Lo-Fi-gahara/math-shepherd
收藏Hugging Face2024-11-30 更新2024-12-14 收录
下载链接：
https://hf-mirror.com/datasets/Lo-Fi-gahara/math-shepherd
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是从HuggingFace上的peiyi9979/Math-Shepherd构建的，并将推理步骤转换为LlaMA3.1-8b-Instruct（基础预训练版本）每一层的隐藏状态。该数据集用于训练分类器和干预器，以从基础模型中学习内部表示。
The dataset is built from peiyi9979/Math-Shepherd on HuggingFace, with reasoning steps converted into hidden states from each layer of LlaMA3.1-8b-Instruct (base pretrained version). The dataset is for training classifiers and interventors to learn inner representations from the base model.
提供机构：
Lo-Fi-gahara
