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_aanns_dtanns-seed_1024-1e-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_aanns_dtanns-seed_1024-1e-4
|
| 15 |
|
| 16 |
-
This model was trained from scratch on
|
| 17 |
It achieves the following results on the evaluation set:
|
| 18 |
- Loss: 3.4264
|
| 19 |
- Accuracy: 0.4056
|
|
|
|
| 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_aanns_dtanns-seed_1024-1e-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: 0.4055823320854937
|
| 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_aanns_dtanns-seed_1024-1e-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: 3.4264
|
| 31 |
- Accuracy: 0.4056
|
all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"eval_accuracy": 0.4055823320854937,
|
| 4 |
+
"eval_loss": 3.4263744354248047,
|
| 5 |
+
"eval_runtime": 154.08,
|
| 6 |
+
"eval_samples": 57918,
|
| 7 |
+
"eval_samples_per_second": 375.896,
|
| 8 |
+
"eval_steps_per_second": 5.874,
|
| 9 |
+
"perplexity": 30.764900174885234,
|
| 10 |
+
"train_loss": 3.1576745831950372,
|
| 11 |
+
"train_runtime": 81271.0608,
|
| 12 |
+
"train_samples": 595038,
|
| 13 |
+
"train_samples_per_second": 146.433,
|
| 14 |
+
"train_steps_per_second": 4.576
|
| 15 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"eval_accuracy": 0.4055823320854937,
|
| 4 |
+
"eval_loss": 3.4263744354248047,
|
| 5 |
+
"eval_runtime": 154.08,
|
| 6 |
+
"eval_samples": 57918,
|
| 7 |
+
"eval_samples_per_second": 375.896,
|
| 8 |
+
"eval_steps_per_second": 5.874,
|
| 9 |
+
"perplexity": 30.764900174885234
|
| 10 |
+
}
|
runs/Jul07_16-13-29_phyl-ling-p01.la.utexas.edu/events.out.tfevents.1720468607.phyl-ling-p01.la.utexas.edu.2048002.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:528c3049b83af4c3b7d71f2f651a4fda67c7757d71fd6b851a559a7157ec7432
|
| 3 |
+
size 417
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"train_loss": 3.1576745831950372,
|
| 4 |
+
"train_runtime": 81271.0608,
|
| 5 |
+
"train_samples": 595038,
|
| 6 |
+
"train_samples_per_second": 146.433,
|
| 7 |
+
"train_steps_per_second": 4.576
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2807 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 20.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 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": 1.4397599697113037,
|
| 14 |
+
"learning_rate": 3.125e-06,
|
| 15 |
+
"loss": 7.5907,
|
| 16 |
+
"step": 1000
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.11,
|
| 20 |
+
"grad_norm": 0.7548799514770508,
|
| 21 |
+
"learning_rate": 6.25e-06,
|
| 22 |
+
"loss": 5.82,
|
| 23 |
+
"step": 2000
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.16,
|
| 27 |
+
"grad_norm": 0.921891987323761,
|
| 28 |
+
"learning_rate": 9.375000000000001e-06,
|
| 29 |
+
"loss": 5.3784,
|
| 30 |
+
"step": 3000
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.22,
|
| 34 |
+
"grad_norm": 0.9153477549552917,
|
| 35 |
+
"learning_rate": 1.25e-05,
|
| 36 |
+
"loss": 5.1593,
|
| 37 |
+
"step": 4000
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.27,
|
| 41 |
+
"grad_norm": 1.0300954580307007,
|
| 42 |
+
"learning_rate": 1.5625e-05,
|
| 43 |
+
"loss": 5.015,
|
| 44 |
+
"step": 5000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.32,
|
| 48 |
+
"grad_norm": 1.0274837017059326,
|
| 49 |
+
"learning_rate": 1.8750000000000002e-05,
|
| 50 |
+
"loss": 4.8818,
|
| 51 |
+
"step": 6000
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.38,
|
| 55 |
+
"grad_norm": 1.128319263458252,
|
| 56 |
+
"learning_rate": 2.1875e-05,
|
| 57 |
+
"loss": 4.757,
|
| 58 |
+
"step": 7000
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.43,
|
| 62 |
+
"grad_norm": 1.1055282354354858,
|
| 63 |
+
"learning_rate": 2.5e-05,
|
| 64 |
+
"loss": 4.6511,
|
| 65 |
+
"step": 8000
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.48,
|
| 69 |
+
"grad_norm": 1.071848750114441,
|
| 70 |
+
"learning_rate": 2.8125000000000003e-05,
|
| 71 |
+
"loss": 4.5641,
|
| 72 |
+
"step": 9000
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.54,
|
| 76 |
+
"grad_norm": 0.9956591725349426,
|
| 77 |
+
"learning_rate": 3.125e-05,
|
| 78 |
+
"loss": 4.4822,
|
| 79 |
+
"step": 10000
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.59,
|
| 83 |
+
"grad_norm": 1.0482920408248901,
|
| 84 |
+
"learning_rate": 3.4371875e-05,
|
| 85 |
+
"loss": 4.4079,
|
| 86 |
+
"step": 11000
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.65,
|
| 90 |
+
"grad_norm": 1.0127135515213013,
|
| 91 |
+
"learning_rate": 3.7496875e-05,
|
| 92 |
+
"loss": 4.3425,
|
| 93 |
+
"step": 12000
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.7,
|
| 97 |
+
"grad_norm": 1.0505046844482422,
|
| 98 |
+
"learning_rate": 4.0621875e-05,
|
| 99 |
+
"loss": 4.2861,
|
| 100 |
+
"step": 13000
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.75,
|
| 104 |
+
"grad_norm": 1.0681025981903076,
|
| 105 |
+
"learning_rate": 4.3746875000000005e-05,
|
| 106 |
+
"loss": 4.2289,
|
| 107 |
+
"step": 14000
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.81,
|
| 111 |
+
"grad_norm": 1.1013708114624023,
|
| 112 |
+
"learning_rate": 4.686875e-05,
|
| 113 |
+
"loss": 4.1873,
|
| 114 |
+
"step": 15000
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.86,
|
| 118 |
+
"grad_norm": 1.0665634870529175,
|
| 119 |
+
"learning_rate": 4.9990625000000004e-05,
|
| 120 |
+
"loss": 4.1308,
|
| 121 |
+
"step": 16000
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.91,
|
| 125 |
+
"grad_norm": 1.1269334554672241,
|
| 126 |
+
"learning_rate": 5.3115625000000005e-05,
|
| 127 |
+
"loss": 4.0923,
|
| 128 |
+
"step": 17000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.97,
|
| 132 |
+
"grad_norm": 0.9998357892036438,
|
| 133 |
+
"learning_rate": 5.62375e-05,
|
| 134 |
+
"loss": 4.0514,
|
| 135 |
+
"step": 18000
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 1.0,
|
| 139 |
+
"eval_accuracy": 0.3101297245280225,
|
| 140 |
+
"eval_loss": 4.2435173988342285,
|
| 141 |
+
"eval_runtime": 151.7646,
|
| 142 |
+
"eval_samples_per_second": 381.631,
|
| 143 |
+
"eval_steps_per_second": 5.963,
|
| 144 |
+
"step": 18595
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 1.02,
|
| 148 |
+
"grad_norm": 0.9669459462165833,
|
| 149 |
+
"learning_rate": 5.93625e-05,
|
| 150 |
+
"loss": 4.0035,
|
| 151 |
+
"step": 19000
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 1.08,
|
| 155 |
+
"grad_norm": 1.1156255006790161,
|
| 156 |
+
"learning_rate": 6.24875e-05,
|
| 157 |
+
"loss": 3.9592,
|
| 158 |
+
"step": 20000
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.13,
|
| 162 |
+
"grad_norm": 0.9864040613174438,
|
| 163 |
+
"learning_rate": 6.56125e-05,
|
| 164 |
+
"loss": 3.9219,
|
| 165 |
+
"step": 21000
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 1.18,
|
| 169 |
+
"grad_norm": 1.012749433517456,
|
| 170 |
+
"learning_rate": 6.8734375e-05,
|
| 171 |
+
"loss": 3.8856,
|
| 172 |
+
"step": 22000
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 1.24,
|
| 176 |
+
"grad_norm": 1.0206193923950195,
|
| 177 |
+
"learning_rate": 7.185937500000001e-05,
|
| 178 |
+
"loss": 3.8566,
|
| 179 |
+
"step": 23000
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.29,
|
| 183 |
+
"grad_norm": 0.9851728081703186,
|
| 184 |
+
"learning_rate": 7.498125e-05,
|
| 185 |
+
"loss": 3.8218,
|
| 186 |
+
"step": 24000
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 1.34,
|
| 190 |
+
"grad_norm": 0.9994772672653198,
|
| 191 |
+
"learning_rate": 7.810625e-05,
|
| 192 |
+
"loss": 3.7989,
|
| 193 |
+
"step": 25000
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 1.4,
|
| 197 |
+
"grad_norm": 0.9388182163238525,
|
| 198 |
+
"learning_rate": 8.1228125e-05,
|
| 199 |
+
"loss": 3.7688,
|
| 200 |
+
"step": 26000
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 1.45,
|
| 204 |
+
"grad_norm": 0.8769707083702087,
|
| 205 |
+
"learning_rate": 8.435e-05,
|
| 206 |
+
"loss": 3.7461,
|
| 207 |
+
"step": 27000
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 1.51,
|
| 211 |
+
"grad_norm": 0.9455641508102417,
|
| 212 |
+
"learning_rate": 8.747500000000001e-05,
|
| 213 |
+
"loss": 3.7314,
|
| 214 |
+
"step": 28000
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 1.56,
|
| 218 |
+
"grad_norm": 0.932089626789093,
|
| 219 |
+
"learning_rate": 9.0596875e-05,
|
| 220 |
+
"loss": 3.7058,
|
| 221 |
+
"step": 29000
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.61,
|
| 225 |
+
"grad_norm": 0.8848442435264587,
|
| 226 |
+
"learning_rate": 9.3721875e-05,
|
| 227 |
+
"loss": 3.6882,
|
| 228 |
+
"step": 30000
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 1.67,
|
| 232 |
+
"grad_norm": 0.8709959387779236,
|
| 233 |
+
"learning_rate": 9.684375e-05,
|
| 234 |
+
"loss": 3.6697,
|
| 235 |
+
"step": 31000
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 1.72,
|
| 239 |
+
"grad_norm": 0.8558521866798401,
|
| 240 |
+
"learning_rate": 9.9965625e-05,
|
| 241 |
+
"loss": 3.6505,
|
| 242 |
+
"step": 32000
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.77,
|
| 246 |
+
"grad_norm": 0.8082968592643738,
|
| 247 |
+
"learning_rate": 9.970903206825538e-05,
|
| 248 |
+
"loss": 3.6347,
|
| 249 |
+
"step": 33000
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 1.83,
|
| 253 |
+
"grad_norm": 0.849818766117096,
|
| 254 |
+
"learning_rate": 9.941512209473375e-05,
|
| 255 |
+
"loss": 3.6204,
|
| 256 |
+
"step": 34000
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 1.88,
|
| 260 |
+
"grad_norm": 0.8615933060646057,
|
| 261 |
+
"learning_rate": 9.912091791703443e-05,
|
| 262 |
+
"loss": 3.5962,
|
| 263 |
+
"step": 35000
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 1.94,
|
| 267 |
+
"grad_norm": 0.8331784009933472,
|
| 268 |
+
"learning_rate": 9.882671373933511e-05,
|
| 269 |
+
"loss": 3.5822,
|
| 270 |
+
"step": 36000
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 1.99,
|
| 274 |
+
"grad_norm": 0.8385667204856873,
|
| 275 |
+
"learning_rate": 9.853280376581348e-05,
|
| 276 |
+
"loss": 3.5672,
|
| 277 |
+
"step": 37000
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 2.0,
|
| 281 |
+
"eval_accuracy": 0.36221752126950446,
|
| 282 |
+
"eval_loss": 3.765214681625366,
|
| 283 |
+
"eval_runtime": 153.4159,
|
| 284 |
+
"eval_samples_per_second": 377.523,
|
| 285 |
+
"eval_steps_per_second": 5.899,
|
| 286 |
+
"step": 37190
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 2.04,
|
| 290 |
+
"grad_norm": 0.8255208730697632,
|
| 291 |
+
"learning_rate": 9.823859958811416e-05,
|
| 292 |
+
"loss": 3.5267,
|
| 293 |
+
"step": 38000
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 2.1,
|
| 297 |
+
"grad_norm": 0.8173959851264954,
|
| 298 |
+
"learning_rate": 9.794468961459253e-05,
|
| 299 |
+
"loss": 3.509,
|
| 300 |
+
"step": 39000
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 2.15,
|
| 304 |
+
"grad_norm": 0.826338529586792,
|
| 305 |
+
"learning_rate": 9.765048543689321e-05,
|
| 306 |
+
"loss": 3.5054,
|
| 307 |
+
"step": 40000
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 2.2,
|
| 311 |
+
