Training in progress, step 2400, 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 201361312
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:371aa2eba24d0f5f91495c5196526d36db51055b6cdb43e12fe4134f1b4dd4cb
|
3 |
size 201361312
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 102537812
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3bd5a98fb8233303bb539a3f1d8dcbce9cb285843a4e28ac388aa117048b5b4e
|
3 |
size 102537812
|
last-checkpoint/rng_state.pth
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 14244
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:732dadfe5f65735a5f10b3788b0ddf2f5467e863a8a5cd2cddcd862fee24ee9e
|
3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1064
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:631ea393c048f682d46e325b2ee675fb46a8894ea646e45c81e01edbe6269527
|
3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": 1.602339506149292,
|
3 |
"best_model_checkpoint": "miner_id_24/checkpoint-2200",
|
4 |
-
"epoch": 0.
|
5 |
"eval_steps": 200,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -15503,6 +15503,1414 @@
|
|
15503 |
"eval_samples_per_second": 8.733,
|
15504 |
"eval_steps_per_second": 4.372,
|
15505 |
"step": 2200
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
15506 |
}
|
15507 |
],
|
15508 |
"logging_steps": 1,
|
@@ -15517,7 +16925,7 @@
|
|
15517 |
"early_stopping_threshold": 0.0
|
15518 |
},
|
15519 |
"attributes": {
|
15520 |
-
"early_stopping_patience_counter":
|
15521 |
}
|
15522 |
},
|
15523 |
"TrainerControl": {
|
@@ -15531,7 +16939,7 @@
|
|
15531 |
"attributes": {}
|
15532 |
}
|
15533 |
},
|
15534 |
-
"total_flos": 2.
|
15535 |
"train_batch_size": 2,
|
15536 |
"trial_name": null,
|
15537 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": 1.602339506149292,
|
3 |
"best_model_checkpoint": "miner_id_24/checkpoint-2200",
|
4 |
+
"epoch": 0.11454343701900706,
|
5 |
"eval_steps": 200,
|
6 |
+
"global_step": 2400,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
15503 |
"eval_samples_per_second": 8.733,
|
15504 |
"eval_steps_per_second": 4.372,
|
15505 |
"step": 2200
|
15506 |
+
},
|
15507 |
+
{
|
15508 |
+
"epoch": 0.10504587703284772,
|
15509 |
+
"grad_norm": 5.450164318084717,
|
15510 |
+
"learning_rate": 0.00019941131358930373,
|
15511 |
+
"loss": 6.1929,
|
15512 |
+
"step": 2201
|
15513 |
+
},
|
15514 |
+
{
|
15515 |
+
"epoch": 0.10509360346493897,
|
15516 |
+
"grad_norm": 5.591408729553223,
|
15517 |
+
"learning_rate": 0.00019941077167953225,
|
15518 |
+
"loss": 6.1391,
|
15519 |
+
"step": 2202
|
15520 |
+
},
|
15521 |
+
{
|
15522 |
+
"epoch": 0.10514132989703022,
|
15523 |
+
"grad_norm": 5.992808818817139,
|
15524 |
+
"learning_rate": 0.00019941022952118747,
|
15525 |
+
"loss": 4.8693,
|
15526 |
+
"step": 2203
|
15527 |
+
},
|
15528 |
+
{
|
15529 |
+
"epoch": 0.10518905632912147,
|
15530 |
+
"grad_norm": 4.80279016494751,
|
15531 |
+
"learning_rate": 0.0001994096871142708,
|
15532 |
+
"loss": 5.2619,
|
15533 |
+
"step": 2204
|
15534 |
+
},
|
15535 |
+
{
|
15536 |
+
"epoch": 0.10523678276121273,
|
15537 |
+
"grad_norm": 5.691819667816162,
|
15538 |
+
"learning_rate": 0.00019940914445878367,
|
15539 |
+
"loss": 5.3011,
|
15540 |
+
"step": 2205
|
15541 |
+
},
|
15542 |
+
{
|
15543 |
+
"epoch": 0.10528450919330398,
|
15544 |
+
"grad_norm": 5.115384578704834,
|
15545 |
+
"learning_rate": 0.00019940860155472732,
|
15546 |
+
"loss": 4.359,
|
15547 |
+
"step": 2206
|
15548 |
+
},
|
15549 |
+
{
|
15550 |
+
"epoch": 0.10533223562539523,
|
15551 |
+
"grad_norm": 5.720730781555176,
|
15552 |
+
"learning_rate": 0.00019940805840210318,
|
15553 |
+
"loss": 5.9681,
|
15554 |
+
"step": 2207
|
15555 |
+
},
|
15556 |
+
{
|
15557 |
+
"epoch": 0.10537996205748648,
|
15558 |
+
"grad_norm": 4.984778881072998,
|
15559 |
+
"learning_rate": 0.00019940751500091257,
|
15560 |
+
"loss": 5.6348,
|
15561 |
+
"step": 2208
|
15562 |
+
},
|
15563 |
+
{
|
15564 |
+
"epoch": 0.10542768848957774,
|
15565 |
+
"grad_norm": 5.787764072418213,
|
15566 |
+
"learning_rate": 0.00019940697135115688,
|
15567 |
+
"loss": 6.2302,
|
15568 |
+
"step": 2209
|
15569 |
+
},
|
15570 |
+
{
|
15571 |
+
"epoch": 0.105475414921669,
|
15572 |
+
"grad_norm": 5.652523040771484,
|
15573 |
+
"learning_rate": 0.00019940642745283746,
|
15574 |
+
"loss": 5.879,
|
15575 |
+
"step": 2210
|
15576 |
+
},
|
15577 |
+
{
|
15578 |
+
"epoch": 0.10552314135376024,
|
15579 |
+
"grad_norm": 7.101466655731201,
|
15580 |
+
"learning_rate": 0.00019940588330595563,
|
15581 |
+
"loss": 7.963,
|
15582 |
+
"step": 2211
|
15583 |
+
},
|
15584 |
+
{
|
15585 |
+
"epoch": 0.1055708677858515,
|
15586 |
+
"grad_norm": 6.2562713623046875,
|
15587 |
+
"learning_rate": 0.00019940533891051282,
|
15588 |
+
"loss": 6.9716,
|
15589 |
+
"step": 2212
|
15590 |
+
},
|
15591 |
+
{
|
15592 |
+
"epoch": 0.10561859421794276,
|
15593 |
+
"grad_norm": 5.532405376434326,
|
15594 |
+
"learning_rate": 0.00019940479426651034,
|
15595 |
+
"loss": 6.9063,
|
15596 |
+
"step": 2213
|
15597 |
+
},
|
15598 |
+
{
|
15599 |
+
"epoch": 0.105666320650034,
|
15600 |
+
"grad_norm": 6.955551624298096,
|
15601 |
+
"learning_rate": 0.00019940424937394958,
|
15602 |
+
"loss": 6.0961,
|
15603 |
+
"step": 2214
|
15604 |
+
},
|
15605 |
+
{
|
15606 |
+
"epoch": 0.10571404708212526,
|
15607 |
+
"grad_norm": 8.936443328857422,
|
15608 |
+
"learning_rate": 0.00019940370423283187,
|
15609 |
+
"loss": 8.8187,
|
15610 |
+
"step": 2215
|
15611 |
+
},
|
15612 |
+
{
|
15613 |
+
"epoch": 0.1057617735142165,
|
15614 |
+
"grad_norm": 5.99800443649292,
|
15615 |
+
"learning_rate": 0.00019940315884315864,
|
15616 |
+
"loss": 6.5717,
|
15617 |
+
"step": 2216
|
15618 |
+
},
|
15619 |
+
{
|
15620 |
+
"epoch": 0.10580949994630777,
|
15621 |
+
"grad_norm": 6.92852783203125,
|
15622 |
+
"learning_rate": 0.00019940261320493114,
|
15623 |
+
"loss": 6.509,
|
15624 |
+
"step": 2217
|
15625 |
+
},
|
15626 |
