End of training
Browse files
README.md
CHANGED
|
@@ -1,11 +1,23 @@
|
|
| 1 |
---
|
| 2 |
tags:
|
| 3 |
- generated_from_trainer
|
|
|
|
|
|
|
| 4 |
metrics:
|
| 5 |
- accuracy
|
| 6 |
model-index:
|
| 7 |
- name: smolm-autoreg-bpe-counterfactual_babylm_aann_high_variability_numeral-seed_1024-1e-3
|
| 8 |
-
results:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
---
|
| 10 |
|
| 11 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
@@ -13,7 +25,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 13 |
|
| 14 |
# smolm-autoreg-bpe-counterfactual_babylm_aann_high_variability_numeral-seed_1024-1e-3
|
| 15 |
|
| 16 |
-
This model was trained from scratch on
|
| 17 |
It achieves the following results on the evaluation set:
|
| 18 |
- Loss: 3.4236
|
| 19 |
- Accuracy: 0.4102
|
|
|
|
| 1 |
---
|
| 2 |
tags:
|
| 3 |
- generated_from_trainer
|
| 4 |
+
datasets:
|
| 5 |
+
- kanishka/counterfactual_babylm_aann_high_variability_numeral
|
| 6 |
metrics:
|
| 7 |
- accuracy
|
| 8 |
model-index:
|
| 9 |
- name: smolm-autoreg-bpe-counterfactual_babylm_aann_high_variability_numeral-seed_1024-1e-3
|
| 10 |
+
results:
|
| 11 |
+
- task:
|
| 12 |
+
name: Causal Language Modeling
|
| 13 |
+
type: text-generation
|
| 14 |
+
dataset:
|
| 15 |
+
name: kanishka/counterfactual_babylm_aann_high_variability_numeral
|
| 16 |
+
type: kanishka/counterfactual_babylm_aann_high_variability_numeral
|
| 17 |
+
metrics:
|
| 18 |
+
- name: Accuracy
|
| 19 |
+
type: accuracy
|
| 20 |
+
value: 0.41021489963935376
|
| 21 |
---
|
| 22 |
|
| 23 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
|
|
| 25 |
|
| 26 |
# smolm-autoreg-bpe-counterfactual_babylm_aann_high_variability_numeral-seed_1024-1e-3
|
| 27 |
|
| 28 |
+
This model was trained from scratch on the kanishka/counterfactual_babylm_aann_high_variability_numeral dataset.
|
| 29 |
It achieves the following results on the evaluation set:
|
| 30 |
- Loss: 3.4236
|
| 31 |
- Accuracy: 0.4102
|
all_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"eval_accuracy": 0.41021489963935376,
|
| 4 |
+
"eval_loss": 3.423628807067871,
|
| 5 |
+
"eval_runtime": 154.4177,
|
| 6 |
+
"eval_samples": 57917,
|
| 7 |
+
"eval_samples_per_second": 375.067,
|
| 8 |
+
"eval_steps_per_second": 5.861,
|
| 9 |
+
"perplexity": 30.68054704672833,
|
| 10 |
+
"total_flos": 1.5670047538944e+18,
|
| 11 |
+
"train_loss": 3.0279207580395733,
|
| 12 |
+
"train_runtime": 82395.3872,
|
| 13 |
+
"train_samples": 595065,
|
| 14 |
+
"train_samples_per_second": 144.441,
|
| 15 |
+
"train_steps_per_second": 4.514
|
| 16 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"eval_accuracy": 0.41021489963935376,
|
| 4 |
+
"eval_loss": 3.423628807067871,
|
| 5 |
+
"eval_runtime": 154.4177,
|
| 6 |
+
"eval_samples": 57917,
|
| 7 |
+
"eval_samples_per_second": 375.067,
|
| 8 |
+
"eval_steps_per_second": 5.861,
|
| 9 |
+
"perplexity": 30.68054704672833
|
| 10 |
+
}
|
runs/Jul18_08-56-31_phyl-ling-p01.la.utexas.edu/events.out.tfevents.1721393909.phyl-ling-p01.la.utexas.edu.3189178.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:43f215dc6313bef0b581d68c4ada5a3d107c5caf75f6c5e9046f672945ca797a
|
| 3 |
+
size 417
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"total_flos": 1.5670047538944e+18,
|
| 4 |
+
"train_loss": 3.0279207580395733,
|
| 5 |
+
"train_runtime": 82395.3872,
|
| 6 |
+
"train_samples": 595065,
|
| 7 |
+
"train_samples_per_second": 144.441,
|
| 8 |
+
"train_steps_per_second": 4.514
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2819 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 20.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 371920,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.05377500537750054,
|
| 13 |
+
"grad_norm": 0.8324036598205566,
|
| 14 |
+
"learning_rate": 3.125e-05,
|
| 15 |
+
"loss": 6.2314,
|
| 16 |
+
"step": 1000
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.10755001075500108,
|
| 20 |
+
"grad_norm": 0.9016917943954468,
|
| 21 |
+
"learning_rate": 6.25e-05,
|
| 22 |
+
"loss": 5.01,
|
| 23 |
+
"step": 2000
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.1613250161325016,
|
| 27 |
+
"grad_norm": 0.8393586874008179,
|
| 28 |
+
"learning_rate": 9.375e-05,
|
| 29 |
+
"loss": 4.6792,
|
| 30 |
+
"step": 3000
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.21510002151000215,
|
| 34 |
+
"grad_norm": 0.8001790642738342,
|
| 35 |
+
"learning_rate": 0.000125,
|
| 36 |
+
"loss": 4.4675,
|
| 37 |
+
"step": 4000
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.2688750268875027,
|
| 41 |
+
"grad_norm": 0.7500863671302795,
|
| 42 |
+
"learning_rate": 0.00015625,
|
| 43 |
+
"loss": 4.3004,
|
| 44 |
+
"step": 5000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.3226500322650032,
|
| 48 |
+
"grad_norm": 0.6959784626960754,
|
| 49 |
+
"learning_rate": 0.0001875,
|
| 50 |
+
"loss": 4.1762,
|
| 51 |
+
"step": 6000
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.3764250376425038,
|
| 55 |
+
"grad_norm": 0.7082997560501099,
|
| 56 |
+
"learning_rate": 0.00021875,
|
| 57 |
+
"loss": 4.0795,
|
| 58 |
+
"step": 7000
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.4302000430200043,
|
| 62 |
+
"grad_norm": 0.7400528788566589,
|
| 63 |
+
"learning_rate": 0.00025,
|
| 64 |
+
"loss": 3.9794,
|
| 65 |
+
"step": 8000
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.4839750483975048,
|
| 69 |
+
"grad_norm": 0.6886024475097656,
|
| 70 |
+
"learning_rate": 0.00028121875,
|
| 71 |
+
"loss": 3.9062,
|
| 72 |
+
"step": 9000
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.5377500537750054,
|
| 76 |
+
"grad_norm": 0.6196364760398865,
|
| 77 |
+
"learning_rate": 0.0003124375,
|
| 78 |
+
"loss": 3.8427,
|
| 79 |
+
"step": 10000
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.5915250591525059,
|
| 83 |
+
"grad_norm": 0.5815768241882324,
|
| 84 |
+
"learning_rate": 0.00034368749999999997,
|
| 85 |
+
"loss": 3.7992,
|
| 86 |
+
"step": 11000
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.6453000645300064,
|
| 90 |
+
"grad_norm": 0.5629006624221802,
|
| 91 |
+
"learning_rate": 0.0003749375,
|
| 92 |
+
"loss": 3.7502,
|
| 93 |
+
"step": 12000
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.699075069907507,
|
| 97 |
+
"grad_norm": 0.5031692981719971,
|
| 98 |
+
"learning_rate": 0.00040615625,
|
| 99 |
+
"loss": 3.7233,
|
| 100 |
+
"step": 13000
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.7528500752850076,
|
| 104 |
+
"grad_norm": 0.4921340048313141,
|
| 105 |
+
"learning_rate": 0.00043737500000000005,
|
| 106 |
+
"loss": 3.6917,
|
| 107 |
+
"step": 14000
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.806625080662508,
|
| 111 |
+
"grad_norm": 0.45878851413726807,
|
| 112 |
+
"learning_rate": 0.000468625,
|
| 113 |
+
"loss": 3.6641,
|
| 114 |
+
"step": 15000
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.8604000860400086,
|
| 118 |
+
"grad_norm": 0.4047335684299469,
|
| 119 |
+
"learning_rate": 0.000499875,
|
| 120 |
+
"loss": 3.6404,
|
| 121 |
+
"step": 16000
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.9141750914175092,
|
| 125 |
+
"grad_norm": 0.4339119493961334,
|
| 126 |
+
"learning_rate": 0.000531125,
|
| 127 |
+
"loss": 3.6129,
|
| 128 |
+
"step": 17000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.9679500967950097,
|
| 132 |
+
"grad_norm": 0.3588213324546814,
|
| 133 |
+
"learning_rate": 0.00056234375,
|
| 134 |
+
"loss": 3.5932,
|
| 135 |
+
"step": 18000
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 1.0,
|
| 139 |
+
"eval_accuracy": 0.3588065680197524,
|
| 140 |
+
"eval_loss": 3.770080804824829,
|
| 141 |
+
"eval_runtime": 152.9859,
|
| 142 |
+
"eval_samples_per_second": 378.577,
|
| 143 |
+
"eval_steps_per_second": 5.916,
|
| 144 |
+
"step": 18596
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 1.0217251021725102,
|
| 148 |
+
"grad_norm": 0.3476862907409668,
|
| 149 |
+
"learning_rate": 0.00059359375,
|
| 150 |
+
"loss": 3.5726,
|
| 151 |
+
"step": 19000
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 1.0755001075500108,
|
| 155 |
+
"grad_norm": 0.3370579481124878,
|
| 156 |
+
"learning_rate": 0.0006248437500000001,
|
| 157 |
+
"loss": 3.5453,
|
| 158 |
+
"step": 20000
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.1292751129275114,
|
| 162 |
+
"grad_norm": 0.3253530263900757,
|
| 163 |
+
"learning_rate": 0.00065609375,
|
| 164 |
+
"loss": 3.5364,
|
| 165 |
+
"step": 21000
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 1.1830501183050117,
|
| 169 |
+
"grad_norm": 0.3063829839229584,
|
| 170 |
+
"learning_rate": 0.00068728125,
|
| 171 |
+
"loss": 3.5214,
|
| 172 |
+
"step": 22000
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 1.2368251236825123,
|
| 176 |
+
"grad_norm": 0.28737059235572815,
|
| 177 |
+
"learning_rate": 0.00071853125,
|
| 178 |
+
"loss": 3.5158,
|
| 179 |
+
"step": 23000
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.2906001290600129,
|
| 183 |
+
"grad_norm": 0.29937857389450073,
|
| 184 |
+
"learning_rate": 0.00074978125,
|
| 185 |
+
"loss": 3.5014,
|
| 186 |
+
"step": 24000
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 1.3443751344375134,
|
| 190 |
+
"grad_norm": 0.2835935056209564,
|
| 191 |
+
"learning_rate": 0.0007810312499999999,
|
| 192 |
+
"loss": 3.4946,
|
| 193 |
+
"step": 25000
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 1.398150139815014,
|
| 197 |
+
"grad_norm": 0.2764816880226135,
|
| 198 |
+
"learning_rate": 0.00081225,
|
| 199 |
+
"loss": 3.4832,
|
| 200 |
+
"step": 26000
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 1.4519251451925146,
|
| 204 |
+
"grad_norm": 0.2620868384838104,
|
| 205 |
+
"learning_rate": 0.0008435000000000001,
|
| 206 |
+
"loss": 3.4761,
|
| 207 |
+
"step": 27000
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 1.5057001505700152,
|
| 211 |
+
"grad_norm": 0.2731957733631134,
|
| 212 |
+
"learning_rate": 0.00087471875,
|
| 213 |
+
"loss": 3.4653,
|
| 214 |
+
"step": 28000
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 1.5594751559475157,
|
| 218 |
+
"grad_norm": 0.26957619190216064,
|
| 219 |
+
"learning_rate": 0.00090596875,
|
| 220 |
+
"loss": 3.4552,
|
| 221 |
+
"step": 29000
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.613250161325016,
|
| 225 |
+
"grad_norm": 0.24591492116451263,
|
| 226 |
+
"learning_rate": 0.00093721875,
|
| 227 |
+
"loss": 3.4474,
|
| 228 |
+
"step": 30000
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 1.6670251667025167,
|
| 232 |
+
"grad_norm": 0.23927152156829834,
|
| 233 |
+
"learning_rate": 0.00096846875,
|
| 234 |
+
"loss": 3.4443,
|
| 235 |
+
"step": 31000
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 1.7208001720800172,
|
| 239 |
+
"grad_norm": 0.2176426500082016,
|
| 240 |
+
"learning_rate": 0.0009996875,
|
| 241 |
+
"loss": 3.4401,
|
| 242 |
+
"step": 32000
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.7745751774575176,
|
| 246 |
+
"grad_norm": 0.20793931186199188,
|
| 247 |
+
"learning_rate": 0.0009970875500117675,
|
| 248 |
+
"loss": 3.4261,
|
| 249 |
+
"step": 33000
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 1.8283501828350182,
|
| 253 |
+
"grad_norm": 0.2189057469367981,
|
| 254 |
+
"learning_rate": 0.0009941486232054601,
|
| 255 |
+
"loss": 3.419,
|
| 256 |
+
"step": 34000
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 1.8821251882125187,
|
| 260 |
+
"grad_norm": 0.2241194099187851,
|
| 261 |
+
"learning_rate": 0.0009912096963991528,
|
| 262 |
+
"loss": 3.4088,
|
| 263 |
+
"step": 35000
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 1.9359001935900193,
|
| 267 |
+
"grad_norm": 0.23365530371665955,
|
| 268 |
+
"learning_rate": 0.0009882678277241704,
|
| 269 |
+
"loss": 3.3934,
|
| 270 |
+
"step": 36000
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 1.9896751989675199,
|
| 274 |
+
"grad_norm": 0.2019016444683075,
|
| 275 |
+
"learning_rate": 0.000985328900917863,
|
| 276 |
+
"loss": 3.3833,
|
| 277 |
+
"step": 37000
