Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Data found under https://github.com/Santosh-Gupta/datasets is not 700,000 question/answer pairs #15

Open
EphChem opened this issue Aug 24, 2019 · 13 comments

Comments

@EphChem
Copy link

EphChem commented Aug 24, 2019

Hello!
There are 2 links under https://github.com/Santosh-Gupta/datasets:

1 - https://drive.google.com/drive/folders/1PymmjbrgfOIs-HJ7oBmjZKH8j4rYsGZj
2 - https://drive.google.com/drive/folders/1kYD57uStDd4kXyb3JOYCTQd92Al6Il4K

However, there are duplicates in both links; for example each of the AskDocs.csv and icliniqQAs.csv is found in both links. Therefore, when I import all the non-duplicate data, I only see about 200,000 QA pairs, not the 700,000 which your repo mentions. Is the rest of the data somewhere else. Please kindly let me know how to import the entire 700,000 QA Dataset.

Thank you!

@Santosh-Gupta
Copy link
Member

How much data points are in the askdocs , healthtap , and webmd csvs? Each of those should contain several hundred thousand data points each. @ash3n @JayYip Are there updated versions of those files?

@Santosh-Gupta
Copy link
Member

In the meanwhile, here's a link to the binary tfrecords files

https://drive.google.com/drive/u/8/folders/1wRc1jtl5Q0objpfualNFwpg4H575tmks

@EphChem
Copy link
Author

EphChem commented Aug 25, 2019

Number of rows in healthtap, webmd, AskDocs is about 137000, 46200, 78000, respectively. I do not see a few hundred thousand data points per dataset as you mention.
Thanks for the tfrecords but not sure how helpful this will be for my work.

@Santosh-Gupta
Copy link
Member

@ash3n @JayYip I'm going through the chats regarding the data. Unless I missed something, I believe that is the total amount of data we have, ~300,000 question answer pairs, not 700,000. Health tap, askdocs, and webmd are the most significant datasets we have. All the other datasets have very low numbers compared to those three.

@JayYip
Copy link
Collaborator

JayYip commented Aug 26, 2019

I'm not very familiar with the size of data since I'm not the one who collected and parsed them.

@Santosh-Gupta
Copy link
Member

I'll wait until @ash3n comments. If my analysis is correct, we should update the Read.me

@Better-Boy
Copy link

Better-Boy commented Nov 3, 2019

#15 (comment) - link expired
https://drive.google.com/drive/folders/1PymmjbrgfOIs-HJ7oBmjZKH8j4rYsGZj - link expired

Please provide working links

@Santosh-Gupta
Copy link
Member

@JayYip that's your folder

@JayYip
Copy link
Collaborator

JayYip commented Nov 4, 2019

Link restored. Please check.

@Better-Boy
Copy link

@Santosh-Gupta
Copy link
Member

also @JayYip

@JayYip
Copy link
Collaborator

JayYip commented Nov 7, 2019

@Better-Boy restored. I'm running out of google drive space. @Santosh-Gupta @Better-Boy I wonder if you guys can help to store the data and start a PR to update links in README? Thanks.

@Santosh-Gupta
Copy link
Member

Sure, lets chat over the slack

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

4 participants