End of training
Browse files- README.md +4 -4
- all_results.json +15 -0
- eval_results.json +9 -0
- train_results.json +9 -0
- trainer_state.json +1487 -0
README.md
CHANGED
|
@@ -4,7 +4,7 @@ base_model: openai/whisper-small
|
|
| 4 |
tags:
|
| 5 |
- generated_from_trainer
|
| 6 |
datasets:
|
| 7 |
-
- common_voice_17_0
|
| 8 |
metrics:
|
| 9 |
- wer
|
| 10 |
model-index:
|
|
@@ -14,8 +14,8 @@ model-index:
|
|
| 14 |
name: Automatic Speech Recognition
|
| 15 |
type: automatic-speech-recognition
|
| 16 |
dataset:
|
| 17 |
-
name: common_voice_17_0
|
| 18 |
-
type: common_voice_17_0
|
| 19 |
config: id
|
| 20 |
split: None
|
| 21 |
args: id
|
|
@@ -30,7 +30,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 30 |
|
| 31 |
# whisper-small-id
|
| 32 |
|
| 33 |
-
This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the common_voice_17_0 dataset.
|
| 34 |
It achieves the following results on the evaluation set:
|
| 35 |
- Loss: 0.3506
|
| 36 |
- Wer: 0.1617
|
|
|
|
| 4 |
tags:
|
| 5 |
- generated_from_trainer
|
| 6 |
datasets:
|
| 7 |
+
- mozilla-foundation/common_voice_17_0
|
| 8 |
metrics:
|
| 9 |
- wer
|
| 10 |
model-index:
|
|
|
|
| 14 |
name: Automatic Speech Recognition
|
| 15 |
type: automatic-speech-recognition
|
| 16 |
dataset:
|
| 17 |
+
name: mozilla-foundation/common_voice_17_0 id
|
| 18 |
+
type: mozilla-foundation/common_voice_17_0
|
| 19 |
config: id
|
| 20 |
split: None
|
| 21 |
args: id
|
|
|
|
| 30 |
|
| 31 |
# whisper-small-id
|
| 32 |
|
| 33 |
+
This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the mozilla-foundation/common_voice_17_0 id dataset.
|
| 34 |
It achieves the following results on the evaluation set:
|
| 35 |
- Loss: 0.3506
|
| 36 |
- Wer: 0.1617
|
all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 19.23076923076923,
|
| 3 |
+
"eval_loss": 0.35056987404823303,
|
| 4 |
+
"eval_runtime": 413.2463,
|
| 5 |
+
"eval_samples": 3641,
|
| 6 |
+
"eval_samples_per_second": 8.811,
|
| 7 |
+
"eval_steps_per_second": 0.552,
|
| 8 |
+
"eval_wer": 0.16172444196946495,
|
| 9 |
+
"total_flos": 4.616818092859392e+19,
|
| 10 |
+
"train_loss": 0.05540717707388103,
|
| 11 |
+
"train_runtime": 18195.7149,
|
| 12 |
+
"train_samples": 8319,
|
| 13 |
+
"train_samples_per_second": 8.793,
|
| 14 |
+
"train_steps_per_second": 0.275
|
| 15 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 19.23076923076923,
|
| 3 |
+
"eval_loss": 0.35056987404823303,
|
| 4 |
+
"eval_runtime": 413.2463,
|
| 5 |
+
"eval_samples": 3641,
|
| 6 |
+
"eval_samples_per_second": 8.811,
|
| 7 |
+
"eval_steps_per_second": 0.552,
|
| 8 |
+
"eval_wer": 0.16172444196946495
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 19.23076923076923,
|
| 3 |
+
"total_flos": 4.616818092859392e+19,
|
| 4 |
+
"train_loss": 0.05540717707388103,
|
| 5 |
+
"train_runtime": 18195.7149,
|
| 6 |
+
"train_samples": 8319,
|
| 7 |
+
"train_samples_per_second": 8.793,
|
| 8 |
+
"train_steps_per_second": 0.275
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1487 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 19.23076923076923,
|
| 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": 0.09615384615384616,
|
| 13 |
+
"grad_norm": 14.387816429138184,
|
| 14 |
+
"learning_rate": 4.6000000000000004e-07,
|
| 15 |
+
"loss": 1.0674,
|
| 16 |
+
"step": 25
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.19230769230769232,
|
| 20 |
+
"grad_norm": 11.297602653503418,
|
| 21 |
+
"learning_rate": 9.600000000000001e-07,
|
| 22 |
+
"loss": 0.9654,
|
| 23 |
+
"step": 50
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.28846153846153844,
|
| 27 |
+
"grad_norm": 8.031381607055664,
|
| 28 |
+
"learning_rate": 1.46e-06,
|
| 29 |
+
"loss": 0.7994,
|
| 30 |
+
"step": 75
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.38461538461538464,
|
| 34 |
+
"grad_norm": 6.947944641113281,
|
| 35 |
+
"learning_rate": 1.9600000000000003e-06,
|
| 36 |
+
"loss": 0.5243,
|
| 37 |
+
"step": 100
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.4807692307692308,
|
| 41 |
+
"grad_norm": 5.577055931091309,
|
| 42 |
+
"learning_rate": 2.46e-06,
|
| 43 |
+
"loss": 0.4378,
|
| 44 |
+
"step": 125
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.5769230769230769,
|
| 48 |
+
"grad_norm": 5.237943649291992,
|
| 49 |
+
"learning_rate": 2.96e-06,
|
| 50 |
+
"loss": 0.39,
|
| 51 |
+
"step": 150
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.6730769230769231,
|
| 55 |
+
"grad_norm": 5.476370811462402,
|
| 56 |
+
"learning_rate": 3.46e-06,
|
| 57 |
+
"loss": 0.3755,
|
| 58 |
+
"step": 175
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.7692307692307693,
|
| 62 |
+
"grad_norm": 6.1683244705200195,
|
| 63 |
+
"learning_rate": 3.96e-06,
|
| 64 |
+
"loss": 0.3851,
|
| 65 |
+
"step": 200
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.8653846153846154,
|
| 69 |
+
"grad_norm": 5.542613506317139,
|
| 70 |
+
"learning_rate": 4.4600000000000005e-06,
|
| 71 |
+
"loss": 0.3461,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.9615384615384616,
|
| 76 |
+
"grad_norm": 7.256478786468506,
|
| 77 |
+
"learning_rate": 4.960000000000001e-06,
|
| 78 |
+
"loss": 0.3369,
|
| 79 |
+
"step": 250
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 1.0576923076923077,
|
| 83 |
+
"grad_norm": 5.407237529754639,
|
| 84 |
+
"learning_rate": 5.460000000000001e-06,
|
| 85 |
+
"loss": 0.2975,
|
| 86 |
+
"step": 275
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 1.1538461538461537,
|
| 90 |
+
"grad_norm": 5.3987579345703125,
|
| 91 |
+
"learning_rate": 5.9600000000000005e-06,
|
| 92 |
+
"loss": 0.2739,
|
| 93 |
+
"step": 300
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 1.25,
|
| 97 |
+
"grad_norm": 4.13232946395874,
|
| 98 |
+
"learning_rate": 6.460000000000001e-06,
|
| 99 |
+
"loss": 0.2519,
|
| 100 |
+
"step": 325
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 1.3461538461538463,
|
| 104 |
+
"grad_norm": 4.822619438171387,
|
| 105 |
+
"learning_rate": 6.96e-06,
|
| 106 |
+
"loss": 0.2663,
|
| 107 |
+
"step": 350
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 1.4423076923076923,
|
| 111 |
+
"grad_norm": 4.323586463928223,
|
| 112 |
+
"learning_rate": 7.4600000000000006e-06,
|
| 113 |
+
"loss": 0.2482,
|
| 114 |
+
"step": 375
