Model save
Browse files
last-checkpoint/config.json
DELETED
|
@@ -1,31 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"_name_or_path": "Nadav/PretrainedPHD",
|
| 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 sha256:6e4e20c1ef2ac6843c886418e98a47bae638ea7d617febd310a65b1910a8f8e4
|
| 3 |
-
size 893439185
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/pytorch_model.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:65fa8e0f8a447fcc5954aa678e61448bcfe8f8387d2caed08d0f1179f60750d7
|
| 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:fc8cc6cb0c14f159f66977ee3dfa2ab864df7c0e19869ebf5c6d7b0393c02b8d
|
| 3 |
-
size 15587
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/scaler.pt
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:b0c7a6adc45a0eea76cb785076f51f02ab2ddc692d12cb942437e9bf8e23bc08
|
| 3 |
-
size 559
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/scheduler.pt
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:de21128fb4d2d9fbd6335f650a62e3e1299cfe449b8f64957937f253cda36cc0
|
| 3 |
-
size 623
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/trainer_state.json
DELETED
|
@@ -1,2576 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_metric": null,
|
| 3 |
-
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.05,
|
| 5 |
-
"global_step": 200000,
|
| 6 |
-
"is_hyper_param_search": false,
|
| 7 |
-
"is_local_process_zero": true,
|
| 8 |
-
"is_world_process_zero": true,
|
| 9 |
-
"log_history": [
|
| 10 |
-
{
|
| 11 |
-
"epoch": 0.0,
|
| 12 |
-
"learning_rate": 0.00010077985004622052,
|
| 13 |
-
"loss": 0.5399,
|
| 14 |
-
"step": 500
|
| 15 |
-
},
|
| 16 |
-
{
|
| 17 |
-
"epoch": 0.01,
|
| 18 |
-
"learning_rate": 0.00010077565027123787,
|
| 19 |
-
"loss": 0.4951,
|
| 20 |
-
"step": 1000
|
| 21 |
-
},
|
| 22 |
-
{
|
| 23 |
-
"epoch": 0.01,
|
| 24 |
-
"learning_rate": 0.00010076865093411392,
|
| 25 |
-
"loss": 0.487,
|
| 26 |
-
"step": 1500
|
| 27 |
-
},
|
| 28 |
-
{
|
| 29 |
-
"epoch": 0.01,
|
| 30 |
-
"learning_rate": 0.00010075885246660077,
|
| 31 |
-
"loss": 0.4806,
|
| 32 |
-
"step": 2000
|
| 33 |
-
},
|
| 34 |
-
{
|
| 35 |
-
"epoch": 0.01,
|
| 36 |
-
"learning_rate": 0.00010074625547311406,
|
| 37 |
-
"loss": 0.4778,
|
| 38 |
-
"step": 2500
|
| 39 |
-
},
|
| 40 |
-
{
|
| 41 |
-
"epoch": 0.01,
|
| 42 |
-
"learning_rate": 0.00010073086073069567,
|
| 43 |
-
"loss": 0.4747,
|
| 44 |
-
"step": 3000
|
| 45 |
-
},
|
| 46 |
-
{
|
| 47 |
-
"epoch": 0.02,
|
| 48 |
-
"learning_rate": 0.00010071266918896582,
|
| 49 |
-
"loss": 0.4732,
|
| 50 |
-
"step": 3500
|
| 51 |
-
},
|
| 52 |
-
{
|
| 53 |
-
"epoch": 0.02,
|
| 54 |
-
"learning_rate": 0.0001006916819700645,
|
| 55 |
-
"loss": 0.4724,
|
| 56 |
-
"step": 4000
|
| 57 |
-
},
|
| 58 |
-
{
|
| 59 |
-
"epoch": 0.02,
|
| 60 |
-
"learning_rate": 0.00010066790036858225,
|
| 61 |
-
"loss": 0.4708,
|
| 62 |
-
"step": 4500
|
| 63 |
-
},
|
| 64 |
-
{
|
| 65 |
-
"epoch": 0.03,
|
| 66 |
-
"learning_rate": 0.00010064132585148025,
|
| 67 |
-
"loss": 0.4677,
|
| 68 |
-
"step": 5000
|
| 69 |
-
},
|
| 70 |
-
{
|
| 71 |
-
"epoch": 0.03,
|
| 72 |
-
"eval_loss": 0.4321456551551819,
|
| 73 |
-
"eval_runtime": 334.7353,
|
| 74 |
-
"eval_samples_per_second": 128.46,
|
| 75 |
-
"eval_steps_per_second": 2.008,
|
| 76 |
-
"step": 5000
|
| 77 |
-
},
|
| 78 |
-
{
|
| 79 |
-
"epoch": 0.03,
|
| 80 |
-
"learning_rate": 0.0001006119600579999,
|
| 81 |
-
"loss": 0.4673,
|
| 82 |
-
"step": 5500
|
| 83 |
-
},
|
| 84 |
-
{
|
| 85 |
-
"epoch": 0.03,
|
| 86 |
-
"learning_rate": 0.00010057980479956167,
|
| 87 |
-
"loss": 0.4649,
|
| 88 |
-
"step": 6000
|
| 89 |
-
},
|
| 90 |
-
{
|
| 91 |
-
"epoch": 0.03,
|
| 92 |
-
"learning_rate": 0.00010054486205965335,
|
| 93 |
-
"loss": 0.465,
|
| 94 |
-
"step": 6500
|
| 95 |
-
},
|
| 96 |
-
{
|
| 97 |
-
"epoch": 0.04,
|
| 98 |
-
"learning_rate": 0.00010050713399370776,
|
| 99 |
-
"loss": 0.4644,
|
| 100 |
-
"step": 7000
|
| 101 |
-
},
|
| 102 |
-
{
|
| 103 |
-
"epoch": 0.04,
|
| 104 |
-
"learning_rate": 0.00010046662292896969,
|
| 105 |
-
"loss": 0.4639,
|
| 106 |
-
"step": 7500
|
| 107 |
-
},
|
| 108 |
-
{
|
| 109 |
-
"epoch": 0.04,
|
| 110 |
-
"learning_rate": 0.00010042342072067417,
|
| 111 |
-
"loss": 0.4624,
|
| 112 |
-
"step": 8000
|
| 113 |
-
},
|
| 114 |
-
{
|
| 115 |
-
"epoch": 0.04,
|
| 116 |
-
"learning_rate": 0.00010037735687948529,
|
| 117 |
-
"loss": 0.463,
|
| 118 |
-
"step": 8500
|
| 119 |
-
},
|
| 120 |
-
{
|
| 121 |
-
"epoch": 0.04,
|
| 122 |
-
"learning_rate": 0.00010032851804476767,
|
| 123 |
-
"loss": 0.4606,
|
| 124 |
-
"step": 9000
|
| 125 |
-
},
|
| 126 |
-
{
|
| 127 |
-
"epoch": 0.05,
|
| 128 |
-
"learning_rate": 0.00010027690722913066,
|
| 129 |
-
"loss": 0.4595,
|
| 130 |
-
"step": 9500
|
| 131 |
-
},
|
| 132 |
-
{
|
| 133 |
-
"epoch": 0.05,
|
| 134 |
-
"learning_rate": 0.00010022263913645699,
|
| 135 |
-
"loss": 0.4605,
|
| 136 |
-
"step": 10000
|
| 137 |
-
},
|
| 138 |
-
{
|
| 139 |
-
"epoch": 0.05,
|
| 140 |
-
"eval_loss": 0.42437314987182617,
|
| 141 |
-
"eval_runtime": 366.557,
|
| 142 |
-
"eval_samples_per_second": 117.308,
|
| 143 |
-
"eval_steps_per_second": 1.833,
|
| 144 |
-
"step": 10000
|
| 145 |
-
},
|
| 146 |
-
{
|
| 147 |
-
"epoch": 0.05,
|
| 148 |
-
"learning_rate": 0.0001001654996079931,
|
| 149 |
-
"loss": 0.4569,
|
| 150 |
-
"step": 10500
|
| 151 |
-
},
|
| 152 |
-
{
|
| 153 |
-
"epoch": 0.06,
|
| 154 |
-
"learning_rate": 0.00010010559815435491,
|
| 155 |
-
"loss": 0.459,
|
| 156 |
-
"step": 11000
|
| 157 |
-
},
|
| 158 |
-
{
|
| 159 |
-
"epoch": 0.06,
|
| 160 |
-
"learning_rate": 0.00010004293847054622,
|
| 161 |
-
"loss": 0.4571,
|
| 162 |
-
"step": 11500
|
| 163 |
-
},
|
| 164 |
-
{
|
| 165 |
-
"epoch": 0.06,
|
| 166 |
-
"learning_rate": 9.99775244217115e-05,
|
| 167 |
-
"loss": 0.4542,
|
| 168 |
-
"step": 12000
|
| 169 |
-
},
|
| 170 |
-
{
|
| 171 |
-
"epoch": 0.06,
|
| 172 |
-
"learning_rate": 9.990936004289739e-05,
|
| 173 |
-
"loss": 0.4562,
|
| 174 |
-
"step": 12500
|
| 175 |
-
},
|
| 176 |
-
{
|
| 177 |
-
"epoch": 0.07,
|
| 178 |
-
"learning_rate": 9.983859409756594e-05,
|
| 179 |
-
"loss": 0.4549,
|
| 180 |
-
"step": 13000
|
| 181 |
-
},
|
| 182 |
-
{
|
| 183 |
-
"epoch": 0.07,
|
| 184 |
-
"learning_rate": 9.976494732131149e-05,
|
| 185 |
-
"loss": 0.4536,
|
| 186 |
-
"step": 13500
|
| 187 |
-
},
|
| 188 |
-
{
|
| 189 |
-
"epoch": 0.07,
|
| 190 |
-
"learning_rate": 9.968856332783455e-05,
|
| 191 |
-
"loss": 0.4537,
|
| 192 |
-
"step": 14000
|
| 193 |
-
},
|
| 194 |
-
{
|
| 195 |
-
"epoch": 0.07,
|
| 196 |
-
"learning_rate": 9.960944682885961e-05,
|
| 197 |
-
"loss": 0.4536,
|
| 198 |
-
"step": 14500
|
| 199 |
-
},
|
| 200 |
-
{
|
| 201 |
-
"epoch": 0.07,
|
| 202 |
-
"learning_rate": 9.952776911175577e-05,
|
| 203 |
-
"loss": 0.4517,
|
| 204 |
-
"step": 15000
|
| 205 |
-
},
|
| 206 |
-
{
|
| 207 |
-
"epoch": 0.07,
|
| 208 |
-
"eval_loss": 0.4174690246582031,
|
| 209 |
-
"eval_runtime": 401.593,
|
| 210 |
-
"eval_samples_per_second": 107.074,
|
| 211 |
-
"eval_steps_per_second": 1.673,
|
| 212 |
-
"step": 15000
|
| 213 |
-
},
|
| 214 |
-
{
|
| 215 |
-
"epoch": 0.08,
|
| 216 |
-
"learning_rate": 9.944320785087589e-05,
|
| 217 |
-
"loss": 0.4518,
|
| 218 |
-
"step": 15500
|
| 219 |
-
},
|
| 220 |
-
{
|
| 221 |
-
"epoch": 0.08,
|
| 222 |
-
"learning_rate": 9.935592921917959e-05,
|
| 223 |
-
"loss": 0.4509,
|
| 224 |
-
"step": 16000
|
| 225 |
-
},
|
| 226 |
-
{
|
| 227 |
-
"epoch": 0.08,
|
| 228 |
-
"learning_rate": 9.926593860042398e-05,
|
| 229 |
-
"loss": 0.4502,
|
| 230 |
-
"step": 16500
|
| 231 |
-
},
|
| 232 |
-
{
|
| 233 |
-
"epoch": 0.09,
|
| 234 |
-
"learning_rate": 9.91732415456543e-05,
|
| 235 |
-
"loss": 0.4499,
|
| 236 |
-
"step": 17000
|
| 237 |
-
},
|
| 238 |
-
{
|
| 239 |
-
"epoch": 0.09,
|
| 240 |
-
"learning_rate": 9.907784377286154e-05,
|
| 241 |
-
"loss": 0.4497,
|
| 242 |
-
"step": 17500
|
| 243 |
-
},
|
| 244 |
-
{
|
| 245 |
-
"epoch": 0.09,
|
| 246 |
-
"learning_rate": 9.897975116662973e-05,
|
| 247 |
-
"loss": 0.4482,
|
| 248 |
-
"step": 18000
|
| 249 |
-
},
|
| 250 |
-
{
|
| 251 |
-
"epoch": 0.09,
|
| 252 |
-
"learning_rate": 9.887896977777283e-05,
|
| 253 |
-
"loss": 0.4483,
|
| 254 |
-
"step": 18500
|
| 255 |
-
},
|
| 256 |
-
{
|
| 257 |
-
"epoch": 0.1,
|
| 258 |
-
"learning_rate": 9.877571542385757e-05,
|
| 259 |
-
"loss": 0.4471,
|
| 260 |
-
"step": 19000
|
| 261 |
-
},
|
| 262 |
-
{
|
| 263 |
-
"epoch": 0.1,
|
| 264 |
-
"learning_rate": 9.866958063112464e-05,
|
| 265 |
-
"loss": 0.4474,
|
| 266 |
-
"step": 19500
|
| 267 |
-
},
|
| 268 |
-
{
|
| 269 |
-
"epoch": 0.1,
|
| 270 |
-
"learning_rate": 9.856077618854629e-05,
|
| 271 |
-
"loss": 0.4476,
|
| 272 |
-
"step": 20000
|
| 273 |
-
},
|
| 274 |
-
{
|
| 275 |
-
"epoch": 0.1,
|
| 276 |
-
"eval_loss": 0.4115261137485504,
|
| 277 |
-
"eval_runtime": 341.4813,
|
| 278 |
-
"eval_samples_per_second": 125.922,
|
| 279 |
-
"eval_steps_per_second": 1.968,
|
| 280 |
-
"step": 20000
|
| 281 |
-
},
|
| 282 |
-
{
|
| 283 |
-
"epoch": 0.1,
|
| 284 |
-
"learning_rate": 9.844930880769307e-05,
|
| 285 |
-
"loss": 0.4453,
|
| 286 |
-
"step": 20500
|
| 287 |
-
},
|
| 288 |
-
{
|
| 289 |
-
"epoch": 0.1,
|
| 290 |
-
"learning_rate": 9.833518536439808e-05,
|
| 291 |
-
"loss": 0.4436,
|
| 292 |
-
"step": 21000
|
| 293 |
-
