Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +665 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,665 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 19.91111111111111,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 896,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.2222222222222222,
|
| 13 |
+
"grad_norm": 3.4277842044830322,
|
| 14 |
+
"learning_rate": 4.4444444444444447e-05,
|
| 15 |
+
"loss": 1.3797,
|
| 16 |
+
"step": 10
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.4444444444444444,
|
| 20 |
+
"grad_norm": 2.029878854751587,
|
| 21 |
+
"learning_rate": 8.888888888888889e-05,
|
| 22 |
+
"loss": 0.424,
|
| 23 |
+
"step": 20
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.6666666666666666,
|
| 27 |
+
"grad_norm": 1.6295349597930908,
|
| 28 |
+
"learning_rate": 0.00013333333333333334,
|
| 29 |
+
"loss": 0.2302,
|
| 30 |
+
"step": 30
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.8888888888888888,
|
| 34 |
+
"grad_norm": 0.916685163974762,
|
| 35 |
+
"learning_rate": 0.00017777777777777779,
|
| 36 |
+
"loss": 0.1664,
|
| 37 |
+
"step": 40
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 1.1111111111111112,
|
| 41 |
+
"grad_norm": 0.7659821510314941,
|
| 42 |
+
"learning_rate": 0.000199982965150241,
|
| 43 |
+
"loss": 0.1398,
|
| 44 |
+
"step": 50
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 1.3333333333333333,
|
| 48 |
+
"grad_norm": 1.0613561868667603,
|
| 49 |
+
"learning_rate": 0.00019984672117252423,
|
| 50 |
+
"loss": 0.1207,
|
| 51 |
+
"step": 60
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 1.5555555555555556,
|
| 55 |
+
"grad_norm": 0.7991588711738586,
|
| 56 |
+
"learning_rate": 0.00019957441887293156,
|
| 57 |
+
"loss": 0.1119,
|
| 58 |
+
"step": 70
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 1.7777777777777777,
|
| 62 |
+
"grad_norm": 0.6806069016456604,
|
| 63 |
+
"learning_rate": 0.00019916642931015662,
|
| 64 |
+
"loss": 0.0885,
|
| 65 |
+
"step": 80
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 2.0,
|
| 69 |
+
"grad_norm": 0.6726401448249817,
|
| 70 |
+
"learning_rate": 0.00019862330844011466,
|
| 71 |
+
"loss": 0.0849,
|
| 72 |
+
"step": 90
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 2.2222222222222223,
|
| 76 |
+
"grad_norm": 0.44822949171066284,
|
| 77 |
+
"learning_rate": 0.00019794579635835704,
|
| 78 |
+
"loss": 0.0721,
|
| 79 |
+
"step": 100
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 2.4444444444444446,
|
| 83 |
+
"grad_norm": 0.40653809905052185,
|
| 84 |
+
"learning_rate": 0.0001971348162915637,
|
| 85 |
+
"loss": 0.0729,
|
| 86 |
+
"step": 110
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 2.6666666666666665,
|
| 90 |
+
"grad_norm": 0.40992143750190735,
|
| 91 |
+
"learning_rate": 0.00019619147333948823,
|
| 92 |
+
"loss": 0.0666,
|
| 93 |
+
"step": 120
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 2.888888888888889,
|
| 97 |
+
"grad_norm": 0.6051583290100098,
|
| 98 |
+
"learning_rate": 0.00019511705296906945,
|
| 99 |
+
"loss": 0.0606,
|
| 100 |
+
"step": 130
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 3.111111111111111,
|
| 104 |
+
"grad_norm": 0.46464696526527405,
|
| 105 |
+
"learning_rate": 0.00019391301926276156,
|
| 106 |
+
"loss": 0.0631,
|
| 107 |
+
"step": 140
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 3.3333333333333335,
|
| 111 |
+
"grad_norm": 0.5700137615203857,
|
| 112 |
+
"learning_rate": 0.00019258101292347042,
|
| 113 |
+
"loss": 0.0594,
