org.tensorflow

Class SessionFunction

java.lang.Object

org.tensorflow.SessionFunction

All Implemented Interfaces:

TensorFunction

public class SessionFunction
extends Object
implements TensorFunction

A callable function backed by a session. All calls of this function will be ran on the same session.

Does no resource management, the session and all returned tensors are the caller's responsibility.

Does not initialize the session, since it may be shared.

Constructor Summary

Constructors

Constructor and Description
SessionFunction(Signature signature, Session session)

Method Summary

Modifier and Type	Method and Description
Map<String,Tensor>	call(Map<String,Tensor> arguments) Invokes a function using the default eager session.
static SessionFunction	create(Signature signature, Session session)
void	save(String exportDir) Save this function using SavedModelBundle.
Session	session()
Signature	signature() Returns the signature of this function
SessionFunction	withNewSession(Session session) Get a new function with the same signature, but backed by a new session.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Methods inherited from interface org.tensorflow.TensorFunction

call, validateDescription

Constructor Detail

SessionFunction

public SessionFunction(Signature signature,
                       Session session)

Method Detail

create

public static SessionFunction create(Signature signature,
                                     Session session)

save

public void save(String exportDir)
          throws IOException

Save this function using SavedModelBundle.

This is identical to calling SavedModelBundle.exporter(exportDir).withFunction(this).export().

Parameters:

exportDir - the directory path containing a saved model.

Throws:

IOException - if saved model or variable state cannot be written on disk

signature

public Signature signature()

Description copied from interface: TensorFunction

Returns the signature of this function

Specified by:

signature in interface TensorFunction

session

public Session session()

withNewSession

public SessionFunction withNewSession(Session session)

Get a new function with the same signature, but backed by a new session.

Parameters:

session - the new backing session.

call

public Map<String,Tensor> call(Map<String,Tensor> arguments)

Description copied from interface: TensorFunction

Invokes a function using the default eager session.

Caller is responsible for closing all Tensors.

Specified by:

call in interface TensorFunction

Parameters:

arguments - list of tensors to pass in input to the function, mapped by their signature name

Returns:

output tensors resulting from the execution of the function, mapped by their signature name