The open-source code for the NeurIPS 2025 paper, "Beyond the 80/20 Rule: High-Entropy Minority Tokens Drive Effective Reinforcement Learning for LLM Reasoning."

Demystifying Reinforcement Learning in Agentic Reasoning

Official Code for NeurIPS'25 ER Workshop "The Zero-Step Thinking: An Empirical Study of Mode Selection as a Harder Early Exit Problem in Reasoning Models"

SimKO: Simple Pass@K Policy Optimization

MARS: Reinforcing Multi-Agent Reasoning of LLMs through Self-Play in Strategic Games

MemGen: Weaving Generative Latent Memory for Self-Evolving Agents

Post-training with Tinker

The Entropy Mechanism of Reinforcement Learning for Large Language Model Reasoning.

A curated list of awesome resources about reward construction for AI agents. This repository covers cutting-edge research, and practical guides on defining and collecting rewards to build more inte…

Checkpoint-engine is a simple middleware to update model weights in LLM inference engines

Revealing and Mitigating the Challenge of Detecting Character Knowledge Errors in LLM Role-Playing (EMNLP 2025 main)

A Gym for Agentic LLMs

🌾 OAT: A research-friendly framework for LLM online alignment, including reinforcement learning, preference learning, etc.

SGLang is a fast serving framework for large language models and vision language models.

Layer curation pipeline used in LayerAnimate [ICCV 2025]

Verlog: A Multi-turn RL framework for LLM agents

[Preprint] On the Generalization of SFT: A Reinforcement Learning Perspective with Reward Rectification.

A collection of verifiable games in sotopia format

[ACL 2025] Cracking the Code of Hallucination in LVLMs with Vision-aware Head Divergence

Lightning-Fast RL for LLM Reasoning and Agents. Made Simple & Flexible.

Training and inference code for "Training Language Models for Social Deduction with Multi-Agent Reinforcement Learning"

This is the official implementation of Multi-Agent PPO (MAPPO).