Training in progress, step 5000, checkpoint
Browse files- last-checkpoint/config.json +1 -1
- last-checkpoint/model.safetensors +1 -1
- last-checkpoint/optimizer.pt +1 -1
- last-checkpoint/rng_state_0.pth +2 -2
- last-checkpoint/rng_state_1.pth +1 -1
- last-checkpoint/rng_state_2.pth +1 -1
- last-checkpoint/rng_state_3.pth +1 -1
- last-checkpoint/rng_state_4.pth +1 -1
- last-checkpoint/rng_state_5.pth +1 -1
- last-checkpoint/rng_state_6.pth +1 -1
- last-checkpoint/rng_state_7.pth +1 -1
- last-checkpoint/scheduler.pt +1 -1
- last-checkpoint/trainer_state.json +409 -1111
last-checkpoint/config.json
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
{
|
| 2 |
-
"_name_or_path": "x2bee/KoModernBERT-base-mlm-v02-
|
| 3 |
"architectures": [
|
| 4 |
"ModernBertForMaskedLM"
|
| 5 |
],
|
|
|
|
| 1 |
{
|
| 2 |
+
"_name_or_path": "x2bee/KoModernBERT-base-mlm-v02-ckp02",
|
| 3 |
"architectures": [
|
| 4 |
"ModernBertForMaskedLM"
|
| 5 |
],
|
last-checkpoint/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 368988278
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6b42fde2d931440aeb105a7aa23055e1283fb71687fa5d3f2553594d5b2bda4d
|
| 3 |
size 368988278
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1107079290
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:69b75254b559a304e81391980e81b803443819ae3cc5363a1be04e6ff7aeac9a
|
| 3 |
size 1107079290
|
last-checkpoint/rng_state_0.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 sha256:78d3f197f6c6558fa8056324f1563ab9e957255f5a1a959362aa4eed7a9545db
|
| 3 |
+
size 15984
|
last-checkpoint/rng_state_1.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15984
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1c1a9c65c2869356282cad6b4a0f7dff7f4dd68ab3d9d216c72b7d6cb524f860
|
| 3 |
size 15984
|
last-checkpoint/rng_state_2.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15984
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:896febe768e17bae5022a95960c041f6425783774ec8859d99d3b149063b1bf9
|
| 3 |
size 15984
|
last-checkpoint/rng_state_3.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15984
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:eac482d57e966585467c8ef44dae2869bf7e5d92886f69c11ed7bccc34c07efe
|
| 3 |
size 15984
|
last-checkpoint/rng_state_4.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15984
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e1f27d227a20dc320ac283e0938fb2f6e5b475829a583f8c44d1a16a8c828307
|
| 3 |
size 15984
|
last-checkpoint/rng_state_5.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15984
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d05a7106aaeaec4b81704e3f4a998b5123cf9342a6733bd9fd2d578e99108c3b
|
| 3 |
size 15984
|
last-checkpoint/rng_state_6.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15984
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b94120d8d88502ec8d8b623ec7550315caca003b44fcffbb5767ab0de91baefe
|
| 3 |
size 15984
|
last-checkpoint/rng_state_7.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15984
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:332e4d901be380f740b5d8578f7b80ef1865c7fba83bc288c8a35852205cc668
|
| 3 |
size 15984
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1000
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2e7caefe5b3c11741d0fb75575bef604da14e2f80b3dca75692f7742d5d3ab3f
|
| 3 |
size 1000
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,1426 +1,724 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 3000,
|
| 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.
|
| 13 |
-
"grad_norm":
|
| 14 |
-
"learning_rate":
|
| 15 |
-
"loss":
|
| 16 |
"step": 50
|
| 17 |
},
|
| 18 |
{
|
| 19 |
-
"epoch": 0.
|
| 20 |
-
"grad_norm":
|
| 21 |
-
"learning_rate":
|
| 22 |
-
"loss":
|
| 23 |
"step": 100
|
| 24 |
},
|
| 25 |
{
|
| 26 |
-
"epoch": 0.
|
| 27 |
-
"grad_norm":
|
| 28 |
-
"learning_rate":
|
| 29 |
-
"loss":
|
| 30 |
"step": 150
|
| 31 |
},
|
| 32 |
{
|
| 33 |
-
"epoch": 0.
|
| 34 |
-
"grad_norm":
|
| 35 |
-
"learning_rate":
|
| 36 |
-
"loss":
|
| 37 |
"step": 200
|
| 38 |
},
|
| 39 |
{
|
| 40 |
-
"epoch": 0.
|
| 41 |
-
"grad_norm":
|
| 42 |
-
"learning_rate":
|
| 43 |
-
"loss":
|
| 44 |
"step": 250
|
| 45 |
},
|
| 46 |
{
|
| 47 |
-
"epoch": 0.
|
| 48 |
-
"grad_norm":
|
| 49 |
-
"learning_rate":
|
| 50 |
-
"loss":
|
| 51 |
"step": 300
|
| 52 |
},
|
| 53 |
{
|
| 54 |
-
"epoch": 0.
|
| 55 |
-
"grad_norm":
|
| 56 |
-
"learning_rate":
|
| 57 |
-
"loss":
|
| 58 |
"step": 350
|
| 59 |
},
|
| 60 |
{
|
| 61 |
-
"epoch": 0.
|
| 62 |
-
"grad_norm":
|
| 63 |
-
"learning_rate":
|
| 64 |
-
"loss":
|
| 65 |
"step": 400
|
| 66 |
},
|
| 67 |
{
|
| 68 |
-
"epoch": 0.
|
| 69 |
-
"grad_norm":
|
| 70 |
-
"learning_rate":
|
| 71 |
-
"loss":
|
| 72 |
"step": 450
|
| 73 |
},
|
| 74 |
{
|
| 75 |
-
"epoch": 0.
|
| 76 |
-
"grad_norm":
|
| 77 |
-
"learning_rate":
|
| 78 |
-
"loss":
|
| 79 |
"step": 500
|
| 80 |
},
|
| 81 |
{
|
| 82 |
-
"epoch": 0.
|
| 83 |
-
"grad_norm":
|
| 84 |
-
"learning_rate":
|
| 85 |
-
"loss":
|
| 86 |
"step": 550
|
| 87 |
},
|
| 88 |
{
|
| 89 |
-
"epoch": 0.
|
| 90 |
-
"grad_norm":
|
| 91 |
-
"learning_rate":
|
| 92 |
-
"loss":
|
| 93 |
"step": 600
|
| 94 |
},
|
| 95 |
{
|
| 96 |
-
"epoch": 0.
|
| 97 |
-
"grad_norm":
|
| 98 |
-
"learning_rate":
|
| 99 |
-
"loss":
|
| 100 |
"step": 650
|
| 101 |
},
|
| 102 |
{
|
| 103 |
-
"epoch": 0.
|
| 104 |
-
"grad_norm": 21.
|
| 105 |
-
"learning_rate":
|
| 106 |
-
"loss":
|
| 107 |
"step": 700
|
| 108 |
},
|
| 109 |
{
|
| 110 |
-
"epoch": 0.
|
| 111 |
-
"grad_norm":
|
| 112 |
-
"learning_rate":
|
| 113 |
-
"loss":
|
| 114 |
"step": 750
|
| 115 |
},
|
| 116 |
{
|
| 117 |
-
"epoch": 0.
|
| 118 |
-
"grad_norm":
|
| 119 |
-
"learning_rate":
|
| 120 |
-
"loss":
|
| 121 |
"step": 800
|
| 122 |
},
|
| 123 |
{
|
| 124 |
-
"epoch": 0.
|
| 125 |
-
"grad_norm":
|
| 126 |
-
"learning_rate":
|
| 127 |
-
"loss":
|
| 128 |
"step": 850
|
| 129 |
},
|
| 130 |
{
|
| 131 |
-
"epoch": 0.
|
| 132 |
-
"grad_norm":
|
| 133 |
-
"learning_rate":
|
| 134 |
-
"loss":
|
| 135 |
"step": 900
|
| 136 |
},
|
| 137 |
{
|
| 138 |
-
"epoch": 0.
|
| 139 |
-
"grad_norm":
|
| 140 |
-
"learning_rate":
|
| 141 |
-
"loss":
|
| 142 |
"step": 950
|
| 143 |
},
|
| 144 |
{
|
| 145 |
-
"epoch": 0.
|
| 146 |
-
"grad_norm":
|
| 147 |
-
"learning_rate":
|
| 148 |
-
"loss":
|
| 149 |
"step": 1000
|
| 150 |
},
|
| 151 |
{
|
| 152 |
-
"epoch": 0.
|
| 153 |
-
"grad_norm":
|
| 154 |
-
"learning_rate":
|
| 155 |
-
"loss":
|
| 156 |
"step": 1050
|
| 157 |
},
|
| 158 |
{
|
| 159 |
-
"epoch": 0.
|
| 160 |
-
"grad_norm":
|
| 161 |
-
"learning_rate": 1.
|
| 162 |
-
"loss":
|
| 163 |
"step": 1100
|
| 164 |
},
|
| 165 |
{
|
| 166 |
-
"epoch": 0.
|
| 167 |
-
"grad_norm":
|
| 168 |
-
"learning_rate": 1.
|
| 169 |
-
"loss":
|
| 170 |
"step": 1150
|
| 171 |
},
|
| 172 |
{
|
| 173 |
-
"epoch": 0.
|
| 174 |
-
"grad_norm":
|
| 175 |
-
"learning_rate": 1.
|
| 176 |
-
"loss":
|
| 177 |
"step": 1200
|
| 178 |
},
|
| 179 |
{
|
| 180 |
-
"epoch": 0.
|
| 181 |
-
"grad_norm":
|
| 182 |
-
"learning_rate": 1.
|
| 183 |
-
"loss":
|
| 184 |
"step": 1250
|
| 185 |
},
|
| 186 |
{
|
| 187 |
-
"epoch": 0.
|
| 188 |
-
"grad_norm": 20.
|
| 189 |
-
"learning_rate":
|
| 190 |
-
"loss":
|
| 191 |
"step": 1300
|
| 192 |
},
|
| 193 |
{
|
| 194 |
-
"epoch": 0.
|
| 195 |
-
"grad_norm":
|
| 196 |
-
"learning_rate":
|
| 197 |
-
"loss":
|
| 198 |
"step": 1350
|
| 199 |
},
|
| 200 |
{
|
| 201 |
-
"epoch": 0.
|
| 202 |
-
"grad_norm":
|
| 203 |
-
"learning_rate":
|
| 204 |
-
"loss":
|
| 205 |
"step": 1400
|
| 206 |
},
|
| 207 |
{
|
| 208 |
-
"epoch": 0.
|
| 209 |
"grad_norm": 21.15625,
|
| 210 |
-
"learning_rate":
|
| 211 |
-
"loss":
|
| 212 |
"step": 1450
|
| 213 |
},
|
| 214 |
{
|
| 215 |
-
"epoch": 0.
|
| 216 |
-
"grad_norm":
|
| 217 |
-
"learning_rate":
|
| 218 |
-
"loss":
|
| 219 |
"step": 1500
|
| 220 |
},
|
| 221 |
{
|
| 222 |
-
"epoch": 0.
|
| 223 |
-
"grad_norm":
|
| 224 |
-
"learning_rate":
|
| 225 |
-
"loss":
|
| 226 |
"step": 1550
|
| 227 |
},
|
| 228 |
{
|
| 229 |
-
"epoch": 0.
|
| 230 |
-
"grad_norm":
|
| 231 |
-
"learning_rate":
|
| 232 |
-
"loss":
|
| 233 |
"step": 1600
|
| 234 |
},
|
| 235 |
{
|
| 236 |
-
"epoch": 0.
|
| 237 |
-
"grad_norm":
|
| 238 |
-
"learning_rate":
|
| 239 |
-
"loss":
|
| 240 |
"step": 1650
|
| 241 |
},
|
| 242 |
{
|
| 243 |
-
"epoch": 0.
|
| 244 |
-
"grad_norm": 21.
|
| 245 |
-
"learning_rate":
|
| 246 |
-
"loss":
|
| 247 |
"step": 1700
|
| 248 |
},
|
| 249 |
{
|
| 250 |
-
"epoch": 0.
|
| 251 |
-
"grad_norm":
|
| 252 |
-
"learning_rate":
|
| 253 |
-
"loss":
|
| 254 |
"step": 1750
|
| 255 |
},
|
| 256 |
{
|
| 257 |
-
"epoch": 0.
|
| 258 |
-
"grad_norm":
|
| 259 |
-
"learning_rate":
|
| 260 |
-
"loss":
|
| 261 |
"step": 1800
|
| 262 |
},
|
| 263 |
{
|
| 264 |
-
"epoch": 0.
