Commit
•
d53d90b
1
Parent(s):
f79b38b
End of training
Browse files- all_results.json +14 -0
- eval_results.json +9 -0
- train_results.json +8 -0
- trainer_state.json +1900 -0
all_results.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.0,
|
| 3 |
+
"eval_loss": 0.14759068191051483,
|
| 4 |
+
"eval_runtime": 1022.7659,
|
| 5 |
+
"eval_samples": 16007,
|
| 6 |
+
"eval_samples_per_second": 15.651,
|
| 7 |
+
"eval_steps_per_second": 1.956,
|
| 8 |
+
"eval_wer": 0.16119714847590955,
|
| 9 |
+
"train_loss": 0.32739020716330625,
|
| 10 |
+
"train_runtime": 49115.8494,
|
| 11 |
+
"train_samples": 436168,
|
| 12 |
+
"train_samples_per_second": 17.761,
|
| 13 |
+
"train_steps_per_second": 0.555
|
| 14 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.0,
|
| 3 |
+
"eval_loss": 0.14759068191051483,
|
| 4 |
+
"eval_runtime": 1022.7659,
|
| 5 |
+
"eval_samples": 16007,
|
| 6 |
+
"eval_samples_per_second": 15.651,
|
| 7 |
+
"eval_steps_per_second": 1.956,
|
| 8 |
+
"eval_wer": 0.16119714847590955
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.0,
|
| 3 |
+
"train_loss": 0.32739020716330625,
|
| 4 |
+
"train_runtime": 49115.8494,
|
| 5 |
+
"train_samples": 436168,
|
| 6 |
+
"train_samples_per_second": 17.761,
|
| 7 |
+
"train_steps_per_second": 0.555
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1900 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.9999816584435355,
|
| 5 |
+
"global_step": 27260,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.01,
|
| 12 |
+
"learning_rate": 3.675e-06,
|
| 13 |
+
"loss": 11.4989,
|
| 14 |
+
"step": 100
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.01,
|
| 18 |
+
"learning_rate": 7.425e-06,
|
| 19 |
+
"loss": 3.2394,
|
| 20 |
+
"step": 200
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.02,
|
| 24 |
+
"learning_rate": 1.1174999999999999e-05,
|
| 25 |
+
"loss": 3.0303,
|
| 26 |
+
"step": 300
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.03,
|
| 30 |
+
"learning_rate": 1.4925e-05,
|
| 31 |
+
"loss": 2.9052,
|
| 32 |
+
"step": 400
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.04,
|
| 36 |
+
"learning_rate": 1.8675e-05,
|
| 37 |
+
"loss": 2.1033,
|
| 38 |
+
"step": 500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.04,
|
| 42 |
+
"learning_rate": 2.2424999999999996e-05,
|
| 43 |
+
"loss": 1.674,
|
| 44 |
+
"step": 600
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.05,
|
| 48 |
+
"learning_rate": 2.6174999999999996e-05,
|
| 49 |
+
"loss": 1.5568,
|
| 50 |
+
"step": 700
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.06,
|
| 54 |
+
"learning_rate": 2.9925e-05,
|
| 55 |
+
"loss": 1.4654,
|
| 56 |
+
"step": 800
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.07,
|
| 60 |
+
"learning_rate": 3.3675e-05,
|
| 61 |
+
"loss": 1.3031,
|
| 62 |
+
"step": 900
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.07,
|
| 66 |
+
"learning_rate": 3.7424999999999995e-05,
|
| 67 |
+
"loss": 1.1842,
|
| 68 |
+
"step": 1000
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.07,
|
| 72 |
+
"eval_loss": 0.44609957933425903,
|
| 73 |
+
"eval_runtime": 1053.3237,
|
| 74 |
+
"eval_samples_per_second": 15.197,
|
| 75 |
+
"eval_steps_per_second": 1.9,
|
| 76 |
+
"eval_wer": 0.49177182344586473,
|
| 77 |
+
"step": 1000
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 0.08,
|
| 81 |
+
"learning_rate": 4.1175e-05,
|
| 82 |
+
"loss": 1.1329,
|
| 83 |
+
"step": 1100
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 0.09,
|
| 87 |
+
"learning_rate": 4.4924999999999994e-05,
|
| 88 |
+
"loss": 1.1316,
|
| 89 |
+
"step": 1200
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 0.1,
|
| 93 |
+
"learning_rate": 4.8675e-05,
|
| 94 |
+
"loss": 1.1092,
|
| 95 |
+
"step": 1300
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 0.1,
|
| 99 |
+
"learning_rate": 5.2424999999999994e-05,
|
| 100 |
+
"loss": 1.1215,
|
| 101 |
+
"step": 1400
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.11,
|
| 105 |
+
"learning_rate": 5.6175e-05,
|
| 106 |
+
"loss": 1.1165,
|
| 107 |
+
"step": 1500
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.12,
|
| 111 |
+
"learning_rate": 5.9925e-05,
|
| 112 |
+
"loss": 1.0946,
|
| 113 |
+
"step": 1600
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 0.12,
|
| 117 |
+
"learning_rate": 6.367499999999999e-05,
|
| 118 |
+
"loss": 1.1189,
|
| 119 |
+
"step": 1700
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 0.13,
|
| 123 |
+
"learning_rate": 6.7425e-05,
|
| 124 |
+
"loss": 1.1175,
|
| 125 |
+
"step": 1800
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 0.14,
|
| 129 |
+
"learning_rate": 7.1175e-05,
|
| 130 |
+
"loss": 1.1254,
|
| 131 |
+
"step": 1900
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 0.15,
|
| 135 |
+
"learning_rate": 7.492499999999999e-05,
|
| 136 |
+
"loss": 1.1317,
|
| 137 |
+
"step": 2000
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 0.15,
|
| 141 |
+
"eval_loss": 0.2668535113334656,
|
| 142 |
+
"eval_runtime": 988.5751,
|
| 143 |
+
"eval_samples_per_second": 16.192,
|
| 144 |
+
"eval_steps_per_second": 2.024,
|
| 145 |
+
"eval_wer": 0.2748006118212608,
|
| 146 |
+
"step": 2000
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.15,
|
| 150 |
+
"learning_rate": 7.470902612826603e-05,
|
| 151 |
+
"loss": 1.1296,
|
| 152 |
+
"step": 2100
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.16,
|
| 156 |
+
"learning_rate": 7.441211401425178e-05,
|
| 157 |
+
"loss": 1.1406,
|
| 158 |
+
"step": 2200
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.17,
|
| 162 |
+
"learning_rate": 7.411520190023751e-05,
|
| 163 |
+
"loss": 1.1362,
|
| 164 |
+
"step": 2300
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.18,
|
| 168 |
+
"learning_rate": 7.381828978622327e-05,
|
| 169 |
+
"loss": 1.1292,
|
| 170 |
+
"step": 2400
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.18,
|
| 174 |
+
"learning_rate": 7.352137767220902e-05,
|
| 175 |
+
"loss": 1.105,
|
| 176 |
+
"step": 2500
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.19,
|
| 180 |
+
"learning_rate": 7.322446555819477e-05,
|
| 181 |
+
"loss": 1.1231,
|
| 182 |
+
"step": 2600
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.2,
|
| 186 |
+
"learning_rate": 7.292755344418051e-05,
|
| 187 |
+
"loss": 1.1187,
|
| 188 |
+
"step": 2700
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.21,
|
| 192 |
+
"learning_rate": 7.263064133016626e-05,
|
| 193 |
+
"loss": 1.1339,
|
| 194 |
+
"step": 2800
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.21,
|
| 198 |
+
"learning_rate": 7.233372921615201e-05,
|
| 199 |
+
"loss": 1.1241,
|
| 200 |
+
"step": 2900
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.22,
|
| 204 |
+
"learning_rate": 7.203681710213777e-05,
|
| 205 |
+
"loss": 1.1029,
|
| 206 |
+
"step": 3000
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.22,
|
| 210 |
+
"eval_loss": 0.2638496458530426,
|
| 211 |
+
"eval_runtime": 987.5568,
|
| 212 |
+
"eval_samples_per_second": 16.209,
