Initial commit
Browse files- optimizer.pt +1 -1
- pytorch_model.bin +1 -1
- scheduler.pt +1 -1
- trainer_state.json +483 -3
optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 655348487
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6d55c7235936d0242cf392bec1a2ea0817b12aa635dec8779c8a4eacdb4938ed
|
| 3 |
size 655348487
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 333975623
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:492888ed30a69737bf3290cd99f0992bde40cc710fa370b0ccded98a91f294b8
|
| 3 |
size 333975623
|
scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 623
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f51219b2cb4bc18c5893bf94b3f2834354e493062268fb89b16464b4a442a743
|
| 3 |
size 623
|
trainer_state.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch":
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
@@ -5286,11 +5286,491 @@
|
|
| 5286 |
"learning_rate": 4.3991797676008206e-05,
|
| 5287 |
"loss": 1.9818,
|
| 5288 |
"step": 440000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5289 |
}
|
| 5290 |
],
|
| 5291 |
"max_steps": 3658000,
|
| 5292 |
"num_train_epochs": 1000,
|
| 5293 |
-
"total_flos":
|
| 5294 |
"trial_name": null,
|
| 5295 |
"trial_params": null
|
| 5296 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 131.21924548933845,
|
| 5 |
+
"global_step": 480000,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 5286 |
"learning_rate": 4.3991797676008206e-05,
|
| 5287 |
"loss": 1.9818,
|
| 5288 |
"step": 440000
|
| 5289 |
+
},
|
| 5290 |
+
{
|
| 5291 |
+
"epoch": 120.42,
|
| 5292 |
+
"learning_rate": 4.398496240601504e-05,
|
| 5293 |
+
"loss": 1.9914,
|
| 5294 |
+
"step": 440500
|
| 5295 |
+
},
|
| 5296 |
+
{
|
| 5297 |
+
"epoch": 120.56,
|
| 5298 |
+
"learning_rate": 4.3978127136021876e-05,
|
| 5299 |
+
"loss": 1.9949,
|
| 5300 |
+
"step": 441000
|
| 5301 |
+
},
|
| 5302 |
+
{
|
| 5303 |
+
"epoch": 120.69,
|
| 5304 |
+
"learning_rate": 4.397129186602871e-05,
|
| 5305 |
+
"loss": 2.0031,
|
| 5306 |
+
"step": 441500
|
| 5307 |
+
},
|
| 5308 |
+
{
|
| 5309 |
+
"epoch": 120.83,
|
| 5310 |
+
"learning_rate": 4.3964456596035545e-05,
|
| 5311 |
+
"loss": 2.0048,
|
| 5312 |
+
"step": 442000
|
| 5313 |
+
},
|
| 5314 |
+
{
|
| 5315 |
+
"epoch": 120.97,
|
| 5316 |
+
"learning_rate": 4.395762132604238e-05,
|
| 5317 |
+
"loss": 2.018,
|
| 5318 |
+
"step": 442500
|
| 5319 |
+
},
|
| 5320 |
+
{
|
| 5321 |
+
"epoch": 121.1,
|
| 5322 |
+
"learning_rate": 4.3950786056049215e-05,
|
| 5323 |
+
"loss": 1.9803,
|
| 5324 |
+
"step": 443000
|
| 5325 |
+
},
|
| 5326 |
+
{
|
| 5327 |
+
"epoch": 121.24,
|
| 5328 |
+
"learning_rate": 4.3943950786056056e-05,
|
| 5329 |
+
"loss": 1.9769,
|
| 5330 |
+
"step": 443500
|
| 5331 |
+
},
|
| 5332 |
+
{
|
| 5333 |
+
"epoch": 121.38,
|
| 5334 |
+
"learning_rate": 4.393711551606289e-05,
|
| 5335 |
+
"loss": 1.9786,
|
| 5336 |
+
"step": 444000
|
| 5337 |
+
},
|
| 5338 |
+
{
|
| 5339 |
+
"epoch": 121.51,
|
| 5340 |
+
"learning_rate": 4.393028024606972e-05,
|
| 5341 |
+
"loss": 1.99,
|
| 5342 |
+
"step": 444500
|
| 5343 |
+
},
|
| 5344 |
+
{
|
| 5345 |
+
"epoch": 121.65,
|
| 5346 |
+
"learning_rate": 4.3923444976076554e-05,
|
| 5347 |
+
"loss": 1.9929,
|
| 5348 |
+
"step": 445000
|
| 5349 |
+
},
|
| 5350 |
+
{
|
| 5351 |
+
"epoch": 121.79,
|
| 5352 |
+
"learning_rate": 4.3916609706083396e-05,
|
| 5353 |
+
"loss": 2.0034,
|
| 5354 |
+
"step": 445500
|
| 5355 |
+
},
|
| 5356 |
+
{
|
| 5357 |
+
"epoch": 121.92,
|
| 5358 |
+
"learning_rate": 4.390977443609023e-05,
|
| 5359 |
+
"loss": 2.0121,
|
| 5360 |
+
"step": 446000
|
| 5361 |
+
},
|
| 5362 |
+
{
|
| 5363 |
+
"epoch": 122.06,
|
| 5364 |
+
"learning_rate": 4.3902939166097065e-05,
|
| 5365 |
+
"loss": 1.9899,
|
| 5366 |
+
"step": 446500
|
| 5367 |
+
},
|
| 5368 |
+
{
|
| 5369 |
+
"epoch": 122.2,
|
| 5370 |
+
"learning_rate": 4.389610389610389e-05,
|
| 5371 |
+
"loss": 1.9649,
|
| 5372 |
+
"step": 447000
|
| 5373 |
+
},
|
| 5374 |
+
{
|
| 5375 |
+
"epoch": 122.33,
|
| 5376 |
+
"learning_rate": 4.3889268626110735e-05,
|
| 5377 |
+
"loss": 1.9764,
|
| 5378 |
+
"step": 447500
|
| 5379 |
+
},
|
| 5380 |
+
{
|
| 5381 |
+
"epoch": 122.47,
|
| 5382 |
+
"learning_rate": 4.388243335611757e-05,
|
| 5383 |
+
"loss": 1.9795,
|
| 5384 |
+
"step": 448000
|
| 5385 |
+
},
|
| 5386 |
+
{
|
| 5387 |
+
"epoch": 122.61,
|
| 5388 |
+
"learning_rate": 4.3875598086124404e-05,
|
| 5389 |
+
"loss": 1.9988,
|
| 5390 |
+
"step": 448500
|
| 5391 |
+
},
|
| 5392 |
+
{
|
| 5393 |
+
"epoch": 122.74,
|
| 5394 |
+
"learning_rate": 4.386876281613124e-05,
|
| 5395 |
+
"loss": 1.9975,
|
| 5396 |
+
"step": 449000
|
| 5397 |
+
},
|
| 5398 |
+
{
|
| 5399 |
+
"epoch": 122.88,
|
| 5400 |
+
"learning_rate": 4.3861927546138074e-05,
|
| 5401 |
+
"loss": 2.007,
|
| 5402 |
+
"step": 449500
|
| 5403 |
+
},
|
| 5404 |
+
{
|
| 5405 |
+
"epoch": 123.02,
|
| 5406 |
+
"learning_rate": 4.385509227614491e-05,
|
| 5407 |
+
"loss": 2.0015,
|
| 5408 |
+
"step": 450000
|
| 5409 |
+
},
|
| 5410 |
+
{
|
| 5411 |
+
"epoch": 123.15,
|
| 5412 |
+
"learning_rate": 4.3848257006151743e-05,
|
| 5413 |
+
"loss": 1.9606,
|
| 5414 |
+
"step": 450500
|
| 5415 |
+
},
|
| 5416 |
+
{
|
| 5417 |
