inferno.trainers.callbacks.logging package

Submodules

inferno.trainers.callbacks.logging.base module

class inferno.trainers.callbacks.logging.base.Logger(log_directory=None)

Bases: inferno.trainers.callbacks.base.Callback

A special callback for logging.

Loggers are special because they’re required to be serializable, whereas other callbacks have no such guarantees. In this regard, they jointly handled by trainers and the callback engine.

log_directory

set_log_directory(log_directory)

inferno.trainers.callbacks.logging.tensorboard module

class inferno.trainers.callbacks.logging.tensorboard.TensorboardLogger(log_directory=None, log_scalars_every=None, log_images_every=None, send_image_at_batch_indices='all', send_image_at_channel_indices='all', send_volume_at_z_indices='mid')

Bases: inferno.trainers.callbacks.logging.base.Logger

Class to enable logging of training progress to Tensorboard.

Currently supports logging scalars and images.

end_of_training_iteration(**_)

end_of_validation_run(**_)

extract_images_from_batch(batch)

get_config()

log_histogram(tag, values, step, bins=1000)

Logs the histogram of a list/vector of values.

log_image_or_volume_batch(tag, batch, step=None)

log_images(tag, images, step)

Logs a list of images.

log_images_every

log_images_now

log_object(tag, object_, allow_scalar_logging=True, allow_image_logging=True)

log_scalar(tag, value, step)

tag : basestring

Name of the scalar

value step : int

training iteration

log_scalars_every

log_scalars_now

observe_state(key, observe_while='training')

observe_states(keys, observe_while='training')

writer

Module contents

inferno.trainers.callbacks.logging.get_logger(name)