Palash Shah
Palash Shah
possibly distill a neural network using squeezenet and then retrain the model using the best hyperparameters.
right now, we have to put the logger at the top of every file in order to maintain global variable state. This should be fairly straightforward, but we need to...
look into this and how to make this work. possibly by integrating flask?
what if the user wants to visualize the deep learning process they took. We need to create some sort of graph visualizer.
include a pre-trained sentiment classifier workflow of some sort? not sure what this would look like but open to discussion.
store this under a docker folder, check out github.com/palashio/libra for an example docker setup
should we get into confusion matrices etc. maybe possible integration with vega-lite