End of training
Browse files- README.md +1 -1
- all_results.json +15 -0
- eval_results.json +9 -0
- runs/Oct03_07-46-59_d862200e241c/events.out.tfevents.1727994008.d862200e241c.1103.1 +3 -0
- train_results.json +9 -0
- trainer_state.json +1487 -0
README.md
CHANGED
|
@@ -15,7 +15,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 15 |
|
| 16 |
# whisper-large-v3-turbo-ft-cv-cy
|
| 17 |
|
| 18 |
-
This model is a fine-tuned version of [openai/whisper-large-v3-turbo](https://huggingface.co/openai/whisper-large-v3-turbo) on
|
| 19 |
It achieves the following results on the evaluation set:
|
| 20 |
- Loss: 0.4998
|
| 21 |
- Wer: 0.2181
|
|
|
|
| 15 |
|
| 16 |
# whisper-large-v3-turbo-ft-cv-cy
|
| 17 |
|
| 18 |
+
This model is a fine-tuned version of [openai/whisper-large-v3-turbo](https://huggingface.co/openai/whisper-large-v3-turbo) on the DewiBrynJones/commonvoice_18_0_cy train main dataset.
|
| 19 |
It achieves the following results on the evaluation set:
|
| 20 |
- Loss: 0.4998
|
| 21 |
- Wer: 0.2181
|
all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.080321285140563,
|
| 3 |
+
"eval_loss": 0.4997508227825165,
|
| 4 |
+
"eval_runtime": 2084.4536,
|
| 5 |
+
"eval_samples": 5381,
|
| 6 |
+
"eval_samples_per_second": 2.581,
|
| 7 |
+
"eval_steps_per_second": 0.162,
|
| 8 |
+
"eval_wer": 0.2181001232653386,
|
| 9 |
+
"total_flos": 2.727921844224e+20,
|
| 10 |
+
"train_loss": 0.07613397375550121,
|
| 11 |
+
"train_runtime": 53449.9289,
|
| 12 |
+
"train_samples": 7968,
|
| 13 |
+
"train_samples_per_second": 2.993,
|
| 14 |
+
"train_steps_per_second": 0.094
|
| 15 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.080321285140563,
|
| 3 |
+
"eval_loss": 0.4997508227825165,
|
| 4 |
+
"eval_runtime": 2084.4536,
|
| 5 |
+
"eval_samples": 5381,
|
| 6 |
+
"eval_samples_per_second": 2.581,
|
| 7 |
+
"eval_steps_per_second": 0.162,
|
| 8 |
+
"eval_wer": 0.2181001232653386
|
| 9 |
+
}
|
runs/Oct03_07-46-59_d862200e241c/events.out.tfevents.1727994008.d862200e241c.1103.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d812f3a86853432fa80b4ede781c35d9c57c0a7cd905aa20b335950edee6198f
|
| 3 |
+
size 406
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.080321285140563,
|
| 3 |
+
"total_flos": 2.727921844224e+20,
|
| 4 |
+
"train_loss": 0.07613397375550121,
|
| 5 |
+
"train_runtime": 53449.9289,
|
| 6 |
+
"train_samples": 7968,
|
| 7 |
+
"train_samples_per_second": 2.993,
|
| 8 |
+
"train_steps_per_second": 0.094
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1487 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 20.080321285140563,
|
| 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": 0.10040160642570281,
|
| 13 |
+
"grad_norm": 16.52980613708496,
|
| 14 |
+
"learning_rate": 4.800000000000001e-07,
|
| 15 |
+
"loss": 1.5503,
|
| 16 |
+
"step": 25
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.20080321285140562,
|
| 20 |
+
"grad_norm": 11.904419898986816,
|
| 21 |
+
"learning_rate": 9.800000000000001e-07,
|
| 22 |
+
"loss": 0.7686,
|
| 23 |
+
"step": 50
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.30120481927710846,
|
| 27 |
+
"grad_norm": 10.730138778686523,
|
| 28 |
+
"learning_rate": 1.48e-06,
|
| 29 |
+
"loss": 0.5875,
|
| 30 |
+
"step": 75
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.40160642570281124,
|
| 34 |
+
"grad_norm": 9.80690860748291,
|
| 35 |
+
"learning_rate": 1.98e-06,
|
| 36 |
+
"loss": 0.5653,
|
| 37 |
+
"step": 100
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.5020080321285141,
|
| 41 |
+
"grad_norm": 9.938033103942871,
|
| 42 |
+
"learning_rate": 2.4800000000000004e-06,
|
| 43 |
+
"loss": 0.5157,
|
| 44 |
+
"step": 125
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.6024096385542169,
|
| 48 |
+
"grad_norm": 10.42677116394043,
|
| 49 |
+
"learning_rate": 2.9800000000000003e-06,
|
| 50 |
+
"loss": 0.5222,
|
| 51 |
+
"step": 150
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.7028112449799196,
|
| 55 |
+
"grad_norm": 10.295225143432617,
|
| 56 |
+
"learning_rate": 3.46e-06,
|
| 57 |
+
"loss": 0.4931,
|
| 58 |
+
"step": 175
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.8032128514056225,
|
| 62 |
+
"grad_norm": 11.053289413452148,
|
| 63 |
+
"learning_rate": 3.96e-06,
|
| 64 |
+
"loss": 0.4682,
|
| 65 |
+
"step": 200
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.9036144578313253,
|
| 69 |
+
"grad_norm": 10.42480754852295,
|
| 70 |
+
"learning_rate": 4.4600000000000005e-06,
|
| 71 |
+
"loss": 0.4781,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 1.0040160642570282,
|
| 76 |
+
"grad_norm": 8.67918586730957,
|
| 77 |
+
"learning_rate": 4.960000000000001e-06,
|
| 78 |
+
"loss": 0.4537,
|
| 79 |
+
"step": 250
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 1.104417670682731,
|
| 83 |
+
"grad_norm": 10.315282821655273,
|
| 84 |
+
"learning_rate": 5.460000000000001e-06,
|
| 85 |
+
"loss": 0.3555,
|
| 86 |
+
"step": 275
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 1.2048192771084336,
|
| 90 |
+
"grad_norm": 9.206184387207031,
|
| 91 |
+
"learning_rate": 5.9600000000000005e-06,
|
| 92 |
+
"loss": 0.3416,
|
| 93 |
+
"step": 300
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 1.3052208835341366,
|
| 97 |
+
"grad_norm": 9.105973243713379,
|
| 98 |
+
"learning_rate": 6.460000000000001e-06,
|
| 99 |
+
"loss": 0.341,
|
| 100 |
+
"step": 325
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 1.4056224899598393,
|
| 104 |
+
"grad_norm": 9.135680198669434,
|
| 105 |
+
"learning_rate": 6.96e-06,
|
| 106 |
+
"loss": 0.3486,
|
| 107 |
+
"step": 350
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 1.5060240963855422,
|
| 111 |
+
"grad_norm": 8.966490745544434,
|
| 112 |
+
"learning_rate": 7.4600000000000006e-06,
|
| 113 |
+
"loss": 0.3703,
|
| 114 |
+
"step": 375
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 1.606425702811245,
|
| 118 |
+
"grad_norm": 9.549636840820312,
|
| 119 |
+
"learning_rate": 7.960000000000002e-06,
|
| 120 |
+
"loss": 0.3481,
|
| 121 |
+
"step": 400
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 1.7068273092369477,
|
| 125 |
+
"grad_norm": 8.577873229980469,
|
| 126 |
+
"learning_rate": 8.46e-06,
|
| 127 |
+
"loss": 0.3451,
|
| 128 |
