Training in progress, epoch 1
Browse files- .gitattributes +7 -0
- Screenshot 2023-10-21 at 21-56-46 TensorBoard.png +0 -0
- Screenshot 2023-10-21 at 21-58-51 TensorBoard.png +0 -0
- Screenshot 2023-10-21 at 21-59-25 TensorBoard.png +0 -0
- all_results.json +12 -12
- eval_results.json +8 -8
- pytorch_model.bin +1 -1
- runs/Oct21_13-32-53_EE4E077/events.out.tfevents.1697868948.EE4E077.986268.11 +3 -0
- runs/Oct26_14-36-46_EE4E077/1698302209.970762/events.out.tfevents.1698302209.EE4E077.357099.3 +3 -0
- runs/Oct26_14-36-46_EE4E077/events.out.tfevents.1698302209.EE4E077.357099.2 +3 -0
- train_results.json +4 -4
- trainer_state.json +321 -321
- training_args.bin +1 -1
.gitattributes
CHANGED
|
@@ -33,3 +33,10 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
| 34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
| 35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
| 36 |
+
2023-10-21 filter=lfs diff=lfs merge=lfs -text
|
| 37 |
+
at filter=lfs diff=lfs merge=lfs -text
|
| 38 |
+
21-56-46 filter=lfs diff=lfs merge=lfs -text
|
| 39 |
+
TensorBoard.png filter=lfs diff=lfs merge=lfs -text
|
| 40 |
+
Screenshot filter=lfs diff=lfs merge=lfs -text
|
| 41 |
+
21-58-51 filter=lfs diff=lfs merge=lfs -text
|
| 42 |
+
21-59-25 filter=lfs diff=lfs merge=lfs -text
|
Screenshot 2023-10-21 at 21-56-46 TensorBoard.png
ADDED
|
Screenshot 2023-10-21 at 21-58-51 TensorBoard.png
ADDED
|
Screenshot 2023-10-21 at 21-59-25 TensorBoard.png
ADDED
|
all_results.json
CHANGED
|
@@ -1,16 +1,16 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 4.0,
|
| 3 |
-
"eval_accuracy": 0.
|
| 4 |
-
"eval_f1": 0.
|
| 5 |
-
"eval_loss": 0.
|
| 6 |
-
"eval_precision": 0.
|
| 7 |
-
"eval_recall": 0.
|
| 8 |
-
"eval_runtime": 8.
|
| 9 |
-
"eval_samples_per_second":
|
| 10 |
-
"eval_steps_per_second": 10.
|
| 11 |
"total_flos": 3.104468219559813e+18,
|
| 12 |
-
"train_loss": 0.
|
| 13 |
-
"train_runtime":
|
| 14 |
-
"train_samples_per_second": 47.
|
| 15 |
-
"train_steps_per_second": 2.
|
| 16 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 4.0,
|
| 3 |
+
"eval_accuracy": 0.7718253968253969,
|
| 4 |
+
"eval_f1": 0.7520737401500063,
|
| 5 |
+
"eval_loss": 0.8010103702545166,
|
| 6 |
+
"eval_precision": 0.7604704688179398,
|
| 7 |
+
"eval_recall": 0.7718253968253969,
|
| 8 |
+
"eval_runtime": 8.8001,
|
| 9 |
+
"eval_samples_per_second": 171.815,
|
| 10 |
+
"eval_steps_per_second": 10.795,
|
| 11 |
"total_flos": 3.104468219559813e+18,
|
| 12 |
+
"train_loss": 0.06904298590738683,
|
| 13 |
+
"train_runtime": 834.8541,
|
| 14 |
+
"train_samples_per_second": 47.984,
|
| 15 |
+
"train_steps_per_second": 2.999
|
| 16 |
}
|
eval_results.json
CHANGED
|
@@ -1,11 +1,11 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 4.0,
|
| 3 |
-
"eval_accuracy": 0.
|
| 4 |
-
"eval_f1": 0.
|
| 5 |
-
"eval_loss": 0.
|
| 6 |
-
"eval_precision": 0.
|
| 7 |
-
"eval_recall": 0.
|
| 8 |
-
"eval_runtime": 8.
|
| 9 |
-
"eval_samples_per_second":
|
| 10 |
-
"eval_steps_per_second": 10.
|
| 11 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 4.0,
|
| 3 |
+
"eval_accuracy": 0.7718253968253969,
|
| 4 |
+
"eval_f1": 0.7520737401500063,
|
| 5 |
+
"eval_loss": 0.8010103702545166,
|
| 6 |
+
"eval_precision": 0.7604704688179398,
|
| 7 |
+
"eval_recall": 0.7718253968253969,
|
| 8 |
+
"eval_runtime": 8.8001,
|
| 9 |
+
"eval_samples_per_second": 171.815,
|
| 10 |
+
"eval_steps_per_second": 10.795
|
| 11 |
}
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 343284077
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:13d2d5477ba9db242ca7d11efd60356bfad7daea63b6ac969254a9e5fcee5289
|
| 3 |
size 343284077
|
runs/Oct21_13-32-53_EE4E077/events.out.tfevents.1697868948.EE4E077.986268.11
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8cc53c8d917aea24374564c3e501385d59dbfde1f2a39cc168c946ce1dc72748
|
| 3 |
+
size 560
|
runs/Oct26_14-36-46_EE4E077/1698302209.970762/events.out.tfevents.1698302209.EE4E077.357099.3
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3b1d85442bd4444bc32162177b80145b06a655ac0a53929b8218a5e8a9347cef
|
| 3 |
+
size 5913
|
runs/Oct26_14-36-46_EE4E077/events.out.tfevents.1698302209.EE4E077.357099.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:842c37af403c95c822da3cd6449678150989b988e524297d0ca55238eac0ff62
|
| 3 |
+
size 14323
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 4.0,
|
| 3 |
"total_flos": 3.104468219559813e+18,
|
| 4 |
-
"train_loss": 0.
|
| 5 |
-
"train_runtime":
|
| 6 |
-
"train_samples_per_second": 47.
|
| 7 |
-
"train_steps_per_second": 2.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 4.0,
|
| 3 |
"total_flos": 3.104468219559813e+18,
|
| 4 |
+
"train_loss": 0.06904298590738683,
|
| 5 |
+
"train_runtime": 834.8541,
|
| 6 |
+
"train_samples_per_second": 47.984,
|
| 7 |
+
"train_steps_per_second": 2.999
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 0.
|
| 3 |
"best_model_checkpoint": "./vit-base-skin/checkpoint-1252",
|
| 4 |
"epoch": 4.0,
|
| 5 |
"global_step": 2504,
|
|
@@ -10,1559 +10,1559 @@
|
|
| 10 |
{
|
| 11 |
"epoch": 0.02,
|
| 12 |
"learning_rate": 0.00019920127795527157,
|
| 13 |
-
"loss": 0.
|
| 14 |
"step": 10
|
| 15 |
},
|
| 16 |
{
|
| 17 |
"epoch": 0.03,
|
| 18 |
"learning_rate": 0.00019840255591054313,
|
| 19 |
-
"loss": 0.
|
| 20 |
"step": 20
|
| 21 |
},
|
| 22 |
{
|
| 23 |
"epoch": 0.05,
|
| 24 |
"learning_rate": 0.00019760383386581472,
|
| 25 |
-
"loss": 0.
|
| 26 |
"step": 30
|
| 27 |
},
|
| 28 |
{
|
| 29 |
"epoch": 0.06,
|
| 30 |
"learning_rate": 0.00019680511182108628,
|
| 31 |
-
"loss": 0.
|
| 32 |
"step": 40
|
| 33 |
},
|
| 34 |
{
|
| 35 |
"epoch": 0.08,
|
| 36 |
"learning_rate": 0.00019600638977635784,
|
| 37 |
-
"loss": 0.
|
| 38 |
"step": 50
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 0.1,
|
| 42 |
"learning_rate": 0.0001952076677316294,
|
| 43 |
-
"loss": 0.
|
| 44 |
"step": 60
|
| 45 |
},
|
| 46 |
{
|
| 47 |
"epoch": 0.11,
|
| 48 |
"learning_rate": 0.00019440894568690097,
|
| 49 |
-
"loss": 0.
|
| 50 |
"step": 70
|
| 51 |
},
|
| 52 |
{
|
| 53 |
"epoch": 0.13,
|
| 54 |
"learning_rate": 0.00019361022364217253,
|
| 55 |
-
"loss": 0.
|
| 56 |
"step": 80
|
| 57 |
},
|
| 58 |
{
|
| 59 |
"epoch": 0.14,
|
| 60 |
"learning_rate": 0.0001928115015974441,
|
| 61 |
-
"loss": 0.
|
| 62 |
"step": 90
|
| 63 |
},
|
| 64 |
{
|
| 65 |
"epoch": 0.16,
|
| 66 |
"learning_rate": 0.00019201277955271565,
|
| 67 |
-
"loss": 0.
|
| 68 |
"step": 100
|
| 69 |
},
|
| 70 |
{
|
| 71 |
"epoch": 0.18,
|
| 72 |
"learning_rate": 0.0001912140575079872,
|
| 73 |
-
"loss": 0.
|
| 74 |
"step": 110
|
| 75 |
},
|
| 76 |
{
|
| 77 |
"epoch": 0.19,
|
| 78 |
"learning_rate": 0.0001904153354632588,
|
| 79 |
-
"loss": 0.
|
| 80 |
"step": 120
|
| 81 |
},
|
| 82 |
{
|
| 83 |
"epoch": 0.21,
|
| 84 |
"learning_rate": 0.00018961661341853036,
|
| 85 |
-
"loss": 0.
|
| 86 |
"step": 130
|
| 87 |
},
|
| 88 |
{
|
| 89 |
"epoch": 0.22,
|
| 90 |
"learning_rate": 0.00018881789137380192,
|
| 91 |
-
"loss": 0.
|
| 92 |
"step": 140
|
| 93 |
},
|
| 94 |
{
|
| 95 |
"epoch": 0.24,
|
| 96 |
"learning_rate": 0.00018801916932907348,
|
| 97 |
-
"loss": 0.
|
| 98 |
"step": 150
|
| 99 |
},
|
| 100 |
{
|
| 101 |
"epoch": 0.26,
|
| 102 |
"learning_rate": 0.00018722044728434505,
|
| 103 |
-
"loss": 0.
|
| 104 |
"step": 160
|
| 105 |
},
|
| 106 |
{
|
| 107 |
"epoch": 0.27,
|
| 108 |
"learning_rate": 0.00018642172523961663,
|
| 109 |
-
"loss": 0.
|
| 110 |
"step": 170
|
| 111 |
},
|
| 112 |
{
|
| 113 |
"epoch": 0.29,
|
| 114 |
"learning_rate": 0.0001856230031948882,
|
| 115 |
-
"loss": 0.
|
| 116 |
"step": 180
|
| 117 |
},
|
| 118 |
{
|
| 119 |
"epoch": 0.3,
|
| 120 |
"learning_rate": 0.00018482428115015976,
|
| 121 |
-
"loss": 0.
|
| 122 |
"step": 190
|
| 123 |
},
|
| 124 |
{
|
| 125 |
"epoch": 0.32,
|
| 126 |
"learning_rate": 0.00018402555910543132,
|
| 127 |
-
"loss": 0.
|
| 128 |
"step": 200
|
| 129 |
},
|
| 130 |
{
|
| 131 |
"epoch": 0.34,
|
| 132 |
"learning_rate": 0.00018322683706070288,
|
| 133 |
-
"loss": 0.
|
| 134 |
"step": 210
|
| 135 |
},
|
| 136 |
{
|
| 137 |
"epoch": 0.35,
|
| 138 |
"learning_rate": 0.00018242811501597444,
|
| 139 |
-
"loss": 0.
|
| 140 |
"step": 220
|
| 141 |
},
|
| 142 |
{
|
| 143 |
"epoch": 0.37,
|
| 144 |
"learning_rate": 0.000181629392971246,
|
| 145 |
-
"loss": 0.
|
| 146 |
"step": 230
|
| 147 |
},
|
| 148 |
{
|
| 149 |
"epoch": 0.38,
|
| 150 |
"learning_rate": 0.00018083067092651756,
|
| 151 |
-
"loss": 0.
|
| 152 |
"step": 240
|
| 153 |
},
|
| 154 |
{
|
| 155 |
"epoch": 0.4,
|
| 156 |
"learning_rate": 0.00018003194888178913,
|
| 157 |
-
"loss": 0.
|
| 158 |
"step": 250
|
| 159 |
},
|
| 160 |
{
|
| 161 |
"epoch": 0.42,
|
| 162 |
"learning_rate": 0.00017923322683706071,
|
| 163 |
-
"loss": 0.
|
| 164 |
"step": 260
|
| 165 |
},
|
| 166 |
{
|
| 167 |
"epoch": 0.43,
|
| 168 |
"learning_rate": 0.00017843450479233228,
|
| 169 |
-
"loss": 0.
|
| 170 |
"step": 270
|
| 171 |
},
|
| 172 |
{
|
| 173 |
"epoch": 0.45,
|
| 174 |
"learning_rate": 0.00017763578274760384,
|
| 175 |
-
"loss": 0.
|
| 176 |
"step": 280
|
| 177 |
},
|
| 178 |
{
|
| 179 |
"epoch": 0.46,
|
| 180 |
"learning_rate": 0.0001768370607028754,
|
| 181 |
-
"loss": 0.
|
| 182 |
"step": 290
|
| 183 |
},
|
| 184 |
{
|
| 185 |
"epoch": 0.48,
|
| 186 |
"learning_rate": 0.000176038338658147,
|
| 187 |
-
"loss": 0.
|
| 188 |
"step": 300
|
| 189 |
},
|
| 190 |
{
|
| 191 |
"epoch": 0.5,
|
| 192 |
"learning_rate": 0.00017523961661341855,
|
| 193 |
-
"loss": 0.
|
| 194 |
"step": 310
|
| 195 |
},
|
| 196 |
{
|
| 197 |
"epoch": 0.51,
|
| 198 |
"learning_rate": 0.0001744408945686901,
|
| 199 |
-
"loss": 0.
|
| 200 |
"step": 320
|
| 201 |
},
|
| 202 |
{
|
| 203 |
"epoch": 0.53,
|
| 204 |
"learning_rate": 0.00017364217252396167,
|
| 205 |
-
"loss": 0.
|
| 206 |
"step": 330
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 0.54,
|
| 210 |
"learning_rate": 0.00017284345047923323,
|
| 211 |
-
"loss": 0.
|
| 212 |
"step": 340
|
| 213 |
},
|
| 214 |
{
|
| 215 |
"epoch": 0.56,
|
| 216 |
"learning_rate": 0.00017204472843450482,
|
| 217 |
-
"loss": 0.
|
| 218 |
"step": 350
|
| 219 |
},
|
| 220 |
{
|
| 221 |
"epoch": 0.58,
|
| 222 |
"learning_rate": 0.00017124600638977638,
|
| 223 |
-
"loss": 0.
|
| 224 |
"step": 360
|
| 225 |
},
|
| 226 |
{
|
| 227 |
"epoch": 0.59,
|
| 228 |
"learning_rate": 0.00017044728434504792,
|
| 229 |
-
"loss": 0.
|
| 230 |
"step": 370
|
| 231 |
},
|
| 232 |
{
|
| 233 |
"epoch": 0.61,
|
| 234 |
"learning_rate": 0.00016964856230031948,
|
| 235 |
-
"loss": 0.
|
| 236 |
"step": 380
|
| 237 |
},
|
| 238 |
{
|
| 239 |
"epoch": 0.62,
|
| 240 |
"learning_rate": 0.00016884984025559104,
|
| 241 |
-
"loss": 0.
|
| 242 |
"step": 390
|
| 243 |
},
|
| 244 |
{
|
| 245 |
"epoch": 0.64,
|
| 246 |
"learning_rate": 0.00016805111821086263,
|
| 247 |
-
"loss": 0.
|
| 248 |
"step": 400
|
| 249 |
},
|
| 250 |
{
|
| 251 |
"epoch": 0.65,
|
| 252 |
"learning_rate": 0.0001672523961661342,
|
| 253 |
-
"loss": 0.
|
| 254 |
"step": 410
|
| 255 |
},
|
| 256 |
{
|
| 257 |
"epoch": 0.67,
|
| 258 |
"learning_rate": 0.00016645367412140575,
|
| 259 |
-
"loss": 0.
|
| 260 |
"step": 420
|
| 261 |
},
|
| 262 |
{
|
| 263 |
"epoch": 0.69,
|
| 264 |
"learning_rate": 0.0001656549520766773,
|
| 265 |
-
"loss": 0.
|
| 266 |
"step": 430
|
| 267 |
},
|
| 268 |
{
|
| 269 |
"epoch": 0.7,
|
| 270 |
"learning_rate": 0.0001648562300319489,
|
| 271 |
-
"loss": 0.
|
| 272 |
"step": 440
|
| 273 |
},
|
| 274 |
{
|
| 275 |
"epoch": 0.72,
|
| 276 |
"learning_rate": 0.00016405750798722046,
|
| 277 |
-
"loss": 0.
|
| 278 |
"step": 450
|
| 279 |
},
|
| 280 |
{
|
| 281 |
"epoch": 0.73,
|
| 282 |
"learning_rate": 0.00016325878594249202,
|
| 283 |
-
"loss": 0.
