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_211-1e-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_211-1e-4
|
| 15 |
|
| 16 |
-
This model was trained from scratch on
|
| 17 |
It achieves the following results on the evaluation set:
|
| 18 |
- Loss: 3.4323
|
| 19 |
- Accuracy: 0.4055
|
|
|
|
| 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_211-1e-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: 0.40547080986496004
|
| 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_211-1e-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: 3.4323
|
| 31 |
- Accuracy: 0.4055
|
all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"eval_accuracy": 0.40547080986496004,
|
| 4 |
+
"eval_loss": 3.432330846786499,
|
| 5 |
+
"eval_runtime": 153.444,
|
| 6 |
+
"eval_samples": 57918,
|
| 7 |
+
"eval_samples_per_second": 377.454,
|
| 8 |
+
"eval_steps_per_second": 5.898,
|
| 9 |
+
"perplexity": 30.948695412445822,
|
| 10 |
+
"train_loss": 3.1561156885198893,
|
| 11 |
+
"train_runtime": 81199.8655,
|
| 12 |
+
"train_samples": 595038,
|
| 13 |
+
"train_samples_per_second": 146.561,
|
| 14 |
+
"train_steps_per_second": 4.58
|
| 15 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"eval_accuracy": 0.40547080986496004,
|
| 4 |
+
"eval_loss": 3.432330846786499,
|
| 5 |
+
"eval_runtime": 153.444,
|
| 6 |
+
"eval_samples": 57918,
|
| 7 |
+
"eval_samples_per_second": 377.454,
|
| 8 |
+
"eval_steps_per_second": 5.898,
|
| 9 |
+
"perplexity": 30.948695412445822
|
| 10 |
+
}
|
runs/Jul06_17-30-25_phyl-ling-p01.la.utexas.edu/events.out.tfevents.1720386752.phyl-ling-p01.la.utexas.edu.2019449.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:28ac80fc7b24ff4034b4d55e0ce118b9b474df0f7b0df1b5d651b4d0fa9a01a4
|
| 3 |
+
size 417
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"train_loss": 3.1561156885198893,
|
| 4 |
+
"train_runtime": 81199.8655,
|
| 5 |
+
"train_samples": 595038,
|
| 6 |
+
"train_samples_per_second": 146.561,
|
| 7 |
+
"train_steps_per_second": 4.58
|
| 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": 1.1065584421157837,
|
| 14 |
+
"learning_rate": 3.125e-06,
|
| 15 |
+
"loss": 7.5986,
|
| 16 |
+
"step": 1000
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.11,
|
| 20 |
+
"grad_norm": 0.6919564604759216,
|
| 21 |
+
"learning_rate": 6.25e-06,
|
| 22 |
+
"loss": 5.8259,
|
| 23 |
+
"step": 2000
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.16,
|
| 27 |
+
"grad_norm": 0.7977421283721924,
|
| 28 |
+
"learning_rate": 9.375000000000001e-06,
|
| 29 |
+
"loss": 5.3849,
|
| 30 |
+
"step": 3000
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.22,
|
| 34 |
+
"grad_norm": 1.386985421180725,
|
| 35 |
+
"learning_rate": 1.25e-05,
|
| 36 |
+
"loss": 5.1627,
|
| 37 |
+
"step": 4000
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.27,
|
| 41 |
+
"grad_norm": 0.9054310917854309,
|
| 42 |
+
"learning_rate": 1.5625e-05,
|
| 43 |
+
"loss": 5.005,
|
| 44 |
+
"step": 5000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.32,
|
| 48 |
+
"grad_norm": 1.0550600290298462,
|
| 49 |
+
"learning_rate": 1.8750000000000002e-05,
|
| 50 |
+
"loss": 4.8627,
|
| 51 |
+
"step": 6000
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.38,
|
| 55 |
+
"grad_norm": 1.0669718980789185,
|
| 56 |
+
"learning_rate": 2.1875e-05,
|
| 57 |
+
"loss": 4.7454,
|
| 58 |
+
"step": 7000
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.43,
|
| 62 |
+
"grad_norm": 1.0990982055664062,
|
| 63 |
+
"learning_rate": 2.5e-05,
|
| 64 |
+
"loss": 4.6486,
|
| 65 |
+
"step": 8000
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.48,
|
| 69 |
+
"grad_norm": 1.0567301511764526,
|
| 70 |
+
"learning_rate": 2.8125000000000003e-05,
|
| 71 |
+
"loss": 4.5585,
|
| 72 |
+
"step": 9000
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.54,
|
| 76 |
+
"grad_norm": 1.1454998254776,
|
| 77 |
+
"learning_rate": 3.125e-05,
|
| 78 |
+
"loss": 4.4828,
|
| 79 |
+
"step": 10000
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.59,
|
| 83 |
+
"grad_norm": 1.0922003984451294,
|
| 84 |
+
"learning_rate": 3.4375e-05,
|
| 85 |
+
"loss": 4.4099,
|
| 86 |
+
"step": 11000
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.65,
|
| 90 |
+
"grad_norm": 1.0760829448699951,
|
| 91 |
+
"learning_rate": 3.7500000000000003e-05,
|
| 92 |
+
"loss": 4.3435,
|
| 93 |
+
"step": 12000
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.7,
|
| 97 |
+
"grad_norm": 1.0779520273208618,
|
| 98 |
+
"learning_rate": 4.0621875e-05,
|
| 99 |
+
"loss": 4.2871,
|
| 100 |
+
"step": 13000
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.75,
|
| 104 |
+
"grad_norm": 1.0053024291992188,
|
| 105 |
+
"learning_rate": 4.374375e-05,
|
| 106 |
+
"loss": 4.2362,
|
| 107 |
+
"step": 14000
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.81,
|
| 111 |
+
"grad_norm": 1.0250402688980103,
|
| 112 |
+
"learning_rate": 4.686875e-05,
|
| 113 |
+
"loss": 4.1841,
|
| 114 |
+
"step": 15000
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.86,
|
| 118 |
+
"grad_norm": 1.025078296661377,
|
| 119 |
+
"learning_rate": 4.999375e-05,
|
| 120 |
+
"loss": 4.1362,
|
| 121 |
+
"step": 16000
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.91,
|
| 125 |
+
"grad_norm": 1.020418643951416,
|
| 126 |
+
"learning_rate": 5.311875000000001e-05,
|
| 127 |
+
"loss": 4.091,
|
| 128 |
+
"step": 17000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.97,
|
| 132 |
+
"grad_norm": 1.023269772529602,
|
| 133 |
+
"learning_rate": 5.6240625e-05,
|
| 134 |
+
"loss": 4.0524,
|
| 135 |
+
"step": 18000
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 1.0,
|
| 139 |
+
"eval_accuracy": 0.30844237989493534,
|
| 140 |
+
"eval_loss": 4.28109073638916,
|
| 141 |
+
"eval_runtime": 151.382,
|
| 142 |
+
"eval_samples_per_second": 382.595,
|
| 143 |
+
"eval_steps_per_second": 5.978,
|
| 144 |
+
"step": 18595
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 1.02,
|
| 148 |
+
"grad_norm": 1.0309756994247437,
|
| 149 |
+
"learning_rate": 5.9365625e-05,
|
| 150 |
+
"loss": 4.015,
|
| 151 |
+
"step": 19000
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 1.08,
|
| 155 |
+
"grad_norm": 0.9628090262413025,
|
| 156 |
+
"learning_rate": 6.24875e-05,
|
| 157 |
+
"loss": 3.969,
|
| 158 |
+
"step": 20000
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.13,
|
| 162 |
+
"grad_norm": 1.0119906663894653,
|
| 163 |
+
"learning_rate": 6.560937500000001e-05,
|
| 164 |
+
"loss": 3.933,
|
| 165 |
+
"step": 21000
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 1.18,
|
| 169 |
+
"grad_norm": 0.9792044758796692,
|
| 170 |
+
"learning_rate": 6.8734375e-05,
|
| 171 |
+
"loss": 3.8956,
|
| 172 |
+
"step": 22000
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 1.24,
|
| 176 |
+
"grad_norm": 1.0101176500320435,
|
| 177 |
+
"learning_rate": 7.185312499999999e-05,
|
| 178 |
+
"loss": 3.8586,
|
| 179 |
+
"step": 23000
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.29,
|
| 183 |
+
"grad_norm": 0.9678108096122742,
|
| 184 |
+
"learning_rate": 7.4978125e-05,
|
| 185 |
+
"loss": 3.8302,
|
| 186 |
+
"step": 24000
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 1.34,
|
| 190 |
+
"grad_norm": 0.9615768194198608,
|
| 191 |
+
"learning_rate": 7.8103125e-05,
|
| 192 |
+
"loss": 3.801,
|
| 193 |
+
"step": 25000
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 1.4,
|
| 197 |
+
"grad_norm": 0.9395524859428406,
|
| 198 |
+
"learning_rate": 8.1221875e-05,
|
| 199 |
+
"loss": 3.7804,
|
| 200 |
+
"step": 26000
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 1.45,
|
| 204 |
+
"grad_norm": 0.8997763395309448,
|
| 205 |
+
"learning_rate": 8.434687500000001e-05,
|
| 206 |
+
"loss": 3.7585,
|
| 207 |
+
"step": 27000
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 1.51,
|
| 211 |
+
"grad_norm": 0.9433351159095764,
|
| 212 |
+
"learning_rate": 8.746875e-05,
|
| 213 |
+
"loss": 3.732,
|
| 214 |
+
"step": 28000
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 1.56,
|
| 218 |
+
"grad_norm": 0.8686715960502625,
|
| 219 |
+
"learning_rate": 9.059375e-05,
|
| 220 |
+
"loss": 3.7131,
|
| 221 |
+
"step": 29000
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.61,
|
| 225 |
+
"grad_norm": 0.9142166376113892,
|
| 226 |
+
"learning_rate": 9.3715625e-05,
|
| 227 |
+
"loss": 3.6941,
|
| 228 |
+
"step": 30000
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 1.67,
|
| 232 |
+
"grad_norm": 0.9287761449813843,
|
| 233 |
+
"learning_rate": 9.684062500000001e-05,
|
| 234 |
+
"loss": 3.6698,
|
| 235 |
+
"step": 31000
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 1.72,
|
| 239 |
+
"grad_norm": 0.8525364995002747,
|
| 240 |
+
"learning_rate": 9.99625e-05,
|
| 241 |
+
"loss": 3.6574,
|
| 242 |
+
"step": 32000
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.77,
|
| 246 |
+
"grad_norm": 0.8378422260284424,
|
| 247 |
+
"learning_rate": 9.970932627243307e-05,
|
| 248 |
+
"loss": 3.6384,
|
| 249 |
+
"step": 33000
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 1.83,
|
| 253 |
+
"grad_norm": 0.8161050081253052,
|
| 254 |
+
"learning_rate": 9.941512209473375e-05,
|
| 255 |
+
"loss": 3.6184,
|
| 256 |
+
"step": 34000
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 1.88,
|
| 260 |
+
"grad_norm": 0.8268817067146301,
|
| 261 |
+
"learning_rate": 9.912121212121213e-05,
|
| 262 |
+
"loss": 3.5973,
|
| 263 |
+
"step": 35000
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 1.94,
|
| 267 |
+
"grad_norm": 0.7921673059463501,
|
| 268 |
+
"learning_rate": 9.882700794351281e-05,
|
| 269 |
+
"loss": 3.5878,
|
| 270 |
+
"step": 36000
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 1.99,
|
| 274 |
+
"grad_norm": 0.7795915007591248,
|
| 275 |
+
"learning_rate": 9.853309796999118e-05,
|
| 276 |
+
"loss": 3.5716,
|
| 277 |
+
"step": 37000
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 2.0,
|
| 281 |
+
"eval_accuracy": 0.36151083317444777,
|
| 282 |
+
"eval_loss": 3.7749860286712646,
|
| 283 |
+
"eval_runtime": 153.0787,
|
| 284 |
+
"eval_samples_per_second": 378.354,
|
| 285 |
+
"eval_steps_per_second": 5.912,
|
| 286 |
+
"step": 37190
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 2.04,
|
| 290 |
+
"grad_norm": 0.816111147403717,
|
| 291 |
+
"learning_rate": 9.823889379229185e-05,
|
| 292 |
+
"loss": 3.5338,
|
| 293 |
+
"step": 38000
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 2.1,
|
| 297 |
+
"grad_norm": 0.8356756567955017,
|
| 298 |
+
"learning_rate": 9.794498381877023e-05,
|
| 299 |
+
"loss": 3.5187,
|
| 300 |
+
"step": 39000
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 2.15,
|
| 304 |
+
"grad_norm": 0.8443105220794678,
|
| 305 |
+
"learning_rate": 9.765077964107092e-05,
|
| 306 |
+
"loss": 3.5061,
|
| 307 |
+
"step": 40000
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 2.2,
|
| 311 |
+
"grad_norm": 0.7713498473167419,
|
| 312 |
+
"learning_rate": 9.735686966754929e-05,
|
| 313 |
+
"loss": 3.4944,
|
| 314 |
+
"step": 41000
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 2.26,
|
| 318 |
+
