Commit
•
434d56f
1
Parent(s):
2515d8c
End of training
Browse files
all_results.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 10.0,
|
| 3 |
+
"eval_accuracy": 0.752547554347826,
|
| 4 |
+
"eval_loss": 1.0725239515304565,
|
| 5 |
+
"eval_runtime": 145.877,
|
| 6 |
+
"eval_samples_per_second": 40.363,
|
| 7 |
+
"eval_steps_per_second": 2.523,
|
| 8 |
+
"train_loss": 0.8764074790546461,
|
| 9 |
+
"train_runtime": 7685.9678,
|
| 10 |
+
"train_samples_per_second": 28.876,
|
| 11 |
+
"train_steps_per_second": 0.903
|
| 12 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 10.0,
|
| 3 |
+
"eval_accuracy": 0.752547554347826,
|
| 4 |
+
"eval_loss": 1.0725239515304565,
|
| 5 |
+
"eval_runtime": 145.877,
|
| 6 |
+
"eval_samples_per_second": 40.363,
|
| 7 |
+
"eval_steps_per_second": 2.523
|
| 8 |
+
}
|
runs/Feb24_10-11-38_sanchit-single-v100/events.out.tfevents.1677241406.sanchit-single-v100.22712.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:76641783d9a3cd6e5b2ce9f609cad3c70edd14b712ec57dc5ba498fed087ae9f
|
| 3 |
+
size 363
|
train_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 10.0,
|
| 3 |
+
"train_loss": 0.8764074790546461,
|
| 4 |
+
"train_runtime": 7685.9678,
|
| 5 |
+
"train_samples_per_second": 28.876,
|
| 6 |
+
"train_steps_per_second": 0.903
|
| 7 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1777 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.752547554347826,
|
| 3 |
+
"best_model_checkpoint": "whisper-base-ft-common-language-id/checkpoint-6940",
|
| 4 |
+
"epoch": 10.0,
|
| 5 |
+
"global_step": 6940,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.04,
|
| 12 |
+
"learning_rate": 3.602305475504323e-07,
|
| 13 |
+
"loss": 3.8186,
|
| 14 |
+
"step": 25
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.07,
|
| 18 |
+
"learning_rate": 7.204610951008646e-07,
|
| 19 |
+
"loss": 3.819,
|
| 20 |
+
"step": 50
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.11,
|
| 24 |
+
"learning_rate": 1.0806916426512968e-06,
|
| 25 |
+
"loss": 3.8193,
|
| 26 |
+
"step": 75
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.14,
|
| 30 |
+
"learning_rate": 1.4409221902017292e-06,
|
| 31 |
+
"loss": 3.8144,
|
| 32 |
+
"step": 100
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.18,
|
| 36 |
+
"learning_rate": 1.8011527377521614e-06,
|
| 37 |
+
"loss": 3.8163,
|
| 38 |
+
"step": 125
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.22,
|
| 42 |
+
"learning_rate": 2.1613832853025936e-06,
|
| 43 |
+
"loss": 3.8082,
|
| 44 |
+
"step": 150
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.25,
|
| 48 |
+
"learning_rate": 2.521613832853026e-06,
|
| 49 |
+
"loss": 3.8033,
|
| 50 |
+
"step": 175
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.29,
|
| 54 |
+
"learning_rate": 2.8818443804034585e-06,
|
| 55 |
+
"loss": 3.7872,
|
| 56 |
+
"step": 200
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.32,
|
| 60 |
+
"learning_rate": 3.242074927953891e-06,
|
| 61 |
+
"loss": 3.7689,
|
| 62 |
+
"step": 225
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.36,
|
| 66 |
+
"learning_rate": 3.602305475504323e-06,
|
| 67 |
+
"loss": 3.7204,
|
| 68 |
+
"step": 250
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.4,
|
| 72 |
+
"learning_rate": 3.962536023054755e-06,
|
| 73 |
+
"loss": 3.6704,
|
| 74 |
+
"step": 275
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.43,
|
| 78 |
+
"learning_rate": 4.322766570605187e-06,
|
| 79 |
+
"loss": 3.5733,
|
| 80 |
+
"step": 300
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.47,
|
| 84 |
+
"learning_rate": 4.68299711815562e-06,
|
| 85 |
+
"loss": 3.4915,
|
| 86 |
+
"step": 325
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.5,
|
| 90 |
+
"learning_rate": 5.043227665706052e-06,
|
| 91 |
+
"loss": 3.4345,
|
| 92 |
+
"step": 350
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.54,
|
| 96 |
+
"learning_rate": 5.403458213256485e-06,
|
| 97 |
+
"loss": 3.3252,
|
| 98 |
+
"step": 375
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.58,
|
| 102 |
+
"learning_rate": 5.763688760806917e-06,
|
| 103 |
+
"loss": 3.2767,
|
| 104 |
+
"step": 400
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.61,
|
| 108 |
+
"learning_rate": 6.123919308357349e-06,
|
| 109 |
+
"loss": 3.2485,
|
| 110 |
+
"step": 425
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.65,
|
| 114 |
+
"learning_rate": 6.484149855907782e-06,
|
| 115 |
+
"loss": 3.1341,
|
| 116 |
+
"step": 450
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.68,
|
| 120 |
+
"learning_rate": 6.844380403458214e-06,
|
| 121 |
+
"loss": 3.0653,
|
| 122 |
+
"step": 475
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.72,
|
| 126 |
+
"learning_rate": 7.190201729106629e-06,
|
| 127 |
+
"loss": 3.0425,
|
| 128 |
+
"step": 500
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.76,
|
| 132 |
+
"learning_rate": 7.550432276657062e-06,
|
| 133 |
+
"loss": 2.9207,
|
| 134 |
+
"step": 525
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.79,
|
| 138 |
+
"learning_rate": 7.910662824207493e-06,
|
| 139 |
+
"loss": 2.8703,
|
| 140 |
+
"step": 550
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.83,
|
| 144 |
+
"learning_rate": 8.270893371757926e-06,
|
| 145 |
+
"loss": 2.836,
|
| 146 |
+
"step": 575
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.86,
|
| 150 |
+
"learning_rate": 8.631123919308358e-06,
|
| 151 |
+
"loss": 2.7754,
|
| 152 |
+
"step": 600
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.9,
|
| 156 |
+
"learning_rate": 8.991354466858791e-06,
|
| 157 |
+
"loss": 2.6743,
|
| 158 |
+
"step": 625
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.94,
|
| 162 |
+
"learning_rate": 9.351585014409222e-06,
|
| 163 |
+
"loss": 2.625,
|
| 164 |
+
"step": 650
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.97,
|
| 168 |
+
"learning_rate": 9.711815561959655e-06,
|
| 169 |
+
"loss": 2.5291,
|
| 170 |
+
"step": 675
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 1.0,
|
| 174 |
+
"eval_accuracy": 0.48063858695652173,
|
| 175 |
+
"eval_loss": 2.4787049293518066,
|
| 176 |
+
"eval_runtime": 144.8732,
|
| 177 |
+
"eval_samples_per_second": 40.642,
|
| 178 |
+
"eval_steps_per_second": 2.54,
|
| 179 |
+
"step": 694
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 1.01,
|
| 183 |
+
"learning_rate": 9.991994876721104e-06,
|
| 184 |
+
"loss": 2.5502,
|
| 185 |
+
"step": 700
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 1.04,
|
| 189 |
+
"learning_rate": 9.951969260326609e-06,
|
| 190 |
+
"loss": 2.4174,
|
| 191 |
+
"step": 725
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 1.08,
|
| 195 |
+
"learning_rate": 9.911943643932117e-06,
|
| 196 |
+
"loss": 2.3037,
|
| 197 |
+
"step": 750
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 1.12,
|
| 201 |
+
"learning_rate": 9.871918027537624e-06,
|
| 202 |
+
"loss": 2.257,
|
| 203 |
+
"step": 775
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 1.15,
|
| 207 |
+
"learning_rate": 9.831892411143133e-06,
|
| 208 |
+
"loss": 2.2217,
|
| 209 |
+
"step": 800
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 1.19,
|
| 213 |
+
"learning_rate": 9.79186679474864e-06,
|
| 214 |
+
"loss": 2.1369,
|
| 215 |
+
"step": 825
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 1.22,
|
| 219 |
+
"learning_rate": 9.751841178354148e-06,
|
| 220 |
+
"loss": 2.0743,
|
| 221 |
+
"step": 850
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.26,
|
| 225 |
+
"learning_rate": 9.711815561959655e-06,
|
| 226 |
+
"loss": 2.09,
|
| 227 |
+
"step": 875
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 1.3,
|
| 231 |
+
"learning_rate": 9.671789945565162e-06,
|
| 232 |
+
"loss": 2.0694,
|
| 233 |
+
"step": 900
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 1.33,
|
| 237 |
+
"learning_rate": 9.631764329170669e-06,
|
| 238 |
+
"loss": 2.0017,
|
| 239 |
+
"step": 925
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 1.37,
|
| 243 |
+
"learning_rate": 9.591738712776178e-06,
|
| 244 |
+
"loss": 1.9875,
|
| 245 |
+
"step": 950
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 1.4,
|
| 249 |
+
"learning_rate": 9.551713096381685e-06,
|
| 250 |
+
"loss": 1.9366,
|
| 251 |
+
"step": 975
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 1.44,
|
| 255 |
+
"learning_rate": 9.511687479987193e-06,
|
| 256 |
+
"loss": 1.8543,
|
| 257 |
+
"step": 1000
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 1.48,
|
| 261 |
+
"learning_rate": 9.4716618635927e-06,
|
| 262 |
+
"loss": 1.8858,
|
| 263 |
+
"step": 1025
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 1.51,
|
| 267 |
+
"learning_rate": 9.431636247198209e-06,
|
| 268 |
+
"loss": 1.8539,
|
| 269 |
+
"step": 1050
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 1.55,
|
| 273 |
+
"learning_rate": 9.391610630803715e-06,
|
| 274 |
+
"loss": 1.8339,
|
| 275 |
+
"step": 1075
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 1.59,
|
| 279 |
+
"learning_rate": 9.351585014409222e-06,
|
| 280 |
+
"loss": 1.8022,
|
| 281 |
+
"step": 1100
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 1.62,
|
| 285 |
+
"learning_rate": 9.31155939801473e-06,
|
| 286 |
+
"loss": 1.7081,
|
| 287 |
+
"step": 1125
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 1.66,
|
| 291 |
+
"learning_rate": 9.271533781620238e-06,
|
| 292 |
+
"loss": 1.7609,
|
| 293 |
+
"step": 1150
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 1.69,
|
| 297 |
+
"learning_rate": 9.233109189881525e-06,
|
| 298 |
+
"loss": 1.7326,
|
| 299 |
+
"step": 1175
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 1.73,
|
| 303 |
+
"learning_rate": 9.193083573487034e-06,
|
| 304 |
+
"loss": 1.7605,
|
| 305 |
+
"step": 1200
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 1.77,
|
| 309 |
+
"learning_rate": 9.15305795709254e-06,
|
| 310 |
+
"loss": 1.7325,
|
| 311 |
+
"step": 1225
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 1.8,
|
| 315 |
+
"learning_rate": 9.113032340698047e-06,
|
| 316 |
+
"loss": 1.6518,
|
| 317 |
+
"step": 1250
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 1.84,
|
| 321 |
+
"learning_rate": 9.073006724303554e-06,
|
| 322 |
+
"loss": 1.5706,
|
| 323 |
+
"step": 1275
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 1.87,
|
| 327 |
+
"learning_rate": 9.032981107909063e-06,
|
| 328 |
+
"loss": 1.6398,
|
| 329 |
+
"step": 1300
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 1.91,
|
| 333 |
+
"learning_rate": 8.99295549151457e-06,
|
| 334 |
+
"loss": 1.58,
|
| 335 |
+
"step": 1325
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 1.95,
|
| 339 |
+
"learning_rate": 8.952929875120078e-06,
|
| 340 |
+
"loss": 1.4585,
|
| 341 |
+
"step": 1350
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 1.98,
|
| 345 |
+
"learning_rate": 8.912904258725585e-06,
|
| 346 |
+
"loss": 1.5801,
|
| 347 |
+
"step": 1375
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 2.0,
|
| 351 |
+
"eval_accuracy": 0.6260190217391305,
|
| 352 |
+
"eval_loss": 1.6258126497268677,
|
| 353 |
+
"eval_runtime": 142.9849,
|
| 354 |
+
"eval_samples_per_second": 41.179,
|
| 355 |
+
"eval_steps_per_second": 2.574,
|
| 356 |
+
"step": 1388
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.02,
|
| 360 |
+
"learning_rate": 8.872878642331094e-06,
|
| 361 |
+
"loss": 1.4591,
|
| 362 |
+
"step": 1400
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 2.05,
|
| 366 |
+
"learning_rate": 8.832853025936599e-06,
|
| 367 |
+
"loss": 1.3388,
|
| 368 |
+
"step": 1425
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 2.09,
|
| 372 |
+
"learning_rate": 8.792827409542108e-06,
|
| 373 |
+
"loss": 1.2865,
|
| 374 |
+
"step": 1450
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 2.13,
|
| 378 |
+
"learning_rate": 8.752801793147614e-06,
|
| 379 |
+
"loss": 1.3659,
|
| 380 |
+
"step": 1475
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 2.16,
|
| 384 |
+
"learning_rate": 8.712776176753123e-06,
|
| 385 |
+
"loss": 1.3075,
|
| 386 |
+
"step": 1500
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 2.2,
|
| 390 |
+
"learning_rate": 8.67275056035863e-06,
|
| 391 |
+
"loss": 1.2352,
|
| 392 |
+
"step": 1525
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 2.23,
|
| 396 |
+
"learning_rate": 8.632724943964139e-06,
|
| 397 |
+
"loss": 1.299,
|
| 398 |
+
"step": 1550
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 2.27,
|
| 402 |
+
"learning_rate": 8.592699327569645e-06,
|
| 403 |
+
"loss": 1.2533,
|
| 404 |
+
"step": 1575
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 2.31,
|
| 408 |
+
"learning_rate": 8.552673711175152e-06,
|
| 409 |
+
"loss": 1.2684,
|
| 410 |
+
"step": 1600
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 2.34,
|
| 414 |
+
"learning_rate": 8.51264809478066e-06,
|
| 415 |
+
"loss": 1.2821,
|
| 416 |
+
"step": 1625
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 2.38,
