Problems with cuda...

#1
by Kelmeilia - opened

Hello and thanks for your efforts...

I am a beginner in these things, so bear with me if the questions are stupid;

I guess that there is no finnish whisper model in huggingface format, that I could use following this example:
https://huggingface.co/docs/transformers/model_doc/whisper

So as openai is not an option for me, I tried this whisperx format model.

However, I keep getting errors with cuda, even though I have installed the nvidia toolkit.

The latest problem is
RuntimeError: Library cublas64_12.dll is not found or cannot be loaded

I don't know if I need some even more tools from nvidia, butt I couldn't get them without nvidia account.

As I am a beginner, is there any advice how to proceed further? Bang further with nvidia tools, or take a different approach - maybe there is a huggingface type model, that I just don't see? the gglm format is not familiar to me, so there should probably be a really simple example to follow, if I would try that..

thanks!

Finnish-NLP org
edited Apr 1

Hi @Kelmeilia I believe the Huggingface format of the whisper-large-finnish-v3 is this one: https://huggingface.co/Finnish-NLP/whisper-large-finnish-v3 so you could try that

Sign up or log in to comment