Skip to content

ci: publish wheels for torch 2.11, 2.12.1, and 2.13.0 #1005

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
Chessing234 wants to merge 3 commits into
base: main
Choose a base branch
from
Open

ci: publish wheels for torch 2.11, 2.12.1, and 2.13.0 #1005

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
353c219
ci: publish wheels for torch 2.11, 2.12.1, and 2.13.0
Chessing234 Jul 31, 2026
e18802c
docs: note how prebuilt wheel coverage tracks the release matrix
Chessing234 Jul 31, 2026
d7ca4de
docs: note publish matrix must stay aligned with setup wheel URLs
Chessing234 Jul 31, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 15 additions & 1 deletion .github/workflows/publish.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -41,7 +41,9 @@ jobs:
# manylinux docker image, but I haven't figured out how to install CUDA on manylinux.
os: [ubuntu-22.04, ubuntu-22.04-arm]
python-version: ["3.10", "3.11", "3.12", "3.13"]
torch-version: ["2.6.0", "2.7.1", "2.8.0", "2.9.1", "2.10.0"]
# Keep in sync with setup.py get_wheel_url() torch major.minor tags.
# New torch minors require a tagged release to publish matching wheels (#999).
torch-version: ["2.6.0", "2.7.1", "2.8.0", "2.9.1", "2.10.0", "2.11.0", "2.12.1", "2.13.0"]
cuda-version: ["11.8.0", "12.9.1", "13.0.1"]
# We need separate wheels that either uses C++11 ABI (-D_GLIBCXX_USE_CXX11_ABI) or not.
# Pytorch wheels currently don't use it, but nvcr images have Pytorch compiled with C++11 ABI.
Expand All @@ -56,6 +58,12 @@ jobs:
cuda-version: "11.8.0"
- torch-version: "2.10.0"
cuda-version: "11.8.0"
- torch-version: "2.11.0"
cuda-version: "11.8.0"
- torch-version: "2.12.1"
cuda-version: "11.8.0"
- torch-version: "2.13.0"
cuda-version: "11.8.0"
# CUDA 13.0 is only supported by PyTorch 2.9+
- torch-version: "2.6.0"
cuda-version: "13.0.1"
Expand All @@ -78,6 +86,12 @@ jobs:
cxx11_abi: "FALSE"
- torch-version: "2.10.0"
cxx11_abi: "FALSE"
- torch-version: "2.11.0"
cxx11_abi: "FALSE"
- torch-version: "2.12.1"
cxx11_abi: "FALSE"
- torch-version: "2.13.0"
cxx11_abi: "FALSE"
uses: ./.github/workflows/_build.yml
with:
runs-on: ${{ matrix.os }}
Expand Down
7 changes: 7 additions & 0 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -36,6 +36,13 @@ Install PyTorch first. By default, `mamba-ssm` installs the core package without
| CUDA selective scan opt-in | `MAMBA_KEEP_CUDA_BUILD=TRUE pip install mamba-ssm --no-build-isolation` | Installs `selective_scan_cuda`; pip first tries a matching prebuilt CUDA/HIP wheel, then compiles locally if no wheel is available. |
| Force local CUDA selective scan build | `MAMBA_FORCE_BUILD=TRUE MAMBA_KEEP_CUDA_BUILD=TRUE pip install mamba-ssm --no-build-isolation` | Skips cached wheels and compiles `selective_scan_cuda` locally. |

Prebuilt CUDA wheels (when using `MAMBA_KEEP_CUDA_BUILD=TRUE`) are published on
[GitHub releases](https://github.com/state-spaces/mamba/releases) for select
torch/CUDA/Python/platform combinations. The release CI matrix in
`.github/workflows/publish.yaml` defines which torch versions are built; if your
stack is newer than the latest release (e.g. torch 2.11+ on Colab), pip falls
back to a local `selective_scan_cuda` compile with no warning beyond setup logs.

`--no-build-isolation` is required for CUDA builds so that pip uses your existing CUDA-enabled
PyTorch instead of installing torch-cpu in an isolated build environment.

Expand Down