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_measure_nps_as_singular_new-1e-3
|
| 8 |
-
results:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
---
|
| 10 |
|
| 11 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
@@ -13,7 +25,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 13 |
|
| 14 |
# smolm-autoreg-bpe-counterfactual_babylm_measure_nps_as_singular_new-1e-3
|
| 15 |
|
| 16 |
-
This model was trained from scratch on
|
| 17 |
It achieves the following results on the evaluation set:
|
| 18 |
- Loss: 3.4116
|
| 19 |
- Accuracy: 0.4111
|
|
|
|
| 1 |
---
|
| 2 |
tags:
|
| 3 |
- generated_from_trainer
|
| 4 |
+
datasets:
|
| 5 |
+
- kanishka/counterfactual_babylm_measure_nps_as_singular_new
|
| 6 |
metrics:
|
| 7 |
- accuracy
|
| 8 |
model-index:
|
| 9 |
- name: smolm-autoreg-bpe-counterfactual_babylm_measure_nps_as_singular_new-1e-3
|
| 10 |
+
results:
|
| 11 |
+
- task:
|
| 12 |
+
name: Causal Language Modeling
|
| 13 |
+
type: text-generation
|
| 14 |
+
dataset:
|
| 15 |
+
name: kanishka/counterfactual_babylm_measure_nps_as_singular_new
|
| 16 |
+
type: kanishka/counterfactual_babylm_measure_nps_as_singular_new
|
| 17 |
+
metrics:
|
| 18 |
+
- name: Accuracy
|
| 19 |
+
type: accuracy
|
| 20 |
+
value: 0.4110677518157195
|
| 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_measure_nps_as_singular_new-1e-3
|
| 27 |
|
| 28 |
+
This model was trained from scratch on the kanishka/counterfactual_babylm_measure_nps_as_singular_new dataset.
|
| 29 |
It achieves the following results on the evaluation set:
|
| 30 |
- Loss: 3.4116
|
| 31 |
- Accuracy: 0.4111
|
all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"eval_accuracy": 0.4110677518157195,
|
| 4 |
+
"eval_loss": 3.411550760269165,
|
| 5 |
+
"eval_runtime": 154.9473,
|
| 6 |
+
"eval_samples": 57921,
|
| 7 |
+
"eval_samples_per_second": 373.811,
|
| 8 |
+
"eval_steps_per_second": 5.847,
|
| 9 |
+
"perplexity": 30.31221480835158,
|
| 10 |
+
"train_loss": 3.012804690259411,
|
| 11 |
+
"train_runtime": 81277.7938,
|
| 12 |
+
"train_samples": 595248,
|
| 13 |
+
"train_samples_per_second": 146.472,
|
| 14 |
+
"train_steps_per_second": 4.577
|
| 15 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"eval_accuracy": 0.4110677518157195,
|
| 4 |
+
"eval_loss": 3.411550760269165,
|
| 5 |
+
"eval_runtime": 154.9473,
|
| 6 |
+
"eval_samples": 57921,
|
| 7 |
+
"eval_samples_per_second": 373.811,
|
| 8 |
+
"eval_steps_per_second": 5.847,
|
| 9 |
+
"perplexity": 30.31221480835158
|
| 10 |
+
}
|
runs/Jun18_18-45-08_phyl-ling-p01.la.utexas.edu/events.out.tfevents.1718836117.phyl-ling-p01.la.utexas.edu.752720.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:325bbc8c5023d417d6fc387d30cf0286a44ade99cbfcd6012d534ee8e1853603
|
| 3 |
+
size 417
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"train_loss": 3.012804690259411,
|
| 4 |
+
"train_runtime": 81277.7938,
|
| 5 |
+
"train_samples": 595248,
|
| 6 |
+
"train_samples_per_second": 146.472,
|
| 7 |
+
"train_steps_per_second": 4.577
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2814 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 20.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 372040,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.05,
|
| 13 |
+
"grad_norm": 0.772857129573822,
|
| 14 |
+
"learning_rate": 3.125e-05,
|
| 15 |
+
"loss": 6.1994,
|
| 16 |
+
"step": 1000
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.11,
|
| 20 |
+
"grad_norm": 0.9309808611869812,
|
| 21 |
+
"learning_rate": 6.25e-05,
|
| 22 |
+
"loss": 4.9973,
|
| 23 |
+
"step": 2000
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.16,
|
| 27 |
+
"grad_norm": 0.893162190914154,
|
| 28 |
+
"learning_rate": 9.375e-05,
|
| 29 |
+
"loss": 4.6739,
|
| 30 |
+
"step": 3000
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.22,
|
| 34 |
+
"grad_norm": 0.7998523116111755,
|
| 35 |
+
"learning_rate": 0.000125,
|
| 36 |
+
"loss": 4.4521,
|
| 37 |
+
"step": 4000
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.27,
|
| 41 |
+
"grad_norm": 0.7740798592567444,
|
| 42 |
+
"learning_rate": 0.00015625,
|
| 43 |
+
"loss": 4.2992,
|
| 44 |
+
"step": 5000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.32,
|
| 48 |
+
"grad_norm": 0.725793719291687,
|
| 49 |
+
"learning_rate": 0.0001875,
|
| 50 |
+
"loss": 4.181,
|
| 51 |
+
"step": 6000
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.38,
|
| 55 |
+
"grad_norm": 0.7274988889694214,
|
| 56 |
+
"learning_rate": 0.00021875,
|
| 57 |
+
"loss": 4.0768,
|
| 58 |
+
"step": 7000
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.43,
|
| 62 |
+
"grad_norm": 0.6781344413757324,
|
| 63 |
+
"learning_rate": 0.00025,
|
| 64 |
+
"loss": 3.9903,
|
| 65 |
+
"step": 8000
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.48,
|
| 69 |
+
"grad_norm": 0.6776793599128723,
|
| 70 |
+
"learning_rate": 0.00028121875,
|
| 71 |
+
"loss": 3.9081,
|
| 72 |
+
"step": 9000
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.54,
|
| 76 |
+
"grad_norm": 0.6176994442939758,
|
| 77 |
+
"learning_rate": 0.00031246875000000003,
|
| 78 |
+
"loss": 3.8477,
|
| 79 |
+
"step": 10000
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.59,
|
| 83 |
+
"grad_norm": 0.5840777158737183,
|
| 84 |
+
"learning_rate": 0.00034368749999999997,
|
| 85 |
+
"loss": 3.8126,
|
| 86 |
+
"step": 11000
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.65,
|
| 90 |
+
"grad_norm": 0.5407571792602539,
|
| 91 |
+
"learning_rate": 0.00037490625,
|
| 92 |
+
"loss": 3.7654,
|
| 93 |
+
"step": 12000
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.7,
|
| 97 |
+
"grad_norm": 0.5007668137550354,
|
| 98 |
+
"learning_rate": 0.00040615625,
|
| 99 |
+
"loss": 3.728,
|
| 100 |
+
"step": 13000
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.75,
|
| 104 |
+
"grad_norm": 0.46616649627685547,
|
| 105 |
+
"learning_rate": 0.00043737500000000005,
|
| 106 |
+
"loss": 3.6994,
|
| 107 |
+
"step": 14000
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.81,
|
| 111 |
+
"grad_norm": 0.40752965211868286,
|
| 112 |
+
"learning_rate": 0.000468625,
|
| 113 |
+
"loss": 3.6746,
|
| 114 |
+
"step": 15000
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.86,
|
| 118 |
+
"grad_norm": 0.4074413478374481,
|
| 119 |
+
"learning_rate": 0.00049984375,
|
| 120 |
+
"loss": 3.6544,
|
| 121 |
+
"step": 16000
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.91,
|
| 125 |
+
"grad_norm": 0.38538795709609985,
|
| 126 |
+
"learning_rate": 0.00053109375,
|
| 127 |
+
"loss": 3.6247,
|
| 128 |
+
"step": 17000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.97,
|
| 132 |
+
"grad_norm": 0.39283987879753113,
|
| 133 |
+
"learning_rate": 0.0005623125,
|
| 134 |
+
"loss": 3.6071,
|
| 135 |
+
"step": 18000
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 1.0,
|
| 139 |
+
"eval_accuracy": 0.35732314046161223,
|
| 140 |
+
"eval_loss": 3.788609266281128,
|
| 141 |
+
"eval_runtime": 153.8947,
|
| 142 |
+
"eval_samples_per_second": 376.368,
|
| 143 |
+
"eval_steps_per_second": 5.887,
|
| 144 |
+
"step": 18602
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 1.02,
|
| 148 |
+
"grad_norm": 0.33553165197372437,
|
| 149 |
+
"learning_rate": 0.0005935625,
|
| 150 |
+
"loss": 3.5753,
|
| 151 |
+
"step": 19000
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 1.08,
|
| 155 |
+
"grad_norm": 0.33198973536491394,
|
| 156 |
+
"learning_rate": 0.0006248125,
|
| 157 |
+
"loss": 3.549,
|
| 158 |
+
"step": 20000
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.13,
|
| 162 |
+
"grad_norm": 0.30177804827690125,
|
| 163 |
+
"learning_rate": 0.000656,
|
| 164 |
+
"loss": 3.5373,
|
| 165 |
+
"step": 21000
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 1.18,
|
| 169 |
+
"grad_norm": 0.2844315767288208,
|
| 170 |
+
"learning_rate": 0.00068725,
|
| 171 |
+
"loss": 3.5339,
|
| 172 |
+
"step": 22000
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 1.24,
|
| 176 |
+
"grad_norm": 0.2813923954963684,
|
| 177 |
+
"learning_rate": 0.0007185000000000001,
|
| 178 |
+
"loss": 3.5241,
|
| 179 |
+
"step": 23000
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.29,
|
| 183 |
+
"grad_norm": 0.25613418221473694,
|
| 184 |
+
"learning_rate": 0.00074971875,
|
| 185 |
+
"loss": 3.5079,
|
| 186 |
+
"step": 24000
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 1.34,
|
| 190 |
+
"grad_norm": 0.26424020528793335,
|
| 191 |
+
"learning_rate": 0.00078096875,
|
| 192 |
+
"loss": 3.4997,
|
| 193 |
+
"step": 25000
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 1.4,
|
| 197 |
+
"grad_norm": 0.24137265980243683,
|
| 198 |
+
"learning_rate": 0.00081221875,
|
| 199 |
+
"loss": 3.4857,
|
| 200 |
+
"step": 26000
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 1.45,
|
| 204 |
+
"grad_norm": 0.23364698886871338,
|
| 205 |
+
"learning_rate": 0.00084346875,
|
| 206 |
+
"loss": 3.4751,
|
| 207 |
+
"step": 27000
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 1.51,
|
| 211 |
+
"grad_norm": 0.21730710566043854,
|
| 212 |
+
"learning_rate": 0.0008746874999999999,
|
| 213 |
+
"loss": 3.4681,
|
| 214 |
+
"step": 28000
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 1.56,
|
| 218 |
+
"grad_norm": 0.2167159467935562,
|
| 219 |
+
"learning_rate": 0.0009059375,
|
| 220 |
+
"loss": 3.4576,
|
| 221 |
+
"step": 29000
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.61,
|
| 225 |
+
"grad_norm": 0.2263418734073639,
|
| 226 |
+
"learning_rate": 0.0009371562500000001,
|
| 227 |
+
"loss": 3.4514,
|
| 228 |
+
"step": 30000
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 1.67,
|
| 232 |
+
"grad_norm": 0.2026355117559433,
|
| 233 |
+
"learning_rate": 0.0009684062500000001,
|
| 234 |
+
"loss": 3.4421,
|
| 235 |
+
"step": 31000
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 1.72,
|
| 239 |
+
"grad_norm": 0.2083364725112915,
|
| 240 |
+
"learning_rate": 0.0009996562500000001,
|
| 241 |
+
"loss": 3.4402,
|
| 242 |
+
"step": 32000
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.77,
|
| 246 |
+
"grad_norm": 0.1914134919643402,
|
| 247 |
+
"learning_rate": 0.0009970944594753558,
|
| 248 |
+
"loss": 3.4246,
|
| 249 |
+
"step": 33000
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 1.83,
|
| 253 |
+
"grad_norm": 0.2077130675315857,
|
| 254 |
+
"learning_rate": 0.0009941536289848254,
|
| 255 |
+
"loss": 3.4173,
|
| 256 |
+
"step": 34000
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 1.88,
|
| 260 |
+
"grad_norm": 0.19742459058761597,
|
| 261 |
+
"learning_rate": 0.0009912127984942948,
|
| 262 |
+
"loss": 3.4043,
|
| 263 |
+
"step": 35000
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 1.94,
|
| 267 |
+
"grad_norm": 0.18328115344047546,
|
| 268 |
+
"learning_rate": 0.000988274908834255,
|
| 269 |
+
"loss": 3.3924,
|
| 270 |
+
"step": 36000
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 1.99,
|
| 274 |
+
"grad_norm": 0.17804497480392456,
|
| 275 |
+
"learning_rate": 0.0009853340783437243,
|
| 276 |
+
"loss": 3.379,
|
| 277 |
+
"step": 37000
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 2.0,
|
| 281 |
+
"eval_accuracy": 0.3797682433009351,
|
| 282 |
+
"eval_loss": 3.5653116703033447,
|
| 283 |
+
"eval_runtime": 155.3698,
|
| 284 |
+
"eval_samples_per_second": 372.794,
|
| 285 |
+
"eval_steps_per_second": 5.831,
