def feed_input(self, ref, data):
"""Bind the NumPy array to a tensor produced by ExternalSource
operator."""
if not self._built:
raise RuntimeError("Pipeline must be built first.")
if not isinstance(ref, Edge.EdgeReference):
raise TypeError(
("Expected argument one to "
"be EdgeReference. "
"Received output type {}").format(type(ref).__name__))
if isinstance(data, list):
inputs = []
for datum in data:
inputs.append(Edge.TensorCPU(datum))
self._pipe.SetExternalTensorInput(ref.name, inputs)
else:
inp = Edge.TensorListCPU(data)
self._pipe.SetExternalTLInput(ref.name, inp)
def feed_input(self, ref, data, layout=types.NHWC):
"""Bind the NumPy array to a tensor produced by ExternalSource
operator. It is worth mentioning that `ref` should not be overriden
with other operator outputs."""
if not self._built:
raise RuntimeError("Pipeline must be built first.")
if not isinstance(ref, Edge.EdgeReference):
raise TypeError(
("Expected argument one to "
"be EdgeReference. "
"Received output type {}").format(type(ref).__name__))
if isinstance(data, list):
if self._batch_size != len(data):
raise RuntimeError(
"Data list provided to feed_input needs to have batch_size length"
)
inputs = []
for datum in data:
inputs.append(Edge.TensorCPU(datum, layout))
self._pipe.SetExternalTensorInput(ref.name, inputs)
else:
inp = Edge.TensorListCPU(data, layout)
self._pipe.SetExternalTLInput(ref.name, inp)