"grad_norm": 0.8249008655548096,
|
| 312 |
+
"learning_rate": 9.735628125919388e-05,
|
| 313 |
+
"loss": 3.4969,
|
| 314 |
+
"step": 41000
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 2.26,
|
| 318 |
+
"grad_norm": 0.7982118725776672,
|
| 319 |
+
"learning_rate": 9.706207708149456e-05,
|
| 320 |
+
"loss": 3.4873,
|
| 321 |
+
"step": 42000
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 2.31,
|
| 325 |
+
"grad_norm": 0.796518862247467,
|
| 326 |
+
"learning_rate": 9.676816710797295e-05,
|
| 327 |
+
"loss": 3.4752,
|
| 328 |
+
"step": 43000
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 2.37,
|
| 332 |
+
"grad_norm": 0.7944287061691284,
|
| 333 |
+
"learning_rate": 9.647396293027361e-05,
|
| 334 |
+
"loss": 3.4747,
|
| 335 |
+
"step": 44000
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 2.42,
|
| 339 |
+
"grad_norm": 0.801404595375061,
|
| 340 |
+
"learning_rate": 9.61797587525743e-05,
|
| 341 |
+
"loss": 3.464,
|
| 342 |
+
"step": 45000
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 2.47,
|
| 346 |
+
"grad_norm": 0.782745361328125,
|
| 347 |
+
"learning_rate": 9.588584877905267e-05,
|
| 348 |
+
"loss": 3.4534,
|
| 349 |
+
"step": 46000
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 2.53,
|
| 353 |
+
"grad_norm": 0.9451091289520264,
|
| 354 |
+
"learning_rate": 9.559164460135335e-05,
|
| 355 |
+
"loss": 3.4443,
|
| 356 |
+
"step": 47000
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.58,
|
| 360 |
+
"grad_norm": 0.7674155235290527,
|
| 361 |
+
"learning_rate": 9.529773462783173e-05,
|
| 362 |
+
"loss": 3.4388,
|
| 363 |
+
"step": 48000
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 2.64,
|
| 367 |
+
"grad_norm": 0.7728825211524963,
|
| 368 |
+
"learning_rate": 9.50035304501324e-05,
|
| 369 |
+
"loss": 3.4275,
|
| 370 |
+
"step": 49000
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 2.69,
|
| 374 |
+
"grad_norm": 0.7956791520118713,
|
| 375 |
+
"learning_rate": 9.470932627243308e-05,
|
| 376 |
+
"loss": 3.4232,
|
| 377 |
+
"step": 50000
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 2.74,
|
| 381 |
+
"grad_norm": 0.799748420715332,
|
| 382 |
+
"learning_rate": 9.441541629891145e-05,
|
| 383 |
+
"loss": 3.4107,
|
| 384 |
+
"step": 51000
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 2.8,
|
| 388 |
+
"grad_norm": 0.7880944013595581,
|
| 389 |
+
"learning_rate": 9.412121212121212e-05,
|
| 390 |
+
"loss": 3.4125,
|
| 391 |
+
"step": 52000
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 2.85,
|
| 395 |
+
"grad_norm": 0.7877218127250671,
|
| 396 |
+
"learning_rate": 9.38273021476905e-05,
|
| 397 |
+
"loss": 3.4058,
|
| 398 |
+
"step": 53000
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 2.9,
|
| 402 |
+
"grad_norm": 0.784431517124176,
|
| 403 |
+
"learning_rate": 9.353309796999118e-05,
|
| 404 |
+
"loss": 3.3989,
|
| 405 |
+
"step": 54000
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 2.96,
|
| 409 |
+
"grad_norm": 0.7350291609764099,
|
| 410 |
+
"learning_rate": 9.323889379229186e-05,
|
| 411 |
+
"loss": 3.3933,
|
| 412 |
+
"step": 55000
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 3.0,
|
| 416 |
+
"eval_accuracy": 0.37918697076492014,
|
| 417 |
+
"eval_loss": 3.5858614444732666,
|
| 418 |
+
"eval_runtime": 153.7959,
|
| 419 |
+
"eval_samples_per_second": 376.59,
|
| 420 |
+
"eval_steps_per_second": 5.884,
|
| 421 |
+
"step": 55785
|
| 422 |
+
},
|
| 423 |
+
{
|
| 424 |
+
"epoch": 3.01,
|
| 425 |
+
"grad_norm": 0.7721139788627625,
|
| 426 |
+
"learning_rate": 9.294468961459253e-05,
|
| 427 |
+
"loss": 3.3787,
|
| 428 |
+
"step": 56000
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 3.07,
|
| 432 |
+
"grad_norm": 0.7534685730934143,
|
| 433 |
+
"learning_rate": 9.26507796410709e-05,
|
| 434 |
+
"loss": 3.3358,
|
| 435 |
+
"step": 57000
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 3.12,
|
| 439 |
+
"grad_norm": 0.7684205770492554,
|
| 440 |
+
"learning_rate": 9.235657546337158e-05,
|
| 441 |
+
"loss": 3.3365,
|
| 442 |
+
"step": 58000
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 3.17,
|
| 446 |
+
"grad_norm": 0.7345510125160217,
|
| 447 |
+
"learning_rate": 9.206237128567227e-05,
|
| 448 |
+
"loss": 3.3292,
|
| 449 |
+
"step": 59000
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 3.23,
|
| 453 |
+
"grad_norm": 0.772286593914032,
|
| 454 |
+
"learning_rate": 9.176846131215064e-05,
|
| 455 |
+
"loss": 3.3285,
|
| 456 |
+
"step": 60000
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 3.28,
|
| 460 |
+
"grad_norm": 0.7555109262466431,
|
| 461 |
+
"learning_rate": 9.147425713445132e-05,
|
| 462 |
+
"loss": 3.3297,
|
| 463 |
+
"step": 61000
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 3.33,
|
| 467 |
+
"grad_norm": 0.7533125281333923,
|
| 468 |
+
"learning_rate": 9.118034716092969e-05,
|
| 469 |
+
"loss": 3.3248,
|
| 470 |
+
"step": 62000
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 3.39,
|
| 474 |
+
"grad_norm": 0.7516513466835022,
|
| 475 |
+
"learning_rate": 9.088614298323037e-05,
|
| 476 |
+
"loss": 3.3264,
|
| 477 |
+
"step": 63000
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 3.44,
|
| 481 |
+
"grad_norm": 0.7214173078536987,
|
| 482 |
+
"learning_rate": 9.059193880553104e-05,
|
| 483 |
+
"loss": 3.3206,
|
| 484 |
+
"step": 64000
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 3.5,
|
| 488 |
+
"grad_norm": 0.7576092481613159,
|
| 489 |
+
"learning_rate": 9.029802883200942e-05,
|
| 490 |
+
"loss": 3.316,
|
| 491 |
+
"step": 65000
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 3.55,
|
| 495 |
+
"grad_norm": 0.7526379823684692,
|
| 496 |
+
"learning_rate": 9.000411885848779e-05,
|
| 497 |
+
"loss": 3.313,
|
| 498 |
+
"step": 66000
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 3.6,
|
| 502 |
+
"grad_norm": 0.7402895092964172,
|
| 503 |
+
"learning_rate": 8.970991468078847e-05,
|
| 504 |
+
"loss": 3.3134,
|
| 505 |
+
"step": 67000
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 3.66,
|
| 509 |
+
"grad_norm": 0.7132677435874939,
|
| 510 |
+
"learning_rate": 8.941571050308914e-05,
|
| 511 |
+
"loss": 3.3069,
|
| 512 |
+
"step": 68000
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 3.71,
|
| 516 |
+
"grad_norm": 0.8066958785057068,
|
| 517 |
+
"learning_rate": 8.912150632538982e-05,
|
| 518 |
+
"loss": 3.307,
|
| 519 |
+
"step": 69000
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 3.76,
|
| 523 |
+
"grad_norm": 0.7379071712493896,
|
| 524 |
+
"learning_rate": 8.88273021476905e-05,
|
| 525 |
+
"loss": 3.302,
|
| 526 |
+
"step": 70000
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 3.82,
|
| 530 |
+
"grad_norm": 0.7558432817459106,
|
| 531 |
+
"learning_rate": 8.853339217416887e-05,
|
| 532 |
+
"loss": 3.3019,
|
| 533 |
+
"step": 71000
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 3.87,
|
| 537 |
+
"grad_norm": 0.7654557228088379,
|
| 538 |
+
"learning_rate": 8.823918799646955e-05,
|
| 539 |
+
"loss": 3.2993,
|
| 540 |
+
"step": 72000
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 3.93,
|
| 544 |
+
"grad_norm": 0.7307350635528564,
|
| 545 |
+
"learning_rate": 8.794527802294792e-05,
|
| 546 |
+
"loss": 3.2961,
|
| 547 |
+
"step": 73000
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 3.98,
|
| 551 |
+
"grad_norm": 0.7173799872398376,
|
| 552 |
+
"learning_rate": 8.76510738452486e-05,
|
| 553 |
+
"loss": 3.2939,
|
| 554 |
+
"step": 74000
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 4.0,
|
| 558 |
+
"eval_accuracy": 0.3863495861517887,
|
| 559 |
+
"eval_loss": 3.5397000312805176,
|
| 560 |
+
"eval_runtime": 153.3856,
|
| 561 |
+
"eval_samples_per_second": 377.597,
|
| 562 |
+
"eval_steps_per_second": 5.9,
|
| 563 |
+
"step": 74380
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 4.03,
|
| 567 |
+
"grad_norm": 0.7380111217498779,
|
| 568 |
+
"learning_rate": 8.735686966754927e-05,
|
| 569 |
+
"loss": 3.2551,
|
| 570 |
+
"step": 75000
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 4.09,
|
| 574 |
+
"grad_norm": 0.7536616921424866,
|
| 575 |
+
"learning_rate": 8.706266548984995e-05,
|
| 576 |
+
"loss": 3.2352,
|
| 577 |
+
"step": 76000
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 4.14,
|
| 581 |
+
"grad_norm": 0.7641943097114563,
|
| 582 |
+
"learning_rate": 8.676875551632834e-05,
|
| 583 |
+
"loss": 3.2392,
|
| 584 |
+
"step": 77000
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 4.19,
|
| 588 |
+
"grad_norm": 0.7541372776031494,
|
| 589 |
+
"learning_rate": 8.647455133862902e-05,
|
| 590 |
+
"loss": 3.2448,
|
| 591 |
+
"step": 78000
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 4.25,
|
| 595 |
+
"grad_norm": 0.7148016691207886,
|
| 596 |
+
"learning_rate": 8.618064136510739e-05,
|
| 597 |
+
"loss": 3.2369,
|
| 598 |
+
"step": 79000
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 4.3,
|
| 602 |
+
"grad_norm": 0.7816110253334045,
|
| 603 |
+
"learning_rate": 8.588643718740806e-05,
|
| 604 |
+
"loss": 3.238,
|
| 605 |
+
"step": 80000
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 4.36,
|
| 609 |
+
"grad_norm": 0.7324749231338501,
|
| 610 |
+
"learning_rate": 8.559223300970874e-05,
|
| 611 |
+
"loss": 3.2387,
|
| 612 |
+
"step": 81000
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 4.41,
|
| 616 |
+
"grad_norm": 0.7476484179496765,
|
| 617 |
+
"learning_rate": 8.529802883200941e-05,
|
| 618 |
+
"loss": 3.2402,
|
| 619 |
+
"step": 82000
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 4.46,
|
| 623 |
+
"grad_norm": 0.7432982325553894,
|
| 624 |
+
"learning_rate": 8.500382465431009e-05,
|
| 625 |
+
"loss": 3.2346,
|
| 626 |
+
"step": 83000
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 4.52,
|
| 630 |
+
"grad_norm": 0.7481192350387573,
|
| 631 |
+
"learning_rate": 8.470991468078847e-05,
|
| 632 |
+
"loss": 3.2354,
|
| 633 |
+
"step": 84000
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 4.57,
|
| 637 |
+
"grad_norm": 0.7329268455505371,
|
| 638 |
+
"learning_rate": 8.441571050308915e-05,
|
| 639 |
+
"loss": 3.2349,
|
| 640 |
+
"step": 85000
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 4.62,
|
| 644 |
+
"grad_norm": 0.7217251062393188,
|
| 645 |
+
"learning_rate": 8.412150632538982e-05,
|
| 646 |
+
"loss": 3.2331,
|
| 647 |
+
"step": 86000
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 4.68,
|
| 651 |
+
"grad_norm": 0.702445387840271,
|
| 652 |
+
"learning_rate": 8.382759635186819e-05,
|
| 653 |
+
"loss": 3.2324,
|
| 654 |
+
"step": 87000
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 4.73,
|
| 658 |
+
"grad_norm": 0.7485525012016296,
|
| 659 |
+
"learning_rate": 8.353339217416887e-05,
|
| 660 |
+
"loss": 3.2318,
|
| 661 |
+
"step": 88000
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 4.79,
|
| 665 |
+
"grad_norm": 0.7345052361488342,
|
| 666 |
+
"learning_rate": 8.323948220064726e-05,
|
| 667 |
+
"loss": 3.2277,
|
| 668 |
+
"step": 89000
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 4.84,
|
| 672 |
+
"grad_norm": 0.7187086343765259,
|
| 673 |
+
"learning_rate": 8.294527802294793e-05,
|
| 674 |
+
