End of training
Browse files- README.md +6 -3
- all_results.json +16 -0
- eval_results.json +10 -0
- train_results.json +9 -0
- trainer_state.json +932 -0
README.md
CHANGED
|
@@ -1,7 +1,10 @@
|
|
| 1 |
---
|
| 2 |
library_name: transformers
|
|
|
|
| 3 |
base_model: rinna/japanese-hubert-base
|
| 4 |
tags:
|
|
|
|
|
|
|
| 5 |
- generated_from_trainer
|
| 6 |
metrics:
|
| 7 |
- wer
|
|
@@ -15,11 +18,11 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 15 |
|
| 16 |
# Hubert-kakeiken-W-closed_add_ver2
|
| 17 |
|
| 18 |
-
This model is a fine-tuned version of [rinna/japanese-hubert-base](https://huggingface.co/rinna/japanese-hubert-base) on the
|
| 19 |
It achieves the following results on the evaluation set:
|
| 20 |
-
- Loss: 0.
|
| 21 |
- Wer: 0.9988
|
| 22 |
-
- Cer: 1.
|
| 23 |
|
| 24 |
## Model description
|
| 25 |
|
|
|
|
| 1 |
---
|
| 2 |
library_name: transformers
|
| 3 |
+
license: apache-2.0
|
| 4 |
base_model: rinna/japanese-hubert-base
|
| 5 |
tags:
|
| 6 |
+
- automatic-speech-recognition
|
| 7 |
+
- original_kakeiken_W_closed_add_ver2
|
| 8 |
- generated_from_trainer
|
| 9 |
metrics:
|
| 10 |
- wer
|
|
|
|
| 18 |
|
| 19 |
# Hubert-kakeiken-W-closed_add_ver2
|
| 20 |
|
| 21 |
+
This model is a fine-tuned version of [rinna/japanese-hubert-base](https://huggingface.co/rinna/japanese-hubert-base) on the ORIGINAL_KAKEIKEN_W_CLOSED_ADD_VER2 - JA dataset.
|
| 22 |
It achieves the following results on the evaluation set:
|
| 23 |
+
- Loss: 0.0617
|
| 24 |
- Wer: 0.9988
|
| 25 |
+
- Cer: 1.0129
|
| 26 |
|
| 27 |
## Model description
|
| 28 |
|
all_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 39.955088118249,
|
| 3 |
+
"eval_cer": 1.0129208155523945,
|
| 4 |
+
"eval_loss": 0.061716873198747635,
|
| 5 |
+
"eval_runtime": 59.3704,
|
| 6 |
+
"eval_samples": 6840,
|
| 7 |
+
"eval_samples_per_second": 115.209,
|
| 8 |
+
"eval_steps_per_second": 14.401,
|
| 9 |
+
"eval_wer": 0.9988304093567252,
|
| 10 |
+
"total_flos": 1.8269796433195942e+19,
|
| 11 |
+
"train_loss": 1.1798896302406563,
|
| 12 |
+
"train_runtime": 28544.574,
|
| 13 |
+
"train_samples": 56280,
|
| 14 |
+
"train_samples_per_second": 78.866,
|
| 15 |
+
"train_steps_per_second": 1.232
|
| 16 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 39.955088118249,
|
| 3 |
+
"eval_cer": 1.0129208155523945,
|
| 4 |
+
"eval_loss": 0.061716873198747635,
|
| 5 |
+
"eval_runtime": 59.3704,
|
| 6 |
+
"eval_samples": 6840,
|
| 7 |
+
"eval_samples_per_second": 115.209,
|
| 8 |
+
"eval_steps_per_second": 14.401,
|
| 9 |
+
"eval_wer": 0.9988304093567252
|
| 10 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 39.955088118249,
|
| 3 |
+
"total_flos": 1.8269796433195942e+19,
|
| 4 |
+
"train_loss": 1.1798896302406563,
|
| 5 |
+
"train_runtime": 28544.574,
|
| 6 |
+
"train_samples": 56280,
|
| 7 |
+
"train_samples_per_second": 78.866,
|
| 8 |
+
"train_steps_per_second": 1.232
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,932 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 39.955088118249,
|
| 5 |
+
"eval_steps": 100.0,
|
| 6 |
+
"global_step": 35160,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.5685048322910745,
|
| 13 |
+
"grad_norm": 58.098880767822266,
|
| 14 |
+
"learning_rate": 1.1904e-06,
|
| 15 |
+
"loss": 28.4059,
|
| 16 |
+
"step": 500
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 1.0,
|
| 20 |
+
"eval_cer": 1.1284080132764343,
|
| 21 |
+
"eval_loss": 10.672073364257812,
|
| 22 |
+
"eval_runtime": 62.3734,
|
| 23 |
+
"eval_samples_per_second": 109.662,
|
| 24 |
+
"eval_steps_per_second": 13.708,
|
| 25 |
+
"eval_wer": 1.0,
|
| 26 |
+
"step": 880
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 1.1364411597498578,
|
| 30 |
+
"grad_norm": 39.93675231933594,
|
| 31 |
+
"learning_rate": 2.3880000000000003e-06,
|
| 32 |
+
"loss": 11.3029,
|
| 33 |
+
"step": 1000
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 1.7049459920409324,
|
| 37 |
+
"grad_norm": 36.53426742553711,
|
| 38 |
+
"learning_rate": 3.588e-06,
|
| 39 |
+
"loss": 9.1792,
|
| 40 |
+
"step": 1500
|
| 41 |
+
},
|
| 42 |
+
{
|
| 43 |
+
"epoch": 2.0,
|
| 44 |
+
"eval_cer": 1.1283783783783783,
|
| 45 |
+
"eval_loss": 6.992434978485107,
