30617 lines
554 KiB
JSON
30617 lines
554 KiB
JSON
{
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 1.9933554817275747,
|
|
"global_step": 10200,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.166911338316773e-06,
|
|
"loss": 0.8613,
|
|
"step": 2
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 8.333822676633545e-06,
|
|
"loss": 0.8862,
|
|
"step": 4
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.0771309553378665e-05,
|
|
"loss": 0.671,
|
|
"step": 6
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.2500734014950318e-05,
|
|
"loss": 0.6001,
|
|
"step": 8
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.384217984365919e-05,
|
|
"loss": 0.5201,
|
|
"step": 10
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.493822089169544e-05,
|
|
"loss": 0.5519,
|
|
"step": 12
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.5864910393718002e-05,
|
|
"loss": 0.5087,
|
|
"step": 14
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.666764535326709e-05,
|
|
"loss": 0.5112,
|
|
"step": 16
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.7375707768440558e-05,
|
|
"loss": 0.5437,
|
|
"step": 18
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.800909118197596e-05,
|
|
"loss": 0.4583,
|
|
"step": 20
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.8582056174148076e-05,
|
|
"loss": 0.5037,
|
|
"step": 22
|
|
},
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 1.9105132230012212e-05,
|
|
"loss": 0.4601,
|
|
"step": 24
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.9586315556596615e-05,
|
|
"loss": 0.4039,
|
|
"step": 26
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0031821732034778e-05,
|
|
"loss": 0.5029,
|
|
"step": 28
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.044657805872108e-05,
|
|
"loss": 0.4617,
|
|
"step": 30
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.0834556691583863e-05,
|
|
"loss": 0.4431,
|
|
"step": 32
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.1199006596471306e-05,
|
|
"loss": 0.4069,
|
|
"step": 34
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.154261910675733e-05,
|
|
"loss": 0.4889,
|
|
"step": 36
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.1867648658432627e-05,
|
|
"loss": 0.4755,
|
|
"step": 38
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.2176002520292734e-05,
|
|
"loss": 0.4828,
|
|
"step": 40
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.24693086087799e-05,
|
|
"loss": 0.423,
|
|
"step": 42
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.274896751246485e-05,
|
|
"loss": 0.4388,
|
|
"step": 44
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.301619294258914e-05,
|
|
"loss": 0.4351,
|
|
"step": 46
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3272043568328985e-05,
|
|
"loss": 0.4718,
|
|
"step": 48
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.35174483490016e-05,
|
|
"loss": 0.4652,
|
|
"step": 50
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3753226894913388e-05,
|
|
"loss": 0.4678,
|
|
"step": 52
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.3980105983502452e-05,
|
|
"loss": 0.4508,
|
|
"step": 54
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.4198733070351554e-05,
|
|
"loss": 0.4224,
|
|
"step": 56
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.4409687428241253e-05,
|
|
"loss": 0.4264,
|
|
"step": 58
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.461348939703785e-05,
|
|
"loss": 0.3688,
|
|
"step": 60
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.481060811631497e-05,
|
|
"loss": 0.4238,
|
|
"step": 62
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5001468029900636e-05,
|
|
"loss": 0.4525,
|
|
"step": 64
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5186454389209963e-05,
|
|
"loss": 0.4149,
|
|
"step": 66
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5365917934788082e-05,
|
|
"loss": 0.514,
|
|
"step": 68
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5540178899060418e-05,
|
|
"loss": 0.465,
|
|
"step": 70
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.5709530445074103e-05,
|
|
"loss": 0.476,
|
|
"step": 72
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.587424163398852e-05,
|
|
"loss": 0.4485,
|
|
"step": 74
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.6034559996749396e-05,
|
|
"loss": 0.4395,
|
|
"step": 76
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.6190713771658503e-05,
|
|
"loss": 0.4288,
|
|
"step": 78
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.6342913858609503e-05,
|
|
"loss": 0.4274,
|
|
"step": 80
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.6491355531981626e-05,
|
|
"loss": 0.3966,
|
|
"step": 82
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.663621994709667e-05,
|
|
"loss": 0.454,
|
|
"step": 84
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.6777675469393626e-05,
|
|
"loss": 0.4132,
|
|
"step": 86
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.6915878850781624e-05,
|
|
"loss": 0.3928,
|
|
"step": 88
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7050976273782973e-05,
|
|
"loss": 0.4085,
|
|
"step": 90
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7183104280905912e-05,
|
|
"loss": 0.4634,
|
|
"step": 92
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7312390604060268e-05,
|
|
"loss": 0.428,
|
|
"step": 94
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7438954906645757e-05,
|
|
"loss": 0.4706,
|
|
"step": 96
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.756290944911924e-05,
|
|
"loss": 0.4132,
|
|
"step": 98
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.768435968731838e-05,
|
|
"loss": 0.4888,
|
|
"step": 100
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7803404811533196e-05,
|
|
"loss": 0.4356,
|
|
"step": 102
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.7920138233230157e-05,
|
|
"loss": 0.4141,
|
|
"step": 104
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.803464802541273e-05,
|
|
"loss": 0.4459,
|
|
"step": 106
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.8147017321819225e-05,
|
|
"loss": 0.4526,
|
|
"step": 108
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.8257324679490488e-05,
|
|
"loss": 0.45,
|
|
"step": 110
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.8365644408668323e-05,
|
|
"loss": 0.4713,
|
|
"step": 112
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.8472046873494518e-05,
|
|
"loss": 0.4479,
|
|
"step": 114
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.8576598766558025e-05,
|
|
"loss": 0.4359,
|
|
"step": 116
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.8679363359972988e-05,
|
|
"loss": 0.4395,
|
|
"step": 118
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.8780400735354624e-05,
|
|
"loss": 0.3921,
|
|
"step": 120
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.8879767994786194e-05,
|
|
"loss": 0.4476,
|
|
"step": 122
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.897751945463174e-05,
|
|
"loss": 0.4707,
|
|
"step": 124
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.907370682384179e-05,
|
|
"loss": 0.4324,
|
|
"step": 126
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 2.916837936821741e-05,
|
|
"loss": 0.4741,
|
|
"step": 128
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 2.9261584061939028e-05,
|
|
"loss": 0.4003,
|
|
"step": 130
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 2.935336572752674e-05,
|
|
"loss": 0.4531,
|
|
"step": 132
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 2.944376716527608e-05,
|
|
"loss": 0.4572,
|
|
"step": 134
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 2.953282927310485e-05,
|
|
"loss": 0.4388,
|
|
"step": 136
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 2.9620591157651034e-05,
|
|
"loss": 0.4437,
|
|
"step": 138
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 2.970709023737719e-05,
|
|
"loss": 0.431,
|
|
"step": 140
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 2.979236233836175e-05,
|
|
"loss": 0.4008,
|
|
"step": 142
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 2.987644178339088e-05,
|
|
"loss": 0.4243,
|
|
"step": 144
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 2.9959361474905535e-05,
|
|
"loss": 0.3868,
|
|
"step": 146
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 3.0041152972305286e-05,
|
|
"loss": 0.4398,
|
|
"step": 148
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 3.0121846564063495e-05,
|
|
"loss": 0.4377,
|
|
"step": 150
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 3.0201471335066172e-05,
|
|
"loss": 0.4769,
|
|
"step": 152
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 3.028005522954931e-05,
|
|
"loss": 0.4213,
|
|
"step": 154
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 3.035762510997528e-05,
|
|
"loss": 0.4177,
|
|
"step": 156
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 3.0434206812158934e-05,
|
|
"loss": 0.4468,
|
|
"step": 158
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 3.050982519692628e-05,
|
|
"loss": 0.4779,
|
|
"step": 160
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 3.058450419856434e-05,
|
|
"loss": 0.4385,
|
|
"step": 162
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 3.06582668702984e-05,
|
|
"loss": 0.4241,
|
|
"step": 164
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 3.073113542701279e-05,
|
|
"loss": 0.4619,
|
|
"step": 166
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 3.0803131285413445e-05,
|
|
"loss": 0.4441,
|
|
"step": 168
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 3.087427510181372e-05,
|
|
"loss": 0.4419,
|
|
"step": 170
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 3.09445868077104e-05,
|
|
"loss": 0.4652,
|
|
"step": 172
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 3.101408564330315e-05,
|
|
"loss": 0.4457,
|
|
"step": 174
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 3.108279018909839e-05,
|
|
"loss": 0.4712,
|
|
"step": 176
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 3.1150718395727636e-05,
|
|
"loss": 0.4075,
|
|
"step": 178
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 3.121788761209974e-05,
|
|
"loss": 0.4359,
|
|
"step": 180
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 3.1284314611997845e-05,
|
|
"loss": 0.4196,
|
|
"step": 182
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 3.135001561922269e-05,
|
|
"loss": 0.3966,
|
|
"step": 184
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 3.141500633137686e-05,
|
|
"loss": 0.4343,
|
|
"step": 186
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 3.147930194237704e-05,
|
|
"loss": 0.4342,
|
|
"step": 188
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 3.1542917163775046e-05,
|
|
"loss": 0.4633,
|
|
"step": 190
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 3.160586624496253e-05,
|
|
"loss": 0.433,
|
|
"step": 192
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 3.16681629923288e-05,
|
|
"loss": 0.4554,
|
|
"step": 194
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 3.1729820787436004e-05,
|
|
"loss": 0.4449,
|
|
"step": 196
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 3.179085260427186e-05,
|
|
"loss": 0.4346,
|
|
"step": 198
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 3.1851271025635146e-05,
|
|
"loss": 0.458,
|
|
"step": 200
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 3.1911088258706163e-05,
|
|
"loss": 0.4696,
|
|
"step": 202
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 3.1970316149849966e-05,
|
|
"loss": 0.4346,
|
|
"step": 204
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 3.202896619869761e-05,
|
|
"loss": 0.4246,
|
|
"step": 206
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 3.208704957154693e-05,
|
|
"loss": 0.4651,
|
|
"step": 208
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 3.214457711412231e-05,
|
|
"loss": 0.4081,
|
|
"step": 210
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 3.220155936372951e-05,
|
|
"loss": 0.4123,
|
|
"step": 212
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 3.225800656083991e-05,
|
|
"loss": 0.4275,
|
|
"step": 214
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 3.2313928660136e-05,
|
|
"loss": 0.4508,
|
|
"step": 216
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 3.23693353410476e-05,
|
|
"loss": 0.481,
|
|
"step": 218
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 3.242423601780726e-05,
|
|
"loss": 0.4551,
|
|
"step": 220
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 3.247863984905041e-05,
|
|
"loss": 0.4171,
|
|
"step": 222
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 3.253255574698509e-05,
|
|
"loss": 0.3974,
|
|
"step": 224
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 3.258599238615415e-05,
|
|
"loss": 0.4297,
|
|
"step": 226
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 3.263895821181129e-05,
|
|
"loss": 0.4089,
|
|
"step": 228
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 3.269146144793155e-05,
|
|
"loss": 0.4097,
|
|
"step": 230
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 3.27435101048748e-05,
|
|
"loss": 0.4415,
|
|
"step": 232
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 3.27951119867204e-05,
|
|
"loss": 0.3976,
|
|
"step": 234
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 3.284627469828977e-05,
|
|
"loss": 0.4255,
|
|
"step": 236
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 3.289700565187254e-05,
|
|
"loss": 0.4073,
|
|
"step": 238
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 3.29473120736714e-05,
|
|
"loss": 0.4419,
|
|
"step": 240
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 3.2997201009979375e-05,
|
|
"loss": 0.4231,
|
|
"step": 242
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 3.304667933310297e-05,
|
|
"loss": 0.4432,
|
|
"step": 244
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 3.3095753747043514e-05,
|
|
"loss": 0.3816,
|
|
"step": 246
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 3.3144430792948514e-05,
|
|
"loss": 0.4064,
|
|
"step": 248
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 3.3192716854344016e-05,
|
|
"loss": 0.3994,
|
|
"step": 250
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 3.324061816215856e-05,
|
|
"loss": 0.4493,
|
|
"step": 252
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 3.328814079954852e-05,
|
|
"loss": 0.4332,
|
|
"step": 254
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 3.333529070653418e-05,
|
|
"loss": 0.4734,
|
|
"step": 256
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 3.338207368445552e-05,
|
|
"loss": 0.4186,
|
|
"step": 258
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 3.34284954002558e-05,
|
|
"loss": 0.3688,
|
|
"step": 260
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 3.347456139060116e-05,
|
|
"loss": 0.4121,
|
|
"step": 262
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 3.352027706584351e-05,
|
|
"loss": 0.459,
|
|
"step": 264
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 3.3565647713833856e-05,
|
|
"loss": 0.4092,
|
|
"step": 266
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 3.3610678503592854e-05,
|
|
"loss": 0.4398,
|
|
"step": 268
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 3.3655374488844864e-05,
|
|
"loss": 0.4355,
|
|
"step": 270
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 3.369974061142163e-05,
|
|
"loss": 0.4172,
|
|
"step": 272
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 3.374378170454122e-05,
|
|
"loss": 0.4399,
|
|
"step": 274
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 3.37875024959678e-05,
|
|
"loss": 0.4326,
|
|
"step": 276
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 3.383090761105733e-05,
|
|
"loss": 0.4358,
|
|
"step": 278
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 3.387400157569396e-05,
|
|
"loss": 0.449,
|
|
"step": 280
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 3.3916788819122155e-05,
|
|
"loss": 0.4253,
|
|
"step": 282
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 3.395927367667852e-05,
|
|
"loss": 0.4261,
|
|
"step": 284
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 3.400146039242791e-05,
|
|
"loss": 0.3969,
|
|
"step": 286
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 3.404335312170765e-05,
|
|
"loss": 0.4168,
|
|
"step": 288
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 3.408495593358367e-05,
|
|
"loss": 0.3839,
|
|
"step": 290
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 3.412627281322231e-05,
|
|
"loss": 0.3979,
|
|
"step": 292
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 3.4167307664181126e-05,
|
|
"loss": 0.4237,
|
|
"step": 294
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 3.420806431062206e-05,
|
|
"loss": 0.4489,
|
|
"step": 296
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 3.424854649945008e-05,
|
|
"loss": 0.4149,
|
|
"step": 298
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 3.428875790238027e-05,
|
|
"loss": 0.4254,
|
|
"step": 300
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 3.4328702117936315e-05,
|
|
"loss": 0.4333,
|
|
"step": 302
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 3.436838267338295e-05,
|
|
"loss": 0.4568,
|
|
"step": 304
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 3.440780302659509e-05,
|
|
"loss": 0.4558,
|
|
"step": 306
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 3.444696656786607e-05,
|
|
"loss": 0.4667,
|
|
"step": 308
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 3.4485876621657384e-05,
|
|
"loss": 0.4117,
|
|
"step": 310
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 3.4524536448292055e-05,
|
|
"loss": 0.4336,
|
|
"step": 312
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 3.456294924559408e-05,
|
|
"loss": 0.4391,
|
|
"step": 314
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 3.460111815047571e-05,
|
|
"loss": 0.4188,
|
|
"step": 316
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 3.463904624047462e-05,
|
|
"loss": 0.4353,
|
|
"step": 318
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 3.467673653524305e-05,
|
|
"loss": 0.4296,
|
|
"step": 320
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 3.4714191997990376e-05,
|
|
"loss": 0.4324,
|
|
"step": 322
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 3.4751415536881115e-05,
|
|
"loss": 0.4608,
|
|
"step": 324
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 3.478841000638994e-05,
|
|
"loss": 0.438,
|
|
"step": 326
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 3.4825178208615175e-05,
|
|
"loss": 0.4108,
|
|
"step": 328
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 3.486172289455238e-05,
|
|
"loss": 0.3646,
|
|
"step": 330
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 3.4898046765329555e-05,
|
|
"loss": 0.4075,
|
|
"step": 332
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 3.493415247340513e-05,
|
|
"loss": 0.3971,
|
|
"step": 334
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 3.4970042623730214e-05,
|
|
"loss": 0.4009,
|
|
"step": 336
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 3.5005719774876455e-05,
|
|
"loss": 0.4304,
|
|
"step": 338
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 3.50411864401305e-05,
|
|
"loss": 0.4904,
|
|
"step": 340
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 3.507644508855641e-05,
|
|
"loss": 0.4373,
|
|
"step": 342
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 3.511149814602717e-05,
|
|
"loss": 0.4512,
|
|
"step": 344
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 3.514634799622618e-05,
|
|
"loss": 0.39,
|
|
"step": 346
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 3.518099698161992e-05,
|
|
"loss": 0.4095,
|
|
"step": 348
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 3.5215447404402826e-05,
|
|
"loss": 0.4341,
|
|
"step": 350
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 3.5249701527415166e-05,
|
|
"loss": 0.4224,
|
|
"step": 352
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 3.528376157503488e-05,
|
|
"loss": 0.4611,
|
|
"step": 354
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 3.5317629734044405e-05,
|
|
"loss": 0.3829,
|
|
"step": 356
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 3.535130815447316e-05,
|
|
"loss": 0.4337,
|
|
"step": 358
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 3.538479895041652e-05,
|
|
"loss": 0.4108,
|
|
"step": 360
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 3.54181042008322e-05,
|
|
"loss": 0.4385,
|
|
"step": 362
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 3.545122595031462e-05,
|
|
"loss": 0.4148,
|
|
"step": 364
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 3.5484166209848085e-05,
|
|
"loss": 0.4165,
|
|
"step": 366
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 3.551692695753946e-05,
|
|
"loss": 0.4503,
|
|
"step": 368
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 3.554951013933093e-05,
|
|
"loss": 0.4548,
|
|
"step": 370
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 3.558191766969363e-05,
|
|
"loss": 0.4064,
|
|
"step": 372
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 3.5614151432302605e-05,
|
|
"loss": 0.4248,
|
|
"step": 374
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 3.564621328069381e-05,
|
|
"loss": 0.4637,
|
|
"step": 376
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 3.567810503890368e-05,
|
|
"loss": 0.4363,
|
|
"step": 378
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 3.5709828502091815e-05,
|
|
"loss": 0.4037,
|
|
"step": 380
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 3.574138543714731e-05,
|
|
"loss": 0.4579,
|
|
"step": 382
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.57727775832793e-05,
|
|
"loss": 0.4178,
|
|
"step": 384
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.580400665259212e-05,
|
|
"loss": 0.3998,
|
|
"step": 386
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.5835074330645566e-05,
|
|
"loss": 0.415,
|
|
"step": 388
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.586598227700092e-05,
|
|
"loss": 0.4077,
|
|
"step": 390
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.589673212575278e-05,
|
|
"loss": 0.452,
|
|
"step": 392
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.5927325486047535e-05,
|
|
"loss": 0.4371,
|
|
"step": 394
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.5957763942588633e-05,
|
|
"loss": 0.4142,
|
|
"step": 396
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.598804905612913e-05,
|
|
"loss": 0.3987,
|
|
"step": 398
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.601818236395192e-05,
|
|
"loss": 0.4279,
|
|
"step": 400
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.6048165380337976e-05,
|
|
"loss": 0.4226,
|
|
"step": 402
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.607799959702294e-05,
|
|
"loss": 0.4547,
|
|
"step": 404
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.6107686483642486e-05,
|
|
"loss": 0.4287,
|
|
"step": 406
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.613722748816674e-05,
|
|
"loss": 0.4406,
|
|
"step": 408
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.616662403732404e-05,
|
|
"loss": 0.4685,
|
|
"step": 410
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.619587753701438e-05,
|
|
"loss": 0.4758,
|
|
"step": 412
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.6224989372712924e-05,
|
|
"loss": 0.4358,
|
|
"step": 414
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.62539609098637e-05,
|
|
"loss": 0.4749,
|
|
"step": 416
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.628279349426393e-05,
|
|
"loss": 0.4383,
|
|
"step": 418
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.6311488452439085e-05,
|
|
"loss": 0.4353,
|
|
"step": 420
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.634004709200918e-05,
|
|
"loss": 0.384,
|
|
"step": 422
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.6368470702046284e-05,
|
|
"loss": 0.4243,
|
|
"step": 424
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.639676055342364e-05,
|
|
"loss": 0.4099,
|
|
"step": 426
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.642491789915669e-05,
|
|
"loss": 0.4266,
|
|
"step": 428
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.645294397473604e-05,
|
|
"loss": 0.3943,
|
|
"step": 430
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.648083999845277e-05,
|
|
"loss": 0.436,
|
|
"step": 432
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 3.6508607171716195e-05,
|
|
"loss": 0.3705,
|
|
"step": 434
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.653624667936437e-05,
|
|
"loss": 0.4243,
|
|
"step": 436
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.656375968996743e-05,
|
|
"loss": 0.4124,
|
|
"step": 438
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.6591147356124037e-05,
|
|
"loss": 0.4257,
|
|
"step": 440
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.661841081475114e-05,
|
|
"loss": 0.4119,
|
|
"step": 442
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.6645551187367184e-05,
|
|
"loss": 0.4335,
|
|
"step": 444
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.667256958036887e-05,
|
|
"loss": 0.381,
|
|
"step": 446
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.669946708530187e-05,
|
|
"loss": 0.4088,
|
|
"step": 448
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.672624477912538e-05,
|
|
"loss": 0.4084,
|
|
"step": 450
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.675290372447092e-05,
|
|
"loss": 0.4113,
|
|
"step": 452
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.6779444969895275e-05,
|
|
"loss": 0.4076,
|
|
"step": 454
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.6805869550128066e-05,
|
|
"loss": 0.4133,
|
|
"step": 456
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.683217848631373e-05,
|
|
"loss": 0.4448,
|
|
"step": 458
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.685837278624833e-05,
|
|
"loss": 0.4133,
|
|
"step": 460
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.688445344461119e-05,
|
|
"loss": 0.3952,
|
|
"step": 462
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.6910421443191574e-05,
|
|
"loss": 0.4047,
|
|
"step": 464
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.693627775111036e-05,
|
|
"loss": 0.4089,
|
|
"step": 466
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.6962023325037176e-05,
|
|
"loss": 0.4404,
|
|
"step": 468
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.698765910940268e-05,
|
|
"loss": 0.4141,
|
|
"step": 470
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.701318603660653e-05,
|
|
"loss": 0.4017,
|
|
"step": 472
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.703860502722083e-05,
|
|
"loss": 0.4318,
|
|
"step": 474
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.7063916990189315e-05,
|
|
"loss": 0.4291,
|
|
"step": 476
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.7089122823022406e-05,
|
|
"loss": 0.4266,
|
|
"step": 478
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.711422341198817e-05,
|
|
"loss": 0.4271,
|
|
"step": 480
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.7139219632299284e-05,
|
|
"loss": 0.4285,
|
|
"step": 482
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.716411234829615e-05,
|
|
"loss": 0.4426,
|
|
"step": 484
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 3.718890241362624e-05,
|
|
"loss": 0.4451,
|
|
"step": 486
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.7213590671419746e-05,
|
|
"loss": 0.4036,
|
|
"step": 488
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.723817795446165e-05,
|
|
"loss": 0.402,
|
|
"step": 490
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.726266508536029e-05,
|
|
"loss": 0.4011,
|
|
"step": 492
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.7287052876712466e-05,
|
|
"loss": 0.4331,
|
|
"step": 494
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.731134213126529e-05,
|
|
"loss": 0.4114,
|
|
"step": 496
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.7335533642074676e-05,
|
|
"loss": 0.4336,
|
|
"step": 498
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.735962819266079e-05,
|
|
"loss": 0.4479,
|
|
"step": 500
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.738362655716023e-05,
|
|
"loss": 0.3843,
|
|
"step": 502
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.7407529500475336e-05,
|
|
"loss": 0.4518,
|
|
"step": 504
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.743133777842044e-05,
|
|
"loss": 0.4371,
|
|
"step": 506
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.7455052137865294e-05,
|
|
"loss": 0.4536,
|
|
"step": 508
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.747867331687561e-05,
|
|
"loss": 0.4263,
|
|
"step": 510
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.750220204485096e-05,
|
|
"loss": 0.4155,
|
|
"step": 512
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.752563904265985e-05,
|
|
"loss": 0.5803,
|
|
"step": 514
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.754898502277229e-05,
|
|
"loss": 0.394,
|
|
"step": 516
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.757224068938975e-05,
|
|
"loss": 0.4158,
|
|
"step": 518
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.759540673857257e-05,
|
|
"loss": 0.4178,
|
|
"step": 520
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.761848385836504e-05,
|
|
"loss": 0.4119,
|
|
"step": 522
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.7641472728917935e-05,
|
|
"loss": 0.4251,
|
|
"step": 524
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.766437402260888e-05,
|
|
"loss": 0.4089,
|
|
"step": 526
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.768718840416028e-05,
|
|
"loss": 0.4341,
|
|
"step": 528
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.770991653075515e-05,
|
|
"loss": 0.4369,
|
|
"step": 530
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.7732559052150626e-05,
|
|
"loss": 0.4084,
|
|
"step": 532
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.7755116610789527e-05,
|
|
"loss": 0.3943,
|
|
"step": 534
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 3.777758984190963e-05,
|
|
"loss": 0.4171,
|
|
"step": 536
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 3.779997937365106e-05,
|
|
"loss": 0.3927,
|
|
"step": 538
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 3.782228582716164e-05,
|
|
"loss": 0.4158,
|
|
"step": 540
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 3.7844509816700255e-05,
|
|
"loss": 0.4473,
|
|
"step": 542
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 3.7866651949738396e-05,
|
|
"loss": 0.4312,
|
|
"step": 544
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 3.7888712827059736e-05,
|
|
"loss": 0.433,
|
|
"step": 546
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 3.791069304285799e-05,
|
|
"loss": 0.4697,
|
|
"step": 548
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 3.79325931848329e-05,
|
|
"loss": 0.4625,
|
|
"step": 550
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 3.795441383428457e-05,
|
|
"loss": 0.4318,
|
|
"step": 552
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 3.7976155566206025e-05,
|
|
"loss": 0.427,
|
|
"step": 554
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 3.79978189493741e-05,
|
|
"loss": 0.4155,
|
|
"step": 556
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 3.801940454643875e-05,
|
|
"loss": 0.4121,
|
|
"step": 558
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 3.804091291401073e-05,
|
|
"loss": 0.4604,
|
|
"step": 560
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 3.8062344602747734e-05,
|
|
"loss": 0.3972,
|
|
"step": 562
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 3.808370015743893e-05,
|
|
"loss": 0.4404,
|
|
"step": 564
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 3.8104980117088116e-05,
|
|
"loss": 0.4082,
|
|
"step": 566
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 3.8126185014995295e-05,
|
|
"loss": 0.4301,
|
|
"step": 568
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 3.814731537883694e-05,
|
|
"loss": 0.3864,
|
|
"step": 570
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 3.816837173074469e-05,
|
|
"loss": 0.4008,
|
|
"step": 572
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 3.8189354587382856e-05,
|
|
"loss": 0.4584,
|
|
"step": 574
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 3.8210264460024424e-05,
|
|
"loss": 0.4276,
|
|
"step": 576
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 3.823110185462584e-05,
|
|
"loss": 0.4529,
|
|
"step": 578
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 3.8251867271900444e-05,
|
|
"loss": 0.5079,
|
|
"step": 580
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 3.827256120739069e-05,
|
|
"loss": 0.4164,
|
|
"step": 582
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 3.829318415153908e-05,
|
|
"loss": 0.4133,
|
|
"step": 584
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 3.831373658975793e-05,
|
|
"loss": 0.4506,
|
|
"step": 586
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 3.83342190024979e-05,
|
|
"loss": 0.3982,
|
|
"step": 588
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 3.83546318653154e-05,
|
|
"loss": 0.3819,
|
|
"step": 590
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 3.837497564893884e-05,
|
|
"loss": 0.4127,
|
|
"step": 592
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 3.839525081933375e-05,
|
|
"loss": 0.433,
|
|
"step": 594
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 3.841545783776685e-05,
|
|
"loss": 0.4191,
|
|
"step": 596
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 3.843559716086898e-05,
|
|
"loss": 0.3725,
|
|
"step": 598
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 3.8455669240697036e-05,
|
|
"loss": 0.4454,
|
|
"step": 600
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 3.847567452479486e-05,
|
|
"loss": 0.431,
|
|
"step": 602
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 3.8495613456253084e-05,
|
|
"loss": 0.4133,
|
|
"step": 604
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 3.8515486473768054e-05,
|
|
"loss": 0.3997,
|
|
"step": 606
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 3.853529401169972e-05,
|
|
"loss": 0.4156,
|
|
"step": 608
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 3.855503650012861e-05,
|
|
"loss": 0.4344,
|
|
"step": 610
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 3.857471436491186e-05,
|
|
"loss": 0.4282,
|
|
"step": 612
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 3.8594328027738344e-05,
|
|
"loss": 0.4011,
|
|
"step": 614
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 3.861387790618285e-05,
|
|
"loss": 0.4354,
|
|
"step": 616
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 3.86333644137595e-05,
|
|
"loss": 0.3984,
|
|
"step": 618
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 3.865278795997416e-05,
|
|
"loss": 0.4125,
|
|
"step": 620
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 3.86721489503761e-05,
|
|
"loss": 0.4651,
|
|
"step": 622
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 3.8691447786608824e-05,
|
|
"loss": 0.4323,
|
|
"step": 624
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 3.871068486646004e-05,
|
|
"loss": 0.4473,
|
|
"step": 626
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 3.872986058391086e-05,
|
|
"loss": 0.4612,
|
|
"step": 628
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 3.8748975329184206e-05,
|
|
"loss": 0.416,
|
|
"step": 630
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 3.876802948879248e-05,
|
|
"loss": 0.4281,
|
|
"step": 632
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 3.878702344558445e-05,
|
|
"loss": 0.3876,
|
|
"step": 634
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 3.88059575787914e-05,
|
|
"loss": 0.3845,
|
|
"step": 636
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 3.882483226407255e-05,
|
|
"loss": 0.377,
|
|
"step": 638
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 3.884364787355983e-05,
|
|
"loss": 0.4018,
|
|
"step": 640
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 3.886240477590181e-05,
|
|
"loss": 0.4475,
|
|
"step": 642
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 3.8881103336307145e-05,
|
|
"loss": 0.4387,
|
|
"step": 644
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 3.889974391658715e-05,
|
|
"loss": 0.4329,
|
|
"step": 646
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 3.8918326875197885e-05,
|
|
"loss": 0.4165,
|
|
"step": 648
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 3.893685256728144e-05,
|
|
"loss": 0.4237,
|
|
"step": 650
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 3.895532134470672e-05,
|
|
"loss": 0.4422,
|
|
"step": 652
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 3.897373355610949e-05,
|
|
"loss": 0.4543,
|
|
"step": 654
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 3.8992089546931944e-05,
|
|
"loss": 0.4482,
|
|
"step": 656
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 3.90103896594615e-05,
|
|
"loss": 0.4514,
|
|
"step": 658
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 3.902863423286915e-05,
|
|
"loss": 0.4344,
|
|
"step": 660
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 3.90468236032472e-05,
|
|
"loss": 0.438,
|
|
"step": 662
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 3.906495810364633e-05,
|
|
"loss": 0.4642,
|
|
"step": 664
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 3.9083038064112306e-05,
|
|
"loss": 0.3972,
|
|
"step": 666
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 3.91010638117219e-05,
|
|
"loss": 0.4286,
|
|
"step": 668
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 3.9119035670618493e-05,
|
|
"loss": 0.3901,
|
|
"step": 670
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 3.913695396204699e-05,
|
|
"loss": 0.4341,
|
|
"step": 672
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 3.9154819004388286e-05,
|
|
"loss": 0.4188,
|
|
"step": 674
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 3.917263111319323e-05,
|
|
"loss": 0.4272,
|
|
"step": 676
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 3.919039060121604e-05,
|
|
"loss": 0.4415,
|
|
"step": 678
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 3.920809777844726e-05,
|
|
"loss": 0.4354,
|
|
"step": 680
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 3.9225752952146275e-05,
|
|
"loss": 0.4507,
|
|
"step": 682
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 3.924335642687319e-05,
|
|
"loss": 0.4286,
|
|
"step": 684
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 3.926090850452047e-05,
|
|
"loss": 0.4164,
|
|
"step": 686
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 3.927840948434395e-05,
|
|
"loss": 0.4136,
|
|
"step": 688
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 3.929585966299344e-05,
|
|
"loss": 0.4235,
|
|
"step": 690
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 3.931325933454294e-05,
|
|
"loss": 0.4381,
|
|
"step": 692
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 3.933060879052033e-05,
|
|
"loss": 0.4119,
|
|
"step": 694
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 3.9347908319936696e-05,
|
|
"loss": 0.3929,
|
|
"step": 696
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 3.93651582093152e-05,
|
|
"loss": 0.3846,
|
|
"step": 698
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 3.93823587427196e-05,
|
|
"loss": 0.4556,
|
|
"step": 700
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 3.939951020178229e-05,
|
|
"loss": 0.4286,
|
|
"step": 702
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 3.9416612865731936e-05,
|
|
"loss": 0.4532,
|
|
"step": 704
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 3.9433667011420814e-05,
|
|
"loss": 0.4554,
|
|
"step": 706
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 3.945067291335165e-05,
|
|
"loss": 0.4207,
|
|
"step": 708
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 3.946763084370416e-05,
|
|
"loss": 0.4486,
|
|
"step": 710
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 3.948454107236118e-05,
|
|
"loss": 0.406,
|
|
"step": 712
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 3.950140386693443e-05,
|
|
"loss": 0.4077,
|
|
"step": 714
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 3.9518219492789934e-05,
|
|
"loss": 0.39,
|
|
"step": 716
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 3.953498821307306e-05,
|
|
"loss": 0.3861,
|
|
"step": 718
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 3.955171028873329e-05,
|
|
"loss": 0.42,
|
|
"step": 720
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 3.956838597854848e-05,
|
|
"loss": 0.4241,
|
|
"step": 722
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 3.958501553914897e-05,
|
|
"loss": 0.423,
|
|
"step": 724
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 3.9601599225041266e-05,
|
|
"loss": 0.4125,
|
|
"step": 726
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 3.961813728863139e-05,
|
|
"loss": 0.4138,
|
|
"step": 728
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 3.963462998024795e-05,
|
|
"loss": 0.4459,
|
|
"step": 730
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 3.965107754816486e-05,
|
|
"loss": 0.4076,
|
|
"step": 732
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 3.9667480238623775e-05,
|
|
"loss": 0.3984,
|
|
"step": 734
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 3.968383829585623e-05,
|
|
"loss": 0.4379,
|
|
"step": 736
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 3.970015196210541e-05,
|
|
"loss": 0.4066,
|
|
"step": 738
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 3.97164214776477e-05,
|
|
"loss": 0.4365,
|
|
"step": 740
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 3.9732647080813965e-05,
|
|
"loss": 0.4431,
|
|
"step": 742
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 3.9748829008010405e-05,
|
|
"loss": 0.4331,
|
|
"step": 744
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 3.9764967493739306e-05,
|
|
"loss": 0.4296,
|
|
"step": 746
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 3.978106277061938e-05,
|
|
"loss": 0.4106,
|
|
"step": 748
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 3.9797115069405913e-05,
|
|
"loss": 0.439,
|
|
"step": 750
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 3.9813124619010586e-05,
|
|
"loss": 0.4238,
|
|
"step": 752
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 3.9829091646521095e-05,
|
|
"loss": 0.4362,
|
|
"step": 754
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 3.984501637722046e-05,
|
|
"loss": 0.4045,
|
|
"step": 756
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 3.9860899034606074e-05,
|
|
"loss": 0.4302,
|
|
"step": 758
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 3.987673984040859e-05,
|
|
"loss": 0.558,
|
|
"step": 760
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 3.9892539014610415e-05,
|
|
"loss": 0.3887,
|
|
"step": 762
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 3.9908296775464085e-05,
|
|
"loss": 0.4415,
|
|
"step": 764
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 3.992401333951037e-05,
|
|
"loss": 0.4647,
|
|
"step": 766
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 3.993968892159608e-05,
|
|
"loss": 0.4195,
|
|
"step": 768
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 3.995532373489172e-05,
|
|
"loss": 0.4479,
|
|
"step": 770
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 3.9970917990908887e-05,
|
|
"loss": 0.4335,
|
|
"step": 772
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 3.9986471899517414e-05,
|
|
"loss": 0.4152,
|
|
"step": 774
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 4.000198566896234e-05,
|
|
"loss": 0.434,
|
|
"step": 776
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 4.001745950588063e-05,
|
|
"loss": 0.4864,
|
|
"step": 778
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 4.0032893615317694e-05,
|
|
"loss": 0.443,
|
|
"step": 780
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 4.004828820074367e-05,
|
|
"loss": 0.4165,
|
|
"step": 782
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 4.0063643464069556e-05,
|
|
"loss": 0.3849,
|
|
"step": 784
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 4.007895960566306e-05,
|
|
"loss": 0.4286,
|
|
"step": 786
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 4.009423682436431e-05,
|
|
"loss": 0.3933,
|
|
"step": 788
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 4.0109475317501347e-05,
|
|
"loss": 0.3834,
|
|
"step": 790
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 4.01246752809054e-05,
|
|
"loss": 0.4431,
|
|
"step": 792
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.0139836908926014e-05,
|
|
"loss": 0.3909,
|
|
"step": 794
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.01549603944459e-05,
|
|
"loss": 0.4358,
|
|
"step": 796
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.017004592889575e-05,
|
|
"loss": 0.4235,
|
|
"step": 798
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.01850937022687e-05,
|
|
"loss": 0.4691,
|
|
"step": 800
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.020010390313471e-05,
|
|
"loss": 0.4348,
|
|
"step": 802
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.021507671865475e-05,
|
|
"loss": 0.4835,
|
|
"step": 804
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.0230012334594805e-05,
|
|
"loss": 0.3982,
|
|
"step": 806
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.02449109353397e-05,
|
|
"loss": 0.4139,
|
|
"step": 808
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.0259772703906755e-05,
|
|
"loss": 0.4662,
|
|
"step": 810
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.027459782195926e-05,
|
|
"loss": 0.426,
|
|
"step": 812
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.028938646981982e-05,
|
|
"loss": 0.4504,
|
|
"step": 814
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.030413882648352e-05,
|
|
"loss": 0.4559,
|
|
"step": 816
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.0318855069630894e-05,
|
|
"loss": 0.4498,
|
|
"step": 818
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.033353537564081e-05,
|
|
"loss": 0.4011,
|
|
"step": 820
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.03481799196031e-05,
|
|
"loss": 0.4006,
|
|
"step": 822
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.036278887533115e-05,
|
|
"loss": 0.4337,
|
|
"step": 824
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.037736241537422e-05,
|
|
"loss": 0.3834,
|
|
"step": 826
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.0391900711029694e-05,
|
|
"loss": 0.3872,
|
|
"step": 828
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.04064039323552e-05,
|
|
"loss": 0.4382,
|
|
"step": 830
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.042087224818048e-05,
|
|
"loss": 0.4095,
|
|
"step": 832
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.043530582611922e-05,
|
|
"loss": 0.3883,
|
|
"step": 834
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.04497048325807e-05,
|
|
"loss": 0.4379,
|
|
"step": 836
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.046406943278129e-05,
|
|
"loss": 0.4359,
|
|
"step": 838
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.0478399790755854e-05,
|
|
"loss": 0.4147,
|
|
"step": 840
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.0492696069368944e-05,
|
|
"loss": 0.3956,
|
|
"step": 842
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 4.0506958430325957e-05,
|
|
"loss": 0.4261,
|
|
"step": 844
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.0521187034184046e-05,
|
|
"loss": 0.4113,
|
|
"step": 846
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.0535382040363054e-05,
|
|
"loss": 0.4057,
|
|
"step": 848
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.054954360715614e-05,
|
|
"loss": 0.4058,
|
|
"step": 850
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.056367189174041e-05,
|
|
"loss": 0.4188,
|
|
"step": 852
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.0577767050187427e-05,
|
|
"loss": 0.4246,
|
|
"step": 854
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.0591829237473465e-05,
|
|
"loss": 0.4521,
|
|
"step": 856
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.060585860748981e-05,
|
|
"loss": 0.3981,
|
|
"step": 858
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.061985531305281e-05,
|
|
"loss": 0.4489,
|
|
"step": 860
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.063381950591393e-05,
|
|
"loss": 0.4677,
|
|
"step": 862
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.064775133676954e-05,
|
|
"loss": 0.3962,
|
|
"step": 864
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.066165095527074e-05,
|
|
"loss": 0.396,
|
|
"step": 866
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.067551851003297e-05,
|
|
"loss": 0.3868,
|
|
"step": 868
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.0689354148645566e-05,
|
|
"loss": 0.377,
|
|
"step": 870
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.070315801768115e-05,
|
|
"loss": 0.4171,
|
|
"step": 872
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.0716930262705e-05,
|
|
"loss": 0.4397,
|
|
"step": 874
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.07306710282842e-05,
|
|
"loss": 0.4056,
|
|
"step": 876
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.0744380457996814e-05,
|
|
"loss": 0.4335,
|
|
"step": 878
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.0758058694440806e-05,
|
|
"loss": 0.3782,
|
|
"step": 880
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.077170587924302e-05,
|
|
"loss": 0.4139,
|
|
"step": 882
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.078532215306792e-05,
|
|
"loss": 0.3988,
|
|
"step": 884
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.079890765562631e-05,
|
|
"loss": 0.412,
|
|
"step": 886
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.081246252568396e-05,
|
|
"loss": 0.434,
|
|
"step": 888
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.082598690107005e-05,
|
|
"loss": 0.4021,
|
|
"step": 890
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.0839480918685646e-05,
|
|
"loss": 0.4447,
|
|
"step": 892
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 4.085294471451196e-05,
|
|
"loss": 0.399,
|
|
"step": 894
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.086637842361864e-05,
|
|
"loss": 0.3915,
|
|
"step": 896
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.087978218017181e-05,
|
|
"loss": 0.4229,
|
|
"step": 898
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.089315611744216e-05,
|
|
"loss": 0.396,
|
|
"step": 900
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.090650036781293e-05,
|
|
"loss": 0.3907,
|
|
"step": 902
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.091981506278769e-05,
|
|
"loss": 0.4402,
|
|
"step": 904
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.0933100332998206e-05,
|
|
"loss": 0.4026,
|
|
"step": 906
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.094635630821205e-05,
|
|
"loss": 0.4257,
|
|
"step": 908
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.095958311734026e-05,
|
|
"loss": 0.4132,
|
|
"step": 910
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.0972780888444836e-05,
|
|
"loss": 0.4021,
|
|
"step": 912
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.098594974874619e-05,
|
|
"loss": 0.3859,
|
|
"step": 914
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.09990898246305e-05,
|
|
"loss": 0.3994,
|
|
"step": 916
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.1012201241656985e-05,
|
|
"loss": 0.4463,
|
|
"step": 918
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.1025284124565104e-05,
|
|
"loss": 0.4418,
|
|
"step": 920
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.103833859728169e-05,
|
|
"loss": 0.4191,
|
|
"step": 922
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.105136478292797e-05,
|
|
"loss": 0.4027,
|
|
"step": 924
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.1064362803826564e-05,
|
|
"loss": 0.431,
|
|
"step": 926
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.107733278150835e-05,
|
|
"loss": 0.4086,
|
|
"step": 928
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.1090274836719275e-05,
|
|
"loss": 0.4249,
|
|
"step": 930
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.1103189089427136e-05,
|
|
"loss": 0.4215,
|
|
"step": 932
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.1116075658828226e-05,
|
|
"loss": 0.4227,
|
|
"step": 934
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.112893466335394e-05,
|
|
"loss": 0.4083,
|
|
"step": 936
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.114176622067731e-05,
|
|
"loss": 0.4034,
|
|
"step": 938
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.1154570447719456e-05,
|
|
"loss": 0.443,
|
|
"step": 940
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.116734746065598e-05,
|
|
"loss": 0.4171,
|
|
"step": 942
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.1180097374923306e-05,
|
|
"loss": 0.424,
|
|
"step": 944
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 4.1192820305224925e-05,
|
|
"loss": 0.4236,
|
|
"step": 946
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.12055163655376e-05,
|
|
"loss": 0.4448,
|
|
"step": 948
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.1218185669117455e-05,
|
|
"loss": 0.4174,
|
|
"step": 950
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.1230828328506084e-05,
|
|
"loss": 0.4186,
|
|
"step": 952
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.124344445553652e-05,
|
|
"loss": 0.4058,
|
|
"step": 954
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.1256034161339176e-05,
|
|
"loss": 0.3964,
|
|
"step": 956
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.126859755634774e-05,
|
|
"loss": 0.4072,
|
|
"step": 958
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.128113475030494e-05,
|
|
"loss": 0.38,
|
|
"step": 960
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.129364585226836e-05,
|
|
"loss": 0.3773,
|
|
"step": 962
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.130613097061606e-05,
|
|
"loss": 0.4274,
|
|
"step": 964
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.131859021305226e-05,
|
|
"loss": 0.4777,
|
|
"step": 966
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.133102368661292e-05,
|
|
"loss": 0.4189,
|
|
"step": 968
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.1343431497671206e-05,
|
|
"loss": 0.4009,
|
|
"step": 970
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.1355813751943006e-05,
|
|
"loss": 0.3787,
|
|
"step": 972
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.13681705544923e-05,
|
|
"loss": 0.4092,
|
|
"step": 974
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.138050200973651e-05,
|
|
"loss": 0.4042,
|
|
"step": 976
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.139280822145184e-05,
|
|
"loss": 0.4091,
|
|
"step": 978
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.140508929277842e-05,
|
|
"loss": 0.4119,
|
|
"step": 980
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.1417345326225616e-05,
|
|
"loss": 0.4028,
|
|
"step": 982
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.142957642367706e-05,
|
|
"loss": 0.4015,
|
|
"step": 984
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.144178268639579e-05,
|
|
"loss": 0.4297,
|
|
"step": 986
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.145396421502924e-05,
|
|
"loss": 0.4955,
|
|
"step": 988
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.146612110961427e-05,
|
|
"loss": 0.4266,
|
|
"step": 990
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.147825346958206e-05,
|
|
"loss": 0.3773,
|
|
"step": 992
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.149036139376298e-05,
|
|
"loss": 0.4366,
|
|
"step": 994
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 4.150244498039145e-05,
|
|
"loss": 0.459,
|
|
"step": 996
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.151450432711072e-05,
|
|
"loss": 0.4298,
|
|
"step": 998
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.152653953097756e-05,
|
|
"loss": 0.4095,
|
|
"step": 1000
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.153855068846702e-05,
|
|
"loss": 0.4166,
|
|
"step": 1002
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.155053789547701e-05,
|
|
"loss": 0.3569,
|
|
"step": 1004
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.15625012473329e-05,
|
|
"loss": 0.4314,
|
|
"step": 1006
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.157444083879211e-05,
|
|
"loss": 0.4341,
|
|
"step": 1008
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.158635676404857e-05,
|
|
"loss": 0.4484,
|
|
"step": 1010
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.159824911673721e-05,
|
|
"loss": 0.3762,
|
|
"step": 1012
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.161011798993835e-05,
|
|
"loss": 0.3983,
|
|
"step": 1014
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.162196347618206e-05,
|
|
"loss": 0.3804,
|
|
"step": 1016
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.163378566745255e-05,
|
|
"loss": 0.3868,
|
|
"step": 1018
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.164558465519238e-05,
|
|
"loss": 0.403,
|
|
"step": 1020
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.1657360530306765e-05,
|
|
"loss": 0.4648,
|
|
"step": 1022
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.1669113383167727e-05,
|
|
"loss": 0.4285,
|
|
"step": 1024
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.16808433036183e-05,
|
|
"loss": 0.4155,
|
|
"step": 1026
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.169255038097662e-05,
|
|
