Loading deepchem/models/losses.py +25 −0 Original line number Diff line number Diff line Loading @@ -94,6 +94,31 @@ class HingeLoss(Loss): return loss class SquaredHingeLoss(Loss): """The Squared Hinge loss function. Defined as the square of the hinge loss between y_true and y_pred. The Squared Hinge Loss is differentiable. """ def _compute_tf_loss(self, output, labels): import tensorflow as tf output, labels = _make_tf_shapes_consistent(output, labels) return tf.keras.losses.SquaredHinge(reduction='none')(labels, output) def _create_pytorch_loss(self): import torch def loss(output, labels): output, labels = _make_pytorch_shapes_consistent(output, labels) return torch.mean( torch.pow( torch.maximum(1 - torch.multiply(labels, output), torch.tensor(0)), 2), dim=-1) return loss class PoissonLoss(Loss): """The Poisson loss function is defined as the mean of the elements of y_pred - (y_true * log(y_pred) for an input of (y_true, y_pred). Poisson loss is generally used for regression tasks where the data follows the poisson Loading deepchem/models/tests/test_losses.py +19 −0 Original line number Diff line number Diff line Loading @@ -99,6 +99,25 @@ class TestLosses(unittest.TestCase): assert np.allclose(expected, result) @unittest.skipIf(not has_tensorflow, 'TensorFlow is not installed') def test_squared_hinge_loss_tf(self): """Test SquaredHingeLoss.""" loss = losses.SquaredHingeLoss() outputs = tf.constant([[0.1, 0.8], [0.4, 0.6]]) labels = tf.constant([[1.0, -1.0], [-1.0, 1.0]]) result = loss._compute_tf_loss(outputs, labels).numpy() expected = [np.mean([0.8100, 3.2400]), np.mean([1.9600, 0.1600])] assert np.allclose(expected, result) @unittest.skipIf(not has_pytorch, 'PyTorch is not installed') def test_squared_hinge_loss_pytorch(self): """Test SquaredHingeLoss.""" loss = losses.SquaredHingeLoss() outputs = torch.tensor([[0.1, 0.8], [0.4, 0.6]]) labels = torch.tensor([[1.0, -1.0], [-1.0, 1.0]]) result = loss._create_pytorch_loss()(outputs, labels).numpy() expected = [np.mean([0.8100, 3.2400]), np.mean([1.9600, 0.1600])] assert np.allclose(expected, result) def test_poisson_loss_tf(self): """Test PoissonLoss.""" loss = losses.PoissonLoss() Loading docs/source/api_reference/models.rst +3 −0 Original line number Diff line number Diff line Loading @@ -196,6 +196,9 @@ Losses .. autoclass:: deepchem.models.losses.HingeLoss :members: .. autoclass:: deepchem.models.losses.SquaredHingeLoss :members: .. autoclass:: deepchem.models.losses.PoissonLoss :members: Loading env.common.yml +1 −1 Original line number Diff line number Diff line Loading @@ -12,7 +12,7 @@ dependencies: - pip==20.2.* - pip: - biopython - dgllife==0.2.* - dgllife==0.2.8 - lightgbm==3.* - matminer - mordred Loading env.cpu.mac.yml +1 −1 Original line number Diff line number Diff line Loading @@ -2,7 +2,7 @@ dependencies: - pip: - -f https://pytorch-geometric.com/whl/torch-1.6.0.html - -f https://pytorch-geometric.com/whl/torch-1.6.0+cpu.html - dgl==0.5.* - dgl==0.6.* - torch==1.6.0 - torchvision==0.7.0 - torch-cluster Loading Loading
deepchem/models/losses.py +25 −0 Original line number Diff line number Diff line Loading @@ -94,6 +94,31 @@ class HingeLoss(Loss): return loss class SquaredHingeLoss(Loss): """The Squared Hinge loss function. Defined as the square of the hinge loss between y_true and y_pred. The Squared Hinge Loss is differentiable. """ def _compute_tf_loss(self, output, labels): import tensorflow as tf output, labels = _make_tf_shapes_consistent(output, labels) return tf.keras.losses.SquaredHinge(reduction='none')(labels, output) def _create_pytorch_loss(self): import torch def loss(output, labels): output, labels = _make_pytorch_shapes_consistent(output, labels) return torch.mean( torch.pow( torch.maximum(1 - torch.multiply(labels, output), torch.tensor(0)), 2), dim=-1) return loss class PoissonLoss(Loss): """The Poisson loss function is defined as the mean of the elements of y_pred - (y_true * log(y_pred) for an input of (y_true, y_pred). Poisson loss is generally used for regression tasks where the data follows the poisson Loading
deepchem/models/tests/test_losses.py +19 −0 Original line number Diff line number Diff line Loading @@ -99,6 +99,25 @@ class TestLosses(unittest.TestCase): assert np.allclose(expected, result) @unittest.skipIf(not has_tensorflow, 'TensorFlow is not installed') def test_squared_hinge_loss_tf(self): """Test SquaredHingeLoss.""" loss = losses.SquaredHingeLoss() outputs = tf.constant([[0.1, 0.8], [0.4, 0.6]]) labels = tf.constant([[1.0, -1.0], [-1.0, 1.0]]) result = loss._compute_tf_loss(outputs, labels).numpy() expected = [np.mean([0.8100, 3.2400]), np.mean([1.9600, 0.1600])] assert np.allclose(expected, result) @unittest.skipIf(not has_pytorch, 'PyTorch is not installed') def test_squared_hinge_loss_pytorch(self): """Test SquaredHingeLoss.""" loss = losses.SquaredHingeLoss() outputs = torch.tensor([[0.1, 0.8], [0.4, 0.6]]) labels = torch.tensor([[1.0, -1.0], [-1.0, 1.0]]) result = loss._create_pytorch_loss()(outputs, labels).numpy() expected = [np.mean([0.8100, 3.2400]), np.mean([1.9600, 0.1600])] assert np.allclose(expected, result) def test_poisson_loss_tf(self): """Test PoissonLoss.""" loss = losses.PoissonLoss() Loading
docs/source/api_reference/models.rst +3 −0 Original line number Diff line number Diff line Loading @@ -196,6 +196,9 @@ Losses .. autoclass:: deepchem.models.losses.HingeLoss :members: .. autoclass:: deepchem.models.losses.SquaredHingeLoss :members: .. autoclass:: deepchem.models.losses.PoissonLoss :members: Loading
env.common.yml +1 −1 Original line number Diff line number Diff line Loading @@ -12,7 +12,7 @@ dependencies: - pip==20.2.* - pip: - biopython - dgllife==0.2.* - dgllife==0.2.8 - lightgbm==3.* - matminer - mordred Loading
env.cpu.mac.yml +1 −1 Original line number Diff line number Diff line Loading @@ -2,7 +2,7 @@ dependencies: - pip: - -f https://pytorch-geometric.com/whl/torch-1.6.0.html - -f https://pytorch-geometric.com/whl/torch-1.6.0+cpu.html - dgl==0.5.* - dgl==0.6.* - torch==1.6.0 - torchvision==0.7.0 - torch-cluster Loading
