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_anans_new-3e-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_anans_new-3e-4
|
15 |
|
16 |
-
This model was trained from scratch on
|
17 |
It achieves the following results on the evaluation set:
|
18 |
- Loss: 3.4257
|
19 |
- Accuracy: 0.4096
|
|
|
1 |
---
|
2 |
tags:
|
3 |
- generated_from_trainer
|
4 |
+
datasets:
|
5 |
+
- kanishka/counterfactual_babylm_anans_new
|
6 |
metrics:
|
7 |
- accuracy
|
8 |
model-index:
|
9 |
- name: smolm-autoreg-bpe-counterfactual_babylm_anans_new-3e-4
|
10 |
+
results:
|
11 |
+
- task:
|
12 |
+
name: Causal Language Modeling
|
13 |
+
type: text-generation
|
14 |
+
dataset:
|
15 |
+
name: kanishka/counterfactual_babylm_anans_new
|
16 |
+
type: kanishka/counterfactual_babylm_anans_new
|
17 |
+
metrics:
|
18 |
+
- name: Accuracy
|
19 |
+
type: accuracy
|
20 |
+
value: 0.4096272415057219
|
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_anans_new-3e-4
|
27 |
|
28 |
+
This model was trained from scratch on the kanishka/counterfactual_babylm_anans_new dataset.
|
29 |
It achieves the following results on the evaluation set:
|
30 |
- Loss: 3.4257
|
31 |
- Accuracy: 0.4096
|
all_results.json
ADDED
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 20.0,
|
3 |
+
"eval_accuracy": 0.4096272415057219,
|
4 |
+
"eval_loss": 3.4256882667541504,
|
5 |
+
"eval_runtime": 154.1089,
|
6 |
+
"eval_samples": 57921,
|
7 |
+
"eval_samples_per_second": 375.845,
|
8 |
+
"eval_steps_per_second": 5.879,
|
9 |
+
"perplexity": 30.74379750505285,
|
10 |
+
"total_flos": 1.5669257538816e+18,
|
11 |
+
"train_loss": 3.0592986363049124,
|
12 |
+
"train_runtime": 81068.1718,
|
13 |
+
"train_samples": 595035,
|
14 |
+
"train_samples_per_second": 146.799,
|
15 |
+
"train_steps_per_second": 4.587
|
16 |
+
}
|
eval_results.json
ADDED
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 20.0,
|
3 |
+
"eval_accuracy": 0.4096272415057219,
|
4 |
+
"eval_loss": 3.4256882667541504,
|
5 |
+
"eval_runtime": 154.1089,
|
6 |
+
"eval_samples": 57921,
|
7 |
+
"eval_samples_per_second": 375.845,
|
8 |
+
"eval_steps_per_second": 5.879,
|
9 |
+
"perplexity": 30.74379750505285
|
10 |
+
}
|
runs/Jun05_07-51-28_phyl-ling-p01.la.utexas.edu/events.out.tfevents.1717673474.phyl-ling-p01.la.utexas.edu.175344.1
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e6e284846fb41be8706eb3897d2cb1f75fe68484b2e10c6bd43cc920b0a8b1ed
|
3 |
+
size 417
|
train_results.json
ADDED
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 20.0,
|
3 |
+
"total_flos": 1.5669257538816e+18,
|
4 |
+
"train_loss": 3.0592986363049124,
|
5 |
+
"train_runtime": 81068.1718,
|
6 |
+
"train_samples": 595035,
|
7 |
+
"train_samples_per_second": 146.799,
|
8 |
+
"train_steps_per_second": 4.587
|
9 |
+
}
|
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": 0.05377789728421619,
|
13 |
+
"grad_norm": 0.7097817659378052,
|
14 |
+
"learning_rate": 9.375e-06,
|
15 |
+
"loss": 6.8536,
|
16 |
+
"step": 1000
|
17 |
+
},
|
18 |
+
{
|
19 |
+
"epoch": 0.10755579456843238,
|
20 |
+
"grad_norm": 0.8630837798118591,
|
21 |
+
"learning_rate": 1.875e-05,
|
22 |
+
"loss": 5.3656,
|
23 |
+
"step": 2000
|
24 |
+
},
|
25 |
+
{
|
26 |
+
"epoch": 0.16133369185264856,
|
27 |
+
"grad_norm": 0.9362223148345947,
|
28 |
+
"learning_rate": 2.8125e-05,
|
29 |
+
"loss": 5.0405,
|
30 |
+
"step": 3000
|
31 |
+
},
|
32 |
+
{
|
33 |
+
"epoch": 0.21511158913686476,
|
34 |
+
"grad_norm": 0.9098538756370544,
|
35 |
+
"learning_rate": 3.75e-05,
|
36 |
+
"loss": 4.8207,
|
37 |
+
"step": 4000
|
38 |
+
},
|
39 |
+
{
|
40 |
+
"epoch": 0.26888948642108096,
|
41 |
+
"grad_norm": 1.0826491117477417,
|
42 |
+
"learning_rate": 4.6874999999999994e-05,
|
43 |
+
"loss": 4.6468,
|
44 |
+
"step": 5000
|
45 |
+
},
|
46 |
+
{
|
47 |
+
"epoch": 0.32266738370529713,
|
48 |
+
"grad_norm": 0.9499918222427368,
|
49 |
+
"learning_rate": 5.625e-05,
|
50 |
+
"loss": 4.4983,
|
51 |
+
"step": 6000
|
52 |
+
},
|
53 |
+
{
|
54 |
+
"epoch": 0.3764452809895133,
|
55 |
+
"grad_norm": 0.9495302438735962,
|
56 |
+
"learning_rate": 6.5625e-05,
|
57 |
+
"loss": 4.39,
|
58 |
+
"step": 7000
|
59 |
+
},
|
60 |
+
{
|
61 |
+
"epoch": 0.4302231782737295,
|
62 |
+
"grad_norm": 0.8918095231056213,
|
63 |
+
"learning_rate": 7.5e-05,
|
64 |
+
"loss": 4.2828,
|
65 |
+
"step": 8000
|
66 |
+
},
|
67 |
+
{
|
68 |
+
"epoch": 0.4840010755579457,
|
69 |
+
"grad_norm": 0.8999391794204712,
|
70 |
+
"learning_rate": 8.437499999999999e-05,
|
71 |
+
"loss": 4.2052,
|
72 |
+
"step": 9000
|
73 |
+
},
|
74 |
+
{
|
75 |
+
"epoch": 0.5377789728421619,
|
76 |
+
"grad_norm": 0.872922956943512,
|
77 |
+
"learning_rate": 9.374999999999999e-05,
|
78 |
+
"loss": 4.1356,
|
79 |
+
"step": 10000
|
80 |
+
},
|
81 |
+
{
|
82 |
+
"epoch": 0.591556870126378,
|
83 |
+
"grad_norm": 0.8972263932228088,
|
84 |
+
"learning_rate": 0.00010312499999999999,
|
85 |
+
"loss": 4.0696,
|
86 |
+
"step": 11000
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 0.6453347674105943,
|
90 |
+
"grad_norm": 0.853663980960846,
|
91 |
+
"learning_rate": 0.000112490625,
|
92 |
+
"loss": 4.008,
|
93 |
+
"step": 12000
|
94 |
+
},
|
95 |
+
{
|
96 |
+
"epoch": 0.6991126646948105,
|
97 |
+
"grad_norm": 0.8272613286972046,
|
98 |
+
"learning_rate": 0.000121865625,
|
99 |
+
"loss": 3.9457,
|
100 |
+
"step": 13000
|
101 |
+
},
|
102 |
+
{
|
103 |
+
"epoch": 0.7528905619790266,
|
104 |
+
"grad_norm": 0.8427807092666626,
|
105 |
+
"learning_rate": 0.00013123125,
|
106 |
+
"loss": 3.8928,
|
107 |
+
"step": 14000
|
108 |
+
},
|
109 |
+
{
|
110 |
+
"epoch": 0.8066684592632428,
|
111 |
+
"grad_norm": 0.7996278405189514,
|
112 |
+
"learning_rate": 0.00014060625,
|
113 |
+
"loss": 3.8373,
|
114 |
+
"step": 15000
|
115 |
+
},
|
116 |
+
{
|
117 |
+
"epoch": 0.860446356547459,
|
118 |
+
"grad_norm": 0.7708893418312073,
|
119 |
+
"learning_rate": 0.000149971875,
|
120 |
+
"loss": 3.8015,
|
121 |
+
"step": 16000
|
122 |
+
},
|
123 |
+
{
|
124 |
+
"epoch": 0.9142242538316752,
|
125 |
+
"grad_norm": 0.750762403011322,
|
126 |
+
"learning_rate": 0.000159346875,
|
127 |
+
"loss": 3.7619,
|
128 |
+
"step": 17000
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"epoch": 0.9680021511158914,
|
132 |
+
"grad_norm": 0.8043661117553711,
|
133 |
+
"learning_rate": 0.000168703125,
|
134 |
+
"loss": 3.7355,
|
135 |
+
"step": 18000
|
136 |
+
},
|
137 |
+
{
|
138 |
+
"epoch": 1.0,
|
139 |
+
"eval_accuracy": 0.3460631369831053,
|
140 |
+
"eval_loss": 3.899465799331665,
|
141 |
+
"eval_runtime": 151.8767,
|
142 |
+
"eval_samples_per_second": 381.369,
|
143 |
+
"eval_steps_per_second": 5.965,
|
144 |
+
"step": 18595
|
145 |
+
},
|
146 |
+
{
|
147 |
+
"epoch": 1.0217800484001076,
|
148 |
+
"grad_norm": 0.7180488109588623,
|
149 |
+
"learning_rate": 0.000178078125,
|
150 |
+
"loss": 3.6982,
|
151 |
+
"step": 19000
|
152 |
+
},
|
153 |
+
{
|
154 |
+
"epoch": 1.0755579456843238,
|
155 |
+
"grad_norm": 0.6873713135719299,
|
156 |
+
"learning_rate": 0.00018745312499999998,
|
157 |
+
"loss": 3.6651,
|
158 |
+
"step": 20000
|
159 |
+
},
|
160 |
+
{
|
161 |
+
"epoch": 1.1293358429685398,
|
162 |
+
"grad_norm": 0.6932269334793091,
|
163 |
+
"learning_rate": 0.00019680937499999996,
|
164 |
+
"loss": 3.6514,
|
165 |
+
"step": 21000
|
166 |
+
},
|
167 |
+
{
|
168 |
+
"epoch": 1.183113740252756,
|
169 |
+
"grad_norm": 0.668073296546936,
|
170 |
+
"learning_rate": 0.00020618437499999995,
|
171 |
+
"loss": 3.6328,
|
172 |
+
"step": 22000
|
173 |
+
},
|
174 |
+
{
|
175 |
+
"epoch": 1.2368916375369723,
|
176 |
+
"grad_norm": 0.6708812713623047,
|
177 |
+
"learning_rate": 0.00021555937499999998,
|
178 |
+
"loss": 3.6162,
|
179 |
+
"step": 23000
|
180 |
+
},
|
181 |
+
{
|
182 |
+
"epoch": 1.2906695348211885,
|
183 |
+
"grad_norm": 0.6274955868721008,
|
184 |
+
"learning_rate": 0.00022493437499999998,
|
185 |
+
"loss": 3.592,
|
186 |
+
"step": 24000
|
187 |
+
},
|
188 |
+
{
|
189 |
+
"epoch": 1.3444474321054047,
|
190 |
+
"grad_norm": 0.6207209229469299,
|
191 |
+
"learning_rate": 0.00023430937499999997,
|
192 |
+
"loss": 3.5712,
|
193 |
+
"step": 25000
|
194 |
+
},
|
195 |
+
{
|
196 |
+
"epoch": 1.398225329389621,
|
197 |
+
"grad_norm": 0.5920027494430542,
|
198 |
+
"learning_rate": 0.00024367499999999997,
|
199 |
+
"loss": 3.5621,
|
200 |
+
"step": 26000
|
201 |
+
},
|
202 |
+
{
|
203 |
+
"epoch": 1.452003226673837,
|
204 |
+
"grad_norm": 0.5753495097160339,
|
205 |
+
"learning_rate": 0.00025305,
|
206 |
+
"loss": 3.5494,
|
207 |
+
"step": 27000
|
208 |
+
},
|
209 |
+
{
|
210 |
+
"epoch": 1.5057811239580532,
|
211 |
+
"grad_norm": 0.5462520718574524,
|
212 |
+
"learning_rate": 0.000262425,
|
213 |
+
"loss": 3.5336,
|
214 |
+
"step": 28000
|
215 |
+
},
|
216 |
+
{
|
217 |
+
"epoch": 1.5595590212422694,
|
218 |
+
"grad_norm": 0.5698980689048767,
|
219 |
+
"learning_rate": 0.000271790625,
|
220 |
+
"loss": 3.5201,
|
221 |
+
"step": 29000
|
222 |
+
},
|
223 |
+
{
|
224 |
+
"epoch": 1.6133369185264856,
|
225 |
+
"grad_norm": 0.5474157333374023,
|
226 |
+
"learning_rate": 0.00028115624999999994,
|
227 |
+
"loss": 3.5162,
|
228 |
+
"step": 30000
|
229 |
+
},
|
230 |
+
{
|
231 |
+
"epoch": 1.6671148158107019,
|
232 |
+
"grad_norm": 0.5613803267478943,
|
233 |
+
"learning_rate": 0.00029053124999999994,
|
234 |
+
"loss": 3.5054,
|
235 |
+
"step": 31000
|
236 |
+
},
|
237 |
+
{
|
238 |
+
"epoch": 1.7208927130949179,
|
239 |
+
"grad_norm": 0.5235997438430786,
|
240 |
+
"learning_rate": 0.00029990624999999993,
|
241 |
+
"loss": 3.4928,
|
242 |
+
"step": 32000
|
243 |
+
},
|
244 |
+
{
|
245 |
+
"epoch": 1.7746706103791343,
|
246 |
+
"grad_norm": 0.5065760016441345,
|
247 |
+
"learning_rate": 0.00029912621359223297,
|
248 |
+
"loss": 3.4811,
|
249 |
+
"step": 33000
|
250 |
+
},
|
251 |
+
{
|
252 |
+
"epoch": 1.8284485076633503,
|
253 |
+
"grad_norm": 0.5031436085700989,
|
254 |
+
"learning_rate": 0.0002982444836716681,
|
255 |
+
"loss": 3.4693,
|
256 |
+
"step": 34000
|
257 |
+
},
|
258 |
+
{
|
259 |
+
"epoch": 1.8822264049475665,
|
260 |
+
"grad_norm": 0.47748515009880066,
|
261 |
+
"learning_rate": 0.00029736187113857014,
|
262 |
+
"loss": 3.4572,
|
263 |
+
"step": 35000
|
264 |
+
},
|
265 |
+
{
|
266 |
+
"epoch": 1.9360043022317828,
|
267 |
+
"grad_norm": 0.47696492075920105,
|
268 |
+
"learning_rate": 0.0002964801412180053,
|
269 |
+
"loss": 3.4431,
|
270 |
+
"step": 36000
|
271 |
+
},
|
272 |
+
{
|
273 |
+
"epoch": 1.9897821995159988,
|
274 |
+
"grad_norm": 0.43063884973526,
|
275 |
+
"learning_rate": 0.0002955975286849073,
|
276 |
+
"loss": 3.4292,
|
277 |
+
"step": 37000
|
278 |
+
},
|
279 |
+
{
|
280 |
+
"epoch": 2.0,
|
281 |
+
"eval_accuracy": 0.37681863334985255,
|
282 |
+
"eval_loss": 3.60871958732605,
|
283 |
+
"eval_runtime": 153.6559,
|
284 |
+
"eval_samples_per_second": 376.953,
|
285 |
+
"eval_steps_per_second": 5.896,
|
286 |
+
"step": 37190
|
287 |
+
},
|
288 |
+
{
|
289 |
+
"epoch": 2.043560096800215,
|
290 |
+
"grad_norm": 0.4518415331840515,
|
291 |
+
"learning_rate": 0.0002947149161518093,
|
292 |
+
