Mitchel Hsu
commited on
Commit
•
507c812
1
Parent(s):
e77bf98
add: Update model
Browse filesThis view is limited to 50 files because it contains too many changes.
See raw diff
- adapter_config.json +0 -22
- checkpoint-1400/scaler.pt +0 -3
- checkpoint-1400/scheduler.pt +0 -3
- checkpoint-1400/trainer_state.json +0 -912
- checkpoint-1400/training_args.bin +0 -3
- checkpoint-3600/optimizer.pt +0 -3
- checkpoint-3600/pytorch_model.bin +0 -3
- checkpoint-3600/rng_state.pth +0 -3
- checkpoint-3600/scaler.pt +0 -3
- checkpoint-3600/scheduler.pt +0 -3
- checkpoint-3600/trainer_state.json +0 -2320
- checkpoint-3600/training_args.bin +0 -3
- checkpoint-3800/optimizer.pt +0 -3
- checkpoint-3800/pytorch_model.bin +0 -3
- checkpoint-3800/rng_state.pth +0 -3
- checkpoint-3800/scaler.pt +0 -3
- checkpoint-3800/scheduler.pt +0 -3
- checkpoint-3800/trainer_state.json +0 -2448
- checkpoint-3800/training_args.bin +0 -3
- config.json +24 -0
- generation_config.json +7 -0
- checkpoint-1400/rng_state.pth → pytorch_model-00001-of-00039.bin +2 -2
- adapter_model.bin → pytorch_model-00002-of-00039.bin +2 -2
- checkpoint-1400/optimizer.pt → pytorch_model-00003-of-00039.bin +2 -2
- checkpoint-1400/pytorch_model.bin → pytorch_model-00004-of-00039.bin +2 -2
- pytorch_model-00005-of-00039.bin +3 -0
- pytorch_model-00006-of-00039.bin +3 -0
- pytorch_model-00007-of-00039.bin +3 -0
- pytorch_model-00008-of-00039.bin +3 -0
- pytorch_model-00009-of-00039.bin +3 -0
- pytorch_model-00010-of-00039.bin +3 -0
- pytorch_model-00011-of-00039.bin +3 -0
- pytorch_model-00012-of-00039.bin +3 -0
- pytorch_model-00013-of-00039.bin +3 -0
- pytorch_model-00014-of-00039.bin +3 -0
- pytorch_model-00015-of-00039.bin +3 -0
- pytorch_model-00016-of-00039.bin +3 -0
- pytorch_model-00017-of-00039.bin +3 -0
- pytorch_model-00018-of-00039.bin +3 -0
- pytorch_model-00019-of-00039.bin +3 -0
- pytorch_model-00020-of-00039.bin +3 -0
- pytorch_model-00021-of-00039.bin +3 -0
- pytorch_model-00022-of-00039.bin +3 -0
- pytorch_model-00023-of-00039.bin +3 -0
- pytorch_model-00024-of-00039.bin +3 -0
- pytorch_model-00025-of-00039.bin +3 -0
- pytorch_model-00026-of-00039.bin +3 -0
- pytorch_model-00027-of-00039.bin +3 -0
- pytorch_model-00028-of-00039.bin +3 -0
- pytorch_model-00029-of-00039.bin +3 -0
adapter_config.json
DELETED
|
@@ -1,22 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"base_model_name_or_path": "decapoda-research/llama-7b-hf",
|
| 3 |
-
"bias": "none",
|
| 4 |
-
"fan_in_fan_out": false,
|
| 5 |
-
"inference_mode": true,
|
| 6 |
-
"init_lora_weights": true,
|
| 7 |
-
"layers_pattern": null,
|
| 8 |
-
"layers_to_transform": null,
|
| 9 |
-
"lora_alpha": 16,
|
| 10 |
-
"lora_dropout": 0.05,
|
| 11 |
-
"modules_to_save": null,
|
| 12 |
-
"peft_type": "LORA",
|
| 13 |
-
"r": 16,
|
| 14 |
-
"revision": null,
|
| 15 |
-
"target_modules": [
|
| 16 |
-
"q_proj",
|
| 17 |
-
"k_proj",
|
| 18 |
-
"v_proj",
|
| 19 |
-
"o_proj"
|
| 20 |
-
],
|
| 21 |
-
"task_type": "CAUSAL_LM"
|
| 22 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-1400/scaler.pt
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:16fdfc03b58220402968eacaac23fb5471cdb9061302380bd3c8d4d326c02ade
|
| 3 |
-
size 557
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-1400/scheduler.pt
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:0dcd27fef07230ceb8ed85e2a9692df56c7d01b0fe9f962e0a8e2690c9acc9b3
|
| 3 |
-
size 627
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-1400/trainer_state.json
DELETED
|
@@ -1,912 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_metric": 0.8923280239105225,
|
| 3 |
-
"best_model_checkpoint": "./lora-alpaca-hc8/checkpoint-1400",
|
| 4 |
-
"epoch": 3.6012861736334405,
|
| 5 |
-
"global_step": 1400,
|
| 6 |
-
"is_hyper_param_search": false,
|
| 7 |
-
"is_local_process_zero": true,
|
| 8 |
-
"is_world_process_zero": true,
|
| 9 |
-
"log_history": [
|
| 10 |
-
{
|
| 11 |
-
"epoch": 0.03,
|
| 12 |
-
"learning_rate": 2.9999999999999997e-05,
|
| 13 |
-
"loss": 1.3719,
|
| 14 |
-
"step": 10
|
| 15 |
-
},
|
| 16 |
-
{
|
| 17 |
-
"epoch": 0.05,
|
| 18 |
-
"learning_rate": 5.9999999999999995e-05,
|
| 19 |
-
"loss": 1.5358,
|
| 20 |
-
"step": 20
|
| 21 |
-
},
|
| 22 |
-
{
|
| 23 |
-
"epoch": 0.08,
|
| 24 |
-
"learning_rate": 8.999999999999999e-05,
|
| 25 |
-
"loss": 1.5633,
|
| 26 |
-
"step": 30
|
| 27 |
-
},
|
| 28 |
-
{
|
| 29 |
-
"epoch": 0.1,
|
| 30 |
-
"learning_rate": 0.00011999999999999999,
|
| 31 |
-
"loss": 1.2433,
|
| 32 |
-
"step": 40
|
| 33 |
-
},
|
| 34 |
-
{
|
| 35 |
-
"epoch": 0.13,
|
| 36 |
-
"learning_rate": 0.00015,
|
| 37 |
-
"loss": 0.9843,
|
| 38 |
-
"step": 50
|
| 39 |
-
},
|
| 40 |
-
{
|
| 41 |
-
"epoch": 0.15,
|
| 42 |
-
"learning_rate": 0.00017999999999999998,
|
| 43 |
-
"loss": 1.012,
|
| 44 |
-
"step": 60
|
| 45 |
-
},
|
| 46 |
-
{
|
| 47 |
-
"epoch": 0.18,
|
| 48 |
-
"learning_rate": 0.00020999999999999998,
|
| 49 |
-
"loss": 1.0392,
|
| 50 |
-
"step": 70
|
| 51 |
-
},
|
| 52 |
-
{
|
| 53 |
-
"epoch": 0.21,
|
| 54 |
-
"learning_rate": 0.00023999999999999998,
|
| 55 |
-
"loss": 1.0541,
|
| 56 |
-
"step": 80
|
| 57 |
-
},
|
| 58 |
-
{
|
| 59 |
-
"epoch": 0.23,
|
| 60 |
-
"learning_rate": 0.00027,
|
| 61 |
-
"loss": 0.9608,
|
| 62 |
-
"step": 90
|
| 63 |
-
},
|
| 64 |
-
{
|
| 65 |
-
"epoch": 0.26,
|
| 66 |
-
"learning_rate": 0.0003,
|
| 67 |
-
"loss": 0.8697,
|
| 68 |
-
"step": 100
|
| 69 |
-
},
|
| 70 |
-
{
|
| 71 |
-
"epoch": 0.28,
|
| 72 |
-
"learning_rate": 0.00029920634920634916,
|
| 73 |
-
"loss": 0.9836,
|
| 74 |
-
"step": 110
|
| 75 |
-
},
|
| 76 |
-
{
|
| 77 |
-
"epoch": 0.31,
|
| 78 |
-
"learning_rate": 0.00029841269841269835,
|
| 79 |
-
"loss": 1.003,
|
| 80 |
-
"step": 120
|
| 81 |
-
},
|
| 82 |
-
{
|
| 83 |
-
"epoch": 0.33,
|
| 84 |
-
"learning_rate": 0.0002976190476190476,
|
| 85 |
-
"loss": 1.0052,
|
| 86 |
-
"step": 130
|
| 87 |
-
},
|
| 88 |
-
{
|
| 89 |
-
"epoch": 0.36,
|
| 90 |
-
"learning_rate": 0.0002968253968253968,
|
| 91 |
-
"loss": 0.9267,
|
| 92 |
-
"step": 140
|
| 93 |
-
},
|
| 94 |
-
{
|
| 95 |
-
"epoch": 0.39,
|
| 96 |
-
"learning_rate": 0.000296031746031746,
|
| 97 |
-
"loss": 0.8318,
|
| 98 |
-
"step": 150
|
| 99 |
-
},
|
| 100 |
-
{
|
| 101 |
-
"epoch": 0.41,
|
| 102 |
-
"learning_rate": 0.0002952380952380952,
|
| 103 |
-
"loss": 0.9594,
|
| 104 |
-
"step": 160
|
| 105 |
-
},
|
| 106 |
-
{
|
| 107 |
-
"epoch": 0.44,
|
| 108 |
-
"learning_rate": 0.00029444444444444445,
|
| 109 |
-
"loss": 0.987,
|
| 110 |
-
"step": 170
|
| 111 |
-
},
|
| 112 |
-
{
|
| 113 |
-
"epoch": 0.46,
|
| 114 |
-
"learning_rate": 0.00029365079365079364,
|
| 115 |
-
"loss": 0.9646,
|
| 116 |
-
"step": 180
|
| 117 |
-
},
|
| 118 |
-
{
|
| 119 |
-
"epoch": 0.49,
|
| 120 |
-
"learning_rate": 0.00029285714285714283,
|
| 121 |
-
"loss": 0.8501,
|
| 122 |
-
"step": 190
|
| 123 |
-
},
|
| 124 |
-
{
|
| 125 |
-
"epoch": 0.51,
|
| 126 |
-
"learning_rate": 0.000292063492063492,
|
| 127 |
-
"loss": 0.7568,
|
| 128 |
-
"step": 200
|
| 129 |
-
},
|
| 130 |
-
{
|
| 131 |
-
"epoch": 0.51,
|
| 132 |
-
"eval_loss": 0.9958714246749878,
|
| 133 |
-
"eval_runtime": 189.2223,
|
| 134 |
-
"eval_samples_per_second": 10.57,
|
| 135 |
-
"eval_steps_per_second": 1.321,
|
| 136 |
-
"step": 200
|
| 137 |
-
},
|
| 138 |
-
{
|
| 139 |
-
"epoch": 0.54,
|
| 140 |
-
"learning_rate": 0.00029126984126984126,
|
| 141 |
-
"loss": 0.949,
|
| 142 |
-
"step": 210
|
| 143 |
-
},
|
| 144 |
-
{
|
| 145 |
-
"epoch": 0.57,
|
| 146 |
-
"learning_rate": 0.00029047619047619045,
|
| 147 |
-
"loss": 0.9581,
|
| 148 |
-
"step": 220
|
| 149 |
-
},
|
| 150 |
-
{
|
| 151 |
-
"epoch": 0.59,
|
| 152 |
-
"learning_rate": 0.00028968253968253963,
|
| 153 |
-
"loss": 0.9526,
|
| 154 |
-
"step": 230
|
| 155 |
-
},
|
| 156 |
-
{
|
| 157 |
-
"epoch": 0.62,
|
| 158 |
-
"learning_rate": 0.0002888888888888888,
|
| 159 |
-
"loss": 0.847,
|
| 160 |
-
"step": 240
|
| 161 |
-
},
|
| 162 |
-
{
|
| 163 |
-
"epoch": 0.64,
|
| 164 |
-
"learning_rate": 0.00028809523809523806,
|
| 165 |
-
"loss": 0.7414,
|
| 166 |
-
"step": 250
|
| 167 |
-
},
|
| 168 |
-
{
|
| 169 |
-
"epoch": 0.67,
|
| 170 |
-
"learning_rate": 0.00028730158730158725,
|
| 171 |
-
"loss": 0.9449,
|
| 172 |
-
"step": 260
|
| 173 |
-
},
|
| 174 |
-
{
|
| 175 |
-
"epoch": 0.69,
|
| 176 |
-
"learning_rate": 0.0002865079365079365,
|
| 177 |
-
"loss": 0.9607,
|
| 178 |
-
"step": 270
|
| 179 |
-
},
|
| 180 |
-
{
|
| 181 |
-
"epoch": 0.72,
|
| 182 |
-
"learning_rate": 0.0002857142857142857,
|
| 183 |
-
"loss": 0.9456,
|
| 184 |
-
"step": 280
|
| 185 |
-
},
|
| 186 |
-
{
|
| 187 |
-
"epoch": 0.75,
|
| 188 |
-
"learning_rate": 0.0002849206349206349,
|
| 189 |
-
"loss": 0.8346,
|
| 190 |
-
"step": 290
|
| 191 |
-
},
|
| 192 |
-
{
|
| 193 |
-
"epoch": 0.77,
|
| 194 |
-
"learning_rate": 0.0002841269841269841,
|
| 195 |
-
"loss": 0.7275,
|
| 196 |
-
"step": 300
|
| 197 |
-
},
|
| 198 |
-
{
|
| 199 |
-
"epoch": 0.8,
|
| 200 |
-
"learning_rate": 0.0002833333333333333,
|
| 201 |
-
"loss": 0.9337,
|
| 202 |
-
"step": 310
|
| 203 |
-
},
|
| 204 |
-
{
|
| 205 |
-
"epoch": 0.82,
|
| 206 |
-
"learning_rate": 0.0002825396825396825,
|
| 207 |
-
"loss": 0.9466,
|
| 208 |
-
"step": 320
|
| 209 |
-
},
|
| 210 |
-
{
|
| 211 |
-
"epoch": 0.85,
|
| 212 |
-
"learning_rate": 0.00028174603174603173,
|
| 213 |
-
"loss": 0.9386,
|
| 214 |
-
"step": 330
|
| 215 |
-
},
|
| 216 |
-
{
|
| 217 |
-
"epoch": 0.87,
|
| 218 |
-
"learning_rate": 0.0002809523809523809,
|
| 219 |
-
"loss": 0.8254,
|
| 220 |
-
"step": 340
|
| 221 |
-
},
|
| 222 |
-
{
|
| 223 |
-
"epoch": 0.9,
|
| 224 |
-
"learning_rate": 0.0002801587301587301,
|
| 225 |
-
"loss": 0.723,
|
| 226 |
-
"step": 350
|
| 227 |
-
},
|
| 228 |
-
{
|
| 229 |
-
"epoch": 0.93,
|
| 230 |
-
"learning_rate": 0.00027936507936507935,
|
| 231 |
-
"loss": 0.9274,
|
| 232 |
-
"step": 360
|
| 233 |
-
},
|
| 234 |
-
{
|
| 235 |
-
"epoch": 0.95,
|
| 236 |
-
"learning_rate": 0.00027857142857142854,
|
| 237 |
-
"loss": 0.9492,
|
| 238 |
-
"step": 370
|
| 239 |
-
},
|
| 240 |
-
{
|
| 241 |
-
"epoch": 0.98,
|
| 242 |
-
"learning_rate": 0.0002777777777777778,
|
| 243 |
-
"loss": 0.871,
|
| 244 |
-
"step": 380
|
| 245 |
-
},
|
| 246 |
-
{
|
| 247 |
-
"epoch": 1.0,
|
| 248 |
-
"learning_rate": 0.00027698412698412697,
|
| 249 |
-
"loss": 0.7533,
|
| 250 |
-
"step": 390
|
| 251 |
-
},
|
| 252 |
-
{
|
| 253 |
-
"epoch": 1.03,
|
| 254 |
-
"learning_rate": 0.00027619047619047615,
|
| 255 |
-
"loss": 0.9284,
|
| 256 |
-
"step": 400
|
| 257 |
-
},
|
| 258 |
-
{
|
| 259 |
-
"epoch": 1.03,
|
| 260 |
-
"eval_loss": 0.9276881217956543,
|
| 261 |
-
"eval_runtime": 189.2181,
|
| 262 |
-
"eval_samples_per_second": 10.57,
|
| 263 |
-
"eval_steps_per_second": 1.321,
|
| 264 |
-
"step": 400
|
| 265 |
-
},
|
| 266 |
-
{
|
| 267 |
-
"epoch": 1.05,
|
| 268 |
-
"learning_rate": 0.0002753968253968254,
|
| 269 |
-
"loss": 0.9402,
|
| 270 |
-
"step": 410
|
| 271 |
-
},
|
| 272 |
-
{
|
| 273 |
-
"epoch": 1.08,
|
| 274 |
-
"learning_rate": 0.0002746031746031746,
|
| 275 |
-
"loss": 0.9188,
|
| 276 |
-
"step": 420
|
| 277 |
-
},
|
| 278 |
-
{
|
| 279 |
-
"epoch": 1.11,
|
| 280 |
-
"learning_rate": 0.00027380952380952377,
|
| 281 |
-
"loss": 0.7976,
|
| 282 |
-
"step": 430
|
| 283 |
-
},
|
| 284 |
-
{
|
| 285 |
-
"epoch": 1.13,
|
| 286 |
-
"learning_rate": 0.00027301587301587296,
|
| 287 |
-
"loss": 0.7172,
|
| 288 |
-
"step": 440
|
| 289 |
-
},
|
| 290 |
-
{
|
| 291 |
-
"epoch": 1.16,
|
| 292 |
-
"learning_rate": 0.0002722222222222222,
|
| 293 |
-
"loss": 0.9195,
|
| 294 |
-
"step": 450
|
| 295 |
-
},
|
| 296 |
-
{
|
| 297 |
-
"epoch": 1.18,
|
| 298 |
-
"learning_rate": 0.0002714285714285714,
|
| 299 |
-
"loss": 0.9426,
|
| 300 |
-
"step": 460
|
| 301 |
-
},
|
| 302 |
-
{
|
| 303 |
-
"epoch": 1.21,
|
| 304 |
-
"learning_rate": 0.00027063492063492063,
|
| 305 |
-
"loss": 0.9034,
|
| 306 |
-
"step": 470
|
| 307 |
-
},
|
| 308 |
-
{
|
| 309 |
-
"epoch": 1.23,
|
| 310 |
-
"learning_rate": 0.0002698412698412698,
|
| 311 |
-
"loss": 0.788,
|
| 312 |
-
"step": 480
|
| 313 |
-
},
|
| 314 |
-
{
|
| 315 |
-
"epoch": 1.26,
|
| 316 |
-
"learning_rate": 0.000269047619047619,
|
| 317 |
-
"loss": 0.7213,
|
| 318 |
-
"step": 490
|
| 319 |
-
},
|
| 320 |
-
{
|
| 321 |
-
"epoch": 1.29,
|
| 322 |
-
"learning_rate": 0.00026825396825396825,
|
| 323 |
-
"loss": 0.9149,
|
| 324 |
-
"step": 500
|
| 325 |
-
},
|
| 326 |
-
{
|
| 327 |
-
"epoch": 1.31,
|
| 328 |
-
"learning_rate": 0.00026746031746031744,
|
| 329 |
-
"loss": 0.9386,
|
| 330 |
-
"step": 510
|
| 331 |
-
},
|
| 332 |
-
{
|
| 333 |
-
"epoch": 1.34,
|
| 334 |
-
"learning_rate": 0.0002666666666666666,
|
| 335 |
-
"loss": 0.9099,
|
| 336 |
-
"step": 520
|
| 337 |
-
},
|
| 338 |
-
{
|
| 339 |
-
"epoch": 1.36,
|
| 340 |
-
"learning_rate": 0.00026587301587301587,
|
| 341 |
-
"loss": 0.7802,
|
| 342 |
-
"step": 530
|
| 343 |
-
},
|
| 344 |
-
{
|
| 345 |
-
"epoch": 1.39,
|
| 346 |
-
"learning_rate": 0.00026507936507936506,
|
| 347 |
-
"loss": 0.7246,
|
| 348 |
-
"step": 540
|
| 349 |
-
},
|
| 350 |
-
{
|
| 351 |
-
"epoch": 1.41,
|
| 352 |
-
"learning_rate": 0.00026428571428571424,
|
| 353 |
-
"loss": 0.9213,
|
| 354 |
-
"step": 550
|
| 355 |
-
},
|
| 356 |
-
{
|
| 357 |
-
"epoch": 1.44,
|
| 358 |
-
"learning_rate": 0.00026349206349206343,
|
| 359 |
-
"loss": 0.9348,
|
| 360 |
-
"step": 560
|
| 361 |
-
},
|
| 362 |
-
{
|
| 363 |
-
"epoch": 1.47,
|
| 364 |
-
"learning_rate": 0.0002626984126984127,
|
| 365 |
-
"loss": 0.8995,
|
| 366 |
-
"step": 570
|
| 367 |
-
},
|
| 368 |
-
{
|
| 369 |
-
"epoch": 1.49,
|
| 370 |
-
"learning_rate": 0.00026190476190476186,
|
| 371 |
-
"loss": 0.7866,
|
| 372 |
-
"step": 580
|
| 373 |
-
},
|
| 374 |
-
{
|
| 375 |
-
"epoch": 1.52,
|
| 376 |
-
"learning_rate": 0.0002611111111111111,
|
| 377 |
-
"loss": 0.7257,
|
| 378 |
-
"step": 590
|
| 379 |
-
},
|
| 380 |
-
{
|
| 381 |
-
"epoch": 1.54,
|
| 382 |
-
"learning_rate": 0.0002603174603174603,
|
| 383 |
-
"loss": 0.918,
|
| 384 |
-
"step": 600
|
| 385 |
-
},
|
| 386 |
-
{
|
| 387 |
-
"epoch": 1.54,
|
| 388 |
-
"eval_loss": 0.9236659407615662,
|
| 389 |
-
"eval_runtime": 189.1236,
|
| 390 |
-
"eval_samples_per_second": 10.575,
|
| 391 |
-
"eval_steps_per_second": 1.322,
|
| 392 |
-
"step": 600
|
| 393 |
-
},
|
| 394 |
-
{
|
| 395 |
-
"epoch": 1.57,
|
| 396 |
-
"learning_rate": 0.00025952380952380953,
|
| 397 |
-
"loss": 0.933,
|
| 398 |
-
"step": 610
|
| 399 |
-
},
|
| 400 |
-
{
|
| 401 |
-
"epoch": 1.59,
|
| 402 |
-
"learning_rate": 0.0002587301587301587,
|
| 403 |
-
"loss": 0.9085,
|
| 404 |
-
"step": 620
|
| 405 |
-
},
|
| 406 |
-
{
|
| 407 |
-
"epoch": 1.62,
|
| 408 |
-
"learning_rate": 0.0002579365079365079,
|
| 409 |
-
"loss": 0.7928,
|
| 410 |
-
"step": 630
|
| 411 |
-
},
|
| 412 |
-
{
|
| 413 |
-
"epoch": 1.65,
|
| 414 |
-
"learning_rate": 0.0002571428571428571,
|
| 415 |
-
"loss": 0.7162,
|
| 416 |
-
"step": 640
|
| 417 |
-
},
|
| 418 |
-
{
|
| 419 |
-
"epoch": 1.67,
|
| 420 |
-
"learning_rate": 0.00025634920634920634,
|
| 421 |
-
"loss": 0.9076,
|
| 422 |
-
"step": 650
|
| 423 |
-
},
|
| 424 |
-
{
|
| 425 |
-
"epoch": 1.7,
|
| 426 |
-
"learning_rate": 0.00025555555555555553,
|
| 427 |
-
"loss": 0.9345,
|
| 428 |
-
"step": 660
|
| 429 |
-
},
|
| 430 |
-
{
|
| 431 |
-
"epoch": 1.72,
|
| 432 |
-
"learning_rate": 0.0002547619047619047,
|
| 433 |
-
"loss": 0.9107,
|
| 434 |
-
"step": 670
|
| 435 |
-
},
|
| 436 |
-
{
|
| 437 |
-
"epoch": 1.75,
|
| 438 |
-
"learning_rate": 0.00025396825396825396,
|
| 439 |
-
"loss": 0.7721,
|
| 440 |
-
"step": 680
|
| 441 |
-
},
|
| 442 |
-
{
|
| 443 |
-
"epoch": 1.77,
|
| 444 |
-
"learning_rate": 0.00025317460317460315,
|
| 445 |
-
"loss": 0.7112,
|
| 446 |
-
"step": 690
|
| 447 |
-
},
|
| 448 |
-
{
|
| 449 |
-
"epoch": 1.8,
|
| 450 |
-
"learning_rate": 0.0002523809523809524,
|
| 451 |
-
"loss": 0.9118,
|
| 452 |
-
"step": 700
|
| 453 |
-
},
|
| 454 |
-
{
|
| 455 |
-
"epoch": 1.83,
|
| 456 |
-
"learning_rate": 0.0002515873015873016,
|
| 457 |
-
"loss": 0.9205,
|
| 458 |
-
"step": 710
|
| 459 |
-
},
|
| 460 |
-
{
|
| 461 |
-
"epoch": 1.85,
|
| 462 |
-
"learning_rate": 0.00025079365079365076,
|
| 463 |
-
"loss": 0.9004,
|
| 464 |
-
"step": 720
|
| 465 |
-
},
|
| 466 |
-
{
|
| 467 |
-
"epoch": 1.88,
|
| 468 |
-
"learning_rate": 0.00025,
|
| 469 |
-
"loss": 0.7741,
|
| 470 |
-
"step": 730
|
| 471 |
-
},
|
| 472 |
-
{
|
| 473 |
-
"epoch": 1.9,
|
| 474 |
-
"learning_rate": 0.0002492063492063492,
|
| 475 |
-
"loss": 0.7186,
|
| 476 |
-
"step": 740
|
| 477 |
-
},
|
| 478 |
-
{
|
| 479 |
-
"epoch": 1.93,
|
| 480 |
-
"learning_rate": 0.0002484126984126984,
|
| 481 |
-
"loss": 0.9002,
|
| 482 |
-
"step": 750
|
| 483 |
-
},
|
| 484 |
-
{
|
| 485 |
-
"epoch": 1.95,
|
| 486 |
-
"learning_rate": 0.00024761904761904757,
|
| 487 |
-
"loss": 0.9066,
|
| 488 |
-
"step": 760
|
| 489 |
-
},
|
| 490 |
-
{
|
| 491 |
-
"epoch": 1.98,
|
| 492 |
-
"learning_rate": 0.0002468253968253968,
|
| 493 |
-
"loss": 0.8127,
|
| 494 |
-
"step": 770
|
| 495 |
-
},
|
| 496 |
-
{
|
| 497 |
-
"epoch": 2.01,
|
| 498 |
-
"learning_rate": 0.000246031746031746,
|
| 499 |
-
"loss": 0.7305,
|
| 500 |
-
"step": 780
|
| 501 |
-
},
|
| 502 |
-
{
|
| 503 |
-
"epoch": 2.03,
|
| 504 |
-
"learning_rate": 0.0002452380952380952,
|
| 505 |
-
"loss": 0.8921,
|
| 506 |
-
"step": 790
|
| 507 |
-
},
|
| 508 |
-
{
|
| 509 |
-
"epoch": 2.06,
|
| 510 |
-
"learning_rate": 0.00024444444444444443,
|
| 511 |
-
"loss": 0.9178,
|
| 512 |
-
"step": 800
|
| 513 |
-
},
|
| 514 |
-
{
|
| 515 |
-
"epoch": 2.06,
|
| 516 |
-
"eval_loss": 0.901778519153595,
|
| 517 |
-
"eval_runtime": 189.2038,
|
| 518 |
-
"eval_samples_per_second": 10.571,
|
| 519 |
-
"eval_steps_per_second": 1.321,
|
| 520 |
-
"step": 800
|
| 521 |
-
},
|
| 522 |
-
{
|
| 523 |
-
"epoch": 2.08,
|
| 524 |
-
"learning_rate": 0.00024365079365079364,
|
| 525 |
-
"loss": 0.8823,
|
| 526 |
-
"step": 810
|
| 527 |
-
},
|
| 528 |
-
{
|
| 529 |
-
"epoch": 2.11,
|
| 530 |
-
"learning_rate": 0.00024285714285714283,
|
| 531 |
-
"loss": 0.7521,
|
| 532 |
-
"step": 820
|
| 533 |
-
},
|
| 534 |
-
{
|
| 535 |
-
"epoch": 2.14,
|
| 536 |
-
"learning_rate": 0.00024206349206349205,
|
| 537 |
-
"loss": 0.717,
|
| 538 |
-
"step": 830
|
| 539 |
-
},
|
| 540 |
-
{
|
| 541 |
-
"epoch": 2.16,
|
| 542 |
-
"learning_rate": 0.00024126984126984123,
|
| 543 |
-
"loss": 0.9045,
|
| 544 |
-
"step": 840
|
| 545 |
-
},
|
| 546 |
-
{
|
| 547 |
-
"epoch": 2.19,
|
| 548 |
-
"learning_rate": 0.00024047619047619048,
|
| 549 |
-
"loss": 0.9146,
|
| 550 |
-
"step": 850
|
| 551 |
-
},
|
| 552 |
-
{
|
| 553 |
-
"epoch": 2.21,
|
| 554 |
-
"learning_rate": 0.00023968253968253966,
|
| 555 |
-
"loss": 0.8678,
|
| 556 |
-
"step": 860
|
| 557 |
-
},
|
| 558 |
-
{
|
| 559 |
-
"epoch": 2.24,
|
| 560 |
-
"learning_rate": 0.00023888888888888885,
|
| 561 |
-
"loss": 0.745,
|
| 562 |
-
"step": 870
|
| 563 |
-
},
|
| 564 |
-
{
|
| 565 |
-
"epoch": 2.26,
|
| 566 |
-
"learning_rate": 0.00023809523809523807,
|
| 567 |
-
"loss": 0.7226,
|
| 568 |
-
"step": 880
|
| 569 |
-
},
|
| 570 |
-
{
|
| 571 |
-
"epoch": 2.29,
|
| 572 |
-
"learning_rate": 0.00023730158730158728,
|
| 573 |
-
"loss": 0.9002,
|
| 574 |
-
"step": 890
|
| 575 |
-
},
|
| 576 |
-
{
|
| 577 |
-
"epoch": 2.32,
|
| 578 |
-
"learning_rate": 0.0002365079365079365,
|
| 579 |
-
"loss": 0.9035,
|
| 580 |
-
"step": 900
|
| 581 |
-
},
|
| 582 |
-
{
|
| 583 |
-
"epoch": 2.34,
|
| 584 |
-
"learning_rate": 0.00023571428571428569,
|
| 585 |
-
"loss": 0.878,
|
| 586 |
-
"step": 910
|
| 587 |
-
},
|
| 588 |
-
{
|
| 589 |
-
"epoch": 2.37,
|
| 590 |
-
"learning_rate": 0.00023492063492063487,
|
| 591 |
-
"loss": 0.7397,
|
| 592 |
-
"step": 920
|
| 593 |
-
},
|
| 594 |
-
{
|
| 595 |
-
"epoch": 2.39,
|
| 596 |
-
"learning_rate": 0.00023412698412698412,
|
| 597 |
-
"loss": 0.7251,
|
| 598 |
-
"step": 930
|
| 599 |
-
},
|
| 600 |
-
{
|
| 601 |
-
"epoch": 2.42,
|
| 602 |
-
"learning_rate": 0.0002333333333333333,
|
| 603 |
-
"loss": 0.9014,
|
| 604 |
-
"step": 940
|
| 605 |
-
},
|
| 606 |
-
{
|
| 607 |
-
"epoch": 2.44,
|
| 608 |
-
"learning_rate": 0.00023253968253968252,
|
| 609 |
-
"loss": 0.9158,
|
| 610 |
-
"step": 950
|
| 611 |
-
},
|
| 612 |
-
{
|
| 613 |
-
"epoch": 2.47,
|
| 614 |
-
"learning_rate": 0.0002317460317460317,
|
| 615 |
-
"loss": 0.8596,
|
| 616 |
-
"step": 960
|
| 617 |
-
},
|
| 618 |
-
{
|
| 619 |
-
"epoch": 2.5,
|
| 620 |
-
"learning_rate": 0.00023095238095238095,
|
| 621 |
-
"loss": 0.7312,
|
| 622 |
-
"step": 970
|
| 623 |
-
},
|
| 624 |
-
{
|
| 625 |
-
"epoch": 2.52,
|
| 626 |
-
"learning_rate": 0.00023015873015873014,
|
| 627 |
-
"loss": 0.7271,
|
| 628 |
-
"step": 980
|
| 629 |
-
},
|
| 630 |
-
{
|
| 631 |
-
"epoch": 2.55,
|
| 632 |
-
"learning_rate": 0.00022936507936507935,
|
| 633 |
-
"loss": 0.9007,
|
| 634 |
-
"step": 990
|
| 635 |
-
},
|
| 636 |
-
{
|
| 637 |
-
"epoch": 2.57,
|
| 638 |
-
"learning_rate": 0.00022857142857142854,
|
| 639 |
-
"loss": 0.9186,
|
| 640 |
-
"step": 1000
|
| 641 |
-
},
|
| 642 |
-
{
|
| 643 |
-
"epoch": 2.57,
|
| 644 |
-
"eval_loss": 0.8995742201805115,
|
| 645 |
-
"eval_runtime": 189.2401,
|
| 646 |
-
"eval_samples_per_second": 10.569,
|
| 647 |
-
"eval_steps_per_second": 1.321,
|
| 648 |
-
"step": 1000
|
| 649 |
-
},
|
| 650 |
-
{
|
| 651 |
-
"epoch": 2.6,
|
| 652 |
-
"learning_rate": 0.00022777777777777778,
|
| 653 |
