss": 1.186,
|
||
|
|
"step": 162
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4,
|
||
|
|
"learning_rate": 1.9634149007681894e-05,
|
||
|
|
"loss": 1.1621,
|
||
|
|
"step": 163
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.4,
|
||
|
|
"learning_rate": 1.962698544494261e-05,
|
||
|
|
"loss": 1.198,
|
||
|
|
"step": 164
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.41,
|
||
|
|
"learning_rate": 1.962698544494261e-05,
|
||
|
|
"loss": 1.1155,
|
||
|
|
"step": 165
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.41,
|
||
|
|
"learning_rate": 1.9619753759792466e-05,
|
||
|
|
"loss": 1.2091,
|
||
|
|
"step": 166
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.41,
|
||
|
|
"learning_rate": 1.9619753759792466e-05,
|
||
|
|
"loss": 1.2163,
|
||
|
|
"step": 167
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.41,
|
||
|
|
"learning_rate": 1.961245400340427e-05,
|
||
|
|
"loss": 1.1876,
|
||
|
|
"step": 168
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.42,
|
||
|
|
"learning_rate": 1.961245400340427e-05,
|
||
|
|
"loss": 1.0896,
|
||
|
|
"step": 169
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.42,
|
||
|
|
"learning_rate": 1.9605086227432512e-05,
|
||
|
|
"loss": 1.0972,
|
||
|
|
"step": 170
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.42,
|
||
|
|
"learning_rate": 1.9597650484012997e-05,
|
||
|
|
"loss": 1.131,
|
||
|
|
"step": 171
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.42,
|
||
|
|
"learning_rate": 1.9590146825762476e-05,
|
||
|
|
"loss": 1.1095,
|
||
|
|
"step": 172
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.43,
|
||
|
|
"learning_rate": 1.9582575305778297e-05,
|
||
|
|
"loss": 1.119,
|
||
|
|
"step": 173
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.43,
|
||
|
|
"learning_rate": 1.9582575305778297e-05,
|
||
|
|
"loss": 1.153,
|
||
|
|
"step": 174
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.43,
|
||
|
|
"learning_rate": 1.9574935977637994e-05,
|
||
|
|
"loss": 1.0994,
|
||
|
|
"step": 175
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.43,
|
||
|
|
"learning_rate": 1.9567228895398936e-05,
|
||
|
|
"loss": 1.1747,
|
||
|
|
"step": 176
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.44,
|
||
|
|
"learning_rate": 1.955945411359792e-05,
|
||
|
|
"loss": 1.1566,
|
||
|
|
"step": 177
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.44,
|
||
|
|
"learning_rate": 1.9551611687250808e-05,
|
||
|
|
"loss": 1.1407,
|
||
|
|
"step": 178
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.44,
|
||
|
|
"learning_rate": 1.9543701671852127e-05,
|
||
|
|
"loss": 1.2034,
|
||
|
|
"step": 179
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.44,
|
||
|
|
"learning_rate": 1.9535724123374674e-05,
|
||
|
|
"loss": 1.1647,
|
||
|
|
"step": 180
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.45,
|
||
|
|
"learning_rate": 1.952767909826913e-05,
|
||
|
|
"loss": 1.2311,
|
||
|
|
"step": 181
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.45,
|
||
|
|
"learning_rate": 1.951956665346364e-05,
|
||
|
|
"loss": 1.1375,
|
||
|
|
"step": 182
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.45,
|
||
|
|
"learning_rate": 1.951138684636344e-05,
|
||
|
|
"loss": 1.2173,
|
||
|
|
"step": 183
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.45,
|
||
|
|
"learning_rate": 1.9503139734850426e-05,
|
||
|
|
"loss": 1.1083,
|
||
|
|
"step": 184
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.46,
|
||
|
|
"learning_rate": 1.9494825377282746e-05,
|
||
|
|
"loss": 1.1478,
|
||
|
|
"step": 185
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.46,
|
||
|
|
"learning_rate": 1.9486443832494414e-05,
|
||
|
|
"loss": 1.1635,
|
||
|
|
"step": 186
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.46,
|
||
|
|
"learning_rate": 1.9477995159794854e-05,
|
||
|
|
"loss": 1.2343,
|
||
|
|
"step": 187
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.46,
|
||
|
|
"learning_rate": 1.9469479418968506e-05,
|
||
|
|
"loss": 1.1554,
|
||
|
|
"step": 188
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.47,
|
||
|
|
"learning_rate": 1.9460896670274408e-05,
|
||
|
|
"loss": 1.1989,
|
||
|
|
"step": 189
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.47,
|
||
|
|
"learning_rate": 1.9452246974445743e-05,
|
||
|
|
"loss": 1.1154,
|
||
|
|
"step": 190
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.47,
|
||
|
|
"learning_rate": 1.9443530392689434e-05,
|
||
|
|
"loss": 1.1306,
|
||
|
|
"step": 191
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.47,
|
||
|
|
"learning_rate": 1.94347469866857e-05,
|
||
|
|
"loss": 1.124,
|
||
|
|
"step": 192
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.47,
|
||
|
|
"learning_rate": 1.9425896818587615e-05,
|
||
|
|
"loss": 1.1231,
|
||
|
|
"step": 193
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.48,
|
||
|
|
"learning_rate": 1.941697995102069e-05,
|
||
|
|
"loss": 1.1638,
|
||
|
|
"step": 194
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.48,
|
||
|
|
"learning_rate": 1.9407996447082394e-05,
|
||
|
|
"loss": 1.0767,
|
||
|
|
"step": 195
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.48,
|
||
|
|
"learning_rate": 1.939894637034174e-05,
|
||
|
|
"loss": 1.0996,
|
||
|
|
"step": 196
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.48,
|
||
|
|
"learning_rate": 1.9389829784838833e-05,
|
||
|
|
"loss": 1.1234,
|
||
|
|
"step": 197
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.49,
|
||
|
|
"learning_rate": 1.938064675508438e-05,
|
||
|
|
"loss": 1.0935,
|
||
|
|
"step": 198
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.49,
|
||
|
|
"learning_rate": 1.9371397346059286e-05,
|
||
|
|
"loss": 1.1513,
|
||
|
|
"step": 199
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.49,
|
||
|
|
"learning_rate": 1.936208162321415e-05,
|
||
|
|
"loss": 1.1359,
|
||
|
|
"step": 200
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.49,
|
||
|
|
"learning_rate": 1.9352699652468835e-05,
|
||
|
|
"loss": 1.2211,
|
||
|
|
"step": 201
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5,
|
||
|
|
"learning_rate": 1.9343251500211977e-05,
|
||
|
|
"loss": 1.1504,
|
||
|
|
"step": 202
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5,
|
||
|
|
"learning_rate": 1.933373723330053e-05,
|
||
|
|
"loss": 1.1768,
|
||
|
|
"step": 203
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5,
|
||
|
|
"learning_rate": 1.9324156919059286e-05,
|
||
|
|
"loss": 1.1934,
|
||
|
|
"step": 204
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5,
|
||
|
|
"learning_rate": 1.93145106252804e-05,
|
||
|
|
"loss": 1.1678,
|
||
|
|
"step": 205
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.51,
|
||
|
|
"learning_rate": 1.93145106252804e-05,
|
||
|
|
"loss": 1.1403,
|
||
|
|
"step": 206
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.51,
|
||
|
|
"learning_rate": 1.9304798420222918e-05,
|
||
|
|
"loss": 1.2295,
|
||
|
|
"step": 207
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.51,
|
||
|
|
"learning_rate": 1.9295020372612276e-05,
|
||
|
|
"loss": 1.1167,
|
||
|
|
"step": 208
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.51,
|
||
|
|
"learning_rate": 1.9285176551639826e-05,
|
||
|
|
"loss": 1.1411,
|
||
|
|
"step": 209
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.52,
|
||
|
|
"learning_rate": 1.9275267026962358e-05,
|
||
|
|
"loss": 1.1209,
|
||
|
|
"step": 210
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.52,
|
||
|
|
"learning_rate": 1.9265291868701584e-05,
|
||
|
|
"loss": 1.1431,
|
||
|
|
"step": 211
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.52,
|
||
|
|
"learning_rate": 1.9255251147443646e-05,
|
||
|
|
"loss": 1.0942,
|
||
|
|
"step": 212
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.52,
|
||
|
|
"learning_rate": 1.924514493423864e-05,
|
||
|
|
"loss": 1.2042,
|
||
|
|
"step": 213
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.53,
|
||
|
|
"learning_rate": 1.9234973300600074e-05,
|
||
|
|
"loss": 1.2075,
|
||
|
|
"step": 214
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.53,
|
||
|
|
"learning_rate": 1.92247363185044e-05,
|
||
|
|
"loss": 1.1006,
|
||
|
|
"step": 215
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.53,
|
||
|
|
"learning_rate": 1.9214434060390484e-05,
|
||
|
|
"loss": 1.1622,
|
||
|
|
"step": 216
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.53,
|
||
|
|
"learning_rate": 1.9204066599159094e-05,
|
||
|
|
"loss": 1.203,
|
||
|
|
"step": 217
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.54,
|
||
|
|
"learning_rate": 1.9193634008172396e-05,
|
||
|
|
"loss": 1.094,
|
||
|
|
"step": 218
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.54,
|
||
|
|
"learning_rate": 1.9183136361253417e-05,
|
||
|
|
"loss": 1.1161,
|
||
|
|
"step": 219
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.54,
|
||
|
|
"learning_rate": 1.917257373268554e-05,
|
||
|
|
"loss": 1.2007,
|
||
|
|
"step": 220
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.54,
|
||
|
|
"learning_rate": 1.916194619721196e-05,
|
||
|
|
"loss": 1.2133,
|
||
|
|
"step": 221
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.55,
|
||
|
|
"learning_rate": 1.915125383003518e-05,
|
||
|
|
"loss": 1.2108,
|
||
|
|
"step": 222
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.55,
|
||
|
|
"learning_rate": 1.914049670681646e-05,
|
||
|
|
"loss": 1.0996,
|
||
|
|
"step": 223
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.55,
|
||
|
|
"learning_rate": 1.912967490367528e-05,
|
||
|
|
"loss": 1.2501,
|
||
|
|
"step": 224
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.55,
|
||
|
|
"learning_rate": 1.9118788497188815e-05,
|
||
|
|
"loss": 1.1252,
|
||
|
|
"step": 225
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.56,
|
||
|
|
"learning_rate": 1.9107837564391376e-05,
|
||
|
|
"loss": 1.1589,
|
||
|
|
"step": 226
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.56,
|
||
|
|
"learning_rate": 1.9096822182773887e-05,
|
||
|
|
"loss": 1.2532,
|
||
|
|
"step": 227
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.56,
|
||
|
|
"learning_rate": 1.9085742430283322e-05,
|
||
|
|
"loss": 1.1482,
|
||
|
|
"step": 228
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.56,
|
||
|
|
"learning_rate": 1.907459838532215e-05,
|
||
|
|
"loss": 1.1188,
|
||
|
|
"step": 229
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.57,
|
||
|
|
"learning_rate": 1.9063390126747778e-05,
|
||
|
|
"loss": 1.1482,
|
||
|
|
"step": 230
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.57,
|
||
|
|
"learning_rate": 1.9052117733872025e-05,
|
||
|
|
"loss": 1.1335,
|
||
|
|
"step": 231
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.57,
|
||
|
|
"learning_rate": 1.904078128646052e-05,
|
||
|
|
"loss": 1.1648,
|
||
|
|
"step": 232
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.57,
|
||
|
|
"learning_rate": 1.902938086473215e-05,
|
||
|
|
"loss": 1.1477,
|
||
|
|
"step": 233
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.58,
|
||
|
|
"learning_rate": 1.901791654935852e-05,
|
||
|
|
"loss": 1.1836,
|
||
|
|
"step": 234
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.58,
|
||
|
|
"learning_rate": 1.9006388421463322e-05,
|
||
|
|
"loss": 1.2014,
|
||
|
|
"step": 235
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.58,
|
||
|
|
"learning_rate": 1.899479656262183e-05,
|
||
|
|
"loss": 1.1979,
|
||
|
|
"step": 236
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.58,
|
||
|
|
"learning_rate": 1.898314105486028e-05,
|
||
|
|
"loss": 1.1352,
|
||
|
|
"step": 237
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.59,
|
||
|
|
"learning_rate": 1.8971421980655295e-05,
|
||
|
|
"loss": 1.1553,
|
||
|
|
"step": 238
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.59,
|
||
|
|
"learning_rate": 1.8959639422933316e-05,
|
||
|
|
"loss": 1.1289,
|
||
|
|
"step": 239
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.59,
|
||
|
|
"learning_rate": 1.894779346506999e-05,
|
||
|
|
"loss": 1.1257,
|
||
|
|
"step": 240
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.59,
|
||
|
|
"learning_rate": 1.893588419088962e-05,
|
||
|
|
"loss": 1.1614,
|
||
|
|
"step": 241
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6,
|
||
|
|
"learning_rate": 1.892391168466452e-05,
|
||
|
|
"loss": 1.1656,
|
||
|
|
"step": 242
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6,
|
||
|
|
"learning_rate": 1.891187603111447e-05,
|
||
|
|
"loss": 1.1423,
|
||
|
|
"step": 243
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6,
|
||
|
|
"learning_rate": 1.8899777315406073e-05,
|
||
|
|
"loss": 1.1309,
|
||
|
|
"step": 244
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.6,
|
||
|
|
"learning_rate": 1.8887615623152188e-05,
|
||
|
|
"loss": 1.0227,
|
||
|
|
"step": 245
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.61,
|
||
|
|
"learning_rate": 1.88753910404113e-05,
|
||
|
|
"loss": 1.1151,
|
||
|
|
"step": 246
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.61,
|
||
|
|
"learning_rate": 1.8863103653686917e-05,
|
||
|
|
"loss": 1.1514,
|
||
|
|
"step": 247
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.61,
|
||
|
|
"learning_rate": 1.8850753549926967e-05,
|
||
|
|
"loss": 1.2559,
|
||
|
|
"step": 248
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.61,
|
||
|
|
"learning_rate": 1.8838340816523175e-05,
|
||
|
|
"loss": 1.2324,
|
||
|
|
"step": 249
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.62,
|
||
|
|
"learning_rate": 1.8838340816523175e-05,
|
||
|
|
"loss": 1.2488,
|
||
|
|
"step": 250
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.62,
|
||
|
|
"learning_rate": 1.8838340816523175e-05,
|
||
|
|
"loss": 1.1292,
|
||
|
|
"step": 251
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.62,
|
||
|
|
"learning_rate": 1.8825865541310438e-05,
|
||
|
|
"loss": 1.1782,
|
||
|
|
"step": 252
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.62,
|
||
|
|
"learning_rate": 1.8813327812566217e-05,
|
||
|
|
"loss": 1.1617,
|
||
|
|
"step": 253
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.63,
|
||
|
|
"learning_rate": 1.880072771900991e-05,
|
||
|
|
"loss": 1.1814,
|
||
|
|
"step": 254
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.63,
|
||
|
|
"learning_rate": 1.878806534980221e-05,
|
||
|
|
"loss": 1.1925,
|
||
|
|
"step": 255
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.63,
|
||
|
|
"learning_rate": 1.8775340794544497e-05,
|
||
|
|
"loss": 1.1433,
|
||
|
|
"step": 256
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.63,
|
||
|
|
"learning_rate": 1.876255414327818e-05,
|
||
|
|
"loss": 1.1688,
|
||
|
|
"step": 257
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.63,
|
||
|
|
"learning_rate": 1.8749705486484074e-05,
|
||
|
|
"loss": 1.1577,
|
||
|
|
"step": 258
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.64,
|
||
|
|
"learning_rate": 1.8736794915081765e-05,
|
||
|
|
"loss": 1.1985,
|
||
|
|
"step": 259
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.64,
|
||
|
|
"learning_rate": 1.8723822520428954e-05,
|
||
|
|
"loss": 1.0873,
|
||
|
|
"step": 260
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.64,
|
||
|
|
"learning_rate": 1.8710788394320807e-05,
|
||
|
|
"loss": 1.1821,
|
||
|
|
"step": 261
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.64,
|
||
|
|
"learning_rate": 1.8697692628989327e-05,
|
||
|
|
"loss": 1.1282,
|
||
|
|
"step": 262
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.65,
|
||
|
|
"learning_rate": 1.868453531710268e-05,
|
||
|
|
"loss": 1.1346,
|
||
|
|
"step": 263
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.65,
|
||
|
|
"learning_rate": 1.8671316551764552e-05,
|
||
|
|
"loss": 1.105,
|
||
|
|
"step": 264
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.65,
|
||
|
|
"learning_rate": 1.865803642651348e-05,
|
||
|
|
"loss": 1.2267,
|
||
|
|
"step": 265
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.65,
|
||
|
|
"learning_rate": 1.8644695035322203e-05,
|
||
|
|
"loss": 1.1812,
|
||
|
|
"step": 266
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.66,
|
||
|
|
"learning_rate": 1.8644695035322203e-05,
|
||
|
|
"loss": 1.2009,
|
||
|
|
"step": 267
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.66,
|
||
|
|
"learning_rate": 1.8631292472596978e-05,
|
||
|
|
"loss": 1.2195,
|
||
|
|
"step": 268
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.66,
|
||
|
|
"learning_rate": 1.8617828833176935e-05,
|
||
|
|
"loss": 1.1591,
|
||
|
|
"step": 269
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.66,
|
||
|
|
"learning_rate": 1.860430421233339e-05,
|
||
|
|
"loss": 1.087,
|
||
|
|
"step": 270
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.67,
|
||
|
|
"learning_rate": 1.859071870576918e-05,
|
||
|
|
"loss": 1.1992,
|
||
|
|
"step": 271
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.67,
|
||
|
|
"learning_rate": 1.857707240961797e-05,
|
||
|
|
"loss": 1.1432,
|
||
|
|
"step": 272
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.67,
|
||
|
|
"learning_rate": 1.8563365420443594e-05,
|
||
|
|
"loss": 1.131,
|
||
|
|
"step": 273
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.67,
|
||
|
|
"learning_rate": 1.854959783523936e-05,
|
||
|
|
"loss": 1.2095,
|
||
|
|
"step": 274
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.68,
|
||
|
|
"learning_rate": 1.854959783523936e-05,
|
||
|
|
"loss": 1.1089,
|
||
|
|
"step": 275
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.68,
|
||
|
|
"learning_rate": 1.853576975142736e-05,
|
||
|
|
"loss": 1.1882,
|
||
|
|
"step": 276
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.68,
|
||
|
|
"learning_rate": 1.852188126685779e-05,
|
||
|
|
"loss": 1.145,
|
||
|
|
"step": 277
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.68,
|
||
|
|
"learning_rate": 1.8507932479808254e-05,
|
||
|
|
"loss": 1.1407,
|
||
|
|
"step": 278
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.69,
|
||
|
|
"learning_rate": 1.8493923488983066e-05,
|
||
|
|
"loss": 1.219,
|
||
|
|
"step": 279
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.69,
|
||
|
|
"learning_rate": 1.847985439351256e-05,
|
||
|
|
"loss": 1.1448,
|
||
|
|
"step": 280
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.69,
|
||
|
|
"learning_rate": 1.846572529295237e-05,
|
||
|
|
"loss": 1.1969,
|
||
|
|
"step": 281
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.69,
|
||
|
|
"learning_rate": 1.845153628728274e-05,
|
||
|
|
"loss": 1.1975,
|
||
|
|
"step": 282
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7,
|
||
|
|
"learning_rate": 1.8437287476907828e-05,
|
||
|
|
"loss": 1.1798,
|
||
|
|
"step": 283
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7,
|
||
|
|
"learning_rate": 1.842297896265497e-05,
|
||
|
|
"loss": 1.1761,
|
||
|
|
"step": 284
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7,
|
||
|
|
"learning_rate": 1.8408610845773974e-05,
|
||
|
|
"loss": 1.1433,
|
||
|
|
"step": 285
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.7,
|
||
|
|
"learning_rate": 1.8394183227936418e-05,
|
||
|
|
"loss": 1.1364,
|
||
|
|
"step": 286
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.71,
|
||
|
|
"learning_rate": 1.8379696211234918e-05,
|
||
|
|
"loss": 1.1431,
|
||
|
|
"step": 287
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.71,
|
||
|
|
"learning_rate": 1.8365149898182403e-05,
|
||
|
|
"loss": 1.1093,
|
||
|
|
"step": 288
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.71,
|
||
|
|
"learning_rate": 1.8350544391711396e-05,
|
||
|
|
"loss": 1.0717,
|
||
|
|
"step": 289
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.71,
|
||
|
|
"learning_rate": 1.833587979517329e-05,
|
||
|
|
"loss": 1.1857,
|
||
|
|
"step": 290
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.72,
|
||
|
|
"learning_rate": 1.8321156212337604e-05,
|
||
|
|
"loss": 1.0927,
|
||
|
|
"step": 291
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.72,
|
||
|
|
"learning_rate": 1.8321156212337604e-05,
|
||
|
|
"loss": 1.1124,
|
||
|
|
"step": 292
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.72,
|
||
|
|
"learning_rate": 1.830637374739126e-05,
|
||
|
|
"loss": 1.2296,
|
||
|
|
"step": 293
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.72,
|
||
|
|
"learning_rate": 1.829153250493783e-05,
|
||
|
|
"loss": 1.1223,
|
||
|
|
"step": 294
