Delete folder shared/checkpoints/latest with huggingface_hub
Browse files
shared/checkpoints/latest/config.json
DELETED
|
@@ -1,24 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"architectures": [
|
| 3 |
-
"OllaiForCausalLM"
|
| 4 |
-
],
|
| 5 |
-
"attention_dropout": 0.0,
|
| 6 |
-
"dtype": "float32",
|
| 7 |
-
"hidden_dropout": 0.0,
|
| 8 |
-
"hidden_size": 768,
|
| 9 |
-
"initializer_range": 0.02,
|
| 10 |
-
"intermediate_size": 3072,
|
| 11 |
-
"max_position_embeddings": 128,
|
| 12 |
-
"model_type": "ollai",
|
| 13 |
-
"neftune_alpha": 0.0,
|
| 14 |
-
"num_attention_heads": 12,
|
| 15 |
-
"num_hidden_layers": 8,
|
| 16 |
-
"num_key_value_heads": 6,
|
| 17 |
-
"qk_norm": true,
|
| 18 |
-
"rezero_init": 1.0,
|
| 19 |
-
"rms_norm_eps": 1e-06,
|
| 20 |
-
"rope_theta": 10000.0,
|
| 21 |
-
"transformers_version": "5.0.0",
|
| 22 |
-
"use_cache": false,
|
| 23 |
-
"vocab_size": 32000
|
| 24 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
shared/checkpoints/latest/model.safetensors
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:2f89bfb7f7677f84a98c35f52f3f72edfb9b094e0860d8191bad183ce312fdaf
|
| 3 |
-
size 498786448
|
|
|
|
|
|
|
|
|
|
|
|
shared/checkpoints/latest/trainer_state.json
DELETED
|
@@ -1,454 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_global_step": null,
|
| 3 |
-
"best_metric": null,
|
| 4 |
-
"best_model_checkpoint": null,
|
| 5 |
-
"epoch": 0.4175413118577684,
|
| 6 |
-
"eval_steps": 500,
|
| 7 |
-
"global_step": 1186,
|
| 8 |
-
"is_hyper_param_search": false,
|
| 9 |
-
"is_local_process_zero": true,
|
| 10 |
-
"is_world_process_zero": true,
|
| 11 |
-
"log_history": [
|
| 12 |
-
{
|
| 13 |
-
"epoch": 0.0003520584417013224,
|
| 14 |
-
"grad_norm": 118.3223648071289,
|
| 15 |
-
"learning_rate": 5e-05,
|
| 16 |
-
"loss": 256.8507080078125,
|
| 17 |
-
"step": 1
|
| 18 |
-
},
|
| 19 |
-
{
|
| 20 |
-
"epoch": 0.007041168834026448,
|
| 21 |
-
"grad_norm": 751.973388671875,
|
| 22 |
-
"learning_rate": 4.998468123209239e-05,
|
| 23 |
-
"loss": 430.4501182154605,
|
| 24 |
-
"step": 20
|
| 25 |
-
},
|
| 26 |
-
{
|
| 27 |
-
"epoch": 0.014082337668052896,
|
| 28 |
-
"grad_norm": 773.4529418945312,
|
| 29 |
-
"learning_rate": 4.9935478670313695e-05,
|
| 30 |
-
"loss": 292.0135498046875,
|
| 31 |
-
"step": 40
|
| 32 |
-
},
|
| 33 |
-
{
|
| 34 |
-
"epoch": 0.021123506502079344,
|
| 35 |
-
"grad_norm": 488.6176452636719,
|
| 36 |
-
"learning_rate": 4.985241668732203e-05,
|
| 37 |
-
"loss": 290.304443359375,
|
| 38 |
-
"step": 60
|
| 39 |
-
},
|
| 40 |
-
{
|
| 41 |
-
"epoch": 0.028164675336105793,
|
| 42 |
-
"grad_norm": 483.17303466796875,
|
| 43 |
-
"learning_rate": 4.9735608071434246e-05,
|
| 44 |
-