+
{
|
15627 |
+
"epoch": 0.10585722637839902,
|
15628 |
+
"grad_norm": 6.274646282196045,
|
15629 |
+
"learning_rate": 0.00019940206731815086,
|
15630 |
+
"loss": 7.7704,
|
15631 |
+
"step": 2218
|
15632 |
+
},
|
15633 |
+
{
|
15634 |
+
"epoch": 0.10590495281049027,
|
15635 |
+
"grad_norm": 6.350011348724365,
|
15636 |
+
"learning_rate": 0.00019940152118281908,
|
15637 |
+
"loss": 7.1323,
|
15638 |
+
"step": 2219
|
15639 |
+
},
|
15640 |
+
{
|
15641 |
+
"epoch": 0.10595267924258152,
|
15642 |
+
"grad_norm": 5.615455150604248,
|
15643 |
+
"learning_rate": 0.00019940097479893718,
|
15644 |
+
"loss": 6.8173,
|
15645 |
+
"step": 2220
|
15646 |
+
},
|
15647 |
+
{
|
15648 |
+
"epoch": 0.10600040567467278,
|
15649 |
+
"grad_norm": 6.136722087860107,
|
15650 |
+
"learning_rate": 0.00019940042816650656,
|
15651 |
+
"loss": 7.6495,
|
15652 |
+
"step": 2221
|
15653 |
+
},
|
15654 |
+
{
|
15655 |
+
"epoch": 0.10604813210676403,
|
15656 |
+
"grad_norm": 6.436617374420166,
|
15657 |
+
"learning_rate": 0.00019939988128552853,
|
15658 |
+
"loss": 6.2319,
|
15659 |
+
"step": 2222
|
15660 |
+
},
|
15661 |
+
{
|
15662 |
+
"epoch": 0.10609585853885528,
|
15663 |
+
"grad_norm": 5.910126686096191,
|
15664 |
+
"learning_rate": 0.00019939933415600452,
|
15665 |
+
"loss": 6.2024,
|
15666 |
+
"step": 2223
|
15667 |
+
},
|
15668 |
+
{
|
15669 |
+
"epoch": 0.10614358497094653,
|
15670 |
+
"grad_norm": 6.70936393737793,
|
15671 |
+
"learning_rate": 0.0001993987867779359,
|
15672 |
+
"loss": 7.4669,
|
15673 |
+
"step": 2224
|
15674 |
+
},
|
15675 |
+
{
|
15676 |
+
"epoch": 0.1061913114030378,
|
15677 |
+
"grad_norm": 6.343003273010254,
|
15678 |
+
"learning_rate": 0.00019939823915132395,
|
15679 |
+
"loss": 6.854,
|
15680 |
+
"step": 2225
|
15681 |
+
},
|
15682 |
+
{
|
15683 |
+
"epoch": 0.10623903783512904,
|
15684 |
+
"grad_norm": 5.298125743865967,
|
15685 |
+
"learning_rate": 0.00019939769127617015,
|
15686 |
+
"loss": 5.9439,
|
15687 |
+
"step": 2226
|
15688 |
+
},
|
15689 |
+
{
|
15690 |
+
"epoch": 0.1062867642672203,
|
15691 |
+
"grad_norm": 5.3317108154296875,
|
15692 |
+
"learning_rate": 0.0001993971431524758,
|
15693 |
+
"loss": 5.5212,
|
15694 |
+
"step": 2227
|
15695 |
+
},
|
15696 |
+
{
|
15697 |
+
"epoch": 0.10633449069931154,
|
15698 |
+
"grad_norm": 7.667644500732422,
|
15699 |
+
"learning_rate": 0.00019939659478024227,
|
15700 |
+
"loss": 8.6898,
|
15701 |
+
"step": 2228
|
15702 |
+
},
|
15703 |
+
{
|
15704 |
+
"epoch": 0.10638221713140279,
|
15705 |
+
"grad_norm": 5.852370262145996,
|
15706 |
+
"learning_rate": 0.00019939604615947095,
|
15707 |
+
"loss": 5.7408,
|
15708 |
+
"step": 2229
|
15709 |
+
},
|
15710 |
+
{
|
15711 |
+
"epoch": 0.10642994356349406,
|
15712 |
+
"grad_norm": 5.767739772796631,
|
15713 |
+
"learning_rate": 0.00019939549729016326,
|
15714 |
+
"loss": 6.2188,
|
15715 |
+
"step": 2230
|
15716 |
+
},
|
15717 |
+
{
|
15718 |
+
"epoch": 0.1064776699955853,
|
15719 |
+
"grad_norm": 24.669878005981445,
|
15720 |
+
"learning_rate": 0.00019939494817232048,
|
15721 |
+
"loss": 5.4811,
|
15722 |
+
"step": 2231
|
15723 |
+
},
|
15724 |
+
{
|
15725 |
+
"epoch": 0.10652539642767656,
|
15726 |
+
"grad_norm": 6.24949836730957,
|
15727 |
+
"learning_rate": 0.00019939439880594403,
|
15728 |
+
"loss": 6.7981,
|
15729 |
+
"step": 2232
|
15730 |
+
},
|
15731 |
+
{
|
15732 |
+
"epoch": 0.1065731228597678,
|
15733 |
+
"grad_norm": 4.919851303100586,
|
15734 |
+
"learning_rate": 0.0001993938491910353,
|
15735 |
+
"loss": 5.7305,
|
15736 |
+
"step": 2233
|
15737 |
+
},
|
15738 |
+
{
|
15739 |
+
"epoch": 0.10662084929185907,
|
15740 |
+
"grad_norm": 4.707479000091553,
|
15741 |
+
"learning_rate": 0.00019939329932759566,
|
15742 |
+
"loss": 4.4757,
|
15743 |
+
"step": 2234
|
15744 |
+
},
|
15745 |
+
{
|
15746 |
+
"epoch": 0.10666857572395032,
|
15747 |
+
"grad_norm": 5.307102680206299,
|
15748 |
+
"learning_rate": 0.00019939274921562645,
|
15749 |
+
"loss": 6.0117,
|
15750 |
+
"step": 2235
|
15751 |
+
},
|
15752 |
+
{
|
15753 |
+
"epoch": 0.10671630215604157,
|
15754 |
+
"grad_norm": 5.719334602355957,
|
15755 |
+
"learning_rate": 0.00019939219885512906,
|
15756 |
+
"loss": 5.803,
|
15757 |
+
"step": 2236
|
15758 |
+
},
|
15759 |
+
{
|
15760 |
+
"epoch": 0.10676402858813282,
|
15761 |
+
"grad_norm": 4.507400035858154,
|
15762 |
+
"learning_rate": 0.00019939164824610487,
|
15763 |
+
"loss": 4.8236,
|
15764 |
+
"step": 2237
|
15765 |
+
},
|
15766 |
+
{
|
15767 |
+
"epoch": 0.10681175502022408,
|
15768 |
+
"grad_norm": 5.215142726898193,
|
15769 |
+
"learning_rate": 0.00019939109738855527,
|
15770 |
+
"loss": 6.0824,
|
15771 |
+
"step": 2238
|
15772 |
+
},
|
15773 |
+
{
|
15774 |
+
"epoch": 0.10685948145231533,
|
15775 |
+
"grad_norm": 5.201887130737305,
|
15776 |
+
"learning_rate": 0.00019939054628248163,
|
15777 |
+
"loss": 5.8014,
|
15778 |
+
"step": 2239
|
15779 |
+
},
|
15780 |
+
{
|
15781 |
+
"epoch": 0.10690720788440658,
|
15782 |
+
"grad_norm": 5.401710510253906,
|
15783 |
+
"learning_rate": 0.00019938999492788532,
|
15784 |
+
"loss": 7.0493,
|
15785 |
+
"step": 2240
|
15786 |
+
},
|
15787 |
+
{
|
15788 |
+
"epoch": 0.10695493431649783,
|
15789 |
+
"grad_norm": 6.155087471008301,
|
15790 |
+
"learning_rate": 0.0001993894433247677,
|
15791 |
+
"loss": 7.3769,
|
15792 |
+
"step": 2241
|
15793 |
+
},
|
15794 |
+
{
|
15795 |
+
"epoch": 0.10700266074858909,
|
15796 |
+
"grad_norm": 5.142300128936768,
|
15797 |
+
"learning_rate": 0.0001993888914731302,
|
15798 |
+
"loss": 5.3895,
|
15799 |
+
"step": 2242
|
15800 |
+
},
|
15801 |
+
{
|
15802 |
+
"epoch": 0.10705038718068034,
|
15803 |
+
"grad_norm": 8.86355972290039,
|
15804 |
+
"learning_rate": 0.00019938833937297416,
|
15805 |
+
"loss": 8.0173,
|
15806 |
+
"step": 2243
|
15807 |
+
},
|
15808 |
+
{
|
15809 |
+
"epoch": 0.10709811361277159,
|
15810 |
+
"grad_norm": 5.095742225646973,
|
15811 |
+
"learning_rate": 0.00019938778702430096,
