OpenAssistant/oasst1

OpenAssistant Conversations Dataset (OASST1) Dataset Summary In an effort to democratize research on large-scale alignment, we release OpenAssistant Conversations (OASST1), a human-generated, human-annotated assistant-style conversation corpus consisting of 161,443 messages in 35 different languages, annotated with 461,292 quality ratings, resulting in over 10,000 fully annotated conversation trees. The corpus is a product of a worldwide crowd-sourcing effort… See the full description on the dataset page: https://huggingface.co/datasets/OpenAssistant/oasst1.