Training in progress, epoch 1
Browse files- all_results.json +5 -5
- pytorch_model.bin +1 -1
- runs/Oct27_21-22-14_EE4E077/1698412937.0288403/events.out.tfevents.1698412937.EE4E077.103115.4 +3 -0
- runs/Oct27_21-22-14_EE4E077/events.out.tfevents.1698412937.EE4E077.103115.3 +3 -0
- train_results.json +5 -6
- trainer_state.json +132 -2256
- training_args.bin +1 -1
all_results.json
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
"eval_accuracy": 0.794973544973545,
|
| 4 |
"eval_f1": 0.7911822260194087,
|
| 5 |
"eval_loss": 0.8158386945724487,
|
|
@@ -9,8 +9,8 @@
|
|
| 9 |
"eval_samples_per_second": 170.604,
|
| 10 |
"eval_steps_per_second": 10.719,
|
| 11 |
"total_flos": 4.65670232933972e+18,
|
| 12 |
-
"train_loss": 0.
|
| 13 |
-
"train_runtime":
|
| 14 |
-
"train_samples_per_second":
|
| 15 |
-
"train_steps_per_second":
|
| 16 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 4.0,
|
| 3 |
"eval_accuracy": 0.794973544973545,
|
| 4 |
"eval_f1": 0.7911822260194087,
|
| 5 |
"eval_loss": 0.8158386945724487,
|
|
|
|
| 9 |
"eval_samples_per_second": 170.604,
|
| 10 |
"eval_steps_per_second": 10.719,
|
| 11 |
"total_flos": 4.65670232933972e+18,
|
| 12 |
+
"train_loss": 0.2716429328141264,
|
| 13 |
+
"train_runtime": 832.7663,
|
| 14 |
+
"train_samples_per_second": 48.105,
|
| 15 |
+
"train_steps_per_second": 3.007
|
| 16 |
}
|
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 sha256:7d79839abeb277b7780036a9e3dea1bbe1251ea27e5404b8a7c432b58f121393
|
| 3 |
size 343284077
|
runs/Oct27_21-22-14_EE4E077/1698412937.0288403/events.out.tfevents.1698412937.EE4E077.103115.4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:98e5db06d2ba6108ae9786a2439529560fe99b41e23a1665e443f7b4fbc5dbcd
|
| 3 |
+
size 5919
|
runs/Oct27_21-22-14_EE4E077/events.out.tfevents.1698412937.EE4E077.103115.3
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:702ebd3f670ed964a41a1eb061ce5d0424bf6cb5fe615e5c414664aad4ca6a10
|
| 3 |
+
size 5660
|
train_results.json
CHANGED
|
@@ -1,8 +1,7 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"
|
| 4 |
-
"
|
| 5 |
-
"
|
| 6 |
-
"
|
| 7 |
-
"train_steps_per_second": 2.975
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 4.0,
|
| 3 |
+
"train_loss": 0.2716429328141264,
|
| 4 |
+
"train_runtime": 832.7663,
|
| 5 |
+
"train_samples_per_second": 48.105,
|
| 6 |
+
"train_steps_per_second": 3.007
|
|
|
|
| 7 |
}
|
trainer_state.json
CHANGED
|
@@ -1,2347 +1,223 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 0.
|
| 3 |
-
"best_model_checkpoint": "./vit-focal-skin/checkpoint-
|
| 4 |
-
"epoch":
|
| 5 |
-
"global_step":
|
| 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": 0.0001994675186368477,
|
| 13 |
-
"loss": 0.2668,
|
| 14 |
-
"step": 10
|
| 15 |
-
},
|
| 16 |
-
{
|
| 17 |
-
"epoch": 0.03,
|
| 18 |
-
"learning_rate": 0.00019893503727369543,
|
| 19 |
-
"loss": 0.3261,
|
| 20 |
-
"step": 20
|
| 21 |
-
},
|
| 22 |
-
{
|
| 23 |
-
"epoch": 0.05,
|
| 24 |
-
"learning_rate": 0.00019840255591054313,
|
| 25 |
-
"loss": 0.1579,
|
| 26 |
-
"step": 30
|
| 27 |
-
},
|
| 28 |
-
{
|
| 29 |
-
"epoch": 0.06,
|
| 30 |
-
"learning_rate": 0.00019787007454739086,
|
| 31 |
-
"loss": 0.256,
|
| 32 |
-
"step": 40
|
| 33 |
-
},
|
| 34 |
-
{
|
| 35 |
-
"epoch": 0.08,
|
| 36 |
-
"learning_rate": 0.00019733759318423856,
|
| 37 |
-
"loss": 0.2145,
|
| 38 |
-
"step": 50
|
| 39 |
-
},
|
| 40 |
-
{
|
| 41 |
-
"epoch": 0.1,
|
| 42 |
-
"learning_rate": 0.00019680511182108628,
|
| 43 |
-
"loss": 0.2802,
|
| 44 |
-
"step": 60
|
| 45 |
-
},
|
| 46 |
-
{
|
| 47 |
-
"epoch": 0.11,
|
| 48 |
-
"learning_rate": 0.00019627263045793398,
|
| 49 |
-
"loss": 0.3481,
|
| 50 |
-
"step": 70
|
| 51 |
-
},
|
| 52 |
-
{
|
| 53 |
-
"epoch": 0.13,
|
| 54 |
-
"learning_rate": 0.00019574014909478168,
|
| 55 |
-
"loss": 0.2614,
|
| 56 |
-
"step": 80
|
| 57 |
-
},
|
| 58 |
-
{
|
| 59 |
-
"epoch": 0.14,
|
| 60 |
-
"learning_rate": 0.0001952076677316294,
|
| 61 |
-
"loss": 0.3515,
|
| 62 |
-
"step": 90
|
| 63 |
-
},
|
| 64 |
{
|
| 65 |
"epoch": 0.16,
|
| 66 |
-
"learning_rate": 0.0001946751863684771,
|
| 67 |
-
"loss": 0.2308,
|
| 68 |
-
"step": 100
|
| 69 |
-
},
|
| 70 |
-
{
|
| 71 |
-
"epoch": 0.18,
|
| 72 |
-
"learning_rate": 0.00019414270500532483,
|
| 73 |
-
"loss": 0.3011,
|
| 74 |
-
"step": 110
|
| 75 |
-
},
|
| 76 |
-
{
|
| 77 |
-
"epoch": 0.19,
|
| 78 |
-
"learning_rate": 0.00019361022364217253,
|
| 79 |
-
"loss": 0.21,
|
| 80 |
-
"step": 120
|
| 81 |
-
},
|
| 82 |
-
{
|
| 83 |
-
"epoch": 0.21,
|
| 84 |
-
"learning_rate": 0.00019307774227902025,
|
| 85 |
-
"loss": 0.1883,
|
| 86 |
-
"step": 130
|
| 87 |
-
},
|
| 88 |
-
{
|
| 89 |
-
"epoch": 0.22,
|
| 90 |
-
"learning_rate": 0.00019254526091586795,
|
| 91 |
-
"loss": 0.1768,
|
| 92 |
-
"step": 140
|
| 93 |
-
},
|
| 94 |
-
{
|
| 95 |
-
"epoch": 0.24,
|
| 96 |
"learning_rate": 0.00019201277955271565,
|
| 97 |
-
"loss": 0.
|
| 98 |
-
"step":
|
| 99 |
-
},
|
| 100 |
-
{
|
| 101 |
-
"epoch": 0.26,
|
| 102 |
-
"learning_rate": 0.00019148029818956338,
|
| 103 |
-
"loss": 0.2239,
|
| 104 |
-
"step": 160
|
| 105 |
-
},
|
| 106 |
-
{
|
| 107 |
-
"epoch": 0.27,
|
| 108 |
-
"learning_rate": 0.00019094781682641108,
|
| 109 |
-
"loss": 0.1363,
|
| 110 |
-
"step": 170
|
| 111 |
-
},
|
| 112 |
-
{
|
| 113 |
-
"epoch": 0.29,
|
| 114 |
-
"learning_rate": 0.0001904153354632588,
|
| 115 |
-
"loss": 0.2123,
|
| 116 |
-
"step": 180
|
| 117 |
-
},
|
| 118 |
-
{
|
| 119 |
-
"epoch": 0.3,
|
| 120 |
-
"learning_rate": 0.0001898828541001065,
|
| 121 |
-
"loss": 0.1508,
|
| 122 |
-
"step": 190
|
| 123 |
},
|
| 124 |
{
|
| 125 |
"epoch": 0.32,
|
| 126 |
-
"learning_rate": 0.
|
| 127 |
-
"loss": 0.
|
| 128 |
"step": 200
|
| 129 |
},
|
| 130 |
-
{
|
| 131 |
-
"epoch": 0.34,
|
| 132 |
-
"learning_rate": 0.00018881789137380192,
|
| 133 |
-
"loss": 0.244,
|
| 134 |
-
"step": 210
|
| 135 |
-
},
|
| 136 |
-
{
|
| 137 |
-
"epoch": 0.35,
|
| 138 |
-
"learning_rate": 0.00018828541001064962,
|
| 139 |
-
"loss": 0.2445,
|
| 140 |
-
"step": 220
|
| 141 |
-
},
|
| 142 |
-
{
|
| 143 |
-
"epoch": 0.37,
|
| 144 |
-
"learning_rate": 0.00018775292864749735,
|
| 145 |
-
"loss": 0.211,
|
| 146 |
-
"step": 230
|
| 147 |
-
},
|
| 148 |
-
{
|
| 149 |
-
"epoch": 0.38,
|
| 150 |
-
"learning_rate": 0.00018722044728434505,
|
| 151 |
-
"loss": 0.1967,
|
| 152 |
-
"step": 240
|
| 153 |
-
},
|
| 154 |
-
{
|
| 155 |
-
"epoch": 0.4,
|
| 156 |
-
"learning_rate": 0.00018668796592119277,
|
| 157 |
-
"loss": 0.1522,
|
| 158 |
-
"step": 250
|
| 159 |
-
},
|
| 160 |
-
{
|
| 161 |
-
"epoch": 0.42,
|
| 162 |
-
"learning_rate": 0.00018615548455804047,
|
| 163 |
-
"loss": 0.155,
|
| 164 |
-
"step": 260
|
| 165 |
-
},
|
| 166 |
-
{
|
| 167 |
-
"epoch": 0.43,
|
| 168 |
-
"learning_rate": 0.0001856230031948882,
|
| 169 |
-
"loss": 0.3733,
|
| 170 |
-
"step": 270
|
| 171 |
-
},
|
| 172 |
-
{
|
| 173 |
-
"epoch": 0.45,
|
| 174 |
-
"learning_rate": 0.0001850905218317359,
|
| 175 |
-
"loss": 0.201,
|
| 176 |
-
"step": 280
|
| 177 |
-
},
|
| 178 |
-
{
|
| 179 |
-
"epoch": 0.46,
|
| 180 |
-
"learning_rate": 0.0001845580404685836,
|
| 181 |
-
"loss": 0.2162,
|
| 182 |
-
"step": 290
|
| 183 |
-
},
|
| 184 |
{
|
| 185 |
"epoch": 0.48,
|
| 186 |
-
"learning_rate": 0.
|
| 187 |
-
"loss": 0.
|
| 188 |
"step": 300
|
| 189 |
},
|
| 190 |
-
{
|
| 191 |
-
"epoch": 0.5,
|
| 192 |
-
"learning_rate": 0.00018349307774227902,
|
| 193 |
-
"loss": 0.1749,
|
| 194 |
-
"step": 310
|
| 195 |
-
},
|
| 196 |
-
{
|
| 197 |
-
"epoch": 0.51,
|
| 198 |
-
"learning_rate": 0.00018296059637912674,
|
| 199 |
-
"loss": 0.1627,
|
| 200 |
-
"step": 320
|
| 201 |
-
},
|
| 202 |
-
{
|
| 203 |
-
"epoch": 0.53,
|
| 204 |
-
"learning_rate": 0.00018242811501597444,
|
| 205 |
-
"loss": 0.3643,
|
| 206 |
-
"step": 330
|
| 207 |
-
},
|
| 208 |
-
{
|
| 209 |
-
"epoch": 0.54,
|
| 210 |
-
"learning_rate": 0.00018189563365282217,
|
| 211 |
-
"loss": 0.3076,
|
| 212 |
-
"step": 340
|
| 213 |
-
},
|
| 214 |
-
{
|
| 215 |
-
"epoch": 0.56,
|
| 216 |
-
"learning_rate": 0.00018136315228966987,
|
| 217 |
-
"loss": 0.1679,
|
| 218 |
-
"step": 350
|
| 219 |
-
},
|
| 220 |
-
{
|
| 221 |
-
"epoch": 0.58,
|
| 222 |
-
"learning_rate": 0.00018083067092651756,
|
| 223 |
-
"loss": 0.2569,
|
| 224 |
-
"step": 360
|
| 225 |
-
},
|
| 226 |
-
{
|
| 227 |
-
"epoch": 0.59,
|
| 228 |
-
"learning_rate": 0.0001802981895633653,
|
| 229 |
-
"loss": 0.2105,
|
| 230 |
-
"step": 370
|
| 231 |
-
},
|
| 232 |
-
{
|
| 233 |
-
"epoch": 0.61,
|
| 234 |
-
"learning_rate": 0.00017976570820021302,
|
| 235 |
-
"loss": 0.2794,
|
| 236 |
-
"step": 380
|
| 237 |
-
},
|
| 238 |
-
{
|
| 239 |
-
"epoch": 0.62,
|
| 240 |
-
"learning_rate": 0.00017923322683706071,
|
| 241 |
-
"loss": 0.1752,
|
| 242 |
-
"step": 390
|
| 243 |
-
},
|
| 244 |
{
|
| 245 |
"epoch": 0.64,
|
| 246 |
-
"learning_rate": 0.
|
| 247 |
-
"loss": 0.
|
| 248 |
"step": 400
|
| 249 |
},
|
| 250 |
-
{
|
| 251 |
-
"epoch": 0.65,
|
| 252 |
-
"learning_rate": 0.00017816826411075614,
|
| 253 |
-
"loss": 0.2519,
|
| 254 |
-
"step": 410
|
| 255 |
-
},
|
| 256 |
-
{
|
| 257 |
-
"epoch": 0.67,
|
| 258 |
-
"learning_rate": 0.00017763578274760384,
|
| 259 |
-
"loss": 0.2182,
|
| 260 |
-
"step": 420
|
| 261 |
-
},
|
| 262 |
-
{
|
| 263 |
-
"epoch": 0.69,
|
| 264 |
-
"learning_rate": 0.00017710330138445154,
|
| 265 |
-
"loss": 0.2159,
|
| 266 |
-
"step": 430
|
| 267 |
-
},
|
| 268 |
-
{
|
| 269 |
-
"epoch": 0.7,
|
| 270 |
-
"learning_rate": 0.00017657082002129926,
|
| 271 |
-
"loss": 0.1561,
|
| 272 |
-
"step": 440
|
| 273 |
-
},
|
| 274 |
-
{
|
| 275 |
-
"epoch": 0.72,
|
| 276 |
-
"learning_rate": 0.000176038338658147,
|
| 277 |
-
"loss": 0.1129,
|
| 278 |
-
"step": 450
|
| 279 |
-
},
|
| 280 |
-
{
|
| 281 |
-
"epoch": 0.73,
|
| 282 |
-
"learning_rate": 0.00017550585729499469,
|
| 283 |
-
"loss": 0.1696,
|
| 284 |
-
"step": 460
|
| 285 |
-
},
|
| 286 |
-
{
|
| 287 |
-
"epoch": 0.75,
|
| 288 |
-
"learning_rate": 0.0001749733759318424,
|
| 289 |
-
"loss": 0.2304,
|
| 290 |
-
"step": 470
|
| 291 |
-
},
|
| 292 |
-
{
|
| 293 |
-
"epoch": 0.77,
|
| 294 |
-
"learning_rate": 0.0001744408945686901,
|
| 295 |
-
"loss": 0.2481,
|
| 296 |
-
"step": 480
|
| 297 |
-
},
|
| 298 |
-
{
|
| 299 |
-
"epoch": 0.78,
|
| 300 |
-
"learning_rate": 0.0001739084132055378,
|
| 301 |
-
"loss": 0.2593,
|
| 302 |
-
"step": 490
|
| 303 |
-
},
|
| 304 |
{
|
| 305 |
"epoch": 0.8,
|
| 306 |
-
"learning_rate": 0.
