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_naans_new-seed_1024-1e-4
|
| 8 |
-
results:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
---
|
| 10 |
|
| 11 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
@@ -13,7 +25,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 13 |
|
| 14 |
# smolm-autoreg-bpe-counterfactual_babylm_naans_new-seed_1024-1e-4
|
| 15 |
|
| 16 |
-
This model was trained from scratch on
|
| 17 |
It achieves the following results on the evaluation set:
|
| 18 |
- Loss: 3.4373
|
| 19 |
- Accuracy: 0.4072
|
|
|
|
| 1 |
---
|
| 2 |
tags:
|
| 3 |
- generated_from_trainer
|
| 4 |
+
datasets:
|
| 5 |
+
- kanishka/counterfactual_babylm_naans_new
|
| 6 |
metrics:
|
| 7 |
- accuracy
|
| 8 |
model-index:
|
| 9 |
- name: smolm-autoreg-bpe-counterfactual_babylm_naans_new-seed_1024-1e-4
|
| 10 |
+
results:
|
| 11 |
+
- task:
|
| 12 |
+
name: Causal Language Modeling
|
| 13 |
+
type: text-generation
|
| 14 |
+
dataset:
|
| 15 |
+
name: kanishka/counterfactual_babylm_naans_new
|
| 16 |
+
type: kanishka/counterfactual_babylm_naans_new
|
| 17 |
+
metrics:
|
| 18 |
+
- name: Accuracy
|
| 19 |
+
type: accuracy
|
| 20 |
+
value: 0.40723568402608223
|
| 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_naans_new-seed_1024-1e-4
|
| 27 |
|
| 28 |
+
This model was trained from scratch on the kanishka/counterfactual_babylm_naans_new dataset.
|
| 29 |
It achieves the following results on the evaluation set:
|
| 30 |
- Loss: 3.4373
|
| 31 |
- Accuracy: 0.4072
|
all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"eval_accuracy": 0.40723568402608223,
|
| 4 |
+
"eval_loss": 3.4373319149017334,
|
| 5 |
+
"eval_runtime": 154.3696,
|
| 6 |
+
"eval_samples": 57921,
|
| 7 |
+
"eval_samples_per_second": 375.21,
|
| 8 |
+
"eval_steps_per_second": 5.869,
|
| 9 |
+
"perplexity": 31.103859616260518,
|
| 10 |
+
"train_loss": 3.1537178330448374,
|
| 11 |
+
"train_runtime": 81513.9803,
|
| 12 |
+
"train_samples": 595035,
|
| 13 |
+
"train_samples_per_second": 145.996,
|
| 14 |
+
"train_steps_per_second": 4.562
|
| 15 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"eval_accuracy": 0.40723568402608223,
|
| 4 |
+
"eval_loss": 3.4373319149017334,
|
| 5 |
+
"eval_runtime": 154.3696,
|
| 6 |
+
"eval_samples": 57921,
|
| 7 |
+
"eval_samples_per_second": 375.21,
|
| 8 |
+
"eval_steps_per_second": 5.869,
|
| 9 |
+
"perplexity": 31.103859616260518
|
| 10 |
+
}
|
runs/Jun18_21-19-07_phyl-ling-p01.la.utexas.edu/events.out.tfevents.1718845604.phyl-ling-p01.la.utexas.edu.761819.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4c92680540f911265c66187aaf05702f6ef01871ff00b5bd40bf9054fd1ba99d
|
| 3 |
+
size 417
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"train_loss": 3.1537178330448374,
|
| 4 |
+
"train_runtime": 81513.9803,
|
| 5 |
+
"train_samples": 595035,
|
| 6 |
+
"train_samples_per_second": 145.996,
|
| 7 |
+
"train_steps_per_second": 4.562
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2807 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 20.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 371900,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.05,
|
| 13 |
+
"grad_norm": 1.0875967741012573,
|
| 14 |
+
"learning_rate": 3.125e-06,
|
| 15 |
+
"loss": 7.5939,
|
| 16 |
+
"step": 1000
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.11,
|
| 20 |
+
"grad_norm": 0.8077093362808228,
|
| 21 |
+
"learning_rate": 6.25e-06,
|
| 22 |
+
"loss": 5.8309,
|
| 23 |
+
"step": 2000
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.16,
|
| 27 |
+
"grad_norm": 0.8327828049659729,
|
| 28 |
+
"learning_rate": 9.375000000000001e-06,
|
| 29 |
+
"loss": 5.3685,
|
| 30 |
+
"step": 3000
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.22,
|
| 34 |
+
"grad_norm": 0.9203261137008667,
|
| 35 |
+
"learning_rate": 1.25e-05,
|
| 36 |
+
"loss": 5.1581,
|
| 37 |
+
"step": 4000
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.27,
|
| 41 |
+
"grad_norm": 0.9855815768241882,
|
| 42 |
+
"learning_rate": 1.5625e-05,
|
| 43 |
+
"loss": 5.0109,
|
| 44 |
+
"step": 5000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.32,
|
| 48 |
+
"grad_norm": 1.1521941423416138,
|
| 49 |
+
"learning_rate": 1.8750000000000002e-05,
|
| 50 |
+
"loss": 4.8668,
|
| 51 |
+
"step": 6000
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.38,
|
| 55 |
+
"grad_norm": 1.1727802753448486,
|
| 56 |
+
"learning_rate": 2.1875e-05,
|
| 57 |
+
"loss": 4.759,
|
| 58 |
+
"step": 7000
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.43,
|
| 62 |
+
"grad_norm": 1.1063674688339233,
|
| 63 |
+
"learning_rate": 2.5e-05,
|
| 64 |
+
"loss": 4.6561,
|
| 65 |
+
"step": 8000
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.48,
|
| 69 |
+
"grad_norm": 1.178877592086792,
|
| 70 |
+
"learning_rate": 2.8125000000000003e-05,
|
| 71 |
+
"loss": 4.5671,
|
| 72 |
+
"step": 9000
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.54,
|
| 76 |
+
"grad_norm": 1.0879851579666138,
|
| 77 |
+
"learning_rate": 3.125e-05,
|
| 78 |
+
"loss": 4.4872,
|
| 79 |
+
"step": 10000
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.59,
|
| 83 |
+
"grad_norm": 1.1763601303100586,
|
| 84 |
+
"learning_rate": 3.4371875e-05,
|
| 85 |
+
"loss": 4.4182,
|
| 86 |
+
"step": 11000
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.65,
|
| 90 |
+
"grad_norm": 1.0679717063903809,
|
| 91 |
+
"learning_rate": 3.7496875e-05,
|
| 92 |
+
"loss": 4.3542,
|
| 93 |
+
"step": 12000
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.7,
|
| 97 |
+
"grad_norm": 1.051241159439087,
|
| 98 |
+
"learning_rate": 4.061875e-05,
|
| 99 |
+
"loss": 4.2985,
|
| 100 |
+
"step": 13000
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.75,
|
| 104 |
+
"grad_norm": 1.0657079219818115,
|
| 105 |
+
"learning_rate": 4.374375e-05,
|
| 106 |
+
"loss": 4.2397,
|
| 107 |
+
"step": 14000
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.81,
|
| 111 |
+
"grad_norm": 1.0396409034729004,
|
| 112 |
+
"learning_rate": 4.686875e-05,
|
| 113 |
+
"loss": 4.1923,
|
| 114 |
+
"step": 15000
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.86,
|
| 118 |
+
"grad_norm": 0.9875408411026001,
|
| 119 |
+
"learning_rate": 4.999375e-05,
|
| 120 |
+
"loss": 4.1418,
|
| 121 |
+
"step": 16000
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.91,
|
| 125 |
+
"grad_norm": 1.003056526184082,
|
| 126 |
+
"learning_rate": 5.3115625000000005e-05,
|
| 127 |
+
"loss": 4.091,
|
| 128 |
+
"step": 17000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.97,
|
| 132 |
+
"grad_norm": 0.9668901562690735,
|
| 133 |
+
"learning_rate": 5.62375e-05,
|
| 134 |
+
"loss": 4.0553,
|
| 135 |
+
"step": 18000
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 1.0,
|
| 139 |
+
"eval_accuracy": 0.3076893208292497,
|
| 140 |
+
"eval_loss": 4.2751569747924805,
|
| 141 |
+
"eval_runtime": 153.8934,
|
| 142 |
+
"eval_samples_per_second": 376.371,
|
| 143 |
+
"eval_steps_per_second": 5.887,
|
| 144 |
+
"step": 18595
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 1.02,
|
| 148 |
+
"grad_norm": 1.0316466093063354,
|
| 149 |
+
"learning_rate": 5.93625e-05,
|
| 150 |
+
"loss": 4.0151,
|
| 151 |
+
"step": 19000
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 1.08,
|
| 155 |
+
"grad_norm": 1.0109944343566895,
|
| 156 |
+
"learning_rate": 6.24875e-05,
|
| 157 |
+
"loss": 3.9733,
|
| 158 |
+
"step": 20000
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.13,
|
| 162 |
+
"grad_norm": 1.0080300569534302,
|
| 163 |
+
"learning_rate": 6.56125e-05,
|
| 164 |
+
"loss": 3.9328,
|
| 165 |
+
"step": 21000
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 1.18,
|
| 169 |
+
"grad_norm": 1.0754659175872803,
|
| 170 |
+
"learning_rate": 6.873125000000001e-05,
|
| 171 |
+
"loss": 3.8969,
|
| 172 |
+
"step": 22000
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 1.24,
|
| 176 |
+
"grad_norm": 1.018548607826233,
|
| 177 |
+
"learning_rate": 7.185625e-05,
|
| 178 |
+
"loss": 3.8657,
|
| 179 |
+
"step": 23000
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.29,
|
| 183 |
+
"grad_norm": 0.9804821014404297,
|
| 184 |
+
"learning_rate": 7.4975e-05,
|
| 185 |
+
"loss": 3.8338,
|
| 186 |
+
"step": 24000
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 1.34,
|
| 190 |
+
"grad_norm": 1.0100795030593872,
|
| 191 |
+
"learning_rate": 7.81e-05,
|
| 192 |
+
"loss": 3.7995,
|
| 193 |
+
"step": 25000
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 1.4,
|
| 197 |
+
"grad_norm": 0.9405222535133362,
|
| 198 |
+
"learning_rate": 8.1221875e-05,
|
| 199 |
+
"loss": 3.773,
|
| 200 |
+
"step": 26000
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 1.45,
|
| 204 |
+
"grad_norm": 0.9475222826004028,
|
| 205 |
+
"learning_rate": 8.434687500000001e-05,
|
| 206 |
+
"loss": 3.751,
|
| 207 |
+
"step": 27000
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 1.51,
|
| 211 |
+
"grad_norm": 0.890032172203064,
|
| 212 |
+
"learning_rate": 8.746875e-05,
|
| 213 |
+
"loss": 3.7279,
|
| 214 |
+
"step": 28000
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 1.56,
|
| 218 |
+
"grad_norm": 0.9058054685592651,
|
| 219 |
+
"learning_rate": 9.059375e-05,
|
| 220 |
+
"loss": 3.7031,
|
| 221 |
+
"step": 29000
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.61,
|
| 225 |
+
"grad_norm": 0.8877508044242859,
|
| 226 |
+
"learning_rate": 9.3715625e-05,
|
| 227 |
+
"loss": 3.6828,
|
| 228 |
+
"step": 30000
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 1.67,
|
| 232 |
+
"grad_norm": 0.8418898582458496,
|
| 233 |
+
"learning_rate": 9.684062500000001e-05,
|
| 234 |
+
"loss": 3.6672,
|
| 235 |
+
"step": 31000
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 1.72,
|
| 239 |
+
"grad_norm": 0.8882458806037903,
|
| 240 |
+
"learning_rate": 9.9965625e-05,
|
| 241 |
+
"loss": 3.6463,
|
| 242 |
+
"step": 32000
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.77,
|
| 246 |
+
"grad_norm": 0.8538274168968201,
|
| 247 |
+
"learning_rate": 9.970903206825538e-05,
|
| 248 |
+
"loss": 3.6331,
|
| 249 |
+
"step": 33000
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 1.83,
|
| 253 |
+
"grad_norm": 0.8449269533157349,
|
| 254 |
+
"learning_rate": 9.941512209473375e-05,
|
| 255 |
+
"loss": 3.6114,
|
| 256 |
+
"step": 34000
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 1.88,
|
| 260 |
+
"grad_norm": 0.8731977343559265,
|
| 261 |
+
"learning_rate": 9.912150632538982e-05,
|
| 262 |
+
"loss": 3.5938,
|
| 263 |
+
"step": 35000
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 1.94,
|
| 267 |
+
"grad_norm": 0.8085331916809082,
|
| 268 |
+
"learning_rate": 9.88273021476905e-05,
|
| 269 |
+
"loss": 3.5792,
|
| 270 |
+
"step": 36000
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 1.99,
|
| 274 |
+
"grad_norm": 0.8539865612983704,
|
| 275 |
+
"learning_rate": 9.853309796999118e-05,
|
| 276 |
+
"loss": 3.5587,
|
| 277 |
+
"step": 37000
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 2.0,
|
| 281 |
+
"eval_accuracy": 0.3633328691293394,
|
| 282 |
+
"eval_loss": 3.7501370906829834,
|
| 283 |
+
"eval_runtime": 155.089,
|
| 284 |
+
"eval_samples_per_second": 373.47,
|
| 285 |
+
"eval_steps_per_second": 5.842,
|
| 286 |
+
"step": 37190