|
| 46 |
+
"eval_runtime": 59.9626,
|
| 47 |
+
"eval_samples_per_second": 114.071,
|
| 48 |
+
"eval_steps_per_second": 14.259,
|
| 49 |
+
"eval_wer": 1.0,
|
| 50 |
+
"step": 1760
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 2.2728823194997156,
|
| 54 |
+
"grad_norm": 24.559185028076172,
|
| 55 |
+
"learning_rate": 4.788e-06,
|
| 56 |
+
"loss": 7.001,
|
| 57 |
+
"step": 2000
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 2.8413871517907903,
|
| 61 |
+
"grad_norm": 10.105766296386719,
|
| 62 |
+
"learning_rate": 5.988e-06,
|
| 63 |
+
"loss": 4.9143,
|
| 64 |
+
"step": 2500
|
| 65 |
+
},
|
| 66 |
+
{
|
| 67 |
+
"epoch": 3.0,
|
| 68 |
+
"eval_cer": 1.1283783783783783,
|
| 69 |
+
"eval_loss": 3.8166255950927734,
|
| 70 |
+
"eval_runtime": 59.1681,
|
| 71 |
+
"eval_samples_per_second": 115.603,
|
| 72 |
+
"eval_steps_per_second": 14.45,
|
| 73 |
+
"eval_wer": 1.0,
|
| 74 |
+
"step": 2640
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 3.4093234792495735,
|
| 78 |
+
"grad_norm": 1.8796437978744507,
|
| 79 |
+
"learning_rate": 7.1880000000000005e-06,
|
| 80 |
+
"loss": 3.6813,
|
| 81 |
+
"step": 3000
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 3.977828311540648,
|
| 85 |
+
"grad_norm": 2.5441930294036865,
|
| 86 |
+
"learning_rate": 8.388e-06,
|
| 87 |
+
"loss": 3.1394,
|
| 88 |
+
"step": 3500
|
| 89 |
+
},
|
| 90 |
+
{
|
| 91 |
+
"epoch": 4.0,
|
| 92 |
+
"eval_cer": 1.1283191085822666,
|
| 93 |
+
"eval_loss": 2.882925033569336,
|
| 94 |
+
"eval_runtime": 60.4525,
|
| 95 |
+
"eval_samples_per_second": 113.147,
|
| 96 |
+
"eval_steps_per_second": 14.143,
|
| 97 |
+
"eval_wer": 1.0,
|
| 98 |
+
"step": 3520
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 4.545764638999431,
|
| 102 |
+
"grad_norm": 1.8130204677581787,
|
| 103 |
+
"learning_rate": 9.588e-06,
|
| 104 |
+
"loss": 2.7266,
|
| 105 |
+
"step": 4000
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 5.0,
|
| 109 |
+
"eval_cer": 1.1444108582266477,
|
| 110 |
+
"eval_loss": 1.9608492851257324,
|
| 111 |
+
"eval_runtime": 61.7902,
|
| 112 |
+
"eval_samples_per_second": 110.697,
|
| 113 |
+
"eval_steps_per_second": 13.837,
|
| 114 |
+
"eval_wer": 1.0,
|
| 115 |
+
"step": 4400
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 5.1137009664582145,
|
| 119 |
+
"grad_norm": 5.103806972503662,
|
| 120 |
+
"learning_rate": 1.0787999999999999e-05,
|
| 121 |
+
"loss": 2.1714,
|
| 122 |
+
"step": 4500
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 5.6822057987492895,
|
| 126 |
+
"grad_norm": 6.647071838378906,
|
| 127 |
+
"learning_rate": 1.1988000000000001e-05,
|
| 128 |
+
"loss": 1.4314,
|
| 129 |
+
"step": 5000
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 6.0,
|
| 133 |
+
"eval_cer": 1.0662043622569939,
|
| 134 |
+
"eval_loss": 0.8433689475059509,
|
| 135 |
+
"eval_runtime": 62.0284,
|
| 136 |
+
"eval_samples_per_second": 110.272,
|
| 137 |
+
"eval_steps_per_second": 13.784,
|
| 138 |
+
"eval_wer": 0.9998538011695907,
|
| 139 |
+
"step": 5280
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"epoch": 6.250142126208073,
|
| 143 |
+
"grad_norm": 5.186183929443359,
|
| 144 |
+
"learning_rate": 1.3188e-05,
|
| 145 |
+
"loss": 0.9139,
|
| 146 |
+
"step": 5500
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 6.818646958499147,
|
| 150 |
+
"grad_norm": 5.24096155166626,
|
| 151 |
+
"learning_rate": 1.4388000000000002e-05,
|
| 152 |
+
"loss": 0.6837,
|
| 153 |
+
"step": 6000
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 7.0,
|
| 157 |
+
"eval_cer": 1.0329540066382172,
|
| 158 |
+
"eval_loss": 0.4582904875278473,
|
| 159 |
+
"eval_runtime": 60.9484,
|
| 160 |
+
"eval_samples_per_second": 112.226,
|
| 161 |
+
"eval_steps_per_second": 14.028,
|
| 162 |
+
"eval_wer": 0.9997076023391813,
|
| 163 |
+
"step": 6160
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 7.386583285957931,
|
| 167 |
+
"grad_norm": 5.6131510734558105,
|
| 168 |
+
"learning_rate": 1.5588e-05,
|
| 169 |
+
"loss": 0.5497,
|
| 170 |
+
"step": 6500
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 7.955088118249005,
|
| 174 |
+
"grad_norm": 7.207092761993408,
|
| 175 |
+
"learning_rate": 1.6788e-05,
|
| 176 |
+
"loss": 0.403,
