Ray AIR Configurations#

See also

See this Ray Train configuration user guide for more details.

air.RunConfig([name, local_dir, callbacks, ...])

Runtime configuration for training and tuning runs.

air.ScalingConfig([trainer_resources, ...])

Configuration for scaling training.

air.DatasetConfig([fit, split, required, ...])

Configuration for ingest of a single Dataset.

air.CheckpointConfig([num_to_keep, ...])

Configurable parameters for defining the checkpointing strategy.

air.FailureConfig([max_failures, fail_fast])

Configuration related to failure handling of each training/tuning run.

tune.TuneConfig([mode, metric, search_alg, ...])

Tune specific configs.

tune.syncer.SyncConfig([upload_dir, syncer, ...])

Configuration object for Tune syncing.