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 2.04,
|
| 290 |
+
"grad_norm": 0.8294937610626221,
|
| 291 |
+
"learning_rate": 9.823889379229185e-05,
|
| 292 |
+
"loss": 3.5228,
|
| 293 |
+
"step": 38000
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 2.1,
|
| 297 |
+
"grad_norm": 0.7889467477798462,
|
| 298 |
+
"learning_rate": 9.794468961459253e-05,
|
| 299 |
+
"loss": 3.5132,
|
| 300 |
+
"step": 39000
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 2.15,
|
| 304 |
+
"grad_norm": 0.7929906249046326,
|
| 305 |
+
"learning_rate": 9.765048543689321e-05,
|
| 306 |
+
"loss": 3.495,
|
| 307 |
+
"step": 40000
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 2.2,
|
| 311 |
+
"grad_norm": 0.8031061887741089,
|
| 312 |
+
"learning_rate": 9.735657546337158e-05,
|
| 313 |
+
"loss": 3.4887,
|
| 314 |
+
"step": 41000
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 2.26,
|
| 318 |
+
"grad_norm": 0.8079201579093933,
|
| 319 |
+
"learning_rate": 9.706266548984995e-05,
|
| 320 |
+
"loss": 3.4837,
|
| 321 |
+
"step": 42000
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 2.31,
|
| 325 |
+
"grad_norm": 0.8008145689964294,
|
| 326 |
+
"learning_rate": 9.676846131215064e-05,
|
| 327 |
+
"loss": 3.471,
|
| 328 |
+
"step": 43000
|
| 329 |
+
},
|
| 330 |
+
{
|
| 331 |
+
"epoch": 2.37,
|
| 332 |
+
"grad_norm": 0.7869082689285278,
|
| 333 |
+
"learning_rate": 9.647425713445132e-05,
|
| 334 |
+
"loss": 3.4643,
|
| 335 |
+
"step": 44000
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 2.42,
|
| 339 |
+
"grad_norm": 0.8000088930130005,
|
| 340 |
+
"learning_rate": 9.618005295675198e-05,
|
| 341 |
+
"loss": 3.4542,
|
| 342 |
+
"step": 45000
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 2.47,
|
| 346 |
+
"grad_norm": 0.7834064960479736,
|
| 347 |
+
"learning_rate": 9.588614298323037e-05,
|
| 348 |
+
"loss": 3.4491,
|
| 349 |
+
"step": 46000
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 2.53,
|
| 353 |
+
"grad_norm": 0.7652507424354553,
|
| 354 |
+
"learning_rate": 9.559193880553105e-05,
|
| 355 |
+
"loss": 3.4387,
|
| 356 |
+
"step": 47000
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.58,
|
| 360 |
+
"grad_norm": 0.8015010356903076,
|
| 361 |
+
"learning_rate": 9.529802883200942e-05,
|
| 362 |
+
"loss": 3.4331,
|
| 363 |
+
"step": 48000
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 2.64,
|
| 367 |
+
"grad_norm": 0.8380176424980164,
|
| 368 |
+
"learning_rate": 9.500382465431009e-05,
|
| 369 |
+
"loss": 3.4261,
|
| 370 |
+
"step": 49000
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 2.69,
|
| 374 |
+
"grad_norm": 0.7744593024253845,
|
| 375 |
+
"learning_rate": 9.470962047661077e-05,
|
| 376 |
+
"loss": 3.4181,
|
| 377 |
+
"step": 50000
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 2.74,
|
| 381 |
+
"grad_norm": 0.7821061015129089,
|
| 382 |
+
"learning_rate": 9.441541629891145e-05,
|
| 383 |
+
"loss": 3.412,
|
| 384 |
+
"step": 51000
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 2.8,
|
| 388 |
+
"grad_norm": 0.7764970660209656,
|
| 389 |
+
"learning_rate": 9.412150632538983e-05,
|
| 390 |
+
"loss": 3.4048,
|
| 391 |
+
"step": 52000
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 2.85,
|
| 395 |
+
"grad_norm": 0.7456904649734497,
|
| 396 |
+
"learning_rate": 9.38273021476905e-05,
|
| 397 |
+
"loss": 3.3998,
|
| 398 |
+
"step": 53000
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 2.9,
|
| 402 |
+
"grad_norm": 0.7432152032852173,
|
| 403 |
+
"learning_rate": 9.353339217416887e-05,
|
| 404 |
+
"loss": 3.394,
|
| 405 |
+
"step": 54000
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 2.96,
|
| 409 |
+
"grad_norm": 0.7415820956230164,
|
| 410 |
+
"learning_rate": 9.323948220064726e-05,
|
| 411 |
+
"loss": 3.3865,
|
| 412 |
+
"step": 55000
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 3.0,
|
| 416 |
+
"eval_accuracy": 0.3815021224736739,
|
| 417 |
+
"eval_loss": 3.5883822441101074,
|
| 418 |
+
"eval_runtime": 154.6982,
|
| 419 |
+
"eval_samples_per_second": 374.413,
|
| 420 |
+
"eval_steps_per_second": 5.857,
|
| 421 |
+
"step": 55785
|
| 422 |
+
},
|
| 423 |
+
{
|
| 424 |
+
"epoch": 3.01,
|
| 425 |
+
"grad_norm": 0.7807851433753967,
|
| 426 |
+
"learning_rate": 9.294527802294794e-05,
|
| 427 |
+
"loss": 3.3755,
|
| 428 |
+
"step": 56000
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 3.07,
|
| 432 |
+
"grad_norm": 0.7891095876693726,
|
| 433 |
+
"learning_rate": 9.26510738452486e-05,
|
| 434 |
+
"loss": 3.3311,
|
| 435 |
+
"step": 57000
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 3.12,
|
| 439 |
+
"grad_norm": 0.7911957502365112,
|
| 440 |
+
"learning_rate": 9.235686966754929e-05,
|
| 441 |
+
"loss": 3.3342,
|
| 442 |
+
"step": 58000
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 3.17,
|
| 446 |
+
"grad_norm": 0.7487536072731018,
|
| 447 |
+
"learning_rate": 9.206295969402766e-05,
|
| 448 |
+
"loss": 3.3276,
|
| 449 |
+
"step": 59000
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 3.23,
|
| 453 |
+
"grad_norm": 0.7860606908798218,
|
| 454 |
+
"learning_rate": 9.176875551632832e-05,
|
| 455 |
+
"loss": 3.3254,
|
| 456 |
+
"step": 60000
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 3.28,
|
| 460 |
+
"grad_norm": 0.7669396996498108,
|
| 461 |
+
"learning_rate": 9.147484554280671e-05,
|
| 462 |
+
"loss": 3.3237,
|
| 463 |
+
"step": 61000
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 3.33,
|
| 467 |
+
"grad_norm": 0.7570671439170837,
|
| 468 |
+
"learning_rate": 9.118064136510739e-05,
|
| 469 |
+
"loss": 3.3181,
|
| 470 |
+
"step": 62000
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 3.39,
|
| 474 |
+
"grad_norm": 0.764175295829773,
|
| 475 |
+
"learning_rate": 9.088643718740807e-05,
|
| 476 |
+
"loss": 3.3194,
|
| 477 |
+
"step": 63000
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 3.44,
|
| 481 |
+
"grad_norm": 0.723421573638916,
|
| 482 |
+
"learning_rate": 9.059252721388644e-05,
|
| 483 |
+
"loss": 3.3182,
|
| 484 |
+
"step": 64000
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 3.5,
|
| 488 |
+
"grad_norm": 0.7632123231887817,
|
| 489 |
+
"learning_rate": 9.029861724036481e-05,
|
| 490 |
+
"loss": 3.3158,
|
| 491 |
+
"step": 65000
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 3.55,
|
| 495 |
+
"grad_norm": 0.7372169494628906,
|
| 496 |
+
"learning_rate": 9.00044130626655e-05,
|
| 497 |
+
"loss": 3.3099,
|
| 498 |
+
"step": 66000
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 3.6,
|
| 502 |
+
"grad_norm": 0.7337743639945984,
|
| 503 |
+
"learning_rate": 8.971050308914388e-05,
|
| 504 |
+
"loss": 3.3064,
|
| 505 |
+
"step": 67000
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 3.66,
|
| 509 |
+
"grad_norm": 0.7390604019165039,
|
| 510 |
+
"learning_rate": 8.941629891144455e-05,
|
| 511 |
+
"loss": 3.3011,
|
| 512 |
+
"step": 68000
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 3.71,
|
| 516 |
+
"grad_norm": 0.7338905334472656,
|
| 517 |
+
"learning_rate": 8.912238893792292e-05,
|
| 518 |
+
"loss": 3.3003,
|
| 519 |
+
"step": 69000
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 3.76,
|
| 523 |
+
"grad_norm": 0.7679952383041382,
|
| 524 |
+
"learning_rate": 8.88281847602236e-05,
|
| 525 |
+
"loss": 3.2998,
|
| 526 |
+
"step": 70000
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 3.82,
|
| 530 |
+
"grad_norm": 0.7354390025138855,
|
| 531 |
+
"learning_rate": 8.853398058252428e-05,
|
| 532 |
+
"loss": 3.2903,
|
| 533 |
+
"step": 71000
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 3.87,
|
| 537 |
+
"grad_norm": 0.769103467464447,
|
| 538 |
+
"learning_rate": 8.823977640482495e-05,
|
| 539 |
+
"loss": 3.2916,
|
| 540 |
+
"step": 72000
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 3.93,
|
| 544 |
+
"grad_norm": 0.7130835652351379,
|
| 545 |
+
"learning_rate": 8.794616063548102e-05,
|
| 546 |
+
"loss": 3.2913,
|
| 547 |
+
"step": 73000
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 3.98,
|
| 551 |
+
"grad_norm": 0.7574236989021301,
|
| 552 |
+
"learning_rate": 8.76519564577817e-05,
|
| 553 |
+
"loss": 3.2906,
|
| 554 |
+
"step": 74000
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 4.0,
|
| 558 |
+
"eval_accuracy": 0.389078925897793,
|
| 559 |
+
"eval_loss": 3.505403757095337,
|
| 560 |
+
"eval_runtime": 154.7175,
|
| 561 |
+
"eval_samples_per_second": 374.366,
|
| 562 |
+
"eval_steps_per_second": 5.856,
|
| 563 |
+
"step": 74380
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 4.03,
|
| 567 |
+
"grad_norm": 0.7582383155822754,
|
| 568 |
+
"learning_rate": 8.735775228008238e-05,
|
| 569 |
+
"loss": 3.2494,
|
| 570 |
+
"step": 75000
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 4.09,
|
| 574 |
+
"grad_norm": 0.7709288597106934,
|
| 575 |
+
"learning_rate": 8.706354810238306e-05,
|
| 576 |
+
"loss": 3.2266,
|
| 577 |
+
"step": 76000
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 4.14,
|
| 581 |
+
"grad_norm": 0.7542001605033875,
|
| 582 |
+
"learning_rate": 8.676934392468373e-05,
|
| 583 |
+
"loss": 3.2335,
|
| 584 |
+
"step": 77000
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 4.19,
|
| 588 |
+
"grad_norm": 0.7459376454353333,
|
| 589 |
+
"learning_rate": 8.647513974698441e-05,
|
| 590 |
+
"loss": 3.2389,
|
| 591 |
+
"step": 78000
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 4.25,
|
| 595 |
+
"grad_norm": 0.7790886163711548,
|
| 596 |
+
"learning_rate": 8.61812297734628e-05,
|
| 597 |
+
"loss": 3.2343,
|
| 598 |
+
"step": 79000
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 4.3,
|
| 602 |
+
"grad_norm": 0.7570230960845947,
|
| 603 |
+
"learning_rate": 8.588702559576346e-05,
|
| 604 |
+
"loss": 3.236,
|
| 605 |
+
"step": 80000
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 4.36,
|
| 609 |
+
"grad_norm": 0.7441847324371338,
|
| 610 |
+
"learning_rate": 8.559311562224183e-05,
|
| 611 |
+
"loss": 3.2345,
|
| 612 |
+
"step": 81000
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 4.41,
|
| 616 |
+
"grad_norm": 0.7472123503684998,
|
| 617 |
+
"learning_rate": 8.529891144454252e-05,
|
| 618 |
+
"loss": 3.2338,
|
| 619 |
+
"step": 82000
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 4.46,
|
| 623 |
+
"grad_norm": 0.8000940680503845,
|
| 624 |
+
"learning_rate": 8.50050014710209e-05,
|
| 625 |
+
"loss": 3.2329,
|
| 626 |
+
"step": 83000
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 4.52,
|
| 630 |
+
"grad_norm": 0.763556957244873,
|
| 631 |
+
"learning_rate": 8.471079729332157e-05,
|
| 632 |
+
"loss": 3.2309,
|
| 633 |
+
"step": 84000
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 4.57,
|
| 637 |
+
"grad_norm": 0.7414514422416687,
|
| 638 |
+
"learning_rate": 8.441659311562225e-05,
|
| 639 |
+
"loss": 3.2346,
|
| 640 |
+
"step": 85000
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 4.62,
|
| 644 |
+
"grad_norm": 0.7634936571121216,
|
| 645 |
+
"learning_rate": 8.412238893792293e-05,
|
| 646 |
+
"loss": 3.228,
|
| 647 |
+
"step": 86000
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 4.68,
|
| 651 |
+
"grad_norm": 0.7772691249847412,
|
| 652 |
+
"learning_rate": 8.38284789644013e-05,
|
| 653 |
+
"loss": 3.2234,
|
| 654 |
+
"step": 87000
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 4.73,
