Unable to download the fever database and issue in requirements file

[amanagr6697]

Please provide an alternative link to download the database there are issues in the requirements.txt file, it is not getting downloaded successfully. Thankyou

[jayzzhou-thu]

Hi, the download link provided by the authors of FEVER is unavailable now. You can go to the original repo (https://github.com/sheffieldnlp/naacl2018-fever) and try to build the fever.db file by yourself according to its README.

That is to run bash scripts/download-raw-wiki.sh bash scripts/process-wiki.sh after you clone the repo.

The process takes time and the fever.db file is about 9GB. Hope this will help you.

[amanagr6697]

I also found an alternative way, i.e fever.ai/fever/fever.db and it's file size is approx 5 GB, do you think this would also work like intended? Thanks for the early reply.

[jayzzhou-thu]

I also found an alternative way, i.e fever.ai/fever/fever.db and it's file size is approx 5 GB, do you think this would also work like intended? Thanks for the early reply.

Hi, I am not quite sure because I have not used it. It may be used for other tasks such as FEVER 2.0. To reproduce the results in this repo, I think it is better to rebuild the file from scratch.

[amanagr6697]

Unable to run bash scripts/process-wiki.sh file In build_tfidf.py there is this line 'from drqascripts.retriever.build_tfidf import ' which isn't able to resolve so I cloned this repo: https://github.com/j6mes/DrQA and changed it to from DrQA.scripts.retriever.build_tfidf import to make it work but then it is unable to parse TfIdfBuilder corresponding to the line 'tfidf_builder = TfIdfBuilder(args,'sqlite', {'db_path': args.db_path})'

Would be grateful for a solution.

[jayzzhou-thu]

https://github.com/j6mes/DrQA

I guess you can try this: git+git://github.com/j6mes/drqa@parallel or just clone the repo and change to the parallel branch manually to install it. git checkout parallel

[amanagr6697]

Sir, so I did everything and got the files... Now, ig the model would be corresponding to best.pth.tar file in outputs folder, how to use that data.pkl file to predict accept/refuted by passing a sentence... Can you please provide info on this... This would be my last query, thanks a lot. Thanks!!!

[amanagr6697]

Or tell me, where is the model stored and how can I use it to predict if it's not data.pkl in outputs best path folder.

[WhynotLei]

我还找到了另一种方法，即 fever.ai/fever/fever.db，它的文件大小约为 5 GB，您认为这也能按预期工作吗？ 感谢您的早期回复。

Hi, how do you get fever.ai/fever/fever.db? Can you tell me the method? Thank you very much.

[amanagr6697]

我还找到了另一种方法，即 fever.ai/fever/fever.db，它的文件大小约为 5 GB，您认为这也能按预期工作吗？ 感谢您的早期回复。

Hi, how do you get fever.ai/fever/fever.db? Can you tell me the method? Thank you very much.

Go by the method they provided, fever.db would be generated at the designated folder, and then you can proceed normally...