Commit
•
723677d
1
Parent(s):
65067c6
End of training
Browse files- README.md +1 -1
- all_results.json +15 -0
- eval_results.json +9 -0
- runs/Oct04_06-51-13_11dd62ab45fa/events.out.tfevents.1728084823.11dd62ab45fa.115.1 +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-train-all-plus-other-with-excluded
|
| 17 |
|
| 18 |
-
This model is a fine-tuned version of [openai/whisper-large-v3](https://huggingface.co/openai/whisper-large-v3) on
|
| 19 |
It achieves the following results on the evaluation set:
|
| 20 |
- Loss: 0.3181
|
| 21 |
- Wer: 0.1683
|
|
|
|
| 15 |
|
| 16 |
# whisper-large-v3-turbo-ft-cv-cy-train-all-plus-other-with-excluded
|
| 17 |
|
| 18 |
+
This model is a fine-tuned version of [openai/whisper-large-v3](https://huggingface.co/openai/whisper-large-v3) on the DewiBrynJones/commonvoice_18_0_cy train_all+other_with_excluded main dataset.
|
| 19 |
It achieves the following results on the evaluation set:
|
| 20 |
- Loss: 0.3181
|
| 21 |
- Wer: 0.1683
|
all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 7.072135785007072,
|
| 3 |
+
"eval_loss": 0.3181316554546356,
|
| 4 |
+
"eval_runtime": 2245.0023,
|
| 5 |
+
"eval_samples": 5381,
|
| 6 |
+
"eval_samples_per_second": 2.397,
|
| 7 |
+
"eval_steps_per_second": 0.15,
|
| 8 |
+
"eval_wer": 0.16825718716449958,
|
| 9 |
+
"total_flos": 5.4329055670370304e+20,
|
| 10 |
+
"train_loss": 0.09240171913430095,
|
| 11 |
+
"train_runtime": 60495.4593,
|
| 12 |
+
"train_samples": 22611,
|
| 13 |
+
"train_samples_per_second": 2.645,
|
| 14 |
+
"train_steps_per_second": 0.083
|
| 15 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 7.072135785007072,
|
| 3 |
+
"eval_loss": 0.3181316554546356,
|
| 4 |
+
"eval_runtime": 2245.0023,
|
| 5 |
+
"eval_samples": 5381,
|
| 6 |
+
"eval_samples_per_second": 2.397,
|
| 7 |
+
"eval_steps_per_second": 0.15,
|
| 8 |
+
"eval_wer": 0.16825718716449958
|
| 9 |
+
}
|
runs/Oct04_06-51-13_11dd62ab45fa/events.out.tfevents.1728084823.11dd62ab45fa.115.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e293d80628ad9a2a46cc0e8f0d84508832ed15360520ab2106f9536b8f35e5a8
|
| 3 |
+
size 406
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 7.072135785007072,
|
| 3 |
+
"total_flos": 5.4329055670370304e+20,
|
| 4 |
+
"train_loss": 0.09240171913430095,
|
| 5 |
+
"train_runtime": 60495.4593,
|
| 6 |
+
"train_samples": 22611,
|
| 7 |
+
"train_samples_per_second": 2.645,
|
| 8 |
+
"train_steps_per_second": 0.083
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1487 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 7.072135785007072,
|
| 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": 0.03536067892503536,
|
| 13 |
+
"grad_norm": 5.833098411560059,
|
| 14 |
+
"learning_rate": 5.000000000000001e-07,
|
| 15 |
+
"loss": 1.0626,
|
| 16 |
+
"step": 25
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.07072135785007072,
|
| 20 |
+
"grad_norm": 5.129093647003174,
|
| 21 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 22 |
+
"loss": 0.7439,
|
| 23 |
+
"step": 50
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.10608203677510608,
|
| 27 |
+
"grad_norm": 4.212141513824463,
|
| 28 |
+
"learning_rate": 1.5e-06,
|
| 29 |
+
"loss": 0.5203,
|
| 30 |
+
"step": 75
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.14144271570014144,
|
| 34 |
+
"grad_norm": 3.9624226093292236,
|
| 35 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 36 |
+
"loss": 0.4623,
|
| 37 |
+
"step": 100
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.1768033946251768,
|
| 41 |
+
"grad_norm": 4.4477338790893555,
|
| 42 |
+
"learning_rate": 2.5e-06,
|
| 43 |
+
"loss": 0.4498,
|
| 44 |
+
"step": 125
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.21216407355021216,
|
| 48 |
+
"grad_norm": 3.65051007270813,
|
| 49 |
+
"learning_rate": 3e-06,
|
| 50 |
+
"loss": 0.4272,
|
| 51 |
+
"step": 150
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.24752475247524752,
|
| 55 |
+
"grad_norm": 4.097436904907227,
|
| 56 |
+
"learning_rate": 3.5e-06,
|
| 57 |
+
"loss": 0.4245,
|
| 58 |
+
"step": 175
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.2828854314002829,
|
| 62 |
+
"grad_norm": 3.8615634441375732,
|
| 63 |
+
"learning_rate": 4.000000000000001e-06,
|
| 64 |
+
"loss": 0.4303,
|
| 65 |
+
"step": 200
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.31824611032531824,
|
| 69 |
+
"grad_norm": 3.634308099746704,
|
| 70 |
+
"learning_rate": 4.5e-06,
|
| 71 |
+
"loss": 0.3885,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.3536067892503536,
|
| 76 |
+
"grad_norm": 3.808622360229492,
|
| 77 |
+
"learning_rate": 5e-06,
|
| 78 |
+
"loss": 0.3891,
|
| 79 |
+
"step": 250
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.38896746817538896,
|
| 83 |
+
"grad_norm": 3.767564058303833,
|
| 84 |
+
"learning_rate": 5.500000000000001e-06,
|
| 85 |
+
"loss": 0.3737,
|
| 86 |
+
"step": 275
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.4243281471004243,
|
| 90 |
+
"grad_norm": 4.20762825012207,
|
| 91 |
+
"learning_rate": 6e-06,
|
| 92 |
+
"loss": 0.3549,
|
| 93 |
+
"step": 300
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.4596888260254597,
|
| 97 |
+
"grad_norm": 3.8779282569885254,
|
| 98 |
+
"learning_rate": 6.5000000000000004e-06,
|
| 99 |
+
"loss": 0.3757,
|
| 100 |
+
"step": 325
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.49504950495049505,
|
| 104 |
+
"grad_norm": 3.9535462856292725,
|
| 105 |
+
"learning_rate": 7e-06,
|
| 106 |
+
"loss": 0.3454,
|
| 107 |
+
"step": 350
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.5304101838755304,
|
| 111 |
+
"grad_norm": 3.7318906784057617,
|
| 112 |
+
"learning_rate": 7.500000000000001e-06,
|
| 113 |
+
"loss": 0.3216,
|
| 114 |
+
"step": 375
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.5657708628005658,
|
| 118 |
+
"grad_norm": 4.168917655944824,
|
| 119 |
+
"learning_rate": 8.000000000000001e-06,
|
| 120 |
+
"loss": 0.3194,
|
| 121 |
+
"step": 400
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.6011315417256011,
|
| 125 |
+
"grad_norm": 3.1413733959198,
|
| 126 |
+
"learning_rate": 8.5e-06,
|
| 127 |
+
"loss": 0.3188,
|
| 128 |
+
"step": 425
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.6364922206506365,
|