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.73,
|
||
|
|
"learning_rate": 1.827663258999683e-05,
|
||
|
|
"loss": 1.1712,
|
||
|
|
"step": 295
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.73,
|
||
|
|
"learning_rate": 1.8261674108002925e-05,
|
||
|
|
"loss": 1.1244,
|
||
|
|
"step": 296
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.73,
|
||
|
|
"learning_rate": 1.8261674108002925e-05,
|
||
|
|
"loss": 1.1804,
|
||
|
|
"step": 297
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.73,
|
||
|
|
"learning_rate": 1.824665716480524e-05,
|
||
|
|
"loss": 1.0104,
|
||
|
|
"step": 298
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.74,
|
||
|
|
"learning_rate": 1.823158186666656e-05,
|
||
|
|
"loss": 1.1202,
|
||
|
|
"step": 299
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.74,
|
||
|
|
"learning_rate": 1.821644832026261e-05,
|
||
|
|
"loss": 1.1698,
|
||
|
|
"step": 300
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.74,
|
||
|
|
"learning_rate": 1.82012566326813e-05,
|
||
|
|
"loss": 1.1408,
|
||
|
|
"step": 301
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.74,
|
||
|
|
"learning_rate": 1.8186006911421937e-05,
|
||
|
|
"loss": 1.0759,
|
||
|
|
"step": 302
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.75,
|
||
|
|
"learning_rate": 1.817069926439451e-05,
|
||
|
|
"loss": 1.1292,
|
||
|
|
"step": 303
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.75,
|
||
|
|
"learning_rate": 1.8155333799918883e-05,
|
||
|
|
"loss": 1.201,
|
||
|
|
"step": 304
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.75,
|
||
|
|
"learning_rate": 1.8155333799918883e-05,
|
||
|
|
"loss": 1.1038,
|
||
|
|
"step": 305
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.75,
|
||
|
|
"learning_rate": 1.8139910626724058e-05,
|
||
|
|
"loss": 1.1499,
|
||
|
|
"step": 306
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.76,
|
||
|
|
"learning_rate": 1.8124429853947387e-05,
|
||
|
|
"loss": 1.1365,
|
||
|
|
"step": 307
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.76,
|
||
|
|
"learning_rate": 1.8108891591133812e-05,
|
||
|
|
"loss": 1.1646,
|
||
|
|
"step": 308
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.76,
|
||
|
|
"learning_rate": 1.809329594823509e-05,
|
||
|
|
"loss": 1.1382,
|
||
|
|
"step": 309
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.76,
|
||
|
|
"learning_rate": 1.8077643035609006e-05,
|
||
|
|
"loss": 1.1458,
|
||
|
|
"step": 310
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.77,
|
||
|
|
"learning_rate": 1.806193296401859e-05,
|
||
|
|
"loss": 1.1096,
|
||
|
|
"step": 311
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.77,
|
||
|
|
"learning_rate": 1.804616584463136e-05,
|
||
|
|
"loss": 1.1937,
|
||
|
|
"step": 312
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.77,
|
||
|
|
"learning_rate": 1.803034178901849e-05,
|
||
|
|
"loss": 1.1677,
|
||
|
|
"step": 313
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.77,
|
||
|
|
"learning_rate": 1.8014460909154058e-05,
|
||
|
|
"loss": 1.2202,
|
||
|
|
"step": 314
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.78,
|
||
|
|
"learning_rate": 1.799852331741425e-05,
|
||
|
|
"loss": 1.2097,
|
||
|
|
"step": 315
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.78,
|
||
|
|
"learning_rate": 1.799852331741425e-05,
|
||
|
|
"loss": 1.1376,
|
||
|
|
"step": 316
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.78,
|
||
|
|
"learning_rate": 1.7982529126576543e-05,
|
||
|
|
"loss": 1.1526,
|
||
|
|
"step": 317
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.78,
|
||
|
|
"learning_rate": 1.7966478449818925e-05,
|
||
|
|
"loss": 1.1528,
|
||
|
|
"step": 318
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.78,
|
||
|
|
"learning_rate": 1.7950371400719087e-05,
|
||
|
|
"loss": 1.1735,
|
||
|
|
"step": 319
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.79,
|
||
|
|
"learning_rate": 1.7934208093253625e-05,
|
||
|
|
"loss": 1.1243,
|
||
|
|
"step": 320
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.79,
|
||
|
|
"learning_rate": 1.7917988641797227e-05,
|
||
|
|
"loss": 1.15,
|
||
|
|
"step": 321
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.79,
|
||
|
|
"learning_rate": 1.7901713161121873e-05,
|
||
|
|
"loss": 1.1155,
|
||
|
|
"step": 322
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.79,
|
||
|
|
"learning_rate": 1.7885381766396008e-05,
|
||
|
|
"loss": 1.174,
|
||
|
|
"step": 323
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8,
|
||
|
|
"learning_rate": 1.786899457318374e-05,
|
||
|
|
"loss": 1.2247,
|
||
|
|
"step": 324
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8,
|
||
|
|
"learning_rate": 1.7852551697444017e-05,
|
||
|
|
"loss": 1.0521,
|
||
|
|
"step": 325
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8,
|
||
|
|
"learning_rate": 1.783605325552981e-05,
|
||
|
|
"loss": 1.1616,
|
||
|
|
"step": 326
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.8,
|
||
|
|
"learning_rate": 1.7819499364187282e-05,
|
||
|
|
"loss": 1.1757,
|
||
|
|
"step": 327
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.81,
|
||
|
|
"learning_rate": 1.780289014055497e-05,
|
||
|
|
"loss": 1.1687,
|
||
|
|
"step": 328
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.81,
|
||
|
|
"learning_rate": 1.7786225702162955e-05,
|
||
|
|
"loss": 1.1133,
|
||
|
|
"step": 329
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.81,
|
||
|
|
"learning_rate": 1.7769506166932026e-05,
|
||
|
|
"loss": 1.2429,
|
||
|
|
"step": 330
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.81,
|
||
|
|
"learning_rate": 1.7752731653172847e-05,
|
||
|
|
"loss": 1.0541,
|
||
|
|
"step": 331
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.82,
|
||
|
|
"learning_rate": 1.7735902279585118e-05,
|
||
|
|
"loss": 1.1295,
|
||
|
|
"step": 332
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.82,
|
||
|
|
"learning_rate": 1.7719018165256745e-05,
|
||
|
|
"loss": 1.0616,
|
||
|
|
"step": 333
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.82,
|
||
|
|
"learning_rate": 1.7702079429662986e-05,
|
||
|
|
"loss": 1.2172,
|
||
|
|
"step": 334
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.82,
|
||
|
|
"learning_rate": 1.7685086192665605e-05,
|
||
|
|
"loss": 1.0925,
|
||
|
|
"step": 335
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.83,
|
||
|
|
"learning_rate": 1.7668038574512045e-05,
|
||
|
|
"loss": 1.1621,
|
||
|
|
"step": 336
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.83,
|
||
|
|
"learning_rate": 1.7668038574512045e-05,
|
||
|
|
"loss": 1.1228,
|
||
|
|
"step": 337
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.83,
|
||
|
|
"learning_rate": 1.7650936695834536e-05,
|
||
|
|
"loss": 1.2114,
|
||
|
|
"step": 338
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.83,
|
||
|
|
"learning_rate": 1.763378067764929e-05,
|
||
|
|
"loss": 1.1061,
|
||
|
|
"step": 339
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.84,
|
||
|
|
"learning_rate": 1.7616570641355602e-05,
|
||
|
|
"loss": 1.1365,
|
||
|
|
"step": 340
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.84,
|
||
|
|
"learning_rate": 1.759930670873502e-05,
|
||
|
|
"loss": 1.1526,
|
||
|
|
"step": 341
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.84,
|
||
|
|
"learning_rate": 1.758198900195047e-05,
|
||
|
|
"loss": 1.1304,
|
||
|
|
"step": 342
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.84,
|
||
|
|
"learning_rate": 1.7564617643545395e-05,
|
||
|
|
"loss": 1.1383,
|
||
|
|
"step": 343
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.85,
|
||
|
|
"learning_rate": 1.7547192756442887e-05,
|
||
|
|
"loss": 1.1379,
|
||
|
|
"step": 344
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.85,
|
||
|
|
"learning_rate": 1.7547192756442887e-05,
|
||
|
|
"loss": 1.0709,
|
||
|
|
"step": 345
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.85,
|
||
|
|
"learning_rate": 1.7529714463944815e-05,
|
||
|
|
"loss": 1.1121,
|
||
|
|
"step": 346
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.85,
|
||
|
|
"learning_rate": 1.751218288973096e-05,
|
||
|
|
"loss": 1.2317,
|
||
|
|
"step": 347
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.86,
|
||
|
|
"learning_rate": 1.751218288973096e-05,
|
||
|
|
"loss": 1.1517,
|
||
|
|
"step": 348
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.86,
|
||
|
|
"learning_rate": 1.751218288973096e-05,
|
||
|
|
"loss": 1.142,
|
||
|
|
"step": 349
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.86,
|
||
|
|
"learning_rate": 1.7494598157858127e-05,
|
||
|
|
"loss": 1.1774,
|
||
|
|
"step": 350
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.86,
|
||
|
|
"learning_rate": 1.7476960392759284e-05,
|
||
|
|
"loss": 1.1382,
|
||
|
|
"step": 351
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.87,
|
||
|
|
"learning_rate": 1.7459269719242665e-05,
|
||
|
|
"loss": 1.1377,
|
||
|
|
"step": 352
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.87,
|
||
|
|
"learning_rate": 1.74415262624909e-05,
|
||
|
|
"loss": 1.1379,
|
||
|
|
"step": 353
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.87,
|
||
|
|
"learning_rate": 1.742373014806012e-05,
|
||
|
|
"loss": 1.1947,
|
||
|
|
"step": 354
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.87,
|
||
|
|
"learning_rate": 1.740588150187907e-05,
|
||
|
|
"loss": 1.1375,
|
||
|
|
"step": 355
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.88,
|
||
|
|
"learning_rate": 1.7387980450248222e-05,
|
||
|
|
"loss": 1.1951,
|
||
|
|
"step": 356
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.88,
|
||
|
|
"learning_rate": 1.7370027119838884e-05,
|
||
|
|
"loss": 1.1919,
|
||
|
|
"step": 357
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.88,
|
||
|
|
"learning_rate": 1.735202163769229e-05,
|
||
|
|
"loss": 1.0869,
|
||
|
|
"step": 358
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.88,
|
||
|
|
"learning_rate": 1.7333964131218714e-05,
|
||
|
|
"loss": 1.1057,
|
||
|
|
"step": 359
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.89,
|
||
|
|
"learning_rate": 1.7315854728196568e-05,
|
||
|
|
"loss": 1.2241,
|
||
|
|
"step": 360
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.89,
|
||
|
|
"learning_rate": 1.729769355677149e-05,
|
||
|
|
"loss": 1.0972,
|
||
|
|
"step": 361
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.89,
|
||
|
|
"learning_rate": 1.7279480745455433e-05,
|
||
|
|
"loss": 1.1912,
|
||
|
|
"step": 362
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.89,
|
||
|
|
"learning_rate": 1.7261216423125782e-05,
|
||
|
|
"loss": 1.0613,
|
||
|
|
"step": 363
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9,
|
||
|
|
"learning_rate": 1.724290071902441e-05,
|
||
|
|
"loss": 1.1275,
|
||
|
|
"step": 364
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9,
|
||
|
|
"learning_rate": 1.7224533762756775e-05,
|
||
|
|
"loss": 1.1917,
|
||
|
|
"step": 365
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9,
|
||
|
|
"learning_rate": 1.720611568429103e-05,
|
||
|
|
"loss": 1.1288,
|
||
|
|
"step": 366
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.9,
|
||
|
|
"learning_rate": 1.720611568429103e-05,
|
||
|
|
"loss": 1.1787,
|
||
|
|
"step": 367
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.91,
|
||
|
|
"learning_rate": 1.718764661395704e-05,
|
||
|
|
"loss": 1.1365,
|
||
|
|
"step": 368
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.91,
|
||
|
|
"learning_rate": 1.716912668244553e-05,
|
||
|
|
"loss": 1.2281,
|
||
|
|
"step": 369
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.91,
|
||
|
|
"learning_rate": 1.715055602080711e-05,
|
||
|
|
"loss": 1.1344,
|
||
|
|
"step": 370
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.91,
|
||
|
|
"learning_rate": 1.7131934760451385e-05,
|
||
|
|
"loss": 1.1644,
|
||
|
|
"step": 371
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.92,
|
||
|
|
"learning_rate": 1.7113263033145985e-05,
|
||
|
|
"loss": 1.13,
|
||
|
|
"step": 372
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.92,
|
||
|
|
"learning_rate": 1.7094540971015663e-05,
|
||
|
|
"loss": 1.1143,
|
||
|
|
"step": 373
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.92,
|
||
|
|
"learning_rate": 1.7075768706541355e-05,
|
||
|
|
"loss": 1.1438,
|
||
|
|
"step": 374
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.92,
|
||
|
|
"learning_rate": 1.7056946372559234e-05,
|
||
|
|
"loss": 1.137,
|
||
|
|
"step": 375
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.93,
|
||
|
|
"learning_rate": 1.7056946372559234e-05,
|
||
|
|
"loss": 1.0825,
|
||
|
|
"step": 376
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.93,
|
||
|
|
"learning_rate": 1.7038074102259775e-05,
|
||
|
|
"loss": 1.1002,
|
||
|
|
"step": 377
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.93,
|
||
|
|
"learning_rate": 1.7019152029186817e-05,
|
||
|
|
"loss": 1.0477,
|
||
|
|
"step": 378
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.93,
|
||
|
|
"learning_rate": 1.70001802872366e-05,
|
||
|
|
"loss": 1.0862,
|
||
|
|
"step": 379
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.94,
|
||
|
|
"learning_rate": 1.6981159010656847e-05,
|
||
|
|
"loss": 1.2227,
|
||
|
|
"step": 380
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.94,
|
||
|
|
"learning_rate": 1.6962088334045785e-05,
|
||
|
|
"loss": 1.1665,
|
||
|
|
"step": 381
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.94,
|
||
|
|
"learning_rate": 1.6962088334045785e-05,
|
||
|
|
"loss": 1.1262,
|
||
|
|
"step": 382
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.94,
|
||
|
|
"learning_rate": 1.694296839235121e-05,
|
||
|
|
"loss": 1.1451,
|
||
|
|
"step": 383
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.94,
|
||
|
|
"learning_rate": 1.692379932086953e-05,
|
||
|
|
"loss": 1.1458,
|
||
|
|
"step": 384
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.95,
|
||
|
|
"learning_rate": 1.692379932086953e-05,
|
||
|
|
"loss": 1.1896,
|
||
|
|
"step": 385
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.95,
|
||
|
|
"learning_rate": 1.6904581255244802e-05,
|
||
|
|
"loss": 1.1233,
|
||
|
|
"step": 386
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.95,
|
||
|
|
"learning_rate": 1.688531433146777e-05,
|
||
|
|
"loss": 1.1583,
|
||
|
|
"step": 387
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.95,
|
||
|
|
"learning_rate": 1.6865998685874923e-05,
|
||
|
|
"loss": 1.2036,
|
||
|
|
"step": 388
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.96,
|
||
|
|
"learning_rate": 1.6865998685874923e-05,
|
||
|
|
"loss": 1.1578,
|
||
|
|
"step": 389
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.96,
|
||
|
|
"learning_rate": 1.6846634455147498e-05,
|
||
|
|
"loss": 1.1066,
|
||
|
|
"step": 390
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.96,
|
||
|
|
"learning_rate": 1.6827221776310532e-05,
|
||
|
|
"loss": 1.1191,
|
||
|
|
"step": 391
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.96,
|
||
|
|
"learning_rate": 1.6807760786731905e-05,
|
||
|
|
"loss": 1.1265,
|
||
|
|
"step": 392
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.97,
|
||
|
|
"learning_rate": 1.6788251624121335e-05,
|
||
|
|
"loss": 1.2217,
|
||
|
|
"step": 393
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.97,
|
||
|
|
"learning_rate": 1.6768694426529432e-05,
|
||
|
|
"loss": 1.1626,
|
||
|
|
"step": 394
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.97,
|
||
|
|
"learning_rate": 1.6749089332346714e-05,
|
||
|
|
"loss": 1.1616,
|
||
|
|
"step": 395
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.97,
|
||
|
|
"learning_rate": 1.672943648030261e-05,
|
||
|
|
"loss": 1.1559,
|
||
|
|
"step": 396
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.98,
|
||
|
|
"learning_rate": 1.6709736009464504e-05,
|
||
|
|
"loss": 1.193,
|
||
|
|
"step": 397
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.98,
|
||
|
|
"learning_rate": 1.668998805923675e-05,
|
||
|
|
"loss": 1.1917,
|
||
|
|
"step": 398
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.98,
|
||
|
|
"learning_rate": 1.6670192769359643e-05,
|
||
|
|
"loss": 1.1331,
|
||
|
|
"step": 399
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.98,
|
||
|
|
"learning_rate": 1.6670192769359643e-05,
|
||
|
|
"loss": 1.1127,
|
||
|
|
"step": 400
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.99,
|
||
|
|
"learning_rate": 1.6670192769359643e-05,
|
||
|
|
"loss": 1.2107,
|
||
|
|
"step": 401
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.99,
|
||
|
|
"learning_rate": 1.6650350279908497e-05,
|
||
|
|
"loss": 1.0895,
|
||
|
|
"step": 402
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.99,
|
||
|
|
"learning_rate": 1.6630460731292597e-05,
|
||
|
|
"loss": 1.146,
|
||
|
|
"step": 403
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.99,
|
||
|
|
"learning_rate": 1.661052426425424e-05,
|
||
|
|
"loss": 1.1694,
|
||
|
|
"step": 404
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0,
|
||
|
|
"learning_rate": 1.6590541019867722e-05,
|
||
|
|
"loss": 1.1629,
|
||
|
|
"step": 405
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0,
|
||
|
|
"learning_rate": 1.6570511139538348e-05,
|
||
|
|
"loss": 1.265,
|
||
|
|
"step": 406
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0,
|
||
|
|
"learning_rate": 1.655043476500142e-05,
|
||
|
|
"loss": 1.0617,
|
||
|
|
"step": 407
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.0,
|
||
|
|
"learning_rate": 1.6530312038321247e-05,
|
||
|
|
"loss": 1.0527,
|
||
|
|
"step": 408
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.01,
|
||
|
|
"learning_rate": 1.6510143101890136e-05,
|
||
|
|
"loss": 1.1541,
|
||
|
|
"step": 409
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.01,
|
||
|
|
"learning_rate": 1.6489928098427383e-05,
|
||
|
|
"loss": 1.1049,
|
||
|
|
"step": 410
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.01,
|
||
|
|
"learning_rate": 1.6469667170978258e-05,
|
||
|
|
"loss": 1.1096,
|
||
|
|
"step": 411
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.01,
|
||
|
|
"learning_rate": 1.6449360462913005e-05,
|
||
|
|
"loss": 1.1301,
|
||
|
|
"step": 412
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.02,
|
||
|
|
"learning_rate": 1.642900811792582e-05,
|
||
|
|
"loss": 1.0167,
|
||
|
|
"step": 413
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.02,
|
||
|
|
"learning_rate": 1.640861028003383e-05,
|
||
|
|
"loss": 1.132,
|
||
|
|
"step": 414
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.02,
|
||
|
|
"learning_rate": 1.6388167093576083e-05,
|
||
|
|
"loss": 1.0682,
|
||
|
|
"step": 415
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.02,
|
||
|
|
"learning_rate": 1.6388167093576083e-05,
|
||
|
|
"loss": 1.0953,
|
||
|
|
"step": 416
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.03,
|
||
|
|
"learning_rate": 1.6367678703212515e-05,
|
||
|
|
"loss": 1.0795,
|
||
|
|
"step": 417
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.03,
|
||
|
|
"learning_rate": 1.6367678703212515e-05,
|
||
|
|
"loss": 1.0682,
|
||
|
|
"step": 418
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.03,
|
||
|
|
"learning_rate": 1.6347145253922942e-05,
|
||
|
|
"loss": 1.2072,
|
||
|
|
"step": 419
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.03,
|
||
|
|
"learning_rate": 1.6347145253922942e-05,
|
||
|
|
"loss": 1.0967,
|
||
|
|
"step": 420
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.04,
|
||
|
|
"learning_rate": 1.632656689100602e-05,
|
||
|
|
"loss": 1.0267,
|
||
|
|
"step": 421
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.04,
|
||
|
|
"learning_rate": 1.632656689100602e-05,
|
||
|
|
"loss": 1.0864,
|
||
|
|
"step": 422
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.04,
|
||
|
|
"learning_rate": 1.6305943760078226e-05,
|
||
|
|
"loss": 1.0798,
|
||
|
|
"step": 423
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.04,
|
||
|
|
"learning_rate": 1.628527600707283e-05,
|
||
|
|
"loss": 1.1086,
|
||
|
|
"step": 424
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.05,
|
||
|
|
"learning_rate": 1.6264563778238834e-05,
|
||
|
|
"loss": 1.063,
|
||
|
|
"step": 425
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.05,
|
||
|
|
"learning_rate": 1.6243807220139988e-05,
|
||
|
|
"loss": 1.1038,
|
||
|
|
"step": 426
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.05,
