Training in progress, step 150, checkpoint
Browse files
last-checkpoint/adapter_config.json
CHANGED
|
@@ -20,12 +20,12 @@
|
|
| 20 |
"rank_pattern": {},
|
| 21 |
"revision": null,
|
| 22 |
"target_modules": [
|
| 23 |
-
"lm_head",
|
| 24 |
"dense",
|
| 25 |
-
"query_key_value",
|
| 26 |
-
"dense_4h_to_h",
|
| 27 |
"word_embeddings",
|
| 28 |
-
"
|
|
|
|
|
|
|
|
|
|
| 29 |
],
|
| 30 |
"task_type": "CAUSAL_LM",
|
| 31 |
"use_dora": false,
|
|
|
|
| 20 |
"rank_pattern": {},
|
| 21 |
"revision": null,
|
| 22 |
"target_modules": [
|
|
|
|
| 23 |
"dense",
|
|
|
|
|
|
|
| 24 |
"word_embeddings",
|
| 25 |
+
"lm_head",
|
| 26 |
+
"dense_h_to_4h",
|
| 27 |
+
"dense_4h_to_h",
|
| 28 |
+
"query_key_value"
|
| 29 |
],
|
| 30 |
"task_type": "CAUSAL_LM",
|
| 31 |
"use_dora": false,
|
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2083942096
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:16fc430ea02afa629e866a5346641b6836e8f0a1231a21e35108a75b62918c10
|
| 3 |
size 2083942096
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 57574138
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:df6dda378d442e7a76885cd08dbfbaa9ba73630ffbd0a668c07fc0c4eb07ccf6
|
| 3 |
size 57574138
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -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 sha256:04d21b9a22376a76ea15a7649e8a85c3cec7dfde7d72a28d681a78aaa424be2e
|
| 3 |
+
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:acbf40b939fcabf512d1392be3cf484e890b0399ab711748f7ead69b1e6e3930
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,2159 +1,125 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric":
|
| 3 |
-
"best_model_checkpoint": "./output/checkpoint-
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 150,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.007716049382716049,
|
| 13 |
-
"grad_norm":
|
| 14 |
"learning_rate": 8.000000000000001e-06,
|
| 15 |
-
"loss":
|
| 16 |
"step": 10
|
| 17 |
},
|
| 18 |
{
|
| 19 |
"epoch": 0.015432098765432098,
|
| 20 |
-
"grad_norm":
|
| 21 |
"learning_rate": 1.6000000000000003e-05,
|
| 22 |
-
"loss": 4.
|
| 23 |
"step": 20
|
| 24 |
},
|
| 25 |
{
|
| 26 |
"epoch": 0.023148148148148147,
|
| 27 |
-
"grad_norm":
|
| 28 |
"learning_rate": 2.4e-05,
|
| 29 |
-
"loss": 4.
|
| 30 |
"step": 30
|
| 31 |
},
|
| 32 |
{
|
| 33 |
"epoch": 0.030864197530864196,
|
| 34 |
-
"grad_norm":
|
| 35 |
"learning_rate": 3.2000000000000005e-05,
|
| 36 |
-
"loss": 4.
|
| 37 |
"step": 40
|
| 38 |
},
|
| 39 |
{
|
| 40 |
"epoch": 0.038580246913580245,
|
| 41 |
-
"grad_norm":
|
| 42 |
"learning_rate": 4e-05,
|
| 43 |
-
"loss": 4.
|
| 44 |
"step": 50
|
| 45 |
},
|
| 46 |
{
|
| 47 |
"epoch": 0.046296296296296294,
|
| 48 |
-
"grad_norm":
|
| 49 |
"learning_rate": 4.8e-05,
|
| 50 |
-
"loss": 4.
|
| 51 |
"step": 60
|
| 52 |
},
|
| 53 |
{
|
| 54 |
"epoch": 0.05401234567901234,
|
| 55 |
-
"grad_norm":
|
| 56 |
"learning_rate": 5.6e-05,
|
| 57 |
-
"loss": 4.
|
| 58 |
"step": 70
|
| 59 |
},
|
| 60 |
{
|
| 61 |
"epoch": 0.06172839506172839,
|
| 62 |
-
"grad_norm":
|
| 63 |
"learning_rate": 6.400000000000001e-05,
|
| 64 |
-
"loss": 4.
|
| 65 |
"step": 80
|
| 66 |
},
|
| 67 |
{
|
| 68 |
"epoch": 0.06944444444444445,
|
| 69 |
-
"grad_norm":
|
| 70 |
"learning_rate": 7.2e-05,
|
| 71 |
-
"loss": 4.
|
| 72 |
"step": 90
|
| 73 |
},
|
| 74 |
{
|
| 75 |
"epoch": 0.07716049382716049,
|
| 76 |
-
"grad_norm":
|
| 77 |
"learning_rate": 8e-05,
|
| 78 |
-
"loss": 4.
|
| 79 |
"step": 100
|
| 80 |
},
|
| 81 |
{
|
| 82 |
"epoch": 0.08487654320987655,
|
| 83 |
-
"grad_norm":
|
| 84 |
"learning_rate": 7.999917787833465e-05,
|
| 85 |
-
"loss": 4.
|
| 86 |
"step": 110
|
| 87 |
},
|
| 88 |
{
|
| 89 |
"epoch": 0.09259259259259259,
|
| 90 |
-
"grad_norm":
|
| 91 |
"learning_rate": 7.999671154713278e-05,
|
| 92 |
-
"loss": 4.
|
| 93 |
"step": 120
|
| 94 |
},
|
| 95 |
{
|
| 96 |
"epoch": 0.10030864197530864,
|
| 97 |
-
"grad_norm":
|
| 98 |
"learning_rate": 7.99926011077756e-05,
|
| 99 |
-
"loss": 4.
|
| 100 |
"step": 130
|
| 101 |
},
|
| 102 |
{
|
| 103 |
"epoch": 0.10802469135802469,
|
| 104 |
-
"grad_norm":
|
| 105 |
"learning_rate": 7.99868467292272e-05,
|
| 106 |
-
"loss": 4.
|
| 107 |
"step": 140
|
| 108 |
},
|
| 109 |
{
|
| 110 |
"epoch": 0.11574074074074074,
|
| 111 |
-
"grad_norm":
|
| 112 |
"learning_rate": 7.997944864802752e-05,
|
| 113 |
-
"loss": 4.
|
| 114 |
"step": 150
|
| 115 |
},
|
| 116 |
{
|
| 117 |
"epoch": 0.11574074074074074,
|
| 118 |
-
"eval_loss": 4.
|
| 119 |
-
"eval_runtime": 17.
|
| 120 |
-
"eval_samples_per_second": 28.
|
| 121 |
-
"eval_steps_per_second": 28.
|
| 122 |
"step": 150
|
| 123 |
-
},
|
| 124 |
-
{
|
| 125 |
-
"epoch": 0.12345679012345678,
|
| 126 |
-
"grad_norm": 16.790706634521484,
|
| 127 |
-
"learning_rate": 7.997040716828271e-05,
|
| 128 |
-
"loss": 4.2613,
|
| 129 |
-
"step": 160
|
| 130 |
-
},
|
| 131 |
-
{
|
| 132 |
-
"epoch": 0.13117283950617284,
|
| 133 |
-
"grad_norm": 18.06231117248535,
|
| 134 |
-
"learning_rate": 7.995972266165259e-05,
|
| 135 |
-
"loss": 4.2768,
|
| 136 |
-
"step": 170
|
| 137 |
-
},
|
| 138 |
-
{
|
| 139 |
-
"epoch": 0.1388888888888889,
|
| 140 |
-
"grad_norm": 22.725183486938477,
|
| 141 |
-
"learning_rate": 7.994739556733538e-05,
|
| 142 |
-
"loss": 4.1748,
|
| 143 |
-
"step": 180
|
| 144 |
-
},
|
| 145 |
-
{
|
| 146 |
-
"epoch": 0.14660493827160495,
|
| 147 |
-
"grad_norm": 25.361064910888672,
|
| 148 |
-
"learning_rate": 7.993342639204965e-05,
|
| 149 |
-
"loss": 4.2136,
|
| 150 |
-
"step": 190
|
| 151 |
-
},
|
| 152 |
-
{
|
| 153 |
-
"epoch": 0.15432098765432098,
|
| 154 |
-
"grad_norm": 15.649086952209473,
|
| 155 |
-
"learning_rate": 7.991781571001347e-05,
|
| 156 |
-
"loss": 4.2655,
|
| 157 |
-
"step": 200
|
| 158 |
-
},
|
| 159 |
-
{
|
| 160 |
-
"epoch": 0.16203703703703703,
|
| 161 |
-
"grad_norm": 13.647220611572266,
|
| 162 |
-
"learning_rate": 7.990056416292084e-05,
|
| 163 |
-
"loss": 4.1017,
|
| 164 |
-
"step": 210
|
| 165 |
-
},
|
| 166 |
-
{
|
| 167 |
-
"epoch": 0.1697530864197531,
|
| 168 |
-
"grad_norm": 45.323265075683594,
|
| 169 |
-
"learning_rate": 7.988167245991528e-05,
|
| 170 |
-
"loss": 4.1013,
|
| 171 |
-
"step": 220
|
| 172 |
-
},
|
| 173 |
-
{
|
| 174 |
-
"epoch": 0.17746913580246915,
|
| 175 |
-
"grad_norm": 15.971923828125,
|
| 176 |
-
"learning_rate": 7.986114137756074e-05,
|
| 177 |
-
"loss": 4.2096,
|
| 178 |
-
"step": 230
|
| 179 |
-
},
|
| 180 |
-
{
|
| 181 |
-
"epoch": 0.18518518518518517,
|
| 182 |
-
"grad_norm": 16.353431701660156,
|
| 183 |
-
"learning_rate": 7.983897175980957e-05,
|
| 184 |
-
"loss": 4.1852,
|
| 185 |
-
"step": 240
|
| 186 |
-
},
|
| 187 |
-
{
|
| 188 |
-
"epoch": 0.19290123456790123,
|
| 189 |
-
"grad_norm": 16.62486457824707,
|
| 190 |
-
"learning_rate": 7.981516451796794e-05,
|
| 191 |
-
"loss": 4.2644,
|
| 192 |
-
"step": 250
|
| 193 |
-
},
|
| 194 |
-
{
|
| 195 |
-
"epoch": 0.2006172839506173,
|
| 196 |
-
"grad_norm": 11.310361862182617,
|
| 197 |
-
"learning_rate": 7.97897206306583e-05,
|
| 198 |
-
"loss": 4.1296,
|
| 199 |
-
"step": 260
|
| 200 |
-
},
|
| 201 |
-
{
|
| 202 |
-
"epoch": 0.20833333333333334,
|
| 203 |
-
"grad_norm": 17.337621688842773,
|
| 204 |
-
"learning_rate": 7.976264114377922e-05,
|
| 205 |
-
"loss": 4.1704,
|
| 206 |
-
"step": 270
|
| 207 |
-
},
|
| 208 |
-
{
|
| 209 |
-
"epoch": 0.21604938271604937,
|
| 210 |
-
"grad_norm": 15.199752807617188,
|
| 211 |
-
"learning_rate": 7.973392717046233e-05,
|
| 212 |
-
"loss": 4.1354,
|
| 213 |
-
"step": 280
|
| 214 |
-
},
|
| 215 |
-
{
|
| 216 |
-
"epoch": 0.22376543209876543,
|
| 217 |
-
"grad_norm": 18.04800796508789,
|
| 218 |
-
"learning_rate": 7.97035798910266e-05,
|
| 219 |
-
"loss": 4.0614,
|
| 220 |
-
"step": 290
|
| 221 |
-
},
|
| 222 |
-
{
|
| 223 |
-
"epoch": 0.23148148148148148,
|
| 224 |
-
"grad_norm": 14.9419527053833,
|
| 225 |
-
"learning_rate": 7.967160055292984e-05,
|
| 226 |
-
"loss": 4.3118,
|
| 227 |
-
"step": 300
|
| 228 |
-
},
|
| 229 |
-
{
|
| 230 |
-
"epoch": 0.23148148148148148,
|
| 231 |
-
"eval_loss": 4.240542888641357,
|
| 232 |
-
"eval_runtime": 20.5897,
|
| 233 |
-
"eval_samples_per_second": 24.284,
|
| 234 |
-
"eval_steps_per_second": 24.284,
|
| 235 |
-
"step": 300
|
| 236 |
-
},
|
| 237 |
-
{
|
| 238 |
-
"epoch": 0.23919753086419754,
|
| 239 |
-
"grad_norm": 19.11252212524414,
|
| 240 |
-
"learning_rate": 7.96379904707174e-05,
|
| 241 |
-
"loss": 4.1267,
|
| 242 |
-
"step": 310
|
| 243 |
-
},
|
| 244 |
-
{
|
| 245 |
-
"epoch": 0.24691358024691357,
|
| 246 |
-
"grad_norm": 12.28123950958252,
|
| 247 |
-
"learning_rate": 7.960275102596809e-05,
|
| 248 |
-
"loss": 4.2135,
|
| 249 |
-
"step": 320
|
| 250 |
-
},
|
| 251 |
-
{
|
| 252 |
-
"epoch": 0.25462962962962965,
|
| 253 |
-
"grad_norm": 30.027751922607422,
|
| 254 |
-
"learning_rate": 7.956588366723745e-05,
|
| 255 |
-
"loss": 4.2535,
|
| 256 |
-
"step": 330
|
| 257 |
-
},
|
| 258 |
-
{
|
| 259 |
-
"epoch": 0.2623456790123457,
|
| 260 |
-
"grad_norm": 11.198431015014648,
|
| 261 |
-
"learning_rate": 7.952738990999824e-05,
|
| 262 |
-
"loss": 4.2223,
|
| 263 |
-
"step": 340
|
| 264 |
-
},
|
| 265 |
-
{
|
| 266 |
-
"epoch": 0.2700617283950617,
|
| 267 |
-
"grad_norm": 24.937889099121094,
|
| 268 |
-
"learning_rate": 7.948727133657802e-05,
|
| 269 |
-
"loss": 4.0935,
|
| 270 |
-
"step": 350
|
| 271 |
-
