python-ml

Version	2026.144.1443
Repo	superproject

A CUDA-ready Python 3.13 ML environment with PyTorch, transformers, and a vLLM inference runtime The python-ml candy owns the pixi.toml that builds the core ML Python environment under ~/.pixi/envs/default: PyTorch (cu130), torchvision/torchaudio, transformers, accelerate and the full vLLM runtime-dependency closure. After the pixi environment is built, a user-phase task pip-installs the vLLM nightly wheel (–no-deps) into that same interpreter. Every interpreter and library lands at a fixed path under ~/.pixi/envs/default, so its presence and importability are directly checkable.

Acceptance plan

This candy’s plan: — the runnable spec charly check executes against a live deployment. check: steps are idempotent probes; run: steps change state.

Intent	Step
run	command=~/.pixi/envs/default/bin/python -m pip install –no-deps \ ‘https://wheels.vllm.ai/f53fa26e05c476a43f6db048a9e3b43bcb2b72fb/vllm-0.19.1rc1.dev39%2Bgf53fa26e0.cu130-cp38-abi3-manylinux_2_35_x86_64.whl’
check	the pixi default-environment Python interpreter is present
check	PyTorch imports inside the pixi environment and prints its version
check	the transformers library imports inside the pixi environment and prints its version
check	the vLLM nightly wheel installed by the user-phase task imports and prints its version
check	command=${HOME}/.pixi/envs/default/bin/python -c “import torch; print(torch.version)”
check	command=${HOME}/.pixi/envs/default/bin/python -c “import vllm; print(vllm.version)”
check	command=${HOME}/.pixi/envs/default/bin/python -c “import transformers; print(transformers.version)”