},
|
| 294 |
-
{
|
| 295 |
-
"epoch": 0.11,
|
| 296 |
-
"learning_rate": 9.821864908222922e-05,
|
| 297 |
-
"loss": 0.4462,
|
| 298 |
-
"step": 21500
|
| 299 |
-
},
|
| 300 |
-
{
|
| 301 |
-
"epoch": 0.11,
|
| 302 |
-
"learning_rate": 9.809924007281187e-05,
|
| 303 |
-
"loss": 0.4439,
|
| 304 |
-
"step": 22000
|
| 305 |
-
},
|
| 306 |
-
{
|
| 307 |
-
"epoch": 0.11,
|
| 308 |
-
"learning_rate": 9.797744330602011e-05,
|
| 309 |
-
"loss": 0.4443,
|
| 310 |
-
"step": 22500
|
| 311 |
-
},
|
| 312 |
-
{
|
| 313 |
-
"epoch": 0.12,
|
| 314 |
-
"learning_rate": 9.785277813397263e-05,
|
| 315 |
-
"loss": 0.4448,
|
| 316 |
-
"step": 23000
|
| 317 |
-
},
|
| 318 |
-
{
|
| 319 |
-
"epoch": 0.12,
|
| 320 |
-
"learning_rate": 9.772549369630525e-05,
|
| 321 |
-
"loss": 0.4444,
|
| 322 |
-
"step": 23500
|
| 323 |
-
},
|
| 324 |
-
{
|
| 325 |
-
"epoch": 0.12,
|
| 326 |
-
"learning_rate": 9.759559784452165e-05,
|
| 327 |
-
"loss": 0.4422,
|
| 328 |
-
"step": 24000
|
| 329 |
-
},
|
| 330 |
-
{
|
| 331 |
-
"epoch": 0.12,
|
| 332 |
-
"learning_rate": 9.746309859120983e-05,
|
| 333 |
-
"loss": 0.4442,
|
| 334 |
-
"step": 24500
|
| 335 |
-
},
|
| 336 |
-
{
|
| 337 |
-
"epoch": 0.12,
|
| 338 |
-
"learning_rate": 9.732827688303682e-05,
|
| 339 |
-
"loss": 0.4429,
|
| 340 |
-
"step": 25000
|
| 341 |
-
},
|
| 342 |
-
{
|
| 343 |
-
"epoch": 0.12,
|
| 344 |
-
"eval_loss": 0.40562233328819275,
|
| 345 |
-
"eval_runtime": 357.2273,
|
| 346 |
-
"eval_samples_per_second": 120.372,
|
| 347 |
-
"eval_steps_per_second": 1.881,
|
| 348 |
-
"step": 25000
|
| 349 |
-
},
|
| 350 |
-
{
|
| 351 |
-
"epoch": 0.13,
|
| 352 |
-
"learning_rate": 9.71906006716561e-05,
|
| 353 |
-
"loss": 0.4406,
|
| 354 |
-
"step": 25500
|
| 355 |
-
},
|
| 356 |
-
{
|
| 357 |
-
"epoch": 0.13,
|
| 358 |
-
"learning_rate": 9.705034604088048e-05,
|
| 359 |
-
"loss": 0.4407,
|
| 360 |
-
"step": 26000
|
| 361 |
-
},
|
| 362 |
-
{
|
| 363 |
-
"epoch": 0.13,
|
| 364 |
-
"learning_rate": 9.690752164227625e-05,
|
| 365 |
-
"loss": 0.4411,
|
| 366 |
-
"step": 26500
|
| 367 |
-
},
|
| 368 |
-
{
|
| 369 |
-
"epoch": 0.14,
|
| 370 |
-
"learning_rate": 9.676213628592508e-05,
|
| 371 |
-
"loss": 0.4398,
|
| 372 |
-
"step": 27000
|
| 373 |
-
},
|
| 374 |
-
{
|
| 375 |
-
"epoch": 0.14,
|
| 376 |
-
"learning_rate": 9.66141989398806e-05,
|
| 377 |
-
"loss": 0.4405,
|
| 378 |
-
"step": 27500
|
| 379 |
-
},
|
| 380 |
-
{
|
| 381 |
-
"epoch": 0.14,
|
| 382 |
-
"learning_rate": 9.64637187296151e-05,
|
| 383 |
-
"loss": 0.4387,
|
| 384 |
-
"step": 28000
|
| 385 |
-
},
|
| 386 |
-
{
|
| 387 |
-
"epoch": 0.14,
|
| 388 |
-
"learning_rate": 9.631070493745684e-05,
|
| 389 |
-
"loss": 0.439,
|
| 390 |
-
"step": 28500
|
| 391 |
-
},
|
| 392 |
-
{
|
| 393 |
-
"epoch": 0.14,
|
| 394 |
-
"learning_rate": 9.615548059063195e-05,
|
| 395 |
-
"loss": 0.439,
|
| 396 |
-
"step": 29000
|
| 397 |
-
},
|
| 398 |
-
{
|
| 399 |
-
"epoch": 0.15,
|
| 400 |
-
"learning_rate": 9.599743312564315e-05,
|
| 401 |
-
"loss": 0.438,
|
| 402 |
-
"step": 29500
|
| 403 |
-
},
|
| 404 |
-
{
|
| 405 |
-
"epoch": 0.15,
|
| 406 |
-
"learning_rate": 9.583688084145402e-05,
|
| 407 |
-
"loss": 0.4369,
|
| 408 |
-
"step": 30000
|
| 409 |
-
},
|
| 410 |
-
{
|
| 411 |
-
"epoch": 0.15,
|
| 412 |
-
"eval_loss": 0.40220412611961365,
|
| 413 |
-
"eval_runtime": 351.2494,
|
| 414 |
-
"eval_samples_per_second": 122.42,
|
| 415 |
-
"eval_steps_per_second": 1.913,
|
| 416 |
-
"step": 30000
|
| 417 |
-
},
|
| 418 |
-
{
|
| 419 |
-
"epoch": 0.0,
|
| 420 |
-
"learning_rate": 9.56738336416857e-05,
|
| 421 |
-
"loss": 0.4374,
|
| 422 |
-
"step": 30500
|
| 423 |
-
},
|
| 424 |
-
{
|
| 425 |
-
"epoch": 0.01,
|
| 426 |
-
"learning_rate": 9.550896864844414e-05,
|
| 427 |
-
"loss": 0.4373,
|
| 428 |
-
"step": 31000
|
| 429 |
-
},
|
| 430 |
-
{
|
| 431 |
-
"epoch": 0.01,
|
| 432 |
-
"learning_rate": 9.534097182139975e-05,
|
| 433 |
-
"loss": 0.4377,
|
| 434 |
-
"step": 31500
|
| 435 |
-
},
|
| 436 |
-
{
|
| 437 |
-
"epoch": 0.01,
|
| 438 |
-
"learning_rate": 9.517051066878048e-05,
|
| 439 |
-
"loss": 0.4352,
|
| 440 |
-
"step": 32000
|
| 441 |
-
},
|
| 442 |
-
{
|
| 443 |
-
"epoch": 0.01,
|
| 444 |
-
"learning_rate": 9.499759570543318e-05,
|
| 445 |
-
"loss": 0.4347,
|
| 446 |
-
"step": 32500
|
| 447 |
-
},
|
| 448 |
-
{
|
| 449 |
-
"epoch": 0.01,
|
| 450 |
-
"learning_rate": 9.482223759756727e-05,
|
| 451 |
-
"loss": 0.4336,
|
| 452 |
-
"step": 33000
|
| 453 |
-
},
|
| 454 |
-
{
|
| 455 |
-
"epoch": 0.02,
|
| 456 |
-
"learning_rate": 9.464444716209686e-05,
|
| 457 |
-
"loss": 0.4337,
|
| 458 |
-
"step": 33500
|
| 459 |
-
},
|
| 460 |
-
{
|
| 461 |
-
"epoch": 0.02,
|
| 462 |
-
"learning_rate": 9.446423536597349e-05,
|
| 463 |
-
"loss": 0.4346,
|
| 464 |
-
"step": 34000
|
| 465 |
-
},
|
| 466 |
-
{
|
| 467 |
-
"epoch": 0.02,
|
| 468 |
-
"learning_rate": 9.42816133255096e-05,
|
| 469 |
-
"loss": 0.4345,
|
| 470 |
-
"step": 34500
|
| 471 |
-
},
|
| 472 |
-
{
|
| 473 |
-
"epoch": 0.03,
|
| 474 |
-
"learning_rate": 9.409659230569288e-05,
|
| 475 |
-
"loss": 0.4329,
|
| 476 |
-
"step": 35000
|
| 477 |
-
},
|
| 478 |
-
{
|
| 479 |
-
"epoch": 0.03,
|
| 480 |
-
"eval_loss": 0.4000852108001709,
|
| 481 |
-
"eval_runtime": 332.6875,
|
| 482 |
-
"eval_samples_per_second": 129.25,
|
| 483 |
-
"eval_steps_per_second": 2.02,
|
| 484 |
-
"step": 35000
|
| 485 |
-
},
|
| 486 |
-
{
|
| 487 |
-
"epoch": 0.03,
|
| 488 |
-
"learning_rate": 9.390918371949136e-05,
|
| 489 |
-
"loss": 0.433,
|
| 490 |
-
"step": 35500
|
| 491 |
-
},
|
| 492 |
-
{
|
| 493 |
-
"epoch": 0.03,
|
| 494 |
-
"learning_rate": 9.371939912714942e-05,
|
| 495 |
-
"loss": 0.4327,
|
| 496 |
-
"step": 36000
|
| 497 |
-
},
|
| 498 |
-
{
|
| 499 |
-
"epoch": 0.03,
|
| 500 |
-
"learning_rate": 9.352725023547466e-05,
|
| 501 |
-
"loss": 0.4324,
|
| 502 |
-
"step": 36500
|
| 503 |
-
},
|
| 504 |
-
{
|
| 505 |
-
"epoch": 0.04,
|
| 506 |
-
"learning_rate": 9.333314023958391e-05,
|
| 507 |
-
"loss": 0.4331,
|
| 508 |
-
"step": 37000
|
| 509 |
-
},
|
| 510 |
-
{
|
| 511 |
-
"epoch": 0.04,
|
| 512 |
-
"learning_rate": 9.313669912311046e-05,
|
| 513 |
-
"loss": 0.4322,
|
| 514 |
-
"step": 37500
|
| 515 |
-
},
|
| 516 |
-
{
|
| 517 |
-
"epoch": 0.04,
|
| 518 |
-
"learning_rate": 9.293753831787908e-05,
|
| 519 |
-
"loss": 0.4312,
|
| 520 |
-
"step": 38000
|
| 521 |
-
},
|
| 522 |
-
{
|
| 523 |
-
"epoch": 0.04,
|
| 524 |
-
"learning_rate": 9.27360614421725e-05,
|
| 525 |
-
"loss": 0.4324,
|
| 526 |
-
"step": 38500
|
| 527 |
-
},
|
| 528 |
-
{
|
| 529 |
-
"epoch": 0.04,
|
| 530 |
-
"learning_rate": 9.253228092403345e-05,
|
| 531 |
-
"loss": 0.4301,
|
| 532 |
-
"step": 39000
|
| 533 |
-
},
|
| 534 |
-
{
|
| 535 |
-
"epoch": 0.05,
|
| 536 |
-
"learning_rate": 9.23262093336042e-05,
|
| 537 |
-
"loss": 0.4301,
|
| 538 |
-
"step": 39500
|
| 539 |
-
},
|
| 540 |
-
{
|
| 541 |
-
"epoch": 0.05,
|
| 542 |
-
"learning_rate": 9.211827834753872e-05,
|
| 543 |
-
"loss": 0.4303,
|
| 544 |
-
"step": 40000
|
| 545 |
-
},
|
| 546 |
-
{
|
| 547 |
-
"epoch": 0.05,
|
| 548 |
-
"eval_loss": 0.39626842737197876,
|
| 549 |
-
"eval_runtime": 338.3932,
|
| 550 |
-
"eval_samples_per_second": 127.071,
|
| 551 |
-
"eval_steps_per_second": 1.986,
|
| 552 |
-
"step": 40000
|
| 553 |
-
},
|
| 554 |
-
{
|
| 555 |
-
"epoch": 0.05,
|
| 556 |
-
"learning_rate": 9.190766740556652e-05,
|
| 557 |
-
"loss": 0.4278,
|
| 558 |
-
"step": 40500
|
| 559 |
-
},
|
| 560 |
-
{
|
| 561 |
-
"epoch": 0.06,
|
| 562 |
-
"learning_rate": 9.169480392040811e-05,
|
| 563 |
-
"loss": 0.4294,
|
| 564 |
-
"step": 41000
|
| 565 |
-
},
|
| 566 |
-
{
|
| 567 |
-
"epoch": 0.06,
|
| 568 |
-
"learning_rate": 9.147970102248595e-05,
|
| 569 |
-
"loss": 0.4292,
|
| 570 |
-
"step": 41500
|
| 571 |
-
},
|
| 572 |
-
{
|
| 573 |
-
"epoch": 0.06,
|
| 574 |
-
"learning_rate": 9.126237198035999e-05,
|
| 575 |
-
"loss": 0.4266,
|
| 576 |
-
"step": 42000
|
| 577 |
-
},
|
| 578 |
-
{
|
| 579 |
-
"epoch": 0.06,
|
| 580 |
-
"learning_rate": 9.104327148280447e-05,
|
| 581 |
-
"loss": 0.4282,
|
| 582 |
-
"step": 42500
|
| 583 |
-
},
|
| 584 |
-
{
|
| 585 |
-
"epoch": 0.07,
|
| 586 |
-
"learning_rate": 9.082153489118645e-05,
|
| 587 |
-
"loss": 0.4275,
|
| 588 |
-
"step": 43000
|
| 589 |
-
},
|
| 590 |
-
{
|
| 591 |
-
"epoch": 0.07,
|
| 592 |
-
"learning_rate": 9.059761275415186e-05,
|
| 593 |
-
"loss": 0.4269,
|
| 594 |
-
"step": 43500
|
| 595 |
-
},
|
| 596 |
-
{
|
| 597 |
-
"epoch": 0.07,
|
| 598 |
-
"learning_rate": 9.037151888427288e-05,
|
| 599 |
-
"loss": 0.4275,
|
| 600 |
-
"step": 44000
|
| 601 |
-
},
|
| 602 |
-
{
|
| 603 |
-
"epoch": 0.07,
|
| 604 |
-
"learning_rate": 9.014372587553127e-05,
|
| 605 |
-
"loss": 0.427,
|
| 606 |
-
"step": 44500
|
| 607 |
-
},
|
| 608 |
-
{
|
| 609 |
-
"epoch": 0.07,
|
| 610 |
-
"learning_rate": 8.991333478594443e-05,
|
| 611 |
-
"loss": 0.4262,
|
| 612 |
-
"step": 45000
|
| 613 |
-
},
|
| 614 |
-
{
|
| 615 |
-
"epoch": 0.07,
|
| 616 |
-
"eval_loss": 0.39185377955436707,
|
| 617 |
-
"eval_runtime": 338.8452,
|
| 618 |
-
"eval_samples_per_second": 126.902,
|
| 619 |
-
"eval_steps_per_second": 1.983,
|
| 620 |
-
"step": 45000
|
| 621 |
-
},
|
| 622 |
-
{
|
| 623 |
-
"epoch": 0.08,
|
| 624 |
-
"learning_rate": 8.968081417300147e-05,
|
| 625 |
-
"loss": 0.4267,
|
| 626 |
-
"step": 45500
