def test_backward3(self):
n, c, h, w = 1, 5, 20, 15
o, k, s, p = 3, (5, 3), 1, 3
x = np.random.randn(n, c, h, w)
W = np.random.randn(o, c, k[0], k[1])
b = np.random.randn(o)
f = lambda W: F.conv2d_simple(x, W, b, s, p)
self.assertTrue(gradient_check(f, W))
def test_backward3(self):
x_data = np.random.randn(10, 10)
def f(x):
np.random.seed(0)
return F.dropout(x, 0.0)
self.assertTrue(gradient_check(f, x_data))
def test_backward3(self):
n, c_i, c_o = 10, 1, 3
h_i, w_i = 5, 10
h_k, w_k = 10, 10
h_p, w_p = 5, 5
s_y, s_x = 5, 5
x = np.random.uniform(0, 1, (n, c_i, h_i, w_i))
W = np.random.uniform(0, 1, (c_i, c_o, h_k, w_k))
b = np.random.uniform(0, 1, c_o)
f = lambda b: F.deconv2d(x, W, b, stride=(s_y, s_x), pad=(h_p, w_p))
self.assertTrue(gradient_check(f, b))
def test_backward2(self):
n, c, h, w = 1, 1, 3, 3
x = np.random.rand(1, 1, 3, 3, 1, 1)
f = lambda x: F.col2im(x, (n, c, h, w), 3, 3, 0, to_matrix=False)
self.assertTrue(gradient_check(f, x))
def test_backward1(self):
x = np.array([[1, 2, 3], [4, 5, 6]])
self.assertTrue(gradient_check(F.transpose, x))
def test_backward3(self):
np.random.seed(0)
x_data = np.random.rand(10, 10, 10)
self.assertTrue(gradient_check(F.sigmoid, x_data))
def test_backward4(self):
x_data = np.random.rand(10, 20, 20)
f = lambda x: F.sum(x, axis=None)
self.assertTrue(gradient_check(f, x_data))
def test_backward3(self):
N, C = 8, 3
x, gamma, beta, mean, var = get_params(N, C, dtype=np.float64)
f = lambda beta: F.batch_nrom(x, gamma, beta, mean, var)
self.assertTrue(gradient_check(f, beta))
def test_backward3(self):
x_data = np.random.rand(10, 20, 20) * 100
f = lambda x: F.sum_to(x, (10, ))
self.assertTrue(gradient_check(f, x_data))
def test_backward1(self):
x = np.array([[-1, 0, 1, 2], [2, 0, 1, -1]])
f = lambda x: F.log_softmax(x)
self.assertTrue(gradient_check(f, x))
def test_backward1(self):
x_data = np.array([[0, 1, 2], [0, 2, 4]])
f = lambda x: F.softmax(x, axis=1)
self.assertTrue(gradient_check(f, x_data))
def test_backward3(self):
x = np.random.randn(3, 3)
y = np.random.randn(3, 1)
f = lambda x: x / y
self.assertTrue(gradient_check(f, x))
def test_backward1(self):
x = np.array([[-1, 0, 1, 2], [2, 0, 1, -1]], np.float32)
t = np.array([3, 0]).astype(np.int32)
f = lambda x: F.softmax_cross_entropy_simple(x, Variable(t))
self.assertTrue(gradient_check(f, x))
def test_backward3(self):
x = np.random.randn(3, 3)
y = np.random.randn(3, 1)
self.assertTrue(gradient_check(F.add, x, y))
def test_backward2(self):
x = Variable(np.random.randn(3, 3))
y = np.random.randn(3, 1)
f = lambda x: x + y
self.assertTrue(gradient_check(f, x))
def test_backward2(self):
x = np.random.randn(100, 200)
W = np.random.randn(200, 300)
b = None
f = lambda x: F.linear(x, W, b)
self.assertTrue(gradient_check(f, x))
def test_backward1(self):
x = np.random.randn(3, 2)
W = np.random.randn(2, 3)
b = np.random.randn(3)
f = lambda x: F.linear(x, W, b)
self.assertTrue(gradient_check(f, x))
def test_backward4(self):
x_data = np.random.rand(10)
f = lambda x: F.sum_to(x, (10, )) + 1
self.assertTrue(gradient_check(f, x_data))
def test_backward3(self):
N, CLS_NUM = 100, 10
x = np.random.randn(N, CLS_NUM)
t = np.random.randint(0, CLS_NUM, (N, ))
f = lambda x: F.softmax_cross_entropy_simple(x, t)
self.assertTrue(gradient_check(f, x))
def test_backward3(self):
x = np.random.randn(10, 5)
self.assertTrue(gradient_check(F.transpose, x))
def test_backward2(self):
x = np.random.randn(10, 10)
f = lambda x: F.log_softmax(x)
self.assertTrue(gradient_check(f, x))
def test_backward4(self):
x = np.array([1, 2])
self.assertTrue(gradient_check(F.transpose, x))
def test_backward3(self):
np.random.seed(0)
x_data = np.random.rand(10, 10, 10)
f = lambda x: F.softmax(x, axis=1)
self.assertTrue(gradient_check(f, x_data))
def test_backward1(self):
x_data = np.array([[1, 2, 3], [4, 5, 6]])
slices = 1
f = lambda x: F.get_item(x, slices)
gradient_check(f, x_data)
def test_backward6(self):
params = 10, 20, 5, 5
x, gamma, beta, mean, var = get_params(*params, dtype=np.float64)
f = lambda beta: F.batch_nrom(x, gamma, beta, mean, var)
self.assertTrue(gradient_check(f, beta))
def test_backward2(self):
x_data = np.arange(12).reshape(4, 3)
slices = slice(1, 3)
f = lambda x: F.get_item(x, slices)
gradient_check(f, x_data)
def test_backward1(self):
x_data = np.array([[0, 1, 2], [0, 2, 4]])
self.assertTrue(gradient_check(F.sigmoid, x_data))
def test_backward1(self):
n, c, h, w = 1, 5, 16, 16
ksize, stride, pad = 2, 2, 0
x = np.random.randn(n, c, h, w).astype('f') * 1000
f = lambda x: F.pooling(x, ksize, stride, pad)
self.assertTrue(gradient_check(f, x))
def test_backward2(self):
x0 = np.random.rand(100)
x1 = np.random.rand(100)
f = lambda x0: F.mean_squared_error(x0, x1)
self.assertTrue(gradient_check(f, x0))
def test_backward2(self):
n, c, h, w = 1, 1, 3, 3
x = np.arange(n * c * h * w).reshape((n, c, h, w))
f = lambda x: F.im2col(x, 3, 3, 0, to_matrix=False)
self.assertTrue(gradient_check(f, x))