Commit
•
2254c3b
1
Parent(s):
4b9e48b
End of training
Browse files- README.md +1 -1
- all_results.json +15 -0
- eval_results.json +9 -0
- runs/Nov15_15-52-21_4398118e3bd3/events.out.tfevents.1731747029.4398118e3bd3.574.1 +3 -0
- train_results.json +9 -0
- trainer_state.json +1487 -0
README.md
CHANGED
|
@@ -16,7 +16,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 16 |
|
| 17 |
# whisper-large-v3-ft-btb-cv-ca-cy
|
| 18 |
|
| 19 |
-
This model is a fine-tuned version of [openai/whisper-large-v3](https://huggingface.co/openai/whisper-large-v3) on
|
| 20 |
It achieves the following results on the evaluation set:
|
| 21 |
- Loss: 0.3666
|
| 22 |
- Wer: 0.2773
|
|
|
|
| 16 |
|
| 17 |
# whisper-large-v3-ft-btb-cv-ca-cy
|
| 18 |
|
| 19 |
+
This model is a fine-tuned version of [openai/whisper-large-v3](https://huggingface.co/openai/whisper-large-v3) on the DewiBrynJones/banc-trawsgrifiadau-bangor-clean train main, DewiBrynJones/commonvoice_18_0_cy train+dev+other_with_excluded main, cymen-arfor/25awr train+dev main dataset.
|
| 20 |
It achieves the following results on the evaluation set:
|
| 21 |
- Loss: 0.3666
|
| 22 |
- Wer: 0.2773
|
all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.034174125305126,
|
| 3 |
+
"eval_loss": 0.36655664443969727,
|
| 4 |
+
"eval_runtime": 1777.7961,
|
| 5 |
+
"eval_samples": 3901,
|
| 6 |
+
"eval_samples_per_second": 2.194,
|
| 7 |
+
"eval_steps_per_second": 0.137,
|
| 8 |
+
"eval_wer": 0.27733507195981844,
|
| 9 |
+
"total_flos": 5.435045990970163e+20,
|
| 10 |
+
"train_loss": 0.38478662071228026,
|
| 11 |
+
"train_runtime": 58003.8059,
|
| 12 |
+
"train_samples": 78642,
|
| 13 |
+
"train_samples_per_second": 2.758,
|
| 14 |
+
"train_steps_per_second": 0.086
|
| 15 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.034174125305126,
|
| 3 |
+
"eval_loss": 0.36655664443969727,
|
| 4 |
+
"eval_runtime": 1777.7961,
|
| 5 |
+
"eval_samples": 3901,
|
| 6 |
+
"eval_samples_per_second": 2.194,
|
| 7 |
+
"eval_steps_per_second": 0.137,
|
| 8 |
+
"eval_wer": 0.27733507195981844
|
| 9 |
+
}
|
runs/Nov15_15-52-21_4398118e3bd3/events.out.tfevents.1731747029.4398118e3bd3.574.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cf3741ba9fcd42c8e461e5f297b0878315593e60f66e5b50ac6bf52678b816cd
|
| 3 |
+
size 406
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.034174125305126,
|
| 3 |
+
"total_flos": 5.435045990970163e+20,
|
| 4 |
+
"train_loss": 0.38478662071228026,
|
| 5 |
+
"train_runtime": 58003.8059,
|
| 6 |
+
"train_samples": 78642,
|
| 7 |
+
"train_samples_per_second": 2.758,
|
| 8 |
+
"train_steps_per_second": 0.086
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1487 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 2.034174125305126,
|
| 5 |
+
"eval_steps": 1000,
|
| 6 |
+
"global_step": 5000,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.01017087062652563,
|
| 13 |
+
"grad_norm": 16.184080123901367,
|
| 14 |
+
"learning_rate": 4.800000000000001e-07,
|
| 15 |
+
"loss": 1.6215,
|
| 16 |
+
"step": 25
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.02034174125305126,
|
| 20 |
+
"grad_norm": 14.08956241607666,
|
| 21 |
+
"learning_rate": 9.800000000000001e-07,
|
| 22 |
+
"loss": 1.2721,
|
| 23 |
+
"step": 50
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.030512611879576892,
|
| 27 |
+
"grad_norm": 11.49959945678711,
|
| 28 |
+
"learning_rate": 1.48e-06,
|
| 29 |
+
"loss": 0.9884,
|
| 30 |
+
"step": 75
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.04068348250610252,
|
| 34 |
+
"grad_norm": 8.895913124084473,
|
| 35 |
+
"learning_rate": 1.98e-06,
|
| 36 |
+
"loss": 0.8726,
|
| 37 |
+
"step": 100
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.050854353132628156,
|
| 41 |
+
"grad_norm": 9.52476692199707,
|
| 42 |
+
"learning_rate": 2.4800000000000004e-06,
|
| 43 |
+
"loss": 0.8062,
|
| 44 |
+
"step": 125
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.061025223759153785,
|
| 48 |
+
"grad_norm": 9.508084297180176,
|
| 49 |
+
"learning_rate": 2.9800000000000003e-06,
|
| 50 |
+
"loss": 0.7792,
|
| 51 |
+
"step": 150
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.07119609438567942,
|
| 55 |
+
"grad_norm": 9.352806091308594,
|
| 56 |
+
"learning_rate": 3.48e-06,
|
| 57 |
+
"loss": 0.7541,
|
| 58 |
+
"step": 175
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.08136696501220504,
|
| 62 |
+
"grad_norm": 10.18264102935791,
|
| 63 |
+
"learning_rate": 3.980000000000001e-06,
|
| 64 |
+
"loss": 0.6983,
|
| 65 |
+
"step": 200
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.09153783563873068,
|
| 69 |
+
"grad_norm": 10.840176582336426,
|
| 70 |
+
"learning_rate": 4.48e-06,
|
| 71 |
+
"loss": 0.7416,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.10170870626525631,
|
| 76 |
+
"grad_norm": 9.403505325317383,
|
| 77 |
+
"learning_rate": 4.980000000000001e-06,
|
| 78 |
+
"loss": 0.7154,
|
| 79 |
+
"step": 250
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.11187957689178193,
|
| 83 |
+
"grad_norm": 8.170647621154785,
|
| 84 |
+
"learning_rate": 5.480000000000001e-06,
|
| 85 |
+
"loss": 0.6564,
|
| 86 |
+
"step": 275
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.12205044751830757,
|
| 90 |
+
"grad_norm": 9.974289894104004,
|
| 91 |
+
"learning_rate": 5.98e-06,
|
| 92 |
+
"loss": 0.6401,
|
| 93 |
+
"step": 300
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.1322213181448332,
|
| 97 |
+
"grad_norm": 10.3145112991333,
|
| 98 |
+
"learning_rate": 6.480000000000001e-06,
|
| 99 |
+
"loss": 0.6564,
|
| 100 |
+
"step": 325
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.14239218877135884,
|
| 104 |
+
"grad_norm": 8.214014053344727,
|
| 105 |
+
"learning_rate": 6.98e-06,
|
| 106 |
+
"loss": 0.6191,
|
| 107 |
+
"step": 350
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.15256305939788445,
|
| 111 |
+
"grad_norm": 8.800423622131348,
|
| 112 |
+
"learning_rate": 7.48e-06,
|
| 113 |
+
"loss": 0.6492,
|
| 114 |
+
"step": 375
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.16273393002441008,
|
| 118 |
+
"grad_norm": 8.481072425842285,
|
| 119 |
+
"learning_rate": 7.980000000000002e-06,
|
| 120 |
+
"loss": 0.6295,
|
| 121 |
+
"step": 400
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.17290480065093572,