|
| 114 |
+
"step": 150
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 3.5555555555555554,
|
| 118 |
+
"grad_norm": 0.5737802386283875,
|
| 119 |
+
"learning_rate": 0.0001911228490388136,
|
| 120 |
+
"loss": 0.0609,
|
| 121 |
+
"step": 160
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 3.7777777777777777,
|
| 125 |
+
"grad_norm": 0.4904007315635681,
|
| 126 |
+
"learning_rate": 0.0001895405146077514,
|
| 127 |
+
"loss": 0.0539,
|
| 128 |
+
"step": 170
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 4.0,
|
| 132 |
+
"grad_norm": 0.4444526731967926,
|
| 133 |
+
"learning_rate": 0.00018783616583295943,
|
| 134 |
+
"loss": 0.0514,
|
| 135 |
+
"step": 180
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 4.222222222222222,
|
| 139 |
+
"grad_norm": 0.3706776201725006,
|
| 140 |
+
"learning_rate": 0.00018601212518263156,
|
| 141 |
+
"loss": 0.0495,
|
| 142 |
+
"step": 190
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 4.444444444444445,
|
| 146 |
+
"grad_norm": 0.29509109258651733,
|
| 147 |
+
"learning_rate": 0.00018407087822571794,
|
| 148 |
+
"loss": 0.0477,
|
| 149 |
+
"step": 200
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 4.666666666666667,
|
| 153 |
+
"grad_norm": 0.4063428044319153,
|
| 154 |
+
"learning_rate": 0.00018201507024490988,
|
| 155 |
+
"loss": 0.0485,
|
| 156 |
+
"step": 210
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 4.888888888888889,
|
| 160 |
+
"grad_norm": 0.4438524544239044,
|
| 161 |
+
"learning_rate": 0.0001798475026319875,
|
| 162 |
+
"loss": 0.046,
|
| 163 |
+
"step": 220
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 5.111111111111111,
|
| 167 |
+
"grad_norm": 0.26472207903862,
|
| 168 |
+
"learning_rate": 0.000177571129070442,
|
| 169 |
+
"loss": 0.0452,
|
| 170 |
+
"step": 230
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 5.333333333333333,
|
| 174 |
+
"grad_norm": 0.3637102246284485,
|
| 175 |
+
"learning_rate": 0.0001751890515105738,
|
| 176 |
+
"loss": 0.0471,
|
| 177 |
+
"step": 240
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 5.555555555555555,
|
| 181 |
+
"grad_norm": 0.2891571819782257,
|
| 182 |
+
"learning_rate": 0.00017270451594255233,
|
| 183 |
+
"loss": 0.0436,
|
| 184 |
+
"step": 250
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 5.777777777777778,
|
| 188 |
+
"grad_norm": 0.380941778421402,
|
| 189 |
+
"learning_rate": 0.00017012090797319628,
|
| 190 |
+
"loss": 0.0439,
|
| 191 |
+
"step": 260
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 6.0,
|
| 195 |
+
"grad_norm": 0.573648989200592,
|
| 196 |
+
"learning_rate": 0.00016744174821250237,
|
| 197 |
+
"loss": 0.0434,
|
| 198 |
+
"step": 270
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 6.222222222222222,
|
| 202 |
+
"grad_norm": 0.34533384442329407,
|
| 203 |
+
"learning_rate": 0.0001646706874762089,
|
| 204 |
+
"loss": 0.0369,
|
| 205 |
+
"step": 280
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 6.444444444444445,
|
| 209 |
+
"grad_norm": 0.5272351503372192,
|
| 210 |
+
"learning_rate": 0.0001618115018109318,
|
| 211 |
+
"loss": 0.0403,
|
| 212 |
+
"step": 290
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 6.666666666666667,
|
| 216 |
+
"grad_norm": 0.2849768400192261,
|
| 217 |
+
"learning_rate": 0.00015886808734865202,
|
| 218 |
+
"loss": 0.0388,
|
| 219 |
+
"step": 300
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 6.888888888888889,
|
| 223 |
+
"grad_norm": 0.301411896944046,
|
| 224 |
+
"learning_rate": 0.00015584445499756578,
