Commit
•
0e8e70f
1
Parent(s):
765b605
End of training
Browse files- README.md +1 -1
- all_results.json +15 -0
- eval_results.json +9 -0
- runs/Nov07_16-48-55_8cf4d7876f53/events.out.tfevents.1731058123.8cf4d7876f53.36.1 +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-turbo-ft-btb-cv-cy
|
| 18 |
|
| 19 |
-
This model is a fine-tuned version of [openai/whisper-large-v3-turbo](https://huggingface.co/openai/whisper-large-v3-turbo) on
|
| 20 |
It achieves the following results on the evaluation set:
|
| 21 |
- Loss: 0.4072
|
| 22 |
- Wer: 0.2801
|
|
|
|
| 16 |
|
| 17 |
# whisper-large-v3-turbo-ft-btb-cv-cy
|
| 18 |
|
| 19 |
+
This model is a fine-tuned version of [openai/whisper-large-v3-turbo](https://huggingface.co/openai/whisper-large-v3-turbo) on the DewiBrynJones/banc-trawsgrifiadau-bangor-clean train main, DewiBrynJones/commonvoice_18_0_cy train+dev main dataset.
|
| 20 |
It achieves the following results on the evaluation set:
|
| 21 |
- Loss: 0.4072
|
| 22 |
- Wer: 0.2801
|
all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.1595576619273302,
|
| 3 |
+
"eval_loss": 0.4071974754333496,
|
| 4 |
+
"eval_runtime": 1291.4438,
|
| 5 |
+
"eval_samples": 3901,
|
| 6 |
+
"eval_samples_per_second": 3.021,
|
| 7 |
+
"eval_steps_per_second": 0.189,
|
| 8 |
+
"eval_wer": 0.2801361924079977,
|
| 9 |
+
"total_flos": 2.727921844224e+20,
|
| 10 |
+
"train_loss": 0.3573408980369568,
|
| 11 |
+
"train_runtime": 57234.8416,
|
| 12 |
+
"train_samples": 50640,
|
| 13 |
+
"train_samples_per_second": 2.796,
|
| 14 |
+
"train_steps_per_second": 0.087
|
| 15 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.1595576619273302,
|
| 3 |
+
"eval_loss": 0.4071974754333496,
|
| 4 |
+
"eval_runtime": 1291.4438,
|
| 5 |
+
"eval_samples": 3901,
|
| 6 |
+
"eval_samples_per_second": 3.021,
|
| 7 |
+
"eval_steps_per_second": 0.189,
|
| 8 |
+
"eval_wer": 0.2801361924079977
|
| 9 |
+
}
|
runs/Nov07_16-48-55_8cf4d7876f53/events.out.tfevents.1731058123.8cf4d7876f53.36.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4eae7d20a5b201bc954bff1e5da205536fc0d3bf2bf76aaed17506f42caf1424
|
| 3 |
+
size 406
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.1595576619273302,
|
| 3 |
+
"total_flos": 2.727921844224e+20,
|
| 4 |
+
"train_loss": 0.3573408980369568,
|
| 5 |
+
"train_runtime": 57234.8416,
|
| 6 |
+
"train_samples": 50640,
|
| 7 |
+
"train_samples_per_second": 2.796,
|
| 8 |
+
"train_steps_per_second": 0.087
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1487 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.1595576619273302,
|
| 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": 0.01579778830963665,
|
| 13 |
+
"grad_norm": 39.78215789794922,
|
| 14 |
+
"learning_rate": 4.6000000000000004e-07,
|
| 15 |
+
"loss": 1.9397,
|
| 16 |
+
"step": 25
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.0315955766192733,
|
| 20 |
+
"grad_norm": 27.938426971435547,
|
| 21 |
+
"learning_rate": 9.600000000000001e-07,
|
| 22 |
+
"loss": 1.2937,
|
| 23 |
+
"step": 50
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.04739336492890995,
|
| 27 |
+
"grad_norm": 25.550430297851562,
|
| 28 |
+
"learning_rate": 1.46e-06,
|
| 29 |
+
"loss": 0.9786,
|
| 30 |
+
"step": 75
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.0631911532385466,
|
| 34 |
+
"grad_norm": 26.635549545288086,
|
| 35 |
+
"learning_rate": 1.9600000000000003e-06,
|
| 36 |
+
"loss": 0.8818,
|
| 37 |
+
"step": 100
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.07898894154818326,
|
| 41 |
+
"grad_norm": 25.53322410583496,
|
| 42 |
+
"learning_rate": 2.46e-06,
|
| 43 |
+
"loss": 0.8233,
|
| 44 |
+
"step": 125
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.0947867298578199,
|
| 48 |
+
"grad_norm": 26.468778610229492,
|
| 49 |
+
"learning_rate": 2.96e-06,
|
| 50 |
+
"loss": 0.7945,
|
| 51 |
+
"step": 150
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.11058451816745656,
|
| 55 |
+
"grad_norm": 22.71087074279785,
|
| 56 |
+
"learning_rate": 3.46e-06,
|
| 57 |
+
"loss": 0.752,
|
| 58 |
+
"step": 175
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.1263823064770932,
|
| 62 |
+
"grad_norm": 21.494964599609375,
|
| 63 |
+
"learning_rate": 3.96e-06,
|
| 64 |
+
"loss": 0.7655,
|
| 65 |
+
"step": 200
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.14218009478672985,
|
| 69 |
+
"grad_norm": 22.558645248413086,
|
| 70 |
+
"learning_rate": 4.4600000000000005e-06,
|
| 71 |
+
"loss": 0.72,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.1579778830963665,
|
| 76 |
+
"grad_norm": 22.013551712036133,
|
| 77 |
+
"learning_rate": 4.960000000000001e-06,
|
| 78 |
+
"loss": 0.6961,
|
| 79 |
+
"step": 250
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.17377567140600317,
|
| 83 |
+
"grad_norm": 25.887876510620117,
|
| 84 |
+
"learning_rate": 5.460000000000001e-06,
|
| 85 |
+
"loss": 0.6897,
|
| 86 |
+
"step": 275
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.1895734597156398,
|
| 90 |
+
"grad_norm": 19.806230545043945,
|
| 91 |
+
"learning_rate": 5.9600000000000005e-06,
|
| 92 |
+
"loss": 0.6839,
|
| 93 |
+
"step": 300
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.20537124802527645,
|
| 97 |
+
"grad_norm": 21.767576217651367,
|
| 98 |
+
"learning_rate": 6.460000000000001e-06,
|
| 99 |
+
"loss": 0.6767,
|
| 100 |
+
"step": 325
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.2211690363349131,
|
| 104 |
+
"grad_norm": 19.838890075683594,
|
| 105 |
+
"learning_rate": 6.96e-06,
|
| 106 |
+
"loss": 0.6637,
|
| 107 |
+
"step": 350
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.23696682464454977,
|
| 111 |
+
"grad_norm": 22.119140625,
|
| 112 |
+
"learning_rate": 7.4600000000000006e-06,
|
| 113 |
+
"loss": 0.6811,
|
| 114 |
+
"step": 375
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.2527646129541864,
|
| 118 |
+
"grad_norm": 21.972688674926758,
|
| 119 |
+
"learning_rate": 7.960000000000002e-06,
|
| 120 |
+
"loss": 0.6843,
|
| 121 |
+
"step": 400
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.2685624012638231,
|
| 125 |
+
"grad_norm": 21.99839973449707,
|
| 126 |
+
"learning_rate": 8.46e-06,
|
| 127 |
+
"loss": 0.6487,
|
