Commit
•
6366826
1
Parent(s):
04b0cc1
End of training
Browse files- all_results.json +14 -0
- eval_results.json +9 -0
- pytorch_model.bin +1 -1
- train_results.json +8 -0
- trainer_state.json +1231 -0
all_results.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 100.0,
|
| 3 |
+
"eval_loss": 0.5878269672393799,
|
| 4 |
+
"eval_runtime": 127.4128,
|
| 5 |
+
"eval_samples": 2095,
|
| 6 |
+
"eval_samples_per_second": 16.443,
|
| 7 |
+
"eval_steps_per_second": 1.028,
|
| 8 |
+
"eval_wer": 0.3419183497004509,
|
| 9 |
+
"train_loss": 0.9074438333835732,
|
| 10 |
+
"train_runtime": 47510.6321,
|
| 11 |
+
"train_samples": 4711,
|
| 12 |
+
"train_samples_per_second": 9.916,
|
| 13 |
+
"train_steps_per_second": 0.309
|
| 14 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 100.0,
|
| 3 |
+
"eval_loss": 0.5878269672393799,
|
| 4 |
+
"eval_runtime": 127.4128,
|
| 5 |
+
"eval_samples": 2095,
|
| 6 |
+
"eval_samples_per_second": 16.443,
|
| 7 |
+
"eval_steps_per_second": 1.028,
|
| 8 |
+
"eval_wer": 0.3419183497004509
|
| 9 |
+
}
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3850650929
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bca46f16fe2f42e9926176e5fb128a15246d287a6e09e1ec444fc5636a8ecfac
|
| 3 |
size 3850650929
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 100.0,
|
| 3 |
+
"train_loss": 0.9074438333835732,
|
| 4 |
+
"train_runtime": 47510.6321,
|
| 5 |
+
"train_samples": 4711,
|
| 6 |
+
"train_samples_per_second": 9.916,
|
| 7 |
+
"train_steps_per_second": 0.309
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1231 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 99.99830220713073,
|
| 5 |
+
"global_step": 14700,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.68,
|
| 12 |
+
"learning_rate": 3.6375e-06,
|
| 13 |
+
"loss": 11.6851,
|
| 14 |
+
"step": 100
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 1.36,
|
| 18 |
+
"learning_rate": 7.3875e-06,
|
| 19 |
+
"loss": 3.4316,
|
| 20 |
+
"step": 200
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 2.04,
|
| 24 |
+
"learning_rate": 1.1137499999999998e-05,
|
| 25 |
+
"loss": 2.9224,
|
| 26 |
+
"step": 300
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 2.72,
|
| 30 |
+
"learning_rate": 1.48875e-05,
|
| 31 |
+
"loss": 1.9859,
|
| 32 |
+
"step": 400
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 2.72,
|
| 36 |
+
"eval_loss": 1.1663333177566528,
|
| 37 |
+
"eval_runtime": 129.56,
|
| 38 |
+
"eval_samples_per_second": 16.17,
|
| 39 |
+
"eval_steps_per_second": 1.011,
|
| 40 |
+
"eval_wer": 0.7947625223889815,
|
| 41 |
+
"step": 400
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"epoch": 3.4,
|
| 45 |
+
"learning_rate": 1.86375e-05,
|
| 46 |
+
"loss": 1.625,
|
| 47 |
+
"step": 500
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"epoch": 4.08,
|
| 51 |
+
"learning_rate": 2.23875e-05,
|
| 52 |
+
"loss": 1.4408,
|
| 53 |
+
"step": 600
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 4.76,
|
| 57 |
+
"learning_rate": 2.6137499999999995e-05,
|
| 58 |
+
"loss": 1.3534,
|
| 59 |
+
"step": 700
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 5.44,
|
| 63 |
+
"learning_rate": 2.9887499999999998e-05,
|
| 64 |
+
"loss": 1.2969,
|
| 65 |
+
"step": 800
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 5.44,
|
| 69 |
+
"eval_loss": 0.7725260853767395,
|
| 70 |
+
"eval_runtime": 128.7049,
|
| 71 |
+
"eval_samples_per_second": 16.278,
|
| 72 |
+
"eval_steps_per_second": 1.018,
|
| 73 |
+
"eval_wer": 0.656228769069236,
|
| 74 |
+
"step": 800
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 6.12,
|
| 78 |
+
"learning_rate": 3.36375e-05,
|
| 79 |
+
"loss": 1.2621,
|
| 80 |
+
"step": 900
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 6.8,
|
| 84 |
+
"learning_rate": 3.7387499999999994e-05,
|
| 85 |
+
"loss": 1.2048,
|
| 86 |
+
"step": 1000
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 7.48,
|
| 90 |
+
"learning_rate": 4.11375e-05,
|
| 91 |
+
"loss": 1.1914,
|
| 92 |
+
"step": 1100
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 8.16,
|
| 96 |
+
"learning_rate": 4.48875e-05,
|
| 97 |
+
"loss": 1.1954,
|
| 98 |
+
"step": 1200
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 8.16,
|
| 102 |
+
"eval_loss": 0.5940413475036621,
|
| 103 |
+
"eval_runtime": 129.7001,
|
| 104 |
+
"eval_samples_per_second": 16.153,
|
| 105 |
+
"eval_steps_per_second": 1.01,
|
| 106 |
+
"eval_wer": 0.4903958989562102,
|
| 107 |
+
"step": 1200
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 8.84,
|
| 111 |
+
"learning_rate": 4.8637499999999996e-05,
|
| 112 |
+
"loss": 1.1675,
|
| 113 |
+
"step": 1300
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 9.52,
|
| 117 |
+
"learning_rate": 5.23875e-05,