|
| 658 |
+
"grad_norm": 0.7693132758140564,
|
| 659 |
+
"learning_rate": 8.353427478670197e-05,
|
| 660 |
+
"loss": 3.2299,
|
| 661 |
+
"step": 88000
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 4.79,
|
| 665 |
+
"grad_norm": 0.7822285294532776,
|
| 666 |
+
"learning_rate": 8.324007060900265e-05,
|
| 667 |
+
"loss": 3.2258,
|
| 668 |
+
"step": 89000
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 4.84,
|
| 672 |
+
"grad_norm": 0.7286272644996643,
|
| 673 |
+
"learning_rate": 8.294616063548103e-05,
|
| 674 |
+
"loss": 3.2252,
|
| 675 |
+
"step": 90000
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 4.89,
|
| 679 |
+
"grad_norm": 0.7409440279006958,
|
| 680 |
+
"learning_rate": 8.265195645778171e-05,
|
| 681 |
+
"loss": 3.2183,
|
| 682 |
+
"step": 91000
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 4.95,
|
| 686 |
+
"grad_norm": 0.7559691667556763,
|
| 687 |
+
"learning_rate": 8.235775228008238e-05,
|
| 688 |
+
"loss": 3.2202,
|
| 689 |
+
"step": 92000
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 5.0,
|
| 693 |
+
"eval_accuracy": 0.39429668627542264,
|
| 694 |
+
"eval_loss": 3.4769208431243896,
|
| 695 |
+
"eval_runtime": 155.051,
|
| 696 |
+
"eval_samples_per_second": 373.561,
|
| 697 |
+
"eval_steps_per_second": 5.843,
|
| 698 |
+
"step": 92975
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 5.0,
|
| 702 |
+
"grad_norm": 0.7476261854171753,
|
| 703 |
+
"learning_rate": 8.206384230656075e-05,
|
| 704 |
+
"loss": 3.2131,
|
| 705 |
+
"step": 93000
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 5.06,
|
| 709 |
+
"grad_norm": 0.7305045127868652,
|
| 710 |
+
"learning_rate": 8.176993233303914e-05,
|
| 711 |
+
"loss": 3.1643,
|
| 712 |
+
"step": 94000
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 5.11,
|
| 716 |
+
"grad_norm": 0.758934497833252,
|
| 717 |
+
"learning_rate": 8.147572815533982e-05,
|
| 718 |
+
"loss": 3.1671,
|
| 719 |
+
"step": 95000
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 5.16,
|
| 723 |
+
"grad_norm": 0.7448575496673584,
|
| 724 |
+
"learning_rate": 8.118152397764049e-05,
|
| 725 |
+
"loss": 3.1674,
|
| 726 |
+
"step": 96000
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 5.22,
|
| 730 |
+
"grad_norm": 0.7979652881622314,
|
| 731 |
+
"learning_rate": 8.088731979994117e-05,
|
| 732 |
+
"loss": 3.1695,
|
| 733 |
+
"step": 97000
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 5.27,
|
| 737 |
+
"grad_norm": 0.7594915628433228,
|
| 738 |
+
"learning_rate": 8.059311562224185e-05,
|
| 739 |
+
"loss": 3.1731,
|
| 740 |
+
"step": 98000
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 5.32,
|
| 744 |
+
"grad_norm": 0.7524967789649963,
|
| 745 |
+
"learning_rate": 8.02992056487202e-05,
|
| 746 |
+
"loss": 3.1724,
|
| 747 |
+
"step": 99000
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 5.38,
|
| 751 |
+
"grad_norm": 0.7075839042663574,
|
| 752 |
+
"learning_rate": 8.000500147102089e-05,
|
| 753 |
+
"loss": 3.1744,
|
| 754 |
+
"step": 100000
|
| 755 |
+
},
|
| 756 |
+
{
|
| 757 |
+
"epoch": 5.43,
|
| 758 |
+
"grad_norm": 0.7107969522476196,
|
| 759 |
+
"learning_rate": 7.971109149749927e-05,
|
| 760 |
+
"loss": 3.1691,
|
| 761 |
+
"step": 101000
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 5.49,
|
| 765 |
+
"grad_norm": 0.7338487505912781,
|
| 766 |
+
"learning_rate": 7.941688731979995e-05,
|
| 767 |
+
"loss": 3.1714,
|
| 768 |
+
"step": 102000
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 5.54,
|
| 772 |
+
"grad_norm": 0.7563539743423462,
|
| 773 |
+
"learning_rate": 7.912268314210062e-05,
|
| 774 |
+
"loss": 3.1702,
|
| 775 |
+
"step": 103000
|
| 776 |
+
},
|
| 777 |
+
{
|
| 778 |
+
"epoch": 5.59,
|
| 779 |
+
"grad_norm": 0.7359240055084229,
|
| 780 |
+
"learning_rate": 7.882877316857899e-05,
|
| 781 |
+
"loss": 3.1703,
|
| 782 |
+
"step": 104000
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 5.65,
|
| 786 |
+
"grad_norm": 0.7448987364768982,
|
| 787 |
+
"learning_rate": 7.853456899087967e-05,
|
| 788 |
+
"loss": 3.172,
|
| 789 |
+
"step": 105000
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 5.7,
|
| 793 |
+
"grad_norm": 0.7606428265571594,
|
| 794 |
+
"learning_rate": 7.824036481318035e-05,
|
| 795 |
+
"loss": 3.1726,
|
| 796 |
+
"step": 106000
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"epoch": 5.75,
|
| 800 |
+
"grad_norm": 0.7235444784164429,
|
| 801 |
+
"learning_rate": 7.794616063548102e-05,
|
| 802 |
+
"loss": 3.1733,
|
| 803 |
+
"step": 107000
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 5.81,
|
| 807 |
+
"grad_norm": 0.7561464905738831,
|
| 808 |
+
"learning_rate": 7.76522506619594e-05,
|
| 809 |
+
"loss": 3.1696,
|
| 810 |
+
"step": 108000
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 5.86,
|
| 814 |
+
"grad_norm": 0.7387487292289734,
|
| 815 |
+
"learning_rate": 7.735804648426009e-05,
|
| 816 |
+
"loss": 3.1705,
|
| 817 |
+
"step": 109000
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 5.92,
|
| 821 |
+
"grad_norm": 0.7437654137611389,
|
| 822 |
+
"learning_rate": 7.706443071491616e-05,
|
| 823 |
+
"loss": 3.1642,
|
| 824 |
+
"step": 110000
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 5.97,
|
| 828 |
+
"grad_norm": 0.7192903161048889,
|
| 829 |
+
"learning_rate": 7.677022653721683e-05,
|
| 830 |
+
"loss": 3.1699,
|
| 831 |
+
"step": 111000
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 6.0,
|
| 835 |
+
"eval_accuracy": 0.3977428802964349,
|
| 836 |
+
"eval_loss": 3.4425718784332275,
|
| 837 |
+
"eval_runtime": 154.7206,
|
| 838 |
+
"eval_samples_per_second": 374.359,
|
| 839 |
+
"eval_steps_per_second": 5.856,
|
| 840 |
+
"step": 111570
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"epoch": 6.02,
|
| 844 |
+
"grad_norm": 0.7664169073104858,
|
| 845 |
+
"learning_rate": 7.647602235951751e-05,
|
| 846 |
+
"loss": 3.1431,
|
| 847 |
+
"step": 112000
|
| 848 |
+
},
|
| 849 |
+
{
|
| 850 |
+
"epoch": 6.08,
|
| 851 |
+
"grad_norm": 0.7941229343414307,
|
| 852 |
+
"learning_rate": 7.618181818181819e-05,
|
| 853 |
+
"loss": 3.1107,
|
| 854 |
+
"step": 113000
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 6.13,
|
| 858 |
+
"grad_norm": 0.7529132962226868,
|
| 859 |
+
"learning_rate": 7.588761400411886e-05,
|
| 860 |
+
"loss": 3.1141,
|
| 861 |
+
"step": 114000
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 6.18,
|
| 865 |
+
"grad_norm": 0.7258307933807373,
|
| 866 |
+
"learning_rate": 7.559370403059724e-05,
|
| 867 |
+
"loss": 3.1197,
|
| 868 |
+
"step": 115000
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 6.24,
|
| 872 |
+
"grad_norm": 0.7766565680503845,
|
| 873 |
+
"learning_rate": 7.529949985289791e-05,
|
| 874 |
+
"loss": 3.1211,
|
| 875 |
+
"step": 116000
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 6.29,
|
| 879 |
+
"grad_norm": 0.7603769302368164,
|
| 880 |
+
"learning_rate": 7.5005884083554e-05,
|
| 881 |
+
"loss": 3.1226,
|
| 882 |
+
"step": 117000
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 6.35,
|
| 886 |
+
"grad_norm": 0.7358362674713135,
|
| 887 |
+
"learning_rate": 7.471167990585468e-05,
|
| 888 |
+
"loss": 3.1224,
|
| 889 |
+
"step": 118000
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 6.4,
|
| 893 |
+
"grad_norm": 0.7426795363426208,
|
| 894 |
+
"learning_rate": 7.441747572815534e-05,
|
| 895 |
+
"loss": 3.1258,
|
| 896 |
+
"step": 119000
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 6.45,
|
| 900 |
+
"grad_norm": 0.760974645614624,
|
| 901 |
+
"learning_rate": 7.412327155045603e-05,
|
| 902 |
+
"loss": 3.1284,
|
| 903 |
+
"step": 120000
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 6.51,
|
| 907 |
+
"grad_norm": 0.7280532717704773,
|
| 908 |
+
"learning_rate": 7.38293615769344e-05,
|
| 909 |
+
"loss": 3.1269,
|
| 910 |
+
"step": 121000
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 6.56,
|
| 914 |
+
"grad_norm": 0.7262037396430969,
|
| 915 |
+
"learning_rate": 7.353515739923508e-05,
|
| 916 |
+
"loss": 3.1293,
|
| 917 |
+
"step": 122000
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 6.61,
|
| 921 |
+
"grad_norm": 0.7488662600517273,
|
| 922 |
+
"learning_rate": 7.324124742571345e-05,
|
| 923 |
+
"loss": 3.1244,
|
| 924 |
+
"step": 123000
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 6.67,
|
| 928 |
+
"grad_norm": 0.7436386346817017,
|
| 929 |
+
"learning_rate": 7.294704324801413e-05,
|
| 930 |
+
"loss": 3.1273,
|
| 931 |
+
"step": 124000
|
| 932 |
+
},
|
| 933 |
+
{
|
| 934 |
+
"epoch": 6.72,
|
| 935 |
+
"grad_norm": 0.7284784913063049,
|
| 936 |
+
"learning_rate": 7.265283907031481e-05,
|
| 937 |
+
"loss": 3.1301,
|
| 938 |
+
"step": 125000
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 6.78,
|
| 942 |
+
"grad_norm": 0.7190147638320923,
|
| 943 |
+
"learning_rate": 7.235863489261548e-05,
|
| 944 |
+
"loss": 3.1281,
|
| 945 |
+
"step": 126000
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 6.83,
|
| 949 |
+
"grad_norm": 0.7366654276847839,
|
| 950 |
+
"learning_rate": 7.206472491909385e-05,
|
| 951 |
+
"loss": 3.1286,
|
| 952 |
+
"step": 127000
|
| 953 |
+
},
|
| 954 |
+
{
|
| 955 |
+
"epoch": 6.88,
|
| 956 |
+
"grad_norm": 0.7567622065544128,
|
| 957 |
+
"learning_rate": 7.177081494557223e-05,
|
| 958 |
+
"loss": 3.1262,
|
| 959 |
+
"step": 128000
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 6.94,
|
| 963 |
+
"grad_norm": 0.7190493941307068,
|
| 964 |
+
"learning_rate": 7.147661076787291e-05,
|
| 965 |
+
"loss": 3.1309,
|
| 966 |
+
"step": 129000
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 6.99,
|
| 970 |
+
"grad_norm": 0.7200164794921875,
|
| 971 |
+
"learning_rate": 7.11824065901736e-05,
|
| 972 |
+
"loss": 3.1232,
|
| 973 |
+
"step": 130000
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 7.0,
|
| 977 |
+
"eval_accuracy": 0.399363026131729,
|
| 978 |
+
"eval_loss": 3.447763442993164,
|
| 979 |
+
"eval_runtime": 155.0052,
|
| 980 |
+
"eval_samples_per_second": 373.671,
|
| 981 |
+
"eval_steps_per_second": 5.845,
|
| 982 |
+
"step": 130165
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 7.04,
|
| 986 |
+
"grad_norm": 0.7389765977859497,
|
| 987 |
+
"learning_rate": 7.088820241247426e-05,
|
| 988 |
+
"loss": 3.0769,
|
| 989 |
+
"step": 131000
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 7.1,
|
| 993 |
+
"grad_norm": 0.7703937888145447,
|
| 994 |
+
"learning_rate": 7.059429243895263e-05,
|
| 995 |
+
"loss": 3.0708,
|
| 996 |
+
"step": 132000
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 7.15,
|
| 1000 |
+
"grad_norm": 0.7399336099624634,
|
| 1001 |
+
"learning_rate": 7.030008826125331e-05,
|
| 1002 |
+
"loss": 3.0763,
|
| 1003 |
+
"step": 133000
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 7.21,
|
| 1007 |
+
"grad_norm": 0.7627936005592346,
|
| 1008 |
+
"learning_rate": 7.000588408355398e-05,
|
| 1009 |
+
"loss": 3.0798,
|
| 1010 |
+
"step": 134000
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 7.26,
|
| 1014 |
+
"grad_norm": 0.7476488351821899,
|
| 1015 |
+
"learning_rate": 6.971167990585466e-05,
|
| 1016 |
+
"loss": 3.0854,
|
| 1017 |
+
"step": 135000
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 7.31,
|
| 1021 |
+
"grad_norm": 0.7593668103218079,
|
| 1022 |
+
