End of training
Browse files- README.md +7 -2
- all_results.json +14 -1
- phase_1_train_results.json +9 -0
- runs/Apr08_09-20-26_dgx-a100/events.out.tfevents.1744080802.dgx-a100.554420.1 +3 -0
- trainer_state.json +2243 -0
- val_results.json +9 -0
README.md
CHANGED
|
@@ -1,20 +1,25 @@
|
|
| 1 |
---
|
| 2 |
library_name: transformers
|
|
|
|
|
|
|
| 3 |
license: apache-2.0
|
| 4 |
base_model: openai/whisper-tiny
|
| 5 |
tags:
|
|
|
|
|
|
|
|
|
|
| 6 |
- generated_from_trainer
|
| 7 |
metrics:
|
| 8 |
- wer
|
| 9 |
model-index:
|
| 10 |
-
- name:
|
| 11 |
results: []
|
| 12 |
---
|
| 13 |
|
| 14 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 15 |
should probably proofread and complete it, then remove this comment. -->
|
| 16 |
|
| 17 |
-
#
|
| 18 |
|
| 19 |
This model is a fine-tuned version of [openai/whisper-tiny](https://huggingface.co/openai/whisper-tiny) on the None dataset.
|
| 20 |
It achieves the following results on the evaluation set:
|
|
|
|
| 1 |
---
|
| 2 |
library_name: transformers
|
| 3 |
+
language:
|
| 4 |
+
- jv
|
| 5 |
license: apache-2.0
|
| 6 |
base_model: openai/whisper-tiny
|
| 7 |
tags:
|
| 8 |
+
- whisper
|
| 9 |
+
- javanese
|
| 10 |
+
- asr
|
| 11 |
- generated_from_trainer
|
| 12 |
metrics:
|
| 13 |
- wer
|
| 14 |
model-index:
|
| 15 |
+
- name: Whisper-Tiny-Java-v5
|
| 16 |
results: []
|
| 17 |
---
|
| 18 |
|
| 19 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 20 |
should probably proofread and complete it, then remove this comment. -->
|
| 21 |
|
| 22 |
+
# Whisper-Tiny-Java-v5
|
| 23 |
|
| 24 |
This model is a fine-tuned version of [openai/whisper-tiny](https://huggingface.co/openai/whisper-tiny) on the None dataset.
|
| 25 |
It achieves the following results on the evaluation set:
|
all_results.json
CHANGED
|
@@ -1,4 +1,10 @@
|
|
| 1 |
{
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2 |
"pretrained_eval_loss": 4.033740043640137,
|
| 3 |
"pretrained_eval_model_preparation_time": 0.0019,
|
| 4 |
"pretrained_eval_runtime": 1108.6741,
|
|
@@ -9,5 +15,12 @@
|
|
| 9 |
"pretrained_val_runtime": 1685.7446,
|
| 10 |
"pretrained_val_samples_per_second": 5.498,
|
| 11 |
"pretrained_val_steps_per_second": 0.688,
|
| 12 |
-
"pretrained_val_wer": 1.3147204165302655
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 13 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 10.813365051903114,
|
| 3 |
+
"phase_1_val_loss": 0.2532360255718231,
|
| 4 |
+
"phase_1_val_runtime": 1660.0444,
|
| 5 |
+
"phase_1_val_samples_per_second": 5.583,
|
| 6 |
+
"phase_1_val_steps_per_second": 1.396,
|
| 7 |
+
"phase_1_val_wer": 0.1760276855445915,
|
| 8 |
"pretrained_eval_loss": 4.033740043640137,
|
| 9 |
"pretrained_eval_model_preparation_time": 0.0019,
|
| 10 |
"pretrained_eval_runtime": 1108.6741,
|
|
|
|
| 15 |
"pretrained_val_runtime": 1685.7446,
|
| 16 |
"pretrained_val_samples_per_second": 5.498,
|
| 17 |
"pretrained_val_steps_per_second": 0.688,
|
| 18 |
+
"pretrained_val_wer": 1.3147204165302655,
|
| 19 |
+
"total_flos": 2.1271110511959736e+19,
|
| 20 |
+
"train_loss": 3.7687225894058625e-07,
|
| 21 |
+
"train_runtime": 8.535,
|
| 22 |
+
"train_samples": 73984,
|
| 23 |
+
"train_samples_per_second": 93731.391,
|
| 24 |
+
"train_steps_per_second": 5858.212,
|
| 25 |
+
"val_samples": 9268
|
| 26 |
}
|
phase_1_train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 10.813365051903114,
|
| 3 |
+
"total_flos": 2.1271110511959736e+19,
|
| 4 |
+
"train_loss": 3.7687225894058625e-07,
|
| 5 |
+
"train_runtime": 8.535,
|
| 6 |
+
"train_samples": 73984,
|
| 7 |
+
"train_samples_per_second": 93731.391,
|
| 8 |
+
"train_steps_per_second": 5858.212
|
| 9 |
+
}
|
runs/Apr08_09-20-26_dgx-a100/events.out.tfevents.1744080802.dgx-a100.554420.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cbd12df97cbfe46dfff3e2fa6d12b60f007227d7d802e2e29bc27e7ff9087302
|
| 3 |
+
size 477
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2243 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 49000,
|
| 3 |
+
"best_metric": 0.1760276855445915,
|
| 4 |
+
"best_model_checkpoint": "/home/cluster-dgxa100/slp01/bagas-fine-tune-whisper/whisper-tiny-javanese-openslr-v5/checkpoint-49000",
|
| 5 |
+
"epoch": 10.813365051903114,
|
| 6 |
+
"eval_steps": 1000,
|
| 7 |
+
"global_step": 50001,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.08650519031141868,
|
| 14 |
+
"grad_norm": 18.705734252929688,
|
| 15 |
+
"learning_rate": 7.88e-07,
|
| 16 |
+
"loss": 3.5867,
|
| 17 |
+
"step": 200
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.17301038062283736,
|
| 21 |
+
"grad_norm": 17.47745132446289,
|
| 22 |
+
"learning_rate": 1.588e-06,
|
| 23 |
+
"loss": 1.9747,
|
| 24 |
+
"step": 400
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.25951557093425603,
|
| 28 |
+
"grad_norm": 13.15998649597168,
|
| 29 |
+
"learning_rate": 2.3880000000000003e-06,
|
| 30 |
+
"loss": 1.4998,
|
| 31 |
+
"step": 600
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.3460207612456747,
|
| 35 |
+
"grad_norm": 11.823753356933594,
|
| 36 |
+
"learning_rate": 3.188e-06,
|
| 37 |
+
"loss": 1.2625,
|
| 38 |
+
"step": 800
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.43252595155709345,
|
| 42 |
+
"grad_norm": 14.030326843261719,
|
| 43 |
+
"learning_rate": 3.988000000000001e-06,
|
| 44 |
+
"loss": 1.0859,
|
| 45 |
+
"step": 1000
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.43252595155709345,
|
| 49 |
+
"eval_loss": 0.9075252413749695,
|
| 50 |
+
"eval_runtime": 886.547,
|
| 51 |
+
"eval_samples_per_second": 10.454,
|
| 52 |
+
"eval_steps_per_second": 1.307,
|
| 53 |
+
"eval_wer": 0.6418833647539068,
|
| 54 |
+
"step": 1000
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.5190311418685121,
|
| 58 |
+
"grad_norm": 11.967138290405273,
|
| 59 |
+
"learning_rate": 4.7880000000000006e-06,
|
| 60 |
+
"loss": 0.9761,
|
| 61 |
+
"step": 1200
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 0.6055363321799307,
|
| 65 |
+
"grad_norm": 10.42967414855957,
|
| 66 |
+
"learning_rate": 5.588e-06,
|
| 67 |
+
"loss": 0.88,
|
| 68 |
+
"step": 1400
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.6920415224913494,
|
| 72 |
+
"grad_norm": 11.49984073638916,
|
| 73 |
+
"learning_rate": 6.3880000000000005e-06,
|
| 74 |
+
"loss": 0.817,
|
| 75 |
+
"step": 1600
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.7785467128027682,
|
| 79 |
+
"grad_norm": 11.280434608459473,
|
| 80 |
+
"learning_rate": 7.1880000000000005e-06,
|
| 81 |
+
"loss": 0.7575,
|
| 82 |
+
"step": 1800
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"epoch": 0.8650519031141869,
|
| 86 |
+
"grad_norm": 9.452693939208984,
|
| 87 |
+
"learning_rate": 7.988e-06,
|
| 88 |
+
"loss": 0.7132,
|
| 89 |
+
"step": 2000
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 0.8650519031141869,
|
| 93 |
+
"eval_loss": 0.6100018620491028,
|
| 94 |
+
"eval_runtime": 837.4552,
|
| 95 |
+
"eval_samples_per_second": 11.067,
|
| 96 |
+
"eval_steps_per_second": 1.384,
|
| 97 |
+
"eval_wer": 0.5481357991991649,
|
| 98 |
+
"step": 2000
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.9515570934256056,
|
| 102 |
+
"grad_norm": 7.524653911590576,
|
| 103 |
+
"learning_rate": 8.788000000000001e-06,
|
| 104 |
+
"loss": 0.6768,
|
| 105 |
+
"step": 2200
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 1.0380622837370241,
|
| 109 |
+
"grad_norm": 10.51672649383545,
|
| 110 |
+
"learning_rate": 9.588e-06,
|
| 111 |
+
"loss": 0.6132,
|
| 112 |
+
"step": 2400
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 1.1245674740484428,
|
| 116 |
+
"grad_norm": 9.797249794006348,
|
| 117 |
+
"learning_rate": 1.0388e-05,
|
| 118 |
+
"loss": 0.5662,
|
| 119 |
+
"step": 2600
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 1.2110726643598615,
|
| 123 |
+
"grad_norm": 9.872428894042969,
|
| 124 |
+
"learning_rate": 1.1188e-05,
|
| 125 |
+
"loss": 0.5447,
|
| 126 |
+
"step": 2800
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 1.2975778546712804,
|
| 130 |
+
"grad_norm": 11.230769157409668,
|
| 131 |
+
"learning_rate": 1.1988000000000001e-05,
|
| 132 |
+
"loss": 0.5258,
|
| 133 |
+
"step": 3000
|
| 134 |
+
},
|
| 135 |
+
{
|
| 136 |
+
"epoch": 1.2975778546712804,
|
| 137 |
+
"eval_loss": 0.48219427466392517,
|
| 138 |
+
"eval_runtime": 848.682,
|
| 139 |
+
"eval_samples_per_second": 10.92,
|
| 140 |
+
"eval_steps_per_second": 1.366,
|
| 141 |
+
"eval_wer": 0.6351526105043391,
|
| 142 |
+
"step": 3000
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 1.3840830449826989,
|
| 146 |
+
"grad_norm": 10.13039779663086,
|
| 147 |
+
"learning_rate": 1.2788e-05,
|
| 148 |
+
"loss": 0.5046,
|
| 149 |
+
"step": 3200
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 1.4705882352941178,
|
| 153 |
+
"grad_norm": 7.22476863861084,
|
| 154 |
+
"learning_rate": 1.3588000000000001e-05,
|
| 155 |
+
"loss": 0.4905,
|
| 156 |
+
"step": 3400
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 1.5570934256055362,
|
| 160 |
+
"grad_norm": 6.499698162078857,
|
| 161 |
+
"learning_rate": 1.4388000000000002e-05,
|
| 162 |
+
"loss": 0.475,
|
| 163 |
+
"step": 3600
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 1.643598615916955,
|
| 167 |
+
"grad_norm": 8.378984451293945,
|
| 168 |
+
"learning_rate": 1.5188000000000001e-05,
|
| 169 |
+
"loss": 0.4595,
|
| 170 |
+
"step": 3800
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 1.7301038062283736,
|
| 174 |
+
"grad_norm": 7.435483455657959,
|
| 175 |
+
"learning_rate": 1.5988e-05,
|
| 176 |
+
"loss": 0.4521,
|
| 177 |
+
"step": 4000
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 1.7301038062283736,
|
| 181 |
+
"eval_loss": 0.4058275520801544,
|
| 182 |
+
"eval_runtime": 826.7651,
|
| 183 |
+
"eval_samples_per_second": 11.21,
|
| 184 |
+
