VAGEN is a multi-turn reinforcement learning framework designed specifically for training VLM Agents. VAGEN leverages Turn-aware Reason-Interaction Chain Optimization (TRICO) to efficiently train VLMs for visual agentic tasks.
VAGEN is a multi-turn reinforcement learning framework designed specifically for training VLM Agents. VAGEN leverages the TRICO algorithm to efficiently train VLMs for visual agentic tasks.