},
|
| 272 |
-
{
|
| 273 |
-
"epoch": 0.2777777777777778,
|
| 274 |
-
"grad_norm": 14.272343635559082,
|
| 275 |
-
"learning_rate": 7.94455295960942e-05,
|
| 276 |
-
"loss": 4.2069,
|
| 277 |
-
"step": 360
|
| 278 |
-
},
|
| 279 |
-
{
|
| 280 |
-
"epoch": 0.2854938271604938,
|
| 281 |
-
"grad_norm": 17.262426376342773,
|
| 282 |
-
"learning_rate": 7.940216640438628e-05,
|
| 283 |
-
"loss": 4.2168,
|
| 284 |
-
"step": 370
|
| 285 |
-
},
|
| 286 |
-
{
|
| 287 |
-
"epoch": 0.2932098765432099,
|
| 288 |
-
"grad_norm": 27.980453491210938,
|
| 289 |
-
"learning_rate": 7.93571835439452e-05,
|
| 290 |
-
"loss": 4.1711,
|
| 291 |
-
"step": 380
|
| 292 |
-
},
|
| 293 |
-
{
|
| 294 |
-
"epoch": 0.30092592592592593,
|
| 295 |
-
"grad_norm": 17.98458480834961,
|
| 296 |
-
"learning_rate": 7.931058286384016e-05,
|
| 297 |
-
"loss": 4.2067,
|
| 298 |
-
"step": 390
|
| 299 |
-
},
|
| 300 |
-
{
|
| 301 |
-
"epoch": 0.30864197530864196,
|
| 302 |
-
"grad_norm": 21.41556167602539,
|
| 303 |
-
"learning_rate": 7.926236627964262e-05,
|
| 304 |
-
"loss": 4.0364,
|
| 305 |
-
"step": 400
|
| 306 |
-
},
|
| 307 |
-
{
|
| 308 |
-
"epoch": 0.31635802469135804,
|
| 309 |
-
"grad_norm": 16.788623809814453,
|
| 310 |
-
"learning_rate": 7.92125357733475e-05,
|
| 311 |
-
"loss": 4.1445,
|
| 312 |
-
"step": 410
|
| 313 |
-
},
|
| 314 |
-
{
|
| 315 |
-
"epoch": 0.32407407407407407,
|
| 316 |
-
"grad_norm": 29.33296775817871,
|
| 317 |
-
"learning_rate": 7.916109339329173e-05,
|
| 318 |
-
"loss": 4.0973,
|
| 319 |
-
"step": 420
|
| 320 |
-
},
|
| 321 |
-
{
|
| 322 |
-
"epoch": 0.3317901234567901,
|
| 323 |
-
"grad_norm": 15.776802062988281,
|
| 324 |
-
"learning_rate": 7.910804125407007e-05,
|
| 325 |
-
"loss": 3.9347,
|
| 326 |
-
"step": 430
|
| 327 |
-
},
|
| 328 |
-
{
|
| 329 |
-
"epoch": 0.3395061728395062,
|
| 330 |
-
"grad_norm": 18.816022872924805,
|
| 331 |
-
"learning_rate": 7.905338153644818e-05,
|
| 332 |
-
"loss": 4.1418,
|
| 333 |
-
"step": 440
|
| 334 |
-
},
|
| 335 |
-
{
|
| 336 |
-
"epoch": 0.3472222222222222,
|
| 337 |
-
"grad_norm": 19.884754180908203,
|
| 338 |
-
"learning_rate": 7.899711648727294e-05,
|
| 339 |
-
"loss": 4.0071,
|
| 340 |
-
"step": 450
|
| 341 |
-
},
|
| 342 |
-
{
|
| 343 |
-
"epoch": 0.3472222222222222,
|
| 344 |
-
"eval_loss": 4.1484150886535645,
|
| 345 |
-
"eval_runtime": 18.0768,
|
| 346 |
-
"eval_samples_per_second": 27.66,
|
| 347 |
-
"eval_steps_per_second": 27.66,
|
| 348 |
-
"step": 450
|
| 349 |
-
},
|
| 350 |
-
{
|
| 351 |
-
"epoch": 0.3549382716049383,
|
| 352 |
-
"grad_norm": 23.56029510498047,
|
| 353 |
-
"learning_rate": 7.89392484193802e-05,
|
| 354 |
-
"loss": 4.1422,
|
| 355 |
-
"step": 460
|
| 356 |
-
},
|
| 357 |
-
{
|
| 358 |
-
"epoch": 0.3626543209876543,
|
| 359 |
-
"grad_norm": 18.750913619995117,
|
| 360 |
-
"learning_rate": 7.887977971149952e-05,
|
| 361 |
-
"loss": 4.114,
|
| 362 |
-
"step": 470
|
| 363 |
-
},
|
| 364 |
-
{
|
| 365 |
-
"epoch": 0.37037037037037035,
|
| 366 |
-
"grad_norm": 17.39931297302246,
|
| 367 |
-
"learning_rate": 7.881871280815659e-05,
|
| 368 |
-
"loss": 4.1022,
|
| 369 |
-
"step": 480
|
| 370 |
-
},
|
| 371 |
-
{
|
| 372 |
-
"epoch": 0.37808641975308643,
|
| 373 |
-
"grad_norm": 15.805721282958984,
|
| 374 |
-
"learning_rate": 7.875605021957262e-05,
|
| 375 |
-
"loss": 4.1326,
|
| 376 |
-
"step": 490
|
| 377 |
-
},
|
| 378 |
-
{
|
| 379 |
-
"epoch": 0.38580246913580246,
|
| 380 |
-
"grad_norm": 15.227653503417969,
|
| 381 |
-
"learning_rate": 7.869179452156118e-05,
|
| 382 |
-
"loss": 3.9761,
|
| 383 |
-
"step": 500
|
| 384 |
-
},
|
| 385 |
-
{
|
| 386 |
-
"epoch": 0.39351851851851855,
|
| 387 |
-
"grad_norm": 24.661602020263672,
|
| 388 |
-
"learning_rate": 7.862594835542236e-05,
|
| 389 |
-
"loss": 4.1084,
|
| 390 |
-
"step": 510
|
| 391 |
-
},
|
| 392 |
-
{
|
| 393 |
-
"epoch": 0.4012345679012346,
|
| 394 |
-
"grad_norm": 18.671737670898438,
|
| 395 |
-
"learning_rate": 7.855851442783414e-05,
|
| 396 |
-
"loss": 3.9883,
|
| 397 |
-
"step": 520
|
| 398 |
-
},
|
| 399 |
-
{
|
| 400 |
-
"epoch": 0.4089506172839506,
|
| 401 |
-
"grad_norm": 16.544588088989258,
|
| 402 |
-
"learning_rate": 7.848949551074116e-05,
|
| 403 |
-
"loss": 4.106,
|
| 404 |
-
"step": 530
|
| 405 |
-
},
|
| 406 |
-
{
|
| 407 |
-
"epoch": 0.4166666666666667,
|
| 408 |
-
"grad_norm": 27.335189819335938,
|
| 409 |
-
"learning_rate": 7.841889444124078e-05,
|
| 410 |
-
"loss": 4.0624,
|
| 411 |
-
"step": 540
|
| 412 |
-
},
|
| 413 |
-
{
|
| 414 |
-
"epoch": 0.4243827160493827,
|
| 415 |
-
"grad_norm": 21.533105850219727,
|
| 416 |
-
"learning_rate": 7.834671412146643e-05,
|
| 417 |
-
"loss": 4.1846,
|
| 418 |
-
"step": 550
|
| 419 |
-
},
|
| 420 |
-
{
|
| 421 |
-
"epoch": 0.43209876543209874,
|
| 422 |
-
"grad_norm": 13.441323280334473,
|
| 423 |
-
"learning_rate": 7.827295751846836e-05,
|
| 424 |
-
"loss": 4.1276,
|
| 425 |
-
"step": 560
|
| 426 |
-
},
|
| 427 |
-
{
|
| 428 |
-
"epoch": 0.4398148148148148,
|
| 429 |
-
"grad_norm": 23.776737213134766,
|
| 430 |
-
"learning_rate": 7.819762766409162e-05,
|
| 431 |
-
"loss": 4.0718,
|
| 432 |
-
"step": 570
|
| 433 |
-
},
|
| 434 |
-
{
|
| 435 |
-
"epoch": 0.44753086419753085,
|
| 436 |
-
"grad_norm": 14.89003849029541,
|
| 437 |
-
"learning_rate": 7.81207276548515e-05,
|
| 438 |
-
"loss": 4.0092,
|
| 439 |
-
"step": 580
|
| 440 |
-
},
|
| 441 |
-
{
|
| 442 |
-
"epoch": 0.45524691358024694,
|
| 443 |
-
"grad_norm": 18.61737632751465,
|
| 444 |
-
"learning_rate": 7.804226065180615e-05,
|
| 445 |
-
"loss": 3.9711,
|
| 446 |
-
"step": 590
|
| 447 |
-
},
|
| 448 |
-
{
|
| 449 |
-
"epoch": 0.46296296296296297,
|
| 450 |
-
"grad_norm": 22.550548553466797,
|
| 451 |
-
"learning_rate": 7.796222988042676e-05,
|
| 452 |
-
"loss": 3.8619,
|
| 453 |
-
"step": 600
|
| 454 |
-
},
|
| 455 |
-
{
|
| 456 |
-
"epoch": 0.46296296296296297,
|
| 457 |
-
"eval_loss": 4.064351558685303,
|
| 458 |
-
"eval_runtime": 20.247,
|
| 459 |
-
"eval_samples_per_second": 24.695,
|
| 460 |
-
"eval_steps_per_second": 24.695,
|
| 461 |
-
"step": 600
|
| 462 |
-
},
|
| 463 |
-
{
|
| 464 |
-
"epoch": 0.470679012345679,
|
| 465 |
-
"grad_norm": 24.837818145751953,
|
| 466 |
-
"learning_rate": 7.788063863046486e-05,
|
| 467 |
-
"loss": 4.0238,
|
| 468 |
-
"step": 610
|
| 469 |
-
},
|
| 470 |
-
{
|
| 471 |
-
"epoch": 0.4783950617283951,
|
| 472 |
-
"grad_norm": 17.935625076293945,
|
| 473 |
-
"learning_rate": 7.779749025581717e-05,
|
| 474 |
-
"loss": 4.035,
|
| 475 |
-
"step": 620
|
| 476 |
-
},
|
| 477 |
-
{
|
| 478 |
-
"epoch": 0.4861111111111111,
|
| 479 |
-
"grad_norm": 18.532503128051758,
|
| 480 |
-
"learning_rate": 7.771278817438773e-05,
|
| 481 |
-
"loss": 4.0214,
|
| 482 |
-
"step": 630
|
| 483 |
-
},
|
| 484 |
-
{
|
| 485 |
-
"epoch": 0.49382716049382713,
|
| 486 |
-
"grad_norm": 18.070255279541016,
|
| 487 |
-
"learning_rate": 7.762653586794731e-05,
|
| 488 |
-
"loss": 4.0554,
|
| 489 |
-
"step": 640
|
| 490 |
-
},
|
| 491 |
-
{
|
| 492 |
-
"epoch": 0.5015432098765432,
|
| 493 |
-
"grad_norm": 19.590288162231445,
|
| 494 |
-
"learning_rate": 7.753873688199042e-05,
|
| 495 |
-
"loss": 3.9853,
|
| 496 |
-
"step": 650
|
| 497 |
-
},
|
| 498 |
-
{
|
| 499 |
-
"epoch": 0.5092592592592593,
|
| 500 |
-
"grad_norm": 19.89542007446289,
|
| 501 |
-
"learning_rate": 7.74493948255895e-05,
|
| 502 |
-
"loss": 4.0259,
|
| 503 |
-
"step": 660
|
| 504 |
-
},
|
| 505 |
-
{
|
| 506 |
-
"epoch": 0.5169753086419753,
|
| 507 |
-
"grad_norm": 14.403157234191895,
|
| 508 |
-
"learning_rate": 7.735851337124654e-05,
|
| 509 |
-
"loss": 3.9273,
|
| 510 |
-
"step": 670
|
| 511 |
-
},
|
| 512 |
-
{
|
| 513 |
-
"epoch": 0.5246913580246914,
|
| 514 |
-
"grad_norm": 22.67559051513672,
|
| 515 |
-
"learning_rate": 7.726609625474218e-05,
|
| 516 |
-
"loss": 4.0133,
|
| 517 |
-
"step": 680
|
| 518 |
-
},
|
| 519 |
-
{
|
| 520 |
-
"epoch": 0.5324074074074074,
|
| 521 |
-
"grad_norm": 11.595963478088379,
|
| 522 |
-
"learning_rate": 7.717214727498209e-05,
|
| 523 |
-
"loss": 3.9475,
|
| 524 |
-
"step": 690
|
| 525 |
-
},
|
| 526 |
-
{
|
| 527 |
-
"epoch": 0.5401234567901234,
|
| 528 |
-
"grad_norm": 17.296403884887695,
|
| 529 |
-
"learning_rate": 7.707667029384088e-05,
|
| 530 |
-
"loss": 3.8421,
|
| 531 |
-
"step": 700
|
| 532 |
-
},
|
| 533 |
-
{
|
| 534 |
-
"epoch": 0.5478395061728395,
|
| 535 |
-
"grad_norm": 21.876863479614258,
|
| 536 |
-
"learning_rate": 7.697966923600327e-05,
|
| 537 |
-
"loss": 3.9917,
|
| 538 |
-
"step": 710
|
| 539 |
-
},
|
| 540 |
-
{
|
| 541 |
-
"epoch": 0.5555555555555556,
|
| 542 |
-
"grad_norm": 17.589155197143555,
|
| 543 |
-
"learning_rate": 7.688114808880283e-05,
|
| 544 |
-
"loss": 3.9218,
|
| 545 |
-
"step": 720
|
| 546 |
-
},
|
| 547 |
-
{
|
| 548 |
-
"epoch": 0.5632716049382716,
|
| 549 |
-
"grad_norm": 17.723289489746094,
|
| 550 |
-
"learning_rate": 7.678111090205804e-05,
|
| 551 |
-
"loss": 3.9183,
|
| 552 |
-
"step": 730
|
| 553 |
-
},
|
| 554 |
-
{
|
| 555 |
-
"epoch": 0.5709876543209876,
|
| 556 |
-
"grad_norm": 16.053651809692383,
|
| 557 |
-
"learning_rate": 7.667956178790582e-05,
|
| 558 |
-
"loss": 4.0021,
|
| 559 |
-
"step": 740
|
| 560 |
-
},
|
| 561 |
-
{
|
| 562 |
-
"epoch": 0.5787037037037037,
|
| 563 |
-
"grad_norm": 17.192163467407227,
|
| 564 |
-
"learning_rate": 7.65765049206325e-05,
|
| 565 |
-
