AlignmentResearch/diverse-deception download history
AlignmentResearch/diverse-deception is a text classification dataset on the Hugging Face Hub. In the last 30 days it was downloaded 159 times (39 in the last 7 days), and 345 times in total. It ranks #106,255 among datasets by monthly downloads.
Diverse Deception A selection of diverse deception datasets, collected to train and evaluate probes that detect when a language model's answer is deceptive. Every row is one conversation whose final assistant turn is labeled honest or deceptive. The rows come from ten public sources t
Open AlignmentResearch/diverse-deception on Hugging Face
Sister project: Paper Pulse, the upvote history of every Hugging Face Daily Paper.