Copertina

Mechanistic Steering of LLMs Reveals Layer-wise Feature Vulnerabilities in Adversarial Settings

Fuente: arXiv

Salvato in:

Dettagli Bibliografici
Autori principali:	Das, Nilanjana, Gaur, Manas
Natura:	Preprint
Pubblicazione:	2026
Soggetti:	Computation and Language Artificial Intelligence
Accesso online:	Acceder al recurso
Tags:	Aggiungi Tag Nessun Tag, puoi essere il primo ad aggiungerne!!

Documenti analoghi

Human-Readable Adversarial Prompts: An Investigation into LLM Vulnerabilities Using Situational Context
di: Das, Nilanjana, et al.
Pubblicazione: (2024)

Human-Interpretable Adversarial Prompt Attack on Large Language Models with Situational Context
di: Das, Nilanjana, et al.
Pubblicazione: (2024)

Mechanistic Knobs in LLMs: Retrieving and Steering High-Order Semantic Features via Sparse Autoencoders
di: Zhang, Ruikang, et al.
Pubblicazione: (2026)

Mental Health Equity in LLMs: Leveraging Multi-Hop Question Answering to Detect Amplified and Silenced Perspectives
di: Haider, Batool, et al.
Pubblicazione: (2025)

Experiments or Outcomes? Probing Scientific Feasibility in Large Language Models
di: Mohammadi, Seyedali, et al.
Pubblicazione: (2026)

SymLoc: Symbolic Localization of Hallucination across HaluEval and TruthfulQA
di: Lamba, Naveen, et al.
Pubblicazione: (2025)

Investigating Symbolic Triggers of Hallucination in Gemma Models Across HaluEval and TruthfulQA
di: Lamba, Naveen, et al.
Pubblicazione: (2025)

Exploring the Personality Traits of LLMs through Latent Features Steering
di: Yang, Shu, et al.
Pubblicazione: (2024)

Neurosymbolic Retrievers for Retrieval-augmented Generation
di: Saxena, Yash, et al.
Pubblicazione: (2026)

Focus On This, Not That! Steering LLMs with Adaptive Feature Specification
di: Lamb, Tom A., et al.
Pubblicazione: (2024)

Exploitation Without Deception: Dark Triad Feature Steering Reveals Separable Antisocial Circuits in Language Models
di: Berg, Cameron, et al.
Pubblicazione: (2026)

Towards Inference-time Category-wise Safety Steering for Large Language Models
di: Bhattacharjee, Amrita, et al.
Pubblicazione: (2024)

Neural FOXP2 -- Language Specific Neuron Steering for Targeted Language Improvement in LLMs
di: Saha, Anusa, et al.
Pubblicazione: (2026)

What Drives Representation Steering? A Mechanistic Case Study on Steering Refusal
di: Cheng, Stephen, et al.
Pubblicazione: (2026)

Beyond Memorization: Testing LLM Reasoning on Unseen Theory of Computation Tasks
di: Shelat, Shlok, et al.
Pubblicazione: (2026)

From Guessing to Asking: An Approach to Resolving the Persona Knowledge Gap in LLMs during Multi-Turn Conversations
di: Baskar, Sarvesh, et al.
Pubblicazione: (2025)

Layer-wise Regularized Dropout for Neural Language Models
di: Ni, Shiwen, et al.
Pubblicazione: (2024)

SaGE: Evaluating Moral Consistency in Large Language Models
di: Bonagiri, Vamshi Krishna, et al.
Pubblicazione: (2024)

Towards Robust Evaluation of Unlearning in LLMs via Data Transformations
di: Joshi, Abhinav, et al.
Pubblicazione: (2024)

Do LLMs Adhere to Label Definitions? Examining Their Receptivity to External Label Definitions
di: Mohammadi, Seyedali, et al.
Pubblicazione: (2025)

CogSteer: Cognition-Inspired Selective Layer Intervention for Efficiently Steering Large Language Models
di: Wang, Xintong, et al.
Pubblicazione: (2024)

DESTEIN: Navigating Detoxification of Language Models via Universal Steering Pairs and Head-wise Activation Fusion
di: Li, Yu, et al.
Pubblicazione: (2024)

Layer-wise Positional Bias in Short-Context Language Modeling
di: Rahimi, Maryam, et al.
Pubblicazione: (2026)

Unsupervised Layer-wise Score Aggregation for Textual OOD Detection
di: Darrin, Maxime, et al.
Pubblicazione: (2023)

IoT-Based Preventive Mental Health Using Knowledge Graphs and Standards for Better Well-Being
di: Gyrard, Amelie, et al.
Pubblicazione: (2024)

WellDunn: On the Robustness and Explainability of Language Models and Large Language Models in Identifying Wellness Dimensions
di: Mohammadi, Seyedali, et al.
Pubblicazione: (2024)

Psychological Steering in LLMs: An Evaluation of Effectiveness and Trustworthiness
di: Banayeeanzade, Amin, et al.
Pubblicazione: (2025)

KV Cache Steering for Controlling Frozen LLMs
di: Belitsky, Max, et al.
Pubblicazione: (2025)

Unpacking Robustness in Inflectional Languages: Adversarial Evaluation and Mechanistic Insights
di: Walkowiak, Paweł, et al.
Pubblicazione: (2025)

Evolutionary Feature-wise Thresholding for Binary Representation of NLP Embeddings
di: Sinha, Soumen, et al.
Pubblicazione: (2025)

Efficient Layer-wise LLM Fine-tuning for Revision Intention Prediction
di: Liu, Zhexiong, et al.
Pubblicazione: (2025)

Steering Towards Fairness: Mitigating Political Bias in LLMs
di: Nadeem, Afrozah, et al.
Pubblicazione: (2025)

Towards Reliable Evaluation of Behavior Steering Interventions in LLMs
di: Pres, Itamar, et al.
Pubblicazione: (2024)

Dissecting Bias in LLMs: A Mechanistic Interpretability Perspective
di: Chandna, Bhavik, et al.
Pubblicazione: (2025)

Revealing the Intrinsic Ethical Vulnerability of Aligned Large Language Models
di: Lian, Jiawei, et al.
Pubblicazione: (2025)

Can LLMs Obfuscate Code? A Systematic Analysis of Large Language Models into Assembly Code Obfuscation
di: Mohseni, Seyedreza, et al.
Pubblicazione: (2024)

A Framework to Assess Multilingual Vulnerabilities of LLMs
di: Tang, Likai, et al.
Pubblicazione: (2025)

Mechanistic Interpretability of GPT-2: Lexical and Contextual Layers in Sentiment Analysis
di: Hatua, Amartya
Pubblicazione: (2025)

Steering Without Breaking: Mechanistically Informed Interventions for Discrete Diffusion Language Models
di: Zhou, Hanhan, et al.
Pubblicazione: (2026)

Analysing Moral Bias in Finetuned LLMs through Mechanistic Interpretability
di: Raimondi, Bianca, et al.
Pubblicazione: (2025)