Skip to content

Add latest AWQ CUDA fp16 int4 kernels #63

Add latest AWQ CUDA fp16 int4 kernels

Add latest AWQ CUDA fp16 int4 kernels #63

name: Linux CUDA tests
on:
push:
branches:
- main
paths:
- "quanto/**"
- "test/**"
- "pyproject.toml"
- "setup.py"
pull_request:
types: [assigned, opened, synchronize, reopened]
paths:
- "quanto/**"
- "test/**"
- "pyproject.toml"
- "setup.py"
jobs:
conventional-commits:
uses: ./.github/workflows/conventional-commits.yml
python-quality:
uses: ./.github/workflows/python-quality.yml
test-ubuntu-cuda:
needs: [conventional-commits, python-quality]
runs-on: [self-hosted, single-gpu , nvidia-gpu, a10, ci]
strategy:
fail-fast: false
matrix:
cuda-version: ["11.8", "12.1"]
container:
image: pytorch/pytorch:2.2.1-cuda${{ matrix.cuda-version }}-cudnn8-devel
steps:
- uses: actions/checkout@v2
- name: Check CUDA installation
run: |
nvcc -V
- name: Build and install quanto
run: |
pip install --upgrade pip
pip install .[dev]
- name: Run tests
run: |
python -m pytest test