KomeijiForce/Cuckoo_MetaIE_MultiNTE_Fineweb

Cuckoo数据集是一个用于训练信息提取模型的文本数据集，包含转换为下一个标记提取（NTE）实例的数据。这些数据用于训练Cuckoo模型，该模型能够模仿大规模语言模型的下一个标记预测范式，通过在给定的输入上下文中对标记进行标注来预测下一个标记。The Cuckoo dataset is a collection of text data for training information extraction models, containing data converted into next-token extraction (NTE) instances. This data is used to train the Cuckoo model, which is capable of mimicking the next-token prediction paradigm of large-scale language models by tagging tokens in the given input context to predict the next tokens.