|
| 225 |
+
"loss": 0.0405,
|
| 226 |
+
"step": 310
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 7.111111111111111,
|
| 230 |
+
"grad_norm": 0.32372748851776123,
|
| 231 |
+
"learning_rate": 0.0001527447249765329,
|
| 232 |
+
"loss": 0.0362,
|
| 233 |
+
"step": 320
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 7.333333333333333,
|
| 237 |
+
"grad_norm": 0.26995155215263367,
|
| 238 |
+
"learning_rate": 0.00014957312120057005,
|
| 239 |
+
"loss": 0.0335,
|
| 240 |
+
"step": 330
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 7.555555555555555,
|
| 244 |
+
"grad_norm": 0.35481250286102295,
|
| 245 |
+
"learning_rate": 0.00014633396552504063,
|
| 246 |
+
"loss": 0.0424,
|
| 247 |
+
"step": 340
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 7.777777777777778,
|
| 251 |
+
"grad_norm": 0.3813558518886566,
|
| 252 |
+
"learning_rate": 0.00014303167185638366,
|
| 253 |
+
"loss": 0.0378,
|
| 254 |
+
"step": 350
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 8.0,
|
| 258 |
+
"grad_norm": 0.3790789842605591,
|
| 259 |
+
"learning_rate": 0.0001396707401374078,
|
| 260 |
+
"loss": 0.0403,
|
| 261 |
+
"step": 360
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 8.222222222222221,
|
| 265 |
+
"grad_norm": 0.28233516216278076,
|
| 266 |
+
"learning_rate": 0.00013625575021534536,
|
| 267 |
+
"loss": 0.0304,
|
| 268 |
+
"step": 370
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 8.444444444444445,
|
| 272 |
+
"grad_norm": 0.4000020921230316,
|
| 273 |
+
"learning_rate": 0.00013279135560102337,
|
| 274 |
+
"loss": 0.033,
|
| 275 |
+
"step": 380
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 8.666666666666666,
|
| 279 |
+
"grad_norm": 0.29253950715065,
|
| 280 |
+
"learning_rate": 0.00012928227712765504,
|
| 281 |
+
"loss": 0.033,
|
| 282 |
+
"step": 390
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 8.88888888888889,
|
| 286 |
+
"grad_norm": 0.33955538272857666,
|
| 287 |
+
"learning_rate": 0.00012573329651789297,
|
| 288 |
+
"loss": 0.0338,
|
| 289 |
+
"step": 400
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 9.11111111111111,
|
| 293 |
+
"grad_norm": 0.3720948398113251,
|
| 294 |
+
"learning_rate": 0.00012214924986791003,
|
| 295 |
+
"loss": 0.0353,
|
| 296 |
+
"step": 410
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 9.333333333333334,
|
| 300 |
+
"grad_norm": 0.2159433513879776,
|
| 301 |
+
"learning_rate": 0.00011853502105738692,
|
| 302 |
+
"loss": 0.0328,
|
| 303 |
+
"step": 420
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 9.555555555555555,
|
| 307 |
+
"grad_norm": 0.20489104092121124,
|
| 308 |
+
"learning_rate": 0.00011489553509438657,
|
| 309 |
+
"loss": 0.0351,
|
| 310 |
+
"step": 430
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 9.777777777777779,
|
| 314 |
+
"grad_norm": 0.19427303969860077,
|
| 315 |
+
"learning_rate": 0.00011123575140418414,
|
| 316 |
+
"loss": 0.0302,
|
| 317 |
+
"step": 440
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 10.0,
|
| 321 |
+
"grad_norm": 0.2984829545021057,
|
| 322 |
+
"learning_rate": 0.00010756065707119729,
|
| 323 |
+
"loss": 0.0285,
|
| 324 |
+
"step": 450
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 10.222222222222221,
|
| 328 |
+
"grad_norm": 0.19880931079387665,
|
| 329 |
+
"learning_rate": 0.0001038752600432265,
|
| 330 |
+
"loss": 0.0283,
|
| 331 |
+
"step": 460
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 10.444444444444445,
|
| 335 |
+
"grad_norm": 0.30339422821998596,
