🍻 cheers
Browse files- all_results.json +8 -0
- train_results.json +8 -0
- trainer_state.json +325 -0
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 10.0,
|
| 3 |
+
"total_flos": 5.354696482368307e+17,
|
| 4 |
+
"train_loss": 0.011267724910496988,
|
| 5 |
+
"train_runtime": 422.1478,
|
| 6 |
+
"train_samples_per_second": 16.369,
|
| 7 |
+
"train_steps_per_second": 1.042
|
| 8 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 10.0,
|
| 3 |
+
"total_flos": 5.354696482368307e+17,
|
| 4 |
+
"train_loss": 0.011267724910496988,
|
| 5 |
+
"train_runtime": 422.1478,
|
| 6 |
+
"train_samples_per_second": 16.369,
|
| 7 |
+
"train_steps_per_second": 1.042
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,325 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.0006729721208103001,
|
| 3 |
+
"best_model_checkpoint": "./vit-base-blur/checkpoint-400",
|
| 4 |
+
"epoch": 10.0,
|
| 5 |
+
"global_step": 440,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.23,
|
| 12 |
+
"learning_rate": 0.00019545454545454548,
|
| 13 |
+
"loss": 0.0392,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.45,
|
| 18 |
+
"learning_rate": 0.00019090909090909092,
|
| 19 |
+
"loss": 0.1092,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.68,
|
| 24 |
+
"learning_rate": 0.00018636363636363636,
|
| 25 |
+
"loss": 0.064,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.91,
|
| 30 |
+
"learning_rate": 0.00018181818181818183,
|
| 31 |
+
"loss": 0.0254,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 1.14,
|
| 36 |
+
"learning_rate": 0.00017727272727272728,
|
| 37 |
+
"loss": 0.0662,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 1.36,
|
| 42 |
+
"learning_rate": 0.00017272727272727275,
|
| 43 |
+
"loss": 0.0306,
|
| 44 |
+
"step": 60
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 1.59,
|
| 48 |
+
"learning_rate": 0.0001681818181818182,
|
| 49 |
+
"loss": 0.0073,
|
| 50 |
+
"step": 70
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 1.82,
|
| 54 |
+
"learning_rate": 0.00016363636363636366,
|
| 55 |
+
"loss": 0.0231,
|
| 56 |
+
"step": 80
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 2.05,
|
| 60 |
+
"learning_rate": 0.0001590909090909091,
|
| 61 |
+
"loss": 0.0029,
|
| 62 |
+
"step": 90
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 2.27,
|
| 66 |
+
"learning_rate": 0.00015454545454545454,
|
| 67 |
+
"loss": 0.0027,
|
| 68 |
+
"step": 100
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 2.27,
|
| 72 |
+
"eval_accuracy": 0.987012987012987,
|
| 73 |
+
"eval_loss": 0.05653372034430504,
|
| 74 |
+
"eval_runtime": 2.757,
|
| 75 |
+
"eval_samples_per_second": 27.928,
|
| 76 |
+
"eval_steps_per_second": 3.627,
|
| 77 |
+
"step": 100
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 2.5,
|
| 81 |
+
"learning_rate": 0.00015000000000000001,
|
| 82 |
+
"loss": 0.0022,
|
| 83 |
+
"step": 110
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 2.73,
|
| 87 |
+
"learning_rate": 0.00014545454545454546,
|
| 88 |
+
"loss": 0.0226,
|
| 89 |
+
"step": 120
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 2.95,
|
| 93 |
+
"learning_rate": 0.00014090909090909093,
|
| 94 |
+
"loss": 0.02,
|
| 95 |
+
"step": 130
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 3.18,
|
| 99 |
+
"learning_rate": 0.00013636363636363637,
|
| 100 |
+
"loss": 0.0137,
|
| 101 |
+
"step": 140
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 3.41,
|
| 105 |
+
"learning_rate": 0.0001318181818181818,
|
| 106 |
+
"loss": 0.0016,
|
| 107 |
+
"step": 150
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 3.64,
|
| 111 |
+
"learning_rate": 0.00012727272727272728,
|
| 112 |
+
"loss": 0.0016,
|
| 113 |
+
"step": 160
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 3.86,
|
| 117 |
+
"learning_rate": 0.00012272727272727272,
|
| 118 |
+
"loss": 0.037,
|
| 119 |
+
"step": 170
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 4.09,
|
| 123 |
+
"learning_rate": 0.0001181818181818182,
|
| 124 |
+
"loss": 0.0016,
|
| 125 |
+
"step": 180
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 4.32,
|
| 129 |
+
"learning_rate": 0.00011363636363636365,
|
| 130 |
+
"loss": 0.0041,
|
| 131 |
+
"step": 190
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 4.55,
|
| 135 |
+
"learning_rate": 0.00010909090909090909,
|
| 136 |
+
"loss": 0.0013,
|
| 137 |
+
"step": 200
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 4.55,
|
| 141 |
+
"eval_accuracy": 1.0,
|
| 142 |
+
"eval_loss": 0.001206980086863041,
|
| 143 |
+
"eval_runtime": 2.4286,
|
| 144 |
+
"eval_samples_per_second": 31.706,
|
| 145 |
+
"eval_steps_per_second": 4.118,
|
| 146 |
+
"step": 200
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 4.77,
|
| 150 |
+
"learning_rate": 0.00010454545454545455,
|
| 151 |
+
"loss": 0.0012,
|
| 152 |
+
"step": 210
