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_dtanns-1e-4
|
| 8 |
-
results:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
---
|
| 10 |
|
| 11 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
@@ -13,7 +25,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 13 |
|
| 14 |
# smolm-autoreg-bpe-counterfactual_babylm_aann_dtanns-1e-4
|
| 15 |
|
| 16 |
-
This model was trained from scratch on
|
| 17 |
It achieves the following results on the evaluation set:
|
| 18 |
- Loss: 3.4122
|
| 19 |
- Accuracy: 0.4058
|
|
|
|
| 1 |
---
|
| 2 |
tags:
|
| 3 |
- generated_from_trainer
|
| 4 |
+
datasets:
|
| 5 |
+
- kanishka/counterfactual_babylm_aann_dtanns
|
| 6 |
metrics:
|
| 7 |
- accuracy
|
| 8 |
model-index:
|
| 9 |
- name: smolm-autoreg-bpe-counterfactual_babylm_aann_dtanns-1e-4
|
| 10 |
+
results:
|
| 11 |
+
- task:
|
| 12 |
+
name: Causal Language Modeling
|
| 13 |
+
type: text-generation
|
| 14 |
+
dataset:
|
| 15 |
+
name: kanishka/counterfactual_babylm_aann_dtanns
|
| 16 |
+
type: kanishka/counterfactual_babylm_aann_dtanns
|
| 17 |
+
metrics:
|
| 18 |
+
- name: Accuracy
|
| 19 |
+
type: accuracy
|
| 20 |
+
value: 0.40580987772461885
|
| 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_dtanns-1e-4
|
| 27 |
|
| 28 |
+
This model was trained from scratch on the kanishka/counterfactual_babylm_aann_dtanns dataset.
|
| 29 |
It achieves the following results on the evaluation set:
|
| 30 |
- Loss: 3.4122
|
| 31 |
- Accuracy: 0.4058
|
all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"eval_accuracy": 0.40580987772461885,
|
| 4 |
+
"eval_loss": 3.4122142791748047,
|
| 5 |
+
"eval_runtime": 153.3537,
|
| 6 |
+
"eval_samples": 57918,
|
| 7 |
+
"eval_samples_per_second": 377.676,
|
| 8 |
+
"eval_steps_per_second": 5.901,
|
| 9 |
+
"perplexity": 30.332334210012274,
|
| 10 |
+
"train_loss": 3.1577374407077676,
|
| 11 |
+
"train_runtime": 81393.9304,
|
| 12 |
+
"train_samples": 595038,
|
| 13 |
+
"train_samples_per_second": 146.212,
|
| 14 |
+
"train_steps_per_second": 4.569
|
| 15 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"eval_accuracy": 0.40580987772461885,
|
| 4 |
+
"eval_loss": 3.4122142791748047,
|
| 5 |
+
"eval_runtime": 153.3537,
|
| 6 |
+
"eval_samples": 57918,
|
| 7 |
+
"eval_samples_per_second": 377.676,
|
| 8 |
+
"eval_steps_per_second": 5.901,
|
| 9 |
+
"perplexity": 30.332334210012274
|
| 10 |
+
}
|
runs/Jul04_14-15-05_phyl-ling-p01.la.utexas.edu/events.out.tfevents.1720202427.phyl-ling-p01.la.utexas.edu.1962779.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1a555dc07decd8a7a36a2574c3ecaac48b470d30272689fbe6d2bf7bf5a796b8
|
| 3 |
+
size 417
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"train_loss": 3.1577374407077676,
|
| 4 |
+
"train_runtime": 81393.9304,
|
| 5 |
+
"train_samples": 595038,
|
| 6 |
+
"train_samples_per_second": 146.212,
|
| 7 |
+
"train_steps_per_second": 4.569
|
| 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": 371900,
|
| 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": 1.089185118675232,
|
| 14 |
+
"learning_rate": 3.125e-06,
|
| 15 |
+
"loss": 7.5412,
|
| 16 |
+
"step": 1000
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.11,
|
| 20 |
+
"grad_norm": 0.7695654630661011,
|
| 21 |
+
"learning_rate": 6.25e-06,
|
| 22 |
+
"loss": 5.8108,
|
| 23 |
+
"step": 2000
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.16,
|
| 27 |
+
"grad_norm": 0.8033701777458191,
|
| 28 |
+
"learning_rate": 9.375000000000001e-06,
|
| 29 |
+
"loss": 5.3841,
|
| 30 |
+
"step": 3000
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.22,
|
| 34 |
+
"grad_norm": 0.8803929686546326,
|
| 35 |
+
"learning_rate": 1.25e-05,
|
| 36 |
+
"loss": 5.1671,
|
| 37 |
+
"step": 4000
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.27,
|
| 41 |
+
"grad_norm": 0.9745013117790222,
|
| 42 |
+
"learning_rate": 1.5625e-05,
|
| 43 |
+
"loss": 5.0134,
|
| 44 |
+
"step": 5000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.32,
|
| 48 |
+
"grad_norm": 1.0176976919174194,
|
| 49 |
+
"learning_rate": 1.8750000000000002e-05,
|
| 50 |
+
"loss": 4.8773,
|
| 51 |
+
"step": 6000
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.38,
|
| 55 |
+
"grad_norm": 1.1663941144943237,
|
| 56 |
+
"learning_rate": 2.1875e-05,
|
| 57 |
+
"loss": 4.7565,
|
| 58 |
+
"step": 7000
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.43,
|
| 62 |
+
"grad_norm": 1.0743473768234253,
|
| 63 |
+
"learning_rate": 2.5e-05,
|
| 64 |
+
"loss": 4.65,
|
| 65 |
+
"step": 8000
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.48,
|
| 69 |
+
"grad_norm": 1.0922328233718872,
|
| 70 |
+
"learning_rate": 2.8125000000000003e-05,
|
| 71 |
+
"loss": 4.5631,
|
| 72 |
+
"step": 9000
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.54,
|
| 76 |
+
"grad_norm": 1.0371665954589844,
|
| 77 |
+
"learning_rate": 3.125e-05,
|
| 78 |
+
"loss": 4.4855,
|
| 79 |
+
"step": 10000
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.59,
|
| 83 |
+
"grad_norm": 1.0929738283157349,
|
| 84 |
+
"learning_rate": 3.4371875e-05,
|
| 85 |
+
"loss": 4.4152,
|
| 86 |
+
"step": 11000
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.65,
|
| 90 |
+
"grad_norm": 1.0536746978759766,
|
| 91 |
+
"learning_rate": 3.7496875e-05,
|
| 92 |
+
"loss": 4.3493,
|
| 93 |
+
"step": 12000
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.7,
|
| 97 |
+
"grad_norm": 1.0247684717178345,
|
| 98 |
+
"learning_rate": 4.061875e-05,
|
| 99 |
+
"loss": 4.2825,
|
| 100 |
+
"step": 13000
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.75,
|
| 104 |
+
"grad_norm": 1.0697104930877686,
|
| 105 |
+
"learning_rate": 4.374375e-05,
|
| 106 |
+
"loss": 4.2326,
|
| 107 |
+
"step": 14000
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.81,
|
| 111 |
+
"grad_norm": 1.010930061340332,
|
| 112 |
+
"learning_rate": 4.686875e-05,
|
| 113 |
+
"loss": 4.1806,
|
| 114 |
+
"step": 15000
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.86,
|
| 118 |
+
"grad_norm": 1.0303874015808105,
|
| 119 |
+
"learning_rate": 4.999375e-05,
|
| 120 |
+
"loss": 4.1356,
|
| 121 |
+
"step": 16000
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.91,
|
| 125 |
+
"grad_norm": 0.9643944501876831,
|
| 126 |
+
"learning_rate": 5.31125e-05,
|
| 127 |
+
"loss": 4.0973,
|
| 128 |
+
"step": 17000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.97,
|
| 132 |
+
"grad_norm": 1.0404605865478516,
|
| 133 |
+
"learning_rate": 5.62375e-05,
|
| 134 |
+
"loss": 4.0526,
|
| 135 |
+
"step": 18000
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 1.0,
|
| 139 |
+
"eval_accuracy": 0.30970782118769014,
|
| 140 |
+
"eval_loss": 4.265280246734619,
|
| 141 |
+
"eval_runtime": 152.7025,
|
| 142 |
+
"eval_samples_per_second": 379.287,
|
| 143 |
+
"eval_steps_per_second": 5.927,
|
| 144 |
+
"step": 18595
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 1.02,
|
| 148 |
+
"grad_norm": 1.0451867580413818,
|
| 149 |
+
"learning_rate": 5.93625e-05,
|
| 150 |
+
"loss": 4.0032,
|
| 151 |
+
"step": 19000
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 1.08,
|
| 155 |
+
"grad_norm": 0.9735515713691711,
|
| 156 |
+
"learning_rate": 6.24875e-05,
|
| 157 |
+
"loss": 3.956,
|
| 158 |
+
"step": 20000
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.13,
|
| 162 |
+
"grad_norm": 1.0315481424331665,
|
| 163 |
+
"learning_rate": 6.56125e-05,
|
| 164 |
+
"loss": 3.9214,
|
| 165 |
+
"step": 21000
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 1.18,
|
| 169 |
+
"grad_norm": 0.9913306832313538,
|
| 170 |
+
"learning_rate": 6.8734375e-05,
|
| 171 |
+
"loss": 3.8926,
|
| 172 |
+
"step": 22000
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 1.24,
|
| 176 |
+
"grad_norm": 1.0101070404052734,
|
| 177 |
+
"learning_rate": 7.185937500000001e-05,
|
| 178 |
+
"loss": 3.8547,
|
| 179 |
+
"step": 23000
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.29,
|
| 183 |
+
"grad_norm": 1.1149094104766846,
|
| 184 |
+
"learning_rate": 7.4984375e-05,
|
| 185 |
+
"loss": 3.8259,
|
| 186 |
+
"step": 24000
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 1.34,
|
| 190 |
+
"grad_norm": 0.912232518196106,
|
| 191 |
+
"learning_rate": 7.810625e-05,
|
| 192 |
+
"loss": 3.7991,
|
| 193 |
+
"step": 25000
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 1.4,
|
| 197 |
+
"grad_norm": 0.9199328422546387,
|
| 198 |
+
"learning_rate": 8.1228125e-05,
|
| 199 |
+
"loss": 3.7719,
|
| 200 |
+
"step": 26000
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 1.45,
|
| 204 |
+
"grad_norm": 0.9232264161109924,
|
| 205 |
+
"learning_rate": 8.4353125e-05,
|
| 206 |
+
"loss": 3.7442,
|
| 207 |
+
"step": 27000
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 1.51,
|
| 211 |
+
"grad_norm": 0.9053943157196045,
|
| 212 |
+
"learning_rate": 8.7478125e-05,
|
| 213 |
+
"loss": 3.7263,
|
| 214 |
+
"step": 28000
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 1.56,
|
| 218 |
+
"grad_norm": 0.8791221976280212,
|
| 219 |
+
"learning_rate": 9.06e-05,
|
| 220 |
+
"loss": 3.7071,
|
| 221 |
+
"step": 29000
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.61,
|
| 225 |
+
"grad_norm": 0.8786524534225464,
|
| 226 |
+
"learning_rate": 9.3725e-05,
|
| 227 |
+
"loss": 3.6924,
|
| 228 |
+
"step": 30000
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 1.67,
|
| 232 |
+
"grad_norm": 0.9131940603256226,
|
| 233 |
+
"learning_rate": 9.6846875e-05,
|
| 234 |
+
"loss": 3.6627,
|
| 235 |
+
"step": 31000
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 1.72,
|
| 239 |
+
"grad_norm": 0.8522538542747498,
|
| 240 |
+
"learning_rate": 9.996875e-05,
|
| 241 |
+
"loss": 3.6484,
|
| 242 |
+
"step": 32000
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.77,
|
| 246 |
+
"grad_norm": 0.8646347522735596,
|
| 247 |
+
"learning_rate": 9.970873786407767e-05,
|
| 248 |
+
"loss": 3.6358,
|
| 249 |
+
"step": 33000
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 1.83,
|
| 253 |
+
"grad_norm": 0.8322715163230896,
|
| 254 |
+
"learning_rate": 9.941453368637835e-05,
|
| 255 |
+
"loss": 3.6135,
|
| 256 |
+
"step": 34000
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 1.88,
|
| 260 |
+
"grad_norm": 0.8093035817146301,
|
| 261 |
+
"learning_rate": 9.912032950867904e-05,
|
| 262 |
+
"loss": 3.5995,
|
| 263 |
+
"step": 35000
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 1.94,
|
| 267 |
+
"grad_norm": 0.8063403964042664,
|
| 268 |
+
"learning_rate": 9.882641953515739e-05,
|
| 269 |
+
"loss": 3.5866,
|
| 270 |
+
"step": 36000
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 1.99,
|
| 274 |
+
"grad_norm": 0.8168238401412964,
|
| 275 |
+
"learning_rate": 9.853221535745807e-05,
|
| 276 |
+
"loss": 3.5753,
|
| 277 |
+
"step": 37000
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 2.0,
|
| 281 |
+
"eval_accuracy": 0.3615619587225358,
|
| 282 |
+
"eval_loss": 3.748784303665161,
|
| 283 |
+
"eval_runtime": 154.4253,
|
| 284 |
+
"eval_samples_per_second": 375.055,
|
| 285 |
+
"eval_steps_per_second": 5.86,
|
| 286 |
+
"step": 37190
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 2.04,
|
| 290 |
+
"grad_norm": 0.8043264150619507,
|
| 291 |
+
"learning_rate": 9.823830538393646e-05,
|
| 292 |
+
"loss": 3.5292,
