PPL
Sachin Kumar
Papers (4)
Pressure-Testing Deception Probes in LLMs: Scaling, Robustness, and the Geometry of Deceptive RepresentationsHalluWorld: A Controlled Benchmark for Hallucination via Reference World ModelsActivation Differences Reveal Backdoors: A Comparison of SAE ArchitecturesLeveraging Pretrained Language Models as Energy Functions for Glauber Dynamics Text Diffusion