|
| 118 |
+
"loss": 1.1484,
|
| 119 |
+
"step": 1400
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 10.2,
|
| 123 |
+
"learning_rate": 5.61375e-05,
|
| 124 |
+
"loss": 1.1564,
|
| 125 |
+
"step": 1500
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 10.88,
|
| 129 |
+
"learning_rate": 5.988749999999999e-05,
|
| 130 |
+
"loss": 1.164,
|
| 131 |
+
"step": 1600
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 10.88,
|
| 135 |
+
"eval_loss": 0.5338293313980103,
|
| 136 |
+
"eval_runtime": 129.1415,
|
| 137 |
+
"eval_samples_per_second": 16.223,
|
| 138 |
+
"eval_steps_per_second": 1.014,
|
| 139 |
+
"eval_wer": 0.4315978012476067,
|
| 140 |
+
"step": 1600
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 11.56,
|
| 144 |
+
"learning_rate": 6.36375e-05,
|
| 145 |
+
"loss": 1.1467,
|
| 146 |
+
"step": 1700
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 12.24,
|
| 150 |
+
"learning_rate": 6.738749999999999e-05,
|
| 151 |
+
"loss": 1.1664,
|
| 152 |
+
"step": 1800
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 12.92,
|
| 156 |
+
"learning_rate": 7.11375e-05,
|
| 157 |
+
"loss": 1.1658,
|
| 158 |
+
"step": 1900
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 13.6,
|
| 162 |
+
"learning_rate": 7.48875e-05,
|
| 163 |
+
"loss": 1.1464,
|
| 164 |
+
"step": 2000
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 13.6,
|
| 168 |
+
"eval_loss": 0.5432471632957458,
|
| 169 |
+
"eval_runtime": 128.5422,
|
| 170 |
+
"eval_samples_per_second": 16.298,
|
| 171 |
+
"eval_steps_per_second": 1.019,
|
| 172 |
+
"eval_wer": 0.42264220863442653,
|
| 173 |
+
"step": 2000
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 14.29,
|
| 177 |
+
"learning_rate": 7.44271653543307e-05,
|
| 178 |
+
"loss": 1.1256,
|
| 179 |
+
"step": 2100
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 14.96,
|
| 183 |
+
"learning_rate": 7.383661417322834e-05,
|
| 184 |
+
"loss": 1.1464,
|
| 185 |
+
"step": 2200
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 15.65,
|
| 189 |
+
"learning_rate": 7.324606299212597e-05,
|
| 190 |
+
"loss": 1.1545,
|
| 191 |
+
"step": 2300
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 16.33,
|
| 195 |
+
"learning_rate": 7.265551181102362e-05,
|
| 196 |
+
"loss": 1.1553,
|
| 197 |
+
"step": 2400
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 16.33,
|
| 201 |
+
"eval_loss": 0.5470633506774902,
|
| 202 |
+
"eval_runtime": 128.2784,
|
| 203 |
+
"eval_samples_per_second": 16.332,
|
| 204 |
+
"eval_steps_per_second": 1.021,
|
| 205 |
+
"eval_wer": 0.42603915755666727,
|
| 206 |
+
"step": 2400
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 17.01,
|
| 210 |
+
"learning_rate": 7.206496062992125e-05,
|
| 211 |
+
"loss": 1.1332,
|
| 212 |
+
"step": 2500
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 17.69,
|
| 216 |
+
"learning_rate": 7.14744094488189e-05,
|
| 217 |
+
"loss": 1.1317,
|
| 218 |
+
"step": 2600
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 18.37,
|
| 222 |
+
"learning_rate": 7.088385826771653e-05,
|
| 223 |
+
"loss": 1.1197,
|
| 224 |
+
"step": 2700
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 19.05,
|
| 228 |
+
"learning_rate": 7.029330708661417e-05,
|
| 229 |
+
"loss": 1.0985,
|
| 230 |
+
"step": 2800
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 19.05,
|
| 234 |
+
"eval_loss": 0.5289891958236694,
|
| 235 |
+
"eval_runtime": 128.1965,
|
| 236 |
+
"eval_samples_per_second": 16.342,
|
| 237 |
+
"eval_steps_per_second": 1.022,
|
| 238 |
+
"eval_wer": 0.407572107961213,
|
| 239 |
+
"step": 2800
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 19.73,
|
| 243 |
+
"learning_rate": 6.970275590551181e-05,
|
| 244 |
+
"loss": 1.0637,
|
| 245 |
+
"step": 2900
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 20.41,
|
| 249 |
+
"learning_rate": 6.911220472440943e-05,
|
| 250 |
+
"loss": 1.0786,
|
| 251 |
+
"step": 3000
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 21.09,
|
| 255 |
+
"learning_rate": 6.852165354330708e-05,
|
| 256 |
+
"loss": 1.0638,
|
| 257 |
+
"step": 3100
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 21.77,
|
| 261 |
+
"learning_rate": 6.793110236220472e-05,
|
| 262 |
+
"loss": 1.0421,
|
| 263 |
+
"step": 3200
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 21.77,
|
| 267 |
+
"eval_loss": 0.5671822428703308,
|
| 268 |
+
"eval_runtime": 128.6139,
|
| 269 |
+
"eval_samples_per_second": 16.289,
|
| 270 |
+
"eval_steps_per_second": 1.019,
|
| 271 |
+
"eval_wer": 0.4181335309739979,
|
| 272 |
+
"step": 3200
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 22.45,
|
| 276 |
+
"learning_rate": 6.734055118110236e-05,
|
| 277 |
+
"loss": 1.0587,
|
| 278 |
+
"step": 3300
