Training in progress, step 1000, 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 335604696
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a3a650650762ad6d1017be9f7bd95ea05f6b6a0917941b2a0e535d61b3168829
|
3 |
size 335604696
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 170920532
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ce86e3c4e159ad015bd74dd719eb9ddd5b917b8187f6db5fd8613b47e6e01f39
|
3 |
size 170920532
|
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:bf0da1d3a2604b7650e9974ff3f4fb732d1e165970e5bd7cd09cad712f5bab99
|
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 sha256:4077036d99500a708f700f75da24d51b5300e184ad35fda49dc5a4df5596cca2
|
3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
-
"best_metric": 0.
|
3 |
-
"best_model_checkpoint": "miner_id_24_1/checkpoint-
|
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,
|
@@ -5639,6 +5639,1414 @@
|
|
5639 |
"eval_samples_per_second": 8.793,
|
5640 |
"eval_steps_per_second": 2.198,
|
5641 |
"step": 800
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5642 |
}
|
5643 |
],
|
5644 |
"logging_steps": 1,
|
@@ -5662,12 +7070,12 @@
|
|
5662 |
"should_evaluate": false,
|
5663 |
"should_log": false,
|
5664 |
"should_save": true,
|
5665 |
-
"should_training_stop":
|
5666 |
},
|
5667 |
"attributes": {}
|
5668 |
}
|
5669 |
},
|
5670 |
-
"total_flos":
|
5671 |
"train_batch_size": 8,
|
5672 |
"trial_name": null,
|
5673 |
"trial_params": null
|
|
|
1 |
{
|
2 |
+
"best_metric": 0.22850316762924194,
|
3 |
+
"best_model_checkpoint": "miner_id_24_1/checkpoint-1000",
|
4 |
+
"epoch": 0.5092297899427116,
|
5 |
"eval_steps": 200,
|
6 |
+
"global_step": 1000,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
5639 |
"eval_samples_per_second": 8.793,
|
5640 |
"eval_steps_per_second": 2.198,
|
5641 |
"step": 800
|
5642 |
+
},
|
5643 |
+
{
|
5644 |
+
"epoch": 0.40789306174411205,
|
5645 |
+
"grad_norm": 4.510726451873779,
|
5646 |
+
"learning_rate": 9.642615503142926e-06,
|
5647 |
+
"loss": 0.7091,
|
5648 |
+
"step": 801
|
5649 |
+
},
|
5650 |
+
{
|
5651 |
+
"epoch": 0.4084022915340547,
|
5652 |
+
"grad_norm": 4.7646942138671875,
|
5653 |
+
"learning_rate": 9.549150281252633e-06,
|
5654 |
+
"loss": 0.4744,
|
5655 |
+
"step": 802
|
5656 |
+
},
|
5657 |
+
{
|
5658 |
+
"epoch": 0.40891152132399744,
|
5659 |
+
"grad_norm": 6.916759014129639,
|
5660 |
+
"learning_rate": 9.456092398969902e-06,
|
5661 |
+
"loss": 0.6527,
|
5662 |
+
"step": 803
|
5663 |
+
},
|
5664 |
+
{
|
5665 |
+
"epoch": 0.40942075111394016,
|
5666 |
+
"grad_norm": 4.222986221313477,
|
5667 |
+
"learning_rate": 9.363442793386606e-06,
|
5668 |
+
"loss": 0.3524,
|
5669 |
+
"step": 804
|
5670 |
+
},
|
5671 |
+
{
|
5672 |
+
"epoch": 0.4099299809038829,
|
5673 |
+
"grad_norm": 8.524075508117676,
|
5674 |
+
"learning_rate": 9.271202397483215e-06,
|
5675 |
+
"loss": 0.9119,
|
5676 |
+
"step": 805
|
5677 |
+
},
|
5678 |
+
{
|
5679 |
+
"epoch": 0.4104392106938256,
|
5680 |
+
"grad_norm": 12.714594841003418,
|
5681 |
+
"learning_rate": 9.179372140119525e-06,
|
5682 |
+
"loss": 0.4737,
|
5683 |
+
"step": 806
|
5684 |
+
},
|
5685 |
+
{
|
5686 |
+
"epoch": 0.4109484404837683,
|
5687 |
+
"grad_norm": 8.219633102416992,
|
5688 |
+
"learning_rate": 9.087952946025175e-06,
|
5689 |
+
"loss": 0.4039,
|
5690 |
+
"step": 807
|
5691 |
+
},
|
5692 |
+
{
|
5693 |
+
"epoch": 0.411457670273711,
|
5694 |
+
"grad_norm": 11.758675575256348,
|
5695 |
+
"learning_rate": 8.996945735790447e-06,
|
5696 |
+
"loss": 0.8365,
|
5697 |
+
"step": 808
|
5698 |
+
},
|
5699 |
+
{
|
5700 |
+
"epoch": 0.4119669000636537,
|
5701 |
+
"grad_norm": 10.59483528137207,
|
5702 |
+
"learning_rate": 8.906351425856952e-06,
|
5703 |
+
"loss": 1.0241,
|
5704 |
+
"step": 809
|
5705 |
+
},
|
5706 |
+
{
|
5707 |
+
"epoch": 0.41247612985359644,
|
5708 |
+
"grad_norm": 5.28472375869751,
|
5709 |
+
"learning_rate": 8.816170928508365e-06,
|
5710 |
+
"loss": 0.5595,
|
5711 |
+
"step": 810
|
5712 |
+
},
|
5713 |
+
{
|
5714 |
+
"epoch": 0.41298535964353916,
|
5715 |
+
"grad_norm": 8.57630729675293,
|
5716 |
+
"learning_rate": 8.7264051518613e-06,
|
5717 |
+
"loss": 0.5879,
|
5718 |
+
"step": 811
|
5719 |
+
},
|
5720 |
+
{
|
5721 |
+
"epoch": 0.4134945894334819,
|
5722 |
+
"grad_norm": 8.637681007385254,
|
5723 |
+
"learning_rate": 8.637054999856148e-06,
|
5724 |
+
"loss": 0.8988,
|
5725 |
+
"step": 812
|
5726 |
+
},
|
5727 |
+
{
|
5728 |
+
"epoch": 0.41400381922342455,
|
5729 |
+
"grad_norm": 1.3487462997436523,
|
5730 |
+
"learning_rate": 8.548121372247918e-06,
|
5731 |
+
"loss": 0.1034,
|
5732 |
+
"step": 813
|
5733 |
+
},
|
5734 |
+
{
|
5735 |
+
"epoch": 0.4145130490133673,
|
5736 |
+
"grad_norm": 0.9683374762535095,
|
5737 |
+
"learning_rate": 8.459605164597267e-06,
|
5738 |
+
"loss": 0.0647,
|
5739 |
+
"step": 814
|
5740 |
+
},
|
5741 |
+
{
|
5742 |
+
"epoch": 0.41502227880331,
|
5743 |
+
"grad_norm": 0.556820273399353,
|
5744 |
+
"learning_rate": 8.371507268261437e-06,
|
5745 |
+
"loss": 0.0749,
|
5746 |
+
"step": 815
|
5747 |
+
},
|
5748 |
+
{
|
5749 |
+
"epoch": 0.4155315085932527,
|
5750 |
+
"grad_norm": 2.4960896968841553,
|
5751 |
+
"learning_rate": 8.283828570385238e-06,
|
5752 |
+
"loss": 0.1691,
|
5753 |
+
"step": 816
|
5754 |
+
},
|
5755 |
+
{
|
5756 |
+
"epoch": 0.41604073838319544,
|
5757 |
+
"grad_norm": 1.6859462261199951,
|
5758 |
+
"learning_rate": 8.196569953892202e-06,
|
5759 |
+
"loss": 0.1085,
|
5760 |
+
"step": 817
|
5761 |
+
},
|
5762 |
+
{
|
5763 |
+
"epoch": 0.4165499681731381,
|
5764 |
+
"grad_norm": 3.0034708976745605,
|
5765 |
+
"learning_rate": 8.109732297475635e-06,
|
5766 |
+
"loss": 0.0917,