|
| 177 |
+
"step": 7000
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 8.0,
|
| 181 |
+
"eval_cer": 1.0478603603603605,
|
| 182 |
+
"eval_loss": 0.25122731924057007,
|
| 183 |
+
"eval_runtime": 62.9796,
|
| 184 |
+
"eval_samples_per_second": 108.607,
|
| 185 |
+
"eval_steps_per_second": 13.576,
|
| 186 |
+
"eval_wer": 0.9991228070175439,
|
| 187 |
+
"step": 7040
|
| 188 |
+
},
|
| 189 |
+
{
|
| 190 |
+
"epoch": 8.523024445707788,
|
| 191 |
+
"grad_norm": 1.6712586879730225,
|
| 192 |
+
"learning_rate": 1.7988e-05,
|
| 193 |
+
"loss": 0.3035,
|
| 194 |
+
"step": 7500
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 9.0,
|
| 198 |
+
"eval_cer": 1.0364805595068753,
|
| 199 |
+
"eval_loss": 0.19720108807086945,
|
| 200 |
+
"eval_runtime": 61.791,
|
| 201 |
+
"eval_samples_per_second": 110.696,
|
| 202 |
+
"eval_steps_per_second": 13.837,
|
| 203 |
+
"eval_wer": 0.9992690058479532,
|
| 204 |
+
"step": 7920
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 9.090960773166572,
|
| 208 |
+
"grad_norm": 5.482681751251221,
|
| 209 |
+
"learning_rate": 1.9188e-05,
|
| 210 |
+
"loss": 0.2585,
|
| 211 |
+
"step": 8000
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 9.659465605457646,
|
| 215 |
+
"grad_norm": 2.9059221744537354,
|
| 216 |
+
"learning_rate": 2.0388e-05,
|
| 217 |
+
"loss": 0.229,
|
| 218 |
+
"step": 8500
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 10.0,
|
| 222 |
+
"eval_cer": 1.026434329065908,
|
| 223 |
+
"eval_loss": 0.08719063550233841,
|
| 224 |
+
"eval_runtime": 64.0548,
|
| 225 |
+
"eval_samples_per_second": 106.784,
|
| 226 |
+
"eval_steps_per_second": 13.348,
|
| 227 |
+
"eval_wer": 0.9991228070175439,
|
| 228 |
+
"step": 8800
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 10.227401932916429,
|
| 232 |
+
"grad_norm": 4.925757884979248,
|
| 233 |
+
"learning_rate": 2.1588e-05,
|
| 234 |
+
"loss": 0.2205,
|
| 235 |
+
"step": 9000
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 10.795906765207505,
|
| 239 |
+
"grad_norm": 4.320206642150879,
|
| 240 |
+
"learning_rate": 2.2788000000000003e-05,
|
| 241 |
+
"loss": 0.1995,
|
| 242 |
+
"step": 9500
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 11.0,
|
| 246 |
+
"eval_cer": 1.0261972498814604,
|
| 247 |
+
"eval_loss": 0.09591592103242874,
|
| 248 |
+
"eval_runtime": 63.0889,
|
| 249 |
+
"eval_samples_per_second": 108.418,
|
| 250 |
+
"eval_steps_per_second": 13.552,
|
| 251 |
+
"eval_wer": 0.9988304093567252,
|
| 252 |
+
"step": 9680
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 11.363843092666288,
|
| 256 |
+
"grad_norm": 4.490440368652344,
|
| 257 |
+
"learning_rate": 2.3988e-05,
|
| 258 |
+
"loss": 0.1909,
|
| 259 |
+
"step": 10000
|
| 260 |
+
},
|
| 261 |
+
{
|
| 262 |
+
"epoch": 11.932347924957362,
|
| 263 |
+
"grad_norm": 4.226001739501953,
|
| 264 |
+
"learning_rate": 2.5188e-05,
|
| 265 |
+
"loss": 0.1824,
|
| 266 |
+
"step": 10500
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 12.0,
|
| 270 |
+
"eval_cer": 1.0316500711237553,
|
| 271 |
+
"eval_loss": 0.1011653020977974,
|
| 272 |
+
"eval_runtime": 60.7879,
|
| 273 |
+
"eval_samples_per_second": 112.522,
|
| 274 |
+
"eval_steps_per_second": 14.065,
|
| 275 |
+
"eval_wer": 0.9988304093567252,
|
| 276 |
+
"step": 10560
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 12.500284252416145,
|
| 280 |
+
"grad_norm": 4.8370466232299805,
|
| 281 |
+
"learning_rate": 2.6388000000000002e-05,
|
| 282 |
+
"loss": 0.1774,
|
| 283 |
+
"step": 11000
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 13.0,
|
| 287 |
+
"eval_cer": 1.0220483641536273,
|
| 288 |
+
"eval_loss": 0.05406388267874718,
|
| 289 |
+
"eval_runtime": 60.5368,
|
| 290 |
+
"eval_samples_per_second": 112.989,
|
| 291 |
+
"eval_steps_per_second": 14.124,
|
| 292 |
+
"eval_wer": 0.9991228070175439,
|
| 293 |
+
"step": 11440
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 13.068220579874929,
|
| 297 |
+
"grad_norm": 4.82689094543457,
|
| 298 |
+
"learning_rate": 2.7585600000000002e-05,
|
| 299 |
+
"loss": 0.1761,
|
| 300 |
+
"step": 11500
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"epoch": 13.636725412166003,
|
| 304 |
+
"grad_norm": 0.785372793674469,
