Custom Script Changes Not Reflected During Execution in NVFlare

[falibabaei]

I'm using a custom script for tf/fedopt_ctl with the example in my_new_home/NVFlare/examples/getting_started/tf. I commented out line 127 of the code in fedopt_ctl_10 , which should cause an error since num_trainable_weights is not defined, but strangely, I don't get any error and the execution continues.

To Reproduce
Steps to reproduce the behavior:

1. Copy fed_ctl_10.py.
2. In line 128 of tf_fl_script_runner_cifar10.py, change from fedopt_ctl_10 import FedOpt.
3. Execute the code.
4. Modify fed_ctl_10.py and commented out line 127 of the code.
5. Execute the code again.

Expected behavior
get the error num_trainable_weights is not defined

Desktop (please complete the following information):

• ubuntu 22.04
• Python 3.9.16
• NVFlare version is 2.5.1

Here are the custom code for fedopt and the logs before and after modification
fedopt_ctl_10.txt
fedopt_ctl_log_before_modification.txt
fedopt_ctl_log_after_modification.txt

[yhwen]

@falibabaei The root cause of this issue is because of the wrong indentation in your fedopt_ctl_10.py. The "def update_model(self, global_model: FLModel, aggr_result: FLModel):" has an extra indentation which makes this method unreachable. So your modified codes were not got executed.

[falibabaei]

ah it happend during the copy. Sorry