Commit
·
b18cc90
1
Parent(s):
63f545b
Update to latest model
Browse files- README.md +6 -6
- checkpoint-10763/trainer_state.json +0 -3065
- {checkpoint-10763 → checkpoint-11092}/config.json +0 -0
- {checkpoint-10763 → checkpoint-11092}/model.safetensors +1 -1
- {checkpoint-10763 → checkpoint-11092}/optimizer.pt +1 -1
- {checkpoint-10763 → checkpoint-11092}/rng_state.pth +1 -1
- {checkpoint-10763 → checkpoint-11092}/scheduler.pt +1 -1
- checkpoint-11092/trainer_state.json +3156 -0
- {checkpoint-10763 → checkpoint-11092}/training_args.bin +1 -1
- model.safetensors +1 -1
- runs/{Dec04_07-28-34_f529bc929283/events.out.tfevents.1733297327.f529bc929283.82.0 → Dec07_05-45-42_74694ba9b4ba/events.out.tfevents.1733550355.74694ba9b4ba.167.0} +2 -2
- runs/{Dec04_07-28-34_f529bc929283/events.out.tfevents.1733332341.f529bc929283.82.1 → Dec07_05-45-42_74694ba9b4ba/events.out.tfevents.1733585593.74694ba9b4ba.167.1} +1 -1
- training_args.bin +1 -1
- training_params.json +3 -3
README.md
CHANGED
|
@@ -18,14 +18,14 @@ widget:
|
|
| 18 |
- Problem type: Image Classification
|
| 19 |
|
| 20 |
## Validation Metrics
|
| 21 |
-
loss: 0.
|
| 22 |
|
| 23 |
-
f1: 0.
|
| 24 |
|
| 25 |
-
precision: 0.
|
| 26 |
|
| 27 |
-
recall: 0.
|
| 28 |
|
| 29 |
-
auc: 0.
|
| 30 |
|
| 31 |
-
accuracy: 0.
|
|
|
|
| 18 |
- Problem type: Image Classification
|
| 19 |
|
| 20 |
## Validation Metrics
|
| 21 |
+
loss: 0.16320379078388214
|
| 22 |
|
| 23 |
+
f1: 0.9712272460364063
|
| 24 |
|
| 25 |
+
precision: 0.9649941656942824
|
| 26 |
|
| 27 |
+
recall: 0.9775413711583925
|
| 28 |
|
| 29 |
+
auc: 0.9916877882693556
|
| 30 |
|
| 31 |
+
accuracy: 0.9683073539874523
|
checkpoint-10763/trainer_state.json
DELETED
|
@@ -1,3065 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_metric": 0.17713692784309387,
|
| 3 |
-
"best_model_checkpoint": "autotrain-ai-image-detect-20241204-0210/checkpoint-10763",
|
| 4 |
-
"epoch": 1.0,
|
| 5 |
-
"eval_steps": 500,
|
| 6 |
-
"global_step": 10763,
|
| 7 |
-
"is_hyper_param_search": false,
|
| 8 |
-
"is_local_process_zero": true,
|
| 9 |
-
"is_world_process_zero": true,
|
| 10 |
-
"log_history": [
|
| 11 |
-
{
|
| 12 |
-
"epoch": 0.0023227724612097,
|
| 13 |
-
"grad_norm": 0.20149391889572144,
|
| 14 |
-
"learning_rate": 3.9999467510369354e-05,
|
| 15 |
-
"loss": 0.1422,
|
| 16 |
-
"step": 25
|
| 17 |
-
},
|
| 18 |
-
{
|
| 19 |
-
"epoch": 0.0046455449224194,
|
| 20 |
-
"grad_norm": 0.05961635708808899,
|
| 21 |
-
"learning_rate": 3.999787006983194e-05,
|
| 22 |
-
"loss": 0.1565,
|
| 23 |
-
"step": 50
|
| 24 |
-
},
|
| 25 |
-
{
|
| 26 |
-
"epoch": 0.0069683173836291,
|
| 27 |
-
"grad_norm": 0.10839387774467468,
|
| 28 |
-
"learning_rate": 3.99952077634498e-05,
|
| 29 |
-
"loss": 0.0725,
|
| 30 |
-
"step": 75
|
| 31 |
-
},
|
| 32 |
-
{
|
| 33 |
-
"epoch": 0.0092910898448388,
|
| 34 |
-
"grad_norm": 0.4582352340221405,
|
| 35 |
-
"learning_rate": 3.9991480732987994e-05,
|
| 36 |
-
"loss": 0.1475,
|
| 37 |
-
"step": 100
|
| 38 |
-
},
|
| 39 |
-
{
|
| 40 |
-
"epoch": 0.0116138623060485,
|
| 41 |
-
"grad_norm": 0.25347715616226196,
|
| 42 |
-
"learning_rate": 3.998668917690703e-05,
|
| 43 |
-
"loss": 0.2518,
|
| 44 |
-
"step": 125
|
| 45 |
-
},
|
| 46 |
-
{
|
| 47 |
-
"epoch": 0.0139366347672582,
|
| 48 |
-
"grad_norm": 4.189545154571533,
|
| 49 |
-
"learning_rate": 3.99808333503523e-05,
|
| 50 |
-
"loss": 0.1282,
|
| 51 |
-
"step": 150
|
| 52 |
-
},
|
| 53 |
-
{
|
| 54 |
-
"epoch": 0.0162594072284679,
|
| 55 |
-
"grad_norm": 0.2086610645055771,
|
| 56 |
-
"learning_rate": 3.99739135651405e-05,
|
| 57 |
-
"loss": 0.1091,
|
| 58 |
-
"step": 175
|
| 59 |
-
},
|
| 60 |
-
{
|
| 61 |
-
"epoch": 0.0185821796896776,
|
| 62 |
-
"grad_norm": 2.7873218059539795,
|
| 63 |
-
"learning_rate": 3.9965930189743e-05,
|
| 64 |
-
"loss": 0.1279,
|
| 65 |
-
"step": 200
|
| 66 |
-
},
|
| 67 |
-
{
|
| 68 |
-
"epoch": 0.0209049521508873,
|
| 69 |
-
"grad_norm": 0.28166231513023376,
|
| 70 |
-
"learning_rate": 3.995688364926629e-05,
|
| 71 |
-
"loss": 0.1408,
|
| 72 |
-
"step": 225
|
| 73 |
-
},
|
| 74 |
-
{
|
| 75 |
-
"epoch": 0.023227724612097,
|
| 76 |
-
"grad_norm": 0.15007774531841278,
|
| 77 |
-
"learning_rate": 3.9946774425429254e-05,
|
| 78 |
-
"loss": 0.1222,
|
| 79 |
-
"step": 250
|
| 80 |
-
},
|
| 81 |
-
{
|
| 82 |
-
"epoch": 0.0255504970733067,
|
| 83 |
-
"grad_norm": 0.21874751150608063,
|
| 84 |
-
"learning_rate": 3.993560305653757e-05,
|
| 85 |
-
"loss": 0.1683,
|
| 86 |
-
"step": 275
|
| 87 |
-
},
|
| 88 |
-
{
|
| 89 |
-
"epoch": 0.0278732695345164,
|
| 90 |
-
"grad_norm": 0.1976601779460907,
|
| 91 |
-
"learning_rate": 3.9923370137455066e-05,
|
| 92 |
-
"loss": 0.0464,
|
| 93 |
-
"step": 300
|
| 94 |
-
},
|
| 95 |
-
{
|
| 96 |
-
"epoch": 0.0301960419957261,
|
| 97 |
-
"grad_norm": 0.07817498594522476,
|
| 98 |
-
"learning_rate": 3.991007631957199e-05,
|
| 99 |
-
"loss": 0.0606,
|
| 100 |
-
"step": 325
|
| 101 |
-
},
|
| 102 |
-
{
|
| 103 |
-
"epoch": 0.0325188144569358,
|
| 104 |
-
"grad_norm": 0.7152191400527954,
|
| 105 |
-
"learning_rate": 3.989572231077035e-05,
|
| 106 |
-
"loss": 0.0953,
|
| 107 |
-
"step": 350
|
| 108 |
-
},
|
| 109 |
-
{
|
| 110 |
-
"epoch": 0.0348415869181455,
|
| 111 |
-
"grad_norm": 0.10070320218801498,
|
| 112 |
-
"learning_rate": 3.9880308875386244e-05,
|
| 113 |
-
"loss": 0.2266,
|
| 114 |
-
"step": 375
|
| 115 |
-
},
|
| 116 |
-
{
|
| 117 |
-
"epoch": 0.0371643593793552,
|
| 118 |
-
"grad_norm": 9.900060653686523,
|
| 119 |
-
"learning_rate": 3.986383683416912e-05,
|
| 120 |
-
"loss": 0.194,
|
| 121 |
-
"step": 400
|
| 122 |
-
},
|
| 123 |
-
{
|
| 124 |
-
"epoch": 0.0394871318405649,
|
| 125 |
-
"grad_norm": 0.0980474203824997,
|
| 126 |
-
"learning_rate": 3.984630706423809e-05,
|
| 127 |
-
"loss": 0.2128,
|
| 128 |
-
"step": 425
|
| 129 |
-
},
|
| 130 |
-
{
|
| 131 |
-
"epoch": 0.0418099043017746,
|
| 132 |
-
"grad_norm": 0.23585061728954315,
|
| 133 |
-
"learning_rate": 3.982772049903522e-05,
|
| 134 |
-
"loss": 0.1366,
|
| 135 |
-
"step": 450
|
| 136 |
-
},
|
| 137 |
-
{
|
| 138 |
-
"epoch": 0.044132676762984296,
|
| 139 |
-
"grad_norm": 67.88980102539062,
|
| 140 |
-
"learning_rate": 3.9808078128275844e-05,
|
| 141 |
-
"loss": 0.1038,
|
| 142 |
-
"step": 475
|
| 143 |
-
},
|
| 144 |
-
{
|
| 145 |
-
"epoch": 0.046455449224194,
|
| 146 |
-
"grad_norm": 0.933418333530426,
|
| 147 |
-
"learning_rate": 3.978738099789584e-05,
|
| 148 |
-
"loss": 0.1748,
|
| 149 |
-
"step": 500
|
| 150 |
-
},
|
| 151 |
-
{
|
| 152 |
-
"epoch": 0.048778221685403696,
|
| 153 |
-
"grad_norm": 14.157394409179688,
|
| 154 |
-
"learning_rate": 3.9765630209995934e-05,
|
| 155 |
-
"loss": 0.1259,
|
| 156 |
-
"step": 525
|
| 157 |
-
},
|
| 158 |
-
{
|
| 159 |
-
"epoch": 0.0511009941466134,
|
| 160 |
-
"grad_norm": 2.6177523136138916,
|
| 161 |
-
"learning_rate": 3.9742826922783015e-05,
|
| 162 |
-
"loss": 0.1423,
|
| 163 |
-
"step": 550
|
| 164 |
-
},
|
| 165 |
-
{
|
| 166 |
-
"epoch": 0.053423766607823095,
|
| 167 |
-
"grad_norm": 18.699522018432617,
|
| 168 |
-
"learning_rate": 3.97189723505085e-05,
|
| 169 |
-
"loss": 0.1405,
|
| 170 |
-
"step": 575
|
| 171 |
-
},
|
| 172 |
-
{
|
| 173 |
-
"epoch": 0.0557465390690328,
|
| 174 |
-
"grad_norm": 24.496646881103516,
|
| 175 |
-
"learning_rate": 3.969406776340362e-05,
|
| 176 |
-
"loss": 0.2092,
|
| 177 |
-
"step": 600
|
| 178 |
-
},
|
| 179 |
-
{
|
| 180 |
-
"epoch": 0.058069311530242494,
|
| 181 |
-
"grad_norm": 0.09642871469259262,
|
| 182 |
-
"learning_rate": 3.9668114487611815e-05,
|
| 183 |
-
"loss": 0.1268,
|
| 184 |
-
"step": 625
|
| 185 |
-
},
|
| 186 |
-
{
|
| 187 |
-
"epoch": 0.0603920839914522,
|
| 188 |
-
"grad_norm": 162.56884765625,
|
| 189 |
-
"learning_rate": 3.964111390511811e-05,
|
| 190 |
-
"loss": 0.1217,
|
| 191 |
-
"step": 650
|
| 192 |
-
},
|
| 193 |
-
{
|
| 194 |
-
"epoch": 0.0627148564526619,
|
| 195 |
-
"grad_norm": 1.033023715019226,
|
| 196 |
-
"learning_rate": 3.9613067453675514e-05,
|
| 197 |
-
"loss": 0.1637,
|
| 198 |
-
"step": 675
|
| 199 |
-
},
|
| 200 |
-
{
|
| 201 |
-
"epoch": 0.0650376289138716,
|
| 202 |
-
"grad_norm": 0.08476819843053818,
|
| 203 |
-
"learning_rate": 3.95839766267285e-05,
|
| 204 |
-
"loss": 0.1528,
|
| 205 |
-
"step": 700
|
| 206 |
-
},
|
| 207 |
-
{
|
| 208 |
-
"epoch": 0.0673604013750813,
|
| 209 |
-
"grad_norm": 1.8063493967056274,
|
| 210 |
-
"learning_rate": 3.955384297333343e-05,
|
| 211 |
-
"loss": 0.2262,
|
| 212 |
-
"step": 725
|
| 213 |
-
},
|
| 214 |
-
{
|
| 215 |
-
"epoch": 0.069683173836291,
|
| 216 |
-
"grad_norm": 14.230472564697266,
|
| 217 |
-
"learning_rate": 3.9522668098076094e-05,
|
| 218 |
-
"loss": 0.1468,
|
| 219 |
-
"step": 750
|
| 220 |
-
},
|
| 221 |
-
{
|
| 222 |
-
"epoch": 0.07200594629750069,
|
| 223 |
-
"grad_norm": 48.162841796875,
|
| 224 |
-
"learning_rate": 3.949045366098628e-05,
|
| 225 |
-
"loss": 0.115,
|
| 226 |
-
"step": 775
|
| 227 |
-
},
|
| 228 |
-
{
|
| 229 |
-
"epoch": 0.0743287187587104,
|
| 230 |
-
"grad_norm": 0.45269152522087097,
|
| 231 |
-
"learning_rate": 3.945720137744937e-05,
|
| 232 |
-
"loss": 0.1645,
|
| 233 |
-
"step": 800
|
| 234 |
-
},
|
| 235 |
-
{
|
| 236 |
-
"epoch": 0.0766514912199201,
|
| 237 |
-
"grad_norm": 0.2054939568042755,
|
| 238 |
-
"learning_rate": 3.942291301811496e-05,
|
| 239 |
-
"loss": 0.1242,
|
| 240 |
-
"step": 825
|
| 241 |
-
},
|
| 242 |
-
{
|
| 243 |
-
"epoch": 0.0789742636811298,
|
| 244 |
-
"grad_norm": 19.454818725585938,
|
| 245 |
-
"learning_rate": 3.938759040880265e-05,
|
| 246 |
-
"loss": 0.1045,
|
| 247 |
-
"step": 850
|
| 248 |
-
},
|
| 249 |
-
{
|
| 250 |
-
"epoch": 0.08129703614233949,
|
| 251 |
-
"grad_norm": 0.30843499302864075,
|
| 252 |
-
"learning_rate": 3.935123543040474e-05,
|
| 253 |
-
"loss": 0.21,
|
| 254 |
-
"step": 875
|
| 255 |
-
},
|
| 256 |
-
{
|
| 257 |
-
"epoch": 0.0836198086035492,
|
| 258 |
-
"grad_norm": 13.597454071044922,
|
| 259 |
-
"learning_rate": 3.931385001878615e-05,
|
| 260 |
-
"loss": 0.1963,
|
| 261 |
-
"step": 900
|
| 262 |
-
},
|
| 263 |
-
{
|
| 264 |
-
"epoch": 0.0859425810647589,
|
| 265 |
-
"grad_norm": 16.082290649414062,
|
| 266 |
-
"learning_rate": 3.927543616468127e-05,
|
| 267 |
-
"loss": 0.1375,
|
| 268 |
-
"step": 925
|
| 269 |
-
},
|
| 270 |
-
{
|
| 271 |
-
"epoch": 0.08826535352596859,
|
| 272 |
-
"grad_norm": 9.569624900817871,
|
| 273 |
-
"learning_rate": 3.923599591358799e-05,
|
| 274 |
-
"loss": 0.1325,
|
| 275 |
-
"step": 950
|
| 276 |
-
},
|
| 277 |
-
{
|
| 278 |
-
"epoch": 0.0905881259871783,
|
| 279 |
-
"grad_norm": 24.68838882446289,
|
| 280 |
-
"learning_rate": 3.919553136565881e-05,
|
| 281 |
-
"loss": 0.103,
|
| 282 |
-
"step": 975
|
| 283 |
-
},
|
| 284 |
-
{
|
| 285 |
-
"epoch": 0.092910898448388,
|
| 286 |
-
"grad_norm": 1.7922626733779907,
|
| 287 |
-
"learning_rate": 3.915404467558893e-05,
|
| 288 |
-
"loss": 0.1538,
|
| 289 |
-
"step": 1000
|
| 290 |
-
},
|
| 291 |
-
{
|
| 292 |
-
"epoch": 0.0952336709095977,
|
| 293 |
-
"grad_norm": 0.22084029018878937,
|
| 294 |
-
"learning_rate": 3.911153805250158e-05,
|
| 295 |
-
"loss": 0.157,
|
| 296 |
-
"step": 1025
|
| 297 |
-
},
|
| 298 |
-
{
|
| 299 |
-
"epoch": 0.09755644337080739,
|
| 300 |
-
"grad_norm": 19.544288635253906,
|
| 301 |
-
"learning_rate": 3.9068013759830365e-05,
|
| 302 |
-
"loss": 0.1838,
|
| 303 |
-
"step": 1050
|
| 304 |
-
},
|
| 305 |
-
{
|
| 306 |
-
"epoch": 0.0998792158320171,
|
| 307 |
-
"grad_norm": 16.32280158996582,
|
| 308 |
-
"learning_rate": 3.9023474115198736e-05,
|
| 309 |
-
"loss": 0.1329,
|
| 310 |
-
"step": 1075
|
| 311 |
-
},
|
| 312 |
-
{
|
| 313 |
-
"epoch": 0.1022019882932268,
|
| 314 |
-
"grad_norm": 14.392629623413086,
|
| 315 |
-
"learning_rate": 3.897792149029658e-05,
|
| 316 |
-
"loss": 0.1432,
|
| 317 |
-
"step": 1100
|
| 318 |
-
},
|
| 319 |
-
{
|
| 320 |
-
"epoch": 0.1045247607544365,
|
| 321 |
-
"grad_norm": 0.19446851313114166,
|
| 322 |
-
"learning_rate": 3.8931358310753955e-05,
|
| 323 |
-
"loss": 0.1693,
|
| 324 |
-
"step": 1125
|
| 325 |
-
},
|
| 326 |
-
{
|
| 327 |
-
"epoch": 0.10684753321564619,
|
| 328 |
-
"grad_norm": 17.14286994934082,
|
| 329 |
-
"learning_rate": 3.888378705601187e-05,
|
| 330 |
-
"loss": 0.2075,
|
| 331 |
-
"step": 1150
|
| 332 |
-
},
|
| 333 |
-
{
|
| 334 |
-
"epoch": 0.1091703056768559,
|
| 335 |
-
"grad_norm": 0.026348428800702095,
|
| 336 |
-
"learning_rate": 3.883521025919032e-05,
|
| 337 |
-
"loss": 0.0875,
|
| 338 |
-
"step": 1175
|
| 339 |
-
},
|
| 340 |
-
{
|
| 341 |
-
"epoch": 0.1114930781380656,
|
| 342 |
-
"grad_norm": 71.3205337524414,
|
| 343 |
-
"learning_rate": 3.878563050695336e-05,
|
| 344 |
-
"loss": 0.1481,
|
| 345 |
-
"step": 1200
|
| 346 |
-
},
|
| 347 |
-
{
|
| 348 |
-
"epoch": 0.11381585059927529,
|
| 349 |
-
"grad_norm": 0.09060646593570709,
|
| 350 |
-
"learning_rate": 3.87350504393714e-05,
|
| 351 |
-
"loss": 0.1001,
|
| 352 |
-
"step": 1225
|
| 353 |
-
},
|
| 354 |
-
{
|
| 355 |
-
"epoch": 0.11613862306048499,
|
| 356 |
-
"grad_norm": 20.108421325683594,
|
| 357 |
-
"learning_rate": 3.868347274978056e-05,
|
| 358 |
-
"loss": 0.0816,
|
| 359 |
-
"step": 1250
|
| 360 |
-
},
|
| 361 |
-
{
|
| 362 |
-
"epoch": 0.1184613955216947,
|
| 363 |
-
"grad_norm": 0.18188175559043884,
|
| 364 |
-
"learning_rate": 3.863090018463936e-05,
|
| 365 |
-
"loss": 0.1652,
|
| 366 |
-
"step": 1275
|
| 367 |
-
},
|
| 368 |
-
{
|
| 369 |
-
"epoch": 0.1207841679829044,
|
| 370 |
-
"grad_norm": 18.32872200012207,
|
| 371 |
-
"learning_rate": 3.8577335543382366e-05,
|
| 372 |
-
"loss": 0.232,
|
| 373 |
-
"step": 1300
|
| 374 |
-
},
|
| 375 |
-
{
|
| 376 |
-
"epoch": 0.12310694044411409,
|
| 377 |
-
"grad_norm": 0.790989339351654,
|
| 378 |
-
"learning_rate": 3.8522781678271185e-05,
|
| 379 |
-
"loss": 0.1301,
|
| 380 |
-
"step": 1325
|
| 381 |
-
},
|
| 382 |
-
{
|
| 383 |
-
"epoch": 0.1254297129053238,
|
| 384 |
-
"grad_norm": 96.62675476074219,
|
| 385 |
-
"learning_rate": 3.8467241494242555e-05,
|
| 386 |
-
"loss": 0.1407,
|
| 387 |
-
"step": 1350
|
| 388 |
-
},
|
| 389 |
-
{
|
| 390 |
-
"epoch": 0.1277524853665335,
|
| 391 |
-
"grad_norm": 0.3573750853538513,
|
| 392 |
-
"learning_rate": 3.8410717948753704e-05,
|
| 393 |
-
"loss": 0.1055,
|
| 394 |
-
"step": 1375
|
| 395 |
-
},
|
| 396 |
-
{
|
| 397 |
-
"epoch": 0.1300752578277432,
|
| 398 |
-
"grad_norm": 0.42159709334373474,
|
| 399 |
-
"learning_rate": 3.83532140516248e-05,
|
| 400 |
-
"loss": 0.1364,
|
| 401 |
-
"step": 1400
|
| 402 |
-
},
|
| 403 |
-
{
|
| 404 |
-
"epoch": 0.1323980302889529,
|
| 405 |
-
"grad_norm": 0.9740263223648071,
|
| 406 |
-
"learning_rate": 3.829473286487874e-05,
|
| 407 |
-
"loss": 0.0819,
|
| 408 |
-
"step": 1425
|
| 409 |
-
},
|
| 410 |
-
{
|
| 411 |
-
"epoch": 0.1347208027501626,
|
| 412 |
-
"grad_norm": 0.6651044487953186,
|
| 413 |
-
"learning_rate": 3.823527750257809e-05,
|
| 414 |
-
"loss": 0.1713,
|
| 415 |
-
"step": 1450
|
| 416 |
-
},
|
| 417 |
-
{
|
| 418 |
-
"epoch": 0.13704357521137228,
|
| 419 |
-
"grad_norm": 1.3060804605484009,
|
| 420 |
-
"learning_rate": 3.817485113065922e-05,
|
| 421 |
-
"loss": 0.1494,
|
| 422 |
-
"step": 1475
|
| 423 |
-
},
|
| 424 |
-
{
|
| 425 |
-
"epoch": 0.139366347672582,
|
| 426 |
-
"grad_norm": 9.927212715148926,
|
| 427 |
-
"learning_rate": 3.811345696676379e-05,
|
| 428 |
-
"loss": 0.2857,
|
| 429 |
-
"step": 1500
|
| 430 |
-
},
|
| 431 |
-
{
|
| 432 |
-
"epoch": 0.1416891201337917,
|
| 433 |
-
"grad_norm": 17.118671417236328,
|
| 434 |
-
"learning_rate": 3.805109828006737e-05,
|
| 435 |
-
"loss": 0.1425,
|
| 436 |
-
"step": 1525
|
| 437 |
-
},
|
| 438 |
-
{
|
| 439 |
-
"epoch": 0.14401189259500138,
|
| 440 |
-
"grad_norm": 18.74208641052246,
|
| 441 |
-
"learning_rate": 3.7987778391105356e-05,
|
| 442 |
-
"loss": 0.1199,
|
| 443 |
-
"step": 1550
|
| 444 |
-
},
|
| 445 |
-
{
|
| 446 |
-
"epoch": 0.1463346650562111,
|
| 447 |
-
"grad_norm": 58.87640380859375,
|
| 448 |
-
"learning_rate": 3.792350067159618e-05,
|
| 449 |
-
"loss": 0.2119,
|
| 450 |
-
"step": 1575
|
| 451 |
-
},
|
| 452 |
-
{
|
| 453 |
-
"epoch": 0.1486574375174208,
|
| 454 |
-
"grad_norm": 0.20524978637695312,
|
| 455 |
-
"learning_rate": 3.785826854426175e-05,
|
| 456 |
-
"loss": 0.1775,
|
| 457 |
-
"step": 1600
|
| 458 |
-
},
|
| 459 |
-
{
|
| 460 |
-
"epoch": 0.1509802099786305,
|
| 461 |
-
"grad_norm": 0.3944374620914459,
|
| 462 |
-
"learning_rate": 3.779208548264521e-05,
|
| 463 |
-
"loss": 0.1784,
|
| 464 |
-
"step": 1625
|
| 465 |
-
},
|
| 466 |
-
{
|
| 467 |
-
"epoch": 0.1533029824398402,
|
| 468 |
-
"grad_norm": 19.79619598388672,
|
| 469 |
-
"learning_rate": 3.772495501092595e-05,
|
| 470 |
-
"loss": 0.1513,
|
| 471 |
-
"step": 1650
|
| 472 |
-
},
|
| 473 |
-
{
|
| 474 |
-
"epoch": 0.1556257549010499,
|
| 475 |
-
"grad_norm": 13.110577583312988,
|
| 476 |
-
"learning_rate": 3.765688070373201e-05,
|
| 477 |
-
"loss": 0.1402,
|
| 478 |
-
"step": 1675
|
| 479 |
-
},
|
| 480 |
-
{
|
| 481 |
-
"epoch": 0.1579485273622596,
|
| 482 |
-
"grad_norm": 10.218574523925781,
|
| 483 |
-
"learning_rate": 3.758786618594963e-05,
|
| 484 |
-
"loss": 0.171,
|
| 485 |
-
"step": 1700
|
| 486 |
-
},
|
| 487 |
-
{
|
| 488 |
-
"epoch": 0.1602712998234693,
|
| 489 |
-
"grad_norm": 3.4088237285614014,
|
| 490 |
-
"learning_rate": 3.751791513253033e-05,
|
| 491 |
-
"loss": 0.0969,
|
| 492 |
-
"step": 1725
|
| 493 |
-
},
|
| 494 |
-
{
|
| 495 |
-
"epoch": 0.16259407228467898,
|
| 496 |
-
"grad_norm": 0.2767065167427063,
|
| 497 |
-
"learning_rate": 3.7447031268295166e-05,
|
| 498 |
-
"loss": 0.1048,
|
| 499 |
-
"step": 1750
|
| 500 |
-
},
|
| 501 |
-
{
|
| 502 |
-
"epoch": 0.1649168447458887,
|
| 503 |
-
"grad_norm": 16.004505157470703,
|
| 504 |
-
"learning_rate": 3.737521836773641e-05,
|
| 505 |
-
"loss": 0.2263,
|
| 506 |
-
"step": 1775
|
| 507 |
-
},
|
| 508 |
-
{
|
| 509 |
-
"epoch": 0.1672396172070984,
|
| 510 |
-
"grad_norm": 0.22232015430927277,
|
| 511 |
-
"learning_rate": 3.730248025481655e-05,
|
| 512 |
-
"loss": 0.204,
|
| 513 |
-
"step": 1800
|
| 514 |
-
},
|
| 515 |
-
{
|
| 516 |
-
"epoch": 0.16956238966830808,
|
| 517 |
-
"grad_norm": 0.9239380359649658,
|
| 518 |
-
"learning_rate": 3.722882080276468e-05,
|
| 519 |
-
"loss": 0.1623,
|
| 520 |
-
"step": 1825
|
| 521 |
-
},
|
| 522 |
-
{
|
| 523 |
-
"epoch": 0.1718851621295178,
|
| 524 |
-
"grad_norm": 8.507405281066895,
|
| 525 |
-
"learning_rate": 3.7154243933870225e-05,
|
| 526 |
-
"loss": 0.1364,
|
| 527 |
-
"step": 1850
|
| 528 |
-
},
|
| 529 |
-
{
|
| 530 |
-
"epoch": 0.1742079345907275,
|
| 531 |
-
"grad_norm": 0.10597966611385345,
|
| 532 |
-
"learning_rate": 3.707875361927414e-05,
|
| 533 |
-
"loss": 0.1983,
|
| 534 |
-
"step": 1875
|
| 535 |
-
},
|
| 536 |
-
{
|
| 537 |
-
"epoch": 0.17653070705193719,
|
| 538 |
-
"grad_norm": 13.092436790466309,
|
| 539 |
-
"learning_rate": 3.700235387875739e-05,
|
| 540 |
-
"loss": 0.1143,
|
| 541 |
-
"step": 1900
|
| 542 |
-
},
|
| 543 |
-
{
|
| 544 |
-
"epoch": 0.1788534795131469,
|
| 545 |
-
"grad_norm": 15.583847045898438,
|
| 546 |
-
"learning_rate": 3.6925048780526943e-05,
|
| 547 |
-
"loss": 0.1108,
|
| 548 |
-
"step": 1925
|
| 549 |
-
},
|
| 550 |
-
{
|
| 551 |
-
"epoch": 0.1811762519743566,
|
| 552 |
-
"grad_norm": 0.2391364425420761,
|
| 553 |
-
"learning_rate": 3.6846842440999114e-05,
|
| 554 |
-
"loss": 0.0628,
|
| 555 |
-
"step": 1950
|
| 556 |
-
},
|
| 557 |
-
{
|
| 558 |
-
"epoch": 0.1834990244355663,
|
| 559 |
-
"grad_norm": 36.44248962402344,
|
| 560 |
-
"learning_rate": 3.6767739024580385e-05,
|
| 561 |
-
"loss": 0.1776,
|
| 562 |
-
"step": 1975
|
| 563 |
-
},
|
| 564 |
-
{
|
| 565 |
-
"epoch": 0.185821796896776,
|
| 566 |
-
"grad_norm": 247.0557098388672,
|
| 567 |
-
"learning_rate": 3.668774274344565e-05,
|
| 568 |
-
"loss": 0.1573,
|
| 569 |
-
"step": 2000
|
| 570 |
-
},
|
| 571 |
-
{
|
| 572 |
-
"epoch": 0.18814456935798568,
|
| 573 |
-
"grad_norm": 0.03785315155982971,
|
| 574 |
-
"learning_rate": 3.660685785731395e-05,
|
| 575 |
-
"loss": 0.213,
|
| 576 |
-
"step": 2025
|
| 577 |
-
},
|
| 578 |
-
{
|
| 579 |
-
"epoch": 0.1904673418191954,
|
| 580 |
-
"grad_norm": 1.2548052072525024,
|
| 581 |
-
"learning_rate": 3.652508867322158e-05,
|
| 582 |
-
"loss": 0.0668,
|
| 583 |
-
"step": 2050
|
| 584 |
-
},
|
| 585 |
-
{
|
| 586 |
-
"epoch": 0.1927901142804051,
|
| 587 |
-
"grad_norm": 8.225103378295898,
|
| 588 |
-
"learning_rate": 3.644243954529281e-05,
|
| 589 |
-
"loss": 0.1935,
|
| 590 |
-
"step": 2075
|
| 591 |
-
},
|
| 592 |
-
{
|
| 593 |
-
"epoch": 0.19511288674161478,
|
| 594 |
-
"grad_norm": 9.357488632202148,
|
| 595 |
-
"learning_rate": 3.6358914874508e-05,
|
| 596 |
-
"loss": 0.1361,
|
| 597 |
-
"step": 2100
|
| 598 |
-
},
|
| 599 |
-
{
|
| 600 |
-
"epoch": 0.1974356592028245,
|
| 601 |
-
"grad_norm": 0.07872194051742554,
|
| 602 |
-
"learning_rate": 3.627451910846926e-05,
|
| 603 |
-
"loss": 0.1057,
|
| 604 |
-
"step": 2125
|
| 605 |
-
},
|
| 606 |
-
{
|
| 607 |
-
"epoch": 0.1997584316640342,
|
| 608 |
-
"grad_norm": 0.03234529867768288,
|
| 609 |
-
"learning_rate": 3.618925674116362e-05,
|
| 610 |
-
"loss": 0.1579,
|
| 611 |
-
"step": 2150
|
| 612 |
-
},
|
| 613 |
-
{
|
| 614 |
-
"epoch": 0.20208120412524389,
|
| 615 |
-
"grad_norm": 12.699563980102539,
|
| 616 |
-
"learning_rate": 3.610313231272373e-05,
|
| 617 |
-
"loss": 0.2416,
|
| 618 |
-
"step": 2175
|
| 619 |
-
},
|
| 620 |
-
{
|
| 621 |
-
"epoch": 0.2044039765864536,
|
| 622 |
-
"grad_norm": 17.83196449279785,
|
| 623 |
-
"learning_rate": 3.6016150409186085e-05,
|
| 624 |
-
"loss": 0.1646,
|
| 625 |
-
"step": 2200
|
| 626 |
-
},
|
| 627 |
-
{
|
| 628 |
-
"epoch": 0.20672674904766328,
|
| 629 |
-
"grad_norm": 43.95789337158203,
|
| 630 |
-
"learning_rate": 3.592831566224686e-05,
|
| 631 |
-
"loss": 0.0791,
|
| 632 |
-
"step": 2225
|
| 633 |
-
},
|
| 634 |
-
{
|
| 635 |
-
"epoch": 0.209049521508873,
|
| 636 |
-
"grad_norm": 0.7434149980545044,
|
| 637 |
-
"learning_rate": 3.5839632749015266e-05,
|
| 638 |
-
"loss": 0.0972,
|
| 639 |
-
"step": 2250
|
| 640 |
-
},
|
| 641 |
-
{
|
| 642 |
-
"epoch": 0.2113722939700827,
|
| 643 |
-
"grad_norm": 0.3992946147918701,
|
| 644 |
-
"learning_rate": 3.575010639176445e-05,
|
| 645 |
-
"loss": 0.1382,
|
| 646 |
-
"step": 2275
|
| 647 |
-
},
|
| 648 |
-
{
|
| 649 |
-
"epoch": 0.21369506643129238,
|
| 650 |
-
"grad_norm": 35.988037109375,
|
| 651 |
-
"learning_rate": 3.565974135768013e-05,
|
| 652 |
-
"loss": 0.2227,
|
| 653 |
-
"step": 2300
|
| 654 |
-
},
|
| 655 |
-
{
|
| 656 |
-
"epoch": 0.2160178388925021,
|
| 657 |
-
"grad_norm": 0.24125997722148895,
|
| 658 |
-
"learning_rate": 3.556854245860664e-05,
|
| 659 |
-
"loss": 0.1475,
|
| 660 |
-
"step": 2325
|
| 661 |
-
},
|
| 662 |
-
{
|
| 663 |
-
"epoch": 0.2183406113537118,
|
| 664 |
-
"grad_norm": 0.13932979106903076,
|
| 665 |
-
"learning_rate": 3.547651455079081e-05,
|
| 666 |
-
"loss": 0.0912,
|
| 667 |
-
"step": 2350
|
| 668 |
-
},
|
| 669 |
-
{
|
| 670 |
-
"epoch": 0.22066338381492148,
|
| 671 |
-
"grad_norm": 8.023921012878418,
|
| 672 |
-
"learning_rate": 3.5383662534623284e-05,
|
| 673 |
-
"loss": 0.2458,
|
| 674 |
-
"step": 2375
|
| 675 |
-
},
|
| 676 |
-
{
|
| 677 |
-
"epoch": 0.2229861562761312,
|
| 678 |
-
"grad_norm": 88.30912780761719,
|
| 679 |
-
"learning_rate": 3.5289991354377665e-05,
|
| 680 |
-
"loss": 0.1557,
|
| 681 |
-
"step": 2400
|
| 682 |
-
},
|
| 683 |
-
{
|
| 684 |
-
"epoch": 0.2253089287373409,
|
| 685 |
-
"grad_norm": 0.1918799728155136,
|
| 686 |
-
"learning_rate": 3.519550599794715e-05,
|
| 687 |
-
"loss": 0.0811,
|
| 688 |
-
"step": 2425
|
| 689 |
-
},
|
| 690 |
-
{
|
| 691 |
-
"epoch": 0.22763170119855058,
|
| 692 |
-
"grad_norm": 23.40361976623535,
|
| 693 |
-
"learning_rate": 3.5100211496579005e-05,
|
| 694 |
-
"loss": 0.1506,
|
| 695 |
-
"step": 2450
|
| 696 |
-
},
|
| 697 |
-
{
|
| 698 |
-
"epoch": 0.2299544736597603,
|
| 699 |
-
"grad_norm": 1.437442421913147,
|
| 700 |
-
"learning_rate": 3.500411292460661e-05,
|
| 701 |
-
"loss": 0.2609,
|
| 702 |
-
"step": 2475
|
| 703 |
-
},
|
| 704 |
-
{
|
| 705 |
-
"epoch": 0.23227724612096998,
|
| 706 |
-
"grad_norm": 12.71546459197998,
|
| 707 |
-
"learning_rate": 3.490721539917928e-05,
|
| 708 |
-
"loss": 0.1528,
|
| 709 |
-
"step": 2500
|
| 710 |
-
},
|
| 711 |
-
{
|
| 712 |
-
"epoch": 0.2346000185821797,
|
| 713 |
-
"grad_norm": 0.8931025266647339,
|
| 714 |
-
"learning_rate": 3.480952407998975e-05,
|
| 715 |
-
"loss": 0.0943,
|
| 716 |
-
"step": 2525
|
| 717 |
-
},
|
| 718 |
-
{
|
| 719 |
-
"epoch": 0.2369227910433894,
|
| 720 |
-
"grad_norm": 0.08648817241191864,
|
| 721 |
-
"learning_rate": 3.471104416899949e-05,
|
| 722 |
-
"loss": 0.1637,
|
| 723 |
-
"step": 2550
|
| 724 |
-
},
|
| 725 |
-
{
|
| 726 |
-
"epoch": 0.23924556350459908,
|
| 727 |
-
"grad_norm": 18.01279067993164,
|
| 728 |
-
"learning_rate": 3.4611780910161627e-05,
|
| 729 |
-
"loss": 0.1993,
|
| 730 |
-
"step": 2575
|
| 731 |
-
},
|
| 732 |
-
{
|
| 733 |
-
"epoch": 0.2415683359658088,
|
| 734 |
-
"grad_norm": 0.10815341770648956,
|
| 735 |
-
"learning_rate": 3.4511739589141776e-05,
|
| 736 |
-
"loss": 0.14,
|
| 737 |
-
"step": 2600
|
| 738 |
-
},
|
| 739 |
-
{
|
| 740 |
-
"epoch": 0.2438911084270185,
|
| 741 |
-
"grad_norm": 52.11756896972656,
|
| 742 |
-
"learning_rate": 3.4410925533036534e-05,
|
| 743 |
-
"loss": 0.1619,
|
| 744 |
-
"step": 2625
|
| 745 |
-
},
|
| 746 |
-
{
|
| 747 |
-
"epoch": 0.24621388088822818,
|
| 748 |
-
"grad_norm": 0.6608179211616516,
|
| 749 |
-
"learning_rate": 3.430934411008985e-05,
|
| 750 |
-
"loss": 0.0616,
|
| 751 |
-
"step": 2650
|
| 752 |
-
},
|
| 753 |
-
{
|
| 754 |
-
"epoch": 0.2485366533494379,
|
| 755 |
-
"grad_norm": 0.13757063448429108,
|
| 756 |
-
"learning_rate": 3.4207000729407176e-05,
|
| 757 |
-
"loss": 0.2123,
|
| 758 |
-
"step": 2675
|
| 759 |
-
},
|
| 760 |
-
{
|
| 761 |
-
"epoch": 0.2508594258106476,
|
| 762 |
-
"grad_norm": 0.048304133117198944,
|
| 763 |
-
"learning_rate": 3.4103900840667404e-05,
|
| 764 |
-