+
"epoch": 123.29,
|
| 5418 |
+
"learning_rate": 4.384142173615858e-05,
|
| 5419 |
+
"loss": 1.9646,
|
| 5420 |
+
"step": 451000
|
| 5421 |
+
},
|
| 5422 |
+
{
|
| 5423 |
+
"epoch": 123.43,
|
| 5424 |
+
"learning_rate": 4.383458646616542e-05,
|
| 5425 |
+
"loss": 1.9758,
|
| 5426 |
+
"step": 451500
|
| 5427 |
+
},
|
| 5428 |
+
{
|
| 5429 |
+
"epoch": 123.56,
|
| 5430 |
+
"learning_rate": 4.382775119617225e-05,
|
| 5431 |
+
"loss": 1.9893,
|
| 5432 |
+
"step": 452000
|
| 5433 |
+
},
|
| 5434 |
+
{
|
| 5435 |
+
"epoch": 123.7,
|
| 5436 |
+
"learning_rate": 4.382091592617908e-05,
|
| 5437 |
+
"loss": 1.9943,
|
| 5438 |
+
"step": 452500
|
| 5439 |
+
},
|
| 5440 |
+
{
|
| 5441 |
+
"epoch": 123.84,
|
| 5442 |
+
"learning_rate": 4.3814080656185924e-05,
|
| 5443 |
+
"loss": 1.9997,
|
| 5444 |
+
"step": 453000
|
| 5445 |
+
},
|
| 5446 |
+
{
|
| 5447 |
+
"epoch": 123.97,
|
| 5448 |
+
"learning_rate": 4.380724538619276e-05,
|
| 5449 |
+
"loss": 2.0071,
|
| 5450 |
+
"step": 453500
|
| 5451 |
+
},
|
| 5452 |
+
{
|
| 5453 |
+
"epoch": 124.11,
|
| 5454 |
+
"learning_rate": 4.3800410116199594e-05,
|
| 5455 |
+
"loss": 1.9643,
|
| 5456 |
+
"step": 454000
|
| 5457 |
+
},
|
| 5458 |
+
{
|
| 5459 |
+
"epoch": 124.25,
|
| 5460 |
+
"learning_rate": 4.379357484620643e-05,
|
| 5461 |
+
"loss": 1.9644,
|
| 5462 |
+
"step": 454500
|
| 5463 |
+
},
|
| 5464 |
+
{
|
| 5465 |
+
"epoch": 124.38,
|
| 5466 |
+
"learning_rate": 4.378673957621326e-05,
|
| 5467 |
+
"loss": 1.9718,
|
| 5468 |
+
"step": 455000
|
| 5469 |
+
},
|
| 5470 |
+
{
|
| 5471 |
+
"epoch": 124.52,
|
| 5472 |
+
"learning_rate": 4.37799043062201e-05,
|
| 5473 |
+
"loss": 1.9801,
|
| 5474 |
+
"step": 455500
|
| 5475 |
+
},
|
| 5476 |
+
{
|
| 5477 |
+
"epoch": 124.66,
|
| 5478 |
+
"learning_rate": 4.377306903622693e-05,
|
| 5479 |
+
"loss": 1.9836,
|
| 5480 |
+
"step": 456000
|
| 5481 |
+
},
|
| 5482 |
+
{
|
| 5483 |
+
"epoch": 124.79,
|
| 5484 |
+
"learning_rate": 4.376623376623377e-05,
|
| 5485 |
+
"loss": 1.9961,
|
| 5486 |
+
"step": 456500
|
| 5487 |
+
},
|
| 5488 |
+
{
|
| 5489 |
+
"epoch": 124.93,
|
| 5490 |
+
"learning_rate": 4.37593984962406e-05,
|
| 5491 |
+
"loss": 1.9996,
|
| 5492 |
+
"step": 457000
|
| 5493 |
+
},
|
| 5494 |
+
{
|
| 5495 |
+
"epoch": 125.07,
|
| 5496 |
+
"learning_rate": 4.375256322624744e-05,
|
| 5497 |
+
"loss": 1.9744,
|
| 5498 |
+
"step": 457500
|
| 5499 |
+
},
|
| 5500 |
+
{
|
| 5501 |
+
"epoch": 125.21,
|
| 5502 |
+
"learning_rate": 4.374572795625427e-05,
|
| 5503 |
+
"loss": 1.9558,
|
| 5504 |
+
"step": 458000
|
| 5505 |
+
},
|
| 5506 |
+
{
|
| 5507 |
+