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 23.13,
|
| 282 |
+
"learning_rate": 6.675e-05,
|
| 283 |
+
"loss": 1.0431,
|
| 284 |
+
"step": 3400
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 23.81,
|
| 288 |
+
"learning_rate": 6.615944881889763e-05,
|
| 289 |
+
"loss": 1.0282,
|
| 290 |
+
"step": 3500
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 24.49,
|
| 294 |
+
"learning_rate": 6.556889763779528e-05,
|
| 295 |
+
"loss": 0.9831,
|
| 296 |
+
"step": 3600
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 24.49,
|
| 300 |
+
"eval_loss": 0.5740545392036438,
|
| 301 |
+
"eval_runtime": 128.3992,
|
| 302 |
+
"eval_samples_per_second": 16.316,
|
| 303 |
+
"eval_steps_per_second": 1.02,
|
| 304 |
+
"eval_wer": 0.41405719226730897,
|
| 305 |
+
"step": 3600
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 25.17,
|
| 309 |
+
"learning_rate": 6.497834645669291e-05,
|
| 310 |
+
"loss": 1.0409,
|
| 311 |
+
"step": 3700
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 25.85,
|
| 315 |
+
"learning_rate": 6.438779527559054e-05,
|
| 316 |
+
"loss": 1.0023,
|
| 317 |
+
"step": 3800
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 26.53,
|
| 321 |
+
"learning_rate": 6.379724409448818e-05,
|
| 322 |
+
"loss": 0.9724,
|
| 323 |
+
"step": 3900
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 27.21,
|
| 327 |
+
"learning_rate": 6.320669291338583e-05,
|
| 328 |
+
"loss": 0.9827,
|
| 329 |
+
"step": 4000
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 27.21,
|
| 333 |
+
"eval_loss": 0.5753942131996155,
|
| 334 |
+
"eval_runtime": 129.7692,
|
| 335 |
+
"eval_samples_per_second": 16.144,
|
| 336 |
+
"eval_steps_per_second": 1.009,
|
| 337 |
+
"eval_wer": 0.4178864801432895,
|
| 338 |
+
"step": 4000
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 27.89,
|
| 342 |
+
"learning_rate": 6.261614173228346e-05,
|
| 343 |
+
"loss": 0.9974,
|
| 344 |
+
"step": 4100
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 28.57,
|
| 348 |
+
"learning_rate": 6.203149606299212e-05,
|
| 349 |
+
"loss": 0.9666,
|
| 350 |
+
"step": 4200
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 29.25,
|
| 354 |
+
"learning_rate": 6.144685039370079e-05,
|
| 355 |
+
"loss": 0.9571,
|
| 356 |
+
"step": 4300
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 29.93,
|
| 360 |
+
"learning_rate": 6.085629921259842e-05,
|
| 361 |
+
"loss": 0.9669,
|
| 362 |
+
"step": 4400
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 29.93,
|
| 366 |
+
"eval_loss": 0.5309818387031555,
|
| 367 |
+
"eval_runtime": 127.6098,
|
| 368 |
+
"eval_samples_per_second": 16.417,
|
| 369 |
+
"eval_steps_per_second": 1.027,
|
| 370 |
+
"eval_wer": 0.38885800753505034,
|
| 371 |
+
"step": 4400
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 30.61,
|
| 375 |
+
"learning_rate": 6.026574803149606e-05,
|
| 376 |
+
"loss": 0.9642,
|
| 377 |
+
"step": 4500
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 31.29,
|
| 381 |
+
"learning_rate": 5.967519685039369e-05,
|
| 382 |
+
"loss": 0.9843,
|
| 383 |
+
"step": 4600
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 31.97,
|
| 387 |
+
"learning_rate": 5.908464566929133e-05,
|
| 388 |
+
"loss": 0.9589,
|
| 389 |
+
"step": 4700
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 32.65,
|
| 393 |
+
"learning_rate": 5.8494094488188974e-05,
|
| 394 |
+
"loss": 0.9496,
|
| 395 |
+
"step": 4800
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 32.65,
|
| 399 |
+
"eval_loss": 0.5648738145828247,
|
| 400 |
+
"eval_runtime": 128.3031,
|
| 401 |
+
"eval_samples_per_second": 16.329,
|
| 402 |
+
"eval_steps_per_second": 1.021,
|
| 403 |
+
"eval_wer": 0.4061515656846396,
|
| 404 |
+
"step": 4800
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 33.33,
|
| 408 |
+
"learning_rate": 5.7903543307086614e-05,
|
| 409 |
+
"loss": 0.9264,
|
| 410 |
+
"step": 4900
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 34.01,
|
| 414 |
+
"learning_rate": 5.731299212598425e-05,
|
| 415 |
+
"loss": 0.9117,
|
| 416 |
+
"step": 5000
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 34.69,
|
| 420 |
+
"learning_rate": 5.672244094488189e-05,
|
| 421 |
+
"loss": 0.9176,
|
| 422 |
+
"step": 5100
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 35.37,
|
| 426 |
+
"learning_rate": 5.613188976377952e-05,
|
| 427 |
+
"loss": 0.9112,
|
| 428 |
+
"step": 5200
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 35.37,
|
| 432 |
+
"eval_loss": 0.5737842321395874,
|
| 433 |
+
"eval_runtime": 127.713,
|
| 434 |
+
"eval_samples_per_second": 16.404,
|
| 435 |
+
"eval_steps_per_second": 1.026,
|
| 436 |
+
"eval_wer": 0.3925637699956766,
|
| 437 |
+
"step": 5200
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 36.05,