|
| 125 |
+
"grad_norm": 8.669546127319336,
|
| 126 |
+
"learning_rate": 8.48e-06,
|
| 127 |
+
"loss": 0.6167,
|
| 128 |
+
"step": 425
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.18307567127746135,
|
| 132 |
+
"grad_norm": 7.645444869995117,
|
| 133 |
+
"learning_rate": 8.98e-06,
|
| 134 |
+
"loss": 0.6021,
|
| 135 |
+
"step": 450
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.193246541903987,
|
| 139 |
+
"grad_norm": 11.719562530517578,
|
| 140 |
+
"learning_rate": 9.48e-06,
|
| 141 |
+
"loss": 0.6006,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.20341741253051263,
|
| 146 |
+
"grad_norm": 9.42712116241455,
|
| 147 |
+
"learning_rate": 9.980000000000001e-06,
|
| 148 |
+
"loss": 0.5777,
|
| 149 |
+
"step": 500
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.21358828315703823,
|
| 153 |
+
"grad_norm": 7.9745988845825195,
|
| 154 |
+
"learning_rate": 9.946666666666667e-06,
|
| 155 |
+
"loss": 0.5928,
|
| 156 |
+
"step": 525
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.22375915378356387,
|
| 160 |
+
"grad_norm": 7.762495994567871,
|
| 161 |
+
"learning_rate": 9.891111111111113e-06,
|
| 162 |
+
"loss": 0.5735,
|
| 163 |
+
"step": 550
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.2339300244100895,
|
| 167 |
+
"grad_norm": 8.57645320892334,
|
| 168 |
+
"learning_rate": 9.835555555555556e-06,
|
| 169 |
+
"loss": 0.5613,
|
| 170 |
+
"step": 575
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.24410089503661514,
|
| 174 |
+
"grad_norm": 7.481997489929199,
|
| 175 |
+
"learning_rate": 9.780000000000001e-06,
|
| 176 |
+
"loss": 0.5496,
|
| 177 |
+
"step": 600
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.25427176566314075,
|
| 181 |
+
"grad_norm": 6.589779376983643,
|
| 182 |
+
"learning_rate": 9.724444444444445e-06,
|
| 183 |
+
"loss": 0.5746,
|
| 184 |
+
"step": 625
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.2644426362896664,
|
| 188 |
+
"grad_norm": 8.590812683105469,
|
| 189 |
+
"learning_rate": 9.66888888888889e-06,
|
| 190 |
+
"loss": 0.5385,
|
| 191 |
+
"step": 650
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.274613506916192,
|
| 195 |
+
"grad_norm": 7.343937397003174,
|
| 196 |
+
"learning_rate": 9.613333333333335e-06,
|
| 197 |
+
"loss": 0.5432,
|
| 198 |
+
"step": 675
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.2847843775427177,
|
| 202 |
+
"grad_norm": 7.952889919281006,
|
| 203 |
+
"learning_rate": 9.557777777777777e-06,
|
| 204 |
+
"loss": 0.5213,
|
| 205 |
+
"step": 700
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.2949552481692433,
|
| 209 |
+
"grad_norm": 8.083431243896484,
|
| 210 |
+
"learning_rate": 9.502222222222223e-06,
|
| 211 |
+
"loss": 0.5367,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.3051261187957689,
|
| 216 |
+
"grad_norm": 7.172378063201904,
|
| 217 |
+
"learning_rate": 9.446666666666667e-06,
|
| 218 |
+
"loss": 0.5317,
|
| 219 |
+
"step": 750
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.31529698942229456,
|
| 223 |
+
"grad_norm": 7.955025672912598,
|
| 224 |
+
"learning_rate": 9.391111111111111e-06,
|
| 225 |
+
"loss": 0.5208,
|
| 226 |
+
"step": 775
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.32546786004882017,
|
| 230 |
+
"grad_norm": 7.041996479034424,
|
| 231 |
+
"learning_rate": 9.335555555555557e-06,
|
| 232 |
+
"loss": 0.5123,
|
| 233 |
+
"step": 800
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.33563873067534583,
|
| 237 |
+
"grad_norm": 7.7593607902526855,
|
| 238 |
+
"learning_rate": 9.280000000000001e-06,
|
| 239 |
+
"loss": 0.5019,
|
| 240 |
+
"step": 825
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.34580960130187144,
|
| 244 |
+
"grad_norm": 9.114460945129395,
|
| 245 |
+
"learning_rate": 9.224444444444445e-06,
|
| 246 |
+
"loss": 0.5189,
|
| 247 |
+
"step": 850
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.35598047192839705,
|
| 251 |
+
"grad_norm": 8.199577331542969,
|
| 252 |
+
"learning_rate": 9.171111111111112e-06,
|
| 253 |
+
"loss": 0.4844,
|
| 254 |
+
"step": 875
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.3661513425549227,
|
| 258 |
+
"grad_norm": 7.192209243774414,
|
| 259 |
+
"learning_rate": 9.115555555555556e-06,
|
| 260 |
+
"loss": 0.4867,
|
| 261 |
+
"step": 900
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.3763222131814483,
|
| 265 |
+
"grad_norm": 7.112555980682373,
|
| 266 |
+
"learning_rate": 9.060000000000001e-06,
|
| 267 |
+
"loss": 0.4673,
|
| 268 |
+
"step": 925
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.386493083807974,
|
| 272 |
+
"grad_norm": 6.723269462585449,
|
| 273 |
+
"learning_rate": 9.004444444444445e-06,
|
| 274 |
+
"loss": 0.4886,
|
| 275 |
+
"step": 950
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.3966639544344996,
|
| 279 |
+
"grad_norm": 6.772889137268066,
|
| 280 |
+
"learning_rate": 8.94888888888889e-06,
|
| 281 |
+
"loss": 0.5116,
|
| 282 |
+
"step": 975
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.40683482506102525,
|
| 286 |
+
"grad_norm": 7.985610485076904,
|
| 287 |
+
"learning_rate": 8.893333333333333e-06,
|
| 288 |
+
"loss": 0.5072,
|
| 289 |
+
"step": 1000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.40683482506102525,
|
| 293 |
+
"eval_loss": 0.5020337700843811,
|
| 294 |
+
"eval_runtime": 1789.7876,
|
| 295 |
+
"eval_samples_per_second": 2.18,
|
| 296 |
+
"eval_steps_per_second": 0.136,
|
| 297 |
+
"eval_wer": 0.35670820052158797,
|
| 298 |
+
"step": 1000
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.41700569568755086,
|
| 302 |
+
"grad_norm": 10.103032112121582,
|
| 303 |
+
"learning_rate": 8.83777777777778e-06,
|
| 304 |
+
"loss": 0.4951,
|
| 305 |
+
"step": 1025
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.42717656631407647,
|
| 309 |
+
"grad_norm": 6.815390586853027,
|
| 310 |
+
"learning_rate": 8.782222222222223e-06,
|
| 311 |
+
"loss": 0.4682,
|
| 312 |
+
"step": 1050
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.43734743694060213,
|
| 316 |
+
"grad_norm": 7.63904333114624,
|
| 317 |
+
"learning_rate": 8.726666666666667e-06,
|
| 318 |
+
"loss": 0.4842,
|
| 319 |
+
"step": 1075
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.44751830756712774,
|
| 323 |
+
"grad_norm": 7.750236988067627,