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 1.5384615384615383,
|
| 118 |
+
"grad_norm": 4.278296947479248,
|
| 119 |
+
"learning_rate": 7.960000000000002e-06,
|
| 120 |
+
"loss": 0.2525,
|
| 121 |
+
"step": 400
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 1.6346153846153846,
|
| 125 |
+
"grad_norm": 5.396093368530273,
|
| 126 |
+
"learning_rate": 8.46e-06,
|
| 127 |
+
"loss": 0.2575,
|
| 128 |
+
"step": 425
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 1.7307692307692308,
|
| 132 |
+
"grad_norm": 4.330433368682861,
|
| 133 |
+
"learning_rate": 8.96e-06,
|
| 134 |
+
"loss": 0.2502,
|
| 135 |
+
"step": 450
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 1.8269230769230769,
|
| 139 |
+
"grad_norm": 4.134779930114746,
|
| 140 |
+
"learning_rate": 9.460000000000001e-06,
|
| 141 |
+
"loss": 0.2615,
|
| 142 |
+
"step": 475
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 1.9230769230769231,
|
| 146 |
+
"grad_norm": 4.702469348907471,
|
| 147 |
+
"learning_rate": 9.960000000000001e-06,
|
| 148 |
+
"loss": 0.2542,
|
| 149 |
+
"step": 500
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 2.019230769230769,
|
| 153 |
+
"grad_norm": 2.5487704277038574,
|
| 154 |
+
"learning_rate": 9.94888888888889e-06,
|
| 155 |
+
"loss": 0.2502,
|
| 156 |
+
"step": 525
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 2.1153846153846154,
|
| 160 |
+
"grad_norm": 3.5824356079101562,
|
| 161 |
+
"learning_rate": 9.893333333333334e-06,
|
| 162 |
+
"loss": 0.1373,
|
| 163 |
+
"step": 550
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 2.2115384615384617,
|
| 167 |
+
"grad_norm": 3.7268402576446533,
|
| 168 |
+
"learning_rate": 9.837777777777778e-06,
|
| 169 |
+
"loss": 0.1367,
|
| 170 |
+
"step": 575
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 2.3076923076923075,
|
| 174 |
+
"grad_norm": 3.243474245071411,
|
| 175 |
+
"learning_rate": 9.782222222222222e-06,
|
| 176 |
+
"loss": 0.1298,
|
| 177 |
+
"step": 600
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 2.4038461538461537,
|
| 181 |
+
"grad_norm": 3.1370737552642822,
|
| 182 |
+
"learning_rate": 9.726666666666668e-06,
|
| 183 |
+
"loss": 0.135,
|
| 184 |
+
"step": 625
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 2.5,
|
| 188 |
+
"grad_norm": 2.788139581680298,
|
| 189 |
+
"learning_rate": 9.671111111111112e-06,
|
| 190 |
+
"loss": 0.1239,
|
| 191 |
+
"step": 650
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 2.5961538461538463,
|
| 195 |
+
"grad_norm": 3.4849274158477783,
|
| 196 |
+
"learning_rate": 9.615555555555558e-06,
|
| 197 |
+
"loss": 0.1226,
|
| 198 |
+
"step": 675
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 2.6923076923076925,
|
| 202 |
+
"grad_norm": 3.447479009628296,
|
| 203 |
+
"learning_rate": 9.56e-06,
|
| 204 |
+
"loss": 0.1281,
|
| 205 |
+
"step": 700
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 2.7884615384615383,
|
| 209 |
+
"grad_norm": 3.737367868423462,
|
| 210 |
+
"learning_rate": 9.504444444444446e-06,
|
| 211 |
+
"loss": 0.121,
|
| 212 |
+
"step": 725
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 2.8846153846153846,
|
| 216 |
+
"grad_norm": 3.7365550994873047,
|
| 217 |
+
"learning_rate": 9.44888888888889e-06,
|
| 218 |
+
"loss": 0.1284,
|
| 219 |
+
"step": 750
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 2.980769230769231,
|
| 223 |
+
"grad_norm": 4.205287933349609,
|
| 224 |
+
"learning_rate": 9.393333333333334e-06,
|
| 225 |
+
"loss": 0.1393,
|
| 226 |
+
"step": 775
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 3.076923076923077,
|
| 230 |
+
"grad_norm": 2.6460373401641846,
|
| 231 |
+
"learning_rate": 9.33777777777778e-06,
|
| 232 |
+
"loss": 0.0834,
|
| 233 |
+
"step": 800
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 3.173076923076923,
|
| 237 |
+
"grad_norm": 2.1553828716278076,
|
| 238 |
+
"learning_rate": 9.282222222222222e-06,
|
| 239 |
+
"loss": 0.0541,
|
| 240 |
+
"step": 825
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 3.269230769230769,
|
| 244 |
+
"grad_norm": 1.773935079574585,
|
| 245 |
+
"learning_rate": 9.226666666666668e-06,
|
| 246 |
+
"loss": 0.0559,
|
| 247 |
+
"step": 850
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 3.3653846153846154,
|
| 251 |
+
"grad_norm": 2.047421455383301,
|
| 252 |
+
"learning_rate": 9.171111111111112e-06,
|
| 253 |
+
"loss": 0.0618,
|
| 254 |
+
"step": 875
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 3.4615384615384617,
|
| 258 |
+
"grad_norm": 2.5151689052581787,
|
| 259 |
+
"learning_rate": 9.115555555555556e-06,
|
| 260 |
+
"loss": 0.0604,
|
| 261 |
+
"step": 900
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 3.5576923076923075,
|
| 265 |
+
"grad_norm": 1.7141085863113403,
|
| 266 |
+
"learning_rate": 9.060000000000001e-06,
|
| 267 |
+
"loss": 0.0567,
|
| 268 |
+
"step": 925
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 3.6538461538461537,
|
| 272 |
+
"grad_norm": 2.3757855892181396,
|
| 273 |
+
"learning_rate": 9.004444444444445e-06,
|
| 274 |
+
"loss": 0.0567,
|
| 275 |
+
"step": 950
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 3.75,
|
| 279 |
+
"grad_norm": 2.048637866973877,
|
| 280 |
+
"learning_rate": 8.94888888888889e-06,
|
| 281 |
+
"loss": 0.0595,
|
| 282 |
+
"step": 975
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 3.8461538461538463,
|
| 286 |
+
"grad_norm": 2.78825306892395,
|
| 287 |
+
"learning_rate": 8.893333333333333e-06,
|
| 288 |
+
"loss": 0.0547,
|
| 289 |
+
"step": 1000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 3.8461538461538463,
|
| 293 |
+
"eval_loss": 0.2570088803768158,
|
| 294 |
+
"eval_runtime": 418.585,
|
| 295 |
+
"eval_samples_per_second": 8.698,
|
| 296 |
+
"eval_steps_per_second": 0.545,
|
| 297 |
+
"eval_wer": 0.16873172769038006,
|
| 298 |
+
"step": 1000
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 3.9423076923076925,
|
| 302 |
+
"grad_norm": 2.7837064266204834,
|
| 303 |
+
"learning_rate": 8.83777777777778e-06,
|
| 304 |
+
"loss": 0.0566,
|
| 305 |
+
"step": 1025
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 4.038461538461538,
|
| 309 |
+
"grad_norm": 1.4186588525772095,
|
| 310 |
+
"learning_rate": 8.782222222222223e-06,
|
| 311 |
+
"loss": 0.0472,
|
| 312 |
+
"step": 1050
