open-paws/reasoning-llama-format

Open Paws推理Llama格式数据集是Open Paws项目的一部分，用于训练理解并促进动物福利、权利和解放的AI系统。该数据集采用JSONL格式，主要包含多语言（以英语为主）的动物倡导和伦理推理内容。数据集的结构包括Parquet格式的样本，包含带有Llama格式<think>标签的输入提示和推理链，以及质量评估分数。

The Open Paws Reasoning Llama Format dataset is part of the Open Paws initiative, designed to train AI systems that understand and promote animal welfare, rights, and liberation. The dataset is in JSONL format, primarily containing multilingual (mainly English) content focused on animal advocacy and ethical reasoning. The dataset structure includes samples in Parquet format, which contain input prompts and reasoning chains with <think> tags in Llama format, along with quality assessment scores.