|
| 293 |
+
"step": 38000
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 2.1,
|
| 297 |
+
"grad_norm": 0.7958018779754639,
|
| 298 |
+
"learning_rate": 9.794410120623714e-05,
|
| 299 |
+
"loss": 3.5159,
|
| 300 |
+
"step": 39000
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 2.15,
|
| 304 |
+
"grad_norm": 0.804074227809906,
|
| 305 |
+
"learning_rate": 9.764989702853781e-05,
|
| 306 |
+
"loss": 3.5049,
|
| 307 |
+
"step": 40000
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 2.2,
|
| 311 |
+
"grad_norm": 0.8031198978424072,
|
| 312 |
+
"learning_rate": 9.735598705501619e-05,
|
| 313 |
+
"loss": 3.4974,
|
| 314 |
+
"step": 41000
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 2.26,
|
| 318 |
+
"grad_norm": 0.8043462038040161,
|
| 319 |
+
"learning_rate": 9.706178287731686e-05,
|
| 320 |
+
"loss": 3.4901,
|
| 321 |
+
"step": 42000
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 2.31,
|
| 325 |
+
"grad_norm": 0.7754331827163696,
|
| 326 |
+
"learning_rate": 9.676757869961754e-05,
|
| 327 |
+
"loss": 3.4794,
|
| 328 |
+
"step": 43000
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 2.37,
|
| 332 |
+
"grad_norm": 0.7999666333198547,
|
| 333 |
+
"learning_rate": 9.647396293027361e-05,
|
| 334 |
+
"loss": 3.4736,
|
| 335 |
+
"step": 44000
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 2.42,
|
| 339 |
+
"grad_norm": 0.819854199886322,
|
| 340 |
+
"learning_rate": 9.61797587525743e-05,
|
| 341 |
+
"loss": 3.4656,
|
| 342 |
+
"step": 45000
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 2.47,
|
| 346 |
+
"grad_norm": 0.7657862305641174,
|
| 347 |
+
"learning_rate": 9.588555457487498e-05,
|
| 348 |
+
"loss": 3.4553,
|
| 349 |
+
"step": 46000
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 2.53,
|
| 353 |
+
"grad_norm": 0.7784642577171326,
|
| 354 |
+
"learning_rate": 9.559135039717564e-05,
|
| 355 |
+
"loss": 3.4457,
|
| 356 |
+
"step": 47000
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.58,
|
| 360 |
+
"grad_norm": 0.7481061816215515,
|
| 361 |
+
"learning_rate": 9.529714621947633e-05,
|
| 362 |
+
"loss": 3.4389,
|
| 363 |
+
"step": 48000
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 2.64,
|
| 367 |
+
"grad_norm": 0.7543549537658691,
|
| 368 |
+
"learning_rate": 9.50032362459547e-05,
|
| 369 |
+
"loss": 3.4357,
|
| 370 |
+
"step": 49000
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 2.69,
|
| 374 |
+
"grad_norm": 0.8025472164154053,
|
| 375 |
+
"learning_rate": 9.470903206825538e-05,
|
| 376 |
+
"loss": 3.4243,
|
| 377 |
+
"step": 50000
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 2.74,
|
| 381 |
+
"grad_norm": 0.729120671749115,
|
| 382 |
+
"learning_rate": 9.441482789055604e-05,
|
| 383 |
+
"loss": 3.4173,
|
| 384 |
+
"step": 51000
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 2.8,
|
| 388 |
+
"grad_norm": 0.7592692971229553,
|
| 389 |
+
"learning_rate": 9.412091791703443e-05,
|
| 390 |
+
"loss": 3.4168,
|
| 391 |
+
"step": 52000
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 2.85,
|
| 395 |
+
"grad_norm": 0.7497263550758362,
|
| 396 |
+
"learning_rate": 9.382671373933511e-05,
|
| 397 |
+
"loss": 3.4062,
|
| 398 |
+
"step": 53000
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 2.9,
|
| 402 |
+
"grad_norm": 0.7167881727218628,
|
| 403 |
+
"learning_rate": 9.353250956163578e-05,
|
| 404 |
+
"loss": 3.3978,
|
| 405 |
+
"step": 54000
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 2.96,
|
| 409 |
+
"grad_norm": 0.7573443651199341,
|
| 410 |
+
"learning_rate": 9.323859958811415e-05,
|
| 411 |
+
"loss": 3.3983,
|
| 412 |
+
"step": 55000
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 3.0,
|
| 416 |
+
"eval_accuracy": 0.3787864313198467,
|
| 417 |
+
"eval_loss": 3.594850540161133,
|
| 418 |
+
"eval_runtime": 153.9373,
|
| 419 |
+
"eval_samples_per_second": 376.244,
|
| 420 |
+
"eval_steps_per_second": 5.879,
|
| 421 |
+
"step": 55785
|
| 422 |
+
},
|
| 423 |
+
{
|
| 424 |
+
"epoch": 3.01,
|
| 425 |
+
"grad_norm": 0.763841450214386,
|
| 426 |
+
"learning_rate": 9.294439541041483e-05,
|
| 427 |
+
"loss": 3.3722,
|
| 428 |
+
"step": 56000
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 3.07,
|
| 432 |
+
"grad_norm": 0.7948548197746277,
|
| 433 |
+
"learning_rate": 9.265019123271551e-05,
|
| 434 |
+
"loss": 3.3362,
|
| 435 |
+
"step": 57000
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 3.12,
|
| 439 |
+
"grad_norm": 0.7683813571929932,
|
| 440 |
+
"learning_rate": 9.23562812591939e-05,
|
| 441 |
+
"loss": 3.3373,
|
| 442 |
+
"step": 58000
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 3.17,
|
| 446 |
+
"grad_norm": 0.752254068851471,
|
| 447 |
+
"learning_rate": 9.206237128567227e-05,
|
| 448 |
+
"loss": 3.3367,
|
| 449 |
+
"step": 59000
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 3.23,
|
| 453 |
+
"grad_norm": 0.7324814200401306,
|
| 454 |
+
"learning_rate": 9.176816710797293e-05,
|
| 455 |
+
"loss": 3.3346,
|
| 456 |
+
"step": 60000
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 3.28,
|
| 460 |
+
"grad_norm": 0.7799642086029053,
|
| 461 |
+
"learning_rate": 9.147396293027361e-05,
|
| 462 |
+
"loss": 3.3315,
|
| 463 |
+
"step": 61000
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 3.33,
|
| 467 |
+
"grad_norm": 0.7740771174430847,
|
| 468 |
+
"learning_rate": 9.11797587525743e-05,
|
| 469 |
+
"loss": 3.3299,
|
| 470 |
+
"step": 62000
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 3.39,
|
| 474 |
+
"grad_norm": 0.7809312343597412,
|
| 475 |
+
"learning_rate": 9.088584877905267e-05,
|
| 476 |
+
"loss": 3.3258,
|
| 477 |
+
"step": 63000
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 3.44,
|
| 481 |
+
"grad_norm": 0.7324674129486084,
|
| 482 |
+
"learning_rate": 9.059164460135335e-05,
|
| 483 |
+
"loss": 3.3217,
|
| 484 |
+
"step": 64000
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 3.5,
|
| 488 |
+
"grad_norm": 0.7670503854751587,
|
| 489 |
+
"learning_rate": 9.029744042365403e-05,
|
| 490 |
+
"loss": 3.322,
|
| 491 |
+
"step": 65000
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 3.55,
|
| 495 |
+
"grad_norm": 0.7658038139343262,
|
| 496 |
+
"learning_rate": 9.00035304501324e-05,
|
| 497 |
+
"loss": 3.3149,
|
| 498 |
+
"step": 66000
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 3.6,
|
| 502 |
+
"grad_norm": 0.7420821189880371,
|
| 503 |
+
"learning_rate": 8.970932627243307e-05,
|
| 504 |
+
"loss": 3.3142,
|
| 505 |
+
"step": 67000
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 3.66,
|
| 509 |
+
"grad_norm": 0.7639199495315552,
|
| 510 |
+
"learning_rate": 8.941512209473375e-05,
|
| 511 |
+
"loss": 3.3054,
|
| 512 |
+
"step": 68000
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 3.71,
|
| 516 |
+
"grad_norm": 0.7637389302253723,
|
| 517 |
+
"learning_rate": 8.912091791703443e-05,
|
| 518 |
+
"loss": 3.3052,
|
| 519 |
+
"step": 69000
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 3.76,
|
| 523 |
+
"grad_norm": 0.7303109765052795,
|
| 524 |
+
"learning_rate": 8.88270079435128e-05,
|
| 525 |
+
"loss": 3.3012,
|
| 526 |
+
"step": 70000
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 3.82,
|
| 530 |
+
"grad_norm": 0.7220088839530945,
|
| 531 |
+
"learning_rate": 8.853280376581348e-05,
|
| 532 |
+
"loss": 3.3067,
|
| 533 |
+
"step": 71000
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 3.87,
|
| 537 |
+
"grad_norm": 0.7170850038528442,
|
| 538 |
+
"learning_rate": 8.823859958811416e-05,
|
| 539 |
+
"loss": 3.2974,
|
| 540 |
+
"step": 72000
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 3.93,
|
| 544 |
+
"grad_norm": 0.7323712706565857,
|
| 545 |
+
"learning_rate": 8.794439541041484e-05,
|
| 546 |
+
"loss": 3.2987,
|
| 547 |
+
"step": 73000
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 3.98,
|
| 551 |
+
"grad_norm": 0.7505171298980713,
|
| 552 |
+
"learning_rate": 8.76504854368932e-05,
|
| 553 |
+
"loss": 3.2911,
|
| 554 |
+
"step": 74000
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 4.0,
|
| 558 |
+
"eval_accuracy": 0.3863417930327635,
|
| 559 |
+
"eval_loss": 3.541649580001831,
|
| 560 |
+
"eval_runtime": 153.6538,
|
| 561 |
+
"eval_samples_per_second": 376.938,
|
| 562 |
+
"eval_steps_per_second": 5.89,
|
| 563 |
+
"step": 74380
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 4.03,
|
| 567 |
+
"grad_norm": 0.7474557161331177,
|
| 568 |
+
"learning_rate": 8.735628125919388e-05,
|
| 569 |
+
"loss": 3.2588,
|
| 570 |
+
"step": 75000
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 4.09,
|
| 574 |
+
"grad_norm": 0.7711236476898193,
|
| 575 |
+
"learning_rate": 8.706207708149456e-05,
|
| 576 |
+
"loss": 3.241,
|
| 577 |
+
"step": 76000
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 4.14,
|
| 581 |
+
"grad_norm": 0.7377588152885437,
|
| 582 |
+
"learning_rate": 8.676787290379523e-05,
|
| 583 |
+
"loss": 3.2395,
|
| 584 |
+
"step": 77000
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 4.19,
|
| 588 |
+
"grad_norm": 0.7175911664962769,
|
| 589 |
+
"learning_rate": 8.647396293027361e-05,
|
| 590 |
+
"loss": 3.2375,
|
| 591 |
+
"step": 78000
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 4.25,
|
| 595 |
+
"grad_norm": 0.739957869052887,
|
| 596 |
+
"learning_rate": 8.618005295675198e-05,
|
| 597 |
+
"loss": 3.2379,
|
| 598 |
+
"step": 79000
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 4.3,
|
| 602 |
+
"grad_norm": 0.7332589626312256,
|
| 603 |
+
"learning_rate": 8.588584877905267e-05,
|
| 604 |
+
"loss": 3.2408,
|
| 605 |
+
"step": 80000
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 4.36,
|
| 609 |
+
"grad_norm": 0.7471742630004883,
|
| 610 |
+
"learning_rate": 8.559164460135333e-05,
|
| 611 |
+
"loss": 3.2421,
|
| 612 |
+
"step": 81000
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 4.41,
|
| 616 |
+
"grad_norm": 0.7418613433837891,
|
| 617 |
+
"learning_rate": 8.529773462783172e-05,
|
| 618 |
+
"loss": 3.2391,
|
| 619 |
+
"step": 82000
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 4.46,
|
| 623 |
+
"grad_norm": 0.719301700592041,
|
| 624 |
+
"learning_rate": 8.50035304501324e-05,
|
| 625 |
+
"loss": 3.2396,
|
| 626 |
+
"step": 83000
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 4.52,
|
| 630 |
+
"grad_norm": 0.7350900173187256,
|
| 631 |
+
"learning_rate": 8.470932627243308e-05,
|
| 632 |
+
"loss": 3.2334,
|
| 633 |
+
"step": 84000
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 4.57,
|
| 637 |
+
"grad_norm": 0.7193138003349304,
|
| 638 |
+
"learning_rate": 8.441571050308915e-05,
|
| 639 |
+
"loss": 3.2365,
|
| 640 |
+
"step": 85000
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 4.62,
|
| 644 |
+
"grad_norm": 0.7358419895172119,
|
| 645 |
+
"learning_rate": 8.412150632538982e-05,
|
| 646 |
+
"loss": 3.2275,
|
| 647 |
+
"step": 86000
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 4.68,
|
| 651 |
+
"grad_norm": 0.7091595530509949,
|
| 652 |
+
"learning_rate": 8.38273021476905e-05,
|
| 653 |
+
"loss": 3.2346,
|
| 654 |
+
"step": 87000
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 4.73,
|
| 658 |
+
"grad_norm": 0.7487188577651978,
|
| 659 |
+
"learning_rate": 8.353309796999118e-05,
|
| 660 |
+
"loss": 3.2315,