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 4.134615384615385,
|
| 316 |
+
"grad_norm": 1.4111158847808838,
|
| 317 |
+
"learning_rate": 8.726666666666667e-06,
|
| 318 |
+
"loss": 0.026,
|
| 319 |
+
"step": 1075
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 4.230769230769231,
|
| 323 |
+
"grad_norm": 0.8355940580368042,
|
| 324 |
+
"learning_rate": 8.671111111111113e-06,
|
| 325 |
+
"loss": 0.025,
|
| 326 |
+
"step": 1100
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 4.326923076923077,
|
| 330 |
+
"grad_norm": 2.8829832077026367,
|
| 331 |
+
"learning_rate": 8.615555555555555e-06,
|
| 332 |
+
"loss": 0.0281,
|
| 333 |
+
"step": 1125
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 4.423076923076923,
|
| 337 |
+
"grad_norm": 1.8035304546356201,
|
| 338 |
+
"learning_rate": 8.560000000000001e-06,
|
| 339 |
+
"loss": 0.0275,
|
| 340 |
+
"step": 1150
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 4.519230769230769,
|
| 344 |
+
"grad_norm": 2.558990240097046,
|
| 345 |
+
"learning_rate": 8.504444444444445e-06,
|
| 346 |
+
"loss": 0.0284,
|
| 347 |
+
"step": 1175
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 4.615384615384615,
|
| 351 |
+
"grad_norm": 3.0063929557800293,
|
| 352 |
+
"learning_rate": 8.448888888888889e-06,
|
| 353 |
+
"loss": 0.0262,
|
| 354 |
+
"step": 1200
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 4.711538461538462,
|
| 358 |
+
"grad_norm": 1.181315541267395,
|
| 359 |
+
"learning_rate": 8.393333333333335e-06,
|
| 360 |
+
"loss": 0.0224,
|
| 361 |
+
"step": 1225
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 4.8076923076923075,
|
| 365 |
+
"grad_norm": 1.4534687995910645,
|
| 366 |
+
"learning_rate": 8.337777777777777e-06,
|
| 367 |
+
"loss": 0.026,
|
| 368 |
+
"step": 1250
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 4.903846153846154,
|
| 372 |
+
"grad_norm": 1.941493034362793,
|
| 373 |
+
"learning_rate": 8.282222222222223e-06,
|
| 374 |
+
"loss": 0.0276,
|
| 375 |
+
"step": 1275
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 5.0,
|
| 379 |
+
"grad_norm": 1.048115849494934,
|
| 380 |
+
"learning_rate": 8.226666666666667e-06,
|
| 381 |
+
"loss": 0.0278,
|
| 382 |
+
"step": 1300
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 5.096153846153846,
|
| 386 |
+
"grad_norm": 1.3306798934936523,
|
| 387 |
+
"learning_rate": 8.171111111111113e-06,
|
| 388 |
+
"loss": 0.0136,
|
| 389 |
+
"step": 1325
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 5.1923076923076925,
|
| 393 |
+
"grad_norm": 1.5832746028900146,
|
| 394 |
+
"learning_rate": 8.115555555555557e-06,
|
| 395 |
+
"loss": 0.0128,
|
| 396 |
+
"step": 1350
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 5.288461538461538,
|
| 400 |
+
"grad_norm": 1.175854206085205,
|
| 401 |
+
"learning_rate": 8.06e-06,
|
| 402 |
+
"loss": 0.0115,
|
| 403 |
+
"step": 1375
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 5.384615384615385,
|
| 407 |
+
"grad_norm": 1.0961897373199463,
|
| 408 |
+
"learning_rate": 8.004444444444445e-06,
|
| 409 |
+
"loss": 0.0128,
|
| 410 |
+
"step": 1400
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 5.480769230769231,
|
| 414 |
+
"grad_norm": 1.1296463012695312,
|
| 415 |
+
"learning_rate": 7.948888888888889e-06,
|
| 416 |
+
"loss": 0.0116,
|
| 417 |
+
"step": 1425
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 5.576923076923077,
|
| 421 |
+
"grad_norm": 1.3500127792358398,
|
| 422 |
+
"learning_rate": 7.893333333333335e-06,
|
| 423 |
+
"loss": 0.0127,
|
| 424 |
+
"step": 1450
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 5.673076923076923,
|
| 428 |
+
"grad_norm": 1.6575067043304443,
|
| 429 |
+
"learning_rate": 7.837777777777779e-06,
|
| 430 |
+
"loss": 0.0126,
|
| 431 |
+
"step": 1475
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 5.769230769230769,
|
| 435 |
+
"grad_norm": 2.729482412338257,
|
| 436 |
+
"learning_rate": 7.782222222222223e-06,
|
| 437 |
+
"loss": 0.0125,
|
| 438 |
+
"step": 1500
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 5.865384615384615,
|
| 442 |
+
"grad_norm": 1.4455863237380981,
|
| 443 |
+
"learning_rate": 7.726666666666667e-06,
|
| 444 |
+
"loss": 0.0122,
|
| 445 |
+
"step": 1525
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 5.961538461538462,
|
| 449 |
+
"grad_norm": 1.5944123268127441,
|
| 450 |
+
"learning_rate": 7.67111111111111e-06,
|
| 451 |
+
"loss": 0.0147,
|
| 452 |
+
"step": 1550
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 6.0576923076923075,
|
| 456 |
+
"grad_norm": 0.8240692615509033,
|
| 457 |
+
"learning_rate": 7.6155555555555564e-06,
|
| 458 |
+
"loss": 0.0094,
|
| 459 |
+
"step": 1575
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 6.153846153846154,
|
| 463 |
+
"grad_norm": 0.41078874468803406,
|
| 464 |
+
"learning_rate": 7.5600000000000005e-06,
|
| 465 |
+
"loss": 0.0063,
|
| 466 |
+
"step": 1600
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 6.25,
|
| 470 |
+
"grad_norm": 0.2937760651111603,
|
| 471 |
+
"learning_rate": 7.504444444444445e-06,
|
| 472 |
+
"loss": 0.008,
|
| 473 |
+
"step": 1625
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 6.346153846153846,
|
| 477 |
+
"grad_norm": 0.2572005093097687,
|
| 478 |
+
"learning_rate": 7.44888888888889e-06,
|
| 479 |
+
"loss": 0.006,
|
| 480 |
+
"step": 1650
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 6.4423076923076925,
|
| 484 |
+
"grad_norm": 0.5730597376823425,
|
| 485 |
+
"learning_rate": 7.393333333333333e-06,
|
| 486 |
+
"loss": 0.0083,
|
| 487 |
+
"step": 1675
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 6.538461538461538,
|
| 491 |
+
"grad_norm": 0.9147136211395264,
|
| 492 |
+
"learning_rate": 7.337777777777778e-06,
|
| 493 |
+
"loss": 0.0066,
|
| 494 |
+
"step": 1700
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 6.634615384615385,
|
| 498 |
+
"grad_norm": 0.45330244302749634,
|
| 499 |
+
"learning_rate": 7.282222222222222e-06,
|
| 500 |
+
"loss": 0.0065,
|
| 501 |
+
"step": 1725
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 6.730769230769231,
|
| 505 |
+
"grad_norm": 0.6328800916671753,
|
| 506 |
+
"learning_rate": 7.226666666666667e-06,
|
| 507 |
+
"loss": 0.0078,
|
| 508 |
+
"step": 1750
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 6.826923076923077,