|
| 286 |
+
"step": 37204
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 2.04,
|
| 290 |
+
"grad_norm": 0.1852262169122696,
|
| 291 |
+
"learning_rate": 0.0009823932478531937,
|
| 292 |
+
"loss": 3.3362,
|
| 293 |
+
"step": 38000
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 2.1,
|
| 297 |
+
"grad_norm": 0.18120849132537842,
|
| 298 |
+
"learning_rate": 0.0009794524173626633,
|
| 299 |
+
"loss": 3.3286,
|
| 300 |
+
"step": 39000
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 2.15,
|
| 304 |
+
"grad_norm": 0.1793203055858612,
|
| 305 |
+
"learning_rate": 0.0009765145277026233,
|
| 306 |
+
"loss": 3.3231,
|
| 307 |
+
"step": 40000
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 2.2,
|
| 311 |
+
"grad_norm": 0.17847037315368652,
|
| 312 |
+
"learning_rate": 0.0009735736972120928,
|
| 313 |
+
"loss": 3.3184,
|
| 314 |
+
"step": 41000
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 2.26,
|
| 318 |
+
"grad_norm": 0.17294944822788239,
|
| 319 |
+
"learning_rate": 0.0009706358075520528,
|
| 320 |
+
"loss": 3.3086,
|
| 321 |
+
"step": 42000
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 2.31,
|
| 325 |
+
"grad_norm": 0.16597671806812286,
|
| 326 |
+
"learning_rate": 0.0009676949770615222,
|
| 327 |
+
"loss": 3.3067,
|
| 328 |
+
"step": 43000
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 2.37,
|
| 332 |
+
"grad_norm": 0.16891473531723022,
|
| 333 |
+
"learning_rate": 0.0009647570874014821,
|
| 334 |
+
"loss": 3.3,
|
| 335 |
+
"step": 44000
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 2.42,
|
| 339 |
+
"grad_norm": 0.17510074377059937,
|
| 340 |
+
"learning_rate": 0.0009618162569109516,
|
| 341 |
+
"loss": 3.3023,
|
| 342 |
+
"step": 45000
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 2.47,
|
| 346 |
+
"grad_norm": 0.17169316112995148,
|
| 347 |
+
"learning_rate": 0.0009588754264204211,
|
| 348 |
+
"loss": 3.295,
|
| 349 |
+
"step": 46000
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 2.53,
|
| 353 |
+
"grad_norm": 0.174598827958107,
|
| 354 |
+
"learning_rate": 0.0009559345959298906,
|
| 355 |
+
"loss": 3.2861,
|
| 356 |
+
"step": 47000
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.58,
|
| 360 |
+
"grad_norm": 0.18967097997665405,
|
| 361 |
+
"learning_rate": 0.0009529967062698506,
|
| 362 |
+
"loss": 3.2845,
|
| 363 |
+
"step": 48000
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 2.63,
|
| 367 |
+
"grad_norm": 0.1692247986793518,
|
| 368 |
+
"learning_rate": 0.0009500617574403012,
|
| 369 |
+
"loss": 3.2772,
|
| 370 |
+
"step": 49000
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 2.69,
|
| 374 |
+
"grad_norm": 0.1717061698436737,
|
| 375 |
+
"learning_rate": 0.0009471209269497706,
|
| 376 |
+
"loss": 3.2767,
|
| 377 |
+
"step": 50000
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 2.74,
|
| 381 |
+
"grad_norm": 0.18360279500484467,
|
| 382 |
+
"learning_rate": 0.0009441800964592401,
|
| 383 |
+
"loss": 3.2632,
|
| 384 |
+
"step": 51000
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 2.8,
|
| 388 |
+
"grad_norm": 0.19667457044124603,
|
| 389 |
+
"learning_rate": 0.0009412392659687096,
|
| 390 |
+
"loss": 3.2638,
|
| 391 |
+
"step": 52000
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 2.85,
|
| 395 |
+
"grad_norm": 0.16959011554718018,
|
| 396 |
+
"learning_rate": 0.0009382984354781791,
|
| 397 |
+
"loss": 3.2597,
|
| 398 |
+
"step": 53000
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 2.9,
|
| 402 |
+
"grad_norm": 0.19635465741157532,
|
| 403 |
+
"learning_rate": 0.0009353576049876486,
|
| 404 |
+
"loss": 3.2503,
|
| 405 |
+
"step": 54000
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 2.96,
|
| 409 |
+
"grad_norm": 0.17817912995815277,
|
| 410 |
+
"learning_rate": 0.0009324167744971181,
|
| 411 |
+
"loss": 3.2515,
|
| 412 |
+
"step": 55000
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 3.0,
|
| 416 |
+
"eval_accuracy": 0.39181961046231156,
|
| 417 |
+
"eval_loss": 3.4691600799560547,
|
| 418 |
+
"eval_runtime": 155.8093,
|
| 419 |
+
"eval_samples_per_second": 371.743,
|
| 420 |
+
"eval_steps_per_second": 5.815,
|
| 421 |
+
"step": 55806
|
| 422 |
+
},
|
| 423 |
+
{
|
| 424 |
+
"epoch": 3.01,
|
| 425 |
+
"grad_norm": 0.18061339855194092,
|
| 426 |
+
"learning_rate": 0.0009294759440065876,
|
| 427 |
+
"loss": 3.2344,
|
| 428 |
+
"step": 56000
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 3.06,
|
| 432 |
+
"grad_norm": 0.17083428800106049,
|
| 433 |
+
"learning_rate": 0.0009265380543465475,
|
| 434 |
+
"loss": 3.1818,
|
| 435 |
+
"step": 57000
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 3.12,
|
| 439 |
+
"grad_norm": 0.18623469769954681,
|
| 440 |
+
"learning_rate": 0.000923597223856017,
|
| 441 |
+
"loss": 3.1867,
|
| 442 |
+
"step": 58000
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 3.17,
|
| 446 |
+
"grad_norm": 0.16947512328624725,
|
| 447 |
+
"learning_rate": 0.0009206593341959769,
|
| 448 |
+
"loss": 3.1896,
|
| 449 |
+
"step": 59000
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 3.23,
|
| 453 |
+
"grad_norm": 0.18858687579631805,
|
| 454 |
+
"learning_rate": 0.0009177185037054464,
|
| 455 |
+
"loss": 3.1933,
|
| 456 |
+
"step": 60000
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 3.28,
|
| 460 |
+
"grad_norm": 0.1784023642539978,
|
| 461 |
+
"learning_rate": 0.0009147776732149159,
|
| 462 |
+
"loss": 3.1876,
|
| 463 |
+
"step": 61000
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 3.33,
|
| 467 |
+
"grad_norm": 0.18037760257720947,
|
| 468 |
+
"learning_rate": 0.0009118427243853664,
|
| 469 |
+
"loss": 3.1871,
|
| 470 |
+
"step": 62000
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 3.39,
|
| 474 |
+
"grad_norm": 0.1854478120803833,
|
| 475 |
+
"learning_rate": 0.0009089018938948359,
|
| 476 |
+
"loss": 3.1915,
|
| 477 |
+
"step": 63000
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 3.44,
|
| 481 |
+
"grad_norm": 0.17140689492225647,
|
| 482 |
+
"learning_rate": 0.0009059610634043054,
|
| 483 |
+
"loss": 3.1875,
|
| 484 |
+
"step": 64000
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 3.49,
|
| 488 |
+
"grad_norm": 0.18045330047607422,
|
| 489 |
+
"learning_rate": 0.0009030202329137749,
|
| 490 |
+
"loss": 3.1863,
|
| 491 |
+
"step": 65000
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 3.55,
|
| 495 |
+
"grad_norm": 0.1827496588230133,
|
| 496 |
+
"learning_rate": 0.0009000794024232444,
|
| 497 |
+
"loss": 3.1871,
|
| 498 |
+
"step": 66000
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 3.6,
|
| 502 |
+
"grad_norm": 0.20431484282016754,
|
| 503 |
+
"learning_rate": 0.0008971415127632044,
|
| 504 |
+
"loss": 3.1825,
|
| 505 |
+
"step": 67000
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 3.66,
|
| 509 |
+
"grad_norm": 0.17922812700271606,
|
| 510 |
+
"learning_rate": 0.0008942006822726738,
|
| 511 |
+
"loss": 3.181,
|
| 512 |
+
"step": 68000
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 3.71,
|
| 516 |
+
"grad_norm": 0.18560312688350677,
|
| 517 |
+
"learning_rate": 0.0008912627926126339,
|
| 518 |
+
"loss": 3.1767,
|
| 519 |
+
"step": 69000
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 3.76,
|
| 523 |
+
"grad_norm": 0.182236447930336,
|
| 524 |
+
"learning_rate": 0.0008883219621221033,
|
| 525 |
+
"loss": 3.1808,
|
| 526 |
+
"step": 70000
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 3.82,
|
| 530 |
+
"grad_norm": 0.1626577526330948,
|
| 531 |
+
"learning_rate": 0.0008853870132925539,
|
| 532 |
+
"loss": 3.1743,
|
| 533 |
+
"step": 71000
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 3.87,
|
| 537 |
+
"grad_norm": 0.1721421480178833,
|
| 538 |
+
"learning_rate": 0.0008824461828020232,
|
| 539 |
+
"loss": 3.1734,
|
| 540 |
+
"step": 72000
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 3.92,
|
| 544 |
+
"grad_norm": 0.17298929393291473,
|
| 545 |
+
"learning_rate": 0.0008795053523114927,
|
| 546 |
+
"loss": 3.1718,
|
| 547 |
+
"step": 73000
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 3.98,
|
| 551 |
+
"grad_norm": 0.19776953756809235,
|
| 552 |
+
"learning_rate": 0.0008765645218209622,
|
| 553 |
+
"loss": 3.1729,
|
| 554 |
+
"step": 74000
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 4.0,
|
| 558 |
+
"eval_accuracy": 0.3983427178451906,
|
| 559 |
+
"eval_loss": 3.419321060180664,
|
| 560 |
+
"eval_runtime": 154.6724,
|
| 561 |
+
"eval_samples_per_second": 374.475,
|
| 562 |
+
"eval_steps_per_second": 5.858,
|
| 563 |
+
"step": 74408
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 4.03,
|
| 567 |
+
"grad_norm": 0.19919687509536743,
|
| 568 |
+
"learning_rate": 0.0008736236913304317,
|
| 569 |
+
"loss": 3.1245,
|
| 570 |
+
"step": 75000
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 4.09,
|
| 574 |
+
"grad_norm": 0.18255846202373505,
|
| 575 |
+
"learning_rate": 0.0008706858016703917,
|
| 576 |
+
"loss": 3.1058,
|
| 577 |
+
"step": 76000
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 4.14,
|
| 581 |
+
"grad_norm": 0.22742678225040436,
|
| 582 |
+
"learning_rate": 0.0008677449711798612,
|
| 583 |
+
"loss": 3.108,
|
| 584 |
+
"step": 77000
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 4.19,
|
| 588 |
+
"grad_norm": 0.23507148027420044,
|
| 589 |
+
"learning_rate": 0.0008648041406893307,
|
| 590 |
+
"loss": 3.1162,
|
| 591 |
+
"step": 78000
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 4.25,
|
| 595 |
+
"grad_norm": 0.17946326732635498,
|
| 596 |
+
"learning_rate": 0.0008618662510292908,
|
| 597 |
+
"loss": 3.1139,
|
| 598 |
+
"step": 79000
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 4.3,
|
| 602 |
+
"grad_norm": 0.20196717977523804,
|
| 603 |
+
"learning_rate": 0.0008589254205387602,
|
| 604 |
+
"loss": 3.1171,
|
| 605 |
+
"step": 80000
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 4.35,
|
| 609 |
+
"grad_norm": 0.1969883292913437,
|
| 610 |
+
"learning_rate": 0.0008559845900482296,
|
| 611 |
+
"loss": 3.1167,
|
| 612 |
+
"step": 81000
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 4.41,
|
| 616 |
+
"grad_norm": 0.20422372221946716,
|
| 617 |
+
"learning_rate": 0.0008530437595576991,
|
| 618 |
+
"loss": 3.1144,
|
| 619 |
+
"step": 82000
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 4.46,
|
| 623 |
+
"grad_norm": 0.2068530172109604,
|
| 624 |
+
"learning_rate": 0.0008501088107281496,
|
| 625 |
+
"loss": 3.1169,
|
| 626 |
+
"step": 83000
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 4.52,
|
| 630 |
+
"grad_norm": 0.20425015687942505,
|
| 631 |
+
"learning_rate": 0.0008471679802376191,
|
| 632 |
+
"loss": 3.1188,
|
| 633 |
+
"step": 84000
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 4.57,
|
| 637 |
+
"grad_norm": 0.20677131414413452,
|
| 638 |
+
"learning_rate": 0.0008442271497470886,
|
| 639 |
+
"loss": 3.1181,
|
| 640 |
+
"step": 85000
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 4.62,
|
| 644 |
+
"grad_norm": 0.1946183443069458,
|
| 645 |
+
"learning_rate": 0.000841286319256558,
|
| 646 |
+
"loss": 3.1165,
|
| 647 |
+
"step": 86000
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 4.68,
|
| 651 |
+
"grad_norm": 0.19427254796028137,
|
| 652 |
+
"learning_rate": 0.000838348429596518,
|
| 653 |
+
"loss": 3.1205,
|
| 654 |
+
"step": 87000