| 132 |
+
"grad_norm": 3.511763334274292,
|
| 133 |
+
"learning_rate": 9e-06,
|
| 134 |
+
"loss": 0.3129,
|
| 135 |
+
"step": 450
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.6718528995756718,
|
| 139 |
+
"grad_norm": 3.362168073654175,
|
| 140 |
+
"learning_rate": 9.5e-06,
|
| 141 |
+
"loss": 0.3028,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.7072135785007072,
|
| 146 |
+
"grad_norm": 3.6675732135772705,
|
| 147 |
+
"learning_rate": 1e-05,
|
| 148 |
+
"loss": 0.3189,
|
| 149 |
+
"step": 500
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.7425742574257426,
|
| 153 |
+
"grad_norm": 3.5820181369781494,
|
| 154 |
+
"learning_rate": 9.944444444444445e-06,
|
| 155 |
+
"loss": 0.297,
|
| 156 |
+
"step": 525
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.7779349363507779,
|
| 160 |
+
"grad_norm": 3.423774003982544,
|
| 161 |
+
"learning_rate": 9.88888888888889e-06,
|
| 162 |
+
"loss": 0.2935,
|
| 163 |
+
"step": 550
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.8132956152758133,
|
| 167 |
+
"grad_norm": 3.7895655632019043,
|
| 168 |
+
"learning_rate": 9.833333333333333e-06,
|
| 169 |
+
"loss": 0.2718,
|
| 170 |
+
"step": 575
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.8486562942008486,
|
| 174 |
+
"grad_norm": 3.2421023845672607,
|
| 175 |
+
"learning_rate": 9.777777777777779e-06,
|
| 176 |
+
"loss": 0.2741,
|
| 177 |
+
"step": 600
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.884016973125884,
|
| 181 |
+
"grad_norm": 3.0554616451263428,
|
| 182 |
+
"learning_rate": 9.722222222222223e-06,
|
| 183 |
+
"loss": 0.2661,
|
| 184 |
+
"step": 625
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.9193776520509194,
|
| 188 |
+
"grad_norm": 2.754528760910034,
|
| 189 |
+
"learning_rate": 9.666666666666667e-06,
|
| 190 |
+
"loss": 0.2573,
|
| 191 |
+
"step": 650
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.9547383309759547,
|
| 195 |
+
"grad_norm": 2.9093120098114014,
|
| 196 |
+
"learning_rate": 9.611111111111112e-06,
|
| 197 |
+
"loss": 0.2766,
|
| 198 |
+
"step": 675
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.9900990099009901,
|
| 202 |
+
"grad_norm": 2.9693257808685303,
|
| 203 |
+
"learning_rate": 9.555555555555556e-06,
|
| 204 |
+
"loss": 0.2506,
|
| 205 |
+
"step": 700
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 1.0254596888260255,
|
| 209 |
+
"grad_norm": 2.879349946975708,
|
| 210 |
+
"learning_rate": 9.5e-06,
|
| 211 |
+
"loss": 0.2021,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 1.0608203677510608,
|
| 216 |
+
"grad_norm": 2.6136014461517334,
|
| 217 |
+
"learning_rate": 9.444444444444445e-06,
|
| 218 |
+
"loss": 0.181,
|
| 219 |
+
"step": 750
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 1.0961810466760962,
|
| 223 |
+
"grad_norm": 2.8699588775634766,
|
| 224 |
+
"learning_rate": 9.38888888888889e-06,
|
| 225 |
+
"loss": 0.1648,
|
| 226 |
+
"step": 775
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 1.1315417256011315,
|
| 230 |
+
"grad_norm": 2.2577948570251465,
|
| 231 |
+
"learning_rate": 9.333333333333334e-06,
|
| 232 |
+
"loss": 0.171,
|
| 233 |
+
"step": 800
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 1.166902404526167,
|
| 237 |
+
"grad_norm": 2.6615614891052246,
|
| 238 |
+
"learning_rate": 9.277777777777778e-06,
|
| 239 |
+
"loss": 0.1759,
|
| 240 |
+
"step": 825
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 1.2022630834512023,
|
| 244 |
+
"grad_norm": 3.6334331035614014,
|
| 245 |
+
"learning_rate": 9.222222222222224e-06,
|
| 246 |
+
"loss": 0.1703,
|
| 247 |
+
"step": 850
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 1.2376237623762376,
|
| 251 |
+
"grad_norm": 2.0051722526550293,
|
| 252 |
+
"learning_rate": 9.166666666666666e-06,
|
| 253 |
+
"loss": 0.1666,
|
| 254 |
+
"step": 875
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 1.272984441301273,
|
| 258 |
+
"grad_norm": 2.5868630409240723,
|
| 259 |
+
"learning_rate": 9.111111111111112e-06,
|
| 260 |
+
"loss": 0.1631,
|
| 261 |
+
"step": 900
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 1.3083451202263083,
|
| 265 |
+
"grad_norm": 2.7406225204467773,
|
| 266 |
+
"learning_rate": 9.055555555555556e-06,
|
| 267 |
+
"loss": 0.1625,
|
| 268 |
+
"step": 925
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 1.3437057991513437,
|
| 272 |
+
"grad_norm": 2.9130780696868896,
|
| 273 |
+
"learning_rate": 9e-06,
|
| 274 |
+
"loss": 0.1779,
|
| 275 |
+
"step": 950
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 1.379066478076379,
|
| 279 |
+
"grad_norm": 2.973248243331909,
|
| 280 |
+
"learning_rate": 8.944444444444446e-06,
|
| 281 |
+
"loss": 0.1642,
|
| 282 |
+
"step": 975
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 1.4144271570014144,
|
| 286 |
+
"grad_norm": 2.1427431106567383,
|
| 287 |
+
"learning_rate": 8.888888888888888e-06,
|
| 288 |
+
"loss": 0.1599,
|
| 289 |
+
"step": 1000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 1.4144271570014144,
|
| 293 |
+
"eval_loss": 0.2516109049320221,
|
| 294 |
+
"eval_runtime": 2179.6002,
|
| 295 |
+
"eval_samples_per_second": 2.469,
|
| 296 |
+
"eval_steps_per_second": 0.155,
|
| 297 |
+
"eval_wer": 0.2035468607101674,
|
| 298 |
+
"step": 1000
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 1.4497878359264498,
|
| 302 |
+
"grad_norm": 3.01638126373291,
|
| 303 |
+
"learning_rate": 8.833333333333334e-06,
|
| 304 |
+
"loss": 0.1525,
|
| 305 |
+
"step": 1025
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 1.4851485148514851,
|
| 309 |
+
"grad_norm": 2.6232805252075195,
|
| 310 |
+
"learning_rate": 8.777777777777778e-06,
|
| 311 |
+
"loss": 0.1565,
|
| 312 |
+
"step": 1050
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 1.5205091937765205,
|
| 316 |
+
"grad_norm": 2.569593906402588,
|
| 317 |
+
"learning_rate": 8.722222222222224e-06,
|
| 318 |
+
"loss": 0.1546,
|
| 319 |
+
"step": 1075
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 1.5558698727015559,
|
| 323 |
+
"grad_norm": 2.730534076690674,
|
| 324 |
+
"learning_rate": 8.666666666666668e-06,
|
| 325 |
+
"loss": 0.1523,
|
| 326 |
+
"step": 1100
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 1.5912305516265912,
|
| 330 |
+