"loss": 3.385,
|
293 |
+
"step": 38000
|
294 |
+
},
|
295 |
+
{
|
296 |
+
"epoch": 2.0973379940844312,
|
297 |
+
"grad_norm": 0.45232945680618286,
|
298 |
+
"learning_rate": 0.0002938331862312445,
|
299 |
+
"loss": 3.3727,
|
300 |
+
"step": 39000
|
301 |
+
},
|
302 |
+
{
|
303 |
+
"epoch": 2.1511158913686477,
|
304 |
+
"grad_norm": 0.4550546407699585,
|
305 |
+
"learning_rate": 0.0002929505736981465,
|
306 |
+
"loss": 3.366,
|
307 |
+
"step": 40000
|
308 |
+
},
|
309 |
+
{
|
310 |
+
"epoch": 2.2048937886528637,
|
311 |
+
"grad_norm": 0.43207359313964844,
|
312 |
+
"learning_rate": 0.0002920679611650485,
|
313 |
+
"loss": 3.3641,
|
314 |
+
"step": 41000
|
315 |
+
},
|
316 |
+
{
|
317 |
+
"epoch": 2.2586716859370797,
|
318 |
+
"grad_norm": 0.43842893838882446,
|
319 |
+
"learning_rate": 0.0002911853486319506,
|
320 |
+
"loss": 3.3596,
|
321 |
+
"step": 42000
|
322 |
+
},
|
323 |
+
{
|
324 |
+
"epoch": 2.312449583221296,
|
325 |
+
"grad_norm": 0.420186311006546,
|
326 |
+
"learning_rate": 0.0002903036187113857,
|
327 |
+
"loss": 3.3504,
|
328 |
+
"step": 43000
|
329 |
+
},
|
330 |
+
{
|
331 |
+
"epoch": 2.366227480505512,
|
332 |
+
"grad_norm": 0.4150792062282562,
|
333 |
+
"learning_rate": 0.0002894210061782877,
|
334 |
+
"loss": 3.3515,
|
335 |
+
"step": 44000
|
336 |
+
},
|
337 |
+
{
|
338 |
+
"epoch": 2.4200053777897286,
|
339 |
+
"grad_norm": 0.4212990701198578,
|
340 |
+
"learning_rate": 0.0002885401588702559,
|
341 |
+
"loss": 3.3412,
|
342 |
+
"step": 45000
|
343 |
+
},
|
344 |
+
{
|
345 |
+
"epoch": 2.4737832750739446,
|
346 |
+
"grad_norm": 0.4120098054409027,
|
347 |
+
"learning_rate": 0.00028765754633715794,
|
348 |
+
"loss": 3.3399,
|
349 |
+
"step": 46000
|
350 |
+
},
|
351 |
+
{
|
352 |
+
"epoch": 2.527561172358161,
|
353 |
+
"grad_norm": 0.38389134407043457,
|
354 |
+
"learning_rate": 0.00028677493380406,
|
355 |
+
"loss": 3.3321,
|
356 |
+
"step": 47000
|
357 |
+
},
|
358 |
+
{
|
359 |
+
"epoch": 2.581339069642377,
|
360 |
+
"grad_norm": 0.38017722964286804,
|
361 |
+
"learning_rate": 0.00028589232127096203,
|
362 |
+
"loss": 3.3214,
|
363 |
+
"step": 48000
|
364 |
+
},
|
365 |
+
{
|
366 |
+
"epoch": 2.635116966926593,
|
367 |
+
"grad_norm": 0.3919270634651184,
|
368 |
+
"learning_rate": 0.00028500970873786405,
|
369 |
+
"loss": 3.3228,
|
370 |
+
"step": 49000
|
371 |
+
},
|
372 |
+
{
|
373 |
+
"epoch": 2.6888948642108095,
|
374 |
+
"grad_norm": 0.4108082354068756,
|
375 |
+
"learning_rate": 0.0002841279788172992,
|
376 |
+
"loss": 3.3138,
|
377 |
+
"step": 50000
|
378 |
+
},
|
379 |
+
{
|
380 |
+
"epoch": 2.7426727614950255,
|
381 |
+
"grad_norm": 0.3886261582374573,
|
382 |
+
"learning_rate": 0.0002832453662842012,
|
383 |
+
"loss": 3.3056,
|
384 |
+
"step": 51000
|
385 |
+
},
|
386 |
+
{
|
387 |
+
"epoch": 2.796450658779242,
|
388 |
+
"grad_norm": 0.3850669264793396,
|
389 |
+
"learning_rate": 0.00028236275375110323,
|
390 |
+
"loss": 3.3032,
|
391 |
+
"step": 52000
|
392 |
+
},
|
393 |
+
{
|
394 |
+
"epoch": 2.850228556063458,
|
395 |
+
"grad_norm": 0.3835964798927307,
|
396 |
+
"learning_rate": 0.0002814810238305384,
|
397 |
+
"loss": 3.2969,
|
398 |
+
"step": 53000
|
399 |
+
},
|
400 |
+
{
|
401 |
+
"epoch": 2.904006453347674,
|
402 |
+
"grad_norm": 0.38047948479652405,
|
403 |
+
"learning_rate": 0.0002805992939099735,
|
404 |
+
"loss": 3.2936,
|
405 |
+
"step": 54000
|
406 |
+
},
|
407 |
+
{
|
408 |
+
"epoch": 2.9577843506318904,
|
409 |
+
"grad_norm": 0.3762407600879669,
|
410 |
+
"learning_rate": 0.00027971668137687555,
|
411 |
+
"loss": 3.2927,
|
412 |
+
"step": 55000
|
413 |
+
},
|
414 |
+
{
|
415 |
+
"epoch": 3.0,
|
416 |
+
"eval_accuracy": 0.3895145789948566,
|
417 |
+
"eval_loss": 3.5000998973846436,
|
418 |
+
"eval_runtime": 153.5704,
|
419 |
+
"eval_samples_per_second": 377.162,
|
420 |
+
"eval_steps_per_second": 5.9,
|
421 |
+
"step": 55785
|
422 |
+
},
|
423 |
+
{
|
424 |
+
"epoch": 3.0115622479161064,
|
425 |
+
"grad_norm": 0.4009055495262146,
|
426 |
+
"learning_rate": 0.00027883406884377757,
|
427 |
+
"loss": 3.2765,
|
428 |
+
"step": 56000
|
429 |
+
},
|
430 |
+
{
|
431 |
+
"epoch": 3.065340145200323,
|
432 |
+
"grad_norm": 0.37349995970726013,
|
433 |
+
"learning_rate": 0.0002779514563106796,
|
434 |
+
"loss": 3.2286,
|
435 |
+
"step": 57000
|
436 |
+
},
|
437 |
+
{
|
438 |
+
"epoch": 3.119118042484539,
|
439 |
+
"grad_norm": 0.3908737599849701,
|
440 |
+
"learning_rate": 0.00027706972639011474,
|
441 |
+
"loss": 3.2289,
|
442 |
+
"step": 58000
|
443 |
+
},
|
444 |
+
{
|
445 |
+
"epoch": 3.172895939768755,
|
446 |
+
"grad_norm": 0.407253235578537,
|
447 |
+
"learning_rate": 0.00027618711385701675,
|
448 |
+
"loss": 3.2265,
|
449 |
+
"step": 59000
|
450 |
+
},
|
451 |
+
{
|
452 |
+
"epoch": 3.2266738370529713,
|
453 |
+
"grad_norm": 0.381302148103714,
|
454 |
+
"learning_rate": 0.00027530538393645185,
|
455 |
+
"loss": 3.231,
|
456 |
+
"step": 60000
|
457 |
+
},
|
458 |
+
{
|
459 |
+
"epoch": 3.2804517343371873,
|
460 |
+
"grad_norm": 0.3845687508583069,
|
461 |
+
"learning_rate": 0.0002744227714033539,
|
462 |
+
"loss": 3.228,
|
463 |
+
"step": 61000
|
464 |
+
},
|
465 |
+
{
|
466 |
+
"epoch": 3.3342296316214037,
|
467 |
+
"grad_norm": 0.37134280800819397,
|
468 |
+
"learning_rate": 0.00027354015887025594,
|
469 |
+
"loss": 3.2294,
|
470 |
+
"step": 62000
|
471 |
+
},
|
472 |
+
{
|
473 |
+
"epoch": 3.3880075289056197,
|
474 |
+
"grad_norm": 0.3631034195423126,
|
475 |
+
"learning_rate": 0.0002726584289496911,
|
476 |
+
"loss": 3.2269,
|
477 |
+
"step": 63000
|
478 |
+
},
|
479 |
+
{
|
480 |
+
"epoch": 3.4417854261898357,
|
481 |
+
"grad_norm": 0.39031264185905457,
|
482 |
+
"learning_rate": 0.0002717758164165931,
|
483 |
+
"loss": 3.2245,
|
484 |
+
"step": 64000
|
485 |
+
},
|
486 |
+
{
|
487 |
+
"epoch": 3.495563323474052,
|
488 |
+
"grad_norm": 0.3664679527282715,
|
489 |
+
"learning_rate": 0.0002708932038834951,
|
490 |
+
"loss": 3.2219,
|
491 |
+
"step": 65000
|
492 |
+
},
|
493 |
+
{
|
494 |
+
"epoch": 3.549341220758268,
|
495 |
+
"grad_norm": 0.37691643834114075,
|
496 |
+
"learning_rate": 0.00027001059135039714,
|
497 |
+
"loss": 3.2152,
|
498 |
+
"step": 66000
|
499 |
+
},
|
500 |
+
{
|
501 |
+
"epoch": 3.6031191180424846,
|
502 |
+
"grad_norm": 0.3464328944683075,
|
503 |
+
"learning_rate": 0.00026912886142983224,
|
504 |
+
"loss": 3.2158,
|
505 |
+
"step": 67000
|
506 |
+
},
|
507 |
+
{
|
508 |
+
"epoch": 3.6568970153267006,
|
509 |
+
"grad_norm": 0.360546350479126,
|
510 |
+
"learning_rate": 0.0002682462488967343,
|
511 |
+
"loss": 3.2173,
|
512 |
+
"step": 68000
|
513 |
+
},
|
514 |
+
{
|
515 |
+
"epoch": 3.7106749126109166,
|
516 |
+
"grad_norm": 0.39308884739875793,
|
517 |
+
"learning_rate": 0.00026736451897616946,
|
518 |
+
"loss": 3.2144,
|
519 |
+
"step": 69000
|
520 |
+
},
|
521 |
+
{
|
522 |
+
"epoch": 3.764452809895133,
|
523 |
+
"grad_norm": 0.35938188433647156,
|
524 |
+
"learning_rate": 0.0002664819064430715,
|
525 |
+
"loss": 3.2169,
|
526 |
+
"step": 70000
|
527 |
+
},
|
528 |
+
{
|
529 |
+
"epoch": 3.8182307071793495,
|
530 |
+
"grad_norm": 0.3498573899269104,
|
531 |
+
"learning_rate": 0.0002656001765225066,
|
532 |
+
"loss": 3.2095,
|
533 |
+
"step": 71000
|
534 |
+
},
|
535 |
+
{
|
536 |
+
"epoch": 3.8720086044635655,
|
537 |
+
"grad_norm": 0.3463761806488037,
|
538 |
+
"learning_rate": 0.0002647175639894086,
|
539 |
+
"loss": 3.2078,
|
540 |
+
"step": 72000
|
541 |
+
},
|
542 |
+
{
|
543 |
+
"epoch": 3.9257865017477815,
|
544 |
+
"grad_norm": 0.3538376986980438,
|
545 |
+
"learning_rate": 0.00026383583406884375,
|
546 |
+
"loss": 3.2037,
|
547 |
+
"step": 73000
|
548 |
+
},
|
549 |
+
{
|
550 |
+
"epoch": 3.979564399031998,
|
551 |
+
"grad_norm": 0.350891649723053,
|
552 |
+
"learning_rate": 0.0002629541041482789,
|
553 |
+
"loss": 3.1999,
|
554 |
+
"step": 74000
|
555 |
+
},
|
556 |
+
{
|
557 |
+
"epoch": 4.0,
|
558 |
+
"eval_accuracy": 0.39521783897656926,
|
559 |
+
"eval_loss": 3.474266767501831,
|
560 |
+
"eval_runtime": 154.7996,
|
561 |
+
"eval_samples_per_second": 374.168,
|
562 |
+
"eval_steps_per_second": 5.853,
|
563 |
+
"step": 74380
|
564 |
+
},
|
565 |
+
{
|
566 |
+
"epoch": 4.033342296316214,
|
567 |
+
"grad_norm": 0.3689132034778595,
|
568 |
+
"learning_rate": 0.0002620714916151809,
|
569 |
+
"loss": 3.1644,
|
570 |
+
"step": 75000
|
571 |
+
},
|
572 |
+
{
|
573 |
+
"epoch": 4.08712019360043,
|
574 |
+
"grad_norm": 0.3693839907646179,
|
575 |
+
"learning_rate": 0.00026118887908208293,
|
576 |
+
"loss": 3.1393,
|
577 |
+
"step": 76000
|
578 |
+
},
|
579 |
+
{
|
580 |
+
"epoch": 4.140898090884646,
|
581 |
+
"grad_norm": 0.3670288622379303,
|
582 |
+
"learning_rate": 0.000260306266548985,
|
583 |
+
"loss": 3.1455,
|
584 |
+
"step": 77000
|
585 |
+
},
|
586 |
+
{
|
587 |
+
"epoch": 4.1946759881688624,
|
588 |
+
"grad_norm": 0.3569948673248291,
|
589 |
+
"learning_rate": 0.0002594245366284201,
|
590 |
+
"loss": 3.1473,
|
591 |
+
"step": 78000
|
592 |
+
},
|
593 |
+
{
|
594 |
+
"epoch": 4.2484538854530784,
|
595 |
+
"grad_norm": 0.37387335300445557,
|
596 |
+
"learning_rate": 0.0002585419240953221,
|
597 |
+
"loss": 3.144,
|
598 |
+
"step": 79000
|
599 |
+
},
|
600 |
+
{
|
601 |
+
"epoch": 4.302231782737295,
|
602 |
+
"grad_norm": 0.35249531269073486,
|
603 |
+
"learning_rate": 0.00025765931156222413,
|
604 |
+
"loss": 3.1482,
|
605 |
+
"step": 80000
|
606 |
+
},
|
607 |
+
{
|
608 |
+
"epoch": 4.356009680021511,
|
609 |
+
"grad_norm": 0.3619791269302368,
|
610 |
+
"learning_rate": 0.0002567775816416593,
|
611 |
+
"loss": 3.1468,
|
612 |
+
"step": 81000
|
613 |
+
},
|
614 |
+
{
|
615 |
+
"epoch": 4.409787577305727,
|
616 |
+
"grad_norm": 0.3486943542957306,
|
617 |
+
"learning_rate": 0.00025589585172109444,
|
618 |
+
"loss": 3.1543,
|
619 |
+
"step": 82000
|
620 |
+
},
|
621 |
+
{
|
622 |
+
"epoch": 4.463565474589943,
|
623 |
+
"grad_norm": 0.3588618040084839,
|
624 |
+
"learning_rate": 0.00025501323918799646,
|
625 |
+
"loss": 3.1509,
|
626 |
+
"step": 83000
|
627 |
+
},
|
628 |
+
{
|
629 |
+
"epoch": 4.517343371874159,
|
630 |
+
"grad_norm": 0.36453118920326233,
|
631 |
+
"learning_rate": 0.00025413062665489847,
|
632 |
+
"loss": 3.1487,
|
633 |
+
"step": 84000
|
634 |
+
},
|
635 |
+
{
|
636 |
+
"epoch": 4.571121269158376,
|
637 |
+
"grad_norm": 0.3435422480106354,
|
638 |
+
"learning_rate": 0.0002532480141218005,
|
639 |
+
"loss": 3.1485,
|
640 |
+
"step": 85000
|
641 |
+
},
|
642 |
+
{
|
643 |
+
"epoch": 4.624899166442592,
|
644 |
+
"grad_norm": 0.3634852468967438,
|
645 |
+
"learning_rate": 0.00025236540158870256,
|
646 |
+
"loss": 3.1543,
|
647 |
+
"step": 86000
|
648 |
+
},
|
649 |
+
{
|
650 |
+
"epoch": 4.678677063726808,
|
651 |
+
"grad_norm": 0.3428556025028229,
|
652 |
+
"learning_rate": 0.00025148367166813766,
|
653 |
+
"loss": 3.1482,
|
654 |
+
"step": 87000
|
655 |
+
},
|
656 |
+
{
|
657 |
+
"epoch": 4.732454961011024,
|
658 |
+
"grad_norm": 0.36531776189804077,
|
659 |
+
"learning_rate": 0.0002506010591350397,
|
660 |
+
"loss": 3.1467,
|
661 |
+
