Training in progress, step 911, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 60010048
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:be633fd5b9a184214787cf8de804cd47d2d08122c6886a59a13dd6075ceb7cb7
|
3 |
size 60010048
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 30428180
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9edb22bd00df21c10ec53c4bbfa2093db527d86ac59bfb6f02154aa19d5da6c1
|
3 |
size 30428180
|
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 sha256:04e2e9566872b589840ca5036c48f962e578ce55c36abdd70f84617113393bbb
|
3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch": 0.
|
5 |
"eval_steps": 500,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -5313,6 +5313,1077 @@
|
|
5313 |
"learning_rate": 3.773119605425401e-06,
|
5314 |
"loss": 1.2571,
|
5315 |
"step": 758
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5316 |
}
|
5317 |
],
|
5318 |
"logging_steps": 1,
|
@@ -5327,12 +6398,12 @@
|
|
5327 |
"should_evaluate": false,
|
5328 |
"should_log": false,
|
5329 |
"should_save": true,
|
5330 |
-
"should_training_stop":
|
5331 |
},
|
5332 |
"attributes": {}
|
5333 |
}
|
5334 |
},
|
5335 |
-
"total_flos":
|
5336 |
"train_batch_size": 4,
|
5337 |
"trial_name": null,
|
5338 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.9997256515775035,
|
5 |
"eval_steps": 500,
|
6 |
+
"global_step": 911,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
5313 |
"learning_rate": 3.773119605425401e-06,
|
5314 |
"loss": 1.2571,
|
5315 |
"step": 758
|
5316 |
+
},
|
5317 |
+
{
|
5318 |
+
"epoch": 0.8329218106995885,
|
5319 |
+
"grad_norm": 0.2670471668243408,
|
5320 |
+
"learning_rate": 3.7484586929716402e-06,
|
5321 |
+
"loss": 1.1585,
|
5322 |
+
"step": 759
|
5323 |
+
},
|
5324 |
+
{
|
5325 |
+
"epoch": 0.8340192043895748,
|
5326 |
+
"grad_norm": 0.2522693872451782,
|
5327 |
+
"learning_rate": 3.723797780517879e-06,
|
5328 |
+
"loss": 1.2441,
|
5329 |
+
"step": 760
|
5330 |
+
},
|
5331 |
+
{
|
5332 |
+
"epoch": 0.8351165980795611,
|
5333 |
+
"grad_norm": 0.26325473189353943,
|
5334 |
+
"learning_rate": 3.699136868064119e-06,
|
5335 |
+
"loss": 1.231,
|
5336 |
+
"step": 761
|
5337 |
+
},
|
5338 |
+
{
|
5339 |
+
"epoch": 0.8362139917695474,
|
5340 |
+
"grad_norm": 0.2654995024204254,
|
5341 |
+
"learning_rate": 3.674475955610358e-06,
|
5342 |
+
"loss": 1.2638,
|
5343 |
+
"step": 762
|
5344 |
+
},
|
5345 |
+
{
|
5346 |
+
"epoch": 0.8373113854595337,
|
5347 |
+
"grad_norm": 0.24272461235523224,
|
5348 |
+
"learning_rate": 3.649815043156597e-06,
|
5349 |
+
"loss": 1.2356,
|
5350 |
+
"step": 763
|
5351 |
+
},
|
5352 |
+
{
|
5353 |
+
"epoch": 0.83840877914952,
|
5354 |
+
"grad_norm": 0.3264164924621582,
|
5355 |
+
"learning_rate": 3.6251541307028365e-06,
|
5356 |
+
"loss": 1.2741,
|
5357 |
+
"step": 764
|
5358 |
+
},
|
5359 |
+
{
|
5360 |
+
"epoch": 0.8395061728395061,
|
5361 |
+
"grad_norm": 0.2254675030708313,
|
5362 |
+
"learning_rate": 3.6004932182490754e-06,
|
5363 |
+
"loss": 1.1681,
|
5364 |
+
"step": 765
|
5365 |
+
},
|
5366 |
+
{
|
5367 |
+
"epoch": 0.8406035665294924,
|
5368 |
+
"grad_norm": 0.25938257575035095,
|
5369 |
+
"learning_rate": 3.5758323057953147e-06,
|
5370 |
+
"loss": 1.1809,
|
5371 |
+
"step": 766
|
5372 |
+
},
|
5373 |
+
{
|
5374 |
+
"epoch": 0.8417009602194787,
|
5375 |
+
"grad_norm": 0.27756911516189575,
|
5376 |
+
"learning_rate": 3.5511713933415536e-06,
|
5377 |
+
"loss": 1.2611,
|
5378 |
+
"step": 767
|
5379 |
+
},
|
5380 |
+
{
|
5381 |
+
"epoch": 0.842798353909465,
|
5382 |
+
"grad_norm": 0.2840558886528015,
|
5383 |
+
"learning_rate": 3.526510480887793e-06,
|
5384 |
+
"loss": 1.2804,
|
5385 |
+
"step": 768
|
5386 |
+
},
|
5387 |
+
{
|
5388 |
+
"epoch": 0.8438957475994513,
|
5389 |
+
"grad_norm": 0.28387823700904846,
|
5390 |
+
"learning_rate": 3.5018495684340327e-06,
|
5391 |
+
"loss": 1.2598,
|
5392 |
+
"step": 769
|
5393 |
+
},
|
5394 |
+
{
|
5395 |
+
"epoch": 0.8449931412894376,
|
5396 |
+
"grad_norm": 0.27318140864372253,
|
5397 |
+
"learning_rate": 3.4771886559802716e-06,
|
5398 |
+
"loss": 1.2617,
|
5399 |
+
"step": 770
|
5400 |
+
},
|
5401 |
+
{
|
5402 |
+
"epoch": 0.8460905349794239,
|
5403 |
+
"grad_norm": 0.24867716431617737,
|
5404 |
+
"learning_rate": 3.452527743526511e-06,
|
5405 |
+
"loss": 1.172,
|
5406 |
+
"step": 771
|
5407 |
+
},
|
5408 |
+
{
|
