eugenelet/Learning-to-Select-Visual-In-Context-Demonstrations

Learning to Select Visual In-Context Demonstrations

Authors: Eugene Lee, Yu-Chi Lin, Jiajie Diao

Paper: Learning to Select Visual In-Context Demonstrations (CVPR 2026 Findings)

This paper proposes a Dueling DQN agent that learns to select visual in-context demonstrations for multimodal LLMs.

Official Code

The official implementation is available on GitHub: eugenelet/Learning-to-Select-Visual-In-Context-Demonstrations

Note: This Hugging Face Hub page points to the official implementation on GitHub; no model weights are hosted here.