|
5767 |
+
"step": 818
|
5768 |
+
},
|
5769 |
+
{
|
5770 |
+
"epoch": 0.41705919796308083,
|
5771 |
+
"grad_norm": 1.4022327661514282,
|
5772 |
+
"learning_rate": 8.023316475589754e-06,
|
5773 |
+
"loss": 0.1208,
|
5774 |
+
"step": 819
|
5775 |
+
},
|
5776 |
+
{
|
5777 |
+
"epoch": 0.41756842775302355,
|
5778 |
+
"grad_norm": 0.8047292232513428,
|
5779 |
+
"learning_rate": 7.937323358440935e-06,
|
5780 |
+
"loss": 0.0582,
|
5781 |
+
"step": 820
|
5782 |
+
},
|
5783 |
+
{
|
5784 |
+
"epoch": 0.4180776575429663,
|
5785 |
+
"grad_norm": 1.763879656791687,
|
5786 |
+
"learning_rate": 7.851753811978924e-06,
|
5787 |
+
"loss": 0.1686,
|
5788 |
+
"step": 821
|
5789 |
+
},
|
5790 |
+
{
|
5791 |
+
"epoch": 0.418586887332909,
|
5792 |
+
"grad_norm": 1.504451036453247,
|
5793 |
+
"learning_rate": 7.766608697888095e-06,
|
5794 |
+
"loss": 0.0718,
|
5795 |
+
"step": 822
|
5796 |
+
},
|
5797 |
+
{
|
5798 |
+
"epoch": 0.41909611712285166,
|
5799 |
+
"grad_norm": 1.7512840032577515,
|
5800 |
+
"learning_rate": 7.681888873578786e-06,
|
5801 |
+
"loss": 0.1577,
|
5802 |
+
"step": 823
|
5803 |
+
},
|
5804 |
+
{
|
5805 |
+
"epoch": 0.4196053469127944,
|
5806 |
+
"grad_norm": 10.731456756591797,
|
5807 |
+
"learning_rate": 7.597595192178702e-06,
|
5808 |
+
"loss": 0.1255,
|
5809 |
+
"step": 824
|
5810 |
+
},
|
5811 |
+
{
|
5812 |
+
"epoch": 0.4201145767027371,
|
5813 |
+
"grad_norm": 4.191404819488525,
|
5814 |
+
"learning_rate": 7.513728502524286e-06,
|
5815 |
+
"loss": 0.1061,
|
5816 |
+
"step": 825
|
5817 |
+
},
|
5818 |
+
{
|
5819 |
+
"epoch": 0.42062380649267983,
|
5820 |
+
"grad_norm": 6.975319862365723,
|
5821 |
+
"learning_rate": 7.430289649152156e-06,
|
5822 |
+
"loss": 0.1258,
|
5823 |
+
"step": 826
|
5824 |
+
},
|
5825 |
+
{
|
5826 |
+
"epoch": 0.42113303628262255,
|
5827 |
+
"grad_norm": 5.663224697113037,
|
5828 |
+
"learning_rate": 7.347279472290647e-06,
|
5829 |
+
"loss": 0.1082,
|
5830 |
+
"step": 827
|
5831 |
+
},
|
5832 |
+
{
|
5833 |
+
"epoch": 0.4216422660725652,
|
5834 |
+
"grad_norm": 6.514274597167969,
|
5835 |
+
"learning_rate": 7.264698807851328e-06,
|
5836 |
+
"loss": 0.1209,
|
5837 |
+
"step": 828
|
5838 |
+
},
|
5839 |
+
{
|
5840 |
+
"epoch": 0.42215149586250794,
|
5841 |
+
"grad_norm": 5.257199764251709,
|
5842 |
+
"learning_rate": 7.182548487420554e-06,
|
5843 |
+
"loss": 0.0995,
|
5844 |
+
"step": 829
|
5845 |
+
},
|
5846 |
+
{
|
5847 |
+
"epoch": 0.42266072565245066,
|
5848 |
+
"grad_norm": 3.9015791416168213,
|
5849 |
+
"learning_rate": 7.100829338251147e-06,
|
5850 |
+
"loss": 0.1605,
|
5851 |
+
"step": 830
|
5852 |
+
},
|
5853 |
+
{
|
5854 |
+
"epoch": 0.4231699554423934,
|
5855 |
+
"grad_norm": 5.840038776397705,
|
5856 |
+
"learning_rate": 7.019542183254046e-06,
|
5857 |
+
"loss": 0.1126,
|
5858 |
+
"step": 831
|
5859 |
+
},
|
5860 |
+
{
|
5861 |
+
"epoch": 0.4236791852323361,
|
5862 |
+
"grad_norm": 11.773760795593262,
|
5863 |
+
"learning_rate": 6.9386878409899715e-06,
|
5864 |
+
"loss": 0.1449,
|
5865 |
+
"step": 832
|
5866 |
+
},
|
5867 |
+
{
|
5868 |
+
"epoch": 0.42418841502227883,
|
5869 |
+
"grad_norm": 6.619400978088379,
|
5870 |
+
"learning_rate": 6.858267125661272e-06,
|
5871 |
+
"loss": 0.1207,
|
5872 |
+
"step": 833
|
5873 |
+
},
|
5874 |
+
{
|
5875 |
+
"epoch": 0.4246976448122215,
|
5876 |
+
"grad_norm": 5.580018520355225,
|
5877 |
+
"learning_rate": 6.778280847103669e-06,
|
5878 |
+
"loss": 0.0902,
|
5879 |
+
"step": 834
|
5880 |
+
},
|
5881 |
+
{
|
5882 |
+
"epoch": 0.4252068746021642,
|
5883 |
+
"grad_norm": 5.762646675109863,
|
5884 |
+
"learning_rate": 6.698729810778065e-06,
|
5885 |
+
"loss": 0.0773,
|
5886 |
+
"step": 835
|
5887 |
+
},
|
5888 |
+
{
|
5889 |
+
"epoch": 0.42571610439210694,
|
5890 |
+
"grad_norm": 5.908243179321289,
|
5891 |
+
"learning_rate": 6.619614817762537e-06,
|
5892 |
+
"loss": 0.0891,
|
5893 |
+
"step": 836
|
5894 |
+
},
|
5895 |
+
{
|
5896 |
+
"epoch": 0.42622533418204966,
|
5897 |
+
"grad_norm": 4.759620666503906,
|
5898 |
+
"learning_rate": 6.540936664744196e-06,
|
5899 |
+
"loss": 0.092,
|
5900 |
+
"step": 837
|
5901 |
+
},
|
5902 |
+
{
|
5903 |
+
"epoch": 0.4267345639719924,
|
5904 |
+
"grad_norm": 3.2303318977355957,
|
5905 |
+
"learning_rate": 6.462696144011149e-06,
|
5906 |
+
"loss": 0.0409,
|
5907 |
+
"step": 838
|
5908 |
+
},
|
5909 |
+
{
|
5910 |
+
"epoch": 0.42724379376193505,
|
5911 |
+
"grad_norm": 4.1941986083984375,
|
5912 |
+
"learning_rate": 6.384894043444567e-06,
|
5913 |
+
"loss": 0.1015,
|
5914 |
+
"step": 839
|
5915 |
+
},
|
5916 |
+
{
|
5917 |
+
"epoch": 0.4277530235518778,
|
5918 |
+
"grad_norm": 2.3727965354919434,
|
5919 |
+
"learning_rate": 6.3075311465107535e-06,
|
5920 |
+
"loss": 0.0168,
|
5921 |
+
"step": 840
|
5922 |
+
},
|
5923 |
+
{
|
5924 |
+
"epoch": 0.4282622533418205,
|
5925 |
+
"grad_norm": 6.094393730163574,
|
5926 |
+
"learning_rate": 6.230608232253227e-06,
|
5927 |
+
"loss": 0.098,
|
5928 |
+
"step": 841
|
5929 |
+
},
|
5930 |
+
{
|
5931 |
+
"epoch": 0.4287714831317632,
|
5932 |
+
"grad_norm": 6.916983127593994,
|
5933 |
+
"learning_rate": 6.154126075284855e-06,
|
5934 |
+
"loss": 0.0278,
|
5935 |
+
"step": 842
|
5936 |
+
},
|
5937 |
+
{
|
5938 |
+
"epoch": 0.42928071292170594,
|
5939 |
+
"grad_norm": 1.7201404571533203,
|
5940 |
+
"learning_rate": 6.078085445780129e-06,
|
5941 |
+
"loss": 0.0032,
|
5942 |
+
"step": 843
|
5943 |
+
},
|
5944 |
+
{
|
5945 |
+
"epoch": 0.4297899427116486,
|
5946 |
+
"grad_norm": 9.973702430725098,
|
5947 |
+
"learning_rate": 6.002487109467347e-06,
|
5948 |
+
"loss": 0.0374,
|
5949 |
+
"step": 844
|
5950 |
+
},
|
5951 |
+
{
|
5952 |
+
"epoch": 0.43029917250159133,
