Module distributed_training 

Distributed training support for the array protocol.

This module provides utilities for distributed training of neural networks using the array protocol. It includes data-parallel and model-parallel training strategies, parameter synchronization, and distributed optimization.

Structs

CommunicationChannel

A wrapper type that makes Box<dyn DistributedCommunication> cloneable

DistributedDataset

Distributed Dataset that partitions data across workers.

DistributedNode

A node in a distributed training cluster.

DistributedTrainer

Distributed Trainer for handling distributed training.

DistributedTrainingConfig

Configuration for distributed training.

DistributedTrainingFactory

Factory for creating distributed training components.

MockDistributedCommunication

A mock implementation of distributed communication for testing.

ParameterSyncCallback

Callback for synchronizing parameters between workers.

Enums

DistributedStrategy

Distributed training strategy.

Traits

DistributedCommunication

Trait for distributed communication between nodes.