"loss": 0.1133,
|
| 765 |
-
"step": 2700
|
| 766 |
-
},
|
| 767 |
-
{
|
| 768 |
-
"epoch": 0.2531821982718573,
|
| 769 |
-
"grad_norm": 0.24736209213733673,
|
| 770 |
-
"learning_rate": 3.4000049933832696e-05,
|
| 771 |
-
"loss": 0.1665,
|
| 772 |
-
"step": 2725
|
| 773 |
-
},
|
| 774 |
-
{
|
| 775 |
-
"epoch": 0.255504970733067,
|
| 776 |
-
"grad_norm": 0.21393662691116333,
|
| 777 |
-
"learning_rate": 3.389545353885616e-05,
|
| 778 |
-
"loss": 0.2,
|
| 779 |
-
"step": 2750
|
| 780 |
-
},
|
| 781 |
-
{
|
| 782 |
-
"epoch": 0.2578277431942767,
|
| 783 |
-
"grad_norm": 0.2487454116344452,
|
| 784 |
-
"learning_rate": 3.379011722538736e-05,
|
| 785 |
-
"loss": 0.0749,
|
| 786 |
-
"step": 2775
|
| 787 |
-
},
|
| 788 |
-
{
|
| 789 |
-
"epoch": 0.2601505156554864,
|
| 790 |
-
"grad_norm": 54.294708251953125,
|
| 791 |
-
"learning_rate": 3.3684046602475776e-05,
|
| 792 |
-
"loss": 0.1565,
|
| 793 |
-
"step": 2800
|
| 794 |
-
},
|
| 795 |
-
{
|
| 796 |
-
"epoch": 0.26247328811669607,
|
| 797 |
-
"grad_norm": 0.33187443017959595,
|
| 798 |
-
"learning_rate": 3.3577247318272084e-05,
|
| 799 |
-
"loss": 0.1133,
|
| 800 |
-
"step": 2825
|
| 801 |
-
},
|
| 802 |
-
{
|
| 803 |
-
"epoch": 0.2647960605779058,
|
| 804 |
-
"grad_norm": 0.05274425074458122,
|
| 805 |
-
"learning_rate": 3.346972505972742e-05,
|
| 806 |
-
"loss": 0.2052,
|
| 807 |
-
"step": 2850
|
| 808 |
-
},
|
| 809 |
-
{
|
| 810 |
-
"epoch": 0.2671188330391155,
|
| 811 |
-
"grad_norm": 7.938210964202881,
|
| 812 |
-
"learning_rate": 3.3361485552290555e-05,
|
| 813 |
-
"loss": 0.1114,
|
| 814 |
-
"step": 2875
|
| 815 |
-
},
|
| 816 |
-
{
|
| 817 |
-
"epoch": 0.2694416055003252,
|
| 818 |
-
"grad_norm": 0.2889136075973511,
|
| 819 |
-
"learning_rate": 3.325253455960303e-05,
|
| 820 |
-
"loss": 0.1215,
|
| 821 |
-
"step": 2900
|
| 822 |
-
},
|
| 823 |
-
{
|
| 824 |
-
"epoch": 0.2717643779615349,
|
| 825 |
-
"grad_norm": 15.914332389831543,
|
| 826 |
-
"learning_rate": 3.314287788319223e-05,
|
| 827 |
-
"loss": 0.1785,
|
| 828 |
-
"step": 2925
|
| 829 |
-
},
|
| 830 |
-
{
|
| 831 |
-
"epoch": 0.27408715042274456,
|
| 832 |
-
"grad_norm": 0.1932603269815445,
|
| 833 |
-
"learning_rate": 3.303252136216246e-05,
|
| 834 |
-
"loss": 0.1068,
|
| 835 |
-
"step": 2950
|
| 836 |
-
},
|
| 837 |
-
{
|
| 838 |
-
"epoch": 0.2764099228839543,
|
| 839 |
-
"grad_norm": 15.732447624206543,
|
| 840 |
-
"learning_rate": 3.292147087288404e-05,
|
| 841 |
-
"loss": 0.1513,
|
| 842 |
-
"step": 2975
|
| 843 |
-
},
|
| 844 |
-
{
|
| 845 |
-
"epoch": 0.278732695345164,
|
| 846 |
-
"grad_norm": 0.1670047640800476,
|
| 847 |
-
"learning_rate": 3.280973232868037e-05,
|
| 848 |
-
"loss": 0.1499,
|
| 849 |
-
"step": 3000
|
| 850 |
-
},
|
| 851 |
-
{
|
| 852 |
-
"epoch": 0.2810554678063737,
|
| 853 |
-
"grad_norm": 0.16926802694797516,
|
| 854 |
-
"learning_rate": 3.269731167951306e-05,
|
| 855 |
-
"loss": 0.0826,
|
| 856 |
-
"step": 3025
|
| 857 |
-
},
|
| 858 |
-
{
|
| 859 |
-
"epoch": 0.2833782402675834,
|
| 860 |
-
"grad_norm": 17.729829788208008,
|
| 861 |
-
"learning_rate": 3.258421491166511e-05,
|
| 862 |
-
"loss": 0.1894,
|
| 863 |
-
"step": 3050
|
| 864 |
-
},
|
| 865 |
-
{
|
| 866 |
-
"epoch": 0.2857010127287931,
|
| 867 |
-
"grad_norm": 11.645220756530762,
|
| 868 |
-
"learning_rate": 3.247044804742212e-05,
|
| 869 |
-
"loss": 0.149,
|
| 870 |
-
"step": 3075
|
| 871 |
-
},
|
| 872 |
-
{
|
| 873 |
-
"epoch": 0.28802378519000277,
|
| 874 |
-
"grad_norm": 0.42598363757133484,
|
| 875 |
-
"learning_rate": 3.235601714475166e-05,
|
| 876 |
-
"loss": 0.201,
|
| 877 |
-
"step": 3100
|
| 878 |
-
},
|
| 879 |
-
{
|
| 880 |
-
"epoch": 0.2903465576512125,
|
| 881 |
-
"grad_norm": 0.1250600516796112,
|
| 882 |
-
"learning_rate": 3.224092829698064e-05,
|
| 883 |
-
"loss": 0.0886,
|
| 884 |
-
"step": 3125
|
| 885 |
-
},
|
| 886 |
-
{
|
| 887 |
-
"epoch": 0.2926693301124222,
|
| 888 |
-
"grad_norm": 2.5853846073150635,
|
| 889 |
-
"learning_rate": 3.212518763247085e-05,
|
| 890 |
-
"loss": 0.1521,
|
| 891 |
-
"step": 3150
|
| 892 |
-
},
|
| 893 |
-
{
|
| 894 |
-
"epoch": 0.2949921025736319,
|
| 895 |
-
"grad_norm": 0.3828018009662628,
|
| 896 |
-
"learning_rate": 3.2008801314292666e-05,
|
| 897 |
-
"loss": 0.0521,
|
| 898 |
-
"step": 3175
|
| 899 |
-
},
|
| 900 |
-
{
|
| 901 |
-
"epoch": 0.2973148750348416,
|
| 902 |
-
"grad_norm": 19.541994094848633,
|
| 903 |
-
"learning_rate": 3.1891775539896846e-05,
|
| 904 |
-
"loss": 0.1929,
|
| 905 |
-
"step": 3200
|
| 906 |
-
},
|
| 907 |
-
{
|
| 908 |
-
"epoch": 0.29963764749605126,
|
| 909 |
-
"grad_norm": 0.40449729561805725,
|
| 910 |
-
"learning_rate": 3.1774116540784534e-05,
|
| 911 |
-
"loss": 0.1464,
|
| 912 |
-
"step": 3225
|
| 913 |
-
},
|
| 914 |
-
{
|
| 915 |
-
"epoch": 0.301960419957261,
|
| 916 |
-
"grad_norm": 128.0067138671875,
|
| 917 |
-
"learning_rate": 3.165583058217542e-05,
|
| 918 |
-
"loss": 0.1499,
|
| 919 |
-
"step": 3250
|
| 920 |
-
},
|
| 921 |
-
{
|
| 922 |
-
"epoch": 0.3042831924184707,
|
| 923 |
-
"grad_norm": 0.9499270915985107,
|
| 924 |
-
"learning_rate": 3.153692396267415e-05,
|
| 925 |
-
"loss": 0.1087,
|
| 926 |
-
"step": 3275
|
| 927 |
-
},
|
| 928 |
-
{
|
| 929 |
-
"epoch": 0.3066059648796804,
|
| 930 |
-
"grad_norm": 28.435699462890625,
|
| 931 |
-
"learning_rate": 3.141740301393491e-05,
|
| 932 |
-
"loss": 0.2194,
|
| 933 |
-
"step": 3300
|
| 934 |
-
},
|
| 935 |
-
{
|
| 936 |
-
"epoch": 0.3089287373408901,
|
| 937 |
-
"grad_norm": 11.707649230957031,
|
| 938 |
-
"learning_rate": 3.129727410032429e-05,
|
| 939 |
-
"loss": 0.1698,
|
| 940 |
-
"step": 3325
|
| 941 |
-
},
|
| 942 |
-
{
|
| 943 |
-
"epoch": 0.3112515098020998,
|
| 944 |
-
"grad_norm": 0.7132061719894409,
|
| 945 |
-
"learning_rate": 3.117654361858236e-05,
|
| 946 |
-
"loss": 0.092,
|
| 947 |
-
"step": 3350
|
| 948 |
-
},
|
| 949 |
-
{
|
| 950 |
-
"epoch": 0.31357428226330947,
|
| 951 |
-
"grad_norm": 0.1964050829410553,
|
| 952 |
-
"learning_rate": 3.1055217997482106e-05,
|
| 953 |
-
"loss": 0.1247,
|
| 954 |
-
"step": 3375
|
| 955 |
-
},
|
| 956 |
-
{
|
| 957 |
-
"epoch": 0.3158970547245192,
|
| 958 |
-
"grad_norm": 0.0794818252325058,
|
| 959 |
-
"learning_rate": 3.093330369748702e-05,
|
| 960 |
-
"loss": 0.1326,
|
| 961 |
-
"step": 3400
|
| 962 |
-
},
|
| 963 |
-
{
|
| 964 |
-
"epoch": 0.3182198271857289,
|
| 965 |
-
"grad_norm": 0.5441995859146118,
|
| 966 |
-
"learning_rate": 3.081080721040718e-05,
|
| 967 |
-
"loss": 0.1546,
|
| 968 |
-
"step": 3425
|
| 969 |
-
},
|
| 970 |
-
{
|
| 971 |
-
"epoch": 0.3205425996469386,
|
| 972 |
-
"grad_norm": 0.7919977903366089,
|
| 973 |
-
"learning_rate": 3.06877350590535e-05,
|
| 974 |
-
"loss": 0.1182,
|
| 975 |
-
"step": 3450
|
| 976 |
-
},
|
| 977 |
-
{
|
| 978 |
-
"epoch": 0.3228653721081483,
|
| 979 |
-
"grad_norm": 26.345603942871094,
|
| 980 |
-
"learning_rate": 3.05640937968904e-05,
|
| 981 |
-
"loss": 0.118,
|
| 982 |
-
"step": 3475
|
| 983 |
-
},
|
| 984 |
-
{
|
| 985 |
-
"epoch": 0.32518814456935796,
|
| 986 |
-
"grad_norm": 30.931652069091797,
|
| 987 |
-
"learning_rate": 3.0439890007686918e-05,
|
| 988 |
-
"loss": 0.1658,
|
| 989 |
-
"step": 3500
|
| 990 |
-
},
|
| 991 |
-
{
|
| 992 |
-
"epoch": 0.32751091703056767,
|
| 993 |
-
"grad_norm": 105.8614730834961,
|
| 994 |
-
"learning_rate": 3.0315130305166004e-05,
|
| 995 |
-
"loss": 0.0959,
|
| 996 |
-
"step": 3525
|
| 997 |
-
},
|
| 998 |
-
{
|
| 999 |
-
"epoch": 0.3298336894917774,
|
| 1000 |
-
"grad_norm": 1.398895263671875,
|
| 1001 |
-
"learning_rate": 3.0189821332652465e-05,
|
| 1002 |
-
"loss": 0.1249,
|
| 1003 |
-
"step": 3550
|
| 1004 |
-
},
|
| 1005 |
-
{
|
| 1006 |
-
"epoch": 0.3321564619529871,
|
| 1007 |
-
"grad_norm": 0.14747412502765656,
|
| 1008 |
-
"learning_rate": 3.0063969762719146e-05,
|
| 1009 |
-
"loss": 0.2129,
|
| 1010 |
-
"step": 3575
|
| 1011 |
-
},
|
| 1012 |
-
{
|
| 1013 |
-
"epoch": 0.3344792344141968,
|
| 1014 |
-
"grad_norm": 0.22834627330303192,
|
| 1015 |
-
"learning_rate": 2.9937582296831655e-05,
|
| 1016 |
-
"loss": 0.094,
|
| 1017 |
-
"step": 3600
|
| 1018 |
-
},
|
| 1019 |
-
{
|
| 1020 |
-
"epoch": 0.3368020068754065,
|
| 1021 |
-
"grad_norm": 0.31678980588912964,
|
| 1022 |
-
"learning_rate": 2.981066566499148e-05,
|
| 1023 |
-
"loss": 0.0861,
|
| 1024 |
-
"step": 3625
|
| 1025 |
-
},
|
| 1026 |
-
{
|
| 1027 |
-
"epoch": 0.33912477933661617,
|
| 1028 |
-
"grad_norm": 0.04181404039263725,
|
| 1029 |
-
"learning_rate": 2.9683226625377678e-05,
|
| 1030 |
-
"loss": 0.1529,
|
| 1031 |
-
"step": 3650
|
| 1032 |
-
},
|
| 1033 |
-
{
|
| 1034 |
-
"epoch": 0.3414475517978259,
|
| 1035 |
-
"grad_norm": 11.233120918273926,
|
| 1036 |
-
"learning_rate": 2.955527196398695e-05,
|
| 1037 |
-
"loss": 0.0785,
|
| 1038 |
-
"step": 3675
|
| 1039 |
-
},
|
| 1040 |
-
{
|
| 1041 |
-
"epoch": 0.3437703242590356,
|
| 1042 |
-
"grad_norm": 6.489948272705078,
|
| 1043 |
-
"learning_rate": 2.942680849427234e-05,
|
| 1044 |
-
"loss": 0.2541,
|
| 1045 |
-
"step": 3700
|
| 1046 |
-
},
|
| 1047 |
-
{
|
| 1048 |
-
"epoch": 0.3460930967202453,
|
| 1049 |
-
"grad_norm": 10.235221862792969,
|
| 1050 |
-
"learning_rate": 2.9297843056780403e-05,
|
| 1051 |
-
"loss": 0.142,
|
| 1052 |
-
"step": 3725
|
| 1053 |
-
},
|
| 1054 |
-
{
|
| 1055 |
-
"epoch": 0.348415869181455,
|
| 1056 |
-
"grad_norm": 11.741005897521973,
|
| 1057 |
-
"learning_rate": 2.9168382518786946e-05,
|
| 1058 |
-
"loss": 0.1051,
|
| 1059 |
-
"step": 3750
|
| 1060 |
-
},
|
| 1061 |
-
{
|
| 1062 |
-
"epoch": 0.35073864164266466,
|
| 1063 |
-
"grad_norm": 38.80546188354492,
|
| 1064 |
-
"learning_rate": 2.9038433773931395e-05,
|
| 1065 |
-
"loss": 0.1485,
|
| 1066 |
-
"step": 3775
|
| 1067 |
-
},
|
| 1068 |
-
{
|
| 1069 |
-
"epoch": 0.35306141410387437,
|
| 1070 |
-
"grad_norm": 59.40329360961914,
|
| 1071 |
-
"learning_rate": 2.8908003741849646e-05,
|
| 1072 |
-
"loss": 0.0815,
|
| 1073 |
-
"step": 3800
|
| 1074 |
-
},
|
| 1075 |
-
{
|
| 1076 |
-
"epoch": 0.3553841865650841,
|
| 1077 |
-
"grad_norm": 21.45525360107422,
|
| 1078 |
-
"learning_rate": 2.8777099367805673e-05,
|
| 1079 |
-
"loss": 0.1425,
|
| 1080 |
-
"step": 3825
|
| 1081 |
-
},
|
| 1082 |
-
{
|
| 1083 |
-
"epoch": 0.3577069590262938,
|
| 1084 |
-
"grad_norm": 0.09223873913288116,
|
| 1085 |
-
"learning_rate": 2.8645727622321642e-05,
|
| 1086 |
-
"loss": 0.1474,
|
| 1087 |
-
"step": 3850
|
| 1088 |
-
},
|
| 1089 |
-
{
|
| 1090 |
-
"epoch": 0.3600297314875035,
|
| 1091 |
-
"grad_norm": 0.4064118266105652,
|
| 1092 |
-
"learning_rate": 2.8513895500806788e-05,
|
| 1093 |
-
"loss": 0.1377,
|
| 1094 |
-
"step": 3875
|
| 1095 |
-
},
|
| 1096 |
-
{
|
| 1097 |
-
"epoch": 0.3623525039487132,
|
| 1098 |
-
"grad_norm": 37.108123779296875,
|
| 1099 |
-
"learning_rate": 2.8381610023184874e-05,
|
| 1100 |
-
"loss": 0.1292,
|
| 1101 |
-
"step": 3900
|
| 1102 |
-
},
|
| 1103 |
-
{
|
| 1104 |
-
"epoch": 0.36467527640992287,
|
| 1105 |
-
"grad_norm": 0.03851945698261261,
|
| 1106 |
-
"learning_rate": 2.8248878233520417e-05,
|
| 1107 |
-
"loss": 0.1399,
|
| 1108 |
-
"step": 3925
|
| 1109 |
-
},
|
| 1110 |
-
{
|
| 1111 |
-
"epoch": 0.3669980488711326,
|
| 1112 |
-
"grad_norm": 0.34805092215538025,
|
| 1113 |
-
"learning_rate": 2.8115707199643573e-05,
|
| 1114 |
-
"loss": 0.1182,
|
| 1115 |
-
"step": 3950
|
| 1116 |
-
},
|
| 1117 |
-
{
|
| 1118 |
-
"epoch": 0.3693208213323423,
|
| 1119 |
-
"grad_norm": 18.120176315307617,
|
| 1120 |
-
"learning_rate": 2.7982104012773814e-05,
|
| 1121 |
-
"loss": 0.136,
|
| 1122 |
-
"step": 3975
|
| 1123 |
-
},
|
| 1124 |
-
{
|
| 1125 |
-
"epoch": 0.371643593793552,
|
| 1126 |
-
"grad_norm": 70.94572448730469,
|
| 1127 |
-
"learning_rate": 2.78480757871423e-05,
|
| 1128 |
-
"loss": 0.1545,
|
| 1129 |
-
"step": 4000
|
| 1130 |
-
},
|
| 1131 |
-
{
|
| 1132 |
-
"epoch": 0.3739663662547617,
|
| 1133 |
-
"grad_norm": 29.617115020751953,
|
| 1134 |
-
"learning_rate": 2.7713629659613065e-05,
|
| 1135 |
-
"loss": 0.152,
|
| 1136 |
-
"step": 4025
|
| 1137 |
-
},
|
| 1138 |
-
{
|
| 1139 |
-
"epoch": 0.37628913871597136,
|
| 1140 |
-
"grad_norm": 1.17340087890625,
|
| 1141 |
-
"learning_rate": 2.7578772789302998e-05,
|
| 1142 |
-
"loss": 0.1545,
|
| 1143 |
-
"step": 4050
|
| 1144 |
-
},
|
| 1145 |
-
{
|
| 1146 |
-
"epoch": 0.37861191117718107,
|
| 1147 |
-
"grad_norm": 6.033279895782471,
|
| 1148 |
-
"learning_rate": 2.744351235720059e-05,
|
| 1149 |
-
"loss": 0.1953,
|
| 1150 |
-
"step": 4075
|
| 1151 |
-
},
|
| 1152 |
-
{
|
| 1153 |
-
"epoch": 0.3809346836383908,
|
| 1154 |
-
"grad_norm": 8.570503234863281,
|
| 1155 |
-
"learning_rate": 2.7307855565783606e-05,
|
| 1156 |
-
"loss": 0.1828,
|
| 1157 |
-
"step": 4100
|
| 1158 |
-
},
|
| 1159 |
-
{
|
| 1160 |
-
"epoch": 0.3832574560996005,
|
| 1161 |
-
"grad_norm": 1.1502509117126465,
|
| 1162 |
-
"learning_rate": 2.7171809638635518e-05,
|
| 1163 |
-
"loss": 0.1591,
|
| 1164 |
-
"step": 4125
|
| 1165 |
-
},
|
| 1166 |
-
{
|
| 1167 |
-
"epoch": 0.3855802285608102,
|
| 1168 |
-
"grad_norm": 11.07186508178711,
|
| 1169 |
-
"learning_rate": 2.7035381820060877e-05,
|
| 1170 |
-
"loss": 0.1317,
|
| 1171 |
-
"step": 4150
|
| 1172 |
-
},
|
| 1173 |
-
{
|
| 1174 |
-
"epoch": 0.38790300102201986,
|
| 1175 |
-
"grad_norm": 0.5804902911186218,
|
| 1176 |
-
"learning_rate": 2.6898579374699552e-05,
|
| 1177 |
-
"loss": 0.1289,
|
| 1178 |
-
"step": 4175
|
| 1179 |
-
},
|
| 1180 |
-
{
|
| 1181 |
-
"epoch": 0.39022577348322957,
|
| 1182 |
-
"grad_norm": 7.553713321685791,
|
| 1183 |
-
"learning_rate": 2.6761409587139907e-05,
|
| 1184 |
-
"loss": 0.0621,
|
| 1185 |
-
"step": 4200
|
| 1186 |
-
},
|
| 1187 |
-
{
|
| 1188 |
-
"epoch": 0.3925485459444393,
|
| 1189 |
-
"grad_norm": 0.1034901887178421,
|
| 1190 |
-
"learning_rate": 2.6623879761530887e-05,
|
| 1191 |
-
"loss": 0.1567,
|
| 1192 |
-
"step": 4225
|
| 1193 |
-
},
|
| 1194 |
-
{
|
| 1195 |
-
"epoch": 0.394871318405649,
|
| 1196 |
-
"grad_norm": 18.33272933959961,
|
| 1197 |
-
"learning_rate": 2.64859972211931e-05,
|
| 1198 |
-
"loss": 0.1147,
|
| 1199 |
-
"step": 4250
|
| 1200 |
-
},
|
| 1201 |
-
{
|
| 1202 |
-
"epoch": 0.3971940908668587,
|
| 1203 |
-
"grad_norm": 0.11582828313112259,
|
| 1204 |
-
"learning_rate": 2.6347769308228853e-05,
|
| 1205 |
-
"loss": 0.2627,
|
| 1206 |
-
"step": 4275
|
| 1207 |
-
},
|
| 1208 |
-
{
|
| 1209 |
-
"epoch": 0.3995168633280684,
|
| 1210 |
-
"grad_norm": 0.764303982257843,
|
| 1211 |
-
"learning_rate": 2.6209203383131164e-05,
|
| 1212 |
-
"loss": 0.1529,
|
| 1213 |
-
"step": 4300
|
| 1214 |
-
},
|
| 1215 |
-
{
|
| 1216 |
-
"epoch": 0.40183963578927806,
|
| 1217 |
-
"grad_norm": 0.1588120013475418,
|
| 1218 |
-
"learning_rate": 2.6070306824391868e-05,
|
| 1219 |
-
"loss": 0.0722,
|
| 1220 |
-
"step": 4325
|
| 1221 |
-
},
|
| 1222 |
-
{
|
| 1223 |
-
"epoch": 0.40416240825048777,
|
| 1224 |
-
"grad_norm": 12.770928382873535,
|
| 1225 |
-
"learning_rate": 2.5931087028108684e-05,
|
| 1226 |
-
"loss": 0.1529,
|
| 1227 |
-
"step": 4350
|
| 1228 |
-
},
|
| 1229 |
-
{
|
| 1230 |
-
"epoch": 0.4064851807116975,
|
| 1231 |
-
"grad_norm": 0.11627359688282013,
|
| 1232 |
-
"learning_rate": 2.5791551407591405e-05,
|
| 1233 |
-
"loss": 0.1253,
|
| 1234 |
-
"step": 4375
|
| 1235 |
-
},
|
| 1236 |
-
{
|
| 1237 |
-
"epoch": 0.4088079531729072,
|
| 1238 |
-
"grad_norm": 0.06876557320356369,
|
| 1239 |
-
"learning_rate": 2.5651707392967146e-05,
|
| 1240 |
-
"loss": 0.0528,
|
| 1241 |
-
"step": 4400
|
| 1242 |
-
},
|
| 1243 |
-
{
|
| 1244 |
-
"epoch": 0.4111307256341169,
|
| 1245 |
-
"grad_norm": 3.2142481803894043,
|
| 1246 |
-
"learning_rate": 2.5511562430784658e-05,
|
| 1247 |
-
"loss": 0.1729,
|
| 1248 |
-
"step": 4425
|
| 1249 |
-
},
|
| 1250 |
-
{
|
| 1251 |
-
"epoch": 0.41345349809532655,
|
| 1252 |
-
"grad_norm": 0.10223668813705444,
|
| 1253 |
-
"learning_rate": 2.537112398361787e-05,
|
| 1254 |
-
"loss": 0.1409,
|
| 1255 |
-
"step": 4450
|
| 1256 |
-
},
|
| 1257 |
-
{
|
| 1258 |
-
"epoch": 0.41577627055653626,
|
| 1259 |
-
"grad_norm": 50.742740631103516,
|
| 1260 |
-
"learning_rate": 2.5230399529668457e-05,
|
| 1261 |
-
"loss": 0.1075,
|
| 1262 |
-
"step": 4475
|
| 1263 |
-
},
|
| 1264 |
-
{
|
| 1265 |
-
"epoch": 0.418099043017746,
|
| 1266 |
-
"grad_norm": 63.920711517333984,
|
| 1267 |
-
"learning_rate": 2.508939656236768e-05,
|
| 1268 |
-
"loss": 0.1804,
|
| 1269 |
-
"step": 4500
|
| 1270 |
-
},
|
| 1271 |
-
{
|
| 1272 |
-
"epoch": 0.4204218154789557,
|
| 1273 |
-
"grad_norm": 0.054011210799217224,
|
| 1274 |
-
"learning_rate": 2.4948122589977333e-05,
|
| 1275 |
-
"loss": 0.1272,
|
| 1276 |
-
"step": 4525
|
| 1277 |
-
},
|
| 1278 |
-
{
|
| 1279 |
-
"epoch": 0.4227445879401654,
|
| 1280 |
-
"grad_norm": 0.40179839730262756,
|
| 1281 |
-
"learning_rate": 2.4806585135189948e-05,
|
| 1282 |
-
"loss": 0.0517,
|
| 1283 |
-
"step": 4550
|
| 1284 |
-
},
|
| 1285 |
-
{
|
| 1286 |
-
"epoch": 0.4250673604013751,
|
| 1287 |
-
"grad_norm": 32.72251510620117,
|
| 1288 |
-
"learning_rate": 2.4664791734728234e-05,
|
| 1289 |
-
"loss": 0.1406,
|
| 1290 |
-
"step": 4575
|
| 1291 |
-
},
|
| 1292 |
-
{
|
| 1293 |
-
"epoch": 0.42739013286258476,
|
| 1294 |
-
"grad_norm": 0.17425654828548431,
|
| 1295 |
-
"learning_rate": 2.4522749938943736e-05,
|
| 1296 |
-
"loss": 0.169,
|
| 1297 |
-
"step": 4600
|
| 1298 |
-
},
|
| 1299 |
-
{
|
| 1300 |
-
"epoch": 0.42971290532379447,
|
| 1301 |
-
"grad_norm": 114.681884765625,
|
| 1302 |
-
"learning_rate": 2.438046731141479e-05,
|
| 1303 |
-
"loss": 0.118,
|
| 1304 |
-
"step": 4625
|
| 1305 |
-
},
|
| 1306 |
-
{
|
| 1307 |
-
"epoch": 0.4320356777850042,
|
| 1308 |
-
"grad_norm": 0.1432952582836151,
|
| 1309 |
-
"learning_rate": 2.4237951428543767e-05,
|
| 1310 |
-
"loss": 0.1776,
|
| 1311 |
-
"step": 4650
|
| 1312 |
-
},
|
| 1313 |
-
{
|
| 1314 |
-
"epoch": 0.4343584502462139,
|
| 1315 |
-
"grad_norm": 0.20776917040348053,
|
| 1316 |
-
"learning_rate": 2.4095209879153657e-05,
|
| 1317 |
-
"loss": 0.0692,
|
| 1318 |
-
"step": 4675
|
| 1319 |
-
},
|
| 1320 |
-
{
|
| 1321 |
-
"epoch": 0.4366812227074236,
|
| 1322 |
-
"grad_norm": 6.756126880645752,
|
| 1323 |
-
"learning_rate": 2.395225026408395e-05,
|
| 1324 |
-
"loss": 0.261,
|
| 1325 |
-
"step": 4700
|
| 1326 |
-
},
|
| 1327 |
-
{
|
| 1328 |
-
"epoch": 0.43900399516863325,
|
| 1329 |
-
"grad_norm": 1.958385705947876,
|
| 1330 |
-
"learning_rate": 2.3809080195785917e-05,
|
| 1331 |
-
"loss": 0.1665,
|
| 1332 |
-
"step": 4725
|
| 1333 |
-
},
|
| 1334 |
-
{
|
| 1335 |
-
"epoch": 0.44132676762984296,
|
| 1336 |
-
"grad_norm": 0.18160893023014069,
|
| 1337 |
-
"learning_rate": 2.3665707297917217e-05,
|
| 1338 |
-
"loss": 0.1139,
|
| 1339 |
-
"step": 4750
|
| 1340 |
-
},
|
| 1341 |
-
{
|
| 1342 |
-
"epoch": 0.4436495400910527,
|
| 1343 |
-
"grad_norm": 0.2513156831264496,
|
| 1344 |
-
"learning_rate": 2.352213920493601e-05,
|
| 1345 |
-
"loss": 0.087,
|
| 1346 |
-
"step": 4775
|
| 1347 |
-
},
|
| 1348 |
-
{
|
| 1349 |
-
"epoch": 0.4459723125522624,
|
| 1350 |
-
"grad_norm": 0.4901924729347229,
|
| 1351 |
-
"learning_rate": 2.3378383561694364e-05,
|
| 1352 |
-
"loss": 0.1155,
|
| 1353 |
-
"step": 4800
|
| 1354 |
-
},
|
| 1355 |
-
{
|
| 1356 |
-
"epoch": 0.4482950850134721,
|
| 1357 |
-
"grad_norm": 0.11683409661054611,
|
| 1358 |
-
"learning_rate": 2.323444802303123e-05,
|
| 1359 |
-
"loss": 0.1469,
|
| 1360 |
-
"step": 4825
|
| 1361 |
-
},
|
| 1362 |
-
{
|
| 1363 |
-
"epoch": 0.4506178574746818,
|
| 1364 |
-
"grad_norm": 2.1729214191436768,
|
| 1365 |
-
"learning_rate": 2.3090340253364777e-05,
|
| 1366 |
-
"loss": 0.1271,
|
| 1367 |
-
"step": 4850
|
| 1368 |
-
},
|
| 1369 |
-
{
|
| 1370 |
-
"epoch": 0.45294062993589146,
|
| 1371 |
-
"grad_norm": 160.86656188964844,
|
| 1372 |
-
"learning_rate": 2.2946067926284325e-05,
|
| 1373 |
-
"loss": 0.1358,
|
| 1374 |
-
"step": 4875
|
| 1375 |
-
},
|
| 1376 |
-
{
|
| 1377 |
-
"epoch": 0.45526340239710117,
|
| 1378 |
-
"grad_norm": 0.05550852045416832,
|
| 1379 |
-
"learning_rate": 2.2801638724141673e-05,
|
| 1380 |
-
"loss": 0.1778,
|
| 1381 |
-
"step": 4900
|
| 1382 |
-
},
|
| 1383 |
-
{
|
| 1384 |
-
"epoch": 0.4575861748583109,
|
| 1385 |
-
"grad_norm": 0.141483336687088,
|
| 1386 |
-
"learning_rate": 2.265706033764208e-05,
|
| 1387 |
-
"loss": 0.1143,
|
| 1388 |
-
"step": 4925
|
| 1389 |
-
},
|
| 1390 |
-
{
|
| 1391 |
-
"epoch": 0.4599089473195206,
|
| 1392 |
-
"grad_norm": 0.6958547234535217,
|
| 1393 |
-
"learning_rate": 2.251234046543471e-05,
|
| 1394 |
-
"loss": 0.1853,
|
| 1395 |
-
"step": 4950
|
| 1396 |
-
},
|
| 1397 |
-
{
|
| 1398 |
-
"epoch": 0.4622317197807303,
|
| 1399 |
-
"grad_norm": 0.052512262016534805,
|
| 1400 |
-
"learning_rate": 2.2367486813702682e-05,
|
| 1401 |
-
"loss": 0.1543,
|
| 1402 |
-
"step": 4975
|
| 1403 |
-
},
|
| 1404 |
-
{
|
| 1405 |
-
"epoch": 0.46455449224193995,
|
| 1406 |
-
"grad_norm": 24.44770622253418,
|
| 1407 |
-
"learning_rate": 2.2222507095752767e-05,
|
| 1408 |
-
"loss": 0.1154,
|
| 1409 |
-
"step": 5000
|
| 1410 |
-
},
|
| 1411 |
-
{
|
| 1412 |
-
"epoch": 0.46687726470314966,
|
| 1413 |
-
"grad_norm": 0.6328644752502441,
|
| 1414 |
-
"learning_rate": 2.2077409031604593e-05,
|
| 1415 |
-
"loss": 0.1172,
|
| 1416 |
-
"step": 5025
|
| 1417 |
-
},
|
| 1418 |
-
{
|
| 1419 |
-
"epoch": 0.4692000371643594,
|
| 1420 |
-
"grad_norm": 0.19583556056022644,
|
| 1421 |
-
"learning_rate": 2.1932200347579628e-05,
|
| 1422 |
-
"loss": 0.0929,
|
| 1423 |
-
"step": 5050
|
| 1424 |
-
},
|
| 1425 |
-
{
|
| 1426 |
-
"epoch": 0.4715228096255691,
|
| 1427 |
-
"grad_norm": 0.03581807389855385,
|
| 1428 |
-
"learning_rate": 2.1786888775889712e-05,
|
| 1429 |
-
"loss": 0.16,
|
| 1430 |
-
"step": 5075
|
| 1431 |
-
},
|
| 1432 |
-
{
|
| 1433 |
-
"epoch": 0.4738455820867788,
|
| 1434 |
-
"grad_norm": 0.059658344835042953,
|
| 1435 |
-
"learning_rate": 2.164148205422538e-05,
|
| 1436 |
-
"loss": 0.1197,
|
| 1437 |
-
"step": 5100
|
| 1438 |
-
},
|
| 1439 |
-
{
|
| 1440 |
-
"epoch": 0.4761683545479885,
|
| 1441 |
-
"grad_norm": 18.901350021362305,
|
| 1442 |
-
"learning_rate": 2.1495987925343762e-05,
|
| 1443 |
-
"loss": 0.1258,
|
| 1444 |
-
"step": 5125
|
| 1445 |
-
},
|
| 1446 |
-
{
|
| 1447 |
-
"epoch": 0.47849112700919816,
|
| 1448 |
-
"grad_norm": 0.3663087785243988,
|
| 1449 |
-
"learning_rate": 2.135041413665637e-05,
|
| 1450 |
-
"loss": 0.103,
|
| 1451 |
-
"step": 5150
|
| 1452 |
-
},
|
| 1453 |
-
{
|
| 1454 |
-
"epoch": 0.48081389947040787,
|
| 1455 |
-
"grad_norm": 1.0026763677597046,
|
| 1456 |
-
"learning_rate": 2.1204768439816487e-05,
|
| 1457 |
-
"loss": 0.1339,
|
| 1458 |
-
"step": 5175
|
| 1459 |
-
},
|
| 1460 |
-
{
|
| 1461 |
-
"epoch": 0.4831366719316176,
|
| 1462 |
-
"grad_norm": 0.8399305939674377,
|
| 1463 |
-
"learning_rate": 2.1059058590306454e-05,
|
| 1464 |
-
"loss": 0.2015,
|
| 1465 |
-
"step": 5200
|
| 1466 |
-
},
|
| 1467 |
-
{
|
| 1468 |
-
"epoch": 0.4854594443928273,
|
| 1469 |
-
"grad_norm": 7.705094337463379,
|
| 1470 |
-
"learning_rate": 2.0913292347024662e-05,
|
| 1471 |
-
"loss": 0.0892,
|
| 1472 |
-
"step": 5225
|
| 1473 |
-
},
|
| 1474 |
-
{
|
| 1475 |
-
"epoch": 0.487782216854037,
|
| 1476 |
-
"grad_norm": 0.28549614548683167,
|
| 1477 |
-
"learning_rate": 2.0767477471872417e-05,
|
| 1478 |
-
"loss": 0.2087,
|
| 1479 |
-
"step": 5250
|
| 1480 |
-
},
|
| 1481 |
-
{
|
| 1482 |
-
"epoch": 0.49010498931524665,
|
| 1483 |
-
"grad_norm": 0.018054131418466568,
|
| 1484 |
-
"learning_rate": 2.0621621729340616e-05,
|
| 1485 |
-
"loss": 0.1061,
|
| 1486 |
-
"step": 5275
|
| 1487 |
-
},
|
| 1488 |
-
{
|
| 1489 |
-
"epoch": 0.49242776177645636,
|
| 1490 |
-
"grad_norm": 17.231143951416016,
|
| 1491 |
-
"learning_rate": 2.047573288609631e-05,
|
| 1492 |
-
"loss": 0.2437,
|
| 1493 |
-
"step": 5300
|
| 1494 |
-
},
|
| 1495 |
-
{
|
| 1496 |
-
"epoch": 0.4947505342376661,
|
| 1497 |
-
"grad_norm": 0.08126140385866165,
|
| 1498 |
-
"learning_rate": 2.0329818710569125e-05,
|
| 1499 |
-
"loss": 0.1395,
|
| 1500 |
-
"step": 5325
|
| 1501 |
-
},
|
| 1502 |
-
{
|
| 1503 |
-
"epoch": 0.4970733066988758,
|
| 1504 |
-
"grad_norm": 0.13981200754642487,
|
| 1505 |
-
"learning_rate": 2.0183886972537602e-05,
|
| 1506 |
-
"loss": 0.061,
|
| 1507 |
-
"step": 5350
|
| 1508 |
-
},
|
| 1509 |
-
{
|
| 1510 |
-
"epoch": 0.4993960791600855,
|
| 1511 |
-
"grad_norm": 0.13427792489528656,
|
| 1512 |
-
"learning_rate": 2.003794544271547e-05,
|
| 1513 |
-
"loss": 0.1639,
|
| 1514 |
-
"step": 5375
|
| 1515 |
-
},
|
| 1516 |
-
{
|
| 1517 |
-
"epoch": 0.5017188516212951,
|
| 1518 |
-
"grad_norm": 0.0515735000371933,
|
| 1519 |
-
"learning_rate": 1.9892001892337864e-05,
|
| 1520 |
-
"loss": 0.0613,
|
| 1521 |
-
"step": 5400
|
| 1522 |
-
},
|
| 1523 |
-
{
|
| 1524 |
-
"epoch": 0.5040416240825049,
|
| 1525 |
-
"grad_norm": 0.3217664659023285,
|
| 1526 |
-
"learning_rate": 1.9746064092747504e-05,
|
| 1527 |
-
"loss": 0.1095,
|
| 1528 |
-
"step": 5425
|
| 1529 |
-
},
|
| 1530 |
-
{
|
| 1531 |
-
"epoch": 0.5063643965437146,
|
| 1532 |
-
"grad_norm": 0.5333023071289062,
|
| 1533 |
-
"learning_rate": 1.9600139814980884e-05,
|
| 1534 |
-
"loss": 0.1136,
|
| 1535 |
-
"step": 5450
|
| 1536 |
-
},
|
| 1537 |
-
{
|
| 1538 |
-
"epoch": 0.5086871690049243,
|
| 1539 |
-
"grad_norm": 6.019542217254639,
|
| 1540 |
-
"learning_rate": 1.945423682935449e-05,
|
| 1541 |
-
"loss": 0.1337,
|
| 1542 |
-
"step": 5475
|
| 1543 |
-
},
|
| 1544 |
-
{
|
| 1545 |
-
"epoch": 0.511009941466134,
|
| 1546 |
-
"grad_norm": 10.347049713134766,
|
| 1547 |
-
"learning_rate": 1.930836290505101e-05,
|
| 1548 |
-
"loss": 0.1969,
|
| 1549 |
-
"step": 5500
|
| 1550 |
-
},
|
| 1551 |
-
{
|
| 1552 |
-
"epoch": 0.5133327139273437,
|
| 1553 |
-
"grad_norm": 0.13794614374637604,
|
| 1554 |
-
"learning_rate": 1.916252580970566e-05,
|
| 1555 |
-
"loss": 0.1853,
|
| 1556 |
-
"step": 5525
|
| 1557 |
-
},
|
| 1558 |
-
{
|
| 1559 |
-
"epoch": 0.5156554863885534,
|
| 1560 |
-
"grad_norm": 0.5816416144371033,
|
| 1561 |
-
"learning_rate": 1.9016733308992527e-05,
|
| 1562 |
-