|
| 324 |
+
"learning_rate": 8.671111111111113e-06,
|
| 325 |
+
"loss": 0.4823,
|
| 326 |
+
"step": 1100
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.4576891781936534,
|
| 330 |
+
"grad_norm": 6.178125858306885,
|
| 331 |
+
"learning_rate": 8.615555555555555e-06,
|
| 332 |
+
"loss": 0.4764,
|
| 333 |
+
"step": 1125
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.467860048820179,
|
| 337 |
+
"grad_norm": 6.7011284828186035,
|
| 338 |
+
"learning_rate": 8.560000000000001e-06,
|
| 339 |
+
"loss": 0.487,
|
| 340 |
+
"step": 1150
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.4780309194467046,
|
| 344 |
+
"grad_norm": 6.630568504333496,
|
| 345 |
+
"learning_rate": 8.504444444444445e-06,
|
| 346 |
+
"loss": 0.444,
|
| 347 |
+
"step": 1175
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 0.4882017900732303,
|
| 351 |
+
"grad_norm": 7.386712074279785,
|
| 352 |
+
"learning_rate": 8.448888888888889e-06,
|
| 353 |
+
"loss": 0.4372,
|
| 354 |
+
"step": 1200
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.4983726606997559,
|
| 358 |
+
"grad_norm": 6.448077201843262,
|
| 359 |
+
"learning_rate": 8.393333333333335e-06,
|
| 360 |
+
"loss": 0.4474,
|
| 361 |
+
"step": 1225
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 0.5085435313262815,
|
| 365 |
+
"grad_norm": 6.787522315979004,
|
| 366 |
+
"learning_rate": 8.337777777777777e-06,
|
| 367 |
+
"loss": 0.4692,
|
| 368 |
+
"step": 1250
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 0.5187144019528072,
|
| 372 |
+
"grad_norm": 6.614894866943359,
|
| 373 |
+
"learning_rate": 8.282222222222223e-06,
|
| 374 |
+
"loss": 0.4401,
|
| 375 |
+
"step": 1275
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.5288852725793328,
|
| 379 |
+
"grad_norm": 7.09251594543457,
|
| 380 |
+
"learning_rate": 8.226666666666667e-06,
|
| 381 |
+
"loss": 0.4419,
|
| 382 |
+
"step": 1300
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 0.5390561432058584,
|
| 386 |
+
"grad_norm": 7.047774791717529,
|
| 387 |
+
"learning_rate": 8.171111111111113e-06,
|
| 388 |
+
"loss": 0.4323,
|
| 389 |
+
"step": 1325
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 0.549227013832384,
|
| 393 |
+
"grad_norm": 6.616914749145508,
|
| 394 |
+
"learning_rate": 8.115555555555557e-06,
|
| 395 |
+
"loss": 0.4354,
|
| 396 |
+
"step": 1350
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 0.5593978844589097,
|
| 400 |
+
"grad_norm": 8.16896915435791,
|
| 401 |
+
"learning_rate": 8.06e-06,
|
| 402 |
+
"loss": 0.4299,
|
| 403 |
+
"step": 1375
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 0.5695687550854354,
|
| 407 |
+
"grad_norm": 7.04884672164917,
|
| 408 |
+
"learning_rate": 8.004444444444445e-06,
|
| 409 |
+
"loss": 0.4477,
|
| 410 |
+
"step": 1400
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.5797396257119609,
|
| 414 |
+
"grad_norm": 6.052219867706299,
|
| 415 |
+
"learning_rate": 7.948888888888889e-06,
|
| 416 |
+
"loss": 0.4561,
|
| 417 |
+
"step": 1425
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.5899104963384866,
|
| 421 |
+
"grad_norm": 6.257965087890625,
|
| 422 |
+
"learning_rate": 7.893333333333335e-06,
|
| 423 |
+
"loss": 0.4343,
|
| 424 |
+
"step": 1450
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 0.6000813669650122,
|
| 428 |
+
"grad_norm": 7.182107925415039,
|
| 429 |
+
"learning_rate": 7.837777777777779e-06,
|
| 430 |
+
"loss": 0.4215,
|
| 431 |
+
"step": 1475
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.6102522375915378,
|
| 435 |
+
"grad_norm": 6.276819705963135,
|
| 436 |
+
"learning_rate": 7.782222222222223e-06,
|
| 437 |
+
"loss": 0.4188,
|
| 438 |
+
"step": 1500
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 0.6204231082180635,
|
| 442 |
+
"grad_norm": 6.533692359924316,
|
| 443 |
+
"learning_rate": 7.726666666666667e-06,
|
| 444 |
+
"loss": 0.4855,
|
| 445 |
+
"step": 1525
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 0.6305939788445891,
|
| 449 |
+
"grad_norm": 6.108315467834473,
|
| 450 |
+
"learning_rate": 7.67111111111111e-06,
|
| 451 |
+
"loss": 0.4255,
|
| 452 |
+
"step": 1550
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 0.6407648494711147,
|
| 456 |
+
"grad_norm": 6.741276741027832,
|
| 457 |
+
"learning_rate": 7.6155555555555564e-06,
|
| 458 |
+
"loss": 0.4529,
|
| 459 |
+
"step": 1575
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.6509357200976403,
|
| 463 |
+
"grad_norm": 6.165073871612549,
|
| 464 |
+
"learning_rate": 7.5600000000000005e-06,
|
| 465 |
+
"loss": 0.4541,
|
| 466 |
+
"step": 1600
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 0.661106590724166,
|
| 470 |
+
"grad_norm": 7.246621608734131,
|
| 471 |
+
"learning_rate": 7.504444444444445e-06,
|
| 472 |
+
"loss": 0.4073,
|
| 473 |
+
"step": 1625
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 0.6712774613506917,
|
| 477 |
+
"grad_norm": 7.50054407119751,
|
| 478 |
+
"learning_rate": 7.44888888888889e-06,
|
| 479 |
+
"loss": 0.4143,
|
| 480 |
+
"step": 1650
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 0.6814483319772172,
|
| 484 |
+
"grad_norm": 6.225817680358887,
|
| 485 |
+
"learning_rate": 7.393333333333333e-06,
|
| 486 |
+
"loss": 0.438,
|
| 487 |
+
"step": 1675
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 0.6916192026037429,
|
| 491 |
+
"grad_norm": 6.704942226409912,
|
| 492 |
+
"learning_rate": 7.337777777777778e-06,
|
| 493 |
+
"loss": 0.4209,
|
| 494 |
+
"step": 1700
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.7017900732302685,
|
| 498 |
+
"grad_norm": 6.276001930236816,
|
| 499 |
+
"learning_rate": 7.282222222222222e-06,
|
| 500 |
+
"loss": 0.4033,
|
| 501 |
+
"step": 1725
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.7119609438567941,
|
| 505 |
+
"grad_norm": 7.256411075592041,
|
| 506 |
+
"learning_rate": 7.226666666666667e-06,
|
| 507 |
+
"loss": 0.416,
|
| 508 |
+
"step": 1750
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 0.7221318144833198,
|
| 512 |
+
"grad_norm": 6.347288608551025,
|
| 513 |
+
"learning_rate": 7.171111111111112e-06,
|
| 514 |
+
"loss": 0.4373,
|
| 515 |
+
"step": 1775
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 0.7323026851098454,
|
| 519 |
+
"grad_norm": 6.241955757141113,
|
| 520 |
+
