Closed gartland98 closed 2 years ago
I think you mean the feml_core folder of Federated_single_model directory. For example, as you can see in the nn-dist-train/Federated_single_model/fedml_core/trainer/model_trainer.py, it is composed of abstract class.
The goal of this abstract class is to be compatible to any deep learning frameworks such as PyTorch, TensorFlow, Keras, MXNET, etc.
These are reason why I make a fedml_core. But you can change the structure in ease of use.
I'm curious what's the role of codes in core folder. Can you explain it?