Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +399 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,399 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 19.807692307692307,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 515,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.38461538461538464,
|
| 13 |
+
"grad_norm": 3.364804983139038,
|
| 14 |
+
"learning_rate": 7.692307692307693e-05,
|
| 15 |
+
"loss": 1.4034,
|
| 16 |
+
"step": 10
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.7692307692307693,
|
| 20 |
+
"grad_norm": 1.2244789600372314,
|
| 21 |
+
"learning_rate": 0.00015384615384615385,
|
| 22 |
+
"loss": 0.4222,
|
| 23 |
+
"step": 20
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 1.1538461538461537,
|
| 27 |
+
"grad_norm": 1.535311222076416,
|
| 28 |
+
"learning_rate": 0.00019996698220143899,
|
| 29 |
+
"loss": 0.2313,
|
| 30 |
+
"step": 30
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 1.5384615384615383,
|
| 34 |
+
"grad_norm": 0.9641413688659668,
|
| 35 |
+
"learning_rate": 0.0001995957823217325,
|
| 36 |
+
"loss": 0.1714,
|
| 37 |
+
"step": 40
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 1.9230769230769231,
|
| 41 |
+
"grad_norm": 1.1909210681915283,
|
| 42 |
+
"learning_rate": 0.00019881364700727823,
|
| 43 |
+
"loss": 0.148,
|
| 44 |
+
"step": 50
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 2.3076923076923075,
|
| 48 |
+
"grad_norm": 0.9179659485816956,
|
| 49 |
+
"learning_rate": 0.0001976238033738033,
|
| 50 |
+
"loss": 0.1358,
|
| 51 |
+
"step": 60
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 2.6923076923076925,
|
| 55 |
+
"grad_norm": 1.524248719215393,
|
| 56 |
+
"learning_rate": 0.00019603116075480955,
|
| 57 |
+
"loss": 0.1366,
|
| 58 |
+
"step": 70
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 3.076923076923077,
|
| 62 |
+
"grad_norm": 0.7116435766220093,
|
| 63 |
+
"learning_rate": 0.00019404229044550433,
|
| 64 |
+
"loss": 0.1132,
|
| 65 |
+
"step": 80
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 3.4615384615384617,
|
| 69 |
+
"grad_norm": 0.4233417510986328,
|
| 70 |
+
"learning_rate": 0.00019166539858942255,
|
| 71 |
+
"loss": 0.1099,
|
| 72 |
+
"step": 90
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 3.8461538461538463,
|
| 76 |
+
"grad_norm": 0.3895157277584076,
|
| 77 |
+
"learning_rate": 0.00018891029231961207,
|
| 78 |
+
"loss": 0.0973,
|
| 79 |
+
"step": 100
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 4.230769230769231,
|
| 83 |
+
"grad_norm": 0.6012492179870605,
|
| 84 |
+
"learning_rate": 0.0001857883392940837,
|
| 85 |
+
"loss": 0.0974,
|
| 86 |
+
"step": 110
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 4.615384615384615,
|
| 90 |
+
"grad_norm": 0.4354020059108734,
|
| 91 |
+
"learning_rate": 0.00018231242079248512,
|
| 92 |
+
"loss": 0.0853,
|
| 93 |
+
"step": 120
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 5.0,
|
| 97 |
+
"grad_norm": 0.6374463438987732,
|
| 98 |
+
"learning_rate": 0.00017849687856752208,
|
| 99 |
+
"loss": 0.0814,
|
| 100 |
+
"step": 130
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 5.384615384615385,
|
| 104 |
+
"grad_norm": 0.4828743636608124,
|
| 105 |
+
"learning_rate": 0.00017435745567042095,
|
| 106 |
+
"loss": 0.083,
|
| 107 |
+
"step": 140
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 5.769230769230769,
|
| 111 |
+
"grad_norm": 0.2791459560394287,
|
| 112 |
+
"learning_rate": 0.00016991123149458738,
|
| 113 |
+
"loss": 0.0769,
|
| 114 |
+
"step": 150
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 6.153846153846154,
|
| 118 |
+
"grad_norm": 0.3161262273788452,
|
| 119 |
+
"learning_rate": 0.00016517655130547437,
|
| 120 |
+
"loss": 0.0819,
|
| 121 |
+
"step": 160
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 6.538461538461538,
|
| 125 |
+
"grad_norm": 0.5798579454421997,
|
| 126 |
+
"learning_rate": 0.00016017295054742046,
|
| 127 |
+
"loss": 0.0688,
|
| 128 |
+
"step": 170
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 6.923076923076923,
|
| 132 |
+
"grad_norm": 0.38132503628730774,
|
| 133 |
+
"learning_rate": 0.00015492107423977167,
|
| 134 |
+
"loss": 0.0618,
|
| 135 |
+
"step": 180