"step": 88000
|
662 |
+
},
|
663 |
+
{
|
664 |
+
"epoch": 4.786232858295241,
|
665 |
+
"grad_norm": 0.34510383009910583,
|
666 |
+
"learning_rate": 0.00024971844660194174,
|
667 |
+
"loss": 3.1497,
|
668 |
+
"step": 89000
|
669 |
+
},
|
670 |
+
{
|
671 |
+
"epoch": 4.840010755579457,
|
672 |
+
"grad_norm": 0.33938872814178467,
|
673 |
+
"learning_rate": 0.00024883583406884376,
|
674 |
+
"loss": 3.1455,
|
675 |
+
"step": 90000
|
676 |
+
},
|
677 |
+
{
|
678 |
+
"epoch": 4.893788652863673,
|
679 |
+
"grad_norm": 0.373588502407074,
|
680 |
+
"learning_rate": 0.000247954986760812,
|
681 |
+
"loss": 3.1474,
|
682 |
+
"step": 91000
|
683 |
+
},
|
684 |
+
{
|
685 |
+
"epoch": 4.947566550147889,
|
686 |
+
"grad_norm": 0.34171566367149353,
|
687 |
+
"learning_rate": 0.000247072374227714,
|
688 |
+
"loss": 3.1448,
|
689 |
+
"step": 92000
|
690 |
+
},
|
691 |
+
{
|
692 |
+
"epoch": 5.0,
|
693 |
+
"eval_accuracy": 0.39957591505456547,
|
694 |
+
"eval_loss": 3.4088761806488037,
|
695 |
+
"eval_runtime": 155.1687,
|
696 |
+
"eval_samples_per_second": 373.278,
|
697 |
+
"eval_steps_per_second": 5.839,
|
698 |
+
"step": 92975
|
699 |
+
},
|
700 |
+
{
|
701 |
+
"epoch": 5.001344447432105,
|
702 |
+
"grad_norm": 0.3516329526901245,
|
703 |
+
"learning_rate": 0.00024619064430714916,
|
704 |
+
"loss": 3.1468,
|
705 |
+
"step": 93000
|
706 |
+
},
|
707 |
+
{
|
708 |
+
"epoch": 5.055122344716321,
|
709 |
+
"grad_norm": 0.3526703715324402,
|
710 |
+
"learning_rate": 0.0002453080317740512,
|
711 |
+
"loss": 3.0784,
|
712 |
+
"step": 94000
|
713 |
+
},
|
714 |
+
{
|
715 |
+
"epoch": 5.108900242000538,
|
716 |
+
"grad_norm": 0.3785659670829773,
|
717 |
+
"learning_rate": 0.0002444254192409532,
|
718 |
+
"loss": 3.0804,
|
719 |
+
"step": 95000
|
720 |
+
},
|
721 |
+
{
|
722 |
+
"epoch": 5.162678139284754,
|
723 |
+
"grad_norm": 0.3553713262081146,
|
724 |
+
"learning_rate": 0.00024354280670785524,
|
725 |
+
"loss": 3.0885,
|
726 |
+
"step": 96000
|
727 |
+
},
|
728 |
+
{
|
729 |
+
"epoch": 5.21645603656897,
|
730 |
+
"grad_norm": 0.3591731786727905,
|
731 |
+
"learning_rate": 0.00024266107678729037,
|
732 |
+
"loss": 3.0916,
|
733 |
+
"step": 97000
|
734 |
+
},
|
735 |
+
{
|
736 |
+
"epoch": 5.270233933853186,
|
737 |
+
"grad_norm": 0.3515339195728302,
|
738 |
+
"learning_rate": 0.00024177846425419238,
|
739 |
+
"loss": 3.0929,
|
740 |
+
"step": 98000
|
741 |
+
},
|
742 |
+
{
|
743 |
+
"epoch": 5.324011831137403,
|
744 |
+
"grad_norm": 0.3388463854789734,
|
745 |
+
"learning_rate": 0.0002408958517210944,
|
746 |
+
"loss": 3.0946,
|
747 |
+
"step": 99000
|
748 |
+
},
|
749 |
+
{
|
750 |
+
"epoch": 5.377789728421619,
|
751 |
+
"grad_norm": 0.35906437039375305,
|
752 |
+
"learning_rate": 0.00024001500441306263,
|
753 |
+
"loss": 3.0936,
|
754 |
+
"step": 100000
|
755 |
+
},
|
756 |
+
{
|
757 |
+
"epoch": 5.431567625705835,
|
758 |
+
"grad_norm": 0.35187384486198425,
|
759 |
+
"learning_rate": 0.00023913239187996468,
|
760 |
+
"loss": 3.0965,
|
761 |
+
"step": 101000
|
762 |
+
},
|
763 |
+
{
|
764 |
+
"epoch": 5.485345522990051,
|
765 |
+
"grad_norm": 0.3588041067123413,
|
766 |
+
"learning_rate": 0.00023824977934686672,
|
767 |
+
"loss": 3.0954,
|
768 |
+
"step": 102000
|
769 |
+
},
|
770 |
+
{
|
771 |
+
"epoch": 5.539123420274267,
|
772 |
+
"grad_norm": 0.37199875712394714,
|
773 |
+
"learning_rate": 0.00023736716681376874,
|
774 |
+
"loss": 3.0958,
|
775 |
+
"step": 103000
|
776 |
+
},
|
777 |
+
{
|
778 |
+
"epoch": 5.592901317558484,
|
779 |
+
"grad_norm": 0.36916086077690125,
|
780 |
+
"learning_rate": 0.00023648543689320386,
|
781 |
+
"loss": 3.0974,
|
782 |
+
"step": 104000
|
783 |
+
},
|
784 |
+
{
|
785 |
+
"epoch": 5.6466792148427,
|
786 |
+
"grad_norm": 0.3353819251060486,
|
787 |
+
"learning_rate": 0.0002356028243601059,
|
788 |
+
"loss": 3.1009,
|
789 |
+
"step": 105000
|
790 |
+
},
|
791 |
+
{
|
792 |
+
"epoch": 5.700457112126916,
|
793 |
+
"grad_norm": 0.35051777958869934,
|
794 |
+
"learning_rate": 0.00023472021182700792,
|
795 |
+
"loss": 3.1009,
|
796 |
+
"step": 106000
|
797 |
+
},
|
798 |
+
{
|
799 |
+
"epoch": 5.754235009411132,
|
800 |
+
"grad_norm": 0.36077558994293213,
|
801 |
+
"learning_rate": 0.00023383848190644305,
|
802 |
+
"loss": 3.0988,
|
803 |
+
"step": 107000
|
804 |
+
},
|
805 |
+
{
|
806 |
+
"epoch": 5.808012906695348,
|
807 |
+
"grad_norm": 0.3492220640182495,
|
808 |
+
"learning_rate": 0.0002329558693733451,
|
809 |
+
"loss": 3.1008,
|
810 |
+
"step": 108000
|
811 |
+
},
|
812 |
+
{
|
813 |
+
"epoch": 5.861790803979565,
|
814 |
+
"grad_norm": 0.3349301517009735,
|
815 |
+
"learning_rate": 0.00023207325684024714,
|
816 |
+
"loss": 3.1004,
|
817 |
+
"step": 109000
|
818 |
+
},
|
819 |
+
{
|
820 |
+
"epoch": 5.915568701263781,
|
821 |
+
"grad_norm": 0.3572138249874115,
|
822 |
+
"learning_rate": 0.00023119152691968223,
|
823 |
+
"loss": 3.0991,
|
824 |
+
"step": 110000
|
825 |
+
},
|
826 |
+
{
|
827 |
+
"epoch": 5.969346598547997,
|
828 |
+
"grad_norm": 0.3292669653892517,
|
829 |
+
"learning_rate": 0.00023030891438658428,
|
830 |
+
"loss": 3.0986,
|
831 |
+
"step": 111000
|
832 |
+
},
|
833 |
+
{
|
834 |
+
"epoch": 6.0,
|
835 |
+
"eval_accuracy": 0.4021510044171932,
|
836 |
+
"eval_loss": 3.4165823459625244,
|
837 |
+
"eval_runtime": 153.9568,
|
838 |
+
"eval_samples_per_second": 376.216,
|
839 |
+
"eval_steps_per_second": 5.885,
|
840 |
+
"step": 111570
|
841 |
+
},
|
842 |
+
{
|
843 |
+
"epoch": 6.023124495832213,
|
844 |
+
"grad_norm": 0.3583333492279053,
|
845 |
+
"learning_rate": 0.0002294263018534863,
|
846 |
+
"loss": 3.0671,
|
847 |
+
"step": 112000
|
848 |
+
},
|
849 |
+
{
|
850 |
+
"epoch": 6.076902393116429,
|
851 |
+
"grad_norm": 0.3764239251613617,
|
852 |
+
"learning_rate": 0.00022854368932038834,
|
853 |
+
"loss": 3.035,
|
854 |
+
"step": 113000
|
855 |
+
},
|
856 |
+
{
|
857 |
+
"epoch": 6.130680290400646,
|
858 |
+
"grad_norm": 0.3759283721446991,
|
859 |
+
"learning_rate": 0.00022766195939982344,
|
860 |
+
"loss": 3.034,
|
861 |
+
"step": 114000
|
862 |
+
},
|
863 |
+
{
|
864 |
+
"epoch": 6.184458187684862,
|
865 |
+
"grad_norm": 0.3840562701225281,
|
866 |
+
"learning_rate": 0.0002267802294792586,
|
867 |
+
"loss": 3.0451,
|
868 |
+
"step": 115000
|
869 |
+
},
|
870 |
+
{
|
871 |
+
"epoch": 6.238236084969078,
|
872 |
+
"grad_norm": 0.35660070180892944,
|
873 |
+
"learning_rate": 0.00022589761694616063,
|
874 |
+
"loss": 3.0489,
|
875 |
+
"step": 116000
|
876 |
+
},
|
877 |
+
{
|
878 |
+
"epoch": 6.292013982253294,
|
879 |
+
"grad_norm": 0.3423716723918915,
|
880 |
+
"learning_rate": 0.00022501500441306265,
|
881 |
+
"loss": 3.0487,
|
882 |
+
"step": 117000
|
883 |
+
},
|
884 |
+
{
|
885 |
+
"epoch": 6.34579187953751,
|
886 |
+
"grad_norm": 0.37959685921669006,
|
887 |
+
"learning_rate": 0.00022413327449249777,
|
888 |
+
"loss": 3.0516,
|
889 |
+
"step": 118000
|
890 |
+
},
|
891 |
+
{
|
892 |
+
"epoch": 6.399569776821727,
|
893 |
+
"grad_norm": 0.3404131531715393,
|
894 |
+
"learning_rate": 0.0002232506619593998,
|
895 |
+
"loss": 3.0495,
|
896 |
+
"step": 119000
|
897 |
+
},
|
898 |
+
{
|
899 |
+
"epoch": 6.453347674105943,
|
900 |
+
"grad_norm": 0.34124448895454407,
|
901 |
+
"learning_rate": 0.00022236804942630183,
|
902 |
+
"loss": 3.0556,
|
903 |
+
"step": 120000
|
904 |
+
},
|
905 |
+
{
|
906 |
+
"epoch": 6.507125571390159,
|
907 |
+
"grad_norm": 0.366603821516037,
|
908 |
+
"learning_rate": 0.00022148543689320385,
|
909 |
+
"loss": 3.0585,
|
910 |
+
"step": 121000
|
911 |
+
},
|
912 |
+
{
|
913 |
+
"epoch": 6.560903468674375,
|
914 |
+
"grad_norm": 0.3521302342414856,
|
915 |
+
"learning_rate": 0.00022060370697263897,
|
916 |
+
"loss": 3.0594,
|
917 |
+
"step": 122000
|
918 |
+
},
|
919 |
+
{
|
920 |
+
"epoch": 6.614681365958591,
|
921 |
+
"grad_norm": 0.3434734642505646,
|
922 |
+
"learning_rate": 0.00021972109443954102,
|
923 |
+
"loss": 3.0561,
|
924 |
+
"step": 123000
|
925 |
+
},
|
926 |
+
{
|
927 |
+
"epoch": 6.6684592632428075,
|
928 |
+
"grad_norm": 0.3365206718444824,
|
929 |
+
"learning_rate": 0.00021883936451897614,
|
930 |
+
"loss": 3.0577,
|
931 |
+
"step": 124000
|
932 |
+
},
|
933 |
+
{
|
934 |
+
"epoch": 6.7222371605270235,
|
935 |
+
"grad_norm": 0.34190618991851807,
|
936 |
+
"learning_rate": 0.0002179576345984113,
|
937 |
+
"loss": 3.0585,
|
938 |
+
"step": 125000
|
939 |
+
},
|
940 |
+
{
|
941 |
+
"epoch": 6.7760150578112395,
|
942 |
+
"grad_norm": 0.3417965769767761,
|
943 |
+
"learning_rate": 0.0002170759046778464,
|
944 |
+
"loss": 3.0599,
|
945 |
+
"step": 126000
|
946 |
+
},
|
947 |
+
{
|
948 |
+
"epoch": 6.8297929550954555,
|
949 |
+
"grad_norm": 0.3413834571838379,
|
950 |
+
"learning_rate": 0.00021619329214474844,
|
951 |
+
"loss": 3.0575,
|
952 |
+
"step": 127000
|
953 |
+
},
|
954 |
+
{
|
955 |
+
"epoch": 6.8835708523796715,
|
956 |
+
"grad_norm": 0.3608095347881317,
|
957 |
+
"learning_rate": 0.00021531067961165048,
|
958 |
+
"loss": 3.0588,
|
959 |
+
"step": 128000
|
960 |
+
},
|
961 |
+
{
|
962 |
+
"epoch": 6.937348749663888,
|
963 |
+
"grad_norm": 0.360249400138855,
|
964 |
+
"learning_rate": 0.00021442806707855253,
|
965 |
+
"loss": 3.0598,
|
966 |
+
"step": 129000
|
967 |
+
},
|
968 |
+
{
|
969 |
+
"epoch": 6.991126646948104,
|
970 |
+
"grad_norm": 0.3597511351108551,
|
971 |
+
"learning_rate": 0.00021354633715798762,
|
972 |
+
"loss": 3.0592,
|
973 |
+
"step": 130000
|
974 |
+
},
|
975 |
+
{
|
976 |
+
"epoch": 7.0,
|
977 |
+
"eval_accuracy": 0.40360871244389834,
|
978 |
+
"eval_loss": 3.403170108795166,
|
979 |
+
"eval_runtime": 153.6701,
|
980 |
+
"eval_samples_per_second": 376.918,
|
981 |
+
"eval_steps_per_second": 5.896,
|
982 |
+
"step": 130165
|
983 |
+
},
|
984 |
+
{
|
985 |
+
"epoch": 7.04490454423232,
|
986 |
+
"grad_norm": 0.364961177110672,
|
987 |
+
"learning_rate": 0.00021266372462488964,
|
988 |
+
"loss": 3.0023,
|
989 |
+
"step": 131000
|
990 |
+
},
|
991 |
+
{
|
992 |
+
"epoch": 7.098682441516536,
|
993 |
+
"grad_norm": 0.3606811463832855,
|
994 |
+
"learning_rate": 0.00021178111209179168,
|
995 |
+
"loss": 2.9945,
|
996 |
+
"step": 132000
|
997 |
+
},
|
998 |
+
{
|
999 |
+
"epoch": 7.152460338800753,
|
1000 |
+
"grad_norm": 0.3595397472381592,
|
1001 |
+
"learning_rate": 0.0002108984995586937,
|
1002 |
+
"loss": 3.0056,
|
1003 |
+
"step": 133000
|
1004 |
+
},
|
1005 |
+
{
|
1006 |
+
"epoch": 7.206238236084969,
|
1007 |
+
"grad_norm": 0.35375916957855225,
|
1008 |
+
"learning_rate": 0.00021001676963812883,
|
1009 |
+
"loss": 3.0093,
|
1010 |
+
"step": 134000
|
1011 |
+
},
|
1012 |
+
{
|
1013 |
+
"epoch": 7.260016133369185,
|
1014 |
+
"grad_norm": 0.3839136064052582,
|
1015 |
+
"learning_rate": 0.00020913415710503087,
|
1016 |
+
"loss": 3.0092,
|
1017 |
+
"step": 135000
|
1018 |
+
},
|
1019 |
+
{
|
1020 |
+
"epoch": 7.313794030653401,
|
1021 |
+
"grad_norm": 0.36542221903800964,
|
1022 |
+
"learning_rate": 0.0002082515445719329,
|
1023 |
+
"loss": 3.0137,
|
1024 |
+
"step": 136000
|
1025 |
+
},
|
1026 |
+
{
|
1027 |
+
"epoch": 7.367571927937617,
|
1028 |
+