+
"step": 425
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 1.8072289156626506,
|
| 132 |
+
"grad_norm": 7.0459980964660645,
|
| 133 |
+
"learning_rate": 8.96e-06,
|
| 134 |
+
"loss": 0.3492,
|
| 135 |
+
"step": 450
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 1.9076305220883534,
|
| 139 |
+
"grad_norm": 8.674671173095703,
|
| 140 |
+
"learning_rate": 9.460000000000001e-06,
|
| 141 |
+
"loss": 0.3398,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 2.0080321285140563,
|
| 146 |
+
"grad_norm": 6.492681980133057,
|
| 147 |
+
"learning_rate": 9.960000000000001e-06,
|
| 148 |
+
"loss": 0.3226,
|
| 149 |
+
"step": 500
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 2.108433734939759,
|
| 153 |
+
"grad_norm": 8.931197166442871,
|
| 154 |
+
"learning_rate": 9.94888888888889e-06,
|
| 155 |
+
"loss": 0.2213,
|
| 156 |
+
"step": 525
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 2.208835341365462,
|
| 160 |
+
"grad_norm": 8.153618812561035,
|
| 161 |
+
"learning_rate": 9.893333333333334e-06,
|
| 162 |
+
"loss": 0.199,
|
| 163 |
+
"step": 550
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 2.3092369477911645,
|
| 167 |
+
"grad_norm": 6.562689304351807,
|
| 168 |
+
"learning_rate": 9.837777777777778e-06,
|
| 169 |
+
"loss": 0.2146,
|
| 170 |
+
"step": 575
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 2.4096385542168672,
|
| 174 |
+
"grad_norm": 6.018052577972412,
|
| 175 |
+
"learning_rate": 9.782222222222222e-06,
|
| 176 |
+
"loss": 0.2127,
|
| 177 |
+
"step": 600
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 2.5100401606425704,
|
| 181 |
+
"grad_norm": 6.111701011657715,
|
| 182 |
+
"learning_rate": 9.726666666666668e-06,
|
| 183 |
+
"loss": 0.2061,
|
| 184 |
+
"step": 625
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 2.610441767068273,
|
| 188 |
+
"grad_norm": 6.124913215637207,
|
| 189 |
+
"learning_rate": 9.671111111111112e-06,
|
| 190 |
+
"loss": 0.2056,
|
| 191 |
+
"step": 650
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 2.710843373493976,
|
| 195 |
+
"grad_norm": 5.902347087860107,
|
| 196 |
+
"learning_rate": 9.615555555555558e-06,
|
| 197 |
+
"loss": 0.203,
|
| 198 |
+
"step": 675
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 2.8112449799196786,
|
| 202 |
+
"grad_norm": 6.789490699768066,
|
| 203 |
+
"learning_rate": 9.56e-06,
|
| 204 |
+
"loss": 0.2187,
|
| 205 |
+
"step": 700
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 2.9116465863453813,
|
| 209 |
+
"grad_norm": 5.213627338409424,
|
| 210 |
+
"learning_rate": 9.504444444444446e-06,
|
| 211 |
+
"loss": 0.2085,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 3.0120481927710845,
|
| 216 |
+
"grad_norm": 4.021098613739014,
|
| 217 |
+
"learning_rate": 9.44888888888889e-06,
|
| 218 |
+
"loss": 0.2026,
|
| 219 |
+
"step": 750
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 3.112449799196787,
|
| 223 |
+
"grad_norm": 5.916001319885254,
|
| 224 |
+
"learning_rate": 9.393333333333334e-06,
|
| 225 |
+
"loss": 0.1129,
|
| 226 |
+
"step": 775
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 3.21285140562249,
|
| 230 |
+
"grad_norm": 5.063960552215576,
|
| 231 |
+
"learning_rate": 9.33777777777778e-06,
|
| 232 |
+
"loss": 0.116,
|
| 233 |
+
"step": 800
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 3.3132530120481927,
|
| 237 |
+
"grad_norm": 5.580726146697998,
|
| 238 |
+
"learning_rate": 9.282222222222222e-06,
|
| 239 |
+
"loss": 0.1109,
|
| 240 |
+
"step": 825
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 3.4136546184738954,
|
| 244 |
+
"grad_norm": 4.686521053314209,
|
| 245 |
+
"learning_rate": 9.226666666666668e-06,
|
| 246 |
+
"loss": 0.1187,
|
| 247 |
+
"step": 850
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 3.5140562248995986,
|
| 251 |
+
"grad_norm": 5.128270149230957,
|
| 252 |
+
"learning_rate": 9.171111111111112e-06,
|
| 253 |
+
"loss": 0.1133,
|
| 254 |
+
"step": 875
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 3.6144578313253013,
|
| 258 |
+
"grad_norm": 4.7233662605285645,
|
| 259 |
+
"learning_rate": 9.115555555555556e-06,
|
| 260 |
+
"loss": 0.1241,
|
| 261 |
+
"step": 900
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 3.714859437751004,
|
| 265 |
+
"grad_norm": 5.689490795135498,
|
| 266 |
+
"learning_rate": 9.060000000000001e-06,
|
| 267 |
+
"loss": 0.1297,
|
| 268 |
+
"step": 925
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 3.8152610441767068,
|
| 272 |
+
"grad_norm": 5.967415809631348,
|
| 273 |
+
"learning_rate": 9.004444444444445e-06,
|
| 274 |
+
"loss": 0.1251,
|
| 275 |
+
"step": 950
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 3.9156626506024095,
|
| 279 |
+
"grad_norm": 4.376407146453857,
|
| 280 |
+
"learning_rate": 8.94888888888889e-06,
|
| 281 |
+
"loss": 0.1207,
|
| 282 |
+
"step": 975
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 4.016064257028113,
|
| 286 |
+
"grad_norm": 4.3209662437438965,
|
| 287 |
+
"learning_rate": 8.893333333333333e-06,
|
| 288 |
+
"loss": 0.1125,
|
| 289 |
+
"step": 1000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 4.016064257028113,
|
| 293 |
+
"eval_loss": 0.3742842376232147,
|
| 294 |
+
"eval_runtime": 1816.0709,
|
| 295 |
+
"eval_samples_per_second": 2.963,
|
| 296 |
+
"eval_steps_per_second": 0.186,
|
| 297 |
+
"eval_wer": 0.2641258101713786,
|
| 298 |
+
"step": 1000
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 4.116465863453815,
|
| 302 |
+
"grad_norm": 4.307711124420166,
|
| 303 |
+
"learning_rate": 8.83777777777778e-06,
|
| 304 |
+
"loss": 0.0669,
|
| 305 |
+
"step": 1025
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 4.216867469879518,
|
| 309 |
+
"grad_norm": 3.720428228378296,
|
| 310 |
+
"learning_rate": 8.782222222222223e-06,
|
| 311 |
+
"loss": 0.0672,
|
| 312 |
+
"step": 1050
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 4.317269076305221,
|
| 316 |
+
"grad_norm": 4.094714641571045,
|
| 317 |
+
"learning_rate": 8.726666666666667e-06,
|
| 318 |
+
"loss": 0.0644,
|
| 319 |
+
"step": 1075
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 4.417670682730924,
|
| 323 |
+
"grad_norm": 4.306257247924805,
|
| 324 |
+
"learning_rate": 8.671111111111113e-06,
|