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 2.0,
|
| 281 |
+
"eval_accuracy": 0.38185108449506,
|
| 282 |
+
"eval_loss": 3.5596837997436523,
|
| 283 |
+
"eval_runtime": 154.0793,
|
| 284 |
+
"eval_samples_per_second": 375.891,
|
| 285 |
+
"eval_steps_per_second": 5.874,
|
| 286 |
+
"step": 37192
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 2.0434502043450204,
|
| 290 |
+
"grad_norm": 0.20308424532413483,
|
| 291 |
+
"learning_rate": 0.0009823870322428808,
|
| 292 |
+
"loss": 3.3404,
|
| 293 |
+
"step": 38000
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 2.097225209722521,
|
| 297 |
+
"grad_norm": 0.21531735360622406,
|
| 298 |
+
"learning_rate": 0.0009794451635678984,
|
| 299 |
+
"loss": 3.3266,
|
| 300 |
+
"step": 39000
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 2.1510002151000216,
|
| 304 |
+
"grad_norm": 0.26003631949424744,
|
| 305 |
+
"learning_rate": 0.000976503294892916,
|
| 306 |
+
"loss": 3.3242,
|
| 307 |
+
"step": 40000
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 2.204775220477522,
|
| 311 |
+
"grad_norm": 0.24143873155117035,
|
| 312 |
+
"learning_rate": 0.0009735643680866086,
|
| 313 |
+
"loss": 3.3166,
|
| 314 |
+
"step": 41000
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 2.2585502258550227,
|
| 318 |
+
"grad_norm": 0.19083160161972046,
|
| 319 |
+
"learning_rate": 0.0009706224994116263,
|
| 320 |
+
"loss": 3.3187,
|
| 321 |
+
"step": 42000
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 2.3123252312325233,
|
| 325 |
+
"grad_norm": 0.22694003582000732,
|
| 326 |
+
"learning_rate": 0.000967680630736644,
|
| 327 |
+
"loss": 3.3112,
|
| 328 |
+
"step": 43000
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 2.3661002366100234,
|
| 332 |
+
"grad_norm": 0.21774055063724518,
|
| 333 |
+
"learning_rate": 0.0009647417039303365,
|
| 334 |
+
"loss": 3.3075,
|
| 335 |
+
"step": 44000
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 2.419875241987524,
|
| 339 |
+
"grad_norm": 0.2047697901725769,
|
| 340 |
+
"learning_rate": 0.0009617998352553542,
|
| 341 |
+
"loss": 3.2992,
|
| 342 |
+
"step": 45000
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 2.4736502473650246,
|
| 346 |
+
"grad_norm": 0.21876117587089539,
|
| 347 |
+
"learning_rate": 0.0009588579665803719,
|
| 348 |
+
"loss": 3.2983,
|
| 349 |
+
"step": 46000
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 2.527425252742525,
|
| 353 |
+
"grad_norm": 0.21647591888904572,
|
| 354 |
+
"learning_rate": 0.0009559190397740644,
|
| 355 |
+
"loss": 3.2876,
|
| 356 |
+
"step": 47000
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.5812002581200257,
|
| 360 |
+
"grad_norm": 0.20933736860752106,
|
| 361 |
+
"learning_rate": 0.0009529771710990821,
|
| 362 |
+
"loss": 3.2814,
|
| 363 |
+
"step": 48000
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 2.6349752634975263,
|
| 367 |
+
"grad_norm": 0.1911548376083374,
|
| 368 |
+
"learning_rate": 0.0009500382442927748,
|
| 369 |
+
"loss": 3.2797,
|
| 370 |
+
"step": 49000
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 2.688750268875027,
|
| 374 |
+
"grad_norm": 0.22081832587718964,
|
| 375 |
+
"learning_rate": 0.0009470963756177925,
|
| 376 |
+
"loss": 3.2783,
|
| 377 |
+
"step": 50000
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 2.7425252742525275,
|
| 381 |
+
"grad_norm": 0.21164289116859436,
|
| 382 |
+
"learning_rate": 0.0009441545069428101,
|
| 383 |
+
"loss": 3.2752,
|
| 384 |
+
"step": 51000
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 2.796300279630028,
|
| 388 |
+
"grad_norm": 0.21225039660930634,
|
| 389 |
+
"learning_rate": 0.0009412126382678278,
|
| 390 |
+
"loss": 3.2681,
|
| 391 |
+
"step": 52000
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 2.8500752850075286,
|
| 395 |
+
"grad_norm": 0.1924898326396942,
|
| 396 |
+
"learning_rate": 0.0009382707695928455,
|
| 397 |
+
"loss": 3.2629,
|
| 398 |
+
"step": 53000
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 2.903850290385029,
|
| 402 |
+
"grad_norm": 0.19862565398216248,
|
| 403 |
+
"learning_rate": 0.000935331842786538,
|
| 404 |
+
"loss": 3.2634,
|
| 405 |
+
"step": 54000
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 2.9576252957625293,
|
| 409 |
+
"grad_norm": 0.19020138680934906,
|
| 410 |
+
"learning_rate": 0.0009323899741115557,
|
| 411 |
+
"loss": 3.2597,
|
| 412 |
+
"step": 55000
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 3.0,
|
| 416 |
+
"eval_accuracy": 0.39273080241152825,
|
| 417 |
+
"eval_loss": 3.4648334980010986,
|
| 418 |
+
"eval_runtime": 154.5571,
|
| 419 |
+
"eval_samples_per_second": 374.729,
|
| 420 |
+
"eval_steps_per_second": 5.855,
|
| 421 |
+
"step": 55788
|
| 422 |
+
},
|
| 423 |
+
{
|
| 424 |
+
"epoch": 3.0114003011400303,
|
| 425 |
+
"grad_norm": 0.19856859743595123,
|
| 426 |
+
"learning_rate": 0.0009294481054365734,
|
| 427 |
+
"loss": 3.239,
|
| 428 |
+
"step": 56000
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 3.0651753065175305,
|
| 432 |
+
"grad_norm": 0.22371411323547363,
|
| 433 |
+
"learning_rate": 0.0009265091786302659,
|
| 434 |
+
"loss": 3.1886,
|
| 435 |
+
"step": 57000
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 3.118950311895031,
|
| 439 |
+
"grad_norm": 0.21350081264972687,
|
| 440 |
+
"learning_rate": 0.0009235673099552836,
|
| 441 |
+
"loss": 3.1941,
|
| 442 |
+
"step": 58000
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 3.1727253172725316,
|
| 446 |
+
"grad_norm": 0.219674214720726,
|
| 447 |
+
"learning_rate": 0.0009206254412803013,
|
| 448 |
+
"loss": 3.1942,
|
| 449 |
+
"step": 59000
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 3.226500322650032,
|
| 453 |
+
"grad_norm": 0.19072189927101135,
|
| 454 |
+
"learning_rate": 0.0009176865144739939,
|
| 455 |
+
"loss": 3.1973,
|
| 456 |
+
"step": 60000
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 3.2802753280275327,
|
| 460 |
+
"grad_norm": 0.205557718873024,
|
| 461 |
+
"learning_rate": 0.0009147475876676865,
|
| 462 |
+
"loss": 3.1932,
|
| 463 |
+
"step": 61000
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 3.3340503334050333,
|
| 467 |
+
"grad_norm": 0.2098790556192398,
|
| 468 |
+
"learning_rate": 0.0009118057189927041,
|
| 469 |
+
"loss": 3.1935,
|
| 470 |
+
"step": 62000
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 3.387825338782534,
|
| 474 |
+
"grad_norm": 0.196111798286438,
|
| 475 |
+
"learning_rate": 0.0009088638503177218,
|
| 476 |
+
"loss": 3.1954,
|
| 477 |
+
"step": 63000
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 3.4416003441600345,
|
| 481 |
+
"grad_norm": 0.19440898299217224,
|
| 482 |
+
"learning_rate": 0.0009059219816427395,
|
| 483 |
+
"loss": 3.1924,
|
| 484 |
+
"step": 64000
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 3.495375349537535,
|
| 488 |
+
"grad_norm": 0.21081770956516266,
|
| 489 |
+
"learning_rate": 0.0009029801129677572,
|
| 490 |
+
"loss": 3.1952,
|
| 491 |
+
"step": 65000
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 3.5491503549150356,
|
| 495 |
+
"grad_norm": 0.21867215633392334,
|
| 496 |
+
"learning_rate": 0.0009000411861614498,
|
| 497 |
+
"loss": 3.195,
|
| 498 |
+
"step": 66000
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 3.602925360292536,
|
| 502 |
+
"grad_norm": 0.22000326216220856,
|
| 503 |
+
"learning_rate": 0.0008970993174864674,
|
| 504 |
+
"loss": 3.1911,
|
| 505 |
+
"step": 67000
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 3.6567003656700363,
|
| 509 |
+
"grad_norm": 0.1891467422246933,
|
| 510 |
+
"learning_rate": 0.0008941574488114851,
|
| 511 |
+
"loss": 3.1934,
|
| 512 |
+
"step": 68000
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 3.7104753710475373,
|
| 516 |
+
"grad_norm": 0.18787287175655365,
|
| 517 |
+
"learning_rate": 0.0008912185220051777,
|
| 518 |
+
"loss": 3.191,
|
| 519 |
+
"step": 69000
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 3.7642503764250375,
|
| 523 |
+
"grad_norm": 0.23694172501564026,
|
| 524 |
+
"learning_rate": 0.0008882766533301954,
|
| 525 |
+
"loss": 3.1831,
|
| 526 |
+
"step": 70000
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 3.818025381802538,
|
| 530 |
+
"grad_norm": 0.19812917709350586,
|
| 531 |
+
"learning_rate": 0.000885334784655213,
|
| 532 |
+
"loss": 3.1815,
|
| 533 |
+
"step": 71000
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 3.8718003871800386,
|
| 537 |
+
"grad_norm": 0.2005423903465271,
|
| 538 |
+
"learning_rate": 0.0008823958578489056,
|
| 539 |
+
"loss": 3.1801,
|
| 540 |
+
"step": 72000
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 3.925575392557539,
|
| 544 |
+
"grad_norm": 0.21525584161281586,
|
| 545 |
+
"learning_rate": 0.0008794539891739233,
|
| 546 |
+
"loss": 3.1795,
|
| 547 |
+
"step": 73000
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 3.9793503979350398,
|
| 551 |
+
"grad_norm": 0.19802774488925934,
|
| 552 |
+
"learning_rate": 0.0008765150623676159,
|
| 553 |
+
"loss": 3.1741,
|
| 554 |
+
"step": 74000
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 4.0,
|
| 558 |
+
"eval_accuracy": 0.3976694409529698,
|
| 559 |
+
"eval_loss": 3.419067859649658,
|
| 560 |
+
"eval_runtime": 155.1328,
|
| 561 |
+
"eval_samples_per_second": 373.338,
|
| 562 |
+
"eval_steps_per_second": 5.834,
|
| 563 |
+
"step": 74384
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 4.033125403312541,
|
| 567 |
+
"grad_norm": 0.2355957329273224,
|
| 568 |
+
"learning_rate": 0.0008735731936926335,
|
| 569 |
+
"loss": 3.1356,
|
| 570 |
+
"step": 75000
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 4.086900408690041,
|
| 574 |
+
"grad_norm": 0.20892471075057983,
|
| 575 |
+
"learning_rate": 0.0008706313250176512,
|
| 576 |
+
"loss": 3.1124,
|
| 577 |
+
"step": 76000
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 4.140675414067541,
|
| 581 |
+
"grad_norm": 0.24330681562423706,
|
| 582 |
+
"learning_rate": 0.0008676923982113439,
|
| 583 |
+
"loss": 3.1228,
|
| 584 |
+
"step": 77000
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 4.194450419445042,
|
| 588 |
+
"grad_norm": 0.30532753467559814,
|
| 589 |
+
"learning_rate": 0.0008647505295363614,
|
| 590 |
+
"loss": 3.1191,
|
| 591 |
+
"step": 78000
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 4.248225424822542,
|
| 595 |
+
"grad_norm": 0.2023121416568756,
|
| 596 |
+
"learning_rate": 0.0008618116027300541,
|
| 597 |
+
"loss": 3.1185,
|
| 598 |
+
"step": 79000
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 4.302000430200043,
|
| 602 |
+
"grad_norm": 0.20023038983345032,
|
| 603 |
+
"learning_rate": 0.0008588697340550719,
|
| 604 |
+
"loss": 3.1248,
|
| 605 |
+
"step": 80000
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 4.355775435577543,
|
| 609 |
+
"grad_norm": 0.20664258301258087,
|
| 610 |
+
"learning_rate": 0.0008559278653800895,
|
| 611 |
+
"loss": 3.1305,
|
| 612 |
+
"step": 81000
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 4.409550440955044,
|
| 616 |
+
"grad_norm": 0.21807469427585602,
|
| 617 |
+
"learning_rate": 0.0008529889385737821,
|
| 618 |
+
"loss": 3.1265,
|
| 619 |
+
"step": 82000
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 4.4633254463325445,
|
| 623 |
+
"grad_norm": 0.20922619104385376,
|
| 624 |
+
"learning_rate": 0.0008500470698987998,
|
| 625 |
+
"loss": 3.1287,
|
| 626 |
+
"step": 83000
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 4.5171004517100455,
|
| 630 |
+
"grad_norm": 0.22318531572818756,
|
| 631 |
+
"learning_rate": 0.0008471052012238174,
|
| 632 |
+
"loss": 3.1265,
|
| 633 |
+
"step": 84000
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 4.570875457087546,
|
| 637 |
+
"grad_norm": 0.20071184635162354,
|
| 638 |
+
"learning_rate": 0.000844163332548835,
|
| 639 |
+
"loss": 3.1244,
|
| 640 |
+
"step": 85000
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 4.624650462465047,
|
| 644 |
+
"grad_norm": 0.23887498676776886,
|
| 645 |
+
"learning_rate": 0.0008412244057425277,
|
| 646 |
+
"loss": 3.1309,
|
| 647 |
+
