Upload trainer_state.json
Browse files- trainer_state.json +412 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,412 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.0011106810998171568,
|
| 3 |
+
"best_model_checkpoint": "Result/OffsetTwo_batch16_step2000\\checkpoint-600",
|
| 4 |
+
"epoch": 28.571428571428573,
|
| 5 |
+
"global_step": 600,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.48,
|
| 12 |
+
"learning_rate": 0.000199,
|
| 13 |
+
"loss": 0.6712,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.95,
|
| 18 |
+
"learning_rate": 0.00019800000000000002,
|
| 19 |
+
"loss": 0.7073,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 1.43,
|
| 24 |
+
"learning_rate": 0.00019700000000000002,
|
| 25 |
+
"loss": 0.3994,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 1.9,
|
| 30 |
+
"learning_rate": 0.000196,
|
| 31 |
+
"loss": 0.5316,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 2.38,
|
| 36 |
+
"learning_rate": 0.000195,
|
| 37 |
+
"loss": 0.6121,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 2.86,
|
| 42 |
+
"learning_rate": 0.000194,
|
| 43 |
+
"loss": 0.4975,
|
| 44 |
+
"step": 60
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 3.33,
|
| 48 |
+
"learning_rate": 0.000193,
|
| 49 |
+
"loss": 0.5137,
|
| 50 |
+
"step": 70
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 3.81,
|
| 54 |
+
"learning_rate": 0.000192,
|
| 55 |
+
"loss": 0.2739,
|
| 56 |
+
"step": 80
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 4.29,
|
| 60 |
+
"learning_rate": 0.000191,
|
| 61 |
+
"loss": 0.6669,
|
| 62 |
+
"step": 90
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 4.76,
|
| 66 |
+
"learning_rate": 0.00019,
|
| 67 |
+
"loss": 0.3004,
|
| 68 |
+
"step": 100
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 5.24,
|
| 72 |
+
"learning_rate": 0.00018899999999999999,
|
| 73 |
+
"loss": 0.1061,
|
| 74 |
+
"step": 110
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 5.71,
|
| 78 |
+
"learning_rate": 0.000188,
|
| 79 |
+
"loss": 0.2727,
|
| 80 |
+
"step": 120
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 6.19,
|
| 84 |
+
"learning_rate": 0.00018700000000000002,
|
| 85 |
+
"loss": 0.0707,
|
| 86 |
+
"step": 130
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 6.67,
|
| 90 |
+
"learning_rate": 0.00018600000000000002,
|
| 91 |
+
"loss": 0.2655,
|
| 92 |
+
"step": 140
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 7.14,
|
| 96 |
+
"learning_rate": 0.00018500000000000002,
|
| 97 |
+
"loss": 0.1479,
|
| 98 |
+
"step": 150
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 7.62,
|
| 102 |
+
"learning_rate": 0.00018400000000000003,
|
| 103 |
+
"loss": 0.1231,
|
| 104 |
+
"step": 160
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 8.1,
|
| 108 |
+
"learning_rate": 0.000183,
|
| 109 |
+
"loss": 0.0183,
|
| 110 |
+
"step": 170
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 8.57,
|
| 114 |
+
"learning_rate": 0.000182,
|
| 115 |
+
"loss": 0.0423,
|
| 116 |
+
"step": 180
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 9.05,
|
| 120 |
+
"learning_rate": 0.000181,
|
| 121 |
+
"loss": 0.057,
|
| 122 |
+
"step": 190
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 9.52,
|
| 126 |
+
"learning_rate": 0.00018,
|
| 127 |
+
"loss": 0.0082,
|
| 128 |
+
"step": 200
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 9.52,
|
| 132 |
+
"eval_accuracy": 1.0,
|
| 133 |
+
"eval_f1": 1.0,
|
| 134 |
+
"eval_loss": 0.006428680382668972,
|
| 135 |
+
"eval_precision": 1.0,
|
| 136 |
+
"eval_recall": 1.0,
|
| 137 |
+
"eval_runtime": 0.738,
|
| 138 |
+
"eval_samples_per_second": 75.879,
|
| 139 |
+