"loss": 0.3823,
|
|
"step": 1028
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.170423470404002e-05,
|
|
"loss": 0.4236,
|
|
"step": 1030
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.171589636108907e-05,
|
|
"loss": 0.4016,
|
|
"step": 1032
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.172753543989157e-05,
|
|
"loss": 0.4258,
|
|
"step": 1034
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.1739152027706526e-05,
|
|
"loss": 0.4235,
|
|
"step": 1036
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.1750746211288065e-05,
|
|
"loss": 0.416,
|
|
"step": 1038
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.176231807688935e-05,
|
|
"loss": 0.4463,
|
|
"step": 1040
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.1773867710266366e-05,
|
|
"loss": 0.3806,
|
|
"step": 1042
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.178539519668181e-05,
|
|
"loss": 0.4489,
|
|
"step": 1044
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.179690062090882e-05,
|
|
"loss": 0.4072,
|
|
"step": 1046
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 4.180838406723471e-05,
|
|
"loss": 0.3915,
|
|
"step": 1048
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.1819845619464724e-05,
|
|
"loss": 0.4,
|
|
"step": 1050
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.183128536092566e-05,
|
|
"loss": 0.4016,
|
|
"step": 1052
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.18427033744695e-05,
|
|
"loss": 0.4092,
|
|
"step": 1054
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.185409974247706e-05,
|
|
"loss": 0.4283,
|
|
"step": 1056
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.186547454686151e-05,
|
|
"loss": 0.4972,
|
|
"step": 1058
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.1876827869071924e-05,
|
|
"loss": 0.4158,
|
|
"step": 1060
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.188815979009677e-05,
|
|
"loss": 0.4085,
|
|
"step": 1062
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.18994703904674e-05,
|
|
"loss": 0.405,
|
|
"step": 1064
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.1910759750261466e-05,
|
|
"loss": 0.4675,
|
|
"step": 1066
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.19220279491063e-05,
|
|
"loss": 0.4348,
|
|
"step": 1068
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.193327506618233e-05,
|
|
"loss": 0.4107,
|
|
"step": 1070
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.19445011802264e-05,
|
|
"loss": 0.4001,
|
|
"step": 1072
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.1955706369535056e-05,
|
|
"loss": 0.4198,
|
|
"step": 1074
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.196689071196783e-05,
|
|
"loss": 0.4285,
|
|
"step": 1076
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.1978054284950535e-05,
|
|
"loss": 0.4176,
|
|
"step": 1078
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.198919716547841e-05,
|
|
"loss": 0.3881,
|
|
"step": 1080
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.200031943011934e-05,
|
|
"loss": 0.3973,
|
|
"step": 1082
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.201142115501703e-05,
|
|
"loss": 0.4103,
|
|
"step": 1084
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.202250241589409e-05,
|
|
"loss": 0.4142,
|
|
"step": 1086
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.2033563288055166e-05,
|
|
"loss": 0.4537,
|
|
"step": 1088
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.204460384639001e-05,
|
|
"loss": 0.4097,
|
|
"step": 1090
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.205562416537651e-05,
|
|
"loss": 0.4053,
|
|
"step": 1092
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.2066624319083695e-05,
|
|
"loss": 0.3938,
|
|
"step": 1094
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.207760438117476e-05,
|
|
"loss": 0.4059,
|
|
"step": 1096
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.208856442490998e-05,
|
|
"loss": 0.4105,
|
|
"step": 1098
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 4.2099504523149676e-05,
|
|
"loss": 0.4215,
|
|
"step": 1100
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.21104247483571e-05,
|
|
"loss": 0.4072,
|
|
"step": 1102
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.212132517260135e-05,
|
|
"loss": 0.4169,
|
|
"step": 1104
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.2132205867560164e-05,
|
|
"loss": 0.39,
|
|
"step": 1106
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.21430669045228e-05,
|
|
"loss": 0.4276,
|
|
"step": 1108
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.2153908354392824e-05,
|
|
"loss": 0.4126,
|
|
"step": 1110
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.216473028769087e-05,
|
|
"loss": 0.3785,
|
|
"step": 1112
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.217553277455743e-05,
|
|
"loss": 0.4351,
|
|
"step": 1114
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.218631588475552e-05,
|
|
"loss": 0.4343,
|
|
"step": 1116
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.219707968767347e-05,
|
|
"loss": 0.4218,
|
|
"step": 1118
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.220782425232751e-05,
|
|
"loss": 0.3915,
|
|
"step": 1120
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.2218549647364496e-05,
|
|
"loss": 0.4481,
|
|
"step": 1122
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.222925594106451e-05,
|
|
"loss": 0.4034,
|
|
"step": 1124
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.2239943201343465e-05,
|
|
"loss": 0.4274,
|
|
"step": 1126
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.225061149575571e-05,
|
|
"loss": 0.3944,
|
|
"step": 1128
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.226126089149656e-05,
|
|
"loss": 0.4011,
|
|
"step": 1130
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.2271891455404885e-05,
|
|
"loss": 0.4637,
|
|
"step": 1132
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.228250325396557e-05,
|
|
"loss": 0.3798,
|
|
"step": 1134
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.2293096353312065e-05,
|
|
"loss": 0.4163,
|
|
"step": 1136
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.230367081922881e-05,
|
|
"loss": 0.4191,
|
|
"step": 1138
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.231422671715371e-05,
|
|
"loss": 0.4288,
|
|
"step": 1140
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.232476411218056e-05,
|
|
"loss": 0.402,
|
|
"step": 1142
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.233528306906146e-05,
|
|
"loss": 0.4143,
|
|
"step": 1144
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.234578365220921e-05,
|
|
"loss": 0.392,
|
|
"step": 1146
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.2356265925699625e-05,
|
|
"loss": 0.4192,
|
|
"step": 1148
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 4.236672995327397e-05,
|
|
"loss": 0.4271,
|
|
"step": 1150
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.23771757983412e-05,
|
|
"loss": 0.4475,
|
|
"step": 1152
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.238760352398031e-05,
|
|
"loss": 0.4031,
|
|
"step": 1154
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.239801319294261e-05,
|
|
"loss": 0.4504,
|
|
"step": 1156
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.2408404867654e-05,
|
|
"loss": 0.4072,
|
|
"step": 1158
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.2418778610217214e-05,
|
|
"loss": 0.4197,
|
|
"step": 1160
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.242913448241402e-05,
|
|
"loss": 0.4078,
|
|
"step": 1162
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.243947254570746e-05,
|
|
"loss": 0.3883,
|
|
"step": 1164
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.244979286124403e-05,
|
|
"loss": 0.3939,
|
|
"step": 1166
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.2460095489855853e-05,
|
|
"loss": 0.4188,
|
|
"step": 1168
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.2470380492062816e-05,
|
|
"loss": 0.3838,
|
|
"step": 1170
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.2480647928074705e-05,
|
|
"loss": 0.4363,
|
|
"step": 1172
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.249089785779335e-05,
|
|
"loss": 0.4296,
|
|
"step": 1174
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.250113034081468e-05,
|
|
"loss": 0.4247,
|
|
"step": 1176
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.251134543643082e-05,
|
|
"loss": 0.426,
|
|
"step": 1178
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.252154320363217e-05,
|
|
"loss": 0.3717,
|
|
"step": 1180
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.253172370110943e-05,
|
|
"loss": 0.3987,
|
|
"step": 1182
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.254188698725561e-05,
|
|
"loss": 0.4263,
|
|
"step": 1184
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.2552033120168075e-05,
|
|
"loss": 0.3967,
|
|
"step": 1186
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.2562162157650524e-05,
|
|
"loss": 0.426,
|
|
"step": 1188
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.257227415721495e-05,
|
|
"loss": 0.434,
|
|
"step": 1190
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.2582369176083624e-05,
|
|
"loss": 0.3923,
|
|
"step": 1192
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.259244727119102e-05,
|
|
"loss": 0.4758,
|
|
"step": 1194
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.260250849918575e-05,
|
|
"loss": 0.3831,
|
|
"step": 1196
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.2612552916432486e-05,
|
|
"loss": 0.4099,
|
|
"step": 1198
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.262258057901381e-05,
|
|
"loss": 0.3856,
|
|
"step": 1200
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 4.2632591542732164e-05,
|
|
"loss": 0.4101,
|
|
"step": 1202
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.2642585863111635e-05,
|
|
"loss": 0.4311,
|
|
"step": 1204
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.2652563595399866e-05,
|
|
"loss": 0.4353,
|
|
"step": 1206
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.266252479456986e-05,
|
|
"loss": 0.4064,
|
|
"step": 1208
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.267246951532179e-05,
|
|
"loss": 0.4004,
|
|
"step": 1210
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.2682397812084823e-05,
|
|
"loss": 0.4524,
|
|
"step": 1212
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.26923097390189e-05,
|
|
"loss": 0.4236,
|
|
"step": 1214
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.27022053500165e-05,
|
|
"loss": 0.4152,
|
|
"step": 1216
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.271208469870438e-05,
|
|
"loss": 0.4011,
|
|
"step": 1218
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.2721947838445385e-05,
|
|
"loss": 0.4237,
|
|
"step": 1220
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.273179482234011e-05,
|
|
"loss": 0.414,
|
|
"step": 1222
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.274162570322864e-05,
|
|
"loss": 0.4059,
|
|
"step": 1224
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.275144053369225e-05,
|
|
"loss": 0.4322,
|
|
"step": 1226
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.2761239366055114e-05,
|
|
"loss": 0.4512,
|
|
"step": 1228
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.277102225238593e-05,
|
|
"loss": 0.3945,
|
|
"step": 1230
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.278078924449962e-05,
|
|
"loss": 0.4418,
|
|
"step": 1232
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.279054039395898e-05,
|
|
"loss": 0.3977,
|
|
"step": 1234
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.280027575207627e-05,
|
|
"loss": 0.3976,
|
|
"step": 1236
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.2809995369914885e-05,
|
|
"loss": 0.4142,
|
|
"step": 1238
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.281969929829093e-05,
|
|
"loss": 0.4042,
|
|
"step": 1240
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.2829387587774815e-05,
|
|
"loss": 0.4376,
|
|
"step": 1242
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.2839060288692876e-05,
|
|
"loss": 0.4404,
|
|
"step": 1244
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.284871745112887e-05,
|
|
"loss": 0.4039,
|
|
"step": 1246
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.28583591249256e-05,
|
|
"loss": 0.4156,
|
|
"step": 1248
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.286798535968643e-05,
|
|
"loss": 0.4456,
|
|
"step": 1250
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 4.2877596204776815e-05,
|
|
"loss": 0.3924,
|
|
"step": 1252
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.288719170932582e-05,
|
|
"loss": 0.38,
|
|
"step": 1254
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.289677192222763e-05,
|
|
"loss": 0.3981,
|
|
"step": 1256
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.290633689214305e-05,
|
|
"loss": 0.3901,
|
|
"step": 1258
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.2915886667500975e-05,
|
|
"loss": 0.4321,
|
|
"step": 1260
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.2925421296499877e-05,
|
|
"loss": 0.3576,
|
|
"step": 1262
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.293494082710925e-05,
|
|
"loss": 0.4222,
|
|
"step": 1264
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.294444530707107e-05,
|
|
"loss": 0.4119,
|
|
"step": 1266
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.2953934783901225e-05,
|
|
"loss": 0.3854,
|
|
"step": 1268
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.296340930489093e-05,
|
|
"loss": 0.4685,
|
|
"step": 1270
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.2972868917108175e-05,
|
|
"loss": 0.3788,
|
|
"step": 1272
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.298231366739908e-05,
|
|
"loss": 0.4121,
|
|
"step": 1274
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.299174360238933e-05,
|
|
"loss": 0.429,
|
|
"step": 1276
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.300115876848553e-05,
|
|
"loss": 0.4035,
|
|
"step": 1278
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.30105592118766e-05,
|
|
"loss": 0.4093,
|
|
"step": 1280
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.3019944978535095e-05,
|
|
"loss": 0.4355,
|
|
"step": 1282
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.302931611421859e-05,
|
|
"loss": 0.4436,
|
|
"step": 1284
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.3038672664470996e-05,
|
|
"loss": 0.3966,
|
|
"step": 1286
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.304801467462392e-05,
|
|
"loss": 0.4382,
|
|
"step": 1288
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.305734218979793e-05,
|
|
"loss": 0.3781,
|
|
"step": 1290
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.306665525490393e-05,
|
|
"loss": 0.4337,
|
|
"step": 1292
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.307595391464439e-05,
|
|
"loss": 0.3724,
|
|
"step": 1294
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.308523821351466e-05,
|
|
"loss": 0.429,
|
|
"step": 1296
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.3094508195804284e-05,
|
|
"loss": 0.4266,
|
|
"step": 1298
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.310376390559821e-05,
|
|
"loss": 0.3781,
|
|
"step": 1300
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.3113005386778086e-05,
|
|
"loss": 0.4023,
|
|
"step": 1302
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 4.312223268302349e-05,
|
|
"loss": 0.4028,
|
|
"step": 1304
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.3131445837813165e-05,
|
|
"loss": 0.3757,
|
|
"step": 1306
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.314064489442627e-05,
|
|
"loss": 0.4487,
|
|
"step": 1308
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.3149829895943575e-05,
|
|
"loss": 0.4473,
|
|
"step": 1310
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.3159000885248714e-05,
|
|
"loss": 0.4364,
|
|
"step": 1312
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.3168157905029314e-05,
|
|
"loss": 0.417,
|
|
"step": 1314
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.317730099777827e-05,
|
|
"loss": 0.4939,
|
|
"step": 1316
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.3186430205794846e-05,
|
|
"loss": 0.4143,
|
|
"step": 1318
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.319554557118593e-05,
|
|
"loss": 0.4498,
|
|
"step": 1320
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.320464713586712e-05,
|
|
"loss": 0.4373,
|
|
"step": 1322
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.3213734941563975e-05,
|
|
"loss": 0.4366,
|
|
"step": 1324
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.322280902981304e-05,
|
|
"loss": 0.4152,
|
|
"step": 1326
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.323186944196311e-05,
|
|
"loss": 0.439,
|
|
"step": 1328
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.324091621917627e-05,
|
|
"loss": 0.4048,
|
|
"step": 1330
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.324994940242908e-05,
|
|
"loss": 0.4099,
|
|
"step": 1332
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.325896903251362e-05,
|
|
"loss": 0.4357,
|
|
"step": 1334
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.326797515003868e-05,
|
|
"loss": 0.4281,
|
|
"step": 1336
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.327696779543076e-05,
|
|
"loss": 0.4325,
|
|
"step": 1338
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.328594700893527e-05,
|
|
"loss": 0.4338,
|
|
"step": 1340
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.3294912830617493e-05,
|
|
"loss": 0.3955,
|
|
"step": 1342
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.330386530036376e-05,
|
|
"loss": 0.3828,
|
|
"step": 1344
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.331280445788245e-05,
|
|
"loss": 0.4191,
|
|
"step": 1346
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.3321730342705056e-05,
|
|
"loss": 0.4371,
|
|
"step": 1348
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.333064299418728e-05,
|
|
"loss": 0.4177,
|
|
"step": 1350
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.333954245151e-05,
|
|
"loss": 0.4214,
|
|
"step": 1352
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.334842875368037e-05,
|
|
"loss": 0.4283,
|
|
"step": 1354
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 4.335730193953281e-05,
|
|
"loss": 0.3759,
|
|
"step": 1356
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.336616204773003e-05,
|
|
"loss": 0.4375,
|
|
"step": 1358
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.3375009116764036e-05,
|
|
"loss": 0.3871,
|
|
"step": 1360
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.3383843184957166e-05,
|
|
"loss": 0.3933,
|
|
"step": 1362
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.3392664290463044e-05,
|
|
"loss": 0.4353,
|
|
"step": 1364
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.3401472471267574e-05,
|
|
"loss": 0.4012,
|
|
"step": 1366
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.341026776518996e-05,
|
|
"loss": 0.4283,
|
|
"step": 1368
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.341905020988363e-05,
|
|
"loss": 0.432,
|
|
"step": 1370
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.342781984283724e-05,
|
|
"loss": 0.4313,
|
|
"step": 1372
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.343657670137562e-05,
|
|
"loss": 0.4001,
|
|
"step": 1374
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.344532082266072e-05,
|
|
"loss": 0.4315,
|
|
"step": 1376
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.3454052243692575e-05,
|
|
"loss": 0.3938,
|
|
"step": 1378
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.346277100131022e-05,
|
|
"loss": 0.4342,
|
|
"step": 1380
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.3471477132192653e-05,
|
|
"loss": 0.4199,
|
|
"step": 1382
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.348017067285972e-05,
|
|
"loss": 0.4801,
|
|
"step": 1384
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.348885165967309e-05,
|
|
"loss": 0.4286,
|
|
"step": 1386
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.349752012883711e-05,
|
|
"loss": 0.4463,
|
|
"step": 1388
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.350617611639974e-05,
|
|
"loss": 0.4041,
|
|
"step": 1390
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.351481965825347e-05,
|
|
"loss": 0.3823,
|
|
"step": 1392
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.352345079013615e-05,
|
|
"loss": 0.4101,
|
|
"step": 1394
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.353206954763198e-05,
|
|
"loss": 0.4202,
|
|
"step": 1396
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.354067596617226e-05,
|
|
"loss": 0.4435,
|
|
"step": 1398
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.354927008103638e-05,
|
|
"loss": 0.4121,
|
|
"step": 1400
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.355785192735263e-05,
|
|
"loss": 0.4137,
|
|
"step": 1402
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.356642154009906e-05,
|
|
"loss": 0.3904,
|
|
"step": 1404
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 4.357497895410437e-05,
|
|
"loss": 0.4299,
|
|
"step": 1406
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.358352420404871e-05,
|
|
"loss": 0.3948,
|
|
"step": 1408
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.359205732446458e-05,
|
|
"loss": 0.4066,
|
|
"step": 1410
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.3600578349737583e-05,
|
|
"loss": 0.4021,
|
|
"step": 1412
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.360908731410739e-05,
|
|
"loss": 0.4307,
|
|
"step": 1414
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.361758425166843e-05,
|
|
"loss": 0.4443,
|
|
"step": 1416
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.3626069196370766e-05,
|
|
"loss": 0.3856,
|
|
"step": 1418
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.3634542182020935e-05,
|
|
"loss": 0.3988,
|
|
"step": 1420
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.364300324228272e-05,
|
|
"loss": 0.3776,
|
|
"step": 1422
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.365145241067796e-05,
|
|
"loss": 0.3995,
|
|
"step": 1424
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.3659889720587335e-05,
|
|
"loss": 0.4141,
|
|
"step": 1426
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.36683152052512e-05,
|
|
"loss": 0.4506,
|
|
"step": 1428
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.367672889777033e-05,
|
|
"loss": 0.4084,
|
|
"step": 1430
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.3685130831106703e-05,
|
|
"loss": 0.4236,
|
|
"step": 1432
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.36935210380843e-05,
|
|
"loss": 0.3917,
|
|
"step": 1434
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.370189955138984e-05,
|
|
"loss": 0.3972,
|
|
"step": 1436
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.371026640357359e-05,
|
|
"loss": 0.374,
|
|
"step": 1438
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.3718621627050064e-05,
|
|
"loss": 0.4183,
|
|
"step": 1440
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.372696525409884e-05,
|
|
"loss": 0.4146,
|
|
"step": 1442
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.3735297316865254e-05,
|
|
"loss": 0.4316,
|
|
"step": 1444
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.374361784736118e-05,
|
|
"loss": 0.3862,
|
|
"step": 1446
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.3751926877465745e-05,
|
|
"loss": 0.4226,
|
|
"step": 1448
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.3760224438926084e-05,
|
|
"loss": 0.3723,
|
|
"step": 1450
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.376851056335804e-05,
|
|
"loss": 0.4093,
|
|
"step": 1452
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.377678528224691e-05,
|
|
"loss": 0.4116,
|
|
"step": 1454
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.3785048626948166e-05,
|
|
"loss": 0.4136,
|
|
"step": 1456
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 4.379330062868813e-05,
|
|
"loss": 0.3824,
|
|
"step": 1458
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.3801541318564724e-05,
|
|
"loss": 0.4008,
|
|
"step": 1460
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.380977072754816e-05,
|
|
"loss": 0.4091,
|
|
"step": 1462
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.381798888648163e-05,
|
|
"loss": 0.3978,
|
|
"step": 1464
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.382619582608202e-05,
|
|
"loss": 0.3772,
|
|
"step": 1466
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.383439157694055e-05,
|
|
"loss": 0.4243,
|
|
"step": 1468
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.384257616952354e-05,
|
|
"loss": 0.3477,
|
|
"step": 1470
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.3850749634173e-05,
|
|
"loss": 0.438,
|
|
"step": 1472
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.3858912001107384e-05,
|
|
"loss": 0.4043,
|
|
"step": 1474
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.386706330042218e-05,
|
|
"loss": 0.4298,
|
|
"step": 1476
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.387520356209066e-05,
|
|
"loss": 0.4314,
|
|
"step": 1478
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.388333281596448e-05,
|
|
"loss": 0.4114,
|
|
"step": 1480
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.389145109177436e-05,
|
|
"loss": 0.4027,
|
|
"step": 1482
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.3899558419130734e-05,
|
|
"loss": 0.4181,
|
|
"step": 1484
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.390765482752441e-05,
|
|
"loss": 0.4045,
|
|
"step": 1486
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.391574034632718e-05,
|
|
"loss": 0.4233,
|
|
"step": 1488
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.3923815004792494e-05,
|
|
"loss": 0.4053,
|
|
"step": 1490
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.3931878832056075e-05,
|
|
"loss": 0.3867,
|
|
"step": 1492
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.3939931857136574e-05,
|
|
"loss": 0.4557,
|
|
"step": 1494
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.394797410893615e-05,
|
|
"loss": 0.3776,
|
|
"step": 1496
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.395600561624115e-05,
|
|
"loss": 0.429,
|
|
"step": 1498
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.396402640772268e-05,
|
|
"loss": 0.3827,
|
|
"step": 1500
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.397203651193725e-05,
|
|
"loss": 0.3911,
|
|
"step": 1502
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.398003595732736e-05,
|
|
"loss": 0.4183,
|
|
"step": 1504
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.398802477222212e-05,
|
|
"loss": 0.4893,
|
|
"step": 1506
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 4.399600298483787e-05,
|
|
"loss": 0.4136,
|
|
"step": 1508
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.400397062327873e-05,
|
|
"loss": 0.3963,
|
|
"step": 1510
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.401192771553723e-05,
|
|
"loss": 0.4279,
|
|
"step": 1512
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.40198742894949e-05,
|
|
"loss": 0.3694,
|
|
"step": 1514
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.402781037292285e-05,
|
|
"loss": 0.4332,
|
|
"step": 1516
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.403573599348233e-05,
|
|
"loss": 0.4169,
|
|
"step": 1518
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.404365117872536e-05,
|
|
"loss": 0.3999,
|
|
"step": 1520
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.405155595609524e-05,
|
|
"loss": 0.3579,
|
|
"step": 1522
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.4059450352927184e-05,
|
|
"loss": 0.4282,
|
|
"step": 1524
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.4067334396448835e-05,
|
|
"loss": 0.41,
|
|
"step": 1526
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.407520811378086e-05,
|
|
"loss": 0.4163,
|
|
"step": 1528
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.40830715319375e-05,
|
|
"loss": 0.4021,
|
|
"step": 1530
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.409092467782714e-05,
|
|
"loss": 0.4321,
|
|
"step": 1532
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.409876757825283e-05,
|
|
"loss": 0.4132,
|
|
"step": 1534
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.410660025991285e-05,
|
|
"loss": 0.4219,
|
|
"step": 1536
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.411442274940127e-05,
|
|
"loss": 0.3956,
|
|
"step": 1538
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.412223507320849e-05,
|
|
"loss": 0.42,
|
|
"step": 1540
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.413003725772174e-05,
|
|
"loss": 0.3729,
|
|
"step": 1542
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.413782932922566e-05,
|
|
"loss": 0.4064,
|
|
"step": 1544
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.4145611313902804e-05,
|
|
"loss": 0.4148,
|
|
"step": 1546
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.415338323783419e-05,
|
|
"loss": 0.4608,
|
|
"step": 1548
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.41611451269998e-05,
|
|
"loss": 0.4198,
|
|
"step": 1550
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.416889700727911e-05,
|
|
"loss": 0.3994,
|
|
"step": 1552
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.417663890445164e-05,
|
|
"loss": 0.4298,
|
|
"step": 1554
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.41843708441974e-05,
|
|
"loss": 0.4102,
|
|
"step": 1556
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.419209285209748e-05,
|
|
"loss": 0.375,
|
|
"step": 1558
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 4.4199804953634464e-05,
|
|
"loss": 0.4442,
|
|
"step": 1560
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.420750717419305e-05,
|
|
"loss": 0.3857,
|
|
"step": 1562
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.421519953906045e-05,
|
|
"loss": 0.4135,
|
|
"step": 1564
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.422288207342693e-05,
|
|
"loss": 0.4276,
|
|
"step": 1566
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.4230554802386326e-05,
|
|
"loss": 0.4305,
|
|
"step": 1568
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.42382177509365e-05,
|
|
"loss": 0.4143,
|
|
"step": 1570
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.424587094397983e-05,
|
|
"loss": 0.4009,
|
|
"step": 1572
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.425351440632373e-05,
|
|
"loss": 0.3962,
|
|
"step": 1574
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.426114816268109e-05,
|
|
"loss": 0.3993,
|
|
"step": 1576
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.426877223767077e-05,
|
|
"loss": 0.4022,
|
|
"step": 1578
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.427638665581812e-05,
|
|
"loss": 0.4137,
|
|
"step": 1580
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.4283991441555375e-05,
|
|
"loss": 0.3969,
|
|
"step": 1582
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.429158661922218e-05,
|
|
"loss": 0.3967,
|
|
"step": 1584
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.4299172213066036e-05,
|
|
"loss": 0.438,
|
|
"step": 1586
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.4306748247242776e-05,
|
|
"loss": 0.3956,
|
|
"step": 1588
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.4314314745817045e-05,
|
|
"loss": 0.4115,
|
|
"step": 1590
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.4321871732762674e-05,
|
|
"loss": 0.4054,
|
|
"step": 1592
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.432941923196326e-05,
|
|
"loss": 0.4237,
|
|
"step": 1594
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.433695726721252e-05,
|
|
"loss": 0.4312,
|
|
"step": 1596
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.4344485862214794e-05,
|
|
"loss": 0.4115,
|
|
"step": 1598
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.435200504058547e-05,
|
|
"loss": 0.4012,
|
|
"step": 1600
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.435951482585142e-05,
|
|
"loss": 0.4167,
|
|
"step": 1602
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.4367015241451476e-05,
|
|
"loss": 0.4283,
|
|
"step": 1604
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.437450631073684e-05,
|
|
"loss": 0.4393,
|
|
"step": 1606
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.438198805697152e-05,
|
|
"loss": 0.3964,
|
|
"step": 1608
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 4.4389460503332784e-05,
|
|
"loss": 0.4082,
|
|
"step": 1610
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.439692367291158e-05,
|
|
"loss": 0.4078,
|
|
"step": 1612
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.440437758871295e-05,
|
|
"loss": 0.4415,
|
|
"step": 1614
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.441182227365648e-05,
|
|
"loss": 0.3941,
|
|
"step": 1616
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.44192577505767e-05,
|
|
"loss": 0.3972,
|
|
"step": 1618
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.4426684042223524e-05,
|
|
"loss": 0.4083,
|
|
"step": 1620
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.4434101171262674e-05,
|
|
"loss": 0.4113,
|
|
"step": 1622
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.444150916027604e-05,
|
|
"loss": 0.4121,
|
|
"step": 1624
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.4448908031762146e-05,
|
|
"loss": 0.4357,
|
|
"step": 1626
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.445629780813659e-05,
|
|
"loss": 0.425,
|
|
"step": 1628
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.446367851173236e-05,
|
|
"loss": 0.4089,
|
|
"step": 1630
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.447105016480029e-05,
|
|
"loss": 0.4319,
|
|
"step": 1632
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.447841278950948e-05,
|
|
"loss": 0.4266,
|
|
"step": 1634
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.448576640794766e-05,
|
|
"loss": 0.4315,
|
|
"step": 1636
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.449311104212163e-05,
|
|
"loss": 0.4261,
|
|
"step": 1638
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.4500446713957584e-05,
|
|
"loss": 0.4238,
|
|
"step": 1640
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.450777344530158e-05,
|
|
"loss": 0.3994,
|
|
"step": 1642
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.451509125791988e-05,
|
|
"loss": 0.4143,
|
|
"step": 1644
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.452240017349937e-05,
|
|
"loss": 0.4132,
|
|
"step": 1646
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.452970021364793e-05,
|
|
"loss": 0.3768,
|
|
"step": 1648
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.45369913998948e-05,
|
|
"loss": 0.3995,
|
|
"step": 1650
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.454427375369099e-05,
|
|
"loss": 0.4532,
|
|
"step": 1652
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.455154729640966e-05,
|
|
"loss": 0.3864,
|
|
"step": 1654
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.455881204934647e-05,
|
|
"loss": 0.3718,
|
|
"step": 1656
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.4566068033719974e-05,
|
|
"loss": 0.4261,
|
|
"step": 1658
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.457331527067197e-05,
|
|
"loss": 0.3778,
|
|
"step": 1660
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 4.458055378126792e-05,
|
|
"loss": 0.4224,
|
|
"step": 1662
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.4587783586497255e-05,
|
|
"loss": 0.4012,
|
|
"step": 1664
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.4595004707273765e-05,
|
|
"loss": 0.391,
|
|
"step": 1666
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.4602217164435995e-05,
|
|
"loss": 0.468,
|
|
"step": 1668
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.460942097874754e-05,
|
|
"loss": 0.4088,
|
|
"step": 1670
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.4616616170897475e-05,
|
|
"loss": 0.4303,
|
|
"step": 1672
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.462380276150064e-05,
|
|
"loss": 0.4058,
|
|
"step": 1674
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.463098077109807e-05,
|
|
"loss": 0.3846,
|
|
"step": 1676
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.463815022015727e-05,
|
|
"loss": 0.433,
|
|
"step": 1678
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.464531112907263e-05,
|
|
"loss": 0.4092,
|
|
"step": 1680
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.4652463518165736e-05,
|
|
"loss": 0.4015,
|
|
"step": 1682
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.465960740768572e-05,
|
|
"loss": 0.3942,
|
|
"step": 1684
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.466674281780963e-05,
|
|
"loss": 0.4275,
|
|
"step": 1686
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.4673869768642726e-05,
|
|
"loss": 0.4006,
|
|
"step": 1688
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.468098828021887e-05,
|
|
"loss": 0.4241,
|
|
"step": 1690
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.468809837250082e-05,
|
|
"loss": 0.4255,
|
|
"step": 1692
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.46952000653806e-05,
|
|
"loss": 0.3882,
|
|
"step": 1694
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.470229337867983e-05,
|
|
"loss": 0.3973,
|
|
"step": 1696
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.470937833215e-05,
|
|
"loss": 0.3903,
|
|
"step": 1698
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.47164549454729e-05,
|
|
"loss": 0.4517,
|
|
"step": 1700
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.472352323826089e-05,
|
|
"loss": 0.4259,
|
|
"step": 1702
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.473058323005718e-05,
|
|
"loss": 0.4127,
|
|
"step": 1704
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.4737634940336286e-05,
|
|
"loss": 0.3812,
|
|
"step": 1706
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.47446783885042e-05,
|
|
"loss": 0.4323,
|
|
"step": 1708
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.475171359389883e-05,
|
|
"loss": 0.4085,
|
|
"step": 1710
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.4758740575790234e-05,
|
|
"loss": 0.4003,
|
|
"step": 1712
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 4.4765759353381026e-05,
|
|
"loss": 0.4269,
|
|
"step": 1714
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.477276994580658e-05,
|
|
"loss": 0.4214,
|
|
"step": 1716
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.477977237213544e-05,
|
|
"loss": 0.4049,
|
|
"step": 1718
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.4786766651369587e-05,
|
|
"loss": 0.4229,
|
|
"step": 1720
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.479375280244475e-05,
|
|
"loss": 0.4025,
|
|
"step": 1722
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.4800730844230706e-05,
|
|
"loss": 0.4205,
|
|
"step": 1724
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.4807700795531624e-05,
|
|
"loss": 0.4641,
|
|
"step": 1726
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.4814662675086315e-05,
|
|
"loss": 0.4131,
|
|
"step": 1728
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.482161650156859e-05,
|
|
"loss": 0.4139,
|
|
"step": 1730
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.482856229358752e-05,
|
|
"loss": 0.389,
|
|
"step": 1732
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.4835500069687726e-05,
|
|
"loss": 0.4015,
|
|
"step": 1734
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.484242984834975e-05,
|
|
"loss": 0.3763,
|
|
"step": 1736
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.484935164799024e-05,
|
|
"loss": 0.4221,
|
|
"step": 1738
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.4856265486962335e-05,
|
|
"loss": 0.3976,
|
|
"step": 1740
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.486317138355593e-05,
|
|
"loss": 0.3992,
|
|
"step": 1742
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.4870069355997917e-05,
|
|
"loss": 0.4117,
|
|
"step": 1744
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.487695942245258e-05,
|
|
"loss": 0.3766,
|
|
"step": 1746
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.4883841601021766e-05,
|
|
"loss": 0.382,
|
|
"step": 1748
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.489071590974525e-05,
|
|
"loss": 0.3691,
|
|
"step": 1750
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.4897582366600975e-05,
|
|
"loss": 0.4192,
|
|
"step": 1752
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.4904440989505366e-05,
|
|
"loss": 0.4153,
|
|
"step": 1754
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.491129179631359e-05,
|
|
"loss": 0.3981,
|
|
"step": 1756
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.491813480481982e-05,
|
|
"loss": 0.3717,
|
|
"step": 1758
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.492497003275758e-05,
|
|
"loss": 0.4121,
|
|
"step": 1760
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.493179749779993e-05,
|
|
"loss": 0.4128,
|
|
"step": 1762
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 4.49386172175598e-05,
|
|
"loss": 0.4232,
|
|
"step": 1764
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.494542920959023e-05,
|
|
"loss": 0.376,
|
|
"step": 1766
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.4952233491384694e-05,
|
|
"loss": 0.397,
|
|
"step": 1768
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.495903008037729e-05,
|
|
"loss": 0.3749,
|
|
"step": 1770
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.4965818993943085e-05,
|
|
"loss": 0.4048,
|
|
"step": 1772
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.4972600249398325e-05,
|
|
"loss": 0.3936,
|
|
"step": 1774
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.497937386400073e-05,
|
|
"loss": 0.3508,
|
|
"step": 1776
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.498613985494975e-05,
|
|
"loss": 0.3737,
|
|
"step": 1778
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.499289823938682e-05,
|
|
"loss": 0.389,
|
|
"step": 1780
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.499964903439564e-05,
|
|
"loss": 0.3713,
|
|
"step": 1782
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.500639225700242e-05,
|
|
"loss": 0.3785,
|
|
"step": 1784
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.501312792417612e-05,
|
|
"loss": 0.4031,
|
|
"step": 1786
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.501985605282874e-05,
|
|
"loss": 0.4181,
|
|
"step": 1788
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.5026576659815574e-05,
|
|
"loss": 0.3619,
|
|
"step": 1790
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.5033289761935414e-05,
|
|
"loss": 0.4164,
|
|
"step": 1792
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.503999537593087e-05,
|
|
"loss": 0.396,
|
|
"step": 1794
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.5046693518488584e-05,
|
|
"loss": 0.3898,
|
|
"step": 1796
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.5053384206239445e-05,
|
|
"loss": 0.3926,
|
|
"step": 1798
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.5060067455758934e-05,
|
|
"loss": 0.3563,
|
|
"step": 1800
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.506674328356726e-05,
|
|
"loss": 0.4279,
|
|
"step": 1802
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.507341170612969e-05,
|
|
"loss": 0.3916,
|
|
"step": 1804
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.5080072739856756e-05,
|
|
"loss": 0.4427,
|
|
"step": 1806
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.5086726401104464e-05,
|
|
"loss": 0.403,
|
|
"step": 1808
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.5093372706174615e-05,
|
|
"loss": 0.4052,
|
|
"step": 1810
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.5100011671314975e-05,
|
|
"loss": 0.4315,
|
|
"step": 1812
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.5106643312719553e-05,
|
|
"loss": 0.4607,
|
|
"step": 1814
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 4.5113267646528814e-05,
|
|
"loss": 0.3792,
|
|
"step": 1816
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.5119884688829945e-05,
|
|
"loss": 0.4329,
|
|
"step": 1818
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.512649445565703e-05,
|
|
"loss": 0.4476,
|
|
"step": 1820
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.5133096962991364e-05,
|
|
"loss": 0.4164,
|
|
"step": 1822
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.513969222676161e-05,
|
|
"loss": 0.3951,
|
|
"step": 1824
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.5146280262844085e-05,
|
|
"loss": 0.4236,
|
|
"step": 1826
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.5152861087062965e-05,
|
|
"loss": 0.3997,
|
|
"step": 1828
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.51594347151905e-05,
|
|
"loss": 0.4188,
|
|
"step": 1830
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.516600116294727e-05,
|
|
"loss": 0.3977,
|
|
"step": 1832
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.517256044600239e-05,
|
|
"loss": 0.3982,
|
|
"step": 1834
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.517911257997376e-05,
|
|
"loss": 0.4314,
|
|
"step": 1836
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.518565758042822e-05,
|
|
"loss": 0.402,
|
|
"step": 1838
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.519219546288187e-05,
|
|
"loss": 0.4145,
|
|
"step": 1840
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.5198726242800235e-05,
|
|
"loss": 0.4197,
|
|
"step": 1842
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.520524993559846e-05,
|
|
"loss": 0.3993,
|
|
"step": 1844
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.5211766556641586e-05,
|
|
"loss": 0.392,
|
|
"step": 1846
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.5218276121244745e-05,
|
|
"loss": 0.4246,
|
|
"step": 1848
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.522477864467334e-05,
|
|
"loss": 0.3687,
|
|
"step": 1850
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.523127414214334e-05,
|
|
"loss": 0.4215,
|
|
"step": 1852
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.523776262882139e-05,
|
|
"loss": 0.4099,
|
|
"step": 1854
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.524424411982512e-05,
|
|
"loss": 0.4205,
|
|
"step": 1856
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.525071863022329e-05,
|
|
"loss": 0.3977,
|
|
"step": 1858
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.5257186175036044e-05,
|
|
"loss": 0.4136,
|
|
"step": 1860
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.526364676923509e-05,
|
|
"loss": 0.4146,
|
|
"step": 1862
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.527010042774391e-05,
|
|
"loss": 0.4286,
|
|
"step": 1864
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 4.527654716543799e-05,
|
|
"loss": 0.3843,
|
|
"step": 1866
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.5282986997145e-05,
|
|
"loss": 0.3942,
|
|
"step": 1868
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.528941993764502e-05,
|
|
"loss": 0.3979,
|
|
"step": 1870
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.5295846001670715e-05,
|
|
"loss": 0.3792,
|
|
"step": 1872
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.530226520390758e-05,
|
|
"loss": 0.4792,
|
|
"step": 1874
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.530867755899409e-05,
|
|
"loss": 0.4165,
|
|
"step": 1876
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.5315083081521937e-05,
|
|
"loss": 0.3567,
|
|
"step": 1878
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.532148178603623e-05,
|
|
"loss": 0.4027,
|
|
"step": 1880
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.532787368703566e-05,
|
|
"loss": 0.4102,
|
|
"step": 1882
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.533425879897275e-05,
|
|
"loss": 0.4355,
|
|
"step": 1884
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.5340637136253996e-05,
|
|
"loss": 0.4364,
|
|
"step": 1886
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.534700871324008e-05,
|
|
"loss": 0.3995,
|
|
"step": 1888
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.53533735442461e-05,
|
|
"loss": 0.3723,
|
|
"step": 1890
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.53597316435417e-05,
|
|
"loss": 0.4166,
|
|
"step": 1892
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.536608302535133e-05,
|
|
"loss": 0.3877,
|
|
"step": 1894
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.537242770385437e-05,
|
|
"loss": 0.4039,
|
|
"step": 1896
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.5378765693185375e-05,
|
|
"loss": 0.4048,
|
|
"step": 1898
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.538509700743423e-05,
|
|
"loss": 0.4291,
|
|
"step": 1900
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.539142166064635e-05,
|
|
"loss": 0.4486,
|
|
"step": 1902
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.539773966682285e-05,
|
|
"loss": 0.4451,
|
|
"step": 1904
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.54040510399208e-05,
|
|
"loss": 0.4062,
|
|
"step": 1906
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.541035579385329e-05,
|
|
"loss": 0.4247,
|
|
"step": 1908
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.541665394248973e-05,
|
|
"loss": 0.433,
|
|
"step": 1910
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.542294549965595e-05,
|
|
"loss": 0.4517,
|
|
"step": 1912
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.542923047913445e-05,
|
|
"loss": 0.4101,
|
|
"step": 1914
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.5435508894664514e-05,
|
|
"loss": 0.4139,
|
|
"step": 1916
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 4.544178075994245e-05,
|
|
"loss": 0.4008,
|
|
"step": 1918
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.544804608862172e-05,
|
|
"loss": 0.4002,
|
|
"step": 1920
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.545430489431316e-05,
|
|
"loss": 0.3896,
|
|
"step": 1922
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.5460557190585136e-05,
|
|
"loss": 0.4164,
|
|
"step": 1924
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.546680299096371e-05,
|
|
"loss": 0.4165,
|
|
"step": 1926
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.5473042308932836e-05,
|
|
"loss": 0.4165,
|
|
"step": 1928
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.5479275157934526e-05,
|
|
"loss": 0.3811,
|
|
"step": 1930
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.5485501551369036e-05,
|
|
"loss": 0.4088,
|
|
"step": 1932
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.5491721502595017e-05,
|
|
"loss": 0.408,
|
|
"step": 1934
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.54979350249297e-05,
|
|
"loss": 0.3635,
|
|
"step": 1936
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.550414213164905e-05,
|
|
"loss": 0.4222,
|
|
"step": 1938
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.551034283598798e-05,
|
|
"loss": 0.434,
|
|
"step": 1940
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.5516537151140485e-05,
|
|
"loss": 0.4304,
|
|
"step": 1942
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.552272509025978e-05,
|
|
"loss": 0.4323,
|
|
"step": 1944
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.552890666645856e-05,
|
|
"loss": 0.4176,
|
|
"step": 1946
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.553508189280907e-05,
|
|
"loss": 0.4083,
|
|
"step": 1948
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.5541250782343334e-05,
|
|
"loss": 0.3801,
|
|
"step": 1950
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.5547413348053284e-05,
|
|
"loss": 0.4105,
|
|
"step": 1952
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.555356960289095e-05,
|
|
"loss": 0.4182,
|
|
"step": 1954
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.55597195597686e-05,
|
|
"loss": 0.455,
|
|
"step": 1956
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.5565863231558935e-05,
|
|
"loss": 0.3899,
|
|
"step": 1958
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.5572000631095196e-05,
|
|
"loss": 0.3902,
|
|
"step": 1960
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.5578131771171384e-05,
|
|
"loss": 0.4343,
|
|
"step": 1962
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.558425666454239e-05,
|
|
"loss": 0.3998,
|
|
"step": 1964
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.559037532392416e-05,
|
|
"loss": 0.4094,
|
|
"step": 1966
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.5596487761993835e-05,
|
|
"loss": 0.4091,
|
|
"step": 1968
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 4.560259399138995e-05,
|
|
"loss": 0.4197,
|
|
"step": 1970
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.560869402471256e-05,
|
|
"loss": 0.4261,
|
|
"step": 1972
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.561478787452339e-05,
|
|
"loss": 0.3993,
|
|
"step": 1974
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.562087555334601e-05,
|
|
"loss": 0.373,
|
|
"step": 1976
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.562695707366599e-05,
|
|
"loss": 0.3928,
|
|
"step": 1978
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.563303244793105e-05,
|
|
"loss": 0.4211,
|
|
"step": 1980
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.563910168855117e-05,
|
|
"loss": 0.4484,
|
|
"step": 1982
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.5645164807898835e-05,
|
|
"loss": 0.3854,
|
|
"step": 1984
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.565122181830909e-05,
|
|
"loss": 0.3953,
|
|
"step": 1986
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.565727273207975e-05,
|
|
"loss": 0.3984,
|
|
"step": 1988
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.5663317561471544e-05,
|
|
"loss": 0.4366,
|
|
"step": 1990
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.566935631870823e-05,
|
|
"loss": 0.3896,
|
|
"step": 1992
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.567538901597677e-05,
|
|
"loss": 0.3877,
|
|
"step": 1994
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.568141566542749e-05,
|
|
"loss": 0.4197,
|
|
"step": 1996
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.5687436279174196e-05,
|
|
"loss": 0.4075,
|
|
"step": 1998
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.569345086929434e-05,
|
|
"loss": 0.4011,
|
|
"step": 2000
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.5699459447829145e-05,
|
|
"loss": 0.3876,
|
|
"step": 2002
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.570546202678379e-05,
|
|
"loss": 0.4078,
|
|
"step": 2004
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.571145861812752e-05,
|
|
"loss": 0.428,
|
|
"step": 2006
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.5717449233793776e-05,
|
|
"loss": 0.391,
|
|
"step": 2008
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.572343388568039e-05,
|
|
"loss": 0.3708,
|
|
"step": 2010
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.5729412585649675e-05,
|
|
"loss": 0.4315,
|
|
"step": 2012
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.573538534552858e-05,
|
|
"loss": 0.4342,
|
|
"step": 2014
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.574135217710888e-05,
|
|
"loss": 0.392,
|
|
"step": 2016
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.574731309214723e-05,
|
|
"loss": 0.3972,
|
|
"step": 2018
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 4.575326810236534e-05,
|
|
"loss": 0.3911,
|
|
"step": 2020
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.575921721945018e-05,
|
|
"loss": 0.3904,
|
|
"step": 2022
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.576516045505399e-05,
|
|
"loss": 0.4301,
|
|
"step": 2024
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.5771097820794514e-05,
|
|
"loss": 0.3813,
|
|
"step": 2026
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.5777029328255115e-05,
|
|
"loss": 0.4364,
|
|
"step": 2028
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.57829549889849e-05,
|
|
"loss": 0.4167,
|
|
"step": 2030
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.578887481449884e-05,
|
|
"loss": 0.3893,
|
|
"step": 2032
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.579478881627793e-05,
|
|
"loss": 0.4084,
|
|
"step": 2034
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.580069700576932e-05,
|
|
"loss": 0.3787,
|
|
"step": 2036
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.580659939438645e-05,
|
|
"loss": 0.4336,
|
|
"step": 2038
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.581249599350916e-05,
|
|
"loss": 0.4369,
|
|
"step": 2040
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.581838681448383e-05,
|
|
"loss": 0.4166,
|
|
"step": 2042
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.582427186862354e-05,
|
|
"loss": 0.4194,
|
|
"step": 2044
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.583015116720816e-05,
|
|
"loss": 0.4896,
|
|
"step": 2046
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.58360247214845e-05,
|
|
"loss": 0.3553,
|
|
"step": 2048
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.584189254266645e-05,
|
|
"loss": 0.416,
|
|
"step": 2050
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.584775464193508e-05,
|
|
"loss": 0.4453,
|
|
"step": 2052
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.5853611030438774e-05,
|
|
"loss": 0.3856,
|
|
"step": 2054
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.5859461719293396e-05,
|
|
"loss": 0.4336,
|
|
"step": 2056
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.586530671958236e-05,