|
| 307 |
-
"loss": 0.
|
| 308 |
"step": 500
|
| 309 |
},
|
| 310 |
-
{
|
| 311 |
-
"epoch": 0.81,
|
| 312 |
-
"learning_rate": 0.00017284345047923323,
|
| 313 |
-
"loss": 0.2631,
|
| 314 |
-
"step": 510
|
| 315 |
-
},
|
| 316 |
-
{
|
| 317 |
-
"epoch": 0.83,
|
| 318 |
-
"learning_rate": 0.00017231096911608096,
|
| 319 |
-
"loss": 0.1654,
|
| 320 |
-
"step": 520
|
| 321 |
-
},
|
| 322 |
-
{
|
| 323 |
-
"epoch": 0.85,
|
| 324 |
-
"learning_rate": 0.00017177848775292866,
|
| 325 |
-
"loss": 0.147,
|
| 326 |
-
"step": 530
|
| 327 |
-
},
|
| 328 |
-
{
|
| 329 |
-
"epoch": 0.86,
|
| 330 |
-
"learning_rate": 0.00017124600638977638,
|
| 331 |
-
"loss": 0.2521,
|
| 332 |
-
"step": 540
|
| 333 |
-
},
|
| 334 |
-
{
|
| 335 |
-
"epoch": 0.88,
|
| 336 |
-
"learning_rate": 0.00017071352502662408,
|
| 337 |
-
"loss": 0.2311,
|
| 338 |
-
"step": 550
|
| 339 |
-
},
|
| 340 |
-
{
|
| 341 |
-
"epoch": 0.89,
|
| 342 |
-
"learning_rate": 0.00017018104366347178,
|
| 343 |
-
"loss": 0.1705,
|
| 344 |
-
"step": 560
|
| 345 |
-
},
|
| 346 |
-
{
|
| 347 |
-
"epoch": 0.91,
|
| 348 |
-
"learning_rate": 0.00016964856230031948,
|
| 349 |
-
"loss": 0.1595,
|
| 350 |
-
"step": 570
|
| 351 |
-
},
|
| 352 |
-
{
|
| 353 |
-
"epoch": 0.93,
|
| 354 |
-
"learning_rate": 0.0001691160809371672,
|
| 355 |
-
"loss": 0.1939,
|
| 356 |
-
"step": 580
|
| 357 |
-
},
|
| 358 |
-
{
|
| 359 |
-
"epoch": 0.94,
|
| 360 |
-
"learning_rate": 0.00016858359957401493,
|
| 361 |
-
"loss": 0.1967,
|
| 362 |
-
"step": 590
|
| 363 |
-
},
|
| 364 |
{
|
| 365 |
"epoch": 0.96,
|
| 366 |
-
"learning_rate": 0.
|
| 367 |
-
"loss": 0.
|
| 368 |
"step": 600
|
| 369 |
},
|
| 370 |
-
{
|
| 371 |
-
"epoch": 0.97,
|
| 372 |
-
"learning_rate": 0.00016751863684771035,
|
| 373 |
-
"loss": 0.302,
|
| 374 |
-
"step": 610
|
| 375 |
-
},
|
| 376 |
-
{
|
| 377 |
-
"epoch": 0.99,
|
| 378 |
-
"learning_rate": 0.00016698615548455805,
|
| 379 |
-
"loss": 0.1024,
|
| 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": 0.00016645367412140575,
|
| 397 |
-
"loss": 0.1753,
|
| 398 |
-
"step": 630
|
| 399 |
-
},
|
| 400 |
-
{
|
| 401 |
-
"epoch": 1.02,
|
| 402 |
-
"learning_rate": 0.00016592119275825345,
|
| 403 |
-
"loss": 0.1077,
|
| 404 |
-
"step": 640
|
| 405 |
-
},
|
| 406 |
-
{
|
| 407 |
-
"epoch": 1.04,
|
| 408 |
-
"learning_rate": 0.00016538871139510118,
|
| 409 |
-
"loss": 0.1503,
|
| 410 |
-
"step": 650
|
| 411 |
-
},
|
| 412 |
-
{
|
| 413 |
-
"epoch": 1.05,
|
| 414 |
-
"learning_rate": 0.0001648562300319489,
|
| 415 |
-
"loss": 0.1275,
|
| 416 |
-
"step": 660
|
| 417 |
-
},
|
| 418 |
-
{
|
| 419 |
-
"epoch": 1.07,
|
| 420 |
-
"learning_rate": 0.0001643237486687966,
|
| 421 |
-
"loss": 0.1194,
|
| 422 |
-
"step": 670
|
| 423 |
-
},
|
| 424 |
-
{
|
| 425 |
-
"epoch": 1.09,
|
| 426 |
-
"learning_rate": 0.00016379126730564433,
|
| 427 |
-
"loss": 0.1587,
|
| 428 |
-
"step": 680
|
| 429 |
-
},
|
| 430 |
-
{
|
| 431 |
-
"epoch": 1.1,
|
| 432 |
-
"learning_rate": 0.00016325878594249202,
|
| 433 |
-
"loss": 0.2023,
|
| 434 |
-
"step": 690
|
| 435 |
-
},
|
| 436 |
{
|
| 437 |
"epoch": 1.12,
|
| 438 |
-
"learning_rate": 0.
|
| 439 |
-
"loss": 0.
|
| 440 |
"step": 700
|
| 441 |
},
|
| 442 |
-
{
|
| 443 |
-
"epoch": 1.13,
|
| 444 |
-
"learning_rate": 0.00016219382321618745,
|
| 445 |
-
"loss": 0.1739,
|
| 446 |
-
"step": 710
|
| 447 |
-
},
|
| 448 |
-
{
|
| 449 |
-
"epoch": 1.15,
|
| 450 |
-
"learning_rate": 0.00016166134185303515,
|
| 451 |
-
"loss": 0.0996,
|
| 452 |
-
"step": 720
|
| 453 |
-
},
|
| 454 |
-
{
|
| 455 |
-
"epoch": 1.17,
|
| 456 |
-
"learning_rate": 0.00016112886048988287,
|
| 457 |
-
"loss": 0.1538,
|
| 458 |
-
"step": 730
|
| 459 |
-
},
|
| 460 |
-
{
|
| 461 |
-
"epoch": 1.18,
|
| 462 |
-
"learning_rate": 0.00016059637912673057,
|
| 463 |
-
"loss": 0.215,
|
| 464 |
-
"step": 740
|
| 465 |
-
},
|
| 466 |
-
{
|
| 467 |
-
"epoch": 1.2,
|
| 468 |
-
"learning_rate": 0.0001600638977635783,
|
| 469 |
-
"loss": 0.0868,
|
| 470 |
-
"step": 750
|
| 471 |
-
},
|
| 472 |
-
{
|
| 473 |
-
"epoch": 1.21,
|
| 474 |
-
"learning_rate": 0.000159531416400426,
|
| 475 |
-
"loss": 0.1199,
|
| 476 |
-
"step": 760
|
| 477 |
-
},
|
| 478 |
-
{
|
| 479 |
-
"epoch": 1.23,
|
| 480 |
-
"learning_rate": 0.0001589989350372737,
|
| 481 |
-
"loss": 0.0801,
|
| 482 |
-
"step": 770
|
| 483 |
-
},
|
| 484 |
-
{
|
| 485 |
-
"epoch": 1.25,
|
| 486 |
-
"learning_rate": 0.00015846645367412142,
|
| 487 |
-
"loss": 0.1025,
|
| 488 |
-
"step": 780
|
| 489 |
-
},
|
| 490 |
-
{
|
| 491 |
-
"epoch": 1.26,
|
| 492 |
-
"learning_rate": 0.00015793397231096912,
|
| 493 |
-
"loss": 0.1524,
|
| 494 |
-
"step": 790
|
| 495 |
-
},
|
| 496 |
{
|
| 497 |
"epoch": 1.28,
|
| 498 |
-
"learning_rate": 0.
|
| 499 |
-
"loss": 0.
|
| 500 |
"step": 800
|
| 501 |
},
|
| 502 |
-
{
|
| 503 |
-
"epoch": 1.29,
|
| 504 |
-
"learning_rate": 0.00015686900958466454,
|
| 505 |
-
"loss": 0.1643,
|
| 506 |
-
"step": 810
|
| 507 |
-
},
|
| 508 |
-
{
|
| 509 |
-
"epoch": 1.31,
|
| 510 |
-
"learning_rate": 0.00015633652822151227,
|
| 511 |
-
"loss": 0.114,
|
| 512 |
-
"step": 820
|
| 513 |
-
},
|
| 514 |
-
{
|
| 515 |
-
"epoch": 1.33,
|
| 516 |
-
"learning_rate": 0.00015580404685835997,
|
| 517 |
-
"loss": 0.1284,
|
| 518 |
-
"step": 830
|
| 519 |
-
},
|
| 520 |
-
{
|
| 521 |
-
"epoch": 1.34,
|
| 522 |
-
"learning_rate": 0.00015527156549520767,
|
| 523 |
-
"loss": 0.1069,
|
| 524 |
-
"step": 840
|
| 525 |
-
},
|
| 526 |
-
{
|
| 527 |
-
"epoch": 1.36,
|
| 528 |
-
"learning_rate": 0.0001547390841320554,
|
| 529 |
-
"loss": 0.0643,
|
| 530 |
-
"step": 850
|
| 531 |
-
},
|
| 532 |
-
{
|
| 533 |
-
"epoch": 1.37,
|
| 534 |
-
"learning_rate": 0.0001542066027689031,
|
| 535 |
-
"loss": 0.1653,
|
| 536 |
-
"step": 860
|
| 537 |
-
},
|
| 538 |
-
{
|
| 539 |
-
"epoch": 1.39,
|
| 540 |
-
"learning_rate": 0.00015367412140575082,
|
| 541 |
-
"loss": 0.1917,
|
| 542 |
-
"step": 870
|
| 543 |
-
},
|
| 544 |
-
{
|
| 545 |
-
"epoch": 1.41,
|
| 546 |
-
"learning_rate": 0.00015314164004259851,
|
| 547 |
-
"loss": 0.1041,
|
| 548 |
-
"step": 880
|
| 549 |
-
},
|
| 550 |
-
{
|
| 551 |
-
"epoch": 1.42,
|
| 552 |
-
"learning_rate": 0.00015260915867944624,
|
| 553 |
-
"loss": 0.1186,
|
| 554 |
-
"step": 890
|
| 555 |
-
},
|
| 556 |
{
|
| 557 |
"epoch": 1.44,
|
| 558 |
-
"learning_rate": 0.
|
| 559 |
-
"loss": 0.
|
| 560 |
"step": 900
|
| 561 |
},
|
| 562 |
-
{
|
| 563 |
-
"epoch": 1.45,
|
| 564 |
-
"learning_rate": 0.00015154419595314164,
|
| 565 |
-
"loss": 0.0971,
|
| 566 |
-
"step": 910
|
| 567 |
-
},
|
| 568 |
-
{
|
| 569 |
-
"epoch": 1.47,
|
| 570 |
-
"learning_rate": 0.00015101171458998936,
|
| 571 |
-
"loss": 0.11,
|
| 572 |
-
"step": 920
|
| 573 |
-
},
|
| 574 |
-
{
|
| 575 |
-
"epoch": 1.49,
|
| 576 |
-
"learning_rate": 0.00015047923322683706,
|
| 577 |
-
"loss": 0.1126,
|
| 578 |
-
"step": 930
|
| 579 |
-
},
|
| 580 |
-
{
|
| 581 |
-
"epoch": 1.5,
|
| 582 |
-
"learning_rate": 0.0001499467518636848,
|
| 583 |
-
"loss": 0.1095,
|
| 584 |
-
"step": 940
|
| 585 |
-
},
|
| 586 |
-
{
|
| 587 |
-
"epoch": 1.52,
|
| 588 |
-
"learning_rate": 0.00014941427050053249,
|
| 589 |
-
"loss": 0.145,
|
| 590 |
-
"step": 950
|
| 591 |
-
},
|
| 592 |
-
{
|
| 593 |
-
"epoch": 1.53,
|
| 594 |
-
"learning_rate": 0.0001488817891373802,
|
| 595 |
-
"loss": 0.163,
|
| 596 |
-
"step": 960
|
| 597 |
-
},
|
| 598 |
-
{
|
| 599 |
-
"epoch": 1.55,
|
| 600 |
-
"learning_rate": 0.0001483493077742279,
|
| 601 |
-
"loss": 0.0869,
|
| 602 |
-
"step": 970
|
| 603 |
-
},
|
| 604 |
-
{
|
| 605 |
-
"epoch": 1.57,
|
| 606 |
-
"learning_rate": 0.0001478168264110756,
|
| 607 |
-
"loss": 0.0842,
|
| 608 |
-
"step": 980
|
| 609 |
-
},
|
| 610 |
-
{
|
| 611 |
-
"epoch": 1.58,
|
| 612 |
-
"learning_rate": 0.00014728434504792333,
|
| 613 |
-
"loss": 0.15,
|
| 614 |
-
"step": 990
|
| 615 |
-
},
|
| 616 |
{
|
| 617 |
"epoch": 1.6,
|
| 618 |
-
"learning_rate": 0.
|
| 619 |
-
"loss": 0.
|
| 620 |
"step": 1000
|
| 621 |
},
|
| 622 |
{
|
| 623 |
-
"epoch": 1.
|
| 624 |
-
"learning_rate": 0.
|
| 625 |
-
"loss": 0.
