End of training
Browse files
README.md
CHANGED
@@ -1,11 +1,23 @@
|
|
1 |
---
|
2 |
tags:
|
3 |
- generated_from_trainer
|
|
|
|
|
4 |
metrics:
|
5 |
- accuracy
|
6 |
model-index:
|
7 |
- name: smolm-autoreg-bpe-counterfactual_babylm_aann_low_variability_numeral_1024-1e-3
|
8 |
-
results:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
9 |
---
|
10 |
|
11 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
@@ -13,7 +25,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
13 |
|
14 |
# smolm-autoreg-bpe-counterfactual_babylm_aann_low_variability_numeral_1024-1e-3
|
15 |
|
16 |
-
This model was trained from scratch on
|
17 |
It achieves the following results on the evaluation set:
|
18 |
- Loss: 3.3975
|
19 |
- Accuracy: 0.4113
|
|
|
1 |
---
|
2 |
tags:
|
3 |
- generated_from_trainer
|
4 |
+
datasets:
|
5 |
+
- kanishka/counterfactual_babylm_aann_low_variability_numeral
|
6 |
metrics:
|
7 |
- accuracy
|
8 |
model-index:
|
9 |
- name: smolm-autoreg-bpe-counterfactual_babylm_aann_low_variability_numeral_1024-1e-3
|
10 |
+
results:
|
11 |
+
- task:
|
12 |
+
name: Causal Language Modeling
|
13 |
+
type: text-generation
|
14 |
+
dataset:
|
15 |
+
name: kanishka/counterfactual_babylm_aann_low_variability_numeral
|
16 |
+
type: kanishka/counterfactual_babylm_aann_low_variability_numeral
|
17 |
+
metrics:
|
18 |
+
- name: Accuracy
|
19 |
+
type: accuracy
|
20 |
+
value: 0.41125847180686265
|
21 |
---
|
22 |
|
23 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
|
25 |
|
26 |
# smolm-autoreg-bpe-counterfactual_babylm_aann_low_variability_numeral_1024-1e-3
|
27 |
|
28 |
+
This model was trained from scratch on the kanishka/counterfactual_babylm_aann_low_variability_numeral dataset.
|
29 |
It achieves the following results on the evaluation set:
|
30 |
- Loss: 3.3975
|
31 |
- Accuracy: 0.4113
|
all_results.json
ADDED
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 20.0,
|
3 |
+
"eval_accuracy": 0.41125847180686265,
|
4 |
+
"eval_loss": 3.397514581680298,
|
5 |
+
"eval_runtime": 153.7992,
|
6 |
+
"eval_samples": 57921,
|
7 |
+
"eval_samples_per_second": 376.601,
|
8 |
+
"eval_steps_per_second": 5.891,
|
9 |
+
"perplexity": 29.889719196260273,
|
10 |
+
"train_loss": 3.027371880730273,
|
11 |
+
"train_runtime": 81125.0441,
|
12 |
+
"train_samples": 594966,
|
13 |
+
"train_samples_per_second": 146.679,
|
14 |
+
"train_steps_per_second": 4.584
|
15 |
+
}
|
eval_results.json
ADDED
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 20.0,
|
3 |
+
"eval_accuracy": 0.41125847180686265,
|
4 |
+
"eval_loss": 3.397514581680298,
|
5 |
+
"eval_runtime": 153.7992,
|
6 |
+
"eval_samples": 57921,
|
7 |
+
"eval_samples_per_second": 376.601,
|
8 |
+
"eval_steps_per_second": 5.891,
|
9 |
+
"perplexity": 29.889719196260273
|
10 |
+
}
|
runs/Jul15_13-15-43_phyl-ling-p01.la.utexas.edu/events.out.tfevents.1721148997.phyl-ling-p01.la.utexas.edu.2792777.1
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:97cd137b88f3e1a0c27a21f3b82a0c3a1821d1f6cd71a50438d9d98093e5a4eb
|
3 |
+
size 417
|
train_results.json
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 20.0,
|
3 |
+
"train_loss": 3.027371880730273,
|
4 |
+
"train_runtime": 81125.0441,
|
5 |
+
"train_samples": 594966,
|
6 |
+
"train_samples_per_second": 146.679,
|
7 |
+
"train_steps_per_second": 4.584
|
8 |
+
}
|
trainer_state.json
ADDED
@@ -0,0 +1,2807 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": null,
|
3 |
+
"best_model_checkpoint": null,
|
4 |
+
"epoch": 20.0,
|
5 |
+
"eval_steps": 500,
|
6 |
+
"global_step": 371860,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.05,
|
13 |
+
"grad_norm": 0.8706823587417603,
|
14 |
+
"learning_rate": 3.125e-05,
|
15 |
+
"loss": 6.2283,
|
16 |
+
"step": 1000
|
17 |
+
},
|
18 |
+
{
|
19 |
+
"epoch": 0.11,
|
20 |
+
"grad_norm": 0.9266456365585327,
|
21 |
+
"learning_rate": 6.25e-05,
|
22 |
+
"loss": 5.0066,
|
23 |
+
"step": 2000
|
24 |
+
},
|
25 |
+
{
|
26 |
+
"epoch": 0.16,
|
27 |
+
"grad_norm": 0.9616362452507019,
|
28 |
+
"learning_rate": 9.375e-05,
|
29 |
+
"loss": 4.6763,
|
30 |
+
"step": 3000
|
31 |
+
},
|
32 |
+
{
|
33 |
+
"epoch": 0.22,
|
34 |
+
"grad_norm": 0.7885107398033142,
|
35 |
+
"learning_rate": 0.000125,
|
36 |
+
"loss": 4.4613,
|
37 |
+
"step": 4000
|
38 |
+
},
|
39 |
+
{
|
40 |
+
"epoch": 0.27,
|
41 |
+
"grad_norm": 0.7654786705970764,
|
42 |
+
"learning_rate": 0.00015625,
|
43 |
+
"loss": 4.3065,
|
44 |
+
"step": 5000
|
45 |
+
},
|
46 |
+
{
|
47 |
+
"epoch": 0.32,
|
48 |
+
"grad_norm": 0.7073574662208557,
|
49 |
+
"learning_rate": 0.0001875,
|
50 |
+
"loss": 4.1746,
|
51 |
+
"step": 6000
|
52 |
+
},
|
53 |
+
{
|
54 |
+
"epoch": 0.38,
|
55 |
+
"grad_norm": 0.6433243751525879,
|
56 |
+
"learning_rate": 0.00021875,
|
57 |
+
"loss": 4.075,
|
58 |
+
"step": 7000
|
59 |
+
},
|
60 |
+
{
|
61 |
+
"epoch": 0.43,
|
62 |
+
"grad_norm": 0.6497963070869446,
|
63 |
+
"learning_rate": 0.00025,
|
64 |
+
"loss": 3.9794,
|
65 |
+
"step": 8000
|
66 |
+
},
|
67 |
+
{
|
68 |
+
"epoch": 0.48,
|
69 |
+
"grad_norm": 0.6403814554214478,
|
70 |
+
"learning_rate": 0.00028121875,
|
71 |
+
"loss": 3.9048,
|
72 |
+
"step": 9000
|
73 |
+
},
|
74 |
+
{
|
75 |
+
"epoch": 0.54,
|
76 |
+
"grad_norm": 0.6297944188117981,
|
77 |
+
"learning_rate": 0.00031246875000000003,
|
78 |
+
"loss": 3.8448,
|
79 |
+
"step": 10000
|
80 |
+
},
|
81 |
+
{
|
82 |
+
"epoch": 0.59,
|
83 |
+
"grad_norm": 0.5447595715522766,
|
84 |
+
"learning_rate": 0.00034371875,
|
85 |
+
"loss": 3.7899,
|
86 |
+
"step": 11000
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 0.65,
|
90 |
+
"grad_norm": 0.5269156098365784,
|
91 |
+
"learning_rate": 0.0003749375,
|
92 |
+
"loss": 3.7537,
|
93 |
+
"step": 12000
|
94 |
+
},
|
95 |
+
{
|
96 |
+
"epoch": 0.7,
|
97 |
+
"grad_norm": 0.5088472366333008,
|
98 |
+
"learning_rate": 0.0004061875,
|
99 |
+
"loss": 3.7246,
|
100 |
+
"step": 13000
|
101 |
+
},
|
102 |
+
{
|
103 |
+
"epoch": 0.75,
|
104 |
+
"grad_norm": 0.4877682030200958,
|
105 |
+
"learning_rate": 0.00043737500000000005,
|
106 |
+
"loss": 3.6885,
|
107 |
+
"step": 14000
|
108 |
+
},
|
109 |
+
{
|
110 |
+
"epoch": 0.81,
|
111 |
+
"grad_norm": 0.45420414209365845,
|
112 |
+
"learning_rate": 0.000468625,
|
113 |
+
"loss": 3.6676,
|
114 |
+
"step": 15000
|
115 |
+
},
|
116 |
+
{
|
117 |
+
"epoch": 0.86,
|
118 |
+
"grad_norm": 0.44210752844810486,
|
119 |
+
"learning_rate": 0.000499875,
|
120 |
+
"loss": 3.6451,
|
121 |
+
"step": 16000
|
122 |
+
},
|
123 |
+
{
|
124 |
+
"epoch": 0.91,
|
125 |
+
"grad_norm": 0.41903868317604065,
|
126 |
+
"learning_rate": 0.000531125,
|
127 |
+
"loss": 3.6174,
|
128 |
+
"step": 17000
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"epoch": 0.97,
|
132 |
+
"grad_norm": 0.4097944498062134,
|
133 |
+
"learning_rate": 0.0005623749999999999,
|
134 |
+
"loss": 3.598,
|
135 |
+
"step": 18000
|
136 |
+
},
|
137 |
+
{
|
138 |
+
"epoch": 1.0,
|
139 |
+
"eval_accuracy": 0.35856486935076,
|
140 |
+
"eval_loss": 3.8055434226989746,
|
141 |
+
"eval_runtime": 153.4601,
|
142 |
+
"eval_samples_per_second": 377.434,
|
143 |
+
"eval_steps_per_second": 5.904,
|
144 |
+
"step": 18593
|
145 |
+
},
|
146 |
+
{
|
147 |
+
"epoch": 1.02,
|
148 |
+
"grad_norm": 0.36287716031074524,
|
149 |
+
"learning_rate": 0.0005935625,
|
150 |
+
"loss": 3.5686,
|
151 |
+
"step": 19000
|
152 |
+
},
|
153 |
+
{
|
154 |
+
"epoch": 1.08,
|
155 |
+
"grad_norm": 0.34093010425567627,
|
156 |
+
"learning_rate": 0.0006248125,
|
157 |
+
"loss": 3.5426,
|
158 |
+
"step": 20000
|
159 |
+
},
|
160 |
+
{
|
161 |
+
"epoch": 1.13,
|
162 |
+
"grad_norm": 0.3330960273742676,
|
163 |
+
"learning_rate": 0.0006560625,
|
164 |
+
"loss": 3.5376,
|
165 |
+
"step": 21000
|
166 |
+
},
|
167 |
+
{
|
168 |
+
"epoch": 1.18,
|
169 |
+
"grad_norm": 0.30257412791252136,
|
170 |
+
"learning_rate": 0.00068728125,
|
171 |
+
"loss": 3.5228,
|
172 |
+
"step": 22000
|
173 |
+
},
|
174 |
+
{
|
175 |
+
"epoch": 1.24,
|
176 |
+
"grad_norm": 0.2785021960735321,
|
177 |
+
"learning_rate": 0.00071853125,
|
178 |
+
"loss": 3.5095,
|
179 |
+
"step": 23000
|
180 |
+
},
|
181 |
+
{
|
182 |
+
"epoch": 1.29,
|
183 |
+
"grad_norm": 0.2798495292663574,
|
184 |
+
"learning_rate": 0.00074978125,
|
185 |
+
"loss": 3.502,
|
186 |
+
"step": 24000
|
187 |
+
},
|
188 |
+
{
|
189 |
+
"epoch": 1.34,
|
190 |
+
"grad_norm": 0.27469247579574585,
|
191 |
+
"learning_rate": 0.0007810312499999999,
|
192 |
+
"loss": 3.4931,
|
193 |
+
"step": 25000
|
194 |
+
},
|
195 |
+
{
|
196 |
+
"epoch": 1.4,
|
197 |
+
"grad_norm": 0.28253281116485596,
|
198 |
+
"learning_rate": 0.00081225,
|
199 |
+
"loss": 3.4818,
|
200 |
+
"step": 26000
|
201 |
+
},
|
202 |
+
{
|
203 |
+
"epoch": 1.45,
|
204 |
+
"grad_norm": 0.2425285279750824,
|
205 |
+
"learning_rate": 0.0008435000000000001,
|
206 |
+
"loss": 3.4757,
|
207 |
+
"step": 27000
|
208 |
+
},
|
209 |
+
{
|
210 |
+
"epoch": 1.51,
|
211 |
+
"grad_norm": 0.26154839992523193,
|
212 |
+
"learning_rate": 0.00087471875,
|
213 |
+
"loss": 3.465,
|
214 |
+
"step": 28000
|
215 |
+
},
|
216 |
+
{
|
217 |
+
"epoch": 1.56,
|
218 |
+
"grad_norm": 0.24984851479530334,
|
219 |
+
"learning_rate": 0.00090596875,
|
220 |
+
"loss": 3.4591,
|
221 |
+
"step": 29000
|
222 |
+
},
|
223 |
+
{
|
224 |
+
"epoch": 1.61,
|
225 |
+
"grad_norm": 0.24280162155628204,
|
226 |
+
"learning_rate": 0.00093721875,
|
227 |
+
"loss": 3.4481,
|
228 |
+
"step": 30000
|
229 |
+
},
|
230 |
+
{
|
231 |
+
"epoch": 1.67,
|
232 |
+
"grad_norm": 0.2471531629562378,
|
233 |
+
"learning_rate": 0.00096846875,
|
234 |
+
"loss": 3.4382,
|
235 |
+
"step": 31000
|
236 |
+
},
|
237 |
+
{
|
238 |
+
"epoch": 1.72,
|
239 |
+
"grad_norm": 0.21386803686618805,
|
240 |
+
"learning_rate": 0.0009996875,
|
241 |
+
"loss": 3.4374,
|
242 |
+
"step": 32000
|
243 |
+
},
|
244 |
+
{
|
245 |
+
"epoch": 1.77,
|
246 |
+
"grad_norm": 0.23056912422180176,
|
247 |
+
"learning_rate": 0.0009970870358382863,
|
248 |
+
"loss": 3.4262,
|
249 |
+
"step": 33000
|
250 |
+
},
|
251 |
+
{
|
252 |
+
"epoch": 1.83,
|
253 |
+
"grad_norm": 0.22734513878822327,
|
254 |
+
"learning_rate": 0.0009941446477961513,
|
255 |
+
"loss": 3.4206,
|
256 |
+
"step": 34000
|
257 |
+
},
|
258 |
+
{
|
259 |
+
"epoch": 1.88,
|
260 |
+
"grad_norm": 0.27341413497924805,
|
261 |
+
"learning_rate": 0.0009912052021420585,
|
262 |
+
"loss": 3.4028,
|
263 |
+
"step": 35000
|
264 |
+
},
|
265 |
+
{
|
266 |
+
"epoch": 1.94,
|
267 |
+
"grad_norm": 0.2197876125574112,
|
268 |
+
"learning_rate": 0.0009882628140999235,
|
269 |
+
"loss": 3.3941,
|
270 |
+