|
15812 |
+
"loss": 5.1925,
|
15813 |
+
"step": 2244
|
15814 |
+
},
|
15815 |
+
{
|
15816 |
+
"epoch": 0.10714584004486284,
|
15817 |
+
"grad_norm": 6.408175945281982,
|
15818 |
+
"learning_rate": 0.00019938723442711203,
|
15819 |
+
"loss": 6.9968,
|
15820 |
+
"step": 2245
|
15821 |
+
},
|
15822 |
+
{
|
15823 |
+
"epoch": 0.1071935664769541,
|
15824 |
+
"grad_norm": 7.479118824005127,
|
15825 |
+
"learning_rate": 0.0001993866815814087,
|
15826 |
+
"loss": 7.813,
|
15827 |
+
"step": 2246
|
15828 |
+
},
|
15829 |
+
{
|
15830 |
+
"epoch": 0.10724129290904535,
|
15831 |
+
"grad_norm": 5.595974922180176,
|
15832 |
+
"learning_rate": 0.00019938612848719237,
|
15833 |
+
"loss": 5.5132,
|
15834 |
+
"step": 2247
|
15835 |
+
},
|
15836 |
+
{
|
15837 |
+
"epoch": 0.1072890193411366,
|
15838 |
+
"grad_norm": 5.417313575744629,
|
15839 |
+
"learning_rate": 0.0001993855751444644,
|
15840 |
+
"loss": 5.7089,
|
15841 |
+
"step": 2248
|
15842 |
+
},
|
15843 |
+
{
|
15844 |
+
"epoch": 0.10733674577322785,
|
15845 |
+
"grad_norm": 5.91009521484375,
|
15846 |
+
"learning_rate": 0.0001993850215532262,
|
15847 |
+
"loss": 6.8627,
|
15848 |
+
"step": 2249
|
15849 |
+
},
|
15850 |
+
{
|
15851 |
+
"epoch": 0.10738447220531912,
|
15852 |
+
"grad_norm": 6.116936683654785,
|
15853 |
+
"learning_rate": 0.00019938446771347915,
|
15854 |
+
"loss": 7.3076,
|
15855 |
+
"step": 2250
|
15856 |
+
},
|
15857 |
+
{
|
15858 |
+
"epoch": 0.10743219863741037,
|
15859 |
+
"grad_norm": 5.593759059906006,
|
15860 |
+
"learning_rate": 0.00019938391362522468,
|
15861 |
+
"loss": 6.352,
|
15862 |
+
"step": 2251
|
15863 |
+
},
|
15864 |
+
{
|
15865 |
+
"epoch": 0.10747992506950162,
|
15866 |
+
"grad_norm": 6.780364513397217,
|
15867 |
+
"learning_rate": 0.00019938335928846408,
|
15868 |
+
"loss": 7.6296,
|
15869 |
+
"step": 2252
|
15870 |
+
},
|
15871 |
+
{
|
15872 |
+
"epoch": 0.10752765150159287,
|
15873 |
+
"grad_norm": 7.030237674713135,
|
15874 |
+
"learning_rate": 0.00019938280470319878,
|
15875 |
+
"loss": 7.8474,
|
15876 |
+
"step": 2253
|
15877 |
+
},
|
15878 |
+
{
|
15879 |
+
"epoch": 0.10757537793368412,
|
15880 |
+
"grad_norm": 8.07816219329834,
|
15881 |
+
"learning_rate": 0.00019938224986943022,
|
15882 |
+
"loss": 7.5247,
|
15883 |
+
"step": 2254
|
15884 |
+
},
|
15885 |
+
{
|
15886 |
+
"epoch": 0.10762310436577538,
|
15887 |
+
"grad_norm": 7.469743728637695,
|
15888 |
+
"learning_rate": 0.0001993816947871597,
|
15889 |
+
"loss": 8.2702,
|
15890 |
+
"step": 2255
|
15891 |
+
},
|
15892 |
+
{
|
15893 |
+
"epoch": 0.10767083079786663,
|
15894 |
+
"grad_norm": 5.907149791717529,
|
15895 |
+
"learning_rate": 0.00019938113945638865,
|
15896 |
+
"loss": 6.131,
|
15897 |
+
"step": 2256
|
15898 |
+
},
|
15899 |
+
{
|
15900 |
+
"epoch": 0.10771855722995788,
|
15901 |
+
"grad_norm": 6.085369110107422,
|
15902 |
+
"learning_rate": 0.00019938058387711845,
|
15903 |
+
"loss": 5.569,
|
15904 |
+
"step": 2257
|
15905 |
+
},
|
15906 |
+
{
|
15907 |
+
"epoch": 0.10776628366204913,
|
15908 |
+
"grad_norm": 5.916553497314453,
|
15909 |
+
"learning_rate": 0.00019938002804935053,
|
15910 |
+
"loss": 6.8827,
|
15911 |
+
"step": 2258
|
15912 |
+
},
|
15913 |
+
{
|
15914 |
+
"epoch": 0.10781401009414039,
|
15915 |
+
"grad_norm": 5.946355819702148,
|
15916 |
+
"learning_rate": 0.00019937947197308623,
|
15917 |
+
"loss": 6.3029,
|
15918 |
+
"step": 2259
|
15919 |
+
},
|
15920 |
+
{
|
15921 |
+
"epoch": 0.10786173652623164,
|
15922 |
+
"grad_norm": 5.058459281921387,
|
15923 |
+
"learning_rate": 0.00019937891564832694,
|
15924 |
+
"loss": 6.0389,
|
15925 |
+
"step": 2260
|
15926 |
+
},
|
15927 |
+
{
|
15928 |
+
"epoch": 0.10790946295832289,
|
15929 |
+
"grad_norm": 5.020669460296631,
|
15930 |
+
"learning_rate": 0.00019937835907507404,
|
15931 |
+
"loss": 7.4807,
|
15932 |
+
"step": 2261
|
15933 |
+
},
|
15934 |
+
{
|
15935 |
+
"epoch": 0.10795718939041414,
|
15936 |
+
"grad_norm": 5.872036933898926,
|
15937 |
+
"learning_rate": 0.000199377802253329,
|
15938 |
+
"loss": 7.3551,
|
15939 |
+
"step": 2262
|
15940 |
+
},
|
15941 |
+
{
|
15942 |
+
"epoch": 0.1080049158225054,
|
15943 |
+
"grad_norm": 6.965964317321777,
|
15944 |
+
"learning_rate": 0.00019937724518309316,
|
15945 |
+
"loss": 5.4708,
|
15946 |
+
"step": 2263
|
15947 |
+
},
|
15948 |
+
{
|
15949 |
+
"epoch": 0.10805264225459665,
|
15950 |
+
"grad_norm": 5.607543468475342,
|
15951 |
+
"learning_rate": 0.00019937668786436787,
|
15952 |
+
"loss": 7.897,
|
15953 |
+
"step": 2264
|
15954 |
+
},
|
15955 |
+
{
|
15956 |
+
"epoch": 0.1081003686866879,
|
15957 |
+
"grad_norm": 5.302212715148926,
|
15958 |
+
"learning_rate": 0.0001993761302971546,
|
15959 |
+
"loss": 5.4608,
|
15960 |
+
"step": 2265
|
15961 |
+
},
|
15962 |
+
{
|
15963 |
+
"epoch": 0.10814809511877915,
|
15964 |
+
"grad_norm": 4.950501918792725,
|
15965 |
+
"learning_rate": 0.00019937557248145472,
|
15966 |
+
"loss": 5.9577,
|
15967 |
+
"step": 2266
|
15968 |
+
},
|
15969 |
+
{
|
15970 |
+
"epoch": 0.10819582155087042,
|
15971 |
+
"grad_norm": 6.807582855224609,
|
15972 |
+
"learning_rate": 0.0001993750144172696,
|
15973 |
+
"loss": 9.1126,
|
15974 |
+
"step": 2267
|
15975 |
+
},
|
15976 |
+
{
|
15977 |
+
"epoch": 0.10824354798296167,
|
15978 |
+
"grad_norm": 4.571179389953613,
|
15979 |
+
"learning_rate": 0.00019937445610460067,
|
15980 |
+
"loss": 5.1932,
|
15981 |
+
"step": 2268
|
15982 |
+
},
|
15983 |
+
{
|
15984 |
+
"epoch": 0.10829127441505292,
|
15985 |
+
"grad_norm": 6.727152347564697,
|
15986 |
+
"learning_rate": 0.0001993738975434493,
|
15987 |
+
"loss": 7.5464,
|
15988 |
+
"step": 2269
|
15989 |
+
},
|
15990 |
+
{
|
15991 |
+
"epoch": 0.10833900084714417,
|
15992 |
+
"grad_norm": 6.7499871253967285,
|
15993 |
+
"learning_rate": 0.00019937333873381685,
|
15994 |
+
"loss": 7.0233,
|
15995 |
+
"step": 2270
|
15996 |
+
},
|
15997 |
+
{
|
15998 |
+
"epoch": 0.10838672727923543,