"learning_rate": 6.941776993233305e-05,
|
| 1023 |
+
"loss": 3.0837,
|
| 1024 |
+
"step": 136000
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 7.37,
|
| 1028 |
+
"grad_norm": 0.7753123044967651,
|
| 1029 |
+
"learning_rate": 6.912356575463373e-05,
|
| 1030 |
+
"loss": 3.0842,
|
| 1031 |
+
"step": 137000
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 7.42,
|
| 1035 |
+
"grad_norm": 0.7626084089279175,
|
| 1036 |
+
"learning_rate": 6.882965578111209e-05,
|
| 1037 |
+
"loss": 3.0883,
|
| 1038 |
+
"step": 138000
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 7.48,
|
| 1042 |
+
"grad_norm": 0.7539852857589722,
|
| 1043 |
+
"learning_rate": 6.853545160341277e-05,
|
| 1044 |
+
"loss": 3.0873,
|
| 1045 |
+
"step": 139000
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 7.53,
|
| 1049 |
+
"grad_norm": 0.7760306000709534,
|
| 1050 |
+
"learning_rate": 6.824154162989115e-05,
|
| 1051 |
+
"loss": 3.0932,
|
| 1052 |
+
"step": 140000
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 7.58,
|
| 1056 |
+
"grad_norm": 0.7804057598114014,
|
| 1057 |
+
"learning_rate": 6.794733745219183e-05,
|
| 1058 |
+
"loss": 3.0908,
|
| 1059 |
+
"step": 141000
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 7.64,
|
| 1063 |
+
"grad_norm": 0.7768170833587646,
|
| 1064 |
+
"learning_rate": 6.76531332744925e-05,
|
| 1065 |
+
"loss": 3.0889,
|
| 1066 |
+
"step": 142000
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 7.69,
|
| 1070 |
+
"grad_norm": 0.7796506881713867,
|
| 1071 |
+
"learning_rate": 6.735892909679318e-05,
|
| 1072 |
+
"loss": 3.0864,
|
| 1073 |
+
"step": 143000
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 7.74,
|
| 1077 |
+
"grad_norm": 0.7837468981742859,
|
| 1078 |
+
"learning_rate": 6.706472491909386e-05,
|
| 1079 |
+
"loss": 3.0948,
|
| 1080 |
+
"step": 144000
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 7.8,
|
| 1084 |
+
"grad_norm": 0.7560127377510071,
|
| 1085 |
+
"learning_rate": 6.677081494557223e-05,
|
| 1086 |
+
"loss": 3.0946,
|
| 1087 |
+
"step": 145000
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 7.85,
|
| 1091 |
+
"grad_norm": 0.7457463145256042,
|
| 1092 |
+
"learning_rate": 6.64769049720506e-05,
|
| 1093 |
+
"loss": 3.092,
|
| 1094 |
+
"step": 146000
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 7.91,
|
| 1098 |
+
"grad_norm": 0.7344046235084534,
|
| 1099 |
+
"learning_rate": 6.618270079435128e-05,
|
| 1100 |
+
"loss": 3.0904,
|
| 1101 |
+
"step": 147000
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 7.96,
|
| 1105 |
+
"grad_norm": 0.7419469356536865,
|
| 1106 |
+
"learning_rate": 6.588849661665197e-05,
|
| 1107 |
+
"loss": 3.091,
|
| 1108 |
+
"step": 148000
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 8.0,
|
| 1112 |
+
"eval_accuracy": 0.4014489882469057,
|
| 1113 |
+
"eval_loss": 3.42431378364563,
|
| 1114 |
+
"eval_runtime": 154.6897,
|
| 1115 |
+
"eval_samples_per_second": 374.433,
|
| 1116 |
+
"eval_steps_per_second": 5.857,
|
| 1117 |
+
"step": 148760
|
| 1118 |
+
},
|
| 1119 |
+
{
|
| 1120 |
+
"epoch": 8.01,
|
| 1121 |
+
"grad_norm": 0.7680034041404724,
|
| 1122 |
+
"learning_rate": 6.559458664313034e-05,
|
| 1123 |
+
"loss": 3.0783,
|
| 1124 |
+
"step": 149000
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 8.07,
|
| 1128 |
+
"grad_norm": 0.7723122239112854,
|
| 1129 |
+
"learning_rate": 6.5300382465431e-05,
|
| 1130 |
+
"loss": 3.0382,
|
| 1131 |
+
"step": 150000
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 8.12,
|
| 1135 |
+
"grad_norm": 0.8067519068717957,
|
| 1136 |
+
"learning_rate": 6.500617828773169e-05,
|
| 1137 |
+
"loss": 3.0417,
|
| 1138 |
+
"step": 151000
|
| 1139 |
+
},
|
| 1140 |
+
{
|
| 1141 |
+
"epoch": 8.17,
|
| 1142 |
+
"grad_norm": 0.7817720770835876,
|
| 1143 |
+
"learning_rate": 6.471197411003237e-05,
|
| 1144 |
+
"loss": 3.0428,
|
| 1145 |
+
"step": 152000
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 8.23,
|
| 1149 |
+
"grad_norm": 0.785193145275116,
|
| 1150 |
+
"learning_rate": 6.441806413651074e-05,
|
| 1151 |
+
"loss": 3.0471,
|
| 1152 |
+
"step": 153000
|
| 1153 |
+
},
|
| 1154 |
+
{
|
| 1155 |
+
"epoch": 8.28,
|
| 1156 |
+
"grad_norm": 0.7578865885734558,
|
| 1157 |
+
"learning_rate": 6.412385995881142e-05,
|
| 1158 |
+
"loss": 3.049,
|
| 1159 |
+
"step": 154000
|
| 1160 |
+
},
|
| 1161 |
+
{
|
| 1162 |
+
"epoch": 8.34,
|
| 1163 |
+
"grad_norm": 0.75961834192276,
|
| 1164 |
+
"learning_rate": 6.382994998528979e-05,
|
| 1165 |
+
"loss": 3.0465,
|
| 1166 |
+
"step": 155000
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 8.39,
|
| 1170 |
+
"grad_norm": 0.7525945901870728,
|
| 1171 |
+
"learning_rate": 6.353574580759047e-05,
|
| 1172 |
+
"loss": 3.0529,
|
| 1173 |
+
"step": 156000
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 8.44,
|
| 1177 |
+
"grad_norm": 0.7648568153381348,
|
| 1178 |
+
"learning_rate": 6.324183583406885e-05,
|
| 1179 |
+
"loss": 3.055,
|
| 1180 |
+
"step": 157000
|
| 1181 |
+
},
|
| 1182 |
+
{
|
| 1183 |
+
"epoch": 8.5,
|
| 1184 |
+
"grad_norm": 0.7817012071609497,
|
| 1185 |
+
"learning_rate": 6.294763165636952e-05,
|
| 1186 |
+
"loss": 3.059,
|
| 1187 |
+
"step": 158000
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 8.55,
|
| 1191 |
+
"grad_norm": 0.7443410158157349,
|
| 1192 |
+
"learning_rate": 6.26534274786702e-05,
|
| 1193 |
+
"loss": 3.0597,
|
| 1194 |
+
"step": 159000
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 8.6,
|
| 1198 |
+
"grad_norm": 0.7928184866905212,
|
| 1199 |
+
"learning_rate": 6.235951750514857e-05,
|
| 1200 |
+
"loss": 3.0572,
|
| 1201 |
+
"step": 160000
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 8.66,
|
| 1205 |
+
"grad_norm": 0.7541539669036865,
|
| 1206 |
+
"learning_rate": 6.206531332744924e-05,
|
| 1207 |
+
"loss": 3.0582,
|
| 1208 |
+
"step": 161000
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 8.71,
|
| 1212 |
+
"grad_norm": 0.7730678915977478,
|
| 1213 |
+
"learning_rate": 6.177110914974992e-05,
|
| 1214 |
+
"loss": 3.0563,
|
| 1215 |
+
"step": 162000
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 8.77,
|
| 1219 |
+
"grad_norm": 0.7646264433860779,
|
| 1220 |
+
"learning_rate": 6.14769049720506e-05,
|
| 1221 |
+
"loss": 3.0607,
|
| 1222 |
+
"step": 163000
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"epoch": 8.82,
|
| 1226 |
+
"grad_norm": 0.7786790728569031,
|
| 1227 |
+
"learning_rate": 6.118270079435127e-05,
|
| 1228 |
+
"loss": 3.0632,
|
| 1229 |
+
"step": 164000
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 8.87,
|
| 1233 |
+
"grad_norm": 0.7602713704109192,
|
| 1234 |
+
"learning_rate": 6.0888790820829656e-05,
|
| 1235 |
+
"loss": 3.0601,
|
| 1236 |
+
"step": 165000
|
| 1237 |
+
},
|
| 1238 |
+
{
|
| 1239 |
+
"epoch": 8.93,
|
| 1240 |
+
"grad_norm": 0.7597308158874512,
|
| 1241 |
+
"learning_rate": 6.059458664313034e-05,
|
| 1242 |
+
"loss": 3.0615,
|
| 1243 |
+
"step": 166000
|
| 1244 |
+
},
|
| 1245 |
+
{
|
| 1246 |
+
"epoch": 8.98,
|
| 1247 |
+
"grad_norm": 0.8023492693901062,
|
| 1248 |
+
"learning_rate": 6.030038246543102e-05,
|
| 1249 |
+
"loss": 3.0613,
|
| 1250 |
+
"step": 167000
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 9.0,
|
| 1254 |
+
"eval_accuracy": 0.4030179440035626,
|
| 1255 |
+
"eval_loss": 3.4169299602508545,
|
| 1256 |
+
"eval_runtime": 154.797,
|
| 1257 |
+
"eval_samples_per_second": 374.174,
|
| 1258 |
+
"eval_steps_per_second": 5.853,
|
| 1259 |
+
"step": 167355
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 9.03,
|
| 1263 |
+
"grad_norm": 0.7707279324531555,
|
| 1264 |
+
"learning_rate": 6.000676669608709e-05,
|
| 1265 |
+
"loss": 3.0235,
|
| 1266 |
+
"step": 168000
|
| 1267 |
+
},
|
| 1268 |
+
{
|
| 1269 |
+
"epoch": 9.09,
|
| 1270 |
+
"grad_norm": 0.8100076913833618,
|
| 1271 |
+
"learning_rate": 5.971256251838776e-05,
|
| 1272 |
+
"loss": 3.0067,
|
| 1273 |
+
"step": 169000
|
| 1274 |
+
},
|
| 1275 |
+
{
|
| 1276 |
+
"epoch": 9.14,
|
| 1277 |
+
"grad_norm": 0.761734127998352,
|
| 1278 |
+
"learning_rate": 5.941835834068844e-05,
|
| 1279 |
+
"loss": 3.0091,
|
| 1280 |
+
"step": 170000
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 9.2,
|
| 1284 |
+
"grad_norm": 0.8010926842689514,
|
| 1285 |
+
"learning_rate": 5.912415416298912e-05,
|
| 1286 |
+
"loss": 3.0168,
|
| 1287 |
+
"step": 171000
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 9.25,
|
| 1291 |
+
"grad_norm": 0.8410947322845459,
|
| 1292 |
+
"learning_rate": 5.88302441894675e-05,
|
| 1293 |
+
"loss": 3.014,
|
| 1294 |
+
"step": 172000
|
| 1295 |
+
},
|
| 1296 |
+
{
|
| 1297 |
+
"epoch": 9.3,
|
| 1298 |
+
"grad_norm": 0.7763057351112366,
|
| 1299 |
+
"learning_rate": 5.853633421594586e-05,
|
| 1300 |
+
"loss": 3.019,
|
| 1301 |
+
"step": 173000
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 9.36,
|
| 1305 |
+
"grad_norm": 0.7941156625747681,
|
| 1306 |
+
"learning_rate": 5.8242130038246544e-05,
|
| 1307 |
+
"loss": 3.0273,
|
| 1308 |
+
"step": 174000
|
| 1309 |
+
},
|
| 1310 |
+
{
|
| 1311 |
+
"epoch": 9.41,
|
| 1312 |
+
"grad_norm": 0.821797251701355,
|
| 1313 |
+
"learning_rate": 5.7947925860547225e-05,
|
| 1314 |
+
"loss": 3.0265,
|
| 1315 |
+
"step": 175000
|
| 1316 |
+
},
|
| 1317 |
+
{
|
| 1318 |
+
"epoch": 9.46,
|
| 1319 |
+
"grad_norm": 0.7570799589157104,
|
| 1320 |
+
"learning_rate": 5.76540158870256e-05,
|
| 1321 |
+
"loss": 3.0298,
|
| 1322 |
+
"step": 176000
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 9.52,
|
| 1326 |
+
"grad_norm": 0.7869284152984619,
|
| 1327 |
+
"learning_rate": 5.735981170932627e-05,
|
| 1328 |
+
"loss": 3.0312,
|
| 1329 |
+
"step": 177000
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 9.57,
|
| 1333 |
+
"grad_norm": 0.8196205496788025,
|
| 1334 |
+
"learning_rate": 5.706560753162695e-05,
|
| 1335 |
+
"loss": 3.0273,
|
| 1336 |
+
"step": 178000
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"epoch": 9.63,
|
| 1340 |
+
"grad_norm": 0.7756869792938232,
|
| 1341 |
+
"learning_rate": 5.677169755810533e-05,
|
| 1342 |
+
"loss": 3.0253,
|
| 1343 |
+
"step": 179000
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 9.68,
|
| 1347 |
+
"grad_norm": 0.8128073811531067,
|
| 1348 |
+
"learning_rate": 5.6477493380406e-05,
|
| 1349 |
+
"loss": 3.0288,
|
| 1350 |
+
"step": 180000
|
| 1351 |
+
},
|
| 1352 |
+
{
|
| 1353 |
+
"epoch": 9.73,
|
| 1354 |
+
"grad_norm": 0.7797793745994568,
|
| 1355 |
+
"learning_rate": 5.618328920270668e-05,
|
| 1356 |
+
"loss": 3.0322,
|
| 1357 |
+
"step": 181000
|
| 1358 |
+
},
|
| 1359 |
+
{
|
| 1360 |
+
"epoch": 9.79,
|
| 1361 |
+
"grad_norm": 0.789199948310852,
|
| 1362 |
+
"learning_rate": 5.588908502500736e-05,
|
| 1363 |
+
"loss": 3.036,
|
| 1364 |
+
"step": 182000
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 9.84,
|
| 1368 |
+
"grad_norm": 0.7418550848960876,
|
| 1369 |
+
"learning_rate": 5.5595175051485736e-05,
|
| 1370 |
+
"loss": 3.0368,
|
| 1371 |
+
"step": 183000
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 9.9,
|
| 1375 |
+
"grad_norm": 0.8195221424102783,
|
| 1376 |
+
"learning_rate": 5.5300970873786404e-05,
|
| 1377 |
+
"loss": 3.0334,
|
| 1378 |
+
"step": 184000
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 9.95,
|
| 1382 |
+
"grad_norm": 0.7761098742485046,