|
| 213 |
+
"eval_steps_per_second": 2.026,
|
| 214 |
+
"eval_wer": 0.2705875122910521,
|
| 215 |
+
"step": 3000
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 0.23,
|
| 219 |
+
"learning_rate": 7.173990498812351e-05,
|
| 220 |
+
"loss": 1.1215,
|
| 221 |
+
"step": 3100
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.23,
|
| 225 |
+
"learning_rate": 7.144299287410925e-05,
|
| 226 |
+
"loss": 1.1067,
|
| 227 |
+
"step": 3200
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.24,
|
| 231 |
+
"learning_rate": 7.114608076009501e-05,
|
| 232 |
+
"loss": 1.1126,
|
| 233 |
+
"step": 3300
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.25,
|
| 237 |
+
"learning_rate": 7.084916864608076e-05,
|
| 238 |
+
"loss": 1.109,
|
| 239 |
+
"step": 3400
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 0.26,
|
| 243 |
+
"learning_rate": 7.05522565320665e-05,
|
| 244 |
+
"loss": 1.1077,
|
| 245 |
+
"step": 3500
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 0.26,
|
| 249 |
+
"learning_rate": 7.025534441805225e-05,
|
| 250 |
+
"loss": 1.1,
|
| 251 |
+
"step": 3600
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 0.27,
|
| 255 |
+
"learning_rate": 6.9958432304038e-05,
|
| 256 |
+
"loss": 1.1061,
|
| 257 |
+
"step": 3700
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 0.28,
|
| 261 |
+
"learning_rate": 6.966152019002374e-05,
|
| 262 |
+
"loss": 1.103,
|
| 263 |
+
"step": 3800
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 0.29,
|
| 267 |
+
"learning_rate": 6.936460807600949e-05,
|
| 268 |
+
"loss": 1.0947,
|
| 269 |
+
"step": 3900
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.29,
|
| 273 |
+
"learning_rate": 6.906769596199525e-05,
|
| 274 |
+
"loss": 1.0949,
|
| 275 |
+
"step": 4000
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.29,
|
| 279 |
+
"eval_loss": 0.25188884139060974,
|
| 280 |
+
"eval_runtime": 996.0428,
|
| 281 |
+
"eval_samples_per_second": 16.071,
|
| 282 |
+
"eval_steps_per_second": 2.009,
|
| 283 |
+
"eval_wer": 0.26274172402490986,
|
| 284 |
+
"step": 4000
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 0.3,
|
| 288 |
+
"learning_rate": 6.8770783847981e-05,
|
| 289 |
+
"loss": 1.1076,
|
| 290 |
+
"step": 4100
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.31,
|
| 294 |
+
"learning_rate": 6.847387173396674e-05,
|
| 295 |
+
"loss": 1.1012,
|
| 296 |
+
"step": 4200
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.32,
|
| 300 |
+
"learning_rate": 6.817695961995249e-05,
|
| 301 |
+
"loss": 1.081,
|
| 302 |
+
"step": 4300
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 0.32,
|
| 306 |
+
"learning_rate": 6.788004750593824e-05,
|
| 307 |
+
"loss": 1.0868,
|
| 308 |
+
"step": 4400
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 0.33,
|
| 312 |
+
"learning_rate": 6.758313539192398e-05,
|
| 313 |
+
"loss": 1.0956,
|
| 314 |
+
"step": 4500
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.34,
|
| 318 |
+
"learning_rate": 6.728622327790973e-05,
|
| 319 |
+
"loss": 1.0953,
|
| 320 |
+
"step": 4600
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 0.34,
|
| 324 |
+
"learning_rate": 6.698931116389548e-05,
|
| 325 |
+
"loss": 1.0952,
|
| 326 |
+
"step": 4700
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.35,
|
| 330 |
+
"learning_rate": 6.669239904988122e-05,
|
| 331 |
+
"loss": 1.0968,
|
| 332 |
+
"step": 4800
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.36,
|
| 336 |
+
"learning_rate": 6.639548693586698e-05,
|
| 337 |
+
"loss": 1.0827,
|
| 338 |
+
"step": 4900
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.37,
|
| 342 |
+
"learning_rate": 6.609857482185273e-05,
|
| 343 |
+
"loss": 1.0923,
|
| 344 |
+
"step": 5000
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 0.37,
|
| 348 |
+
"eval_loss": 0.24751192331314087,
|
| 349 |
+
"eval_runtime": 984.8205,
|
| 350 |
+
"eval_samples_per_second": 16.254,
|
| 351 |
+
"eval_steps_per_second": 2.032,
|
| 352 |
+
"eval_wer": 0.25769556429585927,
|
| 353 |
+
"step": 5000
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.37,
|
| 357 |
+
"learning_rate": 6.580166270783846e-05,
|
| 358 |
+
"loss": 1.0895,
|
| 359 |
+
"step": 5100
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.38,
|
| 363 |
+
"learning_rate": 6.550771971496436e-05,
|
| 364 |
+
"loss": 1.0851,
|
| 365 |
+
"step": 5200
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 0.39,
|
| 369 |
+
"learning_rate": 6.521080760095011e-05,
|
| 370 |
+
"loss": 1.1124,
|
| 371 |
+
"step": 5300
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 0.4,
|
| 375 |
+
"learning_rate": 6.491686460807601e-05,
|
| 376 |
+
"loss": 1.0809,
|
| 377 |
+
"step": 5400
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.4,
|
| 381 |
+
"learning_rate": 6.461995249406176e-05,
|
| 382 |
+
"loss": 1.0985,
|
| 383 |
+
"step": 5500
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 0.41,
|
| 387 |
+
"learning_rate": 6.432304038004749e-05,
|
| 388 |
+
"loss": 1.086,
|
| 389 |
+
"step": 5600
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 0.42,
|
| 393 |
+
"learning_rate": 6.402612826603325e-05,
|
| 394 |
+
"loss": 1.0823,
|
| 395 |
+
"step": 5700
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.43,
|
| 399 |
+
"learning_rate": 6.3729216152019e-05,
|
| 400 |
+
"loss": 1.0732,
|
| 401 |
+
"step": 5800
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.43,
|
| 405 |
+
"learning_rate": 6.343230403800475e-05,
|
| 406 |
+
"loss": 1.076,
|
| 407 |
+
"step": 5900
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 0.44,
|
| 411 |
+
"learning_rate": 6.313539192399049e-05,
|
| 412 |
+
"loss": 1.0847,
|
| 413 |
+
"step": 6000
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 0.44,
|
| 417 |
+
"eval_loss": 0.24355509877204895,
|
| 418 |
+
"eval_runtime": 984.3756,
|
| 419 |
+
"eval_samples_per_second": 16.261,
|
| 420 |
+
"eval_steps_per_second": 2.033,
|
| 421 |
+
"eval_wer": 0.26121217087293785,
|
| 422 |
+
"step": 6000
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.45,
|
| 426 |
+
"learning_rate": 6.283847980997624e-05,
|
| 427 |
+
"loss": 1.0748,
|
| 428 |
+
"step": 6100
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 0.45,
|
| 432 |
+
"learning_rate": 6.254156769596199e-05,
|
| 433 |
+
"loss": 1.0836,
|
| 434 |
+
"step": 6200
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 0.46,
|
| 438 |
+
"learning_rate": 6.224465558194773e-05,
|
| 439 |
+
"loss": 1.084,
|
| 440 |
+
"step": 6300
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 0.47,
|
| 444 |
+
"learning_rate": 6.194774346793349e-05,
|
| 445 |
+
"loss": 1.0649,
|
| 446 |
+
"step": 6400
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 0.48,
|
| 450 |
+
"learning_rate": 6.165083135391923e-05,
|
| 451 |
+
"loss": 1.0751,
|
| 452 |
+
"step": 6500
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 0.48,
|
| 456 |
+
"learning_rate": 6.135391923990499e-05,
|
| 457 |
+
"loss": 1.0773,
|
| 458 |
+
"step": 6600