|
| 512 |
+
"grad_norm": 0.2872777581214905,
|
| 513 |
+
"learning_rate": 7.171111111111112e-06,
|
| 514 |
+
"loss": 0.0079,
|
| 515 |
+
"step": 1775
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 6.923076923076923,
|
| 519 |
+
"grad_norm": 1.7963801622390747,
|
| 520 |
+
"learning_rate": 7.115555555555557e-06,
|
| 521 |
+
"loss": 0.0078,
|
| 522 |
+
"step": 1800
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 7.019230769230769,
|
| 526 |
+
"grad_norm": 0.2437676042318344,
|
| 527 |
+
"learning_rate": 7.06e-06,
|
| 528 |
+
"loss": 0.0068,
|
| 529 |
+
"step": 1825
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 7.115384615384615,
|
| 533 |
+
"grad_norm": 0.20286525785923004,
|
| 534 |
+
"learning_rate": 7.004444444444445e-06,
|
| 535 |
+
"loss": 0.0039,
|
| 536 |
+
"step": 1850
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 7.211538461538462,
|
| 540 |
+
"grad_norm": 0.16271603107452393,
|
| 541 |
+
"learning_rate": 6.948888888888889e-06,
|
| 542 |
+
"loss": 0.0038,
|
| 543 |
+
"step": 1875
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 7.3076923076923075,
|
| 547 |
+
"grad_norm": 0.12690773606300354,
|
| 548 |
+
"learning_rate": 6.893333333333334e-06,
|
| 549 |
+
"loss": 0.0037,
|
| 550 |
+
"step": 1900
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 7.403846153846154,
|
| 554 |
+
"grad_norm": 0.41975799202919006,
|
| 555 |
+
"learning_rate": 6.837777777777779e-06,
|
| 556 |
+
"loss": 0.0042,
|
| 557 |
+
"step": 1925
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 7.5,
|
| 561 |
+
"grad_norm": 0.2555113732814789,
|
| 562 |
+
"learning_rate": 6.782222222222222e-06,
|
| 563 |
+
"loss": 0.0033,
|
| 564 |
+
"step": 1950
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 7.596153846153846,
|
| 568 |
+
"grad_norm": 1.6203404664993286,
|
| 569 |
+
"learning_rate": 6.726666666666667e-06,
|
| 570 |
+
"loss": 0.0049,
|
| 571 |
+
"step": 1975
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 7.6923076923076925,
|
| 575 |
+
"grad_norm": 0.16367614269256592,
|
| 576 |
+
"learning_rate": 6.671111111111112e-06,
|
| 577 |
+
"loss": 0.0034,
|
| 578 |
+
"step": 2000
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 7.6923076923076925,
|
| 582 |
+
"eval_loss": 0.31435152888298035,
|
| 583 |
+
"eval_runtime": 414.9289,
|
| 584 |
+
"eval_samples_per_second": 8.775,
|
| 585 |
+
"eval_steps_per_second": 0.549,
|
| 586 |
+
"eval_wer": 0.15889368416167804,
|
| 587 |
+
"step": 2000
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 7.788461538461538,
|
| 591 |
+
"grad_norm": 0.1244119256734848,
|
| 592 |
+
"learning_rate": 6.615555555555556e-06,
|
| 593 |
+
"loss": 0.0035,
|
| 594 |
+
"step": 2025
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 7.884615384615385,
|
| 598 |
+
"grad_norm": 0.7467623353004456,
|
| 599 |
+
"learning_rate": 6.560000000000001e-06,
|
| 600 |
+
"loss": 0.0042,
|
| 601 |
+
"step": 2050
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 7.980769230769231,
|
| 605 |
+
"grad_norm": 0.26792147755622864,
|
| 606 |
+
"learning_rate": 6.504444444444446e-06,
|
| 607 |
+
"loss": 0.0043,
|
| 608 |
+
"step": 2075
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 8.076923076923077,
|
| 612 |
+
"grad_norm": 0.0888567715883255,
|
| 613 |
+
"learning_rate": 6.448888888888889e-06,
|
| 614 |
+
"loss": 0.0029,
|
| 615 |
+
"step": 2100
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 8.173076923076923,
|
| 619 |
+
"grad_norm": 0.22752462327480316,
|
| 620 |
+
"learning_rate": 6.393333333333334e-06,
|
| 621 |
+
"loss": 0.003,
|
| 622 |
+
"step": 2125
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 8.26923076923077,
|
| 626 |
+
"grad_norm": 0.11363103985786438,
|
| 627 |
+
"learning_rate": 6.3377777777777786e-06,
|
| 628 |
+
"loss": 0.0023,
|
| 629 |
+
"step": 2150
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 8.365384615384615,
|
| 633 |
+
"grad_norm": 0.4889778196811676,
|
| 634 |
+
"learning_rate": 6.282222222222223e-06,
|
| 635 |
+
"loss": 0.0024,
|
| 636 |
+
"step": 2175
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 8.461538461538462,
|
| 640 |
+
"grad_norm": 0.08544881641864777,
|
| 641 |
+
"learning_rate": 6.2266666666666675e-06,
|
| 642 |
+
"loss": 0.0023,
|
| 643 |
+
"step": 2200
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 8.557692307692308,
|
| 647 |
+
"grad_norm": 0.1377352476119995,
|
| 648 |
+
"learning_rate": 6.171111111111112e-06,
|
| 649 |
+
"loss": 0.0032,
|
| 650 |
+
"step": 2225
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 8.653846153846153,
|
| 654 |
+
"grad_norm": 0.49440711736679077,
|
| 655 |
+
"learning_rate": 6.1155555555555555e-06,
|
| 656 |
+
"loss": 0.0022,
|
| 657 |
+
"step": 2250
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 8.75,
|
| 661 |
+
"grad_norm": 0.10170517861843109,
|
| 662 |
+
"learning_rate": 6.0600000000000004e-06,
|
| 663 |
+
"loss": 0.0029,
|
| 664 |
+
"step": 2275
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 8.846153846153847,
|
| 668 |
+
"grad_norm": 0.0798933133482933,
|
| 669 |
+
"learning_rate": 6.004444444444445e-06,
|
| 670 |
+
"loss": 0.0025,
|
| 671 |
+
"step": 2300
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 8.942307692307692,
|
| 675 |
+
"grad_norm": 0.06800112873315811,
|
| 676 |
+
"learning_rate": 5.948888888888889e-06,
|
| 677 |
+
"loss": 0.0023,
|
| 678 |
+
"step": 2325
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 9.038461538461538,
|
| 682 |
+
"grad_norm": 0.05597899481654167,
|
| 683 |
+
"learning_rate": 5.893333333333334e-06,
|
| 684 |
+
"loss": 0.0019,
|
| 685 |
+
"step": 2350
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 9.134615384615385,
|
| 689 |
+
"grad_norm": 0.055818960070610046,
|
| 690 |
+
"learning_rate": 5.837777777777777e-06,
|
| 691 |
+
"loss": 0.0016,
|
| 692 |
+
"step": 2375
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 9.23076923076923,
|
| 696 |
+
"grad_norm": 0.07546485215425491,
|
| 697 |
+
"learning_rate": 5.782222222222222e-06,
|
| 698 |
+
"loss": 0.002,
|
| 699 |
+
"step": 2400
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 9.326923076923077,
|
| 703 |
+
"grad_norm": 0.06564825028181076,
|
| 704 |
+
"learning_rate": 5.726666666666667e-06,
|
| 705 |
+
"loss": 0.0016,
|
