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_high_variability_numeral-seed_211-1e-3
|
| 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_high_variability_numeral-seed_211-1e-3
|
| 15 |
|
| 16 |
-
This model was trained from scratch on
|
| 17 |
It achieves the following results on the evaluation set:
|
| 18 |
- Loss: 3.4248
|
| 19 |
- Accuracy: 0.4100
|
|
|
|
| 1 |
---
|
| 2 |
tags:
|
| 3 |
- generated_from_trainer
|
| 4 |
+
datasets:
|
| 5 |
+
- kanishka/counterfactual_babylm_aann_high_variability_numeral
|
| 6 |
metrics:
|
| 7 |
- accuracy
|
| 8 |
model-index:
|
| 9 |
- name: smolm-autoreg-bpe-counterfactual_babylm_aann_high_variability_numeral-seed_211-1e-3
|
| 10 |
+
results:
|
| 11 |
+
- task:
|
| 12 |
+
name: Causal Language Modeling
|
| 13 |
+
type: text-generation
|
| 14 |
+
dataset:
|
| 15 |
+
name: kanishka/counterfactual_babylm_aann_high_variability_numeral
|
| 16 |
+
type: kanishka/counterfactual_babylm_aann_high_variability_numeral
|
| 17 |
+
metrics:
|
| 18 |
+
- name: Accuracy
|
| 19 |
+
type: accuracy
|
| 20 |
+
value: 0.40996444059320364
|
| 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_high_variability_numeral-seed_211-1e-3
|
| 27 |
|
| 28 |
+
This model was trained from scratch on the kanishka/counterfactual_babylm_aann_high_variability_numeral dataset.
|
| 29 |
It achieves the following results on the evaluation set:
|
| 30 |
- Loss: 3.4248
|
| 31 |
- Accuracy: 0.4100
|
all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"eval_accuracy": 0.40996444059320364,
|
| 4 |
+
"eval_loss": 3.424755573272705,
|
| 5 |
+
"eval_runtime": 154.2813,
|
| 6 |
+
"eval_samples": 57917,
|
| 7 |
+
"eval_samples_per_second": 375.399,
|
| 8 |
+
"eval_steps_per_second": 5.866,
|
| 9 |
+
"perplexity": 30.715136333646623,
|
| 10 |
+
"train_loss": 3.0296430246833577,
|
| 11 |
+
"train_runtime": 81436.2617,
|
| 12 |
+
"train_samples": 595065,
|
| 13 |
+
"train_samples_per_second": 146.143,
|
| 14 |
+
"train_steps_per_second": 4.567
|
| 15 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"eval_accuracy": 0.40996444059320364,
|
| 4 |
+
"eval_loss": 3.424755573272705,
|
| 5 |
+
"eval_runtime": 154.2813,
|
| 6 |
+
"eval_samples": 57917,
|
| 7 |
+
"eval_samples_per_second": 375.399,
|
| 8 |
+
"eval_steps_per_second": 5.866,
|
| 9 |
+
"perplexity": 30.715136333646623
|
| 10 |
+
}
|
runs/Jul15_12-37-22_phyl-ling-p01.la.utexas.edu/events.out.tfevents.1721147010.phyl-ling-p01.la.utexas.edu.2780613.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:19ed5edbd3a4a768b9c5ef2f45a93df60a2ddf5341b63f3be480ceb2a963432f
|
| 3 |
+
size 417
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"train_loss": 3.0296430246833577,
|
| 4 |
+
"train_runtime": 81436.2617,
|
| 5 |
+
"train_samples": 595065,
|
| 6 |
+
"train_samples_per_second": 146.143,
|
| 7 |
+
"train_steps_per_second": 4.567
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2807 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 20.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 371920,
|
| 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": 0.8909159302711487,
|
| 14 |
+
"learning_rate": 3.125e-05,
|
| 15 |
+
"loss": 6.228,
|
| 16 |
+
"step": 1000
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.11,
|
| 20 |
+
"grad_norm": 1.025515079498291,
|
| 21 |
+
"learning_rate": 6.25e-05,
|
| 22 |
+
"loss": 5.0018,
|
| 23 |
+
"step": 2000
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.16,
|
| 27 |
+
"grad_norm": 0.9241665005683899,
|
| 28 |
+
"learning_rate": 9.375e-05,
|
| 29 |
+
"loss": 4.6753,
|
| 30 |
+
"step": 3000
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.22,
|
| 34 |
+
"grad_norm": 0.8754993677139282,
|
| 35 |
+
"learning_rate": 0.000125,
|
| 36 |
+
"loss": 4.4594,
|
| 37 |
+
"step": 4000
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.27,
|
| 41 |
+
"grad_norm": 0.7558003664016724,
|
| 42 |
+
"learning_rate": 0.00015625,
|
| 43 |
+
"loss": 4.3007,
|
| 44 |
+
"step": 5000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.32,
|
| 48 |
+
"grad_norm": 0.7471924424171448,
|
| 49 |
+
"learning_rate": 0.0001875,
|
| 50 |
+
"loss": 4.1771,
|
| 51 |
+
"step": 6000
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.38,
|
| 55 |
+
"grad_norm": 0.6875430345535278,
|
| 56 |
+
"learning_rate": 0.00021875,
|
| 57 |
+
"loss": 4.0791,
|
| 58 |
+
"step": 7000
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.43,
|
| 62 |
+
"grad_norm": 0.6728542447090149,
|
| 63 |
+
"learning_rate": 0.00025,
|
| 64 |
+
"loss": 3.9784,
|
| 65 |
+
"step": 8000
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.48,
|
| 69 |
+
"grad_norm": 0.6585622429847717,
|
| 70 |
+
"learning_rate": 0.00028125000000000003,
|
| 71 |
+
"loss": 3.9146,
|
| 72 |
+
"step": 9000
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.54,
|
| 76 |
+
"grad_norm": 0.5960397124290466,
|
| 77 |
+
"learning_rate": 0.00031246875000000003,
|
| 78 |
+
"loss": 3.847,
|
| 79 |
+
"step": 10000
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.59,
|
| 83 |
+
"grad_norm": 0.576011061668396,
|
| 84 |
+
"learning_rate": 0.00034368749999999997,
|
| 85 |
+
"loss": 3.8063,
|
| 86 |
+
"step": 11000
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.65,
|
| 90 |
+
"grad_norm": 0.5716710686683655,
|
| 91 |
+
"learning_rate": 0.0003749375,
|
| 92 |
+
"loss": 3.7672,
|
| 93 |
+
"step": 12000
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.7,
|
| 97 |
+
"grad_norm": 0.5205326080322266,
|
| 98 |
+
"learning_rate": 0.00040615625,
|
| 99 |
+
"loss": 3.7304,
|
| 100 |
+
"step": 13000
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.75,
|
| 104 |
+
"grad_norm": 0.47286731004714966,
|
| 105 |
+
"learning_rate": 0.00043740625,
|
| 106 |
+
"loss": 3.6967,
|
| 107 |
+
"step": 14000
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.81,
|
| 111 |
+
"grad_norm": 0.4275617301464081,
|
| 112 |
+
"learning_rate": 0.00046865625,
|
| 113 |
+
"loss": 3.6717,
|
| 114 |
+
"step": 15000
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.86,
|
| 118 |
+
"grad_norm": 0.4085034430027008,
|
| 119 |
+
"learning_rate": 0.000499875,
|
| 120 |
+
"loss": 3.6497,
|
| 121 |
+
"step": 16000
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.91,
|
| 125 |
+
"grad_norm": 0.40013808012008667,
|
| 126 |
+
"learning_rate": 0.00053109375,
|
| 127 |
+
"loss": 3.6225,
|
| 128 |
+
"step": 17000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.97,
|
| 132 |
+
"grad_norm": 0.3701670169830322,
|
| 133 |
+
"learning_rate": 0.00056234375,
|
| 134 |
+
"loss": 3.61,
|
| 135 |
+
"step": 18000
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 1.0,
|
| 139 |
+
"eval_accuracy": 0.35835879185489444,
|
| 140 |
+
"eval_loss": 3.795727014541626,
|
| 141 |
+
"eval_runtime": 153.3517,
|
| 142 |
+
"eval_samples_per_second": 377.674,
|
| 143 |
+
"eval_steps_per_second": 5.901,
|
| 144 |
+
"step": 18596
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 1.02,
|
| 148 |
+
"grad_norm": 0.34899449348449707,
|
| 149 |
+
"learning_rate": 0.00059359375,
|
| 150 |
+
"loss": 3.579,
|
| 151 |
+
"step": 19000
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 1.08,
|
| 155 |
+
"grad_norm": 0.37780293822288513,
|
| 156 |
+
"learning_rate": 0.0006248437500000001,
|
| 157 |
+
"loss": 3.5464,
|
| 158 |
+
"step": 20000
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.13,
|
| 162 |
+
"grad_norm": 0.3226105272769928,
|
| 163 |
+
"learning_rate": 0.0006560625,
|
| 164 |
+
"loss": 3.5387,
|
| 165 |
+
"step": 21000
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 1.18,
|
| 169 |
+
"grad_norm": 0.3026217222213745,
|
| 170 |
+
"learning_rate": 0.0006873125,
|
| 171 |
+
"loss": 3.5284,
|
| 172 |
+
"step": 22000
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 1.24,
|
| 176 |
+
"grad_norm": 0.2845563590526581,
|
| 177 |
+
"learning_rate": 0.0007185625,
|
| 178 |
+
"loss": 3.5209,
|
| 179 |
+
"step": 23000
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.29,
|
| 183 |
+
"grad_norm": 0.28052306175231934,
|
| 184 |
+
"learning_rate": 0.0007498125,
|
| 185 |
+
"loss": 3.5094,
|
| 186 |
+
"step": 24000
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 1.34,
|
| 190 |
+
"grad_norm": 0.34453320503234863,
|
| 191 |
+
"learning_rate": 0.0007810312499999999,
|
| 192 |
+
"loss": 3.4934,
|
| 193 |
+
"step": 25000
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 1.4,
|
| 197 |
+
"grad_norm": 0.30819350481033325,
|
| 198 |
+
"learning_rate": 0.00081228125,
|
| 199 |
+
"loss": 3.4931,
|
| 200 |
+
"step": 26000
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 1.45,
|
| 204 |
+
"grad_norm": 0.30646008253097534,
|
| 205 |
+
"learning_rate": 0.0008435000000000001,
|
| 206 |
+
"loss": 3.477,
|
| 207 |
+
"step": 27000
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 1.51,
|
| 211 |
+
"grad_norm": 0.24530355632305145,
|
| 212 |
+
"learning_rate": 0.0008747500000000001,
|
| 213 |
+
"loss": 3.4684,
|
| 214 |
+
"step": 28000
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 1.56,
|
| 218 |
+
"grad_norm": 0.287979394197464,
|
| 219 |
+
"learning_rate": 0.0009059375,
|
| 220 |
+
"loss": 3.4633,
|
| 221 |
+
"step": 29000
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.61,
|
| 225 |
+
"grad_norm": 0.2318340390920639,
|
| 226 |
+
"learning_rate": 0.0009371875,
|
| 227 |
+
"loss": 3.4523,
|
| 228 |
+
"step": 30000
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 1.67,
|
| 232 |
+
"grad_norm": 0.21640178561210632,
|
| 233 |
+
"learning_rate": 0.0009684375,
|
| 234 |
+
"loss": 3.4511,
|
| 235 |
+
"step": 31000
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 1.72,
|
| 239 |
+
"grad_norm": 0.23932267725467682,
|
| 240 |
+
"learning_rate": 0.0009996562500000001,
|
| 241 |
+
"loss": 3.44,
|
| 242 |
+
"step": 32000
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.77,
|
| 246 |
+
"grad_norm": 0.23123665153980255,
|
| 247 |
+
"learning_rate": 0.0009970904918804424,
|
| 248 |
+
"loss": 3.4308,
|
| 249 |
+
"step": 33000
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 1.83,
|
| 253 |
+
"grad_norm": 0.22664424777030945,
|
| 254 |
+
"learning_rate": 0.0009941486232054601,
|
| 255 |
+
"loss": 3.4153,
|
| 256 |
+
"step": 34000
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 1.88,
|
| 260 |
+
"grad_norm": 0.21310342848300934,
|
| 261 |
+
"learning_rate": 0.0009912067545304777,
|
| 262 |
+
"loss": 3.4106,
|
| 263 |
+
"step": 35000
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 1.94,
|
| 267 |
+
"grad_norm": 0.2162679135799408,
|
| 268 |
+
"learning_rate": 0.0009882678277241704,
|
| 269 |
+
"loss": 3.4031,
|
| 270 |
+
"step": 36000
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 1.99,
|
| 274 |
+
"grad_norm": 0.21360158920288086,
|
| 275 |
+
"learning_rate": 0.0009853259590491882,
|
| 276 |
+
"loss": 3.3861,
|
| 277 |
+
"step": 37000
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 2.0,
|
| 281 |
+
"eval_accuracy": 0.38014852732029175,
|
| 282 |
+
"eval_loss": 3.564697742462158,
|
| 283 |
+
"eval_runtime": 154.5153,
|
| 284 |
+
"eval_samples_per_second": 374.83,
|
| 285 |
+
"eval_steps_per_second": 5.857,
|
| 286 |
+
"step": 37192
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 2.04,
|
| 290 |
+
"grad_norm": 0.2235065996646881,