"step": 86000
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 4.678425467842547,
|
| 651 |
+
"grad_norm": 0.21280068159103394,
|
| 652 |
+
"learning_rate": 0.0008382825370675454,
|
| 653 |
+
"loss": 3.1261,
|
| 654 |
+
"step": 87000
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 4.732200473220047,
|
| 658 |
+
"grad_norm": 0.20855990052223206,
|
| 659 |
+
"learning_rate": 0.0008353406683925629,
|
| 660 |
+
"loss": 3.1227,
|
| 661 |
+
"step": 88000
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 4.785975478597548,
|
| 665 |
+
"grad_norm": 0.23701632022857666,
|
| 666 |
+
"learning_rate": 0.0008324017415862556,
|
| 667 |
+
"loss": 3.1274,
|
| 668 |
+
"step": 89000
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 4.839750483975048,
|
| 672 |
+
"grad_norm": 0.22062337398529053,
|
| 673 |
+
"learning_rate": 0.0008294598729112733,
|
| 674 |
+
"loss": 3.1259,
|
| 675 |
+
"step": 90000
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 4.893525489352549,
|
| 679 |
+
"grad_norm": 0.21007812023162842,
|
| 680 |
+
"learning_rate": 0.0008265209461049658,
|
| 681 |
+
"loss": 3.1241,
|
| 682 |
+
"step": 91000
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 4.947300494730049,
|
| 686 |
+
"grad_norm": 0.3277081847190857,
|
| 687 |
+
"learning_rate": 0.0008235790774299835,
|
| 688 |
+
"loss": 3.1213,
|
| 689 |
+
"step": 92000
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 5.0,
|
| 693 |
+
"eval_accuracy": 0.4008789177643117,
|
| 694 |
+
"eval_loss": 3.396653652191162,
|
| 695 |
+
"eval_runtime": 155.4385,
|
| 696 |
+
"eval_samples_per_second": 372.604,
|
| 697 |
+
"eval_steps_per_second": 5.822,
|
| 698 |
+
"step": 92980
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 5.00107550010755,
|
| 702 |
+
"grad_norm": 0.20310255885124207,
|
| 703 |
+
"learning_rate": 0.0008206401506236762,
|
| 704 |
+
"loss": 3.1191,
|
| 705 |
+
"step": 93000
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 5.05485050548505,
|
| 709 |
+
"grad_norm": 0.20080606639385223,
|
| 710 |
+
"learning_rate": 0.0008176982819486937,
|
| 711 |
+
"loss": 3.0521,
|
| 712 |
+
"step": 94000
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 5.108625510862551,
|
| 716 |
+
"grad_norm": 0.21395424008369446,
|
| 717 |
+
"learning_rate": 0.0008147593551423864,
|
| 718 |
+
"loss": 3.0656,
|
| 719 |
+
"step": 95000
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 5.1624005162400515,
|
| 723 |
+
"grad_norm": 0.22563432157039642,
|
| 724 |
+
"learning_rate": 0.0008118174864674042,
|
| 725 |
+
"loss": 3.0641,
|
| 726 |
+
"step": 96000
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 5.2161755216175525,
|
| 730 |
+
"grad_norm": 0.21297597885131836,
|
| 731 |
+
"learning_rate": 0.0008088785596610967,
|
| 732 |
+
"loss": 3.07,
|
| 733 |
+
"step": 97000
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 5.269950526995053,
|
| 737 |
+
"grad_norm": 0.20302899181842804,
|
| 738 |
+
"learning_rate": 0.0008059366909861144,
|
| 739 |
+
"loss": 3.0717,
|
| 740 |
+
"step": 98000
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 5.323725532372554,
|
| 744 |
+
"grad_norm": 0.2152853012084961,
|
| 745 |
+
"learning_rate": 0.0008029948223111321,
|
| 746 |
+
"loss": 3.0759,
|
| 747 |
+
"step": 99000
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 5.377500537750054,
|
| 751 |
+
"grad_norm": 0.2148328423500061,
|
| 752 |
+
"learning_rate": 0.0008000529536361497,
|
| 753 |
+
"loss": 3.0728,
|
| 754 |
+
"step": 100000
|
| 755 |
+
},
|
| 756 |
+
{
|
| 757 |
+
"epoch": 5.431275543127555,
|
| 758 |
+
"grad_norm": 0.20232610404491425,
|
| 759 |
+
"learning_rate": 0.0007971140268298423,
|
| 760 |
+
"loss": 3.0798,
|
| 761 |
+
"step": 101000
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 5.485050548505055,
|
| 765 |
+
"grad_norm": 0.22732730209827423,
|
| 766 |
+
"learning_rate": 0.000794175100023535,
|
| 767 |
+
"loss": 3.0756,
|
| 768 |
+
"step": 102000
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 5.538825553882555,
|
| 772 |
+
"grad_norm": 0.2203952670097351,
|
| 773 |
+
"learning_rate": 0.0007912332313485526,
|
| 774 |
+
"loss": 3.0776,
|
| 775 |
+
"step": 103000
|
| 776 |
+
},
|
| 777 |
+
{
|
| 778 |
+
"epoch": 5.592600559260056,
|
| 779 |
+
"grad_norm": 0.21848390996456146,
|
| 780 |
+
"learning_rate": 0.0007882943045422453,
|
| 781 |
+
"loss": 3.0763,
|
| 782 |
+
"step": 104000
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 5.646375564637556,
|
| 786 |
+
"grad_norm": 0.22204072773456573,
|
| 787 |
+
"learning_rate": 0.0007853524358672629,
|
| 788 |
+
"loss": 3.0797,
|
| 789 |
+
"step": 105000
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 5.700150570015057,
|
| 793 |
+
"grad_norm": 0.20933043956756592,
|
| 794 |
+
"learning_rate": 0.0007824135090609555,
|
| 795 |
+
"loss": 3.0763,
|
| 796 |
+
"step": 106000
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"epoch": 5.753925575392557,
|
| 800 |
+
"grad_norm": 0.19925065338611603,
|
| 801 |
+
"learning_rate": 0.0007794716403859732,
|
| 802 |
+
"loss": 3.0802,
|
| 803 |
+
"step": 107000
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 5.807700580770058,
|
| 807 |
+
"grad_norm": 0.20748205482959747,
|
| 808 |
+
"learning_rate": 0.0007765297717109908,
|
| 809 |
+
"loss": 3.081,
|
| 810 |
+
"step": 108000
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 5.8614755861475585,
|
| 814 |
+
"grad_norm": 0.2089342474937439,
|
| 815 |
+
"learning_rate": 0.0007735908449046834,
|
| 816 |
+
"loss": 3.0787,
|
| 817 |
+
"step": 109000
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 5.9152505915250595,
|
| 821 |
+
"grad_norm": 0.20147345960140228,
|
| 822 |
+
"learning_rate": 0.0007706489762297011,
|
| 823 |
+
"loss": 3.0829,
|
| 824 |
+
"step": 110000
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 5.96902559690256,
|
| 828 |
+
"grad_norm": 0.2211214154958725,
|
| 829 |
+
"learning_rate": 0.0007677071075547188,
|
| 830 |
+
"loss": 3.0783,
|
| 831 |
+
"step": 111000
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 6.0,
|
| 835 |
+
"eval_accuracy": 0.405020628943781,
|
| 836 |
+
"eval_loss": 3.3772811889648438,
|
| 837 |
+
"eval_runtime": 156.2536,
|
| 838 |
+
"eval_samples_per_second": 370.66,
|
| 839 |
+
"eval_steps_per_second": 5.792,
|
| 840 |
+
"step": 111576
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 6.022800602280061,
|
| 844 |
+
"grad_norm": 0.21148885786533356,
|
| 845 |
+
"learning_rate": 0.0007647681807484113,
|
| 846 |
+
"loss": 3.0451,
|
| 847 |
+
"step": 112000
|
| 848 |
+
},
|
| 849 |
+
{
|
| 850 |
+
"epoch": 6.076575607657561,
|
| 851 |
+
"grad_norm": 0.2195775806903839,
|
| 852 |
+
"learning_rate": 0.000761826312073429,
|
| 853 |
+
"loss": 3.0147,
|
| 854 |
+
"step": 113000
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 6.130350613035061,
|
| 858 |
+
"grad_norm": 0.20522399246692657,
|
| 859 |
+
"learning_rate": 0.0007588873852671217,
|
| 860 |
+
"loss": 3.0194,
|
| 861 |
+
"step": 114000
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 6.184125618412562,
|
| 865 |
+
"grad_norm": 0.20723003149032593,
|
| 866 |
+
"learning_rate": 0.0007559455165921393,
|
| 867 |
+
"loss": 3.026,
|
| 868 |
+
"step": 115000
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 6.237900623790062,
|
| 872 |
+
"grad_norm": 0.23514005541801453,
|
| 873 |
+
"learning_rate": 0.000753003647917157,
|
| 874 |
+
"loss": 3.0231,
|
| 875 |
+
"step": 116000
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 6.291675629167563,
|
| 879 |
+
"grad_norm": 0.20580914616584778,
|
| 880 |
+
"learning_rate": 0.0007500647211108497,
|
| 881 |
+
"loss": 3.0321,
|
| 882 |
+
"step": 117000
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 6.345450634545063,
|
| 886 |
+
"grad_norm": 0.2240120768547058,
|
| 887 |
+
"learning_rate": 0.0007471228524358674,
|
| 888 |
+
"loss": 3.0332,
|
| 889 |
+
"step": 118000
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 6.399225639922564,
|
| 893 |
+
"grad_norm": 0.23184897005558014,
|
| 894 |
+
"learning_rate": 0.0007441839256295599,
|
| 895 |
+
"loss": 3.0369,
|
| 896 |
+
"step": 119000
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 6.453000645300064,
|
| 900 |
+
"grad_norm": 0.22646069526672363,
|
| 901 |
+
"learning_rate": 0.0007412449988232526,
|
| 902 |
+
"loss": 3.0357,
|
| 903 |
+
"step": 120000
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 6.506775650677565,
|
| 907 |
+
"grad_norm": 0.21927151083946228,
|
| 908 |
+
"learning_rate": 0.0007383031301482702,
|
| 909 |
+
"loss": 3.0398,
|
| 910 |
+
"step": 121000
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 6.5605506560550655,
|
| 914 |
+
"grad_norm": 0.24726586043834686,
|
| 915 |
+
"learning_rate": 0.0007353612614732878,
|
| 916 |
+
"loss": 3.0373,
|
| 917 |
+
"step": 122000
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 6.6143256614325665,
|
| 921 |
+
"grad_norm": 0.21686062216758728,
|
| 922 |
+
"learning_rate": 0.0007324193927983055,
|
| 923 |
+
"loss": 3.0399,
|
| 924 |
+
"step": 123000
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 6.668100666810067,
|
| 928 |
+
"grad_norm": 0.21142247319221497,
|
| 929 |
+
"learning_rate": 0.0007294804659919982,
|
| 930 |
+
"loss": 3.0446,
|
| 931 |
+
"step": 124000
|
| 932 |
+
},
|
| 933 |
+
{
|
| 934 |
+
"epoch": 6.721875672187567,
|
| 935 |
+
"grad_norm": 0.21460475027561188,
|
| 936 |
+
"learning_rate": 0.0007265385973170157,
|
| 937 |
+
"loss": 3.0403,
|
| 938 |
+
"step": 125000
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 6.775650677565068,
|
| 942 |
+
"grad_norm": 0.22398121654987335,
|
| 943 |
+
"learning_rate": 0.0007235967286420334,
|
| 944 |
+
"loss": 3.0426,
|
| 945 |
+
"step": 126000
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 6.829425682942568,
|
| 949 |
+
"grad_norm": 0.23123160004615784,
|
| 950 |
+
"learning_rate": 0.0007206548599670511,
|
| 951 |
+
"loss": 3.0443,
|
| 952 |
+
"step": 127000
|
| 953 |
+
},
|
| 954 |
+
{
|
| 955 |
+
"epoch": 6.883200688320069,
|
| 956 |
+
"grad_norm": 0.21254226565361023,
|
| 957 |
+
"learning_rate": 0.0007177159331607437,
|
| 958 |
+
"loss": 3.0424,
|
| 959 |
+
"step": 128000
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 6.936975693697569,
|
| 963 |
+
"grad_norm": 0.21302445232868195,
|
| 964 |
+
"learning_rate": 0.0007147740644857613,
|
| 965 |
+
"loss": 3.0472,
|
| 966 |
+
"step": 129000
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 6.99075069907507,
|
| 970 |
+
"grad_norm": 0.2217877358198166,
|
| 971 |
+
"learning_rate": 0.0007118321958107791,
|
| 972 |
+
"loss": 3.0456,
|
| 973 |
+
"step": 130000
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 7.0,
|
| 977 |
+
"eval_accuracy": 0.4055193299898036,
|
| 978 |
+
"eval_loss": 3.3825550079345703,
|
| 979 |
+
"eval_runtime": 155.2208,
|
| 980 |
+
"eval_samples_per_second": 373.126,
|
| 981 |
+
"eval_steps_per_second": 5.83,
|
| 982 |
+
"step": 130172
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 7.04452570445257,
|
| 986 |
+
"grad_norm": 0.23532521724700928,
|
| 987 |
+
"learning_rate": 0.0007088903271357967,
|
| 988 |
+
"loss": 2.9839,
|
| 989 |
+
"step": 131000
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 7.098300709830071,
|
| 993 |
+
"grad_norm": 0.22214515507221222,
|
| 994 |
+
"learning_rate": 0.0007059514003294893,
|
| 995 |
+
"loss": 2.9765,
|
| 996 |
+
"step": 132000
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 7.152075715207571,
|
| 1000 |
+
"grad_norm": 0.2383100390434265,
|
| 1001 |
+
"learning_rate": 0.0007030154153918568,
|
| 1002 |
+
"loss": 2.9843,
|
| 1003 |
+
"step": 133000
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 7.205850720585072,
|
| 1007 |
+
"grad_norm": 0.22472046315670013,
|
| 1008 |
+
"learning_rate": 0.0007000735467168746,
|
| 1009 |
+
"loss": 2.9925,
|
| 1010 |
+
"step": 134000
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 7.2596257259625725,
|
| 1014 |
+
"grad_norm": 0.26296138763427734,
|
| 1015 |
+
"learning_rate": 0.0006971316780418923,
