def testCumSum(self):
I = K.variable(np.array([1., 2., 3., 4., 5., 6.]))
code = """function (I[N]) -> (O) {
O[i: N] = +(I[i - j]), j < N;
}"""
op = K._Op('cumulative_sum', I.dtype, (6,), code,
OrderedDict([('I', I)]), ['O'])
code2 = """function (I[N]) -> (O) {
O[i: N] = +(I[k]), i - k < N;
}"""
op2 = K._Op('cumulative_sum2', I.dtype, (6,), code,
OrderedDict([('I', I)]), ['O'])
reference = K.cumsum(I)
npt.assert_allclose(op.eval(), reference.eval())
npt.assert_allclose(op2.eval(), reference.eval())
def testSkipping(self):
I = K.variable(np.array([[1., 2.], [3., 4.], [5., -4.], [-5., 6.], [-7., 9.]]))
code = """function (I[N, M]) -> (O) {
O[2 * i: N] = +(I[2 * i, j]);
}"""
op = K._Op('skip', I.dtype, (5,), code,
OrderedDict([('I', I)]), ['O'])
npt.assert_allclose(op.eval(), np.array([3., 0., 1., 0., 2.]))
def testSameMaxPool(self):
I = K.variable(np.array([1., 2., 3., 4., 5.]))
code = """function (I[N]) -> (O) {
O[i: (N + 1) / 2] = >(I[2 * i + j]), j < 2;
}"""
op = K._Op('valid_maxpool', I.dtype, ((I.shape[0] + 1)/2,), code,
OrderedDict([('I', I)]), ['O'])
npt.assert_allclose(op.eval(), np.array([2., 4., 5.]))
def testMaxOverAxis(self):
I = K.variable(np.array([[1., 2., 3.], [4., 5., 6.]]))
code = """function (I[M, N]) -> (O) {
O[n: N] = >(I[m, n]);
}"""
op = K._Op('max_over_axis', I.dtype, I.shape[1], code,
OrderedDict([('I', I)]), ['O'])
reference = K.max(I, axis=0)
npt.assert_allclose(op.eval(), reference.eval())
def testBrokenMaxPool(self):
I = K.variable(np.array([1., 2., 3., 4., 5.]))
code = """function (I[N]) -> (O) {
O[i: N / 2] = >(I[2 * i + j]);
}"""
op = K._Op('broken_maxpool', I.dtype, (I.shape[0]/2,), code,
OrderedDict([('I', I)]), ['O'])
reference = K.max(I)
npt.assert_allclose(op.eval(), [5.] * 2)
def testGlobalMean(self):
I = K.variable(np.array([[1., 2., 3.], [4., 5., 6.]]))
code = """function (I[X, Y]) -> (O) {
Sum[] = +(I[x, y]);
O = Sum / (X * Y);
}"""
op = K._Op('global_mean', I.dtype, tuple(), code,
OrderedDict([('I', I)]), ['O'])
reference = K.mean(I, axis=[0, 1])
npt.assert_allclose(op.eval(), reference.eval())
def testMeanOverAxis(self):
I = K.variable(np.array([[1., 2., 3.], [4., 5., 6.]]))
code = """function (I[X, Y]) -> (O) {
Sum[y: Y] = +(I[x, y]);
O = Sum / X;
}"""
op = K._Op('mean_over_axis', I.dtype, I.shape[1], code,
OrderedDict([('I', I)]), ['O'])
reference = K.mean(I, axis=0)
npt.assert_allclose(op.eval(), reference.eval())
def testMatMul(self):
A = K.variable(np.array([[1., 2., 3.], [4., 5., 6.]]))
B = K.variable(np.array([[1., -2.], [-3., 4.], [5., -6.]]))
code = """function (A[M, L], B[L, N]) -> (C) {
C[i, j: M, N] = +(A[i, k] * B[k, j]);
}"""
op = K._Op('matmul', A.dtype, (2, 2), code,
OrderedDict([('A', A), ('B', B)]), ['C'])
reference = K.dot(A, B)
npt.assert_allclose(op.eval(), reference.eval())
def testConv1D(self):
I = K.variable(m(2, 8, 3))
kernel = K.variable(m(3, 3, 2))
code = """function (I[N, L, CI], K[LK, CI, CO]) -> (O) {
O[n, x, co: N, L - LK + 1, CO] = +(I[n, x + k, ci] * K[k, ci, co]);
}"""
op = K._Op('cumulative_sum', I.dtype, (2, 6, 3), code,
OrderedDict([('I', I), ('K', kernel)]), ['O'])
reference = K.conv1d(I, kernel, padding='valid')
npt.assert_allclose(op.eval(), reference.eval())
def testGlobalMin(self):
I = K.variable(np.array([[[1., 2., 3.], [4., 5., 6.]]]))
code = """function (I) -> (O) {
Neg = -I;
O_Neg[] = >(Neg[i, j, k]);
O = -O_Neg;
}"""
op = K._Op('global_min', I.dtype, tuple(), code,
OrderedDict([('I', I)]), ['O'])
reference = K.min(I, axis=[0, 1, 2])
npt.assert_allclose(op.eval(), reference.eval())
def testDilatedConv2D(self):
I = K.variable(m(2, 6, 10, 3))
kernel = K.variable(m(3, 2, 3, 2))
code = """function (I[N, Lx, Ly, CI], K[LKx, LKy, CI, CO]) -> (O) {
O[n, x, y, co: N, Lx - 2 * (LKx - 1), Ly - 3 * (LKy - 1), CO] =
+(I[n, x + 2 * kx, y + 3 * ky, ci] * K[kx, ky, ci, co]);
}"""
op = K._Op('cumulative_sum', I.dtype, (2, 2, 7, 2), code,
OrderedDict([('I', I), ('K', kernel)]), ['O'])
reference = K.conv2d(I, kernel, padding='valid', dilation_rate=(2, 3))
npt.assert_allclose(op.eval(), reference.eval())
def main(code, tensor_A, tensor_B, output_shape):
print(K.backend())
op = K._Op('sandbox_op', A.dtype, output_shape, code,
OrderedDict([('A', tensor_A), ('B', tensor_B)]), ['O'])
print(op.eval())