|
| 265 |
-
"grad_norm":
|
| 266 |
-
"learning_rate":
|
| 267 |
-
"loss":
|
| 268 |
"step": 1850
|
| 269 |
},
|
| 270 |
{
|
| 271 |
-
"epoch": 0.
|
| 272 |
-
"grad_norm":
|
| 273 |
-
"learning_rate":
|
| 274 |
-
"loss":
|
| 275 |
"step": 1900
|
| 276 |
},
|
| 277 |
{
|
| 278 |
-
"epoch": 0.
|
| 279 |
-
"grad_norm":
|
| 280 |
-
"learning_rate":
|
| 281 |
-
"loss":
|
| 282 |
"step": 1950
|
| 283 |
},
|
| 284 |
{
|
| 285 |
-
"epoch": 0.
|
| 286 |
-
"grad_norm":
|
| 287 |
-
"learning_rate":
|
| 288 |
-
"loss":
|
| 289 |
"step": 2000
|
| 290 |
},
|
| 291 |
{
|
| 292 |
-
"epoch": 0.
|
| 293 |
-
"grad_norm":
|
| 294 |
-
"learning_rate":
|
| 295 |
-
"loss":
|
| 296 |
"step": 2050
|
| 297 |
},
|
| 298 |
{
|
| 299 |
-
"epoch": 0.
|
| 300 |
-
"grad_norm":
|
| 301 |
-
"learning_rate":
|
| 302 |
-
"loss":
|
| 303 |
"step": 2100
|
| 304 |
},
|
| 305 |
{
|
| 306 |
-
"epoch": 0.
|
| 307 |
-
"grad_norm":
|
| 308 |
-
"learning_rate":
|
| 309 |
-
"loss":
|
| 310 |
"step": 2150
|
| 311 |
},
|
| 312 |
{
|
| 313 |
-
"epoch": 0.
|
| 314 |
-
"grad_norm": 20.
|
| 315 |
-
"learning_rate":
|
| 316 |
-
"loss":
|
| 317 |
"step": 2200
|
| 318 |
},
|
| 319 |
{
|
| 320 |
-
"epoch": 0.
|
| 321 |
-
"grad_norm":
|
| 322 |
-
"learning_rate":
|
| 323 |
-
"loss":
|
| 324 |
"step": 2250
|
| 325 |
},
|
| 326 |
{
|
| 327 |
-
"epoch": 0.
|
| 328 |
-
"grad_norm":
|
| 329 |
-
"learning_rate":
|
| 330 |
-
"loss":
|
| 331 |
"step": 2300
|
| 332 |
},
|
| 333 |
{
|
| 334 |
-
"epoch": 0.
|
| 335 |
-
"grad_norm":
|
| 336 |
-
"learning_rate":
|
| 337 |
-
"loss":
|
| 338 |
"step": 2350
|
| 339 |
},
|
| 340 |
{
|
| 341 |
-
"epoch": 0.
|
| 342 |
-
"grad_norm":
|
| 343 |
-
"learning_rate":
|
| 344 |
-
"loss":
|
| 345 |
"step": 2400
|
| 346 |
},
|
| 347 |
{
|
| 348 |
-
"epoch": 0.
|
| 349 |
-
"grad_norm":
|
| 350 |
-
"learning_rate":
|
| 351 |
-
"loss":
|
| 352 |
"step": 2450
|
| 353 |
},
|
| 354 |
{
|
| 355 |
-
"epoch": 0.
|
| 356 |
-
"grad_norm":
|
| 357 |
-
"learning_rate":
|
| 358 |
-
"loss":
|
| 359 |
"step": 2500
|
| 360 |
},
|
| 361 |
{
|
| 362 |
-
"epoch": 0.
|
| 363 |
-
"grad_norm": 21.
|
| 364 |
-
"learning_rate":
|
| 365 |
-
"loss":
|
| 366 |
"step": 2550
|
| 367 |
},
|
| 368 |
{
|
| 369 |
-
"epoch": 0.
|
| 370 |
-
"grad_norm":
|
| 371 |
-
"learning_rate":
|
| 372 |
-
"loss":
|
| 373 |
"step": 2600
|
| 374 |
},
|
| 375 |
{
|
| 376 |
-
"epoch": 0.
|
| 377 |
-
"grad_norm":
|
| 378 |
-
"learning_rate":
|
| 379 |
-
"loss":
|
| 380 |
"step": 2650
|
| 381 |
},
|
| 382 |
{
|
| 383 |
-
"epoch": 0.
|
| 384 |
-
"grad_norm":
|
| 385 |
-
"learning_rate":
|
| 386 |
-
"loss":
|
| 387 |
"step": 2700
|
| 388 |
},
|
| 389 |
{
|
| 390 |
-
"epoch": 0.
|
| 391 |
-
"grad_norm":
|
| 392 |
-
"learning_rate":
|
| 393 |
-
"loss":
|
| 394 |
"step": 2750
|
| 395 |
},
|
| 396 |
{
|
| 397 |
-
"epoch": 0.
|
| 398 |
-
"grad_norm":
|
| 399 |
-
"learning_rate":
|
| 400 |
-
"loss":
|
| 401 |
"step": 2800
|
| 402 |
},
|
| 403 |
{
|
| 404 |
-
"epoch": 0.
|
| 405 |
-
"grad_norm":
|
| 406 |
-
"learning_rate":
|
| 407 |
-
"loss":
|
| 408 |
"step": 2850
|
| 409 |
},
|
| 410 |
{
|
| 411 |
-
"epoch": 0.
|
| 412 |
-
"grad_norm":
|
| 413 |
-
"learning_rate":
|
| 414 |
-
"loss":
|
| 415 |
"step": 2900
|
| 416 |
},
|
| 417 |
{
|
| 418 |
-
"epoch": 0.
|
| 419 |
-
"grad_norm":
|
| 420 |
-
"learning_rate":
|
| 421 |
-
"loss":
|
| 422 |
"step": 2950
|
| 423 |
},
|
| 424 |
{
|
| 425 |
-
"epoch": 0.
|
| 426 |
-
"grad_norm":
|
| 427 |
-
"learning_rate":
|
| 428 |
-
"loss":
|
| 429 |
"step": 3000
|
| 430 |
},
|
| 431 |
{
|
| 432 |
-
"epoch": 0.
|
| 433 |
-
"eval_loss": 2.
|
| 434 |
-
"eval_runtime":
|
| 435 |
-
"eval_samples_per_second":
|
| 436 |
-
"eval_steps_per_second":
|
| 437 |
"step": 3000
|
| 438 |
},
|
| 439 |
{
|
| 440 |
-
"epoch": 0.
|
| 441 |
-
"grad_norm":
|
| 442 |
-
"learning_rate":
|
| 443 |
-
"loss":
|
| 444 |
"step": 3050
|
| 445 |
},
|
| 446 |
{
|
| 447 |
-
"epoch": 0.
|
| 448 |
-
"grad_norm":
|
| 449 |
-
"learning_rate":
|
| 450 |
-
"loss":
|
| 451 |
"step": 3100
|
| 452 |
},
|
| 453 |
{
|
| 454 |
-
"epoch": 0.
|
| 455 |
-
"grad_norm":
|
| 456 |
-
"learning_rate":
|
| 457 |
-
"loss":
|
| 458 |
"step": 3150
|
| 459 |
},
|
| 460 |
{
|
| 461 |
-
"epoch": 0.
|
| 462 |
-
"grad_norm":
|
| 463 |
-
"learning_rate":
|
| 464 |
-
"loss":
|
| 465 |
"step": 3200
|
| 466 |
},
|
| 467 |
{
|
| 468 |
-
"epoch": 0.
|
| 469 |
-
"grad_norm":
|
| 470 |
-
"learning_rate":
|
| 471 |
-
"loss":
|
| 472 |
"step": 3250
|
| 473 |
},
|
| 474 |
{
|
| 475 |
-
"epoch": 0.
|
| 476 |
-
"grad_norm":
|
| 477 |
-
"learning_rate":
|
| 478 |
-
"loss":
|
| 479 |
"step": 3300
|
| 480 |
},
|
| 481 |
{
|
| 482 |
-
"epoch": 0.
|
| 483 |
-
"grad_norm":
|
| 484 |
-
"learning_rate":
|
| 485 |
-
"loss":
|
| 486 |
"step": 3350
|
| 487 |
},
|
| 488 |
{
|
| 489 |
-
"epoch": 0.
|
| 490 |
-
"grad_norm":
|
| 491 |
-
"learning_rate":
|
| 492 |
-
"loss":
|
| 493 |
"step": 3400
|
| 494 |
},
|
| 495 |
{
|
| 496 |
-
"epoch": 0.
|
| 497 |
-
"grad_norm":
|
| 498 |
-
"learning_rate":
|
| 499 |
-
"loss":
|
| 500 |
"step": 3450
|
| 501 |
},
|
| 502 |
{
|
| 503 |
-
"epoch": 0.
|
| 504 |
-
"grad_norm":
|
| 505 |
-
"learning_rate":
|
| 506 |
-
"loss":
|
| 507 |
"step": 3500
|
| 508 |
},
|
| 509 |
{
|
| 510 |
-
"epoch": 0.
|
| 511 |
-
"grad_norm":
|
| 512 |
-
"learning_rate":
|
| 513 |
-
"loss":
|
| 514 |
"step": 3550
|
| 515 |
},
|
| 516 |
{
|
| 517 |
-
"epoch": 0.
|
| 518 |
-
"grad_norm":
|
| 519 |
-
"learning_rate":
|
| 520 |
-
"loss":
|
| 521 |
"step": 3600
|
| 522 |
},
|
| 523 |
{
|
| 524 |
-
"epoch": 0.
|
| 525 |
-
"grad_norm": 21.
|
| 526 |
-
"learning_rate":
|
| 527 |
-
"loss":
|
| 528 |
"step": 3650
|
| 529 |
},
|
| 530 |
{
|
| 531 |
-
"epoch": 0.
|
| 532 |
-
"grad_norm":
|
| 533 |
-
"learning_rate":
|
| 534 |
-
"loss":
|
| 535 |
"step": 3700
|
| 536 |
},
|
| 537 |
{
|
| 538 |
-
"epoch": 0.
|
| 539 |
-
"grad_norm":
|
| 540 |
-
"learning_rate":
|
| 541 |
-
"loss":
|
| 542 |
"step": 3750
|
| 543 |
},
|
| 544 |
{
|
| 545 |
-
"epoch": 0.
|
| 546 |
-
"grad_norm": 22.
|
| 547 |
-
"learning_rate":
|
| 548 |
-
"loss":
|
| 549 |
"step": 3800
|
| 550 |
},
|
| 551 |
{
|
| 552 |
-
"epoch": 0.
|
| 553 |
-
"grad_norm":
|
| 554 |
-
"learning_rate":
|
| 555 |
-
"loss":
|
| 556 |
"step": 3850
|
| 557 |
},
|
| 558 |
{
|
| 559 |
-
"epoch": 0.
|
| 560 |
-
"grad_norm":
|
| 561 |
-
"learning_rate":
|
| 562 |
-
"loss":
|
| 563 |
"step": 3900
|
| 564 |
},
|
| 565 |
{
|
| 566 |
-
"epoch": 0.
|
| 567 |
-
"grad_norm":
|
| 568 |
-
"learning_rate":
|
| 569 |
-
"loss":
|
| 570 |
"step": 3950
|
| 571 |
},
|
| 572 |
{
|
| 573 |
-
"epoch": 0.
|
| 574 |
-
"grad_norm":
|
| 575 |
-
"learning_rate":
|
| 576 |
-
"loss":
|
| 577 |
"step": 4000
|
| 578 |
},
|
| 579 |
{
|
| 580 |
-
"epoch": 0.
|
| 581 |
-
"grad_norm":
|
| 582 |
-
"learning_rate":
|
| 583 |
-
"loss":
|
| 584 |
"step": 4050
|
| 585 |
},
|
| 586 |
{
|
| 587 |
-
"epoch": 0.
|
| 588 |
-
"grad_norm":
|
| 589 |
-
"learning_rate":
|
| 590 |
-
"loss":
|
| 591 |
"step": 4100
|
| 592 |
},
|
| 593 |
{
|
| 594 |
-
"epoch": 0.
|
| 595 |
-
"grad_norm": 21.
|
| 596 |
-
"learning_rate":
|
| 597 |
-
"loss":
|
| 598 |
"step": 4150
|
| 599 |
},
|
| 600 |
{
|
| 601 |
-
"epoch": 0.