"eval_steps_per_second": 1.402,
|
| 185 |
+
"eval_wer": 0.4619447517255348,
|
| 186 |
+
"step": 4000
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 0.9083044982698962,
|
| 190 |
+
"grad_norm": 11.933815956115723,
|
| 191 |
+
"learning_rate": 1.6788000000000003e-05,
|
| 192 |
+
"loss": 0.3833,
|
| 193 |
+
"step": 4200
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 0.9515570934256056,
|
| 197 |
+
"grad_norm": 8.0033597946167,
|
| 198 |
+
"learning_rate": 1.7588e-05,
|
| 199 |
+
"loss": 0.3927,
|
| 200 |
+
"step": 4400
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.9948096885813149,
|
| 204 |
+
"grad_norm": 10.064678192138672,
|
| 205 |
+
"learning_rate": 1.8388e-05,
|
| 206 |
+
"loss": 0.3647,
|
| 207 |
+
"step": 4600
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 1.0380622837370241,
|
| 211 |
+
"grad_norm": 11.17820930480957,
|
| 212 |
+
"learning_rate": 1.9188000000000003e-05,
|
| 213 |
+
"loss": 0.3884,
|
| 214 |
+
"step": 4800
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 1.0813148788927336,
|
| 218 |
+
"grad_norm": 11.01378059387207,
|
| 219 |
+
"learning_rate": 1.9988000000000002e-05,
|
| 220 |
+
"loss": 0.3848,
|
| 221 |
+
"step": 5000
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.0813148788927336,
|
| 225 |
+
"eval_loss": 0.40224945545196533,
|
| 226 |
+
"eval_runtime": 1629.692,
|
| 227 |
+
"eval_samples_per_second": 5.687,
|
| 228 |
+
"eval_steps_per_second": 1.422,
|
| 229 |
+
"eval_wer": 0.37777673853060845,
|
| 230 |
+
"step": 5000
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 1.1245674740484428,
|
| 234 |
+
"grad_norm": 11.379752159118652,
|
| 235 |
+
"learning_rate": 1.9912444444444445e-05,
|
| 236 |
+
"loss": 0.3676,
|
| 237 |
+
"step": 5200
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 1.1678200692041523,
|
| 241 |
+
"grad_norm": 11.409998893737793,
|
| 242 |
+
"learning_rate": 1.982355555555556e-05,
|
| 243 |
+
"loss": 0.3696,
|
| 244 |
+
"step": 5400
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 1.2110726643598615,
|
| 248 |
+
"grad_norm": 10.616972923278809,
|
| 249 |
+
"learning_rate": 1.973466666666667e-05,
|
| 250 |
+
"loss": 0.3715,
|
| 251 |
+
"step": 5600
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 1.254325259515571,
|
| 255 |
+
"grad_norm": 11.608460426330566,
|
| 256 |
+
"learning_rate": 1.964577777777778e-05,
|
| 257 |
+
"loss": 0.3602,
|
| 258 |
+
"step": 5800
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 1.2975778546712804,
|
| 262 |
+
"grad_norm": 11.068222045898438,
|
| 263 |
+
"learning_rate": 1.955688888888889e-05,
|
| 264 |
+
"loss": 0.351,
|
| 265 |
+
"step": 6000
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 1.2975778546712804,
|
| 269 |
+
"eval_loss": 0.3710990846157074,
|
| 270 |
+
"eval_runtime": 1563.0786,
|
| 271 |
+
"eval_samples_per_second": 5.929,
|
| 272 |
+
"eval_steps_per_second": 1.482,
|
| 273 |
+
"eval_wer": 0.33260066407894123,
|
| 274 |
+
"step": 6000
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"epoch": 1.3408304498269896,
|
| 278 |
+
"grad_norm": 9.681553840637207,
|
| 279 |
+
"learning_rate": 1.9468444444444444e-05,
|
| 280 |
+
"loss": 0.3345,
|
| 281 |
+
"step": 6200
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 1.3840830449826989,
|
| 285 |
+
"grad_norm": 11.10158634185791,
|
| 286 |
+
"learning_rate": 1.938e-05,
|
| 287 |
+
"loss": 0.3499,
|
| 288 |
+
"step": 6400
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 1.4273356401384083,
|
| 292 |
+
"grad_norm": 9.192054748535156,
|
| 293 |
+
"learning_rate": 1.9291111111111115e-05,
|
| 294 |
+
"loss": 0.338,
|
| 295 |
+
"step": 6600
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 1.4705882352941178,
|
| 299 |
+
"grad_norm": 7.681017875671387,
|
| 300 |
+
"learning_rate": 1.9202222222222225e-05,
|
| 301 |
+
"loss": 0.3299,
|
| 302 |
+
"step": 6800
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 1.513840830449827,
|
| 306 |
+
"grad_norm": 13.422704696655273,
|
| 307 |
+
"learning_rate": 1.9113333333333336e-05,
|
| 308 |
+
"loss": 0.3277,
|
| 309 |
+
"step": 7000
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 1.513840830449827,
|
| 313 |
+
"eval_loss": 0.3545876443386078,
|
| 314 |
+
"eval_runtime": 1575.5397,
|
| 315 |
+
"eval_samples_per_second": 5.882,
|
| 316 |
+
"eval_steps_per_second": 1.471,
|
| 317 |
+
"eval_wer": 0.30530483717594975,
|
| 318 |
+
"step": 7000
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 1.5570934256055362,
|
| 322 |
+
"grad_norm": 8.128918647766113,
|
| 323 |
+
"learning_rate": 1.9024444444444446e-05,
|
| 324 |
+
"loss": 0.3199,
|
| 325 |
+
"step": 7200
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 1.6003460207612457,
|
| 329 |
+
"grad_norm": 7.922735214233398,
|
| 330 |
+
"learning_rate": 1.8935555555555556e-05,
|
| 331 |
+
"loss": 0.3204,
|
| 332 |
+
"step": 7400
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 1.643598615916955,
|
| 336 |
+
"grad_norm": 10.091095924377441,
|
| 337 |
+
"learning_rate": 1.884666666666667e-05,
|
| 338 |
+
"loss": 0.3146,
|
| 339 |
+
"step": 7600
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 1.6868512110726643,
|
| 343 |
+
"grad_norm": 10.206018447875977,
|
| 344 |
+
"learning_rate": 1.875777777777778e-05,
|
| 345 |
+
"loss": 0.3077,
|
| 346 |
+
"step": 7800
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 1.7301038062283736,
|
| 350 |
+
"grad_norm": 7.157899856567383,
|
| 351 |
+
"learning_rate": 1.866888888888889e-05,
|
| 352 |
+
"loss": 0.3122,
|
| 353 |
+
"step": 8000
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 1.7301038062283736,
|
| 357 |
+
"eval_loss": 0.33699458837509155,
|
| 358 |
+
"eval_runtime": 1571.4771,
|
| 359 |
+
"eval_samples_per_second": 5.898,
|
| 360 |
+
"eval_steps_per_second": 1.474,
|
| 361 |
+
"eval_wer": 0.2861774930240534,
|
| 362 |
+
"step": 8000
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 1.773356401384083,
|
| 366 |
+
"grad_norm": 10.22962474822998,
|
| 367 |
+
"learning_rate": 1.858e-05,
|
| 368 |
+
"loss": 0.3459,
|
| 369 |
+
"step": 8200
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 1.8166089965397925,
|
| 373 |
+
"grad_norm": 12.204736709594727,
|
| 374 |
+
"learning_rate": 1.8491111111111112e-05,
|
| 375 |
+
"loss": 0.3592,
|
| 376 |
+
"step": 8400
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"epoch": 1.8598615916955017,
|
| 380 |
+
"grad_norm": 8.272552490234375,
|
| 381 |
+
"learning_rate": 1.8402222222222223e-05,
|
| 382 |
+
"loss": 0.338,
|
| 383 |
+
"step": 8600
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 1.903114186851211,
|
| 387 |
+
"grad_norm": 11.011015892028809,
|
| 388 |
+
"learning_rate": 1.8313333333333333e-05,
|
| 389 |
+
"loss": 0.3503,
|
| 390 |
+
"step": 8800
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 1.9463667820069204,
|
| 394 |
+
"grad_norm": 11.080639839172363,
|
| 395 |
+
"learning_rate": 1.8224444444444447e-05,
|
| 396 |
+
"loss": 0.3433,
|
| 397 |
+
"step": 9000
|
| 398 |
+
},
|
| 399 |
+
{
|
| 400 |
+
"epoch": 1.9463667820069204,
|
| 401 |
+
"eval_loss": 0.3173498213291168,
|
| 402 |
+
"eval_runtime": 1578.1615,
|
| 403 |
+
"eval_samples_per_second": 5.873,
|
| 404 |
+
"eval_steps_per_second": 1.468,
|
| 405 |
+
"eval_wer": 0.25005845765327595,
|
| 406 |
+
"step": 9000
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 1.9896193771626298,
|
| 410 |
+
"grad_norm": 10.65349006652832,
|
| 411 |
+
"learning_rate": 1.8135555555555557e-05,
|
| 412 |
+
"loss": 0.3353,
|
| 413 |
+
"step": 9200
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 2.0328719723183393,
|
| 417 |
+
"grad_norm": 8.191920280456543,
|
| 418 |
+
"learning_rate": 1.8047111111111114e-05,
|
| 419 |
+
"loss": 0.2603,
|
| 420 |
+
"step": 9400
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 2.0761245674740483,
|
| 424 |
+
"grad_norm": 6.512417793273926,
|
| 425 |
+
"learning_rate": 1.7958222222222225e-05,
|
| 426 |
+
"loss": 0.2309,
|
| 427 |
+
"step": 9600
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 2.1193771626297577,
|
| 431 |
+
"grad_norm": 7.245334148406982,
|
| 432 |
+
"learning_rate": 1.7869333333333335e-05,
|
| 433 |
+
"loss": 0.2298,
|
| 434 |
+
"step": 9800
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 2.162629757785467,
|
| 438 |
+
"grad_norm": 7.289510726928711,
|
| 439 |
+
"learning_rate": 1.7780444444444446e-05,
|
| 440 |
+
"loss": 0.2336,
|
| 441 |
+
"step": 10000
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 2.162629757785467,
|
| 445 |
+
"eval_loss": 0.31440219283103943,
|
| 446 |
+
"eval_runtime": 1571.2049,
|
| 447 |
+
"eval_samples_per_second": 5.899,
|
| 448 |
+
"eval_steps_per_second": 1.475,
|
| 449 |
+
"eval_wer": 0.25627835196183885,
|
| 450 |
+
"step": 10000
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 2.2058823529411766,
|
| 454 |
+
"grad_norm": 8.776716232299805,
|
| 455 |
+
"learning_rate": 1.7691555555555556e-05,
|
| 456 |
+
"loss": 0.2304,
|
| 457 |
+
"step": 10200
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 2.2491349480968856,
|
| 461 |
+
"grad_norm": 14.445012092590332,
|
| 462 |
+
"learning_rate": 1.7602666666666667e-05,
|
| 463 |
+
"loss": 0.2328,
|
| 464 |
+
"step": 10400
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 2.292387543252595,
|
| 468 |
+
"grad_norm": 7.246095180511475,
|
| 469 |
+
"learning_rate": 1.7513777777777777e-05,
|
| 470 |
+
"loss": 0.2306,
|
| 471 |
+
"step": 10600
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 2.3356401384083045,
|
| 475 |
+
"grad_norm": 7.7789201736450195,
|
| 476 |
+
"learning_rate": 1.742488888888889e-05,
|
| 477 |
+
"loss": 0.2231,
|
| 478 |
+
"step": 10800
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 2.378892733564014,
|
| 482 |
+
"grad_norm": 8.133916854858398,
|
| 483 |
+
"learning_rate": 1.7336e-05,
|
| 484 |
+
"loss": 0.2238,
|
| 485 |
+
"step": 11000
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 2.378892733564014,
|
| 489 |
+
"eval_loss": 0.3043256402015686,
|
| 490 |
+
"eval_runtime": 1653.7123,
|
| 491 |
+
"eval_samples_per_second": 5.604,
|
| 492 |
+
"eval_steps_per_second": 1.401,
|
| 493 |
+
"eval_wer": 0.23548301610313488,
|
| 494 |
+
"step": 11000
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 2.422145328719723,
|
| 498 |
+
"grad_norm": 8.60897445678711,
|
| 499 |
+
"learning_rate": 1.7247111111111112e-05,
|
| 500 |
+
"loss": 0.2267,
|
| 501 |
+
"step": 11200
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 2.4653979238754324,
|
| 505 |
+
"grad_norm": 8.54346752166748,
|
| 506 |
+
"learning_rate": 1.7158222222222222e-05,
|
| 507 |
+
"loss": 0.2284,
|
| 508 |
+
"step": 11400
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 2.508650519031142,
|
| 512 |
+
"grad_norm": 9.328478813171387,
|
| 513 |
+
"learning_rate": 1.7069333333333336e-05,
|
| 514 |
+
"loss": 0.2312,
|
| 515 |
+
"step": 11600
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 2.5519031141868513,
|
| 519 |
+
"grad_norm": 8.939164161682129,
|
| 520 |
+
"learning_rate": 1.6980444444444447e-05,
|
| 521 |
+
"loss": 0.2242,
|
| 522 |
+
"step": 11800
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 2.595155709342561,
|
| 526 |
+
"grad_norm": 8.84998893737793,
|
| 527 |
+
"learning_rate": 1.6891555555555557e-05,
|
| 528 |
+
"loss": 0.2225,
|
| 529 |
+
"step": 12000
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 2.595155709342561,
|
| 533 |
+
"eval_loss": 0.29693683981895447,
|
| 534 |
+
"eval_runtime": 1661.0721,
|
| 535 |
+
"eval_samples_per_second": 5.58,
|
| 536 |
+
"eval_steps_per_second": 1.395,
|
| 537 |
+
"eval_wer": 0.2402531606104538,
|
| 538 |
+
"step": 12000
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 2.63840830449827,
|
| 542 |
+
"grad_norm": 9.050106048583984,
|
| 543 |
+
"learning_rate": 1.6802666666666668e-05,
|
| 544 |
+
"loss": 0.2317,
|
| 545 |
+
"step": 12200
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 2.6816608996539792,
|
| 549 |
+
"grad_norm": 8.452465057373047,
|
| 550 |
+
"learning_rate": 1.671377777777778e-05,
|
| 551 |
+
"loss": 0.2217,
|
| 552 |
+
"step": 12400
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 2.7249134948096887,
|
| 556 |
+
"grad_norm": 6.295560836791992,
|
| 557 |
+
"learning_rate": 1.6624888888888892e-05,
|
| 558 |
+
"loss": 0.2173,
|
| 559 |
+
"step": 12600
|
| 560 |
+
},
|
| 561 |
+
{
|
| 562 |
+
"epoch": 2.7681660899653977,
|
| 563 |
+
"grad_norm": 8.163686752319336,
|
| 564 |
+
"learning_rate": 1.6536000000000002e-05,
|
| 565 |
+
"loss": 0.2262,
|
| 566 |
+
"step": 12800
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 2.811418685121107,
|
| 570 |
+
"grad_norm": 7.768032550811768,
|
| 571 |
+
"learning_rate": 1.6447111111111113e-05,
|
| 572 |
+
"loss": 0.218,
|
| 573 |
+
"step": 13000
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 2.811418685121107,
|
| 577 |
+
"eval_loss": 0.2880776524543762,
|
| 578 |
+
"eval_runtime": 1675.1405,
|
| 579 |
+
"eval_samples_per_second": 5.533,
|
| 580 |
+
"eval_steps_per_second": 1.383,
|
| 581 |
+
"eval_wer": 0.2325991052081872,
|
| 582 |
+
"step": 13000
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 2.8546712802768166,
|
| 586 |
+
"grad_norm": 7.9054388999938965,
|
| 587 |
+
"learning_rate": 1.6358666666666666e-05,
|
| 588 |
+
"loss": 0.2201,
|
| 589 |
+
"step": 13200
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 2.897923875432526,
|
| 593 |
+
"grad_norm": 10.126465797424316,
|
| 594 |
+
"learning_rate": 1.626977777777778e-05,
|
| 595 |
+
"loss": 0.2168,
|
| 596 |
+
"step": 13400
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 2.9411764705882355,
|
| 600 |
+
"grad_norm": 6.711848735809326,
|
| 601 |
+
"learning_rate": 1.618088888888889e-05,
|
| 602 |
+
"loss": 0.225,
|
| 603 |
+
"step": 13600
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 2.9844290657439445,
|
| 607 |
+
"grad_norm": 5.138487339019775,
|
| 608 |
+
"learning_rate": 1.6092e-05,
|
| 609 |
+
"loss": 0.2148,
|
| 610 |
+
"step": 13800
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"epoch": 3.027681660899654,
|
| 614 |
+
"grad_norm": 15.961743354797363,
|
| 615 |
+
"learning_rate": 1.600311111111111e-05,
|
| 616 |
+
"loss": 0.1778,
|
| 617 |
+
"step": 14000
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 3.027681660899654,
|
| 621 |
+
"eval_loss": 0.2847885191440582,
|
| 622 |
+
"eval_runtime": 1615.0853,
|
| 623 |
+
"eval_samples_per_second": 5.738,
|
| 624 |
+
"eval_steps_per_second": 1.435,
|
| 625 |
+
"eval_wer": 0.21417325289560243,
|
| 626 |
+
"step": 14000
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 3.0709342560553634,
|
| 630 |
+
"grad_norm": 7.719923973083496,
|
| 631 |
+
"learning_rate": 1.5914222222222225e-05,
|
| 632 |
+
"loss": 0.1494,
|
| 633 |
+
"step": 14200
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 3.114186851211073,
|
| 637 |
+
"grad_norm": 4.328163146972656,
|
| 638 |
+
"learning_rate": 1.5825333333333336e-05,
|
| 639 |
+
"loss": 0.1561,
|
| 640 |
+
"step": 14400
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 3.157439446366782,
|
| 644 |
+
"grad_norm": 7.591590881347656,
|
| 645 |
+
"learning_rate": 1.5736444444444446e-05,
|
| 646 |
+
"loss": 0.1585,
|
| 647 |
+
"step": 14600
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 3.2006920415224913,
|
| 651 |
+
"grad_norm": 6.677422523498535,
|
| 652 |
+
"learning_rate": 1.5647555555555557e-05,
|
| 653 |
+
"loss": 0.163,
|
| 654 |
+
"step": 14800
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 3.2439446366782008,
|
| 658 |
+
"grad_norm": 8.380745887756348,
|
| 659 |
+
"learning_rate": 1.5558666666666667e-05,
|
| 660 |
+
"loss": 0.1669,
|
| 661 |
+
"step": 15000
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 3.2439446366782008,
|
| 665 |
+
"eval_loss": 0.28238075971603394,
|
| 666 |
+
"eval_runtime": 1623.36,
|
| 667 |
+
"eval_samples_per_second": 5.709,
|
| 668 |
+
"eval_steps_per_second": 1.427,
|
| 669 |
+
"eval_wer": 0.21135169683081576,
|
| 670 |
+
"step": 15000
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"epoch": 3.28719723183391,
|
| 674 |
+
"grad_norm": 7.155370235443115,
|
| 675 |
+
"learning_rate": 1.5469777777777778e-05,
|
| 676 |
+
"loss": 0.1541,
|
| 677 |
+
"step": 15200
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 3.330449826989619,
|
| 681 |
+
"grad_norm": 8.143542289733887,
|
| 682 |
+
"learning_rate": 1.5380888888888888e-05,
|
| 683 |
+
"loss": 0.1635,
|
| 684 |
+
"step": 15400
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 3.3737024221453287,
|
| 688 |
+
"grad_norm": 5.822775840759277,
|
| 689 |
+
"learning_rate": 1.5292e-05,
|
| 690 |
+
"loss": 0.1563,
|
| 691 |
+
"step": 15600
|
| 692 |
+
},
|
| 693 |
+
{
|
| 694 |
+
"epoch": 3.416955017301038,
|
| 695 |
+
"grad_norm": 6.631377696990967,
|
| 696 |
+
"learning_rate": 1.5203111111111112e-05,
|
| 697 |
+
"loss": 0.1579,
|
| 698 |
+
"step": 15800
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 3.4602076124567476,
|
| 702 |
+
"grad_norm": 5.501554012298584,
|
| 703 |
+
"learning_rate": 1.5114222222222223e-05,
|
| 704 |
+
"loss": 0.1621,
|
| 705 |
+
"step": 16000
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 3.4602076124567476,
|
| 709 |
+
"eval_loss": 0.28122928738594055,
|
| 710 |
+
"eval_runtime": 1660.3516,
|
| 711 |
+
"eval_samples_per_second": 5.582,
|
| 712 |
+
"eval_steps_per_second": 1.395,
|
| 713 |
+
"eval_wer": 0.21305086595270387,
|
| 714 |
+
"step": 16000
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 3.5034602076124566,
|
| 718 |
+
"grad_norm": 6.517296314239502,
|
| 719 |
+
"learning_rate": 1.5025333333333333e-05,
|
| 720 |
+
"loss": 0.1682,
|
| 721 |
+
"step": 16200
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 3.546712802768166,
|
| 725 |
+
"grad_norm": 8.255217552185059,
|
| 726 |
+
"learning_rate": 1.4936444444444447e-05,
|
| 727 |
+
"loss": 0.1578,
|
| 728 |
+
"step": 16400
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 3.5899653979238755,
|
| 732 |
+
"grad_norm": 8.096115112304688,
|
| 733 |
+
"learning_rate": 1.4847555555555558e-05,
|
| 734 |
+
"loss": 0.1595,
|
| 735 |
+
"step": 16600
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 3.633217993079585,
|
| 739 |
+
"grad_norm": 6.509264945983887,
|
| 740 |
+
"learning_rate": 1.4758666666666668e-05,
|
| 741 |
+
"loss": 0.1531,
|
| 742 |
+
"step": 16800
|
| 743 |
+
},
|
| 744 |
+
{
|
| 745 |
+
"epoch": 3.6764705882352944,
|
| 746 |
+
"grad_norm": 7.939600944519043,
|
| 747 |
+
"learning_rate": 1.4669777777777779e-05,
|
| 748 |
+
"loss": 0.1585,
|
| 749 |
+
"step": 17000
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 3.6764705882352944,
|
| 753 |
+
"eval_loss": 0.2753097414970398,
|
| 754 |
+
"eval_runtime": 1673.4744,
|
| 755 |
+
"eval_samples_per_second": 5.538,
|
| 756 |
+
"eval_steps_per_second": 1.385,
|
| 757 |
+
"eval_wer": 0.21139846295343653,
|
| 758 |
+
"step": 17000
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 3.7197231833910034,
|
| 762 |
+
"grad_norm": 8.923968315124512,
|
| 763 |
+
"learning_rate": 1.4581333333333334e-05,
|
| 764 |
+
"loss": 0.1542,
|
| 765 |
+
"step": 17200
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 3.762975778546713,
|
| 769 |
+
"grad_norm": 5.017932891845703,
|
| 770 |
+
"learning_rate": 1.4492444444444444e-05,
|
| 771 |
+
"loss": 0.1606,
|
| 772 |
+
"step": 17400
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 3.8062283737024223,
|
| 776 |
+
"grad_norm": 5.521350383758545,
|
| 777 |
+
"learning_rate": 1.4403555555555556e-05,
|
| 778 |
+
"loss": 0.1514,
|
| 779 |
+
"step": 17600
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 3.8494809688581313,
|
| 783 |
+
"grad_norm": 9.742095947265625,
|
| 784 |
+
"learning_rate": 1.4314666666666669e-05,
|
| 785 |
+
"loss": 0.1605,
|
| 786 |
+
"step": 17800
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 3.8927335640138407,
|
| 790 |
+
"grad_norm": 7.891471862792969,
|
| 791 |
+
"learning_rate": 1.4225777777777779e-05,
|
| 792 |
+
"loss": 0.1567,
|
| 793 |
+
"step": 18000
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 3.8927335640138407,
|
| 797 |
+
"eval_loss": 0.27232253551483154,
|
| 798 |
+
"eval_runtime": 1700.8958,
|
| 799 |
+
"eval_samples_per_second": 5.449,
|
| 800 |
+
"eval_steps_per_second": 1.362,
|
| 801 |
+
"eval_wer": 0.19725950521442268,
|
| 802 |
+
"step": 18000
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 3.93598615916955,
|
| 806 |
+
"grad_norm": 7.626640319824219,
|
| 807 |
+
"learning_rate": 1.4136888888888891e-05,
|
| 808 |
+
"loss": 0.1614,
|
| 809 |
+
"step": 18200
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 3.9792387543252596,
|
| 813 |
+
"grad_norm": 6.852673053741455,
|
| 814 |
+
"learning_rate": 1.4048000000000002e-05,
|
| 815 |
+
"loss": 0.1591,
|
| 816 |
+
"step": 18400
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 4.022491349480969,
|
| 820 |
+
"grad_norm": 8.06204605102539,
|
| 821 |
+
"learning_rate": 1.3959111111111112e-05,
|
| 822 |
+
"loss": 0.1323,
|
| 823 |
+
"step": 18600
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 4.0657439446366785,
|
| 827 |
+
"grad_norm": 6.888394355773926,
|
| 828 |
+
"learning_rate": 1.3870222222222223e-05,
|
| 829 |
+
"loss": 0.1097,
|
| 830 |
+
"step": 18800
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 4.108996539792388,
|
| 834 |
+
"grad_norm": 6.363215446472168,
|
| 835 |
+
"learning_rate": 1.3781333333333335e-05,
|
| 836 |
+
"loss": 0.1092,
|
| 837 |
+
"step": 19000
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 4.108996539792388,
|
| 841 |
+
"eval_loss": 0.2706485986709595,
|
| 842 |
+
"eval_runtime": 1679.429,
|
| 843 |
+
"eval_samples_per_second": 5.519,
|
| 844 |
+
"eval_steps_per_second": 1.38,
|
| 845 |
+
"eval_wer": 0.20047077896771578,
|
| 846 |
+
"step": 19000
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 4.1522491349480966,
|
| 850 |
+
"grad_norm": 6.086292743682861,
|
| 851 |
+
"learning_rate": 1.3692444444444445e-05,
|
| 852 |
+
"loss": 0.1131,
|
| 853 |
+
"step": 19200
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 4.195501730103806,
|
| 857 |
+
"grad_norm": 7.441946983337402,
|
| 858 |
+
"learning_rate": 1.3603555555555556e-05,
|
| 859 |
+
"loss": 0.1118,
|
| 860 |
+
"step": 19400
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 4.2387543252595155,
|
| 864 |
+
"grad_norm": 5.773116588592529,
|
| 865 |
+
"learning_rate": 1.3514666666666668e-05,
|
| 866 |
+
"loss": 0.1161,
|
| 867 |
+
"step": 19600
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 4.282006920415225,
|
| 871 |
+
"grad_norm": 5.2627081871032715,
|
| 872 |
+
"learning_rate": 1.342577777777778e-05,
|
| 873 |
+
"loss": 0.1142,
|
| 874 |
+
"step": 19800
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 4.325259515570934,
|
| 878 |
+
"grad_norm": 5.706352233886719,
|
| 879 |
+
"learning_rate": 1.333688888888889e-05,
|
| 880 |
+
"loss": 0.1122,
|
| 881 |
+
"step": 20000
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 4.325259515570934,
|
| 885 |
+
"eval_loss": 0.27044767141342163,
|
| 886 |
+
"eval_runtime": 1698.3705,
|
| 887 |
+
"eval_samples_per_second": 5.457,
|
| 888 |
+
"eval_steps_per_second": 1.364,
|
| 889 |
+
"eval_wer": 0.20921604389780044,
|
| 890 |
+
"step": 20000
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 4.368512110726644,
|
| 894 |
+
"grad_norm": 5.661611557006836,
|
| 895 |
+
"learning_rate": 1.3248000000000001e-05,
|
| 896 |
+
"loss": 0.1104,
|
| 897 |
+
"step": 20200
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 4.411764705882353,
|
| 901 |
+
"grad_norm": 4.871212959289551,
|
| 902 |
+
"learning_rate": 1.3159111111111111e-05,
|
| 903 |
+
"loss": 0.1117,
|
| 904 |
+
"step": 20400
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 4.455017301038062,
|
| 908 |
+
"grad_norm": 4.58662223815918,
|
| 909 |
+
"learning_rate": 1.3070222222222223e-05,
|
| 910 |
+
"loss": 0.1142,
|
| 911 |
+
"step": 20600
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 4.498269896193771,
|
| 915 |
+
"grad_norm": 7.396908760070801,
|
| 916 |
+
"learning_rate": 1.2981333333333334e-05,
|
| 917 |
+
"loss": 0.1178,
|
| 918 |
+
"step": 20800
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 4.541522491349481,
|
| 922 |
+
"grad_norm": 4.963231086730957,
|
| 923 |
+
"learning_rate": 1.2892444444444444e-05,
|
| 924 |
+
"loss": 0.1138,
|
| 925 |
+
"step": 21000
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 4.541522491349481,
|
| 929 |
+
"eval_loss": 0.27063047885894775,
|
| 930 |
+
"eval_runtime": 1664.8229,
|
| 931 |
+
"eval_samples_per_second": 5.567,
|
| 932 |
+
"eval_steps_per_second": 1.392,
|
| 933 |
+
"eval_wer": 0.1959188763659605,
|
| 934 |
+
"step": 21000
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 4.58477508650519,
|
| 938 |
+
"grad_norm": 5.799412727355957,
|
| 939 |
+
"learning_rate": 1.2803555555555557e-05,
|
| 940 |
+
"loss": 0.1135,
|
| 941 |
+
"step": 21200
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 4.6280276816609,
|
| 945 |
+
"grad_norm": 10.018841743469238,
|
| 946 |
+
"learning_rate": 1.2715111111111112e-05,
|
| 947 |
+
"loss": 0.1138,
|
| 948 |
+
"step": 21400
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 4.671280276816609,
|
| 952 |
+
"grad_norm": 11.115982055664062,
|
| 953 |
+
"learning_rate": 1.2626222222222224e-05,
|
| 954 |
+
"loss": 0.1173,
|
| 955 |
+
"step": 21600
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 4.7145328719723185,
|
| 959 |
+
"grad_norm": 6.210891246795654,
|
| 960 |
+
"learning_rate": 1.2537333333333334e-05,
|
| 961 |
+
"loss": 0.116,
|
| 962 |
+
"step": 21800
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 4.757785467128028,
|
| 966 |
+
"grad_norm": 8.309006690979004,
|
| 967 |
+
"learning_rate": 1.2448444444444445e-05,
|
| 968 |
+
"loss": 0.121,
|
| 969 |
+
"step": 22000
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 4.757785467128028,
|
| 973 |
+
"eval_loss": 0.2650456726551056,
|
| 974 |
+
"eval_runtime": 1684.8264,
|
| 975 |
+
"eval_samples_per_second": 5.501,
|
| 976 |
+
"eval_steps_per_second": 1.375,
|
| 977 |
+
"eval_wer": 0.19517061840402813,
|
| 978 |
+
"step": 22000
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 4.801038062283737,
|
| 982 |
+
"grad_norm": 6.640094757080078,
|
| 983 |
+
"learning_rate": 1.2359555555555555e-05,
|
| 984 |
+
"loss": 0.1182,
|
| 985 |
+
"step": 22200
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 4.844290657439446,
|
| 989 |
+
"grad_norm": 8.134920120239258,
|
| 990 |
+
"learning_rate": 1.227066666666667e-05,
|
| 991 |
+
"loss": 0.1136,
|
| 992 |
+
"step": 22400
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 4.887543252595155,
|
| 996 |
+
"grad_norm": 7.265510082244873,
|
| 997 |
+
"learning_rate": 1.218177777777778e-05,
|
| 998 |
+
"loss": 0.1225,
|
| 999 |
+
"step": 22600
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 4.930795847750865,
|
| 1003 |
+
"grad_norm": 8.816773414611816,
|
| 1004 |
+
"learning_rate": 1.209288888888889e-05,
|
| 1005 |
+
"loss": 0.118,
|
| 1006 |
+
"step": 22800
|
| 1007 |
+
},
|
| 1008 |
+
{
|
| 1009 |
+
"epoch": 4.974048442906574,
|
| 1010 |
+
"grad_norm": 8.387493133544922,
|
| 1011 |
+
"learning_rate": 1.2004e-05,
|
| 1012 |
+
"loss": 0.11,
|
| 1013 |
+
"step": 23000
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 4.974048442906574,
|
| 1017 |
+
"eval_loss": 0.26424792408943176,
|
| 1018 |
+
"eval_runtime": 1682.4354,
|
| 1019 |
+
"eval_samples_per_second": 5.509,
|
| 1020 |
+
"eval_steps_per_second": 1.377,
|
| 1021 |
+
"eval_wer": 0.1935182154047608,
|
| 1022 |
+
"step": 23000
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 5.017301038062284,
|
| 1026 |
+
"grad_norm": 6.013669013977051,
|
| 1027 |
+
"learning_rate": 1.1915111111111113e-05,
|
| 1028 |
+
"loss": 0.1005,
|
| 1029 |
+
"step": 23200
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 5.060553633217993,
|
| 1033 |
+
"grad_norm": 5.462082862854004,
|
| 1034 |
+
"learning_rate": 1.1826222222222223e-05,
|
| 1035 |
+
"loss": 0.0816,
|
| 1036 |
+
"step": 23400
|
| 1037 |
+
},
|
| 1038 |
+
{
|
| 1039 |
+
"epoch": 5.103806228373703,
|
| 1040 |
+
"grad_norm": 4.537853717803955,
|
| 1041 |
+
"learning_rate": 1.1737333333333334e-05,
|
| 1042 |
+
"loss": 0.0789,
|
| 1043 |
+
"step": 23600
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 5.147058823529412,
|
| 1047 |
+
"grad_norm": 3.594006299972534,
|
| 1048 |
+
"learning_rate": 1.1648444444444444e-05,
|
| 1049 |
+
"loss": 0.0779,
|
| 1050 |
+
"step": 23800
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 5.190311418685121,
|
| 1054 |
+
"grad_norm": 4.96687650680542,
|
| 1055 |
+
"learning_rate": 1.1559555555555558e-05,
|
| 1056 |
+
"loss": 0.0848,
|
| 1057 |
+
"step": 24000
|
| 1058 |
+
},
|
| 1059 |
+
{
|
| 1060 |
+
"epoch": 5.190311418685121,
|
| 1061 |
+
"eval_loss": 0.2654721140861511,
|
| 1062 |
+
"eval_runtime": 1692.4989,
|
| 1063 |
+
"eval_samples_per_second": 5.476,
|
| 1064 |
+
"eval_steps_per_second": 1.369,
|
| 1065 |
+
"eval_wer": 0.1916008043773091,
|
| 1066 |
+
"step": 24000
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 5.23356401384083,
|
| 1070 |
+
"grad_norm": 6.847875595092773,
|
| 1071 |
+
"learning_rate": 1.1471111111111113e-05,
|
| 1072 |
+
"loss": 0.0818,
|
| 1073 |
+
"step": 24200
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 5.27681660899654,
|
| 1077 |
+
"grad_norm": 6.794471263885498,
|
| 1078 |
+
"learning_rate": 1.1382222222222224e-05,