"grad_norm": 0.795734167098999,
|
| 319 |
+
"learning_rate": 9.706266548984995e-05,
|
| 320 |
+
"loss": 3.496,
|
| 321 |
+
"step": 42000
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 2.31,
|
| 325 |
+
"grad_norm": 0.8273535370826721,
|
| 326 |
+
"learning_rate": 9.676875551632834e-05,
|
| 327 |
+
"loss": 3.4833,
|
| 328 |
+
"step": 43000
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 2.37,
|
| 332 |
+
"grad_norm": 0.7914260625839233,
|
| 333 |
+
"learning_rate": 9.647455133862902e-05,
|
| 334 |
+
"loss": 3.4718,
|
| 335 |
+
"step": 44000
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 2.42,
|
| 339 |
+
"grad_norm": 0.782301664352417,
|
| 340 |
+
"learning_rate": 9.618064136510739e-05,
|
| 341 |
+
"loss": 3.46,
|
| 342 |
+
"step": 45000
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 2.47,
|
| 346 |
+
"grad_norm": 0.7781464457511902,
|
| 347 |
+
"learning_rate": 9.588643718740806e-05,
|
| 348 |
+
"loss": 3.4537,
|
| 349 |
+
"step": 46000
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 2.53,
|
| 353 |
+
"grad_norm": 0.774398148059845,
|
| 354 |
+
"learning_rate": 9.559223300970874e-05,
|
| 355 |
+
"loss": 3.4478,
|
| 356 |
+
"step": 47000
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.58,
|
| 360 |
+
"grad_norm": 0.7705032229423523,
|
| 361 |
+
"learning_rate": 9.529832303618712e-05,
|
| 362 |
+
"loss": 3.4398,
|
| 363 |
+
"step": 48000
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 2.64,
|
| 367 |
+
"grad_norm": 0.773110568523407,
|
| 368 |
+
"learning_rate": 9.50041188584878e-05,
|
| 369 |
+
"loss": 3.4366,
|
| 370 |
+
"step": 49000
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 2.69,
|
| 374 |
+
"grad_norm": 0.7716487646102905,
|
| 375 |
+
"learning_rate": 9.471050308914386e-05,
|
| 376 |
+
"loss": 3.4256,
|
| 377 |
+
"step": 50000
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 2.74,
|
| 381 |
+
"grad_norm": 0.7782188653945923,
|
| 382 |
+
"learning_rate": 9.441629891144455e-05,
|
| 383 |
+
"loss": 3.417,
|
| 384 |
+
"step": 51000
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 2.8,
|
| 388 |
+
"grad_norm": 0.7639435529708862,
|
| 389 |
+
"learning_rate": 9.412209473374523e-05,
|
| 390 |
+
"loss": 3.4123,
|
| 391 |
+
"step": 52000
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 2.85,
|
| 395 |
+
"grad_norm": 0.7292437553405762,
|
| 396 |
+
"learning_rate": 9.382789055604591e-05,
|
| 397 |
+
"loss": 3.405,
|
| 398 |
+
"step": 53000
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 2.9,
|
| 402 |
+
"grad_norm": 0.7682722210884094,
|
| 403 |
+
"learning_rate": 9.353368637834658e-05,
|
| 404 |
+
"loss": 3.399,
|
| 405 |
+
"step": 54000
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 2.96,
|
| 409 |
+
"grad_norm": 0.7343226671218872,
|
| 410 |
+
"learning_rate": 9.323977640482495e-05,
|
| 411 |
+
"loss": 3.3913,
|
| 412 |
+
"step": 55000
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 3.0,
|
| 416 |
+
"eval_accuracy": 0.3779394670823355,
|
| 417 |
+
"eval_loss": 3.600841760635376,
|
| 418 |
+
"eval_runtime": 153.5925,
|
| 419 |
+
"eval_samples_per_second": 377.089,
|
| 420 |
+
"eval_steps_per_second": 5.892,
|
| 421 |
+
"step": 55785
|
| 422 |
+
},
|
| 423 |
+
{
|
| 424 |
+
"epoch": 3.01,
|
| 425 |
+
"grad_norm": 0.7667288184165955,
|
| 426 |
+
"learning_rate": 9.294557222712563e-05,
|
| 427 |
+
"loss": 3.3741,
|
| 428 |
+
"step": 56000
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 3.07,
|
| 432 |
+
"grad_norm": 0.791042685508728,
|
| 433 |
+
"learning_rate": 9.265166225360401e-05,
|
| 434 |
+
"loss": 3.3318,
|
| 435 |
+
"step": 57000
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 3.12,
|
| 439 |
+
"grad_norm": 0.7432328462600708,
|
| 440 |
+
"learning_rate": 9.235745807590468e-05,
|
| 441 |
+
"loss": 3.3377,
|
| 442 |
+
"step": 58000
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 3.17,
|
| 446 |
+
"grad_norm": 0.7701253294944763,
|
| 447 |
+
"learning_rate": 9.206325389820536e-05,
|
| 448 |
+
"loss": 3.3324,
|
| 449 |
+
"step": 59000
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 3.23,
|
| 453 |
+
"grad_norm": 0.7645742297172546,
|
| 454 |
+
"learning_rate": 9.176963812886143e-05,
|
| 455 |
+
"loss": 3.327,
|
| 456 |
+
"step": 60000
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 3.28,
|
| 460 |
+
"grad_norm": 0.7636393904685974,
|
| 461 |
+
"learning_rate": 9.147543395116211e-05,
|
| 462 |
+
"loss": 3.3306,
|
| 463 |
+
"step": 61000
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 3.33,
|
| 467 |
+
"grad_norm": 0.7321722507476807,
|
| 468 |
+
"learning_rate": 9.118122977346278e-05,
|
| 469 |
+
"loss": 3.3303,
|
| 470 |
+
"step": 62000
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 3.39,
|
| 474 |
+
"grad_norm": 0.790723443031311,
|
| 475 |
+
"learning_rate": 9.088702559576346e-05,
|
| 476 |
+
"loss": 3.3243,
|
| 477 |
+
"step": 63000
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 3.44,
|
| 481 |
+
"grad_norm": 0.7484601140022278,
|
| 482 |
+
"learning_rate": 9.059311562224185e-05,
|
| 483 |
+
"loss": 3.3244,
|
| 484 |
+
"step": 64000
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 3.5,
|
| 488 |
+
"grad_norm": 0.746504008769989,
|
| 489 |
+
"learning_rate": 9.029891144454253e-05,
|
| 490 |
+
"loss": 3.3176,
|
| 491 |
+
"step": 65000
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 3.55,
|
| 495 |
+
"grad_norm": 0.7723560929298401,
|
| 496 |
+
"learning_rate": 9.000500147102089e-05,
|
| 497 |
+
"loss": 3.3123,
|
| 498 |
+
"step": 66000
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 3.6,
|
| 502 |
+
"grad_norm": 0.7322754859924316,
|
| 503 |
+
"learning_rate": 8.971079729332157e-05,
|
| 504 |
+
"loss": 3.3116,
|
| 505 |
+
"step": 67000
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 3.66,
|
| 509 |
+
"grad_norm": 0.7716720104217529,
|
| 510 |
+
"learning_rate": 8.941659311562225e-05,
|
| 511 |
+
"loss": 3.3073,
|
| 512 |
+
"step": 68000
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 3.71,
|
| 516 |
+
"grad_norm": 0.7449876666069031,
|
| 517 |
+
"learning_rate": 8.912268314210063e-05,
|
| 518 |
+
"loss": 3.3009,
|
| 519 |
+
"step": 69000
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 3.76,
|
| 523 |
+
"grad_norm": 0.7449329495429993,
|
| 524 |
+
"learning_rate": 8.88284789644013e-05,
|
| 525 |
+
"loss": 3.306,
|
| 526 |
+
"step": 70000
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 3.82,
|
| 530 |
+
"grad_norm": 0.701579213142395,
|
| 531 |
+
"learning_rate": 8.853427478670198e-05,
|
| 532 |
+
"loss": 3.2996,
|
| 533 |
+
"step": 71000
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 3.87,
|
| 537 |
+
"grad_norm": 0.7269728183746338,
|
| 538 |
+
"learning_rate": 8.824036481318035e-05,
|
| 539 |
+
"loss": 3.294,
|
| 540 |
+
"step": 72000
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 3.93,
|
| 544 |
+
"grad_norm": 0.7257309556007385,
|
| 545 |
+
"learning_rate": 8.794616063548102e-05,
|
| 546 |
+
"loss": 3.2955,
|
| 547 |
+
"step": 73000
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 3.98,
|
| 551 |
+
"grad_norm": 0.6977525353431702,
|
| 552 |
+
"learning_rate": 8.76522506619594e-05,
|
| 553 |
+
"loss": 3.2921,
|
| 554 |
+
"step": 74000
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 4.0,
|
| 558 |
+
"eval_accuracy": 0.3883133849640905,
|
| 559 |
+
"eval_loss": 3.5102479457855225,
|
| 560 |
+
"eval_runtime": 153.2986,
|
| 561 |
+
"eval_samples_per_second": 377.812,
|
| 562 |
+
"eval_steps_per_second": 5.904,
|
| 563 |
+
"step": 74380
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 4.03,
|
| 567 |
+
"grad_norm": 0.7187395691871643,
|
| 568 |
+
"learning_rate": 8.735804648426008e-05,
|
| 569 |
+
"loss": 3.2537,
|
| 570 |
+
"step": 75000
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 4.09,
|
| 574 |
+
"grad_norm": 0.7566428184509277,
|
| 575 |
+
"learning_rate": 8.706413651073846e-05,
|
| 576 |
+
"loss": 3.2417,
|
| 577 |
+
"step": 76000
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 4.14,
|
| 581 |
+
"grad_norm": 0.7315700650215149,
|
| 582 |
+
"learning_rate": 8.676993233303912e-05,
|
| 583 |
+
"loss": 3.2394,
|
| 584 |
+
"step": 77000
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 4.19,
|
| 588 |
+
"grad_norm": 0.7529075741767883,
|
| 589 |
+
"learning_rate": 8.64757281553398e-05,
|
| 590 |
+
"loss": 3.2415,
|
| 591 |
+
"step": 78000
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 4.25,
|
| 595 |
+
"grad_norm": 0.7442134022712708,
|
| 596 |
+
"learning_rate": 8.618152397764049e-05,
|
| 597 |
+
"loss": 3.2399,
|
| 598 |
+
"step": 79000
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 4.3,
|
| 602 |
+
"grad_norm": 0.7772579193115234,
|
| 603 |
+
"learning_rate": 8.588790820829656e-05,
|
| 604 |
+
"loss": 3.2352,
|
| 605 |
+
"step": 80000
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 4.36,
|
| 609 |
+
"grad_norm": 0.7373689413070679,
|
| 610 |
+
"learning_rate": 8.559370403059724e-05,
|
| 611 |
+
"loss": 3.2348,
|
| 612 |
+
"step": 81000
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 4.41,
|
| 616 |
+
"grad_norm": 0.7681490778923035,
|
| 617 |
+
"learning_rate": 8.529949985289791e-05,
|
| 618 |
+
"loss": 3.2282,
|
| 619 |
+
"step": 82000
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 4.46,
|
| 623 |
+
"grad_norm": 0.7327556014060974,
|
| 624 |
+
"learning_rate": 8.500529567519859e-05,
|
| 625 |
+
"loss": 3.2349,
|
| 626 |
+
"step": 83000
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 4.52,
|
| 630 |
+
"grad_norm": 0.7320055961608887,
|
| 631 |
+
"learning_rate": 8.471109149749927e-05,
|
| 632 |
+
"loss": 3.2342,
|
| 633 |
+
"step": 84000
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 4.57,
|
| 637 |
+
"grad_norm": 0.7526534199714661,
|
| 638 |
+
"learning_rate": 8.441718152397764e-05,
|
| 639 |
+
"loss": 3.2329,
|
| 640 |
+
"step": 85000
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 4.62,
|
| 644 |
+
"grad_norm": 0.7839016914367676,
|
| 645 |
+
"learning_rate": 8.412297734627832e-05,
|
| 646 |
+
"loss": 3.2371,
|
| 647 |
+
"step": 86000
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 4.68,
|
| 651 |
+
"grad_norm": 0.7079972624778748,
|
| 652 |
+
"learning_rate": 8.382906737275669e-05,
|
| 653 |
+
"loss": 3.2293,
|
| 654 |
+
"step": 87000
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 4.73,
|
| 658 |
+
"grad_norm": 0.7379385232925415,
|
| 659 |
+
"learning_rate": 8.353486319505737e-05,
|
| 660 |
+
"loss": 3.2315,
|
| 661 |
+
"step": 88000
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 4.79,
|
| 665 |
+
"grad_norm": 0.7435966730117798,
|
| 666 |
+
"learning_rate": 8.324065901735804e-05,
|
| 667 |
+
"loss": 3.2285,
|
| 668 |
+
"step": 89000
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 4.84,
|
| 672 |
+
"grad_norm": 0.7578746676445007,
|
| 673 |
+
"learning_rate": 8.294645483965872e-05,
|
| 674 |
+
"loss": 3.2233,
|
| 675 |
+
"step": 90000
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 4.89,
|
| 679 |
+
"grad_norm": 0.7228344678878784,
|
| 680 |
+
"learning_rate": 8.265254486613711e-05,