|
| 291 |
+
"learning_rate": 0.0009823899741115557,
|
| 292 |
+
"loss": 3.3397,
|
| 293 |
+
"step": 38000
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 2.1,
|
| 297 |
+
"grad_norm": 0.2304013967514038,
|
| 298 |
+
"learning_rate": 0.0009794481054365733,
|
| 299 |
+
"loss": 3.3289,
|
| 300 |
+
"step": 39000
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 2.15,
|
| 304 |
+
"grad_norm": 0.22438882291316986,
|
| 305 |
+
"learning_rate": 0.000976506236761591,
|
| 306 |
+
"loss": 3.326,
|
| 307 |
+
"step": 40000
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 2.2,
|
| 311 |
+
"grad_norm": 0.21580879390239716,
|
| 312 |
+
"learning_rate": 0.0009735643680866086,
|
| 313 |
+
"loss": 3.3208,
|
| 314 |
+
"step": 41000
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 2.26,
|
| 318 |
+
"grad_norm": 0.21414010226726532,
|
| 319 |
+
"learning_rate": 0.0009706283831489763,
|
| 320 |
+
"loss": 3.3202,
|
| 321 |
+
"step": 42000
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 2.31,
|
| 325 |
+
"grad_norm": 0.20920276641845703,
|
| 326 |
+
"learning_rate": 0.0009676865144739939,
|
| 327 |
+
"loss": 3.3111,
|
| 328 |
+
"step": 43000
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 2.37,
|
| 332 |
+
"grad_norm": 0.19549404084682465,
|
| 333 |
+
"learning_rate": 0.0009647446457990115,
|
| 334 |
+
"loss": 3.308,
|
| 335 |
+
"step": 44000
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 2.42,
|
| 339 |
+
"grad_norm": 0.2047666609287262,
|
| 340 |
+
"learning_rate": 0.0009618027771240292,
|
| 341 |
+
"loss": 3.307,
|
| 342 |
+
"step": 45000
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 2.47,
|
| 346 |
+
"grad_norm": 0.19435334205627441,
|
| 347 |
+
"learning_rate": 0.0009588609084490469,
|
| 348 |
+
"loss": 3.3007,
|
| 349 |
+
"step": 46000
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 2.53,
|
| 353 |
+
"grad_norm": 0.20967498421669006,
|
| 354 |
+
"learning_rate": 0.0009559190397740644,
|
| 355 |
+
"loss": 3.293,
|
| 356 |
+
"step": 47000
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.58,
|
| 360 |
+
"grad_norm": 0.22619305551052094,
|
| 361 |
+
"learning_rate": 0.0009529801129677571,
|
| 362 |
+
"loss": 3.2911,
|
| 363 |
+
"step": 48000
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 2.63,
|
| 367 |
+
"grad_norm": 0.1997978389263153,
|
| 368 |
+
"learning_rate": 0.0009500382442927748,
|
| 369 |
+
"loss": 3.286,
|
| 370 |
+
"step": 49000
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 2.69,
|
| 374 |
+
"grad_norm": 0.22306489944458008,
|
| 375 |
+
"learning_rate": 0.0009470963756177925,
|
| 376 |
+
"loss": 3.2839,
|
| 377 |
+
"step": 50000
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 2.74,
|
| 381 |
+
"grad_norm": 0.20853149890899658,
|
| 382 |
+
"learning_rate": 0.000944157448811485,
|
| 383 |
+
"loss": 3.2745,
|
| 384 |
+
"step": 51000
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 2.8,
|
| 388 |
+
"grad_norm": 0.20844997465610504,
|
| 389 |
+
"learning_rate": 0.0009412155801365028,
|
| 390 |
+
"loss": 3.2657,
|
| 391 |
+
"step": 52000
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 2.85,
|
| 395 |
+
"grad_norm": 0.19329632818698883,
|
| 396 |
+
"learning_rate": 0.0009382766533301954,
|
| 397 |
+
"loss": 3.2683,
|
| 398 |
+
"step": 53000
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 2.9,
|
| 402 |
+
"grad_norm": 0.23765866458415985,
|
| 403 |
+
"learning_rate": 0.000935334784655213,
|
| 404 |
+
"loss": 3.2637,
|
| 405 |
+
"step": 54000
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 2.96,
|
| 409 |
+
"grad_norm": 0.2281341701745987,
|
| 410 |
+
"learning_rate": 0.0009323929159802307,
|
| 411 |
+
"loss": 3.2591,
|
| 412 |
+
"step": 55000
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 3.0,
|
| 416 |
+
"eval_accuracy": 0.39145331347307755,
|
| 417 |
+
"eval_loss": 3.4933156967163086,
|
| 418 |
+
"eval_runtime": 154.9004,
|
| 419 |
+
"eval_samples_per_second": 373.898,
|
| 420 |
+
"eval_steps_per_second": 5.842,
|
| 421 |
+
"step": 55788
|
| 422 |
+
},
|
| 423 |
+
{
|
| 424 |
+
"epoch": 3.01,
|
| 425 |
+
"grad_norm": 0.2204052358865738,
|
| 426 |
+
"learning_rate": 0.0009294539891739233,
|
| 427 |
+
"loss": 3.2431,
|
| 428 |
+
"step": 56000
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 3.07,
|
| 432 |
+
"grad_norm": 0.2415602058172226,
|
| 433 |
+
"learning_rate": 0.000926512120498941,
|
| 434 |
+
"loss": 3.1919,
|
| 435 |
+
"step": 57000
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 3.12,
|
| 439 |
+
"grad_norm": 0.19874005019664764,
|
| 440 |
+
"learning_rate": 0.0009235702518239586,
|
| 441 |
+
"loss": 3.196,
|
| 442 |
+
"step": 58000
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 3.17,
|
| 446 |
+
"grad_norm": 0.1932218074798584,
|
| 447 |
+
"learning_rate": 0.0009206283831489762,
|
| 448 |
+
"loss": 3.1988,
|
| 449 |
+
"step": 59000
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 3.23,
|
| 453 |
+
"grad_norm": 0.21343468129634857,
|
| 454 |
+
"learning_rate": 0.0009176894563426689,
|
| 455 |
+
"loss": 3.1961,
|
| 456 |
+
"step": 60000
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 3.28,
|
| 460 |
+
"grad_norm": 0.2427150458097458,
|
| 461 |
+
"learning_rate": 0.0009147475876676865,
|
| 462 |
+
"loss": 3.1967,
|
| 463 |
+
"step": 61000
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 3.33,
|
| 467 |
+
"grad_norm": 0.24537405371665955,
|
| 468 |
+
"learning_rate": 0.0009118086608613791,
|
| 469 |
+
"loss": 3.1992,
|
| 470 |
+
"step": 62000
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 3.39,
|
| 474 |
+
"grad_norm": 0.20690293610095978,
|
| 475 |
+
"learning_rate": 0.0009088667921863968,
|
| 476 |
+
"loss": 3.1979,
|
| 477 |
+
"step": 63000
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 3.44,
|
| 481 |
+
"grad_norm": 0.25561264157295227,
|
| 482 |
+
"learning_rate": 0.0009059249235114144,
|
| 483 |
+
"loss": 3.1956,
|
| 484 |
+
"step": 64000
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 3.5,
|
| 488 |
+
"grad_norm": 0.20544248819351196,
|
| 489 |
+
"learning_rate": 0.000902985996705107,
|
| 490 |
+
"loss": 3.1969,
|
| 491 |
+
"step": 65000
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 3.55,
|
| 495 |
+
"grad_norm": 0.19939689338207245,
|
| 496 |
+
"learning_rate": 0.0009000441280301248,
|
| 497 |
+
"loss": 3.1913,
|
| 498 |
+
"step": 66000
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 3.6,
|
| 502 |
+
"grad_norm": 0.2552586495876312,
|
| 503 |
+
"learning_rate": 0.0008971052012238174,
|
| 504 |
+
"loss": 3.1915,
|
| 505 |
+
"step": 67000
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 3.66,
|
| 509 |
+
"grad_norm": 0.23190921545028687,
|
| 510 |
+
"learning_rate": 0.000894163332548835,
|
| 511 |
+
"loss": 3.1966,
|
| 512 |
+
"step": 68000
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 3.71,
|
| 516 |
+
"grad_norm": 0.24762412905693054,
|
| 517 |
+
"learning_rate": 0.0008912244057425277,
|
| 518 |
+
"loss": 3.1867,
|
| 519 |
+
"step": 69000
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 3.76,
|
| 523 |
+
"grad_norm": 0.2145838439464569,
|
| 524 |
+
"learning_rate": 0.0008882825370675454,
|
| 525 |
+
"loss": 3.1905,
|
| 526 |
+
"step": 70000
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 3.82,
|
| 530 |
+
"grad_norm": 0.2245662808418274,
|
| 531 |
+
"learning_rate": 0.0008853436102612379,
|
| 532 |
+
"loss": 3.1846,
|
| 533 |
+
"step": 71000
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 3.87,
|
| 537 |
+
"grad_norm": 0.2066437155008316,
|
| 538 |
+
"learning_rate": 0.0008824017415862556,
|
| 539 |
+
"loss": 3.1832,
|
| 540 |
+
"step": 72000
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 3.93,
|
| 544 |
+
"grad_norm": 0.21306444704532623,
|
| 545 |
+
"learning_rate": 0.0008794628147799483,
|
| 546 |
+
"loss": 3.1834,
|
| 547 |
+
"step": 73000
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 3.98,
|
| 551 |
+
"grad_norm": 0.22630509734153748,
|
| 552 |
+
"learning_rate": 0.000876520946104966,
|
| 553 |
+
"loss": 3.1836,
|
| 554 |
+
"step": 74000
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 4.0,
|
| 558 |
+
"eval_accuracy": 0.3972446414495344,
|
| 559 |
+
"eval_loss": 3.4326329231262207,
|
| 560 |
+
"eval_runtime": 154.0547,
|
| 561 |
+
"eval_samples_per_second": 375.951,
|
| 562 |
+
"eval_steps_per_second": 5.875,
|
| 563 |
+
"step": 74384
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 4.03,
|
| 567 |
+
"grad_norm": 0.19157204031944275,
|
| 568 |
+
"learning_rate": 0.0008735790774299835,
|
| 569 |
+
"loss": 3.1363,
|
| 570 |
+
"step": 75000
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 4.09,
|
| 574 |
+
"grad_norm": 0.20826587080955505,
|
| 575 |
+
"learning_rate": 0.0008706401506236762,
|
| 576 |
+
"loss": 3.1164,
|
| 577 |
+
"step": 76000
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 4.14,
|
| 581 |
+
"grad_norm": 0.20335273444652557,
|
| 582 |
+
"learning_rate": 0.0008676982819486939,
|
| 583 |
+
"loss": 3.119,
|
| 584 |
+
"step": 77000
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 4.19,
|
| 588 |
+
"grad_norm": 0.20953166484832764,
|
| 589 |
+
"learning_rate": 0.0008647564132737114,
|
| 590 |
+
"loss": 3.1245,
|
| 591 |
+
"step": 78000
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 4.25,
|
| 595 |
+
"grad_norm": 0.20731399953365326,
|
| 596 |
+
"learning_rate": 0.0008618145445987291,
|
| 597 |
+
"loss": 3.1238,
|
| 598 |
+
"step": 79000
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 4.3,
|
| 602 |
+
"grad_norm": 0.23300333321094513,
|
| 603 |
+
"learning_rate": 0.0008588756177924218,
|
| 604 |
+
"loss": 3.1252,
|
| 605 |
+
"step": 80000
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 4.36,
|
| 609 |
+
"grad_norm": 0.20891304314136505,
|
| 610 |
+
"learning_rate": 0.0008559337491174393,
|
| 611 |
+
"loss": 3.1288,
|
| 612 |
+
"step": 81000
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 4.41,
|
| 616 |
+
"grad_norm": 0.21623115241527557,
|
| 617 |
+
"learning_rate": 0.000852994822311132,
|
| 618 |
+
"loss": 3.1285,
|
| 619 |
+
"step": 82000
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 4.46,
|
| 623 |
+
"grad_norm": 0.2474987506866455,
|
| 624 |
+
"learning_rate": 0.0008500529536361497,
|
| 625 |
+
"loss": 3.1262,
|
| 626 |
+
"step": 83000
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 4.52,
|
| 630 |
+
"grad_norm": 0.2118910700082779,
|
| 631 |
+
"learning_rate": 0.0008471140268298423,
|
| 632 |
+
"loss": 3.127,
|
| 633 |
+
"step": 84000
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 4.57,
|
| 637 |
+
"grad_norm": 0.24061526358127594,
|
| 638 |
+
"learning_rate": 0.00084417215815486,
|
| 639 |
+
"loss": 3.131,
|
| 640 |
+
"step": 85000
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 4.62,
|
| 644 |
+
"grad_norm": 0.22474774718284607,
|
| 645 |
+
"learning_rate": 0.0008412302894798777,
|
| 646 |
+
"loss": 3.1276,
|
| 647 |
+
"step": 86000
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 4.68,
|
| 651 |
+
"grad_norm": 0.19421574473381042,
|
| 652 |
+
"learning_rate": 0.0008382884208048953,
|
| 653 |
+
"loss": 3.127,
|
| 654 |
+
"step": 87000
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 4.73,
|
| 658 |
+
"grad_norm": 0.20476099848747253,
|
