Generates and updates professional PyTorch-style docstrings using Sphinx and reStructuredText conventions.
The PyTorch Docstring Writer skill equips Claude with the specific domain knowledge required to document deep learning libraries according to the rigorous standards of the PyTorch project. It automates the creation of raw-string docstrings that include standardized function signatures, LaTeX mathematical formulas for tensor shapes, and Sphinx cross-references. This skill is essential for developers working on PyTorch forks, custom neural network modules, or high-performance C++/Rust extensions who need their documentation to be visually consistent and technically compatible with the official PyTorch ecosystem.
主な機能
01Sphinx cross-referencing for classes, methods, and external research papers
021 GitHub stars
03LaTeX math integration for complex formulas and tensor dimension shapes
04Standardized PyTorch function signature formatting with type annotations
05Support for C-bound function documentation and in-place operation variants
06Structured Args and Keyword Args sections with precise indentation rules
ユースケース
01Standardizing documentation across a large-scale deep learning codebase
02Generating Python-side documentation for high-performance Rust or C++ tensor operations
03Writing documentation for new neural network layers in a PyTorch-based framework
