Skip to content

[NPUW] Suppport 4d float attention_mask - #38932

Draft
OrestChura wants to merge 33 commits into
openvinotoolkit:masterfrom
OrestChura:oc/qwen3-tts-rmetcalfe-attention-mask-4d-float
Draft

OrestChura wants to merge 33 commits into
openvinotoolkit:masterfrom
OrestChura:oc/qwen3-tts-rmetcalfe-attention-mask-4d-float

Conversation

@OrestChura

Copy link
Copy Markdown
Contributor

Details:

  • handle 4d attention mask for NPUW_LLM
  • serialize/deserialize m_attention_mask_is_4d_float

Tickets:

AI Assistance:

  • AI assistance used: yes
  • AI used for porting changes, writing a test, review; human validation: review, local tests, reproducer check

RyanMetcalfeInt8 and others added 3 commits October 16, 2026 23:08
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
- Bump the serialization version for the mask-mode flag
- cover static 4D mask shapes assigment by a test
- align carried code with current formatting

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
@github-actions github-actions Bot added category: build OpenVINO cmake script / infra category: transformations OpenVINO Runtime library - Transformations category: NPU OpenVINO NPU plugin category: NPUW NPUW plugin labels Oct 16, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

category: build OpenVINO cmake script / infra category: NPU OpenVINO NPU plugin category: NPUW NPUW plugin category: transformations OpenVINO Runtime library - Transformations

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants