---
license: mit
task_categories:
- question-answering
language:
- en
tags:
- HotpotQA
- TriviaQA
- NaturalQuestions
- MuSiQue
- 2WikiMultihopQA
- PAI-2
size_categories:
- 1K<n<10K
---
This is a repository with a preprocessed well known benchmarks that were used for PAI-2 (GraphRAG-method from [PAI framework](https://github.com/Dzigen/PersonalAI)) evaluation on QA tasks. More details in our paper: arxiv.org/abs/2605.13481.