| 706 |
+
"step": 2425
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 9.423076923076923,
|
| 710 |
+
"grad_norm": 0.0399746373295784,
|
| 711 |
+
"learning_rate": 5.671111111111112e-06,
|
| 712 |
+
"loss": 0.0016,
|
| 713 |
+
"step": 2450
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 9.51923076923077,
|
| 717 |
+
"grad_norm": 0.059612762182950974,
|
| 718 |
+
"learning_rate": 5.615555555555556e-06,
|
| 719 |
+
"loss": 0.0017,
|
| 720 |
+
"step": 2475
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 9.615384615384615,
|
| 724 |
+
"grad_norm": 0.06947290152311325,
|
| 725 |
+
"learning_rate": 5.560000000000001e-06,
|
| 726 |
+
"loss": 0.0016,
|
| 727 |
+
"step": 2500
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 9.711538461538462,
|
| 731 |
+
"grad_norm": 0.989651083946228,
|
| 732 |
+
"learning_rate": 5.504444444444444e-06,
|
| 733 |
+
"loss": 0.0017,
|
| 734 |
+
"step": 2525
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 9.807692307692308,
|
| 738 |
+
"grad_norm": 0.07500547170639038,
|
| 739 |
+
"learning_rate": 5.448888888888889e-06,
|
| 740 |
+
"loss": 0.0016,
|
| 741 |
+
"step": 2550
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 9.903846153846153,
|
| 745 |
+
"grad_norm": 0.09042783081531525,
|
| 746 |
+
"learning_rate": 5.393333333333334e-06,
|
| 747 |
+
"loss": 0.0017,
|
| 748 |
+
"step": 2575
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 10.0,
|
| 752 |
+
"grad_norm": 0.3791869282722473,
|
| 753 |
+
"learning_rate": 5.337777777777779e-06,
|
| 754 |
+
"loss": 0.0017,
|
| 755 |
+
"step": 2600
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 10.096153846153847,
|
| 759 |
+
"grad_norm": 0.043280228972435,
|
| 760 |
+
"learning_rate": 5.282222222222223e-06,
|
| 761 |
+
"loss": 0.0013,
|
| 762 |
+
"step": 2625
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 10.192307692307692,
|
| 766 |
+
"grad_norm": 0.04278066009283066,
|
| 767 |
+
"learning_rate": 5.226666666666667e-06,
|
| 768 |
+
"loss": 0.0014,
|
| 769 |
+
"step": 2650
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 10.288461538461538,
|
| 773 |
+
"grad_norm": 0.043745964765548706,
|
| 774 |
+
"learning_rate": 5.171111111111111e-06,
|
| 775 |
+
"loss": 0.0013,
|
| 776 |
+
"step": 2675
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 10.384615384615385,
|
| 780 |
+
"grad_norm": 0.05145561695098877,
|
| 781 |
+
"learning_rate": 5.115555555555556e-06,
|
| 782 |
+
"loss": 0.0013,
|
| 783 |
+
"step": 2700
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 10.48076923076923,
|
| 787 |
+
"grad_norm": 0.04810141772031784,
|
| 788 |
+
"learning_rate": 5.060000000000001e-06,
|
| 789 |
+
"loss": 0.0013,
|
| 790 |
+
"step": 2725
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 10.576923076923077,
|
| 794 |
+
"grad_norm": 0.046568796038627625,
|
| 795 |
+
"learning_rate": 5.004444444444445e-06,
|
| 796 |
+
"loss": 0.002,
|
| 797 |
+
"step": 2750
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 10.673076923076923,
|
| 801 |
+
"grad_norm": 0.05206083878874779,
|
| 802 |
+
"learning_rate": 4.94888888888889e-06,
|
| 803 |
+
"loss": 0.0013,
|
| 804 |
+
"step": 2775
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 10.76923076923077,
|
| 808 |
+
"grad_norm": 0.04524935781955719,
|
| 809 |
+
"learning_rate": 4.893333333333334e-06,
|
| 810 |
+
"loss": 0.0013,
|
| 811 |
+
"step": 2800
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 10.865384615384615,
|
| 815 |
+
"grad_norm": 0.05039019510149956,
|
| 816 |
+
"learning_rate": 4.837777777777778e-06,
|
| 817 |
+
"loss": 0.0012,
|
| 818 |
+
"step": 2825
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 10.961538461538462,
|
| 822 |
+
"grad_norm": 0.048185594379901886,
|
| 823 |
+
"learning_rate": 4.7822222222222226e-06,
|
| 824 |
+
"loss": 0.0024,
|
| 825 |
+
"step": 2850
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 11.057692307692308,
|
| 829 |
+
"grad_norm": 0.04220257326960564,
|
| 830 |
+
"learning_rate": 4.7266666666666674e-06,
|
| 831 |
+
"loss": 0.0012,
|
| 832 |
+
"step": 2875
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 11.153846153846153,
|
| 836 |
+
"grad_norm": 0.04200848937034607,
|
| 837 |
+
"learning_rate": 4.6711111111111115e-06,
|
| 838 |
+
"loss": 0.0011,
|
| 839 |
+
"step": 2900
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 11.25,
|
| 843 |
+
"grad_norm": 0.04376620426774025,
|
| 844 |
+
"learning_rate": 4.6155555555555555e-06,
|
| 845 |
+
"loss": 0.0011,
|
| 846 |
+
"step": 2925
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 11.346153846153847,
|
| 850 |
+
"grad_norm": 0.03870835155248642,
|
| 851 |
+
"learning_rate": 4.56e-06,
|
| 852 |
+
"loss": 0.0011,
|
| 853 |
+
"step": 2950
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 11.442307692307692,
|
| 857 |
+
"grad_norm": 0.03991697356104851,
|
| 858 |
+
"learning_rate": 4.504444444444444e-06,
|
| 859 |
+
"loss": 0.0011,
|
| 860 |
+
"step": 2975
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 11.538461538461538,
|
| 864 |
+
"grad_norm": 0.03659369796514511,
|
| 865 |
+
"learning_rate": 4.448888888888889e-06,
|
| 866 |
+
"loss": 0.0011,
|
| 867 |
+
"step": 3000
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 11.538461538461538,
|
| 871 |
+
"eval_loss": 0.33577749133110046,
|
| 872 |
+
"eval_runtime": 414.076,
|
| 873 |
+
"eval_samples_per_second": 8.793,
|
| 874 |
+
"eval_steps_per_second": 0.551,
|
| 875 |
+
"eval_wer": 0.16153881850665924,
|
| 876 |
+
"step": 3000
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 11.634615384615385,
|
| 880 |
+
"grad_norm": 0.035982731729745865,
|
| 881 |
+
"learning_rate": 4.393333333333334e-06,
|
| 882 |
+
"loss": 0.0011,
|
| 883 |
+
"step": 3025
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 11.73076923076923,
|
| 887 |
+
"grad_norm": 0.04060469567775726,
|
| 888 |
+
"learning_rate": 4.337777777777778e-06,
|
| 889 |
+
"loss": 0.0011,
|
| 890 |
+
"step": 3050
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 11.826923076923077,
|
| 894 |
+
"grad_norm": 0.03556708246469498,
|
| 895 |
+
"learning_rate": 4.282222222222222e-06,
|
| 896 |
+
"loss": 0.0011,
|
| 897 |
+
"step": 3075
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 11.923076923076923,
|
| 901 |
+