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.49,
|
| 462 |
+
"learning_rate": 6.105700712589073e-05,
|
| 463 |
+
"loss": 1.095,
|
| 464 |
+
"step": 6700
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.5,
|
| 468 |
+
"learning_rate": 6.076009501187648e-05,
|
| 469 |
+
"loss": 1.0629,
|
| 470 |
+
"step": 6800
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 0.51,
|
| 474 |
+
"learning_rate": 6.0463182897862234e-05,
|
| 475 |
+
"loss": 1.0904,
|
| 476 |
+
"step": 6900
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 0.51,
|
| 480 |
+
"learning_rate": 6.0166270783847974e-05,
|
| 481 |
+
"loss": 1.0667,
|
| 482 |
+
"step": 7000
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 0.51,
|
| 486 |
+
"eval_loss": 0.24724909663200378,
|
| 487 |
+
"eval_runtime": 983.1677,
|
| 488 |
+
"eval_samples_per_second": 16.281,
|
| 489 |
+
"eval_steps_per_second": 2.035,
|
| 490 |
+
"eval_wer": 0.26608762154484866,
|
| 491 |
+
"step": 7000
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.52,
|
| 495 |
+
"learning_rate": 5.986935866983372e-05,
|
| 496 |
+
"loss": 1.0825,
|
| 497 |
+
"step": 7100
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 0.53,
|
| 501 |
+
"learning_rate": 5.9572446555819474e-05,
|
| 502 |
+
"loss": 1.0811,
|
| 503 |
+
"step": 7200
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 0.54,
|
| 507 |
+
"learning_rate": 5.927553444180522e-05,
|
| 508 |
+
"loss": 1.0906,
|
| 509 |
+
"step": 7300
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 0.54,
|
| 513 |
+
"learning_rate": 5.8978622327790975e-05,
|
| 514 |
+
"loss": 1.0784,
|
| 515 |
+
"step": 7400
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 0.55,
|
| 519 |
+
"learning_rate": 5.8681710213776715e-05,
|
| 520 |
+
"loss": 1.0822,
|
| 521 |
+
"step": 7500
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.56,
|
| 525 |
+
"learning_rate": 5.838479809976246e-05,
|
| 526 |
+
"loss": 1.0802,
|
| 527 |
+
"step": 7600
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.56,
|
| 531 |
+
"learning_rate": 5.8087885985748215e-05,
|
| 532 |
+
"loss": 1.0805,
|
| 533 |
+
"step": 7700
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 0.57,
|
| 537 |
+
"learning_rate": 5.779097387173396e-05,
|
| 538 |
+
"loss": 1.093,
|
| 539 |
+
"step": 7800
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 0.58,
|
| 543 |
+
"learning_rate": 5.749406175771971e-05,
|
| 544 |
+
"loss": 1.0456,
|
| 545 |
+
"step": 7900
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 0.59,
|
| 549 |
+
"learning_rate": 5.7197149643705455e-05,
|
| 550 |
+
"loss": 1.0709,
|
| 551 |
+
"step": 8000
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 0.59,
|
| 555 |
+
"eval_loss": 0.24887976050376892,
|
| 556 |
+
"eval_runtime": 982.4054,
|
| 557 |
+
"eval_samples_per_second": 16.294,
|
| 558 |
+
"eval_steps_per_second": 2.037,
|
| 559 |
+
"eval_wer": 0.26095269310608543,
|
| 560 |
+
"step": 8000
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 0.59,
|
| 564 |
+
"learning_rate": 5.690023752969121e-05,
|
| 565 |
+
"loss": 1.0677,
|
| 566 |
+
"step": 8100
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 0.6,
|
| 570 |
+
"learning_rate": 5.6603325415676956e-05,
|
| 571 |
+
"loss": 1.0659,
|
| 572 |
+
"step": 8200
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 0.61,
|
| 576 |
+
"learning_rate": 5.630641330166271e-05,
|
| 577 |
+
"loss": 1.0788,
|
| 578 |
+
"step": 8300
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 0.62,
|
| 582 |
+
"learning_rate": 5.600950118764845e-05,
|
| 583 |
+
"loss": 1.071,
|
| 584 |
+
"step": 8400
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.62,
|
| 588 |
+
"learning_rate": 5.5712589073634196e-05,
|
| 589 |
+
"loss": 1.0669,
|
| 590 |
+
"step": 8500
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.63,
|
| 594 |
+
"learning_rate": 5.541567695961995e-05,
|
| 595 |
+
"loss": 1.0728,
|
| 596 |
+
"step": 8600
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 0.64,
|
| 600 |
+
"learning_rate": 5.512173396674584e-05,
|
| 601 |
+
"loss": 1.0652,
|
| 602 |
+
"step": 8700
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 0.65,
|
| 606 |
+
"learning_rate": 5.482482185273159e-05,
|
| 607 |
+
"loss": 1.0632,
|
| 608 |
+
"step": 8800
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 0.65,
|
| 612 |
+
"learning_rate": 5.452790973871733e-05,
|
| 613 |
+
"loss": 1.048,
|
| 614 |
+
"step": 8900
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 0.66,
|
| 618 |
+
"learning_rate": 5.4230997624703083e-05,
|
| 619 |
+
"loss": 1.0472,
|
| 620 |
+
"step": 9000
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 0.66,
|
| 624 |
+
"eval_loss": 0.23543120920658112,
|
| 625 |
+
"eval_runtime": 985.8901,
|
| 626 |
+
"eval_samples_per_second": 16.236,
|
| 627 |
+
"eval_steps_per_second": 2.03,
|
| 628 |
+
"eval_wer": 0.24997268655085764,
|
| 629 |
+
"step": 9000
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 0.67,
|
| 633 |
+
"learning_rate": 5.393408551068883e-05,
|
| 634 |
+
"loss": 1.0552,
|
| 635 |
+
"step": 9100
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 0.67,
|
| 639 |
+
"learning_rate": 5.3637173396674584e-05,
|
| 640 |
+
"loss": 1.0581,
|
| 641 |
+
"step": 9200
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 0.68,
|
| 645 |
+
"learning_rate": 5.3340261282660324e-05,
|
| 646 |
+
"loss": 1.0658,
|
| 647 |
+
"step": 9300
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.69,
|
| 651 |
+
"learning_rate": 5.304334916864607e-05,
|
| 652 |
+
"loss": 1.0603,
|
| 653 |
+
"step": 9400
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.7,
|
| 657 |
+
"learning_rate": 5.2746437054631824e-05,
|
| 658 |
+
"loss": 1.0661,
|
| 659 |
+
"step": 9500
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 0.7,
|
| 663 |
+
"learning_rate": 5.244952494061757e-05,
|
| 664 |
+
"loss": 1.0554,
|
| 665 |
+
"step": 9600
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 0.71,
|
| 669 |
+
"learning_rate": 5.2152612826603325e-05,
|
| 670 |
+
"loss": 1.0728,
|
| 671 |
+
"step": 9700
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 0.72,
|
| 675 |
+
"learning_rate": 5.1855700712589065e-05,
|
| 676 |
+
"loss": 1.0513,
|
| 677 |
+
"step": 9800
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 0.73,
|
| 681 |
+
"learning_rate": 5.155878859857482e-05,
|
| 682 |
+
"loss": 1.0379,
|
| 683 |
+
"step": 9900
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 0.73,
|
| 687 |
+
"learning_rate": 5.1261876484560565e-05,
|
| 688 |
+
"loss": 1.0604,
|
| 689 |
+
"step": 10000
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.73,
|
| 693 |
+
"eval_loss": 0.23458585143089294,
|
| 694 |
+
"eval_runtime": 986.1525,
|
| 695 |
+
"eval_samples_per_second": 16.232,
|
| 696 |
+
"eval_steps_per_second": 2.029,
|
| 697 |
+
"eval_wer": 0.2485182453840271,
|
| 698 |
+
"step": 10000
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 0.74,
|
| 702 |
+
"learning_rate": 5.096496437054632e-05,
|