|
||
|
|
"learning_rate": 1.6223006479653708e-05,
|
||
|
|
"loss": 1.0511,
|
||
|
|
"step": 427
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.05,
|
||
|
|
"learning_rate": 1.6202161703970057e-05,
|
||
|
|
"loss": 1.1235,
|
||
|
|
"step": 428
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.06,
|
||
|
|
"learning_rate": 1.6181273040590696e-05,
|
||
|
|
"loss": 1.1478,
|
||
|
|
"step": 429
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.06,
|
||
|
|
"learning_rate": 1.616034063732785e-05,
|
||
|
|
"loss": 1.0928,
|
||
|
|
"step": 430
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.06,
|
||
|
|
"learning_rate": 1.613936464230325e-05,
|
||
|
|
"loss": 1.1063,
|
||
|
|
"step": 431
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.06,
|
||
|
|
"learning_rate": 1.6118345203947093e-05,
|
||
|
|
"loss": 1.094,
|
||
|
|
"step": 432
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.07,
|
||
|
|
"learning_rate": 1.6118345203947093e-05,
|
||
|
|
"loss": 1.0798,
|
||
|
|
"step": 433
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.07,
|
||
|
|
"learning_rate": 1.6097282470996997e-05,
|
||
|
|
"loss": 1.14,
|
||
|
|
"step": 434
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.07,
|
||
|
|
"learning_rate": 1.6076176592496926e-05,
|
||
|
|
"loss": 1.1188,
|
||
|
|
"step": 435
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.07,
|
||
|
|
"learning_rate": 1.605502771779616e-05,
|
||
|
|
"loss": 1.1085,
|
||
|
|
"step": 436
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.08,
|
||
|
|
"learning_rate": 1.603383599654823e-05,
|
||
|
|
"loss": 1.1473,
|
||
|
|
"step": 437
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.08,
|
||
|
|
"learning_rate": 1.603383599654823e-05,
|
||
|
|
"loss": 1.0793,
|
||
|
|
"step": 438
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.08,
|
||
|
|
"learning_rate": 1.601260157870985e-05,
|
||
|
|
"loss": 1.0635,
|
||
|
|
"step": 439
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.08,
|
||
|
|
"learning_rate": 1.599132461453987e-05,
|
||
|
|
"loss": 1.0208,
|
||
|
|
"step": 440
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.09,
|
||
|
|
"learning_rate": 1.5970005254598204e-05,
|
||
|
|
"loss": 1.0608,
|
||
|
|
"step": 441
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.09,
|
||
|
|
"learning_rate": 1.594864364974476e-05,
|
||
|
|
"loss": 1.0763,
|
||
|
|
"step": 442
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.09,
|
||
|
|
"learning_rate": 1.592723995113839e-05,
|
||
|
|
"loss": 1.1966,
|
||
|
|
"step": 443
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.09,
|
||
|
|
"learning_rate": 1.592723995113839e-05,
|
||
|
|
"loss": 1.1704,
|
||
|
|
"step": 444
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1,
|
||
|
|
"learning_rate": 1.5905794310235808e-05,
|
||
|
|
"loss": 1.1497,
|
||
|
|
"step": 445
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1,
|
||
|
|
"learning_rate": 1.5884306878790512e-05,
|
||
|
|
"loss": 1.1819,
|
||
|
|
"step": 446
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1,
|
||
|
|
"learning_rate": 1.586277780885172e-05,
|
||
|
|
"loss": 1.0923,
|
||
|
|
"step": 447
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1,
|
||
|
|
"learning_rate": 1.58412072527633e-05,
|
||
|
|
"loss": 1.0444,
|
||
|
|
"step": 448
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.1,
|
||
|
|
"learning_rate": 1.5819595363162682e-05,
|
||
|
|
"loss": 1.0953,
|
||
|
|
"step": 449
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.11,
|
||
|
|
"learning_rate": 1.5797942292979767e-05,
|
||
|
|
"loss": 1.119,
|
||
|
|
"step": 450
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.11,
|
||
|
|
"learning_rate": 1.577624819543587e-05,
|
||
|
|
"loss": 1.0497,
|
||
|
|
"step": 451
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.11,
|
||
|
|
"learning_rate": 1.5754513224042625e-05,
|
||
|
|
"loss": 1.1818,
|
||
|
|
"step": 452
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.11,
|
||
|
|
"learning_rate": 1.573273753260089e-05,
|
||
|
|
"loss": 1.1151,
|
||
|
|
"step": 453
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.12,
|
||
|
|
"learning_rate": 1.571092127519967e-05,
|
||
|
|
"loss": 1.063,
|
||
|
|
"step": 454
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.12,
|
||
|
|
"learning_rate": 1.568906460621502e-05,
|
||
|
|
"loss": 1.1638,
|
||
|
|
"step": 455
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.12,
|
||
|
|
"learning_rate": 1.566716768030896e-05,
|
||
|
|
"loss": 1.0469,
|
||
|
|
"step": 456
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.12,
|
||
|
|
"learning_rate": 1.5645230652428367e-05,
|
||
|
|
"loss": 1.1851,
|
||
|
|
"step": 457
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.13,
|
||
|
|
"learning_rate": 1.5623253677803897e-05,
|
||
|
|
"loss": 1.1276,
|
||
|
|
"step": 458
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.13,
|
||
|
|
"learning_rate": 1.5601236911948876e-05,
|
||
|
|
"loss": 1.1357,
|
||
|
|
"step": 459
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.13,
|
||
|
|
"learning_rate": 1.5579180510658187e-05,
|
||
|
|
"loss": 1.1378,
|
||
|
|
"step": 460
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.13,
|
||
|
|
"learning_rate": 1.5557084630007206e-05,
|
||
|
|
"loss": 1.0875,
|
||
|
|
"step": 461
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.14,
|
||
|
|
"learning_rate": 1.5534949426350642e-05,
|
||
|
|
"loss": 1.1506,
|
||
|
|
"step": 462
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.14,
|
||
|
|
"learning_rate": 1.551277505632149e-05,
|
||
|
|
"loss": 1.0844,
|
||
|
|
"step": 463
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.14,
|
||
|
|
"learning_rate": 1.549056167682987e-05,
|
||
|
|
"loss": 1.0969,
|
||
|
|
"step": 464
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.14,
|
||
|
|
"learning_rate": 1.546830944506196e-05,
|
||
|
|
"loss": 1.1575,
|
||
|
|
"step": 465
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.15,
|
||
|
|
"learning_rate": 1.544601851847885e-05,
|
||
|
|
"loss": 1.1057,
|
||
|
|
"step": 466
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.15,
|
||
|
|
"learning_rate": 1.542368905481545e-05,
|
||
|
|
"loss": 1.0743,
|
||
|
|
"step": 467
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.15,
|
||
|
|
"learning_rate": 1.5401321212079366e-05,
|
||
|
|
"loss": 1.126,
|
||
|
|
"step": 468
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.15,
|
||
|
|
"learning_rate": 1.5378915148549772e-05,
|
||
|
|
"loss": 1.1383,
|
||
|
|
"step": 469
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.16,
|
||
|
|
"learning_rate": 1.5356471022776315e-05,
|
||
|
|
"loss": 1.0252,
|
||
|
|
"step": 470
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.16,
|
||
|
|
"learning_rate": 1.5333988993577958e-05,
|
||
|
|
"loss": 1.0526,
|
||
|
|
"step": 471
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.16,
|
||
|
|
"learning_rate": 1.5311469220041903e-05,
|
||
|
|
"loss": 1.092,
|
||
|
|
"step": 472
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.16,
|
||
|
|
"learning_rate": 1.5288911861522413e-05,
|
||
|
|
"loss": 1.1172,
|
||
|
|
"step": 473
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.17,
|
||
|
|
"learning_rate": 1.526631707763972e-05,
|
||
|
|
"loss": 1.1228,
|
||
|
|
"step": 474
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.17,
|
||
|
|
"learning_rate": 1.5243685028278888e-05,
|
||
|
|
"loss": 1.1605,
|
||
|
|
"step": 475
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.17,
|
||
|
|
"learning_rate": 1.5221015873588672e-05,
|
||
|
|
"loss": 1.1598,
|
||
|
|
"step": 476
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.17,
|
||
|
|
"learning_rate": 1.5198309773980397e-05,
|
||
|
|
"loss": 1.0659,
|
||
|
|
"step": 477
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.18,
|
||
|
|
"learning_rate": 1.5175566890126812e-05,
|
||
|
|
"loss": 1.1201,
|
||
|
|
"step": 478
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.18,
|
||
|
|
"learning_rate": 1.5152787382960968e-05,
|
||
|
|
"loss": 1.1096,
|
||
|
|
"step": 479
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.18,
|
||
|
|
"learning_rate": 1.5152787382960968e-05,
|
||
|
|
"loss": 1.1627,
|
||
|
|
"step": 480
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.18,
|
||
|
|
"learning_rate": 1.5152787382960968e-05,
|
||
|
|
"loss": 1.0902,
|
||
|
|
"step": 481
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.19,
|
||
|
|
"learning_rate": 1.5129971413675055e-05,
|
||
|
|
"loss": 1.1086,
|
||
|
|
"step": 482
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.19,
|
||
|
|
"learning_rate": 1.510711914371929e-05,
|
||
|
|
"loss": 1.0882,
|
||
|
|
"step": 483
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.19,
|
||
|
|
"learning_rate": 1.510711914371929e-05,
|
||
|
|
"loss": 1.1937,
|
||
|
|
"step": 484
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.19,
|
||
|
|
"learning_rate": 1.5084230734800754e-05,
|
||
|
|
"loss": 1.1105,
|
||
|
|
"step": 485
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2,
|
||
|
|
"learning_rate": 1.5084230734800754e-05,
|
||
|
|
"loss": 1.0682,
|
||
|
|
"step": 486
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2,
|
||
|
|
"learning_rate": 1.5084230734800754e-05,
|
||
|
|
"loss": 1.0745,
|
||
|
|
"step": 487
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2,
|
||
|
|
"learning_rate": 1.5084230734800754e-05,
|
||
|
|
"loss": 1.1605,
|
||
|
|
"step": 488
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.2,
|
||
|
|
"learning_rate": 1.5061306348882252e-05,
|
||
|
|
"loss": 1.1498,
|
||
|
|
"step": 489
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.21,
|
||
|
|
"learning_rate": 1.5038346148181178e-05,
|
||
|
|
"loss": 1.0581,
|
||
|
|
"step": 490
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.21,
|
||
|
|
"learning_rate": 1.5015350295168344e-05,
|
||
|
|
"loss": 1.152,
|
||
|
|
"step": 491
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.21,
|
||
|
|
"learning_rate": 1.4992318952566862e-05,
|
||
|
|
"loss": 1.1908,
|
||
|
|
"step": 492
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.21,
|
||
|
|
"learning_rate": 1.4992318952566862e-05,
|
||
|
|
"loss": 1.2114,
|
||
|
|
"step": 493
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.22,
|
||
|
|
"learning_rate": 1.4969252283350964e-05,
|
||
|
|
"loss": 1.1433,
|
||
|
|
"step": 494
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.22,
|
||
|
|
"learning_rate": 1.4969252283350964e-05,
|
||
|
|
"loss": 1.1324,
|
||
|
|
"step": 495
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.22,
|
||
|
|
"learning_rate": 1.4946150450744859e-05,
|
||
|
|
"loss": 1.1062,
|
||
|
|
"step": 496
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.22,
|
||
|
|
"learning_rate": 1.4923013618221584e-05,
|
||
|
|
"loss": 1.0513,
|
||
|
|
"step": 497
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.23,
|
||
|
|
"learning_rate": 1.4899841949501845e-05,
|
||
|
|
"loss": 1.1276,
|
||
|
|
"step": 498
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.23,
|
||
|
|
"learning_rate": 1.4876635608552845e-05,
|
||
|
|
"loss": 1.0997,
|
||
|
|
"step": 499
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.23,
|
||
|
|
"learning_rate": 1.4876635608552845e-05,
|
||
|
|
"loss": 1.0966,
|
||
|
|
"step": 500
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.23,
|
||
|
|
"learning_rate": 1.4853394759587146e-05,
|
||
|
|
"loss": 1.0841,
|
||
|
|
"step": 501
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.24,
|
||
|
|
"learning_rate": 1.4830119567061484e-05,
|
||
|
|
"loss": 1.0822,
|
||
|
|
"step": 502
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.24,
|
||
|
|
"learning_rate": 1.4806810195675627e-05,
|
||
|
|
"loss": 1.1052,
|
||
|
|
"step": 503
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.24,
|
||
|
|
"learning_rate": 1.4783466810371195e-05,
|
||
|
|
"loss": 1.1415,
|
||
|
|
"step": 504
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.24,
|
||
|
|
"learning_rate": 1.4760089576330493e-05,
|
||
|
|
"loss": 1.1243,
|
||
|
|
"step": 505
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.25,
|
||
|
|
"learning_rate": 1.4736678658975357e-05,
|
||
|
|
"loss": 1.078,
|
||
|
|
"step": 506
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.25,
|
||
|
|
"learning_rate": 1.471323422396596e-05,
|
||
|
|
"loss": 1.0579,
|
||
|
|
"step": 507
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.25,
|
||
|
|
"learning_rate": 1.4689756437199658e-05,
|
||
|
|
"loss": 1.1387,
|
||
|
|
"step": 508
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.25,
|
||
|
|
"learning_rate": 1.4689756437199658e-05,
|
||
|
|
"loss": 1.0594,
|
||
|
|
"step": 509
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.25,
|
||
|
|
"learning_rate": 1.4666245464809818e-05,
|
||
|
|
"loss": 1.1052,
|
||
|
|
"step": 510
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.26,
|
||
|
|
"learning_rate": 1.4666245464809818e-05,
|
||
|
|
"loss": 1.0599,
|
||
|
|
"step": 511
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.26,
|
||
|
|
"learning_rate": 1.4642701473164618e-05,
|
||
|
|
"loss": 1.1142,
|
||
|
|
"step": 512
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.26,
|
||
|
|
"learning_rate": 1.4619124628865904e-05,
|
||
|
|
"loss": 1.0957,
|
||
|
|
"step": 513
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.26,
|
||
|
|
"learning_rate": 1.4619124628865904e-05,
|
||
|
|
"loss": 1.1183,
|
||
|
|
"step": 514
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.27,
|
||
|
|
"learning_rate": 1.459551509874798e-05,
|
||
|
|
"loss": 1.1294,
|
||
|
|
"step": 515
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.27,
|
||
|
|
"learning_rate": 1.4571873049876452e-05,
|
||
|
|
"loss": 1.1012,
|
||
|
|
"step": 516
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.27,
|
||
|
|
"learning_rate": 1.454819864954703e-05,
|
||
|
|
"loss": 1.0487,
|
||
|
|
"step": 517
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.27,
|
||
|
|
"learning_rate": 1.4524492065284344e-05,
|
||
|
|
"loss": 1.0667,
|
||
|
|
"step": 518
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.28,
|
||
|
|
"learning_rate": 1.4500753464840775e-05,
|
||
|
|
"loss": 1.0578,
|
||
|
|
"step": 519
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.28,
|
||
|
|
"learning_rate": 1.4476983016195245e-05,
|
||
|
|
"loss": 1.0845,
|
||
|
|
"step": 520
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.28,
|
||
|
|
"learning_rate": 1.4453180887552052e-05,
|
||
|
|
"loss": 1.1243,
|
||
|
|
"step": 521
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.28,
|
||
|
|
"learning_rate": 1.4429347247339656e-05,
|
||
|
|
"loss": 0.9873,
|
||
|
|
"step": 522
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.29,
|
||
|
|
"learning_rate": 1.4405482264209512e-05,
|
||
|
|
"loss": 1.1144,
|
||
|
|
"step": 523
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.29,
|
||
|
|
"learning_rate": 1.4381586107034849e-05,
|
||
|
|
"loss": 1.072,
|
||
|
|
"step": 524
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.29,
|
||
|
|
"learning_rate": 1.4381586107034849e-05,
|
||
|
|
"loss": 1.0671,
|
||
|
|
"step": 525
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.29,
|
||
|
|
"learning_rate": 1.4357658944909496e-05,
|
||
|
|
"loss": 1.0966,
|
||
|
|
"step": 526
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3,
|
||
|
|
"learning_rate": 1.4333700947146686e-05,
|
||
|
|
"loss": 1.0818,
|
||
|
|
"step": 527
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3,
|
||
|
|
"learning_rate": 1.4309712283277839e-05,
|
||
|
|
"loss": 1.0918,
|
||
|
|
"step": 528
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3,
|
||
|
|
"learning_rate": 1.4309712283277839e-05,
|
||
|
|
"loss": 1.0109,
|
||
|
|
"step": 529
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.3,
|
||
|
|
"learning_rate": 1.4285693123051385e-05,
|
||
|
|
"loss": 1.0969,
|
||
|
|
"step": 530
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.31,
|
||
|
|
"learning_rate": 1.4261643636431539e-05,
|
||
|
|
"loss": 1.119,
|
||
|
|
"step": 531
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.31,
|
||
|
|
"learning_rate": 1.4237563993597133e-05,
|
||
|
|
"loss": 1.1245,
|
||
|
|
"step": 532
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.31,
|
||
|
|
"learning_rate": 1.4237563993597133e-05,
|
||
|
|
"loss": 1.152,
|
||
|
|
"step": 533
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.31,
|
||
|
|
"learning_rate": 1.4213454364940362e-05,
|
||
|
|
"loss": 1.1732,
|
||
|
|
"step": 534
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.32,
|
||
|
|
"learning_rate": 1.4189314921065629e-05,
|
||
|
|
"loss": 1.062,
|
||
|
|
"step": 535
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.32,
|
||
|
|
"learning_rate": 1.4165145832788305e-05,
|
||
|
|
"loss": 1.1283,
|
||
|
|
"step": 536
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.32,
|
||
|
|
"learning_rate": 1.4140947271133536e-05,
|
||
|
|
"loss": 1.0471,
|
||
|
|
"step": 537
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.32,
|
||
|
|
"learning_rate": 1.4116719407335022e-05,
|
||
|
|
"loss": 1.0625,
|
||
|
|
"step": 538
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.33,
|
||
|
|
"learning_rate": 1.4092462412833811e-05,
|
||
|
|
"loss": 1.1946,
|
||
|
|
"step": 539
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.33,
|
||
|
|
"learning_rate": 1.406817645927709e-05,
|
||
|
|
"loss": 1.0807,
|
||
|
|
"step": 540
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.33,
|
||
|
|
"learning_rate": 1.4043861718516964e-05,
|
||
|
|
"loss": 1.1708,
|
||
|
|
"step": 541
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.33,
|
||
|
|
"learning_rate": 1.4019518362609239e-05,
|
||
|
|
"loss": 1.0088,
|
||
|
|
"step": 542
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.34,
|
||
|
|
"learning_rate": 1.399514656381221e-05,
|
||
|
|
"loss": 1.0797,
|
||
|
|
"step": 543
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.34,
|
||
|
|
"learning_rate": 1.3970746494585439e-05,
|
||
|
|
"loss": 1.0838,
|
||
|
|
"step": 544
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.34,
|
||
|
|
"learning_rate": 1.3946318327588534e-05,
|
||
|
|
"loss": 1.0151,
|
||
|
|
"step": 545
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.34,
|
||
|
|
"learning_rate": 1.3946318327588534e-05,
|
||
|
|
"loss": 1.0555,
|
||
|
|
"step": 546
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.35,
|
||
|
|
"learning_rate": 1.3921862235679929e-05,
|
||
|
|
"loss": 1.1403,
|
||
|
|
"step": 547
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.35,
|
||
|
|
"learning_rate": 1.389737839191566e-05,
|
||
|
|
"loss": 1.0558,
|
||
|
|
"step": 548
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.35,
|
||
|
|
"learning_rate": 1.3872866969548143e-05,
|
||
|
|
"loss": 1.1357,
|
||
|
|
"step": 549
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.35,
|
||
|
|
"learning_rate": 1.3872866969548143e-05,
|
||
|
|
"loss": 1.1351,
|
||
|
|
"step": 550
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.36,
|
||
|
|
"learning_rate": 1.384832814202494e-05,
|
||
|
|
"loss": 1.0569,
|
||
|
|
"step": 551
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.36,
|
||
|
|
"learning_rate": 1.3823762082987544e-05,
|
||
|
|
"loss": 1.0458,
|
||
|
|
"step": 552
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.36,
|
||
|
|
"learning_rate": 1.3799168966270139e-05,
|
||
|
|
"loss": 1.1161,
|
||
|
|
"step": 553
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.36,
|
||
|
|
"learning_rate": 1.3774548965898371e-05,
|
||
|
|
"loss": 1.1268,
|
||
|
|
"step": 554
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.37,
|
||
|
|
"learning_rate": 1.3749902256088125e-05,
|
||
|
|
"loss": 1.0403,
|
||
|
|
"step": 555
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.37,
|
||
|
|
"learning_rate": 1.3725229011244294e-05,
|
||
|
|
"loss": 1.0115,
|
||
|
|
"step": 556
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.37,
|
||
|
|
"learning_rate": 1.3700529405959517e-05,
|
||
|
|
"loss": 1.15,
|
||
|
|
"step": 557
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.37,
|
||
|
|
"learning_rate": 1.3675803615012993e-05,
|
||
|
|
"loss": 1.053,
|
||
|
|
"step": 558
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.38,
|
||
|
|