|
| 420 |
+
"learning_rate": 8.472622478386168e-06,
|
| 421 |
+
"loss": 1.165,
|
| 422 |
+
"step": 1650
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 2.41,
|
| 426 |
+
"learning_rate": 8.432596861991675e-06,
|
| 427 |
+
"loss": 1.2305,
|
| 428 |
+
"step": 1675
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 2.45,
|
| 432 |
+
"learning_rate": 8.392571245597183e-06,
|
| 433 |
+
"loss": 1.1679,
|
| 434 |
+
"step": 1700
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 2.49,
|
| 438 |
+
"learning_rate": 8.35254562920269e-06,
|
| 439 |
+
"loss": 1.135,
|
| 440 |
+
"step": 1725
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 2.52,
|
| 444 |
+
"learning_rate": 8.312520012808199e-06,
|
| 445 |
+
"loss": 1.134,
|
| 446 |
+
"step": 1750
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 2.56,
|
| 450 |
+
"learning_rate": 8.272494396413706e-06,
|
| 451 |
+
"loss": 1.111,
|
| 452 |
+
"step": 1775
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 2.59,
|
| 456 |
+
"learning_rate": 8.232468780019213e-06,
|
| 457 |
+
"loss": 1.0913,
|
| 458 |
+
"step": 1800
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 2.63,
|
| 462 |
+
"learning_rate": 8.192443163624721e-06,
|
| 463 |
+
"loss": 1.0577,
|
| 464 |
+
"step": 1825
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 2.67,
|
| 468 |
+
"learning_rate": 8.152417547230228e-06,
|
| 469 |
+
"loss": 1.0247,
|
| 470 |
+
"step": 1850
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 2.7,
|
| 474 |
+
"learning_rate": 8.112391930835735e-06,
|
| 475 |
+
"loss": 1.0808,
|
| 476 |
+
"step": 1875
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 2.74,
|
| 480 |
+
"learning_rate": 8.072366314441244e-06,
|
| 481 |
+
"loss": 1.0829,
|
| 482 |
+
"step": 1900
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 2.77,
|
| 486 |
+
"learning_rate": 8.03234069804675e-06,
|
| 487 |
+
"loss": 1.0763,
|
| 488 |
+
"step": 1925
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 2.81,
|
| 492 |
+
"learning_rate": 7.992315081652257e-06,
|
| 493 |
+
"loss": 1.1263,
|
| 494 |
+
"step": 1950
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 2.85,
|
| 498 |
+
"learning_rate": 7.952289465257766e-06,
|
| 499 |
+
"loss": 1.13,
|
| 500 |
+
"step": 1975
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 2.88,
|
| 504 |
+
"learning_rate": 7.912263848863273e-06,
|
| 505 |
+
"loss": 1.0864,
|
| 506 |
+
"step": 2000
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 2.92,
|
| 510 |
+
"learning_rate": 7.872238232468781e-06,
|
| 511 |
+
"loss": 0.9774,
|
| 512 |
+
"step": 2025
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 2.95,
|
| 516 |
+
"learning_rate": 7.832212616074288e-06,
|
| 517 |
+
"loss": 1.0433,
|
| 518 |
+
"step": 2050
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 2.99,
|
| 522 |
+
"learning_rate": 7.792186999679795e-06,
|
| 523 |
+
"loss": 1.0144,
|
| 524 |
+
"step": 2075
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 3.0,
|
| 528 |
+
"eval_accuracy": 0.6815557065217391,
|
| 529 |
+
"eval_loss": 1.288601279258728,
|
| 530 |
+
"eval_runtime": 143.9643,
|
| 531 |
+
"eval_samples_per_second": 40.899,
|
| 532 |
+
"eval_steps_per_second": 2.556,
|
| 533 |
+
"step": 2082
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 3.03,
|
| 537 |
+
"learning_rate": 7.752161383285304e-06,
|
| 538 |
+
"loss": 0.8903,
|
| 539 |
+
"step": 2100
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 3.06,
|
| 543 |
+
"learning_rate": 7.71213576689081e-06,
|
| 544 |
+
"loss": 0.8757,
|
| 545 |
+
"step": 2125
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 3.1,
|
| 549 |
+
"learning_rate": 7.672110150496318e-06,
|
| 550 |
+
"loss": 0.8285,
|
| 551 |
+
"step": 2150
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 3.13,
|
| 555 |
+
"learning_rate": 7.632084534101826e-06,
|
| 556 |
+
"loss": 0.7831,
|
| 557 |
+
"step": 2175
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 3.17,
|
| 561 |
+
"learning_rate": 7.592058917707333e-06,
|
| 562 |
+
"loss": 0.8158,
|
| 563 |
+
"step": 2200
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 3.21,
|
| 567 |
+
"learning_rate": 7.552033301312841e-06,
|
| 568 |
+
"loss": 0.8271,
|
| 569 |
+
"step": 2225
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 3.24,
|
| 573 |
+
"learning_rate": 7.512007684918349e-06,
|
| 574 |
+
"loss": 0.8424,
|
| 575 |
+
"step": 2250
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 3.28,
|
| 579 |
+
"learning_rate": 7.4719820685238555e-06,
|
| 580 |
+
"loss": 0.7962,
|
| 581 |
+
"step": 2275
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 3.31,
|
| 585 |
+
"learning_rate": 7.431956452129363e-06,
|
| 586 |
+
"loss": 0.7954,
|
| 587 |
+
"step": 2300
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 3.35,
|
| 591 |
+
"learning_rate": 7.391930835734871e-06,
|
| 592 |
+
"loss": 0.7962,
|
| 593 |
+
"step": 2325
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 3.39,
|
| 597 |
+
"learning_rate": 7.351905219340379e-06,
|
| 598 |
+
"loss": 0.8528,
|
| 599 |
+
"step": 2350
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 3.42,
|
| 603 |
+
"learning_rate": 7.311879602945886e-06,
|
| 604 |
+
"loss": 0.7725,
|
| 605 |
+
"step": 2375
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 3.46,
|
| 609 |
+
"learning_rate": 7.271853986551393e-06,
|
| 610 |
+
"loss": 0.8015,
|
| 611 |
+
"step": 2400
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 3.49,
|
| 615 |
+
"learning_rate": 7.231828370156901e-06,
|
| 616 |
+
"loss": 0.6676,
|
| 617 |
+
"step": 2425
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 3.53,
|
| 621 |
+
"learning_rate": 7.191802753762408e-06,
|
| 622 |
+
"loss": 0.7208,
|
| 623 |
+
"step": 2450
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 3.57,
|
| 627 |
+
"learning_rate": 7.151777137367916e-06,
|
| 628 |
+
"loss": 0.8135,
|
| 629 |
+
"step": 2475
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 3.6,
|
| 633 |
+
"learning_rate": 7.1117515209734235e-06,
|
| 634 |
+
"loss": 0.8313,
|
| 635 |
+
"step": 2500
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 3.64,
|
| 639 |
+
"learning_rate": 7.071725904578931e-06,
|
| 640 |
+
"loss": 0.7266,
|
| 641 |
+
"step": 2525
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 3.67,
|
| 645 |
+
"learning_rate": 7.031700288184439e-06,
|
| 646 |
+
"loss": 0.7825,
|
| 647 |
+
"step": 2550