"learning_rate": 7.115555555555557e-06,
|
| 521 |
+
"loss": 0.403,
|
| 522 |
+
"step": 1800
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.7424735557363711,
|
| 526 |
+
"grad_norm": 5.548867702484131,
|
| 527 |
+
"learning_rate": 7.06e-06,
|
| 528 |
+
"loss": 0.4034,
|
| 529 |
+
"step": 1825
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 0.7526444263628966,
|
| 533 |
+
"grad_norm": 7.0312113761901855,
|
| 534 |
+
"learning_rate": 7.004444444444445e-06,
|
| 535 |
+
"loss": 0.4037,
|
| 536 |
+
"step": 1850
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 0.7628152969894223,
|
| 540 |
+
"grad_norm": 7.7608256340026855,
|
| 541 |
+
"learning_rate": 6.948888888888889e-06,
|
| 542 |
+
"loss": 0.3939,
|
| 543 |
+
"step": 1875
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 0.772986167615948,
|
| 547 |
+
"grad_norm": 6.072271823883057,
|
| 548 |
+
"learning_rate": 6.893333333333334e-06,
|
| 549 |
+
"loss": 0.4067,
|
| 550 |
+
"step": 1900
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 0.7831570382424735,
|
| 554 |
+
"grad_norm": 6.9571123123168945,
|
| 555 |
+
"learning_rate": 6.837777777777779e-06,
|
| 556 |
+
"loss": 0.4107,
|
| 557 |
+
"step": 1925
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 0.7933279088689992,
|
| 561 |
+
"grad_norm": 5.884681224822998,
|
| 562 |
+
"learning_rate": 6.782222222222222e-06,
|
| 563 |
+
"loss": 0.3995,
|
| 564 |
+
"step": 1950
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 0.8034987794955248,
|
| 568 |
+
"grad_norm": 5.850441932678223,
|
| 569 |
+
"learning_rate": 6.726666666666667e-06,
|
| 570 |
+
"loss": 0.4017,
|
| 571 |
+
"step": 1975
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 0.8136696501220505,
|
| 575 |
+
"grad_norm": 6.021271705627441,
|
| 576 |
+
"learning_rate": 6.671111111111112e-06,
|
| 577 |
+
"loss": 0.3708,
|
| 578 |
+
"step": 2000
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 0.8136696501220505,
|
| 582 |
+
"eval_loss": 0.42604002356529236,
|
| 583 |
+
"eval_runtime": 1798.3084,
|
| 584 |
+
"eval_samples_per_second": 2.169,
|
| 585 |
+
"eval_steps_per_second": 0.136,
|
| 586 |
+
"eval_wer": 0.32584758041147494,
|
| 587 |
+
"step": 2000
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 0.823840520748576,
|
| 591 |
+
"grad_norm": 6.479699611663818,
|
| 592 |
+
"learning_rate": 6.615555555555556e-06,
|
| 593 |
+
"loss": 0.3914,
|
| 594 |
+
"step": 2025
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 0.8340113913751017,
|
| 598 |
+
"grad_norm": 6.589993476867676,
|
| 599 |
+
"learning_rate": 6.560000000000001e-06,
|
| 600 |
+
"loss": 0.401,
|
| 601 |
+
"step": 2050
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 0.8441822620016274,
|
| 605 |
+
"grad_norm": 5.026104927062988,
|
| 606 |
+
"learning_rate": 6.504444444444446e-06,
|
| 607 |
+
"loss": 0.38,
|
| 608 |
+
"step": 2075
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 0.8543531326281529,
|
| 612 |
+
"grad_norm": 6.458845138549805,
|
| 613 |
+
"learning_rate": 6.448888888888889e-06,
|
| 614 |
+
"loss": 0.3885,
|
| 615 |
+
"step": 2100
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 0.8645240032546786,
|
| 619 |
+
"grad_norm": 6.570550441741943,
|
| 620 |
+
"learning_rate": 6.393333333333334e-06,
|
| 621 |
+
"loss": 0.3992,
|
| 622 |
+
"step": 2125
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 0.8746948738812043,
|
| 626 |
+
"grad_norm": 6.160183906555176,
|
| 627 |
+
"learning_rate": 6.3377777777777786e-06,
|
| 628 |
+
"loss": 0.385,
|
| 629 |
+
"step": 2150
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 0.8848657445077298,
|
| 633 |
+
"grad_norm": 5.925890922546387,
|
| 634 |
+
"learning_rate": 6.282222222222223e-06,
|
| 635 |
+
"loss": 0.4082,
|
| 636 |
+
"step": 2175
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 0.8950366151342555,
|
| 640 |
+
"grad_norm": 6.508133888244629,
|
| 641 |
+
"learning_rate": 6.2266666666666675e-06,
|
| 642 |
+
"loss": 0.4043,
|
| 643 |
+
"step": 2200
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 0.9052074857607811,
|
| 647 |
+
"grad_norm": 6.300657272338867,
|
| 648 |
+
"learning_rate": 6.171111111111112e-06,
|
| 649 |
+
"loss": 0.3729,
|
| 650 |
+
"step": 2225
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 0.9153783563873068,
|
| 654 |
+
"grad_norm": 5.284305572509766,
|
| 655 |
+
"learning_rate": 6.1155555555555555e-06,
|
| 656 |
+
"loss": 0.4046,
|
| 657 |
+
"step": 2250
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 0.9255492270138324,
|
| 661 |
+
"grad_norm": 5.809286594390869,
|
| 662 |
+
"learning_rate": 6.0600000000000004e-06,
|
| 663 |
+
"loss": 0.3864,
|
| 664 |
+
"step": 2275
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 0.935720097640358,
|
| 668 |
+
"grad_norm": 6.667562007904053,
|
| 669 |
+
"learning_rate": 6.004444444444445e-06,
|
| 670 |
+
"loss": 0.3929,
|
| 671 |
+
"step": 2300
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 0.9458909682668837,
|
| 675 |
+
"grad_norm": 6.4545392990112305,
|
| 676 |
+
"learning_rate": 5.948888888888889e-06,
|
| 677 |
+
"loss": 0.3975,
|
| 678 |
+
"step": 2325
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 0.9560618388934092,
|
| 682 |
+
"grad_norm": 5.876199245452881,
|
| 683 |
+
"learning_rate": 5.893333333333334e-06,
|
| 684 |
+
"loss": 0.3769,
|
| 685 |
+
"step": 2350
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 0.9662327095199349,
|
| 689 |
+
"grad_norm": 5.008429527282715,
|
| 690 |
+
"learning_rate": 5.837777777777777e-06,
|
| 691 |
+
"loss": 0.4031,
|
| 692 |
+
"step": 2375
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 0.9764035801464606,
|
| 696 |
+
"grad_norm": 7.304080963134766,
|
| 697 |
+
"learning_rate": 5.782222222222222e-06,
|
| 698 |
+
"loss": 0.3804,
|
| 699 |
+
"step": 2400
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.9865744507729862,
|
| 703 |
+
"grad_norm": 7.11293888092041,
|
| 704 |
+
"learning_rate": 5.726666666666667e-06,
|
| 705 |
+
"loss": 0.3849,
|
| 706 |
+
"step": 2425
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 0.9967453213995118,
|
| 710 |
+
"grad_norm": 6.156993389129639,
|
| 711 |
+
"learning_rate": 5.671111111111112e-06,
|
| 712 |
+
"loss": 0.3804,
|
| 713 |
+
"step": 2450
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 1.0069161920260374,