"learning_rate": 1.3651051813369188e-05,
|
||
|
|
"loss": 1.1024,
|
||
|
|
"step": 559
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.38,
|
||
|
|
"learning_rate": 1.3626274176176645e-05,
|
||
|
|
"loss": 1.0702,
|
||
|
|
"step": 560
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.38,
|
||
|
|
"learning_rate": 1.3601470878766714e-05,
|
||
|
|
"loss": 1.0325,
|
||
|
|
"step": 561
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.38,
|
||
|
|
"learning_rate": 1.3601470878766714e-05,
|
||
|
|
"loss": 1.1449,
|
||
|
|
"step": 562
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.39,
|
||
|
|
"learning_rate": 1.3576642096652322e-05,
|
||
|
|
"loss": 1.1638,
|
||
|
|
"step": 563
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.39,
|
||
|
|
"learning_rate": 1.3551788005526738e-05,
|
||
|
|
"loss": 1.0787,
|
||
|
|
"step": 564
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.39,
|
||
|
|
"learning_rate": 1.3526908781262314e-05,
|
||
|
|
"loss": 1.1074,
|
||
|
|
"step": 565
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.39,
|
||
|
|
"learning_rate": 1.3526908781262314e-05,
|
||
|
|
"loss": 0.9868,
|
||
|
|
"step": 566
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4,
|
||
|
|
"learning_rate": 1.3502004599909255e-05,
|
||
|
|
"loss": 1.0782,
|
||
|
|
"step": 567
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4,
|
||
|
|
"learning_rate": 1.3477075637694362e-05,
|
||
|
|
"loss": 1.1022,
|
||
|
|
"step": 568
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4,
|
||
|
|
"learning_rate": 1.3452122071019797e-05,
|
||
|
|
"loss": 1.1311,
|
||
|
|
"step": 569
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.4,
|
||
|
|
"learning_rate": 1.3427144076461818e-05,
|
||
|
|
"loss": 1.1138,
|
||
|
|
"step": 570
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.41,
|
||
|
|
"learning_rate": 1.3402141830769551e-05,
|
||
|
|
"loss": 1.0972,
|
||
|
|
"step": 571
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.41,
|
||
|
|
"learning_rate": 1.3377115510863716e-05,
|
||
|
|
"loss": 1.0664,
|
||
|
|
"step": 572
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.41,
|
||
|
|
"learning_rate": 1.3352065293835399e-05,
|
||
|
|
"loss": 1.1277,
|
||
|
|
"step": 573
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.41,
|
||
|
|
"learning_rate": 1.3326991356944776e-05,
|
||
|
|
"loss": 1.1859,
|
||
|
|
"step": 574
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.41,
|
||
|
|
"learning_rate": 1.3301893877619874e-05,
|
||
|
|
"loss": 1.0863,
|
||
|
|
"step": 575
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.42,
|
||
|
|
"learning_rate": 1.3276773033455312e-05,
|
||
|
|
"loss": 1.1276,
|
||
|
|
"step": 576
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.42,
|
||
|
|
"learning_rate": 1.3251629002211042e-05,
|
||
|
|
"loss": 1.1127,
|
||
|
|
"step": 577
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.42,
|
||
|
|
"learning_rate": 1.322646196181109e-05,
|
||
|
|
"loss": 1.0815,
|
||
|
|
"step": 578
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.42,
|
||
|
|
"learning_rate": 1.3201272090342303e-05,
|
||
|
|
"loss": 1.0121,
|
||
|
|
"step": 579
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.43,
|
||
|
|
"learning_rate": 1.3176059566053083e-05,
|
||
|
|
"loss": 1.0896,
|
||
|
|
"step": 580
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.43,
|
||
|
|
"learning_rate": 1.3150824567352128e-05,
|
||
|
|
"loss": 1.1186,
|
||
|
|
"step": 581
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.43,
|
||
|
|
"learning_rate": 1.3125567272807167e-05,
|
||
|
|
"loss": 1.1824,
|
||
|
|
"step": 582
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.43,
|
||
|
|
"learning_rate": 1.3100287861143703e-05,
|
||
|
|
"loss": 1.1333,
|
||
|
|
"step": 583
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.44,
|
||
|
|
"learning_rate": 1.3074986511243741e-05,
|
||
|
|
"loss": 1.1035,
|
||
|
|
"step": 584
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.44,
|
||
|
|
"learning_rate": 1.3074986511243741e-05,
|
||
|
|
"loss": 1.0916,
|
||
|
|
"step": 585
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.44,
|
||
|
|
"learning_rate": 1.3049663402144528e-05,
|
||
|
|
"loss": 1.2061,
|
||
|
|
"step": 586
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.44,
|
||
|
|
"learning_rate": 1.302431871303728e-05,
|
||
|
|
"loss": 1.1206,
|
||
|
|
"step": 587
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.45,
|
||
|
|
"learning_rate": 1.2998952623265917e-05,
|
||
|
|
"loss": 1.1941,
|
||
|
|
"step": 588
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.45,
|
||
|
|
"learning_rate": 1.2973565312325798e-05,
|
||
|
|
"loss": 1.1687,
|
||
|
|
"step": 589
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.45,
|
||
|
|
"learning_rate": 1.2948156959862446e-05,
|
||
|
|
"loss": 1.0668,
|
||
|
|
"step": 590
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.45,
|
||
|
|
"learning_rate": 1.2922727745670276e-05,
|
||
|
|
"loss": 1.1493,
|
||
|
|
"step": 591
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.46,
|
||
|
|
"learning_rate": 1.2897277849691326e-05,
|
||
|
|
"loss": 1.0341,
|
||
|
|
"step": 592
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.46,
|
||
|
|
"learning_rate": 1.2871807452013977e-05,
|
||
|
|
"loss": 1.1093,
|
||
|
|
"step": 593
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.46,
|
||
|
|
"learning_rate": 1.2871807452013977e-05,
|
||
|
|
"loss": 1.0038,
|
||
|
|
"step": 594
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.46,
|
||
|
|
"learning_rate": 1.28463167328717e-05,
|
||
|
|
"loss": 1.1001,
|
||
|
|
"step": 595
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.47,
|
||
|
|
"learning_rate": 1.2820805872641745e-05,
|
||
|
|
"loss": 1.0977,
|
||
|
|
"step": 596
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.47,
|
||
|
|
"learning_rate": 1.2795275051843893e-05,
|
||
|
|
"loss": 1.1301,
|
||
|
|
"step": 597
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.47,
|
||
|
|
"learning_rate": 1.276972445113917e-05,
|
||
|
|
"loss": 1.0771,
|
||
|
|
"step": 598
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.47,
|
||
|
|
"learning_rate": 1.2744154251328573e-05,
|
||
|
|
"loss": 1.1446,
|
||
|
|
"step": 599
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.48,
|
||
|
|
"learning_rate": 1.2718564633351773e-05,
|
||
|
|
"loss": 1.0806,
|
||
|
|
"step": 600
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.48,
|
||
|
|
"learning_rate": 1.2692955778285865e-05,
|
||
|
|
"loss": 1.0326,
|
||
|
|
"step": 601
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.48,
|
||
|
|
"learning_rate": 1.266732786734405e-05,
|
||
|
|
"loss": 1.0992,
|
||
|
|
"step": 602
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.48,
|
||
|
|
"learning_rate": 1.266732786734405e-05,
|
||
|
|
"loss": 1.0253,
|
||
|
|
"step": 603
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.49,
|
||
|
|
"learning_rate": 1.2641681081874394e-05,
|
||
|
|
"loss": 1.1312,
|
||
|
|
"step": 604
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.49,
|
||
|
|
"learning_rate": 1.2616015603358497e-05,
|
||
|
|
"loss": 1.1362,
|
||
|
|
"step": 605
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.49,
|
||
|
|
"learning_rate": 1.2590331613410261e-05,
|
||
|
|
"loss": 1.1169,
|
||
|
|
"step": 606
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.49,
|
||
|
|
"learning_rate": 1.2564629293774561e-05,
|
||
|
|
"loss": 1.0657,
|
||
|
|
"step": 607
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5,
|
||
|
|
"learning_rate": 1.253890882632598e-05,
|
||
|
|
"loss": 1.0873,
|
||
|
|
"step": 608
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5,
|
||
|
|
"learning_rate": 1.2513170393067527e-05,
|
||
|
|
"loss": 1.1539,
|
||
|
|
"step": 609
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5,
|
||
|
|
"learning_rate": 1.2513170393067527e-05,
|
||
|
|
"loss": 1.1084,
|
||
|
|
"step": 610
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.5,
|
||
|
|
"learning_rate": 1.2513170393067527e-05,
|
||
|
|
"loss": 1.1232,
|
||
|
|
"step": 611
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.51,
|
||
|
|
"learning_rate": 1.2487414176129322e-05,
|
||
|
|
"loss": 1.0896,
|
||
|
|
"step": 612
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.51,
|
||
|
|
"learning_rate": 1.246164035776735e-05,
|
||
|
|
"loss": 1.1459,
|
||
|
|
"step": 613
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.51,
|
||
|
|
"learning_rate": 1.2435849120362123e-05,
|
||
|
|
"loss": 1.2163,
|
||
|
|
"step": 614
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.51,
|
||
|
|
"learning_rate": 1.2410040646417431e-05,
|
||
|
|
"loss": 1.135,
|
||
|
|
"step": 615
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.52,
|
||
|
|
"learning_rate": 1.2384215118559027e-05,
|
||
|
|
"loss": 1.1274,
|
||
|
|
"step": 616
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.52,
|
||
|
|
"learning_rate": 1.235837271953334e-05,
|
||
|
|
"loss": 1.1326,
|
||
|
|
"step": 617
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.52,
|
||
|
|
"learning_rate": 1.2332513632206183e-05,
|
||
|
|
"loss": 1.1253,
|
||
|
|
"step": 618
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.52,
|
||
|
|
"learning_rate": 1.2306638039561455e-05,
|
||
|
|
"loss": 1.0119,
|
||
|
|
"step": 619
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.53,
|
||
|
|
"learning_rate": 1.2280746124699864e-05,
|
||
|
|
"loss": 1.02,
|
||
|
|
"step": 620
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.53,
|
||
|
|
"learning_rate": 1.2280746124699864e-05,
|
||
|
|
"loss": 1.1543,
|
||
|
|
"step": 621
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.53,
|
||
|
|
"learning_rate": 1.2254838070837596e-05,
|
||
|
|
"loss": 1.0669,
|
||
|
|
"step": 622
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.53,
|
||
|
|
"learning_rate": 1.2228914061305059e-05,
|
||
|
|
"loss": 1.064,
|
||
|
|
"step": 623
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.54,
|
||
|
|
"learning_rate": 1.2202974279545554e-05,
|
||
|
|
"loss": 1.0327,
|
||
|
|
"step": 624
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.54,
|
||
|
|
"learning_rate": 1.2177018909113994e-05,
|
||
|
|
"loss": 1.0624,
|
||
|
|
"step": 625
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.54,
|
||
|
|
"learning_rate": 1.21510481336756e-05,
|
||
|
|
"loss": 1.0961,
|
||
|
|
"step": 626
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.54,
|
||
|
|
"learning_rate": 1.2125062137004602e-05,
|
||
|
|
"loss": 1.1146,
|
||
|
|
"step": 627
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.55,
|
||
|
|
"learning_rate": 1.2099061102982939e-05,
|
||
|
|
"loss": 1.0938,
|
||
|
|
"step": 628
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.55,
|
||
|
|
"learning_rate": 1.2073045215598953e-05,
|
||
|
|
"loss": 1.0607,
|
||
|
|
"step": 629
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.55,
|
||
|
|
"learning_rate": 1.2073045215598953e-05,
|
||
|
|
"loss": 1.1032,
|
||
|
|
"step": 630
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.55,
|
||
|
|
"learning_rate": 1.20470146589461e-05,
|
||
|
|
"loss": 1.1542,
|
||
|
|
"step": 631
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.56,
|
||
|
|
"learning_rate": 1.2020969617221627e-05,
|
||
|
|
"loss": 1.1351,
|
||
|
|
"step": 632
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.56,
|
||
|
|
"learning_rate": 1.199491027472529e-05,
|
||
|
|
"loss": 1.1101,
|
||
|
|
"step": 633
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.56,
|
||
|
|
"learning_rate": 1.1968836815858038e-05,
|
||
|
|
"loss": 1.1252,
|
||
|
|
"step": 634
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.56,
|
||
|
|
"learning_rate": 1.1942749425120704e-05,
|
||
|
|
"loss": 1.0601,
|
||
|
|
"step": 635
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.57,
|
||
|
|
"learning_rate": 1.1916648287112714e-05,
|
||
|
|
"loss": 1.1064,
|
||
|
|
"step": 636
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.57,
|
||
|
|
"learning_rate": 1.1890533586530766e-05,
|
||
|
|
"loss": 1.1221,
|
||
|
|
"step": 637
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.57,
|
||
|
|
"learning_rate": 1.1864405508167532e-05,
|
||
|
|
"loss": 1.0848,
|
||
|
|
"step": 638
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.57,
|
||
|
|
"learning_rate": 1.1864405508167532e-05,
|
||
|
|
"loss": 1.1277,
|
||
|
|
"step": 639
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.57,
|
||
|
|
"learning_rate": 1.1838264236910348e-05,
|
||
|
|
"loss": 1.0979,
|
||
|
|
"step": 640
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.58,
|
||
|
|
"learning_rate": 1.1838264236910348e-05,
|
||
|
|
"loss": 1.0496,
|
||
|
|
"step": 641
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.58,
|
||
|
|
"learning_rate": 1.1812109957739907e-05,
|
||
|
|
"loss": 1.0531,
|
||
|
|
"step": 642
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.58,
|
||
|
|
"learning_rate": 1.1785942855728945e-05,
|
||
|
|
"loss": 1.0413,
|
||
|
|
"step": 643
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.58,
|
||
|
|
"learning_rate": 1.1759763116040936e-05,
|
||
|
|
"loss": 1.1525,
|
||
|
|
"step": 644
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.59,
|
||
|
|
"learning_rate": 1.1733570923928785e-05,
|
||
|
|
"loss": 1.0346,
|
||
|
|
"step": 645
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.59,
|
||
|
|
"learning_rate": 1.1733570923928785e-05,
|
||
|
|
"loss": 1.0079,
|
||
|
|
"step": 646
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.59,
|
||
|
|
"learning_rate": 1.1707366464733501e-05,
|
||
|
|
"loss": 1.1522,
|
||
|
|
"step": 647
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.59,
|
||
|
|
"learning_rate": 1.1681149923882913e-05,
|
||
|
|
"loss": 1.1947,
|
||
|
|
"step": 648
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6,
|
||
|
|
"learning_rate": 1.1654921486890327e-05,
|
||
|
|
"loss": 1.1049,
|
||
|
|
"step": 649
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6,
|
||
|
|
"learning_rate": 1.1628681339353244e-05,
|
||
|
|
"loss": 1.1212,
|
||
|
|
"step": 650
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6,
|
||
|
|
"learning_rate": 1.1602429666952015e-05,
|
||
|
|
"loss": 1.0158,
|
||
|
|
"step": 651
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.6,
|
||
|
|
"learning_rate": 1.1576166655448558e-05,
|
||
|
|
"loss": 1.125,
|
||
|
|
"step": 652
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.61,
|
||
|
|
"learning_rate": 1.1549892490685018e-05,
|
||
|
|
"loss": 1.1191,
|
||
|
|
"step": 653
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.61,
|
||
|
|
"learning_rate": 1.1523607358582462e-05,
|
||
|
|
"loss": 0.9951,
|
||
|
|
"step": 654
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.61,
|
||
|
|
"learning_rate": 1.149731144513958e-05,
|
||
|
|
"loss": 1.0928,
|
||
|
|
"step": 655
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.61,
|
||
|
|
"learning_rate": 1.1471004936431327e-05,
|
||
|
|
"loss": 1.0634,
|
||
|
|
"step": 656
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.62,
|
||
|
|
"learning_rate": 1.144468801860766e-05,
|
||
|
|
"loss": 1.081,
|
||
|
|
"step": 657
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.62,
|
||
|
|
"learning_rate": 1.1418360877892165e-05,
|
||
|
|
"loss": 1.0573,
|
||
|
|
"step": 658
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.62,
|
||
|
|
"learning_rate": 1.1392023700580796e-05,
|
||
|
|
"loss": 1.0894,
|
||
|
|
"step": 659
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.62,
|
||
|
|
"learning_rate": 1.1365676673040502e-05,
|
||
|
|
"loss": 1.0959,
|
||
|
|
"step": 660
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.63,
|
||
|
|
"learning_rate": 1.133931998170795e-05,
|
||
|
|
"loss": 1.116,
|
||
|
|
"step": 661
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.63,
|
||
|
|
"learning_rate": 1.1312953813088183e-05,
|
||
|
|
"loss": 1.0826,
|
||
|
|
"step": 662
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.63,
|
||
|
|
"learning_rate": 1.1286578353753313e-05,
|
||
|
|
"loss": 1.079,
|
||
|
|
"step": 663
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.63,
|
||
|
|
"learning_rate": 1.1260193790341186e-05,
|
||
|
|
"loss": 1.114,
|
||
|
|
"step": 664
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.64,
|
||
|
|
"learning_rate": 1.1233800309554083e-05,
|
||
|
|
"loss": 1.0547,
|
||
|
|
"step": 665
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.64,
|
||
|
|
"learning_rate": 1.1207398098157371e-05,
|
||
|
|
"loss": 1.0623,
|
||
|
|
"step": 666
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.64,
|
||
|
|
"learning_rate": 1.1180987342978209e-05,
|
||
|
|
"loss": 0.9716,
|
||
|
|
"step": 667
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.64,
|
||
|
|
"learning_rate": 1.1154568230904204e-05,
|
||
|
|
"loss": 1.0553,
|
||
|
|
"step": 668
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.65,
|
||
|
|
"learning_rate": 1.1128140948882107e-05,
|
||
|
|
"loss": 1.084,
|
||
|
|
"step": 669
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.65,
|
||
|
|
"learning_rate": 1.1101705683916473e-05,
|
||
|
|
"loss": 1.1569,
|
||
|
|
"step": 670
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.65,
|
||
|
|
"learning_rate": 1.1101705683916473e-05,
|
||
|
|
"loss": 1.1716,
|
||
|
|
"step": 671
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.65,
|
||
|
|
"learning_rate": 1.1075262623068352e-05,
|
||
|
|
"loss": 1.1011,
|
||
|
|
"step": 672
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.66,
|
||
|
|
"learning_rate": 1.1048811953453955e-05,
|
||
|
|
"loss": 1.0421,
|
||
|
|
"step": 673
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.66,
|
||
|
|
"learning_rate": 1.1022353862243338e-05,
|
||
|
|
"loss": 1.1387,
|
||
|
|
"step": 674
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.66,
|
||
|
|
"learning_rate": 1.0995888536659067e-05,
|
||
|
|
"loss": 1.1285,
|
||
|
|
"step": 675
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.66,
|
||
|
|
"learning_rate": 1.096941616397491e-05,
|
||
|
|
"loss": 1.0415,
|
||
|
|
"step": 676
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.67,
|
||
|
|
"learning_rate": 1.0942936931514492e-05,
|
||
|
|
"loss": 1.0013,
|
||
|
|
"step": 677
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.67,
|
||
|
|
"learning_rate": 1.0916451026649981e-05,
|
||
|
|
"loss": 1.0699,
|
||
|
|
"step": 678
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.67,
|
||
|
|
"learning_rate": 1.0916451026649981e-05,
|
||
|
|
"loss": 1.1105,
|
||
|
|
"step": 679
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.67,
|
||
|
|
"learning_rate": 1.088995863680077e-05,
|
||
|
|
"loss": 1.1356,
|
||
|
|
"step": 680
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.68,
|
||
|
|
"learning_rate": 1.0863459949432122e-05,
|
||
|
|
"loss": 1.0638,
|
||
|
|
"step": 681
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.68,
|
||
|
|
"learning_rate": 1.0836955152053883e-05,
|
||
|
|
"loss": 1.0228,
|
||
|
|
"step": 682
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.68,
|
||
|
|
"learning_rate": 1.081044443221912e-05,
|
||
|
|
"loss": 1.0095,
|
||
|
|
"step": 683
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.68,
|
||
|
|
"learning_rate": 1.0783927977522819e-05,
|
||
|
|
"loss": 1.0925,
|
||
|
|
"step": 684
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.69,
|
||
|
|
"learning_rate": 1.0757405975600534e-05,
|
||
|
|
"loss": 1.1006,
|
||
|
|
"step": 685
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.69,
|
||
|
|
"learning_rate": 1.0730878614127087e-05,
|
||
|
|
"loss": 1.1201,
|
||
|
|
"step": 686
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.69,
|
||
|
|
"learning_rate": 1.0704346080815218e-05,
|
||
|
|
"loss": 1.1514,
|
||
|
|
"step": 687
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.69,
|
||
|
|
"learning_rate": 1.0677808563414256e-05,
|
||
|
|
"loss": 1.0657,
|
||
|
|
"step": 688
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7,
|
||
|
|
"learning_rate": 1.0651266249708816e-05,
|
||
|
|
"loss": 1.0907,
|
||
|
|
"step": 689
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7,
|
||
|
|
"learning_rate": 1.0624719327517434e-05,
|
||
|
|
"loss": 1.1276,
|
||
|
|
"step": 690
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7,
|
||
|
|