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 3.71,
|
| 651 |
+
"learning_rate": 6.991674671789947e-06,
|
| 652 |
+
"loss": 0.7119,
|
| 653 |
+
"step": 2575
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 3.75,
|
| 657 |
+
"learning_rate": 6.951649055395454e-06,
|
| 658 |
+
"loss": 0.8349,
|
| 659 |
+
"step": 2600
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 3.78,
|
| 663 |
+
"learning_rate": 6.9116234390009606e-06,
|
| 664 |
+
"loss": 0.6898,
|
| 665 |
+
"step": 2625
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 3.82,
|
| 669 |
+
"learning_rate": 6.871597822606468e-06,
|
| 670 |
+
"loss": 0.7068,
|
| 671 |
+
"step": 2650
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 3.85,
|
| 675 |
+
"learning_rate": 6.831572206211976e-06,
|
| 676 |
+
"loss": 0.7628,
|
| 677 |
+
"step": 2675
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 3.89,
|
| 681 |
+
"learning_rate": 6.791546589817484e-06,
|
| 682 |
+
"loss": 0.6544,
|
| 683 |
+
"step": 2700
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 3.93,
|
| 687 |
+
"learning_rate": 6.7515209734229915e-06,
|
| 688 |
+
"loss": 0.7577,
|
| 689 |
+
"step": 2725
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 3.96,
|
| 693 |
+
"learning_rate": 6.711495357028499e-06,
|
| 694 |
+
"loss": 0.7262,
|
| 695 |
+
"step": 2750
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 4.0,
|
| 699 |
+
"learning_rate": 6.671469740634007e-06,
|
| 700 |
+
"loss": 0.7442,
|
| 701 |
+
"step": 2775
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 4.0,
|
| 705 |
+
"eval_accuracy": 0.7236752717391305,
|
| 706 |
+
"eval_loss": 1.0782954692840576,
|
| 707 |
+
"eval_runtime": 142.9904,
|
| 708 |
+
"eval_samples_per_second": 41.178,
|
| 709 |
+
"eval_steps_per_second": 2.574,
|
| 710 |
+
"step": 2776
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 4.03,
|
| 714 |
+
"learning_rate": 6.631444124239513e-06,
|
| 715 |
+
"loss": 0.5559,
|
| 716 |
+
"step": 2800
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 4.07,
|
| 720 |
+
"learning_rate": 6.591418507845021e-06,
|
| 721 |
+
"loss": 0.5587,
|
| 722 |
+
"step": 2825
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 4.11,
|
| 726 |
+
"learning_rate": 6.551392891450529e-06,
|
| 727 |
+
"loss": 0.5495,
|
| 728 |
+
"step": 2850
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 4.14,
|
| 732 |
+
"learning_rate": 6.511367275056036e-06,
|
| 733 |
+
"loss": 0.519,
|
| 734 |
+
"step": 2875
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 4.18,
|
| 738 |
+
"learning_rate": 6.471341658661544e-06,
|
| 739 |
+
"loss": 0.5675,
|
| 740 |
+
"step": 2900
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 4.21,
|
| 744 |
+
"learning_rate": 6.431316042267052e-06,
|
| 745 |
+
"loss": 0.509,
|
| 746 |
+
"step": 2925
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 4.25,
|
| 750 |
+
"learning_rate": 6.3912904258725596e-06,
|
| 751 |
+
"loss": 0.5276,
|
| 752 |
+
"step": 2950
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 4.29,
|
| 756 |
+
"learning_rate": 6.351264809478066e-06,
|
| 757 |
+
"loss": 0.5094,
|
| 758 |
+
"step": 2975
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 4.32,
|
| 762 |
+
"learning_rate": 6.311239193083573e-06,
|
| 763 |
+
"loss": 0.5092,
|
| 764 |
+
"step": 3000
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 4.36,
|
| 768 |
+
"learning_rate": 6.271213576689081e-06,
|
| 769 |
+
"loss": 0.5025,
|
| 770 |
+
"step": 3025
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 4.39,
|
| 774 |
+
"learning_rate": 6.231187960294589e-06,
|
| 775 |
+
"loss": 0.5393,
|
| 776 |
+
"step": 3050
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 4.43,
|
| 780 |
+
"learning_rate": 6.191162343900097e-06,
|
| 781 |
+
"loss": 0.5807,
|
| 782 |
+
"step": 3075
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 4.47,
|
| 786 |
+
"learning_rate": 6.151136727505604e-06,
|
| 787 |
+
"loss": 0.5582,
|
| 788 |
+
"step": 3100
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 4.5,
|
| 792 |
+
"learning_rate": 6.111111111111112e-06,
|
| 793 |
+
"loss": 0.5373,
|
| 794 |
+
"step": 3125
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 4.54,
|
| 798 |
+
"learning_rate": 6.071085494716619e-06,
|
| 799 |
+
"loss": 0.5123,
|
| 800 |
+
"step": 3150
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 4.57,
|
| 804 |
+
"learning_rate": 6.031059878322127e-06,
|
| 805 |
+
"loss": 0.5087,
|
| 806 |
+
"step": 3175
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 4.61,
|
| 810 |
+
"learning_rate": 5.992635286583414e-06,
|
| 811 |
+
"loss": 0.5189,
|
| 812 |
+
"step": 3200
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 4.65,
|
| 816 |
+
"learning_rate": 5.9526096701889215e-06,
|
| 817 |
+
"loss": 0.5056,
|
| 818 |
+
"step": 3225
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 4.68,
|
| 822 |
+
"learning_rate": 5.912584053794429e-06,
|
| 823 |
+
"loss": 0.5393,
|
| 824 |
+
"step": 3250
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 4.72,
|
| 828 |
+
"learning_rate": 5.872558437399937e-06,
|
| 829 |
+
"loss": 0.5533,
|
| 830 |
+
"step": 3275
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 4.76,
|
| 834 |
+
"learning_rate": 5.832532821005445e-06,
|
| 835 |
+
"loss": 0.535,
|
| 836 |
+
"step": 3300
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 4.79,
|
| 840 |
+
"learning_rate": 5.794108229266732e-06,
|
| 841 |
+
"loss": 0.618,
|
| 842 |
+
"step": 3325
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 4.83,
|
| 846 |
+
"learning_rate": 5.754082612872239e-06,
|
| 847 |
+
"loss": 0.4511,
|
| 848 |
+
"step": 3350
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 4.86,
|
| 852 |
+
"learning_rate": 5.7140569964777456e-06,
|
| 853 |
+
"loss": 0.49,
|
| 854 |
+
"step": 3375
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 4.9,
|
| 858 |
+
"learning_rate": 5.674031380083253e-06,
|
| 859 |
+
"loss": 0.5176,
|
| 860 |
+
"step": 3400
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 4.94,
|
| 864 |
+
"learning_rate": 5.634005763688761e-06,
|
| 865 |
+
"loss": 0.4858,
|
| 866 |
+
"step": 3425
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 4.97,
|
| 870 |
+
"learning_rate": 5.593980147294269e-06,
|
| 871 |
+
"loss": 0.4802,
|
| 872 |
+
"step": 3450
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 5.0,
|
| 876 |
+
"eval_accuracy": 0.7265625,
|
| 877 |
+
"eval_loss": 1.0582159757614136,
|
| 878 |
+