"step": 36000
|
271 |
+
},
|
272 |
+
{
|
273 |
+
"epoch": 1.99,
|
274 |
+
"grad_norm": 0.2100793868303299,
|
275 |
+
"learning_rate": 0.0009853204260577885,
|
276 |
+
"loss": 3.3819,
|
277 |
+
"step": 37000
|
278 |
+
},
|
279 |
+
{
|
280 |
+
"epoch": 2.0,
|
281 |
+
"eval_accuracy": 0.3806616512481747,
|
282 |
+
"eval_loss": 3.5619659423828125,
|
283 |
+
"eval_runtime": 155.2921,
|
284 |
+
"eval_samples_per_second": 372.981,
|
285 |
+
"eval_steps_per_second": 5.834,
|
286 |
+
"step": 37186
|
287 |
+
},
|
288 |
+
{
|
289 |
+
"epoch": 2.04,
|
290 |
+
"grad_norm": 0.2338385134935379,
|
291 |
+
"learning_rate": 0.0009823839227917379,
|
292 |
+
"loss": 3.3373,
|
293 |
+
"step": 38000
|
294 |
+
},
|
295 |
+
{
|
296 |
+
"epoch": 2.1,
|
297 |
+
"grad_norm": 0.20092932879924774,
|
298 |
+
"learning_rate": 0.0009794415347496028,
|
299 |
+
"loss": 3.3246,
|
300 |
+
"step": 39000
|
301 |
+
},
|
302 |
+
{
|
303 |
+
"epoch": 2.15,
|
304 |
+
"grad_norm": 0.2200673222541809,
|
305 |
+
"learning_rate": 0.0009764991467074677,
|
306 |
+
"loss": 3.3266,
|
307 |
+
"step": 40000
|
308 |
+
},
|
309 |
+
{
|
310 |
+
"epoch": 2.21,
|
311 |
+
"grad_norm": 0.2211606204509735,
|
312 |
+
"learning_rate": 0.0009735567586653328,
|
313 |
+
"loss": 3.3149,
|
314 |
+
"step": 41000
|
315 |
+
},
|
316 |
+
{
|
317 |
+
"epoch": 2.26,
|
318 |
+
"grad_norm": 0.19502712786197662,
|
319 |
+
"learning_rate": 0.0009706143706231978,
|
320 |
+
"loss": 3.3163,
|
321 |
+
"step": 42000
|
322 |
+
},
|
323 |
+
{
|
324 |
+
"epoch": 2.31,
|
325 |
+
"grad_norm": 0.2075120508670807,
|
326 |
+
"learning_rate": 0.000967674924969105,
|
327 |
+
"loss": 3.3074,
|
328 |
+
"step": 43000
|
329 |
+
},
|
330 |
+
{
|
331 |
+
"epoch": 2.37,
|
332 |
+
"grad_norm": 0.22068411111831665,
|
333 |
+
"learning_rate": 0.0009647325369269699,
|
334 |
+
"loss": 3.3013,
|
335 |
+
"step": 44000
|
336 |
+
},
|
337 |
+
{
|
338 |
+
"epoch": 2.42,
|
339 |
+
"grad_norm": 0.19794422388076782,
|
340 |
+
"learning_rate": 0.000961793091272877,
|
341 |
+
"loss": 3.3064,
|
342 |
+
"step": 45000
|
343 |
+
},
|
344 |
+
{
|
345 |
+
"epoch": 2.47,
|
346 |
+
"grad_norm": 0.2203867882490158,
|
347 |
+
"learning_rate": 0.0009588507032307421,
|
348 |
+
"loss": 3.2953,
|
349 |
+
"step": 46000
|
350 |
+
},
|
351 |
+
{
|
352 |
+
"epoch": 2.53,
|
353 |
+
"grad_norm": 0.20799222588539124,
|
354 |
+
"learning_rate": 0.0009559112575766492,
|
355 |
+
"loss": 3.2912,
|
356 |
+
"step": 47000
|
357 |
+
},
|
358 |
+
{
|
359 |
+
"epoch": 2.58,
|
360 |
+
"grad_norm": 0.22151567041873932,
|
361 |
+
"learning_rate": 0.0009529688695345143,
|
362 |
+
"loss": 3.2828,
|
363 |
+
"step": 48000
|
364 |
+
},
|
365 |
+
{
|
366 |
+
"epoch": 2.64,
|
367 |
+
"grad_norm": 0.21492493152618408,
|
368 |
+
"learning_rate": 0.0009500264814923793,
|
369 |
+
"loss": 3.2807,
|
370 |
+
"step": 49000
|
371 |
+
},
|
372 |
+
{
|
373 |
+
"epoch": 2.69,
|
374 |
+
"grad_norm": 0.22774606943130493,
|
375 |
+
"learning_rate": 0.0009470840934502442,
|
376 |
+
"loss": 3.277,
|
377 |
+
"step": 50000
|
378 |
+
},
|
379 |
+
{
|
380 |
+
"epoch": 2.74,
|
381 |
+
"grad_norm": 0.20704419910907745,
|
382 |
+
"learning_rate": 0.0009441417054081093,
|
383 |
+
"loss": 3.2756,
|
384 |
+
"step": 51000
|
385 |
+
},
|
386 |
+
{
|
387 |
+
"epoch": 2.8,
|
388 |
+
"grad_norm": 0.18878625333309174,
|
389 |
+
"learning_rate": 0.0009412052021420585,
|
390 |
+
"loss": 3.2698,
|
391 |
+
"step": 52000
|
392 |
+
},
|
393 |
+
{
|
394 |
+
"epoch": 2.85,
|
395 |
+
"grad_norm": 0.21784694492816925,
|
396 |
+
"learning_rate": 0.0009382628140999236,
|
397 |
+
"loss": 3.2633,
|
398 |
+
"step": 53000
|
399 |
+
},
|
400 |
+
{
|
401 |
+
"epoch": 2.9,
|
402 |
+
"grad_norm": 0.19541580975055695,
|
403 |
+
"learning_rate": 0.0009353204260577886,
|
404 |
+
"loss": 3.2603,
|
405 |
+
"step": 54000
|
406 |
+
},
|
407 |
+
{
|
408 |
+
"epoch": 2.96,
|
409 |
+
"grad_norm": 0.1930837631225586,
|
410 |
+
"learning_rate": 0.0009323780380156535,
|
411 |
+
"loss": 3.2622,
|
412 |
+
"step": 55000
|
413 |
+
},
|
414 |
+
{
|
415 |
+
"epoch": 3.0,
|
416 |
+
"eval_accuracy": 0.3929298036900791,
|
417 |
+
"eval_loss": 3.4625182151794434,
|
418 |
+
"eval_runtime": 154.3214,
|
419 |
+
"eval_samples_per_second": 375.327,
|
420 |
+
"eval_steps_per_second": 5.871,
|
421 |
+
"step": 55779
|
422 |
+
},
|
423 |
+
{
|
424 |
+
"epoch": 3.01,
|
425 |
+
"grad_norm": 0.20553229749202728,
|
426 |
+
"learning_rate": 0.0009294356499735185,
|
427 |
+
"loss": 3.2366,
|
428 |
+
"step": 56000
|
429 |
+
},
|
430 |
+
{
|
431 |
+
"epoch": 3.07,
|
432 |
+
"grad_norm": 0.20729704201221466,
|
433 |
+
"learning_rate": 0.0009264962043194257,
|
434 |
+
"loss": 3.1896,
|
435 |
+
"step": 57000
|
436 |
+
},
|
437 |
+
{
|
438 |
+
"epoch": 3.12,
|
439 |
+
"grad_norm": 0.20280729234218597,
|
440 |
+
"learning_rate": 0.0009235538162772906,
|
441 |
+
"loss": 3.1933,
|
442 |
+
"step": 58000
|
443 |
+
},
|
444 |
+
{
|
445 |
+
"epoch": 3.17,
|
446 |
+
"grad_norm": 0.1880486160516739,
|
447 |
+
"learning_rate": 0.0009206114282351557,
|
448 |
+
"loss": 3.1987,
|
449 |
+
"step": 59000
|
450 |
+
},
|
451 |
+
{
|
452 |
+
"epoch": 3.23,
|
453 |
+
"grad_norm": 0.2021629512310028,
|
454 |
+
"learning_rate": 0.0009176719825810627,
|
455 |
+
"loss": 3.2011,
|
456 |
+
"step": 60000
|
457 |
+
},
|
458 |
+
{
|
459 |
+
"epoch": 3.28,
|
460 |
+
"grad_norm": 0.24562139809131622,
|
461 |
+
"learning_rate": 0.0009147295945389278,
|
462 |
+
"loss": 3.1928,
|
463 |
+
"step": 61000
|
464 |
+
},
|
465 |
+
{
|
466 |
+
"epoch": 3.33,
|
467 |
+
"grad_norm": 0.18961690366268158,
|
468 |
+
"learning_rate": 0.0009117901488848349,
|
469 |
+
"loss": 3.1966,
|
470 |
+
"step": 62000
|
471 |
+
},
|
472 |
+
{
|
473 |
+
"epoch": 3.39,
|
474 |
+
"grad_norm": 0.2056853473186493,
|
475 |
+
"learning_rate": 0.0009088477608426999,
|
476 |
+
"loss": 3.1888,
|
477 |
+
"step": 63000
|
478 |
+
},
|
479 |
+
{
|
480 |
+
"epoch": 3.44,
|
481 |
+
"grad_norm": 0.18212585151195526,
|
482 |
+
"learning_rate": 0.000905905372800565,
|
483 |
+
"loss": 3.1906,
|
484 |
+
"step": 64000
|
485 |
+
},
|
486 |
+
{
|
487 |
+
"epoch": 3.5,
|
488 |
+
"grad_norm": 0.21010248363018036,
|
489 |
+
"learning_rate": 0.000902965927146472,
|
490 |
+
"loss": 3.1925,
|
491 |
+
"step": 65000
|
492 |
+
},
|
493 |
+
{
|
494 |
+
"epoch": 3.55,
|
495 |
+
"grad_norm": 0.21237117052078247,
|
496 |
+
"learning_rate": 0.0009000235391043371,
|
497 |
+
"loss": 3.1918,
|
498 |
+
"step": 66000
|
499 |
+
},
|
500 |
+
{
|
501 |
+
"epoch": 3.6,
|
502 |
+
"grad_norm": 0.24303843080997467,
|
503 |
+
"learning_rate": 0.0008970840934502442,
|
504 |
+
"loss": 3.1918,
|
505 |
+
"step": 67000
|
506 |
+
},
|
507 |
+
{
|
508 |
+
"epoch": 3.66,
|
509 |
+
"grad_norm": 0.19125622510910034,
|
510 |
+
"learning_rate": 0.0008941417054081092,
|
511 |
+
"loss": 3.1866,
|
512 |
+
"step": 68000
|
513 |
+
},
|
514 |
+
{
|
515 |
+
"epoch": 3.71,
|
516 |
+
"grad_norm": 0.23382623493671417,
|
517 |
+
"learning_rate": 0.0008912022597540164,
|
518 |
+
"loss": 3.1844,
|
519 |
+
"step": 69000
|
520 |
+
},
|
521 |
+
{
|
522 |
+
"epoch": 3.76,
|
523 |
+
"grad_norm": 0.21234822273254395,
|
524 |
+
"learning_rate": 0.0008882598717118814,
|
525 |
+
"loss": 3.1852,
|
526 |
+
"step": 70000
|
527 |
+
},
|
528 |
+
{
|
529 |
+
"epoch": 3.82,
|
530 |
+
"grad_norm": 0.2681824564933777,
|
531 |
+
"learning_rate": 0.0008853204260577885,
|
532 |
+
"loss": 3.1855,
|
533 |
+
"step": 71000
|
534 |
+
},
|
535 |
+
{
|
536 |
+
"epoch": 3.87,
|
537 |
+
"grad_norm": 0.19967304170131683,
|
538 |
+
"learning_rate": 0.0008823780380156535,
|
539 |
+
"loss": 3.1781,
|
540 |
+
"step": 72000
|
541 |
+
},
|
542 |
+
{
|
543 |
+
"epoch": 3.93,
|
544 |
+
"grad_norm": 0.19710466265678406,
|
545 |
+
"learning_rate": 0.0008794385923615607,
|
546 |
+
"loss": 3.1804,
|
547 |
+
"step": 73000
|
548 |
+
},
|
549 |
+
{
|
550 |
+
"epoch": 3.98,
|
551 |
+
"grad_norm": 0.2174214869737625,
|
552 |
+
"learning_rate": 0.0008764962043194257,
|
553 |
+
"loss": 3.1788,
|
554 |
+
"step": 74000
|
555 |
+
},
|
556 |
+
{
|
557 |
+
"epoch": 4.0,
|
558 |
+
"eval_accuracy": 0.39836750674153854,
|
559 |
+
"eval_loss": 3.4131386280059814,
|
560 |
+
"eval_runtime": 154.2916,
|
561 |
+
"eval_samples_per_second": 375.4,
|
562 |
+
"eval_steps_per_second": 5.872,
|
563 |
+
"step": 74372
|
564 |
+
},
|
565 |
+
{
|
566 |
+
"epoch": 4.03,
|
567 |
+
"grad_norm": 0.23124143481254578,
|
568 |
+
"learning_rate": 0.0008735567586653328,
|
569 |
+
"loss": 3.1339,
|
570 |
+
"step": 75000
|
571 |
+
},
|
572 |
+
{
|
573 |
+
"epoch": 4.09,
|
574 |
+
"grad_norm": 0.2329677790403366,
|
575 |
+
"learning_rate": 0.0008706143706231978,
|
576 |
+
"loss": 3.1142,
|
577 |
+
"step": 76000
|
578 |
+
},
|
579 |
+
{
|
580 |
+
"epoch": 4.14,
|
581 |
+
"grad_norm": 0.21693040430545807,
|
582 |
+
"learning_rate": 0.0008676749249691049,
|
583 |
+
"loss": 3.1174,
|
584 |
+
"step": 77000
|
585 |
+
},
|
586 |
+
{
|
587 |
+
"epoch": 4.2,
|
588 |
+
"grad_norm": 0.2263847142457962,
|
589 |
+
"learning_rate": 0.00086473253692697,
|
590 |
+
"loss": 3.1182,
|
591 |
+
"step": 78000
|
592 |
+
},
|
593 |
+
{
|
594 |
+
"epoch": 4.25,
|
595 |
+
"grad_norm": 0.23874357342720032,
|
596 |
+
"learning_rate": 0.0008617930912728771,
|
597 |
+
"loss": 3.1247,
|
598 |
+
"step": 79000
|
599 |
+
},
|
600 |
+
{
|
601 |
+
"epoch": 4.3,
|
602 |
+
"grad_norm": 0.21914631128311157,
|
603 |
+
"learning_rate": 0.0008588507032307421,
|
604 |
+
"loss": 3.1239,
|
605 |
+
"step": 80000
|
606 |
+
},
|
607 |
+
{
|
608 |
+
"epoch": 4.36,
|
609 |
+
"grad_norm": 0.24384135007858276,
|
610 |
+
"learning_rate": 0.0008559083151886072,
|
611 |
+
"loss": 3.1273,
|
612 |
+
"step": 81000
|
613 |
+
},
|
614 |
+
{
|
615 |
+
"epoch": 4.41,
|
616 |
+
"grad_norm": 0.20350953936576843,
|
617 |
+
"learning_rate": 0.0008529688695345142,
|
618 |
+
"loss": 3.1251,
|
619 |
+
"step": 82000
|
620 |
+
},
|
621 |
+
{
|
622 |
+
"epoch": 4.46,
|
623 |
+
"grad_norm": 0.19013893604278564,
|
624 |
+
"learning_rate": 0.0008500264814923793,
|
625 |
+
"loss": 3.1249,
|
626 |
+
"step": 83000
|
627 |
+
},
|
628 |
+
{
|
629 |
+
"epoch": 4.52,
|
630 |
+
"grad_norm": 0.20296384394168854,
|
631 |
+
"learning_rate": 0.0008470840934502443,
|
632 |
+
"loss": 3.1283,
|
633 |
+
"step": 84000
|
634 |
+
},
|
635 |
+
{
|
636 |
+
"epoch": 4.57,
|
637 |
+
"grad_norm": 0.19276568293571472,
|
638 |
+
"learning_rate": 0.0008441417054081092,
|
639 |
+
"loss": 3.1257,
|
640 |
+
"step": 85000
|
641 |
+
},
|
642 |
+
{
|
643 |
+
"epoch": 4.63,
|
644 |
+
"grad_norm": 0.196748286485672,
|
645 |
+