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 4.73,
|
| 658 |
+
"grad_norm": 0.2612309455871582,
|
| 659 |
+
"learning_rate": 0.0008354075991059875,
|
| 660 |
+
"loss": 3.1168,
|
| 661 |
+
"step": 88000
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 4.78,
|
| 665 |
+
"grad_norm": 0.19651499390602112,
|
| 666 |
+
"learning_rate": 0.0008324697094459476,
|
| 667 |
+
"loss": 3.1181,
|
| 668 |
+
"step": 89000
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 4.84,
|
| 672 |
+
"grad_norm": 0.21289727091789246,
|
| 673 |
+
"learning_rate": 0.0008295288789554171,
|
| 674 |
+
"loss": 3.1144,
|
| 675 |
+
"step": 90000
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 4.89,
|
| 679 |
+
"grad_norm": 0.18825019896030426,
|
| 680 |
+
"learning_rate": 0.000826590989295377,
|
| 681 |
+
"loss": 3.1141,
|
| 682 |
+
"step": 91000
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 4.95,
|
| 686 |
+
"grad_norm": 0.20688407123088837,
|
| 687 |
+
"learning_rate": 0.0008236501588048465,
|
| 688 |
+
"loss": 3.114,
|
| 689 |
+
"step": 92000
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 5.0,
|
| 693 |
+
"grad_norm": 0.18017753958702087,
|
| 694 |
+
"learning_rate": 0.000820709328314316,
|
| 695 |
+
"loss": 3.1139,
|
| 696 |
+
"step": 93000
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 5.0,
|
| 700 |
+
"eval_accuracy": 0.40260902798169274,
|
| 701 |
+
"eval_loss": 3.3907480239868164,
|
| 702 |
+
"eval_runtime": 155.5488,
|
| 703 |
+
"eval_samples_per_second": 372.365,
|
| 704 |
+
"eval_steps_per_second": 5.825,
|
| 705 |
+
"step": 93010
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 5.05,
|
| 709 |
+
"grad_norm": 0.2159438282251358,
|
| 710 |
+
"learning_rate": 0.0008177714386542759,
|
| 711 |
+
"loss": 3.0438,
|
| 712 |
+
"step": 94000
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 5.11,
|
| 716 |
+
"grad_norm": 0.1938624382019043,
|
| 717 |
+
"learning_rate": 0.0008148306081637454,
|
| 718 |
+
"loss": 3.0471,
|
| 719 |
+
"step": 95000
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 5.16,
|
| 723 |
+
"grad_norm": 0.1948939859867096,
|
| 724 |
+
"learning_rate": 0.0008118897776732149,
|
| 725 |
+
"loss": 3.0519,
|
| 726 |
+
"step": 96000
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 5.21,
|
| 730 |
+
"grad_norm": 0.21813590824604034,
|
| 731 |
+
"learning_rate": 0.000808951888013175,
|
| 732 |
+
"loss": 3.0615,
|
| 733 |
+
"step": 97000
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 5.27,
|
| 737 |
+
"grad_norm": 0.25585538148880005,
|
| 738 |
+
"learning_rate": 0.0008060110575226443,
|
| 739 |
+
"loss": 3.0615,
|
| 740 |
+
"step": 98000
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 5.32,
|
| 744 |
+
"grad_norm": 0.2105061262845993,
|
| 745 |
+
"learning_rate": 0.0008030702270321138,
|
| 746 |
+
"loss": 3.0617,
|
| 747 |
+
"step": 99000
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 5.38,
|
| 751 |
+
"grad_norm": 0.18835408985614777,
|
| 752 |
+
"learning_rate": 0.0008001293965415833,
|
| 753 |
+
"loss": 3.0677,
|
| 754 |
+
"step": 100000
|
| 755 |
+
},
|
| 756 |
+
{
|
| 757 |
+
"epoch": 5.43,
|
| 758 |
+
"grad_norm": 0.1814804971218109,
|
| 759 |
+
"learning_rate": 0.0007971915068815434,
|
| 760 |
+
"loss": 3.07,
|
| 761 |
+
"step": 101000
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 5.48,
|
| 765 |
+
"grad_norm": 0.2165517359972,
|
| 766 |
+
"learning_rate": 0.0007942536172215034,
|
| 767 |
+
"loss": 3.0654,
|
| 768 |
+
"step": 102000
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 5.54,
|
| 772 |
+
"grad_norm": 0.21530425548553467,
|
| 773 |
+
"learning_rate": 0.0007913127867309729,
|
| 774 |
+
"loss": 3.0686,
|
| 775 |
+
"step": 103000
|
| 776 |
+
},
|
| 777 |
+
{
|
| 778 |
+
"epoch": 5.59,
|
| 779 |
+
"grad_norm": NaN,
|
| 780 |
+
"learning_rate": 0.0007883748970709328,
|
| 781 |
+
"loss": 3.0676,
|
| 782 |
+
"step": 104000
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 5.64,
|
| 786 |
+
"grad_norm": 0.20055846869945526,
|
| 787 |
+
"learning_rate": 0.0007854340665804023,
|
| 788 |
+
"loss": 3.0705,
|
| 789 |
+
"step": 105000
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 5.7,
|
| 793 |
+
"grad_norm": 0.19475945830345154,
|
| 794 |
+
"learning_rate": 0.0007824932360898718,
|
| 795 |
+
"loss": 3.0684,
|
| 796 |
+
"step": 106000
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"epoch": 5.75,
|
| 800 |
+
"grad_norm": 0.2444075047969818,
|
| 801 |
+
"learning_rate": 0.0007795553464298318,
|
| 802 |
+
"loss": 3.0644,
|
| 803 |
+
"step": 107000
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 5.81,
|
| 807 |
+
"grad_norm": 0.3142812252044678,
|
| 808 |
+
"learning_rate": 0.0007766145159393013,
|
| 809 |
+
"loss": 3.0701,
|
| 810 |
+
"step": 108000
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 5.86,
|
| 814 |
+
"grad_norm": 0.1972387135028839,
|
| 815 |
+
"learning_rate": 0.0007736736854487707,
|
| 816 |
+
"loss": 3.0648,
|
| 817 |
+
"step": 109000
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 5.91,
|
| 821 |
+
"grad_norm": 0.18938925862312317,
|
| 822 |
+
"learning_rate": 0.0007707328549582402,
|
| 823 |
+
"loss": 3.0686,
|
| 824 |
+
"step": 110000
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 5.97,
|
| 828 |
+
"grad_norm": 0.22070257365703583,
|
| 829 |
+
"learning_rate": 0.0007677949652982002,
|
| 830 |
+
"loss": 3.0709,
|
| 831 |
+
"step": 111000
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 6.0,
|
| 835 |
+
"eval_accuracy": 0.40427472089482946,
|
| 836 |
+
"eval_loss": 3.3642256259918213,
|
| 837 |
+
"eval_runtime": 154.6657,
|
| 838 |
+
"eval_samples_per_second": 374.492,
|
| 839 |
+
"eval_steps_per_second": 5.858,
|
| 840 |
+
"step": 111612
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 6.02,
|
| 844 |
+
"grad_norm": 0.20117591321468353,
|
| 845 |
+
"learning_rate": 0.0007648541348076697,
|
| 846 |
+
"loss": 3.0385,
|
| 847 |
+
"step": 112000
|
| 848 |
+
},
|
| 849 |
+
{
|
| 850 |
+
"epoch": 6.07,
|
| 851 |
+
"grad_norm": 0.2254948914051056,
|
| 852 |
+
"learning_rate": 0.0007619133043171391,
|
| 853 |
+
"loss": 2.9981,
|
| 854 |
+
"step": 113000
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 6.13,
|
| 858 |
+
"grad_norm": 0.21827402710914612,
|
| 859 |
+
"learning_rate": 0.0007589754146570992,
|
| 860 |
+
"loss": 3.007,
|
| 861 |
+
"step": 114000
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 6.18,
|
| 865 |
+
"grad_norm": 0.2094143182039261,
|
| 866 |
+
"learning_rate": 0.0007560345841665687,
|
| 867 |
+
"loss": 3.0119,
|
| 868 |
+
"step": 115000
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 6.24,
|
| 872 |
+
"grad_norm": 0.23713943362236023,
|
| 873 |
+
"learning_rate": 0.0007530937536760382,
|
| 874 |
+
"loss": 3.0158,
|
| 875 |
+
"step": 116000
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 6.29,
|
| 879 |
+
"grad_norm": 0.21203161776065826,
|
| 880 |
+
"learning_rate": 0.0007501558640159981,
|
| 881 |
+
"loss": 3.0179,
|
| 882 |
+
"step": 117000
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 6.34,
|
| 886 |
+
"grad_norm": 0.223785862326622,
|
| 887 |
+
"learning_rate": 0.0007472150335254676,
|
| 888 |
+
"loss": 3.0211,
|
| 889 |
+
"step": 118000
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 6.4,
|
| 893 |
+
"grad_norm": 0.3047547936439514,
|
| 894 |
+
"learning_rate": 0.0007442742030349371,
|
| 895 |
+
"loss": 3.0232,
|
| 896 |
+
"step": 119000
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 6.45,
|
| 900 |
+
"grad_norm": 0.22039389610290527,
|
| 901 |
+
"learning_rate": 0.0007413333725444066,
|
| 902 |
+
"loss": 3.0261,
|
| 903 |
+
"step": 120000
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 6.5,
|
| 907 |
+
"grad_norm": 0.22225935757160187,
|
| 908 |
+
"learning_rate": 0.0007383954828843665,
|
| 909 |
+
"loss": 3.0268,
|
| 910 |
+
"step": 121000
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 6.56,
|
| 914 |
+
"grad_norm": 0.30233263969421387,
|
| 915 |
+
"learning_rate": 0.000735454652393836,
|
| 916 |
+
"loss": 3.0299,
|
| 917 |
+
"step": 122000
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 6.61,
|
| 921 |
+
"grad_norm": 0.23030564188957214,
|
| 922 |
+
"learning_rate": 0.0007325138219033055,
|
| 923 |
+
"loss": 3.0301,
|
| 924 |
+
"step": 123000
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 6.67,
|
| 928 |
+
"grad_norm": 0.2047109752893448,
|
| 929 |
+
"learning_rate": 0.0007295759322432654,
|
| 930 |
+
"loss": 3.027,
|
| 931 |
+
"step": 124000
|
| 932 |
+
},
|
| 933 |
+
{
|
| 934 |
+
"epoch": 6.72,
|
| 935 |
+
"grad_norm": 0.21085889637470245,
|
| 936 |
+
"learning_rate": 0.0007266380425832255,
|
| 937 |
+
"loss": 3.0278,
|
| 938 |
+
"step": 125000
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 6.77,
|
| 942 |
+
"grad_norm": 0.2596847712993622,
|
| 943 |
+
"learning_rate": 0.000723697212092695,
|
| 944 |
+
"loss": 3.0358,
|
| 945 |
+
"step": 126000
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 6.83,
|
| 949 |
+
"grad_norm": 0.1978030651807785,
|
| 950 |
+
"learning_rate": 0.0007207563816021645,
|
| 951 |
+
"loss": 3.0315,
|
| 952 |
+
"step": 127000
|
| 953 |
+
},
|
| 954 |
+
{
|
| 955 |
+
"epoch": 6.88,
|
| 956 |
+
"grad_norm": 0.21782337129116058,
|
| 957 |
+
"learning_rate": 0.000717815551111634,
|
| 958 |
+
"loss": 3.028,
|
| 959 |
+
"step": 128000
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 6.93,
|
| 963 |
+
"grad_norm": 0.2578461766242981,
|
| 964 |
+
"learning_rate": 0.000714877661451594,
|
| 965 |
+
"loss": 3.0329,
|
| 966 |
+
"step": 129000
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 6.99,
|
| 970 |
+
"grad_norm": 0.20211657881736755,
|
| 971 |
+
"learning_rate": 0.0007119368309610634,
|
| 972 |
+
"loss": 3.0297,
|
| 973 |
+
"step": 130000
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 7.0,
|
| 977 |
+
"eval_accuracy": 0.40667091369655045,
|
| 978 |
+
"eval_loss": 3.3544673919677734,
|
| 979 |
+
"eval_runtime": 155.1034,
|
| 980 |
+
"eval_samples_per_second": 373.435,
|
| 981 |
+
"eval_steps_per_second": 5.841,
|
| 982 |
+
"step": 130214
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 7.04,
|
| 986 |
+
"grad_norm": 0.2451675981283188,
|
| 987 |
+
"learning_rate": 0.0007089989413010234,
|
| 988 |
+
"loss": 2.9744,
|
| 989 |
+
"step": 131000
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 7.1,
|
| 993 |
+
"grad_norm": 0.26030394434928894,
|
| 994 |
+
"learning_rate": 0.0007060581108104929,
|
| 995 |
+
"loss": 2.9677,
|
| 996 |
+
"step": 132000
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 7.15,
|
| 1000 |
+
"grad_norm": 0.22699125111103058,
|
| 1001 |
+
"learning_rate": 0.0007031172803199624,
|
| 1002 |
+
"loss": 2.9753,
|
| 1003 |
+
"step": 133000
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 7.2,
|
| 1007 |
+
"grad_norm": 0.22514161467552185,
|
| 1008 |
+
"learning_rate": 0.0007001764498294319,
|
| 1009 |
+
"loss": 2.974,
|
| 1010 |
+
"step": 134000
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 7.26,
|
| 1014 |
+
"grad_norm": 0.23178595304489136,
|
| 1015 |
+
"learning_rate": 0.0006972356193389014,
|
| 1016 |
+
"loss": 2.9784,
|
| 1017 |
+
"step": 135000
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 7.31,
|
| 1021 |
+
"grad_norm": 0.27629485726356506,
|
| 1022 |
+
"learning_rate": 0.0006942977296788612,
