GeezerVison

Official implementation of CVPR'26 paper A Multi-Agent Perception-Action Alliance for Efficient Long Video Reasoning

This is the code for AttentionPack (CVPR 2026)

Dynamically selecting the most diverse ensemble set of multiple LLMs and fusing the outputs with Multi Agent Reinforcement Learning. Overal system can switch between models according to the tasks and boost the overall accuracy and robustness

A survey on harmful fine-tuning attack for large language model (ACM CSUR)

A Survey on Large Language Model-Based Game Agents (ACM CSUR)

A Behavioral Trace Dataset for High-Risk Memecoin Launch Detection

MemeTrans: A Dataset for Detecting High-Risk Meme Coin Launches on Blockchain