Ning Yang

Token-Importance Guided Direct Preference Optimization

Breaking the MoE LLM Trilemma: Dynamic Expert Clustering with Structured Compression

Evolving LLMs' Self-Refinement Capability via Synergistic Training-Inference Optimization

Turn-Level Trajectory Optimization for Robust Multi-Turn LLM Reasoning

Enhancing Mathematical Reasoning in Language Models Through Focused Differentiation Training

FedADM: Adaptive Federated Learning via Dissimilarity Measure