Update README.md
Browse files
README.md
CHANGED
|
@@ -62,6 +62,23 @@ Also, currently whisper.cpp and faster-whisper support the [sequential long-form
|
|
| 62 |
and only Huggingface pipeline supports the [chunked long-form decoding](https://huggingface.co/distil-whisper/distil-large-v3#chunked-long-form), which we empirically
|
| 63 |
found better than the sequnential long-form decoding.
|
| 64 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 65 |
### Quantized Model
|
| 66 |
To use the quantized model, download the quantized GGML weights:
|
| 67 |
|
|
|
|
| 62 |
and only Huggingface pipeline supports the [chunked long-form decoding](https://huggingface.co/distil-whisper/distil-large-v3#chunked-long-form), which we empirically
|
| 63 |
found better than the sequnential long-form decoding.
|
| 64 |
|
| 65 |
+
### Conversion details
|
| 66 |
+
The original model was converted with the following command:
|
| 67 |
+
|
| 68 |
+
```
|
| 69 |
+
# clone OpenAI whisper and whisper.cpp
|
| 70 |
+
git clone https://github.com/openai/whisper
|
| 71 |
+
git clone https://github.com/ggerganov/whisper.cpp
|
| 72 |
+
|
| 73 |
+
# get the models
|
| 74 |
+
cd whisper.cpp/models
|
| 75 |
+
git clone https://huggingface.co/kotoba-tech/kotoba-whisper-v1.0
|
| 76 |
+
|
| 77 |
+
# convert to ggml
|
| 78 |
+
python3 ./convert-h5-to-ggml.py ./kotoba-whisper-v1.0/ ../../whisper .
|
| 79 |
+
mv ggml-model.bin ggml-kotoba-whisper-v1.0
|
| 80 |
+
```
|
| 81 |
+
|
| 82 |
### Quantized Model
|
| 83 |
To use the quantized model, download the quantized GGML weights:
|
| 84 |
|