Yongxian Wei

OptMerge: Unifying Multimodal LLM Capabilities and Modalities via Model Merging

Mixture of Neuron Experts

Learn to Learn Consistently via Meta Self-distillation for Few-shot Classification

Adaptive Defense against Harmful Fine-Tuning for Large Language Models via Bayesian Data Scheduler

Open-Vocabulary Customization from CLIP via Data-Free Knowledge Distillation

Whoever Started the interference Should End It: Guiding Data-Free Model Merging via Task Vectors

Modeling Multi-Task Model Merging as Adaptive Projective Gradient Descent

LoRA Recycle: Towards Fine-Tuning-Free Visual Foundation Model via Double-Efficient Data-Free Meta-Learning

LEARN TO LEARN CONSISTENTLY