|
| 681 |
+
"loss": 3.2261,
|
| 682 |
+
"step": 91000
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 4.95,
|
| 686 |
+
"grad_norm": 0.7377974987030029,
|
| 687 |
+
"learning_rate": 8.235863489261548e-05,
|
| 688 |
+
"loss": 3.2283,
|
| 689 |
+
"step": 92000
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 5.0,
|
| 693 |
+
"eval_accuracy": 0.3929738045053096,
|
| 694 |
+
"eval_loss": 3.478940010070801,
|
| 695 |
+
"eval_runtime": 154.0106,
|
| 696 |
+
"eval_samples_per_second": 376.065,
|
| 697 |
+
"eval_steps_per_second": 5.876,
|
| 698 |
+
"step": 92975
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 5.0,
|
| 702 |
+
"grad_norm": 0.7944506406784058,
|
| 703 |
+
"learning_rate": 8.206443071491615e-05,
|
| 704 |
+
"loss": 3.2153,
|
| 705 |
+
"step": 93000
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 5.06,
|
| 709 |
+
"grad_norm": 0.752309262752533,
|
| 710 |
+
"learning_rate": 8.177022653721683e-05,
|
| 711 |
+
"loss": 3.16,
|
| 712 |
+
"step": 94000
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 5.11,
|
| 716 |
+
"grad_norm": 0.7849130630493164,
|
| 717 |
+
"learning_rate": 8.147631656369521e-05,
|
| 718 |
+
"loss": 3.172,
|
| 719 |
+
"step": 95000
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 5.16,
|
| 723 |
+
"grad_norm": 0.7549823522567749,
|
| 724 |
+
"learning_rate": 8.118211238599589e-05,
|
| 725 |
+
"loss": 3.1717,
|
| 726 |
+
"step": 96000
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 5.22,
|
| 730 |
+
"grad_norm": 0.761482834815979,
|
| 731 |
+
"learning_rate": 8.088790820829656e-05,
|
| 732 |
+
"loss": 3.1693,
|
| 733 |
+
"step": 97000
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 5.27,
|
| 737 |
+
"grad_norm": 0.7370353937149048,
|
| 738 |
+
"learning_rate": 8.059399823477494e-05,
|
| 739 |
+
"loss": 3.1749,
|
| 740 |
+
"step": 98000
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 5.32,
|
| 744 |
+
"grad_norm": 0.7203201651573181,
|
| 745 |
+
"learning_rate": 8.029979405707561e-05,
|
| 746 |
+
"loss": 3.1721,
|
| 747 |
+
"step": 99000
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 5.38,
|
| 751 |
+
"grad_norm": 0.7289775609970093,
|
| 752 |
+
"learning_rate": 8.0005884083554e-05,
|
| 753 |
+
"loss": 3.1809,
|
| 754 |
+
"step": 100000
|
| 755 |
+
},
|
| 756 |
+
{
|
| 757 |
+
"epoch": 5.43,
|
| 758 |
+
"grad_norm": 0.7284764051437378,
|
| 759 |
+
"learning_rate": 7.971167990585466e-05,
|
| 760 |
+
"loss": 3.1722,
|
| 761 |
+
"step": 101000
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 5.49,
|
| 765 |
+
"grad_norm": 0.7419421672821045,
|
| 766 |
+
"learning_rate": 7.941747572815534e-05,
|
| 767 |
+
"loss": 3.178,
|
| 768 |
+
"step": 102000
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 5.54,
|
| 772 |
+
"grad_norm": 0.7295576930046082,
|
| 773 |
+
"learning_rate": 7.912356575463373e-05,
|
| 774 |
+
"loss": 3.1777,
|
| 775 |
+
"step": 103000
|
| 776 |
+
},
|
| 777 |
+
{
|
| 778 |
+
"epoch": 5.59,
|
| 779 |
+
"grad_norm": 0.7482835054397583,
|
| 780 |
+
"learning_rate": 7.88293615769344e-05,
|
| 781 |
+
"loss": 3.178,
|
| 782 |
+
"step": 104000
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 5.65,
|
| 786 |
+
"grad_norm": 0.778439998626709,
|
| 787 |
+
"learning_rate": 7.853515739923508e-05,
|
| 788 |
+
"loss": 3.1735,
|
| 789 |
+
"step": 105000
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 5.7,
|
| 793 |
+
"grad_norm": 0.7204304337501526,
|
| 794 |
+
"learning_rate": 7.824124742571345e-05,
|
| 795 |
+
"loss": 3.1719,
|
| 796 |
+
"step": 106000
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"epoch": 5.75,
|
| 800 |
+
"grad_norm": 0.7353498339653015,
|
| 801 |
+
"learning_rate": 7.794704324801413e-05,
|
| 802 |
+
"loss": 3.1761,
|
| 803 |
+
"step": 107000
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 5.81,
|
| 807 |
+
"grad_norm": 0.7248005867004395,
|
| 808 |
+
"learning_rate": 7.76528390703148e-05,
|
| 809 |
+
"loss": 3.1752,
|
| 810 |
+
"step": 108000
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 5.86,
|
| 814 |
+
"grad_norm": 0.7356109619140625,
|
| 815 |
+
"learning_rate": 7.735863489261548e-05,
|
| 816 |
+
"loss": 3.1754,
|
| 817 |
+
"step": 109000
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 5.92,
|
| 821 |
+
"grad_norm": 0.708113968372345,
|
| 822 |
+
"learning_rate": 7.706472491909386e-05,
|
| 823 |
+
"loss": 3.1744,
|
| 824 |
+
"step": 110000
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 5.97,
|
| 828 |
+
"grad_norm": 0.7147455215454102,
|
| 829 |
+
"learning_rate": 7.677052074139453e-05,
|
| 830 |
+
"loss": 3.1673,
|
| 831 |
+
"step": 111000
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 6.0,
|
| 835 |
+
"eval_accuracy": 0.3961948483989776,
|
| 836 |
+
"eval_loss": 3.4379465579986572,
|
| 837 |
+
"eval_runtime": 153.8126,
|
| 838 |
+
"eval_samples_per_second": 376.549,
|
| 839 |
+
"eval_steps_per_second": 5.884,
|
| 840 |
+
"step": 111570
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 6.02,
|
| 844 |
+
"grad_norm": 0.7331594824790955,
|
| 845 |
+
"learning_rate": 7.64766107678729e-05,
|
| 846 |
+
"loss": 3.1401,
|
| 847 |
+
"step": 112000
|
| 848 |
+
},
|
| 849 |
+
{
|
| 850 |
+
"epoch": 6.08,
|
| 851 |
+
"grad_norm": 0.7715177536010742,
|
| 852 |
+
"learning_rate": 7.618240659017358e-05,
|
| 853 |
+
"loss": 3.1162,
|
| 854 |
+
"step": 113000
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 6.13,
|
| 858 |
+
"grad_norm": 0.7466005682945251,
|
| 859 |
+
"learning_rate": 7.588820241247426e-05,
|
| 860 |
+
"loss": 3.1136,
|
| 861 |
+
"step": 114000
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 6.18,
|
| 865 |
+
"grad_norm": 0.734748899936676,
|
| 866 |
+
"learning_rate": 7.559399823477493e-05,
|
| 867 |
+
"loss": 3.1227,
|
| 868 |
+
"step": 115000
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 6.24,
|
| 872 |
+
"grad_norm": 0.7688254714012146,
|
| 873 |
+
"learning_rate": 7.530008826125331e-05,
|
| 874 |
+
"loss": 3.1252,
|
| 875 |
+
"step": 116000
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 6.29,
|
| 879 |
+
"grad_norm": 0.7620559930801392,
|
| 880 |
+
"learning_rate": 7.5005884083554e-05,
|
| 881 |
+
"loss": 3.1243,
|
| 882 |
+
"step": 117000
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 6.35,
|
| 886 |
+
"grad_norm": 0.7437037825584412,
|
| 887 |
+
"learning_rate": 7.471167990585468e-05,
|
| 888 |
+
"loss": 3.1286,
|
| 889 |
+
"step": 118000
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 6.4,
|
| 893 |
+
"grad_norm": 0.7445530295372009,
|
| 894 |
+
"learning_rate": 7.441747572815534e-05,
|
| 895 |
+
"loss": 3.1279,
|
| 896 |
+
"step": 119000
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 6.45,
|
| 900 |
+
"grad_norm": 0.7443315386772156,
|
| 901 |
+
"learning_rate": 7.412356575463371e-05,
|
| 902 |
+
"loss": 3.1314,
|
| 903 |
+
"step": 120000
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 6.51,
|
| 907 |
+
"grad_norm": 0.7500354647636414,
|
| 908 |
+
"learning_rate": 7.38293615769344e-05,
|
| 909 |
+
"loss": 3.1288,
|
| 910 |
+
"step": 121000
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 6.56,
|
| 914 |
+
"grad_norm": 0.7664729952812195,
|
| 915 |
+
"learning_rate": 7.353545160341278e-05,
|
| 916 |
+
"loss": 3.1307,
|
| 917 |
+
"step": 122000
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 6.61,
|
| 921 |
+
"grad_norm": 0.7711361646652222,
|
| 922 |
+
"learning_rate": 7.324154162989115e-05,
|
| 923 |
+
"loss": 3.132,
|
| 924 |
+
"step": 123000
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 6.67,
|
| 928 |
+
"grad_norm": 0.7488456964492798,
|
| 929 |
+
"learning_rate": 7.294733745219182e-05,
|
| 930 |
+
"loss": 3.1303,
|
| 931 |
+
"step": 124000
|
| 932 |
+
},
|
| 933 |
+
{
|
| 934 |
+
"epoch": 6.72,
|
| 935 |
+
"grad_norm": 0.7189005017280579,
|
| 936 |
+
"learning_rate": 7.26531332744925e-05,
|
| 937 |
+
"loss": 3.1285,
|
| 938 |
+
"step": 125000
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 6.78,
|
| 942 |
+
"grad_norm": 0.7578772902488708,
|
| 943 |
+
"learning_rate": 7.235922330097088e-05,
|
| 944 |
+
"loss": 3.1281,
|
| 945 |
+
"step": 126000
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 6.83,
|
| 949 |
+
"grad_norm": 0.7351284027099609,
|
| 950 |
+
"learning_rate": 7.206501912327155e-05,
|
| 951 |
+
"loss": 3.1305,
|
| 952 |
+
"step": 127000
|
| 953 |
+
},
|
| 954 |
+
{
|
| 955 |
+
"epoch": 6.88,
|
| 956 |
+
"grad_norm": 0.7461217641830444,
|
| 957 |
+
"learning_rate": 7.177081494557223e-05,
|
| 958 |
+
"loss": 3.1327,
|
| 959 |
+
"step": 128000
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 6.94,
|
| 963 |
+
"grad_norm": 0.7394551038742065,
|
| 964 |
+
"learning_rate": 7.14769049720506e-05,
|
| 965 |
+
"loss": 3.133,
|
| 966 |
+
"step": 129000
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 6.99,
|
| 970 |
+
"grad_norm": 0.7317260503768921,
|
| 971 |
+
"learning_rate": 7.118270079435128e-05,
|
| 972 |
+
"loss": 3.127,
|
| 973 |
+
"step": 130000
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 7.0,
|
| 977 |
+
"eval_accuracy": 0.39883100527338866,
|
| 978 |
+
"eval_loss": 3.41751766204834,
|
| 979 |
+
"eval_runtime": 153.2748,
|
| 980 |
+
"eval_samples_per_second": 377.87,
|
| 981 |
+
"eval_steps_per_second": 5.904,
|
| 982 |
+
"step": 130165
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 7.04,
|
| 986 |
+
"grad_norm": 0.7801479697227478,
|
| 987 |
+
"learning_rate": 7.088879082082967e-05,
|
| 988 |
+
"loss": 3.0813,
|
| 989 |
+
"step": 131000
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 7.1,
|
| 993 |
+
"grad_norm": 0.7538677453994751,
|
| 994 |
+
"learning_rate": 7.059458664313034e-05,
|
| 995 |
+
"loss": 3.077,
|
| 996 |
+
"step": 132000
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 7.15,
|
| 1000 |
+
"grad_norm": 0.7728538513183594,
|
| 1001 |
+
"learning_rate": 7.030038246543102e-05,
|
| 1002 |
+
"loss": 3.0807,
|
| 1003 |
+
"step": 133000
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 7.21,
|
| 1007 |
+
"grad_norm": 0.7852578163146973,
|
| 1008 |
+
"learning_rate": 7.00061782877317e-05,
|
| 1009 |
+
"loss": 3.0811,
|
| 1010 |
+
"step": 134000
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 7.26,
|
| 1014 |
+
"grad_norm": 0.7735338807106018,
|
| 1015 |
+
"learning_rate": 6.971226831421006e-05,
|
| 1016 |
+
"loss": 3.0861,
|
| 1017 |
+
"step": 135000
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 7.31,
|
| 1021 |
+
"grad_norm": 0.7543830871582031,
|
| 1022 |
+
"learning_rate": 6.941835834068844e-05,
|
| 1023 |
+
"loss": 3.0846,
|
| 1024 |
+
"step": 136000
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 7.37,
|
| 1028 |
+
"grad_norm": 0.7399442195892334,
|
| 1029 |
+
"learning_rate": 6.912415416298912e-05,
|
| 1030 |
+
"loss": 3.0866,
|
| 1031 |
+
"step": 137000
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 7.42,
|
| 1035 |
+
"grad_norm": 0.7640860676765442,
|
| 1036 |
+
"learning_rate": 6.88299499852898e-05,
|
| 1037 |
+
"loss": 3.0893,
|
| 1038 |
+
"step": 138000
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 7.48,