"learning_rate": 0.0008412022597540165,
|
646 |
+
"loss": 3.1242,
|
647 |
+
"step": 86000
|
648 |
+
},
|
649 |
+
{
|
650 |
+
"epoch": 4.68,
|
651 |
+
"grad_norm": 0.20028959214687347,
|
652 |
+
"learning_rate": 0.0008382598717118813,
|
653 |
+
"loss": 3.1246,
|
654 |
+
"step": 87000
|
655 |
+
},
|
656 |
+
{
|
657 |
+
"epoch": 4.73,
|
658 |
+
"grad_norm": 0.20650836825370789,
|
659 |
+
"learning_rate": 0.0008353174836697463,
|
660 |
+
"loss": 3.1273,
|
661 |
+
"step": 88000
|
662 |
+
},
|
663 |
+
{
|
664 |
+
"epoch": 4.79,
|
665 |
+
"grad_norm": 0.2691803574562073,
|
666 |
+
"learning_rate": 0.0008323780380156535,
|
667 |
+
"loss": 3.1219,
|
668 |
+
"step": 89000
|
669 |
+
},
|
670 |
+
{
|
671 |
+
"epoch": 4.84,
|
672 |
+
"grad_norm": 0.20094037055969238,
|
673 |
+
"learning_rate": 0.0008294385923615606,
|
674 |
+
"loss": 3.1236,
|
675 |
+
"step": 90000
|
676 |
+
},
|
677 |
+
{
|
678 |
+
"epoch": 4.89,
|
679 |
+
"grad_norm": 0.23795652389526367,
|
680 |
+
"learning_rate": 0.0008264962043194257,
|
681 |
+
"loss": 3.1256,
|
682 |
+
"step": 91000
|
683 |
+
},
|
684 |
+
{
|
685 |
+
"epoch": 4.95,
|
686 |
+
"grad_norm": 0.20850218832492828,
|
687 |
+
"learning_rate": 0.0008235538162772906,
|
688 |
+
"loss": 3.1251,
|
689 |
+
"step": 92000
|
690 |
+
},
|
691 |
+
{
|
692 |
+
"epoch": 5.0,
|
693 |
+
"eval_accuracy": 0.40224626364489346,
|
694 |
+
"eval_loss": 3.385948419570923,
|
695 |
+
"eval_runtime": 154.3892,
|
696 |
+
"eval_samples_per_second": 375.162,
|
697 |
+
"eval_steps_per_second": 5.868,
|
698 |
+
"step": 92965
|
699 |
+
},
|
700 |
+
{
|
701 |
+
"epoch": 5.0,
|
702 |
+
"grad_norm": 0.19185079634189606,
|
703 |
+
"learning_rate": 0.0008206114282351556,
|
704 |
+
"loss": 3.1207,
|
705 |
+
"step": 93000
|
706 |
+
},
|
707 |
+
{
|
708 |
+
"epoch": 5.06,
|
709 |
+
"grad_norm": 0.2044549584388733,
|
710 |
+
"learning_rate": 0.0008176719825810628,
|
711 |
+
"loss": 3.0552,
|
712 |
+
"step": 94000
|
713 |
+
},
|
714 |
+
{
|
715 |
+
"epoch": 5.11,
|
716 |
+
"grad_norm": 0.2207447588443756,
|
717 |
+
"learning_rate": 0.0008147295945389278,
|
718 |
+
"loss": 3.0595,
|
719 |
+
"step": 95000
|
720 |
+
},
|
721 |
+
{
|
722 |
+
"epoch": 5.16,
|
723 |
+
"grad_norm": 0.2116185575723648,
|
724 |
+
"learning_rate": 0.0008117872064967929,
|
725 |
+
"loss": 3.0654,
|
726 |
+
"step": 96000
|
727 |
+
},
|
728 |
+
{
|
729 |
+
"epoch": 5.22,
|
730 |
+
"grad_norm": 0.20827573537826538,
|
731 |
+
"learning_rate": 0.0008088477608427,
|
732 |
+
"loss": 3.0653,
|
733 |
+
"step": 97000
|
734 |
+
},
|
735 |
+
{
|
736 |
+
"epoch": 5.27,
|
737 |
+
"grad_norm": 0.21885184943675995,
|
738 |
+
"learning_rate": 0.0008059053728005649,
|
739 |
+
"loss": 3.0689,
|
740 |
+
"step": 98000
|
741 |
+
},
|
742 |
+
{
|
743 |
+
"epoch": 5.32,
|
744 |
+
"grad_norm": 0.22063520550727844,
|
745 |
+
"learning_rate": 0.0008029659271464721,
|
746 |
+
"loss": 3.0734,
|
747 |
+
"step": 99000
|
748 |
+
},
|
749 |
+
{
|
750 |
+
"epoch": 5.38,
|
751 |
+
"grad_norm": 0.21756470203399658,
|
752 |
+
"learning_rate": 0.0008000235391043371,
|
753 |
+
"loss": 3.0761,
|
754 |
+
"step": 100000
|
755 |
+
},
|
756 |
+
{
|
757 |
+
"epoch": 5.43,
|
758 |
+
"grad_norm": 0.2308843582868576,
|
759 |
+
"learning_rate": 0.0007970811510622022,
|
760 |
+
"loss": 3.0764,
|
761 |
+
"step": 101000
|
762 |
+
},
|
763 |
+
{
|
764 |
+
"epoch": 5.49,
|
765 |
+
"grad_norm": 0.2255343496799469,
|
766 |
+
"learning_rate": 0.0007941387630200672,
|
767 |
+
"loss": 3.0786,
|
768 |
+
"step": 102000
|
769 |
+
},
|
770 |
+
{
|
771 |
+
"epoch": 5.54,
|
772 |
+
"grad_norm": 0.21067161858081818,
|
773 |
+
"learning_rate": 0.0007912022597540164,
|
774 |
+
"loss": 3.0786,
|
775 |
+
"step": 103000
|
776 |
+
},
|
777 |
+
{
|
778 |
+
"epoch": 5.59,
|
779 |
+
"grad_norm": 0.21768967807292938,
|
780 |
+
"learning_rate": 0.0007882598717118814,
|
781 |
+
"loss": 3.0773,
|
782 |
+
"step": 104000
|
783 |
+
},
|
784 |
+
{
|
785 |
+
"epoch": 5.65,
|
786 |
+
"grad_norm": 0.20667089521884918,
|
787 |
+
"learning_rate": 0.0007853174836697464,
|
788 |
+
"loss": 3.0795,
|
789 |
+
"step": 105000
|
790 |
+
},
|
791 |
+
{
|
792 |
+
"epoch": 5.7,
|
793 |
+
"grad_norm": 0.21367168426513672,
|
794 |
+
"learning_rate": 0.0007823750956276114,
|
795 |
+
"loss": 3.082,
|
796 |
+
"step": 106000
|
797 |
+
},
|
798 |
+
{
|
799 |
+
"epoch": 5.75,
|
800 |
+
"grad_norm": 0.2038014829158783,
|
801 |
+
"learning_rate": 0.0007794327075854764,
|
802 |
+
"loss": 3.0785,
|
803 |
+
"step": 107000
|
804 |
+
},
|
805 |
+
{
|
806 |
+
"epoch": 5.81,
|
807 |
+
"grad_norm": 0.21575333178043365,
|
808 |
+
"learning_rate": 0.0007764932619313834,
|
809 |
+
"loss": 3.0808,
|
810 |
+
"step": 108000
|
811 |
+
},
|
812 |
+
{
|
813 |
+
"epoch": 5.86,
|
814 |
+
"grad_norm": 0.20344999432563782,
|
815 |
+
"learning_rate": 0.0007735508738892485,
|
816 |
+
"loss": 3.0829,
|
817 |
+
"step": 109000
|
818 |
+
},
|
819 |
+
{
|
820 |
+
"epoch": 5.92,
|
821 |
+
"grad_norm": 0.2332640290260315,
|
822 |
+
"learning_rate": 0.0007706114282351556,
|
823 |
+
"loss": 3.0769,
|
824 |
+
"step": 110000
|
825 |
+
},
|
826 |
+
{
|
827 |
+
"epoch": 5.97,
|
828 |
+
"grad_norm": 0.23659543693065643,
|
829 |
+
"learning_rate": 0.0007676690401930207,
|
830 |
+
"loss": 3.0765,
|
831 |
+
"step": 111000
|
832 |
+
},
|
833 |
+
{
|
834 |
+
"epoch": 6.0,
|
835 |
+
"eval_accuracy": 0.4050011228899796,
|
836 |
+
"eval_loss": 3.3707284927368164,
|
837 |
+
"eval_runtime": 153.9554,
|
838 |
+
"eval_samples_per_second": 376.219,
|
839 |
+
"eval_steps_per_second": 5.885,
|
840 |
+
"step": 111558
|
841 |
+
},
|
842 |
+
{
|
843 |
+
"epoch": 6.02,
|
844 |
+
"grad_norm": 0.21739870309829712,
|
845 |
+
"learning_rate": 0.0007647266521508857,
|
846 |
+
"loss": 3.0492,
|
847 |
+
"step": 112000
|
848 |
+
},
|
849 |
+
{
|
850 |
+
"epoch": 6.08,
|
851 |
+
"grad_norm": 0.23045282065868378,
|
852 |
+
"learning_rate": 0.0007617872064967927,
|
853 |
+
"loss": 3.0123,
|
854 |
+
"step": 113000
|
855 |
+
},
|
856 |
+
{
|
857 |
+
"epoch": 6.13,
|
858 |
+
"grad_norm": 0.2114897221326828,
|
859 |
+
"learning_rate": 0.0007588448184546578,
|
860 |
+
"loss": 3.0196,
|
861 |
+
"step": 114000
|
862 |
+
},
|
863 |
+
{
|
864 |
+
"epoch": 6.19,
|
865 |
+
"grad_norm": 0.21402007341384888,
|
866 |
+
"learning_rate": 0.0007559053728005649,
|
867 |
+
"loss": 3.0229,
|
868 |
+
"step": 115000
|
869 |
+
},
|
870 |
+
{
|
871 |
+
"epoch": 6.24,
|
872 |
+
"grad_norm": 0.22187209129333496,
|
873 |
+
"learning_rate": 0.00075296298475843,
|
874 |
+
"loss": 3.0269,
|
875 |
+
"step": 116000
|
876 |
+
},
|
877 |
+
{
|
878 |
+
"epoch": 6.29,
|
879 |
+
"grad_norm": 0.2102159708738327,
|
880 |
+
"learning_rate": 0.000750020596716295,
|
881 |
+
"loss": 3.0273,
|
882 |
+
"step": 117000
|
883 |
+
},
|
884 |
+
{
|
885 |
+
"epoch": 6.35,
|
886 |
+
"grad_norm": 0.2769564092159271,
|
887 |
+
"learning_rate": 0.000747081151062202,
|
888 |
+
"loss": 3.0318,
|
889 |
+
"step": 118000
|
890 |
+
},
|
891 |
+
{
|
892 |
+
"epoch": 6.4,
|
893 |
+
"grad_norm": 0.2173321396112442,
|
894 |
+
"learning_rate": 0.0007441387630200671,
|
895 |
+
"loss": 3.0331,
|
896 |
+
"step": 119000
|
897 |
+
},
|
898 |
+
{
|
899 |
+
"epoch": 6.45,
|
900 |
+
"grad_norm": 0.24418728053569794,
|
901 |
+
"learning_rate": 0.0007411993173659742,
|
902 |
+
"loss": 3.0376,
|
903 |
+
"step": 120000
|
904 |
+
},
|
905 |
+
{
|
906 |
+
"epoch": 6.51,
|
907 |
+
"grad_norm": 0.23827865719795227,
|
908 |
+
"learning_rate": 0.0007382569293238393,
|
909 |
+
"loss": 3.0404,
|
910 |
+
"step": 121000
|
911 |
+
},
|
912 |
+
{
|
913 |
+
"epoch": 6.56,
|
914 |
+
"grad_norm": 0.2125864326953888,
|
915 |
+
"learning_rate": 0.0007353174836697464,
|
916 |
+
"loss": 3.0366,
|
917 |
+
"step": 122000
|
918 |
+
},
|
919 |
+
{
|
920 |
+
"epoch": 6.62,
|
921 |
+
"grad_norm": 0.21209298074245453,
|
922 |
+
"learning_rate": 0.0007323750956276114,
|
923 |
+
"loss": 3.0408,
|
924 |
+
"step": 123000
|
925 |
+
},
|
926 |
+
{
|
927 |
+
"epoch": 6.67,
|
928 |
+
"grad_norm": 0.21926939487457275,
|
929 |
+
"learning_rate": 0.0007294327075854764,
|
930 |
+
"loss": 3.0435,
|
931 |
+
"step": 124000
|
932 |
+
},
|
933 |
+
{
|
934 |
+
"epoch": 6.72,
|
935 |
+
"grad_norm": 0.2240251749753952,
|
936 |
+
"learning_rate": 0.0007264932619313835,
|
937 |
+
"loss": 3.0439,
|
938 |
+
"step": 125000
|
939 |
+
},
|
940 |
+
{
|
941 |
+
"epoch": 6.78,
|
942 |
+
"grad_norm": 0.21099534630775452,
|
943 |
+
"learning_rate": 0.0007235508738892486,
|
944 |
+
"loss": 3.0415,
|
945 |
+
"step": 126000
|
946 |
+
},
|
947 |
+
{
|
948 |
+
"epoch": 6.83,
|
949 |
+
"grad_norm": 0.20424066483974457,
|
950 |
+
"learning_rate": 0.0007206084858471136,
|
951 |
+
"loss": 3.0444,
|
952 |
+
"step": 127000
|
953 |
+
},
|
954 |
+
{
|
955 |
+
"epoch": 6.88,
|
956 |
+
"grad_norm": 0.22034712135791779,
|
957 |
+
"learning_rate": 0.0007176660978049785,
|
958 |
+
"loss": 3.0448,
|
959 |
+
"step": 128000
|
960 |
+
},
|
961 |
+
{
|
962 |
+
"epoch": 6.94,
|
963 |
+
"grad_norm": 0.23952622711658478,
|
964 |
+
"learning_rate": 0.0007147237097628436,
|
965 |
+
"loss": 3.0427,
|
966 |
+
"step": 129000
|
967 |
+
},
|
968 |
+
{
|
969 |
+
"epoch": 6.99,
|
970 |
+
"grad_norm": 0.21370893716812134,
|
971 |
+
"learning_rate": 0.0007117842641087506,
|
972 |
+
"loss": 3.0458,
|
973 |
+
"step": 130000
|
974 |
+
},
|
975 |
+
{
|
976 |
+
"epoch": 7.0,
|
977 |
+
"eval_accuracy": 0.4076835636248676,
|
978 |
+
"eval_loss": 3.3522255420684814,
|
979 |
+
"eval_runtime": 154.4925,
|
980 |
+
"eval_samples_per_second": 374.911,
|
981 |
+
"eval_steps_per_second": 5.864,
|
982 |
+
"step": 130151
|
983 |
+
},
|
984 |
+
{
|
985 |
+
"epoch": 7.05,
|
986 |
+
"grad_norm": 0.22439521551132202,
|
987 |
+
"learning_rate": 0.0007088448184546579,
|
988 |
+
"loss": 2.9871,
|
989 |
+
"step": 131000
|
990 |
+
},
|
991 |
+
{
|
992 |
+
"epoch": 7.1,
|
993 |
+
"grad_norm": 0.24310387670993805,
|
994 |
+
"learning_rate": 0.0007059024304125228,
|
995 |
+
"loss": 2.98,
|
996 |
+
"step": 132000
|
997 |
+
},
|
998 |
+
{
|
999 |
+
"epoch": 7.15,
|
1000 |
+
"grad_norm": 0.22332601249217987,
|
1001 |
+
"learning_rate": 0.0007029600423703878,
|
1002 |
+
"loss": 2.9863,
|
1003 |
+
"step": 133000
|
1004 |
+
},
|
1005 |
+
{
|
1006 |
+
"epoch": 7.21,
|
1007 |
+
"grad_norm": 0.24013490974903107,
|
1008 |
+
"learning_rate": 0.0007000176543282529,
|
1009 |
+
"loss": 2.9879,
|
1010 |
+
"step": 134000
|
1011 |
+
},
|
1012 |
+
{
|
1013 |
+
"epoch": 7.26,
|
1014 |
+
"grad_norm": 0.23142094910144806,
|
1015 |
+
"learning_rate": 0.0006970782086741599,
|
1016 |
+
"loss": 2.9914,