"grad_norm": 0.3783646523952484,
|
1029 |
+
"learning_rate": 0.00020736893203883493,
|
1030 |
+
"loss": 3.0178,
|
1031 |
+
"step": 137000
|
1032 |
+
},
|
1033 |
+
{
|
1034 |
+
"epoch": 7.421349825221834,
|
1035 |
+
"grad_norm": 0.3594056963920593,
|
1036 |
+
"learning_rate": 0.00020648720211827005,
|
1037 |
+
"loss": 3.0204,
|
1038 |
+
"step": 138000
|
1039 |
+
},
|
1040 |
+
{
|
1041 |
+
"epoch": 7.47512772250605,
|
1042 |
+
"grad_norm": 0.3637217581272125,
|
1043 |
+
"learning_rate": 0.0002056045895851721,
|
1044 |
+
"loss": 3.0189,
|
1045 |
+
"step": 139000
|
1046 |
+
},
|
1047 |
+
{
|
1048 |
+
"epoch": 7.528905619790266,
|
1049 |
+
"grad_norm": 0.36873236298561096,
|
1050 |
+
"learning_rate": 0.00020472197705207414,
|
1051 |
+
"loss": 3.0193,
|
1052 |
+
"step": 140000
|
1053 |
+
},
|
1054 |
+
{
|
1055 |
+
"epoch": 7.582683517074482,
|
1056 |
+
"grad_norm": 0.3408534824848175,
|
1057 |
+
"learning_rate": 0.00020384024713150924,
|
1058 |
+
"loss": 3.0235,
|
1059 |
+
"step": 141000
|
1060 |
+
},
|
1061 |
+
{
|
1062 |
+
"epoch": 7.636461414358699,
|
1063 |
+
"grad_norm": 0.34803324937820435,
|
1064 |
+
"learning_rate": 0.00020295763459841128,
|
1065 |
+
"loss": 3.0237,
|
1066 |
+
"step": 142000
|
1067 |
+
},
|
1068 |
+
{
|
1069 |
+
"epoch": 7.690239311642915,
|
1070 |
+
"grad_norm": 0.3650054931640625,
|
1071 |
+
"learning_rate": 0.0002020759046778464,
|
1072 |
+
"loss": 3.0214,
|
1073 |
+
"step": 143000
|
1074 |
+
},
|
1075 |
+
{
|
1076 |
+
"epoch": 7.744017208927131,
|
1077 |
+
"grad_norm": 0.3592528998851776,
|
1078 |
+
"learning_rate": 0.00020119329214474845,
|
1079 |
+
"loss": 3.0261,
|
1080 |
+
"step": 144000
|
1081 |
+
},
|
1082 |
+
{
|
1083 |
+
"epoch": 7.797795106211347,
|
1084 |
+
"grad_norm": 0.37202954292297363,
|
1085 |
+
"learning_rate": 0.00020031067961165047,
|
1086 |
+
"loss": 3.0253,
|
1087 |
+
"step": 145000
|
1088 |
+
},
|
1089 |
+
{
|
1090 |
+
"epoch": 7.851573003495563,
|
1091 |
+
"grad_norm": 0.35640275478363037,
|
1092 |
+
"learning_rate": 0.0001994280670785525,
|
1093 |
+
"loss": 3.027,
|
1094 |
+
"step": 146000
|
1095 |
+
},
|
1096 |
+
{
|
1097 |
+
"epoch": 7.90535090077978,
|
1098 |
+
"grad_norm": 0.34525027871131897,
|
1099 |
+
"learning_rate": 0.00019854633715798764,
|
1100 |
+
"loss": 3.0276,
|
1101 |
+
"step": 147000
|
1102 |
+
},
|
1103 |
+
{
|
1104 |
+
"epoch": 7.959128798063996,
|
1105 |
+
"grad_norm": 0.33090052008628845,
|
1106 |
+
"learning_rate": 0.00019766372462488965,
|
1107 |
+
"loss": 3.0279,
|
1108 |
+
"step": 148000
|
1109 |
+
},
|
1110 |
+
{
|
1111 |
+
"epoch": 8.0,
|
1112 |
+
"eval_accuracy": 0.4061836674493643,
|
1113 |
+
"eval_loss": 3.3745980262756348,
|
1114 |
+
"eval_runtime": 155.0808,
|
1115 |
+
"eval_samples_per_second": 373.489,
|
1116 |
+
"eval_steps_per_second": 5.842,
|
1117 |
+
"step": 148760
|
1118 |
+
},
|
1119 |
+
{
|
1120 |
+
"epoch": 8.012906695348212,
|
1121 |
+
"grad_norm": 0.3575202524662018,
|
1122 |
+
"learning_rate": 0.00019678199470432478,
|
1123 |
+
"loss": 3.0055,
|
1124 |
+
"step": 149000
|
1125 |
+
},
|
1126 |
+
{
|
1127 |
+
"epoch": 8.066684592632429,
|
1128 |
+
"grad_norm": 0.3376094400882721,
|
1129 |
+
"learning_rate": 0.00019589938217122682,
|
1130 |
+
"loss": 2.9616,
|
1131 |
+
"step": 150000
|
1132 |
+
},
|
1133 |
+
{
|
1134 |
+
"epoch": 8.120462489916644,
|
1135 |
+
"grad_norm": 0.36086979508399963,
|
1136 |
+
"learning_rate": 0.00019501676963812887,
|
1137 |
+
"loss": 2.9642,
|
1138 |
+
"step": 151000
|
1139 |
+
},
|
1140 |
+
{
|
1141 |
+
"epoch": 8.17424038720086,
|
1142 |
+
"grad_norm": 0.36575761437416077,
|
1143 |
+
"learning_rate": 0.00019413503971756397,
|
1144 |
+
"loss": 2.9755,
|
1145 |
+
"step": 152000
|
1146 |
+
},
|
1147 |
+
{
|
1148 |
+
"epoch": 8.228018284485076,
|
1149 |
+
"grad_norm": 0.35903316736221313,
|
1150 |
+
"learning_rate": 0.000193252427184466,
|
1151 |
+
"loss": 2.9758,
|
1152 |
+
"step": 153000
|
1153 |
+
},
|
1154 |
+
{
|
1155 |
+
"epoch": 8.281796181769293,
|
1156 |
+
"grad_norm": 0.3707205653190613,
|
1157 |
+
"learning_rate": 0.00019237157987643422,
|
1158 |
+
"loss": 2.9812,
|
1159 |
+
"step": 154000
|
1160 |
+
},
|
1161 |
+
{
|
1162 |
+
"epoch": 8.33557407905351,
|
1163 |
+
"grad_norm": 0.3687371611595154,
|
1164 |
+
"learning_rate": 0.00019148896734333626,
|
1165 |
+
"loss": 2.9833,
|
1166 |
+
"step": 155000
|
1167 |
+
},
|
1168 |
+
{
|
1169 |
+
"epoch": 8.389351976337725,
|
1170 |
+
"grad_norm": 0.3716333508491516,
|
1171 |
+
"learning_rate": 0.0001906063548102383,
|
1172 |
+
"loss": 2.988,
|
1173 |
+
"step": 156000
|
1174 |
+
},
|
1175 |
+
{
|
1176 |
+
"epoch": 8.443129873621942,
|
1177 |
+
"grad_norm": 0.3598574101924896,
|
1178 |
+
"learning_rate": 0.00018972374227714032,
|
1179 |
+
"loss": 2.9878,
|
1180 |
+
"step": 157000
|
1181 |
+
},
|
1182 |
+
{
|
1183 |
+
"epoch": 8.496907770906157,
|
1184 |
+
"grad_norm": 0.34485259652137756,
|
1185 |
+
"learning_rate": 0.00018884112974404236,
|
1186 |
+
"loss": 2.9891,
|
1187 |
+
"step": 158000
|
1188 |
+
},
|
1189 |
+
{
|
1190 |
+
"epoch": 8.550685668190374,
|
1191 |
+
"grad_norm": 0.3574945628643036,
|
1192 |
+
"learning_rate": 0.00018795851721094438,
|
1193 |
+
"loss": 2.9923,
|
1194 |
+
"step": 159000
|
1195 |
+
},
|
1196 |
+
{
|
1197 |
+
"epoch": 8.60446356547459,
|
1198 |
+
"grad_norm": 0.3579365611076355,
|
1199 |
+
"learning_rate": 0.0001870767872903795,
|
1200 |
+
"loss": 2.9908,
|
1201 |
+
"step": 160000
|
1202 |
+
},
|
1203 |
+
{
|
1204 |
+
"epoch": 8.658241462758806,
|
1205 |
+
"grad_norm": 0.3805078864097595,
|
1206 |
+
"learning_rate": 0.00018619505736981463,
|
1207 |
+
"loss": 2.9896,
|
1208 |
+
"step": 161000
|
1209 |
+
},
|
1210 |
+
{
|
1211 |
+
"epoch": 8.712019360043023,
|
1212 |
+
"grad_norm": 0.3955886960029602,
|
1213 |
+
"learning_rate": 0.00018531244483671667,
|
1214 |
+
"loss": 2.9956,
|
1215 |
+
"step": 162000
|
1216 |
+
},
|
1217 |
+
{
|
1218 |
+
"epoch": 8.765797257327238,
|
1219 |
+
"grad_norm": 0.35813698172569275,
|
1220 |
+
"learning_rate": 0.00018442983230361872,
|
1221 |
+
"loss": 2.9987,
|
1222 |
+
"step": 163000
|
1223 |
+
},
|
1224 |
+
{
|
1225 |
+
"epoch": 8.819575154611455,
|
1226 |
+
"grad_norm": 0.3720499277114868,
|
1227 |
+
"learning_rate": 0.0001835472197705207,
|
1228 |
+
"loss": 2.9956,
|
1229 |
+
"step": 164000
|
1230 |
+
},
|
1231 |
+
{
|
1232 |
+
"epoch": 8.873353051895672,
|
1233 |
+
"grad_norm": 0.35912513732910156,
|
1234 |
+
"learning_rate": 0.00018266548984995586,
|
1235 |
+
"loss": 2.9985,
|
1236 |
+
"step": 165000
|
1237 |
+
},
|
1238 |
+
{
|
1239 |
+
"epoch": 8.927130949179887,
|
1240 |
+
"grad_norm": 0.36929354071617126,
|
1241 |
+
"learning_rate": 0.00018178287731685788,
|
1242 |
+
"loss": 2.9972,
|
1243 |
+
"step": 166000
|
1244 |
+
},
|
1245 |
+
{
|
1246 |
+
"epoch": 8.980908846464104,
|
1247 |
+
"grad_norm": 0.34516987204551697,
|
1248 |
+
"learning_rate": 0.0001809020300088261,
|
1249 |
+
"loss": 2.9977,
|
1250 |
+
"step": 167000
|
1251 |
+
},
|
1252 |
+
{
|
1253 |
+
"epoch": 9.0,
|
1254 |
+
"eval_accuracy": 0.4067501172434183,
|
1255 |
+
"eval_loss": 3.3709380626678467,
|
1256 |
+
"eval_runtime": 154.3682,
|
1257 |
+
"eval_samples_per_second": 375.213,
|
1258 |
+
"eval_steps_per_second": 5.869,
|
1259 |
+
"step": 167355
|
1260 |
+
},
|
1261 |
+
{
|
1262 |
+
"epoch": 9.034686743748319,
|
1263 |
+
"grad_norm": 0.38715243339538574,
|
1264 |
+
"learning_rate": 0.00018001941747572815,
|
1265 |
+
"loss": 2.9529,
|
1266 |
+
"step": 168000
|
1267 |
+
},
|
1268 |
+
{
|
1269 |
+
"epoch": 9.088464641032536,
|
1270 |
+
"grad_norm": 0.39728203415870667,
|
1271 |
+
"learning_rate": 0.00017913680494263017,
|
1272 |
+
"loss": 2.9363,
|
1273 |
+
"step": 169000
|
1274 |
+
},
|
1275 |
+
{
|
1276 |
+
"epoch": 9.142242538316752,
|
1277 |
+
"grad_norm": 0.3717023730278015,
|
1278 |
+
"learning_rate": 0.00017825419240953221,
|
1279 |
+
"loss": 2.9421,
|
1280 |
+
"step": 170000
|
1281 |
+
},
|
1282 |
+
{
|
1283 |
+
"epoch": 9.196020435600968,
|
1284 |
+
"grad_norm": 0.3885519504547119,
|
1285 |
+
"learning_rate": 0.00017737157987643426,
|
1286 |
+
"loss": 2.9453,
|
1287 |
+
"step": 171000
|
1288 |
+
},
|
1289 |
+
{
|
1290 |
+
"epoch": 9.249798332885184,
|
1291 |
+
"grad_norm": 0.3886348307132721,
|
1292 |
+
"learning_rate": 0.00017648896734333625,
|
1293 |
+
"loss": 2.9489,
|
1294 |
+
"step": 172000
|
1295 |
+
},
|
1296 |
+
{
|
1297 |
+
"epoch": 9.3035762301694,
|
1298 |
+
"grad_norm": 0.3586263656616211,
|
1299 |
+
"learning_rate": 0.00017560723742277137,
|
1300 |
+
"loss": 2.9544,
|
1301 |
+
"step": 173000
|
1302 |
+
},
|
1303 |
+
{
|
1304 |
+
"epoch": 9.357354127453616,
|
1305 |
+
"grad_norm": 0.37699151039123535,
|
1306 |
+
"learning_rate": 0.00017472462488967342,
|
1307 |
+
"loss": 2.9584,
|
1308 |
+
"step": 174000
|
1309 |
+
},
|
1310 |
+
{
|
1311 |
+
"epoch": 9.411132024737833,
|
1312 |
+
"grad_norm": 0.36616960167884827,
|
1313 |
+
"learning_rate": 0.00017384289496910857,
|
1314 |
+
"loss": 2.9624,
|
1315 |
+
"step": 175000
|
1316 |
+
},
|
1317 |
+
{
|
1318 |
+
"epoch": 9.464909922022049,
|
1319 |
+
"grad_norm": 0.36951467394828796,
|
1320 |
+
"learning_rate": 0.00017296028243601056,
|
1321 |
+
"loss": 2.9585,
|
1322 |
+
"step": 176000
|
1323 |
+
},
|
1324 |
+
{
|
1325 |
+
"epoch": 9.518687819306265,
|
1326 |
+
"grad_norm": 0.36193570494651794,
|
1327 |
+
"learning_rate": 0.0001720776699029126,
|
1328 |
+
"loss": 2.9649,
|
1329 |
+
"step": 177000
|
1330 |
+
},
|
1331 |
+
{
|
1332 |
+
"epoch": 9.57246571659048,
|
1333 |
+
"grad_norm": 0.3632647395133972,
|
1334 |
+
"learning_rate": 0.00017119593998234775,
|
1335 |
+
"loss": 2.9573,
|
1336 |
+
"step": 178000
|
1337 |
+
},
|
1338 |
+
{
|
1339 |
+
"epoch": 9.626243613874697,
|
1340 |
+
"grad_norm": 0.3749377727508545,
|
1341 |
+
"learning_rate": 0.00017031332744924974,
|
1342 |
+
"loss": 2.9642,
|
1343 |
+
"step": 179000
|
1344 |
+
},
|
1345 |
+
{
|
1346 |
+
"epoch": 9.680021511158914,
|
1347 |
+
"grad_norm": 0.3834667205810547,
|
1348 |
+
"learning_rate": 0.00016943159752868487,
|
1349 |
+
"loss": 2.9663,
|
1350 |
+
"step": 180000
|
1351 |
+
},
|
1352 |
+
{
|
1353 |
+
"epoch": 9.73379940844313,
|
1354 |
+
"grad_norm": 0.3825732171535492,
|
1355 |
+
"learning_rate": 0.0001685489849955869,
|
1356 |
+
"loss": 2.9665,
|
1357 |
+
"step": 181000
|
1358 |
+
},
|
1359 |
+
{
|
1360 |
+
"epoch": 9.787577305727346,
|
1361 |
+
"grad_norm": 0.3804017901420593,
|
1362 |
+
"learning_rate": 0.00016766725507502206,
|
1363 |
+
"loss": 2.9656,
|
1364 |
+
"step": 182000
|
1365 |
+
},
|
1366 |
+
{
|
1367 |
+
"epoch": 9.841355203011563,
|
1368 |
+
"grad_norm": 0.34586480259895325,
|
1369 |
+
"learning_rate": 0.0001667846425419241,
|
1370 |
+
"loss": 2.9733,
|
1371 |
+
"step": 183000
|
1372 |
+
},
|
1373 |
+
{
|
1374 |
+
"epoch": 9.895133100295778,
|
1375 |
+
"grad_norm": 0.3566882312297821,
|
1376 |
+
"learning_rate": 0.0001659020300088261,
|
1377 |
+
"loss": 2.9731,
|
1378 |
+
"step": 184000
|
1379 |
+
},
|
1380 |
+
{
|
1381 |
+
"epoch": 9.948910997579995,
|
1382 |
+
"grad_norm": 0.36795541644096375,
|
1383 |
