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_adj-seed_1024-1e-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_adj-seed_1024-1e-3
|
| 15 |
|
| 16 |
-
This model was trained from scratch on
|
| 17 |
It achieves the following results on the evaluation set:
|
| 18 |
- Loss: 3.4042
|
| 19 |
- Accuracy: 0.4105
|
|
|
|
| 1 |
---
|
| 2 |
tags:
|
| 3 |
- generated_from_trainer
|
| 4 |
+
datasets:
|
| 5 |
+
- kanishka/counterfactual_babylm_aann_high_variability_adj
|
| 6 |
metrics:
|
| 7 |
- accuracy
|
| 8 |
model-index:
|
| 9 |
- name: smolm-autoreg-bpe-counterfactual_babylm_aann_high_variability_adj-seed_1024-1e-3
|
| 10 |
+
results:
|
| 11 |
+
- task:
|
| 12 |
+
name: Causal Language Modeling
|
| 13 |
+
type: text-generation
|
| 14 |
+
dataset:
|
| 15 |
+
name: kanishka/counterfactual_babylm_aann_high_variability_adj
|
| 16 |
+
type: kanishka/counterfactual_babylm_aann_high_variability_adj
|
| 17 |
+
metrics:
|
| 18 |
+
- name: Accuracy
|
| 19 |
+
type: accuracy
|
| 20 |
+
value: 0.41050110032238096
|
| 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_adj-seed_1024-1e-3
|
| 27 |
|
| 28 |
+
This model was trained from scratch on the kanishka/counterfactual_babylm_aann_high_variability_adj dataset.
|
| 29 |
It achieves the following results on the evaluation set:
|
| 30 |
- Loss: 3.4042
|
| 31 |
- Accuracy: 0.4105
|
all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"eval_accuracy": 0.41050110032238096,
|
| 4 |
+
"eval_loss": 3.4041664600372314,
|
| 5 |
+
"eval_runtime": 154.211,
|
| 6 |
+
"eval_samples": 57919,
|
| 7 |
+
"eval_samples_per_second": 375.583,
|
| 8 |
+
"eval_steps_per_second": 5.869,
|
| 9 |
+
"perplexity": 30.0892047136139,
|
| 10 |
+
"train_loss": 3.0269814196563325,
|
| 11 |
+
"train_runtime": 81274.1105,
|
| 12 |
+
"train_samples": 594994,
|
| 13 |
+
"train_samples_per_second": 146.417,
|
| 14 |
+
"train_steps_per_second": 4.576
|
| 15 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"eval_accuracy": 0.41050110032238096,
|
| 4 |
+
"eval_loss": 3.4041664600372314,
|
| 5 |
+
"eval_runtime": 154.211,
|
| 6 |
+
"eval_samples": 57919,
|
| 7 |
+
"eval_samples_per_second": 375.583,
|
| 8 |
+
"eval_steps_per_second": 5.869,
|
| 9 |
+
"perplexity": 30.0892047136139
|
| 10 |
+
}
|
runs/Jul17_11-20-59_phyl-ling-p01.la.utexas.edu/events.out.tfevents.1721315061.phyl-ling-p01.la.utexas.edu.2998199.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:145328aebaab4629c719aa02854acd260c52c5acd683cb1560392d5b77834e95
|
| 3 |
+
size 417
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"train_loss": 3.0269814196563325,
|
| 4 |
+
"train_runtime": 81274.1105,
|
| 5 |
+
"train_samples": 594994,
|
| 6 |
+
"train_samples_per_second": 146.417,
|
| 7 |
+
"train_steps_per_second": 4.576
|
| 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": 371880,
|
| 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": 1.0183484554290771,
|
| 14 |
+
"learning_rate": 3.125e-05,
|
| 15 |
+
"loss": 6.2316,
|
| 16 |
+
"step": 1000
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.11,
|
| 20 |
+
"grad_norm": 0.9109435677528381,
|
| 21 |
+
"learning_rate": 6.25e-05,
|
| 22 |
+
"loss": 5.0067,
|
| 23 |
+
"step": 2000
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.16,
|
| 27 |
+
"grad_norm": 0.8983614444732666,
|
| 28 |
+
"learning_rate": 9.375e-05,
|
| 29 |
+
"loss": 4.675,
|
| 30 |
+
"step": 3000
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.22,
|
| 34 |
+
"grad_norm": 0.8370816111564636,
|
| 35 |
+
"learning_rate": 0.000125,
|
| 36 |
+
"loss": 4.4579,
|
| 37 |
+
"step": 4000
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.27,
|
| 41 |
+
"grad_norm": 0.7996561527252197,
|
| 42 |
+
"learning_rate": 0.00015625,
|
| 43 |
+
"loss": 4.301,
|
| 44 |
+
"step": 5000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.32,
|
| 48 |
+
"grad_norm": 0.7139356732368469,
|
| 49 |
+
"learning_rate": 0.0001875,
|
| 50 |
+
"loss": 4.1791,
|
| 51 |
+
"step": 6000
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.38,
|
| 55 |
+
"grad_norm": 0.720668375492096,
|
| 56 |
+
"learning_rate": 0.00021875,
|
| 57 |
+
"loss": 4.0705,
|
| 58 |
+
"step": 7000
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.43,
|
| 62 |
+
"grad_norm": 0.6427111029624939,
|
| 63 |
+
"learning_rate": 0.00025,
|
| 64 |
+
"loss": 3.9831,
|
| 65 |
+
"step": 8000
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.48,
|
| 69 |
+
"grad_norm": 0.6335527300834656,
|
| 70 |
+
"learning_rate": 0.00028121875,
|
| 71 |
+
"loss": 3.9099,
|
| 72 |
+
"step": 9000
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.54,
|
| 76 |
+
"grad_norm": 0.6659609079360962,
|
| 77 |
+
"learning_rate": 0.00031246875000000003,
|
| 78 |
+
"loss": 3.8399,
|
| 79 |
+
"step": 10000
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.59,
|
| 83 |
+
"grad_norm": 0.5713762044906616,
|
| 84 |
+
"learning_rate": 0.00034371875,
|
| 85 |
+
"loss": 3.7978,
|
| 86 |
+
"step": 11000
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.65,
|
| 90 |
+
"grad_norm": 0.5290955901145935,
|
| 91 |
+
"learning_rate": 0.00037490625,
|
| 92 |
+
"loss": 3.7599,
|
| 93 |
+
"step": 12000
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.7,
|
| 97 |
+
"grad_norm": 0.4921986162662506,
|
| 98 |
+
"learning_rate": 0.00040615625,
|
| 99 |
+
"loss": 3.7187,
|
| 100 |
+
"step": 13000
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.75,
|
| 104 |
+
"grad_norm": 0.5067416429519653,
|
| 105 |
+
"learning_rate": 0.00043740625,
|
| 106 |
+
"loss": 3.6894,
|
| 107 |
+
"step": 14000
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.81,
|
| 111 |
+
"grad_norm": 0.4463486671447754,
|
| 112 |
+
"learning_rate": 0.000468625,
|
| 113 |
+
"loss": 3.6571,
|
| 114 |
+
"step": 15000
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.86,
|
| 118 |
+
"grad_norm": 0.402967244386673,
|
| 119 |
+
"learning_rate": 0.000499875,
|
| 120 |
+
"loss": 3.6394,
|
| 121 |
+
"step": 16000
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.91,
|
| 125 |
+
"grad_norm": 0.39186057448387146,
|
| 126 |
+
"learning_rate": 0.000531125,
|
| 127 |
+
"loss": 3.6159,
|
| 128 |
+
"step": 17000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.97,
|
| 132 |
+
"grad_norm": 0.35502514243125916,
|
| 133 |
+
"learning_rate": 0.00056234375,
|
| 134 |
+
"loss": 3.5971,
|
| 135 |
+
"step": 18000
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 1.0,
|
| 139 |
+
"eval_accuracy": 0.3617747930945827,
|
| 140 |
+
"eval_loss": 3.736064910888672,
|
| 141 |
+
"eval_runtime": 152.7106,
|
| 142 |
+
"eval_samples_per_second": 379.273,
|
| 143 |
+
"eval_steps_per_second": 5.926,
|
| 144 |
+
"step": 18594
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 1.02,
|
| 148 |
+
"grad_norm": 0.35789868235588074,
|
| 149 |
+
"learning_rate": 0.00059359375,
|
| 150 |
+
"loss": 3.5664,
|
| 151 |
+
"step": 19000
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 1.08,
|
| 155 |
+
"grad_norm": 0.33985644578933716,
|
| 156 |
+
"learning_rate": 0.0006248125,
|
| 157 |
+
"loss": 3.5412,
|
| 158 |
+
"step": 20000
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.13,
|
| 162 |
+
"grad_norm": 0.32881930470466614,
|
| 163 |
+
"learning_rate": 0.0006560625,
|
| 164 |
+
"loss": 3.5403,
|
| 165 |
+
"step": 21000
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 1.18,
|
| 169 |
+
"grad_norm": 0.3065691292285919,
|
| 170 |
+
"learning_rate": 0.0006873125,
|
| 171 |
+
"loss": 3.5256,
|
| 172 |
+
"step": 22000
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 1.24,
|
| 176 |
+
"grad_norm": 0.27938657999038696,
|
| 177 |
+
"learning_rate": 0.00071853125,
|
| 178 |
+
"loss": 3.5151,
|
| 179 |
+
"step": 23000
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.29,
|
| 183 |
+
"grad_norm": 0.3042667508125305,
|
| 184 |
+
"learning_rate": 0.00074978125,
|
| 185 |
+
"loss": 3.4991,
|
| 186 |
+
"step": 24000
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 1.34,
|
| 190 |
+
"grad_norm": 0.3247887194156647,
|
| 191 |
+
"learning_rate": 0.0007810312499999999,
|
| 192 |
+
"loss": 3.4982,
|
| 193 |
+
"step": 25000
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 1.4,
|
| 197 |
+
"grad_norm": 0.27074122428894043,
|
| 198 |
+
"learning_rate": 0.00081225,
|
| 199 |
+
"loss": 3.4822,
|
| 200 |
+
"step": 26000
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 1.45,
|
| 204 |
+
"grad_norm": 0.27802783250808716,
|
| 205 |
+
"learning_rate": 0.0008435000000000001,
|
| 206 |
+
"loss": 3.4749,
|
| 207 |
+
"step": 27000
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 1.51,
|
| 211 |
+
"grad_norm": 0.26988905668258667,
|
| 212 |
+
"learning_rate": 0.0008746874999999999,
|
| 213 |
+
"loss": 3.4671,
|
| 214 |
+
"step": 28000
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 1.56,
|
| 218 |
+
"grad_norm": 0.25848186016082764,
|
| 219 |
+
"learning_rate": 0.0009059375,
|
| 220 |
+
"loss": 3.4568,
|
| 221 |
+
"step": 29000
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.61,
|
| 225 |
+
"grad_norm": 0.24958929419517517,
|
| 226 |
+
"learning_rate": 0.0009371875,
|
| 227 |
+
"loss": 3.4476,
|
| 228 |
+
"step": 30000
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 1.67,
|
| 232 |
+
"grad_norm": 0.2539744973182678,
|
| 233 |
+
"learning_rate": 0.0009684375,
|
| 234 |
+
"loss": 3.4437,
|
| 235 |
+
"step": 31000
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 1.72,
|
| 239 |
+
"grad_norm": 0.23232826590538025,
|
| 240 |
+
"learning_rate": 0.0009996562500000001,
|
| 241 |
+
"loss": 3.4327,
|
| 242 |
+
"step": 32000
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.77,
|
| 246 |
+
"grad_norm": 0.24366217851638794,
|
| 247 |
+
"learning_rate": 0.0009970901494645169,
|
| 248 |
+
"loss": 3.4251,
|
| 249 |
+
"step": 33000
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 1.83,
|
| 253 |
+
"grad_norm": 0.2503926753997803,
|
| 254 |
+
"learning_rate": 0.0009941479345651407,
|
| 255 |
+
"loss": 3.4122,
|
| 256 |
+
"step": 34000
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 1.88,
|
| 260 |
+
"grad_norm": 0.24379245936870575,
|
| 261 |
+
"learning_rate": 0.0009912057196657645,
|
| 262 |
+
"loss": 3.3993,
|
| 263 |
+
"step": 35000
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 1.94,
|
| 267 |
+
"grad_norm": 0.22166165709495544,
|
| 268 |
+
"learning_rate": 0.000988263504766388,
|
| 269 |
+
"loss": 3.3923,
|
| 270 |
+
"step": 36000
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 1.99,
|
| 274 |
+
"grad_norm": 0.23829974234104156,
|
| 275 |
+
"learning_rate": 0.0009853242320819113,
|
| 276 |
+
"loss": 3.3831,
|
| 277 |
+
"step": 37000
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 2.0,
|
| 281 |
+
"eval_accuracy": 0.38096548762618504,
|
| 282 |
+
"eval_loss": 3.564042091369629,
|
| 283 |
+
"eval_runtime": 153.9584,
|
| 284 |
+
"eval_samples_per_second": 376.199,
|
| 285 |
+
"eval_steps_per_second": 5.878,
|
| 286 |
+
"step": 37188
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 2.04,
|
| 290 |
+
"grad_norm": 0.22480344772338867,
|
| 291 |
+
"learning_rate": 0.000982382017182535,
|
| 292 |
+
"loss": 3.3353,
|
| 293 |
+
"step": 38000
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 2.1,
|
| 297 |
+