| 659 |
+
"learning_rate": 0.0008353494939985879,
|
| 660 |
+
"loss": 3.128,
|
| 661 |
+
"step": 88000
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 4.79,
|
| 665 |
+
"grad_norm": 0.23917576670646667,
|
| 666 |
+
"learning_rate": 0.0008324076253236056,
|
| 667 |
+
"loss": 3.1285,
|
| 668 |
+
"step": 89000
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 4.84,
|
| 672 |
+
"grad_norm": 0.213016539812088,
|
| 673 |
+
"learning_rate": 0.0008294657566486232,
|
| 674 |
+
"loss": 3.1281,
|
| 675 |
+
"step": 90000
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 4.89,
|
| 679 |
+
"grad_norm": 0.20960000157356262,
|
| 680 |
+
"learning_rate": 0.0008265238879736408,
|
| 681 |
+
"loss": 3.1314,
|
| 682 |
+
"step": 91000
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 4.95,
|
| 686 |
+
"grad_norm": 0.1963956207036972,
|
| 687 |
+
"learning_rate": 0.0008235849611673335,
|
| 688 |
+
"loss": 3.1278,
|
| 689 |
+
"step": 92000
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 5.0,
|
| 693 |
+
"eval_accuracy": 0.40181854228669783,
|
| 694 |
+
"eval_loss": 3.4132988452911377,
|
| 695 |
+
"eval_runtime": 154.1416,
|
| 696 |
+
"eval_samples_per_second": 375.739,
|
| 697 |
+
"eval_steps_per_second": 5.871,
|
| 698 |
+
"step": 92980
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 5.0,
|
| 702 |
+
"grad_norm": 0.26301759481430054,
|
| 703 |
+
"learning_rate": 0.0008206430924923511,
|
| 704 |
+
"loss": 3.121,
|
| 705 |
+
"step": 93000
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 5.05,
|
| 709 |
+
"grad_norm": 0.21968483924865723,
|
| 710 |
+
"learning_rate": 0.0008177012238173688,
|
| 711 |
+
"loss": 3.0579,
|
| 712 |
+
"step": 94000
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 5.11,
|
| 716 |
+
"grad_norm": 0.23824404180049896,
|
| 717 |
+
"learning_rate": 0.0008147622970110614,
|
| 718 |
+
"loss": 3.0621,
|
| 719 |
+
"step": 95000
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 5.16,
|
| 723 |
+
"grad_norm": 0.20740585029125214,
|
| 724 |
+
"learning_rate": 0.000811820428336079,
|
| 725 |
+
"loss": 3.0653,
|
| 726 |
+
"step": 96000
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 5.22,
|
| 730 |
+
"grad_norm": 0.21885009109973907,
|
| 731 |
+
"learning_rate": 0.0008088785596610967,
|
| 732 |
+
"loss": 3.0721,
|
| 733 |
+
"step": 97000
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 5.27,
|
| 737 |
+
"grad_norm": 0.27921879291534424,
|
| 738 |
+
"learning_rate": 0.0008059425747234644,
|
| 739 |
+
"loss": 3.0724,
|
| 740 |
+
"step": 98000
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 5.32,
|
| 744 |
+
"grad_norm": 0.2494860142469406,
|
| 745 |
+
"learning_rate": 0.0008030007060484819,
|
| 746 |
+
"loss": 3.0771,
|
| 747 |
+
"step": 99000
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 5.38,
|
| 751 |
+
"grad_norm": 0.2149578034877777,
|
| 752 |
+
"learning_rate": 0.0008000588373734997,
|
| 753 |
+
"loss": 3.0778,
|
| 754 |
+
"step": 100000
|
| 755 |
+
},
|
| 756 |
+
{
|
| 757 |
+
"epoch": 5.43,
|
| 758 |
+
"grad_norm": 0.24940645694732666,
|
| 759 |
+
"learning_rate": 0.0007971169686985174,
|
| 760 |
+
"loss": 3.079,
|
| 761 |
+
"step": 101000
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 5.49,
|
| 765 |
+
"grad_norm": 0.2192409187555313,
|
| 766 |
+
"learning_rate": 0.00079417804189221,
|
| 767 |
+
"loss": 3.0803,
|
| 768 |
+
"step": 102000
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 5.54,
|
| 772 |
+
"grad_norm": 0.21941109001636505,
|
| 773 |
+
"learning_rate": 0.0007912361732172276,
|
| 774 |
+
"loss": 3.0771,
|
| 775 |
+
"step": 103000
|
| 776 |
+
},
|
| 777 |
+
{
|
| 778 |
+
"epoch": 5.59,
|
| 779 |
+
"grad_norm": 0.21313978731632233,
|
| 780 |
+
"learning_rate": 0.0007882943045422453,
|
| 781 |
+
"loss": 3.0784,
|
| 782 |
+
"step": 104000
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 5.65,
|
| 786 |
+
"grad_norm": 0.27852264046669006,
|
| 787 |
+
"learning_rate": 0.0007853553777359379,
|
| 788 |
+
"loss": 3.0842,
|
| 789 |
+
"step": 105000
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 5.7,
|
| 793 |
+
"grad_norm": 0.27452683448791504,
|
| 794 |
+
"learning_rate": 0.0007824164509296305,
|
| 795 |
+
"loss": 3.0822,
|
| 796 |
+
"step": 106000
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"epoch": 5.75,
|
| 800 |
+
"grad_norm": 0.22761721909046173,
|
| 801 |
+
"learning_rate": 0.0007794745822546482,
|
| 802 |
+
"loss": 3.0824,
|
| 803 |
+
"step": 107000
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 5.81,
|
| 807 |
+
"grad_norm": 0.21608448028564453,
|
| 808 |
+
"learning_rate": 0.0007765356554483409,
|
| 809 |
+
"loss": 3.0859,
|
| 810 |
+
"step": 108000
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 5.86,
|
| 814 |
+
"grad_norm": 0.26219895482063293,
|
| 815 |
+
"learning_rate": 0.0007735937867733584,
|
| 816 |
+
"loss": 3.0817,
|
| 817 |
+
"step": 109000
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 5.92,
|
| 821 |
+
"grad_norm": 0.2277887463569641,
|
| 822 |
+
"learning_rate": 0.0007706519180983761,
|
| 823 |
+
"loss": 3.0799,
|
| 824 |
+
"step": 110000
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 5.97,
|
| 828 |
+
"grad_norm": 0.20768152177333832,
|
| 829 |
+
"learning_rate": 0.0007677100494233938,
|
| 830 |
+
"loss": 3.0795,
|
| 831 |
+
"step": 111000
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 6.0,
|
| 835 |
+
"eval_accuracy": 0.40326990139989005,
|
| 836 |
+
"eval_loss": 3.4124019145965576,
|
| 837 |
+
"eval_runtime": 154.0264,
|
| 838 |
+
"eval_samples_per_second": 376.02,
|
| 839 |
+
"eval_steps_per_second": 5.876,
|
| 840 |
+
"step": 111576
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 6.02,
|
| 844 |
+
"grad_norm": 0.20890597999095917,
|
| 845 |
+
"learning_rate": 0.0007647681807484113,
|
| 846 |
+
"loss": 3.049,
|
| 847 |
+
"step": 112000
|
| 848 |
+
},
|
| 849 |
+
{
|
| 850 |
+
"epoch": 6.08,
|
| 851 |
+
"grad_norm": 0.23515057563781738,
|
| 852 |
+
"learning_rate": 0.000761829253942104,
|
| 853 |
+
"loss": 3.0134,
|
| 854 |
+
"step": 113000
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 6.13,
|
| 858 |
+
"grad_norm": 0.23229765892028809,
|
| 859 |
+
"learning_rate": 0.0007588873852671217,
|
| 860 |
+
"loss": 3.0212,
|
| 861 |
+
"step": 114000
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 6.18,
|
| 865 |
+
"grad_norm": 0.26751187443733215,
|
| 866 |
+
"learning_rate": 0.0007559455165921393,
|
| 867 |
+
"loss": 3.0246,
|
| 868 |
+
"step": 115000
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 6.24,
|
| 872 |
+
"grad_norm": 0.25669464468955994,
|
| 873 |
+
"learning_rate": 0.0007530095316545069,
|
| 874 |
+
"loss": 3.0349,
|
| 875 |
+
"step": 116000
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 6.29,
|
| 879 |
+
"grad_norm": 0.2544041872024536,
|
| 880 |
+
"learning_rate": 0.0007500676629795246,
|
| 881 |
+
"loss": 3.0371,
|
| 882 |
+
"step": 117000
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 6.35,
|
| 886 |
+
"grad_norm": 0.23453743755817413,
|
| 887 |
+
"learning_rate": 0.0007471257943045423,
|
| 888 |
+
"loss": 3.0346,
|
| 889 |
+
"step": 118000
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 6.4,
|
| 893 |
+
"grad_norm": 0.2783219814300537,
|
| 894 |
+
"learning_rate": 0.0007441839256295599,
|
| 895 |
+
"loss": 3.0379,
|
| 896 |
+
"step": 119000
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 6.45,
|
| 900 |
+
"grad_norm": 0.24288877844810486,
|
| 901 |
+
"learning_rate": 0.0007412420569545776,
|
| 902 |
+
"loss": 3.0353,
|
| 903 |
+
"step": 120000
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 6.51,
|
| 907 |
+
"grad_norm": 0.21807478368282318,
|
| 908 |
+
"learning_rate": 0.0007383031301482702,
|
| 909 |
+
"loss": 3.0422,
|
| 910 |
+
"step": 121000
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 6.56,
|
| 914 |
+
"grad_norm": 0.2313011735677719,
|
| 915 |
+
"learning_rate": 0.0007353642033419628,
|
| 916 |
+
"loss": 3.0417,
|
| 917 |
+
"step": 122000
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 6.61,
|
| 921 |
+
"grad_norm": 0.23170366883277893,
|
| 922 |
+
"learning_rate": 0.0007324223346669805,
|
| 923 |
+
"loss": 3.0423,
|
| 924 |
+
"step": 123000
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 6.67,
|
| 928 |
+
"grad_norm": 0.21878008544445038,
|
| 929 |
+
"learning_rate": 0.0007294804659919982,
|
| 930 |
+
"loss": 3.0422,
|
| 931 |
+
"step": 124000
|
| 932 |
+
},
|
| 933 |
+
{
|
| 934 |
+
"epoch": 6.72,
|
| 935 |
+
"grad_norm": 0.2428036332130432,
|
| 936 |
+
"learning_rate": 0.0007265385973170157,
|
| 937 |
+
"loss": 3.0479,
|
| 938 |
+
"step": 125000
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 6.78,
|
| 942 |
+
"grad_norm": 0.22587792575359344,
|
| 943 |
+
"learning_rate": 0.0007235967286420334,
|
| 944 |
+
"loss": 3.0425,
|
| 945 |
+
"step": 126000
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 6.83,
|
| 949 |
+
"grad_norm": 0.22501161694526672,
|
| 950 |
+
"learning_rate": 0.000720657801835726,
|
| 951 |
+
"loss": 3.0446,
|
| 952 |
+
"step": 127000
|
| 953 |
+
},
|
| 954 |
+
{
|
| 955 |
+
"epoch": 6.88,
|
| 956 |
+
"grad_norm": 0.23849289119243622,
|
| 957 |
+
"learning_rate": 0.0007177159331607437,
|
| 958 |
+
"loss": 3.0415,
|
| 959 |
+
"step": 128000
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 6.94,
|
| 963 |
+
"grad_norm": 0.23556004464626312,
|
| 964 |
+
"learning_rate": 0.0007147770063544363,
|
| 965 |
+
"loss": 3.0478,
|
| 966 |
+
"step": 129000
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 6.99,
|
| 970 |
+
"grad_norm": 0.22832949459552765,
|
| 971 |
+
"learning_rate": 0.000711835137679454,
|
| 972 |
+
"loss": 3.0444,
|
| 973 |
+
"step": 130000
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 7.0,
|
| 977 |
+
"eval_accuracy": 0.4052325248213447,
|
| 978 |
+
"eval_loss": 3.3837218284606934,
|
| 979 |
+
"eval_runtime": 154.0952,
|
| 980 |
+
"eval_samples_per_second": 375.852,
|
| 981 |
+
"eval_steps_per_second": 5.873,
|
| 982 |
+
"step": 130172
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 7.04,
|
| 986 |
+
"grad_norm": 0.21719269454479218,
|
| 987 |
+
"learning_rate": 0.0007088932690044717,
|
| 988 |
+
"loss": 2.9885,
|
| 989 |
+
"step": 131000
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 7.1,
|
| 993 |
+
"grad_norm": 0.24770499765872955,
|
| 994 |
+
"learning_rate": 0.0007059543421981642,
|
| 995 |
+
"loss": 2.9813,
|
| 996 |
+
"step": 132000
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 7.15,
|
| 1000 |
+
"grad_norm": 0.23300251364707947,
|
| 1001 |
+
"learning_rate": 0.000703012473523182,
|
| 1002 |
+
"loss": 2.9884,
|
| 1003 |
+
"step": 133000
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 7.21,
|
| 1007 |
+
"grad_norm": 0.2238881140947342,
|
| 1008 |
+
"learning_rate": 0.0007000735467168746,
|
| 1009 |
+
"loss": 2.9932,
|
| 1010 |
+
"step": 134000
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 7.26,
|
| 1014 |
+
"grad_norm": 0.22716933488845825,
|
| 1015 |
+
"learning_rate": 0.0006971316780418923,
|
| 1016 |
+
"loss": 2.9953,
|
| 1017 |
+
"step": 135000
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 7.31,
|
| 1021 |
+
"grad_norm": 0.2647615075111389,
|
| 1022 |
+
"learning_rate": 0.0006941898093669099,
|
| 1023 |
+