-
"loss": 0.8685,
|
| 654 |
-
"step": 1010
|
| 655 |
-
},
|
| 656 |
-
{
|
| 657 |
-
"epoch": 2.62,
|
| 658 |
-
"learning_rate": 0.00022698412698412697,
|
| 659 |
-
"loss": 0.7359,
|
| 660 |
-
"step": 1020
|
| 661 |
-
},
|
| 662 |
-
{
|
| 663 |
-
"epoch": 2.65,
|
| 664 |
-
"learning_rate": 0.00022619047619047616,
|
| 665 |
-
"loss": 0.7166,
|
| 666 |
-
"step": 1030
|
| 667 |
-
},
|
| 668 |
-
{
|
| 669 |
-
"epoch": 2.68,
|
| 670 |
-
"learning_rate": 0.00022539682539682537,
|
| 671 |
-
"loss": 0.9012,
|
| 672 |
-
"step": 1040
|
| 673 |
-
},
|
| 674 |
-
{
|
| 675 |
-
"epoch": 2.7,
|
| 676 |
-
"learning_rate": 0.0002246031746031746,
|
| 677 |
-
"loss": 0.9195,
|
| 678 |
-
"step": 1050
|
| 679 |
-
},
|
| 680 |
-
{
|
| 681 |
-
"epoch": 2.73,
|
| 682 |
-
"learning_rate": 0.0002238095238095238,
|
| 683 |
-
"loss": 0.8733,
|
| 684 |
-
"step": 1060
|
| 685 |
-
},
|
| 686 |
-
{
|
| 687 |
-
"epoch": 2.75,
|
| 688 |
-
"learning_rate": 0.000223015873015873,
|
| 689 |
-
"loss": 0.7488,
|
| 690 |
-
"step": 1070
|
| 691 |
-
},
|
| 692 |
-
{
|
| 693 |
-
"epoch": 2.78,
|
| 694 |
-
"learning_rate": 0.00022222222222222218,
|
| 695 |
-
"loss": 0.7223,
|
| 696 |
-
"step": 1080
|
| 697 |
-
},
|
| 698 |
-
{
|
| 699 |
-
"epoch": 2.8,
|
| 700 |
-
"learning_rate": 0.00022142857142857142,
|
| 701 |
-
"loss": 0.9034,
|
| 702 |
-
"step": 1090
|
| 703 |
-
},
|
| 704 |
-
{
|
| 705 |
-
"epoch": 2.83,
|
| 706 |
-
"learning_rate": 0.0002206349206349206,
|
| 707 |
-
"loss": 0.9174,
|
| 708 |
-
"step": 1100
|
| 709 |
-
},
|
| 710 |
-
{
|
| 711 |
-
"epoch": 2.86,
|
| 712 |
-
"learning_rate": 0.00021984126984126982,
|
| 713 |
-
"loss": 0.868,
|
| 714 |
-
"step": 1110
|
| 715 |
-
},
|
| 716 |
-
{
|
| 717 |
-
"epoch": 2.88,
|
| 718 |
-
"learning_rate": 0.000219047619047619,
|
| 719 |
-
"loss": 0.7394,
|
| 720 |
-
"step": 1120
|
| 721 |
-
},
|
| 722 |
-
{
|
| 723 |
-
"epoch": 2.91,
|
| 724 |
-
"learning_rate": 0.00021825396825396825,
|
| 725 |
-
"loss": 0.7198,
|
| 726 |
-
"step": 1130
|
| 727 |
-
},
|
| 728 |
-
{
|
| 729 |
-
"epoch": 2.93,
|
| 730 |
-
"learning_rate": 0.00021746031746031744,
|
| 731 |
-
"loss": 0.9085,
|
| 732 |
-
"step": 1140
|
| 733 |
-
},
|
| 734 |
-
{
|
| 735 |
-
"epoch": 2.96,
|
| 736 |
-
"learning_rate": 0.00021666666666666666,
|
| 737 |
-
"loss": 0.9148,
|
| 738 |
-
"step": 1150
|
| 739 |
-
},
|
| 740 |
-
{
|
| 741 |
-
"epoch": 2.98,
|
| 742 |
-
"learning_rate": 0.00021587301587301584,
|
| 743 |
-
"loss": 0.7775,
|
| 744 |
-
"step": 1160
|
| 745 |
-
},
|
| 746 |
-
{
|
| 747 |
-
"epoch": 3.01,
|
| 748 |
-
"learning_rate": 0.0002150793650793651,
|
| 749 |
-
"loss": 0.7366,
|
| 750 |
-
"step": 1170
|
| 751 |
-
},
|
| 752 |
-
{
|
| 753 |
-
"epoch": 3.04,
|
| 754 |
-
"learning_rate": 0.00021428571428571427,
|
| 755 |
-
"loss": 0.8865,
|
| 756 |
-
"step": 1180
|
| 757 |
-
},
|
| 758 |
-
{
|
| 759 |
-
"epoch": 3.06,
|
| 760 |
-
"learning_rate": 0.00021349206349206346,
|
| 761 |
-
"loss": 0.8989,
|
| 762 |
-
"step": 1190
|
| 763 |
-
},
|
| 764 |
-
{
|
| 765 |
-
"epoch": 3.09,
|
| 766 |
-
"learning_rate": 0.00021269841269841268,
|
| 767 |
-
"loss": 0.8376,
|
| 768 |
-
"step": 1200
|
| 769 |
-
},
|
| 770 |
-
{
|
| 771 |
-
"epoch": 3.09,
|
| 772 |
-
"eval_loss": 0.8965018391609192,
|
| 773 |
-
"eval_runtime": 189.2693,
|
| 774 |
-
"eval_samples_per_second": 10.567,
|
| 775 |
-
"eval_steps_per_second": 1.321,
|
| 776 |
-
"step": 1200
|
| 777 |
-
},
|
| 778 |
-
{
|
| 779 |
-
"epoch": 3.11,
|
| 780 |
-
"learning_rate": 0.0002119047619047619,
|
| 781 |
-
"loss": 0.7012,
|
| 782 |
-
"step": 1210
|
| 783 |
-
},
|
| 784 |
-
{
|
| 785 |
-
"epoch": 3.14,
|
| 786 |
-
"learning_rate": 0.0002111111111111111,
|
| 787 |
-
"loss": 0.7288,
|
| 788 |
-
"step": 1220
|
| 789 |
-
},
|
| 790 |
-
{
|
| 791 |
-
"epoch": 3.16,
|
| 792 |
-
"learning_rate": 0.0002103174603174603,
|
| 793 |
-
"loss": 0.8904,
|
| 794 |
-
"step": 1230
|
| 795 |
-
},
|
| 796 |
-
{
|
| 797 |
-
"epoch": 3.19,
|
| 798 |
-
"learning_rate": 0.00020952380952380948,
|
| 799 |
-
"loss": 0.9081,
|
| 800 |
-
"step": 1240
|
| 801 |
-
},
|
| 802 |
-
{
|
| 803 |
-
"epoch": 3.22,
|
| 804 |
-
"learning_rate": 0.00020873015873015873,
|
| 805 |
-
"loss": 0.8461,
|
| 806 |
-
"step": 1250
|
| 807 |
-
},
|
| 808 |
-
{
|
| 809 |
-
"epoch": 3.24,
|
| 810 |
-
"learning_rate": 0.00020793650793650791,
|
| 811 |
-
"loss": 0.6997,
|
| 812 |
-
"step": 1260
|
| 813 |
-
},
|
| 814 |
-
{
|
| 815 |
-
"epoch": 3.27,
|
| 816 |
-
"learning_rate": 0.00020714285714285713,
|
| 817 |
-
"loss": 0.7189,
|
| 818 |
-
"step": 1270
|
| 819 |
-
},
|
| 820 |
-
{
|
| 821 |
-
"epoch": 3.29,
|
| 822 |
-
"learning_rate": 0.00020634920634920632,
|
| 823 |
-
"loss": 0.8863,
|
| 824 |
-
"step": 1280
|
| 825 |
-
},
|
| 826 |
-
{
|
| 827 |
-
"epoch": 3.32,
|
| 828 |
-
"learning_rate": 0.00020555555555555556,
|
| 829 |
-
"loss": 0.906,
|
| 830 |
-
"step": 1290
|
| 831 |
-
},
|
| 832 |
-
{
|
| 833 |
-
"epoch": 3.34,
|
| 834 |
-
"learning_rate": 0.00020476190476190475,
|
| 835 |
-
"loss": 0.8287,
|
| 836 |
-
"step": 1300
|
| 837 |
-
},
|
| 838 |
-
{
|
| 839 |
-
"epoch": 3.37,
|
| 840 |
-
"learning_rate": 0.00020396825396825393,
|
| 841 |
-
"loss": 0.7015,
|
| 842 |
-
"step": 1310
|
| 843 |
-
},
|
| 844 |
-
{
|
| 845 |
-
"epoch": 3.4,
|
| 846 |
-
"learning_rate": 0.00020317460317460315,
|
| 847 |
-
"loss": 0.7325,
|
| 848 |
-
"step": 1320
|
| 849 |
-
},
|
| 850 |
-
{
|
| 851 |
-
"epoch": 3.42,
|
| 852 |
-
"learning_rate": 0.00020238095238095236,
|
| 853 |
-
"loss": 0.8878,
|
| 854 |
-
"step": 1330
|
| 855 |
-
},
|
| 856 |
-
{
|
| 857 |
-
"epoch": 3.45,
|
| 858 |
-
"learning_rate": 0.00020158730158730158,
|
| 859 |
-
"loss": 0.9057,
|
| 860 |
-
"step": 1340
|
| 861 |
-
},
|
| 862 |
-
{
|
| 863 |
-
"epoch": 3.47,
|
| 864 |
-
"learning_rate": 0.00020079365079365077,
|
| 865 |
-
"loss": 0.8399,
|
| 866 |
-
"step": 1350
|
| 867 |
-
},
|
| 868 |
-
{
|
| 869 |
-
"epoch": 3.5,
|
| 870 |
-
"learning_rate": 0.00019999999999999998,
|
| 871 |
-
"loss": 0.7073,
|
| 872 |
-
"step": 1360
|
| 873 |
-
},
|
| 874 |
-
{
|
| 875 |
-
"epoch": 3.52,
|
| 876 |
-
"learning_rate": 0.0001992063492063492,
|
| 877 |
-
"loss": 0.7281,
|
| 878 |
-
"step": 1370
|
| 879 |
-
},
|
| 880 |
-
{
|
| 881 |
-
"epoch": 3.55,
|
| 882 |
-
"learning_rate": 0.0001984126984126984,
|
| 883 |
-
"loss": 0.8829,
|
| 884 |
-
"step": 1380
|
| 885 |
-
},
|
| 886 |
-
{
|
| 887 |
-
"epoch": 3.58,
|
| 888 |
-
"learning_rate": 0.0001976190476190476,
|
| 889 |
-
"loss": 0.8923,
|
| 890 |
-
"step": 1390
|
| 891 |
-
},
|
| 892 |
-
{
|
| 893 |
-
"epoch": 3.6,
|
| 894 |
-
"learning_rate": 0.0001968253968253968,
|
| 895 |
-
"loss": 0.8389,
|
| 896 |
-
"step": 1400
|
| 897 |
-
},
|
| 898 |
-
{
|
| 899 |
-
"epoch": 3.6,
|
| 900 |
-
"eval_loss": 0.8923280239105225,
|
| 901 |
-
"eval_runtime": 189.1693,
|
| 902 |
-
"eval_samples_per_second": 10.573,
|
| 903 |
-
"eval_steps_per_second": 1.322,
|
| 904 |
-
"step": 1400
|
| 905 |
-
}
|
| 906 |
-
],
|
| 907 |
-
"max_steps": 3880,
|
| 908 |
-
"num_train_epochs": 10,
|
| 909 |
-
"total_flos": 1.627337272190042e+18,
|
| 910 |
-
"trial_name": null,
|
| 911 |
-
"trial_params": null
|
| 912 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-1400/training_args.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:3e9adb78996a536c4aa514741768e2b05cafc3e20ac4a0a0fe98e38b91109396
|
| 3 |
-
size 3899
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-3600/optimizer.pt
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:33b24cfa8e04448aada6bd0f35798f33e4037e4f5c5e4d990ab5168be1ac720c
|
| 3 |
-
size 134433093
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-3600/pytorch_model.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:58653c0888eea93a18cfef68476391e6cf3aaabd0a866d5bd1a63232af5da325
|
| 3 |
-
size 67201357
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-3600/rng_state.pth
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:efd17e8763b5b06bd50bbfccd0ef28c0faa97523b4dd293e16d443dfb1f74431
|
| 3 |
-
size 14575
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-3600/scaler.pt
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:7fa181fa360d46feed4180ea17c8b6a4a879a9b4231c2e91aff2be20be9076cc
|
| 3 |
-
size 557
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-3600/scheduler.pt
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:f0b82598b73ddc100e1ea38611348b8278822263257c909d620c6530bad49649
|
| 3 |
-
size 627
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-3600/trainer_state.json
DELETED
|
@@ -1,2320 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_metric": 0.8923280239105225,
|
| 3 |
-
"best_model_checkpoint": "./lora-alpaca-hc8/checkpoint-1400",
|
| 4 |
-
"epoch": 9.260450160771704,
|
| 5 |
-
"global_step": 3600,
|
| 6 |
-
"is_hyper_param_search": false,
|
| 7 |
-
"is_local_process_zero": true,
|
| 8 |
-
"is_world_process_zero": true,
|
| 9 |
-
"log_history": [
|
| 10 |
-
{
|
| 11 |
-
"epoch": 0.03,
|
| 12 |
-
"learning_rate": 2.9999999999999997e-05,
|
| 13 |
-
"loss": 1.3719,
|
| 14 |
-
"step": 10
|
| 15 |
-
},
|
| 16 |
-
{
|
| 17 |
-
"epoch": 0.05,
|
| 18 |
-
"learning_rate": 5.9999999999999995e-05,
|
| 19 |
-
"loss": 1.5358,
|
| 20 |
-
"step": 20
|
| 21 |
-
},
|
| 22 |
-
{
|
| 23 |
-
"epoch": 0.08,
|
| 24 |
-
"learning_rate": 8.999999999999999e-05,
|
| 25 |
-
"loss": 1.5633,
|
| 26 |
-
"step": 30
|
| 27 |
-
},
|
| 28 |
-
{
|
| 29 |
-
"epoch": 0.1,
|
| 30 |
-
"learning_rate": 0.00011999999999999999,
|
| 31 |
-
"loss": 1.2433,
|
| 32 |
-
"step": 40
|
| 33 |
-
},
|
| 34 |
-
{
|
| 35 |
-
"epoch": 0.13,
|
| 36 |
-
"learning_rate": 0.00015,
|
| 37 |
-
"loss": 0.9843,
|
| 38 |
-
"step": 50
|
| 39 |
-
},
|
| 40 |
-
{
|
| 41 |
-
"epoch": 0.15,
|
| 42 |
-
"learning_rate": 0.00017999999999999998,
|
| 43 |
-
"loss": 1.012,
|
| 44 |
-
"step": 60
|
| 45 |
-
},
|
| 46 |
-
{
|
| 47 |
-
"epoch": 0.18,
|
| 48 |
-
"learning_rate": 0.00020999999999999998,
|
| 49 |
-
"loss": 1.0392,
|
| 50 |
-
"step": 70
|
| 51 |
-
},
|
| 52 |
-
{
|
| 53 |
-
"epoch": 0.21,
|
| 54 |
-
"learning_rate": 0.00023999999999999998,
|
| 55 |
-
"loss": 1.0541,
|
| 56 |
-
"step": 80
|
| 57 |
-
},
|
| 58 |
-
{
|
| 59 |
-
"epoch": 0.23,
|
| 60 |
-
"learning_rate": 0.00027,
|
| 61 |
-
"loss": 0.9608,
|
| 62 |
-
"step": 90
|
| 63 |
-
},
|
| 64 |
-
{
|
| 65 |
-
"epoch": 0.26,
|
| 66 |
-
"learning_rate": 0.0003,
|
| 67 |
-
"loss": 0.8697,
|
| 68 |
-
"step": 100
|
| 69 |
-
},
|
| 70 |
-
{
|
| 71 |
-
"epoch": 0.28,
|
| 72 |
-
"learning_rate": 0.00029920634920634916,
|
| 73 |
-
"loss": 0.9836,
|
| 74 |
-
"step": 110
|
| 75 |
-
},
|
| 76 |
-
{
|
| 77 |
-
"epoch": 0.31,
|
| 78 |
-
"learning_rate": 0.00029841269841269835,
|
| 79 |
-
"loss": 1.003,
|
| 80 |
-
"step": 120
|
| 81 |
-
},
|
| 82 |
-
{
|
| 83 |
-
"epoch": 0.33,
|
| 84 |
-
"learning_rate": 0.0002976190476190476,
|
| 85 |
-
"loss": 1.0052,
|
| 86 |
-
"step": 130
|
| 87 |
-
},
|
| 88 |
-
{
|
| 89 |
-
"epoch": 0.36,
|
| 90 |
-
"learning_rate": 0.0002968253968253968,
|
| 91 |
-
"loss": 0.9267,
|
| 92 |
-
"step": 140
|
| 93 |
-
},
|
| 94 |
-
{
|
| 95 |
-
"epoch": 0.39,
|
| 96 |
-
"learning_rate": 0.000296031746031746,
|
| 97 |
-
"loss": 0.8318,
|
| 98 |
-
"step": 150
|
| 99 |
-
},
|
| 100 |
-
{
|
| 101 |
-
"epoch": 0.41,
|
| 102 |
-
"learning_rate": 0.0002952380952380952,
|
| 103 |
-
"loss": 0.9594,
|
| 104 |
-
"step": 160
|
| 105 |
-
},
|
| 106 |
-
{
|
| 107 |
-
"epoch": 0.44,
|
| 108 |
-
"learning_rate": 0.00029444444444444445,
|
| 109 |
-
"loss": 0.987,
|
| 110 |
-
"step": 170
|
| 111 |
-
},
|
| 112 |
-
{
|
| 113 |
-
"epoch": 0.46,
|
| 114 |
-
"learning_rate": 0.00029365079365079364,
|
| 115 |
-
"loss": 0.9646,
|
| 116 |
-
"step": 180
|
| 117 |
-
},
|
| 118 |
-
{
|
| 119 |
-
"epoch": 0.49,
|
| 120 |
-
"learning_rate": 0.00029285714285714283,
|
| 121 |
-
"loss": 0.8501,
|
| 122 |
-
"step": 190
|
| 123 |
-
},
|
| 124 |
-
{
|
| 125 |
-
"epoch": 0.51,
|
| 126 |
-
"learning_rate": 0.000292063492063492,
|
| 127 |
-
"loss": 0.7568,
|
| 128 |
-
"step": 200
|
| 129 |
-
},
|
| 130 |
-
{
|
| 131 |
-
"epoch": 0.51,
|
| 132 |
-
"eval_loss": 0.9958714246749878,
|
| 133 |
-
"eval_runtime": 189.2223,
|
| 134 |
-
"eval_samples_per_second": 10.57,
|
| 135 |
-
"eval_steps_per_second": 1.321,
|
| 136 |
-
"step": 200
|
| 137 |
-
},
|
| 138 |
-
{
|
| 139 |
-
"epoch": 0.54,
|
| 140 |
-
"learning_rate": 0.00029126984126984126,
|
| 141 |
-
"loss": 0.949,
|
| 142 |
-
"step": 210
|
| 143 |
-
},
|
| 144 |
-
{
|
| 145 |
-
"epoch": 0.57,
|
| 146 |
-
"learning_rate": 0.00029047619047619045,
|
| 147 |
-
"loss": 0.9581,
|
| 148 |
-
"step": 220
|
| 149 |
-
},
|
| 150 |
-
{
|
| 151 |
-
"epoch": 0.59,
|
| 152 |
-
"learning_rate": 0.00028968253968253963,
|
| 153 |
-
"loss": 0.9526,
|
| 154 |
-
"step": 230
|
| 155 |
-
},
|
| 156 |
-
{
|
| 157 |
-
"epoch": 0.62,
|
| 158 |
-
"learning_rate": 0.0002888888888888888,
|
| 159 |
-
"loss": 0.847,
|
| 160 |
-
"step": 240
|
| 161 |
-
},
|
| 162 |
-
{
|
| 163 |
-
"epoch": 0.64,
|
| 164 |
-
"learning_rate": 0.00028809523809523806,
|
| 165 |
-
"loss": 0.7414,
|
| 166 |
-
"step": 250
|
| 167 |
-
},
|
| 168 |
-
{
|
| 169 |
-
"epoch": 0.67,
|
| 170 |
-
"learning_rate": 0.00028730158730158725,
|
| 171 |
-
"loss": 0.9449,
|
| 172 |
-
"step": 260
|
| 173 |
-
},
|
| 174 |
-
{
|
| 175 |
-
"epoch": 0.69,
|
| 176 |
-
"learning_rate": 0.0002865079365079365,
|
| 177 |
-
"loss": 0.9607,
|
| 178 |
-
"step": 270
|
| 179 |
-
},
|
| 180 |
-
{
|
| 181 |
-
"epoch": 0.72,
|
| 182 |
-
"learning_rate": 0.0002857142857142857,
|
| 183 |
-
"loss": 0.9456,
|
| 184 |
-
"step": 280
|
| 185 |
-
},
|
| 186 |
-
{
|
| 187 |
-
"epoch": 0.75,
|
| 188 |
-
"learning_rate": 0.0002849206349206349,
|
| 189 |
-
"loss": 0.8346,
|
| 190 |
-
"step": 290
|
| 191 |
-
},
|
| 192 |
-
{
|
| 193 |
-
"epoch": 0.77,
|
| 194 |
-
"learning_rate": 0.0002841269841269841,
|
| 195 |
-
"loss": 0.7275,
|
| 196 |
-
"step": 300
|
| 197 |
-
},
|
| 198 |
-
{
|
| 199 |
-
"epoch": 0.8,
|
| 200 |
-
"learning_rate": 0.0002833333333333333,
|
| 201 |
-
"loss": 0.9337,
|
| 202 |
-
"step": 310
|
| 203 |
-
},
|
| 204 |
-
{
|
| 205 |
-
"epoch": 0.82,
|
| 206 |
-
"learning_rate": 0.0002825396825396825,
|
| 207 |
-
"loss": 0.9466,
|
| 208 |
-
"step": 320
|
| 209 |
-
},
|
| 210 |
-
{
|
| 211 |
-
"epoch": 0.85,
|
| 212 |
-
"learning_rate": 0.00028174603174603173,
|
| 213 |
-
"loss": 0.9386,
|
| 214 |
-
"step": 330
|
| 215 |
-
},
|
| 216 |
-
{
|
| 217 |
-
"epoch": 0.87,
|
| 218 |
-
"learning_rate": 0.0002809523809523809,
|
| 219 |
-
"loss": 0.8254,
|
| 220 |
-
"step": 340
|
| 221 |
-
},
|
| 222 |
-
{
|
| 223 |
-
"epoch": 0.9,
|
| 224 |
-
"learning_rate": 0.0002801587301587301,
|
| 225 |
-
"loss": 0.723,
|
| 226 |
-
"step": 350
|
| 227 |
-
},
|
| 228 |
-
{
|
| 229 |
-
"epoch": 0.93,
|
| 230 |
-
"learning_rate": 0.00027936507936507935,
|
| 231 |
-
"loss": 0.9274,
|
| 232 |
-
"step": 360
|
| 233 |
-
},
|
| 234 |
-
{
|
| 235 |
-
"epoch": 0.95,
|
| 236 |
-
"learning_rate": 0.00027857142857142854,
|
| 237 |
-
"loss": 0.9492,
|
| 238 |
-
"step": 370
|
| 239 |
-
},
|
| 240 |
-
{
|
| 241 |
-
"epoch": 0.98,
|
| 242 |
-
"learning_rate": 0.0002777777777777778,
|
| 243 |
-
"loss": 0.871,
|
| 244 |
-
"step": 380
|
| 245 |
-
},
|
| 246 |
-
{
|
| 247 |
-
"epoch": 1.0,
|
| 248 |
-
"learning_rate": 0.00027698412698412697,
|
| 249 |
-
"loss": 0.7533,
|
| 250 |
-
"step": 390
|
| 251 |
-
},
|
| 252 |
-
{
|
| 253 |
-
"epoch": 1.03,
|
| 254 |
-
"learning_rate": 0.00027619047619047615,
|
| 255 |
-
"loss": 0.9284,
|
| 256 |
-
"step": 400
|
| 257 |
-
},
|
| 258 |
-
{
|
| 259 |
-
"epoch": 1.03,
|
| 260 |
-
"eval_loss": 0.9276881217956543,
|
| 261 |
-
"eval_runtime": 189.2181,
|
| 262 |
-
"eval_samples_per_second": 10.57,
|
| 263 |
-
"eval_steps_per_second": 1.321,
|
| 264 |
-
"step": 400
|
| 265 |
-
},
|
| 266 |
-
{
|
| 267 |
-
"epoch": 1.05,
|
| 268 |
-
"learning_rate": 0.0002753968253968254,
|
| 269 |
-
"loss": 0.9402,
|
| 270 |
-
"step": 410
|
| 271 |
-
},
|
| 272 |
-
{
|
| 273 |
-
"epoch": 1.08,
|
| 274 |
-
"learning_rate": 0.0002746031746031746,
|
| 275 |
-
"loss": 0.9188,
|
| 276 |
-
"step": 420
|
| 277 |
-
},
|
| 278 |
-
{
|
| 279 |
-
"epoch": 1.11,
|
| 280 |
-
"learning_rate": 0.00027380952380952377,
|
| 281 |
-
"loss": 0.7976,
|
| 282 |
-
"step": 430
|
| 283 |
-
},
|
| 284 |
-
{
|
| 285 |
-
"epoch": 1.13,
|
| 286 |
-
"learning_rate": 0.00027301587301587296,
|
| 287 |
-
"loss": 0.7172,
|
| 288 |
-
"step": 440
|
| 289 |
-
},
|
| 290 |
-
{
|
| 291 |
-
"epoch": 1.16,
|
| 292 |
-
"learning_rate": 0.0002722222222222222,
|
| 293 |
-
"loss": 0.9195,
|
| 294 |
-
"step": 450
|
| 295 |
-
},
|
| 296 |
-
{
|
| 297 |
-
"epoch": 1.18,
|
| 298 |
-
"learning_rate": 0.0002714285714285714,
|
| 299 |
-
"loss": 0.9426,
|
| 300 |
-
"step": 460
|
| 301 |
-
},
|
| 302 |
-
{
|
| 303 |
-
"epoch": 1.21,
|
| 304 |
-
"learning_rate": 0.00027063492063492063,
|
| 305 |
-
"loss": 0.9034,
|
| 306 |
-
"step": 470
|
| 307 |
-
},
|
| 308 |
-
{
|
| 309 |
-
"epoch": 1.23,
|
| 310 |
-
"learning_rate": 0.0002698412698412698,
|
| 311 |
-
"loss": 0.788,
|
| 312 |
-
"step": 480
|
| 313 |
-
},
|
| 314 |
-
{
|
| 315 |
-
"epoch": 1.26,
|
| 316 |
-
"learning_rate": 0.000269047619047619,
|
| 317 |
-
"loss": 0.7213,
|
| 318 |
-
"step": 490
|
| 319 |
-
},
|
| 320 |
-
{
|
| 321 |
-
"epoch": 1.29,
|
| 322 |
-
"learning_rate": 0.00026825396825396825,
|
| 323 |
-
"loss": 0.9149,
|
| 324 |
-
"step": 500
|
| 325 |
-
},
|
| 326 |
-
{
|
| 327 |
-
"epoch": 1.31,
|
| 328 |
-
"learning_rate": 0.00026746031746031744,
|
| 329 |
-
"loss": 0.9386,
|
| 330 |
-
"step": 510
|
| 331 |
-
},
|
| 332 |
-
{
|
| 333 |
-
"epoch": 1.34,
|
| 334 |
-
"learning_rate": 0.0002666666666666666,
|
| 335 |
-
"loss": 0.9099,
|
| 336 |
-
"step": 520
|
| 337 |
-
},
|
| 338 |
-
{
|
| 339 |
-
"epoch": 1.36,
|
| 340 |
-
"learning_rate": 0.00026587301587301587,
|
| 341 |
-
"loss": 0.7802,
|
| 342 |
-
"step": 530
|
| 343 |
-
},
|
| 344 |
-
{
|
| 345 |
-
"epoch": 1.39,
|
| 346 |
-
"learning_rate": 0.00026507936507936506,
|
| 347 |
-
"loss": 0.7246,
|
| 348 |
-
"step": 540
|
| 349 |
-
},
|
| 350 |
-
{
|
| 351 |
-
"epoch": 1.41,
|
| 352 |
-
"learning_rate": 0.00026428571428571424,
|
| 353 |
-
"loss": 0.9213,
|
| 354 |
-
"step": 550
|
| 355 |
-
},
|
| 356 |
-
{
|
| 357 |
-
"epoch": 1.44,
|
| 358 |
-
"learning_rate": 0.00026349206349206343,
|
| 359 |
-
"loss": 0.9348,
|
| 360 |
-
"step": 560
|
| 361 |
-
},
|
| 362 |
-
{
|
| 363 |
-
"epoch": 1.47,
|
| 364 |
-
"learning_rate": 0.0002626984126984127,
|
| 365 |
-
"loss": 0.8995,
|
| 366 |
-
"step": 570
|
| 367 |
-
},
|
| 368 |
-
{
|
| 369 |
-
"epoch": 1.49,
|
| 370 |
-
"learning_rate": 0.00026190476190476186,
|
| 371 |
-
"loss": 0.7866,
|
| 372 |
-
"step": 580
|
| 373 |
-
},
|
| 374 |
-
{
|
| 375 |
-
"epoch": 1.52,
|
| 376 |
-
"learning_rate": 0.0002611111111111111,
|
| 377 |
-
"loss": 0.7257,
|
| 378 |
-
"step": 590
|
| 379 |
-
},
|
| 380 |
-
{
|
| 381 |
-
"epoch": 1.54,
|
| 382 |
-
"learning_rate": 0.0002603174603174603,
|
| 383 |
-
"loss": 0.918,
|
| 384 |
-
"step": 600
|
| 385 |
-
},
|
| 386 |
-
{
|
| 387 |
-
"epoch": 1.54,
|
| 388 |
-
"eval_loss": 0.9236659407615662,
|
| 389 |
-
"eval_runtime": 189.1236,
|
| 390 |
-
"eval_samples_per_second": 10.575,
|
| 391 |
-
"eval_steps_per_second": 1.322,
|
| 392 |
-
"step": 600
|
| 393 |
-
},
|
| 394 |
-
{
|
| 395 |
-
"epoch": 1.57,
|
| 396 |
-
"learning_rate": 0.00025952380952380953,
|
| 397 |
-
"loss": 0.933,
|
| 398 |
-
"step": 610
|
| 399 |
-
},
|
| 400 |
-
{
|
| 401 |
-
"epoch": 1.59,
|
| 402 |
-
"learning_rate": 0.0002587301587301587,
|
| 403 |
-
"loss": 0.9085,
|
| 404 |
-
"step": 620
|
| 405 |
-
},
|
| 406 |
-
{
|
| 407 |
-
"epoch": 1.62,
|
| 408 |
-
"learning_rate": 0.0002579365079365079,
|
| 409 |
-
"loss": 0.7928,
|
| 410 |
-
"step": 630
|
| 411 |
-
},
|
| 412 |
-
{
|
| 413 |
-
"epoch": 1.65,
|
| 414 |
-
"learning_rate": 0.0002571428571428571,
|
| 415 |
-
"loss": 0.7162,
|
| 416 |
-
"step": 640
|
| 417 |
-
},
|
| 418 |
-
{
|
| 419 |
-
"epoch": 1.67,
|
| 420 |
-
"learning_rate": 0.00025634920634920634,
|
| 421 |
-
"loss": 0.9076,
|
| 422 |
-
"step": 650
|
| 423 |
-
},
|
| 424 |
-
{
|
| 425 |
-
"epoch": 1.7,
|
| 426 |
-
"learning_rate": 0.00025555555555555553,
|
| 427 |
-
"loss": 0.9345,
|
| 428 |
-
"step": 660
|
| 429 |
-
},
|
| 430 |
-
{
|
| 431 |
-
"epoch": 1.72,
|
| 432 |
-
"learning_rate": 0.0002547619047619047,
|
| 433 |
-
"loss": 0.9107,
|
| 434 |
-
"step": 670
|
| 435 |
-
},
|
| 436 |
-
{
|
| 437 |
-
"epoch": 1.75,
|
| 438 |
-
"learning_rate": 0.00025396825396825396,
|
| 439 |
-
"loss": 0.7721,
|
| 440 |
-
"step": 680
|
| 441 |
-
},
|
| 442 |
-
{
|
| 443 |
-
"epoch": 1.77,
|
| 444 |
-
"learning_rate": 0.00025317460317460315,
|
| 445 |
-
"loss": 0.7112,
|
| 446 |
-
"step": 690
|
| 447 |
-
},
|
| 448 |
-
{
|
| 449 |
-
"epoch": 1.8,
|
| 450 |
-
"learning_rate": 0.0002523809523809524,
|
| 451 |
-
"loss": 0.9118,
|
| 452 |
-
"step": 700
|
| 453 |
-
},
|
| 454 |
-
{
|
| 455 |
-
"epoch": 1.83,
|
| 456 |
-
"learning_rate": 0.0002515873015873016,
|
| 457 |
-
"loss": 0.9205,
|
| 458 |
-
"step": 710
|
| 459 |
-
},
|
| 460 |
-
{
|
| 461 |
-
"epoch": 1.85,
|
| 462 |
-
"learning_rate": 0.00025079365079365076,
|
| 463 |
-
"loss": 0.9004,
|
| 464 |
-
"step": 720
|
| 465 |
-
},
|
| 466 |
-
{
|
| 467 |
-
"epoch": 1.88,
|
| 468 |
-
"learning_rate": 0.00025,
|
| 469 |
-
"loss": 0.7741,
|
| 470 |
-
"step": 730
|
| 471 |
-
