Vis-CoT: A Human-in-the-Loop Framework for Interactive Visualization and Intervention in LLM Chain-of-Thought Reasoning

Kaviraj Pather, Elena Hadjigeorgiou, Arben Krasniqi, Claire Schmit, Irina Rusu, Marc Pons, Kabir Khan · 2025

Read Vis-CoT: A Human-in-the-Loop Framework for Interactive Visualization and Intervention in LLM Chain-of-Thought Reasoning by Kaviraj Pather, Elena Hadjigeorgiou, Arben Krasniqi, Claire Schmit, Irina Rusu, Marc Pons, Kabir Khan (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-01412.txt.

Description

Chain-of-thought reasoning lets language models show their steps, but plain text traces are hard to check or steer in high-stakes work. Kaviraj Pather, with six co-authors, presents Vis-CoT, a framework that turns a linear reasoning trace into an interactive graph. Users can see the logical flow, spot flawed steps, and act on them directly.

Within the graph, users prune wrong paths and graft in new premises of their own, shifting the work from passive watching to active collaboration. On two benchmarks, a math word problem set and a strategy question set, Vis-CoT raised final-answer accuracy by up to twenty-four points over non-interactive baselines. A user study also found large gains in felt usability and trust. The authors present the tool as a practical path to more reliable model reasoning.

Catalog record

CREATORKaviraj Pather, Elena Hadjigeorgiou, Arben Krasniqi, Claire Schmit, Irina Rusu, Marc Pons, Kabir Khan
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE42 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2509-01412

Browse the full library · About the Tower of Babel