End of training
Browse files
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"eval_accuracy": 0.9833333333333333,
|
| 4 |
+
"eval_loss": 0.05558411777019501,
|
| 5 |
+
"eval_runtime": 38.4928,
|
| 6 |
+
"eval_samples_per_second": 155.873,
|
| 7 |
+
"eval_steps_per_second": 4.884,
|
| 8 |
+
"total_flos": 1.342523444871168e+18,
|
| 9 |
+
"train_loss": 0.6594652506694975,
|
| 10 |
+
"train_runtime": 822.8009,
|
| 11 |
+
"train_samples_per_second": 65.629,
|
| 12 |
+
"train_steps_per_second": 0.513
|
| 13 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"eval_accuracy": 0.9833333333333333,
|
| 4 |
+
"eval_loss": 0.05558411777019501,
|
| 5 |
+
"eval_runtime": 38.4928,
|
| 6 |
+
"eval_samples_per_second": 155.873,
|
| 7 |
+
"eval_steps_per_second": 4.884
|
| 8 |
+
}
|
runs/Jun21_08-56-34_15d408464ff9/events.out.tfevents.1655802836.15d408464ff9.75.8
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c6fa3a3eb5bc736a4eccbd43fcdd83150ea5a737aefb417c12a5cbb678d19b64
|
| 3 |
+
size 363
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"total_flos": 1.342523444871168e+18,
|
| 4 |
+
"train_loss": 0.6594652506694975,
|
| 5 |
+
"train_runtime": 822.8009,
|
| 6 |
+
"train_samples_per_second": 65.629,
|
| 7 |
+
"train_steps_per_second": 0.513
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,286 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.9833333333333333,
|
| 3 |
+
"best_model_checkpoint": "image-classification/checkpoint-422",
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
+
"global_step": 422,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.02,
|
| 12 |
+
"learning_rate": 1.1627906976744187e-05,
|
| 13 |
+
"loss": 2.3142,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.05,
|
| 18 |
+
"learning_rate": 2.3255813953488374e-05,
|
| 19 |
+
"loss": 2.2058,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.07,
|
| 24 |
+
"learning_rate": 3.488372093023256e-05,
|
| 25 |
+
"loss": 1.893,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.09,
|
| 30 |
+
"learning_rate": 4.651162790697675e-05,
|
| 31 |
+
"loss": 1.4224,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.12,
|
| 36 |
+
"learning_rate": 4.907651715039578e-05,
|
| 37 |
+
"loss": 1.0782,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.14,
|
| 42 |
+
"learning_rate": 4.775725593667547e-05,
|
| 43 |
+
"loss": 0.877,
|
| 44 |
+
"step": 60
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.17,
|
| 48 |
+
"learning_rate": 4.6437994722955145e-05,
|
| 49 |
+
"loss": 0.778,
|
| 50 |
+
"step": 70
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.19,
|
| 54 |
+
"learning_rate": 4.511873350923483e-05,
|
| 55 |
+
"loss": 0.6735,
|
| 56 |
+
"step": 80
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.21,
|
| 60 |
+
"learning_rate": 4.379947229551452e-05,
|
| 61 |
+
"loss": 0.6688,
|
| 62 |
+
"step": 90
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.24,
|
| 66 |
+
"learning_rate": 4.2480211081794194e-05,
|
| 67 |
+
"loss": 0.6841,
|
| 68 |
+
"step": 100
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.26,
|
| 72 |
+
"learning_rate": 4.116094986807388e-05,
|
| 73 |
+
"loss": 0.5935,
|
| 74 |
+
"step": 110
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.28,
|
| 78 |
+
"learning_rate": 3.984168865435357e-05,
|
| 79 |
+
"loss": 0.5521,
|
| 80 |
+
"step": 120
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.31,
|
| 84 |
+
"learning_rate": 3.8522427440633244e-05,
|
| 85 |
+
"loss": 0.5555,
|
| 86 |
+
"step": 130
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.33,
|
| 90 |
+
"learning_rate": 3.7203166226912934e-05,
|
| 91 |
+
"loss": 0.5483,
|
| 92 |
+
"step": 140
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.36,
|
| 96 |
+
"learning_rate": 3.588390501319262e-05,
|
| 97 |
+
"loss": 0.5371,
|
| 98 |
+
"step": 150
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.38,
|
| 102 |
+
"learning_rate": 3.456464379947229e-05,
|
| 103 |
+
"loss": 0.5485,
|
| 104 |
+
"step": 160
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.4,
|
| 108 |
+
"learning_rate": 3.324538258575198e-05,
|
| 109 |
+
"loss": 0.526,
|
| 110 |
+
"step": 170
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.43,
|
| 114 |
+
"learning_rate": 3.1926121372031666e-05,
|
| 115 |
+
"loss": 0.5255,
|
| 116 |
+
"step": 180
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.45,
|
| 120 |
+
"learning_rate": 3.060686015831134e-05,
|
| 121 |
+
"loss": 0.5133,
|
| 122 |
+
"step": 190
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.47,
|
