search

Aifred-Health / VulcanAI

A high level deep learning framework for quickly prototyping networks with added tools in data visualisation, model interpretability and performance metrics
https://www.aifredhealth.com
GNU Affero General Public License v3.0
17 stars 8 forks source link

Load Model Records for Display in Visualization #109

Open joemehltretter opened 5 years ago

joemehltretter commented 5 years ago

In visualization.py we need to add functionality that can load a model and collect it's records to display in display_record. Basenetwork has a load_model function; however, it can't be used in visualization.py. Therefore, we need to discuss a way to load models in visualization.

joemehltretter commented 5 years ago

Here is the first option where we just load a graphic and use display_record to plot it.

def load_network_record_visualization(load_path):
        with open(load_path) as in_file:
            record = pickle.load(in_file)
            display_record(record = record)
joemehltretter commented 5 years ago

Here is the second option we thought of, but doesn't work due to circular dependency issues:

def load_network_record_visualization(load_path):
   model = basenetwork.load_model(load_path)
   display_record(record=model.record)

I think the benefit of this functionality is that we are pulling straight from the model.

nirtiac commented 5 years ago

@joemehltretter is this still an issue? I don't think it was addressed in remote server plotting but I'm not sure. If it wasn't I can take it on.

nirtiac commented 5 years ago

@joemehltretter bump. I think this has been addressed but want your feedback.