|
| 661 |
+
"step": 88000
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 4.79,
|
| 665 |
+
"grad_norm": 0.7331488728523254,
|
| 666 |
+
"learning_rate": 8.323918799646955e-05,
|
| 667 |
+
"loss": 3.2307,
|
| 668 |
+
"step": 89000
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 4.84,
|
| 672 |
+
"grad_norm": 0.7424489855766296,
|
| 673 |
+
"learning_rate": 8.294498381877022e-05,
|
| 674 |
+
"loss": 3.2303,
|
| 675 |
+
"step": 90000
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 4.89,
|
| 679 |
+
"grad_norm": 0.7525271773338318,
|
| 680 |
+
"learning_rate": 8.26507796410709e-05,
|
| 681 |
+
"loss": 3.2294,
|
| 682 |
+
"step": 91000
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 4.95,
|
| 686 |
+
"grad_norm": 0.7213485240936279,
|
| 687 |
+
"learning_rate": 8.235657546337158e-05,
|
| 688 |
+
"loss": 3.2282,
|
| 689 |
+
"step": 92000
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 5.0,
|
| 693 |
+
"eval_accuracy": 0.39235109398595597,
|
| 694 |
+
"eval_loss": 3.4672956466674805,
|
| 695 |
+
"eval_runtime": 154.0738,
|
| 696 |
+
"eval_samples_per_second": 375.911,
|
| 697 |
+
"eval_steps_per_second": 5.874,
|
| 698 |
+
"step": 92975
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 5.0,
|
| 702 |
+
"grad_norm": 0.7489692568778992,
|
| 703 |
+
"learning_rate": 8.206237128567225e-05,
|
| 704 |
+
"loss": 3.2178,
|
| 705 |
+
"step": 93000
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 5.06,
|
| 709 |
+
"grad_norm": 0.7288196086883545,
|
| 710 |
+
"learning_rate": 8.176816710797293e-05,
|
| 711 |
+
"loss": 3.1657,
|
| 712 |
+
"step": 94000
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 5.11,
|
| 716 |
+
"grad_norm": 0.7864417433738708,
|
| 717 |
+
"learning_rate": 8.147425713445132e-05,
|
| 718 |
+
"loss": 3.1688,
|
| 719 |
+
"step": 95000
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 5.16,
|
| 723 |
+
"grad_norm": 0.7475498914718628,
|
| 724 |
+
"learning_rate": 8.1180052956752e-05,
|
| 725 |
+
"loss": 3.1716,
|
| 726 |
+
"step": 96000
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 5.22,
|
| 730 |
+
"grad_norm": 0.7414774894714355,
|
| 731 |
+
"learning_rate": 8.088584877905267e-05,
|
| 732 |
+
"loss": 3.175,
|
| 733 |
+
"step": 97000
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 5.27,
|
| 737 |
+
"grad_norm": 0.7480049133300781,
|
| 738 |
+
"learning_rate": 8.059164460135335e-05,
|
| 739 |
+
"loss": 3.1784,
|
| 740 |
+
"step": 98000
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 5.32,
|
| 744 |
+
"grad_norm": 0.753226637840271,
|
| 745 |
+
"learning_rate": 8.029773462783172e-05,
|
| 746 |
+
"loss": 3.1763,
|
| 747 |
+
"step": 99000
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 5.38,
|
| 751 |
+
"grad_norm": 0.7331398725509644,
|
| 752 |
+
"learning_rate": 8.00038246543101e-05,
|
| 753 |
+
"loss": 3.1789,
|
| 754 |
+
"step": 100000
|
| 755 |
+
},
|
| 756 |
+
{
|
| 757 |
+
"epoch": 5.43,
|
| 758 |
+
"grad_norm": 0.7280023694038391,
|
| 759 |
+
"learning_rate": 7.970962047661077e-05,
|
| 760 |
+
"loss": 3.1787,
|
| 761 |
+
"step": 101000
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 5.49,
|
| 765 |
+
"grad_norm": 0.7298089861869812,
|
| 766 |
+
"learning_rate": 7.941541629891145e-05,
|
| 767 |
+
"loss": 3.1784,
|
| 768 |
+
"step": 102000
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 5.54,
|
| 772 |
+
"grad_norm": 0.7451596856117249,
|
| 773 |
+
"learning_rate": 7.912150632538982e-05,
|
| 774 |
+
"loss": 3.1801,
|
| 775 |
+
"step": 103000
|
| 776 |
+
},
|
| 777 |
+
{
|
| 778 |
+
"epoch": 5.59,
|
| 779 |
+
"grad_norm": 0.7099459767341614,
|
| 780 |
+
"learning_rate": 7.882730214769049e-05,
|
| 781 |
+
"loss": 3.1799,
|
| 782 |
+
"step": 104000
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 5.65,
|
| 786 |
+
"grad_norm": 0.755178689956665,
|
| 787 |
+
"learning_rate": 7.853309796999117e-05,
|
| 788 |
+
"loss": 3.1799,
|
| 789 |
+
"step": 105000
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 5.7,
|
| 793 |
+
"grad_norm": 0.7444600462913513,
|
| 794 |
+
"learning_rate": 7.823889379229185e-05,
|
| 795 |
+
"loss": 3.1756,
|
| 796 |
+
"step": 106000
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"epoch": 5.75,
|
| 800 |
+
"grad_norm": 0.7303113341331482,
|
| 801 |
+
"learning_rate": 7.794498381877024e-05,
|
| 802 |
+
"loss": 3.1766,
|
| 803 |
+
"step": 107000
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 5.81,
|
| 807 |
+
"grad_norm": 0.7302135229110718,
|
| 808 |
+
"learning_rate": 7.76507796410709e-05,
|
| 809 |
+
"loss": 3.1782,
|
| 810 |
+
"step": 108000
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 5.86,
|
| 814 |
+
"grad_norm": 0.733063817024231,
|
| 815 |
+
"learning_rate": 7.735686966754927e-05,
|
| 816 |
+
"loss": 3.1744,
|
| 817 |
+
"step": 109000
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 5.92,
|
| 821 |
+
"grad_norm": 0.7234317064285278,
|
| 822 |
+
"learning_rate": 7.706266548984996e-05,
|
| 823 |
+
"loss": 3.176,
|
| 824 |
+
"step": 110000
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 5.97,
|
| 828 |
+
"grad_norm": 0.751522421836853,
|
| 829 |
+
"learning_rate": 7.676846131215064e-05,
|
| 830 |
+
"loss": 3.1707,
|
| 831 |
+
"step": 111000
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 6.0,
|
| 835 |
+
"eval_accuracy": 0.39479154951794854,
|
| 836 |
+
"eval_loss": 3.4627251625061035,
|
| 837 |
+
"eval_runtime": 154.0269,
|
| 838 |
+
"eval_samples_per_second": 376.025,
|
| 839 |
+
"eval_steps_per_second": 5.876,
|
| 840 |
+
"step": 111570
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 6.02,
|
| 844 |
+
"grad_norm": 0.7434809803962708,
|
| 845 |
+
"learning_rate": 7.647455133862901e-05,
|
| 846 |
+
"loss": 3.1472,
|
| 847 |
+
"step": 112000
|
| 848 |
+
},
|
| 849 |
+
{
|
| 850 |
+
"epoch": 6.08,
|
| 851 |
+
"grad_norm": 0.7743415832519531,
|
| 852 |
+
"learning_rate": 7.618034716092969e-05,
|
| 853 |
+
"loss": 3.1205,
|
| 854 |
+
"step": 113000
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 6.13,
|
| 858 |
+
"grad_norm": 0.7340639233589172,
|
| 859 |
+
"learning_rate": 7.588614298323037e-05,
|
| 860 |
+
"loss": 3.1193,
|
| 861 |
+
"step": 114000
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 6.18,
|
| 865 |
+
"grad_norm": 0.7659904956817627,
|
| 866 |
+
"learning_rate": 7.559193880553104e-05,
|
| 867 |
+
"loss": 3.1192,
|
| 868 |
+
"step": 115000
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 6.24,
|
| 872 |
+
"grad_norm": 0.7478824257850647,
|
| 873 |
+
"learning_rate": 7.529802883200942e-05,
|
| 874 |
+
"loss": 3.1262,
|
| 875 |
+
"step": 116000
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 6.29,
|
| 879 |
+
"grad_norm": 0.8000442385673523,
|
| 880 |
+
"learning_rate": 7.500411885848779e-05,
|
| 881 |
+
"loss": 3.1281,
|
| 882 |
+
"step": 117000
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 6.35,
|
| 886 |
+
"grad_norm": 0.7282128930091858,
|
| 887 |
+
"learning_rate": 7.470991468078847e-05,
|
| 888 |
+
"loss": 3.1288,
|
| 889 |
+
"step": 118000
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 6.4,
|
| 893 |
+
"grad_norm": 0.7605059742927551,
|
| 894 |
+
"learning_rate": 7.441571050308914e-05,
|
| 895 |
+
"loss": 3.1297,
|
| 896 |
+
"step": 119000
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 6.45,
|
| 900 |
+
"grad_norm": 0.7170723676681519,
|
| 901 |
+
"learning_rate": 7.412150632538982e-05,
|
| 902 |
+
"loss": 3.1322,
|
| 903 |
+
"step": 120000
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 6.51,
|
| 907 |
+
"grad_norm": 0.7601571679115295,
|
| 908 |
+
"learning_rate": 7.38273021476905e-05,
|
| 909 |
+
"loss": 3.1317,
|
| 910 |
+
"step": 121000
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 6.56,
|
| 914 |
+
"grad_norm": 0.7484211921691895,
|
| 915 |
+
"learning_rate": 7.353339217416887e-05,
|
| 916 |
+
"loss": 3.1347,
|
| 917 |
+
"step": 122000
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 6.61,
|
| 921 |
+
"grad_norm": 0.7147959470748901,
|
| 922 |
+
"learning_rate": 7.323918799646956e-05,
|
| 923 |
+
"loss": 3.137,
|
| 924 |
+
"step": 123000
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 6.67,
|
| 928 |
+
"grad_norm": 0.754641592502594,
|
| 929 |
+
"learning_rate": 7.294527802294793e-05,
|
| 930 |
+
"loss": 3.1313,
|
| 931 |
+
"step": 124000
|
| 932 |
+
},
|
| 933 |
+
{
|
| 934 |
+
"epoch": 6.72,
|
| 935 |
+
"grad_norm": 0.7599353194236755,
|
| 936 |
+
"learning_rate": 7.265107384524861e-05,
|
| 937 |
+
"loss": 3.1307,
|
| 938 |
+
"step": 125000
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 6.78,
|
| 942 |
+
"grad_norm": 0.7601804137229919,
|
| 943 |
+
"learning_rate": 7.235686966754929e-05,
|
| 944 |
+
"loss": 3.1334,
|
| 945 |
+
"step": 126000
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 6.83,
|
| 949 |
+
"grad_norm": 0.7083086967468262,
|
| 950 |
+
"learning_rate": 7.206266548984996e-05,
|
| 951 |
+
"loss": 3.1338,
|
| 952 |
+
"step": 127000
|
| 953 |
+
},
|
| 954 |
+
{
|
| 955 |
+
"epoch": 6.88,
|
| 956 |
+
"grad_norm": 0.7496170997619629,
|
| 957 |
+
"learning_rate": 7.176846131215064e-05,
|
| 958 |
+
"loss": 3.1326,
|
| 959 |
+
"step": 128000
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 6.94,
|
| 963 |
+
"grad_norm": 0.7311526536941528,
|
| 964 |
+
"learning_rate": 7.147455133862901e-05,
|
| 965 |
+
"loss": 3.1331,
|
| 966 |
+
"step": 129000
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 6.99,
|
| 970 |
+
"grad_norm": 0.7198259234428406,
|
| 971 |
+
"learning_rate": 7.118034716092969e-05,
|
| 972 |
+
"loss": 3.1335,
|
| 973 |
+
"step": 130000
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 7.0,
|
| 977 |
+
"eval_accuracy": 0.39886237929567137,
|
| 978 |
+
"eval_loss": 3.426504135131836,
|
| 979 |
+
"eval_runtime": 154.0887,
|
| 980 |
+
"eval_samples_per_second": 375.874,
|
| 981 |
+
"eval_steps_per_second": 5.873,
|
| 982 |
+
"step": 130165
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 7.04,
|
| 986 |
+
"grad_norm": 0.769108235836029,
|
| 987 |
+
"learning_rate": 7.088614298323037e-05,
|
| 988 |
+
"loss": 3.0785,
|
| 989 |
+
"step": 131000
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 7.1,
|
| 993 |
+
"grad_norm": 0.7577947974205017,
|
| 994 |
+
"learning_rate": 7.059193880553104e-05,
|
| 995 |
+
"loss": 3.0786,
|
| 996 |
+
"step": 132000
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 7.15,
|
| 1000 |
+
"grad_norm": 0.7779477834701538,
|
| 1001 |
+
"learning_rate": 7.029802883200942e-05,
|
| 1002 |
+
"loss": 3.079,
|
| 1003 |
+
"step": 133000
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 7.21,
|
| 1007 |
+
"grad_norm": 0.7599722146987915,
|
| 1008 |
+
"learning_rate": 7.000411885848779e-05,
|
| 1009 |
+
"loss": 3.0878,
|
| 1010 |
+
"step": 134000
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 7.26,
|
| 1014 |
+
"grad_norm": 0.7357174158096313,
|
| 1015 |
+
"learning_rate": 6.970991468078847e-05,
|
| 1016 |
+
"loss": 3.0888,
|
| 1017 |
+
"step": 135000
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 7.31,
|
| 1021 |
+
"grad_norm": 0.7348117828369141,
|
| 1022 |
+
"learning_rate": 6.941600470726684e-05,
|
| 1023 |
+
"loss": 3.0882,
|
| 1024 |
+
"step": 136000
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 7.37,
