About the revised version
#1
by
Wangyh
- opened
Thanks for making this revised version. I was wondering what's the main difference between the old version and revised version. And what' s the motivation to make this new version?
Hi, thanks for your comment. Indeed we added the details of the updated datasets but we were missing a summary of the main differences.
We have now updated the dataset descriptions to include the main updates to the older version. I am adding them below for clarity:
Keypoints about the updated datasets:
- (1) it maintains similar tasks, number of classes and train and test set sizes as the previous benchmark
- (2) all the data is now from human samples and from recent and high-quality datasets (see descriptions of each dataset source in the README)
- (3) randomly generated negative samples have been replaced by existing chunks of genomes not containing the respective elements
- (4) all datasets have now proper chromosome held-out test sets.
Let us know if you have more questions. Thanks