+
"learning_rate": 0.00016502030008826125,
|
1384 |
+
"loss": 2.9761,
|
1385 |
+
"step": 185000
|
1386 |
+
},
|
1387 |
+
{
|
1388 |
+
"epoch": 10.0,
|
1389 |
+
"eval_accuracy": 0.4070907126485243,
|
1390 |
+
"eval_loss": 3.379540205001831,
|
1391 |
+
"eval_runtime": 153.7409,
|
1392 |
+
"eval_samples_per_second": 376.744,
|
1393 |
+
"eval_steps_per_second": 5.893,
|
1394 |
+
"step": 185950
|
1395 |
+
},
|
1396 |
+
{
|
1397 |
+
"epoch": 10.00268889486421,
|
1398 |
+
"grad_norm": 0.37020811438560486,
|
1399 |
+
"learning_rate": 0.00016413768755516327,
|
1400 |
+
"loss": 2.9715,
|
1401 |
+
"step": 186000
|
1402 |
+
},
|
1403 |
+
{
|
1404 |
+
"epoch": 10.056466792148427,
|
1405 |
+
"grad_norm": 0.36936092376708984,
|
1406 |
+
"learning_rate": 0.00016325507502206528,
|
1407 |
+
"loss": 2.905,
|
1408 |
+
"step": 187000
|
1409 |
+
},
|
1410 |
+
{
|
1411 |
+
"epoch": 10.110244689432642,
|
1412 |
+
"grad_norm": 0.3851975202560425,
|
1413 |
+
"learning_rate": 0.0001623733451015004,
|
1414 |
+
"loss": 2.9174,
|
1415 |
+
"step": 188000
|
1416 |
+
},
|
1417 |
+
{
|
1418 |
+
"epoch": 10.16402258671686,
|
1419 |
+
"grad_norm": 0.37237951159477234,
|
1420 |
+
"learning_rate": 0.00016149073256840245,
|
1421 |
+
"loss": 2.9218,
|
1422 |
+
"step": 189000
|
1423 |
+
},
|
1424 |
+
{
|
1425 |
+
"epoch": 10.217800484001076,
|
1426 |
+
"grad_norm": 0.39296218752861023,
|
1427 |
+
"learning_rate": 0.0001606081200353045,
|
1428 |
+
"loss": 2.9218,
|
1429 |
+
"step": 190000
|
1430 |
+
},
|
1431 |
+
{
|
1432 |
+
"epoch": 10.271578381285291,
|
1433 |
+
"grad_norm": 0.427022248506546,
|
1434 |
+
"learning_rate": 0.0001597255075022065,
|
1435 |
+
"loss": 2.9235,
|
1436 |
+
"step": 191000
|
1437 |
+
},
|
1438 |
+
{
|
1439 |
+
"epoch": 10.325356278569508,
|
1440 |
+
"grad_norm": 0.40037524700164795,
|
1441 |
+
"learning_rate": 0.00015884377758164164,
|
1442 |
+
"loss": 2.9287,
|
1443 |
+
"step": 192000
|
1444 |
+
},
|
1445 |
+
{
|
1446 |
+
"epoch": 10.379134175853725,
|
1447 |
+
"grad_norm": 0.37520840764045715,
|
1448 |
+
"learning_rate": 0.00015796116504854368,
|
1449 |
+
"loss": 2.9303,
|
1450 |
+
"step": 193000
|
1451 |
+
},
|
1452 |
+
{
|
1453 |
+
"epoch": 10.43291207313794,
|
1454 |
+
"grad_norm": 0.38916105031967163,
|
1455 |
+
"learning_rate": 0.00015707855251544572,
|
1456 |
+
"loss": 2.9352,
|
1457 |
+
"step": 194000
|
1458 |
+
},
|
1459 |
+
{
|
1460 |
+
"epoch": 10.486689970422157,
|
1461 |
+
"grad_norm": 0.36788520216941833,
|
1462 |
+
"learning_rate": 0.00015619682259488082,
|
1463 |
+
"loss": 2.9363,
|
1464 |
+
"step": 195000
|
1465 |
+
},
|
1466 |
+
{
|
1467 |
+
"epoch": 10.540467867706372,
|
1468 |
+
"grad_norm": 0.3871680796146393,
|
1469 |
+
"learning_rate": 0.00015531421006178287,
|
1470 |
+
"loss": 2.9393,
|
1471 |
+
"step": 196000
|
1472 |
+
},
|
1473 |
+
{
|
1474 |
+
"epoch": 10.594245764990589,
|
1475 |
+
"grad_norm": 0.3886531889438629,
|
1476 |
+
"learning_rate": 0.0001544333627537511,
|
1477 |
+
"loss": 2.938,
|
1478 |
+
"step": 197000
|
1479 |
+
},
|
1480 |
+
{
|
1481 |
+
"epoch": 10.648023662274806,
|
1482 |
+
"grad_norm": 0.3601588010787964,
|
1483 |
+
"learning_rate": 0.00015355075022065312,
|
1484 |
+
"loss": 2.9447,
|
1485 |
+
"step": 198000
|
1486 |
+
},
|
1487 |
+
{
|
1488 |
+
"epoch": 10.701801559559021,
|
1489 |
+
"grad_norm": 0.36893409490585327,
|
1490 |
+
"learning_rate": 0.00015266813768755513,
|
1491 |
+
"loss": 2.9398,
|
1492 |
+
"step": 199000
|
1493 |
+
},
|
1494 |
+
{
|
1495 |
+
"epoch": 10.755579456843238,
|
1496 |
+
"grad_norm": 0.3649713397026062,
|
1497 |
+
"learning_rate": 0.00015178552515445718,
|
1498 |
+
"loss": 2.9445,
|
1499 |
+
"step": 200000
|
1500 |
+
},
|
1501 |
+
{
|
1502 |
+
"epoch": 10.809357354127453,
|
1503 |
+
"grad_norm": 0.3681996762752533,
|
1504 |
+
"learning_rate": 0.0001509037952338923,
|
1505 |
+
"loss": 2.9483,
|
1506 |
+
"step": 201000
|
1507 |
+
},
|
1508 |
+
{
|
1509 |
+
"epoch": 10.86313525141167,
|
1510 |
+
"grad_norm": 0.38864666223526,
|
1511 |
+
"learning_rate": 0.00015002206531332745,
|
1512 |
+
"loss": 2.9449,
|
1513 |
+
"step": 202000
|
1514 |
+
},
|
1515 |
+
{
|
1516 |
+
"epoch": 10.916913148695887,
|
1517 |
+
"grad_norm": 0.3854575753211975,
|
1518 |
+
"learning_rate": 0.00014913945278022947,
|
1519 |
+
"loss": 2.95,
|
1520 |
+
"step": 203000
|
1521 |
+
},
|
1522 |
+
{
|
1523 |
+
"epoch": 10.970691045980102,
|
1524 |
+
"grad_norm": 0.3969317376613617,
|
1525 |
+
"learning_rate": 0.0001482568402471315,
|
1526 |
+
"loss": 2.9464,
|
1527 |
+
"step": 204000
|
1528 |
+
},
|
1529 |
+
{
|
1530 |
+
"epoch": 11.0,
|
1531 |
+
"eval_accuracy": 0.4079866062032567,
|
1532 |
+
"eval_loss": 3.378293991088867,
|
1533 |
+
"eval_runtime": 154.03,
|
1534 |
+
"eval_samples_per_second": 376.037,
|
1535 |
+
"eval_steps_per_second": 5.882,
|
1536 |
+
"step": 204545
|
1537 |
+
},
|
1538 |
+
{
|
1539 |
+
"epoch": 11.024468943264319,
|
1540 |
+
"grad_norm": 0.4027111232280731,
|
1541 |
+
"learning_rate": 0.0001473751103265666,
|
1542 |
+
"loss": 2.9169,
|
1543 |
+
"step": 205000
|
1544 |
+
},
|
1545 |
+
{
|
1546 |
+
"epoch": 11.078246840548534,
|
1547 |
+
"grad_norm": 0.38684701919555664,
|
1548 |
+
"learning_rate": 0.00014649249779346866,
|
1549 |
+
"loss": 2.8892,
|
1550 |
+
"step": 206000
|
1551 |
+
},
|
1552 |
+
{
|
1553 |
+
"epoch": 11.13202473783275,
|
1554 |
+
"grad_norm": 0.4004645049571991,
|
1555 |
+
"learning_rate": 0.00014561076787290378,
|
1556 |
+
"loss": 2.8914,
|
1557 |
+
"step": 207000
|
1558 |
+
},
|
1559 |
+
{
|
1560 |
+
"epoch": 11.185802635116968,
|
1561 |
+
"grad_norm": 0.39821696281433105,
|
1562 |
+
"learning_rate": 0.0001447281553398058,
|
1563 |
+
"loss": 2.8959,
|
1564 |
+
"step": 208000
|
1565 |
+
},
|
1566 |
+
{
|
1567 |
+
"epoch": 11.239580532401183,
|
1568 |
+
"grad_norm": 0.37264591455459595,
|
1569 |
+
"learning_rate": 0.00014384554280670784,
|
1570 |
+
"loss": 2.9001,
|
1571 |
+
"step": 209000
|
1572 |
+
},
|
1573 |
+
{
|
1574 |
+
"epoch": 11.2933584296854,
|
1575 |
+
"grad_norm": 0.40092912316322327,
|
1576 |
+
"learning_rate": 0.00014296293027360986,
|
1577 |
+
"loss": 2.903,
|
1578 |
+
"step": 210000
|
1579 |
+
},
|
1580 |
+
{
|
1581 |
+
"epoch": 11.347136326969615,
|
1582 |
+
"grad_norm": 0.4014595150947571,
|
1583 |
+
"learning_rate": 0.000142081200353045,
|
1584 |
+
"loss": 2.9087,
|
1585 |
+
"step": 211000
|
1586 |
+
},
|
1587 |
+
{
|
1588 |
+
"epoch": 11.400914224253832,
|
1589 |
+
"grad_norm": 0.4027450978755951,
|
1590 |
+
"learning_rate": 0.00014119858781994703,
|
1591 |
+
"loss": 2.9107,
|
1592 |
+
"step": 212000
|
1593 |
+
},
|
1594 |
+
{
|
1595 |
+
"epoch": 11.454692121538049,
|
1596 |
+
"grad_norm": 0.3879305124282837,
|
1597 |
+
"learning_rate": 0.00014031597528684904,
|
1598 |
+
"loss": 2.9139,
|
1599 |
+
"step": 213000
|
1600 |
+
},
|
1601 |
+
{
|
1602 |
+
"epoch": 11.508470018822264,
|
1603 |
+
"grad_norm": 0.42585039138793945,
|
1604 |
+
"learning_rate": 0.0001394333627537511,
|
1605 |
+
"loss": 2.9174,
|
1606 |
+
"step": 214000
|
1607 |
+
},
|
1608 |
+
{
|
1609 |
+
"epoch": 11.56224791610648,
|
1610 |
+
"grad_norm": 0.41681861877441406,
|
1611 |
+
"learning_rate": 0.00013855163283318624,
|
1612 |
+
"loss": 2.918,
|
1613 |
+
"step": 215000
|
1614 |
+
},
|
1615 |
+
{
|
1616 |
+
"epoch": 11.616025813390696,
|
1617 |
+
"grad_norm": 0.3900226950645447,
|
1618 |
+
"learning_rate": 0.00013766990291262134,
|
1619 |
+
"loss": 2.919,
|
1620 |
+
"step": 216000
|
1621 |
+
},
|
1622 |
+
{
|
1623 |
+
"epoch": 11.669803710674913,
|
1624 |
+
"grad_norm": 0.3816623091697693,
|
1625 |
+
"learning_rate": 0.00013678729037952338,
|
1626 |
+
"loss": 2.9215,
|
1627 |
+
"step": 217000
|
1628 |
+
},
|
1629 |
+
{
|
1630 |
+
"epoch": 11.72358160795913,
|
1631 |
+
"grad_norm": 0.37244102358818054,
|
1632 |
+
"learning_rate": 0.0001359046778464254,
|
1633 |
+
"loss": 2.9189,
|
1634 |
+
"step": 218000
|
1635 |
+
},
|
1636 |
+
{
|
1637 |
+
"epoch": 11.777359505243345,
|
1638 |
+
"grad_norm": 0.39248353242874146,
|
1639 |
+
"learning_rate": 0.00013502206531332744,
|
1640 |
+
"loss": 2.9165,
|
1641 |
+
"step": 219000
|
1642 |
+
},
|
1643 |
+
{
|
1644 |
+
"epoch": 11.831137402527562,
|
1645 |
+
"grad_norm": 0.3882627785205841,
|
1646 |
+
"learning_rate": 0.00013414033539276257,
|
1647 |
+
"loss": 2.9226,
|
1648 |
+
"step": 220000
|
1649 |
+
},
|
1650 |
+
{
|
1651 |
+
"epoch": 11.884915299811777,
|
1652 |
+
"grad_norm": 0.39201658964157104,
|
1653 |
+
"learning_rate": 0.00013325772285966458,
|
1654 |
+
"loss": 2.9234,
|
1655 |
+
"step": 221000
|
1656 |
+
},
|
1657 |
+
{
|
1658 |
+
"epoch": 11.938693197095994,
|
1659 |
+
"grad_norm": 0.39154133200645447,
|
1660 |
+
"learning_rate": 0.00013237599293909974,
|
1661 |
+
"loss": 2.9267,
|
1662 |
+
"step": 222000
|
1663 |
+
},
|
1664 |
+
{
|
1665 |
+
"epoch": 11.99247109438021,
|
1666 |
+
"grad_norm": 0.39426833391189575,
|
1667 |
+
"learning_rate": 0.00013149338040600175,
|
1668 |
+
"loss": 2.9234,
|
1669 |
+
"step": 223000
|
1670 |
+
},
|
1671 |
+
{
|
1672 |
+
"epoch": 12.0,
|
1673 |
+
"eval_accuracy": 0.40836213447042485,
|
1674 |
+
"eval_loss": 3.383244037628174,
|
1675 |
+
"eval_runtime": 154.3587,
|
1676 |
+
"eval_samples_per_second": 375.236,
|
1677 |
+
"eval_steps_per_second": 5.869,
|
1678 |
+
"step": 223140
|
1679 |
+
},
|
1680 |
+
{
|
1681 |
+
"epoch": 12.046248991664426,
|
1682 |
+
"grad_norm": 0.3944448232650757,
|
1683 |
+
"learning_rate": 0.00013061165048543688,
|
1684 |
+
"loss": 2.8663,
|
1685 |
+
"step": 224000
|
1686 |
+
},
|
1687 |
+
{
|
1688 |
+
"epoch": 12.100026888948642,
|
1689 |
+
"grad_norm": 0.409795880317688,
|
1690 |
+
"learning_rate": 0.0001297290379523389,
|
1691 |
+
"loss": 2.867,
|
1692 |
+
"step": 225000
|
1693 |
+
},
|
1694 |
+
{
|
1695 |
+
"epoch": 12.153804786232858,
|
1696 |
+
"grad_norm": 0.42699024081230164,
|
1697 |
+
"learning_rate": 0.00012884642541924094,
|
1698 |
+
"loss": 2.8737,
|
1699 |
+
"step": 226000
|
1700 |
+
},
|
1701 |
+
{
|
1702 |
+
"epoch": 12.207582683517074,
|
1703 |
+
"grad_norm": 0.40281254053115845,
|
1704 |
+
"learning_rate": 0.00012796557811120917,
|
1705 |
+
"loss": 2.8802,
|
1706 |
+
"step": 227000
|
1707 |
+
},
|
1708 |
+
{
|
1709 |
+
"epoch": 12.261360580801291,
|
1710 |
+
"grad_norm": 0.41500693559646606,
|
1711 |
+
"learning_rate": 0.0001270829655781112,
|
1712 |
+
"loss": 2.8801,
|
1713 |
+
"step": 228000
|
1714 |
+
},
|
1715 |
+
{
|
1716 |
+
"epoch": 12.315138478085506,
|
1717 |
+
"grad_norm": 0.3833758533000946,
|
1718 |
+
"learning_rate": 0.00012620035304501323,
|
1719 |
+
"loss": 2.8872,
|
1720 |
+
"step": 229000
|
1721 |
+
},
|
1722 |
+
{
|
1723 |
+
"epoch": 12.368916375369723,
|
1724 |
+
"grad_norm": 0.40702077746391296,
|
1725 |
+
"learning_rate": 0.00012531774051191525,
|
1726 |
+
"loss": 2.8852,
|
1727 |
+
"step": 230000
|
1728 |
+
},
|
1729 |
+
{
|
1730 |
+
"epoch": 12.422694272653938,
|
1731 |
+
"grad_norm": 0.38716500997543335,
|
1732 |
+
"learning_rate": 0.0001244351279788173,
|
1733 |
+
"loss": 2.8888,
|
1734 |
+
"step": 231000
|
1735 |
+
},
|
1736 |
+
{
|
1737 |
