Preference Tuning For Toxicity Mitigation Generalizes Across Languages (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Detoxifying multilingual Large Language Models (LLMs) has become crucial due to their increasing global use. |
| Approach: | They propose to use English preference tuning to study cross-lingual detoxification of LLMs. |
| Outcome: | The proposed method reduces toxicity in multilingual LLMs by reducing the probability of mGPT-1.3B generating toxic continuations across 17 languages. |
Similar Papers
A Survey of Toxicity Mitigation Strategies for Multilingual Language Models (2026.findings-acl)
Copied to clipboard
| Challenge: | Large language models can reproduce and amplify toxic content, including hate speech, harassment, and bias. |
| Approach: | They propose a comprehensive survey of the many detoxification methods tailored to multilingual LLMs. |
| Outcome: | The proposed methods are based on data filtering, style transfer, expert-based logit steering, retrieval augmentation, and human feedback. |
Cross-Lingual Transfer of Debiasing and Detoxification in Multilingual LLMs: An Extensive Investigation (2025.findings-acl)
Copied to clipboard
| Challenge: | Prior work has shown that finetuning on specialized datasets can mitigate this behavior, and doing so in English can transfer to other languages. |
| Approach: | They propose to fine tune generative large language models to provide safe responses to harmful user input and to use direct preference optimization to mitigate toxicity. |
| Outcome: | The proposed models show that finetuning on specialized datasets reduces biases but also produces fluent and diverse text in non-English languages. |
Implicit Cross-Lingual Rewarding for Efficient Multilingual Preference Alignment (2025.findings-acl)
Copied to clipboard
| Challenge: | Existing approaches to align English LLMs with human preferences rely on expensive human annotations or advanced multilingual preference alignment models. |
| Approach: | They propose a method that captures learned preferences from English models by implicit rewards . they annotate preference relations in cross-lingual instruction-following pairs using English . |
| Outcome: | The proposed approach captures learned preferences from well-aligned English models by implicit rewards and transfers them to other languages through iterative training. |
How Does DPO Reduce Toxicity? A Mechanistic Neuron-Level Analysis (2025.emnlp-main)
Copied to clipboard
| Challenge: | Direct Preference Optimization (DPO) is a popular choice of safety fine-tuning algorithms, but prior explanations of its effects only account for dampened toxic neurons in the MLP layers. |
| Approach: | They analysed four language models and found that toxic neurons only account for 2.5% to 24% of DPO’s effects across models. |
| Outcome: | The proposed method outperforms DPO in reducing toxicity while preserving perplexity, without requiring any weight updates. |
Exploring Cross-lingual Text Detoxification with Large Multilingual Language Models. (2022.acl-srw)
Copied to clipboard
| Challenge: | Existing methods of textual style transfer are monolingual i.e. designed to work in one exact language. |
| Approach: | They propose to make large multilingual models capable of performing multilingual style transfer without direct fine-tuning in a given language. |
| Outcome: | The proposed model can generate text in a given language without fine-tuning and is able to perform cross-lingual detoxification without direct fine- tuning. |
RLHF Can Speak Many Languages: Unlocking Multilingual Preference Optimization for LLMs (2024.emnlp-main)
Copied to clipboard
| Challenge: | Preference optimization is a widely adopted post-training technique to align large language models with human preferences. |
| Approach: | They propose a method for generating multilingual feedback data to balance data coverage. |
| Outcome: | The proposed method achieves 54.4% win-rate against current state-of-the-art multilingual LLM in its parameter class and 69.5% win- rate or higher against widely used models like Gemma, Mistral and Llama 3. |
Direct Preference Optimization for Neural Machine Translation with Minimum Bayes Risk Decoding (2024.naacl-short)
Copied to clipboard
| Challenge: | Recent work shows that MBR decoding can significantly improve translation performance of Multilingual Large Language Models. |
| Approach: | They propose a method that uses a monolingual fine-tuning set to fine- tune MLLMs to get the gains of MBR without additional computation in inference. |
| Outcome: | The proposed method outperforms greedy decoding and beam search on multiple NMT tests. |
Detoxifying Large Language Models via the Diversity of Toxic Samples (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods for analyzing and utilizing toxic samples are limited . current methods fail to fully harness their potential . |
| Approach: | They propose a diverse detoxification framework that leverages toxic samples' diversity . they propose MPSG strategy and SC-DPO approach to elicit personalized toxic responses . |
| Outcome: | The proposed framework could be used to optimize large language models for user safety . it incorporates two components: MPSG strategy and SC-DPO approach . |
SynthDetoxM: Modern LLMs are Few-Shot Parallel Detoxification Data Annotators (2025.naacl-long)
Copied to clipboard
| Challenge: | Existing approaches to multilingual text detoxification are hampered by the scarcity of parallel multilingual datasets. |
| Approach: | They propose a pipeline for the generation of multilingual parallel detoxification data and a dataset for SynthDetoxM which is manually generated and rewritten with open-source LLMs. |
| Outcome: | The proposed pipeline outperforms human-annotated datasets even in data limited setting. |
FuxiTranyu: A Multilingual Large Language Model Trained with Balanced Data (2024.emnlp-industry)
Copied to clipboard
Haoran Sun, Renren Jin, Shaoyang Xu, Leiyu Pan, null Supryadi, Menglong Cui, Jiangcun Du, Yikun Lei, Lei Yang, Ling Shi, Juesi Xiao, Shaolin Zhu, Deyi Xiong
| Challenge: | Large language models exhibit significant performance discrepancies between high- and low-resource languages. |
| Approach: | They present an open-source multilingual LLM with 8 billion parameters and a multilingual instruction dataset. |
| Outcome: | The proposed model achieves consistent multilingual representations across languages. |