Model save
Browse files
last-checkpoint/config.json
DELETED
|
@@ -1,31 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"_name_or_path": "Nadav/PretrainedPHD-v2",
|
| 3 |
-
"architectures": [
|
| 4 |
-
"PIXELForPreTraining"
|
| 5 |
-
],
|
| 6 |
-
"attention_probs_dropout_prob": 0.1,
|
| 7 |
-
"decoder_hidden_size": 512,
|
| 8 |
-
"decoder_intermediate_size": 2048,
|
| 9 |
-
"decoder_num_attention_heads": 16,
|
| 10 |
-
"decoder_num_hidden_layers": 8,
|
| 11 |
-
"hidden_act": "gelu",
|
| 12 |
-
"hidden_dropout_prob": 0.1,
|
| 13 |
-
"hidden_size": 768,
|
| 14 |
-
"image_size": [
|
| 15 |
-
368,
|
| 16 |
-
368
|
| 17 |
-
],
|
| 18 |
-
"initializer_range": 0.02,
|
| 19 |
-
"intermediate_size": 3072,
|
| 20 |
-
"layer_norm_eps": 1e-12,
|
| 21 |
-
"mask_ratio": 0.29,
|
| 22 |
-
"model_type": "pixel",
|
| 23 |
-
"norm_pix_loss": true,
|
| 24 |
-
"num_attention_heads": 12,
|
| 25 |
-
"num_channels": 3,
|
| 26 |
-
"num_hidden_layers": 12,
|
| 27 |
-
"patch_size": 16,
|
| 28 |
-
"qkv_bias": true,
|
| 29 |
-
"torch_dtype": "float32",
|
| 30 |
-
"transformers_version": "4.17.0"
|
| 31 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/optimizer.pt
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:e0083228159a9684fe6abf7fc7f3682abd4e091c336c63a24ba76d9736a69586
|
| 3 |
-
size 893439185
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/pytorch_model.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:9e59ec613175c3d9dd17cb09bb4ec1c875973ee63862b916c52950e17b26a470
|
| 3 |
-
size 449471589
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/rng_state.pth
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:cccf7b9fd3b0a19dfeea5197c3f0bc5fb356c93fa129ae332264dec9fa909108
|
| 3 |
-
size 15459
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/scaler.pt
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:4db330b47705f38a44d9ca503524acd867a7f5b9c7d09cea471cc2d7ef717581
|
| 3 |
-
size 559
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/scheduler.pt
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:53825b8848113cf8bc0172552a5393f9bdd93287620ad2f38f87ba251b8fb391
|
| 3 |
-
size 623
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/trainer_state.json
DELETED
|
@@ -1,1376 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_metric": null,
|
| 3 |
-
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 7.964394532950291,
|
| 5 |
-
"global_step": 100000,
|
| 6 |
-
"is_hyper_param_search": false,
|
| 7 |
-
"is_local_process_zero": true,
|
| 8 |
-
"is_world_process_zero": true,
|
| 9 |
-
"log_history": [
|
| 10 |
-
{
|
| 11 |
-
"epoch": 0.04,
|
| 12 |
-
"learning_rate": 0.00010077565027123787,
|
| 13 |
-
"loss": 0.4732,
|
| 14 |
-
"step": 500
|
| 15 |
-
},
|
| 16 |
-
{
|
| 17 |
-
"epoch": 0.09,
|
| 18 |
-
"learning_rate": 0.00010075885246660077,
|
| 19 |
-
"loss": 0.4445,
|
| 20 |
-
"step": 1000
|
| 21 |
-
},
|
| 22 |
-
{
|
| 23 |
-
"epoch": 0.13,
|
| 24 |
-
"learning_rate": 0.00010073086073069567,
|
| 25 |
-
"loss": 0.4329,
|
| 26 |
-
"step": 1500
|
| 27 |
-
},
|
| 28 |
-
{
|
| 29 |
-
"epoch": 0.17,
|
| 30 |
-
"learning_rate": 0.0001006916819700645,
|
| 31 |
-
"loss": 0.4256,
|
| 32 |
-
"step": 2000
|
| 33 |
-
},
|
| 34 |
-
{
|
| 35 |
-
"epoch": 0.22,
|
| 36 |
-
"learning_rate": 0.0001006414377109213,
|
| 37 |
-
"loss": 0.421,
|
| 38 |
-
"step": 2500
|
| 39 |
-
},
|
| 40 |
-
{
|
| 41 |
-
"epoch": 0.26,
|
| 42 |
-
"learning_rate": 0.00010057993897463803,
|
| 43 |
-
"loss": 0.4167,
|
| 44 |
-
"step": 3000
|
| 45 |
-
},
|
| 46 |
-
{
|
| 47 |
-
"epoch": 0.3,
|
| 48 |
-
"learning_rate": 0.00010050729045131372,
|
| 49 |
-
"loss": 0.4132,
|
| 50 |
-
"step": 3500
|
| 51 |
-
},
|
| 52 |
-
{
|
| 53 |
-
"epoch": 0.35,
|
| 54 |
-
"learning_rate": 0.00010042351006588448,
|
| 55 |
-
"loss": 0.4107,
|
| 56 |
-
"step": 4000
|
| 57 |
-
},
|
| 58 |
-
{
|
| 59 |
-
"epoch": 0.39,
|
| 60 |
-
"learning_rate": 0.00010032861848990674,
|
| 61 |
-
"loss": 0.4091,
|
| 62 |
-
"step": 4500
|
| 63 |
-
},
|
| 64 |
-
{
|
| 65 |
-
"epoch": 0.44,
|
| 66 |
-
"learning_rate": 0.0001002228621438404,
|
| 67 |
-
"loss": 0.4067,
|
| 68 |
-
"step": 5000
|
| 69 |
-
},
|
| 70 |
-
{
|
| 71 |
-
"epoch": 0.44,
|
| 72 |
-
"eval_loss": 0.3877949118614197,
|
| 73 |
-
"eval_runtime": 108.0468,
|
| 74 |
-
"eval_samples_per_second": 46.276,
|
| 75 |
-
"eval_steps_per_second": 0.731,
|
| 76 |
-
"step": 5000
|
| 77 |
-
},
|
| 78 |
-
{
|
| 79 |
-
"epoch": 0.48,
|
| 80 |
-
"learning_rate": 0.00010010584325708389,
|
| 81 |
-
"loss": 0.4038,
|
| 82 |
-
"step": 5500
|
| 83 |
-
},
|
| 84 |
-
{
|
| 85 |
-
"epoch": 0.52,
|
| 86 |
-
"learning_rate": 9.997779155931062e-05,
|
| 87 |
-
"loss": 0.402,
|
| 88 |
-
"step": 6000
|
| 89 |
-
},
|
| 90 |
-
{
|
| 91 |
-
"epoch": 0.57,
|
| 92 |
-
"learning_rate": 9.983873864536092e-05,
|
| 93 |
-
"loss": 0.401,
|
| 94 |
-
"step": 6500
|
| 95 |
-
},
|
| 96 |
-
{
|
| 97 |
-
"epoch": 0.61,
|
| 98 |
-
"learning_rate": 9.968871882446063e-05,
|
| 99 |
-
"loss": 0.4001,
|
| 100 |
-
"step": 7000
|
| 101 |
-
},
|
| 102 |
-
{
|
| 103 |
-
"epoch": 0.65,
|
| 104 |
-
"learning_rate": 9.952776911175577e-05,
|
| 105 |
-