|
15999 |
+
"grad_norm": 6.338409900665283,
|
16000 |
+
"learning_rate": 0.0001993727796757048,
|
16001 |
+
"loss": 7.9402,
|
16002 |
+
"step": 2271
|
16003 |
+
},
|
16004 |
+
{
|
16005 |
+
"epoch": 0.10843445371132668,
|
16006 |
+
"grad_norm": 5.187378406524658,
|
16007 |
+
"learning_rate": 0.0001993722203691145,
|
16008 |
+
"loss": 6.3124,
|
16009 |
+
"step": 2272
|
16010 |
+
},
|
16011 |
+
{
|
16012 |
+
"epoch": 0.10848218014341793,
|
16013 |
+
"grad_norm": 6.8389997482299805,
|
16014 |
+
"learning_rate": 0.00019937166081404736,
|
16015 |
+
"loss": 5.4769,
|
16016 |
+
"step": 2273
|
16017 |
+
},
|
16018 |
+
{
|
16019 |
+
"epoch": 0.10852990657550918,
|
16020 |
+
"grad_norm": 6.336965560913086,
|
16021 |
+
"learning_rate": 0.00019937110101050478,
|
16022 |
+
"loss": 7.4738,
|
16023 |
+
"step": 2274
|
16024 |
+
},
|
16025 |
+
{
|
16026 |
+
"epoch": 0.10857763300760044,
|
16027 |
+
"grad_norm": 5.328413963317871,
|
16028 |
+
"learning_rate": 0.00019937054095848814,
|
16029 |
+
"loss": 5.5843,
|
16030 |
+
"step": 2275
|
16031 |
+
},
|
16032 |
+
{
|
16033 |
+
"epoch": 0.10862535943969169,
|
16034 |
+
"grad_norm": 6.723316669464111,
|
16035 |
+
"learning_rate": 0.0001993699806579989,
|
16036 |
+
"loss": 7.3747,
|
16037 |
+
"step": 2276
|
16038 |
+
},
|
16039 |
+
{
|
16040 |
+
"epoch": 0.10867308587178294,
|
16041 |
+
"grad_norm": 8.220144271850586,
|
16042 |
+
"learning_rate": 0.00019936942010903837,
|
16043 |
+
"loss": 7.368,
|
16044 |
+
"step": 2277
|
16045 |
+
},
|
16046 |
+
{
|
16047 |
+
"epoch": 0.10872081230387419,
|
16048 |
+
"grad_norm": 5.896597862243652,
|
16049 |
+
"learning_rate": 0.00019936885931160801,
|
16050 |
+
"loss": 7.4897,
|
16051 |
+
"step": 2278
|
16052 |
+
},
|
16053 |
+
{
|
16054 |
+
"epoch": 0.10876853873596544,
|
16055 |
+
"grad_norm": 5.127669334411621,
|
16056 |
+
"learning_rate": 0.00019936829826570923,
|
16057 |
+
"loss": 5.2628,
|
16058 |
+
"step": 2279
|
16059 |
+
},
|
16060 |
+
{
|
16061 |
+
"epoch": 0.1088162651680567,
|
16062 |
+
"grad_norm": 5.313838481903076,
|
16063 |
+
"learning_rate": 0.0001993677369713434,
|
16064 |
+
"loss": 5.643,
|
16065 |
+
"step": 2280
|
16066 |
+
},
|
16067 |
+
{
|
16068 |
+
"epoch": 0.10886399160014795,
|
16069 |
+
"grad_norm": 5.881006240844727,
|
16070 |
+
"learning_rate": 0.00019936717542851199,
|
16071 |
+
"loss": 6.6248,
|
16072 |
+
"step": 2281
|
16073 |
+
},
|
16074 |
+
{
|
16075 |
+
"epoch": 0.1089117180322392,
|
16076 |
+
"grad_norm": 4.9843244552612305,
|
16077 |
+
"learning_rate": 0.00019936661363721626,
|
16078 |
+
"loss": 5.4418,
|
16079 |
+
"step": 2282
|
16080 |
+
},
|
16081 |
+
{
|
16082 |
+
"epoch": 0.10895944446433045,
|
16083 |
+
"grad_norm": 6.7661943435668945,
|
16084 |
+
"learning_rate": 0.00019936605159745778,
|
16085 |
+
"loss": 7.1359,
|
16086 |
+
"step": 2283
|
16087 |
+
},
|
16088 |
+
{
|
16089 |
+
"epoch": 0.10900717089642172,
|
16090 |
+
"grad_norm": 4.885817527770996,
|
16091 |
+
"learning_rate": 0.00019936548930923786,
|
16092 |
+
"loss": 6.0781,
|
16093 |
+
"step": 2284
|
16094 |
+
},
|
16095 |
+
{
|
16096 |
+
"epoch": 0.10905489732851296,
|
16097 |
+
"grad_norm": 6.349702835083008,
|
16098 |
+
"learning_rate": 0.00019936492677255792,
|
16099 |
+
"loss": 6.7283,
|
16100 |
+
"step": 2285
|
16101 |
+
},
|
16102 |
+
{
|
16103 |
+
"epoch": 0.10910262376060421,
|
16104 |
+
"grad_norm": 7.6741461753845215,
|
16105 |
+
"learning_rate": 0.00019936436398741939,
|
16106 |
+
"loss": 6.876,
|
16107 |
+
"step": 2286
|
16108 |
+
},
|
16109 |
+
{
|
16110 |
+
"epoch": 0.10915035019269546,
|
16111 |
+
"grad_norm": 4.658005714416504,
|
16112 |
+
"learning_rate": 0.00019936380095382365,
|
16113 |
+
"loss": 5.0804,
|
16114 |
+
"step": 2287
|
16115 |
+
},
|
16116 |
+
{
|
16117 |
+
"epoch": 0.10919807662478673,
|
16118 |
+
"grad_norm": 6.007474899291992,
|
16119 |
+
"learning_rate": 0.00019936323767177214,
|
16120 |
+
"loss": 5.9884,
|
16121 |
+
"step": 2288
|
16122 |
+
},
|
16123 |
+
{
|
16124 |
+
"epoch": 0.10924580305687798,
|
16125 |
+
"grad_norm": 7.069979190826416,
|
16126 |
+
"learning_rate": 0.00019936267414126621,
|
16127 |
+
"loss": 5.87,
|
16128 |
+
"step": 2289
|
16129 |
+
},
|
16130 |
+
{
|
16131 |
+
"epoch": 0.10929352948896923,
|
16132 |
+
"grad_norm": 7.070275783538818,
|
16133 |
+
"learning_rate": 0.00019936211036230734,
|
16134 |
+
"loss": 7.335,
|
16135 |
+
"step": 2290
|
16136 |
+
},
|
16137 |
+
{
|
16138 |
+
"epoch": 0.10934125592106048,
|
16139 |
+
"grad_norm": 5.51244592666626,
|
16140 |
+
"learning_rate": 0.0001993615463348969,
|
16141 |
+
"loss": 5.12,
|
16142 |
+
"step": 2291
|
16143 |
+
},
|
16144 |
+
{
|
16145 |
+
"epoch": 0.10938898235315174,
|
16146 |
+
"grad_norm": 6.348170757293701,
|
16147 |
+
"learning_rate": 0.0001993609820590363,
|
16148 |
+
"loss": 6.1544,
|
16149 |
+
"step": 2292
|
16150 |
+
},
|
16151 |
+
{
|
16152 |
+
"epoch": 0.10943670878524299,
|
16153 |
+
"grad_norm": 7.591483116149902,
|
16154 |
+
"learning_rate": 0.00019936041753472697,
|
16155 |
+
"loss": 7.2094,
|
16156 |
+
"step": 2293
|
16157 |
+
},
|
16158 |
+
{
|
16159 |
+
"epoch": 0.10948443521733424,
|
16160 |
+
"grad_norm": 6.927452564239502,
|
16161 |
+
"learning_rate": 0.0001993598527619703,
|
16162 |
+
"loss": 7.4317,
|
16163 |
+
"step": 2294
|
16164 |
+
},
|
16165 |
+
{
|
16166 |
+
"epoch": 0.10953216164942549,
|
16167 |
+
"grad_norm": 6.805290699005127,
|
16168 |
+
"learning_rate": 0.00019935928774076774,
|
16169 |
+
"loss": 7.6707,
|
16170 |
+
"step": 2295
|
16171 |
+
},
|
16172 |
+
{
|
16173 |
+
"epoch": 0.10957988808151675,
|
16174 |
+
"grad_norm": 4.375084400177002,
|
16175 |
+
"learning_rate": 0.00019935872247112065,
|
16176 |
+
"loss": 4.3418,
|
16177 |
+
"step": 2296
|
16178 |
+
},
|
16179 |
+
{
|
16180 |
+
"epoch": 0.109627614513608,
|
16181 |
+
"grad_norm": 5.1622633934021,
|
16182 |
+
"learning_rate": 0.00019935815695303047,
|
16183 |
+
"loss": 6.8221,