| 703 |
+
"loss": 1.0632,
|
| 704 |
+
"step": 10100
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 0.75,
|
| 708 |
+
"learning_rate": 5.0668052256532065e-05,
|
| 709 |
+
"loss": 1.0526,
|
| 710 |
+
"step": 10200
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.76,
|
| 714 |
+
"learning_rate": 5.0371140142517805e-05,
|
| 715 |
+
"loss": 1.0314,
|
| 716 |
+
"step": 10300
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.76,
|
| 720 |
+
"learning_rate": 5.007422802850356e-05,
|
| 721 |
+
"loss": 1.0508,
|
| 722 |
+
"step": 10400
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 0.77,
|
| 726 |
+
"learning_rate": 4.9777315914489306e-05,
|
| 727 |
+
"loss": 1.0446,
|
| 728 |
+
"step": 10500
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 0.78,
|
| 732 |
+
"learning_rate": 4.948040380047506e-05,
|
| 733 |
+
"loss": 1.0361,
|
| 734 |
+
"step": 10600
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 0.79,
|
| 738 |
+
"learning_rate": 4.91834916864608e-05,
|
| 739 |
+
"loss": 1.0319,
|
| 740 |
+
"step": 10700
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 0.79,
|
| 744 |
+
"learning_rate": 4.8886579572446546e-05,
|
| 745 |
+
"loss": 1.0178,
|
| 746 |
+
"step": 10800
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 0.8,
|
| 750 |
+
"learning_rate": 4.85896674584323e-05,
|
| 751 |
+
"loss": 1.0301,
|
| 752 |
+
"step": 10900
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.81,
|
| 756 |
+
"learning_rate": 4.8292755344418046e-05,
|
| 757 |
+
"loss": 1.0375,
|
| 758 |
+
"step": 11000
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.81,
|
| 762 |
+
"eval_loss": 0.2285824865102768,
|
| 763 |
+
"eval_runtime": 979.8277,
|
| 764 |
+
"eval_samples_per_second": 16.337,
|
| 765 |
+
"eval_steps_per_second": 2.042,
|
| 766 |
+
"eval_wer": 0.23898585163334427,
|
| 767 |
+
"step": 11000
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 0.81,
|
| 771 |
+
"learning_rate": 4.79958432304038e-05,
|
| 772 |
+
"loss": 1.0398,
|
| 773 |
+
"step": 11100
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 0.82,
|
| 777 |
+
"learning_rate": 4.769893111638954e-05,
|
| 778 |
+
"loss": 1.0308,
|
| 779 |
+
"step": 11200
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 0.83,
|
| 783 |
+
"learning_rate": 4.7402019002375294e-05,
|
| 784 |
+
"loss": 1.0309,
|
| 785 |
+
"step": 11300
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 0.84,
|
| 789 |
+
"learning_rate": 4.710510688836104e-05,
|
| 790 |
+
"loss": 1.0287,
|
| 791 |
+
"step": 11400
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 0.84,
|
| 795 |
+
"learning_rate": 4.6808194774346794e-05,
|
| 796 |
+
"loss": 1.0195,
|
| 797 |
+
"step": 11500
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 0.85,
|
| 801 |
+
"learning_rate": 4.651128266033254e-05,
|
| 802 |
+
"loss": 1.0292,
|
| 803 |
+
"step": 11600
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 0.86,
|
| 807 |
+
"learning_rate": 4.621437054631828e-05,
|
| 808 |
+
"loss": 1.0147,
|
| 809 |
+
"step": 11700
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 0.87,
|
| 813 |
+
"learning_rate": 4.5917458432304034e-05,
|
| 814 |
+
"loss": 1.0242,
|
| 815 |
+
"step": 11800
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 0.87,
|
| 819 |
+
"learning_rate": 4.562054631828978e-05,
|
| 820 |
+
"loss": 1.029,
|
| 821 |
+
"step": 11900
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 0.88,
|
| 825 |
+
"learning_rate": 4.5326603325415675e-05,
|
| 826 |
+
"loss": 1.0193,
|
| 827 |
+
"step": 12000
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 0.88,
|
| 831 |
+
"eval_loss": 0.22122837603092194,
|
| 832 |
+
"eval_runtime": 981.4673,
|
| 833 |
+
"eval_samples_per_second": 16.309,
|
| 834 |
+
"eval_steps_per_second": 2.039,
|
| 835 |
+
"eval_wer": 0.23376215448486834,
|
| 836 |
+
"step": 12000
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 0.89,
|
| 840 |
+
"learning_rate": 4.502969121140143e-05,
|
| 841 |
+
"loss": 1.0249,
|
| 842 |
+
"step": 12100
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.9,
|
| 846 |
+
"learning_rate": 4.473277909738717e-05,
|
| 847 |
+
"loss": 1.0165,
|
| 848 |
+
"step": 12200
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 0.9,
|
| 852 |
+
"learning_rate": 4.4435866983372915e-05,
|
| 853 |
+
"loss": 1.0303,
|
| 854 |
+
"step": 12300
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 0.91,
|
| 858 |
+
"learning_rate": 4.413895486935867e-05,
|
| 859 |
+
"loss": 1.0295,
|
| 860 |
+
"step": 12400
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 0.92,
|
| 864 |
+
"learning_rate": 4.3842042755344415e-05,
|
| 865 |
+
"loss": 1.0112,
|
| 866 |
+
"step": 12500
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 0.92,
|
| 870 |
+
"learning_rate": 4.35480997624703e-05,
|
| 871 |
+
"loss": 1.0056,
|
| 872 |
+
"step": 12600
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 0.93,
|
| 876 |
+
"learning_rate": 4.325118764845605e-05,
|
| 877 |
+
"loss": 1.0108,
|
| 878 |
+
"step": 12700
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 0.94,
|
| 882 |
+
"learning_rate": 4.29542755344418e-05,
|
| 883 |
+
"loss": 1.0133,
|
| 884 |
+
"step": 12800
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.95,
|
| 888 |
+
"learning_rate": 4.265736342042755e-05,
|
| 889 |
+
"loss": 1.0063,
|
| 890 |
+
"step": 12900
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 0.95,
|
| 894 |
+
"learning_rate": 4.23604513064133e-05,
|
| 895 |
+
"loss": 1.0077,
|
| 896 |
+
"step": 13000
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 0.95,
|
| 900 |
+
"eval_loss": 0.21520280838012695,
|
| 901 |
+
"eval_runtime": 983.9086,
|
| 902 |
+
"eval_samples_per_second": 16.269,
|
| 903 |
+
"eval_steps_per_second": 2.034,
|
| 904 |
+
"eval_wer": 0.22689282202556538,
|
| 905 |
+
"step": 13000
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 0.96,
|
| 909 |
+
"learning_rate": 4.206353919239904e-05,
|
| 910 |
+
"loss": 1.0085,
|
| 911 |
+
"step": 13100
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 0.97,
|
| 915 |
+
"learning_rate": 4.176662707838479e-05,
|
| 916 |
+
"loss": 1.011,
|
| 917 |
+
"step": 13200
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 0.98,
|
| 921 |
+
"learning_rate": 4.146971496437054e-05,
|
| 922 |
+
"loss": 1.0131,
|
| 923 |
+
"step": 13300
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 0.98,
|
| 927 |
+
"learning_rate": 4.117280285035629e-05,
|
| 928 |
+
"loss": 0.998,
|
| 929 |
+
"step": 13400
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 0.99,
|
| 933 |
+
"learning_rate": 4.0875890736342043e-05,
|
| 934 |
+
"loss": 1.0002,
|
| 935 |
+
"step": 13500
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 1.0,
|
| 939 |
+
"learning_rate": 4.0578978622327783e-05,
|
| 940 |
+
"loss": 0.9916,
|
| 941 |
+
"step": 13600
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 1.01,
|
| 945 |
+
"learning_rate": 4.028206650831354e-05,
|
| 946 |
+
"loss": 0.9662,
|
| 947 |
+
"step": 13700