|
| 305 |
+
"learning_rate": 2.87856e-05,
|
| 306 |
+
"loss": 0.1739,
|
| 307 |
+
"step": 12000
|
| 308 |
+
},
|
| 309 |
+
{
|
| 310 |
+
"epoch": 14.0,
|
| 311 |
+
"eval_cer": 1.026997392128971,
|
| 312 |
+
"eval_loss": 0.0703384280204773,
|
| 313 |
+
"eval_runtime": 60.3222,
|
| 314 |
+
"eval_samples_per_second": 113.391,
|
| 315 |
+
"eval_steps_per_second": 14.174,
|
| 316 |
+
"eval_wer": 0.9989766081871345,
|
| 317 |
+
"step": 12320
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 14.204661739624786,
|
| 321 |
+
"grad_norm": 2.4106199741363525,
|
| 322 |
+
"learning_rate": 2.99856e-05,
|
| 323 |
+
"loss": 0.1642,
|
| 324 |
+
"step": 12500
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 14.773166571915862,
|
| 328 |
+
"grad_norm": 2.8759591579437256,
|
| 329 |
+
"learning_rate": 2.996483380918142e-05,
|
| 330 |
+
"loss": 0.1609,
|
| 331 |
+
"step": 13000
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 15.0,
|
| 335 |
+
"eval_cer": 1.0202702702702702,
|
| 336 |
+
"eval_loss": 0.048034194856882095,
|
| 337 |
+
"eval_runtime": 61.6445,
|
| 338 |
+
"eval_samples_per_second": 110.959,
|
| 339 |
+
"eval_steps_per_second": 13.87,
|
| 340 |
+
"eval_wer": 0.9988304093567252,
|
| 341 |
+
"step": 13200
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 15.341102899374645,
|
| 345 |
+
"grad_norm": 5.875328540802002,
|
| 346 |
+
"learning_rate": 2.9857791176729968e-05,
|
| 347 |
+
"loss": 0.1583,
|
| 348 |
+
"step": 13500
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 15.90960773166572,
|
| 352 |
+
"grad_norm": 3.2317707538604736,
|
| 353 |
+
"learning_rate": 2.9679381078280773e-05,
|
| 354 |
+
"loss": 0.1512,
|
| 355 |
+
"step": 14000
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 16.0,
|
| 359 |
+
"eval_cer": 1.016239924134661,
|
| 360 |
+
"eval_loss": 0.053960736840963364,
|
| 361 |
+
"eval_runtime": 61.5915,
|
| 362 |
+
"eval_samples_per_second": 111.054,
|
| 363 |
+
"eval_steps_per_second": 13.882,
|
| 364 |
+
"eval_wer": 0.9988304093567252,
|
| 365 |
+
"step": 14080
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 16.4775440591245,
|
| 369 |
+
"grad_norm": 3.0755059719085693,
|
| 370 |
+
"learning_rate": 2.9430460483519525e-05,
|
| 371 |
+
"loss": 0.1412,
|
| 372 |
+
"step": 14500
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 17.0,
|
| 376 |
+
"eval_cer": 1.0187885253674727,
|
| 377 |
+
"eval_loss": 0.03960481286048889,
|
| 378 |
+
"eval_runtime": 61.6801,
|
| 379 |
+
"eval_samples_per_second": 110.895,
|
| 380 |
+
"eval_steps_per_second": 13.862,
|
| 381 |
+
"eval_wer": 0.9988304093567252,
|
| 382 |
+
"step": 14960
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 17.045480386583286,
|
| 386 |
+
"grad_norm": 1.7888600826263428,
|
| 387 |
+
"learning_rate": 2.911222505012316e-05,
|
| 388 |
+
"loss": 0.1411,
|
| 389 |
+
"step": 15000
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 17.61398521887436,
|
| 393 |
+
"grad_norm": 5.4633941650390625,
|
| 394 |
+
"learning_rate": 2.872704189552075e-05,
|
| 395 |
+
"loss": 0.1391,
|
| 396 |
+
"step": 15500
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 18.0,
|
| 400 |
+
"eval_cer": 1.0194997629208156,
|
| 401 |
+
"eval_loss": 0.04934508726000786,
|
| 402 |
+
"eval_runtime": 60.3205,
|
| 403 |
+
"eval_samples_per_second": 113.394,
|
| 404 |
+
"eval_steps_per_second": 14.174,
|
| 405 |
+
"eval_wer": 0.9988304093567252,
|
| 406 |
+
"step": 15840
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 18.181921546333143,
|
| 410 |
+
"grad_norm": 4.8292951583862305,
|
| 411 |
+
"learning_rate": 2.8275217996094984e-05,
|
| 412 |
+
"loss": 0.1363,
|
| 413 |
+
"step": 16000
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 18.75042637862422,
|
| 417 |
+
"grad_norm": 1.6074540615081787,
|
| 418 |
+
"learning_rate": 2.775962831495378e-05,
|
| 419 |
+
"loss": 0.1325,
|
| 420 |
+
"step": 16500
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 19.0,
|
| 424 |
+
"eval_cer": 1.0185810810810811,
|
| 425 |
+
"eval_loss": 0.03655907139182091,
|
| 426 |
+
"eval_runtime": 60.6228,
|
| 427 |
+
"eval_samples_per_second": 112.829,
|
| 428 |
+
"eval_steps_per_second": 14.104,
|
| 429 |
+
"eval_wer": 0.9988304093567252,