"loss": 3.0028,
|
| 1024 |
+
"step": 136000
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 7.37,
|
| 1028 |
+
"grad_norm": 0.2218490093946457,
|
| 1029 |
+
"learning_rate": 0.0006912508825606025,
|
| 1030 |
+
"loss": 3.0034,
|
| 1031 |
+
"step": 137000
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 7.42,
|
| 1035 |
+
"grad_norm": 0.20896905660629272,
|
| 1036 |
+
"learning_rate": 0.0006883090138856202,
|
| 1037 |
+
"loss": 3.0015,
|
| 1038 |
+
"step": 138000
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 7.47,
|
| 1042 |
+
"grad_norm": 0.24288326501846313,
|
| 1043 |
+
"learning_rate": 0.0006853700870793128,
|
| 1044 |
+
"loss": 3.0057,
|
| 1045 |
+
"step": 139000
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 7.53,
|
| 1049 |
+
"grad_norm": 0.23336203396320343,
|
| 1050 |
+
"learning_rate": 0.0006824282184043304,
|
| 1051 |
+
"loss": 3.0104,
|
| 1052 |
+
"step": 140000
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 7.58,
|
| 1056 |
+
"grad_norm": 0.22705671191215515,
|
| 1057 |
+
"learning_rate": 0.0006794863497293481,
|
| 1058 |
+
"loss": 3.0079,
|
| 1059 |
+
"step": 141000
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 7.64,
|
| 1063 |
+
"grad_norm": 0.24445344507694244,
|
| 1064 |
+
"learning_rate": 0.0006765474229230408,
|
| 1065 |
+
"loss": 3.0095,
|
| 1066 |
+
"step": 142000
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 7.69,
|
| 1070 |
+
"grad_norm": 0.24330651760101318,
|
| 1071 |
+
"learning_rate": 0.0006736055542480583,
|
| 1072 |
+
"loss": 3.0088,
|
| 1073 |
+
"step": 143000
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 7.74,
|
| 1077 |
+
"grad_norm": 0.23043465614318848,
|
| 1078 |
+
"learning_rate": 0.000670666627441751,
|
| 1079 |
+
"loss": 3.0093,
|
| 1080 |
+
"step": 144000
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 7.8,
|
| 1084 |
+
"grad_norm": 0.21775782108306885,
|
| 1085 |
+
"learning_rate": 0.0006677247587667687,
|
| 1086 |
+
"loss": 3.0126,
|
| 1087 |
+
"step": 145000
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 7.85,
|
| 1091 |
+
"grad_norm": 0.2638954222202301,
|
| 1092 |
+
"learning_rate": 0.0006647858319604612,
|
| 1093 |
+
"loss": 3.014,
|
| 1094 |
+
"step": 146000
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 7.9,
|
| 1098 |
+
"grad_norm": 0.24298764765262604,
|
| 1099 |
+
"learning_rate": 0.0006618439632854789,
|
| 1100 |
+
"loss": 3.0144,
|
| 1101 |
+
"step": 147000
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 7.96,
|
| 1105 |
+
"grad_norm": 0.2365158647298813,
|
| 1106 |
+
"learning_rate": 0.0006589020946104966,
|
| 1107 |
+
"loss": 3.0115,
|
| 1108 |
+
"step": 148000
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 8.0,
|
| 1112 |
+
"eval_accuracy": 0.4068079041596424,
|
| 1113 |
+
"eval_loss": 3.366936683654785,
|
| 1114 |
+
"eval_runtime": 153.7426,
|
| 1115 |
+
"eval_samples_per_second": 376.714,
|
| 1116 |
+
"eval_steps_per_second": 5.886,
|
| 1117 |
+
"step": 148768
|
| 1118 |
+
},
|
| 1119 |
+
{
|
| 1120 |
+
"epoch": 8.01,
|
| 1121 |
+
"grad_norm": 0.2332482486963272,
|
| 1122 |
+
"learning_rate": 0.0006559631678041892,
|
| 1123 |
+
"loss": 3.0018,
|
| 1124 |
+
"step": 149000
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 8.07,
|
| 1128 |
+
"grad_norm": 0.2459622025489807,
|
| 1129 |
+
"learning_rate": 0.0006530242409978818,
|
| 1130 |
+
"loss": 2.9471,
|
| 1131 |
+
"step": 150000
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 8.12,
|
| 1135 |
+
"grad_norm": 0.2830921411514282,
|
| 1136 |
+
"learning_rate": 0.0006500823723228996,
|
| 1137 |
+
"loss": 2.9543,
|
| 1138 |
+
"step": 151000
|
| 1139 |
+
},
|
| 1140 |
+
{
|
| 1141 |
+
"epoch": 8.17,
|
| 1142 |
+
"grad_norm": 0.267229825258255,
|
| 1143 |
+
"learning_rate": 0.0006471434455165921,
|
| 1144 |
+
"loss": 2.9631,
|
| 1145 |
+
"step": 152000
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 8.23,
|
| 1149 |
+
"grad_norm": 0.259920597076416,
|
| 1150 |
+
"learning_rate": 0.0006442015768416098,
|
| 1151 |
+
"loss": 2.9617,
|
| 1152 |
+
"step": 153000
|
| 1153 |
+
},
|
| 1154 |
+
{
|
| 1155 |
+
"epoch": 8.28,
|
| 1156 |
+
"grad_norm": 0.2159169465303421,
|
| 1157 |
+
"learning_rate": 0.0006412597081666275,
|
| 1158 |
+
"loss": 2.9668,
|
| 1159 |
+
"step": 154000
|
| 1160 |
+
},
|
| 1161 |
+
{
|
| 1162 |
+
"epoch": 8.34,
|
| 1163 |
+
"grad_norm": 0.2374085932970047,
|
| 1164 |
+
"learning_rate": 0.0006383178394916452,
|
| 1165 |
+
"loss": 2.9728,
|
| 1166 |
+
"step": 155000
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 8.39,
|
| 1170 |
+
"grad_norm": 0.24743901193141937,
|
| 1171 |
+
"learning_rate": 0.0006353759708166627,
|
| 1172 |
+
"loss": 2.9693,
|
| 1173 |
+
"step": 156000
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 8.44,
|
| 1177 |
+
"grad_norm": 0.22295930981636047,
|
| 1178 |
+
"learning_rate": 0.0006324341021416804,
|
| 1179 |
+
"loss": 2.9758,
|
| 1180 |
+
"step": 157000
|
| 1181 |
+
},
|
| 1182 |
+
{
|
| 1183 |
+
"epoch": 8.5,
|
| 1184 |
+
"grad_norm": 0.2578998804092407,
|
| 1185 |
+
"learning_rate": 0.0006294922334666981,
|
| 1186 |
+
"loss": 2.9743,
|
| 1187 |
+
"step": 158000
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 8.55,
|
| 1191 |
+
"grad_norm": 0.23009610176086426,
|
| 1192 |
+
"learning_rate": 0.0006265533066603907,
|
| 1193 |
+
"loss": 2.9815,
|
| 1194 |
+
"step": 159000
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 8.6,
|
| 1198 |
+
"grad_norm": 0.2383948415517807,
|
| 1199 |
+
"learning_rate": 0.0006236143798540833,
|
| 1200 |
+
"loss": 2.9798,
|
| 1201 |
+
"step": 160000
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 8.66,
|
| 1205 |
+
"grad_norm": 0.2457112967967987,
|
| 1206 |
+
"learning_rate": 0.000620675453047776,
|
| 1207 |
+
"loss": 2.9815,
|
| 1208 |
+
"step": 161000
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 8.71,
|
| 1212 |
+
"grad_norm": 0.24992822110652924,
|
| 1213 |
+
"learning_rate": 0.0006177335843727936,
|
| 1214 |
+
"loss": 2.9797,
|
| 1215 |
+
"step": 162000
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 8.77,
|
| 1219 |
+
"grad_norm": 0.24688223004341125,
|
| 1220 |
+
"learning_rate": 0.0006147946575664862,
|
| 1221 |
+
"loss": 2.9869,
|
| 1222 |
+
"step": 163000
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"epoch": 8.82,
|
| 1226 |
+
"grad_norm": 0.23152810335159302,
|
| 1227 |
+
"learning_rate": 0.0006118527888915038,
|
| 1228 |
+
"loss": 2.9833,
|
| 1229 |
+
"step": 164000
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 8.87,
|
| 1233 |
+
"grad_norm": 0.2560078501701355,
|
| 1234 |
+
"learning_rate": 0.0006089109202165215,
|
| 1235 |
+
"loss": 2.9869,
|
| 1236 |
+
"step": 165000
|
| 1237 |
+
},
|
| 1238 |
+
{
|
| 1239 |
+
"epoch": 8.93,
|
| 1240 |
+
"grad_norm": 0.2425604909658432,
|
| 1241 |
+
"learning_rate": 0.0006059690515415393,
|
| 1242 |
+
"loss": 2.9856,
|
| 1243 |
+
"step": 166000
|
| 1244 |
+
},
|
| 1245 |
+
{
|
| 1246 |
+
"epoch": 8.98,
|
| 1247 |
+
"grad_norm": 0.22484102845191956,
|
| 1248 |
+
"learning_rate": 0.0006030271828665569,
|
| 1249 |
+
"loss": 2.9869,
|
| 1250 |
+
"step": 167000
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 9.0,
|
| 1254 |
+
"eval_accuracy": 0.40730411942650524,
|
| 1255 |
+
"eval_loss": 3.3864247798919678,
|
| 1256 |
+
"eval_runtime": 154.1912,
|
| 1257 |
+
"eval_samples_per_second": 375.618,
|
| 1258 |
+
"eval_steps_per_second": 5.869,
|
| 1259 |
+
"step": 167364
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 9.03,
|
| 1263 |
+
"grad_norm": 0.24037963151931763,
|
| 1264 |
+
"learning_rate": 0.0006000882560602495,
|
| 1265 |
+
"loss": 2.9434,
|
| 1266 |
+
"step": 168000
|
| 1267 |
+
},
|
| 1268 |
+
{
|
| 1269 |
+
"epoch": 9.09,
|
| 1270 |
+
"grad_norm": 0.23906952142715454,
|
| 1271 |
+
"learning_rate": 0.0005971463873852672,
|
| 1272 |
+
"loss": 2.9315,
|
| 1273 |
+
"step": 169000
|
| 1274 |
+
},
|
| 1275 |
+
{
|
| 1276 |
+
"epoch": 9.14,
|
| 1277 |
+
"grad_norm": 0.2632203698158264,
|
| 1278 |
+
"learning_rate": 0.0005942045187102848,
|
| 1279 |
+
"loss": 2.9304,
|
| 1280 |
+
"step": 170000
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 9.2,
|
| 1284 |
+
"grad_norm": 0.2545742392539978,
|
| 1285 |
+
"learning_rate": 0.0005912655919039774,
|
| 1286 |
+
"loss": 2.9354,
|
| 1287 |
+
"step": 171000
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 9.25,
|
| 1291 |
+
"grad_norm": 0.2435782551765442,
|
| 1292 |
+
"learning_rate": 0.0005883237232289951,
|
| 1293 |
+
"loss": 2.937,
|
| 1294 |
+
"step": 172000
|
| 1295 |
+
},
|
| 1296 |
+
{
|
| 1297 |
+
"epoch": 9.3,
|
| 1298 |
+
"grad_norm": 0.25050055980682373,
|
| 1299 |
+
"learning_rate": 0.0005853847964226877,
|
| 1300 |
+
"loss": 2.9435,
|
| 1301 |
+
"step": 173000
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 9.36,
|
| 1305 |
+
"grad_norm": 0.30542534589767456,
|
| 1306 |
+
"learning_rate": 0.0005824429277477053,
|
| 1307 |
+
"loss": 2.9454,
|
| 1308 |
+
"step": 174000
|
| 1309 |
+
},
|
| 1310 |
+
{
|
| 1311 |
+
"epoch": 9.41,
|
| 1312 |
+
"grad_norm": 0.23577359318733215,
|
| 1313 |
+
"learning_rate": 0.000579501059072723,
|
| 1314 |
+
"loss": 2.9473,
|
| 1315 |
+
"step": 175000
|
| 1316 |
+
},
|
| 1317 |
+
{
|
| 1318 |
+
"epoch": 9.46,
|
| 1319 |
+
"grad_norm": 0.2357962429523468,
|
| 1320 |
+
"learning_rate": 0.0005765591903977406,
|
| 1321 |
+
"loss": 2.9438,
|
| 1322 |
+
"step": 176000
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 9.52,
|
| 1326 |
+
"grad_norm": 0.27557680010795593,
|
| 1327 |
+
"learning_rate": 0.0005736202635914332,
|
| 1328 |
+
"loss": 2.9519,
|
| 1329 |
+
"step": 177000
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 9.57,
|
| 1333 |
+
"grad_norm": 0.2616015672683716,
|
| 1334 |
+
"learning_rate": 0.0005706813367851259,
|
| 1335 |
+
"loss": 2.9517,
|
| 1336 |
+
"step": 178000
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"epoch": 9.63,
|
| 1340 |
+
"grad_norm": 0.2544803321361542,
|
| 1341 |
+
"learning_rate": 0.0005677394681101436,
|
| 1342 |
+
"loss": 2.9528,
|
| 1343 |
+
"step": 179000
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 9.68,
|
| 1347 |
+
"grad_norm": 0.2523456811904907,
|
| 1348 |
+
"learning_rate": 0.0005647975994351611,
|
| 1349 |
+
"loss": 2.9551,
|
| 1350 |
+
"step": 180000
|
| 1351 |
+
},
|
| 1352 |
+
{
|
| 1353 |
+
"epoch": 9.73,
|
| 1354 |
+
"grad_norm": 0.24971884489059448,
|
| 1355 |
+
"learning_rate": 0.0005618557307601789,
|
| 1356 |
+
"loss": 2.9546,
|
| 1357 |
+
"step": 181000
|
| 1358 |
+
},
|
| 1359 |
+
{
|
| 1360 |
+
"epoch": 9.79,
|
| 1361 |
+
"grad_norm": 0.24730221927165985,
|
| 1362 |
+
"learning_rate": 0.0005589138620851966,
|
| 1363 |
+
"loss": 2.9569,
|
| 1364 |
+
"step": 182000
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 9.84,
|
| 1368 |
+
"grad_norm": 0.230006605386734,
|
| 1369 |
+
"learning_rate": 0.0005559749352788892,
|
| 1370 |
+
"loss": 2.9582,
|
| 1371 |
+
"step": 183000
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 9.89,
|
| 1375 |
+
"grad_norm": 0.22560349106788635,
|
| 1376 |
+
"learning_rate": 0.0005530389503412567,
|
| 1377 |
+
"loss": 2.9606,
|
| 1378 |
+
"step": 184000
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 9.95,
|
| 1382 |
+
"grad_norm": 0.28621554374694824,
|
| 1383 |
+
