Is it possible to re-train the model within glasspy?

[sarah-allec]

I'm interested in doing transfer learning from the original GlassNet model to the prediction of other glass properties - is this possible within the current glasspy framework?

[drcassar]

Hi @sarah-allec. Sure, that would be a great use of GlassNet! GlassNet is based on a lightning module, which itself is based on pytorch. You can perhaps freeze the weights and bias of the "hidden_layers", delete the "output_layer", and then add new layers to train for your transfer learning. I am planning on doing a tutorial on this, but I am not sure when it will be ready.

[sarah-allec]

Thank you! I would greatly appreciate the tutorial but will attempt it on my own in the meantime.