|
| 1079 |
+
"loss": 0.0823,
|
| 1080 |
+
"step": 24400
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 5.320069204152249,
|
| 1084 |
+
"grad_norm": 5.965904712677002,
|
| 1085 |
+
"learning_rate": 1.1293333333333334e-05,
|
| 1086 |
+
"loss": 0.0819,
|
| 1087 |
+
"step": 24600
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 5.3633217993079585,
|
| 1091 |
+
"grad_norm": 5.61757230758667,
|
| 1092 |
+
"learning_rate": 1.1204444444444445e-05,
|
| 1093 |
+
"loss": 0.0856,
|
| 1094 |
+
"step": 24800
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 5.406574394463668,
|
| 1098 |
+
"grad_norm": 6.113648891448975,
|
| 1099 |
+
"learning_rate": 1.1115555555555557e-05,
|
| 1100 |
+
"loss": 0.0844,
|
| 1101 |
+
"step": 25000
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 5.406574394463668,
|
| 1105 |
+
"eval_loss": 0.26443904638290405,
|
| 1106 |
+
"eval_runtime": 1715.5202,
|
| 1107 |
+
"eval_samples_per_second": 5.402,
|
| 1108 |
+
"eval_steps_per_second": 1.351,
|
| 1109 |
+
"eval_wer": 0.18901307892562627,
|
| 1110 |
+
"step": 25000
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"epoch": 5.449826989619377,
|
| 1114 |
+
"grad_norm": 5.7227983474731445,
|
| 1115 |
+
"learning_rate": 1.1026666666666667e-05,
|
| 1116 |
+
"loss": 0.0852,
|
| 1117 |
+
"step": 25200
|
| 1118 |
+
},
|
| 1119 |
+
{
|
| 1120 |
+
"epoch": 5.493079584775087,
|
| 1121 |
+
"grad_norm": 4.039446830749512,
|
| 1122 |
+
"learning_rate": 1.0937777777777778e-05,
|
| 1123 |
+
"loss": 0.0875,
|
| 1124 |
+
"step": 25400
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 5.536332179930795,
|
| 1128 |
+
"grad_norm": 6.3002400398254395,
|
| 1129 |
+
"learning_rate": 1.0849333333333335e-05,
|
| 1130 |
+
"loss": 0.0853,
|
| 1131 |
+
"step": 25600
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 5.579584775086505,
|
| 1135 |
+
"grad_norm": 4.92480993270874,
|
| 1136 |
+
"learning_rate": 1.0760444444444445e-05,
|
| 1137 |
+
"loss": 0.0856,
|
| 1138 |
+
"step": 25800
|
| 1139 |
+
},
|
| 1140 |
+
{
|
| 1141 |
+
"epoch": 5.622837370242214,
|
| 1142 |
+
"grad_norm": 5.624736309051514,
|
| 1143 |
+
"learning_rate": 1.0671555555555557e-05,
|
| 1144 |
+
"loss": 0.0836,
|
| 1145 |
+
"step": 26000
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 5.622837370242214,
|
| 1149 |
+
"eval_loss": 0.2625565826892853,
|
| 1150 |
+
"eval_runtime": 1696.592,
|
| 1151 |
+
"eval_samples_per_second": 5.463,
|
| 1152 |
+
"eval_steps_per_second": 1.366,
|
| 1153 |
+
"eval_wer": 0.19049400614195078,
|
| 1154 |
+
"step": 26000
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 5.666089965397924,
|
| 1158 |
+
"grad_norm": 5.235467910766602,
|
| 1159 |
+
"learning_rate": 1.0582666666666668e-05,
|
| 1160 |
+
"loss": 0.0853,
|
| 1161 |
+
"step": 26200
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 5.709342560553633,
|
| 1165 |
+
"grad_norm": 7.8722147941589355,
|
| 1166 |
+
"learning_rate": 1.0493777777777778e-05,
|
| 1167 |
+
"loss": 0.0832,
|
| 1168 |
+
"step": 26400
|
| 1169 |
+
},
|
| 1170 |
+
{
|
| 1171 |
+
"epoch": 5.752595155709343,
|
| 1172 |
+
"grad_norm": 6.0298752784729,
|
| 1173 |
+
"learning_rate": 1.0404888888888889e-05,
|
| 1174 |
+
"loss": 0.0888,
|
| 1175 |
+
"step": 26600
|
| 1176 |
+
},
|
| 1177 |
+
{
|
| 1178 |
+
"epoch": 5.795847750865052,
|
| 1179 |
+
"grad_norm": 4.088235378265381,
|
| 1180 |
+
"learning_rate": 1.0316e-05,
|
| 1181 |
+
"loss": 0.0808,
|
| 1182 |
+
"step": 26800
|
| 1183 |
+
},
|
| 1184 |
+
{
|
| 1185 |
+
"epoch": 5.839100346020762,
|
| 1186 |
+
"grad_norm": 5.005467891693115,
|
| 1187 |
+
"learning_rate": 1.0227111111111111e-05,
|
| 1188 |
+
"loss": 0.087,
|
| 1189 |
+
"step": 27000
|
| 1190 |
+
},
|
| 1191 |
+
{
|
| 1192 |
+
"epoch": 5.839100346020762,
|
| 1193 |
+
"eval_loss": 0.2586536407470703,
|
| 1194 |
+
"eval_runtime": 1722.7058,
|
| 1195 |
+
"eval_samples_per_second": 5.38,
|
| 1196 |
+
"eval_steps_per_second": 1.345,
|
| 1197 |
+
"eval_wer": 0.1884830628692575,
|
| 1198 |
+
"step": 27000
|
| 1199 |
+
},
|
| 1200 |
+
{
|
| 1201 |
+
"epoch": 5.882352941176471,
|
| 1202 |
+
"grad_norm": 7.082780838012695,
|
| 1203 |
+
"learning_rate": 1.0138222222222223e-05,
|
| 1204 |
+
"loss": 0.0845,
|
| 1205 |
+
"step": 27200
|
| 1206 |
+
},
|
| 1207 |
+
{
|
| 1208 |
+
"epoch": 5.92560553633218,
|
| 1209 |
+
"grad_norm": 5.2374982833862305,
|
| 1210 |
+
"learning_rate": 1.0049333333333334e-05,
|
| 1211 |
+
"loss": 0.0863,
|
| 1212 |
+
"step": 27400
|
| 1213 |
+
},
|
| 1214 |
+
{
|
| 1215 |
+
"epoch": 5.968858131487889,
|
| 1216 |
+
"grad_norm": 3.091181516647339,
|
| 1217 |
+
"learning_rate": 9.960444444444444e-06,
|
| 1218 |
+
"loss": 0.0858,
|
| 1219 |
+
"step": 27600
|
| 1220 |
+
},
|
| 1221 |
+
{
|
| 1222 |
+
"epoch": 6.0121107266435985,
|
| 1223 |
+
"grad_norm": 3.6232330799102783,
|
| 1224 |
+
"learning_rate": 9.871555555555557e-06,
|
| 1225 |
+
"loss": 0.078,
|
| 1226 |
+
"step": 27800
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 6.055363321799308,
|
| 1230 |
+
"grad_norm": 6.693848133087158,
|
| 1231 |
+
"learning_rate": 9.782666666666667e-06,
|
| 1232 |
+
"loss": 0.059,
|
| 1233 |
+
"step": 28000
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 6.055363321799308,
|
| 1237 |
+
"eval_loss": 0.25941041111946106,
|
| 1238 |
+
"eval_runtime": 1696.975,
|
| 1239 |
+
"eval_samples_per_second": 5.461,
|
| 1240 |
+
"eval_steps_per_second": 1.365,
|
| 1241 |
+
"eval_wer": 0.18268406366428158,
|
| 1242 |
+
"step": 28000
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 6.098615916955017,
|
| 1246 |
+
"grad_norm": 3.9847519397735596,
|
| 1247 |
+
"learning_rate": 9.693777777777779e-06,
|
| 1248 |
+
"loss": 0.0611,
|
| 1249 |
+
"step": 28200
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 6.141868512110727,
|
| 1253 |
+
"grad_norm": 4.18721342086792,
|
| 1254 |
+
"learning_rate": 9.60488888888889e-06,
|
| 1255 |
+
"loss": 0.0596,
|
| 1256 |
+
"step": 28400
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 6.185121107266436,
|
| 1260 |
+
"grad_norm": 3.4485201835632324,
|
| 1261 |
+
"learning_rate": 9.516e-06,
|
| 1262 |
+
"loss": 0.0599,
|
| 1263 |
+
"step": 28600
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 6.228373702422146,
|
| 1267 |
+
"grad_norm": 4.734513759613037,
|
| 1268 |
+
"learning_rate": 9.427111111111112e-06,
|
| 1269 |
+
"loss": 0.0637,
|
| 1270 |
+
"step": 28800
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 6.271626297577854,
|
| 1274 |
+
"grad_norm": 5.914642333984375,
|
| 1275 |
+
"learning_rate": 9.338222222222223e-06,
|
| 1276 |
+
"loss": 0.0596,
|
| 1277 |
+
"step": 29000
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 6.271626297577854,
|
| 1281 |
+
"eval_loss": 0.2605881690979004,
|
| 1282 |
+
"eval_runtime": 1692.8882,
|
| 1283 |
+
"eval_samples_per_second": 5.475,
|
| 1284 |
+
"eval_steps_per_second": 1.369,
|
| 1285 |
+
"eval_wer": 0.18354144257899577,
|
| 1286 |
+
"step": 29000
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 6.314878892733564,
|
| 1290 |
+
"grad_norm": 4.263479709625244,
|
| 1291 |
+
"learning_rate": 9.249333333333335e-06,
|
| 1292 |
+
"loss": 0.0628,
|
| 1293 |
+
"step": 29200
|
| 1294 |
+
},
|
| 1295 |
+
{
|
| 1296 |
+
"epoch": 6.358131487889273,
|
| 1297 |
+
"grad_norm": 4.9372735023498535,
|
| 1298 |
+
"learning_rate": 9.160444444444445e-06,
|
| 1299 |
+
"loss": 0.0588,
|
| 1300 |
+
"step": 29400
|
| 1301 |
+
},
|
| 1302 |
+
{
|
| 1303 |
+
"epoch": 6.401384083044983,
|
| 1304 |
+
"grad_norm": 3.16778302192688,
|
| 1305 |
+
"learning_rate": 9.071555555555557e-06,
|
| 1306 |
+
"loss": 0.0635,
|
| 1307 |
+
"step": 29600
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 6.444636678200692,
|
| 1311 |
+
"grad_norm": 4.998132705688477,
|
| 1312 |
+
"learning_rate": 8.982666666666668e-06,
|
| 1313 |
+
"loss": 0.0613,
|
| 1314 |
+
"step": 29800
|
| 1315 |
+
},
|
| 1316 |
+
{
|
| 1317 |
+
"epoch": 6.4878892733564015,
|
| 1318 |
+
"grad_norm": 2.108086585998535,
|
| 1319 |
+
"learning_rate": 8.893777777777778e-06,
|
| 1320 |
+
"loss": 0.0616,
|
| 1321 |
+
"step": 30000
|
| 1322 |
+
},
|
| 1323 |
+
{
|
| 1324 |
+
"epoch": 6.4878892733564015,
|
| 1325 |
+
"eval_loss": 0.25865623354911804,
|
| 1326 |
+
"eval_runtime": 1695.2212,
|
| 1327 |
+
"eval_samples_per_second": 5.467,
|
| 1328 |
+
"eval_steps_per_second": 1.367,
|
| 1329 |
+
"eval_wer": 0.18951191756691452,
|
| 1330 |
+
"step": 30000
|
| 1331 |
+
},
|
| 1332 |
+
{
|
| 1333 |
+
"epoch": 6.531141868512111,
|
| 1334 |
+
"grad_norm": 4.51666784286499,
|
| 1335 |
+
"learning_rate": 8.804888888888889e-06,
|
| 1336 |
+
"loss": 0.0639,
|
| 1337 |
+
"step": 30200
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 6.57439446366782,
|
| 1341 |
+
"grad_norm": 5.595043182373047,
|
| 1342 |
+
"learning_rate": 8.716000000000001e-06,
|
| 1343 |
+
"loss": 0.0629,
|
| 1344 |
+
"step": 30400
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"epoch": 6.617647058823529,
|
| 1348 |
+
"grad_norm": 2.728165864944458,
|
| 1349 |
+
"learning_rate": 8.627555555555556e-06,
|
| 1350 |
+
"loss": 0.0613,
|
| 1351 |
+
"step": 30600
|
| 1352 |
+
},
|
| 1353 |
+
{
|
| 1354 |
+
"epoch": 6.660899653979238,
|
| 1355 |
+
"grad_norm": 5.515061378479004,
|
| 1356 |
+
"learning_rate": 8.538666666666667e-06,
|
| 1357 |
+
"loss": 0.0627,
|
| 1358 |
+
"step": 30800
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 6.704152249134948,
|
| 1362 |
+
"grad_norm": 4.086916446685791,
|
| 1363 |
+
"learning_rate": 8.449777777777779e-06,
|
| 1364 |
+
"loss": 0.0634,
|
| 1365 |
+
"step": 31000
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 6.704152249134948,
|
| 1369 |
+
"eval_loss": 0.25769299268722534,
|
| 1370 |
+
"eval_runtime": 1839.2375,
|
| 1371 |
+
"eval_samples_per_second": 5.039,
|
| 1372 |
+
"eval_steps_per_second": 1.26,
