Loading
Cookies help us deliver our services. By using our services, you agree to our use of cookies. Learn more

crowdAI is shutting down - please read our blog post for more information

League of Nations Archives Digitization Challenge

Help us share the archives of the League of Nations, a vital part of world history


Completed
469
Submissions
135
Participants
7417
Views

Binary label does not fit the data

Posted by ViktorF over 1 year ago

Check training sample: train/en/0a0b9dcefa8339b0a1bc7e7e79bcecc3.jpg

This is equally English and French, so 0.5, 0.5 would be a perfect prediction.

But it was categorized as English in the training set.

How to resolve these samples?

Does English have precedence over non-English?

Posted by nshreyasvi  over 1 year ago |  Quote

There are some errors in the dataset, this one should qualify for bilingual. The dataset was prepared using crowdsourcing so there are some errors. Sorry for that.