"learning_rate": 1.0598167984691276e-05,
|
||
|
|
"loss": 1.1104,
|
||
|
|
"step": 691
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.7,
|
||
|
|
"learning_rate": 1.057161240911277e-05,
|
||
|
|
"loss": 1.0683,
|
||
|
|
"step": 692
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.71,
|
||
|
|
"learning_rate": 1.0545052788694312e-05,
|
||
|
|
"loss": 1.1155,
|
||
|
|
"step": 693
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.71,
|
||
|
|
"learning_rate": 1.0518489311376905e-05,
|
||
|
|
"loss": 1.0613,
|
||
|
|
"step": 694
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.71,
|
||
|
|
"learning_rate": 1.0491922165128853e-05,
|
||
|
|
"loss": 1.0657,
|
||
|
|
"step": 695
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.71,
|
||
|
|
"learning_rate": 1.0465351537944429e-05,
|
||
|
|
"loss": 1.0921,
|
||
|
|
"step": 696
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.72,
|
||
|
|
"learning_rate": 1.043877761784252e-05,
|
||
|
|
"loss": 1.143,
|
||
|
|
"step": 697
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.72,
|
||
|
|
"learning_rate": 1.043877761784252e-05,
|
||
|
|
"loss": 1.0966,
|
||
|
|
"step": 698
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.72,
|
||
|
|
"learning_rate": 1.0412200592865331e-05,
|
||
|
|
"loss": 1.0698,
|
||
|
|
"step": 699
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.72,
|
||
|
|
"learning_rate": 1.0412200592865331e-05,
|
||
|
|
"loss": 1.0695,
|
||
|
|
"step": 700
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.73,
|
||
|
|
"learning_rate": 1.0385620651077024e-05,
|
||
|
|
"loss": 1.121,
|
||
|
|
"step": 701
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.73,
|
||
|
|
"learning_rate": 1.0359037980562416e-05,
|
||
|
|
"loss": 1.0819,
|
||
|
|
"step": 702
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.73,
|
||
|
|
"learning_rate": 1.0332452769425619e-05,
|
||
|
|
"loss": 1.134,
|
||
|
|
"step": 703
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.73,
|
||
|
|
"learning_rate": 1.0305865205788728e-05,
|
||
|
|
"loss": 1.0645,
|
||
|
|
"step": 704
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.73,
|
||
|
|
"learning_rate": 1.0279275477790487e-05,
|
||
|
|
"loss": 1.1163,
|
||
|
|
"step": 705
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.74,
|
||
|
|
"learning_rate": 1.0252683773584953e-05,
|
||
|
|
"loss": 1.1289,
|
||
|
|
"step": 706
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.74,
|
||
|
|
"learning_rate": 1.0226090281340168e-05,
|
||
|
|
"loss": 1.0632,
|
||
|
|
"step": 707
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.74,
|
||
|
|
"learning_rate": 1.0199495189236828e-05,
|
||
|
|
"loss": 1.0571,
|
||
|
|
"step": 708
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.74,
|
||
|
|
"learning_rate": 1.0172898685466947e-05,
|
||
|
|
"loss": 1.1067,
|
||
|
|
"step": 709
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.75,
|
||
|
|
"learning_rate": 1.0146300958232528e-05,
|
||
|
|
"loss": 1.1304,
|
||
|
|
"step": 710
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.75,
|
||
|
|
"learning_rate": 1.0119702195744236e-05,
|
||
|
|
"loss": 1.0648,
|
||
|
|
"step": 711
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.75,
|
||
|
|
"learning_rate": 1.0093102586220056e-05,
|
||
|
|
"loss": 1.0753,
|
||
|
|
"step": 712
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.75,
|
||
|
|
"learning_rate": 1.0066502317883969e-05,
|
||
|
|
"loss": 1.1494,
|
||
|
|
"step": 713
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.76,
|
||
|
|
"learning_rate": 1.0039901578964619e-05,
|
||
|
|
"loss": 1.103,
|
||
|
|
"step": 714
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.76,
|
||
|
|
"learning_rate": 1.0013300557693981e-05,
|
||
|
|
"loss": 1.105,
|
||
|
|
"step": 715
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.76,
|
||
|
|
"learning_rate": 9.986699442306025e-06,
|
||
|
|
"loss": 1.1234,
|
||
|
|
"step": 716
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.76,
|
||
|
|
"learning_rate": 9.960098421035383e-06,
|
||
|
|
"loss": 1.1089,
|
||
|
|
"step": 717
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.77,
|
||
|
|
"learning_rate": 9.960098421035383e-06,
|
||
|
|
"loss": 1.1128,
|
||
|
|
"step": 718
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.77,
|
||
|
|
"learning_rate": 9.933497682116035e-06,
|
||
|
|
"loss": 1.0996,
|
||
|
|
"step": 719
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.77,
|
||
|
|
"learning_rate": 9.906897413779949e-06,
|
||
|
|
"loss": 1.121,
|
||
|
|
"step": 720
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.77,
|
||
|
|
"learning_rate": 9.88029780425577e-06,
|
||
|
|
"loss": 1.0723,
|
||
|
|
"step": 721
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.78,
|
||
|
|
"learning_rate": 9.853699041767473e-06,
|
||
|
|
"loss": 1.085,
|
||
|
|
"step": 722
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.78,
|
||
|
|
"learning_rate": 9.827101314533056e-06,
|
||
|
|
"loss": 1.0817,
|
||
|
|
"step": 723
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.78,
|
||
|
|
"learning_rate": 9.800504810763176e-06,
|
||
|
|
"loss": 1.0649,
|
||
|
|
"step": 724
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.78,
|
||
|
|
"learning_rate": 9.773909718659831e-06,
|
||
|
|
"loss": 1.1562,
|
||
|
|
"step": 725
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.79,
|
||
|
|
"learning_rate": 9.747316226415052e-06,
|
||
|
|
"loss": 1.0294,
|
||
|
|
"step": 726
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.79,
|
||
|
|
"learning_rate": 9.720724522209518e-06,
|
||
|
|
"loss": 1.0389,
|
||
|
|
"step": 727
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.79,
|
||
|
|
"learning_rate": 9.720724522209518e-06,
|
||
|
|
"loss": 1.0775,
|
||
|
|
"step": 728
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.79,
|
||
|
|
"learning_rate": 9.694134794211277e-06,
|
||
|
|
"loss": 1.0804,
|
||
|
|
"step": 729
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8,
|
||
|
|
"learning_rate": 9.667547230574386e-06,
|
||
|
|
"loss": 1.0589,
|
||
|
|
"step": 730
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8,
|
||
|
|
"learning_rate": 9.64096201943759e-06,
|
||
|
|
"loss": 1.1451,
|
||
|
|
"step": 731
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8,
|
||
|
|
"learning_rate": 9.61437934892298e-06,
|
||
|
|
"loss": 1.0446,
|
||
|
|
"step": 732
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.8,
|
||
|
|
"learning_rate": 9.587799407134672e-06,
|
||
|
|
"loss": 1.088,
|
||
|
|
"step": 733
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.81,
|
||
|
|
"learning_rate": 9.587799407134672e-06,
|
||
|
|
"loss": 1.1246,
|
||
|
|
"step": 734
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.81,
|
||
|
|
"learning_rate": 9.56122238215748e-06,
|
||
|
|
"loss": 1.0897,
|
||
|
|
"step": 735
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.81,
|
||
|
|
"learning_rate": 9.534648462055576e-06,
|
||
|
|
"loss": 1.1113,
|
||
|
|
"step": 736
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.81,
|
||
|
|
"learning_rate": 9.50807783487115e-06,
|
||
|
|
"loss": 1.1456,
|
||
|
|
"step": 737
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.82,
|
||
|
|
"learning_rate": 9.481510688623098e-06,
|
||
|
|
"loss": 1.0886,
|
||
|
|
"step": 738
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.82,
|
||
|
|
"learning_rate": 9.454947211305691e-06,
|
||
|
|
"loss": 1.1339,
|
||
|
|
"step": 739
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.82,
|
||
|
|
"learning_rate": 9.42838759088723e-06,
|
||
|
|
"loss": 1.1105,
|
||
|
|
"step": 740
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.82,
|
||
|
|
"learning_rate": 9.401832015308728e-06,
|
||
|
|
"loss": 1.098,
|
||
|
|
"step": 741
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.83,
|
||
|
|
"learning_rate": 9.375280672482567e-06,
|
||
|
|
"loss": 1.0945,
|
||
|
|
"step": 742
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.83,
|
||
|
|
"learning_rate": 9.348733750291186e-06,
|
||
|
|
"loss": 1.0507,
|
||
|
|
"step": 743
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.83,
|
||
|
|
"learning_rate": 9.322191436585745e-06,
|
||
|
|
"loss": 1.0621,
|
||
|
|
"step": 744
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.83,
|
||
|
|
"learning_rate": 9.295653919184787e-06,
|
||
|
|
"loss": 1.0752,
|
||
|
|
"step": 745
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.84,
|
||
|
|
"learning_rate": 9.269121385872915e-06,
|
||
|
|
"loss": 1.0413,
|
||
|
|
"step": 746
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.84,
|
||
|
|
"learning_rate": 9.242594024399467e-06,
|
||
|
|
"loss": 1.0219,
|
||
|
|
"step": 747
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.84,
|
||
|
|
"learning_rate": 9.216072022477183e-06,
|
||
|
|
"loss": 1.0307,
|
||
|
|
"step": 748
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.84,
|
||
|
|
"learning_rate": 9.189555567780882e-06,
|
||
|
|
"loss": 1.1017,
|
||
|
|
"step": 749
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.85,
|
||
|
|
"learning_rate": 9.189555567780882e-06,
|
||
|
|
"loss": 1.1019,
|
||
|
|
"step": 750
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.85,
|
||
|
|
"learning_rate": 9.16304484794612e-06,
|
||
|
|
"loss": 1.0908,
|
||
|
|
"step": 751
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.85,
|
||
|
|
"learning_rate": 9.13654005056788e-06,
|
||
|
|
"loss": 1.1263,
|
||
|
|
"step": 752
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.85,
|
||
|
|
"learning_rate": 9.110041363199233e-06,
|
||
|
|
"loss": 1.0468,
|
||
|
|
"step": 753
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.86,
|
||
|
|
"learning_rate": 9.083548973350019e-06,
|
||
|
|
"loss": 1.113,
|
||
|
|
"step": 754
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.86,
|
||
|
|
"learning_rate": 9.057063068485513e-06,
|
||
|
|
"loss": 1.0335,
|
||
|
|
"step": 755
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.86,
|
||
|
|
"learning_rate": 9.030583836025093e-06,
|
||
|
|
"loss": 1.1527,
|
||
|
|
"step": 756
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.86,
|
||
|
|
"learning_rate": 9.004111463340935e-06,
|
||
|
|
"loss": 1.1044,
|
||
|
|
"step": 757
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.87,
|
||
|
|
"learning_rate": 8.977646137756662e-06,
|
||
|
|
"loss": 1.0624,
|
||
|
|
"step": 758
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.87,
|
||
|
|
"learning_rate": 8.951188046546048e-06,
|
||
|
|
"loss": 1.057,
|
||
|
|
"step": 759
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.87,
|
||
|
|
"learning_rate": 8.924737376931651e-06,
|
||
|
|
"loss": 1.0724,
|
||
|
|
"step": 760
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.87,
|
||
|
|
"learning_rate": 8.898294316083529e-06,
|
||
|
|
"loss": 1.0964,
|
||
|
|
"step": 761
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.88,
|
||
|
|
"learning_rate": 8.871859051117896e-06,
|
||
|
|
"loss": 1.0759,
|
||
|
|
"step": 762
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.88,
|
||
|
|
"learning_rate": 8.8454317690958e-06,
|
||
|
|
"loss": 0.9316,
|
||
|
|
"step": 763
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.88,
|
||
|
|
"learning_rate": 8.819012657021794e-06,
|
||
|
|
"loss": 1.0518,
|
||
|
|
"step": 764
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.88,
|
||
|
|
"learning_rate": 8.79260190184263e-06,
|
||
|
|
"loss": 1.1317,
|
||
|
|
"step": 765
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.88,
|
||
|
|
"learning_rate": 8.76619969044592e-06,
|
||
|
|
"loss": 1.0858,
|
||
|
|
"step": 766
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.89,
|
||
|
|
"learning_rate": 8.739806209658812e-06,
|
||
|
|
"loss": 1.0621,
|
||
|
|
"step": 767
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.89,
|
||
|
|
"learning_rate": 8.713421646246692e-06,
|
||
|
|
"loss": 1.0204,
|
||
|
|
"step": 768
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.89,
|
||
|
|
"learning_rate": 8.687046186911819e-06,
|
||
|
|
"loss": 1.092,
|
||
|
|
"step": 769
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.89,
|
||
|
|
"learning_rate": 8.660680018292053e-06,
|
||
|
|
"loss": 1.0981,
|
||
|
|
"step": 770
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9,
|
||
|
|
"learning_rate": 8.634323326959501e-06,
|
||
|
|
"loss": 1.1185,
|
||
|
|
"step": 771
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9,
|
||
|
|
"learning_rate": 8.60797629941921e-06,
|
||
|
|
"loss": 1.0391,
|
||
|
|
"step": 772
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9,
|
||
|
|
"learning_rate": 8.581639122107837e-06,
|
||
|
|
"loss": 1.0094,
|
||
|
|
"step": 773
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.9,
|
||
|
|
"learning_rate": 8.555311981392342e-06,
|
||
|
|
"loss": 1.0877,
|
||
|
|
"step": 774
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.91,
|
||
|
|
"learning_rate": 8.528995063568673e-06,
|
||
|
|
"loss": 1.0701,
|
||
|
|
"step": 775
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.91,
|
||
|
|
"learning_rate": 8.502688554860426e-06,
|
||
|
|
"loss": 1.1157,
|
||
|
|
"step": 776
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.91,
|
||
|
|
"learning_rate": 8.47639264141754e-06,
|
||
|
|
"loss": 1.1232,
|
||
|
|
"step": 777
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.91,
|
||
|
|
"learning_rate": 8.450107509314983e-06,
|
||
|
|
"loss": 1.0751,
|
||
|
|
"step": 778
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.92,
|
||
|
|
"learning_rate": 8.423833344551443e-06,
|
||
|
|
"loss": 1.1229,
|
||
|
|
"step": 779
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.92,
|
||
|
|
"learning_rate": 8.423833344551443e-06,
|
||
|
|
"loss": 1.0607,
|
||
|
|
"step": 780
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.92,
|
||
|
|
"learning_rate": 8.423833344551443e-06,
|
||
|
|
"loss": 1.123,
|
||
|
|
"step": 781
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.92,
|
||
|
|
"learning_rate": 8.397570333047985e-06,
|
||
|
|
"loss": 1.1089,
|
||
|
|
"step": 782
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.93,
|
||
|
|
"learning_rate": 8.37131866064676e-06,
|
||
|
|
"loss": 1.114,
|
||
|
|
"step": 783
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.93,
|
||
|
|
"learning_rate": 8.345078513109677e-06,
|
||
|
|
"loss": 1.1008,
|
||
|
|
"step": 784
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.93,
|
||
|
|
"learning_rate": 8.31885007611709e-06,
|
||
|
|
"loss": 1.0697,
|
||
|
|
"step": 785
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.93,
|
||
|
|
"learning_rate": 8.2926335352665e-06,
|
||
|
|
"loss": 1.0889,
|
||
|
|
"step": 786
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.94,
|
||
|
|
"learning_rate": 8.2926335352665e-06,
|
||
|
|
"loss": 1.125,
|
||
|
|
"step": 787
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.94,
|
||
|
|
"learning_rate": 8.266429076071221e-06,
|
||
|
|
"loss": 1.0054,
|
||
|
|
"step": 788
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.94,
|
||
|
|
"learning_rate": 8.240236883959067e-06,
|
||
|
|
"loss": 1.0789,
|
||
|
|
"step": 789
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.94,
|
||
|
|
"learning_rate": 8.214057144271058e-06,
|
||
|
|
"loss": 1.0938,
|
||
|
|
"step": 790
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.95,
|
||
|
|
"learning_rate": 8.187890042260094e-06,
|
||
|
|
"loss": 1.1036,
|
||
|
|
"step": 791
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.95,
|
||
|
|
"learning_rate": 8.161735763089654e-06,
|
||
|
|
"loss": 1.1085,
|
||
|
|
"step": 792
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.95,
|
||
|
|
"learning_rate": 8.13559449183247e-06,
|
||
|
|
"loss": 1.1132,
|
||
|
|
"step": 793
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.95,
|
||
|
|
"learning_rate": 8.109466413469238e-06,
|
||
|
|
"loss": 1.0786,
|
||
|
|
"step": 794
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.96,
|
||
|
|
"learning_rate": 8.083351712887288e-06,
|
||
|
|
"loss": 1.0557,
|
||
|
|
"step": 795
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.96,
|
||
|
|
"learning_rate": 8.057250574879296e-06,
|
||
|
|
"loss": 1.0675,
|
||
|
|
"step": 796
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.96,
|
||
|
|
"learning_rate": 8.031163184141965e-06,
|
||
|
|
"loss": 1.1093,
|
||
|
|
"step": 797
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.96,
|
||
|
|
"learning_rate": 8.005089725274711e-06,
|
||
|
|
"loss": 1.1161,
|
||
|
|
"step": 798
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.97,
|
||
|
|
"learning_rate": 7.979030382778376e-06,
|
||
|
|
"loss": 1.06,
|
||
|
|
"step": 799
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.97,
|
||
|
|
"learning_rate": 7.952985341053902e-06,
|
||
|
|
"loss": 1.0706,
|
||
|
|
"step": 800
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.97,
|
||
|
|
"learning_rate": 7.92695478440105e-06,
|
||
|
|
"loss": 1.0718,
|
||
|
|
"step": 801
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.97,
|
||
|
|
"learning_rate": 7.900938897017064e-06,
|
||
|
|
"loss": 1.069,
|
||
|
|
"step": 802
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.98,
|
||
|
|
"learning_rate": 7.874937862995401e-06,
|
||
|
|
"loss": 1.1215,
|
||
|
|
"step": 803
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.98,
|
||
|
|
"learning_rate": 7.848951866324402e-06,
|
||
|
|
"loss": 1.1296,
|
||
|
|
"step": 804
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.98,
|
||
|
|
"learning_rate": 7.822981090886011e-06,
|
||
|
|
"loss": 1.075,
|
||
|
|
"step": 805
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.98,
|
||
|
|
"learning_rate": 7.822981090886011e-06,
|
||
|
|
"loss": 1.0761,
|
||
|
|
"step": 806
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.99,
|
||
|
|
"learning_rate": 7.79702572045445e-06,
|
||
|
|
"loss": 1.0116,
|
||
|
|
"step": 807
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.99,
|
||
|
|
"learning_rate": 7.771085938694943e-06,
|
||
|
|
"loss": 1.0345,
|
||
|
|
"step": 808
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.99,
|
||
|
|
"learning_rate": 7.745161929162405e-06,
|
||
|
|
"loss": 1.026,
|
||
|
|
"step": 809
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.99,
|
||
|
|
"learning_rate": 7.719253875300138e-06,
|
||
|
|
"loss": 1.1431,
|
||
|
|
"step": 810
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0,
|
||
|
|
"learning_rate": 7.693361960438548e-06,
|
||
|
|
"loss": 1.1076,
|
||
|
|
"step": 811
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0,
|
||
|
|
"learning_rate": 7.667486367793822e-06,
|
||
|
|
"loss": 1.1333,
|
||
|
|
"step": 812
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0,
|
||
|
|
"learning_rate": 7.641627280466663e-06,
|
||
|
|
"loss": 1.064,
|
||
|
|
"step": 813
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.0,
|
||
|
|
"learning_rate": 7.615784881440975e-06,
|
||
|
|
"loss": 0.9942,
|
||
|
|
"step": 814
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.01,
|
||
|
|
"learning_rate": 7.589959353582574e-06,
|
||
|
|
"loss": 1.0166,
|
||
|
|
"step": 815
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.01,
|
||
|
|
"learning_rate": 7.564150879637882e-06,
|
||
|
|
"loss": 1.061,
|
||
|
|
"step": 816
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.01,
|
||
|
|
"learning_rate": 7.538359642232654e-06,
|
||
|
|
"loss": 1.0256,
|
||
|
|
"step": 817
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.01,
|
||
|
|
"learning_rate": 7.5125858238706785e-06,
|
||
|
|
"loss": 1.0732,
|
||
|
|
"step": 818
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.02,
|
||
|
|
"learning_rate": 7.486829606932478e-06,
|
||
|
|
"loss": 0.9878,
|
||
|
|
"step": 819
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.02,
|
||
|
|
"learning_rate": 7.486829606932478e-06,
|
||
|
|
"loss": 1.0634,
|
||
|
|
"step": 820
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.02,
|
||
|
|
"learning_rate": 7.461091173674022e-06,
|
||
|
|
"loss": 1.086,
|
||
|
|
"step": 821
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.02,
|
||
|
|
"learning_rate": 7.43537070622544e-06,
|
||
|
|
"loss": 1.1077,
|
||
|
|
"step": 822
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.03,
|
||
|
|
"learning_rate": 7.40966838658974e-06,
|
||
|
|
"loss": 1.0418,
|
||
|
|
"step": 823
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.03,