| 128 |
+
"step": 425
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.2843601895734597,
|
| 132 |
+
"grad_norm": 18.968303680419922,
|
| 133 |
+
"learning_rate": 8.96e-06,
|
| 134 |
+
"loss": 0.6528,
|
| 135 |
+
"step": 450
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.3001579778830964,
|
| 139 |
+
"grad_norm": 20.77776336669922,
|
| 140 |
+
"learning_rate": 9.460000000000001e-06,
|
| 141 |
+
"loss": 0.6346,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.315955766192733,
|
| 146 |
+
"grad_norm": 17.947195053100586,
|
| 147 |
+
"learning_rate": 9.960000000000001e-06,
|
| 148 |
+
"loss": 0.6105,
|
| 149 |
+
"step": 500
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.33175355450236965,
|
| 153 |
+
"grad_norm": 17.2167911529541,
|
| 154 |
+
"learning_rate": 9.94888888888889e-06,
|
| 155 |
+
"loss": 0.6399,
|
| 156 |
+
"step": 525
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.34755134281200634,
|
| 160 |
+
"grad_norm": 19.390045166015625,
|
| 161 |
+
"learning_rate": 9.893333333333334e-06,
|
| 162 |
+
"loss": 0.5825,
|
| 163 |
+
"step": 550
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.36334913112164297,
|
| 167 |
+
"grad_norm": 17.195106506347656,
|
| 168 |
+
"learning_rate": 9.837777777777778e-06,
|
| 169 |
+
"loss": 0.5895,
|
| 170 |
+
"step": 575
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.3791469194312796,
|
| 174 |
+
"grad_norm": 18.29102325439453,
|
| 175 |
+
"learning_rate": 9.782222222222222e-06,
|
| 176 |
+
"loss": 0.621,
|
| 177 |
+
"step": 600
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.3949447077409163,
|
| 181 |
+
"grad_norm": 16.213546752929688,
|
| 182 |
+
"learning_rate": 9.726666666666668e-06,
|
| 183 |
+
"loss": 0.587,
|
| 184 |
+
"step": 625
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.4107424960505529,
|
| 188 |
+
"grad_norm": 15.86738109588623,
|
| 189 |
+
"learning_rate": 9.671111111111112e-06,
|
| 190 |
+
"loss": 0.5765,
|
| 191 |
+
"step": 650
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.4265402843601896,
|
| 195 |
+
"grad_norm": 17.446897506713867,
|
| 196 |
+
"learning_rate": 9.617777777777778e-06,
|
| 197 |
+
"loss": 0.5914,
|
| 198 |
+
"step": 675
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.4423380726698262,
|
| 202 |
+
"grad_norm": 15.107172966003418,
|
| 203 |
+
"learning_rate": 9.562222222222223e-06,
|
| 204 |
+
"loss": 0.5517,
|
| 205 |
+
"step": 700
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.45813586097946285,
|
| 209 |
+
"grad_norm": 16.07261848449707,
|
| 210 |
+
"learning_rate": 9.506666666666667e-06,
|
| 211 |
+
"loss": 0.5523,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.47393364928909953,
|
| 216 |
+
"grad_norm": 15.2976655960083,
|
| 217 |
+
"learning_rate": 9.451111111111112e-06,
|
| 218 |
+
"loss": 0.5419,
|
| 219 |
+
"step": 750
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.48973143759873616,
|
| 223 |
+
"grad_norm": 16.560869216918945,
|
| 224 |
+
"learning_rate": 9.395555555555556e-06,
|
| 225 |
+
"loss": 0.5763,
|
| 226 |
+
"step": 775
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.5055292259083728,
|
| 230 |
+
"grad_norm": 18.58966064453125,
|
| 231 |
+
"learning_rate": 9.340000000000002e-06,
|
| 232 |
+
"loss": 0.5497,
|
| 233 |
+
"step": 800
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.5213270142180095,
|
| 237 |
+
"grad_norm": 16.940420150756836,
|
| 238 |
+
"learning_rate": 9.284444444444444e-06,
|
| 239 |
+
"loss": 0.5177,
|
| 240 |
+
"step": 825
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.5371248025276462,
|
| 244 |
+
"grad_norm": 12.964641571044922,
|
| 245 |
+
"learning_rate": 9.22888888888889e-06,
|
| 246 |
+
"loss": 0.5293,
|
| 247 |
+
"step": 850
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.5529225908372828,
|
| 251 |
+
"grad_norm": 16.428470611572266,
|
| 252 |
+
"learning_rate": 9.173333333333334e-06,
|
| 253 |
+
"loss": 0.5062,
|
| 254 |
+
"step": 875
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.5687203791469194,
|
| 258 |
+
"grad_norm": 17.921024322509766,
|
| 259 |
+
"learning_rate": 9.117777777777778e-06,
|
| 260 |
+
"loss": 0.5471,
|
| 261 |
+
"step": 900
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.584518167456556,
|
| 265 |
+
"grad_norm": 14.068251609802246,
|
| 266 |
+
"learning_rate": 9.062222222222224e-06,
|
| 267 |
+
"loss": 0.5072,
|
| 268 |
+
"step": 925
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.6003159557661928,
|
| 272 |
+
"grad_norm": 13.403594017028809,
|
| 273 |
+
"learning_rate": 9.006666666666666e-06,
|
| 274 |
+
"loss": 0.5051,
|
| 275 |
+
"step": 950
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.6161137440758294,
|
| 279 |
+
"grad_norm": 14.557646751403809,
|
| 280 |
+
"learning_rate": 8.951111111111112e-06,
|
| 281 |
+
"loss": 0.5391,
|
| 282 |
+
"step": 975
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.631911532385466,
|
| 286 |
+
"grad_norm": 15.385436058044434,
|
| 287 |
+
"learning_rate": 8.895555555555556e-06,
|
| 288 |
+
"loss": 0.5037,
|
| 289 |
+
"step": 1000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.631911532385466,
|
| 293 |
+
"eval_loss": 0.551193118095398,
|
| 294 |
+
"eval_runtime": 1363.4745,
|
| 295 |
+
"eval_samples_per_second": 2.861,
|
| 296 |
+
"eval_steps_per_second": 0.179,
|
| 297 |
+
"eval_wer": 0.38095238095238093,
|
| 298 |
+
"step": 1000
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.6477093206951027,
|
| 302 |
+
"grad_norm": 13.815781593322754,
|
| 303 |
+
"learning_rate": 8.84e-06,
|
| 304 |
+
"loss": 0.4888,
|
| 305 |
+
"step": 1025
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.6635071090047393,
|
| 309 |
+
"grad_norm": 16.213642120361328,
|
| 310 |
+
"learning_rate": 8.784444444444446e-06,
|
| 311 |
+
"loss": 0.4865,
|
| 312 |
+
"step": 1050
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.6793048973143759,
|
| 316 |
+
"grad_norm": 14.779074668884277,
|
| 317 |
+
"learning_rate": 8.72888888888889e-06,
|
| 318 |
+
"loss": 0.4963,
|
| 319 |
+
"step": 1075
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.6951026856240127,
|
| 323 |
+
"grad_norm": 13.794785499572754,
|
| 324 |
+