5409 |
+
"epoch": 0.8471879286694102,
|
5410 |
+
"grad_norm": 0.24067752063274384,
|
5411 |
+
"learning_rate": 3.42786683107275e-06,
|
5412 |
+
"loss": 1.2622,
|
5413 |
+
"step": 772
|
5414 |
+
},
|
5415 |
+
{
|
5416 |
+
"epoch": 0.8482853223593965,
|
5417 |
+
"grad_norm": 0.23819519579410553,
|
5418 |
+
"learning_rate": 3.403205918618989e-06,
|
5419 |
+
"loss": 1.2249,
|
5420 |
+
"step": 773
|
5421 |
+
},
|
5422 |
+
{
|
5423 |
+
"epoch": 0.8493827160493828,
|
5424 |
+
"grad_norm": 0.2725595533847809,
|
5425 |
+
"learning_rate": 3.3785450061652285e-06,
|
5426 |
+
"loss": 1.1911,
|
5427 |
+
"step": 774
|
5428 |
+
},
|
5429 |
+
{
|
5430 |
+
"epoch": 0.850480109739369,
|
5431 |
+
"grad_norm": 0.27524641156196594,
|
5432 |
+
"learning_rate": 3.3538840937114674e-06,
|
5433 |
+
"loss": 1.255,
|
5434 |
+
"step": 775
|
5435 |
+
},
|
5436 |
+
{
|
5437 |
+
"epoch": 0.8515775034293552,
|
5438 |
+
"grad_norm": 0.24099332094192505,
|
5439 |
+
"learning_rate": 3.3292231812577068e-06,
|
5440 |
+
"loss": 1.2023,
|
5441 |
+
"step": 776
|
5442 |
+
},
|
5443 |
+
{
|
5444 |
+
"epoch": 0.8526748971193415,
|
5445 |
+
"grad_norm": 0.2646848261356354,
|
5446 |
+
"learning_rate": 3.3045622688039457e-06,
|
5447 |
+
"loss": 1.197,
|
5448 |
+
"step": 777
|
5449 |
+
},
|
5450 |
+
{
|
5451 |
+
"epoch": 0.8537722908093278,
|
5452 |
+
"grad_norm": 0.24707616865634918,
|
5453 |
+
"learning_rate": 3.2799013563501854e-06,
|
5454 |
+
"loss": 1.1769,
|
5455 |
+
"step": 778
|
5456 |
+
},
|
5457 |
+
{
|
5458 |
+
"epoch": 0.8548696844993141,
|
5459 |
+
"grad_norm": 0.2570493817329407,
|
5460 |
+
"learning_rate": 3.2552404438964248e-06,
|
5461 |
+
"loss": 1.2645,
|
5462 |
+
"step": 779
|
5463 |
+
},
|
5464 |
+
{
|
5465 |
+
"epoch": 0.8559670781893004,
|
5466 |
+
"grad_norm": 0.2723713517189026,
|
5467 |
+
"learning_rate": 3.2305795314426637e-06,
|
5468 |
+
"loss": 1.2125,
|
5469 |
+
"step": 780
|
5470 |
+
},
|
5471 |
+
{
|
5472 |
+
"epoch": 0.8570644718792867,
|
5473 |
+
"grad_norm": 0.23607034981250763,
|
5474 |
+
"learning_rate": 3.205918618988903e-06,
|
5475 |
+
"loss": 1.2958,
|
5476 |
+
"step": 781
|
5477 |
+
},
|
5478 |
+
{
|
5479 |
+
"epoch": 0.858161865569273,
|
5480 |
+
"grad_norm": 0.24119311571121216,
|
5481 |
+
"learning_rate": 3.181257706535142e-06,
|
5482 |
+
"loss": 1.2814,
|
5483 |
+
"step": 782
|
5484 |
+
},
|
5485 |
+
{
|
5486 |
+
"epoch": 0.8592592592592593,
|
5487 |
+
"grad_norm": 0.2970694303512573,
|
5488 |
+
"learning_rate": 3.1565967940813812e-06,
|
5489 |
+
"loss": 1.1962,
|
5490 |
+
"step": 783
|
5491 |
+
},
|
5492 |
+
{
|
5493 |
+
"epoch": 0.8603566529492456,
|
5494 |
+
"grad_norm": 0.24748800694942474,
|
5495 |
+
"learning_rate": 3.1319358816276206e-06,
|
5496 |
+
"loss": 1.2468,
|
5497 |
+
"step": 784
|
5498 |
+
},
|
5499 |
+
{
|
5500 |
+
"epoch": 0.8614540466392319,
|
5501 |
+
"grad_norm": 0.2634020745754242,
|
5502 |
+
"learning_rate": 3.1072749691738595e-06,
|
5503 |
+
"loss": 1.2125,
|
5504 |
+
"step": 785
|
5505 |
+
},
|
5506 |
+
{
|
5507 |
+
"epoch": 0.8625514403292182,
|
5508 |
+
"grad_norm": 0.24719679355621338,
|
5509 |
+
"learning_rate": 3.0826140567200992e-06,
|
5510 |
+
"loss": 1.2097,
|
5511 |
+
"step": 786
|
5512 |
+
},
|
5513 |
+
{
|
5514 |
+
"epoch": 0.8636488340192043,
|
5515 |
+
"grad_norm": 0.28501445055007935,
|
5516 |
+
"learning_rate": 3.057953144266338e-06,
|
5517 |
+
"loss": 1.1259,
|
5518 |
+
"step": 787
|
5519 |
+
},
|
5520 |
+
{
|
5521 |
+
"epoch": 0.8647462277091906,
|
5522 |
+
"grad_norm": 0.25702810287475586,
|
5523 |
+
"learning_rate": 3.0332922318125775e-06,
|
5524 |
+
"loss": 1.2083,
|
5525 |
+
"step": 788
|
5526 |
+
},
|
5527 |
+
{
|
5528 |
+
"epoch": 0.8658436213991769,
|
5529 |
+
"grad_norm": 0.25063031911849976,
|
5530 |
+
"learning_rate": 3.0086313193588164e-06,
|
5531 |
+
"loss": 1.172,
|
5532 |
+
"step": 789
|
5533 |
+
},
|
5534 |
+
{
|
5535 |
+
"epoch": 0.8669410150891632,
|
5536 |
+
"grad_norm": 0.2768210172653198,
|
5537 |
+
"learning_rate": 2.9839704069050557e-06,
|
5538 |
+
"loss": 1.181,
|
5539 |
+
"step": 790
|
5540 |
+
},
|
5541 |
+
{
|
5542 |
+
"epoch": 0.8680384087791495,
|
5543 |
+
"grad_norm": 0.24431820213794708,
|
5544 |
+
"learning_rate": 2.959309494451295e-06,
|
5545 |
+
"loss": 1.2165,
|
5546 |
+
"step": 791
|
5547 |
+
},
|
5548 |
+
{
|
5549 |
+
"epoch": 0.8691358024691358,
|
5550 |