|
| 1383 |
+
"learning_rate": 5.5006766696087085e-05,
|
| 1384 |
+
"loss": 3.0322,
|
| 1385 |
+
"step": 185000
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 10.0,
|
| 1389 |
+
"eval_accuracy": 0.4050055566763182,
|
| 1390 |
+
"eval_loss": 3.4141781330108643,
|
| 1391 |
+
"eval_runtime": 154.9579,
|
| 1392 |
+
"eval_samples_per_second": 373.785,
|
| 1393 |
+
"eval_steps_per_second": 5.847,
|
| 1394 |
+
"step": 185950
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 10.0,
|
| 1398 |
+
"grad_norm": 0.7680907249450684,
|
| 1399 |
+
"learning_rate": 5.471256251838777e-05,
|
| 1400 |
+
"loss": 3.0339,
|
| 1401 |
+
"step": 186000
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 10.06,
|
| 1405 |
+
"grad_norm": 0.7804239392280579,
|
| 1406 |
+
"learning_rate": 5.441894674904384e-05,
|
| 1407 |
+
"loss": 2.9809,
|
| 1408 |
+
"step": 187000
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"epoch": 10.11,
|
| 1412 |
+
"grad_norm": 0.8002775311470032,
|
| 1413 |
+
"learning_rate": 5.4124742571344514e-05,
|
| 1414 |
+
"loss": 2.9844,
|
| 1415 |
+
"step": 188000
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 10.16,
|
| 1419 |
+
"grad_norm": 0.8047283291816711,
|
| 1420 |
+
"learning_rate": 5.383083259782289e-05,
|
| 1421 |
+
"loss": 2.9912,
|
| 1422 |
+
"step": 189000
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 10.22,
|
| 1426 |
+
"grad_norm": 0.8343998193740845,
|
| 1427 |
+
"learning_rate": 5.353692262430127e-05,
|
| 1428 |
+
"loss": 2.989,
|
| 1429 |
+
"step": 190000
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 10.27,
|
| 1433 |
+
"grad_norm": 0.798310399055481,
|
| 1434 |
+
"learning_rate": 5.3242718446601943e-05,
|
| 1435 |
+
"loss": 2.9925,
|
| 1436 |
+
"step": 191000
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 10.33,
|
| 1440 |
+
"grad_norm": 0.8054454326629639,
|
| 1441 |
+
"learning_rate": 5.294880847308033e-05,
|
| 1442 |
+
"loss": 2.9986,
|
| 1443 |
+
"step": 192000
|
| 1444 |
+
},
|
| 1445 |
+
{
|
| 1446 |
+
"epoch": 10.38,
|
| 1447 |
+
"grad_norm": 0.810845136642456,
|
| 1448 |
+
"learning_rate": 5.2654604295380995e-05,
|
| 1449 |
+
"loss": 2.9996,
|
| 1450 |
+
"step": 193000
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 10.43,
|
| 1454 |
+
"grad_norm": 0.7934587597846985,
|
| 1455 |
+
"learning_rate": 5.2360400117681677e-05,
|
| 1456 |
+
"loss": 3.0,
|
| 1457 |
+
"step": 194000
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 10.49,
|
| 1461 |
+
"grad_norm": 0.8145301938056946,
|
| 1462 |
+
"learning_rate": 5.206619593998235e-05,
|
| 1463 |
+
"loss": 2.9977,
|
| 1464 |
+
"step": 195000
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 10.54,
|
| 1468 |
+
"grad_norm": 0.7856894135475159,
|
| 1469 |
+
"learning_rate": 5.177228596646072e-05,
|
| 1470 |
+
"loss": 3.0025,
|
| 1471 |
+
"step": 196000
|
| 1472 |
+
},
|
| 1473 |
+
{
|
| 1474 |
+
"epoch": 10.59,
|
| 1475 |
+
"grad_norm": 0.8008938431739807,
|
| 1476 |
+
"learning_rate": 5.14780817887614e-05,
|
| 1477 |
+
"loss": 3.0059,
|
| 1478 |
+
"step": 197000
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 10.65,
|
| 1482 |
+
"grad_norm": 0.8000041246414185,
|
| 1483 |
+
"learning_rate": 5.1183877611062084e-05,
|
| 1484 |
+
"loss": 3.0025,
|
| 1485 |
+
"step": 198000
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 10.7,
|
| 1489 |
+
"grad_norm": 0.8157011866569519,
|
| 1490 |
+
"learning_rate": 5.088967343336276e-05,
|
| 1491 |
+
"loss": 3.0018,
|
| 1492 |
+
"step": 199000
|
| 1493 |
+
},
|
| 1494 |
+
{
|
| 1495 |
+
"epoch": 10.76,
|
| 1496 |
+
"grad_norm": 0.7584720253944397,
|
| 1497 |
+
"learning_rate": 5.059576345984113e-05,
|
| 1498 |
+
"loss": 3.0051,
|
| 1499 |
+
"step": 200000
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 10.81,
|
| 1503 |
+
"grad_norm": 0.7773544192314148,
|
| 1504 |
+
"learning_rate": 5.030155928214181e-05,
|
| 1505 |
+
"loss": 3.0052,
|
| 1506 |
+
"step": 201000
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 10.86,
|
| 1510 |
+
"grad_norm": 0.7875356078147888,
|
| 1511 |
+
"learning_rate": 5.000764930862019e-05,
|
| 1512 |
+
"loss": 3.0137,
|
| 1513 |
+
"step": 202000
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 10.92,
|
| 1517 |
+
"grad_norm": 0.7761446833610535,
|
| 1518 |
+
"learning_rate": 4.971344513092086e-05,
|
| 1519 |
+
"loss": 3.0143,
|
| 1520 |
+
"step": 203000
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 10.97,
|
| 1524 |
+
"grad_norm": 0.8193157315254211,
|
| 1525 |
+
"learning_rate": 4.941924095322154e-05,
|
| 1526 |
+
"loss": 3.0107,
|
| 1527 |
+
"step": 204000
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 11.0,
|
| 1531 |
+
"eval_accuracy": 0.40485742790545853,
|
| 1532 |
+
"eval_loss": 3.4051809310913086,
|
| 1533 |
+
"eval_runtime": 154.8665,
|
| 1534 |
+
"eval_samples_per_second": 374.006,
|
| 1535 |
+
"eval_steps_per_second": 5.85,
|
| 1536 |
+
"step": 204545
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 11.02,
|
| 1540 |
+
"grad_norm": 0.8065283298492432,
|
| 1541 |
+
"learning_rate": 4.9125330979699914e-05,
|
| 1542 |
+
"loss": 2.9828,
|
| 1543 |
+
"step": 205000
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 11.08,
|
| 1547 |
+
"grad_norm": 0.8200777769088745,
|
| 1548 |
+
"learning_rate": 4.8831126802000595e-05,
|
| 1549 |
+
"loss": 2.96,
|
| 1550 |
+
"step": 206000
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 11.13,
|
| 1554 |
+
"grad_norm": 0.8049452900886536,
|
| 1555 |
+
"learning_rate": 4.8537216828478965e-05,
|
| 1556 |
+
"loss": 2.9624,
|
| 1557 |
+
"step": 207000
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 11.19,
|
| 1561 |
+
"grad_norm": 0.8497307300567627,
|
| 1562 |
+
"learning_rate": 4.824301265077965e-05,
|
| 1563 |
+
"loss": 2.971,
|
| 1564 |
+
"step": 208000
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 11.24,
|
| 1568 |
+
"grad_norm": 0.815825343132019,
|
| 1569 |
+
"learning_rate": 4.794880847308032e-05,
|
| 1570 |
+
"loss": 2.9686,
|
| 1571 |
+
"step": 209000
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 11.29,
|
| 1575 |
+
"grad_norm": 0.8359244465827942,
|
| 1576 |
+
"learning_rate": 4.7654604295380996e-05,
|
| 1577 |
+
"loss": 2.972,
|
| 1578 |
+
"step": 210000
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 11.35,
|
| 1582 |
+
"grad_norm": 0.804949164390564,
|
| 1583 |
+
"learning_rate": 4.736069432185937e-05,
|
| 1584 |
+
"loss": 2.9756,
|
| 1585 |
+
"step": 211000
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 11.4,
|
| 1589 |
+
"grad_norm": 0.8300769925117493,
|
| 1590 |
+
"learning_rate": 4.706678434833775e-05,
|
| 1591 |
+
"loss": 2.9777,
|
| 1592 |
+
"step": 212000
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 11.45,
|
| 1596 |
+
"grad_norm": 0.8034248352050781,
|
| 1597 |
+
"learning_rate": 4.6772580170638425e-05,
|
| 1598 |
+
"loss": 2.9736,
|
| 1599 |
+
"step": 213000
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 11.51,
|
| 1603 |
+
"grad_norm": 0.7853664755821228,
|
| 1604 |
+
"learning_rate": 4.64783759929391e-05,
|
| 1605 |
+
"loss": 2.9769,
|
| 1606 |
+
"step": 214000
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 11.56,
|
| 1610 |
+
"grad_norm": 0.8291251063346863,
|
| 1611 |
+
"learning_rate": 4.618417181523978e-05,
|
| 1612 |
+
"loss": 2.9748,
|
| 1613 |
+
"step": 215000
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 11.62,
|
| 1617 |
+
"grad_norm": 0.8467937111854553,
|
| 1618 |
+
"learning_rate": 4.5890556045895854e-05,
|
| 1619 |
+
"loss": 2.9849,
|
| 1620 |
+
"step": 216000
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 11.67,
|
| 1624 |
+
"grad_norm": 0.8191725015640259,
|
| 1625 |
+
"learning_rate": 4.559635186819653e-05,
|
| 1626 |
+
"loss": 2.983,
|
| 1627 |
+
"step": 217000
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 11.72,
|
| 1631 |
+
"grad_norm": 0.7857011556625366,
|
| 1632 |
+
"learning_rate": 4.530214769049721e-05,
|
| 1633 |
+
"loss": 2.9797,
|
| 1634 |
+
"step": 218000
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 11.78,
|
| 1638 |
+
"grad_norm": 0.8048397302627563,
|
| 1639 |
+
"learning_rate": 4.500823771697558e-05,
|
| 1640 |
+
"loss": 2.9844,
|
| 1641 |
+
"step": 219000
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 11.83,
|
| 1645 |
+
"grad_norm": 0.816554069519043,
|
| 1646 |
+
"learning_rate": 4.471403353927626e-05,
|
| 1647 |
+
"loss": 2.986,
|
| 1648 |
+
"step": 220000
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 11.88,
|
| 1652 |
+
"grad_norm": 0.8050833940505981,
|
| 1653 |
+
"learning_rate": 4.442012356575463e-05,
|
| 1654 |
+
"loss": 2.9887,
|
| 1655 |
+
"step": 221000
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 11.94,
|
| 1659 |
+
"grad_norm": 0.8382700681686401,
|
| 1660 |
+
"learning_rate": 4.412621359223301e-05,
|
| 1661 |
+
"loss": 2.9892,
|
| 1662 |
+
"step": 222000
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 11.99,
|
| 1666 |
+
"grad_norm": 0.8233689665794373,
|
| 1667 |
+
"learning_rate": 4.3832009414533684e-05,
|
| 1668 |
+
"loss": 2.9844,
|
| 1669 |
+
"step": 223000
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 12.0,
|
| 1673 |
+
"eval_accuracy": 0.4053066510758616,
|
| 1674 |
+
"eval_loss": 3.4128060340881348,
|
| 1675 |
+
"eval_runtime": 155.24,
|
| 1676 |
+
"eval_samples_per_second": 373.106,
|
| 1677 |
+
"eval_steps_per_second": 5.836,
|
| 1678 |
+
"step": 223140
|
| 1679 |
+
},
|
| 1680 |
+
{
|
| 1681 |
+
"epoch": 12.05,
|
| 1682 |
+
"grad_norm": 0.8400596976280212,
|
| 1683 |
+
"learning_rate": 4.353809944101207e-05,
|
| 1684 |
+
"loss": 2.9383,
|
| 1685 |
+
"step": 224000
|
| 1686 |
+
},
|
| 1687 |
+
{
|
| 1688 |
+
"epoch": 12.1,
|
| 1689 |
+
"grad_norm": 0.8800355792045593,
|
| 1690 |
+
"learning_rate": 4.324389526331274e-05,
|
| 1691 |
+
"loss": 2.9441,
|
| 1692 |
+
"step": 225000
|
| 1693 |
+
},
|
| 1694 |
+
{
|
| 1695 |
+
"epoch": 12.15,
|
| 1696 |
+
"grad_norm": 0.8398945927619934,
|
| 1697 |
+
"learning_rate": 4.2949691085613417e-05,
|
| 1698 |
+
"loss": 2.9484,
|
| 1699 |
+
"step": 226000
|
| 1700 |
+
},
|
| 1701 |
+
{
|
| 1702 |
+
"epoch": 12.21,
|
| 1703 |
+
"grad_norm": 0.8549145460128784,
|
| 1704 |
+
"learning_rate": 4.265548690791409e-05,
|
| 1705 |
+
"loss": 2.9443,
|
| 1706 |
+
"step": 227000
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 12.26,
|
| 1710 |
+
"grad_norm": 0.8680046796798706,
|
| 1711 |
+
"learning_rate": 4.2361282730214765e-05,
|
| 1712 |
+
"loss": 2.9485,
|
| 1713 |
+
"step": 228000
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 12.32,
|
| 1717 |
+
"grad_norm": 0.8534417748451233,
|
| 1718 |
+
"learning_rate": 4.206737275669315e-05,
|
| 1719 |
+
"loss": 2.9479,
|
| 1720 |
+
"step": 229000
|
| 1721 |
+
},
|
| 1722 |
+
{
|
| 1723 |
+
"epoch": 12.37,
|
| 1724 |
+
"grad_norm": 0.8083067536354065,
|
| 1725 |
+
"learning_rate": 4.1773168578993824e-05,
|
| 1726 |
+
"loss": 2.9497,
|
| 1727 |
+
"step": 230000
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 12.42,
|
| 1731 |
+
"grad_norm": 0.810694694519043,
|
| 1732 |
+
"learning_rate": 4.14789644012945e-05,
|
| 1733 |
+
"loss": 2.959,
|
| 1734 |
+
"step": 231000
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 12.48,
|
| 1738 |
+
"grad_norm": 0.8079540133476257,
|
