Skip to content

[tests] tests for compilation + quantization (bnb) #11672

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 8 commits into
base: main
Choose a base branch
from

Conversation

sayakpaul
Copy link
Member

@sayakpaul sayakpaul commented Jun 6, 2025

What does this PR do?

Adds tests for

  • quant + compilation
  • quant + compilation + model CPU offloading
  • quant + compilation + group offloading

Does this for bitsandbytes for now.

@sayakpaul sayakpaul added quantization performance Anything related to performance improvements, profiling and benchmarking torch.compile labels Jun 6, 2025
@HuggingFaceDocBuilderDev

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.

@sayakpaul sayakpaul requested a review from matthewdouglas June 6, 2025 09:36
@sayakpaul sayakpaul changed the title [wip][tests] start adding tests for compilation + quantization [tests] tests for compilation + quantization (bnb) Jun 7, 2025
@sayakpaul sayakpaul marked this pull request as ready for review June 7, 2025 04:45
@sayakpaul sayakpaul requested review from DN6 and matthewdouglas June 7, 2025 04:45
@sayakpaul
Copy link
Member Author

@DN6 LMK what you think of the test suite. The combinations target consumer GPU where using quantization is beneficial (cc: @asomoza). Would you be able to add this for GGUF, too?

Also cc: @stevhliu I think we should try to document these combos of optims in an easy manner now that we know they work (I can help get latency and memory numbers).

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
performance Anything related to performance improvements, profiling and benchmarking quantization torch.compile
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants