Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,20 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: apache-2.0
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
language:
|
| 4 |
+
- ru
|
| 5 |
+
library_name: transformers
|
| 6 |
+
pipeline_tag: automatic-speech-recognition
|
| 7 |
+
base_model: openai/whisper-base
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
# Whisper-base-ru-pruned
|
| 11 |
+
This is a pruned version of openai/whisper-base model with only russian tokens left.
|
| 12 |
+
Pruning was made without any fine-tuning. Method from [this post](https://medium.com/m/global-identity-2?redirectUrl=https%3A%2F%2Ftowardsdatascience.com%2Fhow-to-adapt-a-multilingual-t5-model-for-a-single-language-b9f94f3d9c90) was used.
|
| 13 |
+
|
| 14 |
+
Model size is 30% less then original whisper-base:
|
| 15 |
+
| | openai/whisper-base | waveletdeboshir/whisper-base-ru-pruned |
|
| 16 |
+
| :------ | :------ | :------ |
|
| 17 |
+
| n of parameters | 74 M | 48.5 M |
|
| 18 |
+
| n of parameters (with proj_out layer) | 99 M | 51 M |
|
| 19 |
+
| model file size | 290 Mb | 203 Mb |
|
| 20 |
+
|