"loss": 3.9775,
|
| 566 |
-
"step": 750
|
| 567 |
-
},
|
| 568 |
-
{
|
| 569 |
-
"epoch": 0.5787037037037037,
|
| 570 |
-
"eval_loss": 3.9868149757385254,
|
| 571 |
-
"eval_runtime": 17.6035,
|
| 572 |
-
"eval_samples_per_second": 28.403,
|
| 573 |
-
"eval_steps_per_second": 28.403,
|
| 574 |
-
"step": 750
|
| 575 |
-
},
|
| 576 |
-
{
|
| 577 |
-
"epoch": 0.5864197530864198,
|
| 578 |
-
"grad_norm": 29.302610397338867,
|
| 579 |
-
"learning_rate": 7.647194453650228e-05,
|
| 580 |
-
"loss": 4.0921,
|
| 581 |
-
"step": 760
|
| 582 |
-
},
|
| 583 |
-
{
|
| 584 |
-
"epoch": 0.5941358024691358,
|
| 585 |
-
"grad_norm": 19.50672721862793,
|
| 586 |
-
"learning_rate": 7.6365884933583e-05,
|
| 587 |
-
"loss": 3.7631,
|
| 588 |
-
"step": 770
|
| 589 |
-
},
|
| 590 |
-
{
|
| 591 |
-
"epoch": 0.6018518518518519,
|
| 592 |
-
"grad_norm": 25.308528900146484,
|
| 593 |
-
"learning_rate": 7.625833047156953e-05,
|
| 594 |
-
"loss": 3.9659,
|
| 595 |
-
"step": 780
|
| 596 |
-
},
|
| 597 |
-
{
|
| 598 |
-
"epoch": 0.6095679012345679,
|
| 599 |
-
"grad_norm": 20.864965438842773,
|
| 600 |
-
"learning_rate": 7.614928557160454e-05,
|
| 601 |
-
"loss": 3.902,
|
| 602 |
-
"step": 790
|
| 603 |
-
},
|
| 604 |
-
{
|
| 605 |
-
"epoch": 0.6172839506172839,
|
| 606 |
-
"grad_norm": 26.428804397583008,
|
| 607 |
-
"learning_rate": 7.603875471609677e-05,
|
| 608 |
-
"loss": 3.9137,
|
| 609 |
-
"step": 800
|
| 610 |
-
},
|
| 611 |
-
{
|
| 612 |
-
"epoch": 0.625,
|
| 613 |
-
"grad_norm": 18.294261932373047,
|
| 614 |
-
"learning_rate": 7.592674244853676e-05,
|
| 615 |
-
"loss": 3.9768,
|
| 616 |
-
"step": 810
|
| 617 |
-
},
|
| 618 |
-
{
|
| 619 |
-
"epoch": 0.6327160493827161,
|
| 620 |
-
"grad_norm": 13.396001815795898,
|
| 621 |
-
"learning_rate": 7.581325337331013e-05,
|
| 622 |
-
"loss": 3.979,
|
| 623 |
-
"step": 820
|
| 624 |
-
},
|
| 625 |
-
{
|
| 626 |
-
"epoch": 0.6404320987654321,
|
| 627 |
-
"grad_norm": 17.813318252563477,
|
| 628 |
-
"learning_rate": 7.569829215550825e-05,
|
| 629 |
-
"loss": 3.9316,
|
| 630 |
-
"step": 830
|
| 631 |
-
},
|
| 632 |
-
{
|
| 633 |
-
"epoch": 0.6481481481481481,
|
| 634 |
-
"grad_norm": 27.784278869628906,
|
| 635 |
-
"learning_rate": 7.558186352073648e-05,
|
| 636 |
-
"loss": 3.9523,
|
| 637 |
-
"step": 840
|
| 638 |
-
},
|
| 639 |
-
{
|
| 640 |
-
"epoch": 0.6558641975308642,
|
| 641 |
-
"grad_norm": 24.91390609741211,
|
| 642 |
-
"learning_rate": 7.546397225492001e-05,
|
| 643 |
-
"loss": 3.9183,
|
| 644 |
-
"step": 850
|
| 645 |
-
},
|
| 646 |
-
{
|
| 647 |
-
"epoch": 0.6635802469135802,
|
| 648 |
-
"grad_norm": 18.925460815429688,
|
| 649 |
-
"learning_rate": 7.534462320410702e-05,
|
| 650 |
-
"loss": 4.0088,
|
| 651 |
-
"step": 860
|
| 652 |
-
},
|
| 653 |
-
{
|
| 654 |
-
"epoch": 0.6712962962962963,
|
| 655 |
-
"grad_norm": 23.20028305053711,
|
| 656 |
-
"learning_rate": 7.522382127426952e-05,
|
| 657 |
-
"loss": 3.8521,
|
| 658 |
-
"step": 870
|
| 659 |
-
},
|
| 660 |
-
{
|
| 661 |
-
"epoch": 0.6790123456790124,
|
| 662 |
-
"grad_norm": 25.306800842285156,
|
| 663 |
-
"learning_rate": 7.510157143110172e-05,
|
| 664 |
-
"loss": 3.931,
|
| 665 |
-
"step": 880
|
| 666 |
-
},
|
| 667 |
-
{
|
| 668 |
-
"epoch": 0.6867283950617284,
|
| 669 |
-
"grad_norm": 14.286781311035156,
|
| 670 |
-
"learning_rate": 7.497787869981583e-05,
|
| 671 |
-
"loss": 3.7833,
|
| 672 |
-
"step": 890
|
| 673 |
-
},
|
| 674 |
-
{
|
| 675 |
-
"epoch": 0.6944444444444444,
|
| 676 |
-
"grad_norm": 27.823122024536133,
|
| 677 |
-
"learning_rate": 7.485274816493558e-05,
|
| 678 |
-
"loss": 3.929,
|
| 679 |
-
"step": 900
|
| 680 |
-
},
|
| 681 |
-
{
|
| 682 |
-
"epoch": 0.6944444444444444,
|
| 683 |
-
"eval_loss": 3.928668737411499,
|
| 684 |
-
"eval_runtime": 18.7881,
|
| 685 |
-
"eval_samples_per_second": 26.613,
|
| 686 |
-
"eval_steps_per_second": 26.613,
|
| 687 |
-
"step": 900
|
| 688 |
-
},
|
| 689 |
-
{
|
| 690 |
-
"epoch": 0.7021604938271605,
|
| 691 |
-
"grad_norm": 19.728225708007812,
|
| 692 |
-
"learning_rate": 7.472618497008714e-05,
|
| 693 |
-
"loss": 3.8653,
|
| 694 |
-
"step": 910
|
| 695 |
-
},
|
| 696 |
-
{
|
| 697 |
-
"epoch": 0.7098765432098766,
|
| 698 |
-
"grad_norm": 18.80716896057129,
|
| 699 |
-
"learning_rate": 7.459819431778775e-05,
|
| 700 |
-
"loss": 3.8751,
|
| 701 |
-
"step": 920
|
| 702 |
-
},
|
| 703 |
-
{
|
| 704 |
-
"epoch": 0.7175925925925926,
|
| 705 |
-
"grad_norm": 22.028133392333984,
|
| 706 |
-
"learning_rate": 7.44687814692318e-05,
|
| 707 |
-
"loss": 3.9937,
|
| 708 |
-
"step": 930
|
| 709 |
-
},
|
| 710 |
-
{
|
| 711 |
-
"epoch": 0.7253086419753086,
|
| 712 |
-
"grad_norm": 16.667814254760742,
|
| 713 |
-
"learning_rate": 7.433795174407465e-05,
|
| 714 |
-
"loss": 3.8102,
|
| 715 |
-
"step": 940
|
| 716 |
-
},
|
| 717 |
-
{
|
| 718 |
-
"epoch": 0.7330246913580247,
|
| 719 |
-
"grad_norm": 20.56685447692871,
|
| 720 |
-
"learning_rate": 7.420571052021386e-05,
|
| 721 |
-
"loss": 3.8229,
|
| 722 |
-
"step": 950
|
| 723 |
-
},
|
| 724 |
-
{
|
| 725 |
-
"epoch": 0.7407407407407407,
|
| 726 |
-
"grad_norm": 19.397369384765625,
|
| 727 |
-
"learning_rate": 7.407206323356818e-05,
|
| 728 |
-
"loss": 3.807,
|
| 729 |
-
"step": 960
|
| 730 |
-
},
|
| 731 |
-
{
|
| 732 |
-
"epoch": 0.7484567901234568,
|
| 733 |
-
"grad_norm": 19.5135498046875,
|
| 734 |
-
"learning_rate": 7.393701537785411e-05,
|
| 735 |
-
"loss": 4.049,
|
| 736 |
-
"step": 970
|
| 737 |
-
},
|
| 738 |
-
{
|
| 739 |
-
"epoch": 0.7561728395061729,
|
| 740 |
-
"grad_norm": 15.916143417358398,
|
| 741 |
-
"learning_rate": 7.380057250436006e-05,
|
| 742 |
-
"loss": 3.7301,
|
| 743 |
-
"step": 980
|
| 744 |
-
},
|
| 745 |
-
{
|
| 746 |
-
"epoch": 0.7638888888888888,
|
| 747 |
-
"grad_norm": 19.677715301513672,
|
| 748 |
-
"learning_rate": 7.366274022171814e-05,
|
| 749 |
-
"loss": 3.8505,
|
| 750 |
-
"step": 990
|
| 751 |
-
},
|
| 752 |
-
{
|
| 753 |
-
"epoch": 0.7716049382716049,
|
| 754 |
-
"grad_norm": 15.923828125,
|
| 755 |
-
"learning_rate": 7.352352419567362e-05,
|
| 756 |
-
"loss": 3.8464,
|
| 757 |
-
"step": 1000
|
| 758 |
-
},
|
| 759 |
-
{
|
| 760 |
-
"epoch": 0.779320987654321,
|
| 761 |
-
"grad_norm": 15.452752113342285,
|
| 762 |
-
"learning_rate": 7.338293014885212e-05,
|
| 763 |
-
"loss": 3.6698,
|
| 764 |
-
"step": 1010
|
| 765 |
-
},
|
| 766 |
-
{
|
| 767 |
-
"epoch": 0.7870370370370371,
|
| 768 |
-
"grad_norm": 34.56717300415039,
|
| 769 |
-
"learning_rate": 7.324096386052416e-05,
|
| 770 |
-
"loss": 3.8096,
|
| 771 |
-
"step": 1020
|
| 772 |
-
},
|
| 773 |
-
{
|
| 774 |
-
"epoch": 0.7947530864197531,
|
| 775 |
-
"grad_norm": 21.44382095336914,
|
| 776 |
-
"learning_rate": 7.309763116636786e-05,
|
| 777 |
-
"loss": 3.9437,
|
| 778 |
-
"step": 1030
|
| 779 |
-
},
|
| 780 |
-
{
|
| 781 |
-
"epoch": 0.8024691358024691,
|
| 782 |
-
"grad_norm": 19.292604446411133,
|
| 783 |
-
"learning_rate": 7.295293795822887e-05,
|
| 784 |
-
"loss": 3.9738,
|
| 785 |
-
"step": 1040
|
| 786 |
-
},
|
| 787 |
-
{
|
| 788 |
-
"epoch": 0.8101851851851852,
|
| 789 |
-
"grad_norm": 15.463872909545898,
|
| 790 |
-
"learning_rate": 7.280689018387824e-05,
|
| 791 |
-
"loss": 3.861,
|
| 792 |
-
"step": 1050
|
| 793 |
-
},
|
| 794 |
-
{
|
| 795 |
-
"epoch": 0.8101851851851852,
|
| 796 |
-
"eval_loss": 3.859124183654785,
|
| 797 |
-
"eval_runtime": 17.5741,
|
| 798 |
-
"eval_samples_per_second": 28.451,
|
| 799 |
-
"eval_steps_per_second": 28.451,
|
| 800 |
-
"step": 1050
|
| 801 |
-
},
|
| 802 |
-
{
|
| 803 |
-
"epoch": 0.8179012345679012,
|
| 804 |
-
"grad_norm": 20.17243003845215,
|
| 805 |
-
"learning_rate": 7.265949384676795e-05,
|
| 806 |
-
"loss": 3.7003,
|
| 807 |
-
"step": 1060
|
| 808 |
-
},
|
| 809 |
-
{
|
| 810 |
-
"epoch": 0.8256172839506173,
|
| 811 |
-
"grad_norm": 17.895933151245117,
|
| 812 |
-
"learning_rate": 7.251075500578411e-05,
|
| 813 |
-
"loss": 3.8089,
|
| 814 |
-
"step": 1070
|
| 815 |
-
},
|
| 816 |
-
{
|
| 817 |
-
"epoch": 0.8333333333333334,
|
| 818 |
-
"grad_norm": 16.239850997924805,
|
| 819 |
-
"learning_rate": 7.236067977499791e-05,
|
| 820 |
-
"loss": 3.8402,
|
| 821 |
-
"step": 1080
|
| 822 |
-
},
|
| 823 |
-
{
|
| 824 |
-
"epoch": 0.8410493827160493,
|
| 825 |
-
"grad_norm": 23.776884078979492,
|
| 826 |
-
"learning_rate": 7.220927432341426e-05,
|
| 827 |
-
"loss": 3.7538,
|
| 828 |
-
"step": 1090
|
| 829 |
-
},
|
| 830 |
-
{
|
| 831 |
-
"epoch": 0.8487654320987654,
|
| 832 |
-
"grad_norm": 15.736024856567383,
|
| 833 |
-
"learning_rate": 7.205654487471826e-05,
|
| 834 |
-
"loss": 3.8258,
|
| 835 |
-
"step": 1100
|
| 836 |
-
},
|
| 837 |
-
{
|
| 838 |
-
"epoch": 0.8564814814814815,
|
| 839 |
-
"grad_norm": 26.04443359375,
|
| 840 |
-
"learning_rate": 7.190249770701939e-05,
|
| 841 |
-
"loss": 4.0184,
|
| 842 |
-
"step": 1110
|
| 843 |
-
},
|
| 844 |
-
{
|
| 845 |
-
"epoch": 0.8641975308641975,
|
| 846 |
-
"grad_norm": 21.772159576416016,
|
| 847 |
-
"learning_rate": 7.174713915259331e-05,
|
| 848 |
-
"loss": 3.7938,
|
| 849 |
-
"step": 1120
|
| 850 |
-
},
|
| 851 |
-
{
|
| 852 |
-
"epoch": 0.8719135802469136,
|
| 853 |
-
"grad_norm": 19.530372619628906,
|
| 854 |
-
"learning_rate": 7.15904755976217e-05,
|
| 855 |
-
"loss": 3.6955,
|
| 856 |
-
"step": 1130
|
| 857 |
-
},
|
| 858 |
-
{
|
| 859 |
-
"epoch": 0.8796296296296297,
|
| 860 |
-
"grad_norm": 19.027843475341797,