|
| 626 |
-
"step":
|
| 627 |
-
},
|
| 628 |
-
{
|
| 629 |
-
"epoch": 1.63,
|
| 630 |
-
"learning_rate": 0.00014568690095846646,
|
| 631 |
-
"loss": 0.0986,
|
| 632 |
-
"step": 1020
|
| 633 |
-
},
|
| 634 |
-
{
|
| 635 |
-
"epoch": 1.65,
|
| 636 |
-
"learning_rate": 0.00014515441959531418,
|
| 637 |
-
"loss": 0.1126,
|
| 638 |
-
"step": 1030
|
| 639 |
-
},
|
| 640 |
-
{
|
| 641 |
-
"epoch": 1.66,
|
| 642 |
-
"learning_rate": 0.00014462193823216188,
|
| 643 |
-
"loss": 0.1521,
|
| 644 |
-
"step": 1040
|
| 645 |
-
},
|
| 646 |
-
{
|
| 647 |
-
"epoch": 1.68,
|
| 648 |
-
"learning_rate": 0.00014408945686900958,
|
| 649 |
-
"loss": 0.1989,
|
| 650 |
-
"step": 1050
|
| 651 |
-
},
|
| 652 |
-
{
|
| 653 |
-
"epoch": 1.69,
|
| 654 |
-
"learning_rate": 0.0001435569755058573,
|
| 655 |
-
"loss": 0.1445,
|
| 656 |
-
"step": 1060
|
| 657 |
},
|
| 658 |
{
|
| 659 |
-
"epoch": 1.
|
| 660 |
-
"learning_rate": 0.
|
| 661 |
-
"loss": 0.
|
| 662 |
-
"step":
|
| 663 |
},
|
| 664 |
{
|
| 665 |
-
"epoch":
|
| 666 |
-
"
|
| 667 |
-
"
|
| 668 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 669 |
},
|
| 670 |
{
|
| 671 |
-
"epoch":
|
| 672 |
-
"learning_rate":
|
| 673 |
-
"loss": 0.
|
| 674 |
-
"step":
|
| 675 |
},
|
| 676 |
{
|
| 677 |
-
"epoch":
|
| 678 |
-
"learning_rate":
|
| 679 |
-
"loss": 0.
|
| 680 |
-
"step":
|
| 681 |
},
|
| 682 |
{
|
| 683 |
-
"epoch":
|
| 684 |
-
"learning_rate":
|
| 685 |
-
"loss": 0.
|
| 686 |
-
"step":
|
| 687 |
},
|
| 688 |
{
|
| 689 |
-
"epoch":
|
| 690 |
-
"learning_rate":
|
| 691 |
-
"loss": 0.
|
| 692 |
-
"step":
|
| 693 |
},
|
| 694 |
{
|
| 695 |
-
"epoch":
|
| 696 |
-
"learning_rate":
|
| 697 |
-
"loss": 0.
|
| 698 |
-
"step":
|
| 699 |
},
|
| 700 |
{
|
| 701 |
-
"epoch":
|
| 702 |
-
"learning_rate":
|
| 703 |
-
"loss": 0.
|
| 704 |
-
"step":
|
| 705 |
},
|
| 706 |
{
|
| 707 |
-
"epoch":
|
| 708 |
-
"
|
| 709 |
-
"
|
| 710 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 711 |
},
|
| 712 |
{
|
| 713 |
-
"epoch":
|
| 714 |
-
"learning_rate":
|
| 715 |
-
"loss": 0.
|
| 716 |
-
"step":
|
| 717 |
},
|
| 718 |
{
|
| 719 |
-
"epoch":
|
| 720 |
-
"learning_rate":
|
| 721 |
-
"loss": 0.
|
| 722 |
-
"step":
|
| 723 |
},
|
| 724 |
{
|
| 725 |
-
"epoch":
|
| 726 |
-
"learning_rate":
|
| 727 |
-
"loss": 0.
|
| 728 |
-
"step":
|
| 729 |
},
|
| 730 |
{
|
| 731 |
-
"epoch":
|
| 732 |
-
"learning_rate":
|
| 733 |
-
"loss": 0.
|
| 734 |
-
"step":
|
| 735 |
},
|
| 736 |
{
|
| 737 |
-
"epoch":
|
| 738 |
-
"learning_rate":
|
| 739 |
-
"loss": 0.
|
| 740 |
-
"step":
|
| 741 |
},
|
| 742 |
{
|
| 743 |
-
"epoch":
|
| 744 |
-
"learning_rate":
|
| 745 |
-
"loss": 0.
|
| 746 |
-
"step":
|
| 747 |
},
|
| 748 |
{
|
| 749 |
-
"epoch":
|
| 750 |
-
"learning_rate":
|
| 751 |
-
"loss": 0.
|
| 752 |
-
"step":
|
| 753 |
},
|
| 754 |
{
|
| 755 |
-
"epoch":
|
| 756 |
-
"
|
| 757 |
-
"
|
| 758 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 759 |
},
|
| 760 |
{
|
| 761 |
-
"epoch":
|
| 762 |
-
"
|
| 763 |
-
"
|
| 764 |
-
"
|
| 765 |
-
|
| 766 |
-
|
| 767 |
-
"
|
| 768 |
-
"learning_rate": 0.0001334398296059638,
|
| 769 |
-
"loss": 0.1279,
|
| 770 |
-
"step": 1250
|
| 771 |
-
},
|
| 772 |
-
{
|
| 773 |
-
"epoch": 2.0,
|
| 774 |
-
"eval_accuracy": 0.8186528497409327,
|
| 775 |
-
"eval_f1": 0.8223400297767379,
|
| 776 |
-
"eval_loss": 0.4645093083381653,
|
| 777 |
-
"eval_precision": 0.8294534912410559,
|
| 778 |
-
"eval_recall": 0.8186528497409327,
|
| 779 |
-
"eval_runtime": 1.2384,
|
| 780 |
-
"eval_samples_per_second": 155.847,
|
| 781 |
-
"eval_steps_per_second": 10.497,
|
| 782 |
-
"step": 1252
|
| 783 |
-
},
|
| 784 |
-
{
|
| 785 |
-
"epoch": 2.01,
|
| 786 |
-
"learning_rate": 0.0001329073482428115,
|
| 787 |
-
"loss": 0.1003,
|
| 788 |
-
"step": 1260
|
| 789 |
-
},
|
| 790 |
-
{
|
| 791 |
-
"epoch": 2.03,
|
| 792 |
-
"learning_rate": 0.00013237486687965922,
|
| 793 |
-
"loss": 0.0551,
|
| 794 |
-
"step": 1270
|
| 795 |
-
},
|
| 796 |
-
{
|
| 797 |
-
"epoch": 2.04,
|
| 798 |
-
"learning_rate": 0.00013184238551650694,
|
| 799 |
-
"loss": 0.1244,
|
| 800 |
-
"step": 1280
|
| 801 |
-
},
|
| 802 |
-
{
|
| 803 |
-
"epoch": 2.06,
|
| 804 |
-
"learning_rate": 0.00013130990415335464,
|
| 805 |
-
"loss": 0.0669,
|
| 806 |
-
"step": 1290
|
| 807 |
-
},
|
| 808 |
-
{
|
| 809 |
-
"epoch": 2.08,
|
| 810 |
-
"learning_rate": 0.00013077742279020234,
|
| 811 |
-
"loss": 0.0603,
|
| 812 |
-
"step": 1300
|
| 813 |
-
},
|
| 814 |
-
{
|
| 815 |
-
"epoch": 2.09,
|
| 816 |
-
"learning_rate": 0.00013024494142705007,
|
| 817 |
-
"loss": 0.0074,
|
| 818 |
-
"step": 1310
|
| 819 |
-
},
|
| 820 |
-
{
|
| 821 |
-
"epoch": 2.11,
|
| 822 |
-
"learning_rate": 0.00012971246006389777,
|
| 823 |
-
"loss": 0.0144,
|
| 824 |
-
"step": 1320
|
| 825 |
-
},
|
| 826 |
-
{
|
| 827 |
-
"epoch": 2.12,
|
| 828 |
-
"learning_rate": 0.00012917997870074546,
|
| 829 |
-
"loss": 0.0741,
|
| 830 |
-
"step": 1330
|
| 831 |
-
},
|
| 832 |
-
{
|
| 833 |
-
"epoch": 2.14,
|
| 834 |
-
"learning_rate": 0.0001286474973375932,
|
| 835 |
-
"loss": 0.053,
|
| 836 |
-
"step": 1340
|
| 837 |
-
},
|
| 838 |
-
{
|
| 839 |
-
"epoch": 2.16,
|
| 840 |
-
"learning_rate": 0.00012811501597444092,
|
| 841 |
-
"loss": 0.0377,
|
| 842 |
-
"step": 1350
|
| 843 |
-
},
|
| 844 |
-
{
|
| 845 |
-
"epoch": 2.17,
|
| 846 |
-
"learning_rate": 0.00012758253461128861,
|
| 847 |
-
"loss": 0.0137,
|
| 848 |
-
"step": 1360
|
| 849 |
-
},
|
| 850 |
-
{
|
| 851 |
-
"epoch": 2.19,
|
| 852 |
-
"learning_rate": 0.0001270500532481363,
|
| 853 |
-
"loss": 0.0116,
|
| 854 |
-
"step": 1370
|
| 855 |
-
},
|
| 856 |
-
{
|
| 857 |
-
"epoch": 2.2,
|
| 858 |
-
"learning_rate": 0.00012651757188498404,
|
| 859 |
-
"loss": 0.0615,
|
| 860 |
-
"step": 1380
|
| 861 |
-
},
|
| 862 |
-
{
|
| 863 |
-
"epoch": 2.22,
|
| 864 |
-
"learning_rate": 0.00012598509052183174,
|
| 865 |
-
"loss": 0.057,
|
| 866 |
-
"step": 1390
|
| 867 |
-
},
|
| 868 |
-
{
|
| 869 |
-
"epoch": 2.24,
|
| 870 |
-
"learning_rate": 0.00012545260915867944,
|
| 871 |
-
"loss": 0.013,
|
| 872 |
-
"step": 1400
|
| 873 |
-
},
|
| 874 |
-
{
|
| 875 |
-
"epoch": 2.25,
|
| 876 |
-
"learning_rate": 0.00012492012779552716,
|
| 877 |
-
"loss": 0.0113,
|
| 878 |
-
"step": 1410
|
| 879 |
-
},
|
| 880 |
-
{
|
| 881 |
-
"epoch": 2.27,
|
| 882 |
-
"learning_rate": 0.0001243876464323749,
|
| 883 |
-
"loss": 0.0665,
|
| 884 |
-
"step": 1420
|
| 885 |
-
},
|
| 886 |
-
{
|
| 887 |
-
"epoch": 2.28,
|
| 888 |
-
"learning_rate": 0.00012385516506922259,
|
| 889 |
-
"loss": 0.0143,
|
| 890 |
-
"step": 1430
|
| 891 |
-
},
|
| 892 |
-
{
|
| 893 |
-
"epoch": 2.3,
|
| 894 |
-
"learning_rate": 0.00012332268370607028,
|
| 895 |
-
"loss": 0.016,
|
| 896 |
-
"step": 1440
|
| 897 |
-
},
|
| 898 |
-
{
|
| 899 |
-
"epoch": 2.32,
|
| 900 |
-
"learning_rate": 0.000122790202342918,
|
| 901 |
-
"loss": 0.0449,
|
| 902 |
-
"step": 1450
|
| 903 |
-
},
|
| 904 |
-
{
|
| 905 |
-
"epoch": 2.33,
|
| 906 |
-
"learning_rate": 0.0001222577209797657,
|
| 907 |
-
"loss": 0.0312,
|
| 908 |
-
"step": 1460
|
| 909 |
-
},
|
| 910 |
-
{
|
| 911 |
-
"epoch": 2.35,
|
| 912 |
-
"learning_rate": 0.00012172523961661342,
|
| 913 |
-
"loss": 0.0664,
|
| 914 |
-
"step": 1470
|
| 915 |
-
},
|
| 916 |
-
{
|
| 917 |
-
"epoch": 2.36,
|
| 918 |
-
"learning_rate": 0.00012119275825346112,
|
| 919 |
-
"loss": 0.0901,
|
| 920 |
-
"step": 1480
|
| 921 |
-
},
|
| 922 |
-
{
|
| 923 |
-
"epoch": 2.38,
|
| 924 |
-
"learning_rate": 0.00012066027689030886,
|
| 925 |
-
"loss": 0.0564,
|
| 926 |
-
"step": 1490
|
| 927 |
-
},
|
| 928 |
-
{
|
| 929 |
-
"epoch": 2.4,
|
| 930 |
-
"learning_rate": 0.00012012779552715656,
|
| 931 |
-
"loss": 0.0304,
|
| 932 |
-
"step": 1500
|
| 933 |
-
},
|
| 934 |
-
{
|
| 935 |
-
"epoch": 2.41,
|
| 936 |
-
"learning_rate": 0.00011959531416400427,
|
| 937 |
-
"loss": 0.0852,
|
| 938 |
-
"step": 1510
|
| 939 |
-
},
|
| 940 |
-
{
|
| 941 |
-
"epoch": 2.43,
|
| 942 |
-
"learning_rate": 0.00011906283280085198,
|
| 943 |
-
"loss": 0.0405,
|
| 944 |
-
"step": 1520
|
| 945 |
-
},
|
| 946 |
-
{
|
| 947 |
-
"epoch": 2.44,
|
| 948 |
-
"learning_rate": 0.00011853035143769968,
|
| 949 |
-
"loss": 0.0812,
|
| 950 |
-
"step": 1530
|
| 951 |
-
},
|
| 952 |
-
{
|
| 953 |
-
"epoch": 2.46,
|
| 954 |
-
"learning_rate": 0.00011799787007454739,
|
| 955 |
-
"loss": 0.0343,
|
| 956 |
-
"step": 1540
|
| 957 |
-
},
|
| 958 |
-
{
|
| 959 |
-
"epoch": 2.48,
|
| 960 |
-
"learning_rate": 0.00011746538871139509,
|
| 961 |
-
"loss": 0.0313,
|
| 962 |
-
"step": 1550
|
| 963 |
-
},
|
| 964 |
-
{
|
| 965 |
-
"epoch": 2.49,
|
| 966 |
-
"learning_rate": 0.00011693290734824283,
|
| 967 |
-
"loss": 0.0683,
|
| 968 |
-
"step": 1560
|
| 969 |
-
},
|
| 970 |
-
{
|
| 971 |
-
"epoch": 2.51,
|
| 972 |
-
"learning_rate": 0.00011640042598509053,
|
| 973 |
-
"loss": 0.0378,
|
| 974 |
-
"step": 1570
|
| 975 |
-
},
|
| 976 |
-
{
|
| 977 |
-
"epoch": 2.52,
|
| 978 |
-
"learning_rate": 0.00011586794462193824,
|
| 979 |
-
"loss": 0.0404,
|
| 980 |
-
"step": 1580
|
| 981 |
-
},
|
| 982 |
-
{
|
| 983 |
-
"epoch": 2.54,
|
| 984 |
-
"learning_rate": 0.00011533546325878595,
|
| 985 |
-
"loss": 0.0786,
|
| 986 |
-
"step": 1590
|
| 987 |
-
},
|
| 988 |
-
{
|
| 989 |
