"""
Test for the model generation
These tests check that the generated NN are as expected
It checks that both the number of layers and the shape
of the weights of the layers are what is expected
"""
from n3fit.backends import NN_PREFIX
from n3fit.model_gen import generate_nn
INSIZE = 16
OUT_SIZES = (4, 3)
BASIS_SIZE = 3
COMMON_ARGS = {
"nodes_in": INSIZE,
"nodes": OUT_SIZES,
"activations": ["sigmoid", "tanh"],
"initializer_name": "glorot_uniform",
"replica_seeds": [0],
"dropout": 0.0,
"regularizer": None,
"regularizer_args": {},
"last_layer_nodes": BASIS_SIZE,
}
[docs]
def test_generate_dense_network():
nn = generate_nn("dense", **COMMON_ARGS)
# The number of layers should be input layer + len(OUT_SIZES)
assert len(nn.layers) == len(OUT_SIZES) + 1
# Check that the number of parameters is as expected
expected_sizes = [(INSIZE, OUT_SIZES[0]), (OUT_SIZES[0]), (*OUT_SIZES,), (OUT_SIZES[1])]
for weight, esize in zip(nn.weights, expected_sizes):
assert weight.shape == esize
[docs]
def test_generate_multi_dense_network():
nn = generate_nn("multidense", **COMMON_ARGS)
# The number of layers should be input layer + len(OUT_SIZES)
assert len(nn.layers) == len(OUT_SIZES) + 1
# Check that the number of parameters is as expected
expected_sizes = [
(1, INSIZE, OUT_SIZES[0]),
(1, 1, OUT_SIZES[0]),
(1, *OUT_SIZES),
(1, 1, OUT_SIZES[1]),
]
for weight, esize in zip(nn.weights, expected_sizes):
assert weight.shape == esize
[docs]
def test_generate_dense_per_flavour_network():
nn = generate_nn("dense_per_flavour", **COMMON_ARGS).get_layer(f"{NN_PREFIX}_0")
# The number of layers should be input + BASIS_SIZE*len(OUT_SIZES) + concatenate
assert len(nn.layers) == BASIS_SIZE * len(OUT_SIZES) + 2
# The shape for this network of denses for flavours will depend on the basis_size
expected_sizes = []
expected_sizes += BASIS_SIZE * [(INSIZE, OUT_SIZES[0]), (OUT_SIZES[0],)]
expected_sizes += BASIS_SIZE * [(OUT_SIZES[0], 1), (1,)]
for weight, esize in zip(nn.weights, expected_sizes):
assert weight.shape == esize