"learning_rate": 8.673333333333334e-06,
|
| 325 |
+
"loss": 0.482,
|
| 326 |
+
"step": 1100
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.7109004739336493,
|
| 330 |
+
"grad_norm": 14.701066970825195,
|
| 331 |
+
"learning_rate": 8.617777777777778e-06,
|
| 332 |
+
"loss": 0.5027,
|
| 333 |
+
"step": 1125
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.7266982622432859,
|
| 337 |
+
"grad_norm": 14.537113189697266,
|
| 338 |
+
"learning_rate": 8.562222222222224e-06,
|
| 339 |
+
"loss": 0.4798,
|
| 340 |
+
"step": 1150
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.7424960505529226,
|
| 344 |
+
"grad_norm": 14.886212348937988,
|
| 345 |
+
"learning_rate": 8.506666666666668e-06,
|
| 346 |
+
"loss": 0.4777,
|
| 347 |
+
"step": 1175
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 0.7582938388625592,
|
| 351 |
+
"grad_norm": 13.01016616821289,
|
| 352 |
+
"learning_rate": 8.451111111111112e-06,
|
| 353 |
+
"loss": 0.4736,
|
| 354 |
+
"step": 1200
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.7740916271721959,
|
| 358 |
+
"grad_norm": 14.213628768920898,
|
| 359 |
+
"learning_rate": 8.395555555555557e-06,
|
| 360 |
+
"loss": 0.4808,
|
| 361 |
+
"step": 1225
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 0.7898894154818326,
|
| 365 |
+
"grad_norm": 14.541191101074219,
|
| 366 |
+
"learning_rate": 8.34e-06,
|
| 367 |
+
"loss": 0.5008,
|
| 368 |
+
"step": 1250
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 0.8056872037914692,
|
| 372 |
+
"grad_norm": 14.092310905456543,
|
| 373 |
+
"learning_rate": 8.284444444444446e-06,
|
| 374 |
+
"loss": 0.4693,
|
| 375 |
+
"step": 1275
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.8214849921011058,
|
| 379 |
+
"grad_norm": 14.289324760437012,
|
| 380 |
+
"learning_rate": 8.22888888888889e-06,
|
| 381 |
+
"loss": 0.4418,
|
| 382 |
+
"step": 1300
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 0.8372827804107424,
|
| 386 |
+
"grad_norm": 13.257657051086426,
|
| 387 |
+
"learning_rate": 8.173333333333334e-06,
|
| 388 |
+
"loss": 0.4643,
|
| 389 |
+
"step": 1325
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 0.8530805687203792,
|
| 393 |
+
"grad_norm": 15.05517864227295,
|
| 394 |
+
"learning_rate": 8.11777777777778e-06,
|
| 395 |
+
"loss": 0.4751,
|
| 396 |
+
"step": 1350
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 0.8688783570300158,
|
| 400 |
+
"grad_norm": 13.352594375610352,
|
| 401 |
+
"learning_rate": 8.062222222222222e-06,
|
| 402 |
+
"loss": 0.442,
|
| 403 |
+
"step": 1375
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 0.8846761453396524,
|
| 407 |
+
"grad_norm": 12.487988471984863,
|
| 408 |
+
"learning_rate": 8.006666666666667e-06,
|
| 409 |
+
"loss": 0.4496,
|
| 410 |
+
"step": 1400
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.9004739336492891,
|
| 414 |
+
"grad_norm": 13.963912963867188,
|
| 415 |
+
"learning_rate": 7.951111111111111e-06,
|
| 416 |
+
"loss": 0.4415,
|
| 417 |
+
"step": 1425
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.9162717219589257,
|
| 421 |
+
"grad_norm": 13.90829086303711,
|
| 422 |
+
"learning_rate": 7.895555555555557e-06,
|
| 423 |
+
"loss": 0.4349,
|
| 424 |
+
"step": 1450
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 0.9320695102685624,
|
| 428 |
+
"grad_norm": 16.863481521606445,
|
| 429 |
+
"learning_rate": 7.840000000000001e-06,
|
| 430 |
+
"loss": 0.4722,
|
| 431 |
+
"step": 1475
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.9478672985781991,
|
| 435 |
+
"grad_norm": 13.991209983825684,
|
| 436 |
+
"learning_rate": 7.784444444444445e-06,
|
| 437 |
+
"loss": 0.4397,
|
| 438 |
+
"step": 1500
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 0.9636650868878357,
|
| 442 |
+
"grad_norm": 12.423737525939941,
|
| 443 |
+
"learning_rate": 7.72888888888889e-06,
|
| 444 |
+
"loss": 0.423,
|
| 445 |
+
"step": 1525
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 0.9794628751974723,
|
| 449 |
+
"grad_norm": 13.574849128723145,
|
| 450 |
+
"learning_rate": 7.673333333333333e-06,
|
| 451 |
+
"loss": 0.4472,
|
| 452 |
+
"step": 1550
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 0.995260663507109,
|
| 456 |
+
"grad_norm": 10.542879104614258,
|
| 457 |
+
"learning_rate": 7.617777777777778e-06,
|
| 458 |
+
"loss": 0.4319,
|
| 459 |
+
"step": 1575
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 1.0110584518167456,
|
| 463 |
+
"grad_norm": 12.750446319580078,
|
| 464 |
+
"learning_rate": 7.562222222222223e-06,
|
| 465 |
+
"loss": 0.3401,
|
| 466 |
+
"step": 1600
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 1.0268562401263823,
|
| 470 |
+
"grad_norm": 10.037585258483887,
|
| 471 |
+
"learning_rate": 7.506666666666668e-06,
|
| 472 |
+
"loss": 0.3236,
|
| 473 |
+
"step": 1625
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 1.042654028436019,
|
| 477 |
+
"grad_norm": 12.519506454467773,
|
| 478 |
+
"learning_rate": 7.451111111111111e-06,
|
| 479 |
+
"loss": 0.3379,
|
| 480 |
+
"step": 1650
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 1.0584518167456556,
|
| 484 |
+
"grad_norm": 11.666909217834473,
|
| 485 |
+
"learning_rate": 7.395555555555556e-06,
|
| 486 |
+
"loss": 0.3082,
|
| 487 |
+
"step": 1675
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 1.0742496050552923,
|
| 491 |
+
"grad_norm": 14.614953994750977,
|
| 492 |
+
"learning_rate": 7.340000000000001e-06,
|
| 493 |
+
"loss": 0.3272,
|
| 494 |
+
"step": 1700
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 1.0900473933649288,
|
| 498 |
+
"grad_norm": 11.243821144104004,
|
| 499 |
+
"learning_rate": 7.284444444444445e-06,
|
| 500 |
+
"loss": 0.3081,
|
| 501 |
+
"step": 1725
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 1.1058451816745656,
|
| 505 |
+
"grad_norm": 10.896337509155273,
|
| 506 |
+
"learning_rate": 7.22888888888889e-06,
|
| 507 |
+
"loss": 0.3212,
|
| 508 |
+
"step": 1750
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 1.1216429699842023,
|
| 512 |
+
"grad_norm": 11.159407615661621,
|
| 513 |
+
"learning_rate": 7.173333333333335e-06,
|
| 514 |
+
"loss": 0.3316,
|
| 515 |
+
"step": 1775
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 1.1374407582938388,