"loss": 0.1323,
|
| 1563 |
-
"step": 5550
|
| 1564 |
-
},
|
| 1565 |
-
{
|
| 1566 |
-
"epoch": 0.5179782588497631,
|
| 1567 |
-
"grad_norm": 0.3536951243877411,
|
| 1568 |
-
"learning_rate": 1.8870993166211114e-05,
|
| 1569 |
-
"loss": 0.1425,
|
| 1570 |
-
"step": 5575
|
| 1571 |
-
},
|
| 1572 |
-
{
|
| 1573 |
-
"epoch": 0.5203010313109728,
|
| 1574 |
-
"grad_norm": 0.28000468015670776,
|
| 1575 |
-
"learning_rate": 1.8725313141872886e-05,
|
| 1576 |
-
"loss": 0.1609,
|
| 1577 |
-
"step": 5600
|
| 1578 |
-
},
|
| 1579 |
-
{
|
| 1580 |
-
"epoch": 0.5226238037721824,
|
| 1581 |
-
"grad_norm": 0.07619854062795639,
|
| 1582 |
-
"learning_rate": 1.8579700993288094e-05,
|
| 1583 |
-
"loss": 0.1499,
|
| 1584 |
-
"step": 5625
|
| 1585 |
-
},
|
| 1586 |
-
{
|
| 1587 |
-
"epoch": 0.5249465762333921,
|
| 1588 |
-
"grad_norm": 13.564208984375,
|
| 1589 |
-
"learning_rate": 1.843416447415265e-05,
|
| 1590 |
-
"loss": 0.1227,
|
| 1591 |
-
"step": 5650
|
| 1592 |
-
},
|
| 1593 |
-
{
|
| 1594 |
-
"epoch": 0.5272693486946018,
|
| 1595 |
-
"grad_norm": 0.15668974816799164,
|
| 1596 |
-
"learning_rate": 1.828871133413528e-05,
|
| 1597 |
-
"loss": 0.0853,
|
| 1598 |
-
"step": 5675
|
| 1599 |
-
},
|
| 1600 |
-
{
|
| 1601 |
-
"epoch": 0.5295921211558116,
|
| 1602 |
-
"grad_norm": 2.117285966873169,
|
| 1603 |
-
"learning_rate": 1.8143349318464876e-05,
|
| 1604 |
-
"loss": 0.0532,
|
| 1605 |
-
"step": 5700
|
| 1606 |
-
},
|
| 1607 |
-
{
|
| 1608 |
-
"epoch": 0.5319148936170213,
|
| 1609 |
-
"grad_norm": 0.030197104439139366,
|
| 1610 |
-
"learning_rate": 1.7998086167518033e-05,
|
| 1611 |
-
"loss": 0.1767,
|
| 1612 |
-
"step": 5725
|
| 1613 |
-
},
|
| 1614 |
-
{
|
| 1615 |
-
"epoch": 0.534237666078231,
|
| 1616 |
-
"grad_norm": 15.744141578674316,
|
| 1617 |
-
"learning_rate": 1.7852929616406922e-05,
|
| 1618 |
-
"loss": 0.0922,
|
| 1619 |
-
"step": 5750
|
| 1620 |
-
},
|
| 1621 |
-
{
|
| 1622 |
-
"epoch": 0.5365604385394407,
|
| 1623 |
-
"grad_norm": 30.61720848083496,
|
| 1624 |
-
"learning_rate": 1.7707887394567355e-05,
|
| 1625 |
-
"loss": 0.201,
|
| 1626 |
-
"step": 5775
|
| 1627 |
-
},
|
| 1628 |
-
{
|
| 1629 |
-
"epoch": 0.5388832110006504,
|
| 1630 |
-
"grad_norm": 0.2354450225830078,
|
| 1631 |
-
"learning_rate": 1.7562967225347262e-05,
|
| 1632 |
-
"loss": 0.1371,
|
| 1633 |
-
"step": 5800
|
| 1634 |
-
},
|
| 1635 |
-
{
|
| 1636 |
-
"epoch": 0.5412059834618601,
|
| 1637 |
-
"grad_norm": 0.06641119718551636,
|
| 1638 |
-
"learning_rate": 1.741817682559537e-05,
|
| 1639 |
-
"loss": 0.1178,
|
| 1640 |
-
"step": 5825
|
| 1641 |
-
},
|
| 1642 |
-
{
|
| 1643 |
-
"epoch": 0.5435287559230698,
|
| 1644 |
-
"grad_norm": 0.09019186347723007,
|
| 1645 |
-
"learning_rate": 1.7273523905250334e-05,
|
| 1646 |
-
"loss": 0.0818,
|
| 1647 |
-
"step": 5850
|
| 1648 |
-
},
|
| 1649 |
-
{
|
| 1650 |
-
"epoch": 0.5458515283842795,
|
| 1651 |
-
"grad_norm": 0.14201141893863678,
|
| 1652 |
-
"learning_rate": 1.712901616693016e-05,
|
| 1653 |
-
"loss": 0.1256,
|
| 1654 |
-
"step": 5875
|
| 1655 |
-
},
|
| 1656 |
-
{
|
| 1657 |
-
"epoch": 0.5481743008454891,
|
| 1658 |
-
"grad_norm": 0.2438458800315857,
|
| 1659 |
-
"learning_rate": 1.698466130552208e-05,
|
| 1660 |
-
"loss": 0.1056,
|
| 1661 |
-
"step": 5900
|
| 1662 |
-
},
|
| 1663 |
-
{
|
| 1664 |
-
"epoch": 0.5504970733066988,
|
| 1665 |
-
"grad_norm": 0.7838077545166016,
|
| 1666 |
-
"learning_rate": 1.6840467007772766e-05,
|
| 1667 |
-
"loss": 0.127,
|
| 1668 |
-
"step": 5925
|
| 1669 |
-
},
|
| 1670 |
-
{
|
| 1671 |
-
"epoch": 0.5528198457679085,
|
| 1672 |
-
"grad_norm": 312.2820129394531,
|
| 1673 |
-
"learning_rate": 1.6696440951879056e-05,
|
| 1674 |
-
"loss": 0.0944,
|
| 1675 |
-
"step": 5950
|
| 1676 |
-
},
|
| 1677 |
-
{
|
| 1678 |
-
"epoch": 0.5551426182291183,
|
| 1679 |
-
"grad_norm": 0.30949246883392334,
|
| 1680 |
-
"learning_rate": 1.655259080707908e-05,
|
| 1681 |
-
"loss": 0.0798,
|
| 1682 |
-
"step": 5975
|
| 1683 |
-
},
|
| 1684 |
-
{
|
| 1685 |
-
"epoch": 0.557465390690328,
|
| 1686 |
-
"grad_norm": 0.029276683926582336,
|
| 1687 |
-
"learning_rate": 1.6408924233243885e-05,
|
| 1688 |
-
"loss": 0.0775,
|
| 1689 |
-
"step": 6000
|
| 1690 |
-
},
|
| 1691 |
-
{
|
| 1692 |
-
"epoch": 0.5597881631515377,
|
| 1693 |
-
"grad_norm": 4.421539783477783,
|
| 1694 |
-
"learning_rate": 1.6265448880469562e-05,
|
| 1695 |
-
"loss": 0.0698,
|
| 1696 |
-
"step": 6025
|
| 1697 |
-
},
|
| 1698 |
-
{
|
| 1699 |
-
"epoch": 0.5621109356127474,
|
| 1700 |
-
"grad_norm": 0.6649914979934692,
|
| 1701 |
-
"learning_rate": 1.6122172388669863e-05,
|
| 1702 |
-
"loss": 0.1132,
|
| 1703 |
-
"step": 6050
|
| 1704 |
-
},
|
| 1705 |
-
{
|
| 1706 |
-
"epoch": 0.5644337080739571,
|
| 1707 |
-
"grad_norm": 0.07938749343156815,
|
| 1708 |
-
"learning_rate": 1.597910238716941e-05,
|
| 1709 |
-
"loss": 0.1072,
|
| 1710 |
-
"step": 6075
|
| 1711 |
-
},
|
| 1712 |
-
{
|
| 1713 |
-
"epoch": 0.5667564805351668,
|
| 1714 |
-
"grad_norm": 17.54945182800293,
|
| 1715 |
-
"learning_rate": 1.583624649429743e-05,
|
| 1716 |
-
"loss": 0.1522,
|
| 1717 |
-
"step": 6100
|
| 1718 |
-
},
|
| 1719 |
-
{
|
| 1720 |
-
"epoch": 0.5690792529963765,
|
| 1721 |
-
"grad_norm": 0.1551954746246338,
|
| 1722 |
-
"learning_rate": 1.569361231698208e-05,
|
| 1723 |
-
"loss": 0.1588,
|
| 1724 |
-
"step": 6125
|
| 1725 |
-
},
|
| 1726 |
-
{
|
| 1727 |
-
"epoch": 0.5714020254575862,
|
| 1728 |
-
"grad_norm": 11.957656860351562,
|
| 1729 |
-
"learning_rate": 1.5551207450345412e-05,
|
| 1730 |
-
"loss": 0.0767,
|
| 1731 |
-
"step": 6150
|
| 1732 |
-
},
|
| 1733 |
-
{
|
| 1734 |
-
"epoch": 0.5737247979187958,
|
| 1735 |
-
"grad_norm": 0.04506440460681915,
|
| 1736 |
-
"learning_rate": 1.540903947729891e-05,
|
| 1737 |
-
"loss": 0.1059,
|
| 1738 |
-
"step": 6175
|
| 1739 |
-
},
|
| 1740 |
-
{
|
| 1741 |
-
"epoch": 0.5760475703800055,
|
| 1742 |
-
"grad_norm": 0.23562848567962646,
|
| 1743 |
-
"learning_rate": 1.52671159681397e-05,
|
| 1744 |
-
"loss": 0.0996,
|
| 1745 |
-
"step": 6200
|
| 1746 |
-
},
|
| 1747 |
-
{
|
| 1748 |
-
"epoch": 0.5783703428412152,
|
| 1749 |
-
"grad_norm": 0.6889678239822388,
|
| 1750 |
-
"learning_rate": 1.5125444480147497e-05,
|
| 1751 |
-
"loss": 0.1168,
|
| 1752 |
-
"step": 6225
|
| 1753 |
-
},
|
| 1754 |
-
{
|
| 1755 |
-
"epoch": 0.580693115302425,
|
| 1756 |
-
"grad_norm": 0.376047283411026,
|
| 1757 |
-
"learning_rate": 1.4984032557182125e-05,
|
| 1758 |
-
"loss": 0.1244,
|
| 1759 |
-
"step": 6250
|
| 1760 |
-
},
|
| 1761 |
-
{
|
| 1762 |
-
"epoch": 0.5830158877636347,
|
| 1763 |
-
"grad_norm": 0.16828536987304688,
|
| 1764 |
-
"learning_rate": 1.484288772928185e-05,
|
| 1765 |
-
"loss": 0.092,
|
| 1766 |
-
"step": 6275
|
| 1767 |
-
},
|
| 1768 |
-
{
|
| 1769 |
-
"epoch": 0.5853386602248444,
|
| 1770 |
-
"grad_norm": 0.04910039156675339,
|
| 1771 |
-
"learning_rate": 1.4702017512262396e-05,
|
| 1772 |
-
"loss": 0.209,
|
| 1773 |
-
"step": 6300
|
| 1774 |
-
},
|
| 1775 |
-
{
|
| 1776 |
-
"epoch": 0.5876614326860541,
|
| 1777 |
-
"grad_norm": 0.16525715589523315,
|
| 1778 |
-
"learning_rate": 1.4561429407316752e-05,
|
| 1779 |
-
"loss": 0.0471,
|
| 1780 |
-
"step": 6325
|
| 1781 |
-
},
|
| 1782 |
-
{
|
| 1783 |
-
"epoch": 0.5899842051472638,
|
| 1784 |
-
"grad_norm": 0.4152815043926239,
|
| 1785 |
-
"learning_rate": 1.4421130900615722e-05,
|
| 1786 |
-
"loss": 0.0902,
|
| 1787 |
-
"step": 6350
|
| 1788 |
-
},
|
| 1789 |
-
{
|
| 1790 |
-
"epoch": 0.5923069776084735,
|
| 1791 |
-
"grad_norm": 0.9569645524024963,
|
| 1792 |
-
"learning_rate": 1.4281129462909307e-05,
|
| 1793 |
-
"loss": 0.1231,
|
| 1794 |
-
"step": 6375
|
| 1795 |
-
},
|
| 1796 |
-
{
|
| 1797 |
-
"epoch": 0.5946297500696832,
|
| 1798 |
-
"grad_norm": 0.1703191101551056,
|
| 1799 |
-
"learning_rate": 1.414143254912889e-05,
|
| 1800 |
-
"loss": 0.0772,
|
| 1801 |
-
"step": 6400
|
| 1802 |
-
},
|
| 1803 |
-
{
|
| 1804 |
-
"epoch": 0.5969525225308929,
|
| 1805 |
-
"grad_norm": 20.376893997192383,
|
| 1806 |
-
"learning_rate": 1.4002047597990283e-05,
|
| 1807 |
-
"loss": 0.1987,
|
| 1808 |
-
"step": 6425
|
| 1809 |
-
},
|
| 1810 |
-
{
|
| 1811 |
-
"epoch": 0.5992752949921025,
|
| 1812 |
-
"grad_norm": 26.851207733154297,
|
| 1813 |
-
"learning_rate": 1.3862982031597593e-05,
|
| 1814 |
-
"loss": 0.1681,
|
| 1815 |
-
"step": 6450
|
| 1816 |
-
},
|
| 1817 |
-
{
|
| 1818 |
-
"epoch": 0.6015980674533122,
|
| 1819 |
-
"grad_norm": 13.581109046936035,
|
| 1820 |
-
"learning_rate": 1.3724243255048028e-05,
|
| 1821 |
-
"loss": 0.2194,
|
| 1822 |
-
"step": 6475
|
| 1823 |
-
},
|
| 1824 |
-
{
|
| 1825 |
-
"epoch": 0.603920839914522,
|
| 1826 |
-
"grad_norm": 0.042847760021686554,
|
| 1827 |
-
"learning_rate": 1.3585838656037577e-05,
|
| 1828 |
-
"loss": 0.0724,
|
| 1829 |
-
"step": 6500
|
| 1830 |
-
},
|
| 1831 |
-
{
|
| 1832 |
-
"epoch": 0.6062436123757317,
|
| 1833 |
-
"grad_norm": 0.04649437963962555,
|
| 1834 |
-
"learning_rate": 1.344777560446762e-05,
|
| 1835 |
-
"loss": 0.108,
|
| 1836 |
-
"step": 6525
|
| 1837 |
-
},
|
| 1838 |
-
{
|
| 1839 |
-
"epoch": 0.6085663848369414,
|
| 1840 |
-
"grad_norm": 0.1492205411195755,
|
| 1841 |
-
"learning_rate": 1.3310061452052493e-05,
|
| 1842 |
-
"loss": 0.1783,
|
| 1843 |
-
"step": 6550
|
| 1844 |
-
},
|
| 1845 |
-
{
|
| 1846 |
-
"epoch": 0.6108891572981511,
|
| 1847 |
-
"grad_norm": 0.9922373294830322,
|
| 1848 |
-
"learning_rate": 1.317270353192801e-05,
|
| 1849 |
-
"loss": 0.0934,
|
| 1850 |
-
"step": 6575
|
| 1851 |
-
},
|
| 1852 |
-
{
|
| 1853 |
-
"epoch": 0.6132119297593608,
|
| 1854 |
-
"grad_norm": 1.0141760110855103,
|
| 1855 |
-
"learning_rate": 1.3035709158260987e-05,
|
| 1856 |
-
"loss": 0.0989,
|
| 1857 |
-
"step": 6600
|
| 1858 |
-
},
|
| 1859 |
-
{
|
| 1860 |
-
"epoch": 0.6155347022205705,
|
| 1861 |
-
"grad_norm": 0.17959514260292053,
|
| 1862 |
-
"learning_rate": 1.2899085625859767e-05,
|
| 1863 |
-
"loss": 0.0344,
|
| 1864 |
-
"step": 6625
|
| 1865 |
-
},
|
| 1866 |
-
{
|
| 1867 |
-
"epoch": 0.6178574746817802,
|
| 1868 |
-
"grad_norm": 2.6247718334198,
|
| 1869 |
-
"learning_rate": 1.2762840209785784e-05,
|
| 1870 |
-
"loss": 0.1481,
|
| 1871 |
-
"step": 6650
|
| 1872 |
-
},
|
| 1873 |
-
{
|
| 1874 |
-
"epoch": 0.6201802471429899,
|
| 1875 |
-
"grad_norm": 15.859939575195312,
|
| 1876 |
-
"learning_rate": 1.2626980164966156e-05,
|
| 1877 |
-
"loss": 0.1347,
|
| 1878 |
-
"step": 6675
|
| 1879 |
-
},
|
| 1880 |
-
{
|
| 1881 |
-
"epoch": 0.6225030196041996,
|
| 1882 |
-
"grad_norm": 0.06661898642778397,
|
| 1883 |
-
"learning_rate": 1.2491512725807406e-05,
|
| 1884 |
-
"loss": 0.1138,
|
| 1885 |
-
"step": 6700
|
| 1886 |
-
},
|
| 1887 |
-
{
|
| 1888 |
-
"epoch": 0.6248257920654092,
|
| 1889 |
-
"grad_norm": 0.03395082801580429,
|
| 1890 |
-
"learning_rate": 1.2356445105810196e-05,
|
| 1891 |
-
"loss": 0.1807,
|
| 1892 |
-
"step": 6725
|
| 1893 |
-
},
|
| 1894 |
-
{
|
| 1895 |
-
"epoch": 0.6271485645266189,
|
| 1896 |
-
"grad_norm": 0.443010538816452,
|
| 1897 |
-
"learning_rate": 1.2221784497185216e-05,
|
| 1898 |
-
"loss": 0.0829,
|
| 1899 |
-
"step": 6750
|
| 1900 |
-
},
|
| 1901 |
-
{
|
| 1902 |
-
"epoch": 0.6294713369878286,
|
| 1903 |
-
"grad_norm": 10.890091896057129,
|
| 1904 |
-
"learning_rate": 1.2087538070470261e-05,
|
| 1905 |
-
"loss": 0.2418,
|
| 1906 |
-
"step": 6775
|
| 1907 |
-
},
|
| 1908 |
-
{
|
| 1909 |
-
"epoch": 0.6317941094490384,
|
| 1910 |
-
"grad_norm": 0.17275892198085785,
|
| 1911 |
-
"learning_rate": 1.1953712974148339e-05,
|
| 1912 |
-
"loss": 0.1344,
|
| 1913 |
-
"step": 6800
|
| 1914 |
-
},
|
| 1915 |
-
{
|
| 1916 |
-
"epoch": 0.6341168819102481,
|
| 1917 |
-
"grad_norm": 13.537515640258789,
|
| 1918 |
-
"learning_rate": 1.1820316334267068e-05,
|
| 1919 |
-
"loss": 0.113,
|
| 1920 |
-
"step": 6825
|
| 1921 |
-
},
|
| 1922 |
-
{
|
| 1923 |
-
"epoch": 0.6364396543714578,
|
| 1924 |
-
"grad_norm": 9.592342376708984,
|
| 1925 |
-
"learning_rate": 1.1687355254059189e-05,
|
| 1926 |
-
"loss": 0.1304,
|
| 1927 |
-
"step": 6850
|
| 1928 |
-
},
|
| 1929 |
-
{
|
| 1930 |
-
"epoch": 0.6387624268326675,
|
| 1931 |
-
"grad_norm": 18.435609817504883,
|
| 1932 |
-
"learning_rate": 1.1554836813564356e-05,
|
| 1933 |
-
"loss": 0.0742,
|
| 1934 |
-
"step": 6875
|
| 1935 |
-
},
|
| 1936 |
-
{
|
| 1937 |
-
"epoch": 0.6410851992938772,
|
| 1938 |
-
"grad_norm": 10.369148254394531,
|
| 1939 |
-
"learning_rate": 1.1422768069252124e-05,
|
| 1940 |
-
"loss": 0.0829,
|
| 1941 |
-
"step": 6900
|
| 1942 |
-
},
|
| 1943 |
-
{
|
| 1944 |
-
"epoch": 0.6434079717550869,
|
| 1945 |
-
"grad_norm": 0.5392451882362366,
|
| 1946 |
-
"learning_rate": 1.1291156053646157e-05,
|
| 1947 |
-
"loss": 0.1044,
|
| 1948 |
-
"step": 6925
|
| 1949 |
-
},
|
| 1950 |
-
{
|
| 1951 |
-
"epoch": 0.6457307442162966,
|
| 1952 |
-
"grad_norm": 0.29093310236930847,
|
| 1953 |
-
"learning_rate": 1.1160007774949828e-05,
|
| 1954 |
-
"loss": 0.1483,
|
| 1955 |
-
"step": 6950
|
| 1956 |
-
},
|
| 1957 |
-
{
|
| 1958 |
-
"epoch": 0.6480535166775063,
|
| 1959 |
-
"grad_norm": 0.259670615196228,
|
| 1960 |
-
"learning_rate": 1.1029330216672988e-05,
|
| 1961 |
-
"loss": 0.1249,
|
| 1962 |
-
"step": 6975
|
| 1963 |
-
},
|
| 1964 |
-
{
|
| 1965 |
-
"epoch": 0.6503762891387159,
|
| 1966 |
-
"grad_norm": 21.401098251342773,
|
| 1967 |
-
"learning_rate": 1.0899130337260106e-05,
|
| 1968 |
-
"loss": 0.1463,
|
| 1969 |
-
"step": 7000
|
| 1970 |
-
},
|
| 1971 |
-
{
|
| 1972 |
-
"epoch": 0.6526990615999256,
|
| 1973 |
-
"grad_norm": 8.750685691833496,
|
| 1974 |
-
"learning_rate": 1.0769415069719744e-05,
|
| 1975 |
-
"loss": 0.1642,
|
| 1976 |
-
"step": 7025
|
| 1977 |
-
},
|
| 1978 |
-
{
|
| 1979 |
-
"epoch": 0.6550218340611353,
|
| 1980 |
-
"grad_norm": 1.4688172340393066,
|
| 1981 |
-
"learning_rate": 1.0640191321255401e-05,
|
| 1982 |
-
"loss": 0.1255,
|
| 1983 |
-
"step": 7050
|
| 1984 |
-
},
|
| 1985 |
-
{
|
| 1986 |
-
"epoch": 0.657344606522345,
|
| 1987 |
-
"grad_norm": 0.08441878110170364,
|
| 1988 |
-
"learning_rate": 1.0511465972897685e-05,
|
| 1989 |
-
"loss": 0.0936,
|
| 1990 |
-
"step": 7075
|
| 1991 |
-
},
|
| 1992 |
-
{
|
| 1993 |
-
"epoch": 0.6596673789835548,
|
| 1994 |
-
"grad_norm": 0.051643550395965576,
|
| 1995 |
-
"learning_rate": 1.0383245879137922e-05,
|
| 1996 |
-
"loss": 0.0948,
|
| 1997 |
-
"step": 7100
|
| 1998 |
-
},
|
| 1999 |
-
{
|
| 2000 |
-
"epoch": 0.6619901514447645,
|
| 2001 |
-
"grad_norm": 0.16709965467453003,
|
| 2002 |
-
"learning_rate": 1.0255537867563135e-05,
|
| 2003 |
-
"loss": 0.1037,
|
| 2004 |
-
"step": 7125
|
| 2005 |
-
},
|
| 2006 |
-
{
|
| 2007 |
-
"epoch": 0.6643129239059742,
|
| 2008 |
-
"grad_norm": 0.1178256943821907,
|
| 2009 |
-
"learning_rate": 1.0128348738492523e-05,
|
| 2010 |
-
"loss": 0.0922,
|
| 2011 |
-
"step": 7150
|
| 2012 |
-
},
|
| 2013 |
-
{
|
| 2014 |
-
"epoch": 0.6666356963671839,
|
| 2015 |
-
"grad_norm": 40.31432342529297,
|
| 2016 |
-
"learning_rate": 1.0001685264615322e-05,
|
| 2017 |
-
"loss": 0.0666,
|
| 2018 |
-
"step": 7175
|
| 2019 |
-
},
|
| 2020 |
-
{
|
| 2021 |
-
"epoch": 0.6689584688283936,
|
| 2022 |
-
"grad_norm": 0.8184968829154968,
|
| 2023 |
-
"learning_rate": 9.875554190630184e-06,
|
| 2024 |
-
"loss": 0.0335,
|
| 2025 |
-
"step": 7200
|
| 2026 |
-
},
|
| 2027 |
-
{
|
| 2028 |
-
"epoch": 0.6712812412896033,
|
| 2029 |
-
"grad_norm": 0.09114021807909012,
|
| 2030 |
-
"learning_rate": 9.749962232885987e-06,
|
| 2031 |
-
"loss": 0.0207,
|
| 2032 |
-
"step": 7225
|
| 2033 |
-
},
|
| 2034 |
-
{
|
| 2035 |
-
"epoch": 0.673604013750813,
|
| 2036 |
-
"grad_norm": 0.22469620406627655,
|
| 2037 |
-
"learning_rate": 9.624916079024266e-06,
|
| 2038 |
-
"loss": 0.116,
|
| 2039 |
-
"step": 7250
|
| 2040 |
-
},
|
| 2041 |
-
{
|
| 2042 |
-
"epoch": 0.6759267862120226,
|
| 2043 |
-
"grad_norm": 10.686841011047363,
|
| 2044 |
-
"learning_rate": 9.500422387623054e-06,
|
| 2045 |
-
"loss": 0.1441,
|
| 2046 |
-
"step": 7275
|
| 2047 |
-
},
|
| 2048 |
-
{
|
| 2049 |
-
"epoch": 0.6782495586732323,
|
| 2050 |
-
"grad_norm": 16.609556198120117,
|
| 2051 |
-
"learning_rate": 9.376487787842307e-06,
|
| 2052 |
-
"loss": 0.1856,
|
| 2053 |
-
"step": 7300
|
| 2054 |
-
},
|
| 2055 |
-
{
|
| 2056 |
-
"epoch": 0.680572331134442,
|
| 2057 |
-
"grad_norm": 14.723616600036621,
|
| 2058 |
-
"learning_rate": 9.253118879070973e-06,
|
| 2059 |
-
"loss": 0.0956,
|
| 2060 |
-
"step": 7325
|
| 2061 |
-
},
|
| 2062 |
-
{
|
| 2063 |
-
"epoch": 0.6828951035956518,
|
| 2064 |
-
"grad_norm": 1.1138169765472412,
|
| 2065 |
-
"learning_rate": 9.130322230575497e-06,
|
| 2066 |
-
"loss": 0.0808,
|
| 2067 |
-
"step": 7350
|
| 2068 |
-
},
|
| 2069 |
-
{
|
| 2070 |
-
"epoch": 0.6852178760568615,
|
| 2071 |
-
"grad_norm": 0.119639553129673,
|
| 2072 |
-
"learning_rate": 9.008104381150092e-06,
|
| 2073 |
-
"loss": 0.2254,
|
| 2074 |
-
"step": 7375
|
| 2075 |
-
},
|
| 2076 |
-
{
|
| 2077 |
-
"epoch": 0.6875406485180712,
|
| 2078 |
-
"grad_norm": 0.07603152841329575,
|
| 2079 |
-
"learning_rate": 8.886471838768504e-06,
|
| 2080 |
-
"loss": 0.0681,
|
| 2081 |
-
"step": 7400
|
| 2082 |
-
},
|
| 2083 |
-
{
|
| 2084 |
-
"epoch": 0.6898634209792809,
|
| 2085 |
-
"grad_norm": 2.726926326751709,
|
| 2086 |
-
"learning_rate": 8.765431080237499e-06,
|
| 2087 |
-
"loss": 0.1014,
|
| 2088 |
-
"step": 7425
|
| 2089 |
-
},
|
| 2090 |
-
{
|
| 2091 |
-
"epoch": 0.6921861934404906,
|
| 2092 |
-
"grad_norm": 7.617353916168213,
|
| 2093 |
-
"learning_rate": 8.644988550851943e-06,
|
| 2094 |
-
"loss": 0.0662,
|
| 2095 |
-
"step": 7450
|
| 2096 |
-
},
|
| 2097 |
-
{
|
| 2098 |
-
"epoch": 0.6945089659017003,
|
| 2099 |
-
"grad_norm": 0.20352880656719208,
|
| 2100 |
-
"learning_rate": 8.525150664051643e-06,
|
| 2101 |
-
"loss": 0.0973,
|
| 2102 |
-
"step": 7475
|
| 2103 |
-
},
|
| 2104 |
-
{
|
| 2105 |
-
"epoch": 0.69683173836291,
|
| 2106 |
-
"grad_norm": 12.325729370117188,
|
| 2107 |
-
"learning_rate": 8.405923801079802e-06,
|
| 2108 |
-
"loss": 0.1741,
|
| 2109 |
-
"step": 7500
|
| 2110 |
-
},
|
| 2111 |
-
{
|
| 2112 |
-
"epoch": 0.6991545108241197,
|
| 2113 |
-
"grad_norm": 0.7742374539375305,
|
| 2114 |
-
"learning_rate": 8.287314310643254e-06,
|
| 2115 |
-
"loss": 0.0956,
|
| 2116 |
-
"step": 7525
|
| 2117 |
-
},
|
| 2118 |
-
{
|
| 2119 |
-
"epoch": 0.7014772832853293,
|
| 2120 |
-
"grad_norm": 264.7276916503906,
|
| 2121 |
-
"learning_rate": 8.169328508574368e-06,
|
| 2122 |
-
"loss": 0.1212,
|
| 2123 |
-
"step": 7550
|
| 2124 |
-
},
|
| 2125 |
-
{
|
| 2126 |
-
"epoch": 0.703800055746539,
|
| 2127 |
-
"grad_norm": 18.052600860595703,
|
| 2128 |
-
"learning_rate": 8.051972677494753e-06,
|
| 2129 |
-
"loss": 0.1594,
|
| 2130 |
-
"step": 7575
|
| 2131 |
-
},
|
| 2132 |
-
{
|
| 2133 |
-
"epoch": 0.7061228282077487,
|
| 2134 |
-
"grad_norm": 0.3958927094936371,
|
| 2135 |
-
"learning_rate": 7.935253066480726e-06,
|
| 2136 |
-
"loss": 0.1112,
|
| 2137 |
-
"step": 7600
|
| 2138 |
-
},
|
| 2139 |
-
{
|
| 2140 |
-
"epoch": 0.7084456006689585,
|
| 2141 |
-
"grad_norm": 0.22095303237438202,
|
| 2142 |
-
"learning_rate": 7.819175890730548e-06,
|
| 2143 |
-
"loss": 0.1599,
|
| 2144 |
-
"step": 7625
|
| 2145 |
-
},
|
| 2146 |
-
{
|
| 2147 |
-
"epoch": 0.7107683731301682,
|
| 2148 |
-
"grad_norm": 0.03430924192070961,
|
| 2149 |
-
"learning_rate": 7.703747331233468e-06,
|
| 2150 |
-
"loss": 0.0925,
|
| 2151 |
-
"step": 7650
|
| 2152 |
-
},
|
| 2153 |
-
{
|
| 2154 |
-
"epoch": 0.7130911455913779,
|
| 2155 |
-
"grad_norm": 0.022623928263783455,
|
| 2156 |
-
"learning_rate": 7.588973534440575e-06,
|
| 2157 |
-
"loss": 0.1126,
|
| 2158 |
-
"step": 7675
|
| 2159 |
-
},
|
| 2160 |
-
{
|
| 2161 |
-
"epoch": 0.7154139180525876,
|
| 2162 |
-
"grad_norm": 0.41440054774284363,
|
| 2163 |
-
"learning_rate": 7.47486061193754e-06,
|
| 2164 |
-
"loss": 0.1558,
|
| 2165 |
-
"step": 7700
|
| 2166 |
-
},
|
| 2167 |
-
{
|
| 2168 |
-
"epoch": 0.7177366905137973,
|
| 2169 |
-
"grad_norm": 0.09115582704544067,
|
| 2170 |
-
"learning_rate": 7.36141464011916e-06,
|
| 2171 |
-
"loss": 0.1127,
|
| 2172 |
-
"step": 7725
|
| 2173 |
-
},
|
| 2174 |
-
{
|
| 2175 |
-
"epoch": 0.720059462975007,
|
| 2176 |
-
"grad_norm": 0.9181798696517944,
|
| 2177 |
-
"learning_rate": 7.248641659865805e-06,
|
| 2178 |
-
"loss": 0.1852,
|
| 2179 |
-
"step": 7750
|
| 2180 |
-
},
|
| 2181 |
-
{
|
| 2182 |
-
"epoch": 0.7223822354362167,
|
| 2183 |
-
"grad_norm": 0.46478185057640076,
|
| 2184 |
-
"learning_rate": 7.136547676221719e-06,
|
| 2185 |
-
"loss": 0.1609,
|
| 2186 |
-
"step": 7775
|
| 2187 |
-
},
|
| 2188 |
-
{
|
| 2189 |
-
"epoch": 0.7247050078974264,
|
| 2190 |
-
"grad_norm": 0.1365162432193756,
|
| 2191 |
-
"learning_rate": 7.025138658075308e-06,
|
| 2192 |
-
"loss": 0.1068,
|
| 2193 |
-
"step": 7800
|
| 2194 |
-
},
|
| 2195 |
-
{
|
| 2196 |
-
"epoch": 0.727027780358636,
|
| 2197 |
-
"grad_norm": 0.03270541876554489,
|
| 2198 |
-
"learning_rate": 6.9144205378412686e-06,
|
| 2199 |
-
"loss": 0.0687,
|
| 2200 |
-
"step": 7825
|
| 2201 |
-
},
|
| 2202 |
-
{
|
| 2203 |
-
"epoch": 0.7293505528198457,
|
| 2204 |
-
"grad_norm": 0.0757010206580162,
|
| 2205 |
-
"learning_rate": 6.804399211144688e-06,
|
| 2206 |
-
"loss": 0.0987,
|
| 2207 |
-
"step": 7850
|
| 2208 |
-
},
|
| 2209 |
-
{
|
| 2210 |
-
"epoch": 0.7316733252810554,
|
| 2211 |
-
"grad_norm": 0.15775972604751587,
|
| 2212 |
-
"learning_rate": 6.695080536507124e-06,
|
| 2213 |
-
"loss": 0.0559,
|
| 2214 |
-
"step": 7875
|
| 2215 |
-
},
|
| 2216 |
-
{
|
| 2217 |
-
"epoch": 0.7339960977422652,
|
| 2218 |
-
"grad_norm": 9.302044868469238,
|
| 2219 |
-
"learning_rate": 6.586470335034647e-06,
|
| 2220 |
-
"loss": 0.2095,
|
| 2221 |
-
"step": 7900
|
| 2222 |
-
},
|
| 2223 |
-
{
|
| 2224 |
-
"epoch": 0.7363188702034749,
|
| 2225 |
-
"grad_norm": 0.04565757140517235,
|
| 2226 |
-
"learning_rate": 6.47857439010787e-06,
|
| 2227 |
-
"loss": 0.1644,
|
| 2228 |
-
"step": 7925
|
| 2229 |
-
},
|
| 2230 |
-
{
|
| 2231 |
-
"epoch": 0.7386416426646846,
|
| 2232 |
-
"grad_norm": 2.6939070224761963,
|
| 2233 |
-
"learning_rate": 6.3713984470739795e-06,
|
| 2234 |
-
"loss": 0.0546,
|
| 2235 |
-
"step": 7950
|
| 2236 |
-
},
|
| 2237 |
-
{
|
| 2238 |
-
"epoch": 0.7409644151258943,
|
| 2239 |
-
"grad_norm": 0.3147173523902893,
|
| 2240 |
-
"learning_rate": 6.264948212940798e-06,
|
| 2241 |
-
"loss": 0.0747,
|
| 2242 |
-
"step": 7975
|
| 2243 |
-
},
|
| 2244 |
-
{
|
| 2245 |
-
"epoch": 0.743287187587104,
|
| 2246 |
-
"grad_norm": 14.228339195251465,
|
| 2247 |
-
"learning_rate": 6.159229356072916e-06,
|
| 2248 |
-
"loss": 0.0715,
|
| 2249 |
-
"step": 8000
|
| 2250 |
-
},
|
| 2251 |
-
{
|
| 2252 |
-
"epoch": 0.7456099600483137,
|
| 2253 |
-
"grad_norm": 0.07007303833961487,
|
| 2254 |
-
"learning_rate": 6.054247505889839e-06,
|
| 2255 |
-
"loss": 0.0914,
|
| 2256 |
-
"step": 8025
|
| 2257 |
-
},
|
| 2258 |
-
{
|
| 2259 |
-
"epoch": 0.7479327325095234,
|
| 2260 |
-
"grad_norm": 0.06615108996629715,
|
| 2261 |
-
"learning_rate": 5.950008252566237e-06,
|
| 2262 |
-
"loss": 0.1166,
|
| 2263 |
-
"step": 8050
|
| 2264 |
-
},
|
| 2265 |
-
{
|
| 2266 |
-
"epoch": 0.750255504970733,
|
| 2267 |
-
"grad_norm": 2.1876168251037598,
|
| 2268 |
-
"learning_rate": 5.846517146734252e-06,
|
| 2269 |
-
"loss": 0.0528,
|
| 2270 |
-
"step": 8075
|
| 2271 |
-
},
|
| 2272 |
-
{
|
| 2273 |
-
"epoch": 0.7525782774319427,
|
| 2274 |
-
"grad_norm": 0.08167368173599243,
|
| 2275 |
-
"learning_rate": 5.74377969918795e-06,
|
| 2276 |
-
"loss": 0.1514,
|
| 2277 |
-
"step": 8100
|
| 2278 |
-
},
|
| 2279 |
-
{
|
| 2280 |
-
"epoch": 0.7549010498931524,
|
| 2281 |
-
"grad_norm": 0.30077704787254333,
|
| 2282 |
-
"learning_rate": 5.641801380589888e-06,
|
| 2283 |
-
"loss": 0.0796,
|
| 2284 |
-
"step": 8125
|
| 2285 |
-
},
|
| 2286 |
-
{
|
| 2287 |
-
"epoch": 0.7572238223543621,
|
| 2288 |
-
"grad_norm": 0.038170505315065384,
|
| 2289 |
-
"learning_rate": 5.540587621179787e-06,
|
| 2290 |
-
"loss": 0.1391,
|
| 2291 |
-
"step": 8150
|
| 2292 |
-
},
|
| 2293 |
-
{
|
| 2294 |
-
"epoch": 0.7595465948155719,
|
| 2295 |
-
"grad_norm": 0.6518100500106812,
|
| 2296 |
-
"learning_rate": 5.4401438104853856e-06,
|
| 2297 |
-
"loss": 0.1197,
|
| 2298 |
-
"step": 8175
|
| 2299 |
-
},
|
| 2300 |
-
{
|
| 2301 |
-
"epoch": 0.7618693672767816,
|
| 2302 |
-
"grad_norm": 1.5173622369766235,
|
| 2303 |
-
"learning_rate": 5.340475297035444e-06,
|
| 2304 |
-
"loss": 0.057,
|
| 2305 |
-
"step": 8200
|
| 2306 |
-
},
|
| 2307 |
-
{
|
| 2308 |
-
"epoch": 0.7641921397379913,
|
| 2309 |
-
"grad_norm": 0.33258476853370667,
|
| 2310 |
-
"learning_rate": 5.2415873880749534e-06,
|
| 2311 |
-
"loss": 0.1659,
|
| 2312 |
-
"step": 8225
|
| 2313 |
-
},
|
| 2314 |
-
{
|
| 2315 |
-
"epoch": 0.766514912199201,
|
| 2316 |
-
"grad_norm": 0.1752271056175232,
|
| 2317 |
-
"learning_rate": 5.143485349282529e-06,
|
| 2318 |
-
"loss": 0.0992,
|
| 2319 |
-
"step": 8250
|
| 2320 |
-
},
|
| 2321 |
-
{
|
| 2322 |
-
"epoch": 0.7688376846604107,
|
| 2323 |
-
"grad_norm": 0.04892320558428764,
|
| 2324 |
-
"learning_rate": 5.0461744044900145e-06,
|
| 2325 |
-
"loss": 0.0912,
|
| 2326 |
-
"step": 8275
|
| 2327 |
-
},
|
| 2328 |
-
{
|
| 2329 |
-
"epoch": 0.7711604571216204,
|
| 2330 |
-
"grad_norm": 0.06182296574115753,
|
| 2331 |
-
"learning_rate": 4.949659735404306e-06,
|
| 2332 |
-
"loss": 0.0315,
|
| 2333 |
-
"step": 8300
|
| 2334 |
-
},
|
| 2335 |
-
{
|
| 2336 |
-
"epoch": 0.7734832295828301,
|
| 2337 |
-
"grad_norm": 0.057482581585645676,
|
| 2338 |
-
"learning_rate": 4.853946481331457e-06,
|
| 2339 |
-
"loss": 0.0648,
|
| 2340 |
-
"step": 8325
|
| 2341 |
-
},
|
| 2342 |
-
{
|
| 2343 |
-
"epoch": 0.7758060020440397,
|
| 2344 |
-
"grad_norm": 0.21633560955524445,
|
| 2345 |
-
"learning_rate": 4.7590397389030045e-06,
|
| 2346 |
-
"loss": 0.083,
|
| 2347 |
-
"step": 8350
|
| 2348 |
-
},
|
| 2349 |
-
{
|
| 2350 |
-
"epoch": 0.7781287745052494,
|
| 2351 |
-
"grad_norm": 1.5439313650131226,
|
| 2352 |
-
"learning_rate": 4.664944561804559e-06,
|
| 2353 |
-
"loss": 0.0333,
