Use released vllm with pytorch

pyproject.toml

# git clone https://gist.github.com/vadimkantorov/07a421f676f8af94457732f8e9127dde
# cd 07a421f676f8af94457732f8e9127dde
# uv venv
# uv sync
# source .venv/bin/activate # uv run python

# https://github.com/pytorch/pytorch/issues/160875#issuecomment-4183789164
# https://github.com/vllm-project/vllm/issues/24218

[build-system]
requires = ["setuptools>=65", "wheel"]
build-backend = "setuptools.build_meta"

[project]
name =  "vllm_serve"
version = "0.0.0.1"
requires-python = ">=3.12,<3.13"
dependencies = [
    "torch==2.10.0",
    "flash-attn==2.8.3",
    "vllm==0.19.0",
]

[tool.uv.sources]
torch = { index = "torch_cu13" }
flash-attn = { url = "https://github.com/Dao-AILab/flash-attention/releases/download/v2.8.3/flash_attn-2.8.3+cu12torch2.8cxx11abiTRUE-cp312-cp312-linux_x86_64.whl" }

[[tool.uv.index]]
name = "torch_cu13"
url = "https://download.pytorch.org/whl/cu130"

[tool.uv]
index-strategy = "unsafe-best-match" # for setuptools resolution not from pytorch index which has setuptools too old for vllm