Vishal Pramanik

Jailbreaking the Matrix: Nullspace Steering for Controlled Model Subversion

Hessian-Enhanced Token Attribution (HETA): Interpreting Autoregressive LLMs

Fact or Hallucination? An Entropy-Based Framework for Attention-Wise Usable Information in LLMs

Fact or Hallucination? An Entropy-Based Framework for Attention-Wise Usable Information in LLMs

Enhancing Integrated Gradients Using Emphasis Factors and Attention for Effective Explainability of Large Language Models