qingy2024/Pristine-SFT-500k download history

qingy2024/Pristine-SFT-500k is a dataset on the Hugging Face Hub. In the last 30 days it was downloaded 98 times (25 in the last 7 days), and 1,571 times in total. It ranks #115,511 among datasets by monthly downloads.

Pristine is a dataset of 500k prompts/reasoning traces/responses for SFT. It contains responses from GPT-OSS-120B and 20B with a variety of reasoning efforts. For math and code domains, it uses high reasoning effort. For chat domain it uses low / medium reasoning efforts. Dataset Sources: qingy

Open qingy2024/Pristine-SFT-500k on Hugging Face

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