google-deepmind / google-deepmind/multi_object_datasets
Number of images in CLEVR and Multi_dSprites_on_gray datasets are different from IODINE
Open
- Dominant language
- Python
- Stars
- 290
- Forks
- 27
- PR merge metrics
- No merged PRs in 30d
Description
In the paper `Multi-Object Representation Learning with Iterative Variational Inference Klaus` (**IODINE**)
1. The paper claims there are 70000 images for CLEVR but there are 100000+ images in its tfrecords
2. The paper claims there are 60000 images for Multi dSprites but there are much more than that.
How to choose the training and test dataset in the same way as this paper?
Contributor guide
Assessment
This issue has not been assessed yet.