|
| 627 |
-
},
|
| 628 |
-
{
|
| 629 |
-
"epoch": 0.08,
|
| 630 |
-
"learning_rate": 8.944664975260744e-05,
|
| 631 |
-
"loss": 0.4257,
|
| 632 |
-
"step": 46000
|
| 633 |
-
},
|
| 634 |
-
{
|
| 635 |
-
"epoch": 0.08,
|
| 636 |
-
"learning_rate": 8.920991743920979e-05,
|
| 637 |
-
"loss": 0.4268,
|
| 638 |
-
"step": 46500
|
| 639 |
-
},
|
| 640 |
-
{
|
| 641 |
-
"epoch": 0.09,
|
| 642 |
-
"learning_rate": 8.89710989925514e-05,
|
| 643 |
-
"loss": 0.4258,
|
| 644 |
-
"step": 47000
|
| 645 |
-
},
|
| 646 |
-
{
|
| 647 |
-
"epoch": 0.09,
|
| 648 |
-
"learning_rate": 8.8730209144079e-05,
|
| 649 |
-
"loss": 0.4253,
|
| 650 |
-
"step": 47500
|
| 651 |
-
},
|
| 652 |
-
{
|
| 653 |
-
"epoch": 0.09,
|
| 654 |
-
"learning_rate": 8.848726275301312e-05,
|
| 655 |
-
"loss": 0.4244,
|
| 656 |
-
"step": 48000
|
| 657 |
-
},
|
| 658 |
-
{
|
| 659 |
-
"epoch": 0.09,
|
| 660 |
-
"learning_rate": 8.824227480543154e-05,
|
| 661 |
-
"loss": 0.423,
|
| 662 |
-
"step": 48500
|
| 663 |
-
},
|
| 664 |
-
{
|
| 665 |
-
"epoch": 0.1,
|
| 666 |
-
"learning_rate": 8.799526041334489e-05,
|
| 667 |
-
"loss": 0.4228,
|
| 668 |
-
"step": 49000
|
| 669 |
-
},
|
| 670 |
-
{
|
| 671 |
-
"epoch": 0.1,
|
| 672 |
-
"learning_rate": 8.77462348137644e-05,
|
| 673 |
-
"loss": 0.4234,
|
| 674 |
-
"step": 49500
|
| 675 |
-
},
|
| 676 |
-
{
|
| 677 |
-
"epoch": 0.1,
|
| 678 |
-
"learning_rate": 8.749571739223735e-05,
|
| 679 |
-
"loss": 0.4251,
|
| 680 |
-
"step": 50000
|
| 681 |
-
},
|
| 682 |
-
{
|
| 683 |
-
"epoch": 0.1,
|
| 684 |
-
"eval_loss": 0.3901826739311218,
|
| 685 |
-
"eval_runtime": 323.5249,
|
| 686 |
-
"eval_samples_per_second": 132.911,
|
| 687 |
-
"eval_steps_per_second": 2.077,
|
| 688 |
-
"step": 50000
|
| 689 |
-
},
|
| 690 |
-
{
|
| 691 |
-
"epoch": 0.1,
|
| 692 |
-
"learning_rate": 8.724271952918828e-05,
|
| 693 |
-
"loss": 0.4238,
|
| 694 |
-
"step": 50500
|
| 695 |
-
},
|
| 696 |
-
{
|
| 697 |
-
"epoch": 0.1,
|
| 698 |
-
"learning_rate": 8.69877568789119e-05,
|
| 699 |
-
"loss": 0.4222,
|
| 700 |
-
"step": 51000
|
| 701 |
-
},
|
| 702 |
-
{
|
| 703 |
-
"epoch": 0.11,
|
| 704 |
-
"learning_rate": 8.673084516870541e-05,
|
| 705 |
-
"loss": 0.4233,
|
| 706 |
-
"step": 51500
|
| 707 |
-
},
|
| 708 |
-
{
|
| 709 |
-
"epoch": 0.11,
|
| 710 |
-
"learning_rate": 8.647200024609333e-05,
|
| 711 |
-
"loss": 0.4223,
|
| 712 |
-
"step": 52000
|
| 713 |
-
},
|
| 714 |
-
{
|
| 715 |
-
"epoch": 0.11,
|
| 716 |
-
"learning_rate": 8.62112380778497e-05,
|
| 717 |
-
"loss": 0.4218,
|
| 718 |
-
"step": 52500
|
| 719 |
-
},
|
| 720 |
-
{
|
| 721 |
-
"epoch": 0.12,
|
| 722 |
-
"learning_rate": 8.594910196227962e-05,
|
| 723 |
-
"loss": 0.4214,
|
| 724 |
-
"step": 53000
|
| 725 |
-
},
|
| 726 |
-
{
|
| 727 |
-
"epoch": 0.12,
|
| 728 |
-
"learning_rate": 8.568455742883119e-05,
|
| 729 |
-
"loss": 0.4216,
|
| 730 |
-
"step": 53500
|
| 731 |
-
},
|
| 732 |
-
{
|
| 733 |
-
"epoch": 0.12,
|
| 734 |
-
"learning_rate": 8.54181442229334e-05,
|
| 735 |
-
"loss": 0.4215,
|
| 736 |
-
"step": 54000
|
| 737 |
-
},
|
| 738 |
-
{
|
| 739 |
-
"epoch": 0.12,
|
| 740 |
-
"learning_rate": 8.51498787782077e-05,
|
| 741 |
-
"loss": 0.4211,
|
| 742 |
-
"step": 54500
|
| 743 |
-
},
|
| 744 |
-
{
|
| 745 |
-
"epoch": 0.12,
|
| 746 |
-
"learning_rate": 8.487977764253034e-05,
|
| 747 |
-
"loss": 0.4207,
|
| 748 |
-
"step": 55000
|
| 749 |
-
},
|
| 750 |
-
{
|
| 751 |
-
"epoch": 0.12,
|
| 752 |
-
"eval_loss": 0.38710343837738037,
|
| 753 |
-
"eval_runtime": 326.6784,
|
| 754 |
-
"eval_samples_per_second": 131.628,
|
| 755 |
-
"eval_steps_per_second": 2.057,
|
| 756 |
-
"step": 55000
|
| 757 |
-
},
|
| 758 |
-
{
|
| 759 |
-
"epoch": 0.13,
|
| 760 |
-
"learning_rate": 8.460785747701169e-05,
|
| 761 |
-
"loss": 0.4191,
|
| 762 |
-
"step": 55500
|
| 763 |
-
},
|
| 764 |
-
{
|
| 765 |
-
"epoch": 0.13,
|
| 766 |
-
"learning_rate": 8.433468428726034e-05,
|
| 767 |
-
"loss": 0.4192,
|
| 768 |
-
"step": 56000
|
| 769 |
-
},
|
| 770 |
-
{
|
| 771 |
-
"epoch": 0.13,
|
| 772 |
-
"learning_rate": 8.405918004700074e-05,
|
| 773 |
-
"loss": 0.4197,
|
| 774 |
-
"step": 56500
|
| 775 |
-
},
|
| 776 |
-
{
|
| 777 |
-
"epoch": 0.14,
|
| 778 |
-
"learning_rate": 8.378190739522489e-05,
|
| 779 |
-
"loss": 0.4192,
|
| 780 |
-
"step": 57000
|
| 781 |
-
},
|
| 782 |
-
{
|
| 783 |
-
"epoch": 0.14,
|
| 784 |
-
"learning_rate": 8.350288343541602e-05,
|
| 785 |
-
"loss": 0.42,
|
| 786 |
-
"step": 57500
|
| 787 |
-
},
|
| 788 |
-
{
|
| 789 |
-
"epoch": 0.14,
|
| 790 |
-
"learning_rate": 8.322212537908629e-05,
|
| 791 |
-
"loss": 0.419,
|
| 792 |
-
"step": 58000
|
| 793 |
-
},
|
| 794 |
-
{
|
| 795 |
-
"epoch": 0.14,
|
| 796 |
-
"learning_rate": 8.294021719616442e-05,
|
| 797 |
-
"loss": 0.4191,
|
| 798 |
-
"step": 58500
|
| 799 |
-
},
|
| 800 |
-
{
|
| 801 |
-
"epoch": 0.14,
|
| 802 |
-
"learning_rate": 8.26560463893781e-05,
|
| 803 |
-
"loss": 0.4187,
|
| 804 |
-
"step": 59000
|
| 805 |
-
},
|
| 806 |
-
{
|
| 807 |
-
"epoch": 0.15,
|
| 808 |
-
"learning_rate": 8.237019372296887e-05,
|
| 809 |
-
"loss": 0.4172,
|
| 810 |
-
"step": 59500
|
| 811 |
-
},
|
| 812 |
-
{
|
| 813 |
-
"epoch": 0.15,
|
| 814 |
-
"learning_rate": 8.208325351258954e-05,
|
| 815 |
-
"loss": 0.4169,
|
| 816 |
-
"step": 60000
|
| 817 |
-
},
|
| 818 |
-
{
|
| 819 |
-
"epoch": 0.15,
|
| 820 |
-
"eval_loss": 0.38494616746902466,
|
| 821 |
-
"eval_runtime": 334.5237,
|
| 822 |
-
"eval_samples_per_second": 128.541,
|
| 823 |
-
"eval_steps_per_second": 2.009,
|
| 824 |
-
"step": 60000
|
| 825 |
-
},
|
| 826 |
-
{
|
| 827 |
-
"epoch": 0.0,
|
| 828 |
-
"learning_rate": 8.179409340302407e-05,
|
| 829 |
-
"loss": 0.4176,
|
| 830 |
-
"step": 60500
|
| 831 |
-
},
|
| 832 |
-
{
|
| 833 |
-
"epoch": 0.01,
|
| 834 |
-
"learning_rate": 8.150330460315606e-05,
|
| 835 |
-
"loss": 0.4161,
|
| 836 |
-
"step": 61000
|
| 837 |
-
},
|
| 838 |
-
{
|
| 839 |
-
"epoch": 0.01,
|
| 840 |
-
"learning_rate": 8.121090505020843e-05,
|
| 841 |
-
"loss": 0.416,
|
| 842 |
-
"step": 61500
|
| 843 |
-
},
|
| 844 |
-
{
|
| 845 |
-
"epoch": 0.01,
|
| 846 |
-
"learning_rate": 8.091691278076293e-05,
|
| 847 |
-
"loss": 0.4147,
|
| 848 |
-
"step": 62000
|
| 849 |
-
},
|
| 850 |
-
{
|
| 851 |
-
"epoch": 0.01,
|
| 852 |
-
"learning_rate": 8.062134592964756e-05,
|
| 853 |
-
"loss": 0.4148,
|
| 854 |
-
"step": 62500
|
| 855 |
-
},
|
| 856 |
-
{
|
| 857 |
-
"epoch": 0.01,
|
| 858 |
-
"learning_rate": 8.032422272881793e-05,
|
| 859 |
-
"loss": 0.4146,
|
| 860 |
-
"step": 63000
|
| 861 |
-
},
|
| 862 |
-
{
|
| 863 |
-
"epoch": 0.02,
|
| 864 |
-
"learning_rate": 8.002616035139417e-05,
|
| 865 |
-
"loss": 0.4143,
|
| 866 |
-
"step": 63500
|
| 867 |
-
},
|
| 868 |
-
{
|
| 869 |
-
"epoch": 0.02,
|
| 870 |
-
"learning_rate": 7.972598255063367e-05,
|
| 871 |
-
"loss": 0.4145,
|
| 872 |
-
"step": 64000
|
| 873 |
-
},
|
| 874 |
-
{
|
| 875 |
-
"epoch": 0.02,
|
| 876 |
-
"learning_rate": 7.942430363038965e-05,
|
| 877 |
-
"loss": 0.4153,
|
| 878 |
-
"step": 64500
|
| 879 |
-
},
|
| 880 |
-
{
|
| 881 |
-
"epoch": 0.03,
|
| 882 |
-
"learning_rate": 7.912174998963153e-05,
|
| 883 |
-
"loss": 0.4136,
|
| 884 |
-
"step": 65000
|
| 885 |
-
},
|
| 886 |
-
{
|
| 887 |
-
"epoch": 0.03,
|
| 888 |
-
"eval_loss": 0.3819546103477478,
|
| 889 |
-
"eval_runtime": 325.1658,
|
| 890 |
-
"eval_samples_per_second": 132.24,
|
| 891 |
-
"eval_steps_per_second": 2.067,
|
| 892 |
-
"step": 65000
|
| 893 |
-
},
|
| 894 |
-
{
|
| 895 |
-
"epoch": 0.03,
|
| 896 |
-
"learning_rate": 7.88171276576962e-05,
|
| 897 |
-
"loss": 0.4139,
|
| 898 |
-
"step": 65500
|
| 899 |
-
},
|
| 900 |
-
{
|
| 901 |
-
"epoch": 0.03,
|
| 902 |
-
"learning_rate": 7.851106026872828e-05,
|
| 903 |
-
"loss": 0.4134,
|
| 904 |
-
"step": 66000
|
| 905 |
-
},
|
| 906 |
-
{
|
| 907 |
-
"epoch": 0.03,
|
| 908 |
-
"learning_rate": 7.820356670240599e-05,
|
| 909 |
-
"loss": 0.413,
|
| 910 |
-
"step": 66500
|
| 911 |
-
},
|
| 912 |
-
{
|
| 913 |
-
"epoch": 0.04,
|
| 914 |
-
"learning_rate": 7.789466592638097e-05,
|
| 915 |
-
"loss": 0.4136,
|
| 916 |
-
"step": 67000
|
| 917 |
-
},
|
| 918 |
-
{
|
| 919 |
-
"epoch": 0.04,
|
| 920 |
-
"learning_rate": 7.75843769951081e-05,
|
| 921 |
-
"loss": 0.4141,
|
| 922 |
-
"step": 67500
|
| 923 |
-
},
|
| 924 |
-
{
|
| 925 |
-
"epoch": 0.04,
|
| 926 |
-
"learning_rate": 7.727271904867026e-05,
|
| 927 |
-
"loss": 0.4129,
|
| 928 |
-
"step": 68000
|
| 929 |
-
},
|
| 930 |
-
{
|
| 931 |
-
"epoch": 0.04,
|
| 932 |
-
"learning_rate": 7.695971131159764e-05,
|
| 933 |
-
"loss": 0.4139,
|
| 934 |
-
"step": 68500
|
| 935 |
-
},
|
| 936 |
-
{
|
| 937 |
-
"epoch": 0.04,
|
| 938 |
-
"learning_rate": 7.664537309168187e-05,
|
| 939 |
-
"loss": 0.4114,
|
| 940 |
-
"step": 69000
|
| 941 |
-
},
|
| 942 |
-
{
|
| 943 |
-
"epoch": 0.05,
|
| 944 |
-
"learning_rate": 7.632972377878499e-05,
|
| 945 |
-
"loss": 0.4124,
|
| 946 |
-
"step": 69500
|
| 947 |
-
},
|
| 948 |
-
{
|
| 949 |
-
"epoch": 0.05,
|
| 950 |
-
"learning_rate": 7.601278284364346e-05,
|
| 951 |
-
"loss": 0.4133,
|
| 952 |
-
"step": 70000
|
| 953 |
-
},
|
| 954 |
-
{
|
| 955 |
-
"epoch": 0.05,
|
| 956 |
-
"eval_loss": 0.38163504004478455,
|
| 957 |
-
"eval_runtime": 341.6253,
|
| 958 |
-
"eval_samples_per_second": 125.869,
|
| 959 |
-
"eval_steps_per_second": 1.967,
|
| 960 |
-
