Apply to speak at TensorFlow World. Deadline April 23rd. Propose talk

tf.contrib.checkpoint.capture_dependencies

tf.contrib.checkpoint.capture_dependencies(template)

Defined in tensorflow/python/training/checkpointable/util.py.

Capture variables created within this scope as Template dependencies.

Requires that template.variable_scope is active.

This scope is intended as a compatibility measure, allowing a checkpointable object to add dependencies on variables created in a block of code which is not aware of object-based saving (and instead uses variable names heavily). This is how Template objects add dependencies on variables and sub-Templates. Where possible, use tf.make_template directly.

Args:

  • template: The Template object to register dependencies with.

Yields:

None (when used as a context manager).