"loss": 3.2262,
|
| 675 |
+
"step": 90000
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 4.89,
|
| 679 |
+
"grad_norm": 0.7242982983589172,
|
| 680 |
+
"learning_rate": 8.26510738452486e-05,
|
| 681 |
+
"loss": 3.2304,
|
| 682 |
+
"step": 91000
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 4.95,
|
| 686 |
+
"grad_norm": 0.7377817630767822,
|
| 687 |
+
"learning_rate": 8.235716387172698e-05,
|
| 688 |
+
"loss": 3.2248,
|
| 689 |
+
"step": 92000
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 5.0,
|
| 693 |
+
"eval_accuracy": 0.3919453815222192,
|
| 694 |
+
"eval_loss": 3.4727883338928223,
|
| 695 |
+
"eval_runtime": 153.4803,
|
| 696 |
+
"eval_samples_per_second": 377.364,
|
| 697 |
+
"eval_steps_per_second": 5.897,
|
| 698 |
+
"step": 92975
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 5.0,
|
| 702 |
+
"grad_norm": 0.7362983822822571,
|
| 703 |
+
"learning_rate": 8.206295969402766e-05,
|
| 704 |
+
"loss": 3.2199,
|
| 705 |
+
"step": 93000
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 5.06,
|
| 709 |
+
"grad_norm": 0.7538195848464966,
|
| 710 |
+
"learning_rate": 8.176904972050603e-05,
|
| 711 |
+
"loss": 3.1732,
|
| 712 |
+
"step": 94000
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 5.11,
|
| 716 |
+
"grad_norm": 0.7742282748222351,
|
| 717 |
+
"learning_rate": 8.147484554280671e-05,
|
| 718 |
+
"loss": 3.1708,
|
| 719 |
+
"step": 95000
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 5.16,
|
| 723 |
+
"grad_norm": 0.7649984955787659,
|
| 724 |
+
"learning_rate": 8.118064136510739e-05,
|
| 725 |
+
"loss": 3.17,
|
| 726 |
+
"step": 96000
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 5.22,
|
| 730 |
+
"grad_norm": 0.7696561217308044,
|
| 731 |
+
"learning_rate": 8.088643718740806e-05,
|
| 732 |
+
"loss": 3.1709,
|
| 733 |
+
"step": 97000
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 5.27,
|
| 737 |
+
"grad_norm": 0.7525352835655212,
|
| 738 |
+
"learning_rate": 8.059252721388644e-05,
|
| 739 |
+
"loss": 3.1775,
|
| 740 |
+
"step": 98000
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 5.32,
|
| 744 |
+
"grad_norm": 0.7478598952293396,
|
| 745 |
+
"learning_rate": 8.029832303618712e-05,
|
| 746 |
+
"loss": 3.1765,
|
| 747 |
+
"step": 99000
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 5.38,
|
| 751 |
+
"grad_norm": 0.717166543006897,
|
| 752 |
+
"learning_rate": 8.000411885848779e-05,
|
| 753 |
+
"loss": 3.1735,
|
| 754 |
+
"step": 100000
|
| 755 |
+
},
|
| 756 |
+
{
|
| 757 |
+
"epoch": 5.43,
|
| 758 |
+
"grad_norm": 0.743232786655426,
|
| 759 |
+
"learning_rate": 7.971020888496618e-05,
|
| 760 |
+
"loss": 3.1801,
|
| 761 |
+
"step": 101000
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 5.49,
|
| 765 |
+
"grad_norm": 0.7384679913520813,
|
| 766 |
+
"learning_rate": 7.941600470726684e-05,
|
| 767 |
+
"loss": 3.1796,
|
| 768 |
+
"step": 102000
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 5.54,
|
| 772 |
+
"grad_norm": 0.7780394554138184,
|
| 773 |
+
"learning_rate": 7.912209473374523e-05,
|
| 774 |
+
"loss": 3.1745,
|
| 775 |
+
"step": 103000
|
| 776 |
+
},
|
| 777 |
+
{
|
| 778 |
+
"epoch": 5.59,
|
| 779 |
+
"grad_norm": 0.7568414211273193,
|
| 780 |
+
"learning_rate": 7.88281847602236e-05,
|
| 781 |
+
"loss": 3.1776,
|
| 782 |
+
"step": 104000
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 5.65,
|
| 786 |
+
"grad_norm": 0.7363548278808594,
|
| 787 |
+
"learning_rate": 7.853398058252428e-05,
|
| 788 |
+
"loss": 3.176,
|
| 789 |
+
"step": 105000
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 5.7,
|
| 793 |
+
"grad_norm": 0.7693814635276794,
|
| 794 |
+
"learning_rate": 7.824007060900265e-05,
|
| 795 |
+
"loss": 3.1777,
|
| 796 |
+
"step": 106000
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"epoch": 5.75,
|
| 800 |
+
"grad_norm": 0.7400150299072266,
|
| 801 |
+
"learning_rate": 7.794586643130333e-05,
|
| 802 |
+
"loss": 3.1755,
|
| 803 |
+
"step": 107000
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 5.81,
|
| 807 |
+
"grad_norm": 0.7456902861595154,
|
| 808 |
+
"learning_rate": 7.76519564577817e-05,
|
| 809 |
+
"loss": 3.1769,
|
| 810 |
+
"step": 108000
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 5.86,
|
| 814 |
+
"grad_norm": 0.7355201244354248,
|
| 815 |
+
"learning_rate": 7.735775228008238e-05,
|
| 816 |
+
"loss": 3.1732,
|
| 817 |
+
"step": 109000
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 5.92,
|
| 821 |
+
"grad_norm": 0.6928702592849731,
|
| 822 |
+
"learning_rate": 7.706354810238305e-05,
|
| 823 |
+
"loss": 3.1744,
|
| 824 |
+
"step": 110000
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 5.97,
|
| 828 |
+
"grad_norm": 0.7330304384231567,
|
| 829 |
+
"learning_rate": 7.676934392468373e-05,
|
| 830 |
+
"loss": 3.173,
|
| 831 |
+
"step": 111000
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 6.0,
|
| 835 |
+
"eval_accuracy": 0.3949880570450938,
|
| 836 |
+
"eval_loss": 3.467240571975708,
|
| 837 |
+
"eval_runtime": 153.7673,
|
| 838 |
+
"eval_samples_per_second": 376.66,
|
| 839 |
+
"eval_steps_per_second": 5.886,
|
| 840 |
+
"step": 111570
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 6.02,
|
| 844 |
+
"grad_norm": 0.7390985488891602,
|
| 845 |
+
"learning_rate": 7.647543395116212e-05,
|
| 846 |
+
"loss": 3.1468,
|
| 847 |
+
"step": 112000
|
| 848 |
+
},
|
| 849 |
+
{
|
| 850 |
+
"epoch": 6.08,
|
| 851 |
+
"grad_norm": 0.775302529335022,
|
| 852 |
+
"learning_rate": 7.618152397764049e-05,
|
| 853 |
+
"loss": 3.1168,
|
| 854 |
+
"step": 113000
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 6.13,
|
| 858 |
+
"grad_norm": 0.7603143453598022,
|
| 859 |
+
"learning_rate": 7.588731979994115e-05,
|
| 860 |
+
"loss": 3.1219,
|
| 861 |
+
"step": 114000
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 6.18,
|
| 865 |
+
"grad_norm": 0.7535637617111206,
|
| 866 |
+
"learning_rate": 7.559311562224184e-05,
|
| 867 |
+
"loss": 3.123,
|
| 868 |
+
"step": 115000
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 6.24,
|
| 872 |
+
"grad_norm": 0.7628033757209778,
|
| 873 |
+
"learning_rate": 7.529891144454252e-05,
|
| 874 |
+
"loss": 3.1289,
|
| 875 |
+
"step": 116000
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 6.29,
|
| 879 |
+
"grad_norm": 0.7593590617179871,
|
| 880 |
+
"learning_rate": 7.50050014710209e-05,
|
| 881 |
+
"loss": 3.124,
|
| 882 |
+
"step": 117000
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 6.35,
|
| 886 |
+
"grad_norm": 0.7466164231300354,
|
| 887 |
+
"learning_rate": 7.471109149749926e-05,
|
| 888 |
+
"loss": 3.126,
|
| 889 |
+
"step": 118000
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 6.4,
|
| 893 |
+
"grad_norm": 0.7527124881744385,
|
| 894 |
+
"learning_rate": 7.441688731979994e-05,
|
| 895 |
+
"loss": 3.1335,
|
| 896 |
+
"step": 119000
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 6.45,
|
| 900 |
+
"grad_norm": 0.7380790114402771,
|
| 901 |
+
"learning_rate": 7.412268314210062e-05,
|
| 902 |
+
"loss": 3.1333,
|
| 903 |
+
"step": 120000
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 6.51,
|
| 907 |
+
"grad_norm": 0.7652768492698669,
|
| 908 |
+
"learning_rate": 7.382847896440129e-05,
|
| 909 |
+
"loss": 3.1307,
|
| 910 |
+
"step": 121000
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 6.56,
|
| 914 |
+
"grad_norm": 0.7151010036468506,
|
| 915 |
+
"learning_rate": 7.353456899087967e-05,
|
| 916 |
+
"loss": 3.1341,
|
| 917 |
+
"step": 122000
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 6.61,
|
| 921 |
+
"grad_norm": 0.75682532787323,
|
| 922 |
+
"learning_rate": 7.324065901735804e-05,
|
| 923 |
+
"loss": 3.1289,
|
| 924 |
+
"step": 123000
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 6.67,
|
| 928 |
+
"grad_norm": 0.762443482875824,
|
| 929 |
+
"learning_rate": 7.294645483965872e-05,
|
| 930 |
+
"loss": 3.13,
|
| 931 |
+
"step": 124000
|
| 932 |
+
},
|
| 933 |
+
{
|
| 934 |
+
"epoch": 6.72,
|
| 935 |
+
"grad_norm": 0.7441132068634033,
|
| 936 |
+
"learning_rate": 7.265225066195939e-05,
|
| 937 |
+
"loss": 3.1321,
|
| 938 |
+
"step": 125000
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 6.78,
|
| 942 |
+
"grad_norm": 0.7210931777954102,
|
| 943 |
+
"learning_rate": 7.235834068843778e-05,
|
| 944 |
+
"loss": 3.1297,
|
| 945 |
+
"step": 126000
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 6.83,
|
| 949 |
+
"grad_norm": 0.757232129573822,
|
| 950 |
+
"learning_rate": 7.206413651073846e-05,
|
| 951 |
+
"loss": 3.1349,
|
| 952 |
+
"step": 127000
|
| 953 |
+
},
|
| 954 |
+
{
|
| 955 |
+
"epoch": 6.88,
|
| 956 |
+
"grad_norm": 0.7122140526771545,
|
| 957 |
+
"learning_rate": 7.176993233303914e-05,
|
| 958 |
+
"loss": 3.1319,
|
| 959 |
+
"step": 128000
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 6.94,
|
| 963 |
+
"grad_norm": 0.7202630639076233,
|
| 964 |
+
"learning_rate": 7.14757281553398e-05,
|
| 965 |
+
"loss": 3.1335,
|
| 966 |
+
"step": 129000
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 6.99,
|
| 970 |
+
"grad_norm": 0.7444923520088196,
|
| 971 |
+
"learning_rate": 7.118181818181818e-05,
|
| 972 |
+
"loss": 3.1332,
|
| 973 |
+
"step": 130000
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 7.0,
|
| 977 |
+
"eval_accuracy": 0.39866983551009927,
|
| 978 |
+
"eval_loss": 3.424886465072632,
|
| 979 |
+
"eval_runtime": 153.9395,
|
| 980 |
+
"eval_samples_per_second": 376.239,
|
| 981 |
+
"eval_steps_per_second": 5.879,
|
| 982 |
+
"step": 130165
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 7.04,
|
| 986 |
+
"grad_norm": 0.7459632754325867,
|
| 987 |
+
"learning_rate": 7.088790820829656e-05,
|
| 988 |
+
"loss": 3.0788,
|
| 989 |
+
"step": 131000
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 7.1,
|
| 993 |
+
"grad_norm": 0.781525194644928,
|
| 994 |
+
"learning_rate": 7.059370403059724e-05,
|
| 995 |
+
"loss": 3.0751,
|
| 996 |
+
"step": 132000
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 7.15,
|
| 1000 |
+
"grad_norm": 0.7643530368804932,
|
| 1001 |
+
"learning_rate": 7.029979405707561e-05,
|
| 1002 |
+
"loss": 3.0821,
|
| 1003 |
+
"step": 133000
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 7.21,
|
| 1007 |
+
"grad_norm": 0.7509598731994629,
|
| 1008 |
+
"learning_rate": 7.000558987937629e-05,
|
| 1009 |
+
"loss": 3.0821,
|
| 1010 |
+
"step": 134000
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 7.26,
|
| 1014 |
+
"grad_norm": 0.7690565586090088,
|
| 1015 |
+
"learning_rate": 6.971138570167697e-05,
|
| 1016 |
+
"loss": 3.0853,
|
| 1017 |
+
"step": 135000
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 7.31,
|
| 1021 |
+
"grad_norm": 0.76784348487854,
|
| 1022 |
+
"learning_rate": 6.941718152397764e-05,
|
| 1023 |
+
"loss": 3.0896,
|
| 1024 |
+
"step": 136000
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 7.37,
|
| 1028 |
+
"grad_norm": 0.756600558757782,
|
| 1029 |
+
"learning_rate": 6.912356575463373e-05,
|
| 1030 |
+
"loss": 3.0854,
|
| 1031 |
+
"step": 137000
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 7.42,
|
| 1035 |
+
"grad_norm": 0.7614617347717285,
|
| 1036 |
+