|
| 717 |
+
"grad_norm": 4.09175968170166,
|
| 718 |
+
"learning_rate": 5.615555555555556e-06,
|
| 719 |
+
"loss": 0.3138,
|
| 720 |
+
"step": 2475
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 1.017087062652563,
|
| 724 |
+
"grad_norm": 5.104438304901123,
|
| 725 |
+
"learning_rate": 5.560000000000001e-06,
|
| 726 |
+
"loss": 0.2747,
|
| 727 |
+
"step": 2500
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 1.0272579332790888,
|
| 731 |
+
"grad_norm": 5.911495685577393,
|
| 732 |
+
"learning_rate": 5.504444444444444e-06,
|
| 733 |
+
"loss": 0.2825,
|
| 734 |
+
"step": 2525
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 1.0374288039056143,
|
| 738 |
+
"grad_norm": 5.619592189788818,
|
| 739 |
+
"learning_rate": 5.448888888888889e-06,
|
| 740 |
+
"loss": 0.2991,
|
| 741 |
+
"step": 2550
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 1.0475996745321399,
|
| 745 |
+
"grad_norm": 5.471043109893799,
|
| 746 |
+
"learning_rate": 5.393333333333334e-06,
|
| 747 |
+
"loss": 0.2766,
|
| 748 |
+
"step": 2575
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 1.0577705451586656,
|
| 752 |
+
"grad_norm": 4.353448867797852,
|
| 753 |
+
"learning_rate": 5.337777777777779e-06,
|
| 754 |
+
"loss": 0.2883,
|
| 755 |
+
"step": 2600
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 1.0679414157851912,
|
| 759 |
+
"grad_norm": 4.982877731323242,
|
| 760 |
+
"learning_rate": 5.282222222222223e-06,
|
| 761 |
+
"loss": 0.2888,
|
| 762 |
+
"step": 2625
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 1.0781122864117167,
|
| 766 |
+
"grad_norm": 5.228238582611084,
|
| 767 |
+
"learning_rate": 5.226666666666667e-06,
|
| 768 |
+
"loss": 0.2802,
|
| 769 |
+
"step": 2650
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 1.0882831570382425,
|
| 773 |
+
"grad_norm": 5.883535385131836,
|
| 774 |
+
"learning_rate": 5.171111111111111e-06,
|
| 775 |
+
"loss": 0.2779,
|
| 776 |
+
"step": 2675
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 1.098454027664768,
|
| 780 |
+
"grad_norm": 4.599557876586914,
|
| 781 |
+
"learning_rate": 5.115555555555556e-06,
|
| 782 |
+
"loss": 0.2817,
|
| 783 |
+
"step": 2700
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 1.1086248982912936,
|
| 787 |
+
"grad_norm": 5.081583023071289,
|
| 788 |
+
"learning_rate": 5.060000000000001e-06,
|
| 789 |
+
"loss": 0.2804,
|
| 790 |
+
"step": 2725
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 1.1187957689178194,
|
| 794 |
+
"grad_norm": 5.387673854827881,
|
| 795 |
+
"learning_rate": 5.004444444444445e-06,
|
| 796 |
+
"loss": 0.2676,
|
| 797 |
+
"step": 2750
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 1.128966639544345,
|
| 801 |
+
"grad_norm": 5.010275840759277,
|
| 802 |
+
"learning_rate": 4.94888888888889e-06,
|
| 803 |
+
"loss": 0.2752,
|
| 804 |
+
"step": 2775
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 1.1391375101708707,
|
| 808 |
+
"grad_norm": 5.828437805175781,
|
| 809 |
+
"learning_rate": 4.893333333333334e-06,
|
| 810 |
+
"loss": 0.2795,
|
| 811 |
+
"step": 2800
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 1.1493083807973963,
|
| 815 |
+
"grad_norm": 5.313021183013916,
|
| 816 |
+
"learning_rate": 4.837777777777778e-06,
|
| 817 |
+
"loss": 0.2876,
|
| 818 |
+
"step": 2825
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 1.1594792514239218,
|
| 822 |
+
"grad_norm": 5.790019512176514,
|
| 823 |
+
"learning_rate": 4.7822222222222226e-06,
|
| 824 |
+
"loss": 0.283,
|
| 825 |
+
"step": 2850
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 1.1696501220504476,
|
| 829 |
+
"grad_norm": 5.2214250564575195,
|
| 830 |
+
"learning_rate": 4.7266666666666674e-06,
|
| 831 |
+
"loss": 0.2762,
|
| 832 |
+
"step": 2875
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 1.1798209926769732,
|
| 836 |
+
"grad_norm": 6.338456630706787,
|
| 837 |
+
"learning_rate": 4.6711111111111115e-06,
|
| 838 |
+
"loss": 0.2588,
|
| 839 |
+
"step": 2900
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 1.1899918633034987,
|
| 843 |
+
"grad_norm": 6.858678340911865,
|
| 844 |
+
"learning_rate": 4.6155555555555555e-06,
|
| 845 |
+
"loss": 0.2724,
|
| 846 |
+
"step": 2925
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 1.2001627339300245,
|
| 850 |
+
"grad_norm": 5.54526948928833,
|
| 851 |
+
"learning_rate": 4.56e-06,
|
| 852 |
+
"loss": 0.2893,
|
| 853 |
+
"step": 2950
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 1.21033360455655,
|
| 857 |
+
"grad_norm": 5.165172100067139,
|
| 858 |
+
"learning_rate": 4.504444444444444e-06,
|
| 859 |
+
"loss": 0.2751,
|
| 860 |
+
"step": 2975
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 1.2205044751830756,
|
| 864 |
+
"grad_norm": 6.359204292297363,
|
| 865 |
+
"learning_rate": 4.448888888888889e-06,
|
| 866 |
+
"loss": 0.2599,
|
| 867 |
+
"step": 3000
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 1.2205044751830756,
|
| 871 |
+
"eval_loss": 0.39731866121292114,
|
| 872 |
+
"eval_runtime": 1796.0978,
|
| 873 |
+
"eval_samples_per_second": 2.172,
|
| 874 |
+
"eval_steps_per_second": 0.136,
|
| 875 |
+
"eval_wer": 0.30034772529701537,
|
| 876 |
+
"step": 3000
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 1.2306753458096014,
|
| 880 |
+
"grad_norm": 4.958346366882324,
|
| 881 |
+
"learning_rate": 4.393333333333334e-06,
|
| 882 |
+
"loss": 0.2754,
|
| 883 |
+
"step": 3025
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 1.240846216436127,
|
| 887 |
+
"grad_norm": 7.284126281738281,
|
| 888 |
+
"learning_rate": 4.337777777777778e-06,
|
| 889 |
+
"loss": 0.2764,
|
| 890 |
+
"step": 3050
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 1.2510170870626527,
|
| 894 |
+
"grad_norm": 5.106224536895752,
|
| 895 |
+
"learning_rate": 4.282222222222222e-06,
|
| 896 |
+
"loss": 0.2679,
|
| 897 |
+
"step": 3075
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 1.2611879576891782,
|
| 901 |
+
"grad_norm": 4.638324737548828,
|
| 902 |
+
"learning_rate": 4.226666666666667e-06,
|
| 903 |
+
"loss": 0.2678,
|
| 904 |
+
"step": 3100
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 1.2713588283157038,
|
| 908 |
+
"grad_norm": 5.0809245109558105,
|
| 909 |
+
"learning_rate": 4.171111111111111e-06,
