Function test_SkipConnectionConvModule

numpy_ml/tests/test_nn.py:1822–1987

(N=15)

def test_SkipConnectionConvModule(N=15):
    from numpy_ml.neural_nets.modules import SkipConnectionConvModule
    from numpy_ml.neural_nets.activations import Tanh, ReLU, Sigmoid, Affine

    N = np.inf if N is None else N

    np.random.seed(12345)

    acts = [
        (Tanh(), nn.Tanh(), "Tanh"),
        (Sigmoid(), nn.Sigmoid(), "Sigmoid"),
        (ReLU(), nn.ReLU(), "ReLU"),
        (Affine(), TorchLinearActivation(), "Affine"),
    ]

    i = 1
    while i < N + 1:
        n_ex = np.random.randint(2, 10)
        in_rows = np.random.randint(2, 10)
        in_cols = np.random.randint(2, 10)
        n_in = np.random.randint(2, 5)
        n_out1 = np.random.randint(2, 5)
        n_out2 = np.random.randint(2, 5)
        f_shape1 = (
            min(in_rows, np.random.randint(1, 5)),
            min(in_cols, np.random.randint(1, 5)),
        )
        f_shape2 = (
            min(in_rows, np.random.randint(1, 5)),
            min(in_cols, np.random.randint(1, 5)),
        )
        f_shape_skip = (
            min(in_rows, np.random.randint(1, 5)),
            min(in_cols, np.random.randint(1, 5)),
        )

        s1 = np.random.randint(1, 5)
        s2 = np.random.randint(1, 5)
        s_skip = np.random.randint(1, 5)

        # randomly select an activation function
        act_fn, torch_fn, act_fn_name = acts[np.random.randint(0, len(acts))]

        X = random_tensor((n_ex, in_rows, in_cols, n_in), standardize=True)

        p1 = (np.random.randint(1, 5), np.random.randint(1, 5))
        p2 = (np.random.randint(1, 5), np.random.randint(1, 5))

        # initialize SkipConnectionConv module
        L1 = SkipConnectionConvModule(
            out_ch1=n_out1,
            out_ch2=n_out2,
            kernel_shape1=f_shape1,
            kernel_shape2=f_shape2,
            kernel_shape_skip=f_shape_skip,
            stride1=s1,
            stride2=s2,
            stride_skip=s_skip,