Basic variable class Arguments: lifecycle: specific lifecycle of the attribute

The data exchanged during an FL course are abstracted as 'Message' in FederatedScope. A message object includes: msg_type: The ty

An extended configuration system based on [yacs]( \ https://github.com/rbgirshick/yacs). \ The two-level tree structure consists of sever

``ClientData`` converts split data to ``DataLoader``. Args: loader: ``Dataloader`` class or data dict which have been built

Multilayer Perceptron

r"""GraphSAGE model from the "Inductive Representation Learning on Large Graphs" paper, in NeurIPS'17 Source: https://github.com/pyg-team

Wrap raw data into pytorch Dataset Arguments: dataset (dict): raw data dictionary contains "x" and "y"

Create a new_dict to ensure we can access the dictionary with one bracket only e.g., dict[key1][key2][key3] --> dict[key1.key2.key3]

AdditiveSecretSharing class, which can split a number into frames and recover it by summing up

r""" Args: root (string): Root directory where the dataset should be saved. FL (Bool): Federated setting, `0` for DBLP, `1` for FL

To automatic allocate the gpu, which returns the gpu with the largest free memory rate, unless the specified_device has been set up When

Provide the monitoring functionalities such as formatting the \ evaluation results into diverse metrics. \ Besides the prediction related

r"""GAT model from the "Graph Attention Networks" paper, in ICLR'18 Arguments: in_channels (int): dimension of input. out_channel

r""" GCN model from the "Semi-supervised Classification with Graph Convolutional Networks" paper, in ICLR'17. Arguments: in_channels

r"""Graph Isomorphism Network model from the "How Powerful are Graph Neural Networks?" paper, in ICLR'19 Arguments: in_channels (int)

r"""GPR-GNN model from the "Adaptive Universal Generalized PageRank Graph Neural Network" paper, in ICLR'21 Arguments: in_channels (i

Bleu scorer.

Represents a discrete search space, e.g., {'abc', 'ijk', 'xyz'}.

Track the history of metric (e.g., validation loss), \ check whether should stop (training) process if the metric doesn't \ improve after

This class is used to construct an FL course, which includes `_set_up` and `run`. Arguments: data: The data used in the FL cours

r"""Implementation of ``CanonicalAtomFeaturizer`` and ``CanonicalBondFeaturizer`` in DGL. \ Source: https://lifesci.dgl.ai/_modules/dgllife/ut

Convert data list to torch Dataset to save memory usage.

Split Data into small data via louvain algorithm. Args: client_num (int): Split data into ``client_num`` of pieces. delta (i

Initializes the metric functions for the monitor. Use ``eval(ctx)`` \ to get evaluation results. Args: eval_metric: set of metri

r""" Arguments: root (string): Root directory where the dataset should be saved. name (string): The name of the dataset (:obj:`"ep

The communicator used for standalone mode

The communicator used for standalone mode with multigpu

An object to compute the acquisition functions.

VerticalDataSampler is used to sample a subset from data Arguments: data(dict): data replace (bool): Whether the sample is w

r"""Create synthetic dataset based on the contextual SBM from the paper: https://arxiv.org/pdf/1807.09596.pdf Use the similar class as InMemo

The implementation of gRPCCommManager is referred to the tutorial on https://grpc.io/docs/languages/python/

r"""Analyzer for raw graph and split subgraphs. Arguments: raw_data (PyG.data): raw graph. split_data (list): the list for subgra

The aggregator used in asynchronous training, which discounts the \ staled model updates

y = {1, 0} L = -yln(p)-(1-y)ln(1-p)

The implementation of BLEU refer to 'Bleu: a method for automatic evaluation of machine translation.' [Papineni et al., 2002] (https://ac

Record and pass variables among different hook functions. Arguments: model: training model cfg: config data (dict):

Represents a continuous search space, e.g., in the range [0.001, 0.1].

Logging colored formatter, adapted from https://stackoverflow.com/a/56944256/3638629

Implementation of the paper "Deep Leakage from Gradients": https://papers.nips.cc/paper/2019/file/ \ 60a6c4002cc7b29142def8871531281a-Paper.pd

Data collator used denoising language modeling task in BART. The implementation is based on https://github.com/pytorch/fairseq/blob/ 1bba7

``DummyDataTranslator`` convert datadict to ``StandaloneDataDict``. \ Compared to ``core.data.base_translator.BaseDataTranslator``, it do not

Dummy regularizer that only returns zero.

The FedEM implementation, "Federated Multi-Task Learning under a \ Mixture of Distributions (NeurIPS 2021)" \ based on the Algorithm 1 in

Implementation of FedOpt refer to `Adaptive Federated Optimization` \ [Reddi et al., 2021](https://openreview.net/forum?id=LkFG3lB13U5)

This splitter split according to what FedSAM provides Args: client_num: the dataset will be split into ``client_num`` pieces

The implementation of GAN based class representative attack. https://dl.acm.org/doi/abs/10.1145/3133956.3134012 References: Hit

r"""GNN model with pre-linear layer, pooling layer and output layer for graph classification tasks. Arguments: in_channels (int):

propagation class for GPR_GNN source: https://github.com/jianhao2016/GPRGNN/blob/master/src/GNN_models.py