DSCD: Large Language Model Detoxification with Self-Constrained Decoding (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods for decoding large language models (LLMs) are based on external constraints and require additional resource overhead and loss of generation fluency. |
| Approach: | They propose a method for LLMs detoxification without parameter fine-tuning that strengthens the inner token distribution while weakening that of hallucination and toxic layer during output generation. |
| Outcome: | Extensive experiments on open-source LLMs and public datasets demonstrate DSCD's state-of-the-art (SOTA) performance in detoxification and generation fluency, with superior efficiency compared to existing methods. |
Similar Papers
Detoxification for LLM: From Dataset Itself (2026.acl-long)
Copied to clipboard
| Challenge: | Existing methods for large language models focus on post-training stage or inference time, while few tackle the source of toxicity, namely, the dataset itself. |
| Approach: | They propose to localize and rewrite toxic spans in raw corpora with SoCD, which guides an LLM to localized and preserving semantics while preserving toxicity. |
| Outcome: | The proposed method reduces TP from 0.42 to 0.18 and Expected Maximum Toxicity (EMT) from 0.43 to 0.20 on three LLMs. |
MICo: Preventative Detoxification of Large Language Models through Inhibition Control (2024.findings-naacl)
Copied to clipboard
Roy Siegelmann, Ninareh Mehrabi, Palash Goyal, Prasoon Goyal, Lisa Bauer, Jwala Dhamala, Aram Galstyan, Rahul Gupta, Reza Ghanadan
| Challenge: | Large Language Models (LLMs) have a tendency to devolve into toxic degeneration . model may classify prompts as toxic or non-toxic and categorically refuse to respond to those deemed toxic. |
| Approach: | They propose a mechanism for LLM detoxification by labeling acceptable and unacceptable examples and including a corresponding acceptable rewrite with every unacceptable example. |
| Outcome: | The proposed model improves on the baseline model and shows that it detects and rewrites toxic and harmful examples. |
Attribute Controlled Fine-tuning for Large Language Models: A Case Study on Detoxification (2024.findings-emnlp)
Copied to clipboard
Tao Meng, Ninareh Mehrabi, Palash Goyal, Anil Ramakrishna, Aram Galstyan, Richard Zemel, Kai-Wei Chang, Rahul Gupta, Charith Peris
| Challenge: | Using a sequence-level constraint, we regularize the LLMtraining by penalizing the KL divergence between the desired output distribution and the LRM’s posterior. |
| Approach: | They propose a constraint learning schema forfine-tuning Large Language Models with attribute control by penalizing the KL divergence be-tween the desired output distribution and the LLM's posterior. |
| Outcome: | The proposed approach improves the performance of large language models while enhancing their utility and generation quality. |
Detoxifying Large Language Models via Knowledge Editing (2024.acl-long)
Copied to clipboard
Mengru Wang, Ningyu Zhang, Ziwen Xu, Zekun Xi, Shumin Deng, Yunzhi Yao, Qishen Zhang, Linyi Yang, Jindong Wang, Huajun Chen
| Challenge: | Existing methods to detoxify Large Language Models (LLMs) are limiting, but knowledge editing can be effective. |
| Approach: | They propose a baseline method to detoxify Large Language Models (LLMs) they propose supervised fine-tuning and reinforcement learning from human feedback (RLHF) |
| Outcome: | The proposed method reduces toxicity of large language models with one instance of tuning . it reduces the toxicity, while minimizing the toxins, the authors show . |
Self-Detoxifying Language Models via Toxification Reversal (2023.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods to generate toxic content in pretrained language models are resource-intensive and require additional components. |
| Approach: | They propose a method that enables the PLM itself to achieve "self-detoxification" they identify the toxification direction from the normal generation process to the one prompted with the negative prefix and then steer the generation to the reverse direction by manipulating the information movement within the attention layers. |
| Outcome: | The proposed method can achieve comparable performance with state-of-the-art methods without any fine-tuning or extra components. |
CMD: a framework for Context-aware Model self-Detoxification (2024.emnlp-main)
Copied to clipboard
| Challenge: | Existing methods of text detoxification fail to achieve a decent balance between effectiveness and generation quality. |
| Approach: | They propose a text detoxification framework that pays attention to both context and detoxification process. |
| Outcome: | Experiments on various LLMs show that the proposed framework can yield the best performance compared to baselines. |
Adaptive Detoxification: Safeguarding General Capabilities of LLMs through Toxicity-Aware Knowledge Editing (2025.findings-acl)
Copied to clipboard
Yifan Lu, Jing Li, Yigeng Zhou, Yihui Zhang, Wenya Wang, Xiucheng Li, Meishan Zhang, Fangming Liu, Jun Yu, Min Zhang
| Challenge: | Existing knowledge editing methods for large language models (LLMs) suffer from over-editing, where detoxified models reject legitimate queries, compromising overall performance. |
| Approach: | They propose a toxicity-aware knowledge editing approach that dynamically detects toxic activation patterns during forward propagation and then routes computations through adaptive inter-layer pathways to mitigate toxicity effectively. |
| Outcome: | The proposed method outperforms existing methods on large language models and enhances the SafeEdit benchmark. |
Breaking Bad Tokens: Detoxification of LLMs Using Sparse Autoencoders (2025.emnlp-main)
Copied to clipboard
| Challenge: | Large language models (LLMs) are ubiquitous in user-facing applications, yet they still generate undesirable toxic outputs, including profanity, vulgarity, and derogatory remarks. |
| Approach: | They leverage sparse autoencoders to identify toxicity-related directions in residual stream of large language models and perform targeted activation steering using the corresponding decoder vectors. |
| Outcome: | The proposed models surpass baselines in reducing toxicity by up to 20%, though fluency can degrade noticeably on GPT-2 Small and Gemma-2-2B. |
Unlocking Anticipatory Text Generation: A Constrained Approach for Large Language Models Decoding (2024.emnlp-main)
Copied to clipboard
| Challenge: | Large language models have shown a powerful ability for text generation, but undesired behaviors such as toxicity and hallucinations can manifest. |
| Approach: | They propose to formalize text generation as a future-constrained generation problem to minimize undesirable behaviors and enforce faithfulness to instructions. |
| Outcome: | The proposed approach is effective across three tasks, including keyword-constrained generation, toxicity reduction, and factual correctness in question-answering. |
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. |