"step": 70000
|
| 961 |
-
},
|
| 962 |
-
{
|
| 963 |
-
"epoch": 0.0,
|
| 964 |
-
"learning_rate": 7.56945698366671e-05,
|
| 965 |
-
"loss": 0.4079,
|
| 966 |
-
"step": 70500
|
| 967 |
-
},
|
| 968 |
-
{
|
| 969 |
-
"epoch": 0.01,
|
| 970 |
-
"learning_rate": 7.537574455448575e-05,
|
| 971 |
-
"loss": 0.4064,
|
| 972 |
-
"step": 71000
|
| 973 |
-
},
|
| 974 |
-
{
|
| 975 |
-
"epoch": 0.01,
|
| 976 |
-
"learning_rate": 7.505504881348406e-05,
|
| 977 |
-
"loss": 0.4059,
|
| 978 |
-
"step": 71500
|
| 979 |
-
},
|
| 980 |
-
{
|
| 981 |
-
"epoch": 0.01,
|
| 982 |
-
"learning_rate": 7.473314007819383e-05,
|
| 983 |
-
"loss": 0.4039,
|
| 984 |
-
"step": 72000
|
| 985 |
-
},
|
| 986 |
-
{
|
| 987 |
-
"epoch": 0.01,
|
| 988 |
-
"learning_rate": 7.441068558672384e-05,
|
| 989 |
-
"loss": 0.4042,
|
| 990 |
-
"step": 72500
|
| 991 |
-
},
|
| 992 |
-
{
|
| 993 |
-
"epoch": 0.01,
|
| 994 |
-
"learning_rate": 7.408641283347138e-05,
|
| 995 |
-
"loss": 0.4036,
|
| 996 |
-
"step": 73000
|
| 997 |
-
},
|
| 998 |
-
{
|
| 999 |
-
"epoch": 0.02,
|
| 1000 |
-
"learning_rate": 7.376098683596032e-05,
|
| 1001 |
-
"loss": 0.4038,
|
| 1002 |
-
"step": 73500
|
| 1003 |
-
},
|
| 1004 |
-
{
|
| 1005 |
-
"epoch": 0.02,
|
| 1006 |
-
"learning_rate": 7.343442766799908e-05,
|
| 1007 |
-
"loss": 0.4044,
|
| 1008 |
-
"step": 74000
|
| 1009 |
-
},
|
| 1010 |
-
{
|
| 1011 |
-
"epoch": 0.02,
|
| 1012 |
-
"learning_rate": 7.31067554732954e-05,
|
| 1013 |
-
"loss": 0.4047,
|
| 1014 |
-
"step": 74500
|
| 1015 |
-
},
|
| 1016 |
-
{
|
| 1017 |
-
"epoch": 0.03,
|
| 1018 |
-
"learning_rate": 7.27779904642137e-05,
|
| 1019 |
-
"loss": 0.4028,
|
| 1020 |
-
"step": 75000
|
| 1021 |
-
},
|
| 1022 |
-
{
|
| 1023 |
-
"epoch": 0.03,
|
| 1024 |
-
"eval_loss": 0.3773901164531708,
|
| 1025 |
-
"eval_runtime": 331.7531,
|
| 1026 |
-
"eval_samples_per_second": 129.614,
|
| 1027 |
-
"eval_steps_per_second": 2.026,
|
| 1028 |
-
"step": 75000
|
| 1029 |
-
},
|
| 1030 |
-
{
|
| 1031 |
-
"epoch": 0.03,
|
| 1032 |
-
"learning_rate": 7.244815292052833e-05,
|
| 1033 |
-
"loss": 0.4063,
|
| 1034 |
-
"step": 75500
|
| 1035 |
-
},
|
| 1036 |
-
{
|
| 1037 |
-
"epoch": 0.03,
|
| 1038 |
-
"learning_rate": 7.211726318817271e-05,
|
| 1039 |
-
"loss": 0.4059,
|
| 1040 |
-
"step": 76000
|
| 1041 |
-
},
|
| 1042 |
-
{
|
| 1043 |
-
"epoch": 0.03,
|
| 1044 |
-
"learning_rate": 7.178534167798416e-05,
|
| 1045 |
-
"loss": 0.4063,
|
| 1046 |
-
"step": 76500
|
| 1047 |
-
},
|
| 1048 |
-
{
|
| 1049 |
-
"epoch": 0.04,
|
| 1050 |
-
"learning_rate": 7.145240886444487e-05,
|
| 1051 |
-
"loss": 0.4066,
|
| 1052 |
-
"step": 77000
|
| 1053 |
-
},
|
| 1054 |
-
{
|
| 1055 |
-
"epoch": 0.04,
|
| 1056 |
-
"learning_rate": 7.11184852844191e-05,
|
| 1057 |
-
"loss": 0.4071,
|
| 1058 |
-
"step": 77500
|
| 1059 |
-
},
|
| 1060 |
-
{
|
| 1061 |
-
"epoch": 0.04,
|
| 1062 |
-
"learning_rate": 7.078359153588611e-05,
|
| 1063 |
-
"loss": 0.4056,
|
| 1064 |
-
"step": 78000
|
| 1065 |
-
},
|
| 1066 |
-
{
|
| 1067 |
-
"epoch": 0.04,
|
| 1068 |
-
"learning_rate": 7.044842089704005e-05,
|
| 1069 |
-
"loss": 0.4069,
|
| 1070 |
-
"step": 78500
|
| 1071 |
-
},
|
| 1072 |
-
{
|
| 1073 |
-
"epoch": 0.04,
|
| 1074 |
-
"learning_rate": 7.011165068041005e-05,
|
| 1075 |
-
"loss": 0.4046,
|
| 1076 |
-
"step": 79000
|
| 1077 |
-
},
|
| 1078 |
-
{
|
| 1079 |
-
"epoch": 0.05,
|
| 1080 |
-
"learning_rate": 6.977397240157385e-05,
|
| 1081 |
-
"loss": 0.4055,
|
| 1082 |
-
"step": 79500
|
| 1083 |
-
},
|
| 1084 |
-
{
|
| 1085 |
-
"epoch": 0.05,
|
| 1086 |
-
"learning_rate": 6.94354068901183e-05,
|
| 1087 |
-
"loss": 0.406,
|
| 1088 |
-
"step": 80000
|
| 1089 |
-
},
|
| 1090 |
-
{
|
| 1091 |
-
"epoch": 0.05,
|
| 1092 |
-
"eval_loss": 0.3769119679927826,
|
| 1093 |
-
"eval_runtime": 327.8086,
|
| 1094 |
-
"eval_samples_per_second": 131.174,
|
| 1095 |
-
"eval_steps_per_second": 2.05,
|
| 1096 |
-
"step": 80000
|
| 1097 |
-
},
|
| 1098 |
-
{
|
| 1099 |
-
"epoch": 0.05,
|
| 1100 |
-
"learning_rate": 6.909665474478617e-05,
|
| 1101 |
-
"loss": 0.4082,
|
| 1102 |
-
"step": 80500
|
| 1103 |
-
},
|
| 1104 |
-
{
|
| 1105 |
-
"epoch": 0.06,
|
| 1106 |
-
"learning_rate": 6.875706052538903e-05,
|
| 1107 |
-
"loss": 0.4096,
|
| 1108 |
-
"step": 81000
|
| 1109 |
-
},
|
| 1110 |
-
{
|
| 1111 |
-
"epoch": 0.06,
|
| 1112 |
-
"learning_rate": 6.841596209024172e-05,
|
| 1113 |
-
"loss": 0.4093,
|
| 1114 |
-
"step": 81500
|
| 1115 |
-
},
|
| 1116 |
-
{
|
| 1117 |
-
"epoch": 0.06,
|
| 1118 |
-
"learning_rate": 6.807406019094787e-05,
|
| 1119 |
-
"loss": 0.4066,
|
| 1120 |
-
"step": 82000
|
| 1121 |
-
},
|
| 1122 |
-
{
|
| 1123 |
-
"epoch": 0.06,
|
| 1124 |
-
"learning_rate": 6.773137591762715e-05,
|
| 1125 |
-
"loss": 0.4087,
|
| 1126 |
-
"step": 82500
|
| 1127 |
-
},
|
| 1128 |
-
{
|
| 1129 |
-
"epoch": 0.07,
|
| 1130 |
-
"learning_rate": 6.73879304086597e-05,
|
| 1131 |
-
"loss": 0.4073,
|
| 1132 |
-
"step": 83000
|
| 1133 |
-
},
|
| 1134 |
-
{
|
| 1135 |
-
"epoch": 0.07,
|
| 1136 |
-
"learning_rate": 6.704374484938226e-05,
|
| 1137 |
-
"loss": 0.4071,
|
| 1138 |
-
"step": 83500
|
| 1139 |
-
},
|
| 1140 |
-
{
|
| 1141 |
-
"epoch": 0.07,
|
| 1142 |
-
"learning_rate": 6.669884047078145e-05,
|
| 1143 |
-
"loss": 0.4079,
|
| 1144 |
-
"step": 84000
|
| 1145 |
-
},
|
| 1146 |
-
{
|
| 1147 |
-
"epoch": 0.07,
|
| 1148 |
-
"learning_rate": 6.635323854818394e-05,
|
| 1149 |
-
"loss": 0.408,
|
| 1150 |
-
"step": 84500
|
| 1151 |
-
},
|
| 1152 |
-
{
|
| 1153 |
-
"epoch": 0.07,
|
| 1154 |
-
"learning_rate": 6.600765361692337e-05,
|
| 1155 |
-
"loss": 0.4064,
|
| 1156 |
-
"step": 85000
|
| 1157 |
-
},
|
| 1158 |
-
{
|
| 1159 |
-
"epoch": 0.07,
|
| 1160 |
-
"eval_loss": 0.37409472465515137,
|
| 1161 |
-
"eval_runtime": 338.9849,
|
| 1162 |
-
"eval_samples_per_second": 126.849,
|
| 1163 |
-
"eval_steps_per_second": 1.982,
|
| 1164 |
-
"step": 85000
|
| 1165 |
-
},
|
| 1166 |
-
{
|
| 1167 |
-
"epoch": 0.08,
|
| 1168 |
-
"learning_rate": 6.566072189149555e-05,
|
| 1169 |
-
"loss": 0.4074,
|
| 1170 |
-
"step": 85500
|
| 1171 |
-
},
|
| 1172 |
-
{
|
| 1173 |
-
"epoch": 0.08,
|
| 1174 |
-
"learning_rate": 6.531454815259442e-05,
|
| 1175 |
-
"loss": 0.4061,
|
| 1176 |
-
"step": 86000
|
| 1177 |
-
},
|
| 1178 |
-
{
|
| 1179 |
-
"epoch": 0.08,
|
| 1180 |
-
"learning_rate": 6.496637325623796e-05,
|
| 1181 |
-
"loss": 0.4061,
|
| 1182 |
-
"step": 86500
|
| 1183 |
-
},
|
| 1184 |
-
{
|
| 1185 |
-
"epoch": 0.09,
|
| 1186 |
-
"learning_rate": 6.461760768262325e-05,
|
| 1187 |
-
"loss": 0.4065,
|
| 1188 |
-
"step": 87000
|
| 1189 |
-
},
|
| 1190 |
-
{
|
| 1191 |
-
"epoch": 0.09,
|
| 1192 |
-
"learning_rate": 6.426827294525372e-05,
|
| 1193 |
-
"loss": 0.4061,
|
| 1194 |
-
"step": 87500
|
| 1195 |
-
},
|
| 1196 |
-
{
|
| 1197 |
-
"epoch": 0.09,
|
| 1198 |
-
"learning_rate": 6.391839059274147e-05,
|
| 1199 |
-
"loss": 0.405,
|
| 1200 |
-
"step": 88000
|
| 1201 |
-
},
|
| 1202 |
-
{
|
| 1203 |
-
"epoch": 0.09,
|
| 1204 |
-
"learning_rate": 6.35679822074781e-05,
|
| 1205 |
-
"loss": 0.4044,
|
| 1206 |
-
"step": 88500
|
| 1207 |
-
},
|
| 1208 |
-
{
|
| 1209 |
-
"epoch": 0.1,
|
| 1210 |
-
"learning_rate": 6.321706940430336e-05,
|
| 1211 |
-
"loss": 0.4043,
|
| 1212 |
-
"step": 89000
|
| 1213 |
-
},
|
| 1214 |
-
{
|
| 1215 |
-
"epoch": 0.1,
|
| 1216 |
-
"learning_rate": 6.286637708772627e-05,
|
| 1217 |
-
"loss": 0.4048,
|
| 1218 |
-
"step": 89500
|
| 1219 |
-
},
|
| 1220 |
-
{
|
| 1221 |
-
"epoch": 0.1,
|
| 1222 |
-
"learning_rate": 6.251452131691338e-05,
|
| 1223 |
-
"loss": 0.4059,
|
| 1224 |
-
"step": 90000
|
| 1225 |
-
},
|
| 1226 |
-
{
|
| 1227 |
-
"epoch": 0.1,
|
| 1228 |
-
"eval_loss": 0.3708394467830658,
|
| 1229 |
-
"eval_runtime": 331.4133,
|
| 1230 |
-
"eval_samples_per_second": 129.747,
|
| 1231 |
-
"eval_steps_per_second": 2.028,
|
| 1232 |
-
"step": 90000
|
| 1233 |
-
},
|
| 1234 |
-
{
|
| 1235 |
-
"epoch": 0.1,
|
| 1236 |
-
"learning_rate": 6.21622261106191e-05,
|
| 1237 |
-
"loss": 0.4039,
|
| 1238 |
-
"step": 90500
|
| 1239 |
-
},
|
| 1240 |
-
{
|
| 1241 |
-
"epoch": 0.1,
|
| 1242 |
-
"learning_rate": 6.180951320007123e-05,
|
| 1243 |
-
"loss": 0.4031,
|
| 1244 |
-
"step": 91000
|
| 1245 |
-
},
|
| 1246 |
-
{
|
| 1247 |
-
"epoch": 0.11,
|
| 1248 |
-
"learning_rate": 6.145640434226352e-05,
|
| 1249 |
-
"loss": 0.4045,
|
| 1250 |
-
"step": 91500
|
| 1251 |
-
},
|
| 1252 |
-
{
|
| 1253 |
-
"epoch": 0.11,
|
| 1254 |
-
"learning_rate": 6.11029213186136e-05,
|
| 1255 |
-
"loss": 0.4029,
|
| 1256 |
-
"step": 92000
|
| 1257 |
-
},
|
| 1258 |
-
{
|
| 1259 |
-
"epoch": 0.11,
|
| 1260 |
-
"learning_rate": 6.074908593361942e-05,
|
| 1261 |
-
"loss": 0.4031,
|
| 1262 |
-
"step": 92500
|
| 1263 |
-
},
|
| 1264 |
-
{
|
| 1265 |
-
"epoch": 0.12,
|
| 1266 |
-
"learning_rate": 6.039562866071103e-05,
|
| 1267 |
-
"loss": 0.4034,
|
| 1268 |
-
"step": 93000
|
| 1269 |
-
},
|
| 1270 |
-
{
|
| 1271 |
-
"epoch": 0.12,
|
| 1272 |
-
"learning_rate": 6.004115464767828e-05,
|
| 1273 |
-
"loss": 0.4033,
|
| 1274 |
-
"step": 93500
|
| 1275 |
-
},
|
| 1276 |
-
{
|
| 1277 |
-
"epoch": 0.12,
|
| 1278 |
-
"learning_rate": 5.9686393768070895e-05,
|
| 1279 |
-
"loss": 0.4021,
|
| 1280 |
-
"step": 94000
|
| 1281 |
-
},
|
| 1282 |
-
{
|
| 1283 |
-
"epoch": 0.12,
|
| 1284 |
-
"learning_rate": 5.933136790521103e-05,
|
| 1285 |