|
| 1023 |
+
"loss": 2.9856,
|
| 1024 |
+
"step": 136000
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 7.36,
|
| 1028 |
+
"grad_norm": 0.2211887091398239,
|
| 1029 |
+
"learning_rate": 0.0006913598400188213,
|
| 1030 |
+
"loss": 2.9859,
|
| 1031 |
+
"step": 137000
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 7.42,
|
| 1035 |
+
"grad_norm": 0.2578117847442627,
|
| 1036 |
+
"learning_rate": 0.0006884190095282908,
|
| 1037 |
+
"loss": 2.985,
|
| 1038 |
+
"step": 138000
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 7.47,
|
| 1042 |
+
"grad_norm": 0.25054463744163513,
|
| 1043 |
+
"learning_rate": 0.0006854781790377603,
|
| 1044 |
+
"loss": 2.9902,
|
| 1045 |
+
"step": 139000
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 7.53,
|
| 1049 |
+
"grad_norm": 0.21288973093032837,
|
| 1050 |
+
"learning_rate": 0.0006825402893777203,
|
| 1051 |
+
"loss": 2.9922,
|
| 1052 |
+
"step": 140000
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 7.58,
|
| 1056 |
+
"grad_norm": 0.21266783773899078,
|
| 1057 |
+
"learning_rate": 0.0006795994588871898,
|
| 1058 |
+
"loss": 2.994,
|
| 1059 |
+
"step": 141000
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 7.63,
|
| 1063 |
+
"grad_norm": 0.26795753836631775,
|
| 1064 |
+
"learning_rate": 0.0006766615692271497,
|
| 1065 |
+
"loss": 2.9945,
|
| 1066 |
+
"step": 142000
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 7.69,
|
| 1070 |
+
"grad_norm": 0.2051943689584732,
|
| 1071 |
+
"learning_rate": 0.0006737207387366192,
|
| 1072 |
+
"loss": 2.994,
|
| 1073 |
+
"step": 143000
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 7.74,
|
| 1077 |
+
"grad_norm": 0.2313968390226364,
|
| 1078 |
+
"learning_rate": 0.0006707799082460887,
|
| 1079 |
+
"loss": 2.9988,
|
| 1080 |
+
"step": 144000
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 7.79,
|
| 1084 |
+
"grad_norm": 0.24470819532871246,
|
| 1085 |
+
"learning_rate": 0.0006678390777555582,
|
| 1086 |
+
"loss": 2.998,
|
| 1087 |
+
"step": 145000
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 7.85,
|
| 1091 |
+
"grad_norm": 0.203665092587471,
|
| 1092 |
+
"learning_rate": 0.0006649011880955182,
|
| 1093 |
+
"loss": 2.9962,
|
| 1094 |
+
"step": 146000
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 7.9,
|
| 1098 |
+
"grad_norm": 0.21225391328334808,
|
| 1099 |
+
"learning_rate": 0.0006619603576049877,
|
| 1100 |
+
"loss": 2.9975,
|
| 1101 |
+
"step": 147000
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 7.96,
|
| 1105 |
+
"grad_norm": 0.25866907835006714,
|
| 1106 |
+
"learning_rate": 0.0006590195271144572,
|
| 1107 |
+
"loss": 2.9988,
|
| 1108 |
+
"step": 148000
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 8.0,
|
| 1112 |
+
"eval_accuracy": 0.4079528153770697,
|
| 1113 |
+
"eval_loss": 3.359600782394409,
|
| 1114 |
+
"eval_runtime": 154.5906,
|
| 1115 |
+
"eval_samples_per_second": 374.673,
|
| 1116 |
+
"eval_steps_per_second": 5.861,
|
| 1117 |
+
"step": 148816
|
| 1118 |
+
},
|
| 1119 |
+
{
|
| 1120 |
+
"epoch": 8.01,
|
| 1121 |
+
"grad_norm": 0.25523641705513,
|
| 1122 |
+
"learning_rate": 0.0006560816374544171,
|
| 1123 |
+
"loss": 2.9863,
|
| 1124 |
+
"step": 149000
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 8.06,
|
| 1128 |
+
"grad_norm": 0.22232286632061005,
|
| 1129 |
+
"learning_rate": 0.0006531408069638866,
|
| 1130 |
+
"loss": 2.9316,
|
| 1131 |
+
"step": 150000
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 8.12,
|
| 1135 |
+
"grad_norm": 0.22783271968364716,
|
| 1136 |
+
"learning_rate": 0.000650199976473356,
|
| 1137 |
+
"loss": 2.9369,
|
| 1138 |
+
"step": 151000
|
| 1139 |
+
},
|
| 1140 |
+
{
|
| 1141 |
+
"epoch": 8.17,
|
| 1142 |
+
"grad_norm": 0.21875810623168945,
|
| 1143 |
+
"learning_rate": 0.0006472591459828255,
|
| 1144 |
+
"loss": 2.9418,
|
| 1145 |
+
"step": 152000
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 8.22,
|
| 1149 |
+
"grad_norm": 0.2610374093055725,
|
| 1150 |
+
"learning_rate": 0.0006443212563227856,
|
| 1151 |
+
"loss": 2.9468,
|
| 1152 |
+
"step": 153000
|
| 1153 |
+
},
|
| 1154 |
+
{
|
| 1155 |
+
"epoch": 8.28,
|
| 1156 |
+
"grad_norm": 0.2271444946527481,
|
| 1157 |
+
"learning_rate": 0.0006413804258322551,
|
| 1158 |
+
"loss": 2.9532,
|
| 1159 |
+
"step": 154000
|
| 1160 |
+
},
|
| 1161 |
+
{
|
| 1162 |
+
"epoch": 8.33,
|
| 1163 |
+
"grad_norm": 0.23558945953845978,
|
| 1164 |
+
"learning_rate": 0.000638442536172215,
|
| 1165 |
+
"loss": 2.9513,
|
| 1166 |
+
"step": 155000
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 8.39,
|
| 1170 |
+
"grad_norm": 0.23239487409591675,
|
| 1171 |
+
"learning_rate": 0.0006355046465121751,
|
| 1172 |
+
"loss": 2.9544,
|
| 1173 |
+
"step": 156000
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 8.44,
|
| 1177 |
+
"grad_norm": 0.2163064181804657,
|
| 1178 |
+
"learning_rate": 0.0006325638160216445,
|
| 1179 |
+
"loss": 2.9594,
|
| 1180 |
+
"step": 157000
|
| 1181 |
+
},
|
| 1182 |
+
{
|
| 1183 |
+
"epoch": 8.49,
|
| 1184 |
+
"grad_norm": 0.25110113620758057,
|
| 1185 |
+
"learning_rate": 0.000629622985531114,
|
| 1186 |
+
"loss": 2.9587,
|
| 1187 |
+
"step": 158000
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 8.55,
|
| 1191 |
+
"grad_norm": 0.23023970425128937,
|
| 1192 |
+
"learning_rate": 0.0006266850958710741,
|
| 1193 |
+
"loss": 2.9635,
|
| 1194 |
+
"step": 159000
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 8.6,
|
| 1198 |
+
"grad_norm": 0.2888554036617279,
|
| 1199 |
+
"learning_rate": 0.0006237442653805435,
|
| 1200 |
+
"loss": 2.9645,
|
| 1201 |
+
"step": 160000
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 8.65,
|
| 1205 |
+
"grad_norm": 0.22880807518959045,
|
| 1206 |
+
"learning_rate": 0.000620803434890013,
|
| 1207 |
+
"loss": 2.9623,
|
| 1208 |
+
"step": 161000
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 8.71,
|
| 1212 |
+
"grad_norm": 0.25806036591529846,
|
| 1213 |
+
"learning_rate": 0.0006178626043994823,
|
| 1214 |
+
"loss": 2.971,
|
| 1215 |
+
"step": 162000
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 8.76,
|
| 1219 |
+
"grad_norm": 0.25819092988967896,
|
| 1220 |
+
"learning_rate": 0.0006149247147394424,
|
| 1221 |
+
"loss": 2.9689,
|
| 1222 |
+
"step": 163000
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"epoch": 8.82,
|
| 1226 |
+
"grad_norm": 0.2252822071313858,
|
| 1227 |
+
"learning_rate": 0.0006119838842489119,
|
| 1228 |
+
"loss": 2.9672,
|
| 1229 |
+
"step": 164000
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 8.87,
|
| 1233 |
+
"grad_norm": 0.22955825924873352,
|
| 1234 |
+
"learning_rate": 0.0006090489354193625,
|
| 1235 |
+
"loss": 2.9666,
|
| 1236 |
+
"step": 165000
|
| 1237 |
+
},
|
| 1238 |
+
{
|
| 1239 |
+
"epoch": 8.92,
|
| 1240 |
+
"grad_norm": 0.33369356393814087,
|
| 1241 |
+
"learning_rate": 0.000606108104928832,
|
| 1242 |
+
"loss": 2.9677,
|
| 1243 |
+
"step": 166000
|
| 1244 |
+
},
|
| 1245 |
+
{
|
| 1246 |
+
"epoch": 8.98,
|
| 1247 |
+
"grad_norm": 0.2391689568758011,
|
| 1248 |
+
"learning_rate": 0.0006031672744383015,
|
| 1249 |
+
"loss": 2.9717,
|
| 1250 |
+
"step": 167000
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 9.0,
|
| 1254 |
+
"eval_accuracy": 0.40866631908468914,
|
| 1255 |
+
"eval_loss": 3.3723206520080566,
|
| 1256 |
+
"eval_runtime": 155.2809,
|
| 1257 |
+
"eval_samples_per_second": 373.008,
|
| 1258 |
+
"eval_steps_per_second": 5.835,
|
| 1259 |
+
"step": 167418
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 9.03,
|
| 1263 |
+
"grad_norm": 0.22217118740081787,
|
| 1264 |
+
"learning_rate": 0.0006002293847782614,
|
| 1265 |
+
"loss": 2.9298,
|
| 1266 |
+
"step": 168000
|
| 1267 |
+
},
|
| 1268 |
+
{
|
| 1269 |
+
"epoch": 9.09,
|
| 1270 |
+
"grad_norm": 0.249114990234375,
|
| 1271 |
+
"learning_rate": 0.0005972885542877309,
|
| 1272 |
+
"loss": 2.9048,
|
| 1273 |
+
"step": 169000
|
| 1274 |
+
},
|
| 1275 |
+
{
|
| 1276 |
+
"epoch": 9.14,
|
| 1277 |
+
"grad_norm": 0.28138816356658936,
|
| 1278 |
+
"learning_rate": 0.0005943506646276909,
|
| 1279 |
+
"loss": 2.9122,
|
| 1280 |
+
"step": 170000
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 9.19,
|
| 1284 |
+
"grad_norm": 0.26497504115104675,
|
| 1285 |
+
"learning_rate": 0.0005914098341371603,
|
| 1286 |
+
"loss": 2.9165,
|
| 1287 |
+
"step": 171000
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 9.25,
|
| 1291 |
+
"grad_norm": 0.33365172147750854,
|
| 1292 |
+
"learning_rate": 0.0005884690036466298,
|
| 1293 |
+
"loss": 2.9174,
|
| 1294 |
+
"step": 172000
|
| 1295 |
+
},
|
| 1296 |
+
{
|
| 1297 |
+
"epoch": 9.3,
|
| 1298 |
+
"grad_norm": 0.257129043340683,
|
| 1299 |
+
"learning_rate": 0.0005855281731560992,
|
| 1300 |
+
"loss": 2.924,
|
| 1301 |
+
"step": 173000
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 9.35,
|
| 1305 |
+
"grad_norm": 0.29947561025619507,
|
| 1306 |
+
"learning_rate": 0.0005825902834960592,
|
| 1307 |
+
"loss": 2.9282,
|
| 1308 |
+
"step": 174000
|
| 1309 |
+
},
|
| 1310 |
+
{
|
| 1311 |
+
"epoch": 9.41,
|
| 1312 |
+
"grad_norm": 0.23127929866313934,
|
| 1313 |
+
"learning_rate": 0.0005796494530055287,
|
| 1314 |
+
"loss": 2.9272,
|
| 1315 |
+
"step": 175000
|
| 1316 |
+
},
|
| 1317 |
+
{
|
| 1318 |
+
"epoch": 9.46,
|
| 1319 |
+
"grad_norm": 0.2609017491340637,
|
| 1320 |
+
"learning_rate": 0.0005767086225149982,
|
| 1321 |
+
"loss": 2.9311,
|
| 1322 |
+
"step": 176000
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 9.52,
|
| 1326 |
+
"grad_norm": NaN,
|
| 1327 |
+
"learning_rate": 0.0005737707328549583,
|
| 1328 |
+
"loss": 2.9321,
|
| 1329 |
+
"step": 177000
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 9.57,
|
| 1333 |
+
"grad_norm": 0.27535927295684814,
|
| 1334 |
+
"learning_rate": 0.0005708299023644278,
|
| 1335 |
+
"loss": 2.9343,
|
| 1336 |
+
"step": 178000
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"epoch": 9.62,
|
| 1340 |
+
"grad_norm": 0.2315855473279953,
|
| 1341 |
+
"learning_rate": 0.0005678890718738972,
|
| 1342 |
+
"loss": 2.9342,
|
| 1343 |
+
"step": 179000
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 9.68,
|
| 1347 |
+
"grad_norm": 0.2345881462097168,
|
| 1348 |
+
"learning_rate": 0.0005649482413833666,
|
| 1349 |
+
"loss": 2.9377,
|
| 1350 |
+
"step": 180000
|
| 1351 |
+
},
|
| 1352 |
+
{
|
| 1353 |
+
"epoch": 9.73,
|
| 1354 |
+
"grad_norm": 0.36270418763160706,
|
| 1355 |
+
"learning_rate": 0.0005620103517233267,
|
| 1356 |
+
"loss": 2.9415,
|
| 1357 |
+
"step": 181000
|
| 1358 |
+
},
|
| 1359 |
+
{
|
| 1360 |
+
"epoch": 9.78,
|
| 1361 |
+
"grad_norm": 0.22433358430862427,
|
| 1362 |
+
"learning_rate": 0.0005590695212327962,
|
| 1363 |
+
"loss": 2.9433,
|
| 1364 |
+
"step": 182000
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 9.84,
|
| 1368 |
+
"grad_norm": 0.23198537528514862,
|
| 1369 |
+
"learning_rate": 0.0005561286907422656,
|
| 1370 |
+
"loss": 2.9435,
|
| 1371 |
+
"step": 183000
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 9.89,
|
| 1375 |
+
"grad_norm": 0.3433922827243805,
|
| 1376 |
+
"learning_rate": 0.0005531908010822257,
|
| 1377 |
+
"loss": 2.9422,
|
| 1378 |
+
"step": 184000
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 9.95,
|
| 1382 |
+
"grad_norm": 0.2841480076313019,
|
| 1383 |
+
"learning_rate": 0.0005502499705916952,
|
| 1384 |
