Projects

Open-source framework for advanced LLM reasoning via process reward models, tree search, and reinforcement learning.

🏆 Best System Paper, CoRL 2020

Scalable multi-agent RL training school for autonomous driving. Powered NeurIPS 2022 driving challenge.

JMLR 2023

Parallel framework for population-based multi-agent RL. Supports self-play, league training, and PSRO.

NeurIPS 2022

Casting multi-agent RL as a sequence modeling problem. SOTA on cooperative benchmarks. Deployed in real-world multi-AGV warehouse systems.

NeurIPS 2024

Evaluation toolkit and benchmark for multi-agent zero-shot coordination.

ICML 2023

GPU-centric experience replay system for large RL models. 6× throughput over DeepMind Reverb.

Intelligently sync, transcribe, and organize Apple Voice Memos with AI. OpenClaw Skill.