"learning_rate": 6.88293615769344e-05,
|
| 1037 |
+
"loss": 3.0915,
|
| 1038 |
+
"step": 138000
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 7.48,
|
| 1042 |
+
"grad_norm": 0.763688862323761,
|
| 1043 |
+
"learning_rate": 6.853515739923508e-05,
|
| 1044 |
+
"loss": 3.0949,
|
| 1045 |
+
"step": 139000
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 7.53,
|
| 1049 |
+
"grad_norm": 0.7517523169517517,
|
| 1050 |
+
"learning_rate": 6.824124742571345e-05,
|
| 1051 |
+
"loss": 3.0958,
|
| 1052 |
+
"step": 140000
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 7.58,
|
| 1056 |
+
"grad_norm": 0.7513327598571777,
|
| 1057 |
+
"learning_rate": 6.794704324801413e-05,
|
| 1058 |
+
"loss": 3.0964,
|
| 1059 |
+
"step": 141000
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 7.64,
|
| 1063 |
+
"grad_norm": 0.7641260623931885,
|
| 1064 |
+
"learning_rate": 6.76528390703148e-05,
|
| 1065 |
+
"loss": 3.0958,
|
| 1066 |
+
"step": 142000
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 7.69,
|
| 1070 |
+
"grad_norm": 0.7590881586074829,
|
| 1071 |
+
"learning_rate": 6.735892909679318e-05,
|
| 1072 |
+
"loss": 3.094,
|
| 1073 |
+
"step": 143000
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 7.74,
|
| 1077 |
+
"grad_norm": 0.7330321073532104,
|
| 1078 |
+
"learning_rate": 6.706472491909386e-05,
|
| 1079 |
+
"loss": 3.0967,
|
| 1080 |
+
"step": 144000
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 7.8,
|
| 1084 |
+
"grad_norm": 0.749355673789978,
|
| 1085 |
+
"learning_rate": 6.677052074139453e-05,
|
| 1086 |
+
"loss": 3.0984,
|
| 1087 |
+
"step": 145000
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 7.85,
|
| 1091 |
+
"grad_norm": 0.7481942176818848,
|
| 1092 |
+
"learning_rate": 6.647631656369521e-05,
|
| 1093 |
+
"loss": 3.0988,
|
| 1094 |
+
"step": 146000
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 7.91,
|
| 1098 |
+
"grad_norm": 0.7246766090393066,
|
| 1099 |
+
"learning_rate": 6.618240659017358e-05,
|
| 1100 |
+
"loss": 3.1007,
|
| 1101 |
+
"step": 147000
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 7.96,
|
| 1105 |
+
"grad_norm": 0.7423948049545288,
|
| 1106 |
+
"learning_rate": 6.588849661665197e-05,
|
| 1107 |
+
"loss": 3.0958,
|
| 1108 |
+
"step": 148000
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 8.0,
|
| 1112 |
+
"eval_accuracy": 0.3998490150370919,
|
| 1113 |
+
"eval_loss": 3.4232170581817627,
|
| 1114 |
+
"eval_runtime": 153.6442,
|
| 1115 |
+
"eval_samples_per_second": 376.962,
|
| 1116 |
+
"eval_steps_per_second": 5.89,
|
| 1117 |
+
"step": 148760
|
| 1118 |
+
},
|
| 1119 |
+
{
|
| 1120 |
+
"epoch": 8.01,
|
| 1121 |
+
"grad_norm": 0.754693865776062,
|
| 1122 |
+
"learning_rate": 6.559458664313034e-05,
|
| 1123 |
+
"loss": 3.0855,
|
| 1124 |
+
"step": 149000
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 8.07,
|
| 1128 |
+
"grad_norm": 0.7467632293701172,
|
| 1129 |
+
"learning_rate": 6.5300382465431e-05,
|
| 1130 |
+
"loss": 3.0395,
|
| 1131 |
+
"step": 150000
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 8.12,
|
| 1135 |
+
"grad_norm": 0.7670211791992188,
|
| 1136 |
+
"learning_rate": 6.500617828773169e-05,
|
| 1137 |
+
"loss": 3.0438,
|
| 1138 |
+
"step": 151000
|
| 1139 |
+
},
|
| 1140 |
+
{
|
| 1141 |
+
"epoch": 8.17,
|
| 1142 |
+
"grad_norm": 0.7683054804801941,
|
| 1143 |
+
"learning_rate": 6.471197411003237e-05,
|
| 1144 |
+
"loss": 3.046,
|
| 1145 |
+
"step": 152000
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 8.23,
|
| 1149 |
+
"grad_norm": 0.7673893570899963,
|
| 1150 |
+
"learning_rate": 6.441776993233303e-05,
|
| 1151 |
+
"loss": 3.0528,
|
| 1152 |
+
"step": 153000
|
| 1153 |
+
},
|
| 1154 |
+
{
|
| 1155 |
+
"epoch": 8.28,
|
| 1156 |
+
"grad_norm": 0.7494843602180481,
|
| 1157 |
+
"learning_rate": 6.412356575463372e-05,
|
| 1158 |
+
"loss": 3.0557,
|
| 1159 |
+
"step": 154000
|
| 1160 |
+
},
|
| 1161 |
+
{
|
| 1162 |
+
"epoch": 8.34,
|
| 1163 |
+
"grad_norm": 0.7590730786323547,
|
| 1164 |
+
"learning_rate": 6.38296557811121e-05,
|
| 1165 |
+
"loss": 3.0553,
|
| 1166 |
+
"step": 155000
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 8.39,
|
| 1170 |
+
"grad_norm": 0.7913221120834351,
|
| 1171 |
+
"learning_rate": 6.353545160341278e-05,
|
| 1172 |
+
"loss": 3.0609,
|
| 1173 |
+
"step": 156000
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 8.44,
|
| 1177 |
+
"grad_norm": 0.7681291699409485,
|
| 1178 |
+
"learning_rate": 6.324154162989114e-05,
|
| 1179 |
+
"loss": 3.0592,
|
| 1180 |
+
"step": 157000
|
| 1181 |
+
},
|
| 1182 |
+
{
|
| 1183 |
+
"epoch": 8.5,
|
| 1184 |
+
"grad_norm": 0.7939081788063049,
|
| 1185 |
+
"learning_rate": 6.294733745219182e-05,
|
| 1186 |
+
"loss": 3.0597,
|
| 1187 |
+
"step": 158000
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 8.55,
|
| 1191 |
+
"grad_norm": 0.7624540328979492,
|
| 1192 |
+
"learning_rate": 6.26534274786702e-05,
|
| 1193 |
+
"loss": 3.0635,
|
| 1194 |
+
"step": 159000
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 8.6,
|
| 1198 |
+
"grad_norm": 0.7824980020523071,
|
| 1199 |
+
"learning_rate": 6.235922330097088e-05,
|
| 1200 |
+
"loss": 3.0587,
|
| 1201 |
+
"step": 160000
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 8.66,
|
| 1205 |
+
"grad_norm": 0.777001142501831,
|
| 1206 |
+
"learning_rate": 6.206531332744924e-05,
|
| 1207 |
+
"loss": 3.0606,
|
| 1208 |
+
"step": 161000
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 8.71,
|
| 1212 |
+
"grad_norm": 0.7782078385353088,
|
| 1213 |
+
"learning_rate": 6.177110914974992e-05,
|
| 1214 |
+
"loss": 3.0672,
|
| 1215 |
+
"step": 162000
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 8.77,
|
| 1219 |
+
"grad_norm": 0.7476241588592529,
|
| 1220 |
+
"learning_rate": 6.14769049720506e-05,
|
| 1221 |
+
"loss": 3.058,
|
| 1222 |
+
"step": 163000
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"epoch": 8.82,
|
| 1226 |
+
"grad_norm": 0.7654275298118591,
|
| 1227 |
+
"learning_rate": 6.118299499852899e-05,
|
| 1228 |
+
"loss": 3.0678,
|
| 1229 |
+
"step": 164000
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 8.87,
|
| 1233 |
+
"grad_norm": 0.7455354928970337,
|
| 1234 |
+
"learning_rate": 6.0888790820829656e-05,
|
| 1235 |
+
"loss": 3.0664,
|
| 1236 |
+
"step": 165000
|
| 1237 |
+
},
|
| 1238 |
+
{
|
| 1239 |
+
"epoch": 8.93,
|
| 1240 |
+
"grad_norm": 0.7513304948806763,
|
| 1241 |
+
"learning_rate": 6.059458664313034e-05,
|
| 1242 |
+
"loss": 3.0674,
|
| 1243 |
+
"step": 166000
|
| 1244 |
+
},
|
| 1245 |
+
{
|
| 1246 |
+
"epoch": 8.98,
|
| 1247 |
+
"grad_norm": 0.730146586894989,
|
| 1248 |
+
"learning_rate": 6.030038246543102e-05,
|
| 1249 |
+
"loss": 3.0709,
|
| 1250 |
+
"step": 167000
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 9.0,
|
| 1254 |
+
"eval_accuracy": 0.40124062423958307,
|
| 1255 |
+
"eval_loss": 3.413806915283203,
|
| 1256 |
+
"eval_runtime": 153.6351,
|
| 1257 |
+
"eval_samples_per_second": 376.984,
|
| 1258 |
+
"eval_steps_per_second": 5.891,
|
| 1259 |
+
"step": 167355
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 9.03,
|
| 1263 |
+
"grad_norm": 0.7589735388755798,
|
| 1264 |
+
"learning_rate": 6.0006472491909396e-05,
|
| 1265 |
+
"loss": 3.03,
|
| 1266 |
+
"step": 168000
|
| 1267 |
+
},
|
| 1268 |
+
{
|
| 1269 |
+
"epoch": 9.09,
|
| 1270 |
+
"grad_norm": 0.8093643188476562,
|
| 1271 |
+
"learning_rate": 5.9712268314210063e-05,
|
| 1272 |
+
"loss": 3.0117,
|
| 1273 |
+
"step": 169000
|
| 1274 |
+
},
|
| 1275 |
+
{
|
| 1276 |
+
"epoch": 9.14,
|
| 1277 |
+
"grad_norm": 0.7802730798721313,
|
| 1278 |
+
"learning_rate": 5.941835834068844e-05,
|
| 1279 |
+
"loss": 3.0147,
|
| 1280 |
+
"step": 170000
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 9.2,
|
| 1284 |
+
"grad_norm": 0.8009030222892761,
|
| 1285 |
+
"learning_rate": 5.912415416298912e-05,
|
| 1286 |
+
"loss": 3.0232,
|
| 1287 |
+
"step": 171000
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 9.25,
|
| 1291 |
+
"grad_norm": 0.7963329553604126,
|
| 1292 |
+
"learning_rate": 5.88302441894675e-05,
|
| 1293 |
+
"loss": 3.024,
|
| 1294 |
+
"step": 172000
|
| 1295 |
+
},
|
| 1296 |
+
{
|
| 1297 |
+
"epoch": 9.3,
|
| 1298 |
+
"grad_norm": 0.7515285611152649,
|
| 1299 |
+
"learning_rate": 5.853604001176817e-05,
|
| 1300 |
+
"loss": 3.025,
|
| 1301 |
+
"step": 173000
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 9.36,
|
| 1305 |
+
"grad_norm": 0.7860144972801208,
|
| 1306 |
+
"learning_rate": 5.824183583406885e-05,
|
| 1307 |
+
"loss": 3.029,
|
| 1308 |
+
"step": 174000
|
| 1309 |
+
},
|
| 1310 |
+
{
|
| 1311 |
+
"epoch": 9.41,
|
| 1312 |
+
"grad_norm": 0.766737163066864,
|
| 1313 |
+
"learning_rate": 5.794763165636953e-05,
|
| 1314 |
+
"loss": 3.0294,
|
| 1315 |
+
"step": 175000
|
| 1316 |
+
},
|
| 1317 |
+
{
|
| 1318 |
+
"epoch": 9.46,
|
| 1319 |
+
"grad_norm": 0.7924851775169373,
|
| 1320 |
+
"learning_rate": 5.765372168284789e-05,
|
| 1321 |
+
"loss": 3.0303,
|
| 1322 |
+
"step": 176000
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 9.52,
|
| 1326 |
+
"grad_norm": 0.805383563041687,
|
| 1327 |
+
"learning_rate": 5.735981170932627e-05,
|
| 1328 |
+
"loss": 3.0318,
|
| 1329 |
+
"step": 177000
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 9.57,
|
| 1333 |
+
"grad_norm": 0.781491756439209,
|
| 1334 |
+
"learning_rate": 5.706560753162695e-05,
|
| 1335 |
+
"loss": 3.0328,
|
| 1336 |
+
"step": 178000
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"epoch": 9.63,
|
| 1340 |
+
"grad_norm": 0.7737943530082703,
|
| 1341 |
+
"learning_rate": 5.677169755810533e-05,
|
| 1342 |
+
"loss": 3.0322,
|
| 1343 |
+
"step": 179000
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 9.68,
|
| 1347 |
+
"grad_norm": 0.7793260812759399,
|
| 1348 |
+
"learning_rate": 5.6477493380406e-05,
|
| 1349 |
+
"loss": 3.0339,
|
| 1350 |
+
"step": 180000
|
| 1351 |
+
},
|
| 1352 |
+
{
|
| 1353 |
+
"epoch": 9.73,
|
| 1354 |
+
"grad_norm": 0.7405052185058594,
|
| 1355 |
+
"learning_rate": 5.6183583406884374e-05,
|
| 1356 |
+
"loss": 3.0366,
|
| 1357 |
+
"step": 181000
|
| 1358 |
+
},
|
| 1359 |
+
{
|
| 1360 |
+
"epoch": 9.79,
|
| 1361 |
+
"grad_norm": 0.798646092414856,
|
| 1362 |
+
"learning_rate": 5.5889379229185055e-05,
|
| 1363 |
+
"loss": 3.04,
|
| 1364 |
+
"step": 182000
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 9.84,
|
| 1368 |
+
"grad_norm": 0.7703911662101746,
|
| 1369 |
+
"learning_rate": 5.559546925566343e-05,
|
| 1370 |
+
"loss": 3.0363,
|
| 1371 |
+
"step": 183000
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 9.9,
|
| 1375 |
+
"grad_norm": 0.7801164984703064,
|
| 1376 |
+
"learning_rate": 5.530155928214181e-05,
|
| 1377 |
+
"loss": 3.0383,
|
| 1378 |
+
"step": 184000
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 9.95,
|
| 1382 |
+
"grad_norm": 0.7496475577354431,
|
| 1383 |
+
"learning_rate": 5.5007355104442484e-05,
|
| 1384 |
+
"loss": 3.0426,
|
| 1385 |
+
"step": 185000
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 10.0,
|
| 1389 |
+
"eval_accuracy": 0.40137438372216294,
|
| 1390 |
+
"eval_loss": 3.426945924758911,
|