+
"epoch": 12.476472169938155,
|
1738 |
+
"grad_norm": 0.36926084756851196,
|
1739 |
+
"learning_rate": 0.00012355251544571934,
|
1740 |
+
"loss": 2.8933,
|
1741 |
+
"step": 232000
|
1742 |
+
},
|
1743 |
+
{
|
1744 |
+
"epoch": 12.530250067222372,
|
1745 |
+
"grad_norm": 0.40624773502349854,
|
1746 |
+
"learning_rate": 0.00012267078552515443,
|
1747 |
+
"loss": 2.8932,
|
1748 |
+
"step": 233000
|
1749 |
+
},
|
1750 |
+
{
|
1751 |
+
"epoch": 12.584027964506587,
|
1752 |
+
"grad_norm": 0.3890169858932495,
|
1753 |
+
"learning_rate": 0.00012178817299205648,
|
1754 |
+
"loss": 2.8969,
|
1755 |
+
"step": 234000
|
1756 |
+
},
|
1757 |
+
{
|
1758 |
+
"epoch": 12.637805861790804,
|
1759 |
+
"grad_norm": 0.403980553150177,
|
1760 |
+
"learning_rate": 0.0001209064430714916,
|
1761 |
+
"loss": 2.895,
|
1762 |
+
"step": 235000
|
1763 |
+
},
|
1764 |
+
{
|
1765 |
+
"epoch": 12.69158375907502,
|
1766 |
+
"grad_norm": 0.3942379057407379,
|
1767 |
+
"learning_rate": 0.00012002383053839363,
|
1768 |
+
"loss": 2.8956,
|
1769 |
+
"step": 236000
|
1770 |
+
},
|
1771 |
+
{
|
1772 |
+
"epoch": 12.745361656359236,
|
1773 |
+
"grad_norm": 0.3773936629295349,
|
1774 |
+
"learning_rate": 0.00011914121800529566,
|
1775 |
+
"loss": 2.8996,
|
1776 |
+
"step": 237000
|
1777 |
+
},
|
1778 |
+
{
|
1779 |
+
"epoch": 12.799139553643453,
|
1780 |
+
"grad_norm": 0.40187960863113403,
|
1781 |
+
"learning_rate": 0.0001182594880847308,
|
1782 |
+
"loss": 2.9019,
|
1783 |
+
"step": 238000
|
1784 |
+
},
|
1785 |
+
{
|
1786 |
+
"epoch": 12.852917450927668,
|
1787 |
+
"grad_norm": 0.3979538381099701,
|
1788 |
+
"learning_rate": 0.00011737687555163282,
|
1789 |
+
"loss": 2.9017,
|
1790 |
+
"step": 239000
|
1791 |
+
},
|
1792 |
+
{
|
1793 |
+
"epoch": 12.906695348211885,
|
1794 |
+
"grad_norm": 0.37608620524406433,
|
1795 |
+
"learning_rate": 0.00011649426301853485,
|
1796 |
+
"loss": 2.9048,
|
1797 |
+
"step": 240000
|
1798 |
+
},
|
1799 |
+
{
|
1800 |
+
"epoch": 12.9604732454961,
|
1801 |
+
"grad_norm": 0.39789021015167236,
|
1802 |
+
"learning_rate": 0.00011561253309796997,
|
1803 |
+
"loss": 2.9068,
|
1804 |
+
"step": 241000
|
1805 |
+
},
|
1806 |
+
{
|
1807 |
+
"epoch": 13.0,
|
1808 |
+
"eval_accuracy": 0.4087216070567606,
|
1809 |
+
"eval_loss": 3.383761167526245,
|
1810 |
+
"eval_runtime": 154.4144,
|
1811 |
+
"eval_samples_per_second": 375.101,
|
1812 |
+
"eval_steps_per_second": 5.867,
|
1813 |
+
"step": 241735
|
1814 |
+
},
|
1815 |
+
{
|
1816 |
+
"epoch": 13.014251142780317,
|
1817 |
+
"grad_norm": 0.39842721819877625,
|
1818 |
+
"learning_rate": 0.00011472992056487202,
|
1819 |
+
"loss": 2.889,
|
1820 |
+
"step": 242000
|
1821 |
+
},
|
1822 |
+
{
|
1823 |
+
"epoch": 13.068029040064534,
|
1824 |
+
"grad_norm": 0.40080752968788147,
|
1825 |
+
"learning_rate": 0.00011384730803177405,
|
1826 |
+
"loss": 2.851,
|
1827 |
+
"step": 243000
|
1828 |
+
},
|
1829 |
+
{
|
1830 |
+
"epoch": 13.12180693734875,
|
1831 |
+
"grad_norm": 0.40195319056510925,
|
1832 |
+
"learning_rate": 0.00011296557811120917,
|
1833 |
+
"loss": 2.8539,
|
1834 |
+
"step": 244000
|
1835 |
+
},
|
1836 |
+
{
|
1837 |
+
"epoch": 13.175584834632966,
|
1838 |
+
"grad_norm": 0.385681688785553,
|
1839 |
+
"learning_rate": 0.0001120829655781112,
|
1840 |
+
"loss": 2.8565,
|
1841 |
+
"step": 245000
|
1842 |
+
},
|
1843 |
+
{
|
1844 |
+
"epoch": 13.229362731917181,
|
1845 |
+
"grad_norm": 0.383973091840744,
|
1846 |
+
"learning_rate": 0.00011120035304501322,
|
1847 |
+
"loss": 2.8581,
|
1848 |
+
"step": 246000
|
1849 |
+
},
|
1850 |
+
{
|
1851 |
+
"epoch": 13.283140629201398,
|
1852 |
+
"grad_norm": 0.3838016092777252,
|
1853 |
+
"learning_rate": 0.00011031774051191526,
|
1854 |
+
"loss": 2.8625,
|
1855 |
+
"step": 247000
|
1856 |
+
},
|
1857 |
+
{
|
1858 |
+
"epoch": 13.336918526485615,
|
1859 |
+
"grad_norm": 0.38851508498191833,
|
1860 |
+
"learning_rate": 0.00010943601059135037,
|
1861 |
+
"loss": 2.8618,
|
1862 |
+
"step": 248000
|
1863 |
+
},
|
1864 |
+
{
|
1865 |
+
"epoch": 13.39069642376983,
|
1866 |
+
"grad_norm": 0.40345388650894165,
|
1867 |
+
"learning_rate": 0.00010855428067078551,
|
1868 |
+
"loss": 2.8634,
|
1869 |
+
"step": 249000
|
1870 |
+
},
|
1871 |
+
{
|
1872 |
+
"epoch": 13.444474321054047,
|
1873 |
+
"grad_norm": 0.41579893231391907,
|
1874 |
+
"learning_rate": 0.00010767166813768754,
|
1875 |
+
"loss": 2.8678,
|
1876 |
+
"step": 250000
|
1877 |
+
},
|
1878 |
+
{
|
1879 |
+
"epoch": 13.498252218338264,
|
1880 |
+
"grad_norm": 0.4044778347015381,
|
1881 |
+
"learning_rate": 0.00010678905560458957,
|
1882 |
+
"loss": 2.87,
|
1883 |
+
"step": 251000
|
1884 |
+
},
|
1885 |
+
{
|
1886 |
+
"epoch": 13.552030115622479,
|
1887 |
+
"grad_norm": 0.39501869678497314,
|
1888 |
+
"learning_rate": 0.0001059064430714916,
|
1889 |
+
"loss": 2.8777,
|
1890 |
+
"step": 252000
|
1891 |
+
},
|
1892 |
+
{
|
1893 |
+
"epoch": 13.605808012906696,
|
1894 |
+
"grad_norm": 0.406240850687027,
|
1895 |
+
"learning_rate": 0.00010502471315092674,
|
1896 |
+
"loss": 2.8775,
|
1897 |
+
"step": 253000
|
1898 |
+
},
|
1899 |
+
{
|
1900 |
+
"epoch": 13.659585910190911,
|
1901 |
+
"grad_norm": 0.39923685789108276,
|
1902 |
+
"learning_rate": 0.00010414210061782876,
|
1903 |
+
"loss": 2.8752,
|
1904 |
+
"step": 254000
|
1905 |
+
},
|
1906 |
+
{
|
1907 |
+
"epoch": 13.713363807475128,
|
1908 |
+
"grad_norm": 0.4037391245365143,
|
1909 |
+
"learning_rate": 0.0001032603706972639,
|
1910 |
+
"loss": 2.8808,
|
1911 |
+
"step": 255000
|
1912 |
+
},
|
1913 |
+
{
|
1914 |
+
"epoch": 13.767141704759345,
|
1915 |
+
"grad_norm": 0.3982155919075012,
|
1916 |
+
"learning_rate": 0.00010237775816416591,
|
1917 |
+
"loss": 2.8814,
|
1918 |
+
"step": 256000
|
1919 |
+
},
|
1920 |
+
{
|
1921 |
+
"epoch": 13.82091960204356,
|
1922 |
+
"grad_norm": 0.40128931403160095,
|
1923 |
+
"learning_rate": 0.00010149691085613415,
|
1924 |
+
"loss": 2.8843,
|
1925 |
+
"step": 257000
|
1926 |
+
},
|
1927 |
+
{
|
1928 |
+
"epoch": 13.874697499327777,
|
1929 |
+
"grad_norm": 0.4084520936012268,
|
1930 |
+
"learning_rate": 0.00010061429832303619,
|
1931 |
+
"loss": 2.8802,
|
1932 |
+
"step": 258000
|
1933 |
+
},
|
1934 |
+
{
|
1935 |
+
"epoch": 13.928475396611992,
|
1936 |
+
"grad_norm": 0.38339564204216003,
|
1937 |
+
"learning_rate": 9.973168578993821e-05,
|
1938 |
+
"loss": 2.8805,
|
1939 |
+
"step": 259000
|
1940 |
+
},
|
1941 |
+
{
|
1942 |
+
"epoch": 13.982253293896209,
|
1943 |
+
"grad_norm": 0.39040228724479675,
|
1944 |
+
"learning_rate": 9.884907325684024e-05,
|
1945 |
+
"loss": 2.88,
|
1946 |
+
"step": 260000
|
1947 |
+
},
|
1948 |
+
{
|
1949 |
+
"epoch": 14.0,
|
1950 |
+
"eval_accuracy": 0.4090670393196906,
|
1951 |
+
"eval_loss": 3.3881213665008545,
|
1952 |
+
"eval_runtime": 153.9195,
|
1953 |
+
"eval_samples_per_second": 376.307,
|
1954 |
+
"eval_steps_per_second": 5.886,
|
1955 |
+
"step": 260330
|
1956 |
+
},
|
1957 |
+
{
|
1958 |
+
"epoch": 14.036031191180426,
|
1959 |
+
"grad_norm": 0.4209740459918976,
|
1960 |
+
"learning_rate": 9.796734333627536e-05,
|
1961 |
+
"loss": 2.8478,
|
1962 |
+
"step": 261000
|
1963 |
+
},
|
1964 |
+
{
|
1965 |
+
"epoch": 14.08980908846464,
|
1966 |
+
"grad_norm": 0.40589722990989685,
|
1967 |
+
"learning_rate": 9.70847308031774e-05,
|
1968 |
+
"loss": 2.8301,
|
1969 |
+
"step": 262000
|
1970 |
+
},
|
1971 |
+
{
|
1972 |
+
"epoch": 14.143586985748858,
|
1973 |
+
"grad_norm": 0.4551624357700348,
|
1974 |
+
"learning_rate": 9.620300088261252e-05,
|
1975 |
+
"loss": 2.8316,
|
1976 |
+
"step": 263000
|
1977 |
+
},
|
1978 |
+
{
|
1979 |
+
"epoch": 14.197364883033073,
|
1980 |
+
"grad_norm": 0.4008908271789551,
|
1981 |
+
"learning_rate": 9.532038834951455e-05,
|
1982 |
+
"loss": 2.8363,
|
1983 |
+
"step": 264000
|
1984 |
+
},
|
1985 |
+
{
|
1986 |
+
"epoch": 14.25114278031729,
|
1987 |
+
"grad_norm": 0.4118269681930542,
|
1988 |
+
"learning_rate": 9.443865842894969e-05,
|
1989 |
+
"loss": 2.8439,
|
1990 |
+
"step": 265000
|
1991 |
+
},
|
1992 |
+
{
|
1993 |
+
"epoch": 14.304920677601507,
|
1994 |
+
"grad_norm": 0.4041500985622406,
|
1995 |
+
"learning_rate": 9.35560458958517e-05,
|
1996 |
+
"loss": 2.8449,
|
1997 |
+
"step": 266000
|
1998 |
+
},
|
1999 |
+
{
|
2000 |
+
"epoch": 14.358698574885722,
|
2001 |
+
"grad_norm": 0.41177451610565186,
|
2002 |
+
"learning_rate": 9.267431597528684e-05,
|
2003 |
+
"loss": 2.8455,
|
2004 |
+
"step": 267000
|
2005 |
+
},
|
2006 |
+
{
|
2007 |
+
"epoch": 14.412476472169939,
|
2008 |
+
"grad_norm": 0.399616539478302,
|
2009 |
+
"learning_rate": 9.179170344218886e-05,
|
2010 |
+
"loss": 2.8519,
|
2011 |
+
"step": 268000
|
2012 |
+
},
|
2013 |
+
{
|
2014 |
+
"epoch": 14.466254369454154,
|
2015 |
+
"grad_norm": 0.41544005274772644,
|
2016 |
+
"learning_rate": 9.09090909090909e-05,
|
2017 |
+
"loss": 2.8493,
|
2018 |
+
"step": 269000
|
2019 |
+
},
|
2020 |
+
{
|
2021 |
+
"epoch": 14.52003226673837,
|
2022 |
+
"grad_norm": 0.41849789023399353,
|
2023 |
+
"learning_rate": 9.002647837599293e-05,
|
2024 |
+
"loss": 2.8533,
|
2025 |
+
"step": 270000
|
2026 |
+
},
|
2027 |
+
{
|
2028 |
+
"epoch": 14.573810164022587,
|
2029 |
+
"grad_norm": 0.43560275435447693,
|
2030 |
+
"learning_rate": 8.914474845542806e-05,
|
2031 |
+
"loss": 2.8566,
|
2032 |
+
"step": 271000
|
2033 |
+
},
|
2034 |
+
{
|
2035 |
+
"epoch": 14.627588061306803,
|
2036 |
+
"grad_norm": 0.40364864468574524,
|
2037 |
+
"learning_rate": 8.82630185348632e-05,
|
2038 |
+
"loss": 2.8566,
|
2039 |
+
"step": 272000
|
2040 |
+
},
|
2041 |
+
{
|
2042 |
+
"epoch": 14.68136595859102,
|
2043 |
+
"grad_norm": 0.4094676375389099,
|
2044 |
+
"learning_rate": 8.738040600176521e-05,
|
2045 |
+
"loss": 2.8584,
|
2046 |
+
"step": 273000
|
2047 |
+
},
|
2048 |
+
{
|
2049 |
+
"epoch": 14.735143855875235,
|
2050 |
+
"grad_norm": 0.40439966320991516,
|
2051 |
+
"learning_rate": 8.649867608120035e-05,
|
2052 |
+
"loss": 2.8592,
|
2053 |
+
"step": 274000
|
2054 |
+
},
|
2055 |
+
{
|
2056 |
+
"epoch": 14.788921753159451,
|
2057 |
+
"grad_norm": 0.4156148433685303,
|
2058 |
+
"learning_rate": 8.561606354810237e-05,
|
2059 |
+
"loss": 2.861,
|
2060 |
+
"step": 275000
|
2061 |
+
},
|
2062 |
+
{
|
2063 |
+
"epoch": 14.842699650443668,
|
2064 |
+
"grad_norm": 0.4287779927253723,
|
2065 |
+
"learning_rate": 8.47334510150044e-05,
|
2066 |
+
"loss": 2.8616,
|
2067 |
+
"step": 276000
|
2068 |
+
},
|
2069 |
+
{
|
2070 |
+
"epoch": 14.896477547727883,
|
2071 |
+
"grad_norm": 0.4273608922958374,
|
2072 |
+
"learning_rate": 8.385083848190644e-05,
|
2073 |
+
"loss": 2.865,
|
2074 |
+
"step": 277000
|
2075 |
+
},
|
2076 |
+
{
|
2077 |
+
"epoch": 14.9502554450121,
|
2078 |
+
"grad_norm": 0.4030059278011322,
|
2079 |
+
"learning_rate": 8.296910856134155e-05,
|
2080 |
+
"loss": 2.8614,
|
2081 |
+
"step": 278000
|
2082 |
+
},
|
2083 |
+
{
|
2084 |
+
"epoch": 15.0,
|
2085 |
+
"eval_accuracy": 0.40971853719714973,
|
2086 |
+
"eval_loss": 3.3862648010253906,
|
2087 |
+
"eval_runtime": 154.422,
|
2088 |
+
"eval_samples_per_second": 375.083,
|
2089 |
+
"eval_steps_per_second": 5.867,
