Save and Load model

[Badal-Shrestha]

I used pickle to save and load the model. The pickle file has been saved but I am unable to load the pickle model due to the ELM attribute error not defined. I even import the ELM module while loading pickle model it still throws the same error.

Is there any best way to load and save this model?

[Blimpyway]

The class is so simple that I think you can save all init parameters including the self._w, self_beta and self._bias which are easy to serialize numpy arrays.

Furthermore, you can spare a lot of data if instead of the full _w and _bias you use and save a random seed to generate them and save in full only the _beta matrix (last layer) as it is the only one computed by fit().

In the mnist example the _beta is almost 1/80 the size of the entire model. beta size is 10x512 floats vs (784+1)x512 size for W and bias arrays.

PS sorry bias is all zero, you may as well skip it.