End of training
Browse files
README.md
CHANGED
|
@@ -1,11 +1,23 @@
|
|
| 1 |
---
|
| 2 |
tags:
|
| 3 |
- generated_from_trainer
|
|
|
|
|
|
|
| 4 |
metrics:
|
| 5 |
- accuracy
|
| 6 |
model-index:
|
| 7 |
- name: smolm-autoreg-bpe-counterfactual_babylm_aann_indef_articles_with_pl_nouns_removal_new-3e-4
|
| 8 |
-
results:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
---
|
| 10 |
|
| 11 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
@@ -13,7 +25,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 13 |
|
| 14 |
# smolm-autoreg-bpe-counterfactual_babylm_aann_indef_articles_with_pl_nouns_removal_new-3e-4
|
| 15 |
|
| 16 |
-
This model was trained from scratch on
|
| 17 |
It achieves the following results on the evaluation set:
|
| 18 |
- Loss: 3.4250
|
| 19 |
- Accuracy: 0.4092
|
|
|
|
| 1 |
---
|
| 2 |
tags:
|
| 3 |
- generated_from_trainer
|
| 4 |
+
datasets:
|
| 5 |
+
- kanishka/counterfactual_babylm_aann_indef_articles_with_pl_nouns_removal_new
|
| 6 |
metrics:
|
| 7 |
- accuracy
|
| 8 |
model-index:
|
| 9 |
- name: smolm-autoreg-bpe-counterfactual_babylm_aann_indef_articles_with_pl_nouns_removal_new-3e-4
|
| 10 |
+
results:
|
| 11 |
+
- task:
|
| 12 |
+
name: Causal Language Modeling
|
| 13 |
+
type: text-generation
|
| 14 |
+
dataset:
|
| 15 |
+
name: kanishka/counterfactual_babylm_aann_indef_articles_with_pl_nouns_removal_new
|
| 16 |
+
type: kanishka/counterfactual_babylm_aann_indef_articles_with_pl_nouns_removal_new
|
| 17 |
+
metrics:
|
| 18 |
+
- name: Accuracy
|
| 19 |
+
type: accuracy
|
| 20 |
+
value: 0.4091656007481136
|
| 21 |
---
|
| 22 |
|
| 23 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
|
|
| 25 |
|
| 26 |
# smolm-autoreg-bpe-counterfactual_babylm_aann_indef_articles_with_pl_nouns_removal_new-3e-4
|
| 27 |
|
| 28 |
+
This model was trained from scratch on the kanishka/counterfactual_babylm_aann_indef_articles_with_pl_nouns_removal_new dataset.
|
| 29 |
It achieves the following results on the evaluation set:
|
| 30 |
- Loss: 3.4250
|
| 31 |
- Accuracy: 0.4092
|
all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"eval_accuracy": 0.4091656007481136,
|
| 4 |
+
"eval_loss": 3.4249629974365234,
|
| 5 |
+
"eval_runtime": 155.4739,
|
| 6 |
+
"eval_samples": 57920,
|
| 7 |
+
"eval_samples_per_second": 372.538,
|
| 8 |
+
"eval_steps_per_second": 5.821,
|
| 9 |
+
"perplexity": 30.72150805591893,
|
| 10 |
+
"train_loss": 3.055753937752016,
|
| 11 |
+
"train_runtime": 81288.5778,
|
| 12 |
+
"train_samples": 595173,
|
| 13 |
+
"train_samples_per_second": 146.435,
|
| 14 |
+
"train_steps_per_second": 4.576
|
| 15 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"eval_accuracy": 0.4091656007481136,
|
| 4 |
+
"eval_loss": 3.4249629974365234,
|
| 5 |
+
"eval_runtime": 155.4739,
|
| 6 |
+
"eval_samples": 57920,
|
| 7 |
+
"eval_samples_per_second": 372.538,
|
| 8 |
+
"eval_steps_per_second": 5.821,
|
| 9 |
+
"perplexity": 30.72150805591893
|
| 10 |
+
}
|
runs/Jun21_17-29-53_phyl-ling-p01.la.utexas.edu/events.out.tfevents.1719090849.phyl-ling-p01.la.utexas.edu.1078584.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:176c53e59369deeb3b3afd7578a977f14dfc98ce86905ebfe845e3d6d2831ad9
|
| 3 |
+
size 417
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"train_loss": 3.055753937752016,
|
| 4 |
+
"train_runtime": 81288.5778,
|
| 5 |
+
"train_samples": 595173,
|
| 6 |
+
"train_samples_per_second": 146.435,
|
| 7 |
+
"train_steps_per_second": 4.576
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2814 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 20.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 372000,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.05,
|
| 13 |
+
"grad_norm": 0.8259170651435852,
|
| 14 |
+
"learning_rate": 9.375e-06,
|
| 15 |
+
"loss": 6.8606,
|
| 16 |
+
"step": 1000
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.11,
|
| 20 |
+
"grad_norm": 0.7919405102729797,
|
| 21 |
+
"learning_rate": 1.875e-05,
|
| 22 |
+
"loss": 5.3621,
|
| 23 |
+
"step": 2000
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.16,
|
| 27 |
+
"grad_norm": 0.9477707147598267,
|
| 28 |
+
"learning_rate": 2.8125e-05,
|
| 29 |
+
"loss": 5.0308,
|
| 30 |
+
"step": 3000
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.22,
|
| 34 |
+
"grad_norm": 0.9951394200325012,
|
| 35 |
+
"learning_rate": 3.75e-05,
|
| 36 |
+
"loss": 4.8123,
|
| 37 |
+
"step": 4000
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.27,
|
| 41 |
+
"grad_norm": 0.9227204918861389,
|
| 42 |
+
"learning_rate": 4.6874999999999994e-05,
|
| 43 |
+
"loss": 4.6357,
|
| 44 |
+
"step": 5000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.32,
|
| 48 |
+
"grad_norm": 0.9371578097343445,
|
| 49 |
+
"learning_rate": 5.625e-05,
|
| 50 |
+
"loss": 4.4935,
|
| 51 |
+
"step": 6000
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.38,
|
| 55 |
+
"grad_norm": 0.9119696617126465,
|
| 56 |
+
"learning_rate": 6.5625e-05,
|
| 57 |
+
"loss": 4.3815,
|
| 58 |
+
"step": 7000
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.43,
|
| 62 |
+
"grad_norm": 0.9198476672172546,
|
| 63 |
+
"learning_rate": 7.5e-05,
|
| 64 |
+
"loss": 4.2884,
|
| 65 |
+
"step": 8000
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.48,
|
| 69 |
+
"grad_norm": 0.8627898693084717,
|
| 70 |
+
"learning_rate": 8.437499999999999e-05,
|
| 71 |
+
"loss": 4.208,
|
| 72 |
+
"step": 9000
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.54,
|
| 76 |
+
"grad_norm": 0.8556337356567383,
|
| 77 |
+
"learning_rate": 9.374999999999999e-05,
|
| 78 |
+
"loss": 4.1353,
|
| 79 |
+
"step": 10000
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.59,
|
| 83 |
+
"grad_norm": 0.8405832052230835,
|
| 84 |
+
"learning_rate": 0.000103115625,
|
| 85 |
+
"loss": 4.0695,
|
| 86 |
+
"step": 11000
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.65,
|
| 90 |
+
"grad_norm": 0.8061829805374146,
|
| 91 |
+
"learning_rate": 0.000112490625,
|
| 92 |
+
"loss": 4.0086,
|
| 93 |
+
"step": 12000
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.7,
|
| 97 |
+
"grad_norm": 0.8495123386383057,
|
| 98 |
+
"learning_rate": 0.000121865625,
|
| 99 |
+
"loss": 3.946,
|
| 100 |
+
"step": 13000
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.75,
|
| 104 |
+
"grad_norm": 0.9331762790679932,
|
| 105 |
+
"learning_rate": 0.000131221875,
|
| 106 |
+
"loss": 3.8919,
|
| 107 |
+
"step": 14000
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.81,
|
| 111 |
+
"grad_norm": 0.8087450265884399,
|
| 112 |
+
"learning_rate": 0.000140596875,
|
| 113 |
+
"loss": 3.8453,
|
| 114 |
+
"step": 15000
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.86,
|
| 118 |
+
"grad_norm": 0.78471440076828,
|
| 119 |
+
"learning_rate": 0.0001499625,
|
| 120 |
+
"loss": 3.8107,
|
| 121 |
+
"step": 16000
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.91,
|
| 125 |
+
"grad_norm": 0.7369292974472046,
|
| 126 |
+
"learning_rate": 0.000159328125,
|
| 127 |
+
"loss": 3.7708,
|
| 128 |
+
"step": 17000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.97,
|
| 132 |
+
"grad_norm": 0.7355145812034607,
|
| 133 |
+
"learning_rate": 0.000168703125,
|
| 134 |
+
"loss": 3.7386,
|
| 135 |
+
"step": 18000
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 1.0,
|
| 139 |
+
"eval_accuracy": 0.3445651589741385,
|
| 140 |
+
"eval_loss": 3.938964366912842,
|
| 141 |
+
"eval_runtime": 153.6743,
|
| 142 |
+
"eval_samples_per_second": 376.901,
|
| 143 |
+
"eval_steps_per_second": 5.889,
|
| 144 |
+
"step": 18600
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 1.02,
|
| 148 |
+
"grad_norm": 0.745087206363678,
|
| 149 |
+
"learning_rate": 0.00017806875,
|
| 150 |
+
"loss": 3.7067,
|
| 151 |
+
"step": 19000
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 1.08,
|
| 155 |
+
"grad_norm": 0.9550297260284424,
|
| 156 |
+
"learning_rate": 0.00018744374999999999,
|
| 157 |
+
"loss": 3.6708,
|
| 158 |
+
"step": 20000
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.13,
|
| 162 |
+
"grad_norm": 0.6702427864074707,
|
| 163 |
+
"learning_rate": 0.00019680937499999996,
|
| 164 |
+
"loss": 3.6515,
|
| 165 |
+
"step": 21000
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 1.18,
|
| 169 |
+
"grad_norm": 0.7245630025863647,
|
| 170 |
+
"learning_rate": 0.00020617499999999998,
|
| 171 |
+
"loss": 3.6354,
|
| 172 |
+
"step": 22000
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 1.24,
|
| 176 |
+
"grad_norm": 0.6476539969444275,
|
| 177 |
+
"learning_rate": 0.00021554999999999998,
|
| 178 |
+
"loss": 3.6117,
|
| 179 |
+
"step": 23000
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.29,
|
| 183 |
+
"grad_norm": 0.6578693389892578,
|
| 184 |
+
"learning_rate": 0.00022492499999999998,
|
| 185 |
+
"loss": 3.5979,
|
| 186 |
+
"step": 24000
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 1.34,
|
| 190 |
+
"grad_norm": 0.6044876575469971,
|
| 191 |
+
"learning_rate": 0.00023429999999999998,
|
| 192 |
+
"loss": 3.5757,
|
| 193 |
+
"step": 25000
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 1.4,
|
| 197 |
+
"grad_norm": 0.6016374826431274,
|
| 198 |
+
"learning_rate": 0.00024366562499999997,
|
| 199 |
+
"loss": 3.572,
|
| 200 |
+
"step": 26000
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 1.45,
|
| 204 |
+
"grad_norm": 0.56680828332901,
|
| 205 |
+
"learning_rate": 0.000253040625,
|
| 206 |
+
"loss": 3.552,
|
| 207 |
+
"step": 27000
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 1.51,
|
| 211 |
+
"grad_norm": 0.5599876046180725,
|
| 212 |
+
"learning_rate": 0.00026240624999999994,
|
| 213 |
+
"loss": 3.541,
|
| 214 |
+
"step": 28000
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 1.56,
|
| 218 |
+
"grad_norm": 0.5561350584030151,
|
| 219 |
+
"learning_rate": 0.00027178124999999994,
|
| 220 |
+
"loss": 3.5302,
|
| 221 |
+
"step": 29000
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.61,
|
| 225 |
+
"grad_norm": 0.5627046227455139,
|
| 226 |
+
"learning_rate": 0.00028115624999999994,
|
| 227 |
+
"loss": 3.5212,
|
| 228 |
+
"step": 30000
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 1.67,
|
| 232 |
+
"grad_norm": 0.5255282521247864,
|
| 233 |
+
"learning_rate": 0.000290521875,
|
| 234 |
+
"loss": 3.5043,
|
| 235 |
+
"step": 31000
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 1.72,
|
| 239 |
+
"grad_norm": 0.5291178822517395,
|
| 240 |
+
"learning_rate": 0.000299896875,
|
| 241 |
+
"loss": 3.4957,
|
| 242 |
+
"step": 32000
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.77,
|
| 246 |
+
"grad_norm": 0.48417210578918457,
|
| 247 |
+
"learning_rate": 0.0002991282352941176,
|
| 248 |
+
"loss": 3.4827,
|
| 249 |
+
"step": 33000
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 1.83,
|
| 253 |
+
"grad_norm": 0.49497950077056885,
|
| 254 |
+
"learning_rate": 0.00029824588235294113,
|
| 255 |
+
"loss": 3.4696,
|
| 256 |
+
"step": 34000
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 1.88,
|
| 260 |
+
"grad_norm": 0.45140179991722107,
|
| 261 |
+
"learning_rate": 0.0002973635294117647,
|
| 262 |
+
"loss": 3.4628,
|
| 263 |
+
"step": 35000
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 1.94,
|
| 267 |
+
"grad_norm": 0.4502553939819336,
|
| 268 |
+
"learning_rate": 0.00029648205882352937,
|
| 269 |
+
"loss": 3.4448,
|
| 270 |
+
"step": 36000
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 1.99,