|
| 1028 |
+
"grad_norm": 0.7608048915863037,
|
| 1029 |
+
"learning_rate": 6.912180052956753e-05,
|
| 1030 |
+
"loss": 3.0924,
|
| 1031 |
+
"step": 137000
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 7.42,
|
| 1035 |
+
"grad_norm": 0.7683416604995728,
|
| 1036 |
+
"learning_rate": 6.88275963518682e-05,
|
| 1037 |
+
"loss": 3.0926,
|
| 1038 |
+
"step": 138000
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 7.48,
|
| 1042 |
+
"grad_norm": 0.7787294387817383,
|
| 1043 |
+
"learning_rate": 6.853368637834658e-05,
|
| 1044 |
+
"loss": 3.0963,
|
| 1045 |
+
"step": 139000
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 7.53,
|
| 1049 |
+
"grad_norm": 0.7591497898101807,
|
| 1050 |
+
"learning_rate": 6.823948220064726e-05,
|
| 1051 |
+
"loss": 3.0984,
|
| 1052 |
+
"step": 140000
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 7.58,
|
| 1056 |
+
"grad_norm": 0.7495740652084351,
|
| 1057 |
+
"learning_rate": 6.794527802294794e-05,
|
| 1058 |
+
"loss": 3.093,
|
| 1059 |
+
"step": 141000
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 7.64,
|
| 1063 |
+
"grad_norm": 0.7411355972290039,
|
| 1064 |
+
"learning_rate": 6.76513680494263e-05,
|
| 1065 |
+
"loss": 3.096,
|
| 1066 |
+
"step": 142000
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 7.69,
|
| 1070 |
+
"grad_norm": 0.7169022560119629,
|
| 1071 |
+
"learning_rate": 6.735716387172698e-05,
|
| 1072 |
+
"loss": 3.0967,
|
| 1073 |
+
"step": 143000
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 7.74,
|
| 1077 |
+
"grad_norm": 0.7618188261985779,
|
| 1078 |
+
"learning_rate": 6.706325389820536e-05,
|
| 1079 |
+
"loss": 3.094,
|
| 1080 |
+
"step": 144000
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 7.8,
|
| 1084 |
+
"grad_norm": 0.7514373064041138,
|
| 1085 |
+
"learning_rate": 6.676904972050604e-05,
|
| 1086 |
+
"loss": 3.0972,
|
| 1087 |
+
"step": 145000
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 7.85,
|
| 1091 |
+
"grad_norm": 0.7369088530540466,
|
| 1092 |
+
"learning_rate": 6.64751397469844e-05,
|
| 1093 |
+
"loss": 3.0963,
|
| 1094 |
+
"step": 146000
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 7.91,
|
| 1098 |
+
"grad_norm": 0.7700458765029907,
|
| 1099 |
+
"learning_rate": 6.618093556928508e-05,
|
| 1100 |
+
"loss": 3.0932,
|
| 1101 |
+
"step": 147000
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 7.96,
|
| 1105 |
+
"grad_norm": 0.727824330329895,
|
| 1106 |
+
"learning_rate": 6.588673139158576e-05,
|
| 1107 |
+
"loss": 3.1003,
|
| 1108 |
+
"step": 148000
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 8.0,
|
| 1112 |
+
"eval_accuracy": 0.400063661720589,
|
| 1113 |
+
"eval_loss": 3.413517713546753,
|
| 1114 |
+
"eval_runtime": 154.023,
|
| 1115 |
+
"eval_samples_per_second": 376.035,
|
| 1116 |
+
"eval_steps_per_second": 5.876,
|
| 1117 |
+
"step": 148760
|
| 1118 |
+
},
|
| 1119 |
+
{
|
| 1120 |
+
"epoch": 8.01,
|
| 1121 |
+
"grad_norm": 0.7588272094726562,
|
| 1122 |
+
"learning_rate": 6.559252721388644e-05,
|
| 1123 |
+
"loss": 3.0802,
|
| 1124 |
+
"step": 149000
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 8.07,
|
| 1128 |
+
"grad_norm": 0.7554176449775696,
|
| 1129 |
+
"learning_rate": 6.529832303618711e-05,
|
| 1130 |
+
"loss": 3.0383,
|
| 1131 |
+
"step": 150000
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 8.12,
|
| 1135 |
+
"grad_norm": 0.7648897171020508,
|
| 1136 |
+
"learning_rate": 6.50044130626655e-05,
|
| 1137 |
+
"loss": 3.0442,
|
| 1138 |
+
"step": 151000
|
| 1139 |
+
},
|
| 1140 |
+
{
|
| 1141 |
+
"epoch": 8.17,
|
| 1142 |
+
"grad_norm": 0.7691752314567566,
|
| 1143 |
+
"learning_rate": 6.471050308914387e-05,
|
| 1144 |
+
"loss": 3.0502,
|
| 1145 |
+
"step": 152000
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 8.23,
|
| 1149 |
+
"grad_norm": 0.7566884756088257,
|
| 1150 |
+
"learning_rate": 6.441629891144455e-05,
|
| 1151 |
+
"loss": 3.0523,
|
| 1152 |
+
"step": 153000
|
| 1153 |
+
},
|
| 1154 |
+
{
|
| 1155 |
+
"epoch": 8.28,
|
| 1156 |
+
"grad_norm": 0.7847636938095093,
|
| 1157 |
+
"learning_rate": 6.412209473374522e-05,
|
| 1158 |
+
"loss": 3.0538,
|
| 1159 |
+
"step": 154000
|
| 1160 |
+
},
|
| 1161 |
+
{
|
| 1162 |
+
"epoch": 8.34,
|
| 1163 |
+
"grad_norm": 0.7558983564376831,
|
| 1164 |
+
"learning_rate": 6.38278905560459e-05,
|
| 1165 |
+
"loss": 3.0585,
|
| 1166 |
+
"step": 155000
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 8.39,
|
| 1170 |
+
"grad_norm": 0.7530951499938965,
|
| 1171 |
+
"learning_rate": 6.353398058252428e-05,
|
| 1172 |
+
"loss": 3.0571,
|
| 1173 |
+
"step": 156000
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 8.44,
|
| 1177 |
+
"grad_norm": 0.7423450350761414,
|
| 1178 |
+
"learning_rate": 6.323977640482495e-05,
|
| 1179 |
+
"loss": 3.0542,
|
| 1180 |
+
"step": 157000
|
| 1181 |
+
},
|
| 1182 |
+
{
|
| 1183 |
+
"epoch": 8.5,
|
| 1184 |
+
"grad_norm": 0.7695331573486328,
|
| 1185 |
+
"learning_rate": 6.294557222712563e-05,
|
| 1186 |
+
"loss": 3.0607,
|
| 1187 |
+
"step": 158000
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 8.55,
|
| 1191 |
+
"grad_norm": 0.7808562517166138,
|
| 1192 |
+
"learning_rate": 6.2651662253604e-05,
|
| 1193 |
+
"loss": 3.0603,
|
| 1194 |
+
"step": 159000
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 8.6,
|
| 1198 |
+
"grad_norm": 0.710171639919281,
|
| 1199 |
+
"learning_rate": 6.235775228008238e-05,
|
| 1200 |
+
"loss": 3.0643,
|
| 1201 |
+
"step": 160000
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 8.66,
|
| 1205 |
+
"grad_norm": 0.7685097455978394,
|
| 1206 |
+
"learning_rate": 6.206384230656075e-05,
|
| 1207 |
+
"loss": 3.0632,
|
| 1208 |
+
"step": 161000
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 8.71,
|
| 1212 |
+
"grad_norm": 0.7420626878738403,
|
| 1213 |
+
"learning_rate": 6.176963812886142e-05,
|
| 1214 |
+
"loss": 3.0651,
|
| 1215 |
+
"step": 162000
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 8.77,
|
| 1219 |
+
"grad_norm": 0.7664393186569214,
|
| 1220 |
+
"learning_rate": 6.14754339511621e-05,
|
| 1221 |
+
"loss": 3.0685,
|
| 1222 |
+
"step": 163000
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"epoch": 8.82,
|
| 1226 |
+
"grad_norm": 0.7250008583068848,
|
| 1227 |
+
"learning_rate": 6.118122977346278e-05,
|
| 1228 |
+
"loss": 3.0656,
|
| 1229 |
+
"step": 164000
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 8.87,
|
| 1233 |
+
"grad_norm": 0.7701383233070374,
|
| 1234 |
+
"learning_rate": 6.088702559576346e-05,
|
| 1235 |
+
"loss": 3.0684,
|
| 1236 |
+
"step": 165000
|
| 1237 |
+
},
|
| 1238 |
+
{
|
| 1239 |
+
"epoch": 8.93,
|
| 1240 |
+
"grad_norm": 0.7626206874847412,
|
| 1241 |
+
"learning_rate": 6.059311562224184e-05,
|
| 1242 |
+
"loss": 3.0652,
|
| 1243 |
+
"step": 166000
|
| 1244 |
+
},
|
| 1245 |
+
{
|
| 1246 |
+
"epoch": 8.98,
|
| 1247 |
+
"grad_norm": 0.7680991291999817,
|
| 1248 |
+
"learning_rate": 6.029891144454252e-05,
|
| 1249 |
+
"loss": 3.0686,
|
| 1250 |
+
"step": 167000
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 9.0,
|
| 1254 |
+
"eval_accuracy": 0.40090236256465095,
|
| 1255 |
+
"eval_loss": 3.401263475418091,
|
| 1256 |
+
"eval_runtime": 154.0702,
|
| 1257 |
+
"eval_samples_per_second": 375.92,
|
| 1258 |
+
"eval_steps_per_second": 5.874,
|
| 1259 |
+
"step": 167355
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 9.03,
|
| 1263 |
+
"grad_norm": 0.7986507415771484,
|
| 1264 |
+
"learning_rate": 6.00047072668432e-05,
|
| 1265 |
+
"loss": 3.0289,
|
| 1266 |
+
"step": 168000
|
| 1267 |
+
},
|
| 1268 |
+
{
|
| 1269 |
+
"epoch": 9.09,
|
| 1270 |
+
"grad_norm": 0.8043704032897949,
|
| 1271 |
+
"learning_rate": 5.971050308914387e-05,
|
| 1272 |
+
"loss": 3.0153,
|
| 1273 |
+
"step": 169000
|
| 1274 |
+
},
|
| 1275 |
+
{
|
| 1276 |
+
"epoch": 9.14,
|
| 1277 |
+
"grad_norm": 0.7941464781761169,
|
| 1278 |
+
"learning_rate": 5.9416593115622244e-05,
|
| 1279 |
+
"loss": 3.0172,
|
| 1280 |
+
"step": 170000
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 9.2,
|
| 1284 |
+
"grad_norm": 0.7718305587768555,
|
| 1285 |
+
"learning_rate": 5.9122388937922925e-05,
|
| 1286 |
+
"loss": 3.0171,
|
| 1287 |
+
"step": 171000
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 9.25,
|
| 1291 |
+
"grad_norm": 0.7973867654800415,
|
| 1292 |
+
"learning_rate": 5.88284789644013e-05,
|
| 1293 |
+
"loss": 3.0233,
|
| 1294 |
+
"step": 172000
|
| 1295 |
+
},
|
| 1296 |
+
{
|
| 1297 |
+
"epoch": 9.3,
|
| 1298 |
+
"grad_norm": 0.8227545022964478,
|
| 1299 |
+
"learning_rate": 5.8534568990879666e-05,
|
| 1300 |
+
"loss": 3.0247,
|
| 1301 |
+
"step": 173000
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 9.36,
|
| 1305 |
+
"grad_norm": 0.7878923416137695,
|
| 1306 |
+
"learning_rate": 5.824036481318035e-05,
|
| 1307 |
+
"loss": 3.0264,
|
| 1308 |
+
"step": 174000
|
| 1309 |
+
},
|
| 1310 |
+
{
|
| 1311 |
+
"epoch": 9.41,
|
| 1312 |
+
"grad_norm": 0.7934916019439697,
|
| 1313 |
+
"learning_rate": 5.794616063548103e-05,
|
| 1314 |
+
"loss": 3.0296,
|
| 1315 |
+
"step": 175000
|
| 1316 |
+
},
|
| 1317 |
+
{
|
| 1318 |
+
"epoch": 9.46,
|
| 1319 |
+
"grad_norm": 0.79119873046875,
|
| 1320 |
+
"learning_rate": 5.765195645778171e-05,
|
| 1321 |
+
"loss": 3.0311,
|
| 1322 |
+
"step": 176000
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 9.52,
|
| 1326 |
+
"grad_norm": 0.7816409468650818,
|
| 1327 |
+
"learning_rate": 5.7358046484260074e-05,
|
| 1328 |
+
"loss": 3.0362,
|
| 1329 |
+
"step": 177000
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 9.57,
|
| 1333 |
+
"grad_norm": 0.7821895480155945,
|
| 1334 |
+
"learning_rate": 5.7063842306560755e-05,
|
| 1335 |
+
"loss": 3.0322,
|
| 1336 |
+
"step": 178000
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"epoch": 9.63,
|
| 1340 |
+
"grad_norm": 0.7676215171813965,
|
| 1341 |
+
"learning_rate": 5.677022653721683e-05,
|
| 1342 |
+
"loss": 3.0384,
|
| 1343 |
+
"step": 179000
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 9.68,
|
| 1347 |
+
"grad_norm": 0.7703620791435242,
|
| 1348 |
+
"learning_rate": 5.647602235951751e-05,
|
| 1349 |
+
"loss": 3.0354,
|
| 1350 |
+
"step": 180000
|
| 1351 |
+
},
|
| 1352 |
+
{
|
| 1353 |
+
"epoch": 9.73,
|
| 1354 |
+
"grad_norm": 0.7742705345153809,
|
| 1355 |
+
"learning_rate": 5.618181818181818e-05,
|
| 1356 |
+
"loss": 3.0364,
|
| 1357 |
+
"step": 181000
|
| 1358 |
+
},
|
| 1359 |
+
{
|
| 1360 |
+
"epoch": 9.79,
|
| 1361 |
+
"grad_norm": 0.7501013278961182,
|
| 1362 |
+
"learning_rate": 5.588761400411886e-05,
|
| 1363 |
+
"loss": 3.038,
|
| 1364 |
+
"step": 182000
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 9.84,
|
| 1368 |
+
"grad_norm": 0.7982527017593384,
|
| 1369 |
+
"learning_rate": 5.5593704030597236e-05,
|
| 1370 |
+
"loss": 3.0358,
|
| 1371 |
+
"step": 183000
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 9.9,
|
| 1375 |
+
"grad_norm": 0.7525540590286255,
|
| 1376 |
+
"learning_rate": 5.529949985289792e-05,
|
| 1377 |
+
"loss": 3.0391,
|
| 1378 |
+
"step": 184000
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 9.95,
|
| 1382 |
+
"grad_norm": 0.7513368725776672,
|
| 1383 |
+
"learning_rate": 5.5005295675198585e-05,
|
| 1384 |
+
"loss": 3.0407,
|
