tf.contrib.framework.init_from_checkpoint

tf.contrib.framework.init_from_checkpoint(
    checkpoint_dir,
    assignment_map
)

Defined in tensorflow/contrib/framework/python/framework/checkpoint_utils.py.

See the guide: Framework (contrib) > Checkpoint utilities

Using assignment map initializes current variables with loaded tensors.

Assignment map supports following syntax:

  • 'checkpoint_scope_name/': 'scope_name/' - will load all variables in current scope_name from checkpoint_scope_name with matching variable names.
  • 'checkpoint_scope_name/some_other_variable': 'scope_name/variable_name' - will initialize scope_name/variable_name variable from checkpoint_scope_name/some_other_variable.
  • 'scope_variable_name': variable - will initialize given tf.Variable object with variable from the checkpoint.
  • 'scope_variable_name': list(variable) - will initialize list of partitioned variables with variable from the checkpoint.
  • '/': 'scope_name/' - will load all variables in current scope_name from checkpoint's root (e.g. no scope).

Supports loading into partitioned variables, which are represented as '<variable>/part_<part #>'.

Example:

  # Create variables.
  with tf.variable_scope('test'):
    m = tf.get_variable('my_var')
  with tf.variable_scope('test2'):
    var2 = tf.get_variable('my_var')
  var3 = tf.get_variable(name="my1", shape=[100, 100],
                         partitioner=lambda shape, dtype: [5, 1])
  ...
  # Specify which variables to initialize from checkpoint.
  init_from_checkpoint(checkpoint_dir, {
    'some_var': 'test/my_var',
    'some_scope/': 'test2/'})
  ...
  # Or use `Variable` objects to identify what to initialize.
  init_from_checkpoint(checkpoint_dir, {
    'some_scope/var2': var2,
  })
  # Initialize partitioned variables
  init_from_checkpoint(checkpoint_dir, {
    'some_var_from_ckpt': 'part_var',
  })
  # Or specifying the list of `Variable` objects.
  init_from_checkpoint(checkpoint_dir, {
    'some_var_from_ckpt': var3._get_variable_list(),
  })
  ...
  # Initialize variables as usual.
  session.run(tf.get_all_variables())

Args:

  • checkpoint_dir: Directory with checkpoints file or path to checkpoint.
  • assignment_map: Dict, where keys are names of the variables in the checkpoint and values are current variables or names of current variables (in default graph).

Raises:

  • tf.errors.OpError: If missing checkpoints or tensors in checkpoints.
  • ValueError: If missing variables in current graph.