End of training
Browse files- README.md +16 -4
- all_results.json +15 -0
- eval_results.json +9 -0
- train_results.json +9 -0
- trainer_state.json +1691 -0
README.md
CHANGED
|
@@ -4,11 +4,23 @@ license: apache-2.0
|
|
| 4 |
base_model: Leonel-Maia/fongbe-whisper-small
|
| 5 |
tags:
|
| 6 |
- generated_from_trainer
|
|
|
|
|
|
|
| 7 |
metrics:
|
| 8 |
- wer
|
| 9 |
model-index:
|
| 10 |
- name: whisper-small-transfer
|
| 11 |
-
results:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
---
|
| 13 |
|
| 14 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
@@ -16,10 +28,10 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 16 |
|
| 17 |
# whisper-small-transfer
|
| 18 |
|
| 19 |
-
This model is a fine-tuned version of [Leonel-Maia/fongbe-whisper-small](https://huggingface.co/Leonel-Maia/fongbe-whisper-small) on
|
| 20 |
It achieves the following results on the evaluation set:
|
| 21 |
-
- Loss: 0.
|
| 22 |
-
- Wer: 0.
|
| 23 |
|
| 24 |
## Model description
|
| 25 |
|
|
|
|
| 4 |
base_model: Leonel-Maia/fongbe-whisper-small
|
| 5 |
tags:
|
| 6 |
- generated_from_trainer
|
| 7 |
+
datasets:
|
| 8 |
+
- Leonel-Maia/ewe_dataset_splitted
|
| 9 |
metrics:
|
| 10 |
- wer
|
| 11 |
model-index:
|
| 12 |
- name: whisper-small-transfer
|
| 13 |
+
results:
|
| 14 |
+
- task:
|
| 15 |
+
name: Automatic Speech Recognition
|
| 16 |
+
type: automatic-speech-recognition
|
| 17 |
+
dataset:
|
| 18 |
+
name: Leonel-Maia/ewe_dataset_splitted
|
| 19 |
+
type: Leonel-Maia/ewe_dataset_splitted
|
| 20 |
+
metrics:
|
| 21 |
+
- name: Wer
|
| 22 |
+
type: wer
|
| 23 |
+
value: 0.21356341934578732
|
| 24 |
---
|
| 25 |
|
| 26 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
|
|
| 28 |
|
| 29 |
# whisper-small-transfer
|
| 30 |
|
| 31 |
+
This model is a fine-tuned version of [Leonel-Maia/fongbe-whisper-small](https://huggingface.co/Leonel-Maia/fongbe-whisper-small) on the Leonel-Maia/ewe_dataset_splitted dataset.
|
| 32 |
It achieves the following results on the evaluation set:
|
| 33 |
+
- Loss: 0.2392
|
| 34 |
+
- Wer: 0.2136
|
| 35 |
|
| 36 |
## Model description
|
| 37 |
|
all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 6.641930618401207,
|
| 3 |
+
"eval_loss": 0.23916485905647278,
|
| 4 |
+
"eval_runtime": 2446.3074,
|
| 5 |
+
"eval_samples": 3315,
|
| 6 |
+
"eval_samples_per_second": 1.355,
|
| 7 |
+
"eval_steps_per_second": 0.339,
|
| 8 |
+
"eval_wer": 0.21356341934578732,
|
| 9 |
+
"total_flos": 5.08256607043584e+19,
|
| 10 |
+
"train_loss": 0.25465128779411317,
|
| 11 |
+
"train_runtime": 113391.8356,
|
| 12 |
+
"train_samples": 26517,
|
| 13 |
+
"train_samples_per_second": 14.031,
|
| 14 |
+
"train_steps_per_second": 0.438
|
| 15 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 6.641930618401207,
|
| 3 |
+
"eval_loss": 0.23916485905647278,
|
| 4 |
+
"eval_runtime": 2446.3074,
|
| 5 |
+
"eval_samples": 3315,
|
| 6 |
+
"eval_samples_per_second": 1.355,
|
| 7 |
+
"eval_steps_per_second": 0.339,
|
| 8 |
+
"eval_wer": 0.21356341934578732
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 6.641930618401207,
|
| 3 |
+
"total_flos": 5.08256607043584e+19,
|
| 4 |
+
"train_loss": 0.25465128779411317,
|
| 5 |
+
"train_runtime": 113391.8356,
|
| 6 |
+
"train_samples": 26517,
|
| 7 |
+
"train_samples_per_second": 14.031,
|
| 8 |
+
"train_steps_per_second": 0.438
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1691 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 3000,
|
| 3 |
+
"best_metric": 0.23916485905647278,
|
| 4 |
+
"best_model_checkpoint": "./whisper-small-transfer/checkpoint-3000",
|
| 5 |
+
"epoch": 6.641930618401207,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 5500,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.030165912518853696,
|
| 14 |
+
"grad_norm": 26.211389541625977,
|
| 15 |
+
"learning_rate": 4.800000000000001e-07,
|
| 16 |
+
"loss": 4.3528,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.06033182503770739,
|
| 21 |
+
"grad_norm": 16.204137802124023,
|
| 22 |
+
"learning_rate": 9.800000000000001e-07,
|
| 23 |
+
"loss": 3.4167,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.09049773755656108,
|
| 28 |
+
"grad_norm": 10.840243339538574,
|
| 29 |
+
"learning_rate": 1.48e-06,
|
| 30 |
+
"loss": 2.3923,
|
| 31 |
+
"step": 75
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.12066365007541478,
|
| 35 |
+
"grad_norm": 7.455395698547363,
|
| 36 |
+
"learning_rate": 1.98e-06,
|
| 37 |
+
"loss": 1.7404,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.15082956259426847,
|
| 42 |
+
"grad_norm": 5.6613593101501465,
|
| 43 |
+
"learning_rate": 2.4800000000000004e-06,
|
| 44 |
+
"loss": 1.3135,
|
| 45 |
+
"step": 125
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.18099547511312217,
|
| 49 |
+
"grad_norm": 4.958397388458252,
|
| 50 |
+
"learning_rate": 2.9800000000000003e-06,
|
| 51 |
+
"loss": 1.0967,
|
| 52 |
+
"step": 150
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.21116138763197587,
|
| 56 |
+
"grad_norm": 4.388453006744385,
|
| 57 |
+
"learning_rate": 3.48e-06,
|
| 58 |
+
"loss": 0.9164,
|
| 59 |
+
"step": 175
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.24132730015082957,
|
| 63 |
+
"grad_norm": 4.023995399475098,
|
| 64 |
+
"learning_rate": 3.980000000000001e-06,
|
| 65 |
+
"loss": 0.8129,
|
| 66 |
+
"step": 200
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.27149321266968324,
|
| 70 |
+
"grad_norm": 3.9679815769195557,
|
| 71 |
+
"learning_rate": 4.48e-06,
|
| 72 |
+
"loss": 0.7049,
|
| 73 |
+
"step": 225
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.30165912518853694,
|
| 77 |
+
"grad_norm": 4.201240062713623,
|
| 78 |
+
"learning_rate": 4.980000000000001e-06,
|
| 79 |
+
"loss": 0.6658,
|
| 80 |
+
"step": 250
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.33182503770739064,
|
| 84 |
+
"grad_norm": 4.384641647338867,
|
| 85 |
+
"learning_rate": 5.480000000000001e-06,
|
| 86 |
+
"loss": 0.6323,
|
| 87 |
+
"step": 275
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.36199095022624433,
|
| 91 |
+
"grad_norm": 3.6530189514160156,
|
| 92 |
+
"learning_rate": 5.98e-06,
|
| 93 |
+
"loss": 0.5616,
|
| 94 |
+
"step": 300
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.39215686274509803,
|
| 98 |
+
"grad_norm": 4.075361728668213,
|
| 99 |
+
"learning_rate": 6.480000000000001e-06,
|
| 100 |
+
"loss": 0.5227,
|
| 101 |
+
"step": 325
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.42232277526395173,
|
| 105 |
+
"grad_norm": 4.275745868682861,
|
| 106 |
+
"learning_rate": 6.98e-06,
|
| 107 |
+
"loss": 0.5387,
|
| 108 |
+
"step": 350
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.45248868778280543,
|
| 112 |
+
"grad_norm": 3.494539737701416,
|
| 113 |
+
"learning_rate": 7.48e-06,
|
| 114 |
+
"loss": 0.5019,
|
| 115 |
+
"step": 375
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.48265460030165913,
|
| 119 |
+
"grad_norm": 3.5438411235809326,
|
| 120 |
+
"learning_rate": 7.980000000000002e-06,
|
| 121 |
+
"loss": 0.4708,
|
| 122 |
+
"step": 400
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.5128205128205128,
|
| 126 |
+
"grad_norm": 3.6554458141326904,
|
| 127 |
+
"learning_rate": 8.48e-06,
|
| 128 |
+
"loss": 0.441,
|
| 129 |
