Saver

A saver can save and restore variables and other saveable objects.

This class adds ops to save and restore variables to and from *checkpoints*. It also provides convenience methods to run these ops. Checkpoints are binary files in a proprietary format which map variable names to tensor values. The best way to examine the contents of a checkpoint is to load it using a Saver.

Savers can automatically number checkpoint filenames. This lets you keep multiple checkpoints at different steps while training a model. For example, you can number the checkpoint filenames with the training step number. To avoid filling up disks, savers manage checkpoint files automatically. For example, they can make sure to keep only the N most recent files, or one checkpoint for every N hours of training.

You may number checkpoint filenames by passing a value to the optional globalStep argument of the save method. For example:

// Using a slight abuse of notation for paths:
saver.save(session, "my-model", globalStep = 0) ==> filename: "my-model-0"
saver.save(session, "my-model", globalStep = 1000) ==> filename: "my-model-1000"

Also, optional arguments to the Saver constructor let you control the proliferation of checkpoint files on disk:

maxToKeep: The maximum number of recent checkpoint files to keep. As new files are created, older files are deleted. If 0, no checkpoints are deleted from the filesystem but only the last one is kept in the checkpoint file. Defaults to 5 (i.e., only the 5 most recent checkpoint files are kept).
keepCheckpointEveryNHours: In addition to keeping the most recent maxToKeep checkpoint files, you might want to keep one checkpoint file for every N hours of training. This can be useful if you want to later analyze how a model progressed during a long training session. For example, passing keepCheckpointEveryNHours = 2 ensures that you keep one checkpoint file for every 2 hours of training. The default value of 10000 hours effectively disables the feature. Note that you still have to call the save method every time you want to save the model. Passing these arguments to the constructor will not save variables automatically for you.

An example training program that saves regularly looks like this:

// Using a slight abuse of notation.
// Create a saver.
val saver = tf.Saver(variables)
// Launch the graph and train, saving the model every 1,000 steps.
for (step <- 0 to 1000000) {
  session.run(trainOp)
  if (step % 1000 == 0) {
    // Append the step number to the checkpoint name.
    saver.save(session, "my-model", globalStep = step)
  }
}

In addition to checkpoint files, savers keep a protocol buffer on disk with the list of recent checkpoints. This is used to manage numbered checkpoint files. The latestCheckpoint method makes it easy to discover the path to the most recent checkpoint. That protocol buffer is stored in a file next to the checkpoint files, with default name "checkpoint" (can be provided using the checkpointStateFilename argument of the save method).

Linear Supertypes

Serializable, AnyRef, Any

Value Members

final def !=(arg0: Any): Boolean

Definition Classes
AnyRef → Any
final def ##(): Int

Definition Classes
AnyRef → Any
final def ==(arg0: Any): Boolean

Definition Classes
AnyRef → Any
final def asInstanceOf[T0]: T0

Definition Classes
Any
def clone(): AnyRef

Attributes
protected[java.lang]
Definition Classes
AnyRef
Annotations
@throws( ... )
final def eq(arg0: AnyRef): Boolean

Definition Classes
AnyRef
def equals(arg0: Any): Boolean

Definition Classes
AnyRef → Any
def finalize(): Unit

Attributes
protected[java.lang]
Definition Classes
AnyRef
Annotations
@throws( classOf[java.lang.Throwable] )
final def getClass(): Class[_]

Definition Classes
AnyRef → Any
def hashCode(): Int

Definition Classes
AnyRef → Any
final def isInstanceOf[T0]: Boolean

Definition Classes
Any
def latestCheckpoints: Seq[Path]

Returns the sequence of the latest and not-yet-deleted checkpoint filenames, sorted from oldest to newest.
Returns the sequence of the latest and not-yet-deleted checkpoint filenames, sorted from oldest to newest. You can pass any of the returned values to restore.
final def ne(arg0: AnyRef): Boolean

Definition Classes
AnyRef
final def notify(): Unit

Definition Classes
AnyRef
final def notifyAll(): Unit

Definition Classes
AnyRef
def recoverLastCheckpoints(checkpoints: Seq[Path]): Unit

Recovers the internal saver state (holding the last checkpoints) after a crash.
Recovers the internal saver state (holding the last checkpoints) after a crash.
This method searches for the checkpoints pointed to by checkpoints (which can also be glob patterns). If the files exist, the method uses their last modification time as the checkpoint stamp.
checkpoints
Sequence of checkpoint filenames (can also be glob patterns).
def restore(session: core.client.Session, savePath: Path): Unit