|
| 430 |
+
"step": 16720
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 19.318362706083,
|
| 434 |
+
"grad_norm": 2.9544260501861572,
|
| 435 |
+
"learning_rate": 2.7182749420020325e-05,
|
| 436 |
+
"loss": 0.1243,
|
| 437 |
+
"step": 17000
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 19.886867538374077,
|
| 441 |
+
"grad_norm": 1.7803765535354614,
|
| 442 |
+
"learning_rate": 2.6547352273978724e-05,
|
| 443 |
+
"loss": 0.1242,
|
| 444 |
+
"step": 17500
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 20.0,
|
| 448 |
+
"eval_cer": 1.0178105737316263,
|
| 449 |
+
"eval_loss": 0.03915562480688095,
|
| 450 |
+
"eval_runtime": 63.37,
|
| 451 |
+
"eval_samples_per_second": 107.937,
|
| 452 |
+
"eval_steps_per_second": 13.492,
|
| 453 |
+
"eval_wer": 0.9988304093567252,
|
| 454 |
+
"step": 17600
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 20.454803865832858,
|
| 458 |
+
"grad_norm": 7.015552520751953,
|
| 459 |
+
"learning_rate": 2.5857923843413123e-05,
|
| 460 |
+
"loss": 0.122,
|
| 461 |
+
"step": 18000
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 21.0,
|
| 465 |
+
"eval_cer": 1.0192923186344238,
|
| 466 |
+
"eval_loss": 0.05453035235404968,
|
| 467 |
+
"eval_runtime": 62.1534,
|
| 468 |
+
"eval_samples_per_second": 110.05,
|
| 469 |
+
"eval_steps_per_second": 13.756,
|
| 470 |
+
"eval_wer": 0.9988304093567252,
|
| 471 |
+
"step": 18480
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 21.022740193291643,
|
| 475 |
+
"grad_norm": 0.5277901887893677,
|
| 476 |
+
"learning_rate": 2.511654911570264e-05,
|
| 477 |
+
"loss": 0.1154,
|
| 478 |
+
"step": 18500
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 21.59124502558272,
|
| 482 |
+
"grad_norm": 2.2439563274383545,
|
| 483 |
+
"learning_rate": 2.432514615070941e-05,
|
| 484 |
+
"loss": 0.1143,
|
| 485 |
+
"step": 19000
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 22.0,
|
| 489 |
+
"eval_cer": 1.0184625414888573,
|
| 490 |
+
"eval_loss": 0.04077836126089096,
|
| 491 |
+
"eval_runtime": 62.3438,
|
| 492 |
+
"eval_samples_per_second": 109.714,
|
| 493 |
+
"eval_steps_per_second": 13.714,
|
| 494 |
+
"eval_wer": 0.9988304093567252,
|
| 495 |
+
"step": 19360
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 22.1591813530415,
|
| 499 |
+
"grad_norm": 3.2656147480010986,
|
| 500 |
+
"learning_rate": 2.3488951059960833e-05,
|
| 501 |
+
"loss": 0.108,
|
| 502 |
+
"step": 19500
|
| 503 |
+
},
|
| 504 |
+
{
|
| 505 |
+
"epoch": 22.727686185332576,
|
| 506 |
+
"grad_norm": 1.6540584564208984,
|
| 507 |
+
"learning_rate": 2.261198039773451e-05,
|
| 508 |
+
"loss": 0.1087,
|
| 509 |
+
"step": 20000
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 23.0,
|
| 513 |
+
"eval_cer": 1.0176031294452348,
|
| 514 |
+
"eval_loss": 0.03096814453601837,
|
| 515 |
+
"eval_runtime": 61.14,
|
| 516 |
+
"eval_samples_per_second": 111.874,
|
| 517 |
+
"eval_steps_per_second": 13.984,
|
| 518 |
+
"eval_wer": 0.9988304093567252,
|
| 519 |
+
"step": 20240
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 23.295622512791358,
|
| 523 |
+
"grad_norm": 0.029715025797486305,
|
| 524 |
+
"learning_rate": 2.1698446578458188e-05,
|
| 525 |
+
"loss": 0.1011,
|
| 526 |
+
"step": 20500
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 23.864127345082434,
|
| 530 |
+
"grad_norm": 1.3068634271621704,
|
| 531 |
+
"learning_rate": 2.0752737642925386e-05,
|
| 532 |
+
"loss": 0.1013,
|
| 533 |
+
"step": 21000
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 24.0,
|
| 537 |
+
"eval_cer": 1.0165955429113325,
|
| 538 |
+
"eval_loss": 0.02615249529480934,
|
| 539 |
+
"eval_runtime": 60.7558,
|
| 540 |
+
"eval_samples_per_second": 112.582,
|
| 541 |
+
"eval_steps_per_second": 14.073,
|
| 542 |
+
"eval_wer": 0.9988304093567252,
|
| 543 |
+
"step": 21120
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 24.432063672541215,
|
| 547 |
+
"grad_norm": 2.296964168548584,
|
| 548 |
+
"learning_rate": 1.9779396180912585e-05,
|
| 549 |
+
"loss": 0.0952,
|
| 550 |
+
"step": 21500
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 25.0,
|
| 554 |
+
"grad_norm": 0.6059958934783936,
|
| 555 |
+
"learning_rate": 1.8783097511440484e-05,
