[Q&A] - Global and Local model for xgboost histogram-based

[ramiro350]

Python version (python3 -V)

3.10.12

NVFlare version (python3 -m pip list | grep "nvflare")

2.4.0

NVFlare branch (if running examples, please use the branch that corresponds to the NVFlare version, git branch)

main

Operating system

Ubuntu 22.04

Have you successfully run any of the following examples?

• hello-numpy-sag with simulator
• hello-pt with simulator
• hello-numpy-sag with POC
• hello-pt with POC

Please describe your question

Hello guys, i've been running the xgboost histogram-based example(this one https://github.com/NVIDIA/NVFlare/tree/main/examples/advanced/xgboost/histogram-based) i'm running it in real world, one server and one client, i've launched on the admin cmd and everything went well i believe, i was checking in the client and in the job directory i've found the test.model which i believe to be the local model, my question is, is this actually the local model ? and where is the global model stored in the server ?

[YuanTingHsieh]

@ramiro350 thanks for the interest!

For your question:

is this actually the local model ?
and where is the global model stored in the server ?

@ZiyueXu77 can you help answer these questions?
The local model was saved in here: https://github.com/NVIDIA/NVFlare/blob/2.4/nvflare/app_opt/xgboost/histogram_based/executor.py#L288
Is the global model the same as the local model we get in federated xgboost (the insecure version) after training is done?

[ZiyueXu77]

In this case (histogram-based) server only performs necessary communication during tree-building, so it does not hold any model. The outcome model is the same across all clients and will be stored locally.