|
| 861 |
-
"learning_rate": 7.143251348192971e-05,
|
| 862 |
-
"loss": 3.8163,
|
| 863 |
-
"step": 1140
|
| 864 |
-
},
|
| 865 |
-
{
|
| 866 |
-
"epoch": 0.8873456790123457,
|
| 867 |
-
"grad_norm": 24.277997970581055,
|
| 868 |
-
"learning_rate": 7.12732592987212e-05,
|
| 869 |
-
"loss": 3.7664,
|
| 870 |
-
"step": 1150
|
| 871 |
-
},
|
| 872 |
-
{
|
| 873 |
-
"epoch": 0.8950617283950617,
|
| 874 |
-
"grad_norm": 24.98284149169922,
|
| 875 |
-
"learning_rate": 7.111271959431189e-05,
|
| 876 |
-
"loss": 3.5373,
|
| 877 |
-
"step": 1160
|
| 878 |
-
},
|
| 879 |
-
{
|
| 880 |
-
"epoch": 0.9027777777777778,
|
| 881 |
-
"grad_norm": 27.04857063293457,
|
| 882 |
-
"learning_rate": 7.095090096786027e-05,
|
| 883 |
-
"loss": 3.6208,
|
| 884 |
-
"step": 1170
|
| 885 |
-
},
|
| 886 |
-
{
|
| 887 |
-
"epoch": 0.9104938271604939,
|
| 888 |
-
"grad_norm": 19.242406845092773,
|
| 889 |
-
"learning_rate": 7.078781007109625e-05,
|
| 890 |
-
"loss": 3.8045,
|
| 891 |
-
"step": 1180
|
| 892 |
-
},
|
| 893 |
-
{
|
| 894 |
-
"epoch": 0.9182098765432098,
|
| 895 |
-
"grad_norm": 52.25917053222656,
|
| 896 |
-
"learning_rate": 7.062345360804779e-05,
|
| 897 |
-
"loss": 3.8328,
|
| 898 |
-
"step": 1190
|
| 899 |
-
},
|
| 900 |
-
{
|
| 901 |
-
"epoch": 0.9259259259259259,
|
| 902 |
-
"grad_norm": 19.197126388549805,
|
| 903 |
-
"learning_rate": 7.045783833476538e-05,
|
| 904 |
-
"loss": 3.6963,
|
| 905 |
-
"step": 1200
|
| 906 |
-
},
|
| 907 |
-
{
|
| 908 |
-
"epoch": 0.9259259259259259,
|
| 909 |
-
"eval_loss": 3.787529945373535,
|
| 910 |
-
"eval_runtime": 18.9623,
|
| 911 |
-
"eval_samples_per_second": 26.368,
|
| 912 |
-
"eval_steps_per_second": 26.368,
|
| 913 |
-
"step": 1200
|
| 914 |
-
},
|
| 915 |
-
{
|
| 916 |
-
"epoch": 0.933641975308642,
|
| 917 |
-
"grad_norm": 25.777538299560547,
|
| 918 |
-
"learning_rate": 7.029097105904422e-05,
|
| 919 |
-
"loss": 3.8159,
|
| 920 |
-
"step": 1210
|
| 921 |
-
},
|
| 922 |
-
{
|
| 923 |
-
"epoch": 0.941358024691358,
|
| 924 |
-
"grad_norm": 21.648229598999023,
|
| 925 |
-
"learning_rate": 7.012285864014445e-05,
|
| 926 |
-
"loss": 3.6991,
|
| 927 |
-
"step": 1220
|
| 928 |
-
},
|
| 929 |
-
{
|
| 930 |
-
"epoch": 0.9490740740740741,
|
| 931 |
-
"grad_norm": 18.975547790527344,
|
| 932 |
-
"learning_rate": 6.995350798850913e-05,
|
| 933 |
-
"loss": 3.787,
|
| 934 |
-
"step": 1230
|
| 935 |
-
},
|
| 936 |
-
{
|
| 937 |
-
"epoch": 0.9567901234567902,
|
| 938 |
-
"grad_norm": 17.65053939819336,
|
| 939 |
-
"learning_rate": 6.978292606548029e-05,
|
| 940 |
-
"loss": 3.7765,
|
| 941 |
-
"step": 1240
|
| 942 |
-
},
|
| 943 |
-
{
|
| 944 |
-
"epoch": 0.9645061728395061,
|
| 945 |
-
"grad_norm": 17.642724990844727,
|
| 946 |
-
"learning_rate": 6.961111988301262e-05,
|
| 947 |
-
"loss": 3.6433,
|
| 948 |
-
"step": 1250
|
| 949 |
-
},
|
| 950 |
-
{
|
| 951 |
-
"epoch": 0.9722222222222222,
|
| 952 |
-
"grad_norm": 31.705036163330078,
|
| 953 |
-
"learning_rate": 6.943809650338541e-05,
|
| 954 |
-
"loss": 3.7943,
|
| 955 |
-
"step": 1260
|
| 956 |
-
},
|
| 957 |
-
{
|
| 958 |
-
"epoch": 0.9799382716049383,
|
| 959 |
-
"grad_norm": 15.860248565673828,
|
| 960 |
-
"learning_rate": 6.926386303891205e-05,
|
| 961 |
-
"loss": 3.7749,
|
| 962 |
-
"step": 1270
|
| 963 |
-
},
|
| 964 |
-
{
|
| 965 |
-
"epoch": 0.9876543209876543,
|
| 966 |
-
"grad_norm": 20.174976348876953,
|
| 967 |
-
"learning_rate": 6.908842665164789e-05,
|
| 968 |
-
"loss": 3.6612,
|
| 969 |
-
"step": 1280
|
| 970 |
-
},
|
| 971 |
-
{
|
| 972 |
-
"epoch": 0.9953703703703703,
|
| 973 |
-
"grad_norm": 24.97418785095215,
|
| 974 |
-
"learning_rate": 6.891179455309567e-05,
|
| 975 |
-
"loss": 3.7852,
|
| 976 |
-
"step": 1290
|
| 977 |
-
},
|
| 978 |
-
{
|
| 979 |
-
"epoch": 1.0030864197530864,
|
| 980 |
-
"grad_norm": 26.189205169677734,
|
| 981 |
-
"learning_rate": 6.873397400390911e-05,
|
| 982 |
-
"loss": 3.6042,
|
| 983 |
-
"step": 1300
|
| 984 |
-
},
|
| 985 |
-
{
|
| 986 |
-
"epoch": 1.0108024691358024,
|
| 987 |
-
"grad_norm": 15.729840278625488,
|
| 988 |
-
"learning_rate": 6.855497231359453e-05,
|
| 989 |
-
"loss": 3.5714,
|
| 990 |
-
"step": 1310
|
| 991 |
-
},
|
| 992 |
-
{
|
| 993 |
-
"epoch": 1.0185185185185186,
|
| 994 |
-
"grad_norm": 16.016050338745117,
|
| 995 |
-
"learning_rate": 6.837479684021032e-05,
|
| 996 |
-
"loss": 3.6184,
|
| 997 |
-
"step": 1320
|
| 998 |
-
},
|
| 999 |
-
{
|
| 1000 |
-
"epoch": 1.0262345679012346,
|
| 1001 |
-
"grad_norm": 21.100793838500977,
|
| 1002 |
-
"learning_rate": 6.819345499006448e-05,
|
| 1003 |
-
"loss": 3.4837,
|
| 1004 |
-
"step": 1330
|
| 1005 |
-
},
|
| 1006 |
-
{
|
| 1007 |
-
"epoch": 1.0339506172839505,
|
| 1008 |
-
"grad_norm": 34.890785217285156,
|
| 1009 |
-
"learning_rate": 6.80109542174102e-05,
|
| 1010 |
-
"loss": 3.5829,
|
| 1011 |
-
"step": 1340
|
| 1012 |
-
},
|
| 1013 |
-
{
|
| 1014 |
-
"epoch": 1.0416666666666667,
|
| 1015 |
-
"grad_norm": 22.613712310791016,
|
| 1016 |
-
"learning_rate": 6.782730202413946e-05,
|
| 1017 |
-
"loss": 3.4799,
|
| 1018 |
-
"step": 1350
|
| 1019 |
-
},
|
| 1020 |
-
{
|
| 1021 |
-
"epoch": 1.0416666666666667,
|
| 1022 |
-
"eval_loss": 3.742891550064087,
|
| 1023 |
-
"eval_runtime": 17.5683,
|
| 1024 |
-
"eval_samples_per_second": 28.46,
|
| 1025 |
-
"eval_steps_per_second": 28.46,
|
| 1026 |
-
"step": 1350
|
| 1027 |
-
},
|
| 1028 |
-
{
|
| 1029 |
-
"epoch": 1.0493827160493827,
|
| 1030 |
-
"grad_norm": 20.22081184387207,
|
| 1031 |
-
"learning_rate": 6.76425059594746e-05,
|
| 1032 |
-
"loss": 3.6744,
|
| 1033 |
-
"step": 1360
|
| 1034 |
-
},
|
| 1035 |
-
{
|
| 1036 |
-
"epoch": 1.0570987654320987,
|
| 1037 |
-
"grad_norm": 16.30259895324707,
|
| 1038 |
-
"learning_rate": 6.745657361965803e-05,
|
| 1039 |
-
"loss": 3.3617,
|
| 1040 |
-
"step": 1370
|
| 1041 |
-
},
|
| 1042 |
-
{
|
| 1043 |
-
"epoch": 1.0648148148148149,
|
| 1044 |
-
"grad_norm": 32.47397994995117,
|
| 1045 |
-
"learning_rate": 6.726951264763998e-05,
|
| 1046 |
-
"loss": 3.3692,
|
| 1047 |
-
"step": 1380
|
| 1048 |
-
},
|
| 1049 |
-
{
|
| 1050 |
-
"epoch": 1.0725308641975309,
|
| 1051 |
-
"grad_norm": 16.126924514770508,
|
| 1052 |
-
"learning_rate": 6.70813307327644e-05,
|
| 1053 |
-
"loss": 3.4547,
|
| 1054 |
-
"step": 1390
|
| 1055 |
-
},
|
| 1056 |
-
{
|
| 1057 |
-
"epoch": 1.0802469135802468,
|
| 1058 |
-
"grad_norm": 35.09951400756836,
|
| 1059 |
-
"learning_rate": 6.689203561045268e-05,
|
| 1060 |
-
"loss": 3.4555,
|
| 1061 |
-
"step": 1400
|
| 1062 |
-
},
|
| 1063 |
-
{
|
| 1064 |
-
"epoch": 1.087962962962963,
|
| 1065 |
-
"grad_norm": 27.35624122619629,
|
| 1066 |
-
"learning_rate": 6.670163506188593e-05,
|
| 1067 |
-
"loss": 3.639,
|
| 1068 |
-
"step": 1410
|
| 1069 |
-
},
|
| 1070 |
-
{
|
| 1071 |
-
"epoch": 1.095679012345679,
|
| 1072 |
-
"grad_norm": 20.523944854736328,
|
| 1073 |
-
"learning_rate": 6.651013691368492e-05,
|
| 1074 |
-
"loss": 3.6066,
|
| 1075 |
-
"step": 1420
|
| 1076 |
-
},
|
| 1077 |
-
{
|
| 1078 |
-
"epoch": 1.103395061728395,
|
| 1079 |
-
"grad_norm": 29.101484298706055,
|
| 1080 |
-
"learning_rate": 6.631754903758851e-05,
|
| 1081 |
-
"loss": 3.5652,
|
| 1082 |
-
"step": 1430
|
| 1083 |
-
},
|
| 1084 |
-
{
|
| 1085 |
-
"epoch": 1.1111111111111112,
|
| 1086 |
-
"grad_norm": 24.4724063873291,
|
| 1087 |
-
"learning_rate": 6.612387935012995e-05,
|
| 1088 |
-
"loss": 3.4425,
|
| 1089 |
-
"step": 1440
|
| 1090 |
-
},
|
| 1091 |
-
{
|
| 1092 |
-
"epoch": 1.1188271604938271,
|
| 1093 |
-
"grad_norm": 24.631128311157227,
|
| 1094 |
-
"learning_rate": 6.592913581231155e-05,
|
| 1095 |
-
"loss": 3.5492,
|
| 1096 |
-
"step": 1450
|
| 1097 |
-
},
|
| 1098 |
-
{
|
| 1099 |
-
"epoch": 1.126543209876543,
|
| 1100 |
-
"grad_norm": 39.123775482177734,
|
| 1101 |
-
"learning_rate": 6.573332642927737e-05,
|
| 1102 |
-
"loss": 3.3953,
|
| 1103 |
-
"step": 1460
|
| 1104 |
-
},
|
| 1105 |
-
{
|
| 1106 |
-
"epoch": 1.1342592592592593,
|
| 1107 |
-
"grad_norm": 30.524311065673828,
|
| 1108 |
-
"learning_rate": 6.553645924998422e-05,
|
| 1109 |
-
"loss": 3.3948,
|
| 1110 |
-
"step": 1470
|
| 1111 |
-
},
|
| 1112 |
-
{
|
| 1113 |
-
"epoch": 1.1419753086419753,
|
| 1114 |
-
"grad_norm": 23.212848663330078,
|
| 1115 |
-
"learning_rate": 6.53385423668708e-05,
|
| 1116 |
-
"loss": 3.3809,
|
| 1117 |
-
"step": 1480
|
| 1118 |
-
},
|
| 1119 |
-
{
|
| 1120 |
-
"epoch": 1.1496913580246915,
|
| 1121 |
-
"grad_norm": 32.31661605834961,
|
| 1122 |
-
"learning_rate": 6.513958391552494e-05,
|
| 1123 |
-
"loss": 3.4081,
|
| 1124 |
-
"step": 1490
|
| 1125 |
-
},
|
| 1126 |
-
{
|
| 1127 |
-
"epoch": 1.1574074074074074,
|
| 1128 |
-
"grad_norm": 27.419097900390625,
|
| 1129 |
-
"learning_rate": 6.493959207434934e-05,
|
| 1130 |
-
"loss": 3.3998,
|
| 1131 |
-
"step": 1500
|
| 1132 |
-
},
|
| 1133 |
-
{
|
| 1134 |
-
"epoch": 1.1574074074074074,
|
| 1135 |
-
"eval_loss": 3.70896053314209,
|
| 1136 |
-
"eval_runtime": 19.7466,
|
| 1137 |
-
"eval_samples_per_second": 25.321,
|
| 1138 |
-
"eval_steps_per_second": 25.321,
|
| 1139 |
-
"step": 1500
|
| 1140 |
-
},
|
| 1141 |
-
{
|
| 1142 |
-
"epoch": 1.1651234567901234,
|
| 1143 |
-
"grad_norm": 24.480224609375,
|
| 1144 |
-
"learning_rate": 6.473857506422526e-05,
|
| 1145 |
-
"loss": 3.3748,
|
| 1146 |
-
"step": 1510
|
| 1147 |
-
},
|
| 1148 |
-
{
|
| 1149 |
-