| 325 |
+
"loss": 0.0701,
|
| 326 |
+
"step": 1100
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 4.518072289156627,
|
| 330 |
+
"grad_norm": 3.7728941440582275,
|
| 331 |
+
"learning_rate": 8.615555555555555e-06,
|
| 332 |
+
"loss": 0.0707,
|
| 333 |
+
"step": 1125
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 4.618473895582329,
|
| 337 |
+
"grad_norm": 4.781138896942139,
|
| 338 |
+
"learning_rate": 8.560000000000001e-06,
|
| 339 |
+
"loss": 0.0722,
|
| 340 |
+
"step": 1150
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 4.718875502008032,
|
| 344 |
+
"grad_norm": 3.8744442462921143,
|
| 345 |
+
"learning_rate": 8.504444444444445e-06,
|
| 346 |
+
"loss": 0.0708,
|
| 347 |
+
"step": 1175
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 4.8192771084337345,
|
| 351 |
+
"grad_norm": 4.639457702636719,
|
| 352 |
+
"learning_rate": 8.448888888888889e-06,
|
| 353 |
+
"loss": 0.0712,
|
| 354 |
+
"step": 1200
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 4.919678714859438,
|
| 358 |
+
"grad_norm": 2.9719793796539307,
|
| 359 |
+
"learning_rate": 8.393333333333335e-06,
|
| 360 |
+
"loss": 0.0748,
|
| 361 |
+
"step": 1225
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 5.020080321285141,
|
| 365 |
+
"grad_norm": 2.4250712394714355,
|
| 366 |
+
"learning_rate": 8.337777777777777e-06,
|
| 367 |
+
"loss": 0.0601,
|
| 368 |
+
"step": 1250
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 5.120481927710843,
|
| 372 |
+
"grad_norm": 3.8074028491973877,
|
| 373 |
+
"learning_rate": 8.282222222222223e-06,
|
| 374 |
+
"loss": 0.0396,
|
| 375 |
+
"step": 1275
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 5.220883534136546,
|
| 379 |
+
"grad_norm": 2.799952507019043,
|
| 380 |
+
"learning_rate": 8.226666666666667e-06,
|
| 381 |
+
"loss": 0.0366,
|
| 382 |
+
"step": 1300
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 5.321285140562249,
|
| 386 |
+
"grad_norm": 2.802058219909668,
|
| 387 |
+
"learning_rate": 8.171111111111113e-06,
|
| 388 |
+
"loss": 0.0452,
|
| 389 |
+
"step": 1325
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 5.421686746987952,
|
| 393 |
+
"grad_norm": 2.9185872077941895,
|
| 394 |
+
"learning_rate": 8.115555555555557e-06,
|
| 395 |
+
"loss": 0.0429,
|
| 396 |
+
"step": 1350
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 5.522088353413655,
|
| 400 |
+
"grad_norm": 4.0047526359558105,
|
| 401 |
+
"learning_rate": 8.06e-06,
|
| 402 |
+
"loss": 0.045,
|
| 403 |
+
"step": 1375
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 5.622489959839357,
|
| 407 |
+
"grad_norm": 2.992342710494995,
|
| 408 |
+
"learning_rate": 8.004444444444445e-06,
|
| 409 |
+
"loss": 0.0466,
|
| 410 |
+
"step": 1400
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 5.72289156626506,
|
| 414 |
+
"grad_norm": 4.288572311401367,
|
| 415 |
+
"learning_rate": 7.948888888888889e-06,
|
| 416 |
+
"loss": 0.0471,
|
| 417 |
+
"step": 1425
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 5.823293172690763,
|
| 421 |
+
"grad_norm": 2.9833133220672607,
|
| 422 |
+
"learning_rate": 7.893333333333335e-06,
|
| 423 |
+
"loss": 0.0427,
|
| 424 |
+
"step": 1450
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 5.923694779116466,
|
| 428 |
+
"grad_norm": 3.362567663192749,
|
| 429 |
+
"learning_rate": 7.837777777777779e-06,
|
| 430 |
+
"loss": 0.0503,
|
| 431 |
+
"step": 1475
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 6.024096385542169,
|
| 435 |
+
"grad_norm": 3.4157121181488037,
|
| 436 |
+
"learning_rate": 7.782222222222223e-06,
|
| 437 |
+
"loss": 0.04,
|
| 438 |
+
"step": 1500
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 6.124497991967871,
|
| 442 |
+
"grad_norm": 2.642629623413086,
|
| 443 |
+
"learning_rate": 7.726666666666667e-06,
|
| 444 |
+
"loss": 0.0266,
|
| 445 |
+
"step": 1525
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 6.224899598393574,
|
| 449 |
+
"grad_norm": 2.336653470993042,
|
| 450 |
+
"learning_rate": 7.67111111111111e-06,
|
| 451 |
+
"loss": 0.028,
|
| 452 |
+
"step": 1550
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 6.325301204819277,
|
| 456 |
+
"grad_norm": 2.498267889022827,
|
| 457 |
+
"learning_rate": 7.6155555555555564e-06,
|
| 458 |
+
"loss": 0.0268,
|
| 459 |
+
"step": 1575
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 6.42570281124498,
|
| 463 |
+
"grad_norm": 2.4469289779663086,
|
| 464 |
+
"learning_rate": 7.5600000000000005e-06,
|
| 465 |
+
"loss": 0.0261,
|
| 466 |
+
"step": 1600
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 6.526104417670683,
|
| 470 |
+
"grad_norm": 2.872375726699829,
|
| 471 |
+
"learning_rate": 7.504444444444445e-06,
|
| 472 |
+
"loss": 0.03,
|
| 473 |
+
"step": 1625
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 6.626506024096385,
|
| 477 |
+
"grad_norm": 4.681485652923584,
|
| 478 |
+
"learning_rate": 7.44888888888889e-06,
|
| 479 |
+
"loss": 0.0279,
|
| 480 |
+
"step": 1650
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 6.7269076305220885,
|
| 484 |
+
"grad_norm": 3.043527841567993,
|
| 485 |
+
"learning_rate": 7.393333333333333e-06,
|
| 486 |
+
"loss": 0.0305,
|
| 487 |
+
"step": 1675
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 6.827309236947791,
|
| 491 |
+
"grad_norm": 3.1976921558380127,
|
| 492 |
+
"learning_rate": 7.337777777777778e-06,
|
| 493 |
+
"loss": 0.0339,
|
| 494 |
+
"step": 1700
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 6.927710843373494,
|
| 498 |
+
"grad_norm": 2.7758069038391113,
|
| 499 |
+
"learning_rate": 7.282222222222222e-06,
|
| 500 |
+
"loss": 0.0299,
|
| 501 |
+
"step": 1725
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 7.028112449799197,
|
| 505 |
+
"grad_norm": 2.653137445449829,
|
| 506 |
+
"learning_rate": 7.226666666666667e-06,
|
| 507 |
+
"loss": 0.0289,
|
| 508 |
+
"step": 1750
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 7.128514056224899,
|
| 512 |
+
"grad_norm": 2.6387505531311035,
|
| 513 |
+
"learning_rate": 7.171111111111112e-06,
|
| 514 |
+
"loss": 0.0188,
|
| 515 |
+
"step": 1775
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 7.228915662650603,
|
| 519 |
+
"grad_norm": 2.1044726371765137,
|
| 520 |
+