+
"step": 425
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.5429864253393665,
|
| 133 |
+
"grad_norm": 3.1188902854919434,
|
| 134 |
+
"learning_rate": 8.98e-06,
|
| 135 |
+
"loss": 0.4363,
|
| 136 |
+
"step": 450
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.5731523378582202,
|
| 140 |
+
"grad_norm": 3.156195878982544,
|
| 141 |
+
"learning_rate": 9.48e-06,
|
| 142 |
+
"loss": 0.4245,
|
| 143 |
+
"step": 475
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.6033182503770739,
|
| 147 |
+
"grad_norm": 3.006938934326172,
|
| 148 |
+
"learning_rate": 9.980000000000001e-06,
|
| 149 |
+
"loss": 0.3978,
|
| 150 |
+
"step": 500
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.6033182503770739,
|
| 154 |
+
"eval_loss": 0.3847188353538513,
|
| 155 |
+
"eval_runtime": 2461.5392,
|
| 156 |
+
"eval_samples_per_second": 1.347,
|
| 157 |
+
"eval_steps_per_second": 0.337,
|
| 158 |
+
"eval_wer": 0.34837741876301787,
|
| 159 |
+
"step": 500
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.6334841628959276,
|
| 163 |
+
"grad_norm": 3.300306797027588,
|
| 164 |
+
"learning_rate": 9.99511996746645e-06,
|
| 165 |
+
"loss": 0.4099,
|
| 166 |
+
"step": 525
|
| 167 |
+
},
|
| 168 |
+
{
|
| 169 |
+
"epoch": 0.6636500754147813,
|
| 170 |
+
"grad_norm": 2.924450635910034,
|
| 171 |
+
"learning_rate": 9.990036600244002e-06,
|
| 172 |
+
"loss": 0.3736,
|
| 173 |
+
"step": 550
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 0.693815987933635,
|
| 177 |
+
"grad_norm": 3.143441677093506,
|
| 178 |
+
"learning_rate": 9.984953233021555e-06,
|
| 179 |
+
"loss": 0.3785,
|
| 180 |
+
"step": 575
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 0.7239819004524887,
|
| 184 |
+
"grad_norm": 3.0799312591552734,
|
| 185 |
+
"learning_rate": 9.979869865799107e-06,
|
| 186 |
+
"loss": 0.3559,
|
| 187 |
+
"step": 600
|
| 188 |
+
},
|
| 189 |
+
{
|
| 190 |
+
"epoch": 0.7541478129713424,
|
| 191 |
+
"grad_norm": 2.933881998062134,
|
| 192 |
+
"learning_rate": 9.974786498576659e-06,
|
| 193 |
+
"loss": 0.3554,
|
| 194 |
+
"step": 625
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.7843137254901961,
|
| 198 |
+
"grad_norm": 2.6328113079071045,
|
| 199 |
+
"learning_rate": 9.96970313135421e-06,
|
| 200 |
+
"loss": 0.3393,
|
| 201 |
+
"step": 650
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.8144796380090498,
|
| 205 |
+
"grad_norm": 2.767172336578369,
|
| 206 |
+
"learning_rate": 9.964619764131762e-06,
|
| 207 |
+
"loss": 0.3539,
|
| 208 |
+
"step": 675
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 0.8446455505279035,
|
| 212 |
+
"grad_norm": 3.040672779083252,
|
| 213 |
+
"learning_rate": 9.959536396909314e-06,
|
| 214 |
+
"loss": 0.3347,
|
| 215 |
+
"step": 700
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 0.8748114630467572,
|
| 219 |
+
"grad_norm": 2.708042860031128,
|
| 220 |
+
"learning_rate": 9.954453029686866e-06,
|
| 221 |
+
"loss": 0.3208,
|
| 222 |
+
"step": 725
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 0.9049773755656109,
|
| 226 |
+
"grad_norm": 2.599907875061035,
|
| 227 |
+
"learning_rate": 9.949369662464417e-06,
|
| 228 |
+
"loss": 0.3314,
|
| 229 |
+
"step": 750
|
| 230 |
+
},
|
| 231 |
+
{
|
| 232 |
+
"epoch": 0.9351432880844646,
|
| 233 |
+
"grad_norm": 2.7246906757354736,
|
| 234 |
+
"learning_rate": 9.944286295241969e-06,
|
| 235 |
+
"loss": 0.3212,
|
| 236 |
+
"step": 775
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.9653092006033183,
|
| 240 |
+
"grad_norm": 2.789820909500122,
|
| 241 |
+
"learning_rate": 9.93920292801952e-06,
|
| 242 |
+
"loss": 0.3197,
|
| 243 |
+
"step": 800
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.995475113122172,
|
| 247 |
+
"grad_norm": 2.7562570571899414,
|
| 248 |
+
"learning_rate": 9.934119560797073e-06,
|
| 249 |
+
"loss": 0.3075,
|
| 250 |
+
"step": 825
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 1.0265460030165912,
|
| 254 |
+
"grad_norm": 2.808879852294922,
|
| 255 |
+
"learning_rate": 9.929036193574624e-06,
|
| 256 |
+
"loss": 0.2848,
|
| 257 |
+
"step": 850
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 1.056711915535445,
|
| 261 |
+
"grad_norm": 2.3943305015563965,
|
| 262 |
+
"learning_rate": 9.923952826352176e-06,
|
| 263 |
+
"loss": 0.2741,
|
| 264 |
+
"step": 875
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 1.0868778280542986,
|
| 268 |
+
"grad_norm": 2.515653133392334,
|
| 269 |
+
"learning_rate": 9.91886945912973e-06,
|
| 270 |
+
"loss": 0.2812,
|
| 271 |
+
"step": 900
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"epoch": 1.1170437405731524,
|
| 275 |
+
"grad_norm": 2.7121636867523193,
|
| 276 |
+
"learning_rate": 9.913786091907281e-06,
|
| 277 |
+
"loss": 0.2805,
|
| 278 |
+
"step": 925
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 1.147209653092006,
|
| 282 |
+
"grad_norm": 2.8646414279937744,
|
| 283 |
+
"learning_rate": 9.908702724684833e-06,
|
| 284 |
+
"loss": 0.2726,
|
| 285 |
+
"step": 950
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 1.1773755656108598,
|
| 289 |
+
"grad_norm": 2.6251399517059326,
|
| 290 |
+
"learning_rate": 9.903619357462384e-06,
|
| 291 |
+
"loss": 0.2747,
|
| 292 |
+
"step": 975
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 1.2075414781297134,
|
| 296 |
+
"grad_norm": 2.397496461868286,
|
| 297 |
+
"learning_rate": 9.898535990239936e-06,
|
| 298 |
+
"loss": 0.249,
|
| 299 |
+
"step": 1000
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 1.2075414781297134,
|
| 303 |
+
"eval_loss": 0.28903013467788696,
|
| 304 |
+
"eval_runtime": 2466.5989,
|
| 305 |
+
"eval_samples_per_second": 1.344,
|
| 306 |
+
"eval_steps_per_second": 0.336,
|
| 307 |
+
"eval_wer": 0.25850141915153085,
|
| 308 |
+
"step": 1000
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 1.2377073906485672,
|
| 312 |
+
"grad_norm": 2.6788296699523926,
|
| 313 |
+
"learning_rate": 9.893452623017488e-06,
|
| 314 |
+
"loss": 0.2737,
|
| 315 |
+
"step": 1025
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 1.2678733031674208,
|
| 319 |
+
"grad_norm": 2.5809574127197266,
|
| 320 |
+
"learning_rate": 9.88836925579504e-06,
|
| 321 |
+
"loss": 0.2609,
|
| 322 |
+
"step": 1050
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 1.2980392156862746,
|
| 326 |
+
"grad_norm": 2.419163227081299,
|
| 327 |
+
"learning_rate": 9.883285888572591e-06,
|
| 328 |
+
"loss": 0.2673,
|
| 329 |
+
"step": 1075
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 1.3282051282051281,
|
| 333 |
+
"grad_norm": 2.5561363697052,
|
| 334 |
+
"learning_rate": 9.878202521350143e-06,
|
| 335 |
+
"loss": 0.2535,
|
| 336 |
+
"step": 1100
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 1.358371040723982,
|
| 340 |
+
"grad_norm": 2.6312174797058105,
|
| 341 |
+
"learning_rate": 9.873119154127695e-06,
|
| 342 |
+
"loss": 0.2693,
|
| 343 |
+
"step": 1125
|
| 344 |
+
},
|
| 345 |
+
{
|
| 346 |
+
"epoch": 1.3885369532428355,
|
| 347 |
+
"grad_norm": 2.4085021018981934,
|
| 348 |
+
"learning_rate": 9.868035786905246e-06,
|
| 349 |
+
"loss": 0.2598,
|
| 350 |
+
"step": 1150
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 1.4187028657616894,
|
| 354 |
+
"grad_norm": 2.352027654647827,
|
| 355 |
+