-
"loss": 0.4029,
|
| 1286 |
-
"step": 94500
|
| 1287 |
-
},
|
| 1288 |
-
{
|
| 1289 |
-
"epoch": 0.12,
|
| 1290 |
-
"learning_rate": 5.897609895876622e-05,
|
| 1291 |
-
"loss": 0.4022,
|
| 1292 |
-
"step": 95000
|
| 1293 |
-
},
|
| 1294 |
-
{
|
| 1295 |
-
"epoch": 0.12,
|
| 1296 |
-
"eval_loss": 0.3686982989311218,
|
| 1297 |
-
"eval_runtime": 327.5187,
|
| 1298 |
-
"eval_samples_per_second": 131.29,
|
| 1299 |
-
"eval_steps_per_second": 2.052,
|
| 1300 |
-
"step": 95000
|
| 1301 |
-
},
|
| 1302 |
-
{
|
| 1303 |
-
"epoch": 0.13,
|
| 1304 |
-
"learning_rate": 5.862060884339855e-05,
|
| 1305 |
-
"loss": 0.4009,
|
| 1306 |
-
"step": 95500
|
| 1307 |
-
},
|
| 1308 |
-
{
|
| 1309 |
-
"epoch": 0.13,
|
| 1310 |
-
"learning_rate": 5.826491948741281e-05,
|
| 1311 |
-
"loss": 0.4004,
|
| 1312 |
-
"step": 96000
|
| 1313 |
-
},
|
| 1314 |
-
{
|
| 1315 |
-
"epoch": 0.13,
|
| 1316 |
-
"learning_rate": 5.790905283140396e-05,
|
| 1317 |
-
"loss": 0.401,
|
| 1318 |
-
"step": 96500
|
| 1319 |
-
},
|
| 1320 |
-
{
|
| 1321 |
-
"epoch": 0.14,
|
| 1322 |
-
"learning_rate": 5.755303082690363e-05,
|
| 1323 |
-
"loss": 0.4009,
|
| 1324 |
-
"step": 97000
|
| 1325 |
-
},
|
| 1326 |
-
{
|
| 1327 |
-
"epoch": 0.14,
|
| 1328 |
-
"learning_rate": 5.719687543502604e-05,
|
| 1329 |
-
"loss": 0.4009,
|
| 1330 |
-
"step": 97500
|
| 1331 |
-
},
|
| 1332 |
-
{
|
| 1333 |
-
"epoch": 0.14,
|
| 1334 |
-
"learning_rate": 5.6840608625113427e-05,
|
| 1335 |
-
"loss": 0.3994,
|
| 1336 |
-
"step": 98000
|
| 1337 |
-
},
|
| 1338 |
-
{
|
| 1339 |
-
"epoch": 0.14,
|
| 1340 |
-
"learning_rate": 5.64849651605374e-05,
|
| 1341 |
-
"loss": 0.4009,
|
| 1342 |
-
"step": 98500
|
| 1343 |
-
},
|
| 1344 |
-
{
|
| 1345 |
-
"epoch": 0.14,
|
| 1346 |
-
"learning_rate": 5.612854156169634e-05,
|
| 1347 |
-
"loss": 0.4004,
|
| 1348 |
-
"step": 99000
|
| 1349 |
-
},
|
| 1350 |
-
{
|
| 1351 |
-
"epoch": 0.15,
|
| 1352 |
-
"learning_rate": 5.5772072444685844e-05,
|
| 1353 |
-
"loss": 0.3994,
|
| 1354 |
-
"step": 99500
|
| 1355 |
-
},
|
| 1356 |
-
{
|
| 1357 |
-
"epoch": 0.15,
|
| 1358 |
-
"learning_rate": 5.5415579798200205e-05,
|
| 1359 |
-
"loss": 0.3986,
|
| 1360 |
-
"step": 100000
|
| 1361 |
-
},
|
| 1362 |
-
{
|
| 1363 |
-
"epoch": 0.15,
|
| 1364 |
-
"eval_loss": 0.3672657907009125,
|
| 1365 |
-
"eval_runtime": 324.7481,
|
| 1366 |
-
"eval_samples_per_second": 132.41,
|
| 1367 |
-
"eval_steps_per_second": 2.069,
|
| 1368 |
-
"step": 100000
|
| 1369 |
-
},
|
| 1370 |
-
{
|
| 1371 |
-
"epoch": 0.0,
|
| 1372 |
-
"learning_rate": 5.505908561238503e-05,
|
| 1373 |
-
"loss": 0.3955,
|
| 1374 |
-
"step": 100500
|
| 1375 |
-
},
|
| 1376 |
-
{
|
| 1377 |
-
"epoch": 0.01,
|
| 1378 |
-
"learning_rate": 5.4702611877481e-05,
|
| 1379 |
-
"loss": 0.3939,
|
| 1380 |
-
"step": 101000
|
| 1381 |
-
},
|
| 1382 |
-
{
|
| 1383 |
-
"epoch": 0.01,
|
| 1384 |
-
"learning_rate": 5.4346893388088e-05,
|
| 1385 |
-
"loss": 0.3941,
|
| 1386 |
-
"step": 101500
|
| 1387 |
-
},
|
| 1388 |
-
{
|
| 1389 |
-
"epoch": 0.01,
|
| 1390 |
-
"learning_rate": 5.399052636853316e-05,
|
| 1391 |
-
"loss": 0.3924,
|
| 1392 |
-
"step": 102000
|
| 1393 |
-
},
|
| 1394 |
-
{
|
| 1395 |
-
"epoch": 0.01,
|
| 1396 |
-
"learning_rate": 5.3634245713656956e-05,
|
| 1397 |
-
"loss": 0.3929,
|
| 1398 |
-
"step": 102500
|
| 1399 |
-
},
|
| 1400 |
-
{
|
| 1401 |
-
"epoch": 0.01,
|
| 1402 |
-
"learning_rate": 5.327807340052843e-05,
|
| 1403 |
-
"loss": 0.3922,
|
| 1404 |
-
"step": 103000
|
| 1405 |
-
},
|
| 1406 |
-
{
|
| 1407 |
-
"epoch": 0.02,
|
| 1408 |
-
"learning_rate": 5.2922031399533566e-05,
|
| 1409 |
-
"loss": 0.3918,
|
| 1410 |
-
"step": 103500
|
| 1411 |
-
},
|
| 1412 |
-
{
|
| 1413 |
-
"epoch": 0.02,
|
| 1414 |
-
"learning_rate": 5.256685328591006e-05,
|
| 1415 |
-
"loss": 0.3923,
|
| 1416 |
-
"step": 104000
|
| 1417 |
-
},
|
| 1418 |
-
{
|
| 1419 |
-
"epoch": 0.02,
|
| 1420 |
-
"learning_rate": 5.2211137416475685e-05,
|
| 1421 |
-
"loss": 0.3929,
|
| 1422 |
-
"step": 104500
|
| 1423 |
-
},
|
| 1424 |
-
{
|
| 1425 |
-
"epoch": 0.03,
|
| 1426 |
-
"learning_rate": 5.1855617672812194e-05,
|
| 1427 |
-
"loss": 0.3912,
|
| 1428 |
-
"step": 105000
|
| 1429 |
-
},
|
| 1430 |
-
{
|
| 1431 |
-
"epoch": 0.03,
|
| 1432 |
-
"learning_rate": 5.150031598505203e-05,
|
| 1433 |
-
"loss": 0.3965,
|
| 1434 |
-
"step": 105500
|
| 1435 |
-
},
|
| 1436 |
-
{
|
| 1437 |
-
"epoch": 0.03,
|
| 1438 |
-
"learning_rate": 5.1145254269876875e-05,
|
| 1439 |
-
"loss": 0.3966,
|
| 1440 |
-
"step": 106000
|
| 1441 |
-
},
|
| 1442 |
-
{
|
| 1443 |
-
"epoch": 0.03,
|
| 1444 |
-
"learning_rate": 5.079045442916584e-05,
|
| 1445 |
-
"loss": 0.3964,
|
| 1446 |
-
"step": 106500
|
| 1447 |
-
},
|
| 1448 |
-
{
|
| 1449 |
-
"epoch": 0.04,
|
| 1450 |
-
"learning_rate": 5.0435938348644345e-05,
|
| 1451 |
-
"loss": 0.397,
|
| 1452 |
-
"step": 107000
|
| 1453 |
-
},
|
| 1454 |
-
{
|
| 1455 |
-
"epoch": 0.04,
|
| 1456 |
-
"learning_rate": 5.00824359978954e-05,
|
| 1457 |
-
"loss": 0.3974,
|
| 1458 |
-
"step": 107500
|
| 1459 |
-
},
|
| 1460 |
-
{
|
| 1461 |
-
"epoch": 0.04,
|
| 1462 |
-
"learning_rate": 4.972855234681767e-05,
|
| 1463 |
-
"loss": 0.396,
|
| 1464 |
-
"step": 108000
|
| 1465 |
-
},
|
| 1466 |
-
{
|
| 1467 |
-
"epoch": 0.04,
|
| 1468 |
-
"learning_rate": 4.937501795905196e-05,
|
| 1469 |
-
"loss": 0.3969,
|
| 1470 |
-
"step": 108500
|
| 1471 |
-
},
|
| 1472 |
-
{
|
| 1473 |
-
"epoch": 0.04,
|
| 1474 |
-
"learning_rate": 4.902185464226454e-05,
|
| 1475 |
-
"loss": 0.3949,
|
| 1476 |
-
"step": 109000
|
| 1477 |
-
},
|
| 1478 |
-
{
|
| 1479 |
-
"epoch": 0.05,
|
| 1480 |
-
"learning_rate": 4.8669084181232404e-05,
|
| 1481 |
-
"loss": 0.396,
|
| 1482 |
-
"step": 109500
|
| 1483 |
-
},
|
| 1484 |
-
{
|
| 1485 |
-
"epoch": 0.05,
|
| 1486 |
-
"learning_rate": 4.831672833649923e-05,
|
| 1487 |
-
"loss": 0.3963,
|
| 1488 |
-
"step": 110000
|
| 1489 |
-
},
|
| 1490 |
-
{
|
| 1491 |
-
"epoch": 0.05,
|
| 1492 |
-
"learning_rate": 4.79648088430333e-05,
|
| 1493 |
-
"loss": 0.3957,
|
| 1494 |
-
"step": 110500
|
| 1495 |
-
},
|
| 1496 |
-
{
|
| 1497 |
-
"epoch": 0.06,
|
| 1498 |
-
"learning_rate": 4.761334740888664e-05,
|
| 1499 |
-
"loss": 0.3964,
|
| 1500 |
-
"step": 111000
|
| 1501 |
-
},
|
| 1502 |
-
{
|
| 1503 |
-
"epoch": 0.06,
|
| 1504 |
-
"learning_rate": 4.7262365713856065e-05,
|
| 1505 |
-
"loss": 0.3962,
|
| 1506 |
-
"step": 111500
|
| 1507 |
-
},
|
| 1508 |
-
{
|
| 1509 |
-
"epoch": 0.06,
|
| 1510 |
-
"learning_rate": 4.691328630193755e-05,
|
| 1511 |
-
"loss": 0.3937,
|
| 1512 |
-
"step": 112000
|
| 1513 |
-
},
|
| 1514 |
-
{
|
| 1515 |
-
"epoch": 0.06,
|
| 1516 |
-
"learning_rate": 4.656402625240025e-05,
|
| 1517 |
-
"loss": 0.3958,
|
| 1518 |
-
"step": 112500
|
| 1519 |
-
},
|
| 1520 |
-
{
|
| 1521 |
-
"epoch": 0.07,
|
| 1522 |
-
"learning_rate": 4.6214610219017164e-05,
|
| 1523 |
-
"loss": 0.3951,
|
| 1524 |
-
"step": 113000
|
| 1525 |
-
},
|
| 1526 |
-
{
|
| 1527 |
-
"epoch": 0.07,
|
| 1528 |
-
"learning_rate": 4.586576020544867e-05,
|
| 1529 |
-
"loss": 0.3945,
|
| 1530 |
-
"step": 113500
|
| 1531 |
-
},
|
| 1532 |
-
{
|
| 1533 |
-
"epoch": 0.07,
|
| 1534 |
-
"learning_rate": 4.551749773040681e-05,
|
| 1535 |
-
"loss": 0.3952,
|
| 1536 |
-
"step": 114000
|
| 1537 |
-
},
|
| 1538 |
-
{
|
| 1539 |
-
"epoch": 0.07,
|
| 1540 |
-
"learning_rate": 4.516984427636154e-05,
|
| 1541 |
-
"loss": 0.3951,
|
| 1542 |
-
"step": 114500
|
| 1543 |
-
},
|
| 1544 |
-
{
|
| 1545 |
-
"epoch": 0.07,
|
| 1546 |
-
"learning_rate": 4.4822821288215474e-05,
|
| 1547 |
-
"loss": 0.3938,
|
| 1548 |
-
"step": 115000
|
| 1549 |
-
},
|
| 1550 |
-
{
|
| 1551 |
-
"epoch": 0.08,
|
| 1552 |
-
"learning_rate": 4.447645017198114e-05,
|
| 1553 |
-
"loss": 0.3944,
|
| 1554 |
-
"step": 115500
|
| 1555 |
-
},
|
| 1556 |
-
{
|
| 1557 |
-
"epoch": 0.08,
|
| 1558 |
-
"learning_rate": 4.4130752293460565e-05,
|
| 1559 |
-
"loss": 0.3932,
|
| 1560 |
-
"step": 116000
|
| 1561 |
-
},
|
| 1562 |
-
{
|
| 1563 |
-
"epoch": 0.08,
|
| 1564 |
-
"learning_rate": 4.3785748976927253e-05,
|
| 1565 |
-
"loss": 0.3934,
|
| 1566 |
-
"step": 116500
|
| 1567 |
-
},
|
| 1568 |
-
{
|
| 1569 |
-
"epoch": 0.09,
|
| 1570 |
-
"learning_rate": 4.344214935022219e-05,
|
| 1571 |
-
"loss": 0.3937,
|
| 1572 |
-
"step": 117000
|
| 1573 |
-
},
|
| 1574 |
-
{
|
| 1575 |
-
"epoch": 0.09,
|
| 1576 |
-
"learning_rate": 4.309859746246968e-05,
|
| 1577 |
-
"loss": 0.3937,
|
| 1578 |
-
"step": 117500
|
| 1579 |
-
},
|
| 1580 |
-
{
|
| 1581 |
-
"epoch": 0.09,
|
| 1582 |
-
"learning_rate": 4.275580380487631e-05,
|
| 1583 |
-
"loss": 0.3927,
|
| 1584 |
-
"step": 118000
|
| 1585 |
-
},
|
| 1586 |
-
{
|
| 1587 |
-
"epoch": 0.09,
|
| 1588 |
-
"learning_rate": 4.2413789522569584e-05,
|
| 1589 |
-
"loss": 0.3915,
|
| 1590 |
-
"step": 118500
|
| 1591 |
-
},
|
| 1592 |
-
{
|
| 1593 |
-
"epoch": 0.1,
|
| 1594 |
-
"learning_rate": 4.2072575712601445e-05,
|
| 1595 |
-
"loss": 0.3919,
|
| 1596 |
-
"step": 119000
|
| 1597 |
-
},
|
| 1598 |
-
{
|
| 1599 |
-
"epoch": 0.1,
|
| 1600 |
-
"learning_rate": 4.1732863373385324e-05,
|
| 1601 |
-
"loss": 0.3928,
|
| 1602 |
-
"step": 119500
|
| 1603 |
-
},
|
| 1604 |
-
{
|
| 1605 |
-
"epoch": 0.1,
|
| 1606 |
-
"learning_rate": 4.1393311894489634e-05,
|
| 1607 |
-
"loss": 0.3933,
|
| 1608 |
-
"step": 120000
|
| 1609 |
