def forward_cpu(self, inputs):
gy, = inputs
gx1 = utils.force_array(numpy.where(self.cond, gy, gy.dtype.type(0)))
gx2 = utils.force_array(numpy.where(self.cond, gy.dtype.type(0), gy))
return (
utils.sum_to(gx1, self.x1_shape),
utils.sum_to(gx2, self.x2_shape))
def forward_gpu(self, inputs):
gy, = inputs
gx1, gx2 = cuda.elementwise(
'S cond, T gy', 'T gx1, T gx2', '''
gx1 = cond ? gy : (T)0.0;
gx2 = cond ? (T)0.0 : gy;
''', 'maximum_bwd1')(self.cond, gy)
return (utils.sum_to(gx1,
self.x1_shape), utils.sum_to(gx2, self.x2_shape))
def forward_gpu(self, inputs):
self.retain_inputs((0, 1, 2))
x0, x1, gy = inputs
gx0, gx1 = cuda.elementwise(
'T x0, T x1, T gy', 'T gx0, T gx1', '''
gx0 = gy / x1;
gx1 = -gx0 * x0 / x1;
''', 'div_bwd')(x0, x1, gy)
return utils.sum_to(gx0, x0.shape), utils.sum_to(gx1, x1.shape)
def forward_cpu(self, inputs):
self.retain_inputs((0, 1, 2))
x0, x1, gy = inputs
one = x1.dtype.type(1)
gx0 = utils.sum_to(utils.force_array(x1 * (x0**(x1 - one)) * gy),
x0.shape)
gx1 = utils.sum_to(utils.force_array(numpy.log(x0) * self.y * gy),
x1.shape)
return gx0, gx1
def forward_cpu(self, inputs):
self.retain_inputs((0, 1, 2))
x0, x1, gy = inputs
one = x1.dtype.type(1)
gx0 = utils.sum_to(
utils.force_array(x1 * (x0 ** (x1 - one)) * gy), x0.shape)
gx1 = utils.sum_to(
utils.force_array(numpy.log(x0) * self.y * gy), x1.shape)
return gx0, gx1
def forward_gpu(self, inputs):
gy, = inputs
x1, x2 = self.x1, self.x2
gx1 = cuda.elementwise('T x1, T x2, T gy', 'T gx1',
'gx1 = (x1 <= x2) ? gy : (T)0.0',
'minimum_bwd1')(x1, x2, gy)
gx2 = cuda.elementwise('T x1, T x2, T gy', 'T gx1',
'gx1 = (x1 > x2) ? gy : (T)0.0',
'minimum_bwd2')(x1, x2, gy)
return utils.sum_to(gx1, x1.shape), utils.sum_to(gx2, x2.shape)
def forward_gpu(self, inputs):
self.retain_inputs((0, 1, 2))
x0, x1, gy = inputs
gx0, gx1 = cuda.elementwise(
'T x0, T x1, T gy',
'T gx0, T gx1',
'''
gx0 = gy / x1;
gx1 = -gx0 * x0 / x1;
''', 'div_bwd')(x0, x1, gy)
return utils.sum_to(gx0, x0.shape), utils.sum_to(gx1, x1.shape)
def forward_gpu(self, inputs):
gy, = inputs
x1, x2 = self.x1, self.x2
gx1 = cuda.elementwise(
'T x1, T x2, T gy', 'T gx1',
'gx1 = (x1 <= x2) ? gy : (T)0.0',
'minimum_bwd1')(x1, x2, gy)
gx2 = cuda.elementwise(
'T x1, T x2, T gy', 'T gx1',
'gx1 = (x1 > x2) ? gy : (T)0.0',
'minimum_bwd2')(x1, x2, gy)
return utils.sum_to(gx1, x1.shape), utils.sum_to(gx2, x2.shape)
def forward_gpu(self, inputs):
gy, = inputs
gx1, gx2 = cuda.elementwise(
'S cond, T gy', 'T gx1, T gx2',
'''
gx1 = cond ? gy : (T)0.0;
gx2 = cond ? (T)0.0 : gy;
''',
'maximum_bwd1')(self.cond, gy)
return (
utils.sum_to(gx1, self.x1_shape),
utils.sum_to(gx2, self.x2_shape))
def forward_gpu(self, inputs):
self.retain_inputs((0, 1, 2))
x0, x1, gy = inputs
gx0, gx1 = cuda.elementwise(
'T x0, T x1, T gy, T y', 'T gx0, T gx1', '''
gx0 = x1 * pow(x0, x1 - 1) * gy;
gx1 = log(x0) * y * gy;
''', 'pow_var_var_bwd')(x0, x1, gy, self.y)
gx0 = utils.sum_to(gx0, x0.shape)
gx1 = utils.sum_to(gx1, x1.shape)
return gx0, gx1
def forward_gpu(self, inputs):
self.retain_inputs((0, 1, 2))
x0, x1, gy = inputs
gx0, gx1 = cuda.elementwise(
'T x0, T x1, T gy, T y', 'T gx0, T gx1',
'''
gx0 = x1 * pow(x0, x1 - 1) * gy;
gx1 = log(x0) * y * gy;
''', 'pow_var_var_bwd')(x0, x1, gy, self.y)
gx0 = utils.sum_to(gx0, x0.shape)
gx1 = utils.sum_to(gx1, x1.shape)
return gx0, gx1
def forward_cpu(self, inputs):
self.retain_inputs((0, 1, 2))
x0, x1, gy = inputs
gx0 = utils.force_array(gy / x1)
gx1 = utils.force_array(-gx0 * x0 / x1)
return utils.sum_to(gx0, x0.shape), utils.sum_to(gx1, x1.shape)
def forward(self, inputs):
x, = inputs
return utils.sum_to(x, self._shape),
def forward_cpu(self, inputs):
gy, = inputs
x1, x2 = self.x1, self.x2
gx1 = utils.force_array(gy * (x1 <= x2))
gx2 = utils.force_array(gy * (x1 > x2))
return utils.sum_to(gx1, x1.shape), utils.sum_to(gx2, x2.shape)
def forward_cpu(self, inputs):
self.retain_inputs((0, 1, 2))
x0, x1, gy = inputs
gx0 = utils.force_array(gy / x1)
gx1 = utils.force_array(-gx0 * x0 / x1)
return utils.sum_to(gx0, x0.shape), utils.sum_to(gx1, x1.shape)
def forward(self, inputs):
x, = inputs
return utils.sum_to(x, self._shape),