"loss": 278.869140625,
|
| 45 |
-
"step": 80
|
| 46 |
-
},
|
| 47 |
-
{
|
| 48 |
-
"epoch": 0.03520584417013224,
|
| 49 |
-
"grad_norm": 393.8902893066406,
|
| 50 |
-
"learning_rate": 4.958521143489032e-05,
|
| 51 |
-
"loss": 276.3343017578125,
|
| 52 |
-
"step": 100
|
| 53 |
-
},
|
| 54 |
-
{
|
| 55 |
-
"epoch": 0.04224701300415869,
|
| 56 |
-
"grad_norm": 281.7361755371094,
|
| 57 |
-
"learning_rate": 4.940143099847674e-05,
|
| 58 |
-
"loss": 254.011474609375,
|
| 59 |
-
"step": 120
|
| 60 |
-
},
|
| 61 |
-
{
|
| 62 |
-
"epoch": 0.04928818183818514,
|
| 63 |
-
"grad_norm": 380.287841796875,
|
| 64 |
-
"learning_rate": 4.9184516314218934e-05,
|
| 65 |
-
"loss": 294.28583984375,
|
| 66 |
-
"step": 140
|
| 67 |
-
},
|
| 68 |
-
{
|
| 69 |
-
"epoch": 0.056329350672211585,
|
| 70 |
-
"grad_norm": 458.5362548828125,
|
| 71 |
-
"learning_rate": 4.8934761926519235e-05,
|
| 72 |
-
"loss": 241.399560546875,
|
| 73 |
-
"step": 160
|
| 74 |
-
},
|
| 75 |
-
{
|
| 76 |
-
"epoch": 0.06337051950623804,
|
| 77 |
-
"grad_norm": 425.85931396484375,
|
| 78 |
-
"learning_rate": 4.8652506972200495e-05,
|
| 79 |
-
"loss": 239.3857177734375,
|
| 80 |
-
"step": 180
|
| 81 |
-
},
|
| 82 |
-
{
|
| 83 |
-
"epoch": 0.07041168834026448,
|
| 84 |
-
"grad_norm": 539.6701049804688,
|
| 85 |
-
"learning_rate": 4.833813471999858e-05,
|
| 86 |
-
"loss": 242.390869140625,
|
| 87 |
-
"step": 200
|
| 88 |
-
},
|
| 89 |
-
{
|
| 90 |
-
"epoch": 0.07745285717429093,
|
| 91 |
-
"grad_norm": 484.08203125,
|
| 92 |
-
"learning_rate": 4.7992072050128834e-05,
|
| 93 |
-
"loss": 249.8130126953125,
|
| 94 |
-
"step": 220
|
| 95 |
-
},
|
| 96 |
-
{
|
| 97 |
-
"epoch": 0.08449402600831737,
|
| 98 |
-
"grad_norm": 358.29730224609375,
|
| 99 |
-
"learning_rate": 4.761478887463347e-05,
|
| 100 |
-
"loss": 229.807470703125,
|
| 101 |
-
"step": 240
|
| 102 |
-
},
|
| 103 |
-
{
|
| 104 |
-
"epoch": 0.09153519484234383,
|
| 105 |
-
"grad_norm": 303.1641540527344,
|
| 106 |
-
"learning_rate": 4.720679749929673e-05,
|
| 107 |
-
"loss": 254.2503173828125,
|
| 108 |
-
"step": 260
|
| 109 |
-
},
|
| 110 |
-
{
|
| 111 |
-
"epoch": 0.09857636367637028,
|
| 112 |
-
"grad_norm": 219.22288513183594,
|
| 113 |
-
"learning_rate": 4.6768651927994434e-05,
|
| 114 |
-
"loss": 219.1611328125,
|
| 115 |
-
"step": 280
|
| 116 |
-
},
|
| 117 |
-
{
|
| 118 |
-
"epoch": 0.10561753251039672,
|
| 119 |
-
"grad_norm": 226.80201721191406,
|
| 120 |
-
"learning_rate": 4.6300947110422445e-05,
|
| 121 |
-
"loss": 228.465625,
|
| 122 |
-
"step": 300
|
| 123 |
-
},
|
| 124 |
-
{
|
| 125 |
-
"epoch": 0.11265870134442317,
|
| 126 |
-
"grad_norm": 424.9483337402344,
|
| 127 |
-
"learning_rate": 4.58043181342256e-05,