-
"epoch": 2.56,
|
| 990 |
-
"learning_rate": 0.00011480298189563365,
|
| 991 |
-
"loss": 0.0353,
|
| 992 |
-
"step": 1600
|
| 993 |
-
},
|
| 994 |
-
{
|
| 995 |
-
"epoch": 2.57,
|
| 996 |
-
"learning_rate": 0.00011427050053248136,
|
| 997 |
-
"loss": 0.0458,
|
| 998 |
-
"step": 1610
|
| 999 |
-
},
|
| 1000 |
-
{
|
| 1001 |
-
"epoch": 2.59,
|
| 1002 |
-
"learning_rate": 0.00011373801916932908,
|
| 1003 |
-
"loss": 0.0624,
|
| 1004 |
-
"step": 1620
|
| 1005 |
-
},
|
| 1006 |
-
{
|
| 1007 |
-
"epoch": 2.6,
|
| 1008 |
-
"learning_rate": 0.0001132055378061768,
|
| 1009 |
-
"loss": 0.0878,
|
| 1010 |
-
"step": 1630
|
| 1011 |
-
},
|
| 1012 |
-
{
|
| 1013 |
-
"epoch": 2.62,
|
| 1014 |
-
"learning_rate": 0.0001126730564430245,
|
| 1015 |
-
"loss": 0.0245,
|
| 1016 |
-
"step": 1640
|
| 1017 |
-
},
|
| 1018 |
-
{
|
| 1019 |
-
"epoch": 2.64,
|
| 1020 |
-
"learning_rate": 0.00011214057507987221,
|
| 1021 |
-
"loss": 0.079,
|
| 1022 |
-
"step": 1650
|
| 1023 |
-
},
|
| 1024 |
-
{
|
| 1025 |
-
"epoch": 2.65,
|
| 1026 |
-
"learning_rate": 0.00011160809371671992,
|
| 1027 |
-
"loss": 0.0374,
|
| 1028 |
-
"step": 1660
|
| 1029 |
-
},
|
| 1030 |
-
{
|
| 1031 |
-
"epoch": 2.67,
|
| 1032 |
-
"learning_rate": 0.00011107561235356762,
|
| 1033 |
-
"loss": 0.0572,
|
| 1034 |
-
"step": 1670
|
| 1035 |
-
},
|
| 1036 |
-
{
|
| 1037 |
-
"epoch": 2.68,
|
| 1038 |
-
"learning_rate": 0.00011054313099041533,
|
| 1039 |
-
"loss": 0.1096,
|
| 1040 |
-
"step": 1680
|
| 1041 |
-
},
|
| 1042 |
-
{
|
| 1043 |
-
"epoch": 2.7,
|
| 1044 |
-
"learning_rate": 0.00011001064962726305,
|
| 1045 |
-
"loss": 0.1099,
|
| 1046 |
-
"step": 1690
|
| 1047 |
-
},
|
| 1048 |
-
{
|
| 1049 |
-
"epoch": 2.72,
|
| 1050 |
-
"learning_rate": 0.00010947816826411077,
|
| 1051 |
-
"loss": 0.074,
|
| 1052 |
-
"step": 1700
|
| 1053 |
-
},
|
| 1054 |
-
{
|
| 1055 |
-
"epoch": 2.73,
|
| 1056 |
-
"learning_rate": 0.00010894568690095847,
|
| 1057 |
-
"loss": 0.0348,
|
| 1058 |
-
"step": 1710
|
| 1059 |
-
},
|
| 1060 |
-
{
|
| 1061 |
-
"epoch": 2.75,
|
| 1062 |
-
"learning_rate": 0.00010841320553780618,
|
| 1063 |
-
"loss": 0.1437,
|
| 1064 |
-
"step": 1720
|
| 1065 |
-
},
|
| 1066 |
-
{
|
| 1067 |
-
"epoch": 2.76,
|
| 1068 |
-
"learning_rate": 0.0001078807241746539,
|
| 1069 |
-
"loss": 0.0798,
|
| 1070 |
-
"step": 1730
|
| 1071 |
-
},
|
| 1072 |
-
{
|
| 1073 |
-
"epoch": 2.78,
|
| 1074 |
-
"learning_rate": 0.0001073482428115016,
|
| 1075 |
-
"loss": 0.0662,
|
| 1076 |
-
"step": 1740
|
| 1077 |
-
},
|
| 1078 |
-
{
|
| 1079 |
-
"epoch": 2.8,
|
| 1080 |
-
"learning_rate": 0.0001068157614483493,
|
| 1081 |
-
"loss": 0.0207,
|
| 1082 |
-
"step": 1750
|
| 1083 |
-
},
|
| 1084 |
-
{
|
| 1085 |
-
"epoch": 2.81,
|
| 1086 |
-
"learning_rate": 0.00010628328008519702,
|
| 1087 |
-
"loss": 0.1088,
|
| 1088 |
-
"step": 1760
|
| 1089 |
-
},
|
| 1090 |
-
{
|
| 1091 |
-
"epoch": 2.83,
|
| 1092 |
-
"learning_rate": 0.00010575079872204474,
|
| 1093 |
-
"loss": 0.0871,
|
| 1094 |
-
"step": 1770
|
| 1095 |
-
},
|
| 1096 |
-
{
|
| 1097 |
-
"epoch": 2.84,
|
| 1098 |
-
"learning_rate": 0.00010521831735889244,
|
| 1099 |
-
"loss": 0.0846,
|
| 1100 |
-
"step": 1780
|
| 1101 |
-
},
|
| 1102 |
-
{
|
| 1103 |
-
"epoch": 2.86,
|
| 1104 |
-
"learning_rate": 0.00010468583599574015,
|
| 1105 |
-
"loss": 0.0644,
|
| 1106 |
-
"step": 1790
|
| 1107 |
-
},
|
| 1108 |
-
{
|
| 1109 |
-
"epoch": 2.88,
|
| 1110 |
-
"learning_rate": 0.00010415335463258787,
|
| 1111 |
-
"loss": 0.056,
|
| 1112 |
-
"step": 1800
|
| 1113 |
-
},
|
| 1114 |
-
{
|
| 1115 |
-
"epoch": 2.89,
|
| 1116 |
-
"learning_rate": 0.00010362087326943557,
|
| 1117 |
-
"loss": 0.1148,
|
| 1118 |
-
"step": 1810
|
| 1119 |
-
},
|
| 1120 |
-
{
|
| 1121 |
-
"epoch": 2.91,
|
| 1122 |
-
"learning_rate": 0.00010308839190628328,
|
| 1123 |
-
"loss": 0.0919,
|
| 1124 |
-
"step": 1820
|
| 1125 |
-
},
|
| 1126 |
-
{
|
| 1127 |
-
"epoch": 2.92,
|
| 1128 |
-
"learning_rate": 0.000102555910543131,
|
| 1129 |
-
"loss": 0.0343,
|
| 1130 |
-
"step": 1830
|
| 1131 |
-
},
|
| 1132 |
-
{
|
| 1133 |
-
"epoch": 2.94,
|
| 1134 |
-
"learning_rate": 0.00010202342917997871,
|
| 1135 |
-
"loss": 0.0103,
|
| 1136 |
-
"step": 1840
|
| 1137 |
-
},
|
| 1138 |
-
{
|
| 1139 |
-
"epoch": 2.96,
|
| 1140 |
-
"learning_rate": 0.00010149094781682643,
|
| 1141 |
-
"loss": 0.0287,
|
| 1142 |
-
"step": 1850
|
| 1143 |
-
},
|
| 1144 |
-
{
|
| 1145 |
-
"epoch": 2.97,
|
| 1146 |
-
"learning_rate": 0.00010095846645367413,
|
| 1147 |
-
"loss": 0.0382,
|
| 1148 |
-
"step": 1860
|
| 1149 |
-
},
|
| 1150 |
-
{
|
| 1151 |
-
"epoch": 2.99,
|
| 1152 |
-
"learning_rate": 0.00010042598509052184,
|
| 1153 |
-
"loss": 0.0912,
|
| 1154 |
-
"step": 1870
|
| 1155 |
-
},
|
| 1156 |
-
{
|
| 1157 |
-
"epoch": 3.0,
|
| 1158 |
-
"eval_accuracy": 0.8497409326424871,
|
| 1159 |
-
"eval_f1": 0.8454290011413006,
|
| 1160 |
-
"eval_loss": 0.4882831573486328,
|
| 1161 |
-
"eval_precision": 0.8461923629555449,
|
| 1162 |
-
"eval_recall": 0.8497409326424871,
|
| 1163 |
-
"eval_runtime": 1.2092,
|
| 1164 |
-
"eval_samples_per_second": 159.604,
|
| 1165 |
-
"eval_steps_per_second": 10.751,
|
| 1166 |
-
"step": 1878
|
| 1167 |
-
},
|
| 1168 |
-
{
|
| 1169 |
-
"epoch": 3.0,
|
| 1170 |
-
"learning_rate": 9.989350372736954e-05,
|
| 1171 |
-
"loss": 0.0122,
|
| 1172 |
-
"step": 1880
|
| 1173 |
-
},
|
| 1174 |
-
{
|
| 1175 |
-
"epoch": 3.02,
|
| 1176 |
-
"learning_rate": 9.936102236421726e-05,
|
| 1177 |
-
"loss": 0.0255,
|
| 1178 |
-
"step": 1890
|
| 1179 |
-
},
|
| 1180 |
-
{
|
| 1181 |
-
"epoch": 3.04,
|
| 1182 |
-
"learning_rate": 9.882854100106496e-05,
|
| 1183 |
-
"loss": 0.0246,
|
| 1184 |
-
"step": 1900
|
| 1185 |
-
},
|
| 1186 |
-
{
|
| 1187 |
-
"epoch": 3.05,
|
| 1188 |
-
"learning_rate": 9.829605963791267e-05,
|
| 1189 |
-
"loss": 0.0134,
|
| 1190 |
-
"step": 1910
|
| 1191 |
-
},
|
| 1192 |
-
{
|
| 1193 |
-
"epoch": 3.07,
|
| 1194 |
-
"learning_rate": 9.77635782747604e-05,
|
| 1195 |
-
"loss": 0.013,
|
| 1196 |
-
"step": 1920
|
| 1197 |
-
},
|
| 1198 |
-
{
|
| 1199 |
-
"epoch": 3.08,
|
| 1200 |
-
"learning_rate": 9.72310969116081e-05,
|
| 1201 |
-
"loss": 0.0057,
|
| 1202 |
-
"step": 1930
|
| 1203 |
-
},
|
| 1204 |
-
{
|
| 1205 |
-
"epoch": 3.1,
|
| 1206 |
-
"learning_rate": 9.669861554845581e-05,
|
| 1207 |
-
"loss": 0.0149,
|
| 1208 |
-
"step": 1940
|
| 1209 |
-
},
|
| 1210 |
-
{
|
| 1211 |
-
"epoch": 3.12,
|
| 1212 |
-
"learning_rate": 9.616613418530351e-05,
|
| 1213 |
-
"loss": 0.0025,
|
| 1214 |
-
"step": 1950
|
| 1215 |
-
},
|
| 1216 |
-
{
|
| 1217 |
-
"epoch": 3.13,
|
| 1218 |
-
"learning_rate": 9.563365282215123e-05,
|
| 1219 |
-
"loss": 0.0288,
|
| 1220 |
-
"step": 1960
|
| 1221 |
-
},
|
| 1222 |
-
{
|
| 1223 |
-
"epoch": 3.15,
|
| 1224 |
-
"learning_rate": 9.510117145899895e-05,
|
| 1225 |
-
"loss": 0.0014,
|
| 1226 |
-
"step": 1970
|
| 1227 |
-
},
|
| 1228 |
-
{
|
| 1229 |
-
"epoch": 3.16,
|
| 1230 |
-
"learning_rate": 9.456869009584664e-05,
|
| 1231 |
-
"loss": 0.0104,
|
| 1232 |
-
"step": 1980
|
| 1233 |
-
},
|
| 1234 |
-
{
|
| 1235 |
-
"epoch": 3.18,
|
| 1236 |
-
"learning_rate": 9.403620873269437e-05,
|
| 1237 |
-
"loss": 0.0309,
|
| 1238 |
-
"step": 1990
|
| 1239 |
-
},
|
| 1240 |
-
{
|
| 1241 |
-
"epoch": 3.19,
|
| 1242 |
-
"learning_rate": 9.350372736954207e-05,
|
| 1243 |
-
"loss": 0.0175,
|
| 1244 |
-
"step": 2000
|
| 1245 |
-
},
|
| 1246 |
-
{
|
| 1247 |
-
"epoch": 3.21,
|
| 1248 |
-
"learning_rate": 9.297124600638978e-05,
|
| 1249 |
-
"loss": 0.0238,
|
| 1250 |
-
"step": 2010
|
| 1251 |
-
},
|
| 1252 |
-
{
|
| 1253 |
-
"epoch": 3.23,
|
| 1254 |
-
"learning_rate": 9.243876464323749e-05,
|
| 1255 |
-
"loss": 0.0439,
|
| 1256 |
-
"step": 2020
|
| 1257 |
-
},
|
| 1258 |
-
{
|
| 1259 |
-
"epoch": 3.24,
|
| 1260 |
-
"learning_rate": 9.19062832800852e-05,
|
| 1261 |
-
"loss": 0.0286,
|
| 1262 |
-
"step": 2030
|
| 1263 |
-
},
|
| 1264 |
-
{
|
| 1265 |
-
"epoch": 3.26,
|
| 1266 |
-
"learning_rate": 9.137380191693292e-05,
|
| 1267 |
-
"loss": 0.0165,
|
| 1268 |
-
"step": 2040
|
| 1269 |
-
},
|
| 1270 |
-
{
|
| 1271 |
-
"epoch": 3.27,
|
| 1272 |
-
"learning_rate": 9.084132055378062e-05,
|
| 1273 |
-
"loss": 0.0108,
|
| 1274 |
-
"step": 2050
|
| 1275 |
-
},
|
| 1276 |
-
{
|
| 1277 |
-
"epoch": 3.29,
|
| 1278 |
-
"learning_rate": 9.030883919062834e-05,
|
| 1279 |
-
"loss": 0.0024,
|
| 1280 |
-
"step": 2060
|
| 1281 |
-
},
|
| 1282 |
-
{
|
| 1283 |
-
"epoch": 3.31,
|
| 1284 |
-
"learning_rate": 8.977635782747604e-05,
|
| 1285 |
-
"loss": 0.003,
|
| 1286 |
-
"step": 2070
|
| 1287 |
-
},
|
| 1288 |
-
{
|
| 1289 |
-
"epoch": 3.32,
|
| 1290 |
-
"learning_rate": 8.924387646432375e-05,
|
| 1291 |
-
"loss": 0.0028,
|
| 1292 |
-
"step": 2080
|
| 1293 |
-
},
|
| 1294 |
-
{
|
| 1295 |
-
"epoch": 3.34,
|
| 1296 |
-
"learning_rate": 8.871139510117146e-05,
|
| 1297 |
-
"loss": 0.078,
|
| 1298 |
-
"step": 2090
|
| 1299 |
-
},
|
| 1300 |
-
{
|
| 1301 |
-
"epoch": 3.35,
|
| 1302 |
-
"learning_rate": 8.817891373801918e-05,
|
| 1303 |
-
"loss": 0.0013,
|
| 1304 |
-
"step": 2100
|
| 1305 |
-
},
|
| 1306 |
-
{
|
| 1307 |
-
"epoch": 3.37,
|
| 1308 |
-
"learning_rate": 8.764643237486689e-05,
|
| 1309 |
-
"loss": 0.0027,
|
| 1310 |
-
"step": 2110
