MCPcopy Create free account
hub / github.com/DeepRec-AI/DeepRec / register

Function register

tensorflow/python/eager/function.py:2152–2176  ·  view source on GitHub ↗

Register a specialization of a `Function` into the graph. This won't actually call the function with the inputs, and only put the function definition into graph. Register function with different input param will result into multiple version of functions registered in graph. Args: func:

(func, *args, **kwargs)

Source from the content-addressed store, hash-verified

2150
2151
2152def register(func, *args, **kwargs):
2153 """Register a specialization of a `Function` into the graph.
2154
2155 This won't actually call the function with the inputs, and only put the
2156 function definition into graph. Register function with different input param
2157 will result into multiple version of functions registered in graph.
2158
2159 Args:
2160 func: the `Function` instance that generated by a @defun
2161 *args: input arguments for the Python function.
2162 **kwargs: input keyword arguments for the Python function.
2163
2164 Returns:
2165 a `ConcreteFunction` object specialized to inputs and execution context.
2166
2167 Raises:
2168 ValueError: When the input function is not a defun wrapped python function.
2169 """
2170 if not isinstance(func, Function):
2171 raise ValueError("Only defun function is allowed to be registered. "
2172 "Got type: %s" % type(func))
2173 concrete_func = func.get_concrete_function(*args, **kwargs)
2174 concrete_func.add_to_graph()
2175 concrete_func.add_gradient_functions_to_graph()
2176 return concrete_func
2177
2178
2179def validate_signature(signature):

Callers

nothing calls this directly

Calls 4

typeFunction · 0.85
get_concrete_functionMethod · 0.45
add_to_graphMethod · 0.45

Tested by

no test coverage detected