|
| 128 |
-
"loss": 232.2102294921875,
|
| 129 |
-
"step": 320
|
| 130 |
-
},
|
| 131 |
-
{
|
| 132 |
-
"epoch": 0.11969987017844962,
|
| 133 |
-
"grad_norm": 381.19781494140625,
|
| 134 |
-
"learning_rate": 4.527943936262403e-05,
|
| 135 |
-
"loss": 217.78046875,
|
| 136 |
-
"step": 340
|
| 137 |
-
},
|
| 138 |
-
{
|
| 139 |
-
"epoch": 0.12674103901247608,
|
| 140 |
-
"grad_norm": 332.9507141113281,
|
| 141 |
-
"learning_rate": 4.472702351870794e-05,
|
| 142 |
-
"loss": 206.33056640625,
|
| 143 |
-
"step": 360
|
| 144 |
-
},
|
| 145 |
-
{
|
| 146 |
-
"epoch": 0.13378220784650252,
|
| 147 |
-
"grad_norm": 378.1746520996094,
|
| 148 |
-
"learning_rate": 4.414782071764419e-05,
|
| 149 |
-
"loss": 218.1360107421875,
|
| 150 |
-
"step": 380
|
| 151 |
-
},
|
| 152 |
-
{
|
| 153 |
-
"epoch": 0.14082337668052897,
|
| 154 |
-
"grad_norm": 461.5779113769531,
|
| 155 |
-
"learning_rate": 4.354261744810895e-05,
|
| 156 |
-
"loss": 186.132275390625,
|
| 157 |
-
"step": 400
|
| 158 |
-
},
|
| 159 |
-
{
|
| 160 |
-
"epoch": 0.1478645455145554,
|
| 161 |
-
"grad_norm": 430.0111999511719,
|
| 162 |
-
"learning_rate": 4.2912235504329374e-05,
|
| 163 |
-
"loss": 192.41707763671874,
|
| 164 |
-
"step": 420
|
| 165 |
-
},
|
| 166 |
-
{
|
| 167 |
-
"epoch": 0.15490571434858186,
|
| 168 |
-
"grad_norm": 293.7337951660156,
|
| 169 |
-
"learning_rate": 4.225753087018457e-05,
|
| 170 |
-
"loss": 232.38330078125,
|
| 171 |
-
"step": 440
|
| 172 |
-
},
|
| 173 |
-
{
|
| 174 |
-
"epoch": 0.1619468831826083,
|
| 175 |
-
"grad_norm": 458.7125244140625,
|
| 176 |
-
"learning_rate": 4.1579392556881034e-05,
|
| 177 |
-
"loss": 187.08692626953126,
|
| 178 |
-
"step": 460
|
| 179 |
-
},
|
| 180 |
-
{
|
| 181 |
-
"epoch": 0.16898805201663475,
|
| 182 |
-
"grad_norm": 422.7680969238281,
|
| 183 |
-
"learning_rate": 4.087874139578098e-05,
|
| 184 |
-
"loss": 179.99237060546875,
|
| 185 |
-
"step": 480
|
| 186 |
-
},
|
| 187 |
-
{
|
| 188 |
-
"epoch": 0.17602922085066122,
|
| 189 |
-
"grad_norm": 504.5588684082031,
|
| 190 |
-
"learning_rate": 4.015652878802253e-05,
|
| 191 |
-
"loss": 193.752587890625,
|
| 192 |
-
"step": 500
|
| 193 |
-
},
|
| 194 |
-
{
|
| 195 |
-
"epoch": 0.18307038968468767,
|
| 196 |
-
"grad_norm": 327.46630859375,
|
| 197 |
-
"learning_rate": 3.941373541262992e-05,
|
| 198 |
-
"loss": 194.81287841796876,
|
| 199 |
-
"step": 520
|
| 200 |
-
},
|
| 201 |
-
{
|
| 202 |
-
"epoch": 0.1901115585187141,
|
| 203 |
-
"grad_norm": 284.4969482421875,
|
| 204 |
-
"learning_rate": 3.865136989486776e-05,
|
| 205 |
-
"loss": 160.69073486328125,
|
| 206 |
-
"step": 540
|
| 207 |
-
},
|
| 208 |
-
{
|
| 209 |
-
"epoch": 0.19715272735274056,
|
| 210 |
-