"grad_norm": 0.2300485223531723,
|
| 298 |
+
"learning_rate": 0.0009794427444980581,
|
| 299 |
+
"loss": 3.3295,
|
| 300 |
+
"step": 39000
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 2.15,
|
| 304 |
+
"grad_norm": 0.21135911345481873,
|
| 305 |
+
"learning_rate": 0.0009765005295986819,
|
| 306 |
+
"loss": 3.3236,
|
| 307 |
+
"step": 40000
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 2.21,
|
| 311 |
+
"grad_norm": 0.20723822712898254,
|
| 312 |
+
"learning_rate": 0.0009735583146993057,
|
| 313 |
+
"loss": 3.319,
|
| 314 |
+
"step": 41000
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 2.26,
|
| 318 |
+
"grad_norm": 0.2093385010957718,
|
| 319 |
+
"learning_rate": 0.0009706160997999294,
|
| 320 |
+
"loss": 3.3158,
|
| 321 |
+
"step": 42000
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 2.31,
|
| 325 |
+
"grad_norm": 0.21909525990486145,
|
| 326 |
+
"learning_rate": 0.0009676768271154526,
|
| 327 |
+
"loss": 3.3127,
|
| 328 |
+
"step": 43000
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 2.37,
|
| 332 |
+
"grad_norm": 0.23994240164756775,
|
| 333 |
+
"learning_rate": 0.0009647375544309756,
|
| 334 |
+
"loss": 3.3092,
|
| 335 |
+
"step": 44000
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 2.42,
|
| 339 |
+
"grad_norm": 0.2161034643650055,
|
| 340 |
+
"learning_rate": 0.0009617953395315993,
|
| 341 |
+
"loss": 3.2979,
|
| 342 |
+
"step": 45000
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 2.47,
|
| 346 |
+
"grad_norm": 0.23253212869167328,
|
| 347 |
+
"learning_rate": 0.0009588531246322232,
|
| 348 |
+
"loss": 3.291,
|
| 349 |
+
"step": 46000
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 2.53,
|
| 353 |
+
"grad_norm": 0.22771930694580078,
|
| 354 |
+
"learning_rate": 0.0009559109097328469,
|
| 355 |
+
"loss": 3.2867,
|
| 356 |
+
"step": 47000
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.58,
|
| 360 |
+
"grad_norm": 0.18900541961193085,
|
| 361 |
+
"learning_rate": 0.00095297163704837,
|
| 362 |
+
"loss": 3.2844,
|
| 363 |
+
"step": 48000
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 2.64,
|
| 367 |
+
"grad_norm": 0.23678931593894958,
|
| 368 |
+
"learning_rate": 0.0009500294221489938,
|
| 369 |
+
"loss": 3.2812,
|
| 370 |
+
"step": 49000
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 2.69,
|
| 374 |
+
"grad_norm": 0.24454990029335022,
|
| 375 |
+
"learning_rate": 0.0009470872072496175,
|
| 376 |
+
"loss": 3.277,
|
| 377 |
+
"step": 50000
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 2.74,
|
| 381 |
+
"grad_norm": 0.22431446611881256,
|
| 382 |
+
"learning_rate": 0.0009441479345651407,
|
| 383 |
+
"loss": 3.2715,
|
| 384 |
+
"step": 51000
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 2.8,
|
| 388 |
+
"grad_norm": 0.22084662318229675,
|
| 389 |
+
"learning_rate": 0.0009412057196657645,
|
| 390 |
+
"loss": 3.2696,
|
| 391 |
+
"step": 52000
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 2.85,
|
| 395 |
+
"grad_norm": 0.20824232697486877,
|
| 396 |
+
"learning_rate": 0.0009382635047663881,
|
| 397 |
+
"loss": 3.2631,
|
| 398 |
+
"step": 53000
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 2.9,
|
| 402 |
+
"grad_norm": 0.20131325721740723,
|
| 403 |
+
"learning_rate": 0.0009353271742968107,
|
| 404 |
+
"loss": 3.2608,
|
| 405 |
+
"step": 54000
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 2.96,
|
| 409 |
+
"grad_norm": 0.2070458084344864,
|
| 410 |
+
"learning_rate": 0.0009323849593974344,
|
| 411 |
+
"loss": 3.2562,
|
| 412 |
+
"step": 55000
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 3.0,
|
| 416 |
+
"eval_accuracy": 0.3928006125285796,
|
| 417 |
+
"eval_loss": 3.451838970184326,
|
| 418 |
+
"eval_runtime": 153.9331,
|
| 419 |
+
"eval_samples_per_second": 376.261,
|
| 420 |
+
"eval_steps_per_second": 5.879,
|
| 421 |
+
"step": 55782
|
| 422 |
+
},
|
| 423 |
+
{
|
| 424 |
+
"epoch": 3.01,
|
| 425 |
+
"grad_norm": 0.20725230872631073,
|
| 426 |
+
"learning_rate": 0.0009294427444980582,
|
| 427 |
+
"loss": 3.2396,
|
| 428 |
+
"step": 56000
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 3.07,
|
| 432 |
+
"grad_norm": 0.2388351708650589,
|
| 433 |
+
"learning_rate": 0.0009265005295986819,
|
| 434 |
+
"loss": 3.1887,
|
| 435 |
+
"step": 57000
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 3.12,
|
| 439 |
+
"grad_norm": 0.21209891140460968,
|
| 440 |
+
"learning_rate": 0.000923561256914205,
|
| 441 |
+
"loss": 3.1954,
|
| 442 |
+
"step": 58000
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 3.17,
|
| 446 |
+
"grad_norm": 0.2077139914035797,
|
| 447 |
+
"learning_rate": 0.0009206190420148288,
|
| 448 |
+
"loss": 3.191,
|
| 449 |
+
"step": 59000
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 3.23,
|
| 453 |
+
"grad_norm": 0.18837687373161316,
|
| 454 |
+
"learning_rate": 0.0009176797693303519,
|
| 455 |
+
"loss": 3.1978,
|
| 456 |
+
"step": 60000
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 3.28,
|
| 460 |
+
"grad_norm": 0.21532927453517914,
|
| 461 |
+
"learning_rate": 0.0009147375544309756,
|
| 462 |
+
"loss": 3.1977,
|
| 463 |
+
"step": 61000
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 3.33,
|
| 467 |
+
"grad_norm": 0.2525598704814911,
|
| 468 |
+
"learning_rate": 0.0009117982817464989,
|
| 469 |
+
"loss": 3.1913,
|
| 470 |
+
"step": 62000
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 3.39,
|
| 474 |
+
"grad_norm": 0.200853168964386,
|
| 475 |
+
"learning_rate": 0.0009088560668471226,
|
| 476 |
+
"loss": 3.1919,
|
| 477 |
+
"step": 63000
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 3.44,
|
| 481 |
+
"grad_norm": 0.229068785905838,
|
| 482 |
+
"learning_rate": 0.0009059138519477463,
|
| 483 |
+
"loss": 3.1943,
|
| 484 |
+
"step": 64000
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 3.5,
|
| 488 |
+
"grad_norm": 0.21732106804847717,
|
| 489 |
+
"learning_rate": 0.00090297163704837,
|
| 490 |
+
"loss": 3.1888,
|
| 491 |
+
"step": 65000
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 3.55,
|
| 495 |
+
"grad_norm": 0.22062081098556519,
|
| 496 |
+
"learning_rate": 0.0009000294221489937,
|
| 497 |
+
"loss": 3.1915,
|
| 498 |
+
"step": 66000
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 3.6,
|
| 502 |
+
"grad_norm": 0.1934981495141983,
|
| 503 |
+
"learning_rate": 0.0008970872072496176,
|
| 504 |
+
"loss": 3.1932,
|
| 505 |
+
"step": 67000
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 3.66,
|
| 509 |
+
"grad_norm": 0.2158234715461731,
|
| 510 |
+
"learning_rate": 0.0008941449923502413,
|
| 511 |
+
"loss": 3.1886,
|
| 512 |
+
"step": 68000
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 3.71,
|
| 516 |
+
"grad_norm": 0.20966586470603943,
|
| 517 |
+
"learning_rate": 0.0008912086618806638,
|
| 518 |
+
"loss": 3.1874,
|
| 519 |
+
"step": 69000
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 3.76,
|
| 523 |
+
"grad_norm": 0.22365672886371613,
|
| 524 |
+
"learning_rate": 0.0008882664469812876,
|
| 525 |
+
"loss": 3.1846,
|
| 526 |
+
"step": 70000
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 3.82,
|
| 530 |
+
"grad_norm": 0.21290117502212524,
|
| 531 |
+
"learning_rate": 0.0008853242320819112,
|
| 532 |
+
"loss": 3.1814,
|
| 533 |
+
"step": 71000
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 3.87,
|
| 537 |
+
"grad_norm": 0.24976381659507751,
|
| 538 |
+
"learning_rate": 0.0008823849593974345,
|
| 539 |
+
"loss": 3.1807,
|
| 540 |
+
"step": 72000
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 3.93,
|
| 544 |
+
"grad_norm": 0.22274042665958405,
|
| 545 |
+
"learning_rate": 0.0008794427444980582,
|
| 546 |
+
"loss": 3.1793,
|
| 547 |
+
"step": 73000
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 3.98,
|
| 551 |
+
"grad_norm": 0.2167842984199524,
|
| 552 |
+
"learning_rate": 0.0008765034718135813,
|
| 553 |
+
"loss": 3.1823,
|
| 554 |
+
"step": 74000
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 4.0,
|
| 558 |
+
"eval_accuracy": 0.39791865508136515,
|
| 559 |
+
"eval_loss": 3.415066957473755,
|
| 560 |
+
"eval_runtime": 153.3194,
|
| 561 |
+
"eval_samples_per_second": 377.767,
|
| 562 |
+
"eval_steps_per_second": 5.903,
|
| 563 |
+
"step": 74376
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 4.03,
|
| 567 |
+
"grad_norm": 0.21291407942771912,
|
| 568 |
+
"learning_rate": 0.0008735612569142051,
|
| 569 |
+
"loss": 3.1343,
|
| 570 |
+
"step": 75000
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 4.09,
|
| 574 |
+
"grad_norm": 0.20777927339076996,
|
| 575 |
+
"learning_rate": 0.0008706190420148288,
|
| 576 |
+
"loss": 3.1141,
|
| 577 |
+
"step": 76000
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 4.14,
|
| 581 |
+
"grad_norm": 0.26749035716056824,
|
| 582 |
+
"learning_rate": 0.0008676797693303519,
|
| 583 |
+
"loss": 3.1196,
|
| 584 |
+
"step": 77000
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 4.19,
|
| 588 |
+
"grad_norm": 0.23727400600910187,
|
| 589 |
+
"learning_rate": 0.000864740496645875,
|
| 590 |
+
"loss": 3.1233,
|
| 591 |
+
"step": 78000
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 4.25,
|
| 595 |
+
"grad_norm": 0.2067500203847885,
|
| 596 |
+
"learning_rate": 0.0008617982817464987,
|
| 597 |
+
"loss": 3.1214,
|
| 598 |
+
"step": 79000
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 4.3,
|
| 602 |
+
"grad_norm": 0.22707048058509827,
|
| 603 |
+
"learning_rate": 0.0008588560668471226,
|
| 604 |
+
"loss": 3.1273,
|
| 605 |
+
"step": 80000
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 4.36,
|
| 609 |
+
"grad_norm": 0.24729251861572266,
|
| 610 |
+
"learning_rate": 0.0008559138519477463,
|
| 611 |
+
"loss": 3.125,
|
| 612 |
+
"step": 81000
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 4.41,
|
| 616 |
+
"grad_norm": 0.22690938413143158,
|
| 617 |
+
"learning_rate": 0.00085297163704837,
|
| 618 |
+
"loss": 3.1231,
|
| 619 |
+
"step": 82000
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 4.46,
|
| 623 |
+
"grad_norm": 0.18175360560417175,
|
| 624 |
+
"learning_rate": 0.0008500323643638932,
|
| 625 |
+
"loss": 3.1261,
|
| 626 |
+
"step": 83000
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 4.52,
|
| 630 |
+
"grad_norm": 0.19980421662330627,
|
| 631 |
+
"learning_rate": 0.0008470901494645169,
|
| 632 |
+
"loss": 3.1266,
|
| 633 |
+
"step": 84000
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 4.57,
|
| 637 |
+
"grad_norm": 0.2160082459449768,
|
| 638 |
+
"learning_rate": 0.00084415087678004,
|
| 639 |
+
"loss": 3.1267,
|
| 640 |
+
"step": 85000
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 4.63,
|
| 644 |
+
"grad_norm": 0.2302197962999344,
|
| 645 |
+
"learning_rate": 0.0008412086618806639,
|
| 646 |
+
"loss": 3.1255,
|
| 647 |
+
"step": 86000
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 4.68,
|
| 651 |
+
"grad_norm": 0.2090998739004135,
|
| 652 |
+
"learning_rate": 0.0008382664469812875,
|
| 653 |
+
"loss": 3.1287,
|
| 654 |
+
"step": 87000
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 4.73,
|
| 658 |
+
"grad_norm": 0.23025181889533997,
|
| 659 |
+
"learning_rate": 0.0008353271742968107,
|
| 660 |
+
"loss": 3.1243,
|
| 661 |
+
"step": 88000
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 4.79,