|
5953 |
+
"grad_norm": 1.6437783241271973,
|
5954 |
+
"learning_rate": 5.927331827620903e-06,
|
5955 |
+
"loss": 0.0081,
|
5956 |
+
"step": 845
|
5957 |
+
},
|
5958 |
+
{
|
5959 |
+
"epoch": 0.43080840229153405,
|
5960 |
+
"grad_norm": 4.856413841247559,
|
5961 |
+
"learning_rate": 5.852620357053651e-06,
|
5962 |
+
"loss": 0.0392,
|
5963 |
+
"step": 846
|
5964 |
+
},
|
5965 |
+
{
|
5966 |
+
"epoch": 0.4313176320814768,
|
5967 |
+
"grad_norm": 5.264596939086914,
|
5968 |
+
"learning_rate": 5.778353450109286e-06,
|
5969 |
+
"loss": 0.0289,
|
5970 |
+
"step": 847
|
5971 |
+
},
|
5972 |
+
{
|
5973 |
+
"epoch": 0.4318268618714195,
|
5974 |
+
"grad_norm": 4.01646089553833,
|
5975 |
+
"learning_rate": 5.704531854654721e-06,
|
5976 |
+
"loss": 0.0063,
|
5977 |
+
"step": 848
|
5978 |
+
},
|
5979 |
+
{
|
5980 |
+
"epoch": 0.43233609166136217,
|
5981 |
+
"grad_norm": 5.703643798828125,
|
5982 |
+
"learning_rate": 5.631156314072605e-06,
|
5983 |
+
"loss": 0.0557,
|
5984 |
+
"step": 849
|
5985 |
+
},
|
5986 |
+
{
|
5987 |
+
"epoch": 0.4328453214513049,
|
5988 |
+
"grad_norm": 9.367152214050293,
|
5989 |
+
"learning_rate": 5.558227567253832e-06,
|
5990 |
+
"loss": 0.1552,
|
5991 |
+
"step": 850
|
5992 |
+
},
|
5993 |
+
{
|
5994 |
+
"epoch": 0.4333545512412476,
|
5995 |
+
"grad_norm": 4.262320518493652,
|
5996 |
+
"learning_rate": 5.485746348590048e-06,
|
5997 |
+
"loss": 0.6515,
|
5998 |
+
"step": 851
|
5999 |
+
},
|
6000 |
+
{
|
6001 |
+
"epoch": 0.43386378103119033,
|
6002 |
+
"grad_norm": 5.231907367706299,
|
6003 |
+
"learning_rate": 5.413713387966329e-06,
|
6004 |
+
"loss": 0.5992,
|
6005 |
+
"step": 852
|
6006 |
+
},
|
6007 |
+
{
|
6008 |
+
"epoch": 0.43437301082113305,
|
6009 |
+
"grad_norm": 4.903708457946777,
|
6010 |
+
"learning_rate": 5.34212941075381e-06,
|
6011 |
+
"loss": 0.3601,
|
6012 |
+
"step": 853
|
6013 |
+
},
|
6014 |
+
{
|
6015 |
+
"epoch": 0.4348822406110757,
|
6016 |
+
"grad_norm": 6.984781742095947,
|
6017 |
+
"learning_rate": 5.270995137802315e-06,
|
6018 |
+
"loss": 0.5169,
|
6019 |
+
"step": 854
|
6020 |
+
},
|
6021 |
+
{
|
6022 |
+
"epoch": 0.43539147040101844,
|
6023 |
+
"grad_norm": 7.836650371551514,
|
6024 |
+
"learning_rate": 5.200311285433213e-06,
|
6025 |
+
"loss": 0.6677,
|
6026 |
+
"step": 855
|
6027 |
+
},
|
6028 |
+
{
|
6029 |
+
"epoch": 0.43590070019096117,
|
6030 |
+
"grad_norm": 5.570544719696045,
|
6031 |
+
"learning_rate": 5.13007856543209e-06,
|
6032 |
+
"loss": 0.3495,
|
6033 |
+
"step": 856
|
6034 |
+
},
|
6035 |
+
{
|
6036 |
+
"epoch": 0.4364099299809039,
|
6037 |
+
"grad_norm": 21.38435935974121,
|
6038 |
+
"learning_rate": 5.060297685041659e-06,
|
6039 |
+
"loss": 1.0934,
|
6040 |
+
"step": 857
|
6041 |
+
},
|
6042 |
+
{
|
6043 |
+
"epoch": 0.4369191597708466,
|
6044 |
+
"grad_norm": 11.60948371887207,
|
6045 |
+
"learning_rate": 4.99096934695461e-06,
|
6046 |
+
"loss": 0.8771,
|
6047 |
+
"step": 858
|
6048 |
+
},
|
6049 |
+
{
|
6050 |
+
"epoch": 0.43742838956078933,
|
6051 |
+
"grad_norm": 11.168967247009277,
|
6052 |
+
"learning_rate": 4.922094249306558e-06,
|
6053 |
+
"loss": 0.822,
|
6054 |
+
"step": 859
|
6055 |
+
},
|
6056 |
+
{
|
6057 |
+
"epoch": 0.437937619350732,
|
6058 |
+
"grad_norm": 9.118983268737793,
|
6059 |
+
"learning_rate": 4.853673085668947e-06,
|
6060 |
+
"loss": 0.7252,
|
6061 |
+
"step": 860
|
6062 |
+
},
|
6063 |
+
{
|
6064 |
+
"epoch": 0.4384468491406747,
|
6065 |
+
"grad_norm": 8.641736030578613,
|
6066 |
+
"learning_rate": 4.78570654504214e-06,
|
6067 |
+
"loss": 0.8113,
|
6068 |
+
"step": 861
|
6069 |
+
},
|
6070 |
+
{
|
6071 |
+
"epoch": 0.43895607893061744,
|
6072 |
+
"grad_norm": 7.092012882232666,
|
6073 |
+
"learning_rate": 4.7181953118484556e-06,
|
6074 |
+
"loss": 0.6584,
|
6075 |
+
"step": 862
|
6076 |
+
},
|
6077 |
+
{
|
6078 |
+
"epoch": 0.43946530872056017,
|
6079 |
+
"grad_norm": 1.0570333003997803,
|
6080 |
+
"learning_rate": 4.651140065925269e-06,
|
6081 |
+
"loss": 0.1191,
|
6082 |
+
"step": 863
|
6083 |
+
},
|
6084 |
+
{
|
6085 |
+
"epoch": 0.4399745385105029,
|
6086 |
+
"grad_norm": 2.1120429039001465,
|
6087 |
+
"learning_rate": 4.58454148251814e-06,
|
6088 |
+
"loss": 0.1825,
|
6089 |
+
"step": 864
|
6090 |
+
},
|
6091 |
+
{
|
6092 |
+
"epoch": 0.44048376830044556,
|
6093 |
+
"grad_norm": 0.5101509094238281,
|
6094 |
+
"learning_rate": 4.5184002322740785e-06,
|
6095 |
+
"loss": 0.0776,
|
6096 |
+
"step": 865
|
6097 |
+
},
|
6098 |
+
{
|
6099 |
+
"epoch": 0.4409929980903883,
|
6100 |
+
"grad_norm": 1.4004946947097778,
|
6101 |
+
"learning_rate": 4.452716981234744e-06,
|
6102 |
+
"loss": 0.1078,
|
6103 |
+
"step": 866
|
6104 |
+
},
|
6105 |
+
{
|
6106 |
+
"epoch": 0.441502227880331,
|
6107 |
+
"grad_norm": 1.1376584768295288,
|
6108 |
+
"learning_rate": 4.387492390829734e-06,
|
6109 |
+
"loss": 0.0968,
|
6110 |
+
"step": 867
|
6111 |
+
},
|
6112 |
+
{
|
6113 |
+
"epoch": 0.4420114576702737,
|
6114 |
+
"grad_norm": 1.968505620956421,
|
6115 |
+
"learning_rate": 4.322727117869951e-06,
|
6116 |
+
"loss": 0.1204,
|
6117 |
+
"step": 868
|
6118 |
+
},
|
6119 |
+
{
|
6120 |
+
"epoch": 0.44252068746021644,
|
6121 |
+
"grad_norm": 0.5514800548553467,
|
6122 |
+
"learning_rate": 4.258421814540992e-06,
|
6123 |
+
"loss": 0.0586,
|
6124 |
+
"step": 869
|
6125 |
+
},
|
6126 |
+
{
|
6127 |
+
"epoch": 0.4430299172501591,
|
6128 |
+
"grad_norm": 2.0910208225250244,
|
6129 |
+
"learning_rate": 4.19457712839652e-06,
|
6130 |
+
"loss": 0.1634,
|
6131 |
+
"step": 870
|
6132 |
+
},
|
6133 |
+
{
|
6134 |
+
"epoch": 0.44353914704010183,
|
6135 |
+
"grad_norm": 3.2385759353637695,
|
6136 |
+
"learning_rate": 4.131193702351827e-06,
|
6137 |
