Pytorch

This skill guides developers through implementing Metal kernels for PyTorch operators on Apple Silicon (MPS). It covers the complete workflow including updating dispatch configuration in native_functions.yaml, writing Metal shader code with type-specialized functors, implementing host-side stubs in Objective-C++, and testing the implementation.

Write docstrings for PyTorch functions and methods following PyTorch conventions. Use when writing or updating docstrings in PyTorch code.

Add unsigned integer (uint) type support to PyTorch operators by updating AT_DISPATCH macros. Use when adding support for uint16, uint32, uint64 types to operators, kernels, or when user mentions enabling unsigned types, barebones unsigned types, or uint support.

Convert PyTorch AT_DISPATCH macros to AT_DISPATCH_V2 format in ATen C++ code. Use when porting AT_DISPATCH_ALL_TYPES_AND*, AT_DISPATCH_FLOATING_TYPES*, or other dispatch macros to the new v2 API. For ATen kernel files, CUDA kernels, and native operator implementations.

TorchDrug is a documentation skill that provides comprehensive guidance for using the TorchDrug PyTorch library in drug discovery and molecular science. It covers graph neural networks for molecules and proteins, including molecular property prediction, protein modeling, knowledge graph reasoning, molecular generation, and retrosynthesis planning with 40+ curated datasets and 20+ model architectures.

This skill provides comprehensive guidance for PyTorch Geometric (PyG), a library for developing Graph Neural Networks. It covers graph creation, GNN architectures (GCN, GAT, GraphSAGE, GIN), node/graph classification, molecular property prediction, and large-scale graph learning with extensive reference documentation and utility scripts.

A comprehensive reference skill for reinforcement learning with Stable Baselines3. It provides algorithm selection guides, training templates, custom environment creation tutorials, callback documentation, and vectorized environment usage patterns for efficient RL agent development.

PufferLib is a high-performance reinforcement learning framework optimized for fast parallel training. It provides templates for creating custom environments, training scripts with PPO, and seamless integration with popular RL frameworks like Gymnasium and PettingZoo, achieving millions of steps per second.