},
|
| 472 |
-
{
|
| 473 |
-
"epoch": 1.9,
|
| 474 |
-
"learning_rate": 0.0002492063492063492,
|
| 475 |
-
"loss": 0.7186,
|
| 476 |
-
"step": 740
|
| 477 |
-
},
|
| 478 |
-
{
|
| 479 |
-
"epoch": 1.93,
|
| 480 |
-
"learning_rate": 0.0002484126984126984,
|
| 481 |
-
"loss": 0.9002,
|
| 482 |
-
"step": 750
|
| 483 |
-
},
|
| 484 |
-
{
|
| 485 |
-
"epoch": 1.95,
|
| 486 |
-
"learning_rate": 0.00024761904761904757,
|
| 487 |
-
"loss": 0.9066,
|
| 488 |
-
"step": 760
|
| 489 |
-
},
|
| 490 |
-
{
|
| 491 |
-
"epoch": 1.98,
|
| 492 |
-
"learning_rate": 0.0002468253968253968,
|
| 493 |
-
"loss": 0.8127,
|
| 494 |
-
"step": 770
|
| 495 |
-
},
|
| 496 |
-
{
|
| 497 |
-
"epoch": 2.01,
|
| 498 |
-
"learning_rate": 0.000246031746031746,
|
| 499 |
-
"loss": 0.7305,
|
| 500 |
-
"step": 780
|
| 501 |
-
},
|
| 502 |
-
{
|
| 503 |
-
"epoch": 2.03,
|
| 504 |
-
"learning_rate": 0.0002452380952380952,
|
| 505 |
-
"loss": 0.8921,
|
| 506 |
-
"step": 790
|
| 507 |
-
},
|
| 508 |
-
{
|
| 509 |
-
"epoch": 2.06,
|
| 510 |
-
"learning_rate": 0.00024444444444444443,
|
| 511 |
-
"loss": 0.9178,
|
| 512 |
-
"step": 800
|
| 513 |
-
},
|
| 514 |
-
{
|
| 515 |
-
"epoch": 2.06,
|
| 516 |
-
"eval_loss": 0.901778519153595,
|
| 517 |
-
"eval_runtime": 189.2038,
|
| 518 |
-
"eval_samples_per_second": 10.571,
|
| 519 |
-
"eval_steps_per_second": 1.321,
|
| 520 |
-
"step": 800
|
| 521 |
-
},
|
| 522 |
-
{
|
| 523 |
-
"epoch": 2.08,
|
| 524 |
-
"learning_rate": 0.00024365079365079364,
|
| 525 |
-
"loss": 0.8823,
|
| 526 |
-
"step": 810
|
| 527 |
-
},
|
| 528 |
-
{
|
| 529 |
-
"epoch": 2.11,
|
| 530 |
-
"learning_rate": 0.00024285714285714283,
|
| 531 |
-
"loss": 0.7521,
|
| 532 |
-
"step": 820
|
| 533 |
-
},
|
| 534 |
-
{
|
| 535 |
-
"epoch": 2.14,
|
| 536 |
-
"learning_rate": 0.00024206349206349205,
|
| 537 |
-
"loss": 0.717,
|
| 538 |
-
"step": 830
|
| 539 |
-
},
|
| 540 |
-
{
|
| 541 |
-
"epoch": 2.16,
|
| 542 |
-
"learning_rate": 0.00024126984126984123,
|
| 543 |
-
"loss": 0.9045,
|
| 544 |
-
"step": 840
|
| 545 |
-
},
|
| 546 |
-
{
|
| 547 |
-
"epoch": 2.19,
|
| 548 |
-
"learning_rate": 0.00024047619047619048,
|
| 549 |
-
"loss": 0.9146,
|
| 550 |
-
"step": 850
|
| 551 |
-
},
|
| 552 |
-
{
|
| 553 |
-
"epoch": 2.21,
|
| 554 |
-
"learning_rate": 0.00023968253968253966,
|
| 555 |
-
"loss": 0.8678,
|
| 556 |
-
"step": 860
|
| 557 |
-
},
|
| 558 |
-
{
|
| 559 |
-
"epoch": 2.24,
|
| 560 |
-
"learning_rate": 0.00023888888888888885,
|
| 561 |
-
"loss": 0.745,
|
| 562 |
-
"step": 870
|
| 563 |
-
},
|
| 564 |
-
{
|
| 565 |
-
"epoch": 2.26,
|
| 566 |
-
"learning_rate": 0.00023809523809523807,
|
| 567 |
-
"loss": 0.7226,
|
| 568 |
-
"step": 880
|
| 569 |
-
},
|
| 570 |
-
{
|
| 571 |
-
"epoch": 2.29,
|
| 572 |
-
"learning_rate": 0.00023730158730158728,
|
| 573 |
-
"loss": 0.9002,
|
| 574 |
-
"step": 890
|
| 575 |
-
},
|
| 576 |
-
{
|
| 577 |
-
"epoch": 2.32,
|
| 578 |
-
"learning_rate": 0.0002365079365079365,
|
| 579 |
-
"loss": 0.9035,
|
| 580 |
-
"step": 900
|
| 581 |
-
},
|
| 582 |
-
{
|
| 583 |
-
"epoch": 2.34,
|
| 584 |
-
"learning_rate": 0.00023571428571428569,
|
| 585 |
-
"loss": 0.878,
|
| 586 |
-
"step": 910
|
| 587 |
-
},
|
| 588 |
-
{
|
| 589 |
-
"epoch": 2.37,
|
| 590 |
-
"learning_rate": 0.00023492063492063487,
|
| 591 |
-
"loss": 0.7397,
|
| 592 |
-
"step": 920
|
| 593 |
-
},
|
| 594 |
-
{
|
| 595 |
-
"epoch": 2.39,
|
| 596 |
-
"learning_rate": 0.00023412698412698412,
|
| 597 |
-
"loss": 0.7251,
|
| 598 |
-
"step": 930
|
| 599 |
-
},
|
| 600 |
-
{
|
| 601 |
-
"epoch": 2.42,
|
| 602 |
-
"learning_rate": 0.0002333333333333333,
|
| 603 |
-
"loss": 0.9014,
|
| 604 |
-
"step": 940
|
| 605 |
-
},
|
| 606 |
-
{
|
| 607 |
-
"epoch": 2.44,
|
| 608 |
-
"learning_rate": 0.00023253968253968252,
|
| 609 |
-
"loss": 0.9158,
|
| 610 |
-
"step": 950
|
| 611 |
-
},
|
| 612 |
-
{
|
| 613 |
-
"epoch": 2.47,
|
| 614 |
-
"learning_rate": 0.0002317460317460317,
|
| 615 |
-
"loss": 0.8596,
|
| 616 |
-
"step": 960
|
| 617 |
-
},
|
| 618 |
-
{
|
| 619 |
-
"epoch": 2.5,
|
| 620 |
-
"learning_rate": 0.00023095238095238095,
|
| 621 |
-
"loss": 0.7312,
|
| 622 |
-
"step": 970
|
| 623 |
-
},
|
| 624 |
-
{
|
| 625 |
-
"epoch": 2.52,
|
| 626 |
-
"learning_rate": 0.00023015873015873014,
|
| 627 |
-
"loss": 0.7271,
|
| 628 |
-
"step": 980
|
| 629 |
-
},
|
| 630 |
-
{
|
| 631 |
-
"epoch": 2.55,
|
| 632 |
-
"learning_rate": 0.00022936507936507935,
|
| 633 |
-
"loss": 0.9007,
|
| 634 |
-
"step": 990
|
| 635 |
-
},
|
| 636 |
-
{
|
| 637 |
-
"epoch": 2.57,
|
| 638 |
-
"learning_rate": 0.00022857142857142854,
|
| 639 |
-
"loss": 0.9186,
|
| 640 |
-
"step": 1000
|
| 641 |
-
},
|
| 642 |
-
{
|
| 643 |
-
"epoch": 2.57,
|
| 644 |
-
"eval_loss": 0.8995742201805115,
|
| 645 |
-
"eval_runtime": 189.2401,
|
| 646 |
-
"eval_samples_per_second": 10.569,
|
| 647 |
-
"eval_steps_per_second": 1.321,
|
| 648 |
-
"step": 1000
|
| 649 |
-
},
|
| 650 |
-
{
|
| 651 |
-
"epoch": 2.6,
|
| 652 |
-
"learning_rate": 0.00022777777777777778,
|
| 653 |
-
"loss": 0.8685,
|
| 654 |
-
"step": 1010
|
| 655 |
-
},
|
| 656 |
-
{
|
| 657 |
-
"epoch": 2.62,
|
| 658 |
-
"learning_rate": 0.00022698412698412697,
|
| 659 |
-
"loss": 0.7359,
|
| 660 |
-
"step": 1020
|
| 661 |
-
},
|
| 662 |
-
{
|
| 663 |
-
"epoch": 2.65,
|
| 664 |
-
"learning_rate": 0.00022619047619047616,
|
| 665 |
-
"loss": 0.7166,
|
| 666 |
-
"step": 1030
|
| 667 |
-
},
|
| 668 |
-
{
|
| 669 |
-
"epoch": 2.68,
|
| 670 |
-
"learning_rate": 0.00022539682539682537,
|
| 671 |
-
"loss": 0.9012,
|
| 672 |
-
"step": 1040
|
| 673 |
-
},
|
| 674 |
-
{
|
| 675 |
-
"epoch": 2.7,
|
| 676 |
-
"learning_rate": 0.0002246031746031746,
|
| 677 |
-
"loss": 0.9195,
|
| 678 |
-
"step": 1050
|
| 679 |
-
},
|
| 680 |
-
{
|
| 681 |
-
"epoch": 2.73,
|
| 682 |
-
"learning_rate": 0.0002238095238095238,
|
| 683 |
-
"loss": 0.8733,
|
| 684 |
-
"step": 1060
|
| 685 |
-
},
|
| 686 |
-
{
|
| 687 |
-
"epoch": 2.75,
|
| 688 |
-
"learning_rate": 0.000223015873015873,
|
| 689 |
-
"loss": 0.7488,
|
| 690 |
-
"step": 1070
|
| 691 |
-
},
|
| 692 |
-
{
|
| 693 |
-
"epoch": 2.78,
|
| 694 |
-
"learning_rate": 0.00022222222222222218,
|
| 695 |
-
"loss": 0.7223,
|
| 696 |
-
"step": 1080
|
| 697 |
-
},
|
| 698 |
-
{
|
| 699 |
-
"epoch": 2.8,
|
| 700 |
-
"learning_rate": 0.00022142857142857142,
|
| 701 |
-
"loss": 0.9034,
|
| 702 |
-
"step": 1090
|
| 703 |
-
},
|
| 704 |
-
{
|
| 705 |
-
"epoch": 2.83,
|
| 706 |
-
"learning_rate": 0.0002206349206349206,
|
| 707 |
-
"loss": 0.9174,
|
| 708 |
-
"step": 1100
|
| 709 |
-
},
|
| 710 |
-
{
|
| 711 |
-
"epoch": 2.86,
|
| 712 |
-
"learning_rate": 0.00021984126984126982,
|
| 713 |
-
"loss": 0.868,
|
| 714 |
-
"step": 1110
|
| 715 |
-
},
|
| 716 |
-
{
|
| 717 |
-
"epoch": 2.88,
|
| 718 |
-
"learning_rate": 0.000219047619047619,
|
| 719 |
-
"loss": 0.7394,
|
| 720 |
-
"step": 1120
|
| 721 |
-
},
|
| 722 |
-
{
|
| 723 |
-
"epoch": 2.91,
|
| 724 |
-
"learning_rate": 0.00021825396825396825,
|
| 725 |
-
"loss": 0.7198,
|
| 726 |
-
"step": 1130
|
| 727 |
-
},
|
| 728 |
-
{
|
| 729 |
-
"epoch": 2.93,
|
| 730 |
-
"learning_rate": 0.00021746031746031744,
|
| 731 |
-
"loss": 0.9085,
|
| 732 |
-
"step": 1140
|
| 733 |
-
},
|
| 734 |
-
{
|
| 735 |
-
"epoch": 2.96,
|
| 736 |
-
"learning_rate": 0.00021666666666666666,
|
| 737 |
-
"loss": 0.9148,
|
| 738 |
-
"step": 1150
|
| 739 |
-
},
|
| 740 |
-
{
|
| 741 |
-
"epoch": 2.98,
|
| 742 |
-
"learning_rate": 0.00021587301587301584,
|
| 743 |
-
"loss": 0.7775,
|
| 744 |
-
"step": 1160
|
| 745 |
-
},
|
| 746 |
-
{
|
| 747 |
-
"epoch": 3.01,
|
| 748 |
-
"learning_rate": 0.0002150793650793651,
|
| 749 |
-
"loss": 0.7366,
|
| 750 |
-
"step": 1170
|
| 751 |
-
},
|
| 752 |
-
{
|
| 753 |
-
"epoch": 3.04,
|
| 754 |
-
"learning_rate": 0.00021428571428571427,
|
| 755 |
-
"loss": 0.8865,
|
| 756 |
-
"step": 1180
|
| 757 |
-
},
|
| 758 |
-
{
|
| 759 |
-
"epoch": 3.06,
|
| 760 |
-
"learning_rate": 0.00021349206349206346,
|
| 761 |
-
"loss": 0.8989,
|
| 762 |
-
"step": 1190
|
| 763 |
-
},
|
| 764 |
-
{
|
| 765 |
-
"epoch": 3.09,
|
| 766 |
-
"learning_rate": 0.00021269841269841268,
|
| 767 |
-
"loss": 0.8376,
|
| 768 |
-
"step": 1200
|
| 769 |
-
},
|
| 770 |
-
{
|
| 771 |
-
"epoch": 3.09,
|
| 772 |
-
"eval_loss": 0.8965018391609192,
|
| 773 |
-
"eval_runtime": 189.2693,
|
| 774 |
-
"eval_samples_per_second": 10.567,
|
| 775 |
-
"eval_steps_per_second": 1.321,
|
| 776 |
-
"step": 1200
|
| 777 |
-
},
|
| 778 |
-
{
|
| 779 |
-
"epoch": 3.11,
|
| 780 |
-
"learning_rate": 0.0002119047619047619,
|
| 781 |
-
"loss": 0.7012,
|
| 782 |
-
"step": 1210
|
| 783 |
-
},
|
| 784 |
-
{
|
| 785 |
-
"epoch": 3.14,
|
| 786 |
-
"learning_rate": 0.0002111111111111111,
|
| 787 |
-
"loss": 0.7288,
|
| 788 |
-
"step": 1220
|
| 789 |
-
},
|
| 790 |
-
{
|
| 791 |
-
"epoch": 3.16,
|
| 792 |
-
"learning_rate": 0.0002103174603174603,
|
| 793 |
-
"loss": 0.8904,
|
| 794 |
-
"step": 1230
|
| 795 |
-
},
|
| 796 |
-
{
|
| 797 |
-
"epoch": 3.19,
|
| 798 |
-
"learning_rate": 0.00020952380952380948,
|
| 799 |
-
"loss": 0.9081,
|
| 800 |
-
"step": 1240
|
| 801 |
-
},
|
| 802 |
-
{
|
| 803 |
-
"epoch": 3.22,
|
| 804 |
-
"learning_rate": 0.00020873015873015873,
|
| 805 |
-
"loss": 0.8461,
|
| 806 |
-
"step": 1250
|
| 807 |
-
},
|
| 808 |
-
{
|
| 809 |
-
"epoch": 3.24,
|
| 810 |
-
"learning_rate": 0.00020793650793650791,
|
| 811 |
-
"loss": 0.6997,
|
| 812 |
-
"step": 1260
|
| 813 |
-
},
|
| 814 |
-
{
|
| 815 |
-
"epoch": 3.27,
|
| 816 |
-
"learning_rate": 0.00020714285714285713,
|
| 817 |
-
"loss": 0.7189,
|
| 818 |
-
"step": 1270
|
| 819 |
-
},
|
| 820 |
-
{
|
| 821 |
-
"epoch": 3.29,
|
| 822 |
-
"learning_rate": 0.00020634920634920632,
|
| 823 |
-
"loss": 0.8863,
|
| 824 |
-
"step": 1280
|
| 825 |
-
},
|
| 826 |
-
{
|
| 827 |
-
"epoch": 3.32,
|
| 828 |
-
"learning_rate": 0.00020555555555555556,
|
| 829 |
-
"loss": 0.906,
|
| 830 |
-
"step": 1290
|
| 831 |
-
},
|
| 832 |
-
{
|
| 833 |
-
"epoch": 3.34,
|
| 834 |
-
"learning_rate": 0.00020476190476190475,
|
| 835 |
-
"loss": 0.8287,
|
| 836 |
-
"step": 1300
|
| 837 |
-
},
|
| 838 |
-
{
|
| 839 |
-
"epoch": 3.37,
|
| 840 |
-
"learning_rate": 0.00020396825396825393,
|
| 841 |
-
"loss": 0.7015,
|
| 842 |
-
"step": 1310
|
| 843 |
-
},
|
| 844 |
-
{
|
| 845 |
-
"epoch": 3.4,
|
| 846 |
-
"learning_rate": 0.00020317460317460315,
|
| 847 |
-
"loss": 0.7325,
|
| 848 |
-
"step": 1320
|
| 849 |
-
},
|
| 850 |
-
{
|
| 851 |
-
"epoch": 3.42,
|
| 852 |
-
"learning_rate": 0.00020238095238095236,
|
| 853 |
-
"loss": 0.8878,
|
| 854 |
-
"step": 1330
|
| 855 |
-
},
|
| 856 |
-
{
|
| 857 |
-
"epoch": 3.45,
|
| 858 |
-
"learning_rate": 0.00020158730158730158,
|
| 859 |
-
"loss": 0.9057,
|
| 860 |
-
"step": 1340
|
| 861 |
-
},
|
| 862 |
-
{
|
| 863 |
-
"epoch": 3.47,
|
| 864 |
-
"learning_rate": 0.00020079365079365077,
|
| 865 |
-
"loss": 0.8399,
|
| 866 |
-
"step": 1350
|
| 867 |
-
},
|
| 868 |
-
{
|
| 869 |
-
"epoch": 3.5,
|
| 870 |
-
"learning_rate": 0.00019999999999999998,
|
| 871 |
-
"loss": 0.7073,
|
| 872 |
-
"step": 1360
|
| 873 |
-
},
|
| 874 |
-
{
|
| 875 |
-
"epoch": 3.52,
|
| 876 |
-
"learning_rate": 0.0001992063492063492,
|
| 877 |
-
"loss": 0.7281,
|
| 878 |
-
"step": 1370
|
| 879 |
-
},
|
| 880 |
-
{
|
| 881 |
-
"epoch": 3.55,
|
| 882 |
-
"learning_rate": 0.0001984126984126984,
|
| 883 |
-
"loss": 0.8829,
|
| 884 |
-
"step": 1380
|
| 885 |
-
},
|
| 886 |
-
{
|
| 887 |
-
"epoch": 3.58,
|
| 888 |
-
"learning_rate": 0.0001976190476190476,
|
| 889 |
-
"loss": 0.8923,
|
| 890 |
-
"step": 1390
|
| 891 |
-
},
|
| 892 |
-
{
|
| 893 |
-
"epoch": 3.6,
|
| 894 |
-
"learning_rate": 0.0001968253968253968,
|
| 895 |
-
"loss": 0.8389,
|
| 896 |
-
"step": 1400
|
| 897 |
-
},
|
| 898 |
-
{
|
| 899 |
-
"epoch": 3.6,
|
| 900 |
-
"eval_loss": 0.8923280239105225,
|
| 901 |
-
"eval_runtime": 189.1693,
|
| 902 |
-
"eval_samples_per_second": 10.573,
|
| 903 |
-
"eval_steps_per_second": 1.322,
|
| 904 |
-
"step": 1400
|
| 905 |
-
},
|
| 906 |
-
{
|
| 907 |
-
"epoch": 3.63,
|
| 908 |
-
"learning_rate": 0.00019603174603174603,
|
| 909 |
-
"loss": 0.7148,
|
| 910 |
-
"step": 1410
|
| 911 |
-
},
|
| 912 |
-
{
|
| 913 |
-
"epoch": 3.65,
|
| 914 |
-
"learning_rate": 0.00019523809523809522,
|
| 915 |
-
"loss": 0.7331,
|
| 916 |
-
"step": 1420
|
| 917 |
-
},
|
| 918 |
-
{
|
| 919 |
-
"epoch": 3.68,
|
| 920 |
-
"learning_rate": 0.00019444444444444443,
|
| 921 |
-
"loss": 0.8944,
|
| 922 |
-
"step": 1430
|
| 923 |
-
},
|
| 924 |
-
{
|
| 925 |
-
"epoch": 3.7,
|
| 926 |
-
"learning_rate": 0.00019365079365079362,
|
| 927 |
-
"loss": 0.9001,
|
| 928 |
-
"step": 1440
|
| 929 |
-
},
|
| 930 |
-
{
|
| 931 |
-
"epoch": 3.73,
|
| 932 |
-
"learning_rate": 0.00019285714285714286,
|
| 933 |
-
"loss": 0.8397,
|
| 934 |
-
"step": 1450
|
| 935 |
-
},
|
| 936 |
-
{
|
| 937 |
-
"epoch": 3.76,
|
| 938 |
-
"learning_rate": 0.00019206349206349205,
|
| 939 |
-
"loss": 0.6953,
|
| 940 |
-
"step": 1460
|
| 941 |
-
},
|
| 942 |
-
{
|
| 943 |
-
"epoch": 3.78,
|
| 944 |
-
"learning_rate": 0.00019126984126984124,
|
| 945 |
-
"loss": 0.7229,
|
| 946 |
-
"step": 1470
|
| 947 |
-
},
|
| 948 |
-
{
|
| 949 |
-
"epoch": 3.81,
|
| 950 |
-
"learning_rate": 0.00019047619047619045,
|
| 951 |
-
"loss": 0.8833,
|
| 952 |
-
"step": 1480
|
| 953 |
-
},
|
| 954 |
-
{
|
| 955 |
-
"epoch": 3.83,
|
| 956 |
-
"learning_rate": 0.00018968253968253967,
|
| 957 |
-
"loss": 0.9028,
|
| 958 |
-
"step": 1490
|
| 959 |
-
},
|
| 960 |
-
{
|
| 961 |
-
"epoch": 3.86,
|
| 962 |
-
"learning_rate": 0.00018888888888888888,
|
| 963 |
-
"loss": 0.8412,
|
| 964 |
-
"step": 1500
|
| 965 |
-
},
|
| 966 |
-
{
|
| 967 |
-
"epoch": 3.88,
|
| 968 |
-
"learning_rate": 0.00018809523809523807,
|
| 969 |
-
"loss": 0.7024,
|
| 970 |
-
"step": 1510
|
| 971 |
-
},
|
| 972 |
-
{
|
| 973 |
-
"epoch": 3.91,
|
| 974 |
-
"learning_rate": 0.0001873015873015873,
|
| 975 |
-
"loss": 0.7262,
|
| 976 |
-
"step": 1520
|
| 977 |
-
},
|
| 978 |
-
{
|
| 979 |
-
"epoch": 3.94,
|
| 980 |
-
"learning_rate": 0.0001865079365079365,
|
| 981 |
-
"loss": 0.8926,
|
| 982 |
-
"step": 1530
|
| 983 |
-
},
|
| 984 |
-
{
|
| 985 |
-
"epoch": 3.96,
|
| 986 |
-
"learning_rate": 0.00018571428571428572,
|
| 987 |
-
"loss": 0.8703,
|
| 988 |
-
"step": 1540
|
| 989 |
-
},
|
| 990 |
-
{
|
| 991 |
-
"epoch": 3.99,
|
| 992 |
-
"learning_rate": 0.0001849206349206349,
|
| 993 |
-
"loss": 0.7311,
|
| 994 |
-
"step": 1550
|
| 995 |
-
},
|
| 996 |
-
{
|
| 997 |
-
"epoch": 4.01,
|
| 998 |
-
"learning_rate": 0.0001841269841269841,
|
| 999 |
-
"loss": 0.7535,
|
| 1000 |
-
"step": 1560
|
| 1001 |
-
},
|
| 1002 |
-
{
|
| 1003 |
-
"epoch": 4.04,
|
| 1004 |
-
"learning_rate": 0.00018333333333333334,
|
| 1005 |
-
"loss": 0.8779,
|
| 1006 |
-
"step": 1570
|
| 1007 |
-
},
|
| 1008 |
-
{
|
| 1009 |
-
"epoch": 4.06,
|
| 1010 |
-
"learning_rate": 0.00018253968253968252,
|
| 1011 |
-
"loss": 0.887,
|
| 1012 |
-
"step": 1580
|
| 1013 |
-
},
|
| 1014 |
-
{
|
| 1015 |
-
"epoch": 4.09,
|
| 1016 |
-
"learning_rate": 0.00018174603174603174,
|
| 1017 |
-
"loss": 0.797,
|
| 1018 |
-
"step": 1590
|
| 1019 |
-
},
|
| 1020 |
-
{
|
| 1021 |
-
"epoch": 4.12,
|
| 1022 |
-
"learning_rate": 0.00018095238095238093,
|
| 1023 |
-
"loss": 0.6651,
|
| 1024 |
-
"step": 1600
|
| 1025 |
-
},
|
| 1026 |
-
{
|
| 1027 |
-
"epoch": 4.12,
|
| 1028 |
-
"eval_loss": 0.904344916343689,
|
| 1029 |
-
"eval_runtime": 189.1226,
|
| 1030 |
-
"eval_samples_per_second": 10.575,
|
| 1031 |
-
"eval_steps_per_second": 1.322,
|
| 1032 |
-
"step": 1600
|
| 1033 |
-
},
|
| 1034 |
-
{
|
| 1035 |
-
"epoch": 4.14,
|
| 1036 |
-
"learning_rate": 0.00018015873015873017,
|
| 1037 |
-
"loss": 0.7348,
|
| 1038 |
-
"step": 1610
|
| 1039 |
-
},
|
| 1040 |
-
{
|
| 1041 |
-
"epoch": 4.17,
|
| 1042 |
-
"learning_rate": 0.00017936507936507936,
|
| 1043 |
-
"loss": 0.8756,
|
| 1044 |
-
"step": 1620
|
| 1045 |
-
},
|
| 1046 |
-
{
|
| 1047 |
-
"epoch": 4.19,
|
| 1048 |
-
"learning_rate": 0.00017857142857142854,
|
| 1049 |
-
"loss": 0.8934,
|
| 1050 |
-
"step": 1630
|
| 1051 |
-
},
|
| 1052 |
-
{
|
| 1053 |
-
"epoch": 4.22,
|
| 1054 |
-
"learning_rate": 0.00017777777777777776,
|
| 1055 |
-
"loss": 0.8023,
|
| 1056 |
-
"step": 1640
|
| 1057 |
-
},
|
| 1058 |
-
{
|
| 1059 |
-
"epoch": 4.24,
|
| 1060 |
-
"learning_rate": 0.00017698412698412697,
|
| 1061 |
-
"loss": 0.6788,
|
| 1062 |
-
"step": 1650
|
| 1063 |
-
},
|
| 1064 |
-
{
|
| 1065 |
-
"epoch": 4.27,
|
| 1066 |
-
"learning_rate": 0.0001761904761904762,
|
| 1067 |
-
"loss": 0.7387,
|
| 1068 |
-
"step": 1660
|
| 1069 |
-
},
|
| 1070 |
-
{
|
| 1071 |
-
"epoch": 4.3,
|
| 1072 |
-
"learning_rate": 0.00017539682539682538,
|
| 1073 |
-
"loss": 0.885,
|
| 1074 |
-
"step": 1670
|
| 1075 |
-
},
|
| 1076 |
-
{
|
| 1077 |
-
"epoch": 4.32,
|
| 1078 |
-
"learning_rate": 0.00017460317460317457,
|
| 1079 |
-
"loss": 0.8738,
|
| 1080 |
-
"step": 1680
|
| 1081 |
-
},
|
| 1082 |
-
{
|
| 1083 |
-
"epoch": 4.35,
|
| 1084 |
-
"learning_rate": 0.0001738095238095238,
|
| 1085 |
-
"loss": 0.8059,
|
| 1086 |
-
"step": 1690
|
| 1087 |
-
},
|
| 1088 |
-
{
|
| 1089 |
-
"epoch": 4.37,
|
| 1090 |
-
"learning_rate": 0.000173015873015873,
|
| 1091 |
-
"loss": 0.6667,
|
| 1092 |
-
"step": 1700
|
| 1093 |
-
},
|
| 1094 |
-
{
|
| 1095 |
-
"epoch": 4.4,
|
| 1096 |
-
"learning_rate": 0.0001722222222222222,
|
| 1097 |
-
"loss": 0.7406,
|
| 1098 |
-
"step": 1710
|
| 1099 |
-
},
|
| 1100 |
-
{
|
| 1101 |
-
"epoch": 4.42,
|
| 1102 |
-
"learning_rate": 0.0001714285714285714,
|
| 1103 |
-
"loss": 0.8764,
|
| 1104 |
-
"step": 1720
|
| 1105 |
-
},
|
| 1106 |
-
{
|
| 1107 |
-
"epoch": 4.45,
|
| 1108 |
-
"learning_rate": 0.00017063492063492064,
|
| 1109 |
-
"loss": 0.8839,
|
| 1110 |
-
"step": 1730
|
| 1111 |
-
},
|
| 1112 |
-
{
|
| 1113 |
-
"epoch": 4.48,
|
| 1114 |
-
"learning_rate": 0.00016984126984126983,
|
| 1115 |
-
"loss": 0.8009,
|
| 1116 |
-
"step": 1740
|
| 1117 |
-
},
|
| 1118 |
-
{
|
| 1119 |
-
"epoch": 4.5,
|
| 1120 |
-
"learning_rate": 0.00016904761904761904,
|
| 1121 |
-
"loss": 0.6658,
|
| 1122 |
-
"step": 1750
|
| 1123 |
-
},
|
| 1124 |
-
{
|
| 1125 |
-
"epoch": 4.53,
|
| 1126 |
-
"learning_rate": 0.00016825396825396823,
|
| 1127 |
-
"loss": 0.7423,
|
| 1128 |
-
"step": 1760
|
| 1129 |
-
},
|
| 1130 |
-
{
|
| 1131 |
-
"epoch": 4.55,
|
| 1132 |
-
"learning_rate": 0.00016746031746031747,
|
| 1133 |
-
"loss": 0.8748,
|
| 1134 |
-
"step": 1770
|
| 1135 |
-
},
|
| 1136 |
-
{
|
| 1137 |
-
"epoch": 4.58,
|
| 1138 |
-
"learning_rate": 0.00016666666666666666,
|
| 1139 |
-
"loss": 0.887,
|
| 1140 |
-
"step": 1780
|
| 1141 |
-
},
|
| 1142 |
-
{
|
| 1143 |
-
"epoch": 4.6,
|
| 1144 |
-
"learning_rate": 0.00016587301587301585,
|
| 1145 |
-
"loss": 0.8038,
|
| 1146 |
-
"step": 1790
|
| 1147 |
-
},
|
| 1148 |
-
{
|
| 1149 |
-
"epoch": 4.63,
|
| 1150 |
-
"learning_rate": 0.00016507936507936506,
|
| 1151 |
-
"loss": 0.6631,
|
| 1152 |
-
"step": 1800
|
| 1153 |
-
},
|
| 1154 |
-
{
|
| 1155 |
-
"epoch": 4.63,
|
| 1156 |
-
"eval_loss": 0.9004252552986145,
|
| 1157 |
-
"eval_runtime": 189.1263,
|
| 1158 |
-
"eval_samples_per_second": 10.575,
|
| 1159 |
-
"eval_steps_per_second": 1.322,
|
| 1160 |
-
"step": 1800
|
| 1161 |
-
},
|
| 1162 |
-
{
|
| 1163 |
-
"epoch": 4.66,
|
| 1164 |
-
"learning_rate": 0.00016428571428571428,
|
| 1165 |
-
"loss": 0.7327,
|
| 1166 |
-
"step": 1810
|
| 1167 |
-
},
|
| 1168 |
-
{
|
| 1169 |
-
"epoch": 4.68,
|
| 1170 |
-
"learning_rate": 0.0001634920634920635,
|
| 1171 |
-
"loss": 0.8703,
|
| 1172 |
-
"step": 1820
|
| 1173 |
-
},
|
| 1174 |
-
{
|
| 1175 |
-
"epoch": 4.71,
|
| 1176 |
-
"learning_rate": 0.00016269841269841268,
|
| 1177 |
-
"loss": 0.8734,
|
| 1178 |
-
"step": 1830
|
| 1179 |
-
},
|
| 1180 |
-
{
|
| 1181 |
-
"epoch": 4.73,
|
| 1182 |
-
"learning_rate": 0.00016190476190476187,
|
| 1183 |
-
"loss": 0.8066,
|
| 1184 |
-
"step": 1840
|
| 1185 |
-
},
|
| 1186 |
-
{
|
| 1187 |
-
"epoch": 4.76,
|
| 1188 |
-
"learning_rate": 0.0001611111111111111,
|
| 1189 |
-
"loss": 0.6655,
|
| 1190 |
-
"step": 1850
|
| 1191 |
-
},
|
| 1192 |
-
{
|
| 1193 |
-
"epoch": 4.78,
|
| 1194 |
-
"learning_rate": 0.0001603174603174603,
|
| 1195 |
-
"loss": 0.736,
|
| 1196 |
-
"step": 1860
|
| 1197 |
-
},
|
| 1198 |
-
{
|
| 1199 |
-
"epoch": 4.81,
|
| 1200 |
-
"learning_rate": 0.00015952380952380951,
|
| 1201 |
-
"loss": 0.8707,
|
| 1202 |
-
"step": 1870
|
| 1203 |
-
},
|
| 1204 |
-
{
|
| 1205 |
-
"epoch": 4.84,
|
| 1206 |
-
"learning_rate": 0.0001587301587301587,
|
| 1207 |
-
"loss": 0.8789,
|
| 1208 |
-
"step": 1880
|
| 1209 |
-
},
|
| 1210 |
-
{
|
| 1211 |
-
"epoch": 4.86,
|
| 1212 |
-
"learning_rate": 0.00015793650793650795,
|
| 1213 |
-
"loss": 0.804,
|
| 1214 |
-
"step": 1890
|
| 1215 |
-
},
|
| 1216 |
-
{
|
| 1217 |
-
"epoch": 4.89,
|
| 1218 |
-
"learning_rate": 0.00015714285714285713,
|
| 1219 |
-
"loss": 0.6628,
|
| 1220 |
-
"step": 1900
|
| 1221 |
-
},
|
| 1222 |
-
{
|
| 1223 |
-
"epoch": 4.91,
|
| 1224 |
-
"learning_rate": 0.00015634920634920635,
|
| 1225 |
-
"loss": 0.7394,
|
| 1226 |
-
"step": 1910
|
| 1227 |
-
},
|
| 1228 |
-
{
|
| 1229 |
-
"epoch": 4.94,
|
| 1230 |
-
"learning_rate": 0.00015555555555555554,
|
| 1231 |
-
"loss": 0.8856,
|
| 1232 |
-
"step": 1920
|
| 1233 |
-
},
|
| 1234 |
-
{
|
| 1235 |
-
"epoch": 4.96,
|
| 1236 |
-
"learning_rate": 0.00015476190476190478,
|
| 1237 |
-
"loss": 0.8558,
|
| 1238 |
-
"step": 1930
|
| 1239 |
-
