Purfview
Purfview
> [@Purfview](https://github.com/Purfview) I'm a bit confused, you state the following: "In the next version, I will change the compute_type from 'auto' to 'default' and it will be resolved as 'float16'...
Fixed, currently in Pro r3.256.1
> okay first of all, there is the problem that the condition on previous text param is not passed to the transcribe call, the second issue is the mentioned repeat...
Yes, I was thinking to rename it to `vad_timestamps` for batched, but it's "cleaner" as is I think. Maybe just add the notes at the descriptions that it needs different...
>We could change both to Optional[List[dict]] clip_timestamps in sequential mode is for a quick user input, not for "self-defined VAD", btw, you can't have this functionality in batched mode as...
If you change the sequential one then it will lose original functionality which is to replicate the vanilla Whisper's clip_timestamps .
>yep I try it .but the folder doesn't has any file about cudnn_ops64_9.dll. In your screenshot it's cudnn v8, and it's for cuda 11. You need cudnn v9 for cuda...
https://developer.nvidia.com/cublas OR https://github.com/Purfview/whisper-standalone-win/releases/tag/libs
Last week Colab switched from cudnn v8 to cudnn v9. Try CTranslate2==4.5.0 or install cudnn v8.
Try older huggingface-hub: `pip install huggingface-hub==0.19.4`