Xiaoqiang Lin

ActiveDPO: Active Direct Preference Optimization for Sample-Efficient Alignment

REFRAG: Rethinking RAG based Decoding

Provably Sample-Efficient Active Preference Data Collection

NICE Data Selection for Instruction Tuning in LLMs with Non-differentiable Evaluation Metric

Efficient Top-m Data Values Identification for Data Selection

Neural Dueling Bandits: Preference-Based Optimization with Human Feedback

Prompt Optimization with Human Feedback