|
| 910 |
+
"loss": 0.2679,
|
| 911 |
+
"step": 3125
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 1.2815296989422293,
|
| 915 |
+
"grad_norm": 5.177508354187012,
|
| 916 |
+
"learning_rate": 4.115555555555556e-06,
|
| 917 |
+
"loss": 0.2603,
|
| 918 |
+
"step": 3150
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 1.2917005695687551,
|
| 922 |
+
"grad_norm": 4.664310932159424,
|
| 923 |
+
"learning_rate": 4.060000000000001e-06,
|
| 924 |
+
"loss": 0.2659,
|
| 925 |
+
"step": 3175
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 1.3018714401952807,
|
| 929 |
+
"grad_norm": 5.230463981628418,
|
| 930 |
+
"learning_rate": 4.004444444444445e-06,
|
| 931 |
+
"loss": 0.271,
|
| 932 |
+
"step": 3200
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 1.3120423108218064,
|
| 936 |
+
"grad_norm": 5.396975994110107,
|
| 937 |
+
"learning_rate": 3.948888888888889e-06,
|
| 938 |
+
"loss": 0.2653,
|
| 939 |
+
"step": 3225
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 1.322213181448332,
|
| 943 |
+
"grad_norm": 4.997494220733643,
|
| 944 |
+
"learning_rate": 3.893333333333333e-06,
|
| 945 |
+
"loss": 0.2678,
|
| 946 |
+
"step": 3250
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 1.3323840520748575,
|
| 950 |
+
"grad_norm": 6.467133045196533,
|
| 951 |
+
"learning_rate": 3.837777777777778e-06,
|
| 952 |
+
"loss": 0.2628,
|
| 953 |
+
"step": 3275
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 1.342554922701383,
|
| 957 |
+
"grad_norm": 4.6464338302612305,
|
| 958 |
+
"learning_rate": 3.782222222222223e-06,
|
| 959 |
+
"loss": 0.2819,
|
| 960 |
+
"step": 3300
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 1.3527257933279089,
|
| 964 |
+
"grad_norm": 5.477710723876953,
|
| 965 |
+
"learning_rate": 3.726666666666667e-06,
|
| 966 |
+
"loss": 0.2694,
|
| 967 |
+
"step": 3325
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 1.3628966639544344,
|
| 971 |
+
"grad_norm": 5.745607376098633,
|
| 972 |
+
"learning_rate": 3.6711111111111113e-06,
|
| 973 |
+
"loss": 0.2731,
|
| 974 |
+
"step": 3350
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 1.3730675345809602,
|
| 978 |
+
"grad_norm": 4.110312461853027,
|
| 979 |
+
"learning_rate": 3.615555555555556e-06,
|
| 980 |
+
"loss": 0.2726,
|
| 981 |
+
"step": 3375
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 1.3832384052074858,
|
| 985 |
+
"grad_norm": 5.377216815948486,
|
| 986 |
+
"learning_rate": 3.5600000000000002e-06,
|
| 987 |
+
"loss": 0.2677,
|
| 988 |
+
"step": 3400
|
| 989 |
+
},
|
| 990 |
+
{
|
| 991 |
+
"epoch": 1.3934092758340113,
|
| 992 |
+
"grad_norm": 5.4512763023376465,
|
| 993 |
+
"learning_rate": 3.5044444444444447e-06,
|
| 994 |
+
"loss": 0.2621,
|
| 995 |
+
"step": 3425
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 1.403580146460537,
|
| 999 |
+
"grad_norm": 5.338203430175781,
|
| 1000 |
+
"learning_rate": 3.4488888888888896e-06,
|
| 1001 |
+
"loss": 0.2731,
|
| 1002 |
+
"step": 3450
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 1.4137510170870626,
|
| 1006 |
+
"grad_norm": 5.551027774810791,
|
| 1007 |
+
"learning_rate": 3.3933333333333336e-06,
|
| 1008 |
+
"loss": 0.2821,
|
| 1009 |
+
"step": 3475
|
| 1010 |
+
},
|
| 1011 |
+
{
|
| 1012 |
+
"epoch": 1.4239218877135884,
|
| 1013 |
+
"grad_norm": 6.804419040679932,
|
| 1014 |
+
"learning_rate": 3.337777777777778e-06,
|
| 1015 |
+
"loss": 0.2652,
|
| 1016 |
+
"step": 3500
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 1.434092758340114,
|
| 1020 |
+
"grad_norm": 5.15053653717041,
|
| 1021 |
+
"learning_rate": 3.282222222222223e-06,
|
| 1022 |
+
"loss": 0.2626,
|
| 1023 |
+
"step": 3525
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 1.4442636289666395,
|
| 1027 |
+
"grad_norm": 4.987846851348877,
|
| 1028 |
+
"learning_rate": 3.226666666666667e-06,
|
| 1029 |
+
"loss": 0.2558,
|
| 1030 |
+
"step": 3550
|
| 1031 |
+
},
|
| 1032 |
+
{
|
| 1033 |
+
"epoch": 1.454434499593165,
|
| 1034 |
+
"grad_norm": 5.460192680358887,
|
| 1035 |
+
"learning_rate": 3.1711111111111114e-06,
|
| 1036 |
+
"loss": 0.2688,
|
| 1037 |
+
"step": 3575
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 1.4646053702196908,
|
| 1041 |
+
"grad_norm": 5.20620059967041,
|
| 1042 |
+
"learning_rate": 3.1155555555555555e-06,
|
| 1043 |
+
"loss": 0.2541,
|
| 1044 |
+
"step": 3600
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 1.4747762408462164,
|
| 1048 |
+
"grad_norm": 5.177892208099365,
|
| 1049 |
+
"learning_rate": 3.0600000000000003e-06,
|
| 1050 |
+
"loss": 0.2546,
|
| 1051 |
+
"step": 3625
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 1.4849471114727422,
|
| 1055 |
+
"grad_norm": 5.275447845458984,
|
| 1056 |
+
"learning_rate": 3.004444444444445e-06,
|
| 1057 |
+
"loss": 0.2487,
|
| 1058 |
+
"step": 3650
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 1.4951179820992677,
|
| 1062 |
+
"grad_norm": 4.815367698669434,
|
| 1063 |
+
"learning_rate": 2.948888888888889e-06,
|
| 1064 |
+
"loss": 0.2505,
|
| 1065 |
+
"step": 3675
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 1.5052888527257933,
|
| 1069 |
+
"grad_norm": 4.586023330688477,
|
| 1070 |
+
"learning_rate": 2.8933333333333337e-06,
|
| 1071 |
+
"loss": 0.2385,
|
| 1072 |
+
"step": 3700
|
| 1073 |
+
},
|
| 1074 |
+
{
|
| 1075 |
+
"epoch": 1.5154597233523188,
|
| 1076 |
+
"grad_norm": 5.920341491699219,
|
| 1077 |
+
"learning_rate": 2.837777777777778e-06,
|
| 1078 |
+
"loss": 0.2502,
|
| 1079 |
+
"step": 3725
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 1.5256305939788446,
|
| 1083 |
+
"grad_norm": 5.697859764099121,
|
| 1084 |
+
"learning_rate": 2.7822222222222222e-06,
|
| 1085 |
+
"loss": 0.2574,
|
| 1086 |
+
"step": 3750
|
| 1087 |
+
},
|
| 1088 |
+
{
|
| 1089 |
+
"epoch": 1.5358014646053704,
|
| 1090 |
+
"grad_norm": 4.68679666519165,
|
| 1091 |
+
"learning_rate": 2.726666666666667e-06,
|
| 1092 |
+
"loss": 0.2543,
|
| 1093 |
+
"step": 3775
|
| 1094 |
+
},
|
| 1095 |
+
{
|
| 1096 |
+
"epoch": 1.545972335231896,
|
| 1097 |
+
"grad_norm": 5.064055442810059,
|
| 1098 |
+
"learning_rate": 2.6711111111111116e-06,
|
| 1099 |
+
"loss": 0.2639,
|
| 1100 |
+
"step": 3800
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 1.5561432058584215,
|
| 1104 |
+
