LG-FedAvg

LG-FedAvg copied to clipboard

Failed to converge when changing num_users and frac

1

# Description When I change the num_user to 10 and frac to 0.3 **with --iid**, which means each epoch there are 3 client been choosen, I find the model become...

jkup64

In main_mtl.py, the omega is not updated after initialization, with is different from Algorithm 1 line 11 in [MOCHA](https://arxiv.org/abs/1705.10467) I also print the omega during training, but its value does...

TsingZ0

CNNCifar.weight_keys

4

Nice Work! But in Net.py CNNCifar.weight_keys, why are fc layers ahead of conv? ```python self.weight_keys = [['fc1.weight', 'fc1.bias'], ['fc2.weight', 'fc2.bias'], ['fc3.weight', 'fc3.bias'], ['conv2.weight', 'conv2.bias'], ['conv1.weight', 'conv1.bias'], ] ``` If nothing...

KiteFlyKid

Wrong loss function?

1

I notice that the code uses CrossEntropyLoss for local training: https://github.com/pliang279/LG-FedAvg/blob/7af0568b2cae88922ebeacc021b1679815092f4e/models/Update.py#L28 And it accepts the log-probabilities as input: https://github.com/pliang279/LG-FedAvg/blob/7af0568b2cae88922ebeacc021b1679815092f4e/models/Update.py#L50 The output of CNN networks is also logsoftmax: https://github.com/pliang279/LG-FedAvg/blob/7af0568b2cae88922ebeacc021b1679815092f4e/models/Nets.py#L104 But according...

mengcz13

Hi, I can't see local representations learning in main_lg.py. It just records the best local acc and the best local model, then updates the global model.

6

Hi, I can't see local representations learning in main_lg.py. It just records the best local acc and the best local model, then updates the global model. Did I understand it...

wardseptember

Hi, thanks for your code. However, I noticed that there is no implementation of Eq. 1 in your paper. Eq. 1 is leveraged in the training of the local model...

FlyingRoastDuck

Bumps [torch](https://github.com/pytorch/pytorch) from 0.4.1 to 2.2.0. Release notes Sourced from torch's releases. PyTorch 2.2: FlashAttention-v2, AOTInductor PyTorch 2.2 Release Notes Highlights Backwards Incompatible Changes Deprecations New Features Improvements Bug fixes...

dependabot[bot]