|
| 284 |
"step": 460
|
| 285 |
},
|
| 286 |
{
|
| 287 |
"epoch": 0.75,
|
| 288 |
"learning_rate": 0.00016246006389776359,
|
| 289 |
-
"loss": 0.
|
| 290 |
"step": 470
|
| 291 |
},
|
| 292 |
{
|
| 293 |
"epoch": 0.77,
|
| 294 |
"learning_rate": 0.00016166134185303515,
|
| 295 |
-
"loss": 0.
|
| 296 |
"step": 480
|
| 297 |
},
|
| 298 |
{
|
| 299 |
"epoch": 0.78,
|
| 300 |
"learning_rate": 0.00016086261980830674,
|
| 301 |
-
"loss": 0.
|
| 302 |
"step": 490
|
| 303 |
},
|
| 304 |
{
|
| 305 |
"epoch": 0.8,
|
| 306 |
"learning_rate": 0.0001600638977635783,
|
| 307 |
-
"loss": 0.
|
| 308 |
"step": 500
|
| 309 |
},
|
| 310 |
{
|
| 311 |
"epoch": 0.81,
|
| 312 |
"learning_rate": 0.00015926517571884986,
|
| 313 |
-
"loss": 0.
|
| 314 |
"step": 510
|
| 315 |
},
|
| 316 |
{
|
| 317 |
"epoch": 0.83,
|
| 318 |
"learning_rate": 0.00015846645367412142,
|
| 319 |
-
"loss": 0.
|
| 320 |
"step": 520
|
| 321 |
},
|
| 322 |
{
|
| 323 |
"epoch": 0.85,
|
| 324 |
"learning_rate": 0.00015766773162939298,
|
| 325 |
-
"loss": 0.
|
| 326 |
"step": 530
|
| 327 |
},
|
| 328 |
{
|
| 329 |
"epoch": 0.86,
|
| 330 |
"learning_rate": 0.00015686900958466454,
|
| 331 |
-
"loss": 0.
|
| 332 |
"step": 540
|
| 333 |
},
|
| 334 |
{
|
| 335 |
"epoch": 0.88,
|
| 336 |
"learning_rate": 0.0001560702875399361,
|
| 337 |
-
"loss": 0.
|
| 338 |
"step": 550
|
| 339 |
},
|
| 340 |
{
|
| 341 |
"epoch": 0.89,
|
| 342 |
"learning_rate": 0.00015527156549520767,
|
| 343 |
-
"loss": 0.
|
| 344 |
"step": 560
|
| 345 |
},
|
| 346 |
{
|
| 347 |
"epoch": 0.91,
|
| 348 |
"learning_rate": 0.00015447284345047923,
|
| 349 |
-
"loss": 0.
|
| 350 |
"step": 570
|
| 351 |
},
|
| 352 |
{
|
| 353 |
"epoch": 0.93,
|
| 354 |
"learning_rate": 0.00015367412140575082,
|
| 355 |
-
"loss": 0.
|
| 356 |
"step": 580
|
| 357 |
},
|
| 358 |
{
|
| 359 |
"epoch": 0.94,
|
| 360 |
"learning_rate": 0.00015287539936102238,
|
| 361 |
-
"loss": 0.
|
| 362 |
"step": 590
|
| 363 |
},
|
| 364 |
{
|
| 365 |
"epoch": 0.96,
|
| 366 |
"learning_rate": 0.00015207667731629394,
|
| 367 |
-
"loss": 0.
|
| 368 |
"step": 600
|
| 369 |
},
|
| 370 |
{
|
| 371 |
"epoch": 0.97,
|
| 372 |
"learning_rate": 0.0001512779552715655,
|
| 373 |
-
"loss": 0.
|
| 374 |
"step": 610
|
| 375 |
},
|
| 376 |
{
|
| 377 |
"epoch": 0.99,
|
| 378 |
"learning_rate": 0.00015047923322683706,
|
| 379 |
-
"loss": 0.
|
| 380 |
"step": 620
|
| 381 |
},
|
| 382 |
{
|
| 383 |
"epoch": 1.0,
|
| 384 |
-
"eval_accuracy": 0.
|
| 385 |
-
"eval_f1": 0.
|
| 386 |
-
"eval_loss": 0.
|
| 387 |
-
"eval_precision": 0.
|
| 388 |
-
"eval_recall": 0.
|
| 389 |
-
"eval_runtime": 1.
|
| 390 |
-
"eval_samples_per_second":
|
| 391 |
-
"eval_steps_per_second":
|
| 392 |
"step": 626
|
| 393 |
},
|
| 394 |
{
|
| 395 |
"epoch": 1.01,
|
| 396 |
"learning_rate": 0.00014968051118210865,
|
| 397 |
-
"loss": 0.
|
| 398 |
"step": 630
|
| 399 |
},
|
| 400 |
{
|
| 401 |
"epoch": 1.02,
|
| 402 |
"learning_rate": 0.0001488817891373802,
|
| 403 |
-
"loss": 0.
|
| 404 |
"step": 640
|
| 405 |
},
|
| 406 |
{
|
| 407 |
"epoch": 1.04,
|
| 408 |
"learning_rate": 0.00014808306709265177,
|
| 409 |
-
"loss": 0.
|
| 410 |
"step": 650
|
| 411 |
},
|
| 412 |
{
|
| 413 |
"epoch": 1.05,
|
| 414 |
"learning_rate": 0.00014728434504792333,
|
| 415 |
-
"loss": 0.
|
| 416 |
"step": 660
|
| 417 |
},
|
| 418 |
{
|
| 419 |
"epoch": 1.07,
|
| 420 |
"learning_rate": 0.0001464856230031949,
|
| 421 |
-
"loss": 0.
|
| 422 |
"step": 670
|
| 423 |
},
|
| 424 |
{
|
| 425 |
"epoch": 1.09,
|
| 426 |
"learning_rate": 0.00014568690095846646,
|
| 427 |
-
"loss": 0.
|
| 428 |
"step": 680
|
| 429 |
},
|
| 430 |
{
|
| 431 |
"epoch": 1.1,
|
| 432 |
"learning_rate": 0.00014488817891373802,
|
| 433 |
-
"loss": 0.
|
| 434 |
"step": 690
|
| 435 |
},
|
| 436 |
{
|
| 437 |
"epoch": 1.12,
|
| 438 |
"learning_rate": 0.00014408945686900958,
|
| 439 |
-
"loss": 0.
|
| 440 |
"step": 700
|
| 441 |
},
|
| 442 |
{
|
| 443 |
"epoch": 1.13,
|
| 444 |
"learning_rate": 0.00014329073482428114,
|
| 445 |
-
"loss": 0.
|
| 446 |
"step": 710
|
| 447 |
},
|
| 448 |
{
|
| 449 |
"epoch": 1.15,
|
| 450 |
"learning_rate": 0.00014249201277955273,
|
| 451 |
-
"loss": 0.
|
| 452 |
"step": 720
|
| 453 |
},
|
| 454 |
{
|
| 455 |
"epoch": 1.17,
|
| 456 |
"learning_rate": 0.0001416932907348243,
|
| 457 |
-
"loss": 0.
|
| 458 |
"step": 730
|
| 459 |
},
|
| 460 |
{
|
| 461 |
"epoch": 1.18,
|
| 462 |
"learning_rate": 0.00014089456869009585,
|
| 463 |
-
"loss": 0.
|
| 464 |
"step": 740
|
| 465 |
},
|
| 466 |
{
|
| 467 |
"epoch": 1.2,
|
| 468 |
"learning_rate": 0.0001400958466453674,
|
| 469 |
-
"loss": 0.
|
| 470 |
"step": 750
|
| 471 |
},
|
| 472 |
{
|
| 473 |
"epoch": 1.21,
|
| 474 |
"learning_rate": 0.000139297124600639,
|
| 475 |
-
"loss": 0.
|
| 476 |
"step": 760
|
| 477 |
},
|
| 478 |
{
|
| 479 |
"epoch": 1.23,
|
| 480 |
"learning_rate": 0.00013849840255591056,
|
| 481 |
-
"loss": 0.
|
| 482 |
"step": 770
|
| 483 |
},
|
| 484 |
{
|
| 485 |
"epoch": 1.25,
|
| 486 |
"learning_rate": 0.00013769968051118212,
|
| 487 |
-
"loss": 0.
|
| 488 |
"step": 780
|
| 489 |
},
|
| 490 |
{
|
| 491 |
"epoch": 1.26,
|
| 492 |
"learning_rate": 0.00013690095846645369,
|
| 493 |
-
"loss": 0.
|
| 494 |
"step": 790
|
| 495 |
},
|
| 496 |
{
|
| 497 |
"epoch": 1.28,
|
| 498 |
"learning_rate": 0.00013610223642172525,
|
| 499 |
-
"loss": 0.
|
| 500 |
"step": 800
|
| 501 |
},
|
| 502 |
{
|
| 503 |
"epoch": 1.29,
|
| 504 |
"learning_rate": 0.0001353035143769968,
|
| 505 |
-
"loss": 0.
|
| 506 |
"step": 810
|
| 507 |
},
|
| 508 |
{
|
| 509 |
"epoch": 1.31,
|
| 510 |
"learning_rate": 0.00013450479233226837,
|
| 511 |
-
"loss": 0.
|
| 512 |
"step": 820
|
| 513 |
},
|
| 514 |
{
|
| 515 |
"epoch": 1.33,
|
| 516 |
"learning_rate": 0.00013370607028753993,
|
| 517 |
-
"loss": 0.
|
| 518 |
"step": 830
|
| 519 |
},
|
| 520 |
{
|
| 521 |
"epoch": 1.34,
|
| 522 |
"learning_rate": 0.0001329073482428115,
|
| 523 |
-
"loss": 0.
|
| 524 |
"step": 840
|
| 525 |
},
|
| 526 |
{
|
| 527 |
"epoch": 1.36,
|
| 528 |
"learning_rate": 0.00013210862619808305,
|
| 529 |
-
"loss": 0.
|
| 530 |
"step": 850
|
| 531 |
},
|
| 532 |
{
|
| 533 |
"epoch": 1.37,
|
| 534 |
"learning_rate": 0.00013130990415335464,
|
| 535 |
-
"loss": 0.
|
| 536 |
"step": 860
|
| 537 |
},
|
| 538 |
{
|
| 539 |
"epoch": 1.39,
|
| 540 |
"learning_rate": 0.0001305111821086262,
|
| 541 |
-
"loss": 0.
|
| 542 |
"step": 870
|
| 543 |
},
|
| 544 |
{
|
| 545 |
"epoch": 1.41,
|
| 546 |
"learning_rate": 0.00012971246006389777,
|
| 547 |
-
"loss": 0.
|
| 548 |
"step": 880
|
| 549 |
},
|
| 550 |
{
|
| 551 |
"epoch": 1.42,
|
| 552 |
"learning_rate": 0.00012891373801916933,
|
| 553 |
-
"loss": 0.
|
| 554 |
"step": 890
|
| 555 |
},
|
| 556 |
{
|
| 557 |
"epoch": 1.44,
|
| 558 |
"learning_rate": 0.00012811501597444092,
|
| 559 |
-
"loss": 0.
|
| 560 |
"step": 900
|
| 561 |
},
|
| 562 |
{
|
| 563 |
"epoch": 1.45,
|
| 564 |
"learning_rate": 0.00012731629392971248,
|
| 565 |
-
"loss": 0.
|
| 566 |
"step": 910
|
| 567 |
},
|
| 568 |
{
|
| 569 |
"epoch": 1.47,
|
| 570 |
"learning_rate": 0.00012651757188498404,
|
| 571 |
-
"loss": 0.
|
| 572 |
"step": 920
|
| 573 |
},
|
| 574 |
{
|
| 575 |
"epoch": 1.49,
|
| 576 |
"learning_rate": 0.0001257188498402556,
|
| 577 |
-
"loss": 0.
|
| 578 |
"step": 930
|
| 579 |
},
|
| 580 |
{
|
| 581 |
"epoch": 1.5,
|
| 582 |
"learning_rate": 0.00012492012779552716,
|
| 583 |
-
"loss": 0.
|
| 584 |
"step": 940
|
| 585 |
},
|
| 586 |
{
|
| 587 |
"epoch": 1.52,
|
| 588 |
"learning_rate": 0.00012412140575079872,
|
| 589 |
-
"loss": 0.
|
| 590 |
"step": 950
|
| 591 |
},
|
| 592 |
{
|
| 593 |
"epoch": 1.53,
|
| 594 |
"learning_rate": 0.00012332268370607028,
|
| 595 |
-
"loss": 0.
|
| 596 |
"step": 960
|
| 597 |
},
|
| 598 |
{
|
| 599 |
"epoch": 1.55,
|
| 600 |
"learning_rate": 0.00012252396166134185,
|
| 601 |
-
"loss": 0.
|
| 602 |
"step": 970
|
| 603 |
},
|
| 604 |
{
|
| 605 |
"epoch": 1.57,
|
| 606 |
"learning_rate": 0.00012172523961661342,
|
| 607 |
-
"loss": 0.
|
| 608 |
"step": 980
|
| 609 |
},
|
| 610 |
{
|
| 611 |
"epoch": 1.58,
|
| 612 |
"learning_rate": 0.000120926517571885,
|
| 613 |
-
"loss": 0.
|
| 614 |
"step": 990
|
| 615 |
},
|
| 616 |
{
|
| 617 |
"epoch": 1.6,
|
| 618 |
"learning_rate": 0.00012012779552715656,
|
| 619 |
-
"loss": 0.
|
| 620 |
"step": 1000
|
| 621 |
},
|
| 622 |
{
|
| 623 |
"epoch": 1.61,
|
| 624 |
"learning_rate": 0.00011932907348242812,
|
| 625 |
-
"loss": 0.
|
| 626 |
"step": 1010
|
| 627 |
},
|
| 628 |
{
|
| 629 |
"epoch": 1.63,
|
| 630 |
"learning_rate": 0.00011853035143769968,
|
| 631 |
-
"loss": 0.
|
| 632 |
"step": 1020
|
| 633 |
},
|
| 634 |
{
|
| 635 |
"epoch": 1.65,
|
| 636 |
"learning_rate": 0.00011773162939297124,
|
| 637 |
-
"loss": 0.
|
| 638 |
"step": 1030
|
| 639 |
},
|
| 640 |
{
|
| 641 |
"epoch": 1.66,
|
| 642 |
"learning_rate": 0.00011693290734824283,
|
| 643 |
-
"loss": 0.
|
| 644 |
"step": 1040
|
| 645 |
},
|
| 646 |
{
|
| 647 |
"epoch": 1.68,
|
| 648 |
"learning_rate": 0.00011613418530351439,
|
| 649 |
-
"loss": 0.
|
| 650 |
"step": 1050
|
| 651 |
},
|
| 652 |
{
|
| 653 |
"epoch": 1.69,
|
| 654 |
"learning_rate": 0.00011533546325878595,
|
| 655 |
-
"loss": 0.
|
| 656 |
"step": 1060
|
| 657 |
},
|
| 658 |
{
|
| 659 |
"epoch": 1.71,
|
| 660 |
"learning_rate": 0.00011453674121405751,
|
| 661 |
-
"loss": 0.
|
| 662 |
"step": 1070
|
| 663 |
},
|
| 664 |
{
|
| 665 |
"epoch": 1.73,
|
| 666 |
"learning_rate": 0.00011373801916932908,
|
| 667 |
-
"loss": 0.
|
| 668 |
"step": 1080
|
| 669 |
},
|
| 670 |
{
|
| 671 |
"epoch": 1.74,
|
| 672 |
"learning_rate": 0.00011293929712460065,
|
| 673 |
-
"loss": 0.
|
| 674 |
"step": 1090
|
| 675 |
},
|
| 676 |
{
|
| 677 |
"epoch": 1.76,
|
| 678 |
"learning_rate": 0.00011214057507987221,
|
| 679 |
-
"loss": 0.
|
| 680 |
"step": 1100
|
| 681 |
},
|
| 682 |
{
|
| 683 |
"epoch": 1.77,
|
| 684 |
"learning_rate": 0.00011134185303514377,
|
| 685 |
-
"loss": 0.
|
| 686 |
"step": 1110
|
| 687 |
},
|
| 688 |
{
|
| 689 |
"epoch": 1.79,
|
| 690 |
"learning_rate": 0.00011054313099041533,
|
| 691 |
-
"loss": 0.
|
| 692 |
"step": 1120
|
| 693 |
},
|
| 694 |
{
|
| 695 |
"epoch": 1.81,
|
| 696 |
"learning_rate": 0.00010974440894568691,
|
| 697 |
-
"loss": 0.
|
| 698 |
"step": 1130
|
| 699 |
},
|
| 700 |
{
|
| 701 |
"epoch": 1.82,
|
| 702 |
"learning_rate": 0.00010894568690095847,
|
| 703 |
-
"loss": 0.
|
| 704 |
"step": 1140
|
| 705 |
},
|
| 706 |
{
|
| 707 |
"epoch": 1.84,
|
| 708 |
"learning_rate": 0.00010814696485623003,
|
| 709 |
-
"loss": 0.
