tf.compat.v1.train.Saver

Saves and restores variables.

Migrate to TF2

tf.compat.v1.train.Saver is not supported for saving and restoring checkpoints in TF2. Please switch to tf.train.Checkpoint or tf.keras.Model.save_weights, which perform a more robust object-based saving.

How to Rewrite Checkpoints

Please rewrite your checkpoints immediately using the object-based checkpoint APIs.

You can load a name-based checkpoint written by tf.compat.v1.train.Saver using tf.train.Checkpoint.restore or tf.keras.Model.load_weights. However, you may have to change the names of the variables in your model to match the variable names in the name-based checkpoint, which can be viewed with tf.train.list_variables(path).

Another option is to create an assignment_map that maps the name of the variables in the name-based checkpoint to the variables in your model, eg:

{
    'sequential/dense/bias': model.variables[0],
    'sequential/dense/kernel': model.variables[1]
}

and use tf.compat.v1.train.init_from_checkpoint(path, assignment_map) to restore the name-based checkpoint.

After restoring, re-encode your checkpoint using tf.train.Checkpoint.save or tf.keras.Model.save_weights.

See the Checkpoint compatibility section of the migration guide for more details.

Checkpoint Management in TF2

Use tf.train.CheckpointManager to manage checkpoints in TF2. tf.train.CheckpointManager offers equivalent keep_checkpoint_every_n_hours and max_to_keep parameters.

To recover the latest checkpoint,

checkpoint = tf.train.Checkpoint(model)
manager = tf.train.CheckpointManager(checkpoint)
status = checkpoint.restore(manager.latest_checkpoint)

tf.train.CheckpointManager also writes a CheckpointState proto which contains the timestamp when each checkpoint was created.

Writing MetaGraphDefs in TF2

To replace, tf.compat.v1.train.Saver.save(write_meta_graph=True), use tf.saved_model.save to write the MetaGraphDef (which is contained in saved_model.pb).

Description

See Variables for an overview of variables, saving and restoring.

The Saver 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 with a provided counter. 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 keep only the N most recent files, or one checkpoint for every N hours of training.

You number checkpoint filenames by passing a value to the optional global_step argument to save():

saver.save(sess, 'my-model', global_step=0) ==> filename: 'my-model-0'
...
saver.save(sess, 'my-model', global_step=1000) ==> filename: 'my-model-1000'

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

  • max_to_keep indicates the maximum number of recent checkpoint files to keep. As new files are created, older files are deleted. If None or 0, no checkpoints are deleted from the filesystem but only the last one is kept in the checkpoint file. Defaults to 5 (that is, the 5 most recent checkpoint files are kept.)

  • keep_checkpoint_every_n_hours: In addition to keeping the most recent max_to_keep 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 keep_checkpoint_every_n_hours=2 ensures that you keep one checkpoint file for every 2 hours of training. The default value of 10,000 hours effectively disables the feature.

Note that you still have to call the save() method to save the model. Passing these arguments to the constructor will not save variables automatically for you.

A training program that saves regularly looks like:

...
# Create a saver.
saver = tf.compat.v1.train.Saver(...variables...)
# Launch the graph and train, saving the model every 1,000 steps.
sess = tf.compat.v1.Session()
for step in xrange(1000000):
    sess.run(..training_op..)
    if step % 1000 == 0:
        # Append the step number to the checkpoint name:
        saver.save(sess, 'my-model', global_step=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 and by latest_checkpoint(), which makes it easy to discover the path to the most recent checkpoint. That protocol buffer is stored in a file named 'checkpoint' next to the checkpoint files.

If you create several savers, you can specify a different filename for the protocol buffer file in the call to save().

var_list A list of Variable/SaveableObject, or a dictionary mapping names to SaveableObjects. If None, defaults to the list of all saveable objects.
reshape If True, allows restoring parameters from a checkpoint where the variables have a different shape.
sharded If True, shard the checkpoints, one per device.
max_to_keep Maximum number of recent checkpoints to keep. Defaults to 5.
keep_checkpoint_every_n_hours How often to keep checkpoints. Defaults to 10,000 hours.
name String. Optional name to use as a prefix when adding operations.
restore_sequentially A Bool, which if true, causes restore of different variables to happen sequentially within each device. This can lower memory usage when restoring very large models.
saver_def Optional SaverDef proto to use instead of running the builder. This is only useful for specialty code that wants to recreate a Saver object for a previously built Graph that had a Saver. The saver_def proto should be the one returned by the as_saver_def() call of the Saver that was created for that Graph.
builder Optional SaverBuilder to use if a saver_def was not provided. Defaults to BulkSaverBuilder().
defer_build If True, defer adding the save and restore ops to the build() call. In that case build() should be called before finalizing the graph or using the saver.
allow_empty If False (default) raise an error if there are no variables in the graph. Otherwise, construct the saver anyway and make it a no-op.
write_version controls what format to use when saving checkpoints. It also affects certain filepath matching logic. The V2 format is the recommended choice: it is much more optimized than V1 in terms of memory required and latency incurred during restore. Regardless of this flag, the Saver is able to restore from both V2 and V1 checkpoints.
pad_step_number if True, pads the global step number in the checkpoint filepaths to some fixed width (8 by default). This is turned off by default.
save_relative_paths If True, will write relative paths to the checkpoint state file. This is needed if the user wants to copy the checkpoint directory and reload from the copied directory.
filename If known at graph construction time, filename used for variable loading/saving.

