Client-side model parameters NOT updated during the FedEx training

[krishnakanthnakka]

In

FederatedScope/federatedscope/autotune/fedex/client.py

Line 30 in 6dfe8d4

self.trainer.cfg = self._cfg

the sampled hyperparams at the client are updated to self.trainer.cfg but they don't seem to take actual effect in client-side trainer.

For example, the dropout in the model (at https://github.com/alibaba/FederatedScope/blob/6dfe8d4474451c07cd7d69690a2df33a1f455f19/federatedscope/cv/model/cnn.py#L45C15-L45C22) is maintained by self.dropout but this variable is never updated during the FedEx training but instead model_config.dropout is updated which is never reused except during the initialization stage.

In other words, I think, similar to trainer.update() at

FederatedScope/federatedscope/autotune/fedex/client.py

Line 46 in 6dfe8d4

self.trainer.update(model_params)

we should have some function to update the dynamic hyperparameters in torch_trainer.py.

Please correct me if I missed something.

Edit:

The optimizer related hyperparams are indeed updated at

FederatedScope/federatedscope/core/trainers/torch_trainer.py

Line 199 in 6dfe8d4

ctx.optimizer = get_optimizer(ctx.model,

but kindly check if model related hyperparams (dropout) updated somewhere...

[rayrayraykk]

Thanks for the suggestion, we noticed that if the cfg changes, the correct implementation is that the model (and the corresponding optimizer) should be re-instantiated. We will fix it ASAP.