"grad_norm": 3.71217942237854,
|
| 331 |
+
"learning_rate": 8.611111111111112e-06,
|
| 332 |
+
"loss": 0.1601,
|
| 333 |
+
"step": 1125
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 1.6265912305516266,
|
| 337 |
+
"grad_norm": 2.2378976345062256,
|
| 338 |
+
"learning_rate": 8.555555555555556e-06,
|
| 339 |
+
"loss": 0.1517,
|
| 340 |
+
"step": 1150
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 1.661951909476662,
|
| 344 |
+
"grad_norm": 2.2248027324676514,
|
| 345 |
+
"learning_rate": 8.5e-06,
|
| 346 |
+
"loss": 0.1479,
|
| 347 |
+
"step": 1175
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 1.6973125884016973,
|
| 351 |
+
"grad_norm": 1.97626531124115,
|
| 352 |
+
"learning_rate": 8.444444444444446e-06,
|
| 353 |
+
"loss": 0.1448,
|
| 354 |
+
"step": 1200
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 1.7326732673267327,
|
| 358 |
+
"grad_norm": 2.5063023567199707,
|
| 359 |
+
"learning_rate": 8.38888888888889e-06,
|
| 360 |
+
"loss": 0.1493,
|
| 361 |
+
"step": 1225
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 1.768033946251768,
|
| 365 |
+
"grad_norm": 2.230543613433838,
|
| 366 |
+
"learning_rate": 8.333333333333334e-06,
|
| 367 |
+
"loss": 0.1413,
|
| 368 |
+
"step": 1250
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 1.8033946251768034,
|
| 372 |
+
"grad_norm": 2.3696229457855225,
|
| 373 |
+
"learning_rate": 8.277777777777778e-06,
|
| 374 |
+
"loss": 0.1514,
|
| 375 |
+
"step": 1275
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 1.8387553041018387,
|
| 379 |
+
"grad_norm": 2.3994140625,
|
| 380 |
+
"learning_rate": 8.222222222222222e-06,
|
| 381 |
+
"loss": 0.1514,
|
| 382 |
+
"step": 1300
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 1.874115983026874,
|
| 386 |
+
"grad_norm": 2.1594157218933105,
|
| 387 |
+
"learning_rate": 8.166666666666668e-06,
|
| 388 |
+
"loss": 0.1391,
|
| 389 |
+
"step": 1325
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 1.9094766619519095,
|
| 393 |
+
"grad_norm": 2.2677788734436035,
|
| 394 |
+
"learning_rate": 8.111111111111112e-06,
|
| 395 |
+
"loss": 0.1333,
|
| 396 |
+
"step": 1350
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 1.9448373408769448,
|
| 400 |
+
"grad_norm": 2.0969929695129395,
|
| 401 |
+
"learning_rate": 8.055555555555557e-06,
|
| 402 |
+
"loss": 0.141,
|
| 403 |
+
"step": 1375
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 1.9801980198019802,
|
| 407 |
+
"grad_norm": 2.2506678104400635,
|
| 408 |
+
"learning_rate": 8.000000000000001e-06,
|
| 409 |
+
"loss": 0.1393,
|
| 410 |
+
"step": 1400
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 2.0155586987270158,
|
| 414 |
+
"grad_norm": 1.4057811498641968,
|
| 415 |
+
"learning_rate": 7.944444444444445e-06,
|
| 416 |
+
"loss": 0.1088,
|
| 417 |
+
"step": 1425
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 2.050919377652051,
|
| 421 |
+
"grad_norm": 1.8405675888061523,
|
| 422 |
+
"learning_rate": 7.88888888888889e-06,
|
| 423 |
+
"loss": 0.0696,
|
| 424 |
+
"step": 1450
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 2.0862800565770865,
|
| 428 |
+
"grad_norm": 1.7965513467788696,
|
| 429 |
+
"learning_rate": 7.833333333333333e-06,
|
| 430 |
+
"loss": 0.0715,
|
| 431 |
+
"step": 1475
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 2.1216407355021216,
|
| 435 |
+
"grad_norm": 1.533977746963501,
|
| 436 |
+
"learning_rate": 7.77777777777778e-06,
|
| 437 |
+
"loss": 0.0655,
|
| 438 |
+
"step": 1500
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 2.157001414427157,
|
| 442 |
+
"grad_norm": 1.5952216386795044,
|
| 443 |
+
"learning_rate": 7.722222222222223e-06,
|
| 444 |
+
"loss": 0.0701,
|
| 445 |
+
"step": 1525
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 2.1923620933521923,
|
| 449 |
+
"grad_norm": 1.796464443206787,
|
| 450 |
+
"learning_rate": 7.666666666666667e-06,
|
| 451 |
+
"loss": 0.0649,
|
| 452 |
+
"step": 1550
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 2.227722772277228,
|
| 456 |
+
"grad_norm": 1.637083888053894,
|
| 457 |
+
"learning_rate": 7.611111111111111e-06,
|
| 458 |
+
"loss": 0.0729,
|
| 459 |
+
"step": 1575
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 2.263083451202263,
|
| 463 |
+
"grad_norm": 1.361791968345642,
|
| 464 |
+
"learning_rate": 7.555555555555556e-06,
|
| 465 |
+
"loss": 0.0661,
|
| 466 |
+
"step": 1600
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 2.298444130127298,
|
| 470 |
+
"grad_norm": 1.4004043340682983,
|
| 471 |
+
"learning_rate": 7.500000000000001e-06,
|
| 472 |
+
"loss": 0.0635,
|
| 473 |
+
"step": 1625
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 2.333804809052334,
|
| 477 |
+
"grad_norm": 1.415955662727356,
|
| 478 |
+
"learning_rate": 7.444444444444445e-06,
|
| 479 |
+
"loss": 0.0694,
|
| 480 |
+
"step": 1650
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 2.3691654879773694,
|
| 484 |
+
"grad_norm": 1.5830997228622437,
|
| 485 |
+
"learning_rate": 7.38888888888889e-06,
|
| 486 |
+
"loss": 0.0693,
|
| 487 |
+
"step": 1675
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 2.4045261669024045,
|
| 491 |
+
"grad_norm": 1.8817075490951538,
|
| 492 |
+
"learning_rate": 7.333333333333333e-06,
|
| 493 |
+
"loss": 0.0753,
|
| 494 |
+
"step": 1700
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 2.4398868458274396,
|
| 498 |
+
"grad_norm": 2.290759563446045,
|
| 499 |
+
"learning_rate": 7.277777777777778e-06,
|
| 500 |
+
"loss": 0.0705,
|
| 501 |
+
"step": 1725
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 2.4752475247524752,
|
| 505 |
+
"grad_norm": 1.521173119544983,
|
| 506 |
+
"learning_rate": 7.222222222222223e-06,
|
| 507 |
+
"loss": 0.0745,
|
| 508 |
+
"step": 1750
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 2.510608203677511,
|
| 512 |
+
"grad_norm": 2.191132068634033,
|
| 513 |
+
"learning_rate": 7.166666666666667e-06,
|
| 514 |
+
"loss": 0.0678,
|
| 515 |
+
"step": 1775
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 2.545968882602546,
|
| 519 |
+
"grad_norm": 1.374758243560791,
|
| 520 |
+
"learning_rate": 7.111111111111112e-06,
|
| 521 |
+
"loss": 0.0622,
|
| 522 |
+
"step": 1800
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 2.581329561527581,
|