| 1739 |
+
"learning_rate": 4.1185054427772876e-05,
|
| 1740 |
+
"loss": 2.9501,
|
| 1741 |
+
"step": 232000
|
| 1742 |
+
},
|
| 1743 |
+
{
|
| 1744 |
+
"epoch": 12.53,
|
| 1745 |
+
"grad_norm": 0.8427674174308777,
|
| 1746 |
+
"learning_rate": 4.089085025007355e-05,
|
| 1747 |
+
"loss": 2.9553,
|
| 1748 |
+
"step": 233000
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 12.58,
|
| 1752 |
+
"grad_norm": 0.8193045854568481,
|
| 1753 |
+
"learning_rate": 4.0596646072374225e-05,
|
| 1754 |
+
"loss": 2.9604,
|
| 1755 |
+
"step": 234000
|
| 1756 |
+
},
|
| 1757 |
+
{
|
| 1758 |
+
"epoch": 12.64,
|
| 1759 |
+
"grad_norm": 0.8305484652519226,
|
| 1760 |
+
"learning_rate": 4.030273609885261e-05,
|
| 1761 |
+
"loss": 2.9623,
|
| 1762 |
+
"step": 235000
|
| 1763 |
+
},
|
| 1764 |
+
{
|
| 1765 |
+
"epoch": 12.69,
|
| 1766 |
+
"grad_norm": 0.8170352578163147,
|
| 1767 |
+
"learning_rate": 4.0008826125330986e-05,
|
| 1768 |
+
"loss": 2.9616,
|
| 1769 |
+
"step": 236000
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 12.75,
|
| 1773 |
+
"grad_norm": 0.8222061395645142,
|
| 1774 |
+
"learning_rate": 3.971462194763166e-05,
|
| 1775 |
+
"loss": 2.9659,
|
| 1776 |
+
"step": 237000
|
| 1777 |
+
},
|
| 1778 |
+
{
|
| 1779 |
+
"epoch": 12.8,
|
| 1780 |
+
"grad_norm": 0.8384312987327576,
|
| 1781 |
+
"learning_rate": 3.9420417769932335e-05,
|
| 1782 |
+
"loss": 2.9645,
|
| 1783 |
+
"step": 238000
|
| 1784 |
+
},
|
| 1785 |
+
{
|
| 1786 |
+
"epoch": 12.85,
|
| 1787 |
+
"grad_norm": 0.8427971005439758,
|
| 1788 |
+
"learning_rate": 3.9126213592233016e-05,
|
| 1789 |
+
"loss": 2.9628,
|
| 1790 |
+
"step": 239000
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 12.91,
|
| 1794 |
+
"grad_norm": 0.8048679232597351,
|
| 1795 |
+
"learning_rate": 3.883200941453369e-05,
|
| 1796 |
+
"loss": 2.9675,
|
| 1797 |
+
"step": 240000
|
| 1798 |
+
},
|
| 1799 |
+
{
|
| 1800 |
+
"epoch": 12.96,
|
| 1801 |
+
"grad_norm": 0.8325705528259277,
|
| 1802 |
+
"learning_rate": 3.853809944101207e-05,
|
| 1803 |
+
"loss": 2.9671,
|
| 1804 |
+
"step": 241000
|
| 1805 |
+
},
|
| 1806 |
+
{
|
| 1807 |
+
"epoch": 13.0,
|
| 1808 |
+
"eval_accuracy": 0.4061510186590524,
|
| 1809 |
+
"eval_loss": 3.414961576461792,
|
| 1810 |
+
"eval_runtime": 154.7616,
|
| 1811 |
+
"eval_samples_per_second": 374.26,
|
| 1812 |
+
"eval_steps_per_second": 5.854,
|
| 1813 |
+
"step": 241735
|
| 1814 |
+
},
|
| 1815 |
+
{
|
| 1816 |
+
"epoch": 13.01,
|
| 1817 |
+
"grad_norm": 0.856779158115387,
|
| 1818 |
+
"learning_rate": 3.824418946749044e-05,
|
| 1819 |
+
"loss": 2.9534,
|
| 1820 |
+
"step": 242000
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 13.07,
|
| 1824 |
+
"grad_norm": 0.8315408229827881,
|
| 1825 |
+
"learning_rate": 3.794998528979112e-05,
|
| 1826 |
+
"loss": 2.9192,
|
| 1827 |
+
"step": 243000
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 13.12,
|
| 1831 |
+
"grad_norm": 0.8712647557258606,
|
| 1832 |
+
"learning_rate": 3.7655781112091794e-05,
|
| 1833 |
+
"loss": 2.9236,
|
| 1834 |
+
"step": 244000
|
| 1835 |
+
},
|
| 1836 |
+
{
|
| 1837 |
+
"epoch": 13.18,
|
| 1838 |
+
"grad_norm": 0.8575468063354492,
|
| 1839 |
+
"learning_rate": 3.7361576934392476e-05,
|
| 1840 |
+
"loss": 2.9209,
|
| 1841 |
+
"step": 245000
|
| 1842 |
+
},
|
| 1843 |
+
{
|
| 1844 |
+
"epoch": 13.23,
|
| 1845 |
+
"grad_norm": 0.848013699054718,
|
| 1846 |
+
"learning_rate": 3.7067666960870846e-05,
|
| 1847 |
+
"loss": 2.9224,
|
| 1848 |
+
"step": 246000
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 13.28,
|
| 1852 |
+
"grad_norm": 0.8788365721702576,
|
| 1853 |
+
"learning_rate": 3.677346278317153e-05,
|
| 1854 |
+
"loss": 2.9313,
|
| 1855 |
+
"step": 247000
|
| 1856 |
+
},
|
| 1857 |
+
{
|
| 1858 |
+
"epoch": 13.34,
|
| 1859 |
+
"grad_norm": 0.8671311140060425,
|
| 1860 |
+
"learning_rate": 3.64792586054722e-05,
|
| 1861 |
+
"loss": 2.9354,
|
| 1862 |
+
"step": 248000
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 13.39,
|
| 1866 |
+
"grad_norm": 0.8452686667442322,
|
| 1867 |
+
"learning_rate": 3.618534863195058e-05,
|
| 1868 |
+
"loss": 2.9335,
|
| 1869 |
+
"step": 249000
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 13.44,
|
| 1873 |
+
"grad_norm": 0.8493407368659973,
|
| 1874 |
+
"learning_rate": 3.5891144454251254e-05,
|
| 1875 |
+
"loss": 2.9352,
|
| 1876 |
+
"step": 250000
|
| 1877 |
+
},
|
| 1878 |
+
{
|
| 1879 |
+
"epoch": 13.5,
|
| 1880 |
+
"grad_norm": 0.8630385398864746,
|
| 1881 |
+
"learning_rate": 3.559694027655193e-05,
|
| 1882 |
+
"loss": 2.9366,
|
| 1883 |
+
"step": 251000
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 13.55,
|
| 1887 |
+
"grad_norm": 0.8659392595291138,
|
| 1888 |
+
"learning_rate": 3.5303030303030305e-05,
|
| 1889 |
+
"loss": 2.9396,
|
| 1890 |
+
"step": 252000
|
| 1891 |
+
},
|
| 1892 |
+
{
|
| 1893 |
+
"epoch": 13.61,
|
| 1894 |
+
"grad_norm": 0.8576540350914001,
|
| 1895 |
+
"learning_rate": 3.500912032950868e-05,
|
| 1896 |
+
"loss": 2.9371,
|
| 1897 |
+
"step": 253000
|
| 1898 |
+
},
|
| 1899 |
+
{
|
| 1900 |
+
"epoch": 13.66,
|
| 1901 |
+
"grad_norm": 0.839272141456604,
|
| 1902 |
+
"learning_rate": 3.471491615180936e-05,
|
| 1903 |
+
"loss": 2.9399,
|
| 1904 |
+
"step": 254000
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 13.71,
|
| 1908 |
+
"grad_norm": 0.8816892504692078,
|
| 1909 |
+
"learning_rate": 3.4421006178287734e-05,
|
| 1910 |
+
"loss": 2.9414,
|
| 1911 |
+
"step": 255000
|
| 1912 |
+
},
|
| 1913 |
+
{
|
| 1914 |
+
"epoch": 13.77,
|
| 1915 |
+
"grad_norm": 0.8923898339271545,
|
| 1916 |
+
"learning_rate": 3.412680200058841e-05,
|
| 1917 |
+
"loss": 2.9429,
|
| 1918 |
+
"step": 256000
|
| 1919 |
+
},
|
| 1920 |
+
{
|
| 1921 |
+
"epoch": 13.82,
|
| 1922 |
+
"grad_norm": 0.84392249584198,
|
| 1923 |
+
"learning_rate": 3.383259782288909e-05,
|
| 1924 |
+
"loss": 2.9459,
|
| 1925 |
+
"step": 257000
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"epoch": 13.87,
|
| 1929 |
+
"grad_norm": 0.8665987849235535,
|
| 1930 |
+
"learning_rate": 3.3538393645189765e-05,
|
| 1931 |
+
"loss": 2.9467,
|
| 1932 |
+
"step": 258000
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"epoch": 13.93,
|
| 1936 |
+
"grad_norm": 0.8305437564849854,
|
| 1937 |
+
"learning_rate": 3.324448367166814e-05,
|
| 1938 |
+
"loss": 2.9475,
|
| 1939 |
+
"step": 259000
|
| 1940 |
+
},
|
| 1941 |
+
{
|
| 1942 |
+
"epoch": 13.98,
|
| 1943 |
+
"grad_norm": 0.8880412578582764,
|
| 1944 |
+
"learning_rate": 3.2950279493968816e-05,
|
| 1945 |
+
"loss": 2.9477,
|
| 1946 |
+
"step": 260000
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 14.0,
|
| 1950 |
+
"eval_accuracy": 0.4062352605994869,
|
| 1951 |
+
"eval_loss": 3.417443037033081,
|
| 1952 |
+
"eval_runtime": 154.6446,
|
| 1953 |
+
"eval_samples_per_second": 374.543,
|
| 1954 |
+
"eval_steps_per_second": 5.859,
|
| 1955 |
+
"step": 260330
|
| 1956 |
+
},
|
| 1957 |
+
{
|
| 1958 |
+
"epoch": 14.04,
|
| 1959 |
+
"grad_norm": 0.8674964308738708,
|
| 1960 |
+
"learning_rate": 3.265607531626949e-05,
|
| 1961 |
+
"loss": 2.9136,
|
| 1962 |
+
"step": 261000
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 14.09,
|
| 1966 |
+
"grad_norm": 0.8518526554107666,
|
| 1967 |
+
"learning_rate": 3.236216534274787e-05,
|
| 1968 |
+
"loss": 2.9045,
|
| 1969 |
+
"step": 262000
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 14.14,
|
| 1973 |
+
"grad_norm": 0.8985535502433777,
|
| 1974 |
+
"learning_rate": 3.206796116504854e-05,
|
| 1975 |
+
"loss": 2.9094,
|
| 1976 |
+
"step": 263000
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 14.2,
|
| 1980 |
+
"grad_norm": 0.9025124311447144,
|
| 1981 |
+
"learning_rate": 3.177405119152692e-05,
|
| 1982 |
+
"loss": 2.9072,
|
| 1983 |
+
"step": 264000
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 14.25,
|
| 1987 |
+
"grad_norm": 0.866775631904602,
|
| 1988 |
+
"learning_rate": 3.1479847013827594e-05,
|
| 1989 |
+
"loss": 2.912,
|
| 1990 |
+
"step": 265000
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 14.3,
|
| 1994 |
+
"grad_norm": 0.875120222568512,
|
| 1995 |
+
"learning_rate": 3.1185642836128276e-05,
|
| 1996 |
+
"loss": 2.9127,
|
| 1997 |
+
"step": 266000
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 14.36,
|
| 2001 |
+
"grad_norm": 0.8747627139091492,
|
| 2002 |
+
"learning_rate": 3.089143865842895e-05,
|
| 2003 |
+
"loss": 2.9133,
|
| 2004 |
+
"step": 267000
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 14.41,
|
| 2008 |
+
"grad_norm": 0.8658396005630493,
|
| 2009 |
+
"learning_rate": 3.059752868490733e-05,
|
| 2010 |
+
"loss": 2.9134,
|
| 2011 |
+
"step": 268000
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 14.47,
|
| 2015 |
+
"grad_norm": 0.8496713638305664,
|
| 2016 |
+
"learning_rate": 3.0303324507208002e-05,
|
| 2017 |
+
"loss": 2.9196,
|
| 2018 |
+
"step": 269000
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 14.52,
|
| 2022 |
+
"grad_norm": 0.8578941822052002,
|
| 2023 |
+
"learning_rate": 3.0009120329508683e-05,
|
| 2024 |
+
"loss": 2.92,
|
| 2025 |
+
"step": 270000
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 14.57,
|
| 2029 |
+
"grad_norm": 0.8853529691696167,
|
| 2030 |
+
"learning_rate": 2.9715504560164753e-05,
|
| 2031 |
+
"loss": 2.9222,
|
| 2032 |
+
"step": 271000
|
| 2033 |
+
},
|
| 2034 |
+
{
|
| 2035 |
+
"epoch": 14.63,
|
| 2036 |
+
"grad_norm": 0.8526709079742432,
|
| 2037 |
+
"learning_rate": 2.942159458664313e-05,
|
| 2038 |
+
"loss": 2.9165,
|
| 2039 |
+
"step": 272000
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 14.68,
|
| 2043 |
+
"grad_norm": 0.8756287693977356,
|
| 2044 |
+
"learning_rate": 2.912739040894381e-05,
|
| 2045 |
+
"loss": 2.926,
|
| 2046 |
+
"step": 273000
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 14.74,
|
| 2050 |
+
"grad_norm": 0.8718807697296143,
|
| 2051 |
+
"learning_rate": 2.8833186231244486e-05,
|
| 2052 |
+
"loss": 2.924,
|
| 2053 |
+
"step": 274000
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 14.79,
|
| 2057 |
+
"grad_norm": 0.874072790145874,
|
| 2058 |
+
"learning_rate": 2.853898205354516e-05,
|
| 2059 |
+
"loss": 2.9244,
|
| 2060 |
+
"step": 275000
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 14.84,
|
| 2064 |
+
"grad_norm": 0.8762756586074829,
|
| 2065 |
+
"learning_rate": 2.824477787584584e-05,
|
| 2066 |
+
"loss": 2.9211,
|
| 2067 |
+
"step": 276000
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 14.9,
|
| 2071 |
+
"grad_norm": 0.9252573251724243,
|
| 2072 |
+
"learning_rate": 2.7950867902324212e-05,
|
| 2073 |
+
"loss": 2.9225,
|
| 2074 |
+
"step": 277000
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 14.95,
|
| 2078 |
+
"grad_norm": 0.8735837936401367,
|
| 2079 |
+
"learning_rate": 2.7656663724624893e-05,
|
| 2080 |
+
"loss": 2.9272,
|
| 2081 |
+
"step": 278000
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 15.0,
|
| 2085 |
+
"eval_accuracy": 0.4066799828049704,
|
| 2086 |
+
"eval_loss": 3.427541494369507,
|
| 2087 |
+
"eval_runtime": 155.041,
|
| 2088 |
+
"eval_samples_per_second": 373.585,
|
| 2089 |
+
"eval_steps_per_second": 5.844,
|
| 2090 |
+
"step": 278925
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 15.0,
|
| 2094 |
+
"grad_norm": 0.8828846216201782,
|
| 2095 |
+