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 1.01,
|
| 951 |
+
"learning_rate": 3.9985154394299284e-05,
|
| 952 |
+
"loss": 0.9758,
|
| 953 |
+
"step": 13800
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 1.02,
|
| 957 |
+
"learning_rate": 3.968824228028504e-05,
|
| 958 |
+
"loss": 1.013,
|
| 959 |
+
"step": 13900
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 1.03,
|
| 963 |
+
"learning_rate": 3.939133016627078e-05,
|
| 964 |
+
"loss": 1.0004,
|
| 965 |
+
"step": 14000
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 1.03,
|
| 969 |
+
"eval_loss": 0.2093251347541809,
|
| 970 |
+
"eval_runtime": 986.9604,
|
| 971 |
+
"eval_samples_per_second": 16.218,
|
| 972 |
+
"eval_steps_per_second": 2.027,
|
| 973 |
+
"eval_wer": 0.22069949743253578,
|
| 974 |
+
"step": 14000
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 1.03,
|
| 978 |
+
"learning_rate": 3.9094418052256524e-05,
|
| 979 |
+
"loss": 0.9852,
|
| 980 |
+
"step": 14100
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 1.04,
|
| 984 |
+
"learning_rate": 3.879750593824228e-05,
|
| 985 |
+
"loss": 0.9765,
|
| 986 |
+
"step": 14200
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 1.05,
|
| 990 |
+
"learning_rate": 3.8500593824228025e-05,
|
| 991 |
+
"loss": 0.9978,
|
| 992 |
+
"step": 14300
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 1.06,
|
| 996 |
+
"learning_rate": 3.820368171021378e-05,
|
| 997 |
+
"loss": 0.9807,
|
| 998 |
+
"step": 14400
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 1.06,
|
| 1002 |
+
"learning_rate": 3.790676959619952e-05,
|
| 1003 |
+
"loss": 0.9988,
|
| 1004 |
+
"step": 14500
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 1.07,
|
| 1008 |
+
"learning_rate": 3.7609857482185265e-05,
|
| 1009 |
+
"loss": 0.977,
|
| 1010 |
+
"step": 14600
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 1.08,
|
| 1014 |
+
"learning_rate": 3.731294536817102e-05,
|
| 1015 |
+
"loss": 0.9735,
|
| 1016 |
+
"step": 14700
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 1.09,
|
| 1020 |
+
"learning_rate": 3.7016033254156765e-05,
|
| 1021 |
+
"loss": 0.9767,
|
| 1022 |
+
"step": 14800
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 1.09,
|
| 1026 |
+
"learning_rate": 3.671912114014251e-05,
|
| 1027 |
+
"loss": 0.9555,
|
| 1028 |
+
"step": 14900
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 1.1,
|
| 1032 |
+
"learning_rate": 3.6422209026128266e-05,
|
| 1033 |
+
"loss": 0.9649,
|
| 1034 |
+
"step": 15000
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 1.1,
|
| 1038 |
+
"eval_loss": 0.19932541251182556,
|
| 1039 |
+
"eval_runtime": 986.5773,
|
| 1040 |
+
"eval_samples_per_second": 16.225,
|
| 1041 |
+
"eval_steps_per_second": 2.028,
|
| 1042 |
+
"eval_wer": 0.21130367092756475,
|
| 1043 |
+
"step": 15000
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 1.11,
|
| 1047 |
+
"learning_rate": 3.612529691211401e-05,
|
| 1048 |
+
"loss": 0.9608,
|
| 1049 |
+
"step": 15100
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 1.12,
|
| 1053 |
+
"learning_rate": 3.582838479809976e-05,
|
| 1054 |
+
"loss": 0.9549,
|
| 1055 |
+
"step": 15200
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 1.12,
|
| 1059 |
+
"learning_rate": 3.5531472684085506e-05,
|
| 1060 |
+
"loss": 0.9636,
|
| 1061 |
+
"step": 15300
|
| 1062 |
+
},
|
| 1063 |
+
{
|
| 1064 |
+
"epoch": 1.13,
|
| 1065 |
+
"learning_rate": 3.523456057007125e-05,
|
| 1066 |
+
"loss": 0.9605,
|
| 1067 |
+
"step": 15400
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 1.14,
|
| 1071 |
+
"learning_rate": 3.4937648456057006e-05,
|
| 1072 |
+
"loss": 0.962,
|
| 1073 |
+
"step": 15500
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 1.14,
|
| 1077 |
+
"learning_rate": 3.464073634204275e-05,
|
| 1078 |
+
"loss": 0.9565,
|
| 1079 |
+
"step": 15600
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 1.15,
|
| 1083 |
+
"learning_rate": 3.43438242280285e-05,
|
| 1084 |
+
"loss": 0.9609,
|
| 1085 |
+
"step": 15700
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 1.16,
|
| 1089 |
+
"learning_rate": 3.404691211401425e-05,
|
| 1090 |
+
"loss": 0.9552,
|
| 1091 |
+
"step": 15800
|
| 1092 |
+
},
|
| 1093 |
+
{
|
| 1094 |
+
"epoch": 1.17,
|
| 1095 |
+
"learning_rate": 3.375e-05,
|
| 1096 |
+
"loss": 0.9503,
|
| 1097 |
+
"step": 15900
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 1.17,
|
| 1101 |
+
"learning_rate": 3.345308788598574e-05,
|
| 1102 |
+
"loss": 0.9509,
|
| 1103 |
+
"step": 16000
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"epoch": 1.17,
|
| 1107 |
+
"eval_loss": 0.19342663884162903,
|
| 1108 |
+
"eval_runtime": 984.1094,
|
| 1109 |
+
"eval_samples_per_second": 16.265,
|
| 1110 |
+
"eval_steps_per_second": 2.033,
|
| 1111 |
+
"eval_wer": 0.20888643067846607,
|
| 1112 |
+
"step": 16000
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 1.18,
|
| 1116 |
+
"learning_rate": 3.3156175771971494e-05,
|
| 1117 |
+
"loss": 0.9369,
|
| 1118 |
+
"step": 16100
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 1.19,
|
| 1122 |
+
"learning_rate": 3.285926365795724e-05,
|
| 1123 |
+
"loss": 0.9549,
|
| 1124 |
+
"step": 16200
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 1.2,
|
| 1128 |
+
"learning_rate": 3.256235154394299e-05,
|
| 1129 |
+
"loss": 0.9503,
|
| 1130 |
+
"step": 16300
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 1.2,
|
| 1134 |
+
"learning_rate": 3.226543942992874e-05,
|
| 1135 |
+
"loss": 0.9553,
|
| 1136 |
+
"step": 16400
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 1.21,
|
| 1140 |
+
"learning_rate": 3.196852731591449e-05,
|
| 1141 |
+
"loss": 0.9508,
|
| 1142 |
+
"step": 16500
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 1.22,
|
| 1146 |
+
"learning_rate": 3.1671615201900235e-05,
|
| 1147 |
+
"loss": 0.9411,
|
| 1148 |
+
"step": 16600
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 1.23,
|
| 1152 |
+
"learning_rate": 3.137470308788598e-05,
|
| 1153 |
+
"loss": 0.9435,
|
| 1154 |
+
"step": 16700
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 1.23,
|
| 1158 |
+
"learning_rate": 3.107779097387173e-05,
|
| 1159 |
+
"loss": 0.9439,
|
| 1160 |
+
"step": 16800
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 1.24,
|
| 1164 |
+
"learning_rate": 3.078087885985748e-05,
|
| 1165 |
+
"loss": 0.946,
|
| 1166 |
+
"step": 16900
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 1.25,
|
| 1170 |
+
"learning_rate": 3.048396674584323e-05,
|
| 1171 |
+
"loss": 0.9533,
|
| 1172 |
+
"step": 17000
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 1.25,
|
| 1176 |
+
"eval_loss": 0.18736572563648224,
|
| 1177 |
+
"eval_runtime": 984.7341,
|
| 1178 |
+
"eval_samples_per_second": 16.255,
|
| 1179 |
+
"eval_steps_per_second": 2.032,
|
| 1180 |
+
"eval_wer": 0.20231071779744347,
|
| 1181 |
+
"step": 17000
|
| 1182 |
+
},
|
| 1183 |
+
{
|
| 1184 |
+
"epoch": 1.25,
|
| 1185 |
+
"learning_rate": 3.018705463182898e-05,
|
| 1186 |
+
"loss": 0.9322,