|
| 2354 |
-
"step": 8375
|
| 2355 |
-
},
|
| 2356 |
-
{
|
| 2357 |
-
"epoch": 0.7804515469664591,
|
| 2358 |
-
"grad_norm": 0.3675384521484375,
|
| 2359 |
-
"learning_rate": 4.571665960506743e-06,
|
| 2360 |
-
"loss": 0.0554,
|
| 2361 |
-
"step": 8400
|
| 2362 |
-
},
|
| 2363 |
-
{
|
| 2364 |
-
"epoch": 0.7827743194276688,
|
| 2365 |
-
"grad_norm": 60.508705139160156,
|
| 2366 |
-
"learning_rate": 4.479208901998338e-06,
|
| 2367 |
-
"loss": 0.1488,
|
| 2368 |
-
"step": 8425
|
| 2369 |
-
},
|
| 2370 |
-
{
|
| 2371 |
-
"epoch": 0.7850970918888786,
|
| 2372 |
-
"grad_norm": 0.09332936257123947,
|
| 2373 |
-
"learning_rate": 4.387578309521845e-06,
|
| 2374 |
-
"loss": 0.1348,
|
| 2375 |
-
"step": 8450
|
| 2376 |
-
},
|
| 2377 |
-
{
|
| 2378 |
-
"epoch": 0.7874198643500883,
|
| 2379 |
-
"grad_norm": 0.15118038654327393,
|
| 2380 |
-
"learning_rate": 4.2967790623113004e-06,
|
| 2381 |
-
"loss": 0.0818,
|
| 2382 |
-
"step": 8475
|
| 2383 |
-
},
|
| 2384 |
-
{
|
| 2385 |
-
"epoch": 0.789742636811298,
|
| 2386 |
-
"grad_norm": 7.807275772094727,
|
| 2387 |
-
"learning_rate": 4.206815995332465e-06,
|
| 2388 |
-
"loss": 0.1108,
|
| 2389 |
-
"step": 8500
|
| 2390 |
-
},
|
| 2391 |
-
{
|
| 2392 |
-
"epoch": 0.7920654092725077,
|
| 2393 |
-
"grad_norm": 17.519529342651367,
|
| 2394 |
-
"learning_rate": 4.117693899025363e-06,
|
| 2395 |
-
"loss": 0.0786,
|
| 2396 |
-
"step": 8525
|
| 2397 |
-
},
|
| 2398 |
-
{
|
| 2399 |
-
"epoch": 0.7943881817337174,
|
| 2400 |
-
"grad_norm": 24.23104476928711,
|
| 2401 |
-
"learning_rate": 4.029417519049215e-06,
|
| 2402 |
-
"loss": 0.1875,
|
| 2403 |
-
"step": 8550
|
| 2404 |
-
},
|
| 2405 |
-
{
|
| 2406 |
-
"epoch": 0.7967109541949271,
|
| 2407 |
-
"grad_norm": 32.55892562866211,
|
| 2408 |
-
"learning_rate": 3.941991556029714e-06,
|
| 2409 |
-
"loss": 0.1477,
|
| 2410 |
-
"step": 8575
|
| 2411 |
-
},
|
| 2412 |
-
{
|
| 2413 |
-
"epoch": 0.7990337266561368,
|
| 2414 |
-
"grad_norm": 0.068290114402771,
|
| 2415 |
-
"learning_rate": 3.855420665308745e-06,
|
| 2416 |
-
"loss": 0.0466,
|
| 2417 |
-
"step": 8600
|
| 2418 |
-
},
|
| 2419 |
-
{
|
| 2420 |
-
"epoch": 0.8013564991173464,
|
| 2421 |
-
"grad_norm": 0.11146409064531326,
|
| 2422 |
-
"learning_rate": 3.769709456696462e-06,
|
| 2423 |
-
"loss": 0.0914,
|
| 2424 |
-
"step": 8625
|
| 2425 |
-
},
|
| 2426 |
-
{
|
| 2427 |
-
"epoch": 0.8036792715785561,
|
| 2428 |
-
"grad_norm": 1.0052744150161743,
|
| 2429 |
-
"learning_rate": 3.6848624942258425e-06,
|
| 2430 |
-
"loss": 0.0578,
|
| 2431 |
-
"step": 8650
|
| 2432 |
-
},
|
| 2433 |
-
{
|
| 2434 |
-
"epoch": 0.8060020440397658,
|
| 2435 |
-
"grad_norm": 0.023286884650588036,
|
| 2436 |
-
"learning_rate": 3.6008842959096634e-06,
|
| 2437 |
-
"loss": 0.0408,
|
| 2438 |
-
"step": 8675
|
| 2439 |
-
},
|
| 2440 |
-
{
|
| 2441 |
-
"epoch": 0.8083248165009755,
|
| 2442 |
-
"grad_norm": 0.5776505470275879,
|
| 2443 |
-
"learning_rate": 3.517779333499909e-06,
|
| 2444 |
-
"loss": 0.0374,
|
| 2445 |
-
"step": 8700
|
| 2446 |
-
},
|
| 2447 |
-
{
|
| 2448 |
-
"epoch": 0.8106475889621853,
|
| 2449 |
-
"grad_norm": 0.04028211906552315,
|
| 2450 |
-
"learning_rate": 3.435552032249643e-06,
|
| 2451 |
-
"loss": 0.1663,
|
| 2452 |
-
"step": 8725
|
| 2453 |
-
},
|
| 2454 |
-
{
|
| 2455 |
-
"epoch": 0.812970361423395,
|
| 2456 |
-
"grad_norm": 19.693021774291992,
|
| 2457 |
-
"learning_rate": 3.3542067706773996e-06,
|
| 2458 |
-
"loss": 0.1373,
|
| 2459 |
-
"step": 8750
|
| 2460 |
-
},
|
| 2461 |
-
{
|
| 2462 |
-
"epoch": 0.8152931338846047,
|
| 2463 |
-
"grad_norm": 0.016215180978178978,
|
| 2464 |
-
"learning_rate": 3.273747880334004e-06,
|
| 2465 |
-
"loss": 0.0966,
|
| 2466 |
-
"step": 8775
|
| 2467 |
-
},
|
| 2468 |
-
{
|
| 2469 |
-
"epoch": 0.8176159063458144,
|
| 2470 |
-
"grad_norm": 0.2978752553462982,
|
| 2471 |
-
"learning_rate": 3.1941796455719463e-06,
|
| 2472 |
-
"loss": 0.1219,
|
| 2473 |
-
"step": 8800
|
| 2474 |
-
},
|
| 2475 |
-
{
|
| 2476 |
-
"epoch": 0.8199386788070241,
|
| 2477 |
-
"grad_norm": 0.12780235707759857,
|
| 2478 |
-
"learning_rate": 3.115506303317206e-06,
|
| 2479 |
-
"loss": 0.1237,
|
| 2480 |
-
"step": 8825
|
| 2481 |
-
},
|
| 2482 |
-
{
|
| 2483 |
-
"epoch": 0.8222614512682338,
|
| 2484 |
-
"grad_norm": 0.01512450072914362,
|
| 2485 |
-
"learning_rate": 3.0377320428436865e-06,
|
| 2486 |
-
"loss": 0.0794,
|
| 2487 |
-
"step": 8850
|
| 2488 |
-
},
|
| 2489 |
-
{
|
| 2490 |
-
"epoch": 0.8245842237294435,
|
| 2491 |
-
"grad_norm": 0.16190892457962036,
|
| 2492 |
-
"learning_rate": 2.9608610055501087e-06,
|
| 2493 |
-
"loss": 0.0756,
|
| 2494 |
-
"step": 8875
|
| 2495 |
-
},
|
| 2496 |
-
{
|
| 2497 |
-
"epoch": 0.8269069961906531,
|
| 2498 |
-
"grad_norm": 0.018665891140699387,
|
| 2499 |
-
"learning_rate": 2.884897284739505e-06,
|
| 2500 |
-
"loss": 0.0833,
|
| 2501 |
-
"step": 8900
|
| 2502 |
-
},
|
| 2503 |
-
{
|
| 2504 |
-
"epoch": 0.8292297686518628,
|
| 2505 |
-
"grad_norm": 0.08344435691833496,
|
| 2506 |
-
"learning_rate": 2.809844925401235e-06,
|
| 2507 |
-
"loss": 0.1114,
|
| 2508 |
-
"step": 8925
|
| 2509 |
-
},
|
| 2510 |
-
{
|
| 2511 |
-
"epoch": 0.8315525411130725,
|
| 2512 |
-
"grad_norm": 0.2573084533214569,
|
| 2513 |
-
"learning_rate": 2.735707923995603e-06,
|
| 2514 |
-
"loss": 0.0563,
|
| 2515 |
-
"step": 8950
|
| 2516 |
-
},
|
| 2517 |
-
{
|
| 2518 |
-
"epoch": 0.8338753135742822,
|
| 2519 |
-
"grad_norm": 0.07283484190702438,
|
| 2520 |
-
"learning_rate": 2.6624902282410635e-06,
|
| 2521 |
-
"loss": 0.1347,
|
| 2522 |
-
"step": 8975
|
| 2523 |
-
},
|
| 2524 |
-
{
|
| 2525 |
-
"epoch": 0.836198086035492,
|
| 2526 |
-
"grad_norm": 15.993093490600586,
|
| 2527 |
-
"learning_rate": 2.5901957369039934e-06,
|
| 2528 |
-
"loss": 0.1341,
|
| 2529 |
-
"step": 9000
|
| 2530 |
-
},
|
| 2531 |
-
{
|
| 2532 |
-
"epoch": 0.8385208584967017,
|
| 2533 |
-
"grad_norm": 35.06781768798828,
|
| 2534 |
-
"learning_rate": 2.518828299591096e-06,
|
| 2535 |
-
"loss": 0.0813,
|
| 2536 |
-
"step": 9025
|
| 2537 |
-
},
|
| 2538 |
-
{
|
| 2539 |
-
"epoch": 0.8408436309579114,
|
| 2540 |
-
"grad_norm": 21.37590789794922,
|
| 2541 |
-
"learning_rate": 2.448391716544396e-06,
|
| 2542 |
-
"loss": 0.053,
|
| 2543 |
-
"step": 9050
|
| 2544 |
-
},
|
| 2545 |
-
{
|
| 2546 |
-
"epoch": 0.8431664034191211,
|
| 2547 |
-
"grad_norm": 0.06716252118349075,
|
| 2548 |
-
"learning_rate": 2.3788897384389075e-06,
|
| 2549 |
-
"loss": 0.1405,
|
| 2550 |
-
"step": 9075
|
| 2551 |
-
},
|
| 2552 |
-
{
|
| 2553 |
-
"epoch": 0.8454891758803308,
|
| 2554 |
-
"grad_norm": 1.1064941883087158,
|
| 2555 |
-
"learning_rate": 2.310326066182893e-06,
|
| 2556 |
-
"loss": 0.1201,
|
| 2557 |
-
"step": 9100
|
| 2558 |
-
},
|
| 2559 |
-
{
|
| 2560 |
-
"epoch": 0.8478119483415405,
|
| 2561 |
-
"grad_norm": 0.26063647866249084,
|
| 2562 |
-
"learning_rate": 2.2427043507208124e-06,
|
| 2563 |
-
"loss": 0.0925,
|
| 2564 |
-
"step": 9125
|
| 2565 |
-
},
|
| 2566 |
-
{
|
| 2567 |
-
"epoch": 0.8501347208027502,
|
| 2568 |
-
"grad_norm": 0.058635413646698,
|
| 2569 |
-
"learning_rate": 2.176028192838884e-06,
|
| 2570 |
-
"loss": 0.063,
|
| 2571 |
-
"step": 9150
|
| 2572 |
-
},
|
| 2573 |
-
{
|
| 2574 |
-
"epoch": 0.8524574932639598,
|
| 2575 |
-
"grad_norm": 0.30567261576652527,
|
| 2576 |
-
"learning_rate": 2.1103011429733832e-06,
|
| 2577 |
-
"loss": 0.1673,
|
| 2578 |
-
"step": 9175
|
| 2579 |
-
},
|
| 2580 |
-
{
|
| 2581 |
-
"epoch": 0.8547802657251695,
|
| 2582 |
-
"grad_norm": 0.08589446544647217,
|
| 2583 |
-
"learning_rate": 2.045526701021556e-06,
|
| 2584 |
-
"loss": 0.0452,
|
| 2585 |
-
"step": 9200
|
| 2586 |
-
},
|
| 2587 |
-
{
|
| 2588 |
-
"epoch": 0.8571030381863792,
|
| 2589 |
-
"grad_norm": 540.3936767578125,
|
| 2590 |
-
"learning_rate": 1.981708316155271e-06,
|
| 2591 |
-
"loss": 0.0466,
|
| 2592 |
-
"step": 9225
|
| 2593 |
-
},
|
| 2594 |
-
{
|
| 2595 |
-
"epoch": 0.8594258106475889,
|
| 2596 |
-
"grad_norm": 7.8817949295043945,
|
| 2597 |
-
"learning_rate": 1.918849386637349e-06,
|
| 2598 |
-
"loss": 0.131,
|
| 2599 |
-
"step": 9250
|
| 2600 |
-
},
|
| 2601 |
-
{
|
| 2602 |
-
"epoch": 0.8617485831087986,
|
| 2603 |
-
"grad_norm": 0.2560771703720093,
|
| 2604 |
-
"learning_rate": 1.8569532596406014e-06,
|
| 2605 |
-
"loss": 0.1428,
|
| 2606 |
-
"step": 9275
|
| 2607 |
-
},
|
| 2608 |
-
{
|
| 2609 |
-
"epoch": 0.8640713555700084,
|
| 2610 |
-
"grad_norm": 0.07730316370725632,
|
| 2611 |
-
"learning_rate": 1.7960232310696125e-06,
|
| 2612 |
-
"loss": 0.0884,
|
| 2613 |
-
"step": 9300
|
| 2614 |
-
},
|
| 2615 |
-
{
|
| 2616 |
-
"epoch": 0.8663941280312181,
|
| 2617 |
-
"grad_norm": 0.33886778354644775,
|
| 2618 |
-
"learning_rate": 1.736062545385222e-06,
|
| 2619 |
-
"loss": 0.1213,
|
| 2620 |
-
"step": 9325
|
| 2621 |
-
},
|
| 2622 |
-
{
|
| 2623 |
-
"epoch": 0.8687169004924278,
|
| 2624 |
-
"grad_norm": 0.06674880534410477,
|
| 2625 |
-
"learning_rate": 1.6770743954317704e-06,
|
| 2626 |
-
"loss": 0.0947,
|
| 2627 |
-
"step": 9350
|
| 2628 |
-
},
|
| 2629 |
-
{
|
| 2630 |
-
"epoch": 0.8710396729536375,
|
| 2631 |
-
"grad_norm": 0.248457133769989,
|
| 2632 |
-
"learning_rate": 1.6190619222670689e-06,
|
| 2633 |
-
"loss": 0.1217,
|
| 2634 |
-
"step": 9375
|
| 2635 |
-
},
|
| 2636 |
-
{
|
| 2637 |
-
"epoch": 0.8733624454148472,
|
| 2638 |
-
"grad_norm": 0.05154260993003845,
|
| 2639 |
-
"learning_rate": 1.5620282149951639e-06,
|
| 2640 |
-
"loss": 0.0452,
|
| 2641 |
-
"step": 9400
|
| 2642 |
-
},
|
| 2643 |
-
{
|
| 2644 |
-
"epoch": 0.8756852178760569,
|
| 2645 |
-
"grad_norm": 0.050282254815101624,
|
| 2646 |
-
"learning_rate": 1.5059763106018244e-06,
|
| 2647 |
-
"loss": 0.1506,
|
| 2648 |
-
"step": 9425
|
| 2649 |
-
},
|
| 2650 |
-
{
|
| 2651 |
-
"epoch": 0.8780079903372665,
|
| 2652 |
-
"grad_norm": 0.09417497366666794,
|
| 2653 |
-
"learning_rate": 1.4509091937928444e-06,
|
| 2654 |
-
"loss": 0.1395,
|
| 2655 |
-
"step": 9450
|
| 2656 |
-
},
|
| 2657 |
-
{
|
| 2658 |
-
"epoch": 0.8803307627984762,
|
| 2659 |
-
"grad_norm": 32.32343673706055,
|
| 2660 |
-
"learning_rate": 1.396829796835082e-06,
|
| 2661 |
-
"loss": 0.1297,
|
| 2662 |
-
"step": 9475
|
| 2663 |
-
},
|
| 2664 |
-
{
|
| 2665 |
-
"epoch": 0.8826535352596859,
|
| 2666 |
-
"grad_norm": 21.86054039001465,
|
| 2667 |
-
"learning_rate": 1.3437409994003515e-06,
|
| 2668 |
-
"loss": 0.1308,
|
| 2669 |
-
"step": 9500
|
| 2670 |
-
},
|
| 2671 |
-
{
|
| 2672 |
-
"epoch": 0.8849763077208956,
|
| 2673 |
-
"grad_norm": 0.017470579594373703,
|
| 2674 |
-
"learning_rate": 1.2916456284120705e-06,
|
| 2675 |
-
"loss": 0.1107,
|
| 2676 |
-
"step": 9525
|
| 2677 |
-
},
|
| 2678 |
-
{
|
| 2679 |
-
"epoch": 0.8872990801821053,
|
| 2680 |
-
"grad_norm": 25.04646110534668,
|
| 2681 |
-
"learning_rate": 1.2405464578947246e-06,
|
| 2682 |
-
"loss": 0.0748,
|
| 2683 |
-
"step": 9550
|
| 2684 |
-
},
|
| 2685 |
-
{
|
| 2686 |
-
"epoch": 0.8896218526433151,
|
| 2687 |
-
"grad_norm": 0.06748735904693604,
|
| 2688 |
-
"learning_rate": 1.1904462088261525e-06,
|
| 2689 |
-
"loss": 0.157,
|
| 2690 |
-
"step": 9575
|
| 2691 |
-
},
|
| 2692 |
-
{
|
| 2693 |
-
"epoch": 0.8919446251045248,
|
| 2694 |
-
"grad_norm": 37.16495132446289,
|
| 2695 |
-
"learning_rate": 1.1413475489926685e-06,
|
| 2696 |
-
"loss": 0.1406,
|
| 2697 |
-
"step": 9600
|
| 2698 |
-
},
|
| 2699 |
-
{
|
| 2700 |
-
"epoch": 0.8942673975657345,
|
| 2701 |
-
"grad_norm": 0.26858797669410706,
|
| 2702 |
-
"learning_rate": 1.0932530928469974e-06,
|
| 2703 |
-
"loss": 0.151,
|
| 2704 |
-
"step": 9625
|
| 2705 |
-
},
|
| 2706 |
-
{
|
| 2707 |
-
"epoch": 0.8965901700269442,
|
| 2708 |
-
"grad_norm": 0.05599837377667427,
|
| 2709 |
-
"learning_rate": 1.0461654013690614e-06,
|
| 2710 |
-
"loss": 0.0589,
|
| 2711 |
-
"step": 9650
|
| 2712 |
-
},
|
| 2713 |
-
{
|
| 2714 |
-
"epoch": 0.8989129424881539,
|
| 2715 |
-
"grad_norm": 0.1042524203658104,
|
| 2716 |
-
"learning_rate": 1.0000869819295999e-06,
|
| 2717 |
-
"loss": 0.1436,
|
| 2718 |
-
"step": 9675
|
| 2719 |
-
},
|
| 2720 |
-
{
|
| 2721 |
-
"epoch": 0.9012357149493636,
|
| 2722 |
-
"grad_norm": 0.2055070549249649,
|
| 2723 |
-
"learning_rate": 9.550202881566673e-07,
|
| 2724 |
-
"loss": 0.0815,
|
| 2725 |
-
"step": 9700
|
| 2726 |
-
},
|
| 2727 |
-
{
|
| 2728 |
-
"epoch": 0.9035584874105732,
|
| 2729 |
-
"grad_norm": 0.31944283843040466,
|
| 2730 |
-
"learning_rate": 9.109677198049805e-07,
|
| 2731 |
-
"loss": 0.0181,
|
| 2732 |
-
"step": 9725
|
| 2733 |
-
},
|
| 2734 |
-
{
|
| 2735 |
-
"epoch": 0.9058812598717829,
|
| 2736 |
-
"grad_norm": 19.389678955078125,
|
| 2737 |
-
"learning_rate": 8.679316226281198e-07,
|
| 2738 |
-
"loss": 0.1066,
|
| 2739 |
-
"step": 9750
|
| 2740 |
-
},
|
| 2741 |
-
{
|
| 2742 |
-
"epoch": 0.9082040323329926,
|
| 2743 |
-
"grad_norm": 0.28735801577568054,
|
| 2744 |
-
"learning_rate": 8.259142882536375e-07,
|
| 2745 |
-
"loss": 0.0823,
|
| 2746 |
-
"step": 9775
|
| 2747 |
-
},
|
| 2748 |
-
{
|
| 2749 |
-
"epoch": 0.9105268047942023,
|
| 2750 |
-
"grad_norm": 1.9290257692337036,
|
| 2751 |
-
"learning_rate": 7.849179540610174e-07,
|
| 2752 |
-
"loss": 0.0359,
|
| 2753 |
-
"step": 9800
|
| 2754 |
-
},
|
| 2755 |
-
{
|
| 2756 |
-
"epoch": 0.912849577255412,
|
| 2757 |
-
"grad_norm": 0.04476006329059601,
|
| 2758 |
-
"learning_rate": 7.449448030625461e-07,
|
| 2759 |
-
"loss": 0.1055,
|
| 2760 |
-
"step": 9825
|
| 2761 |
-
},
|
| 2762 |
-
{
|
| 2763 |
-
"epoch": 0.9151723497166218,
|
| 2764 |
-
"grad_norm": 0.04544592648744583,
|
| 2765 |
-
"learning_rate": 7.059969637870634e-07,
|
| 2766 |
-
"loss": 0.0734,
|
| 2767 |
-
"step": 9850
|
| 2768 |
-
},
|
| 2769 |
-
{
|
| 2770 |
-
"epoch": 0.9174951221778315,
|
| 2771 |
-
"grad_norm": 0.20919212698936462,
|
| 2772 |
-
"learning_rate": 6.680765101666264e-07,
|
| 2773 |
-
"loss": 0.1067,
|
| 2774 |
-
"step": 9875
|
| 2775 |
-
},
|
| 2776 |
-
{
|
| 2777 |
-
"epoch": 0.9198178946390412,
|
| 2778 |
-
"grad_norm": 0.10152643918991089,
|
| 2779 |
-
"learning_rate": 6.311854614260671e-07,
|
| 2780 |
-
"loss": 0.087,
|
| 2781 |
-
"step": 9900
|
| 2782 |
-
},
|
| 2783 |
-
{
|
| 2784 |
-
"epoch": 0.9221406671002509,
|
| 2785 |
-
"grad_norm": 0.4486711323261261,
|
| 2786 |
-
"learning_rate": 5.953257819754798e-07,
|
| 2787 |
-
"loss": 0.0729,
|
| 2788 |
-
"step": 9925
|
| 2789 |
-
},
|
| 2790 |
-
{
|
| 2791 |
-
"epoch": 0.9244634395614606,
|
| 2792 |
-
"grad_norm": 12.16341781616211,
|
| 2793 |
-
"learning_rate": 5.604993813056081e-07,
|
| 2794 |
-
"loss": 0.1004,
|
| 2795 |
-
"step": 9950
|
| 2796 |
-
},
|
| 2797 |
-
{
|
| 2798 |
-
"epoch": 0.9267862120226703,
|
| 2799 |
-
"grad_norm": 72.28809356689453,
|
| 2800 |
-
"learning_rate": 5.267081138861784e-07,
|
| 2801 |
-
"loss": 0.1114,
|
| 2802 |
-
"step": 9975
|
| 2803 |
-
},
|
| 2804 |
-
{
|
| 2805 |
-
"epoch": 0.9291089844838799,
|
| 2806 |
-
"grad_norm": 14.374608039855957,
|
| 2807 |
-
"learning_rate": 4.939537790671401e-07,
|
| 2808 |
-
"loss": 0.0967,
|
| 2809 |
-
"step": 10000
|
| 2810 |
-
},
|
| 2811 |
-
{
|
| 2812 |
-
"epoch": 0.9314317569450896,
|
| 2813 |
-
"grad_norm": 26.21600341796875,
|
| 2814 |
-
"learning_rate": 4.6223812098285415e-07,
|
| 2815 |
-
"loss": 0.1591,
|
| 2816 |
-
"step": 10025
|
| 2817 |
-
},
|
| 2818 |
-
{
|
| 2819 |
-
"epoch": 0.9337545294062993,
|
| 2820 |
-
"grad_norm": 0.08388189971446991,
|
| 2821 |
-
"learning_rate": 4.31562828459231e-07,
|
| 2822 |
-
"loss": 0.1493,
|
| 2823 |
-
"step": 10050
|
| 2824 |
-
},
|
| 2825 |
-
{
|
| 2826 |
-
"epoch": 0.936077301867509,
|
| 2827 |
-
"grad_norm": 74.03582000732422,
|
| 2828 |
-
"learning_rate": 4.019295349237906e-07,
|
| 2829 |
-
"loss": 0.1424,
|
| 2830 |
-
"step": 10075
|
| 2831 |
-
},
|
| 2832 |
-
{
|
| 2833 |
-
"epoch": 0.9384000743287187,
|
| 2834 |
-
"grad_norm": 11.0938081741333,
|
| 2835 |
-
"learning_rate": 3.7333981831868313e-07,
|
| 2836 |
-
"loss": 0.0904,
|
| 2837 |
-
"step": 10100
|
| 2838 |
-
},
|
| 2839 |
-
{
|
| 2840 |
-
"epoch": 0.9407228467899285,
|
| 2841 |
-
"grad_norm": 0.054299451410770416,
|
| 2842 |
-
"learning_rate": 3.4579520101666987e-07,
|
| 2843 |
-
"loss": 0.0509,
|
| 2844 |
-
"step": 10125
|
| 2845 |
-
},
|
| 2846 |
-
{
|
| 2847 |
-
"epoch": 0.9430456192511382,
|
| 2848 |
-
"grad_norm": 0.11387737840414047,
|
| 2849 |
-
"learning_rate": 3.192971497400632e-07,
|
| 2850 |
-
"loss": 0.1397,
|
| 2851 |
-
"step": 10150
|
| 2852 |
-
},
|
| 2853 |
-
{
|
| 2854 |
-
"epoch": 0.9453683917123479,
|
| 2855 |
-
"grad_norm": 0.03952685371041298,
|
| 2856 |
-
"learning_rate": 2.9384707548261616e-07,
|
| 2857 |
-
"loss": 0.077,
|
| 2858 |
-
"step": 10175
|
| 2859 |
-
},
|
| 2860 |
-
{
|
| 2861 |
-
"epoch": 0.9476911641735576,
|
| 2862 |
-
"grad_norm": 0.9854119420051575,
|
| 2863 |
-
"learning_rate": 2.694463334343955e-07,
|
| 2864 |
-
"loss": 0.1448,
|
| 2865 |
-
"step": 10200
|
| 2866 |
-
},
|
| 2867 |
-
{
|
| 2868 |
-
"epoch": 0.9500139366347673,
|
| 2869 |
-
"grad_norm": 0.05075085535645485,
|
| 2870 |
-
"learning_rate": 2.4609622290961086e-07,
|
| 2871 |
-
"loss": 0.1411,
|
| 2872 |
-
"step": 10225
|
| 2873 |
-
},
|
| 2874 |
-
{
|
| 2875 |
-
"epoch": 0.952336709095977,
|
| 2876 |
-
"grad_norm": 10.095162391662598,
|
| 2877 |
-
"learning_rate": 2.2379798727743207e-07,
|
| 2878 |
-
"loss": 0.0803,
|
| 2879 |
-
"step": 10250
|
| 2880 |
-
},
|
| 2881 |
-
{
|
| 2882 |
-
"epoch": 0.9546594815571866,
|
| 2883 |
-
"grad_norm": 0.7577381730079651,
|
| 2884 |
-
"learning_rate": 2.0255281389579106e-07,
|
| 2885 |
-
"loss": 0.1743,
|
| 2886 |
-
"step": 10275
|
| 2887 |
-
},
|
| 2888 |
-
{
|
| 2889 |
-
"epoch": 0.9569822540183963,
|
| 2890 |
-
"grad_norm": 0.19169993698596954,
|
| 2891 |
-
"learning_rate": 1.823618340481348e-07,
|
| 2892 |
-
"loss": 0.0999,
|
| 2893 |
-
"step": 10300
|
| 2894 |
-
},
|
| 2895 |
-
{
|
| 2896 |
-
"epoch": 0.959305026479606,
|
| 2897 |
-
"grad_norm": 0.18852142989635468,
|
| 2898 |
-
"learning_rate": 1.6322612288320883e-07,
|
| 2899 |
-
"loss": 0.0882,
|
| 2900 |
-
"step": 10325
|
| 2901 |
-
},
|
| 2902 |
-
{
|
| 2903 |
-
"epoch": 0.9616277989408157,
|
| 2904 |
-
"grad_norm": 0.05598045513033867,
|
| 2905 |
-
"learning_rate": 1.4514669935778546e-07,
|
| 2906 |
-
"loss": 0.0533,
|
| 2907 |
-
"step": 10350
|
| 2908 |
-
},
|
| 2909 |
-
{
|
| 2910 |
-
"epoch": 0.9639505714020254,
|
| 2911 |
-
"grad_norm": 0.17084026336669922,
|
| 2912 |
-
"learning_rate": 1.2812452618242045e-07,
|
| 2913 |
-
"loss": 0.0919,
|
| 2914 |
-
"step": 10375
|
| 2915 |
-
},
|
| 2916 |
-
{
|
| 2917 |
-
"epoch": 0.9662733438632352,
|
| 2918 |
-
"grad_norm": 58.29783630371094,
|
| 2919 |
-
"learning_rate": 1.1216050977018944e-07,
|
| 2920 |
-
"loss": 0.1371,
|
| 2921 |
-
"step": 10400
|
| 2922 |
-
},
|
| 2923 |
-
{
|
| 2924 |
-
"epoch": 0.9685961163244449,
|
| 2925 |
-
"grad_norm": 0.11096494644880295,
|
| 2926 |
-
"learning_rate": 9.725550018840679e-08,
|
| 2927 |
-
"loss": 0.0345,
|
| 2928 |
-
"step": 10425
|
| 2929 |
-
},
|
| 2930 |
-
{
|
| 2931 |
-
"epoch": 0.9709188887856546,
|
| 2932 |
-
"grad_norm": 17.48317527770996,
|
| 2933 |
-
"learning_rate": 8.341029111338161e-08,
|
| 2934 |
-
"loss": 0.1682,
|
| 2935 |
-
"step": 10450
|
| 2936 |
-
},
|
| 2937 |
-
{
|
| 2938 |
-
"epoch": 0.9732416612468643,
|
| 2939 |
-
"grad_norm": 0.34725722670555115,
|
| 2940 |
-
"learning_rate": 7.06256197881361e-08,
|
| 2941 |
-
"loss": 0.074,
|
| 2942 |
-
"step": 10475
|
| 2943 |
-
},
|
| 2944 |
-
{
|
| 2945 |
-
"epoch": 0.975564433708074,
|
| 2946 |
-
"grad_norm": 0.04875205457210541,
|
| 2947 |
-
"learning_rate": 5.890216698316798e-08,
|
| 2948 |
-
"loss": 0.1402,
|
| 2949 |
-
"step": 10500
|
| 2950 |
-
},
|
| 2951 |
-
{
|
| 2952 |
-
"epoch": 0.9778872061692837,
|
| 2953 |
-
"grad_norm": 0.07622114568948746,
|
| 2954 |
-
"learning_rate": 4.8240556960175154e-08,
|
| 2955 |
-
"loss": 0.0474,
|
| 2956 |
-
"step": 10525
|
| 2957 |
-
},
|
| 2958 |
-
{
|
| 2959 |
-
"epoch": 0.9802099786304933,
|
| 2960 |
-
"grad_norm": 0.07797674834728241,
|
| 2961 |
-
"learning_rate": 3.864135743884223e-08,
|
| 2962 |
-
"loss": 0.0593,
|
| 2963 |
-
"step": 10550
|
| 2964 |
-
},
|
| 2965 |
-
{
|
| 2966 |
-
"epoch": 0.982532751091703,
|
| 2967 |
-
"grad_norm": 1.8889813423156738,
|
| 2968 |
-
"learning_rate": 3.0105079566584705e-08,
|
| 2969 |
-
"loss": 0.0382,
|
| 2970 |
-
"step": 10575
|
| 2971 |
-
},
|
| 2972 |
-
{
|
| 2973 |
-
"epoch": 0.9848555235529127,
|
| 2974 |
-
"grad_norm": 0.22253760695457458,
|
| 2975 |
-
"learning_rate": 2.2632177891352967e-08,
|
| 2976 |
-
"loss": 0.0638,
|
| 2977 |
-
"step": 10600
|
| 2978 |
-
},
|
| 2979 |
-
{
|
| 2980 |
-
"epoch": 0.9871782960141224,
|
| 2981 |
-
"grad_norm": 0.7944088578224182,
|
| 2982 |
-
"learning_rate": 1.6223050337407232e-08,
|
| 2983 |
-
"loss": 0.0549,
|
| 2984 |
-
"step": 10625
|
| 2985 |
-
},
|
| 2986 |
-
{
|
| 2987 |
-
"epoch": 0.9895010684753321,
|
| 2988 |
-
"grad_norm": 13.208889961242676,
|
| 2989 |
-
"learning_rate": 1.0878038184147787e-08,
|
| 2990 |
-
"loss": 0.1545,
|
| 2991 |
-
"step": 10650
|
| 2992 |
-
},
|
| 2993 |
-
{
|
| 2994 |
-
"epoch": 0.9918238409365419,
|
| 2995 |
-
"grad_norm": 1.312228798866272,
|
| 2996 |
-
"learning_rate": 6.597426047927346e-09,
|
| 2997 |
-
"loss": 0.095,
|
| 2998 |
-
"step": 10675
|
| 2999 |
-
},
|
| 3000 |
-
{
|
| 3001 |
-
"epoch": 0.9941466133977516,
|
| 3002 |
-
"grad_norm": 0.08410879969596863,
|
| 3003 |
-
"learning_rate": 3.381441866905366e-09,
|
| 3004 |
-
"loss": 0.178,
|
| 3005 |
-
"step": 10700
|
| 3006 |
-
},
|
| 3007 |
-
{
|
| 3008 |
-
"epoch": 0.9964693858589613,
|
| 3009 |
-
"grad_norm": 0.03298262879252434,
|
| 3010 |
-
"learning_rate": 1.2302568889022147e-09,
|
| 3011 |
-
"loss": 0.1228,
|
| 3012 |
-
"step": 10725
|
| 3013 |
-
},
|
| 3014 |
-
{
|
| 3015 |
-
"epoch": 0.998792158320171,
|
| 3016 |
-
"grad_norm": 24.537267684936523,
|
| 3017 |
-
"learning_rate": 1.4398566228646105e-10,
|
| 3018 |
-
"loss": 0.034,
|
| 3019 |
-
"step": 10750
|
| 3020 |
-
},
|
| 3021 |
-
{
|
| 3022 |
-
"epoch": 1.0,
|
| 3023 |
-
"eval_accuracy": 0.9658653207597291,
|
| 3024 |
-
"eval_auc": 0.9894164864898246,
|
| 3025 |
-
"eval_f1": 0.9687689877263337,
|
| 3026 |
-
"eval_loss": 0.17713692784309387,
|
| 3027 |
-
"eval_precision": 0.9622208811104406,
|
| 3028 |
-
"eval_recall": 0.9754068273583751,
|
| 3029 |
-
"eval_runtime": 2165.56,
|
| 3030 |
-
"eval_samples_per_second": 6.953,
|
| 3031 |
-
"eval_steps_per_second": 0.29,
|
| 3032 |
-
"step": 10763
|
| 3033 |
-
}
|
| 3034 |
-
],
|
| 3035 |
-
"logging_steps": 25,
|
| 3036 |
-
"max_steps": 10763,
|
| 3037 |
-
"num_input_tokens_seen": 0,
|
| 3038 |
-
"num_train_epochs": 1,
|
| 3039 |
-
"save_steps": 500,
|
| 3040 |
-
"stateful_callbacks": {
|
| 3041 |
-
"EarlyStoppingCallback": {
|
| 3042 |
-
"args": {
|
| 3043 |
-
"early_stopping_patience": 5,
|
| 3044 |
-
"early_stopping_threshold": 0.01
|
| 3045 |
-
},
|
| 3046 |
-
"attributes": {
|
| 3047 |
-
"early_stopping_patience_counter": 0
|
| 3048 |
-
}
|
| 3049 |
-
},
|
| 3050 |
-
"TrainerControl": {
|
| 3051 |
-
"args": {
|
| 3052 |
-
"should_epoch_stop": false,
|
| 3053 |
-
"should_evaluate": false,
|
| 3054 |
-
"should_log": false,
|
| 3055 |
-
"should_save": true,
|
| 3056 |
-
"should_training_stop": true
|
| 3057 |
-
},
|
| 3058 |
-
"attributes": {}
|
| 3059 |
-
}
|
| 3060 |
-
},
|
| 3061 |
-
"total_flos": 2.9740165749559e+19,
|
| 3062 |
-
"train_batch_size": 12,
|
| 3063 |
-
"trial_name": null,
|
| 3064 |
-
"trial_params": null
|
| 3065 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
{checkpoint-10763 → checkpoint-11092}/config.json
RENAMED
|
File without changes
|
{checkpoint-10763 → checkpoint-11092}/model.safetensors
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 780886280
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:eb2da3e13bdcac1efd85b67f172389f48a7d5955c139ee019e5b1af566969611
|
| 3 |
size 780886280
|
{checkpoint-10763 → checkpoint-11092}/optimizer.pt
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1562042973
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ef177cc51f41eaae48c8855f0e5bd1fb082c3838326d0b42cb37b51ee8471487
|
| 3 |
size 1562042973
|
{checkpoint-10763 → checkpoint-11092}/rng_state.pth
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b243e5aa2d23928fd94372e4aa6589a0d1283080ceec2ce84a1581ad5abfe369
|
| 3 |
size 14244
|
{checkpoint-10763 → checkpoint-11092}/scheduler.pt
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cd0f59e53185b00a12bcffab57ca508c924f7e8acfa052743c559a7a3afe1672
|
| 3 |
size 1064
|
checkpoint-11092/trainer_state.json
ADDED
|
@@ -0,0 +1,3156 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.16320379078388214,
|
| 3 |
+
"best_model_checkpoint": "autotrain-ai-image-detect-20241207-0028/checkpoint-11092",
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 11092,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0022538766678687344,
|
| 13 |
+
"grad_norm": 15.676097869873047,
|
| 14 |
+
"learning_rate": 3.999949863013569e-05,
|
| 15 |
+
"loss": 0.1114,
|
| 16 |
+
"step": 25
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.004507753335737469,
|
| 20 |
+
"grad_norm": 0.012626341544091702,
|
| 21 |
+
"learning_rate": 3.999799454567995e-05,
|
| 22 |
+
"loss": 0.0507,
|
| 23 |
+
"step": 50
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.006761630003606202,
|
| 27 |
+
"grad_norm": 0.014825106598436832,
|
| 28 |
+
"learning_rate": 3.999548782204303e-05,
|
| 29 |
+
"loss": 0.1927,
|
| 30 |
+
"step": 75
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.009015506671474938,
|
| 34 |
+
"grad_norm": 0.3768613636493683,
|
| 35 |
+
"learning_rate": 3.999197858490449e-05,
|
| 36 |
+
"loss": 0.1248,
|
| 37 |
+
"step": 100
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.01126938333934367,
|
| 41 |
+
"grad_norm": 1.3044193983078003,
|
| 42 |
+
"learning_rate": 3.998746701020692e-05,
|
| 43 |
+
"loss": 0.1965,
|
| 44 |
+
"step": 125
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.013523260007212405,
|
| 48 |
+
"grad_norm": 0.36453139781951904,