+
"loss": 0.1482,
|
6138 |
+
"step": 871
|
6139 |
+
},
|
6140 |
+
{
|
6141 |
+
"epoch": 0.44404837683004456,
|
6142 |
+
"grad_norm": 1.8257766962051392,
|
6143 |
+
"learning_rate": 4.068272174677335e-06,
|
6144 |
+
"loss": 0.1311,
|
6145 |
+
"step": 872
|
6146 |
+
},
|
6147 |
+
{
|
6148 |
+
"epoch": 0.4445576066199873,
|
6149 |
+
"grad_norm": 1.4228204488754272,
|
6150 |
+
"learning_rate": 4.005813178992091e-06,
|
6151 |
+
"loss": 0.0839,
|
6152 |
+
"step": 873
|
6153 |
+
},
|
6154 |
+
{
|
6155 |
+
"epoch": 0.44506683640993,
|
6156 |
+
"grad_norm": 14.156405448913574,
|
6157 |
+
"learning_rate": 3.9438173442575e-06,
|
6158 |
+
"loss": 0.1061,
|
6159 |
+
"step": 874
|
6160 |
+
},
|
6161 |
+
{
|
6162 |
+
"epoch": 0.44557606619987267,
|
6163 |
+
"grad_norm": 7.2103047370910645,
|
6164 |
+
"learning_rate": 3.8822852947709375e-06,
|
6165 |
+
"loss": 0.1517,
|
6166 |
+
"step": 875
|
6167 |
+
},
|
6168 |
+
{
|
6169 |
+
"epoch": 0.4460852959898154,
|
6170 |
+
"grad_norm": 4.403718948364258,
|
6171 |
+
"learning_rate": 3.821217650159453e-06,
|
6172 |
+
"loss": 0.1108,
|
6173 |
+
"step": 876
|
6174 |
+
},
|
6175 |
+
{
|
6176 |
+
"epoch": 0.4465945257797581,
|
6177 |
+
"grad_norm": 4.354316711425781,
|
6178 |
+
"learning_rate": 3.760615025373543e-06,
|
6179 |
+
"loss": 0.1314,
|
6180 |
+
"step": 877
|
6181 |
+
},
|
6182 |
+
{
|
6183 |
+
"epoch": 0.44710375556970083,
|
6184 |
+
"grad_norm": 3.967412233352661,
|
6185 |
+
"learning_rate": 3.700478030680987e-06,
|
6186 |
+
"loss": 0.1001,
|
6187 |
+
"step": 878
|
6188 |
+
},
|
6189 |
+
{
|
6190 |
+
"epoch": 0.44761298535964356,
|
6191 |
+
"grad_norm": 4.008363246917725,
|
6192 |
+
"learning_rate": 3.6408072716606346e-06,
|
6193 |
+
"loss": 0.0968,
|
6194 |
+
"step": 879
|
6195 |
+
},
|
6196 |
+
{
|
6197 |
+
"epoch": 0.4481222151495863,
|
6198 |
+
"grad_norm": 6.099944114685059,
|
6199 |
+
"learning_rate": 3.581603349196372e-06,
|
6200 |
+
"loss": 0.1582,
|
6201 |
+
"step": 880
|
6202 |
+
},
|
6203 |
+
{
|
6204 |
+
"epoch": 0.44863144493952894,
|
6205 |
+
"grad_norm": 5.96859073638916,
|
6206 |
+
"learning_rate": 3.522866859471047e-06,
|
6207 |
+
"loss": 0.1445,
|
6208 |
+
"step": 881
|
6209 |
+
},
|
6210 |
+
{
|
6211 |
+
"epoch": 0.44914067472947167,
|
6212 |
+
"grad_norm": 3.3854691982269287,
|
6213 |
+
"learning_rate": 3.4645983939604496e-06,
|
6214 |
+
"loss": 0.0567,
|
6215 |
+
"step": 882
|
6216 |
+
},
|
6217 |
+
{
|
6218 |
+
"epoch": 0.4496499045194144,
|
6219 |
+
"grad_norm": 3.6964199542999268,
|
6220 |
+
"learning_rate": 3.406798539427386e-06,
|
6221 |
+
"loss": 0.1331,
|
6222 |
+
"step": 883
|
6223 |
+
},
|
6224 |
+
{
|
6225 |
+
"epoch": 0.4501591343093571,
|
6226 |
+
"grad_norm": 3.9392178058624268,
|
6227 |
+
"learning_rate": 3.349467877915746e-06,
|
6228 |
+
"loss": 0.1401,
|
6229 |
+
"step": 884
|
6230 |
+
},
|
6231 |
+
{
|
6232 |
+
"epoch": 0.45066836409929983,
|
6233 |
+
"grad_norm": 8.413680076599121,
|
6234 |
+
"learning_rate": 3.2926069867446675e-06,
|
6235 |
+
"loss": 0.0784,
|
6236 |
+
"step": 885
|
6237 |
+
},
|
6238 |
+
{
|
6239 |
+
"epoch": 0.4511775938892425,
|
6240 |
+
"grad_norm": 3.3428969383239746,
|
6241 |
+
"learning_rate": 3.2362164385026706e-06,
|
6242 |
+
"loss": 0.1658,
|
6243 |
+
"step": 886
|
6244 |
+
},
|
6245 |
+
{
|
6246 |
+
"epoch": 0.4516868236791852,
|
6247 |
+
"grad_norm": 4.6566386222839355,
|
6248 |
+
"learning_rate": 3.180296801041971e-06,
|
6249 |
+
"loss": 0.1105,
|
6250 |
+
"step": 887
|
6251 |
+
},
|
6252 |
+
{
|
6253 |
+
"epoch": 0.45219605346912795,
|
6254 |
+
"grad_norm": 4.742708683013916,
|
6255 |
+
"learning_rate": 3.1248486374726883e-06,
|
6256 |
+
"loss": 0.0988,
|
6257 |
+
"step": 888
|
6258 |
+
},
|
6259 |
+
{
|
6260 |
+
"epoch": 0.45270528325907067,
|
6261 |
+
"grad_norm": 5.364091873168945,
|
6262 |
+
"learning_rate": 3.069872506157212e-06,
|
6263 |
+
"loss": 0.0267,
|
6264 |
+
"step": 889
|
6265 |
+
},
|
6266 |
+
{
|
6267 |
+
"epoch": 0.4532145130490134,
|
6268 |
+
"grad_norm": 7.658147811889648,
|
6269 |
+
"learning_rate": 3.0153689607045845e-06,
|
6270 |
+
"loss": 0.1993,
|
6271 |
+
"step": 890
|
6272 |
+
},
|
6273 |
+
{
|
6274 |
+
"epoch": 0.45372374283895606,
|
6275 |
+
"grad_norm": 5.430888652801514,
|
6276 |
+
"learning_rate": 2.961338549964893e-06,
|
6277 |
+
"loss": 0.0843,
|
6278 |
+
"step": 891
|
6279 |
+
},
|
6280 |
+
{
|
6281 |
+
"epoch": 0.4542329726288988,
|
6282 |
+
"grad_norm": 4.619664669036865,
|
6283 |
+
"learning_rate": 2.9077818180237693e-06,
|
6284 |
+
"loss": 0.0777,
|
6285 |
+
"step": 892
|
6286 |
+
},
|
6287 |
+
{
|
6288 |
+
"epoch": 0.4547422024188415,
|
6289 |
+
"grad_norm": 0.4149484932422638,
|
6290 |
+
"learning_rate": 2.8546993041969173e-06,
|
6291 |
+
"loss": 0.0021,
|
6292 |
+
"step": 893
|
6293 |
+
},
|
6294 |
+
{
|
6295 |
+
"epoch": 0.4552514322087842,
|
6296 |
+
"grad_norm": 3.0641720294952393,
|
6297 |
+
"learning_rate": 2.802091543024671e-06,
|
6298 |
+
"loss": 0.0088,
|
6299 |
+
"step": 894
|
6300 |
+
},
|
6301 |
+
{
|
6302 |
+
"epoch": 0.45576066199872695,
|
6303 |
+
"grad_norm": 3.299743175506592,
|
6304 |
+
"learning_rate": 2.7499590642665774e-06,
|
6305 |
+
"loss": 0.0406,
|
6306 |
+
"step": 895
|
6307 |
+
},
|
6308 |
+
{
|
6309 |
+
"epoch": 0.4562698917886696,
|
6310 |
+
"grad_norm": 5.375297546386719,
|
6311 |
+
"learning_rate": 2.6983023928961404e-06,
|
6312 |
+
"loss": 0.0383,
|
6313 |
+
"step": 896
|
6314 |
+
},
|
6315 |
+
{
|
6316 |
+
"epoch": 0.45677912157861233,
|
6317 |
+
"grad_norm": 1.769721508026123,
|
6318 |
+
"learning_rate": 2.647122049095463e-06,
|
6319 |
+
"loss": 0.0077,
|
6320 |
+
"step": 897
|
6321 |
+
},
|
6322 |
+
{
|
6323 |
+