|
| 1187 |
+
"step": 17100
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 1.26,
|
| 1191 |
+
"learning_rate": 2.9890142517814722e-05,
|
| 1192 |
+
"loss": 0.94,
|
| 1193 |
+
"step": 17200
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 1.27,
|
| 1197 |
+
"learning_rate": 2.9593230403800473e-05,
|
| 1198 |
+
"loss": 0.9373,
|
| 1199 |
+
"step": 17300
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 1.28,
|
| 1203 |
+
"learning_rate": 2.9299287410926363e-05,
|
| 1204 |
+
"loss": 0.924,
|
| 1205 |
+
"step": 17400
|
| 1206 |
+
},
|
| 1207 |
+
{
|
| 1208 |
+
"epoch": 1.28,
|
| 1209 |
+
"learning_rate": 2.9005344418052253e-05,
|
| 1210 |
+
"loss": 0.9357,
|
| 1211 |
+
"step": 17500
|
| 1212 |
+
},
|
| 1213 |
+
{
|
| 1214 |
+
"epoch": 1.29,
|
| 1215 |
+
"learning_rate": 2.8708432304038003e-05,
|
| 1216 |
+
"loss": 0.9351,
|
| 1217 |
+
"step": 17600
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 1.3,
|
| 1221 |
+
"learning_rate": 2.841152019002375e-05,
|
| 1222 |
+
"loss": 0.9371,
|
| 1223 |
+
"step": 17700
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 1.31,
|
| 1227 |
+
"learning_rate": 2.81146080760095e-05,
|
| 1228 |
+
"loss": 0.9253,
|
| 1229 |
+
"step": 17800
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 1.31,
|
| 1233 |
+
"learning_rate": 2.7817695961995246e-05,
|
| 1234 |
+
"loss": 0.9264,
|
| 1235 |
+
"step": 17900
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 1.32,
|
| 1239 |
+
"learning_rate": 2.7520783847980997e-05,
|
| 1240 |
+
"loss": 0.9248,
|
| 1241 |
+
"step": 18000
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 1.32,
|
| 1245 |
+
"eval_loss": 0.1818237155675888,
|
| 1246 |
+
"eval_runtime": 1114.2718,
|
| 1247 |
+
"eval_samples_per_second": 14.365,
|
| 1248 |
+
"eval_steps_per_second": 1.796,
|
| 1249 |
+
"eval_wer": 0.19742843876324703,
|
| 1250 |
+
"step": 18000
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 1.33,
|
| 1254 |
+
"learning_rate": 2.722387173396674e-05,
|
| 1255 |
+
"loss": 0.9448,
|
| 1256 |
+
"step": 18100
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 1.34,
|
| 1260 |
+
"learning_rate": 2.692695961995249e-05,
|
| 1261 |
+
"loss": 0.9284,
|
| 1262 |
+
"step": 18200
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 1.34,
|
| 1266 |
+
"learning_rate": 2.663004750593824e-05,
|
| 1267 |
+
"loss": 0.9141,
|
| 1268 |
+
"step": 18300
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 1.35,
|
| 1272 |
+
"learning_rate": 2.6333135391923987e-05,
|
| 1273 |
+
"loss": 0.9117,
|
| 1274 |
+
"step": 18400
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 1.36,
|
| 1278 |
+
"learning_rate": 2.6036223277909737e-05,
|
| 1279 |
+
"loss": 0.917,
|
| 1280 |
+
"step": 18500
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 1.36,
|
| 1284 |
+
"learning_rate": 2.5739311163895484e-05,
|
| 1285 |
+
"loss": 0.9165,
|
| 1286 |
+
"step": 18600
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 1.37,
|
| 1290 |
+
"learning_rate": 2.5442399049881234e-05,
|
| 1291 |
+
"loss": 0.9099,
|
| 1292 |
+
"step": 18700
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 1.38,
|
| 1296 |
+
"learning_rate": 2.5145486935866978e-05,
|
| 1297 |
+
"loss": 0.9022,
|
| 1298 |
+
"step": 18800
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 1.39,
|
| 1302 |
+
"learning_rate": 2.4848574821852728e-05,
|
| 1303 |
+
"loss": 0.9246,
|
| 1304 |
+
"step": 18900
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 1.39,
|
| 1308 |
+
"learning_rate": 2.4551662707838478e-05,
|
| 1309 |
+
"loss": 0.9216,
|
| 1310 |
+
"step": 19000
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 1.39,
|
| 1314 |
+
"eval_loss": 0.17756715416908264,
|
| 1315 |
+
"eval_runtime": 1032.2412,
|
| 1316 |
+
"eval_samples_per_second": 15.507,
|
| 1317 |
+
"eval_steps_per_second": 1.939,
|
| 1318 |
+
"eval_wer": 0.19256664481590735,
|
| 1319 |
+
"step": 19000
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 1.4,
|
| 1323 |
+
"learning_rate": 2.4254750593824225e-05,
|
| 1324 |
+
"loss": 0.9142,
|
| 1325 |
+
"step": 19100
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 1.41,
|
| 1329 |
+
"learning_rate": 2.3957838479809975e-05,
|
| 1330 |
+
"loss": 0.9275,
|
| 1331 |
+
"step": 19200
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 1.42,
|
| 1335 |
+
"learning_rate": 2.3660926365795722e-05,
|
| 1336 |
+
"loss": 0.9132,
|
| 1337 |
+
"step": 19300
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 1.42,
|
| 1341 |
+
"learning_rate": 2.3364014251781472e-05,
|
| 1342 |
+
"loss": 0.9111,
|
| 1343 |
+
"step": 19400
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 1.43,
|
| 1347 |
+
"learning_rate": 2.3067102137767216e-05,
|
| 1348 |
+
"loss": 0.8974,
|
| 1349 |
+
"step": 19500
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 1.44,
|
| 1353 |
+
"learning_rate": 2.2770190023752966e-05,
|
| 1354 |
+
"loss": 0.9013,
|
| 1355 |
+
"step": 19600
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 1.45,
|
| 1359 |
+
"learning_rate": 2.2473277909738716e-05,
|
| 1360 |
+
"loss": 0.9093,
|
| 1361 |
+
"step": 19700
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 1.45,
|
| 1365 |
+
"learning_rate": 2.2176365795724463e-05,
|
| 1366 |
+
"loss": 0.8926,
|
| 1367 |
+
"step": 19800
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 1.46,
|
| 1371 |
+
"learning_rate": 2.1879453681710213e-05,
|
| 1372 |
+
"loss": 0.9026,
|
| 1373 |
+
"step": 19900
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 1.47,
|
| 1377 |
+
"learning_rate": 2.158254156769596e-05,
|
| 1378 |
+
"loss": 0.8964,
|
| 1379 |
+
"step": 20000
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 1.47,
|
| 1383 |
+
"eval_loss": 0.1722368746995926,
|
| 1384 |
+
"eval_runtime": 1019.2936,
|
| 1385 |
+
"eval_samples_per_second": 15.704,
|
| 1386 |
+
"eval_steps_per_second": 1.963,
|
| 1387 |
+
"eval_wer": 0.19043619578280346,
|
| 1388 |
+
"step": 20000
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 1.47,
|
| 1392 |
+
"learning_rate": 2.128859857482185e-05,
|
| 1393 |
+
"loss": 0.8906,
|
| 1394 |
+
"step": 20100
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 1.48,
|
| 1398 |
+
"learning_rate": 2.09916864608076e-05,
|
| 1399 |
+
"loss": 0.8878,
|
| 1400 |
+
"step": 20200
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 1.49,
|
| 1404 |
+
"learning_rate": 2.0694774346793347e-05,
|
| 1405 |
+
"loss": 0.9024,
|
| 1406 |
+
"step": 20300
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 1.5,
|
| 1410 |
+
"learning_rate": 2.0397862232779097e-05,
|
| 1411 |
+
"loss": 0.8903,
|
| 1412 |
+
"step": 20400
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 1.5,
|
| 1416 |
+
"learning_rate": 2.0100950118764844e-05,
|
| 1417 |
+
"loss": 0.8843,
|
| 1418 |
+
"step": 20500
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 1.51,
|
| 1422 |
+
"learning_rate": 1.9804038004750594e-05,
|
| 1423 |
+
"loss": 0.8911,
|
| 1424 |
+
"step": 20600
