13 papers
To Erase, or Not to Erase: Robust Training-Free Concept Erasure with Preservation aware Adaptive Ranked Subspace Expansion
Shaswati Saha, Rajasekhar Anguluri, Manas Gaur
Concept erasure techniques (CETs) edit text-to-image diffusion models to erase undesired targets such as NSFW content or copyrighted styles, while preserving model utility on benig…
From Concept-Aligned Tokens to Vulnerable Features: Mechanistic Localization of Jailbreaks
Nilanjana Das, Mathew Dawit, Aman Chadha +1
Jailbreak attacks expose a persistent failure mode in safety-aligned LLMs: models can be pushed into harmful behavior, but the internal representations enabling this shift remain p…
Ranking Free RAG: Replacing Re-ranking with Selection in RAG for Sensitive Domains
Yash Saxena, Ankur Padia, Mandar S Chaudhary +3
Retrieval-Augmented Generation (RAG) systems deployed in sensitive domains must provide interpretable evidence selection and robust safeguards against data poisoning, yet current a…
Structure-Aware Chunking for Tabular Data in Retrieval-Augmented Generation
Pooja Guttal, Varun Magotra, Vasudeva Mahavishnu +3
Tabular documents such as CSV and Excel files are widely used in enterprise data pipelines, yet existing chunking strategies for retrieval-augmented generation (RAG) are primarily…
Analyzing Chain of Thought (CoT) Approaches in Control Flow Code Deobfuscation Tasks
Seyedreza Mohseni, Sarvesh Baskar, Edward Raff +1
Code deobfuscation is the task of recovering a readable version of a program while preserving its original behavior. In practice, this often requires days or even months of manual…
Do LLMs Adhere to Label Definitions? Examining Their Receptivity to External Label Definitions
Seyedali Mohammadi, Bhaskara Hanuma Vedula, Hemank Lamba +4
Do LLMs genuinely incorporate external definitions, or do they primarily rely on their parametric knowledge? To address these questions, we conduct controlled experiments across mu…