Merge pull request #1176 from peastman/eager (adfbb040) · Commits · 钟慕尧 / deepchem

deepchem/models/tensorgraph/initializations.py

+5 −6

Original line number	Diff line number	Diff line
		@@ -4,8 +4,7 @@ from __future__ import unicode_literals

		import numpy as np
		import tensorflow as tf
		from deepchem.models.tensorgraph.model_ops import random_uniform_variable
		from deepchem.models.tensorgraph.model_ops import random_normal_variable
		from deepchem.models.tensorgraph.model_ops import random_uniform_variable, random_normal_variable, create_variable
		from deepchem.models.tensorgraph.activations import get_from_module


		@@ -94,7 +93,7 @@ def orthogonal(shape, scale=1.1, name=None):
		# Pick the one with the correct shape.
		q = u if u.shape == flat_shape else v
		q = q.reshape(shape)
		return tf.Variable(
		return create_variable(
		scale * q[:shape[0], :shape[1]], dtype=tf.float32, name=name)


		@@ -103,16 +102,16 @@ def identity(shape, scale=1, name=None):
		raise ValueError('Identity matrix initialization can only be used '
		'for 2D square matrices.')
		else:
		return tf.Variable(
		return create_variable(
		scale * np.identity(shape[0]), dtype=tf.float32, name=name)


		def zero(shape, name=None):
		return tf.Variable(tf.zeros(shape), dtype=tf.float32, name=name)
		return create_variable(tf.zeros(shape), dtype=tf.float32, name=name)


		def one(shape, name=None):
		return tf.Variable(tf.ones(shape), dtype=tf.float32, name=name)
		return create_variable(tf.ones(shape), dtype=tf.float32, name=name)


		def get(identifier, **kwargs):

deepchem/models/tensorgraph/layers.py

+502 −189

File changed.

Preview size limit exceeded, changes collapsed.

deepchem/models/tensorgraph/model_ops.py

+31 −22

Original line number	Diff line number	Diff line
		@@ -11,6 +11,7 @@ import sys
		import traceback
		import numpy as np
		import tensorflow as tf
		import tensorflow.contrib.eager as tfe
		from tensorflow.python.training import moving_averages
		from collections import defaultdict

		@@ -22,28 +23,36 @@ def _to_tensor(x, dtype):
		return x


		def create_variable(value, dtype=None, name=None):
		"""Create a tf.Variable or tfe.Variable, depending on the current mode."""
		if tfe.in_eager_mode():
		return tfe.Variable(value, dtype=dtype, name=name)
		else:
		return tf.Variable(value, dtype=dtype, name=name)


		def ones(shape, dtype=None, name=None):
		"""Instantiates an all-ones tensor variable and returns it.

		Parameters
		----------
		shape: Tuple of integers, shape of returned Keras variable.
		shape: Tuple of integers, shape of returned Tensorflow variable.
		dtype: Tensorflow dtype
		name: String, name of returned Keras variable.
		name: String, name of returned Tensorflow variable.

		Returns
		-------
		A Keras variable, filled with `1.0`.
		A Tensorflow variable, filled with `1.0`.
		"""
		if dtype is None:
		dtype = tf.float32
		shape = tuple(map(int, shape))
		return tf.Variable(
		return create_variable(
		tf.constant_initializer(1., dtype=dtype)(shape), dtype, name)


		def cast_to_floatx(x):
		"""Cast a Numpy array to the default Keras float type.
		"""Cast a Numpy array to the default Tensorflow float type.

		Parameters
		----------
		@@ -65,7 +74,7 @@ def moving_average_update(variable, value, momentum):


		def int_shape(x):
		"""Returns the shape of a Keras tensor or a Keras variable as a tuple of
		"""Returns the shape of a Tensorflow tensor or a Tensorflow variable as a tuple of
		integers or None entries.

		Arguments
		@@ -193,7 +202,7 @@ def get_ndim(x):


		def get_dtype(x):
		"""Returns the dtype of a Keras tensor or variable, as a string.
		"""Returns the dtype of a Tensorflow tensor or variable, as a string.

		Parameters
		----------
		@@ -259,7 +268,7 @@ def random_uniform_variable(shape,
		seed = np.random.randint(10e8)
		value = tf.random_uniform_initializer(
		low, high, dtype=dtype, seed=seed)(shape)
		return tf.Variable(value, dtype=dtype, name=name)
		return create_variable(value, dtype=dtype, name=name)


		def random_normal_variable(shape,
		@@ -268,16 +277,16 @@ def random_normal_variable(shape,
		dtype=tf.float32,
		name=None,
		seed=None):
		"""Instantiates an Keras variable filled with
		"""Instantiates an Tensorflow variable filled with
		samples drawn from a normal distribution and returns it.

		Parameters
		----------
		shape: Tuple of integers, shape of returned Keras variable.
		shape: Tuple of integers, shape of returned Tensorflow variable.
		mean: Float, mean of the normal distribution.
		scale: Float, standard deviation of the normal distribution.
		dtype: Tensorflow dtype
		name: String, name of returned Keras variable.
		name: String, name of returned Tensorflow variable.
		seed: Integer, random seed.

