chrsunwil
commented
6 years ago @gwaygenomics
- I added docstrings to all files and function headers.
- I added the labels to the figures as per your link, work's reasonably well, but I had to put it on top, rather than the side of the image.
- I changed the order of imports to what is in the document you linked.
- I changed to use the
os.path.joinand Python3 string concatenation. - I added a parameter for number of visualizations to be shown
- I used a linter to have proper formatting (I think)
I wrote some comments in response throughout your review.
Made the code much more general. Now uses the main_file.py to call functions from other files. Each dataset gets it's own file which implements the abstract class form DataSetInfoAbstractClass. This will ensure that each data set has a method for loading and visualizing. These get called in both the main_file.py and the shared_vae_class.py. The model is now also representing in a .png file saved into an output folder. The visualizations are also saved into this folder. I also made the ICVL and MNIST data into pickle files to speed up the process of loading in the data.
Adding new datasets should be a much easier process now as it will mainly involved adding a new file which maintains the invariants of the abstract class.