|
16184 |
+
"step": 2297
|
16185 |
+
},
|
16186 |
+
{
|
16187 |
+
"epoch": 0.10967534094569925,
|
16188 |
+
"grad_norm": 4.2949066162109375,
|
16189 |
+
"learning_rate": 0.00019935759118649862,
|
16190 |
+
"loss": 6.0277,
|
16191 |
+
"step": 2298
|
16192 |
+
},
|
16193 |
+
{
|
16194 |
+
"epoch": 0.1097230673777905,
|
16195 |
+
"grad_norm": 7.153820037841797,
|
16196 |
+
"learning_rate": 0.00019935702517152648,
|
16197 |
+
"loss": 6.82,
|
16198 |
+
"step": 2299
|
16199 |
+
},
|
16200 |
+
{
|
16201 |
+
"epoch": 0.10977079380988176,
|
16202 |
+
"grad_norm": 5.745052814483643,
|
16203 |
+
"learning_rate": 0.0001993564589081155,
|
16204 |
+
"loss": 6.2553,
|
16205 |
+
"step": 2300
|
16206 |
+
},
|
16207 |
+
{
|
16208 |
+
"epoch": 0.10981852024197301,
|
16209 |
+
"grad_norm": 6.87201452255249,
|
16210 |
+
"learning_rate": 0.00019935589239626712,
|
16211 |
+
"loss": 7.7752,
|
16212 |
+
"step": 2301
|
16213 |
+
},
|
16214 |
+
{
|
16215 |
+
"epoch": 0.10986624667406426,
|
16216 |
+
"grad_norm": 5.777369022369385,
|
16217 |
+
"learning_rate": 0.00019935532563598272,
|
16218 |
+
"loss": 6.3414,
|
16219 |
+
"step": 2302
|
16220 |
+
},
|
16221 |
+
{
|
16222 |
+
"epoch": 0.10991397310615551,
|
16223 |
+
"grad_norm": 7.546662330627441,
|
16224 |
+
"learning_rate": 0.0001993547586272637,
|
16225 |
+
"loss": 8.5535,
|
16226 |
+
"step": 2303
|
16227 |
+
},
|
16228 |
+
{
|
16229 |
+
"epoch": 0.10996169953824676,
|
16230 |
+
"grad_norm": 6.240353584289551,
|
16231 |
+
"learning_rate": 0.00019935419137011152,
|
16232 |
+
"loss": 5.8689,
|
16233 |
+
"step": 2304
|
16234 |
+
},
|
16235 |
+
{
|
16236 |
+
"epoch": 0.11000942597033803,
|
16237 |
+
"grad_norm": 6.683908939361572,
|
16238 |
+
"learning_rate": 0.00019935362386452757,
|
16239 |
+
"loss": 6.5744,
|
16240 |
+
"step": 2305
|
16241 |
+
},
|
16242 |
+
{
|
16243 |
+
"epoch": 0.11005715240242928,
|
16244 |
+
"grad_norm": 6.688528060913086,
|
16245 |
+
"learning_rate": 0.0001993530561105133,
|
16246 |
+
"loss": 6.5041,
|
16247 |
+
"step": 2306
|
16248 |
+
},
|
16249 |
+
{
|
16250 |
+
"epoch": 0.11010487883452053,
|
16251 |
+
"grad_norm": 6.669954776763916,
|
16252 |
+
"learning_rate": 0.00019935248810807007,
|
16253 |
+
"loss": 8.4592,
|
16254 |
+
"step": 2307
|
16255 |
+
},
|
16256 |
+
{
|
16257 |
+
"epoch": 0.11015260526661177,
|
16258 |
+
"grad_norm": 5.248333930969238,
|
16259 |
+
"learning_rate": 0.00019935191985719937,
|
16260 |
+
"loss": 5.163,
|
16261 |
+
"step": 2308
|
16262 |
+
},
|
16263 |
+
{
|
16264 |
+
"epoch": 0.11020033169870304,
|
16265 |
+
"grad_norm": 6.443370342254639,
|
16266 |
+
"learning_rate": 0.00019935135135790258,
|
16267 |
+
"loss": 7.1714,
|
16268 |
+
"step": 2309
|
16269 |
+
},
|
16270 |
+
{
|
16271 |
+
"epoch": 0.11024805813079429,
|
16272 |
+
"grad_norm": 5.921457290649414,
|
16273 |
+
"learning_rate": 0.0001993507826101811,
|
16274 |
+
"loss": 6.3727,
|
16275 |
+
"step": 2310
|
16276 |
+
},
|
16277 |
+
{
|
16278 |
+
"epoch": 0.11029578456288554,
|
16279 |
+
"grad_norm": 5.955488204956055,
|
16280 |
+
"learning_rate": 0.00019935021361403642,
|
16281 |
+
"loss": 6.4379,
|
16282 |
+
"step": 2311
|
16283 |
+
},
|
16284 |
+
{
|
16285 |
+
"epoch": 0.11034351099497679,
|
16286 |
+
"grad_norm": 5.717413902282715,
|
16287 |
+
"learning_rate": 0.00019934964436946987,
|
16288 |
+
"loss": 6.7091,
|
16289 |
+
"step": 2312
|
16290 |
+
},
|
16291 |
+
{
|
16292 |
+
"epoch": 0.11039123742706805,
|
16293 |
+
"grad_norm": 6.291435718536377,
|
16294 |
+
"learning_rate": 0.00019934907487648298,
|
16295 |
+
"loss": 7.9312,
|
16296 |
+
"step": 2313
|
16297 |
+
},
|
16298 |
+
{
|
16299 |
+
"epoch": 0.1104389638591593,
|
16300 |
+
"grad_norm": 6.957582950592041,
|
16301 |
+
"learning_rate": 0.00019934850513507712,
|
16302 |
+
"loss": 8.0919,
|
16303 |
+
"step": 2314
|
16304 |
+
},
|
16305 |
+
{
|
16306 |
+
"epoch": 0.11048669029125055,
|
16307 |
+
"grad_norm": 7.492276191711426,
|
16308 |
+
"learning_rate": 0.00019934793514525371,
|
16309 |
+
"loss": 7.1227,
|
16310 |
+
"step": 2315
|
16311 |
+
},
|
16312 |
+
{
|
16313 |
+
"epoch": 0.1105344167233418,
|
16314 |
+
"grad_norm": 6.058623313903809,
|
16315 |
+
"learning_rate": 0.00019934736490701417,
|
16316 |
+
"loss": 6.6442,
|
16317 |
+
"step": 2316
|
16318 |
+
},
|
16319 |
+
{
|
16320 |
+
"epoch": 0.11058214315543306,
|
16321 |
+
"grad_norm": 7.71964693069458,
|
16322 |
+
"learning_rate": 0.00019934679442035997,
|
16323 |
+
"loss": 8.5892,
|
16324 |
+
"step": 2317
|
16325 |
+
},
|
16326 |
+
{
|
16327 |
+
"epoch": 0.11062986958752431,
|
16328 |
+
"grad_norm": 6.302182197570801,
|
16329 |
+
"learning_rate": 0.00019934622368529244,
|
16330 |
+
"loss": 6.2054,
|
16331 |
+
"step": 2318
|
16332 |
+
},
|
16333 |
+
{
|
16334 |
+
"epoch": 0.11067759601961556,
|
16335 |
+
"grad_norm": 5.68353271484375,
|
16336 |
+
"learning_rate": 0.00019934565270181317,
|
16337 |
+
"loss": 5.8535,
|
16338 |
+
"step": 2319
|
16339 |
+
},
|
16340 |
+
{
|
16341 |
+
"epoch": 0.11072532245170681,
|
16342 |
+
"grad_norm": 13.180729866027832,
|
16343 |
+
"learning_rate": 0.00019934508146992341,
|
16344 |
+
"loss": 5.0452,
|
16345 |
+
"step": 2320
|
16346 |
+
},
|
16347 |
+
{
|
16348 |
+
"epoch": 0.11077304888379808,
|
16349 |
+
"grad_norm": 6.95402193069458,
|
16350 |
+
"learning_rate": 0.00019934450998962468,
|
16351 |
+
"loss": 6.5574,
|
16352 |
+
"step": 2321
|
16353 |
+
},
|
16354 |
+
{
|
16355 |
+
"epoch": 0.11082077531588932,
|
16356 |
+
"grad_norm": 4.98831033706665,
|
16357 |
+
"learning_rate": 0.00019934393826091841,
|
16358 |
+
"loss": 4.8938,
|
16359 |
+
"step": 2322
|
16360 |
+
},
|
16361 |
+
{
|
16362 |
+
"epoch": 0.11086850174798057,
|
16363 |
+
"grad_norm": 6.9891510009765625,
|
16364 |
+
"learning_rate": 0.000199343366283806,
|
16365 |
+
"loss": 5.9829,
|
16366 |
+
"step": 2323
|
16367 |
+
},
|
16368 |
+
{
|
16369 |
+
"epoch": 0.11091622818007182,
|
16370 |
+
"grad_norm": 5.644136905670166,