|
| 602 |
-
"grad_norm":
|
| 603 |
-
"learning_rate":
|
| 604 |
-
"loss":
|
| 605 |
"step": 4200
|
| 606 |
},
|
| 607 |
{
|
| 608 |
-
"epoch": 0.
|
| 609 |
-
"grad_norm":
|
| 610 |
-
"learning_rate":
|
| 611 |
-
"loss":
|
| 612 |
"step": 4250
|
| 613 |
},
|
| 614 |
{
|
| 615 |
-
"epoch": 0.
|
| 616 |
-
"grad_norm": 22.
|
| 617 |
-
"learning_rate":
|
| 618 |
-
"loss":
|
| 619 |
"step": 4300
|
| 620 |
},
|
| 621 |
{
|
| 622 |
-
"epoch": 0.
|
| 623 |
-
"grad_norm":
|
| 624 |
-
"learning_rate":
|
| 625 |
-
"loss":
|
| 626 |
"step": 4350
|
| 627 |
},
|
| 628 |
{
|
| 629 |
-
"epoch": 0.
|
| 630 |
-
"grad_norm":
|
| 631 |
-
"learning_rate":
|
| 632 |
-
"loss":
|
| 633 |
"step": 4400
|
| 634 |
},
|
| 635 |
{
|
| 636 |
-
"epoch": 0.
|
| 637 |
-
"grad_norm":
|
| 638 |
-
"learning_rate":
|
| 639 |
-
"loss":
|
| 640 |
"step": 4450
|
| 641 |
},
|
| 642 |
{
|
| 643 |
-
"epoch": 0.
|
| 644 |
-
"grad_norm":
|
| 645 |
-
"learning_rate":
|
| 646 |
-
"loss":
|
| 647 |
"step": 4500
|
| 648 |
},
|
| 649 |
{
|
| 650 |
-
"epoch": 0.
|
| 651 |
-
"grad_norm":
|
| 652 |
-
"learning_rate":
|
| 653 |
-
"loss":
|
| 654 |
"step": 4550
|
| 655 |
},
|
| 656 |
{
|
| 657 |
-
"epoch": 0.
|
| 658 |
-
"grad_norm":
|
| 659 |
-
"learning_rate":
|
| 660 |
-
"loss":
|
| 661 |
"step": 4600
|
| 662 |
},
|
| 663 |
{
|
| 664 |
-
"epoch": 0.
|
| 665 |
-
"grad_norm":
|
| 666 |
-
"learning_rate":
|
| 667 |
-
"loss":
|
| 668 |
"step": 4650
|
| 669 |
},
|
| 670 |
{
|
| 671 |
-
"epoch": 0.
|
| 672 |
-
"grad_norm":
|
| 673 |
-
"learning_rate":
|
| 674 |
-
"loss":
|
| 675 |
"step": 4700
|
| 676 |
},
|
| 677 |
{
|
| 678 |
-
"epoch": 0.
|
| 679 |
-
"grad_norm":
|
| 680 |
-
"learning_rate":
|
| 681 |
-
"loss":
|
| 682 |
"step": 4750
|
| 683 |
},
|
| 684 |
{
|
| 685 |
-
"epoch": 0.
|
| 686 |
-
"grad_norm":
|
| 687 |
-
"learning_rate":
|
| 688 |
-
"loss":
|
| 689 |
"step": 4800
|
| 690 |
},
|
| 691 |
{
|
| 692 |
-
"epoch": 0.
|
| 693 |
-
"grad_norm":
|
| 694 |
-
"learning_rate":
|
| 695 |
-
"loss":
|
| 696 |
"step": 4850
|
| 697 |
},
|
| 698 |
{
|
| 699 |
-
"epoch": 0.
|
| 700 |
-
"grad_norm": 21.
|
| 701 |
-
"learning_rate":
|
| 702 |
-
"loss":
|
| 703 |
"step": 4900
|
| 704 |
},
|
| 705 |
{
|
| 706 |
-
"epoch": 0.
|
| 707 |
-
"grad_norm":
|
| 708 |
-
"learning_rate":
|
| 709 |
-
"loss":
|
| 710 |
"step": 4950
|
| 711 |
},
|
| 712 |
{
|
| 713 |
-
"epoch": 0.
|
| 714 |
-
"grad_norm":
|
| 715 |
-
"learning_rate":
|
| 716 |
-
"loss":
|
| 717 |
"step": 5000
|
| 718 |
-
},
|
| 719 |
-
{
|
| 720 |
-
"epoch": 0.5085087100996879,
|
| 721 |
-
"grad_norm": 22.390625,
|
| 722 |
-
"learning_rate": 7.945623613449344e-08,
|
| 723 |
-
"loss": 20.0789,
|
| 724 |
-
"step": 5050
|
| 725 |
-
},
|
| 726 |
-
{
|
| 727 |
-
"epoch": 0.5135434498036452,
|
| 728 |
-
"grad_norm": 24.96875,
|
| 729 |
-
"learning_rate": 8.024293154176566e-08,
|
| 730 |
-
"loss": 19.9527,
|
| 731 |
-
"step": 5100
|
| 732 |
-
},
|
| 733 |
-
{
|
| 734 |
-
"epoch": 0.5185781895076025,
|
| 735 |
-
"grad_norm": 32.1875,
|
| 736 |
-
"learning_rate": 8.102962694903787e-08,
|
| 737 |
-
"loss": 20.0024,
|
| 738 |
-
"step": 5150
|
| 739 |
-
},
|
| 740 |
-
{
|
| 741 |
-
"epoch": 0.5236129292115598,
|
| 742 |
-
"grad_norm": 28.796875,
|
| 743 |
-
"learning_rate": 8.181632235631009e-08,
|
| 744 |
-
"loss": 19.9463,
|
| 745 |
-
"step": 5200
|
| 746 |
-
},
|
| 747 |
-
{
|
| 748 |
-
"epoch": 0.5286476689155171,
|
| 749 |
-
"grad_norm": 22.859375,
|
| 750 |
-
"learning_rate": 8.260301776358229e-08,
|
| 751 |
-
"loss": 19.9707,
|
| 752 |
-
"step": 5250
|
| 753 |
-
},
|
| 754 |
-
{
|
| 755 |
-
"epoch": 0.5336824086194744,
|
| 756 |
-
"grad_norm": 23.46875,
|
| 757 |
-
"learning_rate": 8.338971317085451e-08,
|
| 758 |
-
"loss": 19.8411,
|
| 759 |
-
"step": 5300
|
| 760 |
-
},
|
| 761 |
-
{
|
| 762 |
-
"epoch": 0.5387171483234316,
|
| 763 |
-
"grad_norm": 25.03125,
|
| 764 |
-
"learning_rate": 8.417640857812672e-08,
|
| 765 |
-
"loss": 19.9652,
|
| 766 |
-
"step": 5350
|
| 767 |
-
},
|
| 768 |
-
{
|
| 769 |
-
"epoch": 0.5437518880273889,
|
| 770 |
-
"grad_norm": 21.65625,
|
| 771 |
-
"learning_rate": 8.496310398539892e-08,
|
| 772 |
-
"loss": 19.7365,
|
| 773 |
-
"step": 5400
|
| 774 |
-
},
|
| 775 |
-
{
|
| 776 |
-
"epoch": 0.5487866277313463,
|
| 777 |
-
"grad_norm": 21.5625,
|
| 778 |
-
"learning_rate": 8.574979939267114e-08,
|
| 779 |
-
"loss": 19.7931,
|
| 780 |
-
"step": 5450
|
| 781 |
-
},
|
| 782 |
-
{
|
| 783 |
-
"epoch": 0.5538213674353036,
|
| 784 |
-
"grad_norm": 21.484375,
|
| 785 |
-
"learning_rate": 8.653649479994335e-08,
|
| 786 |
-
"loss": 19.7417,
|
| 787 |
-
"step": 5500
|
| 788 |
-
},
|
| 789 |
-
{
|
| 790 |
-
"epoch": 0.5588561071392609,
|
| 791 |
-
"grad_norm": 23.546875,
|
| 792 |
-
"learning_rate": 8.732319020721557e-08,
|
| 793 |
-
"loss": 19.6592,
|
| 794 |
-
"step": 5550
|
| 795 |
-
},
|
| 796 |
-
{
|
| 797 |
-
"epoch": 0.5638908468432182,
|
| 798 |
-
"grad_norm": 21.78125,
|
| 799 |
-
"learning_rate": 8.810988561448777e-08,
|
| 800 |
-
"loss": 19.6256,
|
| 801 |
-
"step": 5600
|
| 802 |
-
},
|
| 803 |
-
{
|
| 804 |
-
"epoch": 0.5689255865471755,
|
| 805 |
-
"grad_norm": 24.6875,
|
| 806 |
-
"learning_rate": 8.889658102175999e-08,
|
| 807 |
-
"loss": 19.7998,
|
| 808 |
-
"step": 5650
|
| 809 |
-
},
|
| 810 |
-
{
|
| 811 |
-
"epoch": 0.5739603262511328,
|
| 812 |
-
"grad_norm": 33.0625,
|
| 813 |
-
"learning_rate": 8.96832764290322e-08,
|
| 814 |
-
"loss": 19.6983,
|
| 815 |
-
"step": 5700
|
| 816 |
-
},
|
| 817 |
-
{
|
| 818 |
-
"epoch": 0.5789950659550901,
|
| 819 |
-
"grad_norm": 23.03125,
|
| 820 |
-
"learning_rate": 9.046997183630442e-08,
|
| 821 |
-
"loss": 19.7832,
|
| 822 |
-
"step": 5750
|
| 823 |
-
},
|
| 824 |
-
{
|
| 825 |
-
"epoch": 0.5840298056590474,
|
| 826 |
-
"grad_norm": 24.796875,
|
| 827 |
-
"learning_rate": 9.125666724357662e-08,
|
| 828 |
-
"loss": 19.7117,
|
| 829 |
-
"step": 5800
|
| 830 |
-
},
|
| 831 |
-
{
|
| 832 |
-
"epoch": 0.5890645453630048,
|
| 833 |
-
"grad_norm": 23.109375,
|
| 834 |
-
"learning_rate": 9.204336265084884e-08,
|
| 835 |
-
"loss": 19.726,
|
| 836 |
-
"step": 5850
|
| 837 |
-
},
|
| 838 |
-
{
|
| 839 |
-
"epoch": 0.5940992850669621,
|
| 840 |
-
"grad_norm": 22.484375,
|
| 841 |
-
"learning_rate": 9.283005805812105e-08,
|
| 842 |
-
"loss": 19.6984,
|
| 843 |
-
"step": 5900
|
| 844 |
-
},
|
| 845 |
-
{
|
| 846 |
-
"epoch": 0.5991340247709194,
|
| 847 |
-
"grad_norm": 21.53125,
|
| 848 |
-
"learning_rate": 9.361675346539325e-08,
|
| 849 |
-
"loss": 19.7171,
|
| 850 |
-
"step": 5950
|
| 851 |
-
},
|
| 852 |
-
{
|
| 853 |
-
"epoch": 0.6041687644748767,
|
| 854 |
-
"grad_norm": 23.296875,
|
| 855 |
-
"learning_rate": 9.440344887266547e-08,
|
| 856 |
-
"loss": 19.5465,
|
| 857 |
-
"step": 6000
|
| 858 |
-
},
|
| 859 |
-
{
|
| 860 |
-
"epoch": 0.6041687644748767,
|
| 861 |
-
"eval_loss": 2.448728084564209,
|
| 862 |
-
"eval_runtime": 97.8245,
|
| 863 |
-
"eval_samples_per_second": 2735.644,
|
| 864 |
-
"eval_steps_per_second": 42.75,
|
| 865 |
-
"step": 6000
|
| 866 |
-
},
|
| 867 |
-
{
|
| 868 |
-
"epoch": 0.609203504178834,
|
| 869 |
-
"grad_norm": 24.5625,
|
| 870 |
-
"learning_rate": 9.519014427993768e-08,
|
| 871 |
-
"loss": 19.5921,
|
| 872 |
-
"step": 6050
|
| 873 |
-
},
|
| 874 |
-
{
|
| 875 |
-
"epoch": 0.6142382438827912,
|
| 876 |
-
"grad_norm": 23.28125,
|
| 877 |
-
"learning_rate": 9.59768396872099e-08,
|
| 878 |
-
"loss": 19.5556,
|
| 879 |
-
"step": 6100
|
| 880 |
-
},
|
| 881 |
-
{
|
| 882 |
-
"epoch": 0.6192729835867485,
|
| 883 |
-
"grad_norm": 23.671875,
|
| 884 |
-
"learning_rate": 9.67635350944821e-08,
|
| 885 |
-
"loss": 19.6202,
|
| 886 |
-
"step": 6150
|
| 887 |
-
},
|
| 888 |
-
{
|
| 889 |
-
"epoch": 0.6243077232907058,
|
| 890 |
-
"grad_norm": 26.609375,
|
| 891 |
-