"grad_norm": 5.656905651092529,
|
| 1105 |
+
"learning_rate": 2.6155555555555556e-06,
|
| 1106 |
+
"loss": 0.2639,
|
| 1107 |
+
"step": 3825
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 1.566314076484947,
|
| 1111 |
+
"grad_norm": 5.397951126098633,
|
| 1112 |
+
"learning_rate": 2.56e-06,
|
| 1113 |
+
"loss": 0.271,
|
| 1114 |
+
"step": 3850
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 1.5764849471114726,
|
| 1118 |
+
"grad_norm": 6.378788948059082,
|
| 1119 |
+
"learning_rate": 2.504444444444445e-06,
|
| 1120 |
+
"loss": 0.2542,
|
| 1121 |
+
"step": 3875
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 1.5866558177379984,
|
| 1125 |
+
"grad_norm": 6.6806960105896,
|
| 1126 |
+
"learning_rate": 2.448888888888889e-06,
|
| 1127 |
+
"loss": 0.2471,
|
| 1128 |
+
"step": 3900
|
| 1129 |
+
},
|
| 1130 |
+
{
|
| 1131 |
+
"epoch": 1.5968266883645241,
|
| 1132 |
+
"grad_norm": 4.837161064147949,
|
| 1133 |
+
"learning_rate": 2.3933333333333334e-06,
|
| 1134 |
+
"loss": 0.2497,
|
| 1135 |
+
"step": 3925
|
| 1136 |
+
},
|
| 1137 |
+
{
|
| 1138 |
+
"epoch": 1.6069975589910497,
|
| 1139 |
+
"grad_norm": 5.817267894744873,
|
| 1140 |
+
"learning_rate": 2.337777777777778e-06,
|
| 1141 |
+
"loss": 0.2583,
|
| 1142 |
+
"step": 3950
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 1.6171684296175752,
|
| 1146 |
+
"grad_norm": 5.383573532104492,
|
| 1147 |
+
"learning_rate": 2.2822222222222223e-06,
|
| 1148 |
+
"loss": 0.2535,
|
| 1149 |
+
"step": 3975
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 1.6273393002441008,
|
| 1153 |
+
"grad_norm": 4.556544780731201,
|
| 1154 |
+
"learning_rate": 2.226666666666667e-06,
|
| 1155 |
+
"loss": 0.2618,
|
| 1156 |
+
"step": 4000
|
| 1157 |
+
},
|
| 1158 |
+
{
|
| 1159 |
+
"epoch": 1.6273393002441008,
|
| 1160 |
+
"eval_loss": 0.37833017110824585,
|
| 1161 |
+
"eval_runtime": 1788.9834,
|
| 1162 |
+
"eval_samples_per_second": 2.181,
|
| 1163 |
+
"eval_steps_per_second": 0.136,
|
| 1164 |
+
"eval_wer": 0.2904713609581764,
|
| 1165 |
+
"step": 4000
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 1.6375101708706266,
|
| 1169 |
+
"grad_norm": 5.119606018066406,
|
| 1170 |
+
"learning_rate": 2.1711111111111113e-06,
|
| 1171 |
+
"loss": 0.2613,
|
| 1172 |
+
"step": 4025
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 1.647681041497152,
|
| 1176 |
+
"grad_norm": 5.435574054718018,
|
| 1177 |
+
"learning_rate": 2.1155555555555557e-06,
|
| 1178 |
+
"loss": 0.243,
|
| 1179 |
+
"step": 4050
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 1.6578519121236779,
|
| 1183 |
+
"grad_norm": 5.2191643714904785,
|
| 1184 |
+
"learning_rate": 2.06e-06,
|
| 1185 |
+
"loss": 0.2563,
|
| 1186 |
+
"step": 4075
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 1.6680227827502034,
|
| 1190 |
+
"grad_norm": 4.851512908935547,
|
| 1191 |
+
"learning_rate": 2.0044444444444446e-06,
|
| 1192 |
+
"loss": 0.2608,
|
| 1193 |
+
"step": 4100
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 1.678193653376729,
|
| 1197 |
+
"grad_norm": 5.570041179656982,
|
| 1198 |
+
"learning_rate": 1.948888888888889e-06,
|
| 1199 |
+
"loss": 0.2572,
|
| 1200 |
+
"step": 4125
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 1.6883645240032545,
|
| 1204 |
+
"grad_norm": 4.914021968841553,
|
| 1205 |
+
"learning_rate": 1.8933333333333333e-06,
|
| 1206 |
+
"loss": 0.2502,
|
| 1207 |
+
"step": 4150
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 1.6985353946297803,
|
| 1211 |
+
"grad_norm": 4.91426944732666,
|
| 1212 |
+
"learning_rate": 1.837777777777778e-06,
|
| 1213 |
+
"loss": 0.2484,
|
| 1214 |
+
"step": 4175
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 1.708706265256306,
|
| 1218 |
+
"grad_norm": 5.054934501647949,
|
| 1219 |
+
"learning_rate": 1.7822222222222225e-06,
|
| 1220 |
+
"loss": 0.2543,
|
| 1221 |
+
"step": 4200
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 1.7188771358828316,
|
| 1225 |
+
"grad_norm": 5.295421600341797,
|
| 1226 |
+
"learning_rate": 1.7266666666666667e-06,
|
| 1227 |
+
"loss": 0.2479,
|
| 1228 |
+
"step": 4225
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 1.7290480065093572,
|
| 1232 |
+
"grad_norm": 4.494225978851318,
|
| 1233 |
+
"learning_rate": 1.6711111111111112e-06,
|
| 1234 |
+
"loss": 0.2708,
|
| 1235 |
+
"step": 4250
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 1.7392188771358827,
|
| 1239 |
+
"grad_norm": 5.637074947357178,
|
| 1240 |
+
"learning_rate": 1.6155555555555559e-06,
|
| 1241 |
+
"loss": 0.2536,
|
| 1242 |
+
"step": 4275
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 1.7493897477624083,
|
| 1246 |
+
"grad_norm": 4.7007222175598145,
|
| 1247 |
+
"learning_rate": 1.56e-06,
|
| 1248 |
+
"loss": 0.2445,
|
| 1249 |
+
"step": 4300
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 1.759560618388934,
|
| 1253 |
+
"grad_norm": 5.922358989715576,
|
| 1254 |
+
"learning_rate": 1.5044444444444446e-06,
|
| 1255 |
+
"loss": 0.2629,
|
| 1256 |
+
"step": 4325
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 1.7697314890154598,
|
| 1260 |
+
"grad_norm": 5.818343162536621,
|
| 1261 |
+
"learning_rate": 1.4488888888888892e-06,
|
| 1262 |
+
"loss": 0.2462,
|
| 1263 |
+
"step": 4350
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 1.7799023596419854,
|
| 1267 |
+
"grad_norm": 4.40110969543457,
|
| 1268 |
+
"learning_rate": 1.3933333333333335e-06,
|
| 1269 |
+
"loss": 0.2389,
|
| 1270 |
+
"step": 4375
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 1.790073230268511,
|
| 1274 |
+
"grad_norm": 6.067351818084717,
|
| 1275 |
+
"learning_rate": 1.337777777777778e-06,
|
| 1276 |
+
"loss": 0.2662,
|
| 1277 |
+
"step": 4400
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 1.8002441008950365,
|
| 1281 |
+
"grad_norm": 4.228072166442871,
|
| 1282 |
+
"learning_rate": 1.2822222222222222e-06,
|
| 1283 |
+
"loss": 0.2611,
|
| 1284 |
+
"step": 4425
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 1.8104149715215623,
|
| 1288 |
+
"grad_norm": 5.135393142700195,
|
| 1289 |
+
"learning_rate": 1.2266666666666666e-06,
|
| 1290 |
+
"loss": 0.2607,
|
| 1291 |
+
"step": 4450
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 1.8205858421480878,
|
| 1295 |
+
"grad_norm": 5.195592403411865,
|
| 1296 |
+