|
| 441 |
+
"learning_rate": 5.5541338582677156e-05,
|
| 442 |
+
"loss": 0.903,
|
| 443 |
+
"step": 5300
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 36.73,
|
| 447 |
+
"learning_rate": 5.49507874015748e-05,
|
| 448 |
+
"loss": 0.8615,
|
| 449 |
+
"step": 5400
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 37.41,
|
| 453 |
+
"learning_rate": 5.436023622047244e-05,
|
| 454 |
+
"loss": 0.904,
|
| 455 |
+
"step": 5500
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 38.1,
|
| 459 |
+
"learning_rate": 5.376968503937008e-05,
|
| 460 |
+
"loss": 0.8838,
|
| 461 |
+
"step": 5600
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 38.1,
|
| 465 |
+
"eval_loss": 0.5232195258140564,
|
| 466 |
+
"eval_runtime": 127.5056,
|
| 467 |
+
"eval_samples_per_second": 16.431,
|
| 468 |
+
"eval_steps_per_second": 1.027,
|
| 469 |
+
"eval_wer": 0.37681427953801494,
|
| 470 |
+
"step": 5600
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 38.77,
|
| 474 |
+
"learning_rate": 5.317913385826771e-05,
|
| 475 |
+
"loss": 0.8693,
|
| 476 |
+
"step": 5700
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 39.46,
|
| 480 |
+
"learning_rate": 5.258858267716535e-05,
|
| 481 |
+
"loss": 0.8653,
|
| 482 |
+
"step": 5800
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 40.14,
|
| 486 |
+
"learning_rate": 5.1998031496062986e-05,
|
| 487 |
+
"loss": 0.8862,
|
| 488 |
+
"step": 5900
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 40.81,
|
| 492 |
+
"learning_rate": 5.140748031496062e-05,
|
| 493 |
+
"loss": 0.8666,
|
| 494 |
+
"step": 6000
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 40.81,
|
| 498 |
+
"eval_loss": 0.5510496497154236,
|
| 499 |
+
"eval_runtime": 128.5402,
|
| 500 |
+
"eval_samples_per_second": 16.298,
|
| 501 |
+
"eval_steps_per_second": 1.019,
|
| 502 |
+
"eval_wer": 0.3852140077821012,
|
| 503 |
+
"step": 6000
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 41.5,
|
| 507 |
+
"learning_rate": 5.081692913385826e-05,
|
| 508 |
+
"loss": 0.8378,
|
| 509 |
+
"step": 6100
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 42.18,
|
| 513 |
+
"learning_rate": 5.02263779527559e-05,
|
| 514 |
+
"loss": 0.8636,
|
| 515 |
+
"step": 6200
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 42.86,
|
| 519 |
+
"learning_rate": 4.963582677165354e-05,
|
| 520 |
+
"loss": 0.8464,
|
| 521 |
+
"step": 6300
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 43.54,
|
| 525 |
+
"learning_rate": 4.9045275590551175e-05,
|
| 526 |
+
"loss": 0.8366,
|
| 527 |
+
"step": 6400
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 43.54,
|
| 531 |
+
"eval_loss": 0.5436074733734131,
|
| 532 |
+
"eval_runtime": 127.7975,
|
| 533 |
+
"eval_samples_per_second": 16.393,
|
| 534 |
+
"eval_steps_per_second": 1.025,
|
| 535 |
+
"eval_wer": 0.3837317027978507,
|
| 536 |
+
"step": 6400
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 44.22,
|
| 540 |
+
"learning_rate": 4.846062992125984e-05,
|
| 541 |
+
"loss": 0.8298,
|
| 542 |
+
"step": 6500
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 44.9,
|
| 546 |
+
"learning_rate": 4.7870078740157475e-05,
|
| 547 |
+
"loss": 0.8385,
|
| 548 |
+
"step": 6600
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 45.58,
|
| 552 |
+
"learning_rate": 4.727952755905511e-05,
|
| 553 |
+
"loss": 0.8347,
|
| 554 |
+
"step": 6700
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 46.26,
|
| 558 |
+
"learning_rate": 4.668897637795275e-05,
|
| 559 |
+
"loss": 0.7957,
|
| 560 |
+
"step": 6800
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 46.26,
|
| 564 |
+
"eval_loss": 0.5337450504302979,
|
| 565 |
+
"eval_runtime": 128.0855,
|
| 566 |
+
"eval_samples_per_second": 16.356,
|
| 567 |
+
"eval_steps_per_second": 1.023,
|
| 568 |
+
"eval_wer": 0.3774936693224631,
|
| 569 |
+
"step": 6800
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 46.94,
|
| 573 |
+
"learning_rate": 4.609842519685039e-05,
|
| 574 |
+
"loss": 0.8226,
|
| 575 |
+
"step": 6900
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 47.62,
|
| 579 |
+
"learning_rate": 4.550787401574803e-05,
|
| 580 |
+
"loss": 0.8228,
|
| 581 |
+
"step": 7000
|
| 582 |
+
},
|
| 583 |
+
{
|
| 584 |
+
"epoch": 48.3,
|
| 585 |
+
"learning_rate": 4.4917322834645664e-05,
|
| 586 |
+
"loss": 0.815,
|
| 587 |
+
"step": 7100
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 48.98,
|
| 591 |
+
"learning_rate": 4.4326771653543305e-05,
|
| 592 |
+
"loss": 0.7834,
|
| 593 |
+
"step": 7200
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 48.98,
|
| 597 |
+
"eval_loss": 0.561106264591217,
|
| 598 |
+
"eval_runtime": 127.7689,
|
| 599 |
+
"eval_samples_per_second": 16.397,
|
| 600 |
+