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 1.52,
|
| 1428 |
+
"learning_rate": 1.9507125890736337e-05,
|
| 1429 |
+
"loss": 0.8795,
|
| 1430 |
+
"step": 20700
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 1.53,
|
| 1434 |
+
"learning_rate": 1.9210213776722087e-05,
|
| 1435 |
+
"loss": 0.8777,
|
| 1436 |
+
"step": 20800
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 1.53,
|
| 1440 |
+
"learning_rate": 1.8913301662707838e-05,
|
| 1441 |
+
"loss": 0.889,
|
| 1442 |
+
"step": 20900
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 1.54,
|
| 1446 |
+
"learning_rate": 1.8616389548693584e-05,
|
| 1447 |
+
"loss": 0.8941,
|
| 1448 |
+
"step": 21000
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 1.54,
|
| 1452 |
+
"eval_loss": 0.16895848512649536,
|
| 1453 |
+
"eval_runtime": 1022.9987,
|
| 1454 |
+
"eval_samples_per_second": 15.647,
|
| 1455 |
+
"eval_steps_per_second": 1.956,
|
| 1456 |
+
"eval_wer": 0.18521932699661314,
|
| 1457 |
+
"step": 21000
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 1.55,
|
| 1461 |
+
"learning_rate": 1.831947743467933e-05,
|
| 1462 |
+
"loss": 0.882,
|
| 1463 |
+
"step": 21100
|
| 1464 |
+
},
|
| 1465 |
+
{
|
| 1466 |
+
"epoch": 1.56,
|
| 1467 |
+
"learning_rate": 1.802256532066508e-05,
|
| 1468 |
+
"loss": 0.8801,
|
| 1469 |
+
"step": 21200
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 1.56,
|
| 1473 |
+
"learning_rate": 1.772565320665083e-05,
|
| 1474 |
+
"loss": 0.8718,
|
| 1475 |
+
"step": 21300
|
| 1476 |
+
},
|
| 1477 |
+
{
|
| 1478 |
+
"epoch": 1.57,
|
| 1479 |
+
"learning_rate": 1.742874109263658e-05,
|
| 1480 |
+
"loss": 0.8904,
|
| 1481 |
+
"step": 21400
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 1.58,
|
| 1485 |
+
"learning_rate": 1.7131828978622325e-05,
|
| 1486 |
+
"loss": 0.8729,
|
| 1487 |
+
"step": 21500
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 1.58,
|
| 1491 |
+
"learning_rate": 1.6834916864608075e-05,
|
| 1492 |
+
"loss": 0.8722,
|
| 1493 |
+
"step": 21600
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 1.59,
|
| 1497 |
+
"learning_rate": 1.6538004750593822e-05,
|
| 1498 |
+
"loss": 0.8739,
|
| 1499 |
+
"step": 21700
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 1.6,
|
| 1503 |
+
"learning_rate": 1.624109263657957e-05,
|
| 1504 |
+
"loss": 0.8635,
|
| 1505 |
+
"step": 21800
|
| 1506 |
+
},
|
| 1507 |
+
{
|
| 1508 |
+
"epoch": 1.61,
|
| 1509 |
+
"learning_rate": 1.594418052256532e-05,
|
| 1510 |
+
"loss": 0.8767,
|
| 1511 |
+
"step": 21900
|
| 1512 |
+
},
|
| 1513 |
+
{
|
| 1514 |
+
"epoch": 1.61,
|
| 1515 |
+
"learning_rate": 1.564726840855107e-05,
|
| 1516 |
+
"loss": 0.871,
|
| 1517 |
+
"step": 22000
|
| 1518 |
+
},
|
| 1519 |
+
{
|
| 1520 |
+
"epoch": 1.61,
|
| 1521 |
+
"eval_loss": 0.16269078850746155,
|
| 1522 |
+
"eval_runtime": 1042.6643,
|
| 1523 |
+
"eval_samples_per_second": 15.352,
|
| 1524 |
+
"eval_steps_per_second": 1.919,
|
| 1525 |
+
"eval_wer": 0.17805637495902982,
|
| 1526 |
+
"step": 22000
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 1.62,
|
| 1530 |
+
"learning_rate": 1.5350356294536816e-05,
|
| 1531 |
+
"loss": 0.8663,
|
| 1532 |
+
"step": 22100
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 1.63,
|
| 1536 |
+
"learning_rate": 1.5056413301662706e-05,
|
| 1537 |
+
"loss": 0.8732,
|
| 1538 |
+
"step": 22200
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 1.64,
|
| 1542 |
+
"learning_rate": 1.4759501187648455e-05,
|
| 1543 |
+
"loss": 0.8625,
|
| 1544 |
+
"step": 22300
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 1.64,
|
| 1548 |
+
"learning_rate": 1.4462589073634203e-05,
|
| 1549 |
+
"loss": 0.854,
|
| 1550 |
+
"step": 22400
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 1.65,
|
| 1554 |
+
"learning_rate": 1.416567695961995e-05,
|
| 1555 |
+
"loss": 0.8692,
|
| 1556 |
+
"step": 22500
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 1.66,
|
| 1560 |
+
"learning_rate": 1.38687648456057e-05,
|
| 1561 |
+
"loss": 0.8477,
|
| 1562 |
+
"step": 22600
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 1.67,
|
| 1566 |
+
"learning_rate": 1.3571852731591449e-05,
|
| 1567 |
+
"loss": 0.8494,
|
| 1568 |
+
"step": 22700
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 1.67,
|
| 1572 |
+
"learning_rate": 1.3277909738717339e-05,
|
| 1573 |
+
"loss": 0.8599,
|
| 1574 |
+
"step": 22800
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 1.68,
|
| 1578 |
+
"learning_rate": 1.2980997624703087e-05,
|
| 1579 |
+
"loss": 0.863,
|
| 1580 |
+
"step": 22900
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 1.69,
|
| 1584 |
+
"learning_rate": 1.2684085510688834e-05,
|
| 1585 |
+
"loss": 0.847,
|
| 1586 |
+
"step": 23000
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 1.69,
|
| 1590 |
+
"eval_loss": 0.15907420217990875,
|
| 1591 |
+
"eval_runtime": 1036.4519,
|
| 1592 |
+
"eval_samples_per_second": 15.444,
|
| 1593 |
+
"eval_steps_per_second": 1.931,
|
| 1594 |
+
"eval_wer": 0.17514066426308314,
|
| 1595 |
+
"step": 23000
|
| 1596 |
+
},
|
| 1597 |
+
{
|
| 1598 |
+
"epoch": 1.69,
|
| 1599 |
+
"learning_rate": 1.2387173396674582e-05,
|
| 1600 |
+
"loss": 0.8487,
|
| 1601 |
+
"step": 23100
|
| 1602 |
+
},
|
| 1603 |
+
{
|
| 1604 |
+
"epoch": 1.7,
|
| 1605 |
+
"learning_rate": 1.2090261282660333e-05,
|
| 1606 |
+
"loss": 0.8637,
|
| 1607 |
+
"step": 23200
|
| 1608 |
+
},
|
| 1609 |
+
{
|
| 1610 |
+
"epoch": 1.71,
|
| 1611 |
+
"learning_rate": 1.1793349168646081e-05,
|
| 1612 |
+
"loss": 0.8456,
|
| 1613 |
+
"step": 23300
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 1.72,
|
| 1617 |
+
"learning_rate": 1.1496437054631828e-05,
|
| 1618 |
+
"loss": 0.8518,
|
| 1619 |
+
"step": 23400
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 1.72,
|
| 1623 |
+
"learning_rate": 1.1199524940617576e-05,
|
| 1624 |
+
"loss": 0.8456,
|
| 1625 |
+
"step": 23500
|
| 1626 |
+
},
|
| 1627 |
+
{
|
| 1628 |
+
"epoch": 1.73,
|
| 1629 |
+
"learning_rate": 1.0902612826603325e-05,
|
| 1630 |
+
"loss": 0.8349,
|
| 1631 |
+
"step": 23600
|
| 1632 |
+
},
|
| 1633 |
+
{
|
| 1634 |
+
"epoch": 1.74,
|
| 1635 |
+
"learning_rate": 1.0605700712589072e-05,
|
| 1636 |
+
"loss": 0.8426,
|
| 1637 |
+
"step": 23700
|
| 1638 |
+
},
|
| 1639 |
+
{
|
| 1640 |
+
"epoch": 1.75,
|
| 1641 |
+
"learning_rate": 1.030878859857482e-05,
|
| 1642 |
+
"loss": 0.8503,
|
| 1643 |
+
"step": 23800
|
| 1644 |
+
},
|
| 1645 |
+
{
|
| 1646 |
+
"epoch": 1.75,
|
| 1647 |
+
"learning_rate": 1.001187648456057e-05,
|
| 1648 |
+
"loss": 0.844,
|
| 1649 |
+
"step": 23900
|
| 1650 |
+
},
|
| 1651 |
+
{
|
| 1652 |
+
"epoch": 1.76,
|
| 1653 |
+
"learning_rate": 9.714964370546319e-06,
|
| 1654 |
+
"loss": 0.822,
|
| 1655 |
+
"step": 24000
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 1.76,
|
| 1659 |
+
"eval_loss": 0.1550702005624771,
|
| 1660 |
+
"eval_runtime": 1027.8442,
|
| 1661 |
+
"eval_samples_per_second": 15.573,
|
| 1662 |
+
"eval_steps_per_second": 1.947,
|
| 1663 |
+
"eval_wer": 0.17010133289631815,
