Suraj Srinivas

New methods improve how machine learning handles noisy data.

Introducing Verifiability Tuning for clearer and trustworthy AI explanations.

A new method enhances the safety of language models against harmful prompts.

SpLiCE helps clarify the dense data from CLIP for better understanding.

Research reveals similarities in image models' internal representations.

A straightforward look at large language models and their workings.