"eval_steps_per_second": 1.025,
|
| 601 |
+
"eval_wer": 0.38441109258229883,
|
| 602 |
+
"step": 7200
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 49.66,
|
| 606 |
+
"learning_rate": 4.3736220472440945e-05,
|
| 607 |
+
"loss": 0.7998,
|
| 608 |
+
"step": 7300
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 50.34,
|
| 612 |
+
"learning_rate": 4.314566929133857e-05,
|
| 613 |
+
"loss": 0.7898,
|
| 614 |
+
"step": 7400
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 51.02,
|
| 618 |
+
"learning_rate": 4.255511811023621e-05,
|
| 619 |
+
"loss": 0.7567,
|
| 620 |
+
"step": 7500
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 51.7,
|
| 624 |
+
"learning_rate": 4.1964566929133854e-05,
|
| 625 |
+
"loss": 0.7685,
|
| 626 |
+
"step": 7600
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 51.7,
|
| 630 |
+
"eval_loss": 0.571021556854248,
|
| 631 |
+
"eval_runtime": 126.8927,
|
| 632 |
+
"eval_samples_per_second": 16.51,
|
| 633 |
+
"eval_steps_per_second": 1.032,
|
| 634 |
+
"eval_wer": 0.40077821011673154,
|
| 635 |
+
"step": 7600
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 52.38,
|
| 639 |
+
"learning_rate": 4.1374015748031494e-05,
|
| 640 |
+
"loss": 0.7744,
|
| 641 |
+
"step": 7700
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 53.06,
|
| 645 |
+
"learning_rate": 4.078346456692913e-05,
|
| 646 |
+
"loss": 0.7483,
|
| 647 |
+
"step": 7800
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 53.74,
|
| 651 |
+
"learning_rate": 4.019291338582677e-05,
|
| 652 |
+
"loss": 0.76,
|
| 653 |
+
"step": 7900
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 54.42,
|
| 657 |
+
"learning_rate": 3.960236220472441e-05,
|
| 658 |
+
"loss": 0.7431,
|
| 659 |
+
"step": 8000
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 54.42,
|
| 663 |
+
"eval_loss": 0.5636317729949951,
|
| 664 |
+
"eval_runtime": 128.5647,
|
| 665 |
+
"eval_samples_per_second": 16.295,
|
| 666 |
+
"eval_steps_per_second": 1.019,
|
| 667 |
+
"eval_wer": 0.37261441541597184,
|
| 668 |
+
"step": 8000
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 55.1,
|
| 672 |
+
"learning_rate": 3.901181102362205e-05,
|
| 673 |
+
"loss": 0.746,
|
| 674 |
+
"step": 8100
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 55.78,
|
| 678 |
+
"learning_rate": 3.842125984251968e-05,
|
| 679 |
+
"loss": 0.7162,
|
| 680 |
+
"step": 8200
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 56.46,
|
| 684 |
+
"learning_rate": 3.783070866141732e-05,
|
| 685 |
+
"loss": 0.7253,
|
| 686 |
+
"step": 8300
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 57.14,
|
| 690 |
+
"learning_rate": 3.724015748031496e-05,
|
| 691 |
+
"loss": 0.7353,
|
| 692 |
+
"step": 8400
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 57.14,
|
| 696 |
+
"eval_loss": 0.5937429070472717,
|
| 697 |
+
"eval_runtime": 127.7429,
|
| 698 |
+
"eval_samples_per_second": 16.4,
|
| 699 |
+
"eval_steps_per_second": 1.025,
|
| 700 |
+
"eval_wer": 0.3836081773824965,
|
| 701 |
+
"step": 8400
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 57.82,
|
| 705 |
+
"learning_rate": 3.664960629921259e-05,
|
| 706 |
+
"loss": 0.6947,
|
| 707 |
+
"step": 8500
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 58.5,
|
| 711 |
+
"learning_rate": 3.605905511811023e-05,
|
| 712 |
+
"loss": 0.6965,
|
| 713 |
+
"step": 8600
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 59.18,
|
| 717 |
+
"learning_rate": 3.546850393700787e-05,
|
| 718 |
+
"loss": 0.6975,
|
| 719 |
+
"step": 8700
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 59.86,
|
| 723 |
+
"learning_rate": 3.4877952755905506e-05,
|
| 724 |
+
"loss": 0.7001,
|
| 725 |
+
"step": 8800
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 59.86,
|
| 729 |
+
"eval_loss": 0.5815082788467407,
|
| 730 |
+
"eval_runtime": 127.9156,
|
| 731 |
+
"eval_samples_per_second": 16.378,
|
| 732 |
+
"eval_steps_per_second": 1.024,
|
| 733 |
+
"eval_wer": 0.38583163485887223,
|
| 734 |
+
"step": 8800
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 60.54,
|
| 738 |
+
"learning_rate": 3.428740157480315e-05,
|
| 739 |
+
"loss": 0.694,
|
| 740 |
+
"step": 8900
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 61.22,
|
| 744 |
+
"learning_rate": 3.369685039370079e-05,
|
| 745 |
+
"loss": 0.6867,
|
| 746 |
+
"step": 9000
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 61.9,
|
| 750 |
+
"learning_rate": 3.310629921259842e-05,
|
| 751 |
+
"loss": 0.7046,
|
| 752 |
+
"step": 9100
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 62.58,
|
| 756 |
+
"learning_rate": 3.2515748031496055e-05,
|
| 757 |
+
"loss": 0.6799,
|
| 758 |
+
"step": 9200
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 62.58,
|
