Hotpot QA

Summary

The "hotpot_qa" dataset is designed for the task of Question Answering in English. It is monolingual and falls under the size category of 100K to 1M. The dataset is original and has been crowdsourced.

Size

The dataset consists of different subsets: Distractor: 97.9k rows Fullwiki: 105k rows Train: 90.4k rows Validation: 7.41k rows

Use cases

The "hotpot_qa" dataset is suitable for developing and evaluating models for multi-hop question answering. It can be used to train models to understand and answer complex questions that require reasoning over multiple pieces of information.

License

The dataset is licensed under the CC BY-SA 4.0 License.