PPL
Kaiyuan Liu
Papers (6)
Bandwidth-Efficient and Privacy-Preserving Edge-Cloud Many-to-Many Speech TranslationBacktracking When It Strays: Mitigating Dual Exposure Biases in LLM Reasoning DistillationOpenDeepThink: Parallel Reasoning via Bradley-Terry AggregationFrontierSmith: Synthesizing Open-Ended Coding Problems at ScalePrefix Teach, Suffix Fade: Local Teachability Collapse in Strong-to-Weak On-Policy DistillationMulti-Objective and Mixed-Reward Reinforcement Learning via Reward-Decorrelated Policy Optimization