"grad_norm": 0.037598781287670135,
|
| 902 |
+
"learning_rate": 4.226666666666667e-06,
|
| 903 |
+
"loss": 0.001,
|
| 904 |
+
"step": 3100
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 12.01923076923077,
|
| 908 |
+
"grad_norm": 0.03427188843488693,
|
| 909 |
+
"learning_rate": 4.171111111111111e-06,
|
| 910 |
+
"loss": 0.0011,
|
| 911 |
+
"step": 3125
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 12.115384615384615,
|
| 915 |
+
"grad_norm": 0.03573931008577347,
|
| 916 |
+
"learning_rate": 4.115555555555556e-06,
|
| 917 |
+
"loss": 0.0009,
|
| 918 |
+
"step": 3150
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 12.211538461538462,
|
| 922 |
+
"grad_norm": 0.03210974112153053,
|
| 923 |
+
"learning_rate": 4.060000000000001e-06,
|
| 924 |
+
"loss": 0.0009,
|
| 925 |
+
"step": 3175
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 12.307692307692308,
|
| 929 |
+
"grad_norm": 0.036103058606386185,
|
| 930 |
+
"learning_rate": 4.004444444444445e-06,
|
| 931 |
+
"loss": 0.001,
|
| 932 |
+
"step": 3200
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 12.403846153846153,
|
| 936 |
+
"grad_norm": 0.036252211779356,
|
| 937 |
+
"learning_rate": 3.948888888888889e-06,
|
| 938 |
+
"loss": 0.001,
|
| 939 |
+
"step": 3225
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 12.5,
|
| 943 |
+
"grad_norm": 0.03289297968149185,
|
| 944 |
+
"learning_rate": 3.893333333333333e-06,
|
| 945 |
+
"loss": 0.0009,
|
| 946 |
+
"step": 3250
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 12.596153846153847,
|
| 950 |
+
"grad_norm": 0.035281531512737274,
|
| 951 |
+
"learning_rate": 3.837777777777778e-06,
|
| 952 |
+
"loss": 0.001,
|
| 953 |
+
"step": 3275
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 12.692307692307692,
|
| 957 |
+
"grad_norm": 0.036654986441135406,
|
| 958 |
+
"learning_rate": 3.782222222222223e-06,
|
| 959 |
+
"loss": 0.0009,
|
| 960 |
+
"step": 3300
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 12.788461538461538,
|
| 964 |
+
"grad_norm": 0.028989532962441444,
|
| 965 |
+
"learning_rate": 3.726666666666667e-06,
|
| 966 |
+
"loss": 0.0009,
|
| 967 |
+
"step": 3325
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 12.884615384615385,
|
| 971 |
+
"grad_norm": 0.033068809658288956,
|
| 972 |
+
"learning_rate": 3.6711111111111113e-06,
|
| 973 |
+
"loss": 0.0009,
|
| 974 |
+
"step": 3350
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 12.98076923076923,
|
| 978 |
+
"grad_norm": 0.035067297518253326,
|
| 979 |
+
"learning_rate": 3.615555555555556e-06,
|
| 980 |
+
"loss": 0.0009,
|
| 981 |
+
"step": 3375
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 13.076923076923077,
|
| 985 |
+
"grad_norm": 0.03311024233698845,
|
| 986 |
+
"learning_rate": 3.5600000000000002e-06,
|
| 987 |
+
"loss": 0.0008,
|
| 988 |
+
"step": 3400
|
| 989 |
+
},
|
| 990 |
+
{
|
| 991 |
+
"epoch": 13.173076923076923,
|
| 992 |
+
"grad_norm": 0.0331607460975647,
|
| 993 |
+
"learning_rate": 3.5044444444444447e-06,
|
| 994 |
+
"loss": 0.0009,
|
| 995 |
+
"step": 3425
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 13.26923076923077,
|
| 999 |
+
"grad_norm": 0.031973276287317276,
|
| 1000 |
+
"learning_rate": 3.4488888888888896e-06,
|
| 1001 |
+
"loss": 0.0015,
|
| 1002 |
+
"step": 3450
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 13.365384615384615,
|
| 1006 |
+
"grad_norm": 0.031224718317389488,
|
| 1007 |
+
"learning_rate": 3.3933333333333336e-06,
|
| 1008 |
+
"loss": 0.0008,
|
| 1009 |
+
"step": 3475
|
| 1010 |
+
},
|
| 1011 |
+
{
|
| 1012 |
+
"epoch": 13.461538461538462,
|
| 1013 |
+
"grad_norm": 0.030508123338222504,
|
| 1014 |
+
"learning_rate": 3.337777777777778e-06,
|
| 1015 |
+
"loss": 0.0008,
|
| 1016 |
+
"step": 3500
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 13.557692307692308,
|
| 1020 |
+
"grad_norm": 0.03208326920866966,
|
| 1021 |
+
"learning_rate": 3.282222222222223e-06,
|
| 1022 |
+
"loss": 0.0009,
|
| 1023 |
+
"step": 3525
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 13.653846153846153,
|
| 1027 |
+
"grad_norm": 0.02819848246872425,
|
| 1028 |
+
"learning_rate": 3.226666666666667e-06,
|
| 1029 |
+
"loss": 0.0008,
|
| 1030 |
+
"step": 3550
|
| 1031 |
+
},
|
| 1032 |
+
{
|
| 1033 |
+
"epoch": 13.75,
|
| 1034 |
+
"grad_norm": 0.02973184920847416,
|
| 1035 |
+
"learning_rate": 3.1711111111111114e-06,
|
| 1036 |
+
"loss": 0.0009,
|
| 1037 |
+
"step": 3575
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 13.846153846153847,
|
| 1041 |
+
"grad_norm": 0.026373526081442833,
|
| 1042 |
+
"learning_rate": 3.1155555555555555e-06,
|
| 1043 |
+
"loss": 0.0008,
|
| 1044 |
+
"step": 3600
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 13.942307692307692,
|
| 1048 |
+
"grad_norm": 0.030721429735422134,
|
| 1049 |
+
"learning_rate": 3.0600000000000003e-06,
|
| 1050 |
+
"loss": 0.0008,
|
| 1051 |
+
"step": 3625
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 14.038461538461538,
|
| 1055 |
+
"grad_norm": 0.027194401249289513,
|
| 1056 |
+
"learning_rate": 3.004444444444445e-06,
|
| 1057 |
+
"loss": 0.0008,
|
| 1058 |
+
"step": 3650
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 14.134615384615385,
|
| 1062 |
+
"grad_norm": 0.02467515505850315,
|
| 1063 |
+
"learning_rate": 2.948888888888889e-06,
|
| 1064 |
+
"loss": 0.0008,
|
| 1065 |
+
"step": 3675
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 14.23076923076923,
|
| 1069 |
+
"grad_norm": 0.02789357490837574,
|
| 1070 |
+
"learning_rate": 2.8933333333333337e-06,
|
| 1071 |
+
"loss": 0.0008,
|
| 1072 |
+
"step": 3700
|
| 1073 |
+
},
|
| 1074 |
+
{
|
| 1075 |
+
"epoch": 14.326923076923077,
|
| 1076 |
+
"grad_norm": 0.02929651364684105,
|
| 1077 |
+
"learning_rate": 2.837777777777778e-06,
|
| 1078 |
+
"loss": 0.0008,
|
| 1079 |
+
"step": 3725
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 14.423076923076923,
|
| 1083 |
+
"grad_norm": 0.02698478475213051,
|
| 1084 |
+
"learning_rate": 2.7822222222222222e-06,
|
| 1085 |
+
"loss": 0.0008,
|
| 1086 |
+
"step": 3750
|
| 1087 |
+
},
|
| 1088 |
+
{
|
| 1089 |
+
"epoch": 14.51923076923077,
|
| 1090 |
+
"grad_norm": 0.026549380272626877,
|
| 1091 |
+
"learning_rate": 2.726666666666667e-06,
|
| 1092 |
+
"loss": 0.0008,
|
| 1093 |
+
"step": 3775
|
| 1094 |