"epoch": 125.34,
|
| 5508 |
+
"learning_rate": 4.373889268626111e-05,
|
| 5509 |
+
"loss": 1.9701,
|
| 5510 |
+
"step": 458500
|
| 5511 |
+
},
|
| 5512 |
+
{
|
| 5513 |
+
"epoch": 125.48,
|
| 5514 |
+
"learning_rate": 4.373205741626795e-05,
|
| 5515 |
+
"loss": 1.971,
|
| 5516 |
+
"step": 459000
|
| 5517 |
+
},
|
| 5518 |
+
{
|
| 5519 |
+
"epoch": 125.62,
|
| 5520 |
+
"learning_rate": 4.372522214627478e-05,
|
| 5521 |
+
"loss": 1.9868,
|
| 5522 |
+
"step": 459500
|
| 5523 |
+
},
|
| 5524 |
+
{
|
| 5525 |
+
"epoch": 125.75,
|
| 5526 |
+
"learning_rate": 4.371838687628161e-05,
|
| 5527 |
+
"loss": 1.9827,
|
| 5528 |
+
"step": 460000
|
| 5529 |
+
},
|
| 5530 |
+
{
|
| 5531 |
+
"epoch": 125.89,
|
| 5532 |
+
"learning_rate": 4.3711551606288446e-05,
|
| 5533 |
+
"loss": 1.9944,
|
| 5534 |
+
"step": 460500
|
| 5535 |
+
},
|
| 5536 |
+
{
|
| 5537 |
+
"epoch": 126.03,
|
| 5538 |
+
"learning_rate": 4.370471633629529e-05,
|
| 5539 |
+
"loss": 1.9864,
|
| 5540 |
+
"step": 461000
|
| 5541 |
+
},
|
| 5542 |
+
{
|
| 5543 |
+
"epoch": 126.16,
|
| 5544 |
+
"learning_rate": 4.369788106630212e-05,
|
| 5545 |
+
"loss": 1.9455,
|
| 5546 |
+
"step": 461500
|
| 5547 |
+
},
|
| 5548 |
+
{
|
| 5549 |
+
"epoch": 126.3,
|
| 5550 |
+
"learning_rate": 4.369104579630896e-05,
|
| 5551 |
+
"loss": 1.9598,
|
| 5552 |
+
"step": 462000
|
| 5553 |
+
},
|
| 5554 |
+
{
|
| 5555 |
+
"epoch": 126.44,
|
| 5556 |
+
"learning_rate": 4.368421052631579e-05,
|
| 5557 |
+
"loss": 1.9695,
|
| 5558 |
+
"step": 462500
|
| 5559 |
+
},
|
| 5560 |
+
{
|
| 5561 |
+
"epoch": 126.57,
|
| 5562 |
+
"learning_rate": 4.367737525632263e-05,
|
| 5563 |
+
"loss": 1.9791,
|
| 5564 |
+
"step": 463000
|
| 5565 |
+
},
|
| 5566 |
+
{
|
| 5567 |
+
"epoch": 126.71,
|
| 5568 |
+
"learning_rate": 4.367053998632946e-05,
|
| 5569 |
+
"loss": 1.9733,
|
| 5570 |
+
"step": 463500
|
| 5571 |
+
},
|
| 5572 |
+
{
|
| 5573 |
+
"epoch": 126.85,
|
| 5574 |
+
"learning_rate": 4.3663704716336296e-05,
|
| 5575 |
+
"loss": 1.9899,
|
| 5576 |
+
"step": 464000
|
| 5577 |
+
},
|
| 5578 |
+
{
|
| 5579 |
+
"epoch": 126.98,
|
| 5580 |
+
"learning_rate": 4.365686944634314e-05,
|
| 5581 |
+
"loss": 1.9975,
|
| 5582 |
+
"step": 464500
|
| 5583 |
+
},
|
| 5584 |
+
{
|
| 5585 |
+
"epoch": 127.12,
|
| 5586 |
+
"learning_rate": 4.3650034176349966e-05,
|
| 5587 |
+
"loss": 1.9522,
|
| 5588 |
+
"step": 465000
|
| 5589 |
+
},
|
| 5590 |
+
{
|
| 5591 |
+
"epoch": 127.26,
|
| 5592 |
+
"learning_rate": 4.36431989063568e-05,
|
| 5593 |
+
"loss": 1.9489,
|
| 5594 |
+
"step": 465500
|
| 5595 |
+
},
|
| 5596 |
+
{
|
| 5597 |
+
