def test_forward_consistency(self, nobias=False):
x_cpu = chainer.Variable(self.x)
W_cpu = chainer.Variable(self.W)
b_cpu = None if nobias else chainer.Variable(self.b)
y_cpu = functions.dilated_convolution_2d(x_cpu,
W_cpu,
b_cpu,
stride=self.stride,
pad=self.pad,
dilate=self.dilate,
use_cudnn=self.use_cudnn,
cover_all=self.cover_all)
x_gpu = chainer.Variable(cuda.to_gpu(self.x))
W_gpu = chainer.Variable(cuda.to_gpu(self.W))
b_gpu = None if nobias else chainer.Variable(cuda.to_gpu(self.b))
y_gpu = functions.dilated_convolution_2d(x_gpu,
W_gpu,
b_gpu,
stride=self.stride,
pad=self.pad,
dilate=self.dilate,
use_cudnn=self.use_cudnn,
cover_all=self.cover_all)
testing.assert_allclose(y_cpu.data, y_gpu.data.get(),
**self.check_forward_options)
def test_forward_consistency(self, nobias=False):
x_cpu = chainer.Variable(self.x)
W_cpu = chainer.Variable(self.W)
b_cpu = None if nobias else chainer.Variable(self.b)
y_cpu = functions.dilated_convolution_2d(
x_cpu, W_cpu, b_cpu, stride=self.stride, pad=self.pad,
dilate=self.dilate, cover_all=self.cover_all)
x_gpu = chainer.Variable(cuda.to_gpu(self.x))
W_gpu = chainer.Variable(cuda.to_gpu(self.W))
b_gpu = None if nobias else chainer.Variable(cuda.to_gpu(self.b))
with chainer.using_config('use_cudnn', self.use_cudnn):
y_gpu = functions.dilated_convolution_2d(
x_gpu, W_gpu, b_gpu, stride=self.stride, pad=self.pad,
dilate=self.dilate, cover_all=self.cover_all)
testing.assert_allclose(
y_cpu.data, y_gpu.data.get(), **self.check_forward_options)
def forward(self):
x = chainer.Variable(self.x)
W = chainer.Variable(self.W)
return functions.dilated_convolution_2d(x,
W,
None,
stride=self.stride,
pad=self.pad,
dilate=self.dilate)
def forward(self, x):
if self.W.array is None:
self._initialize_params(x.shape[1])
pad_width = [(0, 0), (0, 0)] + list(map(lambda x: (x, x), self.pad))
x = F.pad(x, pad_width, self.pad_mode)
return F.dilated_convolution_2d(x, self.W, self.b, self.stride, 0,
self.dilate)
def forward(self):
x = chainer.Variable(self.x)
W = chainer.Variable(self.W)
return functions.dilated_convolution_2d(
x, W, None, stride=self.stride, pad=self.pad, dilate=self.dilate)
def f(*args):
return F.dilated_convolution_2d(*args, stride=self.stride,
pad=self.pad, dilate=self.dilate,
cover_all=self.cover_all)
