def _get_fully_connected(cfg):
"""Get a fully connected problem from the given config."""
return (problem_spec.Spec(
pg.FullyConnected, (cfg["n_features"], cfg["n_classes"]), {
"hidden_sizes": tuple(cfg["hidden_sizes"]),
"activation": utils.get_activation(cfg["activation"]),
}), losg_datasets.random_mlp(cfg["n_features"],
cfg["n_samples"]), cfg["bs"])
def testRandomMlp(self):
n_samples = 4
n_features = 3
n_layers = 2
width = 10
ds = datasets.random_mlp(
n_features, n_samples, n_layers=n_layers, width=width, random_seed=200)
self.assertEqual(n_samples, ds.size)
self.assertLen(ds.labels, n_samples)
self.assertLen(ds.data[0], n_features)
self.assertTrue(all([x in [0, 1] for x in ds.labels])) # binary
def _get_outward_snake(cfg):
"""Get an outward snake problem from the given config."""
return (problem_spec.Spec(pg.OutwardSnake, (cfg["dim"], ), {}),
losg_datasets.random_mlp(cfg["dim"], cfg["n_samples"]), cfg["bs"])
def _get_dependency_chain(cfg):
"""Get a dependency chain problem from the given config."""
return (problem_spec.Spec(pg.DependencyChain, (cfg["dim"], ), {}),
losg_datasets.random_mlp(cfg["dim"], cfg["n_samples"]), cfg["bs"])