|
1017 |
+
"step": 135000
|
1018 |
+
},
|
1019 |
+
{
|
1020 |
+
"epoch": 7.31,
|
1021 |
+
"grad_norm": 0.236104354262352,
|
1022 |
+
"learning_rate": 0.0006941358206320249,
|
1023 |
+
"loss": 2.9964,
|
1024 |
+
"step": 136000
|
1025 |
+
},
|
1026 |
+
{
|
1027 |
+
"epoch": 7.37,
|
1028 |
+
"grad_norm": 0.33047351241111755,
|
1029 |
+
"learning_rate": 0.0006911963749779321,
|
1030 |
+
"loss": 2.9986,
|
1031 |
+
"step": 137000
|
1032 |
+
},
|
1033 |
+
{
|
1034 |
+
"epoch": 7.42,
|
1035 |
+
"grad_norm": 0.26023513078689575,
|
1036 |
+
"learning_rate": 0.0006882539869357971,
|
1037 |
+
"loss": 2.9995,
|
1038 |
+
"step": 138000
|
1039 |
+
},
|
1040 |
+
{
|
1041 |
+
"epoch": 7.48,
|
1042 |
+
"grad_norm": 0.26877066493034363,
|
1043 |
+
"learning_rate": 0.0006853115988936622,
|
1044 |
+
"loss": 3.0005,
|
1045 |
+
"step": 139000
|
1046 |
+
},
|
1047 |
+
{
|
1048 |
+
"epoch": 7.53,
|
1049 |
+
"grad_norm": 0.2331291288137436,
|
1050 |
+
"learning_rate": 0.0006823721532395692,
|
1051 |
+
"loss": 3.0057,
|
1052 |
+
"step": 140000
|
1053 |
+
},
|
1054 |
+
{
|
1055 |
+
"epoch": 7.58,
|
1056 |
+
"grad_norm": 0.21710553765296936,
|
1057 |
+
"learning_rate": 0.0006794297651974342,
|
1058 |
+
"loss": 3.0107,
|
1059 |
+
"step": 141000
|
1060 |
+
},
|
1061 |
+
{
|
1062 |
+
"epoch": 7.64,
|
1063 |
+
"grad_norm": 0.24514536559581757,
|
1064 |
+
"learning_rate": 0.0006764903195433414,
|
1065 |
+
"loss": 3.0055,
|
1066 |
+
"step": 142000
|
1067 |
+
},
|
1068 |
+
{
|
1069 |
+
"epoch": 7.69,
|
1070 |
+
"grad_norm": 0.21572381258010864,
|
1071 |
+
"learning_rate": 0.0006735508738892485,
|
1072 |
+
"loss": 3.0102,
|
1073 |
+
"step": 143000
|
1074 |
+
},
|
1075 |
+
{
|
1076 |
+
"epoch": 7.74,
|
1077 |
+
"grad_norm": 0.23894068598747253,
|
1078 |
+
"learning_rate": 0.0006706084858471136,
|
1079 |
+
"loss": 3.0101,
|
1080 |
+
"step": 144000
|
1081 |
+
},
|
1082 |
+
{
|
1083 |
+
"epoch": 7.8,
|
1084 |
+
"grad_norm": 0.20779648423194885,
|
1085 |
+
"learning_rate": 0.0006676660978049786,
|
1086 |
+
"loss": 3.0136,
|
1087 |
+
"step": 145000
|
1088 |
+
},
|
1089 |
+
{
|
1090 |
+
"epoch": 7.85,
|
1091 |
+
"grad_norm": 0.21832165122032166,
|
1092 |
+
"learning_rate": 0.0006647237097628435,
|
1093 |
+
"loss": 3.0144,
|
1094 |
+
"step": 146000
|
1095 |
+
},
|
1096 |
+
{
|
1097 |
+
"epoch": 7.91,
|
1098 |
+
"grad_norm": 0.2295515537261963,
|
1099 |
+
"learning_rate": 0.0006617813217207086,
|
1100 |
+
"loss": 3.0162,
|
1101 |
+
"step": 147000
|
1102 |
+
},
|
1103 |
+
{
|
1104 |
+
"epoch": 7.96,
|
1105 |
+
"grad_norm": 0.21139173209667206,
|
1106 |
+
"learning_rate": 0.0006588418760666157,
|
1107 |
+
"loss": 3.0116,
|
1108 |
+
"step": 148000
|
1109 |
+
},
|
1110 |
+
{
|
1111 |
+
"epoch": 8.0,
|
1112 |
+
"eval_accuracy": 0.4079684679864168,
|
1113 |
+
"eval_loss": 3.3560173511505127,
|
1114 |
+
"eval_runtime": 154.3168,
|
1115 |
+
"eval_samples_per_second": 375.338,
|
1116 |
+
"eval_steps_per_second": 5.871,
|
1117 |
+
"step": 148744
|
1118 |
+
},
|
1119 |
+
{
|
1120 |
+
"epoch": 8.01,
|
1121 |
+
"grad_norm": 0.2352878898382187,
|
1122 |
+
"learning_rate": 0.0006558994880244807,
|
1123 |
+
"loss": 2.9921,
|
1124 |
+
"step": 149000
|
1125 |
+
},
|
1126 |
+
{
|
1127 |
+
"epoch": 8.07,
|
1128 |
+
"grad_norm": 0.26401156187057495,
|
1129 |
+
"learning_rate": 0.0006529600423703879,
|
1130 |
+
"loss": 2.9483,
|
1131 |
+
"step": 150000
|
1132 |
+
},
|
1133 |
+
{
|
1134 |
+
"epoch": 8.12,
|
1135 |
+
"grad_norm": 0.2650126814842224,
|
1136 |
+
"learning_rate": 0.0006500176543282527,
|
1137 |
+
"loss": 2.9538,
|
1138 |
+
"step": 151000
|
1139 |
+
},
|
1140 |
+
{
|
1141 |
+
"epoch": 8.18,
|
1142 |
+
"grad_norm": 0.2173539251089096,
|
1143 |
+
"learning_rate": 0.00064707820867416,
|
1144 |
+
"loss": 2.9592,
|
1145 |
+
"step": 152000
|
1146 |
+
},
|
1147 |
+
{
|
1148 |
+
"epoch": 8.23,
|
1149 |
+
"grad_norm": 0.23582415282726288,
|
1150 |
+
"learning_rate": 0.0006441358206320249,
|
1151 |
+
"loss": 2.9674,
|
1152 |
+
"step": 153000
|
1153 |
+
},
|
1154 |
+
{
|
1155 |
+
"epoch": 8.28,
|
1156 |
+
"grad_norm": 0.24998264014720917,
|
1157 |
+
"learning_rate": 0.0006411963749779322,
|
1158 |
+
"loss": 2.9634,
|
1159 |
+
"step": 154000
|
1160 |
+
},
|
1161 |
+
{
|
1162 |
+
"epoch": 8.34,
|
1163 |
+
"grad_norm": 0.2233651876449585,
|
1164 |
+
"learning_rate": 0.0006382539869357971,
|
1165 |
+
"loss": 2.9725,
|
1166 |
+
"step": 155000
|
1167 |
+
},
|
1168 |
+
{
|
1169 |
+
"epoch": 8.39,
|
1170 |
+
"grad_norm": 0.2446517050266266,
|
1171 |
+
"learning_rate": 0.000635311598893662,
|
1172 |
+
"loss": 2.9668,
|
1173 |
+
"step": 156000
|
1174 |
+
},
|
1175 |
+
{
|
1176 |
+
"epoch": 8.44,
|
1177 |
+
"grad_norm": 0.27410104870796204,
|
1178 |
+
"learning_rate": 0.0006323692108515271,
|
1179 |
+
"loss": 2.9716,
|
1180 |
+
"step": 157000
|
1181 |
+
},
|
1182 |
+
{
|
1183 |
+
"epoch": 8.5,
|
1184 |
+
"grad_norm": 0.19913798570632935,
|
1185 |
+
"learning_rate": 0.0006294268228093921,
|
1186 |
+
"loss": 2.9744,
|
1187 |
+
"step": 158000
|
1188 |
+
},
|
1189 |
+
{
|
1190 |
+
"epoch": 8.55,
|
1191 |
+
"grad_norm": 0.24324330687522888,
|
1192 |
+
"learning_rate": 0.0006264844347672571,
|
1193 |
+
"loss": 2.975,
|
1194 |
+
"step": 159000
|
1195 |
+
},
|
1196 |
+
{
|
1197 |
+
"epoch": 8.61,
|
1198 |
+
"grad_norm": 0.2302367389202118,
|
1199 |
+
"learning_rate": 0.0006235479315012064,
|
1200 |
+
"loss": 2.9765,
|
1201 |
+
"step": 160000
|
1202 |
+
},
|
1203 |
+
{
|
1204 |
+
"epoch": 8.66,
|
1205 |
+
"grad_norm": 0.25450703501701355,
|
1206 |
+
"learning_rate": 0.0006206055434590713,
|
1207 |
+
"loss": 2.9762,
|
1208 |
+
"step": 161000
|
1209 |
+
},
|
1210 |
+
{
|
1211 |
+
"epoch": 8.71,
|
1212 |
+
"grad_norm": 0.2539065480232239,
|
1213 |
+
"learning_rate": 0.0006176631554169364,
|
1214 |
+
"loss": 2.9819,
|
1215 |
+
"step": 162000
|
1216 |
+
},
|
1217 |
+
{
|
1218 |
+
"epoch": 8.77,
|
1219 |
+
"grad_norm": 0.2278948873281479,
|
1220 |
+
"learning_rate": 0.0006147207673748014,
|
1221 |
+
"loss": 2.9809,
|
1222 |
+
"step": 163000
|
1223 |
+
},
|
1224 |
+
{
|
1225 |
+
"epoch": 8.82,
|
1226 |
+
"grad_norm": 0.23980122804641724,
|
1227 |
+
"learning_rate": 0.0006117813217207086,
|
1228 |
+
"loss": 2.9829,
|
1229 |
+
"step": 164000
|
1230 |
+
},
|
1231 |
+
{
|
1232 |
+
"epoch": 8.87,
|
1233 |
+
"grad_norm": 0.25505828857421875,
|
1234 |
+
"learning_rate": 0.0006088389336785736,
|
1235 |
+
"loss": 2.9845,
|
1236 |
+
"step": 165000
|
1237 |
+
},
|
1238 |
+
{
|
1239 |
+
"epoch": 8.93,
|
1240 |
+
"grad_norm": 0.26612409949302673,
|
1241 |
+
"learning_rate": 0.0006058965456364386,
|
1242 |
+
"loss": 2.9874,
|
1243 |
+
"step": 166000
|
1244 |
+
},
|
1245 |
+
{
|
1246 |
+
"epoch": 8.98,
|
1247 |
+
"grad_norm": 0.2189650535583496,
|
1248 |
+
"learning_rate": 0.0006029541575943035,
|
1249 |
+
"loss": 2.9832,
|
1250 |
+
"step": 167000
|
1251 |
+
},
|
1252 |
+
{
|
1253 |
+
"epoch": 9.0,
|
1254 |
+
"eval_accuracy": 0.40796309369994566,
|
1255 |
+
"eval_loss": 3.3773229122161865,
|
1256 |
+
"eval_runtime": 154.2001,
|
1257 |
+
"eval_samples_per_second": 375.622,
|
1258 |
+
"eval_steps_per_second": 5.875,
|
1259 |
+
"step": 167337
|
1260 |
+
},
|
1261 |
+
{
|
1262 |
+
"epoch": 9.04,
|
1263 |
+
"grad_norm": 0.27406546473503113,
|
1264 |
+
"learning_rate": 0.0006000147119402107,
|
1265 |
+
"loss": 2.9377,
|
1266 |
+
"step": 168000
|
1267 |
+
},
|
1268 |
+
{
|
1269 |
+
"epoch": 9.09,
|
1270 |
+
"grad_norm": 0.2886415719985962,
|
1271 |
+
"learning_rate": 0.0005970723238980756,
|
1272 |
+
"loss": 2.9216,
|
1273 |
+
"step": 169000
|
1274 |
+
},
|
1275 |
+
{
|
1276 |
+
"epoch": 9.14,
|
1277 |
+
"grad_norm": 0.30712640285491943,
|
1278 |
+
"learning_rate": 0.0005941299358559407,
|
1279 |
+
"loss": 2.9295,
|
1280 |
+
"step": 170000
|
1281 |
+
},
|
1282 |
+
{
|
1283 |
+
"epoch": 9.2,
|
1284 |
+
"grad_norm": 0.24219301342964172,
|
1285 |
+
"learning_rate": 0.0005911875478138057,
|
1286 |
+
"loss": 2.9333,
|
1287 |
+
"step": 171000
|
1288 |
+
},
|
1289 |
+
{
|
1290 |
+
"epoch": 9.25,
|
1291 |
+
"grad_norm": 0.24113836884498596,
|
1292 |
+
"learning_rate": 0.0005882481021597129,
|
1293 |
+
"loss": 2.9348,
|
1294 |
+
"step": 172000
|
1295 |
+
},
|
1296 |
+
{
|
1297 |
+
"epoch": 9.3,
|
1298 |
+
"grad_norm": 0.26300373673439026,
|
1299 |
+
"learning_rate": 0.0005853057141175778,
|
1300 |
+
"loss": 2.94,
|
1301 |
+
"step": 173000
|
1302 |
+
},
|
1303 |
+
{
|
1304 |
+
"epoch": 9.36,
|
1305 |
+
"grad_norm": 0.2541556656360626,
|
1306 |
+
"learning_rate": 0.0005823662684634849,
|
1307 |
+
"loss": 2.9424,
|
1308 |
+
"step": 174000
|
1309 |
+
},
|
1310 |
+
{
|
1311 |
+
"epoch": 9.41,
|
1312 |
+
"grad_norm": 0.2480422556400299,
|
1313 |
+
"learning_rate": 0.00057942388042135,
|
1314 |
+
"loss": 2.9424,
|
1315 |
+
"step": 175000
|
1316 |
+
},
|
1317 |
+
{
|
1318 |
+
"epoch": 9.47,
|
1319 |
+
"grad_norm": 0.22237752377986908,
|
1320 |
+
"learning_rate": 0.0005764844347672571,
|
1321 |
+
"loss": 2.9463,
|
1322 |
+
"step": 176000
|
1323 |
+
},
|
1324 |
+
{
|
1325 |
+
"epoch": 9.52,
|
1326 |
+
"grad_norm": 0.23950238525867462,
|
1327 |
+
"learning_rate": 0.0005735420467251222,
|
1328 |
+
"loss": 2.9531,
|
1329 |
+
"step": 177000
|
1330 |
+
},
|
1331 |
+
{
|
1332 |
+
"epoch": 9.57,
|
1333 |
+
"grad_norm": 0.2510799467563629,
|
1334 |
+
"learning_rate": 0.0005705996586829871,
|
1335 |
+
"loss": 2.952,
|
1336 |
+
"step": 178000
|
1337 |
+
},
|
1338 |
+
{
|
1339 |
+
"epoch": 9.63,
|
1340 |
+
"grad_norm": 0.23875580728054047,
|
1341 |
+
"learning_rate": 0.0005676572706408521,
|
1342 |
+
"loss": 2.955,
|
1343 |
+
"step": 179000
|
1344 |
+
},
|
1345 |
+
{
|
1346 |
+
"epoch": 9.68,
|
1347 |
+
"grad_norm": 0.2786506414413452,
|
1348 |
+
"learning_rate": 0.0005647178249867593,
|
1349 |
+
"loss": 2.9576,
|
1350 |
+
"step": 180000
|
1351 |
+
},
|
1352 |
+
{
|
1353 |
+
"epoch": 9.73,
|
1354 |
+
"grad_norm": 0.22881832718849182,
|
1355 |
+
"learning_rate": 0.0005617754369446243,
|
1356 |
+
"loss": 2.9578,
|
1357 |
+
"step": 181000
|
1358 |
+
},
|
1359 |
+
{
|
1360 |
+
"epoch": 9.79,
|
1361 |
+
"grad_norm": 0.24316935241222382,
|
1362 |
+
"learning_rate": 0.0005588359912905315,
|
1363 |
+
"loss": 2.9569,
|
1364 |
+
"step": 182000
|
1365 |
+
},
|
1366 |
+
{
|
1367 |
+
"epoch": 9.84,
|
1368 |
+
"grad_norm": 0.2705481946468353,
|
1369 |
+
"learning_rate": 0.0005558936032483965,
|
1370 |
+
"loss": 2.9565,
|
1371 |
+
"step": 183000
|
1372 |
+
},
|
1373 |
+
{
|
1374 |
+
"epoch": 9.9,
|
1375 |
+
"grad_norm": 0.2366609275341034,
|
1376 |
+
"learning_rate": 0.0005529512152062614,
|
1377 |
+