|
| 665 |
+
"grad_norm": 0.18944258987903595,
|
| 666 |
+
"learning_rate": 0.0008323849593974345,
|
| 667 |
+
"loss": 3.1276,
|
| 668 |
+
"step": 89000
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 4.84,
|
| 672 |
+
"grad_norm": 0.23088830709457397,
|
| 673 |
+
"learning_rate": 0.0008294427444980582,
|
| 674 |
+
"loss": 3.119,
|
| 675 |
+
"step": 90000
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 4.89,
|
| 679 |
+
"grad_norm": 0.2036658078432083,
|
| 680 |
+
"learning_rate": 0.0008265034718135813,
|
| 681 |
+
"loss": 3.1229,
|
| 682 |
+
"step": 91000
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 4.95,
|
| 686 |
+
"grad_norm": 0.22064287960529327,
|
| 687 |
+
"learning_rate": 0.0008235612569142051,
|
| 688 |
+
"loss": 3.123,
|
| 689 |
+
"step": 92000
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 5.0,
|
| 693 |
+
"eval_accuracy": 0.40198451036846505,
|
| 694 |
+
"eval_loss": 3.3894715309143066,
|
| 695 |
+
"eval_runtime": 153.8615,
|
| 696 |
+
"eval_samples_per_second": 376.436,
|
| 697 |
+
"eval_steps_per_second": 5.882,
|
| 698 |
+
"step": 92970
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 5.0,
|
| 702 |
+
"grad_norm": 0.23256349563598633,
|
| 703 |
+
"learning_rate": 0.0008206190420148288,
|
| 704 |
+
"loss": 3.1192,
|
| 705 |
+
"step": 93000
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 5.06,
|
| 709 |
+
"grad_norm": 0.22044919431209564,
|
| 710 |
+
"learning_rate": 0.0008176768271154526,
|
| 711 |
+
"loss": 3.0559,
|
| 712 |
+
"step": 94000
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 5.11,
|
| 716 |
+
"grad_norm": 0.23432625830173492,
|
| 717 |
+
"learning_rate": 0.0008147346122160762,
|
| 718 |
+
"loss": 3.0609,
|
| 719 |
+
"step": 95000
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 5.16,
|
| 723 |
+
"grad_norm": 0.20878207683563232,
|
| 724 |
+
"learning_rate": 0.0008117953395315995,
|
| 725 |
+
"loss": 3.0594,
|
| 726 |
+
"step": 96000
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 5.22,
|
| 730 |
+
"grad_norm": 0.22522564232349396,
|
| 731 |
+
"learning_rate": 0.0008088531246322232,
|
| 732 |
+
"loss": 3.0697,
|
| 733 |
+
"step": 97000
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 5.27,
|
| 737 |
+
"grad_norm": 0.21230152249336243,
|
| 738 |
+
"learning_rate": 0.0008059167941626456,
|
| 739 |
+
"loss": 3.0719,
|
| 740 |
+
"step": 98000
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 5.32,
|
| 744 |
+
"grad_norm": 0.21536217629909515,
|
| 745 |
+
"learning_rate": 0.0008029775214781688,
|
| 746 |
+
"loss": 3.0704,
|
| 747 |
+
"step": 99000
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 5.38,
|
| 751 |
+
"grad_norm": 0.2108333706855774,
|
| 752 |
+
"learning_rate": 0.0008000353065787926,
|
| 753 |
+
"loss": 3.0744,
|
| 754 |
+
"step": 100000
|
| 755 |
+
},
|
| 756 |
+
{
|
| 757 |
+
"epoch": 5.43,
|
| 758 |
+
"grad_norm": 0.2334020584821701,
|
| 759 |
+
"learning_rate": 0.0007970960338943156,
|
| 760 |
+
"loss": 3.0798,
|
| 761 |
+
"step": 101000
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 5.49,
|
| 765 |
+
"grad_norm": 0.2081642746925354,
|
| 766 |
+
"learning_rate": 0.0007941538189949394,
|
| 767 |
+
"loss": 3.0768,
|
| 768 |
+
"step": 102000
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 5.54,
|
| 772 |
+
"grad_norm": 0.23282569646835327,
|
| 773 |
+
"learning_rate": 0.0007912116040955632,
|
| 774 |
+
"loss": 3.0777,
|
| 775 |
+
"step": 103000
|
| 776 |
+
},
|
| 777 |
+
{
|
| 778 |
+
"epoch": 5.59,
|
| 779 |
+
"grad_norm": 0.23747290670871735,
|
| 780 |
+
"learning_rate": 0.000788269389196187,
|
| 781 |
+
"loss": 3.0809,
|
| 782 |
+
"step": 104000
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 5.65,
|
| 786 |
+
"grad_norm": 0.24338039755821228,
|
| 787 |
+
"learning_rate": 0.0007853271742968107,
|
| 788 |
+
"loss": 3.0818,
|
| 789 |
+
"step": 105000
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 5.7,
|
| 793 |
+
"grad_norm": 0.2295004278421402,
|
| 794 |
+
"learning_rate": 0.0007823879016123338,
|
| 795 |
+
"loss": 3.0823,
|
| 796 |
+
"step": 106000
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"epoch": 5.75,
|
| 800 |
+
"grad_norm": 0.21398743987083435,
|
| 801 |
+
"learning_rate": 0.0007794456867129576,
|
| 802 |
+
"loss": 3.0817,
|
| 803 |
+
"step": 107000
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 5.81,
|
| 807 |
+
"grad_norm": 0.20917442440986633,
|
| 808 |
+
"learning_rate": 0.0007765034718135813,
|
| 809 |
+
"loss": 3.0757,
|
| 810 |
+
"step": 108000
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 5.86,
|
| 814 |
+
"grad_norm": 0.21766169369220734,
|
| 815 |
+
"learning_rate": 0.000773561256914205,
|
| 816 |
+
"loss": 3.0773,
|
| 817 |
+
"step": 109000
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 5.92,
|
| 821 |
+
"grad_norm": 0.2269996553659439,
|
| 822 |
+
"learning_rate": 0.0007706219842297282,
|
| 823 |
+
"loss": 3.0747,
|
| 824 |
+
"step": 110000
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 5.97,
|
| 828 |
+
"grad_norm": 0.21728727221488953,
|
| 829 |
+
"learning_rate": 0.0007676827115452512,
|
| 830 |
+
"loss": 3.0779,
|
| 831 |
+
"step": 111000
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 6.0,
|
| 835 |
+
"eval_accuracy": 0.40453321937661096,
|
| 836 |
+
"eval_loss": 3.3727822303771973,
|
| 837 |
+
"eval_runtime": 153.703,
|
| 838 |
+
"eval_samples_per_second": 376.824,
|
| 839 |
+
"eval_steps_per_second": 5.888,
|
| 840 |
+
"step": 111564
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 6.02,
|
| 844 |
+
"grad_norm": 0.22364608943462372,
|
| 845 |
+
"learning_rate": 0.0007647434388607743,
|
| 846 |
+
"loss": 3.0463,
|
| 847 |
+
"step": 112000
|
| 848 |
+
},
|
| 849 |
+
{
|
| 850 |
+
"epoch": 6.08,
|
| 851 |
+
"grad_norm": 0.2303326576948166,
|
| 852 |
+
"learning_rate": 0.0007618012239613981,
|
| 853 |
+
"loss": 3.013,
|
| 854 |
+
"step": 113000
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 6.13,
|
| 858 |
+
"grad_norm": 0.22477248311042786,
|
| 859 |
+
"learning_rate": 0.0007588590090620218,
|
| 860 |
+
"loss": 3.0186,
|
| 861 |
+
"step": 114000
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 6.18,
|
| 865 |
+
"grad_norm": 0.2574509084224701,
|
| 866 |
+
"learning_rate": 0.0007559167941626457,
|
| 867 |
+
"loss": 3.0288,
|
| 868 |
+
"step": 115000
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 6.24,
|
| 872 |
+
"grad_norm": 0.2393523007631302,
|
| 873 |
+
"learning_rate": 0.0007529775214781687,
|
| 874 |
+
"loss": 3.0246,
|
| 875 |
+
"step": 116000
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 6.29,
|
| 879 |
+
"grad_norm": 0.2154286503791809,
|
| 880 |
+
"learning_rate": 0.0007500353065787924,
|
| 881 |
+
"loss": 3.032,
|
| 882 |
+
"step": 117000
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 6.35,
|
| 886 |
+
"grad_norm": 0.24074001610279083,
|
| 887 |
+
"learning_rate": 0.0007470960338943157,
|
| 888 |
+
"loss": 3.03,
|
| 889 |
+
"step": 118000
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 6.4,
|
| 893 |
+
"grad_norm": 0.21355246007442474,
|
| 894 |
+
"learning_rate": 0.0007441538189949394,
|
| 895 |
+
"loss": 3.0339,
|
| 896 |
+
"step": 119000
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 6.45,
|
| 900 |
+
"grad_norm": 0.23035864531993866,
|
| 901 |
+
"learning_rate": 0.0007412116040955633,
|
| 902 |
+
"loss": 3.0346,
|
| 903 |
+
"step": 120000
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 6.51,
|
| 907 |
+
"grad_norm": 0.22556589543819427,
|
| 908 |
+
"learning_rate": 0.0007382693891961869,
|
| 909 |
+
"loss": 3.0383,
|
| 910 |
+
"step": 121000
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 6.56,
|
| 914 |
+
"grad_norm": 0.2367316633462906,
|
| 915 |
+
"learning_rate": 0.00073533011651171,
|
| 916 |
+
"loss": 3.0389,
|
| 917 |
+
"step": 122000
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 6.62,
|
| 921 |
+
"grad_norm": 0.23370467126369476,
|
| 922 |
+
"learning_rate": 0.0007323879016123339,
|
| 923 |
+
"loss": 3.0418,
|
| 924 |
+
"step": 123000
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 6.67,
|
| 928 |
+
"grad_norm": 0.24725669622421265,
|
| 929 |
+
"learning_rate": 0.0007294486289278569,
|
| 930 |
+
"loss": 3.043,
|
| 931 |
+
"step": 124000
|
| 932 |
+
},
|
| 933 |
+
{
|
| 934 |
+
"epoch": 6.72,
|
| 935 |
+
"grad_norm": 0.2352810651063919,
|
| 936 |
+
"learning_rate": 0.0007265064140284806,
|
| 937 |
+
"loss": 3.0412,
|
| 938 |
+
"step": 125000
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 6.78,
|
| 942 |
+
"grad_norm": 0.25413230061531067,
|
| 943 |
+
"learning_rate": 0.0007235671413440038,
|
| 944 |
+
"loss": 3.0405,
|
| 945 |
+
"step": 126000
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 6.83,
|
| 949 |
+
"grad_norm": 0.25267255306243896,
|
| 950 |
+
"learning_rate": 0.0007206249264446275,
|
| 951 |
+
"loss": 3.0445,
|
| 952 |
+
"step": 127000
|
| 953 |
+
},
|
| 954 |
+
{
|
| 955 |
+
"epoch": 6.88,
|
| 956 |
+
"grad_norm": 0.22229382395744324,
|
| 957 |
+
"learning_rate": 0.0007176827115452513,
|
| 958 |
+
"loss": 3.0457,
|
| 959 |
+
"step": 128000
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 6.94,
|
| 963 |
+
"grad_norm": 0.23900896310806274,
|
| 964 |
+
"learning_rate": 0.0007147434388607744,
|
| 965 |
+
"loss": 3.0438,
|
| 966 |
+
"step": 129000
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 6.99,
|
| 970 |
+
"grad_norm": 0.26492878794670105,
|
| 971 |
+
"learning_rate": 0.0007118012239613981,
|
| 972 |
+
"loss": 3.0394,
|
| 973 |
+
"step": 130000
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 7.0,
|
| 977 |
+
"eval_accuracy": 0.4067593928808265,
|
| 978 |
+
"eval_loss": 3.3513076305389404,
|
| 979 |
+
"eval_runtime": 154.1811,
|
| 980 |
+
"eval_samples_per_second": 375.656,
|
| 981 |
+
"eval_steps_per_second": 5.87,
|
| 982 |
+
"step": 130158
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 7.05,
|
| 986 |
+
"grad_norm": 0.22051194310188293,
|
| 987 |
+
"learning_rate": 0.0007088590090620219,
|
| 988 |
+
"loss": 2.9883,
|
| 989 |
+
"step": 131000
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 7.1,
|
| 993 |
+
"grad_norm": 0.2120658904314041,
|
| 994 |
+
"learning_rate": 0.000705919736377545,
|
| 995 |
+
"loss": 2.9817,
|
| 996 |
+
"step": 132000
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 7.15,
|
| 1000 |
+
"grad_norm": 0.21563994884490967,
|
| 1001 |
+
"learning_rate": 0.0007029775214781687,
|
| 1002 |
+
"loss": 2.9867,
|
| 1003 |
+
"step": 133000
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 7.21,
|
| 1007 |
+
"grad_norm": 0.228189617395401,
|
| 1008 |
+
"learning_rate": 0.0007000353065787925,
|
| 1009 |
+
"loss": 2.9915,
|
| 1010 |
+
"step": 134000
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 7.26,
|
| 1014 |
+
"grad_norm": 0.27084964513778687,
|
| 1015 |
+
"learning_rate": 0.0006970960338943157,
|
| 1016 |
+
"loss": 2.9942,
|
| 1017 |
+
"step": 135000
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 7.31,
|
| 1021 |
+
"grad_norm": 0.21773342788219452,
|
| 1022 |
+
"learning_rate": 0.0006941538189949394,
|
| 1023 |
+
"loss": 2.994,
|
| 1024 |
+
"step": 136000
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 7.37,
|
| 1028 |
+
"grad_norm": 0.22931870818138123,