|
| 556 |
+
"loss": 0.0998,
|
| 557 |
+
"step": 22000
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 25.0,
|
| 561 |
+
"eval_cer": 1.0199442863916548,
|
| 562 |
+
"eval_loss": 0.0387667752802372,
|
| 563 |
+
"eval_runtime": 62.9295,
|
| 564 |
+
"eval_samples_per_second": 108.693,
|
| 565 |
+
"eval_steps_per_second": 13.587,
|
| 566 |
+
"eval_wer": 0.9988304093567252,
|
| 567 |
+
"step": 22000
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 25.568504832291076,
|
| 571 |
+
"grad_norm": 3.430413007736206,
|
| 572 |
+
"learning_rate": 1.777067107469613e-05,
|
| 573 |
+
"loss": 0.0903,
|
| 574 |
+
"step": 22500
|
| 575 |
+
},
|
| 576 |
+
{
|
| 577 |
+
"epoch": 26.0,
|
| 578 |
+
"eval_cer": 1.0166251778093884,
|
| 579 |
+
"eval_loss": 0.027963772416114807,
|
| 580 |
+
"eval_runtime": 62.2122,
|
| 581 |
+
"eval_samples_per_second": 109.946,
|
| 582 |
+
"eval_steps_per_second": 13.743,
|
| 583 |
+
"eval_wer": 0.9988304093567252,
|
| 584 |
+
"step": 22880
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 26.136441159749857,
|
| 588 |
+
"grad_norm": 3.037529706954956,
|
| 589 |
+
"learning_rate": 1.6742923736196817e-05,
|
| 590 |
+
"loss": 0.0867,
|
| 591 |
+
"step": 23000
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 26.704945992040933,
|
| 595 |
+
"grad_norm": 2.661895751953125,
|
| 596 |
+
"learning_rate": 1.5706804490393117e-05,
|
| 597 |
+
"loss": 0.088,
|
| 598 |
+
"step": 23500
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 27.0,
|
| 602 |
+
"eval_cer": 1.0197072072072073,
|
| 603 |
+
"eval_loss": 0.04922711104154587,
|
| 604 |
+
"eval_runtime": 68.8335,
|
| 605 |
+
"eval_samples_per_second": 99.37,
|
| 606 |
+
"eval_steps_per_second": 12.421,
|
| 607 |
+
"eval_wer": 0.9988304093567252,
|
| 608 |
+
"step": 23760
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 27.272882319499715,
|
| 612 |
+
"grad_norm": 2.580177068710327,
|
| 613 |
+
"learning_rate": 1.4667290201218887e-05,
|
| 614 |
+
"loss": 0.0874,
|
| 615 |
+
"step": 24000
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 27.84138715179079,
|
| 619 |
+
"grad_norm": 6.204455375671387,
|
| 620 |
+
"learning_rate": 1.3629374040256936e-05,
|
| 621 |
+
"loss": 0.0838,
|
| 622 |
+
"step": 24500
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 28.0,
|
| 626 |
+
"eval_cer": 1.0162991939307728,
|
| 627 |
+
"eval_loss": 0.02296304889023304,
|
| 628 |
+
"eval_runtime": 65.6917,
|
| 629 |
+
"eval_samples_per_second": 104.123,
|
| 630 |
+
"eval_steps_per_second": 13.015,
|
| 631 |
+
"eval_wer": 0.9988304093567252,
|
| 632 |
+
"step": 24640
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 28.409323479249572,
|
| 636 |
+
"grad_norm": 0.0685189813375473,
|
| 637 |
+
"learning_rate": 1.2600094296980161e-05,
|
| 638 |
+
"loss": 0.0826,
|
| 639 |
+
"step": 25000
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 28.977828311540648,
|
| 643 |
+
"grad_norm": 6.252466678619385,
|
| 644 |
+
"learning_rate": 1.1580271268352735e-05,
|
| 645 |
+
"loss": 0.079,
|
| 646 |
+
"step": 25500
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 29.0,
|
| 650 |
+
"eval_cer": 1.0169511616880038,
|
| 651 |
+
"eval_loss": 0.0281895250082016,
|
| 652 |
+
"eval_runtime": 61.0,
|
| 653 |
+
"eval_samples_per_second": 112.131,
|
| 654 |
+
"eval_steps_per_second": 14.016,
|
| 655 |
+
"eval_wer": 0.9988304093567252,
|
| 656 |
+
"step": 25520
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 29.545764638999433,
|
| 660 |
+
"grad_norm": 3.142706871032715,
|
| 661 |
+
"learning_rate": 1.0576874461569077e-05,
|
| 662 |
+
"loss": 0.0747,
|
| 663 |
+
"step": 26000
|
| 664 |
+
},
|
| 665 |
+
{
|
| 666 |
+
"epoch": 30.0,
|
| 667 |
+
"eval_cer": 1.016239924134661,
|
| 668 |
+
"eval_loss": 0.027105851098895073,
|
| 669 |
+
"eval_runtime": 59.5418,
|
| 670 |
+
"eval_samples_per_second": 114.877,
|
| 671 |
+
"eval_steps_per_second": 14.36,
|
| 672 |
+
"eval_wer": 0.9988304093567252,
|
| 673 |
+
"step": 26400
|
| 674 |
+
},
|
| 675 |
+
{
|
| 676 |
+
"epoch": 30.113700966458214,
|
| 677 |
+
"grad_norm": 3.5694313049316406,
|
| 678 |
+
"learning_rate": 9.594723562586447e-06,
|
| 679 |
+
"loss": 0.0774,
|
| 680 |
+
"step": 26500
