Flash attention SDPA extension (MHA/GQA prefill & decode, FP16/FP32)#287

Open

student-Jungmin wants to merge 41 commits into

PSAL-POSTECH:masterfrom

student-Jungmin:feature-flashattn-extension

Commits on May 22, 2026

[CI] Set persist-credentials: false on actions/checkout

authored and
root
committed
[Doc] Add CLAUDE.md project reference for AI assistants

authored and
root
committed
[Build] Pin third-party deps to thirdparty/github-releases.json tags

YWHyuk
and
claude
committed
Merge pull request #227 from PSAL-POSTECH/feature/build-pins-and-op-coverage
YWHyuk
authored
[Scripts] Introduce parallel worktree workflow with codegen cache helper

YWHyuk
and
claude
committed
[Doc] Note CI test allowlist and codegen-cache gotcha in CLAUDE.md

YWHyuk
and
claude
committed
[Frontend] Fix index/i64 type mismatch in expert-mask codegen (issue #228)
YWHyuk
committed
[CI] Register 5 previously-uncovered tests in pytorchsim_test.yml

YWHyuk
and
claude
committed

Commits on May 27, 2026

[Frontend] Drop stale single-reduction-axis guard breaking test_diffusion

YWHyuk
and
claude
committed

Commits on Jun 15, 2026

Commits on Jun 17, 2026

[CI] Move CI to a CPU-only image and GitHub-hosted runners

psalwonhyuk
authored and
YWHyuk
committed

Commits on Jun 18, 2026

[Experiment] Fix BERT.py imports and fail-fast in run_cycle.sh

YWHyuk
and
claude
committed

Commits on Jun 19, 2026

Commits on Jun 22, 2026

Commits on Jun 24, 2026

[CI] Make wrapper2/wrapper3 exercise small systolic-array configs
YWHyuk
committed

Commits on Jul 6, 2026

Flash attention SDPA extension (MHA/GQA prefill & decode, FP16/FP32)
psalwonhyuk
committed