McGill-NLP/zsre_qa

该数据集包含多个配置，每个配置用于不同场景的文本处理任务，包括问题回答、任务分类等。具体包含以下几种配置：

1. flan_few_shot：用于少样本学习的问题回答任务。
2. flan_zero_shot：用于零样本学习的问题回答任务。
3. forget_high_count、forget_low_count、forget_medium_count及其对应的paraphrased版本：用于研究在不同n-gram计数条件下的文本遗忘问题。
4. tulu_full_test：包含完整测试数据的问题回答数据集。
5. world_facts：包含世界知识的问题回答数据集。

The dataset consists of multiple configurations, each designed for different text processing tasks such as question answering and task classification. It includes the following configurations:

1. flan_few_shot: For few-shot learning in question answering tasks.
2. flan_zero_shot: For zero-shot learning in question answering tasks.
3. forget_high_count, forget_low_count, forget_medium_count and their paraphrased versions: For studying text forgetting issues under different n-gram count conditions.
4. tulu_full_test: A question answering dataset with complete test data.
5. world_facts: A question answering dataset containing world knowledge.