"eval_steps_per_second": 9.485,
|
| 140 |
+
"step": 200
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 10.0,
|
| 144 |
+
"learning_rate": 0.00017900000000000001,
|
| 145 |
+
"loss": 0.0118,
|
| 146 |
+
"step": 210
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 10.48,
|
| 150 |
+
"learning_rate": 0.00017800000000000002,
|
| 151 |
+
"loss": 0.007,
|
| 152 |
+
"step": 220
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 10.95,
|
| 156 |
+
"learning_rate": 0.00017700000000000002,
|
| 157 |
+
"loss": 0.0056,
|
| 158 |
+
"step": 230
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 11.43,
|
| 162 |
+
"learning_rate": 0.00017600000000000002,
|
| 163 |
+
"loss": 0.0048,
|
| 164 |
+
"step": 240
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 11.9,
|
| 168 |
+
"learning_rate": 0.000175,
|
| 169 |
+
"loss": 0.0045,
|
| 170 |
+
"step": 250
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 12.38,
|
| 174 |
+
"learning_rate": 0.000174,
|
| 175 |
+
"loss": 0.0042,
|
| 176 |
+
"step": 260
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 12.86,
|
| 180 |
+
"learning_rate": 0.000173,
|
| 181 |
+
"loss": 0.004,
|
| 182 |
+
"step": 270
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 13.33,
|
| 186 |
+
"learning_rate": 0.000172,
|
| 187 |
+
"loss": 0.0036,
|
| 188 |
+
"step": 280
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 13.81,
|
| 192 |
+
"learning_rate": 0.000171,
|
| 193 |
+
"loss": 0.0034,
|
| 194 |
+
"step": 290
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 14.29,
|
| 198 |
+
"learning_rate": 0.00017,
|
| 199 |
+
"loss": 0.0032,
|
| 200 |
+
"step": 300
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 14.76,
|
| 204 |
+
"learning_rate": 0.00016900000000000002,
|
| 205 |
+
"loss": 0.0031,
|
| 206 |
+
"step": 310
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 15.24,
|
| 210 |
+
"learning_rate": 0.000168,
|
| 211 |
+
"loss": 0.0029,
|
| 212 |
+
"step": 320
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 15.71,
|
| 216 |
+
"learning_rate": 0.000167,
|
| 217 |
+
"loss": 0.0028,
|
| 218 |
+
"step": 330
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 16.19,
|
| 222 |
+
"learning_rate": 0.000166,
|
| 223 |
+
"loss": 0.0027,
|
| 224 |
+
"step": 340
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 16.67,
|
| 228 |
+
"learning_rate": 0.000165,
|
| 229 |
+
"loss": 0.0025,
|
| 230 |
+
"step": 350
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 17.14,
|
| 234 |
+
"learning_rate": 0.000164,
|
| 235 |
+
"loss": 0.0025,
|
| 236 |
+
"step": 360
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 17.62,
|
| 240 |
+
"learning_rate": 0.000163,
|
| 241 |
+
"loss": 0.0024,
|
| 242 |
+
"step": 370
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 18.1,
|
| 246 |
+
"learning_rate": 0.000162,
|
| 247 |
+
"loss": 0.0023,
|
| 248 |
+
"step": 380
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 18.57,
|
| 252 |
+
"learning_rate": 0.000161,
|
| 253 |
+
"loss": 0.0021,
|
| 254 |
+
"step": 390
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 19.05,
|
| 258 |
+
"learning_rate": 0.00016,
|
| 259 |
+
"loss": 0.0021,
|
| 260 |
+
"step": 400
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 19.05,
|
| 264 |
+
"eval_accuracy": 1.0,
|
| 265 |
+
"eval_f1": 1.0,
|
| 266 |
+
"eval_loss": 0.0019897979218512774,
|
| 267 |
+
"eval_precision": 1.0,
|
| 268 |
+
"eval_recall": 1.0,
|
| 269 |
+
"eval_runtime": 0.713,
|
| 270 |
+
"eval_samples_per_second": 78.541,
|
| 271 |
+
"eval_steps_per_second": 9.818,
|
| 272 |
+
"step": 400
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 19.52,
|
| 276 |
+
"learning_rate": 0.00015900000000000002,