|
| 710 |
"step": 1150
|
| 711 |
},
|
| 712 |
{
|
| 713 |
"epoch": 1.85,
|
| 714 |
"learning_rate": 0.0001073482428115016,
|
| 715 |
-
"loss": 0.
|
| 716 |
"step": 1160
|
| 717 |
},
|
| 718 |
{
|
| 719 |
"epoch": 1.87,
|
| 720 |
"learning_rate": 0.00010654952076677316,
|
| 721 |
-
"loss": 0.
|
| 722 |
"step": 1170
|
| 723 |
},
|
| 724 |
{
|
| 725 |
"epoch": 1.88,
|
| 726 |
"learning_rate": 0.00010575079872204474,
|
| 727 |
-
"loss": 0.
|
| 728 |
"step": 1180
|
| 729 |
},
|
| 730 |
{
|
| 731 |
"epoch": 1.9,
|
| 732 |
"learning_rate": 0.0001049520766773163,
|
| 733 |
-
"loss": 0.
|
| 734 |
"step": 1190
|
| 735 |
},
|
| 736 |
{
|
| 737 |
"epoch": 1.92,
|
| 738 |
"learning_rate": 0.00010415335463258787,
|
| 739 |
-
"loss": 0.
|
| 740 |
"step": 1200
|
| 741 |
},
|
| 742 |
{
|
| 743 |
"epoch": 1.93,
|
| 744 |
"learning_rate": 0.00010335463258785943,
|
| 745 |
-
"loss": 0.
|
| 746 |
"step": 1210
|
| 747 |
},
|
| 748 |
{
|
| 749 |
"epoch": 1.95,
|
| 750 |
"learning_rate": 0.000102555910543131,
|
| 751 |
-
"loss": 0.
|
| 752 |
"step": 1220
|
| 753 |
},
|
| 754 |
{
|
| 755 |
"epoch": 1.96,
|
| 756 |
"learning_rate": 0.00010175718849840256,
|
| 757 |
-
"loss": 0.
|
| 758 |
"step": 1230
|
| 759 |
},
|
| 760 |
{
|
| 761 |
"epoch": 1.98,
|
| 762 |
"learning_rate": 0.00010095846645367413,
|
| 763 |
-
"loss": 0.
|
| 764 |
"step": 1240
|
| 765 |
},
|
| 766 |
{
|
| 767 |
"epoch": 2.0,
|
| 768 |
"learning_rate": 0.00010015974440894569,
|
| 769 |
-
"loss": 0.
|
| 770 |
"step": 1250
|
| 771 |
},
|
| 772 |
{
|
| 773 |
"epoch": 2.0,
|
| 774 |
-
"eval_accuracy": 0.
|
| 775 |
-
"eval_f1": 0.
|
| 776 |
-
"eval_loss": 0.
|
| 777 |
-
"eval_precision": 0.
|
| 778 |
-
"eval_recall": 0.
|
| 779 |
-
"eval_runtime": 1.
|
| 780 |
-
"eval_samples_per_second":
|
| 781 |
-
"eval_steps_per_second":
|
| 782 |
"step": 1252
|
| 783 |
},
|
| 784 |
{
|
| 785 |
"epoch": 2.01,
|
| 786 |
"learning_rate": 9.936102236421726e-05,
|
| 787 |
-
"loss": 0.
|
| 788 |
"step": 1260
|
| 789 |
},
|
| 790 |
{
|
| 791 |
"epoch": 2.03,
|
| 792 |
"learning_rate": 9.856230031948882e-05,
|
| 793 |
-
"loss": 0.
|
| 794 |
"step": 1270
|
| 795 |
},
|
| 796 |
{
|
| 797 |
"epoch": 2.04,
|
| 798 |
"learning_rate": 9.77635782747604e-05,
|
| 799 |
-
"loss": 0.
|
| 800 |
"step": 1280
|
| 801 |
},
|
| 802 |
{
|
| 803 |
"epoch": 2.06,
|
| 804 |
"learning_rate": 9.696485623003195e-05,
|
| 805 |
-
"loss": 0.
|
| 806 |
"step": 1290
|
| 807 |
},
|
| 808 |
{
|
| 809 |
"epoch": 2.08,
|
| 810 |
"learning_rate": 9.616613418530351e-05,
|
| 811 |
-
"loss": 0.
|
| 812 |
"step": 1300
|
| 813 |
},
|
| 814 |
{
|
| 815 |
"epoch": 2.09,
|
| 816 |
"learning_rate": 9.536741214057508e-05,
|
| 817 |
-
"loss": 0.
|
| 818 |
"step": 1310
|
| 819 |
},
|
| 820 |
{
|
| 821 |
"epoch": 2.11,
|
| 822 |
"learning_rate": 9.456869009584664e-05,
|
| 823 |
-
"loss": 0.
|
| 824 |
"step": 1320
|
| 825 |
},
|
| 826 |
{
|
| 827 |
"epoch": 2.12,
|
| 828 |
"learning_rate": 9.376996805111822e-05,
|
| 829 |
-
"loss": 0.
|
| 830 |
"step": 1330
|
| 831 |
},
|
| 832 |
{
|
| 833 |
"epoch": 2.14,
|
| 834 |
"learning_rate": 9.297124600638978e-05,
|
| 835 |
-
"loss": 0.
|
| 836 |
"step": 1340
|
| 837 |
},
|
| 838 |
{
|
| 839 |
"epoch": 2.16,
|
| 840 |
"learning_rate": 9.217252396166136e-05,
|
| 841 |
-
"loss": 0.
|
| 842 |
"step": 1350
|
| 843 |
},
|
| 844 |
{
|
| 845 |
"epoch": 2.17,
|
| 846 |
"learning_rate": 9.137380191693292e-05,
|
| 847 |
-
"loss": 0.
|
| 848 |
"step": 1360
|
| 849 |
},
|
| 850 |
{
|
| 851 |
"epoch": 2.19,
|
| 852 |
"learning_rate": 9.057507987220448e-05,
|
| 853 |
-
"loss": 0.
|
| 854 |
"step": 1370
|
| 855 |
},
|
| 856 |
{
|
| 857 |
"epoch": 2.2,
|
| 858 |
"learning_rate": 8.977635782747604e-05,
|
| 859 |
-
"loss": 0.
|
| 860 |
"step": 1380
|
| 861 |
},
|
| 862 |
{
|
| 863 |
"epoch": 2.22,
|
| 864 |
"learning_rate": 8.89776357827476e-05,
|
| 865 |
-
"loss": 0.
|
| 866 |
"step": 1390
|
| 867 |
},
|
| 868 |
{
|
| 869 |
"epoch": 2.24,
|
| 870 |
"learning_rate": 8.817891373801918e-05,
|
| 871 |
-
"loss": 0.
|
| 872 |
"step": 1400
|
| 873 |
},
|
| 874 |
{
|
| 875 |
"epoch": 2.25,
|
| 876 |
"learning_rate": 8.738019169329074e-05,
|
| 877 |
-
"loss": 0.
|
| 878 |
"step": 1410
|
| 879 |
},
|
| 880 |
{
|
| 881 |
"epoch": 2.27,
|
| 882 |
"learning_rate": 8.658146964856231e-05,
|
| 883 |
-
"loss": 0.
|
| 884 |
"step": 1420
|
| 885 |
},
|
| 886 |
{
|
| 887 |
"epoch": 2.28,
|
| 888 |
"learning_rate": 8.578274760383387e-05,
|
| 889 |
-
"loss": 0.
|
| 890 |
"step": 1430
|
| 891 |
},
|
| 892 |
{
|
| 893 |
"epoch": 2.3,
|
| 894 |
"learning_rate": 8.498402555910544e-05,
|
| 895 |
-
"loss": 0.
|
| 896 |
"step": 1440
|
| 897 |
},
|
| 898 |
{
|
| 899 |
"epoch": 2.32,
|
| 900 |
"learning_rate": 8.4185303514377e-05,
|
| 901 |
-
"loss": 0.
|
| 902 |
"step": 1450
|
| 903 |
},
|
| 904 |
{
|
| 905 |
"epoch": 2.33,
|
| 906 |
"learning_rate": 8.338658146964856e-05,
|
| 907 |
-
"loss": 0.
|
| 908 |
"step": 1460
|
| 909 |
},
|
| 910 |
{
|
| 911 |
"epoch": 2.35,
|
| 912 |
"learning_rate": 8.258785942492013e-05,
|
| 913 |
-
"loss": 0.
|
| 914 |
"step": 1470
|
| 915 |
},
|
| 916 |
{
|
| 917 |
"epoch": 2.36,
|
| 918 |
"learning_rate": 8.17891373801917e-05,
|
| 919 |
-
"loss": 0.
|
| 920 |
"step": 1480
|
| 921 |
},
|
| 922 |
{
|
| 923 |
"epoch": 2.38,
|
| 924 |
"learning_rate": 8.099041533546327e-05,
|
| 925 |
-
"loss": 0.
|
| 926 |
"step": 1490
|
| 927 |
},
|
| 928 |
{
|
| 929 |
"epoch": 2.4,
|
| 930 |
"learning_rate": 8.019169329073483e-05,
|
| 931 |
-
"loss": 0.
|
| 932 |
"step": 1500
|
| 933 |
},
|
| 934 |
{
|
| 935 |
"epoch": 2.41,
|
| 936 |
"learning_rate": 7.939297124600639e-05,
|
| 937 |
-
"loss": 0.
|
| 938 |
"step": 1510
|
| 939 |
},
|
| 940 |
{
|
| 941 |
"epoch": 2.43,
|
| 942 |
"learning_rate": 7.859424920127795e-05,
|
| 943 |
-
"loss": 0.
|
| 944 |
"step": 1520
|
| 945 |
},
|
| 946 |
{
|
| 947 |
"epoch": 2.44,
|
| 948 |
"learning_rate": 7.779552715654951e-05,
|
| 949 |
-
"loss": 0.
|
| 950 |
"step": 1530
|
| 951 |
},
|
| 952 |
{
|
| 953 |
"epoch": 2.46,
|
| 954 |
"learning_rate": 7.699680511182109e-05,
|
| 955 |
-
"loss": 0.
|
| 956 |
"step": 1540
|
| 957 |
},
|
| 958 |
{
|
| 959 |
"epoch": 2.48,
|
| 960 |
"learning_rate": 7.619808306709265e-05,
|
| 961 |
-
"loss": 0.
|
| 962 |
"step": 1550
|
| 963 |
},
|
| 964 |
{
|
| 965 |
"epoch": 2.49,
|
| 966 |
"learning_rate": 7.539936102236423e-05,
|
| 967 |
-
"loss": 0.
|
| 968 |
"step": 1560
|
| 969 |
},
|
| 970 |
{
|
| 971 |
"epoch": 2.51,
|
| 972 |
"learning_rate": 7.460063897763579e-05,
|
| 973 |
-
"loss": 0.
|
| 974 |
"step": 1570
|
| 975 |
},
|
| 976 |
{
|
| 977 |
"epoch": 2.52,
|
| 978 |
"learning_rate": 7.380191693290735e-05,
|
| 979 |
-
"loss": 0.
|
| 980 |
"step": 1580
|
| 981 |
},
|
| 982 |
{
|
| 983 |
"epoch": 2.54,
|
| 984 |
"learning_rate": 7.300319488817891e-05,
|
| 985 |
-
"loss": 0.
|
| 986 |
"step": 1590
|
| 987 |
},
|
| 988 |
{
|
| 989 |
"epoch": 2.56,
|
| 990 |
"learning_rate": 7.220447284345049e-05,
|
| 991 |
-
"loss": 0.
|
| 992 |
"step": 1600
|
| 993 |
},
|
| 994 |
{
|
| 995 |
"epoch": 2.57,
|
| 996 |
"learning_rate": 7.140575079872205e-05,
|
| 997 |
-
"loss": 0.
|
| 998 |
"step": 1610
|
| 999 |
},
|
| 1000 |
{
|
| 1001 |
"epoch": 2.59,
|
| 1002 |
"learning_rate": 7.060702875399361e-05,
|
| 1003 |
-
"loss": 0.
|
| 1004 |
"step": 1620
|
| 1005 |
},
|
| 1006 |
{
|
| 1007 |
"epoch": 2.6,
|
| 1008 |
"learning_rate": 6.980830670926518e-05,
|
| 1009 |
-
"loss": 0.
|
| 1010 |
"step": 1630
|
| 1011 |
},
|
| 1012 |
{
|
| 1013 |
"epoch": 2.62,
|
| 1014 |
"learning_rate": 6.900958466453674e-05,
|
| 1015 |
-
"loss": 0.
|
| 1016 |
"step": 1640
|
| 1017 |
},
|
| 1018 |
{
|
| 1019 |
"epoch": 2.64,
|
| 1020 |
"learning_rate": 6.821086261980832e-05,
|
| 1021 |
-
"loss": 0.
|
| 1022 |
"step": 1650
|
| 1023 |
},
|
| 1024 |
{
|
| 1025 |
"epoch": 2.65,
|
| 1026 |
"learning_rate": 6.741214057507987e-05,
|
| 1027 |
-
"loss": 0.
|
| 1028 |
"step": 1660
|
| 1029 |
},
|
| 1030 |
{
|
| 1031 |
"epoch": 2.67,
|
| 1032 |
"learning_rate": 6.661341853035144e-05,
|
| 1033 |
-
"loss": 0.
|
| 1034 |
"step": 1670
|
| 1035 |
},
|
| 1036 |
{
|
| 1037 |
"epoch": 2.68,
|
| 1038 |
"learning_rate": 6.5814696485623e-05,
|
| 1039 |
-
"loss": 0.
|
| 1040 |
"step": 1680
|
| 1041 |
},
|
| 1042 |
{
|
| 1043 |
"epoch": 2.7,
|
| 1044 |
"learning_rate": 6.501597444089457e-05,
|
| 1045 |
-
"loss": 0.
|
| 1046 |
"step": 1690
|
| 1047 |
},
|
| 1048 |
{
|
| 1049 |
"epoch": 2.72,
|
| 1050 |
"learning_rate": 6.421725239616614e-05,
|
| 1051 |
-
"loss": 0.
|
| 1052 |
"step": 1700
|
| 1053 |
},
|
| 1054 |
{
|
| 1055 |
"epoch": 2.73,
|
| 1056 |
"learning_rate": 6.34185303514377e-05,
|
| 1057 |
-
"loss": 0.
|
| 1058 |
"step": 1710
|
| 1059 |
},
|
| 1060 |
{
|
| 1061 |
"epoch": 2.75,
|
| 1062 |
"learning_rate": 6.261980830670928e-05,
|
| 1063 |
-
"loss": 0.
|
| 1064 |
"step": 1720
|
| 1065 |
},
|
| 1066 |
{
|
| 1067 |
"epoch": 2.76,
|
| 1068 |
"learning_rate": 6.182108626198084e-05,
|
| 1069 |
-
"loss": 0.
|
| 1070 |
"step": 1730
|
| 1071 |
},
|
| 1072 |
{
|
| 1073 |
"epoch": 2.78,
|
| 1074 |
"learning_rate": 6.1022364217252406e-05,
|
| 1075 |
-
"loss": 0.
|
| 1076 |
"step": 1740
|
| 1077 |
},
|
| 1078 |
{
|
| 1079 |
"epoch": 2.8,
|
| 1080 |
"learning_rate": 6.022364217252396e-05,
|
| 1081 |
-
"loss": 0.
|
| 1082 |
"step": 1750
|
| 1083 |
},
|
| 1084 |
{
|
| 1085 |
"epoch": 2.81,
|
| 1086 |
"learning_rate": 5.942492012779552e-05,
|
| 1087 |
-
"loss": 0.
|
| 1088 |
"step": 1760
|
| 1089 |
},
|
| 1090 |
{
|
| 1091 |
"epoch": 2.83,
|
| 1092 |
"learning_rate": 5.86261980830671e-05,
|
| 1093 |
-
"loss": 0.
|
| 1094 |
"step": 1770
|
| 1095 |
},
|
| 1096 |
{
|
| 1097 |
"epoch": 2.84,
|
| 1098 |
"learning_rate": 5.782747603833866e-05,
|
| 1099 |
-
"loss": 0.
|
| 1100 |
"step": 1780
|
| 1101 |
},
|
| 1102 |
{
|
| 1103 |
"epoch": 2.86,
|
| 1104 |
"learning_rate": 5.702875399361023e-05,
|
| 1105 |
-
"loss": 0.
|
| 1106 |
"step": 1790
|
| 1107 |
},
|
| 1108 |
{
|
| 1109 |
"epoch": 2.88,
|
| 1110 |
"learning_rate": 5.623003194888179e-05,
|
| 1111 |
-
"loss": 0.
|
| 1112 |
"step": 1800
|
| 1113 |
},
|
| 1114 |
{
|
| 1115 |
"epoch": 2.89,
|
| 1116 |
"learning_rate": 5.543130990415336e-05,
|
| 1117 |
-
"loss": 0.
|
| 1118 |
"step": 1810
|
| 1119 |
},
|
| 1120 |
{
|
| 1121 |
"epoch": 2.91,
|
| 1122 |
"learning_rate": 5.4632587859424925e-05,
|
| 1123 |
-
"loss": 0.
