Skip to content

Add a recipe for gemma4 qnn vision - #650

Open
ananyaa-ms wants to merge 8 commits into
neilmsft/gemma4-e2b-npu-textfrom
aamancherla/gemma4-npu-vision
Open

Add a recipe for gemma4 qnn vision#650
ananyaa-ms wants to merge 8 commits into
neilmsft/gemma4-e2b-npu-textfrom
aamancherla/gemma4-npu-vision

Conversation

@ananyaa-ms

@ananyaa-ms ananyaa-ms commented Oct 5, 2026 •
edited
Loading

Copy link
Copy Markdown
  • Add a qnn recipe for Gemma-4-E2B-it's vision encoder.
  • Use builds to combine the decoder and vision recipes into a single config.

Copilot stopped work on behalf of neilmsft due to an error October 6, 2026 05:17
The session was cancelled by the user.
Copilot AI requested a review from neilmsft October 6, 2026 05:17
Xiaoyu Z (xiaoyu-work) added a commit that referenced this pull request Oct 6, 2026
@xiaoyu-work
Integrate Gemma 4 QNN vision recipe
ba901a9 
Fold PR #650 into the existing multi_comp QNN configuration and shared user_script.py.

Use CPU calibration and serial QNN builds, precompile vision, and cast only the decoder-facing embedding outputs to FP16.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@neilmsft neilmsft Awaiting requested review from neilmsft

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

1 participant

@ananyaa-ms