|
2090 |
+
"step": 278925
|
2091 |
+
},
|
2092 |
+
{
|
2093 |
+
"epoch": 15.004033342296315,
|
2094 |
+
"grad_norm": 0.42723333835601807,
|
2095 |
+
"learning_rate": 8.20864960282436e-05,
|
2096 |
+
"loss": 2.8594,
|
2097 |
+
"step": 279000
|
2098 |
+
},
|
2099 |
+
{
|
2100 |
+
"epoch": 15.057811239580532,
|
2101 |
+
"grad_norm": 0.420899361371994,
|
2102 |
+
"learning_rate": 8.120653133274491e-05,
|
2103 |
+
"loss": 2.8111,
|
2104 |
+
"step": 280000
|
2105 |
+
},
|
2106 |
+
{
|
2107 |
+
"epoch": 15.11158913686475,
|
2108 |
+
"grad_norm": 0.4493071734905243,
|
2109 |
+
"learning_rate": 8.032391879964694e-05,
|
2110 |
+
"loss": 2.8189,
|
2111 |
+
"step": 281000
|
2112 |
+
},
|
2113 |
+
{
|
2114 |
+
"epoch": 15.165367034148964,
|
2115 |
+
"grad_norm": 0.4198229908943176,
|
2116 |
+
"learning_rate": 7.944130626654899e-05,
|
2117 |
+
"loss": 2.8199,
|
2118 |
+
"step": 282000
|
2119 |
+
},
|
2120 |
+
{
|
2121 |
+
"epoch": 15.219144931433181,
|
2122 |
+
"grad_norm": 0.44265827536582947,
|
2123 |
+
"learning_rate": 7.8558693733451e-05,
|
2124 |
+
"loss": 2.8237,
|
2125 |
+
"step": 283000
|
2126 |
+
},
|
2127 |
+
{
|
2128 |
+
"epoch": 15.272922828717396,
|
2129 |
+
"grad_norm": 0.44851958751678467,
|
2130 |
+
"learning_rate": 7.767608120035305e-05,
|
2131 |
+
"loss": 2.8257,
|
2132 |
+
"step": 284000
|
2133 |
+
},
|
2134 |
+
{
|
2135 |
+
"epoch": 15.326700726001613,
|
2136 |
+
"grad_norm": 0.4241320788860321,
|
2137 |
+
"learning_rate": 7.679435127978816e-05,
|
2138 |
+
"loss": 2.8296,
|
2139 |
+
"step": 285000
|
2140 |
+
},
|
2141 |
+
{
|
2142 |
+
"epoch": 15.38047862328583,
|
2143 |
+
"grad_norm": 0.4106122851371765,
|
2144 |
+
"learning_rate": 7.59117387466902e-05,
|
2145 |
+
"loss": 2.8334,
|
2146 |
+
"step": 286000
|
2147 |
+
},
|
2148 |
+
{
|
2149 |
+
"epoch": 15.434256520570045,
|
2150 |
+
"grad_norm": 0.4082964360713959,
|
2151 |
+
"learning_rate": 7.502912621359223e-05,
|
2152 |
+
"loss": 2.8322,
|
2153 |
+
"step": 287000
|
2154 |
+
},
|
2155 |
+
{
|
2156 |
+
"epoch": 15.488034417854262,
|
2157 |
+
"grad_norm": 0.4180322587490082,
|
2158 |
+
"learning_rate": 7.414739629302736e-05,
|
2159 |
+
"loss": 2.8338,
|
2160 |
+
"step": 288000
|
2161 |
+
},
|
2162 |
+
{
|
2163 |
+
"epoch": 15.541812315138477,
|
2164 |
+
"grad_norm": 0.424429327249527,
|
2165 |
+
"learning_rate": 7.326478375992939e-05,
|
2166 |
+
"loss": 2.8344,
|
2167 |
+
"step": 289000
|
2168 |
+
},
|
2169 |
+
{
|
2170 |
+
"epoch": 15.595590212422694,
|
2171 |
+
"grad_norm": 0.4517696797847748,
|
2172 |
+
"learning_rate": 7.238305383936451e-05,
|
2173 |
+
"loss": 2.8371,
|
2174 |
+
"step": 290000
|
2175 |
+
},
|
2176 |
+
{
|
2177 |
+
"epoch": 15.649368109706911,
|
2178 |
+
"grad_norm": 0.433946430683136,
|
2179 |
+
"learning_rate": 7.150044130626654e-05,
|
2180 |
+
"loss": 2.8344,
|
2181 |
+
"step": 291000
|
2182 |
+
},
|
2183 |
+
{
|
2184 |
+
"epoch": 15.703146006991126,
|
2185 |
+
"grad_norm": 0.43706047534942627,
|
2186 |
+
"learning_rate": 7.061782877316857e-05,
|
2187 |
+
"loss": 2.8371,
|
2188 |
+
"step": 292000
|
2189 |
+
},
|
2190 |
+
{
|
2191 |
+
"epoch": 15.756923904275343,
|
2192 |
+
"grad_norm": 0.4216670095920563,
|
2193 |
+
"learning_rate": 6.97352162400706e-05,
|
2194 |
+
"loss": 2.8416,
|
2195 |
+
"step": 293000
|
2196 |
+
},
|
2197 |
+
{
|
2198 |
+
"epoch": 15.81070180155956,
|
2199 |
+
"grad_norm": 0.42567121982574463,
|
2200 |
+
"learning_rate": 6.885348631950573e-05,
|
2201 |
+
"loss": 2.8412,
|
2202 |
+
"step": 294000
|
2203 |
+
},
|
2204 |
+
{
|
2205 |
+
"epoch": 15.864479698843775,
|
2206 |
+
"grad_norm": 0.45544958114624023,
|
2207 |
+
"learning_rate": 6.797087378640776e-05,
|
2208 |
+
"loss": 2.8425,
|
2209 |
+
"step": 295000
|
2210 |
+
},
|
2211 |
+
{
|
2212 |
+
"epoch": 15.918257596127992,
|
2213 |
+
"grad_norm": 0.42355233430862427,
|
2214 |
+
"learning_rate": 6.708914386584288e-05,
|
2215 |
+
"loss": 2.8427,
|
2216 |
+
"step": 296000
|
2217 |
+
},
|
2218 |
+
{
|
2219 |
+
"epoch": 15.972035493412207,
|
2220 |
+
"grad_norm": 0.40585145354270935,
|
2221 |
+
"learning_rate": 6.620741394527801e-05,
|
2222 |
+
"loss": 2.841,
|
2223 |
+
"step": 297000
|
2224 |
+
},
|
2225 |
+
{
|
2226 |
+
"epoch": 16.0,
|
2227 |
+
"eval_accuracy": 0.40937653104184507,
|
2228 |
+
"eval_loss": 3.4092414379119873,
|
2229 |
+
"eval_runtime": 154.0074,
|
2230 |
+
"eval_samples_per_second": 376.092,
|
2231 |
+
"eval_steps_per_second": 5.883,
|
2232 |
+
"step": 297520
|
2233 |
+
},
|
2234 |
+
{
|
2235 |
+
"epoch": 16.025813390696424,
|
2236 |
+
"grad_norm": 0.43941229581832886,
|
2237 |
+
"learning_rate": 6.532480141218005e-05,
|
2238 |
+
"loss": 2.8215,
|
2239 |
+
"step": 298000
|
2240 |
+
},
|
2241 |
+
{
|
2242 |
+
"epoch": 16.07959128798064,
|
2243 |
+
"grad_norm": 0.4342760145664215,
|
2244 |
+
"learning_rate": 6.444218887908208e-05,
|
2245 |
+
"loss": 2.8046,
|
2246 |
+
"step": 299000
|
2247 |
+
},
|
2248 |
+
{
|
2249 |
+
"epoch": 16.133369185264858,
|
2250 |
+
"grad_norm": 0.4505048394203186,
|
2251 |
+
"learning_rate": 6.356045895851721e-05,
|
2252 |
+
"loss": 2.8076,
|
2253 |
+
"step": 300000
|
2254 |
+
},
|
2255 |
+
{
|
2256 |
+
"epoch": 16.18714708254907,
|
2257 |
+
"grad_norm": 0.45032915472984314,
|
2258 |
+
"learning_rate": 6.267784642541924e-05,
|
2259 |
+
"loss": 2.8041,
|
2260 |
+
"step": 301000
|
2261 |
+
},
|
2262 |
+
{
|
2263 |
+
"epoch": 16.240924979833288,
|
2264 |
+
"grad_norm": 0.44145554304122925,
|
2265 |
+
"learning_rate": 6.179523389232127e-05,
|
2266 |
+
"loss": 2.8097,
|
2267 |
+
"step": 302000
|
2268 |
+
},
|
2269 |
+
{
|
2270 |
+
"epoch": 16.294702877117505,
|
2271 |
+
"grad_norm": 0.4323660433292389,
|
2272 |
+
"learning_rate": 6.091262135922329e-05,
|
2273 |
+
"loss": 2.8111,
|
2274 |
+
"step": 303000
|
2275 |
+
},
|
2276 |
+
{
|
2277 |
+
"epoch": 16.34848077440172,
|
2278 |
+
"grad_norm": 0.43976891040802,
|
2279 |
+
"learning_rate": 6.0030891438658424e-05,
|
2280 |
+
"loss": 2.8105,
|
2281 |
+
"step": 304000
|
2282 |
+
},
|
2283 |
+
{
|
2284 |
+
"epoch": 16.40225867168594,
|
2285 |
+
"grad_norm": 0.4359465539455414,
|
2286 |
+
"learning_rate": 5.914827890556045e-05,
|
2287 |
+
"loss": 2.8109,
|
2288 |
+
"step": 305000
|
2289 |
+
},
|
2290 |
+
{
|
2291 |
+
"epoch": 16.456036568970152,
|
2292 |
+
"grad_norm": 0.43427878618240356,
|
2293 |
+
"learning_rate": 5.8265666372462485e-05,
|
2294 |
+
"loss": 2.8154,
|
2295 |
+
"step": 306000
|
2296 |
+
},
|
2297 |
+
{
|
2298 |
+
"epoch": 16.50981446625437,
|
2299 |
+
"grad_norm": 0.45613136887550354,
|
2300 |
+
"learning_rate": 5.7383053839364515e-05,
|
2301 |
+
"loss": 2.8171,
|
2302 |
+
"step": 307000
|
2303 |
+
},
|
2304 |
+
{
|
2305 |
+
"epoch": 16.563592363538586,
|
2306 |
+
"grad_norm": 0.42773687839508057,
|
2307 |
+
"learning_rate": 5.650220653133274e-05,
|
2308 |
+
"loss": 2.8169,
|
2309 |
+
"step": 308000
|
2310 |
+
},
|
2311 |
+
{
|
2312 |
+
"epoch": 16.617370260822803,
|
2313 |
+
"grad_norm": 0.4209708869457245,
|
2314 |
+
"learning_rate": 5.561959399823477e-05,
|
2315 |
+
"loss": 2.8147,
|
2316 |
+
"step": 309000
|
2317 |
+
},
|
2318 |
+
{
|
2319 |
+
"epoch": 16.67114815810702,
|
2320 |
+
"grad_norm": 0.4385839104652405,
|
2321 |
+
"learning_rate": 5.4736981465136795e-05,
|
2322 |
+
"loss": 2.822,
|
2323 |
+
"step": 310000
|
2324 |
+
},
|
2325 |
+
{
|
2326 |
+
"epoch": 16.724926055391233,
|
2327 |
+
"grad_norm": 0.459045946598053,
|
2328 |
+
"learning_rate": 5.3855251544571934e-05,
|
2329 |
+
"loss": 2.8214,
|
2330 |
+
"step": 311000
|
2331 |
+
},
|
2332 |
+
{
|
2333 |
+
"epoch": 16.77870395267545,
|
2334 |
+
"grad_norm": 0.43635278940200806,
|
2335 |
+
"learning_rate": 5.297263901147396e-05,
|
2336 |
+
"loss": 2.8202,
|
2337 |
+
"step": 312000
|
2338 |
+
},
|
2339 |
+
{
|
2340 |
+
"epoch": 16.832481849959667,
|
2341 |
+
"grad_norm": 0.43077248334884644,
|
2342 |
+
"learning_rate": 5.209002647837599e-05,
|
2343 |
+
"loss": 2.823,
|
2344 |
+
"step": 313000
|
2345 |
+
},
|
2346 |
+
{
|
2347 |
+
"epoch": 16.886259747243884,
|
2348 |
+
"grad_norm": 0.43129387497901917,
|
2349 |
+
"learning_rate": 5.120741394527802e-05,
|
2350 |
+
"loss": 2.8237,
|
2351 |
+
"step": 314000
|
2352 |
+
},
|
2353 |
+
{
|
2354 |
+
"epoch": 16.9400376445281,
|
2355 |
+
"grad_norm": 0.42089757323265076,
|
2356 |
+
"learning_rate": 5.032568402471314e-05,
|
2357 |
+
"loss": 2.8228,
|
2358 |
+
"step": 315000
|
2359 |
+
},
|
2360 |
+
{
|
2361 |
+
"epoch": 16.993815541812314,
|
2362 |
+
"grad_norm": 0.44255971908569336,
|
2363 |
+
"learning_rate": 4.944395410414828e-05,
|
2364 |
+
"loss": 2.8225,
|
2365 |
+
"step": 316000
|
2366 |
+
},
|
2367 |
+
{
|
2368 |
+
"epoch": 17.0,
|
2369 |
+
"eval_accuracy": 0.4098076831739891,
|
2370 |
+
"eval_loss": 3.3965845108032227,
|
2371 |
+
"eval_runtime": 154.459,
|
2372 |
+
"eval_samples_per_second": 374.993,
|
2373 |
+
"eval_steps_per_second": 5.866,
|
2374 |
+
"step": 316115
|
2375 |
+
},
|
2376 |
+
{
|
2377 |
+
"epoch": 17.04759343909653,
|
2378 |
+
"grad_norm": 0.4566737115383148,
|
2379 |
+
"learning_rate": 4.8561341571050305e-05,
|
2380 |
+
"loss": 2.7889,
|
2381 |
+
"step": 317000
|
2382 |
+
},
|
2383 |
+
{
|
2384 |
+
"epoch": 17.101371336380748,
|
2385 |
+
"grad_norm": 0.4266511797904968,
|
2386 |
+
"learning_rate": 4.767961165048544e-05,
|
2387 |
+
"loss": 2.7872,
|
2388 |
+
"step": 318000
|
2389 |
+
},
|
2390 |
+
{
|
2391 |
+
"epoch": 17.155149233664964,
|
2392 |
+
"grad_norm": 0.43798133730888367,
|
2393 |
+
"learning_rate": 4.679699911738746e-05,
|
2394 |
+
"loss": 2.7905,
|
2395 |
+
"step": 319000
|
2396 |
+
},
|
2397 |
+
{
|
2398 |
+
"epoch": 17.20892713094918,
|
2399 |
+
"grad_norm": 0.4472503662109375,
|
2400 |
+
"learning_rate": 4.591438658428949e-05,
|
2401 |
+
"loss": 2.7931,
|
2402 |
+
"step": 320000
|
2403 |
+
},
|
2404 |
+
{
|
2405 |
+
"epoch": 17.262705028233395,
|
2406 |
+
"grad_norm": 0.45179733633995056,
|
2407 |
+
"learning_rate": 4.503177405119153e-05,
|
2408 |
+
"loss": 2.7924,
|
2409 |
+
"step": 321000
|
2410 |
+
},
|
2411 |
+
{
|
2412 |
+
"epoch": 17.31648292551761,
|
2413 |
+
"grad_norm": 0.43849724531173706,
|
2414 |
+
"learning_rate": 4.414916151809356e-05,
|
2415 |
+
"loss": 2.7959,
|
2416 |
+
"step": 322000
|
2417 |
+
},
|
2418 |
+
{
|
2419 |
+
"epoch": 17.37026082280183,
|
2420 |
+
"grad_norm": 0.444363534450531,
|
2421 |
+
"learning_rate": 4.326743159752868e-05,
|
2422 |
+
"loss": 2.7984,
|
2423 |
+
"step": 323000
|
2424 |
+
},
|
2425 |
+
{
|
2426 |
+
"epoch": 17.424038720086045,
|
2427 |
+
"grad_norm": 0.4618755280971527,
|
2428 |
+
"learning_rate": 4.238481906443071e-05,
|
2429 |
+
"loss": 2.7971,
|
2430 |
+
"step": 324000
|
2431 |
+
},
|
2432 |
+
{
|
2433 |
+
"epoch": 17.477816617370262,
|
2434 |
+
"grad_norm": 0.44242867827415466,
|
2435 |
+
"learning_rate": 4.150308914386584e-05,
|
2436 |
+
"loss": 2.7994,
|
2437 |
+
"step": 325000
|
2438 |
+
},
|
2439 |
+
{
|
2440 |
+
"epoch": 17.531594514654476,
|
2441 |
+
"grad_norm": 0.4494096636772156,
|
2442 |
+
"learning_rate": 4.062047661076787e-05,
|
2443 |
+
"loss": 2.7972,
|
2444 |
+
"step": 326000
|