|
| 519 |
+
"grad_norm": 10.72888469696045,
|
| 520 |
+
"learning_rate": 7.117777777777778e-06,
|
| 521 |
+
"loss": 0.3465,
|
| 522 |
+
"step": 1800
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 1.1532385466034756,
|
| 526 |
+
"grad_norm": 11.136720657348633,
|
| 527 |
+
"learning_rate": 7.062222222222223e-06,
|
| 528 |
+
"loss": 0.3219,
|
| 529 |
+
"step": 1825
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 1.169036334913112,
|
| 533 |
+
"grad_norm": 11.062825202941895,
|
| 534 |
+
"learning_rate": 7.006666666666667e-06,
|
| 535 |
+
"loss": 0.3234,
|
| 536 |
+
"step": 1850
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 1.1848341232227488,
|
| 540 |
+
"grad_norm": 13.367472648620605,
|
| 541 |
+
"learning_rate": 6.951111111111112e-06,
|
| 542 |
+
"loss": 0.3215,
|
| 543 |
+
"step": 1875
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 1.2006319115323856,
|
| 547 |
+
"grad_norm": 12.034098625183105,
|
| 548 |
+
"learning_rate": 6.8955555555555565e-06,
|
| 549 |
+
"loss": 0.3195,
|
| 550 |
+
"step": 1900
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 1.216429699842022,
|
| 554 |
+
"grad_norm": 12.74405574798584,
|
| 555 |
+
"learning_rate": 6.8400000000000014e-06,
|
| 556 |
+
"loss": 0.3105,
|
| 557 |
+
"step": 1925
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 1.2322274881516588,
|
| 561 |
+
"grad_norm": 14.234502792358398,
|
| 562 |
+
"learning_rate": 6.784444444444445e-06,
|
| 563 |
+
"loss": 0.3327,
|
| 564 |
+
"step": 1950
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.2480252764612954,
|
| 568 |
+
"grad_norm": 12.721147537231445,
|
| 569 |
+
"learning_rate": 6.7288888888888895e-06,
|
| 570 |
+
"loss": 0.3405,
|
| 571 |
+
"step": 1975
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.263823064770932,
|
| 575 |
+
"grad_norm": 12.109272003173828,
|
| 576 |
+
"learning_rate": 6.6733333333333335e-06,
|
| 577 |
+
"loss": 0.3264,
|
| 578 |
+
"step": 2000
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 1.263823064770932,
|
| 582 |
+
"eval_loss": 0.4630681872367859,
|
| 583 |
+
"eval_runtime": 1327.3738,
|
| 584 |
+
"eval_samples_per_second": 2.939,
|
| 585 |
+
"eval_steps_per_second": 0.184,
|
| 586 |
+
"eval_wer": 0.331425673717763,
|
| 587 |
+
"step": 2000
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 1.2796208530805688,
|
| 591 |
+
"grad_norm": 10.781163215637207,
|
| 592 |
+
"learning_rate": 6.617777777777778e-06,
|
| 593 |
+
"loss": 0.3152,
|
| 594 |
+
"step": 2025
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 1.2954186413902053,
|
| 598 |
+
"grad_norm": 12.081149101257324,
|
| 599 |
+
"learning_rate": 6.562222222222223e-06,
|
| 600 |
+
"loss": 0.334,
|
| 601 |
+
"step": 2050
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 1.311216429699842,
|
| 605 |
+
"grad_norm": 11.082480430603027,
|
| 606 |
+
"learning_rate": 6.5066666666666665e-06,
|
| 607 |
+
"loss": 0.3182,
|
| 608 |
+
"step": 2075
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 1.3270142180094786,
|
| 612 |
+
"grad_norm": 10.400703430175781,
|
| 613 |
+
"learning_rate": 6.451111111111111e-06,
|
| 614 |
+
"loss": 0.3209,
|
| 615 |
+
"step": 2100
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 1.3428120063191153,
|
| 619 |
+
"grad_norm": 14.698405265808105,
|
| 620 |
+
"learning_rate": 6.395555555555556e-06,
|
| 621 |
+
"loss": 0.3358,
|
| 622 |
+
"step": 2125
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 1.358609794628752,
|
| 626 |
+
"grad_norm": 11.301855087280273,
|
| 627 |
+
"learning_rate": 6.34e-06,
|
| 628 |
+
"loss": 0.3274,
|
| 629 |
+
"step": 2150
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 1.3744075829383886,
|
| 633 |
+
"grad_norm": 13.15268611907959,
|
| 634 |
+
"learning_rate": 6.284444444444445e-06,
|
| 635 |
+
"loss": 0.3091,
|
| 636 |
+
"step": 2175
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 1.3902053712480253,
|
| 640 |
+
"grad_norm": 10.712764739990234,
|
| 641 |
+
"learning_rate": 6.22888888888889e-06,
|
| 642 |
+
"loss": 0.3217,
|
| 643 |
+
"step": 2200
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 1.4060031595576619,
|
| 647 |
+
"grad_norm": 9.865320205688477,
|
| 648 |
+
"learning_rate": 6.173333333333333e-06,
|
| 649 |
+
"loss": 0.3341,
|
| 650 |
+
"step": 2225
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 1.4218009478672986,
|
| 654 |
+
"grad_norm": 11.386091232299805,
|
| 655 |
+
"learning_rate": 6.117777777777778e-06,
|
| 656 |
+
"loss": 0.308,
|
| 657 |
+
"step": 2250
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 1.4375987361769353,
|
| 661 |
+
"grad_norm": 10.972213745117188,
|
| 662 |
+
"learning_rate": 6.062222222222223e-06,
|
| 663 |
+
"loss": 0.3184,
|
| 664 |
+
"step": 2275
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 1.4533965244865719,
|
| 668 |
+
"grad_norm": 9.059267044067383,
|
| 669 |
+
"learning_rate": 6.006666666666667e-06,
|
| 670 |
+
"loss": 0.3285,
|
| 671 |
+
"step": 2300
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 1.4691943127962086,
|
| 675 |
+
"grad_norm": 10.708878517150879,
|
| 676 |
+
"learning_rate": 5.951111111111112e-06,
|
| 677 |
+
"loss": 0.288,
|
| 678 |
+
"step": 2325
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 1.4849921011058451,
|
| 682 |
+
"grad_norm": 9.580412864685059,
|
| 683 |
+
"learning_rate": 5.895555555555557e-06,
|
| 684 |
+
"loss": 0.3133,
|
| 685 |
+
"step": 2350
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 1.5007898894154819,
|
| 689 |
+
"grad_norm": 13.611145973205566,
|
| 690 |
+
"learning_rate": 5.84e-06,
|
| 691 |
+
"loss": 0.3398,
|
| 692 |
+
"step": 2375
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 1.5165876777251186,
|
| 696 |
+
"grad_norm": 11.563826560974121,
|
| 697 |
+
"learning_rate": 5.784444444444445e-06,
|
| 698 |
+
"loss": 0.3187,
|
| 699 |
+
"step": 2400
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 1.5323854660347551,
|
| 703 |
+
"grad_norm": 11.612549781799316,
|
| 704 |
+
"learning_rate": 5.72888888888889e-06,
|
| 705 |
+
"loss": 0.3169,
|
| 706 |
+
"step": 2425
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 1.5481832543443916,
|
| 710 |
+
"grad_norm": 9.463521003723145,
