def test_pytorch_module(self):
asset = Linear(4, 2)
path = Path(self.tempdir.name) / 'pytorch-module'
kind, meta = serialize(asset, path)
deserialized = deserialize(kind, path, meta)
self.assertEqual(deserialized.in_features, asset.in_features)
self.assertEqual(deserialized.out_features, asset.out_features)
def test_annoy_index(self):
asset = AnnoyIndex(8)
asset.build(4)
path = Path(self.tempdir.name) / 'annoy-index'
kind, meta = serialize(asset, path)
deserialized = deserialize(kind, path, meta)
self.assertEqual(deserialized.f, asset.f)
def pull_asset(self, experiment_name, name, asset):
namespaced_asset_dir = Path(self.pulled_asset_dir) / experiment_name
namespaced_asset_dir.mkdir(parents=True, exist_ok=True)
filename = str(namespaced_asset_dir / name)
key = self._generate_key(experiment_name, 'assets', name)
self._pull_object(key, filename)
return deserialize(asset['kind'], filename, asset['meta'])
def test_pandas_data_frame(self):
asset = pd.DataFrame({'a': [1]})
path = str(Path(self.tempdir.name) / 'pandas-data-frame')
kind, meta = serialize(asset, path)
deserialized = deserialize(kind, path, meta)
self.assertEqual(deserialized['a'].iloc[0], 1)
def test_sklearn_estimator(self):
asset = RandomForestClassifier(8)
path = Path(self.tempdir.name) / 'sklearn-estimator'
kind, meta = serialize(asset, path)
deserialized = deserialize(kind, path, meta)
self.assertEqual(deserialized.n_estimators, asset.n_estimators)
def test_pytorch_tensor(self):
asset = tensor([[1., -1.], [1., -1.]])
path = Path(self.tempdir.name) / 'pytorch-tensor'
kind, meta = serialize(asset, path)
deserialized = deserialize(kind, path, meta)
self.assertTrue(equal(deserialized, asset))
def test_numpy_array(self):
asset = np.array([1, 2])
path = Path(self.tempdir.name) / 'numpy-array'
kind, meta = serialize(asset, path)
deserialized = deserialize(kind, path, meta)
self.assertEqual(deserialized.all(), asset.all())
def test_python_tuple(self):
asset = (1, 2)
path = Path(self.tempdir.name) / 'python-tuple'
kind, meta = serialize(asset, path)
deserialized = deserialize(kind, path, meta)
self.assertEqual(deserialized, asset)
def test_python_set(self):
asset = {1, 2}
path = Path(self.tempdir.name) / 'python-set'
kind, meta = serialize(asset, path)
deserialized = deserialize(kind, path, meta)
self.assertEqual(deserialized, asset)
def test_python_dict(self):
asset = {'key': 'value'}
path = Path(self.tempdir.name) / 'python-dict'
kind, meta = serialize(asset, path)
deserialized = deserialize(kind, path, meta)
self.assertEqual(deserialized, asset)
def test_python_str(self):
asset = 'value'
path = Path(self.tempdir.name) / 'python-str'
kind, meta = serialize(asset, path)
deserialized = deserialize(kind, path, meta)
self.assertEqual(deserialized, asset)
def test_unknown_deserialize(self):
path = Path(self.tempdir.name) / 'unknown'
with self.assertRaises(SerializerNotFoundError):
deserialize('unknown/unknown', path, None)
def test_pandas_series(self):
asset = pd.Series([1])
path = str(Path(self.tempdir.name) / 'pandas-series')
kind, meta = serialize(asset, path)
deserialized = deserialize(kind, path, meta)
self.assertEqual(deserialized.iloc[0], 1)