Visualizing memorization in RNNs

Visualizing the gradient magnitudes in context, can be a powerful tool to see when recurrent units use short-term or long-term memorization.

  1. Run npm install
  2. Train the models with make train. This will take a very long time :)
  3. Build the data for the article with make build
  4. Bundle the article JavaScript with make bundle
  5. Run the article server with make server

Alternatively, if you just want to render and show the article:

  1. npm install
  2. npm start