+
},
|
| 1095 |
+
{
|
| 1096 |
+
"epoch": 14.615384615384615,
|
| 1097 |
+
"grad_norm": 0.025898369029164314,
|
| 1098 |
+
"learning_rate": 2.6711111111111116e-06,
|
| 1099 |
+
"loss": 0.0008,
|
| 1100 |
+
"step": 3800
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 14.711538461538462,
|
| 1104 |
+
"grad_norm": 0.024109981954097748,
|
| 1105 |
+
"learning_rate": 2.6155555555555556e-06,
|
| 1106 |
+
"loss": 0.0008,
|
| 1107 |
+
"step": 3825
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 14.807692307692308,
|
| 1111 |
+
"grad_norm": 0.028975041583180428,
|
| 1112 |
+
"learning_rate": 2.56e-06,
|
| 1113 |
+
"loss": 0.0007,
|
| 1114 |
+
"step": 3850
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 14.903846153846153,
|
| 1118 |
+
"grad_norm": 0.029065443202853203,
|
| 1119 |
+
"learning_rate": 2.504444444444445e-06,
|
| 1120 |
+
"loss": 0.0008,
|
| 1121 |
+
"step": 3875
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 15.0,
|
| 1125 |
+
"grad_norm": 0.03024737536907196,
|
| 1126 |
+
"learning_rate": 2.448888888888889e-06,
|
| 1127 |
+
"loss": 0.0013,
|
| 1128 |
+
"step": 3900
|
| 1129 |
+
},
|
| 1130 |
+
{
|
| 1131 |
+
"epoch": 15.096153846153847,
|
| 1132 |
+
"grad_norm": 0.020746273919939995,
|
| 1133 |
+
"learning_rate": 2.3933333333333334e-06,
|
| 1134 |
+
"loss": 0.0007,
|
| 1135 |
+
"step": 3925
|
| 1136 |
+
},
|
| 1137 |
+
{
|
| 1138 |
+
"epoch": 15.192307692307692,
|
| 1139 |
+
"grad_norm": 0.022498076781630516,
|
| 1140 |
+
"learning_rate": 2.337777777777778e-06,
|
| 1141 |
+
"loss": 0.0007,
|
| 1142 |
+
"step": 3950
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 15.288461538461538,
|
| 1146 |
+
"grad_norm": 0.025430675595998764,
|
| 1147 |
+
"learning_rate": 2.2822222222222223e-06,
|
| 1148 |
+
"loss": 0.0007,
|
| 1149 |
+
"step": 3975
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 15.384615384615385,
|
| 1153 |
+
"grad_norm": 0.024518582969903946,
|
| 1154 |
+
"learning_rate": 2.226666666666667e-06,
|
| 1155 |
+
"loss": 0.0007,
|
| 1156 |
+
"step": 4000
|
| 1157 |
+
},
|
| 1158 |
+
{
|
| 1159 |
+
"epoch": 15.384615384615385,
|
| 1160 |
+
"eval_loss": 0.34604960680007935,
|
| 1161 |
+
"eval_runtime": 414.6061,
|
| 1162 |
+
"eval_samples_per_second": 8.782,
|
| 1163 |
+
"eval_steps_per_second": 0.55,
|
| 1164 |
+
"eval_wer": 0.1614460067752564,
|
| 1165 |
+
"step": 4000
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 15.48076923076923,
|
| 1169 |
+
"grad_norm": 0.028795845806598663,
|
| 1170 |
+
"learning_rate": 2.1711111111111113e-06,
|
| 1171 |
+
"loss": 0.0007,
|
| 1172 |
+
"step": 4025
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 15.576923076923077,
|
| 1176 |
+
"grad_norm": 0.02770942822098732,
|
| 1177 |
+
"learning_rate": 2.1155555555555557e-06,
|
| 1178 |
+
"loss": 0.0007,
|
| 1179 |
+
"step": 4050
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 15.673076923076923,
|
| 1183 |
+
"grad_norm": 0.026291629299521446,
|
| 1184 |
+
"learning_rate": 2.06e-06,
|
| 1185 |
+
"loss": 0.0008,
|
| 1186 |
+
"step": 4075
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 15.76923076923077,
|
| 1190 |
+
"grad_norm": 0.024974102154374123,
|
| 1191 |
+
"learning_rate": 2.0044444444444446e-06,
|
| 1192 |
+
"loss": 0.0007,
|
| 1193 |
+
"step": 4100
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 15.865384615384615,
|
| 1197 |
+
"grad_norm": 0.02725864201784134,
|
| 1198 |
+
"learning_rate": 1.948888888888889e-06,
|
| 1199 |
+
"loss": 0.0007,
|
| 1200 |
+
"step": 4125
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 15.961538461538462,
|
| 1204 |
+
"grad_norm": 0.027727024629712105,
|
| 1205 |
+
"learning_rate": 1.8933333333333333e-06,
|
| 1206 |
+
"loss": 0.0007,
|
| 1207 |
+
"step": 4150
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 16.057692307692307,
|
| 1211 |
+
"grad_norm": 0.023184822872281075,
|
| 1212 |
+
"learning_rate": 1.837777777777778e-06,
|
| 1213 |
+
"loss": 0.0007,
|
| 1214 |
+
"step": 4175
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 16.153846153846153,
|
| 1218 |
+
"grad_norm": 0.020365368574857712,
|
| 1219 |
+
"learning_rate": 1.7822222222222225e-06,
|
| 1220 |
+
"loss": 0.0007,
|
| 1221 |
+
"step": 4200
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 16.25,
|
| 1225 |
+
"grad_norm": 0.023292195051908493,
|
| 1226 |
+
"learning_rate": 1.7266666666666667e-06,
|
| 1227 |
+
"loss": 0.0006,
|
| 1228 |
+
"step": 4225
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 16.346153846153847,
|
| 1232 |
+
"grad_norm": 0.02580893039703369,
|
| 1233 |
+
"learning_rate": 1.6711111111111112e-06,
|
| 1234 |
+
"loss": 0.0006,
|
| 1235 |
+
"step": 4250
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 16.442307692307693,
|
| 1239 |
+
"grad_norm": 0.02440100722014904,
|
| 1240 |
+
"learning_rate": 1.6155555555555559e-06,
|
| 1241 |
+
"loss": 0.0007,
|
| 1242 |
+
"step": 4275
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 16.53846153846154,
|
| 1246 |
+
"grad_norm": 0.02583976835012436,
|
| 1247 |
+
"learning_rate": 1.56e-06,
|
| 1248 |
+
"loss": 0.0007,
|
| 1249 |
+
"step": 4300
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 16.634615384615383,
|
| 1253 |
+
"grad_norm": 0.024059804156422615,
|
| 1254 |
+
"learning_rate": 1.5044444444444446e-06,
|
| 1255 |
+
"loss": 0.0007,
|
| 1256 |
+
"step": 4325
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 16.73076923076923,
|
| 1260 |
+
"grad_norm": 0.023279106244444847,
|
| 1261 |
+
"learning_rate": 1.4488888888888892e-06,
|
| 1262 |
+
"loss": 0.0007,
|
| 1263 |
+
"step": 4350
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 16.826923076923077,
|
| 1267 |
+
"grad_norm": 0.022877462208271027,
|
| 1268 |
+
"learning_rate": 1.3933333333333335e-06,
|
| 1269 |
+
"loss": 0.0007,
|
| 1270 |
+
"step": 4375
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 16.923076923076923,
|
| 1274 |
+
"grad_norm": 0.02401687018573284,
|
| 1275 |
+
"learning_rate": 1.337777777777778e-06,
|
| 1276 |
+
"loss": 0.0006,
|
| 1277 |
+
"step": 4400
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 17.01923076923077,
|
| 1281 |
+
"grad_norm": 0.021333323791623116,
|
| 1282 |
+
"learning_rate": 1.2822222222222222e-06,
|
| 1283 |
+
"loss": 0.0006,
|
| 1284 |
+
"step": 4425