+
"grad_norm": 0.2454931139945984,
|
5551 |
+
"learning_rate": 2.934648581997534e-06,
|
5552 |
+
"loss": 1.2469,
|
5553 |
+
"step": 792
|
5554 |
+
},
|
5555 |
+
{
|
5556 |
+
"epoch": 0.8702331961591221,
|
5557 |
+
"grad_norm": 0.2588542401790619,
|
5558 |
+
"learning_rate": 2.9099876695437733e-06,
|
5559 |
+
"loss": 1.2271,
|
5560 |
+
"step": 793
|
5561 |
+
},
|
5562 |
+
{
|
5563 |
+
"epoch": 0.8713305898491084,
|
5564 |
+
"grad_norm": 0.22791962325572968,
|
5565 |
+
"learning_rate": 2.885326757090012e-06,
|
5566 |
+
"loss": 1.2226,
|
5567 |
+
"step": 794
|
5568 |
+
},
|
5569 |
+
{
|
5570 |
+
"epoch": 0.8724279835390947,
|
5571 |
+
"grad_norm": 0.25886085629463196,
|
5572 |
+
"learning_rate": 2.860665844636252e-06,
|
5573 |
+
"loss": 1.2904,
|
5574 |
+
"step": 795
|
5575 |
+
},
|
5576 |
+
{
|
5577 |
+
"epoch": 0.873525377229081,
|
5578 |
+
"grad_norm": 0.24065649509429932,
|
5579 |
+
"learning_rate": 2.8360049321824913e-06,
|
5580 |
+
"loss": 1.201,
|
5581 |
+
"step": 796
|
5582 |
+
},
|
5583 |
+
{
|
5584 |
+
"epoch": 0.8746227709190673,
|
5585 |
+
"grad_norm": 0.2704163193702698,
|
5586 |
+
"learning_rate": 2.81134401972873e-06,
|
5587 |
+
"loss": 1.2145,
|
5588 |
+
"step": 797
|
5589 |
+
},
|
5590 |
+
{
|
5591 |
+
"epoch": 0.8757201646090536,
|
5592 |
+
"grad_norm": 0.25260040163993835,
|
5593 |
+
"learning_rate": 2.7866831072749695e-06,
|
5594 |
+
"loss": 1.2277,
|
5595 |
+
"step": 798
|
5596 |
+
},
|
5597 |
+
{
|
5598 |
+
"epoch": 0.8768175582990397,
|
5599 |
+
"grad_norm": 0.2618052363395691,
|
5600 |
+
"learning_rate": 2.7620221948212084e-06,
|
5601 |
+
"loss": 1.2252,
|
5602 |
+
"step": 799
|
5603 |
+
},
|
5604 |
+
{
|
5605 |
+
"epoch": 0.877914951989026,
|
5606 |
+
"grad_norm": 0.29944562911987305,
|
5607 |
+
"learning_rate": 2.7373612823674478e-06,
|
5608 |
+
"loss": 1.2221,
|
5609 |
+
"step": 800
|
5610 |
+
},
|
5611 |
+
{
|
5612 |
+
"epoch": 0.8790123456790123,
|
5613 |
+
"grad_norm": 0.2519710958003998,
|
5614 |
+
"learning_rate": 2.712700369913687e-06,
|
5615 |
+
"loss": 1.2022,
|
5616 |
+
"step": 801
|
5617 |
+
},
|
5618 |
+
{
|
5619 |
+
"epoch": 0.8801097393689986,
|
5620 |
+
"grad_norm": 0.29498758912086487,
|
5621 |
+
"learning_rate": 2.688039457459926e-06,
|
5622 |
+
"loss": 1.1939,
|
5623 |
+
"step": 802
|
5624 |
+
},
|
5625 |
+
{
|
5626 |
+
"epoch": 0.8812071330589849,
|
5627 |
+
"grad_norm": 0.3040591776371002,
|
5628 |
+
"learning_rate": 2.6633785450061657e-06,
|
5629 |
+
"loss": 1.1721,
|
5630 |
+
"step": 803
|
5631 |
+
},
|
5632 |
+
{
|
5633 |
+
"epoch": 0.8823045267489712,
|
5634 |
+
"grad_norm": 0.24407289922237396,
|
5635 |
+
"learning_rate": 2.6387176325524042e-06,
|
5636 |
+
"loss": 1.281,
|
5637 |
+
"step": 804
|
5638 |
+
},
|
5639 |
+
{
|
5640 |
+
"epoch": 0.8834019204389575,
|
5641 |
+
"grad_norm": 0.3141189217567444,
|
5642 |
+
"learning_rate": 2.614056720098644e-06,
|
5643 |
+
"loss": 1.1731,
|
5644 |
+
"step": 805
|
5645 |
+
},
|
5646 |
+
{
|
5647 |
+
"epoch": 0.8844993141289438,
|
5648 |
+
"grad_norm": 0.2557481527328491,
|
5649 |
+
"learning_rate": 2.5893958076448833e-06,
|
5650 |
+
"loss": 1.2577,
|
5651 |
+
"step": 806
|
5652 |
+
},
|
5653 |
+
{
|
5654 |
+
"epoch": 0.8855967078189301,
|
5655 |
+
"grad_norm": 0.256987601518631,
|
5656 |
+
"learning_rate": 2.5647348951911222e-06,
|
5657 |
+
"loss": 1.1678,
|
5658 |
+
"step": 807
|
5659 |
+
},
|
5660 |
+
{
|
5661 |
+
"epoch": 0.8866941015089164,
|
5662 |
+
"grad_norm": 0.26355645060539246,
|
5663 |
+
"learning_rate": 2.5400739827373616e-06,
|
5664 |
+
"loss": 1.2432,
|
5665 |
+
"step": 808
|
5666 |
+
},
|
5667 |
+
{
|
5668 |
+
"epoch": 0.8877914951989027,
|
5669 |
+
"grad_norm": 0.2442682534456253,
|
5670 |
+
"learning_rate": 2.5154130702836005e-06,
|
5671 |
+
"loss": 1.3616,
|
5672 |
+
"step": 809
|
5673 |
+
},
|
5674 |
+
{
|
5675 |
+
"epoch": 0.8888888888888888,
|
5676 |
+
"grad_norm": 0.29615169763565063,
|
5677 |
+
"learning_rate": 2.49075215782984e-06,
|
5678 |
+
"loss": 1.2012,
|
5679 |
+
"step": 810
|
5680 |
+
},
|
5681 |
+
{
|
5682 |
+
"epoch": 0.8899862825788751,
|
5683 |
+
"grad_norm": 0.25238725543022156,
|
5684 |
+
"learning_rate": 2.466091245376079e-06,
|
5685 |
+
"loss": 1.1903,
|
5686 |
+
"step": 811
|
5687 |
+
},
|
5688 |
+
{
|
5689 |
+
"epoch": 0.8910836762688614,
|
5690 |
+
"grad_norm": 0.25462445616722107,
|
5691 |
+