"learning_rate": 9.755023050175432e-08,
|
| 892 |
-
"loss": 19.5439,
|
| 893 |
-
"step": 6200
|
| 894 |
-
},
|
| 895 |
-
{
|
| 896 |
-
"epoch": 0.6293424629946631,
|
| 897 |
-
"grad_norm": 26.3125,
|
| 898 |
-
"learning_rate": 9.833692590902653e-08,
|
| 899 |
-
"loss": 19.4785,
|
| 900 |
-
"step": 6250
|
| 901 |
-
},
|
| 902 |
-
{
|
| 903 |
-
"epoch": 0.6343772026986205,
|
| 904 |
-
"grad_norm": 24.9375,
|
| 905 |
-
"learning_rate": 9.912362131629876e-08,
|
| 906 |
-
"loss": 19.4382,
|
| 907 |
-
"step": 6300
|
| 908 |
-
},
|
| 909 |
-
{
|
| 910 |
-
"epoch": 0.6394119424025778,
|
| 911 |
-
"grad_norm": 25.703125,
|
| 912 |
-
"learning_rate": 9.991031672357097e-08,
|
| 913 |
-
"loss": 19.4496,
|
| 914 |
-
"step": 6350
|
| 915 |
-
},
|
| 916 |
-
{
|
| 917 |
-
"epoch": 0.6444466821065351,
|
| 918 |
-
"grad_norm": 23.046875,
|
| 919 |
-
"learning_rate": 1.0069701213084319e-07,
|
| 920 |
-
"loss": 19.4539,
|
| 921 |
-
"step": 6400
|
| 922 |
-
},
|
| 923 |
-
{
|
| 924 |
-
"epoch": 0.6494814218104924,
|
| 925 |
-
"grad_norm": 23.203125,
|
| 926 |
-
"learning_rate": 1.0148370753811539e-07,
|
| 927 |
-
"loss": 19.4393,
|
| 928 |
-
"step": 6450
|
| 929 |
-
},
|
| 930 |
-
{
|
| 931 |
-
"epoch": 0.6545161615144497,
|
| 932 |
-
"grad_norm": 26.09375,
|
| 933 |
-
"learning_rate": 1.022704029453876e-07,
|
| 934 |
-
"loss": 19.3709,
|
| 935 |
-
"step": 6500
|
| 936 |
-
},
|
| 937 |
-
{
|
| 938 |
-
"epoch": 0.659550901218407,
|
| 939 |
-
"grad_norm": 23.46875,
|
| 940 |
-
"learning_rate": 1.0305709835265982e-07,
|
| 941 |
-
"loss": 19.4344,
|
| 942 |
-
"step": 6550
|
| 943 |
-
},
|
| 944 |
-
{
|
| 945 |
-
"epoch": 0.6645856409223643,
|
| 946 |
-
"grad_norm": 24.84375,
|
| 947 |
-
"learning_rate": 1.0384379375993202e-07,
|
| 948 |
-
"loss": 19.2581,
|
| 949 |
-
"step": 6600
|
| 950 |
-
},
|
| 951 |
-
{
|
| 952 |
-
"epoch": 0.6696203806263216,
|
| 953 |
-
"grad_norm": 24.0,
|
| 954 |
-
"learning_rate": 1.0463048916720424e-07,
|
| 955 |
-
"loss": 19.3535,
|
| 956 |
-
"step": 6650
|
| 957 |
-
},
|
| 958 |
-
{
|
| 959 |
-
"epoch": 0.674655120330279,
|
| 960 |
-
"grad_norm": 23.890625,
|
| 961 |
-
"learning_rate": 1.0541718457447645e-07,
|
| 962 |
-
"loss": 19.1846,
|
| 963 |
-
"step": 6700
|
| 964 |
-
},
|
| 965 |
-
{
|
| 966 |
-
"epoch": 0.6796898600342363,
|
| 967 |
-
"grad_norm": 22.875,
|
| 968 |
-
"learning_rate": 1.0620387998174867e-07,
|
| 969 |
-
"loss": 19.234,
|
| 970 |
-
"step": 6750
|
| 971 |
-
},
|
| 972 |
-
{
|
| 973 |
-
"epoch": 0.6847245997381936,
|
| 974 |
-
"grad_norm": 23.65625,
|
| 975 |
-
"learning_rate": 1.0699057538902087e-07,
|
| 976 |
-
"loss": 19.1321,
|
| 977 |
-
"step": 6800
|
| 978 |
-
},
|
| 979 |
-
{
|
| 980 |
-
"epoch": 0.6897593394421508,
|
| 981 |
-
"grad_norm": 21.328125,
|
| 982 |
-
"learning_rate": 1.0777727079629309e-07,
|
| 983 |
-
"loss": 19.3041,
|
| 984 |
-
"step": 6850
|
| 985 |
-
},
|
| 986 |
-
{
|
| 987 |
-
"epoch": 0.6947940791461081,
|
| 988 |
-
"grad_norm": 30.234375,
|
| 989 |
-
"learning_rate": 1.085639662035653e-07,
|
| 990 |
-
"loss": 19.2101,
|
| 991 |
-
"step": 6900
|
| 992 |
-
},
|
| 993 |
-
{
|
| 994 |
-
"epoch": 0.6998288188500654,
|
| 995 |
-
"grad_norm": 22.609375,
|
| 996 |
-
"learning_rate": 1.0935066161083752e-07,
|
| 997 |
-
"loss": 19.12,
|
| 998 |
-
"step": 6950
|
| 999 |
-
},
|
| 1000 |
-
{
|
| 1001 |
-
"epoch": 0.7048635585540227,
|
| 1002 |
-
"grad_norm": 22.921875,
|
| 1003 |
-
"learning_rate": 1.1013735701810972e-07,
|
| 1004 |
-
"loss": 19.211,
|
| 1005 |
-
"step": 7000
|
| 1006 |
-
},
|
| 1007 |
-
{
|
| 1008 |
-
"epoch": 0.70989829825798,
|
| 1009 |
-
"grad_norm": 27.5,
|
| 1010 |
-
"learning_rate": 1.1092405242538194e-07,
|
| 1011 |
-
"loss": 19.0585,
|
| 1012 |
-
"step": 7050
|
| 1013 |
-
},
|
| 1014 |
-
{
|
| 1015 |
-
"epoch": 0.7149330379619374,
|
| 1016 |
-
"grad_norm": 26.515625,
|
| 1017 |
-
"learning_rate": 1.1171074783265415e-07,
|
| 1018 |
-
"loss": 19.2151,
|
| 1019 |
-
"step": 7100
|
| 1020 |
-
},
|
| 1021 |
-
{
|
| 1022 |
-
"epoch": 0.7199677776658947,
|
| 1023 |
-
"grad_norm": 22.734375,
|
| 1024 |
-
"learning_rate": 1.1249744323992635e-07,
|
| 1025 |
-
"loss": 19.0144,
|
| 1026 |
-
"step": 7150
|
| 1027 |
-
},
|
| 1028 |
-
{
|
| 1029 |
-
"epoch": 0.725002517369852,
|
| 1030 |
-
"grad_norm": 23.359375,
|
| 1031 |
-
"learning_rate": 1.1328413864719857e-07,
|
| 1032 |
-
"loss": 19.1019,
|
| 1033 |
-
"step": 7200
|
| 1034 |
-
},
|
| 1035 |
-
{
|
| 1036 |
-
"epoch": 0.7300372570738093,
|
| 1037 |
-
"grad_norm": 28.09375,
|
| 1038 |
-
"learning_rate": 1.1407083405447078e-07,
|
| 1039 |
-
"loss": 19.1398,
|
| 1040 |
-
"step": 7250
|
| 1041 |
-
},
|
| 1042 |
-
{
|
| 1043 |
-
"epoch": 0.7350719967777666,
|
| 1044 |
-
"grad_norm": 24.09375,
|
| 1045 |
-
"learning_rate": 1.14857529461743e-07,
|
| 1046 |
-
"loss": 19.1944,
|
| 1047 |
-
"step": 7300
|
| 1048 |
-
},
|
| 1049 |
-
{
|
| 1050 |
-
"epoch": 0.7401067364817239,
|
| 1051 |
-
"grad_norm": 22.625,
|
| 1052 |
-
"learning_rate": 1.156442248690152e-07,
|
| 1053 |
-
"loss": 19.0951,
|
| 1054 |
-
"step": 7350
|
| 1055 |
-
},
|
| 1056 |
-
{
|
| 1057 |
-
"epoch": 0.7451414761856812,
|
| 1058 |
-
"grad_norm": 24.8125,
|
| 1059 |
-
"learning_rate": 1.1643092027628742e-07,
|
| 1060 |
-
"loss": 19.0727,
|
| 1061 |
-
"step": 7400
|
| 1062 |
-
},
|
| 1063 |
-
{
|
| 1064 |
-
"epoch": 0.7501762158896385,
|
| 1065 |
-
"grad_norm": 24.375,
|
| 1066 |
-
"learning_rate": 1.1721761568355963e-07,
|
| 1067 |
-
"loss": 19.1014,
|
| 1068 |
-
"step": 7450
|
| 1069 |
-
},
|
| 1070 |
-
{
|
| 1071 |
-
"epoch": 0.7552109555935959,
|
| 1072 |
-
"grad_norm": 22.4375,
|
| 1073 |
-
"learning_rate": 1.1800431109083185e-07,
|
| 1074 |
-
"loss": 18.9976,
|
| 1075 |
-
"step": 7500
|
| 1076 |
-
},
|
| 1077 |
-
{
|
| 1078 |
-
"epoch": 0.7602456952975531,
|
| 1079 |
-
"grad_norm": 23.359375,
|
| 1080 |
-
"learning_rate": 1.1879100649810405e-07,
|
| 1081 |
-
"loss": 19.0317,
|
| 1082 |
-
"step": 7550
|
| 1083 |
-
},
|
| 1084 |
-
{
|
| 1085 |
-
"epoch": 0.7652804350015104,
|
| 1086 |
-
"grad_norm": 22.59375,
|
| 1087 |
-
"learning_rate": 1.1957770190537627e-07,
|
| 1088 |
-
"loss": 19.0108,
|
| 1089 |
-
"step": 7600
|
| 1090 |
-
},
|
| 1091 |
-
{
|
| 1092 |
-
"epoch": 0.7703151747054677,
|
| 1093 |
-
"grad_norm": 28.8125,
|
| 1094 |
-
"learning_rate": 1.2036439731264848e-07,
|
| 1095 |
-
"loss": 18.9483,
|
| 1096 |
-
"step": 7650
|
| 1097 |
-
},
|
| 1098 |
-
{
|
| 1099 |
-
"epoch": 0.775349914409425,
|
| 1100 |
-
"grad_norm": 25.171875,
|
| 1101 |
-
"learning_rate": 1.2115109271992069e-07,
|
| 1102 |
-
"loss": 19.0207,
|
| 1103 |
-
"step": 7700
|
| 1104 |
-
},
|
| 1105 |
-
{
|
| 1106 |
-
"epoch": 0.7803846541133823,
|
| 1107 |
-
"grad_norm": 24.015625,
|
| 1108 |
-
"learning_rate": 1.2193778812719292e-07,
|
| 1109 |
-
"loss": 18.9183,
|
| 1110 |
-
"step": 7750
|
| 1111 |
-
},
|
| 1112 |
-
{
|
| 1113 |
-
"epoch": 0.7854193938173396,
|
| 1114 |
-
"grad_norm": 21.953125,
|
| 1115 |
-
"learning_rate": 1.2272448353446512e-07,
|
| 1116 |
-
"loss": 18.8943,
|
| 1117 |
-
"step": 7800
|
| 1118 |
-
},
|
| 1119 |
-
{
|
| 1120 |
-
"epoch": 0.7904541335212969,
|
| 1121 |
-
"grad_norm": 23.9375,
|
| 1122 |
-
"learning_rate": 1.2351117894173733e-07,
|
| 1123 |
-
"loss": 18.877,
|
| 1124 |
-
"step": 7850
|
| 1125 |
-
},
|
| 1126 |
-
{
|
| 1127 |
-
"epoch": 0.7954888732252543,
|
| 1128 |
-
"grad_norm": 23.796875,
|
| 1129 |
-
"learning_rate": 1.2429787434900954e-07,
|
| 1130 |
-
"loss": 18.7945,
|
| 1131 |
-
"step": 7900
|
| 1132 |
-
},
|
| 1133 |
-
{
|
| 1134 |
-
"epoch": 0.8005236129292116,
|
| 1135 |
-
"grad_norm": 23.484375,
|
| 1136 |
-
"learning_rate": 1.2508456975628177e-07,
|
| 1137 |
-
"loss": 18.8354,
|
| 1138 |
-
"step": 7950
|
| 1139 |
-
},
|
| 1140 |
-
{
|
| 1141 |
-
"epoch": 0.8055583526331689,
|
| 1142 |