|
||
|
|
"learning_rate": 7.383984396641506e-06,
|
||
|
|
"loss": 1.0398,
|
||
|
|
"step": 824
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.03,
|
||
|
|
"learning_rate": 7.383984396641506e-06,
|
||
|
|
"loss": 1.1066,
|
||
|
|
"step": 825
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.03,
|
||
|
|
"learning_rate": 7.358318918125613e-06,
|
||
|
|
"loss": 1.0911,
|
||
|
|
"step": 826
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.04,
|
||
|
|
"learning_rate": 7.332672132655953e-06,
|
||
|
|
"loss": 1.055,
|
||
|
|
"step": 827
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.04,
|
||
|
|
"learning_rate": 7.307044221714139e-06,
|
||
|
|
"loss": 1.0988,
|
||
|
|
"step": 828
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.04,
|
||
|
|
"learning_rate": 7.307044221714139e-06,
|
||
|
|
"loss": 1.03,
|
||
|
|
"step": 829
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.04,
|
||
|
|
"learning_rate": 7.2814353666482276e-06,
|
||
|
|
"loss": 1.0436,
|
||
|
|
"step": 830
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.04,
|
||
|
|
"learning_rate": 7.2814353666482276e-06,
|
||
|
|
"loss": 1.099,
|
||
|
|
"step": 831
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.05,
|
||
|
|
"learning_rate": 7.2558457486714316e-06,
|
||
|
|
"loss": 1.0698,
|
||
|
|
"step": 832
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.05,
|
||
|
|
"learning_rate": 7.230275548860833e-06,
|
||
|
|
"loss": 1.0925,
|
||
|
|
"step": 833
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.05,
|
||
|
|
"learning_rate": 7.2047249481561125e-06,
|
||
|
|
"loss": 0.9996,
|
||
|
|
"step": 834
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.05,
|
||
|
|
"learning_rate": 7.179194127358258e-06,
|
||
|
|
"loss": 0.9369,
|
||
|
|
"step": 835
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.06,
|
||
|
|
"learning_rate": 7.153683267128304e-06,
|
||
|
|
"loss": 1.0355,
|
||
|
|
"step": 836
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.06,
|
||
|
|
"learning_rate": 7.153683267128304e-06,
|
||
|
|
"loss": 1.0375,
|
||
|
|
"step": 837
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.06,
|
||
|
|
"learning_rate": 7.128192547986023e-06,
|
||
|
|
"loss": 1.0004,
|
||
|
|
"step": 838
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.06,
|
||
|
|
"learning_rate": 7.128192547986023e-06,
|
||
|
|
"loss": 1.041,
|
||
|
|
"step": 839
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.07,
|
||
|
|
"learning_rate": 7.102722150308678e-06,
|
||
|
|
"loss": 1.0829,
|
||
|
|
"step": 840
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.07,
|
||
|
|
"learning_rate": 7.077272254329726e-06,
|
||
|
|
"loss": 1.0619,
|
||
|
|
"step": 841
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.07,
|
||
|
|
"learning_rate": 7.051843040137558e-06,
|
||
|
|
"loss": 1.0293,
|
||
|
|
"step": 842
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.07,
|
||
|
|
"learning_rate": 7.026434687674204e-06,
|
||
|
|
"loss": 1.0453,
|
||
|
|
"step": 843
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.08,
|
||
|
|
"learning_rate": 7.001047376734087e-06,
|
||
|
|
"loss": 1.0631,
|
||
|
|
"step": 844
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.08,
|
||
|
|
"learning_rate": 6.975681286962724e-06,
|
||
|
|
"loss": 1.1154,
|
||
|
|
"step": 845
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.08,
|
||
|
|
"learning_rate": 6.9503365978554735e-06,
|
||
|
|
"loss": 0.9798,
|
||
|
|
"step": 846
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.08,
|
||
|
|
"learning_rate": 6.925013488756264e-06,
|
||
|
|
"loss": 0.9492,
|
||
|
|
"step": 847
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.09,
|
||
|
|
"learning_rate": 6.8997121388563e-06,
|
||
|
|
"loss": 0.9858,
|
||
|
|
"step": 848
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.09,
|
||
|
|
"learning_rate": 6.874432727192837e-06,
|
||
|
|
"loss": 1.0972,
|
||
|
|
"step": 849
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.09,
|
||
|
|
"learning_rate": 6.849175432647875e-06,
|
||
|
|
"loss": 1.0957,
|
||
|
|
"step": 850
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.09,
|
||
|
|
"learning_rate": 6.823940433946921e-06,
|
||
|
|
"loss": 0.964,
|
||
|
|
"step": 851
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1,
|
||
|
|
"learning_rate": 6.798727909657698e-06,
|
||
|
|
"loss": 1.0073,
|
||
|
|
"step": 852
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1,
|
||
|
|
"learning_rate": 6.773538038188912e-06,
|
||
|
|
"loss": 1.0884,
|
||
|
|
"step": 853
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1,
|
||
|
|
"learning_rate": 6.773538038188912e-06,
|
||
|
|
"loss": 1.0126,
|
||
|
|
"step": 854
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.1,
|
||
|
|
"learning_rate": 6.74837099778896e-06,
|
||
|
|
"loss": 1.1461,
|
||
|
|
"step": 855
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.11,
|
||
|
|
"learning_rate": 6.723226966544691e-06,
|
||
|
|
"loss": 1.0353,
|
||
|
|
"step": 856
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.11,
|
||
|
|
"learning_rate": 6.69810612238013e-06,
|
||
|
|
"loss": 1.0648,
|
||
|
|
"step": 857
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.11,
|
||
|
|
"learning_rate": 6.673008643055228e-06,
|
||
|
|
"loss": 0.9614,
|
||
|
|
"step": 858
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.11,
|
||
|
|
"learning_rate": 6.6479347061646046e-06,
|
||
|
|
"loss": 0.9475,
|
||
|
|
"step": 859
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.12,
|
||
|
|
"learning_rate": 6.622884489136286e-06,
|
||
|
|
"loss": 0.9858,
|
||
|
|
"step": 860
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.12,
|
||
|
|
"learning_rate": 6.622884489136286e-06,
|
||
|
|
"loss": 0.9516,
|
||
|
|
"step": 861
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.12,
|
||
|
|
"learning_rate": 6.597858169230454e-06,
|
||
|
|
"loss": 1.0605,
|
||
|
|
"step": 862
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.12,
|
||
|
|
"learning_rate": 6.597858169230454e-06,
|
||
|
|
"loss": 1.0432,
|
||
|
|
"step": 863
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.13,
|
||
|
|
"learning_rate": 6.572855923538186e-06,
|
||
|
|
"loss": 1.083,
|
||
|
|
"step": 864
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.13,
|
||
|
|
"learning_rate": 6.547877928980206e-06,
|
||
|
|
"loss": 1.1029,
|
||
|
|
"step": 865
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.13,
|
||
|
|
"learning_rate": 6.522924362305639e-06,
|
||
|
|
"loss": 1.0126,
|
||
|
|
"step": 866
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.13,
|
||
|
|
"learning_rate": 6.497995400090748e-06,
|
||
|
|
"loss": 1.0487,
|
||
|
|
"step": 867
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.14,
|
||
|
|
"learning_rate": 6.4730912187376895e-06,
|
||
|
|
"loss": 1.0179,
|
||
|
|
"step": 868
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.14,
|
||
|
|
"learning_rate": 6.448211994473263e-06,
|
||
|
|
"loss": 1.1008,
|
||
|
|
"step": 869
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.14,
|
||
|
|
"learning_rate": 6.448211994473263e-06,
|
||
|
|
"loss": 0.9999,
|
||
|
|
"step": 870
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.14,
|
||
|
|
"learning_rate": 6.42335790334768e-06,
|
||
|
|
"loss": 1.0426,
|
||
|
|
"step": 871
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.15,
|
||
|
|
"learning_rate": 6.42335790334768e-06,
|
||
|
|
"loss": 1.0961,
|
||
|
|
"step": 872
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.15,
|
||
|
|
"learning_rate": 6.398529121233291e-06,
|
||
|
|
"loss": 1.0348,
|
||
|
|
"step": 873
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.15,
|
||
|
|
"learning_rate": 6.373725823823359e-06,
|
||
|
|
"loss": 1.1038,
|
||
|
|
"step": 874
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.15,
|
||
|
|
"learning_rate": 6.348948186630815e-06,
|
||
|
|
"loss": 1.0372,
|
||
|
|
"step": 875
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.16,
|
||
|
|
"learning_rate": 6.324196384987009e-06,
|
||
|
|
"loss": 1.0386,
|
||
|
|
"step": 876
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.16,
|
||
|
|
"learning_rate": 6.2994705940404825e-06,
|
||
|
|
"loss": 1.0796,
|
||
|
|
"step": 877
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.16,
|
||
|
|
"learning_rate": 6.274770988755712e-06,
|
||
|
|
"loss": 1.0432,
|
||
|
|
"step": 878
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.16,
|
||
|
|
"learning_rate": 6.250097743911877e-06,
|
||
|
|
"loss": 1.0826,
|
||
|
|
"step": 879
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.17,
|
||
|
|
"learning_rate": 6.225451034101631e-06,
|
||
|
|
"loss": 1.0985,
|
||
|
|
"step": 880
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.17,
|
||
|
|
"learning_rate": 6.200831033729864e-06,
|
||
|
|
"loss": 1.0645,
|
||
|
|
"step": 881
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.17,
|
||
|
|
"learning_rate": 6.176237917012459e-06,
|
||
|
|
"loss": 1.0353,
|
||
|
|
"step": 882
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.17,
|
||
|
|
"learning_rate": 6.151671857975061e-06,
|
||
|
|
"loss": 1.0356,
|
||
|
|
"step": 883
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.18,
|
||
|
|
"learning_rate": 6.12713303045186e-06,
|
||
|
|
"loss": 1.0918,
|
||
|
|
"step": 884
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.18,
|
||
|
|
"learning_rate": 6.10262160808434e-06,
|
||
|
|
"loss": 1.1154,
|
||
|
|
"step": 885
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.18,
|
||
|
|
"learning_rate": 6.0781377643200765e-06,
|
||
|
|
"loss": 1.0837,
|
||
|
|
"step": 886
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.18,
|
||
|
|
"learning_rate": 6.053681672411471e-06,
|
||
|
|
"loss": 0.927,
|
||
|
|
"step": 887
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.19,
|
||
|
|
"learning_rate": 6.029253505414565e-06,
|
||
|
|
"loss": 0.9846,
|
||
|
|
"step": 888
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.19,
|
||
|
|
"learning_rate": 6.004853436187794e-06,
|
||
|
|
"loss": 0.9542,
|
||
|
|
"step": 889
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.19,
|
||
|
|
"learning_rate": 5.9804816373907625e-06,
|
||
|
|
"loss": 1.0098,
|
||
|
|
"step": 890
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.19,
|
||
|
|
"learning_rate": 5.956138281483039e-06,
|
||
|
|
"loss": 1.0494,
|
||
|
|
"step": 891
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2,
|
||
|
|
"learning_rate": 5.931823540722912e-06,
|
||
|
|
"loss": 1.0146,
|
||
|
|
"step": 892
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2,
|
||
|
|
"learning_rate": 5.907537587166191e-06,
|
||
|
|
"loss": 1.0438,
|
||
|
|
"step": 893
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2,
|
||
|
|
"learning_rate": 5.883280592664979e-06,
|
||
|
|
"loss": 0.999,
|
||
|
|
"step": 894
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2,
|
||
|
|
"learning_rate": 5.859052728866468e-06,
|
||
|
|
"loss": 1.0596,
|
||
|
|
"step": 895
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.2,
|
||
|
|
"learning_rate": 5.859052728866468e-06,
|
||
|
|
"loss": 0.9974,
|
||
|
|
"step": 896
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.21,
|
||
|
|
"learning_rate": 5.859052728866468e-06,
|
||
|
|
"loss": 1.0249,
|
||
|
|
"step": 897
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.21,
|
||
|
|
"learning_rate": 5.859052728866468e-06,
|
||
|
|
"loss": 1.0427,
|
||
|
|
"step": 898
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.21,
|
||
|
|
"learning_rate": 5.859052728866468e-06,
|
||
|
|
"loss": 1.0657,
|
||
|
|
"step": 899
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.21,
|
||
|
|
"learning_rate": 5.834854167211699e-06,
|
||
|
|
"loss": 1.017,
|
||
|
|
"step": 900
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.22,
|
||
|
|
"learning_rate": 5.810685078934375e-06,
|
||
|
|
"loss": 1.0231,
|
||
|
|
"step": 901
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.22,
|
||
|
|
"learning_rate": 5.810685078934375e-06,
|
||
|
|
"loss": 1.1882,
|
||
|
|
"step": 902
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.22,
|
||
|
|
"learning_rate": 5.78654563505964e-06,
|
||
|
|
"loss": 0.9763,
|
||
|
|
"step": 903
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.22,
|
||
|
|
"learning_rate": 5.762436006402874e-06,
|
||
|
|
"loss": 1.0037,
|
||
|
|
"step": 904
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.23,
|
||
|
|
"learning_rate": 5.738356363568463e-06,
|
||
|
|
"loss": 1.0616,
|
||
|
|
"step": 905
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.23,
|
||
|
|
"learning_rate": 5.714306876948621e-06,
|
||
|
|
"loss": 1.0914,
|
||
|
|
"step": 906
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.23,
|
||
|
|
"learning_rate": 5.69028771672216e-06,
|
||
|
|
"loss": 0.9882,
|
||
|
|
"step": 907
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.23,
|
||
|
|
"learning_rate": 5.666299052853314e-06,
|
||
|
|
"loss": 1.0209,
|
||
|
|
"step": 908
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.24,
|
||
|
|
"learning_rate": 5.642341055090508e-06,
|
||
|
|
"loss": 1.0136,
|
||
|
|
"step": 909
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.24,
|
||
|
|
"learning_rate": 5.618413892965158e-06,
|
||
|
|
"loss": 0.9574,
|
||
|
|
"step": 910
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.24,
|
||
|
|
"learning_rate": 5.5945177357904935e-06,
|
||
|
|
"loss": 1.0981,
|
||
|
|
"step": 911
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.24,
|
||
|
|
"learning_rate": 5.570652752660343e-06,
|
||
|
|
"loss": 1.0165,
|
||
|
|
"step": 912
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.25,
|
||
|
|
"learning_rate": 5.546819112447952e-06,
|
||
|
|
"loss": 1.02,
|
||
|
|
"step": 913
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.25,
|
||
|
|
"learning_rate": 5.523016983804759e-06,
|
||
|
|
"loss": 0.9675,
|
||
|
|
"step": 914
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.25,
|
||
|
|
"learning_rate": 5.499246535159231e-06,
|
||
|
|
"loss": 0.9956,
|
||
|
|
"step": 915
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.25,
|
||
|
|
"learning_rate": 5.47550793471566e-06,
|
||
|
|
"loss": 0.9971,
|
||
|
|
"step": 916
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.26,
|
||
|
|
"learning_rate": 5.451801350452975e-06,
|
||
|
|
"loss": 1.0175,
|
||
|
|
"step": 917
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.26,
|
||
|
|
"learning_rate": 5.428126950123551e-06,
|
||
|
|
"loss": 0.9816,
|
||
|
|
"step": 918
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.26,
|
||
|
|
"learning_rate": 5.404484901252023e-06,
|
||
|
|
"loss": 1.0217,
|
||
|
|
"step": 919
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.26,
|
||
|
|
"learning_rate": 5.3808753711341e-06,
|
||
|
|
"loss": 0.9744,
|
||
|
|
"step": 920
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.27,
|
||
|
|
"learning_rate": 5.357298526835381e-06,
|
||
|
|
"loss": 1.0531,
|
||
|
|
"step": 921
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.27,
|
||
|
|
"learning_rate": 5.333754535190186e-06,
|
||
|
|
"loss": 1.0347,
|
||
|
|
"step": 922
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.27,
|
||
|
|
"learning_rate": 5.3102435628003435e-06,
|
||
|
|
"loss": 0.9996,
|
||
|
|
"step": 923
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.27,
|
||
|
|
"learning_rate": 5.286765776034044e-06,
|
||
|
|
"loss": 0.9865,
|
||
|
|
"step": 924
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.28,
|
||
|
|
"learning_rate": 5.263321341024646e-06,
|
||
|
|
"loss": 1.0154,
|
||
|
|
"step": 925
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.28,
|
||
|
|
"learning_rate": 5.239910423669509e-06,
|
||
|
|
"loss": 1.0845,
|
||
|
|
"step": 926
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.28,
|
||
|
|
"learning_rate": 5.216533189628808e-06,
|
||
|
|
"loss": 1.0039,
|
||
|
|
"step": 927
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.28,
|
||
|
|
"learning_rate": 5.193189804324376e-06,
|
||
|
|
"loss": 1.0424,
|
||
|
|
"step": 928
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.29,
|
||
|
|
"learning_rate": 5.169880432938519e-06,
|
||
|
|
"loss": 1.0209,
|
||
|
|
"step": 929
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.29,
|
||
|
|
"learning_rate": 5.146605240412859e-06,
|
||
|
|
"loss": 0.9603,
|
||
|
|
"step": 930
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.29,
|
||
|
|
"learning_rate": 5.123364391447156e-06,
|
||
|
|
"loss": 0.9627,
|
||
|
|
"step": 931
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.29,
|
||
|
|
"learning_rate": 5.100158050498159e-06,
|
||
|
|
"loss": 1.1093,
|
||
|
|
"step": 932
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3,
|
||
|
|
"learning_rate": 5.076986381778417e-06,
|
||
|
|
"loss": 1.0122,
|
||
|
|
"step": 933
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3,
|
||
|
|
"learning_rate": 5.053849549255143e-06,
|
||
|
|
"loss": 1.0515,
|
||
|
|
"step": 934
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3,
|
||
|
|
"learning_rate": 5.053849549255143e-06,
|
||
|
|
"loss": 1.0038,
|
||
|
|
"step": 935
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.3,
|
||
|
|
"learning_rate": 5.03074771664904e-06,
|
||
|
|
"loss": 1.1328,
|
||
|
|
"step": 936
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.31,
|
||
|
|
"learning_rate": 5.03074771664904e-06,
|
||
|
|
"loss": 1.0808,
|
||
|
|
"step": 937
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.31,
|
||
|
|
"learning_rate": 5.0076810474331395e-06,
|
||
|
|
"loss": 1.0216,
|
||
|
|
"step": 938
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.31,
|
||
|
|
"learning_rate": 4.984649704831658e-06,
|
||
|
|
"loss": 0.9545,
|
||
|
|
"step": 939
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.31,
|
||
|
|
"learning_rate": 4.961653851818827e-06,
|
||
|
|
"loss": 1.0393,
|
||
|
|
"step": 940
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.32,
|
||
|
|
"learning_rate": 4.938693651117751e-06,
|
||
|
|
"loss": 1.0588,
|
||
|
|
"step": 941
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.32,
|
||
|
|
"learning_rate": 4.9157692651992495e-06,
|
||
|
|
"loss": 1.0493,
|
||
|
|
"step": 942
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.32,
|
||
|
|
"learning_rate": 4.892880856280713e-06,
|
||
|
|
"loss": 0.9733,
|
||
|
|
"step": 943
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.32,
|
||
|
|
"learning_rate": 4.892880856280713e-06,
|
||
|
|
"loss": 0.9834,
|
||
|
|
"step": 944
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.33,
|
||
|
|
"learning_rate": 4.870028586324947e-06,
|
||
|
|
"loss": 1.007,
|
||
|
|
"step": 945
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.33,
|
||
|
|
"learning_rate": 4.847212617039037e-06,
|
||
|
|
"loss": 1.0005,
|
||
|
|
"step": 946
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.33,
|
||
|
|
"learning_rate": 4.82443310987319e-06,
|
||
|
|
"loss": 1.0557,
|
||
|
|
"step": 947
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.33,
|
||
|
|
"learning_rate": 4.801690226019606e-06,
|
||
|
|
"loss": 1.0515,
|
||
|
|
"step": 948
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.34,
|
||
|
|
"learning_rate": 4.77898412641133e-06,
|
||
|
|
"loss": 0.9365,
|
||
|
|
"step": 949
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.34,
|
||
|
|
"learning_rate": 4.756314971721115e-06,
|
||
|
|
"loss": 0.999,
|
||
|
|
"step": 950
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.34,
|
||
|
|
"learning_rate": 4.733682922360282e-06,
|
||
|
|
"loss": 1.0896,
|
||
|
|
"step": 951
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.34,
|
||
|
|
"learning_rate": 4.71108813847759e-06,
|
||
|
|
"loss": 1.0972,
|
||
|
|
"step": 952
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.35,
|
||
|
|
"learning_rate": 4.688530779958099e-06,
|
||
|
|
"loss": 1.0062,
|
||
|
|
"step": 953
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.35,
|
||
|
|
"learning_rate": 4.666011006422041e-06,
|
||
|
|
"loss": 0.9758,
|
||
|
|
"step": 954
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.35,
|
||
|
|
"learning_rate": 4.643528977223689e-06,
|
||
|
|
"loss": 1.0303,
|
||
|
|
"step": 955
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.35,
|
||
|
|
"learning_rate": 4.621084851450229e-06,
|
||
|
|
"loss": 1.145,
|
||
|
|
"step": 956
