Reproduce “CLARA: A Cognitively-Driven Clarification Framework for Grounding Ambiguous Requests” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Dimitrios Bekris, Marios Glytsos, Panagiotis P. Filntisis, Niki Efthymiou, George Retsinas, Petros Maragos
CLARA turns vague household requests into scene proposals by asking about object placement and storage preferences. It uses segmentation and image editing to rearrange or remove objects, fill gaps, and render a proposed scene for confirmation. In a six-participant study, clarified prompts raised Gemini’s intent rating from 1.40 to 2.97, but evaluation covered three scenes, not physical task completion.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.