| 526 |
+
"grad_norm": 1.8768402338027954,
|
| 527 |
+
"learning_rate": 7.055555555555557e-06,
|
| 528 |
+
"loss": 0.0596,
|
| 529 |
+
"step": 1825
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 2.6166902404526167,
|
| 533 |
+
"grad_norm": 1.5529979467391968,
|
| 534 |
+
"learning_rate": 7e-06,
|
| 535 |
+
"loss": 0.063,
|
| 536 |
+
"step": 1850
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 2.6520509193776522,
|
| 540 |
+
"grad_norm": 1.5192201137542725,
|
| 541 |
+
"learning_rate": 6.944444444444445e-06,
|
| 542 |
+
"loss": 0.072,
|
| 543 |
+
"step": 1875
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 2.6874115983026874,
|
| 547 |
+
"grad_norm": 1.872154951095581,
|
| 548 |
+
"learning_rate": 6.88888888888889e-06,
|
| 549 |
+
"loss": 0.069,
|
| 550 |
+
"step": 1900
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 2.7227722772277225,
|
| 554 |
+
"grad_norm": 1.8019146919250488,
|
| 555 |
+
"learning_rate": 6.833333333333334e-06,
|
| 556 |
+
"loss": 0.0599,
|
| 557 |
+
"step": 1925
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 2.758132956152758,
|
| 561 |
+
"grad_norm": 1.5129454135894775,
|
| 562 |
+
"learning_rate": 6.777777777777779e-06,
|
| 563 |
+
"loss": 0.0671,
|
| 564 |
+
"step": 1950
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 2.7934936350777937,
|
| 568 |
+
"grad_norm": 1.387603998184204,
|
| 569 |
+
"learning_rate": 6.7222222222222235e-06,
|
| 570 |
+
"loss": 0.059,
|
| 571 |
+
"step": 1975
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 2.828854314002829,
|
| 575 |
+
"grad_norm": 1.5892807245254517,
|
| 576 |
+
"learning_rate": 6.666666666666667e-06,
|
| 577 |
+
"loss": 0.0651,
|
| 578 |
+
"step": 2000
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 2.828854314002829,
|
| 582 |
+
"eval_loss": 0.23385389149188995,
|
| 583 |
+
"eval_runtime": 2134.4668,
|
| 584 |
+
"eval_samples_per_second": 2.521,
|
| 585 |
+
"eval_steps_per_second": 0.158,
|
| 586 |
+
"eval_wer": 0.18308879080679152,
|
| 587 |
+
"step": 2000
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 2.864214992927864,
|
| 591 |
+
"grad_norm": 2.636842727661133,
|
| 592 |
+
"learning_rate": 6.6111111111111115e-06,
|
| 593 |
+
"loss": 0.0683,
|
| 594 |
+
"step": 2025
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 2.8995756718528995,
|
| 598 |
+
"grad_norm": 2.534872531890869,
|
| 599 |
+
"learning_rate": 6.555555555555556e-06,
|
| 600 |
+
"loss": 0.0617,
|
| 601 |
+
"step": 2050
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 2.934936350777935,
|
| 605 |
+
"grad_norm": 1.9003880023956299,
|
| 606 |
+
"learning_rate": 6.5000000000000004e-06,
|
| 607 |
+
"loss": 0.0566,
|
| 608 |
+
"step": 2075
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 2.9702970297029703,
|
| 612 |
+
"grad_norm": 2.0285208225250244,
|
| 613 |
+
"learning_rate": 6.444444444444445e-06,
|
| 614 |
+
"loss": 0.0652,
|
| 615 |
+
"step": 2100
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 3.005657708628006,
|
| 619 |
+
"grad_norm": 1.0410995483398438,
|
| 620 |
+
"learning_rate": 6.3888888888888885e-06,
|
| 621 |
+
"loss": 0.0567,
|
| 622 |
+
"step": 2125
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 3.041018387553041,
|
| 626 |
+
"grad_norm": 1.090918779373169,
|
| 627 |
+
"learning_rate": 6.333333333333333e-06,
|
| 628 |
+
"loss": 0.0304,
|
| 629 |
+
"step": 2150
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 3.0763790664780766,
|
| 633 |
+
"grad_norm": 1.4816592931747437,
|
| 634 |
+
"learning_rate": 6.277777777777778e-06,
|
| 635 |
+
"loss": 0.0255,
|
| 636 |
+
"step": 2175
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 3.1117397454031117,
|
| 640 |
+
"grad_norm": 1.5663459300994873,
|
| 641 |
+
"learning_rate": 6.222222222222223e-06,
|
| 642 |
+
"loss": 0.0279,
|
| 643 |
+
"step": 2200
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 3.1471004243281473,
|
| 647 |
+
"grad_norm": 1.0855896472930908,
|
| 648 |
+
"learning_rate": 6.166666666666667e-06,
|
| 649 |
+
"loss": 0.0257,
|
| 650 |
+
"step": 2225
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 3.1824611032531824,
|
| 654 |
+
"grad_norm": 0.9822140336036682,
|
| 655 |
+
"learning_rate": 6.111111111111112e-06,
|
| 656 |
+
"loss": 0.0274,
|
| 657 |
+
"step": 2250
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 3.217821782178218,
|
| 661 |
+
"grad_norm": 1.3822486400604248,
|
| 662 |
+
"learning_rate": 6.055555555555555e-06,
|
| 663 |
+
"loss": 0.0295,
|
| 664 |
+
"step": 2275
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 3.253182461103253,
|
| 668 |
+
"grad_norm": 0.9994338154792786,
|
| 669 |
+
"learning_rate": 6e-06,
|
| 670 |
+
"loss": 0.0287,
|
| 671 |
+
"step": 2300
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 3.2885431400282887,
|
| 675 |
+
"grad_norm": 1.223861813545227,
|
| 676 |
+
"learning_rate": 5.944444444444445e-06,
|
| 677 |
+
"loss": 0.0288,
|
| 678 |
+
"step": 2325
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 3.323903818953324,
|
| 682 |
+
"grad_norm": 1.0625354051589966,
|
| 683 |
+
"learning_rate": 5.88888888888889e-06,
|
| 684 |
+
"loss": 0.0313,
|
| 685 |
+
"step": 2350
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 3.3592644978783595,
|
| 689 |
+
"grad_norm": 1.2225134372711182,
|
| 690 |
+
"learning_rate": 5.833333333333334e-06,
|
| 691 |
+
"loss": 0.0278,
|
| 692 |
+
"step": 2375
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 3.3946251768033946,
|
| 696 |
+
"grad_norm": 1.0246098041534424,
|
| 697 |
+
"learning_rate": 5.777777777777778e-06,
|
| 698 |
+
"loss": 0.0285,
|
| 699 |
+
"step": 2400
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 3.42998585572843,
|
| 703 |
+
"grad_norm": 1.376129388809204,
|
| 704 |
+
"learning_rate": 5.722222222222222e-06,
|
| 705 |
+
"loss": 0.0326,
|
| 706 |
+
"step": 2425
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 3.4653465346534653,
|
| 710 |
+
"grad_norm": 1.7981650829315186,
|
| 711 |
+
"learning_rate": 5.666666666666667e-06,
|
| 712 |
+
"loss": 0.0326,
|
| 713 |
+
"step": 2450
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 3.500707213578501,
|
| 717 |
+
"grad_norm": 0.6855577230453491,
|
| 718 |
+
"learning_rate": 5.611111111111112e-06,
|
| 719 |
+