|
| 1042 |
+
"grad_norm": 0.7997984886169434,
|
| 1043 |
+
"learning_rate": 6.853574580759047e-05,
|
| 1044 |
+
"loss": 3.0909,
|
| 1045 |
+
"step": 139000
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 7.53,
|
| 1049 |
+
"grad_norm": 0.7792380452156067,
|
| 1050 |
+
"learning_rate": 6.824183583406884e-05,
|
| 1051 |
+
"loss": 3.0939,
|
| 1052 |
+
"step": 140000
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 7.58,
|
| 1056 |
+
"grad_norm": 0.7403313517570496,
|
| 1057 |
+
"learning_rate": 6.794763165636952e-05,
|
| 1058 |
+
"loss": 3.0906,
|
| 1059 |
+
"step": 141000
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 7.64,
|
| 1063 |
+
"grad_norm": 0.7356191873550415,
|
| 1064 |
+
"learning_rate": 6.765342747867019e-05,
|
| 1065 |
+
"loss": 3.0909,
|
| 1066 |
+
"step": 142000
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 7.69,
|
| 1070 |
+
"grad_norm": 0.7950366735458374,
|
| 1071 |
+
"learning_rate": 6.735951750514857e-05,
|
| 1072 |
+
"loss": 3.0965,
|
| 1073 |
+
"step": 143000
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 7.74,
|
| 1077 |
+
"grad_norm": 0.7580872774124146,
|
| 1078 |
+
"learning_rate": 6.706531332744925e-05,
|
| 1079 |
+
"loss": 3.0922,
|
| 1080 |
+
"step": 144000
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 7.8,
|
| 1084 |
+
"grad_norm": 0.7285676002502441,
|
| 1085 |
+
"learning_rate": 6.677140335392763e-05,
|
| 1086 |
+
"loss": 3.0952,
|
| 1087 |
+
"step": 145000
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 7.85,
|
| 1091 |
+
"grad_norm": 0.730719268321991,
|
| 1092 |
+
"learning_rate": 6.64771991762283e-05,
|
| 1093 |
+
"loss": 3.0943,
|
| 1094 |
+
"step": 146000
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 7.91,
|
| 1098 |
+
"grad_norm": 0.7722078561782837,
|
| 1099 |
+
"learning_rate": 6.618299499852897e-05,
|
| 1100 |
+
"loss": 3.0957,
|
| 1101 |
+
"step": 147000
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 7.96,
|
| 1105 |
+
"grad_norm": 0.7444345951080322,
|
| 1106 |
+
"learning_rate": 6.588879082082966e-05,
|
| 1107 |
+
"loss": 3.0935,
|
| 1108 |
+
"step": 148000
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 8.0,
|
| 1112 |
+
"eval_accuracy": 0.39977229312392953,
|
| 1113 |
+
"eval_loss": 3.4271562099456787,
|
| 1114 |
+
"eval_runtime": 154.0268,
|
| 1115 |
+
"eval_samples_per_second": 376.025,
|
| 1116 |
+
"eval_steps_per_second": 5.876,
|
| 1117 |
+
"step": 148760
|
| 1118 |
+
},
|
| 1119 |
+
{
|
| 1120 |
+
"epoch": 8.01,
|
| 1121 |
+
"grad_norm": 0.7648381590843201,
|
| 1122 |
+
"learning_rate": 6.559488084730804e-05,
|
| 1123 |
+
"loss": 3.0798,
|
| 1124 |
+
"step": 149000
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 8.07,
|
| 1128 |
+
"grad_norm": 0.746784508228302,
|
| 1129 |
+
"learning_rate": 6.530067666960871e-05,
|
| 1130 |
+
"loss": 3.0327,
|
| 1131 |
+
"step": 150000
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 8.12,
|
| 1135 |
+
"grad_norm": 0.7783890962600708,
|
| 1136 |
+
"learning_rate": 6.500647249190939e-05,
|
| 1137 |
+
"loss": 3.0398,
|
| 1138 |
+
"step": 151000
|
| 1139 |
+
},
|
| 1140 |
+
{
|
| 1141 |
+
"epoch": 8.17,
|
| 1142 |
+
"grad_norm": 0.7425867915153503,
|
| 1143 |
+
"learning_rate": 6.471226831421007e-05,
|
| 1144 |
+
"loss": 3.0455,
|
| 1145 |
+
"step": 152000
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 8.23,
|
| 1149 |
+
"grad_norm": 0.7634230852127075,
|
| 1150 |
+
"learning_rate": 6.441835834068844e-05,
|
| 1151 |
+
"loss": 3.0529,
|
| 1152 |
+
"step": 153000
|
| 1153 |
+
},
|
| 1154 |
+
{
|
| 1155 |
+
"epoch": 8.28,
|
| 1156 |
+
"grad_norm": 0.747742235660553,
|
| 1157 |
+
"learning_rate": 6.412415416298911e-05,
|
| 1158 |
+
"loss": 3.0487,
|
| 1159 |
+
"step": 154000
|
| 1160 |
+
},
|
| 1161 |
+
{
|
| 1162 |
+
"epoch": 8.34,
|
| 1163 |
+
"grad_norm": 0.7804195284843445,
|
| 1164 |
+
"learning_rate": 6.383024418946749e-05,
|
| 1165 |
+
"loss": 3.0545,
|
| 1166 |
+
"step": 155000
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 8.39,
|
| 1170 |
+
"grad_norm": 0.7713324427604675,
|
| 1171 |
+
"learning_rate": 6.353604001176817e-05,
|
| 1172 |
+
"loss": 3.06,
|
| 1173 |
+
"step": 156000
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 8.44,
|
| 1177 |
+
"grad_norm": 0.7601629495620728,
|
| 1178 |
+
"learning_rate": 6.324183583406885e-05,
|
| 1179 |
+
"loss": 3.0537,
|
| 1180 |
+
"step": 157000
|
| 1181 |
+
},
|
| 1182 |
+
{
|
| 1183 |
+
"epoch": 8.5,
|
| 1184 |
+
"grad_norm": 0.7745895385742188,
|
| 1185 |
+
"learning_rate": 6.294792586054723e-05,
|
| 1186 |
+
"loss": 3.0605,
|
| 1187 |
+
"step": 158000
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 8.55,
|
| 1191 |
+
"grad_norm": 0.7613347172737122,
|
| 1192 |
+
"learning_rate": 6.265372168284789e-05,
|
| 1193 |
+
"loss": 3.0573,
|
| 1194 |
+
"step": 159000
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 8.6,
|
| 1198 |
+
"grad_norm": 0.7753309607505798,
|
| 1199 |
+
"learning_rate": 6.235981170932628e-05,
|
| 1200 |
+
"loss": 3.0627,
|
| 1201 |
+
"step": 160000
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 8.66,
|
| 1205 |
+
"grad_norm": 0.7721776962280273,
|
| 1206 |
+
"learning_rate": 6.206560753162696e-05,
|
| 1207 |
+
"loss": 3.0612,
|
| 1208 |
+
"step": 161000
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 8.71,
|
| 1212 |
+
"grad_norm": 0.8062002062797546,
|
| 1213 |
+
"learning_rate": 6.177140335392763e-05,
|
| 1214 |
+
"loss": 3.0622,
|
| 1215 |
+
"step": 162000
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 8.77,
|
| 1219 |
+
"grad_norm": 0.7599053978919983,
|
| 1220 |
+
"learning_rate": 6.147719917622831e-05,
|
| 1221 |
+
"loss": 3.061,
|
| 1222 |
+
"step": 163000
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"epoch": 8.82,
|
| 1226 |
+
"grad_norm": 0.7855945229530334,
|
| 1227 |
+
"learning_rate": 6.118328920270668e-05,
|
| 1228 |
+
"loss": 3.0638,
|
| 1229 |
+
"step": 164000
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 8.87,
|
| 1233 |
+
"grad_norm": 0.7659214735031128,
|
| 1234 |
+
"learning_rate": 6.088908502500735e-05,
|
| 1235 |
+
"loss": 3.0637,
|
| 1236 |
+
"step": 165000
|
| 1237 |
+
},
|
| 1238 |
+
{
|
| 1239 |
+
"epoch": 8.93,
|
| 1240 |
+
"grad_norm": 0.7432619333267212,
|
| 1241 |
+
"learning_rate": 6.059517505148573e-05,
|
| 1242 |
+
"loss": 3.0661,
|
| 1243 |
+
"step": 166000
|
| 1244 |
+
},
|
| 1245 |
+
{
|
| 1246 |
+
"epoch": 8.98,
|
| 1247 |
+
"grad_norm": 0.7428516745567322,
|
| 1248 |
+
"learning_rate": 6.030097087378641e-05,
|
| 1249 |
+
"loss": 3.0679,
|
| 1250 |
+
"step": 167000
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 9.0,
|
| 1254 |
+
"eval_accuracy": 0.4010507677364335,
|
| 1255 |
+
"eval_loss": 3.4149951934814453,
|
| 1256 |
+
"eval_runtime": 153.7947,
|
| 1257 |
+
"eval_samples_per_second": 376.593,
|
| 1258 |
+
"eval_steps_per_second": 5.884,
|
| 1259 |
+
"step": 167355
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 9.03,
|
| 1263 |
+
"grad_norm": 0.7959059476852417,
|
| 1264 |
+
"learning_rate": 6.000676669608709e-05,
|
| 1265 |
+
"loss": 3.0238,
|
| 1266 |
+
"step": 168000
|
| 1267 |
+
},
|
| 1268 |
+
{
|
| 1269 |
+
"epoch": 9.09,
|
| 1270 |
+
"grad_norm": 0.7902527451515198,
|
| 1271 |
+
"learning_rate": 5.971256251838776e-05,
|
| 1272 |
+
"loss": 3.007,
|
| 1273 |
+
"step": 169000
|
| 1274 |
+
},
|
| 1275 |
+
{
|
| 1276 |
+
"epoch": 9.14,
|
| 1277 |
+
"grad_norm": 0.7812909483909607,
|
| 1278 |
+
"learning_rate": 5.941865254486614e-05,
|
| 1279 |
+
"loss": 3.0162,
|
| 1280 |
+
"step": 170000
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 9.2,
|
| 1284 |
+
"grad_norm": 0.7669194936752319,
|
| 1285 |
+
"learning_rate": 5.9124742571344514e-05,
|
| 1286 |
+
"loss": 3.0162,
|
| 1287 |
+
"step": 171000
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 9.25,
|
| 1291 |
+
"grad_norm": 0.7774065136909485,
|
| 1292 |
+
"learning_rate": 5.8830538393645195e-05,
|
| 1293 |
+
"loss": 3.0219,
|
| 1294 |
+
"step": 172000
|
| 1295 |
+
},
|
| 1296 |
+
{
|
| 1297 |
+
"epoch": 9.3,
|
| 1298 |
+
"grad_norm": 0.7773292064666748,
|
| 1299 |
+
"learning_rate": 5.853662842012357e-05,
|
| 1300 |
+
"loss": 3.0238,
|
| 1301 |
+
"step": 173000
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 9.36,
|
| 1305 |
+
"grad_norm": 0.7975197434425354,
|
| 1306 |
+
"learning_rate": 5.824242424242424e-05,
|
| 1307 |
+
"loss": 3.0246,
|
| 1308 |
+
"step": 174000
|
| 1309 |
+
},
|
| 1310 |
+
{
|
| 1311 |
+
"epoch": 9.41,
|
| 1312 |
+
"grad_norm": 0.7579612731933594,
|
| 1313 |
+
"learning_rate": 5.794822006472492e-05,
|
| 1314 |
+
"loss": 3.0259,
|
| 1315 |
+
"step": 175000
|
| 1316 |
+
},
|
| 1317 |
+
{
|
| 1318 |
+
"epoch": 9.46,
|
| 1319 |
+
"grad_norm": 0.7753346562385559,
|
| 1320 |
+
"learning_rate": 5.76540158870256e-05,
|
| 1321 |
+
"loss": 3.0287,
|
| 1322 |
+
"step": 176000
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 9.52,
|
| 1326 |
+
"grad_norm": 0.7583608031272888,
|
| 1327 |
+
"learning_rate": 5.7360105913503966e-05,
|
| 1328 |
+
"loss": 3.0281,
|
| 1329 |
+
"step": 177000
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 9.57,
|
| 1333 |
+
"grad_norm": 0.7782021760940552,
|
| 1334 |
+
"learning_rate": 5.706590173580465e-05,
|
| 1335 |
+
"loss": 3.0298,
|
| 1336 |
+
"step": 178000
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"epoch": 9.63,
|
| 1340 |
+
"grad_norm": 0.7898275852203369,
|
| 1341 |
+
"learning_rate": 5.677169755810533e-05,
|
| 1342 |
+
"loss": 3.0384,
|
| 1343 |
+
"step": 179000
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 9.68,
|
| 1347 |
+
"grad_norm": 0.7844170928001404,
|
| 1348 |
+
"learning_rate": 5.6477787584583706e-05,
|
| 1349 |
+
"loss": 3.0341,
|
| 1350 |
+
"step": 180000
|
| 1351 |
+
},
|
| 1352 |
+
{
|
| 1353 |
+
"epoch": 9.73,
|
| 1354 |
+
"grad_norm": 0.7562811374664307,
|
| 1355 |
+
"learning_rate": 5.618387761106208e-05,
|
| 1356 |
+
"loss": 3.0346,
|
| 1357 |
+
"step": 181000
|
| 1358 |
+
},
|
| 1359 |
+
{
|
| 1360 |
+
"epoch": 9.79,
|
| 1361 |
+
"grad_norm": 0.7964505553245544,
|
| 1362 |
+
"learning_rate": 5.588967343336276e-05,
|
| 1363 |
+
"loss": 3.0364,
|
| 1364 |
+
"step": 182000
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 9.84,
|
| 1368 |
+
"grad_norm": 0.7780364751815796,
|
| 1369 |
+
"learning_rate": 5.5595763459841135e-05,
|
| 1370 |
+
"loss": 3.0376,
|
| 1371 |
+
"step": 183000
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 9.9,
|
| 1375 |
+
"grad_norm": 0.7689926028251648,
|
| 1376 |
+
"learning_rate": 5.530155928214181e-05,
|
| 1377 |
+
"loss": 3.0343,
|
| 1378 |
+
"step": 184000
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 9.95,
|
| 1382 |
+
"grad_norm": 0.7878050804138184,
|
| 1383 |
+
"learning_rate": 5.5007355104442484e-05,
|
| 1384 |
+
"loss": 3.0402,
|
| 1385 |
+
"step": 185000
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 10.0,
|
| 1389 |
+
"eval_accuracy": 0.4023324670878444,
|
| 1390 |
+
"eval_loss": 3.414785146713257,
|
| 1391 |
+
"eval_runtime": 153.4202,
|
| 1392 |
+
"eval_samples_per_second": 377.512,
|
| 1393 |
+