|
| 711 |
+
"learning_rate": 5.673333333333334e-06,
|
| 712 |
+
"loss": 0.3124,
|
| 713 |
+
"step": 2450
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 1.5639810426540284,
|
| 717 |
+
"grad_norm": 14.442208290100098,
|
| 718 |
+
"learning_rate": 5.617777777777779e-06,
|
| 719 |
+
"loss": 0.3202,
|
| 720 |
+
"step": 2475
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 1.5797788309636651,
|
| 724 |
+
"grad_norm": 11.377900123596191,
|
| 725 |
+
"learning_rate": 5.562222222222222e-06,
|
| 726 |
+
"loss": 0.317,
|
| 727 |
+
"step": 2500
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 1.5955766192733019,
|
| 731 |
+
"grad_norm": 11.9774808883667,
|
| 732 |
+
"learning_rate": 5.506666666666667e-06,
|
| 733 |
+
"loss": 0.3188,
|
| 734 |
+
"step": 2525
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 1.6113744075829384,
|
| 738 |
+
"grad_norm": 11.492107391357422,
|
| 739 |
+
"learning_rate": 5.451111111111112e-06,
|
| 740 |
+
"loss": 0.3109,
|
| 741 |
+
"step": 2550
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 1.627172195892575,
|
| 745 |
+
"grad_norm": 12.987119674682617,
|
| 746 |
+
"learning_rate": 5.3955555555555565e-06,
|
| 747 |
+
"loss": 0.3127,
|
| 748 |
+
"step": 2575
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 1.6429699842022116,
|
| 752 |
+
"grad_norm": 11.403392791748047,
|
| 753 |
+
"learning_rate": 5.3400000000000005e-06,
|
| 754 |
+
"loss": 0.3007,
|
| 755 |
+
"step": 2600
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 1.6587677725118484,
|
| 759 |
+
"grad_norm": 10.354674339294434,
|
| 760 |
+
"learning_rate": 5.2844444444444454e-06,
|
| 761 |
+
"loss": 0.3129,
|
| 762 |
+
"step": 2625
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 1.674565560821485,
|
| 766 |
+
"grad_norm": 9.936201095581055,
|
| 767 |
+
"learning_rate": 5.228888888888889e-06,
|
| 768 |
+
"loss": 0.3226,
|
| 769 |
+
"step": 2650
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 1.6903633491311216,
|
| 773 |
+
"grad_norm": 10.426243782043457,
|
| 774 |
+
"learning_rate": 5.1733333333333335e-06,
|
| 775 |
+
"loss": 0.3021,
|
| 776 |
+
"step": 2675
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 1.7061611374407581,
|
| 780 |
+
"grad_norm": 10.784858703613281,
|
| 781 |
+
"learning_rate": 5.117777777777778e-06,
|
| 782 |
+
"loss": 0.2923,
|
| 783 |
+
"step": 2700
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 1.7219589257503949,
|
| 787 |
+
"grad_norm": 10.411828994750977,
|
| 788 |
+
"learning_rate": 5.062222222222222e-06,
|
| 789 |
+
"loss": 0.2893,
|
| 790 |
+
"step": 2725
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 1.7377567140600316,
|
| 794 |
+
"grad_norm": 12.027934074401855,
|
| 795 |
+
"learning_rate": 5.006666666666667e-06,
|
| 796 |
+
"loss": 0.3136,
|
| 797 |
+
"step": 2750
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 1.7535545023696684,
|
| 801 |
+
"grad_norm": 10.067774772644043,
|
| 802 |
+
"learning_rate": 4.951111111111111e-06,
|
| 803 |
+
"loss": 0.296,
|
| 804 |
+
"step": 2775
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 1.7693522906793049,
|
| 808 |
+
"grad_norm": 10.396674156188965,
|
| 809 |
+
"learning_rate": 4.895555555555556e-06,
|
| 810 |
+
"loss": 0.2973,
|
| 811 |
+
"step": 2800
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 1.7851500789889414,
|
| 815 |
+
"grad_norm": 9.719764709472656,
|
| 816 |
+
"learning_rate": 4.84e-06,
|
| 817 |
+
"loss": 0.2831,
|
| 818 |
+
"step": 2825
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 1.8009478672985781,
|
| 822 |
+
"grad_norm": 11.552470207214355,
|
| 823 |
+
"learning_rate": 4.784444444444445e-06,
|
| 824 |
+
"loss": 0.2802,
|
| 825 |
+
"step": 2850
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 1.8167456556082149,
|
| 829 |
+
"grad_norm": 10.932677268981934,
|
| 830 |
+
"learning_rate": 4.728888888888889e-06,
|
| 831 |
+
"loss": 0.3189,
|
| 832 |
+
"step": 2875
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 1.8325434439178516,
|
| 836 |
+
"grad_norm": 12.281967163085938,
|
| 837 |
+
"learning_rate": 4.673333333333333e-06,
|
| 838 |
+
"loss": 0.3014,
|
| 839 |
+
"step": 2900
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 1.8483412322274881,
|
| 843 |
+
"grad_norm": 12.78361988067627,
|
| 844 |
+
"learning_rate": 4.617777777777778e-06,
|
| 845 |
+
"loss": 0.3265,
|
| 846 |
+
"step": 2925
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 1.8641390205371247,
|
| 850 |
+
"grad_norm": 11.523568153381348,
|
| 851 |
+
"learning_rate": 4.562222222222222e-06,
|
| 852 |
+
"loss": 0.3062,
|
| 853 |
+
"step": 2950
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 1.8799368088467614,
|
| 857 |
+
"grad_norm": 12.009855270385742,
|
| 858 |
+
"learning_rate": 4.506666666666667e-06,
|
| 859 |
+
"loss": 0.294,
|
| 860 |
+
"step": 2975
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 1.8957345971563981,
|
| 864 |
+
"grad_norm": 11.980591773986816,
|
| 865 |
+
"learning_rate": 4.451111111111112e-06,
|
| 866 |
+
"loss": 0.2997,
|
| 867 |
+
"step": 3000
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 1.8957345971563981,
|
| 871 |
+
"eval_loss": 0.420254111289978,
|
| 872 |
+
"eval_runtime": 1304.4467,
|
| 873 |
+
"eval_samples_per_second": 2.991,
|
| 874 |
+
"eval_steps_per_second": 0.187,
|
| 875 |
+
"eval_wer": 0.30416304452815607,
|
| 876 |
+
"step": 3000
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 1.9115323854660349,
|
| 880 |
+
"grad_norm": 9.556490898132324,
|
| 881 |
+
"learning_rate": 4.395555555555556e-06,
|
| 882 |
+
"loss": 0.3128,
|
| 883 |
+
"step": 3025
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 1.9273301737756714,
|
| 887 |
+
"grad_norm": 12.445517539978027,
|
| 888 |
+
"learning_rate": 4.34e-06,
|
| 889 |
+
"loss": 0.2898,
|
| 890 |
+
"step": 3050
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 1.943127962085308,
|
| 894 |
+
"grad_norm": 11.485321998596191,
|
| 895 |
+
"learning_rate": 4.284444444444445e-06,
|
| 896 |
+
"loss": 0.2922,
|
| 897 |
+
"step": 3075
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 1.9589257503949447,
|
| 901 |
+
"grad_norm": 11.148347854614258,
|
| 902 |
+