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 17.115384615384617,
|
| 1288 |
+
"grad_norm": 0.019655166193842888,
|
| 1289 |
+
"learning_rate": 1.2266666666666666e-06,
|
| 1290 |
+
"loss": 0.0006,
|
| 1291 |
+
"step": 4450
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 17.21153846153846,
|
| 1295 |
+
"grad_norm": 0.02106211706995964,
|
| 1296 |
+
"learning_rate": 1.171111111111111e-06,
|
| 1297 |
+
"loss": 0.0006,
|
| 1298 |
+
"step": 4475
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 17.307692307692307,
|
| 1302 |
+
"grad_norm": 0.022571563720703125,
|
| 1303 |
+
"learning_rate": 1.1155555555555558e-06,
|
| 1304 |
+
"loss": 0.0006,
|
| 1305 |
+
"step": 4500
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 17.403846153846153,
|
| 1309 |
+
"grad_norm": 0.023800240829586983,
|
| 1310 |
+
"learning_rate": 1.06e-06,
|
| 1311 |
+
"loss": 0.0006,
|
| 1312 |
+
"step": 4525
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 17.5,
|
| 1316 |
+
"grad_norm": 0.0228471290320158,
|
| 1317 |
+
"learning_rate": 1.0044444444444445e-06,
|
| 1318 |
+
"loss": 0.0006,
|
| 1319 |
+
"step": 4550
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 17.596153846153847,
|
| 1323 |
+
"grad_norm": 0.021810224279761314,
|
| 1324 |
+
"learning_rate": 9.488888888888889e-07,
|
| 1325 |
+
"loss": 0.0006,
|
| 1326 |
+
"step": 4575
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 17.692307692307693,
|
| 1330 |
+
"grad_norm": 0.02282741293311119,
|
| 1331 |
+
"learning_rate": 8.933333333333334e-07,
|
| 1332 |
+
"loss": 0.0008,
|
| 1333 |
+
"step": 4600
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 17.78846153846154,
|
| 1337 |
+
"grad_norm": 0.024272125214338303,
|
| 1338 |
+
"learning_rate": 8.37777777777778e-07,
|
| 1339 |
+
"loss": 0.0006,
|
| 1340 |
+
"step": 4625
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 17.884615384615383,
|
| 1344 |
+
"grad_norm": 0.026012342423200607,
|
| 1345 |
+
"learning_rate": 7.822222222222223e-07,
|
| 1346 |
+
"loss": 0.0006,
|
| 1347 |
+
"step": 4650
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 17.98076923076923,
|
| 1351 |
+
"grad_norm": 0.02298705466091633,
|
| 1352 |
+
"learning_rate": 7.266666666666668e-07,
|
| 1353 |
+
"loss": 0.0006,
|
| 1354 |
+
"step": 4675
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 18.076923076923077,
|
| 1358 |
+
"grad_norm": 0.021737568080425262,
|
| 1359 |
+
"learning_rate": 6.711111111111111e-07,
|
| 1360 |
+
"loss": 0.0006,
|
| 1361 |
+
"step": 4700
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 18.173076923076923,
|
| 1365 |
+
"grad_norm": 0.021294375881552696,
|
| 1366 |
+
"learning_rate": 6.155555555555556e-07,
|
| 1367 |
+
"loss": 0.0006,
|
| 1368 |
+
"step": 4725
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 18.26923076923077,
|
| 1372 |
+
"grad_norm": 0.0215039923787117,
|
| 1373 |
+
"learning_rate": 5.6e-07,
|
| 1374 |
+
"loss": 0.0006,
|
| 1375 |
+
"step": 4750
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 18.365384615384617,
|
| 1379 |
+
"grad_norm": 0.02031785622239113,
|
| 1380 |
+
"learning_rate": 5.044444444444445e-07,
|
| 1381 |
+
"loss": 0.0006,
|
| 1382 |
+
"step": 4775
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 18.46153846153846,
|
| 1386 |
+
"grad_norm": 0.02249339409172535,
|
| 1387 |
+
"learning_rate": 4.488888888888889e-07,
|
| 1388 |
+
"loss": 0.0006,
|
| 1389 |
+
"step": 4800
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 18.557692307692307,
|
| 1393 |
+
"grad_norm": 0.022298617288470268,
|
| 1394 |
+
"learning_rate": 3.9333333333333336e-07,
|
| 1395 |
+
"loss": 0.0006,
|
| 1396 |
+
"step": 4825
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 18.653846153846153,
|
| 1400 |
+
"grad_norm": 0.019332217052578926,
|
| 1401 |
+
"learning_rate": 3.3777777777777777e-07,
|
| 1402 |
+
"loss": 0.0006,
|
| 1403 |
+
"step": 4850
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 18.75,
|
| 1407 |
+
"grad_norm": 0.02518359012901783,
|
| 1408 |
+
"learning_rate": 2.822222222222222e-07,
|
| 1409 |
+
"loss": 0.0006,
|
| 1410 |
+
"step": 4875
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 18.846153846153847,
|
| 1414 |
+
"grad_norm": 0.022782141342759132,
|
| 1415 |
+
"learning_rate": 2.266666666666667e-07,
|
| 1416 |
+
"loss": 0.0006,
|
| 1417 |
+
"step": 4900
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 18.942307692307693,
|
| 1421 |
+
"grad_norm": 0.023741982877254486,
|
| 1422 |
+
"learning_rate": 1.7111111111111114e-07,
|
| 1423 |
+
"loss": 0.0006,
|
| 1424 |
+
"step": 4925
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 19.03846153846154,
|
| 1428 |
+
"grad_norm": 0.02083681896328926,
|
| 1429 |
+
"learning_rate": 1.1555555555555556e-07,
|
| 1430 |
+
"loss": 0.0006,
|
| 1431 |
+
"step": 4950
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 19.134615384615383,
|
| 1435 |
+
"grad_norm": 0.020423738285899162,
|
| 1436 |
+
"learning_rate": 6.000000000000001e-08,
|
| 1437 |
+
"loss": 0.0006,
|
| 1438 |
+
"step": 4975
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 19.23076923076923,
|
| 1442 |
+
"grad_norm": 0.0213633980602026,
|
| 1443 |
+
"learning_rate": 4.444444444444445e-09,
|
| 1444 |
+
"loss": 0.0006,
|
| 1445 |
+
"step": 5000
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 19.23076923076923,
|
| 1449 |
+
"eval_loss": 0.35056987404823303,
|
| 1450 |
+
"eval_runtime": 416.1932,
|
| 1451 |
+
"eval_samples_per_second": 8.748,
|
| 1452 |
+
"eval_steps_per_second": 0.548,
|
| 1453 |
+
"eval_wer": 0.16172444196946495,
|
| 1454 |
+
"step": 5000
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 19.23076923076923,
|
| 1458 |
+
"step": 5000,
|
| 1459 |
+
"total_flos": 4.616818092859392e+19,
|
| 1460 |
+
"train_loss": 0.05540717707388103,
|
| 1461 |
+
"train_runtime": 18195.7149,
|
| 1462 |
+
"train_samples_per_second": 8.793,
|
| 1463 |
+
"train_steps_per_second": 0.275
|
| 1464 |
+
}
|
| 1465 |
+
],
|
| 1466 |
+
"logging_steps": 25,
|
| 1467 |
+
"max_steps": 5000,
|
| 1468 |
+
"num_input_tokens_seen": 0,
|
| 1469 |
+
"num_train_epochs": 20,
|
| 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": 4.616818092859392e+19,
|
| 1484 |
+
"train_batch_size": 16,
|
| 1485 |
+
"trial_name": null,
|
| 1486 |
+
"trial_params": null
|
| 1487 |
+
}
|