|
| 49 |
+
"learning_rate": 3.998195332414708e-05,
|
| 50 |
+
"loss": 0.1072,
|
| 51 |
+
"step": 150
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.01577713667508114,
|
| 55 |
+
"grad_norm": 2.911674737930298,
|
| 56 |
+
"learning_rate": 3.997543780316456e-05,
|
| 57 |
+
"loss": 0.0886,
|
| 58 |
+
"step": 175
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.018031013342949875,
|
| 62 |
+
"grad_norm": 0.29508596658706665,
|
| 63 |
+
"learning_rate": 3.9967920773927955e-05,
|
| 64 |
+
"loss": 0.1546,
|
| 65 |
+
"step": 200
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.020284890010818606,
|
| 69 |
+
"grad_norm": 0.1335642784833908,
|
| 70 |
+
"learning_rate": 3.9959402613318466e-05,
|
| 71 |
+
"loss": 0.2164,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.02253876667868734,
|
| 76 |
+
"grad_norm": 33.6629638671875,
|
| 77 |
+
"learning_rate": 3.9949883748410986e-05,
|
| 78 |
+
"loss": 0.0635,
|
| 79 |
+
"step": 250
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.024792643346556075,
|
| 83 |
+
"grad_norm": 0.02095983922481537,
|
| 84 |
+
"learning_rate": 3.993936465645272e-05,
|
| 85 |
+
"loss": 0.1648,
|
| 86 |
+
"step": 275
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.02704652001442481,
|
| 90 |
+
"grad_norm": 0.10742558538913727,
|
| 91 |
+
"learning_rate": 3.992784586483923e-05,
|
| 92 |
+
"loss": 0.1705,
|
| 93 |
+
"step": 300
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.029300396682293544,
|
| 97 |
+
"grad_norm": 8.841888427734375,
|
| 98 |
+
"learning_rate": 3.9915327951088025e-05,
|
| 99 |
+
"loss": 0.1467,
|
| 100 |
+
"step": 325
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.03155427335016228,
|
| 104 |
+
"grad_norm": 0.5381837487220764,
|
| 105 |
+
"learning_rate": 3.990181154280957e-05,
|
| 106 |
+
"loss": 0.157,
|
| 107 |
+
"step": 350
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.033808150018031016,
|
| 111 |
+
"grad_norm": 13.203426361083984,
|
| 112 |
+
"learning_rate": 3.988729731767585e-05,
|
| 113 |
+
"loss": 0.1299,
|
| 114 |
+
"step": 375
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.03606202668589975,
|
| 118 |
+
"grad_norm": 0.5511285662651062,
|
| 119 |
+
"learning_rate": 3.987178600338637e-05,
|
| 120 |
+
"loss": 0.0957,
|
| 121 |
+
"step": 400
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.038315903353768485,
|
| 125 |
+
"grad_norm": 18.01679801940918,
|
| 126 |
+
"learning_rate": 3.985527837763168e-05,
|
| 127 |
+
"loss": 0.2378,
|
| 128 |
+
"step": 425
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.04056978002163721,
|
| 132 |
+
"grad_norm": 28.235441207885742,
|
| 133 |
+
"learning_rate": 3.9837775268054396e-05,
|
| 134 |
+
"loss": 0.1541,
|
| 135 |
+
"step": 450
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.04282365668950595,
|
| 139 |
+
"grad_norm": 0.40507128834724426,
|
| 140 |
+
"learning_rate": 3.9819277552207684e-05,
|
| 141 |
+
"loss": 0.0903,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.04507753335737468,
|
| 146 |
+
"grad_norm": 35.182228088378906,
|
| 147 |
+
"learning_rate": 3.979978615751127e-05,
|
| 148 |
+
"loss": 0.2194,
|
| 149 |
+
"step": 500
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.047331410025243416,
|
| 153 |
+
"grad_norm": 1.6243983507156372,
|
| 154 |
+
"learning_rate": 3.9779302061204947e-05,
|
| 155 |
+
"loss": 0.2171,
|
| 156 |
+
"step": 525
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.04958528669311215,
|
| 160 |
+
"grad_norm": 0.029595619067549706,
|
| 161 |
+
"learning_rate": 3.975782629029957e-05,
|
| 162 |
+
"loss": 0.0641,
|
| 163 |
+
"step": 550
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.051839163360980885,
|
| 167 |
+
"grad_norm": 19.178627014160156,
|
| 168 |
+
"learning_rate": 3.973535992152557e-05,
|
| 169 |
+
"loss": 0.1061,
|
| 170 |
+
"step": 575
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.05409304002884962,
|
| 174 |
+
"grad_norm": 0.028080996125936508,
|
| 175 |
+
"learning_rate": 3.971190408127899e-05,
|
| 176 |
+
"loss": 0.1615,
|
| 177 |
+
"step": 600
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.056346916696718354,
|
| 181 |
+
"grad_norm": 27.624467849731445,
|
| 182 |
+
"learning_rate": 3.968745994556496e-05,
|
| 183 |
+
"loss": 0.0824,
|
| 184 |
+
"step": 625
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.05860079336458709,
|
| 188 |
+
"grad_norm": 30.725770950317383,
|
| 189 |
+
"learning_rate": 3.966202873993878e-05,
|
| 190 |
+
"loss": 0.1061,
|
| 191 |
+
"step": 650
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.06085467003245582,
|
| 195 |
+
"grad_norm": 23.075244903564453,
|
| 196 |
+
"learning_rate": 3.963561173944447e-05,
|
| 197 |
+
"loss": 0.2112,
|
| 198 |
+
"step": 675
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.06310854670032456,
|
| 202 |
+
"grad_norm": 0.20598366856575012,
|
| 203 |
+
"learning_rate": 3.960821026855082e-05,
|
| 204 |
+
"loss": 0.0591,
|
| 205 |
+
"step": 700
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.0653624233681933,
|
| 209 |
+
"grad_norm": 0.17925754189491272,
|
| 210 |
+
"learning_rate": 3.957982570108501e-05,
|
| 211 |
+
"loss": 0.1284,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.06761630003606203,
|
| 216 |
+
"grad_norm": 19.054052352905273,
|
| 217 |
+
"learning_rate": 3.9550459460163694e-05,
|
| 218 |
+
"loss": 0.1575,
|
| 219 |
+
"step": 750
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.06987017670393077,
|
| 223 |
+
"grad_norm": 0.1789284348487854,
|
| 224 |
+
"learning_rate": 3.9520113018121725e-05,
|
| 225 |
+
"loss": 0.1586,
|
| 226 |
+
"step": 775
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.0721240533717995,
|
| 230 |
+
"grad_norm": 0.5868380665779114,
|
| 231 |
+
"learning_rate": 3.948878789643823e-05,
|
| 232 |
+
"loss": 0.099,
|
| 233 |
+
"step": 800
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.07437793003966824,
|
| 237 |
+
"grad_norm": 103.31700134277344,
|
| 238 |
+
"learning_rate": 3.945648566566043e-05,
|
| 239 |
+
"loss": 0.1047,
|
| 240 |
+
"step": 825
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.07663180670753697,
|
| 244 |
+
"grad_norm": 31.75994873046875,
|
| 245 |
+
"learning_rate": 3.942320794532482e-05,
|
| 246 |
+
"loss": 0.0846,
|
| 247 |
+
"step": 850
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.0788856833754057,
|
| 251 |
+
"grad_norm": 18.438316345214844,
|
| 252 |
+
"learning_rate": 3.938895640387601e-05,
|
| 253 |
+
"loss": 0.0937,
|
| 254 |
+
"step": 875
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.08113956004327443,
|
| 258 |
+
"grad_norm": 0.10990960150957108,
|
| 259 |
+
"learning_rate": 3.9353732758583066e-05,
|
| 260 |
+
"loss": 0.1218,
|
| 261 |
+
"step": 900
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.08339343671114316,
|
| 265 |
+
"grad_norm": 11.226954460144043,
|
| 266 |
+
"learning_rate": 3.931753877545343e-05,
|
| 267 |
+
"loss": 0.1213,
|
| 268 |
+
"step": 925
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.0856473133790119,
|
| 272 |
+
"grad_norm": 26.347089767456055,
|
| 273 |
+
"learning_rate": 3.928037626914433e-05,
|
| 274 |
+
"loss": 0.1545,
|
| 275 |
+
"step": 950
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.08790119004688063,
|
| 279 |
+
"grad_norm": 0.1437385380268097,
|
| 280 |
+
"learning_rate": 3.924224710287186e-05,
|
| 281 |
+
"loss": 0.1783,
|
| 282 |
+
"step": 975
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.09015506671474936,
|
| 286 |
+
"grad_norm": 17.052108764648438,
|
| 287 |
+
"learning_rate": 3.920315318831748e-05,
|
| 288 |
+
"loss": 0.1552,
|
| 289 |
+
"step": 1000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.0924089433826181,
|
| 293 |
+
"grad_norm": 0.31946513056755066,
|
| 294 |
+
"learning_rate": 3.9163096485532276e-05,
|
| 295 |
+
"loss": 0.1392,
|
| 296 |
+
"step": 1025
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.09466282005048683,
|
| 300 |
+
"grad_norm": 9.407455444335938,
|
| 301 |
+
"learning_rate": 3.912207900283861e-05,
|
| 302 |
+
"loss": 0.1396,
|
| 303 |
+
"step": 1050
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.09691669671835557,
|
| 307 |
+
"grad_norm": 41.089141845703125,
|
| 308 |
+
"learning_rate": 3.908010279672945e-05,
|
| 309 |
+
"loss": 0.1832,
|
| 310 |
+
"step": 1075
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.0991705733862243,
|
| 314 |
+
"grad_norm": 0.14397946000099182,
|
| 315 |
+
"learning_rate": 3.9037169971765265e-05,
|
| 316 |
+
"loss": 0.2347,
|
| 317 |
+
"step": 1100
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.10142445005409303,
|
| 321 |
+
"grad_norm": 11.140134811401367,
|
| 322 |
+
"learning_rate": 3.899328268046854e-05,
|
| 323 |
+
"loss": 0.1394,
|
| 324 |
+
"step": 1125
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.10367832672196177,
|
| 328 |
+
"grad_norm": 44.023338317871094,
|
| 329 |
+
"learning_rate": 3.894844312321578e-05,
|
| 330 |
+
"loss": 0.1914,
|
| 331 |
+
"step": 1150
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.1059322033898305,
|
| 335 |
+
"grad_norm": 0.09425333887338638,
|
| 336 |
+
"learning_rate": 3.890265354812727e-05,
|
| 337 |
+
"loss": 0.0997,
|
| 338 |
+
"step": 1175
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.10818608005769924,
|
| 342 |
+
"grad_norm": 0.04344582557678223,
|
| 343 |
+
"learning_rate": 3.885591625095432e-05,
|
| 344 |
+
"loss": 0.1207,
|
| 345 |
+
"step": 1200
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.11043995672556797,
|
| 349 |
+
"grad_norm": 14.171688079833984,
|
| 350 |
+
"learning_rate": 3.880823357496415e-05,
|
| 351 |
+
"loss": 0.1482,
|
| 352 |
+
"step": 1225
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.11269383339343671,
|
| 356 |
+
"grad_norm": 0.5523734092712402,
|
| 357 |
+
"learning_rate": 3.875960791082245e-05,
|
| 358 |
+
"loss": 0.12,
|
| 359 |
+
"step": 1250
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.11494771006130544,
|
| 363 |
+
"grad_norm": 33.0614128112793,
|
| 364 |
+
"learning_rate": 3.8710041696473484e-05,
|
| 365 |
+
"loss": 0.2042,
|
| 366 |
+
"step": 1275
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.11720158672917418,
|
| 370 |
+
"grad_norm": 0.36186420917510986,
|
| 371 |
+
"learning_rate": 3.865953741701785e-05,
|
| 372 |
+
"loss": 0.1783,
|
| 373 |
+
"step": 1300
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.11945546339704291,
|
| 377 |
+
"grad_norm": 92.81356048583984,
|
| 378 |
+
"learning_rate": 3.8608097604587946e-05,
|
| 379 |
+
"loss": 0.1225,
|
| 380 |
+
"step": 1325
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.12170934006491164,
|
| 384 |
+
"grad_norm": 0.8442438840866089,
|
| 385 |
+
"learning_rate": 3.855572483822094e-05,
|
| 386 |
+
"loss": 0.1182,
|
| 387 |
+
"step": 1350
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.12396321673278038,
|
| 391 |
+
"grad_norm": 0.04132000356912613,
|
| 392 |
+
"learning_rate": 3.850242174372951e-05,
|
| 393 |
+
"loss": 0.1182,
|
| 394 |
+
"step": 1375
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.12621709340064913,
|
| 398 |
+
"grad_norm": 91.91195678710938,
|
| 399 |
+
"learning_rate": 3.844819099357017e-05,
|
| 400 |
+
"loss": 0.0642,
|
| 401 |
+
"step": 1400
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.12847097006851785,
|
| 405 |
+
"grad_norm": 16.58065414428711,
|
| 406 |
+
"learning_rate": 3.8393035306709315e-05,
|
| 407 |
+
"loss": 0.2305,
|
| 408 |
+
"step": 1425
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.1307248467363866,
|
| 412 |
+
"grad_norm": 16.433385848999023,
|
| 413 |
+
"learning_rate": 3.8336957448486874e-05,
|
| 414 |
+
"loss": 0.1944,
|
| 415 |
+
"step": 1450
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.13297872340425532,
|
| 419 |
+
"grad_norm": 10.438956260681152,
|
| 420 |
+
"learning_rate": 3.8279960230477655e-05,
|
| 421 |
+
"loss": 0.1467,
|
| 422 |
+
"step": 1475
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.13523260007212407,
|
| 426 |
+
"grad_norm": 0.17678000032901764,
|
| 427 |
+
"learning_rate": 3.822204651035041e-05,
|
| 428 |
+
"loss": 0.0659,
|
| 429 |
+
"step": 1500
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.13748647673999279,
|
| 433 |
+
"grad_norm": 64.35008239746094,
|
| 434 |
+
"learning_rate": 3.816321919172453e-05,
|
| 435 |
+
"loss": 0.1638,
|
| 436 |
+
"step": 1525
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.13974035340786153,
|
| 440 |
+
"grad_norm": 0.048081036657094955,
|
| 441 |
+
"learning_rate": 3.810348122402451e-05,
|
| 442 |
+
"loss": 0.1203,
|
| 443 |
+
"step": 1550
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.14199423007573025,
|
| 447 |
+
"grad_norm": 0.34617549180984497,
|
| 448 |
+
"learning_rate": 3.8042835602332e-05,
|
| 449 |
+
"loss": 0.1667,
|
| 450 |
+
"step": 1575
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.144248106743599,
|
| 454 |
+
"grad_norm": 2.3699162006378174,
|
| 455 |
+
"learning_rate": 3.798128536723572e-05,
|
| 456 |
+
"loss": 0.1016,
|
| 457 |
+
"step": 1600
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.14650198341146772,
|
| 461 |
+
"grad_norm": 0.058592069894075394,
|
| 462 |
+
"learning_rate": 3.791883360467899e-05,
|
| 463 |
+
"loss": 0.1542,
|
| 464 |
+
"step": 1625
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.14875586007933647,
|
| 468 |
+
"grad_norm": 14.057805061340332,
|
| 469 |
+
"learning_rate": 3.785548344580496e-05,
|
| 470 |
+
"loss": 0.1249,
|
| 471 |
+
"step": 1650
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.1510097367472052,
|
| 475 |
+
"grad_norm": 1.7941921949386597,
|
| 476 |
+
"learning_rate": 3.77912380667997e-05,
|
| 477 |
+
"loss": 0.1423,
|
| 478 |
+
"step": 1675
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.15326361341507394,
|
| 482 |
+
"grad_norm": 0.17149803042411804,
|
| 483 |
+
"learning_rate": 3.772610068873289e-05,
|
| 484 |
+
"loss": 0.1051,
|
| 485 |
+
"step": 1700
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.15551749008294266,
|
| 489 |
+
"grad_norm": 0.19008079171180725,
|
| 490 |
+
"learning_rate": 3.7660074577396384e-05,
|
| 491 |
+
"loss": 0.1792,
|
| 492 |
+
"step": 1725
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.1577713667508114,
|
| 496 |
+
"grad_norm": 0.5955036878585815,
|
| 497 |
+
"learning_rate": 3.759316304314042e-05,
|
| 498 |
+
"loss": 0.0913,
|
| 499 |
+
"step": 1750
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.16002524341868013,
|
| 503 |
+
"grad_norm": 0.3940871059894562,
|
| 504 |
+
"learning_rate": 3.752536944070769e-05,
|
| 505 |
+
"loss": 0.1519,
|
| 506 |
+
"step": 1775
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.16227912008654885,
|
| 510 |
+
"grad_norm": 0.3010105788707733,
|
| 511 |
+
"learning_rate": 3.7456697169065135e-05,
|
| 512 |
+
"loss": 0.0829,
|
| 513 |
+
"step": 1800
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.1645329967544176,
|
| 517 |
+
"grad_norm": 28.83148956298828,
|
| 518 |
+
"learning_rate": 3.7387149671233475e-05,
|
| 519 |
+
"loss": 0.1205,
|
| 520 |
+
"step": 1825
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.16678687342228632,
|
| 524 |
+
"grad_norm": 31.214406967163086,
|
| 525 |
+
"learning_rate": 3.731673043411468e-05,
|
| 526 |
+
"loss": 0.14,
|
| 527 |
+
"step": 1850
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.16904075009015507,
|
| 531 |
+
"grad_norm": 15.15921688079834,
|
| 532 |
+
"learning_rate": 3.724544298831709e-05,
|
| 533 |
+
"loss": 0.1614,
|
| 534 |
+
"step": 1875
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.1712946267580238,
|
| 538 |
+
"grad_norm": 4.872321128845215,
|
| 539 |
+
"learning_rate": 3.71732909079784e-05,
|
| 540 |
+
"loss": 0.1176,
|
| 541 |
+
"step": 1900
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.17354850342589254,
|
| 545 |
+
"grad_norm": 16.01061248779297,
|
| 546 |
+
"learning_rate": 3.7100277810586487e-05,
|
| 547 |
+
"loss": 0.1893,
|
| 548 |
+
"step": 1925
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.17580238009376126,
|
| 552 |
+
"grad_norm": 0.6090813875198364,
|
| 553 |
+
"learning_rate": 3.702640735679803e-05,
|
| 554 |
+
"loss": 0.1686,
|
| 555 |
+
"step": 1950
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.17805625676163,
|
| 559 |
+
"grad_norm": 0.49127471446990967,
|
| 560 |
+
"learning_rate": 3.695168325025494e-05,
|
| 561 |
+
"loss": 0.157,
|
| 562 |
+
"step": 1975
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.18031013342949873,
|
| 566 |
+
"grad_norm": 0.2058860808610916,
|
| 567 |
+
"learning_rate": 3.687610923739877e-05,
|
| 568 |
+
"loss": 0.1284,
|
| 569 |
+
"step": 2000
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.18256401009736747,
|
| 573 |
+
"grad_norm": 26.90603256225586,
|
| 574 |
+
"learning_rate": 3.679968910728277e-05,
|
| 575 |
+
"loss": 0.0757,
|
| 576 |
+
"step": 2025
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.1848178867652362,
|
| 580 |
+
"grad_norm": 64.29013061523438,
|
| 581 |
+
"learning_rate": 3.672242669138194e-05,
|
| 582 |
+
"loss": 0.1802,
|
| 583 |
+
"step": 2050
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.18707176343310494,
|
| 587 |
+
"grad_norm": 0.6742556095123291,
|
| 588 |
+
"learning_rate": 3.664432586340101e-05,
|
| 589 |
+
"loss": 0.1707,
|
| 590 |
+
"step": 2075
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.18932564010097366,
|
| 594 |
+
"grad_norm": 15.53222370147705,
|
| 595 |
+
"learning_rate": 3.656539053908011e-05,
|
| 596 |
+
"loss": 0.148,
|
| 597 |
+
"step": 2100
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.1915795167688424,
|
| 601 |
+
"grad_norm": 0.5117244720458984,
|
| 602 |
+
"learning_rate": 3.648562467599854e-05,
|
| 603 |
+
"loss": 0.0652,
|
| 604 |
+
"step": 2125
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.19383339343671113,
|
| 608 |
+
"grad_norm": 0.14978158473968506,
|
| 609 |
+
"learning_rate": 3.640503227337628e-05,
|
| 610 |
+
"loss": 0.086,
|
| 611 |
+
"step": 2150
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.19608727010457988,
|
| 615 |
+
"grad_norm": 0.9616666436195374,
|
| 616 |
+
"learning_rate": 3.632361737187354e-05,
|
| 617 |
+
"loss": 0.2264,
|
| 618 |
+
"step": 2175
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.1983411467724486,
|
| 622 |
+
"grad_norm": 0.0734601691365242,
|
| 623 |
+
"learning_rate": 3.624138405338812e-05,
|
| 624 |
+
"loss": 0.1176,
|
| 625 |
+
"step": 2200
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.20059502344031735,
|
| 629 |
+
"grad_norm": 0.21638202667236328,
|
| 630 |
+
"learning_rate": 3.61583364408508e-05,
|
| 631 |
+
"loss": 0.1449,
|
| 632 |
+
"step": 2225
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.20284890010818607,
|
| 636 |
+
"grad_norm": 2.470684289932251,
|
| 637 |
+
"learning_rate": 3.607447869801861e-05,
|
| 638 |
+
"loss": 0.1816,
|
| 639 |
+
"step": 2250
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.20510277677605482,
|
| 643 |
+
"grad_norm": 9.10354232788086,
|
| 644 |
+
"learning_rate": 3.5989815029266045e-05,
|
| 645 |
+
"loss": 0.2137,
|
| 646 |
+
"step": 2275
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.20735665344392354,
|
| 650 |
+
"grad_norm": 64.91108703613281,
|
| 651 |
+
"learning_rate": 3.590434967937434e-05,
|
| 652 |
+
"loss": 0.154,
|
| 653 |
+
"step": 2300
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.2096105301117923,
|
| 657 |
+
"grad_norm": 14.862517356872559,
|
| 658 |
+
"learning_rate": 3.5818086933318564e-05,
|
| 659 |
+
"loss": 0.1112,
|
| 660 |
+
"step": 2325
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.211864406779661,
|
| 664 |
+
"grad_norm": 1.856042742729187,
|
| 665 |
+
"learning_rate": 3.573103111605286e-05,
|
| 666 |
+
"loss": 0.2006,
|
| 667 |
+
"step": 2350
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.21411828344752976,
|
| 671 |
+
"grad_norm": 49.54050827026367,
|
| 672 |
+
"learning_rate": 3.5643186592293534e-05,
|
| 673 |
+
"loss": 0.1805,
|
| 674 |
+
"step": 2375
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.21637216011539848,
|
| 678 |
+
"grad_norm": 0.0626867488026619,
|
| 679 |
+
"learning_rate": 3.555455776630031e-05,
|
| 680 |
+
"loss": 0.2223,
|
| 681 |
+
"step": 2400
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.21862603678326722,
|
| 685 |
+
"grad_norm": 1.3941552639007568,
|
| 686 |
+
"learning_rate": 3.546514908165542e-05,
|
| 687 |
+
"loss": 0.2031,
|
| 688 |
+
"step": 2425
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.22087991345113595,
|
| 692 |
+
"grad_norm": 11.90259075164795,
|
| 693 |
+
"learning_rate": 3.537496502104087e-05,
|
| 694 |
+
"loss": 0.192,
|
| 695 |
+
"step": 2450
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.2231337901190047,
|
| 699 |
+
"grad_norm": 0.6432807445526123,
|
| 700 |
+
"learning_rate": 3.5284010106013685e-05,
|
| 701 |
+
"loss": 0.1615,
|
| 702 |
+
"step": 2475
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 0.22538766678687341,
|
| 706 |
+
"grad_norm": 17.459768295288086,
|
| 707 |
+
"learning_rate": 3.519228889677922e-05,
|
| 708 |
+
"loss": 0.1193,
|
| 709 |
+
"step": 2500
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.22764154345474216,
|
| 713 |
+
"grad_norm": 0.1881784200668335,
|
| 714 |
+
"learning_rate": 3.509980599196247e-05,
|
| 715 |
+
"loss": 0.1313,
|
| 716 |
+
"step": 2525
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.22989542012261088,
|
| 720 |
+
"grad_norm": 1.5529861450195312,
|
| 721 |
+
"learning_rate": 3.500656602837761e-05,
|
| 722 |
+
"loss": 0.106,
|
| 723 |
+
"step": 2550
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.23214929679047963,
|
| 727 |
+
"grad_norm": 28.792247772216797,
|
| 728 |
+
"learning_rate": 3.491257368079541e-05,
|
| 729 |
+
"loss": 0.14,
|
| 730 |
+
"step": 2575
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.23440317345834835,
|
| 734 |
+
"grad_norm": 8.728706359863281,
|
| 735 |
+
"learning_rate": 3.481783366170892e-05,
|
| 736 |
+
"loss": 0.1256,
|
| 737 |
+
"step": 2600
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.2366570501262171,
|
| 741 |
+
"grad_norm": 24.14179039001465,
|
| 742 |
+
"learning_rate": 3.472235072109721e-05,
|
| 743 |
+
"loss": 0.2334,
|
| 744 |
+
"step": 2625
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.23891092679408582,
|
| 748 |
+
"grad_norm": 13.173722267150879,
|
| 749 |
+
"learning_rate": 3.462612964618716e-05,
|
| 750 |
+
"loss": 0.1811,
|
| 751 |
+
"step": 2650
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.24116480346195457,
|
| 755 |
+
"grad_norm": 9.828522682189941,
|
| 756 |
+
"learning_rate": 3.4529175261213515e-05,
|
| 757 |
+
"loss": 0.2615,
|
| 758 |
+
"step": 2675
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.2434186801298233,
|
| 762 |
+
"grad_norm": 51.55712127685547,
|
| 763 |
+
"learning_rate": 3.4431492427176946e-05,
|
| 764 |
+
"loss": 0.1282,
|
| 765 |
+
"step": 2700
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.24567255679769204,
|
| 769 |
+
"grad_norm": 9.62954044342041,
|
| 770 |
+
"learning_rate": 3.433308604160037e-05,
|
| 771 |
+
"loss": 0.1331,
|
| 772 |
+
"step": 2725
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 0.24792643346556076,
|
| 776 |
+
"grad_norm": 0.8969252705574036,
|
| 777 |
+
"learning_rate": 3.423396103828343e-05,
|
| 778 |
+
"loss": 0.1256,
|
| 779 |
+
"step": 2750
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 0.2501803101334295,
|
| 783 |
+
"grad_norm": 0.12056900560855865,
|
| 784 |
+
"learning_rate": 3.413412238705505e-05,
|
| 785 |
+
"loss": 0.1954,
|
| 786 |
+
"step": 2775
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 0.25243418680129825,
|
| 790 |
+
"grad_norm": 7.533175468444824,
|
| 791 |
+
"learning_rate": 3.403357509352434e-05,
|
| 792 |
+
"loss": 0.1871,
|
| 793 |
+
"step": 2800
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.25468806346916695,
|
| 797 |
+
"grad_norm": 0.23234868049621582,
|
| 798 |
+
"learning_rate": 3.39323241988296e-05,
|
| 799 |
+
"loss": 0.1379,
|
| 800 |
+
"step": 2825
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 0.2569419401370357,
|
| 804 |
+
"grad_norm": 0.08487538993358612,
|
| 805 |
+
"learning_rate": 3.3830374779385545e-05,
|
| 806 |
+
"loss": 0.1037,
|
| 807 |
+
"step": 2850
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.25919581680490444,
|
| 811 |
+
"grad_norm": 0.11238917708396912,
|
| 812 |
+
"learning_rate": 3.3727731946628854e-05,
|
| 813 |
+
"loss": 0.0955,
|
| 814 |
+
"step": 2875
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 0.2614496934727732,
|
| 818 |
+
"grad_norm": 26.237930297851562,
|
| 819 |
+
"learning_rate": 3.3624400846761825e-05,
|
| 820 |
+
"loss": 0.2093,
|
| 821 |
+
"step": 2900
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 0.2637035701406419,
|
| 825 |
+
"grad_norm": 0.647821843624115,
|
| 826 |
+
"learning_rate": 3.352038666049442e-05,
|
| 827 |
+
"loss": 0.0679,
|
| 828 |
+
"step": 2925
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 0.26595744680851063,
|
| 832 |
+
"grad_norm": 41.48789596557617,
|
| 833 |
+
"learning_rate": 3.3415694602784474e-05,
|
| 834 |
+
"loss": 0.1589,
|
| 835 |
+
"step": 2950
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 0.2682113234763794,
|
| 839 |
+
"grad_norm": 0.25765570998191833,
|
| 840 |
+
"learning_rate": 3.331032992257627e-05,
|
| 841 |
+
"loss": 0.1298,
|
| 842 |
+
"step": 2975
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.27046520014424813,
|
| 846 |
+
"grad_norm": 14.330556869506836,
|
| 847 |
+
"learning_rate": 3.320429790253735e-05,
|
| 848 |
+
"loss": 0.1352,
|
| 849 |
+
"step": 3000
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.2727190768121168,
|
| 853 |
+
"grad_norm": 0.18978039920330048,
|
| 854 |
+
"learning_rate": 3.309760385879367e-05,
|
| 855 |
+
"loss": 0.1407,
|
| 856 |
+
"step": 3025
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 0.27497295347998557,
|
| 860 |
+
"grad_norm": 21.670555114746094,
