Commit ·
2d52288
1
Parent(s): 7260fd8
End of training
Browse files- all_results.json +13 -0
- eval_results.json +8 -0
- train_results.json +8 -0
- trainer_state.json +277 -0
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 9.98,
|
| 3 |
+
"eval_accuracy": 0.9933333333333333,
|
| 4 |
+
"eval_loss": 0.030707545578479767,
|
| 5 |
+
"eval_runtime": 47.7279,
|
| 6 |
+
"eval_samples_per_second": 15.714,
|
| 7 |
+
"eval_steps_per_second": 0.503,
|
| 8 |
+
"total_flos": 8.68938785610965e+17,
|
| 9 |
+
"train_loss": 0.19813838097508307,
|
| 10 |
+
"train_runtime": 3233.657,
|
| 11 |
+
"train_samples_per_second": 10.824,
|
| 12 |
+
"train_steps_per_second": 0.083
|
| 13 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 9.98,
|
| 3 |
+
"eval_accuracy": 0.9933333333333333,
|
| 4 |
+
"eval_loss": 0.030707545578479767,
|
| 5 |
+
"eval_runtime": 47.7279,
|
| 6 |
+
"eval_samples_per_second": 15.714,
|
| 7 |
+
"eval_steps_per_second": 0.503
|
| 8 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 9.98,
|
| 3 |
+
"total_flos": 8.68938785610965e+17,
|
| 4 |
+
"train_loss": 0.19813838097508307,
|
| 5 |
+
"train_runtime": 3233.657,
|
| 6 |
+
"train_samples_per_second": 10.824,
|
| 7 |
+
"train_steps_per_second": 0.083
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,277 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.9933333333333333,
|
| 3 |
+
"best_model_checkpoint": "swin-tiny-patch4-window7-224-lcbsi-wbc/checkpoint-135",
|
| 4 |
+
"epoch": 9.981818181818182,
|
| 5 |
+
"global_step": 270,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.36,
|
| 12 |
+
"learning_rate": 1.8518518518518518e-05,
|
| 13 |
+
"loss": 1.5888,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.73,
|
| 18 |
+
"learning_rate": 3.7037037037037037e-05,
|
| 19 |
+
"loss": 1.3668,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.98,
|
| 24 |
+
"eval_accuracy": 0.76,
|
| 25 |
+
"eval_loss": 0.6976749897003174,
|
| 26 |
+
"eval_runtime": 47.8881,
|
| 27 |
+
"eval_samples_per_second": 15.662,
|
| 28 |
+
"eval_steps_per_second": 0.501,
|
| 29 |
+
"step": 27
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"epoch": 1.11,
|
| 33 |
+
"learning_rate": 4.938271604938271e-05,
|
| 34 |
+
"loss": 0.8641,
|
| 35 |
+
"step": 30
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"epoch": 1.47,
|
| 39 |
+
"learning_rate": 4.732510288065844e-05,
|
| 40 |
+
"loss": 0.3824,
|
| 41 |
+
"step": 40
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"epoch": 1.84,
|
| 45 |
+
"learning_rate": 4.5267489711934157e-05,
|
| 46 |
+
"loss": 0.217,
|
| 47 |
+
"step": 50
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"epoch": 1.98,
|
| 51 |
+
"eval_accuracy": 0.972,
|
| 52 |
+
"eval_loss": 0.09920086711645126,
|
| 53 |
+
"eval_runtime": 46.0407,
|
| 54 |
+
"eval_samples_per_second": 16.29,
|
| 55 |
+
"eval_steps_per_second": 0.521,
|
| 56 |
+
"step": 54
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 2.22,
|
| 60 |
+
"learning_rate": 4.3209876543209875e-05,
|
| 61 |
+
"loss": 0.1489,
|
| 62 |
+
"step": 60
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 2.58,
|
| 66 |
+
"learning_rate": 4.11522633744856e-05,