|
| 274 |
+
"grad_norm": 0.47246313095092773,
|
| 275 |
+
"learning_rate": 0.0002955997058823529,
|
| 276 |
+
"loss": 3.4323,
|
| 277 |
+
"step": 37000
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 2.0,
|
| 281 |
+
"eval_accuracy": 0.37465563664896706,
|
| 282 |
+
"eval_loss": 3.6439149379730225,
|
| 283 |
+
"eval_runtime": 155.8711,
|
| 284 |
+
"eval_samples_per_second": 371.589,
|
| 285 |
+
"eval_steps_per_second": 5.806,
|
| 286 |
+
"step": 37200
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 2.04,
|
| 290 |
+
"grad_norm": 0.4650474488735199,
|
| 291 |
+
"learning_rate": 0.00029471735294117645,
|
| 292 |
+
"loss": 3.3917,
|
| 293 |
+
"step": 38000
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 2.1,
|
| 297 |
+
"grad_norm": 0.4563141465187073,
|
| 298 |
+
"learning_rate": 0.00029383588235294114,
|
| 299 |
+
"loss": 3.3726,
|
| 300 |
+
"step": 39000
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 2.15,
|
| 304 |
+
"grad_norm": 0.42616134881973267,
|
| 305 |
+
"learning_rate": 0.0002929535294117647,
|
| 306 |
+
"loss": 3.3674,
|
| 307 |
+
"step": 40000
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 2.2,
|
| 311 |
+
"grad_norm": 0.4414719343185425,
|
| 312 |
+
"learning_rate": 0.00029207117647058817,
|
| 313 |
+
"loss": 3.3679,
|
| 314 |
+
"step": 41000
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 2.26,
|
| 318 |
+
"grad_norm": 0.42499449849128723,
|
| 319 |
+
"learning_rate": 0.0002911897058823529,
|
| 320 |
+
"loss": 3.3584,
|
| 321 |
+
"step": 42000
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 2.31,
|
| 325 |
+
"grad_norm": 0.4124417304992676,
|
| 326 |
+
"learning_rate": 0.00029030735294117646,
|
| 327 |
+
"loss": 3.3522,
|
| 328 |
+
"step": 43000
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 2.37,
|
| 332 |
+
"grad_norm": 0.3980655372142792,
|
| 333 |
+
"learning_rate": 0.00028942499999999995,
|
| 334 |
+
"loss": 3.3481,
|
| 335 |
+
"step": 44000
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 2.42,
|
| 339 |
+
"grad_norm": 0.40041157603263855,
|
| 340 |
+
"learning_rate": 0.0002885426470588235,
|
| 341 |
+
"loss": 3.3422,
|
| 342 |
+
"step": 45000
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 2.47,
|
| 346 |
+
"grad_norm": 0.399441123008728,
|
| 347 |
+
"learning_rate": 0.00028766117647058823,
|
| 348 |
+
"loss": 3.3346,
|
| 349 |
+
"step": 46000
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 2.53,
|
| 353 |
+
"grad_norm": 0.39357373118400574,
|
| 354 |
+
"learning_rate": 0.0002867788235294118,
|
| 355 |
+
"loss": 3.3326,
|
| 356 |
+
"step": 47000
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.58,
|
| 360 |
+
"grad_norm": 0.4075976610183716,
|
| 361 |
+
"learning_rate": 0.00028589647058823526,
|
| 362 |
+
"loss": 3.3236,
|
| 363 |
+
"step": 48000
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 2.63,
|
| 367 |
+
"grad_norm": 0.3738775849342346,
|
| 368 |
+
"learning_rate": 0.00028501499999999995,
|
| 369 |
+
"loss": 3.3247,
|
| 370 |
+
"step": 49000
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 2.69,
|
| 374 |
+
"grad_norm": 0.37619298696517944,
|
| 375 |
+
"learning_rate": 0.0002841326470588235,
|
| 376 |
+
"loss": 3.3113,
|
| 377 |
+
"step": 50000
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 2.74,
|
| 381 |
+
"grad_norm": 0.37718719244003296,
|
| 382 |
+
"learning_rate": 0.00028325029411764704,
|
| 383 |
+
"loss": 3.308,
|
| 384 |
+
"step": 51000
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 2.8,
|
| 388 |
+
"grad_norm": 0.3818044066429138,
|
| 389 |
+
"learning_rate": 0.0002823679411764706,
|
| 390 |
+
"loss": 3.3057,
|
| 391 |
+
"step": 52000
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 2.85,
|
| 395 |
+
"grad_norm": 0.3819046914577484,
|
| 396 |
+
"learning_rate": 0.00028148647058823527,
|
| 397 |
+
"loss": 3.2991,
|
| 398 |
+
"step": 53000
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 2.9,
|
| 402 |
+
"grad_norm": 0.3860973119735718,
|
| 403 |
+
"learning_rate": 0.0002806041176470588,
|
| 404 |
+
"loss": 3.2951,
|
| 405 |
+
"step": 54000
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 2.96,
|
| 409 |
+
"grad_norm": 0.3734324276447296,
|
| 410 |
+
"learning_rate": 0.00027972176470588236,
|
| 411 |
+
"loss": 3.2906,
|
| 412 |
+
"step": 55000
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 3.0,
|
| 416 |
+
"eval_accuracy": 0.3877894103231077,
|
| 417 |
+
"eval_loss": 3.5132408142089844,
|
| 418 |
+
"eval_runtime": 154.2632,
|
| 419 |
+
"eval_samples_per_second": 375.462,
|
| 420 |
+
"eval_steps_per_second": 5.867,
|
| 421 |
+
"step": 55800
|
| 422 |
+
},
|
| 423 |
+
{
|
| 424 |
+
"epoch": 3.01,
|
| 425 |
+
"grad_norm": 0.3738992214202881,
|
| 426 |
+
"learning_rate": 0.00027884029411764705,
|
| 427 |
+
"loss": 3.271,
|
| 428 |
+
"step": 56000
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 3.06,
|
| 432 |
+
"grad_norm": 0.3885354697704315,
|
| 433 |
+
"learning_rate": 0.0002779579411764706,
|
| 434 |
+
"loss": 3.2208,
|
| 435 |
+
"step": 57000
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 3.12,
|
| 439 |
+
"grad_norm": 0.3763429522514343,
|
| 440 |
+
"learning_rate": 0.0002770755882352941,
|
| 441 |
+
"loss": 3.2269,
|
| 442 |
+
"step": 58000
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 3.17,
|
| 446 |
+
"grad_norm": 0.37402838468551636,
|
| 447 |
+
"learning_rate": 0.00027619411764705877,
|
| 448 |
+
"loss": 3.2296,
|
| 449 |
+
"step": 59000
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 3.23,
|
| 453 |
+
"grad_norm": 0.40755459666252136,
|
| 454 |
+
"learning_rate": 0.0002753117647058823,
|
| 455 |
+
"loss": 3.2297,
|
| 456 |
+
"step": 60000
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 3.28,
|
| 460 |
+
"grad_norm": 0.36526888608932495,
|
| 461 |
+
"learning_rate": 0.00027442941176470585,
|
| 462 |
+
"loss": 3.2264,
|
| 463 |
+
"step": 61000
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 3.33,
|
| 467 |
+
"grad_norm": 0.3764183521270752,
|
| 468 |
+
"learning_rate": 0.0002735470588235294,
|
| 469 |
+
"loss": 3.2268,
|
| 470 |
+
"step": 62000
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 3.39,
|
| 474 |
+
"grad_norm": 0.3648715615272522,
|
| 475 |
+
"learning_rate": 0.0002726664705882353,
|
| 476 |
+
"loss": 3.2206,
|
| 477 |
+
"step": 63000
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 3.44,
|
| 481 |
+
"grad_norm": 0.3624918758869171,
|
| 482 |
+
"learning_rate": 0.00027178411764705883,
|
| 483 |
+
"loss": 3.2268,
|
| 484 |
+
"step": 64000
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 3.49,
|
| 488 |
+
"grad_norm": 0.3546142876148224,
|
| 489 |
+
"learning_rate": 0.0002709017647058823,
|
| 490 |
+
"loss": 3.2182,
|
| 491 |
+
"step": 65000
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 3.55,
|
| 495 |
+
"grad_norm": 0.36598387360572815,
|
| 496 |
+
"learning_rate": 0.00027002029411764706,
|
| 497 |
+
"loss": 3.2185,
|
| 498 |
+
"step": 66000
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 3.6,
|
| 502 |
+
"grad_norm": 0.34152278304100037,
|
| 503 |
+
"learning_rate": 0.00026913794117647055,
|
| 504 |
+
"loss": 3.2204,
|
| 505 |
+
"step": 67000
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 3.66,
|
| 509 |
+
"grad_norm": 0.3589687943458557,
|
| 510 |
+
"learning_rate": 0.0002682555882352941,
|
| 511 |
+
"loss": 3.2158,
|
| 512 |
+
"step": 68000
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 3.71,
|
| 516 |
+
"grad_norm": 0.36120596528053284,
|
| 517 |
+
"learning_rate": 0.0002673741176470588,
|
| 518 |
+
"loss": 3.2098,
|
| 519 |
+
"step": 69000
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 3.76,
|
| 523 |
+
"grad_norm": 0.35094866156578064,
|
| 524 |
+
"learning_rate": 0.0002664917647058823,
|
| 525 |
+
"loss": 3.2098,
|
| 526 |
+
"step": 70000
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 3.82,
|
| 530 |
+
"grad_norm": 0.3578512966632843,
|
| 531 |
+
"learning_rate": 0.000265610294117647,
|
| 532 |
+
"loss": 3.2093,
|
| 533 |
+
"step": 71000
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 3.87,
|
| 537 |
+
"grad_norm": 0.34762489795684814,
|
| 538 |
+
"learning_rate": 0.00026472794117647056,
|
| 539 |
+
"loss": 3.2072,
|
| 540 |
+
"step": 72000
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 3.92,
|
| 544 |
+
"grad_norm": 0.3419814109802246,
|
| 545 |
+
"learning_rate": 0.0002638455882352941,
|
| 546 |
+
"loss": 3.2054,
|
| 547 |
+
"step": 73000
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 3.98,
|
| 551 |
+
"grad_norm": 0.3337661027908325,
|
| 552 |
+
"learning_rate": 0.0002629641176470588,
|
| 553 |
+
"loss": 3.2008,
|
| 554 |
+
"step": 74000
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 4.0,
|
| 558 |
+
"eval_accuracy": 0.3951671566309091,
|
| 559 |
+
"eval_loss": 3.4662108421325684,
|
| 560 |
+
"eval_runtime": 153.9091,
|
| 561 |
+
"eval_samples_per_second": 376.326,
|
| 562 |
+
"eval_steps_per_second": 5.88,
|
| 563 |
+
"step": 74400
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 4.03,
|
| 567 |
+
"grad_norm": 0.36316078901290894,
|
| 568 |
+
"learning_rate": 0.00026208176470588233,
|
| 569 |
+
"loss": 3.1617,
|
| 570 |
+
"step": 75000
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 4.09,
|
| 574 |
+
"grad_norm": 0.35584449768066406,
|
| 575 |
+
"learning_rate": 0.0002611994117647058,
|
| 576 |
+
"loss": 3.1371,
|
| 577 |
+
"step": 76000
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 4.14,
|
| 581 |
+
"grad_norm": 0.36811745166778564,
|
| 582 |
+
"learning_rate": 0.00026031705882352936,
|
| 583 |
+
"loss": 3.1406,
|
| 584 |
+
"step": 77000
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 4.19,
|
| 588 |
+
"grad_norm": 0.37321680784225464,
|
| 589 |
+
"learning_rate": 0.0002594355882352941,
|
| 590 |
+
"loss": 3.1471,
|
| 591 |
+
"step": 78000
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 4.25,
|
| 595 |
+
"grad_norm": 0.3459160029888153,
|
| 596 |
+
"learning_rate": 0.00025855323529411765,
|
| 597 |
+
"loss": 3.1461,
|
| 598 |
+
"step": 79000
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 4.3,
|
| 602 |
+
"grad_norm": 0.3662152588367462,
|
| 603 |
+
"learning_rate": 0.00025767176470588234,
|
| 604 |
+
"loss": 3.1455,
|
| 605 |
+
"step": 80000
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 4.35,
|
| 609 |
+
"grad_norm": 0.35658517479896545,
|
| 610 |
+
"learning_rate": 0.0002567894117647059,
|
| 611 |
+
"loss": 3.1517,
|
| 612 |
+
"step": 81000
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 4.41,
|
| 616 |
+
"grad_norm": 0.3443373739719391,
|
| 617 |
+
"learning_rate": 0.00025590794117647057,
|
| 618 |
+
"loss": 3.1475,
|
| 619 |
+
"step": 82000
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 4.46,
|
| 623 |
+
"grad_norm": 0.34918272495269775,
|
| 624 |
+
"learning_rate": 0.0002550264705882353,
|
| 625 |
+
"loss": 3.1517,
|
| 626 |
+
"step": 83000
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 4.52,
|
| 630 |
+
"grad_norm": 0.3572300672531128,
|
| 631 |
+
"learning_rate": 0.0002541441176470588,
|
| 632 |
+
"loss": 3.1507,
|
| 633 |
+
"step": 84000
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 4.57,
|
| 637 |
+
"grad_norm": 0.35121145844459534,
|
| 638 |
+
"learning_rate": 0.00025326176470588234,
|
| 639 |
+
"loss": 3.15,
|
| 640 |
+
"step": 85000
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 4.62,