|
| 861 |
+
"learning_rate": 3.299025314066304e-05,
|
| 862 |
+
"loss": 0.1238,
|
| 863 |
+
"step": 3050
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 0.2772268301478543,
|
| 867 |
+
"grad_norm": 26.64297866821289,
|
| 868 |
+
"learning_rate": 3.288225113038697e-05,
|
| 869 |
+
"loss": 0.1765,
|
| 870 |
+
"step": 3075
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 0.27948070681572307,
|
| 874 |
+
"grad_norm": 108.37388610839844,
|
| 875 |
+
"learning_rate": 3.277360324286077e-05,
|
| 876 |
+
"loss": 0.2013,
|
| 877 |
+
"step": 3100
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 0.28173458348359176,
|
| 881 |
+
"grad_norm": 0.5027166604995728,
|
| 882 |
+
"learning_rate": 3.266431492536212e-05,
|
| 883 |
+
"loss": 0.1436,
|
| 884 |
+
"step": 3125
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.2839884601514605,
|
| 888 |
+
"grad_norm": 18.417152404785156,
|
| 889 |
+
"learning_rate": 3.255439165727789e-05,
|
| 890 |
+
"loss": 0.0876,
|
| 891 |
+
"step": 3150
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.28624233681932926,
|
| 895 |
+
"grad_norm": 0.06666754186153412,
|
| 896 |
+
"learning_rate": 3.244383894982951e-05,
|
| 897 |
+
"loss": 0.1203,
|
| 898 |
+
"step": 3175
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 0.288496213487198,
|
| 902 |
+
"grad_norm": 0.04935978353023529,
|
| 903 |
+
"learning_rate": 3.233266234579655e-05,
|
| 904 |
+
"loss": 0.0907,
|
| 905 |
+
"step": 3200
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 0.2907500901550667,
|
| 909 |
+
"grad_norm": 0.052629631012678146,
|
| 910 |
+
"learning_rate": 3.222086741923891e-05,
|
| 911 |
+
"loss": 0.1407,
|
| 912 |
+
"step": 3225
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 0.29300396682293545,
|
| 916 |
+
"grad_norm": 0.1207326129078865,
|
| 917 |
+
"learning_rate": 3.2108459775217286e-05,
|
| 918 |
+
"loss": 0.1126,
|
| 919 |
+
"step": 3250
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 0.2952578434908042,
|
| 923 |
+
"grad_norm": 46.62343215942383,
|
| 924 |
+
"learning_rate": 3.199544504951222e-05,
|
| 925 |
+
"loss": 0.1659,
|
| 926 |
+
"step": 3275
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.29751172015867294,
|
| 930 |
+
"grad_norm": 8.397638320922852,
|
| 931 |
+
"learning_rate": 3.1881828908341484e-05,
|
| 932 |
+
"loss": 0.1641,
|
| 933 |
+
"step": 3300
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 0.29976559682654164,
|
| 937 |
+
"grad_norm": 1.138248324394226,
|
| 938 |
+
"learning_rate": 3.1767617048075995e-05,
|
| 939 |
+
"loss": 0.1057,
|
| 940 |
+
"step": 3325
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 0.3020194734944104,
|
| 944 |
+
"grad_norm": 0.34413909912109375,
|
| 945 |
+
"learning_rate": 3.165281519495424e-05,
|
| 946 |
+
"loss": 0.121,
|
| 947 |
+
"step": 3350
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 0.30427335016227913,
|
| 951 |
+
"grad_norm": 160.63751220703125,
|
| 952 |
+
"learning_rate": 3.153742910479518e-05,
|
| 953 |
+
"loss": 0.1572,
|
| 954 |
+
"step": 3375
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.3065272268301479,
|
| 958 |
+
"grad_norm": 33.884849548339844,
|
| 959 |
+
"learning_rate": 3.142146456270965e-05,
|
| 960 |
+
"loss": 0.1272,
|
| 961 |
+
"step": 3400
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 0.3087811034980166,
|
| 965 |
+
"grad_norm": 18.125350952148438,
|
| 966 |
+
"learning_rate": 3.130492738281031e-05,
|
| 967 |
+
"loss": 0.147,
|
| 968 |
+
"step": 3425
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 0.3110349801658853,
|
| 972 |
+
"grad_norm": 0.6469773054122925,
|
| 973 |
+
"learning_rate": 3.1187823407920194e-05,
|
| 974 |
+
"loss": 0.2482,
|
| 975 |
+
"step": 3450
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 0.31328885683375407,
|
| 979 |
+
"grad_norm": 123.72215270996094,
|
| 980 |
+
"learning_rate": 3.107015850927967e-05,
|
| 981 |
+
"loss": 0.116,
|
| 982 |
+
"step": 3475
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 0.3155427335016228,
|
| 986 |
+
"grad_norm": 22.499483108520508,
|
| 987 |
+
"learning_rate": 3.095193858625219e-05,
|
| 988 |
+
"loss": 0.1916,
|
| 989 |
+
"step": 3500
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.3177966101694915,
|
| 993 |
+
"grad_norm": 17.948232650756836,
|
| 994 |
+
"learning_rate": 3.083316956602842e-05,
|
| 995 |
+
"loss": 0.179,
|
| 996 |
+
"step": 3525
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 0.32005048683736026,
|
| 1000 |
+
"grad_norm": 9.72963809967041,
|
| 1001 |
+
"learning_rate": 3.071385740332912e-05,
|
| 1002 |
+
"loss": 0.1405,
|
| 1003 |
+
"step": 3550
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 0.322304363505229,
|
| 1007 |
+
"grad_norm": 10.847256660461426,
|
| 1008 |
+
"learning_rate": 3.059400808010656e-05,
|
| 1009 |
+
"loss": 0.1037,
|
| 1010 |
+
"step": 3575
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.3245582401730977,
|
| 1014 |
+
"grad_norm": 0.27193042635917664,
|
| 1015 |
+
"learning_rate": 3.047362760524465e-05,
|
| 1016 |
+
"loss": 0.0492,
|
| 1017 |
+
"step": 3600
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 0.32681211684096645,
|
| 1021 |
+
"grad_norm": 0.46222802996635437,
|
| 1022 |
+
"learning_rate": 3.0352722014257606e-05,
|
| 1023 |
+
"loss": 0.06,
|
| 1024 |
+
"step": 3625
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 0.3290659935088352,
|
| 1028 |
+
"grad_norm": 0.12345420569181442,
|
| 1029 |
+
"learning_rate": 3.023129736898742e-05,
|
| 1030 |
+
"loss": 0.1223,
|
| 1031 |
+
"step": 3650
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 0.33131987017670395,
|
| 1035 |
+
"grad_norm": 0.84328293800354,
|
| 1036 |
+
"learning_rate": 3.0109359757299877e-05,
|
| 1037 |
+
"loss": 0.1467,
|
| 1038 |
+
"step": 3675
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 0.33357374684457264,
|
| 1042 |
+
"grad_norm": 1.759286642074585,
|
| 1043 |
+
"learning_rate": 2.9986915292779354e-05,
|
| 1044 |
+
"loss": 0.1982,
|
| 1045 |
+
"step": 3700
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 0.3358276235124414,
|
| 1049 |
+
"grad_norm": 16.525177001953125,
|
| 1050 |
+
"learning_rate": 2.9863970114422315e-05,
|
| 1051 |
+
"loss": 0.1601,
|
| 1052 |
+
"step": 3725
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 0.33808150018031013,
|
| 1056 |
+
"grad_norm": 0.7221691012382507,
|
| 1057 |
+
"learning_rate": 2.974053038632949e-05,
|
| 1058 |
+
"loss": 0.1717,
|
| 1059 |
+
"step": 3750
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.3403353768481789,
|
| 1063 |
+
"grad_norm": 15.795534133911133,
|
| 1064 |
+
"learning_rate": 2.9616602297396865e-05,
|
| 1065 |
+
"loss": 0.136,
|
| 1066 |
+
"step": 3775
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 0.3425892535160476,
|
| 1070 |
+
"grad_norm": 18.442581176757812,
|
| 1071 |
+
"learning_rate": 2.949219206100535e-05,
|
| 1072 |
+
"loss": 0.1343,
|
| 1073 |
+
"step": 3800
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 0.3448431301839163,
|
| 1077 |
+
"grad_norm": 0.0298846997320652,
|
| 1078 |
+
"learning_rate": 2.9367305914709267e-05,
|
| 1079 |
+
"loss": 0.1249,
|
| 1080 |
+
"step": 3825
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 0.3470970068517851,
|
| 1084 |
+
"grad_norm": 0.21138563752174377,
|
| 1085 |
+
"learning_rate": 2.9241950119923642e-05,
|
| 1086 |
+
"loss": 0.1854,
|
| 1087 |
+
"step": 3850
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 0.3493508835196538,
|
| 1091 |
+
"grad_norm": 0.4857686460018158,
|
| 1092 |
+
"learning_rate": 2.911613096161027e-05,
|
| 1093 |
+
"loss": 0.1047,
|
| 1094 |
+
"step": 3875
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 0.3516047601875225,
|
| 1098 |
+
"grad_norm": 21.95134162902832,
|
| 1099 |
+
"learning_rate": 2.8989854747962574e-05,
|
| 1100 |
+
"loss": 0.1034,
|
| 1101 |
+
"step": 3900
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 0.35385863685539126,
|
| 1105 |
+
"grad_norm": 0.23518282175064087,
|
| 1106 |
+
"learning_rate": 2.8863127810089363e-05,
|
| 1107 |
+
"loss": 0.1115,
|
| 1108 |
+
"step": 3925
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 0.35611251352326,
|
| 1112 |
+
"grad_norm": 0.21235395967960358,
|
| 1113 |
+
"learning_rate": 2.8735956501697398e-05,
|
| 1114 |
+
"loss": 0.124,
|
| 1115 |
+
"step": 3950
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 0.35836639019112876,
|
| 1119 |
+
"grad_norm": 0.4174875319004059,
|
| 1120 |
+
"learning_rate": 2.8608347198772856e-05,
|
| 1121 |
+
"loss": 0.1672,
|
| 1122 |
+
"step": 3975
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 0.36062026685899745,
|
| 1126 |
+
"grad_norm": 0.6259788870811462,
|
| 1127 |
+
"learning_rate": 2.8480306299261618e-05,
|
| 1128 |
+
"loss": 0.1254,
|
| 1129 |
+
"step": 4000
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 0.3628741435268662,
|
| 1133 |
+
"grad_norm": 111.80479431152344,
|
| 1134 |
+
"learning_rate": 2.8351840222748524e-05,
|
| 1135 |
+
"loss": 0.1683,
|
| 1136 |
+
"step": 4025
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 0.36512802019473495,
|
| 1140 |
+
"grad_norm": 0.35139939188957214,
|
| 1141 |
+
"learning_rate": 2.8222955410135512e-05,
|
| 1142 |
+
"loss": 0.0963,
|
| 1143 |
+
"step": 4050
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 0.3673818968626037,
|
| 1147 |
+
"grad_norm": 0.15862537920475006,
|
| 1148 |
+
"learning_rate": 2.8093658323318673e-05,
|
| 1149 |
+
"loss": 0.0996,
|
| 1150 |
+
"step": 4075
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 0.3696357735304724,
|
| 1154 |
+
"grad_norm": 0.2985376715660095,
|
| 1155 |
+
"learning_rate": 2.796395544486432e-05,
|
| 1156 |
+
"loss": 0.1072,
|
| 1157 |
+
"step": 4100
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 0.37188965019834114,
|
| 1161 |
+
"grad_norm": 0.13498641550540924,
|
| 1162 |
+
"learning_rate": 2.783385327768388e-05,
|
| 1163 |
+
"loss": 0.1739,
|
| 1164 |
+
"step": 4125
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 0.3741435268662099,
|
| 1168 |
+
"grad_norm": 0.18899990618228912,
|
| 1169 |
+
"learning_rate": 2.770335834470796e-05,
|
| 1170 |
+
"loss": 0.1572,
|
| 1171 |
+
"step": 4150
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 0.37639740353407863,
|
| 1175 |
+
"grad_norm": 6.3806915283203125,
|
| 1176 |
+
"learning_rate": 2.757247718855924e-05,
|
| 1177 |
+
"loss": 0.165,
|
| 1178 |
+
"step": 4175
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 0.3786512802019473,
|
| 1182 |
+
"grad_norm": 1.358575701713562,
|
| 1183 |
+
"learning_rate": 2.7441216371224472e-05,
|
| 1184 |
+
"loss": 0.0631,
|
| 1185 |
+
"step": 4200
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.3809051568698161,
|
| 1189 |
+
"grad_norm": 21.20967674255371,
|
| 1190 |
+
"learning_rate": 2.7309582473725467e-05,
|
| 1191 |
+
"loss": 0.2438,
|
| 1192 |
+
"step": 4225
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 0.3831590335376848,
|
| 1196 |
+
"grad_norm": 0.9558500051498413,
|
| 1197 |
+
"learning_rate": 2.717758209578917e-05,
|
| 1198 |
+
"loss": 0.1069,
|
| 1199 |
+
"step": 4250
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 0.38541291020555357,
|
| 1203 |
+
"grad_norm": 18.68955421447754,
|
| 1204 |
+
"learning_rate": 2.7045221855516728e-05,
|
| 1205 |
+
"loss": 0.1784,
|
| 1206 |
+
"step": 4275
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 0.38766678687342226,
|
| 1210 |
+
"grad_norm": 0.34952789545059204,
|
| 1211 |
+
"learning_rate": 2.6912508389051715e-05,
|
| 1212 |
+
"loss": 0.1194,
|
| 1213 |
+
"step": 4300
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 0.389920663541291,
|
| 1217 |
+
"grad_norm": 21.433164596557617,
|
| 1218 |
+
"learning_rate": 2.67794483502474e-05,
|
| 1219 |
+
"loss": 0.0592,
|
| 1220 |
+
"step": 4325
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 0.39217454020915976,
|
| 1224 |
+
"grad_norm": 0.13010774552822113,
|
| 1225 |
+
"learning_rate": 2.664604841033313e-05,
|
| 1226 |
+
"loss": 0.1116,
|
| 1227 |
+
"step": 4350
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.3944284168770285,
|
| 1231 |
+
"grad_norm": 23.884571075439453,
|
| 1232 |
+
"learning_rate": 2.6512315257579905e-05,
|
| 1233 |
+
"loss": 0.1544,
|
| 1234 |
+
"step": 4375
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 0.3966822935448972,
|
| 1238 |
+
"grad_norm": 0.5002589225769043,
|
| 1239 |
+
"learning_rate": 2.6378255596964977e-05,
|
| 1240 |
+
"loss": 0.1064,
|
| 1241 |
+
"step": 4400
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 0.39893617021276595,
|
| 1245 |
+
"grad_norm": 1.1411672830581665,
|
| 1246 |
+
"learning_rate": 2.6243876149835733e-05,
|
| 1247 |
+
"loss": 0.1181,
|
| 1248 |
+
"step": 4425
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 0.4011900468806347,
|
| 1252 |
+
"grad_norm": 1.1338392496109009,
|
| 1253 |
+
"learning_rate": 2.6109183653572686e-05,
|
| 1254 |
+
"loss": 0.1916,
|
| 1255 |
+
"step": 4450
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 0.40344392354850345,
|
| 1259 |
+
"grad_norm": 41.243370056152344,
|
| 1260 |
+
"learning_rate": 2.597418486125169e-05,
|
| 1261 |
+
"loss": 0.1372,
|
| 1262 |
+
"step": 4475
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 0.40569780021637214,
|
| 1266 |
+
"grad_norm": 12.509779930114746,
|
| 1267 |
+
"learning_rate": 2.5838886541305383e-05,
|
| 1268 |
+
"loss": 0.1757,
|
| 1269 |
+
"step": 4500
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 0.4079516768842409,
|
| 1273 |
+
"grad_norm": 5.6783037185668945,
|
| 1274 |
+
"learning_rate": 2.570329547718378e-05,
|
| 1275 |
+
"loss": 0.131,
|
| 1276 |
+
"step": 4525
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 0.41020555355210964,
|
| 1280 |
+
"grad_norm": 0.28341442346572876,
|
| 1281 |
+
"learning_rate": 2.5567418467014225e-05,
|
| 1282 |
+
"loss": 0.1154,
|
| 1283 |
+
"step": 4550
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 0.4124594302199784,
|
| 1287 |
+
"grad_norm": 0.15925537049770355,
|
| 1288 |
+
"learning_rate": 2.543126232326054e-05,
|
| 1289 |
+
"loss": 0.1166,
|
| 1290 |
+
"step": 4575
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 0.4147133068878471,
|
| 1294 |
+
"grad_norm": 20.447715759277344,
|
| 1295 |
+
"learning_rate": 2.529483387238145e-05,
|
| 1296 |
+
"loss": 0.1239,
|
| 1297 |
+
"step": 4600
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 0.4169671835557158,
|
| 1301 |
+
"grad_norm": 0.26573672890663147,
|
| 1302 |
+
"learning_rate": 2.515813995448835e-05,
|
| 1303 |
+
"loss": 0.1495,
|
| 1304 |
+
"step": 4625
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 0.4192210602235846,
|
| 1308 |
+
"grad_norm": 11.98095989227295,
|
| 1309 |
+
"learning_rate": 2.5021187423002336e-05,
|
| 1310 |
+
"loss": 0.1204,
|
| 1311 |
+
"step": 4650
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 0.4214749368914533,
|
| 1315 |
+
"grad_norm": 0.027674587443470955,
|
| 1316 |
+
"learning_rate": 2.4883983144310634e-05,
|
| 1317 |
+
"loss": 0.1109,
|
| 1318 |
+
"step": 4675
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 0.423728813559322,
|
| 1322 |
+
"grad_norm": 0.20825082063674927,
|
| 1323 |
+
"learning_rate": 2.4746533997422303e-05,
|
| 1324 |
+
"loss": 0.1738,
|
| 1325 |
+
"step": 4700
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 0.42598269022719076,
|
| 1329 |
+
"grad_norm": 0.0698222666978836,
|
| 1330 |
+
"learning_rate": 2.460884687362334e-05,
|
| 1331 |
+
"loss": 0.1753,
|
| 1332 |
+
"step": 4725
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 0.4282365668950595,
|
| 1336 |
+
"grad_norm": 12.488253593444824,
|
| 1337 |
+
"learning_rate": 2.447092867613122e-05,
|
| 1338 |
+
"loss": 0.1378,
|
| 1339 |
+
"step": 4750
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 0.43049044356292826,
|
| 1343 |
+
"grad_norm": 0.14049705862998962,
|
| 1344 |
+
"learning_rate": 2.433278631974873e-05,
|
| 1345 |
+
"loss": 0.062,
|
| 1346 |
+
"step": 4775
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 0.43274432023079695,
|
| 1350 |
+
"grad_norm": 0.07813134789466858,
|
| 1351 |
+
"learning_rate": 2.419442673051732e-05,
|
| 1352 |
+
"loss": 0.1287,
|
| 1353 |
+
"step": 4800
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 0.4349981968986657,
|
| 1357 |
+
"grad_norm": 5.596477508544922,
|
| 1358 |
+
"learning_rate": 2.4055856845369836e-05,
|
| 1359 |
+
"loss": 0.2271,
|
| 1360 |
+
"step": 4825
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 0.43725207356653445,
|
| 1364 |
+
"grad_norm": 1.3635306358337402,
|
| 1365 |
+
"learning_rate": 2.3917083611782738e-05,
|
| 1366 |
+
"loss": 0.0879,
|
| 1367 |
+
"step": 4850
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 0.4395059502344032,
|
| 1371 |
+
"grad_norm": 25.128684997558594,
|
| 1372 |
+
"learning_rate": 2.3778113987427734e-05,
|
| 1373 |
+
"loss": 0.1598,
|
| 1374 |
+
"step": 4875
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 0.4417598269022719,
|
| 1378 |
+
"grad_norm": 7.198594570159912,
|
| 1379 |
+
"learning_rate": 2.363895493982302e-05,
|
| 1380 |
+
"loss": 0.1722,
|
| 1381 |
+
"step": 4900
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 0.44401370357014064,
|
| 1385 |
+
"grad_norm": 11.872293472290039,
|
| 1386 |
+
"learning_rate": 2.349961344598386e-05,
|
| 1387 |
+
"loss": 0.1307,
|
| 1388 |
+
"step": 4925
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 0.4462675802380094,
|
| 1392 |
+
"grad_norm": 0.17397132515907288,
|
| 1393 |
+
"learning_rate": 2.3360096492072853e-05,
|
| 1394 |
+
"loss": 0.0715,
|
| 1395 |
+
"step": 4950
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 0.44852145690587814,
|
| 1399 |
+
"grad_norm": 1.0778571367263794,
|
| 1400 |
+
"learning_rate": 2.322041107304961e-05,
|
| 1401 |
+
"loss": 0.1564,
|
| 1402 |
+
"step": 4975
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 0.45077533357374683,
|
| 1406 |
+
"grad_norm": 16.016172409057617,
|
| 1407 |
+
"learning_rate": 2.3080564192320103e-05,
|
| 1408 |
+
"loss": 0.1262,
|
| 1409 |
+
"step": 5000
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 0.4530292102416156,
|
| 1413 |
+
"grad_norm": 0.13017550110816956,
|
| 1414 |
+
"learning_rate": 2.294056286138548e-05,
|
| 1415 |
+
"loss": 0.1126,
|
| 1416 |
+
"step": 5025
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 0.4552830869094843,
|
| 1420 |
+
"grad_norm": 0.2517072260379791,
|
| 1421 |
+
"learning_rate": 2.2800414099490577e-05,
|
| 1422 |
+
"loss": 0.0807,
|
| 1423 |
+
"step": 5050
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 0.4575369635773531,
|
| 1427 |
+
"grad_norm": 43.9896354675293,
|
| 1428 |
+
"learning_rate": 2.266012493327197e-05,
|
| 1429 |
+
"loss": 0.1007,
|
| 1430 |
+
"step": 5075
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 0.45979084024522177,
|
| 1434 |
+
"grad_norm": 0.4000146687030792,
|
| 1435 |
+
"learning_rate": 2.251970239640569e-05,
|
| 1436 |
+
"loss": 0.145,
|
| 1437 |
+
"step": 5100
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 0.4620447169130905,
|
| 1441 |
+
"grad_norm": 0.48227617144584656,
|
| 1442 |
+
"learning_rate": 2.2379153529254536e-05,
|
| 1443 |
+
"loss": 0.2186,
|
| 1444 |
+
"step": 5125
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 0.46429859358095926,
|
| 1448 |
+
"grad_norm": 25.047019958496094,
|
| 1449 |
+
"learning_rate": 2.2238485378515175e-05,
|
| 1450 |
+
"loss": 0.1343,
|
| 1451 |
+
"step": 5150
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 0.466552470248828,
|
| 1455 |
+
"grad_norm": 1.820315957069397,
|
| 1456 |
+
"learning_rate": 2.2097704996864766e-05,
|
| 1457 |
+
"loss": 0.1137,
|
| 1458 |
+
"step": 5175
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 0.4688063469166967,
|
| 1462 |
+
"grad_norm": 0.05017359182238579,
|
| 1463 |
+
"learning_rate": 2.1956819442607395e-05,
|
| 1464 |
+
"loss": 0.06,
|
| 1465 |
+
"step": 5200
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 0.47106022358456545,
|
| 1469 |
+
"grad_norm": 0.4226453900337219,
|
| 1470 |
+
"learning_rate": 2.1815835779320183e-05,
|
| 1471 |
+
"loss": 0.1164,
|
| 1472 |
+
"step": 5225
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 0.4733141002524342,
|
| 1476 |
+
"grad_norm": 4.943594455718994,
|
| 1477 |
+
"learning_rate": 2.1674761075499138e-05,
|
| 1478 |
+
"loss": 0.0789,
|
| 1479 |
+
"step": 5250
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 0.47556797692030295,
|
| 1483 |
+
"grad_norm": 9.37826156616211,
|
| 1484 |
+
"learning_rate": 2.1533602404204775e-05,
|
| 1485 |
+
"loss": 0.138,
|
| 1486 |
+
"step": 5275
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 0.47782185358817164,
|
| 1490 |
+
"grad_norm": 0.3179582357406616,
|
| 1491 |
+
"learning_rate": 2.1392366842707495e-05,
|
| 1492 |
+
"loss": 0.1992,
|
| 1493 |
+
"step": 5300
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 0.4800757302560404,
|
| 1497 |
+
"grad_norm": 0.40917298197746277,
|
| 1498 |
+
"learning_rate": 2.1251061472132703e-05,
|
| 1499 |
+
"loss": 0.1084,
|
| 1500 |
+
"step": 5325
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 0.48232960692390914,
|
| 1504 |
+
"grad_norm": 18.944381713867188,
|
| 1505 |
+
"learning_rate": 2.110969337710586e-05,
|
| 1506 |
+
"loss": 0.1311,
|
| 1507 |
+
"step": 5350
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 0.48458348359177783,
|
| 1511 |
+
"grad_norm": 17.96181869506836,
|
| 1512 |
+
"learning_rate": 2.0968269645397234e-05,
|
| 1513 |
+
"loss": 0.1786,
|
| 1514 |
+
"step": 5375
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 0.4868373602596466,
|
| 1518 |
+
"grad_norm": 9.519824981689453,
|
| 1519 |
+
"learning_rate": 2.0826797367566533e-05,
|
| 1520 |
+
"loss": 0.1502,
|
| 1521 |
+
"step": 5400
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 0.4890912369275153,
|
| 1525 |
+
"grad_norm": 0.17551390826702118,
|
| 1526 |
+
"learning_rate": 2.0685283636607432e-05,
|
| 1527 |
+
"loss": 0.0475,
|
| 1528 |
+
"step": 5425
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 0.4913451135953841,
|
| 1532 |
+
"grad_norm": 0.018554313108325005,
|
| 1533 |
+
"learning_rate": 2.0543735547591934e-05,
|
| 1534 |
+
"loss": 0.1057,
|
| 1535 |
+
"step": 5450
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 0.49359899026325277,
|
| 1539 |
+
"grad_norm": 0.16615626215934753,
|
| 1540 |
+
"learning_rate": 2.0402160197314678e-05,
|
| 1541 |
+
"loss": 0.231,
|
| 1542 |
+
"step": 5475
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 0.4958528669311215,
|
| 1546 |
+
"grad_norm": 0.21435368061065674,
|
| 1547 |
+
"learning_rate": 2.0260564683937053e-05,
|
| 1548 |
+
"loss": 0.0496,
|
| 1549 |
+
"step": 5500
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 0.49810674359899026,
|
| 1553 |
+
"grad_norm": 0.2134808897972107,
|
| 1554 |
+
"learning_rate": 2.0118956106631414e-05,
|
| 1555 |
+
"loss": 0.1022,
|
| 1556 |
+
"step": 5525
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 0.500360620266859,
|
| 1560 |
+
"grad_norm": 0.1332341432571411,
|
| 1561 |
+
"learning_rate": 1.997734156522507e-05,
|
| 1562 |
+
"loss": 0.1345,
|
| 1563 |
+
"step": 5550
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 0.5026144969347277,
|
| 1567 |
+
"grad_norm": 80.57794189453125,
|
| 1568 |
+
"learning_rate": 1.9835728159844364e-05,
|
| 1569 |
+
"loss": 0.0603,
|
| 1570 |
+
"step": 5575
|
| 1571 |
+
},
|
| 1572 |
+
{
|
| 1573 |
+
"epoch": 0.5048683736025965,
|
| 1574 |
+
"grad_norm": 0.517909824848175,
|
| 1575 |
+
"learning_rate": 1.969412299055867e-05,
|
| 1576 |
+
"loss": 0.179,
|
| 1577 |
+
"step": 5600
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 0.5071222502704652,
|
| 1581 |
+
"grad_norm": 0.4541005790233612,
|
| 1582 |
+
"learning_rate": 1.9552533157024462e-05,
|
| 1583 |
+
"loss": 0.0843,
|
| 1584 |
+
"step": 5625
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 0.5093761269383339,
|
| 1588 |
+
"grad_norm": 0.42274990677833557,
|
| 1589 |
+
"learning_rate": 1.941096575812928e-05,
|
| 1590 |
+
"loss": 0.1301,
|
| 1591 |
+
"step": 5650
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 0.5116300036062027,
|
| 1595 |
+
"grad_norm": 18.68648338317871,
|
| 1596 |
+
"learning_rate": 1.9269427891635886e-05,
|
| 1597 |
+
"loss": 0.114,
|
| 1598 |
+
"step": 5675
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 0.5138838802740714,
|
| 1602 |
+
"grad_norm": 0.12541067600250244,
|
| 1603 |
+
"learning_rate": 1.912792665382638e-05,
|
| 1604 |
+
"loss": 0.053,
|
| 1605 |
+
"step": 5700
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 0.5161377569419401,
|
| 1609 |
+
"grad_norm": 27.757553100585938,
|
| 1610 |
+
"learning_rate": 1.8986469139146406e-05,
|
| 1611 |
+
"loss": 0.2328,
|
| 1612 |
+
"step": 5725
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 0.5183916336098089,
|
| 1616 |
+
"grad_norm": 0.1179078221321106,
|
| 1617 |
+
"learning_rate": 1.8845062439849437e-05,
|
| 1618 |
+
"loss": 0.2118,
|
| 1619 |
+
"step": 5750
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 0.5206455102776776,
|
| 1623 |
+
"grad_norm": 4.539048194885254,
|
| 1624 |
+
"learning_rate": 1.8703713645641253e-05,
|
| 1625 |
+
"loss": 0.1236,
|
| 1626 |
+
"step": 5775
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 0.5228993869455464,
|
| 1630 |
+
"grad_norm": 0.8501512408256531,
|
| 1631 |
+
"learning_rate": 1.8562429843324423e-05,
|
| 1632 |
+
"loss": 0.1199,
|
| 1633 |
+
"step": 5800
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 0.5251532636134151,
|
| 1637 |
+
"grad_norm": 1.2078758478164673,
|
| 1638 |
+
"learning_rate": 1.8421218116443044e-05,
|
| 1639 |
+
"loss": 0.1576,
|
| 1640 |
+
"step": 5825
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 0.5274071402812838,
|
| 1644 |
+
"grad_norm": 0.25490689277648926,
|
| 1645 |
+
"learning_rate": 1.8280085544927534e-05,
|
| 1646 |
+
"loss": 0.1667,
|
| 1647 |
+
"step": 5850
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 0.5296610169491526,
|
