CarO : C hain-of- A nalogy R easoning O ptimization for Robust Content Moderation

Bingzhe Wu [email protected], Haotian Lu [email protected], Yuchen Mou [email protected] · 2026

Read CarO : C hain-of- A nalogy R easoning O ptimization for Robust Content Moderation by Bingzhe Wu [email protected], Haotian Lu [email protected], Yuchen Mou [email protected] (2026) 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-2604-10504.txt.

Description

Wu, Lu, and Mou harden moderation models against sly context traps. Even reasoning tuned models trip on vague cases when context plants false shortcuts. Their CARO frame first grows analogy chains from moderation data with retrieval plus tuning. A second preference pass then rewards analogy habits outright. Tests beat top reasoning and guard models, lifting mean F1 by 24.9 percent on hard vague sets.

Catalog record

CREATORBingzhe Wu [email protected], Haotian Lu [email protected], Yuchen Mou [email protected]
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE40442
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.10504

Browse the full library · About the Tower of Babel