GPU Support for transcription

#14
by varunkuntal - opened

Hello and thanks for the excellent library!

I used the model for transcribing tasks and observed it defaults to CPU usage.

I'd like it to utilize the dedicated GPU (Nvidia Quadro A3000 GPU on an HP G8 workstation) for better performance. I tried to:

  1. Enable hardware acceleration in Chrome's settings and
  2. Setting Chrome to launch with the Nvidia control panel using the dedicated graphics processor,

but the model continues to operate on CPU.

Could you guide me on how to configure the setup to use the GPU? Am I missing any steps or configurations?

Thank you!

Hi there! This is currently being worked on in Transformers.js v3: https://github.com/xenova/transformers.js/pull/545

Sign up or log in to comment