| 1651 |
+
"grad_norm": 0.6682247519493103,
|
| 1652 |
+
"learning_rate": 1.8139039204739712e-05,
|
| 1653 |
+
"loss": 0.097,
|
| 1654 |
+
"step": 5875
|
| 1655 |
+
},
|
| 1656 |
+
{
|
| 1657 |
+
"epoch": 0.5319148936170213,
|
| 1658 |
+
"grad_norm": 11.402225494384766,
|
| 1659 |
+
"learning_rate": 1.7998086167518033e-05,
|
| 1660 |
+
"loss": 0.1525,
|
| 1661 |
+
"step": 5900
|
| 1662 |
+
},
|
| 1663 |
+
{
|
| 1664 |
+
"epoch": 0.53416877028489,
|
| 1665 |
+
"grad_norm": 0.09847287833690643,
|
| 1666 |
+
"learning_rate": 1.7857233500223018e-05,
|
| 1667 |
+
"loss": 0.1392,
|
| 1668 |
+
"step": 5925
|
| 1669 |
+
},
|
| 1670 |
+
{
|
| 1671 |
+
"epoch": 0.5364226469527588,
|
| 1672 |
+
"grad_norm": 0.20688538253307343,
|
| 1673 |
+
"learning_rate": 1.7716488264782915e-05,
|
| 1674 |
+
"loss": 0.1541,
|
| 1675 |
+
"step": 5950
|
| 1676 |
+
},
|
| 1677 |
+
{
|
| 1678 |
+
"epoch": 0.5386765236206275,
|
| 1679 |
+
"grad_norm": 0.2824791371822357,
|
| 1680 |
+
"learning_rate": 1.75758575177397e-05,
|
| 1681 |
+
"loss": 0.1382,
|
| 1682 |
+
"step": 5975
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 0.5409304002884963,
|
| 1686 |
+
"grad_norm": 0.19064374268054962,
|
| 1687 |
+
"learning_rate": 1.743534830989522e-05,
|
| 1688 |
+
"loss": 0.1389,
|
| 1689 |
+
"step": 6000
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 0.543184276956365,
|
| 1693 |
+
"grad_norm": 0.35777968168258667,
|
| 1694 |
+
"learning_rate": 1.7294967685957733e-05,
|
| 1695 |
+
"loss": 0.1635,
|
| 1696 |
+
"step": 6025
|
| 1697 |
+
},
|
| 1698 |
+
{
|
| 1699 |
+
"epoch": 0.5454381536242336,
|
| 1700 |
+
"grad_norm": 0.03058139607310295,
|
| 1701 |
+
"learning_rate": 1.7154722684188668e-05,
|
| 1702 |
+
"loss": 0.103,
|
| 1703 |
+
"step": 6050
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 0.5476920302921024,
|
| 1707 |
+
"grad_norm": 18.864845275878906,
|
| 1708 |
+
"learning_rate": 1.7014620336049774e-05,
|
| 1709 |
+
"loss": 0.1438,
|
| 1710 |
+
"step": 6075
|
| 1711 |
+
},
|
| 1712 |
+
{
|
| 1713 |
+
"epoch": 0.5499459069599711,
|
| 1714 |
+
"grad_norm": 0.5285549163818359,
|
| 1715 |
+
"learning_rate": 1.6874667665850583e-05,
|
| 1716 |
+
"loss": 0.11,
|
| 1717 |
+
"step": 6100
|
| 1718 |
+
},
|
| 1719 |
+
{
|
| 1720 |
+
"epoch": 0.5521997836278398,
|
| 1721 |
+
"grad_norm": 0.1905856728553772,
|
| 1722 |
+
"learning_rate": 1.673487169039623e-05,
|
| 1723 |
+
"loss": 0.1059,
|
| 1724 |
+
"step": 6125
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 0.5544536602957086,
|
| 1728 |
+
"grad_norm": 0.28477397561073303,
|
| 1729 |
+
"learning_rate": 1.6595239418635616e-05,
|
| 1730 |
+
"loss": 0.138,
|
| 1731 |
+
"step": 6150
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 0.5567075369635773,
|
| 1735 |
+
"grad_norm": 10.22347640991211,
|
| 1736 |
+
"learning_rate": 1.6455777851310074e-05,
|
| 1737 |
+
"loss": 0.1124,
|
| 1738 |
+
"step": 6175
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 0.5589614136314461,
|
| 1742 |
+
"grad_norm": 68.21721649169922,
|
| 1743 |
+
"learning_rate": 1.6316493980602307e-05,
|
| 1744 |
+
"loss": 0.0738,
|
| 1745 |
+
"step": 6200
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 0.5612152902993148,
|
| 1749 |
+
"grad_norm": 9.33055591583252,
|
| 1750 |
+
"learning_rate": 1.6177394789785855e-05,
|
| 1751 |
+
"loss": 0.1692,
|
| 1752 |
+
"step": 6225
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 0.5634691669671835,
|
| 1756 |
+
"grad_norm": 41.284515380859375,
|
| 1757 |
+
"learning_rate": 1.6038487252874964e-05,
|
| 1758 |
+
"loss": 0.1048,
|
| 1759 |
+
"step": 6250
|
| 1760 |
+
},
|
| 1761 |
+
{
|
| 1762 |
+
"epoch": 0.5657230436350523,
|
| 1763 |
+
"grad_norm": 71.44107818603516,
|
| 1764 |
+
"learning_rate": 1.5899778334274908e-05,
|
| 1765 |
+
"loss": 0.1088,
|
| 1766 |
+
"step": 6275
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 0.567976920302921,
|
| 1770 |
+
"grad_norm": 16.360801696777344,
|
| 1771 |
+
"learning_rate": 1.5761274988432877e-05,
|
| 1772 |
+
"loss": 0.2039,
|
| 1773 |
+
"step": 6300
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 0.5702307969707897,
|
| 1777 |
+
"grad_norm": 1.811718463897705,
|
| 1778 |
+
"learning_rate": 1.562298415948924e-05,
|
| 1779 |
+
"loss": 0.1092,
|
| 1780 |
+
"step": 6325
|
| 1781 |
+
},
|
| 1782 |
+
{
|
| 1783 |
+
"epoch": 0.5724846736386585,
|
| 1784 |
+
"grad_norm": 56.19565200805664,
|
| 1785 |
+
"learning_rate": 1.5484912780929392e-05,
|
| 1786 |
+
"loss": 0.0783,
|
| 1787 |
+
"step": 6350
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 0.5747385503065272,
|
| 1791 |
+
"grad_norm": 0.043573249131441116,
|
| 1792 |
+
"learning_rate": 1.534706777523619e-05,
|
| 1793 |
+
"loss": 0.0916,
|
| 1794 |
+
"step": 6375
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 0.576992426974396,
|
| 1798 |
+
"grad_norm": 0.02965572103857994,
|
| 1799 |
+
"learning_rate": 1.5209456053542794e-05,
|
| 1800 |
+
"loss": 0.108,
|
| 1801 |
+
"step": 6400
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 0.5792463036422647,
|
| 1805 |
+
"grad_norm": 0.08491244167089462,
|
| 1806 |
+
"learning_rate": 1.5072084515286245e-05,
|
| 1807 |
+
"loss": 0.1407,
|
| 1808 |
+
"step": 6425
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 0.5815001803101334,
|
| 1812 |
+
"grad_norm": 0.10835108906030655,
|
| 1813 |
+
"learning_rate": 1.4934960047861489e-05,
|
| 1814 |
+
"loss": 0.1707,
|
| 1815 |
+
"step": 6450
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 0.5837540569780022,
|
| 1819 |
+
"grad_norm": 0.23179195821285248,
|
| 1820 |
+
"learning_rate": 1.4798089526276074e-05,
|
| 1821 |
+
"loss": 0.2206,
|
| 1822 |
+
"step": 6475
|
| 1823 |
+
},
|
| 1824 |
+
{
|
| 1825 |
+
"epoch": 0.5860079336458709,
|
| 1826 |
+
"grad_norm": 0.06362619251012802,
|
| 1827 |
+
"learning_rate": 1.4661479812805499e-05,
|
| 1828 |
+
"loss": 0.0975,
|
| 1829 |
+
"step": 6500
|
| 1830 |
+
},
|
| 1831 |
+
{
|
| 1832 |
+
"epoch": 0.5882618103137396,
|
| 1833 |
+
"grad_norm": 0.0836687833070755,
|
| 1834 |
+
"learning_rate": 1.4525137756649113e-05,
|
| 1835 |
+
"loss": 0.1178,
|
| 1836 |
+
"step": 6525
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 0.5905156869816084,
|
| 1840 |
+
"grad_norm": 0.1414812058210373,
|
| 1841 |
+
"learning_rate": 1.4389070193586721e-05,
|
| 1842 |
+
"loss": 0.1111,
|
| 1843 |
+
"step": 6550
|
| 1844 |
+
},
|
| 1845 |
+
{
|
| 1846 |
+
"epoch": 0.5927695636494771,
|
| 1847 |
+
"grad_norm": 0.09423329681158066,
|
| 1848 |
+
"learning_rate": 1.4253283945635907e-05,
|
| 1849 |
+
"loss": 0.0827,
|
| 1850 |
+
"step": 6575
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 0.5950234403173459,
|
| 1854 |
+
"grad_norm": 5.506004333496094,
|
| 1855 |
+
"learning_rate": 1.4117785820709926e-05,
|
| 1856 |
+
"loss": 0.1063,
|
| 1857 |
+
"step": 6600
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"epoch": 0.5972773169852146,
|
| 1861 |
+
"grad_norm": 0.020630862563848495,
|
| 1862 |
+
"learning_rate": 1.3982582612276439e-05,
|
| 1863 |
+
"loss": 0.1304,
|
| 1864 |
+
"step": 6625
|
| 1865 |
+
},
|
| 1866 |
+
{
|
| 1867 |
+
"epoch": 0.5995311936530833,
|
| 1868 |
+
"grad_norm": 0.8215897679328918,
|
| 1869 |
+
"learning_rate": 1.3847681099016872e-05,
|
| 1870 |
+
"loss": 0.0835,
|
| 1871 |
+
"step": 6650
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 0.6017850703209521,
|
| 1875 |
+
"grad_norm": 0.021800508722662926,
|
| 1876 |
+
"learning_rate": 1.371308804448655e-05,
|
| 1877 |
+
"loss": 0.0606,
|
| 1878 |
+
"step": 6675
|
| 1879 |
+
},
|
| 1880 |
+
{
|
| 1881 |
+
"epoch": 0.6040389469888208,
|
| 1882 |
+
"grad_norm": 0.20198146998882294,
|
| 1883 |
+
"learning_rate": 1.3578810196775639e-05,
|
| 1884 |
+
"loss": 0.085,
|
| 1885 |
+
"step": 6700
|
| 1886 |
+
},
|
| 1887 |
+
{
|
| 1888 |
+
"epoch": 0.6062928236566895,
|
| 1889 |
+
"grad_norm": 0.21143490076065063,
|
| 1890 |
+
"learning_rate": 1.3444854288170763e-05,
|
| 1891 |
+
"loss": 0.0751,
|
| 1892 |
+
"step": 6725
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 0.6085467003245583,
|
| 1896 |
+
"grad_norm": 0.243058979511261,
|
| 1897 |
+
"learning_rate": 1.3311227034817482e-05,
|
| 1898 |
+
"loss": 0.2111,
|
| 1899 |
+
"step": 6750
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 0.610800576992427,
|
| 1903 |
+
"grad_norm": 0.13287846744060516,
|
| 1904 |
+
"learning_rate": 1.3177935136383608e-05,
|
| 1905 |
+
"loss": 0.1739,
|
| 1906 |
+
"step": 6775
|
| 1907 |
+
},
|
| 1908 |
+
{
|
| 1909 |
+
"epoch": 0.6130544536602958,
|
| 1910 |
+
"grad_norm": 0.47691506147384644,
|
| 1911 |
+
"learning_rate": 1.3044985275723221e-05,
|
| 1912 |
+
"loss": 0.0745,
|
| 1913 |
+
"step": 6800
|
| 1914 |
+
},
|
| 1915 |
+
{
|
| 1916 |
+
"epoch": 0.6153083303281645,
|
| 1917 |
+
"grad_norm": 40.97590637207031,
|
| 1918 |
+
"learning_rate": 1.2912384118541695e-05,
|
| 1919 |
+
"loss": 0.1734,
|
| 1920 |
+
"step": 6825
|
| 1921 |
+
},
|
| 1922 |
+
{
|
| 1923 |
+
"epoch": 0.6175622069960331,
|
| 1924 |
+
"grad_norm": 0.4189409613609314,
|
| 1925 |
+
"learning_rate": 1.2780138313061448e-05,
|
| 1926 |
+
"loss": 0.1325,
|
| 1927 |
+
"step": 6850
|
| 1928 |
+
},
|
| 1929 |
+
{
|
| 1930 |
+
"epoch": 0.619816083663902,
|
| 1931 |
+
"grad_norm": 0.6649547815322876,
|
| 1932 |
+
"learning_rate": 1.264825448968862e-05,
|
| 1933 |
+
"loss": 0.0759,
|
| 1934 |
+
"step": 6875
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 0.6220699603317706,
|
| 1938 |
+
"grad_norm": 13.50507926940918,
|
| 1939 |
+
"learning_rate": 1.2516739260680693e-05,
|
| 1940 |
+
"loss": 0.1746,
|
| 1941 |
+
"step": 6900
|
| 1942 |
+
},
|
| 1943 |
+
{
|
| 1944 |
+
"epoch": 0.6243238369996393,
|
| 1945 |
+
"grad_norm": 9.8441801071167,
|
| 1946 |
+
"learning_rate": 1.2385599219814907e-05,
|
| 1947 |
+
"loss": 0.1033,
|
| 1948 |
+
"step": 6925
|
| 1949 |
+
},
|
| 1950 |
+
{
|
| 1951 |
+
"epoch": 0.6265777136675081,
|
| 1952 |
+
"grad_norm": 0.06765761971473694,
|
| 1953 |
+
"learning_rate": 1.225484094205771e-05,
|
| 1954 |
+
"loss": 0.23,
|
| 1955 |
+
"step": 6950
|
| 1956 |
+
},
|
| 1957 |
+
{
|
| 1958 |
+
"epoch": 0.6288315903353768,
|
| 1959 |
+
"grad_norm": 0.23089264333248138,
|
| 1960 |
+
"learning_rate": 1.2124470983235117e-05,
|
| 1961 |
+
"loss": 0.1995,
|
| 1962 |
+
"step": 6975
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 0.6310854670032456,
|
| 1966 |
+
"grad_norm": 0.40788424015045166,
|
| 1967 |
+
"learning_rate": 1.1994495879703962e-05,
|
| 1968 |
+
"loss": 0.0751,
|
| 1969 |
+
"step": 7000
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 0.6333393436711143,
|
| 1973 |
+
"grad_norm": 0.09133646637201309,
|
| 1974 |
+
"learning_rate": 1.186492214802426e-05,
|
| 1975 |
+
"loss": 0.1263,
|
| 1976 |
+
"step": 7025
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 0.635593220338983,
|
| 1980 |
+
"grad_norm": 0.17885656654834747,
|
| 1981 |
+
"learning_rate": 1.1735756284632442e-05,
|
| 1982 |
+
"loss": 0.1524,
|
| 1983 |
+
"step": 7050
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 0.6378470970068518,
|
| 1987 |
+
"grad_norm": 28.033615112304688,
|
| 1988 |
+
"learning_rate": 1.1607004765515628e-05,
|
| 1989 |
+
"loss": 0.0855,
|
| 1990 |
+
"step": 7075
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 0.6401009736747205,
|
| 1994 |
+
"grad_norm": 0.42501696944236755,
|
| 1995 |
+
"learning_rate": 1.1478674045887007e-05,
|
| 1996 |
+
"loss": 0.0899,
|
| 1997 |
+
"step": 7100
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 0.6423548503425892,
|
| 2001 |
+
"grad_norm": 0.5233384370803833,
|
| 2002 |
+
"learning_rate": 1.1350770559862125e-05,
|
| 2003 |
+
"loss": 0.0931,
|
| 2004 |
+
"step": 7125
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 0.644608727010458,
|
| 2008 |
+
"grad_norm": 0.028280099853873253,
|
| 2009 |
+
"learning_rate": 1.1223300720136305e-05,
|
| 2010 |
+
"loss": 0.1446,
|
| 2011 |
+
"step": 7150
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 0.6468626036783267,
|
| 2015 |
+
"grad_norm": 0.08879163861274719,
|
| 2016 |
+
"learning_rate": 1.1096270917663195e-05,
|
| 2017 |
+
"loss": 0.0715,
|
| 2018 |
+
"step": 7175
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 0.6491164803461954,
|
| 2022 |
+
"grad_norm": 0.09535069018602371,
|
| 2023 |
+
"learning_rate": 1.0969687521334275e-05,
|
| 2024 |
+
"loss": 0.1956,
|
| 2025 |
+
"step": 7200
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 0.6513703570140642,
|
| 2029 |
+
"grad_norm": 0.407167911529541,
|
| 2030 |
+
"learning_rate": 1.084355687765955e-05,
|
| 2031 |
+
"loss": 0.1472,
|
| 2032 |
+
"step": 7225
|
| 2033 |
+
},
|
| 2034 |
+
{
|
| 2035 |
+
"epoch": 0.6536242336819329,
|
| 2036 |
+
"grad_norm": 0.5132746696472168,
|
| 2037 |
+
"learning_rate": 1.0717885310449407e-05,
|
| 2038 |
+
"loss": 0.0652,
|
| 2039 |
+
"step": 7250
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 0.6558781103498017,
|
| 2043 |
+
"grad_norm": 0.9216197729110718,
|
| 2044 |
+
"learning_rate": 1.0592679120497505e-05,
|
| 2045 |
+
"loss": 0.055,
|
| 2046 |
+
"step": 7275
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 0.6581319870176704,
|
| 2050 |
+
"grad_norm": 0.2732597887516022,
|
| 2051 |
+
"learning_rate": 1.0467944585264886e-05,
|
| 2052 |
+
"loss": 0.1322,
|
| 2053 |
+
"step": 7300
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 0.6603858636855391,
|
| 2057 |
+
"grad_norm": 32.903404235839844,
|
| 2058 |
+
"learning_rate": 1.0343687958565253e-05,
|
| 2059 |
+
"loss": 0.094,
|
| 2060 |
+
"step": 7325
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 0.6626397403534079,
|
| 2064 |
+
"grad_norm": 0.044182177633047104,
|
| 2065 |
+
"learning_rate": 1.0219915470251411e-05,
|
| 2066 |
+
"loss": 0.0713,
|
| 2067 |
+
"step": 7350
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 0.6648936170212766,
|
| 2071 |
+
"grad_norm": 0.1363406777381897,
|
| 2072 |
+
"learning_rate": 1.0096633325902931e-05,
|
| 2073 |
+
"loss": 0.0925,
|
| 2074 |
+
"step": 7375
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 0.6671474936891453,
|
| 2078 |
+
"grad_norm": 16.840986251831055,
|
| 2079 |
+
"learning_rate": 9.973847706515007e-06,
|
| 2080 |
+
"loss": 0.1535,
|
| 2081 |
+
"step": 7400
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 0.6694013703570141,
|
| 2085 |
+
"grad_norm": 30.311649322509766,
|
| 2086 |
+
"learning_rate": 9.851564768188566e-06,
|
| 2087 |
+
"loss": 0.1257,
|
| 2088 |
+
"step": 7425
|
| 2089 |
+
},
|
| 2090 |
+
{
|
| 2091 |
+
"epoch": 0.6716552470248828,
|
| 2092 |
+
"grad_norm": 0.19690652191638947,
|
| 2093 |
+
"learning_rate": 9.729790641821652e-06,
|
| 2094 |
+
"loss": 0.0924,
|
| 2095 |
+
"step": 7450
|
| 2096 |
+
},
|
| 2097 |
+
{
|
| 2098 |
+
"epoch": 0.6739091236927516,
|
| 2099 |
+
"grad_norm": 15.262349128723145,
|
| 2100 |
+
"learning_rate": 9.608531432801953e-06,
|
| 2101 |
+
"loss": 0.0591,
|
| 2102 |
+
"step": 7475
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 0.6761630003606203,
|
| 2106 |
+
"grad_norm": 0.08197861164808273,
|
| 2107 |
+
"learning_rate": 9.48779322070081e-06,
|
| 2108 |
+
"loss": 0.1236,
|
| 2109 |
+
"step": 7500
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 0.678416877028489,
|
| 2113 |
+
"grad_norm": 33.88435363769531,
|
| 2114 |
+
"learning_rate": 9.367582058968325e-06,
|
| 2115 |
+
"loss": 0.147,
|
| 2116 |
+
"step": 7525
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 0.6806707536963578,
|
| 2120 |
+
"grad_norm": 0.18482732772827148,
|
| 2121 |
+
"learning_rate": 9.247903974629862e-06,
|
| 2122 |
+
"loss": 0.1603,
|
| 2123 |
+
"step": 7550
|
| 2124 |
+
},
|
| 2125 |
+
{
|
| 2126 |
+
"epoch": 0.6829246303642265,
|
| 2127 |
+
"grad_norm": 25.463579177856445,
|
| 2128 |
+
"learning_rate": 9.128764967983929e-06,
|
| 2129 |
+
"loss": 0.1208,
|
| 2130 |
+
"step": 7575
|
| 2131 |
+
},
|
| 2132 |
+
{
|
| 2133 |
+
"epoch": 0.6851785070320952,
|
| 2134 |
+
"grad_norm": 670.9367065429688,
|
| 2135 |
+
"learning_rate": 9.010171012301291e-06,
|
| 2136 |
+
"loss": 0.0883,
|
| 2137 |
+
"step": 7600
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 0.687432383699964,
|
| 2141 |
+
"grad_norm": 63.760841369628906,
|
| 2142 |
+
"learning_rate": 8.892128053525471e-06,
|
| 2143 |
+
"loss": 0.1323,
|
| 2144 |
+
"step": 7625
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 0.6896862603678326,
|
| 2148 |
+
"grad_norm": 0.16674941778182983,
|
| 2149 |
+
"learning_rate": 8.77464200997471e-06,
|
| 2150 |
+
"loss": 0.082,
|
| 2151 |
+
"step": 7650
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 0.6919401370357015,
|
| 2155 |
+
"grad_norm": 1.6841052770614624,
|
| 2156 |
+
"learning_rate": 8.657718772045176e-06,
|
| 2157 |
+
"loss": 0.0917,
|
| 2158 |
+
"step": 7675
|
| 2159 |
+
},
|
| 2160 |
+
{
|
| 2161 |
+
"epoch": 0.6941940137035701,
|
| 2162 |
+
"grad_norm": 24.29615592956543,
|
| 2163 |
+
"learning_rate": 8.541364201915662e-06,
|
| 2164 |
+
"loss": 0.0623,
|
| 2165 |
+
"step": 7700
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 0.6964478903714388,
|
| 2169 |
+
"grad_norm": 0.6013920307159424,
|
| 2170 |
+
"learning_rate": 8.42558413325367e-06,
|
| 2171 |
+
"loss": 0.1257,
|
| 2172 |
+
"step": 7725
|
| 2173 |
+
},
|
| 2174 |
+
{
|
| 2175 |
+
"epoch": 0.6987017670393076,
|
| 2176 |
+
"grad_norm": 0.08703339844942093,
|
| 2177 |
+
"learning_rate": 8.310384370922935e-06,
|
| 2178 |
+
"loss": 0.1565,
|
| 2179 |
+
"step": 7750
|
| 2180 |
+
},
|
| 2181 |
+
{
|
| 2182 |
+
"epoch": 0.7009556437071763,
|
| 2183 |
+
"grad_norm": 0.1819688379764557,
|
| 2184 |
+
"learning_rate": 8.195770690692373e-06,
|
| 2185 |
+
"loss": 0.1264,
|
| 2186 |
+
"step": 7775
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 0.703209520375045,
|
| 2190 |
+
"grad_norm": 0.4516184628009796,
|
| 2191 |
+
"learning_rate": 8.081748838946516e-06,
|
| 2192 |
+
"loss": 0.1686,
|
| 2193 |
+
"step": 7800
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 0.7054633970429138,
|
| 2197 |
+
"grad_norm": 11.845215797424316,
|
| 2198 |
+
"learning_rate": 7.96832453239739e-06,
|
| 2199 |
+
"loss": 0.1668,
|
| 2200 |
+
"step": 7825
|
| 2201 |
+
},
|
| 2202 |
+
{
|
| 2203 |
+
"epoch": 0.7077172737107825,
|
| 2204 |
+
"grad_norm": 0.20621220767498016,
|
| 2205 |
+
"learning_rate": 7.85550345779794e-06,
|
| 2206 |
+
"loss": 0.1688,
|
| 2207 |
+
"step": 7850
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 0.7099711503786513,
|
| 2211 |
+
"grad_norm": 0.608947217464447,
|
| 2212 |
+
"learning_rate": 7.743291271656818e-06,
|
| 2213 |
+
"loss": 0.129,
|
| 2214 |
+
"step": 7875
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 0.71222502704652,
|
| 2218 |
+
"grad_norm": 1.3757911920547485,
|
| 2219 |
+
"learning_rate": 7.631693599954901e-06,
|
| 2220 |
+
"loss": 0.1011,
|
| 2221 |
+
"step": 7900
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 0.7144789037143887,
|
| 2225 |
+
"grad_norm": 0.04608175903558731,
|
| 2226 |
+
"learning_rate": 7.520716037863143e-06,
|
| 2227 |
+
"loss": 0.0808,
|
| 2228 |
+
"step": 7925
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 0.7167327803822575,
|
| 2232 |
+
"grad_norm": 0.02408667653799057,
|
| 2233 |
+
"learning_rate": 7.4103641494620434e-06,
|
| 2234 |
+
"loss": 0.0615,
|
| 2235 |
+
"step": 7950
|
| 2236 |
+
},
|
| 2237 |
+
{
|
| 2238 |
+
"epoch": 0.7189866570501262,
|
| 2239 |
+
"grad_norm": 16.15177345275879,
|
| 2240 |
+
"learning_rate": 7.300643467462758e-06,
|
| 2241 |
+
"loss": 0.0886,
|
| 2242 |
+
"step": 7975
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 0.7212405337179949,
|
| 2246 |
+
"grad_norm": 0.06660573929548264,
|
| 2247 |
+
"learning_rate": 7.191559492929623e-06,
|
| 2248 |
+
"loss": 0.0859,
|
| 2249 |
+
"step": 8000
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 0.7234944103858637,
|
| 2253 |
+
"grad_norm": 0.013860706239938736,
|
| 2254 |
+
"learning_rate": 7.083117695004389e-06,
|
| 2255 |
+
"loss": 0.056,
|
| 2256 |
+
"step": 8025
|
| 2257 |
+
},
|
| 2258 |
+
{
|
| 2259 |
+
"epoch": 0.7257482870537324,
|
| 2260 |
+
"grad_norm": 584.7158813476562,
|
| 2261 |
+
"learning_rate": 6.9753235106320084e-06,
|
| 2262 |
+
"loss": 0.1153,
|
| 2263 |
+
"step": 8050
|
| 2264 |
+
},
|
| 2265 |
+
{
|
| 2266 |
+
"epoch": 0.7280021637216012,
|
| 2267 |
+
"grad_norm": 14.605430603027344,
|
| 2268 |
+
"learning_rate": 6.868182344288039e-06,
|
| 2269 |
+
"loss": 0.0537,
|
| 2270 |
+
"step": 8075
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 0.7302560403894699,
|
| 2274 |
+
"grad_norm": 0.12337201833724976,
|
| 2275 |
+
"learning_rate": 6.761699567707685e-06,
|
| 2276 |
+
"loss": 0.1732,
|
| 2277 |
+
"step": 8100
|
| 2278 |
+
},
|
| 2279 |
+
{
|
| 2280 |
+
"epoch": 0.7325099170573386,
|
| 2281 |
+
"grad_norm": 0.1590554416179657,
|
| 2282 |
+
"learning_rate": 6.655880519616469e-06,
|
| 2283 |
+
"loss": 0.049,
|
| 2284 |
+
"step": 8125
|
| 2285 |
+
},
|
| 2286 |
+
{
|
| 2287 |
+
"epoch": 0.7347637937252074,
|
| 2288 |
+
"grad_norm": 0.17607277631759644,
|
| 2289 |
+
"learning_rate": 6.550730505462566e-06,
|
| 2290 |
+
"loss": 0.0315,
|
| 2291 |
+
"step": 8150
|
| 2292 |
+
},
|
| 2293 |
+
{
|
| 2294 |
+
"epoch": 0.7370176703930761,
|
| 2295 |
+
"grad_norm": 0.27538254857063293,
|
| 2296 |
+
"learning_rate": 6.446254797150824e-06,
|
| 2297 |
+
"loss": 0.1013,
|
| 2298 |
+
"step": 8175
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 0.7392715470609448,
|
| 2302 |
+
"grad_norm": 36.80734634399414,
|
| 2303 |
+
"learning_rate": 6.342458632778398e-06,
|
| 2304 |
+
"loss": 0.1319,
|
| 2305 |
+
"step": 8200
|
| 2306 |
+
},
|
| 2307 |
+
{
|
| 2308 |
+
"epoch": 0.7415254237288136,
|
| 2309 |
+
"grad_norm": 0.29385101795196533,
|
| 2310 |
+
"learning_rate": 6.239347216372171e-06,
|
| 2311 |
+
"loss": 0.1119,
|
| 2312 |
+
"step": 8225
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 0.7437793003966823,
|
| 2316 |
+
"grad_norm": 1.9184335470199585,
|
| 2317 |
+
"learning_rate": 6.136925717627846e-06,
|
| 2318 |
+
"loss": 0.1307,
|
| 2319 |
+
"step": 8250
|
| 2320 |
+
},
|
| 2321 |
+
{
|
| 2322 |
+
"epoch": 0.7460331770645511,
|
| 2323 |
+
"grad_norm": 0.8366139531135559,
|
| 2324 |
+
"learning_rate": 6.035199271650694e-06,
|
| 2325 |
+
"loss": 0.0604,
|
| 2326 |
+
"step": 8275
|
| 2327 |
+
},
|
| 2328 |
+
{
|
| 2329 |
+
"epoch": 0.7482870537324198,
|
| 2330 |
+
"grad_norm": 0.009770547039806843,
|
| 2331 |
+
"learning_rate": 5.934172978698172e-06,
|
| 2332 |
+
"loss": 0.0551,
|
| 2333 |
+
"step": 8300
|
| 2334 |
+
},
|
| 2335 |
+
{
|
| 2336 |
+
"epoch": 0.7505409304002885,
|
| 2337 |
+
"grad_norm": 154.11497497558594,
|
| 2338 |
+
"learning_rate": 5.833851903924153e-06,
|
| 2339 |
+
"loss": 0.0861,
|
| 2340 |
+
"step": 8325
|
| 2341 |
+
},
|
| 2342 |
+
{
|
| 2343 |
+
"epoch": 0.7527948070681573,
|
| 2344 |
+
"grad_norm": 18.2222957611084,
|
| 2345 |
+
"learning_rate": 5.7342410771250025e-06,
|
| 2346 |
+
"loss": 0.1007,
|
| 2347 |
+
"step": 8350
|
| 2348 |
+
},
|
| 2349 |
+
{
|
| 2350 |
+
"epoch": 0.755048683736026,
|
| 2351 |
+
"grad_norm": 0.08756579458713531,
|
| 2352 |
+
"learning_rate": 5.6353454924873916e-06,
|
| 2353 |
+
"loss": 0.1254,
|
| 2354 |
+
"step": 8375
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"epoch": 0.7573025604038947,
|
| 2358 |
+
"grad_norm": 0.020753098651766777,
|
| 2359 |
+
"learning_rate": 5.537170108337906e-06,
|
| 2360 |
+
"loss": 0.0878,
|
| 2361 |
+
"step": 8400
|
| 2362 |
+
},
|
| 2363 |
+
{
|
| 2364 |
+
"epoch": 0.7595564370717635,
|
| 2365 |
+
"grad_norm": 0.02824629656970501,
|
| 2366 |
+
"learning_rate": 5.439719846894449e-06,
|
| 2367 |
+
"loss": 0.1191,
|
| 2368 |
+
"step": 8425
|
| 2369 |
+
},
|
| 2370 |
+
{
|
| 2371 |
+
"epoch": 0.7618103137396322,
|
| 2372 |
+
"grad_norm": 0.15957316756248474,
|
| 2373 |
+
"learning_rate": 5.342999594019456e-06,
|
| 2374 |
+
"loss": 0.1476,
|
| 2375 |
+
"step": 8450
|
| 2376 |
+
},
|
| 2377 |
+
{
|
| 2378 |
+
"epoch": 0.764064190407501,
|
| 2379 |
+
"grad_norm": 14.759479522705078,
|
| 2380 |
+
"learning_rate": 5.2470141989749315e-06,
|
| 2381 |
+
"loss": 0.1251,
|
| 2382 |
+
"step": 8475
|
| 2383 |
+
},
|
| 2384 |
+
{
|
| 2385 |
+
"epoch": 0.7663180670753696,
|
| 2386 |
+
"grad_norm": 0.07145761698484421,
|
| 2387 |
+
"learning_rate": 5.151768474179324e-06,
|
| 2388 |
+
"loss": 0.1057,
|
| 2389 |
+
"step": 8500
|
| 2390 |
+
},
|
| 2391 |
+
{
|
| 2392 |
+
"epoch": 0.7685719437432383,
|
| 2393 |
+
"grad_norm": 0.04770238324999809,
|
| 2394 |
+
"learning_rate": 5.057267194966249e-06,
|
| 2395 |
+
"loss": 0.1478,
|
| 2396 |
+
"step": 8525
|
| 2397 |
+
},
|
| 2398 |
+
{
|
| 2399 |
+
"epoch": 0.7708258204111071,
|
| 2400 |
+
"grad_norm": 9.733259201049805,
|
| 2401 |
+
"learning_rate": 4.9635150993450505e-06,
|
| 2402 |
+
"loss": 0.1225,
|
| 2403 |
+
"step": 8550
|
| 2404 |
+
},
|
| 2405 |
+
{
|
| 2406 |
+
"epoch": 0.7730796970789758,
|
| 2407 |
+
"grad_norm": 1.1977856159210205,
|
| 2408 |
+
"learning_rate": 4.870516887763292e-06,
|
| 2409 |
+
"loss": 0.1596,
|
| 2410 |
+
"step": 8575
|
| 2411 |
+
},
|
| 2412 |
+
{
|
| 2413 |
+
"epoch": 0.7753335737468445,
|
| 2414 |
+
"grad_norm": 1.318892002105713,
|
| 2415 |
+
"learning_rate": 4.778277222871033e-06,
|
| 2416 |
+
"loss": 0.1581,
|
| 2417 |
+
"step": 8600
|
| 2418 |
+
},
|
| 2419 |
+
{
|
| 2420 |
+
"epoch": 0.7775874504147133,
|
| 2421 |
+
"grad_norm": 0.619382381439209,
|
| 2422 |
+
"learning_rate": 4.686800729287095e-06,
|
| 2423 |
+
"loss": 0.0913,
|
| 2424 |
+
"step": 8625
|
| 2425 |
+
},
|
| 2426 |
+
{
|
| 2427 |
+
"epoch": 0.779841327082582,
|
| 2428 |
+
"grad_norm": 0.1258872002363205,
|
| 2429 |
+
"learning_rate": 4.596091993367215e-06,
|
| 2430 |
+
"loss": 0.1059,
|
| 2431 |
+
"step": 8650
|
| 2432 |
+
},
|
| 2433 |
+
{
|
| 2434 |
+
"epoch": 0.7820952037504508,
|
| 2435 |
+
"grad_norm": 20.535972595214844,
|
| 2436 |
+
"learning_rate": 4.5061555629740325e-06,
|
| 2437 |
+
"loss": 0.1324,
|
| 2438 |
+
"step": 8675
|
| 2439 |
+
},
|
| 2440 |
+