| 1385 |
+
"step": 185000
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 10.0,
|
| 1389 |
+
"eval_accuracy": 0.40278849891494256,
|
| 1390 |
+
"eval_loss": 3.3952178955078125,
|
| 1391 |
+
"eval_runtime": 153.9135,
|
| 1392 |
+
"eval_samples_per_second": 376.302,
|
| 1393 |
+
"eval_steps_per_second": 5.88,
|
| 1394 |
+
"step": 185950
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 10.0,
|
| 1398 |
+
"grad_norm": 0.7844313979148865,
|
| 1399 |
+
"learning_rate": 5.4711091497499266e-05,
|
| 1400 |
+
"loss": 3.0354,
|
| 1401 |
+
"step": 186000
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 10.06,
|
| 1405 |
+
"grad_norm": 0.7969244122505188,
|
| 1406 |
+
"learning_rate": 5.4417181523977643e-05,
|
| 1407 |
+
"loss": 2.9801,
|
| 1408 |
+
"step": 187000
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"epoch": 10.11,
|
| 1412 |
+
"grad_norm": 0.7915099263191223,
|
| 1413 |
+
"learning_rate": 5.4122977346278325e-05,
|
| 1414 |
+
"loss": 2.9902,
|
| 1415 |
+
"step": 188000
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 10.16,
|
| 1419 |
+
"grad_norm": 0.8039547204971313,
|
| 1420 |
+
"learning_rate": 5.382877316857899e-05,
|
| 1421 |
+
"loss": 2.9976,
|
| 1422 |
+
"step": 189000
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 10.22,
|
| 1426 |
+
"grad_norm": 0.8285199403762817,
|
| 1427 |
+
"learning_rate": 5.3534568990879674e-05,
|
| 1428 |
+
"loss": 2.9953,
|
| 1429 |
+
"step": 190000
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 10.27,
|
| 1433 |
+
"grad_norm": 0.8010717034339905,
|
| 1434 |
+
"learning_rate": 5.324065901735805e-05,
|
| 1435 |
+
"loss": 2.998,
|
| 1436 |
+
"step": 191000
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 10.33,
|
| 1440 |
+
"grad_norm": 0.7704669833183289,
|
| 1441 |
+
"learning_rate": 5.294645483965872e-05,
|
| 1442 |
+
"loss": 3.0017,
|
| 1443 |
+
"step": 192000
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 10.38,
|
| 1447 |
+
"grad_norm": 0.7777411937713623,
|
| 1448 |
+
"learning_rate": 5.26522506619594e-05,
|
| 1449 |
+
"loss": 3.0018,
|
| 1450 |
+
"step": 193000
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 10.43,
|
| 1454 |
+
"grad_norm": 0.8083963394165039,
|
| 1455 |
+
"learning_rate": 5.235863489261548e-05,
|
| 1456 |
+
"loss": 3.0048,
|
| 1457 |
+
"step": 194000
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 10.49,
|
| 1461 |
+
"grad_norm": 0.7807542085647583,
|
| 1462 |
+
"learning_rate": 5.2064430714916154e-05,
|
| 1463 |
+
"loss": 3.0104,
|
| 1464 |
+
"step": 195000
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 10.54,
|
| 1468 |
+
"grad_norm": 0.7731226682662964,
|
| 1469 |
+
"learning_rate": 5.177052074139453e-05,
|
| 1470 |
+
"loss": 3.006,
|
| 1471 |
+
"step": 196000
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 10.59,
|
| 1475 |
+
"grad_norm": 0.7869495153427124,
|
| 1476 |
+
"learning_rate": 5.1476316563695206e-05,
|
| 1477 |
+
"loss": 3.0061,
|
| 1478 |
+
"step": 197000
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 10.65,
|
| 1482 |
+
"grad_norm": 0.7773155570030212,
|
| 1483 |
+
"learning_rate": 5.118211238599588e-05,
|
| 1484 |
+
"loss": 3.0133,
|
| 1485 |
+
"step": 198000
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 10.7,
|
| 1489 |
+
"grad_norm": 0.7991915941238403,
|
| 1490 |
+
"learning_rate": 5.088790820829656e-05,
|
| 1491 |
+
"loss": 3.0093,
|
| 1492 |
+
"step": 199000
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"epoch": 10.76,
|
| 1496 |
+
"grad_norm": 0.8070498704910278,
|
| 1497 |
+
"learning_rate": 5.0593704030597236e-05,
|
| 1498 |
+
"loss": 3.0155,
|
| 1499 |
+
"step": 200000
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 10.81,
|
| 1503 |
+
"grad_norm": 0.7820618152618408,
|
| 1504 |
+
"learning_rate": 5.0299794057075614e-05,
|
| 1505 |
+
"loss": 3.013,
|
| 1506 |
+
"step": 201000
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 10.86,
|
| 1510 |
+
"grad_norm": 0.7867504358291626,
|
| 1511 |
+
"learning_rate": 5.000558987937629e-05,
|
| 1512 |
+
"loss": 3.0112,
|
| 1513 |
+
"step": 202000
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 10.92,
|
| 1517 |
+
"grad_norm": 0.8042470216751099,
|
| 1518 |
+
"learning_rate": 4.971138570167696e-05,
|
| 1519 |
+
"loss": 3.0126,
|
| 1520 |
+
"step": 203000
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 10.97,
|
| 1524 |
+
"grad_norm": 0.8135991096496582,
|
| 1525 |
+
"learning_rate": 4.941718152397764e-05,
|
| 1526 |
+
"loss": 3.0138,
|
| 1527 |
+
"step": 204000
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 11.0,
|
| 1531 |
+
"eval_accuracy": 0.40270257305948315,
|
| 1532 |
+
"eval_loss": 3.3949787616729736,
|
| 1533 |
+
"eval_runtime": 153.9854,
|
| 1534 |
+
"eval_samples_per_second": 376.127,
|
| 1535 |
+
"eval_steps_per_second": 5.877,
|
| 1536 |
+
"step": 204545
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 11.02,
|
| 1540 |
+
"grad_norm": 0.8106272220611572,
|
| 1541 |
+
"learning_rate": 4.912327155045602e-05,
|
| 1542 |
+
"loss": 2.9883,
|
| 1543 |
+
"step": 205000
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 11.08,
|
| 1547 |
+
"grad_norm": 0.8176043629646301,
|
| 1548 |
+
"learning_rate": 4.8829067372756696e-05,
|
| 1549 |
+
"loss": 2.9619,
|
| 1550 |
+
"step": 206000
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 11.13,
|
| 1554 |
+
"grad_norm": 0.8122496008872986,
|
| 1555 |
+
"learning_rate": 4.853515739923507e-05,
|
| 1556 |
+
"loss": 2.9662,
|
| 1557 |
+
"step": 207000
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 11.19,
|
| 1561 |
+
"grad_norm": 0.804396390914917,
|
| 1562 |
+
"learning_rate": 4.824095322153575e-05,
|
| 1563 |
+
"loss": 2.9721,
|
| 1564 |
+
"step": 208000
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 11.24,
|
| 1568 |
+
"grad_norm": 0.8028585314750671,
|
| 1569 |
+
"learning_rate": 4.794674904383643e-05,
|
| 1570 |
+
"loss": 2.9713,
|
| 1571 |
+
"step": 209000
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 11.29,
|
| 1575 |
+
"grad_norm": 0.8111907243728638,
|
| 1576 |
+
"learning_rate": 4.7652839070314806e-05,
|
| 1577 |
+
"loss": 2.9759,
|
| 1578 |
+
"step": 210000
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 11.35,
|
| 1582 |
+
"grad_norm": 0.828597366809845,
|
| 1583 |
+
"learning_rate": 4.735863489261548e-05,
|
| 1584 |
+
"loss": 2.9788,
|
| 1585 |
+
"step": 211000
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 11.4,
|
| 1589 |
+
"grad_norm": 0.8048691153526306,
|
| 1590 |
+
"learning_rate": 4.706472491909386e-05,
|
| 1591 |
+
"loss": 2.9819,
|
| 1592 |
+
"step": 212000
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 11.45,
|
| 1596 |
+
"grad_norm": 0.7922060489654541,
|
| 1597 |
+
"learning_rate": 4.677052074139453e-05,
|
| 1598 |
+
"loss": 2.9851,
|
| 1599 |
+
"step": 213000
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 11.51,
|
| 1603 |
+
"grad_norm": 0.8169183731079102,
|
| 1604 |
+
"learning_rate": 4.6476316563695207e-05,
|
| 1605 |
+
"loss": 2.9814,
|
| 1606 |
+
"step": 214000
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 11.56,
|
| 1610 |
+
"grad_norm": 0.8223658204078674,
|
| 1611 |
+
"learning_rate": 4.618211238599589e-05,
|
| 1612 |
+
"loss": 2.9866,
|
| 1613 |
+
"step": 215000
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 11.62,
|
| 1617 |
+
"grad_norm": 0.7746108174324036,
|
| 1618 |
+
"learning_rate": 4.588820241247426e-05,
|
| 1619 |
+
"loss": 2.9881,
|
| 1620 |
+
"step": 216000
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 11.67,
|
| 1624 |
+
"grad_norm": 0.7992538213729858,
|
| 1625 |
+
"learning_rate": 4.5594292438952636e-05,
|
| 1626 |
+
"loss": 2.9872,
|
| 1627 |
+
"step": 217000
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 11.72,
|
| 1631 |
+
"grad_norm": 0.8094954490661621,
|
| 1632 |
+
"learning_rate": 4.530008826125331e-05,
|
| 1633 |
+
"loss": 2.9878,
|
| 1634 |
+
"step": 218000
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 11.78,
|
| 1638 |
+
"grad_norm": 0.8307532668113708,
|
| 1639 |
+
"learning_rate": 4.500588408355399e-05,
|
| 1640 |
+
"loss": 2.9891,
|
| 1641 |
+
"step": 219000
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 11.83,
|
| 1645 |
+
"grad_norm": 0.8303326964378357,
|
| 1646 |
+
"learning_rate": 4.4711679905854666e-05,
|
| 1647 |
+
"loss": 2.9884,
|
| 1648 |
+
"step": 220000
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 11.88,
|
| 1652 |
+
"grad_norm": 0.7702817916870117,
|
| 1653 |
+
"learning_rate": 4.441747572815535e-05,
|
| 1654 |
+
"loss": 2.9878,
|
| 1655 |
+
"step": 221000
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 11.94,
|
| 1659 |
+
"grad_norm": 0.8154752254486084,
|
| 1660 |
+
"learning_rate": 4.412356575463372e-05,
|
| 1661 |
+
"loss": 2.9861,
|
| 1662 |
+
"step": 222000
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 11.99,
|
| 1666 |
+
"grad_norm": 0.8282615542411804,
|
| 1667 |
+
"learning_rate": 4.38293615769344e-05,
|
| 1668 |
+
"loss": 2.9937,
|
| 1669 |
+
"step": 223000
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 12.0,
|
| 1673 |
+
"eval_accuracy": 0.40359663192144857,
|
| 1674 |
+
"eval_loss": 3.3930435180664062,
|
| 1675 |
+
"eval_runtime": 153.7704,
|
| 1676 |
+
"eval_samples_per_second": 376.653,
|
| 1677 |
+
"eval_steps_per_second": 5.885,
|
| 1678 |
+
"step": 223140
|
| 1679 |
+
},
|
| 1680 |
+
{
|
| 1681 |
+
"epoch": 12.05,
|
| 1682 |
+
"grad_norm": 0.8476201891899109,
|
| 1683 |
+
"learning_rate": 4.353574580759047e-05,
|
| 1684 |
+
"loss": 2.9443,
|
| 1685 |
+
"step": 224000
|
| 1686 |
+
},
|
| 1687 |
+
{
|
| 1688 |
+
"epoch": 12.1,
|
| 1689 |
+
"grad_norm": 0.8455144166946411,
|
| 1690 |
+
"learning_rate": 4.324154162989115e-05,
|
| 1691 |
+
"loss": 2.9452,
|
| 1692 |
+
"step": 225000
|
| 1693 |
+
},
|
| 1694 |
+
{
|
| 1695 |
+
"epoch": 12.15,
|
| 1696 |
+
"grad_norm": 0.8212487101554871,
|
| 1697 |
+
"learning_rate": 4.2947631656369524e-05,
|
| 1698 |
+
"loss": 2.9477,
|
| 1699 |
+
"step": 226000
|
| 1700 |
+
},
|
| 1701 |
+
{
|
| 1702 |
+
"epoch": 12.21,
|
| 1703 |
+
"grad_norm": 0.8192943930625916,
|
| 1704 |
+
"learning_rate": 4.26534274786702e-05,
|
| 1705 |
+
"loss": 2.9546,
|
| 1706 |
+
"step": 227000
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 12.26,
|
| 1710 |
+
"grad_norm": 0.8424757122993469,
|
| 1711 |
+
"learning_rate": 4.235922330097087e-05,
|
| 1712 |
+
"loss": 2.95,
|
| 1713 |
+
"step": 228000
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 12.32,
|
| 1717 |
+
"grad_norm": 0.8195380568504333,
|
| 1718 |
+
"learning_rate": 4.206531332744925e-05,
|
| 1719 |
+
"loss": 2.9545,
|
| 1720 |
+
"step": 229000
|
| 1721 |
+
},
|
| 1722 |
+
{
|
| 1723 |
+
"epoch": 12.37,
|
| 1724 |
+
"grad_norm": 0.8145859837532043,
|
| 1725 |
+
"learning_rate": 4.1771109149749925e-05,
|
| 1726 |
+
"loss": 2.9559,
|
| 1727 |
+
"step": 230000
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 12.42,
|
| 1731 |
+
"grad_norm": 0.8282389640808105,
|
| 1732 |
+
"learning_rate": 4.1476904972050606e-05,
|
| 1733 |
+
"loss": 2.9605,
|
| 1734 |
+
"step": 231000
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 12.48,
|
| 1738 |
+
"grad_norm": 0.8569869995117188,
|
| 1739 |
+
"learning_rate": 4.118270079435128e-05,
|