"loss": 0.3987,
|
| 106 |
-
"step": 7500
|
| 107 |
-
},
|
| 108 |
-
{
|
| 109 |
-
"epoch": 0.7,
|
| 110 |
-
"learning_rate": 9.935592921917959e-05,
|
| 111 |
-
"loss": 0.3974,
|
| 112 |
-
"step": 8000
|
| 113 |
-
},
|
| 114 |
-
{
|
| 115 |
-
"epoch": 0.74,
|
| 116 |
-
"learning_rate": 9.917361771757108e-05,
|
| 117 |
-
"loss": 0.3954,
|
| 118 |
-
"step": 8500
|
| 119 |
-
},
|
| 120 |
-
{
|
| 121 |
-
"epoch": 0.78,
|
| 122 |
-
"learning_rate": 9.898014889719868e-05,
|
| 123 |
-
"loss": 0.3945,
|
| 124 |
-
"step": 9000
|
| 125 |
-
},
|
| 126 |
-
{
|
| 127 |
-
"epoch": 0.83,
|
| 128 |
-
"learning_rate": 9.877592501404869e-05,
|
| 129 |
-
"loss": 0.394,
|
| 130 |
-
"step": 9500
|
| 131 |
-
},
|
| 132 |
-
{
|
| 133 |
-
"epoch": 0.87,
|
| 134 |
-
"learning_rate": 9.856099645730841e-05,
|
| 135 |
-
"loss": 0.3926,
|
| 136 |
-
"step": 10000
|
| 137 |
-
},
|
| 138 |
-
{
|
| 139 |
-
"epoch": 0.87,
|
| 140 |
-
"eval_loss": 0.37431180477142334,
|
| 141 |
-
"eval_runtime": 106.4392,
|
| 142 |
-
"eval_samples_per_second": 46.975,
|
| 143 |
-
"eval_steps_per_second": 0.742,
|
| 144 |
-
"step": 10000
|
| 145 |
-
},
|
| 146 |
-
{
|
| 147 |
-
"epoch": 0.04,
|
| 148 |
-
"learning_rate": 9.833541625738316e-05,
|
| 149 |
-
"loss": 0.3929,
|
| 150 |
-
"step": 10500
|
| 151 |
-
},
|
| 152 |
-
{
|
| 153 |
-
"epoch": 0.09,
|
| 154 |
-
"learning_rate": 9.809924007281187e-05,
|
| 155 |
-
"loss": 0.3906,
|
| 156 |
-
"step": 11000
|
| 157 |
-
},
|
| 158 |
-
{
|
| 159 |
-
"epoch": 0.13,
|
| 160 |
-
"learning_rate": 9.78525261765341e-05,
|
| 161 |
-
"loss": 0.3899,
|
| 162 |
-
"step": 11500
|
| 163 |
-
},
|
| 164 |
-
{
|
| 165 |
-
"epoch": 0.17,
|
| 166 |
-
"learning_rate": 9.759533544151208e-05,
|
| 167 |
-
"loss": 0.3892,
|
| 168 |
-
"step": 12000
|
| 169 |
-
},
|
| 170 |
-
{
|
| 171 |
-
"epoch": 0.22,
|
| 172 |
-
"learning_rate": 9.732773132571125e-05,
|
| 173 |
-
"loss": 0.3889,
|
| 174 |
-
"step": 12500
|
| 175 |
-
},
|
| 176 |
-
{
|
| 177 |
-
"epoch": 0.26,
|
| 178 |
-
"learning_rate": 9.705034604088048e-05,
|
| 179 |
-
"loss": 0.3865,
|
| 180 |
-
"step": 13000
|
| 181 |
-
},
|
| 182 |
-
{
|
| 183 |
-
"epoch": 0.3,
|
| 184 |
-
"learning_rate": 9.676213628592508e-05,
|
| 185 |
-
"loss": 0.3865,
|
| 186 |
-
"step": 13500
|
| 187 |
-
},
|
| 188 |
-
{
|
| 189 |
-
"epoch": 0.35,
|
| 190 |
-
"learning_rate": 9.64637187296151e-05,
|
| 191 |
-
"loss": 0.3867,
|
| 192 |
-
"step": 14000
|
| 193 |
-
},
|
| 194 |
-
{
|
| 195 |
-
"epoch": 0.39,
|
| 196 |
-
"learning_rate": 9.615516700201724e-05,
|
| 197 |
-
"loss": 0.3858,
|
| 198 |
-
"step": 14500
|
| 199 |
-
},
|
| 200 |
-
{
|
| 201 |
-
"epoch": 0.44,
|
| 202 |
-
"learning_rate": 9.583720443927501e-05,
|
| 203 |
-
"loss": 0.3853,
|
| 204 |
-
"step": 15000
|
| 205 |
-
},
|
| 206 |
-
{
|
| 207 |
-
"epoch": 0.44,
|
| 208 |
-
"eval_loss": 0.36506548523902893,
|
| 209 |
-
"eval_runtime": 96.1419,
|
| 210 |
-
"eval_samples_per_second": 52.006,
|
| 211 |
-
"eval_steps_per_second": 0.822,
|
| 212 |
-
"step": 15000
|
| 213 |
-
},
|
| 214 |
-
{
|
| 215 |
-
"epoch": 0.48,
|
| 216 |
-
"learning_rate": 9.550863512110018e-05,
|
| 217 |
-
"loss": 0.3844,
|
| 218 |
-
"step": 15500
|
| 219 |
-
},
|
| 220 |
-
{
|
| 221 |
-
"epoch": 0.52,
|
| 222 |
-
"learning_rate": 9.517016728422667e-05,
|
| 223 |
-
"loss": 0.3834,
|
| 224 |
-
"step": 16000
|
| 225 |
-
},
|
| 226 |
-
{
|
| 227 |
-
"epoch": 0.57,
|
| 228 |
-
"learning_rate": 9.482188444052858e-05,
|
| 229 |
-
"loss": 0.3826,
|
| 230 |
-
"step": 16500
|
| 231 |
-
},
|
| 232 |
-
{
|
| 233 |
-
"epoch": 0.61,
|
| 234 |
-
"learning_rate": 9.446387252358614e-05,
|
| 235 |
-
"loss": 0.3824,
|
| 236 |
-
"step": 17000
|
| 237 |
-
},
|
| 238 |
-
{
|
| 239 |
-
"epoch": 0.65,
|
| 240 |
-
"learning_rate": 9.40962198674828e-05,
|
| 241 |
-
"loss": 0.3825,
|
| 242 |
-
"step": 17500
|
| 243 |
-
},
|
| 244 |
-
{
|
| 245 |
-
"epoch": 0.7,
|
| 246 |
-
"learning_rate": 9.371901718501017e-05,
|
| 247 |
-
"loss": 0.3804,
|
| 248 |
-
"step": 18000
|
| 249 |
-
},
|
| 250 |
-
{
|
| 251 |
-
"epoch": 0.74,
|
| 252 |
-
"learning_rate": 9.333314023958391e-05,
|
| 253 |
-
"loss": 0.3803,
|
| 254 |
-
"step": 18500
|
| 255 |
-
},
|
| 256 |
-
{
|
| 257 |
-
"epoch": 0.78,
|
| 258 |
-
"learning_rate": 9.293713767144707e-05,
|
| 259 |
-
"loss": 0.381,
|
| 260 |
-
"step": 19000
|
| 261 |
-
},
|
| 262 |
-
{
|
| 263 |
-
"epoch": 0.83,
|
| 264 |
-
"learning_rate": 9.253187106312908e-05,
|
| 265 |
-
"loss": 0.3793,
|
| 266 |
-
"step": 19500
|
| 267 |
-
},
|
| 268 |
-
{
|
| 269 |
-
"epoch": 0.87,
|
| 270 |
-
"learning_rate": 9.211744040810141e-05,
|
| 271 |
-
"loss": 0.3793,
|
| 272 |
-
"step": 20000
|
| 273 |
-
},
|
| 274 |
-
{
|
| 275 |
-
"epoch": 0.87,
|
| 276 |
-
"eval_loss": 0.35968872904777527,
|
| 277 |
-
"eval_runtime": 95.0235,
|
| 278 |
-
"eval_samples_per_second": 52.619,
|
| 279 |
-
"eval_steps_per_second": 0.831,
|
| 280 |
-
"step": 20000
|
| 281 |
-
},
|
| 282 |
-
{
|
| 283 |
-
"epoch": 1.04,
|
| 284 |
-
"learning_rate": 9.169394796092694e-05,
|
| 285 |
-
"loss": 0.4465,
|
| 286 |
-
"step": 20500
|
| 287 |
-
},
|
| 288 |
-
{
|
| 289 |
-
"epoch": 1.09,
|
| 290 |
-
"learning_rate": 9.126149821203012e-05,