"epoch": 0.45728835136855506,
|
6324 |
+
"grad_norm": 7.771859645843506,
|
6325 |
+
"learning_rate": 2.596418548250029e-06,
|
6326 |
+
"loss": 0.0876,
|
6327 |
+
"step": 898
|
6328 |
+
},
|
6329 |
+
{
|
6330 |
+
"epoch": 0.4577975811584978,
|
6331 |
+
"grad_norm": 5.935695171356201,
|
6332 |
+
"learning_rate": 2.546192400943537e-06,
|
6333 |
+
"loss": 0.0387,
|
6334 |
+
"step": 899
|
6335 |
+
},
|
6336 |
+
{
|
6337 |
+
"epoch": 0.4583068109484405,
|
6338 |
+
"grad_norm": 5.693915367126465,
|
6339 |
+
"learning_rate": 2.496444112952734e-06,
|
6340 |
+
"loss": 0.1489,
|
6341 |
+
"step": 900
|
6342 |
+
},
|
6343 |
+
{
|
6344 |
+
"epoch": 0.45881604073838317,
|
6345 |
+
"grad_norm": 3.7777862548828125,
|
6346 |
+
"learning_rate": 2.4471741852423237e-06,
|
6347 |
+
"loss": 0.6961,
|
6348 |
+
"step": 901
|
6349 |
+
},
|
6350 |
+
{
|
6351 |
+
"epoch": 0.4593252705283259,
|
6352 |
+
"grad_norm": 5.661333084106445,
|
6353 |
+
"learning_rate": 2.3983831139599287e-06,
|
6354 |
+
"loss": 0.6676,
|
6355 |
+
"step": 902
|
6356 |
+
},
|
6357 |
+
{
|
6358 |
+
"epoch": 0.4598345003182686,
|
6359 |
+
"grad_norm": 3.3415160179138184,
|
6360 |
+
"learning_rate": 2.3500713904311024e-06,
|
6361 |
+
"loss": 0.1848,
|
6362 |
+
"step": 903
|
6363 |
+
},
|
6364 |
+
{
|
6365 |
+
"epoch": 0.46034373010821134,
|
6366 |
+
"grad_norm": 9.008618354797363,
|
6367 |
+
"learning_rate": 2.3022395011543686e-06,
|
6368 |
+
"loss": 0.8188,
|
6369 |
+
"step": 904
|
6370 |
+
},
|
6371 |
+
{
|
6372 |
+
"epoch": 0.46085295989815406,
|
6373 |
+
"grad_norm": 6.555294036865234,
|
6374 |
+
"learning_rate": 2.2548879277963064e-06,
|
6375 |
+
"loss": 0.5172,
|
6376 |
+
"step": 905
|
6377 |
+
},
|
6378 |
+
{
|
6379 |
+
"epoch": 0.4613621896880968,
|
6380 |
+
"grad_norm": 15.021227836608887,
|
6381 |
+
"learning_rate": 2.208017147186736e-06,
|
6382 |
+
"loss": 0.6723,
|
6383 |
+
"step": 906
|
6384 |
+
},
|
6385 |
+
{
|
6386 |
+
"epoch": 0.46187141947803945,
|
6387 |
+
"grad_norm": 9.838711738586426,
|
6388 |
+
"learning_rate": 2.161627631313923e-06,
|
6389 |
+
"loss": 0.7112,
|
6390 |
+
"step": 907
|
6391 |
+
},
|
6392 |
+
{
|
6393 |
+
"epoch": 0.46238064926798217,
|
6394 |
+
"grad_norm": 12.633716583251953,
|
6395 |
+
"learning_rate": 2.1157198473197414e-06,
|
6396 |
+
"loss": 0.9927,
|
6397 |
+
"step": 908
|
6398 |
+
},
|
6399 |
+
{
|
6400 |
+
"epoch": 0.4628898790579249,
|
6401 |
+
"grad_norm": 7.977221488952637,
|
6402 |
+
"learning_rate": 2.070294257495081e-06,
|
6403 |
+
"loss": 0.558,
|
6404 |
+
"step": 909
|
6405 |
+
},
|
6406 |
+
{
|
6407 |
+
"epoch": 0.4633991088478676,
|
6408 |
+
"grad_norm": 9.081966400146484,
|
6409 |
+
"learning_rate": 2.0253513192751373e-06,
|
6410 |
+
"loss": 0.5821,
|
6411 |
+
"step": 910
|
6412 |
+
},
|
6413 |
+
{
|
6414 |
+
"epoch": 0.46390833863781034,
|
6415 |
+
"grad_norm": 8.361538887023926,
|
6416 |
+
"learning_rate": 1.9808914852347813e-06,
|
6417 |
+
"loss": 0.9417,
|
6418 |
+
"step": 911
|
6419 |
+
},
|
6420 |
+
{
|
6421 |
+
"epoch": 0.464417568427753,
|
6422 |
+
"grad_norm": 2.8661348819732666,
|
6423 |
+
"learning_rate": 1.9369152030840556e-06,
|
6424 |
+
"loss": 0.3311,
|
6425 |
+
"step": 912
|
6426 |
+
},
|
6427 |
+
{
|
6428 |
+
"epoch": 0.4649267982176957,
|
6429 |
+
"grad_norm": 1.429822564125061,
|
6430 |
+
"learning_rate": 1.8934229156636452e-06,
|
6431 |
+
"loss": 0.1388,
|
6432 |
+
"step": 913
|
6433 |
+
},
|
6434 |
+
{
|
6435 |
+
"epoch": 0.46543602800763845,
|
6436 |
+
"grad_norm": 1.2748867273330688,
|
6437 |
+
"learning_rate": 1.8504150609403858e-06,
|
6438 |
+
"loss": 0.1259,
|
6439 |
+
"step": 914
|
6440 |
+
},
|
6441 |
+
{
|
6442 |
+
"epoch": 0.46594525779758117,
|
6443 |
+
"grad_norm": 0.8820096850395203,
|
6444 |
+
"learning_rate": 1.807892072002898e-06,
|
6445 |
+
"loss": 0.0854,
|
6446 |
+
"step": 915
|
6447 |
+
},
|
6448 |
+
{
|
6449 |
+
"epoch": 0.4664544875875239,
|
6450 |
+
"grad_norm": 2.467775344848633,
|
6451 |
+
"learning_rate": 1.7658543770572189e-06,
|
6452 |
+
"loss": 0.0945,
|
6453 |
+
"step": 916
|
6454 |
+
},
|
6455 |
+
{
|
6456 |
+
"epoch": 0.46696371737746656,
|
6457 |
+
"grad_norm": 4.9203877449035645,
|
6458 |
+
"learning_rate": 1.724302399422456e-06,
|
6459 |
+
"loss": 0.1121,
|
6460 |
+
"step": 917
|
6461 |
+
},
|
6462 |
+
{
|
6463 |
+
"epoch": 0.4674729471674093,
|
6464 |
+
"grad_norm": 0.4738912284374237,
|
6465 |
+
"learning_rate": 1.6832365575265741e-06,
|
6466 |
+
"loss": 0.0745,
|
6467 |
+
"step": 918
|
6468 |
+
},
|
6469 |
+
{
|
6470 |
+
"epoch": 0.467982176957352,
|
6471 |
+
"grad_norm": 1.508836269378662,
|
6472 |
+
"learning_rate": 1.6426572649021476e-06,
|
6473 |
+
"loss": 0.1326,
|
6474 |
+
"step": 919
|
6475 |
+
},
|
6476 |
+
{
|
6477 |
+
"epoch": 0.4684914067472947,
|
6478 |
+
"grad_norm": 1.2223467826843262,
|
6479 |
+
"learning_rate": 1.6025649301821876e-06,
|
6480 |
+
"loss": 0.1413,
|
6481 |
+
"step": 920
|
6482 |
+
},
|
6483 |
+
{
|
6484 |
+
"epoch": 0.46900063653723745,
|
6485 |
+
"grad_norm": 0.6743190288543701,
|
6486 |
+
"learning_rate": 1.5629599570960718e-06,
|
6487 |
+
"loss": 0.0968,
|
6488 |
+
"step": 921
|
6489 |
+
},
|
6490 |
+
{
|
6491 |
+
"epoch": 0.4695098663271801,
|
6492 |
+
"grad_norm": 1.4873098134994507,
|
6493 |
+
"learning_rate": 1.523842744465437e-06,
|
6494 |
+
"loss": 0.102,
|
6495 |
+
"step": 922
|
6496 |
+
},
|
6497 |
+
{
|
6498 |
+
"epoch": 0.47001909611712284,
|
6499 |
+
"grad_norm": 1.1552408933639526,
|
6500 |
+
"learning_rate": 1.4852136862001764e-06,
|
6501 |
+
"loss": 0.0608,
|
6502 |
+
"step": 923
|
6503 |
+
},
|
6504 |
+
{
|
6505 |
+
"epoch": 0.47052832590706556,
|
6506 |
+
"grad_norm": 5.109307289123535,
|
6507 |
+