|
| 1311 |
-
},
|
| 1312 |
-
{
|
| 1313 |
-
"epoch": 3.39,
|
| 1314 |
-
"learning_rate": 8.711395101171459e-05,
|
| 1315 |
-
"loss": 0.0169,
|
| 1316 |
-
"step": 2120
|
| 1317 |
-
},
|
| 1318 |
-
{
|
| 1319 |
-
"epoch": 3.4,
|
| 1320 |
-
"learning_rate": 8.658146964856231e-05,
|
| 1321 |
-
"loss": 0.0078,
|
| 1322 |
-
"step": 2130
|
| 1323 |
-
},
|
| 1324 |
-
{
|
| 1325 |
-
"epoch": 3.42,
|
| 1326 |
-
"learning_rate": 8.604898828541001e-05,
|
| 1327 |
-
"loss": 0.0016,
|
| 1328 |
-
"step": 2140
|
| 1329 |
-
},
|
| 1330 |
-
{
|
| 1331 |
-
"epoch": 3.43,
|
| 1332 |
-
"learning_rate": 8.551650692225772e-05,
|
| 1333 |
-
"loss": 0.0444,
|
| 1334 |
-
"step": 2150
|
| 1335 |
-
},
|
| 1336 |
-
{
|
| 1337 |
-
"epoch": 3.45,
|
| 1338 |
-
"learning_rate": 8.498402555910544e-05,
|
| 1339 |
-
"loss": 0.003,
|
| 1340 |
-
"step": 2160
|
| 1341 |
-
},
|
| 1342 |
-
{
|
| 1343 |
-
"epoch": 3.47,
|
| 1344 |
-
"learning_rate": 8.445154419595315e-05,
|
| 1345 |
-
"loss": 0.0045,
|
| 1346 |
-
"step": 2170
|
| 1347 |
-
},
|
| 1348 |
-
{
|
| 1349 |
-
"epoch": 3.48,
|
| 1350 |
-
"learning_rate": 8.391906283280086e-05,
|
| 1351 |
-
"loss": 0.0014,
|
| 1352 |
-
"step": 2180
|
| 1353 |
-
},
|
| 1354 |
-
{
|
| 1355 |
-
"epoch": 3.5,
|
| 1356 |
-
"learning_rate": 8.338658146964856e-05,
|
| 1357 |
-
"loss": 0.023,
|
| 1358 |
-
"step": 2190
|
| 1359 |
-
},
|
| 1360 |
-
{
|
| 1361 |
-
"epoch": 3.51,
|
| 1362 |
-
"learning_rate": 8.285410010649628e-05,
|
| 1363 |
-
"loss": 0.0064,
|
| 1364 |
-
"step": 2200
|
| 1365 |
-
},
|
| 1366 |
-
{
|
| 1367 |
-
"epoch": 3.53,
|
| 1368 |
-
"learning_rate": 8.232161874334398e-05,
|
| 1369 |
-
"loss": 0.0225,
|
| 1370 |
-
"step": 2210
|
| 1371 |
-
},
|
| 1372 |
-
{
|
| 1373 |
-
"epoch": 3.55,
|
| 1374 |
-
"learning_rate": 8.17891373801917e-05,
|
| 1375 |
-
"loss": 0.06,
|
| 1376 |
-
"step": 2220
|
| 1377 |
-
},
|
| 1378 |
-
{
|
| 1379 |
-
"epoch": 3.56,
|
| 1380 |
-
"learning_rate": 8.12566560170394e-05,
|
| 1381 |
-
"loss": 0.0516,
|
| 1382 |
-
"step": 2230
|
| 1383 |
-
},
|
| 1384 |
-
{
|
| 1385 |
-
"epoch": 3.58,
|
| 1386 |
-
"learning_rate": 8.072417465388712e-05,
|
| 1387 |
-
"loss": 0.012,
|
| 1388 |
-
"step": 2240
|
| 1389 |
-
},
|
| 1390 |
-
{
|
| 1391 |
-
"epoch": 3.59,
|
| 1392 |
-
"learning_rate": 8.019169329073483e-05,
|
| 1393 |
-
"loss": 0.0067,
|
| 1394 |
-
"step": 2250
|
| 1395 |
-
},
|
| 1396 |
-
{
|
| 1397 |
-
"epoch": 3.61,
|
| 1398 |
-
"learning_rate": 7.965921192758253e-05,
|
| 1399 |
-
"loss": 0.0253,
|
| 1400 |
-
"step": 2260
|
| 1401 |
-
},
|
| 1402 |
-
{
|
| 1403 |
-
"epoch": 3.63,
|
| 1404 |
-
"learning_rate": 7.912673056443025e-05,
|
| 1405 |
-
"loss": 0.0124,
|
| 1406 |
-
"step": 2270
|
| 1407 |
-
},
|
| 1408 |
-
{
|
| 1409 |
-
"epoch": 3.64,
|
| 1410 |
-
"learning_rate": 7.859424920127795e-05,
|
| 1411 |
-
"loss": 0.0084,
|
| 1412 |
-
"step": 2280
|
| 1413 |
-
},
|
| 1414 |
-
{
|
| 1415 |
-
"epoch": 3.66,
|
| 1416 |
-
"learning_rate": 7.806176783812567e-05,
|
| 1417 |
-
"loss": 0.0027,
|
| 1418 |
-
"step": 2290
|
| 1419 |
-
},
|
| 1420 |
-
{
|
| 1421 |
-
"epoch": 3.67,
|
| 1422 |
-
"learning_rate": 7.752928647497338e-05,
|
| 1423 |
-
"loss": 0.0182,
|
| 1424 |
-
"step": 2300
|
| 1425 |
-
},
|
| 1426 |
-
{
|
| 1427 |
-
"epoch": 3.69,
|
| 1428 |
-
"learning_rate": 7.699680511182109e-05,
|
| 1429 |
-
"loss": 0.056,
|
| 1430 |
-
"step": 2310
|
| 1431 |
-
},
|
| 1432 |
-
{
|
| 1433 |
-
"epoch": 3.71,
|
| 1434 |
-
"learning_rate": 7.64643237486688e-05,
|
| 1435 |
-
"loss": 0.0153,
|
| 1436 |
-
"step": 2320
|
| 1437 |
-
},
|
| 1438 |
-
{
|
| 1439 |
-
"epoch": 3.72,
|
| 1440 |
-
"learning_rate": 7.59318423855165e-05,
|
| 1441 |
-
"loss": 0.0509,
|
| 1442 |
-
"step": 2330
|
| 1443 |
-
},
|
| 1444 |
-
{
|
| 1445 |
-
"epoch": 3.74,
|
| 1446 |
-
"learning_rate": 7.539936102236423e-05,
|
| 1447 |
-
"loss": 0.0045,
|
| 1448 |
-
"step": 2340
|
| 1449 |
-
},
|
| 1450 |
-
{
|
| 1451 |
-
"epoch": 3.75,
|
| 1452 |
-
"learning_rate": 7.486687965921192e-05,
|
| 1453 |
-
"loss": 0.0007,
|
| 1454 |
-
"step": 2350
|
| 1455 |
-
},
|
| 1456 |
-
{
|
| 1457 |
-
"epoch": 3.77,
|
| 1458 |
-
"learning_rate": 7.438764643237486e-05,
|
| 1459 |
-
"loss": 0.0583,
|
| 1460 |
-
"step": 2360
|
| 1461 |
-
},
|
| 1462 |
-
{
|
| 1463 |
-
"epoch": 3.79,
|
| 1464 |
-
"learning_rate": 7.385516506922258e-05,
|
| 1465 |
-
"loss": 0.0014,
|
| 1466 |
-
"step": 2370
|
| 1467 |
-
},
|
| 1468 |
-
{
|
| 1469 |
-
"epoch": 3.8,
|
| 1470 |
-
"learning_rate": 7.332268370607029e-05,
|
| 1471 |
-
"loss": 0.0021,
|
| 1472 |
-
"step": 2380
|
| 1473 |
-
},
|
| 1474 |
-
{
|
| 1475 |
-
"epoch": 3.82,
|
| 1476 |
-
"learning_rate": 7.2790202342918e-05,
|
| 1477 |
-
"loss": 0.0408,
|
| 1478 |
-
"step": 2390
|
| 1479 |
-
},
|
| 1480 |
-
{
|
| 1481 |
-
"epoch": 3.83,
|
| 1482 |
-
"learning_rate": 7.225772097976571e-05,
|
| 1483 |
-
"loss": 0.0138,
|
| 1484 |
-
"step": 2400
|
| 1485 |
-
},
|
| 1486 |
-
{
|
| 1487 |
-
"epoch": 3.85,
|
| 1488 |
-
"learning_rate": 7.172523961661343e-05,
|
| 1489 |
-
"loss": 0.0075,
|
| 1490 |
-
"step": 2410
|
| 1491 |
-
},
|
| 1492 |
-
{
|
| 1493 |
-
"epoch": 3.87,
|
| 1494 |
-
"learning_rate": 7.119275825346114e-05,
|
| 1495 |
-
"loss": 0.0769,
|
| 1496 |
-
"step": 2420
|
| 1497 |
-
},
|
| 1498 |
-
{
|
| 1499 |
-
"epoch": 3.88,
|
| 1500 |
-
"learning_rate": 7.066027689030884e-05,
|
| 1501 |
-
"loss": 0.0016,
|
| 1502 |
-
"step": 2430
|
| 1503 |
-
},
|
| 1504 |
-
{
|
| 1505 |
-
"epoch": 3.9,
|
| 1506 |
-
"learning_rate": 7.012779552715655e-05,
|
| 1507 |
-
"loss": 0.0213,
|
| 1508 |
-
"step": 2440
|
| 1509 |
-
},
|
| 1510 |
-
{
|
| 1511 |
-
"epoch": 3.91,
|
| 1512 |
-
"learning_rate": 6.959531416400426e-05,
|
| 1513 |
-
"loss": 0.0101,
|
| 1514 |
-
"step": 2450
|
| 1515 |
-
},
|
| 1516 |
-
{
|
| 1517 |
-
"epoch": 3.93,
|
| 1518 |
-
"learning_rate": 6.906283280085197e-05,
|
| 1519 |
-
"loss": 0.0087,
|
| 1520 |
-
"step": 2460
|
| 1521 |
-
},
|
| 1522 |
-
{
|
| 1523 |
-
"epoch": 3.95,
|
| 1524 |
-
"learning_rate": 6.853035143769968e-05,
|
| 1525 |
-
"loss": 0.0131,
|
| 1526 |
-
"step": 2470
|
| 1527 |
-
},
|
| 1528 |
-
{
|
| 1529 |
-
"epoch": 3.96,
|
| 1530 |
-
"learning_rate": 6.79978700745474e-05,
|
| 1531 |
-
"loss": 0.0012,
|
| 1532 |
-
"step": 2480
|
| 1533 |
-
},
|
| 1534 |
-
{
|
| 1535 |
-
"epoch": 3.98,
|
| 1536 |
-
"learning_rate": 6.746538871139511e-05,
|
| 1537 |
-
"loss": 0.0254,
|
| 1538 |
-
"step": 2490
|
| 1539 |
-
},
|
| 1540 |
-
{
|
| 1541 |
-
"epoch": 3.99,
|
| 1542 |
-
"learning_rate": 6.693290734824281e-05,
|
| 1543 |
-
"loss": 0.0397,
|
| 1544 |
-
"step": 2500
|
| 1545 |
-
},
|
| 1546 |
-
{
|
| 1547 |
-
"epoch": 4.0,
|
| 1548 |
-
"eval_accuracy": 0.8238341968911918,
|
| 1549 |
-
"eval_f1": 0.8273994981457394,
|
| 1550 |
-
"eval_loss": 0.5439336895942688,
|
| 1551 |
-
"eval_precision": 0.8342306448710682,
|
| 1552 |
-
"eval_recall": 0.8238341968911918,
|
| 1553 |
-
"eval_runtime": 1.3389,
|
| 1554 |
-
"eval_samples_per_second": 144.149,
|
| 1555 |
-
"eval_steps_per_second": 9.709,
|
| 1556 |
-
"step": 2504
|
| 1557 |
-
},
|
| 1558 |
-
{
|
| 1559 |
-
"epoch": 4.01,
|
| 1560 |
-
"learning_rate": 6.640042598509052e-05,
|
| 1561 |
-
"loss": 0.0261,
|
| 1562 |
-
"step": 2510
|
| 1563 |
-
},
|
| 1564 |
-
{
|
| 1565 |
-
"epoch": 4.03,
|
| 1566 |
-
"learning_rate": 6.586794462193823e-05,
|
| 1567 |
-
"loss": 0.0007,
|
| 1568 |
-
"step": 2520
|
| 1569 |
-
},
|
| 1570 |
-
{
|
| 1571 |
-
"epoch": 4.04,
|
| 1572 |
-
"learning_rate": 6.533546325878594e-05,
|
| 1573 |
-
"loss": 0.0232,
|
| 1574 |
-
"step": 2530
|
| 1575 |
-
},
|
| 1576 |
-
{
|
| 1577 |
-
"epoch": 4.06,
|
| 1578 |
-
"learning_rate": 6.480298189563366e-05,
|
| 1579 |
-
"loss": 0.0017,
|
| 1580 |
-
"step": 2540
|
| 1581 |
-
},
|
| 1582 |
-
{
|
| 1583 |
-
"epoch": 4.07,
|
| 1584 |
-
"learning_rate": 6.427050053248137e-05,
|
| 1585 |
-
"loss": 0.0188,
|
| 1586 |
-
"step": 2550
|
| 1587 |
-
},
|
| 1588 |
-
{
|
| 1589 |
-
"epoch": 4.09,
|
| 1590 |
-
"learning_rate": 6.373801916932908e-05,
|
| 1591 |
-
"loss": 0.0076,
|
| 1592 |
-
"step": 2560
|
| 1593 |
-
},
|
| 1594 |
-
{
|
| 1595 |
-
"epoch": 4.11,
|
| 1596 |
-
"learning_rate": 6.320553780617678e-05,
|
| 1597 |
-
"loss": 0.002,
|
| 1598 |
-
"step": 2570
|
| 1599 |
-
},
|
| 1600 |
-
{
|
| 1601 |
-
"epoch": 4.12,
|
| 1602 |
-
"learning_rate": 6.26730564430245e-05,
|
| 1603 |
-
"loss": 0.0158,
|
| 1604 |
-
"step": 2580
|
| 1605 |
-
},
|
| 1606 |
-
{
|
| 1607 |
-
"epoch": 4.14,
|
| 1608 |
-
"learning_rate": 6.214057507987222e-05,
|
| 1609 |
-
"loss": 0.0025,
|
| 1610 |
-
"step": 2590
|
| 1611 |
-
},
|
| 1612 |
-
{
|
| 1613 |
-
"epoch": 4.15,
|
| 1614 |
-
"learning_rate": 6.160809371671991e-05,
|
| 1615 |
-
"loss": 0.0189,
|
| 1616 |
-
"step": 2600
|
| 1617 |
-
},
|
| 1618 |
-
{
|
| 1619 |
-
"epoch": 4.17,
|
| 1620 |
-
"learning_rate": 6.107561235356763e-05,
|
| 1621 |
-
"loss": 0.0007,
|
| 1622 |
-
"step": 2610
|
| 1623 |
-
},
|
| 1624 |
-
{
|
| 1625 |
-
"epoch": 4.19,
|
| 1626 |
-
"learning_rate": 6.054313099041534e-05,
|
| 1627 |
-
"loss": 0.0008,
|
| 1628 |
-
"step": 2620
|
| 1629 |
-
},
|
| 1630 |
-
{
|
| 1631 |
-
"epoch": 4.2,
|
| 1632 |
-
"learning_rate": 6.001064962726305e-05,
|
| 1633 |
-
"loss": 0.0007,