"learning_rate": 9.862952419682798e-06,
|
| 356 |
+
"loss": 0.2713,
|
| 357 |
+
"step": 1175
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 1.448868778280543,
|
| 361 |
+
"grad_norm": 2.478139877319336,
|
| 362 |
+
"learning_rate": 9.85786905246035e-06,
|
| 363 |
+
"loss": 0.2783,
|
| 364 |
+
"step": 1200
|
| 365 |
+
},
|
| 366 |
+
{
|
| 367 |
+
"epoch": 1.4790346907993968,
|
| 368 |
+
"grad_norm": 2.542982816696167,
|
| 369 |
+
"learning_rate": 9.852785685237901e-06,
|
| 370 |
+
"loss": 0.2445,
|
| 371 |
+
"step": 1225
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 1.5092006033182503,
|
| 375 |
+
"grad_norm": 2.4919915199279785,
|
| 376 |
+
"learning_rate": 9.847702318015455e-06,
|
| 377 |
+
"loss": 0.2654,
|
| 378 |
+
"step": 1250
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 1.539366515837104,
|
| 382 |
+
"grad_norm": 2.59543776512146,
|
| 383 |
+
"learning_rate": 9.842618950793007e-06,
|
| 384 |
+
"loss": 0.2477,
|
| 385 |
+
"step": 1275
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 1.5695324283559577,
|
| 389 |
+
"grad_norm": 2.4258153438568115,
|
| 390 |
+
"learning_rate": 9.837535583570558e-06,
|
| 391 |
+
"loss": 0.2544,
|
| 392 |
+
"step": 1300
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 1.5996983408748116,
|
| 396 |
+
"grad_norm": 2.4069764614105225,
|
| 397 |
+
"learning_rate": 9.83245221634811e-06,
|
| 398 |
+
"loss": 0.2383,
|
| 399 |
+
"step": 1325
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 1.6298642533936651,
|
| 403 |
+
"grad_norm": 2.3214974403381348,
|
| 404 |
+
"learning_rate": 9.827368849125662e-06,
|
| 405 |
+
"loss": 0.2652,
|
| 406 |
+
"step": 1350
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 1.6600301659125187,
|
| 410 |
+
"grad_norm": 2.626494884490967,
|
| 411 |
+
"learning_rate": 9.822285481903213e-06,
|
| 412 |
+
"loss": 0.2501,
|
| 413 |
+
"step": 1375
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 1.6901960784313725,
|
| 417 |
+
"grad_norm": 2.589961290359497,
|
| 418 |
+
"learning_rate": 9.817202114680765e-06,
|
| 419 |
+
"loss": 0.2465,
|
| 420 |
+
"step": 1400
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 1.7203619909502263,
|
| 424 |
+
"grad_norm": 2.246035099029541,
|
| 425 |
+
"learning_rate": 9.812118747458317e-06,
|
| 426 |
+
"loss": 0.2399,
|
| 427 |
+
"step": 1425
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 1.75052790346908,
|
| 431 |
+
"grad_norm": 2.430635452270508,
|
| 432 |
+
"learning_rate": 9.807035380235868e-06,
|
| 433 |
+
"loss": 0.2421,
|
| 434 |
+
"step": 1450
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 1.7806938159879335,
|
| 438 |
+
"grad_norm": 2.1688945293426514,
|
| 439 |
+
"learning_rate": 9.80195201301342e-06,
|
| 440 |
+
"loss": 0.2433,
|
| 441 |
+
"step": 1475
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 1.8108597285067873,
|
| 445 |
+
"grad_norm": 2.5152928829193115,
|
| 446 |
+
"learning_rate": 9.796868645790972e-06,
|
| 447 |
+
"loss": 0.2481,
|
| 448 |
+
"step": 1500
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"epoch": 1.8108597285067873,
|
| 452 |
+
"eval_loss": 0.2585034966468811,
|
| 453 |
+
"eval_runtime": 2401.9254,
|
| 454 |
+
"eval_samples_per_second": 1.38,
|
| 455 |
+
"eval_steps_per_second": 0.345,
|
| 456 |
+
"eval_wer": 0.23873042596129979,
|
| 457 |
+
"step": 1500
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 1.8410256410256411,
|
| 461 |
+
"grad_norm": 2.3025524616241455,
|
| 462 |
+
"learning_rate": 9.791785278568524e-06,
|
| 463 |
+
"loss": 0.2371,
|
| 464 |
+
"step": 1525
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 1.8711915535444947,
|
| 468 |
+
"grad_norm": 2.131627082824707,
|
| 469 |
+
"learning_rate": 9.786701911346075e-06,
|
| 470 |
+
"loss": 0.2294,
|
| 471 |
+
"step": 1550
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 1.9013574660633483,
|
| 475 |
+
"grad_norm": 2.1070120334625244,
|
| 476 |
+
"learning_rate": 9.781618544123627e-06,
|
| 477 |
+
"loss": 0.2593,
|
| 478 |
+
"step": 1575
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 1.9315233785822021,
|
| 482 |
+
"grad_norm": 2.209496021270752,
|
| 483 |
+
"learning_rate": 9.77653517690118e-06,
|
| 484 |
+
"loss": 0.2415,
|
| 485 |
+
"step": 1600
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 1.961689291101056,
|
| 489 |
+
"grad_norm": 2.3032071590423584,
|
| 490 |
+
"learning_rate": 9.771451809678732e-06,
|
| 491 |
+
"loss": 0.2419,
|
| 492 |
+
"step": 1625
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 1.9918552036199095,
|
| 496 |
+
"grad_norm": 2.4967117309570312,
|
| 497 |
+
"learning_rate": 9.766368442456284e-06,
|
| 498 |
+
"loss": 0.2474,
|
| 499 |
+
"step": 1650
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 2.0229260935143287,
|
| 503 |
+
"grad_norm": 2.2161521911621094,
|
| 504 |
+
"learning_rate": 9.761285075233836e-06,
|
| 505 |
+
"loss": 0.2136,
|
| 506 |
+
"step": 1675
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 2.0530920060331823,
|
| 510 |
+
"grad_norm": 2.219045400619507,
|
| 511 |
+
"learning_rate": 9.756201708011387e-06,
|
| 512 |
+
"loss": 0.1956,
|
| 513 |
+
"step": 1700
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 2.0832579185520363,
|
| 517 |
+
"grad_norm": 2.001629590988159,
|
| 518 |
+
"learning_rate": 9.75111834078894e-06,
|
| 519 |
+
"loss": 0.1886,
|
| 520 |
+
"step": 1725
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 2.11342383107089,
|
| 524 |
+
"grad_norm": 2.072310209274292,
|
| 525 |
+
"learning_rate": 9.746034973566492e-06,
|
| 526 |
+
"loss": 0.2012,
|
| 527 |
+
"step": 1750
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 2.1435897435897435,
|
| 531 |
+
"grad_norm": 2.7330238819122314,
|
| 532 |
+
"learning_rate": 9.740951606344044e-06,
|
| 533 |
+
"loss": 0.2154,
|
| 534 |
+
"step": 1775
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 2.173755656108597,
|
| 538 |
+
"grad_norm": 2.282186985015869,
|
| 539 |
+
"learning_rate": 9.735868239121596e-06,
|
| 540 |
+
"loss": 0.2028,
|
| 541 |
+
"step": 1800
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 2.203921568627451,
|
| 545 |
+
"grad_norm": 2.1114888191223145,
|
| 546 |
+
"learning_rate": 9.730784871899147e-06,
|
| 547 |
+
"loss": 0.2026,
|
| 548 |
+
"step": 1825
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 2.2340874811463047,
|
| 552 |
+
"grad_norm": 2.320906400680542,
|
| 553 |
+
"learning_rate": 9.725701504676699e-06,
|
| 554 |
+
"loss": 0.211,
|
| 555 |
+
"step": 1850
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 2.2642533936651583,
|
| 559 |
+
"grad_norm": 2.435915231704712,
|
| 560 |
+
"learning_rate": 9.72061813745425e-06,
|
| 561 |
+
"loss": 0.1983,
|
| 562 |
+
"step": 1875
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 2.294419306184012,
|
| 566 |
+
"grad_norm": 2.0894196033477783,
|
| 567 |
+
"learning_rate": 9.715534770231803e-06,
|
| 568 |
+
"loss": 0.1852,
|
| 569 |
+
"step": 1900
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 2.324585218702866,
|
| 573 |
+
"grad_norm": 1.7594900131225586,
|
| 574 |
+
"learning_rate": 9.710451403009354e-06,
|
| 575 |
+
"loss": 0.1864,
|
| 576 |
+
"step": 1925
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 2.3547511312217195,
|
| 580 |
+
