Weizhu Chen

Beyond Pass@ 1: Self-Play with Variational Problem Synthesis Sustains RLVR

COSMOS: A Hybrid Adaptive Optimizer for Efficient Training of Large Language Models

NorMuon: Making Muon more efficient and scalable

Mixture of Neuron Experts

InfoAgent: Advancing Autonomous Information‑Seeking Agents

R&D-Agent: An LLM-Agent Framework Towards Autonomous Data Science

Reinforcement Learning for Reasoning in Large Language Models with One Training Example

Decoder-Hybrid-Decoder Architecture for Efficient Reasoning with Long Generation

Samba: Simple Hybrid State Space Models for Efficient Unlimited Context Language Modeling

SlimMoE: Structured Compression of Large MoE Models via Expert Slimming and Distillation

Scaling Laws of Synthetic Data for Language Model

SwS: Self-aware Weakness-driven Problem Synthesis in Reinforcement Learning for LLM Reasoning

LongRoPE2: Near-Lossless LLM Context Window Scaling

Relative Preference Optimization: Enhancing LLM Alignment through Contrasting Responses across Identical and Diverse Prompts

Segmenting Text and Learning Their Rewards for Improved RLHF in Language Models

Self-Augmented Preference Optimization: Off-Policy Paradigms for Language Model Alignment