microsoft / microsoft/winml-cli
Fix e2e regressions in HF export mask handling for QNN path
Open
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 40
- Forks
- 11
- Avg merge
- 1d 8h
- Merged PRs (30d)
- 50
Description
Summary
DistilBERT fill-mask/question-answering and BLIP image-to-text e2e regressions.
Root causes
- SDPA attention mask dtype mismatch
- Decoder kwargs path incompatibility
- BLIP decoder mask shape mismatch
Fixes
- Export IO bool normalization for binary
attention_mask - Robust transformers compatibility patching
decoder_wrapperkwargs support- BLIP 2D mask pass-through
Validation
Focused unit tests and e2e tests passed: question-answering, fill-mask, and image-to-text-fp16.
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
No file paths are named. Start with the focused unit tests and e2e cases for question-answering, fill-mask, and image-to-text-fp16, then trace the HF export QNN path around attention-mask normalization, decoder kwargs, and BLIP mask handling. Done means those three scenarios pass without the listed regressions.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- machine-learning
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Active
- Clarity
- Mostly clear
- Newbie friendliness
- 45/100