def test_round_split(self):
split = 10
batch_size = 32
ntrain, ntest, nin, nout = 100, 10, 10, 5
data = UniformRandom(ntrain, ntest, nin, nout, validation_pct=split)
data.backend = CPU(rng_seed=0)
data.backend.batch_size = batch_size
data.load()
split /= 100.0
nb_batches = ntrain // batch_size
expected_nb_train = floor((1.0 - split) * nb_batches)
expected_nb_valid = floor(split * nb_batches)
assert expected_nb_train == len(data.inputs['train'])
assert expected_nb_train == len(data.targets['train'])
assert expected_nb_valid == len(data.inputs['validation'])
assert expected_nb_valid == len(data.targets['validation'])
def test_round_split(self):
split = 10
batch_size = 32
ntrain, ntest, nin, nout = 100, 10, 10, 5
data = UniformRandom(ntrain, ntest, nin, nout, validation_pct=split)
data.backend = CPU(rng_seed=0)
data.backend.batch_size = batch_size
data.load()
split /= 100.0
nb_batches = ntrain // batch_size
expected_nb_train = floor((1.0 - split) * nb_batches)
expected_nb_valid = floor(split * nb_batches)
assert expected_nb_train == len(data.inputs['train'])
assert expected_nb_train == len(data.targets['train'])
assert expected_nb_valid == len(data.inputs['validation'])
assert expected_nb_valid == len(data.targets['validation'])