+
"loss": 2.9457,
|
| 1385 |
+
"step": 185000
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 10.0,
|
| 1389 |
+
"grad_norm": 0.2232799530029297,
|
| 1390 |
+
"learning_rate": 0.0005473091401011647,
|
| 1391 |
+
"loss": 2.9432,
|
| 1392 |
+
"step": 186000
|
| 1393 |
+
},
|
| 1394 |
+
{
|
| 1395 |
+
"epoch": 10.0,
|
| 1396 |
+
"eval_accuracy": 0.4092650817761506,
|
| 1397 |
+
"eval_loss": 3.357872486114502,
|
| 1398 |
+
"eval_runtime": 155.1007,
|
| 1399 |
+
"eval_samples_per_second": 373.441,
|
| 1400 |
+
"eval_steps_per_second": 5.841,
|
| 1401 |
+
"step": 186020
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 10.05,
|
| 1405 |
+
"grad_norm": 0.2610397934913635,
|
| 1406 |
+
"learning_rate": 0.0005443712504411245,
|
| 1407 |
+
"loss": 2.8782,
|
| 1408 |
+
"step": 187000
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"epoch": 10.11,
|
| 1412 |
+
"grad_norm": 0.3216498792171478,
|
| 1413 |
+
"learning_rate": 0.000541430419950594,
|
| 1414 |
+
"loss": 2.8815,
|
| 1415 |
+
"step": 188000
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 10.16,
|
| 1419 |
+
"grad_norm": 0.2474890649318695,
|
| 1420 |
+
"learning_rate": 0.0005384895894600635,
|
| 1421 |
+
"loss": 2.8881,
|
| 1422 |
+
"step": 189000
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 10.21,
|
| 1426 |
+
"grad_norm": 0.35237178206443787,
|
| 1427 |
+
"learning_rate": 0.000535548758969533,
|
| 1428 |
+
"loss": 2.8933,
|
| 1429 |
+
"step": 190000
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 10.27,
|
| 1433 |
+
"grad_norm": 0.3191303610801697,
|
| 1434 |
+
"learning_rate": 0.000532610869309493,
|
| 1435 |
+
"loss": 2.8989,
|
| 1436 |
+
"step": 191000
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 10.32,
|
| 1440 |
+
"grad_norm": 0.2325054258108139,
|
| 1441 |
+
"learning_rate": 0.0005296700388189625,
|
| 1442 |
+
"loss": 2.8978,
|
| 1443 |
+
"step": 192000
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 10.38,
|
| 1447 |
+
"grad_norm": 0.24588368833065033,
|
| 1448 |
+
"learning_rate": 0.0005267321491589225,
|
| 1449 |
+
"loss": 2.9019,
|
| 1450 |
+
"step": 193000
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 10.43,
|
| 1454 |
+
"grad_norm": 0.2485446035861969,
|
| 1455 |
+
"learning_rate": 0.0005237913186683919,
|
| 1456 |
+
"loss": 2.9035,
|
| 1457 |
+
"step": 194000
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 10.48,
|
| 1461 |
+
"grad_norm": 0.24470220506191254,
|
| 1462 |
+
"learning_rate": 0.000520853429008352,
|
| 1463 |
+
"loss": 2.9077,
|
| 1464 |
+
"step": 195000
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 10.54,
|
| 1468 |
+
"grad_norm": 0.248806893825531,
|
| 1469 |
+
"learning_rate": 0.0005179125985178215,
|
| 1470 |
+
"loss": 2.9089,
|
| 1471 |
+
"step": 196000
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 10.59,
|
| 1475 |
+
"grad_norm": 0.22887374460697174,
|
| 1476 |
+
"learning_rate": 0.000514971768027291,
|
| 1477 |
+
"loss": 2.908,
|
| 1478 |
+
"step": 197000
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 10.64,
|
| 1482 |
+
"grad_norm": 0.2529465854167938,
|
| 1483 |
+
"learning_rate": 0.0005120309375367605,
|
| 1484 |
+
"loss": 2.9127,
|
| 1485 |
+
"step": 198000
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 10.7,
|
| 1489 |
+
"grad_norm": 0.25075119733810425,
|
| 1490 |
+
"learning_rate": 0.0005090901070462299,
|
| 1491 |
+
"loss": 2.911,
|
| 1492 |
+
"step": 199000
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"epoch": 10.75,
|
| 1496 |
+
"grad_norm": 0.23723989725112915,
|
| 1497 |
+
"learning_rate": 0.0005061522173861898,
|
| 1498 |
+
"loss": 2.9152,
|
| 1499 |
+
"step": 200000
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 10.81,
|
| 1503 |
+
"grad_norm": 0.25637319684028625,
|
| 1504 |
+
"learning_rate": 0.0005032143277261498,
|
| 1505 |
+
"loss": 2.9149,
|
| 1506 |
+
"step": 201000
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 10.86,
|
| 1510 |
+
"grad_norm": 0.28713148832321167,
|
| 1511 |
+
"learning_rate": 0.0005002734972356193,
|
| 1512 |
+
"loss": 2.9178,
|
| 1513 |
+
"step": 202000
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 10.91,
|
| 1517 |
+
"grad_norm": 0.27700188755989075,
|
| 1518 |
+
"learning_rate": 0.0004973356075755794,
|
| 1519 |
+
"loss": 2.9214,
|
| 1520 |
+
"step": 203000
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 10.97,
|
| 1524 |
+
"grad_norm": 0.33580395579338074,
|
| 1525 |
+
"learning_rate": 0.0004943947770850489,
|
| 1526 |
+
"loss": 2.9217,
|
| 1527 |
+
"step": 204000
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 11.0,
|
| 1531 |
+
"eval_accuracy": 0.4097510516302999,
|
| 1532 |
+
"eval_loss": 3.370105028152466,
|
| 1533 |
+
"eval_runtime": 155.2297,
|
| 1534 |
+
"eval_samples_per_second": 373.131,
|
| 1535 |
+
"eval_steps_per_second": 5.837,
|
| 1536 |
+
"step": 204622
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 11.02,
|
| 1540 |
+
"grad_norm": 0.23724067211151123,
|
| 1541 |
+
"learning_rate": 0.0004914539465945184,
|
| 1542 |
+
"loss": 2.8953,
|
| 1543 |
+
"step": 205000
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 11.07,
|
| 1547 |
+
"grad_norm": 0.33235830068588257,
|
| 1548 |
+
"learning_rate": 0.0004885160569344783,
|
| 1549 |
+
"loss": 2.8533,
|
| 1550 |
+
"step": 206000
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 11.13,
|
| 1554 |
+
"grad_norm": 0.2660665810108185,
|
| 1555 |
+
"learning_rate": 0.0004855752264439478,
|
| 1556 |
+
"loss": 2.8628,
|
| 1557 |
+
"step": 207000
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 11.18,
|
| 1561 |
+
"grad_norm": 0.24949678778648376,
|
| 1562 |
+
"learning_rate": 0.00048263439595341726,
|
| 1563 |
+
"loss": 2.8692,
|
| 1564 |
+
"step": 208000
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 11.24,
|
| 1568 |
+
"grad_norm": 0.2526322901248932,
|
| 1569 |
+
"learning_rate": 0.00047969356546288675,
|
| 1570 |
+
"loss": 2.8702,
|
| 1571 |
+
"step": 209000
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 11.29,
|
| 1575 |
+
"grad_norm": 0.29348504543304443,
|
| 1576 |
+
"learning_rate": 0.00047676155746382784,
|
| 1577 |
+
"loss": 2.8716,
|
| 1578 |
+
"step": 210000
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 11.34,
|
| 1582 |
+
"grad_norm": 0.2396526336669922,
|
| 1583 |
+
"learning_rate": 0.0004738207269732973,
|
| 1584 |
+
"loss": 2.8788,
|
| 1585 |
+
"step": 211000
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 11.4,
|
| 1589 |
+
"grad_norm": 0.24263983964920044,
|
| 1590 |
+
"learning_rate": 0.00047087989648276677,
|
| 1591 |
+
"loss": 2.8842,
|
| 1592 |
+
"step": 212000
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 11.45,
|
| 1596 |
+
"grad_norm": 0.24339748919010162,
|
| 1597 |
+
"learning_rate": 0.00046793906599223626,
|
| 1598 |
+
"loss": 2.8805,
|
| 1599 |
+
"step": 213000
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 11.5,
|
| 1603 |
+
"grad_norm": 0.35075515508651733,
|
| 1604 |
+
"learning_rate": 0.00046499823550170564,
|
| 1605 |
+
"loss": 2.8861,
|
| 1606 |
+
"step": 214000
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 11.56,
|
| 1610 |
+
"grad_norm": 0.2587752938270569,
|
| 1611 |
+
"learning_rate": 0.00046205740501117514,
|
| 1612 |
+
"loss": 2.8861,
|
| 1613 |
+
"step": 215000
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 11.61,
|
| 1617 |
+
"grad_norm": 0.2334243208169937,
|
| 1618 |
+
"learning_rate": 0.00045911657452064463,
|
| 1619 |
+
"loss": 2.8904,
|
| 1620 |
+
"step": 216000
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 11.67,
|
| 1624 |
+
"grad_norm": 0.2610035836696625,
|
| 1625 |
+
"learning_rate": 0.00045617868486060465,
|
| 1626 |
+
"loss": 2.8888,
|
| 1627 |
+
"step": 217000
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 11.72,
|
| 1631 |
+
"grad_norm": 0.2527429461479187,
|
| 1632 |
+
"learning_rate": 0.00045323785437007414,
|
| 1633 |
+
"loss": 2.8869,
|
| 1634 |
+
"step": 218000
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 11.77,
|
| 1638 |
+
"grad_norm": 0.27201133966445923,
|
| 1639 |
+
"learning_rate": 0.00045030290554052464,
|
| 1640 |
+
"loss": 2.8921,
|
| 1641 |
+
"step": 219000
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 11.83,
|
| 1645 |
+
"grad_norm": 0.25473257899284363,
|
| 1646 |
+
"learning_rate": 0.00044736207504999414,
|
| 1647 |
+
"loss": 2.8933,
|
| 1648 |
+
"step": 220000
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 11.88,
|
| 1652 |
+
"grad_norm": 0.24522466957569122,
|
| 1653 |
+
"learning_rate": 0.0004444212445594636,
|
| 1654 |
+
"loss": 2.8943,
|
| 1655 |
+
"step": 221000
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 11.93,
|
| 1659 |
+
"grad_norm": 0.2609509825706482,
|
| 1660 |
+
"learning_rate": 0.00044148041406893307,
|
| 1661 |
+
"loss": 2.8972,
|
| 1662 |
+
"step": 222000
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 11.99,
|
| 1666 |
+
"grad_norm": 0.252936989068985,
|
| 1667 |
+
"learning_rate": 0.0004385425244088931,
|
| 1668 |
+
"loss": 2.8986,
|
| 1669 |
+
"step": 223000
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 12.0,
|
| 1673 |
+
"eval_accuracy": 0.41030937281606633,
|
| 1674 |
+
"eval_loss": 3.364614963531494,
|
| 1675 |
+
"eval_runtime": 154.8866,
|
| 1676 |
+
"eval_samples_per_second": 373.957,
|
| 1677 |
+
"eval_steps_per_second": 5.849,
|
| 1678 |
+
"step": 223224
|
| 1679 |
+
},
|
| 1680 |
+
{
|
| 1681 |
+
"epoch": 12.04,
|
| 1682 |
+
"grad_norm": 0.2545512914657593,
|
| 1683 |
+
"learning_rate": 0.0004356016939183626,
|
| 1684 |
+
"loss": 2.8465,
|
| 1685 |
+
"step": 224000
|
| 1686 |
+
},
|
| 1687 |
+
{
|
| 1688 |
+
"epoch": 12.1,
|
| 1689 |
+
"grad_norm": 0.28531843423843384,
|
| 1690 |
+
"learning_rate": 0.0004326638042583226,
|
| 1691 |
+
"loss": 2.8391,
|
| 1692 |
+
"step": 225000
|
| 1693 |
+
},
|
| 1694 |
+
{
|
| 1695 |
+
"epoch": 12.15,
|
| 1696 |
+
"grad_norm": 0.26895156502723694,
|
| 1697 |
+
"learning_rate": 0.000429722973767792,
|
| 1698 |
+
"loss": 2.8444,
|
| 1699 |
+
"step": 226000
|
| 1700 |
+
},
|
| 1701 |
+
{
|
| 1702 |
+
"epoch": 12.2,
|
| 1703 |
+
"grad_norm": 0.28283581137657166,
|
| 1704 |
+
"learning_rate": 0.0004267821432772615,
|
| 1705 |
+
"loss": 2.849,
|
| 1706 |
+
"step": 227000
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 12.26,
|
| 1710 |
+
"grad_norm": 0.3257947564125061,
|
| 1711 |
+
"learning_rate": 0.0004238442536172215,
|
| 1712 |
+
"loss": 2.8473,
|
| 1713 |
+
"step": 228000
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 12.31,
|
| 1717 |
+
"grad_norm": 0.26338285207748413,
|
| 1718 |
+
"learning_rate": 0.000420903423126691,
|
| 1719 |
+
"loss": 2.8562,
|
| 1720 |
+
"step": 229000
|
| 1721 |
+
},
|
| 1722 |
+
{
|
| 1723 |
+
"epoch": 12.36,
|
| 1724 |
+
"grad_norm": 0.24919289350509644,
|
| 1725 |
+
"learning_rate": 0.0004179625926361605,
|
| 1726 |
+
"loss": 2.8545,
|
| 1727 |
+
"step": 230000
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 12.42,
|
| 1731 |
+
"grad_norm": 0.24178272485733032,
|
| 1732 |
+
"learning_rate": 0.00041502176214562993,
|
| 1733 |
+
"loss": 2.8579,
|
| 1734 |
+
"step": 231000
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 12.47,
|
| 1738 |
+
"grad_norm": 0.2602218985557556,
|
| 1739 |
+
"learning_rate": 0.00041208387248558995,
|
| 1740 |
+
"loss": 2.8588,
|
| 1741 |
+
"step": 232000
|
