遇见数据集

mksethi/llama-3.1-8b-Instruct_sae_reps

收藏
Hugging Face2025-10-08 更新2025-10-25 收录
官方服务:

资源简介:

该数据集包含问题、知识、答案、虚构信息等字段,以及用于模型处理的其他技术特征,如input_ids、attention_mask和sae。数据集被分为多个层,每层有26211个示例。数据集的下载大小为61.88 GB,实际大小为97.21 GB。

The dataset includes fields such as questions, knowledge, answers, and hallucination, as well as other technical features for model processing like input_ids, attention_mask, and sae. The dataset is divided into multiple layers, each containing 26,211 examples. The download size of the dataset is 61.88 GB, and the actual size is 97.21 GB.

提供机构:
mksethi
二维码
社区交流群
二维码
科研交流群
商业服务