|
| 336 |
+
"learning_rate": 0.00010018458230726523,
|
| 337 |
+
"loss": 0.0268,
|
| 338 |
+
"step": 470
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 10.666666666666666,
|
| 342 |
+
"grad_norm": 0.20127274096012115,
|
| 343 |
+
"learning_rate": 9.649365304617952e-05,
|
| 344 |
+
"loss": 0.031,
|
| 345 |
+
"step": 480
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 10.88888888888889,
|
| 349 |
+
"grad_norm": 0.2175353467464447,
|
| 350 |
+
"learning_rate": 9.280750178558138e-05,
|
| 351 |
+
"loss": 0.0259,
|
| 352 |
+
"step": 490
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 11.11111111111111,
|
| 356 |
+
"grad_norm": 0.26974818110466003,
|
| 357 |
+
"learning_rate": 8.913115154023605e-05,
|
| 358 |
+
"loss": 0.0306,
|
| 359 |
+
"step": 500
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 11.333333333333334,
|
| 363 |
+
"grad_norm": 0.4030226767063141,
|
| 364 |
+
"learning_rate": 8.546961196934043e-05,
|
| 365 |
+
"loss": 0.0248,
|
| 366 |
+
"step": 510
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 11.555555555555555,
|
| 370 |
+
"grad_norm": 0.23470337688922882,
|
| 371 |
+
"learning_rate": 8.182787255000155e-05,
|
| 372 |
+
"loss": 0.0262,
|
| 373 |
+
"step": 520
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 11.777777777777779,
|
| 377 |
+
"grad_norm": 0.40099358558654785,
|
| 378 |
+
"learning_rate": 7.82108957782161e-05,
|
| 379 |
+
"loss": 0.0259,
|
| 380 |
+
"step": 530
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 12.0,
|
| 384 |
+
"grad_norm": 0.17941424250602722,
|
| 385 |
+
"learning_rate": 7.462361040661667e-05,
|
| 386 |
+
"loss": 0.0299,
|
| 387 |
+
"step": 540
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 12.222222222222221,
|
| 391 |
+
"grad_norm": 0.24401065707206726,
|
| 392 |
+
"learning_rate": 7.107090472819896e-05,
|
| 393 |
+
"loss": 0.0231,
|
| 394 |
+
"step": 550
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 12.444444444444445,
|
| 398 |
+
"grad_norm": 0.26819756627082825,
|
| 399 |
+
"learning_rate": 6.755761991518219e-05,
|
| 400 |
+
"loss": 0.0262,
|
| 401 |
+
"step": 560
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 12.666666666666666,
|
| 405 |
+
"grad_norm": 0.300659716129303,
|
| 406 |
+
"learning_rate": 6.408854342207982e-05,
|
| 407 |
+
"loss": 0.0289,
|
| 408 |
+
"step": 570
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 12.88888888888889,
|
| 412 |
+
"grad_norm": 0.2212854027748108,
|
| 413 |
+
"learning_rate": 6.0668402461969807e-05,
|
| 414 |
+
"loss": 0.0243,
|
| 415 |
+
"step": 580
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 13.11111111111111,
|
| 419 |
+
"grad_norm": 0.28721633553504944,
|
| 420 |
+
"learning_rate": 5.730185756485395e-05,
|
| 421 |
+
"loss": 0.0256,
|
| 422 |
+
"step": 590
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 13.333333333333334,
|
| 426 |
+
"grad_norm": 0.2350863218307495,
|
| 427 |
+
"learning_rate": 5.399349622688479e-05,
|
| 428 |
+
"loss": 0.0261,
|
| 429 |
+
"step": 600
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 13.555555555555555,
|
| 433 |
+
"grad_norm": 0.31170952320098877,
|
| 434 |
+
"learning_rate": 5.074782665911341e-05,
|
| 435 |
+
"loss": 0.0247,
|
| 436 |
+
"step": 610
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 13.777777777777779,
|
| 440 |
+
"grad_norm": 0.18656305968761444,
|
| 441 |
+
"learning_rate": 4.756927164427685e-05,
|
| 442 |
+
"loss": 0.0233,
|
| 443 |
+
"step": 620
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 14.0,