"epoch": 1.1728395061728394,
|
| 1150 |
-
"grad_norm": 22.06273078918457,
|
| 1151 |
-
"learning_rate": 6.453654114817467e-05,
|
| 1152 |
-
"loss": 3.7085,
|
| 1153 |
-
"step": 1520
|
| 1154 |
-
},
|
| 1155 |
-
{
|
| 1156 |
-
"epoch": 1.1805555555555556,
|
| 1157 |
-
"grad_norm": 21.491657257080078,
|
| 1158 |
-
"learning_rate": 6.433349863102053e-05,
|
| 1159 |
-
"loss": 3.398,
|
| 1160 |
-
"step": 1530
|
| 1161 |
-
},
|
| 1162 |
-
{
|
| 1163 |
-
"epoch": 1.1882716049382716,
|
| 1164 |
-
"grad_norm": 23.725849151611328,
|
| 1165 |
-
"learning_rate": 6.412945585904545e-05,
|
| 1166 |
-
"loss": 3.4217,
|
| 1167 |
-
"step": 1540
|
| 1168 |
-
},
|
| 1169 |
-
{
|
| 1170 |
-
"epoch": 1.1959876543209877,
|
| 1171 |
-
"grad_norm": 22.040245056152344,
|
| 1172 |
-
"learning_rate": 6.392442121964865e-05,
|
| 1173 |
-
"loss": 3.3477,
|
| 1174 |
-
"step": 1550
|
| 1175 |
-
},
|
| 1176 |
-
{
|
| 1177 |
-
"epoch": 1.2037037037037037,
|
| 1178 |
-
"grad_norm": 20.66607093811035,
|
| 1179 |
-
"learning_rate": 6.371840314100104e-05,
|
| 1180 |
-
"loss": 3.4077,
|
| 1181 |
-
"step": 1560
|
| 1182 |
-
},
|
| 1183 |
-
{
|
| 1184 |
-
"epoch": 1.2114197530864197,
|
| 1185 |
-
"grad_norm": 23.58853530883789,
|
| 1186 |
-
"learning_rate": 6.351141009169893e-05,
|
| 1187 |
-
"loss": 3.3661,
|
| 1188 |
-
"step": 1570
|
| 1189 |
-
},
|
| 1190 |
-
{
|
| 1191 |
-
"epoch": 1.2191358024691359,
|
| 1192 |
-
"grad_norm": 24.576080322265625,
|
| 1193 |
-
"learning_rate": 6.330345058041585e-05,
|
| 1194 |
-
"loss": 3.4321,
|
| 1195 |
-
"step": 1580
|
| 1196 |
-
},
|
| 1197 |
-
{
|
| 1198 |
-
"epoch": 1.2268518518518519,
|
| 1199 |
-
"grad_norm": 24.945005416870117,
|
| 1200 |
-
"learning_rate": 6.309453315555279e-05,
|
| 1201 |
-
"loss": 3.5472,
|
| 1202 |
-
"step": 1590
|
| 1203 |
-
},
|
| 1204 |
-
{
|
| 1205 |
-
"epoch": 1.2345679012345678,
|
| 1206 |
-
"grad_norm": 20.998676300048828,
|
| 1207 |
-
"learning_rate": 6.288466640488679e-05,
|
| 1208 |
-
"loss": 3.4229,
|
| 1209 |
-
"step": 1600
|
| 1210 |
-
},
|
| 1211 |
-
{
|
| 1212 |
-
"epoch": 1.242283950617284,
|
| 1213 |
-
"grad_norm": 25.277576446533203,
|
| 1214 |
-
"learning_rate": 6.2673858955218e-05,
|
| 1215 |
-
"loss": 3.5484,
|
| 1216 |
-
"step": 1610
|
| 1217 |
-
},
|
| 1218 |
-
{
|
| 1219 |
-
"epoch": 1.25,
|
| 1220 |
-
"grad_norm": 35.780555725097656,
|
| 1221 |
-
"learning_rate": 6.2462119472015e-05,
|
| 1222 |
-
"loss": 3.572,
|
| 1223 |
-
"step": 1620
|
| 1224 |
-
},
|
| 1225 |
-
{
|
| 1226 |
-
"epoch": 1.257716049382716,
|
| 1227 |
-
"grad_norm": 24.384044647216797,
|
| 1228 |
-
"learning_rate": 6.22494566590586e-05,
|
| 1229 |
-
"loss": 3.445,
|
| 1230 |
-
"step": 1630
|
| 1231 |
-
},
|
| 1232 |
-
{
|
| 1233 |
-
"epoch": 1.2654320987654322,
|
| 1234 |
-
"grad_norm": 16.81711196899414,
|
| 1235 |
-
"learning_rate": 6.20358792580841e-05,
|
| 1236 |
-
"loss": 3.3449,
|
| 1237 |
-
"step": 1640
|
| 1238 |
-
},
|
| 1239 |
-
{
|
| 1240 |
-
"epoch": 1.2731481481481481,
|
| 1241 |
-
"grad_norm": 19.94586181640625,
|
| 1242 |
-
"learning_rate": 6.182139604842195e-05,
|
| 1243 |
-
"loss": 3.4073,
|
| 1244 |
-
"step": 1650
|
| 1245 |
-
},
|
| 1246 |
-
{
|
| 1247 |
-
"epoch": 1.2731481481481481,
|
| 1248 |
-
"eval_loss": 3.637455463409424,
|
| 1249 |
-
"eval_runtime": 17.8365,
|
| 1250 |
-
"eval_samples_per_second": 28.032,
|
| 1251 |
-
"eval_steps_per_second": 28.032,
|
| 1252 |
-
"step": 1650
|
| 1253 |
-
},
|
| 1254 |
-
{
|
| 1255 |
-
"epoch": 1.2808641975308643,
|
| 1256 |
-
"grad_norm": 28.49296760559082,
|
| 1257 |
-
"learning_rate": 6.160601584663681e-05,
|
| 1258 |
-
"loss": 3.4991,
|
| 1259 |
-
"step": 1660
|
| 1260 |
-
},
|
| 1261 |
-
{
|
| 1262 |
-
"epoch": 1.2885802469135803,
|
| 1263 |
-
"grad_norm": 30.67734146118164,
|
| 1264 |
-
"learning_rate": 6.138974750616519e-05,
|
| 1265 |
-
"loss": 3.3618,
|
| 1266 |
-
"step": 1670
|
| 1267 |
-
},
|
| 1268 |
-
{
|
| 1269 |
-
"epoch": 1.2962962962962963,
|
| 1270 |
-
"grad_norm": 19.116107940673828,
|
| 1271 |
-
"learning_rate": 6.11725999169515e-05,
|
| 1272 |
-
"loss": 3.4404,
|
| 1273 |
-
"step": 1680
|
| 1274 |
-
},
|
| 1275 |
-
{
|
| 1276 |
-
"epoch": 1.3040123456790123,
|
| 1277 |
-
"grad_norm": 28.58710289001465,
|
| 1278 |
-
"learning_rate": 6.0954582005082616e-05,
|
| 1279 |
-
"loss": 3.3612,
|
| 1280 |
-
"step": 1690
|
| 1281 |
-
},
|
| 1282 |
-
{
|
| 1283 |
-
"epoch": 1.3117283950617284,
|
| 1284 |
-
"grad_norm": 37.30510711669922,
|
| 1285 |
-
"learning_rate": 6.0735702732421015e-05,
|
| 1286 |
-
"loss": 3.3507,
|
| 1287 |
-
"step": 1700
|
| 1288 |
-
},
|
| 1289 |
-
{
|
| 1290 |
-
"epoch": 1.3194444444444444,
|
| 1291 |
-
"grad_norm": 25.61612319946289,
|
| 1292 |
-
"learning_rate": 6.0515971096236253e-05,
|
| 1293 |
-
"loss": 3.2735,
|
| 1294 |
-
"step": 1710
|
| 1295 |
-
},
|
| 1296 |
-
{
|
| 1297 |
-
"epoch": 1.3271604938271606,
|
| 1298 |
-
"grad_norm": 33.34403991699219,
|
| 1299 |
-
"learning_rate": 6.029539612883529e-05,
|
| 1300 |
-
"loss": 3.2973,
|
| 1301 |
-
"step": 1720
|
| 1302 |
-
},
|
| 1303 |
-
{
|
| 1304 |
-
"epoch": 1.3348765432098766,
|
| 1305 |
-
"grad_norm": 31.370502471923828,
|
| 1306 |
-
"learning_rate": 6.007398689719111e-05,
|
| 1307 |
-
"loss": 3.327,
|
| 1308 |
-
"step": 1730
|
| 1309 |
-
},
|
| 1310 |
-
{
|
| 1311 |
-
"epoch": 1.3425925925925926,
|
| 1312 |
-
"grad_norm": 31.45387840270996,
|
| 1313 |
-
"learning_rate": 5.9851752502570015e-05,
|
| 1314 |
-
"loss": 3.2595,
|
| 1315 |
-
"step": 1740
|
| 1316 |
-
},
|
| 1317 |
-
{
|
| 1318 |
-
"epoch": 1.3503086419753085,
|
| 1319 |
-
"grad_norm": 28.973865509033203,
|
| 1320 |
-
"learning_rate": 5.9628702080157526e-05,
|
| 1321 |
-
"loss": 3.3104,
|
| 1322 |
-
"step": 1750
|
| 1323 |
-
},
|
| 1324 |
-
{
|
| 1325 |
-
"epoch": 1.3580246913580247,
|
| 1326 |
-
"grad_norm": 21.37240219116211,
|
| 1327 |
-
"learning_rate": 5.940484479868288e-05,
|
| 1328 |
-
"loss": 3.4102,
|
| 1329 |
-
"step": 1760
|
| 1330 |
-
},
|
| 1331 |
-
{
|
| 1332 |
-
"epoch": 1.3657407407407407,
|
| 1333 |
-
"grad_norm": 33.90608215332031,
|
| 1334 |
-
"learning_rate": 5.918018986004216e-05,
|
| 1335 |
-
"loss": 3.5516,
|
| 1336 |
-
"step": 1770
|
| 1337 |
-
},
|
| 1338 |
-
{
|
| 1339 |
-
"epoch": 1.373456790123457,
|
| 1340 |
-
"grad_norm": 24.596567153930664,
|
| 1341 |
-
"learning_rate": 5.895474649891995e-05,
|
| 1342 |
-
"loss": 3.378,
|
| 1343 |
-
"step": 1780
|
| 1344 |
-
},
|
| 1345 |
-
{
|
| 1346 |
-
"epoch": 1.3811728395061729,
|
| 1347 |
-
"grad_norm": 25.496883392333984,
|
| 1348 |
-
"learning_rate": 5.872852398240984e-05,
|
| 1349 |
-
"loss": 3.3586,
|
| 1350 |
-
"step": 1790
|
| 1351 |
-
},
|
| 1352 |
-
{
|
| 1353 |
-
"epoch": 1.3888888888888888,
|
| 1354 |
-
"grad_norm": 27.56167221069336,
|
| 1355 |
-
"learning_rate": 5.8501531609633424e-05,
|
| 1356 |
-
"loss": 3.3986,
|
| 1357 |
-
"step": 1800
|
| 1358 |
-
},
|
| 1359 |
-
{
|
| 1360 |
-
"epoch": 1.3888888888888888,
|
| 1361 |
-
"eval_loss": 3.574371576309204,
|
| 1362 |
-
"eval_runtime": 19.0779,
|
| 1363 |
-
"eval_samples_per_second": 26.208,
|
| 1364 |
-
"eval_steps_per_second": 26.208,
|
| 1365 |
-
"step": 1800
|
| 1366 |
-
},
|
| 1367 |
-
{
|
| 1368 |
-
"epoch": 1.3966049382716048,
|
| 1369 |
-
"grad_norm": 26.979446411132812,
|
| 1370 |
-
"learning_rate": 5.827377871135807e-05,
|
| 1371 |
-
"loss": 3.223,
|
| 1372 |
-
"step": 1810
|
| 1373 |
-
},
|
| 1374 |
-
{
|
| 1375 |
-
"epoch": 1.404320987654321,
|
| 1376 |
-
"grad_norm": 33.906124114990234,
|
| 1377 |
-
"learning_rate": 5.8045274649613386e-05,
|
| 1378 |
-
"loss": 3.2157,
|
| 1379 |
-
"step": 1820
|
| 1380 |
-
},
|
| 1381 |
-
{
|
| 1382 |
-
"epoch": 1.412037037037037,
|
| 1383 |
-
"grad_norm": 30.635848999023438,
|
| 1384 |
-
"learning_rate": 5.781602881730637e-05,
|
| 1385 |
-
"loss": 3.3547,
|
| 1386 |
-
"step": 1830
|
| 1387 |
-
},
|
| 1388 |
-
{
|
| 1389 |
-
"epoch": 1.4197530864197532,
|
| 1390 |
-
"grad_norm": 18.647005081176758,
|
| 1391 |
-
"learning_rate": 5.7586050637835295e-05,
|
| 1392 |
-
"loss": 3.3164,
|
| 1393 |
-
"step": 1840
|
| 1394 |
-
},
|
| 1395 |
-
{
|
| 1396 |
-
"epoch": 1.4274691358024691,
|
| 1397 |
-
"grad_norm": 28.55082130432129,
|
| 1398 |
-
"learning_rate": 5.735534956470233e-05,
|
| 1399 |
-
"loss": 3.392,
|
| 1400 |
-
"step": 1850
|
| 1401 |
-
},
|
| 1402 |
-
{
|
| 1403 |
-
"epoch": 1.4351851851851851,
|
| 1404 |
-
"grad_norm": 24.287002563476562,
|
| 1405 |
-
"learning_rate": 5.7123935081125034e-05,
|
| 1406 |
-
"loss": 3.4614,
|
| 1407 |
-
"step": 1860
|
| 1408 |
-
},
|
| 1409 |
-
{
|
| 1410 |
-
"epoch": 1.4429012345679013,
|
| 1411 |
-
"grad_norm": 23.30496597290039,
|
| 1412 |
-
"learning_rate": 5.68918166996464e-05,
|
| 1413 |
-
"loss": 3.363,
|
| 1414 |
-
"step": 1870
|
| 1415 |
-
},
|
| 1416 |
-
{
|
| 1417 |
-
"epoch": 1.4506172839506173,
|
| 1418 |
-
"grad_norm": 25.783321380615234,
|
| 1419 |
-
"learning_rate": 5.6659003961743965e-05,
|
| 1420 |
-
"loss": 3.4322,
|
| 1421 |
-
"step": 1880
|
| 1422 |
-
},
|
| 1423 |
-
{
|
| 1424 |
-
"epoch": 1.4583333333333333,
|
| 1425 |
-
"grad_norm": 22.144737243652344,
|
| 1426 |
-
"learning_rate": 5.642550643743753e-05,
|
| 1427 |
-
"loss": 3.3036,
|
| 1428 |
-
"step": 1890
|
| 1429 |
-
},
|
| 1430 |
-
{
|
| 1431 |
-
"epoch": 1.4660493827160495,
|
| 1432 |
-
"grad_norm": 29.154598236083984,
|
| 1433 |
-
"learning_rate": 5.619133372489575e-05,