"loss": 2.9613,
|
1378 |
+
"step": 184000
|
1379 |
+
},
|
1380 |
+
{
|
1381 |
+
"epoch": 9.95,
|
1382 |
+
"grad_norm": 0.2798861265182495,
|
1383 |
+
"learning_rate": 0.0005500088271641264,
|
1384 |
+
"loss": 2.9565,
|
1385 |
+
"step": 185000
|
1386 |
+
},
|
1387 |
+
{
|
1388 |
+
"epoch": 10.0,
|
1389 |
+
"eval_accuracy": 0.40983898839268323,
|
1390 |
+
"eval_loss": 3.351311683654785,
|
1391 |
+
"eval_runtime": 153.7734,
|
1392 |
+
"eval_samples_per_second": 376.665,
|
1393 |
+
"eval_steps_per_second": 5.892,
|
1394 |
+
"step": 185930
|
1395 |
+
},
|
1396 |
+
{
|
1397 |
+
"epoch": 10.0,
|
1398 |
+
"grad_norm": 0.26953232288360596,
|
1399 |
+
"learning_rate": 0.0005470693815100336,
|
1400 |
+
"loss": 2.951,
|
1401 |
+
"step": 186000
|
1402 |
+
},
|
1403 |
+
{
|
1404 |
+
"epoch": 10.06,
|
1405 |
+
"grad_norm": 0.2257799357175827,
|
1406 |
+
"learning_rate": 0.0005441269934678985,
|
1407 |
+
"loss": 2.8908,
|
1408 |
+
"step": 187000
|
1409 |
+
},
|
1410 |
+
{
|
1411 |
+
"epoch": 10.11,
|
1412 |
+
"grad_norm": 0.23467403650283813,
|
1413 |
+
"learning_rate": 0.0005411875478138058,
|
1414 |
+
"loss": 2.9006,
|
1415 |
+
"step": 188000
|
1416 |
+
},
|
1417 |
+
{
|
1418 |
+
"epoch": 10.17,
|
1419 |
+
"grad_norm": 0.2749410569667816,
|
1420 |
+
"learning_rate": 0.0005382451597716706,
|
1421 |
+
"loss": 2.9069,
|
1422 |
+
"step": 189000
|
1423 |
+
},
|
1424 |
+
{
|
1425 |
+
"epoch": 10.22,
|
1426 |
+
"grad_norm": 0.2599303722381592,
|
1427 |
+
"learning_rate": 0.0005353057141175779,
|
1428 |
+
"loss": 2.9104,
|
1429 |
+
"step": 190000
|
1430 |
+
},
|
1431 |
+
{
|
1432 |
+
"epoch": 10.27,
|
1433 |
+
"grad_norm": 0.2843225598335266,
|
1434 |
+
"learning_rate": 0.0005323633260754428,
|
1435 |
+
"loss": 2.9137,
|
1436 |
+
"step": 191000
|
1437 |
+
},
|
1438 |
+
{
|
1439 |
+
"epoch": 10.33,
|
1440 |
+
"grad_norm": 0.24611413478851318,
|
1441 |
+
"learning_rate": 0.00052942388042135,
|
1442 |
+
"loss": 2.9161,
|
1443 |
+
"step": 192000
|
1444 |
+
},
|
1445 |
+
{
|
1446 |
+
"epoch": 10.38,
|
1447 |
+
"grad_norm": 0.2826565206050873,
|
1448 |
+
"learning_rate": 0.000526481492379215,
|
1449 |
+
"loss": 2.9175,
|
1450 |
+
"step": 193000
|
1451 |
+
},
|
1452 |
+
{
|
1453 |
+
"epoch": 10.43,
|
1454 |
+
"grad_norm": 0.28213128447532654,
|
1455 |
+
"learning_rate": 0.000523542046725122,
|
1456 |
+
"loss": 2.9164,
|
1457 |
+
"step": 194000
|
1458 |
+
},
|
1459 |
+
{
|
1460 |
+
"epoch": 10.49,
|
1461 |
+
"grad_norm": 0.2640922963619232,
|
1462 |
+
"learning_rate": 0.0005205996586829871,
|
1463 |
+
"loss": 2.9258,
|
1464 |
+
"step": 195000
|
1465 |
+
},
|
1466 |
+
{
|
1467 |
+
"epoch": 10.54,
|
1468 |
+
"grad_norm": 0.2534860670566559,
|
1469 |
+
"learning_rate": 0.0005176572706408521,
|
1470 |
+
"loss": 2.9278,
|
1471 |
+
"step": 196000
|
1472 |
+
},
|
1473 |
+
{
|
1474 |
+
"epoch": 10.6,
|
1475 |
+
"grad_norm": 0.24981430172920227,
|
1476 |
+
"learning_rate": 0.0005147148825987171,
|
1477 |
+
"loss": 2.9292,
|
1478 |
+
"step": 197000
|
1479 |
+
},
|
1480 |
+
{
|
1481 |
+
"epoch": 10.65,
|
1482 |
+
"grad_norm": 0.2640092372894287,
|
1483 |
+
"learning_rate": 0.0005117724945565822,
|
1484 |
+
"loss": 2.9365,
|
1485 |
+
"step": 198000
|
1486 |
+
},
|
1487 |
+
{
|
1488 |
+
"epoch": 10.7,
|
1489 |
+
"grad_norm": 0.27058008313179016,
|
1490 |
+
"learning_rate": 0.0005088330489024892,
|
1491 |
+
"loss": 2.9324,
|
1492 |
+
"step": 199000
|
1493 |
+
},
|
1494 |
+
{
|
1495 |
+
"epoch": 10.76,
|
1496 |
+
"grad_norm": 0.25028932094573975,
|
1497 |
+
"learning_rate": 0.0005058906608603543,
|
1498 |
+
"loss": 2.9324,
|
1499 |
+
"step": 200000
|
1500 |
+
},
|
1501 |
+
{
|
1502 |
+
"epoch": 10.81,
|
1503 |
+
"grad_norm": 0.2733946144580841,
|
1504 |
+
"learning_rate": 0.0005029512152062614,
|
1505 |
+
"loss": 2.9309,
|
1506 |
+
"step": 201000
|
1507 |
+
},
|
1508 |
+
{
|
1509 |
+
"epoch": 10.86,
|
1510 |
+
"grad_norm": 0.2877325117588043,
|
1511 |
+
"learning_rate": 0.0005000088271641264,
|
1512 |
+
"loss": 2.9353,
|
1513 |
+
"step": 202000
|
1514 |
+
},
|
1515 |
+
{
|
1516 |
+
"epoch": 10.92,
|
1517 |
+
"grad_norm": 0.2410973757505417,
|
1518 |
+
"learning_rate": 0.0004970664391219914,
|
1519 |
+
"loss": 2.9363,
|
1520 |
+
"step": 203000
|
1521 |
+
},
|
1522 |
+
{
|
1523 |
+
"epoch": 10.97,
|
1524 |
+
"grad_norm": 0.2605952024459839,
|
1525 |
+
"learning_rate": 0.0004941240510798565,
|
1526 |
+
"loss": 2.9381,
|
1527 |
+
"step": 204000
|
1528 |
+
},
|
1529 |
+
{
|
1530 |
+
"epoch": 11.0,
|
1531 |
+
"eval_accuracy": 0.40963281732793566,
|
1532 |
+
"eval_loss": 3.3482396602630615,
|
1533 |
+
"eval_runtime": 154.276,
|
1534 |
+
"eval_samples_per_second": 375.437,
|
1535 |
+
"eval_steps_per_second": 5.873,
|
1536 |
+
"step": 204523
|
1537 |
+
},
|
1538 |
+
{
|
1539 |
+
"epoch": 11.03,
|
1540 |
+
"grad_norm": 0.2494332194328308,
|
1541 |
+
"learning_rate": 0.0004911846054257635,
|
1542 |
+
"loss": 2.9059,
|
1543 |
+
"step": 205000
|
1544 |
+
},
|
1545 |
+
{
|
1546 |
+
"epoch": 11.08,
|
1547 |
+
"grad_norm": 0.27014607191085815,
|
1548 |
+
"learning_rate": 0.0004882451597716707,
|
1549 |
+
"loss": 2.8743,
|
1550 |
+
"step": 206000
|
1551 |
+
},
|
1552 |
+
{
|
1553 |
+
"epoch": 11.13,
|
1554 |
+
"grad_norm": 0.24216848611831665,
|
1555 |
+
"learning_rate": 0.0004853027717295357,
|
1556 |
+
"loss": 2.8821,
|
1557 |
+
"step": 207000
|
1558 |
+
},
|
1559 |
+
{
|
1560 |
+
"epoch": 11.19,
|
1561 |
+
"grad_norm": 0.2651168704032898,
|
1562 |
+
"learning_rate": 0.0004823633260754429,
|
1563 |
+
"loss": 2.8867,
|
1564 |
+
"step": 208000
|
1565 |
+
},
|
1566 |
+
{
|
1567 |
+
"epoch": 11.24,
|
1568 |
+
"grad_norm": 0.2583751082420349,
|
1569 |
+
"learning_rate": 0.0004794209380333078,
|
1570 |
+
"loss": 2.8907,
|
1571 |
+
"step": 209000
|
1572 |
+
},
|
1573 |
+
{
|
1574 |
+
"epoch": 11.29,
|
1575 |
+
"grad_norm": 0.26034820079803467,
|
1576 |
+
"learning_rate": 0.00047647854999117284,
|
1577 |
+
"loss": 2.8953,
|
1578 |
+
"step": 210000
|
1579 |
+
},
|
1580 |
+
{
|
1581 |
+
"epoch": 11.35,
|
1582 |
+
"grad_norm": 0.3137526214122772,
|
1583 |
+
"learning_rate": 0.0004735361619490379,
|
1584 |
+
"loss": 2.8943,
|
1585 |
+
"step": 211000
|
1586 |
+
},
|
1587 |
+
{
|
1588 |
+
"epoch": 11.4,
|
1589 |
+
"grad_norm": 0.29816457629203796,
|
1590 |
+
"learning_rate": 0.000470596716294945,
|
1591 |
+
"loss": 2.8985,
|
1592 |
+
"step": 212000
|
1593 |
+
},
|
1594 |
+
{
|
1595 |
+
"epoch": 11.46,
|
1596 |
+
"grad_norm": 0.2567712664604187,
|
1597 |
+
"learning_rate": 0.00046765727064085213,
|
1598 |
+
"loss": 2.9024,
|
1599 |
+
"step": 213000
|
1600 |
+
},
|
1601 |
+
{
|
1602 |
+
"epoch": 11.51,
|
1603 |
+
"grad_norm": 0.2891099154949188,
|
1604 |
+
"learning_rate": 0.0004647148825987171,
|
1605 |
+
"loss": 2.9014,
|
1606 |
+
"step": 214000
|
1607 |
+
},
|
1608 |
+
{
|
1609 |
+
"epoch": 11.56,
|
1610 |
+
"grad_norm": 0.26076436042785645,
|
1611 |
+
"learning_rate": 0.00046177249455658215,
|
1612 |
+
"loss": 2.9061,
|
1613 |
+
"step": 215000
|
1614 |
+
},
|
1615 |
+
{
|
1616 |
+
"epoch": 11.62,
|
1617 |
+
"grad_norm": 0.2649165689945221,
|
1618 |
+
"learning_rate": 0.00045883010651444713,
|
1619 |
+
"loss": 2.9061,
|
1620 |
+
"step": 216000
|
1621 |
+
},
|
1622 |
+
{
|
1623 |
+
"epoch": 11.67,
|
1624 |
+
"grad_norm": 0.30298641324043274,
|
1625 |
+
"learning_rate": 0.0004558877184723121,
|
1626 |
+
"loss": 2.9054,
|
1627 |
+
"step": 217000
|
1628 |
+
},
|
1629 |
+
{
|
1630 |
+
"epoch": 11.72,
|
1631 |
+
"grad_norm": 0.2831864058971405,
|
1632 |
+
"learning_rate": 0.00045294827281821924,
|
1633 |
+
"loss": 2.9075,
|
1634 |
+
"step": 218000
|
1635 |
+
},
|
1636 |
+
{
|
1637 |
+
"epoch": 11.78,
|
1638 |
+
"grad_norm": 0.2540026307106018,
|
1639 |
+
"learning_rate": 0.00045000882716412637,
|
1640 |
+
"loss": 2.9112,
|
1641 |
+
"step": 219000
|
1642 |
+
},
|
1643 |
+
{
|
1644 |
+
"epoch": 11.83,
|
1645 |
+
"grad_norm": 0.2612316906452179,
|
1646 |
+
"learning_rate": 0.0004470664391219914,
|
1647 |
+
"loss": 2.9124,
|
1648 |
+
"step": 220000
|
1649 |
+
},
|
1650 |
+
{
|
1651 |
+
"epoch": 11.89,
|
1652 |
+
"grad_norm": 0.26292499899864197,
|
1653 |
+
"learning_rate": 0.00044412405107985644,
|
1654 |
+
"loss": 2.9111,
|
1655 |
+
"step": 221000
|
1656 |
+
},
|
1657 |
+
{
|
1658 |
+
"epoch": 11.94,
|
1659 |
+
"grad_norm": 0.2696852684020996,
|
1660 |
+
"learning_rate": 0.0004411816630377214,
|
1661 |
+
"loss": 2.9151,
|
1662 |
+
"step": 222000
|
1663 |
+
},
|
1664 |
+
{
|
1665 |
+
"epoch": 11.99,
|
1666 |
+
"grad_norm": 0.2645638883113861,
|
1667 |
+
"learning_rate": 0.00043824221738362855,
|
1668 |
+
"loss": 2.9171,
|
1669 |
+
"step": 223000
|
1670 |
+
},
|
1671 |
+
{
|
1672 |
+
"epoch": 12.0,
|
1673 |
+
"eval_accuracy": 0.41176466241385945,
|
1674 |
+
"eval_loss": 3.3354427814483643,
|
1675 |
+
"eval_runtime": 153.951,
|
1676 |
+
"eval_samples_per_second": 376.23,
|
1677 |
+
"eval_steps_per_second": 5.885,
|
1678 |
+
"step": 223116
|
1679 |
+
},
|
1680 |
+
{
|
1681 |
+
"epoch": 12.05,
|
1682 |
+
"grad_norm": 0.289524108171463,
|
1683 |
+
"learning_rate": 0.0004352998293414936,
|
1684 |
+
"loss": 2.8568,
|
1685 |
+
"step": 224000
|
1686 |
+
},
|
1687 |
+
{
|
1688 |
+
"epoch": 12.1,
|
1689 |
+
"grad_norm": 0.28729820251464844,
|
1690 |
+
"learning_rate": 0.00043235744129935857,
|
1691 |
+
"loss": 2.858,
|
1692 |
+
"step": 225000
|
1693 |
+
},
|
1694 |
+
{
|
1695 |
+
"epoch": 12.16,
|
1696 |
+
"grad_norm": 0.2749260663986206,
|
1697 |
+
"learning_rate": 0.00042941505325722355,
|
1698 |
+
"loss": 2.8627,
|
1699 |
+
"step": 226000
|
1700 |
+
},
|
1701 |
+
{
|
1702 |
+
"epoch": 12.21,
|
1703 |
+
"grad_norm": 0.2578941583633423,
|
1704 |
+
"learning_rate": 0.0004264785499911729,
|
1705 |
+
"loss": 2.8695,
|
1706 |
+
"step": 227000
|
1707 |
+
},
|
1708 |
+
{
|
1709 |
+
"epoch": 12.26,
|
1710 |
+
"grad_norm": 0.2701773941516876,
|
1711 |
+
"learning_rate": 0.0004235361619490378,
|
1712 |
+
"loss": 2.8702,
|
1713 |
+
"step": 228000
|
1714 |
+
},
|
1715 |
+
{
|
1716 |
+
"epoch": 12.32,
|
1717 |
+
"grad_norm": 0.2578844428062439,
|
1718 |
+
"learning_rate": 0.00042059377390690284,
|
1719 |
+
"loss": 2.8706,
|
1720 |
+
"step": 229000
|
1721 |
+
},
|
1722 |
+
{
|
1723 |
+
"epoch": 12.37,
|
1724 |
+
"grad_norm": 0.2564055621623993,
|
1725 |
+
"learning_rate": 0.0004176513858647679,
|
1726 |
+
"loss": 2.8719,
|
1727 |
+
"step": 230000
|
1728 |
+
},
|
1729 |
+
{
|
1730 |
+
"epoch": 12.42,
|
1731 |
+
"grad_norm": 0.2670447528362274,
|
1732 |
+
"learning_rate": 0.000414711940210675,
|
1733 |
+
"loss": 2.8779,
|
1734 |
+
"step": 231000
|
1735 |