| 1740 |
+
"loss": 2.961,
|
| 1741 |
+
"step": 232000
|
| 1742 |
+
},
|
| 1743 |
+
{
|
| 1744 |
+
"epoch": 12.53,
|
| 1745 |
+
"grad_norm": 0.794511616230011,
|
| 1746 |
+
"learning_rate": 4.088879082082966e-05,
|
| 1747 |
+
"loss": 2.9611,
|
| 1748 |
+
"step": 233000
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 12.58,
|
| 1752 |
+
"grad_norm": 0.8420530557632446,
|
| 1753 |
+
"learning_rate": 4.059458664313033e-05,
|
| 1754 |
+
"loss": 2.9637,
|
| 1755 |
+
"step": 234000
|
| 1756 |
+
},
|
| 1757 |
+
{
|
| 1758 |
+
"epoch": 12.64,
|
| 1759 |
+
"grad_norm": 0.7839045524597168,
|
| 1760 |
+
"learning_rate": 4.030038246543101e-05,
|
| 1761 |
+
"loss": 2.966,
|
| 1762 |
+
"step": 235000
|
| 1763 |
+
},
|
| 1764 |
+
{
|
| 1765 |
+
"epoch": 12.69,
|
| 1766 |
+
"grad_norm": 0.8528980016708374,
|
| 1767 |
+
"learning_rate": 4.0006472491909384e-05,
|
| 1768 |
+
"loss": 2.9656,
|
| 1769 |
+
"step": 236000
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 12.75,
|
| 1773 |
+
"grad_norm": 0.8356289267539978,
|
| 1774 |
+
"learning_rate": 3.971256251838776e-05,
|
| 1775 |
+
"loss": 2.9676,
|
| 1776 |
+
"step": 237000
|
| 1777 |
+
},
|
| 1778 |
+
{
|
| 1779 |
+
"epoch": 12.8,
|
| 1780 |
+
"grad_norm": 0.7946913838386536,
|
| 1781 |
+
"learning_rate": 3.9418358340688436e-05,
|
| 1782 |
+
"loss": 2.9684,
|
| 1783 |
+
"step": 238000
|
| 1784 |
+
},
|
| 1785 |
+
{
|
| 1786 |
+
"epoch": 12.85,
|
| 1787 |
+
"grad_norm": 0.8257811665534973,
|
| 1788 |
+
"learning_rate": 3.912415416298912e-05,
|
| 1789 |
+
"loss": 2.9691,
|
| 1790 |
+
"step": 239000
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 12.91,
|
| 1794 |
+
"grad_norm": 0.8345640897750854,
|
| 1795 |
+
"learning_rate": 3.882994998528979e-05,
|
| 1796 |
+
"loss": 2.9655,
|
| 1797 |
+
"step": 240000
|
| 1798 |
+
},
|
| 1799 |
+
{
|
| 1800 |
+
"epoch": 12.96,
|
| 1801 |
+
"grad_norm": 0.815122663974762,
|
| 1802 |
+
"learning_rate": 3.853604001176817e-05,
|
| 1803 |
+
"loss": 2.9758,
|
| 1804 |
+
"step": 241000
|
| 1805 |
+
},
|
| 1806 |
+
{
|
| 1807 |
+
"epoch": 13.0,
|
| 1808 |
+
"eval_accuracy": 0.4041131276030529,
|
| 1809 |
+
"eval_loss": 3.4018185138702393,
|
| 1810 |
+
"eval_runtime": 154.088,
|
| 1811 |
+
"eval_samples_per_second": 375.876,
|
| 1812 |
+
"eval_steps_per_second": 5.873,
|
| 1813 |
+
"step": 241735
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 13.01,
|
| 1817 |
+
"grad_norm": 0.8575399518013,
|
| 1818 |
+
"learning_rate": 3.8242130038246546e-05,
|
| 1819 |
+
"loss": 2.9563,
|
| 1820 |
+
"step": 242000
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 13.07,
|
| 1824 |
+
"grad_norm": 0.8668804168701172,
|
| 1825 |
+
"learning_rate": 3.794792586054722e-05,
|
| 1826 |
+
"loss": 2.924,
|
| 1827 |
+
"step": 243000
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 13.12,
|
| 1831 |
+
"grad_norm": 0.8839793801307678,
|
| 1832 |
+
"learning_rate": 3.7653721682847895e-05,
|
| 1833 |
+
"loss": 2.9267,
|
| 1834 |
+
"step": 244000
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 13.18,
|
| 1838 |
+
"grad_norm": 0.8395956754684448,
|
| 1839 |
+
"learning_rate": 3.735951750514857e-05,
|
| 1840 |
+
"loss": 2.931,
|
| 1841 |
+
"step": 245000
|
| 1842 |
+
},
|
| 1843 |
+
{
|
| 1844 |
+
"epoch": 13.23,
|
| 1845 |
+
"grad_norm": 0.8280730843544006,
|
| 1846 |
+
"learning_rate": 3.706531332744925e-05,
|
| 1847 |
+
"loss": 2.9303,
|
| 1848 |
+
"step": 246000
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 13.28,
|
| 1852 |
+
"grad_norm": 0.8762979507446289,
|
| 1853 |
+
"learning_rate": 3.677140335392763e-05,
|
| 1854 |
+
"loss": 2.9349,
|
| 1855 |
+
"step": 247000
|
| 1856 |
+
},
|
| 1857 |
+
{
|
| 1858 |
+
"epoch": 13.34,
|
| 1859 |
+
"grad_norm": 0.8566986918449402,
|
| 1860 |
+
"learning_rate": 3.6477493380406005e-05,
|
| 1861 |
+
"loss": 2.9361,
|
| 1862 |
+
"step": 248000
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 13.39,
|
| 1866 |
+
"grad_norm": 0.8482074737548828,
|
| 1867 |
+
"learning_rate": 3.618328920270668e-05,
|
| 1868 |
+
"loss": 2.9376,
|
| 1869 |
+
"step": 249000
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 13.44,
|
| 1873 |
+
"grad_norm": 0.847722053527832,
|
| 1874 |
+
"learning_rate": 3.5889085025007354e-05,
|
| 1875 |
+
"loss": 2.9431,
|
| 1876 |
+
"step": 250000
|
| 1877 |
+
},
|
| 1878 |
+
{
|
| 1879 |
+
"epoch": 13.5,
|
| 1880 |
+
"grad_norm": 0.8807717561721802,
|
| 1881 |
+
"learning_rate": 3.559517505148574e-05,
|
| 1882 |
+
"loss": 2.9441,
|
| 1883 |
+
"step": 251000
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 13.55,
|
| 1887 |
+
"grad_norm": 0.8653402924537659,
|
| 1888 |
+
"learning_rate": 3.530097087378641e-05,
|
| 1889 |
+
"loss": 2.9429,
|
| 1890 |
+
"step": 252000
|
| 1891 |
+
},
|
| 1892 |
+
{
|
| 1893 |
+
"epoch": 13.61,
|
| 1894 |
+
"grad_norm": 0.8465253114700317,
|
| 1895 |
+
"learning_rate": 3.500676669608709e-05,
|
| 1896 |
+
"loss": 2.9417,
|
| 1897 |
+
"step": 253000
|
| 1898 |
+
},
|
| 1899 |
+
{
|
| 1900 |
+
"epoch": 13.66,
|
| 1901 |
+
"grad_norm": 0.8789597749710083,
|
| 1902 |
+
"learning_rate": 3.471256251838776e-05,
|
| 1903 |
+
"loss": 2.9446,
|
| 1904 |
+
"step": 254000
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 13.71,
|
| 1908 |
+
"grad_norm": 0.8379344940185547,
|
| 1909 |
+
"learning_rate": 3.4418358340688436e-05,
|
| 1910 |
+
"loss": 2.946,
|
| 1911 |
+
"step": 255000
|
| 1912 |
+
},
|
| 1913 |
+
{
|
| 1914 |
+
"epoch": 13.77,
|
| 1915 |
+
"grad_norm": 0.8516440987586975,
|
| 1916 |
+
"learning_rate": 3.412444836716682e-05,
|
| 1917 |
+
"loss": 2.9483,
|
| 1918 |
+
"step": 256000
|
| 1919 |
+
},
|
| 1920 |
+
{
|
| 1921 |
+
"epoch": 13.82,
|
| 1922 |
+
"grad_norm": 0.8952211737632751,
|
| 1923 |
+
"learning_rate": 3.3830244189467495e-05,
|
| 1924 |
+
"loss": 2.9486,
|
| 1925 |
+
"step": 257000
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"epoch": 13.87,
|
| 1929 |
+
"grad_norm": 0.8536117672920227,
|
| 1930 |
+
"learning_rate": 3.353633421594587e-05,
|
| 1931 |
+
"loss": 2.948,
|
| 1932 |
+
"step": 258000
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"epoch": 13.93,
|
| 1936 |
+
"grad_norm": 0.8689549565315247,
|
| 1937 |
+
"learning_rate": 3.3242130038246546e-05,
|
| 1938 |
+
"loss": 2.9495,
|
| 1939 |
+
"step": 259000
|
| 1940 |
+
},
|
| 1941 |
+
{
|
| 1942 |
+
"epoch": 13.98,
|
| 1943 |
+
"grad_norm": 0.8051441311836243,
|
| 1944 |
+
"learning_rate": 3.294792586054722e-05,
|
| 1945 |
+
"loss": 2.9478,
|
| 1946 |
+
"step": 260000
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 14.0,
|
| 1950 |
+
"eval_accuracy": 0.40446637087749043,
|
| 1951 |
+
"eval_loss": 3.402351140975952,
|
| 1952 |
+
"eval_runtime": 153.948,
|
| 1953 |
+
"eval_samples_per_second": 376.218,
|
| 1954 |
+
"eval_steps_per_second": 5.879,
|
| 1955 |
+
"step": 260330
|
| 1956 |
+
},
|
| 1957 |
+
{
|
| 1958 |
+
"epoch": 14.04,
|
| 1959 |
+
"grad_norm": 0.8607249855995178,
|
| 1960 |
+
"learning_rate": 3.26540158870256e-05,
|
| 1961 |
+
"loss": 2.9142,
|
| 1962 |
+
"step": 261000
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 14.09,
|
| 1966 |
+
"grad_norm": 0.9051989912986755,
|
| 1967 |
+
"learning_rate": 3.2360105913503975e-05,
|
| 1968 |
+
"loss": 2.9049,
|
| 1969 |
+
"step": 262000
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 14.14,
|
| 1973 |
+
"grad_norm": 0.8793129920959473,
|
| 1974 |
+
"learning_rate": 3.206590173580465e-05,
|
| 1975 |
+
"loss": 2.9119,
|
| 1976 |
+
"step": 263000
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 14.2,
|
| 1980 |
+
"grad_norm": 0.8799360990524292,
|
| 1981 |
+
"learning_rate": 3.177169755810533e-05,
|
| 1982 |
+
"loss": 2.9128,
|
| 1983 |
+
"step": 264000
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 14.25,
|
| 1987 |
+
"grad_norm": 0.8635936379432678,
|
| 1988 |
+
"learning_rate": 3.1477493380406006e-05,
|
| 1989 |
+
"loss": 2.9162,
|
| 1990 |
+
"step": 265000
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 14.3,
|
| 1994 |
+
"grad_norm": 0.8844926357269287,
|
| 1995 |
+
"learning_rate": 3.118358340688438e-05,
|
| 1996 |
+
"loss": 2.9184,
|
| 1997 |
+
"step": 266000
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 14.36,
|
| 2001 |
+
"grad_norm": 0.8986685276031494,
|
| 2002 |
+
"learning_rate": 3.088937922918506e-05,
|
| 2003 |
+
"loss": 2.9174,
|
| 2004 |
+
"step": 267000
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 14.41,
|
| 2008 |
+
"grad_norm": 0.9062278270721436,
|
| 2009 |
+
"learning_rate": 3.059517505148573e-05,
|
| 2010 |
+
"loss": 2.9196,
|
| 2011 |
+
"step": 268000
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 14.47,
|
| 2015 |
+
"grad_norm": 0.8506668210029602,
|
| 2016 |
+
"learning_rate": 3.030097087378641e-05,
|
| 2017 |
+
"loss": 2.9259,
|
| 2018 |
+
"step": 269000
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 14.52,
|
| 2022 |
+
"grad_norm": 0.8786486983299255,
|
| 2023 |
+
"learning_rate": 3.0007060900264784e-05,
|
| 2024 |
+
"loss": 2.9227,
|
| 2025 |
+
"step": 270000
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 14.57,
|
| 2029 |
+
"grad_norm": 0.8500825762748718,
|
| 2030 |
+
"learning_rate": 2.971285672256546e-05,
|
| 2031 |
+
"loss": 2.9235,
|
| 2032 |
+
"step": 271000
|
| 2033 |
+
},
|
| 2034 |
+
{
|
| 2035 |
+
"epoch": 14.63,
|
| 2036 |
+
"grad_norm": 0.882700502872467,
|
| 2037 |
+
"learning_rate": 2.9418946749043835e-05,
|
| 2038 |
+
"loss": 2.9265,
|
| 2039 |
+
"step": 272000
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 14.68,
|
| 2043 |
+
"grad_norm": 0.8619365096092224,
|
| 2044 |
+
"learning_rate": 2.9124742571344517e-05,
|
| 2045 |
+
"loss": 2.9306,
|
| 2046 |
+
"step": 273000
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 14.74,
|
| 2050 |
+
"grad_norm": 0.8572811484336853,
|
| 2051 |
+
"learning_rate": 2.883053839364519e-05,
|
| 2052 |
+
"loss": 2.9297,
|
| 2053 |
+
"step": 274000
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 14.79,
|
| 2057 |
+
"grad_norm": 0.8639488816261292,
|
| 2058 |
+
"learning_rate": 2.853662842012357e-05,
|
| 2059 |
+
"loss": 2.9243,
|
| 2060 |
+
"step": 275000
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 14.84,
|
| 2064 |
+
"grad_norm": 0.8537041544914246,
|
| 2065 |
+
"learning_rate": 2.8242424242424243e-05,
|
| 2066 |
+
"loss": 2.9282,
|
| 2067 |
+
"step": 276000
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 14.9,
|
| 2071 |
+
"grad_norm": 0.9280806183815002,
|
| 2072 |
+
"learning_rate": 2.794880847308032e-05,
|
| 2073 |
+
"loss": 2.928,
|
| 2074 |
+
"step": 277000
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 14.95,
|
| 2078 |
+
"grad_norm": 0.8643816709518433,
|
| 2079 |
+
"learning_rate": 2.7654604295380997e-05,
|
| 2080 |
+
"loss": 2.9311,
|
| 2081 |
+
"step": 278000
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 15.0,
|
| 2085 |
+
"eval_accuracy": 0.40496674286455975,
|
| 2086 |
+
"eval_loss": 3.4006123542785645,
|
| 2087 |
+
"eval_runtime": 153.7369,
|
| 2088 |
+
"eval_samples_per_second": 376.735,
|
| 2089 |
+
"eval_steps_per_second": 5.887,
|
| 2090 |
+
"step": 278925
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 15.0,
|
| 2094 |
+
"grad_norm": 0.8929987549781799,
|
| 2095 |
+
"learning_rate": 2.7360400117681672e-05,
