multi-agent judge framework

New framework automates LLM creativity evaluation

Researchers have developed a new automated framework to evaluate the creativity of large language models (LLMs) across various open-ended tasks. This domain-agnostic approach uses semantic entropy to measure divergent c…