"learning_rate": 2.441430332922318e-06,
|
5692 |
+
"loss": 1.1539,
|
5693 |
+
"step": 812
|
5694 |
+
},
|
5695 |
+
{
|
5696 |
+
"epoch": 0.8921810699588477,
|
5697 |
+
"grad_norm": 0.2639220356941223,
|
5698 |
+
"learning_rate": 2.416769420468558e-06,
|
5699 |
+
"loss": 1.1004,
|
5700 |
+
"step": 813
|
5701 |
+
},
|
5702 |
+
{
|
5703 |
+
"epoch": 0.893278463648834,
|
5704 |
+
"grad_norm": 0.287663072347641,
|
5705 |
+
"learning_rate": 2.3921085080147967e-06,
|
5706 |
+
"loss": 1.2643,
|
5707 |
+
"step": 814
|
5708 |
+
},
|
5709 |
+
{
|
5710 |
+
"epoch": 0.8943758573388203,
|
5711 |
+
"grad_norm": 0.24938149750232697,
|
5712 |
+
"learning_rate": 2.367447595561036e-06,
|
5713 |
+
"loss": 1.3602,
|
5714 |
+
"step": 815
|
5715 |
+
},
|
5716 |
+
{
|
5717 |
+
"epoch": 0.8954732510288066,
|
5718 |
+
"grad_norm": 0.2602677643299103,
|
5719 |
+
"learning_rate": 2.342786683107275e-06,
|
5720 |
+
"loss": 1.2043,
|
5721 |
+
"step": 816
|
5722 |
+
},
|
5723 |
+
{
|
5724 |
+
"epoch": 0.8965706447187929,
|
5725 |
+
"grad_norm": 0.25690793991088867,
|
5726 |
+
"learning_rate": 2.3181257706535143e-06,
|
5727 |
+
"loss": 1.1586,
|
5728 |
+
"step": 817
|
5729 |
+
},
|
5730 |
+
{
|
5731 |
+
"epoch": 0.8976680384087792,
|
5732 |
+
"grad_norm": 0.2565641701221466,
|
5733 |
+
"learning_rate": 2.2934648581997536e-06,
|
5734 |
+
"loss": 1.2512,
|
5735 |
+
"step": 818
|
5736 |
+
},
|
5737 |
+
{
|
5738 |
+
"epoch": 0.8987654320987655,
|
5739 |
+
"grad_norm": 0.2628938853740692,
|
5740 |
+
"learning_rate": 2.268803945745993e-06,
|
5741 |
+
"loss": 1.1438,
|
5742 |
+
"step": 819
|
5743 |
+
},
|
5744 |
+
{
|
5745 |
+
"epoch": 0.8998628257887518,
|
5746 |
+
"grad_norm": 0.23171505331993103,
|
5747 |
+
"learning_rate": 2.244143033292232e-06,
|
5748 |
+
"loss": 1.1741,
|
5749 |
+
"step": 820
|
5750 |
+
},
|
5751 |
+
{
|
5752 |
+
"epoch": 0.900960219478738,
|
5753 |
+
"grad_norm": 0.2506265640258789,
|
5754 |
+
"learning_rate": 2.219482120838471e-06,
|
5755 |
+
"loss": 1.1697,
|
5756 |
+
"step": 821
|
5757 |
+
},
|
5758 |
+
{
|
5759 |
+
"epoch": 0.9020576131687242,
|
5760 |
+
"grad_norm": 0.27947020530700684,
|
5761 |
+
"learning_rate": 2.1948212083847105e-06,
|
5762 |
+
"loss": 1.1571,
|
5763 |
+
"step": 822
|
5764 |
+
},
|
5765 |
+
{
|
5766 |
+
"epoch": 0.9031550068587105,
|
5767 |
+
"grad_norm": 0.2594459354877472,
|
5768 |
+
"learning_rate": 2.1701602959309494e-06,
|
5769 |
+
"loss": 1.2105,
|
5770 |
+
"step": 823
|
5771 |
+
},
|
5772 |
+
{
|
5773 |
+
"epoch": 0.9042524005486968,
|
5774 |
+
"grad_norm": 0.24918478727340698,
|
5775 |
+
"learning_rate": 2.1454993834771887e-06,
|
5776 |
+
"loss": 1.2497,
|
5777 |
+
"step": 824
|
5778 |
+
},
|
5779 |
+
{
|
5780 |
+
"epoch": 0.9053497942386831,
|
5781 |
+
"grad_norm": 0.2924598157405853,
|
5782 |
+
"learning_rate": 2.120838471023428e-06,
|
5783 |
+
"loss": 1.3235,
|
5784 |
+
"step": 825
|
5785 |
+
},
|
5786 |
+
{
|
5787 |
+
"epoch": 0.9064471879286694,
|
5788 |
+
"grad_norm": 0.24772346019744873,
|
5789 |
+
"learning_rate": 2.0961775585696674e-06,
|
5790 |
+
"loss": 1.2818,
|
5791 |
+
"step": 826
|
5792 |
+
},
|
5793 |
+
{
|
5794 |
+
"epoch": 0.9075445816186557,
|
5795 |
+
"grad_norm": 0.2838573157787323,
|
5796 |
+
"learning_rate": 2.0715166461159063e-06,
|
5797 |
+
"loss": 1.1918,
|
5798 |
+
"step": 827
|
5799 |
+
},
|
5800 |
+
{
|
5801 |
+
"epoch": 0.908641975308642,
|
5802 |
+
"grad_norm": 0.26798808574676514,
|
5803 |
+
"learning_rate": 2.0468557336621456e-06,
|
5804 |
+
"loss": 1.1874,
|
5805 |
+
"step": 828
|
5806 |
+
},
|
5807 |
+
{
|
5808 |
+
"epoch": 0.9097393689986283,
|
5809 |
+
"grad_norm": 0.2984868586063385,
|
5810 |
+
"learning_rate": 2.022194821208385e-06,
|
5811 |
+
"loss": 1.2417,
|
5812 |
+
"step": 829
|
5813 |
+
},
|
5814 |
+
{
|
5815 |
+
"epoch": 0.9108367626886146,
|
5816 |
+
"grad_norm": 0.27059322595596313,
|
5817 |
+
"learning_rate": 1.9975339087546243e-06,
|
5818 |
+
"loss": 1.2273,
|
5819 |
+
"step": 830
|
5820 |
+
},
|
5821 |
+
{
|
5822 |
+
"epoch": 0.9119341563786009,
|
5823 |
+
"grad_norm": 0.2572599947452545,
|
5824 |
+
"learning_rate": 1.9728729963008632e-06,
|
5825 |
+
"loss": 1.186,
|
5826 |
+
"step": 831
|
5827 |
+
},
|
5828 |
+
{
|
5829 |
+
"epoch": 0.9130315500685872,
|
5830 |
+
"grad_norm": 0.27578267455101013,
|
5831 |
+
"learning_rate": 1.9482120838471025e-06,
|
5832 |
+
