Zachary Charles comments

Results 35 comments of


                                            Zachary Charles

How interpret results

@aynesss It looks like it is training, since the loss is decreasing. It's worth noting that a round is only a small model change, so I wouldn't expect the accuracy...

Is it possible to find Precision and Recall of multi-class classification using default metrics?

Hi @Nawrin14. This is certainly suspicious. Can you give a bit more detail? In particular, as part of our bug filing system we encourage you to submit the following: *...

custom_federated_algorithms_2.ipynb - use of possibly outdated version of tff.framework.local_executor_factory function in the tutorial

Hi @Scolpe. Thanks for the detailed information. As for your question, the tutorials are generally updated to reflect the latest version of TFF (in your case, I believe 0.23.0, though...

custom_federated_algorithms_2.ipynb - use of possibly outdated version of tff.framework.local_executor_factory function in the tutorial

Quick follow-up: The tutorial appears to be working at the latest github commit (ab932a8633f0ea4c9c95f46d7e06a58924560bae for posterity). If you run into issues of running it at the latest release, please let...

Asynchronous Federated Learning

There are not currently any asynchronous execution contexts available through TFF. That being said, if your interest is purely in simulations, then you could certainly simulate asynchronous client computations. For...

Asynchronous Federated Learning

Closing due to lack of activity.

Pretrained model: Non-trainable weights seems to be updated during FL training

Hi @mirkohin1991. Can you give more details on the ResNet-50 layers? In particular, I suspect that BatchNorm may be causing problems. It's worth noting that BatchNorm will exhibit much different...

How to change the total number of users in federated learning experiments? (ex) https://github.com/tensorflow/federated/blob/main/tensorflow_federated/python/simulation/datasets/emnist.py)

Hi @Yeojoon. One easy way to do this would be to subsample client IDs from EMNIST. This gives you a smaller total number of clients, but also reduces the total...

How to change the total number of users in federated learning experiments? (ex) https://github.com/tensorflow/federated/blob/main/tensorflow_federated/python/simulation/datasets/emnist.py)

If neither of these solutions are exactly what you're looking for, could you add some details about what kind of feature you're looking for?

How to change the total number of users in federated learning experiments? (ex) https://github.com/tensorflow/federated/blob/main/tensorflow_federated/python/simulation/datasets/emnist.py)

Could you provide a bit more detail here? How were you hoping to do this "re-partitioning", where the number of examples is fixed and the number of clients varies? Note...