About the datasets for WinPrediction

I had analyzed the dataset and found the following:

  • val dataset contains 9 examples with an incorrect label (189, 198, 866, 876, 1326, 1392, 2198, 2942, 3453).
  • train dataset contains 111 examples with an incorrect label(480, 777, 863, …),
    2 examples with a “clear board” (39154, 39394) and 5 examples with a “black board”( black pawns are not visible)(15535, 16587, 32283, 34999, 38336).

Maybe the test dataset also has incorrect labels and we can’t get 100% score?

@victorkras2008: Thanks for pointing these out. We are on it. And indeed, the some of the labels should also be wrong in the test set. We are working on correcting the affected data points, and then re-evaluating all the submissions.
@ayushivani is on it :tada: !

@victorkras2008 We have found the mistake and have updated the dataset including the test set. You can find the information in this post.