"eval_steps_per_second": 5.899,
|
| 1394 |
+
"step": 185950
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 10.0,
|
| 1398 |
+
"grad_norm": 0.7869421243667603,
|
| 1399 |
+
"learning_rate": 5.471344513092086e-05,
|
| 1400 |
+
"loss": 3.0355,
|
| 1401 |
+
"step": 186000
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 10.06,
|
| 1405 |
+
"grad_norm": 0.7897782325744629,
|
| 1406 |
+
"learning_rate": 5.441924095322154e-05,
|
| 1407 |
+
"loss": 2.9791,
|
| 1408 |
+
"step": 187000
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"epoch": 10.11,
|
| 1412 |
+
"grad_norm": 0.810604989528656,
|
| 1413 |
+
"learning_rate": 5.412503677552222e-05,
|
| 1414 |
+
"loss": 2.9856,
|
| 1415 |
+
"step": 188000
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 10.16,
|
| 1419 |
+
"grad_norm": 0.802144467830658,
|
| 1420 |
+
"learning_rate": 5.383112680200059e-05,
|
| 1421 |
+
"loss": 2.9921,
|
| 1422 |
+
"step": 189000
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 10.22,
|
| 1426 |
+
"grad_norm": 0.8142128586769104,
|
| 1427 |
+
"learning_rate": 5.353692262430127e-05,
|
| 1428 |
+
"loss": 2.9952,
|
| 1429 |
+
"step": 190000
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 10.27,
|
| 1433 |
+
"grad_norm": 0.792819082736969,
|
| 1434 |
+
"learning_rate": 5.3243012650779646e-05,
|
| 1435 |
+
"loss": 2.9997,
|
| 1436 |
+
"step": 191000
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 10.33,
|
| 1440 |
+
"grad_norm": 0.7813856601715088,
|
| 1441 |
+
"learning_rate": 5.294880847308033e-05,
|
| 1442 |
+
"loss": 3.0017,
|
| 1443 |
+
"step": 192000
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 10.38,
|
| 1447 |
+
"grad_norm": 0.8088542222976685,
|
| 1448 |
+
"learning_rate": 5.2654604295380995e-05,
|
| 1449 |
+
"loss": 3.0005,
|
| 1450 |
+
"step": 193000
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 10.43,
|
| 1454 |
+
"grad_norm": 0.7641571164131165,
|
| 1455 |
+
"learning_rate": 5.2360400117681677e-05,
|
| 1456 |
+
"loss": 3.0007,
|
| 1457 |
+
"step": 194000
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 10.49,
|
| 1461 |
+
"grad_norm": 0.7936010956764221,
|
| 1462 |
+
"learning_rate": 5.2066490144160054e-05,
|
| 1463 |
+
"loss": 3.0038,
|
| 1464 |
+
"step": 195000
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 10.54,
|
| 1468 |
+
"grad_norm": 0.8503419756889343,
|
| 1469 |
+
"learning_rate": 5.177258017063843e-05,
|
| 1470 |
+
"loss": 3.0049,
|
| 1471 |
+
"step": 196000
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 10.59,
|
| 1475 |
+
"grad_norm": 0.7761277556419373,
|
| 1476 |
+
"learning_rate": 5.14783759929391e-05,
|
| 1477 |
+
"loss": 3.0079,
|
| 1478 |
+
"step": 197000
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 10.65,
|
| 1482 |
+
"grad_norm": 0.7853693962097168,
|
| 1483 |
+
"learning_rate": 5.118417181523978e-05,
|
| 1484 |
+
"loss": 3.0091,
|
| 1485 |
+
"step": 198000
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 10.7,
|
| 1489 |
+
"grad_norm": 0.7898029685020447,
|
| 1490 |
+
"learning_rate": 5.089026184171816e-05,
|
| 1491 |
+
"loss": 3.0047,
|
| 1492 |
+
"step": 199000
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"epoch": 10.76,
|
| 1496 |
+
"grad_norm": 0.7760916948318481,
|
| 1497 |
+
"learning_rate": 5.0596351868196535e-05,
|
| 1498 |
+
"loss": 3.01,
|
| 1499 |
+
"step": 200000
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 10.81,
|
| 1503 |
+
"grad_norm": 0.8151899576187134,
|
| 1504 |
+
"learning_rate": 5.03021476904972e-05,
|
| 1505 |
+
"loss": 3.0079,
|
| 1506 |
+
"step": 201000
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 10.86,
|
| 1510 |
+
"grad_norm": 0.837149977684021,
|
| 1511 |
+
"learning_rate": 5.0007943512797884e-05,
|
| 1512 |
+
"loss": 3.0115,
|
| 1513 |
+
"step": 202000
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 10.92,
|
| 1517 |
+
"grad_norm": 0.7857736945152283,
|
| 1518 |
+
"learning_rate": 4.971373933509856e-05,
|
| 1519 |
+
"loss": 3.0123,
|
| 1520 |
+
"step": 203000
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 10.97,
|
| 1524 |
+
"grad_norm": 0.7611953020095825,
|
| 1525 |
+
"learning_rate": 4.9419829361576935e-05,
|
| 1526 |
+
"loss": 3.0082,
|
| 1527 |
+
"step": 204000
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 11.0,
|
| 1531 |
+
"eval_accuracy": 0.4029490640396869,
|
| 1532 |
+
"eval_loss": 3.4169130325317383,
|
| 1533 |
+
"eval_runtime": 153.2931,
|
| 1534 |
+
"eval_samples_per_second": 377.825,
|
| 1535 |
+
"eval_steps_per_second": 5.904,
|
| 1536 |
+
"step": 204545
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 11.02,
|
| 1540 |
+
"grad_norm": 0.816586971282959,
|
| 1541 |
+
"learning_rate": 4.9125625183877617e-05,
|
| 1542 |
+
"loss": 2.9829,
|
| 1543 |
+
"step": 205000
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 11.08,
|
| 1547 |
+
"grad_norm": 0.7916204929351807,
|
| 1548 |
+
"learning_rate": 4.883171521035599e-05,
|
| 1549 |
+
"loss": 2.9574,
|
| 1550 |
+
"step": 206000
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 11.13,
|
| 1554 |
+
"grad_norm": 0.8013209104537964,
|
| 1555 |
+
"learning_rate": 4.853751103265667e-05,
|
| 1556 |
+
"loss": 2.9652,
|
| 1557 |
+
"step": 207000
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 11.19,
|
| 1561 |
+
"grad_norm": 0.7968335747718811,
|
| 1562 |
+
"learning_rate": 4.824330685495734e-05,
|
| 1563 |
+
"loss": 2.9686,
|
| 1564 |
+
"step": 208000
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 11.24,
|
| 1568 |
+
"grad_norm": 0.8586769104003906,
|
| 1569 |
+
"learning_rate": 4.794910267725802e-05,
|
| 1570 |
+
"loss": 2.9722,
|
| 1571 |
+
"step": 209000
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 11.29,
|
| 1575 |
+
"grad_norm": 0.8471382260322571,
|
| 1576 |
+
"learning_rate": 4.7655192703736395e-05,
|
| 1577 |
+
"loss": 2.9747,
|
| 1578 |
+
"step": 210000
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 11.35,
|
| 1582 |
+
"grad_norm": 0.8427879810333252,
|
| 1583 |
+
"learning_rate": 4.736098852603707e-05,
|
| 1584 |
+
"loss": 2.9741,
|
| 1585 |
+
"step": 211000
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 11.4,
|
| 1589 |
+
"grad_norm": 0.8404529094696045,
|
| 1590 |
+
"learning_rate": 4.706678434833775e-05,
|
| 1591 |
+
"loss": 2.9773,
|
| 1592 |
+
"step": 212000
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 11.45,
|
| 1596 |
+
"grad_norm": 0.8058509826660156,
|
| 1597 |
+
"learning_rate": 4.677287437481612e-05,
|
| 1598 |
+
"loss": 2.9749,
|
| 1599 |
+
"step": 213000
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 11.51,
|
| 1603 |
+
"grad_norm": 0.8321877717971802,
|
| 1604 |
+
"learning_rate": 4.64786701971168e-05,
|
| 1605 |
+
"loss": 2.9794,
|
| 1606 |
+
"step": 214000
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 11.56,
|
| 1610 |
+
"grad_norm": 0.8273854851722717,
|
| 1611 |
+
"learning_rate": 4.618476022359517e-05,
|
| 1612 |
+
"loss": 2.9855,
|
| 1613 |
+
"step": 215000
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 11.62,
|
| 1617 |
+
"grad_norm": 0.8307694792747498,
|
| 1618 |
+
"learning_rate": 4.589085025007355e-05,
|
| 1619 |
+
"loss": 2.9862,
|
| 1620 |
+
"step": 216000
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 11.67,
|
| 1624 |
+
"grad_norm": 0.8184703588485718,
|
| 1625 |
+
"learning_rate": 4.559694027655193e-05,
|
| 1626 |
+
"loss": 2.9836,
|
| 1627 |
+
"step": 217000
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 11.72,
|
| 1631 |
+
"grad_norm": 0.7970172762870789,
|
| 1632 |
+
"learning_rate": 4.530273609885261e-05,
|
| 1633 |
+
"loss": 2.9843,
|
| 1634 |
+
"step": 218000
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 11.78,
|
| 1638 |
+
"grad_norm": 0.8030002117156982,
|
| 1639 |
+
"learning_rate": 4.500853192115328e-05,
|
| 1640 |
+
"loss": 2.9891,
|
| 1641 |
+
"step": 219000
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 11.83,
|
| 1645 |
+
"grad_norm": 0.795408308506012,
|
| 1646 |
+
"learning_rate": 4.471432774345396e-05,
|
| 1647 |
+
"loss": 2.987,
|
| 1648 |
+
"step": 220000
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 11.88,
|
| 1652 |
+
"grad_norm": 0.8084612488746643,
|
| 1653 |
+
"learning_rate": 4.4420417769932335e-05,
|
| 1654 |
+
"loss": 2.9898,
|
| 1655 |
+
"step": 221000
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 11.94,
|
| 1659 |
+
"grad_norm": 0.8061679005622864,
|
| 1660 |
+
"learning_rate": 4.412650779641071e-05,
|
| 1661 |
+
"loss": 2.9941,
|
| 1662 |
+
"step": 222000
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 11.99,
|
| 1666 |
+
"grad_norm": 0.8087071776390076,
|
| 1667 |
+
"learning_rate": 4.383230361871139e-05,
|
| 1668 |
+
"loss": 2.988,
|
| 1669 |
+
"step": 223000
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 12.0,
|
| 1673 |
+
"eval_accuracy": 0.4035294498608861,
|
| 1674 |
+
"eval_loss": 3.412954568862915,
|
| 1675 |
+
"eval_runtime": 153.0197,
|
| 1676 |
+
"eval_samples_per_second": 378.5,
|
| 1677 |
+
"eval_steps_per_second": 5.914,
|
| 1678 |
+
"step": 223140
|
| 1679 |
+
},
|
| 1680 |
+
{
|
| 1681 |
+
"epoch": 12.05,
|
| 1682 |
+
"grad_norm": 0.850972056388855,
|
| 1683 |
+
"learning_rate": 4.353809944101207e-05,
|
| 1684 |
+
"loss": 2.9439,
|
| 1685 |
+
"step": 224000
|
| 1686 |
+
},
|
| 1687 |
+
{
|
| 1688 |
+
"epoch": 12.1,
|
| 1689 |
+
"grad_norm": 0.8421295285224915,
|
| 1690 |
+
"learning_rate": 4.324389526331274e-05,
|
| 1691 |
+
"loss": 2.9402,
|
| 1692 |
+
"step": 225000
|
| 1693 |
+
},
|
| 1694 |
+
{
|
| 1695 |
+
"epoch": 12.15,
|
| 1696 |
+
"grad_norm": 0.8112582564353943,
|
| 1697 |
+
"learning_rate": 4.294998528979112e-05,
|
| 1698 |
+
"loss": 2.9473,
|
| 1699 |
+
"step": 226000
|
| 1700 |
+
},
|
| 1701 |
+
{
|
| 1702 |
+
"epoch": 12.21,
|
| 1703 |
+
"grad_norm": 0.8302417397499084,
|
| 1704 |
+
"learning_rate": 4.2655781112091794e-05,
|
| 1705 |
+
"loss": 2.9491,
|
| 1706 |
+
"step": 227000
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 12.26,
|
| 1710 |
+
"grad_norm": 0.8474385738372803,
|
| 1711 |
+
"learning_rate": 4.2361576934392475e-05,
|
| 1712 |
+
"loss": 2.9484,
|
| 1713 |
+
"step": 228000
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 12.32,
|
| 1717 |
+
"grad_norm": 0.8351003527641296,
|
| 1718 |
+
"learning_rate": 4.206737275669315e-05,
|
| 1719 |
+
"loss": 2.9529,
|
| 1720 |
+
"step": 229000
|
| 1721 |
+
},
|
| 1722 |
+
{
|
| 1723 |
+
"epoch": 12.37,
|
| 1724 |
+
"grad_norm": 0.85118168592453,
|
| 1725 |
+
"learning_rate": 4.177346278317153e-05,
|
| 1726 |
+
"loss": 2.9562,
|
| 1727 |
+
"step": 230000
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 12.42,
|
| 1731 |
+
"grad_norm": 0.8224354982376099,
|
| 1732 |
+
"learning_rate": 4.14792586054722e-05,
|
| 1733 |
+
"loss": 2.9539,
|
| 1734 |
+
"step": 231000
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 12.48,
|
| 1738 |
+
"grad_norm": 0.8389358520507812,
|
| 1739 |
+
"learning_rate": 4.1185054427772876e-05,
|
| 1740 |
+
"loss": 2.9581,
|
| 1741 |
+
"step": 232000
|
| 1742 |
+
},
|
| 1743 |
+
{
|
| 1744 |
+
"epoch": 12.53,
|
| 1745 |
+
"grad_norm": 0.8302854895591736,