|
| 1434 |
-
"loss": 3.3028,
|
| 1435 |
-
"step": 1900
|
| 1436 |
-
},
|
| 1437 |
-
{
|
| 1438 |
-
"epoch": 1.4737654320987654,
|
| 1439 |
-
"grad_norm": 26.780481338500977,
|
| 1440 |
-
"learning_rate": 5.5956495450041675e-05,
|
| 1441 |
-
"loss": 3.4062,
|
| 1442 |
-
"step": 1910
|
| 1443 |
-
},
|
| 1444 |
-
{
|
| 1445 |
-
"epoch": 1.4814814814814814,
|
| 1446 |
-
"grad_norm": 25.359783172607422,
|
| 1447 |
-
"learning_rate": 5.572100126615695e-05,
|
| 1448 |
-
"loss": 3.3956,
|
| 1449 |
-
"step": 1920
|
| 1450 |
-
},
|
| 1451 |
-
{
|
| 1452 |
-
"epoch": 1.4891975308641976,
|
| 1453 |
-
"grad_norm": 36.21287536621094,
|
| 1454 |
-
"learning_rate": 5.5484860853485135e-05,
|
| 1455 |
-
"loss": 3.4237,
|
| 1456 |
-
"step": 1930
|
| 1457 |
-
},
|
| 1458 |
-
{
|
| 1459 |
-
"epoch": 1.4969135802469136,
|
| 1460 |
-
"grad_norm": 30.381282806396484,
|
| 1461 |
-
"learning_rate": 5.524808391883367e-05,
|
| 1462 |
-
"loss": 3.316,
|
| 1463 |
-
"step": 1940
|
| 1464 |
-
},
|
| 1465 |
-
{
|
| 1466 |
-
"epoch": 1.5046296296296298,
|
| 1467 |
-
"grad_norm": 25.306821823120117,
|
| 1468 |
-
"learning_rate": 5.5010680195174975e-05,
|
| 1469 |
-
"loss": 3.3108,
|
| 1470 |
-
"step": 1950
|
| 1471 |
-
},
|
| 1472 |
-
{
|
| 1473 |
-
"epoch": 1.5046296296296298,
|
| 1474 |
-
"eval_loss": 3.5245025157928467,
|
| 1475 |
-
"eval_runtime": 17.6142,
|
| 1476 |
-
"eval_samples_per_second": 28.386,
|
| 1477 |
-
"eval_steps_per_second": 28.386,
|
| 1478 |
-
"step": 1950
|
| 1479 |
-
},
|
| 1480 |
-
{
|
| 1481 |
-
"epoch": 1.5123456790123457,
|
| 1482 |
-
"grad_norm": 19.196184158325195,
|
| 1483 |
-
"learning_rate": 5.477265944124626e-05,
|
| 1484 |
-
"loss": 3.3328,
|
| 1485 |
-
"step": 1960
|
| 1486 |
-
},
|
| 1487 |
-
{
|
| 1488 |
-
"epoch": 1.5200617283950617,
|
| 1489 |
-
"grad_norm": 23.37359619140625,
|
| 1490 |
-
"learning_rate": 5.453403144114846e-05,
|
| 1491 |
-
"loss": 3.2503,
|
| 1492 |
-
"step": 1970
|
| 1493 |
-
},
|
| 1494 |
-
{
|
| 1495 |
-
"epoch": 1.5277777777777777,
|
| 1496 |
-
"grad_norm": 22.228778839111328,
|
| 1497 |
-
"learning_rate": 5.429480600394405e-05,
|
| 1498 |
-
"loss": 3.3659,
|
| 1499 |
-
"step": 1980
|
| 1500 |
-
},
|
| 1501 |
-
{
|
| 1502 |
-
"epoch": 1.5354938271604939,
|
| 1503 |
-
"grad_norm": 33.1472282409668,
|
| 1504 |
-
"learning_rate": 5.4054992963253716e-05,
|
| 1505 |
-
"loss": 3.206,
|
| 1506 |
-
"step": 1990
|
| 1507 |
-
},
|
| 1508 |
-
{
|
| 1509 |
-
"epoch": 1.5432098765432098,
|
| 1510 |
-
"grad_norm": 20.814178466796875,
|
| 1511 |
-
"learning_rate": 5.381460217685231e-05,
|
| 1512 |
-
"loss": 3.3325,
|
| 1513 |
-
"step": 2000
|
| 1514 |
-
},
|
| 1515 |
-
{
|
| 1516 |
-
"epoch": 1.550925925925926,
|
| 1517 |
-
"grad_norm": 25.20581817626953,
|
| 1518 |
-
"learning_rate": 5.357364352626351e-05,
|
| 1519 |
-
"loss": 3.4038,
|
| 1520 |
-
"step": 2010
|
| 1521 |
-
},
|
| 1522 |
-
{
|
| 1523 |
-
"epoch": 1.558641975308642,
|
| 1524 |
-
"grad_norm": 23.560279846191406,
|
| 1525 |
-
"learning_rate": 5.333212691635368e-05,
|
| 1526 |
-
"loss": 3.2845,
|
| 1527 |
-
"step": 2020
|
| 1528 |
-
},
|
| 1529 |
-
{
|
| 1530 |
-
"epoch": 1.566358024691358,
|
| 1531 |
-
"grad_norm": 34.251522064208984,
|
| 1532 |
-
"learning_rate": 5.309006227492468e-05,
|
| 1533 |
-
"loss": 3.2339,
|
| 1534 |
-
"step": 2030
|
| 1535 |
-
},
|
| 1536 |
-
{
|
| 1537 |
-
"epoch": 1.574074074074074,
|
| 1538 |
-
"grad_norm": 28.38330078125,
|
| 1539 |
-
"learning_rate": 5.2847459552305834e-05,
|
| 1540 |
-
"loss": 3.2448,
|
| 1541 |
-
"step": 2040
|
| 1542 |
-
},
|
| 1543 |
-
{
|
| 1544 |
-
"epoch": 1.5817901234567902,
|
| 1545 |
-
"grad_norm": 26.189992904663086,
|
| 1546 |
-
"learning_rate": 5.260432872094484e-05,
|
| 1547 |
-
"loss": 3.3235,
|
| 1548 |
-
"step": 2050
|
| 1549 |
-
},
|
| 1550 |
-
{
|
| 1551 |
-
"epoch": 1.5895061728395061,
|
| 1552 |
-
"grad_norm": 18.80211639404297,
|
| 1553 |
-
"learning_rate": 5.23606797749979e-05,
|
| 1554 |
-
"loss": 3.2259,
|
| 1555 |
-
"step": 2060
|
| 1556 |
-
},
|
| 1557 |
-
{
|
| 1558 |
-
"epoch": 1.5972222222222223,
|
| 1559 |
-
"grad_norm": 30.651729583740234,
|
| 1560 |
-
"learning_rate": 5.211652272991889e-05,
|
| 1561 |
-
"loss": 3.1168,
|
| 1562 |
-
"step": 2070
|
| 1563 |
-
},
|
| 1564 |
-
{
|
| 1565 |
-
"epoch": 1.6049382716049383,
|
| 1566 |
-
"grad_norm": 34.49848175048828,
|
| 1567 |
-
"learning_rate": 5.1871867622047624e-05,
|
| 1568 |
-
"loss": 3.1488,
|
| 1569 |
-
"step": 2080
|
| 1570 |
-
},
|
| 1571 |
-
{
|
| 1572 |
-
"epoch": 1.6126543209876543,
|
| 1573 |
-
"grad_norm": 26.484848022460938,
|
| 1574 |
-
"learning_rate": 5.1626724508197356e-05,
|
| 1575 |
-
"loss": 3.1819,
|
| 1576 |
-
"step": 2090
|
| 1577 |
-
},
|
| 1578 |
-
{
|
| 1579 |
-
"epoch": 1.6203703703703702,
|
| 1580 |
-
"grad_norm": 30.39811897277832,
|
| 1581 |
-
"learning_rate": 5.13811034652413e-05,
|
| 1582 |
-
"loss": 3.4571,
|
| 1583 |
-
"step": 2100
|
| 1584 |
-
},
|
| 1585 |
-
{
|
| 1586 |
-
"epoch": 1.6203703703703702,
|
| 1587 |
-
"eval_loss": 3.4539551734924316,
|
| 1588 |
-
"eval_runtime": 18.5177,
|
| 1589 |
-
"eval_samples_per_second": 27.001,
|
| 1590 |
-
"eval_steps_per_second": 27.001,
|
| 1591 |
-
"step": 2100
|
| 1592 |
-
},
|
| 1593 |
-
{
|
| 1594 |
-
"epoch": 1.6280864197530864,
|
| 1595 |
-
"grad_norm": 27.10554313659668,
|
| 1596 |
-
"learning_rate": 5.113501458969854e-05,
|
| 1597 |
-
"loss": 3.3854,
|
| 1598 |
-
"step": 2110
|
| 1599 |
-
},
|
| 1600 |
-
{
|
| 1601 |
-
"epoch": 1.6358024691358026,
|
| 1602 |
-
"grad_norm": 20.923534393310547,
|
| 1603 |
-
"learning_rate": 5.088846799731885e-05,
|
| 1604 |
-
"loss": 3.3414,
|
| 1605 |
-
"step": 2120
|
| 1606 |
-
},
|
| 1607 |
-
{
|
| 1608 |
-
"epoch": 1.6435185185185186,
|
| 1609 |
-
"grad_norm": 29.248905181884766,
|
| 1610 |
-
"learning_rate": 5.064147382266701e-05,
|
| 1611 |
-
"loss": 3.2729,
|
| 1612 |
-
"step": 2130
|
| 1613 |
-
},
|
| 1614 |
-
{
|
| 1615 |
-
"epoch": 1.6512345679012346,
|
| 1616 |
-
"grad_norm": 25.068252563476562,
|
| 1617 |
-
"learning_rate": 5.039404221870612e-05,
|
| 1618 |
-
"loss": 3.4848,
|
| 1619 |
-
"step": 2140
|
| 1620 |
-
},
|
| 1621 |
-
{
|
| 1622 |
-
"epoch": 1.6589506172839505,
|
| 1623 |
-
"grad_norm": 33.346012115478516,
|
| 1624 |
-
"learning_rate": 5.0146183356380295e-05,
|
| 1625 |
-
"loss": 3.1731,
|
| 1626 |
-
"step": 2150
|
| 1627 |
-
},
|
| 1628 |
-
{
|
| 1629 |
-
"epoch": 1.6666666666666665,
|
| 1630 |
-
"grad_norm": 27.35399055480957,
|
| 1631 |
-
"learning_rate": 4.989790742419658e-05,
|
| 1632 |
-
"loss": 3.2532,
|
| 1633 |
-
"step": 2160
|
| 1634 |
-
},
|
| 1635 |
-
{
|
| 1636 |
-
"epoch": 1.6743827160493827,
|
| 1637 |
-
"grad_norm": 19.2067928314209,
|
| 1638 |
-
"learning_rate": 4.96492246278061e-05,
|
| 1639 |
-
"loss": 3.3465,
|
| 1640 |
-
"step": 2170
|
| 1641 |
-
},
|
| 1642 |
-
{
|
| 1643 |
-
"epoch": 1.682098765432099,
|
| 1644 |
-
"grad_norm": 31.099220275878906,
|
| 1645 |
-
"learning_rate": 4.940014518958461e-05,
|
| 1646 |
-
"loss": 3.311,
|
| 1647 |
-
"step": 2180
|
| 1648 |
-
},
|
| 1649 |
-
{
|
| 1650 |
-
"epoch": 1.6898148148148149,
|
| 1651 |
-
"grad_norm": 20.159053802490234,
|
| 1652 |
-
"learning_rate": 4.915067934821222e-05,
|
| 1653 |
-
"loss": 3.2779,
|
| 1654 |
-
"step": 2190
|
| 1655 |
-
},
|
| 1656 |
-
{
|
| 1657 |
-
"epoch": 1.6975308641975309,
|
| 1658 |
-
"grad_norm": 27.030029296875,
|
| 1659 |
-
"learning_rate": 4.890083735825258e-05,
|
| 1660 |
-
"loss": 2.936,
|
| 1661 |
-
"step": 2200
|
| 1662 |
-
},
|
| 1663 |
-
{
|
| 1664 |
-
"epoch": 1.7052469135802468,
|
| 1665 |
-
"grad_norm": 32.07747268676758,
|
| 1666 |
-
"learning_rate": 4.865062948973134e-05,
|
| 1667 |
-
"loss": 3.2353,
|
| 1668 |
-
"step": 2210
|
| 1669 |
-
},
|
| 1670 |
-
{
|
| 1671 |
-
"epoch": 1.7129629629629628,
|
| 1672 |
-
"grad_norm": 23.422794342041016,
|
| 1673 |
-
"learning_rate": 4.8400066027713974e-05,
|
| 1674 |
-
"loss": 3.1703,
|
| 1675 |
-
"step": 2220
|
| 1676 |
-
},
|
| 1677 |
-
{
|
| 1678 |
-
"epoch": 1.720679012345679,
|
| 1679 |
-
"grad_norm": 26.94754981994629,
|
| 1680 |
-
"learning_rate": 4.8149157271883026e-05,
|
| 1681 |
-
"loss": 3.2279,
|
| 1682 |
-
"step": 2230
|
| 1683 |
-
},
|
| 1684 |
-
{
|
| 1685 |
-
"epoch": 1.7283950617283952,
|
| 1686 |
-
"grad_norm": 26.43851089477539,
|
| 1687 |
-
"learning_rate": 4.789791353611469e-05,
|
| 1688 |
-
"loss": 3.1315,
|
| 1689 |
-
"step": 2240
|
| 1690 |
-
},
|
| 1691 |
-
{
|
| 1692 |
-
"epoch": 1.7361111111111112,
|
| 1693 |
-
"grad_norm": 34.339378356933594,
|
| 1694 |
-
"learning_rate": 4.76463451480549e-05,
|
| 1695 |
-
"loss": 3.2606,
|
| 1696 |
-
"step": 2250
|
| 1697 |
-
},
|
| 1698 |
-
{
|
| 1699 |
-
"epoch": 1.7361111111111112,
|
| 1700 |
-
"eval_loss": 3.3989486694335938,
|
| 1701 |
-
"eval_runtime": 17.7398,
|
| 1702 |
-
"eval_samples_per_second": 28.185,
|
| 1703 |
-
"eval_steps_per_second": 28.185,
|
| 1704 |
-
"step": 2250
|
| 1705 |
-
},
|
| 1706 |
-
{
|
| 1707 |
-
"epoch": 1.7438271604938271,
|
| 1708 |
-
"grad_norm": 28.98934555053711,
|
| 1709 |
-
"learning_rate": 4.7394462448694756e-05,
|
| 1710 |
-
"loss": 3.33,
|
| 1711 |
-
"step": 2260
|
| 1712 |
-
},
|
| 1713 |
-
{
|
| 1714 |
-
"epoch": 1.751543209876543,
|
| 1715 |
-
"grad_norm": 29.47772216796875,
|
| 1716 |
-
"learning_rate": 4.714227579194547e-05,
|
| 1717 |
-