"loss": 1.1343,
|
5833 |
+
"step": 832
|
5834 |
+
},
|
5835 |
+
{
|
5836 |
+
"epoch": 0.9141289437585733,
|
5837 |
+
"grad_norm": 0.25030946731567383,
|
5838 |
+
"learning_rate": 1.9235511713933415e-06,
|
5839 |
+
"loss": 1.2599,
|
5840 |
+
"step": 833
|
5841 |
+
},
|
5842 |
+
{
|
5843 |
+
"epoch": 0.9152263374485596,
|
5844 |
+
"grad_norm": 0.23245501518249512,
|
5845 |
+
"learning_rate": 1.8988902589395808e-06,
|
5846 |
+
"loss": 1.2471,
|
5847 |
+
"step": 834
|
5848 |
+
},
|
5849 |
+
{
|
5850 |
+
"epoch": 0.9163237311385459,
|
5851 |
+
"grad_norm": 0.23760074377059937,
|
5852 |
+
"learning_rate": 1.8742293464858201e-06,
|
5853 |
+
"loss": 1.2457,
|
5854 |
+
"step": 835
|
5855 |
+
},
|
5856 |
+
{
|
5857 |
+
"epoch": 0.9174211248285322,
|
5858 |
+
"grad_norm": 0.2975304424762726,
|
5859 |
+
"learning_rate": 1.8495684340320595e-06,
|
5860 |
+
"loss": 1.2025,
|
5861 |
+
"step": 836
|
5862 |
+
},
|
5863 |
+
{
|
5864 |
+
"epoch": 0.9185185185185185,
|
5865 |
+
"grad_norm": 0.240847647190094,
|
5866 |
+
"learning_rate": 1.8249075215782986e-06,
|
5867 |
+
"loss": 1.2289,
|
5868 |
+
"step": 837
|
5869 |
+
},
|
5870 |
+
{
|
5871 |
+
"epoch": 0.9196159122085048,
|
5872 |
+
"grad_norm": 0.23974715173244476,
|
5873 |
+
"learning_rate": 1.8002466091245377e-06,
|
5874 |
+
"loss": 1.2117,
|
5875 |
+
"step": 838
|
5876 |
+
},
|
5877 |
+
{
|
5878 |
+
"epoch": 0.9207133058984911,
|
5879 |
+
"grad_norm": 0.27474868297576904,
|
5880 |
+
"learning_rate": 1.7755856966707768e-06,
|
5881 |
+
"loss": 1.2863,
|
5882 |
+
"step": 839
|
5883 |
+
},
|
5884 |
+
{
|
5885 |
+
"epoch": 0.9218106995884774,
|
5886 |
+
"grad_norm": 0.28905266523361206,
|
5887 |
+
"learning_rate": 1.7509247842170164e-06,
|
5888 |
+
"loss": 1.2041,
|
5889 |
+
"step": 840
|
5890 |
+
},
|
5891 |
+
{
|
5892 |
+
"epoch": 0.9229080932784637,
|
5893 |
+
"grad_norm": 0.25439831614494324,
|
5894 |
+
"learning_rate": 1.7262638717632555e-06,
|
5895 |
+
"loss": 1.2385,
|
5896 |
+
"step": 841
|
5897 |
+
},
|
5898 |
+
{
|
5899 |
+
"epoch": 0.92400548696845,
|
5900 |
+
"grad_norm": 0.2607545256614685,
|
5901 |
+
"learning_rate": 1.7016029593094946e-06,
|
5902 |
+
"loss": 1.2837,
|
5903 |
+
"step": 842
|
5904 |
+
},
|
5905 |
+
{
|
5906 |
+
"epoch": 0.9251028806584363,
|
5907 |
+
"grad_norm": 0.34511038661003113,
|
5908 |
+
"learning_rate": 1.6769420468557337e-06,
|
5909 |
+
"loss": 1.2155,
|
5910 |
+
"step": 843
|
5911 |
+
},
|
5912 |
+
{
|
5913 |
+
"epoch": 0.9262002743484224,
|
5914 |
+
"grad_norm": 0.26956722140312195,
|
5915 |
+
"learning_rate": 1.6522811344019728e-06,
|
5916 |
+
"loss": 1.1794,
|
5917 |
+
"step": 844
|
5918 |
+
},
|
5919 |
+
{
|
5920 |
+
"epoch": 0.9272976680384087,
|
5921 |
+
"grad_norm": 0.25592947006225586,
|
5922 |
+
"learning_rate": 1.6276202219482124e-06,
|
5923 |
+
"loss": 1.1996,
|
5924 |
+
"step": 845
|
5925 |
+
},
|
5926 |
+
{
|
5927 |
+
"epoch": 0.928395061728395,
|
5928 |
+
"grad_norm": 0.25805869698524475,
|
5929 |
+
"learning_rate": 1.6029593094944515e-06,
|
5930 |
+
"loss": 1.2655,
|
5931 |
+
"step": 846
|
5932 |
+
},
|
5933 |
+
{
|
5934 |
+
"epoch": 0.9294924554183813,
|
5935 |
+
"grad_norm": 0.2707035541534424,
|
5936 |
+
"learning_rate": 1.5782983970406906e-06,
|
5937 |
+
"loss": 1.3274,
|
5938 |
+
"step": 847
|
5939 |
+
},
|
5940 |
+
{
|
5941 |
+
"epoch": 0.9305898491083676,
|
5942 |
+
"grad_norm": 0.2707644999027252,
|
5943 |
+
"learning_rate": 1.5536374845869297e-06,
|
5944 |
+
"loss": 1.2375,
|
5945 |
+
"step": 848
|
5946 |
+
},
|
5947 |
+
{
|
5948 |
+
"epoch": 0.9316872427983539,
|
5949 |
+
"grad_norm": 0.2761702537536621,
|
5950 |
+
"learning_rate": 1.528976572133169e-06,
|
5951 |
+
"loss": 1.1769,
|
5952 |
+
"step": 849
|
5953 |
+
},
|
5954 |
+
{
|
5955 |
+
"epoch": 0.9327846364883402,
|
5956 |
+
"grad_norm": 0.26850977540016174,
|
5957 |
+
"learning_rate": 1.5043156596794082e-06,
|
5958 |
+
"loss": 1.1833,
|
5959 |
+
"step": 850
|
5960 |
+
},
|
5961 |
+
{
|
5962 |
+
"epoch": 0.9338820301783265,
|
5963 |
+
"grad_norm": 0.26352792978286743,
|
5964 |
+
"learning_rate": 1.4796547472256475e-06,
|
5965 |
+
"loss": 1.2284,
|
5966 |
+
"step": 851
|
5967 |
+
},
|
5968 |
+
{
|
5969 |
+
"epoch": 0.9349794238683128,
|
5970 |
+
"grad_norm": 0.24324378371238708,
|
5971 |
+
"learning_rate": 1.4549938347718866e-06,
|
5972 |
+
"loss": 1.2005,
|
5973 |
+
"step": 852
