workflow.ignite package

workflow.ignite.trainer(train_batch, evaluate_batch, evaluate_data_loaders, metrics, optimizers)

Create standard trainer with evaluators.

train_batchfunction

function that trains on given batch

evaluate_batchfunction

function that evaluates a given batch

evaluate_data_loaders: list

data loaders that yield batches to evaluate on

metricsdict

dict with one dict each for ‘train’ and evaluate data loader. Wrap a metric with trainer.Progress to show in progress bar.

optimizersdict

dict with optimizers for logging

tuple

trainer engine list of evaluator engines tensorboard logger

workflow.ignite.evaluator(evaluate_batch, description, metrics, tensorboard_logger)

workflow.ignite.worker_init(seed, trainer, worker_id)

Subpackages

• workflow.ignite.decorators package
• workflow.ignite.handlers package
• workflow.ignite.handlers.learning_rate package
• workflow.ignite.metrics package