| 1391 |
+
"eval_runtime": 153.8591,
|
| 1392 |
+
"eval_samples_per_second": 376.435,
|
| 1393 |
+
"eval_steps_per_second": 5.882,
|
| 1394 |
+
"step": 185950
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 10.0,
|
| 1398 |
+
"grad_norm": 0.7879915237426758,
|
| 1399 |
+
"learning_rate": 5.471315092674316e-05,
|
| 1400 |
+
"loss": 3.0403,
|
| 1401 |
+
"step": 186000
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 10.06,
|
| 1405 |
+
"grad_norm": 0.7904853820800781,
|
| 1406 |
+
"learning_rate": 5.441924095322154e-05,
|
| 1407 |
+
"loss": 2.9838,
|
| 1408 |
+
"step": 187000
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"epoch": 10.11,
|
| 1412 |
+
"grad_norm": 0.7917992472648621,
|
| 1413 |
+
"learning_rate": 5.412503677552222e-05,
|
| 1414 |
+
"loss": 2.9899,
|
| 1415 |
+
"step": 188000
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 10.16,
|
| 1419 |
+
"grad_norm": 0.8023027181625366,
|
| 1420 |
+
"learning_rate": 5.383083259782289e-05,
|
| 1421 |
+
"loss": 2.9929,
|
| 1422 |
+
"step": 189000
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 10.22,
|
| 1426 |
+
"grad_norm": 0.7836917638778687,
|
| 1427 |
+
"learning_rate": 5.3536628420123566e-05,
|
| 1428 |
+
"loss": 2.9918,
|
| 1429 |
+
"step": 190000
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 10.27,
|
| 1433 |
+
"grad_norm": 0.792332112789154,
|
| 1434 |
+
"learning_rate": 5.3242718446601943e-05,
|
| 1435 |
+
"loss": 2.9955,
|
| 1436 |
+
"step": 191000
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 10.33,
|
| 1440 |
+
"grad_norm": 0.8201410174369812,
|
| 1441 |
+
"learning_rate": 5.294851426890262e-05,
|
| 1442 |
+
"loss": 3.0028,
|
| 1443 |
+
"step": 192000
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 10.38,
|
| 1447 |
+
"grad_norm": 0.807855486869812,
|
| 1448 |
+
"learning_rate": 5.2654604295380995e-05,
|
| 1449 |
+
"loss": 3.0062,
|
| 1450 |
+
"step": 193000
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 10.43,
|
| 1454 |
+
"grad_norm": 0.7855746150016785,
|
| 1455 |
+
"learning_rate": 5.2360400117681677e-05,
|
| 1456 |
+
"loss": 3.0054,
|
| 1457 |
+
"step": 194000
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 10.49,
|
| 1461 |
+
"grad_norm": 0.781585156917572,
|
| 1462 |
+
"learning_rate": 5.206619593998235e-05,
|
| 1463 |
+
"loss": 3.0082,
|
| 1464 |
+
"step": 195000
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 10.54,
|
| 1468 |
+
"grad_norm": 0.8157842755317688,
|
| 1469 |
+
"learning_rate": 5.1771991762283025e-05,
|
| 1470 |
+
"loss": 3.0068,
|
| 1471 |
+
"step": 196000
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 10.59,
|
| 1475 |
+
"grad_norm": 0.7841827273368835,
|
| 1476 |
+
"learning_rate": 5.14780817887614e-05,
|
| 1477 |
+
"loss": 3.0081,
|
| 1478 |
+
"step": 197000
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 10.65,
|
| 1482 |
+
"grad_norm": 0.8037407398223877,
|
| 1483 |
+
"learning_rate": 5.1183877611062084e-05,
|
| 1484 |
+
"loss": 3.0082,
|
| 1485 |
+
"step": 198000
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 10.7,
|
| 1489 |
+
"grad_norm": 0.8202306628227234,
|
| 1490 |
+
"learning_rate": 5.088967343336276e-05,
|
| 1491 |
+
"loss": 3.0149,
|
| 1492 |
+
"step": 199000
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"epoch": 10.76,
|
| 1496 |
+
"grad_norm": 0.7935760021209717,
|
| 1497 |
+
"learning_rate": 5.059576345984113e-05,
|
| 1498 |
+
"loss": 3.0136,
|
| 1499 |
+
"step": 200000
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 10.81,
|
| 1503 |
+
"grad_norm": 0.7972075343132019,
|
| 1504 |
+
"learning_rate": 5.030155928214181e-05,
|
| 1505 |
+
"loss": 3.0142,
|
| 1506 |
+
"step": 201000
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 10.86,
|
| 1510 |
+
"grad_norm": 0.7736737728118896,
|
| 1511 |
+
"learning_rate": 5.000764930862019e-05,
|
| 1512 |
+
"loss": 3.0111,
|
| 1513 |
+
"step": 202000
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 10.92,
|
| 1517 |
+
"grad_norm": 0.7707228660583496,
|
| 1518 |
+
"learning_rate": 4.971344513092086e-05,
|
| 1519 |
+
"loss": 3.0161,
|
| 1520 |
+
"step": 203000
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 10.97,
|
| 1524 |
+
"grad_norm": 0.8111630082130432,
|
| 1525 |
+
"learning_rate": 4.941924095322154e-05,
|
| 1526 |
+
"loss": 3.0138,
|
| 1527 |
+
"step": 204000
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 11.0,
|
| 1531 |
+
"eval_accuracy": 0.4036656950797068,
|
| 1532 |
+
"eval_loss": 3.402326822280884,
|
| 1533 |
+
"eval_runtime": 154.1124,
|
| 1534 |
+
"eval_samples_per_second": 375.817,
|
| 1535 |
+
"eval_steps_per_second": 5.872,
|
| 1536 |
+
"step": 204545
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 11.02,
|
| 1540 |
+
"grad_norm": 0.7924003005027771,
|
| 1541 |
+
"learning_rate": 4.9125330979699914e-05,
|
| 1542 |
+
"loss": 2.9853,
|
| 1543 |
+
"step": 205000
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 11.08,
|
| 1547 |
+
"grad_norm": 0.8233866691589355,
|
| 1548 |
+
"learning_rate": 4.8831126802000595e-05,
|
| 1549 |
+
"loss": 2.9635,
|
| 1550 |
+
"step": 206000
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 11.13,
|
| 1554 |
+
"grad_norm": 0.8010469675064087,
|
| 1555 |
+
"learning_rate": 4.853692262430127e-05,
|
| 1556 |
+
"loss": 2.9704,
|
| 1557 |
+
"step": 207000
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 11.19,
|
| 1561 |
+
"grad_norm": 0.8223724365234375,
|
| 1562 |
+
"learning_rate": 4.824301265077965e-05,
|
| 1563 |
+
"loss": 2.9697,
|
| 1564 |
+
"step": 208000
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 11.24,
|
| 1568 |
+
"grad_norm": 0.814317524433136,
|
| 1569 |
+
"learning_rate": 4.794880847308032e-05,
|
| 1570 |
+
"loss": 2.9726,
|
| 1571 |
+
"step": 209000
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 11.29,
|
| 1575 |
+
"grad_norm": 0.8176049590110779,
|
| 1576 |
+
"learning_rate": 4.76548984995587e-05,
|
| 1577 |
+
"loss": 2.9755,
|
| 1578 |
+
"step": 210000
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 11.35,
|
| 1582 |
+
"grad_norm": 0.8245882987976074,
|
| 1583 |
+
"learning_rate": 4.736069432185937e-05,
|
| 1584 |
+
"loss": 2.9801,
|
| 1585 |
+
"step": 211000
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 11.4,
|
| 1589 |
+
"grad_norm": 0.8146276473999023,
|
| 1590 |
+
"learning_rate": 4.706678434833775e-05,
|
| 1591 |
+
"loss": 2.9788,
|
| 1592 |
+
"step": 212000
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 11.45,
|
| 1596 |
+
"grad_norm": 0.8188950419425964,
|
| 1597 |
+
"learning_rate": 4.6772580170638425e-05,
|
| 1598 |
+
"loss": 2.9817,
|
| 1599 |
+
"step": 213000
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 11.51,
|
| 1603 |
+
"grad_norm": 0.7914262413978577,
|
| 1604 |
+
"learning_rate": 4.64783759929391e-05,
|
| 1605 |
+
"loss": 2.9856,
|
| 1606 |
+
"step": 214000
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 11.56,
|
| 1610 |
+
"grad_norm": 0.81853187084198,
|
| 1611 |
+
"learning_rate": 4.618417181523978e-05,
|
| 1612 |
+
"loss": 2.9851,
|
| 1613 |
+
"step": 215000
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 11.62,
|
| 1617 |
+
"grad_norm": 0.8388851881027222,
|
| 1618 |
+
"learning_rate": 4.5889967637540455e-05,
|
| 1619 |
+
"loss": 2.9852,
|
| 1620 |
+
"step": 216000
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 11.67,
|
| 1624 |
+
"grad_norm": 0.8250225782394409,
|
| 1625 |
+
"learning_rate": 4.559605766401883e-05,
|
| 1626 |
+
"loss": 2.9838,
|
| 1627 |
+
"step": 217000
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 11.72,
|
| 1631 |
+
"grad_norm": 0.7764880061149597,
|
| 1632 |
+
"learning_rate": 4.530185348631951e-05,
|
| 1633 |
+
"loss": 2.9913,
|
| 1634 |
+
"step": 218000
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 11.78,
|
| 1638 |
+
"grad_norm": 0.8230167627334595,
|
| 1639 |
+
"learning_rate": 4.5007943512797884e-05,
|
| 1640 |
+
"loss": 2.9888,
|
| 1641 |
+
"step": 219000
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 11.83,
|
| 1645 |
+
"grad_norm": 0.7996993064880371,
|
| 1646 |
+
"learning_rate": 4.471373933509856e-05,
|
| 1647 |
+
"loss": 2.9869,
|
| 1648 |
+
"step": 220000
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 11.88,
|
| 1652 |
+
"grad_norm": 0.8135858178138733,
|
| 1653 |
+
"learning_rate": 4.441953515739924e-05,
|
| 1654 |
+
"loss": 2.9924,
|
| 1655 |
+
"step": 221000
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 11.94,
|
| 1659 |
+
"grad_norm": 0.8289496898651123,
|
| 1660 |
+
"learning_rate": 4.412562518387761e-05,
|
| 1661 |
+
"loss": 2.9901,
|
| 1662 |
+
"step": 222000
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 11.99,
|
| 1666 |
+
"grad_norm": 0.7562257647514343,
|
| 1667 |
+
"learning_rate": 4.383142100617829e-05,
|
| 1668 |
+
"loss": 2.995,
|
| 1669 |
+
"step": 223000
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 12.0,
|
| 1673 |
+
"eval_accuracy": 0.4034856471573994,
|
| 1674 |
+
"eval_loss": 3.4036989212036133,
|
| 1675 |
+
"eval_runtime": 153.8908,
|
| 1676 |
+
"eval_samples_per_second": 376.358,
|
| 1677 |
+
"eval_steps_per_second": 5.881,
|
| 1678 |
+
"step": 223140
|
| 1679 |
+
},
|
| 1680 |
+
{
|
| 1681 |
+
"epoch": 12.05,
|
| 1682 |
+
"grad_norm": 0.8186893463134766,
|
| 1683 |
+
"learning_rate": 4.3537216828478966e-05,
|
| 1684 |
+
"loss": 2.9427,
|
| 1685 |
+
"step": 224000
|
| 1686 |
+
},
|
| 1687 |
+
{
|
| 1688 |
+
"epoch": 12.1,
|
| 1689 |
+
"grad_norm": 0.861786961555481,
|
| 1690 |
+
"learning_rate": 4.324330685495734e-05,
|
| 1691 |
+
"loss": 2.9419,
|
| 1692 |
+
"step": 225000
|
| 1693 |
+
},
|
| 1694 |
+
{
|
| 1695 |
+
"epoch": 12.15,
|
| 1696 |
+
"grad_norm": 0.8424795866012573,
|
| 1697 |
+
"learning_rate": 4.294910267725802e-05,
|
| 1698 |
+
"loss": 2.9471,
|
| 1699 |
+
"step": 226000
|
| 1700 |
+
},
|
| 1701 |
+
{
|
| 1702 |
+
"epoch": 12.21,
|
| 1703 |
+
"grad_norm": 0.822487473487854,
|
| 1704 |
+
"learning_rate": 4.265489849955869e-05,
|
| 1705 |
+
"loss": 2.953,
|
| 1706 |
+
"step": 227000
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 12.26,
|
| 1710 |
+
"grad_norm": 0.8305549025535583,
|
| 1711 |
+
"learning_rate": 4.2360694321859373e-05,
|
| 1712 |
+
"loss": 2.951,
|
| 1713 |
+
"step": 228000
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 12.32,
|
| 1717 |
+
"grad_norm": 0.8566609621047974,
|
| 1718 |
+
"learning_rate": 4.206678434833775e-05,
|
| 1719 |
+
"loss": 2.9529,
|
| 1720 |
+
"step": 229000
|
| 1721 |
+
},
|
| 1722 |
+
{
|
| 1723 |
+
"epoch": 12.37,
|
| 1724 |
+
"grad_norm": 0.8459467887878418,
|
| 1725 |
+
"learning_rate": 4.1772580170638425e-05,
|
| 1726 |
+
"loss": 2.9578,
|
| 1727 |
+
"step": 230000
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 12.42,
|
| 1731 |
+
"grad_norm": 0.8066169023513794,
|
| 1732 |
+
"learning_rate": 4.14786701971168e-05,
|
| 1733 |
+
"loss": 2.965,
|
| 1734 |
+
"step": 231000
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 12.48,
|
| 1738 |
+
"grad_norm": 0.8078910112380981,
|
| 1739 |
+
"learning_rate": 4.118446601941748e-05,
|
| 1740 |
+
"loss": 2.9581,
|
| 1741 |
+
"step": 232000
|
| 1742 |
+
},
|