|
| 291 |
-
"loss": 0.444,
|
| 292 |
-
"step": 21000
|
| 293 |
-
},
|
| 294 |
-
{
|
| 295 |
-
"epoch": 1.13,
|
| 296 |
-
"learning_rate": 9.082019786191564e-05,
|
| 297 |
-
"loss": 0.4435,
|
| 298 |
-
"step": 21500
|
| 299 |
-
},
|
| 300 |
-
{
|
| 301 |
-
"epoch": 1.17,
|
| 302 |
-
"learning_rate": 9.037106452976014e-05,
|
| 303 |
-
"loss": 0.4422,
|
| 304 |
-
"step": 22000
|
| 305 |
-
},
|
| 306 |
-
{
|
| 307 |
-
"epoch": 1.22,
|
| 308 |
-
"learning_rate": 8.991240893598493e-05,
|
| 309 |
-
"loss": 0.4446,
|
| 310 |
-
"step": 22500
|
| 311 |
-
},
|
| 312 |
-
{
|
| 313 |
-
"epoch": 1.26,
|
| 314 |
-
"learning_rate": 8.944523560858406e-05,
|
| 315 |
-
"loss": 0.4418,
|
| 316 |
-
"step": 23000
|
| 317 |
-
},
|
| 318 |
-
{
|
| 319 |
-
"epoch": 1.3,
|
| 320 |
-
"learning_rate": 8.89696598155856e-05,
|
| 321 |
-
"loss": 0.4412,
|
| 322 |
-
"step": 23500
|
| 323 |
-
},
|
| 324 |
-
{
|
| 325 |
-
"epoch": 1.35,
|
| 326 |
-
"learning_rate": 8.848579889820028e-05,
|
| 327 |
-
"loss": 0.4414,
|
| 328 |
-
"step": 24000
|
| 329 |
-
},
|
| 330 |
-
{
|
| 331 |
-
"epoch": 1.39,
|
| 332 |
-
"learning_rate": 8.799377224186928e-05,
|
| 333 |
-
"loss": 0.4407,
|
| 334 |
-
"step": 24500
|
| 335 |
-
},
|
| 336 |
-
{
|
| 337 |
-
"epoch": 1.44,
|
| 338 |
-
"learning_rate": 8.749470933536528e-05,
|
| 339 |
-
"loss": 0.4402,
|
| 340 |
-
"step": 25000
|
| 341 |
-
},
|
| 342 |
-
{
|
| 343 |
-
"epoch": 1.44,
|
| 344 |
-
"eval_loss": 0.42174017429351807,
|
| 345 |
-
"eval_runtime": 91.5901,
|
| 346 |
-
"eval_samples_per_second": 54.591,
|
| 347 |
-
"eval_steps_per_second": 0.863,
|
| 348 |
-
"step": 25000
|
| 349 |
-
},
|
| 350 |
-
{
|
| 351 |
-
"epoch": 1.48,
|
| 352 |
-
"learning_rate": 8.698673310407346e-05,
|
| 353 |
-
"loss": 0.4396,
|
| 354 |
-
"step": 25500
|
| 355 |
-
},
|
| 356 |
-
{
|
| 357 |
-
"epoch": 1.52,
|
| 358 |
-
"learning_rate": 8.647096100588967e-05,
|
| 359 |
-
"loss": 0.4381,
|
| 360 |
-
"step": 26000
|
| 361 |
-
},
|
| 362 |
-
{
|
| 363 |
-
"epoch": 1.57,
|
| 364 |
-
"learning_rate": 8.594752029986149e-05,
|
| 365 |
-
"loss": 0.4382,
|
| 366 |
-
"step": 26500
|
| 367 |
-
},
|
| 368 |
-
{
|
| 369 |
-
"epoch": 1.61,
|
| 370 |
-
"learning_rate": 8.54165401371507e-05,
|
| 371 |
-
"loss": 0.4386,
|
| 372 |
-
"step": 27000
|
| 373 |
-
},
|
| 374 |
-
{
|
| 375 |
-
"epoch": 1.65,
|
| 376 |
-
"learning_rate": 8.487815152916725e-05,
|
| 377 |
-
"loss": 0.4376,
|
| 378 |
-
"step": 27500
|
| 379 |
-
},
|
| 380 |
-
{
|
| 381 |
-
"epoch": 1.7,
|
| 382 |
-
"learning_rate": 8.433248731524403e-05,
|
| 383 |
-
"loss": 0.4361,
|
| 384 |
-
"step": 28000
|
| 385 |
-
},
|
| 386 |
-
{
|
| 387 |
-
"epoch": 1.74,
|
| 388 |
-
"learning_rate": 8.377968212986092e-05,
|
| 389 |
-
"loss": 0.4369,
|
| 390 |
-
"step": 28500
|
| 391 |
-
},
|
| 392 |
-
{
|
| 393 |
-
"epoch": 1.78,
|
| 394 |
-
"learning_rate": 8.321987236942559e-05,
|
| 395 |
-
"loss": 0.4367,
|
| 396 |
-
"step": 29000
|
| 397 |
-
},
|
| 398 |
-
{
|
| 399 |
-
"epoch": 1.83,
|
| 400 |
-
"learning_rate": 8.265433627110443e-05,
|
| 401 |
-
"loss": 0.4345,
|
| 402 |
-
"step": 29500
|
| 403 |
-
},
|
| 404 |
-
{
|
| 405 |
-
"epoch": 1.87,
|
| 406 |
-
"learning_rate": 8.20809467414187e-05,
|
| 407 |
-
"loss": 0.4349,
|
| 408 |
-
"step": 30000
|
| 409 |
-
},
|
| 410 |
-
{
|
| 411 |
-
"epoch": 1.87,
|
| 412 |
-
"eval_loss": 0.41873496770858765,
|
| 413 |
-
"eval_runtime": 67.332,
|
| 414 |
-
"eval_samples_per_second": 74.259,
|
| 415 |
-
"eval_steps_per_second": 1.173,
|
| 416 |
-
"step": 30000
|
| 417 |
-
},
|
| 418 |
-
{
|
| 419 |
-
"epoch": 1.91,
|
| 420 |
-
"learning_rate": 8.150097177421913e-05,
|
| 421 |
-
"loss": 0.4351,
|
| 422 |
-
"step": 30500
|
| 423 |
-
},
|
| 424 |
-
{
|
| 425 |
-
"epoch": 1.96,
|
| 426 |
-
"learning_rate": 8.091455446965055e-05,
|
| 427 |
-
"loss": 0.4346,
|
| 428 |
-
"step": 31000
|
| 429 |
-
},
|
| 430 |
-
{
|
| 431 |
-
"epoch": 2.0,
|
| 432 |
-
"learning_rate": 8.032183951740807e-05,
|
| 433 |
-
"loss": 0.4348,
|
| 434 |
-
"step": 31500
|
| 435 |
-
},
|
| 436 |
-
{
|
| 437 |
-
"epoch": 2.04,
|
| 438 |
-
"learning_rate": 7.972417693488813e-05,
|
| 439 |
-
"loss": 0.4339,
|
| 440 |
-
"step": 32000
|
| 441 |
-
},
|
| 442 |
-
{
|
| 443 |
-
"epoch": 2.09,
|
| 444 |
-
"learning_rate": 7.911931879453039e-05,
|
| 445 |
-
"loss": 0.4321,
|
| 446 |
-
"step": 32500
|
| 447 |
-
},
|
| 448 |
-
{
|
| 449 |
-
"epoch": 2.13,
|
| 450 |
-
"learning_rate": 7.850860595403895e-05,
|
| 451 |
-
"loss": 0.4327,
|
| 452 |
-
"step": 33000
|
| 453 |
-
},
|
| 454 |
-
{
|
| 455 |
-
"epoch": 2.18,
|
| 456 |
-
"learning_rate": 7.789218909766887e-05,
|
| 457 |
-
"loss": 0.432,
|
| 458 |
-
"step": 33500
|
| 459 |
-
},
|
| 460 |
-
{
|
| 461 |
-
"epoch": 2.22,
|
| 462 |
-
"learning_rate": 7.727022031705582e-05,
|
| 463 |
-
"loss": 0.432,
|
| 464 |
-
"step": 34000
|
| 465 |
-
},
|
| 466 |
-
{
|
| 467 |
-
"epoch": 2.26,
|
| 468 |
-
"learning_rate": 7.664537309168187e-05,
|
| 469 |
-
"loss": 0.4318,
|
| 470 |
-
"step": 34500
|
| 471 |
-
},
|
| 472 |
-
{
|
| 473 |
-
"epoch": 2.31,
|
| 474 |
-
"learning_rate": 7.601278284364346e-05,
|
| 475 |
-
"loss": 0.4311,
|
| 476 |
-
