Welcome to my blog
I proudly earned my Ph.D. degree at the University of Toronto. This blog is where I share my thoughts on AI, research, and life. Visit my academic homepage for publications and projects.
分解 MXFP4 量化误差：为大模型强化学习对症下药
MXFP4 量化误差的精确三分解（缩放偏置 / 死区截断 / 网格噪声）以及对症下药的三件套修正（MBS / OF / AQN）。
Decomposing MXFP4 quantization error for LLM RL: reducible bias, recoverable deadzone, and an irreducible floor
An exact three-way decomposition of MXFP4 quantization error into scale bias, deadzone truncation, and grid noise — and the mechanism-targeted corrections (MBS, OF, AQN) that follow.
A-3PO: Accelerating Asynchronous LLM Training with Staleness-aware Proximal Policy Approximation
A simple yet principled approach to eliminate computational overhead in decoupled PPO while maintaining training stability and performance.