},
|
| 1240 |
-
{
|
| 1241 |
-
"epoch": 4.99,
|
| 1242 |
-
"learning_rate": 0.00015396825396825397,
|
| 1243 |
-
"loss": 0.6766,
|
| 1244 |
-
"step": 1940
|
| 1245 |
-
},
|
| 1246 |
-
{
|
| 1247 |
-
"epoch": 5.02,
|
| 1248 |
-
"learning_rate": 0.00015317460317460315,
|
| 1249 |
-
"loss": 0.7662,
|
| 1250 |
-
"step": 1950
|
| 1251 |
-
},
|
| 1252 |
-
{
|
| 1253 |
-
"epoch": 5.04,
|
| 1254 |
-
"learning_rate": 0.00015238095238095237,
|
| 1255 |
-
"loss": 0.8602,
|
| 1256 |
-
"step": 1960
|
| 1257 |
-
},
|
| 1258 |
-
{
|
| 1259 |
-
"epoch": 5.07,
|
| 1260 |
-
"learning_rate": 0.00015158730158730158,
|
| 1261 |
-
"loss": 0.8567,
|
| 1262 |
-
"step": 1970
|
| 1263 |
-
},
|
| 1264 |
-
{
|
| 1265 |
-
"epoch": 5.09,
|
| 1266 |
-
"learning_rate": 0.0001507936507936508,
|
| 1267 |
-
"loss": 0.7482,
|
| 1268 |
-
"step": 1980
|
| 1269 |
-
},
|
| 1270 |
-
{
|
| 1271 |
-
"epoch": 5.12,
|
| 1272 |
-
"learning_rate": 0.00015,
|
| 1273 |
-
"loss": 0.6249,
|
| 1274 |
-
"step": 1990
|
| 1275 |
-
},
|
| 1276 |
-
{
|
| 1277 |
-
"epoch": 5.14,
|
| 1278 |
-
"learning_rate": 0.00014920634920634917,
|
| 1279 |
-
"loss": 0.7506,
|
| 1280 |
-
"step": 2000
|
| 1281 |
-
},
|
| 1282 |
-
{
|
| 1283 |
-
"epoch": 5.14,
|
| 1284 |
-
"eval_loss": 0.9112463593482971,
|
| 1285 |
-
"eval_runtime": 189.2232,
|
| 1286 |
-
"eval_samples_per_second": 10.57,
|
| 1287 |
-
"eval_steps_per_second": 1.321,
|
| 1288 |
-
"step": 2000
|
| 1289 |
-
},
|
| 1290 |
-
{
|
| 1291 |
-
"epoch": 5.17,
|
| 1292 |
-
"learning_rate": 0.0001484126984126984,
|
| 1293 |
-
"loss": 0.8679,
|
| 1294 |
-
"step": 2010
|
| 1295 |
-
},
|
| 1296 |
-
{
|
| 1297 |
-
"epoch": 5.2,
|
| 1298 |
-
"learning_rate": 0.0001476190476190476,
|
| 1299 |
-
"loss": 0.8575,
|
| 1300 |
-
"step": 2020
|
| 1301 |
-
},
|
| 1302 |
-
{
|
| 1303 |
-
"epoch": 5.22,
|
| 1304 |
-
"learning_rate": 0.00014682539682539682,
|
| 1305 |
-
"loss": 0.7545,
|
| 1306 |
-
"step": 2030
|
| 1307 |
-
},
|
| 1308 |
-
{
|
| 1309 |
-
"epoch": 5.25,
|
| 1310 |
-
"learning_rate": 0.000146031746031746,
|
| 1311 |
-
"loss": 0.6237,
|
| 1312 |
-
"step": 2040
|
| 1313 |
-
},
|
| 1314 |
-
{
|
| 1315 |
-
"epoch": 5.27,
|
| 1316 |
-
"learning_rate": 0.00014523809523809522,
|
| 1317 |
-
"loss": 0.7561,
|
| 1318 |
-
"step": 2050
|
| 1319 |
-
},
|
| 1320 |
-
{
|
| 1321 |
-
"epoch": 5.3,
|
| 1322 |
-
"learning_rate": 0.0001444444444444444,
|
| 1323 |
-
"loss": 0.8688,
|
| 1324 |
-
"step": 2060
|
| 1325 |
-
},
|
| 1326 |
-
{
|
| 1327 |
-
"epoch": 5.32,
|
| 1328 |
-
"learning_rate": 0.00014365079365079363,
|
| 1329 |
-
"loss": 0.8578,
|
| 1330 |
-
"step": 2070
|
| 1331 |
-
},
|
| 1332 |
-
{
|
| 1333 |
-
"epoch": 5.35,
|
| 1334 |
-
"learning_rate": 0.00014285714285714284,
|
| 1335 |
-
"loss": 0.7743,
|
| 1336 |
-
"step": 2080
|
| 1337 |
-
},
|
| 1338 |
-
{
|
| 1339 |
-
"epoch": 5.38,
|
| 1340 |
-
"learning_rate": 0.00014206349206349206,
|
| 1341 |
-
"loss": 0.6337,
|
| 1342 |
-
"step": 2090
|
| 1343 |
-
},
|
| 1344 |
-
{
|
| 1345 |
-
"epoch": 5.4,
|
| 1346 |
-
"learning_rate": 0.00014126984126984124,
|
| 1347 |
-
"loss": 0.7518,
|
| 1348 |
-
"step": 2100
|
| 1349 |
-
},
|
| 1350 |
-
{
|
| 1351 |
-
"epoch": 5.43,
|
| 1352 |
-
"learning_rate": 0.00014047619047619046,
|
| 1353 |
-
"loss": 0.8736,
|
| 1354 |
-
"step": 2110
|
| 1355 |
-
},
|
| 1356 |
-
{
|
| 1357 |
-
"epoch": 5.45,
|
| 1358 |
-
"learning_rate": 0.00013968253968253967,
|
| 1359 |
-
"loss": 0.8623,
|
| 1360 |
-
"step": 2120
|
| 1361 |
-
},
|
| 1362 |
-
{
|
| 1363 |
-
"epoch": 5.48,
|
| 1364 |
-
"learning_rate": 0.0001388888888888889,
|
| 1365 |
-
"loss": 0.7719,
|
| 1366 |
-
"step": 2130
|
| 1367 |
-
},
|
| 1368 |
-
{
|
| 1369 |
-
"epoch": 5.5,
|
| 1370 |
-
"learning_rate": 0.00013809523809523808,
|
| 1371 |
-
"loss": 0.6363,
|
| 1372 |
-
"step": 2140
|
| 1373 |
-
},
|
| 1374 |
-
{
|
| 1375 |
-
"epoch": 5.53,
|
| 1376 |
-
"learning_rate": 0.0001373015873015873,
|
| 1377 |
-
"loss": 0.7458,
|
| 1378 |
-
"step": 2150
|
| 1379 |
-
},
|
| 1380 |
-
{
|
| 1381 |
-
"epoch": 5.56,
|
| 1382 |
-
"learning_rate": 0.00013650793650793648,
|
| 1383 |
-
"loss": 0.865,
|
| 1384 |
-
"step": 2160
|
| 1385 |
-
},
|
| 1386 |
-
{
|
| 1387 |
-
"epoch": 5.58,
|
| 1388 |
-
"learning_rate": 0.00013587301587301588,
|
| 1389 |
-
"loss": 2.6458,
|
| 1390 |
-
"step": 2170
|
| 1391 |
-
},
|
| 1392 |
-
{
|
| 1393 |
-
"epoch": 5.61,
|
| 1394 |
-
"learning_rate": 0.00013587301587301588,
|
| 1395 |
-
"loss": 801.9857,
|
| 1396 |
-
"step": 2180
|
| 1397 |
-
},
|
| 1398 |
-
{
|
| 1399 |
-
"epoch": 5.63,
|
| 1400 |
-
"learning_rate": 0.0001357142857142857,
|
| 1401 |
-
"loss": 4068.018,
|
| 1402 |
-
"step": 2190
|
| 1403 |
-
},
|
| 1404 |
-
{
|
| 1405 |
-
"epoch": 5.66,
|
| 1406 |
-
"learning_rate": 0.00013547619047619047,
|
| 1407 |
-
"loss": 41837456.0,
|
| 1408 |
-
"step": 2200
|
| 1409 |
-
},
|
| 1410 |
-
{
|
| 1411 |
-
"epoch": 5.66,
|
| 1412 |
-
"eval_loss": NaN,
|
| 1413 |
-
"eval_runtime": 189.1905,
|
| 1414 |
-
"eval_samples_per_second": 10.571,
|
| 1415 |
-
"eval_steps_per_second": 1.321,
|
| 1416 |
-
"step": 2200
|
| 1417 |
-
},
|
| 1418 |
-
{
|
| 1419 |
-
"epoch": 5.68,
|
| 1420 |
-
"learning_rate": 0.00013547619047619047,
|
| 1421 |
-
"loss": 1.4509301308571507e+26,
|
| 1422 |
-
"step": 2210
|
| 1423 |
-
},
|
| 1424 |
-
{
|
| 1425 |
-
"epoch": 5.71,
|
| 1426 |
-
"learning_rate": 0.00013547619047619047,
|
| 1427 |
-
"loss": 3.3847730092507856e+24,
|
| 1428 |
-
"step": 2220
|
| 1429 |
-
},
|
| 1430 |
-
{
|
| 1431 |
-
"epoch": 5.74,
|
| 1432 |
-
"learning_rate": 0.00013547619047619047,
|
| 1433 |
-
"loss": 1.5211943209070177e+23,
|
| 1434 |
-
"step": 2230
|
| 1435 |
-
},
|
| 1436 |
-
{
|
| 1437 |
-
"epoch": 5.76,
|
| 1438 |
-
"learning_rate": 0.00013547619047619047,
|
| 1439 |
-
"loss": 6.678915709304036e+21,
|
| 1440 |
-
"step": 2240
|
| 1441 |
-
},
|
| 1442 |
-
{
|
| 1443 |
-
"epoch": 5.79,
|
| 1444 |
-
"learning_rate": 0.00013531746031746032,
|
| 1445 |
-
"loss": 9.238063623264189e+19,
|
| 1446 |
-
"step": 2250
|
| 1447 |
-
},
|
| 1448 |
-
{
|
| 1449 |
-
"epoch": 5.81,
|
| 1450 |
-
"learning_rate": 0.00013531746031746032,
|
| 1451 |
-
"loss": 1.0532014646514701e+20,
|
| 1452 |
-
"step": 2260
|
| 1453 |
-
},
|
| 1454 |
-
{
|
| 1455 |
-
"epoch": 5.84,
|
| 1456 |
-
"learning_rate": 0.00013531746031746032,
|
| 1457 |
-
"loss": 3.161435096687031e+25,
|
| 1458 |
-
"step": 2270
|
| 1459 |
-
},
|
| 1460 |
-
{
|
| 1461 |
-
"epoch": 5.86,
|
| 1462 |
-
"learning_rate": 0.00013531746031746032,
|
| 1463 |
-
"loss": 2.0162940987179532e+19,
|
| 1464 |
-
"step": 2280
|
| 1465 |
-
},
|
| 1466 |
-
{
|
| 1467 |
-
"epoch": 5.89,
|
| 1468 |
-
"learning_rate": 0.00013531746031746032,
|
| 1469 |
-
"loss": 2.55694599151234e+20,
|
| 1470 |
-
"step": 2290
|
| 1471 |
-
},
|
| 1472 |
-
{
|
| 1473 |
-
"epoch": 5.92,
|
| 1474 |
-
"learning_rate": 0.00013523809523809522,
|
| 1475 |
-
"loss": 5.808441058207432e+20,
|
| 1476 |
-
"step": 2300
|
| 1477 |
-
},
|
| 1478 |
-
{
|
| 1479 |
-
"epoch": 5.94,
|
| 1480 |
-
"learning_rate": 0.00013523809523809522,
|
| 1481 |
-
"loss": 1.9880088514154103e+22,
|
| 1482 |
-
"step": 2310
|
| 1483 |
-
},
|
| 1484 |
-
{
|
| 1485 |
-
"epoch": 5.97,
|
| 1486 |
-
"learning_rate": 0.00013523809523809522,
|
| 1487 |
-
"loss": 2.0954874078435546e+24,
|
| 1488 |
-
"step": 2320
|
| 1489 |
-
},
|
| 1490 |
-
{
|
| 1491 |
-
"epoch": 5.99,
|
| 1492 |
-
"learning_rate": 0.00013523809523809522,
|
| 1493 |
-
"loss": 6.309141694629275e+20,
|
| 1494 |
-
"step": 2330
|
| 1495 |
-
},
|
| 1496 |
-
{
|
| 1497 |
-
"epoch": 6.02,
|
| 1498 |
-
"learning_rate": 0.00013515873015873016,
|
| 1499 |
-
"loss": 7.353349497283535e+23,
|
| 1500 |
-
"step": 2340
|
| 1501 |
-
},
|
| 1502 |
-
{
|
| 1503 |
-
"epoch": 6.05,
|
| 1504 |
-
"learning_rate": 0.00013468253968253966,
|
| 1505 |
-
"loss": 0.0,
|
| 1506 |
-
"step": 2350
|
| 1507 |
-
},
|
| 1508 |
-
{
|
| 1509 |
-
"epoch": 6.07,
|
| 1510 |
-
"learning_rate": 0.00013388888888888888,
|
| 1511 |
-
"loss": 0.0,
|
| 1512 |
-
"step": 2360
|
| 1513 |
-
},
|
| 1514 |
-
{
|
| 1515 |
-
"epoch": 6.1,
|
| 1516 |
-
"learning_rate": 0.00013309523809523806,
|
| 1517 |
-
"loss": 0.0,
|
| 1518 |
-
"step": 2370
|
| 1519 |
-
},
|
| 1520 |
-
{
|
| 1521 |
-
"epoch": 6.12,
|
| 1522 |
-
"learning_rate": 0.00013230158730158728,
|
| 1523 |
-
"loss": 0.0,
|
| 1524 |
-
"step": 2380
|
| 1525 |
-
},
|
| 1526 |
-
{
|
| 1527 |
-
"epoch": 6.15,
|
| 1528 |
-
"learning_rate": 0.0001315079365079365,
|
| 1529 |
-
"loss": 0.0,
|
| 1530 |
-
"step": 2390
|
| 1531 |
-
},
|
| 1532 |
-
{
|
| 1533 |
-
"epoch": 6.17,
|
| 1534 |
-
"learning_rate": 0.0001307142857142857,
|
| 1535 |
-
"loss": 0.0,
|
| 1536 |
-
"step": 2400
|
| 1537 |
-
},
|
| 1538 |
-
{
|
| 1539 |
-
"epoch": 6.17,
|
| 1540 |
-
"eval_loss": NaN,
|
| 1541 |
-
"eval_runtime": 164.902,
|
| 1542 |
-
"eval_samples_per_second": 12.128,
|
| 1543 |
-
"eval_steps_per_second": 1.516,
|
| 1544 |
-
"step": 2400
|
| 1545 |
-
},
|
| 1546 |
-
{
|
| 1547 |
-
"epoch": 6.2,
|
| 1548 |
-
"learning_rate": 0.0001299206349206349,
|
| 1549 |
-
"loss": 0.0,
|
| 1550 |
-
"step": 2410
|
| 1551 |
-
},
|
| 1552 |
-
{
|
| 1553 |
-
"epoch": 6.23,
|
| 1554 |
-
"learning_rate": 0.0001291269841269841,
|
| 1555 |
-
"loss": 0.0,
|
| 1556 |
-
"step": 2420
|
| 1557 |
-
},
|
| 1558 |
-
{
|
| 1559 |
-
"epoch": 6.25,
|
| 1560 |
-
"learning_rate": 0.00012833333333333333,
|
| 1561 |
-
"loss": 0.0,
|
| 1562 |
-
"step": 2430
|
| 1563 |
-
},
|
| 1564 |
-
{
|
| 1565 |
-
"epoch": 6.28,
|
| 1566 |
-
"learning_rate": 0.00012753968253968254,
|
| 1567 |
-
"loss": 0.0,
|
| 1568 |
-
"step": 2440
|
| 1569 |
-
},
|
| 1570 |
-
{
|
| 1571 |
-
"epoch": 6.3,
|
| 1572 |
-
"learning_rate": 0.00012674603174603173,
|
| 1573 |
-
"loss": 0.0,
|
| 1574 |
-
"step": 2450
|
| 1575 |
-
},
|
| 1576 |
-
{
|
| 1577 |
-
"epoch": 6.33,
|
| 1578 |
-
"learning_rate": 0.00012595238095238094,
|
| 1579 |
-
"loss": 0.0,
|
| 1580 |
-
"step": 2460
|
| 1581 |
-
},
|
| 1582 |
-
{
|
| 1583 |
-
"epoch": 6.35,
|
| 1584 |
-
"learning_rate": 0.00012515873015873013,
|
| 1585 |
-
"loss": 0.0,
|
| 1586 |
-
"step": 2470
|
| 1587 |
-
},
|
| 1588 |
-
{
|
| 1589 |
-
"epoch": 6.38,
|
| 1590 |
-
"learning_rate": 0.00012436507936507935,
|
| 1591 |
-
"loss": 0.0,
|
| 1592 |
-
"step": 2480
|
| 1593 |
-
},
|
| 1594 |
-
{
|
| 1595 |
-
"epoch": 6.41,
|
| 1596 |
-
"learning_rate": 0.00012357142857142856,
|
| 1597 |
-
"loss": 0.0,
|
| 1598 |
-
"step": 2490
|
| 1599 |
-
},
|
| 1600 |
-
{
|
| 1601 |
-
"epoch": 6.43,
|
| 1602 |
-
"learning_rate": 0.00012277777777777778,
|
| 1603 |
-
"loss": 0.0,
|
| 1604 |
-
"step": 2500
|
| 1605 |
-
},
|
| 1606 |
-
{
|
| 1607 |
-
"epoch": 6.46,
|
| 1608 |
-
"learning_rate": 0.00012198412698412697,
|
| 1609 |
-
"loss": 0.0,
|
| 1610 |
-
"step": 2510
|
| 1611 |
-
},
|
| 1612 |
-
{
|
| 1613 |
-
"epoch": 6.48,
|
| 1614 |
-
"learning_rate": 0.00012119047619047618,
|
| 1615 |
-
"loss": 0.0,
|
| 1616 |
-
"step": 2520
|
| 1617 |
-
},
|
| 1618 |
-
{
|
| 1619 |
-
"epoch": 6.51,
|
| 1620 |
-
"learning_rate": 0.00012039682539682538,
|
| 1621 |
-
"loss": 0.0,
|
| 1622 |
-
"step": 2530
|
| 1623 |
-
},
|
| 1624 |
-
{
|
| 1625 |
-
"epoch": 6.53,
|
| 1626 |
-
"learning_rate": 0.0001196031746031746,
|
| 1627 |
-
"loss": 0.0,
|
| 1628 |
-
"step": 2540
|
| 1629 |
-
},
|
| 1630 |
-
{
|
| 1631 |
-
"epoch": 6.56,
|
| 1632 |
-
"learning_rate": 0.0001188095238095238,
|
| 1633 |
-
"loss": 0.0,
|
| 1634 |
-
"step": 2550
|
| 1635 |
-
},
|
| 1636 |
-
{
|
| 1637 |
-
"epoch": 6.59,
|
| 1638 |
-
"learning_rate": 0.00011801587301587301,
|
| 1639 |
-
"loss": 0.0,
|
| 1640 |
-
"step": 2560
|
| 1641 |
-
},
|
| 1642 |
-
{
|
| 1643 |
-
"epoch": 6.61,
|
| 1644 |
-
"learning_rate": 0.0001172222222222222,
|
| 1645 |
-
"loss": 0.0,
|
| 1646 |
-
"step": 2570
|
| 1647 |
-
},
|
| 1648 |
-
{
|
| 1649 |
-
"epoch": 6.64,
|
| 1650 |
-
"learning_rate": 0.00011642857142857142,
|
| 1651 |
-
"loss": 0.0,
|
| 1652 |
-
"step": 2580
|
| 1653 |
-
},
|
| 1654 |
-
{
|
| 1655 |
-
"epoch": 6.66,
|
| 1656 |
-
"learning_rate": 0.00011563492063492062,
|
| 1657 |
-
"loss": 0.0,
|
| 1658 |
-
"step": 2590
|
| 1659 |
-
},
|
| 1660 |
-
{
|
| 1661 |
-
"epoch": 6.69,
|
| 1662 |
-
"learning_rate": 0.00011484126984126983,
|
| 1663 |
-
"loss": 0.0,
|
| 1664 |
-
"step": 2600
|
| 1665 |
-
},
|
| 1666 |
-
{
|
| 1667 |
-
"epoch": 6.69,
|
| 1668 |
-
"eval_loss": NaN,
|
| 1669 |
-
"eval_runtime": 164.9294,
|
| 1670 |
-
"eval_samples_per_second": 12.126,
|
| 1671 |
-
"eval_steps_per_second": 1.516,
|
| 1672 |
-
"step": 2600
|
| 1673 |
-
},
|
| 1674 |
-
{
|
| 1675 |
-
"epoch": 6.71,
|
| 1676 |
-
"learning_rate": 0.00011404761904761903,
|
| 1677 |
-
"loss": 0.0,
|
| 1678 |
-
"step": 2610
|
| 1679 |
-
},
|
| 1680 |
-
{
|
| 1681 |
-
"epoch": 6.74,
|
| 1682 |
-
"learning_rate": 0.00011325396825396825,
|
| 1683 |
-
"loss": 0.0,
|
| 1684 |
-
"step": 2620
|
| 1685 |
-
},
|
| 1686 |
-
{
|
| 1687 |
-
"epoch": 6.77,
|
| 1688 |
-
"learning_rate": 0.00011246031746031745,
|
| 1689 |
-
"loss": 0.0,
|
| 1690 |
-
"step": 2630
|
| 1691 |
-
},
|
| 1692 |
-
{
|
| 1693 |
-
"epoch": 6.79,
|
| 1694 |
-
"learning_rate": 0.00011166666666666667,
|
| 1695 |
-
"loss": 0.0,
|
| 1696 |
-
"step": 2640
|
| 1697 |
-
},
|
| 1698 |
-
{
|
| 1699 |
-
"epoch": 6.82,
|
| 1700 |
-
"learning_rate": 0.00011087301587301585,
|
| 1701 |
-
"loss": 0.0,
|
| 1702 |
-
"step": 2650
|
| 1703 |
-
},
|
| 1704 |
-
{
|
| 1705 |
-
"epoch": 6.84,
|
| 1706 |
-
"learning_rate": 0.00011007936507936507,
|
| 1707 |
-
"loss": 0.0,
|
| 1708 |
-
"step": 2660
|
| 1709 |
-
},
|
| 1710 |
-
{
|
| 1711 |
-
"epoch": 6.87,
|
| 1712 |
-
"learning_rate": 0.00010928571428571427,
|
| 1713 |
-
"loss": 0.0,
|
| 1714 |
-
"step": 2670
|
| 1715 |
-
},
|
| 1716 |
-
{
|
| 1717 |
-
"epoch": 6.89,
|
| 1718 |
-
"learning_rate": 0.00010849206349206349,
|
| 1719 |
-
"loss": 0.0,
|
| 1720 |
-
"step": 2680
|
| 1721 |
-
},
|
| 1722 |
-
{
|
| 1723 |
-
"epoch": 6.92,
|
| 1724 |
-
"learning_rate": 0.00010769841269841269,
|
| 1725 |
-
"loss": 0.0,
|
| 1726 |
-
"step": 2690
|
| 1727 |
-
},
|
| 1728 |
-
{
|
| 1729 |
-
"epoch": 6.95,
|
| 1730 |
-
"learning_rate": 0.0001069047619047619,
|
| 1731 |
-
"loss": 0.0,
|
| 1732 |
-
"step": 2700
|
| 1733 |
-
},
|
| 1734 |
-
{
|
| 1735 |
-
"epoch": 6.97,
|
| 1736 |
-
"learning_rate": 0.0001061111111111111,
|
| 1737 |
-
"loss": 0.0,
|
| 1738 |
-
"step": 2710
|
| 1739 |
-
},
|
| 1740 |
-
{
|
| 1741 |
-
"epoch": 7.0,
|
| 1742 |
-
"learning_rate": 0.0001053174603174603,
|
| 1743 |
-
"loss": 0.0,
|
| 1744 |
-
"step": 2720
|
| 1745 |
-
},
|
| 1746 |
-
{
|
| 1747 |
-
"epoch": 7.02,
|
| 1748 |
-
"learning_rate": 0.0001045238095238095,
|
| 1749 |
-
"loss": 0.0,
|
| 1750 |
-
"step": 2730
|
| 1751 |
-
},
|
| 1752 |
-
{
|
| 1753 |
-
"epoch": 7.05,
|
| 1754 |
-
"learning_rate": 0.00010373015873015872,
|
| 1755 |
-
"loss": 0.0,
|
| 1756 |
-
"step": 2740
|
| 1757 |
-
},
|
| 1758 |
-
{
|
| 1759 |
-
"epoch": 7.07,
|
| 1760 |
-
"learning_rate": 0.00010293650793650792,
|
| 1761 |
-
"loss": 0.0,
|
| 1762 |
-
"step": 2750
|
| 1763 |
-
},
|
| 1764 |
-
{
|
| 1765 |
-
"epoch": 7.1,
|
| 1766 |
-
"learning_rate": 0.00010214285714285714,
|
| 1767 |
-
"loss": 0.0,
|
| 1768 |
-
"step": 2760
|
| 1769 |
-
},
|
| 1770 |
-
{
|
| 1771 |
-
"epoch": 7.13,
|
| 1772 |
-
"learning_rate": 0.00010134920634920634,
|
| 1773 |
-
"loss": 0.0,
|
| 1774 |
-
"step": 2770
|
| 1775 |
-
},
|
| 1776 |
-
{
|
| 1777 |
-
"epoch": 7.15,
|
| 1778 |
-
"learning_rate": 0.00010055555555555555,
|
| 1779 |
-
"loss": 0.0,
|
| 1780 |
-
"step": 2780
|
| 1781 |
-
},
|
| 1782 |
-
{
|
| 1783 |
-
"epoch": 7.18,
|
| 1784 |
-
"learning_rate": 9.976190476190474e-05,
|
| 1785 |
-
"loss": 0.0,
|
| 1786 |
-
"step": 2790
|
| 1787 |
-
},
|
| 1788 |
-
{
|
| 1789 |
-
"epoch": 7.2,
|
| 1790 |
-
"learning_rate": 9.896825396825396e-05,
|
| 1791 |
-
"loss": 0.0,
|
| 1792 |
-
"step": 2800
|
| 1793 |
-
},
|
| 1794 |
-
{
|
| 1795 |
-
"epoch": 7.2,
|
| 1796 |
-
"eval_loss": NaN,
|
| 1797 |
-
"eval_runtime": 164.8859,
|
| 1798 |
-
"eval_samples_per_second": 12.13,
|
| 1799 |
-
"eval_steps_per_second": 1.516,
|
| 1800 |
-
"step": 2800
|
| 1801 |
-
},
|
| 1802 |
-
{
|
| 1803 |
-
"epoch": 7.23,
|
| 1804 |
-
"learning_rate": 9.817460317460316e-05,
|
| 1805 |
-
"loss": 0.0,
|
| 1806 |
-
"step": 2810
|
| 1807 |
-
},
|
| 1808 |
-
{
|
| 1809 |
-
"epoch": 7.25,
|
| 1810 |
-
"learning_rate": 9.738095238095237e-05,
|
| 1811 |
-
"loss": 0.0,
|
| 1812 |
-
"step": 2820
|
| 1813 |
-
},
|
| 1814 |
-
{
|
| 1815 |
-
"epoch": 7.28,
|
| 1816 |
-
"learning_rate": 9.658730158730158e-05,
|
| 1817 |
-
"loss": 0.0,
|
| 1818 |
-
"step": 2830
|
| 1819 |
-
},
|
| 1820 |
-
{
|
| 1821 |
-
"epoch": 7.31,
|
| 1822 |
-
"learning_rate": 9.579365079365079e-05,
|
| 1823 |
-
"loss": 0.0,
|
| 1824 |
-
"step": 2840
|
| 1825 |
-
},
|
| 1826 |
-
{
|
| 1827 |
-
"epoch": 7.33,
|
| 1828 |
-
"learning_rate": 9.499999999999999e-05,
|
| 1829 |
-
"loss": 0.0,
|
| 1830 |
-
"step": 2850
|
| 1831 |
-
},
|
| 1832 |
-
{
|
| 1833 |
-
"epoch": 7.36,
|
| 1834 |
-
"learning_rate": 9.42063492063492e-05,
|
| 1835 |
-
"loss": 0.0,
|
| 1836 |
-
"step": 2860
|
| 1837 |
-
},
|
| 1838 |
-
{
|
| 1839 |
-
"epoch": 7.38,
|
| 1840 |
-
"learning_rate": 9.34126984126984e-05,
|
| 1841 |
-
"loss": 0.0,
|
| 1842 |
-
"step": 2870
|
| 1843 |
-
},
|
| 1844 |
-
{
|
| 1845 |
-
"epoch": 7.41,
|
| 1846 |
-
"learning_rate": 9.261904761904761e-05,
|
| 1847 |
-
"loss": 0.0,
|
| 1848 |
-
"step": 2880
|
| 1849 |
-
},
|
| 1850 |
-
{
|
| 1851 |
-
"epoch": 7.43,
|
| 1852 |
-
"learning_rate": 9.182539682539681e-05,
|
| 1853 |
-
"loss": 0.0,
|
| 1854 |
-
"step": 2890
|
| 1855 |
-
},
|
| 1856 |
-
{
|
| 1857 |
-
"epoch": 7.46,
|
| 1858 |
-
"learning_rate": 9.103174603174603e-05,
|
| 1859 |
-
"loss": 0.0,
|
| 1860 |
-
"step": 2900
|
| 1861 |
-
},
|
| 1862 |
-
{
|
| 1863 |
-
"epoch": 7.49,
|
| 1864 |
-
"learning_rate": 9.023809523809523e-05,
|
| 1865 |
-
"loss": 0.0,
|
| 1866 |
-
"step": 2910
|
| 1867 |
-
},
|
| 1868 |
-
{
|
| 1869 |
-
"epoch": 7.51,
|
| 1870 |
-
"learning_rate": 8.944444444444444e-05,
|
| 1871 |
-
"loss": 0.0,
|
| 1872 |
-
"step": 2920
|
| 1873 |
-
},
|
| 1874 |
-
{
|
| 1875 |
-
"epoch": 7.54,
|
| 1876 |
-
"learning_rate": 8.865079365079364e-05,
|
| 1877 |
-
"loss": 0.0,
|
| 1878 |
-
"step": 2930
|
| 1879 |
-
},
|
| 1880 |
-
{
|
| 1881 |
-
"epoch": 7.56,
|
| 1882 |
-
"learning_rate": 8.785714285714286e-05,
|
| 1883 |
-
"loss": 0.0,
|
| 1884 |
-
"step": 2940
|
| 1885 |
-
},
|
| 1886 |
-
{
|
| 1887 |
-
"epoch": 7.59,
|
| 1888 |
-
"learning_rate": 8.706349206349205e-05,
|
| 1889 |
-
"loss": 0.0,
|
| 1890 |
-
"step": 2950
|
| 1891 |
-
},
|
| 1892 |
-
{
|
| 1893 |
-
"epoch": 7.61,
|
| 1894 |
-
"learning_rate": 8.626984126984126e-05,
|
| 1895 |
-
"loss": 0.0,
|
| 1896 |
-
"step": 2960
|
| 1897 |
-
},
|
| 1898 |
-
{
|
| 1899 |
-
"epoch": 7.64,
|
| 1900 |
-
"learning_rate": 8.547619047619046e-05,
|
| 1901 |
-
"loss": 0.0,
|
| 1902 |
-
"step": 2970
|
| 1903 |
-
},
|
| 1904 |
-
{
|
| 1905 |
-
"epoch": 7.67,
|
| 1906 |
-
"learning_rate": 8.468253968253968e-05,
|
| 1907 |
-
"loss": 0.0,
|
| 1908 |
-
"step": 2980
|
| 1909 |
-
},
|
| 1910 |
-
{
|
| 1911 |
-
"epoch": 7.69,
|
| 1912 |
-
"learning_rate": 8.388888888888888e-05,
|
| 1913 |
-
"loss": 0.0,
|
| 1914 |
-
"step": 2990
|
| 1915 |
-
},
|
| 1916 |
-
{
|
| 1917 |
-
"epoch": 7.72,
|
| 1918 |
-
"learning_rate": 8.30952380952381e-05,
|
| 1919 |
-
"loss": 0.0,
|
| 1920 |
-
"step": 3000
|
| 1921 |
-
},
|
| 1922 |
-
{
|
| 1923 |
-
"epoch": 7.72,
|
| 1924 |
-
"eval_loss": NaN,
|
| 1925 |
-
"eval_runtime": 164.8915,
|
| 1926 |
-
"eval_samples_per_second": 12.129,
|
| 1927 |
-
"eval_steps_per_second": 1.516,
|
| 1928 |
-
"step": 3000
|
| 1929 |
-
},
|
| 1930 |
-
{
|
| 1931 |
-
"epoch": 7.74,
|
| 1932 |
-
"learning_rate": 8.23015873015873e-05,
|
| 1933 |
-
"loss": 0.0,
|
| 1934 |
-
"step": 3010
|
| 1935 |
-
},
|
| 1936 |
-
{
|
| 1937 |
-
"epoch": 7.77,
|
| 1938 |
-
"learning_rate": 8.150793650793651e-05,
|
| 1939 |
-
"loss": 0.0,
|
| 1940 |
-
"step": 3020
|
| 1941 |
-
},
|
| 1942 |
-
{
|
| 1943 |
-
"epoch": 7.79,
|
| 1944 |
-
"learning_rate": 8.07142857142857e-05,
|
| 1945 |
-
"loss": 0.0,
|
| 1946 |
-
"step": 3030
|
| 1947 |
-
},
|
| 1948 |
-
{
|
| 1949 |
-
"epoch": 7.82,
|
| 1950 |
-
"learning_rate": 7.992063492063491e-05,
|
| 1951 |
-
"loss": 0.0,
|
| 1952 |
-
"step": 3040
|
| 1953 |
-
},
|
| 1954 |
-
{
|
| 1955 |
-
"epoch": 7.85,
|
| 1956 |
-
"learning_rate": 7.912698412698412e-05,
|
| 1957 |
-
"loss": 0.0,
|
| 1958 |
-
"step": 3050
|
| 1959 |
-
},
|
| 1960 |
-
{
|
| 1961 |
-
"epoch": 7.87,
|
| 1962 |
-
"learning_rate": 7.833333333333333e-05,
|
| 1963 |
-
"loss": 0.0,
|
| 1964 |
-
"step": 3060
|
| 1965 |
-
},
|
| 1966 |
-
{
|
| 1967 |
-
"epoch": 7.9,
|
| 1968 |
-
"learning_rate": 7.753968253968253e-05,
|
| 1969 |
-
"loss": 0.0,
|
| 1970 |
-
"step": 3070
|
| 1971 |
-
},
|
| 1972 |
-
{
|
| 1973 |
-
"epoch": 7.92,
|
| 1974 |
-
"learning_rate": 7.674603174603175e-05,
|
| 1975 |
-
"loss": 0.0,
|
| 1976 |
-
"step": 3080
|
| 1977 |
-
},
|
| 1978 |
-
{
|
| 1979 |
-
"epoch": 7.95,
|
| 1980 |
-
"learning_rate": 7.595238095238095e-05,
|
| 1981 |
-
"loss": 0.0,
|
| 1982 |
-
"step": 3090
|
| 1983 |
-
},
|
| 1984 |
-
{
|
| 1985 |
-
"epoch": 7.97,
|
| 1986 |
-
"learning_rate": 7.515873015873015e-05,
|
| 1987 |
-
"loss": 0.0,
|
| 1988 |
-
"step": 3100
|
| 1989 |
-
},
|
| 1990 |
-
{
|
| 1991 |
-
"epoch": 8.0,
|