"step": 35000
|
| 477 |
-
},
|
| 478 |
-
{
|
| 479 |
-
"epoch": 2.31,
|
| 480 |
-
"eval_loss": 0.4138348698616028,
|
| 481 |
-
"eval_runtime": 71.9462,
|
| 482 |
-
"eval_samples_per_second": 69.496,
|
| 483 |
-
"eval_steps_per_second": 1.098,
|
| 484 |
-
"step": 35000
|
| 485 |
-
},
|
| 486 |
-
{
|
| 487 |
-
"epoch": 2.35,
|
| 488 |
-
"learning_rate": 7.53751043867331e-05,
|
| 489 |
-
"loss": 0.4304,
|
| 490 |
-
"step": 35500
|
| 491 |
-
},
|
| 492 |
-
{
|
| 493 |
-
"epoch": 2.39,
|
| 494 |
-
"learning_rate": 7.473249505856825e-05,
|
| 495 |
-
"loss": 0.43,
|
| 496 |
-
"step": 36000
|
| 497 |
-
},
|
| 498 |
-
{
|
| 499 |
-
"epoch": 2.44,
|
| 500 |
-
"learning_rate": 7.408511341338508e-05,
|
| 501 |
-
"loss": 0.4295,
|
| 502 |
-
"step": 36500
|
| 503 |
-
},
|
| 504 |
-
{
|
| 505 |
-
"epoch": 2.48,
|
| 506 |
-
"learning_rate": 7.343311918291752e-05,
|
| 507 |
-
"loss": 0.43,
|
| 508 |
-
"step": 37000
|
| 509 |
-
},
|
| 510 |
-
{
|
| 511 |
-
"epoch": 2.52,
|
| 512 |
-
"learning_rate": 7.277667323698597e-05,
|
| 513 |
-
"loss": 0.43,
|
| 514 |
-
"step": 37500
|
| 515 |
-
},
|
| 516 |
-
{
|
| 517 |
-
"epoch": 2.57,
|
| 518 |
-
"learning_rate": 7.211593754380497e-05,
|
| 519 |
-
"loss": 0.429,
|
| 520 |
-
"step": 38000
|
| 521 |
-
},
|
| 522 |
-
{
|
| 523 |
-
"epoch": 2.61,
|
| 524 |
-
"learning_rate": 7.145107513002001e-05,
|
| 525 |
-
"loss": 0.4289,
|
| 526 |
-
"step": 38500
|
| 527 |
-
},
|
| 528 |
-
{
|
| 529 |
-
"epoch": 2.65,
|
| 530 |
-
"learning_rate": 7.078225004048314e-05,
|
| 531 |
-
"loss": 0.4275,
|
| 532 |
-
"step": 39000
|
| 533 |
-
},
|
| 534 |
-
{
|
| 535 |
-
"epoch": 2.7,
|
| 536 |
-
"learning_rate": 7.011097622316453e-05,
|
| 537 |
-
"loss": 0.428,
|
| 538 |
-
"step": 39500
|
| 539 |
-
},
|
| 540 |
-
{
|
| 541 |
-
"epoch": 2.74,
|
| 542 |
-
"learning_rate": 6.943472888404397e-05,
|
| 543 |
-
"loss": 0.4275,
|
| 544 |
-
"step": 40000
|
| 545 |
-
},
|
| 546 |
-
{
|
| 547 |
-
"epoch": 2.74,
|
| 548 |
-
"eval_loss": 0.4106511175632477,
|
| 549 |
-
"eval_runtime": 94.0729,
|
| 550 |
-
"eval_samples_per_second": 53.15,
|
| 551 |
-
"eval_steps_per_second": 0.84,
|
| 552 |
-
"step": 40000
|
| 553 |
-
},
|
| 554 |
-
{
|
| 555 |
-
"epoch": 2.78,
|
| 556 |
-
"learning_rate": 6.875501637243646e-05,
|
| 557 |
-
"loss": 0.4269,
|
| 558 |
-
"step": 40500
|
| 559 |
-
},
|
| 560 |
-
{
|
| 561 |
-
"epoch": 2.83,
|
| 562 |
-
"learning_rate": 6.807200639723347e-05,
|
| 563 |
-
"loss": 0.4256,
|
| 564 |
-
"step": 41000
|
| 565 |
-
},
|
| 566 |
-
{
|
| 567 |
-
"epoch": 2.87,
|
| 568 |
-
"learning_rate": 6.738586748092632e-05,
|
| 569 |
-
"loss": 0.4273,
|
| 570 |
-
"step": 41500
|
| 571 |
-
},
|
| 572 |
-
{
|
| 573 |
-
"epoch": 2.92,
|
| 574 |
-
"learning_rate": 6.669676891802565e-05,
|
| 575 |
-
"loss": 0.4255,
|
| 576 |
-
"step": 42000
|
| 577 |
-
},
|
| 578 |
-
{
|
| 579 |
-
"epoch": 2.96,
|
| 580 |
-
"learning_rate": 6.600626718034563e-05,
|
| 581 |
-
"loss": 0.4255,
|
| 582 |
-
"step": 42500
|
| 583 |
-
},
|
| 584 |
-
{
|
| 585 |
-
"epoch": 3.0,
|
| 586 |
-
"learning_rate": 6.531176515384238e-05,
|
| 587 |
-
"loss": 0.4265,
|
| 588 |
-
"step": 43000
|
| 589 |
-
},
|
| 590 |
-
{
|
| 591 |
-
"epoch": 3.05,
|
| 592 |
-
"learning_rate": 6.461621146309398e-05,
|
| 593 |
-
"loss": 0.4257,
|
| 594 |
-
"step": 43500
|
| 595 |
-
},
|
| 596 |
-
{
|
| 597 |
-
"epoch": 3.09,
|
| 598 |
-
"learning_rate": 6.39169899926586e-05,
|
| 599 |
-
"loss": 0.4242,
|
| 600 |
-
"step": 44000
|
| 601 |
-
},
|
| 602 |
-
{
|
| 603 |
-
"epoch": 3.13,
|
| 604 |
-
"learning_rate": 6.321566476924404e-05,
|
| 605 |
-
"loss": 0.4236,
|
| 606 |
-
"step": 44500
|
| 607 |
-
},
|
| 608 |
-
{
|
| 609 |
-
"epoch": 3.18,
|
| 610 |
-
"learning_rate": 6.2512408834355e-05,
|
| 611 |
-
"loss": 0.4245,
|
| 612 |
-
"step": 45000
|
| 613 |
-
},
|
| 614 |
-
{
|
| 615 |
-
"epoch": 3.18,
|
| 616 |
-
"eval_loss": 0.40728819370269775,
|
| 617 |
-
"eval_runtime": 61.7603,
|
| 618 |
-
"eval_samples_per_second": 80.958,
|
| 619 |
-
"eval_steps_per_second": 1.279,
|
| 620 |
-
"step": 45000
|
| 621 |
-
},
|
| 622 |
-
{
|
| 623 |
-
"epoch": 3.22,
|
| 624 |
-
"learning_rate": 6.180739570587037e-05,
|
| 625 |
-
"loss": 0.4251,
|
| 626 |
-
"step": 45500
|
| 627 |
-
},
|
| 628 |
-
{
|
| 629 |
-
"epoch": 3.26,
|
| 630 |
-
"learning_rate": 6.110079933523028e-05,
|
| 631 |
-
"loss": 0.4228,
|
| 632 |
-
"step": 46000
|
| 633 |
-
},
|
| 634 |
-
{
|
| 635 |
-
"epoch": 3.31,
|
| 636 |
-
"learning_rate": 6.039279406451614e-05,
|
| 637 |
-
"loss": 0.4228,
|
| 638 |
-
"step": 46500
|
| 639 |
-
},
|
| 640 |
-
{
|
| 641 |
-
"epoch": 3.35,
|
| 642 |
-
"learning_rate": 5.968355458343437e-05,
|
| 643 |
-
"loss": 0.422,
|
| 644 |
-
"step": 47000
|
| 645 |
-
},
|
| 646 |
-
{
|
| 647 |
-
"epoch": 3.39,
|
| 648 |
-
"learning_rate": 5.89732558862144e-05,
|
| 649 |
-
"loss": 0.4224,
|
| 650 |
-
"step": 47500
|
| 651 |
-
},
|
| 652 |
-
{
|
| 653 |
-
"epoch": 3.44,
|
| 654 |
-
"learning_rate": 5.826207322843129e-05,
|
| 655 |
-
"loss": 0.4221,
|
| 656 |
-
"step": 48000
|
| 657 |
-
},
|
| 658 |
-
{
|
| 659 |
-
"epoch": 3.48,
|
| 660 |
-
"learning_rate": 5.755160645640033e-05,
|
| 661 |
-
"loss": 0.4215,