| 1743 |
+
{
|
| 1744 |
+
"epoch": 12.53,
|
| 1745 |
+
"grad_norm": 0.8605197668075562,
|
| 1746 |
+
"learning_rate": 4.0890556045895854e-05,
|
| 1747 |
+
"loss": 2.9648,
|
| 1748 |
+
"step": 233000
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 12.58,
|
| 1752 |
+
"grad_norm": 0.8278983235359192,
|
| 1753 |
+
"learning_rate": 4.059635186819653e-05,
|
| 1754 |
+
"loss": 2.9654,
|
| 1755 |
+
"step": 234000
|
| 1756 |
+
},
|
| 1757 |
+
{
|
| 1758 |
+
"epoch": 12.64,
|
| 1759 |
+
"grad_norm": 0.7852072715759277,
|
| 1760 |
+
"learning_rate": 4.03021476904972e-05,
|
| 1761 |
+
"loss": 2.9638,
|
| 1762 |
+
"step": 235000
|
| 1763 |
+
},
|
| 1764 |
+
{
|
| 1765 |
+
"epoch": 12.69,
|
| 1766 |
+
"grad_norm": 0.8360777497291565,
|
| 1767 |
+
"learning_rate": 4.0007943512797884e-05,
|
| 1768 |
+
"loss": 2.9643,
|
| 1769 |
+
"step": 236000
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 12.75,
|
| 1773 |
+
"grad_norm": 0.8392547965049744,
|
| 1774 |
+
"learning_rate": 3.9714033539276255e-05,
|
| 1775 |
+
"loss": 2.9719,
|
| 1776 |
+
"step": 237000
|
| 1777 |
+
},
|
| 1778 |
+
{
|
| 1779 |
+
"epoch": 12.8,
|
| 1780 |
+
"grad_norm": 0.828360915184021,
|
| 1781 |
+
"learning_rate": 3.9419829361576936e-05,
|
| 1782 |
+
"loss": 2.9705,
|
| 1783 |
+
"step": 238000
|
| 1784 |
+
},
|
| 1785 |
+
{
|
| 1786 |
+
"epoch": 12.85,
|
| 1787 |
+
"grad_norm": 0.8145458698272705,
|
| 1788 |
+
"learning_rate": 3.912591938805531e-05,
|
| 1789 |
+
"loss": 2.9723,
|
| 1790 |
+
"step": 239000
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 12.91,
|
| 1794 |
+
"grad_norm": 0.837388813495636,
|
| 1795 |
+
"learning_rate": 3.883200941453369e-05,
|
| 1796 |
+
"loss": 2.9677,
|
| 1797 |
+
"step": 240000
|
| 1798 |
+
},
|
| 1799 |
+
{
|
| 1800 |
+
"epoch": 12.96,
|
| 1801 |
+
"grad_norm": 0.8371777534484863,
|
| 1802 |
+
"learning_rate": 3.8537805236834365e-05,
|
| 1803 |
+
"loss": 2.9702,
|
| 1804 |
+
"step": 241000
|
| 1805 |
+
},
|
| 1806 |
+
{
|
| 1807 |
+
"epoch": 13.0,
|
| 1808 |
+
"eval_accuracy": 0.4042875322322731,
|
| 1809 |
+
"eval_loss": 3.3991425037384033,
|
| 1810 |
+
"eval_runtime": 153.5581,
|
| 1811 |
+
"eval_samples_per_second": 377.173,
|
| 1812 |
+
"eval_steps_per_second": 5.894,
|
| 1813 |
+
"step": 241735
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 13.01,
|
| 1817 |
+
"grad_norm": 0.8215118646621704,
|
| 1818 |
+
"learning_rate": 3.824360105913504e-05,
|
| 1819 |
+
"loss": 2.9556,
|
| 1820 |
+
"step": 242000
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 13.07,
|
| 1824 |
+
"grad_norm": 0.8516760468482971,
|
| 1825 |
+
"learning_rate": 3.7949396881435714e-05,
|
| 1826 |
+
"loss": 2.9217,
|
| 1827 |
+
"step": 243000
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 13.12,
|
| 1831 |
+
"grad_norm": 0.847839891910553,
|
| 1832 |
+
"learning_rate": 3.765548690791409e-05,
|
| 1833 |
+
"loss": 2.9274,
|
| 1834 |
+
"step": 244000
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 13.18,
|
| 1838 |
+
"grad_norm": 0.8394870758056641,
|
| 1839 |
+
"learning_rate": 3.736128273021477e-05,
|
| 1840 |
+
"loss": 2.9288,
|
| 1841 |
+
"step": 245000
|
| 1842 |
+
},
|
| 1843 |
+
{
|
| 1844 |
+
"epoch": 13.23,
|
| 1845 |
+
"grad_norm": 0.8435803055763245,
|
| 1846 |
+
"learning_rate": 3.706707855251545e-05,
|
| 1847 |
+
"loss": 2.9338,
|
| 1848 |
+
"step": 246000
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 13.28,
|
| 1852 |
+
"grad_norm": 0.8720064163208008,
|
| 1853 |
+
"learning_rate": 3.677287437481612e-05,
|
| 1854 |
+
"loss": 2.9338,
|
| 1855 |
+
"step": 247000
|
| 1856 |
+
},
|
| 1857 |
+
{
|
| 1858 |
+
"epoch": 13.34,
|
| 1859 |
+
"grad_norm": 0.8536742329597473,
|
| 1860 |
+
"learning_rate": 3.64789644012945e-05,
|
| 1861 |
+
"loss": 2.9371,
|
| 1862 |
+
"step": 248000
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 13.39,
|
| 1866 |
+
"grad_norm": 0.845740020275116,
|
| 1867 |
+
"learning_rate": 3.6184760223595173e-05,
|
| 1868 |
+
"loss": 2.9405,
|
| 1869 |
+
"step": 249000
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 13.44,
|
| 1873 |
+
"grad_norm": 0.8853806257247925,
|
| 1874 |
+
"learning_rate": 3.589055604589585e-05,
|
| 1875 |
+
"loss": 2.9408,
|
| 1876 |
+
"step": 250000
|
| 1877 |
+
},
|
| 1878 |
+
{
|
| 1879 |
+
"epoch": 13.5,
|
| 1880 |
+
"grad_norm": 0.8576515316963196,
|
| 1881 |
+
"learning_rate": 3.559635186819653e-05,
|
| 1882 |
+
"loss": 2.9412,
|
| 1883 |
+
"step": 251000
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 13.55,
|
| 1887 |
+
"grad_norm": 0.8389966487884521,
|
| 1888 |
+
"learning_rate": 3.5302441894674906e-05,
|
| 1889 |
+
"loss": 2.9444,
|
| 1890 |
+
"step": 252000
|
| 1891 |
+
},
|
| 1892 |
+
{
|
| 1893 |
+
"epoch": 13.61,
|
| 1894 |
+
"grad_norm": 0.8276740908622742,
|
| 1895 |
+
"learning_rate": 3.500882612533098e-05,
|
| 1896 |
+
"loss": 2.9425,
|
| 1897 |
+
"step": 253000
|
| 1898 |
+
},
|
| 1899 |
+
{
|
| 1900 |
+
"epoch": 13.66,
|
| 1901 |
+
"grad_norm": 0.8876621127128601,
|
| 1902 |
+
"learning_rate": 3.471462194763166e-05,
|
| 1903 |
+
"loss": 2.9441,
|
| 1904 |
+
"step": 254000
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 13.71,
|
| 1908 |
+
"grad_norm": 0.8480891585350037,
|
| 1909 |
+
"learning_rate": 3.4420417769932335e-05,
|
| 1910 |
+
"loss": 2.9461,
|
| 1911 |
+
"step": 255000
|
| 1912 |
+
},
|
| 1913 |
+
{
|
| 1914 |
+
"epoch": 13.77,
|
| 1915 |
+
"grad_norm": 0.8812869191169739,
|
| 1916 |
+
"learning_rate": 3.412621359223302e-05,
|
| 1917 |
+
"loss": 2.946,
|
| 1918 |
+
"step": 256000
|
| 1919 |
+
},
|
| 1920 |
+
{
|
| 1921 |
+
"epoch": 13.82,
|
| 1922 |
+
"grad_norm": 0.8724462985992432,
|
| 1923 |
+
"learning_rate": 3.383200941453369e-05,
|
| 1924 |
+
"loss": 2.9463,
|
| 1925 |
+
"step": 257000
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"epoch": 13.87,
|
| 1929 |
+
"grad_norm": 0.837191641330719,
|
| 1930 |
+
"learning_rate": 3.3537805236834366e-05,
|
| 1931 |
+
"loss": 2.9517,
|
| 1932 |
+
"step": 258000
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"epoch": 13.93,
|
| 1936 |
+
"grad_norm": 0.8628780841827393,
|
| 1937 |
+
"learning_rate": 3.324360105913504e-05,
|
| 1938 |
+
"loss": 2.9476,
|
| 1939 |
+
"step": 259000
|
| 1940 |
+
},
|
| 1941 |
+
{
|
| 1942 |
+
"epoch": 13.98,
|
| 1943 |
+
"grad_norm": 0.8205897808074951,
|
| 1944 |
+
"learning_rate": 3.294969108561342e-05,
|
| 1945 |
+
"loss": 2.954,
|
| 1946 |
+
"step": 260000
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 14.0,
|
| 1950 |
+
"eval_accuracy": 0.4042310321193401,
|
| 1951 |
+
"eval_loss": 3.4180350303649902,
|
| 1952 |
+
"eval_runtime": 153.7406,
|
| 1953 |
+
"eval_samples_per_second": 376.725,
|
| 1954 |
+
"eval_steps_per_second": 5.887,
|
| 1955 |
+
"step": 260330
|
| 1956 |
+
},
|
| 1957 |
+
{
|
| 1958 |
+
"epoch": 14.04,
|
| 1959 |
+
"grad_norm": 0.8445749282836914,
|
| 1960 |
+
"learning_rate": 3.26554869079141e-05,
|
| 1961 |
+
"loss": 2.9212,
|
| 1962 |
+
"step": 261000
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 14.09,
|
| 1966 |
+
"grad_norm": 0.8567079901695251,
|
| 1967 |
+
"learning_rate": 3.236157693439247e-05,
|
| 1968 |
+
"loss": 2.9086,
|
| 1969 |
+
"step": 262000
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 14.14,
|
| 1973 |
+
"grad_norm": 0.8678041100502014,
|
| 1974 |
+
"learning_rate": 3.2067666960870846e-05,
|
| 1975 |
+
"loss": 2.9132,
|
| 1976 |
+
"step": 263000
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 14.2,
|
| 1980 |
+
"grad_norm": 0.8548084497451782,
|
| 1981 |
+
"learning_rate": 3.177346278317152e-05,
|
| 1982 |
+
"loss": 2.9131,
|
| 1983 |
+
"step": 264000
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 14.25,
|
| 1987 |
+
"grad_norm": 0.9028843641281128,
|
| 1988 |
+
"learning_rate": 3.14792586054722e-05,
|
| 1989 |
+
"loss": 2.9163,
|
| 1990 |
+
"step": 265000
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 14.3,
|
| 1994 |
+
"grad_norm": 0.912269651889801,
|
| 1995 |
+
"learning_rate": 3.118505442777288e-05,
|
| 1996 |
+
"loss": 2.9202,
|
| 1997 |
+
"step": 266000
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 14.36,
|
| 2001 |
+
"grad_norm": 0.8899259567260742,
|
| 2002 |
+
"learning_rate": 3.0891144454251254e-05,
|
| 2003 |
+
"loss": 2.9205,
|
| 2004 |
+
"step": 267000
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 14.41,
|
| 2008 |
+
"grad_norm": 0.8662390112876892,
|
| 2009 |
+
"learning_rate": 3.059694027655193e-05,
|
| 2010 |
+
"loss": 2.9197,
|
| 2011 |
+
"step": 268000
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 14.47,
|
| 2015 |
+
"grad_norm": 0.8414704203605652,
|
| 2016 |
+
"learning_rate": 3.0303030303030306e-05,
|
| 2017 |
+
"loss": 2.9215,
|
| 2018 |
+
"step": 269000
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 14.52,
|
| 2022 |
+
"grad_norm": 0.8585665225982666,
|
| 2023 |
+
"learning_rate": 3.000882612533098e-05,
|
| 2024 |
+
"loss": 2.9224,
|
| 2025 |
+
"step": 270000
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 14.57,
|
| 2029 |
+
"grad_norm": 0.8473167419433594,
|
| 2030 |
+
"learning_rate": 2.971462194763166e-05,
|
| 2031 |
+
"loss": 2.9261,
|
| 2032 |
+
"step": 271000
|
| 2033 |
+
},
|
| 2034 |
+
{
|
| 2035 |
+
"epoch": 14.63,
|
| 2036 |
+
"grad_norm": 0.8617039918899536,
|
| 2037 |
+
"learning_rate": 2.9420417769932336e-05,
|
| 2038 |
+
"loss": 2.9276,
|
| 2039 |
+
"step": 272000
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 14.68,
|
| 2043 |
+
"grad_norm": 0.8747395873069763,
|
| 2044 |
+
"learning_rate": 2.9126507796410713e-05,
|
| 2045 |
+
"loss": 2.9274,
|
| 2046 |
+
"step": 273000
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 14.74,
|
| 2050 |
+
"grad_norm": 0.8597936034202576,
|
| 2051 |
+
"learning_rate": 2.8832303618711388e-05,
|
| 2052 |
+
"loss": 2.9252,
|
| 2053 |
+
"step": 274000
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 14.79,
|
| 2057 |
+
"grad_norm": 0.8555459976196289,
|
| 2058 |
+
"learning_rate": 2.8538393645189765e-05,
|
| 2059 |
+
"loss": 2.926,
|
| 2060 |
+
"step": 275000
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 14.84,
|
| 2064 |
+
"grad_norm": 0.876461386680603,
|
| 2065 |
+
"learning_rate": 2.824418946749044e-05,
|
| 2066 |
+
"loss": 2.9302,
|
| 2067 |
+
"step": 276000
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 14.9,
|
| 2071 |
+
"grad_norm": 0.8898986577987671,
|
| 2072 |
+
"learning_rate": 2.7950279493968817e-05,
|
| 2073 |
+
"loss": 2.9278,
|
| 2074 |
+
"step": 277000
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 14.95,
|
| 2078 |
+
"grad_norm": 0.8596899509429932,
|
| 2079 |
+
"learning_rate": 2.765607531626949e-05,
|
| 2080 |
+
"loss": 2.9299,
|
| 2081 |
+
"step": 278000
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 15.0,
|
| 2085 |
+
"eval_accuracy": 0.4048775922701934,
|
| 2086 |
+
"eval_loss": 3.4059524536132812,
|
| 2087 |
+
"eval_runtime": 153.5689,
|
| 2088 |
+
"eval_samples_per_second": 377.147,
|
| 2089 |
+
"eval_steps_per_second": 5.893,
|
| 2090 |
+
"step": 278925
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 15.0,
|
| 2094 |
+
"grad_norm": 0.8894351124763489,