|
| 1016 |
+
"loss": 2.997,
|
| 1017 |
+
"step": 135000
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 7.3134007313400735,
|
| 1021 |
+
"grad_norm": 0.2494724839925766,
|
| 1022 |
+
"learning_rate": 0.0006941898093669099,
|
| 1023 |
+
"loss": 2.997,
|
| 1024 |
+
"step": 136000
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 7.367175736717574,
|
| 1028 |
+
"grad_norm": 0.22137367725372314,
|
| 1029 |
+
"learning_rate": 0.0006912508825606025,
|
| 1030 |
+
"loss": 2.9973,
|
| 1031 |
+
"step": 137000
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 7.420950742095075,
|
| 1035 |
+
"grad_norm": 0.22704289853572845,
|
| 1036 |
+
"learning_rate": 0.0006883090138856202,
|
| 1037 |
+
"loss": 3.0066,
|
| 1038 |
+
"step": 138000
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 7.474725747472575,
|
| 1042 |
+
"grad_norm": 0.2145918905735016,
|
| 1043 |
+
"learning_rate": 0.0006853700870793128,
|
| 1044 |
+
"loss": 3.0054,
|
| 1045 |
+
"step": 139000
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 7.528500752850075,
|
| 1049 |
+
"grad_norm": 0.21607990562915802,
|
| 1050 |
+
"learning_rate": 0.0006824282184043304,
|
| 1051 |
+
"loss": 3.0018,
|
| 1052 |
+
"step": 140000
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 7.582275758227576,
|
| 1056 |
+
"grad_norm": 0.2057826817035675,
|
| 1057 |
+
"learning_rate": 0.0006794863497293481,
|
| 1058 |
+
"loss": 3.0101,
|
| 1059 |
+
"step": 141000
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 7.636050763605076,
|
| 1063 |
+
"grad_norm": 0.23032937943935394,
|
| 1064 |
+
"learning_rate": 0.0006765474229230408,
|
| 1065 |
+
"loss": 3.0099,
|
| 1066 |
+
"step": 142000
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 7.689825768982577,
|
| 1070 |
+
"grad_norm": 0.22495923936367035,
|
| 1071 |
+
"learning_rate": 0.0006736055542480583,
|
| 1072 |
+
"loss": 3.008,
|
| 1073 |
+
"step": 143000
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 7.743600774360077,
|
| 1077 |
+
"grad_norm": 0.2345353364944458,
|
| 1078 |
+
"learning_rate": 0.000670666627441751,
|
| 1079 |
+
"loss": 3.0099,
|
| 1080 |
+
"step": 144000
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 7.797375779737578,
|
| 1084 |
+
"grad_norm": 0.23005186021327972,
|
| 1085 |
+
"learning_rate": 0.0006677277006354437,
|
| 1086 |
+
"loss": 3.0125,
|
| 1087 |
+
"step": 145000
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 7.851150785115078,
|
| 1091 |
+
"grad_norm": 0.29431313276290894,
|
| 1092 |
+
"learning_rate": 0.0006647858319604612,
|
| 1093 |
+
"loss": 3.0107,
|
| 1094 |
+
"step": 146000
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 7.904925790492579,
|
| 1098 |
+
"grad_norm": 0.2245541363954544,
|
| 1099 |
+
"learning_rate": 0.0006618439632854789,
|
| 1100 |
+
"loss": 3.0133,
|
| 1101 |
+
"step": 147000
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 7.9587007958700795,
|
| 1105 |
+
"grad_norm": 0.22786079347133636,
|
| 1106 |
+
"learning_rate": 0.0006589020946104966,
|
| 1107 |
+
"loss": 3.0126,
|
| 1108 |
+
"step": 148000
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 8.0,
|
| 1112 |
+
"eval_accuracy": 0.40771172002548395,
|
| 1113 |
+
"eval_loss": 3.3547439575195312,
|
| 1114 |
+
"eval_runtime": 154.7201,
|
| 1115 |
+
"eval_samples_per_second": 374.334,
|
| 1116 |
+
"eval_steps_per_second": 5.849,
|
| 1117 |
+
"step": 148768
|
| 1118 |
+
},
|
| 1119 |
+
{
|
| 1120 |
+
"epoch": 8.01247580124758,
|
| 1121 |
+
"grad_norm": 0.2346237152814865,
|
| 1122 |
+
"learning_rate": 0.0006559602259355142,
|
| 1123 |
+
"loss": 2.9965,
|
| 1124 |
+
"step": 149000
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 8.066250806625082,
|
| 1128 |
+
"grad_norm": 0.23346728086471558,
|
| 1129 |
+
"learning_rate": 0.0006530212991292069,
|
| 1130 |
+
"loss": 2.9479,
|
| 1131 |
+
"step": 150000
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 8.12002581200258,
|
| 1135 |
+
"grad_norm": 0.24378512799739838,
|
| 1136 |
+
"learning_rate": 0.0006500794304542246,
|
| 1137 |
+
"loss": 2.9513,
|
| 1138 |
+
"step": 151000
|
| 1139 |
+
},
|
| 1140 |
+
{
|
| 1141 |
+
"epoch": 8.173800817380082,
|
| 1142 |
+
"grad_norm": 0.24439002573490143,
|
| 1143 |
+
"learning_rate": 0.0006471405036479172,
|
| 1144 |
+
"loss": 2.9601,
|
| 1145 |
+
"step": 152000
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 8.227575822757583,
|
| 1149 |
+
"grad_norm": 0.27081623673439026,
|
| 1150 |
+
"learning_rate": 0.0006441986349729348,
|
| 1151 |
+
"loss": 2.9631,
|
| 1152 |
+
"step": 153000
|
| 1153 |
+
},
|
| 1154 |
+
{
|
| 1155 |
+
"epoch": 8.281350828135082,
|
| 1156 |
+
"grad_norm": 0.2521245777606964,
|
| 1157 |
+
"learning_rate": 0.0006412626500353025,
|
| 1158 |
+
"loss": 2.966,
|
| 1159 |
+
"step": 154000
|
| 1160 |
+
},
|
| 1161 |
+
{
|
| 1162 |
+
"epoch": 8.335125833512583,
|
| 1163 |
+
"grad_norm": 0.21975190937519073,
|
| 1164 |
+
"learning_rate": 0.0006383207813603201,
|
| 1165 |
+
"loss": 2.9678,
|
| 1166 |
+
"step": 155000
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 8.388900838890084,
|
| 1170 |
+
"grad_norm": 0.2267887145280838,
|
| 1171 |
+
"learning_rate": 0.0006353789126853378,
|
| 1172 |
+
"loss": 2.9696,
|
| 1173 |
+
"step": 156000
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 8.442675844267585,
|
| 1177 |
+
"grad_norm": 0.218279168009758,
|
| 1178 |
+
"learning_rate": 0.0006324370440103554,
|
| 1179 |
+
"loss": 2.9713,
|
| 1180 |
+
"step": 157000
|
| 1181 |
+
},
|
| 1182 |
+
{
|
| 1183 |
+
"epoch": 8.496450849645084,
|
| 1184 |
+
"grad_norm": 0.23300865292549133,
|
| 1185 |
+
"learning_rate": 0.0006294951753353731,
|
| 1186 |
+
"loss": 2.9772,
|
| 1187 |
+
"step": 158000
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 8.550225855022585,
|
| 1191 |
+
"grad_norm": 0.21749693155288696,
|
| 1192 |
+
"learning_rate": 0.0006265562485290657,
|
| 1193 |
+
"loss": 2.9773,
|
| 1194 |
+
"step": 159000
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 8.604000860400086,
|
| 1198 |
+
"grad_norm": 0.26928380131721497,
|
| 1199 |
+
"learning_rate": 0.0006236143798540833,
|
| 1200 |
+
"loss": 2.978,
|
| 1201 |
+
"step": 160000
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 8.657775865777587,
|
| 1205 |
+
"grad_norm": 0.22122180461883545,
|
| 1206 |
+
"learning_rate": 0.000620672511179101,
|
| 1207 |
+
"loss": 2.9794,
|
| 1208 |
+
"step": 161000
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 8.711550871155087,
|
| 1212 |
+
"grad_norm": 0.22700442373752594,
|
| 1213 |
+
"learning_rate": 0.0006177306425041186,
|
| 1214 |
+
"loss": 2.9824,
|
| 1215 |
+
"step": 162000
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 8.765325876532588,
|
| 1219 |
+
"grad_norm": 0.2541004419326782,
|
| 1220 |
+
"learning_rate": 0.0006147917156978112,
|
| 1221 |
+
"loss": 2.9841,
|
| 1222 |
+
"step": 163000
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"epoch": 8.819100881910089,
|
| 1226 |
+
"grad_norm": 0.2551893889904022,
|
| 1227 |
+
"learning_rate": 0.0006118498470228289,
|
| 1228 |
+
"loss": 2.9837,
|
| 1229 |
+
"step": 164000
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 8.872875887287588,
|
| 1233 |
+
"grad_norm": 0.25604966282844543,
|
| 1234 |
+
"learning_rate": 0.0006089079783478466,
|
| 1235 |
+
"loss": 2.984,
|
| 1236 |
+
"step": 165000
|
| 1237 |
+
},
|
| 1238 |
+
{
|
| 1239 |
+
"epoch": 8.926650892665089,
|
| 1240 |
+
"grad_norm": 0.24571265280246735,
|
| 1241 |
+
"learning_rate": 0.0006059719934102142,
|
| 1242 |
+
"loss": 2.9849,
|
| 1243 |
+
"step": 166000
|
| 1244 |
+
},
|
| 1245 |
+
{
|
| 1246 |
+
"epoch": 8.98042589804259,
|
| 1247 |
+
"grad_norm": 0.24128392338752747,
|
| 1248 |
+
"learning_rate": 0.0006030301247352318,
|
| 1249 |
+
"loss": 2.9843,
|
| 1250 |
+
"step": 167000
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 9.0,
|
| 1254 |
+
"eval_accuracy": 0.4083424360998555,
|
| 1255 |
+
"eval_loss": 3.3613698482513428,
|
| 1256 |
+
"eval_runtime": 155.0364,
|
| 1257 |
+
"eval_samples_per_second": 373.57,
|
| 1258 |
+
"eval_steps_per_second": 5.837,
|
| 1259 |
+
"step": 167364
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 9.034200903420091,
|
| 1263 |
+
"grad_norm": 0.24964158236980438,
|
| 1264 |
+
"learning_rate": 0.0006000882560602495,
|
| 1265 |
+
"loss": 2.9432,
|
| 1266 |
+
"step": 168000
|
| 1267 |
+
},
|
| 1268 |
+
{
|
| 1269 |
+
"epoch": 9.08797590879759,
|
| 1270 |
+
"grad_norm": 0.2405262142419815,
|
| 1271 |
+
"learning_rate": 0.0005971463873852672,
|
| 1272 |
+
"loss": 2.922,
|
| 1273 |
+
"step": 169000
|
| 1274 |
+
},
|
| 1275 |
+
{
|
| 1276 |
+
"epoch": 9.141750914175091,
|
| 1277 |
+
"grad_norm": 0.22288870811462402,
|
| 1278 |
+
"learning_rate": 0.0005942045187102848,
|
| 1279 |
+
"loss": 2.9269,
|
| 1280 |
+
"step": 170000
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 9.195525919552592,
|
| 1284 |
+
"grad_norm": 0.3041359484195709,
|
| 1285 |
+
"learning_rate": 0.0005912626500353024,
|
| 1286 |
+
"loss": 2.9342,
|
| 1287 |
+
"step": 171000
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 9.249300924930093,
|
| 1291 |
+
"grad_norm": 0.22550632059574127,
|
| 1292 |
+
"learning_rate": 0.0005883237232289951,
|
| 1293 |
+
"loss": 2.9392,
|
| 1294 |
+
"step": 172000
|
| 1295 |
+
},
|
| 1296 |
+
{
|
| 1297 |
+
"epoch": 9.303075930307593,
|
| 1298 |
+
"grad_norm": 0.23584921658039093,
|
| 1299 |
+
"learning_rate": 0.0005853818545540127,
|
| 1300 |
+
"loss": 2.9422,
|
| 1301 |
+
"step": 173000
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 9.356850935685094,
|
| 1305 |
+
"grad_norm": 0.2634640634059906,
|
| 1306 |
+
"learning_rate": 0.0005824458696163803,
|
| 1307 |
+
"loss": 2.945,
|
| 1308 |
+
"step": 174000
|
| 1309 |
+
},
|
| 1310 |
+
{
|
| 1311 |
+
"epoch": 9.410625941062595,
|
| 1312 |
+
"grad_norm": 0.2354883849620819,
|
| 1313 |
+
"learning_rate": 0.000579504000941398,
|
| 1314 |
+
"loss": 2.9485,
|
| 1315 |
+
"step": 175000
|
| 1316 |
+
},
|
| 1317 |
+
{
|
| 1318 |
+
"epoch": 9.464400946440094,
|
| 1319 |
+
"grad_norm": 0.26491352915763855,
|
| 1320 |
+
"learning_rate": 0.0005765621322664157,
|
| 1321 |
+
"loss": 2.9475,
|
| 1322 |
+
"step": 176000
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 9.518175951817595,
|
| 1326 |
+
"grad_norm": 0.2462054342031479,
|
| 1327 |
+
"learning_rate": 0.0005736202635914332,
|
| 1328 |
+
"loss": 2.9482,
|
| 1329 |
+
"step": 177000
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 9.571950957195096,
|
| 1333 |
+
"grad_norm": 0.2643495202064514,
|
| 1334 |
+
"learning_rate": 0.0005706783949164509,
|
| 1335 |
+
"loss": 2.9502,
|
| 1336 |
+
"step": 178000
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"epoch": 9.625725962572597,
|
| 1340 |
+
"grad_norm": 0.24029669165611267,
|
| 1341 |
+
"learning_rate": 0.0005677365262414686,
|
| 1342 |
+
"loss": 2.9518,
|
| 1343 |
+
"step": 179000
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 9.679500967950096,
|
| 1347 |
+
"grad_norm": 0.2550260126590729,
|
| 1348 |
+
"learning_rate": 0.0005648005413038361,
|
| 1349 |
+
"loss": 2.9571,
|
| 1350 |
+
"step": 180000
|
| 1351 |
+
},
|
| 1352 |
+
{
|
| 1353 |
+
"epoch": 9.733275973327597,
|
| 1354 |
+
"grad_norm": 0.23675589263439178,
|
| 1355 |
+
"learning_rate": 0.0005618586726288538,
|
| 1356 |
+
"loss": 2.9543,
|
| 1357 |
+
"step": 181000
|
| 1358 |
+
},
|
| 1359 |
+
{
|
| 1360 |
+
"epoch": 9.787050978705098,
|
| 1361 |
+
"grad_norm": 0.2279191017150879,
|
| 1362 |
+
"learning_rate": 0.0005589197458225465,
|
| 1363 |
+
"loss": 2.9593,
|
| 1364 |
+
"step": 182000
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 9.8408259840826,
|
| 1368 |
+
"grad_norm": 0.27392587065696716,
|
| 1369 |
+
"learning_rate": 0.0005559778771475641,
|
| 1370 |
+
"loss": 2.9561,
|
| 1371 |
+
"step": 183000
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 9.894600989460098,
|
| 1375 |
+
"grad_norm": 0.2388741672039032,
|
| 1376 |
+
"learning_rate": 0.0005530360084725818,
|
| 1377 |
+
"loss": 2.9565,
