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

Gemini 2025 ICPC World Finals Code Submissions

[ICLR 2026] LLM/VLM gaming agents and model evaluation through games.

Command and Conquer: Red Alert

The code for creating the iGSM datasets in papers "Physics of Language Models Part 2.1, Grade-School Math and the Hidden Reasoning Process" (arxiv 2407.20311) and "Physics of Language Models Part 2…

[NeurIPS 2025 Spotlight] Reasoning Environments for Reinforcement Learning with Verifiable Rewards

Benchmarking physical understanding in generative video models

A 15TB Collection of Physics Simulation Datasets

Can Language Models Solve Olympiad Programming?

BricksRL: A Platform for Democratizing Robotics and Reinforcement Learning Research and Education with LEGO

LeanRL is a fork of CleanRL, where selected PyTorch scripts optimized for performance using compile and cudagraphs.

Omniverse Lunar Robotics Simulator

Reference implementation of Megalodon 7B model

(Crafter + NetHack) in JAX. ICML 2024 Spotlight.

A repo to evaluate various LLM's chess playing abilities.

The NetHack Learning Environment

Reverb is an efficient and easy-to-use data storage and transport system designed for machine learning research

Code for "Joint Policy Search for Collaborative Multi-agent Incomplete Information Games"

Command & Conquer: Remastered Collection

CUDA Accelerated Robot Library

Sotopia: an Open-ended Social Learning Environment (ICLR 2024 spotlight)

Google Coding Competitions problem archive

A data-driven, fast driving simulator for multi-agent coordination under partial observability.

A modular, primitive-first, python-first PyTorch library for Reinforcement Learning.

Monte Carlo tree search in JAX

PyTorch implementation of MAE https//arxiv.org/abs/2111.06377

A library for distributed ML training with PyTorch