TypeError If var_list is invalid.
ValueError If any of the keys or values in var_list are not unique.
RuntimeError If eager execution is enabled andvar_list does not specify a list of variables to save.

last_checkpoints List of not-yet-deleted checkpoint filenames.

You can pass any of the returned values to restore().

Methods

as_saver_def

View source

Generates a SaverDef representation of this saver.

Returns
A SaverDef proto.

build

View source

export_meta_graph

View source

Writes MetaGraphDef to save_path/filename.

Args
filename Optional meta_graph filename including the path.
collection_list List of string keys to collect.
as_text If True, writes the meta_graph as an ASCII proto.
export_scope Optional string. Name scope to remove.
clear_devices Whether or not to clear the device field for an Operation or Tensor during export.
clear_extraneous_savers Remove any Saver-related information from the graph (both Save/Restore ops and SaverDefs) that are not associated with this Saver.
strip_default_attrs Boolean. If True, default-valued attributes will be removed from the NodeDefs. For a detailed guide, see Stripping Default-Valued Attributes.
save_debug_info If True, save the GraphDebugInfo to a separate file, which in the same directory of filename and with _debug added before the file extension.

Returns
A MetaGraphDef proto.

from_proto

View source

Returns a Saver object created from saver_def.

Args
saver_def a SaverDef protocol buffer.
import_scope Optional string. Name scope to use.

Returns
A Saver built from saver_def.

recover_last_checkpoints

View source

Recovers the internal saver state after a crash.

This method is useful for recovering the "self._last_checkpoints" state.

Globs for the checkpoints pointed to by checkpoint_paths. If the files exist, use their mtime as the checkpoint timestamp.

Args
checkpoint_paths a list of checkpoint paths.

restore

View source

Restores previously saved variables.

This method runs the ops added by the constructor for restoring variables. It requires a session in which the graph was launched. The variables to restore do not have to have been initialized, as restoring is itself a way to initialize variables.

The save_path argument is typically a value previously returned from a save() call, or a call to latest_checkpoint().

Args
sess A Session to use to restore the parameters. None in eager mode.
save_path Path where parameters were previously saved.

Raises
ValueError If save_path is None or not a valid checkpoint.

save

View source

Saves variables.

This method runs the ops added by the constructor for saving variables. It requires a session in which the graph was launched. The variables to save must also have been initialized.

The method returns the path prefix of the newly created checkpoint files. This string can be passed directly to a call to restore().

Args
sess A Session to use to save the variables.
save_path String. Prefix of filenames created for the checkpoint.
global_step If provided the global step number is appended to save_path to create the checkpoint filenames. The optional argument can be a Tensor, a Tensor name or an integer.
latest_filename Optional name for the protocol buffer file that will contains the list of most recent checkpoints. That file, kept in the same directory as the checkpoint files, is automatically managed by the saver to keep track of recent checkpoints. Defaults to 'checkpoint'.
meta_graph_suffix Suffix for MetaGraphDef file. Defaults to 'meta'.
write_meta_graph Boolean indicating whether or not to write the meta graph file.
write_state Boolean indicating whether or not to write the CheckpointStateProto.
strip_default_attrs Boolean. If True, default-valued attributes will be removed from the NodeDefs. For a detailed guide, see Stripping Default-Valued Attributes.
save_debug_info If True, save the GraphDebugInfo to a separate file, which in the same directory of save_path and with _debug added before the file extension. This is only enabled when write_meta_graph is True

Returns
A string: path prefix used for the checkpoint files. If the saver is sharded, this string ends with: '-?????-of-nnnnn' where 'nnnnn' is the number of shards created. If the saver is empty, returns None.

Raises
TypeError If sess is not a Session.
ValueError If latest_filename contains path components, or if it collides with save_path.
RuntimeError If save and restore ops weren't built.

set_last_checkpoints

View source

Sets the list of old checkpoint filenames.

Args
last_checkpoints A list of checkpoint filenames.

Raises
AssertionError If last_checkpoints is not a list.

set_last_checkpoints_with_time

View source

Sets the list of old checkpoint filenames and timestamps.

Args
last_checkpoints_with_time A list of tuples of checkpoint filenames and timestamps.

Raises
AssertionError If last_checkpoints_with_time is not a list.

to_proto

View source

Converts this Saver to a SaverDef protocol buffer.

Args
export_scope Optional string. Name scope to remove.

Returns
A SaverDef protocol buffer.