Dataloader support for TF

[kashprime]

Description

Hi Team, there is admittedly some overlap between this issue and a previous one. However I thought I would make a new one since the other is fairly old. I am looking for some data loader support in FedScale for TensorFlow. It seems the data classes as they stand are written using PyTorch in mind and I was wondering if anyone has any experience using a TensorFlow data set, particularly using .tfrecord files with a known schema.

Use case

I work in industry and am looking to use FedScale to run a simulation of a specific federated learning model on a specific set of hardware. All of our models are written using tf and keras.

[AmberLJC]

Hi kashprime, we had some tf data loader use cases (not public yet), we will discuss about it by this week and you may hear back soon.

[kashprime]

Awesome, thank you AmberLJC!

[fanlai0990]

Hi kashprime, Thank you for trying FedScale. You can refer here for a quick workaround too, which transforms PyTorch to TF tensors.

[fanlai0990]

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.