2445 |
+
},
|
2446 |
+
{
|
2447 |
+
"epoch": 17.585372411938692,
|
2448 |
+
"grad_norm": 0.4398290514945984,
|
2449 |
+
"learning_rate": 3.97378640776699e-05,
|
2450 |
+
"loss": 2.8002,
|
2451 |
+
"step": 327000
|
2452 |
+
},
|
2453 |
+
{
|
2454 |
+
"epoch": 17.63915030922291,
|
2455 |
+
"grad_norm": 0.43445536494255066,
|
2456 |
+
"learning_rate": 3.885525154457193e-05,
|
2457 |
+
"loss": 2.8015,
|
2458 |
+
"step": 328000
|
2459 |
+
},
|
2460 |
+
{
|
2461 |
+
"epoch": 17.692928206507126,
|
2462 |
+
"grad_norm": 0.43242964148521423,
|
2463 |
+
"learning_rate": 3.797352162400706e-05,
|
2464 |
+
"loss": 2.8037,
|
2465 |
+
"step": 329000
|
2466 |
+
},
|
2467 |
+
{
|
2468 |
+
"epoch": 17.746706103791343,
|
2469 |
+
"grad_norm": 0.43721792101860046,
|
2470 |
+
"learning_rate": 3.7091791703442186e-05,
|
2471 |
+
"loss": 2.8024,
|
2472 |
+
"step": 330000
|
2473 |
+
},
|
2474 |
+
{
|
2475 |
+
"epoch": 17.800484001075557,
|
2476 |
+
"grad_norm": 0.46508708596229553,
|
2477 |
+
"learning_rate": 3.6209179170344216e-05,
|
2478 |
+
"loss": 2.8028,
|
2479 |
+
"step": 331000
|
2480 |
+
},
|
2481 |
+
{
|
2482 |
+
"epoch": 17.854261898359773,
|
2483 |
+
"grad_norm": 0.4641873836517334,
|
2484 |
+
"learning_rate": 3.5326566637246246e-05,
|
2485 |
+
"loss": 2.8042,
|
2486 |
+
"step": 332000
|
2487 |
+
},
|
2488 |
+
{
|
2489 |
+
"epoch": 17.90803979564399,
|
2490 |
+
"grad_norm": 0.4373840391635895,
|
2491 |
+
"learning_rate": 3.4443954104148276e-05,
|
2492 |
+
"loss": 2.8067,
|
2493 |
+
"step": 333000
|
2494 |
+
},
|
2495 |
+
{
|
2496 |
+
"epoch": 17.961817692928207,
|
2497 |
+
"grad_norm": 0.4608209729194641,
|
2498 |
+
"learning_rate": 3.35622241835834e-05,
|
2499 |
+
"loss": 2.8062,
|
2500 |
+
"step": 334000
|
2501 |
+
},
|
2502 |
+
{
|
2503 |
+
"epoch": 18.0,
|
2504 |
+
"eval_accuracy": 0.4095544199240385,
|
2505 |
+
"eval_loss": 3.4095396995544434,
|
2506 |
+
"eval_runtime": 154.3283,
|
2507 |
+
"eval_samples_per_second": 375.31,
|
2508 |
+
"eval_steps_per_second": 5.871,
|
2509 |
+
"step": 334710
|
2510 |
+
},
|
2511 |
+
{
|
2512 |
+
"epoch": 18.015595590212424,
|
2513 |
+
"grad_norm": 0.4512391686439514,
|
2514 |
+
"learning_rate": 3.268049426301853e-05,
|
2515 |
+
"loss": 2.7991,
|
2516 |
+
"step": 335000
|
2517 |
+
},
|
2518 |
+
{
|
2519 |
+
"epoch": 18.069373487496637,
|
2520 |
+
"grad_norm": 0.4583674967288971,
|
2521 |
+
"learning_rate": 3.179788172992056e-05,
|
2522 |
+
"loss": 2.7775,
|
2523 |
+
"step": 336000
|
2524 |
+
},
|
2525 |
+
{
|
2526 |
+
"epoch": 18.123151384780854,
|
2527 |
+
"grad_norm": 0.43765154480934143,
|
2528 |
+
"learning_rate": 3.0915269196822593e-05,
|
2529 |
+
"loss": 2.7777,
|
2530 |
+
"step": 337000
|
2531 |
+
},
|
2532 |
+
{
|
2533 |
+
"epoch": 18.17692928206507,
|
2534 |
+
"grad_norm": 0.4475190341472626,
|
2535 |
+
"learning_rate": 3.003265666372462e-05,
|
2536 |
+
"loss": 2.7755,
|
2537 |
+
"step": 338000
|
2538 |
+
},
|
2539 |
+
{
|
2540 |
+
"epoch": 18.230707179349288,
|
2541 |
+
"grad_norm": 0.47484225034713745,
|
2542 |
+
"learning_rate": 2.9150926743159752e-05,
|
2543 |
+
"loss": 2.7748,
|
2544 |
+
"step": 339000
|
2545 |
+
},
|
2546 |
+
{
|
2547 |
+
"epoch": 18.284485076633505,
|
2548 |
+
"grad_norm": 0.44132038950920105,
|
2549 |
+
"learning_rate": 2.8269196822594877e-05,
|
2550 |
+
"loss": 2.7786,
|
2551 |
+
"step": 340000
|
2552 |
+
},
|
2553 |
+
{
|
2554 |
+
"epoch": 18.33826297391772,
|
2555 |
+
"grad_norm": 0.44409921765327454,
|
2556 |
+
"learning_rate": 2.7386584289496907e-05,
|
2557 |
+
"loss": 2.7843,
|
2558 |
+
"step": 341000
|
2559 |
+
},
|
2560 |
+
{
|
2561 |
+
"epoch": 18.392040871201935,
|
2562 |
+
"grad_norm": 0.4480835795402527,
|
2563 |
+
"learning_rate": 2.6503971756398938e-05,
|
2564 |
+
"loss": 2.7806,
|
2565 |
+
"step": 342000
|
2566 |
+
},
|
2567 |
+
{
|
2568 |
+
"epoch": 18.445818768486152,
|
2569 |
+
"grad_norm": 0.46462225914001465,
|
2570 |
+
"learning_rate": 2.5621359223300968e-05,
|
2571 |
+
"loss": 2.7848,
|
2572 |
+
"step": 343000
|
2573 |
+
},
|
2574 |
+
{
|
2575 |
+
"epoch": 18.49959666577037,
|
2576 |
+
"grad_norm": 0.4442801773548126,
|
2577 |
+
"learning_rate": 2.4740511915269195e-05,
|
2578 |
+
"loss": 2.7872,
|
2579 |
+
"step": 344000
|
2580 |
+
},
|
2581 |
+
{
|
2582 |
+
"epoch": 18.553374563054586,
|
2583 |
+
"grad_norm": 0.45635318756103516,
|
2584 |
+
"learning_rate": 2.3857899382171225e-05,
|
2585 |
+
"loss": 2.7855,
|
2586 |
+
"step": 345000
|
2587 |
+
},
|
2588 |
+
{
|
2589 |
+
"epoch": 18.6071524603388,
|
2590 |
+
"grad_norm": 0.4429308772087097,
|
2591 |
+
"learning_rate": 2.2975286849073255e-05,
|
2592 |
+
"loss": 2.7819,
|
2593 |
+
"step": 346000
|
2594 |
+
},
|
2595 |
+
{
|
2596 |
+
"epoch": 18.660930357623016,
|
2597 |
+
"grad_norm": 0.4405366778373718,
|
2598 |
+
"learning_rate": 2.209355692850838e-05,
|
2599 |
+
"loss": 2.7859,
|
2600 |
+
"step": 347000
|
2601 |
+
},
|
2602 |
+
{
|
2603 |
+
"epoch": 18.714708254907233,
|
2604 |
+
"grad_norm": 0.4760023355484009,
|
2605 |
+
"learning_rate": 2.1210944395410414e-05,
|
2606 |
+
"loss": 2.7845,
|
2607 |
+
"step": 348000
|
2608 |
+
},
|
2609 |
+
{
|
2610 |
+
"epoch": 18.76848615219145,
|
2611 |
+
"grad_norm": 0.43945708870887756,
|
2612 |
+
"learning_rate": 2.032833186231244e-05,
|
2613 |
+
"loss": 2.7851,
|
2614 |
+
"step": 349000
|
2615 |
+
},
|
2616 |
+
{
|
2617 |
+
"epoch": 18.822264049475667,
|
2618 |
+
"grad_norm": 0.4329651892185211,
|
2619 |
+
"learning_rate": 1.9446601941747572e-05,
|
2620 |
+
"loss": 2.7858,
|
2621 |
+
"step": 350000
|
2622 |
+
},
|
2623 |
+
{
|
2624 |
+
"epoch": 18.87604194675988,
|
2625 |
+
"grad_norm": 0.4544674754142761,
|
2626 |
+
"learning_rate": 1.8563989408649602e-05,
|
2627 |
+
"loss": 2.7854,
|
2628 |
+
"step": 351000
|
2629 |
+
},
|
2630 |
+
{
|
2631 |
+
"epoch": 18.929819844044097,
|
2632 |
+
"grad_norm": 0.45160219073295593,
|
2633 |
+
"learning_rate": 1.7681376875551633e-05,
|
2634 |
+
"loss": 2.7868,
|
2635 |
+
"step": 352000
|
2636 |
+
},
|
2637 |
+
{
|
2638 |
+
"epoch": 18.983597741328314,
|
2639 |
+
"grad_norm": 0.4521346092224121,
|
2640 |
+
"learning_rate": 1.6798764342453663e-05,
|
2641 |
+
"loss": 2.7904,
|
2642 |
+
"step": 353000
|
2643 |
+
},
|
2644 |
+
{
|
2645 |
+
"epoch": 19.0,
|
2646 |
+
"eval_accuracy": 0.4098326736060797,
|
2647 |
+
"eval_loss": 3.416917562484741,
|
2648 |
+
"eval_runtime": 154.3288,
|
2649 |
+
"eval_samples_per_second": 375.309,
|
2650 |
+
"eval_steps_per_second": 5.871,
|
2651 |
+
"step": 353305
|
2652 |
+
},
|
2653 |
+
{
|
2654 |
+
"epoch": 19.03737563861253,
|
2655 |
+
"grad_norm": 0.4459937810897827,
|
2656 |
+
"learning_rate": 1.5916151809355693e-05,
|
2657 |
+
"loss": 2.7735,
|
2658 |
+
"step": 354000
|
2659 |
+
},
|
2660 |
+
{
|
2661 |
+
"epoch": 19.091153535896748,
|
2662 |
+
"grad_norm": 0.4429547190666199,
|
2663 |
+
"learning_rate": 1.5035304501323918e-05,
|
2664 |
+
"loss": 2.7666,
|
2665 |
+
"step": 355000
|
2666 |
+
},
|
2667 |
+
{
|
2668 |
+
"epoch": 19.14493143318096,
|
2669 |
+
"grad_norm": 0.45378807187080383,
|
2670 |
+
"learning_rate": 1.4152691968225948e-05,
|
2671 |
+
"loss": 2.7675,
|
2672 |
+
"step": 356000
|
2673 |
+
},
|
2674 |
+
{
|
2675 |
+
"epoch": 19.198709330465178,
|
2676 |
+
"grad_norm": 0.4565734267234802,
|
2677 |
+
"learning_rate": 1.3270079435127978e-05,
|
2678 |
+
"loss": 2.7678,
|
2679 |
+
"step": 357000
|
2680 |
+
},
|
2681 |
+
{
|
2682 |
+
"epoch": 19.252487227749395,
|
2683 |
+
"grad_norm": 0.46085768938064575,
|
2684 |
+
"learning_rate": 1.2387466902030009e-05,
|
2685 |
+
"loss": 2.7671,
|
2686 |
+
"step": 358000
|
2687 |
+
},
|
2688 |
+
{
|
2689 |
+
"epoch": 19.30626512503361,
|
2690 |
+
"grad_norm": 0.45043742656707764,
|
2691 |
+
"learning_rate": 1.1505736981465135e-05,
|
2692 |
+
"loss": 2.7697,
|
2693 |
+
"step": 359000
|
2694 |
+
},
|
2695 |
+
{
|
2696 |
+
"epoch": 19.36004302231783,
|
2697 |
+
"grad_norm": 0.44789180159568787,
|
2698 |
+
"learning_rate": 1.0623124448367166e-05,
|
2699 |
+
"loss": 2.7678,
|
2700 |
+
"step": 360000
|
2701 |
+
},
|
2702 |
+
{
|
2703 |
+
"epoch": 19.413820919602042,
|
2704 |
+
"grad_norm": 0.46358925104141235,
|
2705 |
+
"learning_rate": 9.741394527802292e-06,
|
2706 |
+
"loss": 2.7672,
|
2707 |
+
"step": 361000
|
2708 |
+
},
|
2709 |
+
{
|
2710 |
+
"epoch": 19.46759881688626,
|
2711 |
+
"grad_norm": 0.4664359390735626,
|
2712 |
+
"learning_rate": 8.858781994704324e-06,
|
2713 |
+
"loss": 2.7646,
|
2714 |
+
"step": 362000
|
2715 |
+
},
|
2716 |
+
{
|
2717 |
+
"epoch": 19.521376714170476,
|
2718 |
+
"grad_norm": 0.46864446997642517,
|
2719 |
+
"learning_rate": 7.976169461606355e-06,
|
2720 |
+
"loss": 2.7694,
|
2721 |
+
"step": 363000
|
2722 |
+
},
|
2723 |
+
{
|
2724 |
+
"epoch": 19.575154611454693,
|
2725 |
+
"grad_norm": 0.4606035351753235,
|
2726 |
+
"learning_rate": 7.093556928508385e-06,
|
2727 |
+
"loss": 2.7702,
|
2728 |
+
"step": 364000
|
2729 |
+
},
|
2730 |
+
{
|
2731 |
+
"epoch": 19.62893250873891,
|
2732 |
+
"grad_norm": 0.4529917538166046,
|
2733 |
+
"learning_rate": 6.21270962047661e-06,
|
2734 |
+
"loss": 2.7703,
|
2735 |
+
"step": 365000
|
2736 |
+
},
|
2737 |
+
{
|
2738 |
+
"epoch": 19.682710406023123,
|
2739 |
+
"grad_norm": 0.44931086897850037,
|
2740 |
+
"learning_rate": 5.33009708737864e-06,
|
2741 |
+
"loss": 2.7721,
|
2742 |
+
"step": 366000
|
2743 |
+
},
|
2744 |
+
{
|
2745 |
+
"epoch": 19.73648830330734,
|
2746 |
+
"grad_norm": 0.4593379497528076,
|
2747 |
+
"learning_rate": 4.44748455428067e-06,
|
2748 |
+
"loss": 2.766,
|
2749 |
+
"step": 367000
|
2750 |
+
},
|
2751 |
+
{
|
2752 |
+
"epoch": 19.790266200591557,
|
2753 |
+
"grad_norm": 0.4625178277492523,
|
2754 |
+
"learning_rate": 3.565754633715798e-06,
|
2755 |
+
"loss": 2.7716,
|
2756 |
+
"step": 368000
|
2757 |
+
},
|
2758 |
+
{
|
2759 |
+
"epoch": 19.844044097875774,
|
2760 |
+
"grad_norm": 0.441020667552948,
|
2761 |
+
"learning_rate": 2.6831421006178284e-06,
|
2762 |
+
"loss": 2.771,
|
2763 |
+
"step": 369000
|
2764 |
+
},
|
2765 |
+
{
|
2766 |
+
"epoch": 19.89782199515999,
|
2767 |
+
"grad_norm": 0.48773327469825745,
|
2768 |
+
"learning_rate": 1.8005295675198586e-06,
|
2769 |
+
"loss": 2.7657,
|
2770 |
+
"step": 370000
|
2771 |
+
},
|
2772 |
+
{
|
2773 |
+
"epoch": 19.951599892444204,
|
2774 |
+
"grad_norm": 0.45526123046875,
|
2775 |
+
"learning_rate": 9.179170344218888e-07,
|
2776 |
+
"loss": 2.775,
|
2777 |
+
"step": 371000
|
2778 |
+
},
|
2779 |
+
{
|
2780 |
+
"epoch": 20.0,
|
2781 |
+
"eval_accuracy": 0.4096272415057219,
|
2782 |
+
"eval_loss": 3.4256882667541504,
|
2783 |
+
"eval_runtime": 154.1682,
|
2784 |
+
"eval_samples_per_second": 375.7,
|
2785 |
+
"eval_steps_per_second": 5.877,
|
2786 |
+
"step": 371900
|
2787 |
+
},
|
2788 |
+
{
|
2789 |
+
"epoch": 20.0,
|
2790 |
+
"step": 371900,
|
2791 |
+
"total_flos": 1.5669257538816e+18,
|
2792 |
+
"train_loss": 3.0592986363049124,
|
2793 |
+
"train_runtime": 81068.1718,
|
2794 |
+
"train_samples_per_second": 146.799,
|
2795 |
+
"train_steps_per_second": 4.587
|
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 |
+
}
|