CulFiT: A Fine-grained Cultural-aware LLM Training Paradigm via Multilingual Critique Data Synthesis

Ruixiang Feng, Shen Gao, Xiuying Chen, Lisi Chen, Shuo Shang · 2025

Read CulFiT: A Fine-grained Cultural-aware LLM Training Paradigm via Multilingual Critique Data Synthesis by Ruixiang Feng, Shen Gao, Xiuying Chen, Lisi Chen, Shuo Shang (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-2505-19484.txt.

Description

Ruixiang Feng, Shen Gao, Xiuying Chen, Lisi Chen, and Shuo Shang present CulFiT in 2025, a training path that cuts culture bias in language models. Models lean to rich lands and slight the tongues and values of poor regions. Their path makes culture asks, builds critique data in home tongues, and grades replies in fine units.

A new open ask set, GlobalCultureQA, tests culture-fair replies world-wide. Across that set and three past benches, CulFiT tops open models on culture fit and plain thought scores. The fine grade path keeps the checks clear to read.

Catalog record

CREATORRuixiang Feng, Shen Gao, Xiuying Chen, Lisi Chen, Shuo Shang
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE55 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-19484

Browse the full library · About the Tower of Babel