|
|
"loss": 0.4177,
|
|
"step": 2058
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.587114604235679e-05,
|
|
"loss": 0.4301,
|
|
"step": 2060
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.587697969863565e-05,
|
|
"loss": 0.4077,
|
|
"step": 2062
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.588280769940584e-05,
|
|
"loss": 0.4139,
|
|
"step": 2064
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.5888630055622345e-05,
|
|
"loss": 0.3959,
|
|
"step": 2066
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.589444677820834e-05,
|
|
"loss": 0.4027,
|
|
"step": 2068
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.590025787805534e-05,
|
|
"loss": 0.4433,
|
|
"step": 2070
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 4.5906063366023295e-05,
|
|
"loss": 0.424,
|
|
"step": 2072
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.591186325294073e-05,
|
|
"loss": 0.4077,
|
|
"step": 2074
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.591765754960484e-05,
|
|
"loss": 0.4078,
|
|
"step": 2076
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.5923446266781654e-05,
|
|
"loss": 0.4098,
|
|
"step": 2078
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.592922941520612e-05,
|
|
"loss": 0.3766,
|
|
"step": 2080
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.5935007005582224e-05,
|
|
"loss": 0.3852,
|
|
"step": 2082
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.594077904858314e-05,
|
|
"loss": 0.3986,
|
|
"step": 2084
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.594654555485131e-05,
|
|
"loss": 0.3922,
|
|
"step": 2086
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.5952306534998586e-05,
|
|
"loss": 0.4012,
|
|
"step": 2088
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.5958061999606346e-05,
|
|
"loss": 0.4246,
|
|
"step": 2090
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.596381195922559e-05,
|
|
"loss": 0.3962,
|
|
"step": 2092
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.5969556424377094e-05,
|
|
"loss": 0.415,
|
|
"step": 2094
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.597529540555148e-05,
|
|
"loss": 0.4221,
|
|
"step": 2096
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.5981028913209384e-05,
|
|
"loss": 0.3984,
|
|
"step": 2098
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.598675695778149e-05,
|
|
"loss": 0.424,
|
|
"step": 2100
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.599247954966876e-05,
|
|
"loss": 0.395,
|
|
"step": 2102
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.599819669924243e-05,
|
|
"loss": 0.3857,
|
|
"step": 2104
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.600390841684418e-05,
|
|
"loss": 0.4069,
|
|
"step": 2106
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.6009614712786274e-05,
|
|
"loss": 0.3835,
|
|
"step": 2108
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.6015315597351596e-05,
|
|
"loss": 0.4102,
|
|
"step": 2110
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.602101108079383e-05,
|
|
"loss": 0.421,
|
|
"step": 2112
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.602670117333754e-05,
|
|
"loss": 0.4131,
|
|
"step": 2114
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.603238588517828e-05,
|
|
"loss": 0.4188,
|
|
"step": 2116
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.6038065226482705e-05,
|
|
"loss": 0.4239,
|
|
"step": 2118
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.604373920738869e-05,
|
|
"loss": 0.4084,
|
|
"step": 2120
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 4.604940783800543e-05,
|
|
"loss": 0.4005,
|
|
"step": 2122
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.605507112841355e-05,
|
|
"loss": 0.4017,
|
|
"step": 2124
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.606072908866519e-05,
|
|
"loss": 0.382,
|
|
"step": 2126
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.606638172878418e-05,
|
|
"loss": 0.3785,
|
|
"step": 2128
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.6072029058766056e-05,
|
|
"loss": 0.4229,
|
|
"step": 2130
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.6077671088578235e-05,
|
|
"loss": 0.3741,
|
|
"step": 2132
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.6083307828160097e-05,
|
|
"loss": 0.4261,
|
|
"step": 2134
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.608893928742308e-05,
|
|
"loss": 0.4039,
|
|
"step": 2136
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.609456547625078e-05,
|
|
"loss": 0.3484,
|
|
"step": 2138
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.6100186404499105e-05,
|
|
"loss": 0.3984,
|
|
"step": 2140
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.610580208199632e-05,
|
|
"loss": 0.3866,
|
|
"step": 2142
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.6111412518543175e-05,
|
|
"loss": 0.3927,
|
|
"step": 2144
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.6117017723913e-05,
|
|
"loss": 0.4291,
|
|
"step": 2146
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.6122617707851825e-05,
|
|
"loss": 0.408,
|
|
"step": 2148
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.612821248007846e-05,
|
|
"loss": 0.3971,
|
|
"step": 2150
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.6133802050284606e-05,
|
|
"loss": 0.3681,
|
|
"step": 2152
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.613938642813496e-05,
|
|
"loss": 0.3817,
|
|
"step": 2154
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.6144965623267304e-05,
|
|
"loss": 0.384,
|
|
"step": 2156
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.615053964529263e-05,
|
|
"loss": 0.3646,
|
|
"step": 2158
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.615610850379518e-05,
|
|
"loss": 0.3992,
|
|
"step": 2160
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.616167220833264e-05,
|
|
"loss": 0.3866,
|
|
"step": 2162
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.616723076843611e-05,
|
|
"loss": 0.4041,
|
|
"step": 2164
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.617278419361037e-05,
|
|
"loss": 0.3993,
|
|
"step": 2166
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.617833249333381e-05,
|
|
"loss": 0.4648,
|
|
"step": 2168
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.618387567705861e-05,
|
|
"loss": 0.3933,
|
|
"step": 2170
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.6189413754210866e-05,
|
|
"loss": 0.4463,
|
|
"step": 2172
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 4.619494673419061e-05,
|
|
"loss": 0.402,
|
|
"step": 2174
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.620047462637194e-05,
|
|
"loss": 0.4044,
|
|
"step": 2176
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.620599744010316e-05,
|
|
"loss": 0.4125,
|
|
"step": 2178
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.621151518470679e-05,
|
|
"loss": 0.4014,
|
|
"step": 2180
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.6217027869479715e-05,
|
|
"loss": 0.4747,
|
|
"step": 2182
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.622253550369329e-05,
|
|
"loss": 0.3444,
|
|
"step": 2184
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.622803809659337e-05,
|
|
"loss": 0.4058,
|
|
"step": 2186
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.623353565740047e-05,
|
|
"loss": 0.4154,
|
|
"step": 2188
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.6239028195309845e-05,
|
|
"loss": 0.4845,
|
|
"step": 2190
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.624451571949153e-05,
|
|
"loss": 0.4243,
|
|
"step": 2192
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.624999823909051e-05,
|
|
"loss": 0.4148,
|
|
"step": 2194
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.625547576322675e-05,
|
|
"loss": 0.3995,
|
|
"step": 2196
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.6260948300995314e-05,
|
|
"loss": 0.3824,
|
|
"step": 2198
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.626641586146645e-05,
|
|
"loss": 0.386,
|
|
"step": 2200
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.627187845368567e-05,
|
|
"loss": 0.4142,
|
|
"step": 2202
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.6277336086673876e-05,
|
|
"loss": 0.4016,
|
|
"step": 2204
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.62827887694274e-05,
|
|
"loss": 0.4064,
|
|
"step": 2206
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.6288236510918124e-05,
|
|
"loss": 0.4357,
|
|
"step": 2208
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.629367932009356e-05,
|
|
"loss": 0.3807,
|
|
"step": 2210
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.629911720587694e-05,
|
|
"loss": 0.3939,
|
|
"step": 2212
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.63045501771673e-05,
|
|
"loss": 0.3734,
|
|
"step": 2214
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.630997824283958e-05,
|
|
"loss": 0.4501,
|
|
"step": 2216
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.6315401411744684e-05,
|
|
"loss": 0.3693,
|
|
"step": 2218
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.63208196927096e-05,
|
|
"loss": 0.4106,
|
|
"step": 2220
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.632623309453746e-05,
|
|
"loss": 0.4108,
|
|
"step": 2222
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 4.633164162600765e-05,
|
|
"loss": 0.447,
|
|
"step": 2224
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.6337045295875856e-05,
|
|
"loss": 0.3908,
|
|
"step": 2226
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.63424441128742e-05,
|
|
"loss": 0.3882,
|
|
"step": 2228
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.6347838085711285e-05,
|
|
"loss": 0.4033,
|
|
"step": 2230
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.6353227223072296e-05,
|
|
"loss": 0.4136,
|
|
"step": 2232
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.6358611533619085e-05,
|
|
"loss": 0.4222,
|
|
"step": 2234
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.6363991025990244e-05,
|
|
"loss": 0.3749,
|
|
"step": 2236
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.6369365708801197e-05,
|
|
"loss": 0.4148,
|
|
"step": 2238
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.6374735590644284e-05,
|
|
"loss": 0.4276,
|
|
"step": 2240
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.638010068008884e-05,
|
|
"loss": 0.377,
|
|
"step": 2242
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.638546098568127e-05,
|
|
"loss": 0.382,
|
|
"step": 2244
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.639081651594514e-05,
|
|
"loss": 0.3786,
|
|
"step": 2246
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.639616727938128e-05,
|
|
"loss": 0.4369,
|
|
"step": 2248
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.6401513284467804e-05,
|
|
"loss": 0.3572,
|
|
"step": 2250
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.640685453966024e-05,
|
|
"loss": 0.4116,
|
|
"step": 2252
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.641219105339161e-05,
|
|
"loss": 0.3896,
|
|
"step": 2254
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.641752283407248e-05,
|
|
"loss": 0.3588,
|
|
"step": 2256
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.6422849890091074e-05,
|
|
"loss": 0.428,
|
|
"step": 2258
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.642817222981333e-05,
|
|
"loss": 0.386,
|
|
"step": 2260
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.6433489861582986e-05,
|
|
"loss": 0.3932,
|
|
"step": 2262
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.643880279372166e-05,
|
|
"loss": 0.4,
|
|
"step": 2264
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.6444111034528907e-05,
|
|
"loss": 0.4128,
|
|
"step": 2266
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.644941459228235e-05,
|
|
"loss": 0.4009,
|
|
"step": 2268
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.645471347523769e-05,
|
|
"loss": 0.407,
|
|
"step": 2270
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.646000769162884e-05,
|
|
"loss": 0.3594,
|
|
"step": 2272
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.646529724966797e-05,
|
|
"loss": 0.4431,
|
|
"step": 2274
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 4.6470582157545586e-05,
|
|
"loss": 0.4518,
|
|
"step": 2276
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.647586242343061e-05,
|
|
"loss": 0.431,
|
|
"step": 2278
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.6481138055470475e-05,
|
|
"loss": 0.4072,
|
|
"step": 2280
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.6486409061791176e-05,
|
|
"loss": 0.356,
|
|
"step": 2282
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.649167545049733e-05,
|
|
"loss": 0.4835,
|
|
"step": 2284
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.64969372296723e-05,
|
|
"loss": 0.3894,
|
|
"step": 2286
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.650219440737823e-05,
|
|
"loss": 0.3843,
|
|
"step": 2288
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.650744699165614e-05,
|
|
"loss": 0.4099,
|
|
"step": 2290
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.6512694990525976e-05,
|
|
"loss": 0.4454,
|
|
"step": 2292
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.651793841198671e-05,
|
|
"loss": 0.3818,
|
|
"step": 2294
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.65231772640164e-05,
|
|
"loss": 0.4361,
|
|
"step": 2296
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.652841155457226e-05,
|
|
"loss": 0.4005,
|
|
"step": 2298
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.653364129159074e-05,
|
|
"loss": 0.3986,
|
|
"step": 2300
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.65388664829876e-05,
|
|
"loss": 0.438,
|
|
"step": 2302
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.654408713665797e-05,
|
|
"loss": 0.421,
|
|
"step": 2304
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.654930326047643e-05,
|
|
"loss": 0.385,
|
|
"step": 2306
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.6554514862297085e-05,
|
|
"loss": 0.4093,
|
|
"step": 2308
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.655972194995361e-05,
|
|
"loss": 0.3929,
|
|
"step": 2310
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.656492453125939e-05,
|
|
"loss": 0.4757,
|
|
"step": 2312
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.657012261400748e-05,
|
|
"loss": 0.4221,
|
|
"step": 2314
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.657531620597078e-05,
|
|
"loss": 0.3797,
|
|
"step": 2316
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.658050531490205e-05,
|
|
"loss": 0.3904,
|
|
"step": 2318
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.658568994853399e-05,
|
|
"loss": 0.3914,
|
|
"step": 2320
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.6590870114579305e-05,
|
|
"loss": 0.4304,
|
|
"step": 2322
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.659604582073079e-05,
|
|
"loss": 0.3561,
|
|
"step": 2324
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.6601217074661384e-05,
|
|
"loss": 0.4216,
|
|
"step": 2326
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 4.6606383884024226e-05,
|
|
"loss": 0.3832,
|
|
"step": 2328
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.6611546256452776e-05,
|
|
"loss": 0.4028,
|
|
"step": 2330
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.661670419956081e-05,
|
|
"loss": 0.4072,
|
|
"step": 2332
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.662185772094252e-05,
|
|
"loss": 0.4041,
|
|
"step": 2334
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.662700682817263e-05,
|
|
"loss": 0.4163,
|
|
"step": 2336
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.663215152880636e-05,
|
|
"loss": 0.3976,
|
|
"step": 2338
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.663729183037958e-05,
|
|
"loss": 0.4287,
|
|
"step": 2340
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.664242774040885e-05,
|
|
"loss": 0.4084,
|
|
"step": 2342
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.664755926639148e-05,
|
|
"loss": 0.4355,
|
|
"step": 2344
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.665268641580556e-05,
|
|
"loss": 0.4096,
|
|
"step": 2346
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.665780919611012e-05,
|
|
"loss": 0.4163,
|
|
"step": 2348
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.6662927614745096e-05,
|
|
"loss": 0.4198,
|
|
"step": 2350
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.666804167913145e-05,
|
|
"loss": 0.3615,
|
|
"step": 2352
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.667315139667122e-05,
|
|
"loss": 0.4165,
|
|
"step": 2354
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.667825677474759e-05,
|
|
"loss": 0.4127,
|
|
"step": 2356
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.6683357820724954e-05,
|
|
"loss": 0.4447,
|
|
"step": 2358
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.6688454541948945e-05,
|
|
"loss": 0.4182,
|
|
"step": 2360
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.669354694574657e-05,
|
|
"loss": 0.4052,
|
|
"step": 2362
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.66986350394262e-05,
|
|
"loss": 0.3961,
|
|
"step": 2364
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.670371883027769e-05,
|
|
"loss": 0.4232,
|
|
"step": 2366
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.6708798325572384e-05,
|
|
"loss": 0.4171,
|
|
"step": 2368
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.671387353256324e-05,
|
|
"loss": 0.4203,
|
|
"step": 2370
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.671894445848485e-05,
|
|
"loss": 0.4486,
|
|
"step": 2372
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.672401111055352e-05,
|
|
"loss": 0.4087,
|
|
"step": 2374
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.67290734959673e-05,
|
|
"loss": 0.4069,
|
|
"step": 2376
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 4.67341316219061e-05,
|
|
"loss": 0.3944,
|
|
"step": 2378
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.673918549553172e-05,
|
|
"loss": 0.3939,
|
|
"step": 2380
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.67442351239879e-05,
|
|
"loss": 0.377,
|
|
"step": 2382
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.67492805144004e-05,
|
|
"loss": 0.3882,
|
|
"step": 2384
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.6754321673877036e-05,
|
|
"loss": 0.3881,
|
|
"step": 2386
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.6759358609507795e-05,
|
|
"loss": 0.438,
|
|
"step": 2388
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.676439132836482e-05,
|
|
"loss": 0.4233,
|
|
"step": 2390
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.6769419837502524e-05,
|
|
"loss": 0.4039,
|
|
"step": 2392
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.677444414395764e-05,
|
|
"loss": 0.4115,
|
|
"step": 2394
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.677946425474926e-05,
|
|
"loss": 0.4315,
|
|
"step": 2396
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.67844801768789e-05,
|
|
"loss": 0.4048,
|
|
"step": 2398
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.678949191733059e-05,
|
|
"loss": 0.4258,
|
|
"step": 2400
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.679449948307088e-05,
|
|
"loss": 0.4004,
|
|
"step": 2402
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.6799502881048933e-05,
|
|
"loss": 0.4406,
|
|
"step": 2404
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.680450211819659e-05,
|
|
"loss": 0.4104,
|
|
"step": 2406
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.6809497201428404e-05,
|
|
"loss": 0.3538,
|
|
"step": 2408
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.68144881376417e-05,
|
|
"loss": 0.4276,
|
|
"step": 2410
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.6819474933716636e-05,
|
|
"loss": 0.3774,
|
|
"step": 2412
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.682445759651628e-05,
|
|
"loss": 0.4226,
|
|
"step": 2414
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.6829436132886636e-05,
|
|
"loss": 0.516,
|
|
"step": 2416
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.6834410549656695e-05,
|
|
"loss": 0.3829,
|
|
"step": 2418
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.683938085363856e-05,
|
|
"loss": 0.4081,
|
|
"step": 2420
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.684434705162741e-05,
|
|
"loss": 0.3757,
|
|
"step": 2422
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.68493091504016e-05,
|
|
"loss": 0.4374,
|
|
"step": 2424
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.685426715672273e-05,
|
|
"loss": 0.4353,
|
|
"step": 2426
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.685922107733567e-05,
|
|
"loss": 0.3948,
|
|
"step": 2428
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 4.6864170918968646e-05,
|
|
"loss": 0.4116,
|
|
"step": 2430
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.686911668833327e-05,
|
|
"loss": 0.3866,
|
|
"step": 2432
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.6874058392124576e-05,
|
|
"loss": 0.3681,
|
|
"step": 2434
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.687899603702115e-05,
|
|
"loss": 0.3798,
|
|
"step": 2436
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.68839296296851e-05,
|
|
"loss": 0.4028,
|
|
"step": 2438
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.688885917676215e-05,
|
|
"loss": 0.385,
|
|
"step": 2440
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.6893784684881714e-05,
|
|
"loss": 0.3521,
|
|
"step": 2442
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.6898706160656876e-05,
|
|
"loss": 0.4241,
|
|
"step": 2444
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.690362361068454e-05,
|
|
"loss": 0.3975,
|
|
"step": 2446
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.690853704154541e-05,
|
|
"loss": 0.4237,
|
|
"step": 2448
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.691344645980406e-05,
|
|
"loss": 0.3651,
|
|
"step": 2450
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.691835187200902e-05,
|
|
"loss": 0.38,
|
|
"step": 2452
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.6923253284692785e-05,
|
|
"loss": 0.3836,
|
|
"step": 2454
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.692815070437189e-05,
|
|
"loss": 0.3811,
|
|
"step": 2456
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.6933044137546936e-05,
|
|
"loss": 0.4301,
|
|
"step": 2458
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.6937933590702705e-05,
|
|
"loss": 0.3903,
|
|
"step": 2460
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.694281907030813e-05,
|
|
"loss": 0.3666,
|
|
"step": 2462
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.694770058281639e-05,
|
|
"loss": 0.4111,
|
|
"step": 2464
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.6952578134665e-05,
|
|
"loss": 0.4153,
|
|
"step": 2466
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.6957451732275756e-05,
|
|
"loss": 0.415,
|
|
"step": 2468
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.696232138205488e-05,
|
|
"loss": 0.4263,
|
|
"step": 2470
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.696718709039304e-05,
|
|
"loss": 0.3967,
|
|
"step": 2472
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.69720488636654e-05,
|
|
"loss": 0.4363,
|
|
"step": 2474
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.697690670823166e-05,
|
|
"loss": 0.3629,
|
|
"step": 2476
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.6981760630436115e-05,
|
|
"loss": 0.4099,
|
|
"step": 2478
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 4.69866106366077e-05,
|
|
"loss": 0.4234,
|
|
"step": 2480
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.699145673306007e-05,
|
|
"loss": 0.3932,
|
|
"step": 2482
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.699629892609159e-05,
|
|
"loss": 0.3888,
|
|
"step": 2484
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.700113722198544e-05,
|
|
"loss": 0.4083,
|
|
"step": 2486
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.7005971627009645e-05,
|
|
"loss": 0.4218,
|
|
"step": 2488
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.701080214741709e-05,
|
|
"loss": 0.4127,
|
|
"step": 2490
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.701562878944564e-05,
|
|
"loss": 0.4333,
|
|
"step": 2492
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.7020451559318116e-05,
|
|
"loss": 0.4004,
|
|
"step": 2494
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.7025270463242376e-05,
|
|
"loss": 0.3765,
|
|
"step": 2496
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.7030085507411375e-05,
|
|
"loss": 0.3904,
|
|
"step": 2498
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.70348966980032e-05,
|
|
"loss": 0.4174,
|
|
"step": 2500
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.703970404118111e-05,
|
|
"loss": 0.4437,
|
|
"step": 2502
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.7044507543093584e-05,
|
|
"loss": 0.4222,
|
|
"step": 2504
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.704930720987439e-05,
|
|
"loss": 0.4591,
|
|
"step": 2506
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.70541030476426e-05,
|
|
"loss": 0.3981,
|
|
"step": 2508
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.7058895062502647e-05,
|
|
"loss": 0.3875,
|
|
"step": 2510
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.706368326054441e-05,
|
|
"loss": 0.3762,
|
|
"step": 2512
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.706846764784319e-05,
|
|
"loss": 0.4099,
|
|
"step": 2514
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.707324823045982e-05,
|
|
"loss": 0.3867,
|
|
"step": 2516
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.7078025014440686e-05,
|
|
"loss": 0.4038,
|
|
"step": 2518
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.708279800581775e-05,
|
|
"loss": 0.3901,
|
|
"step": 2520
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.708756721060863e-05,
|
|
"loss": 0.3883,
|
|
"step": 2522
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.709233263481665e-05,
|
|
"loss": 0.4297,
|
|
"step": 2524
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.709709428443084e-05,
|
|
"loss": 0.3997,
|
|
"step": 2526
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.710185216542603e-05,
|
|
"loss": 0.3699,
|
|
"step": 2528
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.710660628376286e-05,
|
|
"loss": 0.4185,
|
|
"step": 2530
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 4.711135664538785e-05,
|
|
"loss": 0.3967,
|
|
"step": 2532
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.711610325623343e-05,
|
|
"loss": 0.4573,
|
|
"step": 2534
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.7120846122218e-05,
|
|
"loss": 0.3971,
|
|
"step": 2536
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.7125585249245944e-05,
|
|
"loss": 0.5222,
|
|
"step": 2538
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.71303206432077e-05,
|
|
"loss": 0.4156,
|
|
"step": 2540
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.7135052309979816e-05,
|
|
"loss": 0.4073,
|
|
"step": 2542
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.713978025542495e-05,
|
|
"loss": 0.3741,
|
|
"step": 2544
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.714450448539194e-05,
|
|
"loss": 0.4135,
|
|
"step": 2546
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.7149225005715854e-05,
|
|
"loss": 0.386,
|
|
"step": 2548
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.715394182221802e-05,
|
|
"loss": 0.4369,
|
|
"step": 2550
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.7158654940706105e-05,
|
|
"loss": 0.4521,
|
|
"step": 2552
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.716336436697407e-05,
|
|
"loss": 0.4269,
|
|
"step": 2554
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.71680701068023e-05,
|
|
"loss": 0.4105,
|
|
"step": 2556
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.717277216595765e-05,
|
|
"loss": 0.3929,
|
|
"step": 2558
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.717747055019337e-05,
|
|
"loss": 0.4241,
|
|
"step": 2560
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.718216526524932e-05,
|
|
"loss": 0.3986,
|
|
"step": 2562
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.7186856316851864e-05,
|
|
"loss": 0.3839,
|
|
"step": 2564
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.7191543710714e-05,
|
|
"loss": 0.4333,
|
|
"step": 2566
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.7196227452535356e-05,
|
|
"loss": 0.4095,
|
|
"step": 2568
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.720090754800226e-05,
|
|
"loss": 0.3768,
|
|
"step": 2570
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.7205584002787765e-05,
|
|
"loss": 0.418,
|
|
"step": 2572
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.72102568225517e-05,
|
|
"loss": 0.4279,
|
|
"step": 2574
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.721492601294069e-05,
|
|
"loss": 0.4127,
|
|
"step": 2576
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.721959157958823e-05,
|
|
"loss": 0.3995,
|
|
"step": 2578
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.722425352811471e-05,
|
|
"loss": 0.4143,
|
|
"step": 2580
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.7228911864127435e-05,
|
|
"loss": 0.3824,
|
|
"step": 2582
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 4.7233566593220705e-05,
|
|
"loss": 0.446,
|
|
"step": 2584
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.723821772097583e-05,
|
|
"loss": 0.4091,
|
|
"step": 2586
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.724286525296116e-05,
|
|
"loss": 0.4326,
|
|
"step": 2588
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.7247509194732166e-05,
|
|
"loss": 0.4283,
|
|
"step": 2590
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.725214955183144e-05,
|
|
"loss": 0.3759,
|
|
"step": 2592
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.725678632978874e-05,
|
|
"loss": 0.3715,
|
|
"step": 2594
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.726141953412106e-05,
|
|
"loss": 0.3738,
|
|
"step": 2596
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.726604917033263e-05,
|
|
"loss": 0.3905,
|
|
"step": 2598
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.727067524391499e-05,
|
|
"loss": 0.4308,
|
|
"step": 2600
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.727529776034699e-05,
|
|
"loss": 0.3986,
|
|
"step": 2602
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.727991672509486e-05,
|
|
"loss": 0.3992,
|
|
"step": 2604
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.728453214361226e-05,
|
|
"loss": 0.4031,
|
|
"step": 2606
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.7289144021340265e-05,
|
|
"loss": 0.4193,
|
|
"step": 2608
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.729375236370745e-05,
|
|
"loss": 0.4105,
|
|
"step": 2610
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.7298357176129934e-05,
|
|
"loss": 0.4419,
|
|
"step": 2612
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.730295846401137e-05,
|
|
"loss": 0.3727,
|
|
"step": 2614
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.730755623274304e-05,
|
|
"loss": 0.381,
|
|
"step": 2616
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.731215048770383e-05,
|
|
"loss": 0.4327,
|
|
"step": 2618
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.731674123426035e-05,
|
|
"loss": 0.4115,
|
|
"step": 2620
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.732132847776689e-05,
|
|
"loss": 0.4009,
|
|
"step": 2622
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.732591222356549e-05,
|
|
"loss": 0.385,
|
|
"step": 2624
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.7330492476986e-05,
|
|
"loss": 0.4468,
|
|
"step": 2626
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.733506924334609e-05,
|
|
"loss": 0.387,
|
|
"step": 2628
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.73396425279513e-05,
|
|
"loss": 0.4195,
|
|
"step": 2630
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.734421233609504e-05,
|
|
"loss": 0.392,
|
|
"step": 2632
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 4.7348778673058705e-05,
|
|
"loss": 0.4396,
|
|
"step": 2634
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.735334154411162e-05,
|
|
"loss": 0.3937,
|
|
"step": 2636
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.735790095451115e-05,
|
|
"loss": 0.4379,
|
|
"step": 2638
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.7362456909502703e-05,
|
|
"loss": 0.365,
|
|
"step": 2640
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.736700941431975e-05,
|
|
"loss": 0.4155,
|
|
"step": 2642
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.7371558474183895e-05,
|
|
"loss": 0.3857,
|
|
"step": 2644
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.737610409430492e-05,
|
|
"loss": 0.4326,
|
|
"step": 2646
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.7380646279880744e-05,
|
|
"loss": 0.368,
|
|
"step": 2648
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.738518503609756e-05,
|
|
"loss": 0.3861,
|
|
"step": 2650
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.7389720368129815e-05,
|
|
"loss": 0.4069,
|
|
"step": 2652
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.739425228114023e-05,
|
|
"loss": 0.4132,
|
|
"step": 2654
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.7398780780279876e-05,
|
|
"loss": 0.3733,
|
|
"step": 2656
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.740330587068821e-05,
|
|
"loss": 0.4584,
|
|
"step": 2658
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.740782755749305e-05,
|
|
"loss": 0.4136,
|
|
"step": 2660
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.7412345845810675e-05,
|
|
"loss": 0.3771,
|
|
"step": 2662
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.741686074074585e-05,
|
|
"loss": 0.3706,
|
|
"step": 2664
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.742137224739182e-05,
|
|
"loss": 0.4363,
|
|
"step": 2666
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.7425880370830395e-05,
|
|
"loss": 0.3931,
|
|
"step": 2668
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.743038511613194e-05,
|
|
"loss": 0.4073,
|
|
"step": 2670
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.743488648835545e-05,
|
|
"loss": 0.3776,
|
|
"step": 2672
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.743938449254854e-05,
|
|
"loss": 0.4103,
|
|
"step": 2674
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.7443879133747537e-05,
|
|
"loss": 0.4493,
|
|
"step": 2676
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.744837041697744e-05,
|
|
"loss": 0.402,
|
|
"step": 2678
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.7452858347252046e-05,
|
|
"loss": 0.379,
|
|
"step": 2680
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.745734292957386e-05,
|
|
"loss": 0.4108,
|
|
"step": 2682
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.746182416893427e-05,
|
|
"loss": 0.5107,
|
|
"step": 2684
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 4.746630207031347e-05,
|
|
"loss": 0.4894,
|
|
"step": 2686
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.7470776638680536e-05,
|
|
"loss": 0.3683,
|
|
"step": 2688
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.747524787899347e-05,
|
|
"loss": 0.4359,
|
|
"step": 2690
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.747971579619922e-05,
|
|
"loss": 0.4221,
|
|
"step": 2692
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.74841803952337e-05,
|
|
"loss": 0.3964,
|
|
"step": 2694
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.748864168102183e-05,
|
|
"loss": 0.4282,
|
|
"step": 2696
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.7493099658477604e-05,
|
|
"loss": 0.3993,
|
|
"step": 2698
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.749755433250405e-05,
|
|
"loss": 0.3751,
|
|
"step": 2700
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.750200570799334e-05,
|
|
"loss": 0.4206,
|
|
"step": 2702
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.7506453789826776e-05,
|
|
"loss": 0.4039,
|
|
"step": 2704
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.7510898582874813e-05,
|
|
"loss": 0.4525,
|
|
"step": 2706
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.751534009199715e-05,
|
|
"loss": 0.4187,
|
|
"step": 2708
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.751977832204268e-05,
|
|
"loss": 0.4237,
|
|
"step": 2710
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.7524213277849585e-05,
|
|
"loss": 0.4333,
|
|
"step": 2712
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.752864496424536e-05,
|
|
"loss": 0.4292,
|
|
"step": 2714
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.75330733860468e-05,
|
|
"loss": 0.4041,
|
|
"step": 2716
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.7537498548060097e-05,
|
|
"loss": 0.4034,
|
|
"step": 2718
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.754192045508081e-05,
|
|
"loss": 0.3914,
|
|
"step": 2720
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.754633911189394e-05,
|
|
"loss": 0.4276,
|
|
"step": 2722
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.7550754523273936e-05,
|
|
"loss": 0.3926,
|
|
"step": 2724
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.755516669398474e-05,
|
|
"loss": 0.4144,
|
|
"step": 2726
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.755957562877982e-05,
|
|
"loss": 0.4241,
|
|
"step": 2728
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.756398133240215e-05,
|
|
"loss": 0.4041,
|
|
"step": 2730
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.756838380958435e-05,
|
|
"loss": 0.4056,
|
|
"step": 2732
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.757278306504859e-05,
|
|
"loss": 0.4344,
|
|
"step": 2734
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 4.757717910350673e-05,
|
|
"loss": 0.4894,
|
|
"step": 2736
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.7581571929660265e-05,
|
|
"loss": 0.4012,
|
|
"step": 2738
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.7585961548200396e-05,
|
|
"loss": 0.4019,
|
|
"step": 2740
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.759034796380809e-05,
|
|
"loss": 0.3726,
|
|
"step": 2742
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.759473118115401e-05,
|
|
"loss": 0.3984,
|
|
"step": 2744
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.759911120489868e-05,
|
|
"loss": 0.433,
|
|
"step": 2746
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.760348803969239e-05,
|
|
"loss": 0.3642,
|
|
"step": 2748
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.7607861690175316e-05,
|
|
"loss": 0.3894,
|
|
"step": 2750
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.761223216097749e-05,
|
|
"loss": 0.4242,
|
|
"step": 2752
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.7616599456718876e-05,
|
|
"loss": 0.3702,
|
|
"step": 2754
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.762096358200935e-05,
|
|
"loss": 0.3887,
|
|
"step": 2756
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.762532454144877e-05,
|
|
"loss": 0.451,
|
|
"step": 2758
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.762968233962699e-05,
|
|
"loss": 0.4139,
|
|
"step": 2760
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.7634036981123904e-05,
|
|
"loss": 0.4099,
|
|
"step": 2762
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.763838847050942e-05,
|
|
"loss": 0.4015,
|
|
"step": 2764
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.7642736812343576e-05,
|
|
"loss": 0.411,
|
|
"step": 2766
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.7647082011176495e-05,
|
|
"loss": 0.4089,
|
|
"step": 2768
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.765142407154845e-05,
|
|
"loss": 0.394,
|
|
"step": 2770
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.7655762997989866e-05,
|
|
"loss": 0.4266,
|
|
"step": 2772
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.7660098795021386e-05,
|
|
"loss": 0.3978,
|
|
"step": 2774
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.7664431467153885e-05,
|
|
"loss": 0.4335,
|
|
"step": 2776
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.7668761018888455e-05,
|
|
"loss": 0.4225,
|
|
"step": 2778
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.767308745471651e-05,
|
|
"loss": 0.4115,
|
|
"step": 2780
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.767741077911976e-05,
|
|
"loss": 0.3983,
|
|
"step": 2782
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.768173099657024e-05,
|
|
"loss": 0.4031,
|
|
"step": 2784
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.768604811153036e-05,
|
|
"loss": 0.411,
|
|
"step": 2786
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 4.769036212845293e-05,
|
|
"loss": 0.3978,
|
|
"step": 2788
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.7694673051781166e-05,
|
|
"loss": 0.4274,
|
|
"step": 2790
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.769898088594875e-05,
|
|
"loss": 0.4025,
|
|
"step": 2792
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.770328563537982e-05,
|
|
"loss": 0.3878,
|
|
"step": 2794
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.7707587304489034e-05,
|
|
"loss": 0.3693,
|
|
"step": 2796
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.771188589768156e-05,
|
|
"loss": 0.3733,
|
|
"step": 2798
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.771618141935315e-05,
|
|
"loss": 0.4188,
|
|
"step": 2800
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.7720473873890124e-05,
|
|
"loss": 0.4026,
|
|
"step": 2802
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.7724763265669403e-05,
|
|
"loss": 0.4135,
|
|
"step": 2804
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.772904959905856e-05,
|
|
"loss": 0.3612,
|
|
"step": 2806
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.7733332878415836e-05,
|
|
"loss": 0.4161,
|
|
"step": 2808
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.773761310809015e-05,
|
|
"loss": 0.3889,
|
|
"step": 2810
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.774189029242115e-05,
|
|
"loss": 0.4067,
|
|
"step": 2812
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.77461644357392e-05,
|
|
"loss": 0.4443,
|
|
"step": 2814
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.775043554236549e-05,
|
|
"loss": 0.4293,
|
|
"step": 2816
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.775470361661194e-05,
|
|
"loss": 0.4059,
|
|
"step": 2818
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.775896866278135e-05,
|
|
"loss": 0.3992,
|
|
"step": 2820
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.7763230685167315e-05,
|
|
"loss": 0.4055,
|
|
"step": 2822
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.776748968805436e-05,
|
|
"loss": 0.4156,
|
|
"step": 2824
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.777174567571787e-05,
|
|
"loss": 0.4089,
|
|
"step": 2826
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.7775998652424166e-05,
|
|
"loss": 0.3756,
|
|
"step": 2828
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.7780248622430525e-05,
|
|
"loss": 0.3907,
|
|
"step": 2830
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.7784495589985196e-05,
|
|
"loss": 0.3971,
|
|
"step": 2832
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.7788739559327445e-05,
|
|
"loss": 0.4442,
|
|
"step": 2834
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.7792980534687535e-05,
|
|
"loss": 0.3779,
|
|
"step": 2836
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 4.779721852028682e-05,
|
|
"loss": 0.4062,
|
|
"step": 2838
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.780145352033771e-05,
|
|
"loss": 0.4148,
|
|
"step": 2840
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.780568553904372e-05,
|
|
"loss": 0.4379,
|
|
"step": 2842
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.780991458059949e-05,
|
|
"loss": 0.3979,
|
|
"step": 2844
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.7814140649190835e-05,
|
|
"loss": 0.387,
|
|
"step": 2846
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.7818363748994726e-05,
|
|
"loss": 0.3877,
|
|
"step": 2848
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.7822583884179345e-05,
|
|
"loss": 0.407,
|
|
"step": 2850
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.782680105890411e-05,
|
|
"loss": 0.4132,
|
|
"step": 2852
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.783101527731967e-05,
|
|
"loss": 0.4321,
|
|
"step": 2854
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.7835226543567975e-05,
|
|
"loss": 0.3734,
|
|
"step": 2856
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.783943486178226e-05,
|
|
"loss": 0.3749,
|
|
"step": 2858
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.78436402360871e-05,
|
|
"loss": 0.4146,
|
|
"step": 2860
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.784784267059841e-05,
|
|
"loss": 0.4484,
|
|
"step": 2862
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.785204216942348e-05,
|
|
"loss": 0.4349,
|
|
"step": 2864
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.7856238736661e-05,
|
|
"loss": 0.4189,
|
|
"step": 2866
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.786043237640107e-05,
|
|
"loss": 0.4371,
|
|
"step": 2868
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.7864623092725265e-05,
|
|
"loss": 0.41,
|
|
"step": 2870
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.7868810889706615e-05,
|
|
"loss": 0.4078,
|
|
"step": 2872
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.7872995771409636e-05,
|
|
"loss": 0.4635,
|
|
"step": 2874
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.7877177741890355e-05,
|
|
"loss": 0.3872,
|
|
"step": 2876
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.788135680519638e-05,
|
|
"loss": 0.4232,
|
|
"step": 2878
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.788553296536684e-05,
|
|
"loss": 0.388,
|
|
"step": 2880
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.7889706226432466e-05,
|
|
"loss": 0.3853,
|
|
"step": 2882
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.7893876592415616e-05,
|
|
"loss": 0.4312,
|
|
"step": 2884
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.789804406733026e-05,
|
|
"loss": 0.4117,
|
|
"step": 2886
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.790220865518202e-05,
|
|
"loss": 0.4213,
|
|
"step": 2888
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 4.790637035996826e-05,
|
|
"loss": 0.4132,
|
|
"step": 2890
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.791052918567796e-05,
|
|
"loss": 0.4235,
|
|
"step": 2892
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.791468513629187e-05,
|
|
"loss": 0.3869,
|
|
"step": 2894
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.791883821578252e-05,
|
|
"loss": 0.3975,
|
|
"step": 2896
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.792298842811416e-05,
|
|
"loss": 0.4282,
|
|
"step": 2898
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.792713577724285e-05,
|
|
"loss": 0.4377,
|
|
"step": 2900
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.79312802671165e-05,
|
|
"loss": 0.3864,
|
|
"step": 2902
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.793542190167482e-05,
|
|
"loss": 0.4462,
|
|
"step": 2904
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.793956068484938e-05,
|
|
"loss": 0.3818,
|
|
"step": 2906
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.794369662056369e-05,
|
|
"loss": 0.4175,
|
|
"step": 2908
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.79478297127331e-05,
|
|
"loss": 0.4443,
|
|
"step": 2910
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.7951959965264935e-05,
|
|
"loss": 0.3994,
|
|
"step": 2912
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.795608738205846e-05,
|
|
"loss": 0.4163,
|
|
"step": 2914
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.7960211967004904e-05,
|
|
"loss": 0.4128,
|
|
"step": 2916
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.7964333723987494e-05,
|
|
"loss": 0.3932,
|
|
"step": 2918
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.796845265688149e-05,
|
|
"loss": 0.3847,
|
|
"step": 2920
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.7972568769554194e-05,
|
|
"loss": 0.3859,
|
|
"step": 2922
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.797668206586493e-05,
|
|
"loss": 0.4163,
|
|
"step": 2924
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.798079254966516e-05,
|
|
"loss": 0.3693,
|
|
"step": 2926
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.7984900224798406e-05,
|
|
"loss": 0.3818,
|
|
"step": 2928
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.7989005095100344e-05,
|
|
"loss": 0.4208,
|
|
"step": 2930
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.7993107164398794e-05,
|
|
"loss": 0.4417,
|
|
"step": 2932
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.7997206436513725e-05,
|
|
"loss": 0.412,
|
|
"step": 2934
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.800130291525733e-05,
|
|
"loss": 0.4061,
|
|
"step": 2936
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.8005396604433985e-05,
|
|
"loss": 0.4526,
|
|
"step": 2938
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.800948750784032e-05,
|
|
"loss": 0.4037,
|
|
"step": 2940
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 4.80135756292652e-05,
|
|
"loss": 0.4137,
|
|
"step": 2942
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.801766097248977e-05,
|
|
"loss": 0.407,
|
|
"step": 2944
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.8021743541287513e-05,
|
|
"loss": 0.4387,
|
|
"step": 2946
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.8025823339424154e-05,
|
|
"loss": 0.4077,
|
|
"step": 2948
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.8029900370657815e-05,
|
|
"loss": 0.381,
|
|
"step": 2950
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.803397463873896e-05,
|
|
"loss": 0.3995,
|
|
"step": 2952
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.803804614741041e-05,
|
|
"loss": 0.3993,
|
|
"step": 2954
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.804211490040743e-05,
|
|
"loss": 0.4491,
|
|
"step": 2956
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.804618090145768e-05,
|
|
"loss": 0.402,
|
|
"step": 2958
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.8050244154281254e-05,
|
|
"loss": 0.424,
|
|
"step": 2960
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.805430466259072e-05,
|
|
"loss": 0.3857,
|
|
"step": 2962
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.805836243009113e-05,
|
|
"loss": 0.4331,
|
|
"step": 2964
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.806241746048004e-05,
|
|
"loss": 0.3931,
|
|
"step": 2966
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.806646975744751e-05,
|
|
"loss": 0.3792,
|
|
"step": 2968
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.8070519324676164e-05,
|
|
"loss": 0.3853,
|
|
"step": 2970
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.8074566165841186e-05,
|
|
"loss": 0.3901,
|
|
"step": 2972
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.807861028461032e-05,
|
|
"loss": 0.3851,
|
|
"step": 2974
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.808265168464395e-05,
|
|
"loss": 0.3977,
|
|
"step": 2976
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.8086690369595055e-05,
|
|
"loss": 0.3613,
|
|
"step": 2978
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.8090726343109264e-05,
|
|
"loss": 0.4363,
|
|
"step": 2980
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.8094759608824874e-05,
|
|
"loss": 0.4467,
|
|
"step": 2982
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.809879017037285e-05,
|
|
"loss": 0.4358,
|
|
"step": 2984
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.810281803137687e-05,
|
|
"loss": 0.3885,
|
|
"step": 2986
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.810684319545334e-05,
|
|
"loss": 0.4094,
|
|
"step": 2988
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.81108656662114e-05,
|
|
"loss": 0.3757,
|
|
"step": 2990
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 4.811488544725292e-05,
|
|
"loss": 0.4695,
|
|
"step": 2992
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.811890254217261e-05,
|
|
"loss": 0.3911,
|
|
"step": 2994
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.812291695455793e-05,
|
|
"loss": 0.3876,
|
|
"step": 2996
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.812692868798916e-05,
|
|
"loss": 0.3719,
|
|
"step": 2998
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.813093774603945e-05,
|
|
"loss": 0.3934,
|
|
"step": 3000
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.813494413227479e-05,
|
|
"loss": 0.413,
|
|
"step": 3002
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.8138947850254027e-05,
|
|
"loss": 0.4293,
|
|
"step": 3004
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.8142948903528914e-05,
|
|
"loss": 0.3813,
|
|
"step": 3006
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.814694729564413e-05,
|
|
"loss": 0.4197,
|
|
"step": 3008
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.815094303013728e-05,
|
|
"loss": 0.4071,
|
|
"step": 3010
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.815493611053889e-05,
|
|
"loss": 0.3986,
|
|
"step": 3012
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.815892654037251e-05,
|
|
"loss": 0.376,
|
|
"step": 3014
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.816291432315465e-05,
|
|
"loss": 0.4479,
|
|
"step": 3016
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.8166899462394796e-05,
|
|
"loss": 0.4181,
|
|
"step": 3018
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.81708819615955e-05,
|
|
"loss": 0.4183,
|
|
"step": 3020
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.817486182425235e-05,
|
|
"loss": 0.3899,
|
|
"step": 3022
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.8178839053854e-05,
|
|
"loss": 0.4025,
|
|
"step": 3024
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.818281365388217e-05,
|
|
"loss": 0.3897,
|
|
"step": 3026
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.8186785627811676e-05,
|
|
"loss": 0.4105,
|
|
"step": 3028
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.819075497911046e-05,
|
|
"loss": 0.4178,
|
|
"step": 3030
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.819472171123961e-05,
|
|
"loss": 0.3859,
|
|
"step": 3032
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.819868582765337e-05,
|
|
"loss": 0.4344,
|
|
"step": 3034
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.82026473317991e-05,
|
|
"loss": 0.4353,
|
|
"step": 3036
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.820660622711743e-05,
|
|
"loss": 0.405,
|
|
"step": 3038
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.821056251704213e-05,
|
|
"loss": 0.4002,
|
|
"step": 3040
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.8214516205000236e-05,
|
|
"loss": 0.4255,
|
|
"step": 3042
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 4.821846729441202e-05,
|
|
"loss": 0.4199,
|
|
"step": 3044
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.822241578869098e-05,
|
|
"loss": 0.3708,
|
|
"step": 3046
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.8226361691243954e-05,
|
|
"loss": 0.4027,
|
|
"step": 3048
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.8230305005471025e-05,
|
|
"loss": 0.4008,
|
|
"step": 3050
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.8234245734765604e-05,
|
|
"loss": 0.4001,
|
|
"step": 3052
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.8238183882514434e-05,
|
|
"loss": 0.4006,
|
|
"step": 3054
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.824211945209763e-05,
|
|
"loss": 0.3896,
|
|
"step": 3056
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.824605244688863e-05,
|
|
"loss": 0.3971,
|
|
"step": 3058
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.824998287025427e-05,
|
|
"loss": 0.4041,
|
|
"step": 3060
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.825391072555482e-05,
|
|
"loss": 0.3986,
|
|
"step": 3062
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.825783601614392e-05,
|
|
"loss": 0.4429,
|
|
"step": 3064
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.8261758745368656e-05,
|
|
"loss": 0.3749,
|
|
"step": 3066
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.826567891656961e-05,
|
|
"loss": 0.3622,
|
|
"step": 3068
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.826959653308076e-05,
|
|
"loss": 0.3814,
|
|
"step": 3070
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.8273511598229624e-05,
|
|
"loss": 0.3777,
|
|
"step": 3072
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.827742411533721e-05,
|
|
"loss": 0.3691,
|
|
"step": 3074
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.828133408771805e-05,
|
|
"loss": 0.372,
|
|
"step": 3076
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.8285241518680194e-05,
|
|
"loss": 0.4142,
|
|
"step": 3078
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.828914641152526e-05,
|
|
"loss": 0.4372,
|
|
"step": 3080
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.829304876954845e-05,
|
|
"loss": 0.4175,
|
|
"step": 3082
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.8296948596038524e-05,
|
|
"loss": 0.4127,
|
|
"step": 3084
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.830084589427785e-05,
|
|
"loss": 0.3911,
|
|
"step": 3086