|
| 1373 |
+
"eval_wer": 0.18051723331618574,
|
| 1374 |
+
"step": 31000
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 6.747404844290657,
|
| 1378 |
+
"grad_norm": 11.221170425415039,
|
| 1379 |
+
"learning_rate": 8.36088888888889e-06,
|
| 1380 |
+
"loss": 0.0676,
|
| 1381 |
+
"step": 31200
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 6.790657439446367,
|
| 1385 |
+
"grad_norm": 3.5501492023468018,
|
| 1386 |
+
"learning_rate": 8.272000000000001e-06,
|
| 1387 |
+
"loss": 0.0618,
|
| 1388 |
+
"step": 31400
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 6.833910034602076,
|
| 1392 |
+
"grad_norm": 5.12147331237793,
|
| 1393 |
+
"learning_rate": 8.183111111111112e-06,
|
| 1394 |
+
"loss": 0.0629,
|
| 1395 |
+
"step": 31600
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 6.877162629757786,
|
| 1399 |
+
"grad_norm": 4.411380290985107,
|
| 1400 |
+
"learning_rate": 8.094222222222224e-06,
|
| 1401 |
+
"loss": 0.0604,
|
| 1402 |
+
"step": 31800
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 6.920415224913495,
|
| 1406 |
+
"grad_norm": 4.010900020599365,
|
| 1407 |
+
"learning_rate": 8.005333333333335e-06,
|
| 1408 |
+
"loss": 0.0647,
|
| 1409 |
+
"step": 32000
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 6.920415224913495,
|
| 1413 |
+
"eval_loss": 0.25568053126335144,
|
| 1414 |
+
"eval_runtime": 1855.895,
|
| 1415 |
+
"eval_samples_per_second": 4.994,
|
| 1416 |
+
"eval_steps_per_second": 1.248,
|
| 1417 |
+
"eval_wer": 0.1858641600024942,
|
| 1418 |
+
"step": 32000
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 6.963667820069205,
|
| 1422 |
+
"grad_norm": 6.230306625366211,
|
| 1423 |
+
"learning_rate": 7.916444444444445e-06,
|
| 1424 |
+
"loss": 0.0642,
|
| 1425 |
+
"step": 32200
|
| 1426 |
+
},
|
| 1427 |
+
{
|
| 1428 |
+
"epoch": 7.006920415224913,
|
| 1429 |
+
"grad_norm": 5.123142242431641,
|
| 1430 |
+
"learning_rate": 7.827555555555555e-06,
|
| 1431 |
+
"loss": 0.06,
|
| 1432 |
+
"step": 32400
|
| 1433 |
+
},
|
| 1434 |
+
{
|
| 1435 |
+
"epoch": 7.050173010380623,
|
| 1436 |
+
"grad_norm": 2.5576796531677246,
|
| 1437 |
+
"learning_rate": 7.738666666666668e-06,
|
| 1438 |
+
"loss": 0.0474,
|
| 1439 |
+
"step": 32600
|
| 1440 |
+
},
|
| 1441 |
+
{
|
| 1442 |
+
"epoch": 7.093425605536332,
|
| 1443 |
+
"grad_norm": 4.169487953186035,
|
| 1444 |
+
"learning_rate": 7.649777777777778e-06,
|
| 1445 |
+
"loss": 0.045,
|
| 1446 |
+
"step": 32800
|
| 1447 |
+
},
|
| 1448 |
+
{
|
| 1449 |
+
"epoch": 7.1366782006920415,
|
| 1450 |
+
"grad_norm": 7.225770950317383,
|
| 1451 |
+
"learning_rate": 7.56088888888889e-06,
|
| 1452 |
+
"loss": 0.0467,
|
| 1453 |
+
"step": 33000
|
| 1454 |
+
},
|
| 1455 |
+
{
|
| 1456 |
+
"epoch": 7.1366782006920415,
|
| 1457 |
+
"eval_loss": 0.25844866037368774,
|
| 1458 |
+
"eval_runtime": 1843.7704,
|
| 1459 |
+
"eval_samples_per_second": 5.027,
|
| 1460 |
+
"eval_steps_per_second": 1.257,
|
| 1461 |
+
"eval_wer": 0.18003398338243776,
|
| 1462 |
+
"step": 33000
|
| 1463 |
+
},
|
| 1464 |
+
{
|
| 1465 |
+
"epoch": 7.179930795847751,
|
| 1466 |
+
"grad_norm": 3.876297950744629,
|
| 1467 |
+
"learning_rate": 7.4724444444444455e-06,
|
| 1468 |
+
"loss": 0.0461,
|
| 1469 |
+
"step": 33200
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 7.22318339100346,
|
| 1473 |
+
"grad_norm": 4.19256067276001,
|
| 1474 |
+
"learning_rate": 7.383555555555556e-06,
|
| 1475 |
+
"loss": 0.0493,
|
| 1476 |
+
"step": 33400
|
| 1477 |
+
},
|
| 1478 |
+
{
|
| 1479 |
+
"epoch": 7.26643598615917,
|
| 1480 |
+
"grad_norm": 4.434962272644043,
|
| 1481 |
+
"learning_rate": 7.294666666666668e-06,
|
| 1482 |
+
"loss": 0.0452,
|
| 1483 |
+
"step": 33600
|
| 1484 |
+
},
|
| 1485 |
+
{
|
| 1486 |
+
"epoch": 7.309688581314879,
|
| 1487 |
+
"grad_norm": 5.4323248863220215,
|
| 1488 |
+
"learning_rate": 7.2057777777777785e-06,
|
| 1489 |
+
"loss": 0.0481,
|
| 1490 |
+
"step": 33800
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 7.352941176470588,
|
| 1494 |
+
"grad_norm": 3.391416072845459,
|
| 1495 |
+
"learning_rate": 7.11688888888889e-06,
|
| 1496 |
+
"loss": 0.0474,
|
| 1497 |
+
"step": 34000
|
| 1498 |
+
},
|
| 1499 |
+
{
|
| 1500 |
+
"epoch": 7.352941176470588,
|
| 1501 |
+
"eval_loss": 0.2545304596424103,
|
| 1502 |
+
"eval_runtime": 1894.5017,
|
| 1503 |
+
"eval_samples_per_second": 4.892,
|
| 1504 |
+
"eval_steps_per_second": 1.223,
|
| 1505 |
+
"eval_wer": 0.1800183946748975,
|
| 1506 |
+
"step": 34000
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 7.396193771626297,
|
| 1510 |
+
"grad_norm": 3.7030251026153564,
|
| 1511 |
+
"learning_rate": 7.028e-06,
|
| 1512 |
+
"loss": 0.0477,
|
| 1513 |
+
"step": 34200
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 7.439446366782007,
|
| 1517 |
+
"grad_norm": 5.492336750030518,
|
| 1518 |
+
"learning_rate": 6.9395555555555565e-06,
|
| 1519 |
+
"loss": 0.0466,
|
| 1520 |
+
"step": 34400
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 7.482698961937716,
|
| 1524 |
+
"grad_norm": 4.130298137664795,
|
| 1525 |
+
"learning_rate": 6.850666666666668e-06,
|
| 1526 |
+
"loss": 0.0481,
|
| 1527 |
+
"step": 34600
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 7.525951557093426,
|
| 1531 |
+
"grad_norm": 4.441675186157227,
|
| 1532 |
+
"learning_rate": 6.761777777777778e-06,
|
| 1533 |
+
"loss": 0.0472,
|
| 1534 |
+
"step": 34800
|
| 1535 |
+
},
|
| 1536 |
+
{
|
| 1537 |
+
"epoch": 7.569204152249135,
|
| 1538 |
+
"grad_norm": 3.9193003177642822,
|
| 1539 |
+
"learning_rate": 6.6728888888888895e-06,
|
| 1540 |
+
"loss": 0.0478,
|
| 1541 |
+
"step": 35000
|
| 1542 |
+
},
|
| 1543 |
+
{
|
| 1544 |
+
"epoch": 7.569204152249135,
|
| 1545 |
+
"eval_loss": 0.25875434279441833,
|
| 1546 |
+
"eval_runtime": 1831.0658,
|
| 1547 |
+
"eval_samples_per_second": 5.062,
|
| 1548 |
+
"eval_steps_per_second": 1.265,
|
| 1549 |
+
"eval_wer": 0.1827152410793621,
|
| 1550 |
+
"step": 35000
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 7.612456747404845,
|
| 1554 |
+
"grad_norm": 5.8942131996154785,
|
| 1555 |
+
"learning_rate": 6.584e-06,
|
| 1556 |
+
"loss": 0.0468,
|
| 1557 |
+
"step": 35200
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 7.655709342560554,
|
| 1561 |
+
"grad_norm": 10.911412239074707,
|
| 1562 |
+
"learning_rate": 6.495111111111112e-06,
|
| 1563 |
+
"loss": 0.0485,
|
| 1564 |
+
"step": 35400
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 7.698961937716263,
|
| 1568 |
+
"grad_norm": 2.6245601177215576,
|
| 1569 |
+
"learning_rate": 6.406222222222223e-06,
|
| 1570 |
+
"loss": 0.0476,
|
| 1571 |
+
"step": 35600
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 7.742214532871972,
|
| 1575 |
+
"grad_norm": 8.821427345275879,
|
| 1576 |
+
"learning_rate": 6.317333333333334e-06,
|
| 1577 |
+
"loss": 0.0499,
|
| 1578 |
+
"step": 35800
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 7.7854671280276815,
|
| 1582 |
+
"grad_norm": 3.949411392211914,
|
| 1583 |
+
"learning_rate": 6.228444444444444e-06,
|
| 1584 |
+
"loss": 0.0485,
|
| 1585 |
+
"step": 36000
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 7.7854671280276815,
|
| 1589 |
+
"eval_loss": 0.2559249699115753,
|
| 1590 |
+
"eval_runtime": 1848.2662,
|
| 1591 |
+
"eval_samples_per_second": 5.014,
|
| 1592 |
+
"eval_steps_per_second": 1.254,
|
| 1593 |
+
"eval_wer": 0.1800183946748975,
|
| 1594 |
+
"step": 36000
|
| 1595 |
+
},
|
| 1596 |
+
{
|
| 1597 |
+
"epoch": 7.828719723183391,
|
| 1598 |
+
"grad_norm": 1.9492051601409912,
|
| 1599 |
+
"learning_rate": 6.1395555555555565e-06,
|
| 1600 |
+
"loss": 0.0471,
|
| 1601 |
+
"step": 36200
|
| 1602 |
+
},
|
| 1603 |
+
{
|
| 1604 |
+
"epoch": 7.8719723183391,
|
| 1605 |
+
"grad_norm": 4.136373996734619,
|
| 1606 |
+
"learning_rate": 6.050666666666667e-06,
|
| 1607 |
+
"loss": 0.0461,
|
| 1608 |
+
"step": 36400
|
| 1609 |
+
},
|
| 1610 |
+
{
|
| 1611 |
+
"epoch": 7.91522491349481,
|
| 1612 |
+
"grad_norm": 4.727861404418945,
|
| 1613 |
+
"learning_rate": 5.961777777777778e-06,
|
| 1614 |
+
"loss": 0.0469,
|
| 1615 |
+
"step": 36600
|
| 1616 |
+
},
|
| 1617 |
+
{
|
| 1618 |
+
"epoch": 7.958477508650519,
|
| 1619 |
+
"grad_norm": 5.108994007110596,
|
| 1620 |
+
"learning_rate": 5.872888888888889e-06,
|
| 1621 |
+
"loss": 0.0461,
|
| 1622 |
+
"step": 36800
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 8.001730103806228,
|
| 1626 |
+
"grad_norm": 2.9737696647644043,
|
| 1627 |
+
"learning_rate": 5.784000000000001e-06,
|
| 1628 |
+
"loss": 0.0456,
|
| 1629 |
+
"step": 37000
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 8.001730103806228,
|
| 1633 |
+
"eval_loss": 0.25564759969711304,
|
| 1634 |
+
"eval_runtime": 1798.8164,
|
| 1635 |
+
"eval_samples_per_second": 5.152,
|
| 1636 |
+
"eval_steps_per_second": 1.288,
|
| 1637 |
+
"eval_wer": 0.18043928977848447,
|
| 1638 |
+
"step": 37000
|
| 1639 |
+
},
|
| 1640 |
+
{
|
| 1641 |
+
"epoch": 8.044982698961938,
|
| 1642 |
+
"grad_norm": 2.9534389972686768,
|
| 1643 |
+
"learning_rate": 5.695111111111111e-06,
|
| 1644 |
+
"loss": 0.0377,
|
| 1645 |
+
"step": 37200
|
| 1646 |
+
},
|
| 1647 |
+
{
|
| 1648 |
+
"epoch": 8.088235294117647,
|
| 1649 |
+
"grad_norm": 3.9319846630096436,
|
| 1650 |
+
"learning_rate": 5.606222222222223e-06,
|
| 1651 |
+
"loss": 0.0373,
|
| 1652 |
+
"step": 37400
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 8.131487889273357,
|
| 1656 |
+
"grad_norm": 3.482482671737671,
|
| 1657 |
+
"learning_rate": 5.517333333333333e-06,
|
| 1658 |
+
"loss": 0.0357,
|
| 1659 |
+
"step": 37600
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 8.174740484429066,
|
| 1663 |
+
"grad_norm": 2.8425943851470947,
|
| 1664 |
+
"learning_rate": 5.428444444444445e-06,
|
| 1665 |
+
"loss": 0.0357,
|
| 1666 |
+
"step": 37800
|
| 1667 |
