kaplan-meier-he example not working

[shammimore]

Describe the bug
kaplan-meier-he example gives an error.
fl_model.params send to server from client is empty
fl_model.params {}

Warning message:
2025-01-30 10:25:12,710 - InProcessClientAPI - INFO - Try to send local model back to peer
2025-01-30 10:25:12,908 - PTToNumpyParamsConverter - WARNING - 2 vars excluded as they were non-tensor type: ['hist_obs', 'hist_cen']

Error message:
File "kaplan-meier-he/src/kaplan_meier_wf.py", line 51, in aggr_hist
hist = fl_model.params["hist_obs"]
~~~~~~~~~~~~~~~^^^^^^^^^^^^
KeyError: 'hist_obs'

To Reproduce
Steps to reproduce the behavior:

1. Go to 'https://github.com/NVIDIA/NVFlare/tree/main/examples/advanced/kaplan-meier-he'
2. Run 'python km_job.py '

Expected behavior
A clear and concise description of what you expected to happen.

Screenshots

Desktop (please complete the following information):

• OS: Ubuntu 24.04.1 LTS
• Python Version [3.12.3]
• NVFlare Version [2.5.2]

Additional context
Add any other context about the problem here.

[ZiyueXu77]

could you try install nvflare from main branch?

pip uninstall nvflare
pip install -e . 

The error seems to be caused by the exchange_format, which is specified here
https://github.com/NVIDIA/NVFlare/blob/main/examples/advanced/kaplan-meier-he/km_job.py#L66

[shammimore]

Thanks! It works after installing nvflare from main branch.

Note:
There was another error 'unable to import torch'. It was quickly fixed by installing torch. Maybe it's good to add that to the requirement.txt.

Error message:

File "/home/smore/nvflare_trial/NVFlare/nvflare/job_config/script_runner.py", line 314, in init
super().init(
File "/home/smore/nvflare_trial/NVFlare/nvflare/job_config/script_runner.py", line 123, in init
raise ValueError("Using FrameworkType.PYTORCH, but unable to import torch")
ValueError: Using FrameworkType.PYTORCH, but unable to import torch

[ZiyueXu77]

Thanks for the update! I also realized the issue and created a PR (#3194) to specify the framework, torch is not necessary in this example.