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.830474066754244e-05,
|
|
"loss": 0.3997,
|
|
"step": 3088
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.830863291910191e-05,
|
|
"loss": 0.4379,
|
|
"step": 3090
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.831252265221957e-05,
|
|
"loss": 0.367,
|
|
"step": 3092
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 4.831640987015238e-05,
|
|
"loss": 0.3997,
|
|
"step": 3094
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.832029457615096e-05,
|
|
"loss": 0.3838,
|
|
"step": 3096
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.832417677345967e-05,
|
|
"loss": 0.3776,
|
|
"step": 3098
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.832805646531657e-05,
|
|
"loss": 0.4003,
|
|
"step": 3100
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.833193365495346e-05,
|
|
"loss": 0.3782,
|
|
"step": 3102
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.833580834559589e-05,
|
|
"loss": 0.4391,
|
|
"step": 3104
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.833968054046318e-05,
|
|
"loss": 0.4045,
|
|
"step": 3106
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.834355024276841e-05,
|
|
"loss": 0.4297,
|
|
"step": 3108
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.8347417455718515e-05,
|
|
"loss": 0.444,
|
|
"step": 3110
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.8351282182514184e-05,
|
|
"loss": 0.4196,
|
|
"step": 3112
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.835514442634996e-05,
|
|
"loss": 0.4187,
|
|
"step": 3114
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.835900419041424e-05,
|
|
"loss": 0.3857,
|
|
"step": 3116
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.83628614778893e-05,
|
|
"loss": 0.3938,
|
|
"step": 3118
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.836671629195124e-05,
|
|
"loss": 0.3876,
|
|
"step": 3120
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.83705686357701e-05,
|
|
"loss": 0.3884,
|
|
"step": 3122
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.837441851250982e-05,
|
|
"loss": 0.4664,
|
|
"step": 3124
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.837826592532826e-05,
|
|
"loss": 0.4257,
|
|
"step": 3126
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.838211087737721e-05,
|
|
"loss": 0.4503,
|
|
"step": 3128
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.8385953371802455e-05,
|
|
"loss": 0.3718,
|
|
"step": 3130
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.838979341174371e-05,
|
|
"loss": 0.4459,
|
|
"step": 3132
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.8393631000334675e-05,
|
|
"loss": 0.3763,
|
|
"step": 3134
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.839746614070311e-05,
|
|
"loss": 0.4042,
|
|
"step": 3136
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.840129883597071e-05,
|
|
"loss": 0.4214,
|
|
"step": 3138
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.840512908925328e-05,
|
|
"loss": 0.3693,
|
|
"step": 3140
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.840895690366061e-05,
|
|
"loss": 0.4092,
|
|
"step": 3142
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.8412782282296595e-05,
|
|
"loss": 0.366,
|
|
"step": 3144
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 4.841660522825922e-05,
|
|
"loss": 0.4133,
|
|
"step": 3146
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.8420425744640504e-05,
|
|
"loss": 0.3981,
|
|
"step": 3148
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.842424383452661e-05,
|
|
"loss": 0.4041,
|
|
"step": 3150
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.842805950099785e-05,
|
|
"loss": 0.4189,
|
|
"step": 3152
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.843187274712864e-05,
|
|
"loss": 0.3711,
|
|
"step": 3154
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.843568357598755e-05,
|
|
"loss": 0.3821,
|
|
"step": 3156
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.843949199063732e-05,
|
|
"loss": 0.4004,
|
|
"step": 3158
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.84432979941349e-05,
|
|
"loss": 0.4272,
|
|
"step": 3160
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.844710158953139e-05,
|
|
"loss": 0.416,
|
|
"step": 3162
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.845090277987215e-05,
|
|
"loss": 0.3729,
|
|
"step": 3164
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.845470156819674e-05,
|
|
"loss": 0.4313,
|
|
"step": 3166
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.845849795753895e-05,
|
|
"loss": 0.4,
|
|
"step": 3168
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.846229195092687e-05,
|
|
"loss": 0.4362,
|
|
"step": 3170
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.846608355138281e-05,
|
|
"loss": 0.4181,
|
|
"step": 3172
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.8469872761923393e-05,
|
|
"loss": 0.3868,
|
|
"step": 3174
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.847365958555956e-05,
|
|
"loss": 0.3761,
|
|
"step": 3176
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.84774440252965e-05,
|
|
"loss": 0.3865,
|
|
"step": 3178
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.848122608413381e-05,
|
|
"loss": 0.4018,
|
|
"step": 3180
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.848500576506538e-05,
|
|
"loss": 0.369,
|
|
"step": 3182
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.848878307107945e-05,
|
|
"loss": 0.3905,
|
|
"step": 3184
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.849255800515866e-05,
|
|
"loss": 0.3936,
|
|
"step": 3186
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.8496330570280037e-05,
|
|
"loss": 0.4216,
|
|
"step": 3188
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.850010076941497e-05,
|
|
"loss": 0.3644,
|
|
"step": 3190
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.850386860552929e-05,
|
|
"loss": 0.3872,
|
|
"step": 3192
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.850763408158326e-05,
|
|
"loss": 0.3655,
|
|
"step": 3194
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.851139720053157e-05,
|
|
"loss": 0.4171,
|
|
"step": 3196
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 4.8515157965323357e-05,
|
|
"loss": 0.3945,
|
|
"step": 3198
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.851891637890224e-05,
|
|
"loss": 0.4045,
|
|
"step": 3200
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.852267244420632e-05,
|
|
"loss": 0.4033,
|
|
"step": 3202
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.85264261641682e-05,
|
|
"loss": 0.3882,
|
|
"step": 3204
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.8530177541714957e-05,
|
|
"loss": 0.4162,
|
|
"step": 3206
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.853392657976825e-05,
|
|
"loss": 0.4007,
|
|
"step": 3208
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.853767328124422e-05,
|
|
"loss": 0.3839,
|
|
"step": 3210
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.854141764905361e-05,
|
|
"loss": 0.4059,
|
|
"step": 3212
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.854515968610168e-05,
|
|
"loss": 0.3991,
|
|
"step": 3214
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.85488993952883e-05,
|
|
"loss": 0.3956,
|
|
"step": 3216
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.8552636779507906e-05,
|
|
"loss": 0.4087,
|
|
"step": 3218
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.8556371841649554e-05,
|
|
"loss": 0.3798,
|
|
"step": 3220
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.8560104584596946e-05,
|
|
"loss": 0.3881,
|
|
"step": 3222
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.856383501122836e-05,
|
|
"loss": 0.3878,
|
|
"step": 3224
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.8567563124416744e-05,
|
|
"loss": 0.3877,
|
|
"step": 3226
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.857128892702973e-05,
|
|
"loss": 0.3706,
|
|
"step": 3228
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.8575012421929575e-05,
|
|
"loss": 0.4218,
|
|
"step": 3230
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.8578733611973254e-05,
|
|
"loss": 0.4022,
|
|
"step": 3232
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.8582452500012425e-05,
|
|
"loss": 0.4585,
|
|
"step": 3234
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.858616908889347e-05,
|
|
"loss": 0.3994,
|
|
"step": 3236
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.8589883381457494e-05,
|
|
"loss": 0.4314,
|
|
"step": 3238
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.859359538054031e-05,
|
|
"loss": 0.3939,
|
|
"step": 3240
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.85973050889725e-05,
|
|
"loss": 0.3939,
|
|
"step": 3242
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.860101250957945e-05,
|
|
"loss": 0.4202,
|
|
"step": 3244
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.8604717645181234e-05,
|
|
"loss": 0.4175,
|
|
"step": 3246
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 4.86084204985928e-05,
|
|
"loss": 0.3984,
|
|
"step": 3248
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.861212107262385e-05,
|
|
"loss": 0.3787,
|
|
"step": 3250
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.861581937007892e-05,
|
|
"loss": 0.408,
|
|
"step": 3252
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.8619515393757366e-05,
|
|
"loss": 0.4028,
|
|
"step": 3254
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.862320914645336e-05,
|
|
"loss": 0.4319,
|
|
"step": 3256
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.862690063095599e-05,
|
|
"loss": 0.3992,
|
|
"step": 3258
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.863058985004913e-05,
|
|
"loss": 0.4329,
|
|
"step": 3260
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.86342768065116e-05,
|
|
"loss": 0.39,
|
|
"step": 3262
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.8637961503117056e-05,
|
|
"loss": 0.4303,
|
|
"step": 3264
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.864164394263411e-05,
|
|
"loss": 0.3956,
|
|
"step": 3266
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.8645324127826256e-05,
|
|
"loss": 0.3866,
|
|
"step": 3268
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.864900206145191e-05,
|
|
"loss": 0.3842,
|
|
"step": 3270
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.865267774626444e-05,
|
|
"loss": 0.4108,
|
|
"step": 3272
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.865635118501217e-05,
|
|
"loss": 0.4263,
|
|
"step": 3274
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.866002238043841e-05,
|
|
"loss": 0.3641,
|
|
"step": 3276
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.866369133528137e-05,
|
|
"loss": 0.3901,
|
|
"step": 3278
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.866735805227436e-05,
|
|
"loss": 0.4167,
|
|
"step": 3280
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.867102253414559e-05,
|
|
"loss": 0.3777,
|
|
"step": 3282
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.867468478361835e-05,
|
|
"loss": 0.3983,
|
|
"step": 3284
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.8678344803410925e-05,
|
|
"loss": 0.3602,
|
|
"step": 3286
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.868200259623666e-05,
|
|
"loss": 0.4023,
|
|
"step": 3288
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.868565816480391e-05,
|
|
"loss": 0.4185,
|
|
"step": 3290
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.8689311511816147e-05,
|
|
"loss": 0.3667,
|
|
"step": 3292
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.8692962639971866e-05,
|
|
"loss": 0.3777,
|
|
"step": 3294
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.8696611551964704e-05,
|
|
"loss": 0.3669,
|
|
"step": 3296
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.870025825048332e-05,
|
|
"loss": 0.4338,
|
|
"step": 3298
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 4.8703902738211574e-05,
|
|
"loss": 0.3894,
|
|
"step": 3300
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.8707545017828353e-05,
|
|
"loss": 0.4463,
|
|
"step": 3302
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.871118509200776e-05,
|
|
"loss": 0.4047,
|
|
"step": 3304
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.8714822963419e-05,
|
|
"loss": 0.3939,
|
|
"step": 3306
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.871845863472643e-05,
|
|
"loss": 0.3572,
|
|
"step": 3308
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.872209210858961e-05,
|
|
"loss": 0.3794,
|
|
"step": 3310
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.872572338766324e-05,
|
|
"loss": 0.4373,
|
|
"step": 3312
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.872935247459725e-05,
|
|
"loss": 0.379,
|
|
"step": 3314
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.8732979372036744e-05,
|
|
"loss": 0.3798,
|
|
"step": 3316
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.873660408262206e-05,
|
|
"loss": 0.3748,
|
|
"step": 3318
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.8740226608988746e-05,
|
|
"loss": 0.4086,
|
|
"step": 3320
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.874384695376761e-05,
|
|
"loss": 0.3993,
|
|
"step": 3322
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.87474651195847e-05,
|
|
"loss": 0.3983,
|
|
"step": 3324
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.87510811090613e-05,
|
|
"loss": 0.4302,
|
|
"step": 3326
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.8754694924814024e-05,
|
|
"loss": 0.3797,
|
|
"step": 3328
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.8758306569454714e-05,
|
|
"loss": 0.4104,
|
|
"step": 3330
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.876191604559054e-05,
|
|
"loss": 0.417,
|
|
"step": 3332
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.876552335582396e-05,
|
|
"loss": 0.3956,
|
|
"step": 3334
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.8769128502752764e-05,
|
|
"loss": 0.4031,
|
|
"step": 3336
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.877273148897006e-05,
|
|
"loss": 0.3797,
|
|
"step": 3338
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.8776332317064324e-05,
|
|
"loss": 0.4144,
|
|
"step": 3340
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.8779930989619315e-05,
|
|
"loss": 0.3842,
|
|
"step": 3342
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.878352750921425e-05,
|
|
"loss": 0.3904,
|
|
"step": 3344
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.878712187842364e-05,
|
|
"loss": 0.3912,
|
|
"step": 3346
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.879071409981742e-05,
|
|
"loss": 0.4167,
|
|
"step": 3348
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 4.87943041759609e-05,
|
|
"loss": 0.375,
|
|
"step": 3350
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.8797892109414845e-05,
|
|
"loss": 0.4077,
|
|
"step": 3352
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.8801477902735366e-05,
|
|
"loss": 0.3649,
|
|
"step": 3354
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.880506155847405e-05,
|
|
"loss": 0.3756,
|
|
"step": 3356
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.88086430791779e-05,
|
|
"loss": 0.4221,
|
|
"step": 3358
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.88122224673894e-05,
|
|
"loss": 0.4133,
|
|
"step": 3360
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.881579972564648e-05,
|
|
"loss": 0.3989,
|
|
"step": 3362
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.8819374856482505e-05,
|
|
"loss": 0.4462,
|
|
"step": 3364
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.882294786242639e-05,
|
|
"loss": 0.3637,
|
|
"step": 3366
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.8826518746002496e-05,
|
|
"loss": 0.4392,
|
|
"step": 3368
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.88300875097307e-05,
|
|
"loss": 0.3851,
|
|
"step": 3370
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.8833654156126394e-05,
|
|
"loss": 0.4038,
|
|
"step": 3372
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.883721868770052e-05,
|
|
"loss": 0.4036,
|
|
"step": 3374
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.88407811069595e-05,
|
|
"loss": 0.4258,
|
|
"step": 3376
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.8844341416405356e-05,
|
|
"loss": 0.437,
|
|
"step": 3378
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.884789961853564e-05,
|
|
"loss": 0.3942,
|
|
"step": 3380
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.885145571584348e-05,
|
|
"loss": 0.4473,
|
|
"step": 3382
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.88550097108176e-05,
|
|
"loss": 0.3876,
|
|
"step": 3384
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.885856160594226e-05,
|
|
"loss": 0.3711,
|
|
"step": 3386
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.886211140369738e-05,
|
|
"loss": 0.3963,
|
|
"step": 3388
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.8865659106558456e-05,
|
|
"loss": 0.4381,
|
|
"step": 3390
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.886920471699659e-05,
|
|
"loss": 0.4218,
|
|
"step": 3392
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.887274823747857e-05,
|
|
"loss": 0.4178,
|
|
"step": 3394
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.8876289670466776e-05,
|
|
"loss": 0.3909,
|
|
"step": 3396
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.8879829018419245e-05,
|
|
"loss": 0.3749,
|
|
"step": 3398
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.8883366283789675e-05,
|
|
"loss": 0.4194,
|
|
"step": 3400
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 4.888690146902746e-05,
|
|
"loss": 0.4027,
|
|
"step": 3402
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.889043457657765e-05,
|
|
"loss": 0.4155,
|
|
"step": 3404
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.8893965608881e-05,
|
|
"loss": 0.3704,
|
|
"step": 3406
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.889749456837396e-05,
|
|
"loss": 0.408,
|
|
"step": 3408
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.8901021457488684e-05,
|
|
"loss": 0.4341,
|
|
"step": 3410
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.8904546278653055e-05,
|
|
"loss": 0.3899,
|
|
"step": 3412
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.89080690342907e-05,
|
|
"loss": 0.3979,
|
|
"step": 3414
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.891158972682097e-05,
|
|
"loss": 0.4318,
|
|
"step": 3416
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.891510835865898e-05,
|
|
"loss": 0.4291,
|
|
"step": 3418
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.89186249322156e-05,
|
|
"loss": 0.4141,
|
|
"step": 3420
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.892213944989747e-05,
|
|
"loss": 0.4003,
|
|
"step": 3422
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.8925651914107004e-05,
|
|
"loss": 0.4149,
|
|
"step": 3424
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.892916232724245e-05,
|
|
"loss": 0.3872,
|
|
"step": 3426
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.8932670691697795e-05,
|
|
"loss": 0.3411,
|
|
"step": 3428
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.893617700986288e-05,
|
|
"loss": 0.4247,
|
|
"step": 3430
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.893968128412335e-05,
|
|
"loss": 0.382,
|
|
"step": 3432
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.89431835168607e-05,
|
|
"loss": 0.403,
|
|
"step": 3434
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.8946683710452215e-05,
|
|
"loss": 0.3948,
|
|
"step": 3436
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.89501818672711e-05,
|
|
"loss": 0.3973,
|
|
"step": 3438
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.895367798968637e-05,
|
|
"loss": 0.3917,
|
|
"step": 3440
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.8957172080062906e-05,
|
|
"loss": 0.4181,
|
|
"step": 3442
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.896066414076152e-05,
|
|
"loss": 0.3934,
|
|
"step": 3444
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.8964154174138856e-05,
|
|
"loss": 0.4128,
|
|
"step": 3446
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.896764218254748e-05,
|
|
"loss": 0.3678,
|
|
"step": 3448
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.8971128168335865e-05,
|
|
"loss": 0.4329,
|
|
"step": 3450
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 4.8974612133848393e-05,
|
|
"loss": 0.425,
|
|
"step": 3452
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.897809408142539e-05,
|
|
"loss": 0.3719,
|
|
"step": 3454
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.898157401340309e-05,
|
|
"loss": 0.3765,
|
|
"step": 3456
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.89850519321137e-05,
|
|
"loss": 0.395,
|
|
"step": 3458
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.8988527839885366e-05,
|
|
"loss": 0.3365,
|
|
"step": 3460
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.899200173904221e-05,
|
|
"loss": 0.4196,
|
|
"step": 3462
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.8995473631904287e-05,
|
|
"loss": 0.406,
|
|
"step": 3464
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.89989435207877e-05,
|
|
"loss": 0.4048,
|
|
"step": 3466
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.90024114080045e-05,
|
|
"loss": 0.3851,
|
|
"step": 3468
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.900587729586275e-05,
|
|
"loss": 0.3721,
|
|
"step": 3470
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.9009341186666516e-05,
|
|
"loss": 0.4286,
|
|
"step": 3472
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.9012803082715906e-05,
|
|
"loss": 0.3818,
|
|
"step": 3474
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.901626298630701e-05,
|
|
"loss": 0.4165,
|
|
"step": 3476
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.901972089973201e-05,
|
|
"loss": 0.4021,
|
|
"step": 3478
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.9023176825279104e-05,
|
|
"loss": 0.407,
|
|
"step": 3480
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.902663076523257e-05,
|
|
"loss": 0.4199,
|
|
"step": 3482
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.90300827218727e-05,
|
|
"loss": 0.394,
|
|
"step": 3484
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.903353269747591e-05,
|
|
"loss": 0.4017,
|
|
"step": 3486
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.903698069431469e-05,
|
|
"loss": 0.4058,
|
|
"step": 3488
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.904042671465761e-05,
|
|
"loss": 0.4275,
|
|
"step": 3490
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.9043870760769355e-05,
|
|
"loss": 0.4005,
|
|
"step": 3492
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.90473128349107e-05,
|
|
"loss": 0.4022,
|
|
"step": 3494
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.905075293933854e-05,
|
|
"loss": 0.4384,
|
|
"step": 3496
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.905419107630593e-05,
|
|
"loss": 0.427,
|
|
"step": 3498
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.905762724806202e-05,
|
|
"loss": 0.4053,
|
|
"step": 3500
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.9061061456852137e-05,
|
|
"loss": 0.3997,
|
|
"step": 3502
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 4.9064493704917744e-05,
|
|
"loss": 0.3596,
|
|
"step": 3504
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.906792399449648e-05,
|
|
"loss": 0.4312,
|
|
"step": 3506
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.9071352327822135e-05,
|
|
"loss": 0.4362,
|
|
"step": 3508
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.907477870712471e-05,
|
|
"loss": 0.3997,
|
|
"step": 3510
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.907820313463036e-05,
|
|
"loss": 0.394,
|
|
"step": 3512
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.908162561256146e-05,
|
|
"loss": 0.3999,
|
|
"step": 3514
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.90850461431366e-05,
|
|
"loss": 0.4,
|
|
"step": 3516
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.908846472857056e-05,
|
|
"loss": 0.3811,
|
|
"step": 3518
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.909188137107436e-05,
|
|
"loss": 0.439,
|
|
"step": 3520
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.909529607285524e-05,
|
|
"loss": 0.3831,
|
|
"step": 3522
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.9098708836116705e-05,
|
|
"loss": 0.4285,
|
|
"step": 3524
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.910211966305848e-05,
|
|
"loss": 0.3862,
|
|
"step": 3526
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.910552855587656e-05,
|
|
"loss": 0.3837,
|
|
"step": 3528
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.910893551676323e-05,
|
|
"loss": 0.3939,
|
|
"step": 3530
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.9112340547907006e-05,
|
|
"loss": 0.4393,
|
|
"step": 3532
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.9115743651492714e-05,
|
|
"loss": 0.4227,
|
|
"step": 3534
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.911914482970147e-05,
|
|
"loss": 0.3938,
|
|
"step": 3536
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.912254408471068e-05,
|
|
"loss": 0.3961,
|
|
"step": 3538
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.9125941418694067e-05,
|
|
"loss": 0.4105,
|
|
"step": 3540
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.9129336833821675e-05,
|
|
"loss": 0.3884,
|
|
"step": 3542
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.913273033225986e-05,
|
|
"loss": 0.3737,
|
|
"step": 3544
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.913612191617132e-05,
|
|
"loss": 0.4211,
|
|
"step": 3546
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.91395115877151e-05,
|
|
"loss": 0.4116,
|
|
"step": 3548
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.9142899349046574e-05,
|
|
"loss": 0.4053,
|
|
"step": 3550
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.91462852023175e-05,
|
|
"loss": 0.3868,
|
|
"step": 3552
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.914966914967599e-05,
|
|
"loss": 0.4242,
|
|
"step": 3554
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 4.9153051193266526e-05,
|
|
"loss": 0.4031,
|
|
"step": 3556
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.9156431335229966e-05,
|
|
"loss": 0.3623,
|
|
"step": 3558
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.915980957770359e-05,
|
|
"loss": 0.3954,
|
|
"step": 3560
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.916318592282106e-05,
|
|
"loss": 0.3932,
|
|
"step": 3562
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.9166560372712415e-05,
|
|
"loss": 0.4083,
|
|
"step": 3564
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.916993292950417e-05,
|
|
"loss": 0.4072,
|
|
"step": 3566
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.917330359531919e-05,
|
|
"loss": 0.3842,
|
|
"step": 3568
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.9176672372276845e-05,
|
|
"loss": 0.3735,
|
|
"step": 3570
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.9180039262492895e-05,
|
|
"loss": 0.4138,
|
|
"step": 3572
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.9183404268079556e-05,
|
|
"loss": 0.3563,
|
|
"step": 3574
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.918676739114552e-05,
|
|
"loss": 0.3872,
|
|
"step": 3576
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.919012863379591e-05,
|
|
"loss": 0.372,
|
|
"step": 3578
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.919348799813234e-05,
|
|
"loss": 0.4264,
|
|
"step": 3580
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.919684548625292e-05,
|
|
"loss": 0.4238,
|
|
"step": 3582
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.9200201100252183e-05,
|
|
"loss": 0.38,
|
|
"step": 3584
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.9203554842221236e-05,
|
|
"loss": 0.4628,
|
|
"step": 3586
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.920690671424764e-05,
|
|
"loss": 0.3824,
|
|
"step": 3588
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.921025671841548e-05,
|
|
"loss": 0.4216,
|
|
"step": 3590
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.921360485680535e-05,
|
|
"loss": 0.4208,
|
|
"step": 3592
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.921695113149438e-05,
|
|
"loss": 0.3976,
|
|
"step": 3594
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.922029554455622e-05,
|
|
"loss": 0.4059,
|
|
"step": 3596
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.922363809806108e-05,
|
|
"loss": 0.604,
|
|
"step": 3598
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.92269787940757e-05,
|
|
"loss": 0.3684,
|
|
"step": 3600
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.9230317634663405e-05,
|
|
"loss": 0.4421,
|
|
"step": 3602
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.923365462188404e-05,
|
|
"loss": 0.4005,
|
|
"step": 3604
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 4.923698975779405e-05,
|
|
"loss": 0.3982,
|
|
"step": 3606
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.924032304444647e-05,
|
|
"loss": 0.423,
|
|
"step": 3608
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.92436544838909e-05,
|
|
"loss": 0.3925,
|
|
"step": 3610
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.924698407817352e-05,
|
|
"loss": 0.3643,
|
|
"step": 3612
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.9250311829337164e-05,
|
|
"loss": 0.4015,
|
|
"step": 3614
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.925363773942124e-05,
|
|
"loss": 0.3947,
|
|
"step": 3616
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.925696181046176e-05,
|
|
"loss": 0.3945,
|
|
"step": 3618
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.926028404449139e-05,
|
|
"loss": 0.4086,
|
|
"step": 3620
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.9263604443539416e-05,
|
|
"loss": 0.3989,
|
|
"step": 3622
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.926692300963176e-05,
|
|
"loss": 0.4363,
|
|
"step": 3624
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.927023974479098e-05,
|
|
"loss": 0.3964,
|
|
"step": 3626
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.9273554651036336e-05,
|
|
"loss": 0.4264,
|
|
"step": 3628
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.927686773038369e-05,
|
|
"loss": 0.3502,
|
|
"step": 3630
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.928017898484559e-05,
|
|
"loss": 0.3656,
|
|
"step": 3632
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.92834884164313e-05,
|
|
"loss": 0.3965,
|
|
"step": 3634
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.928679602714672e-05,
|
|
"loss": 0.3975,
|
|
"step": 3636
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.929010181899445e-05,
|
|
"loss": 0.3712,
|
|
"step": 3638
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.92934057939738e-05,
|
|
"loss": 0.3974,
|
|
"step": 3640
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.929670795408079e-05,
|
|
"loss": 0.4031,
|
|
"step": 3642
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.930000830130814e-05,
|
|
"loss": 0.3565,
|
|
"step": 3644
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.930330683764528e-05,
|
|
"loss": 0.3612,
|
|
"step": 3646
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.930660356507839e-05,
|
|
"loss": 0.379,
|
|
"step": 3648
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.930989848559037e-05,
|
|
"loss": 0.4302,
|
|
"step": 3650
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.931319160116086e-05,
|
|
"loss": 0.4396,
|
|
"step": 3652
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.931648291376627e-05,
|
|
"loss": 0.4396,
|
|
"step": 3654
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.931977242537975e-05,
|
|
"loss": 0.4349,
|
|
"step": 3656
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 4.9323060137971184e-05,
|
|
"loss": 0.3927,
|
|
"step": 3658
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.932634605350727e-05,
|
|
"loss": 0.3564,
|
|
"step": 3660
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.932963017395148e-05,
|
|
"loss": 0.3734,
|
|
"step": 3662
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.9332912501264045e-05,
|
|
"loss": 0.3574,
|
|
"step": 3664
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.9336193037402e-05,
|
|
"loss": 0.4202,
|
|
"step": 3666
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.933947178431917e-05,
|
|
"loss": 0.3741,
|
|
"step": 3668
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.93427487439662e-05,
|
|
"loss": 0.4147,
|
|
"step": 3670
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.9346023918290523e-05,
|
|
"loss": 0.4302,
|
|
"step": 3672
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.9349297309236425e-05,
|
|
"loss": 0.4192,
|
|
"step": 3674
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.935256891874499e-05,
|
|
"loss": 0.3798,
|
|
"step": 3676
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.9355838748754144e-05,
|
|
"loss": 0.3705,
|
|
"step": 3678
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.935910680119865e-05,
|
|
"loss": 0.4005,
|
|
"step": 3680
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.936237307801011e-05,
|
|
"loss": 0.3985,
|
|
"step": 3682
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.9365637581117004e-05,
|
|
"loss": 0.3878,
|
|
"step": 3684
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.936890031244465e-05,
|
|
"loss": 0.4365,
|
|
"step": 3686
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.937216127391523e-05,
|
|
"loss": 0.3853,
|
|
"step": 3688
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.937542046744782e-05,
|
|
"loss": 0.3916,
|
|
"step": 3690
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.937867789495836e-05,
|
|
"loss": 0.4246,
|
|
"step": 3692
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.9381933558359686e-05,
|
|
"loss": 0.3677,
|
|
"step": 3694
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.9385187459561514e-05,
|
|
"loss": 0.4066,
|
|
"step": 3696
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.9388439600470475e-05,
|
|
"loss": 0.4072,
|
|
"step": 3698
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.939168998299012e-05,
|
|
"loss": 0.4273,
|
|
"step": 3700
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.939493860902087e-05,
|
|
"loss": 0.4153,
|
|
"step": 3702
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.939818548046011e-05,
|
|
"loss": 0.3945,
|
|
"step": 3704
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.940143059920214e-05,
|
|
"loss": 0.4027,
|
|
"step": 3706
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 4.940467396713816e-05,
|
|
"loss": 0.3969,
|
|
"step": 3708
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.940791558615638e-05,
|
|
"loss": 0.3594,
|
|
"step": 3710
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.941115545814189e-05,
|
|
"loss": 0.3744,
|
|
"step": 3712
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.941439358497678e-05,
|
|
"loss": 0.4259,
|
|
"step": 3714
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.9417629968540066e-05,
|
|
"loss": 0.391,
|
|
"step": 3716
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.9420864610707754e-05,
|
|
"loss": 0.3755,
|
|
"step": 3718
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.942409751335282e-05,
|
|
"loss": 0.366,
|
|
"step": 3720
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.942732867834521e-05,
|
|
"loss": 0.3929,
|
|
"step": 3722
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.943055810755186e-05,
|
|
"loss": 0.3722,
|
|
"step": 3724
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.9433785802836706e-05,
|
|
"loss": 0.4021,
|
|
"step": 3726
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.943701176606069e-05,
|
|
"loss": 0.412,
|
|
"step": 3728
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.944023599908172e-05,
|
|
"loss": 0.3921,
|
|
"step": 3730
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.9443458503754767e-05,
|
|
"loss": 0.3949,
|
|
"step": 3732
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.9446679281931784e-05,
|
|
"loss": 0.4125,
|
|
"step": 3734
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.944989833546178e-05,
|
|
"loss": 0.3617,
|
|
"step": 3736
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.945311566619075e-05,
|
|
"loss": 0.4385,
|
|
"step": 3738
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.94563312759618e-05,
|
|
"loss": 0.3995,
|
|
"step": 3740
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.9459545166614985e-05,
|
|
"loss": 0.4536,
|
|
"step": 3742
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.94627573399875e-05,
|
|
"loss": 0.3788,
|
|
"step": 3744
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.9465967797913514e-05,
|
|
"loss": 0.3781,
|
|
"step": 3746
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.946917654222435e-05,
|
|
"loss": 0.3726,
|
|
"step": 3748
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.947238357474832e-05,
|
|
"loss": 0.4052,
|
|
"step": 3750
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.9475588897310856e-05,
|
|
"loss": 0.3986,
|
|
"step": 3752
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.9478792511734456e-05,
|
|
"loss": 0.3832,
|
|
"step": 3754
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.9481994419838706e-05,
|
|
"loss": 0.4008,
|
|
"step": 3756
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.9485194623440293e-05,
|
|
"loss": 0.3939,
|
|
"step": 3758
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 4.9488393124353e-05,
|
|
"loss": 0.3995,
|
|
"step": 3760
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.949158992438772e-05,
|
|
"loss": 0.3994,
|
|
"step": 3762
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.949478502535243e-05,
|
|
"loss": 0.4288,
|
|
"step": 3764
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.949797842905229e-05,
|
|
"loss": 0.3783,
|
|
"step": 3766
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.9501170137289524e-05,
|
|
"loss": 0.4587,
|
|
"step": 3768
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.9504360151863504e-05,
|
|
"loss": 0.4019,
|
|
"step": 3770
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.950754847457076e-05,
|
|
"loss": 0.4279,
|
|
"step": 3772
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.9510735107204944e-05,
|
|
"loss": 0.4352,
|
|
"step": 3774
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.951392005155685e-05,
|
|
"loss": 0.3785,
|
|
"step": 3776
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.9517103309414456e-05,
|
|
"loss": 0.4148,
|
|
"step": 3778
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.9520284882562866e-05,
|
|
"loss": 0.377,
|
|
"step": 3780
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.9523464772784386e-05,
|
|
"loss": 0.3953,
|
|
"step": 3782
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.952664298185847e-05,
|
|
"loss": 0.4216,
|
|
"step": 3784
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.952981951156177e-05,
|
|
"loss": 0.4324,
|
|
"step": 3786
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.95329943636681e-05,
|
|
"loss": 0.4076,
|
|
"step": 3788
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.953616753994849e-05,
|
|
"loss": 0.3538,
|
|
"step": 3790
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.953933904217115e-05,
|
|
"loss": 0.4196,
|
|
"step": 3792
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.9542508872101494e-05,
|
|
"loss": 0.3735,
|
|
"step": 3794
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.9545677031502144e-05,
|
|
"loss": 0.4048,
|
|
"step": 3796
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.954884352213296e-05,
|
|
"loss": 0.4165,
|
|
"step": 3798
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.955200834575101e-05,
|
|
"loss": 0.3953,
|
|
"step": 3800
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.955517150411055e-05,
|
|
"loss": 0.4056,
|
|
"step": 3802
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.9558332998963116e-05,
|
|
"loss": 0.4329,
|
|
"step": 3804
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.9561492832057474e-05,
|
|
"loss": 0.3622,
|
|
"step": 3806
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.956465100513963e-05,
|
|
"loss": 0.4104,
|
|
"step": 3808
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.9567807519952824e-05,
|
|
"loss": 0.4233,
|
|
"step": 3810
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 4.9570962378237575e-05,
|
|
"loss": 0.3838,
|
|
"step": 3812
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.957411558173164e-05,
|
|
"loss": 0.4136,
|
|
"step": 3814
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.9577267132170066e-05,
|
|
"loss": 0.4171,
|
|
"step": 3816
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.958041703128516e-05,
|
|
"loss": 0.3917,
|
|
"step": 3818
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.95835652808065e-05,
|
|
"loss": 0.4075,
|
|
"step": 3820
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.958671188246097e-05,
|
|
"loss": 0.4131,
|
|
"step": 3822
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.9589856837972735e-05,
|
|
"loss": 0.3956,
|
|
"step": 3824
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.959300014906323e-05,
|
|
"loss": 0.3829,
|
|
"step": 3826
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.959614181745122e-05,
|
|
"loss": 0.4006,
|
|
"step": 3828
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.959928184485278e-05,
|
|
"loss": 0.3987,
|
|
"step": 3830
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.9602420232981284e-05,
|
|
"loss": 0.4147,
|
|
"step": 3832
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.9605556983547416e-05,
|
|
"loss": 0.4082,
|
|
"step": 3834
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.960869209825922e-05,
|
|
"loss": 0.4052,
|
|
"step": 3836
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.961182557882201e-05,
|
|
"loss": 0.4025,
|
|
"step": 3838
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.961495742693849e-05,
|
|
"loss": 0.4066,
|
|
"step": 3840
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.961808764430867e-05,
|
|
"loss": 0.4776,
|
|
"step": 3842
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.962121623262994e-05,
|
|
"loss": 0.4008,
|
|
"step": 3844
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.9624343193596986e-05,
|
|
"loss": 0.3891,
|
|
"step": 3846
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.96274685289019e-05,
|
|
"loss": 0.4229,
|
|
"step": 3848
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.963059224023413e-05,
|
|
"loss": 0.4497,
|
|
"step": 3850
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.963371432928047e-05,
|
|
"loss": 0.3937,
|
|
"step": 3852
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.963683479772511e-05,
|
|
"loss": 0.4116,
|
|
"step": 3854
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.9639953647249605e-05,
|
|
"loss": 0.3978,
|
|
"step": 3856
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.964307087953288e-05,
|
|
"loss": 0.3787,
|
|
"step": 3858
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.9646186496251295e-05,
|
|
"loss": 0.3665,
|
|
"step": 3860
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 4.964930049907856e-05,
|
|
"loss": 0.3586,
|
|
"step": 3862
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.965241288968581e-05,
|
|
"loss": 0.3903,
|
|
"step": 3864
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.9655523669741574e-05,
|
|
"loss": 0.4027,
|
|
"step": 3866
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.965863284091179e-05,
|
|
"loss": 0.4207,
|
|
"step": 3868
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.966174040485982e-05,
|
|
"loss": 0.3729,
|
|
"step": 3870
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.966484636324647e-05,
|
|
"loss": 0.36,
|
|
"step": 3872
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.966795071772992e-05,
|
|
"loss": 0.3726,
|
|
"step": 3874
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.967105346996583e-05,
|
|
"loss": 0.3635,
|
|
"step": 3876
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.967415462160726e-05,
|
|
"loss": 0.4147,
|
|
"step": 3878
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.967725417430476e-05,
|
|
"loss": 0.411,
|
|
"step": 3880
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.968035212970628e-05,
|
|
"loss": 0.3898,
|
|
"step": 3882
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.9683448489457254e-05,
|
|
"loss": 0.3872,
|
|
"step": 3884
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.9686543255200555e-05,
|
|
"loss": 0.4028,
|
|
"step": 3886
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.9689636428576545e-05,
|
|
"loss": 0.3978,
|
|
"step": 3888
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.969272801122305e-05,
|
|
"loss": 0.3592,
|
|
"step": 3890
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.969581800477534e-05,
|
|
"loss": 0.4215,
|
|
"step": 3892
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.969890641086619e-05,
|
|
"loss": 0.3914,
|
|
"step": 3894
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.970199323112585e-05,
|
|
"loss": 0.3684,
|
|
"step": 3896
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.970507846718208e-05,
|
|
"loss": 0.4271,
|
|
"step": 3898
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.970816212066011e-05,
|
|
"loss": 0.3735,
|
|
"step": 3900
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.971124419318269e-05,
|
|
"loss": 0.4147,
|
|
"step": 3902
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.971432468637006e-05,
|
|
"loss": 0.3918,
|
|
"step": 3904
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.971740360183998e-05,
|
|
"loss": 0.4336,
|
|
"step": 3906
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.972048094120773e-05,
|
|
"loss": 0.3827,
|
|
"step": 3908
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.972355670608609e-05,
|
|
"loss": 0.3992,
|
|
"step": 3910
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.9726630898085386e-05,
|
|
"loss": 0.3992,
|
|
"step": 3912
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 4.972970351881346e-05,
|
|
"loss": 0.3788,
|
|
"step": 3914
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.9732774569875705e-05,
|
|
"loss": 0.3931,
|
|
"step": 3916
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.9735844052875055e-05,
|
|
"loss": 0.4575,
|
|
"step": 3918
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.9738911969411965e-05,
|
|
"loss": 0.3607,
|
|
"step": 3920
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.974197832108447e-05,
|
|
"loss": 0.3779,
|
|
"step": 3922
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.974504310948816e-05,
|
|
"loss": 0.405,
|
|
"step": 3924
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.974810633621616e-05,
|
|
"loss": 0.4406,
|
|
"step": 3926
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.975116800285916e-05,
|
|
"loss": 0.4047,
|
|
"step": 3928
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.975422811100547e-05,
|
|
"loss": 0.3752,
|
|
"step": 3930
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.975728666224093e-05,
|
|
"loss": 0.3881,
|
|
"step": 3932
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.976034365814897e-05,
|
|
"loss": 0.4176,
|
|
"step": 3934
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.9763399100310604e-05,
|
|
"loss": 0.4326,
|
|
"step": 3936
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.976645299030446e-05,
|
|
"loss": 0.3788,
|
|
"step": 3938
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.9769505329706726e-05,
|
|
"loss": 0.3777,
|
|
"step": 3940
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.977255612009121e-05,
|
|
"loss": 0.3798,
|
|
"step": 3942
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.9775605363029334e-05,
|
|
"loss": 0.39,
|
|
"step": 3944
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.97786530600901e-05,
|
|
"loss": 0.3759,
|
|
"step": 3946
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.9781699212840164e-05,
|
|
"loss": 0.3923,
|
|
"step": 3948
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.978474382284376e-05,
|
|
"loss": 0.4044,
|
|
"step": 3950
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.978778689166279e-05,
|
|
"loss": 0.4536,
|
|
"step": 3952
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.979082842085674e-05,
|
|
"loss": 0.3842,
|
|
"step": 3954
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.979386841198277e-05,
|
|
"loss": 0.4044,
|
|
"step": 3956
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.979690686659565e-05,
|
|
"loss": 0.4048,
|
|
"step": 3958
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.979994378624782e-05,
|
|
"loss": 0.3982,
|
|
"step": 3960
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.980297917248934e-05,
|
|
"loss": 0.3966,
|
|
"step": 3962
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 4.9806013026867945e-05,
|
|
"loss": 0.4453,
|
|
"step": 3964
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.9809045350929016e-05,
|
|
"loss": 0.3912,
|
|
"step": 3966
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.9812076146215605e-05,
|
|
"loss": 0.39,
|
|
"step": 3968
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.981510541426842e-05,
|
|
"loss": 0.402,
|
|
"step": 3970
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.981813315662586e-05,
|
|
"loss": 0.3984,
|
|
"step": 3972
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.982115937482398e-05,
|
|
"loss": 0.4215,
|
|
"step": 3974
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.982418407039652e-05,
|
|
"loss": 0.3905,
|
|
"step": 3976
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.982720724487493e-05,
|
|
"loss": 0.3772,
|
|
"step": 3978
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.983022889978831e-05,
|
|
"loss": 0.4175,
|
|
"step": 3980
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.98332490366635e-05,
|
|
"loss": 0.3946,
|
|
"step": 3982
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.9836267657025e-05,
|
|
"loss": 0.3911,
|
|
"step": 3984
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.983928476239504e-05,
|
|
"loss": 0.4135,
|
|
"step": 3986
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.984230035429354e-05,
|
|
"loss": 0.3845,
|
|
"step": 3988
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.984531443423817e-05,
|
|
"loss": 0.3619,
|
|
"step": 3990
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.984832700374426e-05,
|
|
"loss": 0.3676,
|
|
"step": 3992
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.985133806432492e-05,
|
|
"loss": 0.4062,
|
|
"step": 3994
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.9854347617490966e-05,
|
|
"loss": 0.3879,
|
|
"step": 3996
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.9857355664750934e-05,
|
|
"loss": 0.3934,
|
|
"step": 3998
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.986036220761111e-05,
|
|
"loss": 0.375,
|
|
"step": 4000
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.986336724757552e-05,
|
|
"loss": 0.3839,
|
|
"step": 4002
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.986637078614592e-05,
|
|
"loss": 0.4156,
|
|
"step": 4004
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.986937282482184e-05,
|
|
"loss": 0.3841,
|
|
"step": 4006
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.987237336510056e-05,
|
|
"loss": 0.4082,
|
|
"step": 4008
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.987537240847711e-05,
|
|
"loss": 0.4172,
|
|
"step": 4010
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.987836995644429e-05,
|
|
"loss": 0.3908,
|
|
"step": 4012
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.988136601049266e-05,
|
|
"loss": 0.3834,
|
|
"step": 4014
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 4.988436057211055e-05,
|
|
"loss": 0.3821,
|
|
"step": 4016
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.988735364278408e-05,
|
|
"loss": 0.3651,
|
|
"step": 4018
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.989034522399716e-05,
|
|
"loss": 0.3951,
|
|
"step": 4020
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.989333531723145e-05,
|
|
"loss": 0.3833,
|
|
"step": 4022
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.989632392396644e-05,
|
|
"loss": 0.4174,
|
|
"step": 4024
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.989931104567939e-05,
|
|
"loss": 0.4276,
|
|
"step": 4026
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.990229668384536e-05,
|
|
"loss": 0.4238,
|
|
"step": 4028
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.990528083993722e-05,
|
|
"loss": 0.4424,
|
|
"step": 4030
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.990826351542565e-05,
|
|
"loss": 0.4164,
|
|
"step": 4032
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.991124471177915e-05,
|
|
"loss": 0.3972,
|
|
"step": 4034
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.9914224430464e-05,
|
|
"loss": 0.3894,
|
|
"step": 4036
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.991720267294434e-05,
|
|
"loss": 0.3589,
|
|
"step": 4038
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.9920179440682124e-05,
|
|
"loss": 0.3713,
|
|
"step": 4040
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.992315473513712e-05,
|
|
"loss": 0.3947,
|
|
"step": 4042
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.9926128557766947e-05,
|
|
"loss": 0.4264,
|
|
"step": 4044
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.992910091002707e-05,
|
|
"loss": 0.4092,
|
|
"step": 4046
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.9932071793370764e-05,
|
|
"loss": 0.4291,
|
|
"step": 4048
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.993504120924917e-05,
|
|
"loss": 0.3656,
|
|
"step": 4050
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.993800915911129e-05,
|
|
"loss": 0.4232,
|
|
"step": 4052
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.9940975644403964e-05,
|
|
"loss": 0.3817,
|
|
"step": 4054
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.99439406665719e-05,
|
|
"loss": 0.4281,
|
|
"step": 4056
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.994690422705765e-05,
|
|
"loss": 0.3979,
|
|
"step": 4058
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.994986632730168e-05,
|
|
"loss": 0.3712,
|
|
"step": 4060
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.995282696874227e-05,
|
|
"loss": 0.4337,
|
|
"step": 4062
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.995578615281561e-05,
|
|
"loss": 0.4023,
|
|
"step": 4064
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.9958743880955774e-05,
|
|
"loss": 0.42,
|
|
"step": 4066
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 4.99617001545947e-05,
|
|
"loss": 0.4301,
|
|
"step": 4068
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 4.996465497516224e-05,
|
|
"loss": 0.415,
|
|
"step": 4070
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 4.9967608344086096e-05,
|
|
"loss": 0.4125,
|
|
"step": 4072
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 4.9970560262791914e-05,
|
|
"loss": 0.4058,
|
|
"step": 4074
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 4.9973510732703224e-05,
|
|
"loss": 0.413,
|
|
"step": 4076
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 4.997645975524144e-05,
|
|
"loss": 0.3904,
|
|
"step": 4078
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 4.9979407331825927e-05,
|
|
"loss": 0.399,
|
|
"step": 4080
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 4.998235346387392e-05,
|
|
"loss": 0.3942,
|
|
"step": 4082
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 4.99852981528006e-05,
|
|
"loss": 0.3959,
|
|
"step": 4084
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 4.998824140001906e-05,
|
|
"loss": 0.4098,
|
|
"step": 4086
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 4.999118320694031e-05,
|
|
"loss": 0.4044,
|
|
"step": 4088
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 4.999412357497331e-05,
|
|
"loss": 0.4011,
|
|
"step": 4090
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 4.999706250552494e-05,
|
|
"loss": 0.4119,
|
|
"step": 4092
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3974,
|
|
"step": 4094
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4248,
|
|
"step": 4096
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4197,
|
|
"step": 4098
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4181,
|
|
"step": 4100
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3878,
|
|
"step": 4102
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4355,
|
|
"step": 4104
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4232,
|
|
"step": 4106
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3993,
|
|
"step": 4108
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4196,
|
|
"step": 4110
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3572,
|
|
"step": 4112
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3953,
|
|
"step": 4114
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3583,
|
|
"step": 4116
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4193,
|
|
"step": 4118
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4211,
|
|
"step": 4120
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4038,
|
|
"step": 4122
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4198,
|
|
"step": 4124
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4158,
|
|
"step": 4126
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3827,