|
| 1124 |
"step": 1820
|
| 1125 |
},
|
| 1126 |
{
|
| 1127 |
"epoch": 2.92,
|
| 1128 |
"learning_rate": 5.383386581469649e-05,
|
| 1129 |
-
"loss": 0.
|
| 1130 |
"step": 1830
|
| 1131 |
},
|
| 1132 |
{
|
| 1133 |
"epoch": 2.94,
|
| 1134 |
"learning_rate": 5.3035143769968054e-05,
|
| 1135 |
-
"loss": 0.
|
| 1136 |
"step": 1840
|
| 1137 |
},
|
| 1138 |
{
|
| 1139 |
"epoch": 2.96,
|
| 1140 |
"learning_rate": 5.2236421725239616e-05,
|
| 1141 |
-
"loss": 0.
|
| 1142 |
"step": 1850
|
| 1143 |
},
|
| 1144 |
{
|
| 1145 |
"epoch": 2.97,
|
| 1146 |
"learning_rate": 5.1437699680511184e-05,
|
| 1147 |
-
"loss": 0.
|
| 1148 |
"step": 1860
|
| 1149 |
},
|
| 1150 |
{
|
| 1151 |
"epoch": 2.99,
|
| 1152 |
"learning_rate": 5.0638977635782745e-05,
|
| 1153 |
-
"loss": 0.
|
| 1154 |
"step": 1870
|
| 1155 |
},
|
| 1156 |
{
|
| 1157 |
"epoch": 3.0,
|
| 1158 |
-
"eval_accuracy": 0.
|
| 1159 |
-
"eval_f1": 0.
|
| 1160 |
-
"eval_loss": 0.
|
| 1161 |
-
"eval_precision": 0.
|
| 1162 |
-
"eval_recall": 0.
|
| 1163 |
-
"eval_runtime": 1.
|
| 1164 |
-
"eval_samples_per_second":
|
| 1165 |
-
"eval_steps_per_second":
|
| 1166 |
"step": 1878
|
| 1167 |
},
|
| 1168 |
{
|
| 1169 |
"epoch": 3.0,
|
| 1170 |
"learning_rate": 4.984025559105431e-05,
|
| 1171 |
-
"loss": 0.
|
| 1172 |
"step": 1880
|
| 1173 |
},
|
| 1174 |
{
|
| 1175 |
"epoch": 3.02,
|
| 1176 |
"learning_rate": 4.904153354632588e-05,
|
| 1177 |
-
"loss": 0.
|
| 1178 |
"step": 1890
|
| 1179 |
},
|
| 1180 |
{
|
| 1181 |
"epoch": 3.04,
|
| 1182 |
"learning_rate": 4.824281150159744e-05,
|
| 1183 |
-
"loss": 0.
|
| 1184 |
"step": 1900
|
| 1185 |
},
|
| 1186 |
{
|
| 1187 |
"epoch": 3.05,
|
| 1188 |
"learning_rate": 4.744408945686901e-05,
|
| 1189 |
-
"loss": 0.
|
| 1190 |
"step": 1910
|
| 1191 |
},
|
| 1192 |
{
|
| 1193 |
"epoch": 3.07,
|
| 1194 |
"learning_rate": 4.664536741214058e-05,
|
| 1195 |
-
"loss": 0.
|
| 1196 |
"step": 1920
|
| 1197 |
},
|
| 1198 |
{
|
| 1199 |
"epoch": 3.08,
|
| 1200 |
"learning_rate": 4.584664536741215e-05,
|
| 1201 |
-
"loss": 0.
|
| 1202 |
"step": 1930
|
| 1203 |
},
|
| 1204 |
{
|
| 1205 |
"epoch": 3.1,
|
| 1206 |
"learning_rate": 4.504792332268371e-05,
|
| 1207 |
-
"loss": 0.
|
| 1208 |
"step": 1940
|
| 1209 |
},
|
| 1210 |
{
|
| 1211 |
"epoch": 3.12,
|
| 1212 |
"learning_rate": 4.424920127795527e-05,
|
| 1213 |
-
"loss": 0.
|
| 1214 |
"step": 1950
|
| 1215 |
},
|
| 1216 |
{
|
| 1217 |
"epoch": 3.13,
|
| 1218 |
"learning_rate": 4.345047923322684e-05,
|
| 1219 |
-
"loss": 0.
|
| 1220 |
"step": 1960
|
| 1221 |
},
|
| 1222 |
{
|
| 1223 |
"epoch": 3.15,
|
| 1224 |
"learning_rate": 4.265175718849841e-05,
|
| 1225 |
-
"loss": 0.
|
| 1226 |
"step": 1970
|
| 1227 |
},
|
| 1228 |
{
|
| 1229 |
"epoch": 3.16,
|
| 1230 |
"learning_rate": 4.185303514376997e-05,
|
| 1231 |
-
"loss": 0.
|
| 1232 |
"step": 1980
|
| 1233 |
},
|
| 1234 |
{
|
| 1235 |
"epoch": 3.18,
|
| 1236 |
"learning_rate": 4.1054313099041536e-05,
|
| 1237 |
-
"loss": 0.
|
| 1238 |
"step": 1990
|
| 1239 |
},
|
| 1240 |
{
|
| 1241 |
"epoch": 3.19,
|
| 1242 |
"learning_rate": 4.0255591054313104e-05,
|
| 1243 |
-
"loss": 0.
|
| 1244 |
"step": 2000
|
| 1245 |
},
|
| 1246 |
{
|
| 1247 |
"epoch": 3.21,
|
| 1248 |
"learning_rate": 3.9456869009584666e-05,
|
| 1249 |
-
"loss": 0.
|
| 1250 |
"step": 2010
|
| 1251 |
},
|
| 1252 |
{
|
| 1253 |
"epoch": 3.23,
|
| 1254 |
"learning_rate": 3.8658146964856234e-05,
|
| 1255 |
-
"loss": 0.
|
| 1256 |
"step": 2020
|
| 1257 |
},
|
| 1258 |
{
|
| 1259 |
"epoch": 3.24,
|
| 1260 |
"learning_rate": 3.7859424920127795e-05,
|
| 1261 |
-
"loss": 0.
|
| 1262 |
"step": 2030
|
| 1263 |
},
|
| 1264 |
{
|
| 1265 |
"epoch": 3.26,
|
| 1266 |
"learning_rate": 3.7060702875399364e-05,
|
| 1267 |
-
"loss": 0.
|
| 1268 |
"step": 2040
|
| 1269 |
},
|
| 1270 |
{
|
| 1271 |
"epoch": 3.27,
|
| 1272 |
"learning_rate": 3.6261980830670925e-05,
|
| 1273 |
-
"loss": 0.
|
| 1274 |
"step": 2050
|
| 1275 |
},
|
| 1276 |
{
|
| 1277 |
"epoch": 3.29,
|
| 1278 |
"learning_rate": 3.546325878594249e-05,
|
| 1279 |
-
"loss": 0.
|
| 1280 |
"step": 2060
|
| 1281 |
},
|
| 1282 |
{
|
| 1283 |
"epoch": 3.31,
|
| 1284 |
"learning_rate": 3.466453674121406e-05,
|
| 1285 |
-
"loss": 0.
|
| 1286 |
"step": 2070
|
| 1287 |
},
|
| 1288 |
{
|
| 1289 |
"epoch": 3.32,
|
| 1290 |
"learning_rate": 3.386581469648562e-05,
|
| 1291 |
-
"loss": 0.
|
| 1292 |
"step": 2080
|
| 1293 |
},
|
| 1294 |
{
|
| 1295 |
"epoch": 3.34,
|
| 1296 |
"learning_rate": 3.306709265175719e-05,
|
| 1297 |
-
"loss": 0.
|
| 1298 |
"step": 2090
|
| 1299 |
},
|
| 1300 |
{
|
| 1301 |
"epoch": 3.35,
|
| 1302 |
"learning_rate": 3.226837060702875e-05,
|
| 1303 |
-
"loss": 0.
|
| 1304 |
"step": 2100
|
| 1305 |
},
|
| 1306 |
{
|
| 1307 |
"epoch": 3.37,
|
| 1308 |
"learning_rate": 3.146964856230032e-05,
|
| 1309 |
-
"loss": 0.
|
| 1310 |
"step": 2110
|
| 1311 |
},
|
| 1312 |
{
|
| 1313 |
"epoch": 3.39,
|
| 1314 |
"learning_rate": 3.067092651757188e-05,
|
| 1315 |
-
"loss": 0.
|
| 1316 |
"step": 2120
|
| 1317 |
},
|
| 1318 |
{
|
| 1319 |
"epoch": 3.4,
|
| 1320 |
"learning_rate": 2.987220447284345e-05,
|
| 1321 |
-
"loss": 0.
|
| 1322 |
"step": 2130
|
| 1323 |
},
|
| 1324 |
{
|
| 1325 |
"epoch": 3.42,
|
| 1326 |
"learning_rate": 2.907348242811502e-05,
|
| 1327 |
-
"loss": 0.
|
| 1328 |
"step": 2140
|
| 1329 |
},
|
| 1330 |
{
|
| 1331 |
"epoch": 3.43,
|
| 1332 |
"learning_rate": 2.8274760383386583e-05,
|
| 1333 |
-
"loss": 0.
|
| 1334 |
"step": 2150
|
| 1335 |
},
|
| 1336 |
{
|
| 1337 |
"epoch": 3.45,
|
| 1338 |
"learning_rate": 2.747603833865815e-05,
|
| 1339 |
-
"loss": 0.
|
| 1340 |
"step": 2160
|
| 1341 |
},
|
| 1342 |
{
|
| 1343 |
"epoch": 3.47,
|
| 1344 |
-
"learning_rate": 2.
|
| 1345 |
-
"loss": 0.
|
| 1346 |
"step": 2170
|
| 1347 |
},
|
| 1348 |
{
|
| 1349 |
"epoch": 3.48,
|
| 1350 |
-
"learning_rate": 2.
|
| 1351 |
-
"loss": 0.
|
| 1352 |
"step": 2180
|
| 1353 |
},
|
| 1354 |
{
|
| 1355 |
"epoch": 3.5,
|
| 1356 |
-
"learning_rate": 2.
|
| 1357 |
-
"loss": 0.
|
| 1358 |
"step": 2190
|
| 1359 |
},
|
| 1360 |
{
|
| 1361 |
"epoch": 3.51,
|
| 1362 |
-
"learning_rate": 2.
|
| 1363 |
-
"loss": 0.
|
| 1364 |
"step": 2200
|
| 1365 |
},
|
| 1366 |
{
|
| 1367 |
"epoch": 3.53,
|
| 1368 |
-
"learning_rate": 2.
|
| 1369 |
-
"loss": 0.
|
| 1370 |
"step": 2210
|
| 1371 |
},
|
| 1372 |
{
|
| 1373 |
"epoch": 3.55,
|
| 1374 |
-
"learning_rate": 2.
|
| 1375 |
-
"loss": 0.
|
| 1376 |
"step": 2220
|
| 1377 |
},
|
| 1378 |
{
|
| 1379 |
"epoch": 3.56,
|
| 1380 |
-
"learning_rate": 2.
|
| 1381 |
-
"loss": 0.
|
| 1382 |
"step": 2230
|
| 1383 |
},
|
| 1384 |
{
|
| 1385 |
"epoch": 3.58,
|
| 1386 |
-
"learning_rate": 2.
|
| 1387 |
-
"loss": 0.
|
| 1388 |
"step": 2240
|
| 1389 |
},
|
| 1390 |
{
|
| 1391 |
"epoch": 3.59,
|
| 1392 |
-
"learning_rate": 2.
|
| 1393 |
-
"loss": 0.
|
| 1394 |
"step": 2250
|
| 1395 |
},
|
| 1396 |
{
|
| 1397 |
"epoch": 3.61,
|
| 1398 |
-
"learning_rate": 1.
|
| 1399 |
-
"loss": 0.
|
| 1400 |
"step": 2260
|
| 1401 |
},
|
| 1402 |
{
|
| 1403 |
"epoch": 3.63,
|
| 1404 |
-
"learning_rate": 1.
|
| 1405 |
-
"loss": 0.
|
| 1406 |
"step": 2270
|
| 1407 |
},
|
| 1408 |
{
|
| 1409 |
"epoch": 3.64,
|
| 1410 |
-
"learning_rate": 1.
|
| 1411 |
-
"loss": 0.
|
| 1412 |
"step": 2280
|
| 1413 |
},
|
| 1414 |
{
|
| 1415 |
"epoch": 3.66,
|
| 1416 |
-
"learning_rate": 1.
|
| 1417 |
-
"loss": 0.
|
| 1418 |
"step": 2290
|
| 1419 |
},
|
| 1420 |
{
|
| 1421 |
"epoch": 3.67,
|
| 1422 |
-
"learning_rate": 1.
|
| 1423 |
-
"loss": 0.
|
| 1424 |
"step": 2300
|
| 1425 |
},
|
| 1426 |
{
|
| 1427 |
"epoch": 3.69,
|
| 1428 |
-
"learning_rate": 1.
|
| 1429 |
-
"loss": 0.
|
| 1430 |
"step": 2310
|
| 1431 |
},
|
| 1432 |
{
|
| 1433 |
"epoch": 3.71,
|
| 1434 |
-
"learning_rate": 1.
|
| 1435 |
-
"loss": 0.
|
| 1436 |
"step": 2320
|
| 1437 |
},
|
| 1438 |
{
|
| 1439 |
"epoch": 3.72,
|
| 1440 |
-
"learning_rate": 1.
|
| 1441 |
-
"loss": 0.
|
| 1442 |
"step": 2330
|
| 1443 |
},
|
| 1444 |
{
|
| 1445 |
"epoch": 3.74,
|
| 1446 |
-
"learning_rate": 1.
|
| 1447 |
-
"loss": 0.
|
| 1448 |
"step": 2340
|
| 1449 |
},
|
| 1450 |
{
|
| 1451 |
"epoch": 3.75,
|
| 1452 |
-
"learning_rate": 1.
|
| 1453 |
-
"loss": 0.
|
| 1454 |
"step": 2350
|
| 1455 |
},
|
| 1456 |
{
|
| 1457 |
"epoch": 3.77,
|
| 1458 |
-
"learning_rate": 1.
|
| 1459 |
-
"loss": 0.
|
| 1460 |
"step": 2360
|
| 1461 |
},
|
| 1462 |
{
|
| 1463 |
"epoch": 3.79,
|
| 1464 |
-
"learning_rate": 1.
|
| 1465 |
-
"loss": 0.
|
| 1466 |
"step": 2370
|
| 1467 |
},
|
| 1468 |
{
|
| 1469 |
"epoch": 3.8,
|
| 1470 |
-
"learning_rate": 9.
|
| 1471 |
-
"loss": 0.
|
| 1472 |
"step": 2380
|
| 1473 |
},
|
| 1474 |
{
|
| 1475 |
"epoch": 3.82,
|
| 1476 |
-
"learning_rate": 9.
|
| 1477 |
-
"loss": 0.
|
| 1478 |
"step": 2390
|
| 1479 |
},
|
| 1480 |
{
|
| 1481 |
"epoch": 3.83,
|
| 1482 |
-
"learning_rate": 8.
|
| 1483 |
-
"loss": 0.
|
| 1484 |
"step": 2400
|
| 1485 |
},
|
| 1486 |
{
|
| 1487 |
"epoch": 3.85,
|
| 1488 |
-
"learning_rate": 7.
|
| 1489 |
-
"loss": 0.
|
| 1490 |
"step": 2410
|
| 1491 |
},
|
| 1492 |
{
|
| 1493 |
"epoch": 3.87,
|
| 1494 |
-
"learning_rate": 6.
|
| 1495 |
-
"loss": 0.
|
| 1496 |
"step": 2420
|
| 1497 |
},
|
| 1498 |
{
|
| 1499 |
"epoch": 3.88,
|
| 1500 |
-
"learning_rate": 5.
|
| 1501 |
-
"loss": 0.
|
| 1502 |
"step": 2430
|
| 1503 |
},
|
| 1504 |
{
|
| 1505 |
"epoch": 3.9,
|
| 1506 |
-
"learning_rate": 5.
|
| 1507 |
-
"loss": 0.
|
| 1508 |
"step": 2440
|
| 1509 |
},
|
| 1510 |
{
|
| 1511 |
"epoch": 3.91,
|
| 1512 |
-
"learning_rate": 4.
|
| 1513 |
-
"loss": 0.
|
| 1514 |
"step": 2450
|
| 1515 |
},
|
| 1516 |
{
|
| 1517 |
"epoch": 3.93,
|
| 1518 |
-
"learning_rate": 3.
|
| 1519 |
-
"loss": 0.
|
| 1520 |
"step": 2460
|
| 1521 |
},
|
| 1522 |
{
|
| 1523 |
"epoch": 3.95,
|
| 1524 |
-
"learning_rate": 2.
|
| 1525 |
-
"loss": 0.
|
| 1526 |
"step": 2470
|
| 1527 |
},
|
| 1528 |
{
|
| 1529 |
"epoch": 3.96,
|
| 1530 |
-
"learning_rate": 1.
|
| 1531 |
-
"loss": 0.
|
| 1532 |
"step": 2480
|
| 1533 |
},
|
| 1534 |
{
|
| 1535 |
"epoch": 3.98,
|
| 1536 |
-
"learning_rate": 1.
