Run 8. Outer Step 98. Inner Step 188.
Browse files- config.json +72 -3
- gradients.pt +1 -1
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -264,15 +264,84 @@
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
-
|
| 268 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 269 |
],
|
| 270 |
"bos_token_id": 1,
|
| 271 |
"eos_token_id": 2,
|
| 272 |
"hidden_act": "silu",
|
| 273 |
"hidden_size": 2048,
|
| 274 |
"initializer_range": 0.02,
|
| 275 |
-
"inner_step":
|
| 276 |
"intermediate_size": 5632,
|
| 277 |
"last_allreduce_block": 6422539,
|
| 278 |
"max_position_embeddings": 2048,
|
|
|
|
| 264 |
"attention_bias": false,
|
| 265 |
"attention_dropout": 0.0,
|
| 266 |
"block_list": [
|
| 267 |
+
6422666,
|
| 268 |
+
6422675,
|
| 269 |
+
6422684,
|
| 270 |
+
6422691,
|
| 271 |
+
6422699,
|
| 272 |
+
6422705,
|
| 273 |
+
6422711,
|
| 274 |
+
6422717,
|
| 275 |
+
6422724,
|
| 276 |
+
6422730,
|
| 277 |
+
6422736,
|
| 278 |
+
6422743,
|
| 279 |
+
6422751,
|
| 280 |
+
6422759,
|
| 281 |
+
6422766,
|
| 282 |
+
6422773,
|
| 283 |
+
6422780,
|
| 284 |
+
6422788,
|
| 285 |
+
6422796,
|
| 286 |
+
6422803,
|
| 287 |
+
6422810,
|
| 288 |
+
6422818,
|
| 289 |
+
6422827,
|
| 290 |
+
6422836,
|
| 291 |
+
6422845,
|
| 292 |
+
6422853,
|
| 293 |
+
6422861,
|
| 294 |
+
6422869,
|
| 295 |
+
6422877,
|
| 296 |
+
6422885,
|
| 297 |
+
6422894,
|
| 298 |
+
6422902,
|
| 299 |
+
6422910,
|
| 300 |
+
6422917,
|
| 301 |
+
6422924,
|
| 302 |
+
6422932,
|
| 303 |
+
6422940,
|
| 304 |
+
6422948,
|
| 305 |
+
6422957,
|
| 306 |
+
6422966,
|
| 307 |
+
6422976,
|
| 308 |
+
6422985,
|
| 309 |
+
6422992,
|
| 310 |
+
6423000,
|
| 311 |
+
6423007,
|
| 312 |
+
6423013,
|
| 313 |
+
6423019,
|
| 314 |
+
6423026,
|
| 315 |
+
6423034,
|
| 316 |
+
6423040,
|
| 317 |
+
6423047,
|
| 318 |
+
6423055,
|
| 319 |
+
6423062,
|
| 320 |
+
6423070,
|
| 321 |
+
6423077,
|
| 322 |
+
6423084,
|
| 323 |
+
6423091,
|
| 324 |
+
6423100,
|
| 325 |
+
6423108,
|
| 326 |
+
6423116,
|
| 327 |
+
6423125,
|
| 328 |
+
6423134,
|
| 329 |
+
6423143,
|
| 330 |
+
6423152,
|
| 331 |
+
6423163,
|
| 332 |
+
6423172,
|
| 333 |
+
6423179,
|
| 334 |
+
6423185,
|
| 335 |
+
6423191,
|
| 336 |
+
6423199,
|
| 337 |
+
6423206
|
| 338 |
],
|
| 339 |
"bos_token_id": 1,
|
| 340 |
"eos_token_id": 2,
|
| 341 |
"hidden_act": "silu",
|
| 342 |
"hidden_size": 2048,
|
| 343 |
"initializer_range": 0.02,
|
| 344 |
+
"inner_step": 188,
|
| 345 |
"intermediate_size": 5632,
|
| 346 |
"last_allreduce_block": 6422539,
|
| 347 |
"max_position_embeddings": 2048,
|
gradients.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 88389626
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:67f807a85930b2ca4f314bb58d83016617be8b7c0f4d27242f77b0c9234e8439
|
| 3 |
size 88389626
|
inner_optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4924926906
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5b67a117cc21001d6946e0fd1a710760a1f2eb005d91d6c8e1af957dd637388b
|
| 3 |
size 4924926906
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4400216536
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:200bab4467f5c7c72cf22e49e19c1d5b3fde8d1895574466bcabe0088473727a
|
| 3 |
size 4400216536
|