|
| 662 |
-
"step": 48500
|
| 663 |
-
},
|
| 664 |
-
{
|
| 665 |
-
"epoch": 3.52,
|
| 666 |
-
"learning_rate": 5.6839183363622005e-05,
|
| 667 |
-
"loss": 0.4212,
|
| 668 |
-
"step": 49000
|
| 669 |
-
},
|
| 670 |
-
{
|
| 671 |
-
"epoch": 3.57,
|
| 672 |
-
"learning_rate": 5.612640286074268e-05,
|
| 673 |
-
"loss": 0.4204,
|
| 674 |
-
"step": 49500
|
| 675 |
-
},
|
| 676 |
-
{
|
| 677 |
-
"epoch": 3.61,
|
| 678 |
-
"learning_rate": 5.5413440815685914e-05,
|
| 679 |
-
"loss": 0.4213,
|
| 680 |
-
"step": 50000
|
| 681 |
-
},
|
| 682 |
-
{
|
| 683 |
-
"epoch": 3.61,
|
| 684 |
-
"eval_loss": 0.4042368531227112,
|
| 685 |
-
"eval_runtime": 65.4775,
|
| 686 |
-
"eval_samples_per_second": 76.362,
|
| 687 |
-
"eval_steps_per_second": 1.207,
|
| 688 |
-
"step": 50000
|
| 689 |
-
},
|
| 690 |
-
{
|
| 691 |
-
"epoch": 3.66,
|
| 692 |
-
"learning_rate": 5.4701898965206674e-05,
|
| 693 |
-
"loss": 0.4199,
|
| 694 |
-
"step": 50500
|
| 695 |
-
},
|
| 696 |
-
{
|
| 697 |
-
"epoch": 3.7,
|
| 698 |
-
"learning_rate": 5.398910105922212e-05,
|
| 699 |
-
"loss": 0.4199,
|
| 700 |
-
"step": 51000
|
| 701 |
-
},
|
| 702 |
-
{
|
| 703 |
-
"epoch": 3.74,
|
| 704 |
-
"learning_rate": 5.327807340052843e-05,
|
| 705 |
-
"loss": 0.4194,
|
| 706 |
-
"step": 51500
|
| 707 |
-
},
|
| 708 |
-
{
|
| 709 |
-
"epoch": 3.79,
|
| 710 |
-
"learning_rate": 5.25661416730201e-05,
|
| 711 |
-
"loss": 0.4189,
|
| 712 |
-
"step": 52000
|
| 713 |
-
},
|
| 714 |
-
{
|
| 715 |
-
"epoch": 3.83,
|
| 716 |
-
"learning_rate": 5.185490684450904e-05,
|
| 717 |
-
"loss": 0.4188,
|
| 718 |
-
"step": 52500
|
| 719 |
-
},
|
| 720 |
-
{
|
| 721 |
-
"epoch": 3.87,
|
| 722 |
-
"learning_rate": 5.114454440154674e-05,
|
| 723 |
-
"loss": 0.4199,
|
| 724 |
-
"step": 53000
|
| 725 |
-
},
|
| 726 |
-
{
|
| 727 |
-
"epoch": 3.92,
|
| 728 |
-
"learning_rate": 5.043522961543667e-05,
|
| 729 |
-
"loss": 0.4186,
|
| 730 |
-
"step": 53500
|
| 731 |
-
},
|
| 732 |
-
{
|
| 733 |
-
"epoch": 3.96,
|
| 734 |
-
"learning_rate": 4.972713749898855e-05,
|
| 735 |
-
"loss": 0.418,
|
| 736 |
-
"step": 54000
|
| 737 |
-
},
|
| 738 |
-
{
|
| 739 |
-
"epoch": 4.0,
|
| 740 |
-
"learning_rate": 4.902185464226454e-05,
|
| 741 |
-
"loss": 0.419,
|
| 742 |
-
"step": 54500
|
| 743 |
-
},
|
| 744 |
-
{
|
| 745 |
-
"epoch": 4.05,
|
| 746 |
-
"learning_rate": 4.831672833649923e-05,
|
| 747 |
-
"loss": 0.418,
|
| 748 |
-
"step": 55000
|
| 749 |
-
},
|
| 750 |
-
{
|
| 751 |
-
"epoch": 4.05,
|
| 752 |
-
"eval_loss": 0.40235766768455505,
|
| 753 |
-
"eval_runtime": 50.0847,
|
| 754 |
-
"eval_samples_per_second": 99.831,
|
| 755 |
-
"eval_steps_per_second": 1.577,
|
| 756 |
-
"step": 55000
|
| 757 |
-
},
|
| 758 |
-
{
|
| 759 |
-
"epoch": 4.09,
|
| 760 |
-
"learning_rate": 4.761334740888664e-05,
|
| 761 |
-
"loss": 0.4173,
|
| 762 |
-
"step": 55500
|
| 763 |
-
},
|
| 764 |
-
{
|
| 765 |
-
"epoch": 4.13,
|
| 766 |
-
"learning_rate": 4.691188540814576e-05,
|
| 767 |
-
"loss": 0.4176,
|
| 768 |
-
"step": 56000
|
| 769 |
-
},
|
| 770 |
-
{
|
| 771 |
-
"epoch": 4.18,
|
| 772 |
-
"learning_rate": 4.621391194692349e-05,
|
| 773 |
-
"loss": 0.4166,
|
| 774 |
-
"step": 56500
|
| 775 |
-
},
|
| 776 |
-
{
|
| 777 |
-
"epoch": 4.22,
|
| 778 |
-
"learning_rate": 4.551680180854116e-05,
|
| 779 |
-
"loss": 0.4182,
|
| 780 |
-
"step": 57000
|
| 781 |
-
},
|
| 782 |
-
{
|
| 783 |
-
"epoch": 4.26,
|
| 784 |
-
"learning_rate": 4.482212788828615e-05,
|
| 785 |
-
"loss": 0.4169,
|
| 786 |
-
"step": 57500
|
| 787 |
-
},
|
| 788 |
-
{
|
| 789 |
-
"epoch": 4.31,
|
| 790 |
-
"learning_rate": 4.413006158655363e-05,
|
| 791 |
-
"loss": 0.4158,
|
| 792 |
-
"step": 58000
|
| 793 |
-
},
|
| 794 |
-
{
|
| 795 |
-
"epoch": 4.35,
|
| 796 |
-
"learning_rate": 4.3440773660347955e-05,
|
| 797 |
-
"loss": 0.4151,
|
| 798 |
-
"step": 58500
|
| 799 |
-
},
|
| 800 |
-
{
|
| 801 |
-
"epoch": 4.4,
|
| 802 |
-
"learning_rate": 4.275443418115087e-05,
|
| 803 |
-
"loss": 0.4146,
|
| 804 |
-
"step": 59000
|
| 805 |
-
},
|
| 806 |
-
{
|
| 807 |
-
"epoch": 4.44,
|
| 808 |
-
"learning_rate": 4.207121249295896e-05,
|
| 809 |
-
"loss": 0.4152,
|
| 810 |
-
"step": 59500
|
| 811 |
-
},
|
| 812 |
-
{
|
| 813 |
-
"epoch": 4.48,
|
| 814 |
-
"learning_rate": 4.139127717050051e-05,
|
| 815 |
-
"loss": 0.4143,
|
| 816 |
-
"step": 60000
|
| 817 |
-
},
|
| 818 |
-
{
|
| 819 |
-
"epoch": 4.48,
|
| 820 |
-
"eval_loss": 0.39910149574279785,
|
| 821 |
-
"eval_runtime": 73.8959,
|
| 822 |
-
"eval_samples_per_second": 67.663,
|
| 823 |
-
"eval_steps_per_second": 1.069,
|
| 824 |
-
"step": 60000
|
| 825 |
-
},
|
| 826 |
-
{
|
| 827 |
-
"epoch": 4.53,
|
| 828 |
-
"learning_rate": 4.0714795977642255e-05,
|
| 829 |
-
"loss": 0.4153,
|
| 830 |
-
"step": 60500
|
| 831 |
-
},
|
| 832 |
-
{
|
| 833 |
-
"epoch": 4.57,
|
| 834 |
-
"learning_rate": 4.004327782200046e-05,
|
| 835 |
-
"loss": 0.4146,
|
| 836 |
-
"step": 61000
|
| 837 |
-
},
|
| 838 |
-
{
|
| 839 |
-
"epoch": 4.61,
|
| 840 |
-
"learning_rate": 3.937419699055001e-05,
|
| 841 |
-
"loss": 0.4138,
|
| 842 |
-
"step": 61500
|
| 843 |
-
},
|
| 844 |
-
{
|
| 845 |
-
"epoch": 4.66,
|
| 846 |
-