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 30.68220579874929,
|
| 684 |
+
"grad_norm": 5.079456329345703,
|
| 685 |
+
"learning_rate": 8.640419592752059e-06,
|
| 686 |
+
"loss": 0.0692,
|
| 687 |
+
"step": 27000
|
| 688 |
+
},
|
| 689 |
+
{
|
| 690 |
+
"epoch": 31.0,
|
| 691 |
+
"eval_cer": 1.0166844476055001,
|
| 692 |
+
"eval_loss": 0.027171434834599495,
|
| 693 |
+
"eval_runtime": 59.3002,
|
| 694 |
+
"eval_samples_per_second": 115.345,
|
| 695 |
+
"eval_steps_per_second": 14.418,
|
| 696 |
+
"eval_wer": 0.9988304093567252,
|
| 697 |
+
"step": 27280
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 31.250142126208072,
|
| 701 |
+
"grad_norm": 2.114372730255127,
|
| 702 |
+
"learning_rate": 7.714723096178886e-06,
|
| 703 |
+
"loss": 0.0718,
|
| 704 |
+
"step": 27500
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 31.818646958499148,
|
| 708 |
+
"grad_norm": 3.835890531539917,
|
| 709 |
+
"learning_rate": 6.824020478947078e-06,
|
| 710 |
+
"loss": 0.0699,
|
| 711 |
+
"step": 28000
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 32.0,
|
| 715 |
+
"eval_cer": 1.0143432906590801,
|
| 716 |
+
"eval_loss": 0.0426531545817852,
|
| 717 |
+
"eval_runtime": 58.5893,
|
| 718 |
+
"eval_samples_per_second": 116.745,
|
| 719 |
+
"eval_steps_per_second": 14.593,
|
| 720 |
+
"eval_wer": 0.9988304093567252,
|
| 721 |
+
"step": 28160
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 32.38658328595793,
|
| 725 |
+
"grad_norm": 3.8743183612823486,
|
| 726 |
+
"learning_rate": 5.97259011514287e-06,
|
| 727 |
+
"loss": 0.0668,
|
| 728 |
+
"step": 28500
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 32.955088118249,
|
| 732 |
+
"grad_norm": 0.82483971118927,
|
| 733 |
+
"learning_rate": 5.164521739694928e-06,
|
| 734 |
+
"loss": 0.0652,
|
| 735 |
+
"step": 29000
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 33.0,
|
| 739 |
+
"eval_cer": 1.016151019440493,
|
| 740 |
+
"eval_loss": 0.032351747155189514,
|
| 741 |
+
"eval_runtime": 60.0164,
|
| 742 |
+
"eval_samples_per_second": 113.969,
|
| 743 |
+
"eval_steps_per_second": 14.246,
|
| 744 |
+
"eval_wer": 0.9988304093567252,
|
| 745 |
+
"step": 29040
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 33.52302444570779,
|
| 749 |
+
"grad_norm": 0.08226080983877182,
|
| 750 |
+
"learning_rate": 4.403696803864931e-06,
|
| 751 |
+
"loss": 0.0624,
|
| 752 |
+
"step": 29500
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 34.0,
|
| 756 |
+
"eval_cer": 1.0162991939307728,
|
| 757 |
+
"eval_loss": 0.03149048984050751,
|
| 758 |
+
"eval_runtime": 58.4368,
|
| 759 |
+
"eval_samples_per_second": 117.05,
|
| 760 |
+
"eval_steps_per_second": 14.631,
|
| 761 |
+
"eval_wer": 0.9988304093567252,
|
| 762 |
+
"step": 29920
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 34.09096077316657,
|
| 766 |
+
"grad_norm": 0.9844266176223755,
|
| 767 |
+
"learning_rate": 3.6951365800521325e-06,
|
| 768 |
+
"loss": 0.0618,
|
| 769 |
+
"step": 30000
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 34.65946560545765,
|
| 773 |
+
"grad_norm": 2.4211599826812744,
|
| 774 |
+
"learning_rate": 3.039405763913186e-06,
|
| 775 |
+
"loss": 0.0588,
|
| 776 |
+
"step": 30500
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 35.0,
|
| 780 |
+
"eval_cer": 1.0136913229018492,
|
| 781 |
+
"eval_loss": 0.054938483983278275,
|
| 782 |
+
"eval_runtime": 59.6299,
|
| 783 |
+
"eval_samples_per_second": 114.708,
|
| 784 |
+
"eval_steps_per_second": 14.338,
|
| 785 |
+
"eval_wer": 0.9988304093567252,
|
| 786 |
+
"step": 30800
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 35.22740193291643,
|
| 790 |
+
"grad_norm": 3.788679599761963,
|
| 791 |
+
"learning_rate": 2.4411261053725335e-06,
|
| 792 |
+
"loss": 0.059,
|
| 793 |
+
"step": 31000
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 35.795906765207505,
|
| 797 |
+
"grad_norm": 2.701200008392334,
|
| 798 |
+
"learning_rate": 1.904185301084242e-06,
|
| 799 |
+
"loss": 0.0594,
|
| 800 |
+
"step": 31500
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 36.0,
|
| 804 |
+
"eval_cer": 1.0141654812707444,
|
| 805 |
+
"eval_loss": 0.045684415847063065,
|
| 806 |
+
"eval_runtime": 60.0287,
|
| 807 |