-
"grad_norm": 23.84375,
|
| 1143 |
-
"learning_rate": 1.2587126516355397e-07,
|
| 1144 |
-
"loss": 18.9123,
|
| 1145 |
-
"step": 8000
|
| 1146 |
-
},
|
| 1147 |
-
{
|
| 1148 |
-
"epoch": 0.8105930923371262,
|
| 1149 |
-
"grad_norm": 22.8125,
|
| 1150 |
-
"learning_rate": 1.2665796057082618e-07,
|
| 1151 |
-
"loss": 18.7323,
|
| 1152 |
-
"step": 8050
|
| 1153 |
-
},
|
| 1154 |
-
{
|
| 1155 |
-
"epoch": 0.8156278320410835,
|
| 1156 |
-
"grad_norm": 22.4375,
|
| 1157 |
-
"learning_rate": 1.2744465597809839e-07,
|
| 1158 |
-
"loss": 18.7036,
|
| 1159 |
-
"step": 8100
|
| 1160 |
-
},
|
| 1161 |
-
{
|
| 1162 |
-
"epoch": 0.8206625717450408,
|
| 1163 |
-
"grad_norm": 23.953125,
|
| 1164 |
-
"learning_rate": 1.2823135138537062e-07,
|
| 1165 |
-
"loss": 18.7942,
|
| 1166 |
-
"step": 8150
|
| 1167 |
-
},
|
| 1168 |
-
{
|
| 1169 |
-
"epoch": 0.825697311448998,
|
| 1170 |
-
"grad_norm": 23.03125,
|
| 1171 |
-
"learning_rate": 1.2901804679264282e-07,
|
| 1172 |
-
"loss": 18.7461,
|
| 1173 |
-
"step": 8200
|
| 1174 |
-
},
|
| 1175 |
-
{
|
| 1176 |
-
"epoch": 0.8307320511529553,
|
| 1177 |
-
"grad_norm": 24.8125,
|
| 1178 |
-
"learning_rate": 1.2980474219991503e-07,
|
| 1179 |
-
"loss": 18.8245,
|
| 1180 |
-
"step": 8250
|
| 1181 |
-
},
|
| 1182 |
-
{
|
| 1183 |
-
"epoch": 0.8357667908569127,
|
| 1184 |
-
"grad_norm": 22.453125,
|
| 1185 |
-
"learning_rate": 1.3059143760718724e-07,
|
| 1186 |
-
"loss": 18.7245,
|
| 1187 |
-
"step": 8300
|
| 1188 |
-
},
|
| 1189 |
-
{
|
| 1190 |
-
"epoch": 0.84080153056087,
|
| 1191 |
-
"grad_norm": 23.609375,
|
| 1192 |
-
"learning_rate": 1.3137813301445947e-07,
|
| 1193 |
-
"loss": 18.7654,
|
| 1194 |
-
"step": 8350
|
| 1195 |
-
},
|
| 1196 |
-
{
|
| 1197 |
-
"epoch": 0.8458362702648273,
|
| 1198 |
-
"grad_norm": 22.078125,
|
| 1199 |
-
"learning_rate": 1.3216482842173165e-07,
|
| 1200 |
-
"loss": 18.6009,
|
| 1201 |
-
"step": 8400
|
| 1202 |
-
},
|
| 1203 |
-
{
|
| 1204 |
-
"epoch": 0.8508710099687846,
|
| 1205 |
-
"grad_norm": 23.015625,
|
| 1206 |
-
"learning_rate": 1.3295152382900388e-07,
|
| 1207 |
-
"loss": 18.7458,
|
| 1208 |
-
"step": 8450
|
| 1209 |
-
},
|
| 1210 |
-
{
|
| 1211 |
-
"epoch": 0.8559057496727419,
|
| 1212 |
-
"grad_norm": 24.578125,
|
| 1213 |
-
"learning_rate": 1.3373821923627609e-07,
|
| 1214 |
-
"loss": 18.7133,
|
| 1215 |
-
"step": 8500
|
| 1216 |
-
},
|
| 1217 |
-
{
|
| 1218 |
-
"epoch": 0.8609404893766992,
|
| 1219 |
-
"grad_norm": 20.953125,
|
| 1220 |
-
"learning_rate": 1.3452491464354832e-07,
|
| 1221 |
-
"loss": 18.6203,
|
| 1222 |
-
"step": 8550
|
| 1223 |
-
},
|
| 1224 |
-
{
|
| 1225 |
-
"epoch": 0.8659752290806565,
|
| 1226 |
-
"grad_norm": 28.0,
|
| 1227 |
-
"learning_rate": 1.353116100508205e-07,
|
| 1228 |
-
"loss": 18.6754,
|
| 1229 |
-
"step": 8600
|
| 1230 |
-
},
|
| 1231 |
-
{
|
| 1232 |
-
"epoch": 0.8710099687846138,
|
| 1233 |
-
"grad_norm": 23.390625,
|
| 1234 |
-
"learning_rate": 1.3609830545809273e-07,
|
| 1235 |
-
"loss": 18.5753,
|
| 1236 |
-
"step": 8650
|
| 1237 |
-
},
|
| 1238 |
-
{
|
| 1239 |
-
"epoch": 0.8760447084885712,
|
| 1240 |
-
"grad_norm": 21.296875,
|
| 1241 |
-
"learning_rate": 1.3688500086536494e-07,
|
| 1242 |
-
"loss": 18.5705,
|
| 1243 |
-
"step": 8700
|
| 1244 |
-
},
|
| 1245 |
-
{
|
| 1246 |
-
"epoch": 0.8810794481925285,
|
| 1247 |
-
"grad_norm": 22.875,
|
| 1248 |
-
"learning_rate": 1.3767169627263717e-07,
|
| 1249 |
-
"loss": 18.5268,
|
| 1250 |
-
"step": 8750
|
| 1251 |
-
},
|
| 1252 |
-
{
|
| 1253 |
-
"epoch": 0.8861141878964858,
|
| 1254 |
-
"grad_norm": 25.90625,
|
| 1255 |
-
"learning_rate": 1.3845839167990935e-07,
|
| 1256 |
-
"loss": 18.594,
|
| 1257 |
-
"step": 8800
|
| 1258 |
-
},
|
| 1259 |
-
{
|
| 1260 |
-
"epoch": 0.8911489276004431,
|
| 1261 |
-
"grad_norm": 24.171875,
|
| 1262 |
-
"learning_rate": 1.3924508708718158e-07,
|
| 1263 |
-
"loss": 18.4483,
|
| 1264 |
-
"step": 8850
|
| 1265 |
-
},
|
| 1266 |
-
{
|
| 1267 |
-
"epoch": 0.8961836673044004,
|
| 1268 |
-
"grad_norm": 24.59375,
|
| 1269 |
-
"learning_rate": 1.400317824944538e-07,
|
| 1270 |
-
"loss": 18.548,
|
| 1271 |
-
"step": 8900
|
| 1272 |
-
},
|
| 1273 |
-
{
|
| 1274 |
-
"epoch": 0.9012184070083576,
|
| 1275 |
-
"grad_norm": 23.34375,
|
| 1276 |
-
"learning_rate": 1.40818477901726e-07,
|
| 1277 |
-
"loss": 18.3815,
|
| 1278 |
-
"step": 8950
|
| 1279 |
-
},
|
| 1280 |
-
{
|
| 1281 |
-
"epoch": 0.9062531467123149,
|
| 1282 |
-
"grad_norm": 21.953125,
|
| 1283 |
-
"learning_rate": 1.4160517330899822e-07,
|
| 1284 |
-
"loss": 18.5225,
|
| 1285 |
-
"step": 9000
|
| 1286 |
-
},
|
| 1287 |
-
{
|
| 1288 |
-
"epoch": 0.9062531467123149,
|
| 1289 |
-
"eval_loss": 2.3123230934143066,
|
| 1290 |
-
"eval_runtime": 96.9463,
|
| 1291 |
-
"eval_samples_per_second": 2760.426,
|
| 1292 |
-
"eval_steps_per_second": 43.137,
|
| 1293 |
-
"step": 9000
|
| 1294 |
-
},
|
| 1295 |
-
{
|
| 1296 |
-
"epoch": 0.9112878864162722,
|
| 1297 |
-
"grad_norm": 23.21875,
|
| 1298 |
-
"learning_rate": 1.4239186871627043e-07,
|
| 1299 |
-
"loss": 18.4194,
|
| 1300 |
-
"step": 9050
|
| 1301 |
-
},
|
| 1302 |
-
{
|
| 1303 |
-
"epoch": 0.9163226261202296,
|
| 1304 |
-
"grad_norm": 22.3125,
|
| 1305 |
-
"learning_rate": 1.4317856412354266e-07,
|
| 1306 |
-
"loss": 18.4873,
|
| 1307 |
-
"step": 9100
|
| 1308 |
-
},
|
| 1309 |
-
{
|
| 1310 |
-
"epoch": 0.9213573658241869,
|
| 1311 |
-
"grad_norm": 21.03125,
|
| 1312 |
-
"learning_rate": 1.4396525953081484e-07,
|
| 1313 |
-
"loss": 18.2763,
|
| 1314 |
-
"step": 9150
|
| 1315 |
-
},
|
| 1316 |
-
{
|
| 1317 |
-
"epoch": 0.9263921055281442,
|
| 1318 |
-
"grad_norm": 22.40625,
|
| 1319 |
-
"learning_rate": 1.4475195493808707e-07,
|
| 1320 |
-
"loss": 18.5129,
|
| 1321 |
-
"step": 9200
|
| 1322 |
-
},
|
| 1323 |
-
{
|
| 1324 |
-
"epoch": 0.9314268452321015,
|
| 1325 |
-
"grad_norm": 20.890625,
|
| 1326 |
-
"learning_rate": 1.4553865034535928e-07,
|
| 1327 |
-
"loss": 18.3831,
|
| 1328 |
-
"step": 9250
|
| 1329 |
-
},
|
| 1330 |
-
{
|
| 1331 |
-
"epoch": 0.9364615849360588,
|
| 1332 |
-
"grad_norm": 27.0625,
|
| 1333 |
-
"learning_rate": 1.463253457526315e-07,
|
| 1334 |
-
"loss": 18.405,
|
| 1335 |
-
"step": 9300
|
| 1336 |
-
},
|
| 1337 |
-
{
|
| 1338 |
-
"epoch": 0.9414963246400161,
|
| 1339 |
-
"grad_norm": 24.390625,
|
| 1340 |
-
"learning_rate": 1.471120411599037e-07,
|
| 1341 |
-
"loss": 18.3264,
|
| 1342 |
-
"step": 9350
|
| 1343 |
-
},
|
| 1344 |
-
{
|
| 1345 |
-
"epoch": 0.9465310643439734,
|
| 1346 |
-
"grad_norm": 21.734375,
|
| 1347 |
-
"learning_rate": 1.4789873656717592e-07,
|
| 1348 |
-
"loss": 18.3415,
|
| 1349 |
-
"step": 9400
|
| 1350 |
-
},
|
| 1351 |
-
{
|
| 1352 |
-
"epoch": 0.9515658040479307,
|
| 1353 |
-
"grad_norm": 28.890625,
|
| 1354 |
-
"learning_rate": 1.4868543197444813e-07,
|
| 1355 |
-
"loss": 18.3986,
|
| 1356 |
-
"step": 9450
|
| 1357 |
-
},
|
| 1358 |
-
{
|
| 1359 |
-
"epoch": 0.9566005437518881,
|
| 1360 |
-
"grad_norm": 22.859375,
|
| 1361 |
-
"learning_rate": 1.4947212738172034e-07,
|
| 1362 |
-
"loss": 18.3425,
|
| 1363 |
-
"step": 9500
|
| 1364 |
-
},
|
| 1365 |
-
{
|
| 1366 |
-
"epoch": 0.9616352834558454,
|
| 1367 |
-
"grad_norm": 24.515625,
|
| 1368 |
-
"learning_rate": 1.5025882278899254e-07,
|
| 1369 |
-
"loss": 18.4341,
|
| 1370 |
-
"step": 9550
|
| 1371 |
-
},
|
| 1372 |
-
{
|
| 1373 |
-
"epoch": 0.9666700231598027,
|
| 1374 |
-
"grad_norm": 21.734375,
|
| 1375 |
-
"learning_rate": 1.5104551819626477e-07,
|
| 1376 |
-
"loss": 18.3468,
|
| 1377 |
-
"step": 9600
|
| 1378 |
-
},
|
| 1379 |
-
{
|
| 1380 |
-
"epoch": 0.97170476286376,
|
| 1381 |
-
"grad_norm": 21.46875,
|
| 1382 |
-
"learning_rate": 1.5183221360353698e-07,
|
| 1383 |
-
"loss": 18.2916,
|
| 1384 |
-
"step": 9650
|
| 1385 |
-
},
|
| 1386 |
-
{
|
| 1387 |
-
"epoch": 0.9767395025677172,