"learning_rate": 7.115555555555557e-06,
|
| 521 |
+
"loss": 0.0198,
|
| 522 |
+
"step": 1800
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 7.329317269076305,
|
| 526 |
+
"grad_norm": 2.4335100650787354,
|
| 527 |
+
"learning_rate": 7.06e-06,
|
| 528 |
+
"loss": 0.0187,
|
| 529 |
+
"step": 1825
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 7.429718875502008,
|
| 533 |
+
"grad_norm": 2.240412473678589,
|
| 534 |
+
"learning_rate": 7.004444444444445e-06,
|
| 535 |
+
"loss": 0.0209,
|
| 536 |
+
"step": 1850
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 7.530120481927711,
|
| 540 |
+
"grad_norm": 3.036201238632202,
|
| 541 |
+
"learning_rate": 6.948888888888889e-06,
|
| 542 |
+
"loss": 0.0209,
|
| 543 |
+
"step": 1875
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 7.6305220883534135,
|
| 547 |
+
"grad_norm": 1.8761370182037354,
|
| 548 |
+
"learning_rate": 6.893333333333334e-06,
|
| 549 |
+
"loss": 0.0204,
|
| 550 |
+
"step": 1900
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 7.730923694779117,
|
| 554 |
+
"grad_norm": 3.35725474357605,
|
| 555 |
+
"learning_rate": 6.837777777777779e-06,
|
| 556 |
+
"loss": 0.0193,
|
| 557 |
+
"step": 1925
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 7.831325301204819,
|
| 561 |
+
"grad_norm": 2.9908254146575928,
|
| 562 |
+
"learning_rate": 6.782222222222222e-06,
|
| 563 |
+
"loss": 0.0204,
|
| 564 |
+
"step": 1950
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 7.931726907630522,
|
| 568 |
+
"grad_norm": 4.191911220550537,
|
| 569 |
+
"learning_rate": 6.726666666666667e-06,
|
| 570 |
+
"loss": 0.022,
|
| 571 |
+
"step": 1975
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 8.032128514056225,
|
| 575 |
+
"grad_norm": 1.2732858657836914,
|
| 576 |
+
"learning_rate": 6.671111111111112e-06,
|
| 577 |
+
"loss": 0.0178,
|
| 578 |
+
"step": 2000
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 8.032128514056225,
|
| 582 |
+
"eval_loss": 0.42306816577911377,
|
| 583 |
+
"eval_runtime": 1886.6147,
|
| 584 |
+
"eval_samples_per_second": 2.852,
|
| 585 |
+
"eval_steps_per_second": 0.179,
|
| 586 |
+
"eval_wer": 0.249890651715774,
|
| 587 |
+
"step": 2000
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 8.132530120481928,
|
| 591 |
+
"grad_norm": 4.009426593780518,
|
| 592 |
+
"learning_rate": 6.615555555555556e-06,
|
| 593 |
+
"loss": 0.0132,
|
| 594 |
+
"step": 2025
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 8.23293172690763,
|
| 598 |
+
"grad_norm": 3.8768014907836914,
|
| 599 |
+
"learning_rate": 6.560000000000001e-06,
|
| 600 |
+
"loss": 0.0148,
|
| 601 |
+
"step": 2050
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 8.333333333333334,
|
| 605 |
+
"grad_norm": 2.1475863456726074,
|
| 606 |
+
"learning_rate": 6.504444444444446e-06,
|
| 607 |
+
"loss": 0.0145,
|
| 608 |
+
"step": 2075
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 8.433734939759036,
|
| 612 |
+
"grad_norm": 1.8194453716278076,
|
| 613 |
+
"learning_rate": 6.448888888888889e-06,
|
| 614 |
+
"loss": 0.0151,
|
| 615 |
+
"step": 2100
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 8.534136546184738,
|
| 619 |
+
"grad_norm": 2.893296241760254,
|
| 620 |
+
"learning_rate": 6.393333333333334e-06,
|
| 621 |
+
"loss": 0.015,
|
| 622 |
+
"step": 2125
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 8.634538152610443,
|
| 626 |
+
"grad_norm": 1.6417754888534546,
|
| 627 |
+
"learning_rate": 6.3377777777777786e-06,
|
| 628 |
+
"loss": 0.0164,
|
| 629 |
+
"step": 2150
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 8.734939759036145,
|
| 633 |
+
"grad_norm": 2.3964133262634277,
|
| 634 |
+
"learning_rate": 6.282222222222223e-06,
|
| 635 |
+
"loss": 0.0167,
|
| 636 |
+
"step": 2175
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 8.835341365461847,
|
| 640 |
+
"grad_norm": 2.9204821586608887,
|
| 641 |
+
"learning_rate": 6.2266666666666675e-06,
|
| 642 |
+
"loss": 0.0135,
|
| 643 |
+
"step": 2200
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 8.93574297188755,
|
| 647 |
+
"grad_norm": 3.2966790199279785,
|
| 648 |
+
"learning_rate": 6.171111111111112e-06,
|
| 649 |
+
"loss": 0.0133,
|
| 650 |
+
"step": 2225
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 9.036144578313253,
|
| 654 |
+
"grad_norm": 1.6279815435409546,
|
| 655 |
+
"learning_rate": 6.1155555555555555e-06,
|
| 656 |
+
"loss": 0.0146,
|
| 657 |
+
"step": 2250
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 9.136546184738956,
|
| 661 |
+
"grad_norm": 3.7027101516723633,
|
| 662 |
+
"learning_rate": 6.0600000000000004e-06,
|
| 663 |
+
"loss": 0.0104,
|
| 664 |
+
"step": 2275
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 9.236947791164658,
|
| 668 |
+
"grad_norm": 2.011039972305298,
|
| 669 |
+
"learning_rate": 6.006666666666667e-06,
|
| 670 |
+
"loss": 0.0108,
|
| 671 |
+
"step": 2300
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 9.337349397590362,
|
| 675 |
+
"grad_norm": 2.4486405849456787,
|
| 676 |
+
"learning_rate": 5.951111111111112e-06,
|
| 677 |
+
"loss": 0.01,
|
| 678 |
+
"step": 2325
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 9.437751004016064,
|
| 682 |
+
"grad_norm": 1.047418475151062,
|
| 683 |
+
"learning_rate": 5.895555555555557e-06,
|
| 684 |
+
"loss": 0.012,
|
| 685 |
+
"step": 2350
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 9.538152610441767,
|
| 689 |
+
"grad_norm": 2.6720094680786133,
|
| 690 |
+
"learning_rate": 5.84e-06,
|
| 691 |
+
"loss": 0.011,
|
| 692 |
+
"step": 2375
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 9.638554216867469,
|
| 696 |
+
"grad_norm": 1.3718459606170654,
|
| 697 |
+
"learning_rate": 5.784444444444445e-06,
|
| 698 |
+
"loss": 0.0122,
|
| 699 |
+
"step": 2400
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 9.738955823293173,
|
| 703 |
+
"grad_norm": 1.6287081241607666,
|
| 704 |
+
"learning_rate": 5.72888888888889e-06,
|
| 705 |
+
"loss": 0.009,
|
| 706 |
+
"step": 2425
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 9.839357429718875,
|
| 710 |
+
"grad_norm": 2.190797805786133,
|
| 711 |
+
"learning_rate": 5.673333333333334e-06,
|
| 712 |
+
"loss": 0.0132,
|
| 713 |
+
"step": 2450
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 9.939759036144578,