"learning_rate": 3.8709067973051936e-05,
|
| 847 |
-
"loss": 0.4137,
|
| 848 |
-
"step": 62000
|
| 849 |
-
},
|
| 850 |
-
{
|
| 851 |
-
"epoch": 4.7,
|
| 852 |
-
"learning_rate": 3.804805488013876e-05,
|
| 853 |
-
"loss": 0.4131,
|
| 854 |
-
"step": 62500
|
| 855 |
-
},
|
| 856 |
-
{
|
| 857 |
-
"epoch": 4.74,
|
| 858 |
-
"learning_rate": 3.739132080690021e-05,
|
| 859 |
-
"loss": 0.4141,
|
| 860 |
-
"step": 63000
|
| 861 |
-
},
|
| 862 |
-
{
|
| 863 |
-
"epoch": 4.79,
|
| 864 |
-
"learning_rate": 3.673902779264194e-05,
|
| 865 |
-
"loss": 0.4123,
|
| 866 |
-
"step": 63500
|
| 867 |
-
},
|
| 868 |
-
{
|
| 869 |
-
"epoch": 4.83,
|
| 870 |
-
"learning_rate": 3.609133678090475e-05,
|
| 871 |
-
"loss": 0.4126,
|
| 872 |
-
"step": 64000
|
| 873 |
-
},
|
| 874 |
-
{
|
| 875 |
-
"epoch": 4.87,
|
| 876 |
-
"learning_rate": 3.544840757975416e-05,
|
| 877 |
-
"loss": 0.4124,
|
| 878 |
-
"step": 64500
|
| 879 |
-
},
|
| 880 |
-
{
|
| 881 |
-
"epoch": 4.92,
|
| 882 |
-
"learning_rate": 3.481039882235021e-05,
|
| 883 |
-
"loss": 0.4124,
|
| 884 |
-
"step": 65000
|
| 885 |
-
},
|
| 886 |
-
{
|
| 887 |
-
"epoch": 4.92,
|
| 888 |
-
"eval_loss": 0.3964463174343109,
|
| 889 |
-
"eval_runtime": 74.8784,
|
| 890 |
-
"eval_samples_per_second": 66.775,
|
| 891 |
-
"eval_steps_per_second": 1.055,
|
| 892 |
-
"step": 65000
|
| 893 |
-
},
|
| 894 |
-
{
|
| 895 |
-
"epoch": 4.96,
|
| 896 |
-
"learning_rate": 3.4177467927807086e-05,
|
| 897 |
-
"loss": 0.412,
|
| 898 |
-
"step": 65500
|
| 899 |
-
},
|
| 900 |
-
{
|
| 901 |
-
"epoch": 5.0,
|
| 902 |
-
"learning_rate": 3.3549771062352226e-05,
|
| 903 |
-
"loss": 0.4128,
|
| 904 |
-
"step": 66000
|
| 905 |
-
},
|
| 906 |
-
{
|
| 907 |
-
"epoch": 5.05,
|
| 908 |
-
"learning_rate": 3.2928702236310036e-05,
|
| 909 |
-
"loss": 0.411,
|
| 910 |
-
"step": 66500
|
| 911 |
-
},
|
| 912 |
-
{
|
| 913 |
-
"epoch": 5.09,
|
| 914 |
-
"learning_rate": 3.231192548659413e-05,
|
| 915 |
-
"loss": 0.4108,
|
| 916 |
-
"step": 67000
|
| 917 |
-
},
|
| 918 |
-
{
|
| 919 |
-
"epoch": 5.14,
|
| 920 |
-
"learning_rate": 3.170084306064917e-05,
|
| 921 |
-
"loss": 0.4106,
|
| 922 |
-
"step": 67500
|
| 923 |
-
},
|
| 924 |
-
{
|
| 925 |
-
"epoch": 5.18,
|
| 926 |
-
"learning_rate": 3.1096810275663516e-05,
|
| 927 |
-
"loss": 0.4112,
|
| 928 |
-
"step": 68000
|
| 929 |
-
},
|
| 930 |
-
{
|
| 931 |
-
"epoch": 5.22,
|
| 932 |
-
"learning_rate": 3.0497555244481445e-05,
|
| 933 |
-
"loss": 0.4106,
|
| 934 |
-
"step": 68500
|
| 935 |
-
},
|
| 936 |
-
{
|
| 937 |
-
"epoch": 5.27,
|
| 938 |
-
"learning_rate": 2.9904442205778553e-05,
|
| 939 |
-
"loss": 0.4106,
|
| 940 |
-
"step": 69000
|
| 941 |
-
},
|
| 942 |
-
{
|
| 943 |
-
"epoch": 5.31,
|
| 944 |
-
"learning_rate": 2.9317617501322188e-05,
|
| 945 |
-
"loss": 0.4104,
|
| 946 |
-
"step": 69500
|
| 947 |
-
},
|
| 948 |
-
{
|
| 949 |
-
"epoch": 5.35,
|
| 950 |
-
"learning_rate": 2.873722592132734e-05,
|
| 951 |
-
"loss": 0.4096,
|
| 952 |
-
"step": 70000
|
| 953 |
-
},
|
| 954 |
-
{
|
| 955 |
-
"epoch": 5.35,
|
| 956 |
-
"eval_loss": 0.3943169414997101,
|
| 957 |
-
"eval_runtime": 90.0414,
|
| 958 |
-
"eval_samples_per_second": 55.53,
|
| 959 |
-
"eval_steps_per_second": 0.877,
|
| 960 |
-
"step": 70000
|
| 961 |
-
},
|
| 962 |
-
{
|
| 963 |
-
"epoch": 5.4,
|
| 964 |
-
"learning_rate": 2.816341066873177e-05,
|
| 965 |
-
"loss": 0.4098,
|
| 966 |
-
"step": 70500
|
| 967 |
-
},
|
| 968 |
-
{
|
| 969 |
-
"epoch": 5.44,
|
| 970 |
-
"learning_rate": 2.75974407207991e-05,
|
| 971 |
-
"loss": 0.4089,
|
| 972 |
-
"step": 71000
|
| 973 |
-
},
|
| 974 |
-
{
|
| 975 |
-
"epoch": 5.48,
|
| 976 |
-
"learning_rate": 2.7037187352256104e-05,
|
| 977 |
-
"loss": 0.4089,
|
| 978 |
-
"step": 71500
|
| 979 |
-
},
|
| 980 |
-
{
|
| 981 |
-
"epoch": 5.53,
|
| 982 |
-
"learning_rate": 2.6483929770191072e-05,
|
| 983 |
-
"loss": 0.4098,
|
| 984 |
-
"step": 72000
|
| 985 |
-
},
|
| 986 |
-
{
|
| 987 |
-
"epoch": 5.57,
|
| 988 |
-
"learning_rate": 2.5937804482633846e-05,
|
| 989 |
-
"loss": 0.4098,
|
| 990 |
-
"step": 72500
|
| 991 |
-
},
|
| 992 |
-
{
|
| 993 |
-
"epoch": 5.61,
|
| 994 |
-
"learning_rate": 2.5400016613135313e-05,
|
| 995 |
-
"loss": 0.4082,
|
| 996 |
-
"step": 73000
|
| 997 |
-
},
|
| 998 |
-
{
|
| 999 |
-
"epoch": 5.66,
|
| 1000 |
-
"learning_rate": 2.486854343481112e-05,
|
| 1001 |
-
"loss": 0.4086,
|
| 1002 |
-
"step": 73500
|
| 1003 |
-
},
|
| 1004 |
-
{
|
| 1005 |
-
"epoch": 5.7,
|
| 1006 |
-
"learning_rate": 2.4344601123401418e-05,
|
| 1007 |
-
"loss": 0.408,
|
| 1008 |
-
"step": 74000
|
| 1009 |
-
},
|
| 1010 |
-
{
|
| 1011 |
-
"epoch": 5.74,
|
| 1012 |
-
"learning_rate": 2.3828318953831648e-05,
|
| 1013 |
-
"loss": 0.409,
|
| 1014 |
-
"step": 74500
|
| 1015 |
-
},
|
| 1016 |
-
{
|
| 1017 |
-
"epoch": 5.79,
|
| 1018 |
-
"learning_rate": 2.3320833444621267e-05,
|
| 1019 |
-
"loss": 0.4079,
|
| 1020 |
-
"step": 75000
|
| 1021 |
-
},
|
| 1022 |
-
{
|
| 1023 |
-
"epoch": 5.79,
|
| 1024 |
-
"eval_loss": 0.39236390590667725,
|
| 1025 |
-
"eval_runtime": 57.2971,
|
| 1026 |
-
"eval_samples_per_second": 87.264,
|
| 1027 |
-
"eval_steps_per_second": 1.379,
|
| 1028 |
-
"step": 75000
|
| 1029 |