|
| 447 |
+
"grad_norm": 0.2578571140766144,
|
| 448 |
+
"learning_rate": 4.446216250999641e-05,
|
| 449 |
+
"loss": 0.0257,
|
| 450 |
+
"step": 630
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 14.222222222222221,
|
| 454 |
+
"grad_norm": 0.16936838626861572,
|
| 455 |
+
"learning_rate": 4.1430733226599114e-05,
|
| 456 |
+
"loss": 0.0212,
|
| 457 |
+
"step": 640
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 14.444444444444445,
|
| 461 |
+
"grad_norm": 0.15445557236671448,
|
| 462 |
+
"learning_rate": 3.8479114637605285e-05,
|
| 463 |
+
"loss": 0.0246,
|
| 464 |
+
"step": 650
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 14.666666666666666,
|
| 468 |
+
"grad_norm": 0.1869278848171234,
|
| 469 |
+
"learning_rate": 3.561132883074427e-05,
|
| 470 |
+
"loss": 0.0232,
|
| 471 |
+
"step": 660
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 14.88888888888889,
|
| 475 |
+
"grad_norm": 0.20103834569454193,
|
| 476 |
+
"learning_rate": 3.2831283657168275e-05,
|
| 477 |
+
"loss": 0.0226,
|
| 478 |
+
"step": 670
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 15.11111111111111,
|
| 482 |
+
"grad_norm": 0.2504471242427826,
|
| 483 |
+
"learning_rate": 3.0142767406333518e-05,
|
| 484 |
+
"loss": 0.0184,
|
| 485 |
+
"step": 680
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 15.333333333333334,
|
| 489 |
+
"grad_norm": 0.18746283650398254,
|
| 490 |
+
"learning_rate": 2.7549443643804585e-05,
|
| 491 |
+
"loss": 0.019,
|
| 492 |
+
"step": 690
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 15.555555555555555,
|
| 496 |
+
"grad_norm": 0.25808605551719666,
|
| 497 |
+
"learning_rate": 2.505484621901655e-05,
|
| 498 |
+
"loss": 0.0181,
|
| 499 |
+
"step": 700
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 15.777777777777779,
|
| 503 |
+
"grad_norm": 0.2831310033798218,
|
| 504 |
+
"learning_rate": 2.2662374449797664e-05,
|
| 505 |
+
"loss": 0.022,
|
| 506 |
+
"step": 710
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 16.0,
|
| 510 |
+
"grad_norm": 0.18062320351600647,
|
| 511 |
+
"learning_rate": 2.0375288490214404e-05,
|
| 512 |
+
"loss": 0.0205,
|
| 513 |
+
"step": 720
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 16.22222222222222,
|
| 517 |
+
"grad_norm": 0.15133315324783325,
|
| 518 |
+
"learning_rate": 1.819670488805111e-05,
|
| 519 |
+
"loss": 0.0198,
|
| 520 |
+
"step": 730
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 16.444444444444443,
|
| 524 |
+
"grad_norm": 0.20342513918876648,
|
| 525 |
+
"learning_rate": 1.6129592337977995e-05,
|
| 526 |
+
"loss": 0.02,
|
| 527 |
+
"step": 740
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 16.666666666666668,
|
| 531 |
+
"grad_norm": 0.21008381247520447,
|
| 532 |
+
"learning_rate": 1.4176767636194122e-05,
|
| 533 |
+
"loss": 0.0232,
|
| 534 |
+
"step": 750
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 16.88888888888889,
|
| 538 |
+
"grad_norm": 0.25737613439559937,
|
| 539 |
+
"learning_rate": 1.234089184205851e-05,
|
| 540 |
+
"loss": 0.0227,
|
| 541 |
+
"step": 760
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 17.11111111111111,
|
| 545 |
+
"grad_norm": 0.15780089795589447,
|
| 546 |
+
"learning_rate": 1.0624466651939247e-05,
|
| 547 |
+
"loss": 0.0203,
|
| 548 |
+
"step": 770
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 17.333333333333332,
|
| 552 |
+
"grad_norm": 0.16062302887439728,
|
| 553 |
+
"learning_rate": 9.029830990222132e-06,
|
| 554 |
+
"loss": 0.0244,
|
| 555 |
+