|
| 1378 |
+
"step": 184000
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 9.9483759948376,
|
| 1382 |
+
"grad_norm": 0.2503463327884674,
|
| 1383 |
+
"learning_rate": 0.0005500970816662745,
|
| 1384 |
+
"loss": 2.9592,
|
| 1385 |
+
"step": 185000
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 10.0,
|
| 1389 |
+
"eval_accuracy": 0.40854156716551776,
|
| 1390 |
+
"eval_loss": 3.377901077270508,
|
| 1391 |
+
"eval_runtime": 154.6842,
|
| 1392 |
+
"eval_samples_per_second": 374.421,
|
| 1393 |
+
"eval_steps_per_second": 5.851,
|
| 1394 |
+
"step": 185960
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 10.0021510002151,
|
| 1398 |
+
"grad_norm": 0.2553974688053131,
|
| 1399 |
+
"learning_rate": 0.0005471552129912922,
|
| 1400 |
+
"loss": 2.9593,
|
| 1401 |
+
"step": 186000
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 10.055926005592601,
|
| 1405 |
+
"grad_norm": 0.25415316224098206,
|
| 1406 |
+
"learning_rate": 0.0005442133443163097,
|
| 1407 |
+
"loss": 2.8931,
|
| 1408 |
+
"step": 187000
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"epoch": 10.1097010109701,
|
| 1412 |
+
"grad_norm": 0.2477007806301117,
|
| 1413 |
+
"learning_rate": 0.0005412744175100024,
|
| 1414 |
+
"loss": 2.8999,
|
| 1415 |
+
"step": 188000
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 10.163476016347602,
|
| 1419 |
+
"grad_norm": 0.23852670192718506,
|
| 1420 |
+
"learning_rate": 0.0005383325488350201,
|
| 1421 |
+
"loss": 2.9028,
|
| 1422 |
+
"step": 189000
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 10.217251021725103,
|
| 1426 |
+
"grad_norm": 0.2484176605939865,
|
| 1427 |
+
"learning_rate": 0.0005353906801600376,
|
| 1428 |
+
"loss": 2.9117,
|
| 1429 |
+
"step": 190000
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 10.271026027102602,
|
| 1433 |
+
"grad_norm": 0.27494022250175476,
|
| 1434 |
+
"learning_rate": 0.0005324517533537303,
|
| 1435 |
+
"loss": 2.9133,
|
| 1436 |
+
"step": 191000
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 10.324801032480103,
|
| 1440 |
+
"grad_norm": 0.2577020823955536,
|
| 1441 |
+
"learning_rate": 0.000529509884678748,
|
| 1442 |
+
"loss": 2.9214,
|
| 1443 |
+
"step": 192000
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 10.378576037857604,
|
| 1447 |
+
"grad_norm": 0.2292626053094864,
|
| 1448 |
+
"learning_rate": 0.0005265680160037656,
|
| 1449 |
+
"loss": 2.9208,
|
| 1450 |
+
"step": 193000
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 10.432351043235105,
|
| 1454 |
+
"grad_norm": 0.267873615026474,
|
| 1455 |
+
"learning_rate": 0.0005236290891974582,
|
| 1456 |
+
"loss": 2.9213,
|
| 1457 |
+
"step": 194000
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 10.486126048612604,
|
| 1461 |
+
"grad_norm": 0.24749253690242767,
|
| 1462 |
+
"learning_rate": 0.0005206872205224759,
|
| 1463 |
+
"loss": 2.9239,
|
| 1464 |
+
"step": 195000
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 10.539901053990105,
|
| 1468 |
+
"grad_norm": 0.26779085397720337,
|
| 1469 |
+
"learning_rate": 0.0005177453518474935,
|
| 1470 |
+
"loss": 2.929,
|
| 1471 |
+
"step": 196000
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 10.593676059367606,
|
| 1475 |
+
"grad_norm": 0.2552465796470642,
|
| 1476 |
+
"learning_rate": 0.0005148064250411861,
|
| 1477 |
+
"loss": 2.9309,
|
| 1478 |
+
"step": 197000
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 10.647451064745107,
|
| 1482 |
+
"grad_norm": 0.2551726996898651,
|
| 1483 |
+
"learning_rate": 0.0005118645563662038,
|
| 1484 |
+
"loss": 2.9287,
|
| 1485 |
+
"step": 198000
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 10.701226070122607,
|
| 1489 |
+
"grad_norm": 0.24481531977653503,
|
| 1490 |
+
"learning_rate": 0.0005089256295598964,
|
| 1491 |
+
"loss": 2.9326,
|
| 1492 |
+
"step": 199000
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"epoch": 10.755001075500108,
|
| 1496 |
+
"grad_norm": 0.2609283924102783,
|
| 1497 |
+
"learning_rate": 0.0005059837608849142,
|
| 1498 |
+
"loss": 2.9332,
|
| 1499 |
+
"step": 200000
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 10.808776080877609,
|
| 1503 |
+
"grad_norm": 0.22893798351287842,
|
| 1504 |
+
"learning_rate": 0.0005030418922099318,
|
| 1505 |
+
"loss": 2.9361,
|
| 1506 |
+
"step": 201000
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 10.86255108625511,
|
| 1510 |
+
"grad_norm": 0.24516218900680542,
|
| 1511 |
+
"learning_rate": 0.0005001029654036244,
|
| 1512 |
+
"loss": 2.9358,
|
| 1513 |
+
"step": 202000
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 10.916326091632609,
|
| 1517 |
+
"grad_norm": 0.23790410161018372,
|
| 1518 |
+
"learning_rate": 0.0004971610967286421,
|
| 1519 |
+
"loss": 2.9344,
|
| 1520 |
+
"step": 203000
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 10.97010109701011,
|
| 1524 |
+
"grad_norm": 0.2549736797809601,
|
| 1525 |
+
"learning_rate": 0.0004942221699223347,
|
| 1526 |
+
"loss": 2.9367,
|
| 1527 |
+
"step": 204000
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 11.0,
|
| 1531 |
+
"eval_accuracy": 0.40993629082379995,
|
| 1532 |
+
"eval_loss": 3.3604304790496826,
|
| 1533 |
+
"eval_runtime": 154.4315,
|
| 1534 |
+
"eval_samples_per_second": 375.034,
|
| 1535 |
+
"eval_steps_per_second": 5.86,
|
| 1536 |
+
"step": 204556
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 11.02387610238761,
|
| 1540 |
+
"grad_norm": 0.2493738979101181,
|
| 1541 |
+
"learning_rate": 0.0004912803012473523,
|
| 1542 |
+
"loss": 2.9057,
|
| 1543 |
+
"step": 205000
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 11.07765110776511,
|
| 1547 |
+
"grad_norm": 0.245314821600914,
|
| 1548 |
+
"learning_rate": 0.00048833843257237,
|
| 1549 |
+
"loss": 2.8775,
|
| 1550 |
+
"step": 206000
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 11.131426113142611,
|
| 1554 |
+
"grad_norm": 0.24536246061325073,
|
| 1555 |
+
"learning_rate": 0.0004853995057660626,
|
| 1556 |
+
"loss": 2.8783,
|
| 1557 |
+
"step": 207000
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 11.185201118520112,
|
| 1561 |
+
"grad_norm": 0.26116108894348145,
|
| 1562 |
+
"learning_rate": 0.00048245763709108024,
|
| 1563 |
+
"loss": 2.8851,
|
| 1564 |
+
"step": 208000
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 11.238976123897613,
|
| 1568 |
+
"grad_norm": 0.24136824905872345,
|
| 1569 |
+
"learning_rate": 0.0004795187102847729,
|
| 1570 |
+
"loss": 2.8918,
|
| 1571 |
+
"step": 209000
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 11.292751129275112,
|
| 1575 |
+
"grad_norm": 0.275006502866745,
|
| 1576 |
+
"learning_rate": 0.0004765768416097905,
|
| 1577 |
+
"loss": 2.8951,
|
| 1578 |
+
"step": 210000
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 11.346526134652613,
|
| 1582 |
+
"grad_norm": 0.27772292494773865,
|
| 1583 |
+
"learning_rate": 0.0004736349729348082,
|
| 1584 |
+
"loss": 2.8973,
|
| 1585 |
+
"step": 211000
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 11.400301140030114,
|
| 1589 |
+
"grad_norm": 0.2865879237651825,
|
| 1590 |
+
"learning_rate": 0.0004706960461285008,
|
| 1591 |
+
"loss": 2.8977,
|
| 1592 |
+
"step": 212000
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 11.454076145407615,
|
| 1596 |
+
"grad_norm": 0.24032117426395416,
|
| 1597 |
+
"learning_rate": 0.00046775711932219347,
|
| 1598 |
+
"loss": 2.9002,
|
| 1599 |
+
"step": 213000
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 11.507851150785115,
|
| 1603 |
+
"grad_norm": 0.24863946437835693,
|
| 1604 |
+
"learning_rate": 0.00046481525064721114,
|
| 1605 |
+
"loss": 2.9052,
|
| 1606 |
+
"step": 214000
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 11.561626156162616,
|
| 1610 |
+
"grad_norm": 0.25115910172462463,
|
| 1611 |
+
"learning_rate": 0.00046187338197222876,
|
| 1612 |
+
"loss": 2.9081,
|
| 1613 |
+
"step": 215000
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 11.615401161540117,
|
| 1617 |
+
"grad_norm": 0.25204330682754517,
|
| 1618 |
+
"learning_rate": 0.00045893151329724644,
|
| 1619 |
+
"loss": 2.906,
|
| 1620 |
+
"step": 216000
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 11.669176166917616,
|
| 1624 |
+
"grad_norm": 0.25553619861602783,
|
| 1625 |
+
"learning_rate": 0.00045599258649093905,
|
| 1626 |
+
"loss": 2.9074,
|
| 1627 |
+
"step": 217000
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 11.722951172295117,
|
| 1631 |
+
"grad_norm": 0.24004510045051575,
|
| 1632 |
+
"learning_rate": 0.00045305071781595667,
|
| 1633 |
+
"loss": 2.9105,
|
| 1634 |
+
"step": 218000
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 11.776726177672618,
|
| 1638 |
+
"grad_norm": 0.28220391273498535,
|
| 1639 |
+
"learning_rate": 0.00045010884914097434,
|
| 1640 |
+
"loss": 2.9075,
|
| 1641 |
+
"step": 219000
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 11.830501183050119,
|
| 1645 |
+
"grad_norm": 0.2582526206970215,
|
| 1646 |
+
"learning_rate": 0.00044716992233466695,
|
| 1647 |
+
"loss": 2.9129,
|
| 1648 |
+
"step": 220000
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 11.884276188427618,
|
| 1652 |
+
"grad_norm": 0.27006328105926514,
|
| 1653 |
+
"learning_rate": 0.0004442309955283596,
|
| 1654 |
+
"loss": 2.9146,
|
| 1655 |
+
"step": 221000
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 11.93805119380512,
|
| 1659 |
+
"grad_norm": 0.3025253415107727,
|
| 1660 |
+
"learning_rate": 0.0004412891268533773,
|
| 1661 |
+
"loss": 2.9161,
|
| 1662 |
+
"step": 222000
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 11.99182619918262,
|
| 1666 |
+
"grad_norm": 0.2534899115562439,
|
| 1667 |
+
"learning_rate": 0.00043834725817839496,
|
| 1668 |
+
"loss": 2.9145,
|
| 1669 |
+
"step": 223000
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 12.0,
|
| 1673 |
+
"eval_accuracy": 0.40973608482660917,
|
| 1674 |
+
"eval_loss": 3.3759312629699707,
|
| 1675 |
+
"eval_runtime": 154.6971,
|
| 1676 |
+
"eval_samples_per_second": 374.39,
|
| 1677 |
+
"eval_steps_per_second": 5.85,
|
| 1678 |
+
"step": 223152
|
| 1679 |
+
},
|
| 1680 |
+
{
|
| 1681 |
+
"epoch": 12.045601204560121,
|
| 1682 |
+
"grad_norm": 0.27566835284233093,
|
| 1683 |
+
"learning_rate": 0.0004354053895034126,
|
| 1684 |
+
"loss": 2.8626,
|
| 1685 |
+
"step": 224000
|
| 1686 |
+
},
|
| 1687 |
+
{
|
| 1688 |
+
"epoch": 12.09937620993762,
|
| 1689 |
+
"grad_norm": 0.243771031498909,
|
| 1690 |
+
"learning_rate": 0.0004324664626971052,
|
| 1691 |
+
"loss": 2.8583,
|
| 1692 |
+
"step": 225000
|
| 1693 |
+
},
|
| 1694 |
+
{
|
| 1695 |
+
"epoch": 12.153151215315122,
|
| 1696 |
+
"grad_norm": 0.26204991340637207,
|
| 1697 |
+
"learning_rate": 0.00042952459402212286,
|
| 1698 |
+
"loss": 2.8654,
|
| 1699 |
+
"step": 226000
|
| 1700 |
+
},
|
| 1701 |
+
{
|
| 1702 |
+
"epoch": 12.206926220692623,
|
| 1703 |
+
"grad_norm": 0.2582913637161255,
|
| 1704 |
+
"learning_rate": 0.0004265827253471405,
|
| 1705 |
+
"loss": 2.8673,
|
| 1706 |
+
"step": 227000
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 12.260701226070122,
|
| 1710 |
+
"grad_norm": 0.27284786105155945,
|
| 1711 |
+
"learning_rate": 0.00042364674040950813,
|
| 1712 |
+
"loss": 2.8735,
|
| 1713 |
+
"step": 228000
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 12.314476231447623,
|
| 1717 |
+
"grad_norm": 0.25654593110084534,
|
| 1718 |
+
"learning_rate": 0.00042070487173452575,
|
| 1719 |
+
"loss": 2.8751,
|
| 1720 |
+
"step": 229000
|
| 1721 |
+
},
|
| 1722 |
+
{
|
| 1723 |
+
"epoch": 12.368251236825124,
|
| 1724 |
+
"grad_norm": 0.23449215292930603,
|
| 1725 |
+
"learning_rate": 0.0004177630030595435,
|
| 1726 |
+
"loss": 2.8769,
|
| 1727 |
+
"step": 230000
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 12.422026242202625,
|
| 1731 |
+
"grad_norm": 0.24525216221809387,
|
| 1732 |
+
"learning_rate": 0.00041482407625323604,
|
| 1733 |
+
"loss": 2.8763,
|
| 1734 |
+
"step": 231000
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 12.475801247580124,
|
| 1738 |
+