-
},
|
| 1030 |
-
{
|
| 1031 |
-
"epoch": 5.83,
|
| 1032 |
-
"learning_rate": 2.2820235842075488e-05,
|
| 1033 |
-
"loss": 0.4076,
|
| 1034 |
-
"step": 75500
|
| 1035 |
-
},
|
| 1036 |
-
{
|
| 1037 |
-
"epoch": 5.88,
|
| 1038 |
-
"learning_rate": 2.232767449569324e-05,
|
| 1039 |
-
"loss": 0.408,
|
| 1040 |
-
"step": 76000
|
| 1041 |
-
},
|
| 1042 |
-
{
|
| 1043 |
-
"epoch": 5.92,
|
| 1044 |
-
"learning_rate": 2.1843270937616385e-05,
|
| 1045 |
-
"loss": 0.4075,
|
| 1046 |
-
"step": 76500
|
| 1047 |
-
},
|
| 1048 |
-
{
|
| 1049 |
-
"epoch": 5.96,
|
| 1050 |
-
"learning_rate": 2.136808860049628e-05,
|
| 1051 |
-
"loss": 0.4075,
|
| 1052 |
-
"step": 77000
|
| 1053 |
-
},
|
| 1054 |
-
{
|
| 1055 |
-
"epoch": 6.01,
|
| 1056 |
-
"learning_rate": 2.090034022927794e-05,
|
| 1057 |
-
"loss": 0.4083,
|
| 1058 |
-
"step": 77500
|
| 1059 |
-
},
|
| 1060 |
-
{
|
| 1061 |
-
"epoch": 6.05,
|
| 1062 |
-
"learning_rate": 2.044110181973758e-05,
|
| 1063 |
-
"loss": 0.4072,
|
| 1064 |
-
"step": 78000
|
| 1065 |
-
},
|
| 1066 |
-
{
|
| 1067 |
-
"epoch": 6.09,
|
| 1068 |
-
"learning_rate": 1.9990486682081012e-05,
|
| 1069 |
-
"loss": 0.4066,
|
| 1070 |
-
"step": 78500
|
| 1071 |
-
},
|
| 1072 |
-
{
|
| 1073 |
-
"epoch": 6.14,
|
| 1074 |
-
"learning_rate": 1.9549480970403115e-05,
|
| 1075 |
-
"loss": 0.4061,
|
| 1076 |
-
"step": 79000
|
| 1077 |
-
},
|
| 1078 |
-
{
|
| 1079 |
-
"epoch": 6.18,
|
| 1080 |
-
"learning_rate": 1.911642597471847e-05,
|
| 1081 |
-
"loss": 0.4062,
|
| 1082 |
-
"step": 79500
|
| 1083 |
-
},
|
| 1084 |
-
{
|
| 1085 |
-
"epoch": 6.22,
|
| 1086 |
-
"learning_rate": 1.869232109486083e-05,
|
| 1087 |
-
"loss": 0.4071,
|
| 1088 |
-
"step": 80000
|
| 1089 |
-
},
|
| 1090 |
-
{
|
| 1091 |
-
"epoch": 6.22,
|
| 1092 |
-
"eval_loss": 0.39048197865486145,
|
| 1093 |
-
"eval_runtime": 72.8664,
|
| 1094 |
-
"eval_samples_per_second": 68.619,
|
| 1095 |
-
"eval_steps_per_second": 1.084,
|
| 1096 |
-
"step": 80000
|
| 1097 |
-
},
|
| 1098 |
-
{
|
| 1099 |
-
"epoch": 6.27,
|
| 1100 |
-
"learning_rate": 1.8277270972363276e-05,
|
| 1101 |
-
"loss": 0.4063,
|
| 1102 |
-
"step": 80500
|
| 1103 |
-
},
|
| 1104 |
-
{
|
| 1105 |
-
"epoch": 6.31,
|
| 1106 |
-
"learning_rate": 1.787137801463301e-05,
|
| 1107 |
-
"loss": 0.4054,
|
| 1108 |
-
"step": 81000
|
| 1109 |
-
},
|
| 1110 |
-
{
|
| 1111 |
-
"epoch": 6.35,
|
| 1112 |
-
"learning_rate": 1.7474742369683822e-05,
|
| 1113 |
-
"loss": 0.4058,
|
| 1114 |
-
"step": 81500
|
| 1115 |
-
},
|
| 1116 |
-
{
|
| 1117 |
-
"epoch": 6.4,
|
| 1118 |
-
"learning_rate": 1.7087461901426146e-05,
|
| 1119 |
-
"loss": 0.4046,
|
| 1120 |
-
"step": 82000
|
| 1121 |
-
},
|
| 1122 |
-
{
|
| 1123 |
-
"epoch": 6.44,
|
| 1124 |
-
"learning_rate": 1.670963216552051e-05,
|
| 1125 |
-
"loss": 0.4057,
|
| 1126 |
-
"step": 82500
|
| 1127 |
-
},
|
| 1128 |
-
{
|
| 1129 |
-
"epoch": 6.48,
|
| 1130 |
-
"learning_rate": 1.634134638580067e-05,
|
| 1131 |
-
"loss": 0.4044,
|
| 1132 |
-
"step": 83000
|
| 1133 |
-
},
|
| 1134 |
-
{
|
| 1135 |
-
"epoch": 6.53,
|
| 1136 |
-
"learning_rate": 1.5982695431271973e-05,
|
| 1137 |
-
"loss": 0.4059,
|
| 1138 |
-
"step": 83500
|
| 1139 |
-
},
|
| 1140 |
-
{
|
| 1141 |
-
"epoch": 6.57,
|
| 1142 |
-
"learning_rate": 1.563514402048906e-05,
|
| 1143 |
-
"loss": 0.4044,
|
| 1144 |
-
"step": 84000
|
| 1145 |
-
},
|
| 1146 |
-
{
|
| 1147 |
-
"epoch": 6.62,
|
| 1148 |
-
"learning_rate": 1.52959863865947e-05,
|
| 1149 |
-
"loss": 0.4042,
|
| 1150 |
-
"step": 84500
|
| 1151 |
-
},
|
| 1152 |
-
{
|
| 1153 |
-
"epoch": 6.66,
|
| 1154 |
-
"learning_rate": 1.496672150482924e-05,
|
| 1155 |
-
"loss": 0.4046,
|
| 1156 |
-
"step": 85000
|
| 1157 |
-
},
|
| 1158 |
-
{
|
| 1159 |
-
"epoch": 6.66,
|
| 1160 |
-
"eval_loss": 0.38925543427467346,
|
| 1161 |
-
"eval_runtime": 77.3103,
|
| 1162 |
-
"eval_samples_per_second": 64.674,
|
| 1163 |
-
"eval_steps_per_second": 1.022,
|
| 1164 |
-
"step": 85000
|
| 1165 |
-
},
|
| 1166 |
-
{
|
| 1167 |
-
"epoch": 6.7,
|
| 1168 |
-
"learning_rate": 1.4647430616375366e-05,
|
| 1169 |
-
"loss": 0.4042,
|
| 1170 |
-
"step": 85500
|
| 1171 |
-
},
|
| 1172 |
-
{
|
| 1173 |
-
"epoch": 6.75,
|
| 1174 |
-
"learning_rate": 1.433819250148215e-05,
|
| 1175 |
-
"loss": 0.4049,
|
| 1176 |
-
"step": 86000
|
| 1177 |
-
},
|
| 1178 |
-
{
|
| 1179 |
-
"epoch": 6.79,
|
| 1180 |
-
"learning_rate": 1.4039083460027203e-05,
|
| 1181 |
-
"loss": 0.4041,
|
| 1182 |
-
"step": 86500
|
| 1183 |
-
},
|
| 1184 |
-
{
|
| 1185 |
-
"epoch": 6.83,
|
| 1186 |
-
"learning_rate": 1.3750177292690856e-05,
|
| 1187 |
-
"loss": 0.4038,
|
| 1188 |
-
"step": 87000
|
| 1189 |
-
},
|
| 1190 |
-
{
|
| 1191 |
-
"epoch": 6.88,
|
| 1192 |
-
"learning_rate": 1.347154528274688e-05,
|
| 1193 |
-
"loss": 0.4047,
|
| 1194 |
-
"step": 87500
|
| 1195 |
-
},
|
| 1196 |
-
{
|
| 1197 |
-
"epoch": 6.92,
|
| 1198 |
-
"learning_rate": 1.3203782390038048e-05,
|
| 1199 |
-
"loss": 0.4036,
|
| 1200 |
-
"step": 88000
|
| 1201 |
-
},
|
| 1202 |
-
{
|
| 1203 |
-
"epoch": 6.96,
|
| 1204 |
-
"learning_rate": 1.2945881505195709e-05,
|
| 1205 |
-
"loss": 0.4044,
|
| 1206 |
-
"step": 88500
|
| 1207 |
-
},
|
| 1208 |
-
{
|