|
| 2095 |
+
"learning_rate": 2.7361871138570166e-05,
|
| 2096 |
+
"loss": 2.9266,
|
| 2097 |
+
"step": 279000
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 15.06,
|
| 2101 |
+
"grad_norm": 0.8618838787078857,
|
| 2102 |
+
"learning_rate": 2.7067666960870847e-05,
|
| 2103 |
+
"loss": 2.8883,
|
| 2104 |
+
"step": 280000
|
| 2105 |
+
},
|
| 2106 |
+
{
|
| 2107 |
+
"epoch": 15.11,
|
| 2108 |
+
"grad_norm": 0.9119635224342346,
|
| 2109 |
+
"learning_rate": 2.6773756987349217e-05,
|
| 2110 |
+
"loss": 2.8918,
|
| 2111 |
+
"step": 281000
|
| 2112 |
+
},
|
| 2113 |
+
{
|
| 2114 |
+
"epoch": 15.17,
|
| 2115 |
+
"grad_norm": 0.9044109582901001,
|
| 2116 |
+
"learning_rate": 2.64795528096499e-05,
|
| 2117 |
+
"loss": 2.8922,
|
| 2118 |
+
"step": 282000
|
| 2119 |
+
},
|
| 2120 |
+
{
|
| 2121 |
+
"epoch": 15.22,
|
| 2122 |
+
"grad_norm": 0.8991778492927551,
|
| 2123 |
+
"learning_rate": 2.6185348631950573e-05,
|
| 2124 |
+
"loss": 2.8929,
|
| 2125 |
+
"step": 283000
|
| 2126 |
+
},
|
| 2127 |
+
{
|
| 2128 |
+
"epoch": 15.27,
|
| 2129 |
+
"grad_norm": 0.8891802430152893,
|
| 2130 |
+
"learning_rate": 2.589143865842895e-05,
|
| 2131 |
+
"loss": 2.8997,
|
| 2132 |
+
"step": 284000
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 15.33,
|
| 2136 |
+
"grad_norm": 0.9004053473472595,
|
| 2137 |
+
"learning_rate": 2.5597528684907328e-05,
|
| 2138 |
+
"loss": 2.9055,
|
| 2139 |
+
"step": 285000
|
| 2140 |
+
},
|
| 2141 |
+
{
|
| 2142 |
+
"epoch": 15.38,
|
| 2143 |
+
"grad_norm": 0.895466148853302,
|
| 2144 |
+
"learning_rate": 2.5303324507208002e-05,
|
| 2145 |
+
"loss": 2.906,
|
| 2146 |
+
"step": 286000
|
| 2147 |
+
},
|
| 2148 |
+
{
|
| 2149 |
+
"epoch": 15.43,
|
| 2150 |
+
"grad_norm": 0.9080808162689209,
|
| 2151 |
+
"learning_rate": 2.5009414533686383e-05,
|
| 2152 |
+
"loss": 2.9072,
|
| 2153 |
+
"step": 287000
|
| 2154 |
+
},
|
| 2155 |
+
{
|
| 2156 |
+
"epoch": 15.49,
|
| 2157 |
+
"grad_norm": 0.8760538101196289,
|
| 2158 |
+
"learning_rate": 2.4715210355987057e-05,
|
| 2159 |
+
"loss": 2.9057,
|
| 2160 |
+
"step": 288000
|
| 2161 |
+
},
|
| 2162 |
+
{
|
| 2163 |
+
"epoch": 15.54,
|
| 2164 |
+
"grad_norm": 0.8684205412864685,
|
| 2165 |
+
"learning_rate": 2.4421006178287735e-05,
|
| 2166 |
+
"loss": 2.9073,
|
| 2167 |
+
"step": 289000
|
| 2168 |
+
},
|
| 2169 |
+
{
|
| 2170 |
+
"epoch": 15.6,
|
| 2171 |
+
"grad_norm": 0.8719987273216248,
|
| 2172 |
+
"learning_rate": 2.412680200058841e-05,
|
| 2173 |
+
"loss": 2.9046,
|
| 2174 |
+
"step": 290000
|
| 2175 |
+
},
|
| 2176 |
+
{
|
| 2177 |
+
"epoch": 15.65,
|
| 2178 |
+
"grad_norm": 0.8654382824897766,
|
| 2179 |
+
"learning_rate": 2.3832597822889084e-05,
|
| 2180 |
+
"loss": 2.9077,
|
| 2181 |
+
"step": 291000
|
| 2182 |
+
},
|
| 2183 |
+
{
|
| 2184 |
+
"epoch": 15.7,
|
| 2185 |
+
"grad_norm": 0.8941999673843384,
|
| 2186 |
+
"learning_rate": 2.3538393645189762e-05,
|
| 2187 |
+
"loss": 2.9111,
|
| 2188 |
+
"step": 292000
|
| 2189 |
+
},
|
| 2190 |
+
{
|
| 2191 |
+
"epoch": 15.76,
|
| 2192 |
+
"grad_norm": 0.8872021436691284,
|
| 2193 |
+
"learning_rate": 2.324418946749044e-05,
|
| 2194 |
+
"loss": 2.914,
|
| 2195 |
+
"step": 293000
|
| 2196 |
+
},
|
| 2197 |
+
{
|
| 2198 |
+
"epoch": 15.81,
|
| 2199 |
+
"grad_norm": 0.8763778209686279,
|
| 2200 |
+
"learning_rate": 2.2950573698146517e-05,
|
| 2201 |
+
"loss": 2.9142,
|
| 2202 |
+
"step": 294000
|
| 2203 |
+
},
|
| 2204 |
+
{
|
| 2205 |
+
"epoch": 15.86,
|
| 2206 |
+
"grad_norm": 0.8603065609931946,
|
| 2207 |
+
"learning_rate": 2.265636952044719e-05,
|
| 2208 |
+
"loss": 2.9092,
|
| 2209 |
+
"step": 295000
|
| 2210 |
+
},
|
| 2211 |
+
{
|
| 2212 |
+
"epoch": 15.92,
|
| 2213 |
+
"grad_norm": 0.8278149366378784,
|
| 2214 |
+
"learning_rate": 2.236245954692557e-05,
|
| 2215 |
+
"loss": 2.911,
|
| 2216 |
+
"step": 296000
|
| 2217 |
+
},
|
| 2218 |
+
{
|
| 2219 |
+
"epoch": 15.97,
|
| 2220 |
+
"grad_norm": 0.8719545006752014,
|
| 2221 |
+
"learning_rate": 2.2068255369226243e-05,
|
| 2222 |
+
"loss": 2.9106,
|
| 2223 |
+
"step": 297000
|
| 2224 |
+
},
|
| 2225 |
+
{
|
| 2226 |
+
"epoch": 16.0,
|
| 2227 |
+
"eval_accuracy": 0.4049153485882295,
|
| 2228 |
+
"eval_loss": 3.4084088802337646,
|
| 2229 |
+
"eval_runtime": 153.4722,
|
| 2230 |
+
"eval_samples_per_second": 377.384,
|
| 2231 |
+
"eval_steps_per_second": 5.897,
|
| 2232 |
+
"step": 297520
|
| 2233 |
+
},
|
| 2234 |
+
{
|
| 2235 |
+
"epoch": 16.03,
|
| 2236 |
+
"grad_norm": 0.8873878121376038,
|
| 2237 |
+
"learning_rate": 2.177405119152692e-05,
|
| 2238 |
+
"loss": 2.8945,
|
| 2239 |
+
"step": 298000
|
| 2240 |
+
},
|
| 2241 |
+
{
|
| 2242 |
+
"epoch": 16.08,
|
| 2243 |
+
"grad_norm": 0.9068606495857239,
|
| 2244 |
+
"learning_rate": 2.1480141218005298e-05,
|
| 2245 |
+
"loss": 2.8766,
|
| 2246 |
+
"step": 299000
|
| 2247 |
+
},
|
| 2248 |
+
{
|
| 2249 |
+
"epoch": 16.13,
|
| 2250 |
+
"grad_norm": 0.9187499284744263,
|
| 2251 |
+
"learning_rate": 2.1185937040305972e-05,
|
| 2252 |
+
"loss": 2.879,
|
| 2253 |
+
"step": 300000
|
| 2254 |
+
},
|
| 2255 |
+
{
|
| 2256 |
+
"epoch": 16.19,
|
| 2257 |
+
"grad_norm": 0.8892787098884583,
|
| 2258 |
+
"learning_rate": 2.089173286260665e-05,
|
| 2259 |
+
"loss": 2.8821,
|
| 2260 |
+
"step": 301000
|
| 2261 |
+
},
|
| 2262 |
+
{
|
| 2263 |
+
"epoch": 16.24,
|
| 2264 |
+
"grad_norm": 0.8787720203399658,
|
| 2265 |
+
"learning_rate": 2.0597822889085024e-05,
|
| 2266 |
+
"loss": 2.8863,
|
| 2267 |
+
"step": 302000
|
| 2268 |
+
},
|
| 2269 |
+
{
|
| 2270 |
+
"epoch": 16.29,
|
| 2271 |
+
"grad_norm": 0.8958452343940735,
|
| 2272 |
+
"learning_rate": 2.0303618711385702e-05,
|
| 2273 |
+
"loss": 2.8852,
|
| 2274 |
+
"step": 303000
|
| 2275 |
+
},
|
| 2276 |
+
{
|
| 2277 |
+
"epoch": 16.35,
|
| 2278 |
+
"grad_norm": 0.9086764454841614,
|
| 2279 |
+
"learning_rate": 2.000941453368638e-05,
|
| 2280 |
+
"loss": 2.8864,
|
| 2281 |
+
"step": 304000
|
| 2282 |
+
},
|
| 2283 |
+
{
|
| 2284 |
+
"epoch": 16.4,
|
| 2285 |
+
"grad_norm": 0.9159351587295532,
|
| 2286 |
+
"learning_rate": 1.9715210355987058e-05,
|
| 2287 |
+
"loss": 2.8904,
|
| 2288 |
+
"step": 305000
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 16.46,
|
| 2292 |
+
"grad_norm": 0.8615668416023254,
|
| 2293 |
+
"learning_rate": 1.942130038246543e-05,
|
| 2294 |
+
"loss": 2.886,
|
| 2295 |
+
"step": 306000
|
| 2296 |
+
},
|
| 2297 |
+
{
|
| 2298 |
+
"epoch": 16.51,
|
| 2299 |
+
"grad_norm": 0.8897385597229004,
|
| 2300 |
+
"learning_rate": 1.912709620476611e-05,
|
| 2301 |
+
"loss": 2.8863,
|
| 2302 |
+
"step": 307000
|
| 2303 |
+
},
|
| 2304 |
+
{
|
| 2305 |
+
"epoch": 16.56,
|
| 2306 |
+
"grad_norm": 0.8810350298881531,
|
| 2307 |
+
"learning_rate": 1.8833186231244483e-05,
|
| 2308 |
+
"loss": 2.8905,
|
| 2309 |
+
"step": 308000
|
| 2310 |
+
},
|
| 2311 |
+
{
|
| 2312 |
+
"epoch": 16.62,
|
| 2313 |
+
"grad_norm": 0.9394462704658508,
|
| 2314 |
+
"learning_rate": 1.853898205354516e-05,
|
| 2315 |
+
"loss": 2.8913,
|
| 2316 |
+
"step": 309000
|
| 2317 |
+
},
|
| 2318 |
+
{
|
| 2319 |
+
"epoch": 16.67,
|
| 2320 |
+
"grad_norm": 0.8938777446746826,
|
| 2321 |
+
"learning_rate": 1.824507208002354e-05,
|
| 2322 |
+
"loss": 2.8938,
|
| 2323 |
+
"step": 310000
|
| 2324 |
+
},
|
| 2325 |
+
{
|
| 2326 |
+
"epoch": 16.72,
|
| 2327 |
+
"grad_norm": 0.9033794403076172,
|
| 2328 |
+
"learning_rate": 1.7950867902324213e-05,
|
| 2329 |
+
"loss": 2.8934,
|
| 2330 |
+
"step": 311000
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 16.78,
|
| 2334 |
+
"grad_norm": 0.9354248642921448,
|
| 2335 |
+
"learning_rate": 1.765695792880259e-05,
|
| 2336 |
+
"loss": 2.8917,
|
| 2337 |
+
"step": 312000
|
| 2338 |
+
},
|
| 2339 |
+
{
|
| 2340 |
+
"epoch": 16.83,
|
| 2341 |
+
"grad_norm": 0.8875927329063416,
|
| 2342 |
+
"learning_rate": 1.7362753751103268e-05,
|
| 2343 |
+
"loss": 2.8934,
|
| 2344 |
+
"step": 313000
|
| 2345 |
+
},
|
| 2346 |
+
{
|
| 2347 |
+
"epoch": 16.89,
|
| 2348 |
+
"grad_norm": 0.9174014925956726,
|
| 2349 |
+
"learning_rate": 1.7068549573403943e-05,
|
| 2350 |
+
"loss": 2.8956,
|
| 2351 |
+
"step": 314000
|
| 2352 |
+
},
|
| 2353 |
+
{
|
| 2354 |
+
"epoch": 16.94,
|
| 2355 |
+
"grad_norm": 0.8926222920417786,
|
| 2356 |
+
"learning_rate": 1.6774345395704617e-05,
|
| 2357 |
+
"loss": 2.8933,
|
| 2358 |
+
"step": 315000
|
| 2359 |
+
},
|
| 2360 |
+
{
|
| 2361 |
+
"epoch": 16.99,
|
| 2362 |
+
"grad_norm": 0.9039079546928406,
|
| 2363 |
+
"learning_rate": 1.6480435422182998e-05,
|
| 2364 |
+
"loss": 2.8923,
|
| 2365 |
+
"step": 316000
|
| 2366 |
+
},
|
| 2367 |
+
{
|
| 2368 |
+
"epoch": 17.0,
|
| 2369 |
+
"eval_accuracy": 0.4055247570595917,
|
| 2370 |
+
"eval_loss": 3.4154162406921387,
|
| 2371 |
+
"eval_runtime": 153.7223,
|
| 2372 |
+
"eval_samples_per_second": 376.77,
|
| 2373 |
+
"eval_steps_per_second": 5.887,
|
| 2374 |
+
"step": 316115
|
| 2375 |
+
},
|
| 2376 |
+
{
|
| 2377 |
+
"epoch": 17.05,
|
| 2378 |
+
"grad_norm": 0.93231600522995,
|
| 2379 |
+
"learning_rate": 1.6186231244483672e-05,
|
| 2380 |
+
"loss": 2.8687,
|
| 2381 |
+
"step": 317000
|
| 2382 |
+
},
|
| 2383 |
+
{
|
| 2384 |
+
"epoch": 17.1,
|
| 2385 |
+
"grad_norm": 0.8828312158584595,
|
| 2386 |
+
"learning_rate": 1.589202706678435e-05,
|
| 2387 |
+
"loss": 2.865,
|
| 2388 |
+
"step": 318000
|
| 2389 |
+
},
|
| 2390 |
+
{
|
| 2391 |
+
"epoch": 17.16,
|
| 2392 |
+
"grad_norm": 0.9261276721954346,
|
| 2393 |
+
"learning_rate": 1.5598117093262727e-05,
|
| 2394 |
+
"loss": 2.8687,
|
| 2395 |
+
"step": 319000
|
| 2396 |
+
},
|
| 2397 |
+
{
|
| 2398 |
+
"epoch": 17.21,
|
| 2399 |
+
"grad_norm": 0.9356419444084167,
|
| 2400 |
+
"learning_rate": 1.5303912915563402e-05,
|
| 2401 |
+
"loss": 2.8681,
|
| 2402 |
+
"step": 320000
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 17.26,
|
| 2406 |
+
"grad_norm": 0.9402567744255066,
|
| 2407 |
+
"learning_rate": 1.5009708737864078e-05,
|
| 2408 |
+
"loss": 2.8676,
|
| 2409 |
+
"step": 321000
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 17.32,
|
| 2413 |
+
"grad_norm": 0.9250918626785278,
|
| 2414 |
+
"learning_rate": 1.4715504560164756e-05,
|
| 2415 |
+
"loss": 2.8729,
|
| 2416 |
+
"step": 322000
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 17.37,
|
| 2420 |
+
"grad_norm": 0.9387281537055969,
|
| 2421 |
+
"learning_rate": 1.442159458664313e-05,
|
| 2422 |
+
"loss": 2.8744,
|
| 2423 |
+
"step": 323000
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 17.42,
|
| 2427 |
+
"grad_norm": 0.9035524725914001,
|
| 2428 |
+
"learning_rate": 1.4127390408943808e-05,
|
| 2429 |
+
"loss": 2.8718,
|
| 2430 |
+
"step": 324000
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 17.48,
|
| 2434 |
+
"grad_norm": 0.8882308602333069,
|
| 2435 |
+
"learning_rate": 1.3833480435422183e-05,
|
| 2436 |
+
"loss": 2.8704,
|
| 2437 |
+
"step": 325000
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 17.53,
|
| 2441 |
+
"grad_norm": 0.9479586482048035,
|
| 2442 |
+
"learning_rate": 1.3539276257722861e-05,
|
| 2443 |
+
"loss": 2.8766,
|
| 2444 |
+
"step": 326000
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 17.59,
|
| 2448 |
+
"grad_norm": 0.8966055512428284,
