Raschka Blog Corpus
收藏arXiv2025-09-30 收录
下载链接：
https://sebastianraschka.com/blog/2024/new-llm-pre-training-and-post-training.html
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含了从一篇最新博客文章中提取的非重叠文本块，这些文本块被分割成了100个字符和1000个字符的片段。此外，该数据集由英文文本块组成，用于评估不同模型对原始文本的重建能力。在评估中，共使用了20个序列，任务的目的是从嵌入式中重建原始文本。
This dataset contains non-overlapping text chunks extracted from a recent blog post, which are split into segments of 100 characters and 1000 characters respectively. Furthermore, this dataset consists solely of English text chunks and is designed to evaluate the ability of various models to reconstruct the original text. In the evaluation, 20 sequences are utilized, with the task objective being to reconstruct the original text from their embeddings.
提供机构：
Sebastian Raschka