"grad_norm": 2.0977182388305664,
|
| 581 |
+
"learning_rate": 9.705368035786906e-06,
|
| 582 |
+
"loss": 0.1942,
|
| 583 |
+
"step": 1950
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 2.384917043740573,
|
| 587 |
+
"grad_norm": 2.048466205596924,
|
| 588 |
+
"learning_rate": 9.700284668564458e-06,
|
| 589 |
+
"loss": 0.2083,
|
| 590 |
+
"step": 1975
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 2.4150829562594267,
|
| 594 |
+
"grad_norm": 2.275794506072998,
|
| 595 |
+
"learning_rate": 9.69520130134201e-06,
|
| 596 |
+
"loss": 0.1996,
|
| 597 |
+
"step": 2000
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 2.4150829562594267,
|
| 601 |
+
"eval_loss": 0.24703706800937653,
|
| 602 |
+
"eval_runtime": 2360.2803,
|
| 603 |
+
"eval_samples_per_second": 1.404,
|
| 604 |
+
"eval_steps_per_second": 0.351,
|
| 605 |
+
"eval_wer": 0.2232870355381444,
|
| 606 |
+
"step": 2000
|
| 607 |
+
},
|
| 608 |
+
{
|
| 609 |
+
"epoch": 2.4452488687782807,
|
| 610 |
+
"grad_norm": 2.1452808380126953,
|
| 611 |
+
"learning_rate": 9.690117934119561e-06,
|
| 612 |
+
"loss": 0.1945,
|
| 613 |
+
"step": 2025
|
| 614 |
+
},
|
| 615 |
+
{
|
| 616 |
+
"epoch": 2.4754147812971343,
|
| 617 |
+
"grad_norm": 2.2425966262817383,
|
| 618 |
+
"learning_rate": 9.685034566897113e-06,
|
| 619 |
+
"loss": 0.1885,
|
| 620 |
+
"step": 2050
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 2.505580693815988,
|
| 624 |
+
"grad_norm": 2.123422861099243,
|
| 625 |
+
"learning_rate": 9.679951199674666e-06,
|
| 626 |
+
"loss": 0.1817,
|
| 627 |
+
"step": 2075
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 2.5357466063348415,
|
| 631 |
+
"grad_norm": 1.747390866279602,
|
| 632 |
+
"learning_rate": 9.674867832452218e-06,
|
| 633 |
+
"loss": 0.1999,
|
| 634 |
+
"step": 2100
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 2.565912518853695,
|
| 638 |
+
"grad_norm": 1.9593695402145386,
|
| 639 |
+
"learning_rate": 9.66978446522977e-06,
|
| 640 |
+
"loss": 0.188,
|
| 641 |
+
"step": 2125
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 2.596078431372549,
|
| 645 |
+
"grad_norm": 2.0385262966156006,
|
| 646 |
+
"learning_rate": 9.664701098007321e-06,
|
| 647 |
+
"loss": 0.2004,
|
| 648 |
+
"step": 2150
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 2.6262443438914027,
|
| 652 |
+
"grad_norm": 1.9144082069396973,
|
| 653 |
+
"learning_rate": 9.659617730784873e-06,
|
| 654 |
+
"loss": 0.2062,
|
| 655 |
+
"step": 2175
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 2.6564102564102563,
|
| 659 |
+
"grad_norm": 2.052720308303833,
|
| 660 |
+
"learning_rate": 9.654534363562425e-06,
|
| 661 |
+
"loss": 0.1839,
|
| 662 |
+
"step": 2200
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 2.6865761689291103,
|
| 666 |
+
"grad_norm": 1.9951667785644531,
|
| 667 |
+
"learning_rate": 9.649450996339976e-06,
|
| 668 |
+
"loss": 0.1798,
|
| 669 |
+
"step": 2225
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 2.716742081447964,
|
| 673 |
+
"grad_norm": 2.5916664600372314,
|
| 674 |
+
"learning_rate": 9.644367629117528e-06,
|
| 675 |
+
"loss": 0.2058,
|
| 676 |
+
"step": 2250
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 2.7469079939668175,
|
| 680 |
+
"grad_norm": 2.3864409923553467,
|
| 681 |
+
"learning_rate": 9.63928426189508e-06,
|
| 682 |
+
"loss": 0.1993,
|
| 683 |
+
"step": 2275
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 2.777073906485671,
|
| 687 |
+
"grad_norm": 2.1544036865234375,
|
| 688 |
+
"learning_rate": 9.634200894672631e-06,
|
| 689 |
+
"loss": 0.1916,
|
| 690 |
+
"step": 2300
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 2.8072398190045247,
|
| 694 |
+
"grad_norm": 2.513857364654541,
|
| 695 |
+
"learning_rate": 9.629117527450183e-06,
|
| 696 |
+
"loss": 0.1979,
|
| 697 |
+
"step": 2325
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 2.8374057315233787,
|
| 701 |
+
"grad_norm": 2.702158212661743,
|
| 702 |
+
"learning_rate": 9.624034160227735e-06,
|
| 703 |
+
"loss": 0.1976,
|
| 704 |
+
"step": 2350
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 2.8675716440422323,
|
| 708 |
+
"grad_norm": 1.9018394947052002,
|
| 709 |
+
"learning_rate": 9.618950793005287e-06,
|
| 710 |
+
"loss": 0.2007,
|
| 711 |
+
"step": 2375
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 2.897737556561086,
|
| 715 |
+
"grad_norm": 2.1460797786712646,
|
| 716 |
+
"learning_rate": 9.613867425782838e-06,
|
| 717 |
+
"loss": 0.191,
|
| 718 |
+
"step": 2400
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 2.92790346907994,
|
| 722 |
+
"grad_norm": 2.0922977924346924,
|
| 723 |
+
"learning_rate": 9.608784058560392e-06,
|
| 724 |
+
"loss": 0.1956,
|
| 725 |
+
"step": 2425
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 2.9580693815987935,
|
| 729 |
+
"grad_norm": 2.323761463165283,
|
| 730 |
+
"learning_rate": 9.603700691337943e-06,
|
| 731 |
+
"loss": 0.1927,
|
| 732 |
+
"step": 2450
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 2.988235294117647,
|
| 736 |
+
"grad_norm": 2.0030786991119385,
|
| 737 |
+
"learning_rate": 9.598617324115495e-06,
|
| 738 |
+
"loss": 0.1923,
|
| 739 |
+
"step": 2475
|
| 740 |
+
},
|
| 741 |
+
{
|
| 742 |
+
"epoch": 3.0193061840120663,
|
| 743 |
+
"grad_norm": 1.9809656143188477,
|
| 744 |
+
"learning_rate": 9.593533956893047e-06,
|
| 745 |
+
"loss": 0.1669,
|
| 746 |
+
"step": 2500
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 3.0193061840120663,
|
| 750 |
+
"eval_loss": 0.241033673286438,
|
| 751 |
+
"eval_runtime": 2297.6931,
|
| 752 |
+
"eval_samples_per_second": 1.443,
|
| 753 |
+
"eval_steps_per_second": 0.361,
|
| 754 |
+
"eval_wer": 0.2156894486353482,
|
| 755 |
+
"step": 2500
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 3.04947209653092,
|
| 759 |
+
"grad_norm": 1.9472345113754272,
|
| 760 |
+
"learning_rate": 9.588450589670599e-06,
|
| 761 |
+
"loss": 0.1578,
|
| 762 |
+
"step": 2525
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 3.079638009049774,
|
| 766 |
+
"grad_norm": 2.215965509414673,
|
| 767 |
+
"learning_rate": 9.58336722244815e-06,
|
| 768 |
+
"loss": 0.1564,
|
| 769 |
+
"step": 2550
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 3.1098039215686275,
|
| 773 |
+
"grad_norm": 1.8719356060028076,
|
| 774 |
+
"learning_rate": 9.578283855225702e-06,
|
| 775 |
+
"loss": 0.1646,
|
| 776 |
+
"step": 2575
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 3.139969834087481,
|
| 780 |
+
"grad_norm": 1.7623282670974731,
|
| 781 |
+
"learning_rate": 9.573200488003254e-06,
|
| 782 |
+
"loss": 0.149,
|
| 783 |
+
"step": 2600
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 3.1701357466063347,
|
| 787 |
+
"grad_norm": 1.8403270244598389,
|
| 788 |
+
"learning_rate": 9.568117120780805e-06,
|
| 789 |
+
"loss": 0.1543,
|
| 790 |
+
"step": 2625
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 3.2003016591251887,
|
| 794 |
+
"grad_norm": 2.092792510986328,
|
| 795 |
+
"learning_rate": 9.563033753558359e-06,
|
| 796 |
+
"loss": 0.1626,
|
| 797 |
+
"step": 2650
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 3.2304675716440423,
|
| 801 |
+
"grad_norm": 1.7507753372192383,
|
| 802 |
+
"learning_rate": 9.55795038633591e-06,
|
| 803 |
+
"loss": 0.1516,
|