|
| 1746 |
+
"learning_rate": 4.089114445425125e-05,
|
| 1747 |
+
"loss": 2.9604,
|
| 1748 |
+
"step": 233000
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 12.58,
|
| 1752 |
+
"grad_norm": 0.8315780758857727,
|
| 1753 |
+
"learning_rate": 4.0596940276551934e-05,
|
| 1754 |
+
"loss": 2.9655,
|
| 1755 |
+
"step": 234000
|
| 1756 |
+
},
|
| 1757 |
+
{
|
| 1758 |
+
"epoch": 12.64,
|
| 1759 |
+
"grad_norm": 0.8219565749168396,
|
| 1760 |
+
"learning_rate": 4.030273609885261e-05,
|
| 1761 |
+
"loss": 2.9651,
|
| 1762 |
+
"step": 235000
|
| 1763 |
+
},
|
| 1764 |
+
{
|
| 1765 |
+
"epoch": 12.69,
|
| 1766 |
+
"grad_norm": 0.8112459778785706,
|
| 1767 |
+
"learning_rate": 4.0008826125330986e-05,
|
| 1768 |
+
"loss": 2.9647,
|
| 1769 |
+
"step": 236000
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 12.75,
|
| 1773 |
+
"grad_norm": 0.8309066295623779,
|
| 1774 |
+
"learning_rate": 3.9714916151809357e-05,
|
| 1775 |
+
"loss": 2.9637,
|
| 1776 |
+
"step": 237000
|
| 1777 |
+
},
|
| 1778 |
+
{
|
| 1779 |
+
"epoch": 12.8,
|
| 1780 |
+
"grad_norm": 0.818537175655365,
|
| 1781 |
+
"learning_rate": 3.942071197411004e-05,
|
| 1782 |
+
"loss": 2.9648,
|
| 1783 |
+
"step": 238000
|
| 1784 |
+
},
|
| 1785 |
+
{
|
| 1786 |
+
"epoch": 12.85,
|
| 1787 |
+
"grad_norm": 0.829467236995697,
|
| 1788 |
+
"learning_rate": 3.912650779641071e-05,
|
| 1789 |
+
"loss": 2.9676,
|
| 1790 |
+
"step": 239000
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 12.91,
|
| 1794 |
+
"grad_norm": 0.8041892647743225,
|
| 1795 |
+
"learning_rate": 3.883230361871139e-05,
|
| 1796 |
+
"loss": 2.9631,
|
| 1797 |
+
"step": 240000
|
| 1798 |
+
},
|
| 1799 |
+
{
|
| 1800 |
+
"epoch": 12.96,
|
| 1801 |
+
"grad_norm": 0.8085007667541504,
|
| 1802 |
+
"learning_rate": 3.8538393645189764e-05,
|
| 1803 |
+
"loss": 2.967,
|
| 1804 |
+
"step": 241000
|
| 1805 |
+
},
|
| 1806 |
+
{
|
| 1807 |
+
"epoch": 13.0,
|
| 1808 |
+
"eval_accuracy": 0.40449210160668586,
|
| 1809 |
+
"eval_loss": 3.3969216346740723,
|
| 1810 |
+
"eval_runtime": 154.1242,
|
| 1811 |
+
"eval_samples_per_second": 375.788,
|
| 1812 |
+
"eval_steps_per_second": 5.872,
|
| 1813 |
+
"step": 241735
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 13.01,
|
| 1817 |
+
"grad_norm": 0.8514643311500549,
|
| 1818 |
+
"learning_rate": 3.824418946749044e-05,
|
| 1819 |
+
"loss": 2.9558,
|
| 1820 |
+
"step": 242000
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 13.07,
|
| 1824 |
+
"grad_norm": 0.8870530724525452,
|
| 1825 |
+
"learning_rate": 3.794998528979112e-05,
|
| 1826 |
+
"loss": 2.9199,
|
| 1827 |
+
"step": 243000
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 13.12,
|
| 1831 |
+
"grad_norm": 0.8721070885658264,
|
| 1832 |
+
"learning_rate": 3.765607531626949e-05,
|
| 1833 |
+
"loss": 2.9276,
|
| 1834 |
+
"step": 244000
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 13.18,
|
| 1838 |
+
"grad_norm": 0.8669698238372803,
|
| 1839 |
+
"learning_rate": 3.736187113857017e-05,
|
| 1840 |
+
"loss": 2.9317,
|
| 1841 |
+
"step": 245000
|
| 1842 |
+
},
|
| 1843 |
+
{
|
| 1844 |
+
"epoch": 13.23,
|
| 1845 |
+
"grad_norm": 0.8661383390426636,
|
| 1846 |
+
"learning_rate": 3.7067666960870846e-05,
|
| 1847 |
+
"loss": 2.9308,
|
| 1848 |
+
"step": 246000
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 13.28,
|
| 1852 |
+
"grad_norm": 0.860162079334259,
|
| 1853 |
+
"learning_rate": 3.677346278317153e-05,
|
| 1854 |
+
"loss": 2.9322,
|
| 1855 |
+
"step": 247000
|
| 1856 |
+
},
|
| 1857 |
+
{
|
| 1858 |
+
"epoch": 13.34,
|
| 1859 |
+
"grad_norm": 0.8323468565940857,
|
| 1860 |
+
"learning_rate": 3.64795528096499e-05,
|
| 1861 |
+
"loss": 2.9336,
|
| 1862 |
+
"step": 248000
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 13.39,
|
| 1866 |
+
"grad_norm": 0.8697391748428345,
|
| 1867 |
+
"learning_rate": 3.6185642836128275e-05,
|
| 1868 |
+
"loss": 2.9342,
|
| 1869 |
+
"step": 249000
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 13.44,
|
| 1873 |
+
"grad_norm": 0.8538952469825745,
|
| 1874 |
+
"learning_rate": 3.589143865842895e-05,
|
| 1875 |
+
"loss": 2.9373,
|
| 1876 |
+
"step": 250000
|
| 1877 |
+
},
|
| 1878 |
+
{
|
| 1879 |
+
"epoch": 13.5,
|
| 1880 |
+
"grad_norm": 0.848420262336731,
|
| 1881 |
+
"learning_rate": 3.559723448072963e-05,
|
| 1882 |
+
"loss": 2.9357,
|
| 1883 |
+
"step": 251000
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 13.55,
|
| 1887 |
+
"grad_norm": 0.8527274131774902,
|
| 1888 |
+
"learning_rate": 3.5303324507208e-05,
|
| 1889 |
+
"loss": 2.9394,
|
| 1890 |
+
"step": 252000
|
| 1891 |
+
},
|
| 1892 |
+
{
|
| 1893 |
+
"epoch": 13.61,
|
| 1894 |
+
"grad_norm": 0.8854946494102478,
|
| 1895 |
+
"learning_rate": 3.500912032950868e-05,
|
| 1896 |
+
"loss": 2.9423,
|
| 1897 |
+
"step": 253000
|
| 1898 |
+
},
|
| 1899 |
+
{
|
| 1900 |
+
"epoch": 13.66,
|
| 1901 |
+
"grad_norm": 0.8356419801712036,
|
| 1902 |
+
"learning_rate": 3.471491615180936e-05,
|
| 1903 |
+
"loss": 2.9412,
|
| 1904 |
+
"step": 254000
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 13.71,
|
| 1908 |
+
"grad_norm": 0.8529571890830994,
|
| 1909 |
+
"learning_rate": 3.442071197411003e-05,
|
| 1910 |
+
"loss": 2.9442,
|
| 1911 |
+
"step": 255000
|
| 1912 |
+
},
|
| 1913 |
+
{
|
| 1914 |
+
"epoch": 13.77,
|
| 1915 |
+
"grad_norm": 0.8669166564941406,
|
| 1916 |
+
"learning_rate": 3.412650779641071e-05,
|
| 1917 |
+
"loss": 2.945,
|
| 1918 |
+
"step": 256000
|
| 1919 |
+
},
|
| 1920 |
+
{
|
| 1921 |
+
"epoch": 13.82,
|
| 1922 |
+
"grad_norm": 0.8501659035682678,
|
| 1923 |
+
"learning_rate": 3.3832892027066786e-05,
|
| 1924 |
+
"loss": 2.9491,
|
| 1925 |
+
"step": 257000
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"epoch": 13.87,
|
| 1929 |
+
"grad_norm": 0.8519834876060486,
|
| 1930 |
+
"learning_rate": 3.353868784936746e-05,
|
| 1931 |
+
"loss": 2.946,
|
| 1932 |
+
"step": 258000
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"epoch": 13.93,
|
| 1936 |
+
"grad_norm": 0.8449692130088806,
|
| 1937 |
+
"learning_rate": 3.324477787584584e-05,
|
| 1938 |
+
"loss": 2.9439,
|
| 1939 |
+
"step": 259000
|
| 1940 |
+
},
|
| 1941 |
+
{
|
| 1942 |
+
"epoch": 13.98,
|
| 1943 |
+
"grad_norm": 0.8286080360412598,
|
| 1944 |
+
"learning_rate": 3.2950867902324215e-05,
|
| 1945 |
+
"loss": 2.9457,
|
| 1946 |
+
"step": 260000
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 14.0,
|
| 1950 |
+
"eval_accuracy": 0.4048622747603851,
|
| 1951 |
+
"eval_loss": 3.4079537391662598,
|
| 1952 |
+
"eval_runtime": 153.3879,
|
| 1953 |
+
"eval_samples_per_second": 377.592,
|
| 1954 |
+
"eval_steps_per_second": 5.9,
|
| 1955 |
+
"step": 260330
|
| 1956 |
+
},
|
| 1957 |
+
{
|
| 1958 |
+
"epoch": 14.04,
|
| 1959 |
+
"grad_norm": 0.8589282631874084,
|
| 1960 |
+
"learning_rate": 3.265666372462489e-05,
|
| 1961 |
+
"loss": 2.918,
|
| 1962 |
+
"step": 261000
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 14.09,
|
| 1966 |
+
"grad_norm": 0.8760070204734802,
|
| 1967 |
+
"learning_rate": 3.2362459546925564e-05,
|
| 1968 |
+
"loss": 2.9044,
|
| 1969 |
+
"step": 262000
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 14.14,
|
| 1973 |
+
"grad_norm": 0.883084237575531,
|
| 1974 |
+
"learning_rate": 3.2068255369226245e-05,
|
| 1975 |
+
"loss": 2.9054,
|
| 1976 |
+
"step": 263000
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 14.2,
|
| 1980 |
+
"grad_norm": 0.8560861349105835,
|
| 1981 |
+
"learning_rate": 3.177405119152692e-05,
|
| 1982 |
+
"loss": 2.9076,
|
| 1983 |
+
"step": 264000
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 14.25,
|
| 1987 |
+
"grad_norm": 0.9030760526657104,
|
| 1988 |
+
"learning_rate": 3.1479847013827594e-05,
|
| 1989 |
+
"loss": 2.9111,
|
| 1990 |
+
"step": 265000
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 14.3,
|
| 1994 |
+
"grad_norm": 0.8720868229866028,
|
| 1995 |
+
"learning_rate": 3.118593704030597e-05,
|
| 1996 |
+
"loss": 2.9138,
|
| 1997 |
+
"step": 266000
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 14.36,
|
| 2001 |
+
"grad_norm": 0.8849419951438904,
|
| 2002 |
+
"learning_rate": 3.089202706678435e-05,
|
| 2003 |
+
"loss": 2.9211,
|
| 2004 |
+
"step": 267000
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 14.41,
|
| 2008 |
+
"grad_norm": 0.8608791828155518,
|
| 2009 |
+
"learning_rate": 3.059782288908502e-05,
|
| 2010 |
+
"loss": 2.9176,
|
| 2011 |
+
"step": 268000
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 14.47,
|
| 2015 |
+
"grad_norm": 0.868854284286499,
|
| 2016 |
+
"learning_rate": 3.03036187113857e-05,
|
| 2017 |
+
"loss": 2.9216,
|
| 2018 |
+
"step": 269000
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 14.52,
|
| 2022 |
+
"grad_norm": 0.8694607615470886,
|
| 2023 |
+
"learning_rate": 3.0009414533686382e-05,
|
| 2024 |
+
"loss": 2.9244,
|
| 2025 |
+
"step": 270000
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 14.57,
|
| 2029 |
+
"grad_norm": 0.8707175254821777,
|
| 2030 |
+
"learning_rate": 2.9715504560164753e-05,
|
| 2031 |
+
"loss": 2.9216,
|
| 2032 |
+
"step": 271000
|
| 2033 |
+
},
|
| 2034 |
+
{
|
| 2035 |
+
"epoch": 14.63,
|
| 2036 |
+
"grad_norm": 0.907792329788208,
|
| 2037 |
+
"learning_rate": 2.9421300382465434e-05,
|
| 2038 |
+
"loss": 2.927,
|
| 2039 |
+
"step": 272000
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 14.68,
|
| 2043 |
+
"grad_norm": 0.8800376057624817,
|
| 2044 |
+
"learning_rate": 2.912739040894381e-05,
|
| 2045 |
+
"loss": 2.9259,
|
| 2046 |
+
"step": 273000
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 14.74,
|
| 2050 |
+
"grad_norm": 0.8789490461349487,
|
| 2051 |
+
"learning_rate": 2.8833186231244486e-05,
|
| 2052 |
+
"loss": 2.9261,
|
| 2053 |
+
"step": 274000
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 14.79,
|
| 2057 |
+
"grad_norm": 0.8529285192489624,
|
| 2058 |
+
"learning_rate": 2.8539276257722863e-05,
|
| 2059 |
+
"loss": 2.9234,
|
| 2060 |
+
"step": 275000
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 14.84,
|
| 2064 |
+
"grad_norm": 0.8620697855949402,
|
| 2065 |
+
"learning_rate": 2.8245072080023538e-05,
|
| 2066 |
+
"loss": 2.9255,
|
| 2067 |
+
"step": 276000
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 14.9,
|
| 2071 |
+
"grad_norm": 0.8713541626930237,
|
| 2072 |
+
"learning_rate": 2.7950867902324212e-05,
|
| 2073 |
+
"loss": 2.9287,
|
| 2074 |
+
"step": 277000
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 14.95,
|
| 2078 |
+
"grad_norm": 0.8321398496627808,
|
| 2079 |
+
"learning_rate": 2.765695792880259e-05,
|
| 2080 |
+
"loss": 2.9268,
|
| 2081 |
+
"step": 278000
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 15.0,
|
| 2085 |
+
"eval_accuracy": 0.40448188993348033,
|
| 2086 |
+
"eval_loss": 3.4128763675689697,
|
| 2087 |
+
"eval_runtime": 153.2847,
|
| 2088 |
+
"eval_samples_per_second": 377.846,
|
| 2089 |
+
"eval_steps_per_second": 5.904,
|
| 2090 |
+
"step": 278925
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 15.0,
|
| 2094 |
+
"grad_norm": 0.8870017528533936,
|
| 2095 |
+
"learning_rate": 2.7363047955280967e-05,
|