| 1742 |
+
},
|
| 1743 |
+
{
|
| 1744 |
+
"epoch": 12.53,
|
| 1745 |
+
"grad_norm": 0.23942138254642487,
|
| 1746 |
+
"learning_rate": 0.0004091459828255499,
|
| 1747 |
+
"loss": 2.8592,
|
| 1748 |
+
"step": 233000
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 12.58,
|
| 1752 |
+
"grad_norm": 0.24679596722126007,
|
| 1753 |
+
"learning_rate": 0.0004062051523350194,
|
| 1754 |
+
"loss": 2.8692,
|
| 1755 |
+
"step": 234000
|
| 1756 |
+
},
|
| 1757 |
+
{
|
| 1758 |
+
"epoch": 12.63,
|
| 1759 |
+
"grad_norm": 0.24444027245044708,
|
| 1760 |
+
"learning_rate": 0.0004032643218444889,
|
| 1761 |
+
"loss": 2.8655,
|
| 1762 |
+
"step": 235000
|
| 1763 |
+
},
|
| 1764 |
+
{
|
| 1765 |
+
"epoch": 12.69,
|
| 1766 |
+
"grad_norm": 0.30420973896980286,
|
| 1767 |
+
"learning_rate": 0.00040032349135395834,
|
| 1768 |
+
"loss": 2.8658,
|
| 1769 |
+
"step": 236000
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 12.74,
|
| 1773 |
+
"grad_norm": 0.2618982493877411,
|
| 1774 |
+
"learning_rate": 0.00039738560169391837,
|
| 1775 |
+
"loss": 2.8709,
|
| 1776 |
+
"step": 237000
|
| 1777 |
+
},
|
| 1778 |
+
{
|
| 1779 |
+
"epoch": 12.79,
|
| 1780 |
+
"grad_norm": 0.2887171804904938,
|
| 1781 |
+
"learning_rate": 0.00039444477120338786,
|
| 1782 |
+
"loss": 2.8724,
|
| 1783 |
+
"step": 238000
|
| 1784 |
+
},
|
| 1785 |
+
{
|
| 1786 |
+
"epoch": 12.85,
|
| 1787 |
+
"grad_norm": 0.26698777079582214,
|
| 1788 |
+
"learning_rate": 0.00039150394071285735,
|
| 1789 |
+
"loss": 2.8689,
|
| 1790 |
+
"step": 239000
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 12.9,
|
| 1794 |
+
"grad_norm": 0.2446519136428833,
|
| 1795 |
+
"learning_rate": 0.0003885631102223268,
|
| 1796 |
+
"loss": 2.8778,
|
| 1797 |
+
"step": 240000
|
| 1798 |
+
},
|
| 1799 |
+
{
|
| 1800 |
+
"epoch": 12.96,
|
| 1801 |
+
"grad_norm": 0.25453439354896545,
|
| 1802 |
+
"learning_rate": 0.00038562522056228676,
|
| 1803 |
+
"loss": 2.8745,
|
| 1804 |
+
"step": 241000
|
| 1805 |
+
},
|
| 1806 |
+
{
|
| 1807 |
+
"epoch": 13.0,
|
| 1808 |
+
"eval_accuracy": 0.4104802079472664,
|
| 1809 |
+
"eval_loss": 3.367594003677368,
|
| 1810 |
+
"eval_runtime": 154.8681,
|
| 1811 |
+
"eval_samples_per_second": 374.002,
|
| 1812 |
+
"eval_steps_per_second": 5.85,
|
| 1813 |
+
"step": 241826
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 13.01,
|
| 1817 |
+
"grad_norm": 0.2726215124130249,
|
| 1818 |
+
"learning_rate": 0.0003826873309022468,
|
| 1819 |
+
"loss": 2.8634,
|
| 1820 |
+
"step": 242000
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 13.06,
|
| 1824 |
+
"grad_norm": 0.2711837887763977,
|
| 1825 |
+
"learning_rate": 0.0003797465004117163,
|
| 1826 |
+
"loss": 2.8164,
|
| 1827 |
+
"step": 243000
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 13.12,
|
| 1831 |
+
"grad_norm": 0.2655225694179535,
|
| 1832 |
+
"learning_rate": 0.00037680566992118577,
|
| 1833 |
+
"loss": 2.8182,
|
| 1834 |
+
"step": 244000
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 13.17,
|
| 1838 |
+
"grad_norm": 0.3055564761161804,
|
| 1839 |
+
"learning_rate": 0.00037386778026114574,
|
| 1840 |
+
"loss": 2.8278,
|
| 1841 |
+
"step": 245000
|
| 1842 |
+
},
|
| 1843 |
+
{
|
| 1844 |
+
"epoch": 13.22,
|
| 1845 |
+
"grad_norm": 0.3115604817867279,
|
| 1846 |
+
"learning_rate": 0.00037092694977061523,
|
| 1847 |
+
"loss": 2.8282,
|
| 1848 |
+
"step": 246000
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 13.28,
|
| 1852 |
+
"grad_norm": 0.27207228541374207,
|
| 1853 |
+
"learning_rate": 0.00036798611928008467,
|
| 1854 |
+
"loss": 2.8323,
|
| 1855 |
+
"step": 247000
|
| 1856 |
+
},
|
| 1857 |
+
{
|
| 1858 |
+
"epoch": 13.33,
|
| 1859 |
+
"grad_norm": 0.2732410132884979,
|
| 1860 |
+
"learning_rate": 0.00036504528878955416,
|
| 1861 |
+
"loss": 2.8363,
|
| 1862 |
+
"step": 248000
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 13.39,
|
| 1866 |
+
"grad_norm": 0.24663548171520233,
|
| 1867 |
+
"learning_rate": 0.0003621073991295142,
|
| 1868 |
+
"loss": 2.8382,
|
| 1869 |
+
"step": 249000
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 13.44,
|
| 1873 |
+
"grad_norm": 0.2780630588531494,
|
| 1874 |
+
"learning_rate": 0.0003591665686389837,
|
| 1875 |
+
"loss": 2.8379,
|
| 1876 |
+
"step": 250000
|
| 1877 |
+
},
|
| 1878 |
+
{
|
| 1879 |
+
"epoch": 13.49,
|
| 1880 |
+
"grad_norm": 0.2623123526573181,
|
| 1881 |
+
"learning_rate": 0.0003562257381484531,
|
| 1882 |
+
"loss": 2.8383,
|
| 1883 |
+
"step": 251000
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 13.55,
|
| 1887 |
+
"grad_norm": 0.26465824246406555,
|
| 1888 |
+
"learning_rate": 0.00035329078931890367,
|
| 1889 |
+
"loss": 2.8445,
|
| 1890 |
+
"step": 252000
|
| 1891 |
+
},
|
| 1892 |
+
{
|
| 1893 |
+
"epoch": 13.6,
|
| 1894 |
+
"grad_norm": 0.2646311819553375,
|
| 1895 |
+
"learning_rate": 0.0003503499588283731,
|
| 1896 |
+
"loss": 2.8417,
|
| 1897 |
+
"step": 253000
|
| 1898 |
+
},
|
| 1899 |
+
{
|
| 1900 |
+
"epoch": 13.65,
|
| 1901 |
+
"grad_norm": 0.2561984956264496,
|
| 1902 |
+
"learning_rate": 0.00034741206916833313,
|
| 1903 |
+
"loss": 2.8439,
|
| 1904 |
+
"step": 254000
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 13.71,
|
| 1908 |
+
"grad_norm": 0.3358238637447357,
|
| 1909 |
+
"learning_rate": 0.0003444712386778026,
|
| 1910 |
+
"loss": 2.8495,
|
| 1911 |
+
"step": 255000
|
| 1912 |
+
},
|
| 1913 |
+
{
|
| 1914 |
+
"epoch": 13.76,
|
| 1915 |
+
"grad_norm": 0.3044319748878479,
|
| 1916 |
+
"learning_rate": 0.0003415304081872721,
|
| 1917 |
+
"loss": 2.8472,
|
| 1918 |
+
"step": 256000
|
| 1919 |
+
},
|
| 1920 |
+
{
|
| 1921 |
+
"epoch": 13.82,
|
| 1922 |
+
"grad_norm": 0.2482803910970688,
|
| 1923 |
+
"learning_rate": 0.0003385925185272321,
|
| 1924 |
+
"loss": 2.8555,
|
| 1925 |
+
"step": 257000
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"epoch": 13.87,
|
| 1929 |
+
"grad_norm": 0.2835788130760193,
|
| 1930 |
+
"learning_rate": 0.0003356516880367016,
|
| 1931 |
+
"loss": 2.8494,
|
| 1932 |
+
"step": 258000
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"epoch": 13.92,
|
| 1936 |
+
"grad_norm": 0.2810549736022949,
|
| 1937 |
+
"learning_rate": 0.000332710857546171,
|
| 1938 |
+
"loss": 2.8522,
|
| 1939 |
+
"step": 259000
|
| 1940 |
+
},
|
| 1941 |
+
{
|
| 1942 |
+
"epoch": 13.98,
|
| 1943 |
+
"grad_norm": 0.30381470918655396,
|
| 1944 |
+
"learning_rate": 0.0003297700270556405,
|
| 1945 |
+
"loss": 2.8518,
|
| 1946 |
+
"step": 260000
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 14.0,
|
| 1950 |
+
"eval_accuracy": 0.4109941912696463,
|
| 1951 |
+
"eval_loss": 3.3750314712524414,
|
| 1952 |
+
"eval_runtime": 154.897,
|
| 1953 |
+
"eval_samples_per_second": 373.932,
|
| 1954 |
+
"eval_steps_per_second": 5.849,
|
| 1955 |
+
"step": 260428
|
| 1956 |
+
},
|
| 1957 |
+
{
|
| 1958 |
+
"epoch": 14.03,
|
| 1959 |
+
"grad_norm": 0.27919963002204895,
|
| 1960 |
+
"learning_rate": 0.00032683213739560053,
|
| 1961 |
+
"loss": 2.8197,
|
| 1962 |
+
"step": 261000
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 14.08,
|
| 1966 |
+
"grad_norm": 0.27463871240615845,
|
| 1967 |
+
"learning_rate": 0.00032389424773556055,
|
| 1968 |
+
"loss": 2.7989,
|
| 1969 |
+
"step": 262000
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 14.14,
|
| 1973 |
+
"grad_norm": 0.26133376359939575,
|
| 1974 |
+
"learning_rate": 0.00032095341724503005,
|
| 1975 |
+
"loss": 2.8029,
|
| 1976 |
+
"step": 263000
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 14.19,
|
| 1980 |
+
"grad_norm": 0.34036701917648315,
|
| 1981 |
+
"learning_rate": 0.00031801258675449943,
|
| 1982 |
+
"loss": 2.8059,
|
| 1983 |
+
"step": 264000
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 14.25,
|
| 1987 |
+
"grad_norm": 0.293765664100647,
|
| 1988 |
+
"learning_rate": 0.00031507469709445945,
|
| 1989 |
+
"loss": 2.8085,
|
| 1990 |
+
"step": 265000
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 14.3,
|
| 1994 |
+
"grad_norm": 0.26148098707199097,
|
| 1995 |
+
"learning_rate": 0.00031213386660392895,
|
| 1996 |
+
"loss": 2.8132,
|
| 1997 |
+
"step": 266000
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 14.35,
|
| 2001 |
+
"grad_norm": 0.25458094477653503,
|
| 2002 |
+
"learning_rate": 0.00030919303611339844,
|
| 2003 |
+
"loss": 2.8154,
|
| 2004 |
+
"step": 267000
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 14.41,
|
| 2008 |
+
"grad_norm": 0.26752936840057373,
|
| 2009 |
+
"learning_rate": 0.00030625220562286793,
|
| 2010 |
+
"loss": 2.8209,
|
| 2011 |
+
"step": 268000
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 14.46,
|
| 2015 |
+
"grad_norm": 0.26262667775154114,
|
| 2016 |
+
"learning_rate": 0.00030331137513233737,
|
| 2017 |
+
"loss": 2.8175,
|
| 2018 |
+
"step": 269000
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 14.51,
|
| 2022 |
+
"grad_norm": 0.3177816867828369,
|
| 2023 |
+
"learning_rate": 0.00030037054464180686,
|
| 2024 |
+
"loss": 2.8221,
|
| 2025 |
+
"step": 270000
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 14.57,
|
| 2029 |
+
"grad_norm": 0.34930920600891113,
|
| 2030 |
+
"learning_rate": 0.0002974326549817669,
|
| 2031 |
+
"loss": 2.8277,
|
| 2032 |
+
"step": 271000
|
| 2033 |
+
},
|
| 2034 |
+
{
|
| 2035 |
+
"epoch": 14.62,
|
| 2036 |
+
"grad_norm": 0.25501158833503723,
|
| 2037 |
+
"learning_rate": 0.0002944918244912364,
|
| 2038 |
+
"loss": 2.8237,
|
| 2039 |
+
"step": 272000
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 14.68,
|
| 2043 |
+
"grad_norm": 0.25757795572280884,
|
| 2044 |
+
"learning_rate": 0.00029155393483119635,
|
| 2045 |
+
"loss": 2.823,
|
| 2046 |
+
"step": 273000
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 14.73,
|
| 2050 |
+
"grad_norm": 0.262439489364624,
|
| 2051 |
+
"learning_rate": 0.0002886131043406658,
|
| 2052 |
+
"loss": 2.8256,
|
| 2053 |
+
"step": 274000
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 14.78,
|
| 2057 |
+
"grad_norm": 0.27561119198799133,
|
| 2058 |
+
"learning_rate": 0.0002856722738501353,
|
| 2059 |
+
"loss": 2.8325,
|
| 2060 |
+
"step": 275000
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 14.84,
|
| 2064 |
+
"grad_norm": 0.30357280373573303,
|
| 2065 |
+
"learning_rate": 0.0002827343841900953,
|
| 2066 |
+
"loss": 2.8292,
|
| 2067 |
+
"step": 276000
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 14.89,
|
| 2071 |
+
"grad_norm": 0.27511313557624817,
|
| 2072 |
+
"learning_rate": 0.0002797935536995648,
|
| 2073 |
+
"loss": 2.8319,
|
| 2074 |
+
"step": 277000
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 14.94,
|
| 2078 |
+
"grad_norm": 0.26999151706695557,
|
| 2079 |
+
"learning_rate": 0.0002768527232090343,
|
| 2080 |
+
"loss": 2.8341,
|
| 2081 |
+
"step": 278000
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 15.0,
|
| 2085 |
+
"grad_norm": 0.27331700921058655,
|
| 2086 |
+
"learning_rate": 0.00027391483354899425,
|
| 2087 |
+
"loss": 2.8328,
|
| 2088 |
+
"step": 279000
|
| 2089 |
+
},
|
| 2090 |
+
{
|
| 2091 |
+
"epoch": 15.0,
|
| 2092 |
+
"eval_accuracy": 0.4111497096844038,
|
| 2093 |
+
"eval_loss": 3.3722476959228516,
|
| 2094 |
+
"eval_runtime": 154.7468,
|
| 2095 |
+
"eval_samples_per_second": 374.295,
|
| 2096 |
+
"eval_steps_per_second": 5.855,
|
| 2097 |
+
"step": 279030
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 15.05,