|
|
"step": 4128
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4213,
|
|
"step": 4130
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4214,
|
|
"step": 4132
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4508,
|
|
"step": 4134
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4063,
|
|
"step": 4136
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3769,
|
|
"step": 4138
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4104,
|
|
"step": 4140
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4199,
|
|
"step": 4142
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3917,
|
|
"step": 4144
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3736,
|
|
"step": 4146
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3994,
|
|
"step": 4148
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4172,
|
|
"step": 4150
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3744,
|
|
"step": 4152
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.375,
|
|
"step": 4154
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4109,
|
|
"step": 4156
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3843,
|
|
"step": 4158
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4207,
|
|
"step": 4160
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.369,
|
|
"step": 4162
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3843,
|
|
"step": 4164
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3569,
|
|
"step": 4166
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.421,
|
|
"step": 4168
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3778,
|
|
"step": 4170
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3715,
|
|
"step": 4172
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3974,
|
|
"step": 4174
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4196,
|
|
"step": 4176
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3925,
|
|
"step": 4178
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3397,
|
|
"step": 4180
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.366,
|
|
"step": 4182
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.394,
|
|
"step": 4184
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4136,
|
|
"step": 4186
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3913,
|
|
"step": 4188
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.371,
|
|
"step": 4190
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3724,
|
|
"step": 4192
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3774,
|
|
"step": 4194
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4252,
|
|
"step": 4196
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3672,
|
|
"step": 4198
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4042,
|
|
"step": 4200
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3546,
|
|
"step": 4202
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4155,
|
|
"step": 4204
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4074,
|
|
"step": 4206
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3929,
|
|
"step": 4208
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4058,
|
|
"step": 4210
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4272,
|
|
"step": 4212
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4135,
|
|
"step": 4214
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3635,
|
|
"step": 4216
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4082,
|
|
"step": 4218
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3916,
|
|
"step": 4220
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4005,
|
|
"step": 4222
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3934,
|
|
"step": 4224
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3752,
|
|
"step": 4226
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4117,
|
|
"step": 4228
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3951,
|
|
"step": 4230
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3848,
|
|
"step": 4232
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3951,
|
|
"step": 4234
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4092,
|
|
"step": 4236
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4227,
|
|
"step": 4238
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3989,
|
|
"step": 4240
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4578,
|
|
"step": 4242
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4498,
|
|
"step": 4244
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3733,
|
|
"step": 4246
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4257,
|
|
"step": 4248
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3914,
|
|
"step": 4250
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3856,
|
|
"step": 4252
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4114,
|
|
"step": 4254
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3899,
|
|
"step": 4256
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3629,
|
|
"step": 4258
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4058,
|
|
"step": 4260
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4138,
|
|
"step": 4262
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4323,
|
|
"step": 4264
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.443,
|
|
"step": 4266
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3885,
|
|
"step": 4268
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4312,
|
|
"step": 4270
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4018,
|
|
"step": 4272
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4189,
|
|
"step": 4274
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4075,
|
|
"step": 4276
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4136,
|
|
"step": 4278
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4141,
|
|
"step": 4280
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3619,
|
|
"step": 4282
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4323,
|
|
"step": 4284
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3418,
|
|
"step": 4286
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.418,
|
|
"step": 4288
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4231,
|
|
"step": 4290
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3721,
|
|
"step": 4292
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3984,
|
|
"step": 4294
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4318,
|
|
"step": 4296
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3914,
|
|
"step": 4298
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3672,
|
|
"step": 4300
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3727,
|
|
"step": 4302
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.407,
|
|
"step": 4304
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4354,
|
|
"step": 4306
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4128,
|
|
"step": 4308
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4613,
|
|
"step": 4310
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.394,
|
|
"step": 4312
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4199,
|
|
"step": 4314
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3625,
|
|
"step": 4316
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3809,
|
|
"step": 4318
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.411,
|
|
"step": 4320
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4001,
|
|
"step": 4322
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3964,
|
|
"step": 4324
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4066,
|
|
"step": 4326
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3896,
|
|
"step": 4328
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3801,
|
|
"step": 4330
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.377,
|
|
"step": 4332
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3566,
|
|
"step": 4334
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3948,
|
|
"step": 4336
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3993,
|
|
"step": 4338
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.411,
|
|
"step": 4340
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3921,
|
|
"step": 4342
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3885,
|
|
"step": 4344
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3966,
|
|
"step": 4346
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3793,
|
|
"step": 4348
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4036,
|
|
"step": 4350
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4073,
|
|
"step": 4352
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3866,
|
|
"step": 4354
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4061,
|
|
"step": 4356
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3816,
|
|
"step": 4358
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4021,
|
|
"step": 4360
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4193,
|
|
"step": 4362
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4205,
|
|
"step": 4364
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4339,
|
|
"step": 4366
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4102,
|
|
"step": 4368
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4098,
|
|
"step": 4370
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3987,
|
|
"step": 4372
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4098,
|
|
"step": 4374
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4136,
|
|
"step": 4376
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4092,
|
|
"step": 4378
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3783,
|
|
"step": 4380
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3964,
|
|
"step": 4382
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4188,
|
|
"step": 4384
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3801,
|
|
"step": 4386
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3677,
|
|
"step": 4388
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4019,
|
|
"step": 4390
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3883,
|
|
"step": 4392
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3836,
|
|
"step": 4394
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.402,
|
|
"step": 4396
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3575,
|
|
"step": 4398
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.433,
|
|
"step": 4400
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4139,
|
|
"step": 4402
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4125,
|
|
"step": 4404
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4177,
|
|
"step": 4406
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.376,
|
|
"step": 4408
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4551,
|
|
"step": 4410
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4155,
|
|
"step": 4412
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4393,
|
|
"step": 4414
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3746,
|
|
"step": 4416
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3975,
|
|
"step": 4418
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4138,
|
|
"step": 4420
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3642,
|
|
"step": 4422
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3955,
|
|
"step": 4424
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.425,
|
|
"step": 4426
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3868,
|
|
"step": 4428
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4108,
|
|
"step": 4430
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3778,
|
|
"step": 4432
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3905,
|
|
"step": 4434
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3782,
|
|
"step": 4436
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4086,
|
|
"step": 4438
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3817,
|
|
"step": 4440
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4108,
|
|
"step": 4442
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4509,
|
|
"step": 4444
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3743,
|
|
"step": 4446
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3768,
|
|
"step": 4448
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4232,
|
|
"step": 4450
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4037,
|
|
"step": 4452
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3959,
|
|
"step": 4454
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4113,
|
|
"step": 4456
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.423,
|
|
"step": 4458
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.412,
|
|
"step": 4460
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3558,
|
|
"step": 4462
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4016,
|
|
"step": 4464
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.444,
|
|
"step": 4466
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3801,
|
|
"step": 4468
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.396,
|
|
"step": 4470
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3899,
|
|
"step": 4472
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.38,
|
|
"step": 4474
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4314,
|
|
"step": 4476
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4077,
|
|
"step": 4478
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4323,
|
|
"step": 4480
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4198,
|
|
"step": 4482
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3806,
|
|
"step": 4484
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3864,
|
|
"step": 4486
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4281,
|
|
"step": 4488
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3781,
|
|
"step": 4490
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4009,
|
|
"step": 4492
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3866,
|
|
"step": 4494
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3782,
|
|
"step": 4496
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3821,
|
|
"step": 4498
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4133,
|
|
"step": 4500
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3966,
|
|
"step": 4502
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3959,
|
|
"step": 4504
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3771,
|
|
"step": 4506
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3798,
|
|
"step": 4508
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4113,
|
|
"step": 4510
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3565,
|
|
"step": 4512
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4406,
|
|
"step": 4514
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3807,
|
|
"step": 4516
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4153,
|
|
"step": 4518
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3792,
|
|
"step": 4520
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4158,
|
|
"step": 4522
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3912,
|
|
"step": 4524
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4131,
|
|
"step": 4526
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3987,
|
|
"step": 4528
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3977,
|
|
"step": 4530
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4314,
|
|
"step": 4532
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.384,
|
|
"step": 4534
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3891,
|
|
"step": 4536
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3546,
|
|
"step": 4538
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3914,
|
|
"step": 4540
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3956,
|
|
"step": 4542
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4312,
|
|
"step": 4544
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3777,
|
|
"step": 4546
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4075,
|
|
"step": 4548
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4054,
|
|
"step": 4550
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3768,
|
|
"step": 4552
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4272,
|
|
"step": 4554
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3981,
|
|
"step": 4556
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4088,
|
|
"step": 4558
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3854,
|
|
"step": 4560
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3937,
|
|
"step": 4562
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3955,
|
|
"step": 4564
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4082,
|
|
"step": 4566
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.423,
|
|
"step": 4568
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4011,
|
|
"step": 4570
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3933,
|
|
"step": 4572
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3582,
|
|
"step": 4574
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3981,
|
|
"step": 4576
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4291,
|
|
"step": 4578
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3669,
|
|
"step": 4580
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3879,
|
|
"step": 4582
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4059,
|
|
"step": 4584
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4039,
|
|
"step": 4586
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3615,
|
|
"step": 4588
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.396,
|
|
"step": 4590
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3591,
|
|
"step": 4592
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.377,
|
|
"step": 4594
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3885,
|
|
"step": 4596
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3922,
|
|
"step": 4598
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4484,
|
|
"step": 4600
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4393,
|
|
"step": 4602
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3776,
|
|
"step": 4604
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4128,
|
|
"step": 4606
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3976,
|
|
"step": 4608
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.392,
|
|
"step": 4610
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4261,
|
|
"step": 4612
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4353,
|
|
"step": 4614
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4001,
|
|
"step": 4616
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3723,
|
|
"step": 4618
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4515,
|
|
"step": 4620
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3658,
|
|
"step": 4622
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4092,
|
|
"step": 4624
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4005,
|
|
"step": 4626
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4327,
|
|
"step": 4628
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3765,
|
|
"step": 4630
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.41,
|
|
"step": 4632
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4337,
|
|
"step": 4634
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3975,
|
|
"step": 4636
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4033,
|
|
"step": 4638
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3951,
|
|
"step": 4640
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4304,
|
|
"step": 4642
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3695,
|
|
"step": 4644
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3982,
|
|
"step": 4646
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3893,
|
|
"step": 4648
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3797,
|
|
"step": 4650
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4125,
|
|
"step": 4652
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.386,
|
|
"step": 4654
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4072,
|
|
"step": 4656
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3872,
|
|
"step": 4658
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4009,
|
|
"step": 4660
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3885,
|
|
"step": 4662
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4309,
|
|
"step": 4664
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4073,
|
|
"step": 4666
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4011,
|
|
"step": 4668
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4004,
|
|
"step": 4670
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4135,
|
|
"step": 4672
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4103,
|
|
"step": 4674
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3857,
|
|
"step": 4676
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3896,
|
|
"step": 4678
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3907,
|
|
"step": 4680
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4357,
|
|
"step": 4682
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3848,
|
|
"step": 4684
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4008,
|
|
"step": 4686
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.372,
|
|
"step": 4688
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3905,
|
|
"step": 4690
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4188,
|
|
"step": 4692
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3897,
|
|
"step": 4694
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3995,
|
|
"step": 4696
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3748,
|
|
"step": 4698
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4015,
|
|
"step": 4700
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4005,
|
|
"step": 4702
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4263,
|
|
"step": 4704
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4148,
|
|
"step": 4706
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4254,
|
|
"step": 4708
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4043,
|
|
"step": 4710
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4124,
|
|
"step": 4712
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.372,
|
|
"step": 4714
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3573,
|
|
"step": 4716
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4038,
|
|
"step": 4718
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3803,
|
|
"step": 4720
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3908,
|
|
"step": 4722
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4324,
|
|
"step": 4724
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4012,
|
|
"step": 4726
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.379,
|
|
"step": 4728
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4198,
|
|
"step": 4730
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3884,
|
|
"step": 4732
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3967,
|
|
"step": 4734
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3806,
|
|
"step": 4736
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3964,
|
|
"step": 4738
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3707,
|
|
"step": 4740
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4019,
|
|
"step": 4742
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4214,
|
|
"step": 4744
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3698,
|
|
"step": 4746
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4113,
|
|
"step": 4748
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3959,
|
|
"step": 4750
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3981,
|
|
"step": 4752
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4106,
|
|
"step": 4754
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3802,
|
|
"step": 4756
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3922,
|
|
"step": 4758
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4055,
|
|
"step": 4760
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3997,
|
|
"step": 4762
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3805,
|
|
"step": 4764
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4366,
|
|
"step": 4766
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3734,
|
|
"step": 4768
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4161,
|
|
"step": 4770
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4116,
|
|
"step": 4772
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4121,
|
|
"step": 4774
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3868,
|
|
"step": 4776
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.402,
|
|
"step": 4778
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3997,
|
|
"step": 4780
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3687,
|
|
"step": 4782
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3918,
|
|
"step": 4784
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3926,
|
|
"step": 4786
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3873,
|
|
"step": 4788
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3591,
|
|
"step": 4790
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3609,
|
|
"step": 4792
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3984,
|
|
"step": 4794
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3731,
|
|
"step": 4796
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4385,
|
|
"step": 4798
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4142,
|
|
"step": 4800
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3945,
|
|
"step": 4802
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3966,
|
|
"step": 4804
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4,
|
|
"step": 4806
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4019,
|
|
"step": 4808
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3896,
|
|
"step": 4810
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3732,
|
|
"step": 4812
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3716,
|
|
"step": 4814
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4539,
|
|
"step": 4816
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3549,
|
|
"step": 4818
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4097,
|
|
"step": 4820
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.407,
|
|
"step": 4822
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3771,
|
|
"step": 4824
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3895,
|
|
"step": 4826
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3859,
|
|
"step": 4828
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.457,
|
|
"step": 4830
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4182,
|
|
"step": 4832
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4023,
|
|
"step": 4834
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3796,
|
|
"step": 4836
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3966,
|
|
"step": 4838
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.5159,
|
|
"step": 4840
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3937,
|
|
"step": 4842
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4129,
|
|
"step": 4844
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4022,
|
|
"step": 4846
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3716,
|
|
"step": 4848
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4131,
|
|
"step": 4850
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3916,
|
|
"step": 4852
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3903,
|
|
"step": 4854
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3813,
|
|
"step": 4856
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4192,
|
|
"step": 4858
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4236,
|
|
"step": 4860
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4349,
|
|
"step": 4862
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.386,
|
|
"step": 4864
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3937,
|
|
"step": 4866
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4176,
|
|
"step": 4868
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4292,
|
|
"step": 4870
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3536,
|
|
"step": 4872
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3774,
|
|
"step": 4874
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3926,
|
|
"step": 4876
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3536,
|
|
"step": 4878
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.406,
|
|
"step": 4880
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3764,
|
|
"step": 4882
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3453,
|
|
"step": 4884
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4215,
|
|
"step": 4886
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4338,
|
|
"step": 4888
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4216,
|
|
"step": 4890
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3878,
|
|
"step": 4892
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4012,
|
|
"step": 4894
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4296,
|
|
"step": 4896
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.427,
|
|
"step": 4898
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3986,
|
|
"step": 4900
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3501,
|
|
"step": 4902
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3959,
|
|
"step": 4904
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4189,
|
|
"step": 4906
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3943,
|
|
"step": 4908
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3666,
|
|
"step": 4910
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3782,
|
|
"step": 4912
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.395,
|
|
"step": 4914
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3912,
|
|
"step": 4916
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3964,
|
|
"step": 4918
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3804,
|
|
"step": 4920
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.359,
|
|
"step": 4922
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3833,
|
|
"step": 4924
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3919,
|
|
"step": 4926
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4141,
|
|
"step": 4928
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4045,
|
|
"step": 4930
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4065,
|
|
"step": 4932
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3973,
|
|
"step": 4934
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4802,
|
|
"step": 4936
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3713,
|
|
"step": 4938
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4088,
|
|
"step": 4940
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3585,
|
|
"step": 4942
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3921,
|
|
"step": 4944
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3989,
|
|
"step": 4946
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3868,
|
|
"step": 4948
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3783,
|
|
"step": 4950
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3731,
|
|
"step": 4952
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4246,
|
|
"step": 4954
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3967,
|
|
"step": 4956
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4005,
|
|
"step": 4958
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3698,
|
|
"step": 4960
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3875,
|
|
"step": 4962
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4159,
|
|
"step": 4964
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3988,
|
|
"step": 4966
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3995,
|
|
"step": 4968
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4049,
|
|
"step": 4970
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3792,
|
|
"step": 4972
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3894,
|
|
"step": 4974
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3837,
|
|
"step": 4976
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3866,
|
|
"step": 4978
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.374,
|
|
"step": 4980
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4204,
|
|
"step": 4982
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3986,
|
|
"step": 4984
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.376,
|
|
"step": 4986
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4163,
|
|
"step": 4988
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3769,
|
|
"step": 4990
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3609,
|
|
"step": 4992
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3826,
|
|
"step": 4994
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3529,
|
|
"step": 4996
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3851,
|
|
"step": 4998
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3972,
|
|
"step": 5000
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4017,
|
|
"step": 5002
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4143,
|
|
"step": 5004
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3845,
|
|
"step": 5006
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3633,
|
|
"step": 5008
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3649,
|
|
"step": 5010
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4145,
|
|
"step": 5012
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3964,
|
|
"step": 5014
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4258,
|
|
"step": 5016
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3861,
|
|
"step": 5018
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3954,
|
|
"step": 5020
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3879,
|
|
"step": 5022
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4163,
|
|
"step": 5024
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4188,
|
|
"step": 5026
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4119,
|
|
"step": 5028
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3729,
|
|
"step": 5030
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4315,
|
|
"step": 5032
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3611,
|
|
"step": 5034
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4167,
|
|
"step": 5036
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3788,
|
|
"step": 5038
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3907,
|
|
"step": 5040
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4214,
|
|
"step": 5042
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3917,
|
|
"step": 5044
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3966,
|
|
"step": 5046
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4039,
|
|
"step": 5048
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3752,
|
|
"step": 5050
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4131,
|
|
"step": 5052
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3948,
|
|
"step": 5054
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4159,
|
|
"step": 5056
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3463,
|
|
"step": 5058
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3884,
|
|
"step": 5060
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4111,
|
|
"step": 5062
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3792,
|
|
"step": 5064
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3777,
|
|
"step": 5066
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4022,
|
|
"step": 5068
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3916,
|
|
"step": 5070
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3837,
|
|
"step": 5072
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4943,
|
|
"step": 5074
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3894,
|
|
"step": 5076
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3774,
|
|
"step": 5078
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3893,
|
|
"step": 5080
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4177,
|
|
"step": 5082
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3761,
|
|
"step": 5084
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4198,
|
|
"step": 5086
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3596,
|
|
"step": 5088
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3511,
|
|
"step": 5090
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3848,
|
|
"step": 5092
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3765,
|
|
"step": 5094
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4037,
|
|
"step": 5096
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3931,
|
|
"step": 5098
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3887,
|
|
"step": 5100
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4175,
|
|
"step": 5102
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3713,
|
|
"step": 5104
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.428,
|
|
"step": 5106
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3749,
|
|
"step": 5108
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3702,
|
|
"step": 5110
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3533,
|
|
"step": 5112
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3728,
|
|
"step": 5114
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4355,
|
|
"step": 5116
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3447,
|
|
"step": 5118
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.309,
|
|
"step": 5120
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3097,
|
|
"step": 5122
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3116,
|
|
"step": 5124
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3751,
|
|
"step": 5126
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.282,
|
|
"step": 5128
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3335,
|
|
"step": 5130
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2927,
|
|
"step": 5132
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2842,
|
|
"step": 5134
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.287,
|
|
"step": 5136
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3029,
|
|
"step": 5138
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2798,
|
|
"step": 5140
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3102,
|
|
"step": 5142
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2924,
|
|
"step": 5144
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2802,
|
|
"step": 5146
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2713,
|
|
"step": 5148
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.311,
|
|
"step": 5150
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.27,
|
|
"step": 5152
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2803,
|
|
"step": 5154
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.301,
|
|
"step": 5156
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.295,
|
|
"step": 5158
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.308,
|
|
"step": 5160
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.291,
|
|
"step": 5162
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3003,
|
|
"step": 5164
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2742,
|
|
"step": 5166
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.278,
|
|
"step": 5168
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2639,
|
|
"step": 5170
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2432,
|
|
"step": 5172
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3098,
|
|
"step": 5174
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2725,
|
|
"step": 5176
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2766,
|
|
"step": 5178
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2883,
|
|
"step": 5180
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2603,
|
|
"step": 5182
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2745,
|
|
"step": 5184
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2916,
|
|
"step": 5186
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2858,
|
|
"step": 5188
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2924,
|
|
"step": 5190
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2735,
|
|
"step": 5192
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.291,
|
|
"step": 5194
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.303,
|
|
"step": 5196
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2733,
|
|
"step": 5198
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3085,
|
|
"step": 5200
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2894,
|
|
"step": 5202
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2902,
|
|
"step": 5204
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2909,
|
|
"step": 5206
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2988,
|
|
"step": 5208
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2815,
|
|
"step": 5210
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3001,
|
|
"step": 5212
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2889,
|
|
"step": 5214
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3077,
|
|
"step": 5216
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3617,
|
|
"step": 5218
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2959,
|
|
"step": 5220
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2717,
|
|
"step": 5222
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3035,
|
|
"step": 5224
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2777,
|
|
"step": 5226
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3109,
|
|
"step": 5228
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3145,
|
|
"step": 5230
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2706,
|
|
"step": 5232
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2922,
|
|
"step": 5234
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2788,
|
|
"step": 5236
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3128,
|
|
"step": 5238
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2824,
|
|
"step": 5240
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2659,
|
|
"step": 5242
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3856,
|
|
"step": 5244
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3079,
|
|
"step": 5246
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3167,
|
|
"step": 5248
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3185,
|
|
"step": 5250
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2765,
|
|
"step": 5252
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2734,
|
|
"step": 5254
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2867,
|
|
"step": 5256
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.29,
|
|
"step": 5258
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2961,
|
|
"step": 5260
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3007,
|
|
"step": 5262
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.29,
|
|
"step": 5264
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3136,
|
|
"step": 5266
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3131,
|
|
"step": 5268
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3029,
|
|
"step": 5270
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2902,
|
|
"step": 5272
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2936,
|
|
"step": 5274
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2845,
|
|
"step": 5276
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3103,
|
|
"step": 5278
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3013,
|
|
"step": 5280
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3362,
|
|
"step": 5282
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3079,
|
|
"step": 5284
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2799,
|
|
"step": 5286
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3126,
|
|
"step": 5288
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2873,
|
|
"step": 5290
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2817,
|
|
"step": 5292
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2798,
|
|
"step": 5294
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2787,
|
|
"step": 5296
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2781,
|
|
"step": 5298
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3262,
|
|
"step": 5300
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2736,
|
|
"step": 5302
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2955,
|
|
"step": 5304
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2908,
|
|
"step": 5306
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2831,
|
|
"step": 5308
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2934,
|
|
"step": 5310
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3004,
|
|
"step": 5312
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2812,
|
|
"step": 5314
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3045,
|
|
"step": 5316
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3266,
|
|
"step": 5318
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3122,
|
|
"step": 5320
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3045,
|
|
"step": 5322
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2726,
|
|
"step": 5324
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.306,
|
|
"step": 5326
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2978,
|
|
"step": 5328
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3138,
|
|
"step": 5330
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2966,
|
|
"step": 5332
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2877,
|
|
"step": 5334
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3035,
|
|
"step": 5336
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2766,
|
|
"step": 5338
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2998,
|
|
"step": 5340
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.299,
|
|
"step": 5342
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2811,
|
|
"step": 5344
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2815,
|
|
"step": 5346
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3091,
|
|
"step": 5348
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3154,
|
|
"step": 5350
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2836,
|
|
"step": 5352
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.278,
|
|
"step": 5354
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3196,
|
|
"step": 5356
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2786,
|
|
"step": 5358
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3121,
|
|
"step": 5360
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2803,
|
|
"step": 5362
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.269,
|
|
"step": 5364
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2866,
|
|
"step": 5366
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2958,
|
|
"step": 5368
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2821,
|
|
"step": 5370
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2784,
|
|
"step": 5372
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3032,
|
|
"step": 5374
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.315,
|
|
"step": 5376
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2826,
|
|
"step": 5378
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2807,
|
|
"step": 5380
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3091,
|
|
"step": 5382
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2768,
|
|
"step": 5384
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3073,
|
|
"step": 5386
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2744,
|
|
"step": 5388
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3198,
|
|
"step": 5390
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2952,
|
|
"step": 5392
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3024,
|
|
"step": 5394
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2858,
|
|
"step": 5396
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2868,
|
|
"step": 5398
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3182,
|
|
"step": 5400
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.267,
|
|
"step": 5402
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.323,
|
|
"step": 5404
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3198,
|
|
"step": 5406
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2911,
|
|
"step": 5408
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3165,
|
|
"step": 5410
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2923,
|
|
"step": 5412
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2933,
|
|
"step": 5414
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2855,
|
|
"step": 5416
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3019,
|
|
"step": 5418
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3125,
|
|
"step": 5420
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2997,
|
|
"step": 5422
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2787,
|
|
"step": 5424
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3165,
|
|
"step": 5426
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3125,
|
|
"step": 5428
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2825,
|
|
"step": 5430
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3088,
|
|
"step": 5432
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2797,
|
|
"step": 5434
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2966,
|
|
"step": 5436
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3019,
|
|
"step": 5438
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.292,
|
|
"step": 5440
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3315,
|
|
"step": 5442
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2933,
|
|
"step": 5444
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3,
|
|
"step": 5446
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2977,
|
|
"step": 5448
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2916,
|
|
"step": 5450
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3336,
|
|
"step": 5452
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3196,
|
|
"step": 5454
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3083,
|
|
"step": 5456
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2876,
|
|
"step": 5458
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2952,
|
|
"step": 5460
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2985,
|
|
"step": 5462
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2848,
|
|
"step": 5464
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2906,
|
|
"step": 5466
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2921,
|
|
"step": 5468
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2895,
|
|
"step": 5470
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3121,
|
|
"step": 5472
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.287,
|
|
"step": 5474
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3007,
|
|
"step": 5476
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3898,
|
|
"step": 5478
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.28,
|
|
"step": 5480
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3323,
|
|
"step": 5482
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3027,
|
|
"step": 5484
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3181,
|
|
"step": 5486
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3236,
|
|
"step": 5488
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2939,
|
|
"step": 5490
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.299,
|
|
"step": 5492
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2925,
|
|
"step": 5494
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2859,
|
|
"step": 5496
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3061,
|
|
"step": 5498
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3064,
|
|
"step": 5500
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3254,
|
|
"step": 5502
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2794,
|
|
"step": 5504
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2912,
|
|
"step": 5506
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.314,
|
|
"step": 5508
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2966,
|
|
"step": 5510
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.303,
|
|
"step": 5512
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2877,
|
|
"step": 5514
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2847,
|
|
"step": 5516
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3065,
|
|
"step": 5518
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2886,
|
|
"step": 5520
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2792,
|
|
"step": 5522
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2708,
|
|
"step": 5524
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2993,
|
|
"step": 5526
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2499,
|
|
"step": 5528
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3298,
|
|
"step": 5530
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3035,
|
|
"step": 5532
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2848,
|
|
"step": 5534
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2709,
|
|
"step": 5536
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2956,
|
|
"step": 5538
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2975,
|
|
"step": 5540
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3044,
|
|
"step": 5542
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3141,
|
|
"step": 5544
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2981,
|
|
"step": 5546
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3047,
|
|
"step": 5548
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2634,
|
|
"step": 5550
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3201,
|
|
"step": 5552
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3052,
|
|
"step": 5554
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.315,
|
|
"step": 5556
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3013,
|
|
"step": 5558
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2921,
|
|
"step": 5560
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2965,
|
|
"step": 5562
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.291,
|
|
"step": 5564
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2938,
|
|
"step": 5566
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3092,
|
|
"step": 5568
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3042,
|
|
"step": 5570
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3158,
|
|
"step": 5572
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2966,
|
|
"step": 5574
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3129,
|
|
"step": 5576
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2849,
|
|
"step": 5578
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3167,
|
|
"step": 5580
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2885,
|
|
"step": 5582
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2825,
|
|
"step": 5584
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2985,
|
|
"step": 5586
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2769,
|
|
"step": 5588
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3034,
|
|
"step": 5590
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3169,
|
|
"step": 5592
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.27,
|
|
"step": 5594
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2763,
|
|
"step": 5596
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3344,
|
|
"step": 5598
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3177,
|
|
"step": 5600
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2946,
|
|
"step": 5602
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2755,
|
|
"step": 5604
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3029,
|
|
"step": 5606
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3112,
|
|
"step": 5608
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2902,
|
|
"step": 5610
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2859,
|
|
"step": 5612
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2952,
|
|
"step": 5614
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2874,
|
|
"step": 5616
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2799,
|
|
"step": 5618
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3371,
|
|
"step": 5620
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.312,
|
|
"step": 5622
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3137,
|
|
"step": 5624
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3257,
|
|
"step": 5626
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3201,
|
|
"step": 5628
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3151,
|
|
"step": 5630
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.316,
|
|
"step": 5632
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.295,
|
|
"step": 5634
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3066,
|
|
"step": 5636
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3104,
|
|
"step": 5638
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2908,
|
|
"step": 5640
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3054,
|
|
"step": 5642
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2815,
|
|
"step": 5644
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2836,
|
|
"step": 5646
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2782,
