nvidia/Llama-Nemotron-Post-Training-Dataset
收藏Hugging Face2025-04-09 更新2025-04-12 收录
下载链接：
https://hf-mirror.com/datasets/nvidia/Llama-Nemotron-Post-Training-Dataset
下载链接
链接失效反馈官方服务：
资源简介：
Llama-Nemotron后训练数据集是一个旨在提高Llama指令模型能力的 Compilation 数据集，包含数学、代码、通用推理和指令遵循等方面的改进。数据集分为SFT和RL配置，包含不同类型的数据文件，如代码、数学、科学、聊天和安全等类别。该数据集适用于训练AI系统，并遵循CC-BY-4.0许可。
The Llama-Nemotron Post-Training Dataset is a compilation designed to enhance the capabilities of the Llama instruct model, including improvements in math, code, general reasoning, and instruction following. The dataset is split into SFT and RL configurations with various data types such as code, math, science, chat, and safety. It is intended for use in training AI systems and is licensed under CC-BY-4.0.
提供机构：
nvidia
