Skip to content

Latest commit

 

History

History
8 lines (7 loc) · 673 Bytes

File metadata and controls

8 lines (7 loc) · 673 Bytes

This folder contains ipython notebooks used for evaluation. For some notebooks, its purpose is listed below.

  • VisualizeTrainingCurves.ipynb
    • This ipython notebook plots the training curves
  • WikicorpVsFeatures.ipynb
    • This ipython notebook builds the plot that compares the difference between the held-out perplexity obtained with and without finopt as a function of the number of features in the dataset
  • JacobianLatentSpace.ipynb
    • Comparing the latent dimensions utilization with the singular values of the Jacobian matrix