"learning_rate": 1.4470731712944884e-06,
|
6508 |
+
"loss": 0.0851,
|
6509 |
+
"step": 924
|
6510 |
+
},
|
6511 |
+
{
|
6512 |
+
"epoch": 0.4710375556970083,
|
6513 |
+
"grad_norm": 4.766353130340576,
|
6514 |
+
"learning_rate": 1.4094215838229176e-06,
|
6515 |
+
"loss": 0.1149,
|
6516 |
+
"step": 925
|
6517 |
+
},
|
6518 |
+
{
|
6519 |
+
"epoch": 0.471546785486951,
|
6520 |
+
"grad_norm": 14.194299697875977,
|
6521 |
+
"learning_rate": 1.372259302936546e-06,
|
6522 |
+
"loss": 0.2078,
|
6523 |
+
"step": 926
|
6524 |
+
},
|
6525 |
+
{
|
6526 |
+
"epoch": 0.4720560152768937,
|
6527 |
+
"grad_norm": 4.660562038421631,
|
6528 |
+
"learning_rate": 1.3355867028591208e-06,
|
6529 |
+
"loss": 0.0858,
|
6530 |
+
"step": 927
|
6531 |
+
},
|
6532 |
+
{
|
6533 |
+
"epoch": 0.4725652450668364,
|
6534 |
+
"grad_norm": 10.358325004577637,
|
6535 |
+
"learning_rate": 1.2994041528833266e-06,
|
6536 |
+
"loss": 0.0983,
|
6537 |
+
"step": 928
|
6538 |
+
},
|
6539 |
+
{
|
6540 |
+
"epoch": 0.4730744748567791,
|
6541 |
+
"grad_norm": 6.56409215927124,
|
6542 |
+
"learning_rate": 1.2637120173670358e-06,
|
6543 |
+
"loss": 0.0809,
|
6544 |
+
"step": 929
|
6545 |
+
},
|
6546 |
+
{
|
6547 |
+
"epoch": 0.47358370464672184,
|
6548 |
+
"grad_norm": 4.377633094787598,
|
6549 |
+
"learning_rate": 1.2285106557296477e-06,
|
6550 |
+
"loss": 0.094,
|
6551 |
+
"step": 930
|
6552 |
+
},
|
6553 |
+
{
|
6554 |
+
"epoch": 0.47409293443666456,
|
6555 |
+
"grad_norm": 5.809483051300049,
|
6556 |
+
"learning_rate": 1.1938004224484988e-06,
|
6557 |
+
"loss": 0.1196,
|
6558 |
+
"step": 931
|
6559 |
+
},
|
6560 |
+
{
|
6561 |
+
"epoch": 0.4746021642266073,
|
6562 |
+
"grad_norm": 3.2537131309509277,
|
6563 |
+
"learning_rate": 1.1595816670552428e-06,
|
6564 |
+
"loss": 0.1031,
|
6565 |
+
"step": 932
|
6566 |
+
},
|
6567 |
+
{
|
6568 |
+
"epoch": 0.47511139401654995,
|
6569 |
+
"grad_norm": 6.928438186645508,
|
6570 |
+
"learning_rate": 1.1258547341323699e-06,
|
6571 |
+
"loss": 0.068,
|
6572 |
+
"step": 933
|
6573 |
+
},
|
6574 |
+
{
|
6575 |
+
"epoch": 0.47562062380649267,
|
6576 |
+
"grad_norm": 3.240041971206665,
|
6577 |
+
"learning_rate": 1.0926199633097157e-06,
|
6578 |
+
"loss": 0.0739,
|
6579 |
+
"step": 934
|
6580 |
+
},
|
6581 |
+
{
|
6582 |
+
"epoch": 0.4761298535964354,
|
6583 |
+
"grad_norm": 6.833902835845947,
|
6584 |
+
"learning_rate": 1.0598776892610685e-06,
|
6585 |
+
"loss": 0.0857,
|
6586 |
+
"step": 935
|
6587 |
+
},
|
6588 |
+
{
|
6589 |
+
"epoch": 0.4766390833863781,
|
6590 |
+
"grad_norm": 5.259301662445068,
|
6591 |
+
"learning_rate": 1.02762824170074e-06,
|
6592 |
+
"loss": 0.1534,
|
6593 |
+
"step": 936
|
6594 |
+
},
|
6595 |
+
{
|
6596 |
+
"epoch": 0.47714831317632084,
|
6597 |
+
"grad_norm": 11.154175758361816,
|
6598 |
+
"learning_rate": 9.958719453803278e-07,
|
6599 |
+
"loss": 0.1273,
|
6600 |
+
"step": 937
|
6601 |
+
},
|
6602 |
+
{
|
6603 |
+
"epoch": 0.4776575429662635,
|
6604 |
+
"grad_norm": 2.975895404815674,
|
6605 |
+
"learning_rate": 9.646091200853802e-07,
|
6606 |
+
"loss": 0.015,
|
6607 |
+
"step": 938
|
6608 |
+
},
|
6609 |
+
{
|
6610 |
+
"epoch": 0.4781667727562062,
|
6611 |
+
"grad_norm": 3.766637086868286,
|
6612 |
+
"learning_rate": 9.338400806321978e-07,
|
6613 |
+
"loss": 0.0202,
|
6614 |
+
"step": 939
|
6615 |
+
},
|
6616 |
+
{
|
6617 |
+
"epoch": 0.47867600254614895,
|
6618 |
+
"grad_norm": 2.1836998462677,
|
6619 |
+
"learning_rate": 9.035651368646648e-07,
|
6620 |
+
"loss": 0.0426,
|
6621 |
+
"step": 940
|
6622 |
+
},
|
6623 |
+
{
|
6624 |
+
"epoch": 0.47918523233609167,
|
6625 |
+
"grad_norm": 3.746504306793213,
|
6626 |
+
"learning_rate": 8.737845936511335e-07,
|
6627 |
+
"loss": 0.0458,
|
6628 |
+
"step": 941
|
6629 |
+
},
|
6630 |
+
{
|
6631 |
+
"epoch": 0.4796944621260344,
|
6632 |
+
"grad_norm": 4.881604194641113,
|
6633 |
+
"learning_rate": 8.444987508813451e-07,
|
6634 |
+
"loss": 0.0482,
|
6635 |
+
"step": 942
|
6636 |
+
},
|
6637 |
+
{
|
6638 |
+
"epoch": 0.48020369191597706,
|
6639 |
+
"grad_norm": 15.250972747802734,
|
6640 |
+
"learning_rate": 8.157079034633974e-07,
|
6641 |
+
"loss": 0.0486,
|
6642 |
+
"step": 943
|
6643 |
+
},
|
6644 |
+
{
|
6645 |
+
"epoch": 0.4807129217059198,
|
6646 |
+
"grad_norm": 4.671970844268799,
|
6647 |
+
"learning_rate": 7.874123413208145e-07,
|
6648 |
+
"loss": 0.0408,
|
6649 |
+
"step": 944
|
6650 |
+
},
|
6651 |
+
{
|
6652 |
+
"epoch": 0.4812221514958625,
|
6653 |
+
"grad_norm": 3.882068157196045,
|
6654 |
+
"learning_rate": 7.596123493895991e-07,
|
6655 |
+
"loss": 0.0702,
|
6656 |
+
"step": 945
|
6657 |
+
},
|
6658 |
+
{
|
6659 |
+
"epoch": 0.4817313812858052,
|
6660 |
+
"grad_norm": 3.0033390522003174,
|
6661 |
+
"learning_rate": 7.323082076153509e-07,
|
6662 |
+
"loss": 0.0722,
|
6663 |
+
"step": 946
|
6664 |
+
},
|
6665 |
+
{
|
6666 |
+
"epoch": 0.48224061107574795,
|
6667 |
+
"grad_norm": 16.160083770751953,
|
6668 |
+
"learning_rate": 7.055001909504755e-07,
|
6669 |
+
"loss": 0.0567,
|
6670 |
+
"step": 947
|
6671 |
+
},
|
6672 |
+
{
|
6673 |
+
"epoch": 0.4827498408656906,
|
6674 |
+
"grad_norm": 5.195967674255371,
|
6675 |
+
"learning_rate": 6.791885693514133e-07,
|
6676 |
+
"loss": 0.0777,
|
6677 |
+
"step": 948
|
6678 |
+
},
|
6679 |
+
{
|
6680 |
+
"epoch": 0.48325907065563334,
|
6681 |
+
"grad_norm": 15.034565925598145,
|
6682 |
+
"learning_rate": 6.533736077758868e-07,
|
6683 |
+
"loss": 0.073,
|
6684 |
+
"step": 949
|
6685 |
+
},
|
6686 |
+
{
|
6687 |
+
"epoch": 0.48376830044557606,
|
6688 |
+
"grad_norm": 3.8854053020477295,
|
6689 |
+
"learning_rate": 6.280555661802856e-07,
|
6690 |
+
"loss": 0.0394,
|
6691 |
+
"step": 950
|