| 1992 |
-
"learning_rate": 7.436507936507935e-05,
|
| 1993 |
-
"loss": 0.0,
|
| 1994 |
-
"step": 3110
|
| 1995 |
-
},
|
| 1996 |
-
{
|
| 1997 |
-
"epoch": 8.03,
|
| 1998 |
-
"learning_rate": 7.357142857142857e-05,
|
| 1999 |
-
"loss": 0.0,
|
| 2000 |
-
"step": 3120
|
| 2001 |
-
},
|
| 2002 |
-
{
|
| 2003 |
-
"epoch": 8.05,
|
| 2004 |
-
"learning_rate": 7.277777777777777e-05,
|
| 2005 |
-
"loss": 0.0,
|
| 2006 |
-
"step": 3130
|
| 2007 |
-
},
|
| 2008 |
-
{
|
| 2009 |
-
"epoch": 8.08,
|
| 2010 |
-
"learning_rate": 7.198412698412697e-05,
|
| 2011 |
-
"loss": 0.0,
|
| 2012 |
-
"step": 3140
|
| 2013 |
-
},
|
| 2014 |
-
{
|
| 2015 |
-
"epoch": 8.1,
|
| 2016 |
-
"learning_rate": 7.119047619047618e-05,
|
| 2017 |
-
"loss": 0.0,
|
| 2018 |
-
"step": 3150
|
| 2019 |
-
},
|
| 2020 |
-
{
|
| 2021 |
-
"epoch": 8.13,
|
| 2022 |
-
"learning_rate": 7.039682539682539e-05,
|
| 2023 |
-
"loss": 0.0,
|
| 2024 |
-
"step": 3160
|
| 2025 |
-
},
|
| 2026 |
-
{
|
| 2027 |
-
"epoch": 8.15,
|
| 2028 |
-
"learning_rate": 6.960317460317459e-05,
|
| 2029 |
-
"loss": 0.0,
|
| 2030 |
-
"step": 3170
|
| 2031 |
-
},
|
| 2032 |
-
{
|
| 2033 |
-
"epoch": 8.18,
|
| 2034 |
-
"learning_rate": 6.88095238095238e-05,
|
| 2035 |
-
"loss": 0.0,
|
| 2036 |
-
"step": 3180
|
| 2037 |
-
},
|
| 2038 |
-
{
|
| 2039 |
-
"epoch": 8.21,
|
| 2040 |
-
"learning_rate": 6.8015873015873e-05,
|
| 2041 |
-
"loss": 0.0,
|
| 2042 |
-
"step": 3190
|
| 2043 |
-
},
|
| 2044 |
-
{
|
| 2045 |
-
"epoch": 8.23,
|
| 2046 |
-
"learning_rate": 6.722222222222222e-05,
|
| 2047 |
-
"loss": 0.0,
|
| 2048 |
-
"step": 3200
|
| 2049 |
-
},
|
| 2050 |
-
{
|
| 2051 |
-
"epoch": 8.23,
|
| 2052 |
-
"eval_loss": NaN,
|
| 2053 |
-
"eval_runtime": 164.9214,
|
| 2054 |
-
"eval_samples_per_second": 12.127,
|
| 2055 |
-
"eval_steps_per_second": 1.516,
|
| 2056 |
-
"step": 3200
|
| 2057 |
-
},
|
| 2058 |
-
{
|
| 2059 |
-
"epoch": 8.26,
|
| 2060 |
-
"learning_rate": 6.642857142857142e-05,
|
| 2061 |
-
"loss": 0.0,
|
| 2062 |
-
"step": 3210
|
| 2063 |
-
},
|
| 2064 |
-
{
|
| 2065 |
-
"epoch": 8.28,
|
| 2066 |
-
"learning_rate": 6.563492063492062e-05,
|
| 2067 |
-
"loss": 0.0,
|
| 2068 |
-
"step": 3220
|
| 2069 |
-
},
|
| 2070 |
-
{
|
| 2071 |
-
"epoch": 8.31,
|
| 2072 |
-
"learning_rate": 6.484126984126984e-05,
|
| 2073 |
-
"loss": 0.0,
|
| 2074 |
-
"step": 3230
|
| 2075 |
-
},
|
| 2076 |
-
{
|
| 2077 |
-
"epoch": 8.33,
|
| 2078 |
-
"learning_rate": 6.404761904761904e-05,
|
| 2079 |
-
"loss": 0.0,
|
| 2080 |
-
"step": 3240
|
| 2081 |
-
},
|
| 2082 |
-
{
|
| 2083 |
-
"epoch": 8.36,
|
| 2084 |
-
"learning_rate": 6.325396825396824e-05,
|
| 2085 |
-
"loss": 0.0,
|
| 2086 |
-
"step": 3250
|
| 2087 |
-
},
|
| 2088 |
-
{
|
| 2089 |
-
"epoch": 8.39,
|
| 2090 |
-
"learning_rate": 6.246031746031746e-05,
|
| 2091 |
-
"loss": 0.0,
|
| 2092 |
-
"step": 3260
|
| 2093 |
-
},
|
| 2094 |
-
{
|
| 2095 |
-
"epoch": 8.41,
|
| 2096 |
-
"learning_rate": 6.166666666666666e-05,
|
| 2097 |
-
"loss": 0.0,
|
| 2098 |
-
"step": 3270
|
| 2099 |
-
},
|
| 2100 |
-
{
|
| 2101 |
-
"epoch": 8.44,
|
| 2102 |
-
"learning_rate": 6.0873015873015865e-05,
|
| 2103 |
-
"loss": 0.0,
|
| 2104 |
-
"step": 3280
|
| 2105 |
-
},
|
| 2106 |
-
{
|
| 2107 |
-
"epoch": 8.46,
|
| 2108 |
-
"learning_rate": 6.007936507936507e-05,
|
| 2109 |
-
"loss": 0.0,
|
| 2110 |
-
"step": 3290
|
| 2111 |
-
},
|
| 2112 |
-
{
|
| 2113 |
-
"epoch": 8.49,
|
| 2114 |
-
"learning_rate": 5.9285714285714275e-05,
|
| 2115 |
-
"loss": 0.0,
|
| 2116 |
-
"step": 3300
|
| 2117 |
-
},
|
| 2118 |
-
{
|
| 2119 |
-
"epoch": 8.51,
|
| 2120 |
-
"learning_rate": 5.849206349206348e-05,
|
| 2121 |
-
"loss": 0.0,
|
| 2122 |
-
"step": 3310
|
| 2123 |
-
},
|
| 2124 |
-
{
|
| 2125 |
-
"epoch": 8.54,
|
| 2126 |
-
"learning_rate": 5.769841269841269e-05,
|
| 2127 |
-
"loss": 0.0,
|
| 2128 |
-
"step": 3320
|
| 2129 |
-
},
|
| 2130 |
-
{
|
| 2131 |
-
"epoch": 8.57,
|
| 2132 |
-
"learning_rate": 5.69047619047619e-05,
|
| 2133 |
-
"loss": 0.0,
|
| 2134 |
-
"step": 3330
|
| 2135 |
-
},
|
| 2136 |
-
{
|
| 2137 |
-
"epoch": 8.59,
|
| 2138 |
-
"learning_rate": 5.61111111111111e-05,
|
| 2139 |
-
"loss": 0.0,
|
| 2140 |
-
"step": 3340
|
| 2141 |
-
},
|
| 2142 |
-
{
|
| 2143 |
-
"epoch": 8.62,
|
| 2144 |
-
"learning_rate": 5.531746031746031e-05,
|
| 2145 |
-
"loss": 0.0,
|
| 2146 |
-
"step": 3350
|
| 2147 |
-
},
|
| 2148 |
-
{
|
| 2149 |
-
"epoch": 8.64,
|
| 2150 |
-
"learning_rate": 5.452380952380952e-05,
|
| 2151 |
-
"loss": 0.0,
|
| 2152 |
-
"step": 3360
|
| 2153 |
-
},
|
| 2154 |
-
{
|
| 2155 |
-
"epoch": 8.67,
|
| 2156 |
-
"learning_rate": 5.3730158730158726e-05,
|
| 2157 |
-
"loss": 0.0,
|
| 2158 |
-
"step": 3370
|
| 2159 |
-
},
|
| 2160 |
-
{
|
| 2161 |
-
"epoch": 8.69,
|
| 2162 |
-
"learning_rate": 5.293650793650793e-05,
|
| 2163 |
-
"loss": 0.0,
|
| 2164 |
-
"step": 3380
|
| 2165 |
-
},
|
| 2166 |
-
{
|
| 2167 |
-
"epoch": 8.72,
|
| 2168 |
-
"learning_rate": 5.2142857142857135e-05,
|
| 2169 |
-
"loss": 0.0,
|
| 2170 |
-
"step": 3390
|
| 2171 |
-
},
|
| 2172 |
-
{
|
| 2173 |
-
"epoch": 8.75,
|
| 2174 |
-
"learning_rate": 5.1349206349206344e-05,
|
| 2175 |
-
"loss": 0.0,
|
| 2176 |
-
"step": 3400
|
| 2177 |
-
},
|
| 2178 |
-
{
|
| 2179 |
-
"epoch": 8.75,
|
| 2180 |
-
"eval_loss": NaN,
|
| 2181 |
-
"eval_runtime": 164.9331,
|
| 2182 |
-
"eval_samples_per_second": 12.126,
|
| 2183 |
-
"eval_steps_per_second": 1.516,
|
| 2184 |
-
"step": 3400
|
| 2185 |
-
},
|
| 2186 |
-
{
|
| 2187 |
-
"epoch": 8.77,
|
| 2188 |
-
"learning_rate": 5.055555555555555e-05,
|
| 2189 |
-
"loss": 0.0,
|
| 2190 |
-
"step": 3410
|
| 2191 |
-
},
|
| 2192 |
-
{
|
| 2193 |
-
"epoch": 8.8,
|
| 2194 |
-
"learning_rate": 4.976190476190475e-05,
|
| 2195 |
-
"loss": 0.0,
|
| 2196 |
-
"step": 3420
|
| 2197 |
-
},
|
| 2198 |
-
{
|
| 2199 |
-
"epoch": 8.82,
|
| 2200 |
-
"learning_rate": 4.896825396825396e-05,
|
| 2201 |
-
"loss": 0.0,
|
| 2202 |
-
"step": 3430
|
| 2203 |
-
},
|
| 2204 |
-
{
|
| 2205 |
-
"epoch": 8.85,
|
| 2206 |
-
"learning_rate": 4.817460317460317e-05,
|
| 2207 |
-
"loss": 0.0,
|
| 2208 |
-
"step": 3440
|
| 2209 |
-
},
|
| 2210 |
-
{
|
| 2211 |
-
"epoch": 8.87,
|
| 2212 |
-
"learning_rate": 4.738095238095238e-05,
|
| 2213 |
-
"loss": 0.0,
|
| 2214 |
-
"step": 3450
|
| 2215 |
-
},
|
| 2216 |
-
{
|
| 2217 |
-
"epoch": 8.9,
|
| 2218 |
-
"learning_rate": 4.658730158730158e-05,
|
| 2219 |
-
"loss": 0.0,
|
| 2220 |
-
"step": 3460
|
| 2221 |
-
},
|
| 2222 |
-
{
|
| 2223 |
-
"epoch": 8.93,
|
| 2224 |
-
"learning_rate": 4.579365079365079e-05,
|
| 2225 |
-
"loss": 0.0,
|
| 2226 |
-
"step": 3470
|
| 2227 |
-
},
|
| 2228 |
-
{
|
| 2229 |
-
"epoch": 8.95,
|
| 2230 |
-
"learning_rate": 4.4999999999999996e-05,
|
| 2231 |
-
"loss": 0.0,
|
| 2232 |
-
"step": 3480
|
| 2233 |
-
},
|
| 2234 |
-
{
|
| 2235 |
-
"epoch": 8.98,
|
| 2236 |
-
"learning_rate": 4.42063492063492e-05,
|
| 2237 |
-
"loss": 0.0,
|
| 2238 |
-
"step": 3490
|
| 2239 |
-
},
|
| 2240 |
-
{
|
| 2241 |
-
"epoch": 9.0,
|
| 2242 |
-
"learning_rate": 4.3412698412698406e-05,
|
| 2243 |
-
"loss": 0.0,
|
| 2244 |
-
"step": 3500
|
| 2245 |
-
},
|
| 2246 |
-
{
|
| 2247 |
-
"epoch": 9.03,
|
| 2248 |
-
"learning_rate": 4.2619047619047614e-05,
|
| 2249 |
-
"loss": 0.0,
|
| 2250 |
-
"step": 3510
|
| 2251 |
-
},
|
| 2252 |
-
{
|
| 2253 |
-
"epoch": 9.05,
|
| 2254 |
-
"learning_rate": 4.182539682539682e-05,
|
| 2255 |
-
"loss": 0.0,
|
| 2256 |
-
"step": 3520
|
| 2257 |
-
},
|
| 2258 |
-
{
|
| 2259 |
-
"epoch": 9.08,
|
| 2260 |
-
"learning_rate": 4.1031746031746024e-05,
|
| 2261 |
-
"loss": 0.0,
|
| 2262 |
-
"step": 3530
|
| 2263 |
-
},
|
| 2264 |
-
{
|
| 2265 |
-
"epoch": 9.11,
|
| 2266 |
-
"learning_rate": 4.023809523809523e-05,
|
| 2267 |
-
"loss": 0.0,
|
| 2268 |
-
"step": 3540
|
| 2269 |
-
},
|
| 2270 |
-
{
|
| 2271 |
-
"epoch": 9.13,
|
| 2272 |
-
"learning_rate": 3.944444444444444e-05,
|
| 2273 |
-
"loss": 0.0,
|
| 2274 |
-
"step": 3550
|
| 2275 |
-
},
|
| 2276 |
-
{
|
| 2277 |
-
"epoch": 9.16,
|
| 2278 |
-
"learning_rate": 3.865079365079365e-05,
|
| 2279 |
-
"loss": 0.0,
|
| 2280 |
-
"step": 3560
|
| 2281 |
-
},
|
| 2282 |
-
{
|
| 2283 |
-
"epoch": 9.18,
|
| 2284 |
-
"learning_rate": 3.785714285714285e-05,
|
| 2285 |
-
"loss": 0.0,
|
| 2286 |
-
"step": 3570
|
| 2287 |
-
},
|
| 2288 |
-
{
|
| 2289 |
-
"epoch": 9.21,
|
| 2290 |
-
"learning_rate": 3.706349206349206e-05,
|
| 2291 |
-
"loss": 0.0,
|
| 2292 |
-
"step": 3580
|
| 2293 |
-
},
|
| 2294 |
-
{
|
| 2295 |
-
"epoch": 9.23,
|
| 2296 |
-
"learning_rate": 3.6269841269841266e-05,
|
| 2297 |
-
"loss": 0.0,
|
| 2298 |
-
"step": 3590
|
| 2299 |
-
},
|
| 2300 |
-
{
|
| 2301 |
-
"epoch": 9.26,
|
| 2302 |
-
"learning_rate": 3.5476190476190475e-05,
|
| 2303 |
-
"loss": 0.0,
|
| 2304 |
-
"step": 3600
|
| 2305 |
-
},
|
| 2306 |
-
{
|
| 2307 |
-
"epoch": 9.26,
|
| 2308 |
-
"eval_loss": NaN,
|
| 2309 |
-
"eval_runtime": 164.9466,
|
| 2310 |
-
"eval_samples_per_second": 12.125,
|
| 2311 |
-
"eval_steps_per_second": 1.516,
|
| 2312 |
-
"step": 3600
|
| 2313 |
-
}
|
| 2314 |
-
],
|
| 2315 |
-
"max_steps": 3880,
|
| 2316 |
-
"num_train_epochs": 10,
|
| 2317 |
-
"total_flos": 4.1583436794093896e+18,
|
| 2318 |
-
"trial_name": null,
|
| 2319 |
-
"trial_params": null
|
| 2320 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-3600/training_args.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:3e9adb78996a536c4aa514741768e2b05cafc3e20ac4a0a0fe98e38b91109396
|
| 3 |
-
size 3899
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-3800/optimizer.pt
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:18c2f0237ccfb8afd6cd7875393f49b2693b0535a82600fa7a3ed83c4fb9d324
|
| 3 |
-
size 134433093
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-3800/pytorch_model.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:58653c0888eea93a18cfef68476391e6cf3aaabd0a866d5bd1a63232af5da325
|
| 3 |
-
size 67201357
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-3800/rng_state.pth
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:94a7728da6a7fa81be20f01daede1f4c98ccfa8b9f494eba5cd524ec56cdd102
|
| 3 |
-
size 14575
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-3800/scaler.pt
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:7fa181fa360d46feed4180ea17c8b6a4a879a9b4231c2e91aff2be20be9076cc
|
| 3 |
-
size 557
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-3800/scheduler.pt
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:34b8c336e51219d5b13963d925f0201df2c0b333a86c61eb7fe22364210b844d
|
| 3 |
-
size 627
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-3800/trainer_state.json
DELETED
|
@@ -1,2448 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_metric": 0.8923280239105225,
|
| 3 |
-
"best_model_checkpoint": "./lora-alpaca-hc8/checkpoint-1400",
|
| 4 |
-
"epoch": 9.77491961414791,
|
| 5 |
-
"global_step": 3800,
|
| 6 |
-
"is_hyper_param_search": false,
|
| 7 |
-
"is_local_process_zero": true,
|
| 8 |
-
"is_world_process_zero": true,
|
| 9 |
-
"log_history": [
|
| 10 |
-
{
|
| 11 |
-
"epoch": 0.03,
|
| 12 |
-
"learning_rate": 2.9999999999999997e-05,
|
| 13 |
-
"loss": 1.3719,
|
| 14 |
-
"step": 10
|
| 15 |
-
},
|
| 16 |
-
{
|
| 17 |
-
"epoch": 0.05,
|
| 18 |
-
"learning_rate": 5.9999999999999995e-05,
|
| 19 |
-
"loss": 1.5358,
|
| 20 |
-
"step": 20
|
| 21 |
-
},
|
| 22 |
-
{
|
| 23 |
-
"epoch": 0.08,
|
| 24 |
-
"learning_rate": 8.999999999999999e-05,
|
| 25 |
-
"loss": 1.5633,
|
| 26 |
-
"step": 30
|
| 27 |
-
},
|
| 28 |
-
{
|
| 29 |
-
"epoch": 0.1,
|
| 30 |
-
"learning_rate": 0.00011999999999999999,
|
| 31 |
-
"loss": 1.2433,
|
| 32 |
-
"step": 40
|
| 33 |
-
},
|
| 34 |
-
{
|
| 35 |
-
"epoch": 0.13,
|
| 36 |
-
"learning_rate": 0.00015,
|
| 37 |
-
"loss": 0.9843,
|
| 38 |
-
"step": 50
|
| 39 |
-
},
|
| 40 |
-
{
|
| 41 |
-
"epoch": 0.15,
|
| 42 |
-
"learning_rate": 0.00017999999999999998,
|
| 43 |
-
"loss": 1.012,
|
| 44 |
-
"step": 60
|
| 45 |
-
},
|
| 46 |
-
{
|
| 47 |
-
"epoch": 0.18,
|
| 48 |
-
"learning_rate": 0.00020999999999999998,
|
| 49 |
-
"loss": 1.0392,
|
| 50 |
-
"step": 70
|
| 51 |
-
},
|
| 52 |
-
{
|
| 53 |
-
"epoch": 0.21,
|
| 54 |
-
"learning_rate": 0.00023999999999999998,
|
| 55 |
-
"loss": 1.0541,
|
| 56 |
-
"step": 80
|
| 57 |
-
},
|
| 58 |
-
{
|
| 59 |
-
"epoch": 0.23,
|
| 60 |
-
"learning_rate": 0.00027,
|
| 61 |
-
"loss": 0.9608,
|
| 62 |
-
"step": 90
|
| 63 |
-
},
|
| 64 |
-
{
|
| 65 |
-
"epoch": 0.26,
|
| 66 |
-
"learning_rate": 0.0003,
|
| 67 |
-
"loss": 0.8697,
|
| 68 |
-
"step": 100
|
| 69 |
-
},
|
| 70 |
-
{
|
| 71 |
-
"epoch": 0.28,
|
| 72 |
-
"learning_rate": 0.00029920634920634916,
|
| 73 |
-
"loss": 0.9836,
|
| 74 |
-
"step": 110
|
| 75 |
-
},
|
| 76 |
-
{
|
| 77 |
-
"epoch": 0.31,
|
| 78 |
-
"learning_rate": 0.00029841269841269835,
|
| 79 |
-
"loss": 1.003,
|
| 80 |
-
"step": 120
|
| 81 |
-
},
|
| 82 |
-
{
|
| 83 |
-
"epoch": 0.33,
|
| 84 |
-
"learning_rate": 0.0002976190476190476,
|
| 85 |
-
"loss": 1.0052,
|
| 86 |
-
"step": 130
|
| 87 |
-
},
|
| 88 |
-
{
|
| 89 |
-
"epoch": 0.36,
|
| 90 |
-
"learning_rate": 0.0002968253968253968,
|
| 91 |
-
"loss": 0.9267,
|
| 92 |
-
"step": 140
|
| 93 |
-
},
|
| 94 |
-
{
|
| 95 |
-
"epoch": 0.39,
|
| 96 |
-
"learning_rate": 0.000296031746031746,
|
| 97 |
-
"loss": 0.8318,
|
| 98 |
-
"step": 150
|
| 99 |
-
},
|
| 100 |
-
{
|
| 101 |
-
"epoch": 0.41,
|
| 102 |
-
"learning_rate": 0.0002952380952380952,
|
| 103 |
-
"loss": 0.9594,
|
| 104 |
-
"step": 160
|
| 105 |
-
},
|
| 106 |
-
{
|
| 107 |
-
"epoch": 0.44,
|
| 108 |
-
"learning_rate": 0.00029444444444444445,
|
| 109 |
-
"loss": 0.987,
|
| 110 |
-
"step": 170
|
| 111 |
-
},
|
| 112 |
-
{
|
| 113 |
-
"epoch": 0.46,
|
| 114 |
-
"learning_rate": 0.00029365079365079364,
|
| 115 |
-
"loss": 0.9646,
|
| 116 |
-
"step": 180
|
| 117 |
-
},
|
| 118 |
-
{
|
| 119 |
-
"epoch": 0.49,
|
| 120 |
-
"learning_rate": 0.00029285714285714283,
|
| 121 |
-
"loss": 0.8501,
|
| 122 |
-
"step": 190
|
| 123 |
-
},
|
| 124 |
-
{
|
| 125 |
-
"epoch": 0.51,
|
| 126 |
-
"learning_rate": 0.000292063492063492,
|
| 127 |
-
"loss": 0.7568,
|
| 128 |
-
"step": 200
|
| 129 |
-
},
|
| 130 |
-
{
|
| 131 |
-
"epoch": 0.51,
|
| 132 |
-
"eval_loss": 0.9958714246749878,
|
| 133 |
-
"eval_runtime": 189.2223,
|
| 134 |
-
"eval_samples_per_second": 10.57,
|
| 135 |
-
"eval_steps_per_second": 1.321,
|
| 136 |
-
"step": 200
|
| 137 |
-
},
|
| 138 |
-
{
|
| 139 |
-
"epoch": 0.54,
|
| 140 |
-
"learning_rate": 0.00029126984126984126,
|
| 141 |
-
"loss": 0.949,
|
| 142 |
-
"step": 210
|
| 143 |
-
},
|
| 144 |
-
{
|
| 145 |
-
"epoch": 0.57,
|
| 146 |
-
"learning_rate": 0.00029047619047619045,
|
| 147 |
-
"loss": 0.9581,
|
| 148 |
-
"step": 220
|
| 149 |
-
},
|
| 150 |
-
{
|
| 151 |
-
"epoch": 0.59,
|
| 152 |
-
"learning_rate": 0.00028968253968253963,
|
| 153 |
-
"loss": 0.9526,
|
| 154 |
-
"step": 230
|
| 155 |
-
},
|
| 156 |
-
{
|
| 157 |
-
"epoch": 0.62,
|
| 158 |
-
"learning_rate": 0.0002888888888888888,
|
| 159 |
-
"loss": 0.847,
|
| 160 |
-
"step": 240
|
| 161 |
-
},
|
| 162 |
-
{
|
| 163 |
-
"epoch": 0.64,
|
| 164 |
-
"learning_rate": 0.00028809523809523806,
|
| 165 |
-
"loss": 0.7414,
|
| 166 |
-
"step": 250
|
| 167 |
-
},
|
| 168 |
-
{
|
| 169 |
-
"epoch": 0.67,
|
| 170 |
-
"learning_rate": 0.00028730158730158725,
|
| 171 |
-
"loss": 0.9449,
|
| 172 |
-
"step": 260
|
| 173 |
-
},
|
| 174 |
-
{
|
| 175 |
-
"epoch": 0.69,
|
| 176 |
-
"learning_rate": 0.0002865079365079365,
|
| 177 |
-
"loss": 0.9607,
|
| 178 |
-
"step": 270
|
| 179 |
-
},
|
| 180 |
-
{
|
| 181 |
-
"epoch": 0.72,
|
| 182 |
-
"learning_rate": 0.0002857142857142857,
|
| 183 |
-
"loss": 0.9456,
|
| 184 |
-
"step": 280
|
| 185 |
-
},
|
| 186 |
-
{
|
| 187 |
-
"epoch": 0.75,
|
| 188 |
-
"learning_rate": 0.0002849206349206349,
|
| 189 |
-
"loss": 0.8346,
|
| 190 |
-
"step": 290
|
| 191 |
-
},
|
| 192 |
-
{
|
| 193 |
-
"epoch": 0.77,
|
| 194 |
-
"learning_rate": 0.0002841269841269841,
|
| 195 |
-
"loss": 0.7275,
|
| 196 |
-
"step": 300
|
| 197 |
-
},
|
| 198 |
-
{
|
| 199 |
-
"epoch": 0.8,
|
| 200 |
-
"learning_rate": 0.0002833333333333333,
|
| 201 |
-
"loss": 0.9337,
|
| 202 |
-
"step": 310
|
| 203 |
-
},
|
| 204 |
-
{
|
| 205 |
-
"epoch": 0.82,
|
| 206 |
-
"learning_rate": 0.0002825396825396825,
|
| 207 |
-
"loss": 0.9466,
|
| 208 |
-
"step": 320
|
| 209 |
-
},
|
| 210 |
-
{
|
| 211 |
-
"epoch": 0.85,
|
| 212 |
-
"learning_rate": 0.00028174603174603173,
|
| 213 |
-
"loss": 0.9386,
|
| 214 |
-
"step": 330
|
| 215 |
-
},
|
| 216 |
-
{
|
| 217 |
-
"epoch": 0.87,
|
| 218 |
-
"learning_rate": 0.0002809523809523809,
|
| 219 |
-
"loss": 0.8254,
|
| 220 |
-
"step": 340
|
| 221 |
-
},
|
| 222 |
-
{
|
| 223 |
-
"epoch": 0.9,
|
| 224 |
-
"learning_rate": 0.0002801587301587301,
|
| 225 |
-
"loss": 0.723,
|
| 226 |
-
"step": 350
|
| 227 |
-
},
|
| 228 |
-
{
|
| 229 |
-
"epoch": 0.93,
|
| 230 |
-
"learning_rate": 0.00027936507936507935,
|
| 231 |
-
"loss": 0.9274,
|
| 232 |
-
"step": 360
|
| 233 |
-
},
|
| 234 |
-
{
|
| 235 |
-
"epoch": 0.95,
|
| 236 |
-
"learning_rate": 0.00027857142857142854,
|
| 237 |
-
"loss": 0.9492,
|
| 238 |
-
"step": 370
|
| 239 |
-
},
|
| 240 |
-
{
|
| 241 |
-
"epoch": 0.98,
|
| 242 |
-
"learning_rate": 0.0002777777777777778,
|
| 243 |
-
"loss": 0.871,
|
| 244 |
-
"step": 380
|
| 245 |
-
},
|
| 246 |
-
{
|
| 247 |
-
"epoch": 1.0,
|
| 248 |
-
"learning_rate": 0.00027698412698412697,
|
| 249 |
-
"loss": 0.7533,
|
| 250 |
-
"step": 390
|
| 251 |
-
},
|
| 252 |
-
{
|
| 253 |
-
"epoch": 1.03,
|
| 254 |
-
"learning_rate": 0.00027619047619047615,
|
| 255 |
-
"loss": 0.9284,
|
| 256 |
-
"step": 400
|
| 257 |
-
},
|
| 258 |
-
{
|
| 259 |
-
"epoch": 1.03,
|
| 260 |
-
"eval_loss": 0.9276881217956543,
|
| 261 |
-
"eval_runtime": 189.2181,
|
| 262 |
-
"eval_samples_per_second": 10.57,
|
| 263 |
-
"eval_steps_per_second": 1.321,
|
| 264 |
-
"step": 400
|
| 265 |
-
},
|
| 266 |
-
{
|
| 267 |
-
"epoch": 1.05,
|
| 268 |
-
"learning_rate": 0.0002753968253968254,
|
| 269 |
-
"loss": 0.9402,
|
| 270 |
-
"step": 410
|
| 271 |
-
},
|
| 272 |
-
{
|
| 273 |
-
"epoch": 1.08,
|
| 274 |
-
"learning_rate": 0.0002746031746031746,
|
| 275 |
-
"loss": 0.9188,
|
| 276 |
-
"step": 420
|
| 277 |
-
},
|
| 278 |
-
{
|
| 279 |
-
"epoch": 1.11,
|
| 280 |
-
"learning_rate": 0.00027380952380952377,
|
| 281 |
-
"loss": 0.7976,
|
| 282 |
-
"step": 430
|
| 283 |
-
},
|
| 284 |
-
{
|
| 285 |
-
"epoch": 1.13,
|
| 286 |
-
"learning_rate": 0.00027301587301587296,
|
| 287 |
-
"loss": 0.7172,
|
| 288 |
-
"step": 440
|
| 289 |
-
},
|
| 290 |
-
{
|
| 291 |
-
"epoch": 1.16,
|
| 292 |
-
"learning_rate": 0.0002722222222222222,
|
| 293 |
-
"loss": 0.9195,
|
| 294 |
-
"step": 450
|
| 295 |
-
},
|
| 296 |
-
{
|
| 297 |
-
"epoch": 1.18,
|
| 298 |
-
"learning_rate": 0.0002714285714285714,
|
| 299 |
-
"loss": 0.9426,
|
| 300 |
-
"step": 460
|
| 301 |
-
},
|
| 302 |
-
{
|
| 303 |
-
"epoch": 1.21,
|
| 304 |
-
"learning_rate": 0.00027063492063492063,
|
| 305 |
-
"loss": 0.9034,
|
| 306 |
-
"step": 470
|
| 307 |
-
},
|
| 308 |
-
{
|
| 309 |
-
"epoch": 1.23,
|
| 310 |
-
"learning_rate": 0.0002698412698412698,
|
| 311 |
-
"loss": 0.788,
|
| 312 |
-
"step": 480
|
| 313 |
-
},
|
| 314 |
-
{
|
| 315 |
-
"epoch": 1.26,
|
| 316 |
-
"learning_rate": 0.000269047619047619,
|
| 317 |
-
"loss": 0.7213,
|
| 318 |
-
"step": 490
|
| 319 |
-
},
|
| 320 |
-
{
|
| 321 |
-
"epoch": 1.29,
|
| 322 |
-
"learning_rate": 0.00026825396825396825,
|
| 323 |
-
"loss": 0.9149,
|
| 324 |
-
"step": 500
|
| 325 |
-
},
|
| 326 |
-
{
|
| 327 |
-
"epoch": 1.31,
|
| 328 |
-
"learning_rate": 0.00026746031746031744,
|
| 329 |
-
"loss": 0.9386,
|
| 330 |
-
"step": 510
|
| 331 |
-
},
|
| 332 |
-
{
|
| 333 |
-
"epoch": 1.34,
|
| 334 |
-
"learning_rate": 0.0002666666666666666,
|
| 335 |
-
"loss": 0.9099,
|
| 336 |
-
"step": 520
|
| 337 |
-
},
|
| 338 |
-
{
|
| 339 |
-
"epoch": 1.36,
|
| 340 |
-
"learning_rate": 0.00026587301587301587,
|
| 341 |
-
"loss": 0.7802,
|
| 342 |
-
"step": 530
|
| 343 |
-
},
|
| 344 |
-
{
|
| 345 |
-
"epoch": 1.39,
|
| 346 |
-
"learning_rate": 0.00026507936507936506,
|
| 347 |
-
"loss": 0.7246,
|
| 348 |
-
"step": 540
|
| 349 |
-
},
|
| 350 |
-
{
|
| 351 |
-
"epoch": 1.41,
|
| 352 |
-
"learning_rate": 0.00026428571428571424,
|
| 353 |
-
"loss": 0.9213,
|
| 354 |
-
"step": 550
|
| 355 |
-
},
|
| 356 |
-
{
|
| 357 |
-
"epoch": 1.44,
|
| 358 |
-
"learning_rate": 0.00026349206349206343,
|
| 359 |
-
"loss": 0.9348,
|
| 360 |
-
"step": 560
|
| 361 |
-
},
|
| 362 |
-
{
|
| 363 |
-
"epoch": 1.47,
|
| 364 |
-
"learning_rate": 0.0002626984126984127,
|
| 365 |
-
"loss": 0.8995,
|
| 366 |
-
"step": 570
|
| 367 |
-
},
|
| 368 |
-
{
|
| 369 |
-
"epoch": 1.49,
|
| 370 |
-
"learning_rate": 0.00026190476190476186,
|
| 371 |
-
"loss": 0.7866,
|
| 372 |
-
"step": 580
|
| 373 |
-
},
|
| 374 |
-
{
|
| 375 |
-
"epoch": 1.52,
|
| 376 |
-
"learning_rate": 0.0002611111111111111,
|
| 377 |
-
"loss": 0.7257,
|
| 378 |
-
"step": 590
|
| 379 |
-
},
|
| 380 |
-
{
|
| 381 |
-
"epoch": 1.54,
|
| 382 |
-
"learning_rate": 0.0002603174603174603,
|
| 383 |
-
"loss": 0.918,
|
| 384 |
-
"step": 600
|
| 385 |
-
},
|
| 386 |
-
{
|
| 387 |
-
"epoch": 1.54,
|
| 388 |
-
"eval_loss": 0.9236659407615662,
|
| 389 |
-
"eval_runtime": 189.1236,
|
| 390 |
-
"eval_samples_per_second": 10.575,
|
| 391 |
-
"eval_steps_per_second": 1.322,
|
| 392 |
-
"step": 600
|
| 393 |
-
},
|
| 394 |
-
{
|
| 395 |
-