|
| 1388 |
-
"grad_norm": 23.53125,
|
| 1389 |
-
"learning_rate": 1.5261890901080919e-07,
|
| 1390 |
-
"loss": 18.268,
|
| 1391 |
-
"step": 9700
|
| 1392 |
-
},
|
| 1393 |
-
{
|
| 1394 |
-
"epoch": 0.9817742422716745,
|
| 1395 |
-
"grad_norm": 24.78125,
|
| 1396 |
-
"learning_rate": 1.534056044180814e-07,
|
| 1397 |
-
"loss": 18.3326,
|
| 1398 |
-
"step": 9750
|
| 1399 |
-
},
|
| 1400 |
-
{
|
| 1401 |
-
"epoch": 0.9868089819756318,
|
| 1402 |
-
"grad_norm": 25.328125,
|
| 1403 |
-
"learning_rate": 1.5419229982535362e-07,
|
| 1404 |
-
"loss": 18.1188,
|
| 1405 |
-
"step": 9800
|
| 1406 |
-
},
|
| 1407 |
-
{
|
| 1408 |
-
"epoch": 0.9918437216795891,
|
| 1409 |
-
"grad_norm": 23.5625,
|
| 1410 |
-
"learning_rate": 1.5497899523262583e-07,
|
| 1411 |
-
"loss": 18.2302,
|
| 1412 |
-
"step": 9850
|
| 1413 |
-
},
|
| 1414 |
-
{
|
| 1415 |
-
"epoch": 0.9968784613835465,
|
| 1416 |
-
"grad_norm": 22.390625,
|
| 1417 |
-
"learning_rate": 1.5576569063989804e-07,
|
| 1418 |
-
"loss": 18.2507,
|
| 1419 |
-
"step": 9900
|
| 1420 |
}
|
| 1421 |
],
|
| 1422 |
"logging_steps": 50,
|
| 1423 |
-
"max_steps":
|
| 1424 |
"num_input_tokens_seen": 0,
|
| 1425 |
"num_train_epochs": 1,
|
| 1426 |
"save_steps": 5000,
|
|
@@ -1431,12 +729,12 @@
|
|
| 1431 |
"should_evaluate": false,
|
| 1432 |
"should_log": false,
|
| 1433 |
"should_save": true,
|
| 1434 |
-
"should_training_stop":
|
| 1435 |
},
|
| 1436 |
"attributes": {}
|
| 1437 |
}
|
| 1438 |
},
|
| 1439 |
-
"total_flos":
|
| 1440 |
"train_batch_size": 8,
|
| 1441 |
"trial_name": null,
|
| 1442 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.2996793431028799,
|
| 5 |
"eval_steps": 3000,
|
| 6 |
+
"global_step": 5000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
+
"epoch": 0.0029967934310287992,
|
| 13 |
+
"grad_norm": 21.09375,
|
| 14 |
+
"learning_rate": 4.682524817381532e-10,
|
| 15 |
+
"loss": 18.3943,
|
| 16 |
"step": 50
|
| 17 |
},
|
| 18 |
{
|
| 19 |
+
"epoch": 0.0059935868620575984,
|
| 20 |
+
"grad_norm": 20.15625,
|
| 21 |
+
"learning_rate": 9.365049634763063e-10,
|
| 22 |
+
"loss": 18.5436,
|
| 23 |
"step": 100
|
| 24 |
},
|
| 25 |
{
|
| 26 |
+
"epoch": 0.008990380293086398,
|
| 27 |
+
"grad_norm": 20.421875,
|
| 28 |
+
"learning_rate": 1.4047574452144595e-09,
|
| 29 |
+
"loss": 18.5259,
|
| 30 |
"step": 150
|
| 31 |
},
|
| 32 |
{
|
| 33 |
+
"epoch": 0.011987173724115197,
|
| 34 |
+
"grad_norm": 21.171875,
|
| 35 |
+
"learning_rate": 1.8730099269526127e-09,
|
| 36 |
+
"loss": 18.374,
|
| 37 |
"step": 200
|
| 38 |
},
|
| 39 |
{
|
| 40 |
+
"epoch": 0.014983967155143996,
|
| 41 |
+
"grad_norm": 20.640625,
|
| 42 |
+
"learning_rate": 2.341262408690766e-09,
|
| 43 |
+
"loss": 18.3854,
|
| 44 |
"step": 250
|
| 45 |
},
|
| 46 |
{
|
| 47 |
+
"epoch": 0.017980760586172796,
|
| 48 |
+
"grad_norm": 20.0625,
|
| 49 |
+
"learning_rate": 2.809514890428919e-09,
|
| 50 |
+
"loss": 18.3343,
|
| 51 |
"step": 300
|
| 52 |
},
|
| 53 |
{
|
| 54 |
+
"epoch": 0.020977554017201593,
|
| 55 |
+
"grad_norm": 18.984375,
|
| 56 |
+
"learning_rate": 3.277767372167072e-09,
|
| 57 |
+
"loss": 18.3591,
|
| 58 |
"step": 350
|
| 59 |
},
|
| 60 |
{
|
| 61 |
+
"epoch": 0.023974347448230394,
|
| 62 |
+
"grad_norm": 21.90625,
|
| 63 |
+
"learning_rate": 3.746019853905225e-09,
|
| 64 |
+
"loss": 18.2317,
|
| 65 |
"step": 400
|
| 66 |
},
|
| 67 |
{
|
| 68 |
+
"epoch": 0.026971140879259194,
|
| 69 |
+
"grad_norm": 19.484375,
|
| 70 |
+
"learning_rate": 4.214272335643379e-09,
|
| 71 |
+
"loss": 18.2008,
|
| 72 |
"step": 450
|
| 73 |
},
|
| 74 |
{
|
| 75 |
+
"epoch": 0.02996793431028799,
|
| 76 |
+
"grad_norm": 21.765625,
|
| 77 |
+
"learning_rate": 4.682524817381532e-09,
|
| 78 |
+
"loss": 18.2127,
|
| 79 |
"step": 500
|
| 80 |
},
|
| 81 |
{
|
| 82 |
+
"epoch": 0.03296472774131679,
|
| 83 |
+
"grad_norm": 19.15625,
|
| 84 |
+
"learning_rate": 5.150777299119685e-09,
|
| 85 |
+
"loss": 18.2734,
|
| 86 |
"step": 550
|
| 87 |
},
|
| 88 |
{
|
| 89 |
+
"epoch": 0.03596152117234559,
|
| 90 |
+
"grad_norm": 20.171875,
|
| 91 |
+
"learning_rate": 5.619029780857838e-09,
|
| 92 |
+
"loss": 18.2318,
|
| 93 |
"step": 600
|
| 94 |
},
|
| 95 |
{
|
| 96 |
+
"epoch": 0.038958314603374386,
|
| 97 |
+
"grad_norm": 20.90625,
|
| 98 |
+
"learning_rate": 6.087282262595992e-09,
|
| 99 |
+
"loss": 18.179,
|
| 100 |
"step": 650
|
| 101 |
},
|
| 102 |
{
|
| 103 |
+
"epoch": 0.04195510803440319,
|
| 104 |
+
"grad_norm": 21.578125,
|
| 105 |
+
"learning_rate": 6.555534744334144e-09,
|
| 106 |
+
"loss": 18.0297,
|
| 107 |
"step": 700
|
| 108 |
},
|
| 109 |
{
|
| 110 |
+
"epoch": 0.04495190146543199,
|
| 111 |
+
"grad_norm": 20.015625,
|
| 112 |
+
"learning_rate": 7.0237872260722975e-09,
|
| 113 |
+
"loss": 18.0406,
|
| 114 |
"step": 750
|
| 115 |
},
|
| 116 |
{
|
| 117 |
+
"epoch": 0.04794869489646079,
|
| 118 |
+
"grad_norm": 22.359375,
|
| 119 |
+
"learning_rate": 7.49203970781045e-09,
|
| 120 |
+
"loss": 18.1646,
|
| 121 |
"step": 800
|
| 122 |
},
|
| 123 |
{
|
| 124 |
+
"epoch": 0.05094548832748959,
|
| 125 |
+
"grad_norm": 19.578125,
|
| 126 |
+
"learning_rate": 7.960292189548605e-09,
|
| 127 |
+
"loss": 18.065,
|
| 128 |
"step": 850
|
| 129 |
},
|
| 130 |
{
|
| 131 |
+
"epoch": 0.05394228175851839,
|
| 132 |
+
"grad_norm": 20.015625,
|
| 133 |
+
"learning_rate": 8.428544671286759e-09,
|
| 134 |
+
"loss": 18.0043,
|
| 135 |
"step": 900
|
| 136 |
},
|
| 137 |
{
|
| 138 |
+
"epoch": 0.05693907518954718,
|
| 139 |
+
"grad_norm": 18.703125,
|
| 140 |
+
"learning_rate": 8.896797153024911e-09,
|
| 141 |
+
"loss": 17.9912,
|
| 142 |
"step": 950
|
| 143 |
},
|
| 144 |
{
|
| 145 |
+
"epoch": 0.05993586862057598,
|
| 146 |
+
"grad_norm": 19.875,
|
| 147 |
+
"learning_rate": 9.365049634763063e-09,
|
| 148 |
+
"loss": 17.9355,
|
| 149 |
"step": 1000
|
| 150 |
},
|
| 151 |
{
|
| 152 |
+
"epoch": 0.06293266205160478,
|
| 153 |
+
"grad_norm": 19.65625,
|
| 154 |
+
"learning_rate": 9.833302116501217e-09,
|
| 155 |
+
"loss": 17.9403,
|
| 156 |
"step": 1050
|
| 157 |
},
|
| 158 |
{
|
| 159 |
+
"epoch": 0.06592945548263358,
|
| 160 |
+
"grad_norm": 20.71875,
|
| 161 |
+
"learning_rate": 1.030155459823937e-08,
|
| 162 |
+
"loss": 18.0746,
|
| 163 |
"step": 1100
|
| 164 |
},
|
| 165 |
{
|
| 166 |
+
"epoch": 0.06892624891366238,
|
| 167 |
+
"grad_norm": 21.1875,
|
| 168 |
+
"learning_rate": 1.0769807079977524e-08,
|
| 169 |
+
"loss": 18.0771,
|
| 170 |
"step": 1150
|
| 171 |
},
|
| 172 |
{
|
| 173 |
+
"epoch": 0.07192304234469118,
|
| 174 |
+
"grad_norm": 20.546875,
|
| 175 |
+
"learning_rate": 1.1238059561715676e-08,
|
| 176 |
+
"loss": 17.9541,
|
| 177 |
"step": 1200
|
| 178 |
},
|
| 179 |
{
|
| 180 |
+
"epoch": 0.07491983577571998,
|
| 181 |
+
"grad_norm": 20.0,
|
| 182 |
+
"learning_rate": 1.170631204345383e-08,
|
| 183 |
+
"loss": 17.9193,
|
| 184 |
"step": 1250
|
| 185 |
},
|
| 186 |
{
|
| 187 |
+
"epoch": 0.07791662920674877,
|
| 188 |
+
"grad_norm": 20.6875,
|
| 189 |
+
"learning_rate": 1.2174564525191984e-08,
|
| 190 |
+
"loss": 17.8958,
|
| 191 |
"step": 1300
|
| 192 |
},
|
| 193 |
{
|
| 194 |
+
"epoch": 0.08091342263777758,
|
| 195 |
+
"grad_norm": 17.984375,
|
| 196 |
+
"learning_rate": 1.2642817006930135e-08,
|
| 197 |
+
"loss": 17.8691,
|
| 198 |
"step": 1350
|
| 199 |
},
|
| 200 |
{
|
| 201 |
+
"epoch": 0.08391021606880637,
|
| 202 |
+
"grad_norm": 19.96875,
|
| 203 |
+
"learning_rate": 1.3111069488668289e-08,
|
| 204 |
+
"loss": 17.8992,
|
| 205 |
"step": 1400
|
| 206 |
},
|
| 207 |
{
|
| 208 |
+
"epoch": 0.08690700949983518,
|
| 209 |
"grad_norm": 21.15625,
|
| 210 |
+
"learning_rate": 1.3579321970406443e-08,
|
| 211 |
+
"loss": 17.9409,
|
| 212 |
"step": 1450
|
| 213 |
},
|
| 214 |
{
|
| 215 |
+
"epoch": 0.08990380293086397,
|
| 216 |
+
"grad_norm": 17.890625,
|