|
| 1029 |
+
"learning_rate": 0.0006912116040955631,
|
| 1030 |
+
"loss": 2.9985,
|
| 1031 |
+
"step": 137000
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 7.42,
|
| 1035 |
+
"grad_norm": 0.24787016212940216,
|
| 1036 |
+
"learning_rate": 0.0006882693891961869,
|
| 1037 |
+
"loss": 3.0026,
|
| 1038 |
+
"step": 138000
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 7.48,
|
| 1042 |
+
"grad_norm": 0.2541216313838959,
|
| 1043 |
+
"learning_rate": 0.0006853301165117101,
|
| 1044 |
+
"loss": 3.0017,
|
| 1045 |
+
"step": 139000
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 7.53,
|
| 1049 |
+
"grad_norm": 0.2567012906074524,
|
| 1050 |
+
"learning_rate": 0.0006823908438272332,
|
| 1051 |
+
"loss": 3.0041,
|
| 1052 |
+
"step": 140000
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 7.58,
|
| 1056 |
+
"grad_norm": 0.24575956165790558,
|
| 1057 |
+
"learning_rate": 0.0006794486289278569,
|
| 1058 |
+
"loss": 3.0037,
|
| 1059 |
+
"step": 141000
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 7.64,
|
| 1063 |
+
"grad_norm": 0.2628481984138489,
|
| 1064 |
+
"learning_rate": 0.00067650935624338,
|
| 1065 |
+
"loss": 3.0091,
|
| 1066 |
+
"step": 142000
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 7.69,
|
| 1070 |
+
"grad_norm": 0.26017704606056213,
|
| 1071 |
+
"learning_rate": 0.0006735671413440038,
|
| 1072 |
+
"loss": 3.0077,
|
| 1073 |
+
"step": 143000
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 7.74,
|
| 1077 |
+
"grad_norm": 0.2303098738193512,
|
| 1078 |
+
"learning_rate": 0.0006706249264446276,
|
| 1079 |
+
"loss": 3.0087,
|
| 1080 |
+
"step": 144000
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 7.8,
|
| 1084 |
+
"grad_norm": 0.25814709067344666,
|
| 1085 |
+
"learning_rate": 0.0006676856537601506,
|
| 1086 |
+
"loss": 3.0141,
|
| 1087 |
+
"step": 145000
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 7.85,
|
| 1091 |
+
"grad_norm": 0.2519998848438263,
|
| 1092 |
+
"learning_rate": 0.0006647434388607744,
|
| 1093 |
+
"loss": 3.0091,
|
| 1094 |
+
"step": 146000
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 7.91,
|
| 1098 |
+
"grad_norm": 0.23445440828800201,
|
| 1099 |
+
"learning_rate": 0.0006618012239613982,
|
| 1100 |
+
"loss": 3.0127,
|
| 1101 |
+
"step": 147000
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 7.96,
|
| 1105 |
+
"grad_norm": 0.22559377551078796,
|
| 1106 |
+
"learning_rate": 0.0006588590090620219,
|
| 1107 |
+
"loss": 3.0098,
|
| 1108 |
+
"step": 148000
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 8.0,
|
| 1112 |
+
"eval_accuracy": 0.40905872638582946,
|
| 1113 |
+
"eval_loss": 3.33290433883667,
|
| 1114 |
+
"eval_runtime": 153.3565,
|
| 1115 |
+
"eval_samples_per_second": 377.676,
|
| 1116 |
+
"eval_steps_per_second": 5.901,
|
| 1117 |
+
"step": 148752
|
| 1118 |
+
},
|
| 1119 |
+
{
|
| 1120 |
+
"epoch": 8.01,
|
| 1121 |
+
"grad_norm": 0.2889389097690582,
|
| 1122 |
+
"learning_rate": 0.000655919736377545,
|
| 1123 |
+
"loss": 2.9968,
|
| 1124 |
+
"step": 149000
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 8.07,
|
| 1128 |
+
"grad_norm": 0.24556472897529602,
|
| 1129 |
+
"learning_rate": 0.0006529804636930682,
|
| 1130 |
+
"loss": 2.9443,
|
| 1131 |
+
"step": 150000
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 8.12,
|
| 1135 |
+
"grad_norm": 0.23049430549144745,
|
| 1136 |
+
"learning_rate": 0.0006500382487936918,
|
| 1137 |
+
"loss": 2.9508,
|
| 1138 |
+
"step": 151000
|
| 1139 |
+
},
|
| 1140 |
+
{
|
| 1141 |
+
"epoch": 8.17,
|
| 1142 |
+
"grad_norm": 0.23687225580215454,
|
| 1143 |
+
"learning_rate": 0.0006470960338943157,
|
| 1144 |
+
"loss": 2.9559,
|
| 1145 |
+
"step": 152000
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 8.23,
|
| 1149 |
+
"grad_norm": 0.24977627396583557,
|
| 1150 |
+
"learning_rate": 0.0006441538189949394,
|
| 1151 |
+
"loss": 2.9594,
|
| 1152 |
+
"step": 153000
|
| 1153 |
+
},
|
| 1154 |
+
{
|
| 1155 |
+
"epoch": 8.28,
|
| 1156 |
+
"grad_norm": 0.2544060945510864,
|
| 1157 |
+
"learning_rate": 0.0006412145463104624,
|
| 1158 |
+
"loss": 2.9649,
|
| 1159 |
+
"step": 154000
|
| 1160 |
+
},
|
| 1161 |
+
{
|
| 1162 |
+
"epoch": 8.34,
|
| 1163 |
+
"grad_norm": 0.2694548964500427,
|
| 1164 |
+
"learning_rate": 0.0006382723314110863,
|
| 1165 |
+
"loss": 2.9691,
|
| 1166 |
+
"step": 155000
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 8.39,
|
| 1170 |
+
"grad_norm": 0.24550598859786987,
|
| 1171 |
+
"learning_rate": 0.00063533011651171,
|
| 1172 |
+
"loss": 2.9692,
|
| 1173 |
+
"step": 156000
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 8.44,
|
| 1177 |
+
"grad_norm": 0.32097670435905457,
|
| 1178 |
+
"learning_rate": 0.0006323879016123337,
|
| 1179 |
+
"loss": 2.9721,
|
| 1180 |
+
"step": 157000
|
| 1181 |
+
},
|
| 1182 |
+
{
|
| 1183 |
+
"epoch": 8.5,
|
| 1184 |
+
"grad_norm": 0.2392505705356598,
|
| 1185 |
+
"learning_rate": 0.000629448628927857,
|
| 1186 |
+
"loss": 2.976,
|
| 1187 |
+
"step": 158000
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 8.55,
|
| 1191 |
+
"grad_norm": 0.24632832407951355,
|
| 1192 |
+
"learning_rate": 0.0006265064140284806,
|
| 1193 |
+
"loss": 2.9734,
|
| 1194 |
+
"step": 159000
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 8.6,
|
| 1198 |
+
"grad_norm": 0.2416047602891922,
|
| 1199 |
+
"learning_rate": 0.0006235671413440038,
|
| 1200 |
+
"loss": 2.9778,
|
| 1201 |
+
"step": 160000
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 8.66,
|
| 1205 |
+
"grad_norm": 0.2523684799671173,
|
| 1206 |
+
"learning_rate": 0.0006206278686595269,
|
| 1207 |
+
"loss": 2.9851,
|
| 1208 |
+
"step": 161000
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 8.71,
|
| 1212 |
+
"grad_norm": 0.23185420036315918,
|
| 1213 |
+
"learning_rate": 0.0006176856537601506,
|
| 1214 |
+
"loss": 2.9812,
|
| 1215 |
+
"step": 162000
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 8.77,
|
| 1219 |
+
"grad_norm": 0.24621973931789398,
|
| 1220 |
+
"learning_rate": 0.0006147434388607744,
|
| 1221 |
+
"loss": 2.986,
|
| 1222 |
+
"step": 163000
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"epoch": 8.82,
|
| 1226 |
+
"grad_norm": 0.24415548145771027,
|
| 1227 |
+
"learning_rate": 0.0006118012239613982,
|
| 1228 |
+
"loss": 2.9836,
|
| 1229 |
+
"step": 164000
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 8.87,
|
| 1233 |
+
"grad_norm": 0.2517019808292389,
|
| 1234 |
+
"learning_rate": 0.0006088590090620219,
|
| 1235 |
+
"loss": 2.9833,
|
| 1236 |
+
"step": 165000
|
| 1237 |
+
},
|
| 1238 |
+
{
|
| 1239 |
+
"epoch": 8.93,
|
| 1240 |
+
"grad_norm": 0.23550397157669067,
|
| 1241 |
+
"learning_rate": 0.0006059167941626457,
|
| 1242 |
+
"loss": 2.9857,
|
| 1243 |
+
"step": 166000
|
| 1244 |
+
},
|
| 1245 |
+
{
|
| 1246 |
+
"epoch": 8.98,
|
| 1247 |
+
"grad_norm": 0.26258397102355957,
|
| 1248 |
+
"learning_rate": 0.0006029775214781688,
|
| 1249 |
+
"loss": 2.9848,
|
| 1250 |
+
"step": 167000
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 9.0,
|
| 1254 |
+
"eval_accuracy": 0.40959254582224486,
|
| 1255 |
+
"eval_loss": 3.3542630672454834,
|
| 1256 |
+
"eval_runtime": 153.6915,
|
| 1257 |
+
"eval_samples_per_second": 376.852,
|
| 1258 |
+
"eval_steps_per_second": 5.888,
|
| 1259 |
+
"step": 167346
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 9.04,
|
| 1263 |
+
"grad_norm": 0.252084881067276,
|
| 1264 |
+
"learning_rate": 0.0006000353065787926,
|
| 1265 |
+
"loss": 2.9379,
|
| 1266 |
+
"step": 168000
|
| 1267 |
+
},
|
| 1268 |
+
{
|
| 1269 |
+
"epoch": 9.09,
|
| 1270 |
+
"grad_norm": 0.24115502834320068,
|
| 1271 |
+
"learning_rate": 0.0005970960338943156,
|
| 1272 |
+
"loss": 2.9231,
|
| 1273 |
+
"step": 169000
|
| 1274 |
+
},
|
| 1275 |
+
{
|
| 1276 |
+
"epoch": 9.14,
|
| 1277 |
+
"grad_norm": 0.2534337043762207,
|
| 1278 |
+
"learning_rate": 0.0005941538189949394,
|
| 1279 |
+
"loss": 2.9279,
|
| 1280 |
+
"step": 170000
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 9.2,
|
| 1284 |
+
"grad_norm": 0.2916615307331085,
|
| 1285 |
+
"learning_rate": 0.0005912116040955632,
|
| 1286 |
+
"loss": 2.9315,
|
| 1287 |
+
"step": 171000
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 9.25,
|
| 1291 |
+
"grad_norm": 0.27919089794158936,
|
| 1292 |
+
"learning_rate": 0.0005882693891961869,
|
| 1293 |
+
"loss": 2.9369,
|
| 1294 |
+
"step": 172000
|
| 1295 |
+
},
|
| 1296 |
+
{
|
| 1297 |
+
"epoch": 9.3,
|
| 1298 |
+
"grad_norm": 0.2634003460407257,
|
| 1299 |
+
"learning_rate": 0.00058533011651171,
|
| 1300 |
+
"loss": 2.9403,
|
| 1301 |
+
"step": 173000
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 9.36,
|
| 1305 |
+
"grad_norm": 0.24581490457057953,
|
| 1306 |
+
"learning_rate": 0.0005823879016123338,
|
| 1307 |
+
"loss": 2.9463,
|
| 1308 |
+
"step": 174000
|
| 1309 |
+
},
|
| 1310 |
+
{
|
| 1311 |
+
"epoch": 9.41,
|
| 1312 |
+
"grad_norm": 0.24444222450256348,
|
| 1313 |
+
"learning_rate": 0.0005794486289278568,
|
| 1314 |
+
"loss": 2.9446,
|
| 1315 |
+
"step": 175000
|
| 1316 |
+
},
|
| 1317 |
+
{
|
| 1318 |
+
"epoch": 9.47,
|
| 1319 |
+
"grad_norm": 0.24242326617240906,
|
| 1320 |
+
"learning_rate": 0.0005765093562433801,
|
| 1321 |
+
"loss": 2.9463,
|
| 1322 |
+
"step": 176000
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 9.52,
|
| 1326 |
+
"grad_norm": 0.2594030499458313,
|
| 1327 |
+
"learning_rate": 0.0005735671413440038,
|
| 1328 |
+
"loss": 2.9493,
|
| 1329 |
+
"step": 177000
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 9.57,
|
| 1333 |
+
"grad_norm": 0.24433520436286926,
|
| 1334 |
+
"learning_rate": 0.0005706249264446274,
|
| 1335 |
+
"loss": 2.9532,
|
| 1336 |
+
"step": 178000
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"epoch": 9.63,
|
| 1340 |
+
"grad_norm": 0.27443036437034607,
|
| 1341 |
+
"learning_rate": 0.0005676856537601507,
|
| 1342 |
+
"loss": 2.9507,
|
| 1343 |
+
"step": 179000
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 9.68,
|
| 1347 |
+
"grad_norm": 0.2567445933818817,
|
| 1348 |
+
"learning_rate": 0.0005647434388607744,
|
| 1349 |
+
"loss": 2.9515,
|
| 1350 |
+
"step": 180000
|
| 1351 |
+
},
|
| 1352 |
+
{
|
| 1353 |
+
"epoch": 9.73,
|
| 1354 |
+
"grad_norm": 0.24765917658805847,
|
| 1355 |
+
"learning_rate": 0.000561801223961398,
|
| 1356 |
+
"loss": 2.9584,
|
| 1357 |
+
"step": 181000
|
| 1358 |
+
},
|
| 1359 |
+
{
|
| 1360 |
+
"epoch": 9.79,
|
| 1361 |
+
"grad_norm": 0.288038969039917,
|
| 1362 |
+
"learning_rate": 0.000558859009062022,
|
| 1363 |
+
"loss": 2.9558,
|
| 1364 |
+
"step": 182000
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 9.84,
|
| 1368 |
+
"grad_norm": 0.24314820766448975,
|
| 1369 |
+
"learning_rate": 0.000555919736377545,
|
| 1370 |
+
"loss": 2.9545,
|
| 1371 |
+
"step": 183000
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 9.9,
|
| 1375 |
+
"grad_norm": 0.23880471289157867,
|
| 1376 |
+
"learning_rate": 0.0005529775214781688,
|
| 1377 |
+
"loss": 2.9559,
|
| 1378 |
+
"step": 184000
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 9.95,
|
| 1382 |
+
"grad_norm": 0.24601146578788757,
|
| 1383 |
+
"learning_rate": 0.0005500382487936919,
|
| 1384 |
+
"loss": 2.9607,
|
| 1385 |
+
"step": 185000