| 804 |
+
"step": 2675
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 3.260633484162896,
|
| 808 |
+
"grad_norm": 1.7713559865951538,
|
| 809 |
+
"learning_rate": 9.552867019113462e-06,
|
| 810 |
+
"loss": 0.1565,
|
| 811 |
+
"step": 2700
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 3.2907993966817495,
|
| 815 |
+
"grad_norm": 2.3640658855438232,
|
| 816 |
+
"learning_rate": 9.547783651891014e-06,
|
| 817 |
+
"loss": 0.1583,
|
| 818 |
+
"step": 2725
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 3.3209653092006035,
|
| 822 |
+
"grad_norm": 1.9633855819702148,
|
| 823 |
+
"learning_rate": 9.542700284668566e-06,
|
| 824 |
+
"loss": 0.1612,
|
| 825 |
+
"step": 2750
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 3.351131221719457,
|
| 829 |
+
"grad_norm": 1.8450348377227783,
|
| 830 |
+
"learning_rate": 9.537616917446117e-06,
|
| 831 |
+
"loss": 0.1573,
|
| 832 |
+
"step": 2775
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 3.3812971342383107,
|
| 836 |
+
"grad_norm": 1.913599967956543,
|
| 837 |
+
"learning_rate": 9.532533550223669e-06,
|
| 838 |
+
"loss": 0.1623,
|
| 839 |
+
"step": 2800
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 3.4114630467571643,
|
| 843 |
+
"grad_norm": 2.1844587326049805,
|
| 844 |
+
"learning_rate": 9.52745018300122e-06,
|
| 845 |
+
"loss": 0.1562,
|
| 846 |
+
"step": 2825
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 3.4416289592760183,
|
| 850 |
+
"grad_norm": 1.827317237854004,
|
| 851 |
+
"learning_rate": 9.522366815778772e-06,
|
| 852 |
+
"loss": 0.1598,
|
| 853 |
+
"step": 2850
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 3.471794871794872,
|
| 857 |
+
"grad_norm": 1.7931983470916748,
|
| 858 |
+
"learning_rate": 9.517283448556324e-06,
|
| 859 |
+
"loss": 0.1557,
|
| 860 |
+
"step": 2875
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 3.5019607843137255,
|
| 864 |
+
"grad_norm": 1.85126793384552,
|
| 865 |
+
"learning_rate": 9.512200081333877e-06,
|
| 866 |
+
"loss": 0.1588,
|
| 867 |
+
"step": 2900
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 3.532126696832579,
|
| 871 |
+
"grad_norm": 2.0126168727874756,
|
| 872 |
+
"learning_rate": 9.50711671411143e-06,
|
| 873 |
+
"loss": 0.161,
|
| 874 |
+
"step": 2925
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 3.5622926093514327,
|
| 878 |
+
"grad_norm": 1.982439637184143,
|
| 879 |
+
"learning_rate": 9.502033346888981e-06,
|
| 880 |
+
"loss": 0.1477,
|
| 881 |
+
"step": 2950
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 3.5924585218702867,
|
| 885 |
+
"grad_norm": 2.315864086151123,
|
| 886 |
+
"learning_rate": 9.496949979666533e-06,
|
| 887 |
+
"loss": 0.157,
|
| 888 |
+
"step": 2975
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 3.6226244343891403,
|
| 892 |
+
"grad_norm": 1.939032793045044,
|
| 893 |
+
"learning_rate": 9.491866612444084e-06,
|
| 894 |
+
"loss": 0.1535,
|
| 895 |
+
"step": 3000
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 3.6226244343891403,
|
| 899 |
+
"eval_loss": 0.23916485905647278,
|
| 900 |
+
"eval_runtime": 2282.7909,
|
| 901 |
+
"eval_samples_per_second": 1.452,
|
| 902 |
+
"eval_steps_per_second": 0.363,
|
| 903 |
+
"eval_wer": 0.21356341934578732,
|
| 904 |
+
"step": 3000
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 3.652790346907994,
|
| 908 |
+
"grad_norm": 1.8340437412261963,
|
| 909 |
+
"learning_rate": 9.486783245221636e-06,
|
| 910 |
+
"loss": 0.1555,
|
| 911 |
+
"step": 3025
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 3.682956259426848,
|
| 915 |
+
"grad_norm": 2.115091562271118,
|
| 916 |
+
"learning_rate": 9.481699877999188e-06,
|
| 917 |
+
"loss": 0.1589,
|
| 918 |
+
"step": 3050
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 3.7131221719457015,
|
| 922 |
+
"grad_norm": 2.074758768081665,
|
| 923 |
+
"learning_rate": 9.47661651077674e-06,
|
| 924 |
+
"loss": 0.1506,
|
| 925 |
+
"step": 3075
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 3.743288084464555,
|
| 929 |
+
"grad_norm": 2.0023365020751953,
|
| 930 |
+
"learning_rate": 9.471533143554291e-06,
|
| 931 |
+
"loss": 0.1537,
|
| 932 |
+
"step": 3100
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 3.7734539969834087,
|
| 936 |
+
"grad_norm": 1.883928656578064,
|
| 937 |
+
"learning_rate": 9.466449776331843e-06,
|
| 938 |
+
"loss": 0.1598,
|
| 939 |
+
"step": 3125
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 3.8036199095022623,
|
| 943 |
+
"grad_norm": 2.637164354324341,
|
| 944 |
+
"learning_rate": 9.461366409109394e-06,
|
| 945 |
+
"loss": 0.1725,
|
| 946 |
+
"step": 3150
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 3.8337858220211163,
|
| 950 |
+
"grad_norm": 2.0999221801757812,
|
| 951 |
+
"learning_rate": 9.456283041886946e-06,
|
| 952 |
+
"loss": 0.1591,
|
| 953 |
+
"step": 3175
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 3.86395173453997,
|
| 957 |
+
"grad_norm": 2.033339500427246,
|
| 958 |
+
"learning_rate": 9.451199674664498e-06,
|
| 959 |
+
"loss": 0.1622,
|
| 960 |
+
"step": 3200
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 3.8941176470588235,
|
| 964 |
+
"grad_norm": 2.1740732192993164,
|
| 965 |
+
"learning_rate": 9.44611630744205e-06,
|
| 966 |
+
"loss": 0.1606,
|
| 967 |
+
"step": 3225
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 3.9242835595776775,
|
| 971 |
+
"grad_norm": 2.3770601749420166,
|
| 972 |
+
"learning_rate": 9.441032940219603e-06,
|
| 973 |
+
"loss": 0.1593,
|
| 974 |
+
"step": 3250
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 3.954449472096531,
|
| 978 |
+
"grad_norm": 1.9333163499832153,
|
| 979 |
+
"learning_rate": 9.435949572997155e-06,
|
| 980 |
+
"loss": 0.1579,
|
| 981 |
+
"step": 3275
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 3.9846153846153847,
|
| 985 |
+
"grad_norm": 2.231935501098633,
|
| 986 |
+
"learning_rate": 9.430866205774706e-06,
|
| 987 |
+
"loss": 0.165,
|
| 988 |
+
"step": 3300
|
| 989 |
+
},
|
| 990 |
+
{
|
| 991 |
+
"epoch": 4.015686274509804,
|
| 992 |
+
"grad_norm": 1.8732832670211792,
|
| 993 |
+
"learning_rate": 9.425782838552258e-06,
|
| 994 |
+
"loss": 0.1401,
|
| 995 |
+
"step": 3325
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 4.0458521870286575,
|
| 999 |
+
"grad_norm": 1.96370530128479,
|
| 1000 |
+
"learning_rate": 9.42069947132981e-06,
|
| 1001 |
+
"loss": 0.1241,
|
| 1002 |
+
"step": 3350
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 4.076018099547511,
|
| 1006 |
+
"grad_norm": 2.0420496463775635,
|
| 1007 |
+
"learning_rate": 9.415616104107362e-06,
|
| 1008 |
+
"loss": 0.1198,
|
| 1009 |
+
"step": 3375
|
| 1010 |
+
},
|
| 1011 |
+
{
|
| 1012 |
+
"epoch": 4.106184012066365,
|
| 1013 |
+
"grad_norm": 1.689680576324463,
|
| 1014 |
+
"learning_rate": 9.410532736884913e-06,
|
| 1015 |
+
"loss": 0.1225,
|
| 1016 |
+
"step": 3400
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 4.136349924585219,
|
| 1020 |
+
"grad_norm": 2.194132089614868,
|
| 1021 |
+
"learning_rate": 9.405449369662465e-06,
|
| 1022 |
+
"loss": 0.1333,
|
| 1023 |
+
"step": 3425
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 4.166515837104073,
|
| 1027 |
+
"grad_norm": 1.8354153633117676,
|
| 1028 |
+
"learning_rate": 9.400366002440017e-06,
