2 datasets found

The dataset used in this paper is the Anthropic Helpfulness and Harmlessness (HH) dataset, which features a wide variety of user queries and is commonly used for training...

The HH-RLHF dataset is a human preference dataset for reinforcement learning from human feedback.