|
| 644 |
+
"grad_norm": 0.3452964723110199,
|
| 645 |
+
"learning_rate": 0.00025238029411764703,
|
| 646 |
+
"loss": 3.1446,
|
| 647 |
+
"step": 86000
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 4.68,
|
| 651 |
+
"grad_norm": 0.3484768271446228,
|
| 652 |
+
"learning_rate": 0.0002514979411764706,
|
| 653 |
+
"loss": 3.1492,
|
| 654 |
+
"step": 87000
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 4.73,
|
| 658 |
+
"grad_norm": 0.34911030530929565,
|
| 659 |
+
"learning_rate": 0.0002506155882352941,
|
| 660 |
+
"loss": 3.1486,
|
| 661 |
+
"step": 88000
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 4.78,
|
| 665 |
+
"grad_norm": 0.349050909280777,
|
| 666 |
+
"learning_rate": 0.0002497332352941176,
|
| 667 |
+
"loss": 3.1433,
|
| 668 |
+
"step": 89000
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 4.84,
|
| 672 |
+
"grad_norm": 0.35526707768440247,
|
| 673 |
+
"learning_rate": 0.00024885176470588235,
|
| 674 |
+
"loss": 3.1434,
|
| 675 |
+
"step": 90000
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 4.89,
|
| 679 |
+
"grad_norm": 0.32889583706855774,
|
| 680 |
+
"learning_rate": 0.00024797029411764704,
|
| 681 |
+
"loss": 3.1436,
|
| 682 |
+
"step": 91000
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 4.95,
|
| 686 |
+
"grad_norm": 0.34612250328063965,
|
| 687 |
+
"learning_rate": 0.0002470879411764706,
|
| 688 |
+
"loss": 3.1419,
|
| 689 |
+
"step": 92000
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 5.0,
|
| 693 |
+
"grad_norm": 0.8200662136077881,
|
| 694 |
+
"learning_rate": 0.00024620647058823527,
|
| 695 |
+
"loss": 3.1424,
|
| 696 |
+
"step": 93000
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 5.0,
|
| 700 |
+
"eval_accuracy": 0.3987985575166068,
|
| 701 |
+
"eval_loss": 3.425178289413452,
|
| 702 |
+
"eval_runtime": 154.096,
|
| 703 |
+
"eval_samples_per_second": 375.87,
|
| 704 |
+
"eval_steps_per_second": 5.873,
|
| 705 |
+
"step": 93000
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 5.05,
|
| 709 |
+
"grad_norm": 0.37201544642448425,
|
| 710 |
+
"learning_rate": 0.0002453241176470588,
|
| 711 |
+
"loss": 3.0771,
|
| 712 |
+
"step": 94000
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 5.11,
|
| 716 |
+
"grad_norm": 0.34167686104774475,
|
| 717 |
+
"learning_rate": 0.0002444417647058823,
|
| 718 |
+
"loss": 3.0806,
|
| 719 |
+
"step": 95000
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 5.16,
|
| 723 |
+
"grad_norm": 0.3650292158126831,
|
| 724 |
+
"learning_rate": 0.00024355941176470585,
|
| 725 |
+
"loss": 3.0847,
|
| 726 |
+
"step": 96000
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 5.22,
|
| 730 |
+
"grad_norm": 0.3709043860435486,
|
| 731 |
+
"learning_rate": 0.00024267794117647056,
|
| 732 |
+
"loss": 3.0885,
|
| 733 |
+
"step": 97000
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 5.27,
|
| 737 |
+
"grad_norm": 0.3401245176792145,
|
| 738 |
+
"learning_rate": 0.0002417955882352941,
|
| 739 |
+
"loss": 3.0927,
|
| 740 |
+
"step": 98000
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 5.32,
|
| 744 |
+
"grad_norm": 0.37579312920570374,
|
| 745 |
+
"learning_rate": 0.00024091323529411765,
|
| 746 |
+
"loss": 3.0929,
|
| 747 |
+
"step": 99000
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 5.38,
|
| 751 |
+
"grad_norm": 0.3333818316459656,
|
| 752 |
+
"learning_rate": 0.00024003176470588234,
|
| 753 |
+
"loss": 3.0932,
|
| 754 |
+
"step": 100000
|
| 755 |
+
},
|
| 756 |
+
{
|
| 757 |
+
"epoch": 5.43,
|
| 758 |
+
"grad_norm": 0.35665878653526306,
|
| 759 |
+
"learning_rate": 0.00023914941176470585,
|
| 760 |
+
"loss": 3.0917,
|
| 761 |
+
"step": 101000
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 5.48,
|
| 765 |
+
"grad_norm": 0.356718510389328,
|
| 766 |
+
"learning_rate": 0.00023826705882352937,
|
| 767 |
+
"loss": 3.0903,
|
| 768 |
+
"step": 102000
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 5.54,
|
| 772 |
+
"grad_norm": 0.3443525433540344,
|
| 773 |
+
"learning_rate": 0.0002373847058823529,
|
| 774 |
+
"loss": 3.0916,
|
| 775 |
+
"step": 103000
|
| 776 |
+
},
|
| 777 |
+
{
|
| 778 |
+
"epoch": 5.59,
|
| 779 |
+
"grad_norm": 0.35697805881500244,
|
| 780 |
+
"learning_rate": 0.00023650323529411763,
|
| 781 |
+
"loss": 3.0911,
|
| 782 |
+
"step": 104000
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 5.65,
|
| 786 |
+
"grad_norm": 0.3541969060897827,
|
| 787 |
+
"learning_rate": 0.00023562176470588234,
|
| 788 |
+
"loss": 3.0943,
|
| 789 |
+
"step": 105000
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 5.7,
|
| 793 |
+
"grad_norm": 0.3476627767086029,
|
| 794 |
+
"learning_rate": 0.00023473941176470586,
|
| 795 |
+
"loss": 3.0968,
|
| 796 |
+
"step": 106000
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"epoch": 5.75,
|
| 800 |
+
"grad_norm": 0.3471356928348541,
|
| 801 |
+
"learning_rate": 0.0002338570588235294,
|
| 802 |
+
"loss": 3.0948,
|
| 803 |
+
"step": 107000
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 5.81,
|
| 807 |
+
"grad_norm": 0.3633851110935211,
|
| 808 |
+
"learning_rate": 0.00023297470588235292,
|
| 809 |
+
"loss": 3.0946,
|
| 810 |
+
"step": 108000
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 5.86,
|
| 814 |
+
"grad_norm": 0.3264976441860199,
|
| 815 |
+
"learning_rate": 0.00023209323529411763,
|
| 816 |
+
"loss": 3.0992,
|
| 817 |
+
"step": 109000
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 5.91,
|
| 821 |
+
"grad_norm": 0.3441069722175598,
|
| 822 |
+
"learning_rate": 0.00023121176470588232,
|
| 823 |
+
"loss": 3.0985,
|
| 824 |
+
"step": 110000
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 5.97,
|
| 828 |
+
"grad_norm": 0.34033203125,
|
| 829 |
+
"learning_rate": 0.00023032941176470584,
|
| 830 |
+
"loss": 3.0983,
|
| 831 |
+
"step": 111000
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 6.0,
|
| 835 |
+
"eval_accuracy": 0.40229210557430617,
|
| 836 |
+
"eval_loss": 3.4145569801330566,
|
| 837 |
+
"eval_runtime": 154.4814,
|
| 838 |
+
"eval_samples_per_second": 374.932,
|
| 839 |
+
"eval_steps_per_second": 5.858,
|
| 840 |
+
"step": 111600
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 6.02,
|
| 844 |
+
"grad_norm": 0.36109837889671326,
|
| 845 |
+
"learning_rate": 0.00022944705882352938,
|
| 846 |
+
"loss": 3.0671,
|
| 847 |
+
"step": 112000
|
| 848 |
+
},
|
| 849 |
+
{
|
| 850 |
+
"epoch": 6.08,
|
| 851 |
+
"grad_norm": 0.35247114300727844,
|
| 852 |
+
"learning_rate": 0.00022856470588235292,
|
| 853 |
+
"loss": 3.0266,
|
| 854 |
+
"step": 113000
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 6.13,
|
| 858 |
+
"grad_norm": 0.3495902419090271,
|
| 859 |
+
"learning_rate": 0.00022768323529411764,
|
| 860 |
+
"loss": 3.0356,
|
| 861 |
+
"step": 114000
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 6.18,
|
| 865 |
+
"grad_norm": 0.34144967794418335,
|
| 866 |
+
"learning_rate": 0.00022680176470588233,
|
| 867 |
+
"loss": 3.0441,
|
| 868 |
+
"step": 115000
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 6.24,
|
| 872 |
+
"grad_norm": 0.3760474920272827,
|
| 873 |
+
"learning_rate": 0.00022591941176470587,
|
| 874 |
+
"loss": 3.043,
|
| 875 |
+
"step": 116000
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 6.29,
|
| 879 |
+
"grad_norm": 0.34772148728370667,
|
| 880 |
+
"learning_rate": 0.00022503705882352942,
|
| 881 |
+
"loss": 3.0446,
|
| 882 |
+
"step": 117000
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 6.34,
|
| 886 |
+
"grad_norm": 0.3700126111507416,
|
| 887 |
+
"learning_rate": 0.00022415470588235293,
|
| 888 |
+
"loss": 3.0467,
|
| 889 |
+
"step": 118000
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 6.4,
|
| 893 |
+
"grad_norm": 0.34513553977012634,
|
| 894 |
+
"learning_rate": 0.00022327323529411762,
|
| 895 |
+
"loss": 3.0465,
|
| 896 |
+
"step": 119000
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 6.45,
|
| 900 |
+
"grad_norm": 0.3503216505050659,
|
| 901 |
+
"learning_rate": 0.00022239176470588234,
|
| 902 |
+
"loss": 3.0513,
|
| 903 |
+
"step": 120000
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 6.51,
|
| 907 |
+
"grad_norm": 0.35930097103118896,
|
| 908 |
+
"learning_rate": 0.00022150941176470585,
|
| 909 |
+
"loss": 3.0512,
|
| 910 |
+
"step": 121000
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 6.56,
|
| 914 |
+
"grad_norm": 0.3546220362186432,
|
| 915 |
+
"learning_rate": 0.0002206270588235294,
|
| 916 |
+
"loss": 3.0504,
|
| 917 |
+
"step": 122000
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 6.61,
|
| 921 |
+
"grad_norm": 0.36518198251724243,
|
| 922 |
+
"learning_rate": 0.0002197447058823529,
|
| 923 |
+
"loss": 3.0584,
|
| 924 |
+
"step": 123000
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 6.67,
|
| 928 |
+
"grad_norm": 0.34408724308013916,
|
| 929 |
+
"learning_rate": 0.00021886323529411763,
|
| 930 |
+
"loss": 3.0513,
|
| 931 |
+
"step": 124000
|
| 932 |
+
},
|
| 933 |
+
{
|
| 934 |
+
"epoch": 6.72,
|
| 935 |
+
"grad_norm": 0.34966805577278137,
|
| 936 |
+
"learning_rate": 0.00021798088235294117,
|
| 937 |
+
"loss": 3.0594,
|
| 938 |
+
"step": 125000
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 6.77,
|
| 942 |
+
"grad_norm": 0.32780104875564575,
|
| 943 |
+
"learning_rate": 0.00021709941176470589,
|
| 944 |
+
"loss": 3.0573,
|
| 945 |
+
"step": 126000
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 6.83,
|
| 949 |
+
"grad_norm": 0.3445553779602051,
|
| 950 |
+
"learning_rate": 0.00021621794117647055,
|
| 951 |
+
"loss": 3.0567,
|
| 952 |
+
"step": 127000
|
| 953 |
+
},
|
| 954 |
+
{
|
| 955 |
+
"epoch": 6.88,
|
| 956 |
+
"grad_norm": 0.3367459177970886,
|
| 957 |
+
"learning_rate": 0.0002153355882352941,
|
| 958 |
+
"loss": 3.0543,
|
| 959 |
+
"step": 128000
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 6.94,
|
| 963 |
+
"grad_norm": 0.33958423137664795,
|
| 964 |
+
"learning_rate": 0.00021445323529411763,
|
| 965 |
+
"loss": 3.0624,
|
| 966 |
+
"step": 129000
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 6.99,
|
| 970 |
+
"grad_norm": 0.3499571979045868,
|
| 971 |
+
"learning_rate": 0.00021357176470588232,
|
| 972 |
+
"loss": 3.061,
|
| 973 |
+
"step": 130000
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 7.0,
|
| 977 |
+
"eval_accuracy": 0.40386995614506527,
|
| 978 |
+
"eval_loss": 3.3960554599761963,
|
| 979 |
+
"eval_runtime": 154.1063,
|
| 980 |
+
"eval_samples_per_second": 375.844,
|
| 981 |
+
"eval_steps_per_second": 5.873,
|
| 982 |
+
"step": 130200
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 7.04,
|
| 986 |
+
"grad_norm": 0.3685757517814636,
|
| 987 |
+
"learning_rate": 0.00021268941176470586,
|
| 988 |
+
"loss": 2.9998,
|
| 989 |
+
"step": 131000
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 7.1,
|
| 993 |
+
"grad_norm": 0.35535717010498047,
|
| 994 |
+
"learning_rate": 0.0002118070588235294,
|
| 995 |
+
"loss": 2.9926,
|
| 996 |
+
"step": 132000
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 7.15,
|
| 1000 |
+
"grad_norm": 0.36104437708854675,
|
| 1001 |
+
"learning_rate": 0.0002109255882352941,
|
| 1002 |
+
"loss": 2.9975,
|
| 1003 |
+
"step": 133000
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 7.2,
|
| 1007 |
+
"grad_norm": 0.36811739206314087,
|
| 1008 |
+
"learning_rate": 0.00021004323529411764,
|
| 1009 |
+
"loss": 3.0098,
|
| 1010 |
+
"step": 134000
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 7.26,
|
| 1014 |
+
"grad_norm": 0.36876553297042847,
|
| 1015 |
+
"learning_rate": 0.00020916088235294118,
|
| 1016 |
+
"loss": 3.0089,
|
| 1017 |
+
"step": 135000
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 7.31,
|
| 1021 |
+
"grad_norm": 0.35547611117362976,
|
| 1022 |
+
"learning_rate": 0.00020827852941176467,
|
| 1023 |
+
"loss": 3.0134,
|
| 1024 |
+
"step": 136000
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 7.37,
|
| 1028 |
+
"grad_norm": 0.3423675000667572,
|
| 1029 |
+
"learning_rate": 0.0002073970588235294,
|
| 1030 |
+
"loss": 3.0142,
|
| 1031 |
+
"step": 137000
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 7.42,
|
| 1035 |
+
"grad_norm": 0.34084853529930115,
|
| 1036 |
+
"learning_rate": 0.0002065155882352941,
|
| 1037 |
+
"loss": 3.0152,
|
| 1038 |
+
"step": 138000
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 7.47,
|
| 1042 |
+
"grad_norm": 0.3496473431587219,
|
| 1043 |
+
"learning_rate": 0.00020563323529411762,
|
| 1044 |
+
"loss": 3.0168,
|
| 1045 |
+
"step": 139000
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 7.53,
|
| 1049 |
+
"grad_norm": 0.3702958822250366,
|
| 1050 |
+
"learning_rate": 0.00020475176470588234,
|
| 1051 |
+
"loss": 3.0159,
|
| 1052 |
+
"step": 140000
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 7.58,
|
| 1056 |
+
"grad_norm": 0.36530712246894836,
|
| 1057 |
+
"learning_rate": 0.00020386941176470588,
|
| 1058 |
+
"loss": 3.0178,
|
| 1059 |
+
"step": 141000
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 7.63,
|
| 1063 |
+
"grad_norm": 0.38187021017074585,
|
| 1064 |
+
"learning_rate": 0.0002029870588235294,
|
| 1065 |
+
"loss": 3.0227,
|
| 1066 |
+
"step": 142000
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 7.69,
|
| 1070 |
+
"grad_norm": 0.3322228491306305,
|
| 1071 |
+
"learning_rate": 0.00020210470588235294,
|
| 1072 |
+
"loss": 3.0172,
|
| 1073 |
+
"step": 143000
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 7.74,
|
| 1077 |
+
"grad_norm": 0.3619314730167389,
|
| 1078 |
+
"learning_rate": 0.00020122235294117648,
|
| 1079 |
+
"loss": 3.0196,
|
| 1080 |
+
"step": 144000
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 7.8,
|
| 1084 |
+
"grad_norm": 0.3536997437477112,
|
| 1085 |
+
"learning_rate": 0.00020034088235294114,
|
| 1086 |
+
"loss": 3.023,
|
| 1087 |
+
"step": 145000
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 7.85,
|
| 1091 |
+
"grad_norm": 0.34041669964790344,
|
| 1092 |
+
"learning_rate": 0.00019945852941176468,
|
| 1093 |
+
"loss": 3.0198,
|
| 1094 |
+
"step": 146000
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 7.9,
|
| 1098 |
+
"grad_norm": 0.35122403502464294,
|
| 1099 |
+
"learning_rate": 0.0001985770588235294,
|
| 1100 |
+
"loss": 3.0192,
|
| 1101 |
+
"step": 147000
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 7.96,
|
| 1105 |
+
"grad_norm": 0.3664146363735199,
|
| 1106 |
+
"learning_rate": 0.00019769470588235292,
|
| 1107 |
+
"loss": 3.0241,
|
| 1108 |
+
"step": 148000
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 8.0,
|
| 1112 |
+
"eval_accuracy": 0.4060721752262614,
|
| 1113 |
+
"eval_loss": 3.367530584335327,
|
| 1114 |
+
"eval_runtime": 153.8583,
|
| 1115 |
+
"eval_samples_per_second": 376.45,
|
| 1116 |
+
"eval_steps_per_second": 5.882,
|
| 1117 |
+
"step": 148800
|
| 1118 |
+
},
|
| 1119 |
+
{
|
| 1120 |
+
"epoch": 8.01,
|
| 1121 |
+
"grad_norm": 0.36462950706481934,
|
| 1122 |
+
"learning_rate": 0.00019681323529411763,
|
| 1123 |
+
"loss": 3.0094,
|
| 1124 |
+
"step": 149000
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 8.06,
|
| 1128 |
+
"grad_norm": 0.3626813590526581,
|
| 1129 |
+
"learning_rate": 0.00019593088235294118,
|
| 1130 |
+
"loss": 2.9568,
|
| 1131 |
+
"step": 150000
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 8.12,
|
| 1135 |
+
"grad_norm": 0.36206358671188354,
|
| 1136 |
+
"learning_rate": 0.0001950485294117647,
|
| 1137 |
+
"loss": 2.9686,
|
| 1138 |
+
"step": 151000
|
| 1139 |
+
},
|
| 1140 |
+
{
|
| 1141 |
+
"epoch": 8.17,
|
| 1142 |
+
"grad_norm": 0.3388898968696594,
|
| 1143 |
+
"learning_rate": 0.0001941670588235294,
|
| 1144 |
+
"loss": 2.9665,
|
| 1145 |
+
"step": 152000
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 8.23,
|
| 1149 |
+
"grad_norm": 0.3698641359806061,
|
| 1150 |
+
"learning_rate": 0.0001932855882352941,
|
| 1151 |
+
"loss": 2.9736,
|
| 1152 |
+
"step": 153000
|
| 1153 |
+
},
|
| 1154 |
+
{
|
| 1155 |
+
"epoch": 8.28,
|
| 1156 |
+
"grad_norm": 0.3924822509288788,
|
| 1157 |
+
"learning_rate": 0.0001924032352941176,
|
| 1158 |
+
"loss": 2.9774,
|
| 1159 |
+
"step": 154000
|
| 1160 |
+
},
|
| 1161 |
+
{
|
| 1162 |
+
"epoch": 8.33,
|
| 1163 |
+
"grad_norm": 0.35343238711357117,
|
| 1164 |
+
"learning_rate": 0.00019152088235294116,
|
| 1165 |
+
"loss": 2.9752,
|
| 1166 |
+
"step": 155000
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 8.39,
|
| 1170 |
+
"grad_norm": 0.3765462636947632,
|
| 1171 |
+
"learning_rate": 0.00019063852941176467,
|
| 1172 |
+
"loss": 2.9832,
|
| 1173 |
+
"step": 156000
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 8.44,
|
| 1177 |
+
"grad_norm": 0.35611894726753235,
|
| 1178 |
+
"learning_rate": 0.00018975617647058821,
|
| 1179 |
+
"loss": 2.9807,
|
| 1180 |
+
"step": 157000
|
| 1181 |
+
},
|
| 1182 |
+
{
|
| 1183 |
+
"epoch": 8.49,
|
| 1184 |
+
"grad_norm": 0.34774050116539,
|
| 1185 |
+
"learning_rate": 0.00018887470588235293,
|
| 1186 |
+
"loss": 2.983,
|
| 1187 |
+
"step": 158000
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 8.55,
|
| 1191 |
+
"grad_norm": 0.35137563943862915,
|
| 1192 |
+
"learning_rate": 0.00018799235294117647,
|
| 1193 |
+
"loss": 2.986,
|
| 1194 |
+
"step": 159000
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 8.6,
|
| 1198 |
+
"grad_norm": 0.3774917423725128,
|
| 1199 |
+
"learning_rate": 0.00018711088235294116,
|
| 1200 |
+
"loss": 2.9858,
|
| 1201 |
+
"step": 160000
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 8.66,
|
| 1205 |
+
"grad_norm": 0.37547916173934937,
|
| 1206 |
+
"learning_rate": 0.0001862285294117647,
|
| 1207 |
+
"loss": 2.9906,
|
| 1208 |
+
"step": 161000
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 8.71,
|
| 1212 |
+
"grad_norm": 0.35136911273002625,
|
| 1213 |
+
"learning_rate": 0.0001853461764705882,
|
| 1214 |
+
"loss": 2.9897,
|
| 1215 |
+
"step": 162000
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 8.76,
|
| 1219 |
+
"grad_norm": 0.36987271904945374,
|
| 1220 |
+
"learning_rate": 0.0001844647058823529,
|
| 1221 |
+
"loss": 2.9947,
|
| 1222 |
+
"step": 163000
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"epoch": 8.82,
|
| 1226 |
+
"grad_norm": 0.35084474086761475,
|
| 1227 |
+
"learning_rate": 0.00018358235294117645,
|
| 1228 |
+
"loss": 2.9918,
|
| 1229 |
+
"step": 164000
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 8.87,
|
| 1233 |
+
"grad_norm": 0.35935935378074646,
|
| 1234 |
+
"learning_rate": 0.00018269999999999997,
|
| 1235 |
+
"loss": 2.9973,
|
| 1236 |
+
"step": 165000
|
| 1237 |
+
},
|
| 1238 |
+
{
|
| 1239 |
+
"epoch": 8.92,
|
| 1240 |
+
"grad_norm": 0.3475947678089142,
|
| 1241 |
+
"learning_rate": 0.00018181852941176468,
|
| 1242 |
+
"loss": 2.9949,
|
| 1243 |
+
"step": 166000
|
| 1244 |
+
},
|
| 1245 |
+
{
|
| 1246 |
+
"epoch": 8.98,
|
| 1247 |
+
"grad_norm": 0.35769519209861755,
|
| 1248 |
+
"learning_rate": 0.00018093617647058823,
|
| 1249 |
+
"loss": 2.9955,
|
| 1250 |
+
"step": 167000
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 9.0,
|
| 1254 |
+
"eval_accuracy": 0.4071023765505084,
|
| 1255 |
+
"eval_loss": 3.368950366973877,
|
| 1256 |
+
"eval_runtime": 154.162,
|
| 1257 |
+
"eval_samples_per_second": 375.709,
|
| 1258 |
+
"eval_steps_per_second": 5.87,
|
| 1259 |
+
"step": 167400
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 9.03,
|
| 1263 |
+
"grad_norm": 0.36424490809440613,
|
| 1264 |
+
"learning_rate": 0.00018005382352941174,
|
| 1265 |
+
"loss": 2.9539,
|
| 1266 |
+
"step": 168000
|
| 1267 |
+
},
|
| 1268 |
+
{
|
| 1269 |
+
"epoch": 9.09,
|
| 1270 |
+
"grad_norm": 0.3711196184158325,
|
| 1271 |
+
"learning_rate": 0.00017917147058823529,
|
| 1272 |
+
"loss": 2.9351,
|
| 1273 |
+
"step": 169000
|
| 1274 |
+
},
|
| 1275 |
+
{
|
| 1276 |
+
"epoch": 9.14,
|
| 1277 |
+
"grad_norm": 0.37114304304122925,
|
| 1278 |
+
"learning_rate": 0.00017828911764705883,
|
| 1279 |
+
"loss": 2.9387,
|
| 1280 |
+
"step": 170000
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 9.19,
|
| 1284 |
+
"grad_norm": 0.37651318311691284,
|
| 1285 |
+
"learning_rate": 0.00017740852941176466,
|
| 1286 |
+
"loss": 2.9445,
|
| 1287 |
+
"step": 171000
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 9.25,
|
| 1291 |
+
"grad_norm": 0.3873383104801178,
|
| 1292 |
+
"learning_rate": 0.0001765261764705882,
|
| 1293 |
+
"loss": 2.9402,
|
| 1294 |
+
"step": 172000
|
| 1295 |
+
},
|
| 1296 |
+
{
|
| 1297 |
+
"epoch": 9.3,
|
| 1298 |
+
"grad_norm": 0.3732360303401947,
|
| 1299 |
+
"learning_rate": 0.00017564382352941175,
|
| 1300 |
+
"loss": 2.9515,
|
| 1301 |
+
"step": 173000
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 9.35,
|
| 1305 |
+
"grad_norm": 0.3868846595287323,
|
| 1306 |
+
"learning_rate": 0.00017476235294117647,
|
| 1307 |
+
"loss": 2.9519,
|
| 1308 |
+
"step": 174000
|
| 1309 |
+
},
|
| 1310 |
+
{
|
| 1311 |
+
"epoch": 9.41,
|
| 1312 |
+
"grad_norm": 0.3689320981502533,
|
| 1313 |
+
"learning_rate": 0.00017387999999999998,
|
| 1314 |
+
"loss": 2.9559,
|
| 1315 |
+
"step": 175000
|
| 1316 |
+
},
|
| 1317 |
+
{
|
| 1318 |
+
"epoch": 9.46,
|
| 1319 |
+
"grad_norm": 0.3811360001564026,
|
| 1320 |
+
"learning_rate": 0.00017299764705882352,
|
| 1321 |
+
"loss": 2.9543,
|
| 1322 |
+
"step": 176000
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 9.52,
|
| 1326 |
+
"grad_norm": 0.3691989481449127,
|
| 1327 |
+
"learning_rate": 0.00017211529411764704,
|
| 1328 |
+
"loss": 2.9546,
|
| 1329 |
+
"step": 177000
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 9.57,
|
| 1333 |
+
"grad_norm": 0.35540491342544556,
|
| 1334 |
+
"learning_rate": 0.00017123470588235293,
|
| 1335 |
+
"loss": 2.9598,
|
| 1336 |
+
"step": 178000
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"epoch": 9.62,
|
| 1340 |
+
"grad_norm": 0.36182084679603577,
|
| 1341 |
+
"learning_rate": 0.00017035235294117645,
|
| 1342 |
+
"loss": 2.9608,
|
| 1343 |
+
"step": 179000
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 9.68,
|
| 1347 |
+
"grad_norm": 0.3745622932910919,
|
| 1348 |
+
"learning_rate": 0.00016946999999999996,
|
| 1349 |
+
"loss": 2.9601,
|
| 1350 |
+
"step": 180000
|
| 1351 |
+
},
|
| 1352 |
+
{
|
| 1353 |
+
"epoch": 9.73,
|
| 1354 |
+
"grad_norm": 0.3744942247867584,
|
| 1355 |
+
"learning_rate": 0.0001685876470588235,
|
| 1356 |
+
"loss": 2.962,
|
| 1357 |
+
"step": 181000
|
| 1358 |
+
},
|
| 1359 |
+
{
|
| 1360 |
+
"epoch": 9.78,
|
| 1361 |
+
"grad_norm": 0.3739373981952667,
|
| 1362 |
+
"learning_rate": 0.00016770617647058822,
|
| 1363 |
+
"loss": 2.969,
|
| 1364 |
+
"step": 182000
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 9.84,
|
| 1368 |
+
"grad_norm": 0.38088759779930115,
|
| 1369 |
+
"learning_rate": 0.00016682382352941174,
|
| 1370 |
+
"loss": 2.9639,
|
| 1371 |
+
"step": 183000
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 9.89,
