Some progress in refactor (6387e4c4) · Commits · 钟慕尧 / deepchem

deepchem/models/tf_new_models/tests/test_graph_models.py

+5 −4

Original line number	Diff line number	Diff line
		@@ -49,13 +49,14 @@ class TestGraphModels(test_util.TensorFlowTestCase):
		graph_model.add(BatchNormalization(epsilon=1e-5, mode=1))
		graph_model.add(GraphPool())

		# Gather Projection
		graph_model.add(Dense(128, activation='relu'))
		## Gather Projection
		#graph_model.add(Dense(128, activation='relu'))
		graph_model.add(BatchNormalization(epsilon=1e-5, mode=1))
		graph_model.add(GraphGather(batch_size, activation="tanh"))

		# There should be 8 layers in graph_model
		assert len(graph_model.layers) == 6
		#assert len(graph_model.layers) == 6
		assert len(graph_model.layers) == 5

		def test_sample_attn_lstm_architecture(self):
		"""Tests that an attention architecture can be created without crash."""
		@@ -83,7 +84,7 @@ class TestGraphModels(test_util.TensorFlowTestCase):
		support_model.join(AttnLSTMEmbedding(n_test, n_support, max_depth))

		# Gather Projection
		support_model.add(Dense(128, activation='relu'))
		#support_model.add(Dense(128, activation='relu'))
		support_model.add_test(BatchNormalization(epsilon=1e-5, mode=1))
		support_model.add_support(BatchNormalization(epsilon=1e-5, mode=1))
		support_model.add(GraphGather(batch_size, activation="tanh"))

deepchem/nn/activations.py

+1 −2

Original line number	Diff line number	Diff line
		#from __future__ import absolute_import
		from keras import backend as K
		#from .utils.generic_utils import get_from_module

		def get_from_module(identifier, module_params, module_name,
		instantiate=False, kwargs=None):
		@@ -25,7 +24,7 @@ def get_from_module(identifier, module_params, module_name,
		# Raises
		ValueError: if the identifier cannot be found.
		"""
		if isinstance(identifier, str):
		if isinstance(identifier, str) or isinstance(identifier, unicode):
		res = module_params.get(identifier)
		if not res:
		raise ValueError('Invalid ' + str(module_name) + ': ' +

deepchem/nn/constraints.py

0 → 100644

+103 −0

Original line number	Diff line number	Diff line
		from __future__ import absolute_import
		from keras import backend as K
		from .activations import get_from_module


		class Constraint(object):

		def __call__(self, p):
		return p

		def get_config(self):
		return {'name': self.__class__.__name__}


		class MaxNorm(Constraint):
		"""MaxNorm weight constraint.

		Constrains the weights incident to each hidden unit
		to have a norm less than or equal to a desired value.

		# Arguments
		m: the maximum norm for the incoming weights.
		axis: integer, axis along which to calculate weight norms.
		For instance, in a `Dense` layer the weight matrix
		has shape `(input_dim, output_dim)`,
		set `axis` to `0` to constrain each weight vector
		of length `(input_dim,)`.
		In a `Convolution2D` layer with `dim_ordering="tf"`,
		the weight tensor has shape
		`(rows, cols, input_depth, output_depth)`,
		set `axis` to `[0, 1, 2]`
		to constrain the weights of each filter tensor of size
		`(rows, cols, input_depth)`.

		# References
		- [Dropout: A Simple Way to Prevent Neural Networks from Overfitting Srivastava, Hinton, et al. 2014](http://www.cs.toronto.edu/~rsalakhu/papers/srivastava14a.pdf)
		"""

		def __init__(self, m=2, axis=0):
		self.m = m
		self.axis = axis

		def __call__(self, p):
		norms = K.sqrt(K.sum(K.square(p), axis=self.axis, keepdims=True))
		desired = K.clip(norms, 0, self.m)
		p *= (desired / (K.epsilon() + norms))
		return p

		def get_config(self):
		return {'name': self.__class__.__name__,
		'm': self.m,
		'axis': self.axis}


		class NonNeg(Constraint):
		"""Constrains the weights to be non-negative.
		"""

		def __call__(self, p):
		p *= K.cast(p >= 0., K.floatx())
		return p


		class UnitNorm(Constraint):
		"""Constrains the weights incident to each hidden unit to have unit norm.

		# Arguments
		axis: integer, axis along which to calculate weight norms.
		For instance, in a `Dense` layer the weight matrix
		has shape `(input_dim, output_dim)`,
		set `axis` to `0` to constrain each weight vector
		of length `(input_dim,)`.
		In a `Convolution2D` layer with `dim_ordering="tf"`,
		the weight tensor has shape
		`(rows, cols, input_depth, output_depth)`,
		set `axis` to `[0, 1, 2]`
		to constrain the weights of each filter tensor of size
		`(rows, cols, input_depth)`.
		"""

		def __init__(self, axis=0):
		self.axis = axis

		def __call__(self, p):
		return p / (K.epsilon() + K.sqrt(K.sum(K.square(p),
		axis=self.axis,
		keepdims=True)))

		def get_config(self):
		return {'name': self.__class__.__name__,
		'axis': self.axis}


		# Aliases.

		maxnorm = MaxNorm
		nonneg = NonNeg
		unitnorm = UnitNorm


		def get(identifier, kwargs=None):
		return get_from_module(identifier, globals(), 'constraint',
		instantiate=True, kwargs=kwargs)

deepchem/nn/copy.py

+4 −0

Original line number	Diff line number	Diff line
		@@ -9,6 +9,10 @@ __author__ = "Bharath Ramsundar"
		__copyright__ = "Copyright 2016, Stanford University"
		__license__ = "GPL"

		from . import initializations
		from . import regularizers
		from . import activations
		from . import constraints
		from keras import backend as K

		def to_list(x):

deepchem/nn/layers.py

+16 −0

Original line number	Diff line number	Diff line
		@@ -249,6 +249,22 @@ class GraphConv(Layer):
		# Generate the nb_affine weights and biases
		atom_features_shape = input_shape[0]
		n_features = atom_features_shape[1]
		############################################################### DEBUG
		print("self.nb_affine")
		print(self.nb_affine)
		print("type(self.nb_affine)")
		print(type(self.nb_affine))
		print("self.nb_filter, type(self.nb_filter)")
		print(self.nb_filter, type(self.nb_filter))
		print("range(self.nb_affine)")
		print(range(self.nb_affine))
		print("K.zeros(shape=[self.nb_filter,])")
		print(K.zeros(shape=[self.nb_filter,]))
		print("self.init")
		print(self.init)
		print("type(self.init)")
		print(type(self.init))
		############################################################### DEBUG
		self.W_list = [self.init([n_features, self.nb_filter])
		for k in range(self.nb_affine)]
		self.b_list = [K.zeros(shape=[self.nb_filter,])

Admin message