"eval_runtime": 141.9251,
|
| 879 |
+
"eval_samples_per_second": 41.487,
|
| 880 |
+
"eval_steps_per_second": 2.593,
|
| 881 |
+
"step": 3470
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 5.01,
|
| 885 |
+
"learning_rate": 5.5539545308997765e-06,
|
| 886 |
+
"loss": 0.5603,
|
| 887 |
+
"step": 3475
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 5.04,
|
| 891 |
+
"learning_rate": 5.513928914505284e-06,
|
| 892 |
+
"loss": 0.3679,
|
| 893 |
+
"step": 3500
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 5.08,
|
| 897 |
+
"learning_rate": 5.473903298110791e-06,
|
| 898 |
+
"loss": 0.3718,
|
| 899 |
+
"step": 3525
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 5.12,
|
| 903 |
+
"learning_rate": 5.433877681716299e-06,
|
| 904 |
+
"loss": 0.3868,
|
| 905 |
+
"step": 3550
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 5.15,
|
| 909 |
+
"learning_rate": 5.393852065321807e-06,
|
| 910 |
+
"loss": 0.4013,
|
| 911 |
+
"step": 3575
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 5.19,
|
| 915 |
+
"learning_rate": 5.3538264489273136e-06,
|
| 916 |
+
"loss": 0.4519,
|
| 917 |
+
"step": 3600
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 5.22,
|
| 921 |
+
"learning_rate": 5.313800832532821e-06,
|
| 922 |
+
"loss": 0.3731,
|
| 923 |
+
"step": 3625
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 5.26,
|
| 927 |
+
"learning_rate": 5.273775216138329e-06,
|
| 928 |
+
"loss": 0.39,
|
| 929 |
+
"step": 3650
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 5.3,
|
| 933 |
+
"learning_rate": 5.233749599743837e-06,
|
| 934 |
+
"loss": 0.3997,
|
| 935 |
+
"step": 3675
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 5.33,
|
| 939 |
+
"learning_rate": 5.193723983349344e-06,
|
| 940 |
+
"loss": 0.3921,
|
| 941 |
+
"step": 3700
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 5.37,
|
| 945 |
+
"learning_rate": 5.1536983669548514e-06,
|
| 946 |
+
"loss": 0.3662,
|
| 947 |
+
"step": 3725
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 5.4,
|
| 951 |
+
"learning_rate": 5.113672750560359e-06,
|
| 952 |
+
"loss": 0.3706,
|
| 953 |
+
"step": 3750
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 5.44,
|
| 957 |
+
"learning_rate": 5.073647134165867e-06,
|
| 958 |
+
"loss": 0.372,
|
| 959 |
+
"step": 3775
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 5.48,
|
| 963 |
+
"learning_rate": 5.033621517771375e-06,
|
| 964 |
+
"loss": 0.3909,
|
| 965 |
+
"step": 3800
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 5.51,
|
| 969 |
+
"learning_rate": 4.993595901376882e-06,
|
| 970 |
+
"loss": 0.3774,
|
| 971 |
+
"step": 3825
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 5.55,
|
| 975 |
+
"learning_rate": 4.953570284982389e-06,
|
| 976 |
+
"loss": 0.3634,
|
| 977 |
+
"step": 3850
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 5.58,
|
| 981 |
+
"learning_rate": 4.913544668587897e-06,
|
| 982 |
+
"loss": 0.3733,
|
| 983 |
+
"step": 3875
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 5.62,
|
| 987 |
+
"learning_rate": 4.873519052193404e-06,
|
| 988 |
+
"loss": 0.2952,
|
| 989 |
+
"step": 3900
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 5.66,
|
| 993 |
+
"learning_rate": 4.833493435798912e-06,
|
| 994 |
+
"loss": 0.3792,
|
| 995 |
+
"step": 3925
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 5.69,
|
| 999 |
+
"learning_rate": 4.7934678194044195e-06,
|
| 1000 |
+
"loss": 0.404,
|
| 1001 |
+
"step": 3950
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 5.73,
|
| 1005 |
+
"learning_rate": 4.753442203009927e-06,
|
| 1006 |
+
"loss": 0.3741,
|
| 1007 |
+
"step": 3975
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 5.76,
|
| 1011 |
+
"learning_rate": 4.713416586615434e-06,
|
| 1012 |
+
"loss": 0.3327,
|
| 1013 |
+
"step": 4000
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 5.8,
|
| 1017 |
+
"learning_rate": 4.673390970220942e-06,
|
| 1018 |
+
"loss": 0.3472,
|
| 1019 |
+
"step": 4025
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 5.84,
|
| 1023 |
+
"learning_rate": 4.63336535382645e-06,
|
| 1024 |
+
"loss": 0.3365,
|
| 1025 |
+
"step": 4050
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 5.87,
|
| 1029 |
+
"learning_rate": 4.5933397374319565e-06,
|
| 1030 |
+
"loss": 0.3276,
|
| 1031 |
+
"step": 4075
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 5.91,
|
| 1035 |
+
"learning_rate": 4.553314121037464e-06,
|
| 1036 |
+
"loss": 0.3571,
|
| 1037 |
+
"step": 4100
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 5.94,
|
| 1041 |
+
"learning_rate": 4.513288504642972e-06,
|
| 1042 |
+
"loss": 0.3283,
|
| 1043 |
+
"step": 4125
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 5.98,
|
| 1047 |
+
"learning_rate": 4.47326288824848e-06,
|
| 1048 |
+
"loss": 0.3378,
|
| 1049 |
+
"step": 4150
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 6.0,
|
| 1053 |
+
"eval_accuracy": 0.7416779891304348,
|
| 1054 |
+
"eval_loss": 1.017274022102356,
|
| 1055 |
+
"eval_runtime": 145.4891,
|
| 1056 |
+
"eval_samples_per_second": 40.47,
|
| 1057 |
+
"eval_steps_per_second": 2.529,
|
| 1058 |
+
"step": 4164
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 6.02,
|
| 1062 |
+
"learning_rate": 4.433237271853987e-06,
|
| 1063 |
+
"loss": 0.2904,
|
| 1064 |
+
"step": 4175
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 6.05,
|
| 1068 |
+
"learning_rate": 4.393211655459494e-06,
|
| 1069 |
+
"loss": 0.2767,
|
| 1070 |
+
"step": 4200
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 6.09,
|
| 1074 |
+
"learning_rate": 4.353186039065002e-06,
|
| 1075 |
+
"loss": 0.3324,
|
| 1076 |
+
"step": 4225
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 6.12,
|
| 1080 |
+
"learning_rate": 4.313160422670509e-06,
|
| 1081 |
+
"loss": 0.25,
|
| 1082 |
+
"step": 4250
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 6.16,
|
| 1086 |
+
"learning_rate": 4.273134806276017e-06,
|
| 1087 |
+
"loss": 0.2794,
|
| 1088 |
+
"step": 4275
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 6.2,
|
| 1092 |
+
"learning_rate": 4.2331091898815245e-06,
|
| 1093 |
+
"loss": 0.2126,
|
| 1094 |
+
"step": 4300
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 6.23,
|
| 1098 |
+
"learning_rate": 4.193083573487031e-06,
|
| 1099 |
+
"loss": 0.2909,
|
| 1100 |
+
"step": 4325
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 6.27,
|
| 1104 |
+
"learning_rate": 4.153057957092539e-06,
|
| 1105 |
+
