DeepEP: an efficient expert-parallel communication library

AI agents running research on single-GPU nanochat training automatically

Microbenchmarking hyperparameter tuning for JAX functions.

Fast file synchronization and network forwarding for remote development

Tokamax: A GPU and TPU kernel library.

Terminal-based presentations using Pandoc

Scan documents to PDF and more, as simply as possible.

Local and global SPMD type system supporting erasure

Code for implementing central flows

Claude Opus 4.6 wrote a dependency-free C compiler in Rust, with backends targeting x86 (64- and 32-bit), ARM, and RISC-V, capable of compiling a booting Linux kernel.

Your own personal AI assistant. Any OS. Any Platform. The lobster way. 🦞

Anthropic's original performance take-home, now open for you to try!

Flexible and powerful tensor operations for readable and reliable code (for pytorch, jax, TF and others)

Official JAX implementation of End-to-End Test-Time Training for Long Context

Show usage stats for OpenAI Codex and Claude Code, without having to login.

My configuration files, loosely inspired by @sontek

llms can learn their own context compression via RL

Miles is an enterprise-facing reinforcement learning framework for LLM and VLM post-training, forked from and co-evolving with slime.

MLX: An array framework for Apple silicon

Minimal, lightweight JAX implementations of popular models.

Deep learning for dummies. All the practical details and useful utilities that go into working with real models.

Open-source framework for the research and development of foundation models.

Post-training with Tinker

Dion optimizer algorithm

Minimal and scalable research codebase in JAX, designed for rapid iteration on frontier research in LLM and other autoregressive models.