VinceGx33/mistral-legal-french-dataset

Mistral Legal French Dataset是一个为法语法律领域优化的微调数据集，采用了课程学习策略。数据集包含法律问答和案例法律推理两种类型的数据，共14,875个示例，其中10,000个示例为法律问答（LegalKit），4,875个示例为案例法律推理（COT）。数据集的结构遵循ChatML格式，适用于AutoTrain、HuggingFace TRL、Axolotl等工具。数据集的质量经过严格的验证，并提供了详细的指标。数据集的许可为Apache License 2.0。

Mistral Legal French Dataset is a fine-tuning dataset optimized for the French legal domain with curriculum learning strategy. The dataset includes two types of data: legal question answering (LegalKit) and case law reasoning (COT), with a total of 14,875 examples, 10,000 of which are legal question answering and 4,875 are case law reasoning. The dataset follows the ChatML format and is compatible with tools such as AutoTrain, HuggingFace TRL, and Axolotl. The quality of the dataset has been strictly validated with detailed metrics provided. The dataset is licensed under the Apache License 2.0.