6692 |
+
},
|
6693 |
+
{
|
6694 |
+
"epoch": 0.4842775302355188,
|
6695 |
+
"grad_norm": 3.7877042293548584,
|
6696 |
+
"learning_rate": 6.032346995169968e-07,
|
6697 |
+
"loss": 0.7431,
|
6698 |
+
"step": 951
|
6699 |
+
},
|
6700 |
+
{
|
6701 |
+
"epoch": 0.4847867600254615,
|
6702 |
+
"grad_norm": 11.732136726379395,
|
6703 |
+
"learning_rate": 5.78911257731879e-07,
|
6704 |
+
"loss": 0.6649,
|
6705 |
+
"step": 952
|
6706 |
+
},
|
6707 |
+
{
|
6708 |
+
"epoch": 0.4852959898154042,
|
6709 |
+
"grad_norm": 5.163148403167725,
|
6710 |
+
"learning_rate": 5.550854857617193e-07,
|
6711 |
+
"loss": 0.4255,
|
6712 |
+
"step": 953
|
6713 |
+
},
|
6714 |
+
{
|
6715 |
+
"epoch": 0.4858052196053469,
|
6716 |
+
"grad_norm": 7.868194580078125,
|
6717 |
+
"learning_rate": 5.317576235317756e-07,
|
6718 |
+
"loss": 0.7605,
|
6719 |
+
"step": 954
|
6720 |
+
},
|
6721 |
+
{
|
6722 |
+
"epoch": 0.4863144493952896,
|
6723 |
+
"grad_norm": 7.837756633758545,
|
6724 |
+
"learning_rate": 5.089279059533658e-07,
|
6725 |
+
"loss": 0.7603,
|
6726 |
+
"step": 955
|
6727 |
+
},
|
6728 |
+
{
|
6729 |
+
"epoch": 0.48682367918523234,
|
6730 |
+
"grad_norm": 6.926958084106445,
|
6731 |
+
"learning_rate": 4.865965629214819e-07,
|
6732 |
+
"loss": 0.3583,
|
6733 |
+
"step": 956
|
6734 |
+
},
|
6735 |
+
{
|
6736 |
+
"epoch": 0.48733290897517506,
|
6737 |
+
"grad_norm": 6.891408920288086,
|
6738 |
+
"learning_rate": 4.647638193125137e-07,
|
6739 |
+
"loss": 0.5063,
|
6740 |
+
"step": 957
|
6741 |
+
},
|
6742 |
+
{
|
6743 |
+
"epoch": 0.4878421387651178,
|
6744 |
+
"grad_norm": 17.593307495117188,
|
6745 |
+
"learning_rate": 4.434298949819449e-07,
|
6746 |
+
"loss": 1.0916,
|
6747 |
+
"step": 958
|
6748 |
+
},
|
6749 |
+
{
|
6750 |
+
"epoch": 0.48835136855506045,
|
6751 |
+
"grad_norm": 13.12249755859375,
|
6752 |
+
"learning_rate": 4.2259500476214407e-07,
|
6753 |
+
"loss": 1.178,
|
6754 |
+
"step": 959
|
6755 |
+
},
|
6756 |
+
{
|
6757 |
+
"epoch": 0.48886059834500317,
|
6758 |
+
"grad_norm": 8.691963195800781,
|
6759 |
+
"learning_rate": 4.02259358460233e-07,
|
6760 |
+
"loss": 0.6026,
|
6761 |
+
"step": 960
|
6762 |
+
},
|
6763 |
+
{
|
6764 |
+
"epoch": 0.4893698281349459,
|
6765 |
+
"grad_norm": 5.080837726593018,
|
6766 |
+
"learning_rate": 3.824231608559492e-07,
|
6767 |
+
"loss": 0.379,
|
6768 |
+
"step": 961
|
6769 |
+
},
|
6770 |
+
{
|
6771 |
+
"epoch": 0.4898790579248886,
|
6772 |
+
"grad_norm": 10.101880073547363,
|
6773 |
+
"learning_rate": 3.630866116995757e-07,
|
6774 |
+
"loss": 0.8812,
|
6775 |
+
"step": 962
|
6776 |
+
},
|
6777 |
+
{
|
6778 |
+
"epoch": 0.49038828771483134,
|
6779 |
+
"grad_norm": 2.2711408138275146,
|
6780 |
+
"learning_rate": 3.4424990570994797e-07,
|
6781 |
+
"loss": 0.1018,
|
6782 |
+
"step": 963
|
6783 |
+
},
|
6784 |
+
{
|
6785 |
+
"epoch": 0.490897517504774,
|
6786 |
+
"grad_norm": 0.9632775187492371,
|
6787 |
+
"learning_rate": 3.2591323257248893e-07,
|
6788 |
+
"loss": 0.1066,
|
6789 |
+
"step": 964
|
6790 |
+
},
|
6791 |
+
{
|
6792 |
+
"epoch": 0.49140674729471673,
|
6793 |
+
"grad_norm": 2.8695173263549805,
|
6794 |
+
"learning_rate": 3.080767769372939e-07,
|
6795 |
+
"loss": 0.1316,
|
6796 |
+
"step": 965
|
6797 |
+
},
|
6798 |
+
{
|
6799 |
+
"epoch": 0.49191597708465945,
|
6800 |
+
"grad_norm": 1.3276772499084473,
|
6801 |
+
"learning_rate": 2.907407184172706e-07,
|
6802 |
+
"loss": 0.0816,
|
6803 |
+
"step": 966
|
6804 |
+
},
|
6805 |
+
{
|
6806 |
+
"epoch": 0.4924252068746022,
|
6807 |
+
"grad_norm": 4.322078227996826,
|
6808 |
+
"learning_rate": 2.7390523158633554e-07,
|
6809 |
+
"loss": 0.1199,
|
6810 |
+
"step": 967
|
6811 |
+
},
|
6812 |
+
{
|
6813 |
+
"epoch": 0.4929344366645449,
|
6814 |
+
"grad_norm": 0.6163370013237,
|
6815 |
+
"learning_rate": 2.5757048597765396e-07,
|
6816 |
+
"loss": 0.0881,
|
6817 |
+
"step": 968
|
6818 |
+
},
|
6819 |
+
{
|
6820 |
+
"epoch": 0.49344366645448756,
|
6821 |
+
"grad_norm": 0.8970826864242554,
|
6822 |
+
"learning_rate": 2.4173664608193593e-07,
|
6823 |
+
"loss": 0.095,
|
6824 |
+
"step": 969
|
6825 |
+
},
|
6826 |
+
{
|
6827 |
+
"epoch": 0.4939528962444303,
|
6828 |
+
"grad_norm": 1.772135615348816,
|
6829 |
+
"learning_rate": 2.2640387134577058e-07,
|
6830 |
+
"loss": 0.0945,
|
6831 |
+
"step": 970
|
6832 |
+
},
|
6833 |
+
{
|
6834 |
+
"epoch": 0.494462126034373,
|
6835 |
+
"grad_norm": 1.8468323945999146,
|
6836 |
+
"learning_rate": 2.1157231617002783e-07,
|
6837 |
+
"loss": 0.136,
|
6838 |
+
"step": 971
|
6839 |
+
},
|
6840 |
+
{
|
6841 |
+
"epoch": 0.49497135582431573,
|
6842 |
+
"grad_norm": 2.6759517192840576,
|
6843 |
+
"learning_rate": 1.9724212990830938e-07,
|
6844 |
+
"loss": 0.138,
|
6845 |
+
"step": 972
|
6846 |
+
},
|
6847 |
+
{
|
6848 |
+
"epoch": 0.49548058561425845,
|
6849 |
+
"grad_norm": 0.8388791680335999,
|
6850 |
+
"learning_rate": 1.8341345686543332e-07,
|
6851 |
+
"loss": 0.1131,
|
6852 |
+
"step": 973
|
6853 |
+
},
|
6854 |
+
{
|
6855 |
+
"epoch": 0.4959898154042012,
|
6856 |
+
"grad_norm": 5.01349401473999,
|
6857 |
+
"learning_rate": 1.7008643629596866e-07,
|
6858 |
+
"loss": 0.104,
|
6859 |
+
"step": 974
|
6860 |
+
},
|
6861 |
+
{
|
6862 |
+
"epoch": 0.49649904519414384,
|
6863 |
+
"grad_norm": 8.658512115478516,
|
6864 |
+
"learning_rate": 1.5726120240288634e-07,
|
6865 |
+
"loss": 0.088,
|
6866 |
+
"step": 975
|
6867 |
+
},
|
6868 |
+
{
|
6869 |
+
"epoch": 0.49700827498408656,
|
6870 |
+
"grad_norm": 10.031734466552734,
|
6871 |
+
"learning_rate": 1.449378843361271e-07,
|
6872 |
+
"loss": 0.1412,
|
6873 |
+
"step": 976
|
6874 |
+
},
|
6875 |
+
{
|
6876 |
+
"epoch": 0.4975175047740293,
|
6877 |
+
"grad_norm": 5.323798656463623,