"loss": 0.0272,
|
| 720 |
+
"step": 2475
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 3.536067892503536,
|
| 724 |
+
"grad_norm": 1.0780831575393677,
|
| 725 |
+
"learning_rate": 5.555555555555557e-06,
|
| 726 |
+
"loss": 0.026,
|
| 727 |
+
"step": 2500
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 3.571428571428571,
|
| 731 |
+
"grad_norm": 1.4502952098846436,
|
| 732 |
+
"learning_rate": 5.500000000000001e-06,
|
| 733 |
+
"loss": 0.0326,
|
| 734 |
+
"step": 2525
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 3.6067892503536068,
|
| 738 |
+
"grad_norm": 1.2788467407226562,
|
| 739 |
+
"learning_rate": 5.444444444444445e-06,
|
| 740 |
+
"loss": 0.0278,
|
| 741 |
+
"step": 2550
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 3.6421499292786423,
|
| 745 |
+
"grad_norm": 1.4065041542053223,
|
| 746 |
+
"learning_rate": 5.388888888888889e-06,
|
| 747 |
+
"loss": 0.028,
|
| 748 |
+
"step": 2575
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 3.6775106082036775,
|
| 752 |
+
"grad_norm": 0.9822909832000732,
|
| 753 |
+
"learning_rate": 5.333333333333334e-06,
|
| 754 |
+
"loss": 0.024,
|
| 755 |
+
"step": 2600
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 3.7128712871287126,
|
| 759 |
+
"grad_norm": 0.9713716506958008,
|
| 760 |
+
"learning_rate": 5.2777777777777785e-06,
|
| 761 |
+
"loss": 0.0314,
|
| 762 |
+
"step": 2625
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 3.748231966053748,
|
| 766 |
+
"grad_norm": 1.5902820825576782,
|
| 767 |
+
"learning_rate": 5.2222222222222226e-06,
|
| 768 |
+
"loss": 0.0286,
|
| 769 |
+
"step": 2650
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 3.783592644978784,
|
| 773 |
+
"grad_norm": 1.065138816833496,
|
| 774 |
+
"learning_rate": 5.1666666666666675e-06,
|
| 775 |
+
"loss": 0.0237,
|
| 776 |
+
"step": 2675
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 3.818953323903819,
|
| 780 |
+
"grad_norm": 1.4179657697677612,
|
| 781 |
+
"learning_rate": 5.1111111111111115e-06,
|
| 782 |
+
"loss": 0.0265,
|
| 783 |
+
"step": 2700
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 3.854314002828854,
|
| 787 |
+
"grad_norm": 1.563292384147644,
|
| 788 |
+
"learning_rate": 5.0555555555555555e-06,
|
| 789 |
+
"loss": 0.0321,
|
| 790 |
+
"step": 2725
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 3.8896746817538896,
|
| 794 |
+
"grad_norm": 1.1048407554626465,
|
| 795 |
+
"learning_rate": 5e-06,
|
| 796 |
+
"loss": 0.0277,
|
| 797 |
+
"step": 2750
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 3.9250353606789252,
|
| 801 |
+
"grad_norm": 1.585411787033081,
|
| 802 |
+
"learning_rate": 4.944444444444445e-06,
|
| 803 |
+
"loss": 0.0309,
|
| 804 |
+
"step": 2775
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 3.9603960396039604,
|
| 808 |
+
"grad_norm": 1.438583493232727,
|
| 809 |
+
"learning_rate": 4.888888888888889e-06,
|
| 810 |
+
"loss": 0.0229,
|
| 811 |
+
"step": 2800
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 3.9957567185289955,
|
| 815 |
+
"grad_norm": 1.6560890674591064,
|
| 816 |
+
"learning_rate": 4.833333333333333e-06,
|
| 817 |
+
"loss": 0.0265,
|
| 818 |
+
"step": 2825
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 4.0311173974540315,
|
| 822 |
+
"grad_norm": 1.1240108013153076,
|
| 823 |
+
"learning_rate": 4.777777777777778e-06,
|
| 824 |
+
"loss": 0.013,
|
| 825 |
+
"step": 2850
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 4.066478076379067,
|
| 829 |
+
"grad_norm": 1.5441868305206299,
|
| 830 |
+
"learning_rate": 4.722222222222222e-06,
|
| 831 |
+
"loss": 0.013,
|
| 832 |
+
"step": 2875
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 4.101838755304102,
|
| 836 |
+
"grad_norm": 0.6906932592391968,
|
| 837 |
+
"learning_rate": 4.666666666666667e-06,
|
| 838 |
+
"loss": 0.0137,
|
| 839 |
+
"step": 2900
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 4.137199434229137,
|
| 843 |
+
"grad_norm": 0.4889713227748871,
|
| 844 |
+
"learning_rate": 4.611111111111112e-06,
|
| 845 |
+
"loss": 0.0107,
|
| 846 |
+
"step": 2925
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 4.172560113154173,
|
| 850 |
+
"grad_norm": 1.0169159173965454,
|
| 851 |
+
"learning_rate": 4.555555555555556e-06,
|
| 852 |
+
"loss": 0.0105,
|
| 853 |
+
"step": 2950
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 4.207920792079208,
|
| 857 |
+
"grad_norm": 0.5549736022949219,
|
| 858 |
+
"learning_rate": 4.5e-06,
|
| 859 |
+
"loss": 0.0111,
|
| 860 |
+
"step": 2975
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 4.243281471004243,
|
| 864 |
+
"grad_norm": 1.1406062841415405,
|
| 865 |
+
"learning_rate": 4.444444444444444e-06,
|
| 866 |
+
"loss": 0.0119,
|
| 867 |
+
"step": 3000
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 4.243281471004243,
|
| 871 |
+
"eval_loss": 0.2719973921775818,
|
| 872 |
+
"eval_runtime": 2233.3625,
|
| 873 |
+
"eval_samples_per_second": 2.409,
|
| 874 |
+
"eval_steps_per_second": 0.151,
|
| 875 |
+
"eval_wer": 0.1750566622927353,
|
| 876 |
+
"step": 3000
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 4.278642149929278,
|
| 880 |
+
"grad_norm": 0.460374653339386,
|
| 881 |
+
"learning_rate": 4.388888888888889e-06,
|
| 882 |
+
"loss": 0.0085,
|
| 883 |
+
"step": 3025
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 4.314002828854314,
|
| 887 |
+
"grad_norm": 0.334330677986145,
|
| 888 |
+
"learning_rate": 4.333333333333334e-06,
|
| 889 |
+
"loss": 0.01,
|
| 890 |
+
"step": 3050
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 4.3493635077793495,
|
| 894 |
+
"grad_norm": 0.8199196457862854,
|
| 895 |
+
"learning_rate": 4.277777777777778e-06,
|
| 896 |
+
"loss": 0.0105,
|
| 897 |
+
"step": 3075
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 4.384724186704385,
|
| 901 |
+
"grad_norm": 0.9259428977966309,
|
| 902 |
+
"learning_rate": 4.222222222222223e-06,
|
| 903 |
+
"loss": 0.0108,
|
| 904 |
+
"step": 3100
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 4.42008486562942,
|
| 908 |
+
"grad_norm": 1.7561092376708984,
|
| 909 |
+
"learning_rate": 4.166666666666667e-06,
|
| 910 |
+
"loss": 0.0098,
|
| 911 |
+
"step": 3125
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 4.455445544554456,
|
| 915 |