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 10.0,
|
| 1389 |
+
"eval_accuracy": 0.4101945538728009,
|
| 1390 |
+
"eval_loss": 3.3317415714263916,
|
| 1391 |
+
"eval_runtime": 153.5202,
|
| 1392 |
+
"eval_samples_per_second": 377.273,
|
| 1393 |
+
"eval_steps_per_second": 5.895,
|
| 1394 |
+
"step": 185940
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 10.0,
|
| 1398 |
+
"grad_norm": 0.27554717659950256,
|
| 1399 |
+
"learning_rate": 0.0005470960338943156,
|
| 1400 |
+
"loss": 2.9568,
|
| 1401 |
+
"step": 186000
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 10.06,
|
| 1405 |
+
"grad_norm": 0.2996651530265808,
|
| 1406 |
+
"learning_rate": 0.0005441538189949394,
|
| 1407 |
+
"loss": 2.8922,
|
| 1408 |
+
"step": 187000
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"epoch": 10.11,
|
| 1412 |
+
"grad_norm": 0.26455074548721313,
|
| 1413 |
+
"learning_rate": 0.0005412145463104625,
|
| 1414 |
+
"loss": 2.9033,
|
| 1415 |
+
"step": 188000
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 10.16,
|
| 1419 |
+
"grad_norm": 0.2427845299243927,
|
| 1420 |
+
"learning_rate": 0.0005382723314110862,
|
| 1421 |
+
"loss": 2.9029,
|
| 1422 |
+
"step": 189000
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 10.22,
|
| 1426 |
+
"grad_norm": 0.27417176961898804,
|
| 1427 |
+
"learning_rate": 0.00053533011651171,
|
| 1428 |
+
"loss": 2.9121,
|
| 1429 |
+
"step": 190000
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 10.27,
|
| 1433 |
+
"grad_norm": 0.2733907997608185,
|
| 1434 |
+
"learning_rate": 0.0005323908438272331,
|
| 1435 |
+
"loss": 2.9135,
|
| 1436 |
+
"step": 191000
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 10.33,
|
| 1440 |
+
"grad_norm": 0.23740702867507935,
|
| 1441 |
+
"learning_rate": 0.0005294486289278569,
|
| 1442 |
+
"loss": 2.9194,
|
| 1443 |
+
"step": 192000
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 10.38,
|
| 1447 |
+
"grad_norm": 0.2555631101131439,
|
| 1448 |
+
"learning_rate": 0.0005265064140284806,
|
| 1449 |
+
"loss": 2.9217,
|
| 1450 |
+
"step": 193000
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 10.43,
|
| 1454 |
+
"grad_norm": 0.2792896628379822,
|
| 1455 |
+
"learning_rate": 0.0005235641991291044,
|
| 1456 |
+
"loss": 2.9207,
|
| 1457 |
+
"step": 194000
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 10.49,
|
| 1461 |
+
"grad_norm": 0.258811891078949,
|
| 1462 |
+
"learning_rate": 0.0005206219842297282,
|
| 1463 |
+
"loss": 2.9201,
|
| 1464 |
+
"step": 195000
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 10.54,
|
| 1468 |
+
"grad_norm": 0.24523362517356873,
|
| 1469 |
+
"learning_rate": 0.0005176827115452512,
|
| 1470 |
+
"loss": 2.9262,
|
| 1471 |
+
"step": 196000
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 10.59,
|
| 1475 |
+
"grad_norm": 0.2545565664768219,
|
| 1476 |
+
"learning_rate": 0.0005147434388607743,
|
| 1477 |
+
"loss": 2.9292,
|
| 1478 |
+
"step": 197000
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 10.65,
|
| 1482 |
+
"grad_norm": 0.26881828904151917,
|
| 1483 |
+
"learning_rate": 0.0005118012239613981,
|
| 1484 |
+
"loss": 2.9308,
|
| 1485 |
+
"step": 198000
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 10.7,
|
| 1489 |
+
"grad_norm": 0.2478063404560089,
|
| 1490 |
+
"learning_rate": 0.0005088590090620218,
|
| 1491 |
+
"loss": 2.9275,
|
| 1492 |
+
"step": 199000
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"epoch": 10.76,
|
| 1496 |
+
"grad_norm": 0.2736205756664276,
|
| 1497 |
+
"learning_rate": 0.0005059197363775451,
|
| 1498 |
+
"loss": 2.9303,
|
| 1499 |
+
"step": 200000
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 10.81,
|
| 1503 |
+
"grad_norm": 0.2818506360054016,
|
| 1504 |
+
"learning_rate": 0.0005029804636930682,
|
| 1505 |
+
"loss": 2.9356,
|
| 1506 |
+
"step": 201000
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 10.86,
|
| 1510 |
+
"grad_norm": 0.27006179094314575,
|
| 1511 |
+
"learning_rate": 0.0005000382487936919,
|
| 1512 |
+
"loss": 2.934,
|
| 1513 |
+
"step": 202000
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 10.92,
|
| 1517 |
+
"grad_norm": 0.2664320766925812,
|
| 1518 |
+
"learning_rate": 0.0004970960338943157,
|
| 1519 |
+
"loss": 2.933,
|
| 1520 |
+
"step": 203000
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 10.97,
|
| 1524 |
+
"grad_norm": 0.25513389706611633,
|
| 1525 |
+
"learning_rate": 0.0004941538189949394,
|
| 1526 |
+
"loss": 2.9381,
|
| 1527 |
+
"step": 204000
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 11.0,
|
| 1531 |
+
"eval_accuracy": 0.40957044330593717,
|
| 1532 |
+
"eval_loss": 3.350923538208008,
|
| 1533 |
+
"eval_runtime": 153.3706,
|
| 1534 |
+
"eval_samples_per_second": 377.641,
|
| 1535 |
+
"eval_steps_per_second": 5.901,
|
| 1536 |
+
"step": 204534
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 11.03,
|
| 1540 |
+
"grad_norm": 0.26754114031791687,
|
| 1541 |
+
"learning_rate": 0.0004912145463104625,
|
| 1542 |
+
"loss": 2.9064,
|
| 1543 |
+
"step": 205000
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 11.08,
|
| 1547 |
+
"grad_norm": 0.24352851510047913,
|
| 1548 |
+
"learning_rate": 0.0004882752736259857,
|
| 1549 |
+
"loss": 2.8732,
|
| 1550 |
+
"step": 206000
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 11.13,
|
| 1554 |
+
"grad_norm": 0.28722500801086426,
|
| 1555 |
+
"learning_rate": 0.00048533305872660936,
|
| 1556 |
+
"loss": 2.8768,
|
| 1557 |
+
"step": 207000
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 11.19,
|
| 1561 |
+
"grad_norm": 0.2741190195083618,
|
| 1562 |
+
"learning_rate": 0.00048239084382723315,
|
| 1563 |
+
"loss": 2.89,
|
| 1564 |
+
"step": 208000
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 11.24,
|
| 1568 |
+
"grad_norm": 0.24753689765930176,
|
| 1569 |
+
"learning_rate": 0.0004794486289278569,
|
| 1570 |
+
"loss": 2.8889,
|
| 1571 |
+
"step": 209000
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 11.29,
|
| 1575 |
+
"grad_norm": 0.2748933732509613,
|
| 1576 |
+
"learning_rate": 0.0004765064140284807,
|
| 1577 |
+
"loss": 2.8924,
|
| 1578 |
+
"step": 210000
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 11.35,
|
| 1582 |
+
"grad_norm": 0.2736218273639679,
|
| 1583 |
+
"learning_rate": 0.00047356714134400376,
|
| 1584 |
+
"loss": 2.8966,
|
| 1585 |
+
"step": 211000
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 11.4,
|
| 1589 |
+
"grad_norm": 0.2917451858520508,
|
| 1590 |
+
"learning_rate": 0.0004706278686595269,
|
| 1591 |
+
"loss": 2.9008,
|
| 1592 |
+
"step": 212000
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 11.46,
|
| 1596 |
+
"grad_norm": 0.2701798677444458,
|
| 1597 |
+
"learning_rate": 0.00046768565376015063,
|
| 1598 |
+
"loss": 2.9027,
|
| 1599 |
+
"step": 213000
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 11.51,
|
| 1603 |
+
"grad_norm": 0.2652759253978729,
|
| 1604 |
+
"learning_rate": 0.00046474343886077443,
|
| 1605 |
+
"loss": 2.9006,
|
| 1606 |
+
"step": 214000
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 11.56,
|
| 1610 |
+
"grad_norm": 0.26413607597351074,
|
| 1611 |
+
"learning_rate": 0.00046180122396139817,
|
| 1612 |
+
"loss": 2.9066,
|
| 1613 |
+
"step": 215000
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 11.62,
|
| 1617 |
+
"grad_norm": 0.2796526253223419,
|
| 1618 |
+
"learning_rate": 0.0004588590090620219,
|
| 1619 |
+
"loss": 2.9042,
|
| 1620 |
+
"step": 216000
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 11.67,
|
| 1624 |
+
"grad_norm": 0.3159005343914032,
|
| 1625 |
+
"learning_rate": 0.00045591679416264564,
|
| 1626 |
+
"loss": 2.9055,
|
| 1627 |
+
"step": 217000
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 11.72,
|
| 1631 |
+
"grad_norm": 0.25383687019348145,
|
| 1632 |
+
"learning_rate": 0.0004529775214781688,
|
| 1633 |
+
"loss": 2.9056,
|
| 1634 |
+
"step": 218000
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 11.78,
|
| 1638 |
+
"grad_norm": 0.2639722228050232,
|
| 1639 |
+
"learning_rate": 0.00045003530657879257,
|
| 1640 |
+
"loss": 2.9099,
|
| 1641 |
+
"step": 219000
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 11.83,
|
| 1645 |
+
"grad_norm": 0.26410508155822754,
|
| 1646 |
+
"learning_rate": 0.00044709603389431565,
|
| 1647 |
+
"loss": 2.9112,
|
| 1648 |
+
"step": 220000
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 11.89,
|
| 1652 |
+
"grad_norm": 0.24445202946662903,
|
| 1653 |
+
"learning_rate": 0.0004441567612098388,
|
| 1654 |
+
"loss": 2.9146,
|
| 1655 |
+
"step": 221000
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 11.94,
|
| 1659 |
+
"grad_norm": 0.28658202290534973,
|
| 1660 |
+
"learning_rate": 0.0004412145463104625,
|
| 1661 |
+
"loss": 2.9141,
|
| 1662 |
+
"step": 222000
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 11.99,
|
| 1666 |
+
"grad_norm": 0.2550143301486969,
|
| 1667 |
+
"learning_rate": 0.0004382723314110863,
|
| 1668 |
+
"loss": 2.9132,
|
| 1669 |
+
"step": 223000
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 12.0,
|
| 1673 |
+
"eval_accuracy": 0.41063586520904716,
|
| 1674 |
+
"eval_loss": 3.338320255279541,
|
| 1675 |
+
"eval_runtime": 153.6675,
|
| 1676 |
+
"eval_samples_per_second": 376.911,
|
| 1677 |
+
"eval_steps_per_second": 5.889,
|
| 1678 |
+
"step": 223128
|
| 1679 |
+
},
|
| 1680 |
+
{
|
| 1681 |
+
"epoch": 12.05,
|
| 1682 |
+
"grad_norm": 0.2587660253047943,
|
| 1683 |
+
"learning_rate": 0.0004353330587266094,
|
| 1684 |
+
"loss": 2.8589,
|
| 1685 |
+
"step": 224000
|
| 1686 |
+
},
|
| 1687 |
+
{
|
| 1688 |
+
"epoch": 12.1,
|
| 1689 |
+
"grad_norm": 0.3176356554031372,
|
| 1690 |
+
"learning_rate": 0.00043239084382723313,
|
| 1691 |
+
"loss": 2.8592,
|
| 1692 |
+
"step": 225000
|
| 1693 |
+
},
|
| 1694 |
+
{
|
| 1695 |
+
"epoch": 12.15,
|
| 1696 |
+
"grad_norm": 0.27508774399757385,
|
| 1697 |
+
"learning_rate": 0.0004294486289278569,
|
| 1698 |
+
"loss": 2.864,
|
| 1699 |
+
"step": 226000
|
| 1700 |
+
},
|
| 1701 |
+
{
|
| 1702 |
+
"epoch": 12.21,
|
| 1703 |
+
"grad_norm": 0.24548423290252686,
|
| 1704 |
+
"learning_rate": 0.00042650641402848066,
|
| 1705 |
+
"loss": 2.8653,
|
| 1706 |
+
"step": 227000
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 12.26,
|
| 1710 |
+
"grad_norm": 0.2932184338569641,
|
| 1711 |
+
"learning_rate": 0.00042356714134400374,
|
| 1712 |
+
"loss": 2.8716,
|
| 1713 |
+
"step": 228000
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 12.32,
|
| 1717 |
+
"grad_norm": 0.2907082736492157,
|
| 1718 |
+
"learning_rate": 0.00042062492644462753,
|
| 1719 |
+
"loss": 2.8733,
|
| 1720 |
+
"step": 229000
|
| 1721 |
+
},
|
| 1722 |
+
{
|
| 1723 |
+
"epoch": 12.37,
|
| 1724 |
+
"grad_norm": 0.2678309679031372,
|
| 1725 |
+
"learning_rate": 0.00041768565376015067,
|
| 1726 |
+
"loss": 2.8772,
|
| 1727 |
+
"step": 230000
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 12.42,
|
| 1731 |
+
"grad_norm": 0.3048242926597595,
|
| 1732 |
+
"learning_rate": 0.0004147434388607744,
|
| 1733 |
+
"loss": 2.8825,
|
| 1734 |
+
"step": 231000
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 12.48,
|
| 1738 |
+
"grad_norm": 0.30734652280807495,
|
| 1739 |
+
"learning_rate": 0.00041180122396139814,
|
| 1740 |
+
"loss": 2.8826,
|
| 1741 |
+
"step": 232000