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.35,
|
||
|
|
"learning_rate": 4.5986787879206375e-06,
|
||
|
|
"loss": 1.0513,
|
||
|
|
"step": 957
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.36,
|
||
|
|
"learning_rate": 4.5763109451845515e-06,
|
||
|
|
"loss": 1.0194,
|
||
|
|
"step": 958
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.36,
|
||
|
|
"learning_rate": 4.553981481521156e-06,
|
||
|
|
"loss": 1.0725,
|
||
|
|
"step": 959
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.36,
|
||
|
|
"learning_rate": 4.531690554938043e-06,
|
||
|
|
"loss": 1.0504,
|
||
|
|
"step": 960
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.36,
|
||
|
|
"learning_rate": 4.531690554938043e-06,
|
||
|
|
"loss": 1.0375,
|
||
|
|
"step": 961
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.37,
|
||
|
|
"learning_rate": 4.509438323170131e-06,
|
||
|
|
"loss": 1.1302,
|
||
|
|
"step": 962
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.37,
|
||
|
|
"learning_rate": 4.487224943678513e-06,
|
||
|
|
"loss": 1.0631,
|
||
|
|
"step": 963
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.37,
|
||
|
|
"learning_rate": 4.465050573649359e-06,
|
||
|
|
"loss": 0.9878,
|
||
|
|
"step": 964
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.37,
|
||
|
|
"learning_rate": 4.465050573649359e-06,
|
||
|
|
"loss": 0.9993,
|
||
|
|
"step": 965
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.38,
|
||
|
|
"learning_rate": 4.442915369992802e-06,
|
||
|
|
"loss": 0.9874,
|
||
|
|
"step": 966
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.38,
|
||
|
|
"learning_rate": 4.442915369992802e-06,
|
||
|
|
"loss": 1.0806,
|
||
|
|
"step": 967
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.38,
|
||
|
|
"learning_rate": 4.4208194893418125e-06,
|
||
|
|
"loss": 1.0668,
|
||
|
|
"step": 968
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.38,
|
||
|
|
"learning_rate": 4.398763088051127e-06,
|
||
|
|
"loss": 1.1283,
|
||
|
|
"step": 969
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.39,
|
||
|
|
"learning_rate": 4.3767463221961034e-06,
|
||
|
|
"loss": 1.0712,
|
||
|
|
"step": 970
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.39,
|
||
|
|
"learning_rate": 4.354769347571638e-06,
|
||
|
|
"loss": 1.0416,
|
||
|
|
"step": 971
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.39,
|
||
|
|
"learning_rate": 4.332832319691044e-06,
|
||
|
|
"loss": 1.0996,
|
||
|
|
"step": 972
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.39,
|
||
|
|
"learning_rate": 4.3109353937849815e-06,
|
||
|
|
"loss": 1.0031,
|
||
|
|
"step": 973
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4,
|
||
|
|
"learning_rate": 4.289078724800331e-06,
|
||
|
|
"loss": 1.0374,
|
||
|
|
"step": 974
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4,
|
||
|
|
"learning_rate": 4.267262467399114e-06,
|
||
|
|
"loss": 1.0049,
|
||
|
|
"step": 975
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4,
|
||
|
|
"learning_rate": 4.267262467399114e-06,
|
||
|
|
"loss": 1.0352,
|
||
|
|
"step": 976
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.4,
|
||
|
|
"learning_rate": 4.24548677595738e-06,
|
||
|
|
"loss": 1.0891,
|
||
|
|
"step": 977
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.41,
|
||
|
|
"learning_rate": 4.22375180456413e-06,
|
||
|
|
"loss": 1.0007,
|
||
|
|
"step": 978
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.41,
|
||
|
|
"learning_rate": 4.202057707020235e-06,
|
||
|
|
"loss": 1.0688,
|
||
|
|
"step": 979
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.41,
|
||
|
|
"learning_rate": 4.180404636837321e-06,
|
||
|
|
"loss": 1.0779,
|
||
|
|
"step": 980
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.41,
|
||
|
|
"learning_rate": 4.158792747236702e-06,
|
||
|
|
"loss": 1.0325,
|
||
|
|
"step": 981
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.42,
|
||
|
|
"learning_rate": 4.137222191148282e-06,
|
||
|
|
"loss": 0.9851,
|
||
|
|
"step": 982
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.42,
|
||
|
|
"learning_rate": 4.11569312120949e-06,
|
||
|
|
"loss": 1.0092,
|
||
|
|
"step": 983
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.42,
|
||
|
|
"learning_rate": 4.0942056897641934e-06,
|
||
|
|
"loss": 1.0669,
|
||
|
|
"step": 984
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.42,
|
||
|
|
"learning_rate": 4.072760048861614e-06,
|
||
|
|
"loss": 1.0599,
|
||
|
|
"step": 985
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.43,
|
||
|
|
"learning_rate": 4.051356350255246e-06,
|
||
|
|
"loss": 1.092,
|
||
|
|
"step": 986
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.43,
|
||
|
|
"learning_rate": 4.0299947454018e-06,
|
||
|
|
"loss": 1.1094,
|
||
|
|
"step": 987
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.43,
|
||
|
|
"learning_rate": 4.008675385460131e-06,
|
||
|
|
"loss": 1.0148,
|
||
|
|
"step": 988
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.43,
|
||
|
|
"learning_rate": 3.987398421290155e-06,
|
||
|
|
"loss": 1.0314,
|
||
|
|
"step": 989
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.44,
|
||
|
|
"learning_rate": 3.966164003451775e-06,
|
||
|
|
"loss": 0.9866,
|
||
|
|
"step": 990
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.44,
|
||
|
|
"learning_rate": 3.944972282203844e-06,
|
||
|
|
"loss": 0.9976,
|
||
|
|
"step": 991
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.44,
|
||
|
|
"learning_rate": 3.944972282203844e-06,
|
||
|
|
"loss": 0.964,
|
||
|
|
"step": 992
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.44,
|
||
|
|
"learning_rate": 3.923823407503076e-06,
|
||
|
|
"loss": 1.0865,
|
||
|
|
"step": 993
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.45,
|
||
|
|
"learning_rate": 3.902717529003005e-06,
|
||
|
|
"loss": 1.0493,
|
||
|
|
"step": 994
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.45,
|
||
|
|
"learning_rate": 3.902717529003005e-06,
|
||
|
|
"loss": 0.9973,
|
||
|
|
"step": 995
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.45,
|
||
|
|
"learning_rate": 3.88165479605291e-06,
|
||
|
|
"loss": 1.0446,
|
||
|
|
"step": 996
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.45,
|
||
|
|
"learning_rate": 3.860635357696756e-06,
|
||
|
|
"loss": 1.0599,
|
||
|
|
"step": 997
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.46,
|
||
|
|
"learning_rate": 3.839659362672156e-06,
|
||
|
|
"loss": 1.0406,
|
||
|
|
"step": 998
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.46,
|
||
|
|
"learning_rate": 3.818726959409305e-06,
|
||
|
|
"loss": 1.0254,
|
||
|
|
"step": 999
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.46,
|
||
|
|
"learning_rate": 3.7978382960299476e-06,
|
||
|
|
"loss": 1.0693,
|
||
|
|
"step": 1000
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.46,
|
||
|
|
"learning_rate": 3.776993520346295e-06,
|
||
|
|
"loss": 1.0674,
|
||
|
|
"step": 1001
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.47,
|
||
|
|
"learning_rate": 3.756192779860014e-06,
|
||
|
|
"loss": 1.0298,
|
||
|
|
"step": 1002
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.47,
|
||
|
|
"learning_rate": 3.756192779860014e-06,
|
||
|
|
"loss": 0.9912,
|
||
|
|
"step": 1003
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.47,
|
||
|
|
"learning_rate": 3.7354362217611652e-06,
|
||
|
|
"loss": 1.1106,
|
||
|
|
"step": 1004
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.47,
|
||
|
|
"learning_rate": 3.714723992927177e-06,
|
||
|
|
"loss": 1.0397,
|
||
|
|
"step": 1005
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.48,
|
||
|
|
"learning_rate": 3.694056239921776e-06,
|
||
|
|
"loss": 1.1313,
|
||
|
|
"step": 1006
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.48,
|
||
|
|
"learning_rate": 3.6734331089939835e-06,
|
||
|
|
"loss": 1.1178,
|
||
|
|
"step": 1007
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.48,
|
||
|
|
"learning_rate": 3.6528547460770636e-06,
|
||
|
|
"loss": 1.028,
|
||
|
|
"step": 1008
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.48,
|
||
|
|
"learning_rate": 3.6323212967874866e-06,
|
||
|
|
"loss": 0.9894,
|
||
|
|
"step": 1009
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.49,
|
||
|
|
"learning_rate": 3.6118329064239222e-06,
|
||
|
|
"loss": 0.9559,
|
||
|
|
"step": 1010
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.49,
|
||
|
|
"learning_rate": 3.5913897199661716e-06,
|
||
|
|
"loss": 1.0402,
|
||
|
|
"step": 1011
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.49,
|
||
|
|
"learning_rate": 3.5709918820741816e-06,
|
||
|
|
"loss": 1.0309,
|
||
|
|
"step": 1012
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.49,
|
||
|
|
"learning_rate": 3.5709918820741816e-06,
|
||
|
|
"loss": 0.9957,
|
||
|
|
"step": 1013
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5,
|
||
|
|
"learning_rate": 3.5506395370869963e-06,
|
||
|
|
"loss": 0.9978,
|
||
|
|
"step": 1014
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5,
|
||
|
|
"learning_rate": 3.5303328290217453e-06,
|
||
|
|
"loss": 1.0179,
|
||
|
|
"step": 1015
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5,
|
||
|
|
"learning_rate": 3.5100719015726228e-06,
|
||
|
|
"loss": 1.0326,
|
||
|
|
"step": 1016
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.5,
|
||
|
|
"learning_rate": 3.5100719015726228e-06,
|
||
|
|
"loss": 1.0268,
|
||
|
|
"step": 1017
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.51,
|
||
|
|
"learning_rate": 3.4898568981098678e-06,
|
||
|
|
"loss": 1.0515,
|
||
|
|
"step": 1018
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.51,
|
||
|
|
"learning_rate": 3.469687961678757e-06,
|
||
|
|
"loss": 1.0936,
|
||
|
|
"step": 1019
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.51,
|
||
|
|
"learning_rate": 3.4495652349985844e-06,
|
||
|
|
"loss": 1.0403,
|
||
|
|
"step": 1020
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.51,
|
||
|
|
"learning_rate": 3.429488860461655e-06,
|
||
|
|
"loss": 1.0342,
|
||
|
|
"step": 1021
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.51,
|
||
|
|
"learning_rate": 3.4094589801322773e-06,
|
||
|
|
"loss": 1.0026,
|
||
|
|
"step": 1022
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.52,
|
||
|
|
"learning_rate": 3.389475735745761e-06,
|
||
|
|
"loss": 1.0009,
|
||
|
|
"step": 1023
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.52,
|
||
|
|
"learning_rate": 3.3695392687074045e-06,
|
||
|
|
"loss": 0.9617,
|
||
|
|
"step": 1024
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.52,
|
||
|
|
"learning_rate": 3.3496497200915067e-06,
|
||
|
|
"loss": 1.0,
|
||
|
|
"step": 1025
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.52,
|
||
|
|
"learning_rate": 3.3298072306403595e-06,
|
||
|
|
"loss": 1.0145,
|
||
|
|
"step": 1026
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.53,
|
||
|
|
"learning_rate": 3.3100119407632556e-06,
|
||
|
|
"loss": 0.9589,
|
||
|
|
"step": 1027
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.53,
|
||
|
|
"learning_rate": 3.2902639905354948e-06,
|
||
|
|
"loss": 1.0562,
|
||
|
|
"step": 1028
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.53,
|
||
|
|
"learning_rate": 3.2705635196973927e-06,
|
||
|
|
"loss": 1.0662,
|
||
|
|
"step": 1029
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.53,
|
||
|
|
"learning_rate": 3.2509106676532897e-06,
|
||
|
|
"loss": 1.0597,
|
||
|
|
"step": 1030
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.54,
|
||
|
|
"learning_rate": 3.231305573470569e-06,
|
||
|
|
"loss": 1.0901,
|
||
|
|
"step": 1031
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.54,
|
||
|
|
"learning_rate": 3.231305573470569e-06,
|
||
|
|
"loss": 0.9323,
|
||
|
|
"step": 1032
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.54,
|
||
|
|
"learning_rate": 3.2117483758786683e-06,
|
||
|
|
"loss": 1.0963,
|
||
|
|
"step": 1033
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.54,
|
||
|
|
"learning_rate": 3.2117483758786683e-06,
|
||
|
|
"loss": 1.0256,
|
||
|
|
"step": 1034
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.55,
|
||
|
|
"learning_rate": 3.192239213268099e-06,
|
||
|
|
"loss": 0.9955,
|
||
|
|
"step": 1035
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.55,
|
||
|
|
"learning_rate": 3.17277822368947e-06,
|
||
|
|
"loss": 1.0681,
|
||
|
|
"step": 1036
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.55,
|
||
|
|
"learning_rate": 3.1533655448525057e-06,
|
||
|
|
"loss": 1.03,
|
||
|
|
"step": 1037
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.55,
|
||
|
|
"learning_rate": 3.134001314125079e-06,
|
||
|
|
"loss": 0.9993,
|
||
|
|
"step": 1038
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.56,
|
||
|
|
"learning_rate": 3.114685668532229e-06,
|
||
|
|
"loss": 1.0574,
|
||
|
|
"step": 1039
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.56,
|
||
|
|
"learning_rate": 3.0954187447551996e-06,
|
||
|
|
"loss": 0.9894,
|
||
|
|
"step": 1040
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.56,
|
||
|
|
"learning_rate": 3.076200679130471e-06,
|
||
|
|
"loss": 0.9975,
|
||
|
|
"step": 1041
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.56,
|
||
|
|
"learning_rate": 3.0570316076487918e-06,
|
||
|
|
"loss": 1.0756,
|
||
|
|
"step": 1042
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.57,
|
||
|
|
"learning_rate": 3.0379116659542186e-06,
|
||
|
|
"loss": 1.0049,
|
||
|
|
"step": 1043
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.57,
|
||
|
|
"learning_rate": 3.0188409893431556e-06,
|
||
|
|
"loss": 1.0116,
|
||
|
|
"step": 1044
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.57,
|
||
|
|
"learning_rate": 2.999819712763402e-06,
|
||
|
|
"loss": 1.0514,
|
||
|
|
"step": 1045
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.57,
|
||
|
|
"learning_rate": 2.9808479708131864e-06,
|
||
|
|
"loss": 1.0405,
|
||
|
|
"step": 1046
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.58,
|
||
|
|
"learning_rate": 2.9619258977402253e-06,
|
||
|
|
"loss": 1.0393,
|
||
|
|
"step": 1047
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.58,
|
||
|
|
"learning_rate": 2.943053627440771e-06,
|
||
|
|
"loss": 1.0099,
|
||
|
|
"step": 1048
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.58,
|
||
|
|
"learning_rate": 2.943053627440771e-06,
|
||
|
|
"loss": 1.0145,
|
||
|
|
"step": 1049
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.58,
|
||
|
|
"learning_rate": 2.924231293458647e-06,
|
||
|
|
"loss": 1.0731,
|
||
|
|
"step": 1050
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.59,
|
||
|
|
"learning_rate": 2.90545902898434e-06,
|
||
|
|
"loss": 1.0022,
|
||
|
|
"step": 1051
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.59,
|
||
|
|
"learning_rate": 2.886736966854018e-06,
|
||
|
|
"loss": 1.0274,
|
||
|
|
"step": 1052
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.59,
|
||
|
|
"learning_rate": 2.8680652395486198e-06,
|
||
|
|
"loss": 1.0585,
|
||
|
|
"step": 1053
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.59,
|
||
|
|
"learning_rate": 2.849443979192892e-06,
|
||
|
|
"loss": 1.0792,
|
||
|
|
"step": 1054
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6,
|
||
|
|
"learning_rate": 2.8308733175544724e-06,
|
||
|
|
"loss": 1.0762,
|
||
|
|
"step": 1055
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6,
|
||
|
|
"learning_rate": 2.812353386042962e-06,
|
||
|
|
"loss": 1.0184,
|
||
|
|
"step": 1056
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6,
|
||
|
|
"learning_rate": 2.7938843157089734e-06,
|
||
|
|
"loss": 0.9722,
|
||
|
|
"step": 1057
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6,
|
||
|
|
"learning_rate": 2.775466237243226e-06,
|
||
|
|
"loss": 1.1118,
|
||
|
|
"step": 1058
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.61,
|
||
|
|
"learning_rate": 2.7570992809755937e-06,
|
||
|
|
"loss": 1.0514,
|
||
|
|
"step": 1059
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.61,
|
||
|
|
"learning_rate": 2.73878357687422e-06,
|
||
|
|
"loss": 1.0908,
|
||
|
|
"step": 1060
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.61,
|
||
|
|
"learning_rate": 2.720519254544568e-06,
|
||
|
|
"loss": 1.0791,
|
||
|
|
"step": 1061
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.61,
|
||
|
|
"learning_rate": 2.702306443228516e-06,
|
||
|
|
"loss": 1.0933,
|
||
|
|
"step": 1062
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.62,
|
||
|
|
"learning_rate": 2.702306443228516e-06,
|
||
|
|
"loss": 1.0118,
|
||
|
|
"step": 1063
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.62,
|
||
|
|
"learning_rate": 2.6841452718034343e-06,
|
||
|
|
"loss": 1.0542,
|
||
|
|
"step": 1064
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.62,
|
||
|
|
"learning_rate": 2.666035868781285e-06,
|
||
|
|
"loss": 1.0074,
|
||
|
|
"step": 1065
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.62,
|
||
|
|
"learning_rate": 2.6479783623077105e-06,
|
||
|
|
"loss": 1.0274,
|
||
|
|
"step": 1066
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.63,
|
||
|
|
"learning_rate": 2.62997288016112e-06,
|
||
|
|
"loss": 1.0626,
|
||
|
|
"step": 1067
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.63,
|
||
|
|
"learning_rate": 2.6120195497517818e-06,
|
||
|
|
"loss": 1.0031,
|
||
|
|
"step": 1068
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.63,
|
||
|
|
"learning_rate": 2.5941184981209354e-06,
|
||
|
|
"loss": 1.0309,
|
||
|
|
"step": 1069
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.63,
|
||
|
|
"learning_rate": 2.5941184981209354e-06,
|
||
|
|
"loss": 1.0782,
|
||
|
|
"step": 1070
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.64,
|
||
|
|
"learning_rate": 2.5762698519398832e-06,
|
||
|
|
"loss": 1.0939,
|
||
|
|
"step": 1071
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.64,
|
||
|
|
"learning_rate": 2.5584737375091016e-06,
|
||
|
|
"loss": 1.1096,
|
||
|
|
"step": 1072
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.64,
|
||
|
|
"learning_rate": 2.5407302807573387e-06,
|
||
|
|
"loss": 1.0452,
|
||
|
|
"step": 1073
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.64,
|
||
|
|
"learning_rate": 2.5230396072407204e-06,
|
||
|
|
"loss": 1.0232,
|
||
|
|
"step": 1074
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.65,
|
||
|
|
"learning_rate": 2.5054018421418737e-06,
|
||
|
|
"loss": 0.9717,
|
||
|
|
"step": 1075
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.65,
|
||
|
|
"learning_rate": 2.487817110269042e-06,
|
||
|
|
"loss": 1.0756,
|
||
|
|
"step": 1076
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.65,
|
||
|
|
"learning_rate": 2.470285536055188e-06,
|
||
|
|
"loss": 1.0175,
|
||
|
|
"step": 1077
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.65,
|
||
|
|
"learning_rate": 2.4528072435571158e-06,
|
||
|
|
"loss": 1.0394,
|
||
|
|
"step": 1078
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.66,
|
||
|
|
"learning_rate": 2.4353823564546064e-06,
|
||
|
|
"loss": 1.0886,
|
||
|
|
"step": 1079
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.66,
|
||
|
|
"learning_rate": 2.4353823564546064e-06,
|
||
|
|
"loss": 1.0737,
|
||
|
|
"step": 1080
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.66,
|
||
|
|
"learning_rate": 2.4353823564546064e-06,
|
||
|
|
"loss": 0.9719,
|
||
|
|
"step": 1081
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.66,
|
||
|
|
"learning_rate": 2.4353823564546064e-06,
|
||
|
|
"loss": 1.0594,
|
||
|
|
"step": 1082
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.67,
|
||
|
|
"learning_rate": 2.4180109980495293e-06,
|
||
|
|
"loss": 1.1362,
|
||
|
|
"step": 1083
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.67,
|
||
|
|
"learning_rate": 2.4006932912649816e-06,
|
||
|
|
"loss": 1.041,
|
||
|
|
"step": 1084
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.67,
|
||
|
|
"learning_rate": 2.3834293586444e-06,
|
||
|
|
"loss": 1.0417,
|
||
|
|
"step": 1085
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.67,
|
||
|
|
"learning_rate": 2.3662193223507135e-06,
|
||
|
|
"loss": 0.9742,
|
||
|
|
"step": 1086
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.67,
|
||
|
|
"learning_rate": 2.349063304165462e-06,
|
||
|
|
"loss": 1.0592,
|
||
|
|
"step": 1087
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.68,
|
||
|
|
"learning_rate": 2.331961425487956e-06,
|
||
|
|
"loss": 1.0067,
|
||
|
|
"step": 1088