"learning_rate": 0.0005500970816662745,
|
| 1384 |
+
"loss": 2.9651,
|
| 1385 |
+
"step": 185000
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 10.0,
|
| 1389 |
+
"eval_accuracy": 0.4079115901065721,
|
| 1390 |
+
"eval_loss": 3.361621856689453,
|
| 1391 |
+
"eval_runtime": 154.014,
|
| 1392 |
+
"eval_samples_per_second": 376.05,
|
| 1393 |
+
"eval_steps_per_second": 5.876,
|
| 1394 |
+
"step": 185960
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 10.0,
|
| 1398 |
+
"grad_norm": 0.2537492513656616,
|
| 1399 |
+
"learning_rate": 0.0005471552129912922,
|
| 1400 |
+
"loss": 2.9625,
|
| 1401 |
+
"step": 186000
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 10.06,
|
| 1405 |
+
"grad_norm": 0.23095977306365967,
|
| 1406 |
+
"learning_rate": 0.0005442133443163097,
|
| 1407 |
+
"loss": 2.901,
|
| 1408 |
+
"step": 187000
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"epoch": 10.11,
|
| 1412 |
+
"grad_norm": 0.2696865200996399,
|
| 1413 |
+
"learning_rate": 0.0005412714756413274,
|
| 1414 |
+
"loss": 2.9027,
|
| 1415 |
+
"step": 188000
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 10.16,
|
| 1419 |
+
"grad_norm": 0.26662537455558777,
|
| 1420 |
+
"learning_rate": 0.0005383325488350201,
|
| 1421 |
+
"loss": 2.912,
|
| 1422 |
+
"step": 189000
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 10.22,
|
| 1426 |
+
"grad_norm": 0.24653853476047516,
|
| 1427 |
+
"learning_rate": 0.0005353906801600376,
|
| 1428 |
+
"loss": 2.9081,
|
| 1429 |
+
"step": 190000
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 10.27,
|
| 1433 |
+
"grad_norm": 0.2434544712305069,
|
| 1434 |
+
"learning_rate": 0.0005324488114850553,
|
| 1435 |
+
"loss": 2.9184,
|
| 1436 |
+
"step": 191000
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 10.32,
|
| 1440 |
+
"grad_norm": 0.25054100155830383,
|
| 1441 |
+
"learning_rate": 0.000529506942810073,
|
| 1442 |
+
"loss": 2.9183,
|
| 1443 |
+
"step": 192000
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 10.38,
|
| 1447 |
+
"grad_norm": 0.27828437089920044,
|
| 1448 |
+
"learning_rate": 0.0005265650741350905,
|
| 1449 |
+
"loss": 2.9214,
|
| 1450 |
+
"step": 193000
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 10.43,
|
| 1454 |
+
"grad_norm": 0.2823689579963684,
|
| 1455 |
+
"learning_rate": 0.0005236290891974582,
|
| 1456 |
+
"loss": 2.9229,
|
| 1457 |
+
"step": 194000
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 10.49,
|
| 1461 |
+
"grad_norm": 0.26634788513183594,
|
| 1462 |
+
"learning_rate": 0.0005206872205224759,
|
| 1463 |
+
"loss": 2.9256,
|
| 1464 |
+
"step": 195000
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 10.54,
|
| 1468 |
+
"grad_norm": 0.314277708530426,
|
| 1469 |
+
"learning_rate": 0.0005177453518474935,
|
| 1470 |
+
"loss": 2.9326,
|
| 1471 |
+
"step": 196000
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 10.59,
|
| 1475 |
+
"grad_norm": 0.24600285291671753,
|
| 1476 |
+
"learning_rate": 0.0005148034831725112,
|
| 1477 |
+
"loss": 2.9236,
|
| 1478 |
+
"step": 197000
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 10.65,
|
| 1482 |
+
"grad_norm": 0.2303924709558487,
|
| 1483 |
+
"learning_rate": 0.0005118645563662038,
|
| 1484 |
+
"loss": 2.9316,
|
| 1485 |
+
"step": 198000
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 10.7,
|
| 1489 |
+
"grad_norm": 0.23870837688446045,
|
| 1490 |
+
"learning_rate": 0.0005089226876912215,
|
| 1491 |
+
"loss": 2.9352,
|
| 1492 |
+
"step": 199000
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"epoch": 10.76,
|
| 1496 |
+
"grad_norm": 0.2509617805480957,
|
| 1497 |
+
"learning_rate": 0.0005059808190162391,
|
| 1498 |
+
"loss": 2.9302,
|
| 1499 |
+
"step": 200000
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 10.81,
|
| 1503 |
+
"grad_norm": 0.28766971826553345,
|
| 1504 |
+
"learning_rate": 0.0005030389503412568,
|
| 1505 |
+
"loss": 2.9309,
|
| 1506 |
+
"step": 201000
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 10.86,
|
| 1510 |
+
"grad_norm": 0.26924803853034973,
|
| 1511 |
+
"learning_rate": 0.0005001000235349494,
|
| 1512 |
+
"loss": 2.9381,
|
| 1513 |
+
"step": 202000
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 10.92,
|
| 1517 |
+
"grad_norm": 0.26498013734817505,
|
| 1518 |
+
"learning_rate": 0.0004971581548599671,
|
| 1519 |
+
"loss": 2.9364,
|
| 1520 |
+
"step": 203000
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 10.97,
|
| 1524 |
+
"grad_norm": 0.2591700255870819,
|
| 1525 |
+
"learning_rate": 0.0004942162861849847,
|
| 1526 |
+
"loss": 2.9407,
|
| 1527 |
+
"step": 204000
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 11.0,
|
| 1531 |
+
"eval_accuracy": 0.40898988079613996,
|
| 1532 |
+
"eval_loss": 3.3727080821990967,
|
| 1533 |
+
"eval_runtime": 154.0532,
|
| 1534 |
+
"eval_samples_per_second": 375.955,
|
| 1535 |
+
"eval_steps_per_second": 5.875,
|
| 1536 |
+
"step": 204556
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 11.02,
|
| 1540 |
+
"grad_norm": 0.2849581837654114,
|
| 1541 |
+
"learning_rate": 0.0004912744175100024,
|
| 1542 |
+
"loss": 2.9093,
|
| 1543 |
+
"step": 205000
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 11.08,
|
| 1547 |
+
"grad_norm": 0.26025569438934326,
|
| 1548 |
+
"learning_rate": 0.00048833843257237,
|
| 1549 |
+
"loss": 2.8745,
|
| 1550 |
+
"step": 206000
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 11.13,
|
| 1554 |
+
"grad_norm": 0.2565370202064514,
|
| 1555 |
+
"learning_rate": 0.00048539656389738763,
|
| 1556 |
+
"loss": 2.8846,
|
| 1557 |
+
"step": 207000
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 11.19,
|
| 1561 |
+
"grad_norm": 0.29611414670944214,
|
| 1562 |
+
"learning_rate": 0.00048245763709108024,
|
| 1563 |
+
"loss": 2.8862,
|
| 1564 |
+
"step": 208000
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 11.24,
|
| 1568 |
+
"grad_norm": 0.23962269723415375,
|
| 1569 |
+
"learning_rate": 0.0004795157684160979,
|
| 1570 |
+
"loss": 2.8923,
|
| 1571 |
+
"step": 209000
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 11.29,
|
| 1575 |
+
"grad_norm": 0.23374328017234802,
|
| 1576 |
+
"learning_rate": 0.0004765768416097905,
|
| 1577 |
+
"loss": 2.895,
|
| 1578 |
+
"step": 210000
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 11.35,
|
| 1582 |
+
"grad_norm": 0.2842772305011749,
|
| 1583 |
+
"learning_rate": 0.0004736349729348082,
|
| 1584 |
+
"loss": 2.8978,
|
| 1585 |
+
"step": 211000
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 11.4,
|
| 1589 |
+
"grad_norm": 0.2555365562438965,
|
| 1590 |
+
"learning_rate": 0.0004706960461285008,
|
| 1591 |
+
"loss": 2.8982,
|
| 1592 |
+
"step": 212000
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 11.45,
|
| 1596 |
+
"grad_norm": 0.26069942116737366,
|
| 1597 |
+
"learning_rate": 0.0004677541774535185,
|
| 1598 |
+
"loss": 2.9035,
|
| 1599 |
+
"step": 213000
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 11.51,
|
| 1603 |
+
"grad_norm": 0.2919773459434509,
|
| 1604 |
+
"learning_rate": 0.00046481230877853615,
|
| 1605 |
+
"loss": 2.9057,
|
| 1606 |
+
"step": 214000
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 11.56,
|
| 1610 |
+
"grad_norm": 0.26192766427993774,
|
| 1611 |
+
"learning_rate": 0.0004618704401035538,
|
| 1612 |
+
"loss": 2.9039,
|
| 1613 |
+
"step": 215000
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 11.62,
|
| 1617 |
+
"grad_norm": 0.2489086240530014,
|
| 1618 |
+
"learning_rate": 0.0004589285714285714,
|
| 1619 |
+
"loss": 2.9093,
|
| 1620 |
+
"step": 216000
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 11.67,
|
| 1624 |
+
"grad_norm": 0.26663750410079956,
|
| 1625 |
+
"learning_rate": 0.00045598964462226406,
|
| 1626 |
+
"loss": 2.9134,
|
| 1627 |
+
"step": 217000
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 11.72,
|
| 1631 |
+
"grad_norm": 0.26307132840156555,
|
| 1632 |
+
"learning_rate": 0.00045304777594728173,
|
| 1633 |
+
"loss": 2.9088,
|
| 1634 |
+
"step": 218000
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 11.78,
|
| 1638 |
+
"grad_norm": 0.2765272259712219,
|
| 1639 |
+
"learning_rate": 0.00045010884914097434,
|
| 1640 |
+
"loss": 2.9124,
|
| 1641 |
+
"step": 219000
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 11.83,
|
| 1645 |
+
"grad_norm": 0.2534734010696411,
|
| 1646 |
+
"learning_rate": 0.000447166980465992,
|
| 1647 |
+
"loss": 2.9139,
|
| 1648 |
+
"step": 220000
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 11.88,
|
| 1652 |
+
"grad_norm": 0.2891595661640167,
|
| 1653 |
+
"learning_rate": 0.0004442251117910097,
|
| 1654 |
+
"loss": 2.9135,
|
| 1655 |
+
"step": 221000
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 11.94,
|
| 1659 |
+
"grad_norm": 0.27639132738113403,
|
| 1660 |
+
"learning_rate": 0.0004412861849847023,
|
| 1661 |
+
"loss": 2.9144,
|
| 1662 |
+
"step": 222000
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 11.99,
|
| 1666 |
+
"grad_norm": 0.2949223518371582,
|
| 1667 |
+
"learning_rate": 0.0004383443163097199,
|
| 1668 |
+
"loss": 2.9186,
|
| 1669 |
+
"step": 223000
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 12.0,
|
| 1673 |
+
"eval_accuracy": 0.4090768830667313,
|
| 1674 |
+
"eval_loss": 3.386943817138672,
|
| 1675 |
+
"eval_runtime": 154.0531,
|
| 1676 |
+
"eval_samples_per_second": 375.955,
|
| 1677 |
+
"eval_steps_per_second": 5.875,
|
| 1678 |
+
"step": 223152
|
| 1679 |
+
},
|
| 1680 |
+
{
|
| 1681 |
+
"epoch": 12.05,
|
| 1682 |
+
"grad_norm": 0.2919670641422272,
|
| 1683 |
+
"learning_rate": 0.0004354053895034126,
|
| 1684 |
+
"loss": 2.8624,
|
| 1685 |
+
"step": 224000
|
| 1686 |
+
},
|
| 1687 |
+
{
|
| 1688 |
+
"epoch": 12.1,
|
| 1689 |
+
"grad_norm": 0.2778524160385132,
|
| 1690 |
+
"learning_rate": 0.0004324664626971052,
|
| 1691 |
+
"loss": 2.8607,
|
| 1692 |
+
"step": 225000
|
| 1693 |
+
},
|
| 1694 |
+
{
|
| 1695 |
+
"epoch": 12.15,
|
| 1696 |
+
"grad_norm": 0.2688823640346527,
|
| 1697 |
+
"learning_rate": 0.00042952459402212286,
|
| 1698 |
+
"loss": 2.8637,
|
| 1699 |
+
"step": 226000
|
| 1700 |
+
},
|
| 1701 |
+
{
|
| 1702 |
+
"epoch": 12.21,
|
| 1703 |
+
"grad_norm": 0.26661086082458496,
|
| 1704 |
+
"learning_rate": 0.0004265827253471405,
|
| 1705 |
+
"loss": 2.8716,
|
| 1706 |
+
"step": 227000
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 12.26,
|
| 1710 |
+
"grad_norm": 0.2500409483909607,
|
| 1711 |
+
"learning_rate": 0.00042364085667215815,
|
| 1712 |
+
"loss": 2.8719,
|
| 1713 |
+
"step": 228000
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 12.31,
|
| 1717 |
+
"grad_norm": 0.24592752754688263,
|
| 1718 |
+
"learning_rate": 0.0004207019298658508,
|
| 1719 |
+
"loss": 2.8779,
|
| 1720 |
+
"step": 229000
|
| 1721 |
+
},
|
| 1722 |
+
{
|
| 1723 |
+
"epoch": 12.37,
|
| 1724 |
+
"grad_norm": 0.2966313064098358,
|
| 1725 |
+
"learning_rate": 0.00041776006119086844,
|
| 1726 |
+
"loss": 2.8773,
|
| 1727 |
+
"step": 230000
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 12.42,
|
| 1731 |
+
"grad_norm": 0.2876270115375519,
|
| 1732 |
+
"learning_rate": 0.0004148181925158861,
|
| 1733 |
+
"loss": 2.8777,
|
| 1734 |
+
"step": 231000
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 12.48,
|
| 1738 |
+
"grad_norm": 0.25883278250694275,
|
| 1739 |
+