"step": 780
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 17.555555555555557,
|
| 559 |
+
"grad_norm": 0.18100588023662567,
|
| 560 |
+
"learning_rate": 7.55915782212413e-06,
|
| 561 |
+
"loss": 0.0173,
|
| 562 |
+
"step": 790
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 17.77777777777778,
|
| 566 |
+
"grad_norm": 0.20718730986118317,
|
| 567 |
+
"learning_rate": 6.214451192654747e-06,
|
| 568 |
+
"loss": 0.0186,
|
| 569 |
+
"step": 800
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 18.0,
|
| 573 |
+
"grad_norm": 0.18695631623268127,
|
| 574 |
+
"learning_rate": 4.9975434957601264e-06,
|
| 575 |
+
"loss": 0.0177,
|
| 576 |
+
"step": 810
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 18.22222222222222,
|
| 580 |
+
"grad_norm": 0.1640794426202774,
|
| 581 |
+
"learning_rate": 3.910092977371394e-06,
|
| 582 |
+
"loss": 0.0169,
|
| 583 |
+
"step": 820
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 18.444444444444443,
|
| 587 |
+
"grad_norm": 0.18873189389705658,
|
| 588 |
+
"learning_rate": 2.953581475759404e-06,
|
| 589 |
+
"loss": 0.0214,
|
| 590 |
+
"step": 830
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 18.666666666666668,
|
| 594 |
+
"grad_norm": 0.1896572709083557,
|
| 595 |
+
"learning_rate": 2.1293124022754407e-06,
|
| 596 |
+
"loss": 0.0173,
|
| 597 |
+
"step": 840
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 18.88888888888889,
|
| 601 |
+
"grad_norm": 0.09888505190610886,
|
| 602 |
+
"learning_rate": 1.4384089652291543e-06,
|
| 603 |
+
"loss": 0.0185,
|
| 604 |
+
"step": 850
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 19.11111111111111,
|
| 608 |
+
"grad_norm": 0.17348803579807281,
|
| 609 |
+
"learning_rate": 8.818126393241643e-07,
|
| 610 |
+
"loss": 0.0182,
|
| 611 |
+
"step": 860
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 19.333333333333332,
|
| 615 |
+
"grad_norm": 0.20844842493534088,
|
| 616 |
+
"learning_rate": 4.602818827369126e-07,
|
| 617 |
+
"loss": 0.0217,
|
| 618 |
+
"step": 870
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 19.555555555555557,
|
| 622 |
+
"grad_norm": 0.16763387620449066,
|
| 623 |
+
"learning_rate": 1.7439110358704602e-07,
|
| 624 |
+
"loss": 0.0175,
|
| 625 |
+
"step": 880
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 19.77777777777778,
|
| 629 |
+
"grad_norm": 0.13927114009857178,
|
| 630 |
+
"learning_rate": 2.4529877207557505e-08,
|
| 631 |
+
"loss": 0.0221,
|
| 632 |
+
"step": 890
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 19.91111111111111,
|
| 636 |
+
"step": 896,
|
| 637 |
+
"total_flos": 6.431855929189344e+16,
|
| 638 |
+
"train_loss": 0.060006462794262916,
|
| 639 |
+
"train_runtime": 634.4726,
|
| 640 |
+
"train_samples_per_second": 90.381,
|
| 641 |
+
"train_steps_per_second": 1.412
|
| 642 |
+
}
|
| 643 |
+
],
|
| 644 |
+
"logging_steps": 10,
|
| 645 |
+
"max_steps": 896,
|
| 646 |
+
"num_input_tokens_seen": 0,
|
| 647 |
+
"num_train_epochs": 20,
|
| 648 |
+
"save_steps": 10000,
|
| 649 |
+
"stateful_callbacks": {
|
| 650 |
+
"TrainerControl": {
|
| 651 |
+
"args": {
|
| 652 |
+
"should_epoch_stop": false,
|
| 653 |
+
"should_evaluate": false,
|
| 654 |
+
"should_log": false,
|
| 655 |
+
"should_save": true,
|
| 656 |
+
"should_training_stop": true
|
| 657 |
+
},
|
| 658 |
+
"attributes": {}
|
| 659 |
+
}
|
| 660 |
+
},
|
| 661 |
+
"total_flos": 6.431855929189344e+16,
|
| 662 |
+
"train_batch_size": 64,
|
| 663 |
+
"trial_name": null,
|
| 664 |
+
"trial_params": null
|
| 665 |
+
}
|