"loss": 3.3261,
|
| 1718 |
-
"step": 2270
|
| 1719 |
-
},
|
| 1720 |
-
{
|
| 1721 |
-
"epoch": 1.7592592592592593,
|
| 1722 |
-
"grad_norm": 30.95444107055664,
|
| 1723 |
-
"learning_rate": 4.688979554421276e-05,
|
| 1724 |
-
"loss": 3.3706,
|
| 1725 |
-
"step": 2280
|
| 1726 |
-
},
|
| 1727 |
-
{
|
| 1728 |
-
"epoch": 1.7669753086419753,
|
| 1729 |
-
"grad_norm": 30.03018569946289,
|
| 1730 |
-
"learning_rate": 4.663703208397072e-05,
|
| 1731 |
-
"loss": 3.0371,
|
| 1732 |
-
"step": 2290
|
| 1733 |
-
},
|
| 1734 |
-
{
|
| 1735 |
-
"epoch": 1.7746913580246915,
|
| 1736 |
-
"grad_norm": 26.282629013061523,
|
| 1737 |
-
"learning_rate": 4.6383995801335176e-05,
|
| 1738 |
-
"loss": 3.3524,
|
| 1739 |
-
"step": 2300
|
| 1740 |
-
},
|
| 1741 |
-
{
|
| 1742 |
-
"epoch": 1.7824074074074074,
|
| 1743 |
-
"grad_norm": 55.37343978881836,
|
| 1744 |
-
"learning_rate": 4.6130697097636634e-05,
|
| 1745 |
-
"loss": 3.4364,
|
| 1746 |
-
"step": 2310
|
| 1747 |
-
},
|
| 1748 |
-
{
|
| 1749 |
-
"epoch": 1.7901234567901234,
|
| 1750 |
-
"grad_norm": 32.71061325073242,
|
| 1751 |
-
"learning_rate": 4.5877146384992725e-05,
|
| 1752 |
-
"loss": 3.3357,
|
| 1753 |
-
"step": 2320
|
| 1754 |
-
},
|
| 1755 |
-
{
|
| 1756 |
-
"epoch": 1.7978395061728394,
|
| 1757 |
-
"grad_norm": 27.23622703552246,
|
| 1758 |
-
"learning_rate": 4.562335408588012e-05,
|
| 1759 |
-
"loss": 3.1002,
|
| 1760 |
-
"step": 2330
|
| 1761 |
-
},
|
| 1762 |
-
{
|
| 1763 |
-
"epoch": 1.8055555555555556,
|
| 1764 |
-
"grad_norm": 38.7041015625,
|
| 1765 |
-
"learning_rate": 4.5369330632706223e-05,
|
| 1766 |
-
"loss": 3.1582,
|
| 1767 |
-
"step": 2340
|
| 1768 |
-
},
|
| 1769 |
-
{
|
| 1770 |
-
"epoch": 1.8132716049382716,
|
| 1771 |
-
"grad_norm": 35.14750671386719,
|
| 1772 |
-
"learning_rate": 4.5115086467380244e-05,
|
| 1773 |
-
"loss": 3.1838,
|
| 1774 |
-
"step": 2350
|
| 1775 |
-
},
|
| 1776 |
-
{
|
| 1777 |
-
"epoch": 1.8209876543209877,
|
| 1778 |
-
"grad_norm": 22.170042037963867,
|
| 1779 |
-
"learning_rate": 4.486063204088402e-05,
|
| 1780 |
-
"loss": 3.2643,
|
| 1781 |
-
"step": 2360
|
| 1782 |
-
},
|
| 1783 |
-
{
|
| 1784 |
-
"epoch": 1.8287037037037037,
|
| 1785 |
-
"grad_norm": 26.56646156311035,
|
| 1786 |
-
"learning_rate": 4.4605977812842384e-05,
|
| 1787 |
-
"loss": 3.105,
|
| 1788 |
-
"step": 2370
|
| 1789 |
-
},
|
| 1790 |
-
{
|
| 1791 |
-
"epoch": 1.8364197530864197,
|
| 1792 |
-
"grad_norm": 24.756046295166016,
|
| 1793 |
-
"learning_rate": 4.435113425109324e-05,
|
| 1794 |
-
"loss": 3.1878,
|
| 1795 |
-
"step": 2380
|
| 1796 |
-
},
|
| 1797 |
-
{
|
| 1798 |
-
"epoch": 1.8441358024691357,
|
| 1799 |
-
"grad_norm": 34.98149108886719,
|
| 1800 |
-
"learning_rate": 4.409611183125725e-05,
|
| 1801 |
-
"loss": 3.2356,
|
| 1802 |
-
"step": 2390
|
| 1803 |
-
},
|
| 1804 |
-
{
|
| 1805 |
-
"epoch": 1.8518518518518519,
|
| 1806 |
-
"grad_norm": 32.97663497924805,
|
| 1807 |
-
"learning_rate": 4.3840921036307274e-05,
|
| 1808 |
-
"loss": 3.1842,
|
| 1809 |
-
"step": 2400
|
| 1810 |
-
},
|
| 1811 |
-
{
|
| 1812 |
-
"epoch": 1.8518518518518519,
|
| 1813 |
-
"eval_loss": 3.3580269813537598,
|
| 1814 |
-
"eval_runtime": 19.1402,
|
| 1815 |
-
"eval_samples_per_second": 26.123,
|
| 1816 |
-
"eval_steps_per_second": 26.123,
|
| 1817 |
-
"step": 2400
|
| 1818 |
-
},
|
| 1819 |
-
{
|
| 1820 |
-
"epoch": 1.859567901234568,
|
| 1821 |
-
"grad_norm": 32.32783889770508,
|
| 1822 |
-
"learning_rate": 4.358557235613734e-05,
|
| 1823 |
-
"loss": 3.0993,
|
| 1824 |
-
"step": 2410
|
| 1825 |
-
},
|
| 1826 |
-
{
|
| 1827 |
-
"epoch": 1.867283950617284,
|
| 1828 |
-
"grad_norm": 36.21072006225586,
|
| 1829 |
-
"learning_rate": 4.333007628713158e-05,
|
| 1830 |
-
"loss": 3.2311,
|
| 1831 |
-
"step": 2420
|
| 1832 |
-
},
|
| 1833 |
-
{
|
| 1834 |
-
"epoch": 1.875,
|
| 1835 |
-
"grad_norm": 20.204273223876953,
|
| 1836 |
-
"learning_rate": 4.3074443331732674e-05,
|
| 1837 |
-
"loss": 3.0291,
|
| 1838 |
-
"step": 2430
|
| 1839 |
-
},
|
| 1840 |
-
{
|
| 1841 |
-
"epoch": 1.882716049382716,
|
| 1842 |
-
"grad_norm": 30.630725860595703,
|
| 1843 |
-
"learning_rate": 4.281868399801016e-05,
|
| 1844 |
-
"loss": 3.3228,
|
| 1845 |
-
"step": 2440
|
| 1846 |
-
},
|
| 1847 |
-
{
|
| 1848 |
-
"epoch": 1.890432098765432,
|
| 1849 |
-
"grad_norm": 23.674013137817383,
|
| 1850 |
-
"learning_rate": 4.256280879922852e-05,
|
| 1851 |
-
"loss": 3.1248,
|
| 1852 |
-
"step": 2450
|
| 1853 |
-
},
|
| 1854 |
-
{
|
| 1855 |
-
"epoch": 1.8981481481481481,
|
| 1856 |
-
"grad_norm": 30.178142547607422,
|
| 1857 |
-
"learning_rate": 4.230682825341498e-05,
|
| 1858 |
-
"loss": 3.132,
|
| 1859 |
-
"step": 2460
|
| 1860 |
-
},
|
| 1861 |
-
{
|
| 1862 |
-
"epoch": 1.9058641975308643,
|
| 1863 |
-
"grad_norm": 49.78887939453125,
|
| 1864 |
-
"learning_rate": 4.205075288292717e-05,
|
| 1865 |
-
"loss": 3.2772,
|
| 1866 |
-
"step": 2470
|
| 1867 |
-
},
|
| 1868 |
-
{
|
| 1869 |
-
"epoch": 1.9135802469135803,
|
| 1870 |
-
"grad_norm": 22.746013641357422,
|
| 1871 |
-
"learning_rate": 4.17945932140206e-05,
|
| 1872 |
-
"loss": 3.257,
|
| 1873 |
-
"step": 2480
|
| 1874 |
-
},
|
| 1875 |
-
{
|
| 1876 |
-
"epoch": 1.9212962962962963,
|
| 1877 |
-
"grad_norm": 29.011247634887695,
|
| 1878 |
-
"learning_rate": 4.1538359776415936e-05,
|
| 1879 |
-
"loss": 2.9449,
|
| 1880 |
-
"step": 2490
|
| 1881 |
-
},
|
| 1882 |
-
{
|
| 1883 |
-
"epoch": 1.9290123456790123,
|
| 1884 |
-
"grad_norm": 29.394569396972656,
|
| 1885 |
-
"learning_rate": 4.128206310286622e-05,
|
| 1886 |
-
"loss": 3.1816,
|
| 1887 |
-
"step": 2500
|
| 1888 |
-
},
|
| 1889 |
-
{
|
| 1890 |
-
"epoch": 1.9367283950617284,
|
| 1891 |
-
"grad_norm": 41.661983489990234,
|
| 1892 |
-
"learning_rate": 4.102571372872382e-05,
|
| 1893 |
-
"loss": 3.1338,
|
| 1894 |
-
"step": 2510
|
| 1895 |
-
},
|
| 1896 |
-
{
|
| 1897 |
-
"epoch": 1.9444444444444444,
|
| 1898 |
-
"grad_norm": 24.220867156982422,
|
| 1899 |
-
"learning_rate": 4.0769322191507485e-05,
|
| 1900 |
-
"loss": 3.0679,
|
| 1901 |
-
"step": 2520
|
| 1902 |
-
},
|
| 1903 |
-
{
|
| 1904 |
-
"epoch": 1.9521604938271606,
|
| 1905 |
-
"grad_norm": 30.152782440185547,
|
| 1906 |
-
"learning_rate": 4.051289903046909e-05,
|
| 1907 |
-
"loss": 3.0104,
|
| 1908 |
-
"step": 2530
|
| 1909 |
-
},
|
| 1910 |
-
{
|
| 1911 |
-
"epoch": 1.9598765432098766,
|
| 1912 |
-
"grad_norm": 50.978118896484375,
|
| 1913 |
-
"learning_rate": 4.025645478616045e-05,
|
| 1914 |
-
"loss": 3.1816,
|
| 1915 |
-
"step": 2540
|
| 1916 |
-
},
|
| 1917 |
-
{
|
| 1918 |
-
"epoch": 1.9675925925925926,
|
| 1919 |
-
"grad_norm": 30.627073287963867,
|
| 1920 |
-
"learning_rate": 4e-05,
|
| 1921 |
-
"loss": 3.2438,
|
| 1922 |
-
"step": 2550
|
| 1923 |
-
},
|
| 1924 |
-
{
|
| 1925 |
-
"epoch": 1.9675925925925926,
|
| 1926 |
-
"eval_loss": 3.288115978240967,
|
| 1927 |
-
"eval_runtime": 17.6092,
|
| 1928 |
-
"eval_samples_per_second": 28.394,
|
| 1929 |
-
"eval_steps_per_second": 28.394,
|
| 1930 |
-
"step": 2550
|
| 1931 |
-
},
|
| 1932 |
-
{
|
| 1933 |
-
"epoch": 1.9753086419753085,
|
| 1934 |
-
"grad_norm": 43.21097183227539,
|
| 1935 |
-
"learning_rate": 3.974354521383956e-05,
|
| 1936 |
-
"loss": 2.9912,
|
| 1937 |
-
"step": 2560
|
| 1938 |
-
},
|
| 1939 |
-
{
|
| 1940 |
-
"epoch": 1.9830246913580247,
|
| 1941 |
-
"grad_norm": 29.048534393310547,
|
| 1942 |
-
"learning_rate": 3.948710096953091e-05,
|
| 1943 |
-
"loss": 3.2144,
|
| 1944 |
-
"step": 2570
|
| 1945 |
-
},
|
| 1946 |
-
{
|
| 1947 |
-
"epoch": 1.9907407407407407,
|
| 1948 |
-
"grad_norm": 30.724382400512695,
|
| 1949 |
-
"learning_rate": 3.923067780849252e-05,
|
| 1950 |
-
"loss": 3.1849,
|
| 1951 |
-
"step": 2580
|
| 1952 |
-
},
|
| 1953 |
-
{
|
| 1954 |
-
"epoch": 1.998456790123457,
|
| 1955 |
-
"grad_norm": 32.7701530456543,
|
| 1956 |
-
"learning_rate": 3.8974286271276185e-05,
|
| 1957 |
-
"loss": 3.2466,
|
| 1958 |
-
"step": 2590
|
| 1959 |
-
},
|
| 1960 |
-
{
|
| 1961 |
-
"epoch": 2.006172839506173,
|
| 1962 |
-
"grad_norm": 25.29627799987793,
|
| 1963 |
-
"learning_rate": 3.87179368971338e-05,
|
| 1964 |
-
"loss": 3.0169,
|
| 1965 |
-
"step": 2600
|
| 1966 |
-
},
|
| 1967 |
-
{
|
| 1968 |
-
"epoch": 2.013888888888889,
|
| 1969 |
-
"grad_norm": 33.91135787963867,
|
| 1970 |
-
"learning_rate": 3.8461640223584064e-05,
|
| 1971 |
-
"loss": 2.8224,
|
| 1972 |
-
"step": 2610
|
| 1973 |
-
},
|
| 1974 |
-
{
|
| 1975 |
-
"epoch": 2.021604938271605,
|
| 1976 |
-
"grad_norm": 26.389068603515625,
|
| 1977 |
-
"learning_rate": 3.820540678597942e-05,
|
| 1978 |
-
"loss": 2.6429,
|
| 1979 |
-
"step": 2620
|
| 1980 |
-
},
|
| 1981 |
-
{
|
| 1982 |
-
"epoch": 2.029320987654321,
|
| 1983 |
-
"grad_norm": 23.78996467590332,
|
| 1984 |
-
"learning_rate": 3.794924711707284e-05,
|
| 1985 |
-
"loss": 2.7442,
|
| 1986 |
-
"step": 2630
|
| 1987 |
-
},
|
| 1988 |
-
{
|
| 1989 |
-
"epoch": 2.037037037037037,
|
| 1990 |
-
"grad_norm": 30.783973693847656,
|
| 1991 |
-
"learning_rate": 3.769317174658503e-05,
|
| 1992 |
-
"loss": 2.9359,
|
| 1993 |
-
"step": 2640
|
| 1994 |
-
},
|
| 1995 |
-
{
|
| 1996 |
-
"epoch": 2.044753086419753,
|
| 1997 |
-
"grad_norm": 23.419214248657227,
|
| 1998 |
-
"learning_rate": 3.743719120077149e-05,
|
| 1999 |
-
"loss": 2.9642,
|
| 2000 |
-
"step": 2650
|
| 2001 |
-
},
|
| 2002 |
-
{
|
| 2003 |
-
"epoch": 2.052469135802469,