|
16371 |
+
"learning_rate": 0.00019934279405828893,
|
16372 |
+
"loss": 6.7659,
|
16373 |
+
"step": 2324
|
16374 |
+
},
|
16375 |
+
{
|
16376 |
+
"epoch": 0.11096395461216309,
|
16377 |
+
"grad_norm": 5.159231662750244,
|
16378 |
+
"learning_rate": 0.00019934222158436856,
|
16379 |
+
"loss": 5.5547,
|
16380 |
+
"step": 2325
|
16381 |
+
},
|
16382 |
+
{
|
16383 |
+
"epoch": 0.11101168104425434,
|
16384 |
+
"grad_norm": 5.834488391876221,
|
16385 |
+
"learning_rate": 0.0001993416488620464,
|
16386 |
+
"loss": 7.0846,
|
16387 |
+
"step": 2326
|
16388 |
+
},
|
16389 |
+
{
|
16390 |
+
"epoch": 0.11105940747634559,
|
16391 |
+
"grad_norm": 6.442501544952393,
|
16392 |
+
"learning_rate": 0.0001993410758913238,
|
16393 |
+
"loss": 6.5023,
|
16394 |
+
"step": 2327
|
16395 |
+
},
|
16396 |
+
{
|
16397 |
+
"epoch": 0.11110713390843684,
|
16398 |
+
"grad_norm": 6.493159294128418,
|
16399 |
+
"learning_rate": 0.00019934050267220226,
|
16400 |
+
"loss": 6.8592,
|
16401 |
+
"step": 2328
|
16402 |
+
},
|
16403 |
+
{
|
16404 |
+
"epoch": 0.11115486034052809,
|
16405 |
+
"grad_norm": 5.3274712562561035,
|
16406 |
+
"learning_rate": 0.00019933992920468318,
|
16407 |
+
"loss": 5.79,
|
16408 |
+
"step": 2329
|
16409 |
+
},
|
16410 |
+
{
|
16411 |
+
"epoch": 0.11120258677261935,
|
16412 |
+
"grad_norm": 7.033855438232422,
|
16413 |
+
"learning_rate": 0.00019933935548876802,
|
16414 |
+
"loss": 6.8646,
|
16415 |
+
"step": 2330
|
16416 |
+
},
|
16417 |
+
{
|
16418 |
+
"epoch": 0.1112503132047106,
|
16419 |
+
"grad_norm": 7.091256618499756,
|
16420 |
+
"learning_rate": 0.00019933878152445815,
|
16421 |
+
"loss": 8.0637,
|
16422 |
+
"step": 2331
|
16423 |
+
},
|
16424 |
+
{
|
16425 |
+
"epoch": 0.11129803963680185,
|
16426 |
+
"grad_norm": 5.766811370849609,
|
16427 |
+
"learning_rate": 0.00019933820731175508,
|
16428 |
+
"loss": 6.6315,
|
16429 |
+
"step": 2332
|
16430 |
+
},
|
16431 |
+
{
|
16432 |
+
"epoch": 0.1113457660688931,
|
16433 |
+
"grad_norm": 6.875061511993408,
|
16434 |
+
"learning_rate": 0.0001993376328506602,
|
16435 |
+
"loss": 7.5878,
|
16436 |
+
"step": 2333
|
16437 |
+
},
|
16438 |
+
{
|
16439 |
+
"epoch": 0.11139349250098436,
|
16440 |
+
"grad_norm": 6.680387496948242,
|
16441 |
+
"learning_rate": 0.00019933705814117496,
|
16442 |
+
"loss": 5.9366,
|
16443 |
+
"step": 2334
|
16444 |
+
},
|
16445 |
+
{
|
16446 |
+
"epoch": 0.11144121893307561,
|
16447 |
+
"grad_norm": 6.255440711975098,
|
16448 |
+
"learning_rate": 0.00019933648318330081,
|
16449 |
+
"loss": 6.6987,
|
16450 |
+
"step": 2335
|
16451 |
+
},
|
16452 |
+
{
|
16453 |
+
"epoch": 0.11148894536516686,
|
16454 |
+
"grad_norm": 6.5973711013793945,
|
16455 |
+
"learning_rate": 0.0001993359079770392,
|
16456 |
+
"loss": 6.6335,
|
16457 |
+
"step": 2336
|
16458 |
+
},
|
16459 |
+
{
|
16460 |
+
"epoch": 0.11153667179725811,
|
16461 |
+
"grad_norm": 7.028340816497803,
|
16462 |
+
"learning_rate": 0.00019933533252239151,
|
16463 |
+
"loss": 6.9884,
|
16464 |
+
"step": 2337
|
16465 |
+
},
|
16466 |
+
{
|
16467 |
+
"epoch": 0.11158439822934937,
|
16468 |
+
"grad_norm": 5.201953411102295,
|
16469 |
+
"learning_rate": 0.00019933475681935923,
|
16470 |
+
"loss": 6.3241,
|
16471 |
+
"step": 2338
|
16472 |
+
},
|
16473 |
+
{
|
16474 |
+
"epoch": 0.11163212466144062,
|
16475 |
+
"grad_norm": 6.204700469970703,
|
16476 |
+
"learning_rate": 0.0001993341808679438,
|
16477 |
+
"loss": 7.4391,
|
16478 |
+
"step": 2339
|
16479 |
+
},
|
16480 |
+
{
|
16481 |
+
"epoch": 0.11167985109353187,
|
16482 |
+
"grad_norm": 6.416488170623779,
|
16483 |
+
"learning_rate": 0.0001993336046681466,
|
16484 |
+
"loss": 7.4533,
|
16485 |
+
"step": 2340
|
16486 |
+
},
|
16487 |
+
{
|
16488 |
+
"epoch": 0.11172757752562312,
|
16489 |
+
"grad_norm": 4.511433124542236,
|
16490 |
+
"learning_rate": 0.00019933302821996916,
|
16491 |
+
"loss": 4.9382,
|
16492 |
+
"step": 2341
|
16493 |
+
},
|
16494 |
+
{
|
16495 |
+
"epoch": 0.11177530395771439,
|
16496 |
+
"grad_norm": 4.712959289550781,
|
16497 |
+
"learning_rate": 0.00019933245152341288,
|
16498 |
+
"loss": 4.7801,
|
16499 |
+
"step": 2342
|
16500 |
+
},
|
16501 |
+
{
|
16502 |
+
"epoch": 0.11182303038980564,
|
16503 |
+
"grad_norm": 4.192983627319336,
|
16504 |
+
"learning_rate": 0.00019933187457847918,
|
16505 |
+
"loss": 6.1418,
|
16506 |
+
"step": 2343
|
16507 |
+
},
|
16508 |
+
{
|
16509 |
+
"epoch": 0.11187075682189689,
|
16510 |
+
"grad_norm": 8.828011512756348,
|
16511 |
+
"learning_rate": 0.0001993312973851695,
|
16512 |
+
"loss": 7.6646,
|
16513 |
+
"step": 2344
|
16514 |
+
},
|
16515 |
+
{
|
16516 |
+
"epoch": 0.11191848325398814,
|
16517 |
+
"grad_norm": 8.318449020385742,
|
16518 |
+
"learning_rate": 0.00019933071994348534,
|
16519 |
+
"loss": 9.4005,
|
16520 |
+
"step": 2345
|
16521 |
+
},
|
16522 |
+
{
|
16523 |
+
"epoch": 0.1119662096860794,
|
16524 |
+
"grad_norm": 5.3740386962890625,
|
16525 |
+
"learning_rate": 0.00019933014225342806,
|
16526 |
+
"loss": 5.7934,
|
16527 |
+
"step": 2346
|
16528 |
+
},
|
16529 |
+
{
|
16530 |
+
"epoch": 0.11201393611817065,
|
16531 |
+
"grad_norm": 6.731147766113281,
|
16532 |
+
"learning_rate": 0.0001993295643149992,
|
16533 |
+
"loss": 6.8445,
|
16534 |
+
"step": 2347
|
16535 |
+
},
|
16536 |
+
{
|
16537 |
+
"epoch": 0.1120616625502619,
|
16538 |
+
"grad_norm": 4.421756267547607,
|
16539 |
+
"learning_rate": 0.00019932898612820015,
|
16540 |
+
"loss": 4.8008,
|
16541 |
+
"step": 2348
|
16542 |
+
},
|
16543 |
+
{
|
16544 |
+
"epoch": 0.11210938898235315,
|
16545 |
+
"grad_norm": 8.543274879455566,
|
16546 |
+
"learning_rate": 0.00019932840769303236,
|
16547 |
+
"loss": 5.5817,
|
16548 |
+
"step": 2349
|
16549 |
+
},
|
16550 |
+
{
|
16551 |
+
"epoch": 0.11215711541444441,
|
16552 |
+
"grad_norm": 7.155072212219238,
|
16553 |
+
"learning_rate": 0.00019932782900949726,
|
16554 |
+
"loss": 6.2773,
|
16555 |
+
"step": 2350
|
16556 |
