Sequential (ab5a09a1) · Commits · 钟慕尧 / deepchem

deepchem/models/init.py

+1 −0

Original line number	Diff line number	Diff line
		@@ -34,3 +34,4 @@ from deepchem.models.tensorgraph.models.symmetry_function_regression import BPSy
		from deepchem.models.tensorgraph.models.seqtoseq import SeqToSeq
		from deepchem.models.tensorgraph.models.gan import GAN, WGAN
		from deepchem.models.tensorgraph.models.text_cnn import TextCNNTensorGraph
		from deepchem.models.tensorgraph.sequential import Sequential

+1 −1

Original line number	Diff line number	Diff line
		@@ -997,7 +997,7 @@ class SoftMax(Layer):
		def create_tensor(self, in_layers=None, set_tensors=True, **kwargs):
		inputs = self._get_input_tensors(in_layers)
		if len(inputs) != 1:
		raise ValueError("Must only Softmax single parent")
		raise ValueError("Softmax must have a single input layer.")
		parent = inputs[0]
		out_tensor = tf.contrib.layers.softmax(parent)
		if set_tensors:

0 → 100644

+25 −0

Original line number	Diff line number	Diff line
		import unittest
		import numpy as np
		import deepchem as dc
		from deepchem.models.tensorgraph.layers import Dense
		from deepchem.models.tensorgraph.layers import SoftMax
		from nose.tools import assert_true


		class TestSequential(unittest.TestCase):
		"""
		Test that sequential graphs work correctly.
		"""

		def test_single_task_classifier(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())
		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)))

+45 −2

Original line number	Diff line number	Diff line
		"""
		Activations for models.

		Copied over from Keras.
		"""
		from __future__ import print_function
		from __future__ import division
		@@ -87,22 +85,67 @@ def softsign(x):


		def relu(x, alpha=0., max_value=None):
		"""The rectified linear activation function

		Wrapper around model_ops.relu.

		Parameters
		----------
		x: tf.Tensor
		Input tensor
		"""
		return model_ops.relu(x, alpha=alpha, max_value=max_value)


		def tanh(x):
		"""The hyperbolic tanget activation function

		Wrapper around tf.nn.tanh.

		Parameters
		----------
		x: tf.Tensor
		Input tensor
		"""
		return tf.nn.tanh(x)


		def sigmoid(x):
		"""The sigmoidal activation function

		Wrapper around tf.nn.sigmoid.

		Parameters
		----------
		x: tf.Tensor
		Input tensor
		"""
		return tf.nn.sigmoid(x)


		def hard_sigmoid(x):
		"""The hard sigmoidal activation function

		Piecewise-linear approximation to sigmoid.

		Parameters
		----------
		x: tf.Tensor
		Input tensor
		"""
		return model_ops.hard_sigmoid(x)


		def linear(x):
		"""A linear activation function.

		Note that a linear activation function is simply the identity.

		Parameters
		----------
		x: tf.Tensor
		Input tensor
		"""
		return x

+2 −2

Original line number	Diff line number	Diff line
		@@ -690,8 +690,8 @@ def selu(x):

		def hard_sigmoid(x):
		"""Segment-wise linear approximation of sigmoid.
		Faster than sigmoid.
		Returns 0. if x < -2.5, 1. if x > 2.5.

		Faster than sigmoid. Returns 0. if x < -2.5, 1. if x > 2.5.
		In -2.5 <= x <= 2.5, returns 0.2 * x + 0.5.

		Parameters