Ki-Ung song
sk851
AI & ML interests
Generative model / Multimodal
Recent Activity
authored
a paper
2 days ago
Efficient LLaMA-3.2-Vision by Trimming Cross-attended Visual Features