pyproject.toml

[project]
name = "inference-optimization-mastery"
version = "0.1.0"
description = "Production-grade, measured study of LLM inference optimization (serving, batching, KV cache, quantization, distillation, speculative decoding, compiled runtimes, GPU sizing)."
requires-python = ">=3.12"
dependencies = [
    "torch==2.4.1",
    "transformers>=4.44,<4.50",
    "vllm>=0.6.2,<0.7",          # pinned for torch 2.4/cu121 + Ampere; verified at install time
    "datasets>=2.20",
    "matplotlib>=3.8",
    "numpy>=1.26",
]

[project.optional-dependencies]
dev = ["pytest>=8.0"]
# quantization (P2) — installed in that phase
quant = ["autoawq>=0.2.5", "optimum>=1.20", "auto-gptq>=0.7"]

[tool.pytest.ini_options]
testpaths = ["tests"]
addopts = "-q"
pythonpath = ["."]