MINTQA: A Multi-Hop Question Answering Benchmark for Evaluating LLMs on New and Tail Knowledge

We are currently organizing the entire project code and will be releasing it soon.

We public the annotated MINTQA dataset, divided into two files: MINTQA-POP and MINTQA-TI.

The MINTQA Knowledge Graph (a subset of Wikidata) is available on Hugging Face:

🔗 Check the MINTQA KG here