Fan Lai

There might be some misleading variable name. In FedScale, each executor drives the execution of its client. The executor **ping**(pull) the aggregator for the next steps, which may **receive** CLIENT_TRAIN...

Can not reproduce and it is Pytorch issue.

Hi, ahmedcs, Sorry for the late reply. This is very likely due to the insufficient number of clients in this dataset, so please try a larger dataset. We have fixed...

Great! Thanks a lot! Actually, Amber pushed a similar idea yesterday #173, which of course needs more efforts. We will work on this once we have more bandwidth.

Thanks a lot for your contribution! Let's try to run an e2e evaluation to validate the learning curve and Redis overhead first.

Thanks for your contribution. Can you please - remove all prints or replace them with logging.info - summarize the results of the overhead analysis We really appreciate your help.

Thank you for this feedback. Can you @dywsjtu please work on this asap, which may involve code cleanup? Thanks!

Hi kashprime, Thank you for trying FedScale. You can refer [here](https://github.com/SymbioticLab/FedScale/blob/master/examples/tensorflow_engine/tf_client.py#L24-L38) for a quick workaround too, which transforms PyTorch to TF tensors.

Hi @kashprime. We have some updates to share with you: we are actively collaborating with LinkedIn to add TF support. We will merge this #194 soon to support TF.

https://github.com/SymbioticLab/FedScale/blob/master/benchmark/dataset/download.sh#L572