Mosharaf Chowdhury

Definitely, we should keep adding datasets. As for cross-silo, it should certainly be in the roadmap. Before that however, we need to fix simple things like data format #118

I agree that @ewenw's option 2 is more attractive. @fanlai0990 all these are also relevant to the client library architecture we were discussing recently.

@fanlai0990 can you please assign this to the team who were talking about shutdown command yesterday? It's broader than just shutdown but that can be a start.

We should look into hydra (https://hydra.cc/)

Great progress. Another TODO would be including appropriate tensorflow in the `enviornment.yml` file for installation. How do you want to go about it?

@AmberLJC @fanlai0990 I think this is what we were discussing earlier re:having a separate chekcin server that decouples devices checking in and selector picking them.

Installing torch through pip and then `import torch` in python 3.9 doesn't give any error. The same in the fedscale environment segfaults. There is a torch/python versioning problem in our...

Yes, I'm talking about Mac. > Installing torch through pip and then import torch in python 3.9 doesn't give any error. The same in the fedscale environment segfaults. There is...

FWIW, submitting jobs through works in the fedscale environment as well. Only running python in the interactive mode doesn't

I just looked up the trace files themselves. They seem to be in binary format, which isn't ideal for adding new traces easily. Why not keep the data in plain...