|
| 717 |
+
"grad_norm": 2.275399684906006,
|
| 718 |
+
"learning_rate": 5.617777777777779e-06,
|
| 719 |
+
"loss": 0.0105,
|
| 720 |
+
"step": 2475
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 10.040160642570282,
|
| 724 |
+
"grad_norm": 1.9050276279449463,
|
| 725 |
+
"learning_rate": 5.562222222222222e-06,
|
| 726 |
+
"loss": 0.0107,
|
| 727 |
+
"step": 2500
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 10.140562248995984,
|
| 731 |
+
"grad_norm": 0.9361718893051147,
|
| 732 |
+
"learning_rate": 5.506666666666667e-06,
|
| 733 |
+
"loss": 0.0074,
|
| 734 |
+
"step": 2525
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 10.240963855421686,
|
| 738 |
+
"grad_norm": 2.5661396980285645,
|
| 739 |
+
"learning_rate": 5.451111111111112e-06,
|
| 740 |
+
"loss": 0.0081,
|
| 741 |
+
"step": 2550
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 10.34136546184739,
|
| 745 |
+
"grad_norm": 1.1169323921203613,
|
| 746 |
+
"learning_rate": 5.3955555555555565e-06,
|
| 747 |
+
"loss": 0.0088,
|
| 748 |
+
"step": 2575
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 10.441767068273093,
|
| 752 |
+
"grad_norm": 1.0962170362472534,
|
| 753 |
+
"learning_rate": 5.3400000000000005e-06,
|
| 754 |
+
"loss": 0.0077,
|
| 755 |
+
"step": 2600
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 10.542168674698795,
|
| 759 |
+
"grad_norm": 1.0353940725326538,
|
| 760 |
+
"learning_rate": 5.2844444444444454e-06,
|
| 761 |
+
"loss": 0.0096,
|
| 762 |
+
"step": 2625
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 10.642570281124499,
|
| 766 |
+
"grad_norm": 3.781886339187622,
|
| 767 |
+
"learning_rate": 5.228888888888889e-06,
|
| 768 |
+
"loss": 0.0082,
|
| 769 |
+
"step": 2650
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 10.742971887550201,
|
| 773 |
+
"grad_norm": 3.071993112564087,
|
| 774 |
+
"learning_rate": 5.1733333333333335e-06,
|
| 775 |
+
"loss": 0.0076,
|
| 776 |
+
"step": 2675
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 10.843373493975903,
|
| 780 |
+
"grad_norm": 1.5968672037124634,
|
| 781 |
+
"learning_rate": 5.117777777777778e-06,
|
| 782 |
+
"loss": 0.0066,
|
| 783 |
+
"step": 2700
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 10.943775100401606,
|
| 787 |
+
"grad_norm": 1.1753120422363281,
|
| 788 |
+
"learning_rate": 5.062222222222222e-06,
|
| 789 |
+
"loss": 0.0079,
|
| 790 |
+
"step": 2725
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 11.04417670682731,
|
| 794 |
+
"grad_norm": 0.6646733283996582,
|
| 795 |
+
"learning_rate": 5.006666666666667e-06,
|
| 796 |
+
"loss": 0.0068,
|
| 797 |
+
"step": 2750
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 11.144578313253012,
|
| 801 |
+
"grad_norm": 1.3602428436279297,
|
| 802 |
+
"learning_rate": 4.951111111111111e-06,
|
| 803 |
+
"loss": 0.0066,
|
| 804 |
+
"step": 2775
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 11.244979919678714,
|
| 808 |
+
"grad_norm": 0.6632346510887146,
|
| 809 |
+
"learning_rate": 4.895555555555556e-06,
|
| 810 |
+
"loss": 0.0042,
|
| 811 |
+
"step": 2800
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 11.345381526104418,
|
| 815 |
+
"grad_norm": 2.823923349380493,
|
| 816 |
+
"learning_rate": 4.84e-06,
|
| 817 |
+
"loss": 0.0044,
|
| 818 |
+
"step": 2825
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 11.44578313253012,
|
| 822 |
+
"grad_norm": 1.5665075778961182,
|
| 823 |
+
"learning_rate": 4.784444444444445e-06,
|
| 824 |
+
"loss": 0.0048,
|
| 825 |
+
"step": 2850
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 11.546184738955823,
|
| 829 |
+
"grad_norm": 2.2108278274536133,
|
| 830 |
+
"learning_rate": 4.728888888888889e-06,
|
| 831 |
+
"loss": 0.006,
|
| 832 |
+
"step": 2875
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 11.646586345381525,
|
| 836 |
+
"grad_norm": 0.9662758111953735,
|
| 837 |
+
"learning_rate": 4.673333333333333e-06,
|
| 838 |
+
"loss": 0.0049,
|
| 839 |
+
"step": 2900
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 11.74698795180723,
|
| 843 |
+
"grad_norm": 1.4958486557006836,
|
| 844 |
+
"learning_rate": 4.617777777777778e-06,
|
| 845 |
+
"loss": 0.0062,
|
| 846 |
+
"step": 2925
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 11.847389558232932,
|
| 850 |
+
"grad_norm": 2.0174615383148193,
|
| 851 |
+
"learning_rate": 4.562222222222222e-06,
|
| 852 |
+
"loss": 0.0058,
|
| 853 |
+
"step": 2950
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 11.947791164658634,
|
| 857 |
+
"grad_norm": 2.672804594039917,
|
| 858 |
+
"learning_rate": 4.506666666666667e-06,
|
| 859 |
+
"loss": 0.0056,
|
| 860 |
+
"step": 2975
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 12.048192771084338,
|
| 864 |
+
"grad_norm": 0.4241655170917511,
|
| 865 |
+
"learning_rate": 4.451111111111112e-06,
|
| 866 |
+
"loss": 0.0064,
|
| 867 |
+
"step": 3000
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 12.048192771084338,
|
| 871 |
+
"eval_loss": 0.46265721321105957,
|
| 872 |
+
"eval_runtime": 1822.5129,
|
| 873 |
+
"eval_samples_per_second": 2.953,
|
| 874 |
+
"eval_steps_per_second": 0.185,
|
| 875 |
+
"eval_wer": 0.2341246172810052,
|
| 876 |
+
"step": 3000
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 12.14859437751004,
|
| 880 |
+
"grad_norm": 0.3871651589870453,
|
| 881 |
+
"learning_rate": 4.395555555555556e-06,
|
| 882 |
+
"loss": 0.0035,
|
| 883 |
+
"step": 3025
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 12.248995983935743,
|
| 887 |
+
"grad_norm": 0.2253613919019699,
|
| 888 |
+
"learning_rate": 4.34e-06,
|
| 889 |
+
"loss": 0.0029,
|
| 890 |
+
"step": 3050
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 12.349397590361447,
|
| 894 |
+
"grad_norm": 0.38276419043540955,
|
| 895 |
+
"learning_rate": 4.284444444444445e-06,
|
| 896 |
+
"loss": 0.0037,
|
| 897 |
+
"step": 3075
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 12.449799196787149,
|
| 901 |
+
"grad_norm": 0.8364894390106201,
|
| 902 |
+
"learning_rate": 4.228888888888889e-06,
|
| 903 |
+
"loss": 0.0036,
|
| 904 |
+
"step": 3100
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 12.550200803212851,
|
| 908 |
+
"grad_norm": 0.37833768129348755,