|
|
"step": 5648
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3024,
|
|
"step": 5650
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3243,
|
|
"step": 5652
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2761,
|
|
"step": 5654
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3106,
|
|
"step": 5656
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3077,
|
|
"step": 5658
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2909,
|
|
"step": 5660
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3022,
|
|
"step": 5662
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2974,
|
|
"step": 5664
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.279,
|
|
"step": 5666
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3398,
|
|
"step": 5668
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3129,
|
|
"step": 5670
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.298,
|
|
"step": 5672
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2957,
|
|
"step": 5674
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2911,
|
|
"step": 5676
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2787,
|
|
"step": 5678
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3092,
|
|
"step": 5680
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2767,
|
|
"step": 5682
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3568,
|
|
"step": 5684
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3114,
|
|
"step": 5686
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3318,
|
|
"step": 5688
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2975,
|
|
"step": 5690
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2836,
|
|
"step": 5692
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2694,
|
|
"step": 5694
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3036,
|
|
"step": 5696
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3237,
|
|
"step": 5698
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3073,
|
|
"step": 5700
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.274,
|
|
"step": 5702
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2888,
|
|
"step": 5704
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2882,
|
|
"step": 5706
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2915,
|
|
"step": 5708
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3004,
|
|
"step": 5710
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3033,
|
|
"step": 5712
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3058,
|
|
"step": 5714
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2896,
|
|
"step": 5716
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3328,
|
|
"step": 5718
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2987,
|
|
"step": 5720
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2982,
|
|
"step": 5722
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3223,
|
|
"step": 5724
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2955,
|
|
"step": 5726
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3062,
|
|
"step": 5728
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2966,
|
|
"step": 5730
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3075,
|
|
"step": 5732
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3172,
|
|
"step": 5734
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3076,
|
|
"step": 5736
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2943,
|
|
"step": 5738
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2762,
|
|
"step": 5740
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2838,
|
|
"step": 5742
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2991,
|
|
"step": 5744
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2879,
|
|
"step": 5746
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3008,
|
|
"step": 5748
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2805,
|
|
"step": 5750
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3218,
|
|
"step": 5752
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3383,
|
|
"step": 5754
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2847,
|
|
"step": 5756
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3008,
|
|
"step": 5758
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3027,
|
|
"step": 5760
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2851,
|
|
"step": 5762
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2979,
|
|
"step": 5764
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2825,
|
|
"step": 5766
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3129,
|
|
"step": 5768
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2975,
|
|
"step": 5770
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3045,
|
|
"step": 5772
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2883,
|
|
"step": 5774
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2982,
|
|
"step": 5776
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3177,
|
|
"step": 5778
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3051,
|
|
"step": 5780
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.317,
|
|
"step": 5782
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2969,
|
|
"step": 5784
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3153,
|
|
"step": 5786
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2944,
|
|
"step": 5788
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2878,
|
|
"step": 5790
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3081,
|
|
"step": 5792
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2803,
|
|
"step": 5794
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3235,
|
|
"step": 5796
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2999,
|
|
"step": 5798
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.312,
|
|
"step": 5800
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2858,
|
|
"step": 5802
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3023,
|
|
"step": 5804
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3185,
|
|
"step": 5806
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2831,
|
|
"step": 5808
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2766,
|
|
"step": 5810
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2963,
|
|
"step": 5812
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3007,
|
|
"step": 5814
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2935,
|
|
"step": 5816
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.314,
|
|
"step": 5818
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3181,
|
|
"step": 5820
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3051,
|
|
"step": 5822
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.259,
|
|
"step": 5824
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2925,
|
|
"step": 5826
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3018,
|
|
"step": 5828
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2946,
|
|
"step": 5830
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.312,
|
|
"step": 5832
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2878,
|
|
"step": 5834
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2734,
|
|
"step": 5836
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3344,
|
|
"step": 5838
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3335,
|
|
"step": 5840
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2781,
|
|
"step": 5842
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3006,
|
|
"step": 5844
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3015,
|
|
"step": 5846
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3268,
|
|
"step": 5848
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3172,
|
|
"step": 5850
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2715,
|
|
"step": 5852
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3342,
|
|
"step": 5854
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3129,
|
|
"step": 5856
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2864,
|
|
"step": 5858
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2947,
|
|
"step": 5860
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.335,
|
|
"step": 5862
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3036,
|
|
"step": 5864
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.295,
|
|
"step": 5866
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2933,
|
|
"step": 5868
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3302,
|
|
"step": 5870
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3482,
|
|
"step": 5872
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2927,
|
|
"step": 5874
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2803,
|
|
"step": 5876
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3221,
|
|
"step": 5878
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2917,
|
|
"step": 5880
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.303,
|
|
"step": 5882
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2996,
|
|
"step": 5884
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.314,
|
|
"step": 5886
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3102,
|
|
"step": 5888
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2891,
|
|
"step": 5890
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2825,
|
|
"step": 5892
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3303,
|
|
"step": 5894
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3162,
|
|
"step": 5896
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3347,
|
|
"step": 5898
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3105,
|
|
"step": 5900
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.312,
|
|
"step": 5902
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2847,
|
|
"step": 5904
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3089,
|
|
"step": 5906
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2805,
|
|
"step": 5908
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3004,
|
|
"step": 5910
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3478,
|
|
"step": 5912
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2938,
|
|
"step": 5914
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2855,
|
|
"step": 5916
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3125,
|
|
"step": 5918
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2919,
|
|
"step": 5920
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3206,
|
|
"step": 5922
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2968,
|
|
"step": 5924
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2746,
|
|
"step": 5926
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3087,
|
|
"step": 5928
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2912,
|
|
"step": 5930
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3065,
|
|
"step": 5932
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3168,
|
|
"step": 5934
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2991,
|
|
"step": 5936
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3065,
|
|
"step": 5938
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2989,
|
|
"step": 5940
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3243,
|
|
"step": 5942
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2925,
|
|
"step": 5944
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3205,
|
|
"step": 5946
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2844,
|
|
"step": 5948
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2908,
|
|
"step": 5950
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3022,
|
|
"step": 5952
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.287,
|
|
"step": 5954
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3,
|
|
"step": 5956
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3051,
|
|
"step": 5958
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3074,
|
|
"step": 5960
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2952,
|
|
"step": 5962
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3211,
|
|
"step": 5964
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2924,
|
|
"step": 5966
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2941,
|
|
"step": 5968
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2968,
|
|
"step": 5970
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2987,
|
|
"step": 5972
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3174,
|
|
"step": 5974
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2734,
|
|
"step": 5976
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3019,
|
|
"step": 5978
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2628,
|
|
"step": 5980
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2967,
|
|
"step": 5982
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2955,
|
|
"step": 5984
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2844,
|
|
"step": 5986
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3126,
|
|
"step": 5988
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2771,
|
|
"step": 5990
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2985,
|
|
"step": 5992
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3192,
|
|
"step": 5994
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2783,
|
|
"step": 5996
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2974,
|
|
"step": 5998
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2949,
|
|
"step": 6000
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2914,
|
|
"step": 6002
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2996,
|
|
"step": 6004
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3261,
|
|
"step": 6006
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3009,
|
|
"step": 6008
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2956,
|
|
"step": 6010
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.286,
|
|
"step": 6012
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3068,
|
|
"step": 6014
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3271,
|
|
"step": 6016
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2818,
|
|
"step": 6018
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2873,
|
|
"step": 6020
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.294,
|
|
"step": 6022
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2794,
|
|
"step": 6024
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3324,
|
|
"step": 6026
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2855,
|
|
"step": 6028
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3031,
|
|
"step": 6030
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3032,
|
|
"step": 6032
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3029,
|
|
"step": 6034
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3318,
|
|
"step": 6036
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3229,
|
|
"step": 6038
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2938,
|
|
"step": 6040
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.306,
|
|
"step": 6042
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2832,
|
|
"step": 6044
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2697,
|
|
"step": 6046
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2881,
|
|
"step": 6048
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3063,
|
|
"step": 6050
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2806,
|
|
"step": 6052
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2996,
|
|
"step": 6054
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2968,
|
|
"step": 6056
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2999,
|
|
"step": 6058
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2876,
|
|
"step": 6060
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3011,
|
|
"step": 6062
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3039,
|
|
"step": 6064
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.326,
|
|
"step": 6066
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3242,
|
|
"step": 6068
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2858,
|
|
"step": 6070
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3253,
|
|
"step": 6072
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.286,
|
|
"step": 6074
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2886,
|
|
"step": 6076
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3116,
|
|
"step": 6078
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2842,
|
|
"step": 6080
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2841,
|
|
"step": 6082
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2942,
|
|
"step": 6084
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2822,
|
|
"step": 6086
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2947,
|
|
"step": 6088
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2881,
|
|
"step": 6090
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2938,
|
|
"step": 6092
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2837,
|
|
"step": 6094
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2894,
|
|
"step": 6096
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.299,
|
|
"step": 6098
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3036,
|
|
"step": 6100
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2974,
|
|
"step": 6102
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.306,
|
|
"step": 6104
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2862,
|
|
"step": 6106
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3344,
|
|
"step": 6108
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.34,
|
|
"step": 6110
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3145,
|
|
"step": 6112
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3057,
|
|
"step": 6114
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2934,
|
|
"step": 6116
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.29,
|
|
"step": 6118
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3348,
|
|
"step": 6120
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.276,
|
|
"step": 6122
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3365,
|
|
"step": 6124
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.324,
|
|
"step": 6126
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3035,
|
|
"step": 6128
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3002,
|
|
"step": 6130
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2983,
|
|
"step": 6132
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.277,
|
|
"step": 6134
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3339,
|
|
"step": 6136
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2979,
|
|
"step": 6138
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3008,
|
|
"step": 6140
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2913,
|
|
"step": 6142
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2976,
|
|
"step": 6144
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3307,
|
|
"step": 6146
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3365,
|
|
"step": 6148
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3122,
|
|
"step": 6150
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2885,
|
|
"step": 6152
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3146,
|
|
"step": 6154
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3109,
|
|
"step": 6156
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3127,
|
|
"step": 6158
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2802,
|
|
"step": 6160
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3504,
|
|
"step": 6162
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2946,
|
|
"step": 6164
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2902,
|
|
"step": 6166
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3033,
|
|
"step": 6168
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2997,
|
|
"step": 6170
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3247,
|
|
"step": 6172
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2889,
|
|
"step": 6174
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.307,
|
|
"step": 6176
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2994,
|
|
"step": 6178
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2946,
|
|
"step": 6180
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.29,
|
|
"step": 6182
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.292,
|
|
"step": 6184
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.314,
|
|
"step": 6186
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3017,
|
|
"step": 6188
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3162,
|
|
"step": 6190
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2979,
|
|
"step": 6192
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2816,
|
|
"step": 6194
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3357,
|
|
"step": 6196
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2942,
|
|
"step": 6198
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3037,
|
|
"step": 6200
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2994,
|
|
"step": 6202
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3205,
|
|
"step": 6204
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3095,
|
|
"step": 6206
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.348,
|
|
"step": 6208
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2869,
|
|
"step": 6210
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3115,
|
|
"step": 6212
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3156,
|
|
"step": 6214
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3267,
|
|
"step": 6216
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3465,
|
|
"step": 6218
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3107,
|
|
"step": 6220
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2859,
|
|
"step": 6222
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3228,
|
|
"step": 6224
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2946,
|
|
"step": 6226
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2766,
|
|
"step": 6228
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2972,
|
|
"step": 6230
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3113,
|
|
"step": 6232
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3017,
|
|
"step": 6234
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3174,
|
|
"step": 6236
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.312,
|
|
"step": 6238
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3326,
|
|
"step": 6240
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2966,
|
|
"step": 6242
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2982,
|
|
"step": 6244
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3209,
|
|
"step": 6246
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2935,
|
|
"step": 6248
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3043,
|
|
"step": 6250
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2802,
|
|
"step": 6252
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.308,
|
|
"step": 6254
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3093,
|
|
"step": 6256
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3166,
|
|
"step": 6258
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.319,
|
|
"step": 6260
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3267,
|
|
"step": 6262
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3414,
|
|
"step": 6264
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3197,
|
|
"step": 6266
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3028,
|
|
"step": 6268
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2911,
|
|
"step": 6270
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3159,
|
|
"step": 6272
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2778,
|
|
"step": 6274
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3436,
|
|
"step": 6276
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3228,
|
|
"step": 6278
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3106,
|
|
"step": 6280
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3244,
|
|
"step": 6282
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3262,
|
|
"step": 6284
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2988,
|
|
"step": 6286
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3069,
|
|
"step": 6288
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2893,
|
|
"step": 6290
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3203,
|
|
"step": 6292
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2851,
|
|
"step": 6294
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3358,
|
|
"step": 6296
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3176,
|
|
"step": 6298
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2623,
|
|
"step": 6300
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.328,
|
|
"step": 6302
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2916,
|
|
"step": 6304
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3317,
|
|
"step": 6306
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3278,
|
|
"step": 6308
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2921,
|
|
"step": 6310
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.304,
|
|
"step": 6312
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2726,
|
|
"step": 6314
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3072,
|
|
"step": 6316
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2939,
|
|
"step": 6318
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2945,
|
|
"step": 6320
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2932,
|
|
"step": 6322
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3328,
|
|
"step": 6324
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3671,
|
|
"step": 6326
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2998,
|
|
"step": 6328
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3004,
|
|
"step": 6330
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3012,
|
|
"step": 6332
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.308,
|
|
"step": 6334
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3187,
|
|
"step": 6336
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2977,
|
|
"step": 6338
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2785,
|
|
"step": 6340
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2909,
|
|
"step": 6342
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3176,
|
|
"step": 6344
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2864,
|
|
"step": 6346
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3052,
|
|
"step": 6348
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3364,
|
|
"step": 6350
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3264,
|
|
"step": 6352
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3087,
|
|
"step": 6354
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3276,
|
|
"step": 6356
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2829,
|
|
"step": 6358
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2941,
|
|
"step": 6360
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2874,
|
|
"step": 6362
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3162,
|
|
"step": 6364
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2731,
|
|
"step": 6366
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2986,
|
|
"step": 6368
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3096,
|
|
"step": 6370
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3113,
|
|
"step": 6372
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3228,
|
|
"step": 6374
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3231,
|
|
"step": 6376
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3183,
|
|
"step": 6378
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2976,
|
|
"step": 6380
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3088,
|
|
"step": 6382
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2858,
|
|
"step": 6384
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3021,
|
|
"step": 6386
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3424,
|
|
"step": 6388
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3229,
|
|
"step": 6390
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3223,
|
|
"step": 6392
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3165,
|
|
"step": 6394
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3078,
|
|
"step": 6396
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2957,
|
|
"step": 6398
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3196,
|
|
"step": 6400
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3135,
|
|
"step": 6402
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2856,
|
|
"step": 6404
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3232,
|
|
"step": 6406
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.303,
|
|
"step": 6408
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3068,
|
|
"step": 6410
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3039,
|
|
"step": 6412
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3102,
|
|
"step": 6414
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2877,
|
|
"step": 6416
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3148,
|
|
"step": 6418
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3049,
|
|
"step": 6420
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3132,
|
|
"step": 6422
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2942,
|
|
"step": 6424
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3018,
|
|
"step": 6426
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3176,
|
|
"step": 6428
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3109,
|
|
"step": 6430
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2916,
|
|
"step": 6432
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3232,
|
|
"step": 6434
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2934,
|
|
"step": 6436
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3168,
|
|
"step": 6438
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3015,
|
|
"step": 6440
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3074,
|
|
"step": 6442
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2942,
|
|
"step": 6444
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3054,
|
|
"step": 6446
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3063,
|
|
"step": 6448
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2761,
|
|
"step": 6450
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3069,
|
|
"step": 6452
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2823,
|
|
"step": 6454
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3191,
|
|
"step": 6456
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3027,
|
|
"step": 6458
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3069,
|
|
"step": 6460
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3614,
|
|
"step": 6462
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2717,
|
|
"step": 6464
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3381,
|
|
"step": 6466
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3042,
|
|
"step": 6468
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3188,
|
|
"step": 6470
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3278,
|
|
"step": 6472
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3283,
|
|
"step": 6474
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2861,
|
|
"step": 6476
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2981,
|
|
"step": 6478
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3492,
|
|
"step": 6480
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3089,
|
|
"step": 6482
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3114,
|
|
"step": 6484
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2955,
|
|
"step": 6486
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3059,
|
|
"step": 6488
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2964,
|
|
"step": 6490
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2943,
|
|
"step": 6492
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2948,
|
|
"step": 6494
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2807,
|
|
"step": 6496
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2957,
|
|
"step": 6498
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3099,
|
|
"step": 6500
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3147,
|
|
"step": 6502
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3088,
|
|
"step": 6504
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3324,
|
|
"step": 6506
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3057,
|
|
"step": 6508
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2939,
|
|
"step": 6510
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2974,
|
|
"step": 6512
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3088,
|
|
"step": 6514
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3162,
|
|
"step": 6516
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3099,
|
|
"step": 6518
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2814,
|
|
"step": 6520
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3167,
|
|
"step": 6522
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3191,
|
|
"step": 6524
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3362,
|
|
"step": 6526
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3307,
|
|
"step": 6528
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3297,
|
|
"step": 6530
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3156,
|
|
"step": 6532
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3044,
|
|
"step": 6534
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.294,
|
|
"step": 6536
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2908,
|
|
"step": 6538
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2936,
|
|
"step": 6540
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3103,
|
|
"step": 6542
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2929,
|
|
"step": 6544
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3458,
|
|
"step": 6546
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3019,
|
|
"step": 6548
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2805,
|
|
"step": 6550
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3275,
|
|
"step": 6552
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2926,
|
|
"step": 6554
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3182,
|
|
"step": 6556
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3228,
|
|
"step": 6558
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3316,
|
|
"step": 6560
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2991,
|
|
"step": 6562
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2905,
|
|
"step": 6564
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4047,
|
|
"step": 6566
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.327,
|
|
"step": 6568
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2936,
|
|
"step": 6570
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3143,
|
|
"step": 6572
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3352,
|
|
"step": 6574
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3943,
|
|
"step": 6576
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3148,
|
|
"step": 6578
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3271,
|
|
"step": 6580
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3778,
|
|
"step": 6582
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3183,
|
|
"step": 6584
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3121,
|
|
"step": 6586
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2912,
|
|
"step": 6588
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.305,
|
|
"step": 6590
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3094,
|
|
"step": 6592
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2965,
|
|
"step": 6594
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2766,
|
|
"step": 6596
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3067,
|
|
"step": 6598
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2875,
|
|
"step": 6600
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3018,
|
|
"step": 6602
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3163,
|
|
"step": 6604
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3013,
|
|
"step": 6606
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3228,
|
|
"step": 6608
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2971,
|
|
"step": 6610
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2972,
|
|
"step": 6612
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3002,
|
|
"step": 6614
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3014,
|
|
"step": 6616
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3262,
|
|
"step": 6618
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3076,
|
|
"step": 6620
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2936,
|
|
"step": 6622
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3484,
|
|
"step": 6624
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3173,
|
|
"step": 6626
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3004,
|
|
"step": 6628
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2995,
|
|
"step": 6630
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3029,
|
|
"step": 6632
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3361,
|
|
"step": 6634
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3175,
|
|
"step": 6636
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2958,
|
|
"step": 6638
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2961,
|
|
"step": 6640
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3012,
|
|
"step": 6642
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2989,
|
|
"step": 6644
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.322,
|
|
"step": 6646
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3228,
|
|
"step": 6648
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2871,
|
|
"step": 6650
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2901,
|
|
"step": 6652
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2885,
|
|
"step": 6654
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3118,
|
|
"step": 6656
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2968,
|
|
"step": 6658
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3021,
|
|
"step": 6660
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2999,
|
|
"step": 6662
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3172,
|
|
"step": 6664
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3153,
|
|
"step": 6666
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3224,
|
|
"step": 6668
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2974,
|
|
"step": 6670
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3167,
|
|
"step": 6672
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3291,
|
|
"step": 6674
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3103,
|
|
"step": 6676
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.298,
|
|
"step": 6678
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3196,
|
|
"step": 6680
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.327,
|
|
"step": 6682
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3354,
|
|
"step": 6684
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3008,
|
|
"step": 6686
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3143,
|
|
"step": 6688
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2888,
|
|
"step": 6690
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3168,
|
|
"step": 6692
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3035,
|
|
"step": 6694
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3268,
|
|
"step": 6696
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3113,
|
|
"step": 6698
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3036,
|
|
"step": 6700
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3203,
|
|
"step": 6702
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3207,
|
|
"step": 6704
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3036,
|
|
"step": 6706
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2855,
|
|
"step": 6708
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3308,
|
|
"step": 6710
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3085,
|
|
"step": 6712
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2979,
|
|
"step": 6714
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.296,
|
|
"step": 6716
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3119,
|
|
"step": 6718
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3237,
|
|
"step": 6720
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3112,
|
|
"step": 6722
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3237,
|
|
"step": 6724
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2817,
|
|
"step": 6726
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3299,
|
|
"step": 6728
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3095,
|
|
"step": 6730
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.298,
|
|
"step": 6732
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3403,
|
|
"step": 6734
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3099,
|
|
"step": 6736
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3161,
|
|
"step": 6738
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3214,
|
|
"step": 6740
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2922,
|
|
"step": 6742
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3181,
|
|
"step": 6744
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2809,
|
|
"step": 6746
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3122,
|
|
"step": 6748
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3059,
|
|
"step": 6750
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3185,
|
|
"step": 6752
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3038,
|
|
"step": 6754
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3156,
|
|
"step": 6756
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2948,
|
|
"step": 6758
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2959,
|
|
"step": 6760
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3267,
|
|
"step": 6762
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2914,
|
|
"step": 6764
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2939,
|
|
"step": 6766
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3275,
|
|
"step": 6768
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3483,
|
|
"step": 6770
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2957,
|
|
"step": 6772
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3165,
|
|
"step": 6774
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2999,
|
|
"step": 6776
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.315,
|
|
"step": 6778
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3004,
|
|
"step": 6780
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3076,
|
|
"step": 6782
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3126,
|
|
"step": 6784
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2996,
|
|
"step": 6786
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2864,
|
|
"step": 6788
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3056,
|
|
"step": 6790
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2972,
|
|
"step": 6792
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2902,
|
|
"step": 6794
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3147,
|
|
"step": 6796
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3199,
|
|
"step": 6798
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3124,
|
|
"step": 6800
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3333,
|
|
"step": 6802
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3116,
|
|
"step": 6804
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3052,
|
|
"step": 6806
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.33,
|
|
"step": 6808
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3132,
|
|
"step": 6810
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2765,
|
|
"step": 6812
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3124,
|
|
"step": 6814
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3054,
|
|
"step": 6816
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2986,
|
|
"step": 6818
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2977,
|
|
"step": 6820
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3031,
|
|
"step": 6822
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2952,
|
|
"step": 6824
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2999,
|
|
"step": 6826
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3403,
|
|
"step": 6828
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3363,
|
|
"step": 6830
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3344,
|
|
"step": 6832
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3453,
|
|
"step": 6834
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3167,
|
|
"step": 6836
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3126,
|
|
"step": 6838
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3022,
|
|
"step": 6840
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3129,
|
|
"step": 6842
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2939,
|
|
"step": 6844
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3159,
|
|
"step": 6846
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3149,
|
|
"step": 6848
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3123,
|
|
"step": 6850
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3182,
|
|
"step": 6852
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3188,
|
|
"step": 6854
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3243,
|
|
"step": 6856
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3109,
|
|
"step": 6858
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3125,
|
|
"step": 6860
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2894,
|
|
"step": 6862
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2947,
|
|
"step": 6864
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2961,
|
|
"step": 6866
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3042,
|
|
"step": 6868
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2972,
|
|
"step": 6870
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3351,
|
|
"step": 6872
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3077,
|
|
"step": 6874
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3183,
|
|
"step": 6876
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3069,
|
|
"step": 6878
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3069,
|
|
"step": 6880
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3011,
|
|
"step": 6882
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3163,
|
|
"step": 6884
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2761,
|
|
"step": 6886
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3226,
|
|
"step": 6888
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3147,
|
|
"step": 6890
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3072,
|
|
"step": 6892
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3055,
|
|
"step": 6894
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3181,
|
|
"step": 6896
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3211,
|
|
"step": 6898
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3301,
|
|
"step": 6900
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2656,
|
|
"step": 6902
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2896,
|
|
"step": 6904
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3057,
|
|
"step": 6906
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2956,
|
|
"step": 6908
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3077,
|
|
"step": 6910
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3276,
|
|
"step": 6912
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3295,
|
|
"step": 6914
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2845,
|
|
"step": 6916
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2948,
|
|
"step": 6918
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3235,
|
|
"step": 6920
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3168,
|
|
"step": 6922
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3458,
|
|
"step": 6924
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3094,
|
|
"step": 6926
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3337,
|
|
"step": 6928
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.309,
|
|
"step": 6930
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2962,
|
|
"step": 6932
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3151,
|
|
"step": 6934
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3253,
|
|
"step": 6936
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.31,
|
|
"step": 6938
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3164,
|
|
"step": 6940
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2993,
|
|
"step": 6942
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3055,
|
|
"step": 6944
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2978,
|
|
"step": 6946
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3043,
|
|
"step": 6948
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3058,
|
|
"step": 6950
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3071,
|
|
"step": 6952
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2715,
|
|
"step": 6954
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2693,
|
|
"step": 6956
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3216,
|
|
"step": 6958
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3392,
|
|
"step": 6960
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3128,
|
|
"step": 6962
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3197,
|
|
"step": 6964
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3077,
|
|
"step": 6966
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3154,
|
|
"step": 6968
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2993,
|
|
"step": 6970
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3161,
|
|
"step": 6972
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3188,
|
|
"step": 6974
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.317,
|
|
"step": 6976
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3115,
|
|
"step": 6978
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2971,
|
|
"step": 6980
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2913,
|
|
"step": 6982
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3066,
|
|
"step": 6984
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3038,
|
|
"step": 6986
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3163,
|
|
"step": 6988
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2929,
|
|
"step": 6990
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3229,
|
|
"step": 6992
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3665,
|
|
"step": 6994
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.291,
|
|
"step": 6996
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3477,
|
|
"step": 6998
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3148,
|
|
"step": 7000
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2814,
|
|
"step": 7002
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3093,
|
|
"step": 7004
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3184,
|
|
"step": 7006
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2945,
|
|
"step": 7008
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2745,
|
|
"step": 7010
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3219,
|
|
"step": 7012
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3382,
|
|
"step": 7014
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3145,
|
|
"step": 7016
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3307,
|
|
"step": 7018
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3242,
|
|
"step": 7020
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3267,
|
|
"step": 7022
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2977,
|
|
"step": 7024
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2993,
|
|
"step": 7026
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2974,
|
|
"step": 7028
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2795,
|
|
"step": 7030
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2991,
|
|
"step": 7032
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3071,
|
|
"step": 7034
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3237,
|
|
"step": 7036
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3011,
|
|
"step": 7038
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2991,
|
|
"step": 7040
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3266,
|
|
"step": 7042
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3289,
|
|
"step": 7044
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3179,
|
|
"step": 7046
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.312,
|
|
"step": 7048
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3218,
|
|
"step": 7050
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3199,
|
|
"step": 7052
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3308,
|
|
"step": 7054
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3119,
|
|
"step": 7056
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3302,
|
|
"step": 7058
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.388,
|
|
"step": 7060
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2913,
|
|
"step": 7062
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3459,
|
|
"step": 7064
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3156,
|
|
"step": 7066
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3032,
|
|
"step": 7068
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3145,
|
|
"step": 7070
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3032,
|
|
"step": 7072
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2972,
|
|
"step": 7074
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3021,
|
|
"step": 7076
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3061,
|
|
"step": 7078
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3349,
|
|
"step": 7080
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2787,
|
|
"step": 7082
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3241,
|
|
"step": 7084
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3329,
|
|
"step": 7086
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3195,
|
|
"step": 7088
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3193,
|
|
"step": 7090
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2975,
|
|
"step": 7092
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2952,
|
|
"step": 7094
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3096,
|
|
"step": 7096
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3107,
|
|
"step": 7098
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3053,
|
|
"step": 7100
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2883,
|
|
"step": 7102
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3137,
|
|
"step": 7104
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3435,
|
|
"step": 7106
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2935,
|
|
"step": 7108
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3414,
|
|
"step": 7110
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3086,
|
|
"step": 7112
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3171,
|
|
"step": 7114
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3191,
|
|
"step": 7116
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3001,
|
|
"step": 7118
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3149,
|
|
"step": 7120
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.293,
|
|
"step": 7122
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3024,
|
|
"step": 7124
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3264,
|
|
"step": 7126
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2961,
|
|
"step": 7128
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3224,
|
|
"step": 7130
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3102,
|
|
"step": 7132
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3489,
|
|
"step": 7134
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3149,
|
|
"step": 7136
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3189,
|
|
"step": 7138
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2808,
|
|
"step": 7140
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3055,
|
|
"step": 7142
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.337,
|
|
"step": 7144
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.326,
|
|
"step": 7146
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3117,
|
|
"step": 7148
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3306,
|
|
"step": 7150
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3143,
|
|
"step": 7152
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2992,
|
|
"step": 7154
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3271,
|
|
"step": 7156
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3353,
|
|
"step": 7158
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3307,
|
|
"step": 7160
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3218,
|
|
"step": 7162
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3186,
|
|
"step": 7164
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2922,
|
|
"step": 7166
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3246,