Restores previously saved saveables.
Restores previously saved saveables.
This method runs the ops responding for restoring variables. It requires a session in which the saver's graph was launched. The variables to restore do not have to have been initialized, as restoring is itself a way to initialize variables.
The savePath argument is typically a value previously returned from a save call, or a call to Saver.latestCheckpoint.
session
Session to use for restoring the variables.
savePath
Path to the checkpoint filename. If the saver is sharded, this is the prefix of the sharded checkpoint filenames.
def save(session: core.client.Session, savePath: Path, globalStep: Option[Int] = None, checkpointStateFilename: String = "checkpoint", metaGraphSuffix: String = "meta", writeMetaGraph: Boolean = true, writeCheckpointState: Boolean = true): Option[Path]

Saves the current value of the saveables this saver is responsible for.
Saves the current value of the saveables this saver is responsible for.
This method runs the ops responsible for saving variables. It requires a session in which the saver's graph was launched. The variables being saved must also have been initialized.
The method returns the path of the newly created checkpoint file. This path can be passed directly to restore.
session
Session to use for saving the variables.
savePath
Path to the checkpoint filename. If the saver is sharded, this is the prefix of the sharded checkpoint filenames
globalStep
If provided, the global step number is appended to savePath to create the checkpoint filename.
checkpointStateFilename
Optional name for the protocol buffer file that will contain / contains the list of the most recent checkpoint filenames. That file, kept in the same directory as the checkpoint files, and it is automatically managed by the saver to keep track of recent checkpoints.
metaGraphSuffix
Meta graph filename suffix.
writeMetaGraph
Boolean value indicating whether or not to write the graph meta information file.
writeCheckpointState
Boolean value indicating whether or not to write the checkpoint state file.
returns
Path of the newly created checkpoint file, if the save operation was successful; None, otherwise. If the saver is sharded, the filename ends with "-?????-of-nnnnn" where "nnnnn" is the number of shards created.
final def synchronized[T0](arg0: ⇒ T0): T0

Definition Classes
AnyRef
def toProto(exportScope: String = null): SaverDef

Alias for toSaverDef.
def toProto: SaverDef

Converts this object to its corresponding ProtoBuf object.
Converts this object to its corresponding ProtoBuf object.
returns
ProtoBuf object corresponding to this object.

Definition Classes
Saver → Serializable
def toSaverDef(exportScope: String = null): SaverDef

Constructs and returns a SaverDef object that represents this saver.
Constructs and returns a SaverDef object that represents this saver.
exportScope
Optional string specifying the name scope to remove. Only the ops within this name scope will be included in the resulting ProtoBuf object and the export scope will be stripped from their names to allow for easy import into new name scopes.
returns
Constructed SaverDef.
def toString(): String

Definition Classes
AnyRef → Any
final def wait(): Unit

Definition Classes
AnyRef
Annotations
@throws( ... )
final def wait(arg0: Long, arg1: Int): Unit

Definition Classes
AnyRef
Annotations
@throws( ... )
final def wait(arg0: Long): Unit

Definition Classes
AnyRef
Annotations
@throws( ... )
val writerVersion: WriterVersion

Related Docs: object Saver | package variables

class Saver extends Serializable

Value Members

final def !=(arg0: Any): Boolean

final def ##(): Int

final def ==(arg0: Any): Boolean

final def asInstanceOf[T0]: T0

def clone(): AnyRef

final def eq(arg0: AnyRef): Boolean

def equals(arg0: Any): Boolean

def finalize(): Unit

final def getClass(): Class[_]

def hashCode(): Int

final def isInstanceOf[T0]: Boolean

def latestCheckpoints: Seq[Path]

final def ne(arg0: AnyRef): Boolean

final def notify(): Unit

final def notifyAll(): Unit

def recoverLastCheckpoints(checkpoints: Seq[Path]): Unit

def restore(session: core.client.Session, savePath: Path): Unit

def save(session: core.client.Session, savePath: Path, globalStep: Option[Int] = None, checkpointStateFilename: String = "checkpoint", metaGraphSuffix: String = "meta", writeMetaGraph: Boolean = true, writeCheckpointState: Boolean = true): Option[Path]

final def synchronized[T0](arg0: ⇒ T0): T0

def toProto(exportScope: String = null): SaverDef

def toProto: SaverDef

def toSaverDef(exportScope: String = null): SaverDef

def toString(): String

final def wait(): Unit

final def wait(arg0: Long, arg1: Int): Unit

final def wait(arg0: Long): Unit

val writerVersion: WriterVersion

Inherited from Serializable

Inherited from AnyRef

Inherited from Any

Ungrouped