"epoch": 1.57,
|
| 396 |
-
"learning_rate": 0.00025952380952380953,
|
| 397 |
-
"loss": 0.933,
|
| 398 |
-
"step": 610
|
| 399 |
-
},
|
| 400 |
-
{
|
| 401 |
-
"epoch": 1.59,
|
| 402 |
-
"learning_rate": 0.0002587301587301587,
|
| 403 |
-
"loss": 0.9085,
|
| 404 |
-
"step": 620
|
| 405 |
-
},
|
| 406 |
-
{
|
| 407 |
-
"epoch": 1.62,
|
| 408 |
-
"learning_rate": 0.0002579365079365079,
|
| 409 |
-
"loss": 0.7928,
|
| 410 |
-
"step": 630
|
| 411 |
-
},
|
| 412 |
-
{
|
| 413 |
-
"epoch": 1.65,
|
| 414 |
-
"learning_rate": 0.0002571428571428571,
|
| 415 |
-
"loss": 0.7162,
|
| 416 |
-
"step": 640
|
| 417 |
-
},
|
| 418 |
-
{
|
| 419 |
-
"epoch": 1.67,
|
| 420 |
-
"learning_rate": 0.00025634920634920634,
|
| 421 |
-
"loss": 0.9076,
|
| 422 |
-
"step": 650
|
| 423 |
-
},
|
| 424 |
-
{
|
| 425 |
-
"epoch": 1.7,
|
| 426 |
-
"learning_rate": 0.00025555555555555553,
|
| 427 |
-
"loss": 0.9345,
|
| 428 |
-
"step": 660
|
| 429 |
-
},
|
| 430 |
-
{
|
| 431 |
-
"epoch": 1.72,
|
| 432 |
-
"learning_rate": 0.0002547619047619047,
|
| 433 |
-
"loss": 0.9107,
|
| 434 |
-
"step": 670
|
| 435 |
-
},
|
| 436 |
-
{
|
| 437 |
-
"epoch": 1.75,
|
| 438 |
-
"learning_rate": 0.00025396825396825396,
|
| 439 |
-
"loss": 0.7721,
|
| 440 |
-
"step": 680
|
| 441 |
-
},
|
| 442 |
-
{
|
| 443 |
-
"epoch": 1.77,
|
| 444 |
-
"learning_rate": 0.00025317460317460315,
|
| 445 |
-
"loss": 0.7112,
|
| 446 |
-
"step": 690
|
| 447 |
-
},
|
| 448 |
-
{
|
| 449 |
-
"epoch": 1.8,
|
| 450 |
-
"learning_rate": 0.0002523809523809524,
|
| 451 |
-
"loss": 0.9118,
|
| 452 |
-
"step": 700
|
| 453 |
-
},
|
| 454 |
-
{
|
| 455 |
-
"epoch": 1.83,
|
| 456 |
-
"learning_rate": 0.0002515873015873016,
|
| 457 |
-
"loss": 0.9205,
|
| 458 |
-
"step": 710
|
| 459 |
-
},
|
| 460 |
-
{
|
| 461 |
-
"epoch": 1.85,
|
| 462 |
-
"learning_rate": 0.00025079365079365076,
|
| 463 |
-
"loss": 0.9004,
|
| 464 |
-
"step": 720
|
| 465 |
-
},
|
| 466 |
-
{
|
| 467 |
-
"epoch": 1.88,
|
| 468 |
-
"learning_rate": 0.00025,
|
| 469 |
-
"loss": 0.7741,
|
| 470 |
-
"step": 730
|
| 471 |
-
},
|
| 472 |
-
{
|
| 473 |
-
"epoch": 1.9,
|
| 474 |
-
"learning_rate": 0.0002492063492063492,
|
| 475 |
-
"loss": 0.7186,
|
| 476 |
-
"step": 740
|
| 477 |
-
},
|
| 478 |
-
{
|
| 479 |
-
"epoch": 1.93,
|
| 480 |
-
"learning_rate": 0.0002484126984126984,
|
| 481 |
-
"loss": 0.9002,
|
| 482 |
-
"step": 750
|
| 483 |
-
},
|
| 484 |
-
{
|
| 485 |
-
"epoch": 1.95,
|
| 486 |
-
"learning_rate": 0.00024761904761904757,
|
| 487 |
-
"loss": 0.9066,
|
| 488 |
-
"step": 760
|
| 489 |
-
},
|
| 490 |
-
{
|
| 491 |
-
"epoch": 1.98,
|
| 492 |
-
"learning_rate": 0.0002468253968253968,
|
| 493 |
-
"loss": 0.8127,
|
| 494 |
-
"step": 770
|
| 495 |
-
},
|
| 496 |
-
{
|
| 497 |
-
"epoch": 2.01,
|
| 498 |
-
"learning_rate": 0.000246031746031746,
|
| 499 |
-
"loss": 0.7305,
|
| 500 |
-
"step": 780
|
| 501 |
-
},
|
| 502 |
-
{
|
| 503 |
-
"epoch": 2.03,
|
| 504 |
-
"learning_rate": 0.0002452380952380952,
|
| 505 |
-
"loss": 0.8921,
|
| 506 |
-
"step": 790
|
| 507 |
-
},
|
| 508 |
-
{
|
| 509 |
-
"epoch": 2.06,
|
| 510 |
-
"learning_rate": 0.00024444444444444443,
|
| 511 |
-
"loss": 0.9178,
|
| 512 |
-
"step": 800
|
| 513 |
-
},
|
| 514 |
-
{
|
| 515 |
-
"epoch": 2.06,
|
| 516 |
-
"eval_loss": 0.901778519153595,
|
| 517 |
-
"eval_runtime": 189.2038,
|
| 518 |
-
"eval_samples_per_second": 10.571,
|
| 519 |
-
"eval_steps_per_second": 1.321,
|
| 520 |
-
"step": 800
|
| 521 |
-
},
|
| 522 |
-
{
|
| 523 |
-
"epoch": 2.08,
|
| 524 |
-
"learning_rate": 0.00024365079365079364,
|
| 525 |
-
"loss": 0.8823,
|
| 526 |
-
"step": 810
|
| 527 |
-
},
|
| 528 |
-
{
|
| 529 |
-
"epoch": 2.11,
|
| 530 |
-
"learning_rate": 0.00024285714285714283,
|
| 531 |
-
"loss": 0.7521,
|
| 532 |
-
"step": 820
|
| 533 |
-
},
|
| 534 |
-
{
|
| 535 |
-
"epoch": 2.14,
|
| 536 |
-
"learning_rate": 0.00024206349206349205,
|
| 537 |
-
"loss": 0.717,
|
| 538 |
-
"step": 830
|
| 539 |
-
},
|
| 540 |
-
{
|
| 541 |
-
"epoch": 2.16,
|
| 542 |
-
"learning_rate": 0.00024126984126984123,
|
| 543 |
-
"loss": 0.9045,
|
| 544 |
-
"step": 840
|
| 545 |
-
},
|
| 546 |
-
{
|
| 547 |
-
"epoch": 2.19,
|
| 548 |
-
"learning_rate": 0.00024047619047619048,
|
| 549 |
-
"loss": 0.9146,
|
| 550 |
-
"step": 850
|
| 551 |
-
},
|
| 552 |
-
{
|
| 553 |
-
"epoch": 2.21,
|
| 554 |
-
"learning_rate": 0.00023968253968253966,
|
| 555 |
-
"loss": 0.8678,
|
| 556 |
-
"step": 860
|
| 557 |
-
},
|
| 558 |
-
{
|
| 559 |
-
"epoch": 2.24,
|
| 560 |
-
"learning_rate": 0.00023888888888888885,
|
| 561 |
-
"loss": 0.745,
|
| 562 |
-
"step": 870
|
| 563 |
-
},
|
| 564 |
-
{
|
| 565 |
-
"epoch": 2.26,
|
| 566 |
-
"learning_rate": 0.00023809523809523807,
|
| 567 |
-
"loss": 0.7226,
|
| 568 |
-
"step": 880
|
| 569 |
-
},
|
| 570 |
-
{
|
| 571 |
-
"epoch": 2.29,
|
| 572 |
-
"learning_rate": 0.00023730158730158728,
|
| 573 |
-
"loss": 0.9002,
|
| 574 |
-
"step": 890
|
| 575 |
-
},
|
| 576 |
-
{
|
| 577 |
-
"epoch": 2.32,
|
| 578 |
-
"learning_rate": 0.0002365079365079365,
|
| 579 |
-
"loss": 0.9035,
|
| 580 |
-
"step": 900
|
| 581 |
-
},
|
| 582 |
-
{
|
| 583 |
-
"epoch": 2.34,
|
| 584 |
-
"learning_rate": 0.00023571428571428569,
|
| 585 |
-
"loss": 0.878,
|
| 586 |
-
"step": 910
|
| 587 |
-
},
|
| 588 |
-
{
|
| 589 |
-
"epoch": 2.37,
|
| 590 |
-
"learning_rate": 0.00023492063492063487,
|
| 591 |
-
"loss": 0.7397,
|
| 592 |
-
"step": 920
|
| 593 |
-
},
|
| 594 |
-
{
|
| 595 |
-
"epoch": 2.39,
|
| 596 |
-
"learning_rate": 0.00023412698412698412,
|
| 597 |
-
"loss": 0.7251,
|
| 598 |
-
"step": 930
|
| 599 |
-
},
|
| 600 |
-
{
|
| 601 |
-
"epoch": 2.42,
|
| 602 |
-
"learning_rate": 0.0002333333333333333,
|
| 603 |
-
"loss": 0.9014,
|
| 604 |
-
"step": 940
|
| 605 |
-
},
|
| 606 |
-
{
|
| 607 |
-
"epoch": 2.44,
|
| 608 |
-
"learning_rate": 0.00023253968253968252,
|
| 609 |
-
"loss": 0.9158,
|
| 610 |
-
"step": 950
|
| 611 |
-
},
|
| 612 |
-
{
|
| 613 |
-
"epoch": 2.47,
|
| 614 |
-
"learning_rate": 0.0002317460317460317,
|
| 615 |
-
"loss": 0.8596,
|
| 616 |
-
"step": 960
|
| 617 |
-
},
|
| 618 |
-
{
|
| 619 |
-
"epoch": 2.5,
|
| 620 |
-
"learning_rate": 0.00023095238095238095,
|
| 621 |
-
"loss": 0.7312,
|
| 622 |
-
"step": 970
|
| 623 |
-
},
|
| 624 |
-
{
|
| 625 |
-
"epoch": 2.52,
|
| 626 |
-
"learning_rate": 0.00023015873015873014,
|
| 627 |
-
"loss": 0.7271,
|
| 628 |
-
"step": 980
|
| 629 |
-
},
|
| 630 |
-
{
|
| 631 |
-
"epoch": 2.55,
|
| 632 |
-
"learning_rate": 0.00022936507936507935,
|
| 633 |
-
"loss": 0.9007,
|
| 634 |
-
"step": 990
|
| 635 |
-
},
|
| 636 |
-
{
|
| 637 |
-
"epoch": 2.57,
|
| 638 |
-
"learning_rate": 0.00022857142857142854,
|
| 639 |
-
"loss": 0.9186,
|
| 640 |
-
"step": 1000
|
| 641 |
-
},
|
| 642 |
-
{
|
| 643 |
-
"epoch": 2.57,
|
| 644 |
-
"eval_loss": 0.8995742201805115,
|
| 645 |
-
"eval_runtime": 189.2401,
|
| 646 |
-
"eval_samples_per_second": 10.569,
|
| 647 |
-
"eval_steps_per_second": 1.321,
|
| 648 |
-
"step": 1000
|
| 649 |
-
},
|
| 650 |
-
{
|
| 651 |
-
"epoch": 2.6,
|
| 652 |
-
"learning_rate": 0.00022777777777777778,
|
| 653 |
-
"loss": 0.8685,
|
| 654 |
-
"step": 1010
|
| 655 |
-
},
|
| 656 |
-
{
|
| 657 |
-
"epoch": 2.62,
|
| 658 |
-
"learning_rate": 0.00022698412698412697,
|
| 659 |
-
"loss": 0.7359,
|
| 660 |
-
"step": 1020
|
| 661 |
-
},
|
| 662 |
-
{
|
| 663 |
-
"epoch": 2.65,
|
| 664 |
-
"learning_rate": 0.00022619047619047616,
|
| 665 |
-
"loss": 0.7166,
|
| 666 |
-
"step": 1030
|
| 667 |
-
},
|
| 668 |
-
{
|
| 669 |
-
"epoch": 2.68,
|
| 670 |
-
"learning_rate": 0.00022539682539682537,
|
| 671 |
-
"loss": 0.9012,
|
| 672 |
-
"step": 1040
|
| 673 |
-
},
|
| 674 |
-
{
|
| 675 |
-
"epoch": 2.7,
|
| 676 |
-
"learning_rate": 0.0002246031746031746,
|
| 677 |
-
"loss": 0.9195,
|
| 678 |
-
"step": 1050
|
| 679 |
-
},
|
| 680 |
-
{
|
| 681 |
-
"epoch": 2.73,
|
| 682 |
-
"learning_rate": 0.0002238095238095238,
|
| 683 |
-
"loss": 0.8733,
|
| 684 |
-
"step": 1060
|
| 685 |
-
},
|
| 686 |
-
{
|
| 687 |
-
"epoch": 2.75,
|
| 688 |
-
"learning_rate": 0.000223015873015873,
|
| 689 |
-
"loss": 0.7488,
|
| 690 |
-
"step": 1070
|
| 691 |
-
},
|
| 692 |
-
{
|
| 693 |
-
"epoch": 2.78,
|
| 694 |
-
"learning_rate": 0.00022222222222222218,
|
| 695 |
-
"loss": 0.7223,
|
| 696 |
-
"step": 1080
|
| 697 |
-
},
|
| 698 |
-
{
|
| 699 |
-
"epoch": 2.8,
|
| 700 |
-
"learning_rate": 0.00022142857142857142,
|
| 701 |
-
"loss": 0.9034,
|
| 702 |
-
"step": 1090
|
| 703 |
-
},
|
| 704 |
-
{
|
| 705 |
-
"epoch": 2.83,
|
| 706 |
-
"learning_rate": 0.0002206349206349206,
|
| 707 |
-
"loss": 0.9174,
|
| 708 |
-
"step": 1100
|
| 709 |
-
},
|
| 710 |
-
{
|
| 711 |
-
"epoch": 2.86,
|
| 712 |
-
"learning_rate": 0.00021984126984126982,
|
| 713 |
-
"loss": 0.868,
|
| 714 |
-
"step": 1110
|
| 715 |
-
},
|
| 716 |
-
{
|
| 717 |
-
"epoch": 2.88,
|
| 718 |
-
"learning_rate": 0.000219047619047619,
|
| 719 |
-
"loss": 0.7394,
|
| 720 |
-
"step": 1120
|
| 721 |
-
},
|
| 722 |
-
{
|
| 723 |
-
"epoch": 2.91,
|
| 724 |
-
"learning_rate": 0.00021825396825396825,
|
| 725 |
-
"loss": 0.7198,
|
| 726 |
-
"step": 1130
|
| 727 |
-
},
|
| 728 |
-
{
|
| 729 |
-
"epoch": 2.93,
|
| 730 |
-
"learning_rate": 0.00021746031746031744,
|
| 731 |
-
"loss": 0.9085,
|
| 732 |
-
"step": 1140
|
| 733 |
-
},
|
| 734 |
-
{
|
| 735 |
-
"epoch": 2.96,
|
| 736 |
-
"learning_rate": 0.00021666666666666666,
|
| 737 |
-
"loss": 0.9148,
|
| 738 |
-
"step": 1150
|
| 739 |
-
},
|
| 740 |
-
{
|
| 741 |
-
"epoch": 2.98,
|
| 742 |
-
"learning_rate": 0.00021587301587301584,
|
| 743 |
-
"loss": 0.7775,
|
| 744 |
-
"step": 1160
|
| 745 |
-
},
|
| 746 |
-
{
|
| 747 |
-
"epoch": 3.01,
|
| 748 |
-
"learning_rate": 0.0002150793650793651,
|
| 749 |
-
"loss": 0.7366,
|
| 750 |
-
"step": 1170
|
| 751 |
-
},
|
| 752 |
-
{
|
| 753 |
-
"epoch": 3.04,
|
| 754 |
-
"learning_rate": 0.00021428571428571427,
|
| 755 |
-
"loss": 0.8865,
|
| 756 |
-
"step": 1180
|
| 757 |
-
},
|
| 758 |
-
{
|
| 759 |
-
"epoch": 3.06,
|
| 760 |
-
"learning_rate": 0.00021349206349206346,
|
| 761 |
-
"loss": 0.8989,
|
| 762 |
-
"step": 1190
|
| 763 |
-
},
|
| 764 |
-
{
|
| 765 |
-
"epoch": 3.09,
|
| 766 |
-
"learning_rate": 0.00021269841269841268,
|
| 767 |
-
"loss": 0.8376,
|
| 768 |
-
"step": 1200
|
| 769 |
-
},
|
| 770 |
-
{
|
| 771 |
-
"epoch": 3.09,
|
| 772 |
-
"eval_loss": 0.8965018391609192,
|
| 773 |
-
"eval_runtime": 189.2693,
|
| 774 |
-
"eval_samples_per_second": 10.567,
|
| 775 |
-
"eval_steps_per_second": 1.321,
|
| 776 |
-
"step": 1200
|
| 777 |
-
},
|
| 778 |
-
{
|
| 779 |
-
"epoch": 3.11,
|
| 780 |
-
"learning_rate": 0.0002119047619047619,
|
| 781 |
-
"loss": 0.7012,
|
| 782 |
-
"step": 1210
|
| 783 |
-
},
|
| 784 |
-
{
|
| 785 |
-
"epoch": 3.14,
|
| 786 |
-
"learning_rate": 0.0002111111111111111,
|
| 787 |
-
"loss": 0.7288,
|
| 788 |
-
"step": 1220
|
| 789 |
-
},
|
| 790 |
-
{
|
| 791 |
-
"epoch": 3.16,
|
| 792 |
-
"learning_rate": 0.0002103174603174603,
|
| 793 |
-
"loss": 0.8904,
|
| 794 |
-
"step": 1230
|
| 795 |
-
},
|
| 796 |
-
{
|
| 797 |
-
"epoch": 3.19,
|
| 798 |
-
"learning_rate": 0.00020952380952380948,
|
| 799 |
-
"loss": 0.9081,
|
| 800 |
-
"step": 1240
|
| 801 |
-
},
|
| 802 |
-
{
|
| 803 |
-
"epoch": 3.22,
|
| 804 |
-
"learning_rate": 0.00020873015873015873,
|
| 805 |
-
"loss": 0.8461,
|
| 806 |
-
"step": 1250
|
| 807 |
-
},
|
| 808 |
-
{
|
| 809 |
-
"epoch": 3.24,
|
| 810 |
-
"learning_rate": 0.00020793650793650791,
|
| 811 |
-
"loss": 0.6997,
|
| 812 |
-
"step": 1260
|
| 813 |
-
},
|
| 814 |
-
{
|
| 815 |
-
"epoch": 3.27,
|
| 816 |
-
"learning_rate": 0.00020714285714285713,
|
| 817 |
-
"loss": 0.7189,
|
| 818 |
-
"step": 1270
|
| 819 |
-
},
|
| 820 |
-
{
|
| 821 |
-
"epoch": 3.29,
|
| 822 |
-
"learning_rate": 0.00020634920634920632,
|
| 823 |
-
"loss": 0.8863,
|
| 824 |
-
"step": 1280
|
| 825 |
-
},
|
| 826 |
-
{
|
| 827 |
-
"epoch": 3.32,
|
| 828 |
-
"learning_rate": 0.00020555555555555556,
|
| 829 |
-
"loss": 0.906,
|
| 830 |
-
"step": 1290
|
| 831 |
-
},
|
| 832 |
-
{
|
| 833 |
-
"epoch": 3.34,
|
| 834 |
-
"learning_rate": 0.00020476190476190475,
|
| 835 |
-
"loss": 0.8287,
|
| 836 |
-
"step": 1300
|
| 837 |
-
},
|
| 838 |
-
{
|
| 839 |
-
"epoch": 3.37,
|
| 840 |
-
"learning_rate": 0.00020396825396825393,
|
| 841 |
-
"loss": 0.7015,
|
| 842 |
-
"step": 1310
|
| 843 |
-
},
|
| 844 |
-
{
|
| 845 |
-
"epoch": 3.4,
|
| 846 |
-
"learning_rate": 0.00020317460317460315,
|
| 847 |
-
"loss": 0.7325,
|
| 848 |
-
"step": 1320
|
| 849 |
-
},
|
| 850 |
-
{
|
| 851 |
-
"epoch": 3.42,
|
| 852 |
-
"learning_rate": 0.00020238095238095236,
|
| 853 |
-
"loss": 0.8878,
|
| 854 |
-
"step": 1330
|
| 855 |
-
},
|
| 856 |
-
{
|
| 857 |
-
"epoch": 3.45,
|
| 858 |
-
"learning_rate": 0.00020158730158730158,
|
| 859 |
-
"loss": 0.9057,
|
| 860 |
-
"step": 1340
|
| 861 |
-
},
|
| 862 |
-
{
|
| 863 |
-
"epoch": 3.47,
|
| 864 |
-
"learning_rate": 0.00020079365079365077,
|
| 865 |
-
"loss": 0.8399,
|
| 866 |
-
"step": 1350
|
| 867 |
-
},
|
| 868 |
-
{
|
| 869 |
-
"epoch": 3.5,
|
| 870 |
-
"learning_rate": 0.00019999999999999998,
|
| 871 |
-
"loss": 0.7073,
|
| 872 |
-
"step": 1360
|
| 873 |
-
},
|
| 874 |
-
{
|
| 875 |
-
"epoch": 3.52,
|
| 876 |
-
"learning_rate": 0.0001992063492063492,
|
| 877 |
-
"loss": 0.7281,
|
| 878 |
-
"step": 1370
|
| 879 |
-
},
|
| 880 |
-
{
|
| 881 |
-
"epoch": 3.55,
|
| 882 |
-
"learning_rate": 0.0001984126984126984,
|
| 883 |
-
"loss": 0.8829,
|
| 884 |
-
"step": 1380
|
| 885 |
-
},
|
| 886 |
-
{
|
| 887 |
-
"epoch": 3.58,
|
| 888 |
-
"learning_rate": 0.0001976190476190476,
|
| 889 |
-
"loss": 0.8923,
|
| 890 |
-
"step": 1390
|
| 891 |
-
},
|
| 892 |
-
{
|
| 893 |
-
"epoch": 3.6,
|
| 894 |
-
"learning_rate": 0.0001968253968253968,
|
| 895 |
-
"loss": 0.8389,
|
| 896 |
-
"step": 1400
|
| 897 |
-
},
|
| 898 |
-
{
|
| 899 |
-
"epoch": 3.6,
|
| 900 |
-
"eval_loss": 0.8923280239105225,
|
| 901 |
-
"eval_runtime": 189.1693,
|
| 902 |
-
"eval_samples_per_second": 10.573,
|
| 903 |
-
"eval_steps_per_second": 1.322,
|
| 904 |
-
"step": 1400
|
| 905 |
-
},
|
| 906 |
-
{
|
| 907 |
-
"epoch": 3.63,
|
| 908 |
-
"learning_rate": 0.00019603174603174603,
|
| 909 |
-
"loss": 0.7148,
|
| 910 |
-
"step": 1410
|
| 911 |
-
},
|
| 912 |
-
{
|
| 913 |
-
"epoch": 3.65,
|
| 914 |
-
"learning_rate": 0.00019523809523809522,
|
| 915 |
-
"loss": 0.7331,
|
| 916 |
-
"step": 1420
|
| 917 |
-
},
|
| 918 |
-
{
|
| 919 |
-
"epoch": 3.68,
|
| 920 |
-
"learning_rate": 0.00019444444444444443,
|
| 921 |
-
"loss": 0.8944,
|
| 922 |
-
"step": 1430
|
| 923 |
-
},
|
| 924 |
-
{
|
| 925 |
-
"epoch": 3.7,
|
| 926 |
-
"learning_rate": 0.00019365079365079362,
|
| 927 |
-
"loss": 0.9001,
|
| 928 |
-
"step": 1440
|
| 929 |
-
},
|
| 930 |
-
{
|
| 931 |
-
"epoch": 3.73,
|
| 932 |
-
"learning_rate": 0.00019285714285714286,
|
| 933 |
-
"loss": 0.8397,
|
| 934 |
-
"step": 1450
|
| 935 |
-
},
|
| 936 |
-
{
|
| 937 |
-
"epoch": 3.76,
|
| 938 |
-
"learning_rate": 0.00019206349206349205,
|
| 939 |
-
"loss": 0.6953,
|
| 940 |
-
"step": 1460
|
| 941 |
-
},
|
| 942 |
-
{
|
| 943 |
-
"epoch": 3.78,
|
| 944 |
-
"learning_rate": 0.00019126984126984124,
|
| 945 |
-
"loss": 0.7229,
|
| 946 |
-
"step": 1470
|
| 947 |
-
},
|
| 948 |
-
{
|
| 949 |
-
"epoch": 3.81,
|
| 950 |
-
"learning_rate": 0.00019047619047619045,
|
| 951 |
-
"loss": 0.8833,
|
| 952 |
-
"step": 1480
|
| 953 |
-
},
|
| 954 |
-
{
|
| 955 |
-
"epoch": 3.83,
|
| 956 |
-
"learning_rate": 0.00018968253968253967,
|
| 957 |
-
"loss": 0.9028,
|
| 958 |
-
"step": 1490
|
| 959 |
-
},
|
| 960 |
-
{
|
| 961 |
-
"epoch": 3.86,
|
| 962 |
-
"learning_rate": 0.00018888888888888888,
|
| 963 |
-
"loss": 0.8412,
|
| 964 |
-
"step": 1500
|
| 965 |
-
},
|
| 966 |
-
{
|
| 967 |
-
"epoch": 3.88,
|
| 968 |
-
"learning_rate": 0.00018809523809523807,
|
| 969 |
-
"loss": 0.7024,
|
| 970 |
-
"step": 1510
|
| 971 |
-
},
|
| 972 |
-
{
|
| 973 |
-
"epoch": 3.91,
|
| 974 |
-
"learning_rate": 0.0001873015873015873,
|
| 975 |
-
"loss": 0.7262,
|
| 976 |
-
"step": 1520
|
| 977 |
-
},
|
| 978 |
-
{
|
| 979 |
-
"epoch": 3.94,
|
| 980 |
-
"learning_rate": 0.0001865079365079365,
|
| 981 |
-
"loss": 0.8926,
|
| 982 |
-
"step": 1530
|
| 983 |
-
},
|
| 984 |
-
{
|
| 985 |
-
"epoch": 3.96,
|
| 986 |
-
"learning_rate": 0.00018571428571428572,
|
| 987 |
-
"loss": 0.8703,
|
| 988 |
-
"step": 1540
|
| 989 |
-
},
|
| 990 |
-
{
|
| 991 |
-
"epoch": 3.99,
|
| 992 |
-
"learning_rate": 0.0001849206349206349,
|
| 993 |
-
"loss": 0.7311,
|
| 994 |
-
"step": 1550
|
| 995 |
-
},
|
| 996 |
-
{
|
| 997 |
-
"epoch": 4.01,
|
| 998 |
-
"learning_rate": 0.0001841269841269841,
|
| 999 |
-
"loss": 0.7535,
|
| 1000 |
-
"step": 1560
|
| 1001 |
-
},
|
| 1002 |
-
{
|
| 1003 |
-
"epoch": 4.04,
|
| 1004 |
-
"learning_rate": 0.00018333333333333334,
|
| 1005 |
-
"loss": 0.8779,
|
| 1006 |
-
"step": 1570
|
| 1007 |
-
},
|
| 1008 |
-
{
|
| 1009 |
-
"epoch": 4.06,
|
| 1010 |
-
"learning_rate": 0.00018253968253968252,
|
| 1011 |
-
"loss": 0.887,
|
| 1012 |
-
"step": 1580
|
| 1013 |
-
},
|
| 1014 |
-
{
|
| 1015 |
-
"epoch": 4.09,
|
| 1016 |
-
"learning_rate": 0.00018174603174603174,
|
| 1017 |
-
"loss": 0.797,
|
| 1018 |
-
"step": 1590
|
| 1019 |
-
},
|
| 1020 |
-
{
|
| 1021 |
-
"epoch": 4.12,
|
| 1022 |
-
"learning_rate": 0.00018095238095238093,
|
| 1023 |
-
"loss": 0.6651,
|
| 1024 |
-
"step": 1600
|
| 1025 |
-
},
|
| 1026 |
-
{
|
| 1027 |
-
"epoch": 4.12,
|
| 1028 |
-
"eval_loss": 0.904344916343689,
|
| 1029 |
-
"eval_runtime": 189.1226,
|
| 1030 |
-
"eval_samples_per_second": 10.575,
|
| 1031 |
-
"eval_steps_per_second": 1.322,
|
| 1032 |
-
"step": 1600
|
| 1033 |
-
},
|
| 1034 |
-
{
|
| 1035 |
-
"epoch": 4.14,
|
| 1036 |
-
"learning_rate": 0.00018015873015873017,
|
| 1037 |
-
"loss": 0.7348,
|
| 1038 |
-
"step": 1610
|
| 1039 |
-
},
|
| 1040 |
-
{
|
| 1041 |
-
"epoch": 4.17,
|
| 1042 |
-
"learning_rate": 0.00017936507936507936,
|
| 1043 |
-
"loss": 0.8756,
|
| 1044 |
-
"step": 1620
|
| 1045 |
-
},
|
| 1046 |
-
{
|
| 1047 |
-
"epoch": 4.19,
|
| 1048 |
-
"learning_rate": 0.00017857142857142854,
|
| 1049 |
-
"loss": 0.8934,
|
| 1050 |
-
"step": 1630
|
| 1051 |
-
},
|
| 1052 |
-
{
|
| 1053 |
-
"epoch": 4.22,
|
| 1054 |
-
"learning_rate": 0.00017777777777777776,
|
| 1055 |
-
"loss": 0.8023,
|
| 1056 |
-
"step": 1640
|
| 1057 |
-
},
|
| 1058 |
-
{
|
| 1059 |
-
"epoch": 4.24,
|
| 1060 |
-
"learning_rate": 0.00017698412698412697,
|
| 1061 |
-
"loss": 0.6788,
|
| 1062 |
-
"step": 1650
|
| 1063 |
-
},
|
| 1064 |
-
{
|
| 1065 |
-
"epoch": 4.27,
|
| 1066 |
-
"learning_rate": 0.0001761904761904762,
|
| 1067 |
-
"loss": 0.7387,
|
| 1068 |
-
"step": 1660
|
| 1069 |
-
},
|
| 1070 |
-
{
|
| 1071 |
-
"epoch": 4.3,
|
| 1072 |
-
"learning_rate": 0.00017539682539682538,
|
| 1073 |
-
"loss": 0.885,
|
| 1074 |
-
"step": 1670
|
| 1075 |
-
},
|
| 1076 |
-
{
|
| 1077 |
-
"epoch": 4.32,
|
| 1078 |
-
"learning_rate": 0.00017460317460317457,
|
| 1079 |
-
"loss": 0.8738,
|
| 1080 |
-
"step": 1680
|
| 1081 |
-
},
|
| 1082 |
-
{
|
| 1083 |
-
"epoch": 4.35,
|
| 1084 |
-
"learning_rate": 0.0001738095238095238,
|
| 1085 |
-
"loss": 0.8059,
|
| 1086 |
-
"step": 1690
|
| 1087 |
-
},
|
| 1088 |
-
{
|
| 1089 |
-
"epoch": 4.37,
|
| 1090 |
-
"learning_rate": 0.000173015873015873,
|
| 1091 |
-
"loss": 0.6667,
|
| 1092 |
-
"step": 1700
|
| 1093 |
-
},
|
| 1094 |
-
{
|
| 1095 |
-
"epoch": 4.4,
|
| 1096 |
-
"learning_rate": 0.0001722222222222222,
|
| 1097 |
-
"loss": 0.7406,
|
| 1098 |
-
"step": 1710
|
| 1099 |
-
},
|
| 1100 |
-
{
|
| 1101 |
-
"epoch": 4.42,
|
| 1102 |
-
"learning_rate": 0.0001714285714285714,
|
| 1103 |
-
"loss": 0.8764,
|
| 1104 |
-
"step": 1720
|
| 1105 |
-
},
|
| 1106 |
-
{
|
| 1107 |
-
"epoch": 4.45,
|
| 1108 |
-
"learning_rate": 0.00017063492063492064,
|
| 1109 |
-
"loss": 0.8839,
|
| 1110 |
-
"step": 1730
|
| 1111 |
-
},
|
| 1112 |
-
{
|
| 1113 |
-
"epoch": 4.48,
|
| 1114 |
-
"learning_rate": 0.00016984126984126983,
|
| 1115 |
-
"loss": 0.8009,
|
| 1116 |
-
"step": 1740
|
| 1117 |
-
},
|
| 1118 |
-
{
|
| 1119 |
-
"epoch": 4.5,
|
| 1120 |
-
"learning_rate": 0.00016904761904761904,
|
| 1121 |
-
"loss": 0.6658,
|
| 1122 |
-
"step": 1750
|
| 1123 |
-
},
|
| 1124 |
-
{
|
| 1125 |
-
"epoch": 4.53,
|
| 1126 |
-
"learning_rate": 0.00016825396825396823,
|
| 1127 |
-
"loss": 0.7423,
|
| 1128 |
-
"step": 1760
|
| 1129 |
-
},
|
| 1130 |
-
{
|
| 1131 |
-
"epoch": 4.55,
|
| 1132 |
-
"learning_rate": 0.00016746031746031747,
|
| 1133 |
-
"loss": 0.8748,
|
| 1134 |
-
"step": 1770
|
| 1135 |
-
},
|
| 1136 |
-
{
|
| 1137 |
-
"epoch": 4.58,
|
| 1138 |
-
"learning_rate": 0.00016666666666666666,
|
| 1139 |
-
"loss": 0.887,
|
| 1140 |
-
"step": 1780
|
| 1141 |
-
},
|
| 1142 |
-
{
|
| 1143 |
-
"epoch": 4.6,
|
| 1144 |
-
"learning_rate": 0.00016587301587301585,
|
| 1145 |
-
"loss": 0.8038,
|
| 1146 |
-
"step": 1790
|
| 1147 |
-
},
|
| 1148 |
-
{
|
| 1149 |
-
"epoch": 4.63,
|
| 1150 |
-
"learning_rate": 0.00016507936507936506,
|
| 1151 |
-
"loss": 0.6631,
|
| 1152 |
-
"step": 1800
|
| 1153 |
-
},
|
| 1154 |
-
{
|
| 1155 |
-
"epoch": 4.63,
|
| 1156 |
-
"eval_loss": 0.9004252552986145,
|
| 1157 |
-
"eval_runtime": 189.1263,
|
| 1158 |
-
"eval_samples_per_second": 10.575,
|
| 1159 |
-
"eval_steps_per_second": 1.322,
|
| 1160 |
-
"step": 1800
|
| 1161 |
-
},
|
| 1162 |
-
{
|
| 1163 |
-
"epoch": 4.66,
|
| 1164 |
-
"learning_rate": 0.00016428571428571428,
|
| 1165 |
-
"loss": 0.7327,
|
| 1166 |