|
|
"step": 7168
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3381,
|
|
"step": 7170
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3311,
|
|
"step": 7172
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3323,
|
|
"step": 7174
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.326,
|
|
"step": 7176
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3055,
|
|
"step": 7178
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3471,
|
|
"step": 7180
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3116,
|
|
"step": 7182
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3149,
|
|
"step": 7184
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3055,
|
|
"step": 7186
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3076,
|
|
"step": 7188
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3389,
|
|
"step": 7190
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3064,
|
|
"step": 7192
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3045,
|
|
"step": 7194
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3935,
|
|
"step": 7196
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3322,
|
|
"step": 7198
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2966,
|
|
"step": 7200
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3106,
|
|
"step": 7202
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3221,
|
|
"step": 7204
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3384,
|
|
"step": 7206
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.308,
|
|
"step": 7208
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3073,
|
|
"step": 7210
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.303,
|
|
"step": 7212
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3274,
|
|
"step": 7214
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.308,
|
|
"step": 7216
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.304,
|
|
"step": 7218
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3348,
|
|
"step": 7220
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3073,
|
|
"step": 7222
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.288,
|
|
"step": 7224
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.302,
|
|
"step": 7226
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3168,
|
|
"step": 7228
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3184,
|
|
"step": 7230
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3994,
|
|
"step": 7232
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3226,
|
|
"step": 7234
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3011,
|
|
"step": 7236
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2953,
|
|
"step": 7238
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2832,
|
|
"step": 7240
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2924,
|
|
"step": 7242
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2973,
|
|
"step": 7244
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2946,
|
|
"step": 7246
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3442,
|
|
"step": 7248
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3066,
|
|
"step": 7250
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3298,
|
|
"step": 7252
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3177,
|
|
"step": 7254
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.319,
|
|
"step": 7256
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3231,
|
|
"step": 7258
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3146,
|
|
"step": 7260
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3297,
|
|
"step": 7262
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2911,
|
|
"step": 7264
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3331,
|
|
"step": 7266
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2893,
|
|
"step": 7268
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2954,
|
|
"step": 7270
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3075,
|
|
"step": 7272
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.314,
|
|
"step": 7274
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3196,
|
|
"step": 7276
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2997,
|
|
"step": 7278
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3041,
|
|
"step": 7280
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3278,
|
|
"step": 7282
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2946,
|
|
"step": 7284
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3008,
|
|
"step": 7286
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3043,
|
|
"step": 7288
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3095,
|
|
"step": 7290
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3142,
|
|
"step": 7292
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3128,
|
|
"step": 7294
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.305,
|
|
"step": 7296
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3223,
|
|
"step": 7298
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3016,
|
|
"step": 7300
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.313,
|
|
"step": 7302
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2994,
|
|
"step": 7304
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3621,
|
|
"step": 7306
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2956,
|
|
"step": 7308
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3165,
|
|
"step": 7310
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3027,
|
|
"step": 7312
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.313,
|
|
"step": 7314
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3221,
|
|
"step": 7316
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3647,
|
|
"step": 7318
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3187,
|
|
"step": 7320
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3178,
|
|
"step": 7322
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.298,
|
|
"step": 7324
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2968,
|
|
"step": 7326
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2865,
|
|
"step": 7328
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3196,
|
|
"step": 7330
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3173,
|
|
"step": 7332
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3307,
|
|
"step": 7334
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3049,
|
|
"step": 7336
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3206,
|
|
"step": 7338
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3035,
|
|
"step": 7340
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3241,
|
|
"step": 7342
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3275,
|
|
"step": 7344
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3388,
|
|
"step": 7346
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3091,
|
|
"step": 7348
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2933,
|
|
"step": 7350
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3152,
|
|
"step": 7352
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3318,
|
|
"step": 7354
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3281,
|
|
"step": 7356
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2861,
|
|
"step": 7358
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3242,
|
|
"step": 7360
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3242,
|
|
"step": 7362
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3102,
|
|
"step": 7364
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3197,
|
|
"step": 7366
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3207,
|
|
"step": 7368
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3154,
|
|
"step": 7370
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3093,
|
|
"step": 7372
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3183,
|
|
"step": 7374
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2837,
|
|
"step": 7376
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2876,
|
|
"step": 7378
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3208,
|
|
"step": 7380
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3435,
|
|
"step": 7382
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3206,
|
|
"step": 7384
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3109,
|
|
"step": 7386
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3147,
|
|
"step": 7388
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2847,
|
|
"step": 7390
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3099,
|
|
"step": 7392
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3164,
|
|
"step": 7394
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3334,
|
|
"step": 7396
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2882,
|
|
"step": 7398
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3234,
|
|
"step": 7400
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3258,
|
|
"step": 7402
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3036,
|
|
"step": 7404
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3468,
|
|
"step": 7406
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3065,
|
|
"step": 7408
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3377,
|
|
"step": 7410
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3185,
|
|
"step": 7412
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3006,
|
|
"step": 7414
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2793,
|
|
"step": 7416
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3136,
|
|
"step": 7418
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3063,
|
|
"step": 7420
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3098,
|
|
"step": 7422
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2961,
|
|
"step": 7424
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.317,
|
|
"step": 7426
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3074,
|
|
"step": 7428
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3193,
|
|
"step": 7430
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3167,
|
|
"step": 7432
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.334,
|
|
"step": 7434
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3082,
|
|
"step": 7436
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2992,
|
|
"step": 7438
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3451,
|
|
"step": 7440
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3644,
|
|
"step": 7442
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3225,
|
|
"step": 7444
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3192,
|
|
"step": 7446
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2914,
|
|
"step": 7448
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3433,
|
|
"step": 7450
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.326,
|
|
"step": 7452
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3339,
|
|
"step": 7454
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3155,
|
|
"step": 7456
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3301,
|
|
"step": 7458
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.299,
|
|
"step": 7460
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2889,
|
|
"step": 7462
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2937,
|
|
"step": 7464
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3242,
|
|
"step": 7466
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3146,
|
|
"step": 7468
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3278,
|
|
"step": 7470
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3318,
|
|
"step": 7472
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3268,
|
|
"step": 7474
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3006,
|
|
"step": 7476
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.301,
|
|
"step": 7478
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2808,
|
|
"step": 7480
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3123,
|
|
"step": 7482
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3016,
|
|
"step": 7484
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3047,
|
|
"step": 7486
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3113,
|
|
"step": 7488
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2988,
|
|
"step": 7490
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3141,
|
|
"step": 7492
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3087,
|
|
"step": 7494
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3065,
|
|
"step": 7496
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3034,
|
|
"step": 7498
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3187,
|
|
"step": 7500
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3402,
|
|
"step": 7502
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3212,
|
|
"step": 7504
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2958,
|
|
"step": 7506
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3453,
|
|
"step": 7508
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.308,
|
|
"step": 7510
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3154,
|
|
"step": 7512
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2953,
|
|
"step": 7514
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3227,
|
|
"step": 7516
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3257,
|
|
"step": 7518
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3093,
|
|
"step": 7520
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2941,
|
|
"step": 7522
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3011,
|
|
"step": 7524
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3344,
|
|
"step": 7526
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.313,
|
|
"step": 7528
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3131,
|
|
"step": 7530
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3271,
|
|
"step": 7532
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2799,
|
|
"step": 7534
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3176,
|
|
"step": 7536
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3345,
|
|
"step": 7538
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3294,
|
|
"step": 7540
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.314,
|
|
"step": 7542
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3035,
|
|
"step": 7544
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3009,
|
|
"step": 7546
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3282,
|
|
"step": 7548
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3497,
|
|
"step": 7550
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.295,
|
|
"step": 7552
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3108,
|
|
"step": 7554
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3022,
|
|
"step": 7556
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2963,
|
|
"step": 7558
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3068,
|
|
"step": 7560
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3232,
|
|
"step": 7562
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3355,
|
|
"step": 7564
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3121,
|
|
"step": 7566
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3265,
|
|
"step": 7568
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.322,
|
|
"step": 7570
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3018,
|
|
"step": 7572
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2952,
|
|
"step": 7574
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3365,
|
|
"step": 7576
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3256,
|
|
"step": 7578
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3169,
|
|
"step": 7580
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2887,
|
|
"step": 7582
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.303,
|
|
"step": 7584
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2979,
|
|
"step": 7586
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2898,
|
|
"step": 7588
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3022,
|
|
"step": 7590
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3339,
|
|
"step": 7592
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3212,
|
|
"step": 7594
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3341,
|
|
"step": 7596
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3408,
|
|
"step": 7598
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3241,
|
|
"step": 7600
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3293,
|
|
"step": 7602
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3049,
|
|
"step": 7604
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2749,
|
|
"step": 7606
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3445,
|
|
"step": 7608
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.298,
|
|
"step": 7610
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.301,
|
|
"step": 7612
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3259,
|
|
"step": 7614
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3226,
|
|
"step": 7616
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3359,
|
|
"step": 7618
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2999,
|
|
"step": 7620
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3486,
|
|
"step": 7622
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3101,
|
|
"step": 7624
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2955,
|
|
"step": 7626
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2972,
|
|
"step": 7628
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3191,
|
|
"step": 7630
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3303,
|
|
"step": 7632
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3314,
|
|
"step": 7634
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.314,
|
|
"step": 7636
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2979,
|
|
"step": 7638
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.304,
|
|
"step": 7640
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.294,
|
|
"step": 7642
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3234,
|
|
"step": 7644
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.337,
|
|
"step": 7646
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3223,
|
|
"step": 7648
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.316,
|
|
"step": 7650
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3362,
|
|
"step": 7652
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3307,
|
|
"step": 7654
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3221,
|
|
"step": 7656
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3112,
|
|
"step": 7658
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3077,
|
|
"step": 7660
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.313,
|
|
"step": 7662
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3597,
|
|
"step": 7664
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2891,
|
|
"step": 7666
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3208,
|
|
"step": 7668
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3165,
|
|
"step": 7670
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3312,
|
|
"step": 7672
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.355,
|
|
"step": 7674
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3023,
|
|
"step": 7676
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3217,
|
|
"step": 7678
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2932,
|
|
"step": 7680
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3358,
|
|
"step": 7682
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2917,
|
|
"step": 7684
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3132,
|
|
"step": 7686
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3064,
|
|
"step": 7688
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2936,
|
|
"step": 7690
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3077,
|
|
"step": 7692
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3246,
|
|
"step": 7694
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3293,
|
|
"step": 7696
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2899,
|
|
"step": 7698
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3373,
|
|
"step": 7700
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2925,
|
|
"step": 7702
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3257,
|
|
"step": 7704
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2868,
|
|
"step": 7706
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2956,
|
|
"step": 7708
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2979,
|
|
"step": 7710
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2941,
|
|
"step": 7712
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.312,
|
|
"step": 7714
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.326,
|
|
"step": 7716
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3045,
|
|
"step": 7718
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3027,
|
|
"step": 7720
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2988,
|
|
"step": 7722
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3077,
|
|
"step": 7724
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3187,
|
|
"step": 7726
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.296,
|
|
"step": 7728
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3076,
|
|
"step": 7730
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3071,
|
|
"step": 7732
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.313,
|
|
"step": 7734
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2937,
|
|
"step": 7736
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2981,
|
|
"step": 7738
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3121,
|
|
"step": 7740
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3208,
|
|
"step": 7742
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3327,
|
|
"step": 7744
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3258,
|
|
"step": 7746
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3151,
|
|
"step": 7748
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3141,
|
|
"step": 7750
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3068,
|
|
"step": 7752
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3368,
|
|
"step": 7754
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3289,
|
|
"step": 7756
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3145,
|
|
"step": 7758
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3339,
|
|
"step": 7760
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3141,
|
|
"step": 7762
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3088,
|
|
"step": 7764
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3008,
|
|
"step": 7766
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3234,
|
|
"step": 7768
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3057,
|
|
"step": 7770
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3105,
|
|
"step": 7772
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2939,
|
|
"step": 7774
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3088,
|
|
"step": 7776
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2777,
|
|
"step": 7778
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3406,
|
|
"step": 7780
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2775,
|
|
"step": 7782
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3393,
|
|
"step": 7784
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3239,
|
|
"step": 7786
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3325,
|
|
"step": 7788
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3051,
|
|
"step": 7790
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2996,
|
|
"step": 7792
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3238,
|
|
"step": 7794
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3227,
|
|
"step": 7796
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2892,
|
|
"step": 7798
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2856,
|
|
"step": 7800
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3245,
|
|
"step": 7802
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3135,
|
|
"step": 7804
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3341,
|
|
"step": 7806
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2974,
|
|
"step": 7808
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3059,
|
|
"step": 7810
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3187,
|
|
"step": 7812
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2772,
|
|
"step": 7814
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3206,
|
|
"step": 7816
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3067,
|
|
"step": 7818
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2989,
|
|
"step": 7820
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3187,
|
|
"step": 7822
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3236,
|
|
"step": 7824
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.302,
|
|
"step": 7826
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3026,
|
|
"step": 7828
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3131,
|
|
"step": 7830
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3615,
|
|
"step": 7832
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3158,
|
|
"step": 7834
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3085,
|
|
"step": 7836
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3005,
|
|
"step": 7838
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3127,
|
|
"step": 7840
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3158,
|
|
"step": 7842
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3244,
|
|
"step": 7844
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3586,
|
|
"step": 7846
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2919,
|
|
"step": 7848
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3008,
|
|
"step": 7850
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3075,
|
|
"step": 7852
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2916,
|
|
"step": 7854
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3312,
|
|
"step": 7856
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3177,
|
|
"step": 7858
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3316,
|
|
"step": 7860
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3135,
|
|
"step": 7862
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3207,
|
|
"step": 7864
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2951,
|
|
"step": 7866
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3176,
|
|
"step": 7868
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.279,
|
|
"step": 7870
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3398,
|
|
"step": 7872
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3336,
|
|
"step": 7874
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2992,
|
|
"step": 7876
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3204,
|
|
"step": 7878
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3116,
|
|
"step": 7880
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3203,
|
|
"step": 7882
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3328,
|
|
"step": 7884
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.342,
|
|
"step": 7886
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.306,
|
|
"step": 7888
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3262,
|
|
"step": 7890
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3234,
|
|
"step": 7892
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3145,
|
|
"step": 7894
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3117,
|
|
"step": 7896
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3004,
|
|
"step": 7898
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2698,
|
|
"step": 7900
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.334,
|
|
"step": 7902
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3232,
|
|
"step": 7904
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3027,
|
|
"step": 7906
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3134,
|
|
"step": 7908
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3162,
|
|
"step": 7910
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2996,
|
|
"step": 7912
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2918,
|
|
"step": 7914
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3292,
|
|
"step": 7916
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2989,
|
|
"step": 7918
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3508,
|
|
"step": 7920
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3065,
|
|
"step": 7922
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.312,
|
|
"step": 7924
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3345,
|
|
"step": 7926
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2777,
|
|
"step": 7928
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3279,
|
|
"step": 7930
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.33,
|
|
"step": 7932
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3143,
|
|
"step": 7934
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3046,
|
|
"step": 7936
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3142,
|
|
"step": 7938
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2979,
|
|
"step": 7940
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2952,
|
|
"step": 7942
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3217,
|
|
"step": 7944
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3314,
|
|
"step": 7946
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3489,
|
|
"step": 7948
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3348,
|
|
"step": 7950
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.283,
|
|
"step": 7952
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3292,
|
|
"step": 7954
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3165,
|
|
"step": 7956
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3425,
|
|
"step": 7958
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3409,
|
|
"step": 7960
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3837,
|
|
"step": 7962
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3065,
|
|
"step": 7964
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3051,
|
|
"step": 7966
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2858,
|
|
"step": 7968
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3251,
|
|
"step": 7970
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3203,
|
|
"step": 7972
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3096,
|
|
"step": 7974
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3413,
|
|
"step": 7976
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3479,
|
|
"step": 7978
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3035,
|
|
"step": 7980
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3173,
|
|
"step": 7982
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3263,
|
|
"step": 7984
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.294,
|
|
"step": 7986
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3208,
|
|
"step": 7988
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2986,
|
|
"step": 7990
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3322,
|
|
"step": 7992
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3507,
|
|
"step": 7994
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2992,
|
|
"step": 7996
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.314,
|
|
"step": 7998
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3345,
|
|
"step": 8000
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3267,
|
|
"step": 8002
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3218,
|
|
"step": 8004
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3106,
|
|
"step": 8006
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3496,
|
|
"step": 8008
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3191,
|
|
"step": 8010
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3059,
|
|
"step": 8012
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3212,
|
|
"step": 8014
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.307,
|
|
"step": 8016
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3431,
|
|
"step": 8018
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3209,
|
|
"step": 8020
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3115,
|
|
"step": 8022
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3105,
|
|
"step": 8024
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3064,
|
|
"step": 8026
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3017,
|
|
"step": 8028
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3106,
|
|
"step": 8030
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.31,
|
|
"step": 8032
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3301,
|
|
"step": 8034
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3342,
|
|
"step": 8036
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3313,
|
|
"step": 8038
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3278,
|
|
"step": 8040
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3145,
|
|
"step": 8042
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3377,
|
|
"step": 8044
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3388,
|
|
"step": 8046
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.316,
|
|
"step": 8048
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3184,
|
|
"step": 8050
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3017,
|
|
"step": 8052
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2792,
|
|
"step": 8054
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3005,
|
|
"step": 8056
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3182,
|
|
"step": 8058
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.323,
|
|
"step": 8060
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2764,
|
|
"step": 8062
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3217,
|
|
"step": 8064
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3293,
|
|
"step": 8066
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3143,
|
|
"step": 8068
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2993,
|
|
"step": 8070
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2934,
|
|
"step": 8072
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3276,
|
|
"step": 8074
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3168,
|
|
"step": 8076
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3045,
|
|
"step": 8078
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3233,
|
|
"step": 8080
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3389,
|
|
"step": 8082
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3063,
|
|
"step": 8084
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3239,
|
|
"step": 8086
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.309,
|
|
"step": 8088
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3033,
|
|
"step": 8090
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3233,
|
|
"step": 8092
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3162,
|
|
"step": 8094
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3075,
|
|
"step": 8096
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3108,
|
|
"step": 8098
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3379,
|
|
"step": 8100
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3167,
|
|
"step": 8102
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3123,
|
|
"step": 8104
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3126,
|
|
"step": 8106
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3457,
|
|
"step": 8108
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3291,
|
|
"step": 8110
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3498,
|
|
"step": 8112
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3252,
|
|
"step": 8114
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3378,
|
|
"step": 8116
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3284,
|
|
"step": 8118
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3303,
|
|
"step": 8120
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3448,
|
|
"step": 8122
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3375,
|
|
"step": 8124
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2983,
|
|
"step": 8126
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3077,
|
|
"step": 8128
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3302,
|
|
"step": 8130
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3253,
|
|
"step": 8132
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3309,
|
|
"step": 8134
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2997,
|
|
"step": 8136
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3206,
|
|
"step": 8138
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3096,
|
|
"step": 8140
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3234,
|
|
"step": 8142
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3326,
|
|
"step": 8144
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.338,
|
|
"step": 8146
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3322,
|
|
"step": 8148
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.339,
|
|
"step": 8150
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.324,
|
|
"step": 8152
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.338,
|
|
"step": 8154
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3433,
|
|
"step": 8156
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3357,
|
|
"step": 8158
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3169,
|
|
"step": 8160
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.343,
|
|
"step": 8162
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3223,
|
|
"step": 8164
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3161,
|
|
"step": 8166
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2898,
|
|
"step": 8168
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3071,
|
|
"step": 8170
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3112,
|
|
"step": 8172
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3168,
|
|
"step": 8174
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3506,
|
|
"step": 8176
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3055,
|
|
"step": 8178
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3238,
|
|
"step": 8180
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3351,
|
|
"step": 8182
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2891,
|
|
"step": 8184
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3307,
|
|
"step": 8186
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2981,
|
|
"step": 8188
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.327,
|
|
"step": 8190
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.305,
|
|
"step": 8192
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2935,
|
|
"step": 8194
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3106,
|
|
"step": 8196
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3225,
|
|
"step": 8198
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3612,
|
|
"step": 8200
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.302,
|
|
"step": 8202
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2981,
|
|
"step": 8204
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3484,
|
|
"step": 8206
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3007,
|
|
"step": 8208
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3281,
|
|
"step": 8210
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2919,
|
|
"step": 8212
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2979,
|
|
"step": 8214
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3115,
|
|
"step": 8216
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3332,
|
|
"step": 8218
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.377,
|
|
"step": 8220
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3336,
|
|
"step": 8222
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.319,
|
|
"step": 8224
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3831,
|
|
"step": 8226
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.312,
|
|
"step": 8228
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2951,
|
|
"step": 8230
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4032,
|
|
"step": 8232
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3018,
|
|
"step": 8234
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2897,
|
|
"step": 8236
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3063,
|
|
"step": 8238
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3309,
|
|
"step": 8240
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3206,
|
|
"step": 8242
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3204,
|
|
"step": 8244
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.301,
|
|
"step": 8246
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3299,
|
|
"step": 8248
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3297,
|
|
"step": 8250
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2882,
|
|
"step": 8252
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3071,
|
|
"step": 8254
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.322,
|
|
"step": 8256
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2956,
|
|
"step": 8258
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2902,
|
|
"step": 8260
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3059,
|
|
"step": 8262
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3367,
|
|
"step": 8264
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3141,
|
|
"step": 8266
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3266,
|
|
"step": 8268
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3113,
|
|
"step": 8270
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3261,
|
|
"step": 8272
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3168,
|
|
"step": 8274
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.304,
|
|
"step": 8276
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3228,
|
|
"step": 8278
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3739,
|
|
"step": 8280
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3334,
|
|
"step": 8282
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.34,
|
|
"step": 8284
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3205,
|
|
"step": 8286
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2834,
|
|
"step": 8288
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3199,
|
|
"step": 8290
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3359,
|
|
"step": 8292
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3258,
|
|
"step": 8294
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3383,
|
|
"step": 8296
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2912,
|
|
"step": 8298
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3478,
|
|
"step": 8300
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2885,
|
|
"step": 8302
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3116,
|
|
"step": 8304
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3306,
|
|
"step": 8306
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2834,
|
|
"step": 8308
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3273,
|
|
"step": 8310
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3257,
|
|
"step": 8312
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.33,
|
|
"step": 8314
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3118,
|
|
"step": 8316
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3035,
|
|
"step": 8318
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3231,
|
|
"step": 8320
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3164,
|
|
"step": 8322
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3011,
|
|
"step": 8324
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.317,
|
|
"step": 8326
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2972,
|
|
"step": 8328
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.34,
|
|
"step": 8330
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3232,
|
|
"step": 8332
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3114,
|
|
"step": 8334
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3158,
|
|
"step": 8336
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3552,
|
|
"step": 8338
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3027,
|
|
"step": 8340
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2825,
|
|
"step": 8342
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3004,
|
|
"step": 8344
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3141,
|
|
"step": 8346
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3189,
|
|
"step": 8348
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3196,
|
|
"step": 8350
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3265,
|
|
"step": 8352
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3408,
|
|
"step": 8354
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3015,
|
|
"step": 8356
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3293,
|
|
"step": 8358
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3168,
|
|
"step": 8360
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3074,
|
|
"step": 8362
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2983,
|
|
"step": 8364
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.313,
|
|
"step": 8366
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3134,
|
|
"step": 8368
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3207,
|
|
"step": 8370
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3061,
|
|
"step": 8372
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.333,
|
|
"step": 8374
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3364,
|
|
"step": 8376
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3474,
|
|
"step": 8378
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3478,
|
|
"step": 8380
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3132,
|
|
"step": 8382
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3124,
|
|
"step": 8384
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3519,
|
|
"step": 8386
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3597,
|
|
"step": 8388
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3362,
|
|
"step": 8390
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3563,
|
|
"step": 8392
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3243,
|
|
"step": 8394
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3436,
|
|
"step": 8396
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3079,
|
|
"step": 8398
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3104,
|
|
"step": 8400
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3575,
|
|
"step": 8402
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3126,
|
|
"step": 8404
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2909,
|
|
"step": 8406
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.324,
|
|
"step": 8408
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3199,
|
|
"step": 8410
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3225,
|
|
"step": 8412
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2661,
|
|
"step": 8414
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3148,
|
|
"step": 8416
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.308,
|
|
"step": 8418
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3179,
|
|
"step": 8420
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2874,
|
|
"step": 8422
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3203,
|
|
"step": 8424
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2984,
|
|
"step": 8426
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3169,
|
|
"step": 8428
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3369,
|
|
"step": 8430
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3072,
|
|
"step": 8432
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3149,
|
|
"step": 8434
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3174,
|
|
"step": 8436
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3055,
|
|
"step": 8438
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3032,
|
|
"step": 8440
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3217,
|
|
"step": 8442
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3231,
|
|
"step": 8444
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3009,
|
|
"step": 8446
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3215,
|
|
"step": 8448
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3221,
|
|
"step": 8450
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2833,
|
|
"step": 8452
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2797,
|
|
"step": 8454
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.305,
|
|
"step": 8456
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3237,
|
|
"step": 8458
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3387,
|
|
"step": 8460
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2953,
|
|
"step": 8462
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3175,
|
|
"step": 8464
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3235,
|
|
"step": 8466
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2925,
|
|
"step": 8468
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3134,
|
|
"step": 8470
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3033,
|
|
"step": 8472
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3373,
|
|
"step": 8474
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3054,
|
|
"step": 8476
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3072,
|
|
"step": 8478
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3227,
|
|
"step": 8480
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3199,
|
|
"step": 8482
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3106,
|
|
"step": 8484
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3328,
|
|
"step": 8486
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3156,
|
|
"step": 8488
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3356,
|
|
"step": 8490
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3272,
|
|
"step": 8492
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3298,
|
|
"step": 8494
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3173,
|
|
"step": 8496
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2958,
|
|
"step": 8498
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3079,
|
|
"step": 8500
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3221,
|
|
"step": 8502
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3033,
|
|
"step": 8504
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2978,
|
|
"step": 8506
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3293,
|
|
"step": 8508
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3166,
|
|
"step": 8510
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3302,
|
|
"step": 8512
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2997,
|
|
"step": 8514
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3066,
|
|
"step": 8516
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3351,
|
|
"step": 8518
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3295,
|
|
"step": 8520
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3469,
|
|
"step": 8522
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2721,
|
|
"step": 8524
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3422,
|
|
"step": 8526
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3586,
|
|
"step": 8528
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3467,
|
|
"step": 8530
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3242,
|
|
"step": 8532
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3088,
|
|
"step": 8534
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2964,
|
|
"step": 8536
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3027,
|
|
"step": 8538
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3036,
|
|
"step": 8540
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3408,
|
|
"step": 8542
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3062,
|
|
"step": 8544
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.323,
|
|
"step": 8546
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3195,
|
|
"step": 8548
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.319,
|
|
"step": 8550
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2748,
|
|
"step": 8552
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3212,
|
|
"step": 8554
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3222,
|
|
"step": 8556
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3163,
|
|
"step": 8558
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2977,
|
|
"step": 8560
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3257,
|
|
"step": 8562
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3123,
|
|
"step": 8564
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3141,
|
|
"step": 8566
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.321,
|
|
"step": 8568
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.295,
|
|
"step": 8570
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3094,
|
|
"step": 8572
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3079,
|
|
"step": 8574
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3271,
|
|
"step": 8576
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3259,
|
|
"step": 8578
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2954,
|
|
"step": 8580
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3244,
|
|
"step": 8582
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3089,
|
|
"step": 8584
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3021,
|
|
"step": 8586
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3079,
|
|
"step": 8588
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3231,
|
|
"step": 8590
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3099,
|
|
"step": 8592
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3466,
|
|
"step": 8594
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3103,
|
|
"step": 8596
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3186,
|
|
"step": 8598
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3005,
|
|
"step": 8600
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3151,
|
|
"step": 8602
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.343,
|
|
"step": 8604
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3444,
|
|
"step": 8606
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3137,
|
|
"step": 8608
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2966,
|
|
"step": 8610
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3265,
|
|
"step": 8612
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2999,
|
|
"step": 8614
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3232,
|
|
"step": 8616
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2993,
|
|
"step": 8618
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3137,
|
|
"step": 8620
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2971,
|
|
"step": 8622
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3266,
|
|
"step": 8624
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2983,
|
|
"step": 8626
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3273,
|
|
"step": 8628
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.325,
|
|
"step": 8630
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3079,
|
|
"step": 8632
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2918,
|
|
"step": 8634
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3128,
|
|
"step": 8636
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2919,
|
|
"step": 8638
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3373,
|
|
"step": 8640
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3412,
|
|
"step": 8642
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3389,
|
|
"step": 8644
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3026,
|
|
"step": 8646
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3279,
|
|
"step": 8648
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2961,
|
|
"step": 8650
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3322,
|
|
"step": 8652
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3152,
|
|
"step": 8654
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3171,
|
|
"step": 8656
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2786,
|
|
"step": 8658
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.321,
|
|
"step": 8660
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3376,
|
|
"step": 8662
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3184,
|
|
"step": 8664
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3436,
|
|
"step": 8666
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3043,
|
|
"step": 8668
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3466,
|
|
"step": 8670
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3376,
|
|
"step": 8672
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3132,
|
|
"step": 8674
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3124,
|
|
"step": 8676
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3049,
|
|
"step": 8678
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3264,
|
|
"step": 8680
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3091,
|
|
"step": 8682
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3089,
|
|
"step": 8684
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3181,
|
|
"step": 8686
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3531,