"loss": 0.3043,
|
| 1106 |
+
"step": 4350
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 6.3,
|
| 1110 |
+
"learning_rate": 4.113032340698047e-06,
|
| 1111 |
+
"loss": 0.248,
|
| 1112 |
+
"step": 4375
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 6.34,
|
| 1116 |
+
"learning_rate": 4.073006724303555e-06,
|
| 1117 |
+
"loss": 0.2869,
|
| 1118 |
+
"step": 4400
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 6.38,
|
| 1122 |
+
"learning_rate": 4.0329811079090615e-06,
|
| 1123 |
+
"loss": 0.2531,
|
| 1124 |
+
"step": 4425
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 6.41,
|
| 1128 |
+
"learning_rate": 3.992955491514569e-06,
|
| 1129 |
+
"loss": 0.2674,
|
| 1130 |
+
"step": 4450
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 6.45,
|
| 1134 |
+
"learning_rate": 3.952929875120077e-06,
|
| 1135 |
+
"loss": 0.2855,
|
| 1136 |
+
"step": 4475
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 6.48,
|
| 1140 |
+
"learning_rate": 3.912904258725585e-06,
|
| 1141 |
+
"loss": 0.2846,
|
| 1142 |
+
"step": 4500
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 6.52,
|
| 1146 |
+
"learning_rate": 3.8728786423310925e-06,
|
| 1147 |
+
"loss": 0.2807,
|
| 1148 |
+
"step": 4525
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 6.56,
|
| 1152 |
+
"learning_rate": 3.832853025936599e-06,
|
| 1153 |
+
"loss": 0.2874,
|
| 1154 |
+
"step": 4550
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 6.59,
|
| 1158 |
+
"learning_rate": 3.7928274095421076e-06,
|
| 1159 |
+
"loss": 0.2484,
|
| 1160 |
+
"step": 4575
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 6.63,
|
| 1164 |
+
"learning_rate": 3.7528017931476145e-06,
|
| 1165 |
+
"loss": 0.2182,
|
| 1166 |
+
"step": 4600
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 6.66,
|
| 1170 |
+
"learning_rate": 3.7127761767531222e-06,
|
| 1171 |
+
"loss": 0.2364,
|
| 1172 |
+
"step": 4625
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 6.7,
|
| 1176 |
+
"learning_rate": 3.67275056035863e-06,
|
| 1177 |
+
"loss": 0.2318,
|
| 1178 |
+
"step": 4650
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 6.74,
|
| 1182 |
+
"learning_rate": 3.6327249439641373e-06,
|
| 1183 |
+
"loss": 0.2633,
|
| 1184 |
+
"step": 4675
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 6.77,
|
| 1188 |
+
"learning_rate": 3.5926993275696446e-06,
|
| 1189 |
+
"loss": 0.2379,
|
| 1190 |
+
"step": 4700
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 6.81,
|
| 1194 |
+
"learning_rate": 3.5526737111751524e-06,
|
| 1195 |
+
"loss": 0.2937,
|
| 1196 |
+
"step": 4725
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 6.84,
|
| 1200 |
+
"learning_rate": 3.51264809478066e-06,
|
| 1201 |
+
"loss": 0.2859,
|
| 1202 |
+
"step": 4750
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 6.88,
|
| 1206 |
+
"learning_rate": 3.4726224783861674e-06,
|
| 1207 |
+
"loss": 0.2498,
|
| 1208 |
+
"step": 4775
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 6.92,
|
| 1212 |
+
"learning_rate": 3.432596861991675e-06,
|
| 1213 |
+
"loss": 0.2853,
|
| 1214 |
+
"step": 4800
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 6.95,
|
| 1218 |
+
"learning_rate": 3.3925712455971825e-06,
|
| 1219 |
+
"loss": 0.2845,
|
| 1220 |
+
"step": 4825
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 6.99,
|
| 1224 |
+
"learning_rate": 3.35254562920269e-06,
|
| 1225 |
+
"loss": 0.1941,
|
| 1226 |
+
"step": 4850
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 7.0,
|
| 1230 |
+
"eval_accuracy": 0.7445652173913043,
|
| 1231 |
+
"eval_loss": 1.0053629875183105,
|
| 1232 |
+
"eval_runtime": 144.6986,
|
| 1233 |
+
"eval_samples_per_second": 40.691,
|
| 1234 |
+
"eval_steps_per_second": 2.543,
|
| 1235 |
+
"step": 4858
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 7.02,
|
| 1239 |
+
"learning_rate": 3.3125200128081976e-06,
|
| 1240 |
+
"loss": 0.2266,
|
| 1241 |
+
"step": 4875
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 7.06,
|
| 1245 |
+
"learning_rate": 3.2724943964137053e-06,
|
| 1246 |
+
"loss": 0.2138,
|
| 1247 |
+
"step": 4900
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 7.1,
|
| 1251 |
+
"learning_rate": 3.2324687800192126e-06,
|
| 1252 |
+
"loss": 0.2221,
|
| 1253 |
+
"step": 4925
|
| 1254 |
+
},
|
| 1255 |
+
{
|
| 1256 |
+
"epoch": 7.13,
|
| 1257 |
+
"learning_rate": 3.19244316362472e-06,
|
| 1258 |
+
"loss": 0.2113,
|
| 1259 |
+
"step": 4950
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 7.17,
|
| 1263 |
+
"learning_rate": 3.1524175472302277e-06,
|
| 1264 |
+
"loss": 0.1793,
|
| 1265 |
+
"step": 4975
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 7.2,
|
| 1269 |
+
"learning_rate": 3.1123919308357354e-06,
|
| 1270 |
+
"loss": 0.2251,
|
| 1271 |
+
"step": 5000
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 7.24,
|
| 1275 |
+
"learning_rate": 3.0723663144412423e-06,
|
| 1276 |
+
"loss": 0.1947,
|
| 1277 |
+
"step": 5025
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 7.28,
|
| 1281 |
+
"learning_rate": 3.03234069804675e-06,
|
| 1282 |
+
"loss": 0.1985,
|
| 1283 |
+
"step": 5050
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 7.31,
|
| 1287 |
+
"learning_rate": 2.992315081652258e-06,
|
| 1288 |
+
"loss": 0.2318,
|
| 1289 |
+
"step": 5075
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 7.35,
|
| 1293 |
+
"learning_rate": 2.952289465257765e-06,
|
| 1294 |
+
"loss": 0.1684,
|
| 1295 |
+
"step": 5100
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 7.38,
|
| 1299 |
+
"learning_rate": 2.9122638488632725e-06,
|
| 1300 |
+
"loss": 0.1688,
|
| 1301 |
+
"step": 5125
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 7.42,
|
| 1305 |
+
"learning_rate": 2.87223823246878e-06,
|
| 1306 |
+
"loss": 0.2049,
|
| 1307 |
+
"step": 5150
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 7.46,
|
| 1311 |
+
"learning_rate": 2.832212616074288e-06,
|
| 1312 |
+
"loss": 0.1908,
|
| 1313 |
+
"step": 5175
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 7.49,
|
| 1317 |
+
"learning_rate": 2.7921869996797953e-06,
|
| 1318 |
+
"loss": 0.2174,
|
| 1319 |
+
"step": 5200
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 7.53,
|
| 1323 |
+
"learning_rate": 2.7521613832853026e-06,
|
| 1324 |
+
"loss": 0.1624,
|
| 1325 |
+
"step": 5225
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 7.56,
|
| 1329 |
+
"learning_rate": 2.7121357668908103e-06,
|
| 1330 |
+
"loss": 0.178,
|
