def test_reshape(self):
a = np.array([[[1.1, 1.1, 1.4], [1.1, 1.1, 1.1]],
[[1.1, 1.1, 1.3], [1.6, 1.1, 1.2]]])
ta = tensor.from_numpy(a)
tb = tensor.reshape(ta, [2, 6])
self.assertAlmostEqual(tb.shape[0], 2., places=3)
self.assertAlmostEqual(tb.shape[1], 6., places=3)
np.testing.assert_array_almost_equal(tensor.to_numpy(tb),
a.reshape((2, 6)))
def test_gpu_6d_transpose(self, dev=gpu_dev):
s0 = (2, 3, 4, 5, 6, 7)
axes1 = [5, 4, 3, 2, 1, 0]
s1 = (2, 7, 6, 5, 4, 3)
s2 = (2, 4, 3, 5, 7, 6)
a = np.random.random(s1)
ta = tensor.from_numpy(a)
ta.to_device(dev)
ta = tensor.reshape(ta, s1)
ta = tensor.transpose(ta, axes1)
ta = tensor.reshape(ta, s2)
a = np.reshape(a, s1)
a = np.transpose(a, axes1)
a = np.reshape(a, s2)
np.testing.assert_array_almost_equal(tensor.to_numpy(ta), a)
def forward(self, flag, input):
''' Reshape the input tensor into output_shape
Args:
flag: not used
input: a single input tensor
Returns:
output tensor (reshaped input)
'''
assert isinstance(input, tensor.Tensor), 'The input must be Tensor'
self.in_shape = input
outputs = tensor.reshape(
input,
tuple(list([(self.in_shape.size() / self.output_size)])) +
self.output_shape)
return outputs
def backward(self, flag, dy):
''' Return gradient, []'''
dx = tensor.reshape(dy, self.in_shape.shape)
return dx, []
def forward(self, *x):
# if you change to other models, please update the output name here
y = super(MyModel, self).forward(*x, last_layers=-4)[0]
y = self.conv(y)
y = tensor.reshape(y, y.shape[:2])
return y