|
| 1634 |
-
"step": 2630
|
| 1635 |
-
},
|
| 1636 |
-
{
|
| 1637 |
-
"epoch": 4.22,
|
| 1638 |
-
"learning_rate": 5.9478168264110756e-05,
|
| 1639 |
-
"loss": 0.0155,
|
| 1640 |
-
"step": 2640
|
| 1641 |
-
},
|
| 1642 |
-
{
|
| 1643 |
-
"epoch": 4.23,
|
| 1644 |
-
"learning_rate": 5.8945686900958475e-05,
|
| 1645 |
-
"loss": 0.0005,
|
| 1646 |
-
"step": 2650
|
| 1647 |
-
},
|
| 1648 |
-
{
|
| 1649 |
-
"epoch": 4.25,
|
| 1650 |
-
"learning_rate": 5.841320553780618e-05,
|
| 1651 |
-
"loss": 0.0061,
|
| 1652 |
-
"step": 2660
|
| 1653 |
-
},
|
| 1654 |
-
{
|
| 1655 |
-
"epoch": 4.27,
|
| 1656 |
-
"learning_rate": 5.7880724174653886e-05,
|
| 1657 |
-
"loss": 0.0009,
|
| 1658 |
-
"step": 2670
|
| 1659 |
-
},
|
| 1660 |
-
{
|
| 1661 |
-
"epoch": 4.28,
|
| 1662 |
-
"learning_rate": 5.73482428115016e-05,
|
| 1663 |
-
"loss": 0.0024,
|
| 1664 |
-
"step": 2680
|
| 1665 |
-
},
|
| 1666 |
-
{
|
| 1667 |
-
"epoch": 4.3,
|
| 1668 |
-
"learning_rate": 5.681576144834932e-05,
|
| 1669 |
-
"loss": 0.0009,
|
| 1670 |
-
"step": 2690
|
| 1671 |
-
},
|
| 1672 |
-
{
|
| 1673 |
-
"epoch": 4.31,
|
| 1674 |
-
"learning_rate": 5.628328008519702e-05,
|
| 1675 |
-
"loss": 0.0007,
|
| 1676 |
-
"step": 2700
|
| 1677 |
-
},
|
| 1678 |
-
{
|
| 1679 |
-
"epoch": 4.33,
|
| 1680 |
-
"learning_rate": 5.575079872204473e-05,
|
| 1681 |
-
"loss": 0.0021,
|
| 1682 |
-
"step": 2710
|
| 1683 |
-
},
|
| 1684 |
-
{
|
| 1685 |
-
"epoch": 4.35,
|
| 1686 |
-
"learning_rate": 5.521831735889245e-05,
|
| 1687 |
-
"loss": 0.017,
|
| 1688 |
-
"step": 2720
|
| 1689 |
-
},
|
| 1690 |
-
{
|
| 1691 |
-
"epoch": 4.36,
|
| 1692 |
-
"learning_rate": 5.468583599574015e-05,
|
| 1693 |
-
"loss": 0.0009,
|
| 1694 |
-
"step": 2730
|
| 1695 |
-
},
|
| 1696 |
-
{
|
| 1697 |
-
"epoch": 4.38,
|
| 1698 |
-
"learning_rate": 5.415335463258786e-05,
|
| 1699 |
-
"loss": 0.0099,
|
| 1700 |
-
"step": 2740
|
| 1701 |
-
},
|
| 1702 |
-
{
|
| 1703 |
-
"epoch": 4.39,
|
| 1704 |
-
"learning_rate": 5.362087326943557e-05,
|
| 1705 |
-
"loss": 0.0004,
|
| 1706 |
-
"step": 2750
|
| 1707 |
-
},
|
| 1708 |
-
{
|
| 1709 |
-
"epoch": 4.41,
|
| 1710 |
-
"learning_rate": 5.308839190628329e-05,
|
| 1711 |
-
"loss": 0.0016,
|
| 1712 |
-
"step": 2760
|
| 1713 |
-
},
|
| 1714 |
-
{
|
| 1715 |
-
"epoch": 4.42,
|
| 1716 |
-
"learning_rate": 5.2555910543130994e-05,
|
| 1717 |
-
"loss": 0.0012,
|
| 1718 |
-
"step": 2770
|
| 1719 |
-
},
|
| 1720 |
-
{
|
| 1721 |
-
"epoch": 4.44,
|
| 1722 |
-
"learning_rate": 5.20234291799787e-05,
|
| 1723 |
-
"loss": 0.003,
|
| 1724 |
-
"step": 2780
|
| 1725 |
-
},
|
| 1726 |
-
{
|
| 1727 |
-
"epoch": 4.46,
|
| 1728 |
-
"learning_rate": 5.149094781682642e-05,
|
| 1729 |
-
"loss": 0.0004,
|
| 1730 |
-
"step": 2790
|
| 1731 |
-
},
|
| 1732 |
-
{
|
| 1733 |
-
"epoch": 4.47,
|
| 1734 |
-
"learning_rate": 5.095846645367412e-05,
|
| 1735 |
-
"loss": 0.0004,
|
| 1736 |
-
"step": 2800
|
| 1737 |
-
},
|
| 1738 |
-
{
|
| 1739 |
-
"epoch": 4.49,
|
| 1740 |
-
"learning_rate": 5.0425985090521835e-05,
|
| 1741 |
-
"loss": 0.0198,
|
| 1742 |
-
"step": 2810
|
| 1743 |
-
},
|
| 1744 |
-
{
|
| 1745 |
-
"epoch": 4.5,
|
| 1746 |
-
"learning_rate": 4.989350372736955e-05,
|
| 1747 |
-
"loss": 0.0004,
|
| 1748 |
-
"step": 2820
|
| 1749 |
-
},
|
| 1750 |
-
{
|
| 1751 |
-
"epoch": 4.52,
|
| 1752 |
-
"learning_rate": 4.936102236421725e-05,
|
| 1753 |
-
"loss": 0.0003,
|
| 1754 |
-
"step": 2830
|
| 1755 |
-
},
|
| 1756 |
-
{
|
| 1757 |
-
"epoch": 4.54,
|
| 1758 |
-
"learning_rate": 4.8828541001064965e-05,
|
| 1759 |
-
"loss": 0.0006,
|
| 1760 |
-
"step": 2840
|
| 1761 |
-
},
|
| 1762 |
-
{
|
| 1763 |
-
"epoch": 4.55,
|
| 1764 |
-
"learning_rate": 4.829605963791268e-05,
|
| 1765 |
-
"loss": 0.0004,
|
| 1766 |
-
"step": 2850
|
| 1767 |
-
},
|
| 1768 |
-
{
|
| 1769 |
-
"epoch": 4.57,
|
| 1770 |
-
"learning_rate": 4.776357827476038e-05,
|
| 1771 |
-
"loss": 0.0223,
|
| 1772 |
-
"step": 2860
|
| 1773 |
-
},
|
| 1774 |
-
{
|
| 1775 |
-
"epoch": 4.58,
|
| 1776 |
-
"learning_rate": 4.7231096911608095e-05,
|
| 1777 |
-
"loss": 0.0005,
|
| 1778 |
-
"step": 2870
|
| 1779 |
-
},
|
| 1780 |
-
{
|
| 1781 |
-
"epoch": 4.6,
|
| 1782 |
-
"learning_rate": 4.669861554845581e-05,
|
| 1783 |
-
"loss": 0.0006,
|
| 1784 |
-
"step": 2880
|
| 1785 |
-
},
|
| 1786 |
-
{
|
| 1787 |
-
"epoch": 4.62,
|
| 1788 |
-
"learning_rate": 4.616613418530352e-05,
|
| 1789 |
-
"loss": 0.0006,
|
| 1790 |
-
"step": 2890
|
| 1791 |
-
},
|
| 1792 |
-
{
|
| 1793 |
-
"epoch": 4.63,
|
| 1794 |
-
"learning_rate": 4.563365282215123e-05,
|
| 1795 |
-
"loss": 0.0004,
|
| 1796 |
-
"step": 2900
|
| 1797 |
-
},
|
| 1798 |
-
{
|
| 1799 |
-
"epoch": 4.65,
|
| 1800 |
-
"learning_rate": 4.5101171458998936e-05,
|
| 1801 |
-
"loss": 0.0004,
|
| 1802 |
-
"step": 2910
|
| 1803 |
-
},
|
| 1804 |
-
{
|
| 1805 |
-
"epoch": 4.66,
|
| 1806 |
-
"learning_rate": 4.456869009584665e-05,
|
| 1807 |
-
"loss": 0.0012,
|
| 1808 |
-
"step": 2920
|
| 1809 |
-
},
|
| 1810 |
-
{
|
| 1811 |
-
"epoch": 4.68,
|
| 1812 |
-
"learning_rate": 4.4036208732694354e-05,
|
| 1813 |
-
"loss": 0.0033,
|
| 1814 |
-
"step": 2930
|
| 1815 |
-
},
|
| 1816 |
-
{
|
| 1817 |
-
"epoch": 4.7,
|
| 1818 |
-
"learning_rate": 4.3503727369542066e-05,
|
| 1819 |
-
"loss": 0.0007,
|
| 1820 |
-
"step": 2940
|
| 1821 |
-
},
|
| 1822 |
-
{
|
| 1823 |
-
"epoch": 4.71,
|
| 1824 |
-
"learning_rate": 4.297124600638978e-05,
|
| 1825 |
-
"loss": 0.0134,
|
| 1826 |
-
"step": 2950
|
| 1827 |
-
},
|
| 1828 |
-
{
|
| 1829 |
-
"epoch": 4.73,
|
| 1830 |
-
"learning_rate": 4.243876464323749e-05,
|
| 1831 |
-
"loss": 0.0005,
|
| 1832 |
-
"step": 2960
|
| 1833 |
-
},
|
| 1834 |
-
{
|
| 1835 |
-
"epoch": 4.74,
|
| 1836 |
-
"learning_rate": 4.19062832800852e-05,
|
| 1837 |
-
"loss": 0.0005,
|
| 1838 |
-
"step": 2970
|
| 1839 |
-
},
|
| 1840 |
-
{
|
| 1841 |
-
"epoch": 4.76,
|
| 1842 |
-
"learning_rate": 4.137380191693291e-05,
|
| 1843 |
-
"loss": 0.0122,
|
| 1844 |
-
"step": 2980
|
| 1845 |
-
},
|
| 1846 |
-
{
|
| 1847 |
-
"epoch": 4.78,
|
| 1848 |
-
"learning_rate": 4.084132055378062e-05,
|
| 1849 |
-
"loss": 0.0052,
|
| 1850 |
-
"step": 2990
|
| 1851 |
-
},
|
| 1852 |
-
{
|
| 1853 |
-
"epoch": 4.79,
|
| 1854 |
-
"learning_rate": 4.0308839190628325e-05,
|
| 1855 |
-
"loss": 0.0004,
|
| 1856 |
-
"step": 3000
|
| 1857 |
-
},
|
| 1858 |
-
{
|
| 1859 |
-
"epoch": 4.81,
|
| 1860 |
-
"learning_rate": 3.9776357827476044e-05,
|
| 1861 |
-
"loss": 0.0112,
|
| 1862 |
-
"step": 3010
|
| 1863 |
-
},
|
| 1864 |
-
{
|
| 1865 |
-
"epoch": 4.82,
|
| 1866 |
-
"learning_rate": 3.9243876464323756e-05,
|
| 1867 |
-
"loss": 0.0019,
|
| 1868 |
-
"step": 3020
|
| 1869 |
-
},
|
| 1870 |
-
{
|
| 1871 |
-
"epoch": 4.84,
|
| 1872 |
-
"learning_rate": 3.871139510117146e-05,
|
| 1873 |
-
"loss": 0.0016,
|
| 1874 |
-
"step": 3030
|
| 1875 |
-
},
|
| 1876 |
-
{
|
| 1877 |
-
"epoch": 4.86,
|
| 1878 |
-
"learning_rate": 3.8178913738019174e-05,
|
| 1879 |
-
"loss": 0.0076,
|
| 1880 |
-
"step": 3040
|
| 1881 |
-
},
|
| 1882 |
-
{
|
| 1883 |
-
"epoch": 4.87,
|
| 1884 |
-
"learning_rate": 3.764643237486688e-05,
|
| 1885 |
-
"loss": 0.0007,
|
| 1886 |
-
"step": 3050
|
| 1887 |
-
},
|
| 1888 |
-
{
|
| 1889 |
-
"epoch": 4.89,
|
| 1890 |
-
"learning_rate": 3.711395101171459e-05,
|
| 1891 |
-
"loss": 0.0006,
|
| 1892 |
-
"step": 3060
|
| 1893 |
-
},
|
| 1894 |
-
{
|
| 1895 |
-
"epoch": 4.9,
|
| 1896 |
-
"learning_rate": 3.65814696485623e-05,
|
| 1897 |
-
"loss": 0.0025,
|
| 1898 |
-
"step": 3070
|
| 1899 |
-
},
|
| 1900 |
-
{
|
| 1901 |
-
"epoch": 4.92,
|
| 1902 |
-
"learning_rate": 3.6048988285410015e-05,
|
| 1903 |
-
"loss": 0.0008,
|
| 1904 |
-
"step": 3080
|
| 1905 |
-
},
|
| 1906 |
-
{
|
| 1907 |
-
"epoch": 4.94,
|
| 1908 |
-
"learning_rate": 3.551650692225773e-05,
|
| 1909 |
-
"loss": 0.0004,
|
| 1910 |
-
"step": 3090
|
| 1911 |
-
},
|
| 1912 |
-
{
|
| 1913 |
-
"epoch": 4.95,
|
| 1914 |
-
"learning_rate": 3.498402555910543e-05,
|
| 1915 |
-
"loss": 0.0003,
|
| 1916 |
-
"step": 3100
|
| 1917 |
-
},
|
| 1918 |
-
{
|
| 1919 |
-
"epoch": 4.97,
|
| 1920 |
-
"learning_rate": 3.4451544195953145e-05,
|
| 1921 |
-
"loss": 0.0003,
|
| 1922 |
-
"step": 3110
|
| 1923 |
-
},
|
| 1924 |
-
{
|
| 1925 |
-
"epoch": 4.98,
|
| 1926 |
-
"learning_rate": 3.391906283280085e-05,
|
| 1927 |
-
"loss": 0.0004,
|
| 1928 |
-
"step": 3120
|
| 1929 |
-
},
|
| 1930 |
-
{
|
| 1931 |
-
"epoch": 5.0,
|
| 1932 |
-
"learning_rate": 3.338658146964856e-05,
|
| 1933 |
-
"loss": 0.0004,
|
| 1934 |
-
"step": 3130
|
| 1935 |
-
},
|
| 1936 |
-
{
|
| 1937 |
-
"epoch": 5.0,
|
| 1938 |
-
"eval_accuracy": 0.8601036269430051,
|
| 1939 |
-
"eval_f1": 0.8667958211367178,
|
| 1940 |
-
"eval_loss": 0.5794870853424072,
|
| 1941 |
-
"eval_precision": 0.878678756476684,
|
| 1942 |
-
"eval_recall": 0.8601036269430051,
|
| 1943 |
-
"eval_runtime": 1.3078,
|
| 1944 |
-
"eval_samples_per_second": 147.574,
|
| 1945 |
-
"eval_steps_per_second": 9.94,
|
| 1946 |
-
"step": 3130
|
| 1947 |
-
},
|
| 1948 |
-
{
|
| 1949 |
-
"epoch": 5.02,
|
| 1950 |
-
"learning_rate": 3.2854100106496274e-05,
|
| 1951 |
-
"loss": 0.0007,
|
| 1952 |
-
"step": 3140
|
| 1953 |
-
},
|
| 1954 |
-
{
|
| 1955 |
-
"epoch": 5.03,
|
| 1956 |
-