		Returns
		@@ -290,7 +299,7 @@ def random_normal_variable(shape,
		seed = np.random.randint(10e8)
		value = tf.random_normal_initializer(
		mean, scale, dtype=dtype, seed=seed)(shape)
		return tf.Variable(value, dtype=dtype, name=name)
		return create_variable(value, dtype=dtype, name=name)


		def max(x, axis=None, keepdims=False):
		@@ -338,7 +347,7 @@ def categorical_crossentropy(output, target, from_logits=False):
		# TODO(rbharath): Should probably swap this over to tf mode.
		"""
		# Note: tf.nn.softmax_cross_entropy_with_logits
		# expects logits, Keras expects probabilities.
		# expects logits, Tensorflow expects probabilities.
		if not from_logits:
		# scale preds so that the class probas of each sample sum to 1
		output /= tf.reduce_sum(
		@@ -362,7 +371,7 @@ def sparse_categorical_crossentropy(output, target, from_logits=False):
		and a target tensor, where the target is an integer tensor.
		"""
		# Note: tf.nn.softmax_cross_entropy_with_logits
		# expects logits, Keras expects probabilities.
		# expects logits, Tensorflow expects probabilities.
		if not from_logits:
		epsilon = _to_tensor(_EPSILON, output.dtype.base_dtype)
		output = tf.clip_by_value(output, epsilon, 1 - epsilon)
		@@ -398,7 +407,7 @@ def binary_crossentropy(output, target, from_logits=False):
		A tensor.
		"""
		# Note: tf.nn.softmax_cross_entropy_with_logits
		# expects logits, Keras expects probabilities.
		# expects logits, Tensorflow expects probabilities.
		if not from_logits:
		# transform back to logits
		epsilon = _to_tensor(_EPSILON, output.dtype.base_dtype)
		@@ -437,16 +446,16 @@ def zeros(shape, dtype=tf.float32, name=None):

		Parameters
		----------
		shape: Tuple of integers, shape of returned Keras variable
		shape: Tuple of integers, shape of returned Tensorflow variable
		dtype: Tensorflow dtype
		name: String, name of returned Keras variable
		name: String, name of returned Tensorflow variable

		Returns
		-------
		A variable (including Keras metadata), filled with `0.0`.
		A variable (including Tensorflow metadata), filled with `0.0`.
		"""
		shape = tuple(map(int, shape))
		return tf.Variable(
		return create_variable(
		tf.constant_initializer(0., dtype=dtype)(shape), dtype, name)


		@@ -674,7 +683,7 @@ def add_bias(tensor, init=None, name=None):
		if init is None:
		init = tf.zeros([tensor.get_shape()[-1].value])
		with tf.name_scope(name, tensor.op.name, [tensor]):
		b = tf.Variable(init, name='b')
		b = create_variable(init, name='b')
		return tf.nn.bias_add(tensor, b)


		@@ -752,8 +761,8 @@ def fully_connected_layer(tensor,
		bias_init = tf.zeros([size])

		with tf.name_scope(name, 'fully_connected', [tensor]):
		w = tf.Variable(weight_init, name='w', dtype=tf.float32)
		b = tf.Variable(bias_init, name='b', dtype=tf.float32)
		w = create_variable(weight_init, name='w', dtype=tf.float32)
		b = create_variable(bias_init, name='b', dtype=tf.float32)
		return tf.nn.xw_plus_b(tensor, w, b)

deepchem/models/tensorgraph/models/graph_models.py

+3 −3

Original line number	Diff line number	Diff line
		@@ -14,7 +14,7 @@ from deepchem.models.tensorgraph.graph_layers import WeaveGather, \
		from deepchem.models.tensorgraph.graph_layers import WeaveLayerFactory
		from deepchem.models.tensorgraph.layers import Dense, SoftMax, \
		SoftMaxCrossEntropy, GraphConv, BatchNorm, \
		GraphPool, GraphGather, WeightedError, Dropout, BatchNormalization, Stack, Flatten, GraphCNN, GraphCNNPool
		GraphPool, GraphGather, WeightedError, Dropout, BatchNorm, Stack, Flatten, GraphCNN, GraphCNNPool
		from deepchem.models.tensorgraph.layers import L2Loss, Label, Weights, Feature
		from deepchem.models.tensorgraph.tensor_graph import TensorGraph
		from deepchem.trans import undo_transforms
		@@ -86,7 +86,7 @@ class WeaveModel(TensorGraph):
		out_channels=self.n_graph_feat,
		activation_fn=tf.nn.tanh,
		in_layers=weave_layer2A)
		batch_norm1 = BatchNormalization(epsilon=1e-5, mode=1, in_layers=[dense1])
		batch_norm1 = BatchNorm(epsilon=1e-5, in_layers=[dense1])
		weave_gather = WeaveGather(
		self.batch_size,
		n_input=self.n_graph_feat,

deepchem/models/tensorgraph/models/text_cnn.py

+4 −4

Original line number	Diff line number	Diff line
		@@ -9,7 +9,7 @@ import copy

		from deepchem.metrics import to_one_hot, from_one_hot
		from deepchem.models.tensorgraph.layers import Dense, Concat, SoftMax, \
		SoftMaxCrossEntropy, BatchNorm, WeightedError, Dropout, BatchNormalization, \
		SoftMaxCrossEntropy, BatchNorm, WeightedError, Dropout, \
		Conv1D, ReduceMax, Squeeze, Stack, Highway
		from deepchem.models.tensorgraph.graph_layers import DTNNEmbedding

Admin message