Skip to content

mtmd: clip_graph::build_attn flash-attn disabled changes - #19729

Merged
ngxson merged 3 commits into
ggml-org:masterfrom
sfallah:sf/clip_build_attn
Feb 19, 2026
Merged

mtmd: clip_graph::build_attn flash-attn disabled changes #19729
ngxson merged 3 commits into
ggml-org:masterfrom
sfallah:sf/clip_build_attn

Conversation

@sfallah

@sfallah sfallah commented Feb 19, 2026

Copy link
Copy Markdown
Contributor

This PR extracts a small, self-contained change from #17400.

Summary

  • Refactors clip_graph::build_attn.
  • setting flash_attn_type via CLI params
  • Adds a --flash_off option to tools/mtmd/tests.sh to exercise the updated code path.

Testing

  • The changes to clip_graph::build_attn have been validated across all models covered by tools/mtmd/tests.sh.
  • The new --flash_off flag allows running the test suite without flash attention, ensuring the modified path is properly tested.

@sfallah
sfallah requested a review from ngxson as a code owner February 19, 2026 07:31
@sfallah

sfallah commented Feb 19, 2026

Copy link
Copy Markdown
Contributor Author

@ngxson
FYI: these two tests fail.

[vision] FAIL: ggml-org/Qwen2.5-VL-72B-Instruct-GGUF:Q4_K_M
[vision] FAIL: ggml-org/Llama-4-Scout-17B-16E-Instruct-GGUF:IQ1_S

They fail on master too, and they fail both with flash-attn enabled (default).

@ngxson
ngxson merged commit e6267a9 into ggml-org:master Feb 19, 2026
78 checks passed
liparetejas pushed a commit to liparetejas/llama.cpp that referenced this pull request Feb 23, 2026
bartowski1182 pushed a commit to bartowski1182/llama.cpp that referenced this pull request Mar 2, 2026
ArberSephirotheca pushed a commit to ArberSephirotheca/llama.cpp that referenced this pull request Mar 3, 2026
Seunghhon pushed a commit to Seunghhon/llama.cpp that referenced this pull request Apr 26, 2026
ljubomirj pushed a commit to ljubomirj/llama.cpp that referenced this pull request May 6, 2026
my-other-github-account pushed a commit to my-other-github-account/llama.cpp that referenced this pull request May 15, 2026
my-other-github-account pushed a commit to my-other-github-account/llama.cpp that referenced this pull request May 15, 2026
fewtarius pushed a commit to fewtarius/CachyLLama that referenced this pull request May 30, 2026
fukuro-kun pushed a commit to fukuro-kun/fukuro-llama-cpp-turboquant that referenced this pull request Jul 5, 2026
zommiommy pushed a commit to zommiommy/llama.cpp that referenced this pull request Aug 18, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants