Add attention ablation: MHA vs MQA vs GQA vs FlashAttention on perple… #6

	name: CI

	on:
	push:
	branches: [main, dev]
	pull_request:
	branches: [main]

	concurrency:
	group: ${{ github.workflow }}-${{ github.ref }}
	cancel-in-progress: true

	jobs:
	lint:
	name: Lint (black + isort)
	runs-on: ubuntu-latest
	steps:
	- uses: actions/checkout@v4
	- uses: actions/setup-python@v5
	with:
	python-version: "3.10"
	- name: Install linters
	run: pip install black isort
	- name: black check
	run: black --check --diff . --exclude '/(\.git\|__pycache__)/' \|\| true
	- name: isort check
	run: isort --check-only --diff . \|\| true

	test:
	name: Build & Test (A30, SM86)
	runs-on: [self-hosted, gpu, sm86]
	needs: lint
	steps:
	- uses: actions/checkout@v4
	- name: Install deps
	run: pip install torch==2.2.0 triton==2.2.0 ninja packaging pytest tabulate
	- name: Build CUDA extension
	run: TORCH_CUDA_ARCH_LIST="8.6" pip install -e . --no-build-isolation
	- name: Run tests
	run: pytest tests/test_correctness.py -v

Navigation Menu