Yuhang Liu

RAIN-Merging: A Gradient-Free Method to Enhance Instruction Following in Large Reasoning Models with Preserved Thinking Format

Bi-LoRA: Efficient Sharpness-Aware Minimization for Fine-Tuning Large-Scale Models

T2I-ConBench: Text-to-Image Benchmark for Continual Post-training

InjecMEM: Memory Injection Attack on LLM Agent Memory Systems

Simulating Training Dynamics to Reconstruct Training Data from Deep Neural Networks