Magpie-Align/Magpie-Reasoning-V2-250K-CoT-Deepseek-R1-Llama-70B
收藏Hugging Face2025-01-27 更新2025-02-15 收录
下载链接：
https://hf-mirror.com/datasets/Magpie-Align/Magpie-Reasoning-V2-250K-CoT-Deepseek-R1-Llama-70B
下载链接
链接失效反馈官方服务：
资源简介：
Magpie是一个通过提示对齐的语言模型（LLM）从零开始合成大规模对齐数据的方法生成的数据集。该数据集包含了由LLama 3.1 70B Instruct和LLama 3.3 70B Instruct生成的指令，以及由DeepSeek-R1-Distill-Llama-70B生成的响应。数据集的目的是为了增强模型的推理能力，通过使用高质量的指令-响应对。
Magpie is a dataset generated by a method of synthesizing large-scale alignment data from scratch by prompting aligned language models (LLMs). The dataset contains instructions generated by LLama 3.1 70B Instruct and LLama 3.3 70B Instruct, and responses generated by DeepSeek-R1-Distill-Llama-70B. The purpose of the dataset is to enhance the reasoning capabilities of models by using high-quality instruction-response pairs.
提供机构：
Magpie-Align