|
| 67 |
+
"loss": 0.119,
|
| 68 |
+
"step": 70
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 2.95,
|
| 72 |
+
"learning_rate": 3.909465020576132e-05,
|
| 73 |
+
"loss": 0.102,
|
| 74 |
+
"step": 80
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 2.98,
|
| 78 |
+
"eval_accuracy": 0.9853333333333333,
|
| 79 |
+
"eval_loss": 0.05732276290655136,
|
| 80 |
+
"eval_runtime": 45.2518,
|
| 81 |
+
"eval_samples_per_second": 16.574,
|
| 82 |
+
"eval_steps_per_second": 0.53,
|
| 83 |
+
"step": 81
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 3.33,
|
| 87 |
+
"learning_rate": 3.7037037037037037e-05,
|
| 88 |
+
"loss": 0.0703,
|
| 89 |
+
"step": 90
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 3.69,
|
| 93 |
+
"learning_rate": 3.497942386831276e-05,
|
| 94 |
+
"loss": 0.0762,
|
| 95 |
+
"step": 100
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 3.98,
|
| 99 |
+
"eval_accuracy": 0.976,
|
| 100 |
+
"eval_loss": 0.10027103871107101,
|
| 101 |
+
"eval_runtime": 45.6824,
|
| 102 |
+
"eval_samples_per_second": 16.418,
|
| 103 |
+
"eval_steps_per_second": 0.525,
|
| 104 |
+
"step": 108
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 4.07,
|
| 108 |
+
"learning_rate": 3.292181069958848e-05,
|
| 109 |
+
"loss": 0.0979,
|
| 110 |
+
"step": 110
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 4.44,
|
| 114 |
+
"learning_rate": 3.08641975308642e-05,
|
| 115 |
+
"loss": 0.0633,
|
| 116 |
+
"step": 120
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 4.8,
|
| 120 |
+
"learning_rate": 2.880658436213992e-05,
|
| 121 |
+
"loss": 0.0456,
|
| 122 |
+
"step": 130
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 4.98,
|
| 126 |
+
"eval_accuracy": 0.9933333333333333,
|
| 127 |
+
"eval_loss": 0.030707545578479767,
|
| 128 |
+
"eval_runtime": 47.112,
|
| 129 |
+
"eval_samples_per_second": 15.92,
|
| 130 |
+
"eval_steps_per_second": 0.509,
|
| 131 |
+
"step": 135
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 5.18,
|
| 135 |
+
"learning_rate": 2.6748971193415638e-05,
|
| 136 |
+
"loss": 0.0369,
|
| 137 |
+
"step": 140
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 5.55,
|
| 141 |
+
"learning_rate": 2.4691358024691357e-05,
|
| 142 |
+
"loss": 0.036,
|
| 143 |
+
"step": 150
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 5.91,
|
| 147 |
+
"learning_rate": 2.2633744855967078e-05,
|
| 148 |
+
"loss": 0.0219,
|
| 149 |
+
"step": 160
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 5.98,
|
| 153 |
+
"eval_accuracy": 0.9906666666666667,
|
| 154 |
+
"eval_loss": 0.04968710243701935,
|
| 155 |
+
"eval_runtime": 46.2612,
|
| 156 |
+
"eval_samples_per_second": 16.212,
|
| 157 |
+
"eval_steps_per_second": 0.519,
|
| 158 |
+
"step": 162
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 6.29,
|
| 162 |
+
"learning_rate": 2.05761316872428e-05,
|
| 163 |
+
"loss": 0.0149,
|
| 164 |
+
"step": 170
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 6.65,
|
| 168 |
+
"learning_rate": 1.8518518518518518e-05,
|
| 169 |
+
"loss": 0.0106,
|
| 170 |
+
"step": 180
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 6.98,
|
| 174 |
+
