update model
Browse files- README.md +3 -3
- config.json +2 -2
- distil-whisper/events.out.tfevents.1705422649.c175d7640af5.1725615.0 +3 -0
- model.safetensors +2 -2
README.md
CHANGED
@@ -5,7 +5,7 @@ license: mit
|
|
5 |
## Distilled Medium Whisper ASR Model for Thai
|
6 |
|
7 |
### Model Description
|
8 |
-
This is a distilled Automatic Speech Recognition (ASR) model, based on the Whisper architecture. It has been specifically tailored for Thai language speech recognition. The model features
|
9 |
|
10 |
#### Distillation Details
|
11 |
- **Teacher Model**: Medium Whisper ASR model
|
@@ -18,10 +18,10 @@ This is a distilled Automatic Speech Recognition (ASR) model, based on the Whisp
|
|
18 |
|
19 |
### Model Performance
|
20 |
- **DeepCut Tokenized WER on Common Voice 13 Test Set**:
|
21 |
-
- Distilled Model: **
|
22 |
- Teacher Model: **8.92%**
|
23 |
|
24 |
-
|
25 |
|
26 |
### Intended Use
|
27 |
This model is intended for use in applications requiring Thai language speech recognition.
|
|
|
5 |
## Distilled Medium Whisper ASR Model for Thai
|
6 |
|
7 |
### Model Description
|
8 |
+
This is a distilled Automatic Speech Recognition (ASR) model, based on the Whisper architecture. It has been specifically tailored for Thai language speech recognition. The model features 4 decoder layers (vs 24 in teacher model) and has been distilled from a larger teacher model, focusing on enhancing performance and efficiency.
|
9 |
|
10 |
#### Distillation Details
|
11 |
- **Teacher Model**: Medium Whisper ASR model
|
|
|
18 |
|
19 |
### Model Performance
|
20 |
- **DeepCut Tokenized WER on Common Voice 13 Test Set**:
|
21 |
+
- Distilled Model: **10.49%**
|
22 |
- Teacher Model: **8.92%**
|
23 |
|
24 |
+
Additional datasets for distillation or more decoder layers might improve the WER. More to come soon!
|
25 |
|
26 |
### Intended Use
|
27 |
This model is intended for use in applications requiring Thai language speech recognition.
|
config.json
CHANGED
@@ -1,5 +1,5 @@
|
|
1 |
{
|
2 |
-
"_name_or_path": "
|
3 |
"activation_dropout": 0.0,
|
4 |
"activation_function": "gelu",
|
5 |
"apply_spec_augment": true,
|
@@ -17,7 +17,7 @@
|
|
17 |
"decoder_attention_heads": 16,
|
18 |
"decoder_ffn_dim": 4096,
|
19 |
"decoder_layerdrop": 0.0,
|
20 |
-
"decoder_layers":
|
21 |
"decoder_start_token_id": 50258,
|
22 |
"dropout": 0.0,
|
23 |
"encoder_attention_heads": 16,
|
|
|
1 |
{
|
2 |
+
"_name_or_path": "./distil-whisper-medium-init",
|
3 |
"activation_dropout": 0.0,
|
4 |
"activation_function": "gelu",
|
5 |
"apply_spec_augment": true,
|
|
|
17 |
"decoder_attention_heads": 16,
|
18 |
"decoder_ffn_dim": 4096,
|
19 |
"decoder_layerdrop": 0.0,
|
20 |
+
"decoder_layers": 4,
|
21 |
"decoder_start_token_id": 50258,
|
22 |
"dropout": 0.0,
|
23 |
"encoder_attention_heads": 16,
|
distil-whisper/events.out.tfevents.1705422649.c175d7640af5.1725615.0
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:59683a04b97cfdaebbab875018dc06dd25b96e82190613b672dee6676d497fce
|
3 |
+
size 6288
|
model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:280d23deb47a68a4173948b4759cc2d02884df78ef198594560c95a7e48fac10
|
3 |
+
size 1711916448
|