LLaMA2-7B
收藏arXiv2025-09-30 收录
下载链接：
https://arxiv.org/abs/2203.02155
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是对LLaMA模型进行更新的版本，主要用于评估微调方法。在此次研究中，它与RoRA、LoRA和DoRA进行了微调效果的对比。该数据集包含了70亿个参数，并针对常识推理任务进行了优化。
This dataset is an updated version of the LLaMA model, primarily designed for evaluating fine-tuning methods. In this study, its fine-tuning performance is compared against that of RoRA, LoRA, and DoRA. It contains 7 billion parameters and has been optimized for common sense reasoning tasks.
提供机构：
Meta AI
