Source code for fedavg (or fedopt or fedprox) #1727
-
|
Hi, where I can see the code calculating global-model? https://github.com/NVIDIA/NVFlare/blob/2.3/examples/advanced/cifar10/pt/learners/cifar10_learner.py#L396-L409 looks aggregating local model to get global model, but where is the specific (source) code for averaging local weights like |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment
-
|
Please check https://github.com/NVIDIA/NVFlare/blob/dev/nvflare/app_common/aggregators for the aggregation logic on the server side. https://github.com/NVIDIA/NVFlare/blob/2.3/examples/advanced/cifar10/pt/learners/cifar10_learner.py#L396-L409 this is NOT aggregation, this part of code just load the model from server side |
Beta Was this translation helpful? Give feedback.
Please check https://github.com/NVIDIA/NVFlare/blob/dev/nvflare/app_common/aggregators for the aggregation logic on the server side.
https://github.com/NVIDIA/NVFlare/blob/2.3/examples/advanced/cifar10/pt/learners/cifar10_learner.py#L396-L409 this is NOT aggregation, this part of code just load the model from server side