"learning_rate": 3.2321618743343987e-05,
|
| 1957 |
-
"loss": 0.0003,
|
| 1958 |
-
"step": 3150
|
| 1959 |
-
},
|
| 1960 |
-
{
|
| 1961 |
-
"epoch": 5.05,
|
| 1962 |
-
"learning_rate": 3.17891373801917e-05,
|
| 1963 |
-
"loss": 0.0005,
|
| 1964 |
-
"step": 3160
|
| 1965 |
-
},
|
| 1966 |
-
{
|
| 1967 |
-
"epoch": 5.06,
|
| 1968 |
-
"learning_rate": 3.1256656017039404e-05,
|
| 1969 |
-
"loss": 0.0005,
|
| 1970 |
-
"step": 3170
|
| 1971 |
-
},
|
| 1972 |
-
{
|
| 1973 |
-
"epoch": 5.08,
|
| 1974 |
-
"learning_rate": 3.0724174653887116e-05,
|
| 1975 |
-
"loss": 0.0004,
|
| 1976 |
-
"step": 3180
|
| 1977 |
-
},
|
| 1978 |
-
{
|
| 1979 |
-
"epoch": 5.1,
|
| 1980 |
-
"learning_rate": 3.0191693290734825e-05,
|
| 1981 |
-
"loss": 0.0002,
|
| 1982 |
-
"step": 3190
|
| 1983 |
-
},
|
| 1984 |
-
{
|
| 1985 |
-
"epoch": 5.11,
|
| 1986 |
-
"learning_rate": 2.9659211927582537e-05,
|
| 1987 |
-
"loss": 0.0003,
|
| 1988 |
-
"step": 3200
|
| 1989 |
-
},
|
| 1990 |
-
{
|
| 1991 |
-
"epoch": 5.13,
|
| 1992 |
-
"learning_rate": 2.912673056443025e-05,
|
| 1993 |
-
"loss": 0.0003,
|
| 1994 |
-
"step": 3210
|
| 1995 |
-
},
|
| 1996 |
-
{
|
| 1997 |
-
"epoch": 5.14,
|
| 1998 |
-
"learning_rate": 2.8594249201277955e-05,
|
| 1999 |
-
"loss": 0.0002,
|
| 2000 |
-
"step": 3220
|
| 2001 |
-
},
|
| 2002 |
-
{
|
| 2003 |
-
"epoch": 5.16,
|
| 2004 |
-
"learning_rate": 2.806176783812567e-05,
|
| 2005 |
-
"loss": 0.0003,
|
| 2006 |
-
"step": 3230
|
| 2007 |
-
},
|
| 2008 |
-
{
|
| 2009 |
-
"epoch": 5.18,
|
| 2010 |
-
"learning_rate": 2.7529286474973375e-05,
|
| 2011 |
-
"loss": 0.0004,
|
| 2012 |
-
"step": 3240
|
| 2013 |
-
},
|
| 2014 |
-
{
|
| 2015 |
-
"epoch": 5.19,
|
| 2016 |
-
"learning_rate": 2.6996805111821088e-05,
|
| 2017 |
-
"loss": 0.0003,
|
| 2018 |
-
"step": 3250
|
| 2019 |
-
},
|
| 2020 |
-
{
|
| 2021 |
-
"epoch": 5.21,
|
| 2022 |
-
"learning_rate": 2.6464323748668796e-05,
|
| 2023 |
-
"loss": 0.0006,
|
| 2024 |
-
"step": 3260
|
| 2025 |
-
},
|
| 2026 |
-
{
|
| 2027 |
-
"epoch": 5.22,
|
| 2028 |
-
"learning_rate": 2.593184238551651e-05,
|
| 2029 |
-
"loss": 0.0003,
|
| 2030 |
-
"step": 3270
|
| 2031 |
-
},
|
| 2032 |
-
{
|
| 2033 |
-
"epoch": 5.24,
|
| 2034 |
-
"learning_rate": 2.539936102236422e-05,
|
| 2035 |
-
"loss": 0.0003,
|
| 2036 |
-
"step": 3280
|
| 2037 |
-
},
|
| 2038 |
-
{
|
| 2039 |
-
"epoch": 5.26,
|
| 2040 |
-
"learning_rate": 2.486687965921193e-05,
|
| 2041 |
-
"loss": 0.0003,
|
| 2042 |
-
"step": 3290
|
| 2043 |
-
},
|
| 2044 |
-
{
|
| 2045 |
-
"epoch": 5.27,
|
| 2046 |
-
"learning_rate": 2.4334398296059638e-05,
|
| 2047 |
-
"loss": 0.0002,
|
| 2048 |
-
"step": 3300
|
| 2049 |
-
},
|
| 2050 |
-
{
|
| 2051 |
-
"epoch": 5.29,
|
| 2052 |
-
"learning_rate": 2.380191693290735e-05,
|
| 2053 |
-
"loss": 0.0004,
|
| 2054 |
-
"step": 3310
|
| 2055 |
-
},
|
| 2056 |
-
{
|
| 2057 |
-
"epoch": 5.3,
|
| 2058 |
-
"learning_rate": 2.326943556975506e-05,
|
| 2059 |
-
"loss": 0.0003,
|
| 2060 |
-
"step": 3320
|
| 2061 |
-
},
|
| 2062 |
-
{
|
| 2063 |
-
"epoch": 5.32,
|
| 2064 |
-
"learning_rate": 2.273695420660277e-05,
|
| 2065 |
-
"loss": 0.0002,
|
| 2066 |
-
"step": 3330
|
| 2067 |
-
},
|
| 2068 |
-
{
|
| 2069 |
-
"epoch": 5.34,
|
| 2070 |
-
"learning_rate": 2.220447284345048e-05,
|
| 2071 |
-
"loss": 0.0002,
|
| 2072 |
-
"step": 3340
|
| 2073 |
-
},
|
| 2074 |
-
{
|
| 2075 |
-
"epoch": 5.35,
|
| 2076 |
-
"learning_rate": 2.167199148029819e-05,
|
| 2077 |
-
"loss": 0.0003,
|
| 2078 |
-
"step": 3350
|
| 2079 |
-
},
|
| 2080 |
-
{
|
| 2081 |
-
"epoch": 5.37,
|
| 2082 |
-
"learning_rate": 2.11395101171459e-05,
|
| 2083 |
-
"loss": 0.0002,
|
| 2084 |
-
"step": 3360
|
| 2085 |
-
},
|
| 2086 |
-
{
|
| 2087 |
-
"epoch": 5.38,
|
| 2088 |
-
"learning_rate": 2.0607028753993613e-05,
|
| 2089 |
-
"loss": 0.0002,
|
| 2090 |
-
"step": 3370
|
| 2091 |
-
},
|
| 2092 |
-
{
|
| 2093 |
-
"epoch": 5.4,
|
| 2094 |
-
"learning_rate": 2.007454739084132e-05,
|
| 2095 |
-
"loss": 0.0002,
|
| 2096 |
-
"step": 3380
|
| 2097 |
-
},
|
| 2098 |
-
{
|
| 2099 |
-
"epoch": 5.42,
|
| 2100 |
-
"learning_rate": 1.9542066027689034e-05,
|
| 2101 |
-
"loss": 0.0002,
|
| 2102 |
-
"step": 3390
|
| 2103 |
-
},
|
| 2104 |
-
{
|
| 2105 |
-
"epoch": 5.43,
|
| 2106 |
-
"learning_rate": 1.9009584664536742e-05,
|
| 2107 |
-
"loss": 0.0003,
|
| 2108 |
-
"step": 3400
|
| 2109 |
-
},
|
| 2110 |
-
{
|
| 2111 |
-
"epoch": 5.45,
|
| 2112 |
-
"learning_rate": 1.847710330138445e-05,
|
| 2113 |
-
"loss": 0.0004,
|
| 2114 |
-
"step": 3410
|
| 2115 |
-
},
|
| 2116 |
-
{
|
| 2117 |
-
"epoch": 5.46,
|
| 2118 |
-
"learning_rate": 1.7944621938232163e-05,
|
| 2119 |
-
"loss": 0.0003,
|
| 2120 |
-
"step": 3420
|
| 2121 |
-
},
|
| 2122 |
-
{
|
| 2123 |
-
"epoch": 5.48,
|
| 2124 |
-
"learning_rate": 1.7412140575079875e-05,
|
| 2125 |
-
"loss": 0.0003,
|
| 2126 |
-
"step": 3430
|
| 2127 |
-
},
|
| 2128 |
-
{
|
| 2129 |
-
"epoch": 5.5,
|
| 2130 |
-
"learning_rate": 1.6879659211927584e-05,
|
| 2131 |
-
"loss": 0.0003,
|
| 2132 |
-
"step": 3440
|
| 2133 |
-
},
|
| 2134 |
-
{
|
| 2135 |
-
"epoch": 5.51,
|
| 2136 |
-
"learning_rate": 1.6347177848775293e-05,
|
| 2137 |
-
"loss": 0.0002,
|
| 2138 |
-
"step": 3450
|
| 2139 |
-
},
|
| 2140 |
-
{
|
| 2141 |
-
"epoch": 5.53,
|
| 2142 |
-
"learning_rate": 1.5814696485623005e-05,
|
| 2143 |
-
"loss": 0.0036,
|
| 2144 |
-
"step": 3460
|
| 2145 |
-
},
|
| 2146 |
-
{
|
| 2147 |
-
"epoch": 5.54,
|
| 2148 |
-
"learning_rate": 1.5282215122470714e-05,
|
| 2149 |
-
"loss": 0.0003,
|
| 2150 |
-
"step": 3470
|
| 2151 |
-
},
|
| 2152 |
-
{
|
| 2153 |
-
"epoch": 5.56,
|
| 2154 |
-
"learning_rate": 1.4749733759318424e-05,
|
| 2155 |
-
"loss": 0.0003,
|
| 2156 |
-
"step": 3480
|
| 2157 |
-
},
|
| 2158 |
-
{
|
| 2159 |
-
"epoch": 5.58,
|
| 2160 |
-
"learning_rate": 1.4217252396166134e-05,
|
| 2161 |
-
"loss": 0.0003,
|
| 2162 |
-
"step": 3490
|
| 2163 |
-
},
|
| 2164 |
-
{
|
| 2165 |
-
"epoch": 5.59,
|
| 2166 |
-
"learning_rate": 1.3684771033013847e-05,
|
| 2167 |
-
"loss": 0.0003,
|
| 2168 |
-
"step": 3500
|
| 2169 |
-
},
|
| 2170 |
-
{
|
| 2171 |
-
"epoch": 5.61,
|
| 2172 |
-
"learning_rate": 1.3152289669861555e-05,
|
| 2173 |
-
"loss": 0.0002,
|
| 2174 |
-
"step": 3510
|
| 2175 |
-
},
|
| 2176 |
-
{
|
| 2177 |
-
"epoch": 5.62,
|
| 2178 |
-
"learning_rate": 1.2619808306709266e-05,
|
| 2179 |
-
"loss": 0.0003,
|
| 2180 |
-
"step": 3520
|
| 2181 |
-
},
|
| 2182 |
-
{
|
| 2183 |
-
"epoch": 5.64,
|
| 2184 |
-
"learning_rate": 1.2087326943556976e-05,
|
| 2185 |
-
"loss": 0.0002,
|
| 2186 |
-
"step": 3530
|
| 2187 |
-
},
|
| 2188 |
-
{
|
| 2189 |
-
"epoch": 5.65,
|
| 2190 |
-
"learning_rate": 1.1554845580404687e-05,
|
| 2191 |
-
"loss": 0.0002,
|
| 2192 |
-
"step": 3540
|
| 2193 |
-
},
|
| 2194 |
-
{
|
| 2195 |
-
"epoch": 5.67,
|
| 2196 |
-
"learning_rate": 1.1022364217252397e-05,
|
| 2197 |
-
"loss": 0.0006,
|
| 2198 |
-
"step": 3550
|
| 2199 |
-
},
|
| 2200 |
-
{
|
| 2201 |
-
"epoch": 5.69,
|
| 2202 |
-
"learning_rate": 1.0489882854100107e-05,
|
| 2203 |
-
"loss": 0.0002,
|
| 2204 |
-
"step": 3560
|
| 2205 |
-
},
|
| 2206 |
-
{
|
| 2207 |
-
"epoch": 5.7,
|
| 2208 |
-
"learning_rate": 9.957401490947816e-06,
|
| 2209 |
-
"loss": 0.0002,
|
| 2210 |
-
"step": 3570
|
| 2211 |
-
},
|
| 2212 |
-
{
|
| 2213 |
-
"epoch": 5.72,
|
| 2214 |
-
"learning_rate": 9.424920127795528e-06,
|
| 2215 |
-
"loss": 0.0002,
|
| 2216 |
-
"step": 3580
|
| 2217 |
-
},
|
| 2218 |
-
{
|
| 2219 |
-
"epoch": 5.73,
|
| 2220 |
-
"learning_rate": 8.892438764643237e-06,
|
| 2221 |
-
"loss": 0.0002,
|
| 2222 |
-
"step": 3590
|
| 2223 |
-
},
|
| 2224 |
-
{
|
| 2225 |
-
"epoch": 5.75,
|
| 2226 |
-
"learning_rate": 8.359957401490947e-06,
|
| 2227 |
-
"loss": 0.0002,
|
| 2228 |
-
"step": 3600
|
| 2229 |
-
},
|
| 2230 |
-
{
|
| 2231 |
-
"epoch": 5.77,
|
| 2232 |
-
"learning_rate": 7.82747603833866e-06,
|
| 2233 |
-
"loss": 0.0002,
|
| 2234 |
-
"step": 3610
|
| 2235 |
-
},
|
| 2236 |
-
{
|
| 2237 |
-
"epoch": 5.78,
|
| 2238 |
-
"learning_rate": 7.294994675186369e-06,
|
| 2239 |
-
"loss": 0.0002,
|
| 2240 |
-
"step": 3620
|
| 2241 |
-
},
|
| 2242 |
-
{
|
| 2243 |
-
"epoch": 5.8,
|
| 2244 |
-
"learning_rate": 6.762513312034079e-06,
|
| 2245 |
-
"loss": 0.0002,
|
| 2246 |
-
"step": 3630
|
| 2247 |
-
},
|
| 2248 |
-
{
|
| 2249 |
-
"epoch": 5.81,
|
| 2250 |
-
"learning_rate": 6.230031948881789e-06,
|
| 2251 |
-
"loss": 0.0002,
|
| 2252 |
-
"step": 3640
|
| 2253 |
-
},
|
| 2254 |
-
{
|
| 2255 |
-
"epoch": 5.83,
|
| 2256 |
-
"learning_rate": 5.6975505857295e-06,
|
| 2257 |
-
"loss": 0.0002,
|
| 2258 |
-
"step": 3650
|
| 2259 |
-
},
|
| 2260 |
-
{
|
| 2261 |
-
"epoch": 5.85,
|
| 2262 |
-
"learning_rate": 5.16506922257721e-06,
|
| 2263 |
-
"loss": 0.0002,
|
| 2264 |
-
"step": 3660
|
| 2265 |
-
},
|
| 2266 |
-
{
|
| 2267 |
-
"epoch": 5.86,
|
| 2268 |
-
"learning_rate": 4.6325878594249205e-06,
|
| 2269 |
-
"loss": 0.0002,
|
| 2270 |
-
"step": 3670
|
| 2271 |
-
},
|
| 2272 |
-
{
|
| 2273 |
-
"epoch": 5.88,
|
| 2274 |
-
"learning_rate": 4.100106496272631e-06,
|
| 2275 |
-
"loss": 0.0003,
|
| 2276 |
-
"step": 3680
|
| 2277 |
-
},
|
| 2278 |
-
{
|
| 2279 |
-
"epoch": 5.89,
|
| 2280 |
-
"learning_rate": 3.5676251331203413e-06,
|
| 2281 |
-
"loss": 0.0002,
|