+
},
|
1736 |
+
{
|
1737 |
+
"epoch": 12.48,
|
1738 |
+
"grad_norm": 0.25996309518814087,
|
1739 |
+
"learning_rate": 0.00041176955216854,
|
1740 |
+
"loss": 2.8834,
|
1741 |
+
"step": 232000
|
1742 |
+
},
|
1743 |
+
{
|
1744 |
+
"epoch": 12.53,
|
1745 |
+
"grad_norm": 0.2569678723812103,
|
1746 |
+
"learning_rate": 0.00040882716412640497,
|
1747 |
+
"loss": 2.8813,
|
1748 |
+
"step": 233000
|
1749 |
+
},
|
1750 |
+
{
|
1751 |
+
"epoch": 12.59,
|
1752 |
+
"grad_norm": 0.25648847222328186,
|
1753 |
+
"learning_rate": 0.00040588771847231215,
|
1754 |
+
"loss": 2.8887,
|
1755 |
+
"step": 234000
|
1756 |
+
},
|
1757 |
+
{
|
1758 |
+
"epoch": 12.64,
|
1759 |
+
"grad_norm": 0.27697890996932983,
|
1760 |
+
"learning_rate": 0.00040294533043017713,
|
1761 |
+
"loss": 2.8892,
|
1762 |
+
"step": 235000
|
1763 |
+
},
|
1764 |
+
{
|
1765 |
+
"epoch": 12.69,
|
1766 |
+
"grad_norm": 0.2665017247200012,
|
1767 |
+
"learning_rate": 0.0004000058847760843,
|
1768 |
+
"loss": 2.8903,
|
1769 |
+
"step": 236000
|
1770 |
+
},
|
1771 |
+
{
|
1772 |
+
"epoch": 12.75,
|
1773 |
+
"grad_norm": 0.27752918004989624,
|
1774 |
+
"learning_rate": 0.00039706349673394924,
|
1775 |
+
"loss": 2.8866,
|
1776 |
+
"step": 237000
|
1777 |
+
},
|
1778 |
+
{
|
1779 |
+
"epoch": 12.8,
|
1780 |
+
"grad_norm": 0.28366342186927795,
|
1781 |
+
"learning_rate": 0.0003941240510798564,
|
1782 |
+
"loss": 2.8901,
|
1783 |
+
"step": 238000
|
1784 |
+
},
|
1785 |
+
{
|
1786 |
+
"epoch": 12.85,
|
1787 |
+
"grad_norm": 0.268888920545578,
|
1788 |
+
"learning_rate": 0.0003911816630377214,
|
1789 |
+
"loss": 2.8909,
|
1790 |
+
"step": 239000
|
1791 |
+
},
|
1792 |
+
{
|
1793 |
+
"epoch": 12.91,
|
1794 |
+
"grad_norm": 0.2524598240852356,
|
1795 |
+
"learning_rate": 0.00038823927499558644,
|
1796 |
+
"loss": 2.8904,
|
1797 |
+
"step": 240000
|
1798 |
+
},
|
1799 |
+
{
|
1800 |
+
"epoch": 12.96,
|
1801 |
+
"grad_norm": 0.2472527027130127,
|
1802 |
+
"learning_rate": 0.00038529982934149356,
|
1803 |
+
"loss": 2.893,
|
1804 |
+
"step": 241000
|
1805 |
+
},
|
1806 |
+
{
|
1807 |
+
"epoch": 13.0,
|
1808 |
+
"eval_accuracy": 0.4111454102552269,
|
1809 |
+
"eval_loss": 3.3519082069396973,
|
1810 |
+
"eval_runtime": 153.4319,
|
1811 |
+
"eval_samples_per_second": 377.503,
|
1812 |
+
"eval_steps_per_second": 5.905,
|
1813 |
+
"step": 241709
|
1814 |
+
},
|
1815 |
+
{
|
1816 |
+
"epoch": 13.02,
|
1817 |
+
"grad_norm": 0.26184141635894775,
|
1818 |
+
"learning_rate": 0.00038235744129935855,
|
1819 |
+
"loss": 2.8762,
|
1820 |
+
"step": 242000
|
1821 |
+
},
|
1822 |
+
{
|
1823 |
+
"epoch": 13.07,
|
1824 |
+
"grad_norm": 0.26451557874679565,
|
1825 |
+
"learning_rate": 0.00037941799564526567,
|
1826 |
+
"loss": 2.8369,
|
1827 |
+
"step": 243000
|
1828 |
+
},
|
1829 |
+
{
|
1830 |
+
"epoch": 13.12,
|
1831 |
+
"grad_norm": 0.26714783906936646,
|
1832 |
+
"learning_rate": 0.0003764756076031307,
|
1833 |
+
"loss": 2.8415,
|
1834 |
+
"step": 244000
|
1835 |
+
},
|
1836 |
+
{
|
1837 |
+
"epoch": 13.18,
|
1838 |
+
"grad_norm": 0.25869232416152954,
|
1839 |
+
"learning_rate": 0.00037353321956099574,
|
1840 |
+
"loss": 2.8465,
|
1841 |
+
"step": 245000
|
1842 |
+
},
|
1843 |
+
{
|
1844 |
+
"epoch": 13.23,
|
1845 |
+
"grad_norm": 0.30912360548973083,
|
1846 |
+
"learning_rate": 0.00037059377390690287,
|
1847 |
+
"loss": 2.8481,
|
1848 |
+
"step": 246000
|
1849 |
+
},
|
1850 |
+
{
|
1851 |
+
"epoch": 13.28,
|
1852 |
+
"grad_norm": 0.30928054451942444,
|
1853 |
+
"learning_rate": 0.00036765138586476785,
|
1854 |
+
"loss": 2.8496,
|
1855 |
+
"step": 247000
|
1856 |
+
},
|
1857 |
+
{
|
1858 |
+
"epoch": 13.34,
|
1859 |
+
"grad_norm": 0.32625919580459595,
|
1860 |
+
"learning_rate": 0.00036470899782263283,
|
1861 |
+
"loss": 2.8551,
|
1862 |
+
"step": 248000
|
1863 |
+
},
|
1864 |
+
{
|
1865 |
+
"epoch": 13.39,
|
1866 |
+
"grad_norm": 0.2953076958656311,
|
1867 |
+
"learning_rate": 0.00036176955216854,
|
1868 |
+
"loss": 2.8563,
|
1869 |
+
"step": 249000
|
1870 |
+
},
|
1871 |
+
{
|
1872 |
+
"epoch": 13.45,
|
1873 |
+
"grad_norm": 0.2896451950073242,
|
1874 |
+
"learning_rate": 0.000358827164126405,
|
1875 |
+
"loss": 2.8619,
|
1876 |
+
"step": 250000
|
1877 |
+
},
|
1878 |
+
{
|
1879 |
+
"epoch": 13.5,
|
1880 |
+
"grad_norm": 0.271090567111969,
|
1881 |
+
"learning_rate": 0.0003558877184723122,
|
1882 |
+
"loss": 2.8616,
|
1883 |
+
"step": 251000
|
1884 |
+
},
|
1885 |
+
{
|
1886 |
+
"epoch": 13.55,
|
1887 |
+
"grad_norm": 0.2772102952003479,
|
1888 |
+
"learning_rate": 0.0003529453304301771,
|
1889 |
+
"loss": 2.8644,
|
1890 |
+
"step": 252000
|
1891 |
+
},
|
1892 |
+
{
|
1893 |
+
"epoch": 13.61,
|
1894 |
+
"grad_norm": 0.25150129199028015,
|
1895 |
+
"learning_rate": 0.00035000294238804214,
|
1896 |
+
"loss": 2.8627,
|
1897 |
+
"step": 253000
|
1898 |
+
},
|
1899 |
+
{
|
1900 |
+
"epoch": 13.66,
|
1901 |
+
"grad_norm": 0.27617818117141724,
|
1902 |
+
"learning_rate": 0.00034706349673394927,
|
1903 |
+
"loss": 2.861,
|
1904 |
+
"step": 254000
|
1905 |
+
},
|
1906 |
+
{
|
1907 |
+
"epoch": 13.71,
|
1908 |
+
"grad_norm": 0.2900349795818329,
|
1909 |
+
"learning_rate": 0.0003441211086918143,
|
1910 |
+
"loss": 2.8657,
|
1911 |
+
"step": 255000
|
1912 |
+
},
|
1913 |
+
{
|
1914 |
+
"epoch": 13.77,
|
1915 |
+
"grad_norm": 0.2805399000644684,
|
1916 |
+
"learning_rate": 0.0003411787206496793,
|
1917 |
+
"loss": 2.871,
|
1918 |
+
"step": 256000
|
1919 |
+
},
|
1920 |
+
{
|
1921 |
+
"epoch": 13.82,
|
1922 |
+
"grad_norm": 0.2589282691478729,
|
1923 |
+
"learning_rate": 0.00033823633260754427,
|
1924 |
+
"loss": 2.8672,
|
1925 |
+
"step": 257000
|
1926 |
+
},
|
1927 |
+
{
|
1928 |
+
"epoch": 13.88,
|
1929 |
+
"grad_norm": 0.291664183139801,
|
1930 |
+
"learning_rate": 0.00033529688695345145,
|
1931 |
+
"loss": 2.8715,
|
1932 |
+
"step": 258000
|
1933 |
+
},
|
1934 |
+
{
|
1935 |
+
"epoch": 13.93,
|
1936 |
+
"grad_norm": 0.2693411707878113,
|
1937 |
+
"learning_rate": 0.00033235449891131643,
|
1938 |
+
"loss": 2.8763,
|
1939 |
+
"step": 259000
|
1940 |
+
},
|
1941 |
+
{
|
1942 |
+
"epoch": 13.98,
|
1943 |
+
"grad_norm": 0.27113044261932373,
|
1944 |
+
"learning_rate": 0.0003294150532572236,
|
1945 |
+
"loss": 2.876,
|
1946 |
+
"step": 260000
|
1947 |
+
},
|
1948 |
+
{
|
1949 |
+
"epoch": 14.0,
|
1950 |
+
"eval_accuracy": 0.41093769408311887,
|
1951 |
+
"eval_loss": 3.3631937503814697,
|
1952 |
+
"eval_runtime": 153.9063,
|
1953 |
+
"eval_samples_per_second": 376.339,
|
1954 |
+
"eval_steps_per_second": 5.887,
|
1955 |
+
"step": 260302
|
1956 |
+
},
|
1957 |
+
{
|
1958 |
+
"epoch": 14.04,
|
1959 |
+
"grad_norm": 0.2805856168270111,
|
1960 |
+
"learning_rate": 0.00032647266521508854,
|
1961 |
+
"loss": 2.8345,
|
1962 |
+
"step": 261000
|
1963 |
+
},
|
1964 |
+
{
|
1965 |
+
"epoch": 14.09,
|
1966 |
+
"grad_norm": 0.2932230234146118,
|
1967 |
+
"learning_rate": 0.00032353321956099567,
|
1968 |
+
"loss": 2.8245,
|
1969 |
+
"step": 262000
|
1970 |
+
},
|
1971 |
+
{
|
1972 |
+
"epoch": 14.15,
|
1973 |
+
"grad_norm": 0.2711317837238312,
|
1974 |
+
"learning_rate": 0.0003205908315188607,
|
1975 |
+
"loss": 2.8265,
|
1976 |
+
"step": 263000
|
1977 |
+
},
|
1978 |
+
{
|
1979 |
+
"epoch": 14.2,
|
1980 |
+
"grad_norm": 0.30451709032058716,
|
1981 |
+
"learning_rate": 0.00031765138586476783,
|
1982 |
+
"loss": 2.8294,
|
1983 |
+
"step": 264000
|
1984 |
+
},
|
1985 |
+
{
|
1986 |
+
"epoch": 14.25,
|
1987 |
+
"grad_norm": 0.2669803500175476,
|
1988 |
+
"learning_rate": 0.00031470899782263287,
|
1989 |
+
"loss": 2.8349,
|
1990 |
+
"step": 265000
|
1991 |
+
},
|
1992 |
+
{
|
1993 |
+
"epoch": 14.31,
|
1994 |
+
"grad_norm": 0.3362666070461273,
|
1995 |
+
"learning_rate": 0.00031176660978049785,
|
1996 |
+
"loss": 2.8385,
|
1997 |
+
"step": 266000
|
1998 |
+
},
|
1999 |
+
{
|
2000 |
+
"epoch": 14.36,
|
2001 |
+
"grad_norm": 0.2826295495033264,
|
2002 |
+
"learning_rate": 0.000308827164126405,
|
2003 |
+
"loss": 2.8351,
|
2004 |
+
"step": 267000
|
2005 |
+
},
|
2006 |
+
{
|
2007 |
+
"epoch": 14.41,
|
2008 |
+
"grad_norm": 0.29114797711372375,
|
2009 |
+
"learning_rate": 0.00030588477608427,
|
2010 |
+
"loss": 2.8372,
|
2011 |
+
"step": 268000
|
2012 |
+
},
|
2013 |
+
{
|
2014 |
+
"epoch": 14.47,
|
2015 |
+
"grad_norm": 0.29871678352355957,
|
2016 |
+
"learning_rate": 0.00030294533043017714,
|
2017 |
+
"loss": 2.8397,
|
2018 |
+
"step": 269000
|
2019 |
+
},
|
2020 |
+
{
|
2021 |
+
"epoch": 14.52,
|
2022 |
+
"grad_norm": 0.2672843933105469,
|
2023 |
+
"learning_rate": 0.0003000029423880422,
|
2024 |
+
"loss": 2.8448,
|
2025 |
+
"step": 270000
|
2026 |
+
},
|
2027 |
+
{
|
2028 |
+
"epoch": 14.58,
|
2029 |
+
"grad_norm": 0.25932183861732483,
|
2030 |
+
"learning_rate": 0.0002970605543459071,
|
2031 |
+
"loss": 2.8443,
|
2032 |
+
"step": 271000
|
2033 |
+
},
|
2034 |
+
{
|
2035 |
+
"epoch": 14.63,
|
2036 |
+
"grad_norm": 0.2911001443862915,
|
2037 |
+
"learning_rate": 0.00029411816630377214,
|
2038 |
+
"loss": 2.8493,
|
2039 |
+
"step": 272000
|
2040 |
+
},
|
2041 |
+
{
|
2042 |
+
"epoch": 14.68,
|
2043 |
+
"grad_norm": 0.2664991617202759,
|
2044 |
+
"learning_rate": 0.00029117577826163717,
|
2045 |
+
"loss": 2.8422,
|
2046 |
+
"step": 273000
|
2047 |
+
},
|
2048 |
+
{
|
2049 |
+
"epoch": 14.74,
|
2050 |
+
"grad_norm": 0.30084219574928284,
|
2051 |
+
"learning_rate": 0.0002882363326075443,
|
2052 |
+
"loss": 2.8505,
|
2053 |
+
"step": 274000
|
2054 |
+
},
|
2055 |
+
{
|
2056 |
+
"epoch": 14.79,
|
2057 |
+
"grad_norm": 0.2992892861366272,
|
2058 |
+
"learning_rate": 0.0002852939445654093,
|
2059 |
+
"loss": 2.8497,
|
2060 |
+
"step": 275000
|
2061 |
+
},
|
2062 |
+
{
|
2063 |
+
"epoch": 14.84,
|
2064 |
+
"grad_norm": 0.335448682308197,
|
2065 |
+
"learning_rate": 0.0002823544989113164,
|
2066 |
+
"loss": 2.8529,
|
2067 |
+
"step": 276000
|
2068 |
+
},
|
2069 |
+
{
|
2070 |
+
"epoch": 14.9,
|
2071 |
+
"grad_norm": 0.28675752878189087,
|
2072 |
+
"learning_rate": 0.00027941211086918144,
|
2073 |
+
"loss": 2.8497,
|
2074 |
+
"step": 277000
|
2075 |
+
},
|
2076 |
+
{
|
2077 |
+
"epoch": 14.95,
|
2078 |
+
"grad_norm": 0.3506409525871277,
|
2079 |
+
"learning_rate": 0.00027647266521508857,
|
2080 |
+
"loss": 2.8472,
|
2081 |
+
"step": 278000
|
2082 |
+
},
|
2083 |
+
{
|
2084 |
+
"epoch": 15.0,
|
2085 |
+
"eval_accuracy": 0.41203740745226775,
|
2086 |
+
"eval_loss": 3.3530044555664062,
|
2087 |
+
"eval_runtime": 154.3507,
|
2088 |
+
"eval_samples_per_second": 375.256,
|
2089 |
+
"eval_steps_per_second": 5.87,
|
2090 |
+
"step": 278895
|
2091 |
+
},
|
2092 |
+
{
|
2093 |
+