"epoch": 127.39,
|
| 5598 |
+
"learning_rate": 4.3636363636363636e-05,
|
| 5599 |
+
"loss": 1.961,
|
| 5600 |
+
"step": 466000
|
| 5601 |
+
},
|
| 5602 |
+
{
|
| 5603 |
+
"epoch": 127.53,
|
| 5604 |
+
"learning_rate": 4.362952836637048e-05,
|
| 5605 |
+
"loss": 1.9685,
|
| 5606 |
+
"step": 466500
|
| 5607 |
+
},
|
| 5608 |
+
{
|
| 5609 |
+
"epoch": 127.67,
|
| 5610 |
+
"learning_rate": 4.362269309637731e-05,
|
| 5611 |
+
"loss": 1.9775,
|
| 5612 |
+
"step": 467000
|
| 5613 |
+
},
|
| 5614 |
+
{
|
| 5615 |
+
"epoch": 127.8,
|
| 5616 |
+
"learning_rate": 4.361585782638414e-05,
|
| 5617 |
+
"loss": 1.983,
|
| 5618 |
+
"step": 467500
|
| 5619 |
+
},
|
| 5620 |
+
{
|
| 5621 |
+
"epoch": 127.94,
|
| 5622 |
+
"learning_rate": 4.3609022556390975e-05,
|
| 5623 |
+
"loss": 1.9919,
|
| 5624 |
+
"step": 468000
|
| 5625 |
+
},
|
| 5626 |
+
{
|
| 5627 |
+
"epoch": 128.08,
|
| 5628 |
+
"learning_rate": 4.3602187286397816e-05,
|
| 5629 |
+
"loss": 1.9609,
|
| 5630 |
+
"step": 468500
|
| 5631 |
+
},
|
| 5632 |
+
{
|
| 5633 |
+
"epoch": 128.21,
|
| 5634 |
+
"learning_rate": 4.359535201640465e-05,
|
| 5635 |
+
"loss": 1.9508,
|
| 5636 |
+
"step": 469000
|
| 5637 |
+
},
|
| 5638 |
+
{
|
| 5639 |
+
"epoch": 128.35,
|
| 5640 |
+
"learning_rate": 4.3588516746411486e-05,
|
| 5641 |
+
"loss": 1.9594,
|
| 5642 |
+
"step": 469500
|
| 5643 |
+
},
|
| 5644 |
+
{
|
| 5645 |
+
"epoch": 128.49,
|
| 5646 |
+
"learning_rate": 4.358168147641832e-05,
|
| 5647 |
+
"loss": 1.9627,
|
| 5648 |
+
"step": 470000
|
| 5649 |
+
},
|
| 5650 |
+
{
|
| 5651 |
+
"epoch": 128.62,
|
| 5652 |
+
"learning_rate": 4.3574846206425156e-05,
|
| 5653 |
+
"loss": 1.9675,
|
| 5654 |
+
"step": 470500
|
| 5655 |
+
},
|
| 5656 |
+
{
|
| 5657 |
+
"epoch": 128.76,
|
| 5658 |
+
"learning_rate": 4.356801093643199e-05,
|
| 5659 |
+
"loss": 1.9777,
|
| 5660 |
+
"step": 471000
|
| 5661 |
+
},
|
| 5662 |
+
{
|
| 5663 |
+
"epoch": 128.9,
|
| 5664 |
+
"learning_rate": 4.3561175666438825e-05,
|
| 5665 |
+
"loss": 1.9811,
|
| 5666 |
+
"step": 471500
|
| 5667 |
+
},
|
| 5668 |
+
{
|
| 5669 |
+
"epoch": 129.03,
|
| 5670 |
+
"learning_rate": 4.355434039644567e-05,
|
| 5671 |
+
"loss": 1.975,
|
| 5672 |
+
"step": 472000
|
| 5673 |
+
},
|
| 5674 |
+
{
|
| 5675 |
+
"epoch": 129.17,
|
| 5676 |
+
"learning_rate": 4.3547505126452495e-05,
|
| 5677 |
+
"loss": 1.9393,
|
| 5678 |
+
"step": 472500
|
| 5679 |
+
},
|
| 5680 |
+
{
|
| 5681 |
+
"epoch": 129.31,
|
| 5682 |
+
"learning_rate": 4.354066985645933e-05,
|
| 5683 |
+
"loss": 1.9465,
|
| 5684 |
+
"step": 473000
|
| 5685 |
+
},
|
| 5686 |
+
{
|
| 5687 |
+
