Pratyush Maini

Safety Pretraining: Toward the Next Generation of Safe AI

STAMP Your Content: Proving Dataset Membership via Watermarked Rephrasings

Unlocking Post-hoc Dataset Inference with Synthetic Data

Memorization Sinks: Isolating Memorization during LLM Training

TOFU: A Task of Fictitious Unlearning for LLMs

T-MARS: Improving Visual Representations by Circumventing Text Feature Learning

LLM Dataset Inference: Did you train on my dataset?

Rethinking LLM Memorization through the Lens of Adversarial Compression

Understanding Hallucinations in Diffusion Models through Mode Interpolation