IBM / federated-learning-lib

A library for federated learning (a distributed machine learning process) in an enterprise environment.
Other
500 stars 137 forks source link

Fedprox mode update issue #77

Closed jingoohan closed 3 years ago

jingoohan commented 3 years ago

Hi, I am using Fedprox and testing model update at aggreator side. I added 'model' property to 'examples/configs/fedprox/keras/config_agg.yml' as below.

Screen Shot 2021-08-24 at 7 18 55 PM

In this case, 'start_global_training()' in 'iter_avg_fusion_handler.py' calls the following code

Screen Shot 2021-08-24 at 7 21 42 PM

But, it makes following error after typing 'TRAIN' at aggregator side.

Screen Shot 2021-08-24 at 7 23 30 PM

Of course, In FedAvg case, it works properly without any issues. Can you check this issue and let me know how to fix this issue? Thank you.

Yi-Zoey commented 3 years ago

Hi @jingoohan, I think the error is caused by the fact that the model is not properly initialized. You can find an example of how to initialize the model in here.

jingoohan commented 3 years ago

@Yi-Zoey I really appreciate your answer. I just modified source code as you guided. I checked that the issue was solved. Thank you.