"grad_norm": 0.26022928953170776,
|
| 1739 |
+
"learning_rate": 0.0004118822075782537,
|
| 1740 |
+
"loss": 2.8811,
|
| 1741 |
+
"step": 232000
|
| 1742 |
+
},
|
| 1743 |
+
{
|
| 1744 |
+
"epoch": 12.529576252957625,
|
| 1745 |
+
"grad_norm": 0.24519123136997223,
|
| 1746 |
+
"learning_rate": 0.0004089403389032714,
|
| 1747 |
+
"loss": 2.8844,
|
| 1748 |
+
"step": 233000
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 12.583351258335126,
|
| 1752 |
+
"grad_norm": 0.3118698298931122,
|
| 1753 |
+
"learning_rate": 0.000405998470228289,
|
| 1754 |
+
"loss": 2.8837,
|
| 1755 |
+
"step": 234000
|
| 1756 |
+
},
|
| 1757 |
+
{
|
| 1758 |
+
"epoch": 12.637126263712627,
|
| 1759 |
+
"grad_norm": 0.28287139534950256,
|
| 1760 |
+
"learning_rate": 0.0004030566015533066,
|
| 1761 |
+
"loss": 2.8843,
|
| 1762 |
+
"step": 235000
|
| 1763 |
+
},
|
| 1764 |
+
{
|
| 1765 |
+
"epoch": 12.690901269090126,
|
| 1766 |
+
"grad_norm": 0.24490605294704437,
|
| 1767 |
+
"learning_rate": 0.0004001176747469993,
|
| 1768 |
+
"loss": 2.8868,
|
| 1769 |
+
"step": 236000
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 12.744676274467627,
|
| 1773 |
+
"grad_norm": 0.2810141444206238,
|
| 1774 |
+
"learning_rate": 0.00039717580607201696,
|
| 1775 |
+
"loss": 2.8905,
|
| 1776 |
+
"step": 237000
|
| 1777 |
+
},
|
| 1778 |
+
{
|
| 1779 |
+
"epoch": 12.798451279845128,
|
| 1780 |
+
"grad_norm": 0.2660733759403229,
|
| 1781 |
+
"learning_rate": 0.0003942368792657096,
|
| 1782 |
+
"loss": 2.8917,
|
| 1783 |
+
"step": 238000
|
| 1784 |
+
},
|
| 1785 |
+
{
|
| 1786 |
+
"epoch": 12.852226285222628,
|
| 1787 |
+
"grad_norm": 0.2440771758556366,
|
| 1788 |
+
"learning_rate": 0.00039129501059072724,
|
| 1789 |
+
"loss": 2.888,
|
| 1790 |
+
"step": 239000
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 12.906001290600129,
|
| 1794 |
+
"grad_norm": 0.25520211458206177,
|
| 1795 |
+
"learning_rate": 0.0003883560837844199,
|
| 1796 |
+
"loss": 2.8938,
|
| 1797 |
+
"step": 240000
|
| 1798 |
+
},
|
| 1799 |
+
{
|
| 1800 |
+
"epoch": 12.95977629597763,
|
| 1801 |
+
"grad_norm": 0.2661096751689911,
|
| 1802 |
+
"learning_rate": 0.0003854142151094375,
|
| 1803 |
+
"loss": 2.8924,
|
| 1804 |
+
"step": 241000
|
| 1805 |
+
},
|
| 1806 |
+
{
|
| 1807 |
+
"epoch": 13.0,
|
| 1808 |
+
"eval_accuracy": 0.4095556978794759,
|
| 1809 |
+
"eval_loss": 3.3856160640716553,
|
| 1810 |
+
"eval_runtime": 154.4788,
|
| 1811 |
+
"eval_samples_per_second": 374.919,
|
| 1812 |
+
"eval_steps_per_second": 5.858,
|
| 1813 |
+
"step": 241748
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 13.01355130135513,
|
| 1817 |
+
"grad_norm": 0.29137110710144043,
|
| 1818 |
+
"learning_rate": 0.00038247234643445514,
|
| 1819 |
+
"loss": 2.8796,
|
| 1820 |
+
"step": 242000
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 13.06732630673263,
|
| 1824 |
+
"grad_norm": 0.26854678988456726,
|
| 1825 |
+
"learning_rate": 0.0003795304777594728,
|
| 1826 |
+
"loss": 2.837,
|
| 1827 |
+
"step": 243000
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 13.121101312110131,
|
| 1831 |
+
"grad_norm": 0.2614552974700928,
|
| 1832 |
+
"learning_rate": 0.00037659155095316543,
|
| 1833 |
+
"loss": 2.8409,
|
| 1834 |
+
"step": 244000
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 13.174876317487632,
|
| 1838 |
+
"grad_norm": 0.3031397759914398,
|
| 1839 |
+
"learning_rate": 0.00037364968227818316,
|
| 1840 |
+
"loss": 2.8458,
|
| 1841 |
+
"step": 245000
|
| 1842 |
+
},
|
| 1843 |
+
{
|
| 1844 |
+
"epoch": 13.228651322865133,
|
| 1845 |
+
"grad_norm": 0.2507542073726654,
|
| 1846 |
+
"learning_rate": 0.0003707107554718757,
|
| 1847 |
+
"loss": 2.8537,
|
| 1848 |
+
"step": 246000
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 13.282426328242632,
|
| 1852 |
+
"grad_norm": 0.2524189054965973,
|
| 1853 |
+
"learning_rate": 0.0003677688867968934,
|
| 1854 |
+
"loss": 2.852,
|
| 1855 |
+
"step": 247000
|
| 1856 |
+
},
|
| 1857 |
+
{
|
| 1858 |
+
"epoch": 13.336201333620133,
|
| 1859 |
+
"grad_norm": 0.27850231528282166,
|
| 1860 |
+
"learning_rate": 0.00036482701812191106,
|
| 1861 |
+
"loss": 2.8574,
|
| 1862 |
+
"step": 248000
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 13.389976338997634,
|
| 1866 |
+
"grad_norm": 0.2408652901649475,
|
| 1867 |
+
"learning_rate": 0.00036188809131560367,
|
| 1868 |
+
"loss": 2.8563,
|
| 1869 |
+
"step": 249000
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 13.443751344375134,
|
| 1873 |
+
"grad_norm": 0.28609830141067505,
|
| 1874 |
+
"learning_rate": 0.00035894622264062134,
|
| 1875 |
+
"loss": 2.8627,
|
| 1876 |
+
"step": 250000
|
| 1877 |
+
},
|
| 1878 |
+
{
|
| 1879 |
+
"epoch": 13.497526349752635,
|
| 1880 |
+
"grad_norm": 0.2690850496292114,
|
| 1881 |
+
"learning_rate": 0.00035600435396563896,
|
| 1882 |
+
"loss": 2.8619,
|
| 1883 |
+
"step": 251000
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 13.551301355130136,
|
| 1887 |
+
"grad_norm": 0.2862522304058075,
|
| 1888 |
+
"learning_rate": 0.0003530654271593316,
|
| 1889 |
+
"loss": 2.8646,
|
| 1890 |
+
"step": 252000
|
| 1891 |
+
},
|
| 1892 |
+
{
|
| 1893 |
+
"epoch": 13.605076360507637,
|
| 1894 |
+
"grad_norm": 0.2629512548446655,
|
| 1895 |
+
"learning_rate": 0.0003501235584843493,
|
| 1896 |
+
"loss": 2.865,
|
| 1897 |
+
"step": 253000
|
| 1898 |
+
},
|
| 1899 |
+
{
|
| 1900 |
+
"epoch": 13.658851365885136,
|
| 1901 |
+
"grad_norm": 0.2542857825756073,
|
| 1902 |
+
"learning_rate": 0.00034718463167804185,
|
| 1903 |
+
"loss": 2.865,
|
| 1904 |
+
"step": 254000
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 13.712626371262637,
|
| 1908 |
+
"grad_norm": 0.258798211812973,
|
| 1909 |
+
"learning_rate": 0.0003442427630030596,
|
| 1910 |
+
"loss": 2.8686,
|
| 1911 |
+
"step": 255000
|
| 1912 |
+
},
|
| 1913 |
+
{
|
| 1914 |
+
"epoch": 13.766401376640138,
|
| 1915 |
+
"grad_norm": 0.2492339164018631,
|
| 1916 |
+
"learning_rate": 0.0003413008943280772,
|
| 1917 |
+
"loss": 2.8689,
|
| 1918 |
+
"step": 256000
|
| 1919 |
+
},
|
| 1920 |
+
{
|
| 1921 |
+
"epoch": 13.820176382017639,
|
| 1922 |
+
"grad_norm": 0.29779887199401855,
|
| 1923 |
+
"learning_rate": 0.0003383590256530948,
|
| 1924 |
+
"loss": 2.8691,
|
| 1925 |
+
"step": 257000
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"epoch": 13.873951387395138,
|
| 1929 |
+
"grad_norm": 0.2670515179634094,
|
| 1930 |
+
"learning_rate": 0.0003354200988467875,
|
| 1931 |
+
"loss": 2.8711,
|
| 1932 |
+
"step": 258000
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"epoch": 13.92772639277264,
|
| 1936 |
+
"grad_norm": 0.26957201957702637,
|
| 1937 |
+
"learning_rate": 0.0003324782301718051,
|
| 1938 |
+
"loss": 2.8711,
|
| 1939 |
+
"step": 259000
|
| 1940 |
+
},
|
| 1941 |
+
{
|
| 1942 |
+
"epoch": 13.98150139815014,
|
| 1943 |
+
"grad_norm": 0.2824675142765045,
|
| 1944 |
+
"learning_rate": 0.00032953636149682283,
|
| 1945 |
+
"loss": 2.8757,
|
| 1946 |
+
"step": 260000
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 14.0,
|
| 1950 |
+
"eval_accuracy": 0.410457699798363,
|
| 1951 |
+
"eval_loss": 3.384411573410034,
|
| 1952 |
+
"eval_runtime": 154.9443,
|
| 1953 |
+
"eval_samples_per_second": 373.792,
|
| 1954 |
+
"eval_steps_per_second": 5.841,
|
| 1955 |
+
"step": 260344
|
| 1956 |
+
},
|
| 1957 |
+
{
|
| 1958 |
+
"epoch": 14.035276403527641,
|
| 1959 |
+
"grad_norm": 0.29713505506515503,
|
| 1960 |
+
"learning_rate": 0.00032659449282184045,
|
| 1961 |
+
"loss": 2.8354,
|
| 1962 |
+
"step": 261000
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 14.08905140890514,
|
| 1966 |
+
"grad_norm": 0.27730679512023926,
|
| 1967 |
+
"learning_rate": 0.00032365850788420805,
|
| 1968 |
+
"loss": 2.8198,
|
| 1969 |
+
"step": 262000
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 14.142826414282641,
|
| 1973 |
+
"grad_norm": 0.29409125447273254,
|
| 1974 |
+
"learning_rate": 0.0003207166392092257,
|
| 1975 |
+
"loss": 2.824,
|
| 1976 |
+
"step": 263000
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 14.196601419660142,
|
| 1980 |
+
"grad_norm": 0.280676931142807,
|
| 1981 |
+
"learning_rate": 0.00031777477053424334,
|
| 1982 |
+
"loss": 2.8301,
|
| 1983 |
+
"step": 264000
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 14.250376425037642,
|
| 1987 |
+
"grad_norm": 0.3202875554561615,
|
| 1988 |
+
"learning_rate": 0.000314835843727936,
|
| 1989 |
+
"loss": 2.8271,
|
| 1990 |
+
"step": 265000
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 14.304151430415143,
|
| 1994 |
+
"grad_norm": 0.2673070728778839,
|
| 1995 |
+
"learning_rate": 0.0003118939750529536,
|
| 1996 |
+
"loss": 2.8376,
|
| 1997 |
+
"step": 266000
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 14.357926435792644,
|
| 2001 |
+
"grad_norm": 0.2868790030479431,
|
| 2002 |
+
"learning_rate": 0.0003089550482466463,
|
| 2003 |
+
"loss": 2.8416,
|
| 2004 |
+
"step": 267000
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 14.411701441170145,
|
| 2008 |
+
"grad_norm": 0.2689562737941742,
|
| 2009 |
+
"learning_rate": 0.0003060131795716639,
|
| 2010 |
+
"loss": 2.8408,
|
| 2011 |
+
"step": 268000
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 14.465476446547644,
|
| 2015 |
+
"grad_norm": 0.2907434403896332,
|
| 2016 |
+
"learning_rate": 0.0003030713108966815,
|
| 2017 |
+
"loss": 2.8429,
|
| 2018 |
+
"step": 269000
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 14.519251451925145,
|
| 2022 |
+
"grad_norm": 0.26338914036750793,
|
| 2023 |
+
"learning_rate": 0.00030012944222169925,
|
| 2024 |
+
"loss": 2.8465,
|
| 2025 |
+
"step": 270000
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 14.573026457302646,
|
| 2029 |
+
"grad_norm": 0.2925278842449188,
|
| 2030 |
+
"learning_rate": 0.00029719051541539186,
|
| 2031 |
+
"loss": 2.8444,
|
| 2032 |
+
"step": 271000
|
| 2033 |
+
},
|
| 2034 |
+
{
|
| 2035 |
+
"epoch": 14.626801462680147,
|
| 2036 |
+
"grad_norm": 0.2883965075016022,
|
| 2037 |
+
"learning_rate": 0.00029424864674040954,
|
| 2038 |
+
"loss": 2.8466,
|
| 2039 |
+
"step": 272000
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 14.680576468057646,
|
| 2043 |
+
"grad_norm": 0.28422123193740845,
|
| 2044 |
+
"learning_rate": 0.00029130971993410215,
|
| 2045 |
+
"loss": 2.8458,
|
| 2046 |
+
"step": 273000
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 14.734351473435147,
|
| 2050 |
+
"grad_norm": 0.28057223558425903,
|
| 2051 |
+
"learning_rate": 0.0002883678512591198,
|
| 2052 |
+
"loss": 2.8512,
|
| 2053 |
+
"step": 274000
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 14.788126478812648,
|
| 2057 |
+
"grad_norm": 0.2739641070365906,
|
| 2058 |
+
"learning_rate": 0.00028542892445281243,
|
| 2059 |
+
"loss": 2.8508,
|
| 2060 |
+
"step": 275000
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 14.84190148419015,
|
| 2064 |
+
"grad_norm": 0.26283150911331177,
|
| 2065 |
+
"learning_rate": 0.00028248705577783005,
|
| 2066 |
+
"loss": 2.8491,
|
| 2067 |
+
"step": 276000
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 14.895676489567649,
|
| 2071 |
+
"grad_norm": 0.25209805369377136,
|
| 2072 |
+
"learning_rate": 0.0002795451871028477,
|
| 2073 |
+
"loss": 2.8528,
|
| 2074 |
+
"step": 277000
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 14.94945149494515,
|
| 2078 |
+
"grad_norm": 0.29996606707572937,
|
| 2079 |
+
"learning_rate": 0.0002766033184278654,
|
| 2080 |
+
"loss": 2.8545,
|
| 2081 |
+
"step": 278000
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 15.0,
|
| 2085 |
+
"eval_accuracy": 0.4106898178253074,
|
| 2086 |
+
"eval_loss": 3.383195400238037,
|
| 2087 |
+
"eval_runtime": 154.9853,
|
| 2088 |
+
"eval_samples_per_second": 373.693,
|
| 2089 |
+
"eval_steps_per_second": 5.839,
|
| 2090 |
+
"step": 278940
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 15.00322650032265,
|
| 2094 |
+
"grad_norm": 0.30987992882728577,
|
| 2095 |
+
"learning_rate": 0.00027366439162155806,
|
| 2096 |
+
"loss": 2.8502,
|
| 2097 |
+
"step": 279000