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 7.3076923076923075,
|
| 139 |
+
"grad_norm": 0.4480208158493042,
|
| 140 |
+
"learning_rate": 0.00014944259179486066,
|
| 141 |
+
"loss": 0.0678,
|
| 142 |
+
"step": 190
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 7.6923076923076925,
|
| 146 |
+
"grad_norm": 0.3681942820549011,
|
| 147 |
+
"learning_rate": 0.00014376010760930728,
|
| 148 |
+
"loss": 0.0629,
|
| 149 |
+
"step": 200
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 8.076923076923077,
|
| 153 |
+
"grad_norm": 0.5002759695053101,
|
| 154 |
+
"learning_rate": 0.00013789706779754327,
|
| 155 |
+
"loss": 0.0655,
|
| 156 |
+
"step": 210
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 8.461538461538462,
|
| 160 |
+
"grad_norm": 0.4140470325946808,
|
| 161 |
+
"learning_rate": 0.00013187766345238222,
|
| 162 |
+
"loss": 0.0623,
|
| 163 |
+
"step": 220
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 8.846153846153847,
|
| 167 |
+
"grad_norm": 0.37125447392463684,
|
| 168 |
+
"learning_rate": 0.0001257267308317845,
|
| 169 |
+
"loss": 0.0584,
|
| 170 |
+
"step": 230
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 9.23076923076923,
|
| 174 |
+
"grad_norm": 0.3931460976600647,
|
| 175 |
+
"learning_rate": 0.0001194696488836495,
|
| 176 |
+
"loss": 0.0615,
|
| 177 |
+
"step": 240
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 9.615384615384615,
|
| 181 |
+
"grad_norm": 0.4081181287765503,
|
| 182 |
+
"learning_rate": 0.00011313223453145201,
|
| 183 |
+
"loss": 0.0541,
|
| 184 |
+
"step": 250
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 10.0,
|
| 188 |
+
"grad_norm": 0.42426028847694397,
|
| 189 |
+
"learning_rate": 0.0001067406361527768,
|
| 190 |
+
"loss": 0.0591,
|
| 191 |
+
"step": 260
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 10.384615384615385,
|
| 195 |
+
"grad_norm": 0.389825701713562,
|
| 196 |
+
"learning_rate": 0.00010032122569026283,
|
| 197 |
+
"loss": 0.0565,
|
| 198 |
+
"step": 270
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 10.76923076923077,
|
| 202 |
+
"grad_norm": 0.4282422363758087,
|
| 203 |
+
"learning_rate": 9.390048984011096e-05,
|
| 204 |
+
"loss": 0.0492,
|
| 205 |
+
"step": 280
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 11.153846153846153,
|
| 209 |
+
"grad_norm": 0.33358335494995117,
|
| 210 |
+
"learning_rate": 8.750492076711439e-05,
|
| 211 |
+
"loss": 0.0462,
|
| 212 |
+
"step": 290
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 11.538461538461538,
|
| 216 |
+
"grad_norm": 0.3536369204521179,
|
| 217 |
+
"learning_rate": 8.1160906797126e-05,
|
| 218 |
+
"loss": 0.0547,
|
| 219 |
+
"step": 300
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 11.923076923076923,
|
| 223 |
+
"grad_norm": 0.3393729329109192,
|
| 224 |
+
"learning_rate": 7.489462353796793e-05,
|
| 225 |
+
"loss": 0.0539,
|
| 226 |
+
"step": 310
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 12.307692307692308,
|
| 230 |
+
"grad_norm": 0.26623183488845825,
|
| 231 |
+
"learning_rate": 6.87319258780234e-05,
|
| 232 |
+
"loss": 0.04,
|
| 233 |
+
"step": 320
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 12.692307692307692,
|
| 237 |
+
"grad_norm": 0.34395718574523926,
|
| 238 |
+
"learning_rate": 6.269824130812645e-05,
|
| 239 |
+
"loss": 0.0493,
|
| 240 |
+
"step": 330
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 13.076923076923077,
|
| 244 |
+
"grad_norm": 0.2601974904537201,
|
| 245 |
+
"learning_rate": 5.681846500690884e-05,
|
| 246 |
+
"loss": 0.054,
|
| 247 |
+
"step": 340
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 13.461538461538462,
|
| 251 |
+
"grad_norm": 0.20003391802310944,
|
| 252 |
+
"learning_rate": 5.111685712248364e-05,
|
| 253 |
+
"loss": 0.0489,
|
| 254 |
+
"step": 350
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 13.846153846153847,
|
| 258 |
+
"grad_norm": 0.20917710661888123,
|
| 259 |
+
"learning_rate": 4.56169426742856e-05,
|
| 260 |
+
"loss": 0.0536,
|
| 261 |
+
"step": 360
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 14.23076923076923,
|
| 265 |
+
"grad_norm": 0.2096697986125946,
|
| 266 |
+
"learning_rate": 4.03414144880767e-05,
|
| 267 |
+
