A Python-embedded DSL that makes it easy to write fast, scalable ML kernels with minimal boilerplate.
Alerting system for the pytorch org
Fault tolerance for PyTorch (HSDP, LocalSGD, DiLoCo, Streaming DiLoCo)
Collective communications library with various primitives for multi-machine training.
A modular, primitive-first, python-first PyTorch library for Reinforcement Learning.
Run PyTorch LLMs locally on servers, desktop and mobile