"learning_rate": 1.171111111111111e-06,
|
| 1297 |
+
"loss": 0.2534,
|
| 1298 |
+
"step": 4475
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 1.8307567127746136,
|
| 1302 |
+
"grad_norm": 4.900667667388916,
|
| 1303 |
+
"learning_rate": 1.1155555555555558e-06,
|
| 1304 |
+
"loss": 0.2558,
|
| 1305 |
+
"step": 4500
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 1.8409275834011392,
|
| 1309 |
+
"grad_norm": 4.837711811065674,
|
| 1310 |
+
"learning_rate": 1.06e-06,
|
| 1311 |
+
"loss": 0.2476,
|
| 1312 |
+
"step": 4525
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 1.8510984540276647,
|
| 1316 |
+
"grad_norm": 5.072052955627441,
|
| 1317 |
+
"learning_rate": 1.0044444444444445e-06,
|
| 1318 |
+
"loss": 0.2333,
|
| 1319 |
+
"step": 4550
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 1.8612693246541903,
|
| 1323 |
+
"grad_norm": 5.385496139526367,
|
| 1324 |
+
"learning_rate": 9.488888888888889e-07,
|
| 1325 |
+
"loss": 0.2453,
|
| 1326 |
+
"step": 4575
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 1.871440195280716,
|
| 1330 |
+
"grad_norm": 5.36069393157959,
|
| 1331 |
+
"learning_rate": 8.933333333333334e-07,
|
| 1332 |
+
"loss": 0.2477,
|
| 1333 |
+
"step": 4600
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 1.8816110659072418,
|
| 1337 |
+
"grad_norm": 4.781914710998535,
|
| 1338 |
+
"learning_rate": 8.37777777777778e-07,
|
| 1339 |
+
"loss": 0.2494,
|
| 1340 |
+
"step": 4625
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 1.8917819365337674,
|
| 1344 |
+
"grad_norm": 5.366285800933838,
|
| 1345 |
+
"learning_rate": 7.822222222222223e-07,
|
| 1346 |
+
"loss": 0.2519,
|
| 1347 |
+
"step": 4650
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 1.901952807160293,
|
| 1351 |
+
"grad_norm": 4.479983806610107,
|
| 1352 |
+
"learning_rate": 7.266666666666668e-07,
|
| 1353 |
+
"loss": 0.241,
|
| 1354 |
+
"step": 4675
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 1.9121236777868185,
|
| 1358 |
+
"grad_norm": 5.053223133087158,
|
| 1359 |
+
"learning_rate": 6.711111111111111e-07,
|
| 1360 |
+
"loss": 0.2462,
|
| 1361 |
+
"step": 4700
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 1.922294548413344,
|
| 1365 |
+
"grad_norm": 4.833332061767578,
|
| 1366 |
+
"learning_rate": 6.155555555555556e-07,
|
| 1367 |
+
"loss": 0.2698,
|
| 1368 |
+
"step": 4725
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 1.9324654190398698,
|
| 1372 |
+
"grad_norm": 3.715808629989624,
|
| 1373 |
+
"learning_rate": 5.6e-07,
|
| 1374 |
+
"loss": 0.2338,
|
| 1375 |
+
"step": 4750
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 1.9426362896663956,
|
| 1379 |
+
"grad_norm": 5.295703887939453,
|
| 1380 |
+
"learning_rate": 5.044444444444445e-07,
|
| 1381 |
+
"loss": 0.2508,
|
| 1382 |
+
"step": 4775
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 1.9528071602929211,
|
| 1386 |
+
"grad_norm": 5.4455342292785645,
|
| 1387 |
+
"learning_rate": 4.488888888888889e-07,
|
| 1388 |
+
"loss": 0.2468,
|
| 1389 |
+
"step": 4800
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 1.9629780309194467,
|
| 1393 |
+
"grad_norm": 4.946452617645264,
|
| 1394 |
+
"learning_rate": 3.9333333333333336e-07,
|
| 1395 |
+
"loss": 0.2459,
|
| 1396 |
+
"step": 4825
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 1.9731489015459722,
|
| 1400 |
+
"grad_norm": 4.054887294769287,
|
| 1401 |
+
"learning_rate": 3.3777777777777777e-07,
|
| 1402 |
+
"loss": 0.2488,
|
| 1403 |
+
"step": 4850
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 1.983319772172498,
|
| 1407 |
+
"grad_norm": 5.577613830566406,
|
| 1408 |
+
"learning_rate": 2.822222222222222e-07,
|
| 1409 |
+
"loss": 0.2435,
|
| 1410 |
+
"step": 4875
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 1.9934906427990235,
|
| 1414 |
+
"grad_norm": 4.222945213317871,
|
| 1415 |
+
"learning_rate": 2.266666666666667e-07,
|
| 1416 |
+
"loss": 0.2399,
|
| 1417 |
+
"step": 4900
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 2.0036615134255493,
|
| 1421 |
+
"grad_norm": 4.715478897094727,
|
| 1422 |
+
"learning_rate": 1.7111111111111114e-07,
|
| 1423 |
+
"loss": 0.2385,
|
| 1424 |
+
"step": 4925
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 2.013832384052075,
|
| 1428 |
+
"grad_norm": 4.2399678230285645,
|
| 1429 |
+
"learning_rate": 1.1555555555555556e-07,
|
| 1430 |
+
"loss": 0.181,
|
| 1431 |
+
"step": 4950
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 2.0240032546786004,
|
| 1435 |
+
"grad_norm": 4.25973653793335,
|
| 1436 |
+
"learning_rate": 6.000000000000001e-08,
|
| 1437 |
+
"loss": 0.1792,
|
| 1438 |
+
"step": 4975
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 2.034174125305126,
|
| 1442 |
+
"grad_norm": 4.661466598510742,
|
| 1443 |
+
"learning_rate": 4.444444444444445e-09,
|
| 1444 |
+
"loss": 0.1846,
|
| 1445 |
+
"step": 5000
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 2.034174125305126,
|
| 1449 |
+
"eval_loss": 0.36655664443969727,
|
| 1450 |
+
"eval_runtime": 1783.5562,
|
| 1451 |
+
"eval_samples_per_second": 2.187,
|
| 1452 |
+
"eval_steps_per_second": 0.137,
|
| 1453 |
+
"eval_wer": 0.27733507195981844,
|
| 1454 |
+
"step": 5000
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 2.034174125305126,
|
| 1458 |
+
"step": 5000,
|
| 1459 |
+
"total_flos": 5.435045990970163e+20,
|
| 1460 |
+
"train_loss": 0.38478662071228026,
|
| 1461 |
+
"train_runtime": 58003.8059,
|
| 1462 |
+
"train_samples_per_second": 2.758,
|
| 1463 |
+
"train_steps_per_second": 0.086
|
| 1464 |
+
}
|
| 1465 |
+
],
|
| 1466 |
+
"logging_steps": 25,
|
| 1467 |
+
"max_steps": 5000,
|
| 1468 |
+
"num_input_tokens_seen": 0,
|
| 1469 |
+
"num_train_epochs": 3,
|
| 1470 |
+
"save_steps": 1000,
|
| 1471 |
+
"stateful_callbacks": {
|
| 1472 |
+
"TrainerControl": {
|
| 1473 |
+
"args": {
|
| 1474 |
+
"should_epoch_stop": false,
|
| 1475 |
+
"should_evaluate": false,
|
| 1476 |
+
"should_log": false,
|
| 1477 |
+
"should_save": true,
|
| 1478 |
+
"should_training_stop": true
|
| 1479 |
+
},
|
| 1480 |
+
"attributes": {}
|
| 1481 |
+
}
|
| 1482 |
+
},
|
| 1483 |
+
"total_flos": 5.435045990970163e+20,
|
| 1484 |
+
"train_batch_size": 16,
|
| 1485 |
+
"trial_name": null,
|
| 1486 |
+
"trial_params": null
|
| 1487 |
+
}
|