Mario Fritz

Competition reveals vulnerabilities and defenses in language model security.

An overview of sampling methods for text generation.

This article examines how LLMs can identify missing variables in scientific research.

LM-GC offers a new way to compress gradients in machine learning.

Researchers develop methods to teach computers to process invoices while protecting privacy.

ICER framework tests safety measures in text-to-image models effectively.

Innovative methods ensure privacy while generating realistic synthetic data.