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'])