"loss": 0.0436,
|
| 268 |
+
"step": 370
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 14.615384615384615,
|
| 272 |
+
"grad_norm": 0.2338918149471283,
|
| 273 |
+
"learning_rate": 3.53120395646092e-05,
|
| 274 |
+
"loss": 0.0413,
|
| 275 |
+
"step": 380
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 15.0,
|
| 279 |
+
"grad_norm": 0.2506970465183258,
|
| 280 |
+
"learning_rate": 3.054956926827332e-05,
|
| 281 |
+
"loss": 0.0418,
|
| 282 |
+
"step": 390
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 15.384615384615385,
|
| 286 |
+
"grad_norm": 0.16216574609279633,
|
| 287 |
+
"learning_rate": 2.60736537062932e-05,
|
| 288 |
+
"loss": 0.0346,
|
| 289 |
+
"step": 400
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 15.76923076923077,
|
| 293 |
+
"grad_norm": 0.26421189308166504,
|
| 294 |
+
"learning_rate": 2.1902760651745958e-05,
|
| 295 |
+
"loss": 0.0457,
|
| 296 |
+
"step": 410
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 16.153846153846153,
|
| 300 |
+
"grad_norm": 0.2853449881076813,
|
| 301 |
+
"learning_rate": 1.8054099344929832e-05,
|
| 302 |
+
"loss": 0.0432,
|
| 303 |
+
"step": 420
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 16.53846153846154,
|
| 307 |
+
"grad_norm": 0.2742384672164917,
|
| 308 |
+
"learning_rate": 1.454354948747909e-05,
|
| 309 |
+
"loss": 0.0411,
|
| 310 |
+
"step": 430
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 16.923076923076923,
|
| 314 |
+
"grad_norm": 0.2132396250963211,
|
| 315 |
+
"learning_rate": 1.1385595722199438e-05,
|
| 316 |
+
"loss": 0.0372,
|
| 317 |
+
"step": 440
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 17.307692307692307,
|
| 321 |
+
"grad_norm": 0.23929736018180847,
|
| 322 |
+
"learning_rate": 8.593267868960675e-06,
|
| 323 |
+
"loss": 0.0336,
|
| 324 |
+
"step": 450
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 17.692307692307693,
|
| 328 |
+
"grad_norm": 0.16594846546649933,
|
| 329 |
+
"learning_rate": 6.178087163236645e-06,
|
| 330 |
+
"loss": 0.0383,
|
| 331 |
+
"step": 460
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 18.076923076923077,
|
| 335 |
+
"grad_norm": 0.18442106246948242,
|
| 336 |
+
"learning_rate": 4.1500187191131466e-06,
|
| 337 |
+
"loss": 0.0361,
|
| 338 |
+
"step": 470
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 18.46153846153846,
|
| 342 |
+
"grad_norm": 0.20087499916553497,
|
| 343 |
+
"learning_rate": 2.5174304129033653e-06,
|
| 344 |
+
"loss": 0.036,
|
| 345 |
+
"step": 480
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 18.846153846153847,
|
| 349 |
+
"grad_norm": 0.24509386718273163,
|
| 350 |
+
"learning_rate": 1.287058357018278e-06,
|
| 351 |
+
"loss": 0.0335,
|
| 352 |
+
"step": 490
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 19.23076923076923,
|
| 356 |
+
"grad_norm": 0.22717131674289703,
|
| 357 |
+
"learning_rate": 4.639791065478738e-07,
|
| 358 |
+
"loss": 0.0381,
|
| 359 |
+
"step": 500
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 19.615384615384617,
|
| 363 |
+
"grad_norm": 0.14815658330917358,
|
| 364 |
+
"learning_rate": 5.158871322984427e-08,
|
| 365 |
+
"loss": 0.0374,
|
| 366 |
+
"step": 510
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 19.807692307692307,
|
| 370 |
+
"step": 515,
|
| 371 |
+
"total_flos": 7.15348914538104e+16,
|
| 372 |
+
"train_loss": 0.10100355527354675,
|
| 373 |
+
"train_runtime": 643.3659,
|
| 374 |
+
"train_samples_per_second": 51.231,
|
| 375 |
+
"train_steps_per_second": 0.8
|
| 376 |
+
}
|
| 377 |
+
],
|
| 378 |
+
"logging_steps": 10,
|
| 379 |
+
"max_steps": 515,
|
| 380 |
+
"num_input_tokens_seen": 0,
|
| 381 |
+
"num_train_epochs": 20,
|
| 382 |
+
"save_steps": 10000,
|
| 383 |
+
"stateful_callbacks": {
|
| 384 |
+
"TrainerControl": {
|
| 385 |
+
"args": {
|
| 386 |
+
"should_epoch_stop": false,
|
| 387 |
+
"should_evaluate": false,
|
| 388 |
+
"should_log": false,
|
| 389 |
+
"should_save": true,
|
| 390 |
+
"should_training_stop": true
|
| 391 |
+
},
|
| 392 |
+
"attributes": {}
|
| 393 |
+
}
|
| 394 |
+
},
|
| 395 |
+
"total_flos": 7.15348914538104e+16,
|
| 396 |
+
"train_batch_size": 64,
|
| 397 |
+
"trial_name": null,
|
| 398 |
+
"trial_params": null
|
| 399 |
+
}
|