Method _get_variable_name

official/modeling/optimization/lamb.py:247–252

Get the variable name from the tensor name.

(self, param_name)

    return True

  def _get_variable_name(self, param_name):
    """Get the variable name from the tensor name."""
    m = re.match("^(.*):\\d+$", param_name)
    if m is not None:
      param_name = m.group(1)
    return param_name