|
| 2101 |
+
"grad_norm": 0.2742418944835663,
|
| 2102 |
+
"learning_rate": 0.0002709769438889542,
|
| 2103 |
+
"loss": 2.7834,
|
| 2104 |
+
"step": 280000
|
| 2105 |
+
},
|
| 2106 |
+
{
|
| 2107 |
+
"epoch": 15.11,
|
| 2108 |
+
"grad_norm": 0.2747434675693512,
|
| 2109 |
+
"learning_rate": 0.0002680361133984237,
|
| 2110 |
+
"loss": 2.7848,
|
| 2111 |
+
"step": 281000
|
| 2112 |
+
},
|
| 2113 |
+
{
|
| 2114 |
+
"epoch": 15.16,
|
| 2115 |
+
"grad_norm": 0.308196485042572,
|
| 2116 |
+
"learning_rate": 0.0002650952829078932,
|
| 2117 |
+
"loss": 2.7865,
|
| 2118 |
+
"step": 282000
|
| 2119 |
+
},
|
| 2120 |
+
{
|
| 2121 |
+
"epoch": 15.21,
|
| 2122 |
+
"grad_norm": 0.3813375234603882,
|
| 2123 |
+
"learning_rate": 0.0002621544524173627,
|
| 2124 |
+
"loss": 2.7904,
|
| 2125 |
+
"step": 283000
|
| 2126 |
+
},
|
| 2127 |
+
{
|
| 2128 |
+
"epoch": 15.27,
|
| 2129 |
+
"grad_norm": 0.25685274600982666,
|
| 2130 |
+
"learning_rate": 0.00025921656275732267,
|
| 2131 |
+
"loss": 2.7944,
|
| 2132 |
+
"step": 284000
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 15.32,
|
| 2136 |
+
"grad_norm": 0.2853144109249115,
|
| 2137 |
+
"learning_rate": 0.0002562757322667921,
|
| 2138 |
+
"loss": 2.7967,
|
| 2139 |
+
"step": 285000
|
| 2140 |
+
},
|
| 2141 |
+
{
|
| 2142 |
+
"epoch": 15.37,
|
| 2143 |
+
"grad_norm": 0.2811448276042938,
|
| 2144 |
+
"learning_rate": 0.0002533349017762616,
|
| 2145 |
+
"loss": 2.7989,
|
| 2146 |
+
"step": 286000
|
| 2147 |
+
},
|
| 2148 |
+
{
|
| 2149 |
+
"epoch": 15.43,
|
| 2150 |
+
"grad_norm": 0.32674285769462585,
|
| 2151 |
+
"learning_rate": 0.0002503970121162216,
|
| 2152 |
+
"loss": 2.8013,
|
| 2153 |
+
"step": 287000
|
| 2154 |
+
},
|
| 2155 |
+
{
|
| 2156 |
+
"epoch": 15.48,
|
| 2157 |
+
"grad_norm": 0.2677679657936096,
|
| 2158 |
+
"learning_rate": 0.0002474561816256911,
|
| 2159 |
+
"loss": 2.8003,
|
| 2160 |
+
"step": 288000
|
| 2161 |
+
},
|
| 2162 |
+
{
|
| 2163 |
+
"epoch": 15.54,
|
| 2164 |
+
"grad_norm": 0.2848023474216461,
|
| 2165 |
+
"learning_rate": 0.00024451829196565114,
|
| 2166 |
+
"loss": 2.8051,
|
| 2167 |
+
"step": 289000
|
| 2168 |
+
},
|
| 2169 |
+
{
|
| 2170 |
+
"epoch": 15.59,
|
| 2171 |
+
"grad_norm": 0.3296356499195099,
|
| 2172 |
+
"learning_rate": 0.00024157746147512058,
|
| 2173 |
+
"loss": 2.8004,
|
| 2174 |
+
"step": 290000
|
| 2175 |
+
},
|
| 2176 |
+
{
|
| 2177 |
+
"epoch": 15.64,
|
| 2178 |
+
"grad_norm": 0.3256130516529083,
|
| 2179 |
+
"learning_rate": 0.00023863663098459004,
|
| 2180 |
+
"loss": 2.8051,
|
| 2181 |
+
"step": 291000
|
| 2182 |
+
},
|
| 2183 |
+
{
|
| 2184 |
+
"epoch": 15.7,
|
| 2185 |
+
"grad_norm": 0.30205684900283813,
|
| 2186 |
+
"learning_rate": 0.0002357016821550406,
|
| 2187 |
+
"loss": 2.8075,
|
| 2188 |
+
"step": 292000
|
| 2189 |
+
},
|
| 2190 |
+
{
|
| 2191 |
+
"epoch": 15.75,
|
| 2192 |
+
"grad_norm": 0.2870996296405792,
|
| 2193 |
+
"learning_rate": 0.00023276085166451006,
|
| 2194 |
+
"loss": 2.8081,
|
| 2195 |
+
"step": 293000
|
| 2196 |
+
},
|
| 2197 |
+
{
|
| 2198 |
+
"epoch": 15.8,
|
| 2199 |
+
"grad_norm": 0.26530566811561584,
|
| 2200 |
+
"learning_rate": 0.00022982002117397955,
|
| 2201 |
+
"loss": 2.8119,
|
| 2202 |
+
"step": 294000
|
| 2203 |
+
},
|
| 2204 |
+
{
|
| 2205 |
+
"epoch": 15.86,
|
| 2206 |
+
"grad_norm": 0.27719756960868835,
|
| 2207 |
+
"learning_rate": 0.000226879190683449,
|
| 2208 |
+
"loss": 2.8085,
|
| 2209 |
+
"step": 295000
|
| 2210 |
+
},
|
| 2211 |
+
{
|
| 2212 |
+
"epoch": 15.91,
|
| 2213 |
+
"grad_norm": 0.27279022336006165,
|
| 2214 |
+
"learning_rate": 0.00022393836019291848,
|
| 2215 |
+
"loss": 2.8137,
|
| 2216 |
+
"step": 296000
|
| 2217 |
+
},
|
| 2218 |
+
{
|
| 2219 |
+
"epoch": 15.97,
|
| 2220 |
+
"grad_norm": 0.2712692618370056,
|
| 2221 |
+
"learning_rate": 0.0002210004705328785,
|
| 2222 |
+
"loss": 2.8089,
|
| 2223 |
+
"step": 297000
|
| 2224 |
+
},
|
| 2225 |
+
{
|
| 2226 |
+
"epoch": 16.0,
|
| 2227 |
+
"eval_accuracy": 0.4115371957389701,
|
| 2228 |
+
"eval_loss": 3.3797037601470947,
|
| 2229 |
+
"eval_runtime": 154.9333,
|
| 2230 |
+
"eval_samples_per_second": 373.845,
|
| 2231 |
+
"eval_steps_per_second": 5.848,
|
| 2232 |
+
"step": 297632
|
| 2233 |
+
},
|
| 2234 |
+
{
|
| 2235 |
+
"epoch": 16.02,
|
| 2236 |
+
"grad_norm": 0.2609094977378845,
|
| 2237 |
+
"learning_rate": 0.0002180625808728385,
|
| 2238 |
+
"loss": 2.7927,
|
| 2239 |
+
"step": 298000
|
| 2240 |
+
},
|
| 2241 |
+
{
|
| 2242 |
+
"epoch": 16.07,
|
| 2243 |
+
"grad_norm": 0.33733800053596497,
|
| 2244 |
+
"learning_rate": 0.00021512175038230797,
|
| 2245 |
+
"loss": 2.7628,
|
| 2246 |
+
"step": 299000
|
| 2247 |
+
},
|
| 2248 |
+
{
|
| 2249 |
+
"epoch": 16.13,
|
| 2250 |
+
"grad_norm": 0.27339014410972595,
|
| 2251 |
+
"learning_rate": 0.00021218091989177743,
|
| 2252 |
+
"loss": 2.7711,
|
| 2253 |
+
"step": 300000
|
| 2254 |
+
},
|
| 2255 |
+
{
|
| 2256 |
+
"epoch": 16.18,
|
| 2257 |
+
"grad_norm": 0.27283820509910583,
|
| 2258 |
+
"learning_rate": 0.00020924303023173743,
|
| 2259 |
+
"loss": 2.7731,
|
| 2260 |
+
"step": 301000
|
| 2261 |
+
},
|
| 2262 |
+
{
|
| 2263 |
+
"epoch": 16.23,
|
| 2264 |
+
"grad_norm": 0.29361584782600403,
|
| 2265 |
+
"learning_rate": 0.00020630219974120692,
|
| 2266 |
+
"loss": 2.7739,
|
| 2267 |
+
"step": 302000
|
| 2268 |
+
},
|
| 2269 |
+
{
|
| 2270 |
+
"epoch": 16.29,
|
| 2271 |
+
"grad_norm": 0.2751254439353943,
|
| 2272 |
+
"learning_rate": 0.0002033613692506764,
|
| 2273 |
+
"loss": 2.7733,
|
| 2274 |
+
"step": 303000
|
| 2275 |
+
},
|
| 2276 |
+
{
|
| 2277 |
+
"epoch": 16.34,
|
| 2278 |
+
"grad_norm": 0.31571540236473083,
|
| 2279 |
+
"learning_rate": 0.00020042053876014588,
|
| 2280 |
+
"loss": 2.7803,
|
| 2281 |
+
"step": 304000
|
| 2282 |
+
},
|
| 2283 |
+
{
|
| 2284 |
+
"epoch": 16.4,
|
| 2285 |
+
"grad_norm": 0.2743053138256073,
|
| 2286 |
+
"learning_rate": 0.00019748264910010588,
|
| 2287 |
+
"loss": 2.7825,
|
| 2288 |
+
"step": 305000
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 16.45,
|
| 2292 |
+
"grad_norm": 0.30572009086608887,
|
| 2293 |
+
"learning_rate": 0.00019454181860957534,
|
| 2294 |
+
"loss": 2.7831,
|
| 2295 |
+
"step": 306000
|
| 2296 |
+
},
|
| 2297 |
+
{
|
| 2298 |
+
"epoch": 16.5,
|
| 2299 |
+
"grad_norm": 0.27128228545188904,
|
| 2300 |
+
"learning_rate": 0.00019160392894953534,
|
| 2301 |
+
"loss": 2.7858,
|
| 2302 |
+
"step": 307000
|
| 2303 |
+
},
|
| 2304 |
+
{
|
| 2305 |
+
"epoch": 16.56,
|
| 2306 |
+
"grad_norm": 0.28637221455574036,
|
| 2307 |
+
"learning_rate": 0.00018866603928949536,
|
| 2308 |
+
"loss": 2.7873,
|
| 2309 |
+
"step": 308000
|
| 2310 |
+
},
|
| 2311 |
+
{
|
| 2312 |
+
"epoch": 16.61,
|
| 2313 |
+
"grad_norm": 0.30166834592819214,
|
| 2314 |
+
"learning_rate": 0.00018572520879896485,
|
| 2315 |
+
"loss": 2.7859,
|
| 2316 |
+
"step": 309000
|
| 2317 |
+
},
|
| 2318 |
+
{
|
| 2319 |
+
"epoch": 16.66,
|
| 2320 |
+
"grad_norm": 0.26175713539123535,
|
| 2321 |
+
"learning_rate": 0.00018278437830843432,
|
| 2322 |
+
"loss": 2.7902,
|
| 2323 |
+
"step": 310000
|
| 2324 |
+
},
|
| 2325 |
+
{
|
| 2326 |
+
"epoch": 16.72,
|
| 2327 |
+
"grad_norm": 0.3338593542575836,
|
| 2328 |
+
"learning_rate": 0.00017984648864839432,
|
| 2329 |
+
"loss": 2.7876,
|
| 2330 |
+
"step": 311000
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 16.77,
|
| 2334 |
+
"grad_norm": 0.291354775428772,
|
| 2335 |
+
"learning_rate": 0.00017690565815786378,
|
| 2336 |
+
"loss": 2.789,
|
| 2337 |
+
"step": 312000
|
| 2338 |
+
},
|
| 2339 |
+
{
|
| 2340 |
+
"epoch": 16.83,
|
| 2341 |
+
"grad_norm": 0.3241804242134094,
|
| 2342 |
+
"learning_rate": 0.00017396482766733327,
|
| 2343 |
+
"loss": 2.7875,
|
| 2344 |
+
"step": 313000
|
| 2345 |
+
},
|
| 2346 |
+
{
|
| 2347 |
+
"epoch": 16.88,
|
| 2348 |
+
"grad_norm": 0.27583855390548706,
|
| 2349 |
+
"learning_rate": 0.00017102399717680274,
|
| 2350 |
+
"loss": 2.7928,
|
| 2351 |
+
"step": 314000
|
| 2352 |
+
},
|
| 2353 |
+
{
|
| 2354 |
+
"epoch": 16.93,
|
| 2355 |
+
"grad_norm": 0.38700050115585327,
|
| 2356 |
+
"learning_rate": 0.00016808610751676274,
|
| 2357 |
+
"loss": 2.7902,
|
| 2358 |
+
"step": 315000
|
| 2359 |
+
},
|
| 2360 |
+
{
|
| 2361 |
+
"epoch": 16.99,
|
| 2362 |
+
"grad_norm": 0.2652926445007324,
|
| 2363 |
+
"learning_rate": 0.00016514527702623223,
|
| 2364 |
+
"loss": 2.7911,
|
| 2365 |
+
"step": 316000
|
| 2366 |
+
},
|
| 2367 |
+
{
|
| 2368 |
+
"epoch": 17.0,
|
| 2369 |
+
"eval_accuracy": 0.4108795174320692,
|
| 2370 |
+
"eval_loss": 3.3881633281707764,
|
| 2371 |
+
"eval_runtime": 154.5367,
|
| 2372 |
+
"eval_samples_per_second": 374.804,
|
| 2373 |
+
"eval_steps_per_second": 5.863,
|
| 2374 |
+
"step": 316234
|
| 2375 |
+
},
|
| 2376 |
+
{
|
| 2377 |
+
"epoch": 17.04,
|
| 2378 |
+
"grad_norm": 0.2926689684391022,
|
| 2379 |
+
"learning_rate": 0.0001622044465357017,
|
| 2380 |
+
"loss": 2.7624,
|
| 2381 |
+
"step": 317000
|
| 2382 |
+
},
|
| 2383 |
+
{
|
| 2384 |
+
"epoch": 17.09,
|
| 2385 |
+
"grad_norm": 0.2626492977142334,
|
| 2386 |
+
"learning_rate": 0.0001592665568756617,
|
| 2387 |
+
"loss": 2.7546,
|
| 2388 |
+
"step": 318000
|
| 2389 |
+
},
|
| 2390 |
+
{
|
| 2391 |
+
"epoch": 17.15,
|
| 2392 |
+
"grad_norm": 0.27373695373535156,
|
| 2393 |
+
"learning_rate": 0.00015632572638513115,
|
| 2394 |
+
"loss": 2.7568,
|
| 2395 |
+
"step": 319000
|
| 2396 |
+
},
|
| 2397 |
+
{
|
| 2398 |
+
"epoch": 17.2,
|
| 2399 |
+
"grad_norm": 0.3076825737953186,
|
| 2400 |
+
"learning_rate": 0.00015338489589460065,
|
| 2401 |
+
"loss": 2.7601,
|
| 2402 |
+
"step": 320000
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 17.26,
|
| 2406 |
+
"grad_norm": 0.2832266092300415,
|
| 2407 |
+
"learning_rate": 0.0001504440654040701,
|
| 2408 |
+
"loss": 2.7627,
|
| 2409 |
+
"step": 321000
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 17.31,
|
| 2413 |
+
"grad_norm": 0.3134935200214386,
|
| 2414 |
+
"learning_rate": 0.0001475032349135396,
|
| 2415 |
+
"loss": 2.7633,
|
| 2416 |
+
"step": 322000
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 17.36,
|
| 2420 |
+
"grad_norm": 0.28282949328422546,
|
| 2421 |
+
"learning_rate": 0.0001445653452534996,
|
| 2422 |
+
"loss": 2.7616,
|
| 2423 |
+
"step": 323000
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 17.42,
|
| 2427 |
+
"grad_norm": 0.2717437744140625,
|
| 2428 |
+
"learning_rate": 0.00014162451476296906,
|
| 2429 |
+
"loss": 2.7664,
|
| 2430 |
+
"step": 324000
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 17.47,
|
| 2434 |
+
"grad_norm": 0.3018600642681122,
|
| 2435 |
+
"learning_rate": 0.00013868662510292906,
|
| 2436 |
+
"loss": 2.7672,
|
| 2437 |
+
"step": 325000
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 17.52,
|
| 2441 |
+
"grad_norm": 0.41076505184173584,
|
| 2442 |
+
"learning_rate": 0.00013574579461239855,
|
| 2443 |
+
"loss": 2.7662,
|
| 2444 |
+
"step": 326000
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 17.58,
|
| 2448 |
+
"grad_norm": 0.28314197063446045,
|
| 2449 |
+
"learning_rate": 0.00013280790495235855,
|
| 2450 |
+
"loss": 2.7679,
|
| 2451 |
+
"step": 327000
|
| 2452 |
+
},
|
| 2453 |
+
{
|
| 2454 |
+
"epoch": 17.63,
|
| 2455 |
+
"grad_norm": 0.2659911811351776,
|
| 2456 |
+
"learning_rate": 0.00012986707446182802,
|
| 2457 |
+
"loss": 2.7656,
|
| 2458 |
+
"step": 328000
