Commit
•
b329385
1
Parent(s):
7bdb481
Update README.md
Browse files
README.md
CHANGED
|
@@ -19,9 +19,11 @@ This is a pruned version of [openai/whisper-base](https://huggingface.co/openai/
|
|
| 19 |
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.
|
| 20 |
|
| 21 |
## Size
|
|
|
|
| 22 |
Model size is 30% less then original whisper-base:
|
| 23 |
| | openai/whisper-base | waveletdeboshir/whisper-base-ru-pruned |
|
| 24 |
| :------ | :------ | :------ |
|
| 25 |
| n of parameters | 74 M | 48.5 M |
|
| 26 |
| n of parameters (with proj_out layer) | 99 M | 51 M |
|
| 27 |
-
| model file size | 290 Mb | 203 Mb |
|
|
|
|
|
|
| 19 |
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.
|
| 20 |
|
| 21 |
## Size
|
| 22 |
+
Only 10% tokens was left: special whisper tokens, added whisper tokens, 100 most popular tokens from tokenizer and 3000 most popular Russian tokens computed by tokenization of russian text corpus.
|
| 23 |
Model size is 30% less then original whisper-base:
|
| 24 |
| | openai/whisper-base | waveletdeboshir/whisper-base-ru-pruned |
|
| 25 |
| :------ | :------ | :------ |
|
| 26 |
| n of parameters | 74 M | 48.5 M |
|
| 27 |
| n of parameters (with proj_out layer) | 99 M | 51 M |
|
| 28 |
+
| model file size | 290 Mb | 203 Mb |
|
| 29 |
+
| vocab_size | 51865 | 4705 |
|