| 2096 |
+
"loss": 2.9236,
|
| 2097 |
+
"step": 279000
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 15.06,
|
| 2101 |
+
"grad_norm": 0.902341902256012,
|
| 2102 |
+
"learning_rate": 2.706884377758164e-05,
|
| 2103 |
+
"loss": 2.8917,
|
| 2104 |
+
"step": 280000
|
| 2105 |
+
},
|
| 2106 |
+
{
|
| 2107 |
+
"epoch": 15.11,
|
| 2108 |
+
"grad_norm": 0.8734650015830994,
|
| 2109 |
+
"learning_rate": 2.677463959988232e-05,
|
| 2110 |
+
"loss": 2.8882,
|
| 2111 |
+
"step": 281000
|
| 2112 |
+
},
|
| 2113 |
+
{
|
| 2114 |
+
"epoch": 15.17,
|
| 2115 |
+
"grad_norm": 0.8736613392829895,
|
| 2116 |
+
"learning_rate": 2.6480435422182997e-05,
|
| 2117 |
+
"loss": 2.895,
|
| 2118 |
+
"step": 282000
|
| 2119 |
+
},
|
| 2120 |
+
{
|
| 2121 |
+
"epoch": 15.22,
|
| 2122 |
+
"grad_norm": 0.8876488208770752,
|
| 2123 |
+
"learning_rate": 2.618652544866137e-05,
|
| 2124 |
+
"loss": 2.8917,
|
| 2125 |
+
"step": 283000
|
| 2126 |
+
},
|
| 2127 |
+
{
|
| 2128 |
+
"epoch": 15.27,
|
| 2129 |
+
"grad_norm": 0.9243654012680054,
|
| 2130 |
+
"learning_rate": 2.589232127096205e-05,
|
| 2131 |
+
"loss": 2.8952,
|
| 2132 |
+
"step": 284000
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 15.33,
|
| 2136 |
+
"grad_norm": 0.8820899724960327,
|
| 2137 |
+
"learning_rate": 2.5598117093262723e-05,
|
| 2138 |
+
"loss": 2.8973,
|
| 2139 |
+
"step": 285000
|
| 2140 |
+
},
|
| 2141 |
+
{
|
| 2142 |
+
"epoch": 15.38,
|
| 2143 |
+
"grad_norm": 0.9063917398452759,
|
| 2144 |
+
"learning_rate": 2.5304207119741104e-05,
|
| 2145 |
+
"loss": 2.8972,
|
| 2146 |
+
"step": 286000
|
| 2147 |
+
},
|
| 2148 |
+
{
|
| 2149 |
+
"epoch": 15.43,
|
| 2150 |
+
"grad_norm": 0.9035106301307678,
|
| 2151 |
+
"learning_rate": 2.501000294204178e-05,
|
| 2152 |
+
"loss": 2.9064,
|
| 2153 |
+
"step": 287000
|
| 2154 |
+
},
|
| 2155 |
+
{
|
| 2156 |
+
"epoch": 15.49,
|
| 2157 |
+
"grad_norm": 0.8639585971832275,
|
| 2158 |
+
"learning_rate": 2.4716092968520156e-05,
|
| 2159 |
+
"loss": 2.9013,
|
| 2160 |
+
"step": 288000
|
| 2161 |
+
},
|
| 2162 |
+
{
|
| 2163 |
+
"epoch": 15.54,
|
| 2164 |
+
"grad_norm": 0.8969417810440063,
|
| 2165 |
+
"learning_rate": 2.4421888790820833e-05,
|
| 2166 |
+
"loss": 2.9056,
|
| 2167 |
+
"step": 289000
|
| 2168 |
+
},
|
| 2169 |
+
{
|
| 2170 |
+
"epoch": 15.6,
|
| 2171 |
+
"grad_norm": 0.8919821977615356,
|
| 2172 |
+
"learning_rate": 2.4127684613121508e-05,
|
| 2173 |
+
"loss": 2.9073,
|
| 2174 |
+
"step": 290000
|
| 2175 |
+
},
|
| 2176 |
+
{
|
| 2177 |
+
"epoch": 15.65,
|
| 2178 |
+
"grad_norm": 0.8989285230636597,
|
| 2179 |
+
"learning_rate": 2.3833774639599885e-05,
|
| 2180 |
+
"loss": 2.9062,
|
| 2181 |
+
"step": 291000
|
| 2182 |
+
},
|
| 2183 |
+
{
|
| 2184 |
+
"epoch": 15.7,
|
| 2185 |
+
"grad_norm": 0.8888868689537048,
|
| 2186 |
+
"learning_rate": 2.353986466607826e-05,
|
| 2187 |
+
"loss": 2.9088,
|
| 2188 |
+
"step": 292000
|
| 2189 |
+
},
|
| 2190 |
+
{
|
| 2191 |
+
"epoch": 15.76,
|
| 2192 |
+
"grad_norm": 0.9275554418563843,
|
| 2193 |
+
"learning_rate": 2.3245660488378937e-05,
|
| 2194 |
+
"loss": 2.9069,
|
| 2195 |
+
"step": 293000
|
| 2196 |
+
},
|
| 2197 |
+
{
|
| 2198 |
+
"epoch": 15.81,
|
| 2199 |
+
"grad_norm": 0.8558235168457031,
|
| 2200 |
+
"learning_rate": 2.295145631067961e-05,
|
| 2201 |
+
"loss": 2.9037,
|
| 2202 |
+
"step": 294000
|
| 2203 |
+
},
|
| 2204 |
+
{
|
| 2205 |
+
"epoch": 15.86,
|
| 2206 |
+
"grad_norm": 0.8763130307197571,
|
| 2207 |
+
"learning_rate": 2.265725213298029e-05,
|
| 2208 |
+
"loss": 2.9064,
|
| 2209 |
+
"step": 295000
|
| 2210 |
+
},
|
| 2211 |
+
{
|
| 2212 |
+
"epoch": 15.92,
|
| 2213 |
+
"grad_norm": 0.8805882334709167,
|
| 2214 |
+
"learning_rate": 2.2363342159458667e-05,
|
| 2215 |
+
"loss": 2.9079,
|
| 2216 |
+
"step": 296000
|
| 2217 |
+
},
|
| 2218 |
+
{
|
| 2219 |
+
"epoch": 15.97,
|
| 2220 |
+
"grad_norm": 0.9118674993515015,
|
| 2221 |
+
"learning_rate": 2.206943218593704e-05,
|
| 2222 |
+
"loss": 2.911,
|
| 2223 |
+
"step": 297000
|
| 2224 |
+
},
|
| 2225 |
+
{
|
| 2226 |
+
"epoch": 16.0,
|
| 2227 |
+
"eval_accuracy": 0.4047090324802421,
|
| 2228 |
+
"eval_loss": 3.41589617729187,
|
| 2229 |
+
"eval_runtime": 153.0438,
|
| 2230 |
+
"eval_samples_per_second": 378.441,
|
| 2231 |
+
"eval_steps_per_second": 5.913,
|
| 2232 |
+
"step": 297520
|
| 2233 |
+
},
|
| 2234 |
+
{
|
| 2235 |
+
"epoch": 16.03,
|
| 2236 |
+
"grad_norm": 0.9044952392578125,
|
| 2237 |
+
"learning_rate": 2.177522800823772e-05,
|
| 2238 |
+
"loss": 2.8973,
|
| 2239 |
+
"step": 298000
|
| 2240 |
+
},
|
| 2241 |
+
{
|
| 2242 |
+
"epoch": 16.08,
|
| 2243 |
+
"grad_norm": 0.8891879320144653,
|
| 2244 |
+
"learning_rate": 2.1481023830538393e-05,
|
| 2245 |
+
"loss": 2.8726,
|
| 2246 |
+
"step": 299000
|
| 2247 |
+
},
|
| 2248 |
+
{
|
| 2249 |
+
"epoch": 16.13,
|
| 2250 |
+
"grad_norm": 0.9434779286384583,
|
| 2251 |
+
"learning_rate": 2.118681965283907e-05,
|
| 2252 |
+
"loss": 2.8738,
|
| 2253 |
+
"step": 300000
|
| 2254 |
+
},
|
| 2255 |
+
{
|
| 2256 |
+
"epoch": 16.19,
|
| 2257 |
+
"grad_norm": 0.9484353065490723,
|
| 2258 |
+
"learning_rate": 2.0892909679317445e-05,
|
| 2259 |
+
"loss": 2.8755,
|
| 2260 |
+
"step": 301000
|
| 2261 |
+
},
|
| 2262 |
+
{
|
| 2263 |
+
"epoch": 16.24,
|
| 2264 |
+
"grad_norm": 0.9076915979385376,
|
| 2265 |
+
"learning_rate": 2.0598705501618122e-05,
|
| 2266 |
+
"loss": 2.8804,
|
| 2267 |
+
"step": 302000
|
| 2268 |
+
},
|
| 2269 |
+
{
|
| 2270 |
+
"epoch": 16.29,
|
| 2271 |
+
"grad_norm": 0.9162052869796753,
|
| 2272 |
+
"learning_rate": 2.03047955280965e-05,
|
| 2273 |
+
"loss": 2.8824,
|
| 2274 |
+
"step": 303000
|
| 2275 |
+
},
|
| 2276 |
+
{
|
| 2277 |
+
"epoch": 16.35,
|
| 2278 |
+
"grad_norm": 0.9373551607131958,
|
| 2279 |
+
"learning_rate": 2.0010591350397174e-05,
|
| 2280 |
+
"loss": 2.8834,
|
| 2281 |
+
"step": 304000
|
| 2282 |
+
},
|
| 2283 |
+
{
|
| 2284 |
+
"epoch": 16.4,
|
| 2285 |
+
"grad_norm": 0.9191654324531555,
|
| 2286 |
+
"learning_rate": 1.9716387172697852e-05,
|
| 2287 |
+
"loss": 2.8875,
|
| 2288 |
+
"step": 305000
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 16.46,
|
| 2292 |
+
"grad_norm": 0.9333674907684326,
|
| 2293 |
+
"learning_rate": 1.942218299499853e-05,
|
| 2294 |
+
"loss": 2.8867,
|
| 2295 |
+
"step": 306000
|
| 2296 |
+
},
|
| 2297 |
+
{
|
| 2298 |
+
"epoch": 16.51,
|
| 2299 |
+
"grad_norm": 0.9224890470504761,
|
| 2300 |
+
"learning_rate": 1.9128273021476907e-05,
|
| 2301 |
+
"loss": 2.886,
|
| 2302 |
+
"step": 307000
|
| 2303 |
+
},
|
| 2304 |
+
{
|
| 2305 |
+
"epoch": 16.56,
|
| 2306 |
+
"grad_norm": 0.9154072403907776,
|
| 2307 |
+
"learning_rate": 1.8834068843777582e-05,
|
| 2308 |
+
"loss": 2.8859,
|
| 2309 |
+
"step": 308000
|
| 2310 |
+
},
|
| 2311 |
+
{
|
| 2312 |
+
"epoch": 16.62,
|
| 2313 |
+
"grad_norm": 0.8822410702705383,
|
| 2314 |
+
"learning_rate": 1.854015887025596e-05,
|
| 2315 |
+
"loss": 2.8887,
|
| 2316 |
+
"step": 309000
|
| 2317 |
+
},
|
| 2318 |
+
{
|
| 2319 |
+
"epoch": 16.67,
|
| 2320 |
+
"grad_norm": 0.9022771716117859,
|
| 2321 |
+
"learning_rate": 1.8245954692556637e-05,
|
| 2322 |
+
"loss": 2.8884,
|
| 2323 |
+
"step": 310000
|
| 2324 |
+
},
|
| 2325 |
+
{
|
| 2326 |
+
"epoch": 16.72,
|
| 2327 |
+
"grad_norm": 0.8923472762107849,
|
| 2328 |
+
"learning_rate": 1.795175051485731e-05,
|
| 2329 |
+
"loss": 2.893,
|
| 2330 |
+
"step": 311000
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 16.78,
|
| 2334 |
+
"grad_norm": 0.9239156246185303,
|
| 2335 |
+
"learning_rate": 1.765784054133569e-05,
|
| 2336 |
+
"loss": 2.8919,
|
| 2337 |
+
"step": 312000
|
| 2338 |
+
},
|
| 2339 |
+
{
|
| 2340 |
+
"epoch": 16.83,
|
| 2341 |
+
"grad_norm": 0.9343294501304626,
|
| 2342 |
+
"learning_rate": 1.7363636363636366e-05,
|
| 2343 |
+
"loss": 2.8876,
|
| 2344 |
+
"step": 313000
|
| 2345 |
+
},
|
| 2346 |
+
{
|
| 2347 |
+
"epoch": 16.89,
|
| 2348 |
+
"grad_norm": 0.9064814448356628,
|
| 2349 |
+
"learning_rate": 1.706972639011474e-05,
|
| 2350 |
+
"loss": 2.8913,
|
| 2351 |
+
"step": 314000
|
| 2352 |
+
},
|
| 2353 |
+
{
|
| 2354 |
+
"epoch": 16.94,
|
| 2355 |
+
"grad_norm": 0.9277246594429016,
|
| 2356 |
+
"learning_rate": 1.6775522212415418e-05,
|
| 2357 |
+
"loss": 2.8909,
|
| 2358 |
+
"step": 315000
|
| 2359 |
+
},
|
| 2360 |
+
{
|
| 2361 |
+
"epoch": 16.99,
|
| 2362 |
+
"grad_norm": 0.9258543848991394,
|
| 2363 |
+
"learning_rate": 1.6481612238893792e-05,
|
| 2364 |
+
"loss": 2.8964,
|
| 2365 |
+
"step": 316000
|
| 2366 |
+
},
|
| 2367 |
+
{
|
| 2368 |
+
"epoch": 17.0,
|
| 2369 |
+
"eval_accuracy": 0.40512663616869843,
|
| 2370 |
+
"eval_loss": 3.422145366668701,
|
| 2371 |
+
"eval_runtime": 153.2527,
|
| 2372 |
+
"eval_samples_per_second": 377.925,
|
| 2373 |
+
"eval_steps_per_second": 5.905,
|
| 2374 |
+
"step": 316115
|
| 2375 |
+
},
|
| 2376 |
+
{
|
| 2377 |
+
"epoch": 17.05,
|
| 2378 |
+
"grad_norm": 0.9131640195846558,
|
| 2379 |
+
"learning_rate": 1.6187702265372166e-05,
|
| 2380 |
+
"loss": 2.864,
|
| 2381 |
+
"step": 317000
|
| 2382 |
+
},
|
| 2383 |
+
{
|
| 2384 |
+
"epoch": 17.1,
|
| 2385 |
+
"grad_norm": 1.1945741176605225,
|
| 2386 |
+
"learning_rate": 1.5893498087672844e-05,
|
| 2387 |
+
"loss": 2.8617,
|
| 2388 |
+
"step": 318000
|
| 2389 |
+
},
|
| 2390 |
+
{
|
| 2391 |
+
"epoch": 17.16,
|
| 2392 |
+
"grad_norm": 0.9226692914962769,
|
| 2393 |
+
"learning_rate": 1.5599588114151225e-05,
|
| 2394 |
+
"loss": 2.8644,
|
| 2395 |
+
"step": 319000
|
| 2396 |
+
},
|
| 2397 |
+
{
|
| 2398 |
+
"epoch": 17.21,
|
| 2399 |
+
"grad_norm": 0.9040902256965637,
|
| 2400 |
+
"learning_rate": 1.53053839364519e-05,
|
| 2401 |
+
"loss": 2.8643,
|
| 2402 |
+
"step": 320000
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 17.26,
|
| 2406 |
+
"grad_norm": 0.9265015721321106,
|
| 2407 |
+
"learning_rate": 1.5011179758752575e-05,
|
| 2408 |
+
"loss": 2.8684,
|
| 2409 |
+
"step": 321000
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 17.32,
|
| 2413 |
+
"grad_norm": 0.940202534198761,
|
| 2414 |
+
"learning_rate": 1.4716975581053253e-05,
|
| 2415 |
+
"loss": 2.8653,
|
| 2416 |
+
"step": 322000
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 17.37,
|
| 2420 |
+
"grad_norm": 0.9231876730918884,
|
| 2421 |
+
"learning_rate": 1.4423065607531627e-05,
|
| 2422 |
+
"loss": 2.8729,
|
| 2423 |
+
"step": 323000
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 17.42,
|
| 2427 |
+
"grad_norm": 0.9177024960517883,
|
| 2428 |
+
"learning_rate": 1.4128861429832305e-05,
|
| 2429 |
+
"loss": 2.8688,
|
| 2430 |
+
"step": 324000
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 17.48,
|
| 2434 |
+
"grad_norm": 0.9014251828193665,
|
| 2435 |
+
"learning_rate": 1.3834657252132983e-05,
|
| 2436 |
+
"loss": 2.8711,
|
| 2437 |
+
"step": 325000
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 17.53,
|
| 2441 |
+
"grad_norm": 0.9649085402488708,
|
| 2442 |
+
"learning_rate": 1.3540453074433657e-05,
|
| 2443 |
+
"loss": 2.8721,
|
| 2444 |
+
"step": 326000
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 17.59,
|
| 2448 |
+
"grad_norm": 0.9398193955421448,
|
| 2449 |
+
