composer.loggers#

Loggers to store metrics and checkpoints.

In Composer, algorithms and callbacks can make calls to the Logger, which then routes the calls to the appropriate LoggerDestination instances. The LoggerDestination does the actual logging, for example to a file, or Weights and Biases.

See the LoggerDestination documentation for an example of how to define a custom logger and use it when training.

Classes

CometMLLogger

Log to Comet.

ConsoleLogger

Log metrics to the console.

FileLogger

Log data to a file.

InMemoryLogger

Logs metrics to dictionary objects that persist in memory throughout training.

LoggerDestination

Base class for logger destination.

MLFlowLogger

Log to MLFlow.

ProgressBarLogger

Log metrics to the console and optionally show a progress bar.

RemoteUploaderDownloader

Logger destination that uploads (downloads) files to (from) a remote backend.

TensorboardLogger

Log to Tensorboard.

WandBLogger

Log to Weights and Biases.