"learning_rate": 4.228888888888889e-06,
|
| 903 |
+
"loss": 0.2941,
|
| 904 |
+
"step": 3100
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 1.9747235387045814,
|
| 908 |
+
"grad_norm": 8.813661575317383,
|
| 909 |
+
"learning_rate": 4.173333333333334e-06,
|
| 910 |
+
"loss": 0.2738,
|
| 911 |
+
"step": 3125
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 1.9905213270142181,
|
| 915 |
+
"grad_norm": 9.704155921936035,
|
| 916 |
+
"learning_rate": 4.117777777777779e-06,
|
| 917 |
+
"loss": 0.3064,
|
| 918 |
+
"step": 3150
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 2.006319115323855,
|
| 922 |
+
"grad_norm": 8.704689979553223,
|
| 923 |
+
"learning_rate": 4.062222222222223e-06,
|
| 924 |
+
"loss": 0.2478,
|
| 925 |
+
"step": 3175
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 2.022116903633491,
|
| 929 |
+
"grad_norm": 10.47323226928711,
|
| 930 |
+
"learning_rate": 4.006666666666667e-06,
|
| 931 |
+
"loss": 0.1972,
|
| 932 |
+
"step": 3200
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 2.037914691943128,
|
| 936 |
+
"grad_norm": 9.401123046875,
|
| 937 |
+
"learning_rate": 3.953333333333333e-06,
|
| 938 |
+
"loss": 0.2048,
|
| 939 |
+
"step": 3225
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 2.0537124802527646,
|
| 943 |
+
"grad_norm": 8.968587875366211,
|
| 944 |
+
"learning_rate": 3.897777777777778e-06,
|
| 945 |
+
"loss": 0.2047,
|
| 946 |
+
"step": 3250
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 2.0695102685624014,
|
| 950 |
+
"grad_norm": 10.138334274291992,
|
| 951 |
+
"learning_rate": 3.842222222222223e-06,
|
| 952 |
+
"loss": 0.2041,
|
| 953 |
+
"step": 3275
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 2.085308056872038,
|
| 957 |
+
"grad_norm": 9.157357215881348,
|
| 958 |
+
"learning_rate": 3.7866666666666667e-06,
|
| 959 |
+
"loss": 0.1983,
|
| 960 |
+
"step": 3300
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 2.1011058451816744,
|
| 964 |
+
"grad_norm": 11.541014671325684,
|
| 965 |
+
"learning_rate": 3.7311111111111116e-06,
|
| 966 |
+
"loss": 0.2102,
|
| 967 |
+
"step": 3325
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 2.116903633491311,
|
| 971 |
+
"grad_norm": 13.219682693481445,
|
| 972 |
+
"learning_rate": 3.675555555555556e-06,
|
| 973 |
+
"loss": 0.2032,
|
| 974 |
+
"step": 3350
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 2.132701421800948,
|
| 978 |
+
"grad_norm": 9.11551284790039,
|
| 979 |
+
"learning_rate": 3.62e-06,
|
| 980 |
+
"loss": 0.183,
|
| 981 |
+
"step": 3375
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 2.1484992101105846,
|
| 985 |
+
"grad_norm": 7.927098751068115,
|
| 986 |
+
"learning_rate": 3.564444444444445e-06,
|
| 987 |
+
"loss": 0.2022,
|
| 988 |
+
"step": 3400
|
| 989 |
+
},
|
| 990 |
+
{
|
| 991 |
+
"epoch": 2.1642969984202214,
|
| 992 |
+
"grad_norm": 10.464715003967285,
|
| 993 |
+
"learning_rate": 3.508888888888889e-06,
|
| 994 |
+
"loss": 0.1877,
|
| 995 |
+
"step": 3425
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 2.1800947867298577,
|
| 999 |
+
"grad_norm": 8.424384117126465,
|
| 1000 |
+
"learning_rate": 3.4533333333333334e-06,
|
| 1001 |
+
"loss": 0.1879,
|
| 1002 |
+
"step": 3450
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 2.1958925750394944,
|
| 1006 |
+
"grad_norm": 7.617179870605469,
|
| 1007 |
+
"learning_rate": 3.3977777777777783e-06,
|
| 1008 |
+
"loss": 0.1973,
|
| 1009 |
+
"step": 3475
|
| 1010 |
+
},
|
| 1011 |
+
{
|
| 1012 |
+
"epoch": 2.211690363349131,
|
| 1013 |
+
"grad_norm": 9.299885749816895,
|
| 1014 |
+
"learning_rate": 3.3422222222222224e-06,
|
| 1015 |
+
"loss": 0.1982,
|
| 1016 |
+
"step": 3500
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 2.227488151658768,
|
| 1020 |
+
"grad_norm": 8.834092140197754,
|
| 1021 |
+
"learning_rate": 3.286666666666667e-06,
|
| 1022 |
+
"loss": 0.2003,
|
| 1023 |
+
"step": 3525
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 2.2432859399684046,
|
| 1027 |
+
"grad_norm": 8.07299518585205,
|
| 1028 |
+
"learning_rate": 3.2311111111111117e-06,
|
| 1029 |
+
"loss": 0.1971,
|
| 1030 |
+
"step": 3550
|
| 1031 |
+
},
|
| 1032 |
+
{
|
| 1033 |
+
"epoch": 2.259083728278041,
|
| 1034 |
+
"grad_norm": 10.275826454162598,
|
| 1035 |
+
"learning_rate": 3.1755555555555557e-06,
|
| 1036 |
+
"loss": 0.1914,
|
| 1037 |
+
"step": 3575
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 2.2748815165876777,
|
| 1041 |
+
"grad_norm": 9.910749435424805,
|
| 1042 |
+
"learning_rate": 3.12e-06,
|
| 1043 |
+
"loss": 0.2067,
|
| 1044 |
+
"step": 3600
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 2.2906793048973144,
|
| 1048 |
+
"grad_norm": 10.053370475769043,
|
| 1049 |
+
"learning_rate": 3.064444444444445e-06,
|
| 1050 |
+
"loss": 0.2129,
|
| 1051 |
+
"step": 3625
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 2.306477093206951,
|
| 1055 |
+
"grad_norm": 10.744956970214844,
|
| 1056 |
+
"learning_rate": 3.008888888888889e-06,
|
| 1057 |
+
"loss": 0.1984,
|
| 1058 |
+
"step": 3650
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 2.322274881516588,
|
| 1062 |
+
"grad_norm": 9.880094528198242,
|
| 1063 |
+
"learning_rate": 2.9533333333333336e-06,
|
| 1064 |
+
"loss": 0.1852,
|
| 1065 |
+
"step": 3675
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 2.338072669826224,
|
| 1069 |
+
"grad_norm": 10.811684608459473,
|
| 1070 |
+
"learning_rate": 2.8977777777777785e-06,
|
| 1071 |
+
"loss": 0.2019,
|
| 1072 |
+
"step": 3700
|
| 1073 |
+
},
|
| 1074 |
+
{
|
| 1075 |
+
"epoch": 2.353870458135861,
|
| 1076 |
+
"grad_norm": 12.169087409973145,
|
| 1077 |
+
"learning_rate": 2.8422222222222225e-06,
|
| 1078 |
+
"loss": 0.1944,
|
| 1079 |
+
"step": 3725
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 2.3696682464454977,
|
| 1083 |
+
"grad_norm": 10.1768217086792,
|
| 1084 |
+
"learning_rate": 2.786666666666667e-06,
|
| 1085 |
+
"loss": 0.2035,
|
| 1086 |
+
"step": 3750
|
| 1087 |
+
},
|
| 1088 |
+
{
|
| 1089 |
+
"epoch": 2.3854660347551344,
|
| 1090 |
+
"grad_norm": 8.389801979064941,
|
| 1091 |
+
"learning_rate": 2.7311111111111114e-06,
|
| 1092 |
+
"loss": 0.2096,
|
| 1093 |
+
"step": 3775
|
| 1094 |
+
},
|
