optimum-habana icon indicating copy to clipboard operation
optimum-habana copied to clipboard

Published 20 hours ago • verified publisher icon huggingface

Add SVD pipeline

Open dsocek opened this issue 11 months ago • 1 comments

What does this PR do?

  • implements Stable Video Diffusion (SVD) pipeline for Gaudi
  • adds example SVD inference script and README
  • adds SVD related CI tests

dsocek avatar Feb 27 '24 14:02 dsocek

Note: Quality of generated video is really poor with classifier_free_guidance enabled. I override it to False and quality looks good.

dsocek avatar Apr 03 '24 01:04 dsocek

@libinta Updated as follows:

  • Migrated SVD to stable_diffusion folder
  • Added reference for source file in original diffusers
  • Removed default image to None

dsocek avatar May 14 '24 16:05 dsocek

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

HuggingFaceDocBuilderDev avatar May 21 '24 13:05 HuggingFaceDocBuilderDev

@regisss Thanks for review, yes for now HPU graph capture disabled. Fixed all other indicated issues in the amendment.

dsocek avatar May 21 '24 16:05 dsocek

@libinta @regisss Replaced CPU-based randn_tensor call with HPU-based (tested and output video looks good)

dsocek avatar May 21 '24 17:05 dsocek

@dsocek LGTM! Could you also add a test in tests/test_diffusers.py to test the throughput and that the example works please?

regisss avatar May 21 '24 22:05 regisss

@dsocek LGTM! Could you also add a test in tests/test_diffusers.py to test the throughput and that the example works please?

@regisss added SVD test

dsocek avatar May 22 '24 01:05 dsocek