|
| 1537 |
-
"loss": 0.
|
| 1538 |
"step": 2490
|
| 1539 |
},
|
| 1540 |
{
|
| 1541 |
"epoch": 3.99,
|
| 1542 |
-
"learning_rate": 3.
|
| 1543 |
-
"loss": 0.
|
| 1544 |
"step": 2500
|
| 1545 |
},
|
| 1546 |
{
|
| 1547 |
"epoch": 4.0,
|
| 1548 |
-
"eval_accuracy": 0.
|
| 1549 |
-
"eval_f1": 0.
|
| 1550 |
-
"eval_loss": 0.
|
| 1551 |
-
"eval_precision": 0.
|
| 1552 |
-
"eval_recall": 0.
|
| 1553 |
-
"eval_runtime": 1.
|
| 1554 |
-
"eval_samples_per_second":
|
| 1555 |
-
"eval_steps_per_second": 9.
|
| 1556 |
"step": 2504
|
| 1557 |
},
|
| 1558 |
{
|
| 1559 |
"epoch": 4.0,
|
| 1560 |
"step": 2504,
|
| 1561 |
"total_flos": 3.104468219559813e+18,
|
| 1562 |
-
"train_loss": 0.
|
| 1563 |
-
"train_runtime":
|
| 1564 |
-
"train_samples_per_second": 47.
|
| 1565 |
-
"train_steps_per_second": 2.
|
| 1566 |
}
|
| 1567 |
],
|
| 1568 |
"max_steps": 2504,
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.8756476683937824,
|
| 3 |
"best_model_checkpoint": "./vit-base-skin/checkpoint-1252",
|
| 4 |
"epoch": 4.0,
|
| 5 |
"global_step": 2504,
|
|
|
|
| 10 |
{
|
| 11 |
"epoch": 0.02,
|
| 12 |
"learning_rate": 0.00019920127795527157,
|
| 13 |
+
"loss": 0.078,
|
| 14 |
"step": 10
|
| 15 |
},
|
| 16 |
{
|
| 17 |
"epoch": 0.03,
|
| 18 |
"learning_rate": 0.00019840255591054313,
|
| 19 |
+
"loss": 0.2128,
|
| 20 |
"step": 20
|
| 21 |
},
|
| 22 |
{
|
| 23 |
"epoch": 0.05,
|
| 24 |
"learning_rate": 0.00019760383386581472,
|
| 25 |
+
"loss": 0.0939,
|
| 26 |
"step": 30
|
| 27 |
},
|
| 28 |
{
|
| 29 |
"epoch": 0.06,
|
| 30 |
"learning_rate": 0.00019680511182108628,
|
| 31 |
+
"loss": 0.1508,
|
| 32 |
"step": 40
|
| 33 |
},
|
| 34 |
{
|
| 35 |
"epoch": 0.08,
|
| 36 |
"learning_rate": 0.00019600638977635784,
|
| 37 |
+
"loss": 0.1135,
|
| 38 |
"step": 50
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 0.1,
|
| 42 |
"learning_rate": 0.0001952076677316294,
|
| 43 |
+
"loss": 0.242,
|
| 44 |
"step": 60
|
| 45 |
},
|
| 46 |
{
|
| 47 |
"epoch": 0.11,
|
| 48 |
"learning_rate": 0.00019440894568690097,
|
| 49 |
+
"loss": 0.2251,
|
| 50 |
"step": 70
|
| 51 |
},
|
| 52 |
{
|
| 53 |
"epoch": 0.13,
|
| 54 |
"learning_rate": 0.00019361022364217253,
|
| 55 |
+
"loss": 0.1199,
|
| 56 |
"step": 80
|
| 57 |
},
|
| 58 |
{
|
| 59 |
"epoch": 0.14,
|
| 60 |
"learning_rate": 0.0001928115015974441,
|
| 61 |
+
"loss": 0.1629,
|
| 62 |
"step": 90
|
| 63 |
},
|
| 64 |
{
|
| 65 |
"epoch": 0.16,
|
| 66 |
"learning_rate": 0.00019201277955271565,
|
| 67 |
+
"loss": 0.1078,
|
| 68 |
"step": 100
|
| 69 |
},
|
| 70 |
{
|
| 71 |
"epoch": 0.18,
|
| 72 |
"learning_rate": 0.0001912140575079872,
|
| 73 |
+
"loss": 0.2215,
|
| 74 |
"step": 110
|
| 75 |
},
|
| 76 |
{
|
| 77 |
"epoch": 0.19,
|
| 78 |
"learning_rate": 0.0001904153354632588,
|
| 79 |
+
"loss": 0.1648,
|
| 80 |
"step": 120
|
| 81 |
},
|
| 82 |
{
|
| 83 |
"epoch": 0.21,
|
| 84 |
"learning_rate": 0.00018961661341853036,
|
| 85 |
+
"loss": 0.2631,
|
| 86 |
"step": 130
|
| 87 |
},
|
| 88 |
{
|
| 89 |
"epoch": 0.22,
|
| 90 |
"learning_rate": 0.00018881789137380192,
|
| 91 |
+
"loss": 0.1141,
|
| 92 |
"step": 140
|
| 93 |
},
|
| 94 |
{
|
| 95 |
"epoch": 0.24,
|
| 96 |
"learning_rate": 0.00018801916932907348,
|
| 97 |
+
"loss": 0.138,
|
| 98 |
"step": 150
|
| 99 |
},
|
| 100 |
{
|
| 101 |
"epoch": 0.26,
|
| 102 |
"learning_rate": 0.00018722044728434505,
|
| 103 |
+
"loss": 0.1407,
|
| 104 |
"step": 160
|
| 105 |
},
|
| 106 |
{
|
| 107 |
"epoch": 0.27,
|
| 108 |
"learning_rate": 0.00018642172523961663,
|
| 109 |
+
"loss": 0.1169,
|
| 110 |
"step": 170
|
| 111 |
},
|
| 112 |
{
|
| 113 |
"epoch": 0.29,
|
| 114 |
"learning_rate": 0.0001856230031948882,
|
| 115 |
+
"loss": 0.1297,
|
| 116 |
"step": 180
|
| 117 |
},
|
| 118 |
{
|
| 119 |
"epoch": 0.3,
|
| 120 |
"learning_rate": 0.00018482428115015976,
|
| 121 |
+
"loss": 0.1772,
|
| 122 |
"step": 190
|
| 123 |
},
|
| 124 |
{
|
| 125 |
"epoch": 0.32,
|
| 126 |
"learning_rate": 0.00018402555910543132,
|
| 127 |
+
"loss": 0.3667,
|
| 128 |
"step": 200
|
| 129 |
},
|
| 130 |
{
|
| 131 |
"epoch": 0.34,
|
| 132 |
"learning_rate": 0.00018322683706070288,
|
| 133 |
+
"loss": 0.1505,
|
| 134 |
"step": 210
|
| 135 |
},
|
| 136 |
{
|
| 137 |
"epoch": 0.35,
|
| 138 |
"learning_rate": 0.00018242811501597444,
|
| 139 |
+
"loss": 0.134,
|
| 140 |
"step": 220
|
| 141 |
},
|
| 142 |
{
|
| 143 |
"epoch": 0.37,
|
| 144 |
"learning_rate": 0.000181629392971246,
|
| 145 |
+
"loss": 0.1385,
|
| 146 |
"step": 230
|
| 147 |
},
|
| 148 |
{
|
| 149 |
"epoch": 0.38,
|
| 150 |
"learning_rate": 0.00018083067092651756,
|
| 151 |
+
"loss": 0.0883,
|
| 152 |
"step": 240
|
| 153 |
},
|
| 154 |
{
|
| 155 |
"epoch": 0.4,
|
| 156 |
"learning_rate": 0.00018003194888178913,
|
| 157 |
+
"loss": 0.1612,
|
| 158 |
"step": 250
|
| 159 |
},
|
| 160 |
{
|
| 161 |
"epoch": 0.42,
|
| 162 |
"learning_rate": 0.00017923322683706071,
|
| 163 |
+
"loss": 0.1265,
|
| 164 |
"step": 260
|
| 165 |
},
|
| 166 |
{
|
| 167 |
"epoch": 0.43,
|
| 168 |
"learning_rate": 0.00017843450479233228,
|
| 169 |
+
"loss": 0.2464,
|
| 170 |
"step": 270
|
| 171 |
},
|
| 172 |
{
|
| 173 |
"epoch": 0.45,
|
| 174 |
"learning_rate": 0.00017763578274760384,
|
| 175 |
+
"loss": 0.1802,
|
| 176 |
"step": 280
|
| 177 |
},
|
| 178 |
{
|
| 179 |
"epoch": 0.46,
|
| 180 |
"learning_rate": 0.0001768370607028754,
|
| 181 |
+
"loss": 0.1718,
|
| 182 |
"step": 290
|
| 183 |
},
|
| 184 |
{
|
| 185 |
"epoch": 0.48,
|
| 186 |
"learning_rate": 0.000176038338658147,
|
| 187 |
+
"loss": 0.1424,
|
| 188 |
"step": 300
|
| 189 |
},
|
| 190 |
{
|
| 191 |
"epoch": 0.5,
|
| 192 |
"learning_rate": 0.00017523961661341855,
|
| 193 |
+
"loss": 0.1109,
|
| 194 |
"step": 310
|
| 195 |
},
|
| 196 |
{
|
| 197 |
"epoch": 0.51,
|
| 198 |
"learning_rate": 0.0001744408945686901,
|
| 199 |
+
"loss": 0.1093,
|
| 200 |
"step": 320
|
| 201 |
},
|
| 202 |
{
|
| 203 |
"epoch": 0.53,
|
| 204 |
"learning_rate": 0.00017364217252396167,
|
| 205 |
+
"loss": 0.2388,
|
| 206 |
"step": 330
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 0.54,
|
| 210 |
"learning_rate": 0.00017284345047923323,
|
| 211 |
+
"loss": 0.165,
|
| 212 |
"step": 340
|
| 213 |
},
|
| 214 |
{
|
| 215 |
"epoch": 0.56,
|
| 216 |
"learning_rate": 0.00017204472843450482,
|
| 217 |
+
"loss": 0.1271,
|
| 218 |
"step": 350
|
| 219 |
},
|
| 220 |
{
|
| 221 |
"epoch": 0.58,
|
| 222 |
"learning_rate": 0.00017124600638977638,
|
| 223 |
+
"loss": 0.1753,
|
| 224 |
"step": 360
|
| 225 |
},
|
| 226 |
{
|
| 227 |
"epoch": 0.59,
|
| 228 |
"learning_rate": 0.00017044728434504792,
|
| 229 |
+
"loss": 0.078,
|
| 230 |
"step": 370
|
| 231 |
},
|
| 232 |
{
|
| 233 |
"epoch": 0.61,
|
| 234 |
"learning_rate": 0.00016964856230031948,
|
| 235 |
+
"loss": 0.2373,
|
| 236 |
"step": 380
|
| 237 |
},
|
| 238 |
{
|
| 239 |
"epoch": 0.62,
|
| 240 |
"learning_rate": 0.00016884984025559104,
|
| 241 |
+
"loss": 0.1719,
|
| 242 |
"step": 390
|
| 243 |
},
|
| 244 |
{
|
| 245 |
"epoch": 0.64,
|
| 246 |
"learning_rate": 0.00016805111821086263,
|
| 247 |
+
"loss": 0.1963,
|
| 248 |
"step": 400
|
| 249 |
},
|
| 250 |
{
|
| 251 |
"epoch": 0.65,
|
| 252 |
"learning_rate": 0.0001672523961661342,
|
| 253 |
+
"loss": 0.1671,
|
| 254 |
"step": 410
|
| 255 |
},
|
| 256 |
{
|
| 257 |
"epoch": 0.67,
|
| 258 |
"learning_rate": 0.00016645367412140575,
|
| 259 |
+
"loss": 0.1733,
|
| 260 |
"step": 420
|
| 261 |
},
|
| 262 |
{
|
| 263 |
"epoch": 0.69,
|
| 264 |
"learning_rate": 0.0001656549520766773,
|
| 265 |
+
"loss": 0.1268,
|
| 266 |
"step": 430
|
| 267 |
},
|
| 268 |
{
|
| 269 |
"epoch": 0.7,
|
| 270 |
"learning_rate": 0.0001648562300319489,
|
| 271 |
+
"loss": 0.185,
|
| 272 |
"step": 440
|
| 273 |
},
|
| 274 |
{
|
| 275 |
"epoch": 0.72,
|
| 276 |
"learning_rate": 0.00016405750798722046,
|
| 277 |
+
"loss": 0.0595,
|
| 278 |
"step": 450
|
| 279 |
},
|
| 280 |
{
|
| 281 |
"epoch": 0.73,
|
| 282 |
"learning_rate": 0.00016325878594249202,
|
| 283 |
+
"loss": 0.0833,
|
| 284 |
"step": 460
|
| 285 |
},
|
| 286 |
{
|
| 287 |
"epoch": 0.75,
|
| 288 |
"learning_rate": 0.00016246006389776359,
|
| 289 |
+
"loss": 0.2265,
|
| 290 |
"step": 470
|
| 291 |
},
|
| 292 |
{
|
| 293 |
"epoch": 0.77,
|
| 294 |
"learning_rate": 0.00016166134185303515,
|
| 295 |
+
"loss": 0.1407,
|
| 296 |
"step": 480
|
| 297 |
},
|
| 298 |
{
|
| 299 |
"epoch": 0.78,
|
| 300 |
"learning_rate": 0.00016086261980830674,
|
| 301 |
+
"loss": 0.1786,
|
| 302 |
"step": 490
|
| 303 |
},
|
| 304 |
{
|
| 305 |
"epoch": 0.8,
|
| 306 |
"learning_rate": 0.0001600638977635783,
|
| 307 |
+
"loss": 0.1276,
|
| 308 |
"step": 500
|
| 309 |
},
|
| 310 |
{
|
| 311 |
"epoch": 0.81,
|
| 312 |
"learning_rate": 0.00015926517571884986,
|
| 313 |
+
"loss": 0.1227,
|
| 314 |
"step": 510
|
| 315 |
},
|
| 316 |
{
|
| 317 |
"epoch": 0.83,
|
| 318 |
"learning_rate": 0.00015846645367412142,
|
| 319 |
+
"loss": 0.1551,
|
| 320 |
"step": 520
|
| 321 |
},
|
| 322 |
{
|
| 323 |
"epoch": 0.85,
|
| 324 |
"learning_rate": 0.00015766773162939298,
|
| 325 |
+
"loss": 0.1649,
|
| 326 |
"step": 530
|
| 327 |
},
|
| 328 |
{
|
| 329 |
"epoch": 0.86,
|
| 330 |
"learning_rate": 0.00015686900958466454,
|
| 331 |
+
"loss": 0.3069,
|
| 332 |
"step": 540
|
| 333 |
},
|
| 334 |
{
|
| 335 |
"epoch": 0.88,
|
| 336 |
"learning_rate": 0.0001560702875399361,
|
| 337 |
+
"loss": 0.1824,
|
| 338 |
"step": 550
|
| 339 |
},
|
| 340 |
{
|
| 341 |
"epoch": 0.89,
|
| 342 |
"learning_rate": 0.00015527156549520767,
|
| 343 |
+
"loss": 0.1662,
|
| 344 |
"step": 560
|
| 345 |
},
|
| 346 |
{
|
| 347 |
"epoch": 0.91,
|
| 348 |
"learning_rate": 0.00015447284345047923,
|
| 349 |
+
"loss": 0.2483,
|
| 350 |
"step": 570
|
| 351 |
},
|
| 352 |
{
|
| 353 |
"epoch": 0.93,
|
| 354 |
"learning_rate": 0.00015367412140575082,
|
| 355 |
+
"loss": 0.1293,
|
| 356 |
"step": 580
|
| 357 |
},
|
| 358 |
{
|
| 359 |
"epoch": 0.94,
|
| 360 |
"learning_rate": 0.00015287539936102238,
|
| 361 |
+
"loss": 0.171,
|
| 362 |
"step": 590
|
| 363 |
},
|
| 364 |
{
|
| 365 |
"epoch": 0.96,
|
| 366 |
"learning_rate": 0.00015207667731629394,
|
| 367 |
+
"loss": 0.1812,
|
| 368 |
"step": 600
|
| 369 |
},
|
| 370 |
{
|
| 371 |
"epoch": 0.97,
|
| 372 |
"learning_rate": 0.0001512779552715655,
|
| 373 |
+
"loss": 0.1628,
|
| 374 |
"step": 610
|
| 375 |
},
|
| 376 |
{
|
| 377 |
"epoch": 0.99,
|
| 378 |
"learning_rate": 0.00015047923322683706,
|
| 379 |
+
"loss": 0.1259,
|
| 380 |
"step": 620
|
| 381 |
},
|
| 382 |
{
|
| 383 |
"epoch": 1.0,
|
| 384 |
+
"eval_accuracy": 0.8341968911917098,
|
| 385 |
