Skip to content
Permalink

Comparing changes

Choose two branches to see what’s changed or to start a new pull request. If you need to, you can also or learn more about diff comparisons.

Open a pull request

Create a new pull request by comparing changes across two branches. If you need to, you can also . Learn more about diff comparisons here.
base repository: huggingface/diffusers
Failed to load repositories. Confirm that selected base ref is valid, then try again.
Loading
base: main
Choose a base ref
...
head repository: sym-bot/diffusers
Failed to load repositories. Confirm that selected head ref is valid, then try again.
Loading
compare: main
Choose a head ref
Checking mergeability… Don’t worry, you can still create the pull request.
  • 2 commits
  • 2 files changed
  • 2 contributors

Commits on Jan 31, 2026

  1. fix: graceful fallback for attention backend import failures 

    Add try-except wrappers around all external attention backend imports
(flash_attn, flash_attn_3, aiter, sageattention, flex_attention, torch_npu,
torch_xla, xformers) to handle ABI mismatches and import failures gracefully.

When an external attention backend fails to load (e.g., flash_attn compiled
against a different PyTorch version), the code now logs a warning and falls
back to native PyTorch SDPA instead of crashing.

This fixes the "undefined symbol" errors that occur when flash_attn's compiled
binaries are incompatible with the installed PyTorch version.

See CHANGELOG.md for detailed documentation.

Co-Authored-By: Claude Opus 4.5 <[email protected]>
    @sym-bot @claude
    sym-bot and claude committed Jan 31, 2026
    Configuration menu
    Copy the full SHA
    3d2d148 View commit details
    Browse the repository at this point in the history

  2. Merge upstream/main to sync fork

    @sym-bot
    sym-bot committed Jan 31, 2026
    Configuration menu
    Copy the full SHA
    c93a228 View commit details
    Browse the repository at this point in the history
Loading