| 762 |
+
"eval_loss": 0.5861709713935852,
|
| 763 |
+
"eval_runtime": 127.4436,
|
| 764 |
+
"eval_samples_per_second": 16.439,
|
| 765 |
+
"eval_steps_per_second": 1.028,
|
| 766 |
+
"eval_wer": 0.36964980544747084,
|
| 767 |
+
"step": 9200
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 63.26,
|
| 771 |
+
"learning_rate": 3.1925196850393696e-05,
|
| 772 |
+
"loss": 0.6639,
|
| 773 |
+
"step": 9300
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 63.94,
|
| 777 |
+
"learning_rate": 3.1334645669291336e-05,
|
| 778 |
+
"loss": 0.6543,
|
| 779 |
+
"step": 9400
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 64.62,
|
| 783 |
+
"learning_rate": 3.074409448818898e-05,
|
| 784 |
+
"loss": 0.6746,
|
| 785 |
+
"step": 9500
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 65.31,
|
| 789 |
+
"learning_rate": 3.015354330708661e-05,
|
| 790 |
+
"loss": 0.6459,
|
| 791 |
+
"step": 9600
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 65.31,
|
| 795 |
+
"eval_loss": 0.6180748343467712,
|
| 796 |
+
"eval_runtime": 127.2163,
|
| 797 |
+
"eval_samples_per_second": 16.468,
|
| 798 |
+
"eval_steps_per_second": 1.03,
|
| 799 |
+
"eval_wer": 0.3761966524612439,
|
| 800 |
+
"step": 9600
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 65.98,
|
| 804 |
+
"learning_rate": 2.9562992125984248e-05,
|
| 805 |
+
"loss": 0.625,
|
| 806 |
+
"step": 9700
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 66.67,
|
| 810 |
+
"learning_rate": 2.897244094488189e-05,
|
| 811 |
+
"loss": 0.6256,
|
| 812 |
+
"step": 9800
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 67.35,
|
| 816 |
+
"learning_rate": 2.8381889763779526e-05,
|
| 817 |
+
"loss": 0.6284,
|
| 818 |
+
"step": 9900
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 68.03,
|
| 822 |
+
"learning_rate": 2.7791338582677163e-05,
|
| 823 |
+
"loss": 0.6121,
|
| 824 |
+
"step": 10000
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 68.03,
|
| 828 |
+
"eval_loss": 0.5636932253837585,
|
| 829 |
+
"eval_runtime": 128.1438,
|
| 830 |
+
"eval_samples_per_second": 16.349,
|
| 831 |
+
"eval_steps_per_second": 1.022,
|
| 832 |
+
"eval_wer": 0.35896485701933173,
|
| 833 |
+
"step": 10000
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 68.71,
|
| 837 |
+
"learning_rate": 2.7206692913385825e-05,
|
| 838 |
+
"loss": 0.6123,
|
| 839 |
+
"step": 10100
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 69.39,
|
| 843 |
+
"learning_rate": 2.6616141732283462e-05,
|
| 844 |
+
"loss": 0.605,
|
| 845 |
+
"step": 10200
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 70.07,
|
| 849 |
+
"learning_rate": 2.60255905511811e-05,
|
| 850 |
+
"loss": 0.6029,
|
| 851 |
+
"step": 10300
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 70.75,
|
| 855 |
+
"learning_rate": 2.5435039370078737e-05,
|
| 856 |
+
"loss": 0.5942,
|
| 857 |
+
"step": 10400
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 70.75,
|
| 861 |
+
"eval_loss": 0.6374208927154541,
|
| 862 |
+
"eval_runtime": 127.5478,
|
| 863 |
+
"eval_samples_per_second": 16.425,
|
| 864 |
+
"eval_steps_per_second": 1.027,
|
| 865 |
+
"eval_wer": 0.3881786177506022,
|
| 866 |
+
"step": 10400
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 71.43,
|
| 870 |
+
"learning_rate": 2.4844488188976377e-05,
|
| 871 |
+
"loss": 0.6081,
|
| 872 |
+
"step": 10500
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 72.11,
|
| 876 |
+
"learning_rate": 2.4253937007874014e-05,
|
| 877 |
+
"loss": 0.5865,
|
| 878 |
+
"step": 10600
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 72.79,
|
| 882 |
+
"learning_rate": 2.3663385826771648e-05,
|
| 883 |
+
"loss": 0.5695,
|
| 884 |
+
"step": 10700
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 73.47,
|
| 888 |
+
"learning_rate": 2.307283464566929e-05,
|
| 889 |
+
"loss": 0.5769,
|
| 890 |
+
"step": 10800
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 73.47,
|
| 894 |
+
"eval_loss": 0.601510226726532,
|
| 895 |
+
"eval_runtime": 127.2518,
|
| 896 |
+
"eval_samples_per_second": 16.463,
|
| 897 |
+
"eval_steps_per_second": 1.029,
|
| 898 |
+
"eval_wer": 0.3639676363411772,
|
| 899 |
+
"step": 10800
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 74.15,
|
| 903 |
+
"learning_rate": 2.2482283464566926e-05,
|
| 904 |
+
"loss": 0.5827,
|
| 905 |
+
"step": 10900
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 74.83,
|
| 909 |
+
"learning_rate": 2.1891732283464567e-05,
|
| 910 |
+
"loss": 0.5751,
|
| 911 |
+
"step": 11000
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 75.51,
|
| 915 |
+
"learning_rate": 2.1307086614173225e-05,
|
| 916 |
+
"loss": 0.5448,
|
| 917 |
+
"step": 11100
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 76.19,
|
| 921 |
+
"learning_rate": 2.0716535433070866e-05,
|
| 922 |
+
"loss": 0.5689,
|
| 923 |
+
"step": 11200
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 76.19,
|
| 927 |
+
"eval_loss": 0.5669254064559937,
|
| 928 |
+
"eval_runtime": 127.9372,
|
| 929 |
+
"eval_samples_per_second": 16.375,
|
| 930 |
+
"eval_steps_per_second": 1.024,
|
| 931 |
+
"eval_wer": 0.3507504168982768,
|
| 932 |
+
"step": 11200
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 76.87,
|
| 936 |
+
"learning_rate": 2.0125984251968503e-05,
|
| 937 |
+
"loss": 0.5606,
|
| 938 |
+
"step": 11300
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 77.55,
|
| 942 |
+
"learning_rate": 1.9535433070866137e-05,
|
| 943 |
+
"loss": 0.544,
|
| 944 |
+
"step": 11400
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 78.23,
|
| 948 |
+
"learning_rate": 1.8944881889763778e-05,
|
| 949 |
+
"loss": 0.5373,
|
| 950 |
+
"step": 11500
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 78.91,
|
| 954 |
+
"learning_rate": 1.8354330708661415e-05,
|
| 955 |
+
"loss": 0.5461,
|
| 956 |
+
"step": 11600
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 78.91,
|
| 960 |
+
"eval_loss": 0.5967420339584351,
|
| 961 |
+
"eval_runtime": 127.7784,
|
| 962 |
+
"eval_samples_per_second": 16.396,
|
| 963 |
+
"eval_steps_per_second": 1.025,
|
| 964 |
+
"eval_wer": 0.36205299240318695,
|
| 965 |
+
"step": 11600
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 79.59,
|
| 969 |
+
"learning_rate": 1.7763779527559052e-05,
|
| 970 |
+
"loss": 0.5342,
|
| 971 |
+
"step": 11700
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 80.27,
|
| 975 |
+
"learning_rate": 1.7173228346456693e-05,
|
| 976 |
+
"loss": 0.5105,
|
| 977 |
+
"step": 11800
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 80.95,
|
| 981 |
+
"learning_rate": 1.658267716535433e-05,
|
| 982 |
+
"loss": 0.5289,
|
| 983 |
+
"step": 11900
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 81.63,
|
| 987 |
+
"learning_rate": 1.5992125984251967e-05,
|
| 988 |
+
"loss": 0.5286,
|
| 989 |
+
"step": 12000
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 81.63,
|
| 993 |
+
"eval_loss": 0.5839831829071045,
|
| 994 |
+
"eval_runtime": 127.0078,
|
| 995 |
+
"eval_samples_per_second": 16.495,
|
| 996 |
+
"eval_steps_per_second": 1.031,
|
| 997 |
+
"eval_wer": 0.36050892471125934,
|
| 998 |
+
"step": 12000
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 82.31,
|
| 1002 |
+
"learning_rate": 1.5401574803149604e-05,
|
| 1003 |
+
"loss": 0.5179,
|
| 1004 |
+
"step": 12100
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 82.99,
|
| 1008 |
+
"learning_rate": 1.4811023622047243e-05,
|
| 1009 |
+
"loss": 0.502,
|
| 1010 |
+
"step": 12200
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 83.67,
|
| 1014 |
+
"learning_rate": 1.422047244094488e-05,
|
| 1015 |
+
"loss": 0.5153,
|
| 1016 |
+
"step": 12300
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 84.35,
|
| 1020 |
+
"learning_rate": 1.3629921259842519e-05,
|
| 1021 |
+
"loss": 0.5057,
|
| 1022 |
+
"step": 12400
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 84.35,
|
| 1026 |
+
"eval_loss": 0.5848367214202881,
|
| 1027 |
+
"eval_runtime": 128.5953,
|
| 1028 |
+
"eval_samples_per_second": 16.291,
|
| 1029 |
+
"eval_steps_per_second": 1.019,
|
| 1030 |
+
"eval_wer": 0.34889753566796367,
|
| 1031 |
+
"step": 12400
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 85.03,
|
| 1035 |
+
"learning_rate": 1.3039370078740156e-05,
|
| 1036 |
+
"loss": 0.5021,
|
| 1037 |
+
"step": 12500
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 85.71,
|
| 1041 |
+
"learning_rate": 1.2448818897637795e-05,
|
| 1042 |
+
"loss": 0.5002,
|
| 1043 |
+
"step": 12600
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 86.39,
|
| 1047 |
+
"learning_rate": 1.1858267716535432e-05,
|
| 1048 |
+
"loss": 0.495,
|
| 1049 |
+
"step": 12700
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 87.07,
|
| 1053 |
+
"learning_rate": 1.1267716535433071e-05,
|
| 1054 |
+
"loss": 0.482,
|
| 1055 |
+
"step": 12800
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 87.07,
|
| 1059 |
+
"eval_loss": 0.5860108733177185,
|
| 1060 |
+
"eval_runtime": 129.85,
|
| 1061 |
+
"eval_samples_per_second": 16.134,
|
| 1062 |
+
"eval_steps_per_second": 1.009,
|
| 1063 |
+
"eval_wer": 0.34877401025260946,
|
| 1064 |
+
"step": 12800
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 87.75,
|
| 1068 |
+
"learning_rate": 1.0677165354330707e-05,
|
| 1069 |
+
"loss": 0.4651,
|
| 1070 |
+
"step": 12900
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 88.43,
|
| 1074 |
+
"learning_rate": 1.0086614173228345e-05,
|
| 1075 |
+
"loss": 0.4894,
|
| 1076 |
+
"step": 13000