+
"grad_norm": 0.5159019827842712,
|
| 916 |
+
"learning_rate": 4.111111111111111e-06,
|
| 917 |
+
"loss": 0.0114,
|
| 918 |
+
"step": 3150
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 4.490806223479491,
|
| 922 |
+
"grad_norm": 0.6501268148422241,
|
| 923 |
+
"learning_rate": 4.055555555555556e-06,
|
| 924 |
+
"loss": 0.0116,
|
| 925 |
+
"step": 3175
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 4.526166902404526,
|
| 929 |
+
"grad_norm": 0.4661734402179718,
|
| 930 |
+
"learning_rate": 4.000000000000001e-06,
|
| 931 |
+
"loss": 0.0125,
|
| 932 |
+
"step": 3200
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 4.561527581329561,
|
| 936 |
+
"grad_norm": 1.581214427947998,
|
| 937 |
+
"learning_rate": 3.944444444444445e-06,
|
| 938 |
+
"loss": 0.0126,
|
| 939 |
+
"step": 3225
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 4.596888260254596,
|
| 943 |
+
"grad_norm": 0.7272769212722778,
|
| 944 |
+
"learning_rate": 3.88888888888889e-06,
|
| 945 |
+
"loss": 0.0116,
|
| 946 |
+
"step": 3250
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 4.632248939179632,
|
| 950 |
+
"grad_norm": 1.0151731967926025,
|
| 951 |
+
"learning_rate": 3.833333333333334e-06,
|
| 952 |
+
"loss": 0.0089,
|
| 953 |
+
"step": 3275
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 4.667609618104668,
|
| 957 |
+
"grad_norm": 0.4858356714248657,
|
| 958 |
+
"learning_rate": 3.777777777777778e-06,
|
| 959 |
+
"loss": 0.0095,
|
| 960 |
+
"step": 3300
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 4.702970297029703,
|
| 964 |
+
"grad_norm": 1.0517410039901733,
|
| 965 |
+
"learning_rate": 3.7222222222222225e-06,
|
| 966 |
+
"loss": 0.0107,
|
| 967 |
+
"step": 3325
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 4.738330975954739,
|
| 971 |
+
"grad_norm": 1.0615334510803223,
|
| 972 |
+
"learning_rate": 3.6666666666666666e-06,
|
| 973 |
+
"loss": 0.0132,
|
| 974 |
+
"step": 3350
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 4.773691654879774,
|
| 978 |
+
"grad_norm": 1.1842697858810425,
|
| 979 |
+
"learning_rate": 3.6111111111111115e-06,
|
| 980 |
+
"loss": 0.009,
|
| 981 |
+
"step": 3375
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 4.809052333804809,
|
| 985 |
+
"grad_norm": 0.6705470681190491,
|
| 986 |
+
"learning_rate": 3.555555555555556e-06,
|
| 987 |
+
"loss": 0.0109,
|
| 988 |
+
"step": 3400
|
| 989 |
+
},
|
| 990 |
+
{
|
| 991 |
+
"epoch": 4.844413012729844,
|
| 992 |
+
"grad_norm": 0.22446350753307343,
|
| 993 |
+
"learning_rate": 3.5e-06,
|
| 994 |
+
"loss": 0.0113,
|
| 995 |
+
"step": 3425
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 4.879773691654879,
|
| 999 |
+
"grad_norm": 0.6739686131477356,
|
| 1000 |
+
"learning_rate": 3.444444444444445e-06,
|
| 1001 |
+
"loss": 0.0089,
|
| 1002 |
+
"step": 3450
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 4.915134370579915,
|
| 1006 |
+
"grad_norm": 0.9094479084014893,
|
| 1007 |
+
"learning_rate": 3.3888888888888893e-06,
|
| 1008 |
+
"loss": 0.0096,
|
| 1009 |
+
"step": 3475
|
| 1010 |
+
},
|
| 1011 |
+
{
|
| 1012 |
+
"epoch": 4.9504950495049505,
|
| 1013 |
+
"grad_norm": 0.486494243144989,
|
| 1014 |
+
"learning_rate": 3.3333333333333333e-06,
|
| 1015 |
+
"loss": 0.0116,
|
| 1016 |
+
"step": 3500
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 4.985855728429986,
|
| 1020 |
+
"grad_norm": 1.4182993173599243,
|
| 1021 |
+
"learning_rate": 3.277777777777778e-06,
|
| 1022 |
+
"loss": 0.011,
|
| 1023 |
+
"step": 3525
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 5.021216407355022,
|
| 1027 |
+
"grad_norm": 0.35871076583862305,
|
| 1028 |
+
"learning_rate": 3.2222222222222227e-06,
|
| 1029 |
+
"loss": 0.0087,
|
| 1030 |
+
"step": 3550
|
| 1031 |
+
},
|
| 1032 |
+
{
|
| 1033 |
+
"epoch": 5.056577086280057,
|
| 1034 |
+
"grad_norm": 0.08477536588907242,
|
| 1035 |
+
"learning_rate": 3.1666666666666667e-06,
|
| 1036 |
+
"loss": 0.0053,
|
| 1037 |
+
"step": 3575
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 5.091937765205092,
|
| 1041 |
+
"grad_norm": 0.2496667057275772,
|
| 1042 |
+
"learning_rate": 3.1111111111111116e-06,
|
| 1043 |
+
"loss": 0.0033,
|
| 1044 |
+
"step": 3600
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 5.127298444130127,
|
| 1048 |
+
"grad_norm": 0.21481718122959137,
|
| 1049 |
+
"learning_rate": 3.055555555555556e-06,
|
| 1050 |
+
"loss": 0.0039,
|
| 1051 |
+
"step": 3625
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 5.162659123055163,
|
| 1055 |
+
"grad_norm": 1.028740406036377,
|
| 1056 |
+
"learning_rate": 3e-06,
|
| 1057 |
+
"loss": 0.0027,
|
| 1058 |
+
"step": 3650
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 5.198019801980198,
|
| 1062 |
+
"grad_norm": 1.1931138038635254,
|
| 1063 |
+
"learning_rate": 2.944444444444445e-06,
|
| 1064 |
+
"loss": 0.0041,
|
| 1065 |
+
"step": 3675
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 5.233380480905233,
|
| 1069 |
+
"grad_norm": 0.2043699473142624,
|
| 1070 |
+
"learning_rate": 2.888888888888889e-06,
|
| 1071 |
+
"loss": 0.0042,
|
| 1072 |
+
"step": 3700
|
| 1073 |
+
},
|
| 1074 |
+
{
|
| 1075 |
+
"epoch": 5.2687411598302685,
|
| 1076 |
+
"grad_norm": 0.15545345842838287,
|
| 1077 |
+
"learning_rate": 2.8333333333333335e-06,
|
| 1078 |
+
"loss": 0.0036,
|
| 1079 |
+
"step": 3725
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 5.3041018387553045,
|
| 1083 |
+
"grad_norm": 0.127390518784523,
|
| 1084 |
+
"learning_rate": 2.7777777777777783e-06,
|
| 1085 |
+
"loss": 0.0035,
|
| 1086 |
+
"step": 3750
|
| 1087 |
+
},
|
| 1088 |
+
{
|
| 1089 |
+
"epoch": 5.33946251768034,
|
| 1090 |
+
"grad_norm": 0.2568166255950928,
|
| 1091 |
+
"learning_rate": 2.7222222222222224e-06,
|
| 1092 |
+
"loss": 0.0039,
|
| 1093 |
+
"step": 3775
|
| 1094 |
+
},
|
| 1095 |
+
{
|
| 1096 |
+
"epoch": 5.374823196605375,
|
| 1097 |
+
"grad_norm": 0.7207896709442139,
|
| 1098 |
+
"learning_rate": 2.666666666666667e-06,
|
| 1099 |
+
"loss": 0.0037,
|
| 1100 |
+
"step": 3800
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 5.41018387553041,
|
| 1104 |
+
"grad_norm": 0.2093198150396347,
|
| 1105 |
+
"learning_rate": 2.6111111111111113e-06,
|
| 1106 |
+
"loss": 0.0033,
|
| 1107 |
+
"step": 3825