-
},
|
| 1610 |
-
{
|
| 1611 |
-
"epoch": 0.1,
|
| 1612 |
-
"learning_rate": 4.105462383579954e-05,
|
| 1613 |
-
"loss": 0.3916,
|
| 1614 |
-
"step": 120500
|
| 1615 |
-
},
|
| 1616 |
-
{
|
| 1617 |
-
"epoch": 0.1,
|
| 1618 |
-
"learning_rate": 4.0716820089189836e-05,
|
| 1619 |
-
"loss": 0.3908,
|
| 1620 |
-
"step": 121000
|
| 1621 |
-
},
|
| 1622 |
-
{
|
| 1623 |
-
"epoch": 0.11,
|
| 1624 |
-
"learning_rate": 4.038059437201991e-05,
|
| 1625 |
-
"loss": 0.3929,
|
| 1626 |
-
"step": 121500
|
| 1627 |
-
},
|
| 1628 |
-
{
|
| 1629 |
-
"epoch": 0.11,
|
| 1630 |
-
"learning_rate": 4.0044619833152034e-05,
|
| 1631 |
-
"loss": 0.3914,
|
| 1632 |
-
"step": 122000
|
| 1633 |
-
},
|
| 1634 |
-
{
|
| 1635 |
-
"epoch": 0.11,
|
| 1636 |
-
"learning_rate": 3.970959190816897e-05,
|
| 1637 |
-
"loss": 0.3911,
|
| 1638 |
-
"step": 122500
|
| 1639 |
-
},
|
| 1640 |
-
{
|
| 1641 |
-
"epoch": 0.12,
|
| 1642 |
-
"learning_rate": 3.937553126317129e-05,
|
| 1643 |
-
"loss": 0.391,
|
| 1644 |
-
"step": 123000
|
| 1645 |
-
},
|
| 1646 |
-
{
|
| 1647 |
-
"epoch": 0.12,
|
| 1648 |
-
"learning_rate": 3.9043123650533415e-05,
|
| 1649 |
-
"loss": 0.3912,
|
| 1650 |
-
"step": 123500
|
| 1651 |
-
},
|
| 1652 |
-
{
|
| 1653 |
-
"epoch": 0.12,
|
| 1654 |
-
"learning_rate": 3.8711057286544326e-05,
|
| 1655 |
-
"loss": 0.3904,
|
| 1656 |
-
"step": 124000
|
| 1657 |
-
},
|
| 1658 |
-
{
|
| 1659 |
-
"epoch": 0.12,
|
| 1660 |
-
"learning_rate": 3.838001979685956e-05,
|
| 1661 |
-
"loss": 0.3911,
|
| 1662 |
-
"step": 124500
|
| 1663 |
-
},
|
| 1664 |
-
{
|
| 1665 |
-
"epoch": 0.12,
|
| 1666 |
-
"learning_rate": 3.805069051708565e-05,
|
| 1667 |
-
"loss": 0.391,
|
| 1668 |
-
"step": 125000
|
| 1669 |
-
},
|
| 1670 |
-
{
|
| 1671 |
-
"epoch": 0.13,
|
| 1672 |
-
"learning_rate": 3.772176981156942e-05,
|
| 1673 |
-
"loss": 0.3892,
|
| 1674 |
-
"step": 125500
|
| 1675 |
-
},
|
| 1676 |
-
{
|
| 1677 |
-
"epoch": 0.13,
|
| 1678 |
-
"learning_rate": 3.739393900426893e-05,
|
| 1679 |
-
"loss": 0.3894,
|
| 1680 |
-
"step": 126000
|
| 1681 |
-
},
|
| 1682 |
-
{
|
| 1683 |
-
"epoch": 0.13,
|
| 1684 |
-
"learning_rate": 3.706721831733264e-05,
|
| 1685 |
-
"loss": 0.3898,
|
| 1686 |
-
"step": 126500
|
| 1687 |
-
},
|
| 1688 |
-
{
|
| 1689 |
-
"epoch": 0.14,
|
| 1690 |
-
"learning_rate": 3.674162790443148e-05,
|
| 1691 |
-
"loss": 0.389,
|
| 1692 |
-
"step": 127000
|
| 1693 |
-
},
|
| 1694 |
-
{
|
| 1695 |
-
"epoch": 0.14,
|
| 1696 |
-
"learning_rate": 3.6417187849515776e-05,
|
| 1697 |
-
"loss": 0.3898,
|
| 1698 |
-
"step": 127500
|
| 1699 |
-
},
|
| 1700 |
-
{
|
| 1701 |
-
"epoch": 0.14,
|
| 1702 |
-
"learning_rate": 3.60939181655764e-05,
|
| 1703 |
-
"loss": 0.3882,
|
| 1704 |
-
"step": 128000
|
| 1705 |
-
},
|
| 1706 |
-
{
|
| 1707 |
-
"epoch": 0.14,
|
| 1708 |
-
"learning_rate": 3.577183879341017e-05,
|
| 1709 |
-
"loss": 0.3891,
|
| 1710 |
-
"step": 128500
|
| 1711 |
-
},
|
| 1712 |
-
{
|
| 1713 |
-
"epoch": 0.14,
|
| 1714 |
-
"learning_rate": 3.54509696003899e-05,
|
| 1715 |
-
"loss": 0.389,
|
| 1716 |
-
"step": 129000
|
| 1717 |
-
},
|
| 1718 |
-
{
|
| 1719 |
-
"epoch": 0.15,
|
| 1720 |
-
"learning_rate": 3.5131968417051296e-05,
|
| 1721 |
-
"loss": 0.3876,
|
| 1722 |
-
"step": 129500
|
| 1723 |
-
},
|
| 1724 |
-
{
|
| 1725 |
-
"epoch": 0.15,
|
| 1726 |
-
"learning_rate": 3.4813576365618406e-05,
|
| 1727 |
-
"loss": 0.3876,
|
| 1728 |
-
"step": 130000
|
| 1729 |
-
},
|
| 1730 |
-
{
|
| 1731 |
-
"epoch": 0.15,
|
| 1732 |
-
"learning_rate": 3.449645360347186e-05,
|
| 1733 |
-
"loss": 0.3896,
|
| 1734 |
-
"step": 130500
|
| 1735 |
-
},
|
| 1736 |
-
{
|
| 1737 |
-
"epoch": 0.15,
|
| 1738 |
-
"learning_rate": 3.418061969223735e-05,
|
| 1739 |
-
"loss": 0.3907,
|
| 1740 |
-
"step": 131000
|
| 1741 |
-
},
|
| 1742 |
-
{
|
| 1743 |
-
"epoch": 0.16,
|
| 1744 |
-
"learning_rate": 3.386672184656929e-05,
|
| 1745 |
-
"loss": 0.3908,
|
| 1746 |
-
"step": 131500
|
| 1747 |
-
},
|
| 1748 |
-
{
|
| 1749 |
-
"epoch": 0.16,
|
| 1750 |
-
"learning_rate": 3.355352132804756e-05,
|
| 1751 |
-
"loss": 0.389,
|
| 1752 |
-
"step": 132000
|
| 1753 |
-
},
|
| 1754 |
-
{
|
| 1755 |
-
"epoch": 0.16,
|
| 1756 |
-
"learning_rate": 3.32416678249423e-05,
|
| 1757 |
-
"loss": 0.389,
|
| 1758 |
-
"step": 132500
|
| 1759 |
-
},
|
| 1760 |
-
{
|
| 1761 |
-
"epoch": 0.17,
|
| 1762 |
-
"learning_rate": 3.293118057384647e-05,
|
| 1763 |
-
"loss": 0.3894,
|
| 1764 |
-
"step": 133000
|
| 1765 |
-
},
|
| 1766 |
-
{
|
| 1767 |
-
"epoch": 0.17,
|
| 1768 |
-
"learning_rate": 3.262269553544883e-05,
|
| 1769 |
-
"loss": 0.3897,
|
| 1770 |
-
"step": 133500
|
| 1771 |
-
},
|
| 1772 |
-
{
|
| 1773 |
-
"epoch": 0.17,
|
| 1774 |
-
"learning_rate": 3.231499533194849e-05,
|
| 1775 |
-
"loss": 0.3905,
|
| 1776 |
-
"step": 134000
|
| 1777 |
-
},
|
| 1778 |
-
{
|
| 1779 |
-
"epoch": 0.17,
|
| 1780 |
-
"learning_rate": 3.200871854198199e-05,
|
| 1781 |
-
"loss": 0.3901,
|
| 1782 |
-
"step": 134500
|
| 1783 |
-
},
|
| 1784 |
-
{
|
| 1785 |
-
"epoch": 0.17,
|
| 1786 |
-
"learning_rate": 3.170388405814444e-05,
|
| 1787 |
-
"loss": 0.389,
|
| 1788 |
-
"step": 135000
|
| 1789 |
-
},
|
| 1790 |
-
{
|
| 1791 |
-
"epoch": 0.18,
|
| 1792 |
-
"learning_rate": 3.140051068406269e-05,
|
| 1793 |
-
"loss": 0.3897,
|
| 1794 |
-
"step": 135500
|
| 1795 |
-
},
|
| 1796 |
-
{
|
| 1797 |
-
"epoch": 0.18,
|
| 1798 |
-
"learning_rate": 3.109982173491942e-05,
|
| 1799 |
-
"loss": 0.389,
|
| 1800 |
-
"step": 136000
|
| 1801 |
-
},
|
| 1802 |
-
{
|
| 1803 |
-
"epoch": 0.18,
|
| 1804 |
-
"learning_rate": 3.079942059880779e-05,
|
| 1805 |
-
"loss": 0.3887,
|
| 1806 |
-
"step": 136500
|
| 1807 |
-
},
|
| 1808 |
-
{
|
| 1809 |
-
"epoch": 0.18,
|
| 1810 |
-
"learning_rate": 3.0500536364018534e-05,
|
| 1811 |
-
"loss": 0.3884,
|
| 1812 |
-
"step": 137000
|
| 1813 |
-
},
|
| 1814 |
-
{
|
| 1815 |
-
"epoch": 0.19,
|
| 1816 |
-
"learning_rate": 3.0203187467139138e-05,
|
| 1817 |
-
"loss": 0.3885,
|
| 1818 |
-
"step": 137500
|
| 1819 |
-
},
|
| 1820 |
-
{
|
| 1821 |
-
"epoch": 0.19,
|
| 1822 |
-
"learning_rate": 2.9907392250050176e-05,
|
| 1823 |
-
"loss": 0.3884,
|
| 1824 |
-
"step": 138000
|
| 1825 |
-
},
|
| 1826 |
-
{
|
| 1827 |
-
"epoch": 0.19,
|
| 1828 |
-
"learning_rate": 2.9613168958794007e-05,
|
| 1829 |
-
"loss": 0.3882,
|
| 1830 |
-
"step": 138500
|
| 1831 |
-
},
|
| 1832 |
-
{
|
| 1833 |
-
"epoch": 0.2,
|
| 1834 |
-
"learning_rate": 2.9320535742449102e-05,
|
| 1835 |
-
"loss": 0.3878,
|
| 1836 |
-
"step": 139000
|
| 1837 |
-
},
|
| 1838 |
-
{
|
| 1839 |
-
"epoch": 0.2,
|
| 1840 |
-
"learning_rate": 2.903009108533329e-05,
|
| 1841 |
-
"loss": 0.3878,
|
| 1842 |
-
"step": 139500
|
| 1843 |
-
},
|
| 1844 |
-
{
|
| 1845 |
-
"epoch": 0.2,
|
| 1846 |
-
"learning_rate": 2.8740688802594957e-05,
|
| 1847 |
-
"loss": 0.389,
|
| 1848 |
-
"step": 140000
|
| 1849 |
-
},
|
| 1850 |
-
{
|
| 1851 |
-
"epoch": 0.2,
|
| 1852 |
-
"learning_rate": 2.8452930413453704e-05,
|
| 1853 |
-
"loss": 0.3866,
|
| 1854 |
-
"step": 140500
|
| 1855 |
-
},
|
| 1856 |
-
{
|
| 1857 |
-
"epoch": 0.2,
|
| 1858 |
-
"learning_rate": 2.8166833668202425e-05,
|
| 1859 |
-
"loss": 0.3866,
|
| 1860 |
-
"step": 141000
|
| 1861 |
-
},
|
| 1862 |
-
{
|
| 1863 |
-
"epoch": 0.21,
|
| 1864 |
-
"learning_rate": 2.7882983361931612e-05,
|
| 1865 |
-
"loss": 0.3866,
|
| 1866 |
-
"step": 141500
|
| 1867 |
-
},
|
| 1868 |
-
{
|
| 1869 |
-
"epoch": 0.21,
|
| 1870 |
-
"learning_rate": 2.760025933314676e-05,
|
| 1871 |
-
"loss": 0.3875,
|
| 1872 |
-
"step": 142000
|
| 1873 |
-
},
|
| 1874 |
-
{
|
| 1875 |
-
"epoch": 0.21,
|
| 1876 |
-
"learning_rate": 2.7319249545020577e-05,
|
| 1877 |
-
"loss": 0.3874,
|
| 1878 |
-
"step": 142500
|
| 1879 |
-
},
|
| 1880 |
-
{
|
| 1881 |
-
"epoch": 0.21,
|
| 1882 |
-
"learning_rate": 2.7039971331560428e-05,
|
| 1883 |
-
"loss": 0.3857,
|
| 1884 |
-
"step": 143000
|
| 1885 |
-
},
|
| 1886 |
-
{
|
| 1887 |
-
"epoch": 0.22,
|
| 1888 |
-
"learning_rate": 2.676244191996206e-05,
|
| 1889 |
-
"loss": 0.3859,
|
| 1890 |
-
"step": 143500
|
| 1891 |
-
},
|
| 1892 |
-
{
|
| 1893 |
-
"epoch": 0.22,
|
| 1894 |
-
"learning_rate": 2.6486678429546845e-05,
|
| 1895 |
-
"loss": 0.385,
|
| 1896 |
-
"step": 144000
|
| 1897 |
-
},
|
| 1898 |
-
{
|
| 1899 |
-
"epoch": 0.22,
|
| 1900 |
-
"learning_rate": 2.6212697870705843e-05,
|
| 1901 |
-
"loss": 0.3854,
|
| 1902 |
-
"step": 144500
|
| 1903 |
-
},
|
| 1904 |
-
{
|
| 1905 |
-
"epoch": 0.23,
|
| 1906 |
-
"learning_rate": 2.5940517143850467e-05,
|
| 1907 |
-
"loss": 0.3862,
|
| 1908 |
-
"step": 145000
|
| 1909 |
-
},
|
| 1910 |
-
{
|
| 1911 |
-
"epoch": 0.23,
|
| 1912 |
-
"learning_rate": 2.5670153038370093e-05,
|
| 1913 |
-
"loss": 0.3867,
|
| 1914 |
-
"step": 145500
|
| 1915 |
-
},
|
| 1916 |
-
{
|
| 1917 |
-
"epoch": 0.23,
|
| 1918 |
-
"learning_rate": 2.540162223159619e-05,
|
| 1919 |
-
"loss": 0.3847,
|
| 1920 |
-
"step": 146000
|
| 1921 |
-
},
|
| 1922 |
-
{
|
| 1923 |
-
"epoch": 0.23,
|
| 1924 |
-
"learning_rate": 2.513494128777388e-05,
|
| 1925 |
-
"loss": 0.3849,
|
| 1926 |
-
"step": 146500
|
| 1927 |
-
},
|
| 1928 |
-
{
|
| 1929 |
-
"epoch": 0.23,
|
| 1930 |
-
"learning_rate": 2.4870126657039936e-05,
|
| 1931 |
-
"loss": 0.3851,
|
| 1932 |
-
"step": 147000
|
| 1933 |
-
},
|
| 1934 |
-
{
|
| 1935 |
-
"epoch": 0.24,
|
| 1936 |
-