"grad_norm": 295.3741455078125,
|
| 211 |
-
"learning_rate": 3.7870467436647564e-05,
|
| 212 |
-
"loss": 193.65595703125,
|
| 213 |
-
"step": 560
|
| 214 |
-
},
|
| 215 |
-
{
|
| 216 |
-
"epoch": 0.204193896186767,
|
| 217 |
-
"grad_norm": 280.7477111816406,
|
| 218 |
-
"learning_rate": 3.7072088410846407e-05,
|
| 219 |
-
"loss": 209.171728515625,
|
| 220 |
-
"step": 580
|
| 221 |
-
},
|
| 222 |
-
{
|
| 223 |
-
"epoch": 0.21123506502079345,
|
| 224 |
-
"grad_norm": 378.8021545410156,
|
| 225 |
-
"learning_rate": 3.62573169214464e-05,
|
| 226 |
-
"loss": 176.9102294921875,
|
| 227 |
-
"step": 600
|
| 228 |
-
},
|
| 229 |
-
{
|
| 230 |
-
"epoch": 0.2182762338548199,
|
| 231 |
-
"grad_norm": 254.55145263671875,
|
| 232 |
-
"learning_rate": 3.5427259331450104e-05,
|
| 233 |
-
"loss": 166.85062255859376,
|
| 234 |
-
"step": 620
|
| 235 |
-
},
|
| 236 |
-
{
|
| 237 |
-
"epoch": 0.22531740268884634,
|
| 238 |
-
"grad_norm": 214.0546875,
|
| 239 |
-
"learning_rate": 3.458304276057086e-05,
|
| 240 |
-
"loss": 162.18095703125,
|
| 241 |
-
"step": 640
|
| 242 |
-
},
|
| 243 |
-
{
|
| 244 |
-
"epoch": 0.2323585715228728,
|
| 245 |
-
"grad_norm": 289.06134033203125,
|
| 246 |
-
"learning_rate": 3.372581355473793e-05,
|
| 247 |
-
"loss": 175.399755859375,
|
| 248 |
-
"step": 660
|
| 249 |
-
},
|
| 250 |
-
{
|
| 251 |
-
"epoch": 0.23939974035689923,
|
| 252 |
-
"grad_norm": 310.7714538574219,
|
| 253 |
-
"learning_rate": 3.28567357294948e-05,
|
| 254 |
-
"loss": 141.58082275390626,
|
| 255 |
-
"step": 680
|
| 256 |
-
},
|
| 257 |
-
{
|
| 258 |
-
"epoch": 0.2464409091909257,
|
| 259 |
-
"grad_norm": 453.130859375,
|
| 260 |
-
"learning_rate": 3.197698938940418e-05,
|
| 261 |
-
"loss": 152.31658935546875,
|
| 262 |
-
"step": 700
|
| 263 |
-
},
|
| 264 |
-
{
|
| 265 |
-
"epoch": 0.25348207802495215,
|
| 266 |
-
"grad_norm": 203.76687622070312,
|
| 267 |
-
"learning_rate": 3.108776912560598e-05,
|
| 268 |
-
"loss": 153.4617919921875,
|
| 269 |
-
"step": 720
|
| 270 |
-
},
|
| 271 |
-
{
|
| 272 |
-
"epoch": 0.2605232468589786,
|
| 273 |
-
"grad_norm": 254.8419952392578,
|
| 274 |
-
"learning_rate": 3.0190282393704382e-05,
|
| 275 |
-
"loss": 151.0395751953125,
|
| 276 |
-
"step": 740
|
| 277 |
-
},
|
| 278 |
-
{
|
| 279 |
-
"epoch": 0.26756441569300504,
|
| 280 |
-
"grad_norm": 368.3388366699219,
|
| 281 |
-
"learning_rate": 2.9285747874186342e-05,
|
| 282 |
-
"loss": 178.66256103515624,
|
| 283 |
-
"step": 760
|
| 284 |
-
},
|
| 285 |
-
{
|
| 286 |
-
"epoch": 0.2746055845270315,
|
| 287 |
-
"grad_norm": 248.59840393066406,
|
| 288 |
-
"learning_rate": 2.837539381759813e-05,
|
| 289 |
-
"loss": 146.82396240234374,
|
| 290 |
-
"step": 780
|
| 291 |
-
},
|