|
| 1742 |
+
},
|
| 1743 |
+
{
|
| 1744 |
+
"epoch": 12.53,
|
| 1745 |
+
"grad_norm": 0.28315410017967224,
|
| 1746 |
+
"learning_rate": 0.0004088590090620219,
|
| 1747 |
+
"loss": 2.8808,
|
| 1748 |
+
"step": 233000
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 12.58,
|
| 1752 |
+
"grad_norm": 0.2624458074569702,
|
| 1753 |
+
"learning_rate": 0.0004059167941626457,
|
| 1754 |
+
"loss": 2.8826,
|
| 1755 |
+
"step": 234000
|
| 1756 |
+
},
|
| 1757 |
+
{
|
| 1758 |
+
"epoch": 12.64,
|
| 1759 |
+
"grad_norm": 0.2700469195842743,
|
| 1760 |
+
"learning_rate": 0.0004029775214781688,
|
| 1761 |
+
"loss": 2.8829,
|
| 1762 |
+
"step": 235000
|
| 1763 |
+
},
|
| 1764 |
+
{
|
| 1765 |
+
"epoch": 12.69,
|
| 1766 |
+
"grad_norm": 0.27669405937194824,
|
| 1767 |
+
"learning_rate": 0.0004000382487936919,
|
| 1768 |
+
"loss": 2.8901,
|
| 1769 |
+
"step": 236000
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 12.75,
|
| 1773 |
+
"grad_norm": 0.2921361029148102,
|
| 1774 |
+
"learning_rate": 0.0003970960338943156,
|
| 1775 |
+
"loss": 2.8869,
|
| 1776 |
+
"step": 237000
|
| 1777 |
+
},
|
| 1778 |
+
{
|
| 1779 |
+
"epoch": 12.8,
|
| 1780 |
+
"grad_norm": 0.2814725339412689,
|
| 1781 |
+
"learning_rate": 0.00039415970342473815,
|
| 1782 |
+
"loss": 2.8884,
|
| 1783 |
+
"step": 238000
|
| 1784 |
+
},
|
| 1785 |
+
{
|
| 1786 |
+
"epoch": 12.85,
|
| 1787 |
+
"grad_norm": 0.2902793288230896,
|
| 1788 |
+
"learning_rate": 0.0003912174885253619,
|
| 1789 |
+
"loss": 2.8909,
|
| 1790 |
+
"step": 239000
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 12.91,
|
| 1794 |
+
"grad_norm": 0.28354400396347046,
|
| 1795 |
+
"learning_rate": 0.00038827527362598563,
|
| 1796 |
+
"loss": 2.8936,
|
| 1797 |
+
"step": 240000
|
| 1798 |
+
},
|
| 1799 |
+
{
|
| 1800 |
+
"epoch": 12.96,
|
| 1801 |
+
"grad_norm": 0.2816365659236908,
|
| 1802 |
+
"learning_rate": 0.0003853330587266094,
|
| 1803 |
+
"loss": 2.8884,
|
| 1804 |
+
"step": 241000
|
| 1805 |
+
},
|
| 1806 |
+
{
|
| 1807 |
+
"epoch": 13.0,
|
| 1808 |
+
"eval_accuracy": 0.41050620607082894,
|
| 1809 |
+
"eval_loss": 3.377206563949585,
|
| 1810 |
+
"eval_runtime": 153.5063,
|
| 1811 |
+
"eval_samples_per_second": 377.307,
|
| 1812 |
+
"eval_steps_per_second": 5.896,
|
| 1813 |
+
"step": 241722
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 13.01,
|
| 1817 |
+
"grad_norm": 0.2734642028808594,
|
| 1818 |
+
"learning_rate": 0.00038239084382723316,
|
| 1819 |
+
"loss": 2.8709,
|
| 1820 |
+
"step": 242000
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 13.07,
|
| 1824 |
+
"grad_norm": 0.2819451093673706,
|
| 1825 |
+
"learning_rate": 0.0003794486289278569,
|
| 1826 |
+
"loss": 2.8397,
|
| 1827 |
+
"step": 243000
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 13.12,
|
| 1831 |
+
"grad_norm": 0.27505162358283997,
|
| 1832 |
+
"learning_rate": 0.00037650935624338003,
|
| 1833 |
+
"loss": 2.8417,
|
| 1834 |
+
"step": 244000
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 13.18,
|
| 1838 |
+
"grad_norm": 0.29253360629081726,
|
| 1839 |
+
"learning_rate": 0.00037356714134400377,
|
| 1840 |
+
"loss": 2.8452,
|
| 1841 |
+
"step": 245000
|
| 1842 |
+
},
|
| 1843 |
+
{
|
| 1844 |
+
"epoch": 13.23,
|
| 1845 |
+
"grad_norm": 0.3059943914413452,
|
| 1846 |
+
"learning_rate": 0.0003706249264446275,
|
| 1847 |
+
"loss": 2.8485,
|
| 1848 |
+
"step": 246000
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 13.28,
|
| 1852 |
+
"grad_norm": 0.2858084440231323,
|
| 1853 |
+
"learning_rate": 0.0003676827115452513,
|
| 1854 |
+
"loss": 2.8468,
|
| 1855 |
+
"step": 247000
|
| 1856 |
+
},
|
| 1857 |
+
{
|
| 1858 |
+
"epoch": 13.34,
|
| 1859 |
+
"grad_norm": 0.309132844209671,
|
| 1860 |
+
"learning_rate": 0.0003647434388607744,
|
| 1861 |
+
"loss": 2.8587,
|
| 1862 |
+
"step": 248000
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 13.39,
|
| 1866 |
+
"grad_norm": 0.26341453194618225,
|
| 1867 |
+
"learning_rate": 0.0003618012239613981,
|
| 1868 |
+
"loss": 2.8549,
|
| 1869 |
+
"step": 249000
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 13.45,
|
| 1873 |
+
"grad_norm": 0.3062613904476166,
|
| 1874 |
+
"learning_rate": 0.0003588619512769213,
|
| 1875 |
+
"loss": 2.8604,
|
| 1876 |
+
"step": 250000
|
| 1877 |
+
},
|
| 1878 |
+
{
|
| 1879 |
+
"epoch": 13.5,
|
| 1880 |
+
"grad_norm": 0.27354612946510315,
|
| 1881 |
+
"learning_rate": 0.00035591973637754505,
|
| 1882 |
+
"loss": 2.8607,
|
| 1883 |
+
"step": 251000
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 13.55,
|
| 1887 |
+
"grad_norm": 0.2829759418964386,
|
| 1888 |
+
"learning_rate": 0.0003529804636930681,
|
| 1889 |
+
"loss": 2.8609,
|
| 1890 |
+
"step": 252000
|
| 1891 |
+
},
|
| 1892 |
+
{
|
| 1893 |
+
"epoch": 13.61,
|
| 1894 |
+
"grad_norm": 0.28467702865600586,
|
| 1895 |
+
"learning_rate": 0.0003500382487936919,
|
| 1896 |
+
"loss": 2.8668,
|
| 1897 |
+
"step": 253000
|
| 1898 |
+
},
|
| 1899 |
+
{
|
| 1900 |
+
"epoch": 13.66,
|
| 1901 |
+
"grad_norm": 0.2903009057044983,
|
| 1902 |
+
"learning_rate": 0.00034709603389431566,
|
| 1903 |
+
"loss": 2.8623,
|
| 1904 |
+
"step": 254000
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 13.71,
|
| 1908 |
+
"grad_norm": 0.31163880228996277,
|
| 1909 |
+
"learning_rate": 0.0003441538189949394,
|
| 1910 |
+
"loss": 2.8682,
|
| 1911 |
+
"step": 255000
|
| 1912 |
+
},
|
| 1913 |
+
{
|
| 1914 |
+
"epoch": 13.77,
|
| 1915 |
+
"grad_norm": 0.2995510995388031,
|
| 1916 |
+
"learning_rate": 0.00034121454631046253,
|
| 1917 |
+
"loss": 2.8678,
|
| 1918 |
+
"step": 256000
|
| 1919 |
+
},
|
| 1920 |
+
{
|
| 1921 |
+
"epoch": 13.82,
|
| 1922 |
+
"grad_norm": 0.295254111289978,
|
| 1923 |
+
"learning_rate": 0.00033827233141108627,
|
| 1924 |
+
"loss": 2.8731,
|
| 1925 |
+
"step": 257000
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"epoch": 13.88,
|
| 1929 |
+
"grad_norm": 0.2773347795009613,
|
| 1930 |
+
"learning_rate": 0.0003353330587266094,
|
| 1931 |
+
"loss": 2.8722,
|
| 1932 |
+
"step": 258000
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"epoch": 13.93,
|
| 1936 |
+
"grad_norm": 0.2755262851715088,
|
| 1937 |
+
"learning_rate": 0.0003323908438272332,
|
| 1938 |
+
"loss": 2.8707,
|
| 1939 |
+
"step": 259000
|
| 1940 |
+
},
|
| 1941 |
+
{
|
| 1942 |
+
"epoch": 13.98,
|
| 1943 |
+
"grad_norm": 0.27530667185783386,
|
| 1944 |
+
"learning_rate": 0.0003294486289278569,
|
| 1945 |
+
"loss": 2.8698,
|
| 1946 |
+
"step": 260000
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 14.0,
|
| 1950 |
+
"eval_accuracy": 0.4117409910244301,
|
| 1951 |
+
"eval_loss": 3.3456804752349854,
|
| 1952 |
+
"eval_runtime": 153.4816,
|
| 1953 |
+
"eval_samples_per_second": 377.368,
|
| 1954 |
+
"eval_steps_per_second": 5.896,
|
| 1955 |
+
"step": 260316
|
| 1956 |
+
},
|
| 1957 |
+
{
|
| 1958 |
+
"epoch": 14.04,
|
| 1959 |
+
"grad_norm": 0.2921120524406433,
|
| 1960 |
+
"learning_rate": 0.0003265064140284806,
|
| 1961 |
+
"loss": 2.8368,
|
| 1962 |
+
"step": 261000
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 14.09,
|
| 1966 |
+
"grad_norm": 0.28521835803985596,
|
| 1967 |
+
"learning_rate": 0.0003235671413440038,
|
| 1968 |
+
"loss": 2.826,
|
| 1969 |
+
"step": 262000
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 14.14,
|
| 1973 |
+
"grad_norm": 0.3207342326641083,
|
| 1974 |
+
"learning_rate": 0.00032062786865952694,
|
| 1975 |
+
"loss": 2.8255,
|
| 1976 |
+
"step": 263000
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 14.2,
|
| 1980 |
+
"grad_norm": 0.3034232258796692,
|
| 1981 |
+
"learning_rate": 0.0003176856537601506,
|
| 1982 |
+
"loss": 2.8267,
|
| 1983 |
+
"step": 264000
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 14.25,
|
| 1987 |
+
"grad_norm": 0.28970643877983093,
|
| 1988 |
+
"learning_rate": 0.0003147434388607744,
|
| 1989 |
+
"loss": 2.8332,
|
| 1990 |
+
"step": 265000
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 14.31,
|
| 1994 |
+
"grad_norm": 0.31462568044662476,
|
| 1995 |
+
"learning_rate": 0.00031180122396139815,
|
| 1996 |
+
"loss": 2.833,
|
| 1997 |
+
"step": 266000
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 14.36,
|
| 2001 |
+
"grad_norm": 0.31881943345069885,
|
| 2002 |
+
"learning_rate": 0.0003088590090620219,
|
| 2003 |
+
"loss": 2.8384,
|
| 2004 |
+
"step": 267000
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 14.41,
|
| 2008 |
+
"grad_norm": 0.31022316217422485,
|
| 2009 |
+
"learning_rate": 0.0003059167941626457,
|
| 2010 |
+
"loss": 2.8363,
|
| 2011 |
+
"step": 268000
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 14.47,
|
| 2015 |
+
"grad_norm": 0.2837056815624237,
|
| 2016 |
+
"learning_rate": 0.00030297752147816876,
|
| 2017 |
+
"loss": 2.8413,
|
| 2018 |
+
"step": 269000
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 14.52,
|
| 2022 |
+
"grad_norm": 0.2785826325416565,
|
| 2023 |
+
"learning_rate": 0.0003000382487936919,
|
| 2024 |
+
"loss": 2.8413,
|
| 2025 |
+
"step": 270000
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 14.57,
|
| 2029 |
+
"grad_norm": 0.28700125217437744,
|
| 2030 |
+
"learning_rate": 0.0002970960338943157,
|
| 2031 |
+
"loss": 2.8427,
|
| 2032 |
+
"step": 271000
|
| 2033 |
+
},
|
| 2034 |
+
{
|
| 2035 |
+
"epoch": 14.63,
|
| 2036 |
+
"grad_norm": 0.2973004877567291,
|
| 2037 |
+
"learning_rate": 0.00029415676120983877,
|
| 2038 |
+
"loss": 2.843,
|
| 2039 |
+
"step": 272000
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 14.68,
|
| 2043 |
+
"grad_norm": 0.2778867185115814,
|
| 2044 |
+
"learning_rate": 0.0002912145463104625,
|
| 2045 |
+
"loss": 2.8442,
|
| 2046 |
+
"step": 273000
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 14.74,
|
| 2050 |
+
"grad_norm": 0.2906304895877838,
|
| 2051 |
+
"learning_rate": 0.0002882723314110863,
|
| 2052 |
+
"loss": 2.8471,
|
| 2053 |
+
"step": 274000
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 14.79,
|
| 2057 |
+
"grad_norm": 0.2736400365829468,
|
| 2058 |
+
"learning_rate": 0.00028533305872660943,
|
| 2059 |
+
"loss": 2.8513,
|
| 2060 |
+
"step": 275000
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 14.84,
|
| 2064 |
+
"grad_norm": 0.26769936084747314,
|
| 2065 |
+
"learning_rate": 0.0002823908438272331,
|
| 2066 |
+
"loss": 2.8502,
|
| 2067 |
+
"step": 276000
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 14.9,
|
| 2071 |
+
"grad_norm": 0.27321502566337585,
|
| 2072 |
+
"learning_rate": 0.00027945157114275625,
|
| 2073 |
+
"loss": 2.8543,
|
| 2074 |
+
"step": 277000
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 14.95,
|
| 2078 |
+
"grad_norm": 0.3363865911960602,
|
| 2079 |
+
"learning_rate": 0.00027650935624338004,
|
| 2080 |
+
"loss": 2.8512,
|
| 2081 |
+
"step": 278000
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 15.0,
|
| 2085 |
+
"eval_accuracy": 0.41095604938145536,
|
| 2086 |
+
"eval_loss": 3.3591506481170654,
|
| 2087 |
+
"eval_runtime": 153.9239,
|
| 2088 |
+
"eval_samples_per_second": 376.283,
|
| 2089 |
+
"eval_steps_per_second": 5.88,
|
| 2090 |
+
"step": 278910
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 15.0,
|
| 2094 |
+
"grad_norm": 0.2882598042488098,
|
| 2095 |
+
"learning_rate": 0.0002735700835589032,
|
| 2096 |
+
"loss": 2.8467,
