Siming Fu

4

Papers

39

Total Citations

Papers (4)

LLaVA-MoD: Making LLaVA Tiny via MoE-Knowledge Distillation

TFCustom: Customized Image Generation with Time-Aware Frequency Feature Guidance

Resolving Multi-Condition Confusion for Finetuning-Free Personalized Image Generation

MARS: Mixture of Auto-Regressive Models for Fine-grained Text-to-image Synthesis