ReSeTOX: Re-learning attention weights for toxicity mitigation in machine translation

Javier García Gilabert, Carlos Escolano, Marta R. Costa-Jussà
5/19/2023
cs.CL

No confident code implementation found for this paper.

Abstract

Our proposed method, ReSeTOX (REdo SEarch if TOXic), addresses the issue of Neural Machine Translation (NMT) generating translation outputs that contain toxic words not present in the input. The objective is to mitigate the introduction of toxic language without the need for re-training. In the case of identified added toxicity during the inference process, ReSeTOX dynamically adjusts the key-value self-attention weights and re-evaluates the beam search hypotheses. Experimental results demonstrate that ReSeTOX achieves a remarkable 57% reduction in added toxicity while maintaining an average translation quality of 99.5% across 164 languages.

View on arXivView PDF

Code Implementations

No confident code match yet

We couldn't find an author-owned or strongly-evidenced community implementation for this paper. Any repos shown below are weak matches — verify before relying on them.

No code implementations found yet.

Know of an implementation? Let us know in the comments below!

Similar Papers

Added Toxicity Mitigation at Inference Time for Multimodal and Massively Multilingual Translation

89% match2023

Translate, then Detect: Leveraging Machine Translation for Cross-Lingual Toxicity Classification

84% match2025

Adding Instructions during Pretraining: Effective Way of Controlling Toxicity in Language Models

81% match2023

RealToxicityPrompts: Evaluating Neural Toxic Degeneration in Language Models

81% match1 repo2342020

ReMova: Fine-tuning LLMs for English to Belarusian translation

81% match2026

Preference Tuning For Toxicity Mitigation Generalizes Across Languages

81% match1 repo182024

Cite this paper

@article{gilabert2023resetox,
  title  = {ReSeTOX: Re-learning attention weights for toxicity mitigation in machine translation},
  author = {Javier García Gilabert and Carlos Escolano and Marta R. Costa-Jussà},
  year   = {2023},
  eprint = {2305.11761},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2305.11761v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2305.11761.svg)](https://codeofpaper.com/paper/2305.11761)

Discussion