SymDPO: Boosting In-Context Learning of Large Multimodal Models with Symbol Demonstration Direct Preference Optimization Paper • 2411.11909 • Published 11 days ago • 20