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 5.0,
|
| 156 |
+
"learning_rate": 0.0001,
|
| 157 |
+
"loss": 0.0011,
|
| 158 |
+
"step": 220
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 5.23,
|
| 162 |
+
"learning_rate": 9.545454545454546e-05,
|
| 163 |
+
"loss": 0.0011,
|
| 164 |
+
"step": 230
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 5.45,
|
| 168 |
+
"learning_rate": 9.090909090909092e-05,
|
| 169 |
+
"loss": 0.001,
|
| 170 |
+
"step": 240
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 5.68,
|
| 174 |
+
"learning_rate": 8.636363636363637e-05,
|
| 175 |
+
"loss": 0.001,
|
| 176 |
+
"step": 250
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 5.91,
|
| 180 |
+
"learning_rate": 8.181818181818183e-05,
|
| 181 |
+
"loss": 0.0009,
|
| 182 |
+
"step": 260
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 6.14,
|
| 186 |
+
"learning_rate": 7.727272727272727e-05,
|
| 187 |
+
"loss": 0.0009,
|
| 188 |
+
"step": 270
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 6.36,
|
| 192 |
+
"learning_rate": 7.272727272727273e-05,
|
| 193 |
+
"loss": 0.0009,
|
| 194 |
+
"step": 280
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 6.59,
|
| 198 |
+
"learning_rate": 6.818181818181818e-05,
|
| 199 |
+
"loss": 0.0008,
|
| 200 |
+
"step": 290
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 6.82,
|
| 204 |
+
"learning_rate": 6.363636363636364e-05,
|
| 205 |
+
"loss": 0.0008,
|
| 206 |
+
"step": 300
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 6.82,
|
| 210 |
+
"eval_accuracy": 1.0,
|
| 211 |
+
"eval_loss": 0.0008008014992810786,
|
| 212 |
+
"eval_runtime": 1.9546,
|
| 213 |
+
"eval_samples_per_second": 39.395,
|
| 214 |
+
"eval_steps_per_second": 5.116,
|
| 215 |
+
"step": 300
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 7.05,
|
| 219 |
+
"learning_rate": 5.90909090909091e-05,
|
| 220 |
+
"loss": 0.0008,
|
| 221 |
+
"step": 310
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 7.27,
|
| 225 |
+
"learning_rate": 5.4545454545454546e-05,
|
| 226 |
+
"loss": 0.0008,
|
| 227 |
+
"step": 320
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 7.5,
|
| 231 |
+
"learning_rate": 5e-05,
|
| 232 |
+
"loss": 0.0007,
|
| 233 |
+
"step": 330
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 7.73,
|
| 237 |
+
"learning_rate": 4.545454545454546e-05,
|
| 238 |
+
"loss": 0.0007,
|
| 239 |
+
"step": 340
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 7.95,
|
| 243 |
+
"learning_rate": 4.0909090909090915e-05,
|
| 244 |
+
"loss": 0.0007,
|
| 245 |
+
"step": 350
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 8.18,
|
| 249 |
+
"learning_rate": 3.6363636363636364e-05,
|
| 250 |
+
"loss": 0.0007,
|
| 251 |
+
"step": 360
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 8.41,
|
| 255 |
+
"learning_rate": 3.181818181818182e-05,
|
| 256 |
+
"loss": 0.0007,
|
| 257 |
+
"step": 370
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 8.64,
|
| 261 |
+
"learning_rate": 2.7272727272727273e-05,
|
| 262 |
+
"loss": 0.0007,
|
| 263 |
+
"step": 380
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 8.86,
|
| 267 |
+
"learning_rate": 2.272727272727273e-05,
|
| 268 |
+
"loss": 0.0007,
|
| 269 |
+
"step": 390
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 9.09,
|
| 273 |
+
"learning_rate": 1.8181818181818182e-05,
|
| 274 |
+
"loss": 0.0007,
|
| 275 |
+
"step": 400
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 9.09,
|
| 279 |
+
"eval_accuracy": 1.0,
|
| 280 |
+
"eval_loss": 0.0006729721208103001,
|
| 281 |
+
"eval_runtime": 1.9389,
|
| 282 |
+
"eval_samples_per_second": 39.714,
|
| 283 |
+
"eval_steps_per_second": 5.158,
|
| 284 |
+
"step": 400
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 9.32,
|
| 288 |
+
"learning_rate": 1.3636363636363637e-05,
|
| 289 |
+
"loss": 0.0007,
|
| 290 |
+
"step": 410
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 9.55,
|
| 294 |
+
"learning_rate": 9.090909090909091e-06,
|
| 295 |
+
"loss": 0.0007,
|
| 296 |
+
"step": 420
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 9.77,
|
| 300 |
+
"learning_rate": 4.5454545454545455e-06,
|
| 301 |
+
"loss": 0.0007,
|
| 302 |
+
"step": 430
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 10.0,
|
| 306 |
+
"learning_rate": 0.0,
|
| 307 |
+
"loss": 0.0007,
|
| 308 |
+
"step": 440
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 10.0,
|
| 312 |
+
"step": 440,
|
| 313 |
+
"total_flos": 5.354696482368307e+17,
|
| 314 |
+
"train_loss": 0.011267724910496988,
|
| 315 |
+
"train_runtime": 422.1478,
|
| 316 |
+
"train_samples_per_second": 16.369,
|
| 317 |
+
"train_steps_per_second": 1.042
|
| 318 |
+
}
|
| 319 |
+
],
|
| 320 |
+
"max_steps": 440,
|
| 321 |
+
"num_train_epochs": 10,
|
| 322 |
+
"total_flos": 5.354696482368307e+17,
|
| 323 |
+
"trial_name": null,
|
| 324 |
+
"trial_params": null
|
| 325 |
+
}
|