Bingcong Li

Binomial Gradient-Based Meta-Learning for Enhanced Meta-Gradient Estimation

On the Benefits of Weight Normalization for Overparameterized Matrix Sensing

Scalable Variational Bayesian Fine-Tuning of LLMs via Orthogonalized Low-Rank Adapter

PoLAR: Polar-Decomposed Low-Rank Adapter Representation

RefLoRA: Refactored Low-Rank Adaptation for Efficient Fine-Tuning of Large Models

Zeroth-Order Optimization Finds Flat Minima

On the Crucial Role of Initialization for Matrix Factorization

Can RLHF be More Efficient with Imperfect Reward Models? A Policy Coverage Perspective