{
|
| 2441 |
+
"epoch": 0.7843490804183195,
|
| 2442 |
+
"grad_norm": 0.3956652581691742,
|
| 2443 |
+
"learning_rate": 4.416995947249154e-06,
|
| 2444 |
+
"loss": 0.108,
|
| 2445 |
+
"step": 8700
|
| 2446 |
+
},
|
| 2447 |
+
{
|
| 2448 |
+
"epoch": 0.7866029570861882,
|
| 2449 |
+
"grad_norm": 0.295835018157959,
|
| 2450 |
+
"learning_rate": 4.3286176163870186e-06,
|
| 2451 |
+
"loss": 0.0755,
|
| 2452 |
+
"step": 8725
|
| 2453 |
+
},
|
| 2454 |
+
{
|
| 2455 |
+
"epoch": 0.788856833754057,
|
| 2456 |
+
"grad_norm": 0.05923585221171379,
|
| 2457 |
+
"learning_rate": 4.241025001410801e-06,
|
| 2458 |
+
"loss": 0.0998,
|
| 2459 |
+
"step": 8750
|
| 2460 |
+
},
|
| 2461 |
+
{
|
| 2462 |
+
"epoch": 0.7911107104219257,
|
| 2463 |
+
"grad_norm": 0.04538344964385033,
|
| 2464 |
+
"learning_rate": 4.154222493950248e-06,
|
| 2465 |
+
"loss": 0.1605,
|
| 2466 |
+
"step": 8775
|
| 2467 |
+
},
|
| 2468 |
+
{
|
| 2469 |
+
"epoch": 0.7933645870897944,
|
| 2470 |
+
"grad_norm": 0.028473101556301117,
|
| 2471 |
+
"learning_rate": 4.0682144460215015e-06,
|
| 2472 |
+
"loss": 0.0491,
|
| 2473 |
+
"step": 8800
|
| 2474 |
+
},
|
| 2475 |
+
{
|
| 2476 |
+
"epoch": 0.7956184637576632,
|
| 2477 |
+
"grad_norm": 0.0626368597149849,
|
| 2478 |
+
"learning_rate": 3.983005169808891e-06,
|
| 2479 |
+
"loss": 0.0593,
|
| 2480 |
+
"step": 8825
|
| 2481 |
+
},
|
| 2482 |
+
{
|
| 2483 |
+
"epoch": 0.7978723404255319,
|
| 2484 |
+
"grad_norm": 10.354604721069336,
|
| 2485 |
+
"learning_rate": 3.898598937448743e-06,
|
| 2486 |
+
"loss": 0.0766,
|
| 2487 |
+
"step": 8850
|
| 2488 |
+
},
|
| 2489 |
+
{
|
| 2490 |
+
"epoch": 0.8001262170934007,
|
| 2491 |
+
"grad_norm": 19.488628387451172,
|
| 2492 |
+
"learning_rate": 3.81499998081518e-06,
|
| 2493 |
+
"loss": 0.1018,
|
| 2494 |
+
"step": 8875
|
| 2495 |
+
},
|
| 2496 |
+
{
|
| 2497 |
+
"epoch": 0.8023800937612694,
|
| 2498 |
+
"grad_norm": 0.03765766695141792,
|
| 2499 |
+
"learning_rate": 3.7322124913079673e-06,
|
| 2500 |
+
"loss": 0.1289,
|
| 2501 |
+
"step": 8900
|
| 2502 |
+
},
|
| 2503 |
+
{
|
| 2504 |
+
"epoch": 0.8046339704291381,
|
| 2505 |
+
"grad_norm": 13.242864608764648,
|
| 2506 |
+
"learning_rate": 3.650240619642331e-06,
|
| 2507 |
+
"loss": 0.0422,
|
| 2508 |
+
"step": 8925
|
| 2509 |
+
},
|
| 2510 |
+
{
|
| 2511 |
+
"epoch": 0.8068878470970069,
|
| 2512 |
+
"grad_norm": 0.2363833338022232,
|
| 2513 |
+
"learning_rate": 3.5690884756408872e-06,
|
| 2514 |
+
"loss": 0.1273,
|
| 2515 |
+
"step": 8950
|
| 2516 |
+
},
|
| 2517 |
+
{
|
| 2518 |
+
"epoch": 0.8091417237648756,
|
| 2519 |
+
"grad_norm": 0.055782124400138855,
|
| 2520 |
+
"learning_rate": 3.4887601280275973e-06,
|
| 2521 |
+
"loss": 0.0293,
|
| 2522 |
+
"step": 8975
|
| 2523 |
+
},
|
| 2524 |
+
{
|
| 2525 |
+
"epoch": 0.8113956004327443,
|
| 2526 |
+
"grad_norm": 0.32612791657447815,
|
| 2527 |
+
"learning_rate": 3.409259604223716e-06,
|
| 2528 |
+
"loss": 0.084,
|
| 2529 |
+
"step": 9000
|
| 2530 |
+
},
|
| 2531 |
+
{
|
| 2532 |
+
"epoch": 0.8136494771006131,
|
| 2533 |
+
"grad_norm": 0.07187359035015106,
|
| 2534 |
+
"learning_rate": 3.3305908901459282e-06,
|
| 2535 |
+
"loss": 0.15,
|
| 2536 |
+
"step": 9025
|
| 2537 |
+
},
|
| 2538 |
+
{
|
| 2539 |
+
"epoch": 0.8159033537684818,
|
| 2540 |
+
"grad_norm": 0.11757878959178925,
|
| 2541 |
+
"learning_rate": 3.252757930006494e-06,
|
| 2542 |
+
"loss": 0.1501,
|
| 2543 |
+
"step": 9050
|
| 2544 |
+
},
|
| 2545 |
+
{
|
| 2546 |
+
"epoch": 0.8181572304363505,
|
| 2547 |
+
"grad_norm": 0.06753458082675934,
|
| 2548 |
+
"learning_rate": 3.1757646261154785e-06,
|
| 2549 |
+
"loss": 0.0849,
|
| 2550 |
+
"step": 9075
|
| 2551 |
+
},
|
| 2552 |
+
{
|
| 2553 |
+
"epoch": 0.8204111071042193,
|
| 2554 |
+
"grad_norm": 0.1389445662498474,
|
| 2555 |
+
"learning_rate": 3.0996148386851033e-06,
|
| 2556 |
+
"loss": 0.1419,
|
| 2557 |
+
"step": 9100
|
| 2558 |
+
},
|
| 2559 |
+
{
|
| 2560 |
+
"epoch": 0.822664983772088,
|
| 2561 |
+
"grad_norm": 2.3049068450927734,
|
| 2562 |
+
"learning_rate": 3.0243123856362364e-06,
|
| 2563 |
+
"loss": 0.0428,
|
| 2564 |
+
"step": 9125
|
| 2565 |
+
},
|
| 2566 |
+
{
|
| 2567 |
+
"epoch": 0.8249188604399568,
|
| 2568 |
+
"grad_norm": 0.16142496466636658,
|
| 2569 |
+
"learning_rate": 2.949861042406945e-06,
|
| 2570 |
+
"loss": 0.0918,
|
| 2571 |
+
"step": 9150
|
| 2572 |
+
},
|
| 2573 |
+
{
|
| 2574 |
+
"epoch": 0.8271727371078255,
|
| 2575 |
+
"grad_norm": 0.04665221646428108,
|
| 2576 |
+
"learning_rate": 2.8762645417632117e-06,
|
| 2577 |
+
"loss": 0.067,
|
| 2578 |
+
"step": 9175
|
| 2579 |
+
},
|
| 2580 |
+
{
|
| 2581 |
+
"epoch": 0.8294266137756942,
|
| 2582 |
+
"grad_norm": 13.154709815979004,
|
| 2583 |
+
"learning_rate": 2.803526573611792e-06,
|
| 2584 |
+
"loss": 0.0956,
|
| 2585 |
+
"step": 9200
|
| 2586 |
+
},
|
| 2587 |
+
{
|
| 2588 |
+
"epoch": 0.831680490443563,
|
| 2589 |
+
"grad_norm": 19.27960205078125,
|
| 2590 |
+
"learning_rate": 2.731650784815205e-06,
|
| 2591 |
+
"loss": 0.1256,
|
| 2592 |
+
"step": 9225
|
| 2593 |
+
},
|
| 2594 |
+
{
|
| 2595 |
+
"epoch": 0.8339343671114317,
|
| 2596 |
+
"grad_norm": 0.19370368123054504,
|
| 2597 |
+
"learning_rate": 2.660640779008903e-06,
|
| 2598 |
+
"loss": 0.0597,
|
| 2599 |
+
"step": 9250
|
| 2600 |
+
},
|
| 2601 |
+
{
|
| 2602 |
+
"epoch": 0.8361882437793003,
|
| 2603 |
+
"grad_norm": 0.030255110934376717,
|
| 2604 |
+
"learning_rate": 2.5905001164205802e-06,
|
| 2605 |
+
"loss": 0.1069,
|
| 2606 |
+
"step": 9275
|
| 2607 |
+
},
|
| 2608 |
+
{
|
| 2609 |
+
"epoch": 0.8384421204471691,
|
| 2610 |
+
"grad_norm": 42.32724380493164,
|
| 2611 |
+
"learning_rate": 2.521232313691686e-06,
|
| 2612 |
+
"loss": 0.1151,
|
| 2613 |
+
"step": 9300
|
| 2614 |
+
},
|
| 2615 |
+
{
|
| 2616 |
+
"epoch": 0.8406959971150378,
|
| 2617 |
+
"grad_norm": 0.325438916683197,
|
| 2618 |
+
"learning_rate": 2.4528408437011053e-06,
|
| 2619 |
+
"loss": 0.1465,
|
| 2620 |
+
"step": 9325
|
| 2621 |
+
},
|
| 2622 |
+
{
|
| 2623 |
+
"epoch": 0.8429498737829066,
|
| 2624 |
+
"grad_norm": 0.0806034728884697,
|
| 2625 |
+
"learning_rate": 2.3853291353910368e-06,
|
| 2626 |
+
"loss": 0.0626,
|
| 2627 |
+
"step": 9350
|
| 2628 |
+
},
|
| 2629 |
+
{
|
| 2630 |
+
"epoch": 0.8452037504507753,
|
| 2631 |
+
"grad_norm": 0.10465729981660843,
|
| 2632 |
+
"learning_rate": 2.3187005735950917e-06,
|
| 2633 |
+
"loss": 0.1227,
|
| 2634 |
+
"step": 9375
|
| 2635 |
+
},
|
| 2636 |
+
{
|
| 2637 |
+
"epoch": 0.847457627118644,
|
| 2638 |
+
"grad_norm": 39.283180236816406,
|
| 2639 |
+
"learning_rate": 2.2529584988685716e-06,
|
| 2640 |
+
"loss": 0.1182,
|
| 2641 |
+
"step": 9400
|
| 2642 |
+
},
|
| 2643 |
+
{
|
| 2644 |
+
"epoch": 0.8497115037865128,
|
| 2645 |
+
"grad_norm": 0.4078097939491272,
|
| 2646 |
+
"learning_rate": 2.18810620732097e-06,
|
| 2647 |
+
"loss": 0.0491,
|
| 2648 |
+
"step": 9425
|
| 2649 |
+
},
|
| 2650 |
+
{
|
| 2651 |
+
"epoch": 0.8519653804543815,
|
| 2652 |
+
"grad_norm": 0.05473583936691284,
|
| 2653 |
+
"learning_rate": 2.124146950450763e-06,
|
| 2654 |
+
"loss": 0.0623,
|
| 2655 |
+
"step": 9450
|
| 2656 |
+
},
|
| 2657 |
+
{
|
| 2658 |
+
"epoch": 0.8542192571222502,
|
| 2659 |
+
"grad_norm": 10.327305793762207,
|
| 2660 |
+
"learning_rate": 2.0610839349823397e-06,
|
| 2661 |
+
"loss": 0.1621,
|
| 2662 |
+
"step": 9475
|
| 2663 |
+
},
|
| 2664 |
+
{
|
| 2665 |
+
"epoch": 0.856473133790119,
|
| 2666 |
+
"grad_norm": 0.23021194338798523,
|
| 2667 |
+
"learning_rate": 1.9989203227052446e-06,
|
| 2668 |
+
"loss": 0.0396,
|
| 2669 |
+
"step": 9500
|
| 2670 |
+
},
|
| 2671 |
+
{
|
| 2672 |
+
"epoch": 0.8587270104579877,
|
| 2673 |
+
"grad_norm": 9.494355201721191,
|
| 2674 |
+
"learning_rate": 1.9376592303156695e-06,
|
| 2675 |
+
"loss": 0.1125,
|
| 2676 |
+
"step": 9525
|
| 2677 |
+
},
|
| 2678 |
+
{
|
| 2679 |
+
"epoch": 0.8609808871258565,
|
| 2680 |
+
"grad_norm": 0.10281592607498169,
|
| 2681 |
+
"learning_rate": 1.8773037292601716e-06,
|
| 2682 |
+
"loss": 0.1091,
|
| 2683 |
+
"step": 9550
|
| 2684 |
+
},
|
| 2685 |
+
{
|
| 2686 |
+
"epoch": 0.8632347637937252,
|
| 2687 |
+
"grad_norm": 0.1783093959093094,
|
| 2688 |
+
"learning_rate": 1.8178568455816891e-06,
|
| 2689 |
+
"loss": 0.0885,
|
| 2690 |
+
"step": 9575
|
| 2691 |
+
},
|
| 2692 |
+
{
|
| 2693 |
+
"epoch": 0.8654886404615939,
|
| 2694 |
+
"grad_norm": 0.1075611561536789,
|
| 2695 |
+
"learning_rate": 1.7593215597678193e-06,
|
| 2696 |
+
"loss": 0.1054,
|
| 2697 |
+
"step": 9600
|
| 2698 |
+
},
|
| 2699 |
+
{
|
| 2700 |
+
"epoch": 0.8677425171294627,
|
| 2701 |
+
"grad_norm": 32.095333099365234,
|
| 2702 |
+
"learning_rate": 1.7017008066013963e-06,
|
| 2703 |
+
"loss": 0.1152,
|
| 2704 |
+
"step": 9625
|
| 2705 |
+
},
|
| 2706 |
+
{
|
| 2707 |
+
"epoch": 0.8699963937973314,
|
| 2708 |
+
"grad_norm": 0.06124063953757286,
|
| 2709 |
+
"learning_rate": 1.6449974750133346e-06,
|
| 2710 |
+
"loss": 0.1883,
|
| 2711 |
+
"step": 9650
|
| 2712 |
+
},
|
| 2713 |
+
{
|
| 2714 |
+
"epoch": 0.8722502704652001,
|
| 2715 |
+
"grad_norm": 0.20606283843517303,
|
| 2716 |
+
"learning_rate": 1.589214407937807e-06,
|
| 2717 |
+
"loss": 0.206,
|
| 2718 |
+
"step": 9675
|
| 2719 |
+
},
|
| 2720 |
+
{
|
| 2721 |
+
"epoch": 0.8745041471330689,
|
| 2722 |
+
"grad_norm": 0.02182711474597454,
|
| 2723 |
+
"learning_rate": 1.5343544021696866e-06,
|
| 2724 |
+
"loss": 0.0713,
|
| 2725 |
+
"step": 9700
|
| 2726 |
+
},
|
| 2727 |
+
{
|
| 2728 |
+
"epoch": 0.8767580238009376,
|
| 2729 |
+
"grad_norm": 1.0394850969314575,
|
| 2730 |
+
"learning_rate": 1.480420208224338e-06,
|
| 2731 |
+
"loss": 0.1159,
|
| 2732 |
+
"step": 9725
|
| 2733 |
+
},
|
| 2734 |
+
{
|
| 2735 |
+
"epoch": 0.8790119004688064,
|
| 2736 |
+
"grad_norm": 0.033810436725616455,
|
| 2737 |
+
"learning_rate": 1.4274145301997088e-06,
|
| 2738 |
+
"loss": 0.1481,
|
| 2739 |
+
"step": 9750
|
| 2740 |
+
},
|
| 2741 |
+
{
|
| 2742 |
+
"epoch": 0.8812657771366751,
|
| 2743 |
+
"grad_norm": 0.2243296205997467,
|
| 2744 |
+
"learning_rate": 1.3753400256407656e-06,
|
| 2745 |
+
"loss": 0.1388,
|
| 2746 |
+
"step": 9775
|
| 2747 |
+
},
|
| 2748 |
+
{
|
| 2749 |
+
"epoch": 0.8835196538045438,
|
| 2750 |
+
"grad_norm": 1.2577711343765259,
|
| 2751 |
+
"learning_rate": 1.3241993054062375e-06,
|
| 2752 |
+
"loss": 0.1187,
|
| 2753 |
+
"step": 9800
|
| 2754 |
+
},
|
| 2755 |
+
{
|
| 2756 |
+
"epoch": 0.8857735304724126,
|
| 2757 |
+
"grad_norm": 0.44555121660232544,
|
| 2758 |
+
"learning_rate": 1.2739949335377077e-06,
|
| 2759 |
+
"loss": 0.0676,
|
| 2760 |
+
"step": 9825
|
| 2761 |
+
},
|
| 2762 |
+
{
|
| 2763 |
+
"epoch": 0.8880274071402813,
|
| 2764 |
+
"grad_norm": 0.10070797801017761,
|
| 2765 |
+
"learning_rate": 1.2247294271311017e-06,
|
| 2766 |
+
"loss": 0.0753,
|
| 2767 |
+
"step": 9850
|
| 2768 |
+
},
|
| 2769 |
+
{
|
| 2770 |
+
"epoch": 0.89028128380815,
|
| 2771 |
+
"grad_norm": 0.04352548345923424,
|
| 2772 |
+
"learning_rate": 1.1764052562104444e-06,
|
| 2773 |
+
"loss": 0.0726,
|
| 2774 |
+
"step": 9875
|
| 2775 |
+
},
|
| 2776 |
+
{
|
| 2777 |
+
"epoch": 0.8925351604760188,
|
| 2778 |
+
"grad_norm": 0.0819600373506546,
|
| 2779 |
+
"learning_rate": 1.1290248436040274e-06,
|
| 2780 |
+
"loss": 0.14,
|
| 2781 |
+
"step": 9900
|
| 2782 |
+
},
|
| 2783 |
+
{
|
| 2784 |
+
"epoch": 0.8947890371438875,
|
| 2785 |
+
"grad_norm": 0.02376544661819935,
|
| 2786 |
+
"learning_rate": 1.082590564822963e-06,
|
| 2787 |
+
"loss": 0.0779,
|
| 2788 |
+
"step": 9925
|
| 2789 |
+
},
|
| 2790 |
+
{
|
| 2791 |
+
"epoch": 0.8970429138117563,
|
| 2792 |
+
"grad_norm": 0.21197868883609772,
|
| 2793 |
+
"learning_rate": 1.0371047479420547e-06,
|
| 2794 |
+
"loss": 0.1272,
|
| 2795 |
+
"step": 9950
|
| 2796 |
+
},
|
| 2797 |
+
{
|
| 2798 |
+
"epoch": 0.899296790479625,
|
| 2799 |
+
"grad_norm": 0.1868308186531067,
|
| 2800 |
+
"learning_rate": 9.925696734830858e-07,
|
| 2801 |
+
"loss": 0.1221,
|
| 2802 |
+
"step": 9975
|
| 2803 |
+
},
|
| 2804 |
+
{
|
| 2805 |
+
"epoch": 0.9015506671474937,
|
| 2806 |
+
"grad_norm": 0.23166649043560028,
|
| 2807 |
+
"learning_rate": 9.489875743004795e-07,
|
| 2808 |
+
"loss": 0.0928,
|
| 2809 |
+
"step": 10000
|
| 2810 |
+
},
|
| 2811 |
+
{
|
| 2812 |
+
"epoch": 0.9038045438153625,
|
| 2813 |
+
"grad_norm": 0.037723857909440994,
|
| 2814 |
+
"learning_rate": 9.063606354693521e-07,
|
| 2815 |
+
"loss": 0.0648,
|
| 2816 |
+
"step": 10025
|
| 2817 |
+
},
|
| 2818 |
+
{
|
| 2819 |
+
"epoch": 0.9060584204832312,
|
| 2820 |
+
"grad_norm": 34.0538330078125,
|
| 2821 |
+
"learning_rate": 8.646909941759563e-07,
|
| 2822 |
+
"loss": 0.1063,
|
| 2823 |
+
"step": 10050
|
| 2824 |
+
},
|
| 2825 |
+
{
|
| 2826 |
+
"epoch": 0.9083122971510998,
|
| 2827 |
+
"grad_norm": 1.206143856048584,
|
| 2828 |
+
"learning_rate": 8.239807396105326e-07,
|
| 2829 |
+
"loss": 0.0714,
|
| 2830 |
+
"step": 10075
|
| 2831 |
+
},
|
| 2832 |
+
{
|
| 2833 |
+
"epoch": 0.9105661738189686,
|
| 2834 |
+
"grad_norm": 0.10259348154067993,
|
| 2835 |
+
"learning_rate": 7.842319128625608e-07,
|
| 2836 |
+
"loss": 0.0401,
|
| 2837 |
+
"step": 10100
|
| 2838 |
+
},
|
| 2839 |
+
{
|
| 2840 |
+
"epoch": 0.9128200504868373,
|
| 2841 |
+
"grad_norm": 0.3073846697807312,
|
| 2842 |
+
"learning_rate": 7.454465068184347e-07,
|
| 2843 |
+
"loss": 0.1283,
|
| 2844 |
+
"step": 10125
|
| 2845 |
+
},
|
| 2846 |
+
{
|
| 2847 |
+
"epoch": 0.9150739271547061,
|
| 2848 |
+
"grad_norm": 0.19339025020599365,
|
| 2849 |
+
"learning_rate": 7.076264660615196e-07,
|
| 2850 |
+
"loss": 0.0719,
|
| 2851 |
+
"step": 10150
|
| 2852 |
+
},
|
| 2853 |
+
{
|
| 2854 |
+
"epoch": 0.9173278038225748,
|
| 2855 |
+
"grad_norm": 0.9588416814804077,
|
| 2856 |
+
"learning_rate": 6.70773686774695e-07,
|
| 2857 |
+
"loss": 0.0899,
|
| 2858 |
+
"step": 10175
|
| 2859 |
+
},
|
| 2860 |
+
{
|
| 2861 |
+
"epoch": 0.9195816804904435,
|
| 2862 |
+
"grad_norm": 0.32579371333122253,
|
| 2863 |
+
"learning_rate": 6.348900166452555e-07,
|
| 2864 |
+
"loss": 0.0778,
|
| 2865 |
+
"step": 10200
|
| 2866 |
+
},
|
| 2867 |
+
{
|
| 2868 |
+
"epoch": 0.9218355571583123,
|
| 2869 |
+
"grad_norm": 0.567986249923706,
|
| 2870 |
+
"learning_rate": 5.999772547722748e-07,
|
| 2871 |
+
"loss": 0.0846,
|
| 2872 |
+
"step": 10225
|
| 2873 |
+
},
|
| 2874 |
+
{
|
| 2875 |
+
"epoch": 0.924089433826181,
|
| 2876 |
+
"grad_norm": 0.06987219303846359,
|
| 2877 |
+
"learning_rate": 5.660371515764306e-07,
|
| 2878 |
+
"loss": 0.0739,
|
| 2879 |
+
"step": 10250
|
| 2880 |
+
},
|
| 2881 |
+
{
|
| 2882 |
+
"epoch": 0.9263433104940497,
|
| 2883 |
+
"grad_norm": 0.05537739768624306,
|
| 2884 |
+
"learning_rate": 5.330714087122113e-07,
|
| 2885 |
+
"loss": 0.0888,
|
| 2886 |
+
"step": 10275
|
| 2887 |
+
},
|
| 2888 |
+
{
|
| 2889 |
+
"epoch": 0.9285971871619185,
|
| 2890 |
+
"grad_norm": 6.309244155883789,
|
| 2891 |
+
"learning_rate": 5.010816789826223e-07,
|
| 2892 |
+
"loss": 0.1257,
|
| 2893 |
+
"step": 10300
|
| 2894 |
+
},
|
| 2895 |
+
{
|
| 2896 |
+
"epoch": 0.9308510638297872,
|
| 2897 |
+
"grad_norm": 149.58358764648438,
|
| 2898 |
+
"learning_rate": 4.7006956625630596e-07,
|
| 2899 |
+
"loss": 0.0767,
|
| 2900 |
+
"step": 10325
|
| 2901 |
+
},
|
| 2902 |
+
{
|
| 2903 |
+
"epoch": 0.933104940497656,
|
| 2904 |
+
"grad_norm": 0.43270567059516907,
|
| 2905 |
+
"learning_rate": 4.4003662538714133e-07,
|
| 2906 |
+
"loss": 0.0576,
|
| 2907 |
+
"step": 10350
|
| 2908 |
+
},
|
| 2909 |
+
{
|
| 2910 |
+
"epoch": 0.9353588171655247,
|
| 2911 |
+
"grad_norm": 0.15286360681056976,
|
| 2912 |
+
"learning_rate": 4.1098436213627303e-07,
|
| 2913 |
+
"loss": 0.1321,
|
| 2914 |
+
"step": 10375
|
| 2915 |
+
},
|
| 2916 |
+
{
|
| 2917 |
+
"epoch": 0.9376126938333934,
|
| 2918 |
+
"grad_norm": 0.06981105357408524,
|
| 2919 |
+
"learning_rate": 3.829142330966318e-07,
|
| 2920 |
+
"loss": 0.0735,
|
| 2921 |
+
"step": 10400
|
| 2922 |
+
},
|
| 2923 |
+
{
|
| 2924 |
+
"epoch": 0.9398665705012622,
|
| 2925 |
+
"grad_norm": 276.96405029296875,
|
| 2926 |
+
"learning_rate": 3.5582764561989504e-07,
|
| 2927 |
+
"loss": 0.0565,
|
| 2928 |
+
"step": 10425
|
| 2929 |
+
},
|
| 2930 |
+
{
|
| 2931 |
+
"epoch": 0.9421204471691309,
|
| 2932 |
+
"grad_norm": 0.09452413767576218,
|
| 2933 |
+
"learning_rate": 3.297259577459322e-07,
|
| 2934 |
+
"loss": 0.1062,
|
| 2935 |
+
"step": 10450
|
| 2936 |
+
},
|
| 2937 |
+
{
|
| 2938 |
+
"epoch": 0.9443743238369996,
|
| 2939 |
+
"grad_norm": 83.56983947753906,
|
| 2940 |
+
"learning_rate": 3.046104781347148e-07,
|
| 2941 |
+
"loss": 0.2404,
|
| 2942 |
+
"step": 10475
|
| 2943 |
+
},
|
| 2944 |
+
{
|
| 2945 |
+
"epoch": 0.9466282005048684,
|
| 2946 |
+
"grad_norm": 0.1363532841205597,
|
| 2947 |
+
"learning_rate": 2.8048246600070217e-07,
|
| 2948 |
+
"loss": 0.058,
|
| 2949 |
+
"step": 10500
|
| 2950 |
+
},
|
| 2951 |
+
{
|
| 2952 |
+
"epoch": 0.9488820771727371,
|
| 2953 |
+
"grad_norm": 0.0371931754052639,
|
| 2954 |
+
"learning_rate": 2.573431310497143e-07,
|
| 2955 |
+
"loss": 0.1229,
|
| 2956 |
+
"step": 10525
|
| 2957 |
+
},
|
| 2958 |
+
{
|
| 2959 |
+
"epoch": 0.9511359538406059,
|
| 2960 |
+
"grad_norm": 0.08539276570081711,
|
| 2961 |
+
"learning_rate": 2.3519363341826696e-07,
|
| 2962 |
+
"loss": 0.1235,
|
| 2963 |
+
"step": 10550
|
| 2964 |
+
},
|
| 2965 |
+
{
|
| 2966 |
+
"epoch": 0.9533898305084746,
|
| 2967 |
+
"grad_norm": 0.0697568878531456,
|
| 2968 |
+
"learning_rate": 2.1403508361542924e-07,
|
| 2969 |
+
"loss": 0.149,
|
| 2970 |
+
"step": 10575
|
| 2971 |
+
},
|
| 2972 |
+
{
|
| 2973 |
+
"epoch": 0.9556437071763433,
|
| 2974 |
+
"grad_norm": 0.01937931217253208,
|
| 2975 |
+
"learning_rate": 1.9386854246712383e-07,
|
| 2976 |
+
"loss": 0.0301,
|
| 2977 |
+
"step": 10600
|
| 2978 |
+
},
|
| 2979 |
+
{
|
| 2980 |
+
"epoch": 0.9578975838442121,
|
| 2981 |
+
"grad_norm": 18.20211410522461,
|
| 2982 |
+
"learning_rate": 1.7469502106294721e-07,
|
| 2983 |
+
"loss": 0.2033,
|
| 2984 |
+
"step": 10625
|
| 2985 |
+
},
|
| 2986 |
+
{
|
| 2987 |
+
"epoch": 0.9601514605120808,
|
| 2988 |
+
"grad_norm": 0.6471598148345947,
|
| 2989 |
+
"learning_rate": 1.5651548070548362e-07,
|
| 2990 |
+
"loss": 0.0281,
|
| 2991 |
+
"step": 10650
|
| 2992 |
+
},
|
| 2993 |
+
{
|
| 2994 |
+
"epoch": 0.9624053371799495,
|
| 2995 |
+
"grad_norm": 0.04301556572318077,
|
| 2996 |
+
"learning_rate": 1.3933083286210348e-07,
|
| 2997 |
+
"loss": 0.1398,
|
| 2998 |
+
"step": 10675
|
| 2999 |
+
},
|
| 3000 |
+
{
|
| 3001 |
+
"epoch": 0.9646592138478183,
|
| 3002 |
+
"grad_norm": 0.14243409037590027,
|
| 3003 |
+
"learning_rate": 1.2314193911926231e-07,
|
| 3004 |
+
"loss": 0.1188,
|
| 3005 |
+
"step": 10700
|
| 3006 |
+
},
|
| 3007 |
+
{
|
| 3008 |
+
"epoch": 0.966913090515687,
|
| 3009 |
+
"grad_norm": 16.64614486694336,
|
| 3010 |
+
"learning_rate": 1.0794961113930192e-07,
|
| 3011 |
+
"loss": 0.0974,
|
| 3012 |
+
"step": 10725
|
| 3013 |
+
},
|
| 3014 |
+
{
|
| 3015 |
+
"epoch": 0.9691669671835557,
|
| 3016 |
+
"grad_norm": 20.02398681640625,
|
| 3017 |
+
"learning_rate": 9.375461061976732e-08,
|
| 3018 |
+
"loss": 0.0736,
|
| 3019 |
+
"step": 10750
|
| 3020 |
+
},
|
| 3021 |
+
{
|
| 3022 |
+
"epoch": 0.9714208438514245,
|
| 3023 |
+
"grad_norm": 0.11151763796806335,
|
| 3024 |
+
"learning_rate": 8.055764925520848e-08,
|
| 3025 |
+
"loss": 0.0243,
|
| 3026 |
+
"step": 10775
|
| 3027 |
+
},
|
| 3028 |
+
{
|
| 3029 |
+
"epoch": 0.9736747205192932,
|
| 3030 |
+
"grad_norm": 0.4667912423610687,
|
| 3031 |
+
"learning_rate": 6.83593887014955e-08,
|
| 3032 |
+
"loss": 0.1136,
|
| 3033 |
+
"step": 10800
|
| 3034 |
+
},
|
| 3035 |
+
{
|
| 3036 |
+
"epoch": 0.975928597187162,
|
| 3037 |
+
"grad_norm": 0.06938260793685913,
|
| 3038 |
+
"learning_rate": 5.716044054264958e-08,
|
| 3039 |
+
"loss": 0.0681,
|
| 3040 |
+
"step": 10825
|
| 3041 |
+
},
|
| 3042 |
+
{
|
| 3043 |
+
"epoch": 0.9781824738550307,
|
| 3044 |
+
"grad_norm": 0.23027046024799347,
|
| 3045 |
+
"learning_rate": 4.6961366260189765e-08,
|
| 3046 |
+
"loss": 0.0687,
|
| 3047 |
+
"step": 10850
|
| 3048 |
+
},
|
| 3049 |
+
{
|
| 3050 |
+
"epoch": 0.9804363505228993,
|
| 3051 |
+
"grad_norm": 0.14855755865573883,
|
| 3052 |
+
"learning_rate": 3.776267720495996e-08,
|
| 3053 |
+
"loss": 0.1064,
|
| 3054 |
+
"step": 10875
|
| 3055 |
+
},
|
| 3056 |
+
{
|
| 3057 |
+
"epoch": 0.9826902271907682,
|
| 3058 |
+
"grad_norm": 0.029531018808484077,
|
| 3059 |
+
"learning_rate": 2.9564834571509382e-08,
|
| 3060 |
+
"loss": 0.0905,
|
| 3061 |
+
"step": 10900
|
| 3062 |
+
},
|
| 3063 |
+
{
|
| 3064 |
+
"epoch": 0.9849441038586368,
|
| 3065 |
+
"grad_norm": 30.858125686645508,
|
| 3066 |
+
"learning_rate": 2.23682493749644e-08,
|
| 3067 |
+
"loss": 0.1379,
|
| 3068 |
+
"step": 10925
|
| 3069 |
+
},
|
| 3070 |
+
{
|
| 3071 |
+
"epoch": 0.9871979805265055,
|
| 3072 |
+
"grad_norm": 6.944602966308594,
|
| 3073 |
+
"learning_rate": 1.617328243041838e-08,
|
| 3074 |
+
"loss": 0.172,
|
| 3075 |
+
"step": 10950
|
| 3076 |
+
},
|
| 3077 |
+
{
|
| 3078 |
+
"epoch": 0.9894518571943743,
|
| 3079 |
+
"grad_norm": 0.41906285285949707,
|
| 3080 |
+
"learning_rate": 1.0980244334843903e-08,
|
| 3081 |
+
"loss": 0.0967,
|
| 3082 |
+
"step": 10975
|
| 3083 |
+
},
|
| 3084 |
+
{
|
| 3085 |
+
"epoch": 0.991705733862243,
|
| 3086 |
+
"grad_norm": 15.382137298583984,
|
| 3087 |
+
"learning_rate": 6.789395451520797e-09,
|
| 3088 |
+
"loss": 0.0971,
|
| 3089 |
+
"step": 11000
|
| 3090 |
+
},
|
| 3091 |
+
{
|
| 3092 |
+
"epoch": 0.9939596105301118,
|
| 3093 |
+
"grad_norm": 0.10896356403827667,
|
| 3094 |
+
"learning_rate": 3.600945896986563e-09,
|
| 3095 |
+
"loss": 0.094,
|
| 3096 |
+
"step": 11025
|
| 3097 |
+
},
|
| 3098 |
+
{
|
| 3099 |
+
"epoch": 0.9962134871979805,
|
| 3100 |
+
"grad_norm": 0.07236185669898987,
|
| 3101 |
+
"learning_rate": 1.4150555304914826e-09,
|
| 3102 |
+
"loss": 0.0409,
|
| 3103 |
+
"step": 11050
|
| 3104 |
+
},
|
| 3105 |
+
{
|
| 3106 |
+
"epoch": 0.9984673638658492,
|
| 3107 |
+
"grad_norm": 0.9604582190513611,
|
| 3108 |
+
"learning_rate": 2.3183394598724762e-10,
|
| 3109 |
+
"loss": 0.0902,
|
| 3110 |
+
"step": 11075
|
| 3111 |
+
},
|
| 3112 |
+
{
|
| 3113 |
+
"epoch": 1.0,
|
| 3114 |
+
"eval_accuracy": 0.9683073539874523,
|
| 3115 |
+
"eval_auc": 0.9916877882693556,
|
| 3116 |
+
"eval_f1": 0.9712272460364063,
|
| 3117 |
+
"eval_loss": 0.16320379078388214,
|
| 3118 |
+
"eval_precision": 0.9649941656942824,
|
| 3119 |
+
"eval_recall": 0.9775413711583925,
|
| 3120 |
+
"eval_runtime": 2210.9894,
|
| 3121 |
+
"eval_samples_per_second": 6.993,
|
| 3122 |
+
"eval_steps_per_second": 0.292,
|
| 3123 |
+
"step": 11092
|
| 3124 |
+
}
|
| 3125 |
+
],
|
| 3126 |
+
"logging_steps": 25,
|
| 3127 |
+
"max_steps": 11092,
|
| 3128 |
+
"num_input_tokens_seen": 0,
|
| 3129 |
+
"num_train_epochs": 1,
|
| 3130 |
+
"save_steps": 500,
|
| 3131 |
+
"stateful_callbacks": {
|
| 3132 |
+
"EarlyStoppingCallback": {
|
| 3133 |
+
"args": {
|
| 3134 |
+
"early_stopping_patience": 5,
|
| 3135 |
+
"early_stopping_threshold": 0.01
|
| 3136 |
+
},
|
| 3137 |
+
"attributes": {
|
| 3138 |
+
"early_stopping_patience_counter": 0
|
| 3139 |
+
}
|
| 3140 |
+
},
|
| 3141 |
+
"TrainerControl": {
|
| 3142 |
+
"args": {
|
| 3143 |
+
"should_epoch_stop": false,
|
| 3144 |
+
"should_evaluate": false,
|
| 3145 |
+
"should_log": false,
|
| 3146 |
+
"should_save": true,
|
| 3147 |
+
"should_training_stop": true
|
| 3148 |
+
},
|
| 3149 |
+
"attributes": {}
|
| 3150 |
+
}
|
| 3151 |
+
},
|
| 3152 |
+
"total_flos": 3.064790733409773e+19,
|
| 3153 |
+
"train_batch_size": 12,
|
| 3154 |
+
"trial_name": null,
|
| 3155 |
+
"trial_params": null
|
| 3156 |
+
}
|
{checkpoint-10763 → checkpoint-11092}/training_args.bin
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5304
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dd8e89257bd088f9e092a3b931fbaff08575a18da9a15b2cb0fcc14840e244fa
|
| 3 |
size 5304
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 780886280
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:eb2da3e13bdcac1efd85b67f172389f48a7d5955c139ee019e5b1af566969611
|
| 3 |
size 780886280
|
runs/{Dec04_07-28-34_f529bc929283/events.out.tfevents.1733297327.f529bc929283.82.0 → Dec07_05-45-42_74694ba9b4ba/events.out.tfevents.1733550355.74694ba9b4ba.167.0}
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e85e1b6d51efc931ac91c82422750af2f1cdd05210799dc124ae77eb129dc850
|
| 3 |
+
size 99954
|
runs/{Dec04_07-28-34_f529bc929283/events.out.tfevents.1733332341.f529bc929283.82.1 → Dec07_05-45-42_74694ba9b4ba/events.out.tfevents.1733585593.74694ba9b4ba.167.1}
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 607
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:449b7cf99ef0c2f199bcf32751a1a8b75ba34bafad3d262cb76297f3a4366f61
|
| 3 |
size 607
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5304
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dd8e89257bd088f9e092a3b931fbaff08575a18da9a15b2cb0fcc14840e244fa
|
| 3 |
size 5304
|
training_params.json
CHANGED
|
@@ -11,14 +11,14 @@
|
|
| 11 |
"scheduler": "cosine",
|
| 12 |
"weight_decay": 0.001,
|
| 13 |
"max_grad_norm": 1.0,
|
| 14 |
-
"seed":
|
| 15 |
"train_split": "train",
|
| 16 |
"valid_split": "test",
|
| 17 |
"logging_steps": -1,
|
| 18 |
-
"project_name": "autotrain-ai-image-detect-
|
| 19 |
"auto_find_batch_size": true,
|
| 20 |
"mixed_precision": "bf16",
|
| 21 |
-
"save_total_limit":
|
| 22 |
"push_to_hub": true,
|
| 23 |
"eval_strategy": "epoch",
|
| 24 |
"image_column": "image",
|
|
|
|
| 11 |
"scheduler": "cosine",
|
| 12 |
"weight_decay": 0.001,
|
| 13 |
"max_grad_norm": 1.0,
|
| 14 |
+
"seed": 7186,
|
| 15 |
"train_split": "train",
|
| 16 |
"valid_split": "test",
|
| 17 |
"logging_steps": -1,
|
| 18 |
+
"project_name": "autotrain-ai-image-detect-20241207-0028",
|
| 19 |
"auto_find_batch_size": true,
|
| 20 |
"mixed_precision": "bf16",
|
| 21 |
+
"save_total_limit": 1,
|
| 22 |
"push_to_hub": true,
|
| 23 |
"eval_strategy": "epoch",
|
| 24 |
"image_column": "image",
|