| 292 |
-
{
|
| 293 |
-
"epoch": 0.28164675336105793,
|
| 294 |
-
"grad_norm": 329.29925537109375,
|
| 295 |
-
"learning_rate": 2.746045637672678e-05,
|
| 296 |
-
"loss": 135.89444580078126,
|
| 297 |
-
"step": 800
|
| 298 |
-
},
|
| 299 |
-
{
|
| 300 |
-
"epoch": 0.2886879221950844,
|
| 301 |
-
"grad_norm": 247.31663513183594,
|
| 302 |
-
"learning_rate": 2.6542177928051366e-05,
|
| 303 |
-
"loss": 151.56038818359374,
|
| 304 |
-
"step": 820
|
| 305 |
-
},
|
| 306 |
-
{
|
| 307 |
-
"epoch": 0.2957290910291108,
|
| 308 |
-
"grad_norm": 200.099609375,
|
| 309 |
-
"learning_rate": 2.5621805384743096e-05,
|
| 310 |
-
"loss": 148.2068359375,
|
| 311 |
-
"step": 840
|
| 312 |
-
},
|
| 313 |
-
{
|
| 314 |
-
"epoch": 0.30277025986313727,
|
| 315 |
-
"grad_norm": 199.73974609375,
|
| 316 |
-
"learning_rate": 2.470058850350521e-05,
|
| 317 |
-
"loss": 133.66234130859374,
|
| 318 |
-
"step": 860
|
| 319 |
-
},
|
| 320 |
-
{
|
| 321 |
-
"epoch": 0.3098114286971637,
|
| 322 |
-
"grad_norm": 143.12315368652344,
|
| 323 |
-
"learning_rate": 2.3779778187551667e-05,
|
| 324 |
-
"loss": 142.92933349609376,
|
| 325 |
-
"step": 880
|
| 326 |
-
},
|
| 327 |
-
{
|
| 328 |
-
"epoch": 0.31685259753119016,
|
| 329 |
-
"grad_norm": 148.53729248046875,
|
| 330 |
-
"learning_rate": 2.2860624788029013e-05,
|
| 331 |
-
"loss": 123.2794677734375,
|
| 332 |
-
"step": 900
|
| 333 |
-
},
|
| 334 |
-
{
|
| 335 |
-
"epoch": 0.3238937663652166,
|
| 336 |
-
"grad_norm": 230.73326110839844,
|
| 337 |
-
"learning_rate": 2.1944376406187877e-05,
|
| 338 |
-
"loss": 141.41287841796876,
|
| 339 |
-
"step": 920
|
| 340 |
-
},
|
| 341 |
-
{
|
| 342 |
-
"epoch": 0.33093493519924305,
|
| 343 |
-
"grad_norm": 245.84796142578125,
|
| 344 |
-
"learning_rate": 2.1032277198609625e-05,
|
| 345 |
-
"loss": 118.0623046875,
|
| 346 |
-
"step": 940
|
| 347 |
-
},
|
| 348 |
-
{
|
| 349 |
-
"epoch": 0.3379761040332695,
|
| 350 |
-
"grad_norm": 255.589111328125,
|
| 351 |
-
"learning_rate": 2.012556568778926e-05,
|
| 352 |
-
"loss": 117.141064453125,
|
| 353 |
-
"step": 960
|
| 354 |
-
},
|
| 355 |
-
{
|
| 356 |
-
"epoch": 0.345017272867296,
|
| 357 |
-
"grad_norm": 187.10047912597656,
|
| 358 |
-
"learning_rate": 1.9225473080368916e-05,
|
| 359 |
-
"loss": 121.210205078125,
|
| 360 |
-
"step": 980
|
| 361 |
-
},
|
| 362 |
-
{
|
| 363 |
-
"epoch": 0.35205844170132244,
|
| 364 |
-
"grad_norm": 185.55857849121094,
|
| 365 |
-
"learning_rate": 1.8333221595305296e-05,
|
| 366 |
-
"loss": 114.097265625,
|
| 367 |
-
"step": 1000
|
| 368 |
-
},
|
| 369 |
-
{
|
| 370 |
-
"epoch": 0.3590996105353489,
|
| 371 |
-
"grad_norm": 143.88343811035156,
|
| 372 |
-
"learning_rate": 1.745002280424137e-05,
