tasksource/tasksource_dpo_pairs download history

tasksource/tasksource_dpo_pairs is a text generation dataset on the Hugging Face Hub. In the last 30 days it was downloaded 317 times (83 in the last 7 days), and 8,106 times in total. It ranks #47,032 among datasets by monthly downloads.

tasksource_dpo_pairs Preference pairs from ~470 human-annotated classification and multiple-choice tasks of tasksource, for DPO and other preference training. Each pair takes one example from tasksource-instruct. chosen is the gold answer, and rejected is another answer the prompt off

Models trained on tasksource_dpo_pairs

3 models list it as training data.

Open tasksource/tasksource_dpo_pairs on Hugging Face

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