|
6878 |
+
"learning_rate": 1.3311660619138578e-07,
|
6879 |
+
"loss": 0.0799,
|
6880 |
+
"step": 977
|
6881 |
+
},
|
6882 |
+
{
|
6883 |
+
"epoch": 0.498026734563972,
|
6884 |
+
"grad_norm": 14.896018981933594,
|
6885 |
+
"learning_rate": 1.2179748700879012e-07,
|
6886 |
+
"loss": 0.0547,
|
6887 |
+
"step": 978
|
6888 |
+
},
|
6889 |
+
{
|
6890 |
+
"epoch": 0.49853596435391473,
|
6891 |
+
"grad_norm": 2.8598153591156006,
|
6892 |
+
"learning_rate": 1.109806407717462e-07,
|
6893 |
+
"loss": 0.057,
|
6894 |
+
"step": 979
|
6895 |
+
},
|
6896 |
+
{
|
6897 |
+
"epoch": 0.4990451941438574,
|
6898 |
+
"grad_norm": 4.470668792724609,
|
6899 |
+
"learning_rate": 1.0066617640578368e-07,
|
6900 |
+
"loss": 0.1221,
|
6901 |
+
"step": 980
|
6902 |
+
},
|
6903 |
+
{
|
6904 |
+
"epoch": 0.4995544239338001,
|
6905 |
+
"grad_norm": 6.321918964385986,
|
6906 |
+
"learning_rate": 9.085419777743465e-08,
|
6907 |
+
"loss": 0.1274,
|
6908 |
+
"step": 981
|
6909 |
+
},
|
6910 |
+
{
|
6911 |
+
"epoch": 0.5000636537237428,
|
6912 |
+
"grad_norm": 4.125798225402832,
|
6913 |
+
"learning_rate": 8.15448036932176e-08,
|
6914 |
+
"loss": 0.0822,
|
6915 |
+
"step": 982
|
6916 |
+
},
|
6917 |
+
{
|
6918 |
+
"epoch": 0.5005728835136856,
|
6919 |
+
"grad_norm": 5.414987564086914,
|
6920 |
+
"learning_rate": 7.273808789862724e-08,
|
6921 |
+
"loss": 0.0796,
|
6922 |
+
"step": 983
|
6923 |
+
},
|
6924 |
+
{
|
6925 |
+
"epoch": 0.5010821133036283,
|
6926 |
+
"grad_norm": 5.301061153411865,
|
6927 |
+
"learning_rate": 6.443413907720186e-08,
|
6928 |
+
"loss": 0.1011,
|
6929 |
+
"step": 984
|
6930 |
+
},
|
6931 |
+
{
|
6932 |
+
"epoch": 0.501591343093571,
|
6933 |
+
"grad_norm": 6.373870849609375,
|
6934 |
+
"learning_rate": 5.663304084960186e-08,
|
6935 |
+
"loss": 0.1335,
|
6936 |
+
"step": 985
|
6937 |
+
},
|
6938 |
+
{
|
6939 |
+
"epoch": 0.5021005728835137,
|
6940 |
+
"grad_norm": 3.59035062789917,
|
6941 |
+
"learning_rate": 4.933487177280482e-08,
|
6942 |
+
"loss": 0.06,
|
6943 |
+
"step": 986
|
6944 |
+
},
|
6945 |
+
{
|
6946 |
+
"epoch": 0.5026098026734565,
|
6947 |
+
"grad_norm": 4.328834533691406,
|
6948 |
+
"learning_rate": 4.253970533929508e-08,
|
6949 |
+
"loss": 0.0565,
|
6950 |
+
"step": 987
|
6951 |
+
},
|
6952 |
+
{
|
6953 |
+
"epoch": 0.5031190324633991,
|
6954 |
+
"grad_norm": 2.652255058288574,
|
6955 |
+
"learning_rate": 3.624760997631982e-08,
|
6956 |
+
"loss": 0.0428,
|
6957 |
+
"step": 988
|
6958 |
+
},
|
6959 |
+
{
|
6960 |
+
"epoch": 0.5036282622533418,
|
6961 |
+
"grad_norm": 4.667551040649414,
|
6962 |
+
"learning_rate": 3.04586490452119e-08,
|
6963 |
+
"loss": 0.0206,
|
6964 |
+
"step": 989
|
6965 |
+
},
|
6966 |
+
{
|
6967 |
+
"epoch": 0.5041374920432845,
|
6968 |
+
"grad_norm": 7.6786112785339355,
|
6969 |
+
"learning_rate": 2.5172880840745873e-08,
|
6970 |
+
"loss": 0.1095,
|
6971 |
+
"step": 990
|
6972 |
+
},
|
6973 |
+
{
|
6974 |
+
"epoch": 0.5046467218332272,
|
6975 |
+
"grad_norm": 3.08980131149292,
|
6976 |
+
"learning_rate": 2.0390358590538504e-08,
|
6977 |
+
"loss": 0.0337,
|
6978 |
+
"step": 991
|
6979 |
+
},
|
6980 |
+
{
|
6981 |
+
"epoch": 0.50515595162317,
|
6982 |
+
"grad_norm": 2.030238389968872,
|
6983 |
+
"learning_rate": 1.6111130454543598e-08,
|
6984 |
+
"loss": 0.0064,
|
6985 |
+
"step": 992
|
6986 |
+
},
|
6987 |
+
{
|
6988 |
+
"epoch": 0.5056651814131127,
|
6989 |
+
"grad_norm": 0.7045100331306458,
|
6990 |
+
"learning_rate": 1.2335239524541299e-08,
|
6991 |
+
"loss": 0.0105,
|
6992 |
+
"step": 993
|
6993 |
+
},
|
6994 |
+
{
|
6995 |
+
"epoch": 0.5061744112030554,
|
6996 |
+
"grad_norm": 8.8118896484375,
|
6997 |
+
"learning_rate": 9.06272382371065e-09,
|
6998 |
+
"loss": 0.0789,
|
6999 |
+
"step": 994
|
7000 |
+
},
|
7001 |
+
{
|
7002 |
+
"epoch": 0.5066836409929981,
|
7003 |
+
"grad_norm": 8.802488327026367,
|
7004 |
+
"learning_rate": 6.293616306246586e-09,
|
7005 |
+
"loss": 0.0811,
|
7006 |
+
"step": 995
|
7007 |
+
},
|
7008 |
+
{
|
7009 |
+
"epoch": 0.5071928707829408,
|
7010 |
+
"grad_norm": 4.467785358428955,
|
7011 |
+
"learning_rate": 4.0279448570323954e-09,
|
7012 |
+
"loss": 0.0437,
|
7013 |
+
"step": 996
|
7014 |
+
},
|
7015 |
+
{
|
7016 |
+
"epoch": 0.5077021005728836,
|
7017 |
+
"grad_norm": 8.837767601013184,
|
7018 |
+
"learning_rate": 2.265732291356626e-09,
|
7019 |
+
"loss": 0.0425,
|
7020 |
+
"step": 997
|
7021 |
+
},
|
7022 |
+
{
|
7023 |
+
"epoch": 0.5082113303628262,
|
7024 |
+
"grad_norm": 7.292797565460205,
|
7025 |
+
"learning_rate": 1.0069963546743832e-09,
|
7026 |
+
"loss": 0.0808,
|
7027 |
+
"step": 998
|
7028 |
+
},
|
7029 |
+
{
|
7030 |
+
"epoch": 0.5087205601527689,
|
7031 |
+
"grad_norm": 3.702165365219116,
|
7032 |
+
"learning_rate": 2.5174972244634833e-10,
|
7033 |
+
"loss": 0.0469,
|
7034 |
+
"step": 999
|
7035 |
+
},
|
7036 |
+
{
|
7037 |
+
"epoch": 0.5092297899427116,
|
7038 |
+
"grad_norm": 8.40251636505127,
|
7039 |
+
"learning_rate": 0.0,
|
7040 |
+
"loss": 0.0725,
|
7041 |
+
"step": 1000
|
7042 |
+
},
|
7043 |
+
{
|
7044 |
+
"epoch": 0.5092297899427116,
|
7045 |
+
"eval_loss": 0.22850316762924194,
|
7046 |
+
"eval_runtime": 378.274,
|
7047 |
+
"eval_samples_per_second": 8.745,
|
7048 |
+
"eval_steps_per_second": 2.186,
|
7049 |
+
"step": 1000
|
7050 |
}
|
7051 |
],
|
7052 |
"logging_steps": 1,
|
|
|
7070 |
"should_evaluate": false,
|
7071 |
"should_log": false,
|
7072 |
"should_save": true,
|
7073 |
+
"should_training_stop": true
|
7074 |
},
|
7075 |
"attributes": {}
|
7076 |
}
|
7077 |
},
|
7078 |
+
"total_flos": 2.0800300944866673e+18,
|
7079 |
"train_batch_size": 8,
|
7080 |
"trial_name": null,
|
7081 |
"trial_params": null
|