"learning_rate": 0.0004118792657095787,
|
| 1740 |
+
"loss": 2.8838,
|
| 1741 |
+
"step": 232000
|
| 1742 |
+
},
|
| 1743 |
+
{
|
| 1744 |
+
"epoch": 12.53,
|
| 1745 |
+
"grad_norm": 0.25680166482925415,
|
| 1746 |
+
"learning_rate": 0.0004089373970345964,
|
| 1747 |
+
"loss": 2.8847,
|
| 1748 |
+
"step": 233000
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 12.58,
|
| 1752 |
+
"grad_norm": 0.28966328501701355,
|
| 1753 |
+
"learning_rate": 0.000405998470228289,
|
| 1754 |
+
"loss": 2.8847,
|
| 1755 |
+
"step": 234000
|
| 1756 |
+
},
|
| 1757 |
+
{
|
| 1758 |
+
"epoch": 12.64,
|
| 1759 |
+
"grad_norm": 0.2587486803531647,
|
| 1760 |
+
"learning_rate": 0.0004030566015533066,
|
| 1761 |
+
"loss": 2.8868,
|
| 1762 |
+
"step": 235000
|
| 1763 |
+
},
|
| 1764 |
+
{
|
| 1765 |
+
"epoch": 12.69,
|
| 1766 |
+
"grad_norm": 0.2806561291217804,
|
| 1767 |
+
"learning_rate": 0.00040011473287832435,
|
| 1768 |
+
"loss": 2.8887,
|
| 1769 |
+
"step": 236000
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 12.74,
|
| 1773 |
+
"grad_norm": 0.25261130928993225,
|
| 1774 |
+
"learning_rate": 0.00039717580607201696,
|
| 1775 |
+
"loss": 2.8888,
|
| 1776 |
+
"step": 237000
|
| 1777 |
+
},
|
| 1778 |
+
{
|
| 1779 |
+
"epoch": 12.8,
|
| 1780 |
+
"grad_norm": 0.2473638653755188,
|
| 1781 |
+
"learning_rate": 0.00039423393739703463,
|
| 1782 |
+
"loss": 2.8931,
|
| 1783 |
+
"step": 238000
|
| 1784 |
+
},
|
| 1785 |
+
{
|
| 1786 |
+
"epoch": 12.85,
|
| 1787 |
+
"grad_norm": 0.2695077061653137,
|
| 1788 |
+
"learning_rate": 0.00039129206872205225,
|
| 1789 |
+
"loss": 2.8937,
|
| 1790 |
+
"step": 239000
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 12.91,
|
| 1794 |
+
"grad_norm": 0.26754283905029297,
|
| 1795 |
+
"learning_rate": 0.00038835020004706987,
|
| 1796 |
+
"loss": 2.8929,
|
| 1797 |
+
"step": 240000
|
| 1798 |
+
},
|
| 1799 |
+
{
|
| 1800 |
+
"epoch": 12.96,
|
| 1801 |
+
"grad_norm": 0.2615933120250702,
|
| 1802 |
+
"learning_rate": 0.00038541127324076254,
|
| 1803 |
+
"loss": 2.8974,
|
| 1804 |
+
"step": 241000
|
| 1805 |
+
},
|
| 1806 |
+
{
|
| 1807 |
+
"epoch": 13.0,
|
| 1808 |
+
"eval_accuracy": 0.4096338991481772,
|
| 1809 |
+
"eval_loss": 3.3688578605651855,
|
| 1810 |
+
"eval_runtime": 154.4918,
|
| 1811 |
+
"eval_samples_per_second": 374.887,
|
| 1812 |
+
"eval_steps_per_second": 5.858,
|
| 1813 |
+
"step": 241748
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 13.01,
|
| 1817 |
+
"grad_norm": 0.2606102526187897,
|
| 1818 |
+
"learning_rate": 0.00038246940456578016,
|
| 1819 |
+
"loss": 2.8775,
|
| 1820 |
+
"step": 242000
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 13.07,
|
| 1824 |
+
"grad_norm": 0.27312567830085754,
|
| 1825 |
+
"learning_rate": 0.0003795275358907979,
|
| 1826 |
+
"loss": 2.837,
|
| 1827 |
+
"step": 243000
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 13.12,
|
| 1831 |
+
"grad_norm": 0.2866615056991577,
|
| 1832 |
+
"learning_rate": 0.0003765886090844905,
|
| 1833 |
+
"loss": 2.8441,
|
| 1834 |
+
"step": 244000
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 13.17,
|
| 1838 |
+
"grad_norm": 0.26024916768074036,
|
| 1839 |
+
"learning_rate": 0.0003736467404095081,
|
| 1840 |
+
"loss": 2.8453,
|
| 1841 |
+
"step": 245000
|
| 1842 |
+
},
|
| 1843 |
+
{
|
| 1844 |
+
"epoch": 13.23,
|
| 1845 |
+
"grad_norm": 0.3037625551223755,
|
| 1846 |
+
"learning_rate": 0.0003707107554718757,
|
| 1847 |
+
"loss": 2.849,
|
| 1848 |
+
"step": 246000
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 13.28,
|
| 1852 |
+
"grad_norm": 0.25507059693336487,
|
| 1853 |
+
"learning_rate": 0.0003677688867968934,
|
| 1854 |
+
"loss": 2.8511,
|
| 1855 |
+
"step": 247000
|
| 1856 |
+
},
|
| 1857 |
+
{
|
| 1858 |
+
"epoch": 13.34,
|
| 1859 |
+
"grad_norm": 0.26291272044181824,
|
| 1860 |
+
"learning_rate": 0.00036482701812191106,
|
| 1861 |
+
"loss": 2.8608,
|
| 1862 |
+
"step": 248000
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 13.39,
|
| 1866 |
+
"grad_norm": 0.2695772349834442,
|
| 1867 |
+
"learning_rate": 0.0003618851494469287,
|
| 1868 |
+
"loss": 2.8596,
|
| 1869 |
+
"step": 249000
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 13.44,
|
| 1873 |
+
"grad_norm": 0.2734662592411041,
|
| 1874 |
+
"learning_rate": 0.00035894622264062134,
|
| 1875 |
+
"loss": 2.8626,
|
| 1876 |
+
"step": 250000
|
| 1877 |
+
},
|
| 1878 |
+
{
|
| 1879 |
+
"epoch": 13.5,
|
| 1880 |
+
"grad_norm": 0.34274160861968994,
|
| 1881 |
+
"learning_rate": 0.00035600435396563896,
|
| 1882 |
+
"loss": 2.8624,
|
| 1883 |
+
"step": 251000
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 13.55,
|
| 1887 |
+
"grad_norm": 0.2600579857826233,
|
| 1888 |
+
"learning_rate": 0.00035306248529065663,
|
| 1889 |
+
"loss": 2.8641,
|
| 1890 |
+
"step": 252000
|
| 1891 |
+
},
|
| 1892 |
+
{
|
| 1893 |
+
"epoch": 13.61,
|
| 1894 |
+
"grad_norm": 0.27528443932533264,
|
| 1895 |
+
"learning_rate": 0.0003501235584843493,
|
| 1896 |
+
"loss": 2.8663,
|
| 1897 |
+
"step": 253000
|
| 1898 |
+
},
|
| 1899 |
+
{
|
| 1900 |
+
"epoch": 13.66,
|
| 1901 |
+
"grad_norm": 0.26855769753456116,
|
| 1902 |
+
"learning_rate": 0.0003471816898093669,
|
| 1903 |
+
"loss": 2.872,
|
| 1904 |
+
"step": 254000
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 13.71,
|
| 1908 |
+
"grad_norm": 0.26368248462677,
|
| 1909 |
+
"learning_rate": 0.0003442427630030596,
|
| 1910 |
+
"loss": 2.8674,
|
| 1911 |
+
"step": 255000
|
| 1912 |
+
},
|
| 1913 |
+
{
|
| 1914 |
+
"epoch": 13.77,
|
| 1915 |
+
"grad_norm": 0.27387478947639465,
|
| 1916 |
+
"learning_rate": 0.0003413008943280772,
|
| 1917 |
+
"loss": 2.8665,
|
| 1918 |
+
"step": 256000
|
| 1919 |
+
},
|
| 1920 |
+
{
|
| 1921 |
+
"epoch": 13.82,
|
| 1922 |
+
"grad_norm": 0.30804863572120667,
|
| 1923 |
+
"learning_rate": 0.0003383590256530948,
|
| 1924 |
+
"loss": 2.8721,
|
| 1925 |
+
"step": 257000
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"epoch": 13.87,
|
| 1929 |
+
"grad_norm": 0.3087449371814728,
|
| 1930 |
+
"learning_rate": 0.0003354200988467875,
|
| 1931 |
+
"loss": 2.8765,
|
| 1932 |
+
"step": 258000
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"epoch": 13.93,
|
| 1936 |
+
"grad_norm": 0.26736271381378174,
|
| 1937 |
+
"learning_rate": 0.0003324782301718051,
|
| 1938 |
+
"loss": 2.8768,
|
| 1939 |
+
"step": 259000
|
| 1940 |
+
},
|
| 1941 |
+
{
|
| 1942 |
+
"epoch": 13.98,
|
| 1943 |
+
"grad_norm": 0.3191315233707428,
|
| 1944 |
+
"learning_rate": 0.00032953636149682283,
|
| 1945 |
+
"loss": 2.8733,
|
| 1946 |
+
"step": 260000
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 14.0,
|
| 1950 |
+
"eval_accuracy": 0.40978962985337464,
|
| 1951 |
+
"eval_loss": 3.381140947341919,
|
| 1952 |
+
"eval_runtime": 153.947,
|
| 1953 |
+
"eval_samples_per_second": 376.214,
|
| 1954 |
+
"eval_steps_per_second": 5.879,
|
| 1955 |
+
"step": 260344
|
| 1956 |
+
},
|
| 1957 |
+
{
|
| 1958 |
+
"epoch": 14.04,
|
| 1959 |
+
"grad_norm": 0.2935682237148285,
|
| 1960 |
+
"learning_rate": 0.00032659743469051544,
|
| 1961 |
+
"loss": 2.8349,
|
| 1962 |
+
"step": 261000
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 14.09,
|
| 1966 |
+
"grad_norm": 0.32009586691856384,
|
| 1967 |
+
"learning_rate": 0.0003236555660155331,
|
| 1968 |
+
"loss": 2.8245,
|
| 1969 |
+
"step": 262000
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 14.14,
|
| 1973 |
+
"grad_norm": 0.2821045219898224,
|
| 1974 |
+
"learning_rate": 0.00032071369734055073,
|
| 1975 |
+
"loss": 2.8243,
|
| 1976 |
+
"step": 263000
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 14.2,
|
| 1980 |
+
"grad_norm": 0.2877224087715149,
|
| 1981 |
+
"learning_rate": 0.00031777477053424334,
|
| 1982 |
+
"loss": 2.8294,
|
| 1983 |
+
"step": 264000
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 14.25,
|
| 1987 |
+
"grad_norm": 0.28222963213920593,
|
| 1988 |
+
"learning_rate": 0.000314835843727936,
|
| 1989 |
+
"loss": 2.8345,
|
| 1990 |
+
"step": 265000
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 14.3,
|
| 1994 |
+
"grad_norm": 0.2666056156158447,
|
| 1995 |
+
"learning_rate": 0.0003118939750529536,
|
| 1996 |
+
"loss": 2.8345,
|
| 1997 |
+
"step": 266000
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 14.36,
|
| 2001 |
+
"grad_norm": 0.26124948263168335,
|
| 2002 |
+
"learning_rate": 0.0003089521063779713,
|
| 2003 |
+
"loss": 2.8419,
|
| 2004 |
+
"step": 267000
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 14.41,
|
| 2008 |
+
"grad_norm": 0.3096939027309418,
|
| 2009 |
+
"learning_rate": 0.00030601023770298897,
|
| 2010 |
+
"loss": 2.8415,
|
| 2011 |
+
"step": 268000
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 14.47,
|
| 2015 |
+
"grad_norm": 0.27581918239593506,
|
| 2016 |
+
"learning_rate": 0.0003030683690280066,
|
| 2017 |
+
"loss": 2.8435,
|
| 2018 |
+
"step": 269000
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 14.52,
|
| 2022 |
+
"grad_norm": 0.29562026262283325,
|
| 2023 |
+
"learning_rate": 0.00030012944222169925,
|
| 2024 |
+
"loss": 2.8451,
|
| 2025 |
+
"step": 270000
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 14.57,
|
| 2029 |
+
"grad_norm": 0.30838248133659363,
|
| 2030 |
+
"learning_rate": 0.0002971875735467169,
|
| 2031 |
+
"loss": 2.8491,
|
| 2032 |
+
"step": 271000
|
| 2033 |
+
},
|
| 2034 |
+
{
|
| 2035 |
+
"epoch": 14.63,
|
| 2036 |
+
"grad_norm": 0.2891804873943329,
|
| 2037 |
+
"learning_rate": 0.00029424864674040954,
|
| 2038 |
+
"loss": 2.8485,
|
| 2039 |
+
"step": 272000
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 14.68,
|
| 2043 |
+
"grad_norm": 0.2642837166786194,
|
| 2044 |
+
"learning_rate": 0.00029130677806542716,
|
| 2045 |
+
"loss": 2.8478,
|
| 2046 |
+
"step": 273000
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 14.73,
|
| 2050 |
+
"grad_norm": 0.3049638271331787,
|
| 2051 |
+
"learning_rate": 0.0002883649093904448,
|
| 2052 |
+
"loss": 2.8519,
|
| 2053 |
+
"step": 274000
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 14.79,
|
| 2057 |
+
"grad_norm": 0.26046690344810486,
|
| 2058 |
+
"learning_rate": 0.0002854230407154625,
|
| 2059 |
+
"loss": 2.8514,
|
| 2060 |
+
"step": 275000
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 14.84,
|
| 2064 |
+
"grad_norm": 0.2902365028858185,
|
| 2065 |
+
"learning_rate": 0.0002824841139091551,
|
| 2066 |
+
"loss": 2.8505,
|
| 2067 |
+
"step": 276000
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 14.9,
|
| 2071 |
+
"grad_norm": 0.2648141384124756,
|
| 2072 |
+
"learning_rate": 0.0002795422452341728,
|
| 2073 |
+
"loss": 2.8525,
|
| 2074 |
+
"step": 277000
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 14.95,
|
| 2078 |
+
"grad_norm": 0.27711859345436096,
|
| 2079 |
+
"learning_rate": 0.0002766003765591904,
|
| 2080 |
+
"loss": 2.8547,
|
| 2081 |
+
"step": 278000
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 15.0,
|
| 2085 |
+
"eval_accuracy": 0.41021328724206096,
|
| 2086 |
+
"eval_loss": 3.3861238956451416,
|
| 2087 |
+
"eval_runtime": 154.0149,
|
| 2088 |
+
"eval_samples_per_second": 376.048,
|
| 2089 |
+
"eval_steps_per_second": 5.876,
|
| 2090 |
+
"step": 278940
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 15.0,
|
| 2094 |
+
"grad_norm": 0.27470868825912476,
|
| 2095 |
+
