English
Related papers

Related papers: Breaking Bad Tokens: Detoxification of LLMs Using …

200 papers

Sparse Autoencoders (SAEs) have been successfully used to probe Large Language Models (LLMs) and extract interpretable concepts from their internal representations. These concepts are linear combinations of neuron activations that…

Computation and Language · Computer Science 2026-02-23 Mathis Le Bail , Jérémie Dentan , Davide Buscaldi , Sonia Vanier

We use sparse autoencoder (SAE) feature steering to amplify Dark Triad personality traits (Machiavellianism, narcissism, and psychopathy) in Llama-3.3-70B-Instruct and evaluate the resulting behavioral changes across five psychological…

Computation and Language · Computer Science 2026-05-12 Cameron Berg , Roshni Lulla

Sparse autoencoders (SAEs) decompose large language model (LLM) activations into latent features that reveal mechanistic structure. Conventional SAEs train on broad data distributions, forcing a fixed latent budget to capture only…

Machine Learning · Computer Science 2025-08-14 Charles O'Neill , Mudith Jayasekara , Max Kirkby

Sparse Autoencoders (SAEs) can extract interpretable features from large language models (LLMs) without supervision. However, their effectiveness in downstream steering tasks is limited by the requirement for contrastive datasets or large…

Computation and Language · Computer Science 2026-05-05 Seonglae Cho , Zekun Wu , Adriano Koshiyama

We investigate whether sparse autoencoders (SAEs) can be used to remove knowledge from language models. We use the biology subset of the Weapons of Mass Destruction Proxy dataset and test on the gemma-2b-it and gemma-2-2b-it language…

Machine Learning · Computer Science 2024-11-05 Eoin Farrell , Yeu-Tong Lau , Arthur Conmy

Refusal on harmful prompts is a key safety behaviour in instruction-tuned large language models (LLMs), yet the internal causes of this behaviour remain poorly understood. We study two public instruction-tuned models, Gemma-2-2B-IT and…

Computation and Language · Computer Science 2026-04-29 Nirmalendu Prakash , Yeo Wei Jie , Amir Abdullah , Ranjan Satapathy , Erik Cambria , Roy Ka Wei Lee

Large language models (LLMs) exhibit exceptional performance but pose inherent risks of generating toxic content, restricting their safe deployment. While traditional methods (e.g., alignment) adjust output preferences, they fail to…

Machine Learning · Computer Science 2026-01-13 Zenghao Duan , Zhiyi Yin , Zhichao Shi , Liang Pang , Shaoling Jing , Zihe Huang , Jiayi Wu , Yu Yan , Jingcheng Deng , Huawei Shen , Xueqi Cheng

Large language models (LLMs) are increasingly trained in complex Reinforcement Learning, multi-agent environments, making it difficult to understand how behavior changes over training. Sparse Autoencoders (SAEs) have recently shown to be…

Machine Learning · Computer Science 2026-02-09 John Yan , Michael Yu , Yuqi Sun , Alexander Duffy , Tyler Marques , Matthew Lyle Olson

Software vulnerabilities such as buffer overflows and SQL injections are a major source of security breaches. Traditional methods for vulnerability detection remain essential but are limited by high false positive rates, scalability issues,…

Software Engineering · Computer Science 2026-04-10 Rui Melo , Claudia Mamede , Andre Catarino , Rui Abreu , Henrique Lopes Cardoso

Large language models frequently generate toxic, hateful, or harmful content, yet existing mitigation methods rely on costly retraining or output-level filtering with no mechanistic insight into where toxicity originates internally. We…

Computation and Language · Computer Science 2026-05-28 Himanshu Beniwal , Mayank Singh

Ensuring the safety and alignment of large language models (LLMs) with human values is crucial for generating responses that are beneficial to humanity. While LLMs have the capability to identify and avoid harmful queries, they remain…

Computation and Language · Computer Science 2024-10-22 Yihua Zhou , Xiaochuan Shi

Analyzing large-scale text corpora is a core challenge in machine learning, crucial for tasks like identifying undesirable model behaviors or biases in training data. Current methods often rely on costly LLM-based techniques (e.g.…

Artificial Intelligence · Computer Science 2025-12-12 Nick Jiang , Xiaoqing Sun , Lisa Dunlap , Lewis Smith , Neel Nanda

Large Language Models (LLMs) are trained on massive, unstructured corpora, making it unclear which social patterns and biases they absorb and later reproduce. Existing evaluations typically examine outputs or activations, but rarely connect…

Computation and Language · Computer Science 2025-11-14 Mariam Mahran , Katharina Simbeck

The evaluation of large language models (LLMs) relies heavily on standardized benchmarks. These benchmarks provide useful aggregated metrics for a given capability, but those aggregated metrics can obscure (i) particular sub-areas where the…

Computation and Language · Computer Science 2025-12-25 Matyas Bohacek , Nino Scherrer , Nicholas Dufour , Thomas Leung , Christoph Bregler , Stephanie C. Y. Chan

With adversarial or otherwise normal prompts, existing large language models (LLM) can be pushed to generate toxic discourses. One way to reduce the risk of LLMs generating undesired discourses is to alter the training of the LLM. This can…

Computation and Language · Computer Science 2023-02-28 Meng Cao , Mehdi Fatemi , Jackie Chi Kit Cheung , Samira Shabanian

Large Language Models (LLMs) have demonstrated impressive performance across various tasks, yet they remain vulnerable to generating toxic content, necessitating detoxification strategies to ensure safe and responsible deployment. Test-time…

Computation and Language · Computer Science 2025-10-03 Yisong Xiao , Aishan Liu , Siyuan Liang , Zonghao Ying , Xianglong Liu , Dacheng Tao

Sparse autoencoders (SAEs) have become a standard tool for mechanistic interpretability in autoregressive large language models (LLMs), enabling researchers to extract sparse, human-interpretable features and intervene on model behavior.…

Machine Learning · Computer Science 2026-02-06 Xu Wang , Bingqing Jiang , Yu Wan , Baosong Yang , Lingpeng Kong , Difan Zou

Large Language Models (LLMs) can acquire deceptive behaviors through backdoor attacks, where the model executes prohibited actions whenever secret triggers appear in the input. Existing safety training methods largely fail to address this…

Cryptography and Security · Computer Science 2025-10-08 Guangyu Shen , Siyuan Cheng , Xiangzhe Xu , Yuan Zhou , Hanxi Guo , Zhuo Zhang , Xiangyu Zhang

As Large Language Models become integral to software development, with substantial portions of AI-suggested code entering production, understanding their internal correctness mechanisms becomes critical for safe deployment. We apply sparse…

Software Engineering · Computer Science 2025-10-06 Kriz Tahimic , Charibeth Cheng

As large language models (LLMs) become increasingly prevalent in global applications, ensuring that they are toxicity-free across diverse linguistic contexts remains a critical challenge. We explore "Cross-lingual Detoxification", a…

Computation and Language · Computer Science 2025-10-24 Himanshu Beniwal , Youngwoo Kim , Maarten Sap , Soham Dan , Thomas Hartvigsen