Factored Verification: Detecting and reducing hallucinations in frontier models using AI supervision
We evaluate a simple automated method for detecting hallucinations in abstractive summaries of academic papers
Research
A collection of 1 post