| 1209 |
-
"epoch": 7.01,
|
| 1210 |
-
"learning_rate": 1.2698453225696373e-05,
|
| 1211 |
-
"loss": 0.4045,
|
| 1212 |
-
"step": 89000
|
| 1213 |
-
},
|
| 1214 |
-
{
|
| 1215 |
-
"epoch": 7.05,
|
| 1216 |
-
"learning_rate": 1.2461558600765676e-05,
|
| 1217 |
-
"loss": 0.4031,
|
| 1218 |
-
"step": 89500
|
| 1219 |
-
},
|
| 1220 |
-
{
|
| 1221 |
-
"epoch": 7.09,
|
| 1222 |
-
"learning_rate": 1.2235256080607583e-05,
|
| 1223 |
-
"loss": 0.4026,
|
| 1224 |
-
"step": 90000
|
| 1225 |
-
},
|
| 1226 |
-
{
|
| 1227 |
-
"epoch": 7.09,
|
| 1228 |
-
"eval_loss": 0.3881285488605499,
|
| 1229 |
-
"eval_runtime": 133.2371,
|
| 1230 |
-
"eval_samples_per_second": 37.527,
|
| 1231 |
-
"eval_steps_per_second": 0.593,
|
| 1232 |
-
"step": 90000
|
| 1233 |
-
},
|
| 1234 |
-
{
|
| 1235 |
-
"epoch": 7.14,
|
| 1236 |
-
"learning_rate": 1.20196015019827e-05,
|
| 1237 |
-
"loss": 0.4036,
|
| 1238 |
-
"step": 90500
|
| 1239 |
-
},
|
| 1240 |
-
{
|
| 1241 |
-
"epoch": 7.18,
|
| 1242 |
-
"learning_rate": 1.1815047267485115e-05,
|
| 1243 |
-
"loss": 0.4037,
|
| 1244 |
-
"step": 91000
|
| 1245 |
-
},
|
| 1246 |
-
{
|
| 1247 |
-
"epoch": 7.22,
|
| 1248 |
-
"learning_rate": 1.162082400805151e-05,
|
| 1249 |
-
"loss": 0.4048,
|
| 1250 |
-
"step": 91500
|
| 1251 |
-
},
|
| 1252 |
-
{
|
| 1253 |
-
"epoch": 7.27,
|
| 1254 |
-
"learning_rate": 1.1437400292071077e-05,
|
| 1255 |
-
"loss": 0.4032,
|
| 1256 |
-
"step": 92000
|
| 1257 |
-
},
|
| 1258 |
-
{
|
| 1259 |
-
"epoch": 7.31,
|
| 1260 |
-
"learning_rate": 1.126482137660111e-05,
|
| 1261 |
-
"loss": 0.4032,
|
| 1262 |
-
"step": 92500
|
| 1263 |
-
},
|
| 1264 |
-
{
|
| 1265 |
-
"epoch": 7.36,
|
| 1266 |
-
"learning_rate": 1.1103129842906643e-05,
|
| 1267 |
-
"loss": 0.4029,
|
| 1268 |
-
"step": 93000
|
| 1269 |
-
},
|
| 1270 |
-
{
|
| 1271 |
-
"epoch": 7.4,
|
| 1272 |
-
"learning_rate": 1.0952365585954172e-05,
|
| 1273 |
-
"loss": 0.4023,
|
| 1274 |
-
"step": 93500
|
| 1275 |
-
},
|
| 1276 |
-
{
|
| 1277 |
-
"epoch": 7.44,
|
| 1278 |
-
"learning_rate": 1.0812565804568168e-05,
|
| 1279 |
-
"loss": 0.4027,
|
| 1280 |
-
"step": 94000
|
| 1281 |
-
},
|
| 1282 |
-
{
|
| 1283 |
-
"epoch": 7.49,
|
| 1284 |
-
"learning_rate": 1.0683764992252818e-05,
|
| 1285 |
-
"loss": 0.4022,
|
| 1286 |
-
"step": 94500
|
| 1287 |
-
},
|
| 1288 |
-
{
|
| 1289 |
-
"epoch": 7.53,
|
| 1290 |
-
"learning_rate": 1.0566219440353348e-05,
|
| 1291 |
-
"loss": 0.4027,
|
| 1292 |
-
"step": 95000
|
| 1293 |
-
},
|
| 1294 |
-
{
|
| 1295 |
-
"epoch": 7.53,
|
| 1296 |
-
"eval_loss": 0.38710081577301025,
|
| 1297 |
-
"eval_runtime": 65.4468,
|
| 1298 |
-
"eval_samples_per_second": 76.398,
|
| 1299 |
-
"eval_steps_per_second": 1.207,
|
| 1300 |
-
"step": 95000
|
| 1301 |
-
},
|
| 1302 |
-
{
|
| 1303 |
-
"epoch": 7.57,
|
| 1304 |
-
"learning_rate": 1.0459487036725839e-05,
|
| 1305 |
-
"loss": 0.4031,
|
| 1306 |
-
"step": 95500
|
| 1307 |
-
},
|
| 1308 |
-
{
|
| 1309 |
-
"epoch": 7.62,
|
| 1310 |
-
"learning_rate": 1.0363840719071717e-05,
|
| 1311 |
-
"loss": 0.4023,
|
| 1312 |
-
"step": 96000
|
| 1313 |
-
},
|
| 1314 |
-
{
|
| 1315 |
-
"epoch": 7.66,
|
| 1316 |
-
"learning_rate": 1.0279462058166865e-05,
|
| 1317 |
-
"loss": 0.4025,
|
| 1318 |
-
"step": 96500
|
| 1319 |
-
},
|
| 1320 |
-
{
|
| 1321 |
-
"epoch": 7.7,
|
| 1322 |
-
"learning_rate": 1.0206033689125313e-05,
|
| 1323 |
-
"loss": 0.4028,
|
| 1324 |
-
"step": 97000
|
| 1325 |
-
},
|
| 1326 |
-
{
|
| 1327 |
-
"epoch": 7.75,
|
| 1328 |
-
"learning_rate": 1.0143753941878168e-05,
|
| 1329 |
-
"loss": 0.4023,
|
| 1330 |
-
"step": 97500
|
| 1331 |
-
},
|
| 1332 |
-
{
|
| 1333 |
-
"epoch": 7.79,
|
| 1334 |
-
"learning_rate": 1.0092638183021144e-05,
|
| 1335 |
-
"loss": 0.4019,
|
| 1336 |
-
"step": 98000
|
| 1337 |
-
},
|
| 1338 |
-
{
|
| 1339 |
-
"epoch": 7.83,
|
| 1340 |
-
"learning_rate": 1.0052699024602892e-05,
|
| 1341 |
-
"loss": 0.4014,
|
| 1342 |
-
"step": 98500
|
| 1343 |
-
},
|
| 1344 |
-
{
|
| 1345 |
-
"epoch": 7.88,
|
| 1346 |
-
"learning_rate": 1.0023946321013112e-05,
|
| 1347 |
-
"loss": 0.4018,
|
| 1348 |
-
"step": 99000
|
| 1349 |
-
},
|
| 1350 |
-
{
|
| 1351 |
-
"epoch": 7.92,
|
| 1352 |
-
"learning_rate": 1.0006411110358469e-05,
|
| 1353 |
-
"loss": 0.4024,
|
| 1354 |
-
"step": 99500
|
| 1355 |
-
},
|
| 1356 |
-
{
|
| 1357 |
-
"epoch": 7.96,
|
| 1358 |
-
"learning_rate": 1.0000027439232365e-05,
|
| 1359 |
-
"loss": 0.4024,
|
| 1360 |
-
"step": 100000
|
| 1361 |
-
},
|
| 1362 |
-
{
|
| 1363 |
-
"epoch": 7.96,
|
| 1364 |
-
"eval_loss": 0.38628044724464417,
|
| 1365 |
-
"eval_runtime": 91.7648,
|
| 1366 |
-
"eval_samples_per_second": 54.487,
|
| 1367 |
-
"eval_steps_per_second": 0.861,
|
| 1368 |
-
"step": 100000
|
| 1369 |
-
}
|
| 1370 |
-
],
|
| 1371 |
-
"max_steps": 100000,
|
| 1372 |
-
"num_train_epochs": 9,
|
| 1373 |
-
"total_flos": 4.7098350596970145e+21,
|
| 1374 |
-
"trial_name": null,
|
| 1375 |
-
"trial_params": null
|
| 1376 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/training_args.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:da26d9c8c1c726588ec3b1e697e00057dd4bc580fad19e58afb508d1a16ec382
|
| 3 |
-
size 5615
|
|
|
|
|
|
|
|
|
|
|
|