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 15.05700150570015,
|
| 2101 |
+
"grad_norm": 0.2791779041290283,
|
| 2102 |
+
"learning_rate": 0.0002707225229465757,
|
| 2103 |
+
"loss": 2.8054,
|
| 2104 |
+
"step": 280000
|
| 2105 |
+
},
|
| 2106 |
+
{
|
| 2107 |
+
"epoch": 15.11077651107765,
|
| 2108 |
+
"grad_norm": 0.29352006316185,
|
| 2109 |
+
"learning_rate": 0.00026778359614026834,
|
| 2110 |
+
"loss": 2.8087,
|
| 2111 |
+
"step": 281000
|
| 2112 |
+
},
|
| 2113 |
+
{
|
| 2114 |
+
"epoch": 15.164551516455152,
|
| 2115 |
+
"grad_norm": 0.2739843428134918,
|
| 2116 |
+
"learning_rate": 0.00026484466933396095,
|
| 2117 |
+
"loss": 2.8134,
|
| 2118 |
+
"step": 282000
|
| 2119 |
+
},
|
| 2120 |
+
{
|
| 2121 |
+
"epoch": 15.218326521832653,
|
| 2122 |
+
"grad_norm": 0.29397326707839966,
|
| 2123 |
+
"learning_rate": 0.00026190280065897857,
|
| 2124 |
+
"loss": 2.814,
|
| 2125 |
+
"step": 283000
|
| 2126 |
+
},
|
| 2127 |
+
{
|
| 2128 |
+
"epoch": 15.272101527210152,
|
| 2129 |
+
"grad_norm": 0.2891228199005127,
|
| 2130 |
+
"learning_rate": 0.00025896387385267123,
|
| 2131 |
+
"loss": 2.8154,
|
| 2132 |
+
"step": 284000
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 15.325876532587653,
|
| 2136 |
+
"grad_norm": 0.2697419226169586,
|
| 2137 |
+
"learning_rate": 0.00025602200517768885,
|
| 2138 |
+
"loss": 2.815,
|
| 2139 |
+
"step": 285000
|
| 2140 |
+
},
|
| 2141 |
+
{
|
| 2142 |
+
"epoch": 15.379651537965154,
|
| 2143 |
+
"grad_norm": 0.27169767022132874,
|
| 2144 |
+
"learning_rate": 0.00025308013650270653,
|
| 2145 |
+
"loss": 2.8179,
|
| 2146 |
+
"step": 286000
|
| 2147 |
+
},
|
| 2148 |
+
{
|
| 2149 |
+
"epoch": 15.433426543342655,
|
| 2150 |
+
"grad_norm": 0.28420206904411316,
|
| 2151 |
+
"learning_rate": 0.0002501382678277242,
|
| 2152 |
+
"loss": 2.8242,
|
| 2153 |
+
"step": 287000
|
| 2154 |
+
},
|
| 2155 |
+
{
|
| 2156 |
+
"epoch": 15.487201548720154,
|
| 2157 |
+
"grad_norm": 0.29990944266319275,
|
| 2158 |
+
"learning_rate": 0.0002471993410214168,
|
| 2159 |
+
"loss": 2.8284,
|
| 2160 |
+
"step": 288000
|
| 2161 |
+
},
|
| 2162 |
+
{
|
| 2163 |
+
"epoch": 15.540976554097655,
|
| 2164 |
+
"grad_norm": 0.29358208179473877,
|
| 2165 |
+
"learning_rate": 0.0002442574723464345,
|
| 2166 |
+
"loss": 2.8242,
|
| 2167 |
+
"step": 289000
|
| 2168 |
+
},
|
| 2169 |
+
{
|
| 2170 |
+
"epoch": 15.594751559475156,
|
| 2171 |
+
"grad_norm": 0.2709376811981201,
|
| 2172 |
+
"learning_rate": 0.0002413156036714521,
|
| 2173 |
+
"loss": 2.8323,
|
| 2174 |
+
"step": 290000
|
| 2175 |
+
},
|
| 2176 |
+
{
|
| 2177 |
+
"epoch": 15.648526564852656,
|
| 2178 |
+
"grad_norm": 0.27639514207839966,
|
| 2179 |
+
"learning_rate": 0.00023837961873381973,
|
| 2180 |
+
"loss": 2.8295,
|
| 2181 |
+
"step": 291000
|
| 2182 |
+
},
|
| 2183 |
+
{
|
| 2184 |
+
"epoch": 15.702301570230157,
|
| 2185 |
+
"grad_norm": 0.27499568462371826,
|
| 2186 |
+
"learning_rate": 0.00023543775005883738,
|
| 2187 |
+
"loss": 2.8273,
|
| 2188 |
+
"step": 292000
|
| 2189 |
+
},
|
| 2190 |
+
{
|
| 2191 |
+
"epoch": 15.756076575607658,
|
| 2192 |
+
"grad_norm": 0.2614821493625641,
|
| 2193 |
+
"learning_rate": 0.00023249588138385502,
|
| 2194 |
+
"loss": 2.8302,
|
| 2195 |
+
"step": 293000
|
| 2196 |
+
},
|
| 2197 |
+
{
|
| 2198 |
+
"epoch": 15.809851580985159,
|
| 2199 |
+
"grad_norm": 0.30566710233688354,
|
| 2200 |
+
"learning_rate": 0.00022955695457754766,
|
| 2201 |
+
"loss": 2.8315,
|
| 2202 |
+
"step": 294000
|
| 2203 |
+
},
|
| 2204 |
+
{
|
| 2205 |
+
"epoch": 15.863626586362658,
|
| 2206 |
+
"grad_norm": 0.28071853518486023,
|
| 2207 |
+
"learning_rate": 0.00022661508590256533,
|
| 2208 |
+
"loss": 2.8282,
|
| 2209 |
+
"step": 295000
|
| 2210 |
+
},
|
| 2211 |
+
{
|
| 2212 |
+
"epoch": 15.917401591740159,
|
| 2213 |
+
"grad_norm": 0.26871344447135925,
|
| 2214 |
+
"learning_rate": 0.00022367321722758295,
|
| 2215 |
+
"loss": 2.8318,
|
| 2216 |
+
"step": 296000
|
| 2217 |
+
},
|
| 2218 |
+
{
|
| 2219 |
+
"epoch": 15.97117659711766,
|
| 2220 |
+
"grad_norm": 0.27507010102272034,
|
| 2221 |
+
"learning_rate": 0.0002207313485526006,
|
| 2222 |
+
"loss": 2.8339,
|
| 2223 |
+
"step": 297000
|
| 2224 |
+
},
|
| 2225 |
+
{
|
| 2226 |
+
"epoch": 16.0,
|
| 2227 |
+
"eval_accuracy": 0.4098433764298017,
|
| 2228 |
+
"eval_loss": 3.4079394340515137,
|
| 2229 |
+
"eval_runtime": 155.0447,
|
| 2230 |
+
"eval_samples_per_second": 373.55,
|
| 2231 |
+
"eval_steps_per_second": 5.837,
|
| 2232 |
+
"step": 297536
|
| 2233 |
+
},
|
| 2234 |
+
{
|
| 2235 |
+
"epoch": 16.02495160249516,
|
| 2236 |
+
"grad_norm": 0.2846646010875702,
|
| 2237 |
+
"learning_rate": 0.00021778947987761827,
|
| 2238 |
+
"loss": 2.8126,
|
| 2239 |
+
"step": 298000
|
| 2240 |
+
},
|
| 2241 |
+
{
|
| 2242 |
+
"epoch": 16.07872660787266,
|
| 2243 |
+
"grad_norm": 0.30357855558395386,
|
| 2244 |
+
"learning_rate": 0.0002148505530713109,
|
| 2245 |
+
"loss": 2.7901,
|
| 2246 |
+
"step": 299000
|
| 2247 |
+
},
|
| 2248 |
+
{
|
| 2249 |
+
"epoch": 16.132501613250163,
|
| 2250 |
+
"grad_norm": 0.27245578169822693,
|
| 2251 |
+
"learning_rate": 0.00021191162626500354,
|
| 2252 |
+
"loss": 2.791,
|
| 2253 |
+
"step": 300000
|
| 2254 |
+
},
|
| 2255 |
+
{
|
| 2256 |
+
"epoch": 16.186276618627662,
|
| 2257 |
+
"grad_norm": 0.27511173486709595,
|
| 2258 |
+
"learning_rate": 0.0002089697575900212,
|
| 2259 |
+
"loss": 2.7991,
|
| 2260 |
+
"step": 301000
|
| 2261 |
+
},
|
| 2262 |
+
{
|
| 2263 |
+
"epoch": 16.24005162400516,
|
| 2264 |
+
"grad_norm": 0.2818892300128937,
|
| 2265 |
+
"learning_rate": 0.00020602788891503884,
|
| 2266 |
+
"loss": 2.8017,
|
| 2267 |
+
"step": 302000
|
| 2268 |
+
},
|
| 2269 |
+
{
|
| 2270 |
+
"epoch": 16.293826629382664,
|
| 2271 |
+
"grad_norm": 0.29610157012939453,
|
| 2272 |
+
"learning_rate": 0.00020308896210873147,
|
| 2273 |
+
"loss": 2.7992,
|
| 2274 |
+
"step": 303000
|
| 2275 |
+
},
|
| 2276 |
+
{
|
| 2277 |
+
"epoch": 16.347601634760164,
|
| 2278 |
+
"grad_norm": 0.28958022594451904,
|
| 2279 |
+
"learning_rate": 0.00020014709343374912,
|
| 2280 |
+
"loss": 2.8032,
|
| 2281 |
+
"step": 304000
|
| 2282 |
+
},
|
| 2283 |
+
{
|
| 2284 |
+
"epoch": 16.401376640137663,
|
| 2285 |
+
"grad_norm": 0.32356253266334534,
|
| 2286 |
+
"learning_rate": 0.00019720522475876677,
|
| 2287 |
+
"loss": 2.8018,
|
| 2288 |
+
"step": 305000
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 16.455151645515166,
|
| 2292 |
+
"grad_norm": 0.2847062051296234,
|
| 2293 |
+
"learning_rate": 0.0001942633560837844,
|
| 2294 |
+
"loss": 2.8077,
|
| 2295 |
+
"step": 306000
|
| 2296 |
+
},
|
| 2297 |
+
{
|
| 2298 |
+
"epoch": 16.508926650892665,
|
| 2299 |
+
"grad_norm": 0.2739544212818146,
|
| 2300 |
+
"learning_rate": 0.00019132442927747705,
|
| 2301 |
+
"loss": 2.81,
|
| 2302 |
+
"step": 307000
|
| 2303 |
+
},
|
| 2304 |
+
{
|
| 2305 |
+
"epoch": 16.562701656270164,
|
| 2306 |
+
"grad_norm": 0.28697946667671204,
|
| 2307 |
+
"learning_rate": 0.0001883825606024947,
|
| 2308 |
+
"loss": 2.807,
|
| 2309 |
+
"step": 308000
|
| 2310 |
+
},
|
| 2311 |
+
{
|
| 2312 |
+
"epoch": 16.616476661647667,
|
| 2313 |
+
"grad_norm": 0.29771292209625244,
|
| 2314 |
+
"learning_rate": 0.00018544363379618733,
|
| 2315 |
+
"loss": 2.8096,
|
| 2316 |
+
"step": 309000
|
| 2317 |
+
},
|
| 2318 |
+
{
|
| 2319 |
+
"epoch": 16.670251667025166,
|
| 2320 |
+
"grad_norm": 0.29724204540252686,
|
| 2321 |
+
"learning_rate": 0.000182501765121205,
|
| 2322 |
+
"loss": 2.8081,
|
| 2323 |
+
"step": 310000
|
| 2324 |
+
},
|
| 2325 |
+
{
|
| 2326 |
+
"epoch": 16.72402667240267,
|
| 2327 |
+
"grad_norm": 0.2964895963668823,
|
| 2328 |
+
"learning_rate": 0.00017955989644622265,
|
| 2329 |
+
"loss": 2.8133,
|
| 2330 |
+
"step": 311000
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 16.777801677780168,
|
| 2334 |
+
"grad_norm": 0.28841668367385864,
|
| 2335 |
+
"learning_rate": 0.00017662391150859025,
|
| 2336 |
+
"loss": 2.8086,
|
| 2337 |
+
"step": 312000
|
| 2338 |
+
},
|
| 2339 |
+
{
|
| 2340 |
+
"epoch": 16.831576683157667,
|
| 2341 |
+
"grad_norm": 0.2766316831111908,
|
| 2342 |
+
"learning_rate": 0.00017368204283360793,
|
| 2343 |
+
"loss": 2.8152,
|
| 2344 |
+
"step": 313000
|
| 2345 |
+
},
|
| 2346 |
+
{
|
| 2347 |
+
"epoch": 16.88535168853517,
|
| 2348 |
+
"grad_norm": 0.32704171538352966,
|
| 2349 |
+
"learning_rate": 0.00017074017415862554,
|
| 2350 |
+
"loss": 2.8162,
|
| 2351 |
+
"step": 314000
|
| 2352 |
+
},
|
| 2353 |
+
{
|
| 2354 |
+
"epoch": 16.93912669391267,
|
| 2355 |
+
"grad_norm": 0.3360968828201294,
|
| 2356 |
+
"learning_rate": 0.00016779830548364322,
|
| 2357 |
+
"loss": 2.8133,
|
| 2358 |
+
"step": 315000
|
| 2359 |
+
},
|
| 2360 |
+
{
|
| 2361 |
+
"epoch": 16.99290169929017,
|
| 2362 |
+
"grad_norm": 0.31055524945259094,
|
| 2363 |
+
"learning_rate": 0.00016485643680866087,
|
| 2364 |
+
"loss": 2.8157,
|
| 2365 |
+
"step": 316000
|
| 2366 |
+
},
|
| 2367 |
+
{
|
| 2368 |
+
"epoch": 17.0,
|
| 2369 |
+
"eval_accuracy": 0.4103879636154489,
|
| 2370 |
+
"eval_loss": 3.3883779048919678,
|
| 2371 |
+
"eval_runtime": 154.6539,
|
| 2372 |
+
"eval_samples_per_second": 374.494,
|
| 2373 |
+
"eval_steps_per_second": 5.852,
|
| 2374 |
+
"step": 316132
|
| 2375 |
+
},
|
| 2376 |
+
{
|
| 2377 |
+
"epoch": 17.04667670466767,
|
| 2378 |
+
"grad_norm": 0.30388155579566956,
|
| 2379 |
+
"learning_rate": 0.0001619175100023535,
|
| 2380 |
+
"loss": 2.7807,
|
| 2381 |
+
"step": 317000
|
| 2382 |
+
},
|
| 2383 |
+
{
|
| 2384 |
+
"epoch": 17.10045171004517,
|
| 2385 |
+
"grad_norm": 0.2881651818752289,
|
| 2386 |
+
"learning_rate": 0.00015897564132737118,
|
| 2387 |
+
"loss": 2.7798,
|
| 2388 |
+
"step": 318000
|
| 2389 |
+
},
|
| 2390 |
+
{
|
| 2391 |
+
"epoch": 17.15422671542267,
|
| 2392 |
+
"grad_norm": 0.2695824205875397,
|
| 2393 |
+
"learning_rate": 0.0001560337726523888,
|
| 2394 |
+
"loss": 2.7817,
|
| 2395 |
+
"step": 319000
|
| 2396 |
+
},
|
| 2397 |
+
{
|
| 2398 |
+
"epoch": 17.208001720800173,
|
| 2399 |
+
"grad_norm": 0.30308765172958374,
|
| 2400 |
+
"learning_rate": 0.00015309484584608143,
|
| 2401 |
+
"loss": 2.7808,
|
| 2402 |
+
"step": 320000
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 17.261776726177672,
|
| 2406 |
+
"grad_norm": 0.29839852452278137,
|
| 2407 |
+
"learning_rate": 0.00015015297717109908,
|
| 2408 |
+
"loss": 2.7878,
|
| 2409 |
+
"step": 321000
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 17.315551731555175,
|
| 2413 |
+
"grad_norm": 0.2739504873752594,
|
| 2414 |
+
"learning_rate": 0.00014721110849611675,
|
| 2415 |
+
"loss": 2.7849,
|
| 2416 |
+
"step": 322000
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 17.369326736932674,
|
| 2420 |
+
"grad_norm": 0.33289453387260437,
|
| 2421 |
+
"learning_rate": 0.00014427512355848435,
|
| 2422 |
+
"loss": 2.7878,
|
| 2423 |
+
"step": 323000
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 17.423101742310173,
|
| 2427 |
+
"grad_norm": 0.30872535705566406,
|
| 2428 |
+
"learning_rate": 0.00014133325488350202,
|
| 2429 |
+
"loss": 2.7862,
|
| 2430 |
+
"step": 324000
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 17.476876747687676,
|
| 2434 |
+
"grad_norm": 0.30065229535102844,
|
| 2435 |
+
"learning_rate": 0.00013839138620851964,
|
| 2436 |
+
"loss": 2.7903,
|
| 2437 |
+
"step": 325000
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 17.530651753065175,
|
| 2441 |
+
"grad_norm": 0.294783353805542,
|
| 2442 |
+
"learning_rate": 0.0001354495175335373,
|
| 2443 |
+
"loss": 2.7901,
|
| 2444 |
+
"step": 326000
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 17.584426758442675,
|
| 2448 |
+
"grad_norm": 0.3017074167728424,
|
| 2449 |
+
"learning_rate": 0.00013251059072722993,
|
| 2450 |
+
"loss": 2.789,
|
| 2451 |
+
"step": 327000
|
| 2452 |
+
},
|
| 2453 |
+
{
|
| 2454 |
+
"epoch": 17.638201763820177,
|
| 2455 |
+
"grad_norm": 0.28931453824043274,
|
| 2456 |
+
"learning_rate": 0.0001295687220522476,
|
| 2457 |
+
"loss": 2.7962,
|
| 2458 |
+
"step": 328000
