dyyyyyyyy/FAPO-Reasoning-Dataset download history
dyyyyyyyy/FAPO-Reasoning-Dataset is a dataset on the Hugging Face Hub. In the last 30 days it was downloaded 82 times (9 in the last 7 days), and 994 times in total. It ranks #176,070 among datasets by monthly downloads.
FAPO: Flawed-Aware Policy Optimization for Efficient and Reliable Reasoning Dataset used to train the FAPO-32B model. train.parquet: data source from DAPO-Math-17K, with "\boxed{}" instruction format and duplicated 20 times. test.parquet: mixture of AIME24 (duplicated 32 times), AIME
Open dyyyyyyyy/FAPO-Reasoning-Dataset on Hugging Face
Sister project: Paper Pulse, the upvote history of every Hugging Face Daily Paper.