The code requires Python >= 3.6 and PyTorch >= 1.2.0. To install the other dependencies: pip3 install -r requirements.txt.

This code uses the CIFAR10, CIFAR100, mini-ImageNet datasets.

The CIFAR10, CIFAR100 AND MNIST datasets are downloaded automatically by the torchvision package. The mini-ImageNet dataset can be downloaded from this link: https://mega.nz/#!rx0wGQyS!96sFlAr6yyv-9QQPCm5OBFbOm4XSD0t-HlmGaT5GaiE.

FedCACS is run using a command of the following form:

`python main.py --alg fedcacs --dataset [dataset] --num_users [num_users] --model [model] --model [model] --shard_per_user [shard_per_user] --epochs [model] Explanation of parameters:

• alg : algorithm to run, may be fedcacs, fedrep (FedRep), fedavg, fedper (FedPer), or lg (LG-FedAvg)
• dataset : dataset, may be cifar10, cifar100, mini_imagenet
• num_users : number of users
• model : for the CIFAR datasets, we use cnn, for the MNIST datasets, we use mlp
• num_classes : total number of classes
• shard_per_user : number of classes per user (specific to CIFAR datasets and MNIST)
• frac : fraction of participating users in each round (for all experiments we use 0.1)
• local_bs : batch size used locally by each user
• lr : learning rate
• epochs : total number of communication rounds
• local_ep : total number of local epochs
• rep_eps : number of local epochs to execute for the representation (specific to FedCACS and FedRep)
• gpu : GPU ID

A full list of configuration parameters and their descriptions are given in utils/options.py.