+
"eval_samples_per_second": 113.945,
|
| 808 |
+
"eval_steps_per_second": 14.243,
|
| 809 |
+
"eval_wer": 0.9988304093567252,
|
| 810 |
+
"step": 31680
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"epoch": 36.363843092666286,
|
| 814 |
+
"grad_norm": 0.22563552856445312,
|
| 815 |
+
"learning_rate": 1.4290112725289179e-06,
|
| 816 |
+
"loss": 0.0557,
|
| 817 |
+
"step": 32000
|
| 818 |
+
},
|
| 819 |
+
{
|
| 820 |
+
"epoch": 36.93234792495736,
|
| 821 |
+
"grad_norm": 3.3882081508636475,
|
| 822 |
+
"learning_rate": 1.0190237218990893e-06,
|
| 823 |
+
"loss": 0.0619,
|
| 824 |
+
"step": 32500
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 37.0,
|
| 828 |
+
"eval_cer": 1.014402560455192,
|
| 829 |
+
"eval_loss": 0.046258434653282166,
|
| 830 |
+
"eval_runtime": 58.5233,
|
| 831 |
+
"eval_samples_per_second": 116.876,
|
| 832 |
+
"eval_steps_per_second": 14.61,
|
| 833 |
+
"eval_wer": 0.9988304093567252,
|
| 834 |
+
"step": 32560
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 37.500284252416144,
|
| 838 |
+
"grad_norm": 4.180235385894775,
|
| 839 |
+
"learning_rate": 6.761919710294118e-07,
|
| 840 |
+
"loss": 0.058,
|
| 841 |
+
"step": 33000
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"epoch": 38.0,
|
| 845 |
+
"eval_cer": 1.0126837363679468,
|
| 846 |
+
"eval_loss": 0.06653982400894165,
|
| 847 |
+
"eval_runtime": 59.3473,
|
| 848 |
+
"eval_samples_per_second": 115.254,
|
| 849 |
+
"eval_steps_per_second": 14.407,
|
| 850 |
+
"eval_wer": 0.9988304093567252,
|
| 851 |
+
"step": 33440
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 38.06822057987493,
|
| 855 |
+
"grad_norm": 0.7227972745895386,
|
| 856 |
+
"learning_rate": 4.021627676115197e-07,
|
| 857 |
+
"loss": 0.0579,
|
| 858 |
+
"step": 33500
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 38.636725412166,
|
| 862 |
+
"grad_norm": 0.30214107036590576,
|
| 863 |
+
"learning_rate": 1.9825237525585017e-07,
|
| 864 |
+
"loss": 0.059,
|
| 865 |
+
"step": 34000
|
| 866 |
+
},
|
| 867 |
+
{
|
| 868 |
+
"epoch": 39.0,
|
| 869 |
+
"eval_cer": 1.01309862494073,
|
| 870 |
+
"eval_loss": 0.05947383493185043,
|
| 871 |
+
"eval_runtime": 69.1876,
|
| 872 |
+
"eval_samples_per_second": 98.862,
|
| 873 |
+
"eval_steps_per_second": 12.358,
|
| 874 |
+
"eval_wer": 0.9988304093567252,
|
| 875 |
+
"step": 34320
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 39.20466173962479,
|
| 879 |
+
"grad_norm": 4.248073577880859,
|
| 880 |
+
"learning_rate": 6.544025099069761e-08,
|
| 881 |
+
"loss": 0.0539,
|
| 882 |
+
"step": 34500
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 39.77316657191586,
|
| 886 |
+
"grad_norm": 1.873940348625183,
|
| 887 |
+
"learning_rate": 4.364340567880043e-09,
|
| 888 |
+
"loss": 0.0563,
|
| 889 |
+
"step": 35000
|
| 890 |
+
},
|
| 891 |
+
{
|
| 892 |
+
"epoch": 39.955088118249,
|
| 893 |
+
"eval_cer": 1.013276434329066,
|
| 894 |
+
"eval_loss": 0.058061111718416214,
|
| 895 |
+
"eval_runtime": 61.3482,
|
| 896 |
+
"eval_samples_per_second": 111.495,
|
| 897 |
+
"eval_steps_per_second": 13.937,
|
| 898 |
+
"eval_wer": 0.9988304093567252,
|
| 899 |
+
"step": 35160
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 39.955088118249,
|
| 903 |
+
"step": 35160,
|
| 904 |
+
"total_flos": 1.8269796433195942e+19,
|
| 905 |
+
"train_loss": 1.1798896302406563,
|
| 906 |
+
"train_runtime": 28544.574,
|
| 907 |
+
"train_samples_per_second": 78.866,
|
| 908 |
+
"train_steps_per_second": 1.232
|
| 909 |
+
}
|
| 910 |
+
],
|
| 911 |
+
"logging_steps": 500,
|
| 912 |
+
"max_steps": 35160,
|
| 913 |
+
"num_input_tokens_seen": 0,
|
| 914 |
+
"num_train_epochs": 40,
|
| 915 |
+
"save_steps": 400,
|
| 916 |
+
"stateful_callbacks": {
|
| 917 |
+
"TrainerControl": {
|
| 918 |
+
"args": {
|
| 919 |
+
"should_epoch_stop": false,
|
| 920 |
+
"should_evaluate": false,
|
| 921 |
+
"should_log": false,
|
| 922 |
+
"should_save": true,
|
| 923 |
+
"should_training_stop": true
|
| 924 |
+
},
|
| 925 |
+
"attributes": {}
|
| 926 |
+
}
|
| 927 |
+
},
|
| 928 |
+
"total_flos": 1.8269796433195942e+19,
|
| 929 |
+
"train_batch_size": 32,
|
| 930 |
+
"trial_name": null,
|
| 931 |
+
"trial_params": null
|
| 932 |
+
}
|