|
| 1029 |
+
"loss": 0.119,
|
| 1030 |
+
"step": 3450
|
| 1031 |
+
},
|
| 1032 |
+
{
|
| 1033 |
+
"epoch": 4.196681749622926,
|
| 1034 |
+
"grad_norm": 1.5466539859771729,
|
| 1035 |
+
"learning_rate": 9.395282635217568e-06,
|
| 1036 |
+
"loss": 0.1245,
|
| 1037 |
+
"step": 3475
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 4.22684766214178,
|
| 1041 |
+
"grad_norm": 2.1299631595611572,
|
| 1042 |
+
"learning_rate": 9.39019926799512e-06,
|
| 1043 |
+
"loss": 0.1272,
|
| 1044 |
+
"step": 3500
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 4.22684766214178,
|
| 1048 |
+
"eval_loss": 0.24591179192066193,
|
| 1049 |
+
"eval_runtime": 2249.976,
|
| 1050 |
+
"eval_samples_per_second": 1.473,
|
| 1051 |
+
"eval_steps_per_second": 0.368,
|
| 1052 |
+
"eval_wer": 0.21497717486321105,
|
| 1053 |
+
"step": 3500
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 4.2570135746606335,
|
| 1057 |
+
"grad_norm": 2.316577911376953,
|
| 1058 |
+
"learning_rate": 9.385115900772672e-06,
|
| 1059 |
+
"loss": 0.1274,
|
| 1060 |
+
"step": 3525
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 4.287179487179487,
|
| 1064 |
+
"grad_norm": 1.9104264974594116,
|
| 1065 |
+
"learning_rate": 9.380032533550223e-06,
|
| 1066 |
+
"loss": 0.1231,
|
| 1067 |
+
"step": 3550
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 4.317345399698341,
|
| 1071 |
+
"grad_norm": 2.457646369934082,
|
| 1072 |
+
"learning_rate": 9.374949166327775e-06,
|
| 1073 |
+
"loss": 0.1342,
|
| 1074 |
+
"step": 3575
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 4.347511312217194,
|
| 1078 |
+
"grad_norm": 2.089953660964966,
|
| 1079 |
+
"learning_rate": 9.369865799105327e-06,
|
| 1080 |
+
"loss": 0.1186,
|
| 1081 |
+
"step": 3600
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 4.377677224736049,
|
| 1085 |
+
"grad_norm": 2.036520004272461,
|
| 1086 |
+
"learning_rate": 9.36478243188288e-06,
|
| 1087 |
+
"loss": 0.1309,
|
| 1088 |
+
"step": 3625
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 4.407843137254902,
|
| 1092 |
+
"grad_norm": 1.8452465534210205,
|
| 1093 |
+
"learning_rate": 9.359699064660432e-06,
|
| 1094 |
+
"loss": 0.1337,
|
| 1095 |
+
"step": 3650
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 4.438009049773756,
|
| 1099 |
+
"grad_norm": 2.151616096496582,
|
| 1100 |
+
"learning_rate": 9.354615697437984e-06,
|
| 1101 |
+
"loss": 0.1238,
|
| 1102 |
+
"step": 3675
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 4.4681749622926095,
|
| 1106 |
+
"grad_norm": 2.0973825454711914,
|
| 1107 |
+
"learning_rate": 9.349532330215535e-06,
|
| 1108 |
+
"loss": 0.1329,
|
| 1109 |
+
"step": 3700
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 4.498340874811463,
|
| 1113 |
+
"grad_norm": 2.1247801780700684,
|
| 1114 |
+
"learning_rate": 9.344448962993089e-06,
|
| 1115 |
+
"loss": 0.1289,
|
| 1116 |
+
"step": 3725
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 4.528506787330317,
|
| 1120 |
+
"grad_norm": 2.375617027282715,
|
| 1121 |
+
"learning_rate": 9.33936559577064e-06,
|
| 1122 |
+
"loss": 0.1306,
|
| 1123 |
+
"step": 3750
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 4.55867269984917,
|
| 1127 |
+
"grad_norm": 1.6577335596084595,
|
| 1128 |
+
"learning_rate": 9.334282228548192e-06,
|
| 1129 |
+
"loss": 0.1225,
|
| 1130 |
+
"step": 3775
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 4.588838612368024,
|
| 1134 |
+
"grad_norm": 1.7088857889175415,
|
| 1135 |
+
"learning_rate": 9.329198861325744e-06,
|
| 1136 |
+
"loss": 0.1199,
|
| 1137 |
+
"step": 3800
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 4.619004524886877,
|
| 1141 |
+
"grad_norm": 2.1655218601226807,
|
| 1142 |
+
"learning_rate": 9.324115494103296e-06,
|
| 1143 |
+
"loss": 0.1214,
|
| 1144 |
+
"step": 3825
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 4.649170437405732,
|
| 1148 |
+
"grad_norm": 1.614488959312439,
|
| 1149 |
+
"learning_rate": 9.319032126880847e-06,
|
| 1150 |
+
"loss": 0.1129,
|
| 1151 |
+
"step": 3850
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 4.6793363499245855,
|
| 1155 |
+
"grad_norm": 2.1687254905700684,
|
| 1156 |
+
"learning_rate": 9.313948759658399e-06,
|
| 1157 |
+
"loss": 0.1329,
|
| 1158 |
+
"step": 3875
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 4.709502262443439,
|
| 1162 |
+
"grad_norm": 1.8908534049987793,
|
| 1163 |
+
"learning_rate": 9.30886539243595e-06,
|
| 1164 |
+
"loss": 0.1273,
|
| 1165 |
+
"step": 3900
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 4.739668174962293,
|
| 1169 |
+
"grad_norm": 1.9584071636199951,
|
| 1170 |
+
"learning_rate": 9.303782025213502e-06,
|
| 1171 |
+
"loss": 0.1234,
|
| 1172 |
+
"step": 3925
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 4.769834087481146,
|
| 1176 |
+
"grad_norm": 1.8992727994918823,
|
| 1177 |
+
"learning_rate": 9.298698657991054e-06,
|
| 1178 |
+
"loss": 0.1202,
|
| 1179 |
+
"step": 3950
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 4.8,
|
| 1183 |
+
"grad_norm": 2.2295639514923096,
|
| 1184 |
+
"learning_rate": 9.293615290768606e-06,
|
| 1185 |
+
"loss": 0.12,
|
| 1186 |
+
"step": 3975
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 4.830165912518853,
|
| 1190 |
+
"grad_norm": 1.7892181873321533,
|
| 1191 |
+
"learning_rate": 9.288531923546157e-06,
|
| 1192 |
+
"loss": 0.1226,
|
| 1193 |
+
"step": 4000
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 4.830165912518853,
|
| 1197 |
+
"eval_loss": 0.24282999336719513,
|
| 1198 |
+
"eval_runtime": 2246.6955,
|
| 1199 |
+
"eval_samples_per_second": 1.476,
|
| 1200 |
+
"eval_steps_per_second": 0.369,
|
| 1201 |
+
"eval_wer": 0.21186907113024897,
|
| 1202 |
+
"step": 4000
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 4.860331825037708,
|
| 1206 |
+
"grad_norm": 1.6192424297332764,
|
| 1207 |
+
"learning_rate": 9.28344855632371e-06,
|
| 1208 |
+
"loss": 0.1275,
|
| 1209 |
+
"step": 4025
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 4.8904977375565615,
|
| 1213 |
+
"grad_norm": 2.2194361686706543,
|
| 1214 |
+
"learning_rate": 9.278365189101261e-06,
|
| 1215 |
+
"loss": 0.1344,
|
| 1216 |
+
"step": 4050
|
| 1217 |
+
},
|
| 1218 |
+
{
|
| 1219 |
+
"epoch": 4.920663650075415,
|
| 1220 |
+
"grad_norm": 1.8603276014328003,
|
| 1221 |
+
"learning_rate": 9.273281821878813e-06,
|
| 1222 |
+
"loss": 0.1271,
|
| 1223 |
+
"step": 4075
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 4.950829562594269,
|
| 1227 |
+
"grad_norm": 2.0720746517181396,
|
| 1228 |
+
"learning_rate": 9.268198454656366e-06,
|
| 1229 |
+
"loss": 0.1348,
|
| 1230 |
+
"step": 4100
|
| 1231 |
+
},
|
| 1232 |
+
{
|
| 1233 |
+
"epoch": 4.980995475113122,
|
| 1234 |
+
"grad_norm": 2.0119757652282715,
|
| 1235 |
+
"learning_rate": 9.263115087433918e-06,
|
| 1236 |
+
"loss": 0.1246,
|
| 1237 |
+
"step": 4125
|
| 1238 |
+
},
|
| 1239 |
+
{
|
| 1240 |
+
"epoch": 5.012066365007541,
|
| 1241 |
+
"grad_norm": 1.3096059560775757,
|
| 1242 |
+
"learning_rate": 9.25803172021147e-06,
|
| 1243 |
+
"loss": 0.1155,
|
| 1244 |
+
"step": 4150
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 5.042232277526395,
|
| 1248 |
+
"grad_norm": 1.489650011062622,
|
| 1249 |
+