|
| 909 |
+
"learning_rate": 4.173333333333334e-06,
|
| 910 |
+
"loss": 0.0027,
|
| 911 |
+
"step": 3125
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 12.650602409638553,
|
| 915 |
+
"grad_norm": 0.246841162443161,
|
| 916 |
+
"learning_rate": 4.117777777777779e-06,
|
| 917 |
+
"loss": 0.0044,
|
| 918 |
+
"step": 3150
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 12.751004016064257,
|
| 922 |
+
"grad_norm": 0.3080684542655945,
|
| 923 |
+
"learning_rate": 4.062222222222223e-06,
|
| 924 |
+
"loss": 0.0039,
|
| 925 |
+
"step": 3175
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 12.85140562248996,
|
| 929 |
+
"grad_norm": 0.14490999281406403,
|
| 930 |
+
"learning_rate": 4.006666666666667e-06,
|
| 931 |
+
"loss": 0.0045,
|
| 932 |
+
"step": 3200
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 12.951807228915662,
|
| 936 |
+
"grad_norm": 0.6069326996803284,
|
| 937 |
+
"learning_rate": 3.951111111111112e-06,
|
| 938 |
+
"loss": 0.0031,
|
| 939 |
+
"step": 3225
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 13.052208835341366,
|
| 943 |
+
"grad_norm": 0.8227677941322327,
|
| 944 |
+
"learning_rate": 3.895555555555556e-06,
|
| 945 |
+
"loss": 0.0023,
|
| 946 |
+
"step": 3250
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 13.152610441767068,
|
| 950 |
+
"grad_norm": 1.3870292901992798,
|
| 951 |
+
"learning_rate": 3.8400000000000005e-06,
|
| 952 |
+
"loss": 0.0031,
|
| 953 |
+
"step": 3275
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 13.25301204819277,
|
| 957 |
+
"grad_norm": 2.032243490219116,
|
| 958 |
+
"learning_rate": 3.784444444444445e-06,
|
| 959 |
+
"loss": 0.0021,
|
| 960 |
+
"step": 3300
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 13.353413654618475,
|
| 964 |
+
"grad_norm": 3.4899916648864746,
|
| 965 |
+
"learning_rate": 3.728888888888889e-06,
|
| 966 |
+
"loss": 0.0022,
|
| 967 |
+
"step": 3325
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 13.453815261044177,
|
| 971 |
+
"grad_norm": 1.3607572317123413,
|
| 972 |
+
"learning_rate": 3.673333333333334e-06,
|
| 973 |
+
"loss": 0.0017,
|
| 974 |
+
"step": 3350
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 13.55421686746988,
|
| 978 |
+
"grad_norm": 0.21469266712665558,
|
| 979 |
+
"learning_rate": 3.617777777777778e-06,
|
| 980 |
+
"loss": 0.0018,
|
| 981 |
+
"step": 3375
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 13.654618473895582,
|
| 985 |
+
"grad_norm": 0.11461154371500015,
|
| 986 |
+
"learning_rate": 3.5622222222222224e-06,
|
| 987 |
+
"loss": 0.0011,
|
| 988 |
+
"step": 3400
|
| 989 |
+
},
|
| 990 |
+
{
|
| 991 |
+
"epoch": 13.755020080321286,
|
| 992 |
+
"grad_norm": 3.34506893157959,
|
| 993 |
+
"learning_rate": 3.5066666666666673e-06,
|
| 994 |
+
"loss": 0.0018,
|
| 995 |
+
"step": 3425
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 13.855421686746988,
|
| 999 |
+
"grad_norm": 0.4698576331138611,
|
| 1000 |
+
"learning_rate": 3.4511111111111113e-06,
|
| 1001 |
+
"loss": 0.0022,
|
| 1002 |
+
"step": 3450
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 13.95582329317269,
|
| 1006 |
+
"grad_norm": 0.8212801814079285,
|
| 1007 |
+
"learning_rate": 3.3955555555555558e-06,
|
| 1008 |
+
"loss": 0.0014,
|
| 1009 |
+
"step": 3475
|
| 1010 |
+
},
|
| 1011 |
+
{
|
| 1012 |
+
"epoch": 14.056224899598394,
|
| 1013 |
+
"grad_norm": 0.26110076904296875,
|
| 1014 |
+
"learning_rate": 3.3400000000000006e-06,
|
| 1015 |
+
"loss": 0.0014,
|
| 1016 |
+
"step": 3500
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 14.156626506024097,
|
| 1020 |
+
"grad_norm": 0.08139172196388245,
|
| 1021 |
+
"learning_rate": 3.2844444444444447e-06,
|
| 1022 |
+
"loss": 0.0009,
|
| 1023 |
+
"step": 3525
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 14.257028112449799,
|
| 1027 |
+
"grad_norm": 0.040998633950948715,
|
| 1028 |
+
"learning_rate": 3.228888888888889e-06,
|
| 1029 |
+
"loss": 0.0009,
|
| 1030 |
+
"step": 3550
|
| 1031 |
+
},
|
| 1032 |
+
{
|
| 1033 |
+
"epoch": 14.357429718875501,
|
| 1034 |
+
"grad_norm": 0.1332196742296219,
|
| 1035 |
+
"learning_rate": 3.173333333333334e-06,
|
| 1036 |
+
"loss": 0.001,
|
| 1037 |
+
"step": 3575
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 14.457831325301205,
|
| 1041 |
+
"grad_norm": 0.0915873721241951,
|
| 1042 |
+
"learning_rate": 3.117777777777778e-06,
|
| 1043 |
+
"loss": 0.0019,
|
| 1044 |
+
"step": 3600
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 14.558232931726907,
|
| 1048 |
+
"grad_norm": 0.11250325292348862,
|
| 1049 |
+
"learning_rate": 3.0622222222222225e-06,
|
| 1050 |
+
"loss": 0.0013,
|
| 1051 |
+
"step": 3625
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 14.65863453815261,
|
| 1055 |
+
"grad_norm": 0.06163531169295311,
|
| 1056 |
+
"learning_rate": 3.0066666666666674e-06,
|
| 1057 |
+
"loss": 0.0009,
|
| 1058 |
+
"step": 3650
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 14.759036144578314,
|
| 1062 |
+
"grad_norm": 0.2607724368572235,
|
| 1063 |
+
"learning_rate": 2.9511111111111114e-06,
|
| 1064 |
+
"loss": 0.0009,
|
| 1065 |
+
"step": 3675
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 14.859437751004016,
|
| 1069 |
+
"grad_norm": 0.0904630720615387,
|
| 1070 |
+
"learning_rate": 2.895555555555556e-06,
|
| 1071 |
+
"loss": 0.001,
|
| 1072 |
+
"step": 3700
|
| 1073 |
+
},
|
| 1074 |
+
{
|
| 1075 |
+
"epoch": 14.959839357429718,
|
| 1076 |
+
"grad_norm": 0.04550072178244591,
|
| 1077 |
+
"learning_rate": 2.84e-06,
|
| 1078 |
+
"loss": 0.001,
|
| 1079 |
+
"step": 3725
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 15.060240963855422,
|
| 1083 |
+
"grad_norm": 0.48310792446136475,
|
| 1084 |
+
"learning_rate": 2.784444444444445e-06,
|
| 1085 |
+
"loss": 0.0008,
|
| 1086 |
+
"step": 3750
|
| 1087 |
+
},
|
| 1088 |
+
{
|
| 1089 |
+
"epoch": 15.160642570281125,
|
| 1090 |
+
"grad_norm": 0.037569478154182434,
|
| 1091 |
+
"learning_rate": 2.7288888888888893e-06,
|
| 1092 |
+
"loss": 0.0005,
|
| 1093 |
+
"step": 3775
|
| 1094 |
+
},
|
| 1095 |
+
{
|
| 1096 |
+
"epoch": 15.261044176706827,
|
| 1097 |
+
"grad_norm": 0.03933248668909073,
|
| 1098 |
+
"learning_rate": 2.6733333333333333e-06,
|
| 1099 |
+
"loss": 0.0006,
|
| 1100 |