| 1331 |
+
"step": 5250
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 7.6,
|
| 1335 |
+
"learning_rate": 2.6721101504963177e-06,
|
| 1336 |
+
"loss": 0.1888,
|
| 1337 |
+
"step": 5275
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 7.64,
|
| 1341 |
+
"learning_rate": 2.6320845341018254e-06,
|
| 1342 |
+
"loss": 0.173,
|
| 1343 |
+
"step": 5300
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 7.67,
|
| 1347 |
+
"learning_rate": 2.592058917707333e-06,
|
| 1348 |
+
"loss": 0.1769,
|
| 1349 |
+
"step": 5325
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 7.71,
|
| 1353 |
+
"learning_rate": 2.5520333013128405e-06,
|
| 1354 |
+
"loss": 0.1988,
|
| 1355 |
+
"step": 5350
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 7.74,
|
| 1359 |
+
"learning_rate": 2.512007684918348e-06,
|
| 1360 |
+
"loss": 0.2147,
|
| 1361 |
+
"step": 5375
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 7.78,
|
| 1365 |
+
"learning_rate": 2.4719820685238555e-06,
|
| 1366 |
+
"loss": 0.2033,
|
| 1367 |
+
"step": 5400
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 7.82,
|
| 1371 |
+
"learning_rate": 2.431956452129363e-06,
|
| 1372 |
+
"loss": 0.1437,
|
| 1373 |
+
"step": 5425
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 7.85,
|
| 1377 |
+
"learning_rate": 2.3919308357348706e-06,
|
| 1378 |
+
"loss": 0.1775,
|
| 1379 |
+
"step": 5450
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 7.89,
|
| 1383 |
+
"learning_rate": 2.351905219340378e-06,
|
| 1384 |
+
"loss": 0.2166,
|
| 1385 |
+
"step": 5475
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 7.93,
|
| 1389 |
+
"learning_rate": 2.3118796029458857e-06,
|
| 1390 |
+
"loss": 0.1861,
|
| 1391 |
+
"step": 5500
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 7.96,
|
| 1395 |
+
"learning_rate": 2.271853986551393e-06,
|
| 1396 |
+
"loss": 0.1677,
|
| 1397 |
+
"step": 5525
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 8.0,
|
| 1401 |
+
"learning_rate": 2.2318283701569007e-06,
|
| 1402 |
+
"loss": 0.1424,
|
| 1403 |
+
"step": 5550
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 8.0,
|
| 1407 |
+
"eval_accuracy": 0.7508491847826086,
|
| 1408 |
+
"eval_loss": 1.021315574645996,
|
| 1409 |
+
"eval_runtime": 144.0799,
|
| 1410 |
+
"eval_samples_per_second": 40.866,
|
| 1411 |
+
"eval_steps_per_second": 2.554,
|
| 1412 |
+
"step": 5552
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 8.03,
|
| 1416 |
+
"learning_rate": 2.191802753762408e-06,
|
| 1417 |
+
"loss": 0.1602,
|
| 1418 |
+
"step": 5575
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 8.07,
|
| 1422 |
+
"learning_rate": 2.1517771373679154e-06,
|
| 1423 |
+
"loss": 0.1476,
|
| 1424 |
+
"step": 5600
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 8.11,
|
| 1428 |
+
"learning_rate": 2.111751520973423e-06,
|
| 1429 |
+
"loss": 0.1191,
|
| 1430 |
+
"step": 5625
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 8.14,
|
| 1434 |
+
"learning_rate": 2.0717259045789305e-06,
|
| 1435 |
+
"loss": 0.1326,
|
| 1436 |
+
"step": 5650
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 8.18,
|
| 1440 |
+
"learning_rate": 2.031700288184438e-06,
|
| 1441 |
+
"loss": 0.1524,
|
| 1442 |
+
"step": 5675
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 8.21,
|
| 1446 |
+
"learning_rate": 1.9916746717899455e-06,
|
| 1447 |
+
"loss": 0.1016,
|
| 1448 |
+
"step": 5700
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 8.25,
|
| 1452 |
+
"learning_rate": 1.9516490553954533e-06,
|
| 1453 |
+
"loss": 0.1648,
|
| 1454 |
+
"step": 5725
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 8.29,
|
| 1458 |
+
"learning_rate": 1.911623439000961e-06,
|
| 1459 |
+
"loss": 0.1673,
|
| 1460 |
+
"step": 5750
|
| 1461 |
+
},
|
| 1462 |
+
{
|
| 1463 |
+
"epoch": 8.32,
|
| 1464 |
+
"learning_rate": 1.8715978226064681e-06,
|
| 1465 |
+
"loss": 0.1429,
|
| 1466 |
+
"step": 5775
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 8.36,
|
| 1470 |
+
"learning_rate": 1.8315722062119759e-06,
|
| 1471 |
+
"loss": 0.1407,
|
| 1472 |
+
"step": 5800
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 8.39,
|
| 1476 |
+
"learning_rate": 1.7915465898174834e-06,
|
| 1477 |
+
"loss": 0.1412,
|
| 1478 |
+
"step": 5825
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 8.43,
|
| 1482 |
+
"learning_rate": 1.751520973422991e-06,
|
| 1483 |
+
"loss": 0.1647,
|
| 1484 |
+
"step": 5850
|
| 1485 |
+
},
|
| 1486 |
+
{
|
| 1487 |
+
"epoch": 8.47,
|
| 1488 |
+
"learning_rate": 1.7114953570284985e-06,
|
| 1489 |
+
"loss": 0.1708,
|
| 1490 |
+
"step": 5875
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 8.5,
|
| 1494 |
+
"learning_rate": 1.6714697406340058e-06,
|
| 1495 |
+
"loss": 0.0913,
|
| 1496 |
+
"step": 5900
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 8.54,
|
| 1500 |
+
"learning_rate": 1.6314441242395135e-06,
|
| 1501 |
+
"loss": 0.1356,
|
| 1502 |
+
"step": 5925
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 8.57,
|
| 1506 |
+
"learning_rate": 1.5914185078450208e-06,
|
| 1507 |
+
"loss": 0.1485,
|
| 1508 |
+
"step": 5950
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 8.61,
|
| 1512 |
+
"learning_rate": 1.5513928914505286e-06,
|
| 1513 |
+
"loss": 0.1975,
|
| 1514 |
+
"step": 5975
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 8.65,
|
| 1518 |
+
"learning_rate": 1.511367275056036e-06,
|
| 1519 |
+
"loss": 0.142,
|
| 1520 |
+
"step": 6000
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 8.68,
|
| 1524 |
+
"learning_rate": 1.4713416586615434e-06,
|
| 1525 |
+
"loss": 0.1636,
|
| 1526 |
+
"step": 6025
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 8.72,
|
| 1530 |
+
"learning_rate": 1.431316042267051e-06,
|
| 1531 |
+
"loss": 0.1419,
|
| 1532 |
+
"step": 6050
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 8.75,
|
| 1536 |
+
"learning_rate": 1.3912904258725585e-06,
|
| 1537 |
+
"loss": 0.0922,
|
| 1538 |
+
"step": 6075
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 8.79,
|
| 1542 |
+
"learning_rate": 1.351264809478066e-06,
|
| 1543 |
+
"loss": 0.1593,
|
| 1544 |
+
"step": 6100
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 8.83,
|
| 1548 |
+
"learning_rate": 1.3112391930835736e-06,
|
| 1549 |
+
"loss": 0.1294,
|
| 1550 |
+
"step": 6125
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 8.86,
|
| 1554 |
+
"learning_rate": 1.2712135766890813e-06,
|
| 1555 |
+
"loss": 0.1151,
|
| 1556 |
+