+
"eval_f1": 0.8404495414024173,
|
| 386 |
+
"eval_loss": 0.38346073031425476,
|
| 387 |
+
"eval_precision": 0.8537925291610302,
|
| 388 |
+
"eval_recall": 0.8341968911917098,
|
| 389 |
+
"eval_runtime": 1.2354,
|
| 390 |
+
"eval_samples_per_second": 156.22,
|
| 391 |
+
"eval_steps_per_second": 10.523,
|
| 392 |
"step": 626
|
| 393 |
},
|
| 394 |
{
|
| 395 |
"epoch": 1.01,
|
| 396 |
"learning_rate": 0.00014968051118210865,
|
| 397 |
+
"loss": 0.1407,
|
| 398 |
"step": 630
|
| 399 |
},
|
| 400 |
{
|
| 401 |
"epoch": 1.02,
|
| 402 |
"learning_rate": 0.0001488817891373802,
|
| 403 |
+
"loss": 0.0463,
|
| 404 |
"step": 640
|
| 405 |
},
|
| 406 |
{
|
| 407 |
"epoch": 1.04,
|
| 408 |
"learning_rate": 0.00014808306709265177,
|
| 409 |
+
"loss": 0.0718,
|
| 410 |
"step": 650
|
| 411 |
},
|
| 412 |
{
|
| 413 |
"epoch": 1.05,
|
| 414 |
"learning_rate": 0.00014728434504792333,
|
| 415 |
+
"loss": 0.097,
|
| 416 |
"step": 660
|
| 417 |
},
|
| 418 |
{
|
| 419 |
"epoch": 1.07,
|
| 420 |
"learning_rate": 0.0001464856230031949,
|
| 421 |
+
"loss": 0.0659,
|
| 422 |
"step": 670
|
| 423 |
},
|
| 424 |
{
|
| 425 |
"epoch": 1.09,
|
| 426 |
"learning_rate": 0.00014568690095846646,
|
| 427 |
+
"loss": 0.0862,
|
| 428 |
"step": 680
|
| 429 |
},
|
| 430 |
{
|
| 431 |
"epoch": 1.1,
|
| 432 |
"learning_rate": 0.00014488817891373802,
|
| 433 |
+
"loss": 0.0643,
|
| 434 |
"step": 690
|
| 435 |
},
|
| 436 |
{
|
| 437 |
"epoch": 1.12,
|
| 438 |
"learning_rate": 0.00014408945686900958,
|
| 439 |
+
"loss": 0.0632,
|
| 440 |
"step": 700
|
| 441 |
},
|
| 442 |
{
|
| 443 |
"epoch": 1.13,
|
| 444 |
"learning_rate": 0.00014329073482428114,
|
| 445 |
+
"loss": 0.132,
|
| 446 |
"step": 710
|
| 447 |
},
|
| 448 |
{
|
| 449 |
"epoch": 1.15,
|
| 450 |
"learning_rate": 0.00014249201277955273,
|
| 451 |
+
"loss": 0.0619,
|
| 452 |
"step": 720
|
| 453 |
},
|
| 454 |
{
|
| 455 |
"epoch": 1.17,
|
| 456 |
"learning_rate": 0.0001416932907348243,
|
| 457 |
+
"loss": 0.0665,
|
| 458 |
"step": 730
|
| 459 |
},
|
| 460 |
{
|
| 461 |
"epoch": 1.18,
|
| 462 |
"learning_rate": 0.00014089456869009585,
|
| 463 |
+
"loss": 0.0407,
|
| 464 |
"step": 740
|
| 465 |
},
|
| 466 |
{
|
| 467 |
"epoch": 1.2,
|
| 468 |
"learning_rate": 0.0001400958466453674,
|
| 469 |
+
"loss": 0.0492,
|
| 470 |
"step": 750
|
| 471 |
},
|
| 472 |
{
|
| 473 |
"epoch": 1.21,
|
| 474 |
"learning_rate": 0.000139297124600639,
|
| 475 |
+
"loss": 0.0754,
|
| 476 |
"step": 760
|
| 477 |
},
|
| 478 |
{
|
| 479 |
"epoch": 1.23,
|
| 480 |
"learning_rate": 0.00013849840255591056,
|
| 481 |
+
"loss": 0.0884,
|
| 482 |
"step": 770
|
| 483 |
},
|
| 484 |
{
|
| 485 |
"epoch": 1.25,
|
| 486 |
"learning_rate": 0.00013769968051118212,
|
| 487 |
+
"loss": 0.1073,
|
| 488 |
"step": 780
|
| 489 |
},
|
| 490 |
{
|
| 491 |
"epoch": 1.26,
|
| 492 |
"learning_rate": 0.00013690095846645369,
|
| 493 |
+
"loss": 0.0956,
|
| 494 |
"step": 790
|
| 495 |
},
|
| 496 |
{
|
| 497 |
"epoch": 1.28,
|
| 498 |
"learning_rate": 0.00013610223642172525,
|
| 499 |
+
"loss": 0.064,
|
| 500 |
"step": 800
|
| 501 |
},
|
| 502 |
{
|
| 503 |
"epoch": 1.29,
|
| 504 |
"learning_rate": 0.0001353035143769968,
|
| 505 |
+
"loss": 0.0879,
|
| 506 |
"step": 810
|
| 507 |
},
|
| 508 |
{
|
| 509 |
"epoch": 1.31,
|
| 510 |
"learning_rate": 0.00013450479233226837,
|
| 511 |
+
"loss": 0.0781,
|
| 512 |
"step": 820
|
| 513 |
},
|
| 514 |
{
|
| 515 |
"epoch": 1.33,
|
| 516 |
"learning_rate": 0.00013370607028753993,
|
| 517 |
+
"loss": 0.023,
|
| 518 |
"step": 830
|
| 519 |
},
|
| 520 |
{
|
| 521 |
"epoch": 1.34,
|
| 522 |
"learning_rate": 0.0001329073482428115,
|
| 523 |
+
"loss": 0.0874,
|
| 524 |
"step": 840
|
| 525 |
},
|
| 526 |
{
|
| 527 |
"epoch": 1.36,
|
| 528 |
"learning_rate": 0.00013210862619808305,
|
| 529 |
+
"loss": 0.0265,
|
| 530 |
"step": 850
|
| 531 |
},
|
| 532 |
{
|
| 533 |
"epoch": 1.37,
|
| 534 |
"learning_rate": 0.00013130990415335464,
|
| 535 |
+
"loss": 0.1008,
|
| 536 |
"step": 860
|
| 537 |
},
|
| 538 |
{
|
| 539 |
"epoch": 1.39,
|
| 540 |
"learning_rate": 0.0001305111821086262,
|
| 541 |
+
"loss": 0.106,
|
| 542 |
"step": 870
|
| 543 |
},
|
| 544 |
{
|
| 545 |
"epoch": 1.41,
|
| 546 |
"learning_rate": 0.00012971246006389777,
|
| 547 |
+
"loss": 0.1314,
|
| 548 |
"step": 880
|
| 549 |
},
|
| 550 |
{
|
| 551 |
"epoch": 1.42,
|
| 552 |
"learning_rate": 0.00012891373801916933,
|
| 553 |
+
"loss": 0.1214,
|
| 554 |
"step": 890
|
| 555 |
},
|
| 556 |
{
|
| 557 |
"epoch": 1.44,
|
| 558 |
"learning_rate": 0.00012811501597444092,
|
| 559 |
+
"loss": 0.0988,
|
| 560 |
"step": 900
|
| 561 |
},
|
| 562 |
{
|
| 563 |
"epoch": 1.45,
|
| 564 |
"learning_rate": 0.00012731629392971248,
|
| 565 |
+
"loss": 0.1319,
|
| 566 |
"step": 910
|
| 567 |
},
|
| 568 |
{
|
| 569 |
"epoch": 1.47,
|
| 570 |
"learning_rate": 0.00012651757188498404,
|
| 571 |
+
"loss": 0.0993,
|
| 572 |
"step": 920
|
| 573 |
},
|
| 574 |
{
|
| 575 |
"epoch": 1.49,
|
| 576 |
"learning_rate": 0.0001257188498402556,
|
| 577 |
+
"loss": 0.0993,
|
| 578 |
"step": 930
|
| 579 |
},
|
| 580 |
{
|
| 581 |
"epoch": 1.5,
|
| 582 |
"learning_rate": 0.00012492012779552716,
|
| 583 |
+
"loss": 0.0323,
|
| 584 |
"step": 940
|
| 585 |
},
|
| 586 |
{
|
| 587 |
"epoch": 1.52,
|
| 588 |
"learning_rate": 0.00012412140575079872,
|
| 589 |
+
"loss": 0.0544,
|
| 590 |
"step": 950
|
| 591 |
},
|
| 592 |
{
|
| 593 |
"epoch": 1.53,
|
| 594 |
"learning_rate": 0.00012332268370607028,
|
| 595 |
+
"loss": 0.0388,
|
| 596 |
"step": 960
|
| 597 |
},
|
| 598 |
{
|
| 599 |
"epoch": 1.55,
|
| 600 |
"learning_rate": 0.00012252396166134185,
|
| 601 |
+
"loss": 0.0738,
|
| 602 |
"step": 970
|
| 603 |
},
|
| 604 |
{
|
| 605 |
"epoch": 1.57,
|
| 606 |
"learning_rate": 0.00012172523961661342,
|
| 607 |
+
"loss": 0.0705,
|
| 608 |
"step": 980
|
| 609 |
},
|
| 610 |
{
|
| 611 |
"epoch": 1.58,
|
| 612 |
"learning_rate": 0.000120926517571885,
|
| 613 |
+
"loss": 0.1676,
|
| 614 |
"step": 990
|
| 615 |
},
|
| 616 |
{
|
| 617 |
"epoch": 1.6,
|
| 618 |
"learning_rate": 0.00012012779552715656,
|
| 619 |
+
"loss": 0.0677,
|
| 620 |
"step": 1000
|
| 621 |
},
|
| 622 |
{
|
| 623 |
"epoch": 1.61,
|
| 624 |
"learning_rate": 0.00011932907348242812,
|
| 625 |
+
"loss": 0.0742,
|
| 626 |
"step": 1010
|
| 627 |
},
|
| 628 |
{
|
| 629 |
"epoch": 1.63,
|
| 630 |
"learning_rate": 0.00011853035143769968,
|
| 631 |
+
"loss": 0.0339,
|
| 632 |
"step": 1020
|
| 633 |
},
|
| 634 |
{
|
| 635 |
"epoch": 1.65,
|
| 636 |
"learning_rate": 0.00011773162939297124,
|
| 637 |
+
"loss": 0.1323,
|
| 638 |
"step": 1030
|
| 639 |
},
|
| 640 |
{
|
| 641 |
"epoch": 1.66,
|
| 642 |
"learning_rate": 0.00011693290734824283,
|
| 643 |
+
"loss": 0.0707,
|
| 644 |
"step": 1040
|
| 645 |
},
|
| 646 |
{
|
| 647 |
"epoch": 1.68,
|
| 648 |
"learning_rate": 0.00011613418530351439,
|
| 649 |
+
"loss": 0.1936,
|
| 650 |
"step": 1050
|
| 651 |
},
|
| 652 |
{
|
| 653 |
"epoch": 1.69,
|
| 654 |
"learning_rate": 0.00011533546325878595,
|
| 655 |
+
"loss": 0.1448,
|
| 656 |
"step": 1060
|
| 657 |
},
|
| 658 |
{
|
| 659 |
"epoch": 1.71,
|
| 660 |
"learning_rate": 0.00011453674121405751,
|
| 661 |
+
"loss": 0.1723,
|
| 662 |
"step": 1070
|
| 663 |
},
|
| 664 |
{
|
| 665 |
"epoch": 1.73,
|
| 666 |
"learning_rate": 0.00011373801916932908,
|
| 667 |
+
"loss": 0.0485,
|
| 668 |
"step": 1080
|
| 669 |
},
|
| 670 |
{
|
| 671 |
"epoch": 1.74,
|
| 672 |
"learning_rate": 0.00011293929712460065,
|
| 673 |
+
"loss": 0.0797,
|
| 674 |
"step": 1090
|
| 675 |
},
|
| 676 |
{
|
| 677 |
"epoch": 1.76,
|
| 678 |
"learning_rate": 0.00011214057507987221,
|
| 679 |
+
"loss": 0.1008,
|
| 680 |
"step": 1100
|
| 681 |
},
|
| 682 |
{
|
| 683 |
"epoch": 1.77,
|
| 684 |
"learning_rate": 0.00011134185303514377,
|
| 685 |
+
"loss": 0.0858,
|
| 686 |
"step": 1110
|
| 687 |
},
|
| 688 |
{
|
| 689 |
"epoch": 1.79,
|
| 690 |
"learning_rate": 0.00011054313099041533,
|
| 691 |
+
"loss": 0.0537,
|
| 692 |
"step": 1120
|
| 693 |
},
|
| 694 |
{
|
| 695 |
"epoch": 1.81,
|
| 696 |
"learning_rate": 0.00010974440894568691,
|
| 697 |
+
"loss": 0.1089,
|
| 698 |
"step": 1130
|
| 699 |
},
|
| 700 |
{
|
| 701 |
"epoch": 1.82,
|
| 702 |
"learning_rate": 0.00010894568690095847,
|
| 703 |
+
"loss": 0.1016,
|
| 704 |
"step": 1140
|
| 705 |
},
|
| 706 |
{
|
| 707 |
"epoch": 1.84,
|
| 708 |
"learning_rate": 0.00010814696485623003,
|
| 709 |
+
"loss": 0.0771,
|
| 710 |
"step": 1150
|
| 711 |
},
|
| 712 |
{
|
| 713 |
"epoch": 1.85,
|
| 714 |
"learning_rate": 0.0001073482428115016,
|
| 715 |
+
"loss": 0.0549,
|
| 716 |
"step": 1160
|
| 717 |
},
|
| 718 |
{
|
| 719 |
"epoch": 1.87,
|
| 720 |
"learning_rate": 0.00010654952076677316,
|
| 721 |
+
"loss": 0.0593,
|
| 722 |
"step": 1170
|
| 723 |
},
|
| 724 |
{
|
| 725 |
"epoch": 1.88,
|
| 726 |
"learning_rate": 0.00010575079872204474,
|
| 727 |
+
"loss": 0.1017,
|
| 728 |
"step": 1180
|
| 729 |
},
|
| 730 |
{
|
| 731 |
"epoch": 1.9,
|
| 732 |
"learning_rate": 0.0001049520766773163,
|
| 733 |
+
"loss": 0.0265,
|
| 734 |
"step": 1190
|
| 735 |
},
|
| 736 |
{
|
| 737 |
"epoch": 1.92,
|
| 738 |
"learning_rate": 0.00010415335463258787,
|
| 739 |
+
"loss": 0.05,
|
| 740 |
"step": 1200
|
| 741 |
},
|
| 742 |
{
|
| 743 |
"epoch": 1.93,
|
| 744 |
"learning_rate": 0.00010335463258785943,
|
| 745 |
+
"loss": 0.0442,
|
| 746 |
"step": 1210
|
| 747 |
},
|
| 748 |
{
|
| 749 |
"epoch": 1.95,
|
| 750 |
"learning_rate": 0.000102555910543131,
|
| 751 |
+
"loss": 0.0524,
|
| 752 |
"step": 1220
|
| 753 |
},
|
| 754 |
{
|
| 755 |
"epoch": 1.96,
|
| 756 |
"learning_rate": 0.00010175718849840256,
|
| 757 |
+
"loss": 0.0772,
|
| 758 |
"step": 1230
|
| 759 |
},
|
| 760 |
{
|
| 761 |
"epoch": 1.98,
|
| 762 |
"learning_rate": 0.00010095846645367413,
|
| 763 |
+
"loss": 0.182,
|
| 764 |
"step": 1240
|
| 765 |
},
|
| 766 |
{
|
| 767 |
"epoch": 2.0,
|
| 768 |
"learning_rate": 0.00010015974440894569,
|
| 769 |
+
"loss": 0.156,
|
| 770 |
"step": 1250
|
| 771 |
},
|
| 772 |
{
|
| 773 |
"epoch": 2.0,
|
| 774 |
+
"eval_accuracy": 0.8756476683937824,
|
| 775 |
+
"eval_f1": 0.8658135325711,
|
| 776 |
+
"eval_loss": 0.37016573548316956,
|
| 777 |
+
"eval_precision": 0.8651736102970483,
|
| 778 |
+
"eval_recall": 0.8756476683937824,
|
| 779 |
+
"eval_runtime": 1.3092,
|
| 780 |
+
"eval_samples_per_second": 147.419,
|
| 781 |
+
"eval_steps_per_second": 9.93,
|
| 782 |
"step": 1252
|
| 783 |
},
|
| 784 |
{
|
| 785 |
"epoch": 2.01,
|
| 786 |
"learning_rate": 9.936102236421726e-05,
|
| 787 |
+
"loss": 0.0584,
|