"epoch": 15.01,
|
2094 |
+
"grad_norm": 0.2708049416542053,
|
2095 |
+
"learning_rate": 0.0002735302771729536,
|
2096 |
+
"loss": 2.8479,
|
2097 |
+
"step": 279000
|
2098 |
+
},
|
2099 |
+
{
|
2100 |
+
"epoch": 15.06,
|
2101 |
+
"grad_norm": 0.2830442190170288,
|
2102 |
+
"learning_rate": 0.00027059083151886073,
|
2103 |
+
"loss": 2.8018,
|
2104 |
+
"step": 280000
|
2105 |
+
},
|
2106 |
+
{
|
2107 |
+
"epoch": 15.11,
|
2108 |
+
"grad_norm": 0.2702392041683197,
|
2109 |
+
"learning_rate": 0.0002676484434767257,
|
2110 |
+
"loss": 2.805,
|
2111 |
+
"step": 281000
|
2112 |
+
},
|
2113 |
+
{
|
2114 |
+
"epoch": 15.17,
|
2115 |
+
"grad_norm": 0.3452757000923157,
|
2116 |
+
"learning_rate": 0.00026470899782263284,
|
2117 |
+
"loss": 2.8103,
|
2118 |
+
"step": 282000
|
2119 |
+
},
|
2120 |
+
{
|
2121 |
+
"epoch": 15.22,
|
2122 |
+
"grad_norm": 0.28475674986839294,
|
2123 |
+
"learning_rate": 0.0002617666097804979,
|
2124 |
+
"loss": 2.8141,
|
2125 |
+
"step": 283000
|
2126 |
+
},
|
2127 |
+
{
|
2128 |
+
"epoch": 15.27,
|
2129 |
+
"grad_norm": 0.2740528881549835,
|
2130 |
+
"learning_rate": 0.00025882422173836286,
|
2131 |
+
"loss": 2.8162,
|
2132 |
+
"step": 284000
|
2133 |
+
},
|
2134 |
+
{
|
2135 |
+
"epoch": 15.33,
|
2136 |
+
"grad_norm": 0.3067547380924225,
|
2137 |
+
"learning_rate": 0.00025588183369622784,
|
2138 |
+
"loss": 2.8182,
|
2139 |
+
"step": 285000
|
2140 |
+
},
|
2141 |
+
{
|
2142 |
+
"epoch": 15.38,
|
2143 |
+
"grad_norm": 0.3173828721046448,
|
2144 |
+
"learning_rate": 0.00025294238804213497,
|
2145 |
+
"loss": 2.8179,
|
2146 |
+
"step": 286000
|
2147 |
+
},
|
2148 |
+
{
|
2149 |
+
"epoch": 15.44,
|
2150 |
+
"grad_norm": 0.32790786027908325,
|
2151 |
+
"learning_rate": 0.00025,
|
2152 |
+
"loss": 2.8222,
|
2153 |
+
"step": 287000
|
2154 |
+
},
|
2155 |
+
{
|
2156 |
+
"epoch": 15.49,
|
2157 |
+
"grad_norm": 0.3280278444290161,
|
2158 |
+
"learning_rate": 0.00024706055434590713,
|
2159 |
+
"loss": 2.8263,
|
2160 |
+
"step": 288000
|
2161 |
+
},
|
2162 |
+
{
|
2163 |
+
"epoch": 15.54,
|
2164 |
+
"grad_norm": 0.30356574058532715,
|
2165 |
+
"learning_rate": 0.00024411816630377214,
|
2166 |
+
"loss": 2.826,
|
2167 |
+
"step": 289000
|
2168 |
+
},
|
2169 |
+
{
|
2170 |
+
"epoch": 15.6,
|
2171 |
+
"grad_norm": 0.2901437282562256,
|
2172 |
+
"learning_rate": 0.00024117577826163715,
|
2173 |
+
"loss": 2.8254,
|
2174 |
+
"step": 290000
|
2175 |
+
},
|
2176 |
+
{
|
2177 |
+
"epoch": 15.65,
|
2178 |
+
"grad_norm": 0.34759074449539185,
|
2179 |
+
"learning_rate": 0.0002382363326075443,
|
2180 |
+
"loss": 2.8291,
|
2181 |
+
"step": 291000
|
2182 |
+
},
|
2183 |
+
{
|
2184 |
+
"epoch": 15.7,
|
2185 |
+
"grad_norm": 0.38021716475486755,
|
2186 |
+
"learning_rate": 0.00023529394456540929,
|
2187 |
+
"loss": 2.8286,
|
2188 |
+
"step": 292000
|
2189 |
+
},
|
2190 |
+
{
|
2191 |
+
"epoch": 15.76,
|
2192 |
+
"grad_norm": 0.29485049843788147,
|
2193 |
+
"learning_rate": 0.0002323515565232743,
|
2194 |
+
"loss": 2.8312,
|
2195 |
+
"step": 293000
|
2196 |
+
},
|
2197 |
+
{
|
2198 |
+
"epoch": 15.81,
|
2199 |
+
"grad_norm": 0.27850472927093506,
|
2200 |
+
"learning_rate": 0.00022941211086918142,
|
2201 |
+
"loss": 2.8346,
|
2202 |
+
"step": 294000
|
2203 |
+
},
|
2204 |
+
{
|
2205 |
+
"epoch": 15.87,
|
2206 |
+
"grad_norm": 0.2848072052001953,
|
2207 |
+
"learning_rate": 0.00022646972282704646,
|
2208 |
+
"loss": 2.8291,
|
2209 |
+
"step": 295000
|
2210 |
+
},
|
2211 |
+
{
|
2212 |
+
"epoch": 15.92,
|
2213 |
+
"grad_norm": 0.3154236674308777,
|
2214 |
+
"learning_rate": 0.00022353027717295358,
|
2215 |
+
"loss": 2.8338,
|
2216 |
+
"step": 296000
|
2217 |
+
},
|
2218 |
+
{
|
2219 |
+
"epoch": 15.97,
|
2220 |
+
"grad_norm": 0.2986476719379425,
|
2221 |
+
"learning_rate": 0.00022058788913081857,
|
2222 |
+
"loss": 2.8335,
|
2223 |
+
"step": 297000
|
2224 |
+
},
|
2225 |
+
{
|
2226 |
+
"epoch": 16.0,
|
2227 |
+
"eval_accuracy": 0.41131288645738256,
|
2228 |
+
"eval_loss": 3.3726773262023926,
|
2229 |
+
"eval_runtime": 153.5545,
|
2230 |
+
"eval_samples_per_second": 377.201,
|
2231 |
+
"eval_steps_per_second": 5.9,
|
2232 |
+
"step": 297488
|
2233 |
+
},
|
2234 |
+
{
|
2235 |
+
"epoch": 16.03,
|
2236 |
+
"grad_norm": 0.31234875321388245,
|
2237 |
+
"learning_rate": 0.00021764844347672572,
|
2238 |
+
"loss": 2.8076,
|
2239 |
+
"step": 298000
|
2240 |
+
},
|
2241 |
+
{
|
2242 |
+
"epoch": 16.08,
|
2243 |
+
"grad_norm": 0.30626383423805237,
|
2244 |
+
"learning_rate": 0.0002147060554345907,
|
2245 |
+
"loss": 2.7883,
|
2246 |
+
"step": 299000
|
2247 |
+
},
|
2248 |
+
{
|
2249 |
+
"epoch": 16.14,
|
2250 |
+
"grad_norm": 0.2825140655040741,
|
2251 |
+
"learning_rate": 0.00021176366739245574,
|
2252 |
+
"loss": 2.7947,
|
2253 |
+
"step": 300000
|
2254 |
+
},
|
2255 |
+
{
|
2256 |
+
"epoch": 16.19,
|
2257 |
+
"grad_norm": 0.28852617740631104,
|
2258 |
+
"learning_rate": 0.00020882422173836286,
|
2259 |
+
"loss": 2.795,
|
2260 |
+
"step": 301000
|
2261 |
+
},
|
2262 |
+
{
|
2263 |
+
"epoch": 16.24,
|
2264 |
+
"grad_norm": 0.31543779373168945,
|
2265 |
+
"learning_rate": 0.00020588477608427,
|
2266 |
+
"loss": 2.798,
|
2267 |
+
"step": 302000
|
2268 |
+
},
|
2269 |
+
{
|
2270 |
+
"epoch": 16.3,
|
2271 |
+
"grad_norm": 0.2947680652141571,
|
2272 |
+
"learning_rate": 0.000202942388042135,
|
2273 |
+
"loss": 2.8025,
|
2274 |
+
"step": 303000
|
2275 |
+
},
|
2276 |
+
{
|
2277 |
+
"epoch": 16.35,
|
2278 |
+
"grad_norm": 0.291638046503067,
|
2279 |
+
"learning_rate": 0.0002,
|
2280 |
+
"loss": 2.8015,
|
2281 |
+
"step": 304000
|
2282 |
+
},
|
2283 |
+
{
|
2284 |
+
"epoch": 16.4,
|
2285 |
+
"grad_norm": 0.29190531373023987,
|
2286 |
+
"learning_rate": 0.00019705761195786502,
|
2287 |
+
"loss": 2.8063,
|
2288 |
+
"step": 305000
|
2289 |
+
},
|
2290 |
+
{
|
2291 |
+
"epoch": 16.46,
|
2292 |
+
"grad_norm": 0.3504582643508911,
|
2293 |
+
"learning_rate": 0.00019411522391573,
|
2294 |
+
"loss": 2.8084,
|
2295 |
+
"step": 306000
|
2296 |
+
},
|
2297 |
+
{
|
2298 |
+
"epoch": 16.51,
|
2299 |
+
"grad_norm": 0.2807118892669678,
|
2300 |
+
"learning_rate": 0.000191172835873595,
|
2301 |
+
"loss": 2.8028,
|
2302 |
+
"step": 307000
|
2303 |
+
},
|
2304 |
+
{
|
2305 |
+
"epoch": 16.57,
|
2306 |
+
"grad_norm": 0.29211023449897766,
|
2307 |
+
"learning_rate": 0.00018823339021950214,
|
2308 |
+
"loss": 2.8073,
|
2309 |
+
"step": 308000
|
2310 |
+
},
|
2311 |
+
{
|
2312 |
+
"epoch": 16.62,
|
2313 |
+
"grad_norm": 0.28356799483299255,
|
2314 |
+
"learning_rate": 0.0001852939445654093,
|
2315 |
+
"loss": 2.8114,
|
2316 |
+
"step": 309000
|
2317 |
+
},
|
2318 |
+
{
|
2319 |
+
"epoch": 16.67,
|
2320 |
+
"grad_norm": 0.32647237181663513,
|
2321 |
+
"learning_rate": 0.0001823515565232743,
|
2322 |
+
"loss": 2.809,
|
2323 |
+
"step": 310000
|
2324 |
+
},
|
2325 |
+
{
|
2326 |
+
"epoch": 16.73,
|
2327 |
+
"grad_norm": 0.3036649227142334,
|
2328 |
+
"learning_rate": 0.0001794091684811393,
|
2329 |
+
"loss": 2.8088,
|
2330 |
+
"step": 311000
|
2331 |
+
},
|
2332 |
+
{
|
2333 |
+
"epoch": 16.78,
|
2334 |
+
"grad_norm": 0.3010052740573883,
|
2335 |
+
"learning_rate": 0.0001764667804390043,
|
2336 |
+
"loss": 2.8145,
|
2337 |
+
"step": 312000
|
2338 |
+
},
|
2339 |
+
{
|
2340 |
+
"epoch": 16.83,
|
2341 |
+
"grad_norm": 0.2830561101436615,
|
2342 |
+
"learning_rate": 0.00017352733478491144,
|
2343 |
+
"loss": 2.8093,
|
2344 |
+
"step": 313000
|
2345 |
+
},
|
2346 |
+
{
|
2347 |
+
"epoch": 16.89,
|
2348 |
+
"grad_norm": 0.3039667308330536,
|
2349 |
+
"learning_rate": 0.00017058494674277645,
|
2350 |
+
"loss": 2.8143,
|
2351 |
+
"step": 314000
|
2352 |
+
},
|
2353 |
+
{
|
2354 |
+
"epoch": 16.94,
|
2355 |
+
"grad_norm": 0.3185649514198303,
|
2356 |
+
"learning_rate": 0.00016764255870064143,
|
2357 |
+
"loss": 2.8144,
|
2358 |
+
"step": 315000
|
2359 |
+
},
|
2360 |
+
{
|
2361 |
+
"epoch": 17.0,
|
2362 |
+
"grad_norm": 0.2941153049468994,
|
2363 |
+
"learning_rate": 0.00016470017065850644,
|
2364 |
+
"loss": 2.812,
|
2365 |
+
"step": 316000
|
2366 |
+
},
|
2367 |
+
{
|
2368 |
+
"epoch": 17.0,
|
2369 |
+
"eval_accuracy": 0.41104343316943776,
|
2370 |
+
"eval_loss": 3.3814456462860107,
|
2371 |
+
"eval_runtime": 153.7991,
|
2372 |
+
"eval_samples_per_second": 376.602,
|
2373 |
+
"eval_steps_per_second": 5.891,
|
2374 |
+
"step": 316081
|
2375 |
+
},
|
2376 |
+
{
|
2377 |
+
"epoch": 17.05,
|
2378 |
+
"grad_norm": 0.30029386281967163,
|
2379 |
+
"learning_rate": 0.00016176366739245572,
|
2380 |
+
"loss": 2.7825,
|
2381 |
+
"step": 317000
|
2382 |
+
},
|
2383 |
+
{
|
2384 |
+
"epoch": 17.1,
|
2385 |
+
"grad_norm": 0.2988363206386566,
|
2386 |
+
"learning_rate": 0.00015882127935032072,
|
2387 |
+
"loss": 2.7782,
|
2388 |
+
"step": 318000
|
2389 |
+
},
|
2390 |
+
{
|
2391 |
+
"epoch": 17.16,
|
2392 |
+
"grad_norm": 0.31658461689949036,
|
2393 |
+
"learning_rate": 0.00015588183369622785,
|
2394 |
+
"loss": 2.7813,
|
2395 |
+
"step": 319000
|
2396 |
+
},
|
2397 |
+
{
|
2398 |
+
"epoch": 17.21,
|
2399 |
+
"grad_norm": 0.3059552311897278,
|
2400 |
+
"learning_rate": 0.0001529394456540929,
|
2401 |
+
"loss": 2.7835,
|
2402 |
+
"step": 320000
|
2403 |
+
},
|
2404 |
+
{
|
2405 |
+
"epoch": 17.26,
|
2406 |
+
"grad_norm": 0.30100756883621216,
|
2407 |
+
"learning_rate": 0.00014999705761195787,
|
2408 |
+
"loss": 2.7827,
|
2409 |
+
"step": 321000
|
2410 |
+
},
|
2411 |
+
{
|
2412 |
+
"epoch": 17.32,
|
2413 |
+
"grad_norm": 0.30339285731315613,
|
2414 |
+
"learning_rate": 0.00014705466956982285,
|
2415 |
+
"loss": 2.7887,
|
2416 |
+
"step": 322000
|
2417 |
+
},
|
2418 |
+
{
|
2419 |
+
"epoch": 17.37,
|
2420 |
+
"grad_norm": 0.3006446957588196,
|
2421 |
+
"learning_rate": 0.00014411522391573,
|
2422 |
+
"loss": 2.7872,
|
2423 |
+
"step": 323000
|
2424 |
+
},
|
2425 |
+
{
|
2426 |
+
"epoch": 17.43,
|
2427 |
+
"grad_norm": 0.28081661462783813,
|
2428 |
+
"learning_rate": 0.000141172835873595,
|
2429 |
+
"loss": 2.7869,
|
2430 |
+
"step": 324000
|
2431 |
+
},
|
2432 |
+
{
|
2433 |
+
"epoch": 17.48,
|
2434 |
+
"grad_norm": 0.28526371717453003,
|
2435 |
+
"learning_rate": 0.00013823339021950214,
|
2436 |
+
"loss": 2.7881,
|
2437 |
+
"step": 325000
|
2438 |
+
},
|
2439 |
+
{
|
2440 |
+
"epoch": 17.53,
|
2441 |
+
"grad_norm": 0.31043946743011475,
|
2442 |
+
"learning_rate": 0.00013529100217736715,
|
2443 |
+
"loss": 2.7894,
|
2444 |
+
"step": 326000
|
2445 |
+
},
|
2446 |
+
{
|
2447 |
+
"epoch": 17.59,
|
2448 |
+
"grad_norm": 0.3015703558921814,
|
2449 |
+
"learning_rate": 0.00013234861413523216,
|
2450 |
+
"loss": 2.79,
|
2451 |
+
"step": 327000