| 2282 |
-
"step": 3690
|
| 2283 |
-
},
|
| 2284 |
-
{
|
| 2285 |
-
"epoch": 5.91,
|
| 2286 |
-
"learning_rate": 3.0351437699680513e-06,
|
| 2287 |
-
"loss": 0.0002,
|
| 2288 |
-
"step": 3700
|
| 2289 |
-
},
|
| 2290 |
-
{
|
| 2291 |
-
"epoch": 5.93,
|
| 2292 |
-
"learning_rate": 2.5026624068157617e-06,
|
| 2293 |
-
"loss": 0.0002,
|
| 2294 |
-
"step": 3710
|
| 2295 |
-
},
|
| 2296 |
-
{
|
| 2297 |
-
"epoch": 5.94,
|
| 2298 |
-
"learning_rate": 1.9701810436634718e-06,
|
| 2299 |
-
"loss": 0.0003,
|
| 2300 |
-
"step": 3720
|
| 2301 |
-
},
|
| 2302 |
-
{
|
| 2303 |
-
"epoch": 5.96,
|
| 2304 |
-
"learning_rate": 1.4376996805111822e-06,
|
| 2305 |
-
"loss": 0.0002,
|
| 2306 |
-
"step": 3730
|
| 2307 |
-
},
|
| 2308 |
-
{
|
| 2309 |
-
"epoch": 5.97,
|
| 2310 |
-
"learning_rate": 9.052183173588925e-07,
|
| 2311 |
-
"loss": 0.0002,
|
| 2312 |
-
"step": 3740
|
| 2313 |
-
},
|
| 2314 |
-
{
|
| 2315 |
-
"epoch": 5.99,
|
| 2316 |
-
"learning_rate": 3.727369542066028e-07,
|
| 2317 |
-
"loss": 0.0002,
|
| 2318 |
-
"step": 3750
|
| 2319 |
-
},
|
| 2320 |
-
{
|
| 2321 |
-
"epoch": 6.0,
|
| 2322 |
-
"eval_accuracy": 0.8652849740932642,
|
| 2323 |
-
"eval_f1": 0.872574161558089,
|
| 2324 |
-
"eval_loss": 0.5782872438430786,
|
| 2325 |
-
"eval_precision": 0.8851064570906083,
|
| 2326 |
-
"eval_recall": 0.8652849740932642,
|
| 2327 |
-
"eval_runtime": 1.304,
|
| 2328 |
-
"eval_samples_per_second": 148.006,
|
| 2329 |
-
"eval_steps_per_second": 9.969,
|
| 2330 |
-
"step": 3756
|
| 2331 |
-
},
|
| 2332 |
-
{
|
| 2333 |
-
"epoch": 6.0,
|
| 2334 |
-
"step": 3756,
|
| 2335 |
-
"total_flos": 4.65670232933972e+18,
|
| 2336 |
-
"train_loss": 0.07091493732059752,
|
| 2337 |
-
"train_runtime": 1262.5493,
|
| 2338 |
-
"train_samples_per_second": 47.594,
|
| 2339 |
-
"train_steps_per_second": 2.975
|
| 2340 |
}
|
| 2341 |
],
|
| 2342 |
-
"max_steps":
|
| 2343 |
-
"num_train_epochs":
|
| 2344 |
-
"total_flos":
|
| 2345 |
"trial_name": null,
|
| 2346 |
"trial_params": null
|
| 2347 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.8860103626943006,
|
| 3 |
+
"best_model_checkpoint": "./vit-focal-skin/checkpoint-1252",
|
| 4 |
+
"epoch": 4.0,
|
| 5 |
+
"global_step": 2504,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
| 9 |
"log_history": [
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10 |
{
|
| 11 |
"epoch": 0.16,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
"learning_rate": 0.00019201277955271565,
|
| 13 |
+
"loss": 0.7905,
|
| 14 |
+
"step": 100
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 15 |
},
|
| 16 |
{
|
| 17 |
"epoch": 0.32,
|
| 18 |
+
"learning_rate": 0.00018402555910543132,
|
| 19 |
+
"loss": 0.6162,
|
| 20 |
"step": 200
|
| 21 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 22 |
{
|
| 23 |
"epoch": 0.48,
|
| 24 |
+
"learning_rate": 0.000176038338658147,
|
| 25 |
+
"loss": 0.5709,
|
| 26 |
"step": 300
|
| 27 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 28 |
{
|
| 29 |
"epoch": 0.64,
|
| 30 |
+
"learning_rate": 0.00016805111821086263,
|
| 31 |
+
"loss": 0.5101,
|
| 32 |
"step": 400
|
| 33 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 34 |
{
|
| 35 |
"epoch": 0.8,
|
| 36 |
+
"learning_rate": 0.0001600638977635783,
|
| 37 |
+
"loss": 0.4966,
|
| 38 |
"step": 500
|
| 39 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 40 |
{
|
| 41 |
"epoch": 0.96,
|
| 42 |
+
"learning_rate": 0.00015207667731629394,
|
| 43 |
+
"loss": 0.4563,
|
| 44 |
"step": 600
|
| 45 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 46 |
{
|
| 47 |
"epoch": 1.0,
|
| 48 |
+
"eval_accuracy": 0.8341968911917098,
|
| 49 |
+
"eval_f1": 0.8331282144797781,
|
| 50 |
+
"eval_loss": 0.40268635749816895,
|
| 51 |
+
"eval_precision": 0.8791963514680484,
|
| 52 |
+
"eval_recall": 0.8341968911917098,
|
| 53 |
+
"eval_runtime": 1.5386,
|
| 54 |
+
"eval_samples_per_second": 125.44,
|
| 55 |
+
"eval_steps_per_second": 16.249,
|
| 56 |
"step": 626
|
| 57 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 58 |
{
|
| 59 |
"epoch": 1.12,
|
| 60 |
+
"learning_rate": 0.00014408945686900958,
|
| 61 |
+
"loss": 0.385,
|
| 62 |
"step": 700
|
| 63 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 64 |
{
|
| 65 |
"epoch": 1.28,
|
| 66 |
+
"learning_rate": 0.00013610223642172525,
|
| 67 |
+
"loss": 0.3621,
|
| 68 |
"step": 800
|
| 69 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 70 |
{
|
| 71 |
"epoch": 1.44,
|
| 72 |
+
"learning_rate": 0.00012811501597444092,
|
| 73 |
+
"loss": 0.3624,
|
| 74 |
"step": 900
|
| 75 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 76 |
{
|
| 77 |
"epoch": 1.6,
|
| 78 |
+
"learning_rate": 0.00012012779552715656,
|
| 79 |
+
"loss": 0.3403,
|
| 80 |
"step": 1000
|
| 81 |
},
|
| 82 |
{
|
| 83 |
+
"epoch": 1.76,
|
| 84 |
+
"learning_rate": 0.00011214057507987221,
|
| 85 |
+
"loss": 0.3293,
|
| 86 |
+
"step": 1100
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 87 |
},
|
| 88 |
{
|
| 89 |
+
"epoch": 1.92,
|
| 90 |
+
"learning_rate": 0.00010415335463258787,
|
| 91 |
+
"loss": 0.2954,
|
| 92 |
+
"step": 1200
|
| 93 |
},
|
| 94 |
{
|
| 95 |
+
"epoch": 2.0,
|
| 96 |
+
"eval_accuracy": 0.8860103626943006,
|
| 97 |
+
"eval_f1": 0.8891954516071332,
|
| 98 |
+
"eval_loss": 0.3063604235649109,
|
| 99 |
+
"eval_precision": 0.8988400162775354,
|
| 100 |
+
"eval_recall": 0.8860103626943006,
|
| 101 |
+
"eval_runtime": 1.5206,
|
| 102 |
+
"eval_samples_per_second": 126.921,
|
| 103 |
+
"eval_steps_per_second": 16.441,
|
| 104 |
+
"step": 1252
|
| 105 |
},
|
| 106 |
{
|
| 107 |
+
"epoch": 2.08,
|
| 108 |
+
"learning_rate": 9.616613418530351e-05,
|
| 109 |
+
"loss": 0.2258,
|
| 110 |
+
"step": 1300
|
| 111 |
},
|
| 112 |
{
|
| 113 |
+
"epoch": 2.24,
|
| 114 |
+
"learning_rate": 8.817891373801918e-05,
|
| 115 |
+
"loss": 0.1583,
|
| 116 |
+
"step": 1400
|
| 117 |
},
|
| 118 |
{
|
| 119 |
+
"epoch": 2.4,
|
| 120 |
+
"learning_rate": 8.019169329073483e-05,
|
| 121 |
+
"loss": 0.1589,
|
| 122 |
+
"step": 1500
|
| 123 |
},
|
| 124 |
{
|
| 125 |
+
"epoch": 2.56,
|
| 126 |
+
"learning_rate": 7.220447284345049e-05,
|
| 127 |
+
"loss": 0.1271,
|
| 128 |
+
"step": 1600
|
| 129 |
},
|
| 130 |
{
|
| 131 |
+
"epoch": 2.72,
|
| 132 |
+
"learning_rate": 6.421725239616614e-05,
|
| 133 |
+
"loss": 0.1746,
|
| 134 |
+
"step": 1700
|
| 135 |
},
|
| 136 |
{
|
| 137 |
+
"epoch": 2.88,
|
| 138 |
+
"learning_rate": 5.623003194888179e-05,
|
| 139 |
+
"loss": 0.1118,
|
| 140 |
+
"step": 1800
|
| 141 |
},
|
| 142 |
{
|
| 143 |
+
"epoch": 3.0,
|
| 144 |
+
"eval_accuracy": 0.8704663212435233,
|
| 145 |
+
"eval_f1": 0.8663258953141536,
|
| 146 |
+
"eval_loss": 0.30049628019332886,
|
| 147 |
+
"eval_precision": 0.8698602729520133,
|
| 148 |
+
"eval_recall": 0.8704663212435233,
|
| 149 |
+
"eval_runtime": 1.4368,
|
| 150 |
+
"eval_samples_per_second": 134.33,
|
| 151 |
+
"eval_steps_per_second": 17.4,
|
| 152 |
+
"step": 1878
|
| 153 |
},
|
| 154 |
{
|
| 155 |
+
"epoch": 3.04,
|
| 156 |
+
"learning_rate": 4.824281150159744e-05,
|
| 157 |
+
"loss": 0.1193,
|
| 158 |
+
"step": 1900
|
| 159 |
},
|
| 160 |
{
|
| 161 |
+
"epoch": 3.19,
|
| 162 |
+
"learning_rate": 4.0255591054313104e-05,
|
| 163 |
+
"loss": 0.0409,
|
| 164 |
+
"step": 2000
|
| 165 |
},
|
| 166 |
{
|
| 167 |
+
"epoch": 3.35,
|
| 168 |
+
"learning_rate": 3.226837060702875e-05,
|
| 169 |
+
"loss": 0.0327,
|
| 170 |
+
"step": 2100
|
| 171 |
},
|
| 172 |
{
|
| 173 |
+
"epoch": 3.51,
|
| 174 |
+
"learning_rate": 2.428115015974441e-05,
|
| 175 |
+
"loss": 0.0357,
|
| 176 |
+
"step": 2200
|
| 177 |
},
|
| 178 |
{
|
| 179 |
+
"epoch": 3.67,
|
| 180 |
+
"learning_rate": 1.6293929712460065e-05,
|
| 181 |
+
"loss": 0.0386,
|
| 182 |
+
"step": 2300
|
| 183 |
},
|
| 184 |
{
|
| 185 |
+
"epoch": 3.83,
|
| 186 |
+
"learning_rate": 8.306709265175718e-06,
|
| 187 |
+
"loss": 0.0303,
|
| 188 |
+
"step": 2400
|
| 189 |
},
|
| 190 |
{
|
| 191 |
+
"epoch": 3.99,
|
| 192 |
+
"learning_rate": 3.194888178913738e-07,
|
| 193 |
+
"loss": 0.0317,
|
| 194 |
+
"step": 2500
|
| 195 |
},
|
| 196 |
{
|
| 197 |
+
"epoch": 4.0,
|
| 198 |
+
"eval_accuracy": 0.8549222797927462,
|
| 199 |
+
"eval_f1": 0.8560478324319817,
|
| 200 |
+
"eval_loss": 0.3552953004837036,
|
| 201 |
+
"eval_precision": 0.8595484172497833,
|
| 202 |
+
"eval_recall": 0.8549222797927462,
|
| 203 |
+
"eval_runtime": 1.5113,
|
| 204 |
+
"eval_samples_per_second": 127.703,
|
| 205 |
+
"eval_steps_per_second": 16.542,
|
| 206 |
+
"step": 2504
|
| 207 |
},
|
| 208 |
{
|
| 209 |
+
"epoch": 4.0,
|
| 210 |
+
"step": 2504,
|
| 211 |
+
"total_flos": 3.104468219559813e+18,
|
| 212 |
+
"train_loss": 0.2716429328141264,
|
| 213 |
+
"train_runtime": 832.7663,
|
| 214 |
+
"train_samples_per_second": 48.105,
|
| 215 |
+
"train_steps_per_second": 3.007
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 216 |
}
|
| 217 |
],
|
| 218 |
+
"max_steps": 2504,
|
| 219 |
+
"num_train_epochs": 4,
|
| 220 |
+
"total_flos": 3.104468219559813e+18,
|
| 221 |
"trial_name": null,
|
| 222 |
"trial_params": null
|
| 223 |
}
|
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:dae95784d5a6901f55f2c792e11ade68b7f5871a7f402bd5113f06400a3ea472
|
| 3 |
size 3899
|