"learning_rate": 0.000273658507884208,
|
| 2096 |
+
"loss": 2.8524,
|
| 2097 |
+
"step": 279000
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 15.06,
|
| 2101 |
+
"grad_norm": 0.29804325103759766,
|
| 2102 |
+
"learning_rate": 0.0002707225229465757,
|
| 2103 |
+
"loss": 2.8018,
|
| 2104 |
+
"step": 280000
|
| 2105 |
+
},
|
| 2106 |
+
{
|
| 2107 |
+
"epoch": 15.11,
|
| 2108 |
+
"grad_norm": 0.3048715591430664,
|
| 2109 |
+
"learning_rate": 0.0002677806542715933,
|
| 2110 |
+
"loss": 2.8103,
|
| 2111 |
+
"step": 281000
|
| 2112 |
+
},
|
| 2113 |
+
{
|
| 2114 |
+
"epoch": 15.16,
|
| 2115 |
+
"grad_norm": 0.2895219624042511,
|
| 2116 |
+
"learning_rate": 0.00026483878559661097,
|
| 2117 |
+
"loss": 2.812,
|
| 2118 |
+
"step": 282000
|
| 2119 |
+
},
|
| 2120 |
+
{
|
| 2121 |
+
"epoch": 15.22,
|
| 2122 |
+
"grad_norm": 0.31599316000938416,
|
| 2123 |
+
"learning_rate": 0.00026189691692162865,
|
| 2124 |
+
"loss": 2.82,
|
| 2125 |
+
"step": 283000
|
| 2126 |
+
},
|
| 2127 |
+
{
|
| 2128 |
+
"epoch": 15.27,
|
| 2129 |
+
"grad_norm": 0.2961556315422058,
|
| 2130 |
+
"learning_rate": 0.00025895504824664626,
|
| 2131 |
+
"loss": 2.8214,
|
| 2132 |
+
"step": 284000
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 15.33,
|
| 2136 |
+
"grad_norm": 0.3021540343761444,
|
| 2137 |
+
"learning_rate": 0.00025601612144033893,
|
| 2138 |
+
"loss": 2.8174,
|
| 2139 |
+
"step": 285000
|
| 2140 |
+
},
|
| 2141 |
+
{
|
| 2142 |
+
"epoch": 15.38,
|
| 2143 |
+
"grad_norm": 0.2883637845516205,
|
| 2144 |
+
"learning_rate": 0.00025307425276535655,
|
| 2145 |
+
"loss": 2.8197,
|
| 2146 |
+
"step": 286000
|
| 2147 |
+
},
|
| 2148 |
+
{
|
| 2149 |
+
"epoch": 15.43,
|
| 2150 |
+
"grad_norm": 0.2952839136123657,
|
| 2151 |
+
"learning_rate": 0.0002501323840903742,
|
| 2152 |
+
"loss": 2.8187,
|
| 2153 |
+
"step": 287000
|
| 2154 |
+
},
|
| 2155 |
+
{
|
| 2156 |
+
"epoch": 15.49,
|
| 2157 |
+
"grad_norm": 0.2684228718280792,
|
| 2158 |
+
"learning_rate": 0.00024719051541539184,
|
| 2159 |
+
"loss": 2.826,
|
| 2160 |
+
"step": 288000
|
| 2161 |
+
},
|
| 2162 |
+
{
|
| 2163 |
+
"epoch": 15.54,
|
| 2164 |
+
"grad_norm": 0.31188449263572693,
|
| 2165 |
+
"learning_rate": 0.0002442515886090845,
|
| 2166 |
+
"loss": 2.8287,
|
| 2167 |
+
"step": 289000
|
| 2168 |
+
},
|
| 2169 |
+
{
|
| 2170 |
+
"epoch": 15.59,
|
| 2171 |
+
"grad_norm": 0.2904526889324188,
|
| 2172 |
+
"learning_rate": 0.00024130971993410215,
|
| 2173 |
+
"loss": 2.8232,
|
| 2174 |
+
"step": 290000
|
| 2175 |
+
},
|
| 2176 |
+
{
|
| 2177 |
+
"epoch": 15.65,
|
| 2178 |
+
"grad_norm": 0.31149864196777344,
|
| 2179 |
+
"learning_rate": 0.0002383678512591198,
|
| 2180 |
+
"loss": 2.8305,
|
| 2181 |
+
"step": 291000
|
| 2182 |
+
},
|
| 2183 |
+
{
|
| 2184 |
+
"epoch": 15.7,
|
| 2185 |
+
"grad_norm": 0.30907660722732544,
|
| 2186 |
+
"learning_rate": 0.00023542892445281243,
|
| 2187 |
+
"loss": 2.8319,
|
| 2188 |
+
"step": 292000
|
| 2189 |
+
},
|
| 2190 |
+
{
|
| 2191 |
+
"epoch": 15.76,
|
| 2192 |
+
"grad_norm": 0.2970659136772156,
|
| 2193 |
+
"learning_rate": 0.00023248705577783008,
|
| 2194 |
+
"loss": 2.8319,
|
| 2195 |
+
"step": 293000
|
| 2196 |
+
},
|
| 2197 |
+
{
|
| 2198 |
+
"epoch": 15.81,
|
| 2199 |
+
"grad_norm": 0.32185524702072144,
|
| 2200 |
+
"learning_rate": 0.00022954812897152272,
|
| 2201 |
+
"loss": 2.8334,
|
| 2202 |
+
"step": 294000
|
| 2203 |
+
},
|
| 2204 |
+
{
|
| 2205 |
+
"epoch": 15.86,
|
| 2206 |
+
"grad_norm": 0.2747587561607361,
|
| 2207 |
+
"learning_rate": 0.00022660626029654036,
|
| 2208 |
+
"loss": 2.8351,
|
| 2209 |
+
"step": 295000
|
| 2210 |
+
},
|
| 2211 |
+
{
|
| 2212 |
+
"epoch": 15.92,
|
| 2213 |
+
"grad_norm": 0.32215383648872375,
|
| 2214 |
+
"learning_rate": 0.000223664391621558,
|
| 2215 |
+
"loss": 2.8343,
|
| 2216 |
+
"step": 296000
|
| 2217 |
+
},
|
| 2218 |
+
{
|
| 2219 |
+
"epoch": 15.97,
|
| 2220 |
+
"grad_norm": 0.30099472403526306,
|
| 2221 |
+
"learning_rate": 0.00022072252294657566,
|
| 2222 |
+
"loss": 2.8352,
|
| 2223 |
+
"step": 297000
|
| 2224 |
+
},
|
| 2225 |
+
{
|
| 2226 |
+
"epoch": 16.0,
|
| 2227 |
+
"eval_accuracy": 0.4099246009434271,
|
| 2228 |
+
"eval_loss": 3.3998661041259766,
|
| 2229 |
+
"eval_runtime": 154.387,
|
| 2230 |
+
"eval_samples_per_second": 375.142,
|
| 2231 |
+
"eval_steps_per_second": 5.862,
|
| 2232 |
+
"step": 297536
|
| 2233 |
+
},
|
| 2234 |
+
{
|
| 2235 |
+
"epoch": 16.02,
|
| 2236 |
+
"grad_norm": 0.3042096495628357,
|
| 2237 |
+
"learning_rate": 0.0002177835961402683,
|
| 2238 |
+
"loss": 2.8115,
|
| 2239 |
+
"step": 298000
|
| 2240 |
+
},
|
| 2241 |
+
{
|
| 2242 |
+
"epoch": 16.08,
|
| 2243 |
+
"grad_norm": 0.29497480392456055,
|
| 2244 |
+
"learning_rate": 0.00021484172746528597,
|
| 2245 |
+
"loss": 2.7915,
|
| 2246 |
+
"step": 299000
|
| 2247 |
+
},
|
| 2248 |
+
{
|
| 2249 |
+
"epoch": 16.13,
|
| 2250 |
+
"grad_norm": 0.2896369993686676,
|
| 2251 |
+
"learning_rate": 0.0002118998587903036,
|
| 2252 |
+
"loss": 2.7949,
|
| 2253 |
+
"step": 300000
|
| 2254 |
+
},
|
| 2255 |
+
{
|
| 2256 |
+
"epoch": 16.19,
|
| 2257 |
+
"grad_norm": 0.3014323115348816,
|
| 2258 |
+
"learning_rate": 0.00020896093198399625,
|
| 2259 |
+
"loss": 2.7968,
|
| 2260 |
+
"step": 301000
|
| 2261 |
+
},
|
| 2262 |
+
{
|
| 2263 |
+
"epoch": 16.24,
|
| 2264 |
+
"grad_norm": 0.2915323078632355,
|
| 2265 |
+
"learning_rate": 0.00020601906330901387,
|
| 2266 |
+
"loss": 2.7961,
|
| 2267 |
+
"step": 302000
|
| 2268 |
+
},
|
| 2269 |
+
{
|
| 2270 |
+
"epoch": 16.29,
|
| 2271 |
+
"grad_norm": 0.28494659066200256,
|
| 2272 |
+
"learning_rate": 0.00020307719463403154,
|
| 2273 |
+
"loss": 2.7993,
|
| 2274 |
+
"step": 303000
|
| 2275 |
+
},
|
| 2276 |
+
{
|
| 2277 |
+
"epoch": 16.35,
|
| 2278 |
+
"grad_norm": 0.2689177095890045,
|
| 2279 |
+
"learning_rate": 0.00020013826782772418,
|
| 2280 |
+
"loss": 2.8054,
|
| 2281 |
+
"step": 304000
|
| 2282 |
+
},
|
| 2283 |
+
{
|
| 2284 |
+
"epoch": 16.4,
|
| 2285 |
+
"grad_norm": 0.290145605802536,
|
| 2286 |
+
"learning_rate": 0.00019719639915274182,
|
| 2287 |
+
"loss": 2.808,
|
| 2288 |
+
"step": 305000
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 16.46,
|
| 2292 |
+
"grad_norm": 0.306331992149353,
|
| 2293 |
+
"learning_rate": 0.00019425747234643446,
|
| 2294 |
+
"loss": 2.8077,
|
| 2295 |
+
"step": 306000
|
| 2296 |
+
},
|
| 2297 |
+
{
|
| 2298 |
+
"epoch": 16.51,
|
| 2299 |
+
"grad_norm": 0.28849080204963684,
|
| 2300 |
+
"learning_rate": 0.0001913156036714521,
|
| 2301 |
+
"loss": 2.808,
|
| 2302 |
+
"step": 307000
|
| 2303 |
+
},
|
| 2304 |
+
{
|
| 2305 |
+
"epoch": 16.56,
|
| 2306 |
+
"grad_norm": 0.33114251494407654,
|
| 2307 |
+
"learning_rate": 0.00018837373499646975,
|
| 2308 |
+
"loss": 2.8096,
|
| 2309 |
+
"step": 308000
|
| 2310 |
+
},
|
| 2311 |
+
{
|
| 2312 |
+
"epoch": 16.62,
|
| 2313 |
+
"grad_norm": 0.32337045669555664,
|
| 2314 |
+
"learning_rate": 0.0001854348081901624,
|
| 2315 |
+
"loss": 2.8101,
|
| 2316 |
+
"step": 309000
|
| 2317 |
+
},
|
| 2318 |
+
{
|
| 2319 |
+
"epoch": 16.67,
|
| 2320 |
+
"grad_norm": 0.3087981939315796,
|
| 2321 |
+
"learning_rate": 0.00018249293951518004,
|
| 2322 |
+
"loss": 2.8112,
|
| 2323 |
+
"step": 310000
|
| 2324 |
+
},
|
| 2325 |
+
{
|
| 2326 |
+
"epoch": 16.72,
|
| 2327 |
+
"grad_norm": 0.2813110947608948,
|
| 2328 |
+
"learning_rate": 0.00017955401270887267,
|
| 2329 |
+
"loss": 2.8153,
|
| 2330 |
+
"step": 311000
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 16.78,
|
| 2334 |
+
"grad_norm": 0.28998225927352905,
|
| 2335 |
+
"learning_rate": 0.00017661214403389035,
|
| 2336 |
+
"loss": 2.815,
|
| 2337 |
+
"step": 312000
|
| 2338 |
+
},
|
| 2339 |
+
{
|
| 2340 |
+
"epoch": 16.83,
|
| 2341 |
+
"grad_norm": 0.29306545853614807,
|
| 2342 |
+
"learning_rate": 0.00017367321722758298,
|
| 2343 |
+
"loss": 2.8161,
|
| 2344 |
+
"step": 313000
|
| 2345 |
+
},
|
| 2346 |
+
{
|
| 2347 |
+
"epoch": 16.89,
|
| 2348 |
+
"grad_norm": 0.3053112328052521,
|
| 2349 |
+
"learning_rate": 0.0001707313485526006,
|
| 2350 |
+
"loss": 2.8119,
|
| 2351 |
+
"step": 314000
|
| 2352 |
+
},
|
| 2353 |
+
{
|
| 2354 |
+
"epoch": 16.94,
|
| 2355 |
+
"grad_norm": 0.3361365795135498,
|
| 2356 |
+
"learning_rate": 0.00016778947987761828,
|
| 2357 |
+
"loss": 2.8147,
|
| 2358 |
+
"step": 315000
|
| 2359 |
+
},
|
| 2360 |
+
{
|
| 2361 |
+
"epoch": 16.99,
|
| 2362 |
+
"grad_norm": 0.30090630054473877,
|
| 2363 |
+
"learning_rate": 0.00016485349493998588,
|
| 2364 |
+
"loss": 2.8173,
|
| 2365 |
+
"step": 316000
|
| 2366 |
+
},
|
| 2367 |
+
{
|
| 2368 |
+
"epoch": 17.0,
|
| 2369 |
+
"eval_accuracy": 0.4103711006270949,
|
| 2370 |
+
"eval_loss": 3.3992600440979004,
|
| 2371 |
+
"eval_runtime": 154.0327,
|
| 2372 |
+
"eval_samples_per_second": 376.005,
|
| 2373 |
+
"eval_steps_per_second": 5.875,
|
| 2374 |
+
"step": 316132
|
| 2375 |
+
},
|
| 2376 |
+
{
|
| 2377 |
+
"epoch": 17.05,
|
| 2378 |
+
"grad_norm": 0.28460776805877686,
|
| 2379 |
+
"learning_rate": 0.00016191162626500352,
|
| 2380 |
+
"loss": 2.7809,
|
| 2381 |
+
"step": 317000
|
| 2382 |
+
},
|
| 2383 |
+
{
|
| 2384 |
+
"epoch": 17.1,
|
| 2385 |
+
"grad_norm": 0.2945604920387268,
|
| 2386 |
+
"learning_rate": 0.0001589697575900212,
|
| 2387 |
+
"loss": 2.7761,
|
| 2388 |
+
"step": 318000
|
| 2389 |
+
},
|
| 2390 |
+
{
|
| 2391 |
+
"epoch": 17.15,
|
| 2392 |
+
"grad_norm": 0.2811717987060547,
|
| 2393 |
+
"learning_rate": 0.00015603083078371383,
|
| 2394 |
+
"loss": 2.7791,
|
| 2395 |
+
"step": 319000
|
| 2396 |
+
},
|
| 2397 |
+
{
|
| 2398 |
+
"epoch": 17.21,
|
| 2399 |
+
"grad_norm": 0.28703057765960693,
|
| 2400 |
+
"learning_rate": 0.00015308896210873148,
|
| 2401 |
+
"loss": 2.785,
|
| 2402 |
+
"step": 320000
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 17.26,
|
| 2406 |
+
"grad_norm": 0.2974686622619629,
|
| 2407 |
+
"learning_rate": 0.00015014709343374912,
|
| 2408 |
+
"loss": 2.7862,
|
| 2409 |
+
"step": 321000
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 17.32,
|
| 2413 |
+
"grad_norm": 0.29753968119621277,
|
| 2414 |
+
"learning_rate": 0.00014720522475876677,
|
| 2415 |
+
"loss": 2.7875,
|
| 2416 |
+
"step": 322000
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 17.37,
|
| 2420 |
+
"grad_norm": 0.29779624938964844,
|
| 2421 |
+
"learning_rate": 0.0001442662979524594,
|
| 2422 |
+
"loss": 2.7876,
|
| 2423 |
+
"step": 323000
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 17.42,
|
| 2427 |
+
"grad_norm": 0.281829833984375,
|
| 2428 |
+
"learning_rate": 0.00014132442927747705,
|
| 2429 |
+
"loss": 2.7909,
|
| 2430 |
+
"step": 324000
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 17.48,
|
| 2434 |
+
"grad_norm": 0.3030970096588135,
|
| 2435 |
+
"learning_rate": 0.0001383825606024947,
|
| 2436 |
+
"loss": 2.7883,
|
| 2437 |
+
"step": 325000
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 17.53,
|
| 2441 |
+
"grad_norm": 0.28257811069488525,
|
| 2442 |
+
"learning_rate": 0.00013544363379618734,
|
| 2443 |
+
"loss": 2.793,
|
| 2444 |
+
"step": 326000
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 17.58,
|
| 2448 |
+
"grad_norm": 0.3062205910682678,
|
| 2449 |
+
"learning_rate": 0.00013250176512120498,
|
| 2450 |
+
"loss": 2.7923,
|
| 2451 |
+
"step": 327000
|
| 2452 |
+
},
|
| 2453 |
+
{
|
| 2454 |
+