|
| 1375 |
+
"grad_norm": 0.36885249614715576,
|
| 1376 |
+
"learning_rate": 0.00016594147058823528,
|
| 1377 |
+
"loss": 2.9691,
|
| 1378 |
+
"step": 184000
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 9.95,
|
| 1382 |
+
"grad_norm": 0.37970590591430664,
|
| 1383 |
+
"learning_rate": 0.00016506,
|
| 1384 |
+
"loss": 2.9675,
|
| 1385 |
+
"step": 185000
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 10.0,
|
| 1389 |
+
"grad_norm": 0.8675829172134399,
|
| 1390 |
+
"learning_rate": 0.00016417764705882354,
|
| 1391 |
+
"loss": 2.971,
|
| 1392 |
+
"step": 186000
|
| 1393 |
+
},
|
| 1394 |
+
{
|
| 1395 |
+
"epoch": 10.0,
|
| 1396 |
+
"eval_accuracy": 0.4077077331943161,
|
| 1397 |
+
"eval_loss": 3.366788625717163,
|
| 1398 |
+
"eval_runtime": 154.5461,
|
| 1399 |
+
"eval_samples_per_second": 374.775,
|
| 1400 |
+
"eval_steps_per_second": 5.856,
|
| 1401 |
+
"step": 186000
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 10.05,
|
| 1405 |
+
"grad_norm": 0.3905154764652252,
|
| 1406 |
+
"learning_rate": 0.00016329529411764705,
|
| 1407 |
+
"loss": 2.9004,
|
| 1408 |
+
"step": 187000
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"epoch": 10.11,
|
| 1412 |
+
"grad_norm": 0.3730432093143463,
|
| 1413 |
+
"learning_rate": 0.0001624129411764706,
|
| 1414 |
+
"loss": 2.9095,
|
| 1415 |
+
"step": 188000
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 10.16,
|
| 1419 |
+
"grad_norm": 0.3711690306663513,
|
| 1420 |
+
"learning_rate": 0.00016153147058823526,
|
| 1421 |
+
"loss": 2.9136,
|
| 1422 |
+
"step": 189000
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 10.22,
|
| 1426 |
+
"grad_norm": 0.399265855550766,
|
| 1427 |
+
"learning_rate": 0.00016064999999999997,
|
| 1428 |
+
"loss": 2.9178,
|
| 1429 |
+
"step": 190000
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 10.27,
|
| 1433 |
+
"grad_norm": 0.3668770492076874,
|
| 1434 |
+
"learning_rate": 0.00015976764705882352,
|
| 1435 |
+
"loss": 2.9229,
|
| 1436 |
+
"step": 191000
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 10.32,
|
| 1440 |
+
"grad_norm": 0.37367990612983704,
|
| 1441 |
+
"learning_rate": 0.00015888529411764703,
|
| 1442 |
+
"loss": 2.9243,
|
| 1443 |
+
"step": 192000
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 10.38,
|
| 1447 |
+
"grad_norm": 0.385092169046402,
|
| 1448 |
+
"learning_rate": 0.00015800382352941175,
|
| 1449 |
+
"loss": 2.9262,
|
| 1450 |
+
"step": 193000
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 10.43,
|
| 1454 |
+
"grad_norm": 0.3799744248390198,
|
| 1455 |
+
"learning_rate": 0.00015712235294117647,
|
| 1456 |
+
"loss": 2.9308,
|
| 1457 |
+
"step": 194000
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 10.48,
|
| 1461 |
+
"grad_norm": 0.3523291051387787,
|
| 1462 |
+
"learning_rate": 0.00015624,
|
| 1463 |
+
"loss": 2.93,
|
| 1464 |
+
"step": 195000
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 10.54,
|
| 1468 |
+
"grad_norm": 0.38555893301963806,
|
| 1469 |
+
"learning_rate": 0.00015535764705882352,
|
| 1470 |
+
"loss": 2.9368,
|
| 1471 |
+
"step": 196000
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 10.59,
|
| 1475 |
+
"grad_norm": 0.4180794954299927,
|
| 1476 |
+
"learning_rate": 0.00015447529411764704,
|
| 1477 |
+
"loss": 2.9374,
|
| 1478 |
+
"step": 197000
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 10.65,
|
| 1482 |
+
"grad_norm": 0.3763350248336792,
|
| 1483 |
+
"learning_rate": 0.00015359294117647056,
|
| 1484 |
+
"loss": 2.9371,
|
| 1485 |
+
"step": 198000
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 10.7,
|
| 1489 |
+
"grad_norm": 0.3720884919166565,
|
| 1490 |
+
"learning_rate": 0.00015271147058823527,
|
| 1491 |
+
"loss": 2.9383,
|
| 1492 |
+
"step": 199000
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"epoch": 10.75,
|
| 1496 |
+
"grad_norm": 0.4129893481731415,
|
| 1497 |
+
"learning_rate": 0.00015182911764705882,
|
| 1498 |
+
"loss": 2.9436,
|
| 1499 |
+
"step": 200000
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 10.81,
|
| 1503 |
+
"grad_norm": 0.3609713613986969,
|
| 1504 |
+
"learning_rate": 0.00015094676470588233,
|
| 1505 |
+
"loss": 2.9392,
|
| 1506 |
+
"step": 201000
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 10.86,
|
| 1510 |
+
"grad_norm": 0.3772646486759186,
|
| 1511 |
+
"learning_rate": 0.00015006441176470587,
|
| 1512 |
+
"loss": 2.9428,
|
| 1513 |
+
"step": 202000
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 10.91,
|
| 1517 |
+
"grad_norm": 0.37511298060417175,
|
| 1518 |
+
"learning_rate": 0.0001491829411764706,
|
| 1519 |
+
"loss": 2.941,
|
| 1520 |
+
"step": 203000
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 10.97,
|
| 1524 |
+
"grad_norm": 0.37095674872398376,
|
| 1525 |
+
"learning_rate": 0.00014830147058823528,
|
| 1526 |
+
"loss": 2.9425,
|
| 1527 |
+
"step": 204000
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 11.0,
|
| 1531 |
+
"eval_accuracy": 0.40833525915256785,
|
| 1532 |
+
"eval_loss": 3.3716514110565186,
|
| 1533 |
+
"eval_runtime": 154.5582,
|
| 1534 |
+
"eval_samples_per_second": 374.746,
|
| 1535 |
+
"eval_steps_per_second": 5.855,
|
| 1536 |
+
"step": 204600
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 11.02,
|
| 1540 |
+
"grad_norm": 0.4002268612384796,
|
| 1541 |
+
"learning_rate": 0.00014741911764705882,
|
| 1542 |
+
"loss": 2.9159,
|
| 1543 |
+
"step": 205000
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 11.08,
|
| 1547 |
+
"grad_norm": 0.38958606123924255,
|
| 1548 |
+
"learning_rate": 0.00014653676470588234,
|
| 1549 |
+
"loss": 2.8845,
|
| 1550 |
+
"step": 206000
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 11.13,
|
| 1554 |
+
"grad_norm": 0.39512166380882263,
|
| 1555 |
+
"learning_rate": 0.00014565529411764705,
|
| 1556 |
+
"loss": 2.888,
|
| 1557 |
+
"step": 207000
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 11.18,
|
| 1561 |
+
"grad_norm": 0.38474199175834656,
|
| 1562 |
+
"learning_rate": 0.00014477294117647057,
|
| 1563 |
+
"loss": 2.8945,
|
| 1564 |
+
"step": 208000
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 11.24,
|
| 1568 |
+
"grad_norm": 0.3705930709838867,
|
| 1569 |
+
"learning_rate": 0.00014389147058823529,
|
| 1570 |
+
"loss": 2.9003,
|
| 1571 |
+
"step": 209000
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 11.29,
|
| 1575 |
+
"grad_norm": 0.3930906057357788,
|
| 1576 |
+
"learning_rate": 0.0001430091176470588,
|
| 1577 |
+
"loss": 2.8984,
|
| 1578 |
+
"step": 210000
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 11.34,
|
| 1582 |
+
"grad_norm": 0.38158994913101196,
|
| 1583 |
+
"learning_rate": 0.00014212676470588234,
|
| 1584 |
+
"loss": 2.9031,
|
| 1585 |
+
"step": 211000
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 11.4,
|
| 1589 |
+
"grad_norm": 0.4204154312610626,
|
| 1590 |
+
"learning_rate": 0.0001412444117647059,
|
| 1591 |
+
"loss": 2.9029,
|
| 1592 |
+
"step": 212000
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 11.45,
|
| 1596 |
+
"grad_norm": 0.3909905254840851,
|
| 1597 |
+
"learning_rate": 0.00014036294117647058,
|
| 1598 |
+
"loss": 2.9031,
|
| 1599 |
+
"step": 213000
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 11.51,
|
| 1603 |
+
"grad_norm": 0.38486233353614807,
|
| 1604 |
+
"learning_rate": 0.0001394805882352941,
|
| 1605 |
+
"loss": 2.9061,
|
| 1606 |
+
"step": 214000
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 11.56,
|
| 1610 |
+
"grad_norm": 0.3670654594898224,
|
| 1611 |
+
"learning_rate": 0.0001385991176470588,
|
| 1612 |
+
"loss": 2.9114,
|
| 1613 |
+
"step": 215000
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 11.61,
|
| 1617 |
+
"grad_norm": 0.38542190194129944,
|
| 1618 |
+
"learning_rate": 0.00013771676470588235,
|
| 1619 |
+
"loss": 2.913,
|
| 1620 |
+
"step": 216000
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 11.67,
|
| 1624 |
+
"grad_norm": 0.4009072184562683,
|
| 1625 |
+
"learning_rate": 0.00013683441176470587,
|
| 1626 |
+
"loss": 2.9139,
|
| 1627 |
+
"step": 217000
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 11.72,
|
| 1631 |
+
"grad_norm": 0.37454643845558167,
|
| 1632 |
+
"learning_rate": 0.0001359520588235294,
|
| 1633 |
+
"loss": 2.9155,
|
| 1634 |
+
"step": 218000
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 11.77,
|
| 1638 |
+
"grad_norm": 0.38453200459480286,
|
| 1639 |
+
"learning_rate": 0.00013506970588235293,
|
| 1640 |
+
"loss": 2.9198,
|
| 1641 |
+
"step": 219000
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 11.83,
|
| 1645 |
+
"grad_norm": 0.38210123777389526,
|
| 1646 |
+
"learning_rate": 0.00013418735294117647,
|
| 1647 |
+
"loss": 2.9175,
|
| 1648 |
+
"step": 220000
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 11.88,
|
| 1652 |
+
"grad_norm": 0.38491418957710266,
|
| 1653 |
+
"learning_rate": 0.00013330676470588233,
|
| 1654 |
+
"loss": 2.9218,
|
| 1655 |
+
"step": 221000
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 11.94,
|
| 1659 |
+
"grad_norm": 0.40083280205726624,
|
| 1660 |
+
"learning_rate": 0.00013242441176470587,
|
| 1661 |
+
"loss": 2.9197,
|
| 1662 |
+
"step": 222000
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 11.99,
|
| 1666 |
+
"grad_norm": 0.37707728147506714,
|
| 1667 |
+
"learning_rate": 0.0001315420588235294,
|
| 1668 |
+
"loss": 2.9175,
|
| 1669 |
+
"step": 223000
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 12.0,
|
| 1673 |
+
"eval_accuracy": 0.4085028054259733,
|
| 1674 |
+
"eval_loss": 3.383648157119751,
|
| 1675 |
+
"eval_runtime": 153.9053,
|
| 1676 |
+
"eval_samples_per_second": 376.335,
|
| 1677 |
+
"eval_steps_per_second": 5.88,
|
| 1678 |
+
"step": 223200
|
| 1679 |
+
},
|
| 1680 |
+
{
|
| 1681 |
+
"epoch": 12.04,
|
| 1682 |
+
"grad_norm": 0.3987793028354645,
|
| 1683 |
+
"learning_rate": 0.00013065970588235293,
|
| 1684 |
+
"loss": 2.8729,
|
| 1685 |
+
"step": 224000
|
| 1686 |
+
},
|
| 1687 |
+
{
|
| 1688 |
+
"epoch": 12.1,
|
| 1689 |
+
"grad_norm": 0.4100884199142456,
|
| 1690 |
+
"learning_rate": 0.00012977823529411765,
|
| 1691 |
+
"loss": 2.8662,
|
| 1692 |
+
"step": 225000
|
| 1693 |
+
},
|
| 1694 |
+
{
|
| 1695 |
+
"epoch": 12.15,
|
| 1696 |
+
"grad_norm": 0.4005632996559143,
|
| 1697 |
+
"learning_rate": 0.00012889676470588234,
|
| 1698 |
+
"loss": 2.8718,
|
| 1699 |
+
"step": 226000
|
| 1700 |
+
},
|
| 1701 |
+
{
|
| 1702 |
+
"epoch": 12.2,
|
| 1703 |
+
"grad_norm": 0.4106426537036896,
|
| 1704 |
+
"learning_rate": 0.00012801441176470588,
|
| 1705 |
+
"loss": 2.8702,
|
| 1706 |
+
"step": 227000
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 12.26,
|
| 1710 |
+
"grad_norm": 0.4060533046722412,
|
| 1711 |
+
"learning_rate": 0.0001271320588235294,
|
| 1712 |
+
"loss": 2.8727,
|
| 1713 |
+
"step": 228000
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 12.31,
|
| 1717 |
+
"grad_norm": 0.40115979313850403,
|
| 1718 |
+
"learning_rate": 0.0001262505882352941,
|
| 1719 |
+
"loss": 2.8805,
|
| 1720 |
+
"step": 229000
|
| 1721 |
+
},
|
| 1722 |
+
{
|
| 1723 |
+
"epoch": 12.37,
|
| 1724 |
+
"grad_norm": 0.3902336359024048,
|
| 1725 |
+
"learning_rate": 0.00012536823529411763,
|
| 1726 |
+
"loss": 2.8781,
|
| 1727 |
+
"step": 230000
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 12.42,
|
| 1731 |
+
"grad_norm": 0.40399476885795593,
|
| 1732 |
+
"learning_rate": 0.00012448588235294117,
|
| 1733 |