"learning_rate": 2.4607718648691012e-05,
|
| 1937 |
-
"loss": 0.3855,
|
| 1938 |
-
"step": 147500
|
| 1939 |
-
},
|
| 1940 |
-
{
|
| 1941 |
-
"epoch": 0.24,
|
| 1942 |
-
"learning_rate": 2.4346681719201903e-05,
|
| 1943 |
-
"loss": 0.3847,
|
| 1944 |
-
"step": 148000
|
| 1945 |
-
},
|
| 1946 |
-
{
|
| 1947 |
-
"epoch": 0.24,
|
| 1948 |
-
"learning_rate": 2.4087559726364384e-05,
|
| 1949 |
-
"loss": 0.3852,
|
| 1950 |
-
"step": 148500
|
| 1951 |
-
},
|
| 1952 |
-
{
|
| 1953 |
-
"epoch": 0.24,
|
| 1954 |
-
"learning_rate": 2.383036865404354e-05,
|
| 1955 |
-
"loss": 0.3845,
|
| 1956 |
-
"step": 149000
|
| 1957 |
-
},
|
| 1958 |
-
{
|
| 1959 |
-
"epoch": 0.25,
|
| 1960 |
-
"learning_rate": 2.3575632902193518e-05,
|
| 1961 |
-
"loss": 0.384,
|
| 1962 |
-
"step": 149500
|
| 1963 |
-
},
|
| 1964 |
-
{
|
| 1965 |
-
"epoch": 0.25,
|
| 1966 |
-
"learning_rate": 2.3322347204396843e-05,
|
| 1967 |
-
"loss": 0.3829,
|
| 1968 |
-
"step": 150000
|
| 1969 |
-
},
|
| 1970 |
-
{
|
| 1971 |
-
"epoch": 0.0,
|
| 1972 |
-
"learning_rate": 2.3071039629029665e-05,
|
| 1973 |
-
"loss": 0.3777,
|
| 1974 |
-
"step": 150500
|
| 1975 |
-
},
|
| 1976 |
-
{
|
| 1977 |
-
"epoch": 0.01,
|
| 1978 |
-
"learning_rate": 2.2821725677927028e-05,
|
| 1979 |
-
"loss": 0.376,
|
| 1980 |
-
"step": 151000
|
| 1981 |
-
},
|
| 1982 |
-
{
|
| 1983 |
-
"epoch": 0.01,
|
| 1984 |
-
"learning_rate": 2.2574420729947773e-05,
|
| 1985 |
-
"loss": 0.3763,
|
| 1986 |
-
"step": 151500
|
| 1987 |
-
},
|
| 1988 |
-
{
|
| 1989 |
-
"epoch": 0.01,
|
| 1990 |
-
"learning_rate": 2.232914004002604e-05,
|
| 1991 |
-
"loss": 0.3749,
|
| 1992 |
-
"step": 152000
|
| 1993 |
-
},
|
| 1994 |
-
{
|
| 1995 |
-
"epoch": 0.01,
|
| 1996 |
-
"learning_rate": 2.2086383175530457e-05,
|
| 1997 |
-
"loss": 0.375,
|
| 1998 |
-
"step": 152500
|
| 1999 |
-
},
|
| 2000 |
-
{
|
| 2001 |
-
"epoch": 0.01,
|
| 2002 |
-
"learning_rate": 2.1845192142454672e-05,
|
| 2003 |
-
"loss": 0.3744,
|
| 2004 |
-
"step": 153000
|
| 2005 |
-
},
|
| 2006 |
-
{
|
| 2007 |
-
"epoch": 0.02,
|
| 2008 |
-
"learning_rate": 2.1606070349690614e-05,
|
| 2009 |
-
"loss": 0.3744,
|
| 2010 |
-
"step": 153500
|
| 2011 |
-
},
|
| 2012 |
-
{
|
| 2013 |
-
"epoch": 0.02,
|
| 2014 |
-
"learning_rate": 2.136903254739686e-05,
|
| 2015 |
-
"loss": 0.3749,
|
| 2016 |
-
"step": 154000
|
| 2017 |
-
},
|
| 2018 |
-
{
|
| 2019 |
-
"epoch": 0.02,
|
| 2020 |
-
"learning_rate": 2.1134561131493055e-05,
|
| 2021 |
-
"loss": 0.3756,
|
| 2022 |
-
"step": 154500
|
| 2023 |
-
},
|
| 2024 |
-
{
|
| 2025 |
-
"epoch": 0.03,
|
| 2026 |
-
"learning_rate": 2.0901730804927037e-05,
|
| 2027 |
-
"loss": 0.374,
|
| 2028 |
-
"step": 155000
|
| 2029 |
-
},
|
| 2030 |
-
{
|
| 2031 |
-
"epoch": 0.03,
|
| 2032 |
-
"learning_rate": 2.0671027915812168e-05,
|
| 2033 |
-
"loss": 0.3773,
|
| 2034 |
-
"step": 155500
|
| 2035 |
-
},
|
| 2036 |
-
{
|
| 2037 |
-
"epoch": 0.03,
|
| 2038 |
-
"learning_rate": 2.0442466694989354e-05,
|
| 2039 |
-
"loss": 0.3771,
|
| 2040 |
-
"step": 156000
|
| 2041 |
-
},
|
| 2042 |
-
{
|
| 2043 |
-
"epoch": 0.03,
|
| 2044 |
-
"learning_rate": 2.0216061241191315e-05,
|
| 2045 |
-
"loss": 0.377,
|
| 2046 |
-
"step": 156500
|
| 2047 |
-
},
|
| 2048 |
-
{
|
| 2049 |
-
"epoch": 0.04,
|
| 2050 |
-
"learning_rate": 1.9991825520172903e-05,
|
| 2051 |
-
"loss": 0.3773,
|
| 2052 |
-
"step": 157000
|
| 2053 |
-
},
|
| 2054 |
-
{
|
| 2055 |
-
"epoch": 0.04,
|
| 2056 |
-
"learning_rate": 1.9769773363849665e-05,
|
| 2057 |
-
"loss": 0.3782,
|
| 2058 |
-
"step": 157500
|
| 2059 |
-
},
|
| 2060 |
-
{
|
| 2061 |
-
"epoch": 0.04,
|
| 2062 |
-
"learning_rate": 1.954991846944452e-05,
|
| 2063 |
-
"loss": 0.3768,
|
| 2064 |
-
"step": 158000
|
| 2065 |
-
},
|
| 2066 |
-
{
|
| 2067 |
-
"epoch": 0.04,
|
| 2068 |
-
"learning_rate": 1.933227439864297e-05,
|
| 2069 |
-
"loss": 0.3778,
|
| 2070 |
-
"step": 158500
|
| 2071 |
-
},
|
| 2072 |
-
{
|
| 2073 |
-
"epoch": 0.04,
|
| 2074 |
-
"learning_rate": 1.9117283187744766e-05,
|
| 2075 |
-
"loss": 0.3755,
|
| 2076 |
-
"step": 159000
|
| 2077 |
-
},
|
| 2078 |
-
{
|
| 2079 |
-
"epoch": 0.05,
|
| 2080 |
-
"learning_rate": 1.8904096414638772e-05,
|
| 2081 |
-
"loss": 0.3769,
|
| 2082 |
-
"step": 159500
|
| 2083 |
-
},
|
| 2084 |
-
{
|
| 2085 |
-
"epoch": 0.05,
|
| 2086 |
-
"learning_rate": 1.8693160302482794e-05,
|
| 2087 |
-
"loss": 0.3772,
|
| 2088 |
-
"step": 160000
|
| 2089 |
-
},
|
| 2090 |
-
{
|
| 2091 |
-
"epoch": 0.05,
|
| 2092 |
-
"learning_rate": 1.8484487862809858e-05,
|
| 2093 |
-
"loss": 0.3761,
|
| 2094 |
-
"step": 160500
|
| 2095 |
-
},
|
| 2096 |
-
{
|
| 2097 |
-
"epoch": 0.06,
|
| 2098 |
-
"learning_rate": 1.8278502478832604e-05,
|
| 2099 |
-
"loss": 0.3772,
|
| 2100 |
-
"step": 161000
|
| 2101 |
-
},
|
| 2102 |
-
{
|
| 2103 |
-
"epoch": 0.06,
|
| 2104 |
-
"learning_rate": 1.807439126823082e-05,
|
| 2105 |
-
"loss": 0.377,
|
| 2106 |
-
"step": 161500
|
| 2107 |
-
},
|
| 2108 |
-
{
|
| 2109 |
-
"epoch": 0.06,
|
| 2110 |
-
"learning_rate": 1.7872581898700765e-05,
|
| 2111 |
-
"loss": 0.3744,
|
| 2112 |
-
"step": 162000
|
| 2113 |
-
},
|
| 2114 |
-
{
|
| 2115 |
-
"epoch": 0.06,
|
| 2116 |
-
"learning_rate": 1.7673483491090614e-05,
|
| 2117 |
-
"loss": 0.3767,
|
| 2118 |
-
"step": 162500
|
| 2119 |
-
},
|
| 2120 |
-
{
|
| 2121 |
-
"epoch": 0.07,
|
| 2122 |
-
"learning_rate": 1.7476310341228535e-05,
|
| 2123 |
-
"loss": 0.3758,
|
| 2124 |
-
"step": 163000
|
| 2125 |
-
},
|
| 2126 |
-
{
|
| 2127 |
-
"epoch": 0.07,
|
| 2128 |
-
"learning_rate": 1.7281475924886973e-05,
|
| 2129 |
-
"loss": 0.3756,
|
| 2130 |
-
"step": 163500
|
| 2131 |
-
},
|
| 2132 |
-
{
|
| 2133 |
-
"epoch": 0.07,
|
| 2134 |
-
"learning_rate": 1.708899226037049e-05,
|
| 2135 |
-
"loss": 0.3764,
|
| 2136 |
-
"step": 164000
|
| 2137 |
-
},
|
| 2138 |
-
{
|
| 2139 |
-
"epoch": 0.07,
|
| 2140 |
-
"learning_rate": 1.6898871220978173e-05,
|
| 2141 |
-
"loss": 0.3764,
|
| 2142 |
-
"step": 164500
|
| 2143 |
-
},
|
| 2144 |
-
{
|
| 2145 |
-
"epoch": 0.07,
|
| 2146 |
-
"learning_rate": 1.671112453427132e-05,
|
| 2147 |
-
"loss": 0.3751,
|
| 2148 |
-
"step": 165000
|
| 2149 |
-
},
|
| 2150 |
-
{
|
| 2151 |
-
"epoch": 0.08,
|
| 2152 |
-
"learning_rate": 1.6526132114069465e-05,
|
| 2153 |
-
"loss": 0.3761,
|
| 2154 |
-
"step": 165500
|
| 2155 |
-
},
|
| 2156 |
-
{
|
| 2157 |
-
"epoch": 0.08,
|
| 2158 |
-
"learning_rate": 1.6343163922809553e-05,
|
| 2159 |
-
"loss": 0.3748,
|
| 2160 |
-
"step": 166000
|
| 2161 |
-
},
|
| 2162 |
-
{
|
| 2163 |
-
"epoch": 0.08,
|
| 2164 |
-
"learning_rate": 1.6162604362878833e-05,
|
| 2165 |
-
"loss": 0.3749,
|
| 2166 |
-
"step": 166500
|
| 2167 |
-
},
|
| 2168 |
-
{
|
| 2169 |
-
"epoch": 0.09,
|
| 2170 |
-
"learning_rate": 1.59844645720415e-05,
|
| 2171 |
-
"loss": 0.3755,
|
| 2172 |
-
"step": 167000
|
| 2173 |
-
},
|
| 2174 |
-
{
|
| 2175 |
-
"epoch": 0.09,
|
| 2176 |
-
"learning_rate": 1.5809104523740263e-05,
|
| 2177 |
-
"loss": 0.3755,
|
| 2178 |
-
"step": 167500
|
| 2179 |
-
},
|
| 2180 |
-
{
|
| 2181 |
-
"epoch": 0.09,
|
| 2182 |
-
"learning_rate": 1.5636176293585352e-05,
|
| 2183 |
-
"loss": 0.3746,
|
| 2184 |
-
"step": 168000
|
| 2185 |
-
},
|
| 2186 |
-
{
|
| 2187 |
-
"epoch": 0.09,
|
| 2188 |
-
"learning_rate": 1.5465351310379166e-05,
|
| 2189 |
-
"loss": 0.3734,
|
| 2190 |
-
"step": 168500
|
| 2191 |
-
},
|
| 2192 |
-
{
|
| 2193 |
-
"epoch": 0.1,
|
| 2194 |
-
"learning_rate": 1.5296989106118967e-05,
|
| 2195 |
-
"loss": 0.374,
|
| 2196 |
-
"step": 169000
|
| 2197 |
-
},
|
| 2198 |
-
{
|
| 2199 |
-
"epoch": 0.1,
|
| 2200 |
-
"learning_rate": 1.5131100066178594e-05,
|
| 2201 |
-
"loss": 0.3748,
|
| 2202 |
-
"step": 169500
|
| 2203 |
-
},
|
| 2204 |
-
{
|
| 2205 |
-
"epoch": 0.1,
|
| 2206 |
-
"learning_rate": 1.4967694423375435e-05,
|
| 2207 |
-
"loss": 0.3754,
|
| 2208 |
-
"step": 170000
|
| 2209 |
-
},
|
| 2210 |
-
{
|
| 2211 |
-
"epoch": 0.1,
|
| 2212 |
-
"learning_rate": 1.480678225733925e-05,
|
| 2213 |
-
"loss": 0.3739,
|
| 2214 |
-
"step": 170500
|
| 2215 |
-
},
|
| 2216 |
-
{
|
| 2217 |
-
"epoch": 0.1,
|
| 2218 |
-
"learning_rate": 1.4648373493890402e-05,
|
| 2219 |
-
"loss": 0.3728,
|
| 2220 |
-
"step": 171000
|
| 2221 |
-
},
|
| 2222 |
-
{
|
| 2223 |
-
"epoch": 0.11,
|
| 2224 |
-
"learning_rate": 1.4492477904427582e-05,
|
| 2225 |
-
"loss": 0.3748,
|
| 2226 |
-
"step": 171500
|
| 2227 |
-
},
|
| 2228 |
-
{
|
| 2229 |
-
"epoch": 0.11,
|
| 2230 |
-
"learning_rate": 1.4339105105325088e-05,
|
| 2231 |
-
"loss": 0.3735,
|
| 2232 |
-
"step": 172000
|
| 2233 |
-
},
|
| 2234 |
-
{
|
| 2235 |
-
"epoch": 0.11,
|
| 2236 |
-
"learning_rate": 1.4188264557339635e-05,
|
| 2237 |
-
"loss": 0.3737,
|
| 2238 |
-
"step": 172500
|
| 2239 |
-
},
|
| 2240 |
-
{
|
| 2241 |
-
"epoch": 0.12,
|
| 2242 |
-
"learning_rate": 1.403996556502672e-05,
|
| 2243 |
-
"loss": 0.3737,
|
| 2244 |
-
"step": 173000
|
| 2245 |
-
},
|
| 2246 |
-
{
|
| 2247 |
-
"epoch": 0.12,
|
| 2248 |
-
"learning_rate": 1.3894217276166754e-05,
|
| 2249 |
-
"loss": 0.3741,
|
| 2250 |
-
"step": 173500
|
| 2251 |
-
},
|
| 2252 |
-
{
|
| 2253 |
-
"epoch": 0.12,
|
| 2254 |
-
"learning_rate": 1.3751596324911567e-05,
|
| 2255 |
-
"loss": 0.3732,
|
| 2256 |
-
"step": 174000
|
| 2257 |
-
},
|
| 2258 |
-
{
|
| 2259 |
-
"epoch": 0.12,
|
| 2260 |
-
"learning_rate": 1.3610965964896096e-05,
|