+
},
|
16557 |
+
{
|
16558 |
+
"epoch": 0.11220484184653566,
|
16559 |
+
"grad_norm": 9.269936561584473,
|
16560 |
+
"learning_rate": 0.00019932725007759634,
|
16561 |
+
"loss": 8.2326,
|
16562 |
+
"step": 2351
|
16563 |
+
},
|
16564 |
+
{
|
16565 |
+
"epoch": 0.11225256827862691,
|
16566 |
+
"grad_norm": 7.568048477172852,
|
16567 |
+
"learning_rate": 0.000199326670897331,
|
16568 |
+
"loss": 7.7525,
|
16569 |
+
"step": 2352
|
16570 |
+
},
|
16571 |
+
{
|
16572 |
+
"epoch": 0.11230029471071816,
|
16573 |
+
"grad_norm": 5.6857991218566895,
|
16574 |
+
"learning_rate": 0.00019932609146870272,
|
16575 |
+
"loss": 6.557,
|
16576 |
+
"step": 2353
|
16577 |
+
},
|
16578 |
+
{
|
16579 |
+
"epoch": 0.11234802114280942,
|
16580 |
+
"grad_norm": 8.984227180480957,
|
16581 |
+
"learning_rate": 0.000199325511791713,
|
16582 |
+
"loss": 7.3694,
|
16583 |
+
"step": 2354
|
16584 |
+
},
|
16585 |
+
{
|
16586 |
+
"epoch": 0.11239574757490067,
|
16587 |
+
"grad_norm": 5.334461212158203,
|
16588 |
+
"learning_rate": 0.00019932493186636317,
|
16589 |
+
"loss": 6.5095,
|
16590 |
+
"step": 2355
|
16591 |
+
},
|
16592 |
+
{
|
16593 |
+
"epoch": 0.11244347400699192,
|
16594 |
+
"grad_norm": 4.5501251220703125,
|
16595 |
+
"learning_rate": 0.00019932435169265475,
|
16596 |
+
"loss": 5.0719,
|
16597 |
+
"step": 2356
|
16598 |
+
},
|
16599 |
+
{
|
16600 |
+
"epoch": 0.11249120043908317,
|
16601 |
+
"grad_norm": 6.914137363433838,
|
16602 |
+
"learning_rate": 0.0001993237712705892,
|
16603 |
+
"loss": 8.3949,
|
16604 |
+
"step": 2357
|
16605 |
+
},
|
16606 |
+
{
|
16607 |
+
"epoch": 0.11253892687117442,
|
16608 |
+
"grad_norm": 6.041475296020508,
|
16609 |
+
"learning_rate": 0.00019932319060016792,
|
16610 |
+
"loss": 6.8818,
|
16611 |
+
"step": 2358
|
16612 |
+
},
|
16613 |
+
{
|
16614 |
+
"epoch": 0.11258665330326569,
|
16615 |
+
"grad_norm": 4.7477803230285645,
|
16616 |
+
"learning_rate": 0.0001993226096813924,
|
16617 |
+
"loss": 6.2313,
|
16618 |
+
"step": 2359
|
16619 |
+
},
|
16620 |
+
{
|
16621 |
+
"epoch": 0.11263437973535693,
|
16622 |
+
"grad_norm": 7.208108901977539,
|
16623 |
+
"learning_rate": 0.00019932202851426412,
|
16624 |
+
"loss": 7.2228,
|
16625 |
+
"step": 2360
|
16626 |
+
},
|
16627 |
+
{
|
16628 |
+
"epoch": 0.11268210616744818,
|
16629 |
+
"grad_norm": 6.3531670570373535,
|
16630 |
+
"learning_rate": 0.00019932144709878448,
|
16631 |
+
"loss": 6.8451,
|
16632 |
+
"step": 2361
|
16633 |
+
},
|
16634 |
+
{
|
16635 |
+
"epoch": 0.11272983259953943,
|
16636 |
+
"grad_norm": 5.461609840393066,
|
16637 |
+
"learning_rate": 0.00019932086543495496,
|
16638 |
+
"loss": 7.7151,
|
16639 |
+
"step": 2362
|
16640 |
+
},
|
16641 |
+
{
|
16642 |
+
"epoch": 0.1127775590316307,
|
16643 |
+
"grad_norm": 5.988696575164795,
|
16644 |
+
"learning_rate": 0.00019932028352277702,
|
16645 |
+
"loss": 6.6906,
|
16646 |
+
"step": 2363
|
16647 |
+
},
|
16648 |
+
{
|
16649 |
+
"epoch": 0.11282528546372195,
|
16650 |
+
"grad_norm": 4.52699613571167,
|
16651 |
+
"learning_rate": 0.0001993197013622521,
|
16652 |
+
"loss": 5.4023,
|
16653 |
+
"step": 2364
|
16654 |
+
},
|
16655 |
+
{
|
16656 |
+
"epoch": 0.1128730118958132,
|
16657 |
+
"grad_norm": 6.659121036529541,
|
16658 |
+
"learning_rate": 0.00019931911895338162,
|
16659 |
+
"loss": 6.355,
|
16660 |
+
"step": 2365
|
16661 |
+
},
|
16662 |
+
{
|
16663 |
+
"epoch": 0.11292073832790445,
|
16664 |
+
"grad_norm": 4.363926887512207,
|
16665 |
+
"learning_rate": 0.0001993185362961671,
|
16666 |
+
"loss": 4.5778,
|
16667 |
+
"step": 2366
|
16668 |
+
},
|
16669 |
+
{
|
16670 |
+
"epoch": 0.11296846475999571,
|
16671 |
+
"grad_norm": 5.3649115562438965,
|
16672 |
+
"learning_rate": 0.00019931795339061,
|
16673 |
+
"loss": 4.2701,
|
16674 |
+
"step": 2367
|
16675 |
+
},
|
16676 |
+
{
|
16677 |
+
"epoch": 0.11301619119208696,
|
16678 |
+
"grad_norm": 6.213987827301025,
|
16679 |
+
"learning_rate": 0.00019931737023671172,
|
16680 |
+
"loss": 5.5372,
|
16681 |
+
"step": 2368
|
16682 |
+
},
|
16683 |
+
{
|
16684 |
+
"epoch": 0.11306391762417821,
|
16685 |
+
"grad_norm": 5.593975067138672,
|
16686 |
+
"learning_rate": 0.00019931678683447377,
|
16687 |
+
"loss": 6.3435,
|
16688 |
+
"step": 2369
|
16689 |
+
},
|
16690 |
+
{
|
16691 |
+
"epoch": 0.11311164405626946,
|
16692 |
+
"grad_norm": 4.41213846206665,
|
16693 |
+
"learning_rate": 0.00019931620318389755,
|
16694 |
+
"loss": 5.769,
|
16695 |
+
"step": 2370
|
16696 |
+
},
|
16697 |
+
{
|
16698 |
+
"epoch": 0.11315937048836072,
|
16699 |
+
"grad_norm": 6.6761627197265625,
|
16700 |
+
"learning_rate": 0.00019931561928498458,
|
16701 |
+
"loss": 6.7361,
|
16702 |
+
"step": 2371
|
16703 |
+
},
|
16704 |
+
{
|
16705 |
+
"epoch": 0.11320709692045197,
|
16706 |
+
"grad_norm": 5.262270927429199,
|
16707 |
+
"learning_rate": 0.0001993150351377363,
|
16708 |
+
"loss": 7.0394,
|
16709 |
+
"step": 2372
|
16710 |
+
},
|
16711 |
+
{
|
16712 |
+
"epoch": 0.11325482335254322,
|
16713 |
+
"grad_norm": 5.07050895690918,
|
16714 |
+
"learning_rate": 0.00019931445074215418,
|
16715 |
+
"loss": 5.0472,
|
16716 |
+
"step": 2373
|
16717 |
+
},
|
16718 |
+
{
|
16719 |
+
"epoch": 0.11330254978463447,
|
16720 |
+
"grad_norm": 6.702690124511719,
|
16721 |
+
"learning_rate": 0.00019931386609823966,
|
16722 |
+
"loss": 7.3641,
|
16723 |
+
"step": 2374
|
16724 |
+
},
|
16725 |
+
{
|
16726 |
+
"epoch": 0.11335027621672573,
|
16727 |
+
"grad_norm": 6.738053798675537,
|
16728 |
+
"learning_rate": 0.0001993132812059942,
|
16729 |
+
"loss": 6.3667,
|
16730 |
+
"step": 2375
|
16731 |
+
},
|
16732 |
+
{
|
16733 |
+
"epoch": 0.11339800264881698,
|
16734 |
+
"grad_norm": 6.065093517303467,
|
16735 |
+
"learning_rate": 0.0001993126960654193,
|
16736 |
+
"loss": 6.2547,
|
16737 |
+
"step": 2376
|
16738 |
+
},
|
16739 |
+
{
|
16740 |
+
"epoch": 0.11344572908090823,
|
16741 |
+
"grad_norm": 11.0866060256958,
|
16742 |
+