"learning_rate": 9.252948352989021e-06,
|
| 1250 |
+
"loss": 0.0973,
|
| 1251 |
+
"step": 4175
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 5.072398190045249,
|
| 1255 |
+
"grad_norm": 2.003122568130493,
|
| 1256 |
+
"learning_rate": 9.247864985766573e-06,
|
| 1257 |
+
"loss": 0.0949,
|
| 1258 |
+
"step": 4200
|
| 1259 |
+
},
|
| 1260 |
+
{
|
| 1261 |
+
"epoch": 5.102564102564102,
|
| 1262 |
+
"grad_norm": 1.655964970588684,
|
| 1263 |
+
"learning_rate": 9.242781618544125e-06,
|
| 1264 |
+
"loss": 0.0907,
|
| 1265 |
+
"step": 4225
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 5.132730015082957,
|
| 1269 |
+
"grad_norm": 2.134763717651367,
|
| 1270 |
+
"learning_rate": 9.237698251321676e-06,
|
| 1271 |
+
"loss": 0.0955,
|
| 1272 |
+
"step": 4250
|
| 1273 |
+
},
|
| 1274 |
+
{
|
| 1275 |
+
"epoch": 5.16289592760181,
|
| 1276 |
+
"grad_norm": 2.039686918258667,
|
| 1277 |
+
"learning_rate": 9.232614884099228e-06,
|
| 1278 |
+
"loss": 0.0959,
|
| 1279 |
+
"step": 4275
|
| 1280 |
+
},
|
| 1281 |
+
{
|
| 1282 |
+
"epoch": 5.193061840120664,
|
| 1283 |
+
"grad_norm": 2.1623027324676514,
|
| 1284 |
+
"learning_rate": 9.22753151687678e-06,
|
| 1285 |
+
"loss": 0.0971,
|
| 1286 |
+
"step": 4300
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 5.223227752639517,
|
| 1290 |
+
"grad_norm": 2.3452537059783936,
|
| 1291 |
+
"learning_rate": 9.222448149654331e-06,
|
| 1292 |
+
"loss": 0.0882,
|
| 1293 |
+
"step": 4325
|
| 1294 |
+
},
|
| 1295 |
+
{
|
| 1296 |
+
"epoch": 5.253393665158371,
|
| 1297 |
+
"grad_norm": 1.7960082292556763,
|
| 1298 |
+
"learning_rate": 9.217364782431883e-06,
|
| 1299 |
+
"loss": 0.0936,
|
| 1300 |
+
"step": 4350
|
| 1301 |
+
},
|
| 1302 |
+
{
|
| 1303 |
+
"epoch": 5.283559577677225,
|
| 1304 |
+
"grad_norm": 1.9322994947433472,
|
| 1305 |
+
"learning_rate": 9.212281415209435e-06,
|
| 1306 |
+
"loss": 0.099,
|
| 1307 |
+
"step": 4375
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 5.313725490196078,
|
| 1311 |
+
"grad_norm": 2.040149688720703,
|
| 1312 |
+
"learning_rate": 9.207198047986986e-06,
|
| 1313 |
+
"loss": 0.0974,
|
| 1314 |
+
"step": 4400
|
| 1315 |
+
},
|
| 1316 |
+
{
|
| 1317 |
+
"epoch": 5.343891402714932,
|
| 1318 |
+
"grad_norm": 2.1404929161071777,
|
| 1319 |
+
"learning_rate": 9.202114680764538e-06,
|
| 1320 |
+
"loss": 0.0886,
|
| 1321 |
+
"step": 4425
|
| 1322 |
+
},
|
| 1323 |
+
{
|
| 1324 |
+
"epoch": 5.374057315233786,
|
| 1325 |
+
"grad_norm": 1.9556715488433838,
|
| 1326 |
+
"learning_rate": 9.197031313542092e-06,
|
| 1327 |
+
"loss": 0.1003,
|
| 1328 |
+
"step": 4450
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 5.40422322775264,
|
| 1332 |
+
"grad_norm": 2.070523500442505,
|
| 1333 |
+
"learning_rate": 9.191947946319643e-06,
|
| 1334 |
+
"loss": 0.0972,
|
| 1335 |
+
"step": 4475
|
| 1336 |
+
},
|
| 1337 |
+
{
|
| 1338 |
+
"epoch": 5.4343891402714934,
|
| 1339 |
+
"grad_norm": 1.8254801034927368,
|
| 1340 |
+
"learning_rate": 9.186864579097195e-06,
|
| 1341 |
+
"loss": 0.0939,
|
| 1342 |
+
"step": 4500
|
| 1343 |
+
},
|
| 1344 |
+
{
|
| 1345 |
+
"epoch": 5.4343891402714934,
|
| 1346 |
+
"eval_loss": 0.2541460692882538,
|
| 1347 |
+
"eval_runtime": 2247.1988,
|
| 1348 |
+
"eval_samples_per_second": 1.475,
|
| 1349 |
+
"eval_steps_per_second": 0.369,
|
| 1350 |
+
"eval_wer": 0.2141785648762694,
|
| 1351 |
+
"step": 4500
|
| 1352 |
+
},
|
| 1353 |
+
{
|
| 1354 |
+
"epoch": 5.464555052790347,
|
| 1355 |
+
"grad_norm": 2.025676965713501,
|
| 1356 |
+
"learning_rate": 9.181781211874747e-06,
|
| 1357 |
+
"loss": 0.0941,
|
| 1358 |
+
"step": 4525
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 5.494720965309201,
|
| 1362 |
+
"grad_norm": 2.0886032581329346,
|
| 1363 |
+
"learning_rate": 9.176697844652298e-06,
|
| 1364 |
+
"loss": 0.0976,
|
| 1365 |
+
"step": 4550
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 5.524886877828054,
|
| 1369 |
+
"grad_norm": 2.048823118209839,
|
| 1370 |
+
"learning_rate": 9.17161447742985e-06,
|
| 1371 |
+
"loss": 0.1008,
|
| 1372 |
+
"step": 4575
|
| 1373 |
+
},
|
| 1374 |
+
{
|
| 1375 |
+
"epoch": 5.555052790346908,
|
| 1376 |
+
"grad_norm": 2.661174774169922,
|
| 1377 |
+
"learning_rate": 9.166531110207402e-06,
|
| 1378 |
+
"loss": 0.0971,
|
| 1379 |
+
"step": 4600
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 5.585218702865761,
|
| 1383 |
+
"grad_norm": 2.3181405067443848,
|
| 1384 |
+
"learning_rate": 9.161447742984953e-06,
|
| 1385 |
+
"loss": 0.1005,
|
| 1386 |
+
"step": 4625
|
| 1387 |
+
},
|
| 1388 |
+
{
|
| 1389 |
+
"epoch": 5.615384615384615,
|
| 1390 |
+
"grad_norm": 1.9508439302444458,
|
| 1391 |
+
"learning_rate": 9.156364375762505e-06,
|
| 1392 |
+
"loss": 0.098,
|
| 1393 |
+
"step": 4650
|
| 1394 |
+
},
|
| 1395 |
+
{
|
| 1396 |
+
"epoch": 5.6455505279034695,
|
| 1397 |
+
"grad_norm": 2.297891139984131,
|
| 1398 |
+
"learning_rate": 9.151281008540057e-06,
|
| 1399 |
+
"loss": 0.0994,
|
| 1400 |
+
"step": 4675
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 5.675716440422323,
|
| 1404 |
+
"grad_norm": 1.9096143245697021,
|
| 1405 |
+
"learning_rate": 9.146197641317609e-06,
|
| 1406 |
+
"loss": 0.1021,
|
| 1407 |
+
"step": 4700
|
| 1408 |
+
},
|
| 1409 |
+
{
|
| 1410 |
+
"epoch": 5.705882352941177,
|
| 1411 |
+
"grad_norm": 2.3850667476654053,
|
| 1412 |
+
"learning_rate": 9.14111427409516e-06,
|
| 1413 |
+
"loss": 0.0988,
|
| 1414 |
+
"step": 4725
|
| 1415 |
+
},
|
| 1416 |
+
{
|
| 1417 |
+
"epoch": 5.73604826546003,
|
| 1418 |
+
"grad_norm": 1.954728126525879,
|
| 1419 |
+
"learning_rate": 9.136030906872714e-06,
|
| 1420 |
+
"loss": 0.1021,
|
| 1421 |
+
"step": 4750
|
| 1422 |
+
},
|
| 1423 |
+
{
|
| 1424 |
+
"epoch": 5.766214177978884,
|
| 1425 |
+
"grad_norm": 2.3807568550109863,
|
| 1426 |
+
"learning_rate": 9.130947539650265e-06,
|
| 1427 |
+
"loss": 0.0891,
|
| 1428 |
+
"step": 4775
|
| 1429 |
+
},
|
| 1430 |
+
{
|
| 1431 |
+
"epoch": 5.796380090497737,
|
| 1432 |
+
"grad_norm": 2.0435431003570557,
|
| 1433 |
+
"learning_rate": 9.125864172427817e-06,
|
| 1434 |
+
"loss": 0.1055,
|
| 1435 |
+
"step": 4800
|
| 1436 |
+
},
|
| 1437 |
+
{
|
| 1438 |
+
"epoch": 5.826546003016591,
|
| 1439 |
+
"grad_norm": 2.7044458389282227,
|
| 1440 |
+
"learning_rate": 9.120780805205369e-06,
|
| 1441 |
+
"loss": 0.0979,
|
| 1442 |
+
"step": 4825
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 5.856711915535445,
|
| 1446 |
+
"grad_norm": 1.9672693014144897,
|
| 1447 |
+
"learning_rate": 9.11569743798292e-06,
|
| 1448 |
+
"loss": 0.1003,
|
| 1449 |
+
"step": 4850
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 5.886877828054299,
|
| 1453 |
+
"grad_norm": 1.9601274728775024,
|
| 1454 |
+
"learning_rate": 9.110614070760472e-06,
|
| 1455 |
+
"loss": 0.0987,
|
| 1456 |
+
"step": 4875
|
| 1457 |
+
},
|
| 1458 |
+
{
|
| 1459 |
+
"epoch": 5.917043740573153,
|
| 1460 |
+
"grad_norm": 2.0287139415740967,
|
| 1461 |
+
"learning_rate": 9.105530703538024e-06,
|
| 1462 |
+
"loss": 0.0963,
|
| 1463 |
+
"step": 4900
|
| 1464 |
+
},
|
| 1465 |
+
{
|
| 1466 |
+
"epoch": 5.947209653092006,
|
| 1467 |
+
"grad_norm": 1.9941165447235107,
|
| 1468 |
+
"learning_rate": 9.100447336315577e-06,
|
| 1469 |
+
"loss": 0.1005,
|
| 1470 |
+
"step": 4925
|
| 1471 |
