nbeerbower/weasel-dpo download history

nbeerbower/weasel-dpo is a text generation dataset on the Hugging Face Hub. In the last 30 days it was downloaded 46 times (12 in the last 7 days), and 258 times in total. It ranks #283,866 among datasets by monthly downloads.

weasel-dpo A preference dataset targeting hedging and evasion in LLM answers to empirical questions. chosen responses lead with a magnitude and name their sources; rejected responses deliver the same claims with the information stripped out. 144 pairs (130 train / 14 test). Generated

Models trained on weasel-dpo

10 models list it as training data.

Open nbeerbower/weasel-dpo on Hugging Face

Sister project: Paper Pulse, the upvote history of every Hugging Face Daily Paper.