Text Detoxification: Data Efficiency, Semantic Preservation and Model Generalization
Jing Yu, Yibo Zhao, Jiapeng Zhu, Wenming Shao, Bo Pang, Zhao Zhang, Xiang Li · 2025
Read Text Detoxification: Data Efficiency, Semantic Preservation and Model Generalization by Jing Yu, Yibo Zhao, Jiapeng Zhu, Wenming Shao, Bo Pang, Zhao Zhang, Xiang Li (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-2507-01050.txt.
Description
Jing Yu and six co-authors work on cleaning toxic text while keeping its sense, in a paper from two thousand twenty-five. Toxic posts harm online talk, so a tool that rewrites them in a safe way has clear worth. Past tools often lost the sense, failed on new kinds of data, or needed costly hand-made pairs of texts.
Their plan trains in two stages. First, a small set of clean paired texts gives the model a strong start. Then a reward model and unlabeled toxic inputs guide more training. The result beats past work on key tests, holds up on new data, and needs far less hand-made data.
Catalog record
| CREATOR | Jing Yu, Yibo Zhao, Jiapeng Zhu, Wenming Shao, Bo Pang, Zhao Zhang, Xiang Li |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 41 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2507-01050