Merge pull request #2071 from peastman/torchmodel (ee6954ca) · Commits · 钟慕尧 / deepchem

deepchem/metalearning/maml.py

+2 −2

Original line number	Diff line number	Diff line
		@@ -131,9 +131,9 @@ class MAML(object):
		# Create the optimizers for meta-optimization and task optimization.

		self._global_step = tf.Variable(0, trainable=False)
		self._tf_optimizer = optimizer._create_optimizer(self._global_step)
		self._tf_optimizer = optimizer._create_tf_optimizer(self._global_step)
		task_optimizer = GradientDescent(learning_rate=self.learning_rate)
		self._tf_task_optimizer = task_optimizer._create_optimizer(
		self._tf_task_optimizer = task_optimizer._create_tf_optimizer(
		self._global_step)

		# Create a Checkpoint for saving.

deepchem/models/init.py

+5 −0

Original line number	Diff line number	Diff line
		@@ -25,6 +25,11 @@ from deepchem.models.text_cnn import TextCNNModel
		from deepchem.models.atomic_conv import AtomicConvModel
		from deepchem.models.chemnet_models import Smiles2Vec, ChemCeption

		try:
		from deepchem.models.torch_model import TorchModel
		except ModuleNotFoundError:
		pass

		#################### Compatibility imports for renamed TensorGraph models. Remove below with DeepChem 3.0. ####################

		from deepchem.models.text_cnn import TextCNNTensorGraph

deepchem/models/fcnet.py

+2 −2

Original line number	Diff line number	Diff line
		@@ -15,7 +15,7 @@ from deepchem.metrics import to_one_hot
		from tensorflow.keras.layers import Input, Dense, Reshape, Softmax, Dropout, Activation, Lambda

		from typing import Any, Callable, Iterable, List, Optional, Sequence, Tuple, Union
		from deepchem.utils.typing import KerasActivationFn, KerasLossFn, OneOrMany
		from deepchem.utils.typing import KerasActivationFn, LossFn, OneOrMany

		logger = logging.getLogger(__name__)

		@@ -299,7 +299,7 @@ class MultitaskRegressor(KerasModel):
		stddev=weight_init_stddevs[-1]),
		bias_initializer=tf.constant_initializer(
		value=bias_init_consts[-1]))(prev_layer))
		loss: Union[dc.models.losses.Loss, KerasLossFn]
		loss: Union[dc.models.losses.Loss, LossFn]
		if uncertainty:
		log_var = Reshape((n_tasks, 1))(Dense(
		n_tasks,

deepchem/models/graph_models.py

+7 −7

Original line number	Diff line number	Diff line
		@@ -5,7 +5,7 @@ import numpy as np
		import tensorflow as tf

		from typing import List, Union, Tuple, Iterable, Dict
		from deepchem.utils.typing import OneOrMany, KerasLossFn, KerasActivationFn
		from deepchem.utils.typing import OneOrMany, LossFn, KerasActivationFn
		from deepchem.data import Dataset, NumpyDataset, pad_features
		from deepchem.feat.graph_features import ConvMolFeaturizer
		from deepchem.feat.mol_graphs import ConvMol
		@@ -901,7 +901,7 @@ class GraphConvModel(KerasModel):
		batch_size=batch_size)
		if mode == "classification":
		output_types = ['prediction', 'loss', 'embedding']
		loss: Union[Loss, KerasLossFn] = SoftmaxCrossEntropy()
		loss: Union[Loss, LossFn] = SoftmaxCrossEntropy()
		else:
		if self.uncertainty:
		output_types = ['prediction', 'variance', 'loss', 'loss', 'embedding']

deepchem/models/keras_model.py

+20 −9

Original line number	Diff line number	Diff line
		@@ -19,7 +19,7 @@ from deepchem.trans import Transformer, undo_transforms
		from deepchem.utils.evaluate import GeneratorEvaluator

		from typing import Any, Callable, Dict, Iterable, List, Optional, Sequence, Tuple, Union
		from deepchem.utils.typing import KerasLossFn, OneOrMany
		from deepchem.utils.typing import LossFn, OneOrMany

		try:
		import wandb
		@@ -57,6 +57,16 @@ class KerasModel(Model):
		Keras Model class, such as uncertainty prediction and
		saliency mapping.