|
5974 |
+
},
|
5975 |
+
{
|
5976 |
+
"epoch": 0.9360768175582991,
|
5977 |
+
"grad_norm": 0.24241109192371368,
|
5978 |
+
"learning_rate": 1.430332922318126e-06,
|
5979 |
+
"loss": 1.2059,
|
5980 |
+
"step": 853
|
5981 |
+
},
|
5982 |
+
{
|
5983 |
+
"epoch": 0.9371742112482854,
|
5984 |
+
"grad_norm": 0.26090410351753235,
|
5985 |
+
"learning_rate": 1.405672009864365e-06,
|
5986 |
+
"loss": 1.2379,
|
5987 |
+
"step": 854
|
5988 |
+
},
|
5989 |
+
{
|
5990 |
+
"epoch": 0.9382716049382716,
|
5991 |
+
"grad_norm": 0.2450953871011734,
|
5992 |
+
"learning_rate": 1.3810110974106042e-06,
|
5993 |
+
"loss": 1.2946,
|
5994 |
+
"step": 855
|
5995 |
+
},
|
5996 |
+
{
|
5997 |
+
"epoch": 0.9393689986282578,
|
5998 |
+
"grad_norm": 0.24933604896068573,
|
5999 |
+
"learning_rate": 1.3563501849568435e-06,
|
6000 |
+
"loss": 1.1612,
|
6001 |
+
"step": 856
|
6002 |
+
},
|
6003 |
+
{
|
6004 |
+
"epoch": 0.9404663923182441,
|
6005 |
+
"grad_norm": 0.2418948858976364,
|
6006 |
+
"learning_rate": 1.3316892725030829e-06,
|
6007 |
+
"loss": 1.229,
|
6008 |
+
"step": 857
|
6009 |
+
},
|
6010 |
+
{
|
6011 |
+
"epoch": 0.9415637860082304,
|
6012 |
+
"grad_norm": 0.26510903239250183,
|
6013 |
+
"learning_rate": 1.307028360049322e-06,
|
6014 |
+
"loss": 1.1778,
|
6015 |
+
"step": 858
|
6016 |
+
},
|
6017 |
+
{
|
6018 |
+
"epoch": 0.9426611796982167,
|
6019 |
+
"grad_norm": 0.2577672302722931,
|
6020 |
+
"learning_rate": 1.2823674475955611e-06,
|
6021 |
+
"loss": 1.2076,
|
6022 |
+
"step": 859
|
6023 |
+
},
|
6024 |
+
{
|
6025 |
+
"epoch": 0.943758573388203,
|
6026 |
+
"grad_norm": 0.24987109005451202,
|
6027 |
+
"learning_rate": 1.2577065351418002e-06,
|
6028 |
+
"loss": 1.2105,
|
6029 |
+
"step": 860
|
6030 |
+
},
|
6031 |
+
{
|
6032 |
+
"epoch": 0.9448559670781893,
|
6033 |
+
"grad_norm": 0.25255313515663147,
|
6034 |
+
"learning_rate": 1.2330456226880396e-06,
|
6035 |
+
"loss": 1.1777,
|
6036 |
+
"step": 861
|
6037 |
+
},
|
6038 |
+
{
|
6039 |
+
"epoch": 0.9459533607681756,
|
6040 |
+
"grad_norm": 0.2602977752685547,
|
6041 |
+
"learning_rate": 1.208384710234279e-06,
|
6042 |
+
"loss": 1.2409,
|
6043 |
+
"step": 862
|
6044 |
+
},
|
6045 |
+
{
|
6046 |
+
"epoch": 0.9470507544581619,
|
6047 |
+
"grad_norm": 0.2703125476837158,
|
6048 |
+
"learning_rate": 1.183723797780518e-06,
|
6049 |
+
"loss": 1.3063,
|
6050 |
+
"step": 863
|
6051 |
+
},
|
6052 |
+
{
|
6053 |
+
"epoch": 0.9481481481481482,
|
6054 |
+
"grad_norm": 0.23141594231128693,
|
6055 |
+
"learning_rate": 1.1590628853267571e-06,
|
6056 |
+
"loss": 1.223,
|
6057 |
+
"step": 864
|
6058 |
+
},
|
6059 |
+
{
|
6060 |
+
"epoch": 0.9492455418381345,
|
6061 |
+
"grad_norm": 0.26518914103507996,
|
6062 |
+
"learning_rate": 1.1344019728729965e-06,
|
6063 |
+
"loss": 1.2147,
|
6064 |
+
"step": 865
|
6065 |
+
},
|
6066 |
+
{
|
6067 |
+
"epoch": 0.9503429355281207,
|
6068 |
+
"grad_norm": 0.28556376695632935,
|
6069 |
+
"learning_rate": 1.1097410604192356e-06,
|
6070 |
+
"loss": 1.1871,
|
6071 |
+
"step": 866
|
6072 |
+
},
|
6073 |
+
{
|
6074 |
+
"epoch": 0.951440329218107,
|
6075 |
+
"grad_norm": 0.2666340172290802,
|
6076 |
+
"learning_rate": 1.0850801479654747e-06,
|
6077 |
+
"loss": 1.1842,
|
6078 |
+
"step": 867
|
6079 |
+
},
|
6080 |
+
{
|
6081 |
+
"epoch": 0.9525377229080932,
|
6082 |
+
"grad_norm": 0.23198631405830383,
|
6083 |
+
"learning_rate": 1.060419235511714e-06,
|
6084 |
+
"loss": 1.2356,
|
6085 |
+
"step": 868
|
6086 |
+
},
|
6087 |
+
{
|
6088 |
+
"epoch": 0.9536351165980795,
|
6089 |
+
"grad_norm": 0.2616305947303772,
|
6090 |
+
"learning_rate": 1.0357583230579532e-06,
|
6091 |
+
"loss": 1.3017,
|
6092 |
+
"step": 869
|
6093 |
+
},
|
6094 |
+
{
|
6095 |
+
"epoch": 0.9547325102880658,
|
6096 |
+
"grad_norm": 0.2699725925922394,
|
6097 |
+
"learning_rate": 1.0110974106041925e-06,
|
6098 |
+
"loss": 1.1505,
|
6099 |
+
"step": 870
|
6100 |
+
},
|
6101 |
+
{
|
6102 |
+
"epoch": 0.9558299039780521,
|
6103 |
+
"grad_norm": 0.2656715512275696,
|
6104 |
+
"learning_rate": 9.864364981504316e-07,
|
6105 |
+
"loss": 1.2487,
|
6106 |
+
"step": 871
|
6107 |
+
},
|
6108 |
+
{
|
6109 |
+
"epoch": 0.9569272976680384,
|
6110 |
+
"grad_norm": 0.24682074785232544,
|
6111 |
+
"learning_rate": 9.617755856966707e-07,
|
6112 |
+
"loss": 1.2081,
|
6113 |
+
"step": 872
|
6114 |
+
},
|
6115 |
+
{
|
6116 |
+
