Release v0.2.0
Release v0.2.0
π Major Updates
This release expands GPU support with three CUDA variants and updates the entire software stack to the latest versions.
π New Features
Expanded GPU Support
- Added CUDA 12.8 variant (Ubuntu 24.04) with PyTorch 2.9.0 and JAX 0.8.0
- Added CUDA 13 variant (Ubuntu 24.04) with PyTorch 2.9.0 and JAX 0.8.0
- Now providing 8 container variants total (up from 4):
- CPU (Docker & Singularity) - Ubuntu 24.04
- GPU CUDA 12.4 (Docker & Singularity) - Ubuntu 22.04
- GPU CUDA 12.8 (Docker & Singularity) - Ubuntu 24.04
- GPU CUDA 13 (Docker & Singularity) - Ubuntu 24.04
Base Image Updates
- CPU variants: Upgraded to Ubuntu 24.04 (from 22.04)
- CUDA 12.8/13 variants: Ubuntu 24.04
- CUDA 12.4 variant: Remains on Ubuntu 22.04 (for compatibility)
π¦ Package Updates
Core Scientific Stack
- NumPy: 2.3.4
- SciPy: 1.16.3
- Matplotlib: 3.10.7
- Astropy: 7.1.1
- Python: 3.12 (Ubuntu 24.04) / 3.11 (Ubuntu 22.04)
GPU Packages by CUDA Variant
CUDA 12.4 (Ubuntu 22.04):
- PyTorch: 2.3.1+cu121 (with cuDNN 8.9 compatibility)
- JAX: 0.4.26 (with cuDNN 8.9 compatibility)
- CuPy: 13.6.0 (cuda12x)
- CUDA: 12.4.0
CUDA 12.8 (Ubuntu 24.04):
- PyTorch: 2.9.0+cu128
- JAX: 0.8.0
- CuPy: 13.6.0 (cuda12x)
- CUDA: 12.8.1
CUDA 13 (Ubuntu 24.04):
- PyTorch: 2.9.0+cu130
- JAX: 0.8.0
- CuPy: 13.6.0 (cuda13x)
- CUDA: 13.0.1
JAX Ecosystem (CUDA 12.8/13)
- BlackJAX: 1.2.5
- NumPyro: 0.19.0
- PyMC: 5.26.1
- Flax: 0.12.0
- TensorBoard: 2.20.0
π·οΈ Tagging
All images are tagged with version v0.2.0 and include OS information:
v0.2.0-cpu-ubuntu24.04v0.2.0-cpu-singularity-ubuntu24.04v0.2.0-gpu-cu124-ubuntu22.04v0.2.0-gpu-cu124-singularity-ubuntu22.04v0.2.0-gpu-cu128-ubuntu24.04v0.2.0-gpu-cu128-singularity-ubuntu24.04v0.2.0-gpu-cu13-ubuntu24.04v0.2.0-gpu-cu13-singularity-ubuntu24.04
Moving aliases (cpu, gpu-cu124, gpu-cu128, gpu-cu13, etc.) have been updated to point to v0.2.0.
π₯ Quick Start
# Pull CPU variant (multi-arch: amd64/arm64)
docker pull vhaasteren/anpta:cpu
# Pull GPU variants (amd64 only)
docker pull vhaasteren/anpta:gpu-cu124 # CUDA 12.4
docker pull vhaasteren/anpta:gpu-cu128 # CUDA 12.8
docker pull vhaasteren/anpta:gpu-cu13 # CUDA 13π§ Build Instructions
Build and push all variants:
docker login
./scripts/push_to_registry.sh dockerhubπ Documentation
- Updated documentation for all 8 variants
- See
docs/PUBLISHING_DOCKERHUB.mdfor publishing workflow - See
docs/DOCKERHUB_README.mdfor usage instructions
π Links
- Docker Hub: https://hub.docker.com/r/vhaasteren/anpta
- Documentation: See repository
docs/directory - Issues: Report issues on GitHub
Full Changelog: See commit history for detailed changes.