+
"loss": 2.8814,
|
| 1734 |
+
"step": 231000
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 12.47,
|
| 1738 |
+
"grad_norm": 0.39339208602905273,
|
| 1739 |
+
"learning_rate": 0.0001236044117647059,
|
| 1740 |
+
"loss": 2.8871,
|
| 1741 |
+
"step": 232000
|
| 1742 |
+
},
|
| 1743 |
+
{
|
| 1744 |
+
"epoch": 12.53,
|
| 1745 |
+
"grad_norm": 0.38773778080940247,
|
| 1746 |
+
"learning_rate": 0.0001227220588235294,
|
| 1747 |
+
"loss": 2.8931,
|
| 1748 |
+
"step": 233000
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 12.58,
|
| 1752 |
+
"grad_norm": 0.4028586447238922,
|
| 1753 |
+
"learning_rate": 0.00012183970588235292,
|
| 1754 |
+
"loss": 2.892,
|
| 1755 |
+
"step": 234000
|
| 1756 |
+
},
|
| 1757 |
+
{
|
| 1758 |
+
"epoch": 12.63,
|
| 1759 |
+
"grad_norm": 0.38995110988616943,
|
| 1760 |
+
"learning_rate": 0.00012095823529411763,
|
| 1761 |
+
"loss": 2.892,
|
| 1762 |
+
"step": 235000
|
| 1763 |
+
},
|
| 1764 |
+
{
|
| 1765 |
+
"epoch": 12.69,
|
| 1766 |
+
"grad_norm": 0.41991814970970154,
|
| 1767 |
+
"learning_rate": 0.00012007588235294116,
|
| 1768 |
+
"loss": 2.8932,
|
| 1769 |
+
"step": 236000
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 12.74,
|
| 1773 |
+
"grad_norm": 0.39780813455581665,
|
| 1774 |
+
"learning_rate": 0.0001191935294117647,
|
| 1775 |
+
"loss": 2.8923,
|
| 1776 |
+
"step": 237000
|
| 1777 |
+
},
|
| 1778 |
+
{
|
| 1779 |
+
"epoch": 12.8,
|
| 1780 |
+
"grad_norm": 0.40335002541542053,
|
| 1781 |
+
"learning_rate": 0.00011831117647058824,
|
| 1782 |
+
"loss": 2.8947,
|
| 1783 |
+
"step": 238000
|
| 1784 |
+
},
|
| 1785 |
+
{
|
| 1786 |
+
"epoch": 12.85,
|
| 1787 |
+
"grad_norm": 0.4120156168937683,
|
| 1788 |
+
"learning_rate": 0.00011742970588235293,
|
| 1789 |
+
"loss": 2.8976,
|
| 1790 |
+
"step": 239000
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 12.9,
|
| 1794 |
+
"grad_norm": 0.4305998682975769,
|
| 1795 |
+
"learning_rate": 0.00011654735294117645,
|
| 1796 |
+
"loss": 2.8974,
|
| 1797 |
+
"step": 240000
|
| 1798 |
+
},
|
| 1799 |
+
{
|
| 1800 |
+
"epoch": 12.96,
|
| 1801 |
+
"grad_norm": 0.3948858976364136,
|
| 1802 |
+
"learning_rate": 0.00011566499999999998,
|
| 1803 |
+
"loss": 2.8993,
|
| 1804 |
+
"step": 241000
|
| 1805 |
+
},
|
| 1806 |
+
{
|
| 1807 |
+
"epoch": 13.0,
|
| 1808 |
+
"eval_accuracy": 0.4095500704043683,
|
| 1809 |
+
"eval_loss": 3.368474006652832,
|
| 1810 |
+
"eval_runtime": 154.1284,
|
| 1811 |
+
"eval_samples_per_second": 375.791,
|
| 1812 |
+
"eval_steps_per_second": 5.872,
|
| 1813 |
+
"step": 241800
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 13.01,
|
| 1817 |
+
"grad_norm": 0.40451323986053467,
|
| 1818 |
+
"learning_rate": 0.0001147835294117647,
|
| 1819 |
+
"loss": 2.8895,
|
| 1820 |
+
"step": 242000
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 13.06,
|
| 1824 |
+
"grad_norm": 0.40097764134407043,
|
| 1825 |
+
"learning_rate": 0.00011390117647058822,
|
| 1826 |
+
"loss": 2.8413,
|
| 1827 |
+
"step": 243000
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 13.12,
|
| 1831 |
+
"grad_norm": 0.41899681091308594,
|
| 1832 |
+
"learning_rate": 0.00011301970588235293,
|
| 1833 |
+
"loss": 2.8457,
|
| 1834 |
+
"step": 244000
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 13.17,
|
| 1838 |
+
"grad_norm": 0.41395074129104614,
|
| 1839 |
+
"learning_rate": 0.00011213735294117646,
|
| 1840 |
+
"loss": 2.8478,
|
| 1841 |
+
"step": 245000
|
| 1842 |
+
},
|
| 1843 |
+
{
|
| 1844 |
+
"epoch": 13.23,
|
| 1845 |
+
"grad_norm": 0.4069136083126068,
|
| 1846 |
+
"learning_rate": 0.00011125499999999999,
|
| 1847 |
+
"loss": 2.8546,
|
| 1848 |
+
"step": 246000
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 13.28,
|
| 1852 |
+
"grad_norm": 0.4198252260684967,
|
| 1853 |
+
"learning_rate": 0.00011037264705882352,
|
| 1854 |
+
"loss": 2.8586,
|
| 1855 |
+
"step": 247000
|
| 1856 |
+
},
|
| 1857 |
+
{
|
| 1858 |
+
"epoch": 13.33,
|
| 1859 |
+
"grad_norm": 0.4080910384654999,
|
| 1860 |
+
"learning_rate": 0.00010949117647058822,
|
| 1861 |
+
"loss": 2.8595,
|
| 1862 |
+
"step": 248000
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 13.39,
|
| 1866 |
+
"grad_norm": 0.4152771532535553,
|
| 1867 |
+
"learning_rate": 0.00010860970588235294,
|
| 1868 |
+
"loss": 2.8632,
|
| 1869 |
+
"step": 249000
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 13.44,
|
| 1873 |
+
"grad_norm": 0.38884419202804565,
|
| 1874 |
+
"learning_rate": 0.00010772735294117647,
|
| 1875 |
+
"loss": 2.8635,
|
| 1876 |
+
"step": 250000
|
| 1877 |
+
},
|
| 1878 |
+
{
|
| 1879 |
+
"epoch": 13.49,
|
| 1880 |
+
"grad_norm": 0.4043819010257721,
|
| 1881 |
+
"learning_rate": 0.000106845,
|
| 1882 |
+
"loss": 2.8614,
|
| 1883 |
+
"step": 251000
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 13.55,
|
| 1887 |
+
"grad_norm": 0.39720532298088074,
|
| 1888 |
+
"learning_rate": 0.00010596352941176469,
|
| 1889 |
+
"loss": 2.8706,
|
| 1890 |
+
"step": 252000
|
| 1891 |
+
},
|
| 1892 |
+
{
|
| 1893 |
+
"epoch": 13.6,
|
| 1894 |
+
"grad_norm": 0.408447802066803,
|
| 1895 |
+
"learning_rate": 0.00010508117647058823,
|
| 1896 |
+
"loss": 2.8703,
|
| 1897 |
+
"step": 253000
|
| 1898 |
+
},
|
| 1899 |
+
{
|
| 1900 |
+
"epoch": 13.66,
|
| 1901 |
+
"grad_norm": 0.43144288659095764,
|
| 1902 |
+
"learning_rate": 0.00010419882352941176,
|
| 1903 |
+
"loss": 2.8704,
|
| 1904 |
+
"step": 254000
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 13.71,
|
| 1908 |
+
"grad_norm": 0.4294928312301636,
|
| 1909 |
+
"learning_rate": 0.00010331735294117646,
|
| 1910 |
+
"loss": 2.8696,
|
| 1911 |
+
"step": 255000
|
| 1912 |
+
},
|
| 1913 |
+
{
|
| 1914 |
+
"epoch": 13.76,
|
| 1915 |
+
"grad_norm": 0.39523276686668396,
|
| 1916 |
+
"learning_rate": 0.00010243499999999998,
|
| 1917 |
+
"loss": 2.874,
|
| 1918 |
+
"step": 256000
|
| 1919 |
+
},
|
| 1920 |
+
{
|
| 1921 |
+
"epoch": 13.82,
|
| 1922 |
+
"grad_norm": 0.3995026648044586,
|
| 1923 |
+
"learning_rate": 0.00010155352941176469,
|
| 1924 |
+
"loss": 2.8786,
|
| 1925 |
+
"step": 257000
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"epoch": 13.87,
|
| 1929 |
+
"grad_norm": 0.4120596647262573,
|
| 1930 |
+
"learning_rate": 0.00010067117647058822,
|
| 1931 |
+
"loss": 2.8783,
|
| 1932 |
+
"step": 258000
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"epoch": 13.92,
|
| 1936 |
+
"grad_norm": 0.40986043214797974,
|
| 1937 |
+
"learning_rate": 9.978882352941177e-05,
|
| 1938 |
+
"loss": 2.8766,
|
| 1939 |
+
"step": 259000
|
| 1940 |
+
},
|
| 1941 |
+
{
|
| 1942 |
+
"epoch": 13.98,
|
| 1943 |
+
"grad_norm": 0.39782214164733887,
|
| 1944 |
+
"learning_rate": 9.89064705882353e-05,
|
| 1945 |
+
"loss": 2.8802,
|
| 1946 |
+
"step": 260000
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 14.0,
|
| 1950 |
+
"eval_accuracy": 0.4093544429993336,
|
| 1951 |
+
"eval_loss": 3.386864423751831,
|
| 1952 |
+
"eval_runtime": 154.0798,
|
| 1953 |
+
"eval_samples_per_second": 375.909,
|
| 1954 |
+
"eval_steps_per_second": 5.874,
|
| 1955 |
+
"step": 260400
|
| 1956 |
+
},
|
| 1957 |
+
{
|
| 1958 |
+
"epoch": 14.03,
|
| 1959 |
+
"grad_norm": 0.42637449502944946,
|
| 1960 |
+
"learning_rate": 9.802499999999998e-05,
|
| 1961 |
+
"loss": 2.8424,
|
| 1962 |
+
"step": 261000
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 14.09,
|
| 1966 |
+
"grad_norm": 0.4381449818611145,
|
| 1967 |
+
"learning_rate": 9.714264705882351e-05,
|
| 1968 |
+
"loss": 2.827,
|
| 1969 |
+
"step": 262000
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 14.14,
|
| 1973 |
+
"grad_norm": 0.41219663619995117,
|
| 1974 |
+
"learning_rate": 9.626117647058823e-05,
|
| 1975 |
+
"loss": 2.8299,
|
| 1976 |
+
"step": 263000
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 14.19,
|
| 1980 |
+
"grad_norm": 0.3997000455856323,
|
| 1981 |
+
"learning_rate": 9.537882352941176e-05,
|
| 1982 |
+
"loss": 2.8347,
|
| 1983 |
+
"step": 264000
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 14.25,
|
| 1987 |
+
"grad_norm": 0.4551340639591217,
|
| 1988 |
+
"learning_rate": 9.449647058823527e-05,
|
| 1989 |
+
"loss": 2.8429,
|
| 1990 |
+
"step": 265000
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 14.3,
|
| 1994 |
+
"grad_norm": 0.42454150319099426,
|
| 1995 |
+
"learning_rate": 9.361411764705882e-05,
|
| 1996 |
+
"loss": 2.8375,
|
| 1997 |
+
"step": 266000
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 14.35,
|
| 2001 |
+
"grad_norm": 0.4199308454990387,
|
| 2002 |
+
"learning_rate": 9.273264705882352e-05,
|
| 2003 |
+
"loss": 2.8383,
|
| 2004 |
+
"step": 267000
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 14.41,
|
| 2008 |
+
"grad_norm": 0.4250849485397339,
|
| 2009 |
+
"learning_rate": 9.185117647058824e-05,
|
| 2010 |
+
"loss": 2.8448,
|
| 2011 |
+
"step": 268000
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 14.46,
|
| 2015 |
+
"grad_norm": 0.4406766891479492,
|
| 2016 |
+
"learning_rate": 9.096882352941175e-05,
|
| 2017 |
+
"loss": 2.8431,
|
| 2018 |
+
"step": 269000
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 14.52,
|
| 2022 |
+
"grad_norm": 0.40331459045410156,
|
| 2023 |
+
"learning_rate": 9.008647058823528e-05,
|
| 2024 |
+
"loss": 2.8485,
|
| 2025 |
+
"step": 270000
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 14.57,
|
| 2029 |
+
"grad_norm": 0.4189225137233734,
|
| 2030 |
+
"learning_rate": 8.920411764705881e-05,
|
| 2031 |
+
"loss": 2.8496,
|
| 2032 |
+
"step": 271000
|
| 2033 |
+
},
|
| 2034 |
+
{
|
| 2035 |
+
"epoch": 14.62,
|
| 2036 |
+
"grad_norm": 0.4331813454627991,
|
| 2037 |
+
"learning_rate": 8.832264705882353e-05,
|
| 2038 |
+
"loss": 2.8533,
|
| 2039 |
+
"step": 272000
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 14.68,
|
| 2043 |
+
"grad_norm": 0.4302915930747986,
|
| 2044 |
+
"learning_rate": 8.744029411764706e-05,
|
| 2045 |
+
"loss": 2.8538,
|
| 2046 |
+
"step": 273000
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 14.73,
|
| 2050 |
+
"grad_norm": 0.4052475094795227,
|
| 2051 |
+
"learning_rate": 8.655794117647057e-05,
|
| 2052 |
+
"loss": 2.8511,
|
| 2053 |
+
"step": 274000
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 14.78,
|
| 2057 |
+
"grad_norm": 0.41182762384414673,
|
| 2058 |
+
"learning_rate": 8.56755882352941e-05,
|
| 2059 |
+
"loss": 2.8544,
|
| 2060 |
+
"step": 275000
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 14.84,
|
| 2064 |
+
"grad_norm": 0.40615832805633545,
|
| 2065 |
+
"learning_rate": 8.479411764705882e-05,
|
| 2066 |
+
"loss": 2.8531,
|
| 2067 |
+
"step": 276000
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 14.89,
|
| 2071 |
+
"grad_norm": 0.4004068374633789,
|
| 2072 |
+
"learning_rate": 8.391176470588235e-05,
|
| 2073 |
+
"loss": 2.8566,
|
| 2074 |
+
"step": 277000
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 14.95,
|
| 2078 |
+
"grad_norm": 0.4328298270702362,
|
| 2079 |
+
"learning_rate": 8.302941176470588e-05,
|
| 2080 |
+
"loss": 2.8582,
|
| 2081 |
+
"step": 278000
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 15.0,
|
| 2085 |
+
"grad_norm": 1.0608173608779907,
|
| 2086 |
+
"learning_rate": 8.21470588235294e-05,
|
| 2087 |
+
"loss": 2.8591,
|
| 2088 |
+
"step": 279000
|
| 2089 |
+
},
|
| 2090 |
+
{
|
| 2091 |
+
"epoch": 15.0,
|
| 2092 |
+
"eval_accuracy": 0.40931400079540814,
|
| 2093 |
+
"eval_loss": 3.390340566635132,
|
| 2094 |
+
"eval_runtime": 154.3289,
|
| 2095 |
+
"eval_samples_per_second": 375.302,
|
| 2096 |
+
"eval_steps_per_second": 5.864,
|
| 2097 |
+
"step": 279000
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 15.05,
|
| 2101 |
+
"grad_norm": 0.4320835471153259,
|
| 2102 |
+
"learning_rate": 8.126558823529411e-05,
|
| 2103 |
+
"loss": 2.8083,
|
| 2104 |
+
"step": 280000
|
| 2105 |
+
},
|
| 2106 |
+
{
|
| 2107 |
+
"epoch": 15.11,
|
| 2108 |
+
"grad_norm": 0.43483027815818787,
|
| 2109 |
+
"learning_rate": 8.038411764705882e-05,
|
| 2110 |
+
"loss": 2.8112,
|
| 2111 |
+
"step": 281000
|
| 2112 |
+
},
|
| 2113 |
+
{
|
| 2114 |
+
"epoch": 15.16,
|
| 2115 |
+
"grad_norm": 0.4238964021205902,
|
| 2116 |
+
"learning_rate": 7.950176470588235e-05,
|
| 2117 |
+
"loss": 2.8149,
|
| 2118 |
+
"step": 282000
|
| 2119 |
+
},
|
| 2120 |
+
{
|
| 2121 |
+
"epoch": 15.22,
|
| 2122 |
+
"grad_norm": 0.40951046347618103,
|
| 2123 |
+
"learning_rate": 7.862029411764704e-05,
|
| 2124 |
+
"loss": 2.8164,
|
| 2125 |
+
"step": 283000
|
| 2126 |
+
},
|
| 2127 |
+
{
|
| 2128 |
+
"epoch": 15.27,
|
| 2129 |
+
"grad_norm": 0.4388927221298218,
|
| 2130 |
+
"learning_rate": 7.773794117647059e-05,
|
| 2131 |
+
"loss": 2.8185,
|
| 2132 |
+
"step": 284000
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 15.32,
|
| 2136 |
+
"grad_norm": 0.42049768567085266,
|
| 2137 |
+
"learning_rate": 7.685647058823529e-05,
|
| 2138 |
+
"loss": 2.8229,
|
| 2139 |
+
"step": 285000
|
| 2140 |
+
},
|
| 2141 |
+
{
|
| 2142 |
+
"epoch": 15.38,
|
| 2143 |
+
"grad_norm": 0.43744581937789917,
|
| 2144 |
+
"learning_rate": 7.597411764705883e-05,
|
| 2145 |
+
"loss": 2.8277,
|
| 2146 |
+
"step": 286000
|
| 2147 |
+
},
|
| 2148 |
+
{
|
| 2149 |
+
"epoch": 15.43,
|
| 2150 |
+
"grad_norm": 0.45774605870246887,
|
| 2151 |
+
"learning_rate": 7.509176470588235e-05,
|
| 2152 |
+
"loss": 2.8205,
|
| 2153 |
+
"step": 287000
|
| 2154 |
+
},
|
| 2155 |
+
{
|
| 2156 |
+
"epoch": 15.48,
|
| 2157 |
+
"grad_norm": 0.4613402783870697,
|
| 2158 |
+
"learning_rate": 7.421029411764705e-05,
|
| 2159 |
+
"loss": 2.8287,
|
| 2160 |
+
"step": 288000
|
| 2161 |
+
},
|
| 2162 |
+
{
|
| 2163 |
+
"epoch": 15.54,
|
| 2164 |
+
"grad_norm": 0.4393397569656372,
|
| 2165 |
+
"learning_rate": 7.332794117647058e-05,
|
| 2166 |
+
"loss": 2.8314,
|
| 2167 |
+
"step": 289000
|
| 2168 |
+
},
|
| 2169 |
+
{
|
| 2170 |
+
"epoch": 15.59,
|
| 2171 |
+
"grad_norm": 0.4375053942203522,
|
| 2172 |
+
"learning_rate": 7.244558823529411e-05,
|
| 2173 |
+
"loss": 2.8311,
|
| 2174 |
+
"step": 290000
|
| 2175 |
+
},
|
| 2176 |
+
{
|
| 2177 |
+
"epoch": 15.65,
|
| 2178 |
+
"grad_norm": 0.42573311924934387,
|
| 2179 |
+
"learning_rate": 7.156411764705882e-05,
|
| 2180 |
+
"loss": 2.8296,
|
| 2181 |
+
"step": 291000
|
| 2182 |
+
},
|
| 2183 |
+
{
|
| 2184 |
+
"epoch": 15.7,
|
| 2185 |
+
"grad_norm": 0.43389272689819336,
|
| 2186 |
+
"learning_rate": 7.068264705882353e-05,
|
| 2187 |
+
"loss": 2.8284,
|
| 2188 |
+
"step": 292000
|
| 2189 |
+
},
|
| 2190 |
+
{
|
| 2191 |
+
"epoch": 15.75,
|
| 2192 |
+
"grad_norm": 0.43407490849494934,
|
| 2193 |
+
"learning_rate": 6.980029411764706e-05,
|
| 2194 |
+
"loss": 2.8322,
|
| 2195 |
+
"step": 293000
|
| 2196 |
+
},
|
| 2197 |
+
{
|
| 2198 |
+
"epoch": 15.81,
|
| 2199 |
+
"grad_norm": 0.42439669370651245,
|
| 2200 |
+
"learning_rate": 6.891794117647059e-05,
|
| 2201 |
+
"loss": 2.8386,
|
| 2202 |
+
"step": 294000
|
| 2203 |
+
},
|
| 2204 |
+
{
|
| 2205 |
+
"epoch": 15.86,
|
| 2206 |
+
"grad_norm": 0.4558635354042053,
|
| 2207 |
+
"learning_rate": 6.803558823529411e-05,
|
| 2208 |
+
"loss": 2.837,
|
| 2209 |
+
"step": 295000
|
| 2210 |
+
},
|
| 2211 |
+
{
|
| 2212 |
+
"epoch": 15.91,
|
| 2213 |
+
"grad_norm": 0.4102713167667389,
|
| 2214 |
+
"learning_rate": 6.715323529411764e-05,
|
| 2215 |
+
"loss": 2.8375,
|
| 2216 |
+
"step": 296000
|
| 2217 |
+
},
|
| 2218 |
+
{
|
| 2219 |
+
"epoch": 15.97,
|
| 2220 |
+
"grad_norm": 0.4180239737033844,
|
| 2221 |
+
"learning_rate": 6.627176470588235e-05,
|
| 2222 |
+
"loss": 2.8397,
|
| 2223 |
+
"step": 297000
|
| 2224 |
+
},
|
| 2225 |
+
{
|
| 2226 |
+
"epoch": 16.0,
|
| 2227 |
+
"eval_accuracy": 0.4098871111636606,
|
| 2228 |
+
"eval_loss": 3.3899056911468506,
|
| 2229 |
+
"eval_runtime": 154.0201,
|
| 2230 |
+
"eval_samples_per_second": 376.055,
|
| 2231 |
+
"eval_steps_per_second": 5.876,
|
| 2232 |
+
"step": 297600
|
| 2233 |
+
},
|
| 2234 |
+
{
|
| 2235 |
+
"epoch": 16.02,
|
| 2236 |
+
"grad_norm": 0.446920782327652,
|
| 2237 |
+
"learning_rate": 6.538941176470588e-05,
|
| 2238 |
+
"loss": 2.8222,
|
| 2239 |
+
"step": 298000
|
| 2240 |
+
},
|
| 2241 |
+
{
|
| 2242 |
+
"epoch": 16.08,
|
| 2243 |
+
"grad_norm": 0.4178001582622528,
|
| 2244 |
+
"learning_rate": 6.45070588235294e-05,
|
| 2245 |
+
"loss": 2.7924,
|
| 2246 |
+
"step": 299000
|
| 2247 |
+
},
|
| 2248 |
+
{
|
| 2249 |
+
"epoch": 16.13,
|
| 2250 |
+
"grad_norm": 0.45794540643692017,
|
| 2251 |
+
"learning_rate": 6.362558823529411e-05,
|
| 2252 |
+
"loss": 2.7949,
|
| 2253 |
+
"step": 300000
|
| 2254 |
+
},
|
| 2255 |
+
{
|
| 2256 |
+
"epoch": 16.18,
|
| 2257 |
+
"grad_norm": 0.4420151114463806,
|
| 2258 |
+
"learning_rate": 6.274323529411764e-05,
|
| 2259 |
+
"loss": 2.7997,
|
| 2260 |
+
"step": 301000
|
| 2261 |
+
},
|
| 2262 |
+
{
|
| 2263 |
+
"epoch": 16.24,
|
| 2264 |
+
"grad_norm": 0.4682018458843231,
|
| 2265 |
+
"learning_rate": 6.186088235294117e-05,
|
| 2266 |
+
"loss": 2.8017,
|
| 2267 |
+
"step": 302000
|
| 2268 |
+
},
|
| 2269 |
+
{
|
| 2270 |
+
"epoch": 16.29,
|
| 2271 |
+
"grad_norm": 0.44723376631736755,
|
| 2272 |
+
"learning_rate": 6.09785294117647e-05,
|
| 2273 |
+
"loss": 2.807,
|
| 2274 |
+
"step": 303000
|
| 2275 |
+
},
|
| 2276 |
+
{
|
| 2277 |
+
"epoch": 16.34,
|
| 2278 |
+
"grad_norm": 0.44858965277671814,
|
| 2279 |
+
"learning_rate": 6.0097941176470585e-05,
|
| 2280 |
+
"loss": 2.8091,
|
| 2281 |
+
"step": 304000
|
| 2282 |
+
},
|
| 2283 |
+
{
|
| 2284 |
+
"epoch": 16.4,
|
| 2285 |
+
"grad_norm": 0.43858635425567627,
|
| 2286 |
+
"learning_rate": 5.921558823529411e-05,
|
| 2287 |
+
"loss": 2.8055,
|
| 2288 |
+
"step": 305000
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 16.45,
|
| 2292 |
+
"grad_norm": 0.4233960509300232,
|
| 2293 |
+
"learning_rate": 5.8333235294117644e-05,
|
| 2294 |
+
"loss": 2.8104,
|
| 2295 |
+
"step": 306000
|
| 2296 |
+
},
|
| 2297 |
+
{
|
| 2298 |
+
"epoch": 16.51,
|
| 2299 |
+
"grad_norm": 0.4392797648906708,
|
| 2300 |
+
"learning_rate": 5.745088235294117e-05,
|
| 2301 |
+
"loss": 2.8122,
|
| 2302 |
+
"step": 307000
|
| 2303 |
+
},
|
| 2304 |
+
{
|
| 2305 |
+
"epoch": 16.56,
|
| 2306 |
+
"grad_norm": 0.4333525002002716,
|
| 2307 |
+
"learning_rate": 5.6568529411764696e-05,
|
| 2308 |
+
"loss": 2.8119,
|
| 2309 |
+
"step": 308000
|
| 2310 |
+
},
|
| 2311 |
+
{
|
| 2312 |
+
"epoch": 16.61,
|
| 2313 |
+
"grad_norm": 0.4427085518836975,
|
| 2314 |
+
"learning_rate": 5.568617647058823e-05,
|
| 2315 |
+
"loss": 2.8139,
|
| 2316 |
+
"step": 309000
|
| 2317 |
+
},
|
| 2318 |
+
{
|
| 2319 |
+
"epoch": 16.67,
|
| 2320 |
+
"grad_norm": 0.445634663105011,
|
| 2321 |
+
"learning_rate": 5.4804705882352935e-05,
|
| 2322 |
+
"loss": 2.8149,
|
| 2323 |
+
"step": 310000
|
| 2324 |
+
},
|
| 2325 |
+
{
|
| 2326 |
+
"epoch": 16.72,
|
| 2327 |
+
"grad_norm": 0.4446145296096802,
|
| 2328 |
+
"learning_rate": 5.3922352941176464e-05,
|
| 2329 |
+
"loss": 2.8152,
|
| 2330 |
+
"step": 311000
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 16.77,
|
| 2334 |
+
"grad_norm": 0.43026256561279297,
|
| 2335 |
+
"learning_rate": 5.3040882352941166e-05,
|
| 2336 |
+
"loss": 2.8141,
|
| 2337 |
+
"step": 312000
|
| 2338 |
+
},
|
| 2339 |
+
{
|
| 2340 |
+
"epoch": 16.83,
|
| 2341 |
+
"grad_norm": 0.4380964934825897,
|
| 2342 |
+
"learning_rate": 5.21585294117647e-05,
|
| 2343 |
+
"loss": 2.8157,
|
| 2344 |
+
"step": 313000
|
| 2345 |
+
},
|
| 2346 |
+
{
|
| 2347 |
+
"epoch": 16.88,
|
| 2348 |
+
"grad_norm": 0.4361090362071991,
|
| 2349 |
+
"learning_rate": 5.1277058823529405e-05,
|
| 2350 |
+
"loss": 2.8152,
|
| 2351 |
+
"step": 314000
|
| 2352 |
+
},
|
| 2353 |
+
{
|
| 2354 |
+
"epoch": 16.94,
|
| 2355 |
+
"grad_norm": 0.4253467321395874,
|
| 2356 |
+
"learning_rate": 5.0394705882352935e-05,
|
| 2357 |
+
"loss": 2.8229,
|
| 2358 |
+
"step": 315000
|
| 2359 |
+
},
|
| 2360 |
+
{
|
| 2361 |
+
"epoch": 16.99,
|
| 2362 |
+
"grad_norm": 0.4476911425590515,
|
| 2363 |
+
"learning_rate": 4.951323529411764e-05,
|
| 2364 |
+
"loss": 2.8158,
|
| 2365 |
+
"step": 316000
|
| 2366 |
+
},
|
| 2367 |
+
{
|
| 2368 |
+
"epoch": 17.0,
|
| 2369 |
+
"eval_accuracy": 0.40948181578777654,
|
| 2370 |
+
"eval_loss": 3.3992156982421875,
|
| 2371 |
+
"eval_runtime": 154.2196,
|
| 2372 |
+
"eval_samples_per_second": 375.568,
|
| 2373 |
+
"eval_steps_per_second": 5.868,
|
| 2374 |
+
"step": 316200
|
| 2375 |
+
},
|
| 2376 |
+
{
|
| 2377 |
+
"epoch": 17.04,
|
| 2378 |
+
"grad_norm": 0.43660619854927063,
|
| 2379 |
+
"learning_rate": 4.863088235294117e-05,
|
| 2380 |
+
"loss": 2.7885,
|
| 2381 |
+
"step": 317000
|
| 2382 |
+
},
|
| 2383 |
+
{
|
| 2384 |
+
"epoch": 17.1,
|
| 2385 |
+
"grad_norm": 0.46034106612205505,
|
| 2386 |
+
"learning_rate": 4.7749411764705875e-05,
|
| 2387 |
+
"loss": 2.7768,
|
| 2388 |
+
"step": 318000
|
| 2389 |
+
},
|
| 2390 |
+
{
|
| 2391 |
+
"epoch": 17.15,
|
| 2392 |
+
"grad_norm": 0.43989402055740356,
|
| 2393 |
+
"learning_rate": 4.6867058823529405e-05,
|
| 2394 |
+
"loss": 2.7883,
|
| 2395 |
+
"step": 319000
|
| 2396 |
+
},
|
| 2397 |
+
{
|
| 2398 |
+
"epoch": 17.2,
|
| 2399 |
+
"grad_norm": 0.4530872702598572,
|
| 2400 |
+
"learning_rate": 4.598470588235294e-05,
|
| 2401 |
+
"loss": 2.786,
|
| 2402 |
+
"step": 320000
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 17.26,
|
| 2406 |
+
"grad_norm": 0.4461182653903961,
|
| 2407 |
+
"learning_rate": 4.5103235294117643e-05,
|
| 2408 |
+
"loss": 2.7879,
|
| 2409 |
+
"step": 321000
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 17.31,
|
| 2413 |
+
"grad_norm": 0.4403826892375946,
|
| 2414 |
+
"learning_rate": 4.422088235294117e-05,
|
| 2415 |
+
"loss": 2.7923,
|
| 2416 |
+
"step": 322000
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 17.37,
|
| 2420 |
+
"grad_norm": 0.4319212734699249,
|
| 2421 |
+
"learning_rate": 4.33385294117647e-05,
|
| 2422 |
+
"loss": 2.7894,
|
| 2423 |
+
"step": 323000
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 17.42,
|
| 2427 |
+
"grad_norm": 0.4447585344314575,
|
| 2428 |
+
"learning_rate": 4.245617647058823e-05,
|
| 2429 |
+
"loss": 2.7921,
|
| 2430 |
+
"step": 324000
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 17.47,
|
| 2434 |
+
"grad_norm": 0.4347590208053589,
|
| 2435 |
+
"learning_rate": 4.157382352941177e-05,
|
| 2436 |
+
"loss": 2.7921,
|
| 2437 |
+
"step": 325000
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 17.53,
|
| 2441 |
+
"grad_norm": 0.42081958055496216,
|
| 2442 |
+
"learning_rate": 4.069235294117647e-05,
|
| 2443 |
+
"loss": 2.7944,
|
| 2444 |
+
"step": 326000
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 17.58,
|
| 2448 |
+
"grad_norm": 0.4446498155593872,
|
| 2449 |
+
"learning_rate": 3.981e-05,
|
| 2450 |
+
"loss": 2.7944,
|
| 2451 |
+
"step": 327000
|
| 2452 |
+
},
|
| 2453 |
+
{
|
| 2454 |
+
"epoch": 17.63,
|
| 2455 |
+
"grad_norm": 0.44777339696884155,
|
| 2456 |
+
"learning_rate": 3.89285294117647e-05,
|
| 2457 |
+
"loss": 2.7971,
|
| 2458 |
+
"step": 328000
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 17.69,
|
| 2462 |
+
"grad_norm": 0.4410764276981354,
|
| 2463 |
+
"learning_rate": 3.804617647058824e-05,
|
| 2464 |
+
"loss": 2.7989,
|
| 2465 |
+
"step": 329000
|
| 2466 |
+
},
|
| 2467 |
+
{
|
| 2468 |
+
"epoch": 17.74,
|
| 2469 |
+
"grad_norm": 0.4594448208808899,
|
| 2470 |
+
"learning_rate": 3.716470588235294e-05,
|
| 2471 |
+
"loss": 2.7943,
|
| 2472 |
+
"step": 330000
|
| 2473 |
+
},
|
| 2474 |
+
{
|
| 2475 |
+
"epoch": 17.8,
|
| 2476 |
+
"grad_norm": 0.4540853798389435,
|
| 2477 |
+
"learning_rate": 3.628323529411764e-05,
|
| 2478 |
+
"loss": 2.7992,
|
| 2479 |
+
"step": 331000
|
| 2480 |
+
},
|
| 2481 |
+
{
|
| 2482 |
+
"epoch": 17.85,
|
| 2483 |
+
"grad_norm": 0.4680405259132385,
|
| 2484 |
+
"learning_rate": 3.540088235294117e-05,
|
| 2485 |
+
"loss": 2.7983,
|
| 2486 |
+
"step": 332000
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 17.9,
|
| 2490 |
+
"grad_norm": 0.43989479541778564,
|
| 2491 |
+
"learning_rate": 3.451941176470588e-05,
|
| 2492 |
+
"loss": 2.7966,
|
| 2493 |
+
"step": 333000
|
| 2494 |
+
},
|
| 2495 |
+
{
|
| 2496 |
+
"epoch": 17.96,
|
| 2497 |
+
"grad_norm": 0.4248621463775635,
|
| 2498 |
+
"learning_rate": 3.363705882352941e-05,
|
| 2499 |
+
"loss": 2.7994,
|
| 2500 |
+
"step": 334000
|
| 2501 |
+
},
|
| 2502 |
+
{
|
| 2503 |
+
"epoch": 18.0,
|
| 2504 |
+
"eval_accuracy": 0.4089663456370789,
|
| 2505 |
+
"eval_loss": 3.412942409515381,
|
| 2506 |
+
"eval_runtime": 154.3857,
|
| 2507 |
+
"eval_samples_per_second": 375.164,
|
| 2508 |
+
"eval_steps_per_second": 5.862,
|
| 2509 |
+
"step": 334800
|
| 2510 |
+
},
|
| 2511 |
+
{
|
| 2512 |
+
"epoch": 18.01,
|
| 2513 |
+
"grad_norm": 0.4774115979671478,
|
| 2514 |
+
"learning_rate": 3.2754705882352934e-05,
|
| 2515 |
+
"loss": 2.7919,
|
| 2516 |
+
"step": 335000
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 18.06,
|
| 2520 |
+
"grad_norm": 0.45676735043525696,
|
| 2521 |
+
"learning_rate": 3.187323529411764e-05,
|
| 2522 |
+
"loss": 2.7683,
|
| 2523 |
+
"step": 336000
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 18.12,
|
| 2527 |
+
"grad_norm": 0.43984270095825195,
|
| 2528 |
+
"learning_rate": 3.099088235294117e-05,
|
| 2529 |
+
"loss": 2.7712,
|
| 2530 |
+
"step": 337000
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"epoch": 18.17,
|
| 2534 |
+
"grad_norm": 0.4677587151527405,
|
| 2535 |
+
"learning_rate": 3.0108529411764705e-05,
|
| 2536 |
+
"loss": 2.773,
|
| 2537 |
+
"step": 338000
|
| 2538 |
+
},
|
| 2539 |
+
{
|
| 2540 |
+
"epoch": 18.23,
|
| 2541 |
+
"grad_norm": 0.4494146704673767,
|
| 2542 |
+
"learning_rate": 2.9227058823529408e-05,
|
| 2543 |
+
"loss": 2.7703,
|
| 2544 |
+
"step": 339000
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"epoch": 18.28,
|
| 2548 |
+
"grad_norm": 0.4525000751018524,
|
| 2549 |
+
"learning_rate": 2.834470588235294e-05,
|
| 2550 |
+
"loss": 2.7731,
|
| 2551 |
+
"step": 340000
|
| 2552 |
+
},
|
| 2553 |
+
{
|
| 2554 |
+
"epoch": 18.33,
|
| 2555 |
+
"grad_norm": 0.45488810539245605,
|
| 2556 |
+
"learning_rate": 2.7463235294117643e-05,
|
| 2557 |
+
"loss": 2.7729,
|
| 2558 |
+
"step": 341000
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 18.39,
|
| 2562 |
+
"grad_norm": 0.4790239930152893,
|
| 2563 |
+
"learning_rate": 2.6580882352941176e-05,
|
| 2564 |
+
"loss": 2.7788,
|
| 2565 |
+
"step": 342000
|
| 2566 |
+
},
|
| 2567 |
+
{
|
| 2568 |
+
"epoch": 18.44,
|
| 2569 |
+
"grad_norm": 0.44312784075737,
|
| 2570 |
+
"learning_rate": 2.5698529411764705e-05,
|
| 2571 |
+
"loss": 2.7814,
|
| 2572 |
+
"step": 343000
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"epoch": 18.49,
|
| 2576 |
+
"grad_norm": 0.45970168709754944,
|
| 2577 |
+
"learning_rate": 2.481705882352941e-05,
|
| 2578 |
+
"loss": 2.7782,
|
| 2579 |
+
"step": 344000
|
| 2580 |
+
},
|
| 2581 |
+
{
|
| 2582 |
+
"epoch": 18.55,
|
| 2583 |
+
"grad_norm": 0.489257276058197,
|
| 2584 |
+
"learning_rate": 2.393470588235294e-05,
|
| 2585 |
+
"loss": 2.7795,
|
| 2586 |
+
"step": 345000
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"epoch": 18.6,
|
| 2590 |
+
"grad_norm": 0.46976953744888306,
|
| 2591 |
+
"learning_rate": 2.3052352941176467e-05,
|
| 2592 |
+
"loss": 2.7802,
|
| 2593 |
+
"step": 346000
|
| 2594 |
+
},
|
| 2595 |
+
{
|
| 2596 |
+
"epoch": 18.66,
|
| 2597 |
+
"grad_norm": 0.4579465985298157,
|
| 2598 |
+
"learning_rate": 2.2169999999999996e-05,
|
| 2599 |
+
"loss": 2.7777,
|
| 2600 |
+
"step": 347000
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 18.71,
|
| 2604 |
+
"grad_norm": 0.4497370719909668,
|
| 2605 |
+
"learning_rate": 2.1288529411764702e-05,
|
| 2606 |
+
"loss": 2.7802,
|
| 2607 |
+
"step": 348000
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 18.76,
|
| 2611 |
+
"grad_norm": 0.4875960648059845,
|
| 2612 |
+
"learning_rate": 2.0406176470588235e-05,
|
| 2613 |
+
"loss": 2.777,
|
| 2614 |
+
"step": 349000
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"epoch": 18.82,
|
| 2618 |
+
"grad_norm": 0.4670606255531311,
|
| 2619 |
+
"learning_rate": 1.9524705882352937e-05,
|
| 2620 |
+
"loss": 2.7818,
|
| 2621 |
+
"step": 350000
|
| 2622 |
+
},
|
| 2623 |
+
{
|
| 2624 |
+
"epoch": 18.87,
|
| 2625 |
+
"grad_norm": 0.44986578822135925,
|
| 2626 |
+
"learning_rate": 1.864235294117647e-05,
|
| 2627 |
+
"loss": 2.7842,
|
| 2628 |
+
"step": 351000
|
| 2629 |
+
},
|
| 2630 |
+
{
|
| 2631 |
+
"epoch": 18.92,
|
| 2632 |
+
"grad_norm": 0.4309654235839844,
|
| 2633 |
+
"learning_rate": 1.7760882352941176e-05,
|
| 2634 |
+
"loss": 2.778,
|
| 2635 |
+
"step": 352000
|
| 2636 |
+
},
|
| 2637 |
+
{
|
| 2638 |
+
"epoch": 18.98,
|
| 2639 |
+
"grad_norm": 0.4595903158187866,
|
| 2640 |
+
"learning_rate": 1.6878529411764705e-05,
|
| 2641 |
+
"loss": 2.7773,
|
| 2642 |
+
"step": 353000
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 19.0,
|
| 2646 |
+
"eval_accuracy": 0.40923123535481654,
|
| 2647 |
+
"eval_loss": 3.421142339706421,
|
| 2648 |
+
"eval_runtime": 154.3823,
|
| 2649 |
+
"eval_samples_per_second": 375.172,
|
| 2650 |
+
"eval_steps_per_second": 5.862,
|
| 2651 |
+
"step": 353400
|
| 2652 |
+
},
|
| 2653 |
+
{
|
| 2654 |
+
"epoch": 19.03,
|
| 2655 |
+
"grad_norm": 0.4593981206417084,
|
| 2656 |
+
"learning_rate": 1.599705882352941e-05,
|
| 2657 |
+
"loss": 2.7664,
|
| 2658 |
+
"step": 354000
|
| 2659 |
+
},
|
| 2660 |
+
{
|
| 2661 |
+
"epoch": 19.09,
|
| 2662 |
+
"grad_norm": 0.4566170275211334,
|
| 2663 |
+
"learning_rate": 1.5114705882352939e-05,
|
| 2664 |
+
"loss": 2.7573,
|
| 2665 |
+
"step": 355000
|
| 2666 |
+
},
|
| 2667 |
+
{
|
| 2668 |
+
"epoch": 19.14,
|
| 2669 |
+
"grad_norm": 0.475852370262146,
|
| 2670 |
+
"learning_rate": 1.423235294117647e-05,
|
| 2671 |
+
"loss": 2.7558,
|
| 2672 |
+
"step": 356000
|
| 2673 |
+
},
|
| 2674 |
+
{
|
| 2675 |
+
"epoch": 19.19,
|
| 2676 |
+
"grad_norm": 0.4608045518398285,
|
| 2677 |
+
"learning_rate": 1.3349999999999998e-05,
|
| 2678 |
+
"loss": 2.7629,
|
| 2679 |
+
"step": 357000
|
| 2680 |
+
},
|
| 2681 |
+
{
|
| 2682 |
+
"epoch": 19.25,
|
| 2683 |
+
"grad_norm": 0.4460165500640869,
|
| 2684 |
+
"learning_rate": 1.2467647058823529e-05,
|
| 2685 |
+
"loss": 2.7636,
|
| 2686 |
+
"step": 358000
|
| 2687 |
+
},
|
| 2688 |
+
{
|
| 2689 |
+
"epoch": 19.3,
|
| 2690 |
+
"grad_norm": 0.46552643179893494,
|
| 2691 |
+
"learning_rate": 1.1587058823529412e-05,
|
| 2692 |
+
"loss": 2.7604,
|
| 2693 |
+
"step": 359000
|
| 2694 |
+
},
|
| 2695 |
+
{
|
| 2696 |
+
"epoch": 19.35,
|
| 2697 |
+
"grad_norm": 0.43840447068214417,
|
| 2698 |
+
"learning_rate": 1.070470588235294e-05,
|
| 2699 |
+
"loss": 2.7645,
|
| 2700 |
+
"step": 360000
|
| 2701 |
+
},
|
| 2702 |
+
{
|
| 2703 |
+
"epoch": 19.41,
|
| 2704 |
+
"grad_norm": 0.46308407187461853,
|
| 2705 |
+
"learning_rate": 9.822352941176468e-06,
|
| 2706 |
+
"loss": 2.7647,
|
| 2707 |
+
"step": 361000
|
| 2708 |
+
},
|
| 2709 |
+
{
|
| 2710 |
+
"epoch": 19.46,
|
| 2711 |
+
"grad_norm": 0.4742732048034668,
|
| 2712 |
+
"learning_rate": 8.939999999999999e-06,
|
| 2713 |
+
"loss": 2.7654,
|
| 2714 |
+
"step": 362000
|
| 2715 |
+
},
|
| 2716 |
+
{
|
| 2717 |
+
"epoch": 19.52,
|
| 2718 |
+
"grad_norm": 0.47486451268196106,
|
| 2719 |
+
"learning_rate": 8.058529411764705e-06,
|
| 2720 |
+
"loss": 2.7646,
|
| 2721 |
+
"step": 363000
|
| 2722 |
+
},
|
| 2723 |
+
{
|
| 2724 |
+
"epoch": 19.57,
|
| 2725 |
+
"grad_norm": 0.46561646461486816,
|
| 2726 |
+
"learning_rate": 7.176176470588234e-06,
|
| 2727 |
+
"loss": 2.7654,
|
| 2728 |
+
"step": 364000
|
| 2729 |
+
},
|
| 2730 |
+
{
|
| 2731 |
+
"epoch": 19.62,
|
| 2732 |
+
"grad_norm": 0.46853920817375183,
|
| 2733 |
+
"learning_rate": 6.295588235294118e-06,
|
| 2734 |
+
"loss": 2.7647,
|
| 2735 |
+
"step": 365000
|
| 2736 |
+
},
|
| 2737 |
+
{
|
| 2738 |
+
"epoch": 19.68,
|
| 2739 |
+
"grad_norm": 0.48070281744003296,
|
| 2740 |
+
"learning_rate": 5.413235294117647e-06,
|
| 2741 |
+
"loss": 2.7639,
|
| 2742 |
+
"step": 366000
|
| 2743 |
+
},
|
| 2744 |
+
{
|
| 2745 |
+
"epoch": 19.73,
|
| 2746 |
+
"grad_norm": 0.46360719203948975,
|
| 2747 |
+
"learning_rate": 4.530882352941176e-06,
|
| 2748 |
+
"loss": 2.7618,
|
| 2749 |
+
"step": 367000
|
| 2750 |
+
},
|
| 2751 |
+
{
|
| 2752 |
+
"epoch": 19.78,
|
| 2753 |
+
"grad_norm": 0.43898245692253113,
|
| 2754 |
+
"learning_rate": 3.6494117647058817e-06,
|
| 2755 |
+
"loss": 2.7675,
|
| 2756 |
+
"step": 368000
|
| 2757 |
+
},
|
| 2758 |
+
{
|
| 2759 |
+
"epoch": 19.84,
|
| 2760 |
+
"grad_norm": 0.44373300671577454,
|
| 2761 |
+
"learning_rate": 2.7670588235294112e-06,
|
| 2762 |
+
"loss": 2.7632,
|
| 2763 |
+
"step": 369000
|
| 2764 |
+
},
|
| 2765 |
+
{
|
| 2766 |
+
"epoch": 19.89,
|
| 2767 |
+
"grad_norm": 0.45887884497642517,
|
| 2768 |
+
"learning_rate": 1.8847058823529411e-06,
|
| 2769 |
+
"loss": 2.7626,
|
| 2770 |
+
"step": 370000
|
| 2771 |
+
},
|
| 2772 |
+
{
|
| 2773 |
+
"epoch": 19.95,
|
| 2774 |
+
"grad_norm": 0.45596691966056824,
|
| 2775 |
+
"learning_rate": 1.0023529411764706e-06,
|
| 2776 |
+
"loss": 2.7648,
|
| 2777 |
+
"step": 371000
|
| 2778 |
+
},
|
| 2779 |
+
{
|
| 2780 |
+
"epoch": 20.0,
|
| 2781 |
+
"grad_norm": 1.0607672929763794,
|
| 2782 |
+
"learning_rate": 1.2088235294117647e-07,
|
| 2783 |
+
"loss": 2.7599,
|
| 2784 |
+
"step": 372000
|
| 2785 |
+
},
|
| 2786 |
+
{
|
| 2787 |
+
"epoch": 20.0,
|
| 2788 |
+
"eval_accuracy": 0.4091656007481136,
|
| 2789 |
+
"eval_loss": 3.4249629974365234,
|
| 2790 |
+
"eval_runtime": 154.4535,
|
| 2791 |
+
"eval_samples_per_second": 375.0,
|
| 2792 |
+
"eval_steps_per_second": 5.859,
|
| 2793 |
+
"step": 372000
|
| 2794 |
+
},
|
| 2795 |
+
{
|
| 2796 |
+
"epoch": 20.0,
|
| 2797 |
+
"step": 372000,
|
| 2798 |
+
"total_flos": 1.56728915394048e+18,
|
| 2799 |
+
"train_loss": 3.055753937752016,
|
| 2800 |
+
"train_runtime": 81288.5778,
|
| 2801 |
+
"train_samples_per_second": 146.435,
|
| 2802 |
+
"train_steps_per_second": 4.576
|
| 2803 |
+
}
|
| 2804 |
+
],
|
| 2805 |
+
"logging_steps": 1000,
|
| 2806 |
+
"max_steps": 372000,
|
| 2807 |
+
"num_input_tokens_seen": 0,
|
| 2808 |
+
"num_train_epochs": 20,
|
| 2809 |
+
"save_steps": 5000,
|
| 2810 |
+
"total_flos": 1.56728915394048e+18,
|
| 2811 |
+
"train_batch_size": 32,
|
| 2812 |
+
"trial_name": null,
|
| 2813 |
+
"trial_params": null
|
| 2814 |
+
}
|