org.platanios.tensorflow.api.learn.estimators
Creates a replica device setter, if required, to be used as a default device function.
Creates a replica device setter, if required, to be used as a default device function.
Estimators use a ReplicaDevicePlacer as a default device placer. It sets the distributed related arguments such as the number of parameter server replicas based on the provided run configuration.
Configuration.
Constructed replica device placer.
Creates a MonitoredSession to be used for training.
Creates a MonitoredSession to be used for training.
For a chief, this utility sets proper session initializers, savers, and restorers. It also creates hooks related to checkpoint and summary saving. For workers, this utility method sets the proper session creator which waits for the chief to initialize or restore the session. Please refer to MonitoredSession for more information.
NOTE: If you provide any summary saver or checkpoint saver hooks in hooks
or chiefOnlyHooks
, then the
checkpoint configuration in configuration
will be ignored for the chief and those hooks will be used instead.
Configuration to use for this session. Contains information related to the session configuration, the cluster configuration, etc.
Hooks to use while training.
Hooks to use for the chief. These will only be used if isChief
is true
.
Session scaffold used for gathering and/or building supportive ops. If not specified, a default one is created. The session scaffold is used to finalize the graph.
Created monitored session.