| 1095 |
+
{
|
| 1096 |
+
"epoch": 2.401263823064771,
|
| 1097 |
+
"grad_norm": 9.511481285095215,
|
| 1098 |
+
"learning_rate": 2.675555555555556e-06,
|
| 1099 |
+
"loss": 0.1868,
|
| 1100 |
+
"step": 3800
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 2.4170616113744074,
|
| 1104 |
+
"grad_norm": 10.304895401000977,
|
| 1105 |
+
"learning_rate": 2.6200000000000003e-06,
|
| 1106 |
+
"loss": 0.2023,
|
| 1107 |
+
"step": 3825
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 2.432859399684044,
|
| 1111 |
+
"grad_norm": 11.822694778442383,
|
| 1112 |
+
"learning_rate": 2.5644444444444444e-06,
|
| 1113 |
+
"loss": 0.1938,
|
| 1114 |
+
"step": 3850
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 2.448657187993681,
|
| 1118 |
+
"grad_norm": 10.087789535522461,
|
| 1119 |
+
"learning_rate": 2.5088888888888892e-06,
|
| 1120 |
+
"loss": 0.1987,
|
| 1121 |
+
"step": 3875
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 2.4644549763033177,
|
| 1125 |
+
"grad_norm": 8.504409790039062,
|
| 1126 |
+
"learning_rate": 2.4533333333333333e-06,
|
| 1127 |
+
"loss": 0.1845,
|
| 1128 |
+
"step": 3900
|
| 1129 |
+
},
|
| 1130 |
+
{
|
| 1131 |
+
"epoch": 2.4802527646129544,
|
| 1132 |
+
"grad_norm": 9.70301342010498,
|
| 1133 |
+
"learning_rate": 2.397777777777778e-06,
|
| 1134 |
+
"loss": 0.1832,
|
| 1135 |
+
"step": 3925
|
| 1136 |
+
},
|
| 1137 |
+
{
|
| 1138 |
+
"epoch": 2.4960505529225907,
|
| 1139 |
+
"grad_norm": 9.534614562988281,
|
| 1140 |
+
"learning_rate": 2.342222222222222e-06,
|
| 1141 |
+
"loss": 0.182,
|
| 1142 |
+
"step": 3950
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 2.5118483412322274,
|
| 1146 |
+
"grad_norm": 8.997177124023438,
|
| 1147 |
+
"learning_rate": 2.2866666666666667e-06,
|
| 1148 |
+
"loss": 0.205,
|
| 1149 |
+
"step": 3975
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 2.527646129541864,
|
| 1153 |
+
"grad_norm": 9.527833938598633,
|
| 1154 |
+
"learning_rate": 2.2311111111111115e-06,
|
| 1155 |
+
"loss": 0.1851,
|
| 1156 |
+
"step": 4000
|
| 1157 |
+
},
|
| 1158 |
+
{
|
| 1159 |
+
"epoch": 2.527646129541864,
|
| 1160 |
+
"eval_loss": 0.41017213463783264,
|
| 1161 |
+
"eval_runtime": 1287.8118,
|
| 1162 |
+
"eval_samples_per_second": 3.029,
|
| 1163 |
+
"eval_steps_per_second": 0.189,
|
| 1164 |
+
"eval_wer": 0.2870182555780933,
|
| 1165 |
+
"step": 4000
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 2.543443917851501,
|
| 1169 |
+
"grad_norm": 8.116750717163086,
|
| 1170 |
+
"learning_rate": 2.1755555555555556e-06,
|
| 1171 |
+
"loss": 0.1937,
|
| 1172 |
+
"step": 4025
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 2.5592417061611377,
|
| 1176 |
+
"grad_norm": 11.475099563598633,
|
| 1177 |
+
"learning_rate": 2.12e-06,
|
| 1178 |
+
"loss": 0.1795,
|
| 1179 |
+
"step": 4050
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 2.575039494470774,
|
| 1183 |
+
"grad_norm": 7.422430038452148,
|
| 1184 |
+
"learning_rate": 2.064444444444445e-06,
|
| 1185 |
+
"loss": 0.2029,
|
| 1186 |
+
"step": 4075
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 2.5908372827804107,
|
| 1190 |
+
"grad_norm": 8.631953239440918,
|
| 1191 |
+
"learning_rate": 2.008888888888889e-06,
|
| 1192 |
+
"loss": 0.1883,
|
| 1193 |
+
"step": 4100
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 2.6066350710900474,
|
| 1197 |
+
"grad_norm": 9.813713073730469,
|
| 1198 |
+
"learning_rate": 1.9533333333333334e-06,
|
| 1199 |
+
"loss": 0.1901,
|
| 1200 |
+
"step": 4125
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 2.622432859399684,
|
| 1204 |
+
"grad_norm": 10.284896850585938,
|
| 1205 |
+
"learning_rate": 1.8977777777777779e-06,
|
| 1206 |
+
"loss": 0.183,
|
| 1207 |
+
"step": 4150
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 2.638230647709321,
|
| 1211 |
+
"grad_norm": 9.403543472290039,
|
| 1212 |
+
"learning_rate": 1.8422222222222225e-06,
|
| 1213 |
+
"loss": 0.1892,
|
| 1214 |
+
"step": 4175
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 2.654028436018957,
|
| 1218 |
+
"grad_norm": 9.446948051452637,
|
| 1219 |
+
"learning_rate": 1.7866666666666668e-06,
|
| 1220 |
+
"loss": 0.1886,
|
| 1221 |
+
"step": 4200
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 2.669826224328594,
|
| 1225 |
+
"grad_norm": 10.583983421325684,
|
| 1226 |
+
"learning_rate": 1.7311111111111112e-06,
|
| 1227 |
+
"loss": 0.2211,
|
| 1228 |
+
"step": 4225
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 2.6856240126382307,
|
| 1232 |
+
"grad_norm": 10.528802871704102,
|
| 1233 |
+
"learning_rate": 1.675555555555556e-06,
|
| 1234 |
+
"loss": 0.1937,
|
| 1235 |
+
"step": 4250
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 2.7014218009478674,
|
| 1239 |
+
"grad_norm": 8.71202278137207,
|
| 1240 |
+
"learning_rate": 1.6200000000000002e-06,
|
| 1241 |
+
"loss": 0.1936,
|
| 1242 |
+
"step": 4275
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 2.717219589257504,
|
| 1246 |
+
"grad_norm": 8.44046401977539,
|
| 1247 |
+
"learning_rate": 1.5644444444444446e-06,
|
| 1248 |
+
"loss": 0.1994,
|
| 1249 |
+
"step": 4300
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 2.7330173775671405,
|
| 1253 |
+
"grad_norm": 9.856565475463867,
|
| 1254 |
+
"learning_rate": 1.5088888888888889e-06,
|
| 1255 |
+
"loss": 0.1753,
|
| 1256 |
+
"step": 4325
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 2.748815165876777,
|
| 1260 |
+
"grad_norm": 9.842414855957031,
|
| 1261 |
+
"learning_rate": 1.4533333333333335e-06,
|
| 1262 |
+
"loss": 0.1815,
|
| 1263 |
+
"step": 4350
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 2.764612954186414,
|
| 1267 |
+
"grad_norm": 9.333725929260254,
|
| 1268 |
+
"learning_rate": 1.397777777777778e-06,
|
| 1269 |
+
"loss": 0.1851,
|
| 1270 |
+
"step": 4375
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 2.7804107424960507,
|
| 1274 |
+
"grad_norm": 9.97825813293457,
|
| 1275 |
+
"learning_rate": 1.3422222222222222e-06,
|
| 1276 |
+
"loss": 0.1815,
|
| 1277 |
+
"step": 4400
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 2.7962085308056874,
|
| 1281 |
+
"grad_norm": 9.474321365356445,
|
| 1282 |
+
"learning_rate": 1.286666666666667e-06,
|
