Hanling Tian

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

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

GradCFG: Gradient Inversion of Classifier-Free Guidance Diffusion Models

InjecMEM: Memory Injection Attack on LLM Agent Memory Systems

Primphormer: Efficient Graph Transformers with Primal Representations

Simulating Training Dynamics to Reconstruct Training Data from Deep Neural Networks

Primphormer: Leveraging Primal Representation for Graph Transformers