argilla/FinePersonas-v0.1

FinePersonas数据集包含2100万条详细的人物角色描述，用于生成多样化和可控的合成文本。该数据集基于HuggingFaceFW/fineweb-edu数据集中的教育领域网页内容生成，具有强烈的教育和科学领域偏向。数据集分为两个子集：default包含id、persona和labels列，embeddings包含id、model_name_embeddings和embedding列。数据集的主要用途包括生成指令、生成多样化的文本、生成特定角色的问题以及生成图像。

FinePersonas is an open dataset containing 21 million detailed personas for diverse and controllable synthetic text generation. The dataset is grounded on webpages from the HuggingFaceFW/fineweb-edu dataset, with a strong bias towards personas in the education and scientific domain. It includes two subsets: default with columns id, persona, and labels, and embeddings with columns id, model_name_embeddings, and embedding. The dataset is primarily used for generating instructions, diverse text, persona-specific problems, and images.