|
| 1664 |
+
"step": 24000
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 1.77,
|
| 1668 |
+
"learning_rate": 9.418052256532066e-06,
|
| 1669 |
+
"loss": 0.8452,
|
| 1670 |
+
"step": 24100
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 1.78,
|
| 1674 |
+
"learning_rate": 9.121140142517814e-06,
|
| 1675 |
+
"loss": 0.843,
|
| 1676 |
+
"step": 24200
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 1.78,
|
| 1680 |
+
"learning_rate": 8.824228028503563e-06,
|
| 1681 |
+
"loss": 0.8429,
|
| 1682 |
+
"step": 24300
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 1.79,
|
| 1686 |
+
"learning_rate": 8.527315914489311e-06,
|
| 1687 |
+
"loss": 0.8513,
|
| 1688 |
+
"step": 24400
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 1.8,
|
| 1692 |
+
"learning_rate": 8.23040380047506e-06,
|
| 1693 |
+
"loss": 0.834,
|
| 1694 |
+
"step": 24500
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 1.8,
|
| 1698 |
+
"learning_rate": 7.933491686460806e-06,
|
| 1699 |
+
"loss": 0.8383,
|
| 1700 |
+
"step": 24600
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 1.81,
|
| 1704 |
+
"learning_rate": 7.636579572446555e-06,
|
| 1705 |
+
"loss": 0.8294,
|
| 1706 |
+
"step": 24700
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 1.82,
|
| 1710 |
+
"learning_rate": 7.339667458432303e-06,
|
| 1711 |
+
"loss": 0.8335,
|
| 1712 |
+
"step": 24800
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 1.83,
|
| 1716 |
+
"learning_rate": 7.042755344418052e-06,
|
| 1717 |
+
"loss": 0.8207,
|
| 1718 |
+
"step": 24900
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 1.83,
|
| 1722 |
+
"learning_rate": 6.745843230403799e-06,
|
| 1723 |
+
"loss": 0.8188,
|
| 1724 |
+
"step": 25000
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 1.83,
|
| 1728 |
+
"eval_loss": 0.1527515947818756,
|
| 1729 |
+
"eval_runtime": 1034.5359,
|
| 1730 |
+
"eval_samples_per_second": 15.473,
|
| 1731 |
+
"eval_steps_per_second": 1.934,
|
| 1732 |
+
"eval_wer": 0.16672812192723696,
|
| 1733 |
+
"step": 25000
|
| 1734 |
+
},
|
| 1735 |
+
{
|
| 1736 |
+
"epoch": 1.84,
|
| 1737 |
+
"learning_rate": 6.448931116389549e-06,
|
| 1738 |
+
"loss": 0.8289,
|
| 1739 |
+
"step": 25100
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 1.85,
|
| 1743 |
+
"learning_rate": 6.152019002375296e-06,
|
| 1744 |
+
"loss": 0.8306,
|
| 1745 |
+
"step": 25200
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 1.86,
|
| 1749 |
+
"learning_rate": 5.855106888361045e-06,
|
| 1750 |
+
"loss": 0.8335,
|
| 1751 |
+
"step": 25300
|
| 1752 |
+
},
|
| 1753 |
+
{
|
| 1754 |
+
"epoch": 1.86,
|
| 1755 |
+
"learning_rate": 5.5581947743467925e-06,
|
| 1756 |
+
"loss": 0.8291,
|
| 1757 |
+
"step": 25400
|
| 1758 |
+
},
|
| 1759 |
+
{
|
| 1760 |
+
"epoch": 1.87,
|
| 1761 |
+
"learning_rate": 5.261282660332541e-06,
|
| 1762 |
+
"loss": 0.8206,
|
| 1763 |
+
"step": 25500
|
| 1764 |
+
},
|
| 1765 |
+
{
|
| 1766 |
+
"epoch": 1.88,
|
| 1767 |
+
"learning_rate": 4.9643705463182895e-06,
|
| 1768 |
+
"loss": 0.8242,
|
| 1769 |
+
"step": 25600
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 1.89,
|
| 1773 |
+
"learning_rate": 4.667458432304038e-06,
|
| 1774 |
+
"loss": 0.8189,
|
| 1775 |
+
"step": 25700
|
| 1776 |
+
},
|
| 1777 |
+
{
|
| 1778 |
+
"epoch": 1.89,
|
| 1779 |
+
"learning_rate": 4.370546318289786e-06,
|
| 1780 |
+
"loss": 0.8275,
|
| 1781 |
+
"step": 25800
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 1.9,
|
| 1785 |
+
"learning_rate": 4.073634204275534e-06,
|
| 1786 |
+
"loss": 0.8142,
|
| 1787 |
+
"step": 25900
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 1.91,
|
| 1791 |
+
"learning_rate": 3.776722090261282e-06,
|
| 1792 |
+
"loss": 0.8305,
|
| 1793 |
+
"step": 26000
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 1.91,
|
| 1797 |
+
"eval_loss": 0.14921718835830688,
|
| 1798 |
+
"eval_runtime": 1026.6478,
|
| 1799 |
+
"eval_samples_per_second": 15.592,
|
| 1800 |
+
"eval_steps_per_second": 1.949,
|
| 1801 |
+
"eval_wer": 0.16312957500273134,
|
| 1802 |
+
"step": 26000
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 1.91,
|
| 1806 |
+
"learning_rate": 3.4798099762470307e-06,
|
| 1807 |
+
"loss": 0.833,
|
| 1808 |
+
"step": 26100
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 1.92,
|
| 1812 |
+
"learning_rate": 3.1828978622327788e-06,
|
| 1813 |
+
"loss": 0.8175,
|
| 1814 |
+
"step": 26200
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 1.93,
|
| 1818 |
+
"learning_rate": 2.888954869358669e-06,
|
| 1819 |
+
"loss": 0.8259,
|
| 1820 |
+
"step": 26300
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 1.94,
|
| 1824 |
+
"learning_rate": 2.5920427553444177e-06,
|
| 1825 |
+
"loss": 0.8262,
|
| 1826 |
+
"step": 26400
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 1.94,
|
| 1830 |
+
"learning_rate": 2.295130641330166e-06,
|
| 1831 |
+
"loss": 0.8223,
|
| 1832 |
+
"step": 26500
|
| 1833 |
+
},
|
| 1834 |
+
{
|
| 1835 |
+
"epoch": 1.95,
|
| 1836 |
+
"learning_rate": 1.9982185273159142e-06,
|
| 1837 |
+
"loss": 0.8285,
|
| 1838 |
+
"step": 26600
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 1.96,
|
| 1842 |
+
"learning_rate": 1.7013064133016625e-06,
|
| 1843 |
+
"loss": 0.8226,
|
| 1844 |
+
"step": 26700
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 1.97,
|
| 1848 |
+
"learning_rate": 1.404394299287411e-06,
|
| 1849 |
+
"loss": 0.8154,
|
| 1850 |
+
"step": 26800
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 1.97,
|
| 1854 |
+
"learning_rate": 1.107482185273159e-06,
|
| 1855 |
+
"loss": 0.8176,
|
| 1856 |
+
"step": 26900
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 1.98,
|
| 1860 |
+
"learning_rate": 8.105700712589074e-07,
|
| 1861 |
+
"loss": 0.8122,
|
| 1862 |
+
"step": 27000
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 1.98,
|
| 1866 |
+
"eval_loss": 0.14789555966854095,
|
| 1867 |
+
"eval_runtime": 1030.7995,
|
| 1868 |
+
"eval_samples_per_second": 15.529,
|
| 1869 |
+
"eval_steps_per_second": 1.941,
|
| 1870 |
+
"eval_wer": 0.16106740959248333,
|
| 1871 |
+
"step": 27000
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 1.99,
|
| 1875 |
+
"learning_rate": 5.136579572446555e-07,
|
| 1876 |
+
"loss": 0.818,
|
| 1877 |
+
"step": 27100
|
| 1878 |
+
},
|
| 1879 |
+
{
|
| 1880 |
+
"epoch": 2.0,
|
| 1881 |
+
"learning_rate": 2.167458432304038e-07,
|
| 1882 |
+
"loss": 0.8284,
|
| 1883 |
+
"step": 27200
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 2.0,
|
| 1887 |
+
"step": 27260,
|
| 1888 |
+
"total_flos": 4.0396309180498005e+20,
|
| 1889 |
+
"train_loss": 0.32739020716330625,
|
| 1890 |
+
"train_runtime": 49115.8494,
|
| 1891 |
+
"train_samples_per_second": 17.761,
|
| 1892 |
+
"train_steps_per_second": 0.555
|
| 1893 |
+
}
|
| 1894 |
+
],
|
| 1895 |
+
"max_steps": 27260,
|
| 1896 |
+
"num_train_epochs": 2,
|
| 1897 |
+
"total_flos": 4.0396309180498005e+20,
|
| 1898 |
+
"trial_name": null,
|
| 1899 |
+
"trial_params": null
|
| 1900 |
+
}
|