-
"step": 1810
|
| 1167 |
-
},
|
| 1168 |
-
{
|
| 1169 |
-
"epoch": 4.68,
|
| 1170 |
-
"learning_rate": 0.0001634920634920635,
|
| 1171 |
-
"loss": 0.8703,
|
| 1172 |
-
"step": 1820
|
| 1173 |
-
},
|
| 1174 |
-
{
|
| 1175 |
-
"epoch": 4.71,
|
| 1176 |
-
"learning_rate": 0.00016269841269841268,
|
| 1177 |
-
"loss": 0.8734,
|
| 1178 |
-
"step": 1830
|
| 1179 |
-
},
|
| 1180 |
-
{
|
| 1181 |
-
"epoch": 4.73,
|
| 1182 |
-
"learning_rate": 0.00016190476190476187,
|
| 1183 |
-
"loss": 0.8066,
|
| 1184 |
-
"step": 1840
|
| 1185 |
-
},
|
| 1186 |
-
{
|
| 1187 |
-
"epoch": 4.76,
|
| 1188 |
-
"learning_rate": 0.0001611111111111111,
|
| 1189 |
-
"loss": 0.6655,
|
| 1190 |
-
"step": 1850
|
| 1191 |
-
},
|
| 1192 |
-
{
|
| 1193 |
-
"epoch": 4.78,
|
| 1194 |
-
"learning_rate": 0.0001603174603174603,
|
| 1195 |
-
"loss": 0.736,
|
| 1196 |
-
"step": 1860
|
| 1197 |
-
},
|
| 1198 |
-
{
|
| 1199 |
-
"epoch": 4.81,
|
| 1200 |
-
"learning_rate": 0.00015952380952380951,
|
| 1201 |
-
"loss": 0.8707,
|
| 1202 |
-
"step": 1870
|
| 1203 |
-
},
|
| 1204 |
-
{
|
| 1205 |
-
"epoch": 4.84,
|
| 1206 |
-
"learning_rate": 0.0001587301587301587,
|
| 1207 |
-
"loss": 0.8789,
|
| 1208 |
-
"step": 1880
|
| 1209 |
-
},
|
| 1210 |
-
{
|
| 1211 |
-
"epoch": 4.86,
|
| 1212 |
-
"learning_rate": 0.00015793650793650795,
|
| 1213 |
-
"loss": 0.804,
|
| 1214 |
-
"step": 1890
|
| 1215 |
-
},
|
| 1216 |
-
{
|
| 1217 |
-
"epoch": 4.89,
|
| 1218 |
-
"learning_rate": 0.00015714285714285713,
|
| 1219 |
-
"loss": 0.6628,
|
| 1220 |
-
"step": 1900
|
| 1221 |
-
},
|
| 1222 |
-
{
|
| 1223 |
-
"epoch": 4.91,
|
| 1224 |
-
"learning_rate": 0.00015634920634920635,
|
| 1225 |
-
"loss": 0.7394,
|
| 1226 |
-
"step": 1910
|
| 1227 |
-
},
|
| 1228 |
-
{
|
| 1229 |
-
"epoch": 4.94,
|
| 1230 |
-
"learning_rate": 0.00015555555555555554,
|
| 1231 |
-
"loss": 0.8856,
|
| 1232 |
-
"step": 1920
|
| 1233 |
-
},
|
| 1234 |
-
{
|
| 1235 |
-
"epoch": 4.96,
|
| 1236 |
-
"learning_rate": 0.00015476190476190478,
|
| 1237 |
-
"loss": 0.8558,
|
| 1238 |
-
"step": 1930
|
| 1239 |
-
},
|
| 1240 |
-
{
|
| 1241 |
-
"epoch": 4.99,
|
| 1242 |
-
"learning_rate": 0.00015396825396825397,
|
| 1243 |
-
"loss": 0.6766,
|
| 1244 |
-
"step": 1940
|
| 1245 |
-
},
|
| 1246 |
-
{
|
| 1247 |
-
"epoch": 5.02,
|
| 1248 |
-
"learning_rate": 0.00015317460317460315,
|
| 1249 |
-
"loss": 0.7662,
|
| 1250 |
-
"step": 1950
|
| 1251 |
-
},
|
| 1252 |
-
{
|
| 1253 |
-
"epoch": 5.04,
|
| 1254 |
-
"learning_rate": 0.00015238095238095237,
|
| 1255 |
-
"loss": 0.8602,
|
| 1256 |
-
"step": 1960
|
| 1257 |
-
},
|
| 1258 |
-
{
|
| 1259 |
-
"epoch": 5.07,
|
| 1260 |
-
"learning_rate": 0.00015158730158730158,
|
| 1261 |
-
"loss": 0.8567,
|
| 1262 |
-
"step": 1970
|
| 1263 |
-
},
|
| 1264 |
-
{
|
| 1265 |
-
"epoch": 5.09,
|
| 1266 |
-
"learning_rate": 0.0001507936507936508,
|
| 1267 |
-
"loss": 0.7482,
|
| 1268 |
-
"step": 1980
|
| 1269 |
-
},
|
| 1270 |
-
{
|
| 1271 |
-
"epoch": 5.12,
|
| 1272 |
-
"learning_rate": 0.00015,
|
| 1273 |
-
"loss": 0.6249,
|
| 1274 |
-
"step": 1990
|
| 1275 |
-
},
|
| 1276 |
-
{
|
| 1277 |
-
"epoch": 5.14,
|
| 1278 |
-
"learning_rate": 0.00014920634920634917,
|
| 1279 |
-
"loss": 0.7506,
|
| 1280 |
-
"step": 2000
|
| 1281 |
-
},
|
| 1282 |
-
{
|
| 1283 |
-
"epoch": 5.14,
|
| 1284 |
-
"eval_loss": 0.9112463593482971,
|
| 1285 |
-
"eval_runtime": 189.2232,
|
| 1286 |
-
"eval_samples_per_second": 10.57,
|
| 1287 |
-
"eval_steps_per_second": 1.321,
|
| 1288 |
-
"step": 2000
|
| 1289 |
-
},
|
| 1290 |
-
{
|
| 1291 |
-
"epoch": 5.17,
|
| 1292 |
-
"learning_rate": 0.0001484126984126984,
|
| 1293 |
-
"loss": 0.8679,
|
| 1294 |
-
"step": 2010
|
| 1295 |
-
},
|
| 1296 |
-
{
|
| 1297 |
-
"epoch": 5.2,
|
| 1298 |
-
"learning_rate": 0.0001476190476190476,
|
| 1299 |
-
"loss": 0.8575,
|
| 1300 |
-
"step": 2020
|
| 1301 |
-
},
|
| 1302 |
-
{
|
| 1303 |
-
"epoch": 5.22,
|
| 1304 |
-
"learning_rate": 0.00014682539682539682,
|
| 1305 |
-
"loss": 0.7545,
|
| 1306 |
-
"step": 2030
|
| 1307 |
-
},
|
| 1308 |
-
{
|
| 1309 |
-
"epoch": 5.25,
|
| 1310 |
-
"learning_rate": 0.000146031746031746,
|
| 1311 |
-
"loss": 0.6237,
|
| 1312 |
-
"step": 2040
|
| 1313 |
-
},
|
| 1314 |
-
{
|
| 1315 |
-
"epoch": 5.27,
|
| 1316 |
-
"learning_rate": 0.00014523809523809522,
|
| 1317 |
-
"loss": 0.7561,
|
| 1318 |
-
"step": 2050
|
| 1319 |
-
},
|
| 1320 |
-
{
|
| 1321 |
-
"epoch": 5.3,
|
| 1322 |
-
"learning_rate": 0.0001444444444444444,
|
| 1323 |
-
"loss": 0.8688,
|
| 1324 |
-
"step": 2060
|
| 1325 |
-
},
|
| 1326 |
-
{
|
| 1327 |
-
"epoch": 5.32,
|
| 1328 |
-
"learning_rate": 0.00014365079365079363,
|
| 1329 |
-
"loss": 0.8578,
|
| 1330 |
-
"step": 2070
|
| 1331 |
-
},
|
| 1332 |
-
{
|
| 1333 |
-
"epoch": 5.35,
|
| 1334 |
-
"learning_rate": 0.00014285714285714284,
|
| 1335 |
-
"loss": 0.7743,
|
| 1336 |
-
"step": 2080
|
| 1337 |
-
},
|
| 1338 |
-
{
|
| 1339 |
-
"epoch": 5.38,
|
| 1340 |
-
"learning_rate": 0.00014206349206349206,
|
| 1341 |
-
"loss": 0.6337,
|
| 1342 |
-
"step": 2090
|
| 1343 |
-
},
|
| 1344 |
-
{
|
| 1345 |
-
"epoch": 5.4,
|
| 1346 |
-
"learning_rate": 0.00014126984126984124,
|
| 1347 |
-
"loss": 0.7518,
|
| 1348 |
-
"step": 2100
|
| 1349 |
-
},
|
| 1350 |
-
{
|
| 1351 |
-
"epoch": 5.43,
|
| 1352 |
-
"learning_rate": 0.00014047619047619046,
|
| 1353 |
-
"loss": 0.8736,
|
| 1354 |
-
"step": 2110
|
| 1355 |
-
},
|
| 1356 |
-
{
|
| 1357 |
-
"epoch": 5.45,
|
| 1358 |
-
"learning_rate": 0.00013968253968253967,
|
| 1359 |
-
"loss": 0.8623,
|
| 1360 |
-
"step": 2120
|
| 1361 |
-
},
|
| 1362 |
-
{
|
| 1363 |
-
"epoch": 5.48,
|
| 1364 |
-
"learning_rate": 0.0001388888888888889,
|
| 1365 |
-
"loss": 0.7719,
|
| 1366 |
-
"step": 2130
|
| 1367 |
-
},
|
| 1368 |
-
{
|
| 1369 |
-
"epoch": 5.5,
|
| 1370 |
-
"learning_rate": 0.00013809523809523808,
|
| 1371 |
-
"loss": 0.6363,
|
| 1372 |
-
"step": 2140
|
| 1373 |
-
},
|
| 1374 |
-
{
|
| 1375 |
-
"epoch": 5.53,
|
| 1376 |
-
"learning_rate": 0.0001373015873015873,
|
| 1377 |
-
"loss": 0.7458,
|
| 1378 |
-
"step": 2150
|
| 1379 |
-
},
|
| 1380 |
-
{
|
| 1381 |
-
"epoch": 5.56,
|
| 1382 |
-
"learning_rate": 0.00013650793650793648,
|
| 1383 |
-
"loss": 0.865,
|
| 1384 |
-
"step": 2160
|
| 1385 |
-
},
|
| 1386 |
-
{
|
| 1387 |
-
"epoch": 5.58,
|
| 1388 |
-
"learning_rate": 0.00013587301587301588,
|
| 1389 |
-
"loss": 2.6458,
|
| 1390 |
-
"step": 2170
|
| 1391 |
-
},
|
| 1392 |
-
{
|
| 1393 |
-
"epoch": 5.61,
|
| 1394 |
-
"learning_rate": 0.00013587301587301588,
|
| 1395 |
-
"loss": 801.9857,
|
| 1396 |
-
"step": 2180
|
| 1397 |
-
},
|
| 1398 |
-
{
|
| 1399 |
-
"epoch": 5.63,
|
| 1400 |
-
"learning_rate": 0.0001357142857142857,
|
| 1401 |
-
"loss": 4068.018,
|
| 1402 |
-
"step": 2190
|
| 1403 |
-
},
|
| 1404 |
-
{
|
| 1405 |
-
"epoch": 5.66,
|
| 1406 |
-
"learning_rate": 0.00013547619047619047,
|
| 1407 |
-
"loss": 41837456.0,
|
| 1408 |
-
"step": 2200
|
| 1409 |
-
},
|
| 1410 |
-
{
|
| 1411 |
-
"epoch": 5.66,
|
| 1412 |
-
"eval_loss": NaN,
|
| 1413 |
-
"eval_runtime": 189.1905,
|
| 1414 |
-
"eval_samples_per_second": 10.571,
|
| 1415 |
-
"eval_steps_per_second": 1.321,
|
| 1416 |
-
"step": 2200
|
| 1417 |
-
},
|
| 1418 |
-
{
|
| 1419 |
-
"epoch": 5.68,
|
| 1420 |
-
"learning_rate": 0.00013547619047619047,
|
| 1421 |
-
"loss": 1.4509301308571507e+26,
|
| 1422 |
-
"step": 2210
|
| 1423 |
-
},
|
| 1424 |
-
{
|
| 1425 |
-
"epoch": 5.71,
|
| 1426 |
-
"learning_rate": 0.00013547619047619047,
|
| 1427 |
-
"loss": 3.3847730092507856e+24,
|
| 1428 |
-
"step": 2220
|
| 1429 |
-
},
|
| 1430 |
-
{
|
| 1431 |
-
"epoch": 5.74,
|
| 1432 |
-
"learning_rate": 0.00013547619047619047,
|
| 1433 |
-
"loss": 1.5211943209070177e+23,
|
| 1434 |
-
"step": 2230
|
| 1435 |
-
},
|
| 1436 |
-
{
|
| 1437 |
-
"epoch": 5.76,
|
| 1438 |
-
"learning_rate": 0.00013547619047619047,
|
| 1439 |
-
"loss": 6.678915709304036e+21,
|
| 1440 |
-
"step": 2240
|
| 1441 |
-
},
|
| 1442 |
-
{
|
| 1443 |
-
"epoch": 5.79,
|
| 1444 |
-
"learning_rate": 0.00013531746031746032,
|
| 1445 |
-
"loss": 9.238063623264189e+19,
|
| 1446 |
-
"step": 2250
|
| 1447 |
-
},
|
| 1448 |
-
{
|
| 1449 |
-
"epoch": 5.81,
|
| 1450 |
-
"learning_rate": 0.00013531746031746032,
|
| 1451 |
-
"loss": 1.0532014646514701e+20,
|
| 1452 |
-
"step": 2260
|
| 1453 |
-
},
|
| 1454 |
-
{
|
| 1455 |
-
"epoch": 5.84,
|
| 1456 |
-
"learning_rate": 0.00013531746031746032,
|
| 1457 |
-
"loss": 3.161435096687031e+25,
|
| 1458 |
-
"step": 2270
|
| 1459 |
-
},
|
| 1460 |
-
{
|
| 1461 |
-
"epoch": 5.86,
|
| 1462 |
-
"learning_rate": 0.00013531746031746032,
|
| 1463 |
-
"loss": 2.0162940987179532e+19,
|
| 1464 |
-
"step": 2280
|
| 1465 |
-
},
|
| 1466 |
-
{
|
| 1467 |
-
"epoch": 5.89,
|
| 1468 |
-
"learning_rate": 0.00013531746031746032,
|
| 1469 |
-
"loss": 2.55694599151234e+20,
|
| 1470 |
-
"step": 2290
|
| 1471 |
-
},
|
| 1472 |
-
{
|
| 1473 |
-
"epoch": 5.92,
|
| 1474 |
-
"learning_rate": 0.00013523809523809522,
|
| 1475 |
-
"loss": 5.808441058207432e+20,
|
| 1476 |
-
"step": 2300
|
| 1477 |
-
},
|
| 1478 |
-
{
|
| 1479 |
-
"epoch": 5.94,
|
| 1480 |
-
"learning_rate": 0.00013523809523809522,
|
| 1481 |
-
"loss": 1.9880088514154103e+22,
|
| 1482 |
-
"step": 2310
|
| 1483 |
-
},
|
| 1484 |
-
{
|
| 1485 |
-
"epoch": 5.97,
|
| 1486 |
-
"learning_rate": 0.00013523809523809522,
|
| 1487 |
-
"loss": 2.0954874078435546e+24,
|
| 1488 |
-
"step": 2320
|
| 1489 |
-
},
|
| 1490 |
-
{
|
| 1491 |
-
"epoch": 5.99,
|
| 1492 |
-
"learning_rate": 0.00013523809523809522,
|
| 1493 |
-
"loss": 6.309141694629275e+20,
|
| 1494 |
-
"step": 2330
|
| 1495 |
-
},
|
| 1496 |
-
{
|
| 1497 |
-
"epoch": 6.02,
|
| 1498 |
-
"learning_rate": 0.00013515873015873016,
|
| 1499 |
-
"loss": 7.353349497283535e+23,
|
| 1500 |
-
"step": 2340
|
| 1501 |
-
},
|
| 1502 |
-
{
|
| 1503 |
-
"epoch": 6.05,
|
| 1504 |
-
"learning_rate": 0.00013468253968253966,
|
| 1505 |
-
"loss": 0.0,
|
| 1506 |
-
"step": 2350
|
| 1507 |
-
},
|
| 1508 |
-
{
|
| 1509 |
-
"epoch": 6.07,
|
| 1510 |
-
"learning_rate": 0.00013388888888888888,
|
| 1511 |
-
"loss": 0.0,
|
| 1512 |
-
"step": 2360
|
| 1513 |
-
},
|
| 1514 |
-
{
|
| 1515 |
-
"epoch": 6.1,
|
| 1516 |
-
"learning_rate": 0.00013309523809523806,
|
| 1517 |
-
"loss": 0.0,
|
| 1518 |
-
"step": 2370
|
| 1519 |
-
},
|
| 1520 |
-
{
|
| 1521 |
-
"epoch": 6.12,
|
| 1522 |
-
"learning_rate": 0.00013230158730158728,
|
| 1523 |
-
"loss": 0.0,
|
| 1524 |
-
"step": 2380
|
| 1525 |
-
},
|
| 1526 |
-
{
|
| 1527 |
-
"epoch": 6.15,
|
| 1528 |
-
"learning_rate": 0.0001315079365079365,
|
| 1529 |
-
"loss": 0.0,
|
| 1530 |
-
"step": 2390
|
| 1531 |
-
},
|
| 1532 |
-
{
|
| 1533 |
-
"epoch": 6.17,
|
| 1534 |
-
"learning_rate": 0.0001307142857142857,
|
| 1535 |
-
"loss": 0.0,
|
| 1536 |
-
"step": 2400
|
| 1537 |
-
},
|
| 1538 |
-
{
|
| 1539 |
-
"epoch": 6.17,
|
| 1540 |
-
"eval_loss": NaN,
|
| 1541 |
-
"eval_runtime": 164.902,
|
| 1542 |
-
"eval_samples_per_second": 12.128,
|
| 1543 |
-
"eval_steps_per_second": 1.516,
|
| 1544 |
-
"step": 2400
|
| 1545 |
-
},
|
| 1546 |
-
{
|
| 1547 |
-
"epoch": 6.2,
|
| 1548 |
-
"learning_rate": 0.0001299206349206349,
|
| 1549 |
-
"loss": 0.0,
|
| 1550 |
-
"step": 2410
|
| 1551 |
-
},
|
| 1552 |
-
{
|
| 1553 |
-
"epoch": 6.23,
|
| 1554 |
-
"learning_rate": 0.0001291269841269841,
|
| 1555 |
-
"loss": 0.0,
|
| 1556 |
-
"step": 2420
|
| 1557 |
-
},
|
| 1558 |
-
{
|
| 1559 |
-
"epoch": 6.25,
|
| 1560 |
-
"learning_rate": 0.00012833333333333333,
|
| 1561 |
-
"loss": 0.0,
|
| 1562 |
-
"step": 2430
|
| 1563 |
-
},
|
| 1564 |
-
{
|
| 1565 |
-
"epoch": 6.28,
|
| 1566 |
-
"learning_rate": 0.00012753968253968254,
|
| 1567 |
-
"loss": 0.0,
|
| 1568 |
-
"step": 2440
|
| 1569 |
-
},
|
| 1570 |
-
{
|
| 1571 |
-
"epoch": 6.3,
|
| 1572 |
-
"learning_rate": 0.00012674603174603173,
|
| 1573 |
-
"loss": 0.0,
|
| 1574 |
-
"step": 2450
|
| 1575 |
-
},
|
| 1576 |
-
{
|
| 1577 |
-
"epoch": 6.33,
|
| 1578 |
-
"learning_rate": 0.00012595238095238094,
|
| 1579 |
-
"loss": 0.0,
|
| 1580 |
-
"step": 2460
|
| 1581 |
-
},
|
| 1582 |
-
{
|
| 1583 |
-
"epoch": 6.35,
|
| 1584 |
-
"learning_rate": 0.00012515873015873013,
|
| 1585 |
-
"loss": 0.0,
|
| 1586 |
-
"step": 2470
|
| 1587 |
-
},
|
| 1588 |
-
{
|
| 1589 |
-
"epoch": 6.38,
|
| 1590 |
-
"learning_rate": 0.00012436507936507935,
|
| 1591 |
-
"loss": 0.0,
|
| 1592 |
-
"step": 2480
|
| 1593 |
-
},
|
| 1594 |
-
{
|
| 1595 |
-
"epoch": 6.41,
|
| 1596 |
-
"learning_rate": 0.00012357142857142856,
|
| 1597 |
-
"loss": 0.0,
|
| 1598 |
-
"step": 2490
|
| 1599 |
-
},
|
| 1600 |
-
{
|
| 1601 |
-
"epoch": 6.43,
|
| 1602 |
-
"learning_rate": 0.00012277777777777778,
|
| 1603 |
-
"loss": 0.0,
|
| 1604 |
-
"step": 2500
|
| 1605 |
-
},
|
| 1606 |
-
{
|
| 1607 |
-
"epoch": 6.46,
|
| 1608 |
-
"learning_rate": 0.00012198412698412697,
|
| 1609 |
-
"loss": 0.0,
|
| 1610 |
-
"step": 2510
|
| 1611 |
-
},
|
| 1612 |
-
{
|
| 1613 |
-
"epoch": 6.48,
|
| 1614 |
-
"learning_rate": 0.00012119047619047618,
|
| 1615 |
-
"loss": 0.0,
|
| 1616 |
-
"step": 2520
|
| 1617 |
-
},
|
| 1618 |
-
{
|
| 1619 |
-
"epoch": 6.51,
|
| 1620 |
-
"learning_rate": 0.00012039682539682538,
|
| 1621 |
-
"loss": 0.0,
|
| 1622 |
-
"step": 2530
|
| 1623 |
-
},
|
| 1624 |
-
{
|
| 1625 |
-
"epoch": 6.53,
|
| 1626 |
-
"learning_rate": 0.0001196031746031746,
|
| 1627 |
-
"loss": 0.0,
|
| 1628 |
-
"step": 2540
|
| 1629 |
-
},
|
| 1630 |
-
{
|
| 1631 |
-
"epoch": 6.56,
|
| 1632 |
-
"learning_rate": 0.0001188095238095238,
|
| 1633 |
-
"loss": 0.0,
|
| 1634 |
-
"step": 2550
|
| 1635 |
-
},
|
| 1636 |
-
{
|
| 1637 |
-
"epoch": 6.59,
|
| 1638 |
-
"learning_rate": 0.00011801587301587301,
|
| 1639 |
-
"loss": 0.0,
|
| 1640 |
-
"step": 2560
|
| 1641 |
-
},
|
| 1642 |
-
{
|
| 1643 |
-
"epoch": 6.61,
|
| 1644 |
-
"learning_rate": 0.0001172222222222222,
|
| 1645 |
-
"loss": 0.0,
|
| 1646 |
-
"step": 2570
|
| 1647 |
-
},
|
| 1648 |
-
{
|
| 1649 |
-
"epoch": 6.64,
|
| 1650 |
-
"learning_rate": 0.00011642857142857142,
|
| 1651 |
-
"loss": 0.0,
|
| 1652 |
-
"step": 2580
|
| 1653 |
-
},
|
| 1654 |
-
{
|
| 1655 |
-
"epoch": 6.66,
|
| 1656 |
-
"learning_rate": 0.00011563492063492062,
|
| 1657 |
-
"loss": 0.0,
|
| 1658 |
-
"step": 2590
|
| 1659 |
-
},
|
| 1660 |
-
{
|
| 1661 |
-
"epoch": 6.69,
|
| 1662 |
-
"learning_rate": 0.00011484126984126983,
|
| 1663 |
-
"loss": 0.0,
|
| 1664 |
-
"step": 2600
|
| 1665 |
-
},
|
| 1666 |
-
{
|
| 1667 |
-
"epoch": 6.69,
|
| 1668 |
-
"eval_loss": NaN,
|
| 1669 |
-
"eval_runtime": 164.9294,
|
| 1670 |
-
"eval_samples_per_second": 12.126,
|
| 1671 |
-
"eval_steps_per_second": 1.516,
|
| 1672 |
-
"step": 2600
|
| 1673 |
-
},
|
| 1674 |
-
{
|
| 1675 |
-
"epoch": 6.71,
|
| 1676 |
-
"learning_rate": 0.00011404761904761903,
|
| 1677 |
-
"loss": 0.0,
|
| 1678 |
-
"step": 2610
|
| 1679 |
-
},
|
| 1680 |
-
{
|
| 1681 |
-
"epoch": 6.74,
|
| 1682 |
-
"learning_rate": 0.00011325396825396825,
|
| 1683 |
-
"loss": 0.0,
|
| 1684 |
-
"step": 2620
|
| 1685 |
-
},
|
| 1686 |
-
{
|
| 1687 |
-
"epoch": 6.77,
|
| 1688 |
-
"learning_rate": 0.00011246031746031745,
|
| 1689 |
-
"loss": 0.0,
|
| 1690 |
-
"step": 2630
|
| 1691 |
-
},
|
| 1692 |
-
{
|
| 1693 |
-
"epoch": 6.79,
|
| 1694 |
-
"learning_rate": 0.00011166666666666667,
|
| 1695 |
-
"loss": 0.0,
|
| 1696 |
-
"step": 2640
|
| 1697 |
-
},
|
| 1698 |
-
{
|
| 1699 |
-
"epoch": 6.82,
|
| 1700 |
-
"learning_rate": 0.00011087301587301585,
|
| 1701 |
-
"loss": 0.0,
|
| 1702 |
-
"step": 2650
|
| 1703 |
-
},
|
| 1704 |
-
{
|
| 1705 |
-
"epoch": 6.84,
|
| 1706 |
-
"learning_rate": 0.00011007936507936507,
|
| 1707 |
-
"loss": 0.0,
|
| 1708 |
-
"step": 2660
|
| 1709 |
-
},
|
| 1710 |
-
{
|
| 1711 |
-
"epoch": 6.87,
|
| 1712 |
-
"learning_rate": 0.00010928571428571427,
|
| 1713 |
-
"loss": 0.0,
|
| 1714 |
-
"step": 2670
|
| 1715 |
-
},
|
| 1716 |
-
{
|
| 1717 |
-
"epoch": 6.89,
|
| 1718 |
-
"learning_rate": 0.00010849206349206349,
|
| 1719 |
-
"loss": 0.0,
|
| 1720 |
-
"step": 2680
|
| 1721 |
-
},
|
| 1722 |
-
{
|
| 1723 |
-
"epoch": 6.92,
|
| 1724 |
-
"learning_rate": 0.00010769841269841269,
|
| 1725 |
-
"loss": 0.0,
|
| 1726 |
-
"step": 2690
|
| 1727 |
-
},
|
| 1728 |
-
{
|
| 1729 |
-
"epoch": 6.95,
|
| 1730 |
-
"learning_rate": 0.0001069047619047619,
|
| 1731 |
-
"loss": 0.0,
|
| 1732 |
-
"step": 2700
|
| 1733 |
-
},
|
| 1734 |
-
{
|
| 1735 |
-
"epoch": 6.97,
|
| 1736 |
-
"learning_rate": 0.0001061111111111111,
|
| 1737 |
-
"loss": 0.0,
|
| 1738 |
-
"step": 2710
|
| 1739 |
-
},
|
| 1740 |
-
{
|
| 1741 |
-
"epoch": 7.0,
|
| 1742 |
-
"learning_rate": 0.0001053174603174603,
|
| 1743 |
-
"loss": 0.0,
|
| 1744 |
-
"step": 2720
|
| 1745 |
-
},
|
| 1746 |
-
{
|
| 1747 |
-
"epoch": 7.02,
|
| 1748 |
-
"learning_rate": 0.0001045238095238095,
|
| 1749 |
-
"loss": 0.0,
|
| 1750 |
-
"step": 2730
|
| 1751 |
-
},
|
| 1752 |
-
{
|
| 1753 |
-
"epoch": 7.05,
|
| 1754 |
-
"learning_rate": 0.00010373015873015872,
|
| 1755 |
-
"loss": 0.0,
|
| 1756 |
-
"step": 2740
|
| 1757 |
-
},
|
| 1758 |
-
{
|
| 1759 |
-
"epoch": 7.07,
|
| 1760 |
-
"learning_rate": 0.00010293650793650792,
|
| 1761 |
-
"loss": 0.0,
|
| 1762 |
-
"step": 2750
|
| 1763 |
-
},
|
| 1764 |
-
{
|
| 1765 |
-
"epoch": 7.1,
|
| 1766 |
-
"learning_rate": 0.00010214285714285714,
|
| 1767 |
-
"loss": 0.0,
|
| 1768 |
-
"step": 2760
|
| 1769 |
-
},
|
| 1770 |
-
{
|
| 1771 |
-
"epoch": 7.13,
|
| 1772 |
-
"learning_rate": 0.00010134920634920634,
|
| 1773 |
-
"loss": 0.0,
|
| 1774 |
-
"step": 2770
|
| 1775 |
-
},
|
| 1776 |
-
{
|
| 1777 |
-
"epoch": 7.15,
|
| 1778 |
-
"learning_rate": 0.00010055555555555555,
|
| 1779 |
-
"loss": 0.0,
|
| 1780 |
-
"step": 2780
|
| 1781 |
-
},
|
| 1782 |
-
{
|
| 1783 |
-
"epoch": 7.18,
|
| 1784 |
-
"learning_rate": 9.976190476190474e-05,
|
| 1785 |
-
"loss": 0.0,
|
| 1786 |
-
"step": 2790
|
| 1787 |
-
},
|
| 1788 |
-
{
|
| 1789 |
-
"epoch": 7.2,
|
| 1790 |
-
"learning_rate": 9.896825396825396e-05,
|
| 1791 |
-
"loss": 0.0,
|
| 1792 |
-
"step": 2800
|
| 1793 |
-
},
|
| 1794 |
-
{
|
| 1795 |
-
"epoch": 7.2,
|
| 1796 |
-
"eval_loss": NaN,
|
| 1797 |
-
"eval_runtime": 164.8859,
|
| 1798 |
-
"eval_samples_per_second": 12.13,
|
| 1799 |
-
"eval_steps_per_second": 1.516,
|
| 1800 |
-
"step": 2800
|
| 1801 |
-
},
|
| 1802 |
-
{
|
| 1803 |
-
"epoch": 7.23,
|
| 1804 |
-
"learning_rate": 9.817460317460316e-05,
|
| 1805 |
-
"loss": 0.0,
|
| 1806 |
-
"step": 2810
|
| 1807 |
-
},
|
| 1808 |
-
{
|
| 1809 |
-
"epoch": 7.25,
|
| 1810 |
-
"learning_rate": 9.738095238095237e-05,
|
| 1811 |
-
"loss": 0.0,
|
| 1812 |
-
"step": 2820
|
| 1813 |
-
},
|
| 1814 |
-
{
|
| 1815 |
-
"epoch": 7.28,
|
| 1816 |
-
"learning_rate": 9.658730158730158e-05,
|
| 1817 |
-
"loss": 0.0,
|
| 1818 |
-
"step": 2830
|
| 1819 |
-
},
|
| 1820 |
-
{
|
| 1821 |
-
"epoch": 7.31,
|
| 1822 |
-
"learning_rate": 9.579365079365079e-05,
|
| 1823 |
-
"loss": 0.0,
|
| 1824 |
-
"step": 2840
|
| 1825 |
-
},
|
| 1826 |
-
{
|
| 1827 |
-
"epoch": 7.33,
|
| 1828 |
-
"learning_rate": 9.499999999999999e-05,
|
| 1829 |
-
"loss": 0.0,
|
| 1830 |
-
"step": 2850
|
| 1831 |
-
},
|
| 1832 |
-
{
|
| 1833 |
-
"epoch": 7.36,
|
| 1834 |
-
"learning_rate": 9.42063492063492e-05,
|
| 1835 |
-
"loss": 0.0,
|
| 1836 |
-
"step": 2860
|
| 1837 |
-
},
|
| 1838 |
-
{
|
| 1839 |
-
"epoch": 7.38,
|
| 1840 |
-
"learning_rate": 9.34126984126984e-05,
|
| 1841 |
-
"loss": 0.0,
|
| 1842 |
-
"step": 2870
|
| 1843 |
-
},
|
| 1844 |
-
{
|
| 1845 |
-
"epoch": 7.41,
|
| 1846 |
-
"learning_rate": 9.261904761904761e-05,
|
| 1847 |
-
"loss": 0.0,
|
| 1848 |
-
"step": 2880
|
| 1849 |
-
},
|
| 1850 |
-
{
|
| 1851 |
-
"epoch": 7.43,
|
| 1852 |
-
"learning_rate": 9.182539682539681e-05,
|
| 1853 |
-
"loss": 0.0,
|
| 1854 |
-
"step": 2890
|
| 1855 |
-
},
|
| 1856 |
-
{
|
| 1857 |
-
"epoch": 7.46,
|
| 1858 |
-
"learning_rate": 9.103174603174603e-05,
|
| 1859 |
-
"loss": 0.0,
|
| 1860 |
-
"step": 2900
|
| 1861 |
-
},
|
| 1862 |
-
{
|
| 1863 |
-
"epoch": 7.49,
|
| 1864 |
-
"learning_rate": 9.023809523809523e-05,
|
| 1865 |
-
"loss": 0.0,
|
| 1866 |
-
"step": 2910
|
| 1867 |
-
},
|
| 1868 |
-
{
|
| 1869 |
-
"epoch": 7.51,
|
| 1870 |
-
"learning_rate": 8.944444444444444e-05,
|
| 1871 |
-
"loss": 0.0,
|
| 1872 |
-
"step": 2920
|
| 1873 |
-
},
|
| 1874 |
-
{
|
| 1875 |
-
"epoch": 7.54,
|
| 1876 |
-
"learning_rate": 8.865079365079364e-05,
|
| 1877 |
-
"loss": 0.0,
|
| 1878 |
-
"step": 2930
|
| 1879 |
-
},
|
| 1880 |
-
{
|
| 1881 |
-
"epoch": 7.56,
|
| 1882 |
-
"learning_rate": 8.785714285714286e-05,
|
| 1883 |
-
"loss": 0.0,
|
| 1884 |
-
"step": 2940
|
| 1885 |
-
},
|
| 1886 |
-
{
|
| 1887 |
-
"epoch": 7.59,
|
| 1888 |
-
"learning_rate": 8.706349206349205e-05,
|
| 1889 |
-
"loss": 0.0,
|
| 1890 |
-
"step": 2950
|
| 1891 |
-
},
|
| 1892 |
-
{
|
| 1893 |
-
"epoch": 7.61,
|
| 1894 |
-
"learning_rate": 8.626984126984126e-05,
|
| 1895 |
-
"loss": 0.0,
|
| 1896 |
-
"step": 2960
|
| 1897 |
-
},
|
| 1898 |
-
{
|
| 1899 |
-
"epoch": 7.64,
|
| 1900 |
-
"learning_rate": 8.547619047619046e-05,
|
| 1901 |
-
"loss": 0.0,
|
| 1902 |
-
"step": 2970
|
| 1903 |
-
},
|
| 1904 |
-
{
|
| 1905 |
-
"epoch": 7.67,
|
| 1906 |
-
"learning_rate": 8.468253968253968e-05,
|
| 1907 |
-
"loss": 0.0,
|
| 1908 |
-
"step": 2980
|
| 1909 |
-
},
|
| 1910 |
-
{
|
| 1911 |
-
"epoch": 7.69,
|
| 1912 |
-
"learning_rate": 8.388888888888888e-05,
|
| 1913 |
-
"loss": 0.0,
|
| 1914 |
-
"step": 2990
|
| 1915 |
-
},
|
| 1916 |
-
{
|
| 1917 |
-
"epoch": 7.72,
|
| 1918 |