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 17.691976769197677,
|
| 2462 |
+
"grad_norm": 0.30777105689048767,
|
| 2463 |
+
"learning_rate": 0.00012662685337726525,
|
| 2464 |
+
"loss": 2.7905,
|
| 2465 |
+
"step": 329000
|
| 2466 |
+
},
|
| 2467 |
+
{
|
| 2468 |
+
"epoch": 17.745751774575176,
|
| 2469 |
+
"grad_norm": 0.30437570810317993,
|
| 2470 |
+
"learning_rate": 0.00012368792657095788,
|
| 2471 |
+
"loss": 2.7922,
|
| 2472 |
+
"step": 330000
|
| 2473 |
+
},
|
| 2474 |
+
{
|
| 2475 |
+
"epoch": 17.79952677995268,
|
| 2476 |
+
"grad_norm": 0.279985249042511,
|
| 2477 |
+
"learning_rate": 0.00012074605789597553,
|
| 2478 |
+
"loss": 2.7958,
|
| 2479 |
+
"step": 331000
|
| 2480 |
+
},
|
| 2481 |
+
{
|
| 2482 |
+
"epoch": 17.853301785330178,
|
| 2483 |
+
"grad_norm": 0.2660251557826996,
|
| 2484 |
+
"learning_rate": 0.00011780418922099318,
|
| 2485 |
+
"loss": 2.7954,
|
| 2486 |
+
"step": 332000
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 17.90707679070768,
|
| 2490 |
+
"grad_norm": 0.2697618901729584,
|
| 2491 |
+
"learning_rate": 0.00011486526241468581,
|
| 2492 |
+
"loss": 2.7969,
|
| 2493 |
+
"step": 333000
|
| 2494 |
+
},
|
| 2495 |
+
{
|
| 2496 |
+
"epoch": 17.96085179608518,
|
| 2497 |
+
"grad_norm": 0.2890709340572357,
|
| 2498 |
+
"learning_rate": 0.00011192633560837845,
|
| 2499 |
+
"loss": 2.7966,
|
| 2500 |
+
"step": 334000
|
| 2501 |
+
},
|
| 2502 |
+
{
|
| 2503 |
+
"epoch": 18.0,
|
| 2504 |
+
"eval_accuracy": 0.4105063404500295,
|
| 2505 |
+
"eval_loss": 3.4080612659454346,
|
| 2506 |
+
"eval_runtime": 154.9923,
|
| 2507 |
+
"eval_samples_per_second": 373.677,
|
| 2508 |
+
"eval_steps_per_second": 5.839,
|
| 2509 |
+
"step": 334728
|
| 2510 |
+
},
|
| 2511 |
+
{
|
| 2512 |
+
"epoch": 18.01462680146268,
|
| 2513 |
+
"grad_norm": 0.2767917215824127,
|
| 2514 |
+
"learning_rate": 0.0001089844669333961,
|
| 2515 |
+
"loss": 2.7867,
|
| 2516 |
+
"step": 335000
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 18.068401806840182,
|
| 2520 |
+
"grad_norm": 0.29336050152778625,
|
| 2521 |
+
"learning_rate": 0.00010604259825841375,
|
| 2522 |
+
"loss": 2.7639,
|
| 2523 |
+
"step": 336000
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 18.12217681221768,
|
| 2527 |
+
"grad_norm": 0.3294292688369751,
|
| 2528 |
+
"learning_rate": 0.0001031007295834314,
|
| 2529 |
+
"loss": 2.7658,
|
| 2530 |
+
"step": 337000
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"epoch": 18.17595181759518,
|
| 2534 |
+
"grad_norm": 0.30230215191841125,
|
| 2535 |
+
"learning_rate": 0.00010015886090844905,
|
| 2536 |
+
"loss": 2.766,
|
| 2537 |
+
"step": 338000
|
| 2538 |
+
},
|
| 2539 |
+
{
|
| 2540 |
+
"epoch": 18.229726822972683,
|
| 2541 |
+
"grad_norm": 0.28469741344451904,
|
| 2542 |
+
"learning_rate": 9.721993410214168e-05,
|
| 2543 |
+
"loss": 2.7682,
|
| 2544 |
+
"step": 339000
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"epoch": 18.283501828350182,
|
| 2548 |
+
"grad_norm": 0.2829948365688324,
|
| 2549 |
+
"learning_rate": 9.427806542715933e-05,
|
| 2550 |
+
"loss": 2.7747,
|
| 2551 |
+
"step": 340000
|
| 2552 |
+
},
|
| 2553 |
+
{
|
| 2554 |
+
"epoch": 18.33727683372768,
|
| 2555 |
+
"grad_norm": 0.27874138951301575,
|
| 2556 |
+
"learning_rate": 9.133913862085197e-05,
|
| 2557 |
+
"loss": 2.7719,
|
| 2558 |
+
"step": 341000
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 18.391051839105184,
|
| 2562 |
+
"grad_norm": 0.3442845642566681,
|
| 2563 |
+
"learning_rate": 8.839726994586963e-05,
|
| 2564 |
+
"loss": 2.7753,
|
| 2565 |
+
"step": 342000
|
| 2566 |
+
},
|
| 2567 |
+
{
|
| 2568 |
+
"epoch": 18.444826844482684,
|
| 2569 |
+
"grad_norm": 0.36807698011398315,
|
| 2570 |
+
"learning_rate": 8.545540127088727e-05,
|
| 2571 |
+
"loss": 2.7734,
|
| 2572 |
+
"step": 343000
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"epoch": 18.498601849860187,
|
| 2576 |
+
"grad_norm": 0.3008968234062195,
|
| 2577 |
+
"learning_rate": 8.25164744645799e-05,
|
| 2578 |
+
"loss": 2.7742,
|
| 2579 |
+
"step": 344000
|
| 2580 |
+
},
|
| 2581 |
+
{
|
| 2582 |
+
"epoch": 18.552376855237686,
|
| 2583 |
+
"grad_norm": 0.3300212621688843,
|
| 2584 |
+
"learning_rate": 7.957460578959754e-05,
|
| 2585 |
+
"loss": 2.7729,
|
| 2586 |
+
"step": 345000
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"epoch": 18.606151860615185,
|
| 2590 |
+
"grad_norm": 0.2936519682407379,
|
| 2591 |
+
"learning_rate": 7.66327371146152e-05,
|
| 2592 |
+
"loss": 2.7756,
|
| 2593 |
+
"step": 346000
|
| 2594 |
+
},
|
| 2595 |
+
{
|
| 2596 |
+
"epoch": 18.659926865992688,
|
| 2597 |
+
"grad_norm": 0.2997318208217621,
|
| 2598 |
+
"learning_rate": 7.369381030830784e-05,
|
| 2599 |
+
"loss": 2.7787,
|
| 2600 |
+
"step": 347000
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 18.713701871370187,
|
| 2604 |
+
"grad_norm": 0.27832144498825073,
|
| 2605 |
+
"learning_rate": 7.075194163332549e-05,
|
| 2606 |
+
"loss": 2.7736,
|
| 2607 |
+
"step": 348000
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 18.767476876747686,
|
| 2611 |
+
"grad_norm": 0.28568559885025024,
|
| 2612 |
+
"learning_rate": 6.781301482701812e-05,
|
| 2613 |
+
"loss": 2.7752,
|
| 2614 |
+
"step": 349000
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"epoch": 18.82125188212519,
|
| 2618 |
+
"grad_norm": 0.2887098789215088,
|
| 2619 |
+
"learning_rate": 6.487408802071074e-05,
|
| 2620 |
+
"loss": 2.7764,
|
| 2621 |
+
"step": 350000
|
| 2622 |
+
},
|
| 2623 |
+
{
|
| 2624 |
+
"epoch": 18.87502688750269,
|
| 2625 |
+
"grad_norm": 0.31327852606773376,
|
| 2626 |
+
"learning_rate": 6.19322193457284e-05,
|
| 2627 |
+
"loss": 2.7776,
|
| 2628 |
+
"step": 351000
|
| 2629 |
+
},
|
| 2630 |
+
{
|
| 2631 |
+
"epoch": 18.928801892880188,
|
| 2632 |
+
"grad_norm": 0.2947293519973755,
|
| 2633 |
+
"learning_rate": 5.899035067074606e-05,
|
| 2634 |
+
"loss": 2.7778,
|
| 2635 |
+
"step": 352000
|
| 2636 |
+
},
|
| 2637 |
+
{
|
| 2638 |
+
"epoch": 18.98257689825769,
|
| 2639 |
+
"grad_norm": 0.27630892395973206,
|
| 2640 |
+
"learning_rate": 5.604848199576371e-05,
|
| 2641 |
+
"loss": 2.7807,
|
| 2642 |
+
"step": 353000
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 19.0,
|
| 2646 |
+
"eval_accuracy": 0.41043096087659053,
|
| 2647 |
+
"eval_loss": 3.4180543422698975,
|
| 2648 |
+
"eval_runtime": 155.0182,
|
| 2649 |
+
"eval_samples_per_second": 373.614,
|
| 2650 |
+
"eval_steps_per_second": 5.838,
|
| 2651 |
+
"step": 353324
|
| 2652 |
+
},
|
| 2653 |
+
{
|
| 2654 |
+
"epoch": 19.03635190363519,
|
| 2655 |
+
"grad_norm": 0.29157590866088867,
|
| 2656 |
+
"learning_rate": 5.310955518945634e-05,
|
| 2657 |
+
"loss": 2.7661,
|
| 2658 |
+
"step": 354000
|
| 2659 |
+
},
|
| 2660 |
+
{
|
| 2661 |
+
"epoch": 19.090126909012692,
|
| 2662 |
+
"grad_norm": 0.26800793409347534,
|
| 2663 |
+
"learning_rate": 5.0167686514473995e-05,
|
| 2664 |
+
"loss": 2.7558,
|
| 2665 |
+
"step": 355000
|
| 2666 |
+
},
|
| 2667 |
+
{
|
| 2668 |
+
"epoch": 19.14390191439019,
|
| 2669 |
+
"grad_norm": 0.31518709659576416,
|
| 2670 |
+
"learning_rate": 4.722581783949165e-05,
|
| 2671 |
+
"loss": 2.7579,
|
| 2672 |
+
"step": 356000
|
| 2673 |
+
},
|
| 2674 |
+
{
|
| 2675 |
+
"epoch": 19.19767691976769,
|
| 2676 |
+
"grad_norm": 0.2998282313346863,
|
| 2677 |
+
"learning_rate": 4.428689103318428e-05,
|
| 2678 |
+
"loss": 2.7574,
|
| 2679 |
+
"step": 357000
|
| 2680 |
+
},
|
| 2681 |
+
{
|
| 2682 |
+
"epoch": 19.251451925145194,
|
| 2683 |
+
"grad_norm": 0.3034313917160034,
|
| 2684 |
+
"learning_rate": 4.134502235820193e-05,
|
| 2685 |
+
"loss": 2.7606,
|
| 2686 |
+
"step": 358000
|
| 2687 |
+
},
|
| 2688 |
+
{
|
| 2689 |
+
"epoch": 19.305226930522693,
|
| 2690 |
+
"grad_norm": 0.2875937819480896,
|
| 2691 |
+
"learning_rate": 3.8403153683219584e-05,
|
| 2692 |
+
"loss": 2.7586,
|
| 2693 |
+
"step": 359000
|
| 2694 |
+
},
|
| 2695 |
+
{
|
| 2696 |
+
"epoch": 19.359001935900192,
|
| 2697 |
+
"grad_norm": 0.30761197209358215,
|
| 2698 |
+
"learning_rate": 3.546422687691222e-05,
|
| 2699 |
+
"loss": 2.7577,
|
| 2700 |
+
"step": 360000
|
| 2701 |
+
},
|
| 2702 |
+
{
|
| 2703 |
+
"epoch": 19.412776941277695,
|
| 2704 |
+
"grad_norm": 0.29801440238952637,
|
| 2705 |
+
"learning_rate": 3.252530007060485e-05,
|
| 2706 |
+
"loss": 2.7557,
|
| 2707 |
+
"step": 361000
|
| 2708 |
+
},
|
| 2709 |
+
{
|
| 2710 |
+
"epoch": 19.466551946655194,
|
| 2711 |
+
"grad_norm": 0.2939643859863281,
|
| 2712 |
+
"learning_rate": 2.95834313956225e-05,
|
| 2713 |
+
"loss": 2.7571,
|
| 2714 |
+
"step": 362000
|
| 2715 |
+
},
|
| 2716 |
+
{
|
| 2717 |
+
"epoch": 19.520326952032697,
|
| 2718 |
+
"grad_norm": 0.29968592524528503,
|
| 2719 |
+
"learning_rate": 2.6641562720640153e-05,
|
| 2720 |
+
"loss": 2.761,
|
| 2721 |
+
"step": 363000
|
| 2722 |
+
},
|
| 2723 |
+
{
|
| 2724 |
+
"epoch": 19.574101957410196,
|
| 2725 |
+
"grad_norm": 0.28651562333106995,
|
| 2726 |
+
"learning_rate": 2.36996940456578e-05,
|
| 2727 |
+
"loss": 2.7579,
|
| 2728 |
+
"step": 364000
|
| 2729 |
+
},
|
| 2730 |
+
{
|
| 2731 |
+
"epoch": 19.627876962787695,
|
| 2732 |
+
"grad_norm": 0.30474939942359924,
|
| 2733 |
+
"learning_rate": 2.0760767239350436e-05,
|
| 2734 |
+
"loss": 2.7598,
|
| 2735 |
+
"step": 365000
|
| 2736 |
+
},
|
| 2737 |
+
{
|
| 2738 |
+
"epoch": 19.6816519681652,
|
| 2739 |
+
"grad_norm": 0.2941524088382721,
|
| 2740 |
+
"learning_rate": 1.7818898564368086e-05,
|
| 2741 |
+
"loss": 2.7644,
|
| 2742 |
+
"step": 366000
|
| 2743 |
+
},
|
| 2744 |
+
{
|
| 2745 |
+
"epoch": 19.735426973542697,
|
| 2746 |
+
"grad_norm": 0.2570919990539551,
|
| 2747 |
+
"learning_rate": 1.487997175806072e-05,
|
| 2748 |
+
"loss": 2.7549,
|
| 2749 |
+
"step": 367000
|
| 2750 |
+
},
|
| 2751 |
+
{
|
| 2752 |
+
"epoch": 19.789201978920197,
|
| 2753 |
+
"grad_norm": 0.31410789489746094,
|
| 2754 |
+
"learning_rate": 1.193810308307837e-05,
|
| 2755 |
+
"loss": 2.7624,
|
| 2756 |
+
"step": 368000
|
| 2757 |
+
},
|
| 2758 |
+
{
|
| 2759 |
+
"epoch": 19.8429769842977,
|
| 2760 |
+
"grad_norm": 0.29923874139785767,
|
| 2761 |
+
"learning_rate": 8.999176276771005e-06,
|
| 2762 |
+
"loss": 2.7621,
|
| 2763 |
+
"step": 369000
|
| 2764 |
+
},
|
| 2765 |
+
{
|
| 2766 |
+
"epoch": 19.8967519896752,
|
| 2767 |
+
"grad_norm": 0.29635271430015564,
|
| 2768 |
+
"learning_rate": 6.057307601788656e-06,
|
| 2769 |
+
"loss": 2.7601,
|
| 2770 |
+
"step": 370000
|
| 2771 |
+
},
|
| 2772 |
+
{
|
| 2773 |
+
"epoch": 19.950526995052698,
|
| 2774 |
+
"grad_norm": 0.29568880796432495,
|
| 2775 |
+
"learning_rate": 3.1154389268063074e-06,
|
| 2776 |
+
"loss": 2.7595,
|
| 2777 |
+
"step": 371000
|
| 2778 |
+
},
|
| 2779 |
+
{
|
| 2780 |
+
"epoch": 20.0,
|
| 2781 |
+
"eval_accuracy": 0.41021489963935376,
|
| 2782 |
+
"eval_loss": 3.423628807067871,
|
| 2783 |
+
"eval_runtime": 154.8423,
|
| 2784 |
+
"eval_samples_per_second": 374.039,
|
| 2785 |
+
"eval_steps_per_second": 5.845,
|
| 2786 |
+
"step": 371920
|
| 2787 |
+
},
|
| 2788 |
+
{
|
| 2789 |
+
"epoch": 20.0,
|
| 2790 |
+
"step": 371920,
|
| 2791 |
+
"total_flos": 1.5670047538944e+18,
|
| 2792 |
+
"train_loss": 3.0279207580395733,
|
| 2793 |
+
"train_runtime": 82395.3872,
|
| 2794 |
+
"train_samples_per_second": 144.441,
|
| 2795 |
+
"train_steps_per_second": 4.514
|
| 2796 |
+
}
|
| 2797 |
+
],
|
| 2798 |
+
"logging_steps": 1000,
|
| 2799 |
+
"max_steps": 371920,
|
| 2800 |
+
"num_input_tokens_seen": 0,
|
| 2801 |
+
"num_train_epochs": 20,
|
| 2802 |
+
"save_steps": 5000,
|
| 2803 |
+
"stateful_callbacks": {
|
| 2804 |
+
"TrainerControl": {
|
| 2805 |
+
"args": {
|
| 2806 |
+
"should_epoch_stop": false,
|
| 2807 |
+
"should_evaluate": false,
|
| 2808 |
+
"should_log": false,
|
| 2809 |
+
"should_save": true,
|
| 2810 |
+
"should_training_stop": false
|
| 2811 |
+
},
|
| 2812 |
+
"attributes": {}
|
| 2813 |
+
}
|
| 2814 |
+
},
|
| 2815 |
+
"total_flos": 1.5670047538944e+18,
|
| 2816 |
+
"train_batch_size": 32,
|
| 2817 |
+
"trial_name": null,
|
| 2818 |
+
"trial_params": null
|
| 2819 |
+
}
|