| 1283 |
+
"loss": 0.1792,
|
| 1284 |
+
"step": 4425
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 2.8120063191153237,
|
| 1288 |
+
"grad_norm": 8.71677303314209,
|
| 1289 |
+
"learning_rate": 1.2311111111111112e-06,
|
| 1290 |
+
"loss": 0.1921,
|
| 1291 |
+
"step": 4450
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 2.8278041074249605,
|
| 1295 |
+
"grad_norm": 9.69323444366455,
|
| 1296 |
+
"learning_rate": 1.1755555555555556e-06,
|
| 1297 |
+
"loss": 0.1778,
|
| 1298 |
+
"step": 4475
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 2.843601895734597,
|
| 1302 |
+
"grad_norm": 9.335270881652832,
|
| 1303 |
+
"learning_rate": 1.12e-06,
|
| 1304 |
+
"loss": 0.1783,
|
| 1305 |
+
"step": 4500
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 2.859399684044234,
|
| 1309 |
+
"grad_norm": 8.661324501037598,
|
| 1310 |
+
"learning_rate": 1.0644444444444445e-06,
|
| 1311 |
+
"loss": 0.1689,
|
| 1312 |
+
"step": 4525
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 2.8751974723538707,
|
| 1316 |
+
"grad_norm": 9.32027530670166,
|
| 1317 |
+
"learning_rate": 1.008888888888889e-06,
|
| 1318 |
+
"loss": 0.1772,
|
| 1319 |
+
"step": 4550
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 2.890995260663507,
|
| 1323 |
+
"grad_norm": 8.178330421447754,
|
| 1324 |
+
"learning_rate": 9.533333333333335e-07,
|
| 1325 |
+
"loss": 0.1806,
|
| 1326 |
+
"step": 4575
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 2.9067930489731437,
|
| 1330 |
+
"grad_norm": 9.38011646270752,
|
| 1331 |
+
"learning_rate": 8.977777777777778e-07,
|
| 1332 |
+
"loss": 0.1753,
|
| 1333 |
+
"step": 4600
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 2.9225908372827805,
|
| 1337 |
+
"grad_norm": 9.022958755493164,
|
| 1338 |
+
"learning_rate": 8.422222222222224e-07,
|
| 1339 |
+
"loss": 0.1813,
|
| 1340 |
+
"step": 4625
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 2.938388625592417,
|
| 1344 |
+
"grad_norm": 9.93110466003418,
|
| 1345 |
+
"learning_rate": 7.866666666666667e-07,
|
| 1346 |
+
"loss": 0.1825,
|
| 1347 |
+
"step": 4650
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 2.954186413902054,
|
| 1351 |
+
"grad_norm": 9.306452751159668,
|
| 1352 |
+
"learning_rate": 7.311111111111112e-07,
|
| 1353 |
+
"loss": 0.1943,
|
| 1354 |
+
"step": 4675
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 2.9699842022116902,
|
| 1358 |
+
"grad_norm": 7.849850177764893,
|
| 1359 |
+
"learning_rate": 6.755555555555555e-07,
|
| 1360 |
+
"loss": 0.1696,
|
| 1361 |
+
"step": 4700
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 2.985781990521327,
|
| 1365 |
+
"grad_norm": 10.920326232910156,
|
| 1366 |
+
"learning_rate": 6.200000000000001e-07,
|
| 1367 |
+
"loss": 0.1788,
|
| 1368 |
+
"step": 4725
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 3.0015797788309637,
|
| 1372 |
+
"grad_norm": 7.569627285003662,
|
| 1373 |
+
"learning_rate": 5.644444444444445e-07,
|
| 1374 |
+
"loss": 0.1698,
|
| 1375 |
+
"step": 4750
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 3.0173775671406005,
|
| 1379 |
+
"grad_norm": 8.557785987854004,
|
| 1380 |
+
"learning_rate": 5.088888888888889e-07,
|
| 1381 |
+
"loss": 0.1323,
|
| 1382 |
+
"step": 4775
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 3.0331753554502368,
|
| 1386 |
+
"grad_norm": 7.7341132164001465,
|
| 1387 |
+
"learning_rate": 4.533333333333334e-07,
|
| 1388 |
+
"loss": 0.1288,
|
| 1389 |
+
"step": 4800
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 3.0489731437598735,
|
| 1393 |
+
"grad_norm": 6.455957889556885,
|
| 1394 |
+
"learning_rate": 3.9777777777777783e-07,
|
| 1395 |
+
"loss": 0.1253,
|
| 1396 |
+
"step": 4825
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 3.0647709320695102,
|
| 1400 |
+
"grad_norm": 9.151886940002441,
|
| 1401 |
+
"learning_rate": 3.422222222222223e-07,
|
| 1402 |
+
"loss": 0.1369,
|
| 1403 |
+
"step": 4850
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 3.080568720379147,
|
| 1407 |
+
"grad_norm": 8.11299991607666,
|
| 1408 |
+
"learning_rate": 2.866666666666667e-07,
|
| 1409 |
+
"loss": 0.1272,
|
| 1410 |
+
"step": 4875
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 3.0963665086887837,
|
| 1414 |
+
"grad_norm": 6.720188617706299,
|
| 1415 |
+
"learning_rate": 2.3111111111111112e-07,
|
| 1416 |
+
"loss": 0.1329,
|
| 1417 |
+
"step": 4900
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 3.11216429699842,
|
| 1421 |
+
"grad_norm": 7.401381969451904,
|
| 1422 |
+
"learning_rate": 1.7555555555555558e-07,
|
| 1423 |
+
"loss": 0.1257,
|
| 1424 |
+
"step": 4925
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 3.1279620853080567,
|
| 1428 |
+
"grad_norm": 7.703917026519775,
|
| 1429 |
+
"learning_rate": 1.2000000000000002e-07,
|
| 1430 |
+
"loss": 0.1224,
|
| 1431 |
+
"step": 4950
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 3.1437598736176935,
|
| 1435 |
+
"grad_norm": 7.16009521484375,
|
| 1436 |
+
"learning_rate": 6.444444444444445e-08,
|
| 1437 |
+
"loss": 0.1296,
|
| 1438 |
+
"step": 4975
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 3.1595576619273302,
|
| 1442 |
+
"grad_norm": 7.600194931030273,
|
| 1443 |
+
"learning_rate": 8.88888888888889e-09,
|
| 1444 |
+
"loss": 0.1313,
|
| 1445 |
+
"step": 5000
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 3.1595576619273302,
|
| 1449 |
+
"eval_loss": 0.4071974754333496,
|
| 1450 |
+
"eval_runtime": 1289.8235,
|
| 1451 |
+
"eval_samples_per_second": 3.024,
|
| 1452 |
+
"eval_steps_per_second": 0.189,
|
| 1453 |
+
"eval_wer": 0.2801361924079977,
|
| 1454 |
+
"step": 5000
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 3.1595576619273302,
|
| 1458 |
+
"step": 5000,
|
| 1459 |
+
"total_flos": 2.727921844224e+20,
|
| 1460 |
+
"train_loss": 0.3573408980369568,
|
| 1461 |
+
"train_runtime": 57234.8416,
|
| 1462 |
+
"train_samples_per_second": 2.796,
|
| 1463 |
+
"train_steps_per_second": 0.087
|
| 1464 |
+
}
|
| 1465 |
+
],
|
| 1466 |
+
"logging_steps": 25,
|
| 1467 |
+
"max_steps": 5000,
|
| 1468 |
+
"num_input_tokens_seen": 0,
|
| 1469 |
+
"num_train_epochs": 4,
|
| 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 |
+
}
|