| 217 |
+
"learning_rate": 1.4047574452144595e-08,
|
| 218 |
+
"loss": 17.9719,
|
| 219 |
"step": 1500
|
| 220 |
},
|
| 221 |
{
|
| 222 |
+
"epoch": 0.09290059636189278,
|
| 223 |
+
"grad_norm": 24.5,
|
| 224 |
+
"learning_rate": 1.4515826933882749e-08,
|
| 225 |
+
"loss": 17.7969,
|
| 226 |
"step": 1550
|
| 227 |
},
|
| 228 |
{
|
| 229 |
+
"epoch": 0.09589738979292158,
|
| 230 |
+
"grad_norm": 21.21875,
|
| 231 |
+
"learning_rate": 1.49840794156209e-08,
|
| 232 |
+
"loss": 17.7747,
|
| 233 |
"step": 1600
|
| 234 |
},
|
| 235 |
{
|
| 236 |
+
"epoch": 0.09889418322395037,
|
| 237 |
+
"grad_norm": 20.9375,
|
| 238 |
+
"learning_rate": 1.5452331897359057e-08,
|
| 239 |
+
"loss": 17.7449,
|
| 240 |
"step": 1650
|
| 241 |
},
|
| 242 |
{
|
| 243 |
+
"epoch": 0.10189097665497918,
|
| 244 |
+
"grad_norm": 21.75,
|
| 245 |
+
"learning_rate": 1.592058437909721e-08,
|
| 246 |
+
"loss": 17.743,
|
| 247 |
"step": 1700
|
| 248 |
},
|
| 249 |
{
|
| 250 |
+
"epoch": 0.10488777008600797,
|
| 251 |
+
"grad_norm": 21.140625,
|
| 252 |
+
"learning_rate": 1.6388836860835362e-08,
|
| 253 |
+
"loss": 17.7607,
|
| 254 |
"step": 1750
|
| 255 |
},
|
| 256 |
{
|
| 257 |
+
"epoch": 0.10788456351703678,
|
| 258 |
+
"grad_norm": 19.546875,
|
| 259 |
+
"learning_rate": 1.6857089342573517e-08,
|
| 260 |
+
"loss": 17.7645,
|
| 261 |
"step": 1800
|
| 262 |
},
|
| 263 |
{
|
| 264 |
+
"epoch": 0.11088135694806557,
|
| 265 |
+
"grad_norm": 18.90625,
|
| 266 |
+
"learning_rate": 1.7325341824311666e-08,
|
| 267 |
+
"loss": 17.7396,
|
| 268 |
"step": 1850
|
| 269 |
},
|
| 270 |
{
|
| 271 |
+
"epoch": 0.11387815037909436,
|
| 272 |
+
"grad_norm": 19.34375,
|
| 273 |
+
"learning_rate": 1.7793594306049822e-08,
|
| 274 |
+
"loss": 17.6369,
|
| 275 |
"step": 1900
|
| 276 |
},
|
| 277 |
{
|
| 278 |
+
"epoch": 0.11687494381012317,
|
| 279 |
+
"grad_norm": 20.140625,
|
| 280 |
+
"learning_rate": 1.8261846787787974e-08,
|
| 281 |
+
"loss": 17.6624,
|
| 282 |
"step": 1950
|
| 283 |
},
|
| 284 |
{
|
| 285 |
+
"epoch": 0.11987173724115197,
|
| 286 |
+
"grad_norm": 23.890625,
|
| 287 |
+
"learning_rate": 1.8730099269526127e-08,
|
| 288 |
+
"loss": 17.6577,
|
| 289 |
"step": 2000
|
| 290 |
},
|
| 291 |
{
|
| 292 |
+
"epoch": 0.12286853067218077,
|
| 293 |
+
"grad_norm": 18.671875,
|
| 294 |
+
"learning_rate": 1.9198351751264282e-08,
|
| 295 |
+
"loss": 17.6518,
|
| 296 |
"step": 2050
|
| 297 |
},
|
| 298 |
{
|
| 299 |
+
"epoch": 0.12586532410320955,
|
| 300 |
+
"grad_norm": 20.015625,
|
| 301 |
+
"learning_rate": 1.9666604233002435e-08,
|
| 302 |
+
"loss": 17.7459,
|
| 303 |
"step": 2100
|
| 304 |
},
|
| 305 |
{
|
| 306 |
+
"epoch": 0.12886211753423837,
|
| 307 |
+
"grad_norm": 20.640625,
|
| 308 |
+
"learning_rate": 2.0134856714740587e-08,
|
| 309 |
+
"loss": 17.7153,
|
| 310 |
"step": 2150
|
| 311 |
},
|
| 312 |
{
|
| 313 |
+
"epoch": 0.13185891096526717,
|
| 314 |
+
"grad_norm": 20.46875,
|
| 315 |
+
"learning_rate": 2.060310919647874e-08,
|
| 316 |
+
"loss": 17.5973,
|
| 317 |
"step": 2200
|
| 318 |
},
|
| 319 |
{
|
| 320 |
+
"epoch": 0.13485570439629596,
|
| 321 |
+
"grad_norm": 20.765625,
|
| 322 |
+
"learning_rate": 2.1071361678216895e-08,
|
| 323 |
+
"loss": 17.6673,
|
| 324 |
"step": 2250
|
| 325 |
},
|
| 326 |
{
|
| 327 |
+
"epoch": 0.13785249782732475,
|
| 328 |
+
"grad_norm": 22.0,
|
| 329 |
+
"learning_rate": 2.1539614159955047e-08,
|
| 330 |
+
"loss": 17.7575,
|
| 331 |
"step": 2300
|
| 332 |
},
|
| 333 |
{
|
| 334 |
+
"epoch": 0.14084929125835355,
|
| 335 |
+
"grad_norm": 19.015625,
|
| 336 |
+
"learning_rate": 2.20078666416932e-08,
|
| 337 |
+
"loss": 17.6427,
|
| 338 |
"step": 2350
|
| 339 |
},
|
| 340 |
{
|
| 341 |
+
"epoch": 0.14384608468938237,
|
| 342 |
+
"grad_norm": 21.484375,
|
| 343 |
+
"learning_rate": 2.2476119123431352e-08,
|
| 344 |
+
"loss": 17.722,
|
| 345 |
"step": 2400
|
| 346 |
},
|
| 347 |
{
|
| 348 |
+
"epoch": 0.14684287812041116,
|
| 349 |
+
"grad_norm": 21.5,
|
| 350 |
+
"learning_rate": 2.2944371605169504e-08,
|
| 351 |
+
"loss": 17.6005,
|
| 352 |
"step": 2450
|
| 353 |
},
|
| 354 |
{
|
| 355 |
+
"epoch": 0.14983967155143996,
|
| 356 |
+
"grad_norm": 19.859375,
|
| 357 |
+
"learning_rate": 2.341262408690766e-08,
|
| 358 |
+
"loss": 17.4574,
|
| 359 |
"step": 2500
|
| 360 |
},
|
| 361 |
{
|
| 362 |
+
"epoch": 0.15283646498246875,
|
| 363 |
+
"grad_norm": 21.03125,
|
| 364 |
+
"learning_rate": 2.3880876568645812e-08,
|
| 365 |
+
"loss": 17.4844,
|
| 366 |
"step": 2550
|
| 367 |
},
|
| 368 |
{
|
| 369 |
+
"epoch": 0.15583325841349754,
|
| 370 |
+
"grad_norm": 19.828125,
|
| 371 |
+
"learning_rate": 2.4349129050383968e-08,
|
| 372 |
+
"loss": 17.5664,
|
| 373 |
"step": 2600
|
| 374 |
},
|
| 375 |
{
|
| 376 |
+
"epoch": 0.15883005184452637,
|
| 377 |
+
"grad_norm": 22.828125,
|
| 378 |
+
"learning_rate": 2.481738153212212e-08,
|
| 379 |
+
"loss": 17.5743,
|
| 380 |
"step": 2650
|
| 381 |
},
|
| 382 |
{
|
| 383 |
+
"epoch": 0.16182684527555516,
|
| 384 |
+
"grad_norm": 20.984375,
|
| 385 |
+
"learning_rate": 2.528563401386027e-08,
|
| 386 |
+
"loss": 17.551,
|
| 387 |
"step": 2700
|
| 388 |
},
|
| 389 |
{
|
| 390 |
+
"epoch": 0.16482363870658395,
|
| 391 |
+
"grad_norm": 21.546875,
|
| 392 |
+
"learning_rate": 2.5753886495598425e-08,
|
| 393 |
+
"loss": 17.5384,
|
| 394 |
"step": 2750
|
| 395 |
},
|
| 396 |
{
|
| 397 |
+
"epoch": 0.16782043213761275,
|
| 398 |
+
"grad_norm": 22.09375,
|
| 399 |
+
"learning_rate": 2.6222138977336577e-08,
|
| 400 |
+
"loss": 17.5279,
|
| 401 |
"step": 2800
|
| 402 |
},
|
| 403 |
{
|
| 404 |
+
"epoch": 0.17081722556864154,
|
| 405 |
+
"grad_norm": 21.953125,
|
| 406 |
+
"learning_rate": 2.6690391459074733e-08,
|
| 407 |
+
"loss": 17.474,
|
| 408 |
"step": 2850
|
| 409 |
},
|
| 410 |
{
|
| 411 |
+
"epoch": 0.17381401899967036,
|
| 412 |
+
"grad_norm": 18.96875,
|
| 413 |
+
"learning_rate": 2.7158643940812885e-08,
|
| 414 |
+
"loss": 17.5219,
|
| 415 |
"step": 2900
|
| 416 |
},
|
| 417 |
{
|
| 418 |
+
"epoch": 0.17681081243069915,
|
| 419 |
+
"grad_norm": 19.375,
|
| 420 |
+
"learning_rate": 2.762689642255104e-08,
|
| 421 |
+
"loss": 17.4719,
|
| 422 |
"step": 2950
|
| 423 |
},
|
| 424 |
{
|
| 425 |
+
"epoch": 0.17980760586172795,
|
| 426 |
+
"grad_norm": 21.40625,
|
| 427 |
+
"learning_rate": 2.809514890428919e-08,
|
| 428 |
+
"loss": 17.4948,
|
| 429 |
"step": 3000
|
| 430 |
},
|
| 431 |
{
|
| 432 |
+
"epoch": 0.17980760586172795,
|
| 433 |
+
"eval_loss": 2.1850366592407227,
|
| 434 |
+
"eval_runtime": 159.4656,
|
| 435 |
+
"eval_samples_per_second": 2819.436,
|
| 436 |
+
"eval_steps_per_second": 44.06,
|
| 437 |
"step": 3000
|
| 438 |
},
|
| 439 |
{
|
| 440 |
+
"epoch": 0.18280439929275674,
|
| 441 |
+
"grad_norm": 24.109375,
|
| 442 |
+
"learning_rate": 2.8563401386027346e-08,
|
| 443 |
+
"loss": 17.4877,
|
| 444 |
"step": 3050
|
| 445 |
},
|
| 446 |
{
|
| 447 |
+
"epoch": 0.18580119272378556,
|
| 448 |
+
"grad_norm": 20.6875,
|
| 449 |
+
"learning_rate": 2.9031653867765498e-08,
|
| 450 |
+
"loss": 17.3119,
|
| 451 |
"step": 3100
|
| 452 |
},
|
| 453 |
{
|
| 454 |
+
"epoch": 0.18879798615481436,
|
| 455 |
+
"grad_norm": 19.578125,
|
| 456 |
+
"learning_rate": 2.949990634950365e-08,
|
| 457 |
+
"loss": 17.4083,
|
| 458 |
"step": 3150
|
| 459 |
},
|
| 460 |
{
|
| 461 |
+
"epoch": 0.19179477958584315,
|
| 462 |
+
"grad_norm": 20.78125,
|
| 463 |
+
"learning_rate": 2.99681588312418e-08,
|
| 464 |
+
"loss": 17.3361,
|
| 465 |
"step": 3200
|
| 466 |
},
|
| 467 |
{
|
| 468 |
+
"epoch": 0.19479157301687194,
|
| 469 |
+
"grad_norm": 19.859375,
|
| 470 |
+
"learning_rate": 3.0436411312979955e-08,
|
| 471 |
+
"loss": 17.4759,
|
| 472 |
"step": 3250
|
| 473 |
},
|
| 474 |
{
|
| 475 |
+
"epoch": 0.19778836644790074,
|
| 476 |
+
"grad_norm": 20.078125,
|