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 17.69,
|
| 2462 |
+
"grad_norm": 0.28878411650657654,
|
| 2463 |
+
"learning_rate": 0.00012692918480178804,
|
| 2464 |
+
"loss": 2.7627,
|
| 2465 |
+
"step": 329000
|
| 2466 |
+
},
|
| 2467 |
+
{
|
| 2468 |
+
"epoch": 17.74,
|
| 2469 |
+
"grad_norm": 0.2714906930923462,
|
| 2470 |
+
"learning_rate": 0.0001239883543112575,
|
| 2471 |
+
"loss": 2.7715,
|
| 2472 |
+
"step": 330000
|
| 2473 |
+
},
|
| 2474 |
+
{
|
| 2475 |
+
"epoch": 17.79,
|
| 2476 |
+
"grad_norm": 0.3475322425365448,
|
| 2477 |
+
"learning_rate": 0.00012105046465121751,
|
| 2478 |
+
"loss": 2.7704,
|
| 2479 |
+
"step": 331000
|
| 2480 |
+
},
|
| 2481 |
+
{
|
| 2482 |
+
"epoch": 17.85,
|
| 2483 |
+
"grad_norm": 0.2924496531486511,
|
| 2484 |
+
"learning_rate": 0.00011810963416068698,
|
| 2485 |
+
"loss": 2.7695,
|
| 2486 |
+
"step": 332000
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 17.9,
|
| 2490 |
+
"grad_norm": 0.2651541829109192,
|
| 2491 |
+
"learning_rate": 0.00011516880367015646,
|
| 2492 |
+
"loss": 2.7744,
|
| 2493 |
+
"step": 333000
|
| 2494 |
+
},
|
| 2495 |
+
{
|
| 2496 |
+
"epoch": 17.96,
|
| 2497 |
+
"grad_norm": 0.27891284227371216,
|
| 2498 |
+
"learning_rate": 0.00011223091401011645,
|
| 2499 |
+
"loss": 2.773,
|
| 2500 |
+
"step": 334000
|
| 2501 |
+
},
|
| 2502 |
+
{
|
| 2503 |
+
"epoch": 18.0,
|
| 2504 |
+
"eval_accuracy": 0.41146578490748537,
|
| 2505 |
+
"eval_loss": 3.395080804824829,
|
| 2506 |
+
"eval_runtime": 154.4425,
|
| 2507 |
+
"eval_samples_per_second": 375.033,
|
| 2508 |
+
"eval_steps_per_second": 5.866,
|
| 2509 |
+
"step": 334836
|
| 2510 |
+
},
|
| 2511 |
+
{
|
| 2512 |
+
"epoch": 18.01,
|
| 2513 |
+
"grad_norm": 0.3232966363430023,
|
| 2514 |
+
"learning_rate": 0.00010929008351958593,
|
| 2515 |
+
"loss": 2.767,
|
| 2516 |
+
"step": 335000
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 18.06,
|
| 2520 |
+
"grad_norm": 0.3170246481895447,
|
| 2521 |
+
"learning_rate": 0.00010635219385954593,
|
| 2522 |
+
"loss": 2.7387,
|
| 2523 |
+
"step": 336000
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 18.12,
|
| 2527 |
+
"grad_norm": 0.28522101044654846,
|
| 2528 |
+
"learning_rate": 0.00010341136336901541,
|
| 2529 |
+
"loss": 2.7415,
|
| 2530 |
+
"step": 337000
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"epoch": 18.17,
|
| 2534 |
+
"grad_norm": 0.35791322588920593,
|
| 2535 |
+
"learning_rate": 0.00010047347370897542,
|
| 2536 |
+
"loss": 2.741,
|
| 2537 |
+
"step": 338000
|
| 2538 |
+
},
|
| 2539 |
+
{
|
| 2540 |
+
"epoch": 18.22,
|
| 2541 |
+
"grad_norm": 0.2994053065776825,
|
| 2542 |
+
"learning_rate": 9.75326432184449e-05,
|
| 2543 |
+
"loss": 2.7415,
|
| 2544 |
+
"step": 339000
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"epoch": 18.28,
|
| 2548 |
+
"grad_norm": 0.31925782561302185,
|
| 2549 |
+
"learning_rate": 9.459181272791436e-05,
|
| 2550 |
+
"loss": 2.7482,
|
| 2551 |
+
"step": 340000
|
| 2552 |
+
},
|
| 2553 |
+
{
|
| 2554 |
+
"epoch": 18.33,
|
| 2555 |
+
"grad_norm": 0.27817079424858093,
|
| 2556 |
+
"learning_rate": 9.165098223738384e-05,
|
| 2557 |
+
"loss": 2.7465,
|
| 2558 |
+
"step": 341000
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 18.39,
|
| 2562 |
+
"grad_norm": 0.27790531516075134,
|
| 2563 |
+
"learning_rate": 8.871309257734385e-05,
|
| 2564 |
+
"loss": 2.7479,
|
| 2565 |
+
"step": 342000
|
| 2566 |
+
},
|
| 2567 |
+
{
|
| 2568 |
+
"epoch": 18.44,
|
| 2569 |
+
"grad_norm": 0.33701515197753906,
|
| 2570 |
+
"learning_rate": 8.577226208681333e-05,
|
| 2571 |
+
"loss": 2.749,
|
| 2572 |
+
"step": 343000
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"epoch": 18.49,
|
| 2576 |
+
"grad_norm": 0.32259926199913025,
|
| 2577 |
+
"learning_rate": 8.283143159628278e-05,
|
| 2578 |
+
"loss": 2.7484,
|
| 2579 |
+
"step": 344000
|
| 2580 |
+
},
|
| 2581 |
+
{
|
| 2582 |
+
"epoch": 18.55,
|
| 2583 |
+
"grad_norm": 0.2922397255897522,
|
| 2584 |
+
"learning_rate": 7.989060110575226e-05,
|
| 2585 |
+
"loss": 2.7539,
|
| 2586 |
+
"step": 345000
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"epoch": 18.6,
|
| 2590 |
+
"grad_norm": 0.27831122279167175,
|
| 2591 |
+
"learning_rate": 7.694977061522174e-05,
|
| 2592 |
+
"loss": 2.7501,
|
| 2593 |
+
"step": 346000
|
| 2594 |
+
},
|
| 2595 |
+
{
|
| 2596 |
+
"epoch": 18.65,
|
| 2597 |
+
"grad_norm": 0.29930371046066284,
|
| 2598 |
+
"learning_rate": 7.401188095518175e-05,
|
| 2599 |
+
"loss": 2.7528,
|
| 2600 |
+
"step": 347000
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 18.71,
|
| 2604 |
+
"grad_norm": 0.2781750559806824,
|
| 2605 |
+
"learning_rate": 7.107105046465121e-05,
|
| 2606 |
+
"loss": 2.7542,
|
| 2607 |
+
"step": 348000
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 18.76,
|
| 2611 |
+
"grad_norm": 0.34259232878685,
|
| 2612 |
+
"learning_rate": 6.813316080461122e-05,
|
| 2613 |
+
"loss": 2.7519,
|
| 2614 |
+
"step": 349000
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"epoch": 18.82,
|
| 2618 |
+
"grad_norm": 0.27175283432006836,
|
| 2619 |
+
"learning_rate": 6.51923303140807e-05,
|
| 2620 |
+
"loss": 2.7517,
|
| 2621 |
+
"step": 350000
|
| 2622 |
+
},
|
| 2623 |
+
{
|
| 2624 |
+
"epoch": 18.87,
|
| 2625 |
+
"grad_norm": 0.28221696615219116,
|
| 2626 |
+
"learning_rate": 6.225149982355017e-05,
|
| 2627 |
+
"loss": 2.7564,
|
| 2628 |
+
"step": 351000
|
| 2629 |
+
},
|
| 2630 |
+
{
|
| 2631 |
+
"epoch": 18.92,
|
| 2632 |
+
"grad_norm": 0.29943257570266724,
|
| 2633 |
+
"learning_rate": 5.931361016351018e-05,
|
| 2634 |
+
"loss": 2.7532,
|
| 2635 |
+
"step": 352000
|
| 2636 |
+
},
|
| 2637 |
+
{
|
| 2638 |
+
"epoch": 18.98,
|
| 2639 |
+
"grad_norm": 0.31146806478500366,
|
| 2640 |
+
"learning_rate": 5.637277967297965e-05,
|
| 2641 |
+
"loss": 2.7517,
|
| 2642 |
+
"step": 353000
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 19.0,
|
| 2646 |
+
"eval_accuracy": 0.41124107255441245,
|
| 2647 |
+
"eval_loss": 3.40229868888855,
|
| 2648 |
+
"eval_runtime": 154.6135,
|
| 2649 |
+
"eval_samples_per_second": 374.618,
|
| 2650 |
+
"eval_steps_per_second": 5.86,
|
| 2651 |
+
"step": 353438
|
| 2652 |
+
},
|
| 2653 |
+
{
|
| 2654 |
+
"epoch": 19.03,
|
| 2655 |
+
"grad_norm": 0.2778169512748718,
|
| 2656 |
+
"learning_rate": 5.343194918244913e-05,
|
| 2657 |
+
"loss": 2.7402,
|
| 2658 |
+
"step": 354000
|
| 2659 |
+
},
|
| 2660 |
+
{
|
| 2661 |
+
"epoch": 19.08,
|
| 2662 |
+
"grad_norm": 0.27510160207748413,
|
| 2663 |
+
"learning_rate": 5.04911186919186e-05,
|
| 2664 |
+
"loss": 2.733,
|
| 2665 |
+
"step": 355000
|
| 2666 |
+
},
|
| 2667 |
+
{
|
| 2668 |
+
"epoch": 19.14,
|
| 2669 |
+
"grad_norm": 0.29721584916114807,
|
| 2670 |
+
"learning_rate": 4.75532290318786e-05,
|
| 2671 |
+
"loss": 2.7334,
|
| 2672 |
+
"step": 356000
|
| 2673 |
+
},
|
| 2674 |
+
{
|
| 2675 |
+
"epoch": 19.19,
|
| 2676 |
+
"grad_norm": 0.3098326623439789,
|
| 2677 |
+
"learning_rate": 4.4612398541348075e-05,
|
| 2678 |
+
"loss": 2.7346,
|
| 2679 |
+
"step": 357000
|
| 2680 |
+
},
|
| 2681 |
+
{
|
| 2682 |
+
"epoch": 19.25,
|
| 2683 |
+
"grad_norm": 0.3339441418647766,
|
| 2684 |
+
"learning_rate": 4.1671568050817554e-05,
|
| 2685 |
+
"loss": 2.7324,
|
| 2686 |
+
"step": 358000
|
| 2687 |
+
},
|
| 2688 |
+
{
|
| 2689 |
+
"epoch": 19.3,
|
| 2690 |
+
"grad_norm": 0.29926493763923645,
|
| 2691 |
+
"learning_rate": 3.873367839077756e-05,
|
| 2692 |
+
"loss": 2.7342,
|
| 2693 |
+
"step": 359000
|
| 2694 |
+
},
|
| 2695 |
+
{
|
| 2696 |
+
"epoch": 19.35,
|
| 2697 |
+
"grad_norm": 0.32908961176872253,
|
| 2698 |
+
"learning_rate": 3.579578873073756e-05,
|
| 2699 |
+
"loss": 2.7342,
|
| 2700 |
+
"step": 360000
|
| 2701 |
+
},
|
| 2702 |
+
{
|
| 2703 |
+
"epoch": 19.41,
|
| 2704 |
+
"grad_norm": 0.3179979622364044,
|
| 2705 |
+
"learning_rate": 3.285495824020703e-05,
|
| 2706 |
+
"loss": 2.7333,
|
| 2707 |
+
"step": 361000
|
| 2708 |
+
},
|
| 2709 |
+
{
|
| 2710 |
+
"epoch": 19.46,
|
| 2711 |
+
"grad_norm": 0.34039121866226196,
|
| 2712 |
+
"learning_rate": 2.9914127749676508e-05,
|
| 2713 |
+
"loss": 2.7342,
|
| 2714 |
+
"step": 362000
|
| 2715 |
+
},
|
| 2716 |
+
{
|
| 2717 |
+
"epoch": 19.51,
|
| 2718 |
+
"grad_norm": 0.29654234647750854,
|
| 2719 |
+
"learning_rate": 2.6973297259145983e-05,
|
| 2720 |
+
"loss": 2.7365,
|
| 2721 |
+
"step": 363000
|
| 2722 |
+
},
|
| 2723 |
+
{
|
| 2724 |
+
"epoch": 19.57,
|
| 2725 |
+
"grad_norm": 0.26721277832984924,
|
| 2726 |
+
"learning_rate": 2.403540759910599e-05,
|
| 2727 |
+
"loss": 2.7342,
|
| 2728 |
+
"step": 364000
|
| 2729 |
+
},
|
| 2730 |
+
{
|
| 2731 |
+
"epoch": 19.62,
|
| 2732 |
+
"grad_norm": 0.33862730860710144,
|
| 2733 |
+
"learning_rate": 2.109457710857546e-05,
|
| 2734 |
+
"loss": 2.7342,
|
| 2735 |
+
"step": 365000
|
| 2736 |
+
},
|
| 2737 |
+
{
|
| 2738 |
+
"epoch": 19.68,
|
| 2739 |
+
"grad_norm": 0.35417601466178894,
|
| 2740 |
+
"learning_rate": 1.8153746618044937e-05,
|
| 2741 |
+
"loss": 2.7328,
|
| 2742 |
+
"step": 366000
|
| 2743 |
+
},
|
| 2744 |
+
{
|
| 2745 |
+
"epoch": 19.73,
|
| 2746 |
+
"grad_norm": 0.2858126163482666,
|
| 2747 |
+
"learning_rate": 1.5215856958004942e-05,
|
| 2748 |
+
"loss": 2.7381,
|
| 2749 |
+
"step": 367000
|
| 2750 |
+
},
|
| 2751 |
+
{
|
| 2752 |
+
"epoch": 19.78,
|
| 2753 |
+
"grad_norm": 0.28896602988243103,
|
| 2754 |
+
"learning_rate": 1.2275026467474415e-05,
|
| 2755 |
+
"loss": 2.7363,
|
| 2756 |
+
"step": 368000
|
| 2757 |
+
},
|
| 2758 |
+
{
|
| 2759 |
+
"epoch": 19.84,
|
| 2760 |
+
"grad_norm": 0.28532859683036804,
|
| 2761 |
+
"learning_rate": 9.334195976943889e-06,
|
| 2762 |
+
"loss": 2.7336,
|
| 2763 |
+
"step": 369000
|
| 2764 |
+
},
|
| 2765 |
+
{
|
| 2766 |
+
"epoch": 19.89,
|
| 2767 |
+
"grad_norm": 0.28909268975257874,
|
| 2768 |
+
"learning_rate": 6.393365486413364e-06,
|
| 2769 |
+
"loss": 2.7353,
|
| 2770 |
+
"step": 370000
|
| 2771 |
+
},
|
| 2772 |
+
{
|
| 2773 |
+
"epoch": 19.94,
|
| 2774 |
+
"grad_norm": 0.32802271842956543,
|
| 2775 |
+
"learning_rate": 3.455475826373368e-06,
|
| 2776 |
+
"loss": 2.7323,
|
| 2777 |
+
"step": 371000
|
| 2778 |
+
},
|
| 2779 |
+
{
|
| 2780 |
+
"epoch": 20.0,
|
| 2781 |
+
"grad_norm": 0.28372275829315186,
|
| 2782 |
+
"learning_rate": 5.146453358428421e-07,
|
| 2783 |
+
"loss": 2.7342,
|
| 2784 |
+
"step": 372000
|
| 2785 |
+
},
|
| 2786 |
+
{
|
| 2787 |
+
"epoch": 20.0,
|
| 2788 |
+
"eval_accuracy": 0.4110677518157195,
|
| 2789 |
+
"eval_loss": 3.411550760269165,
|
| 2790 |
+
"eval_runtime": 155.3905,
|
| 2791 |
+
"eval_samples_per_second": 372.745,
|
| 2792 |
+
"eval_steps_per_second": 5.83,
|
| 2793 |
+
"step": 372040
|
| 2794 |
+
},
|
| 2795 |
+
{
|
| 2796 |
+
"epoch": 20.0,
|
| 2797 |
+
"step": 372040,
|
| 2798 |
+
"total_flos": 1.56748665397248e+18,
|
| 2799 |
+
"train_loss": 3.012804690259411,
|
| 2800 |
+
"train_runtime": 81277.7938,
|
| 2801 |
+
"train_samples_per_second": 146.472,
|
| 2802 |
+
"train_steps_per_second": 4.577
|
| 2803 |
+
}
|
| 2804 |
+
],
|
| 2805 |
+
"logging_steps": 1000,
|
| 2806 |
+
"max_steps": 372040,
|
| 2807 |
+
"num_input_tokens_seen": 0,
|
| 2808 |
+
"num_train_epochs": 20,
|
| 2809 |
+
"save_steps": 5000,
|
| 2810 |
+
"total_flos": 1.56748665397248e+18,
|
| 2811 |
+
"train_batch_size": 32,
|
| 2812 |
+
"trial_name": null,
|
| 2813 |
+
"trial_params": null
|
| 2814 |
+
}
|