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.68,
|
||
|
|
"learning_rate": 2.331961425487956e-06,
|
||
|
|
"loss": 1.0051,
|
||
|
|
"step": 1089
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.68,
|
||
|
|
"learning_rate": 2.3149138073343958e-06,
|
||
|
|
"loss": 1.0228,
|
||
|
|
"step": 1090
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.68,
|
||
|
|
"learning_rate": 2.297920570337019e-06,
|
||
|
|
"loss": 1.066,
|
||
|
|
"step": 1091
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.69,
|
||
|
|
"learning_rate": 2.2809818347432598e-06,
|
||
|
|
"loss": 1.0886,
|
||
|
|
"step": 1092
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.69,
|
||
|
|
"learning_rate": 2.2640977204148838e-06,
|
||
|
|
"loss": 0.9865,
|
||
|
|
"step": 1093
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.69,
|
||
|
|
"learning_rate": 2.2640977204148838e-06,
|
||
|
|
"loss": 0.9916,
|
||
|
|
"step": 1094
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.69,
|
||
|
|
"learning_rate": 2.2472683468271584e-06,
|
||
|
|
"loss": 1.0139,
|
||
|
|
"step": 1095
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7,
|
||
|
|
"learning_rate": 2.230493833067977e-06,
|
||
|
|
"loss": 1.0251,
|
||
|
|
"step": 1096
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7,
|
||
|
|
"learning_rate": 2.213774297837047e-06,
|
||
|
|
"loss": 1.1165,
|
||
|
|
"step": 1097
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7,
|
||
|
|
"learning_rate": 2.1971098594450315e-06,
|
||
|
|
"loss": 1.0507,
|
||
|
|
"step": 1098
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.7,
|
||
|
|
"learning_rate": 2.1805006358127213e-06,
|
||
|
|
"loss": 1.0245,
|
||
|
|
"step": 1099
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.71,
|
||
|
|
"learning_rate": 2.1805006358127213e-06,
|
||
|
|
"loss": 1.1373,
|
||
|
|
"step": 1100
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.71,
|
||
|
|
"learning_rate": 2.1639467444701934e-06,
|
||
|
|
"loss": 1.0135,
|
||
|
|
"step": 1101
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.71,
|
||
|
|
"learning_rate": 2.1474483025559857e-06,
|
||
|
|
"loss": 1.0316,
|
||
|
|
"step": 1102
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.71,
|
||
|
|
"learning_rate": 2.1310054268162628e-06,
|
||
|
|
"loss": 1.0715,
|
||
|
|
"step": 1103
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.72,
|
||
|
|
"learning_rate": 2.114618233603992e-06,
|
||
|
|
"loss": 1.0695,
|
||
|
|
"step": 1104
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.72,
|
||
|
|
"learning_rate": 2.0982868388781286e-06,
|
||
|
|
"loss": 1.0166,
|
||
|
|
"step": 1105
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.72,
|
||
|
|
"learning_rate": 2.0820113582027734e-06,
|
||
|
|
"loss": 1.1768,
|
||
|
|
"step": 1106
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.72,
|
||
|
|
"learning_rate": 2.0657919067463773e-06,
|
||
|
|
"loss": 1.0106,
|
||
|
|
"step": 1107
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.73,
|
||
|
|
"learning_rate": 2.0496285992809163e-06,
|
||
|
|
"loss": 1.1257,
|
||
|
|
"step": 1108
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.73,
|
||
|
|
"learning_rate": 2.033521550181078e-06,
|
||
|
|
"loss": 1.0428,
|
||
|
|
"step": 1109
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.73,
|
||
|
|
"learning_rate": 2.0174708734234596e-06,
|
||
|
|
"loss": 1.0674,
|
||
|
|
"step": 1110
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.73,
|
||
|
|
"learning_rate": 2.0174708734234596e-06,
|
||
|
|
"loss": 0.9989,
|
||
|
|
"step": 1111
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.74,
|
||
|
|
"learning_rate": 2.0014766825857514e-06,
|
||
|
|
"loss": 0.9387,
|
||
|
|
"step": 1112
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.74,
|
||
|
|
"learning_rate": 1.985539090845943e-06,
|
||
|
|
"loss": 1.0979,
|
||
|
|
"step": 1113
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.74,
|
||
|
|
"learning_rate": 1.9696582109815145e-06,
|
||
|
|
"loss": 1.0745,
|
||
|
|
"step": 1114
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.74,
|
||
|
|
"learning_rate": 1.9538341553686446e-06,
|
||
|
|
"loss": 1.0447,
|
||
|
|
"step": 1115
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.75,
|
||
|
|
"learning_rate": 1.93806703598141e-06,
|
||
|
|
"loss": 1.0419,
|
||
|
|
"step": 1116
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.75,
|
||
|
|
"learning_rate": 1.9223569643909978e-06,
|
||
|
|
"loss": 1.0366,
|
||
|
|
"step": 1117
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.75,
|
||
|
|
"learning_rate": 1.9067040517649115e-06,
|
||
|
|
"loss": 0.9998,
|
||
|
|
"step": 1118
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.75,
|
||
|
|
"learning_rate": 1.8911084088661903e-06,
|
||
|
|
"loss": 1.0471,
|
||
|
|
"step": 1119
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.76,
|
||
|
|
"learning_rate": 1.8911084088661903e-06,
|
||
|
|
"loss": 1.0032,
|
||
|
|
"step": 1120
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.76,
|
||
|
|
"learning_rate": 1.8755701460526166e-06,
|
||
|
|
"loss": 1.0189,
|
||
|
|
"step": 1121
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.76,
|
||
|
|
"learning_rate": 1.860089373275945e-06,
|
||
|
|
"loss": 1.0055,
|
||
|
|
"step": 1122
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.76,
|
||
|
|
"learning_rate": 1.8446662000811177e-06,
|
||
|
|
"loss": 0.9388,
|
||
|
|
"step": 1123
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.77,
|
||
|
|
"learning_rate": 1.8293007356054903e-06,
|
||
|
|
"loss": 1.035,
|
||
|
|
"step": 1124
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.77,
|
||
|
|
"learning_rate": 1.8139930885780621e-06,
|
||
|
|
"loss": 1.0502,
|
||
|
|
"step": 1125
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.77,
|
||
|
|
"learning_rate": 1.7987433673187026e-06,
|
||
|
|
"loss": 0.9969,
|
||
|
|
"step": 1126
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.77,
|
||
|
|
"learning_rate": 1.7835516797373908e-06,
|
||
|
|
"loss": 1.0394,
|
||
|
|
"step": 1127
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.78,
|
||
|
|
"learning_rate": 1.7684181333334437e-06,
|
||
|
|
"loss": 1.0101,
|
||
|
|
"step": 1128
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.78,
|
||
|
|
"learning_rate": 1.7533428351947634e-06,
|
||
|
|
"loss": 0.9991,
|
||
|
|
"step": 1129
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.78,
|
||
|
|
"learning_rate": 1.7383258919970746e-06,
|
||
|
|
"loss": 1.0013,
|
||
|
|
"step": 1130
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.78,
|
||
|
|
"learning_rate": 1.7233674100031728e-06,
|
||
|
|
"loss": 1.0078,
|
||
|
|
"step": 1131
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.79,
|
||
|
|
"learning_rate": 1.7233674100031728e-06,
|
||
|
|
"loss": 1.0005,
|
||
|
|
"step": 1132
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.79,
|
||
|
|
"learning_rate": 1.7084674950621694e-06,
|
||
|
|
"loss": 1.0359,
|
||
|
|
"step": 1133
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.79,
|
||
|
|
"learning_rate": 1.6936262526087432e-06,
|
||
|
|
"loss": 1.0419,
|
||
|
|
"step": 1134
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.79,
|
||
|
|
"learning_rate": 1.6936262526087432e-06,
|
||
|
|
"loss": 1.0236,
|
||
|
|
"step": 1135
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8,
|
||
|
|
"learning_rate": 1.6788437876623963e-06,
|
||
|
|
"loss": 1.0773,
|
||
|
|
"step": 1136
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8,
|
||
|
|
"learning_rate": 1.6641202048267102e-06,
|
||
|
|
"loss": 0.9716,
|
||
|
|
"step": 1137
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8,
|
||
|
|
"learning_rate": 1.6494556082886038e-06,
|
||
|
|
"loss": 1.0103,
|
||
|
|
"step": 1138
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.8,
|
||
|
|
"learning_rate": 1.634850101817601e-06,
|
||
|
|
"loss": 1.09,
|
||
|
|
"step": 1139
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.81,
|
||
|
|
"learning_rate": 1.6203037887650842e-06,
|
||
|
|
"loss": 1.0327,
|
||
|
|
"step": 1140
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.81,
|
||
|
|
"learning_rate": 1.6058167720635832e-06,
|
||
|
|
"loss": 1.0267,
|
||
|
|
"step": 1141
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.81,
|
||
|
|
"learning_rate": 1.5913891542260284e-06,
|
||
|
|
"loss": 1.0975,
|
||
|
|
"step": 1142
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.81,
|
||
|
|
"learning_rate": 1.5770210373450356e-06,
|
||
|
|
"loss": 0.949,
|
||
|
|
"step": 1143
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.82,
|
||
|
|
"learning_rate": 1.5627125230921725e-06,
|
||
|
|
"loss": 1.0099,
|
||
|
|
"step": 1144
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.82,
|
||
|
|
"learning_rate": 1.5484637127172609e-06,
|
||
|
|
"loss": 0.9873,
|
||
|
|
"step": 1145
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.82,
|
||
|
|
"learning_rate": 1.5342747070476339e-06,
|
||
|
|
"loss": 0.9921,
|
||
|
|
"step": 1146
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.82,
|
||
|
|
"learning_rate": 1.520145606487442e-06,
|
||
|
|
"loss": 1.0358,
|
||
|
|
"step": 1147
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.82,
|
||
|
|
"learning_rate": 1.506076511016935e-06,
|
||
|
|
"loss": 1.0532,
|
||
|
|
"step": 1148
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.83,
|
||
|
|
"learning_rate": 1.506076511016935e-06,
|
||
|
|
"loss": 0.9765,
|
||
|
|
"step": 1149
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.83,
|
||
|
|
"learning_rate": 1.506076511016935e-06,
|
||
|
|
"loss": 1.0929,
|
||
|
|
"step": 1150
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.83,
|
||
|
|
"learning_rate": 1.4920675201917467e-06,
|
||
|
|
"loss": 1.002,
|
||
|
|
"step": 1151
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.83,
|
||
|
|
"learning_rate": 1.4781187331422109e-06,
|
||
|
|
"loss": 0.9677,
|
||
|
|
"step": 1152
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.84,
|
||
|
|
"learning_rate": 1.4642302485726423e-06,
|
||
|
|
"loss": 1.0863,
|
||
|
|
"step": 1153
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.84,
|
||
|
|
"learning_rate": 1.4642302485726423e-06,
|
||
|
|
"loss": 1.0279,
|
||
|
|
"step": 1154
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.84,
|
||
|
|
"learning_rate": 1.4504021647606448e-06,
|
||
|
|
"loss": 0.9456,
|
||
|
|
"step": 1155
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.84,
|
||
|
|
"learning_rate": 1.4366345795564084e-06,
|
||
|
|
"loss": 1.0098,
|
||
|
|
"step": 1156
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.85,
|
||
|
|
"learning_rate": 1.4229275903820306e-06,
|
||
|
|
"loss": 0.9929,
|
||
|
|
"step": 1157
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.85,
|
||
|
|
"learning_rate": 1.409281294230821e-06,
|
||
|
|
"loss": 1.0472,
|
||
|
|
"step": 1158
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.85,
|
||
|
|
"learning_rate": 1.395695787666611e-06,
|
||
|
|
"loss": 1.0939,
|
||
|
|
"step": 1159
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.85,
|
||
|
|
"learning_rate": 1.3821711668230675e-06,
|
||
|
|
"loss": 0.9547,
|
||
|
|
"step": 1160
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.86,
|
||
|
|
"learning_rate": 1.3687075274030238e-06,
|
||
|
|
"loss": 0.9937,
|
||
|
|
"step": 1161
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.86,
|
||
|
|
"learning_rate": 1.3553049646777993e-06,
|
||
|
|
"loss": 1.0674,
|
||
|
|
"step": 1162
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.86,
|
||
|
|
"learning_rate": 1.34196357348652e-06,
|
||
|
|
"loss": 0.9952,
|
||
|
|
"step": 1163
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.86,
|
||
|
|
"learning_rate": 1.3286834482354506e-06,
|
||
|
|
"loss": 0.9764,
|
||
|
|
"step": 1164
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.87,
|
||
|
|
"learning_rate": 1.3154646828973217e-06,
|
||
|
|
"loss": 1.0367,
|
||
|
|
"step": 1165
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.87,
|
||
|
|
"learning_rate": 1.3023073710106726e-06,
|
||
|
|
"loss": 1.0518,
|
||
|
|
"step": 1166
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.87,
|
||
|
|
"learning_rate": 1.2892116056791927e-06,
|
||
|
|
"loss": 1.0889,
|
||
|
|
"step": 1167
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.87,
|
||
|
|
"learning_rate": 1.2761774795710502e-06,
|
||
|
|
"loss": 1.1367,
|
||
|
|
"step": 1168
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.88,
|
||
|
|
"learning_rate": 1.2632050849182365e-06,
|
||
|
|
"loss": 1.0241,
|
||
|
|
"step": 1169
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.88,
|
||
|
|
"learning_rate": 1.2502945135159272e-06,
|
||
|
|
"loss": 1.0073,
|
||
|
|
"step": 1170
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.88,
|
||
|
|
"learning_rate": 1.2374458567218217e-06,
|
||
|
|
"loss": 1.0031,
|
||
|
|
"step": 1171
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.88,
|
||
|
|
"learning_rate": 1.2246592054555062e-06,
|
||
|
|
"loss": 1.0898,
|
||
|
|
"step": 1172
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.89,
|
||
|
|
"learning_rate": 1.2119346501977914e-06,
|
||
|
|
"loss": 0.9809,
|
||
|
|
"step": 1173
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.89,
|
||
|
|
"learning_rate": 1.1992722809900925e-06,
|
||
|
|
"loss": 1.0642,
|
||
|
|
"step": 1174
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.89,
|
||
|
|
"learning_rate": 1.1866721874337827e-06,
|
||
|
|
"loss": 1.0443,
|
||
|
|
"step": 1175
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.89,
|
||
|
|
"learning_rate": 1.1741344586895642e-06,
|
||
|
|
"loss": 1.0289,
|
||
|
|
"step": 1176
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9,
|
||
|
|
"learning_rate": 1.1616591834768299e-06,
|
||
|
|
"loss": 1.0584,
|
||
|
|
"step": 1177
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9,
|
||
|
|
"learning_rate": 1.149246450073036e-06,
|
||
|
|
"loss": 1.0363,
|
||
|
|
"step": 1178
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9,
|
||
|
|
"learning_rate": 1.1368963463130866e-06,
|
||
|
|
"loss": 1.0247,
|
||
|
|
"step": 1179
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.9,
|
||
|
|
"learning_rate": 1.1246089595887023e-06,
|
||
|
|
"loss": 1.0164,
|
||
|
|
"step": 1180
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.91,
|
||
|
|
"learning_rate": 1.1123843768478148e-06,
|
||
|
|
"loss": 1.0541,
|
||
|
|
"step": 1181
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.91,
|
||
|
|
"learning_rate": 1.100222684593929e-06,
|
||
|
|
"loss": 1.0542,
|
||
|
|
"step": 1182
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.91,
|
||
|
|
"learning_rate": 1.088123968885534e-06,
|
||
|
|
"loss": 0.9276,
|
||
|
|
"step": 1183
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.91,
|
||
|
|
"learning_rate": 1.0760883153354818e-06,
|
||
|
|
"loss": 1.0496,
|
||
|
|
"step": 1184
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.92,
|
||
|
|
"learning_rate": 1.0641158091103832e-06,
|
||
|
|
"loss": 1.046,
|
||
|
|
"step": 1185
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.92,
|
||
|
|
"learning_rate": 1.0522065349300103e-06,
|
||
|
|
"loss": 0.9732,
|
||
|
|
"step": 1186
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.92,
|
||
|
|
"learning_rate": 1.0522065349300103e-06,
|
||
|
|
"loss": 0.9684,
|
||
|
|
"step": 1187
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.92,
|
||
|
|
"learning_rate": 1.040360577066688e-06,
|
||
|
|
"loss": 1.0563,
|
||
|
|
"step": 1188
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.93,
|
||
|
|
"learning_rate": 1.028578019344706e-06,
|
||
|
|
"loss": 1.035,
|
||
|
|
"step": 1189
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.93,
|
||
|
|
"learning_rate": 1.0168589451397204e-06,
|
||
|
|
"loss": 1.0778,
|
||
|
|
"step": 1190
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.93,
|
||
|
|
"learning_rate": 1.0052034373781716e-06,
|
||
|
|
"loss": 0.955,
|
||
|
|
"step": 1191
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.93,
|
||
|
|
"learning_rate": 1.0052034373781716e-06,
|
||
|
|
"loss": 1.1,
|
||
|
|
"step": 1192
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.94,
|
||
|
|
"learning_rate": 9.936115785366817e-07,
|
||
|
|
"loss": 1.0728,
|
||
|
|
"step": 1193
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.94,
|
||
|
|
"learning_rate": 9.820834506414866e-07,
|
||
|
|
"loss": 1.0634,
|
||
|
|
"step": 1194
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.94,
|
||
|
|
"learning_rate": 9.706191352678495e-07,
|
||
|
|
"loss": 1.0508,
|
||
|
|
"step": 1195
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.94,
|
||
|
|
"learning_rate": 9.592187135394826e-07,
|
||
|
|
"loss": 1.0482,
|
||
|
|
"step": 1196
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.95,
|
||
|
|
"learning_rate": 9.478822661279763e-07,
|
||
|
|
"loss": 1.1091,
|
||
|
|
"step": 1197
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.95,
|
||
|
|
"learning_rate": 9.366098732522233e-07,
|
||
|
|
"loss": 1.099,
|
||
|
|
"step": 1198
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.95,
|
||
|
|
"learning_rate": 9.254016146778555e-07,
|
||
|
|
"loss": 1.049,
|
||
|
|
"step": 1199
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.95,
|
||
|
|
"learning_rate": 9.1425756971668e-07,
|
||
|
|
"loss": 1.0137,
|
||
|
|
"step": 1200
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.96,
|
||
|
|
"learning_rate": 9.1425756971668e-07,
|
||
|
|
"loss": 0.9972,
|
||
|
|
"step": 1201
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.96,
|
||
|
|
"learning_rate": 9.03177817226113e-07,
|
||
|
|
"loss": 0.9507,
|
||
|
|
"step": 1202
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.96,
|
||
|
|
"learning_rate": 8.921624356086256e-07,
|
||
|
|
"loss": 1.0585,
|
||
|
|
"step": 1203
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.96,
|
||
|
|
"learning_rate": 8.81211502811189e-07,
|
||
|
|
"loss": 1.0801,
|
||
|
|
"step": 1204
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.97,
|
||
|
|
"learning_rate": 8.703250963247223e-07,
|
||
|
|
"loss": 1.0114,
|
||
|
|
"step": 1205
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.97,
|
||
|
|
"learning_rate": 8.595032931835423e-07,
|
||
|
|
"loss": 1.0778,
|
||
|
|
"step": 1206
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.97,
|
||
|
|
"learning_rate": 8.595032931835423e-07,
|
||
|
|
"loss": 1.1023,
|
||
|
|
"step": 1207
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.97,
|
||
|
|
"learning_rate": 8.595032931835423e-07,
|
||
|
|
"loss": 1.114,
|
||
|
|
"step": 1208
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.98,
|
||
|
|
"learning_rate": 8.487461699648203e-07,
|
||
|
|
"loss": 0.9095,
|
||
|
|
"step": 1209
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.98,
|
||
|
|
"learning_rate": 8.380538027880425e-07,
|
||
|
|
"loss": 1.0415,
|
||
|
|
"step": 1210
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.98,
|
||
|
|
"learning_rate": 8.274262673144651e-07,
|
||
|
|
"loss": 1.0887,
|
||
|
|
"step": 1211
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.98,
|
||
|
|
"learning_rate": 8.168636387465856e-07,
|
||
|
|
"loss": 1.1132,
|
||
|
|
"step": 1212
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.98,
|
||
|
|
"learning_rate": 8.063659918276056e-07,
|
||
|
|
"loss": 1.0403,
|
||
|
|
"step": 1213
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.99,
|
||
|
|
"learning_rate": 7.95933400840907e-07,
|
||
|
|
"loss": 0.9683,
|
||
|
|
"step": 1214
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.99,
|
||
|
|
"learning_rate": 7.855659396095183e-07,
|
||
|
|
"loss": 1.1194,
|
||
|
|
"step": 1215
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.99,
|
||
|
|
"learning_rate": 7.752636814956027e-07,
|
||
|
|
"loss": 0.9638,
|
||
|
|
"step": 1216
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.99,
|
||
|
|
"learning_rate": 7.6502669939993e-07,
|
||
|
|
"loss": 0.9801,
|
||
|
|
"step": 1217
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.0,
|
||
|
|
"learning_rate": 7.548550657613651e-07,
|
||
|
|
"loss": 1.046,
|
||
|
|
"step": 1218
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"max_steps": 1218,
|
||
|
|
"num_train_epochs": 3,
|
||
|
|
"total_flos": 1.2021253265917215e+18,
|
||
|
|
"trial_name": null,
|
||
|
|
"trial_params": null
|
||
|
|
}
|