"step": 6150
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 8.9,
|
| 1560 |
+
"learning_rate": 1.2311879602945886e-06,
|
| 1561 |
+
"loss": 0.1378,
|
| 1562 |
+
"step": 6175
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 8.93,
|
| 1566 |
+
"learning_rate": 1.1911623439000962e-06,
|
| 1567 |
+
"loss": 0.1484,
|
| 1568 |
+
"step": 6200
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 8.97,
|
| 1572 |
+
"learning_rate": 1.1511367275056037e-06,
|
| 1573 |
+
"loss": 0.1242,
|
| 1574 |
+
"step": 6225
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 9.0,
|
| 1578 |
+
"eval_accuracy": 0.7494904891304348,
|
| 1579 |
+
"eval_loss": 1.0566785335540771,
|
| 1580 |
+
"eval_runtime": 142.7018,
|
| 1581 |
+
"eval_samples_per_second": 41.261,
|
| 1582 |
+
"eval_steps_per_second": 2.579,
|
| 1583 |
+
"step": 6246
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"epoch": 9.01,
|
| 1587 |
+
"learning_rate": 1.112712135766891e-06,
|
| 1588 |
+
"loss": 0.1634,
|
| 1589 |
+
"step": 6250
|
| 1590 |
+
},
|
| 1591 |
+
{
|
| 1592 |
+
"epoch": 9.04,
|
| 1593 |
+
"learning_rate": 1.0726865193723985e-06,
|
| 1594 |
+
"loss": 0.131,
|
| 1595 |
+
"step": 6275
|
| 1596 |
+
},
|
| 1597 |
+
{
|
| 1598 |
+
"epoch": 9.08,
|
| 1599 |
+
"learning_rate": 1.032660902977906e-06,
|
| 1600 |
+
"loss": 0.1299,
|
| 1601 |
+
"step": 6300
|
| 1602 |
+
},
|
| 1603 |
+
{
|
| 1604 |
+
"epoch": 9.11,
|
| 1605 |
+
"learning_rate": 9.926352865834136e-07,
|
| 1606 |
+
"loss": 0.151,
|
| 1607 |
+
"step": 6325
|
| 1608 |
+
},
|
| 1609 |
+
{
|
| 1610 |
+
"epoch": 9.15,
|
| 1611 |
+
"learning_rate": 9.526096701889209e-07,
|
| 1612 |
+
"loss": 0.1339,
|
| 1613 |
+
"step": 6350
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 9.19,
|
| 1617 |
+
"learning_rate": 9.125840537944285e-07,
|
| 1618 |
+
"loss": 0.1128,
|
| 1619 |
+
"step": 6375
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 9.22,
|
| 1623 |
+
"learning_rate": 8.72558437399936e-07,
|
| 1624 |
+
"loss": 0.0972,
|
| 1625 |
+
"step": 6400
|
| 1626 |
+
},
|
| 1627 |
+
{
|
| 1628 |
+
"epoch": 9.26,
|
| 1629 |
+
"learning_rate": 8.325328210054436e-07,
|
| 1630 |
+
"loss": 0.0865,
|
| 1631 |
+
"step": 6425
|
| 1632 |
+
},
|
| 1633 |
+
{
|
| 1634 |
+
"epoch": 9.29,
|
| 1635 |
+
"learning_rate": 7.925072046109511e-07,
|
| 1636 |
+
"loss": 0.1107,
|
| 1637 |
+
"step": 6450
|
| 1638 |
+
},
|
| 1639 |
+
{
|
| 1640 |
+
"epoch": 9.33,
|
| 1641 |
+
"learning_rate": 7.524815882164586e-07,
|
| 1642 |
+
"loss": 0.0914,
|
| 1643 |
+
"step": 6475
|
| 1644 |
+
},
|
| 1645 |
+
{
|
| 1646 |
+
"epoch": 9.37,
|
| 1647 |
+
"learning_rate": 7.124559718219662e-07,
|
| 1648 |
+
"loss": 0.1116,
|
| 1649 |
+
"step": 6500
|
| 1650 |
+
},
|
| 1651 |
+
{
|
| 1652 |
+
"epoch": 9.4,
|
| 1653 |
+
"learning_rate": 6.724303554274736e-07,
|
| 1654 |
+
"loss": 0.1026,
|
| 1655 |
+
"step": 6525
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 9.44,
|
| 1659 |
+
"learning_rate": 6.324047390329811e-07,
|
| 1660 |
+
"loss": 0.1231,
|
| 1661 |
+
"step": 6550
|
| 1662 |
+
},
|
| 1663 |
+
{
|
| 1664 |
+
"epoch": 9.47,
|
| 1665 |
+
"learning_rate": 5.923791226384887e-07,
|
| 1666 |
+
"loss": 0.1046,
|
| 1667 |
+
"step": 6575
|
| 1668 |
+
},
|
| 1669 |
+
{
|
| 1670 |
+
"epoch": 9.51,
|
| 1671 |
+
"learning_rate": 5.523535062439962e-07,
|
| 1672 |
+
"loss": 0.1142,
|
| 1673 |
+
"step": 6600
|
| 1674 |
+
},
|
| 1675 |
+
{
|
| 1676 |
+
"epoch": 9.55,
|
| 1677 |
+
"learning_rate": 5.123278898495037e-07,
|
| 1678 |
+
"loss": 0.086,
|
| 1679 |
+
"step": 6625
|
| 1680 |
+
},
|
| 1681 |
+
{
|
| 1682 |
+
"epoch": 9.58,
|
| 1683 |
+
"learning_rate": 4.723022734550112e-07,
|
| 1684 |
+
"loss": 0.0921,
|
| 1685 |
+
"step": 6650
|
| 1686 |
+
},
|
| 1687 |
+
{
|
| 1688 |
+
"epoch": 9.62,
|
| 1689 |
+
"learning_rate": 4.322766570605188e-07,
|
| 1690 |
+
"loss": 0.1157,
|
| 1691 |
+
"step": 6675
|
| 1692 |
+
},
|
| 1693 |
+
{
|
| 1694 |
+
"epoch": 9.65,
|
| 1695 |
+
"learning_rate": 3.9225104066602624e-07,
|
| 1696 |
+
"loss": 0.1245,
|
| 1697 |
+
"step": 6700
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 9.69,
|
| 1701 |
+
"learning_rate": 3.522254242715338e-07,
|
| 1702 |
+
"loss": 0.113,
|
| 1703 |
+
"step": 6725
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 9.73,
|
| 1707 |
+
"learning_rate": 3.121998078770413e-07,
|
| 1708 |
+
"loss": 0.0821,
|
| 1709 |
+
"step": 6750
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 9.76,
|
| 1713 |
+
"learning_rate": 2.7217419148254884e-07,
|
| 1714 |
+
"loss": 0.1227,
|
| 1715 |
+
"step": 6775
|
| 1716 |
+
},
|
| 1717 |
+
{
|
| 1718 |
+
"epoch": 9.8,
|
| 1719 |
+
"learning_rate": 2.3214857508805637e-07,
|
| 1720 |
+
"loss": 0.1208,
|
| 1721 |
+
"step": 6800
|
| 1722 |
+
},
|
| 1723 |
+
{
|
| 1724 |
+
"epoch": 9.83,
|
| 1725 |
+
"learning_rate": 1.921229586935639e-07,
|
| 1726 |
+
"loss": 0.1323,
|
| 1727 |
+
"step": 6825
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 9.87,
|
| 1731 |
+
"learning_rate": 1.520973422990714e-07,
|
| 1732 |
+
"loss": 0.1217,
|
| 1733 |
+
"step": 6850
|
| 1734 |
+
},
|
| 1735 |
+
{
|
| 1736 |
+
"epoch": 9.91,
|
| 1737 |
+
"learning_rate": 1.1207172590457893e-07,
|
| 1738 |
+
"loss": 0.1162,
|
| 1739 |
+
"step": 6875
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 9.94,
|
| 1743 |
+
"learning_rate": 7.204610951008646e-08,
|
| 1744 |
+
"loss": 0.0977,
|
| 1745 |
+
"step": 6900
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 9.98,
|
| 1749 |
+
"learning_rate": 3.2020493115593984e-08,
|
| 1750 |
+
"loss": 0.1527,
|
| 1751 |
+
"step": 6925
|
| 1752 |
+
},
|
| 1753 |
+
{
|
| 1754 |
+
"epoch": 10.0,
|
| 1755 |
+
"eval_accuracy": 0.752547554347826,
|
| 1756 |
+
"eval_loss": 1.0725239515304565,
|
| 1757 |
+
"eval_runtime": 143.4336,
|
| 1758 |
+
"eval_samples_per_second": 41.05,
|
| 1759 |
+
"eval_steps_per_second": 2.566,
|
| 1760 |
+
"step": 6940
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 10.0,
|
| 1764 |
+
"step": 6940,
|
| 1765 |
+
"total_flos": 6.380841226895999e+18,
|
| 1766 |
+
"train_loss": 0.8764074790546461,
|
| 1767 |
+
"train_runtime": 7685.9678,
|
| 1768 |
+
"train_samples_per_second": 28.876,
|
| 1769 |
+
"train_steps_per_second": 0.903
|
| 1770 |
+
}
|
| 1771 |
+
],
|
| 1772 |
+
"max_steps": 6940,
|
| 1773 |
+
"num_train_epochs": 10,
|
| 1774 |
+
"total_flos": 6.380841226895999e+18,
|
| 1775 |
+
"trial_name": null,
|
| 1776 |
+
"trial_params": null
|
| 1777 |
+
}
|