"learning_rate": 2.7362459546925568e-05,
|
| 2096 |
+
"loss": 2.9242,
|
| 2097 |
+
"step": 279000
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 15.06,
|
| 2101 |
+
"grad_norm": 0.8483282923698425,
|
| 2102 |
+
"learning_rate": 2.7068255369226242e-05,
|
| 2103 |
+
"loss": 2.8881,
|
| 2104 |
+
"step": 280000
|
| 2105 |
+
},
|
| 2106 |
+
{
|
| 2107 |
+
"epoch": 15.11,
|
| 2108 |
+
"grad_norm": 0.8667340278625488,
|
| 2109 |
+
"learning_rate": 2.6774051191526924e-05,
|
| 2110 |
+
"loss": 2.8852,
|
| 2111 |
+
"step": 281000
|
| 2112 |
+
},
|
| 2113 |
+
{
|
| 2114 |
+
"epoch": 15.17,
|
| 2115 |
+
"grad_norm": 0.8646563291549683,
|
| 2116 |
+
"learning_rate": 2.6480435422182997e-05,
|
| 2117 |
+
"loss": 2.8886,
|
| 2118 |
+
"step": 282000
|
| 2119 |
+
},
|
| 2120 |
+
{
|
| 2121 |
+
"epoch": 15.22,
|
| 2122 |
+
"grad_norm": 0.8643233776092529,
|
| 2123 |
+
"learning_rate": 2.618623124448367e-05,
|
| 2124 |
+
"loss": 2.8935,
|
| 2125 |
+
"step": 283000
|
| 2126 |
+
},
|
| 2127 |
+
{
|
| 2128 |
+
"epoch": 15.27,
|
| 2129 |
+
"grad_norm": 0.9026147127151489,
|
| 2130 |
+
"learning_rate": 2.5892027066784353e-05,
|
| 2131 |
+
"loss": 2.8939,
|
| 2132 |
+
"step": 284000
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 15.33,
|
| 2136 |
+
"grad_norm": 0.8915610313415527,
|
| 2137 |
+
"learning_rate": 2.5597822889085027e-05,
|
| 2138 |
+
"loss": 2.8964,
|
| 2139 |
+
"step": 285000
|
| 2140 |
+
},
|
| 2141 |
+
{
|
| 2142 |
+
"epoch": 15.38,
|
| 2143 |
+
"grad_norm": 0.8528517484664917,
|
| 2144 |
+
"learning_rate": 2.53036187113857e-05,
|
| 2145 |
+
"loss": 2.8976,
|
| 2146 |
+
"step": 286000
|
| 2147 |
+
},
|
| 2148 |
+
{
|
| 2149 |
+
"epoch": 15.43,
|
| 2150 |
+
"grad_norm": 0.8766651749610901,
|
| 2151 |
+
"learning_rate": 2.500970873786408e-05,
|
| 2152 |
+
"loss": 2.9009,
|
| 2153 |
+
"step": 287000
|
| 2154 |
+
},
|
| 2155 |
+
{
|
| 2156 |
+
"epoch": 15.49,
|
| 2157 |
+
"grad_norm": 0.897165060043335,
|
| 2158 |
+
"learning_rate": 2.4715504560164757e-05,
|
| 2159 |
+
"loss": 2.9008,
|
| 2160 |
+
"step": 288000
|
| 2161 |
+
},
|
| 2162 |
+
{
|
| 2163 |
+
"epoch": 15.54,
|
| 2164 |
+
"grad_norm": 0.8595399260520935,
|
| 2165 |
+
"learning_rate": 2.442159458664313e-05,
|
| 2166 |
+
"loss": 2.9063,
|
| 2167 |
+
"step": 289000
|
| 2168 |
+
},
|
| 2169 |
+
{
|
| 2170 |
+
"epoch": 15.6,
|
| 2171 |
+
"grad_norm": 0.8562242984771729,
|
| 2172 |
+
"learning_rate": 2.412739040894381e-05,
|
| 2173 |
+
"loss": 2.8988,
|
| 2174 |
+
"step": 290000
|
| 2175 |
+
},
|
| 2176 |
+
{
|
| 2177 |
+
"epoch": 15.65,
|
| 2178 |
+
"grad_norm": 0.8943966627120972,
|
| 2179 |
+
"learning_rate": 2.3833186231244483e-05,
|
| 2180 |
+
"loss": 2.905,
|
| 2181 |
+
"step": 291000
|
| 2182 |
+
},
|
| 2183 |
+
{
|
| 2184 |
+
"epoch": 15.7,
|
| 2185 |
+
"grad_norm": 0.8769930601119995,
|
| 2186 |
+
"learning_rate": 2.353927625772286e-05,
|
| 2187 |
+
"loss": 2.9032,
|
| 2188 |
+
"step": 292000
|
| 2189 |
+
},
|
| 2190 |
+
{
|
| 2191 |
+
"epoch": 15.76,
|
| 2192 |
+
"grad_norm": 0.8928766250610352,
|
| 2193 |
+
"learning_rate": 2.3245072080023535e-05,
|
| 2194 |
+
"loss": 2.907,
|
| 2195 |
+
"step": 293000
|
| 2196 |
+
},
|
| 2197 |
+
{
|
| 2198 |
+
"epoch": 15.81,
|
| 2199 |
+
"grad_norm": 0.8999136686325073,
|
| 2200 |
+
"learning_rate": 2.2950867902324213e-05,
|
| 2201 |
+
"loss": 2.9099,
|
| 2202 |
+
"step": 294000
|
| 2203 |
+
},
|
| 2204 |
+
{
|
| 2205 |
+
"epoch": 15.86,
|
| 2206 |
+
"grad_norm": 0.8893632888793945,
|
| 2207 |
+
"learning_rate": 2.265695792880259e-05,
|
| 2208 |
+
"loss": 2.9039,
|
| 2209 |
+
"step": 295000
|
| 2210 |
+
},
|
| 2211 |
+
{
|
| 2212 |
+
"epoch": 15.92,
|
| 2213 |
+
"grad_norm": 0.9111586213111877,
|
| 2214 |
+
"learning_rate": 2.2362753751103264e-05,
|
| 2215 |
+
"loss": 2.9071,
|
| 2216 |
+
"step": 296000
|
| 2217 |
+
},
|
| 2218 |
+
{
|
| 2219 |
+
"epoch": 15.97,
|
| 2220 |
+
"grad_norm": 0.8878467082977295,
|
| 2221 |
+
"learning_rate": 2.2068549573403942e-05,
|
| 2222 |
+
"loss": 2.9088,
|
| 2223 |
+
"step": 297000
|
| 2224 |
+
},
|
| 2225 |
+
{
|
| 2226 |
+
"epoch": 16.0,
|
| 2227 |
+
"eval_accuracy": 0.4068374493985737,
|
| 2228 |
+
"eval_loss": 3.427147626876831,
|
| 2229 |
+
"eval_runtime": 154.7655,
|
| 2230 |
+
"eval_samples_per_second": 374.25,
|
| 2231 |
+
"eval_steps_per_second": 5.854,
|
| 2232 |
+
"step": 297520
|
| 2233 |
+
},
|
| 2234 |
+
{
|
| 2235 |
+
"epoch": 16.03,
|
| 2236 |
+
"grad_norm": 0.9069207310676575,
|
| 2237 |
+
"learning_rate": 2.177434539570462e-05,
|
| 2238 |
+
"loss": 2.8909,
|
| 2239 |
+
"step": 298000
|
| 2240 |
+
},
|
| 2241 |
+
{
|
| 2242 |
+
"epoch": 16.08,
|
| 2243 |
+
"grad_norm": 0.8874634504318237,
|
| 2244 |
+
"learning_rate": 2.1480435422182994e-05,
|
| 2245 |
+
"loss": 2.8723,
|
| 2246 |
+
"step": 299000
|
| 2247 |
+
},
|
| 2248 |
+
{
|
| 2249 |
+
"epoch": 16.13,
|
| 2250 |
+
"grad_norm": 0.8833279609680176,
|
| 2251 |
+
"learning_rate": 2.1186231244483672e-05,
|
| 2252 |
+
"loss": 2.876,
|
| 2253 |
+
"step": 300000
|
| 2254 |
+
},
|
| 2255 |
+
{
|
| 2256 |
+
"epoch": 16.19,
|
| 2257 |
+
"grad_norm": 0.9021320939064026,
|
| 2258 |
+
"learning_rate": 2.089202706678435e-05,
|
| 2259 |
+
"loss": 2.8793,
|
| 2260 |
+
"step": 301000
|
| 2261 |
+
},
|
| 2262 |
+
{
|
| 2263 |
+
"epoch": 16.24,
|
| 2264 |
+
"grad_norm": 0.8948860764503479,
|
| 2265 |
+
"learning_rate": 2.0598117093262727e-05,
|
| 2266 |
+
"loss": 2.8789,
|
| 2267 |
+
"step": 302000
|
| 2268 |
+
},
|
| 2269 |
+
{
|
| 2270 |
+
"epoch": 16.29,
|
| 2271 |
+
"grad_norm": 0.8972275257110596,
|
| 2272 |
+
"learning_rate": 2.03039129155634e-05,
|
| 2273 |
+
"loss": 2.8778,
|
| 2274 |
+
"step": 303000
|
| 2275 |
+
},
|
| 2276 |
+
{
|
| 2277 |
+
"epoch": 16.35,
|
| 2278 |
+
"grad_norm": 0.8987729549407959,
|
| 2279 |
+
"learning_rate": 2.0009708737864076e-05,
|
| 2280 |
+
"loss": 2.8827,
|
| 2281 |
+
"step": 304000
|
| 2282 |
+
},
|
| 2283 |
+
{
|
| 2284 |
+
"epoch": 16.4,
|
| 2285 |
+
"grad_norm": 0.8773179650306702,
|
| 2286 |
+
"learning_rate": 1.9715798764342457e-05,
|
| 2287 |
+
"loss": 2.8866,
|
| 2288 |
+
"step": 305000
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 16.46,
|
| 2292 |
+
"grad_norm": 0.9324170351028442,
|
| 2293 |
+
"learning_rate": 1.942159458664313e-05,
|
| 2294 |
+
"loss": 2.8842,
|
| 2295 |
+
"step": 306000
|
| 2296 |
+
},
|
| 2297 |
+
{
|
| 2298 |
+
"epoch": 16.51,
|
| 2299 |
+
"grad_norm": 0.9289742708206177,
|
| 2300 |
+
"learning_rate": 1.9127390408943806e-05,
|
| 2301 |
+
"loss": 2.8813,
|
| 2302 |
+
"step": 307000
|
| 2303 |
+
},
|
| 2304 |
+
{
|
| 2305 |
+
"epoch": 16.56,
|
| 2306 |
+
"grad_norm": 0.9093849062919617,
|
| 2307 |
+
"learning_rate": 1.8833480435422186e-05,
|
| 2308 |
+
"loss": 2.8824,
|
| 2309 |
+
"step": 308000
|
| 2310 |
+
},
|
| 2311 |
+
{
|
| 2312 |
+
"epoch": 16.62,
|
| 2313 |
+
"grad_norm": 0.9236767888069153,
|
| 2314 |
+
"learning_rate": 1.853927625772286e-05,
|
| 2315 |
+
"loss": 2.8845,
|
| 2316 |
+
"step": 309000
|
| 2317 |
+
},
|
| 2318 |
+
{
|
| 2319 |
+
"epoch": 16.67,
|
| 2320 |
+
"grad_norm": 0.8720325827598572,
|
| 2321 |
+
"learning_rate": 1.8245366284201238e-05,
|
| 2322 |
+
"loss": 2.8859,
|
| 2323 |
+
"step": 310000
|
| 2324 |
+
},
|
| 2325 |
+
{
|
| 2326 |
+
"epoch": 16.72,
|
| 2327 |
+
"grad_norm": 0.9025481939315796,
|
| 2328 |
+
"learning_rate": 1.7951456310679612e-05,
|
| 2329 |
+
"loss": 2.8921,
|
| 2330 |
+
"step": 311000
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 16.78,
|
| 2334 |
+
"grad_norm": 0.9536554217338562,
|
| 2335 |
+
"learning_rate": 1.765725213298029e-05,
|
| 2336 |
+
"loss": 2.8898,
|
| 2337 |
+
"step": 312000
|
| 2338 |
+
},
|
| 2339 |
+
{
|
| 2340 |
+
"epoch": 16.83,
|
| 2341 |
+
"grad_norm": 0.909812867641449,
|
| 2342 |
+
"learning_rate": 1.7363047955280968e-05,
|
| 2343 |
+
"loss": 2.8901,
|
| 2344 |
+
"step": 313000
|
| 2345 |
+
},
|
| 2346 |
+
{
|
| 2347 |
+
"epoch": 16.89,
|
| 2348 |
+
"grad_norm": 0.9195666313171387,
|
| 2349 |
+
"learning_rate": 1.7068843777581642e-05,
|
| 2350 |
+
"loss": 2.8897,
|
| 2351 |
+
"step": 314000
|
| 2352 |
+
},
|
| 2353 |
+
{
|
| 2354 |
+
"epoch": 16.94,
|
| 2355 |
+
"grad_norm": 0.8690800666809082,
|
| 2356 |
+
"learning_rate": 1.677463959988232e-05,
|
| 2357 |
+
"loss": 2.888,
|
| 2358 |
+
"step": 315000
|
| 2359 |
+
},
|
| 2360 |
+
{
|
| 2361 |
+
"epoch": 16.99,
|
| 2362 |
+
"grad_norm": 0.8886986374855042,
|
| 2363 |
+
"learning_rate": 1.6480435422182998e-05,
|
| 2364 |
+
"loss": 2.8915,
|
| 2365 |
+
"step": 316000
|
| 2366 |
+
},
|
| 2367 |
+
{
|
| 2368 |
+
"epoch": 17.0,
|
| 2369 |
+
"eval_accuracy": 0.4071312885113811,
|
| 2370 |
+
"eval_loss": 3.424515962600708,
|
| 2371 |
+
"eval_runtime": 155.1079,
|
| 2372 |
+
"eval_samples_per_second": 373.424,
|
| 2373 |
+
"eval_steps_per_second": 5.841,
|
| 2374 |
+
"step": 316115
|
| 2375 |
+
},
|
| 2376 |
+
{
|
| 2377 |
+
"epoch": 17.05,
|
| 2378 |
+
"grad_norm": 0.9326255321502686,
|
| 2379 |
+
"learning_rate": 1.6186525448661372e-05,
|
| 2380 |
+
"loss": 2.8626,
|
| 2381 |
+
"step": 317000
|
| 2382 |
+
},
|
| 2383 |
+
{
|
| 2384 |
+
"epoch": 17.1,
|
| 2385 |
+
"grad_norm": 0.9530662298202515,
|
| 2386 |
+
"learning_rate": 1.5892615475139746e-05,
|
| 2387 |
+
"loss": 2.8605,
|
| 2388 |
+
"step": 318000
|
| 2389 |
+
},
|
| 2390 |
+
{
|
| 2391 |
+
"epoch": 17.16,
|
| 2392 |
+
"grad_norm": 0.9299532771110535,
|
| 2393 |
+
"learning_rate": 1.5598411297440423e-05,
|
| 2394 |
+
"loss": 2.8598,
|
| 2395 |
+
"step": 319000
|
| 2396 |
+
},
|
| 2397 |
+
{
|
| 2398 |
+
"epoch": 17.21,
|
| 2399 |
+
"grad_norm": 0.9452057480812073,
|
| 2400 |
+
"learning_rate": 1.53042071197411e-05,
|
| 2401 |
+
"loss": 2.8637,
|
| 2402 |
+
"step": 320000
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 17.26,
|
| 2406 |
+
"grad_norm": 0.9302668571472168,
|
| 2407 |
+
"learning_rate": 1.5010591350397176e-05,
|
| 2408 |
+
"loss": 2.8636,
|
| 2409 |
+
"step": 321000
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 17.32,
|
| 2413 |
+
"grad_norm": 0.9411625266075134,
|
| 2414 |
+
"learning_rate": 1.4716387172697854e-05,
|
| 2415 |
+
"loss": 2.8669,
|
| 2416 |
+
"step": 322000
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 17.37,
|
| 2420 |
+
"grad_norm": 0.9255250096321106,
|
| 2421 |
+
"learning_rate": 1.4422182994998529e-05,
|
| 2422 |
+
"loss": 2.8643,
|
| 2423 |
+
"step": 323000
|
| 2424 |
+
},
|
| 2425 |
+
{
|
| 2426 |
+
"epoch": 17.42,
|
| 2427 |
+
"grad_norm": 0.9399036765098572,
|
| 2428 |
+
"learning_rate": 1.4127978817299207e-05,
|
| 2429 |
+
"loss": 2.8686,
|
| 2430 |
+
"step": 324000
|
| 2431 |
+
},
|
| 2432 |
+
{
|
| 2433 |
+
"epoch": 17.48,
|
| 2434 |
+
"grad_norm": 0.9294721484184265,
|
| 2435 |
+
"learning_rate": 1.3833774639599883e-05,
|
| 2436 |
+
"loss": 2.8693,
|
| 2437 |
+
"step": 325000
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 17.53,
|
| 2441 |
+
"grad_norm": 0.9378616213798523,
|
| 2442 |
+
"learning_rate": 1.3539864666078258e-05,
|
| 2443 |
+
"loss": 2.8689,
|
| 2444 |
+
"step": 326000
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 17.59,
|
| 2448 |
+
"grad_norm": 0.9260066747665405,
|
| 2449 |
+
"learning_rate": 1.3245954692556634e-05,
|
| 2450 |
+
"loss": 2.8697,
|
| 2451 |
+
"step": 327000
|
| 2452 |
+
},
|
| 2453 |
+
{
|
| 2454 |
