Loading deepchem/models/tensorgraph/sequential.py +7 −3 Original line number Diff line number Diff line Loading @@ -29,7 +29,8 @@ class Sequential(TensorGraph): if kwargs["use_queue"]: raise ValueError("Sequential doesn't support queues.") kwargs["use_queue"] = False self.layer_list = [] self._layer_list = [] self._built = False super(Sequential, self).__init__(**kwargs) def add(self, layer): Loading @@ -40,7 +41,7 @@ class Sequential(TensorGraph): layer: Layer Adds layer to this graph. """ self.layer_list.append(layer) self._layer_list.append(layer) def fit(self, dataset, loss, **kwargs): """Fits on the specified dataset. Loading @@ -55,6 +56,8 @@ class Sequential(TensorGraph): Only "binary_crossentropy" for now. """ X_shape, y_shape, _, _ = dataset.get_shape() # Calling fit() for first time if not self._built: feature_shape = X_shape[1:] label_shape = y_shape[1:] # Add in features Loading @@ -66,7 +69,7 @@ class Sequential(TensorGraph): # Add in all layers prev_layer = features for ind, layer in enumerate(self.layer_list): for ind, layer in enumerate(self._layer_list): if not len(layer.in_layers) == 0: raise ValueError("Cannot specify in_layers for Sequential.") layer.in_layers += [prev_layer] Loading @@ -82,5 +85,6 @@ class Sequential(TensorGraph): else: # TODO(rbharath): Add in support for additional losses. raise ValueError("Unsupported loss.") self._built = True super(Sequential, self).fit(dataset, **kwargs) deepchem/models/tensorgraph/tests/test_sequential.py +13 −0 Original line number Diff line number Diff line Loading @@ -23,3 +23,16 @@ class TestSequential(unittest.TestCase): model.fit(dataset, loss="binary_crossentropy", nb_epoch=1000) prediction = np.squeeze(model.predict_on_batch(X)) assert_true(np.all(np.isclose(prediction, y, atol=0.4))) def test_fit_twice(self): n_data_points = 20 n_features = 2 X = np.random.rand(n_data_points, n_features) y = [[0, 1] for x in range(n_data_points)] dataset = dc.data.NumpyDataset(X, y) model = dc.models.Sequential(learning_rate=0.01) model.add(Dense(out_channels=2)) model.add(SoftMax()) # Should be able to call fit twice without failure. model.fit(dataset, loss="binary_crossentropy", nb_epoch=1000) model.fit(dataset, loss="binary_crossentropy", nb_epoch=1000) Loading
deepchem/models/tensorgraph/sequential.py +7 −3 Original line number Diff line number Diff line Loading @@ -29,7 +29,8 @@ class Sequential(TensorGraph): if kwargs["use_queue"]: raise ValueError("Sequential doesn't support queues.") kwargs["use_queue"] = False self.layer_list = [] self._layer_list = [] self._built = False super(Sequential, self).__init__(**kwargs) def add(self, layer): Loading @@ -40,7 +41,7 @@ class Sequential(TensorGraph): layer: Layer Adds layer to this graph. """ self.layer_list.append(layer) self._layer_list.append(layer) def fit(self, dataset, loss, **kwargs): """Fits on the specified dataset. Loading @@ -55,6 +56,8 @@ class Sequential(TensorGraph): Only "binary_crossentropy" for now. """ X_shape, y_shape, _, _ = dataset.get_shape() # Calling fit() for first time if not self._built: feature_shape = X_shape[1:] label_shape = y_shape[1:] # Add in features Loading @@ -66,7 +69,7 @@ class Sequential(TensorGraph): # Add in all layers prev_layer = features for ind, layer in enumerate(self.layer_list): for ind, layer in enumerate(self._layer_list): if not len(layer.in_layers) == 0: raise ValueError("Cannot specify in_layers for Sequential.") layer.in_layers += [prev_layer] Loading @@ -82,5 +85,6 @@ class Sequential(TensorGraph): else: # TODO(rbharath): Add in support for additional losses. raise ValueError("Unsupported loss.") self._built = True super(Sequential, self).fit(dataset, **kwargs)
deepchem/models/tensorgraph/tests/test_sequential.py +13 −0 Original line number Diff line number Diff line Loading @@ -23,3 +23,16 @@ class TestSequential(unittest.TestCase): model.fit(dataset, loss="binary_crossentropy", nb_epoch=1000) prediction = np.squeeze(model.predict_on_batch(X)) assert_true(np.all(np.isclose(prediction, y, atol=0.4))) def test_fit_twice(self): n_data_points = 20 n_features = 2 X = np.random.rand(n_data_points, n_features) y = [[0, 1] for x in range(n_data_points)] dataset = dc.data.NumpyDataset(X, y) model = dc.models.Sequential(learning_rate=0.01) model.add(Dense(out_channels=2)) model.add(SoftMax()) # Should be able to call fit twice without failure. model.fit(dataset, loss="binary_crossentropy", nb_epoch=1000) model.fit(dataset, loss="binary_crossentropy", nb_epoch=1000)