|
2452 |
+
},
|
2453 |
+
{
|
2454 |
+
"epoch": 17.64,
|
2455 |
+
"grad_norm": 0.29077813029289246,
|
2456 |
+
"learning_rate": 0.00012940916848113928,
|
2457 |
+
"loss": 2.7921,
|
2458 |
+
"step": 328000
|
2459 |
+
},
|
2460 |
+
{
|
2461 |
+
"epoch": 17.69,
|
2462 |
+
"grad_norm": 0.29008862376213074,
|
2463 |
+
"learning_rate": 0.00012646972282704644,
|
2464 |
+
"loss": 2.7921,
|
2465 |
+
"step": 329000
|
2466 |
+
},
|
2467 |
+
{
|
2468 |
+
"epoch": 17.75,
|
2469 |
+
"grad_norm": 0.29531145095825195,
|
2470 |
+
"learning_rate": 0.00012352733478491142,
|
2471 |
+
"loss": 2.7908,
|
2472 |
+
"step": 330000
|
2473 |
+
},
|
2474 |
+
{
|
2475 |
+
"epoch": 17.8,
|
2476 |
+
"grad_norm": 0.3173753619194031,
|
2477 |
+
"learning_rate": 0.00012058494674277644,
|
2478 |
+
"loss": 2.7954,
|
2479 |
+
"step": 331000
|
2480 |
+
},
|
2481 |
+
{
|
2482 |
+
"epoch": 17.86,
|
2483 |
+
"grad_norm": 0.28667834401130676,
|
2484 |
+
"learning_rate": 0.00011764550108868357,
|
2485 |
+
"loss": 2.791,
|
2486 |
+
"step": 332000
|
2487 |
+
},
|
2488 |
+
{
|
2489 |
+
"epoch": 17.91,
|
2490 |
+
"grad_norm": 0.3025910556316376,
|
2491 |
+
"learning_rate": 0.00011470311304654858,
|
2492 |
+
"loss": 2.7956,
|
2493 |
+
"step": 333000
|
2494 |
+
},
|
2495 |
+
{
|
2496 |
+
"epoch": 17.96,
|
2497 |
+
"grad_norm": 0.32427656650543213,
|
2498 |
+
"learning_rate": 0.00011176366739245572,
|
2499 |
+
"loss": 2.7944,
|
2500 |
+
"step": 334000
|
2501 |
+
},
|
2502 |
+
{
|
2503 |
+
"epoch": 18.0,
|
2504 |
+
"eval_accuracy": 0.4119177623997049,
|
2505 |
+
"eval_loss": 3.3788328170776367,
|
2506 |
+
"eval_runtime": 153.909,
|
2507 |
+
"eval_samples_per_second": 376.333,
|
2508 |
+
"eval_steps_per_second": 5.887,
|
2509 |
+
"step": 334674
|
2510 |
+
},
|
2511 |
+
{
|
2512 |
+
"epoch": 18.02,
|
2513 |
+
"grad_norm": 0.28937357664108276,
|
2514 |
+
"learning_rate": 0.00010882127935032073,
|
2515 |
+
"loss": 2.785,
|
2516 |
+
"step": 335000
|
2517 |
+
},
|
2518 |
+
{
|
2519 |
+
"epoch": 18.07,
|
2520 |
+
"grad_norm": 0.3173358738422394,
|
2521 |
+
"learning_rate": 0.00010588183369622787,
|
2522 |
+
"loss": 2.7681,
|
2523 |
+
"step": 336000
|
2524 |
+
},
|
2525 |
+
{
|
2526 |
+
"epoch": 18.13,
|
2527 |
+
"grad_norm": 0.3219173550605774,
|
2528 |
+
"learning_rate": 0.00010293944565409286,
|
2529 |
+
"loss": 2.7649,
|
2530 |
+
"step": 337000
|
2531 |
+
},
|
2532 |
+
{
|
2533 |
+
"epoch": 18.18,
|
2534 |
+
"grad_norm": 0.30825838446617126,
|
2535 |
+
"learning_rate": 0.0001,
|
2536 |
+
"loss": 2.7678,
|
2537 |
+
"step": 338000
|
2538 |
+
},
|
2539 |
+
{
|
2540 |
+
"epoch": 18.23,
|
2541 |
+
"grad_norm": 0.2984776198863983,
|
2542 |
+
"learning_rate": 9.7057611957865e-05,
|
2543 |
+
"loss": 2.7691,
|
2544 |
+
"step": 339000
|
2545 |
+
},
|
2546 |
+
{
|
2547 |
+
"epoch": 18.29,
|
2548 |
+
"grad_norm": 0.3033863604068756,
|
2549 |
+
"learning_rate": 9.411522391573001e-05,
|
2550 |
+
"loss": 2.7718,
|
2551 |
+
"step": 340000
|
2552 |
+
},
|
2553 |
+
{
|
2554 |
+
"epoch": 18.34,
|
2555 |
+
"grad_norm": 0.298531711101532,
|
2556 |
+
"learning_rate": 9.117283587359502e-05,
|
2557 |
+
"loss": 2.7688,
|
2558 |
+
"step": 341000
|
2559 |
+
},
|
2560 |
+
{
|
2561 |
+
"epoch": 18.39,
|
2562 |
+
"grad_norm": 0.37657690048217773,
|
2563 |
+
"learning_rate": 8.823339021950214e-05,
|
2564 |
+
"loss": 2.7737,
|
2565 |
+
"step": 342000
|
2566 |
+
},
|
2567 |
+
{
|
2568 |
+
"epoch": 18.45,
|
2569 |
+
"grad_norm": 0.30152538418769836,
|
2570 |
+
"learning_rate": 8.529394456540929e-05,
|
2571 |
+
"loss": 2.7705,
|
2572 |
+
"step": 343000
|
2573 |
+
},
|
2574 |
+
{
|
2575 |
+
"epoch": 18.5,
|
2576 |
+
"grad_norm": 0.2887534499168396,
|
2577 |
+
"learning_rate": 8.23515565232743e-05,
|
2578 |
+
"loss": 2.7716,
|
2579 |
+
"step": 344000
|
2580 |
+
},
|
2581 |
+
{
|
2582 |
+
"epoch": 18.56,
|
2583 |
+
"grad_norm": 0.31988152861595154,
|
2584 |
+
"learning_rate": 7.94091684811393e-05,
|
2585 |
+
"loss": 2.7738,
|
2586 |
+
"step": 345000
|
2587 |
+
},
|
2588 |
+
{
|
2589 |
+
"epoch": 18.61,
|
2590 |
+
"grad_norm": 0.2952527105808258,
|
2591 |
+
"learning_rate": 7.646972282704644e-05,
|
2592 |
+
"loss": 2.7732,
|
2593 |
+
"step": 346000
|
2594 |
+
},
|
2595 |
+
{
|
2596 |
+
"epoch": 18.66,
|
2597 |
+
"grad_norm": 0.318419873714447,
|
2598 |
+
"learning_rate": 7.352733478491143e-05,
|
2599 |
+
"loss": 2.7787,
|
2600 |
+
"step": 347000
|
2601 |
+
},
|
2602 |
+
{
|
2603 |
+
"epoch": 18.72,
|
2604 |
+
"grad_norm": 0.290344774723053,
|
2605 |
+
"learning_rate": 7.058494674277643e-05,
|
2606 |
+
"loss": 2.7724,
|
2607 |
+
"step": 348000
|
2608 |
+
},
|
2609 |
+
{
|
2610 |
+
"epoch": 18.77,
|
2611 |
+
"grad_norm": 0.34806349873542786,
|
2612 |
+
"learning_rate": 6.764550108868357e-05,
|
2613 |
+
"loss": 2.7785,
|
2614 |
+
"step": 349000
|
2615 |
+
},
|
2616 |
+
{
|
2617 |
+
"epoch": 18.82,
|
2618 |
+
"grad_norm": 0.27911612391471863,
|
2619 |
+
"learning_rate": 6.470311304654858e-05,
|
2620 |
+
"loss": 2.7754,
|
2621 |
+
"step": 350000
|
2622 |
+
},
|
2623 |
+
{
|
2624 |
+
"epoch": 18.88,
|
2625 |
+
"grad_norm": 0.36661508679389954,
|
2626 |
+
"learning_rate": 6.176072500441358e-05,
|
2627 |
+
"loss": 2.7737,
|
2628 |
+
"step": 351000
|
2629 |
+
},
|
2630 |
+
{
|
2631 |
+
"epoch": 18.93,
|
2632 |
+
"grad_norm": 0.2794508635997772,
|
2633 |
+
"learning_rate": 5.881833696227858e-05,
|
2634 |
+
"loss": 2.7796,
|
2635 |
+
"step": 352000
|
2636 |
+
},
|
2637 |
+
{
|
2638 |
+
"epoch": 18.99,
|
2639 |
+
"grad_norm": 0.3130718171596527,
|
2640 |
+
"learning_rate": 5.587889130818573e-05,
|
2641 |
+
"loss": 2.7761,
|
2642 |
+
"step": 353000
|
2643 |
+
},
|
2644 |
+
{
|
2645 |
+
"epoch": 19.0,
|
2646 |
+
"eval_accuracy": 0.4113871859678455,
|
2647 |
+
"eval_loss": 3.3925299644470215,
|
2648 |
+
"eval_runtime": 155.9595,
|
2649 |
+
"eval_samples_per_second": 371.385,
|
2650 |
+
"eval_steps_per_second": 5.809,
|
2651 |
+
"step": 353267
|
2652 |
+
},
|
2653 |
+
{
|
2654 |
+
"epoch": 19.04,
|
2655 |
+
"grad_norm": 0.2997041940689087,
|
2656 |
+
"learning_rate": 5.293650326605073e-05,
|
2657 |
+
"loss": 2.7625,
|
2658 |
+
"step": 354000
|
2659 |
+
},
|
2660 |
+
{
|
2661 |
+
"epoch": 19.09,
|
2662 |
+
"grad_norm": 0.31620392203330994,
|
2663 |
+
"learning_rate": 4.999411522391573e-05,
|
2664 |
+
"loss": 2.7545,
|
2665 |
+
"step": 355000
|
2666 |
+
},
|
2667 |
+
{
|
2668 |
+
"epoch": 19.15,
|
2669 |
+
"grad_norm": 0.3096230626106262,
|
2670 |
+
"learning_rate": 4.7051727181780734e-05,
|
2671 |
+
"loss": 2.7559,
|
2672 |
+
"step": 356000
|
2673 |
+
},
|
2674 |
+
{
|
2675 |
+
"epoch": 19.2,
|
2676 |
+
"grad_norm": 0.29621490836143494,
|
2677 |
+
"learning_rate": 4.4115223915730006e-05,
|
2678 |
+
"loss": 2.7603,
|
2679 |
+
"step": 357000
|
2680 |
+
},
|
2681 |
+
{
|
2682 |
+
"epoch": 19.25,
|
2683 |
+
"grad_norm": 0.3432357609272003,
|
2684 |
+
"learning_rate": 4.117283587359501e-05,
|
2685 |
+
"loss": 2.7593,
|
2686 |
+
"step": 358000
|
2687 |
+
},
|
2688 |
+
{
|
2689 |
+
"epoch": 19.31,
|
2690 |
+
"grad_norm": 0.32921677827835083,
|
2691 |
+
"learning_rate": 3.823044783146001e-05,
|
2692 |
+
"loss": 2.7558,
|
2693 |
+
"step": 359000
|
2694 |
+
},
|
2695 |
+
{
|
2696 |
+
"epoch": 19.36,
|
2697 |
+
"grad_norm": 0.3113718032836914,
|
2698 |
+
"learning_rate": 3.528805978932502e-05,
|
2699 |
+
"loss": 2.7584,
|
2700 |
+
"step": 360000
|
2701 |
+
},
|
2702 |
+
{
|
2703 |
+
"epoch": 19.42,
|
2704 |
+
"grad_norm": 0.29590970277786255,
|
2705 |
+
"learning_rate": 3.234861413523216e-05,
|
2706 |
+
"loss": 2.7579,
|
2707 |
+
"step": 361000
|
2708 |
+
},
|
2709 |
+
{
|
2710 |
+
"epoch": 19.47,
|
2711 |
+
"grad_norm": 0.3046259582042694,
|
2712 |
+
"learning_rate": 2.940622609309716e-05,
|
2713 |
+
"loss": 2.757,
|
2714 |
+
"step": 362000
|
2715 |
+
},
|
2716 |
+
{
|
2717 |
+
"epoch": 19.52,
|
2718 |
+
"grad_norm": 0.31513723731040955,
|
2719 |
+
"learning_rate": 2.64667804390043e-05,
|
2720 |
+
"loss": 2.7601,
|
2721 |
+
"step": 363000
|
2722 |
+
},
|
2723 |
+
{
|
2724 |
+
"epoch": 19.58,
|
2725 |
+
"grad_norm": 0.31137752532958984,
|
2726 |
+
"learning_rate": 2.35243923968693e-05,
|
2727 |
+
"loss": 2.7618,
|
2728 |
+
"step": 364000
|
2729 |
+
},
|
2730 |
+
{
|
2731 |
+
"epoch": 19.63,
|
2732 |
+
"grad_norm": 0.3485892117023468,
|
2733 |
+
"learning_rate": 2.0582004354734303e-05,
|
2734 |
+
"loss": 2.7603,
|
2735 |
+
"step": 365000
|
2736 |
+
},
|
2737 |
+
{
|
2738 |
+
"epoch": 19.68,
|
2739 |
+
"grad_norm": 0.3177970051765442,
|
2740 |
+
"learning_rate": 1.7639616312599305e-05,
|
2741 |
+
"loss": 2.7573,
|
2742 |
+
"step": 366000
|
2743 |
+
},
|
2744 |
+
{
|
2745 |
+
"epoch": 19.74,
|
2746 |
+
"grad_norm": 0.3040596842765808,
|
2747 |
+
"learning_rate": 1.4700170658506444e-05,
|
2748 |
+
"loss": 2.7569,
|
2749 |
+
"step": 367000
|
2750 |
+
},
|
2751 |
+
{
|
2752 |
+
"epoch": 19.79,
|
2753 |
+
"grad_norm": 0.31309446692466736,
|
2754 |
+
"learning_rate": 1.1757782616371448e-05,
|
2755 |
+
"loss": 2.7592,
|
2756 |
+
"step": 368000
|
2757 |
+
},
|
2758 |
+
{
|
2759 |
+
"epoch": 19.85,
|
2760 |
+
"grad_norm": 0.3122311234474182,
|
2761 |
+
"learning_rate": 8.815394574236451e-06,
|
2762 |
+
"loss": 2.761,
|
2763 |
+
"step": 369000
|
2764 |
+
},
|
2765 |
+
{
|
2766 |
+
"epoch": 19.9,
|
2767 |
+
"grad_norm": 0.30037108063697815,
|
2768 |
+
"learning_rate": 5.8759489201435885e-06,
|
2769 |
+
"loss": 2.7553,
|
2770 |
+
"step": 370000
|
2771 |
+
},
|
2772 |
+
{
|
2773 |
+
"epoch": 19.95,
|
2774 |
+
"grad_norm": 0.33223336935043335,
|
2775 |
+
"learning_rate": 2.933560878008592e-06,
|
2776 |
+
"loss": 2.7606,
|
2777 |
+
"step": 371000
|
2778 |
+
},
|
2779 |
+
{
|
2780 |
+
"epoch": 20.0,
|
2781 |
+
"eval_accuracy": 0.41125847180686265,
|
2782 |
+
"eval_loss": 3.397514581680298,
|
2783 |
+
"eval_runtime": 153.9424,
|
2784 |
+
"eval_samples_per_second": 376.251,
|
2785 |
+
"eval_steps_per_second": 5.885,
|
2786 |
+
"step": 371860
|
2787 |
+
},
|
2788 |
+
{
|
2789 |
+
"epoch": 20.0,
|
2790 |
+
"step": 371860,
|
2791 |
+
"total_flos": 1.56674405385216e+18,
|
2792 |
+
"train_loss": 3.027371880730273,
|
2793 |
+
"train_runtime": 81125.0441,
|
2794 |
+
"train_samples_per_second": 146.679,
|
2795 |
+
"train_steps_per_second": 4.584
|
2796 |
+
}
|
2797 |
+
],
|
2798 |
+
"logging_steps": 1000,
|
2799 |
+
"max_steps": 371860,
|
2800 |
+
"num_input_tokens_seen": 0,
|
2801 |
+
"num_train_epochs": 20,
|
2802 |
+
"save_steps": 5000,
|
2803 |
+
"total_flos": 1.56674405385216e+18,
|
2804 |
+
"train_batch_size": 32,
|
2805 |
+
"trial_name": null,
|
2806 |
+
"trial_params": null
|
2807 |
+
}
|