stefanbschneider/lfqa-max-answer-length-1024
收藏Hugging Face2025-02-08 更新2025-02-15 收录
下载链接：
https://hf-mirror.com/datasets/stefanbschneider/lfqa-max-answer-length-1024
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含问题和对应的上下文及简单、长篇的答案，是一个过滤后的版本，原始数据集来源于LLukas22/lfqa_preprocessed，该数据集又是基于vblagoje的lfqa_support_docs和lfqa数据集的处理和简化版本。数据集通过LED tokenizer对答案的token数量进行过滤，移除了过长的答案。数据集分为训练集和验证集，适用于问题回答和文本生成任务。
The dataset contains questions, corresponding contexts, and simple, long-form answers. It is a filtered version based on LLukas22/lfqa_preprocessed, which is a processed and simplified version of vblagojes lfqa_support_docs and lfqa datasets. The dataset has been filtered using the LED tokenizer based on the number of tokens in the answers to remove overly long answers. The dataset is split into training and validation sets and is suitable for question-answering and text2text generation tasks.
提供机构：
stefanbschneider