		Here is a simple example of code that uses KerasModel to train
		a Keras model on a DeepChem dataset.

		>> keras_model = tf.keras.Sequential([
		>> tf.keras.layers.Dense(1000, activation='tanh'),
		>> tf.keras.layers.Dense(1)
		>> ])
		>> model = KerasModel(keras_model, loss=dc.models.losses.L2Loss())
		>> model.fit(dataset)

		The loss function for a model can be defined in two different
		ways. For models that have only a single output and use a
		standard loss function, you can simply provide a
		@@ -118,7 +128,7 @@ class KerasModel(Model):

		def __init__(self,
		model: tf.keras.Model,
		loss: Union[Loss, KerasLossFn],
		loss: Union[Loss, LossFn],
		output_types: Optional[List[str]] = None,
		batch_size: int = 100,
		model_dir: Optional[str] = None,
		@@ -157,7 +167,8 @@ class KerasModel(Model):
		log_frequency: int
		The frequency at which to log data. Data is logged using
		`logging` by default. If `tensorboard` is set, data is also
		logged to TensorBoard. Logging happens at global steps. Roughly,
		logged to TensorBoard. If `wandb` is set, data is also logged
		to Weights & Biases. Logging happens at global steps. Roughly,
		a global step corresponds to one batch of training. If you'd
		like a printout every 10 batch steps, you'd set
		`log_frequency=10` for example.
		@@ -166,7 +177,7 @@ class KerasModel(Model):
		model_instance=model, model_dir=model_dir, **kwargs)
		self.model = model
		if isinstance(loss, Loss):
		self._loss_fn: KerasLossFn = _StandardLoss(model, loss)
		self._loss_fn: LossFn = _StandardLoss(model, loss)
		else:
		self._loss_fn = loss
		self.batch_size = batch_size
		@@ -227,7 +238,7 @@ class KerasModel(Model):
		return
		self._built = True
		self._global_step = tf.Variable(0, trainable=False)
		self._tf_optimizer = self.optimizer._create_optimizer(self._global_step)
		self._tf_optimizer = self.optimizer._create_tf_optimizer(self._global_step)
		self._checkpoint = tf.train.Checkpoint(
		optimizer=self._tf_optimizer, model=self.model)

		@@ -271,7 +282,7 @@ class KerasModel(Model):
		deterministic: bool = False,
		restore: bool = False,
		variables: Optional[List[tf.Variable]] = None,
		loss: Optional[KerasLossFn] = None,
		loss: Optional[LossFn] = None,
		callbacks: Union[Callable, List[Callable]] = [],
		all_losses: Optional[List[float]] = None) -> float:
		"""Train this model on a dataset.
		@@ -324,7 +335,7 @@ class KerasModel(Model):
		checkpoint_interval: int = 1000,
		restore: bool = False,
		variables: Optional[List[tf.Variable]] = None,
		loss: Optional[KerasLossFn] = None,
		loss: Optional[LossFn] = None,
		callbacks: Union[Callable, List[Callable]] = [],
		all_losses: Optional[List[float]] = None) -> float:
		"""Train this model on data from a generator.
		@@ -480,7 +491,7 @@ class KerasModel(Model):
		y: Sequence,
		w: Sequence,
		variables: Optional[List[tf.Variable]] = None,
		loss: Optional[KerasLossFn] = None,
		loss: Optional[LossFn] = None,
		callbacks: Union[Callable, List[Callable]] = [],
		checkpoint: bool = True,
		max_checkpoints_to_keep: int = 5) -> float:
		@@ -1203,7 +1214,7 @@ class _StandardLoss(object):
		raise ValueError(
		"Loss functions expects exactly one each of outputs, labels, and weights"
		)
		losses = self.loss(outputs[0], labels[0])
		losses = self.loss._compute_tf_loss(outputs[0], labels[0])
		w = weights[0]
		if len(w.shape) < len(losses.shape):
		if isinstance(w, tf.Tensor):

Admin message