"epoch": 0.9580246913580247,
|
6117 |
+
"grad_norm": 0.26323434710502625,
|
6118 |
+
"learning_rate": 9.371146732429101e-07,
|
6119 |
+
"loss": 1.2228,
|
6120 |
+
"step": 873
|
6121 |
+
},
|
6122 |
+
{
|
6123 |
+
"epoch": 0.959122085048011,
|
6124 |
+
"grad_norm": 0.2442006766796112,
|
6125 |
+
"learning_rate": 9.124537607891493e-07,
|
6126 |
+
"loss": 1.2603,
|
6127 |
+
"step": 874
|
6128 |
+
},
|
6129 |
+
{
|
6130 |
+
"epoch": 0.9602194787379973,
|
6131 |
+
"grad_norm": 0.28010329604148865,
|
6132 |
+
"learning_rate": 8.877928483353884e-07,
|
6133 |
+
"loss": 1.2486,
|
6134 |
+
"step": 875
|
6135 |
+
},
|
6136 |
+
{
|
6137 |
+
"epoch": 0.9613168724279836,
|
6138 |
+
"grad_norm": 0.2618809640407562,
|
6139 |
+
"learning_rate": 8.631319358816277e-07,
|
6140 |
+
"loss": 1.2433,
|
6141 |
+
"step": 876
|
6142 |
+
},
|
6143 |
+
{
|
6144 |
+
"epoch": 0.9624142661179699,
|
6145 |
+
"grad_norm": 0.2693452537059784,
|
6146 |
+
"learning_rate": 8.384710234278669e-07,
|
6147 |
+
"loss": 1.2296,
|
6148 |
+
"step": 877
|
6149 |
+
},
|
6150 |
+
{
|
6151 |
+
"epoch": 0.9635116598079561,
|
6152 |
+
"grad_norm": 0.256401926279068,
|
6153 |
+
"learning_rate": 8.138101109741062e-07,
|
6154 |
+
"loss": 1.2447,
|
6155 |
+
"step": 878
|
6156 |
+
},
|
6157 |
+
{
|
6158 |
+
"epoch": 0.9646090534979423,
|
6159 |
+
"grad_norm": 0.2823677659034729,
|
6160 |
+
"learning_rate": 7.891491985203453e-07,
|
6161 |
+
"loss": 1.1789,
|
6162 |
+
"step": 879
|
6163 |
+
},
|
6164 |
+
{
|
6165 |
+
"epoch": 0.9657064471879286,
|
6166 |
+
"grad_norm": 0.29250475764274597,
|
6167 |
+
"learning_rate": 7.644882860665845e-07,
|
6168 |
+
"loss": 1.1497,
|
6169 |
+
"step": 880
|
6170 |
+
},
|
6171 |
+
{
|
6172 |
+
"epoch": 0.9668038408779149,
|
6173 |
+
"grad_norm": 0.25502100586891174,
|
6174 |
+
"learning_rate": 7.398273736128238e-07,
|
6175 |
+
"loss": 1.2713,
|
6176 |
+
"step": 881
|
6177 |
+
},
|
6178 |
+
{
|
6179 |
+
"epoch": 0.9679012345679012,
|
6180 |
+
"grad_norm": 0.2642868161201477,
|
6181 |
+
"learning_rate": 7.15166461159063e-07,
|
6182 |
+
"loss": 1.2178,
|
6183 |
+
"step": 882
|
6184 |
+
},
|
6185 |
+
{
|
6186 |
+
"epoch": 0.9689986282578875,
|
6187 |
+
"grad_norm": 0.25020915269851685,
|
6188 |
+
"learning_rate": 6.905055487053021e-07,
|
6189 |
+
"loss": 1.2125,
|
6190 |
+
"step": 883
|
6191 |
+
},
|
6192 |
+
{
|
6193 |
+
"epoch": 0.9700960219478738,
|
6194 |
+
"grad_norm": 0.25948819518089294,
|
6195 |
+
"learning_rate": 6.658446362515414e-07,
|
6196 |
+
"loss": 1.2355,
|
6197 |
+
"step": 884
|
6198 |
+
},
|
6199 |
+
{
|
6200 |
+
"epoch": 0.9711934156378601,
|
6201 |
+
"grad_norm": 0.25626036524772644,
|
6202 |
+
"learning_rate": 6.411837237977806e-07,
|
6203 |
+
"loss": 1.227,
|
6204 |
+
"step": 885
|
6205 |
+
},
|
6206 |
+
{
|
6207 |
+
"epoch": 0.9722908093278464,
|
6208 |
+
"grad_norm": 0.24282559752464294,
|
6209 |
+
"learning_rate": 6.165228113440198e-07,
|
6210 |
+
"loss": 1.2019,
|
6211 |
+
"step": 886
|
6212 |
+
},
|
6213 |
+
{
|
6214 |
+
"epoch": 0.9733882030178327,
|
6215 |
+
"grad_norm": 0.2501373291015625,
|
6216 |
+
"learning_rate": 5.91861898890259e-07,
|
6217 |
+
"loss": 1.2002,
|
6218 |
+
"step": 887
|
6219 |
+
},
|
6220 |
+
{
|
6221 |
+
"epoch": 0.974485596707819,
|
6222 |
+
"grad_norm": 0.27020156383514404,
|
6223 |
+
"learning_rate": 5.672009864364982e-07,
|
6224 |
+
"loss": 1.2451,
|
6225 |
+
"step": 888
|
6226 |
+
},
|
6227 |
+
{
|
6228 |
+
"epoch": 0.9755829903978052,
|
6229 |
+
"grad_norm": 0.2579203248023987,
|
6230 |
+
"learning_rate": 5.425400739827374e-07,
|
6231 |
+
"loss": 1.277,
|
6232 |
+
"step": 889
|
6233 |
+
},
|
6234 |
+
{
|
6235 |
+
"epoch": 0.9766803840877915,
|
6236 |
+
"grad_norm": 0.24406872689723969,
|
6237 |
+
"learning_rate": 5.178791615289766e-07,
|
6238 |
+
"loss": 1.2816,
|
6239 |
+
"step": 890
|
6240 |
+
},
|
6241 |
+
{
|
6242 |
+
"epoch": 0.9777777777777777,
|
6243 |
+
"grad_norm": 0.2537878453731537,
|
6244 |
+
"learning_rate": 4.932182490752158e-07,
|
6245 |
+
"loss": 1.23,
|
6246 |
+
"step": 891
|
6247 |
+
},
|
6248 |
+
{
|
6249 |
+
"epoch": 0.978875171467764,
|
6250 |
+
"grad_norm": 0.2799593508243561,
|
6251 |
+
"learning_rate": 4.6855733662145503e-07,
|
6252 |
+
"loss": 1.2479,
|
6253 |
+
"step": 892
|
6254 |
+
},
|
6255 |
+
{
|
6256 |
+
"epoch": 0.9799725651577503,
|
6257 |
+
"grad_norm": 0.23126421868801117,