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 89.12,
|
| 1080 |
+
"learning_rate": 9.496062992125983e-06,
|
| 1081 |
+
"loss": 0.49,
|
| 1082 |
+
"step": 13100
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 89.79,
|
| 1086 |
+
"learning_rate": 8.905511811023621e-06,
|
| 1087 |
+
"loss": 0.4655,
|
| 1088 |
+
"step": 13200
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 89.79,
|
| 1092 |
+
"eval_loss": 0.5780399441719055,
|
| 1093 |
+
"eval_runtime": 126.603,
|
| 1094 |
+
"eval_samples_per_second": 16.548,
|
| 1095 |
+
"eval_steps_per_second": 1.035,
|
| 1096 |
+
"eval_wer": 0.3453152986226916,
|
| 1097 |
+
"step": 13200
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 90.48,
|
| 1101 |
+
"learning_rate": 8.31496062992126e-06,
|
| 1102 |
+
"loss": 0.4661,
|
| 1103 |
+
"step": 13300
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"epoch": 91.16,
|
| 1107 |
+
"learning_rate": 7.724409448818898e-06,
|
| 1108 |
+
"loss": 0.4776,
|
| 1109 |
+
"step": 13400
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 91.84,
|
| 1113 |
+
"learning_rate": 7.133858267716534e-06,
|
| 1114 |
+
"loss": 0.4697,
|
| 1115 |
+
"step": 13500
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 92.52,
|
| 1119 |
+
"learning_rate": 6.543307086614172e-06,
|
| 1120 |
+
"loss": 0.4523,
|
| 1121 |
+
"step": 13600
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 92.52,
|
| 1125 |
+
"eval_loss": 0.6150020956993103,
|
| 1126 |
+
"eval_runtime": 127.4399,
|
| 1127 |
+
"eval_samples_per_second": 16.439,
|
| 1128 |
+
"eval_steps_per_second": 1.028,
|
| 1129 |
+
"eval_wer": 0.353220925205361,
|
| 1130 |
+
"step": 13600
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 93.2,
|
| 1134 |
+
"learning_rate": 5.95275590551181e-06,
|
| 1135 |
+
"loss": 0.4522,
|
| 1136 |
+
"step": 13700
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 93.88,
|
| 1140 |
+
"learning_rate": 5.362204724409448e-06,
|
| 1141 |
+
"loss": 0.4595,
|
| 1142 |
+
"step": 13800
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 94.56,
|
| 1146 |
+
"learning_rate": 4.77755905511811e-06,
|
| 1147 |
+
"loss": 0.4518,
|
| 1148 |
+
"step": 13900
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 95.24,
|
| 1152 |
+
"learning_rate": 4.1870078740157475e-06,
|
| 1153 |
+
"loss": 0.4422,
|
| 1154 |
+
"step": 14000
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 95.24,
|
| 1158 |
+
"eval_loss": 0.5930356383323669,
|
| 1159 |
+
"eval_runtime": 128.0708,
|
| 1160 |
+
"eval_samples_per_second": 16.358,
|
| 1161 |
+
"eval_steps_per_second": 1.023,
|
| 1162 |
+
"eval_wer": 0.3451917732073374,
|
| 1163 |
+
"step": 14000
|
| 1164 |
+
},
|
| 1165 |
+
{
|
| 1166 |
+
"epoch": 95.92,
|
| 1167 |
+
"learning_rate": 3.5964566929133855e-06,
|
| 1168 |
+
"loss": 0.4558,
|
| 1169 |
+
"step": 14100
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 96.6,
|
| 1173 |
+
"learning_rate": 3.0059055118110235e-06,
|
| 1174 |
+
"loss": 0.4427,
|
| 1175 |
+
"step": 14200
|
| 1176 |
+
},
|
| 1177 |
+
{
|
| 1178 |
+
"epoch": 97.28,
|
| 1179 |
+
"learning_rate": 2.415354330708661e-06,
|
| 1180 |
+
"loss": 0.4263,
|
| 1181 |
+
"step": 14300
|
| 1182 |
+
},
|
| 1183 |
+
{
|
| 1184 |
+
"epoch": 97.96,
|
| 1185 |
+
"learning_rate": 1.8248031496062992e-06,
|
| 1186 |
+
"loss": 0.4436,
|
| 1187 |
+
"step": 14400
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 97.96,
|
| 1191 |
+
"eval_loss": 0.5867109894752502,
|
| 1192 |
+
"eval_runtime": 127.0812,
|
| 1193 |
+
"eval_samples_per_second": 16.486,
|
| 1194 |
+
"eval_steps_per_second": 1.031,
|
| 1195 |
+
"eval_wer": 0.34278302760793034,
|
| 1196 |
+
"step": 14400
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 98.64,
|
| 1200 |
+
"learning_rate": 1.234251968503937e-06,
|
| 1201 |
+
"loss": 0.447,
|
| 1202 |
+
"step": 14500
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 99.32,
|
| 1206 |
+
"learning_rate": 6.437007874015748e-07,
|
| 1207 |
+
"loss": 0.4325,
|
| 1208 |
+
"step": 14600
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 100.0,
|
| 1212 |
+
"learning_rate": 5.31496062992126e-08,
|
| 1213 |
+
"loss": 0.421,
|
| 1214 |
+
"step": 14700
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 100.0,
|
| 1218 |
+
"step": 14700,
|
| 1219 |
+
"total_flos": 1.7835239032674828e+20,
|
| 1220 |
+
"train_loss": 0.9074438333835732,
|
| 1221 |
+
"train_runtime": 47510.6321,
|
| 1222 |
+
"train_samples_per_second": 9.916,
|
| 1223 |
+
"train_steps_per_second": 0.309
|
| 1224 |
+
}
|
| 1225 |
+
],
|
| 1226 |
+
"max_steps": 14700,
|
| 1227 |
+
"num_train_epochs": 100,
|
| 1228 |
+
"total_flos": 1.7835239032674828e+20,
|
| 1229 |
+
"trial_name": null,
|
| 1230 |
+
"trial_params": null
|
| 1231 |
+
}
|