|
| 2097 |
+
"step": 279000
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 15.06,
|
| 2101 |
+
"grad_norm": 0.3232513964176178,
|
| 2102 |
+
"learning_rate": 0.00027062786865952686,
|
| 2103 |
+
"loss": 2.8004,
|
| 2104 |
+
"step": 280000
|
| 2105 |
+
},
|
| 2106 |
+
{
|
| 2107 |
+
"epoch": 15.11,
|
| 2108 |
+
"grad_norm": 0.296671986579895,
|
| 2109 |
+
"learning_rate": 0.00026768565376015065,
|
| 2110 |
+
"loss": 2.8081,
|
| 2111 |
+
"step": 281000
|
| 2112 |
+
},
|
| 2113 |
+
{
|
| 2114 |
+
"epoch": 15.17,
|
| 2115 |
+
"grad_norm": 0.2888088822364807,
|
| 2116 |
+
"learning_rate": 0.0002647434388607744,
|
| 2117 |
+
"loss": 2.8074,
|
| 2118 |
+
"step": 282000
|
| 2119 |
+
},
|
| 2120 |
+
{
|
| 2121 |
+
"epoch": 15.22,
|
| 2122 |
+
"grad_norm": 0.30186396837234497,
|
| 2123 |
+
"learning_rate": 0.0002618041661762975,
|
| 2124 |
+
"loss": 2.8142,
|
| 2125 |
+
"step": 283000
|
| 2126 |
+
},
|
| 2127 |
+
{
|
| 2128 |
+
"epoch": 15.27,
|
| 2129 |
+
"grad_norm": 0.310922235250473,
|
| 2130 |
+
"learning_rate": 0.00025886195127692126,
|
| 2131 |
+
"loss": 2.8164,
|
| 2132 |
+
"step": 284000
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 15.33,
|
| 2136 |
+
"grad_norm": 0.2841443419456482,
|
| 2137 |
+
"learning_rate": 0.000255919736377545,
|
| 2138 |
+
"loss": 2.8191,
|
| 2139 |
+
"step": 285000
|
| 2140 |
+
},
|
| 2141 |
+
{
|
| 2142 |
+
"epoch": 15.38,
|
| 2143 |
+
"grad_norm": 0.3377715051174164,
|
| 2144 |
+
"learning_rate": 0.00025298046369306813,
|
| 2145 |
+
"loss": 2.8195,
|
| 2146 |
+
"step": 286000
|
| 2147 |
+
},
|
| 2148 |
+
{
|
| 2149 |
+
"epoch": 15.44,
|
| 2150 |
+
"grad_norm": 0.2865951359272003,
|
| 2151 |
+
"learning_rate": 0.00025003824879369193,
|
| 2152 |
+
"loss": 2.8266,
|
| 2153 |
+
"step": 287000
|
| 2154 |
+
},
|
| 2155 |
+
{
|
| 2156 |
+
"epoch": 15.49,
|
| 2157 |
+
"grad_norm": 0.31996873021125793,
|
| 2158 |
+
"learning_rate": 0.00024709603389431567,
|
| 2159 |
+
"loss": 2.825,
|
| 2160 |
+
"step": 288000
|
| 2161 |
+
},
|
| 2162 |
+
{
|
| 2163 |
+
"epoch": 15.54,
|
| 2164 |
+
"grad_norm": 0.30408111214637756,
|
| 2165 |
+
"learning_rate": 0.0002441538189949394,
|
| 2166 |
+
"loss": 2.8227,
|
| 2167 |
+
"step": 289000
|
| 2168 |
+
},
|
| 2169 |
+
{
|
| 2170 |
+
"epoch": 15.6,
|
| 2171 |
+
"grad_norm": 0.2882407307624817,
|
| 2172 |
+
"learning_rate": 0.0002412174885253619,
|
| 2173 |
+
"loss": 2.8263,
|
| 2174 |
+
"step": 290000
|
| 2175 |
+
},
|
| 2176 |
+
{
|
| 2177 |
+
"epoch": 15.65,
|
| 2178 |
+
"grad_norm": 0.3298856019973755,
|
| 2179 |
+
"learning_rate": 0.00023827527362598564,
|
| 2180 |
+
"loss": 2.8313,
|
| 2181 |
+
"step": 291000
|
| 2182 |
+
},
|
| 2183 |
+
{
|
| 2184 |
+
"epoch": 15.7,
|
| 2185 |
+
"grad_norm": 0.28874561190605164,
|
| 2186 |
+
"learning_rate": 0.0002353330587266094,
|
| 2187 |
+
"loss": 2.8281,
|
| 2188 |
+
"step": 292000
|
| 2189 |
+
},
|
| 2190 |
+
{
|
| 2191 |
+
"epoch": 15.76,
|
| 2192 |
+
"grad_norm": 0.32142964005470276,
|
| 2193 |
+
"learning_rate": 0.00023239084382723312,
|
| 2194 |
+
"loss": 2.8289,
|
| 2195 |
+
"step": 293000
|
| 2196 |
+
},
|
| 2197 |
+
{
|
| 2198 |
+
"epoch": 15.81,
|
| 2199 |
+
"grad_norm": 0.3016972243785858,
|
| 2200 |
+
"learning_rate": 0.00022945157114275628,
|
| 2201 |
+
"loss": 2.8286,
|
| 2202 |
+
"step": 294000
|
| 2203 |
+
},
|
| 2204 |
+
{
|
| 2205 |
+
"epoch": 15.87,
|
| 2206 |
+
"grad_norm": 0.29328015446662903,
|
| 2207 |
+
"learning_rate": 0.00022650935624338002,
|
| 2208 |
+
"loss": 2.826,
|
| 2209 |
+
"step": 295000
|
| 2210 |
+
},
|
| 2211 |
+
{
|
| 2212 |
+
"epoch": 15.92,
|
| 2213 |
+
"grad_norm": 0.2902560830116272,
|
| 2214 |
+
"learning_rate": 0.00022356714134400378,
|
| 2215 |
+
"loss": 2.8349,
|
| 2216 |
+
"step": 296000
|
| 2217 |
+
},
|
| 2218 |
+
{
|
| 2219 |
+
"epoch": 15.97,
|
| 2220 |
+
"grad_norm": 0.32843369245529175,
|
| 2221 |
+
"learning_rate": 0.0002206278686595269,
|
| 2222 |
+
"loss": 2.828,
|
| 2223 |
+
"step": 297000
|
| 2224 |
+
},
|
| 2225 |
+
{
|
| 2226 |
+
"epoch": 16.0,
|
| 2227 |
+
"eval_accuracy": 0.4105747977703734,
|
| 2228 |
+
"eval_loss": 3.378222703933716,
|
| 2229 |
+
"eval_runtime": 153.6775,
|
| 2230 |
+
"eval_samples_per_second": 376.887,
|
| 2231 |
+
"eval_steps_per_second": 5.889,
|
| 2232 |
+
"step": 297504
|
| 2233 |
+
},
|
| 2234 |
+
{
|
| 2235 |
+
"epoch": 16.03,
|
| 2236 |
+
"grad_norm": 0.30167028307914734,
|
| 2237 |
+
"learning_rate": 0.00021768859597505003,
|
| 2238 |
+
"loss": 2.8067,
|
| 2239 |
+
"step": 298000
|
| 2240 |
+
},
|
| 2241 |
+
{
|
| 2242 |
+
"epoch": 16.08,
|
| 2243 |
+
"grad_norm": 0.30875760316848755,
|
| 2244 |
+
"learning_rate": 0.00021474638107567376,
|
| 2245 |
+
"loss": 2.7902,
|
| 2246 |
+
"step": 299000
|
| 2247 |
+
},
|
| 2248 |
+
{
|
| 2249 |
+
"epoch": 16.13,
|
| 2250 |
+
"grad_norm": 0.3411406874656677,
|
| 2251 |
+
"learning_rate": 0.0002118071083911969,
|
| 2252 |
+
"loss": 2.7937,
|
| 2253 |
+
"step": 300000
|
| 2254 |
+
},
|
| 2255 |
+
{
|
| 2256 |
+
"epoch": 16.19,
|
| 2257 |
+
"grad_norm": 0.2929510474205017,
|
| 2258 |
+
"learning_rate": 0.00020886489349182064,
|
| 2259 |
+
"loss": 2.7947,
|
| 2260 |
+
"step": 301000
|
| 2261 |
+
},
|
| 2262 |
+
{
|
| 2263 |
+
"epoch": 16.24,
|
| 2264 |
+
"grad_norm": 0.28705376386642456,
|
| 2265 |
+
"learning_rate": 0.0002059226785924444,
|
| 2266 |
+
"loss": 2.7961,
|
| 2267 |
+
"step": 302000
|
| 2268 |
+
},
|
| 2269 |
+
{
|
| 2270 |
+
"epoch": 16.3,
|
| 2271 |
+
"grad_norm": 0.3195827901363373,
|
| 2272 |
+
"learning_rate": 0.00020298046369306814,
|
| 2273 |
+
"loss": 2.7969,
|
| 2274 |
+
"step": 303000
|
| 2275 |
+
},
|
| 2276 |
+
{
|
| 2277 |
+
"epoch": 16.35,
|
| 2278 |
+
"grad_norm": 0.2913331389427185,
|
| 2279 |
+
"learning_rate": 0.0002000382487936919,
|
| 2280 |
+
"loss": 2.7994,
|
| 2281 |
+
"step": 304000
|
| 2282 |
+
},
|
| 2283 |
+
{
|
| 2284 |
+
"epoch": 16.4,
|
| 2285 |
+
"grad_norm": 0.3066113293170929,
|
| 2286 |
+
"learning_rate": 0.000197098976109215,
|
| 2287 |
+
"loss": 2.805,
|
| 2288 |
+
"step": 305000
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 16.46,
|
| 2292 |
+
"grad_norm": 0.3010548949241638,
|
| 2293 |
+
"learning_rate": 0.00019415676120983878,
|
| 2294 |
+
"loss": 2.8019,
|
| 2295 |
+
"step": 306000
|
| 2296 |
+
},
|
| 2297 |
+
{
|
| 2298 |
+
"epoch": 16.51,
|
| 2299 |
+
"grad_norm": 0.30278268456459045,
|
| 2300 |
+
"learning_rate": 0.00019121748852536188,
|
| 2301 |
+
"loss": 2.8091,
|
| 2302 |
+
"step": 307000
|
| 2303 |
+
},
|
| 2304 |
+
{
|
| 2305 |
+
"epoch": 16.56,
|
| 2306 |
+
"grad_norm": 0.29156890511512756,
|
| 2307 |
+
"learning_rate": 0.00018827527362598565,
|
| 2308 |
+
"loss": 2.809,
|
| 2309 |
+
"step": 308000
|
| 2310 |
+
},
|
| 2311 |
+
{
|
| 2312 |
+
"epoch": 16.62,
|
| 2313 |
+
"grad_norm": 0.3008438050746918,
|
| 2314 |
+
"learning_rate": 0.0001853330587266094,
|
| 2315 |
+
"loss": 2.8096,
|
| 2316 |
+
"step": 309000
|
| 2317 |
+
},
|
| 2318 |
+
{
|
| 2319 |
+
"epoch": 16.67,
|
| 2320 |
+
"grad_norm": 0.3331317603588104,
|
| 2321 |
+
"learning_rate": 0.00018239084382723315,
|
| 2322 |
+
"loss": 2.8093,
|
| 2323 |
+
"step": 310000
|
| 2324 |
+
},
|
| 2325 |
+
{
|
| 2326 |
+
"epoch": 16.73,
|
| 2327 |
+
"grad_norm": 0.3314554691314697,
|
| 2328 |
+
"learning_rate": 0.00017945157114275626,
|
| 2329 |
+
"loss": 2.8056,
|
| 2330 |
+
"step": 311000
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 16.78,
|
| 2334 |
+
"grad_norm": 0.2929142117500305,
|
| 2335 |
+
"learning_rate": 0.00017650935624338002,
|
| 2336 |
+
"loss": 2.8121,
|
| 2337 |
+
"step": 312000
|
| 2338 |
+
},
|
| 2339 |
+
{
|
| 2340 |
+
"epoch": 16.83,
|
| 2341 |
+
"grad_norm": 0.3265311121940613,
|
| 2342 |
+
"learning_rate": 0.00017357008355890313,
|
| 2343 |
+
"loss": 2.8153,
|
| 2344 |
+
"step": 313000
|
| 2345 |
+
},
|
| 2346 |
+
{
|
| 2347 |
+
"epoch": 16.89,
|
| 2348 |
+
"grad_norm": 0.306043416261673,
|
| 2349 |
+
"learning_rate": 0.0001706278686595269,
|
| 2350 |
+
"loss": 2.813,
|
| 2351 |
+
"step": 314000
|
| 2352 |
+
},
|
| 2353 |
+
{
|
| 2354 |
+
"epoch": 16.94,
|
| 2355 |
+
"grad_norm": 0.31978315114974976,
|
| 2356 |
+
"learning_rate": 0.00016768859597505,
|
| 2357 |
+
"loss": 2.8156,
|
| 2358 |
+
"step": 315000
|
| 2359 |
+
},
|
| 2360 |
+
{
|
| 2361 |
+
"epoch": 16.99,
|
| 2362 |
+
"grad_norm": 0.28355807065963745,
|
| 2363 |
+
"learning_rate": 0.00016474638107567377,
|
| 2364 |
+
"loss": 2.813,
|
| 2365 |
+
"step": 316000
|
| 2366 |
+
},
|
| 2367 |
+
{
|
| 2368 |
+
"epoch": 17.0,
|
| 2369 |
+
"eval_accuracy": 0.4109091571128148,
|
| 2370 |
+
"eval_loss": 3.377826452255249,
|
| 2371 |
+
"eval_runtime": 153.6131,
|
| 2372 |
+
"eval_samples_per_second": 377.045,
|
| 2373 |
+
"eval_steps_per_second": 5.891,
|
| 2374 |
+
"step": 316098
|
| 2375 |
+
},
|
| 2376 |
+
{
|
| 2377 |
+
"epoch": 17.05,
|
| 2378 |
+
"grad_norm": 0.296631783246994,
|
| 2379 |
+
"learning_rate": 0.0001618071083911969,
|
| 2380 |
+
"loss": 2.7755,
|
| 2381 |
+
"step": 317000
|
| 2382 |
+
},
|
| 2383 |
+
{
|
| 2384 |
+
"epoch": 17.1,
|
| 2385 |
+
"grad_norm": 0.3535520136356354,
|
| 2386 |
+
"learning_rate": 0.00015886489349182064,
|
| 2387 |
+
"loss": 2.7773,
|
| 2388 |
+
"step": 318000
|
| 2389 |
+
},
|
| 2390 |
+
{
|
| 2391 |
+
"epoch": 17.16,
|
| 2392 |
+
"grad_norm": 0.31634917855262756,
|
| 2393 |
+
"learning_rate": 0.00015592562080734377,
|
| 2394 |
+
"loss": 2.7776,
|
| 2395 |
+
"step": 319000
|
| 2396 |
+
},
|
| 2397 |
+
{
|
| 2398 |
+
"epoch": 17.21,
|
| 2399 |
+
"grad_norm": 0.33776313066482544,
|
| 2400 |
+
"learning_rate": 0.00015298340590796754,
|
| 2401 |
+
"loss": 2.7783,
|
| 2402 |
+
"step": 320000
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 17.26,
|
| 2406 |
+
"grad_norm": 0.3450954258441925,
|
| 2407 |
+
"learning_rate": 0.00015004119100859125,
|
| 2408 |
+
"loss": 2.7816,
|
| 2409 |
+
"step": 321000
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 17.32,
|
| 2413 |
+
"grad_norm": 0.2964255213737488,
|
| 2414 |
+
"learning_rate": 0.00014709897610921502,
|
| 2415 |
+
"loss": 2.7844,
|
| 2416 |
+
"step": 322000
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 17.37,
|
| 2420 |
+
"grad_norm": 0.31515949964523315,
|
| 2421 |
+
"learning_rate": 0.00014415676120983878,
|
| 2422 |
+
"loss": 2.785,
|
| 2423 |
+
"step": 323000
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 17.42,
|
| 2427 |
+
"grad_norm": 0.29529762268066406,
|
| 2428 |
+
"learning_rate": 0.0001412174885253619,
|
| 2429 |
+
"loss": 2.7865,
|
| 2430 |
+
"step": 324000
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 17.48,
|
| 2434 |
+
"grad_norm": 0.28874969482421875,
|
| 2435 |
+
"learning_rate": 0.00013827527362598563,
|
| 2436 |
+
"loss": 2.7872,
|
| 2437 |
+
"step": 325000
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 17.53,
|
| 2441 |
+
"grad_norm": 0.3028223514556885,
|
| 2442 |
+
"learning_rate": 0.0001353330587266094,
|
| 2443 |
+
"loss": 2.7899,
|
| 2444 |
+
"step": 326000
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 17.59,
|
| 2448 |
+
"grad_norm": 0.31474238634109497,
|
| 2449 |
+
"learning_rate": 0.00013239084382723316,
|
| 2450 |
+
"loss": 2.7892,
|
| 2451 |
+
"step": 327000
|
| 2452 |
+
},
|
| 2453 |
+
{
|
| 2454 |
+
"epoch": 17.64,
|
| 2455 |
+