+
"epoch": 17.64,
|
| 2455 |
+
"grad_norm": 0.9042669534683228,
|
| 2456 |
+
"learning_rate": 1.2951750514857312e-05,
|
| 2457 |
+
"loss": 2.8717,
|
| 2458 |
+
"step": 328000
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 17.69,
|
| 2462 |
+
"grad_norm": 0.9497202634811401,
|
| 2463 |
+
"learning_rate": 1.2657546337157988e-05,
|
| 2464 |
+
"loss": 2.8715,
|
| 2465 |
+
"step": 329000
|
| 2466 |
+
},
|
| 2467 |
+
{
|
| 2468 |
+
"epoch": 17.75,
|
| 2469 |
+
"grad_norm": 0.9359462857246399,
|
| 2470 |
+
"learning_rate": 1.2363342159458664e-05,
|
| 2471 |
+
"loss": 2.8724,
|
| 2472 |
+
"step": 330000
|
| 2473 |
+
},
|
| 2474 |
+
{
|
| 2475 |
+
"epoch": 17.8,
|
| 2476 |
+
"grad_norm": 0.9586021900177002,
|
| 2477 |
+
"learning_rate": 1.2069432185937041e-05,
|
| 2478 |
+
"loss": 2.8724,
|
| 2479 |
+
"step": 331000
|
| 2480 |
+
},
|
| 2481 |
+
{
|
| 2482 |
+
"epoch": 17.85,
|
| 2483 |
+
"grad_norm": 0.92441725730896,
|
| 2484 |
+
"learning_rate": 1.1775228008237718e-05,
|
| 2485 |
+
"loss": 2.8739,
|
| 2486 |
+
"step": 332000
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 17.91,
|
| 2490 |
+
"grad_norm": 0.9491459727287292,
|
| 2491 |
+
"learning_rate": 1.1481023830538394e-05,
|
| 2492 |
+
"loss": 2.8785,
|
| 2493 |
+
"step": 333000
|
| 2494 |
+
},
|
| 2495 |
+
{
|
| 2496 |
+
"epoch": 17.96,
|
| 2497 |
+
"grad_norm": 0.9118024110794067,
|
| 2498 |
+
"learning_rate": 1.1187113857016771e-05,
|
| 2499 |
+
"loss": 2.872,
|
| 2500 |
+
"step": 334000
|
| 2501 |
+
},
|
| 2502 |
+
{
|
| 2503 |
+
"epoch": 18.0,
|
| 2504 |
+
"eval_accuracy": 0.4070309908901142,
|
| 2505 |
+
"eval_loss": 3.426220417022705,
|
| 2506 |
+
"eval_runtime": 155.4694,
|
| 2507 |
+
"eval_samples_per_second": 372.556,
|
| 2508 |
+
"eval_steps_per_second": 5.828,
|
| 2509 |
+
"step": 334710
|
| 2510 |
+
},
|
| 2511 |
+
{
|
| 2512 |
+
"epoch": 18.02,
|
| 2513 |
+
"grad_norm": 0.9377329349517822,
|
| 2514 |
+
"learning_rate": 1.0892909679317447e-05,
|
| 2515 |
+
"loss": 2.8661,
|
| 2516 |
+
"step": 335000
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 18.07,
|
| 2520 |
+
"grad_norm": 0.9395567178726196,
|
| 2521 |
+
"learning_rate": 1.0598705501618123e-05,
|
| 2522 |
+
"loss": 2.8488,
|
| 2523 |
+
"step": 336000
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 18.12,
|
| 2527 |
+
"grad_norm": 0.9508826732635498,
|
| 2528 |
+
"learning_rate": 1.0304795528096499e-05,
|
| 2529 |
+
"loss": 2.8486,
|
| 2530 |
+
"step": 337000
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"epoch": 18.18,
|
| 2534 |
+
"grad_norm": 0.9117801785469055,
|
| 2535 |
+
"learning_rate": 1.0010591350397177e-05,
|
| 2536 |
+
"loss": 2.8487,
|
| 2537 |
+
"step": 338000
|
| 2538 |
+
},
|
| 2539 |
+
{
|
| 2540 |
+
"epoch": 18.23,
|
| 2541 |
+
"grad_norm": 0.932178795337677,
|
| 2542 |
+
"learning_rate": 9.716681376875552e-06,
|
| 2543 |
+
"loss": 2.8528,
|
| 2544 |
+
"step": 339000
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"epoch": 18.28,
|
| 2548 |
+
"grad_norm": 0.9204062223434448,
|
| 2549 |
+
"learning_rate": 9.422477199176229e-06,
|
| 2550 |
+
"loss": 2.8579,
|
| 2551 |
+
"step": 340000
|
| 2552 |
+
},
|
| 2553 |
+
{
|
| 2554 |
+
"epoch": 18.34,
|
| 2555 |
+
"grad_norm": 0.9516026377677917,
|
| 2556 |
+
"learning_rate": 9.128273021476906e-06,
|
| 2557 |
+
"loss": 2.8507,
|
| 2558 |
+
"step": 341000
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 18.39,
|
| 2562 |
+
"grad_norm": 0.9432997703552246,
|
| 2563 |
+
"learning_rate": 8.834068843777583e-06,
|
| 2564 |
+
"loss": 2.8571,
|
| 2565 |
+
"step": 342000
|
| 2566 |
+
},
|
| 2567 |
+
{
|
| 2568 |
+
"epoch": 18.45,
|
| 2569 |
+
"grad_norm": 0.9799352288246155,
|
| 2570 |
+
"learning_rate": 8.539864666078259e-06,
|
| 2571 |
+
"loss": 2.8517,
|
| 2572 |
+
"step": 343000
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"epoch": 18.5,
|
| 2576 |
+
"grad_norm": 0.9541887044906616,
|
| 2577 |
+
"learning_rate": 8.245954692556634e-06,
|
| 2578 |
+
"loss": 2.8557,
|
| 2579 |
+
"step": 344000
|
| 2580 |
+
},
|
| 2581 |
+
{
|
| 2582 |
+
"epoch": 18.55,
|
| 2583 |
+
"grad_norm": 0.945749044418335,
|
| 2584 |
+
"learning_rate": 7.951750514857312e-06,
|
| 2585 |
+
"loss": 2.86,
|
| 2586 |
+
"step": 345000
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"epoch": 18.61,
|
| 2590 |
+
"grad_norm": 0.9562386274337769,
|
| 2591 |
+
"learning_rate": 7.657546337157988e-06,
|
| 2592 |
+
"loss": 2.8559,
|
| 2593 |
+
"step": 346000
|
| 2594 |
+
},
|
| 2595 |
+
{
|
| 2596 |
+
"epoch": 18.66,
|
| 2597 |
+
"grad_norm": 0.9685288071632385,
|
| 2598 |
+
"learning_rate": 7.363930567814063e-06,
|
| 2599 |
+
"loss": 2.8546,
|
| 2600 |
+
"step": 347000
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 18.71,
|
| 2604 |
+
"grad_norm": 0.9271679520606995,
|
| 2605 |
+
"learning_rate": 7.06972639011474e-06,
|
| 2606 |
+
"loss": 2.8533,
|
| 2607 |
+
"step": 348000
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 18.77,
|
| 2611 |
+
"grad_norm": 0.9129212498664856,
|
| 2612 |
+
"learning_rate": 6.775522212415417e-06,
|
| 2613 |
+
"loss": 2.8612,
|
| 2614 |
+
"step": 349000
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"epoch": 18.82,
|
| 2618 |
+
"grad_norm": 0.9596994519233704,
|
| 2619 |
+
"learning_rate": 6.481318034716093e-06,
|
| 2620 |
+
"loss": 2.8562,
|
| 2621 |
+
"step": 350000
|
| 2622 |
+
},
|
| 2623 |
+
{
|
| 2624 |
+
"epoch": 18.88,
|
| 2625 |
+
"grad_norm": 0.9252175092697144,
|
| 2626 |
+
"learning_rate": 6.18711385701677e-06,
|
| 2627 |
+
"loss": 2.8577,
|
| 2628 |
+
"step": 351000
|
| 2629 |
+
},
|
| 2630 |
+
{
|
| 2631 |
+
"epoch": 18.93,
|
| 2632 |
+
"grad_norm": 0.9257832169532776,
|
| 2633 |
+
"learning_rate": 5.893203883495145e-06,
|
| 2634 |
+
"loss": 2.8585,
|
| 2635 |
+
"step": 352000
|
| 2636 |
+
},
|
| 2637 |
+
{
|
| 2638 |
+
"epoch": 18.98,
|
| 2639 |
+
"grad_norm": 0.9196575880050659,
|
| 2640 |
+
"learning_rate": 5.598999705795822e-06,
|
| 2641 |
+
"loss": 2.8514,
|
| 2642 |
+
"step": 353000
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 19.0,
|
| 2646 |
+
"eval_accuracy": 0.4072752522102257,
|
| 2647 |
+
"eval_loss": 3.4321706295013428,
|
| 2648 |
+
"eval_runtime": 155.3795,
|
| 2649 |
+
"eval_samples_per_second": 372.771,
|
| 2650 |
+
"eval_steps_per_second": 5.831,
|
| 2651 |
+
"step": 353305
|
| 2652 |
+
},
|
| 2653 |
+
{
|
| 2654 |
+
"epoch": 19.04,
|
| 2655 |
+
"grad_norm": 0.9688698649406433,
|
| 2656 |
+
"learning_rate": 5.305089732274199e-06,
|
| 2657 |
+
"loss": 2.8452,
|
| 2658 |
+
"step": 354000
|
| 2659 |
+
},
|
| 2660 |
+
{
|
| 2661 |
+
"epoch": 19.09,
|
| 2662 |
+
"grad_norm": 0.9426762461662292,
|
| 2663 |
+
"learning_rate": 5.010885554574876e-06,
|
| 2664 |
+
"loss": 2.8401,
|
| 2665 |
+
"step": 355000
|
| 2666 |
+
},
|
| 2667 |
+
{
|
| 2668 |
+
"epoch": 19.14,
|
| 2669 |
+
"grad_norm": 0.9447112679481506,
|
| 2670 |
+
"learning_rate": 4.716975581053251e-06,
|
| 2671 |
+
"loss": 2.839,
|
| 2672 |
+
"step": 356000
|
| 2673 |
+
},
|
| 2674 |
+
{
|
| 2675 |
+
"epoch": 19.2,
|
| 2676 |
+
"grad_norm": 0.9574055671691895,
|
| 2677 |
+
"learning_rate": 4.423065607531627e-06,
|
| 2678 |
+
"loss": 2.8447,
|
| 2679 |
+
"step": 357000
|
| 2680 |
+
},
|
| 2681 |
+
{
|
| 2682 |
+
"epoch": 19.25,
|
| 2683 |
+
"grad_norm": 0.9574621319770813,
|
| 2684 |
+
"learning_rate": 4.128861429832304e-06,
|
| 2685 |
+
"loss": 2.8391,
|
| 2686 |
+
"step": 358000
|
| 2687 |
+
},
|
| 2688 |
+
{
|
| 2689 |
+
"epoch": 19.31,
|
| 2690 |
+
"grad_norm": 0.9345229268074036,
|
| 2691 |
+
"learning_rate": 3.83465725213298e-06,
|
| 2692 |
+
"loss": 2.8444,
|
| 2693 |
+
"step": 359000
|
| 2694 |
+
},
|
| 2695 |
+
{
|
| 2696 |
+
"epoch": 19.36,
|
| 2697 |
+
"grad_norm": 0.9420320391654968,
|
| 2698 |
+
"learning_rate": 3.5404530744336568e-06,
|
| 2699 |
+
"loss": 2.8459,
|
| 2700 |
+
"step": 360000
|
| 2701 |
+
},
|
| 2702 |
+
{
|
| 2703 |
+
"epoch": 19.41,
|
| 2704 |
+
"grad_norm": 0.9578850865364075,
|
| 2705 |
+
"learning_rate": 3.2465431009120332e-06,
|
| 2706 |
+
"loss": 2.8409,
|
| 2707 |
+
"step": 361000
|
| 2708 |
+
},
|
| 2709 |
+
{
|
| 2710 |
+
"epoch": 19.47,
|
| 2711 |
+
"grad_norm": 0.963036060333252,
|
| 2712 |
+
"learning_rate": 2.95233892321271e-06,
|
| 2713 |
+
"loss": 2.8392,
|
| 2714 |
+
"step": 362000
|
| 2715 |
+
},
|
| 2716 |
+
{
|
| 2717 |
+
"epoch": 19.52,
|
| 2718 |
+
"grad_norm": 0.9696721434593201,
|
| 2719 |
+
"learning_rate": 2.6581347455133864e-06,
|
| 2720 |
+
"loss": 2.8451,
|
| 2721 |
+
"step": 363000
|
| 2722 |
+
},
|
| 2723 |
+
{
|
| 2724 |
+
"epoch": 19.58,
|
| 2725 |
+
"grad_norm": 0.9655833840370178,
|
| 2726 |
+
"learning_rate": 2.3642247719917624e-06,
|
| 2727 |
+
"loss": 2.8446,
|
| 2728 |
+
"step": 364000
|
| 2729 |
+
},
|
| 2730 |
+
{
|
| 2731 |
+
"epoch": 19.63,
|
| 2732 |
+
"grad_norm": 0.9735895395278931,
|
| 2733 |
+
"learning_rate": 2.070314798470138e-06,
|
| 2734 |
+
"loss": 2.8436,
|
| 2735 |
+
"step": 365000
|
| 2736 |
+
},
|
| 2737 |
+
{
|
| 2738 |
+
"epoch": 19.68,
|
| 2739 |
+
"grad_norm": 0.9257269501686096,
|
| 2740 |
+
"learning_rate": 1.7761106207708148e-06,
|
| 2741 |
+
"loss": 2.8413,
|
| 2742 |
+
"step": 366000
|
| 2743 |
+
},
|
| 2744 |
+
{
|
| 2745 |
+
"epoch": 19.74,
|
| 2746 |
+
"grad_norm": 0.9528219103813171,
|
| 2747 |
+
"learning_rate": 1.482200647249191e-06,
|
| 2748 |
+
"loss": 2.8403,
|
| 2749 |
+
"step": 367000
|
| 2750 |
+
},
|
| 2751 |
+
{
|
| 2752 |
+
"epoch": 19.79,
|
| 2753 |
+
"grad_norm": 0.9575457572937012,
|
| 2754 |
+
"learning_rate": 1.1879964695498676e-06,
|
| 2755 |
+
"loss": 2.8421,
|
| 2756 |
+
"step": 368000
|
| 2757 |
+
},
|
| 2758 |
+
{
|
| 2759 |
+
"epoch": 19.84,
|
| 2760 |
+
"grad_norm": 0.9463661909103394,
|
| 2761 |
+
"learning_rate": 8.937922918505443e-07,
|
| 2762 |
+
"loss": 2.8405,
|
| 2763 |
+
"step": 369000
|
| 2764 |
+
},
|
| 2765 |
+
{
|
| 2766 |
+
"epoch": 19.9,
|
| 2767 |
+
"grad_norm": 0.9508777856826782,
|
| 2768 |
+
"learning_rate": 5.99588114151221e-07,
|
| 2769 |
+
"loss": 2.842,
|
| 2770 |
+
"step": 370000
|
| 2771 |
+
},
|
| 2772 |
+
{
|
| 2773 |
+
"epoch": 19.95,
|
| 2774 |
+
"grad_norm": 0.93899005651474,
|
| 2775 |
+
"learning_rate": 3.0567814062959694e-07,
|
| 2776 |
+
"loss": 2.8467,
|
| 2777 |
+
"step": 371000
|
| 2778 |
+
},
|
| 2779 |
+
{
|
| 2780 |
+
"epoch": 20.0,
|
| 2781 |
+
"eval_accuracy": 0.40723568402608223,
|
| 2782 |
+
"eval_loss": 3.4373319149017334,
|
| 2783 |
+
"eval_runtime": 154.9871,
|
| 2784 |
+
"eval_samples_per_second": 373.715,
|
| 2785 |
+
"eval_steps_per_second": 5.846,
|
| 2786 |
+
"step": 371900
|
| 2787 |
+
},
|
| 2788 |
+
{
|
| 2789 |
+
"epoch": 20.0,
|
| 2790 |
+
"step": 371900,
|
| 2791 |
+
"total_flos": 1.5669257538816e+18,
|
| 2792 |
+
"train_loss": 3.1537178330448374,
|
| 2793 |
+
"train_runtime": 81513.9803,
|
| 2794 |
+
"train_samples_per_second": 145.996,
|
| 2795 |
+
"train_steps_per_second": 4.562
|
| 2796 |
+
}
|
| 2797 |
+
],
|
| 2798 |
+
"logging_steps": 1000,
|
| 2799 |
+
"max_steps": 371900,
|
| 2800 |
+
"num_input_tokens_seen": 0,
|
| 2801 |
+
"num_train_epochs": 20,
|
| 2802 |
+
"save_steps": 5000,
|
| 2803 |
+
"total_flos": 1.5669257538816e+18,
|
| 2804 |
+
"train_batch_size": 32,
|
| 2805 |
+
"trial_name": null,
|
| 2806 |
+
"trial_params": null
|
| 2807 |
+
}
|