|
| 2096 |
+
"loss": 2.9279,
|
| 2097 |
+
"step": 279000
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 15.06,
|
| 2101 |
+
"grad_norm": 0.9207207560539246,
|
| 2102 |
+
"learning_rate": 2.7066195939982346e-05,
|
| 2103 |
+
"loss": 2.8887,
|
| 2104 |
+
"step": 280000
|
| 2105 |
+
},
|
| 2106 |
+
{
|
| 2107 |
+
"epoch": 15.11,
|
| 2108 |
+
"grad_norm": 0.8984337449073792,
|
| 2109 |
+
"learning_rate": 2.6771991762283028e-05,
|
| 2110 |
+
"loss": 2.891,
|
| 2111 |
+
"step": 281000
|
| 2112 |
+
},
|
| 2113 |
+
{
|
| 2114 |
+
"epoch": 15.17,
|
| 2115 |
+
"grad_norm": 0.9022777080535889,
|
| 2116 |
+
"learning_rate": 2.6478081788761398e-05,
|
| 2117 |
+
"loss": 2.893,
|
| 2118 |
+
"step": 282000
|
| 2119 |
+
},
|
| 2120 |
+
{
|
| 2121 |
+
"epoch": 15.22,
|
| 2122 |
+
"grad_norm": 0.876774251461029,
|
| 2123 |
+
"learning_rate": 2.618387761106208e-05,
|
| 2124 |
+
"loss": 2.892,
|
| 2125 |
+
"step": 283000
|
| 2126 |
+
},
|
| 2127 |
+
{
|
| 2128 |
+
"epoch": 15.27,
|
| 2129 |
+
"grad_norm": 0.8685379028320312,
|
| 2130 |
+
"learning_rate": 2.5889967637540453e-05,
|
| 2131 |
+
"loss": 2.8957,
|
| 2132 |
+
"step": 284000
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 15.33,
|
| 2136 |
+
"grad_norm": 0.8736298680305481,
|
| 2137 |
+
"learning_rate": 2.559605766401883e-05,
|
| 2138 |
+
"loss": 2.9011,
|
| 2139 |
+
"step": 285000
|
| 2140 |
+
},
|
| 2141 |
+
{
|
| 2142 |
+
"epoch": 15.38,
|
| 2143 |
+
"grad_norm": 0.889697790145874,
|
| 2144 |
+
"learning_rate": 2.5301853486319505e-05,
|
| 2145 |
+
"loss": 2.9026,
|
| 2146 |
+
"step": 286000
|
| 2147 |
+
},
|
| 2148 |
+
{
|
| 2149 |
+
"epoch": 15.43,
|
| 2150 |
+
"grad_norm": 0.859124481678009,
|
| 2151 |
+
"learning_rate": 2.5007943512797882e-05,
|
| 2152 |
+
"loss": 2.9034,
|
| 2153 |
+
"step": 287000
|
| 2154 |
+
},
|
| 2155 |
+
{
|
| 2156 |
+
"epoch": 15.49,
|
| 2157 |
+
"grad_norm": 0.9010147452354431,
|
| 2158 |
+
"learning_rate": 2.471373933509856e-05,
|
| 2159 |
+
"loss": 2.9094,
|
| 2160 |
+
"step": 288000
|
| 2161 |
+
},
|
| 2162 |
+
{
|
| 2163 |
+
"epoch": 15.54,
|
| 2164 |
+
"grad_norm": 0.8996730446815491,
|
| 2165 |
+
"learning_rate": 2.4419535157399238e-05,
|
| 2166 |
+
"loss": 2.9058,
|
| 2167 |
+
"step": 289000
|
| 2168 |
+
},
|
| 2169 |
+
{
|
| 2170 |
+
"epoch": 15.6,
|
| 2171 |
+
"grad_norm": 0.916613757610321,
|
| 2172 |
+
"learning_rate": 2.4125330979699913e-05,
|
| 2173 |
+
"loss": 2.9069,
|
| 2174 |
+
"step": 290000
|
| 2175 |
+
},
|
| 2176 |
+
{
|
| 2177 |
+
"epoch": 15.65,
|
| 2178 |
+
"grad_norm": 0.894801914691925,
|
| 2179 |
+
"learning_rate": 2.3831715210355986e-05,
|
| 2180 |
+
"loss": 2.9079,
|
| 2181 |
+
"step": 291000
|
| 2182 |
+
},
|
| 2183 |
+
{
|
| 2184 |
+
"epoch": 15.7,
|
| 2185 |
+
"grad_norm": 0.8849554061889648,
|
| 2186 |
+
"learning_rate": 2.3537511032656664e-05,
|
| 2187 |
+
"loss": 2.9123,
|
| 2188 |
+
"step": 292000
|
| 2189 |
+
},
|
| 2190 |
+
{
|
| 2191 |
+
"epoch": 15.76,
|
| 2192 |
+
"grad_norm": 0.8884662985801697,
|
| 2193 |
+
"learning_rate": 2.324330685495734e-05,
|
| 2194 |
+
"loss": 2.9147,
|
| 2195 |
+
"step": 293000
|
| 2196 |
+
},
|
| 2197 |
+
{
|
| 2198 |
+
"epoch": 15.81,
|
| 2199 |
+
"grad_norm": 0.873546838760376,
|
| 2200 |
+
"learning_rate": 2.294910267725802e-05,
|
| 2201 |
+
"loss": 2.9085,
|
| 2202 |
+
"step": 294000
|
| 2203 |
+
},
|
| 2204 |
+
{
|
| 2205 |
+
"epoch": 15.86,
|
| 2206 |
+
"grad_norm": 0.9005961418151855,
|
| 2207 |
+
"learning_rate": 2.2654898499558694e-05,
|
| 2208 |
+
"loss": 2.9092,
|
| 2209 |
+
"step": 295000
|
| 2210 |
+
},
|
| 2211 |
+
{
|
| 2212 |
+
"epoch": 15.92,
|
| 2213 |
+
"grad_norm": 0.9243658185005188,
|
| 2214 |
+
"learning_rate": 2.236098852603707e-05,
|
| 2215 |
+
"loss": 2.9163,
|
| 2216 |
+
"step": 296000
|
| 2217 |
+
},
|
| 2218 |
+
{
|
| 2219 |
+
"epoch": 15.97,
|
| 2220 |
+
"grad_norm": 0.926802396774292,
|
| 2221 |
+
"learning_rate": 2.2066784348337746e-05,
|
| 2222 |
+
"loss": 2.9098,
|
| 2223 |
+
"step": 297000
|
| 2224 |
+
},
|
| 2225 |
+
{
|
| 2226 |
+
"epoch": 16.0,
|
| 2227 |
+
"eval_accuracy": 0.4050182715050112,
|
| 2228 |
+
"eval_loss": 3.403717041015625,
|
| 2229 |
+
"eval_runtime": 154.0227,
|
| 2230 |
+
"eval_samples_per_second": 376.035,
|
| 2231 |
+
"eval_steps_per_second": 5.876,
|
| 2232 |
+
"step": 297520
|
| 2233 |
+
},
|
| 2234 |
+
{
|
| 2235 |
+
"epoch": 16.03,
|
| 2236 |
+
"grad_norm": 0.8965989947319031,
|
| 2237 |
+
"learning_rate": 2.1772874374816123e-05,
|
| 2238 |
+
"loss": 2.895,
|
| 2239 |
+
"step": 298000
|
| 2240 |
+
},
|
| 2241 |
+
{
|
| 2242 |
+
"epoch": 16.08,
|
| 2243 |
+
"grad_norm": 0.899157702922821,
|
| 2244 |
+
"learning_rate": 2.14786701971168e-05,
|
| 2245 |
+
"loss": 2.8783,
|
| 2246 |
+
"step": 299000
|
| 2247 |
+
},
|
| 2248 |
+
{
|
| 2249 |
+
"epoch": 16.13,
|
| 2250 |
+
"grad_norm": 0.8964145183563232,
|
| 2251 |
+
"learning_rate": 2.1184760223595178e-05,
|
| 2252 |
+
"loss": 2.882,
|
| 2253 |
+
"step": 300000
|
| 2254 |
+
},
|
| 2255 |
+
{
|
| 2256 |
+
"epoch": 16.19,
|
| 2257 |
+
"grad_norm": 0.888133704662323,
|
| 2258 |
+
"learning_rate": 2.0890556045895853e-05,
|
| 2259 |
+
"loss": 2.8782,
|
| 2260 |
+
"step": 301000
|
| 2261 |
+
},
|
| 2262 |
+
{
|
| 2263 |
+
"epoch": 16.24,
|
| 2264 |
+
"grad_norm": 0.8767296075820923,
|
| 2265 |
+
"learning_rate": 2.0596351868196527e-05,
|
| 2266 |
+
"loss": 2.8841,
|
| 2267 |
+
"step": 302000
|
| 2268 |
+
},
|
| 2269 |
+
{
|
| 2270 |
+
"epoch": 16.29,
|
| 2271 |
+
"grad_norm": 0.9565109610557556,
|
| 2272 |
+
"learning_rate": 2.0302147690497205e-05,
|
| 2273 |
+
"loss": 2.8836,
|
| 2274 |
+
"step": 303000
|
| 2275 |
+
},
|
| 2276 |
+
{
|
| 2277 |
+
"epoch": 16.35,
|
| 2278 |
+
"grad_norm": 0.9353139400482178,
|
| 2279 |
+
"learning_rate": 2.000882612533098e-05,
|
| 2280 |
+
"loss": 2.8852,
|
| 2281 |
+
"step": 304000
|
| 2282 |
+
},
|
| 2283 |
+
{
|
| 2284 |
+
"epoch": 16.4,
|
| 2285 |
+
"grad_norm": 0.888659656047821,
|
| 2286 |
+
"learning_rate": 1.9714621947631656e-05,
|
| 2287 |
+
"loss": 2.8865,
|
| 2288 |
+
"step": 305000
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 16.46,
|
| 2292 |
+
"grad_norm": 0.9457473754882812,
|
| 2293 |
+
"learning_rate": 1.9420417769932333e-05,
|
| 2294 |
+
"loss": 2.8847,
|
| 2295 |
+
"step": 306000
|
| 2296 |
+
},
|
| 2297 |
+
{
|
| 2298 |
+
"epoch": 16.51,
|
| 2299 |
+
"grad_norm": 0.8824300765991211,
|
| 2300 |
+
"learning_rate": 1.912621359223301e-05,
|
| 2301 |
+
"loss": 2.8885,
|
| 2302 |
+
"step": 307000
|
| 2303 |
+
},
|
| 2304 |
+
{
|
| 2305 |
+
"epoch": 16.56,
|
| 2306 |
+
"grad_norm": 0.9101892709732056,
|
| 2307 |
+
"learning_rate": 1.883200941453369e-05,
|
| 2308 |
+
"loss": 2.8887,
|
| 2309 |
+
"step": 308000
|
| 2310 |
+
},
|
| 2311 |
+
{
|
| 2312 |
+
"epoch": 16.62,
|
| 2313 |
+
"grad_norm": 0.9139971733093262,
|
| 2314 |
+
"learning_rate": 1.8538099441012063e-05,
|
| 2315 |
+
"loss": 2.8914,
|
| 2316 |
+
"step": 309000
|
| 2317 |
+
},
|
| 2318 |
+
{
|
| 2319 |
+
"epoch": 16.67,
|
| 2320 |
+
"grad_norm": 0.8864375352859497,
|
| 2321 |
+
"learning_rate": 1.824389526331274e-05,
|
| 2322 |
+
"loss": 2.8944,
|
| 2323 |
+
"step": 310000
|
| 2324 |
+
},
|
| 2325 |
+
{
|
| 2326 |
+
"epoch": 16.72,
|
| 2327 |
+
"grad_norm": 0.9049232602119446,
|
| 2328 |
+
"learning_rate": 1.7949985289791115e-05,
|
| 2329 |
+
"loss": 2.8922,
|
| 2330 |
+
"step": 311000
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 16.78,
|
| 2334 |
+
"grad_norm": 0.8852655291557312,
|
| 2335 |
+
"learning_rate": 1.7655781112091793e-05,
|
| 2336 |
+
"loss": 2.8927,
|
| 2337 |
+
"step": 312000
|
| 2338 |
+
},
|
| 2339 |
+
{
|
| 2340 |
+
"epoch": 16.83,
|
| 2341 |
+
"grad_norm": 0.9409217238426208,
|
| 2342 |
+
"learning_rate": 1.7361871138570167e-05,
|
| 2343 |
+
"loss": 2.8909,
|
| 2344 |
+
"step": 313000
|
| 2345 |
+
},
|
| 2346 |
+
{
|
| 2347 |
+
"epoch": 16.89,
|
| 2348 |
+
"grad_norm": 0.9365453124046326,
|
| 2349 |
+
"learning_rate": 1.7067666960870844e-05,
|
| 2350 |
+
"loss": 2.8932,
|
| 2351 |
+
"step": 314000
|
| 2352 |
+
},
|
| 2353 |
+
{
|
| 2354 |
+
"epoch": 16.94,
|
| 2355 |
+
"grad_norm": 0.8982263207435608,
|
| 2356 |
+
"learning_rate": 1.6773462783171522e-05,
|
| 2357 |
+
"loss": 2.8954,
|
| 2358 |
+
"step": 315000
|
| 2359 |
+
},
|
| 2360 |
+
{
|
| 2361 |
+
"epoch": 16.99,
|
| 2362 |
+
"grad_norm": 0.8986501693725586,
|
| 2363 |
+
"learning_rate": 1.64795528096499e-05,
|
| 2364 |
+
"loss": 2.8959,
|
| 2365 |
+
"step": 316000
|
| 2366 |
+
},
|
| 2367 |
+
{
|
| 2368 |
+
"epoch": 17.0,
|
| 2369 |
+
"eval_accuracy": 0.4054019482528835,
|
| 2370 |
+
"eval_loss": 3.3979642391204834,
|
| 2371 |
+
"eval_runtime": 153.8765,
|
| 2372 |
+
"eval_samples_per_second": 376.393,
|
| 2373 |
+
"eval_steps_per_second": 5.881,
|
| 2374 |
+
"step": 316115
|
| 2375 |
+
},
|
| 2376 |
+
{
|
| 2377 |
+
"epoch": 17.05,
|
| 2378 |
+
"grad_norm": 0.918526291847229,
|
| 2379 |
+
"learning_rate": 1.6185348631950574e-05,
|
| 2380 |
+
"loss": 2.8667,
|
| 2381 |
+
"step": 317000
|
| 2382 |
+
},
|
| 2383 |
+
{
|
| 2384 |
+
"epoch": 17.1,
|
| 2385 |
+
"grad_norm": 0.9123374819755554,
|
| 2386 |
+
"learning_rate": 1.589143865842895e-05,
|
| 2387 |
+
"loss": 2.8647,
|
| 2388 |
+
"step": 318000
|
| 2389 |
+
},
|
| 2390 |
+
{
|
| 2391 |
+
"epoch": 17.16,
|
| 2392 |
+
"grad_norm": 0.8885955214500427,
|
| 2393 |
+
"learning_rate": 1.559723448072963e-05,
|
| 2394 |
+
"loss": 2.868,
|
| 2395 |
+
"step": 319000
|
| 2396 |
+
},
|
| 2397 |
+
{
|
| 2398 |
+
"epoch": 17.21,
|
| 2399 |
+
"grad_norm": 0.921298623085022,
|
| 2400 |
+
"learning_rate": 1.5303324507208003e-05,
|
| 2401 |
+
"loss": 2.8696,
|
| 2402 |
+
"step": 320000
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 17.26,
|
| 2406 |
+
"grad_norm": 0.9117158651351929,
|
| 2407 |
+
"learning_rate": 1.5009414533686379e-05,
|
| 2408 |
+
"loss": 2.8702,
|
| 2409 |
+
"step": 321000
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 17.32,
|
| 2413 |
+
"grad_norm": 0.9133957624435425,
|
| 2414 |
+
"learning_rate": 1.4715210355987055e-05,
|
| 2415 |
+
"loss": 2.8688,
|
| 2416 |
+
"step": 322000
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 17.37,
|
| 2420 |
+
"grad_norm": 0.9354053735733032,
|
| 2421 |
+
"learning_rate": 1.4421300382465434e-05,
|
| 2422 |
+
"loss": 2.8734,
|
| 2423 |
+
"step": 323000
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 17.42,
|
| 2427 |
+
"grad_norm": 0.9255644679069519,
|
| 2428 |
+
"learning_rate": 1.4127096204766108e-05,
|
| 2429 |
+
"loss": 2.8742,
|
| 2430 |
+
"step": 324000
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 17.48,
|
| 2434 |
+
"grad_norm": 0.9188544154167175,
|
| 2435 |
+
"learning_rate": 1.3832892027066784e-05,
|
| 2436 |
+
"loss": 2.8702,
|
| 2437 |
+
"step": 325000
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 17.53,
|
| 2441 |
+
"grad_norm": 0.9310522079467773,
|
| 2442 |
+
"learning_rate": 1.3538687849367462e-05,
|
| 2443 |
+
"loss": 2.875,
|
| 2444 |
+
"step": 326000
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 17.59,
|
| 2448 |
+
"grad_norm": 0.9256516098976135,
|
| 2449 |
+
"learning_rate": 1.3244483671668137e-05,
|
| 2450 |
+
"loss": 2.8738,
|
| 2451 |
+
"step": 327000
|
| 2452 |
+
},
|
| 2453 |
+
{
|
| 2454 |
+
"epoch": 17.64,
|