|
|
"step": 8688
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3429,
|
|
"step": 8690
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2991,
|
|
"step": 8692
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3328,
|
|
"step": 8694
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3024,
|
|
"step": 8696
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3286,
|
|
"step": 8698
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3076,
|
|
"step": 8700
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3049,
|
|
"step": 8702
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2948,
|
|
"step": 8704
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3314,
|
|
"step": 8706
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3463,
|
|
"step": 8708
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3237,
|
|
"step": 8710
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3078,
|
|
"step": 8712
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.314,
|
|
"step": 8714
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.329,
|
|
"step": 8716
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3413,
|
|
"step": 8718
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2979,
|
|
"step": 8720
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3573,
|
|
"step": 8722
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3153,
|
|
"step": 8724
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2728,
|
|
"step": 8726
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3394,
|
|
"step": 8728
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3279,
|
|
"step": 8730
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3303,
|
|
"step": 8732
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3347,
|
|
"step": 8734
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3308,
|
|
"step": 8736
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2935,
|
|
"step": 8738
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.316,
|
|
"step": 8740
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3239,
|
|
"step": 8742
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3326,
|
|
"step": 8744
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3064,
|
|
"step": 8746
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.317,
|
|
"step": 8748
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.34,
|
|
"step": 8750
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3154,
|
|
"step": 8752
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2933,
|
|
"step": 8754
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.296,
|
|
"step": 8756
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3199,
|
|
"step": 8758
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3243,
|
|
"step": 8760
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3243,
|
|
"step": 8762
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.289,
|
|
"step": 8764
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3225,
|
|
"step": 8766
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3257,
|
|
"step": 8768
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3311,
|
|
"step": 8770
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3027,
|
|
"step": 8772
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3183,
|
|
"step": 8774
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3226,
|
|
"step": 8776
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3035,
|
|
"step": 8778
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3345,
|
|
"step": 8780
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3271,
|
|
"step": 8782
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.296,
|
|
"step": 8784
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3483,
|
|
"step": 8786
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2997,
|
|
"step": 8788
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3024,
|
|
"step": 8790
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3301,
|
|
"step": 8792
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3046,
|
|
"step": 8794
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3196,
|
|
"step": 8796
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3472,
|
|
"step": 8798
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.316,
|
|
"step": 8800
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3271,
|
|
"step": 8802
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.337,
|
|
"step": 8804
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3064,
|
|
"step": 8806
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.316,
|
|
"step": 8808
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3162,
|
|
"step": 8810
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3157,
|
|
"step": 8812
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3098,
|
|
"step": 8814
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3145,
|
|
"step": 8816
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3218,
|
|
"step": 8818
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3177,
|
|
"step": 8820
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3028,
|
|
"step": 8822
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3505,
|
|
"step": 8824
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3298,
|
|
"step": 8826
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3514,
|
|
"step": 8828
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3281,
|
|
"step": 8830
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3439,
|
|
"step": 8832
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3218,
|
|
"step": 8834
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3065,
|
|
"step": 8836
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3101,
|
|
"step": 8838
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3107,
|
|
"step": 8840
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3092,
|
|
"step": 8842
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3248,
|
|
"step": 8844
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3197,
|
|
"step": 8846
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3372,
|
|
"step": 8848
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2885,
|
|
"step": 8850
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2982,
|
|
"step": 8852
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3033,
|
|
"step": 8854
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3098,
|
|
"step": 8856
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3107,
|
|
"step": 8858
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3299,
|
|
"step": 8860
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3212,
|
|
"step": 8862
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.319,
|
|
"step": 8864
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3218,
|
|
"step": 8866
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3077,
|
|
"step": 8868
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3141,
|
|
"step": 8870
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3242,
|
|
"step": 8872
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3071,
|
|
"step": 8874
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.302,
|
|
"step": 8876
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.4326,
|
|
"step": 8878
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3168,
|
|
"step": 8880
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2948,
|
|
"step": 8882
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3036,
|
|
"step": 8884
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2972,
|
|
"step": 8886
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2974,
|
|
"step": 8888
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3198,
|
|
"step": 8890
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3236,
|
|
"step": 8892
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2983,
|
|
"step": 8894
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3285,
|
|
"step": 8896
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3348,
|
|
"step": 8898
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3514,
|
|
"step": 8900
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3038,
|
|
"step": 8902
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2763,
|
|
"step": 8904
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3087,
|
|
"step": 8906
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2999,
|
|
"step": 8908
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3279,
|
|
"step": 8910
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3463,
|
|
"step": 8912
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3114,
|
|
"step": 8914
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3083,
|
|
"step": 8916
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2863,
|
|
"step": 8918
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3309,
|
|
"step": 8920
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2893,
|
|
"step": 8922
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3089,
|
|
"step": 8924
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.329,
|
|
"step": 8926
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3345,
|
|
"step": 8928
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3081,
|
|
"step": 8930
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3262,
|
|
"step": 8932
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3538,
|
|
"step": 8934
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3276,
|
|
"step": 8936
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3301,
|
|
"step": 8938
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3002,
|
|
"step": 8940
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3074,
|
|
"step": 8942
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3195,
|
|
"step": 8944
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3391,
|
|
"step": 8946
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3362,
|
|
"step": 8948
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3167,
|
|
"step": 8950
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3297,
|
|
"step": 8952
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3195,
|
|
"step": 8954
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3278,
|
|
"step": 8956
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2866,
|
|
"step": 8958
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.318,
|
|
"step": 8960
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3096,
|
|
"step": 8962
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.329,
|
|
"step": 8964
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3406,
|
|
"step": 8966
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3235,
|
|
"step": 8968
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3262,
|
|
"step": 8970
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3121,
|
|
"step": 8972
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3179,
|
|
"step": 8974
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3013,
|
|
"step": 8976
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2993,
|
|
"step": 8978
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3509,
|
|
"step": 8980
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3062,
|
|
"step": 8982
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2922,
|
|
"step": 8984
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3352,
|
|
"step": 8986
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.337,
|
|
"step": 8988
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3398,
|
|
"step": 8990
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3358,
|
|
"step": 8992
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3426,
|
|
"step": 8994
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3313,
|
|
"step": 8996
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.32,
|
|
"step": 8998
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.319,
|
|
"step": 9000
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3093,
|
|
"step": 9002
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.293,
|
|
"step": 9004
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.31,
|
|
"step": 9006
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3194,
|
|
"step": 9008
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2949,
|
|
"step": 9010
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3074,
|
|
"step": 9012
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3104,
|
|
"step": 9014
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3193,
|
|
"step": 9016
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3243,
|
|
"step": 9018
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3113,
|
|
"step": 9020
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2925,
|
|
"step": 9022
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2899,
|
|
"step": 9024
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3361,
|
|
"step": 9026
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3459,
|
|
"step": 9028
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2885,
|
|
"step": 9030
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3354,
|
|
"step": 9032
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3152,
|
|
"step": 9034
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3134,
|
|
"step": 9036
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2863,
|
|
"step": 9038
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3066,
|
|
"step": 9040
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2837,
|
|
"step": 9042
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3351,
|
|
"step": 9044
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3339,
|
|
"step": 9046
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3378,
|
|
"step": 9048
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3291,
|
|
"step": 9050
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3104,
|
|
"step": 9052
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.314,
|
|
"step": 9054
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3243,
|
|
"step": 9056
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3197,
|
|
"step": 9058
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.316,
|
|
"step": 9060
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3263,
|
|
"step": 9062
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3181,
|
|
"step": 9064
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3253,
|
|
"step": 9066
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3445,
|
|
"step": 9068
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3328,
|
|
"step": 9070
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3131,
|
|
"step": 9072
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3247,
|
|
"step": 9074
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3063,
|
|
"step": 9076
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3198,
|
|
"step": 9078
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3048,
|
|
"step": 9080
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3244,
|
|
"step": 9082
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3301,
|
|
"step": 9084
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3568,
|
|
"step": 9086
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2819,
|
|
"step": 9088
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.314,
|
|
"step": 9090
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.305,
|
|
"step": 9092
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3064,
|
|
"step": 9094
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3192,
|
|
"step": 9096
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3204,
|
|
"step": 9098
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3034,
|
|
"step": 9100
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3114,
|
|
"step": 9102
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3149,
|
|
"step": 9104
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.301,
|
|
"step": 9106
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3074,
|
|
"step": 9108
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3007,
|
|
"step": 9110
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3073,
|
|
"step": 9112
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3074,
|
|
"step": 9114
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3215,
|
|
"step": 9116
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3663,
|
|
"step": 9118
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3171,
|
|
"step": 9120
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3309,
|
|
"step": 9122
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3322,
|
|
"step": 9124
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3069,
|
|
"step": 9126
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2963,
|
|
"step": 9128
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3065,
|
|
"step": 9130
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3007,
|
|
"step": 9132
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2926,
|
|
"step": 9134
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3541,
|
|
"step": 9136
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3106,
|
|
"step": 9138
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3216,
|
|
"step": 9140
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2941,
|
|
"step": 9142
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3223,
|
|
"step": 9144
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3189,
|
|
"step": 9146
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3155,
|
|
"step": 9148
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3207,
|
|
"step": 9150
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3062,
|
|
"step": 9152
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3065,
|
|
"step": 9154
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3353,
|
|
"step": 9156
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3003,
|
|
"step": 9158
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3041,
|
|
"step": 9160
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3162,
|
|
"step": 9162
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3021,
|
|
"step": 9164
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3158,
|
|
"step": 9166
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2863,
|
|
"step": 9168
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3342,
|
|
"step": 9170
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.317,
|
|
"step": 9172
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3015,
|
|
"step": 9174
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3117,
|
|
"step": 9176
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3282,
|
|
"step": 9178
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3284,
|
|
"step": 9180
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3107,
|
|
"step": 9182
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.312,
|
|
"step": 9184
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3449,
|
|
"step": 9186
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3126,
|
|
"step": 9188
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3195,
|
|
"step": 9190
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3056,
|
|
"step": 9192
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3314,
|
|
"step": 9194
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.329,
|
|
"step": 9196
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3142,
|
|
"step": 9198
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3179,
|
|
"step": 9200
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2926,
|
|
"step": 9202
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3264,
|
|
"step": 9204
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3423,
|
|
"step": 9206
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.301,
|
|
"step": 9208
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3226,
|
|
"step": 9210
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2858,
|
|
"step": 9212
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3123,
|
|
"step": 9214
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3289,
|
|
"step": 9216
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3245,
|
|
"step": 9218
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3431,
|
|
"step": 9220
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3624,
|
|
"step": 9222
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3171,
|
|
"step": 9224
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3542,
|
|
"step": 9226
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2937,
|
|
"step": 9228
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3411,
|
|
"step": 9230
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2882,
|
|
"step": 9232
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3173,
|
|
"step": 9234
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3327,
|
|
"step": 9236
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3113,
|
|
"step": 9238
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3277,
|
|
"step": 9240
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2926,
|
|
"step": 9242
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2952,
|
|
"step": 9244
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3005,
|
|
"step": 9246
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3625,
|
|
"step": 9248
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3273,
|
|
"step": 9250
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3257,
|
|
"step": 9252
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3096,
|
|
"step": 9254
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3139,
|
|
"step": 9256
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2905,
|
|
"step": 9258
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3613,
|
|
"step": 9260
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3379,
|
|
"step": 9262
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3383,
|
|
"step": 9264
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3202,
|
|
"step": 9266
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3633,
|
|
"step": 9268
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3034,
|
|
"step": 9270
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3109,
|
|
"step": 9272
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2983,
|
|
"step": 9274
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3234,
|
|
"step": 9276
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2972,
|
|
"step": 9278
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2952,
|
|
"step": 9280
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3189,
|
|
"step": 9282
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3008,
|
|
"step": 9284
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3256,
|
|
"step": 9286
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3124,
|
|
"step": 9288
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3018,
|
|
"step": 9290
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3115,
|
|
"step": 9292
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3151,
|
|
"step": 9294
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3146,
|
|
"step": 9296
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2906,
|
|
"step": 9298
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3282,
|
|
"step": 9300
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3138,
|
|
"step": 9302
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2957,
|
|
"step": 9304
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3049,
|
|
"step": 9306
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.321,
|
|
"step": 9308
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3259,
|
|
"step": 9310
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3187,
|
|
"step": 9312
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.308,
|
|
"step": 9314
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2992,
|
|
"step": 9316
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3109,
|
|
"step": 9318
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3287,
|
|
"step": 9320
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3083,
|
|
"step": 9322
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3207,
|
|
"step": 9324
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3109,
|
|
"step": 9326
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3359,
|
|
"step": 9328
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3051,
|
|
"step": 9330
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3288,
|
|
"step": 9332
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.312,
|
|
"step": 9334
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2937,
|
|
"step": 9336
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2897,
|
|
"step": 9338
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3012,
|
|
"step": 9340
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3124,
|
|
"step": 9342
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3273,
|
|
"step": 9344
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.319,
|
|
"step": 9346
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3011,
|
|
"step": 9348
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.324,
|
|
"step": 9350
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3218,
|
|
"step": 9352
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3434,
|
|
"step": 9354
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3413,
|
|
"step": 9356
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3355,
|
|
"step": 9358
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3167,
|
|
"step": 9360
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3208,
|
|
"step": 9362
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.31,
|
|
"step": 9364
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3257,
|
|
"step": 9366
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3086,
|
|
"step": 9368
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3007,
|
|
"step": 9370
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3069,
|
|
"step": 9372
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3129,
|
|
"step": 9374
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3313,
|
|
"step": 9376
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3183,
|
|
"step": 9378
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3263,
|
|
"step": 9380
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3403,
|
|
"step": 9382
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.33,
|
|
"step": 9384
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.311,
|
|
"step": 9386
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2877,
|
|
"step": 9388
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3533,
|
|
"step": 9390
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.316,
|
|
"step": 9392
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3212,
|
|
"step": 9394
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3041,
|
|
"step": 9396
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3292,
|
|
"step": 9398
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3163,
|
|
"step": 9400
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2983,
|
|
"step": 9402
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.319,
|
|
"step": 9404
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3293,
|
|
"step": 9406
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3347,
|
|
"step": 9408
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2941,
|
|
"step": 9410
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3397,
|
|
"step": 9412
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3315,
|
|
"step": 9414
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2887,
|
|
"step": 9416
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3094,
|
|
"step": 9418
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.296,
|
|
"step": 9420
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3182,
|
|
"step": 9422
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3175,
|
|
"step": 9424
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3242,
|
|
"step": 9426
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3074,
|
|
"step": 9428
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.327,
|
|
"step": 9430
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3185,
|
|
"step": 9432
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3337,
|
|
"step": 9434
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3221,
|
|
"step": 9436
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3055,
|
|
"step": 9438
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3193,
|
|
"step": 9440
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3301,
|
|
"step": 9442
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3396,
|
|
"step": 9444
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3281,
|
|
"step": 9446
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.322,
|
|
"step": 9448
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3047,
|
|
"step": 9450
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3256,
|
|
"step": 9452
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.331,
|
|
"step": 9454
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3462,
|
|
"step": 9456
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3422,
|
|
"step": 9458
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3384,
|
|
"step": 9460
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3286,
|
|
"step": 9462
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3361,
|
|
"step": 9464
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3254,
|
|
"step": 9466
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.32,
|
|
"step": 9468
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.323,
|
|
"step": 9470
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3106,
|
|
"step": 9472
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3036,
|
|
"step": 9474
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.314,
|
|
"step": 9476
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2986,
|
|
"step": 9478
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3156,
|
|
"step": 9480
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3169,
|
|
"step": 9482
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3255,
|
|
"step": 9484
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.305,
|
|
"step": 9486
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2944,
|
|
"step": 9488
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3126,
|
|
"step": 9490
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2907,
|
|
"step": 9492
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2845,
|
|
"step": 9494
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3103,
|
|
"step": 9496
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.312,
|
|
"step": 9498
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3307,
|
|
"step": 9500
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3037,
|
|
"step": 9502
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.314,
|
|
"step": 9504
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3245,
|
|
"step": 9506
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3098,
|
|
"step": 9508
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2949,
|
|
"step": 9510
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3194,
|
|
"step": 9512
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3,
|
|
"step": 9514
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3159,
|
|
"step": 9516
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3215,
|
|
"step": 9518
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3037,
|
|
"step": 9520
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2968,
|
|
"step": 9522
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3262,
|
|
"step": 9524
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3092,
|
|
"step": 9526
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.314,
|
|
"step": 9528
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3112,
|
|
"step": 9530
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3192,
|
|
"step": 9532
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3196,
|
|
"step": 9534
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3213,
|
|
"step": 9536
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3147,
|
|
"step": 9538
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3512,
|
|
"step": 9540
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3388,
|
|
"step": 9542
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3141,
|
|
"step": 9544
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.339,
|
|
"step": 9546
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3345,
|
|
"step": 9548
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3328,
|
|
"step": 9550
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.305,
|
|
"step": 9552
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3088,
|
|
"step": 9554
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2993,
|
|
"step": 9556
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3351,
|
|
"step": 9558
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.325,
|
|
"step": 9560
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3253,
|
|
"step": 9562
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3364,
|
|
"step": 9564
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2847,
|
|
"step": 9566
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3168,
|
|
"step": 9568
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3223,
|
|
"step": 9570
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3035,
|
|
"step": 9572
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.303,
|
|
"step": 9574
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3197,
|
|
"step": 9576
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.289,
|
|
"step": 9578
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.36,
|
|
"step": 9580
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.335,
|
|
"step": 9582
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2977,
|
|
"step": 9584
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3006,
|
|
"step": 9586
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3081,
|
|
"step": 9588
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3461,
|
|
"step": 9590
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3343,
|
|
"step": 9592
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3172,
|
|
"step": 9594
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2862,
|
|
"step": 9596
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3239,
|
|
"step": 9598
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3267,
|
|
"step": 9600
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3226,
|
|
"step": 9602
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3633,
|
|
"step": 9604
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3481,
|
|
"step": 9606
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2913,
|
|
"step": 9608
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3188,
|
|
"step": 9610
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3355,
|
|
"step": 9612
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3549,
|
|
"step": 9614
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3158,
|
|
"step": 9616
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3283,
|
|
"step": 9618
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3277,
|
|
"step": 9620
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.317,
|
|
"step": 9622
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3262,
|
|
"step": 9624
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3361,
|
|
"step": 9626
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.328,
|
|
"step": 9628
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3018,
|
|
"step": 9630
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3309,
|
|
"step": 9632
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3057,
|
|
"step": 9634
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3378,
|
|
"step": 9636
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3318,
|
|
"step": 9638
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2965,
|
|
"step": 9640
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3347,
|
|
"step": 9642
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3356,
|
|
"step": 9644
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3332,
|
|
"step": 9646
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3632,
|
|
"step": 9648
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3826,
|
|
"step": 9650
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3343,
|
|
"step": 9652
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3396,
|
|
"step": 9654
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3356,
|
|
"step": 9656
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3105,
|
|
"step": 9658
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3022,
|
|
"step": 9660
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3054,
|
|
"step": 9662
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2987,
|
|
"step": 9664
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3173,
|
|
"step": 9666
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3477,
|
|
"step": 9668
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3663,
|
|
"step": 9670
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3236,
|
|
"step": 9672
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2922,
|
|
"step": 9674
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3238,
|
|
"step": 9676
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3006,
|
|
"step": 9678
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3333,
|
|
"step": 9680
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3147,
|
|
"step": 9682
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3074,
|
|
"step": 9684
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3584,
|
|
"step": 9686
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3076,
|
|
"step": 9688
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3075,
|
|
"step": 9690
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.293,
|
|
"step": 9692
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3355,
|
|
"step": 9694
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3166,
|
|
"step": 9696
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3183,
|
|
"step": 9698
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.327,
|
|
"step": 9700
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3366,
|
|
"step": 9702
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3179,
|
|
"step": 9704
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3431,
|
|
"step": 9706
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3358,
|
|
"step": 9708
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3398,
|
|
"step": 9710
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3305,
|
|
"step": 9712
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3032,
|
|
"step": 9714
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3328,
|
|
"step": 9716
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3456,
|
|
"step": 9718
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3116,
|
|
"step": 9720
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3143,
|
|
"step": 9722
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3327,
|
|
"step": 9724
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.318,
|
|
"step": 9726
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3228,
|
|
"step": 9728
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.334,
|
|
"step": 9730
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3289,
|
|
"step": 9732
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3447,
|
|
"step": 9734
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3394,
|
|
"step": 9736
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.293,
|
|
"step": 9738
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3225,
|
|
"step": 9740
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3287,
|
|
"step": 9742
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2915,
|
|
"step": 9744
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3342,
|
|
"step": 9746
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3205,
|
|
"step": 9748
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3415,
|
|
"step": 9750
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3271,
|
|
"step": 9752
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3481,
|
|
"step": 9754
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3263,
|
|
"step": 9756
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3005,
|
|
"step": 9758
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3019,
|
|
"step": 9760
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2886,
|
|
"step": 9762
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2883,
|
|
"step": 9764
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3375,
|
|
"step": 9766
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3386,
|
|
"step": 9768
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3055,
|
|
"step": 9770
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3561,
|
|
"step": 9772
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3353,
|
|
"step": 9774
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3517,
|
|
"step": 9776
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3243,
|
|
"step": 9778
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3547,
|
|
"step": 9780
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3073,
|
|
"step": 9782
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3334,
|
|
"step": 9784
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3248,
|
|
"step": 9786
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3328,
|
|
"step": 9788
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3098,
|
|
"step": 9790
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3281,
|
|
"step": 9792
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3237,
|
|
"step": 9794
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2739,
|
|
"step": 9796
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.329,
|
|
"step": 9798
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.355,
|
|
"step": 9800
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3157,
|
|
"step": 9802
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2874,
|
|
"step": 9804
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3218,
|
|
"step": 9806
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3342,
|
|
"step": 9808
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3331,
|
|
"step": 9810
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3161,
|
|
"step": 9812
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3424,
|
|
"step": 9814
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3419,
|
|
"step": 9816
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3359,
|
|
"step": 9818
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3309,
|
|
"step": 9820
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3215,
|
|
"step": 9822
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3305,
|
|
"step": 9824
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.318,
|
|
"step": 9826
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3048,
|
|
"step": 9828
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3018,
|
|
"step": 9830
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2963,
|
|
"step": 9832
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2989,
|
|
"step": 9834
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3284,
|
|
"step": 9836
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2944,
|
|
"step": 9838
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2911,
|
|
"step": 9840
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3255,
|
|
"step": 9842
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3051,
|
|
"step": 9844
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3065,
|
|
"step": 9846
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3345,
|
|
"step": 9848
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3276,
|
|
"step": 9850
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3244,
|
|
"step": 9852
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3309,
|
|
"step": 9854
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3314,
|
|
"step": 9856
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3485,
|
|
"step": 9858
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3563,
|
|
"step": 9860
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3366,
|
|
"step": 9862
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3181,
|
|
"step": 9864
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3307,
|
|
"step": 9866
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2772,
|
|
"step": 9868
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3157,
|
|
"step": 9870
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3347,
|
|
"step": 9872
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.322,
|
|
"step": 9874
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2937,
|
|
"step": 9876
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3119,
|
|
"step": 9878
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3746,
|
|
"step": 9880
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3109,
|
|
"step": 9882
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3017,
|
|
"step": 9884
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3403,
|
|
"step": 9886
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2999,
|
|
"step": 9888
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3057,
|
|
"step": 9890
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3064,
|
|
"step": 9892
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3229,
|
|
"step": 9894
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3313,
|
|
"step": 9896
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3423,
|
|
"step": 9898
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3196,
|
|
"step": 9900
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3377,
|
|
"step": 9902
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3386,
|
|
"step": 9904
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3404,
|
|
"step": 9906
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3218,
|
|
"step": 9908
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3311,
|
|
"step": 9910
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3062,
|
|
"step": 9912
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3271,
|
|
"step": 9914
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3197,
|
|
"step": 9916
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3242,
|
|
"step": 9918
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3353,
|
|
"step": 9920
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3,
|
|
"step": 9922
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3231,
|
|
"step": 9924
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2994,
|
|
"step": 9926
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3217,
|
|
"step": 9928
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3359,
|
|
"step": 9930
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3199,
|
|
"step": 9932
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3174,
|
|
"step": 9934
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3333,
|
|
"step": 9936
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3115,
|
|
"step": 9938
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3187,
|
|
"step": 9940
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3152,
|
|
"step": 9942
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3159,
|
|
"step": 9944
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3189,
|
|
"step": 9946
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3109,
|
|
"step": 9948
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3242,
|
|
"step": 9950
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2931,
|
|
"step": 9952
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3323,
|
|
"step": 9954
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3486,
|
|
"step": 9956
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3325,
|
|
"step": 9958
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3223,
|
|
"step": 9960
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3279,
|
|
"step": 9962
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3162,
|
|
"step": 9964
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3135,
|
|
"step": 9966
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3169,
|
|
"step": 9968
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3294,
|
|
"step": 9970
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.293,
|
|
"step": 9972
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3307,
|
|
"step": 9974
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3397,
|
|
"step": 9976
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3029,
|
|
"step": 9978
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2984,
|
|
"step": 9980
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3111,
|
|
"step": 9982
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3287,
|
|
"step": 9984
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3008,
|
|
"step": 9986
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3093,
|
|
"step": 9988
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3613,
|
|
"step": 9990
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3341,
|
|
"step": 9992
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3083,
|
|
"step": 9994
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3254,
|
|
"step": 9996
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3406,
|
|
"step": 9998
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3572,
|
|
"step": 10000
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3103,
|
|
"step": 10002
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3581,
|
|
"step": 10004
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3164,
|
|
"step": 10006
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.321,
|
|
"step": 10008
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3407,
|
|
"step": 10010
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3179,
|
|
"step": 10012
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3167,
|
|
"step": 10014
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3107,
|
|
"step": 10016
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3533,
|
|
"step": 10018
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3255,
|
|
"step": 10020
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3148,
|
|
"step": 10022
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3252,
|
|
"step": 10024
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3211,
|
|
"step": 10026
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3279,
|
|
"step": 10028
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3385,
|
|
"step": 10030
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3146,
|
|
"step": 10032
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3347,
|
|
"step": 10034
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3051,
|
|
"step": 10036
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2955,
|
|
"step": 10038
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3403,
|
|
"step": 10040
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2941,
|
|
"step": 10042
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3456,
|
|
"step": 10044
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3398,
|
|
"step": 10046
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.321,
|
|
"step": 10048
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3212,
|
|
"step": 10050
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2802,
|
|
"step": 10052
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.327,
|
|
"step": 10054
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.324,
|
|
"step": 10056
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3345,
|
|
"step": 10058
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3188,
|
|
"step": 10060
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3444,
|
|
"step": 10062
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2931,
|
|
"step": 10064
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.312,
|
|
"step": 10066
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3221,
|
|
"step": 10068
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3342,
|
|
"step": 10070
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3441,
|
|
"step": 10072
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3127,
|
|
"step": 10074
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2703,
|
|
"step": 10076
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3175,
|
|
"step": 10078
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.305,
|
|
"step": 10080
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3011,
|
|
"step": 10082
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3425,
|
|
"step": 10084
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3401,
|
|
"step": 10086
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2912,
|
|
"step": 10088
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3192,
|
|
"step": 10090
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3263,
|
|
"step": 10092
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3031,
|
|
"step": 10094
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3345,
|
|
"step": 10096
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3037,
|
|
"step": 10098
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.309,
|
|
"step": 10100
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3325,
|
|
"step": 10102
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3268,
|
|
"step": 10104
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3207,
|
|
"step": 10106
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3411,
|
|
"step": 10108
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2937,
|
|
"step": 10110
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.317,
|
|
"step": 10112
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3323,
|
|
"step": 10114
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3417,
|
|
"step": 10116
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3214,
|
|
"step": 10118
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3053,
|
|
"step": 10120
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3198,
|
|
"step": 10122
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.343,
|
|
"step": 10124
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3599,
|
|
"step": 10126
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3356,
|
|
"step": 10128
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3468,
|
|
"step": 10130
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3702,
|
|
"step": 10132
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.312,
|
|
"step": 10134
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3381,
|
|
"step": 10136
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3384,
|
|
"step": 10138
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3248,
|
|
"step": 10140
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3518,
|
|
"step": 10142
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3063,
|
|
"step": 10144
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2751,
|
|
"step": 10146
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.296,
|
|
"step": 10148
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3052,
|
|
"step": 10150
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3267,
|
|
"step": 10152
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3313,
|
|
"step": 10154
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3215,
|
|
"step": 10156
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.279,
|
|
"step": 10158
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3299,
|
|
"step": 10160
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2762,
|
|
"step": 10162
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3298,
|
|
"step": 10164
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3206,
|
|
"step": 10166
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3243,
|
|
"step": 10168
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3199,
|
|
"step": 10170
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3384,
|
|
"step": 10172
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3312,
|
|
"step": 10174
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.379,
|
|
"step": 10176
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3248,
|
|
"step": 10178
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3006,
|
|
"step": 10180
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3389,
|
|
"step": 10182
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3265,
|
|
"step": 10184
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3075,
|
|
"step": 10186
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3105,
|
|
"step": 10188
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3148,
|
|
"step": 10190
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2883,
|
|
"step": 10192
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.2766,
|
|
"step": 10194
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.356,
|
|
"step": 10196
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.3363,
|
|
"step": 10198
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 5e-05,
|
|
"loss": 0.35,
|
|
"step": 10200
|
|
}
|
|
],
|
|
"max_steps": 10234,
|
|
"num_train_epochs": 2,
|
|
"total_flos": 4989428611153920.0,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|