Tasksource

该数据集是一个包含超过500个英语任务的标注框架及其注释，涵盖了输入和标签列的元数据。这些注释能够辅助未来的数据集预处理工作，并且在这些任务上微调的模型，其性能超过了同等规模公开可用的文本编码器。任务范围包括但不限于释义检测、情感分析以及分类任务。

This dataset comprises an annotation framework and associated annotations for over 500 English tasks, along with metadata for input and label columns. These annotations support future dataset preprocessing efforts, and models fine-tuned on these tasks outperform publicly available text encoders of comparable scale. The covered tasks include, but are not limited to, paraphrase detection, sentiment analysis, and classification tasks.