|
| 2004 |
-
"grad_norm": 43.545475006103516,
|
| 2005 |
-
"learning_rate": 3.718131600198984e-05,
|
| 2006 |
-
"loss": 2.7393,
|
| 2007 |
-
"step": 2660
|
| 2008 |
-
},
|
| 2009 |
-
{
|
| 2010 |
-
"epoch": 2.060185185185185,
|
| 2011 |
-
"grad_norm": 27.782304763793945,
|
| 2012 |
-
"learning_rate": 3.6925556668267346e-05,
|
| 2013 |
-
"loss": 3.1423,
|
| 2014 |
-
"step": 2670
|
| 2015 |
-
},
|
| 2016 |
-
{
|
| 2017 |
-
"epoch": 2.067901234567901,
|
| 2018 |
-
"grad_norm": 31.327932357788086,
|
| 2019 |
-
"learning_rate": 3.666992371286843e-05,
|
| 2020 |
-
"loss": 2.8856,
|
| 2021 |
-
"step": 2680
|
| 2022 |
-
},
|
| 2023 |
-
{
|
| 2024 |
-
"epoch": 2.075617283950617,
|
| 2025 |
-
"grad_norm": 27.207284927368164,
|
| 2026 |
-
"learning_rate": 3.6414427643862664e-05,
|
| 2027 |
-
"loss": 3.0525,
|
| 2028 |
-
"step": 2690
|
| 2029 |
-
},
|
| 2030 |
-
{
|
| 2031 |
-
"epoch": 2.0833333333333335,
|
| 2032 |
-
"grad_norm": 27.2406005859375,
|
| 2033 |
-
"learning_rate": 3.615907896369273e-05,
|
| 2034 |
-
"loss": 2.7817,
|
| 2035 |
-
"step": 2700
|
| 2036 |
-
},
|
| 2037 |
-
{
|
| 2038 |
-
"epoch": 2.0833333333333335,
|
| 2039 |
-
"eval_loss": 3.2747247219085693,
|
| 2040 |
-
"eval_runtime": 20.4314,
|
| 2041 |
-
"eval_samples_per_second": 24.472,
|
| 2042 |
-
"eval_steps_per_second": 24.472,
|
| 2043 |
-
"step": 2700
|
| 2044 |
-
},
|
| 2045 |
-
{
|
| 2046 |
-
"epoch": 2.0910493827160495,
|
| 2047 |
-
"grad_norm": 33.61459732055664,
|
| 2048 |
-
"learning_rate": 3.5903888168742754e-05,
|
| 2049 |
-
"loss": 2.8164,
|
| 2050 |
-
"step": 2710
|
| 2051 |
-
},
|
| 2052 |
-
{
|
| 2053 |
-
"epoch": 2.0987654320987654,
|
| 2054 |
-
"grad_norm": 34.80902099609375,
|
| 2055 |
-
"learning_rate": 3.564886574890677e-05,
|
| 2056 |
-
"loss": 2.6692,
|
| 2057 |
-
"step": 2720
|
| 2058 |
-
},
|
| 2059 |
-
{
|
| 2060 |
-
"epoch": 2.1064814814814814,
|
| 2061 |
-
"grad_norm": 40.73919677734375,
|
| 2062 |
-
"learning_rate": 3.539402218715763e-05,
|
| 2063 |
-
"loss": 2.723,
|
| 2064 |
-
"step": 2730
|
| 2065 |
-
},
|
| 2066 |
-
{
|
| 2067 |
-
"epoch": 2.1141975308641974,
|
| 2068 |
-
"grad_norm": 31.271671295166016,
|
| 2069 |
-
"learning_rate": 3.5139367959115986e-05,
|
| 2070 |
-
"loss": 2.7958,
|
| 2071 |
-
"step": 2740
|
| 2072 |
-
},
|
| 2073 |
-
{
|
| 2074 |
-
"epoch": 2.121913580246914,
|
| 2075 |
-
"grad_norm": 29.13555908203125,
|
| 2076 |
-
"learning_rate": 3.4884913532619756e-05,
|
| 2077 |
-
"loss": 2.8345,
|
| 2078 |
-
"step": 2750
|
| 2079 |
-
},
|
| 2080 |
-
{
|
| 2081 |
-
"epoch": 2.1296296296296298,
|
| 2082 |
-
"grad_norm": 24.78423309326172,
|
| 2083 |
-
"learning_rate": 3.4630669367293797e-05,
|
| 2084 |
-
"loss": 2.6955,
|
| 2085 |
-
"step": 2760
|
| 2086 |
-
},
|
| 2087 |
-
{
|
| 2088 |
-
"epoch": 2.1373456790123457,
|
| 2089 |
-
"grad_norm": 33.64024353027344,
|
| 2090 |
-
"learning_rate": 3.43766459141199e-05,
|
| 2091 |
-
"loss": 2.801,
|
| 2092 |
-
"step": 2770
|
| 2093 |
-
},
|
| 2094 |
-
{
|
| 2095 |
-
"epoch": 2.1450617283950617,
|
| 2096 |
-
"grad_norm": 34.713993072509766,
|
| 2097 |
-
"learning_rate": 3.412285361500729e-05,
|
| 2098 |
-
"loss": 2.7948,
|
| 2099 |
-
"step": 2780
|
| 2100 |
-
},
|
| 2101 |
-
{
|
| 2102 |
-
"epoch": 2.1527777777777777,
|
| 2103 |
-
"grad_norm": 158.35733032226562,
|
| 2104 |
-
"learning_rate": 3.386930290236336e-05,
|
| 2105 |
-
"loss": 2.8969,
|
| 2106 |
-
"step": 2790
|
| 2107 |
-
},
|
| 2108 |
-
{
|
| 2109 |
-
"epoch": 2.1604938271604937,
|
| 2110 |
-
"grad_norm": 23.90937614440918,
|
| 2111 |
-
"learning_rate": 3.3616004198664845e-05,
|
| 2112 |
-
"loss": 2.7715,
|
| 2113 |
-
"step": 2800
|
| 2114 |
-
},
|
| 2115 |
-
{
|
| 2116 |
-
"epoch": 2.16820987654321,
|
| 2117 |
-
"grad_norm": 36.69210433959961,
|
| 2118 |
-
"learning_rate": 3.33629679160293e-05,
|
| 2119 |
-
"loss": 3.0046,
|
| 2120 |
-
"step": 2810
|
| 2121 |
-
},
|
| 2122 |
-
{
|
| 2123 |
-
"epoch": 2.175925925925926,
|
| 2124 |
-
"grad_norm": 32.34378433227539,
|
| 2125 |
-
"learning_rate": 3.311020445578725e-05,
|
| 2126 |
-
"loss": 2.7916,
|
| 2127 |
-
"step": 2820
|
| 2128 |
-
},
|
| 2129 |
-
{
|
| 2130 |
-
"epoch": 2.183641975308642,
|
| 2131 |
-
"grad_norm": 27.218441009521484,
|
| 2132 |
-
"learning_rate": 3.285772420805454e-05,
|
| 2133 |
-
"loss": 2.8711,
|
| 2134 |
-
"step": 2830
|
| 2135 |
-
},
|
| 2136 |
-
{
|
| 2137 |
-
"epoch": 2.191358024691358,
|
| 2138 |
-
"grad_norm": 20.864200592041016,
|
| 2139 |
-
"learning_rate": 3.260553755130525e-05,
|
| 2140 |
-
"loss": 2.8068,
|
| 2141 |
-
"step": 2840
|
| 2142 |
-
},
|
| 2143 |
-
{
|
| 2144 |
-
"epoch": 2.199074074074074,
|
| 2145 |
-
"grad_norm": 33.19184494018555,
|
| 2146 |
-
"learning_rate": 3.235365485194512e-05,
|
| 2147 |
-
"loss": 2.6901,
|
| 2148 |
-
"step": 2850
|
| 2149 |
-
},
|
| 2150 |
-
{
|
| 2151 |
-
"epoch": 2.199074074074074,
|
| 2152 |
-
"eval_loss": 3.244508981704712,
|
| 2153 |
-
"eval_runtime": 18.2444,
|
| 2154 |
-
"eval_samples_per_second": 27.406,
|
| 2155 |
-
"eval_steps_per_second": 27.406,
|
| 2156 |
-
"step": 2850
|
| 2157 |
}
|
| 2158 |
],
|
| 2159 |
"logging_steps": 10,
|
|
@@ -2173,7 +139,7 @@
|
|
| 2173 |
"attributes": {}
|
| 2174 |
}
|
| 2175 |
},
|
| 2176 |
-
"total_flos":
|
| 2177 |
"train_batch_size": 8,
|
| 2178 |
"trial_name": null,
|
| 2179 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 4.406219959259033,
|
| 3 |
+
"best_model_checkpoint": "./output/checkpoint-150",
|
| 4 |
+
"epoch": 0.11574074074074074,
|
| 5 |
"eval_steps": 150,
|
| 6 |
+
"global_step": 150,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.007716049382716049,
|
| 13 |
+
"grad_norm": 25.134618759155273,
|
| 14 |
"learning_rate": 8.000000000000001e-06,
|
| 15 |
+
"loss": 5.0051,
|
| 16 |
"step": 10
|
| 17 |
},
|
| 18 |
{
|
| 19 |
"epoch": 0.015432098765432098,
|
| 20 |
+
"grad_norm": 21.143909454345703,
|
| 21 |
"learning_rate": 1.6000000000000003e-05,
|
| 22 |
+
"loss": 4.7303,
|
| 23 |
"step": 20
|
| 24 |
},
|
| 25 |
{
|
| 26 |
"epoch": 0.023148148148148147,
|
| 27 |
+
"grad_norm": 21.639801025390625,
|
| 28 |
"learning_rate": 2.4e-05,
|
| 29 |
+
"loss": 4.6905,
|
| 30 |
"step": 30
|
| 31 |
},
|
| 32 |
{
|
| 33 |
"epoch": 0.030864197530864196,
|
| 34 |
+
"grad_norm": 21.367475509643555,
|
| 35 |
"learning_rate": 3.2000000000000005e-05,
|
| 36 |
+
"loss": 4.6481,
|
| 37 |
"step": 40
|
| 38 |
},
|
| 39 |
{
|
| 40 |
"epoch": 0.038580246913580245,
|
| 41 |
+
"grad_norm": 22.380338668823242,
|
| 42 |
"learning_rate": 4e-05,
|
| 43 |
+
"loss": 4.6698,
|
| 44 |
"step": 50
|
| 45 |
},
|
| 46 |
{
|
| 47 |
"epoch": 0.046296296296296294,
|
| 48 |
+
"grad_norm": 30.744077682495117,
|
| 49 |
"learning_rate": 4.8e-05,
|
| 50 |
+
"loss": 4.5466,
|
| 51 |
"step": 60
|
| 52 |
},
|
| 53 |
{
|
| 54 |
"epoch": 0.05401234567901234,
|
| 55 |
+
"grad_norm": 21.133874893188477,
|
| 56 |
"learning_rate": 5.6e-05,
|
| 57 |
+
"loss": 4.5706,
|
| 58 |
"step": 70
|
| 59 |
},
|
| 60 |
{
|
| 61 |
"epoch": 0.06172839506172839,
|
| 62 |
+
"grad_norm": 19.068326950073242,
|
| 63 |
"learning_rate": 6.400000000000001e-05,
|
| 64 |
+
"loss": 4.3699,
|
| 65 |
"step": 80
|
| 66 |
},
|
| 67 |
{
|
| 68 |
"epoch": 0.06944444444444445,
|
| 69 |
+
"grad_norm": 20.487281799316406,
|
| 70 |
"learning_rate": 7.2e-05,
|
| 71 |
+
"loss": 4.4229,
|
| 72 |
"step": 90
|
| 73 |
},
|
| 74 |
{
|
| 75 |
"epoch": 0.07716049382716049,
|
| 76 |
+
"grad_norm": 24.00373649597168,
|
| 77 |
"learning_rate": 8e-05,
|
| 78 |
+
"loss": 4.4247,
|
| 79 |
"step": 100
|
| 80 |
},
|
| 81 |
{
|
| 82 |
"epoch": 0.08487654320987655,
|
| 83 |
+
"grad_norm": 20.140514373779297,
|
| 84 |
"learning_rate": 7.999917787833465e-05,
|
| 85 |
+
"loss": 4.2552,
|
| 86 |
"step": 110
|
| 87 |
},
|
| 88 |
{
|
| 89 |
"epoch": 0.09259259259259259,
|
| 90 |
+
"grad_norm": 22.779335021972656,
|
| 91 |
"learning_rate": 7.999671154713278e-05,
|
| 92 |
+
"loss": 4.2996,
|
| 93 |
"step": 120
|
| 94 |
},
|
| 95 |
{
|
| 96 |
"epoch": 0.10030864197530864,
|
| 97 |
+
"grad_norm": 15.848005294799805,
|
| 98 |
"learning_rate": 7.99926011077756e-05,
|
| 99 |
+
"loss": 4.359,
|
| 100 |
"step": 130
|
| 101 |
},
|
| 102 |
{
|
| 103 |
"epoch": 0.10802469135802469,
|
| 104 |
+
"grad_norm": 19.472209930419922,
|
| 105 |
"learning_rate": 7.99868467292272e-05,
|
| 106 |
+
"loss": 4.3415,
|
| 107 |
"step": 140
|
| 108 |
},
|
| 109 |
{
|
| 110 |
"epoch": 0.11574074074074074,
|
| 111 |
+
"grad_norm": 13.642756462097168,
|
| 112 |
"learning_rate": 7.997944864802752e-05,
|
| 113 |
+
"loss": 4.4707,
|
| 114 |
"step": 150
|
| 115 |
},
|
| 116 |
{
|
| 117 |
"epoch": 0.11574074074074074,
|
| 118 |
+
"eval_loss": 4.406219959259033,
|
| 119 |
+
"eval_runtime": 17.8179,
|
| 120 |
+
"eval_samples_per_second": 28.062,
|
| 121 |
+
"eval_steps_per_second": 28.062,
|
| 122 |
"step": 150
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 123 |
}
|
| 124 |
],
|
| 125 |
"logging_steps": 10,
|
|
|
|
| 139 |
"attributes": {}
|
| 140 |
}
|
| 141 |
},
|
| 142 |
+
"total_flos": 232102392496128.0,
|
| 143 |
"train_batch_size": 8,
|
| 144 |
"trial_name": null,
|
| 145 |
"trial_params": null
|
last-checkpoint/training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5368
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6605c2de7a844ad29d351b6c36722a29860e2329761ad63ddc14664ce30af171
|
| 3 |
size 5368
|