| 477 |
+
"learning_rate": 3.0904663794718114e-08,
|
| 478 |
+
"loss": 17.3626,
|
| 479 |
"step": 3300
|
| 480 |
},
|
| 481 |
{
|
| 482 |
+
"epoch": 0.20078515987892956,
|
| 483 |
+
"grad_norm": 22.421875,
|
| 484 |
+
"learning_rate": 3.1372916276456266e-08,
|
| 485 |
+
"loss": 17.4349,
|
| 486 |
"step": 3350
|
| 487 |
},
|
| 488 |
{
|
| 489 |
+
"epoch": 0.20378195330995835,
|
| 490 |
+
"grad_norm": 22.21875,
|
| 491 |
+
"learning_rate": 3.184116875819442e-08,
|
| 492 |
+
"loss": 17.4686,
|
| 493 |
"step": 3400
|
| 494 |
},
|
| 495 |
{
|
| 496 |
+
"epoch": 0.20677874674098715,
|
| 497 |
+
"grad_norm": 20.453125,
|
| 498 |
+
"learning_rate": 3.230942123993257e-08,
|
| 499 |
+
"loss": 17.3681,
|
| 500 |
"step": 3450
|
| 501 |
},
|
| 502 |
{
|
| 503 |
+
"epoch": 0.20977554017201594,
|
| 504 |
+
"grad_norm": 19.96875,
|
| 505 |
+
"learning_rate": 3.2777673721670723e-08,
|
| 506 |
+
"loss": 17.4929,
|
| 507 |
"step": 3500
|
| 508 |
},
|
| 509 |
{
|
| 510 |
+
"epoch": 0.21277233360304473,
|
| 511 |
+
"grad_norm": 20.25,
|
| 512 |
+
"learning_rate": 3.3245926203408876e-08,
|
| 513 |
+
"loss": 17.3656,
|
| 514 |
"step": 3550
|
| 515 |
},
|
| 516 |
{
|
| 517 |
+
"epoch": 0.21576912703407355,
|
| 518 |
+
"grad_norm": 20.265625,
|
| 519 |
+
"learning_rate": 3.3714178685147035e-08,
|
| 520 |
+
"loss": 17.3414,
|
| 521 |
"step": 3600
|
| 522 |
},
|
| 523 |
{
|
| 524 |
+
"epoch": 0.21876592046510235,
|
| 525 |
+
"grad_norm": 21.015625,
|
| 526 |
+
"learning_rate": 3.418243116688519e-08,
|
| 527 |
+
"loss": 17.3737,
|
| 528 |
"step": 3650
|
| 529 |
},
|
| 530 |
{
|
| 531 |
+
"epoch": 0.22176271389613114,
|
| 532 |
+
"grad_norm": 20.671875,
|
| 533 |
+
"learning_rate": 3.465068364862333e-08,
|
| 534 |
+
"loss": 17.334,
|
| 535 |
"step": 3700
|
| 536 |
},
|
| 537 |
{
|
| 538 |
+
"epoch": 0.22475950732715994,
|
| 539 |
+
"grad_norm": 23.15625,
|
| 540 |
+
"learning_rate": 3.5118936130361485e-08,
|
| 541 |
+
"loss": 17.445,
|
| 542 |
"step": 3750
|
| 543 |
},
|
| 544 |
{
|
| 545 |
+
"epoch": 0.22775630075818873,
|
| 546 |
+
"grad_norm": 22.046875,
|
| 547 |
+
"learning_rate": 3.5587188612099644e-08,
|
| 548 |
+
"loss": 17.3643,
|
| 549 |
"step": 3800
|
| 550 |
},
|
| 551 |
{
|
| 552 |
+
"epoch": 0.23075309418921755,
|
| 553 |
+
"grad_norm": 21.5,
|
| 554 |
+
"learning_rate": 3.6055441093837797e-08,
|
| 555 |
+
"loss": 17.3034,
|
| 556 |
"step": 3850
|
| 557 |
},
|
| 558 |
{
|
| 559 |
+
"epoch": 0.23374988762024634,
|
| 560 |
+
"grad_norm": 20.75,
|
| 561 |
+
"learning_rate": 3.652369357557595e-08,
|
| 562 |
+
"loss": 17.2343,
|
| 563 |
"step": 3900
|
| 564 |
},
|
| 565 |
{
|
| 566 |
+
"epoch": 0.23674668105127514,
|
| 567 |
+
"grad_norm": 22.140625,
|
| 568 |
+
"learning_rate": 3.69919460573141e-08,
|
| 569 |
+
"loss": 17.322,
|
| 570 |
"step": 3950
|
| 571 |
},
|
| 572 |
{
|
| 573 |
+
"epoch": 0.23974347448230393,
|
| 574 |
+
"grad_norm": 21.6875,
|
| 575 |
+
"learning_rate": 3.7460198539052254e-08,
|
| 576 |
+
"loss": 17.2209,
|
| 577 |
"step": 4000
|
| 578 |
},
|
| 579 |
{
|
| 580 |
+
"epoch": 0.24274026791333272,
|
| 581 |
+
"grad_norm": 20.765625,
|
| 582 |
+
"learning_rate": 3.792845102079041e-08,
|
| 583 |
+
"loss": 17.3421,
|
| 584 |
"step": 4050
|
| 585 |
},
|
| 586 |
{
|
| 587 |
+
"epoch": 0.24573706134436155,
|
| 588 |
+
"grad_norm": 19.3125,
|
| 589 |
+
"learning_rate": 3.8396703502528565e-08,
|
| 590 |
+
"loss": 17.1858,
|
| 591 |
"step": 4100
|
| 592 |
},
|
| 593 |
{
|
| 594 |
+
"epoch": 0.24873385477539034,
|
| 595 |
+
"grad_norm": 21.109375,
|
| 596 |
+
"learning_rate": 3.886495598426672e-08,
|
| 597 |
+
"loss": 17.2446,
|
| 598 |
"step": 4150
|
| 599 |
},
|
| 600 |
{
|
| 601 |
+
"epoch": 0.2517306482064191,
|
| 602 |
+
"grad_norm": 24.671875,
|
| 603 |
+
"learning_rate": 3.933320846600487e-08,
|
| 604 |
+
"loss": 17.3155,
|
| 605 |
"step": 4200
|
| 606 |
},
|
| 607 |
{
|
| 608 |
+
"epoch": 0.2547274416374479,
|
| 609 |
+
"grad_norm": 21.921875,
|
| 610 |
+
"learning_rate": 3.9801460947743015e-08,
|
| 611 |
+
"loss": 17.2362,
|
| 612 |
"step": 4250
|
| 613 |
},
|
| 614 |
{
|
| 615 |
+
"epoch": 0.25772423506847675,
|
| 616 |
+
"grad_norm": 22.1875,
|
| 617 |
+
"learning_rate": 4.0269713429481174e-08,
|
| 618 |
+
"loss": 17.1679,
|
| 619 |
"step": 4300
|
| 620 |
},
|
| 621 |
{
|
| 622 |
+
"epoch": 0.2607210284995055,
|
| 623 |
+
"grad_norm": 20.484375,
|
| 624 |
+
"learning_rate": 4.0737965911219327e-08,
|
| 625 |
+
"loss": 17.0951,
|
| 626 |
"step": 4350
|
| 627 |
},
|
| 628 |
{
|
| 629 |
+
"epoch": 0.26371782193053434,
|
| 630 |
+
"grad_norm": 19.15625,
|
| 631 |
+
"learning_rate": 4.120621839295748e-08,
|
| 632 |
+
"loss": 17.167,
|
| 633 |
"step": 4400
|
| 634 |
},
|
| 635 |
{
|
| 636 |
+
"epoch": 0.2667146153615631,
|
| 637 |
+
"grad_norm": 19.203125,
|
| 638 |
+
"learning_rate": 4.167447087469563e-08,
|
| 639 |
+
"loss": 17.1029,
|
| 640 |
"step": 4450
|
| 641 |
},
|
| 642 |
{
|
| 643 |
+
"epoch": 0.2697114087925919,
|
| 644 |
+
"grad_norm": 20.140625,
|
| 645 |
+
"learning_rate": 4.214272335643379e-08,
|
| 646 |
+
"loss": 17.2224,
|
| 647 |
"step": 4500
|
| 648 |
},
|
| 649 |
{
|
| 650 |
+
"epoch": 0.27270820222362074,
|
| 651 |
+
"grad_norm": 21.28125,
|
| 652 |
+
"learning_rate": 4.261097583817194e-08,
|
| 653 |
+
"loss": 17.1751,
|
| 654 |
"step": 4550
|
| 655 |
},
|
| 656 |
{
|
| 657 |
+
"epoch": 0.2757049956546495,
|
| 658 |
+
"grad_norm": 19.796875,
|
| 659 |
+
"learning_rate": 4.3079228319910095e-08,
|
| 660 |
+
"loss": 17.1403,
|
| 661 |
"step": 4600
|
| 662 |
},
|
| 663 |
{
|
| 664 |
+
"epoch": 0.27870178908567833,
|
| 665 |
+
"grad_norm": 20.59375,
|
| 666 |
+
"learning_rate": 4.354748080164825e-08,
|
| 667 |
+
"loss": 17.1065,
|
| 668 |
"step": 4650
|
| 669 |
},
|
| 670 |
{
|
| 671 |
+
"epoch": 0.2816985825167071,
|
| 672 |
+
"grad_norm": 24.171875,
|
| 673 |
+
"learning_rate": 4.40157332833864e-08,
|
| 674 |
+
"loss": 17.0031,
|
| 675 |
"step": 4700
|
| 676 |
},
|
| 677 |
{
|
| 678 |
+
"epoch": 0.2846953759477359,
|
| 679 |
+
"grad_norm": 20.6875,
|
| 680 |
+
"learning_rate": 4.448398576512456e-08,
|
| 681 |
+
"loss": 17.1772,
|
| 682 |
"step": 4750
|
| 683 |
},
|
| 684 |
{
|
| 685 |
+
"epoch": 0.28769216937876474,
|
| 686 |
+
"grad_norm": 22.0625,
|
| 687 |
+
"learning_rate": 4.4952238246862704e-08,
|
| 688 |
+
"loss": 17.0814,
|
| 689 |
"step": 4800
|
| 690 |
},
|
| 691 |
{
|
| 692 |
+
"epoch": 0.2906889628097935,
|
| 693 |
+
"grad_norm": 19.765625,
|
| 694 |
+
"learning_rate": 4.5420490728600857e-08,
|
| 695 |
+
"loss": 17.1585,
|
| 696 |
"step": 4850
|
| 697 |
},
|
| 698 |
{
|
| 699 |
+
"epoch": 0.2936857562408223,
|
| 700 |
+
"grad_norm": 21.796875,
|
| 701 |
+
"learning_rate": 4.588874321033901e-08,
|
| 702 |
+
"loss": 17.134,
|
| 703 |
"step": 4900
|
| 704 |
},
|
| 705 |
{
|
| 706 |
+
"epoch": 0.2966825496718511,
|
| 707 |
+
"grad_norm": 19.53125,
|
| 708 |
+
"learning_rate": 4.635699569207717e-08,
|
| 709 |
+
"loss": 17.084,
|
| 710 |
"step": 4950
|
| 711 |
},
|
| 712 |
{
|
| 713 |
+
"epoch": 0.2996793431028799,
|
| 714 |
+
"grad_norm": 24.828125,
|
| 715 |
+
"learning_rate": 4.682524817381532e-08,
|
| 716 |
+
"loss": 17.1089,
|
| 717 |
"step": 5000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 718 |
}
|
| 719 |
],
|
| 720 |
"logging_steps": 50,
|
| 721 |
+
"max_steps": 16684,
|
| 722 |
"num_input_tokens_seen": 0,
|
| 723 |
"num_train_epochs": 1,
|
| 724 |
"save_steps": 5000,
|
|
|
|
| 729 |
"should_evaluate": false,
|
| 730 |
"should_log": false,
|
| 731 |
"should_save": true,
|
| 732 |
+
"should_training_stop": false
|
| 733 |
},
|
| 734 |
"attributes": {}
|
| 735 |
}
|
| 736 |
},
|
| 737 |
+
"total_flos": 8.730803076857856e+17,
|
| 738 |
"train_batch_size": 8,
|
| 739 |
"trial_name": null,
|
| 740 |
"trial_params": null
|