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 5.445544554455446,
|
| 1111 |
+
"grad_norm": 1.7021733522415161,
|
| 1112 |
+
"learning_rate": 2.5555555555555557e-06,
|
| 1113 |
+
"loss": 0.0044,
|
| 1114 |
+
"step": 3850
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 5.480905233380481,
|
| 1118 |
+
"grad_norm": 0.4927394390106201,
|
| 1119 |
+
"learning_rate": 2.5e-06,
|
| 1120 |
+
"loss": 0.0033,
|
| 1121 |
+
"step": 3875
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 5.516265912305516,
|
| 1125 |
+
"grad_norm": 0.16159459948539734,
|
| 1126 |
+
"learning_rate": 2.4444444444444447e-06,
|
| 1127 |
+
"loss": 0.0035,
|
| 1128 |
+
"step": 3900
|
| 1129 |
+
},
|
| 1130 |
+
{
|
| 1131 |
+
"epoch": 5.551626591230551,
|
| 1132 |
+
"grad_norm": 0.25875765085220337,
|
| 1133 |
+
"learning_rate": 2.388888888888889e-06,
|
| 1134 |
+
"loss": 0.0049,
|
| 1135 |
+
"step": 3925
|
| 1136 |
+
},
|
| 1137 |
+
{
|
| 1138 |
+
"epoch": 5.586987270155587,
|
| 1139 |
+
"grad_norm": 0.22136487066745758,
|
| 1140 |
+
"learning_rate": 2.3333333333333336e-06,
|
| 1141 |
+
"loss": 0.0045,
|
| 1142 |
+
"step": 3950
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 5.6223479490806225,
|
| 1146 |
+
"grad_norm": 0.2594265341758728,
|
| 1147 |
+
"learning_rate": 2.277777777777778e-06,
|
| 1148 |
+
"loss": 0.0035,
|
| 1149 |
+
"step": 3975
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 5.657708628005658,
|
| 1153 |
+
"grad_norm": 0.46229287981987,
|
| 1154 |
+
"learning_rate": 2.222222222222222e-06,
|
| 1155 |
+
"loss": 0.005,
|
| 1156 |
+
"step": 4000
|
| 1157 |
+
},
|
| 1158 |
+
{
|
| 1159 |
+
"epoch": 5.657708628005658,
|
| 1160 |
+
"eval_loss": 0.2928106486797333,
|
| 1161 |
+
"eval_runtime": 2246.0565,
|
| 1162 |
+
"eval_samples_per_second": 2.396,
|
| 1163 |
+
"eval_steps_per_second": 0.15,
|
| 1164 |
+
"eval_wer": 0.16921149946319933,
|
| 1165 |
+
"step": 4000
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 5.693069306930693,
|
| 1169 |
+
"grad_norm": 0.6756157279014587,
|
| 1170 |
+
"learning_rate": 2.166666666666667e-06,
|
| 1171 |
+
"loss": 0.0033,
|
| 1172 |
+
"step": 4025
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 5.728429985855728,
|
| 1176 |
+
"grad_norm": 0.2873589098453522,
|
| 1177 |
+
"learning_rate": 2.1111111111111114e-06,
|
| 1178 |
+
"loss": 0.0038,
|
| 1179 |
+
"step": 4050
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 5.763790664780764,
|
| 1183 |
+
"grad_norm": 0.2609182596206665,
|
| 1184 |
+
"learning_rate": 2.0555555555555555e-06,
|
| 1185 |
+
"loss": 0.0033,
|
| 1186 |
+
"step": 4075
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 5.799151343705799,
|
| 1190 |
+
"grad_norm": 0.47305575013160706,
|
| 1191 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 1192 |
+
"loss": 0.0026,
|
| 1193 |
+
"step": 4100
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 5.834512022630834,
|
| 1197 |
+
"grad_norm": 0.11515957117080688,
|
| 1198 |
+
"learning_rate": 1.944444444444445e-06,
|
| 1199 |
+
"loss": 0.0028,
|
| 1200 |
+
"step": 4125
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 5.86987270155587,
|
| 1204 |
+
"grad_norm": 0.48609447479248047,
|
| 1205 |
+
"learning_rate": 1.888888888888889e-06,
|
| 1206 |
+
"loss": 0.0034,
|
| 1207 |
+
"step": 4150
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 5.905233380480905,
|
| 1211 |
+
"grad_norm": 1.1658227443695068,
|
| 1212 |
+
"learning_rate": 1.8333333333333333e-06,
|
| 1213 |
+
"loss": 0.0038,
|
| 1214 |
+
"step": 4175
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 5.9405940594059405,
|
| 1218 |
+
"grad_norm": 0.13836970925331116,
|
| 1219 |
+
"learning_rate": 1.777777777777778e-06,
|
| 1220 |
+
"loss": 0.0044,
|
| 1221 |
+
"step": 4200
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 5.975954738330976,
|
| 1225 |
+
"grad_norm": 0.46859803795814514,
|
| 1226 |
+
"learning_rate": 1.7222222222222224e-06,
|
| 1227 |
+
"loss": 0.0041,
|
| 1228 |
+
"step": 4225
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 6.011315417256012,
|
| 1232 |
+
"grad_norm": 0.21296146512031555,
|
| 1233 |
+
"learning_rate": 1.6666666666666667e-06,
|
| 1234 |
+
"loss": 0.0028,
|
| 1235 |
+
"step": 4250
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 6.046676096181047,
|
| 1239 |
+
"grad_norm": 0.05787177383899689,
|
| 1240 |
+
"learning_rate": 1.6111111111111113e-06,
|
| 1241 |
+
"loss": 0.002,
|
| 1242 |
+
"step": 4275
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 6.082036775106082,
|
| 1246 |
+
"grad_norm": 0.09653793275356293,
|
| 1247 |
+
"learning_rate": 1.5555555555555558e-06,
|
| 1248 |
+
"loss": 0.002,
|
| 1249 |
+
"step": 4300
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 6.117397454031117,
|
| 1253 |
+
"grad_norm": 0.40661904215812683,
|
| 1254 |
+
"learning_rate": 1.5e-06,
|
| 1255 |
+
"loss": 0.0015,
|
| 1256 |
+
"step": 4325
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 6.152758132956153,
|
| 1260 |
+
"grad_norm": 0.05121398717164993,
|
| 1261 |
+
"learning_rate": 1.4444444444444445e-06,
|
| 1262 |
+
"loss": 0.0015,
|
| 1263 |
+
"step": 4350
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 6.188118811881188,
|
| 1267 |
+
"grad_norm": 0.06613893806934357,
|
| 1268 |
+
"learning_rate": 1.3888888888888892e-06,
|
| 1269 |
+
"loss": 0.0014,
|
| 1270 |
+
"step": 4375
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 6.223479490806223,
|
| 1274 |
+
"grad_norm": 0.04856628179550171,
|
| 1275 |
+
"learning_rate": 1.3333333333333334e-06,
|
| 1276 |
+
"loss": 0.0023,
|
| 1277 |
+
"step": 4400
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 6.258840169731259,
|
| 1281 |
+
"grad_norm": 0.11592649668455124,
|
| 1282 |
+
"learning_rate": 1.2777777777777779e-06,
|
| 1283 |
+
"loss": 0.0013,
|
| 1284 |
+
"step": 4425
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 6.294200848656295,
|
| 1288 |
+
"grad_norm": 0.0709163099527359,
|
| 1289 |
+
"learning_rate": 1.2222222222222223e-06,
|
| 1290 |
+
"loss": 0.0017,
|
| 1291 |
+
"step": 4450
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 6.32956152758133,
|
| 1295 |
+
"grad_norm": 0.05270463228225708,
|
| 1296 |
+
"learning_rate": 1.1666666666666668e-06,
|
| 1297 |
+
"loss": 0.0016,