|
| 373 |
-
"loss": 121.60416259765626,
|
| 374 |
-
"step": 1020
|
| 375 |
-
},
|
| 376 |
-
{
|
| 377 |
-
"epoch": 0.36614077936937534,
|
| 378 |
-
"grad_norm": 181.9886932373047,
|
| 379 |
-
"learning_rate": 1.6577075986335812e-05,
|
| 380 |
-
"loss": 128.1891845703125,
|
| 381 |
-
"step": 1040
|
| 382 |
-
},
|
| 383 |
-
{
|
| 384 |
-
"epoch": 0.3731819482034018,
|
| 385 |
-
"grad_norm": 162.10708618164062,
|
| 386 |
-
"learning_rate": 1.571556649978427e-05,
|
| 387 |
-
"loss": 117.58974609375,
|
| 388 |
-
"step": 1060
|
| 389 |
-
},
|
| 390 |
-
{
|
| 391 |
-
"epoch": 0.3802231170374282,
|
| 392 |
-
"grad_norm": 173.69859313964844,
|
| 393 |
-
"learning_rate": 1.4866664172243558e-05,
|
| 394 |
-
"loss": 122.39468994140626,
|
| 395 |
-
"step": 1080
|
| 396 |
-
},
|
| 397 |
-
{
|
| 398 |
-
"epoch": 0.38726428587145467,
|
| 399 |
-
"grad_norm": 169.0511016845703,
|
| 400 |
-
"learning_rate": 1.4031521712344487e-05,
|
| 401 |
-
"loss": 109.45206298828126,
|
| 402 |
-
"step": 1100
|
| 403 |
-
},
|
| 404 |
-
{
|
| 405 |
-
"epoch": 0.3943054547054811,
|
| 406 |
-
"grad_norm": 140.88890075683594,
|
| 407 |
-
"learning_rate": 1.3211273144450337e-05,
|
| 408 |
-
"loss": 117.5877197265625,
|
| 409 |
-
"step": 1120
|
| 410 |
-
},
|
| 411 |
-
{
|
| 412 |
-
"epoch": 0.40134662353950756,
|
| 413 |
-
"grad_norm": 152.3916778564453,
|
| 414 |
-
"learning_rate": 1.2407032268786367e-05,
|
| 415 |
-
"loss": 116.1720458984375,
|
| 416 |
-
"step": 1140
|
| 417 |
-
},
|
| 418 |
-
{
|
| 419 |
-
"epoch": 0.408387792373534,
|
| 420 |
-
"grad_norm": 137.79103088378906,
|
| 421 |
-
"learning_rate": 1.1619891149031245e-05,
|
| 422 |
-
"loss": 123.202978515625,
|
| 423 |
-
"step": 1160
|
| 424 |
-
},
|
| 425 |
-
{
|
| 426 |
-
"epoch": 0.41542896120756045,
|
| 427 |
-
"grad_norm": 176.51348876953125,
|
| 428 |
-
"learning_rate": 1.0850918629424127e-05,
|
| 429 |
-
"loss": 109.8859130859375,
|
| 430 |
-
"step": 1180
|
| 431 |
-
}
|
| 432 |
-
],
|
| 433 |
-
"logging_steps": 20,
|
| 434 |
-
"max_steps": 1705,
|
| 435 |
-
"num_input_tokens_seen": 0,
|
| 436 |
-
"num_train_epochs": 1,
|
| 437 |
-
"save_steps": 400,
|
| 438 |
-
"stateful_callbacks": {
|
| 439 |
-
"TrainerControl": {
|
| 440 |
-
"args": {
|
| 441 |
-
"should_epoch_stop": false,
|
| 442 |
-
"should_evaluate": false,
|
| 443 |
-
"should_log": false,
|
| 444 |
-
"should_save": true,
|
| 445 |
-
"should_training_stop": false
|
| 446 |
-
},
|
| 447 |
-
"attributes": {}
|
| 448 |
-
}
|
| 449 |
-
},
|
| 450 |
-
"total_flos": 8754395228798976.0,
|
| 451 |
-
"train_batch_size": 1,
|
| 452 |
-
"trial_name": null,
|
| 453 |
-
"trial_params": null
|
| 454 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|