"eval_accuracy": 0.9866666666666667,
|
| 175 |
+
"eval_loss": 0.05681763216853142,
|
| 176 |
+
"eval_runtime": 46.7081,
|
| 177 |
+
"eval_samples_per_second": 16.057,
|
| 178 |
+
"eval_steps_per_second": 0.514,
|
| 179 |
+
"step": 189
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 7.04,
|
| 183 |
+
"learning_rate": 1.646090534979424e-05,
|
| 184 |
+
"loss": 0.0194,
|
| 185 |
+
"step": 190
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 7.4,
|
| 189 |
+
"learning_rate": 1.440329218106996e-05,
|
| 190 |
+
"loss": 0.009,
|
| 191 |
+
"step": 200
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 7.76,
|
| 195 |
+
"learning_rate": 1.2345679012345678e-05,
|
| 196 |
+
"loss": 0.0112,
|
| 197 |
+
"step": 210
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 7.98,
|
| 201 |
+
"eval_accuracy": 0.9906666666666667,
|
| 202 |
+
"eval_loss": 0.05316087231040001,
|
| 203 |
+
"eval_runtime": 46.5686,
|
| 204 |
+
"eval_samples_per_second": 16.105,
|
| 205 |
+
"eval_steps_per_second": 0.515,
|
| 206 |
+
"step": 216
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 8.15,
|
| 210 |
+
"learning_rate": 1.02880658436214e-05,
|
| 211 |
+
"loss": 0.0081,
|
| 212 |
+
"step": 220
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 8.51,
|
| 216 |
+
"learning_rate": 8.23045267489712e-06,
|
| 217 |
+
"loss": 0.0085,
|
| 218 |
+
"step": 230
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 8.87,
|
| 222 |
+
"learning_rate": 6.172839506172839e-06,
|
| 223 |
+
"loss": 0.0067,
|
| 224 |
+
"step": 240
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 8.98,
|
| 228 |
+
"eval_accuracy": 0.9906666666666667,
|
| 229 |
+
"eval_loss": 0.052792128175497055,
|
| 230 |
+
"eval_runtime": 46.6713,
|
| 231 |
+
"eval_samples_per_second": 16.07,
|
| 232 |
+
"eval_steps_per_second": 0.514,
|
| 233 |
+
"step": 243
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 9.25,
|
| 237 |
+
"learning_rate": 4.11522633744856e-06,
|
| 238 |
+
"loss": 0.0119,
|
| 239 |
+
"step": 250
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 9.62,
|
| 243 |
+
"learning_rate": 2.05761316872428e-06,
|
| 244 |
+
"loss": 0.0043,
|
| 245 |
+
"step": 260
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 9.98,
|
| 249 |
+
"learning_rate": 0.0,
|
| 250 |
+
"loss": 0.008,
|
| 251 |
+
"step": 270
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 9.98,
|
| 255 |
+
"eval_accuracy": 0.992,
|
| 256 |
+
"eval_loss": 0.048162445425987244,
|
| 257 |
+
"eval_runtime": 48.9109,
|
| 258 |
+
"eval_samples_per_second": 15.334,
|
| 259 |
+
"eval_steps_per_second": 0.491,
|
| 260 |
+
"step": 270
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 9.98,
|
| 264 |
+
"step": 270,
|
| 265 |
+
"total_flos": 8.68938785610965e+17,
|
| 266 |
+
"train_loss": 0.19813838097508307,
|
| 267 |
+
"train_runtime": 3233.657,
|
| 268 |
+
"train_samples_per_second": 10.824,
|
| 269 |
+
"train_steps_per_second": 0.083
|
| 270 |
+
}
|
| 271 |
+
],
|
| 272 |
+
"max_steps": 270,
|
| 273 |
+
"num_train_epochs": 10,
|
| 274 |
+
"total_flos": 8.68938785610965e+17,
|
| 275 |
+
"trial_name": null,
|
| 276 |
+
"trial_params": null
|
| 277 |
+
}
|