-
"learning_rate": 8.30952380952381e-05,
|
| 1919 |
-
"loss": 0.0,
|
| 1920 |
-
"step": 3000
|
| 1921 |
-
},
|
| 1922 |
-
{
|
| 1923 |
-
"epoch": 7.72,
|
| 1924 |
-
"eval_loss": NaN,
|
| 1925 |
-
"eval_runtime": 164.8915,
|
| 1926 |
-
"eval_samples_per_second": 12.129,
|
| 1927 |
-
"eval_steps_per_second": 1.516,
|
| 1928 |
-
"step": 3000
|
| 1929 |
-
},
|
| 1930 |
-
{
|
| 1931 |
-
"epoch": 7.74,
|
| 1932 |
-
"learning_rate": 8.23015873015873e-05,
|
| 1933 |
-
"loss": 0.0,
|
| 1934 |
-
"step": 3010
|
| 1935 |
-
},
|
| 1936 |
-
{
|
| 1937 |
-
"epoch": 7.77,
|
| 1938 |
-
"learning_rate": 8.150793650793651e-05,
|
| 1939 |
-
"loss": 0.0,
|
| 1940 |
-
"step": 3020
|
| 1941 |
-
},
|
| 1942 |
-
{
|
| 1943 |
-
"epoch": 7.79,
|
| 1944 |
-
"learning_rate": 8.07142857142857e-05,
|
| 1945 |
-
"loss": 0.0,
|
| 1946 |
-
"step": 3030
|
| 1947 |
-
},
|
| 1948 |
-
{
|
| 1949 |
-
"epoch": 7.82,
|
| 1950 |
-
"learning_rate": 7.992063492063491e-05,
|
| 1951 |
-
"loss": 0.0,
|
| 1952 |
-
"step": 3040
|
| 1953 |
-
},
|
| 1954 |
-
{
|
| 1955 |
-
"epoch": 7.85,
|
| 1956 |
-
"learning_rate": 7.912698412698412e-05,
|
| 1957 |
-
"loss": 0.0,
|
| 1958 |
-
"step": 3050
|
| 1959 |
-
},
|
| 1960 |
-
{
|
| 1961 |
-
"epoch": 7.87,
|
| 1962 |
-
"learning_rate": 7.833333333333333e-05,
|
| 1963 |
-
"loss": 0.0,
|
| 1964 |
-
"step": 3060
|
| 1965 |
-
},
|
| 1966 |
-
{
|
| 1967 |
-
"epoch": 7.9,
|
| 1968 |
-
"learning_rate": 7.753968253968253e-05,
|
| 1969 |
-
"loss": 0.0,
|
| 1970 |
-
"step": 3070
|
| 1971 |
-
},
|
| 1972 |
-
{
|
| 1973 |
-
"epoch": 7.92,
|
| 1974 |
-
"learning_rate": 7.674603174603175e-05,
|
| 1975 |
-
"loss": 0.0,
|
| 1976 |
-
"step": 3080
|
| 1977 |
-
},
|
| 1978 |
-
{
|
| 1979 |
-
"epoch": 7.95,
|
| 1980 |
-
"learning_rate": 7.595238095238095e-05,
|
| 1981 |
-
"loss": 0.0,
|
| 1982 |
-
"step": 3090
|
| 1983 |
-
},
|
| 1984 |
-
{
|
| 1985 |
-
"epoch": 7.97,
|
| 1986 |
-
"learning_rate": 7.515873015873015e-05,
|
| 1987 |
-
"loss": 0.0,
|
| 1988 |
-
"step": 3100
|
| 1989 |
-
},
|
| 1990 |
-
{
|
| 1991 |
-
"epoch": 8.0,
|
| 1992 |
-
"learning_rate": 7.436507936507935e-05,
|
| 1993 |
-
"loss": 0.0,
|
| 1994 |
-
"step": 3110
|
| 1995 |
-
},
|
| 1996 |
-
{
|
| 1997 |
-
"epoch": 8.03,
|
| 1998 |
-
"learning_rate": 7.357142857142857e-05,
|
| 1999 |
-
"loss": 0.0,
|
| 2000 |
-
"step": 3120
|
| 2001 |
-
},
|
| 2002 |
-
{
|
| 2003 |
-
"epoch": 8.05,
|
| 2004 |
-
"learning_rate": 7.277777777777777e-05,
|
| 2005 |
-
"loss": 0.0,
|
| 2006 |
-
"step": 3130
|
| 2007 |
-
},
|
| 2008 |
-
{
|
| 2009 |
-
"epoch": 8.08,
|
| 2010 |
-
"learning_rate": 7.198412698412697e-05,
|
| 2011 |
-
"loss": 0.0,
|
| 2012 |
-
"step": 3140
|
| 2013 |
-
},
|
| 2014 |
-
{
|
| 2015 |
-
"epoch": 8.1,
|
| 2016 |
-
"learning_rate": 7.119047619047618e-05,
|
| 2017 |
-
"loss": 0.0,
|
| 2018 |
-
"step": 3150
|
| 2019 |
-
},
|
| 2020 |
-
{
|
| 2021 |
-
"epoch": 8.13,
|
| 2022 |
-
"learning_rate": 7.039682539682539e-05,
|
| 2023 |
-
"loss": 0.0,
|
| 2024 |
-
"step": 3160
|
| 2025 |
-
},
|
| 2026 |
-
{
|
| 2027 |
-
"epoch": 8.15,
|
| 2028 |
-
"learning_rate": 6.960317460317459e-05,
|
| 2029 |
-
"loss": 0.0,
|
| 2030 |
-
"step": 3170
|
| 2031 |
-
},
|
| 2032 |
-
{
|
| 2033 |
-
"epoch": 8.18,
|
| 2034 |
-
"learning_rate": 6.88095238095238e-05,
|
| 2035 |
-
"loss": 0.0,
|
| 2036 |
-
"step": 3180
|
| 2037 |
-
},
|
| 2038 |
-
{
|
| 2039 |
-
"epoch": 8.21,
|
| 2040 |
-
"learning_rate": 6.8015873015873e-05,
|
| 2041 |
-
"loss": 0.0,
|
| 2042 |
-
"step": 3190
|
| 2043 |
-
},
|
| 2044 |
-
{
|
| 2045 |
-
"epoch": 8.23,
|
| 2046 |
-
"learning_rate": 6.722222222222222e-05,
|
| 2047 |
-
"loss": 0.0,
|
| 2048 |
-
"step": 3200
|
| 2049 |
-
},
|
| 2050 |
-
{
|
| 2051 |
-
"epoch": 8.23,
|
| 2052 |
-
"eval_loss": NaN,
|
| 2053 |
-
"eval_runtime": 164.9214,
|
| 2054 |
-
"eval_samples_per_second": 12.127,
|
| 2055 |
-
"eval_steps_per_second": 1.516,
|
| 2056 |
-
"step": 3200
|
| 2057 |
-
},
|
| 2058 |
-
{
|
| 2059 |
-
"epoch": 8.26,
|
| 2060 |
-
"learning_rate": 6.642857142857142e-05,
|
| 2061 |
-
"loss": 0.0,
|
| 2062 |
-
"step": 3210
|
| 2063 |
-
},
|
| 2064 |
-
{
|
| 2065 |
-
"epoch": 8.28,
|
| 2066 |
-
"learning_rate": 6.563492063492062e-05,
|
| 2067 |
-
"loss": 0.0,
|
| 2068 |
-
"step": 3220
|
| 2069 |
-
},
|
| 2070 |
-
{
|
| 2071 |
-
"epoch": 8.31,
|
| 2072 |
-
"learning_rate": 6.484126984126984e-05,
|
| 2073 |
-
"loss": 0.0,
|
| 2074 |
-
"step": 3230
|
| 2075 |
-
},
|
| 2076 |
-
{
|
| 2077 |
-
"epoch": 8.33,
|
| 2078 |
-
"learning_rate": 6.404761904761904e-05,
|
| 2079 |
-
"loss": 0.0,
|
| 2080 |
-
"step": 3240
|
| 2081 |
-
},
|
| 2082 |
-
{
|
| 2083 |
-
"epoch": 8.36,
|
| 2084 |
-
"learning_rate": 6.325396825396824e-05,
|
| 2085 |
-
"loss": 0.0,
|
| 2086 |
-
"step": 3250
|
| 2087 |
-
},
|
| 2088 |
-
{
|
| 2089 |
-
"epoch": 8.39,
|
| 2090 |
-
"learning_rate": 6.246031746031746e-05,
|
| 2091 |
-
"loss": 0.0,
|
| 2092 |
-
"step": 3260
|
| 2093 |
-
},
|
| 2094 |
-
{
|
| 2095 |
-
"epoch": 8.41,
|
| 2096 |
-
"learning_rate": 6.166666666666666e-05,
|
| 2097 |
-
"loss": 0.0,
|
| 2098 |
-
"step": 3270
|
| 2099 |
-
},
|
| 2100 |
-
{
|
| 2101 |
-
"epoch": 8.44,
|
| 2102 |
-
"learning_rate": 6.0873015873015865e-05,
|
| 2103 |
-
"loss": 0.0,
|
| 2104 |
-
"step": 3280
|
| 2105 |
-
},
|
| 2106 |
-
{
|
| 2107 |
-
"epoch": 8.46,
|
| 2108 |
-
"learning_rate": 6.007936507936507e-05,
|
| 2109 |
-
"loss": 0.0,
|
| 2110 |
-
"step": 3290
|
| 2111 |
-
},
|
| 2112 |
-
{
|
| 2113 |
-
"epoch": 8.49,
|
| 2114 |
-
"learning_rate": 5.9285714285714275e-05,
|
| 2115 |
-
"loss": 0.0,
|
| 2116 |
-
"step": 3300
|
| 2117 |
-
},
|
| 2118 |
-
{
|
| 2119 |
-
"epoch": 8.51,
|
| 2120 |
-
"learning_rate": 5.849206349206348e-05,
|
| 2121 |
-
"loss": 0.0,
|
| 2122 |
-
"step": 3310
|
| 2123 |
-
},
|
| 2124 |
-
{
|
| 2125 |
-
"epoch": 8.54,
|
| 2126 |
-
"learning_rate": 5.769841269841269e-05,
|
| 2127 |
-
"loss": 0.0,
|
| 2128 |
-
"step": 3320
|
| 2129 |
-
},
|
| 2130 |
-
{
|
| 2131 |
-
"epoch": 8.57,
|
| 2132 |
-
"learning_rate": 5.69047619047619e-05,
|
| 2133 |
-
"loss": 0.0,
|
| 2134 |
-
"step": 3330
|
| 2135 |
-
},
|
| 2136 |
-
{
|
| 2137 |
-
"epoch": 8.59,
|
| 2138 |
-
"learning_rate": 5.61111111111111e-05,
|
| 2139 |
-
"loss": 0.0,
|
| 2140 |
-
"step": 3340
|
| 2141 |
-
},
|
| 2142 |
-
{
|
| 2143 |
-
"epoch": 8.62,
|
| 2144 |
-
"learning_rate": 5.531746031746031e-05,
|
| 2145 |
-
"loss": 0.0,
|
| 2146 |
-
"step": 3350
|
| 2147 |
-
},
|
| 2148 |
-
{
|
| 2149 |
-
"epoch": 8.64,
|
| 2150 |
-
"learning_rate": 5.452380952380952e-05,
|
| 2151 |
-
"loss": 0.0,
|
| 2152 |
-
"step": 3360
|
| 2153 |
-
},
|
| 2154 |
-
{
|
| 2155 |
-
"epoch": 8.67,
|
| 2156 |
-
"learning_rate": 5.3730158730158726e-05,
|
| 2157 |
-
"loss": 0.0,
|
| 2158 |
-
"step": 3370
|
| 2159 |
-
},
|
| 2160 |
-
{
|
| 2161 |
-
"epoch": 8.69,
|
| 2162 |
-
"learning_rate": 5.293650793650793e-05,
|
| 2163 |
-
"loss": 0.0,
|
| 2164 |
-
"step": 3380
|
| 2165 |
-
},
|
| 2166 |
-
{
|
| 2167 |
-
"epoch": 8.72,
|
| 2168 |
-
"learning_rate": 5.2142857142857135e-05,
|
| 2169 |
-
"loss": 0.0,
|
| 2170 |
-
"step": 3390
|
| 2171 |
-
},
|
| 2172 |
-
{
|
| 2173 |
-
"epoch": 8.75,
|
| 2174 |
-
"learning_rate": 5.1349206349206344e-05,
|
| 2175 |
-
"loss": 0.0,
|
| 2176 |
-
"step": 3400
|
| 2177 |
-
},
|
| 2178 |
-
{
|
| 2179 |
-
"epoch": 8.75,
|
| 2180 |
-
"eval_loss": NaN,
|
| 2181 |
-
"eval_runtime": 164.9331,
|
| 2182 |
-
"eval_samples_per_second": 12.126,
|
| 2183 |
-
"eval_steps_per_second": 1.516,
|
| 2184 |
-
"step": 3400
|
| 2185 |
-
},
|
| 2186 |
-
{
|
| 2187 |
-
"epoch": 8.77,
|
| 2188 |
-
"learning_rate": 5.055555555555555e-05,
|
| 2189 |
-
"loss": 0.0,
|
| 2190 |
-
"step": 3410
|
| 2191 |
-
},
|
| 2192 |
-
{
|
| 2193 |
-
"epoch": 8.8,
|
| 2194 |
-
"learning_rate": 4.976190476190475e-05,
|
| 2195 |
-
"loss": 0.0,
|
| 2196 |
-
"step": 3420
|
| 2197 |
-
},
|
| 2198 |
-
{
|
| 2199 |
-
"epoch": 8.82,
|
| 2200 |
-
"learning_rate": 4.896825396825396e-05,
|
| 2201 |
-
"loss": 0.0,
|
| 2202 |
-
"step": 3430
|
| 2203 |
-
},
|
| 2204 |
-
{
|
| 2205 |
-
"epoch": 8.85,
|
| 2206 |
-
"learning_rate": 4.817460317460317e-05,
|
| 2207 |
-
"loss": 0.0,
|
| 2208 |
-
"step": 3440
|
| 2209 |
-
},
|
| 2210 |
-
{
|
| 2211 |
-
"epoch": 8.87,
|
| 2212 |
-
"learning_rate": 4.738095238095238e-05,
|
| 2213 |
-
"loss": 0.0,
|
| 2214 |
-
"step": 3450
|
| 2215 |
-
},
|
| 2216 |
-
{
|
| 2217 |
-
"epoch": 8.9,
|
| 2218 |
-
"learning_rate": 4.658730158730158e-05,
|
| 2219 |
-
"loss": 0.0,
|
| 2220 |
-
"step": 3460
|
| 2221 |
-
},
|
| 2222 |
-
{
|
| 2223 |
-
"epoch": 8.93,
|
| 2224 |
-
"learning_rate": 4.579365079365079e-05,
|
| 2225 |
-
"loss": 0.0,
|
| 2226 |
-
"step": 3470
|
| 2227 |
-
},
|
| 2228 |
-
{
|
| 2229 |
-
"epoch": 8.95,
|
| 2230 |
-
"learning_rate": 4.4999999999999996e-05,
|
| 2231 |
-
"loss": 0.0,
|
| 2232 |
-
"step": 3480
|
| 2233 |
-
},
|
| 2234 |
-
{
|
| 2235 |
-
"epoch": 8.98,
|
| 2236 |
-
"learning_rate": 4.42063492063492e-05,
|
| 2237 |
-
"loss": 0.0,
|
| 2238 |
-
"step": 3490
|
| 2239 |
-
},
|
| 2240 |
-
{
|
| 2241 |
-
"epoch": 9.0,
|
| 2242 |
-
"learning_rate": 4.3412698412698406e-05,
|
| 2243 |
-
"loss": 0.0,
|
| 2244 |
-
"step": 3500
|
| 2245 |
-
},
|
| 2246 |
-
{
|
| 2247 |
-
"epoch": 9.03,
|
| 2248 |
-
"learning_rate": 4.2619047619047614e-05,
|
| 2249 |
-
"loss": 0.0,
|
| 2250 |
-
"step": 3510
|
| 2251 |
-
},
|
| 2252 |
-
{
|
| 2253 |
-
"epoch": 9.05,
|
| 2254 |
-
"learning_rate": 4.182539682539682e-05,
|
| 2255 |
-
"loss": 0.0,
|
| 2256 |
-
"step": 3520
|
| 2257 |
-
},
|
| 2258 |
-
{
|
| 2259 |
-
"epoch": 9.08,
|
| 2260 |
-
"learning_rate": 4.1031746031746024e-05,
|
| 2261 |
-
"loss": 0.0,
|
| 2262 |
-
"step": 3530
|
| 2263 |
-
},
|
| 2264 |
-
{
|
| 2265 |
-
"epoch": 9.11,
|
| 2266 |
-
"learning_rate": 4.023809523809523e-05,
|
| 2267 |
-
"loss": 0.0,
|
| 2268 |
-
"step": 3540
|
| 2269 |
-
},
|
| 2270 |
-
{
|
| 2271 |
-
"epoch": 9.13,
|
| 2272 |
-
"learning_rate": 3.944444444444444e-05,
|
| 2273 |
-
"loss": 0.0,
|
| 2274 |
-
"step": 3550
|
| 2275 |
-
},
|
| 2276 |
-
{
|
| 2277 |
-
"epoch": 9.16,
|
| 2278 |
-
"learning_rate": 3.865079365079365e-05,
|
| 2279 |
-
"loss": 0.0,
|
| 2280 |
-
"step": 3560
|
| 2281 |
-
},
|
| 2282 |
-
{
|
| 2283 |
-
"epoch": 9.18,
|
| 2284 |
-
"learning_rate": 3.785714285714285e-05,
|
| 2285 |
-
"loss": 0.0,
|
| 2286 |
-
"step": 3570
|
| 2287 |
-
},
|
| 2288 |
-
{
|
| 2289 |
-
"epoch": 9.21,
|
| 2290 |
-
"learning_rate": 3.706349206349206e-05,
|
| 2291 |
-
"loss": 0.0,
|
| 2292 |
-
"step": 3580
|
| 2293 |
-
},
|
| 2294 |
-
{
|
| 2295 |
-
"epoch": 9.23,
|
| 2296 |
-
"learning_rate": 3.6269841269841266e-05,
|
| 2297 |
-
"loss": 0.0,
|
| 2298 |
-
"step": 3590
|
| 2299 |
-
},
|
| 2300 |
-
{
|
| 2301 |
-
"epoch": 9.26,
|
| 2302 |
-
"learning_rate": 3.5476190476190475e-05,
|
| 2303 |
-
"loss": 0.0,
|
| 2304 |
-
"step": 3600
|
| 2305 |
-
},
|
| 2306 |
-
{
|
| 2307 |
-
"epoch": 9.26,
|
| 2308 |
-
"eval_loss": NaN,
|
| 2309 |
-
"eval_runtime": 164.9466,
|
| 2310 |
-
"eval_samples_per_second": 12.125,
|
| 2311 |
-
"eval_steps_per_second": 1.516,
|
| 2312 |
-
"step": 3600
|
| 2313 |
-
},
|
| 2314 |
-
{
|
| 2315 |
-
"epoch": 9.29,
|
| 2316 |
-
"learning_rate": 3.4682539682539676e-05,
|
| 2317 |
-
"loss": 0.0,
|
| 2318 |
-
"step": 3610
|
| 2319 |
-
},
|
| 2320 |
-
{
|
| 2321 |
-
"epoch": 9.31,
|
| 2322 |
-
"learning_rate": 3.3888888888888884e-05,
|
| 2323 |
-
"loss": 0.0,
|
| 2324 |
-
"step": 3620
|
| 2325 |
-
},
|
| 2326 |
-
{
|
| 2327 |
-
"epoch": 9.34,
|
| 2328 |
-
"learning_rate": 3.309523809523809e-05,
|
| 2329 |
-
"loss": 0.0,
|
| 2330 |
-
"step": 3630
|
| 2331 |
-
},
|
| 2332 |
-
{
|
| 2333 |
-
"epoch": 9.36,
|
| 2334 |
-
"learning_rate": 3.23015873015873e-05,
|
| 2335 |
-
"loss": 0.0,
|
| 2336 |
-
"step": 3640
|
| 2337 |
-
},
|
| 2338 |
-
{
|
| 2339 |
-
"epoch": 9.39,
|
| 2340 |
-
"learning_rate": 3.15079365079365e-05,
|
| 2341 |
-
"loss": 0.0,
|
| 2342 |
-
"step": 3650
|
| 2343 |
-
},
|
| 2344 |
-
{
|
| 2345 |
-
"epoch": 9.41,
|
| 2346 |
-
"learning_rate": 3.071428571428571e-05,
|
| 2347 |
-
"loss": 0.0,
|
| 2348 |
-
"step": 3660
|
| 2349 |
-
},
|
| 2350 |
-
{
|
| 2351 |
-
"epoch": 9.44,
|
| 2352 |
-
"learning_rate": 2.992063492063492e-05,
|
| 2353 |
-
"loss": 0.0,
|
| 2354 |
-
"step": 3670
|
| 2355 |
-
},
|
| 2356 |
-
{
|
| 2357 |
-
"epoch": 9.47,
|
| 2358 |
-
"learning_rate": 2.9126984126984124e-05,
|
| 2359 |
-
"loss": 0.0,
|
| 2360 |
-
"step": 3680
|
| 2361 |
-
},
|
| 2362 |
-
{
|
| 2363 |
-
"epoch": 9.49,
|
| 2364 |
-
"learning_rate": 2.833333333333333e-05,
|
| 2365 |
-
"loss": 0.0,
|
| 2366 |
-
"step": 3690
|
| 2367 |
-
},
|
| 2368 |
-
{
|
| 2369 |
-
"epoch": 9.52,
|
| 2370 |
-
"learning_rate": 2.7539682539682537e-05,
|
| 2371 |
-
"loss": 0.0,
|
| 2372 |
-
"step": 3700
|
| 2373 |
-
},
|
| 2374 |
-
{
|
| 2375 |
-
"epoch": 9.54,
|
| 2376 |
-
"learning_rate": 2.6746031746031742e-05,
|
| 2377 |
-
"loss": 0.0,
|
| 2378 |
-
"step": 3710
|
| 2379 |
-
},
|
| 2380 |
-
{
|
| 2381 |
-
"epoch": 9.57,
|
| 2382 |
-
"learning_rate": 2.595238095238095e-05,
|
| 2383 |
-
"loss": 0.0,
|
| 2384 |
-
"step": 3720
|
| 2385 |
-
},
|
| 2386 |
-
{
|
| 2387 |
-
"epoch": 9.59,
|
| 2388 |
-
"learning_rate": 2.5158730158730155e-05,
|
| 2389 |
-
"loss": 0.0,
|
| 2390 |
-
"step": 3730
|
| 2391 |
-
},
|
| 2392 |
-
{
|
| 2393 |
-
"epoch": 9.62,
|
| 2394 |
-
"learning_rate": 2.4365079365079363e-05,
|
| 2395 |
-
"loss": 0.0,
|
| 2396 |
-
"step": 3740
|
| 2397 |
-
},
|
| 2398 |
-
{
|
| 2399 |
-
"epoch": 9.65,
|
| 2400 |
-
"learning_rate": 2.3571428571428568e-05,
|
| 2401 |
-
"loss": 0.0,
|
| 2402 |
-
"step": 3750
|
| 2403 |
-
},
|
| 2404 |
-
{
|
| 2405 |
-
"epoch": 9.67,
|
| 2406 |
-
"learning_rate": 2.2777777777777776e-05,
|
| 2407 |
-
"loss": 0.0,
|
| 2408 |
-
"step": 3760
|
| 2409 |
-
},
|
| 2410 |
-
{
|
| 2411 |
-
"epoch": 9.7,
|
| 2412 |
-
"learning_rate": 2.198412698412698e-05,
|
| 2413 |
-
"loss": 0.0,
|
| 2414 |
-
"step": 3770
|
| 2415 |
-
},
|
| 2416 |
-
{
|
| 2417 |
-
"epoch": 9.72,
|
| 2418 |
-
"learning_rate": 2.119047619047619e-05,
|
| 2419 |
-
"loss": 0.0,
|
| 2420 |
-
"step": 3780
|
| 2421 |
-
},
|
| 2422 |
-
{
|
| 2423 |
-
"epoch": 9.75,
|
| 2424 |
-
"learning_rate": 2.0396825396825394e-05,
|
| 2425 |
-
"loss": 0.0,
|
| 2426 |
-
"step": 3790
|
| 2427 |
-
},
|
| 2428 |
-
{
|
| 2429 |
-
"epoch": 9.77,
|
| 2430 |
-
"learning_rate": 1.9603174603174602e-05,
|
| 2431 |
-
"loss": 0.0,
|
| 2432 |
-
"step": 3800
|
| 2433 |
-
},
|
| 2434 |
-
{
|
| 2435 |
-
"epoch": 9.77,
|
| 2436 |
-
"eval_loss": NaN,
|
| 2437 |
-
"eval_runtime": 164.8991,
|
| 2438 |
-
"eval_samples_per_second": 12.129,
|
| 2439 |
-
"eval_steps_per_second": 1.516,
|
| 2440 |
-
"step": 3800
|
| 2441 |
-
}
|
| 2442 |
-
],
|
| 2443 |
-
"max_steps": 3880,
|
| 2444 |
-
"num_train_epochs": 10,
|
| 2445 |
-
"total_flos": 4.3886595913710305e+18,
|
| 2446 |
-
"trial_name": null,
|
| 2447 |
-
"trial_params": null
|
| 2448 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
checkpoint-3800/training_args.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:3e9adb78996a536c4aa514741768e2b05cafc3e20ac4a0a0fe98e38b91109396
|
| 3 |
-
size 3899
|
|
|
|
|
|
|
|
|
|
|
|
config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "decapoda-research/llama-7b-hf",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"LlamaForCausalLM"
|
| 5 |
+
],
|
| 6 |
+
"bos_token_id": 0,
|
| 7 |
+
"eos_token_id": 1,
|
| 8 |
+
"hidden_act": "silu",
|
| 9 |
+
"hidden_size": 4096,
|
| 10 |
+
"initializer_range": 0.02,
|
| 11 |
+
"intermediate_size": 11008,
|
| 12 |
+
"max_position_embeddings": 2048,
|
| 13 |
+
"max_sequence_length": 2048,
|
| 14 |
+
"model_type": "llama",
|
| 15 |
+
"num_attention_heads": 32,
|
| 16 |
+
"num_hidden_layers": 32,
|
| 17 |
+
"pad_token_id": -1,
|
| 18 |
+
"rms_norm_eps": 1e-06,
|
| 19 |
+
"tie_word_embeddings": false,
|
| 20 |
+
"torch_dtype": "float16",
|
| 21 |
+
"transformers_version": "4.29.2",
|
| 22 |
+
"use_cache": true,
|
| 23 |
+
"vocab_size": 32000
|
| 24 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 0,
|
| 4 |
+
"eos_token_id": 1,
|
| 5 |
+
"pad_token_id": 0,
|
| 6 |
+
"transformers_version": "4.29.2"
|
| 7 |
+
}
|
checkpoint-1400/rng_state.pth → pytorch_model-00001-of-00039.bin
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:975b4fcce1cc0f5d39f984b7ebcac7a505bb56623bc6eff75df7f381f0007f3e
|
| 3 |
+
size 396364479
|
adapter_model.bin → pytorch_model-00002-of-00039.bin
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:27283e5ff7771322644820763a40e77ee0010486bf6d2ff868b91b593368e54d
|
| 3 |
+
size 371215393
|
checkpoint-1400/optimizer.pt → pytorch_model-00003-of-00039.bin
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e5d680f0fd9b39af959244383207858b3267e2dcc52f1175ae2b4ee02ba89553
|
| 3 |
+
size 371215986
|
checkpoint-1400/pytorch_model.bin → pytorch_model-00004-of-00039.bin
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b473df902ef42eebd86da908b28c363106f9c6d941135a6f1e2f218cc60bc683
|
| 3 |
+
size 371215986
|
pytorch_model-00005-of-00039.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4a2a87142731bdcb6ad27a6a9c38414e3f05789b9d00fabdca75f9c917332171
|
| 3 |
+
size 371215986
|
pytorch_model-00006-of-00039.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:038201b714a5eeddbffc0f7d6401fa7d5105ff0c9480eaeeda9161c7bd3d7a2c
|
| 3 |
+
size 314575888
|
pytorch_model-00007-of-00039.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e89e8728cf12d0cdf2b802153930d91d2ef1839327b7de2b1a16494463e275f9
|
| 3 |
+
size 314592882
|
pytorch_model-00008-of-00039.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:934849ae9e6ef4da628697d02ba984b8ba57c0b3de2a6ebccd4801f26547b74a
|
| 3 |
+
size 314592882
|
pytorch_model-00009-of-00039.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:81e3b2e0d5f945e056c2e7e36f537c5a6630dd3c10525d931e913a04e86cf700
|
| 3 |
+
size 371215393
|
pytorch_model-00010-of-00039.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cc8bcc28c6492ebac9031ea9581290ac6eb08c485b55e5aaea4f0c3ae2d89d6a
|
| 3 |
+
size 371215986
|
pytorch_model-00011-of-00039.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5c5c844df0791f6483bdbb1c61579026529145ca7581eb58dbe33d468cbbad20
|
| 3 |
+
size 371215986
|
pytorch_model-00012-of-00039.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4c503b904ea8996ac79dcb7a8c4a7b5ecee5aa75192280f4ad5b30df9162030a
|
| 3 |
+
size 371215986
|
pytorch_model-00013-of-00039.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fd65605bdc1fa36223bd6df6e74077380b95f4183db2d139bd2b765041e25503
|
| 3 |
+
size 314575888
|
pytorch_model-00014-of-00039.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1ff0bb12cf5798c6fb85b7462cd4a83d1a395c354875fe739f628a693fc3eb04
|
| 3 |
+
size 314592882
|
pytorch_model-00015-of-00039.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bac5ee8a9a61324a31abfbee8c59daaa4c6292581c4e3f28db43686a43f3388d
|
| 3 |
+
size 314592882
|
pytorch_model-00016-of-00039.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:12e14c1ed24eab48141dc66e06c2e4cf4c1fa8b84d571f25f13f1994ea3c4932
|
| 3 |
+
size 371215393
|
pytorch_model-00017-of-00039.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a2c122693cd9e13cdc97eefe31237f04b1aa689bb934c4411465b10400d26c0a
|
| 3 |
+
size 371215986
|
pytorch_model-00018-of-00039.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:be7157b8619d9ee86c047bb142e6a3a7a7e4c0401c733e7446a1b5c10a5b4d3c
|
| 3 |
+
size 371215986
|
pytorch_model-00019-of-00039.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c736d13013d2fd3f627c4068c884950e6fa78fadf08b60afd6162288297a8dd9
|
| 3 |
+
size 371215986
|
pytorch_model-00020-of-00039.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e04f1429986b18496845282f36b032afb172f5172aba0c3fa825a714024f1e34
|
| 3 |
+
size 314575888
|
pytorch_model-00021-of-00039.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d780c33c5bed6d8fac81d50538c06a3348dd34e7ad12c3f8cc61993a1866dbd1
|
| 3 |
+
size 314592882
|
pytorch_model-00022-of-00039.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:822a8b3a34f79d5ca9ad4fa070b700ebe1b687d275b280d19e4b9c1705c586a7
|
| 3 |
+
size 314592882
|
pytorch_model-00023-of-00039.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f44dade49e92d3706ac1207d25b36f19b1fcf150f46deb07ee22d74184050bfb
|
| 3 |
+
size 371215393
|
pytorch_model-00024-of-00039.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5b6d9eff68d0909e909eed722bbe2e263df705ff486c446472c737a2c0fc31a2
|
| 3 |
+
size 371215986
|
pytorch_model-00025-of-00039.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dc292c68330f900426448a4b661ae593347db7cdc664cc354fa965ff1c608e66
|
| 3 |
+
size 371215986
|
pytorch_model-00026-of-00039.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3f9f1a1f95473652f893bd5f339cd6d6d9f75c3d81e525009b5898a6c21e784f
|
| 3 |
+
size 371215986
|
pytorch_model-00027-of-00039.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:63775c85e6504ea5d5d024f7029e90c3a4b9d66079573bd1c36284ce390b7037
|
| 3 |
+
size 314575888
|
pytorch_model-00028-of-00039.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:df35c532086f4e6b6d097bdd038afdf3bd4e271240ededf03d2747bd0544fd6f
|
| 3 |
+
size 314592882
|
pytorch_model-00029-of-00039.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6d4b8da74b497dc675f74e756426fe1cd6384e8df50fe7d392a2e0eb4c4f06fd
|
| 3 |
+
size 314592882
|