|
| 277 |
+
"loss": 0.002,
|
| 278 |
+
"step": 410
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 20.0,
|
| 282 |
+
"learning_rate": 0.00015800000000000002,
|
| 283 |
+
"loss": 0.0019,
|
| 284 |
+
"step": 420
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 20.48,
|
| 288 |
+
"learning_rate": 0.00015700000000000002,
|
| 289 |
+
"loss": 0.0019,
|
| 290 |
+
"step": 430
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 20.95,
|
| 294 |
+
"learning_rate": 0.00015600000000000002,
|
| 295 |
+
"loss": 0.0018,
|
| 296 |
+
"step": 440
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 21.43,
|
| 300 |
+
"learning_rate": 0.000155,
|
| 301 |
+
"loss": 0.0017,
|
| 302 |
+
"step": 450
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 21.9,
|
| 306 |
+
"learning_rate": 0.000154,
|
| 307 |
+
"loss": 0.0017,
|
| 308 |
+
"step": 460
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 22.38,
|
| 312 |
+
"learning_rate": 0.000153,
|
| 313 |
+
"loss": 0.0017,
|
| 314 |
+
"step": 470
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 22.86,
|
| 318 |
+
"learning_rate": 0.000152,
|
| 319 |
+
"loss": 0.0016,
|
| 320 |
+
"step": 480
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 23.33,
|
| 324 |
+
"learning_rate": 0.000151,
|
| 325 |
+
"loss": 0.0015,
|
| 326 |
+
"step": 490
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 23.81,
|
| 330 |
+
"learning_rate": 0.00015000000000000001,
|
| 331 |
+
"loss": 0.0015,
|
| 332 |
+
"step": 500
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 24.29,
|
| 336 |
+
"learning_rate": 0.00014900000000000002,
|
| 337 |
+
"loss": 0.0014,
|
| 338 |
+
"step": 510
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 24.76,
|
| 342 |
+
"learning_rate": 0.000148,
|
| 343 |
+
"loss": 0.0014,
|
| 344 |
+
"step": 520
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 25.24,
|
| 348 |
+
"learning_rate": 0.000147,
|
| 349 |
+
"loss": 0.0014,
|
| 350 |
+
"step": 530
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 25.71,
|
| 354 |
+
"learning_rate": 0.000146,
|
| 355 |
+
"loss": 0.0013,
|
| 356 |
+
"step": 540
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 26.19,
|
| 360 |
+
"learning_rate": 0.000145,
|
| 361 |
+
"loss": 0.0013,
|
| 362 |
+
"step": 550
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 26.67,
|
| 366 |
+
"learning_rate": 0.000144,
|
| 367 |
+
"loss": 0.0013,
|
| 368 |
+
"step": 560
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 27.14,
|
| 372 |
+
"learning_rate": 0.000143,
|
| 373 |
+
"loss": 0.0012,
|
| 374 |
+
"step": 570
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 27.62,
|
| 378 |
+
"learning_rate": 0.000142,
|
| 379 |
+
"loss": 0.0012,
|
| 380 |
+
"step": 580
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 28.1,
|
| 384 |
+
"learning_rate": 0.000141,
|
| 385 |
+
"loss": 0.0012,
|
| 386 |
+
"step": 590
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 28.57,
|
| 390 |
+
"learning_rate": 0.00014,
|
| 391 |
+
"loss": 0.0012,
|
| 392 |
+
"step": 600
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 28.57,
|
| 396 |
+
"eval_accuracy": 1.0,
|
| 397 |
+
"eval_f1": 1.0,
|
| 398 |
+
"eval_loss": 0.0011106810998171568,
|
| 399 |
+
"eval_precision": 1.0,
|
| 400 |
+
"eval_recall": 1.0,
|
| 401 |
+
"eval_runtime": 0.738,
|
| 402 |
+
"eval_samples_per_second": 75.881,
|
| 403 |
+
"eval_steps_per_second": 9.485,
|
| 404 |
+
"step": 600
|
| 405 |
+
}
|
| 406 |
+
],
|
| 407 |
+
"max_steps": 2000,
|
| 408 |
+
"num_train_epochs": 96,
|
| 409 |
+
"total_flos": 7.287346703356232e+17,
|
| 410 |
+
"trial_name": null,
|
| 411 |
+
"trial_params": null
|
| 412 |
+
}
|