|
|
|
@@ -893,11 +893,11 @@ class Cell(Cell_): |
|
|
|
>>> |
|
|
|
>>> context.set_context(mode=context.GRAPH_MODE, device_target="Ascend") |
|
|
|
>>> class reluNet(nn.Cell): |
|
|
|
>>> def __init__(self): |
|
|
|
>>> super(reluNet, self).__init__() |
|
|
|
>>> self.relu = nn.ReLU() |
|
|
|
>>> def construct(self, x): |
|
|
|
>>> return self.relu(x) |
|
|
|
... def __init__(self): |
|
|
|
... super(reluNet, self).__init__() |
|
|
|
... self.relu = nn.ReLU() |
|
|
|
... def construct(self, x): |
|
|
|
... return self.relu(x) |
|
|
|
>>> |
|
|
|
>>> net = reluNet() |
|
|
|
>>> input_dyn = Tensor(shape=[3, None], dtype=ms.float32) |
|
|
|
@@ -909,9 +909,11 @@ class Cell(Cell_): |
|
|
|
This is an experimental interface that is subject to change or deletion. |
|
|
|
""" |
|
|
|
|
|
|
|
for ele in self._dynamic_shape_inputs: |
|
|
|
if isinstance(ele, (str, int, dict)): |
|
|
|
raise TypeError(f"For element in 'set_inputs', the type must be Tensor,\ |
|
|
|
but got {type(ele)}.") |
|
|
|
self._dynamic_shape_inputs = inputs |
|
|
|
if isinstance(self._dynamic_shape_inputs[0], (str, int, dict)): |
|
|
|
raise TypeError(f"For 'set_inputs, the type must be tuple, but got {type(self._dynamic_shape_inputs[0])}.") |
|
|
|
|
|
|
|
def get_inputs(self): |
|
|
|
""" |
|
|
|
|