"epoch": 17.64,
|
| 2455 |
+
"grad_norm": 0.29610031843185425,
|
| 2456 |
+
"learning_rate": 0.00012956283831489762,
|
| 2457 |
+
"loss": 2.7947,
|
| 2458 |
+
"step": 328000
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 17.69,
|
| 2462 |
+
"grad_norm": 0.31379351019859314,
|
| 2463 |
+
"learning_rate": 0.0001266209696399153,
|
| 2464 |
+
"loss": 2.793,
|
| 2465 |
+
"step": 329000
|
| 2466 |
+
},
|
| 2467 |
+
{
|
| 2468 |
+
"epoch": 17.75,
|
| 2469 |
+
"grad_norm": 0.29375967383384705,
|
| 2470 |
+
"learning_rate": 0.00012367910096493294,
|
| 2471 |
+
"loss": 2.7971,
|
| 2472 |
+
"step": 330000
|
| 2473 |
+
},
|
| 2474 |
+
{
|
| 2475 |
+
"epoch": 17.8,
|
| 2476 |
+
"grad_norm": 0.30711761116981506,
|
| 2477 |
+
"learning_rate": 0.00012073723228995057,
|
| 2478 |
+
"loss": 2.7963,
|
| 2479 |
+
"step": 331000
|
| 2480 |
+
},
|
| 2481 |
+
{
|
| 2482 |
+
"epoch": 17.85,
|
| 2483 |
+
"grad_norm": 0.29258862137794495,
|
| 2484 |
+
"learning_rate": 0.00011780124735231819,
|
| 2485 |
+
"loss": 2.7937,
|
| 2486 |
+
"step": 332000
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 17.91,
|
| 2490 |
+
"grad_norm": 0.2702758014202118,
|
| 2491 |
+
"learning_rate": 0.00011485937867733585,
|
| 2492 |
+
"loss": 2.7942,
|
| 2493 |
+
"step": 333000
|
| 2494 |
+
},
|
| 2495 |
+
{
|
| 2496 |
+
"epoch": 17.96,
|
| 2497 |
+
"grad_norm": 0.2748437523841858,
|
| 2498 |
+
"learning_rate": 0.00011191751000235349,
|
| 2499 |
+
"loss": 2.8005,
|
| 2500 |
+
"step": 334000
|
| 2501 |
+
},
|
| 2502 |
+
{
|
| 2503 |
+
"epoch": 18.0,
|
| 2504 |
+
"eval_accuracy": 0.4099377016714312,
|
| 2505 |
+
"eval_loss": 3.4154224395751953,
|
| 2506 |
+
"eval_runtime": 154.4547,
|
| 2507 |
+
"eval_samples_per_second": 374.977,
|
| 2508 |
+
"eval_steps_per_second": 5.859,
|
| 2509 |
+
"step": 334728
|
| 2510 |
+
},
|
| 2511 |
+
{
|
| 2512 |
+
"epoch": 18.01,
|
| 2513 |
+
"grad_norm": 0.29431185126304626,
|
| 2514 |
+
"learning_rate": 0.00010897858319604613,
|
| 2515 |
+
"loss": 2.7895,
|
| 2516 |
+
"step": 335000
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 18.07,
|
| 2520 |
+
"grad_norm": 0.3129688799381256,
|
| 2521 |
+
"learning_rate": 0.00010603965638973876,
|
| 2522 |
+
"loss": 2.7669,
|
| 2523 |
+
"step": 336000
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 18.12,
|
| 2527 |
+
"grad_norm": 0.29429852962493896,
|
| 2528 |
+
"learning_rate": 0.00010309778771475643,
|
| 2529 |
+
"loss": 2.7668,
|
| 2530 |
+
"step": 337000
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"epoch": 18.18,
|
| 2534 |
+
"grad_norm": 0.30273085832595825,
|
| 2535 |
+
"learning_rate": 0.00010015591903977407,
|
| 2536 |
+
"loss": 2.7695,
|
| 2537 |
+
"step": 338000
|
| 2538 |
+
},
|
| 2539 |
+
{
|
| 2540 |
+
"epoch": 18.23,
|
| 2541 |
+
"grad_norm": 0.3032689392566681,
|
| 2542 |
+
"learning_rate": 9.72169922334667e-05,
|
| 2543 |
+
"loss": 2.7712,
|
| 2544 |
+
"step": 339000
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"epoch": 18.28,
|
| 2548 |
+
"grad_norm": 0.3074301779270172,
|
| 2549 |
+
"learning_rate": 9.427512355848434e-05,
|
| 2550 |
+
"loss": 2.7691,
|
| 2551 |
+
"step": 340000
|
| 2552 |
+
},
|
| 2553 |
+
{
|
| 2554 |
+
"epoch": 18.34,
|
| 2555 |
+
"grad_norm": 0.30292806029319763,
|
| 2556 |
+
"learning_rate": 9.1333254883502e-05,
|
| 2557 |
+
"loss": 2.7725,
|
| 2558 |
+
"step": 341000
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 18.39,
|
| 2562 |
+
"grad_norm": 0.31416478753089905,
|
| 2563 |
+
"learning_rate": 8.839432807719464e-05,
|
| 2564 |
+
"loss": 2.7718,
|
| 2565 |
+
"step": 342000
|
| 2566 |
+
},
|
| 2567 |
+
{
|
| 2568 |
+
"epoch": 18.44,
|
| 2569 |
+
"grad_norm": 0.3108297884464264,
|
| 2570 |
+
"learning_rate": 8.545245940221228e-05,
|
| 2571 |
+
"loss": 2.7724,
|
| 2572 |
+
"step": 343000
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"epoch": 18.5,
|
| 2576 |
+
"grad_norm": 0.31622520089149475,
|
| 2577 |
+
"learning_rate": 8.251059072722994e-05,
|
| 2578 |
+
"loss": 2.7771,
|
| 2579 |
+
"step": 344000
|
| 2580 |
+
},
|
| 2581 |
+
{
|
| 2582 |
+
"epoch": 18.55,
|
| 2583 |
+
"grad_norm": 0.2999779284000397,
|
| 2584 |
+
"learning_rate": 7.956872205224759e-05,
|
| 2585 |
+
"loss": 2.7739,
|
| 2586 |
+
"step": 345000
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"epoch": 18.61,
|
| 2590 |
+
"grad_norm": 0.28529179096221924,
|
| 2591 |
+
"learning_rate": 7.662685337726524e-05,
|
| 2592 |
+
"loss": 2.7776,
|
| 2593 |
+
"step": 346000
|
| 2594 |
+
},
|
| 2595 |
+
{
|
| 2596 |
+
"epoch": 18.66,
|
| 2597 |
+
"grad_norm": 0.3071877956390381,
|
| 2598 |
+
"learning_rate": 7.368792657095789e-05,
|
| 2599 |
+
"loss": 2.7764,
|
| 2600 |
+
"step": 347000
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 18.71,
|
| 2604 |
+
"grad_norm": 0.3046530783176422,
|
| 2605 |
+
"learning_rate": 7.074605789597552e-05,
|
| 2606 |
+
"loss": 2.776,
|
| 2607 |
+
"step": 348000
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 18.77,
|
| 2611 |
+
"grad_norm": 0.3025725781917572,
|
| 2612 |
+
"learning_rate": 6.780418922099317e-05,
|
| 2613 |
+
"loss": 2.7787,
|
| 2614 |
+
"step": 349000
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"epoch": 18.82,
|
| 2618 |
+
"grad_norm": 0.3011409044265747,
|
| 2619 |
+
"learning_rate": 6.48652624146858e-05,
|
| 2620 |
+
"loss": 2.7794,
|
| 2621 |
+
"step": 350000
|
| 2622 |
+
},
|
| 2623 |
+
{
|
| 2624 |
+
"epoch": 18.88,
|
| 2625 |
+
"grad_norm": 0.2958693206310272,
|
| 2626 |
+
"learning_rate": 6.192339373970346e-05,
|
| 2627 |
+
"loss": 2.7778,
|
| 2628 |
+
"step": 351000
|
| 2629 |
+
},
|
| 2630 |
+
{
|
| 2631 |
+
"epoch": 18.93,
|
| 2632 |
+
"grad_norm": 0.3003888726234436,
|
| 2633 |
+
"learning_rate": 5.898152506472111e-05,
|
| 2634 |
+
"loss": 2.7786,
|
| 2635 |
+
"step": 352000
|
| 2636 |
+
},
|
| 2637 |
+
{
|
| 2638 |
+
"epoch": 18.98,
|
| 2639 |
+
"grad_norm": 0.29960742592811584,
|
| 2640 |
+
"learning_rate": 5.604554012708873e-05,
|
| 2641 |
+
"loss": 2.7796,
|
| 2642 |
+
"step": 353000
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 19.0,
|
| 2646 |
+
"eval_accuracy": 0.41019803665099974,
|
| 2647 |
+
"eval_loss": 3.4155313968658447,
|
| 2648 |
+
"eval_runtime": 154.2799,
|
| 2649 |
+
"eval_samples_per_second": 375.402,
|
| 2650 |
+
"eval_steps_per_second": 5.866,
|
| 2651 |
+
"step": 353324
|
| 2652 |
+
},
|
| 2653 |
+
{
|
| 2654 |
+
"epoch": 19.04,
|
| 2655 |
+
"grad_norm": 0.29620590806007385,
|
| 2656 |
+
"learning_rate": 5.310367145210638e-05,
|
| 2657 |
+
"loss": 2.7661,
|
| 2658 |
+
"step": 354000
|
| 2659 |
+
},
|
| 2660 |
+
{
|
| 2661 |
+
"epoch": 19.09,
|
| 2662 |
+
"grad_norm": 0.3027678430080414,
|
| 2663 |
+
"learning_rate": 5.016180277712403e-05,
|
| 2664 |
+
"loss": 2.7561,
|
| 2665 |
+
"step": 355000
|
| 2666 |
+
},
|
| 2667 |
+
{
|
| 2668 |
+
"epoch": 19.14,
|
| 2669 |
+
"grad_norm": 0.31339025497436523,
|
| 2670 |
+
"learning_rate": 4.721993410214168e-05,
|
| 2671 |
+
"loss": 2.757,
|
| 2672 |
+
"step": 356000
|
| 2673 |
+
},
|
| 2674 |
+
{
|
| 2675 |
+
"epoch": 19.2,
|
| 2676 |
+
"grad_norm": 0.31296485662460327,
|
| 2677 |
+
"learning_rate": 4.4278065427159335e-05,
|
| 2678 |
+
"loss": 2.7561,
|
| 2679 |
+
"step": 357000
|
| 2680 |
+
},
|
| 2681 |
+
{
|
| 2682 |
+
"epoch": 19.25,
|
| 2683 |
+
"grad_norm": 0.3046360909938812,
|
| 2684 |
+
"learning_rate": 4.1339138620851965e-05,
|
| 2685 |
+
"loss": 2.7601,
|
| 2686 |
+
"step": 358000
|
| 2687 |
+
},
|
| 2688 |
+
{
|
| 2689 |
+
"epoch": 19.31,
|
| 2690 |
+
"grad_norm": 0.3128166198730469,
|
| 2691 |
+
"learning_rate": 3.839726994586962e-05,
|
| 2692 |
+
"loss": 2.7618,
|
| 2693 |
+
"step": 359000
|
| 2694 |
+
},
|
| 2695 |
+
{
|
| 2696 |
+
"epoch": 19.36,
|
| 2697 |
+
"grad_norm": 0.3523208200931549,
|
| 2698 |
+
"learning_rate": 3.545540127088727e-05,
|
| 2699 |
+
"loss": 2.7576,
|
| 2700 |
+
"step": 360000
|
| 2701 |
+
},
|
| 2702 |
+
{
|
| 2703 |
+
"epoch": 19.41,
|
| 2704 |
+
"grad_norm": 0.30089735984802246,
|
| 2705 |
+
"learning_rate": 3.25164744645799e-05,
|
| 2706 |
+
"loss": 2.7602,
|
| 2707 |
+
"step": 361000
|
| 2708 |
+
},
|
| 2709 |
+
{
|
| 2710 |
+
"epoch": 19.47,
|
| 2711 |
+
"grad_norm": 0.3562370538711548,
|
| 2712 |
+
"learning_rate": 2.957460578959755e-05,
|
| 2713 |
+
"loss": 2.7599,
|
| 2714 |
+
"step": 362000
|
| 2715 |
+
},
|
| 2716 |
+
{
|
| 2717 |
+
"epoch": 19.52,
|
| 2718 |
+
"grad_norm": 0.33757689595222473,
|
| 2719 |
+
"learning_rate": 2.6635678983290187e-05,
|
| 2720 |
+
"loss": 2.7591,
|
| 2721 |
+
"step": 363000
|
| 2722 |
+
},
|
| 2723 |
+
{
|
| 2724 |
+
"epoch": 19.57,
|
| 2725 |
+
"grad_norm": 0.313703715801239,
|
| 2726 |
+
"learning_rate": 2.3693810308307837e-05,
|
| 2727 |
+
"loss": 2.7621,
|
| 2728 |
+
"step": 364000
|
| 2729 |
+
},
|
| 2730 |
+
{
|
| 2731 |
+
"epoch": 19.63,
|
| 2732 |
+
"grad_norm": 0.31429997086524963,
|
| 2733 |
+
"learning_rate": 2.075488350200047e-05,
|
| 2734 |
+
"loss": 2.7585,
|
| 2735 |
+
"step": 365000
|
| 2736 |
+
},
|
| 2737 |
+
{
|
| 2738 |
+
"epoch": 19.68,
|
| 2739 |
+
"grad_norm": 0.29419606924057007,
|
| 2740 |
+
"learning_rate": 1.7813014827018123e-05,
|
| 2741 |
+
"loss": 2.7598,
|
| 2742 |
+
"step": 366000
|
| 2743 |
+
},
|
| 2744 |
+
{
|
| 2745 |
+
"epoch": 19.74,
|
| 2746 |
+
"grad_norm": 0.28869205713272095,
|
| 2747 |
+
"learning_rate": 1.4874088020710754e-05,
|
| 2748 |
+
"loss": 2.7631,
|
| 2749 |
+
"step": 367000
|
| 2750 |
+
},
|
| 2751 |
+
{
|
| 2752 |
+
"epoch": 19.79,
|
| 2753 |
+
"grad_norm": 0.2999878227710724,
|
| 2754 |
+
"learning_rate": 1.1932219345728408e-05,
|
| 2755 |
+
"loss": 2.7615,
|
| 2756 |
+
"step": 368000
|
| 2757 |
+
},
|
| 2758 |
+
{
|
| 2759 |
+
"epoch": 19.84,
|
| 2760 |
+
"grad_norm": 0.31875109672546387,
|
| 2761 |
+
"learning_rate": 8.990350670746058e-06,
|
| 2762 |
+
"loss": 2.7607,
|
| 2763 |
+
"step": 369000
|
| 2764 |
+
},
|
| 2765 |
+
{
|
| 2766 |
+
"epoch": 19.9,
|
| 2767 |
+
"grad_norm": 0.2959778308868408,
|
| 2768 |
+
"learning_rate": 6.051423864438692e-06,
|
| 2769 |
+
"loss": 2.7631,
|
| 2770 |
+
"step": 370000
|
| 2771 |
+
},
|
| 2772 |
+
{
|
| 2773 |
+
"epoch": 19.95,
|
| 2774 |
+
"grad_norm": 0.3109806478023529,
|
| 2775 |
+
"learning_rate": 3.1095551894563426e-06,
|
| 2776 |
+
"loss": 2.7606,
|
| 2777 |
+
"step": 371000
|
| 2778 |
+
},
|
| 2779 |
+
{
|
| 2780 |
+
"epoch": 20.0,
|
| 2781 |
+
"eval_accuracy": 0.40996444059320364,
|
| 2782 |
+
"eval_loss": 3.424755573272705,
|
| 2783 |
+
"eval_runtime": 154.1805,
|
| 2784 |
+
"eval_samples_per_second": 375.644,
|
| 2785 |
+
"eval_steps_per_second": 5.87,
|
| 2786 |
+
"step": 371920
|
| 2787 |
+
},
|
| 2788 |
+
{
|
| 2789 |
+
"epoch": 20.0,
|
| 2790 |
+
"step": 371920,
|
| 2791 |
+
"total_flos": 1.5670047538944e+18,
|
| 2792 |
+
"train_loss": 3.0296430246833577,
|
| 2793 |
+
"train_runtime": 81436.2617,
|
| 2794 |
+
"train_samples_per_second": 146.143,
|
| 2795 |
+
"train_steps_per_second": 4.567
|
| 2796 |
+
}
|
| 2797 |
+
],
|
| 2798 |
+
"logging_steps": 1000,
|
| 2799 |
+
"max_steps": 371920,
|
| 2800 |
+
"num_input_tokens_seen": 0,
|
| 2801 |
+
"num_train_epochs": 20,
|
| 2802 |
+
"save_steps": 5000,
|
| 2803 |
+
"total_flos": 1.5670047538944e+18,
|
| 2804 |
+
"train_batch_size": 32,
|
| 2805 |
+
"trial_name": null,
|
| 2806 |
+
"trial_params": null
|
| 2807 |
+
}
|