|
6258 |
+
"learning_rate": 4.438964241676942e-07,
|
6259 |
+
"loss": 1.2648,
|
6260 |
+
"step": 893
|
6261 |
+
},
|
6262 |
+
{
|
6263 |
+
"epoch": 0.9810699588477366,
|
6264 |
+
"grad_norm": 0.2650638818740845,
|
6265 |
+
"learning_rate": 4.1923551171393343e-07,
|
6266 |
+
"loss": 1.2301,
|
6267 |
+
"step": 894
|
6268 |
+
},
|
6269 |
+
{
|
6270 |
+
"epoch": 0.9821673525377229,
|
6271 |
+
"grad_norm": 0.27923524379730225,
|
6272 |
+
"learning_rate": 3.9457459926017265e-07,
|
6273 |
+
"loss": 1.188,
|
6274 |
+
"step": 895
|
6275 |
+
},
|
6276 |
+
{
|
6277 |
+
"epoch": 0.9832647462277092,
|
6278 |
+
"grad_norm": 0.4018268287181854,
|
6279 |
+
"learning_rate": 3.699136868064119e-07,
|
6280 |
+
"loss": 1.234,
|
6281 |
+
"step": 896
|
6282 |
+
},
|
6283 |
+
{
|
6284 |
+
"epoch": 0.9843621399176955,
|
6285 |
+
"grad_norm": 0.26936185359954834,
|
6286 |
+
"learning_rate": 3.4525277435265105e-07,
|
6287 |
+
"loss": 1.2962,
|
6288 |
+
"step": 897
|
6289 |
+
},
|
6290 |
+
{
|
6291 |
+
"epoch": 0.9854595336076818,
|
6292 |
+
"grad_norm": 0.2637123167514801,
|
6293 |
+
"learning_rate": 3.205918618988903e-07,
|
6294 |
+
"loss": 1.231,
|
6295 |
+
"step": 898
|
6296 |
+
},
|
6297 |
+
{
|
6298 |
+
"epoch": 0.9865569272976681,
|
6299 |
+
"grad_norm": 0.2618735730648041,
|
6300 |
+
"learning_rate": 2.959309494451295e-07,
|
6301 |
+
"loss": 1.2333,
|
6302 |
+
"step": 899
|
6303 |
+
},
|
6304 |
+
{
|
6305 |
+
"epoch": 0.9876543209876543,
|
6306 |
+
"grad_norm": 0.2631166875362396,
|
6307 |
+
"learning_rate": 2.712700369913687e-07,
|
6308 |
+
"loss": 1.2565,
|
6309 |
+
"step": 900
|
6310 |
+
},
|
6311 |
+
{
|
6312 |
+
"epoch": 0.9887517146776406,
|
6313 |
+
"grad_norm": 0.25191444158554077,
|
6314 |
+
"learning_rate": 2.466091245376079e-07,
|
6315 |
+
"loss": 1.2109,
|
6316 |
+
"step": 901
|
6317 |
+
},
|
6318 |
+
{
|
6319 |
+
"epoch": 0.9898491083676269,
|
6320 |
+
"grad_norm": 0.24879272282123566,
|
6321 |
+
"learning_rate": 2.219482120838471e-07,
|
6322 |
+
"loss": 1.2384,
|
6323 |
+
"step": 902
|
6324 |
+
},
|
6325 |
+
{
|
6326 |
+
"epoch": 0.9909465020576131,
|
6327 |
+
"grad_norm": 0.2868032455444336,
|
6328 |
+
"learning_rate": 1.9728729963008633e-07,
|
6329 |
+
"loss": 1.1736,
|
6330 |
+
"step": 903
|
6331 |
+
},
|
6332 |
+
{
|
6333 |
+
"epoch": 0.9920438957475994,
|
6334 |
+
"grad_norm": 0.25794628262519836,
|
6335 |
+
"learning_rate": 1.7262638717632553e-07,
|
6336 |
+
"loss": 1.1557,
|
6337 |
+
"step": 904
|
6338 |
+
},
|
6339 |
+
{
|
6340 |
+
"epoch": 0.9931412894375857,
|
6341 |
+
"grad_norm": 0.29972419142723083,
|
6342 |
+
"learning_rate": 1.4796547472256475e-07,
|
6343 |
+
"loss": 1.1637,
|
6344 |
+
"step": 905
|
6345 |
+
},
|
6346 |
+
{
|
6347 |
+
"epoch": 0.994238683127572,
|
6348 |
+
"grad_norm": 0.24752122163772583,
|
6349 |
+
"learning_rate": 1.2330456226880395e-07,
|
6350 |
+
"loss": 1.2375,
|
6351 |
+
"step": 906
|
6352 |
+
},
|
6353 |
+
{
|
6354 |
+
"epoch": 0.9953360768175583,
|
6355 |
+
"grad_norm": 0.2406410425901413,
|
6356 |
+
"learning_rate": 9.864364981504316e-08,
|
6357 |
+
"loss": 1.319,
|
6358 |
+
"step": 907
|
6359 |
+
},
|
6360 |
+
{
|
6361 |
+
"epoch": 0.9964334705075446,
|
6362 |
+
"grad_norm": 0.2575162351131439,
|
6363 |
+
"learning_rate": 7.398273736128238e-08,
|
6364 |
+
"loss": 1.3036,
|
6365 |
+
"step": 908
|
6366 |
+
},
|
6367 |
+
{
|
6368 |
+
"epoch": 0.9975308641975309,
|
6369 |
+
"grad_norm": 0.2430945187807083,
|
6370 |
+
"learning_rate": 4.932182490752158e-08,
|
6371 |
+
"loss": 1.2212,
|
6372 |
+
"step": 909
|
6373 |
+
},
|
6374 |
+
{
|
6375 |
+
"epoch": 0.9986282578875172,
|
6376 |
+
"grad_norm": 0.2730814218521118,
|
6377 |
+
"learning_rate": 2.466091245376079e-08,
|
6378 |
+
"loss": 1.1903,
|
6379 |
+
"step": 910
|
6380 |
+
},
|
6381 |
+
{
|
6382 |
+
"epoch": 0.9997256515775035,
|
6383 |
+
"grad_norm": 0.294842928647995,
|
6384 |
+
"learning_rate": 0.0,
|
6385 |
+
"loss": 1.2108,
|
6386 |
+
"step": 911
|
6387 |
}
|
6388 |
],
|
6389 |
"logging_steps": 1,
|
|
|
6398 |
"should_evaluate": false,
|
6399 |
"should_log": false,
|
6400 |
"should_save": true,
|
6401 |
+
"should_training_stop": true
|
6402 |
},
|
6403 |
"attributes": {}
|
6404 |
}
|
6405 |
},
|
6406 |
+
"total_flos": 5.538444460514181e+17,
|
6407 |
"train_batch_size": 4,
|
6408 |
"trial_name": null,
|
6409 |
"trial_params": null
|