+
},
|
| 1668 |
+
{
|
| 1669 |
+
"epoch": 8.217993079584776,
|
| 1670 |
+
"grad_norm": 2.5396764278411865,
|
| 1671 |
+
"learning_rate": 5.339555555555556e-06,
|
| 1672 |
+
"loss": 0.0361,
|
| 1673 |
+
"step": 38000
|
| 1674 |
+
},
|
| 1675 |
+
{
|
| 1676 |
+
"epoch": 8.217993079584776,
|
| 1677 |
+
"eval_loss": 0.25599434971809387,
|
| 1678 |
+
"eval_runtime": 1833.5427,
|
| 1679 |
+
"eval_samples_per_second": 5.055,
|
| 1680 |
+
"eval_steps_per_second": 1.264,
|
| 1681 |
+
"eval_wer": 0.18442999890879047,
|
| 1682 |
+
"step": 38000
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 8.261245674740485,
|
| 1686 |
+
"grad_norm": 2.4965498447418213,
|
| 1687 |
+
"learning_rate": 5.250666666666667e-06,
|
| 1688 |
+
"loss": 0.0397,
|
| 1689 |
+
"step": 38200
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 8.304498269896193,
|
| 1693 |
+
"grad_norm": 4.19903039932251,
|
| 1694 |
+
"learning_rate": 5.1617777777777774e-06,
|
| 1695 |
+
"loss": 0.0377,
|
| 1696 |
+
"step": 38400
|
| 1697 |
+
},
|
| 1698 |
+
{
|
| 1699 |
+
"epoch": 8.347750865051903,
|
| 1700 |
+
"grad_norm": 5.88726282119751,
|
| 1701 |
+
"learning_rate": 5.07288888888889e-06,
|
| 1702 |
+
"loss": 0.0403,
|
| 1703 |
+
"step": 38600
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 8.391003460207612,
|
| 1707 |
+
"grad_norm": 3.427588939666748,
|
| 1708 |
+
"learning_rate": 4.984444444444445e-06,
|
| 1709 |
+
"loss": 0.0369,
|
| 1710 |
+
"step": 38800
|
| 1711 |
+
},
|
| 1712 |
+
{
|
| 1713 |
+
"epoch": 8.434256055363322,
|
| 1714 |
+
"grad_norm": 3.6567749977111816,
|
| 1715 |
+
"learning_rate": 4.895555555555556e-06,
|
| 1716 |
+
"loss": 0.0354,
|
| 1717 |
+
"step": 39000
|
| 1718 |
+
},
|
| 1719 |
+
{
|
| 1720 |
+
"epoch": 8.434256055363322,
|
| 1721 |
+
"eval_loss": 0.2549572288990021,
|
| 1722 |
+
"eval_runtime": 1795.7951,
|
| 1723 |
+
"eval_samples_per_second": 5.161,
|
| 1724 |
+
"eval_steps_per_second": 1.29,
|
| 1725 |
+
"eval_wer": 0.18057958814634678,
|
| 1726 |
+
"step": 39000
|
| 1727 |
+
},
|
| 1728 |
+
{
|
| 1729 |
+
"epoch": 8.477508650519031,
|
| 1730 |
+
"grad_norm": 4.668037414550781,
|
| 1731 |
+
"learning_rate": 4.8066666666666675e-06,
|
| 1732 |
+
"loss": 0.0382,
|
| 1733 |
+
"step": 39200
|
| 1734 |
+
},
|
| 1735 |
+
{
|
| 1736 |
+
"epoch": 8.520761245674741,
|
| 1737 |
+
"grad_norm": 5.387868881225586,
|
| 1738 |
+
"learning_rate": 4.717777777777778e-06,
|
| 1739 |
+
"loss": 0.0372,
|
| 1740 |
+
"step": 39400
|
| 1741 |
+
},
|
| 1742 |
+
{
|
| 1743 |
+
"epoch": 8.56401384083045,
|
| 1744 |
+
"grad_norm": 2.4888224601745605,
|
| 1745 |
+
"learning_rate": 4.629333333333333e-06,
|
| 1746 |
+
"loss": 0.0387,
|
| 1747 |
+
"step": 39600
|
| 1748 |
+
},
|
| 1749 |
+
{
|
| 1750 |
+
"epoch": 8.607266435986158,
|
| 1751 |
+
"grad_norm": 2.2906575202941895,
|
| 1752 |
+
"learning_rate": 4.540444444444445e-06,
|
| 1753 |
+
"loss": 0.0379,
|
| 1754 |
+
"step": 39800
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 8.650519031141869,
|
| 1758 |
+
"grad_norm": 3.7723381519317627,
|
| 1759 |
+
"learning_rate": 4.451555555555556e-06,
|
| 1760 |
+
"loss": 0.0365,
|
| 1761 |
+
"step": 40000
|
| 1762 |
+
},
|
| 1763 |
+
{
|
| 1764 |
+
"epoch": 8.650519031141869,
|
| 1765 |
+
"eval_loss": 0.25570422410964966,
|
| 1766 |
+
"eval_runtime": 1870.4038,
|
| 1767 |
+
"eval_samples_per_second": 4.955,
|
| 1768 |
+
"eval_steps_per_second": 1.239,
|
| 1769 |
+
"eval_wer": 0.18728273238865767,
|
| 1770 |
+
"step": 40000
|
| 1771 |
+
},
|
| 1772 |
+
{
|
| 1773 |
+
"epoch": 8.693771626297577,
|
| 1774 |
+
"grad_norm": 5.955033779144287,
|
| 1775 |
+
"learning_rate": 4.362666666666667e-06,
|
| 1776 |
+
"loss": 0.0376,
|
| 1777 |
+
"step": 40200
|
| 1778 |
+
},
|
| 1779 |
+
{
|
| 1780 |
+
"epoch": 8.737024221453288,
|
| 1781 |
+
"grad_norm": 5.938442230224609,
|
| 1782 |
+
"learning_rate": 4.2737777777777785e-06,
|
| 1783 |
+
"loss": 0.0359,
|
| 1784 |
+
"step": 40400
|
| 1785 |
+
},
|
| 1786 |
+
{
|
| 1787 |
+
"epoch": 8.780276816608996,
|
| 1788 |
+
"grad_norm": 4.009905815124512,
|
| 1789 |
+
"learning_rate": 4.18488888888889e-06,
|
| 1790 |
+
"loss": 0.0364,
|
| 1791 |
+
"step": 40600
|
| 1792 |
+
},
|
| 1793 |
+
{
|
| 1794 |
+
"epoch": 8.823529411764707,
|
| 1795 |
+
"grad_norm": 4.421857833862305,
|
| 1796 |
+
"learning_rate": 4.096e-06,
|
| 1797 |
+
"loss": 0.041,
|
| 1798 |
+
"step": 40800
|
| 1799 |
+
},
|
| 1800 |
+
{
|
| 1801 |
+
"epoch": 8.866782006920415,
|
| 1802 |
+
"grad_norm": 4.176875591278076,
|
| 1803 |
+
"learning_rate": 4.0071111111111116e-06,
|
| 1804 |
+
"loss": 0.0388,
|
| 1805 |
+
"step": 41000
|
| 1806 |
+
},
|
| 1807 |
+
{
|
| 1808 |
+
"epoch": 8.866782006920415,
|
| 1809 |
+
"eval_loss": 0.25398069620132446,
|
| 1810 |
+
"eval_runtime": 1940.4802,
|
| 1811 |
+
"eval_samples_per_second": 4.776,
|
| 1812 |
+
"eval_steps_per_second": 1.194,
|
| 1813 |
+
"eval_wer": 0.18425852312584765,
|
| 1814 |
+
"step": 41000
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 8.910034602076124,
|
| 1818 |
+
"grad_norm": 6.142982006072998,
|
| 1819 |
+
"learning_rate": 3.918222222222223e-06,
|
| 1820 |
+
"loss": 0.0376,
|
| 1821 |
+
"step": 41200
|
| 1822 |
+
},
|
| 1823 |
+
{
|
| 1824 |
+
"epoch": 8.953287197231834,
|
| 1825 |
+
"grad_norm": 3.8413658142089844,
|
| 1826 |
+
"learning_rate": 3.829333333333334e-06,
|
| 1827 |
+
"loss": 0.0373,
|
| 1828 |
+
"step": 41400
|
| 1829 |
+
},
|
| 1830 |
+
{
|
| 1831 |
+
"epoch": 8.996539792387543,
|
| 1832 |
+
"grad_norm": 4.360950469970703,
|
| 1833 |
+
"learning_rate": 3.740444444444445e-06,
|
| 1834 |
+
"loss": 0.0376,
|
| 1835 |
+
"step": 41600
|
| 1836 |
+
},
|
| 1837 |
+
{
|
| 1838 |
+
"epoch": 9.039792387543253,
|
| 1839 |
+
"grad_norm": 1.8043413162231445,
|
| 1840 |
+
"learning_rate": 3.651555555555556e-06,
|
| 1841 |
+
"loss": 0.0331,
|
| 1842 |
+
"step": 41800
|
| 1843 |
+
},
|
| 1844 |
+
{
|
| 1845 |
+
"epoch": 9.083044982698961,
|
| 1846 |
+
"grad_norm": 4.323986053466797,
|
| 1847 |
+
"learning_rate": 3.5626666666666672e-06,
|
| 1848 |
+
"loss": 0.0317,
|
| 1849 |
+
"step": 42000
|
| 1850 |
+
},
|
| 1851 |
+
{
|
| 1852 |
+
"epoch": 9.083044982698961,
|
| 1853 |
+
"eval_loss": 0.254744291305542,
|
| 1854 |
+
"eval_runtime": 1985.3925,
|
| 1855 |
+
"eval_samples_per_second": 4.668,
|
| 1856 |
+
"eval_steps_per_second": 1.167,
|
| 1857 |
+
"eval_wer": 0.1819046282872687,
|
| 1858 |
+
"step": 42000
|
| 1859 |
+
},
|
| 1860 |
+
{
|
| 1861 |
+
"epoch": 9.126297577854672,
|
| 1862 |
+
"grad_norm": 1.2513147592544556,
|
| 1863 |
+
"learning_rate": 3.473777777777778e-06,
|
| 1864 |
+
"loss": 0.0306,
|
| 1865 |
+
"step": 42200
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 9.16955017301038,
|
| 1869 |
+
"grad_norm": 6.47668981552124,
|
| 1870 |
+
"learning_rate": 3.3848888888888894e-06,
|
| 1871 |
+
"loss": 0.0315,
|
| 1872 |
+
"step": 42400
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 9.21280276816609,
|
| 1876 |
+
"grad_norm": 3.608281135559082,
|
| 1877 |
+
"learning_rate": 3.2960000000000003e-06,
|
| 1878 |
+
"loss": 0.0321,
|
| 1879 |
+
"step": 42600
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 9.2560553633218,
|
| 1883 |
+
"grad_norm": 4.180387496948242,
|
| 1884 |
+
"learning_rate": 3.2071111111111116e-06,
|
| 1885 |
+
"loss": 0.034,
|
| 1886 |
+
"step": 42800
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 9.299307958477508,
|
| 1890 |
+
"grad_norm": 3.6650965213775635,
|
| 1891 |
+
"learning_rate": 3.1182222222222225e-06,
|
| 1892 |
+
"loss": 0.0334,
|
| 1893 |
+
"step": 43000
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 9.299307958477508,
|
| 1897 |
+
"eval_loss": 0.25559520721435547,
|
| 1898 |
+
"eval_runtime": 1997.1707,
|
| 1899 |
+
"eval_samples_per_second": 4.641,
|
| 1900 |
+
"eval_steps_per_second": 1.16,
|
| 1901 |
+
"eval_wer": 0.1780230401097445,
|
| 1902 |
+
"step": 43000
|
| 1903 |
+
},
|
| 1904 |
+
{
|
| 1905 |
+
"epoch": 9.342560553633218,
|
| 1906 |
+
"grad_norm": 6.112791538238525,
|
| 1907 |
+
"learning_rate": 3.0293333333333338e-06,
|
| 1908 |
+
"loss": 0.03,
|
| 1909 |
+
"step": 43200
|
| 1910 |
+
},
|
| 1911 |
+
{
|
| 1912 |
+
"epoch": 9.385813148788927,
|
| 1913 |
+
"grad_norm": 3.127464771270752,
|
| 1914 |
+
"learning_rate": 2.9404444444444447e-06,
|
| 1915 |
+
"loss": 0.0319,
|
| 1916 |
+
"step": 43400
|
| 1917 |
+
},
|
| 1918 |
+
{
|
| 1919 |
+
"epoch": 9.429065743944637,
|
| 1920 |
+
"grad_norm": 4.579778671264648,
|
| 1921 |
+
"learning_rate": 2.8515555555555555e-06,
|
| 1922 |
+
"loss": 0.0305,
|
| 1923 |
+
"step": 43600
|
| 1924 |
+
},
|
| 1925 |
+
{
|
| 1926 |
+
"epoch": 9.472318339100346,
|
| 1927 |
+
"grad_norm": 4.794194221496582,
|
| 1928 |
+
"learning_rate": 2.762666666666667e-06,
|
| 1929 |
+
"loss": 0.0312,
|
| 1930 |
+
"step": 43800
|
| 1931 |
+
},
|
| 1932 |
+
{
|
| 1933 |
+
"epoch": 9.515570934256056,
|
| 1934 |
+
"grad_norm": 1.6962083578109741,
|
| 1935 |
+
"learning_rate": 2.6737777777777777e-06,
|
| 1936 |
+
"loss": 0.033,
|
| 1937 |
+
"step": 44000
|
| 1938 |
+
},
|
| 1939 |
+
{
|
| 1940 |
+
"epoch": 9.515570934256056,
|
| 1941 |
+
"eval_loss": 0.2551681399345398,
|
| 1942 |
+
"eval_runtime": 1926.1067,
|
| 1943 |
+
"eval_samples_per_second": 4.812,
|
| 1944 |
+
"eval_steps_per_second": 1.203,
|
| 1945 |
+
"eval_wer": 0.1801275156276793,
|
| 1946 |
+
"step": 44000
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 9.558823529411764,
|
| 1950 |
+
"grad_norm": 3.76029634475708,
|
| 1951 |
+
"learning_rate": 2.5853333333333335e-06,
|
| 1952 |
+
"loss": 0.0297,
|
| 1953 |
+
"step": 44200
|
| 1954 |
+
},
|
| 1955 |
+
{
|
| 1956 |
+
"epoch": 9.602076124567475,
|
| 1957 |
+
"grad_norm": 2.945800304412842,
|
| 1958 |
+
"learning_rate": 2.4964444444444448e-06,
|
| 1959 |
+
"loss": 0.0302,
|
| 1960 |
+
"step": 44400
|
| 1961 |
+