+
},
|
| 1472 |
+
{
|
| 1473 |
+
"epoch": 5.97737556561086,
|
| 1474 |
+
"grad_norm": 1.8804293870925903,
|
| 1475 |
+
"learning_rate": 9.095363969093129e-06,
|
| 1476 |
+
"loss": 0.1101,
|
| 1477 |
+
"step": 4950
|
| 1478 |
+
},
|
| 1479 |
+
{
|
| 1480 |
+
"epoch": 6.008446455505279,
|
| 1481 |
+
"grad_norm": 1.3988499641418457,
|
| 1482 |
+
"learning_rate": 9.09028060187068e-06,
|
| 1483 |
+
"loss": 0.0912,
|
| 1484 |
+
"step": 4975
|
| 1485 |
+
},
|
| 1486 |
+
{
|
| 1487 |
+
"epoch": 6.038612368024133,
|
| 1488 |
+
"grad_norm": 1.9083130359649658,
|
| 1489 |
+
"learning_rate": 9.085197234648232e-06,
|
| 1490 |
+
"loss": 0.0665,
|
| 1491 |
+
"step": 5000
|
| 1492 |
+
},
|
| 1493 |
+
{
|
| 1494 |
+
"epoch": 6.038612368024133,
|
| 1495 |
+
"eval_loss": 0.2640438675880432,
|
| 1496 |
+
"eval_runtime": 2326.5234,
|
| 1497 |
+
"eval_samples_per_second": 1.425,
|
| 1498 |
+
"eval_steps_per_second": 0.356,
|
| 1499 |
+
"eval_wer": 0.21559232039369314,
|
| 1500 |
+
"step": 5000
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 6.068778280542986,
|
| 1504 |
+
"grad_norm": 1.7908351421356201,
|
| 1505 |
+
"learning_rate": 9.080113867425784e-06,
|
| 1506 |
+
"loss": 0.068,
|
| 1507 |
+
"step": 5025
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 6.09894419306184,
|
| 1511 |
+
"grad_norm": 2.0591347217559814,
|
| 1512 |
+
"learning_rate": 9.075030500203336e-06,
|
| 1513 |
+
"loss": 0.0761,
|
| 1514 |
+
"step": 5050
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 6.129110105580694,
|
| 1518 |
+
"grad_norm": 2.0180952548980713,
|
| 1519 |
+
"learning_rate": 9.069947132980888e-06,
|
| 1520 |
+
"loss": 0.0651,
|
| 1521 |
+
"step": 5075
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 6.159276018099548,
|
| 1525 |
+
"grad_norm": 1.842887282371521,
|
| 1526 |
+
"learning_rate": 9.06486376575844e-06,
|
| 1527 |
+
"loss": 0.0712,
|
| 1528 |
+
"step": 5100
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 6.189441930618401,
|
| 1532 |
+
"grad_norm": 1.8542691469192505,
|
| 1533 |
+
"learning_rate": 9.059780398535991e-06,
|
| 1534 |
+
"loss": 0.0685,
|
| 1535 |
+
"step": 5125
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 6.219607843137255,
|
| 1539 |
+
"grad_norm": 1.998128890991211,
|
| 1540 |
+
"learning_rate": 9.054697031313543e-06,
|
| 1541 |
+
"loss": 0.0663,
|
| 1542 |
+
"step": 5150
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 6.249773755656109,
|
| 1546 |
+
"grad_norm": 2.074847459793091,
|
| 1547 |
+
"learning_rate": 9.049613664091094e-06,
|
| 1548 |
+
"loss": 0.0691,
|
| 1549 |
+
"step": 5175
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 6.279939668174962,
|
| 1553 |
+
"grad_norm": 2.1815929412841797,
|
| 1554 |
+
"learning_rate": 9.044530296868646e-06,
|
| 1555 |
+
"loss": 0.0646,
|
| 1556 |
+
"step": 5200
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 6.310105580693816,
|
| 1560 |
+
"grad_norm": 1.8308629989624023,
|
| 1561 |
+
"learning_rate": 9.039446929646198e-06,
|
| 1562 |
+
"loss": 0.079,
|
| 1563 |
+
"step": 5225
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 6.340271493212669,
|
| 1567 |
+
"grad_norm": 1.6758077144622803,
|
| 1568 |
+
"learning_rate": 9.03436356242375e-06,
|
| 1569 |
+
"loss": 0.0747,
|
| 1570 |
+
"step": 5250
|
| 1571 |
+
},
|
| 1572 |
+
{
|
| 1573 |
+
"epoch": 6.370437405731524,
|
| 1574 |
+
"grad_norm": 1.7715494632720947,
|
| 1575 |
+
"learning_rate": 9.029280195201303e-06,
|
| 1576 |
+
"loss": 0.0699,
|
| 1577 |
+
"step": 5275
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 6.400603318250377,
|
| 1581 |
+
"grad_norm": 1.9695020914077759,
|
| 1582 |
+
"learning_rate": 9.024196827978855e-06,
|
| 1583 |
+
"loss": 0.0706,
|
| 1584 |
+
"step": 5300
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 6.430769230769231,
|
| 1588 |
+
"grad_norm": 2.174950122833252,
|
| 1589 |
+
"learning_rate": 9.019113460756406e-06,
|
| 1590 |
+
"loss": 0.0735,
|
| 1591 |
+
"step": 5325
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 6.460935143288085,
|
| 1595 |
+
"grad_norm": 1.3423354625701904,
|
| 1596 |
+
"learning_rate": 9.014030093533958e-06,
|
| 1597 |
+
"loss": 0.0687,
|
| 1598 |
+
"step": 5350
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 6.491101055806938,
|
| 1602 |
+
"grad_norm": 1.8988721370697021,
|
| 1603 |
+
"learning_rate": 9.00894672631151e-06,
|
| 1604 |
+
"loss": 0.0684,
|
| 1605 |
+
"step": 5375
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 6.521266968325792,
|
| 1609 |
+
"grad_norm": 2.1522958278656006,
|
| 1610 |
+
"learning_rate": 9.003863359089061e-06,
|
| 1611 |
+
"loss": 0.0703,
|
| 1612 |
+
"step": 5400
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 6.551432880844645,
|
| 1616 |
+
"grad_norm": 2.2434656620025635,
|
| 1617 |
+
"learning_rate": 8.998779991866613e-06,
|
| 1618 |
+
"loss": 0.0725,
|
| 1619 |
+
"step": 5425
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 6.581598793363499,
|
| 1623 |
+
"grad_norm": 1.9670661687850952,
|
| 1624 |
+
"learning_rate": 8.993696624644165e-06,
|
| 1625 |
+
"loss": 0.073,
|
| 1626 |
+
"step": 5450
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 6.6117647058823525,
|
| 1630 |
+
"grad_norm": 1.9538111686706543,
|
| 1631 |
+
"learning_rate": 8.988613257421716e-06,
|
| 1632 |
+
"loss": 0.0755,
|
| 1633 |
+
"step": 5475
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 6.641930618401207,
|
| 1637 |
+
"grad_norm": 1.7768446207046509,
|
| 1638 |
+
"learning_rate": 8.983529890199268e-06,
|
| 1639 |
+
"loss": 0.0717,
|
| 1640 |
+
"step": 5500
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 6.641930618401207,
|
| 1644 |
+
"eval_loss": 0.27195391058921814,
|
| 1645 |
+
"eval_runtime": 2447.6492,
|
| 1646 |
+
"eval_samples_per_second": 1.354,
|
| 1647 |
+
"eval_steps_per_second": 0.339,
|
| 1648 |
+
"eval_wer": 0.2185061676433451,
|
| 1649 |
+
"step": 5500
|
| 1650 |
+
},
|
| 1651 |
+
{
|
| 1652 |
+
"epoch": 6.641930618401207,
|
| 1653 |
+
"step": 5500,
|
| 1654 |
+
"total_flos": 5.08256607043584e+19,
|
| 1655 |
+
"train_loss": 0.25465128779411317,
|
| 1656 |
+
"train_runtime": 113391.8356,
|
| 1657 |
+
"train_samples_per_second": 14.031,
|
| 1658 |
+
"train_steps_per_second": 0.438
|
| 1659 |
+
}
|
| 1660 |
+
],
|
| 1661 |
+
"logging_steps": 25,
|
| 1662 |
+
"max_steps": 49680,
|
| 1663 |
+
"num_input_tokens_seen": 0,
|
| 1664 |
+
"num_train_epochs": 60,
|
| 1665 |
+
"save_steps": 500,
|
| 1666 |
+
"stateful_callbacks": {
|
| 1667 |
+
"EarlyStoppingCallback": {
|
| 1668 |
+
"args": {
|
| 1669 |
+
"early_stopping_patience": 5,
|
| 1670 |
+
"early_stopping_threshold": 0.0
|
| 1671 |
+
},
|
| 1672 |
+
"attributes": {
|
| 1673 |
+
"early_stopping_patience_counter": 5
|
| 1674 |
+
}
|
| 1675 |
+
},
|
| 1676 |
+
"TrainerControl": {
|
| 1677 |
+
"args": {
|
| 1678 |
+
"should_epoch_stop": false,
|
| 1679 |
+
"should_evaluate": false,
|
| 1680 |
+
"should_log": false,
|
| 1681 |
+
"should_save": true,
|
| 1682 |
+
"should_training_stop": true
|
| 1683 |
+
},
|
| 1684 |
+
"attributes": {}
|
| 1685 |
+
}
|
| 1686 |
+
},
|
| 1687 |
+
"total_flos": 5.08256607043584e+19,
|
| 1688 |
+
"train_batch_size": 4,
|
| 1689 |
+
"trial_name": null,
|
| 1690 |
+
"trial_params": null
|
| 1691 |
+
}
|