Cover Image

Safety Arithmetic: A Framework for Test-time Safety Alignment of Language Models by Steering Parameters and Activations

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Hazra, Rima, Layek, Sayan, Banerjee, Somnath, Poria, Soujanya
Format:	Preprint
Published:	2024
Subjects:	Computation and Language
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Sowing the Wind, Reaping the Whirlwind: The Impact of Editing Language Models
by: Hazra, Rima, et al.
Published: (2024)

Soteria: Language-Specific Functional Parameter Steering for Multilingual Safety Alignment
by: Banerjee, Somnath, et al.
Published: (2025)

SafeInfer: Context Adaptive Decoding Time Safety Alignment for Large Language Models
by: Banerjee, Somnath, et al.
Published: (2024)

ProSocialAlign: Preference Conditioned Test Time Alignment in Language Models
by: Banerjee, Somnath, et al.
Published: (2025)

Bridging the Multilingual Safety Divide: Efficient, Culturally-Aware Alignment for Global South Languages
by: Banerjee, Somnath, et al.
Published: (2026)

Attributional Safety Failures in Large Language Models under Code-Mixed Perturbations
by: Banerjee, Somnath, et al.
Published: (2025)

How (un)ethical are instruction-centric responses of LLMs? Unveiling the vulnerabilities of safety guardrails to harmful queries
by: Banerjee, Somnath, et al.
Published: (2024)

Language Models are Homer Simpson! Safety Re-Alignment of Fine-tuned Language Models through Task Arithmetic
by: Bhardwaj, Rishabh, et al.
Published: (2024)

SafeTutors: Benchmarking Pedagogical Safety in AI Tutoring Systems
by: Hazra, Rima, et al.
Published: (2026)

Context Matters: Pushing the Boundaries of Open-Ended Answer Generation with Graph-Structured Knowledge Context
by: Banerjee, Somnath, et al.
Published: (2024)

MemeSense: An Adaptive In-Context Framework for Social Commonsense Driven Meme Moderation
by: Adak, Sayantan, et al.
Published: (2025)

Breaking Boundaries: Investigating the Effects of Model Editing on Cross-linguistic Performance
by: Banerjee, Somnath, et al.
Published: (2024)

SafeMath: Inference-time Safety improves Math Accuracy
by: Basu, Sagnik, et al.
Published: (2026)

Navigating the Cultural Kaleidoscope: A Hitchhiker's Guide to Sensitivity in Large Language Models
by: Banerjee, Somnath, et al.
Published: (2024)

AURA: Affordance-Understanding and Risk-aware Alignment Technique for Large Language Models
by: Adak, Sayantan, et al.
Published: (2025)

Lost in Interpretation: The Plausibility-Faithfulness Trade-off in Cross-Lingual Explanations
by: Banerjee, Somnath, et al.
Published: (2026)

DistALANER: Distantly Supervised Active Learning Augmented Named Entity Recognition in the Open Source Software Ecosystem
by: Banerjee, Somnath, et al.
Published: (2024)

Towards Robust Instruction Tuning on Multimodal Large Language Models
by: Han, Wei, et al.
Published: (2024)

Evaluating the Ebb and Flow: An In-depth Analysis of Question-Answering Trends across Diverse Platforms
by: Hazra, Rima, et al.
Published: (2023)

Redefining Developer Assistance: Through Large Language Models in Software Ecosystem
by: Banerjee, Somnath, et al.
Published: (2023)

Trojan Activation Attack: Red-Teaming Large Language Models using Activation Steering for Safety-Alignment
by: Wang, Haoran, et al.
Published: (2023)

Inference Time Alignment with Reward-Guided Tree Search
by: Hung, Chia-Yu, et al.
Published: (2024)

Not All Votes Count! Programs as Verifiers Improve Self-Consistency of Language Models for Math Reasoning
by: Toh, Vernon Y. H., et al.
Published: (2024)

Tutoring Large Language Models to be Domain-adaptive, Precise, and Safe
by: Banerjee, Somnath
Published: (2026)

DELLA-Merging: Reducing Interference in Model Merging through Magnitude-Based Sampling
by: Deep, Pala Tej, et al.
Published: (2024)

SteeringSafety: A Systematic Safety Evaluation Framework of Representation Steering in LLMs
by: Siu, Vincent, et al.
Published: (2025)

WalledEval: A Comprehensive Safety Evaluation Toolkit for Large Language Models
by: Gupta, Prannaya, et al.
Published: (2024)

Can-Do! A Dataset and Neuro-Symbolic Grounded Framework for Embodied Planning with Large Multimodal Models
by: Chia, Yew Ken, et al.
Published: (2024)

DiffPO: Diffusion-styled Preference Optimization for Efficient Inference-Time Alignment of Large Language Models
by: Chen, Ruizhe, et al.
Published: (2025)

Exploring Safety-Utility Trade-Offs in Personalized Language Models
by: Vijjini, Anvesh Rao, et al.
Published: (2024)

PREMISE: Matching-based Prediction for Accurate Review Recommendation
by: Han, Wei, et al.
Published: (2025)

Understanding the Capabilities and Limitations of Large Language Models for Cultural Commonsense
by: Shen, Siqi, et al.
Published: (2024)

Leveraging Parameter-Efficient Transfer Learning for Multi-Lingual Text-to-Speech Adaptation
by: Li, Yingting, et al.
Published: (2024)

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

Steering Language Models with Weight Arithmetic
by: Fierro, Constanza, et al.
Published: (2025)

R1-ACT: Efficient Reasoning Model Safety Alignment by Activating Safety Knowledge
by: In, Yeonjun, et al.
Published: (2025)

Ruby Teaming: Improving Quality Diversity Search with Memory for Automated Red Teaming
by: Han, Vernon Toh Yan, et al.
Published: (2024)

Harnessing Large Language Models for Scientific Novelty Detection
by: Liu, Yan, et al.
Published: (2025)

SafeSteer: Localized On-Policy Distillation for Efficient Safety Alignment
by: Li, Hao, et al.
Published: (2026)

HyperTTS: Parameter Efficient Adaptation in Text to Speech using Hypernetworks
by: Li, Yingting, et al.
Published: (2024)