},
|
| 1962 |
+
{
|
| 1963 |
+
"epoch": 9.645328719723183,
|
| 1964 |
+
"grad_norm": 3.4467453956604004,
|
| 1965 |
+
"learning_rate": 2.4075555555555556e-06,
|
| 1966 |
+
"loss": 0.0293,
|
| 1967 |
+
"step": 44600
|
| 1968 |
+
},
|
| 1969 |
+
{
|
| 1970 |
+
"epoch": 9.688581314878892,
|
| 1971 |
+
"grad_norm": 2.707533121109009,
|
| 1972 |
+
"learning_rate": 2.318666666666667e-06,
|
| 1973 |
+
"loss": 0.0297,
|
| 1974 |
+
"step": 44800
|
| 1975 |
+
},
|
| 1976 |
+
{
|
| 1977 |
+
"epoch": 9.731833910034602,
|
| 1978 |
+
"grad_norm": 2.110910177230835,
|
| 1979 |
+
"learning_rate": 2.229777777777778e-06,
|
| 1980 |
+
"loss": 0.0313,
|
| 1981 |
+
"step": 45000
|
| 1982 |
+
},
|
| 1983 |
+
{
|
| 1984 |
+
"epoch": 9.731833910034602,
|
| 1985 |
+
"eval_loss": 0.254027783870697,
|
| 1986 |
+
"eval_runtime": 1956.344,
|
| 1987 |
+
"eval_samples_per_second": 4.737,
|
| 1988 |
+
"eval_steps_per_second": 1.184,
|
| 1989 |
+
"eval_wer": 0.1787245319490561,
|
| 1990 |
+
"step": 45000
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 9.77508650519031,
|
| 1994 |
+
"grad_norm": 2.0399131774902344,
|
| 1995 |
+
"learning_rate": 2.140888888888889e-06,
|
| 1996 |
+
"loss": 0.0301,
|
| 1997 |
+
"step": 45200
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 9.818339100346021,
|
| 2001 |
+
"grad_norm": 3.487839460372925,
|
| 2002 |
+
"learning_rate": 2.052e-06,
|
| 2003 |
+
"loss": 0.0304,
|
| 2004 |
+
"step": 45400
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 9.86159169550173,
|
| 2008 |
+
"grad_norm": 3.9768216609954834,
|
| 2009 |
+
"learning_rate": 1.9631111111111113e-06,
|
| 2010 |
+
"loss": 0.0306,
|
| 2011 |
+
"step": 45600
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 9.90484429065744,
|
| 2015 |
+
"grad_norm": 2.6950674057006836,
|
| 2016 |
+
"learning_rate": 1.8742222222222222e-06,
|
| 2017 |
+
"loss": 0.0294,
|
| 2018 |
+
"step": 45800
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 9.948096885813149,
|
| 2022 |
+
"grad_norm": 4.673542022705078,
|
| 2023 |
+
"learning_rate": 1.7853333333333333e-06,
|
| 2024 |
+
"loss": 0.0318,
|
| 2025 |
+
"step": 46000
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 9.948096885813149,
|
| 2029 |
+
"eval_loss": 0.25369492173194885,
|
| 2030 |
+
"eval_runtime": 1880.4418,
|
| 2031 |
+
"eval_samples_per_second": 4.929,
|
| 2032 |
+
"eval_steps_per_second": 1.232,
|
| 2033 |
+
"eval_wer": 0.1772436047327316,
|
| 2034 |
+
"step": 46000
|
| 2035 |
+
},
|
| 2036 |
+
{
|
| 2037 |
+
"epoch": 9.991349480968857,
|
| 2038 |
+
"grad_norm": 1.9740992784500122,
|
| 2039 |
+
"learning_rate": 1.6964444444444444e-06,
|
| 2040 |
+
"loss": 0.0274,
|
| 2041 |
+
"step": 46200
|
| 2042 |
+
},
|
| 2043 |
+
{
|
| 2044 |
+
"epoch": 10.034602076124568,
|
| 2045 |
+
"grad_norm": 0.886242687702179,
|
| 2046 |
+
"learning_rate": 1.6075555555555559e-06,
|
| 2047 |
+
"loss": 0.0269,
|
| 2048 |
+
"step": 46400
|
| 2049 |
+
},
|
| 2050 |
+
{
|
| 2051 |
+
"epoch": 10.077854671280276,
|
| 2052 |
+
"grad_norm": 2.5038974285125732,
|
| 2053 |
+
"learning_rate": 1.518666666666667e-06,
|
| 2054 |
+
"loss": 0.0288,
|
| 2055 |
+
"step": 46600
|
| 2056 |
+
},
|
| 2057 |
+
{
|
| 2058 |
+
"epoch": 10.121107266435986,
|
| 2059 |
+
"grad_norm": 2.415977954864502,
|
| 2060 |
+
"learning_rate": 1.429777777777778e-06,
|
| 2061 |
+
"loss": 0.0267,
|
| 2062 |
+
"step": 46800
|
| 2063 |
+
},
|
| 2064 |
+
{
|
| 2065 |
+
"epoch": 10.164359861591695,
|
| 2066 |
+
"grad_norm": 3.4146268367767334,
|
| 2067 |
+
"learning_rate": 1.3408888888888892e-06,
|
| 2068 |
+
"loss": 0.0285,
|
| 2069 |
+
"step": 47000
|
| 2070 |
+
},
|
| 2071 |
+
{
|
| 2072 |
+
"epoch": 10.164359861591695,
|
| 2073 |
+
"eval_loss": 0.2533535957336426,
|
| 2074 |
+
"eval_runtime": 1606.1461,
|
| 2075 |
+
"eval_samples_per_second": 5.77,
|
| 2076 |
+
"eval_steps_per_second": 1.443,
|
| 2077 |
+
"eval_wer": 0.17641740323309793,
|
| 2078 |
+
"step": 47000
|
| 2079 |
+
},
|
| 2080 |
+
{
|
| 2081 |
+
"epoch": 10.207612456747405,
|
| 2082 |
+
"grad_norm": 3.1481282711029053,
|
| 2083 |
+
"learning_rate": 1.2520000000000003e-06,
|
| 2084 |
+
"loss": 0.0274,
|
| 2085 |
+
"step": 47200
|
| 2086 |
+
},
|
| 2087 |
+
{
|
| 2088 |
+
"epoch": 10.250865051903114,
|
| 2089 |
+
"grad_norm": 3.1010215282440186,
|
| 2090 |
+
"learning_rate": 1.1631111111111113e-06,
|
| 2091 |
+
"loss": 0.0265,
|
| 2092 |
+
"step": 47400
|
| 2093 |
+
},
|
| 2094 |
+
{
|
| 2095 |
+
"epoch": 10.294117647058824,
|
| 2096 |
+
"grad_norm": 4.669837474822998,
|
| 2097 |
+
"learning_rate": 1.0742222222222224e-06,
|
| 2098 |
+
"loss": 0.0281,
|
| 2099 |
+
"step": 47600
|
| 2100 |
+
},
|
| 2101 |
+
{
|
| 2102 |
+
"epoch": 10.337370242214533,
|
| 2103 |
+
"grad_norm": 2.6151702404022217,
|
| 2104 |
+
"learning_rate": 9.853333333333333e-07,
|
| 2105 |
+
"loss": 0.0281,
|
| 2106 |
+
"step": 47800
|
| 2107 |
+
},
|
| 2108 |
+
{
|
| 2109 |
+
"epoch": 10.380622837370241,
|
| 2110 |
+
"grad_norm": 10.903833389282227,
|
| 2111 |
+
"learning_rate": 8.964444444444445e-07,
|
| 2112 |
+
"loss": 0.0256,
|
| 2113 |
+
"step": 48000
|
| 2114 |
+
},
|
| 2115 |
+
{
|
| 2116 |
+
"epoch": 10.380622837370241,
|
| 2117 |
+
"eval_loss": 0.25295063853263855,
|
| 2118 |
+
"eval_runtime": 1556.9047,
|
| 2119 |
+
"eval_samples_per_second": 5.953,
|
| 2120 |
+
"eval_steps_per_second": 1.488,
|
| 2121 |
+
"eval_wer": 0.17711889507240955,
|
| 2122 |
+
"step": 48000
|
| 2123 |
+
},
|
| 2124 |
+
{
|
| 2125 |
+
"epoch": 10.423875432525952,
|
| 2126 |
+
"grad_norm": 3.759805679321289,
|
| 2127 |
+
"learning_rate": 8.075555555555556e-07,
|
| 2128 |
+
"loss": 0.0269,
|
| 2129 |
+
"step": 48200
|
| 2130 |
+
},
|
| 2131 |
+
{
|
| 2132 |
+
"epoch": 10.46712802768166,
|
| 2133 |
+
"grad_norm": 2.40134334564209,
|
| 2134 |
+
"learning_rate": 7.191111111111111e-07,
|
| 2135 |
+
"loss": 0.0272,
|
| 2136 |
+
"step": 48400
|
| 2137 |
+
},
|
| 2138 |
+
{
|
| 2139 |
+
"epoch": 10.51038062283737,
|
| 2140 |
+
"grad_norm": 3.8142147064208984,
|
| 2141 |
+
"learning_rate": 6.306666666666666e-07,
|
| 2142 |
+
"loss": 0.0274,
|
| 2143 |
+
"step": 48600
|
| 2144 |
+
},
|
| 2145 |
+
{
|
| 2146 |
+
"epoch": 10.55363321799308,
|
| 2147 |
+
"grad_norm": 10.721040725708008,
|
| 2148 |
+
"learning_rate": 5.417777777777778e-07,
|
| 2149 |
+
"loss": 0.0269,
|
| 2150 |
+
"step": 48800
|
| 2151 |
+
},
|
| 2152 |
+
{
|
| 2153 |
+
"epoch": 10.59688581314879,
|
| 2154 |
+
"grad_norm": 4.5076494216918945,
|
| 2155 |
+
"learning_rate": 4.528888888888889e-07,
|
| 2156 |
+
"loss": 0.0288,
|
| 2157 |
+
"step": 49000
|
| 2158 |
+
},
|
| 2159 |
+
{
|
| 2160 |
+
"epoch": 10.59688581314879,
|
| 2161 |
+
"eval_loss": 0.2532360255718231,
|
| 2162 |
+
"eval_runtime": 1587.8867,
|
| 2163 |
+
"eval_samples_per_second": 5.837,
|
| 2164 |
+
"eval_steps_per_second": 1.459,
|
| 2165 |
+
"eval_wer": 0.1760276855445915,
|
| 2166 |
+
"step": 49000
|
| 2167 |
+
},
|
| 2168 |
+
{
|
| 2169 |
+
"epoch": 10.640138408304498,
|
| 2170 |
+
"grad_norm": 2.6491665840148926,
|
| 2171 |
+
"learning_rate": 3.6400000000000003e-07,
|
| 2172 |
+
"loss": 0.0275,
|
| 2173 |
+
"step": 49200
|
| 2174 |
+
},
|
| 2175 |
+
{
|
| 2176 |
+
"epoch": 10.683391003460208,
|
| 2177 |
+
"grad_norm": 7.777415752410889,
|
| 2178 |
+
"learning_rate": 2.751111111111111e-07,
|
| 2179 |
+
"loss": 0.0293,
|
| 2180 |
+
"step": 49400
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 10.726643598615917,
|
| 2184 |
+
"grad_norm": 2.240342378616333,
|
| 2185 |
+
"learning_rate": 1.8622222222222221e-07,
|
| 2186 |
+
"loss": 0.0254,
|
| 2187 |
+
"step": 49600
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 10.769896193771626,
|
| 2191 |
+
"grad_norm": 4.06622314453125,
|
| 2192 |
+
"learning_rate": 9.733333333333334e-08,
|
| 2193 |
+
"loss": 0.0269,
|
| 2194 |
+
"step": 49800
|
| 2195 |
+
},
|
| 2196 |
+
{
|
| 2197 |
+
"epoch": 10.813148788927336,
|
| 2198 |
+
"grad_norm": 3.918858051300049,
|
| 2199 |
+
"learning_rate": 8.444444444444445e-09,
|
| 2200 |
+
"loss": 0.0265,
|
| 2201 |
+
"step": 50000
|
| 2202 |
+
},
|
| 2203 |
+
{
|
| 2204 |
+
"epoch": 10.813148788927336,
|
| 2205 |
+
"eval_loss": 0.2530483603477478,
|
| 2206 |
+
"eval_runtime": 1579.0697,
|
| 2207 |
+
"eval_samples_per_second": 5.869,
|
| 2208 |
+
"eval_steps_per_second": 1.467,
|
| 2209 |
+
"eval_wer": 0.17626151615769536,
|
| 2210 |
+
"step": 50000
|
| 2211 |
+
},
|
| 2212 |
+
{
|
| 2213 |
+
"epoch": 10.813365051903114,
|
| 2214 |
+
"step": 50001,
|
| 2215 |
+
"total_flos": 2.1271110511959736e+19,
|
| 2216 |
+
"train_loss": 3.7687225894058625e-07,
|
| 2217 |
+
"train_runtime": 8.535,
|
| 2218 |
+
"train_samples_per_second": 93731.391,
|
| 2219 |
+
"train_steps_per_second": 5858.212
|
| 2220 |
+
}
|
| 2221 |
+
],
|
| 2222 |
+
"logging_steps": 200,
|
| 2223 |
+
"max_steps": 50000,
|
| 2224 |
+
"num_input_tokens_seen": 0,
|
| 2225 |
+
"num_train_epochs": 11,
|
| 2226 |
+
"save_steps": 1000,
|
| 2227 |
+
"stateful_callbacks": {
|
| 2228 |
+
"TrainerControl": {
|
| 2229 |
+
"args": {
|
| 2230 |
+
"should_epoch_stop": false,
|
| 2231 |
+
"should_evaluate": false,
|
| 2232 |
+
"should_log": false,
|
| 2233 |
+
"should_save": true,
|
| 2234 |
+
"should_training_stop": true
|
| 2235 |
+
},
|
| 2236 |
+
"attributes": {}
|
| 2237 |
+
}
|
| 2238 |
+
},
|
| 2239 |
+
"total_flos": 2.1271110511959736e+19,
|
| 2240 |
+
"train_batch_size": 8,
|
| 2241 |
+
"trial_name": null,
|
| 2242 |
+
"trial_params": null
|
| 2243 |
+
}
|
val_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 10.813365051903114,
|
| 3 |
+
"phase_1_val_loss": 0.2532360255718231,
|
| 4 |
+
"phase_1_val_runtime": 1660.0444,
|
| 5 |
+
"phase_1_val_samples_per_second": 5.583,
|
| 6 |
+
"phase_1_val_steps_per_second": 1.396,
|
| 7 |
+
"phase_1_val_wer": 0.1760276855445915,
|
| 8 |
+
"val_samples": 9268
|
| 9 |
+
}
|