| 2261 |
-
"loss": 0.374,
|
| 2262 |
-
"step": 174500
|
| 2263 |
-
},
|
| 2264 |
-
{
|
| 2265 |
-
"epoch": 0.12,
|
| 2266 |
-
"learning_rate": 1.3472912771049686e-05,
|
| 2267 |
-
"loss": 0.374,
|
| 2268 |
-
"step": 175000
|
| 2269 |
-
},
|
| 2270 |
-
{
|
| 2271 |
-
"epoch": 0.13,
|
| 2272 |
-
"learning_rate": 1.3337445259143613e-05,
|
| 2273 |
-
"loss": 0.3724,
|
| 2274 |
-
"step": 175500
|
| 2275 |
-
},
|
| 2276 |
-
{
|
| 2277 |
-
"epoch": 0.13,
|
| 2278 |
-
"learning_rate": 1.3204571785452118e-05,
|
| 2279 |
-
"loss": 0.3723,
|
| 2280 |
-
"step": 176000
|
| 2281 |
-
},
|
| 2282 |
-
{
|
| 2283 |
-
"epoch": 0.13,
|
| 2284 |
-
"learning_rate": 1.3074816436839109e-05,
|
| 2285 |
-
"loss": 0.3729,
|
| 2286 |
-
"step": 176500
|
| 2287 |
-
},
|
| 2288 |
-
{
|
| 2289 |
-
"epoch": 0.14,
|
| 2290 |
-
"learning_rate": 1.2947145010969087e-05,
|
| 2291 |
-
"loss": 0.3724,
|
| 2292 |
-
"step": 177000
|
| 2293 |
-
},
|
| 2294 |
-
{
|
| 2295 |
-
"epoch": 0.14,
|
| 2296 |
-
"learning_rate": 1.2822091698871432e-05,
|
| 2297 |
-
"loss": 0.3731,
|
| 2298 |
-
"step": 177500
|
| 2299 |
-
},
|
| 2300 |
-
{
|
| 2301 |
-
"epoch": 0.14,
|
| 2302 |
-
"learning_rate": 1.269966421442348e-05,
|
| 2303 |
-
"loss": 0.3718,
|
| 2304 |
-
"step": 178000
|
| 2305 |
-
},
|
| 2306 |
-
{
|
| 2307 |
-
"epoch": 0.14,
|
| 2308 |
-
"learning_rate": 1.257987010952917e-05,
|
| 2309 |
-
"loss": 0.3725,
|
| 2310 |
-
"step": 178500
|
| 2311 |
-
},
|
| 2312 |
-
{
|
| 2313 |
-
"epoch": 0.14,
|
| 2314 |
-
"learning_rate": 1.2462716773653154e-05,
|
| 2315 |
-
"loss": 0.3725,
|
| 2316 |
-
"step": 179000
|
| 2317 |
-
},
|
| 2318 |
-
{
|
| 2319 |
-
"epoch": 0.15,
|
| 2320 |
-
"learning_rate": 1.2348211433365053e-05,
|
| 2321 |
-
"loss": 0.3714,
|
| 2322 |
-
"step": 179500
|
| 2323 |
-
},
|
| 2324 |
-
{
|
| 2325 |
-
"epoch": 0.15,
|
| 2326 |
-
"learning_rate": 1.2236361151893606e-05,
|
| 2327 |
-
"loss": 0.3715,
|
| 2328 |
-
"step": 180000
|
| 2329 |
-
},
|
| 2330 |
-
{
|
| 2331 |
-
"epoch": 0.15,
|
| 2332 |
-
"learning_rate": 1.2127172828691041e-05,
|
| 2333 |
-
"loss": 0.373,
|
| 2334 |
-
"step": 180500
|
| 2335 |
-
},
|
| 2336 |
-
{
|
| 2337 |
-
"epoch": 0.15,
|
| 2338 |
-
"learning_rate": 1.2020863570515961e-05,
|
| 2339 |
-
"loss": 0.3743,
|
| 2340 |
-
"step": 181000
|
| 2341 |
-
},
|
| 2342 |
-
{
|
| 2343 |
-
"epoch": 0.16,
|
| 2344 |
-
"learning_rate": 1.1917013848007706e-05,
|
| 2345 |
-
"loss": 0.3745,
|
| 2346 |
-
"step": 181500
|
| 2347 |
-
},
|
| 2348 |
-
{
|
| 2349 |
-
"epoch": 0.16,
|
| 2350 |
-
"learning_rate": 1.1815845782614282e-05,
|
| 2351 |
-
"loss": 0.3729,
|
| 2352 |
-
"step": 182000
|
| 2353 |
-
},
|
| 2354 |
-
{
|
| 2355 |
-
"epoch": 0.16,
|
| 2356 |
-
"learning_rate": 1.1717365614858533e-05,
|
| 2357 |
-
"loss": 0.373,
|
| 2358 |
-
"step": 182500
|
| 2359 |
-
},
|
| 2360 |
-
{
|
| 2361 |
-
"epoch": 0.17,
|
| 2362 |
-
"learning_rate": 1.162157941946108e-05,
|
| 2363 |
-
"loss": 0.3733,
|
| 2364 |
-
"step": 183000
|
| 2365 |
-
},
|
| 2366 |
-
{
|
| 2367 |
-
"epoch": 0.17,
|
| 2368 |
-
"learning_rate": 1.1528493104965648e-05,
|
| 2369 |
-
"loss": 0.3736,
|
| 2370 |
-
"step": 183500
|
| 2371 |
-
},
|
| 2372 |
-
{
|
| 2373 |
-
"epoch": 0.17,
|
| 2374 |
-
"learning_rate": 1.1438112413374588e-05,
|
| 2375 |
-
"loss": 0.3744,
|
| 2376 |
-
"step": 184000
|
| 2377 |
-
},
|
| 2378 |
-
{
|
| 2379 |
-
"epoch": 0.17,
|
| 2380 |
-
"learning_rate": 1.1350442919794692e-05,
|
| 2381 |
-
"loss": 0.3743,
|
| 2382 |
-
"step": 184500
|
| 2383 |
-
},
|
| 2384 |
-
{
|
| 2385 |
-
"epoch": 0.17,
|
| 2386 |
-
"learning_rate": 1.126582442516417e-05,
|
| 2387 |
-
"loss": 0.3733,
|
| 2388 |
-
"step": 185000
|
| 2389 |
-
},
|
| 2390 |
-
{
|
| 2391 |
-
"epoch": 0.18,
|
| 2392 |
-
"learning_rate": 1.1183582486034581e-05,
|
| 2393 |
-
"loss": 0.3739,
|
| 2394 |
-
"step": 185500
|
| 2395 |
-
},
|
| 2396 |
-
{
|
| 2397 |
-
"epoch": 0.18,
|
| 2398 |
-
"learning_rate": 1.1104067445521018e-05,
|
| 2399 |
-
"loss": 0.3735,
|
| 2400 |
-
"step": 186000
|
| 2401 |
-
},
|
| 2402 |
-
{
|
| 2403 |
-
"epoch": 0.18,
|
| 2404 |
-
"learning_rate": 1.102728420848572e-05,
|
| 2405 |
-
"loss": 0.3732,
|
| 2406 |
-
"step": 186500
|
| 2407 |
-
},
|
| 2408 |
-
{
|
| 2409 |
-
"epoch": 0.18,
|
| 2410 |
-
"learning_rate": 1.0953237511280449e-05,
|
| 2411 |
-
"loss": 0.3731,
|
| 2412 |
-
"step": 187000
|
| 2413 |
-
},
|
| 2414 |
-
{
|
| 2415 |
-
"epoch": 0.19,
|
| 2416 |
-
"learning_rate": 1.0882071794057046e-05,
|
| 2417 |
-
"loss": 0.3732,
|
| 2418 |
-
"step": 187500
|
| 2419 |
-
},
|
| 2420 |
-
{
|
| 2421 |
-
"epoch": 0.19,
|
| 2422 |
-
"learning_rate": 1.0813506214785774e-05,
|
| 2423 |
-
"loss": 0.3732,
|
| 2424 |
-
"step": 188000
|
| 2425 |
-
},
|
| 2426 |
-
{
|
| 2427 |
-
"epoch": 0.19,
|
| 2428 |
-
"learning_rate": 1.0747690362178142e-05,
|
| 2429 |
-
"loss": 0.3732,
|
| 2430 |
-
"step": 188500
|
| 2431 |
-
},
|
| 2432 |
-
{
|
| 2433 |
-
"epoch": 0.2,
|
| 2434 |
-
"learning_rate": 1.0684628296065977e-05,
|
| 2435 |
-
"loss": 0.373,
|
| 2436 |
-
"step": 189000
|
| 2437 |
-
},
|
| 2438 |
-
{
|
| 2439 |
-
"epoch": 0.2,
|
| 2440 |
-
"learning_rate": 1.062444176053193e-05,
|
| 2441 |
-
"loss": 0.3728,
|
| 2442 |
-
"step": 189500
|
| 2443 |
-
},
|
| 2444 |
-
{
|
| 2445 |
-
"epoch": 0.2,
|
| 2446 |
-
"learning_rate": 1.0566893240808188e-05,
|
| 2447 |
-
"loss": 0.3743,
|
| 2448 |
-
"step": 190000
|
| 2449 |
-
},
|
| 2450 |
-
{
|
| 2451 |
-
"epoch": 0.0,
|
| 2452 |
-
"learning_rate": 1.0512109659997981e-05,
|
| 2453 |
-
"loss": 0.3673,
|
| 2454 |
-
"step": 190500
|
| 2455 |
-
},
|
| 2456 |
-
{
|
| 2457 |
-
"epoch": 0.01,
|
| 2458 |
-
"learning_rate": 1.0460094397410629e-05,
|
| 2459 |
-
"loss": 0.3658,
|
| 2460 |
-
"step": 191000
|
| 2461 |
-
},
|
| 2462 |
-
{
|
| 2463 |
-
"epoch": 0.01,
|
| 2464 |
-
"learning_rate": 1.041085066159254e-05,
|
| 2465 |
-
"loss": 0.3661,
|
| 2466 |
-
"step": 191500
|
| 2467 |
-
},
|
| 2468 |
-
{
|
| 2469 |
-
"epoch": 0.01,
|
| 2470 |
-
"learning_rate": 1.036447165752325e-05,
|
| 2471 |
-
"loss": 0.3648,
|
| 2472 |
-
"step": 192000
|
| 2473 |
-
},
|
| 2474 |
-
{
|
| 2475 |
-
"epoch": 0.01,
|
| 2476 |
-
"learning_rate": 1.03207743592438e-05,
|
| 2477 |
-
"loss": 0.3649,
|
| 2478 |
-
"step": 192500
|
| 2479 |
-
},
|
| 2480 |
-
{
|
| 2481 |
-
"epoch": 0.01,
|
| 2482 |
-
"learning_rate": 1.0279936239738395e-05,
|
| 2483 |
-
"loss": 0.3645,
|
| 2484 |
-
"step": 193000
|
| 2485 |
-
},
|
| 2486 |
-
{
|
| 2487 |
-
"epoch": 0.02,
|
| 2488 |
-
"learning_rate": 1.0241796139099991e-05,
|
| 2489 |
-
"loss": 0.3646,
|
| 2490 |
-
"step": 193500
|
| 2491 |
-
},
|
| 2492 |
-
{
|
| 2493 |
-
"epoch": 0.02,
|
| 2494 |
-
"learning_rate": 1.0206441030895691e-05,
|
| 2495 |
-
"loss": 0.3652,
|
| 2496 |
-
"step": 194000
|
| 2497 |
-
},
|
| 2498 |
-
{
|
| 2499 |
-
"epoch": 0.02,
|
| 2500 |
-
"learning_rate": 1.0173873095995124e-05,
|
| 2501 |
-
"loss": 0.3659,
|
| 2502 |
-
"step": 194500
|
| 2503 |
-
},
|
| 2504 |
-
{
|
| 2505 |
-
"epoch": 0.03,
|
| 2506 |
-
"learning_rate": 1.0144094343341912e-05,
|
| 2507 |
-
"loss": 0.3645,
|
| 2508 |
-
"step": 195000
|
| 2509 |
-
},
|
| 2510 |
-
{
|
| 2511 |
-
"epoch": 0.03,
|
| 2512 |
-
"learning_rate": 1.011715779872454e-05,
|
| 2513 |
-
"loss": 0.3676,
|
| 2514 |
-
"step": 195500
|
| 2515 |
-
},
|
| 2516 |
-
{
|
| 2517 |
-
"epoch": 0.03,
|
| 2518 |
-
"learning_rate": 1.0092957162169376e-05,
|
| 2519 |
-
"loss": 0.3673,
|
| 2520 |
-
"step": 196000
|
| 2521 |
-
},
|
| 2522 |
-
{
|
| 2523 |
-
"epoch": 0.03,
|
| 2524 |
-
"learning_rate": 1.0071550699137517e-05,
|
| 2525 |
-
"loss": 0.3674,
|
| 2526 |
-
"step": 196500
|
| 2527 |
-
},
|
| 2528 |
-
{
|
| 2529 |
-
"epoch": 0.04,
|
| 2530 |
-
"learning_rate": 1.0052939730080435e-05,
|
| 2531 |
-
"loss": 0.3678,
|
| 2532 |
-
"step": 197000
|
| 2533 |
-
},
|
| 2534 |
-
{
|
| 2535 |
-
"epoch": 0.04,
|
| 2536 |
-
"learning_rate": 1.0037125403010368e-05,
|
| 2537 |
-
"loss": 0.3688,
|
| 2538 |
-
"step": 197500
|
| 2539 |
-
},
|
| 2540 |
-
{
|
| 2541 |
-
"epoch": 0.04,
|
| 2542 |
-
"learning_rate": 1.0024108693429497e-05,
|
| 2543 |
-
"loss": 0.3676,
|
| 2544 |
-
"step": 198000
|
| 2545 |
-
},
|
| 2546 |
-
{
|
| 2547 |
-
"epoch": 0.04,
|
| 2548 |
-
"learning_rate": 1.0013890404269793e-05,
|
| 2549 |
-
"loss": 0.3685,
|
| 2550 |
-
"step": 198500
|
| 2551 |
-
},
|
| 2552 |
-
{
|
| 2553 |
-
"epoch": 0.04,
|
| 2554 |
-
"learning_rate": 1.0006471165843453e-05,
|
| 2555 |
-
"loss": 0.3664,
|
| 2556 |
-
"step": 199000
|
| 2557 |
-
},
|
| 2558 |
-
{
|
| 2559 |
-
"epoch": 0.05,
|
| 2560 |
-
"learning_rate": 1.000185143580406e-05,
|
| 2561 |
-
"loss": 0.3678,
|
| 2562 |
-
"step": 199500
|
| 2563 |
-
},
|
| 2564 |
-
{
|
| 2565 |
-
"epoch": 0.05,
|
| 2566 |
-
"learning_rate": 1.0000031499118314e-05,
|
| 2567 |
-
"loss": 0.3682,
|
| 2568 |
-
"step": 200000
|
| 2569 |
-
}
|
| 2570 |
-
],
|
| 2571 |
-
"max_steps": 200000,
|
| 2572 |
-
"num_train_epochs": 9223372036854775807,
|
| 2573 |
-
"total_flos": 4.709861347295232e+21,
|
| 2574 |
-
"trial_name": null,
|
| 2575 |
-
"trial_params": null
|
| 2576 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
last-checkpoint/training_args.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:53c18b4efeb50055b9df3de46e48e5ef92c97dea961e5c1e8f4076e024c3c1db
|
| 3 |
-
size 5551
|
|
|
|
|
|
|
|
|
|
|
|