Transfer Learning

[Connect2Aditya]

Hi @HuguesTHOMAS
I want to implement KPConv in transfer learning mode. I want to use the weights from a pretrained model (trained on dataset1) and train it further on a new dataset (dataset2). It may be noted that there are only 2 classes in both datasets.

But I am getting a training behaviour as if I am training the model from the scratch (on dataset2). I am not sure if the weights are loaded correctly.

I have just provided the path (of trained snapshot) to restore_snap in line 95 of trainer.py. Is there a way to check if the weights are loaded correctly.

Please suggest any other changes I need to make in the code. I have also referred to this issue #43