search

jzbjyb / FLARE

Forward-Looking Active REtrieval-augmented generation (FLARE)
MIT License
545 stars 50 forks source link

Could you share WikiAsp dataset used in the experiments? #5

Closed jihyukkim-nlp closed 11 months ago

jihyukkim-nlp commented 12 months ago

Hi all, I am Jihyuk, a PhD student interested in retrieval-augmented LLMs. I appreciate the open-sourcing of codes!

I am wondering if WikiAsp dataset used in the experiments can also be shared.

I noticed that the original, open-sourced WikiAsp dataset only includes summaries and reference documents. But, it does not include inputs, e.g., "Generate a summary about Joe Biden", which are needed for FLARE.

Best regards, Jihyuk

jzbjyb commented 11 months ago

Thanks for the question and sorry for getting back to you late! I just uploaded the WikiAsp dataset and added instructions on how to run experiments on this dataset in the readme file. You need to setup Bing search before prompting OpenAI models.

jihyukkim-nlp commented 11 months ago

Thank you so much for sharing dataset and providing helpful instructions!