| 126 |
+
"learning_rate": 2.9287598944591033e-05,
|
| 127 |
+
"loss": 0.4857,
|
| 128 |
+
"step": 200
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.5,
|
| 132 |
+
"learning_rate": 2.7968337730870712e-05,
|
| 133 |
+
"loss": 0.4806,
|
| 134 |
+
"step": 210
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.52,
|
| 138 |
+
"learning_rate": 2.6649076517150396e-05,
|
| 139 |
+
"loss": 0.4573,
|
| 140 |
+
"step": 220
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.55,
|
| 144 |
+
"learning_rate": 2.5329815303430082e-05,
|
| 145 |
+
"loss": 0.4995,
|
| 146 |
+
"step": 230
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.57,
|
| 150 |
+
"learning_rate": 2.4010554089709765e-05,
|
| 151 |
+
"loss": 0.4744,
|
| 152 |
+
"step": 240
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.59,
|
| 156 |
+
"learning_rate": 2.269129287598945e-05,
|
| 157 |
+
"loss": 0.4543,
|
| 158 |
+
"step": 250
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.62,
|
| 162 |
+
"learning_rate": 2.1372031662269128e-05,
|
| 163 |
+
"loss": 0.4613,
|
| 164 |
+
"step": 260
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.64,
|
| 168 |
+
"learning_rate": 2.0052770448548815e-05,
|
| 169 |
+
"loss": 0.475,
|
| 170 |
+
"step": 270
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.66,
|
| 174 |
+
"learning_rate": 1.8733509234828498e-05,
|
| 175 |
+
"loss": 0.4324,
|
| 176 |
+
"step": 280
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.69,
|
| 180 |
+
"learning_rate": 1.741424802110818e-05,
|
| 181 |
+
"loss": 0.4003,
|
| 182 |
+
"step": 290
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.71,
|
| 186 |
+
"learning_rate": 1.6094986807387864e-05,
|
| 187 |
+
"loss": 0.4409,
|
| 188 |
+
"step": 300
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.73,
|
| 192 |
+
"learning_rate": 1.4775725593667547e-05,
|
| 193 |
+
"loss": 0.492,
|
| 194 |
+
"step": 310
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.76,
|
| 198 |
+
"learning_rate": 1.345646437994723e-05,
|
| 199 |
+
"loss": 0.4541,
|
| 200 |
+
"step": 320
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.78,
|
| 204 |
+
"learning_rate": 1.2137203166226914e-05,
|
| 205 |
+
"loss": 0.4202,
|
| 206 |
+
"step": 330
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.81,
|
| 210 |
+
"learning_rate": 1.0817941952506597e-05,
|
| 211 |
+
"loss": 0.4371,
|
| 212 |
+
"step": 340
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.83,
|
| 216 |
+
"learning_rate": 9.49868073878628e-06,
|
| 217 |
+
"loss": 0.4339,
|
| 218 |
+
"step": 350
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.85,
|
| 222 |
+
"learning_rate": 8.179419525065963e-06,
|
| 223 |
+
"loss": 0.4083,
|
| 224 |
+
"step": 360
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 0.88,
|
| 228 |
+
"learning_rate": 6.860158311345646e-06,
|
| 229 |
+
"loss": 0.4668,
|
| 230 |
+
"step": 370
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.9,
|
| 234 |
+
"learning_rate": 5.540897097625329e-06,
|
| 235 |
+
"loss": 0.4238,
|
| 236 |
+
"step": 380
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.92,
|
| 240 |
+
"learning_rate": 4.221635883905013e-06,
|
| 241 |
+
"loss": 0.434,
|
| 242 |
+
"step": 390
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.95,
|
| 246 |
+
"learning_rate": 2.9023746701846966e-06,
|
| 247 |
+
"loss": 0.4309,
|
| 248 |
+
"step": 400
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.97,
|
| 252 |
+
"learning_rate": 1.5831134564643801e-06,
|
| 253 |
+
"loss": 0.4218,
|
| 254 |
+
"step": 410
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 1.0,
|
| 258 |
+
"learning_rate": 2.6385224274406334e-07,
|
| 259 |
+
"loss": 0.3743,
|
| 260 |
+
"step": 420
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 1.0,
|
| 264 |
+
"eval_accuracy": 0.9833333333333333,
|
| 265 |
+
"eval_loss": 0.05558411777019501,
|
| 266 |
+
"eval_runtime": 38.6009,
|
| 267 |
+
"eval_samples_per_second": 155.437,
|
| 268 |
+
"eval_steps_per_second": 4.87,
|
| 269 |
+
"step": 422
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 1.0,
|
| 273 |
+
"step": 422,
|
| 274 |
+
"total_flos": 1.342523444871168e+18,
|
| 275 |
+
"train_loss": 0.6594652506694975,
|
| 276 |
+
"train_runtime": 822.8009,
|
| 277 |
+
"train_samples_per_second": 65.629,
|
| 278 |
+
"train_steps_per_second": 0.513
|
| 279 |
+
}
|
| 280 |
+
],
|
| 281 |
+
"max_steps": 422,
|
| 282 |
+
"num_train_epochs": 1,
|
| 283 |
+
"total_flos": 1.342523444871168e+18,
|
| 284 |
+
"trial_name": null,
|
| 285 |
+
"trial_params": null
|
| 286 |
+
}
|