Custom subplot with matplotlib

[stared]

Right now (0.3.4+) there is an option to add extra plots, with extra_plots=[func1, ...] where func1 is a function func1(logs) drawing a subplot (vide a2546b7ca9ae4111835017b227379ad8b35f891a). See e.g. https://github.com/stared/thinking-in-tensors-writing-in-pytorch/blob/master/5%20Nonlinear%20regression.ipynb for an example

The idea is to be able to add any plots. Though to make it more general, some things require initialization, so it is reasonable to create objects (e.g. binding data, or models, or prediction functions).

I want to create subplots in the spirit of TensorFlow Playground and Scikit-learn classifier comparison among other examples. Open to feedback on the class structure, as well as other custom plots.

[stared]

Some other ideas for plots to be implemented:

• https://towardsdatascience.com/visualizing-intermediate-activation-in-convolutional-neural-networks-with-keras-260b36d60d0