"epoch": 129.44,
|
| 5688 |
+
"learning_rate": 4.3533834586466164e-05,
|
| 5689 |
+
"loss": 1.9567,
|
| 5690 |
+
"step": 473500
|
| 5691 |
+
},
|
| 5692 |
+
{
|
| 5693 |
+
"epoch": 129.58,
|
| 5694 |
+
"learning_rate": 4.3526999316473006e-05,
|
| 5695 |
+
"loss": 1.963,
|
| 5696 |
+
"step": 474000
|
| 5697 |
+
},
|
| 5698 |
+
{
|
| 5699 |
+
"epoch": 129.72,
|
| 5700 |
+
"learning_rate": 4.352016404647984e-05,
|
| 5701 |
+
"loss": 1.9731,
|
| 5702 |
+
"step": 474500
|
| 5703 |
+
},
|
| 5704 |
+
{
|
| 5705 |
+
"epoch": 129.85,
|
| 5706 |
+
"learning_rate": 4.3513328776486675e-05,
|
| 5707 |
+
"loss": 1.9787,
|
| 5708 |
+
"step": 475000
|
| 5709 |
+
},
|
| 5710 |
+
{
|
| 5711 |
+
"epoch": 129.99,
|
| 5712 |
+
"learning_rate": 4.3506493506493503e-05,
|
| 5713 |
+
"loss": 1.9779,
|
| 5714 |
+
"step": 475500
|
| 5715 |
+
},
|
| 5716 |
+
{
|
| 5717 |
+
"epoch": 130.13,
|
| 5718 |
+
"learning_rate": 4.3499658236500345e-05,
|
| 5719 |
+
"loss": 1.9351,
|
| 5720 |
+
"step": 476000
|
| 5721 |
+
},
|
| 5722 |
+
{
|
| 5723 |
+
"epoch": 130.26,
|
| 5724 |
+
"learning_rate": 4.349282296650718e-05,
|
| 5725 |
+
"loss": 1.9439,
|
| 5726 |
+
"step": 476500
|
| 5727 |
+
},
|
| 5728 |
+
{
|
| 5729 |
+
"epoch": 130.4,
|
| 5730 |
+
"learning_rate": 4.3485987696514015e-05,
|
| 5731 |
+
"loss": 1.953,
|
| 5732 |
+
"step": 477000
|
| 5733 |
+
},
|
| 5734 |
+
{
|
| 5735 |
+
"epoch": 130.54,
|
| 5736 |
+
"learning_rate": 4.347915242652085e-05,
|
| 5737 |
+
"loss": 1.9584,
|
| 5738 |
+
"step": 477500
|
| 5739 |
+
},
|
| 5740 |
+
{
|
| 5741 |
+
"epoch": 130.67,
|
| 5742 |
+
"learning_rate": 4.3472317156527684e-05,
|
| 5743 |
+
"loss": 1.9635,
|
| 5744 |
+
"step": 478000
|
| 5745 |
+
},
|
| 5746 |
+
{
|
| 5747 |
+
"epoch": 130.81,
|
| 5748 |
+
"learning_rate": 4.346548188653452e-05,
|
| 5749 |
+
"loss": 1.9722,
|
| 5750 |
+
"step": 478500
|
| 5751 |
+
},
|
| 5752 |
+
{
|
| 5753 |
+
"epoch": 130.95,
|
| 5754 |
+
"learning_rate": 4.3458646616541354e-05,
|
| 5755 |
+
"loss": 1.9806,
|
| 5756 |
+
"step": 479000
|
| 5757 |
+
},
|
| 5758 |
+
{
|
| 5759 |
+
"epoch": 131.08,
|
| 5760 |
+
"learning_rate": 4.345181134654819e-05,
|
| 5761 |
+
"loss": 1.9505,
|
| 5762 |
+
"step": 479500
|
| 5763 |
+
},
|
| 5764 |
+
{
|
| 5765 |
+
"epoch": 131.22,
|
| 5766 |
+
"learning_rate": 4.344497607655503e-05,
|
| 5767 |
+
"loss": 1.9355,
|
| 5768 |
+
"step": 480000
|
| 5769 |
}
|
| 5770 |
],
|
| 5771 |
"max_steps": 3658000,
|
| 5772 |
"num_train_epochs": 1000,
|
| 5773 |
+
"total_flos": 966230618169802752,
|
| 5774 |
"trial_name": null,
|
| 5775 |
"trial_params": null
|
| 5776 |
}
|