+
"step": 3800
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 15.36144578313253,
|
| 1104 |
+
"grad_norm": 0.03279777243733406,
|
| 1105 |
+
"learning_rate": 2.617777777777778e-06,
|
| 1106 |
+
"loss": 0.0014,
|
| 1107 |
+
"step": 3825
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 15.461847389558233,
|
| 1111 |
+
"grad_norm": 0.0519753098487854,
|
| 1112 |
+
"learning_rate": 2.5622222222222226e-06,
|
| 1113 |
+
"loss": 0.0007,
|
| 1114 |
+
"step": 3850
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 15.562248995983936,
|
| 1118 |
+
"grad_norm": 0.06775977462530136,
|
| 1119 |
+
"learning_rate": 2.5066666666666667e-06,
|
| 1120 |
+
"loss": 0.001,
|
| 1121 |
+
"step": 3875
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 15.662650602409638,
|
| 1125 |
+
"grad_norm": 0.962402880191803,
|
| 1126 |
+
"learning_rate": 2.451111111111111e-06,
|
| 1127 |
+
"loss": 0.0011,
|
| 1128 |
+
"step": 3900
|
| 1129 |
+
},
|
| 1130 |
+
{
|
| 1131 |
+
"epoch": 15.763052208835342,
|
| 1132 |
+
"grad_norm": 0.03904138505458832,
|
| 1133 |
+
"learning_rate": 2.3955555555555556e-06,
|
| 1134 |
+
"loss": 0.0007,
|
| 1135 |
+
"step": 3925
|
| 1136 |
+
},
|
| 1137 |
+
{
|
| 1138 |
+
"epoch": 15.863453815261044,
|
| 1139 |
+
"grad_norm": 0.041153594851493835,
|
| 1140 |
+
"learning_rate": 2.3400000000000005e-06,
|
| 1141 |
+
"loss": 0.0005,
|
| 1142 |
+
"step": 3950
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 15.963855421686747,
|
| 1146 |
+
"grad_norm": 0.0350770466029644,
|
| 1147 |
+
"learning_rate": 2.2844444444444445e-06,
|
| 1148 |
+
"loss": 0.0005,
|
| 1149 |
+
"step": 3975
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 16.06425702811245,
|
| 1153 |
+
"grad_norm": 0.027113359421491623,
|
| 1154 |
+
"learning_rate": 2.228888888888889e-06,
|
| 1155 |
+
"loss": 0.0004,
|
| 1156 |
+
"step": 4000
|
| 1157 |
+
},
|
| 1158 |
+
{
|
| 1159 |
+
"epoch": 16.06425702811245,
|
| 1160 |
+
"eval_loss": 0.4913391172885895,
|
| 1161 |
+
"eval_runtime": 2057.5949,
|
| 1162 |
+
"eval_samples_per_second": 2.615,
|
| 1163 |
+
"eval_steps_per_second": 0.164,
|
| 1164 |
+
"eval_wer": 0.22008827388762972,
|
| 1165 |
+
"step": 4000
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 16.164658634538153,
|
| 1169 |
+
"grad_norm": 0.027138447389006615,
|
| 1170 |
+
"learning_rate": 2.1733333333333334e-06,
|
| 1171 |
+
"loss": 0.0004,
|
| 1172 |
+
"step": 4025
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 16.265060240963855,
|
| 1176 |
+
"grad_norm": 0.02426302060484886,
|
| 1177 |
+
"learning_rate": 2.117777777777778e-06,
|
| 1178 |
+
"loss": 0.0004,
|
| 1179 |
+
"step": 4050
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 16.365461847389557,
|
| 1183 |
+
"grad_norm": 0.04138059541583061,
|
| 1184 |
+
"learning_rate": 2.0622222222222223e-06,
|
| 1185 |
+
"loss": 0.0004,
|
| 1186 |
+
"step": 4075
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 16.46586345381526,
|
| 1190 |
+
"grad_norm": 0.056056153029203415,
|
| 1191 |
+
"learning_rate": 2.006666666666667e-06,
|
| 1192 |
+
"loss": 0.0004,
|
| 1193 |
+
"step": 4100
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 16.566265060240966,
|
| 1197 |
+
"grad_norm": 0.024884460493922234,
|
| 1198 |
+
"learning_rate": 1.9511111111111113e-06,
|
| 1199 |
+
"loss": 0.0004,
|
| 1200 |
+
"step": 4125
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 16.666666666666668,
|
| 1204 |
+
"grad_norm": 0.023256096988916397,
|
| 1205 |
+
"learning_rate": 1.8955555555555557e-06,
|
| 1206 |
+
"loss": 0.0003,
|
| 1207 |
+
"step": 4150
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 16.76706827309237,
|
| 1211 |
+
"grad_norm": 0.030147407203912735,
|
| 1212 |
+
"learning_rate": 1.8400000000000002e-06,
|
| 1213 |
+
"loss": 0.0005,
|
| 1214 |
+
"step": 4175
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 16.867469879518072,
|
| 1218 |
+
"grad_norm": 0.02672114223241806,
|
| 1219 |
+
"learning_rate": 1.7844444444444444e-06,
|
| 1220 |
+
"loss": 0.0003,
|
| 1221 |
+
"step": 4200
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 16.967871485943775,
|
| 1225 |
+
"grad_norm": 0.02835569903254509,
|
| 1226 |
+
"learning_rate": 1.728888888888889e-06,
|
| 1227 |
+
"loss": 0.0003,
|
| 1228 |
+
"step": 4225
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 17.068273092369477,
|
| 1232 |
+
"grad_norm": 0.02380695752799511,
|
| 1233 |
+
"learning_rate": 1.6733333333333335e-06,
|
| 1234 |
+
"loss": 0.0005,
|
| 1235 |
+
"step": 4250
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 17.16867469879518,
|
| 1239 |
+
"grad_norm": 0.0286069605499506,
|
| 1240 |
+
"learning_rate": 1.6177777777777778e-06,
|
| 1241 |
+
"loss": 0.0004,
|
| 1242 |
+
"step": 4275
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 17.269076305220885,
|
| 1246 |
+
"grad_norm": 0.030239233747124672,
|
| 1247 |
+
"learning_rate": 1.5622222222222225e-06,
|
| 1248 |
+
"loss": 0.0003,
|
| 1249 |
+
"step": 4300
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 17.369477911646587,
|
| 1253 |
+
"grad_norm": 0.02350962907075882,
|
| 1254 |
+
"learning_rate": 1.506666666666667e-06,
|
| 1255 |
+
"loss": 0.0003,
|
| 1256 |
+
"step": 4325
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 17.46987951807229,
|
| 1260 |
+
"grad_norm": 0.021099906414747238,
|
| 1261 |
+
"learning_rate": 1.4511111111111112e-06,
|
| 1262 |
+
"loss": 0.0003,
|
| 1263 |
+
"step": 4350
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 17.570281124497992,
|
| 1267 |
+
"grad_norm": 0.020377565175294876,
|
| 1268 |
+
"learning_rate": 1.3955555555555556e-06,
|
| 1269 |
+
"loss": 0.0004,
|
| 1270 |
+
"step": 4375
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 17.670682730923694,
|
| 1274 |
+
"grad_norm": 0.025648899376392365,
|
| 1275 |
+
"learning_rate": 1.34e-06,
|
| 1276 |
+
"loss": 0.0003,
|
| 1277 |
+
"step": 4400
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 17.771084337349397,
|
| 1281 |
+
"grad_norm": 0.024144701659679413,
|
| 1282 |
+
"learning_rate": 1.2844444444444445e-06,
|
| 1283 |
+
"loss": 0.0003,
|
| 1284 |
+
"step": 4425
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 17.8714859437751,
|
| 1288 |
+
"grad_norm": 0.02264849655330181,