|
| 2449 |
+
"learning_rate": 1.3245072080023537e-05,
|
| 2450 |
+
"loss": 2.874,
|
| 2451 |
+
"step": 327000
|
| 2452 |
+
},
|
| 2453 |
+
{
|
| 2454 |
+
"epoch": 17.64,
|
| 2455 |
+
"grad_norm": 0.9272529482841492,
|
| 2456 |
+
"learning_rate": 1.2951162106501913e-05,
|
| 2457 |
+
"loss": 2.8737,
|
| 2458 |
+
"step": 328000
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 17.69,
|
| 2462 |
+
"grad_norm": 0.9414493441581726,
|
| 2463 |
+
"learning_rate": 1.2657252132980289e-05,
|
| 2464 |
+
"loss": 2.8767,
|
| 2465 |
+
"step": 329000
|
| 2466 |
+
},
|
| 2467 |
+
{
|
| 2468 |
+
"epoch": 17.75,
|
| 2469 |
+
"grad_norm": 0.9303755760192871,
|
| 2470 |
+
"learning_rate": 1.2363047955280965e-05,
|
| 2471 |
+
"loss": 2.8767,
|
| 2472 |
+
"step": 330000
|
| 2473 |
+
},
|
| 2474 |
+
{
|
| 2475 |
+
"epoch": 17.8,
|
| 2476 |
+
"grad_norm": 0.9229914546012878,
|
| 2477 |
+
"learning_rate": 1.2068843777581643e-05,
|
| 2478 |
+
"loss": 2.8778,
|
| 2479 |
+
"step": 331000
|
| 2480 |
+
},
|
| 2481 |
+
{
|
| 2482 |
+
"epoch": 17.85,
|
| 2483 |
+
"grad_norm": 0.9260939955711365,
|
| 2484 |
+
"learning_rate": 1.1774639599882319e-05,
|
| 2485 |
+
"loss": 2.8765,
|
| 2486 |
+
"step": 332000
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 17.91,
|
| 2490 |
+
"grad_norm": 0.9214343428611755,
|
| 2491 |
+
"learning_rate": 1.1480729626360694e-05,
|
| 2492 |
+
"loss": 2.8752,
|
| 2493 |
+
"step": 333000
|
| 2494 |
+
},
|
| 2495 |
+
{
|
| 2496 |
+
"epoch": 17.96,
|
| 2497 |
+
"grad_norm": 0.9373890161514282,
|
| 2498 |
+
"learning_rate": 1.118652544866137e-05,
|
| 2499 |
+
"loss": 2.8795,
|
| 2500 |
+
"step": 334000
|
| 2501 |
+
},
|
| 2502 |
+
{
|
| 2503 |
+
"epoch": 18.0,
|
| 2504 |
+
"eval_accuracy": 0.40571911476079897,
|
| 2505 |
+
"eval_loss": 3.419478178024292,
|
| 2506 |
+
"eval_runtime": 153.6818,
|
| 2507 |
+
"eval_samples_per_second": 376.87,
|
| 2508 |
+
"eval_steps_per_second": 5.889,
|
| 2509 |
+
"step": 334710
|
| 2510 |
+
},
|
| 2511 |
+
{
|
| 2512 |
+
"epoch": 18.02,
|
| 2513 |
+
"grad_norm": 0.9543386101722717,
|
| 2514 |
+
"learning_rate": 1.0892321270962048e-05,
|
| 2515 |
+
"loss": 2.8708,
|
| 2516 |
+
"step": 335000
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 18.07,
|
| 2520 |
+
"grad_norm": 0.893997311592102,
|
| 2521 |
+
"learning_rate": 1.0598411297440426e-05,
|
| 2522 |
+
"loss": 2.8512,
|
| 2523 |
+
"step": 336000
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 18.12,
|
| 2527 |
+
"grad_norm": 0.9292035102844238,
|
| 2528 |
+
"learning_rate": 1.03045013239188e-05,
|
| 2529 |
+
"loss": 2.8533,
|
| 2530 |
+
"step": 337000
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"epoch": 18.18,
|
| 2534 |
+
"grad_norm": 0.9333125948905945,
|
| 2535 |
+
"learning_rate": 1.0010297146219477e-05,
|
| 2536 |
+
"loss": 2.855,
|
| 2537 |
+
"step": 338000
|
| 2538 |
+
},
|
| 2539 |
+
{
|
| 2540 |
+
"epoch": 18.23,
|
| 2541 |
+
"grad_norm": 0.9297951459884644,
|
| 2542 |
+
"learning_rate": 9.716092968520154e-06,
|
| 2543 |
+
"loss": 2.8587,
|
| 2544 |
+
"step": 339000
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"epoch": 18.28,
|
| 2548 |
+
"grad_norm": 0.9800245761871338,
|
| 2549 |
+
"learning_rate": 9.421888790820831e-06,
|
| 2550 |
+
"loss": 2.8595,
|
| 2551 |
+
"step": 340000
|
| 2552 |
+
},
|
| 2553 |
+
{
|
| 2554 |
+
"epoch": 18.34,
|
| 2555 |
+
"grad_norm": 0.9462539553642273,
|
| 2556 |
+
"learning_rate": 9.127684613121506e-06,
|
| 2557 |
+
"loss": 2.8594,
|
| 2558 |
+
"step": 341000
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 18.39,
|
| 2562 |
+
"grad_norm": 0.9382343888282776,
|
| 2563 |
+
"learning_rate": 8.833774639599883e-06,
|
| 2564 |
+
"loss": 2.8603,
|
| 2565 |
+
"step": 342000
|
| 2566 |
+
},
|
| 2567 |
+
{
|
| 2568 |
+
"epoch": 18.45,
|
| 2569 |
+
"grad_norm": 0.9466389417648315,
|
| 2570 |
+
"learning_rate": 8.539864666078259e-06,
|
| 2571 |
+
"loss": 2.8575,
|
| 2572 |
+
"step": 343000
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"epoch": 18.5,
|
| 2576 |
+
"grad_norm": 0.9292024970054626,
|
| 2577 |
+
"learning_rate": 8.245660488378935e-06,
|
| 2578 |
+
"loss": 2.8561,
|
| 2579 |
+
"step": 344000
|
| 2580 |
+
},
|
| 2581 |
+
{
|
| 2582 |
+
"epoch": 18.55,
|
| 2583 |
+
"grad_norm": 0.9281138181686401,
|
| 2584 |
+
"learning_rate": 7.951456310679611e-06,
|
| 2585 |
+
"loss": 2.859,
|
| 2586 |
+
"step": 345000
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"epoch": 18.61,
|
| 2590 |
+
"grad_norm": 0.9447932839393616,
|
| 2591 |
+
"learning_rate": 7.657546337157988e-06,
|
| 2592 |
+
"loss": 2.8596,
|
| 2593 |
+
"step": 346000
|
| 2594 |
+
},
|
| 2595 |
+
{
|
| 2596 |
+
"epoch": 18.66,
|
| 2597 |
+
"grad_norm": 0.9081346988677979,
|
| 2598 |
+
"learning_rate": 7.363342159458665e-06,
|
| 2599 |
+
"loss": 2.8632,
|
| 2600 |
+
"step": 347000
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 18.71,
|
| 2604 |
+
"grad_norm": 0.924645185470581,
|
| 2605 |
+
"learning_rate": 7.0691379817593416e-06,
|
| 2606 |
+
"loss": 2.8609,
|
| 2607 |
+
"step": 348000
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 18.77,
|
| 2611 |
+
"grad_norm": 0.9200485944747925,
|
| 2612 |
+
"learning_rate": 6.774933804060019e-06,
|
| 2613 |
+
"loss": 2.8629,
|
| 2614 |
+
"step": 349000
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"epoch": 18.82,
|
| 2618 |
+
"grad_norm": 0.9209837317466736,
|
| 2619 |
+
"learning_rate": 6.481023830538393e-06,
|
| 2620 |
+
"loss": 2.8599,
|
| 2621 |
+
"step": 350000
|
| 2622 |
+
},
|
| 2623 |
+
{
|
| 2624 |
+
"epoch": 18.88,
|
| 2625 |
+
"grad_norm": 0.9589852094650269,
|
| 2626 |
+
"learning_rate": 6.186819652839071e-06,
|
| 2627 |
+
"loss": 2.8617,
|
| 2628 |
+
"step": 351000
|
| 2629 |
+
},
|
| 2630 |
+
{
|
| 2631 |
+
"epoch": 18.93,
|
| 2632 |
+
"grad_norm": 0.896015465259552,
|
| 2633 |
+
"learning_rate": 5.892615475139747e-06,
|
| 2634 |
+
"loss": 2.8612,
|
| 2635 |
+
"step": 352000
|
| 2636 |
+
},
|
| 2637 |
+
{
|
| 2638 |
+
"epoch": 18.98,
|
| 2639 |
+
"grad_norm": 0.971016526222229,
|
| 2640 |
+
"learning_rate": 5.598705501618123e-06,
|
| 2641 |
+
"loss": 2.8628,
|
| 2642 |
+
"step": 353000
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 19.0,
|
| 2646 |
+
"eval_accuracy": 0.40566355519671377,
|
| 2647 |
+
"eval_loss": 3.422478199005127,
|
| 2648 |
+
"eval_runtime": 153.9588,
|
| 2649 |
+
"eval_samples_per_second": 376.192,
|
| 2650 |
+
"eval_steps_per_second": 5.878,
|
| 2651 |
+
"step": 353305
|
| 2652 |
+
},
|
| 2653 |
+
{
|
| 2654 |
+
"epoch": 19.04,
|
| 2655 |
+
"grad_norm": 0.9984861612319946,
|
| 2656 |
+
"learning_rate": 5.304795528096499e-06,
|
| 2657 |
+
"loss": 2.8502,
|
| 2658 |
+
"step": 354000
|
| 2659 |
+
},
|
| 2660 |
+
{
|
| 2661 |
+
"epoch": 19.09,
|
| 2662 |
+
"grad_norm": 0.9807627201080322,
|
| 2663 |
+
"learning_rate": 5.0105913503971756e-06,
|
| 2664 |
+
"loss": 2.8423,
|
| 2665 |
+
"step": 355000
|
| 2666 |
+
},
|
| 2667 |
+
{
|
| 2668 |
+
"epoch": 19.14,
|
| 2669 |
+
"grad_norm": 0.9239267110824585,
|
| 2670 |
+
"learning_rate": 4.7163871726978526e-06,
|
| 2671 |
+
"loss": 2.8449,
|
| 2672 |
+
"step": 356000
|
| 2673 |
+
},
|
| 2674 |
+
{
|
| 2675 |
+
"epoch": 19.2,
|
| 2676 |
+
"grad_norm": 0.9425461292266846,
|
| 2677 |
+
"learning_rate": 4.422182994998529e-06,
|
| 2678 |
+
"loss": 2.8478,
|
| 2679 |
+
"step": 357000
|
| 2680 |
+
},
|
| 2681 |
+
{
|
| 2682 |
+
"epoch": 19.25,
|
| 2683 |
+
"grad_norm": 0.9452289938926697,
|
| 2684 |
+
"learning_rate": 4.128273021476905e-06,
|
| 2685 |
+
"loss": 2.8426,
|
| 2686 |
+
"step": 358000
|
| 2687 |
+
},
|
| 2688 |
+
{
|
| 2689 |
+
"epoch": 19.31,
|
| 2690 |
+
"grad_norm": 0.8920537233352661,
|
| 2691 |
+
"learning_rate": 3.834068843777581e-06,
|
| 2692 |
+
"loss": 2.8425,
|
| 2693 |
+
"step": 359000
|
| 2694 |
+
},
|
| 2695 |
+
{
|
| 2696 |
+
"epoch": 19.36,
|
| 2697 |
+
"grad_norm": 0.968367338180542,
|
| 2698 |
+
"learning_rate": 3.5401588702559578e-06,
|
| 2699 |
+
"loss": 2.8503,
|
| 2700 |
+
"step": 360000
|
| 2701 |
+
},
|
| 2702 |
+
{
|
| 2703 |
+
"epoch": 19.41,
|
| 2704 |
+
"grad_norm": 0.9560465216636658,
|
| 2705 |
+
"learning_rate": 3.245954692556635e-06,
|
| 2706 |
+
"loss": 2.8474,
|
| 2707 |
+
"step": 361000
|
| 2708 |
+
},
|
| 2709 |
+
{
|
| 2710 |
+
"epoch": 19.47,
|
| 2711 |
+
"grad_norm": 0.9240676760673523,
|
| 2712 |
+
"learning_rate": 2.951750514857311e-06,
|
| 2713 |
+
"loss": 2.8452,
|
| 2714 |
+
"step": 362000
|
| 2715 |
+
},
|
| 2716 |
+
{
|
| 2717 |
+
"epoch": 19.52,
|
| 2718 |
+
"grad_norm": 0.9011372327804565,
|
| 2719 |
+
"learning_rate": 2.657840541335687e-06,
|
| 2720 |
+
"loss": 2.8438,
|
| 2721 |
+
"step": 363000
|
| 2722 |
+
},
|
| 2723 |
+
{
|
| 2724 |
+
"epoch": 19.58,
|
| 2725 |
+
"grad_norm": 0.9103080034255981,
|
| 2726 |
+
"learning_rate": 2.3636363636363636e-06,
|
| 2727 |
+
"loss": 2.85,
|
| 2728 |
+
"step": 364000
|
| 2729 |
+
},
|
| 2730 |
+
{
|
| 2731 |
+
"epoch": 19.63,
|
| 2732 |
+
"grad_norm": 0.9446433782577515,
|
| 2733 |
+
"learning_rate": 2.070020594292439e-06,
|
| 2734 |
+
"loss": 2.848,
|
| 2735 |
+
"step": 365000
|
| 2736 |
+
},
|
| 2737 |
+
{
|
| 2738 |
+
"epoch": 19.68,
|
| 2739 |
+
"grad_norm": 0.927521824836731,
|
| 2740 |
+
"learning_rate": 1.7758164165931158e-06,
|
| 2741 |
+
"loss": 2.8497,
|
| 2742 |
+
"step": 366000
|
| 2743 |
+
},
|
| 2744 |
+
{
|
| 2745 |
+
"epoch": 19.74,
|
| 2746 |
+
"grad_norm": 0.9345549941062927,
|
| 2747 |
+
"learning_rate": 1.4816122388937924e-06,
|
| 2748 |
+
"loss": 2.8488,
|
| 2749 |
+
"step": 367000
|
| 2750 |
+
},
|
| 2751 |
+
{
|
| 2752 |
+
"epoch": 19.79,
|
| 2753 |
+
"grad_norm": 0.9598424434661865,
|
| 2754 |
+
"learning_rate": 1.187408061194469e-06,
|
| 2755 |
+
"loss": 2.8465,
|
| 2756 |
+
"step": 368000
|
| 2757 |
+
},
|
| 2758 |
+
{
|
| 2759 |
+
"epoch": 19.84,
|
| 2760 |
+
"grad_norm": 0.9485284090042114,
|
| 2761 |
+
"learning_rate": 8.932038834951457e-07,
|
| 2762 |
+
"loss": 2.8497,
|
| 2763 |
+
"step": 369000
|
| 2764 |
+
},
|
| 2765 |
+
{
|
| 2766 |
+
"epoch": 19.9,
|
| 2767 |
+
"grad_norm": 0.9556813836097717,
|
| 2768 |
+
"learning_rate": 5.992939099735216e-07,
|
| 2769 |
+
"loss": 2.8439,
|
| 2770 |
+
"step": 370000
|
| 2771 |
+
},
|
| 2772 |
+
{
|
| 2773 |
+
"epoch": 19.95,
|
| 2774 |
+
"grad_norm": 0.9353678226470947,
|
| 2775 |
+
"learning_rate": 3.050897322741983e-07,
|
| 2776 |
+
"loss": 2.8497,
|
| 2777 |
+
"step": 371000
|
| 2778 |
+
},
|
| 2779 |
+
{
|
| 2780 |
+
"epoch": 20.0,
|
| 2781 |
+
"eval_accuracy": 0.4055823320854937,
|
| 2782 |
+
"eval_loss": 3.4263744354248047,
|
| 2783 |
+
"eval_runtime": 153.7409,
|
| 2784 |
+
"eval_samples_per_second": 376.725,
|
| 2785 |
+
"eval_steps_per_second": 5.887,
|
| 2786 |
+
"step": 371900
|
| 2787 |
+
},
|
| 2788 |
+
{
|
| 2789 |
+
"epoch": 20.0,
|
| 2790 |
+
"step": 371900,
|
| 2791 |
+
"total_flos": 1.56693365388288e+18,
|
| 2792 |
+
"train_loss": 3.1576745831950372,
|
| 2793 |
+
"train_runtime": 81271.0608,
|
| 2794 |
+
"train_samples_per_second": 146.433,
|
| 2795 |
+
"train_steps_per_second": 4.576
|
| 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 |
+
}
|