"grad_norm": 0.31951338052749634,
|
| 2456 |
+
"learning_rate": 0.00012945157114275626,
|
| 2457 |
+
"loss": 2.7966,
|
| 2458 |
+
"step": 328000
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 17.69,
|
| 2462 |
+
"grad_norm": 0.32487303018569946,
|
| 2463 |
+
"learning_rate": 0.0001265122984582794,
|
| 2464 |
+
"loss": 2.7909,
|
| 2465 |
+
"step": 329000
|
| 2466 |
+
},
|
| 2467 |
+
{
|
| 2468 |
+
"epoch": 17.75,
|
| 2469 |
+
"grad_norm": 0.2972046732902527,
|
| 2470 |
+
"learning_rate": 0.00012357008355890313,
|
| 2471 |
+
"loss": 2.7907,
|
| 2472 |
+
"step": 330000
|
| 2473 |
+
},
|
| 2474 |
+
{
|
| 2475 |
+
"epoch": 17.8,
|
| 2476 |
+
"grad_norm": 0.3160036504268646,
|
| 2477 |
+
"learning_rate": 0.00012063081087442627,
|
| 2478 |
+
"loss": 2.7965,
|
| 2479 |
+
"step": 331000
|
| 2480 |
+
},
|
| 2481 |
+
{
|
| 2482 |
+
"epoch": 17.86,
|
| 2483 |
+
"grad_norm": 0.31017643213272095,
|
| 2484 |
+
"learning_rate": 0.00011768859597505002,
|
| 2485 |
+
"loss": 2.7938,
|
| 2486 |
+
"step": 332000
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 17.91,
|
| 2490 |
+
"grad_norm": 0.3148263692855835,
|
| 2491 |
+
"learning_rate": 0.00011474638107567377,
|
| 2492 |
+
"loss": 2.799,
|
| 2493 |
+
"step": 333000
|
| 2494 |
+
},
|
| 2495 |
+
{
|
| 2496 |
+
"epoch": 17.96,
|
| 2497 |
+
"grad_norm": 0.30343878269195557,
|
| 2498 |
+
"learning_rate": 0.00011180416617629751,
|
| 2499 |
+
"loss": 2.7978,
|
| 2500 |
+
"step": 334000
|
| 2501 |
+
},
|
| 2502 |
+
{
|
| 2503 |
+
"epoch": 18.0,
|
| 2504 |
+
"eval_accuracy": 0.41052434491399936,
|
| 2505 |
+
"eval_loss": 3.3930857181549072,
|
| 2506 |
+
"eval_runtime": 153.7013,
|
| 2507 |
+
"eval_samples_per_second": 376.828,
|
| 2508 |
+
"eval_steps_per_second": 5.888,
|
| 2509 |
+
"step": 334692
|
| 2510 |
+
},
|
| 2511 |
+
{
|
| 2512 |
+
"epoch": 18.02,
|
| 2513 |
+
"grad_norm": 0.3155461251735687,
|
| 2514 |
+
"learning_rate": 0.00010886489349182064,
|
| 2515 |
+
"loss": 2.785,
|
| 2516 |
+
"step": 335000
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 18.07,
|
| 2520 |
+
"grad_norm": 0.30708247423171997,
|
| 2521 |
+
"learning_rate": 0.0001059226785924444,
|
| 2522 |
+
"loss": 2.7642,
|
| 2523 |
+
"step": 336000
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 18.12,
|
| 2527 |
+
"grad_norm": 0.28757408261299133,
|
| 2528 |
+
"learning_rate": 0.00010298046369306815,
|
| 2529 |
+
"loss": 2.7658,
|
| 2530 |
+
"step": 337000
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"epoch": 18.18,
|
| 2534 |
+
"grad_norm": 0.3023107051849365,
|
| 2535 |
+
"learning_rate": 0.00010004119100859127,
|
| 2536 |
+
"loss": 2.7658,
|
| 2537 |
+
"step": 338000
|
| 2538 |
+
},
|
| 2539 |
+
{
|
| 2540 |
+
"epoch": 18.23,
|
| 2541 |
+
"grad_norm": 0.34044039249420166,
|
| 2542 |
+
"learning_rate": 9.709897610921502e-05,
|
| 2543 |
+
"loss": 2.7702,
|
| 2544 |
+
"step": 339000
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"epoch": 18.29,
|
| 2548 |
+
"grad_norm": 0.3793945014476776,
|
| 2549 |
+
"learning_rate": 9.415970342473815e-05,
|
| 2550 |
+
"loss": 2.7681,
|
| 2551 |
+
"step": 340000
|
| 2552 |
+
},
|
| 2553 |
+
{
|
| 2554 |
+
"epoch": 18.34,
|
| 2555 |
+
"grad_norm": 0.31560423970222473,
|
| 2556 |
+
"learning_rate": 9.121748852536189e-05,
|
| 2557 |
+
"loss": 2.7721,
|
| 2558 |
+
"step": 341000
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 18.39,
|
| 2562 |
+
"grad_norm": 0.3134061098098755,
|
| 2563 |
+
"learning_rate": 8.827527362598564e-05,
|
| 2564 |
+
"loss": 2.7698,
|
| 2565 |
+
"step": 342000
|
| 2566 |
+
},
|
| 2567 |
+
{
|
| 2568 |
+
"epoch": 18.45,
|
| 2569 |
+
"grad_norm": 0.3193502426147461,
|
| 2570 |
+
"learning_rate": 8.533600094150878e-05,
|
| 2571 |
+
"loss": 2.7708,
|
| 2572 |
+
"step": 343000
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"epoch": 18.5,
|
| 2576 |
+
"grad_norm": 0.30446764826774597,
|
| 2577 |
+
"learning_rate": 8.239378604213252e-05,
|
| 2578 |
+
"loss": 2.7713,
|
| 2579 |
+
"step": 344000
|
| 2580 |
+
},
|
| 2581 |
+
{
|
| 2582 |
+
"epoch": 18.55,
|
| 2583 |
+
"grad_norm": 0.32142266631126404,
|
| 2584 |
+
"learning_rate": 7.945157114275627e-05,
|
| 2585 |
+
"loss": 2.773,
|
| 2586 |
+
"step": 345000
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"epoch": 18.61,
|
| 2590 |
+
"grad_norm": 0.31932076811790466,
|
| 2591 |
+
"learning_rate": 7.65122984582794e-05,
|
| 2592 |
+
"loss": 2.7777,
|
| 2593 |
+
"step": 346000
|
| 2594 |
+
},
|
| 2595 |
+
{
|
| 2596 |
+
"epoch": 18.66,
|
| 2597 |
+
"grad_norm": 0.3001977801322937,
|
| 2598 |
+
"learning_rate": 7.357008355890314e-05,
|
| 2599 |
+
"loss": 2.7728,
|
| 2600 |
+
"step": 347000
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 18.72,
|
| 2604 |
+
"grad_norm": 0.3105526566505432,
|
| 2605 |
+
"learning_rate": 7.062786865952689e-05,
|
| 2606 |
+
"loss": 2.774,
|
| 2607 |
+
"step": 348000
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 18.77,
|
| 2611 |
+
"grad_norm": 0.3131723701953888,
|
| 2612 |
+
"learning_rate": 6.768565376015064e-05,
|
| 2613 |
+
"loss": 2.7776,
|
| 2614 |
+
"step": 349000
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"epoch": 18.82,
|
| 2618 |
+
"grad_norm": 0.31600117683410645,
|
| 2619 |
+
"learning_rate": 6.474638107567376e-05,
|
| 2620 |
+
"loss": 2.7742,
|
| 2621 |
+
"step": 350000
|
| 2622 |
+
},
|
| 2623 |
+
{
|
| 2624 |
+
"epoch": 18.88,
|
| 2625 |
+
"grad_norm": 0.34644222259521484,
|
| 2626 |
+
"learning_rate": 6.18071083911969e-05,
|
| 2627 |
+
"loss": 2.7754,
|
| 2628 |
+
"step": 351000
|
| 2629 |
+
},
|
| 2630 |
+
{
|
| 2631 |
+
"epoch": 18.93,
|
| 2632 |
+
"grad_norm": 0.3488774001598358,
|
| 2633 |
+
"learning_rate": 5.886489349182064e-05,
|
| 2634 |
+
"loss": 2.7748,
|
| 2635 |
+
"step": 352000
|
| 2636 |
+
},
|
| 2637 |
+
{
|
| 2638 |
+
"epoch": 18.98,
|
| 2639 |
+
"grad_norm": 0.3060605823993683,
|
| 2640 |
+
"learning_rate": 5.5922678592444393e-05,
|
| 2641 |
+
"loss": 2.7756,
|
| 2642 |
+
"step": 353000
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 19.0,
|
| 2646 |
+
"eval_accuracy": 0.41067926407085487,
|
| 2647 |
+
"eval_loss": 3.3947384357452393,
|
| 2648 |
+
"eval_runtime": 154.2277,
|
| 2649 |
+
"eval_samples_per_second": 375.542,
|
| 2650 |
+
"eval_steps_per_second": 5.868,
|
| 2651 |
+
"step": 353286
|
| 2652 |
+
},
|
| 2653 |
+
{
|
| 2654 |
+
"epoch": 19.04,
|
| 2655 |
+
"grad_norm": 0.31407734751701355,
|
| 2656 |
+
"learning_rate": 5.2980463693068145e-05,
|
| 2657 |
+
"loss": 2.7627,
|
| 2658 |
+
"step": 354000
|
| 2659 |
+
},
|
| 2660 |
+
{
|
| 2661 |
+
"epoch": 19.09,
|
| 2662 |
+
"grad_norm": 0.3081498444080353,
|
| 2663 |
+
"learning_rate": 5.0041191008591265e-05,
|
| 2664 |
+
"loss": 2.756,
|
| 2665 |
+
"step": 355000
|
| 2666 |
+
},
|
| 2667 |
+
{
|
| 2668 |
+
"epoch": 19.15,
|
| 2669 |
+
"grad_norm": 0.32544785737991333,
|
| 2670 |
+
"learning_rate": 4.7098976109215024e-05,
|
| 2671 |
+
"loss": 2.7581,
|
| 2672 |
+
"step": 356000
|
| 2673 |
+
},
|
| 2674 |
+
{
|
| 2675 |
+
"epoch": 19.2,
|
| 2676 |
+
"grad_norm": 0.3303050100803375,
|
| 2677 |
+
"learning_rate": 4.4159703424738144e-05,
|
| 2678 |
+
"loss": 2.7534,
|
| 2679 |
+
"step": 357000
|
| 2680 |
+
},
|
| 2681 |
+
{
|
| 2682 |
+
"epoch": 19.25,
|
| 2683 |
+
"grad_norm": 0.33612754940986633,
|
| 2684 |
+
"learning_rate": 4.122043074026127e-05,
|
| 2685 |
+
"loss": 2.7589,
|
| 2686 |
+
"step": 358000
|
| 2687 |
+
},
|
| 2688 |
+
{
|
| 2689 |
+
"epoch": 19.31,
|
| 2690 |
+
"grad_norm": 0.33169466257095337,
|
| 2691 |
+
"learning_rate": 3.827821584088502e-05,
|
| 2692 |
+
"loss": 2.7556,
|
| 2693 |
+
"step": 359000
|
| 2694 |
+
},
|
| 2695 |
+
{
|
| 2696 |
+
"epoch": 19.36,
|
| 2697 |
+
"grad_norm": 0.31507307291030884,
|
| 2698 |
+
"learning_rate": 3.533600094150877e-05,
|
| 2699 |
+
"loss": 2.7589,
|
| 2700 |
+
"step": 360000
|
| 2701 |
+
},
|
| 2702 |
+
{
|
| 2703 |
+
"epoch": 19.41,
|
| 2704 |
+
"grad_norm": 0.3087073862552643,
|
| 2705 |
+
"learning_rate": 3.239378604213252e-05,
|
| 2706 |
+
"loss": 2.7548,
|
| 2707 |
+
"step": 361000
|
| 2708 |
+
},
|
| 2709 |
+
{
|
| 2710 |
+
"epoch": 19.47,
|
| 2711 |
+
"grad_norm": 0.33943119645118713,
|
| 2712 |
+
"learning_rate": 2.9454513357655643e-05,
|
| 2713 |
+
"loss": 2.7584,
|
| 2714 |
+
"step": 362000
|
| 2715 |
+
},
|
| 2716 |
+
{
|
| 2717 |
+
"epoch": 19.52,
|
| 2718 |
+
"grad_norm": 0.31947916746139526,
|
| 2719 |
+
"learning_rate": 2.6512298458279395e-05,
|
| 2720 |
+
"loss": 2.7604,
|
| 2721 |
+
"step": 363000
|
| 2722 |
+
},
|
| 2723 |
+
{
|
| 2724 |
+
"epoch": 19.58,
|
| 2725 |
+
"grad_norm": 0.2944762408733368,
|
| 2726 |
+
"learning_rate": 2.357302577380252e-05,
|
| 2727 |
+
"loss": 2.7579,
|
| 2728 |
+
"step": 364000
|
| 2729 |
+
},
|
| 2730 |
+
{
|
| 2731 |
+
"epoch": 19.63,
|
| 2732 |
+
"grad_norm": 0.4628629684448242,
|
| 2733 |
+
"learning_rate": 2.0630810874426267e-05,
|
| 2734 |
+
"loss": 2.7568,
|
| 2735 |
+
"step": 365000
|
| 2736 |
+
},
|
| 2737 |
+
{
|
| 2738 |
+
"epoch": 19.68,
|
| 2739 |
+
"grad_norm": 0.34393277764320374,
|
| 2740 |
+
"learning_rate": 1.768859597505002e-05,
|
| 2741 |
+
"loss": 2.759,
|
| 2742 |
+
"step": 366000
|
| 2743 |
+
},
|
| 2744 |
+
{
|
| 2745 |
+
"epoch": 19.74,
|
| 2746 |
+
"grad_norm": 0.3237968683242798,
|
| 2747 |
+
"learning_rate": 1.4746381075673767e-05,
|
| 2748 |
+
"loss": 2.7611,
|
| 2749 |
+
"step": 367000
|
| 2750 |
+
},
|
| 2751 |
+
{
|
| 2752 |
+
"epoch": 19.79,
|
| 2753 |
+
"grad_norm": 0.3589065372943878,
|
| 2754 |
+
"learning_rate": 1.1807108391196892e-05,
|
| 2755 |
+
"loss": 2.7575,
|
| 2756 |
+
"step": 368000
|
| 2757 |
+
},
|
| 2758 |
+
{
|
| 2759 |
+
"epoch": 19.85,
|
| 2760 |
+
"grad_norm": 0.33762767910957336,
|
| 2761 |
+
"learning_rate": 8.86783570672002e-06,
|
| 2762 |
+
"loss": 2.7551,
|
| 2763 |
+
"step": 369000
|
| 2764 |
+
},
|
| 2765 |
+
{
|
| 2766 |
+
"epoch": 19.9,
|
| 2767 |
+
"grad_norm": 0.32348111271858215,
|
| 2768 |
+
"learning_rate": 5.925620807343769e-06,
|
| 2769 |
+
"loss": 2.7589,
|
| 2770 |
+
"step": 370000
|
| 2771 |
+
},
|
| 2772 |
+
{
|
| 2773 |
+
"epoch": 19.95,
|
| 2774 |
+
"grad_norm": 0.3465304970741272,
|
| 2775 |
+
"learning_rate": 2.983405907967518e-06,
|
| 2776 |
+
"loss": 2.7571,
|
| 2777 |
+
"step": 371000
|
| 2778 |
+
},
|
| 2779 |
+
{
|
| 2780 |
+
"epoch": 20.0,
|
| 2781 |
+
"eval_accuracy": 0.41050110032238096,
|
| 2782 |
+
"eval_loss": 3.4041664600372314,
|
| 2783 |
+
"eval_runtime": 154.1848,
|
| 2784 |
+
"eval_samples_per_second": 375.647,
|
| 2785 |
+
"eval_steps_per_second": 5.87,
|
| 2786 |
+
"step": 371880
|
| 2787 |
+
},
|
| 2788 |
+
{
|
| 2789 |
+
"epoch": 20.0,
|
| 2790 |
+
"step": 371880,
|
| 2791 |
+
"total_flos": 1.56681778719744e+18,
|
| 2792 |
+
"train_loss": 3.0269814196563325,
|
| 2793 |
+
"train_runtime": 81274.1105,
|
| 2794 |
+
"train_samples_per_second": 146.417,
|
| 2795 |
+
"train_steps_per_second": 4.576
|
| 2796 |
+
}
|
| 2797 |
+
],
|
| 2798 |
+
"logging_steps": 1000,
|
| 2799 |
+
"max_steps": 371880,
|
| 2800 |
+
"num_input_tokens_seen": 0,
|
| 2801 |
+
"num_train_epochs": 20,
|
| 2802 |
+
"save_steps": 5000,
|
| 2803 |
+
"total_flos": 1.56681778719744e+18,
|
| 2804 |
+
"train_batch_size": 32,
|
| 2805 |
+
"trial_name": null,
|
| 2806 |
+
"trial_params": null
|
| 2807 |
+
}
|