| 788 |
"step": 1260
|
| 789 |
},
|
| 790 |
{
|
| 791 |
"epoch": 2.03,
|
| 792 |
"learning_rate": 9.856230031948882e-05,
|
| 793 |
+
"loss": 0.0477,
|
| 794 |
"step": 1270
|
| 795 |
},
|
| 796 |
{
|
| 797 |
"epoch": 2.04,
|
| 798 |
"learning_rate": 9.77635782747604e-05,
|
| 799 |
+
"loss": 0.1256,
|
| 800 |
"step": 1280
|
| 801 |
},
|
| 802 |
{
|
| 803 |
"epoch": 2.06,
|
| 804 |
"learning_rate": 9.696485623003195e-05,
|
| 805 |
+
"loss": 0.0431,
|
| 806 |
"step": 1290
|
| 807 |
},
|
| 808 |
{
|
| 809 |
"epoch": 2.08,
|
| 810 |
"learning_rate": 9.616613418530351e-05,
|
| 811 |
+
"loss": 0.0543,
|
| 812 |
"step": 1300
|
| 813 |
},
|
| 814 |
{
|
| 815 |
"epoch": 2.09,
|
| 816 |
"learning_rate": 9.536741214057508e-05,
|
| 817 |
+
"loss": 0.0245,
|
| 818 |
"step": 1310
|
| 819 |
},
|
| 820 |
{
|
| 821 |
"epoch": 2.11,
|
| 822 |
"learning_rate": 9.456869009584664e-05,
|
| 823 |
+
"loss": 0.0213,
|
| 824 |
"step": 1320
|
| 825 |
},
|
| 826 |
{
|
| 827 |
"epoch": 2.12,
|
| 828 |
"learning_rate": 9.376996805111822e-05,
|
| 829 |
+
"loss": 0.0107,
|
| 830 |
"step": 1330
|
| 831 |
},
|
| 832 |
{
|
| 833 |
"epoch": 2.14,
|
| 834 |
"learning_rate": 9.297124600638978e-05,
|
| 835 |
+
"loss": 0.0272,
|
| 836 |
"step": 1340
|
| 837 |
},
|
| 838 |
{
|
| 839 |
"epoch": 2.16,
|
| 840 |
"learning_rate": 9.217252396166136e-05,
|
| 841 |
+
"loss": 0.0135,
|
| 842 |
"step": 1350
|
| 843 |
},
|
| 844 |
{
|
| 845 |
"epoch": 2.17,
|
| 846 |
"learning_rate": 9.137380191693292e-05,
|
| 847 |
+
"loss": 0.0221,
|
| 848 |
"step": 1360
|
| 849 |
},
|
| 850 |
{
|
| 851 |
"epoch": 2.19,
|
| 852 |
"learning_rate": 9.057507987220448e-05,
|
| 853 |
+
"loss": 0.05,
|
| 854 |
"step": 1370
|
| 855 |
},
|
| 856 |
{
|
| 857 |
"epoch": 2.2,
|
| 858 |
"learning_rate": 8.977635782747604e-05,
|
| 859 |
+
"loss": 0.0058,
|
| 860 |
"step": 1380
|
| 861 |
},
|
| 862 |
{
|
| 863 |
"epoch": 2.22,
|
| 864 |
"learning_rate": 8.89776357827476e-05,
|
| 865 |
+
"loss": 0.0384,
|
| 866 |
"step": 1390
|
| 867 |
},
|
| 868 |
{
|
| 869 |
"epoch": 2.24,
|
| 870 |
"learning_rate": 8.817891373801918e-05,
|
| 871 |
+
"loss": 0.0097,
|
| 872 |
"step": 1400
|
| 873 |
},
|
| 874 |
{
|
| 875 |
"epoch": 2.25,
|
| 876 |
"learning_rate": 8.738019169329074e-05,
|
| 877 |
+
"loss": 0.0331,
|
| 878 |
"step": 1410
|
| 879 |
},
|
| 880 |
{
|
| 881 |
"epoch": 2.27,
|
| 882 |
"learning_rate": 8.658146964856231e-05,
|
| 883 |
+
"loss": 0.0187,
|
| 884 |
"step": 1420
|
| 885 |
},
|
| 886 |
{
|
| 887 |
"epoch": 2.28,
|
| 888 |
"learning_rate": 8.578274760383387e-05,
|
| 889 |
+
"loss": 0.0021,
|
| 890 |
"step": 1430
|
| 891 |
},
|
| 892 |
{
|
| 893 |
"epoch": 2.3,
|
| 894 |
"learning_rate": 8.498402555910544e-05,
|
| 895 |
+
"loss": 0.0151,
|
| 896 |
"step": 1440
|
| 897 |
},
|
| 898 |
{
|
| 899 |
"epoch": 2.32,
|
| 900 |
"learning_rate": 8.4185303514377e-05,
|
| 901 |
+
"loss": 0.0192,
|
| 902 |
"step": 1450
|
| 903 |
},
|
| 904 |
{
|
| 905 |
"epoch": 2.33,
|
| 906 |
"learning_rate": 8.338658146964856e-05,
|
| 907 |
+
"loss": 0.0235,
|
| 908 |
"step": 1460
|
| 909 |
},
|
| 910 |
{
|
| 911 |
"epoch": 2.35,
|
| 912 |
"learning_rate": 8.258785942492013e-05,
|
| 913 |
+
"loss": 0.0052,
|
| 914 |
"step": 1470
|
| 915 |
},
|
| 916 |
{
|
| 917 |
"epoch": 2.36,
|
| 918 |
"learning_rate": 8.17891373801917e-05,
|
| 919 |
+
"loss": 0.0443,
|
| 920 |
"step": 1480
|
| 921 |
},
|
| 922 |
{
|
| 923 |
"epoch": 2.38,
|
| 924 |
"learning_rate": 8.099041533546327e-05,
|
| 925 |
+
"loss": 0.0428,
|
| 926 |
"step": 1490
|
| 927 |
},
|
| 928 |
{
|
| 929 |
"epoch": 2.4,
|
| 930 |
"learning_rate": 8.019169329073483e-05,
|
| 931 |
+
"loss": 0.0123,
|
| 932 |
"step": 1500
|
| 933 |
},
|
| 934 |
{
|
| 935 |
"epoch": 2.41,
|
| 936 |
"learning_rate": 7.939297124600639e-05,
|
| 937 |
+
"loss": 0.0225,
|
| 938 |
"step": 1510
|
| 939 |
},
|
| 940 |
{
|
| 941 |
"epoch": 2.43,
|
| 942 |
"learning_rate": 7.859424920127795e-05,
|
| 943 |
+
"loss": 0.0617,
|
| 944 |
"step": 1520
|
| 945 |
},
|
| 946 |
{
|
| 947 |
"epoch": 2.44,
|
| 948 |
"learning_rate": 7.779552715654951e-05,
|
| 949 |
+
"loss": 0.0085,
|
| 950 |
"step": 1530
|
| 951 |
},
|
| 952 |
{
|
| 953 |
"epoch": 2.46,
|
| 954 |
"learning_rate": 7.699680511182109e-05,
|
| 955 |
+
"loss": 0.0071,
|
| 956 |
"step": 1540
|
| 957 |
},
|
| 958 |
{
|
| 959 |
"epoch": 2.48,
|
| 960 |
"learning_rate": 7.619808306709265e-05,
|
| 961 |
+
"loss": 0.0059,
|
| 962 |
"step": 1550
|
| 963 |
},
|
| 964 |
{
|
| 965 |
"epoch": 2.49,
|
| 966 |
"learning_rate": 7.539936102236423e-05,
|
| 967 |
+
"loss": 0.0071,
|
| 968 |
"step": 1560
|
| 969 |
},
|
| 970 |
{
|
| 971 |
"epoch": 2.51,
|
| 972 |
"learning_rate": 7.460063897763579e-05,
|
| 973 |
+
"loss": 0.0042,
|
| 974 |
"step": 1570
|
| 975 |
},
|
| 976 |
{
|
| 977 |
"epoch": 2.52,
|
| 978 |
"learning_rate": 7.380191693290735e-05,
|
| 979 |
+
"loss": 0.0075,
|
| 980 |
"step": 1580
|
| 981 |
},
|
| 982 |
{
|
| 983 |
"epoch": 2.54,
|
| 984 |
"learning_rate": 7.300319488817891e-05,
|
| 985 |
+
"loss": 0.032,
|
| 986 |
"step": 1590
|
| 987 |
},
|
| 988 |
{
|
| 989 |
"epoch": 2.56,
|
| 990 |
"learning_rate": 7.220447284345049e-05,
|
| 991 |
+
"loss": 0.0176,
|
| 992 |
"step": 1600
|
| 993 |
},
|
| 994 |
{
|
| 995 |
"epoch": 2.57,
|
| 996 |
"learning_rate": 7.140575079872205e-05,
|
| 997 |
+
"loss": 0.0102,
|
| 998 |
"step": 1610
|
| 999 |
},
|
| 1000 |
{
|
| 1001 |
"epoch": 2.59,
|
| 1002 |
"learning_rate": 7.060702875399361e-05,
|
| 1003 |
+
"loss": 0.0246,
|
| 1004 |
"step": 1620
|
| 1005 |
},
|
| 1006 |
{
|
| 1007 |
"epoch": 2.6,
|
| 1008 |
"learning_rate": 6.980830670926518e-05,
|
| 1009 |
+
"loss": 0.0548,
|
| 1010 |
"step": 1630
|
| 1011 |
},
|
| 1012 |
{
|
| 1013 |
"epoch": 2.62,
|
| 1014 |
"learning_rate": 6.900958466453674e-05,
|
| 1015 |
+
"loss": 0.0279,
|
| 1016 |
"step": 1640
|
| 1017 |
},
|
| 1018 |
{
|
| 1019 |
"epoch": 2.64,
|
| 1020 |
"learning_rate": 6.821086261980832e-05,
|
| 1021 |
+
"loss": 0.0278,
|
| 1022 |
"step": 1650
|
| 1023 |
},
|
| 1024 |
{
|
| 1025 |
"epoch": 2.65,
|
| 1026 |
"learning_rate": 6.741214057507987e-05,
|
| 1027 |
+
"loss": 0.0189,
|
| 1028 |
"step": 1660
|
| 1029 |
},
|
| 1030 |
{
|
| 1031 |
"epoch": 2.67,
|
| 1032 |
"learning_rate": 6.661341853035144e-05,
|
| 1033 |
+
"loss": 0.0516,
|
| 1034 |
"step": 1670
|
| 1035 |
},
|
| 1036 |
{
|
| 1037 |
"epoch": 2.68,
|
| 1038 |
"learning_rate": 6.5814696485623e-05,
|
| 1039 |
+
"loss": 0.0119,
|
| 1040 |
"step": 1680
|
| 1041 |
},
|
| 1042 |
{
|
| 1043 |
"epoch": 2.7,
|
| 1044 |
"learning_rate": 6.501597444089457e-05,
|
| 1045 |
+
"loss": 0.0247,
|
| 1046 |
"step": 1690
|
| 1047 |
},
|
| 1048 |
{
|
| 1049 |
"epoch": 2.72,
|
| 1050 |
"learning_rate": 6.421725239616614e-05,
|
| 1051 |
+
"loss": 0.0216,
|
| 1052 |
"step": 1700
|
| 1053 |
},
|
| 1054 |
{
|
| 1055 |
"epoch": 2.73,
|
| 1056 |
"learning_rate": 6.34185303514377e-05,
|
| 1057 |
+
"loss": 0.0402,
|
| 1058 |
"step": 1710
|
| 1059 |
},
|
| 1060 |
{
|
| 1061 |
"epoch": 2.75,
|
| 1062 |
"learning_rate": 6.261980830670928e-05,
|
| 1063 |
+
"loss": 0.0335,
|
| 1064 |
"step": 1720
|
| 1065 |
},
|
| 1066 |
{
|
| 1067 |
"epoch": 2.76,
|
| 1068 |
"learning_rate": 6.182108626198084e-05,
|
| 1069 |
+
"loss": 0.0109,
|
| 1070 |
"step": 1730
|
| 1071 |
},
|
| 1072 |
{
|
| 1073 |
"epoch": 2.78,
|
| 1074 |
"learning_rate": 6.1022364217252406e-05,
|
| 1075 |
+
"loss": 0.0031,
|
| 1076 |
"step": 1740
|
| 1077 |
},
|
| 1078 |
{
|
| 1079 |
"epoch": 2.8,
|
| 1080 |
"learning_rate": 6.022364217252396e-05,
|
| 1081 |
+
"loss": 0.0062,
|
| 1082 |
"step": 1750
|
| 1083 |
},
|
| 1084 |
{
|
| 1085 |
"epoch": 2.81,
|
| 1086 |
"learning_rate": 5.942492012779552e-05,
|
| 1087 |
+
"loss": 0.0046,
|
| 1088 |
"step": 1760
|
| 1089 |
},
|
| 1090 |
{
|
| 1091 |
"epoch": 2.83,
|
| 1092 |
"learning_rate": 5.86261980830671e-05,
|
| 1093 |
+
"loss": 0.0268,
|
| 1094 |
"step": 1770
|
| 1095 |
},
|
| 1096 |
{
|
| 1097 |
"epoch": 2.84,
|
| 1098 |
"learning_rate": 5.782747603833866e-05,
|
| 1099 |
+
"loss": 0.0225,
|
| 1100 |
"step": 1780
|
| 1101 |
},
|
| 1102 |
{
|
| 1103 |
"epoch": 2.86,
|
| 1104 |
"learning_rate": 5.702875399361023e-05,
|
| 1105 |
+
"loss": 0.0159,
|
| 1106 |
"step": 1790
|
| 1107 |
},
|
| 1108 |
{
|
| 1109 |
"epoch": 2.88,
|
| 1110 |
"learning_rate": 5.623003194888179e-05,
|
| 1111 |
+
"loss": 0.0152,
|
| 1112 |
"step": 1800
|
| 1113 |
},
|
| 1114 |
{
|
| 1115 |
"epoch": 2.89,
|
| 1116 |
"learning_rate": 5.543130990415336e-05,
|
| 1117 |
+
"loss": 0.0338,
|
| 1118 |
"step": 1810
|
| 1119 |
},
|
| 1120 |
{
|
| 1121 |
"epoch": 2.91,
|
| 1122 |
"learning_rate": 5.4632587859424925e-05,
|
| 1123 |
+
"loss": 0.0382,
|
| 1124 |
"step": 1820
|
| 1125 |
},
|
| 1126 |
{
|
| 1127 |
"epoch": 2.92,
|
| 1128 |
"learning_rate": 5.383386581469649e-05,
|
| 1129 |
+
"loss": 0.008,
|
| 1130 |
"step": 1830
|
| 1131 |
},
|
| 1132 |
{
|
| 1133 |
"epoch": 2.94,
|
| 1134 |
"learning_rate": 5.3035143769968054e-05,
|
| 1135 |
+
"loss": 0.004,
|
| 1136 |
"step": 1840
|
| 1137 |
},
|
| 1138 |
{
|
| 1139 |
"epoch": 2.96,
|
| 1140 |
"learning_rate": 5.2236421725239616e-05,
|
| 1141 |
+
"loss": 0.0023,
|
| 1142 |
"step": 1850
|
| 1143 |
},
|
| 1144 |
{
|
| 1145 |
"epoch": 2.97,
|
| 1146 |
"learning_rate": 5.1437699680511184e-05,
|
| 1147 |
+
"loss": 0.0166,
|
| 1148 |
"step": 1860
|
| 1149 |
},
|
| 1150 |
{
|
| 1151 |
"epoch": 2.99,
|
| 1152 |
"learning_rate": 5.0638977635782745e-05,
|
| 1153 |
+
"loss": 0.0349,
|
| 1154 |
"step": 1870
|
| 1155 |
},
|
| 1156 |
{
|
| 1157 |
"epoch": 3.0,
|
| 1158 |
+
"eval_accuracy": 0.8704663212435233,
|
| 1159 |
+
"eval_f1": 0.8714524424651604,
|
| 1160 |
+
"eval_loss": 0.46994760632514954,
|
| 1161 |
+
"eval_precision": 0.8770991838952296,
|
| 1162 |
+
"eval_recall": 0.8704663212435233,
|
| 1163 |
+
"eval_runtime": 1.3201,
|
| 1164 |
+
"eval_samples_per_second": 146.2,
|
| 1165 |
+
"eval_steps_per_second": 9.848,
|
| 1166 |
"step": 1878
|
| 1167 |
},
|
| 1168 |
{
|
| 1169 |
"epoch": 3.0,
|
| 1170 |
"learning_rate": 4.984025559105431e-05,
|
| 1171 |
+
"loss": 0.0296,
|
| 1172 |
"step": 1880
|
| 1173 |
},
|
| 1174 |
{
|
| 1175 |
"epoch": 3.02,
|
| 1176 |
"learning_rate": 4.904153354632588e-05,
|
| 1177 |
+
"loss": 0.0026,
|
| 1178 |
"step": 1890
|
| 1179 |
},
|
| 1180 |
{
|
| 1181 |
"epoch": 3.04,
|
| 1182 |
"learning_rate": 4.824281150159744e-05,
|
| 1183 |
+
"loss": 0.0011,
|
| 1184 |
"step": 1900
|
| 1185 |
},
|
| 1186 |
{
|
| 1187 |
"epoch": 3.05,
|
| 1188 |
"learning_rate": 4.744408945686901e-05,
|