"learning_rate": 0.00019931211067651634,
|
16743 |
+
"loss": 6.9068,
|
16744 |
+
"step": 2377
|
16745 |
+
},
|
16746 |
+
{
|
16747 |
+
"epoch": 0.11349345551299948,
|
16748 |
+
"grad_norm": 6.7252197265625,
|
16749 |
+
"learning_rate": 0.00019931152503928687,
|
16750 |
+
"loss": 5.4051,
|
16751 |
+
"step": 2378
|
16752 |
+
},
|
16753 |
+
{
|
16754 |
+
"epoch": 0.11354118194509075,
|
16755 |
+
"grad_norm": 6.592766284942627,
|
16756 |
+
"learning_rate": 0.00019931093915373236,
|
16757 |
+
"loss": 5.6256,
|
16758 |
+
"step": 2379
|
16759 |
+
},
|
16760 |
+
{
|
16761 |
+
"epoch": 0.113588908377182,
|
16762 |
+
"grad_norm": 5.081091403961182,
|
16763 |
+
"learning_rate": 0.00019931035301985422,
|
16764 |
+
"loss": 5.7141,
|
16765 |
+
"step": 2380
|
16766 |
+
},
|
16767 |
+
{
|
16768 |
+
"epoch": 0.11363663480927325,
|
16769 |
+
"grad_norm": 4.645683288574219,
|
16770 |
+
"learning_rate": 0.0001993097666376539,
|
16771 |
+
"loss": 5.7581,
|
16772 |
+
"step": 2381
|
16773 |
+
},
|
16774 |
+
{
|
16775 |
+
"epoch": 0.1136843612413645,
|
16776 |
+
"grad_norm": 9.910128593444824,
|
16777 |
+
"learning_rate": 0.00019930918000713292,
|
16778 |
+
"loss": 7.054,
|
16779 |
+
"step": 2382
|
16780 |
+
},
|
16781 |
+
{
|
16782 |
+
"epoch": 0.11373208767345575,
|
16783 |
+
"grad_norm": 5.0828447341918945,
|
16784 |
+
"learning_rate": 0.00019930859312829275,
|
16785 |
+
"loss": 6.108,
|
16786 |
+
"step": 2383
|
16787 |
+
},
|
16788 |
+
{
|
16789 |
+
"epoch": 0.11377981410554701,
|
16790 |
+
"grad_norm": 5.93762731552124,
|
16791 |
+
"learning_rate": 0.0001993080060011348,
|
16792 |
+
"loss": 6.5221,
|
16793 |
+
"step": 2384
|
16794 |
+
},
|
16795 |
+
{
|
16796 |
+
"epoch": 0.11382754053763826,
|
16797 |
+
"grad_norm": 5.379035949707031,
|
16798 |
+
"learning_rate": 0.0001993074186256606,
|
16799 |
+
"loss": 5.5101,
|
16800 |
+
"step": 2385
|
16801 |
+
},
|
16802 |
+
{
|
16803 |
+
"epoch": 0.11387526696972951,
|
16804 |
+
"grad_norm": 7.046672344207764,
|
16805 |
+
"learning_rate": 0.00019930683100187157,
|
16806 |
+
"loss": 6.0654,
|
16807 |
+
"step": 2386
|
16808 |
+
},
|
16809 |
+
{
|
16810 |
+
"epoch": 0.11392299340182076,
|
16811 |
+
"grad_norm": 5.484672546386719,
|
16812 |
+
"learning_rate": 0.00019930624312976925,
|
16813 |
+
"loss": 6.6674,
|
16814 |
+
"step": 2387
|
16815 |
+
},
|
16816 |
+
{
|
16817 |
+
"epoch": 0.11397071983391202,
|
16818 |
+
"grad_norm": 5.460198879241943,
|
16819 |
+
"learning_rate": 0.000199305655009355,
|
16820 |
+
"loss": 6.2622,
|
16821 |
+
"step": 2388
|
16822 |
+
},
|
16823 |
+
{
|
16824 |
+
"epoch": 0.11401844626600327,
|
16825 |
+
"grad_norm": 6.094839096069336,
|
16826 |
+
"learning_rate": 0.0001993050666406304,
|
16827 |
+
"loss": 6.5156,
|
16828 |
+
"step": 2389
|
16829 |
+
},
|
16830 |
+
{
|
16831 |
+
"epoch": 0.11406617269809452,
|
16832 |
+
"grad_norm": 6.646234035491943,
|
16833 |
+
"learning_rate": 0.00019930447802359686,
|
16834 |
+
"loss": 7.9221,
|
16835 |
+
"step": 2390
|
16836 |
+
},
|
16837 |
+
{
|
16838 |
+
"epoch": 0.11411389913018577,
|
16839 |
+
"grad_norm": 7.640484809875488,
|
16840 |
+
"learning_rate": 0.00019930388915825585,
|
16841 |
+
"loss": 8.555,
|
16842 |
+
"step": 2391
|
16843 |
+
},
|
16844 |
+
{
|
16845 |
+
"epoch": 0.11416162556227703,
|
16846 |
+
"grad_norm": 5.273614406585693,
|
16847 |
+
"learning_rate": 0.00019930330004460885,
|
16848 |
+
"loss": 6.1841,
|
16849 |
+
"step": 2392
|
16850 |
+
},
|
16851 |
+
{
|
16852 |
+
"epoch": 0.11420935199436828,
|
16853 |
+
"grad_norm": 5.593787670135498,
|
16854 |
+
"learning_rate": 0.00019930271068265736,
|
16855 |
+
"loss": 5.6689,
|
16856 |
+
"step": 2393
|
16857 |
+
},
|
16858 |
+
{
|
16859 |
+
"epoch": 0.11425707842645953,
|
16860 |
+
"grad_norm": 7.496115684509277,
|
16861 |
+
"learning_rate": 0.00019930212107240286,
|
16862 |
+
"loss": 7.1425,
|
16863 |
+
"step": 2394
|
16864 |
+
},
|
16865 |
+
{
|
16866 |
+
"epoch": 0.11430480485855078,
|
16867 |
+
"grad_norm": 4.734747409820557,
|
16868 |
+
"learning_rate": 0.00019930153121384676,
|
16869 |
+
"loss": 4.4183,
|
16870 |
+
"step": 2395
|
16871 |
+
},
|
16872 |
+
{
|
16873 |
+
"epoch": 0.11435253129064205,
|
16874 |
+
"grad_norm": 5.430137634277344,
|
16875 |
+
"learning_rate": 0.0001993009411069906,
|
16876 |
+
"loss": 5.1923,
|
16877 |
+
"step": 2396
|
16878 |
+
},
|
16879 |
+
{
|
16880 |
+
"epoch": 0.1144002577227333,
|
16881 |
+
"grad_norm": 5.818052291870117,
|
16882 |
+
"learning_rate": 0.0001993003507518358,
|
16883 |
+
"loss": 6.9959,
|
16884 |
+
"step": 2397
|
16885 |
+
},
|
16886 |
+
{
|
16887 |
+
"epoch": 0.11444798415482454,
|
16888 |
+
"grad_norm": 8.082784652709961,
|
16889 |
+
"learning_rate": 0.00019929976014838389,
|
16890 |
+
"loss": 6.6515,
|
16891 |
+
"step": 2398
|
16892 |
+
},
|
16893 |
+
{
|
16894 |
+
"epoch": 0.1144957105869158,
|
16895 |
+
"grad_norm": 5.44492244720459,
|
16896 |
+
"learning_rate": 0.0001992991692966363,
|
16897 |
+
"loss": 5.7843,
|
16898 |
+
"step": 2399
|
16899 |
+
},
|
16900 |
+
{
|
16901 |
+
"epoch": 0.11454343701900706,
|
16902 |
+
"grad_norm": 6.5203447341918945,
|
16903 |
+
"learning_rate": 0.00019929857819659454,
|
16904 |
+
"loss": 6.4552,
|
16905 |
+
"step": 2400
|
16906 |
+
},
|
16907 |
+
{
|
16908 |
+
"epoch": 0.11454343701900706,
|
16909 |
+
"eval_loss": 1.6121469736099243,
|
16910 |
+
"eval_runtime": 96.4975,
|
16911 |
+
"eval_samples_per_second": 8.736,
|
16912 |
+
"eval_steps_per_second": 4.373,
|
16913 |
+
"step": 2400
|
16914 |
}
|
16915 |
],
|
16916 |
"logging_steps": 1,
|
|
|
16925 |
"early_stopping_threshold": 0.0
|
16926 |
},
|
16927 |
"attributes": {
|
16928 |
+
"early_stopping_patience_counter": 1
|
16929 |
}
|
16930 |
},
|
16931 |
"TrainerControl": {
|
|
|
16939 |
"attributes": {}
|
16940 |
}
|
16941 |
},
|
16942 |
+
"total_flos": 2.225353217015808e+17,
|
16943 |
"train_batch_size": 2,
|
16944 |
"trial_name": null,
|
16945 |
"trial_params": null
|