"learning_rate": 1.3246543100912034e-05,
|
| 2450 |
+
"loss": 2.8725,
|
| 2451 |
+
"step": 327000
|
| 2452 |
+
},
|
| 2453 |
+
{
|
| 2454 |
+
"epoch": 17.64,
|
| 2455 |
+
"grad_norm": 0.9121034145355225,
|
| 2456 |
+
"learning_rate": 1.2952338923212709e-05,
|
| 2457 |
+
"loss": 2.8771,
|
| 2458 |
+
"step": 328000
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 17.69,
|
| 2462 |
+
"grad_norm": 0.8918709754943848,
|
| 2463 |
+
"learning_rate": 1.2658134745513387e-05,
|
| 2464 |
+
"loss": 2.8721,
|
| 2465 |
+
"step": 329000
|
| 2466 |
+
},
|
| 2467 |
+
{
|
| 2468 |
+
"epoch": 17.75,
|
| 2469 |
+
"grad_norm": 0.8925199508666992,
|
| 2470 |
+
"learning_rate": 1.2363930567814063e-05,
|
| 2471 |
+
"loss": 2.8722,
|
| 2472 |
+
"step": 330000
|
| 2473 |
+
},
|
| 2474 |
+
{
|
| 2475 |
+
"epoch": 17.8,
|
| 2476 |
+
"grad_norm": 0.9330604076385498,
|
| 2477 |
+
"learning_rate": 1.207031479847014e-05,
|
| 2478 |
+
"loss": 2.8763,
|
| 2479 |
+
"step": 331000
|
| 2480 |
+
},
|
| 2481 |
+
{
|
| 2482 |
+
"epoch": 17.85,
|
| 2483 |
+
"grad_norm": 0.9173546433448792,
|
| 2484 |
+
"learning_rate": 1.1776110620770816e-05,
|
| 2485 |
+
"loss": 2.8743,
|
| 2486 |
+
"step": 332000
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 17.91,
|
| 2490 |
+
"grad_norm": 0.9605908393859863,
|
| 2491 |
+
"learning_rate": 1.1481906443071492e-05,
|
| 2492 |
+
"loss": 2.8716,
|
| 2493 |
+
"step": 333000
|
| 2494 |
+
},
|
| 2495 |
+
{
|
| 2496 |
+
"epoch": 17.96,
|
| 2497 |
+
"grad_norm": 0.9416488409042358,
|
| 2498 |
+
"learning_rate": 1.1187702265372168e-05,
|
| 2499 |
+
"loss": 2.8758,
|
| 2500 |
+
"step": 334000
|
| 2501 |
+
},
|
| 2502 |
+
{
|
| 2503 |
+
"epoch": 18.0,
|
| 2504 |
+
"eval_accuracy": 0.4054102116463327,
|
| 2505 |
+
"eval_loss": 3.428605079650879,
|
| 2506 |
+
"eval_runtime": 153.3009,
|
| 2507 |
+
"eval_samples_per_second": 377.806,
|
| 2508 |
+
"eval_steps_per_second": 5.903,
|
| 2509 |
+
"step": 334710
|
| 2510 |
+
},
|
| 2511 |
+
{
|
| 2512 |
+
"epoch": 18.02,
|
| 2513 |
+
"grad_norm": 0.9360882639884949,
|
| 2514 |
+
"learning_rate": 1.0893498087672846e-05,
|
| 2515 |
+
"loss": 2.8649,
|
| 2516 |
+
"step": 335000
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 18.07,
|
| 2520 |
+
"grad_norm": 0.9502159357070923,
|
| 2521 |
+
"learning_rate": 1.0599588114151222e-05,
|
| 2522 |
+
"loss": 2.8497,
|
| 2523 |
+
"step": 336000
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 18.12,
|
| 2527 |
+
"grad_norm": 0.9194748997688293,
|
| 2528 |
+
"learning_rate": 1.0305383936451898e-05,
|
| 2529 |
+
"loss": 2.8492,
|
| 2530 |
+
"step": 337000
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"epoch": 18.18,
|
| 2534 |
+
"grad_norm": 0.9241055250167847,
|
| 2535 |
+
"learning_rate": 1.0011473962930273e-05,
|
| 2536 |
+
"loss": 2.8539,
|
| 2537 |
+
"step": 338000
|
| 2538 |
+
},
|
| 2539 |
+
{
|
| 2540 |
+
"epoch": 18.23,
|
| 2541 |
+
"grad_norm": 0.9047831296920776,
|
| 2542 |
+
"learning_rate": 9.717563989408649e-06,
|
| 2543 |
+
"loss": 2.8555,
|
| 2544 |
+
"step": 339000
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"epoch": 18.28,
|
| 2548 |
+
"grad_norm": 0.9505220651626587,
|
| 2549 |
+
"learning_rate": 9.423359811709327e-06,
|
| 2550 |
+
"loss": 2.8542,
|
| 2551 |
+
"step": 340000
|
| 2552 |
+
},
|
| 2553 |
+
{
|
| 2554 |
+
"epoch": 18.34,
|
| 2555 |
+
"grad_norm": 0.9681676030158997,
|
| 2556 |
+
"learning_rate": 9.129155634010003e-06,
|
| 2557 |
+
"loss": 2.8526,
|
| 2558 |
+
"step": 341000
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 18.39,
|
| 2562 |
+
"grad_norm": 0.9679902791976929,
|
| 2563 |
+
"learning_rate": 8.834951456310681e-06,
|
| 2564 |
+
"loss": 2.8543,
|
| 2565 |
+
"step": 342000
|
| 2566 |
+
},
|
| 2567 |
+
{
|
| 2568 |
+
"epoch": 18.45,
|
| 2569 |
+
"grad_norm": 0.9757332801818848,
|
| 2570 |
+
"learning_rate": 8.540747278611357e-06,
|
| 2571 |
+
"loss": 2.8569,
|
| 2572 |
+
"step": 343000
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"epoch": 18.5,
|
| 2576 |
+
"grad_norm": 0.9353688359260559,
|
| 2577 |
+
"learning_rate": 8.246837305089733e-06,
|
| 2578 |
+
"loss": 2.8563,
|
| 2579 |
+
"step": 344000
|
| 2580 |
+
},
|
| 2581 |
+
{
|
| 2582 |
+
"epoch": 18.55,
|
| 2583 |
+
"grad_norm": 0.9887840747833252,
|
| 2584 |
+
"learning_rate": 7.952633127390409e-06,
|
| 2585 |
+
"loss": 2.8577,
|
| 2586 |
+
"step": 345000
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"epoch": 18.61,
|
| 2590 |
+
"grad_norm": 0.9348433613777161,
|
| 2591 |
+
"learning_rate": 7.658723153868786e-06,
|
| 2592 |
+
"loss": 2.8606,
|
| 2593 |
+
"step": 346000
|
| 2594 |
+
},
|
| 2595 |
+
{
|
| 2596 |
+
"epoch": 18.66,
|
| 2597 |
+
"grad_norm": 0.9570315480232239,
|
| 2598 |
+
"learning_rate": 7.364518976169461e-06,
|
| 2599 |
+
"loss": 2.8603,
|
| 2600 |
+
"step": 347000
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 18.71,
|
| 2604 |
+
"grad_norm": 0.9226345419883728,
|
| 2605 |
+
"learning_rate": 7.0703147984701384e-06,
|
| 2606 |
+
"loss": 2.8586,
|
| 2607 |
+
"step": 348000
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 18.77,
|
| 2611 |
+
"grad_norm": 0.9210349321365356,
|
| 2612 |
+
"learning_rate": 6.776110620770815e-06,
|
| 2613 |
+
"loss": 2.8566,
|
| 2614 |
+
"step": 349000
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"epoch": 18.82,
|
| 2618 |
+
"grad_norm": 0.9562284350395203,
|
| 2619 |
+
"learning_rate": 6.482200647249191e-06,
|
| 2620 |
+
"loss": 2.8609,
|
| 2621 |
+
"step": 350000
|
| 2622 |
+
},
|
| 2623 |
+
{
|
| 2624 |
+
"epoch": 18.88,
|
| 2625 |
+
"grad_norm": 0.9210054874420166,
|
| 2626 |
+
"learning_rate": 6.187996469549868e-06,
|
| 2627 |
+
"loss": 2.8607,
|
| 2628 |
+
"step": 351000
|
| 2629 |
+
},
|
| 2630 |
+
{
|
| 2631 |
+
"epoch": 18.93,
|
| 2632 |
+
"grad_norm": 0.9457345008850098,
|
| 2633 |
+
"learning_rate": 5.893792291850544e-06,
|
| 2634 |
+
"loss": 2.8598,
|
| 2635 |
+
"step": 352000
|
| 2636 |
+
},
|
| 2637 |
+
{
|
| 2638 |
+
"epoch": 18.98,
|
| 2639 |
+
"grad_norm": 0.9460088610649109,
|
| 2640 |
+
"learning_rate": 5.59988231832892e-06,
|
| 2641 |
+
"loss": 2.858,
|
| 2642 |
+
"step": 353000
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 19.0,
|
| 2646 |
+
"eval_accuracy": 0.4053840106427133,
|
| 2647 |
+
"eval_loss": 3.426475763320923,
|
| 2648 |
+
"eval_runtime": 153.612,
|
| 2649 |
+
"eval_samples_per_second": 377.041,
|
| 2650 |
+
"eval_steps_per_second": 5.891,
|
| 2651 |
+
"step": 353305
|
| 2652 |
+
},
|
| 2653 |
+
{
|
| 2654 |
+
"epoch": 19.04,
|
| 2655 |
+
"grad_norm": 0.9623994827270508,
|
| 2656 |
+
"learning_rate": 5.305678140629597e-06,
|
| 2657 |
+
"loss": 2.8452,
|
| 2658 |
+
"step": 354000
|
| 2659 |
+
},
|
| 2660 |
+
{
|
| 2661 |
+
"epoch": 19.09,
|
| 2662 |
+
"grad_norm": 0.9756667613983154,
|
| 2663 |
+
"learning_rate": 5.011473962930274e-06,
|
| 2664 |
+
"loss": 2.842,
|
| 2665 |
+
"step": 355000
|
| 2666 |
+
},
|
| 2667 |
+
{
|
| 2668 |
+
"epoch": 19.14,
|
| 2669 |
+
"grad_norm": 0.9565538763999939,
|
| 2670 |
+
"learning_rate": 4.71726978523095e-06,
|
| 2671 |
+
"loss": 2.8394,
|
| 2672 |
+
"step": 356000
|
| 2673 |
+
},
|
| 2674 |
+
{
|
| 2675 |
+
"epoch": 19.2,
|
| 2676 |
+
"grad_norm": 0.9308847784996033,
|
| 2677 |
+
"learning_rate": 4.4233598117093264e-06,
|
| 2678 |
+
"loss": 2.8404,
|
| 2679 |
+
"step": 357000
|
| 2680 |
+
},
|
| 2681 |
+
{
|
| 2682 |
+
"epoch": 19.25,
|
| 2683 |
+
"grad_norm": 0.9186875224113464,
|
| 2684 |
+
"learning_rate": 4.1291556340100035e-06,
|
| 2685 |
+
"loss": 2.8446,
|
| 2686 |
+
"step": 358000
|
| 2687 |
+
},
|
| 2688 |
+
{
|
| 2689 |
+
"epoch": 19.31,
|
| 2690 |
+
"grad_norm": 0.9542786478996277,
|
| 2691 |
+
"learning_rate": 3.83495145631068e-06,
|
| 2692 |
+
"loss": 2.8453,
|
| 2693 |
+
"step": 359000
|
| 2694 |
+
},
|
| 2695 |
+
{
|
| 2696 |
+
"epoch": 19.36,
|
| 2697 |
+
"grad_norm": 0.9696632027626038,
|
| 2698 |
+
"learning_rate": 3.5410414827890556e-06,
|
| 2699 |
+
"loss": 2.8454,
|
| 2700 |
+
"step": 360000
|
| 2701 |
+
},
|
| 2702 |
+
{
|
| 2703 |
+
"epoch": 19.41,
|
| 2704 |
+
"grad_norm": 0.9518758058547974,
|
| 2705 |
+
"learning_rate": 3.2471315092674317e-06,
|
| 2706 |
+
"loss": 2.8447,
|
| 2707 |
+
"step": 361000
|
| 2708 |
+
},
|
| 2709 |
+
{
|
| 2710 |
+
"epoch": 19.47,
|
| 2711 |
+
"grad_norm": 0.9522430896759033,
|
| 2712 |
+
"learning_rate": 2.9529273315681087e-06,
|
| 2713 |
+
"loss": 2.8461,
|
| 2714 |
+
"step": 362000
|
| 2715 |
+
},
|
| 2716 |
+
{
|
| 2717 |
+
"epoch": 19.52,
|
| 2718 |
+
"grad_norm": 0.9725024700164795,
|
| 2719 |
+
"learning_rate": 2.6587231538687853e-06,
|
| 2720 |
+
"loss": 2.8434,
|
| 2721 |
+
"step": 363000
|
| 2722 |
+
},
|
| 2723 |
+
{
|
| 2724 |
+
"epoch": 19.58,
|
| 2725 |
+
"grad_norm": 0.9441395401954651,
|
| 2726 |
+
"learning_rate": 2.364518976169462e-06,
|
| 2727 |
+
"loss": 2.8441,
|
| 2728 |
+
"step": 364000
|
| 2729 |
+
},
|
| 2730 |
+
{
|
| 2731 |
+
"epoch": 19.63,
|
| 2732 |
+
"grad_norm": 0.9425933957099915,
|
| 2733 |
+
"learning_rate": 2.0706090026478374e-06,
|
| 2734 |
+
"loss": 2.8438,
|
| 2735 |
+
"step": 365000
|
| 2736 |
+
},
|
| 2737 |
+
{
|
| 2738 |
+
"epoch": 19.68,
|
| 2739 |
+
"grad_norm": 0.9448084831237793,
|
| 2740 |
+
"learning_rate": 1.7764048249485142e-06,
|
| 2741 |
+
"loss": 2.8458,
|
| 2742 |
+
"step": 366000
|
| 2743 |
+
},
|
| 2744 |
+
{
|
| 2745 |
+
"epoch": 19.74,
|
| 2746 |
+
"grad_norm": 0.9476189613342285,
|
| 2747 |
+
"learning_rate": 1.4824948514268903e-06,
|
| 2748 |
+
"loss": 2.8462,
|
| 2749 |
+
"step": 367000
|
| 2750 |
+
},
|
| 2751 |
+
{
|
| 2752 |
+
"epoch": 19.79,
|
| 2753 |
+
"grad_norm": 0.9310097694396973,
|
| 2754 |
+
"learning_rate": 1.188290673727567e-06,
|
| 2755 |
+
"loss": 2.8437,
|
| 2756 |
+
"step": 368000
|
| 2757 |
+
},
|
| 2758 |
+
{
|
| 2759 |
+
"epoch": 19.84,
|
| 2760 |
+
"grad_norm": 0.9563276171684265,
|
| 2761 |
+
"learning_rate": 8.940864960282438e-07,
|
| 2762 |
+
"loss": 2.8462,
|
| 2763 |
+
"step": 369000
|
| 2764 |
+
},
|
| 2765 |
+
{
|
| 2766 |
+
"epoch": 19.9,
|
| 2767 |
+
"grad_norm": 0.9808979630470276,
|
| 2768 |
+
"learning_rate": 5.998823183289204e-07,
|
| 2769 |
+
"loss": 2.8412,
|
| 2770 |
+
"step": 370000
|
| 2771 |
+
},
|
| 2772 |
+
{
|
| 2773 |
+
"epoch": 19.95,
|
| 2774 |
+
"grad_norm": 0.9566290378570557,
|
| 2775 |
+
"learning_rate": 3.0597234480729626e-07,
|
| 2776 |
+
"loss": 2.8493,
|
| 2777 |
+
"step": 371000
|
| 2778 |
+
},
|
| 2779 |
+
{
|
| 2780 |
+
"epoch": 20.0,
|
| 2781 |
+
"eval_accuracy": 0.40547080986496004,
|
| 2782 |
+
"eval_loss": 3.432330846786499,
|
| 2783 |
+
"eval_runtime": 154.1,
|
| 2784 |
+
"eval_samples_per_second": 375.847,
|
| 2785 |
+
"eval_steps_per_second": 5.873,
|
| 2786 |
+
"step": 371900
|
| 2787 |
+
},
|
| 2788 |
+
{
|
| 2789 |
+
"epoch": 20.0,
|
| 2790 |
+
"step": 371900,
|
| 2791 |
+
"total_flos": 1.56693365388288e+18,
|
| 2792 |
+
"train_loss": 3.1561156885198893,
|
| 2793 |
+
"train_runtime": 81199.8655,
|
| 2794 |
+
"train_samples_per_second": 146.561,
|
| 2795 |
+
"train_steps_per_second": 4.58
|
| 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 |
+
}
|