| 1189 |
+
"loss": 0.0177,
|
| 1190 |
"step": 1910
|
| 1191 |
},
|
| 1192 |
{
|
| 1193 |
"epoch": 3.07,
|
| 1194 |
"learning_rate": 4.664536741214058e-05,
|
| 1195 |
+
"loss": 0.0018,
|
| 1196 |
"step": 1920
|
| 1197 |
},
|
| 1198 |
{
|
| 1199 |
"epoch": 3.08,
|
| 1200 |
"learning_rate": 4.584664536741215e-05,
|
| 1201 |
+
"loss": 0.0026,
|
| 1202 |
"step": 1930
|
| 1203 |
},
|
| 1204 |
{
|
| 1205 |
"epoch": 3.1,
|
| 1206 |
"learning_rate": 4.504792332268371e-05,
|
| 1207 |
+
"loss": 0.0006,
|
| 1208 |
"step": 1940
|
| 1209 |
},
|
| 1210 |
{
|
| 1211 |
"epoch": 3.12,
|
| 1212 |
"learning_rate": 4.424920127795527e-05,
|
| 1213 |
+
"loss": 0.001,
|
| 1214 |
"step": 1950
|
| 1215 |
},
|
| 1216 |
{
|
| 1217 |
"epoch": 3.13,
|
| 1218 |
"learning_rate": 4.345047923322684e-05,
|
| 1219 |
+
"loss": 0.0037,
|
| 1220 |
"step": 1960
|
| 1221 |
},
|
| 1222 |
{
|
| 1223 |
"epoch": 3.15,
|
| 1224 |
"learning_rate": 4.265175718849841e-05,
|
| 1225 |
+
"loss": 0.0019,
|
| 1226 |
"step": 1970
|
| 1227 |
},
|
| 1228 |
{
|
| 1229 |
"epoch": 3.16,
|
| 1230 |
"learning_rate": 4.185303514376997e-05,
|
| 1231 |
+
"loss": 0.0017,
|
| 1232 |
"step": 1980
|
| 1233 |
},
|
| 1234 |
{
|
| 1235 |
"epoch": 3.18,
|
| 1236 |
"learning_rate": 4.1054313099041536e-05,
|
| 1237 |
+
"loss": 0.0008,
|
| 1238 |
"step": 1990
|
| 1239 |
},
|
| 1240 |
{
|
| 1241 |
"epoch": 3.19,
|
| 1242 |
"learning_rate": 4.0255591054313104e-05,
|
| 1243 |
+
"loss": 0.0219,
|
| 1244 |
"step": 2000
|
| 1245 |
},
|
| 1246 |
{
|
| 1247 |
"epoch": 3.21,
|
| 1248 |
"learning_rate": 3.9456869009584666e-05,
|
| 1249 |
+
"loss": 0.0006,
|
| 1250 |
"step": 2010
|
| 1251 |
},
|
| 1252 |
{
|
| 1253 |
"epoch": 3.23,
|
| 1254 |
"learning_rate": 3.8658146964856234e-05,
|
| 1255 |
+
"loss": 0.0096,
|
| 1256 |
"step": 2020
|
| 1257 |
},
|
| 1258 |
{
|
| 1259 |
"epoch": 3.24,
|
| 1260 |
"learning_rate": 3.7859424920127795e-05,
|
| 1261 |
+
"loss": 0.0014,
|
| 1262 |
"step": 2030
|
| 1263 |
},
|
| 1264 |
{
|
| 1265 |
"epoch": 3.26,
|
| 1266 |
"learning_rate": 3.7060702875399364e-05,
|
| 1267 |
+
"loss": 0.0242,
|
| 1268 |
"step": 2040
|
| 1269 |
},
|
| 1270 |
{
|
| 1271 |
"epoch": 3.27,
|
| 1272 |
"learning_rate": 3.6261980830670925e-05,
|
| 1273 |
+
"loss": 0.0006,
|
| 1274 |
"step": 2050
|
| 1275 |
},
|
| 1276 |
{
|
| 1277 |
"epoch": 3.29,
|
| 1278 |
"learning_rate": 3.546325878594249e-05,
|
| 1279 |
+
"loss": 0.0012,
|
| 1280 |
"step": 2060
|
| 1281 |
},
|
| 1282 |
{
|
| 1283 |
"epoch": 3.31,
|
| 1284 |
"learning_rate": 3.466453674121406e-05,
|
| 1285 |
+
"loss": 0.0008,
|
| 1286 |
"step": 2070
|
| 1287 |
},
|
| 1288 |
{
|
| 1289 |
"epoch": 3.32,
|
| 1290 |
"learning_rate": 3.386581469648562e-05,
|
| 1291 |
+
"loss": 0.0009,
|
| 1292 |
"step": 2080
|
| 1293 |
},
|
| 1294 |
{
|
| 1295 |
"epoch": 3.34,
|
| 1296 |
"learning_rate": 3.306709265175719e-05,
|
| 1297 |
+
"loss": 0.0013,
|
| 1298 |
"step": 2090
|
| 1299 |
},
|
| 1300 |
{
|
| 1301 |
"epoch": 3.35,
|
| 1302 |
"learning_rate": 3.226837060702875e-05,
|
| 1303 |
+
"loss": 0.0026,
|
| 1304 |
"step": 2100
|
| 1305 |
},
|
| 1306 |
{
|
| 1307 |
"epoch": 3.37,
|
| 1308 |
"learning_rate": 3.146964856230032e-05,
|
| 1309 |
+
"loss": 0.0069,
|
| 1310 |
"step": 2110
|
| 1311 |
},
|
| 1312 |
{
|
| 1313 |
"epoch": 3.39,
|
| 1314 |
"learning_rate": 3.067092651757188e-05,
|
| 1315 |
+
"loss": 0.0004,
|
| 1316 |
"step": 2120
|
| 1317 |
},
|
| 1318 |
{
|
| 1319 |
"epoch": 3.4,
|
| 1320 |
"learning_rate": 2.987220447284345e-05,
|
| 1321 |
+
"loss": 0.0006,
|
| 1322 |
"step": 2130
|
| 1323 |
},
|
| 1324 |
{
|
| 1325 |
"epoch": 3.42,
|
| 1326 |
"learning_rate": 2.907348242811502e-05,
|
| 1327 |
+
"loss": 0.0007,
|
| 1328 |
"step": 2140
|
| 1329 |
},
|
| 1330 |
{
|
| 1331 |
"epoch": 3.43,
|
| 1332 |
"learning_rate": 2.8274760383386583e-05,
|
| 1333 |
+
"loss": 0.0011,
|
| 1334 |
"step": 2150
|
| 1335 |
},
|
| 1336 |
{
|
| 1337 |
"epoch": 3.45,
|
| 1338 |
"learning_rate": 2.747603833865815e-05,
|
| 1339 |
+
"loss": 0.0024,
|
| 1340 |
"step": 2160
|
| 1341 |
},
|
| 1342 |
{
|
| 1343 |
"epoch": 3.47,
|
| 1344 |
+
"learning_rate": 2.6677316293929716e-05,
|
| 1345 |
+
"loss": 0.0005,
|
| 1346 |
"step": 2170
|
| 1347 |
},
|
| 1348 |
{
|
| 1349 |
"epoch": 3.48,
|
| 1350 |
+
"learning_rate": 2.5878594249201278e-05,
|
| 1351 |
+
"loss": 0.0011,
|
| 1352 |
"step": 2180
|
| 1353 |
},
|
| 1354 |
{
|
| 1355 |
"epoch": 3.5,
|
| 1356 |
+
"learning_rate": 2.5079872204472842e-05,
|
| 1357 |
+
"loss": 0.0007,
|
| 1358 |
"step": 2190
|
| 1359 |
},
|
| 1360 |
{
|
| 1361 |
"epoch": 3.51,
|
| 1362 |
+
"learning_rate": 2.428115015974441e-05,
|
| 1363 |
+
"loss": 0.0006,
|
| 1364 |
"step": 2200
|
| 1365 |
},
|
| 1366 |
{
|
| 1367 |
"epoch": 3.53,
|
| 1368 |
+
"learning_rate": 2.3482428115015975e-05,
|
| 1369 |
+
"loss": 0.0009,
|
| 1370 |
"step": 2210
|
| 1371 |
},
|
| 1372 |
{
|
| 1373 |
"epoch": 3.55,
|
| 1374 |
+
"learning_rate": 2.268370607028754e-05,
|
| 1375 |
+
"loss": 0.001,
|
| 1376 |
"step": 2220
|
| 1377 |
},
|
| 1378 |
{
|
| 1379 |
"epoch": 3.56,
|
| 1380 |
+
"learning_rate": 2.188498402555911e-05,
|
| 1381 |
+
"loss": 0.0004,
|
| 1382 |
"step": 2230
|
| 1383 |
},
|
| 1384 |
{
|
| 1385 |
"epoch": 3.58,
|
| 1386 |
+
"learning_rate": 2.108626198083067e-05,
|
| 1387 |
+
"loss": 0.0009,
|
| 1388 |
"step": 2240
|
| 1389 |
},
|
| 1390 |
{
|
| 1391 |
"epoch": 3.59,
|
| 1392 |
+
"learning_rate": 2.0287539936102238e-05,
|
| 1393 |
+
"loss": 0.0006,
|
| 1394 |
"step": 2250
|
| 1395 |
},
|
| 1396 |
{
|
| 1397 |
"epoch": 3.61,
|
| 1398 |
+
"learning_rate": 1.9488817891373803e-05,
|
| 1399 |
+
"loss": 0.0304,
|
| 1400 |
"step": 2260
|
| 1401 |
},
|
| 1402 |
{
|
| 1403 |
"epoch": 3.63,
|
| 1404 |
+
"learning_rate": 1.869009584664537e-05,
|
| 1405 |
+
"loss": 0.009,
|
| 1406 |
"step": 2270
|
| 1407 |
},
|
| 1408 |
{
|
| 1409 |
"epoch": 3.64,
|
| 1410 |
+
"learning_rate": 1.7891373801916932e-05,
|
| 1411 |
+
"loss": 0.0146,
|
| 1412 |
"step": 2280
|
| 1413 |
},
|
| 1414 |
{
|
| 1415 |
"epoch": 3.66,
|
| 1416 |
+
"learning_rate": 1.70926517571885e-05,
|
| 1417 |
+
"loss": 0.0007,
|
| 1418 |
"step": 2290
|
| 1419 |
},
|
| 1420 |
{
|
| 1421 |
"epoch": 3.67,
|
| 1422 |
+
"learning_rate": 1.6293929712460065e-05,
|
| 1423 |
+
"loss": 0.0008,
|
| 1424 |
"step": 2300
|
| 1425 |
},
|
| 1426 |
{
|
| 1427 |
"epoch": 3.69,
|
| 1428 |
+
"learning_rate": 1.549520766773163e-05,
|
| 1429 |
+
"loss": 0.0004,
|
| 1430 |
"step": 2310
|
| 1431 |
},
|
| 1432 |
{
|
| 1433 |
"epoch": 3.71,
|
| 1434 |
+
"learning_rate": 1.4696485623003195e-05,
|
| 1435 |
+
"loss": 0.0005,
|
| 1436 |
"step": 2320
|
| 1437 |
},
|
| 1438 |
{
|
| 1439 |
"epoch": 3.72,
|
| 1440 |
+
"learning_rate": 1.3897763578274761e-05,
|
| 1441 |
+
"loss": 0.001,
|
| 1442 |
"step": 2330
|
| 1443 |
},
|
| 1444 |
{
|
| 1445 |
"epoch": 3.74,
|
| 1446 |
+
"learning_rate": 1.3099041533546328e-05,
|
| 1447 |
+
"loss": 0.0004,
|
| 1448 |
"step": 2340
|
| 1449 |
},
|
| 1450 |
{
|
| 1451 |
"epoch": 3.75,
|
| 1452 |
+
"learning_rate": 1.2300319488817893e-05,
|
| 1453 |
+
"loss": 0.0004,
|
| 1454 |
"step": 2350
|
| 1455 |
},
|
| 1456 |
{
|
| 1457 |
"epoch": 3.77,
|
| 1458 |
+
"learning_rate": 1.1501597444089457e-05,
|
| 1459 |
+
"loss": 0.0274,
|
| 1460 |
"step": 2360
|
| 1461 |
},
|
| 1462 |
{
|
| 1463 |
"epoch": 3.79,
|
| 1464 |
+
"learning_rate": 1.0702875399361024e-05,
|
| 1465 |
+
"loss": 0.0006,
|
| 1466 |
"step": 2370
|
| 1467 |
},
|
| 1468 |
{
|
| 1469 |
"epoch": 3.8,
|
| 1470 |
+
"learning_rate": 9.904153354632589e-06,
|
| 1471 |
+
"loss": 0.0004,
|
| 1472 |
"step": 2380
|
| 1473 |
},
|
| 1474 |
{
|
| 1475 |
"epoch": 3.82,
|
| 1476 |
+
"learning_rate": 9.105431309904154e-06,
|
| 1477 |
+
"loss": 0.0004,
|
| 1478 |
"step": 2390
|
| 1479 |
},
|
| 1480 |
{
|
| 1481 |
"epoch": 3.83,
|
| 1482 |
+
"learning_rate": 8.306709265175718e-06,
|
| 1483 |
+
"loss": 0.0005,
|
| 1484 |
"step": 2400
|
| 1485 |
},
|
| 1486 |
{
|
| 1487 |
"epoch": 3.85,
|
| 1488 |
+
"learning_rate": 7.507987220447285e-06,
|
| 1489 |
+
"loss": 0.0086,
|
| 1490 |
"step": 2410
|
| 1491 |
},
|
| 1492 |
{
|
| 1493 |
"epoch": 3.87,
|
| 1494 |
+
"learning_rate": 6.70926517571885e-06,
|
| 1495 |
+
"loss": 0.035,
|
| 1496 |
"step": 2420
|
| 1497 |
},
|
| 1498 |
{
|
| 1499 |
"epoch": 3.88,
|
| 1500 |
+
"learning_rate": 5.910543130990415e-06,
|
| 1501 |
+
"loss": 0.0005,
|
| 1502 |
"step": 2430
|
| 1503 |
},
|
| 1504 |
{
|
| 1505 |
"epoch": 3.9,
|
| 1506 |
+
"learning_rate": 5.111821086261981e-06,
|
| 1507 |
+
"loss": 0.0004,
|
| 1508 |
"step": 2440
|
| 1509 |
},
|
| 1510 |
{
|
| 1511 |
"epoch": 3.91,
|
| 1512 |
+
"learning_rate": 4.3130990415335465e-06,
|
| 1513 |
+
"loss": 0.0005,
|
| 1514 |
"step": 2450
|
| 1515 |
},
|
| 1516 |
{
|
| 1517 |
"epoch": 3.93,
|
| 1518 |
+
"learning_rate": 3.5143769968051118e-06,
|
| 1519 |
+
"loss": 0.0008,
|
| 1520 |
"step": 2460
|
| 1521 |
},
|
| 1522 |
{
|
| 1523 |
"epoch": 3.95,
|
| 1524 |
+
"learning_rate": 2.7156549520766774e-06,
|
| 1525 |
+
"loss": 0.0014,
|
| 1526 |
"step": 2470
|
| 1527 |
},
|
| 1528 |
{
|
| 1529 |
"epoch": 3.96,
|
| 1530 |
+
"learning_rate": 1.9169329073482426e-06,
|
| 1531 |
+
"loss": 0.0006,
|
| 1532 |
"step": 2480
|
| 1533 |
},
|
| 1534 |
{
|
| 1535 |
"epoch": 3.98,
|
| 1536 |
+
"learning_rate": 1.1182108626198083e-06,
|
| 1537 |
+
"loss": 0.0004,
|
| 1538 |
"step": 2490
|
| 1539 |
},
|
| 1540 |
{
|
| 1541 |
"epoch": 3.99,
|
| 1542 |
+
"learning_rate": 3.194888178913738e-07,
|
| 1543 |
+
"loss": 0.0005,
|
| 1544 |
"step": 2500
|
| 1545 |
},
|
| 1546 |
{
|
| 1547 |
"epoch": 4.0,
|
| 1548 |
+
"eval_accuracy": 0.8756476683937824,
|
| 1549 |
+
"eval_f1": 0.8760275990745605,
|
| 1550 |
+
"eval_loss": 0.4526107907295227,
|
| 1551 |
+
"eval_precision": 0.8781157606627441,
|
| 1552 |
+
"eval_recall": 0.8756476683937824,
|
| 1553 |
+
"eval_runtime": 1.3211,
|
| 1554 |
+
"eval_samples_per_second": 146.09,
|
| 1555 |
+
"eval_steps_per_second": 9.84,
|
| 1556 |
"step": 2504
|
| 1557 |
},
|
| 1558 |
{
|
| 1559 |
"epoch": 4.0,
|
| 1560 |
"step": 2504,
|
| 1561 |
"total_flos": 3.104468219559813e+18,
|
| 1562 |
+
"train_loss": 0.06904298590738683,
|
| 1563 |
+
"train_runtime": 834.8541,
|
| 1564 |
+
"train_samples_per_second": 47.984,
|
| 1565 |
+
"train_steps_per_second": 2.999
|
| 1566 |
}
|
| 1567 |
],
|
| 1568 |
"max_steps": 2504,
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3899
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:abe4d4eb25af72f410b3adbbfaeb3fc73806e42f23d9f0d9a2eedb528d41c1d6
|
| 3 |
size 3899
|