|
| 1289 |
+
"learning_rate": 1.228888888888889e-06,
|
| 1290 |
+
"loss": 0.0003,
|
| 1291 |
+
"step": 4450
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 17.971887550200805,
|
| 1295 |
+
"grad_norm": 0.02645409293472767,
|
| 1296 |
+
"learning_rate": 1.1733333333333335e-06,
|
| 1297 |
+
"loss": 0.0003,
|
| 1298 |
+
"step": 4475
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 18.072289156626507,
|
| 1302 |
+
"grad_norm": 0.018070507794618607,
|
| 1303 |
+
"learning_rate": 1.117777777777778e-06,
|
| 1304 |
+
"loss": 0.0003,
|
| 1305 |
+
"step": 4500
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 18.17269076305221,
|
| 1309 |
+
"grad_norm": 0.021406978368759155,
|
| 1310 |
+
"learning_rate": 1.0622222222222222e-06,
|
| 1311 |
+
"loss": 0.0003,
|
| 1312 |
+
"step": 4525
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 18.27309236947791,
|
| 1316 |
+
"grad_norm": 0.01881973072886467,
|
| 1317 |
+
"learning_rate": 1.0066666666666668e-06,
|
| 1318 |
+
"loss": 0.0003,
|
| 1319 |
+
"step": 4550
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 18.373493975903614,
|
| 1323 |
+
"grad_norm": 0.018671726807951927,
|
| 1324 |
+
"learning_rate": 9.511111111111111e-07,
|
| 1325 |
+
"loss": 0.0003,
|
| 1326 |
+
"step": 4575
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 18.473895582329316,
|
| 1330 |
+
"grad_norm": 0.014478115364909172,
|
| 1331 |
+
"learning_rate": 8.955555555555557e-07,
|
| 1332 |
+
"loss": 0.0004,
|
| 1333 |
+
"step": 4600
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 18.57429718875502,
|
| 1337 |
+
"grad_norm": 0.02397969737648964,
|
| 1338 |
+
"learning_rate": 8.400000000000001e-07,
|
| 1339 |
+
"loss": 0.0003,
|
| 1340 |
+
"step": 4625
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 18.674698795180724,
|
| 1344 |
+
"grad_norm": 0.020082278177142143,
|
| 1345 |
+
"learning_rate": 7.844444444444445e-07,
|
| 1346 |
+
"loss": 0.0003,
|
| 1347 |
+
"step": 4650
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 18.775100401606426,
|
| 1351 |
+
"grad_norm": 0.02104588970541954,
|
| 1352 |
+
"learning_rate": 7.28888888888889e-07,
|
| 1353 |
+
"loss": 0.0003,
|
| 1354 |
+
"step": 4675
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 18.87550200803213,
|
| 1358 |
+
"grad_norm": 0.018279341980814934,
|
| 1359 |
+
"learning_rate": 6.733333333333334e-07,
|
| 1360 |
+
"loss": 0.0003,
|
| 1361 |
+
"step": 4700
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 18.97590361445783,
|
| 1365 |
+
"grad_norm": 0.021302426233887672,
|
| 1366 |
+
"learning_rate": 6.177777777777778e-07,
|
| 1367 |
+
"loss": 0.0003,
|
| 1368 |
+
"step": 4725
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 19.076305220883533,
|
| 1372 |
+
"grad_norm": 0.020490070804953575,
|
| 1373 |
+
"learning_rate": 5.622222222222223e-07,
|
| 1374 |
+
"loss": 0.0003,
|
| 1375 |
+
"step": 4750
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 19.176706827309236,
|
| 1379 |
+
"grad_norm": 0.020895950496196747,
|
| 1380 |
+
"learning_rate": 5.066666666666667e-07,
|
| 1381 |
+
"loss": 0.0003,
|
| 1382 |
+
"step": 4775
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 19.27710843373494,
|
| 1386 |
+
"grad_norm": 0.01935204304754734,
|
| 1387 |
+
"learning_rate": 4.511111111111111e-07,
|
| 1388 |
+
"loss": 0.0003,
|
| 1389 |
+
"step": 4800
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 19.377510040160644,
|
| 1393 |
+
"grad_norm": 0.016865380108356476,
|
| 1394 |
+
"learning_rate": 3.9555555555555557e-07,
|
| 1395 |
+
"loss": 0.0003,
|
| 1396 |
+
"step": 4825
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 19.477911646586346,
|
| 1400 |
+
"grad_norm": 0.01524381060153246,
|
| 1401 |
+
"learning_rate": 3.4000000000000003e-07,
|
| 1402 |
+
"loss": 0.0003,
|
| 1403 |
+
"step": 4850
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 19.57831325301205,
|
| 1407 |
+
"grad_norm": 0.014133242890238762,
|
| 1408 |
+
"learning_rate": 2.844444444444445e-07,
|
| 1409 |
+
"loss": 0.0003,
|
| 1410 |
+
"step": 4875
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 19.67871485943775,
|
| 1414 |
+
"grad_norm": 0.014944775030016899,
|
| 1415 |
+
"learning_rate": 2.2888888888888892e-07,
|
| 1416 |
+
"loss": 0.0002,
|
| 1417 |
+
"step": 4900
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 19.779116465863453,
|
| 1421 |
+
"grad_norm": 0.016863036900758743,
|
| 1422 |
+
"learning_rate": 1.7333333333333335e-07,
|
| 1423 |
+
"loss": 0.0003,
|
| 1424 |
+
"step": 4925
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 19.879518072289155,
|
| 1428 |
+
"grad_norm": 0.022053442895412445,
|
| 1429 |
+
"learning_rate": 1.1777777777777778e-07,
|
| 1430 |
+
"loss": 0.0003,
|
| 1431 |
+
"step": 4950
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 19.97991967871486,
|
| 1435 |
+
"grad_norm": 0.01967422477900982,
|
| 1436 |
+
"learning_rate": 6.222222222222223e-08,
|
| 1437 |
+
"loss": 0.0003,
|
| 1438 |
+
"step": 4975
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 20.080321285140563,
|
| 1442 |
+
"grad_norm": 0.017572658136487007,
|
| 1443 |
+
"learning_rate": 6.666666666666667e-09,
|
| 1444 |
+
"loss": 0.0003,
|
| 1445 |
+
"step": 5000
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 20.080321285140563,
|
| 1449 |
+
"eval_loss": 0.4997508227825165,
|
| 1450 |
+
"eval_runtime": 2038.971,
|
| 1451 |
+
"eval_samples_per_second": 2.639,
|
| 1452 |
+
"eval_steps_per_second": 0.165,
|
| 1453 |
+
"eval_wer": 0.2181001232653386,
|
| 1454 |
+
"step": 5000
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 20.080321285140563,
|
| 1458 |
+
"step": 5000,
|
| 1459 |
+
"total_flos": 2.727921844224e+20,
|
| 1460 |
+
"train_loss": 0.07613397375550121,
|
| 1461 |
+
"train_runtime": 53449.9289,
|
| 1462 |
+
"train_samples_per_second": 2.993,
|
| 1463 |
+
"train_steps_per_second": 0.094
|
| 1464 |
+
}
|
| 1465 |
+
],
|
| 1466 |
+
"logging_steps": 25,
|
| 1467 |
+
"max_steps": 5000,
|
| 1468 |
+
"num_input_tokens_seen": 0,
|
| 1469 |
+
"num_train_epochs": 21,
|
| 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": 2.727921844224e+20,
|
| 1484 |
+
"train_batch_size": 16,
|
| 1485 |
+
"trial_name": null,
|
| 1486 |
+
"trial_params": null
|
| 1487 |
+
}
|