|
| 1298 |
+
"step": 4475
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 6.364922206506365,
|
| 1302 |
+
"grad_norm": 0.23842215538024902,
|
| 1303 |
+
"learning_rate": 1.111111111111111e-06,
|
| 1304 |
+
"loss": 0.0014,
|
| 1305 |
+
"step": 4500
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 6.4002828854314,
|
| 1309 |
+
"grad_norm": 0.046483710408210754,
|
| 1310 |
+
"learning_rate": 1.0555555555555557e-06,
|
| 1311 |
+
"loss": 0.0014,
|
| 1312 |
+
"step": 4525
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 6.435643564356436,
|
| 1316 |
+
"grad_norm": 0.050459932535886765,
|
| 1317 |
+
"learning_rate": 1.0000000000000002e-06,
|
| 1318 |
+
"loss": 0.0021,
|
| 1319 |
+
"step": 4550
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 6.471004243281471,
|
| 1323 |
+
"grad_norm": 0.04769385978579521,
|
| 1324 |
+
"learning_rate": 9.444444444444445e-07,
|
| 1325 |
+
"loss": 0.0014,
|
| 1326 |
+
"step": 4575
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 6.506364922206506,
|
| 1330 |
+
"grad_norm": 0.08741884678602219,
|
| 1331 |
+
"learning_rate": 8.88888888888889e-07,
|
| 1332 |
+
"loss": 0.0012,
|
| 1333 |
+
"step": 4600
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 6.5417256011315414,
|
| 1337 |
+
"grad_norm": 0.07672213762998581,
|
| 1338 |
+
"learning_rate": 8.333333333333333e-07,
|
| 1339 |
+
"loss": 0.0013,
|
| 1340 |
+
"step": 4625
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 6.5770862800565775,
|
| 1344 |
+
"grad_norm": 0.3090445399284363,
|
| 1345 |
+
"learning_rate": 7.777777777777779e-07,
|
| 1346 |
+
"loss": 0.0013,
|
| 1347 |
+
"step": 4650
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 6.612446958981613,
|
| 1351 |
+
"grad_norm": 0.06381077319383621,
|
| 1352 |
+
"learning_rate": 7.222222222222222e-07,
|
| 1353 |
+
"loss": 0.0023,
|
| 1354 |
+
"step": 4675
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 6.647807637906648,
|
| 1358 |
+
"grad_norm": 0.10808974504470825,
|
| 1359 |
+
"learning_rate": 6.666666666666667e-07,
|
| 1360 |
+
"loss": 0.0017,
|
| 1361 |
+
"step": 4700
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 6.683168316831683,
|
| 1365 |
+
"grad_norm": 0.06639378517866135,
|
| 1366 |
+
"learning_rate": 6.111111111111112e-07,
|
| 1367 |
+
"loss": 0.0013,
|
| 1368 |
+
"step": 4725
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 6.718528995756719,
|
| 1372 |
+
"grad_norm": 0.05687042698264122,
|
| 1373 |
+
"learning_rate": 5.555555555555555e-07,
|
| 1374 |
+
"loss": 0.0017,
|
| 1375 |
+
"step": 4750
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 6.753889674681754,
|
| 1379 |
+
"grad_norm": 0.050000376999378204,
|
| 1380 |
+
"learning_rate": 5.000000000000001e-07,
|
| 1381 |
+
"loss": 0.0012,
|
| 1382 |
+
"step": 4775
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 6.789250353606789,
|
| 1386 |
+
"grad_norm": 0.043993644416332245,
|
| 1387 |
+
"learning_rate": 4.444444444444445e-07,
|
| 1388 |
+
"loss": 0.0015,
|
| 1389 |
+
"step": 4800
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 6.824611032531824,
|
| 1393 |
+
"grad_norm": 0.0865250751376152,
|
| 1394 |
+
"learning_rate": 3.8888888888888895e-07,
|
| 1395 |
+
"loss": 0.0017,
|
| 1396 |
+
"step": 4825
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 6.85997171145686,
|
| 1400 |
+
"grad_norm": 0.06808679550886154,
|
| 1401 |
+
"learning_rate": 3.3333333333333335e-07,
|
| 1402 |
+
"loss": 0.0011,
|
| 1403 |
+
"step": 4850
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 6.8953323903818955,
|
| 1407 |
+
"grad_norm": 0.03820795193314552,
|
| 1408 |
+
"learning_rate": 2.7777777777777776e-07,
|
| 1409 |
+
"loss": 0.0014,
|
| 1410 |
+
"step": 4875
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 6.930693069306931,
|
| 1414 |
+
"grad_norm": 0.07484640926122665,
|
| 1415 |
+
"learning_rate": 2.2222222222222224e-07,
|
| 1416 |
+
"loss": 0.0014,
|
| 1417 |
+
"step": 4900
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 6.966053748231966,
|
| 1421 |
+
"grad_norm": 0.07878479361534119,
|
| 1422 |
+
"learning_rate": 1.6666666666666668e-07,
|
| 1423 |
+
"loss": 0.0012,
|
| 1424 |
+
"step": 4925
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 7.001414427157002,
|
| 1428 |
+
"grad_norm": 0.0370086245238781,
|
| 1429 |
+
"learning_rate": 1.1111111111111112e-07,
|
| 1430 |
+
"loss": 0.0017,
|
| 1431 |
+
"step": 4950
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 7.036775106082037,
|
| 1435 |
+
"grad_norm": 0.04006652534008026,
|
| 1436 |
+
"learning_rate": 5.555555555555556e-08,
|
| 1437 |
+
"loss": 0.0012,
|
| 1438 |
+
"step": 4975
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 7.072135785007072,
|
| 1442 |
+
"grad_norm": 0.04126604646444321,
|
| 1443 |
+
"learning_rate": 0.0,
|
| 1444 |
+
"loss": 0.0009,
|
| 1445 |
+
"step": 5000
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 7.072135785007072,
|
| 1449 |
+
"eval_loss": 0.3181316554546356,
|
| 1450 |
+
"eval_runtime": 2177.6024,
|
| 1451 |
+
"eval_samples_per_second": 2.471,
|
| 1452 |
+
"eval_steps_per_second": 0.155,
|
| 1453 |
+
"eval_wer": 0.16825718716449958,
|
| 1454 |
+
"step": 5000
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 7.072135785007072,
|
| 1458 |
+
"step": 5000,
|
| 1459 |
+
"total_flos": 5.4329055670370304e+20,
|
| 1460 |
+
"train_loss": 0.09240171913430095,
|
| 1461 |
+
"train_runtime": 60495.4593,
|
| 1462 |
+
"train_samples_per_second": 2.645,
|
| 1463 |
+
"train_steps_per_second": 0.083
|
| 1464 |
+
}
|
| 1465 |
+
],
|
| 1466 |
+
"logging_steps": 25,
|
| 1467 |
+
"max_steps": 5000,
|
| 1468 |
+
"num_input_tokens_seen": 0,
|
| 1469 |
+
"num_train_epochs": 8,
|
| 1470 |
+
"save_steps": 1000,
|
| 1471 |
+
"stateful_callbacks": {
|
| 1472 |
+
"TrainerControl": {
|
| 1473 |
+
"args": {
|
| 1474 |
+
"should_epoch_stop": false,
|
| 1475 |
+
"should_evaluate": false,
|
| 1476 |
+
"should_log": false,
|
| 1477 |
+
"should_save": true,
|
| 1478 |
+
"should_training_stop": true
|
| 1479 |
+
},
|
| 1480 |
+
"attributes": {}
|
| 1481 |
+
}
|
| 1482 |
+
},
|
| 1483 |
+
"total_flos": 5.4329055670370304e+20,
|
| 1484 |
+
"train_batch_size": 16,
|
| 1485 |
+
"trial_name": null,
|
| 1486 |
+
"trial_params": null
|
| 1487 |
+
}
|