| 2455 |
+
"grad_norm": 0.9342946410179138,
|
| 2456 |
+
"learning_rate": 1.2950573698146514e-05,
|
| 2457 |
+
"loss": 2.8739,
|
| 2458 |
+
"step": 328000
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 17.69,
|
| 2462 |
+
"grad_norm": 0.8977386355400085,
|
| 2463 |
+
"learning_rate": 1.2656369520447192e-05,
|
| 2464 |
+
"loss": 2.876,
|
| 2465 |
+
"step": 329000
|
| 2466 |
+
},
|
| 2467 |
+
{
|
| 2468 |
+
"epoch": 17.75,
|
| 2469 |
+
"grad_norm": 0.8941238522529602,
|
| 2470 |
+
"learning_rate": 1.2362459546925568e-05,
|
| 2471 |
+
"loss": 2.8738,
|
| 2472 |
+
"step": 330000
|
| 2473 |
+
},
|
| 2474 |
+
{
|
| 2475 |
+
"epoch": 17.8,
|
| 2476 |
+
"grad_norm": 0.9025571346282959,
|
| 2477 |
+
"learning_rate": 1.2068255369226244e-05,
|
| 2478 |
+
"loss": 2.874,
|
| 2479 |
+
"step": 331000
|
| 2480 |
+
},
|
| 2481 |
+
{
|
| 2482 |
+
"epoch": 17.85,
|
| 2483 |
+
"grad_norm": 0.9084547162055969,
|
| 2484 |
+
"learning_rate": 1.177405119152692e-05,
|
| 2485 |
+
"loss": 2.8777,
|
| 2486 |
+
"step": 332000
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 17.91,
|
| 2490 |
+
"grad_norm": 0.9322535991668701,
|
| 2491 |
+
"learning_rate": 1.1480141218005295e-05,
|
| 2492 |
+
"loss": 2.8748,
|
| 2493 |
+
"step": 333000
|
| 2494 |
+
},
|
| 2495 |
+
{
|
| 2496 |
+
"epoch": 17.96,
|
| 2497 |
+
"grad_norm": 0.9495552182197571,
|
| 2498 |
+
"learning_rate": 1.1185937040305973e-05,
|
| 2499 |
+
"loss": 2.8768,
|
| 2500 |
+
"step": 334000
|
| 2501 |
+
},
|
| 2502 |
+
{
|
| 2503 |
+
"epoch": 18.0,
|
| 2504 |
+
"eval_accuracy": 0.4057570054429562,
|
| 2505 |
+
"eval_loss": 3.407381534576416,
|
| 2506 |
+
"eval_runtime": 153.7887,
|
| 2507 |
+
"eval_samples_per_second": 376.608,
|
| 2508 |
+
"eval_steps_per_second": 5.885,
|
| 2509 |
+
"step": 334710
|
| 2510 |
+
},
|
| 2511 |
+
{
|
| 2512 |
+
"epoch": 18.02,
|
| 2513 |
+
"grad_norm": 0.9260461926460266,
|
| 2514 |
+
"learning_rate": 1.0892027066784349e-05,
|
| 2515 |
+
"loss": 2.875,
|
| 2516 |
+
"step": 335000
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 18.07,
|
| 2520 |
+
"grad_norm": 0.9577099084854126,
|
| 2521 |
+
"learning_rate": 1.0597822889085025e-05,
|
| 2522 |
+
"loss": 2.8539,
|
| 2523 |
+
"step": 336000
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 18.12,
|
| 2527 |
+
"grad_norm": 0.908080518245697,
|
| 2528 |
+
"learning_rate": 1.0303912915563402e-05,
|
| 2529 |
+
"loss": 2.8556,
|
| 2530 |
+
"step": 337000
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"epoch": 18.18,
|
| 2534 |
+
"grad_norm": 0.891520082950592,
|
| 2535 |
+
"learning_rate": 1.0010002942041776e-05,
|
| 2536 |
+
"loss": 2.856,
|
| 2537 |
+
"step": 338000
|
| 2538 |
+
},
|
| 2539 |
+
{
|
| 2540 |
+
"epoch": 18.23,
|
| 2541 |
+
"grad_norm": 0.9053018093109131,
|
| 2542 |
+
"learning_rate": 9.715798764342454e-06,
|
| 2543 |
+
"loss": 2.8559,
|
| 2544 |
+
"step": 339000
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"epoch": 18.28,
|
| 2548 |
+
"grad_norm": 0.913936197757721,
|
| 2549 |
+
"learning_rate": 9.42159458664313e-06,
|
| 2550 |
+
"loss": 2.859,
|
| 2551 |
+
"step": 340000
|
| 2552 |
+
},
|
| 2553 |
+
{
|
| 2554 |
+
"epoch": 18.34,
|
| 2555 |
+
"grad_norm": 0.9781559705734253,
|
| 2556 |
+
"learning_rate": 9.127390408943808e-06,
|
| 2557 |
+
"loss": 2.8576,
|
| 2558 |
+
"step": 341000
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 18.39,
|
| 2562 |
+
"grad_norm": 0.9057850241661072,
|
| 2563 |
+
"learning_rate": 8.833480435422184e-06,
|
| 2564 |
+
"loss": 2.8599,
|
| 2565 |
+
"step": 342000
|
| 2566 |
+
},
|
| 2567 |
+
{
|
| 2568 |
+
"epoch": 18.45,
|
| 2569 |
+
"grad_norm": 0.9551451206207275,
|
| 2570 |
+
"learning_rate": 8.53927625772286e-06,
|
| 2571 |
+
"loss": 2.8593,
|
| 2572 |
+
"step": 343000
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"epoch": 18.5,
|
| 2576 |
+
"grad_norm": 0.9623215198516846,
|
| 2577 |
+
"learning_rate": 8.245366284201237e-06,
|
| 2578 |
+
"loss": 2.8576,
|
| 2579 |
+
"step": 344000
|
| 2580 |
+
},
|
| 2581 |
+
{
|
| 2582 |
+
"epoch": 18.55,
|
| 2583 |
+
"grad_norm": 0.9302039742469788,
|
| 2584 |
+
"learning_rate": 7.951162106501912e-06,
|
| 2585 |
+
"loss": 2.8575,
|
| 2586 |
+
"step": 345000
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"epoch": 18.61,
|
| 2590 |
+
"grad_norm": 0.9487703442573547,
|
| 2591 |
+
"learning_rate": 7.65695792880259e-06,
|
| 2592 |
+
"loss": 2.8563,
|
| 2593 |
+
"step": 346000
|
| 2594 |
+
},
|
| 2595 |
+
{
|
| 2596 |
+
"epoch": 18.66,
|
| 2597 |
+
"grad_norm": 0.9016742706298828,
|
| 2598 |
+
"learning_rate": 7.363342159458665e-06,
|
| 2599 |
+
"loss": 2.8611,
|
| 2600 |
+
"step": 347000
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 18.71,
|
| 2604 |
+
"grad_norm": 0.9533397555351257,
|
| 2605 |
+
"learning_rate": 7.0691379817593416e-06,
|
| 2606 |
+
"loss": 2.8587,
|
| 2607 |
+
"step": 348000
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 18.77,
|
| 2611 |
+
"grad_norm": 0.9059662818908691,
|
| 2612 |
+
"learning_rate": 6.774933804060019e-06,
|
| 2613 |
+
"loss": 2.8612,
|
| 2614 |
+
"step": 349000
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"epoch": 18.82,
|
| 2618 |
+
"grad_norm": 0.9284035563468933,
|
| 2619 |
+
"learning_rate": 6.480729626360695e-06,
|
| 2620 |
+
"loss": 2.8591,
|
| 2621 |
+
"step": 350000
|
| 2622 |
+
},
|
| 2623 |
+
{
|
| 2624 |
+
"epoch": 18.88,
|
| 2625 |
+
"grad_norm": 0.9319166541099548,
|
| 2626 |
+
"learning_rate": 6.186819652839071e-06,
|
| 2627 |
+
"loss": 2.8574,
|
| 2628 |
+
"step": 351000
|
| 2629 |
+
},
|
| 2630 |
+
{
|
| 2631 |
+
"epoch": 18.93,
|
| 2632 |
+
"grad_norm": 0.914257287979126,
|
| 2633 |
+
"learning_rate": 5.892615475139747e-06,
|
| 2634 |
+
"loss": 2.8643,
|
| 2635 |
+
"step": 352000
|
| 2636 |
+
},
|
| 2637 |
+
{
|
| 2638 |
+
"epoch": 18.98,
|
| 2639 |
+
"grad_norm": 0.9576272964477539,
|
| 2640 |
+
"learning_rate": 5.5984112974404235e-06,
|
| 2641 |
+
"loss": 2.8627,
|
| 2642 |
+
"step": 353000
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 19.0,
|
| 2646 |
+
"eval_accuracy": 0.4057549899811393,
|
| 2647 |
+
"eval_loss": 3.409109354019165,
|
| 2648 |
+
"eval_runtime": 153.8346,
|
| 2649 |
+
"eval_samples_per_second": 376.495,
|
| 2650 |
+
"eval_steps_per_second": 5.883,
|
| 2651 |
+
"step": 353305
|
| 2652 |
+
},
|
| 2653 |
+
{
|
| 2654 |
+
"epoch": 19.04,
|
| 2655 |
+
"grad_norm": 0.943903386592865,
|
| 2656 |
+
"learning_rate": 5.3042071197411005e-06,
|
| 2657 |
+
"loss": 2.8517,
|
| 2658 |
+
"step": 354000
|
| 2659 |
+
},
|
| 2660 |
+
{
|
| 2661 |
+
"epoch": 19.09,
|
| 2662 |
+
"grad_norm": 0.9372676014900208,
|
| 2663 |
+
"learning_rate": 5.010002942041777e-06,
|
| 2664 |
+
"loss": 2.8454,
|
| 2665 |
+
"step": 355000
|
| 2666 |
+
},
|
| 2667 |
+
{
|
| 2668 |
+
"epoch": 19.14,
|
| 2669 |
+
"grad_norm": 0.9114904999732971,
|
| 2670 |
+
"learning_rate": 4.716092968520153e-06,
|
| 2671 |
+
"loss": 2.8465,
|
| 2672 |
+
"step": 356000
|
| 2673 |
+
},
|
| 2674 |
+
{
|
| 2675 |
+
"epoch": 19.2,
|
| 2676 |
+
"grad_norm": 0.939071774482727,
|
| 2677 |
+
"learning_rate": 4.42188879082083e-06,
|
| 2678 |
+
"loss": 2.8383,
|
| 2679 |
+
"step": 357000
|
| 2680 |
+
},
|
| 2681 |
+
{
|
| 2682 |
+
"epoch": 19.25,
|
| 2683 |
+
"grad_norm": 0.9586918950080872,
|
| 2684 |
+
"learning_rate": 4.127978817299206e-06,
|
| 2685 |
+
"loss": 2.844,
|
| 2686 |
+
"step": 358000
|
| 2687 |
+
},
|
| 2688 |
+
{
|
| 2689 |
+
"epoch": 19.31,
|
| 2690 |
+
"grad_norm": 0.951808750629425,
|
| 2691 |
+
"learning_rate": 3.834068843777581e-06,
|
| 2692 |
+
"loss": 2.8473,
|
| 2693 |
+
"step": 359000
|
| 2694 |
+
},
|
| 2695 |
+
{
|
| 2696 |
+
"epoch": 19.36,
|
| 2697 |
+
"grad_norm": 0.9387649893760681,
|
| 2698 |
+
"learning_rate": 3.5398646660782584e-06,
|
| 2699 |
+
"loss": 2.8454,
|
| 2700 |
+
"step": 360000
|
| 2701 |
+
},
|
| 2702 |
+
{
|
| 2703 |
+
"epoch": 19.41,
|
| 2704 |
+
"grad_norm": 0.954631507396698,
|
| 2705 |
+
"learning_rate": 3.245660488378935e-06,
|
| 2706 |
+
"loss": 2.8457,
|
| 2707 |
+
"step": 361000
|
| 2708 |
+
},
|
| 2709 |
+
{
|
| 2710 |
+
"epoch": 19.47,
|
| 2711 |
+
"grad_norm": 0.9743807315826416,
|
| 2712 |
+
"learning_rate": 2.951750514857311e-06,
|
| 2713 |
+
"loss": 2.8437,
|
| 2714 |
+
"step": 362000
|
| 2715 |
+
},
|
| 2716 |
+
{
|
| 2717 |
+
"epoch": 19.52,
|
| 2718 |
+
"grad_norm": 0.9627756476402283,
|
| 2719 |
+
"learning_rate": 2.6575463371579876e-06,
|
| 2720 |
+
"loss": 2.8458,
|
| 2721 |
+
"step": 363000
|
| 2722 |
+
},
|
| 2723 |
+
{
|
| 2724 |
+
"epoch": 19.58,
|
| 2725 |
+
"grad_norm": 0.8932241797447205,
|
| 2726 |
+
"learning_rate": 2.3636363636363636e-06,
|
| 2727 |
+
"loss": 2.8447,
|
| 2728 |
+
"step": 364000
|
| 2729 |
+
},
|
| 2730 |
+
{
|
| 2731 |
+
"epoch": 19.63,
|
| 2732 |
+
"grad_norm": 0.943067193031311,
|
| 2733 |
+
"learning_rate": 2.0694321859370406e-06,
|
| 2734 |
+
"loss": 2.8484,
|
| 2735 |
+
"step": 365000
|
| 2736 |
+
},
|
| 2737 |
+
{
|
| 2738 |
+
"epoch": 19.68,
|
| 2739 |
+
"grad_norm": 0.9151442050933838,
|
| 2740 |
+
"learning_rate": 1.7752280082377172e-06,
|
| 2741 |
+
"loss": 2.8481,
|
| 2742 |
+
"step": 366000
|
| 2743 |
+
},
|
| 2744 |
+
{
|
| 2745 |
+
"epoch": 19.74,
|
| 2746 |
+
"grad_norm": 0.9441515803337097,
|
| 2747 |
+
"learning_rate": 1.4810238305383938e-06,
|
| 2748 |
+
"loss": 2.8472,
|
| 2749 |
+
"step": 367000
|
| 2750 |
+
},
|
| 2751 |
+
{
|
| 2752 |
+
"epoch": 19.79,
|
| 2753 |
+
"grad_norm": 0.9662760496139526,
|
| 2754 |
+
"learning_rate": 1.1868196528390704e-06,
|
| 2755 |
+
"loss": 2.8477,
|
| 2756 |
+
"step": 368000
|
| 2757 |
+
},
|
| 2758 |
+
{
|
| 2759 |
+
"epoch": 19.84,
|
| 2760 |
+
"grad_norm": 0.9570898413658142,
|
| 2761 |
+
"learning_rate": 8.932038834951457e-07,
|
| 2762 |
+
"loss": 2.8465,
|
| 2763 |
+
"step": 369000
|
| 2764 |
+
},
|
| 2765 |
+
{
|
| 2766 |
+
"epoch": 19.9,
|
| 2767 |
+
"grad_norm": 0.9315906167030334,
|
| 2768 |
+
"learning_rate": 5.989997057958223e-07,
|
| 2769 |
+
"loss": 2.8504,
|
| 2770 |
+
"step": 370000
|
| 2771 |
+
},
|
| 2772 |
+
{
|
| 2773 |
+
"epoch": 19.95,
|
| 2774 |
+
"grad_norm": 0.954544723033905,
|
| 2775 |
+
"learning_rate": 3.04795528096499e-07,
|
| 2776 |
+
"loss": 2.8505,
|
| 2777 |
+
"step": 371000
|
| 2778 |
+
},
|
| 2779 |
+
{
|
| 2780 |
+
"epoch": 20.0,
|
| 2781 |
+
"eval_accuracy": 0.40580987772461885,
|
| 2782 |
+
"eval_loss": 3.4122142791748047,
|
| 2783 |
+
"eval_runtime": 153.8391,
|
| 2784 |
+
"eval_samples_per_second": 376.484,
|
| 2785 |
+
"eval_steps_per_second": 5.883,
|
| 2786 |
+
"step": 371900
|
| 2787 |
+
},
|
| 2788 |
+
{
|
| 2789 |
+
"epoch": 20.0,
|
| 2790 |
+
"step": 371900,
|
| 2791 |
+
"total_flos": 1.56693365388288e+18,
|
| 2792 |
+
"train_loss": 3.1577374407077676,
|
| 2793 |
+
"train_runtime": 81393.9304,
|
| 2794 |
+
"train_samples_per_second": 146.212,
|
| 2795 |
+
"train_steps_per_second": 4.569
|
| 2796 |
+
}
|
| 2797 |
+
],
|
| 2798 |
+
"logging_steps": 1000,
|
| 2799 |
+
"max_steps": 371900,
|
| 2800 |
+
"num_input_tokens_seen": 0,
|
| 2801 |
+
"num_train_epochs": 20,
|
| 2802 |
+
"save_steps": 5000,
|
| 2803 |
+
"total_flos": 1.56693365388288e+18,
|
| 2804 |
+
"train_batch_size": 32,
|
| 2805 |
+
"trial_name": null,
|
| 2806 |
+
"trial_params": null
|
| 2807 |
+
}
|