DeepYardDeepYard
P

ProMSA

Progressive multimodal search agent for visual question answering with adaptive tool selection

Open SourceFree

About

Research framework for knowledge-based visual question answering using iterative multimodal search. Features adaptive tool selection that progressively queries image and text databases under explicit computational budgets. Designed for scenarios requiring multi-step reasoning across visual and textual knowledge sources, with dynamic strategy adjustment based on intermediate results.

Details

Typecoding-agent
Deploymentself-hosted
Supported Models

Tags

autonomoustool-useopen-sourceragpython