Researcher, Microsoft Research Asia
2 papers at NeurIPS 2025
We propose Pseudo-Zeroth-Order (PseuZO), a framework that estimates Jacobian matrices via model output differentiation and applies EMA for variance reduction, with theoretical convergence guarantees and empirical validation.
We rethink LLMs from the perspective of recommender systems and propose Language System, which exploits the model’s output distribution more efficiently and effectively.
