search

Xiao-Chenguang / FedMind

Federated Learning framework for researcher on top of PyTorch.
https://github.com/Xiao-Chenguang/FedMind
MIT License
1 stars 0 forks source link

New FL develop schema in easy 3 steps. #68

Closed Xiao-Chenguang closed 1 week ago

Xiao-Chenguang commented 1 week ago

This PR introduces a new algorithm developing schema in barely 3 steps:

  1. add extra parameters to the dyn_args group, which will be passed to the client for local training.
  2. define your local training step with the parameters you added.
  3. define your global aggregation function with the extra parameters you want to aggregate.

In this case, you need not worry about anything except the algorithm logic. FedMind will take care of the parallel or serial simulation for you.