Skip to content

Commit

Permalink
Merge pull request #14 from ragavvenkatesan/test
Browse files Browse the repository at this point in the history 
Test
  • Loading branch information
Ragav Venkatesan committed Feb 13, 2017
2 parents a873c3d + df53ace commit 4d31068
Showing 1 changed file with 40 additions and 37 deletions.
77 changes: 40 additions & 37 deletions tests/test_activations.py
View file
Original file line number Diff line number Diff line change
@@ -1,5 +1,6 @@
"""
test_activations.py - Unit tests for YANN core activation functions
defined in yann/core/activations.py
"""

import unittest
Expand All @@ -11,8 +12,11 @@

class TestActivations(unittest.TestCase):
"""
expected_array values are precomputed using a Python interpreter, numpy_input,
and the corresponding activation function.
expected_array values are precomputed using a Python interpreter,
numpy_input, and the corresponding activation function.
.shape is used to check the dimensionality of the result while
.allclose checks the element-wise equality of the result
"""

def setUp(self):
Expand All @@ -21,19 +25,19 @@ def setUp(self):
different activation functions
"""
self.theano_input = T.matrix()
self.numpy_input = np.array([[-1,2,-3,4,5],
[-1,2,-3,4,5],
[-1,2,-3,4,5],
[-1,2,-3,4,5],
[-1,2,-3,4,5]],
dtype=theano.config.floatX)
self.numpy_input = np.array([[-1, 2, -3, 4, 5],
[-1, 2, -3, 4, 5],
[-1, 2, -3, 4, 5],
[-1, 2, -3, 4, 5],
[-1, 2, -3, 4, 5]],
dtype=theano.config.floatX)

def test_abs(self):
expected_array = np.array([[ 1., 2., 3., 4., 5.],
[ 1., 2., 3., 4., 5.],
[ 1., 2., 3., 4., 5.],
[ 1., 2., 3., 4., 5.],
[ 1., 2., 3., 4., 5.]])
expected_array = np.array([[1., 2., 3., 4., 5.],
[1., 2., 3., 4., 5.],
[1., 2., 3., 4., 5.],
[1., 2., 3., 4., 5.],
[1., 2., 3., 4., 5.]])
theano_result = A.Abs(self.theano_input).eval({self.theano_input: self.numpy_input})
self.assertEqual(theano_result.shape, expected_array.shape)
self.assertTrue(np.allclose(theano_result, expected_array))
Expand All @@ -55,16 +59,16 @@ def test_elu(self):
[-0.63212056, 2., -0.95021293, 4., 5.],
[-0.63212056, 2., -0.95021293, 4., 5.]])
alpha = 1
theano_result = A.Elu(self.theano_input,alpha).eval({self.theano_input: self.numpy_input})
theano_result = A.Elu(self.theano_input, alpha).eval({self.theano_input: self.numpy_input})
self.assertEqual(theano_result.shape, expected_array.shape)
self.assertTrue(np.allclose(theano_result, expected_array))

def test_sigmoid(self):
expected_array = np.array([[ 0.26894142, 0.88079708, 0.04742587, 0.98201379, 0.99330715],
[ 0.26894142, 0.88079708, 0.04742587, 0.98201379, 0.99330715],
[ 0.26894142, 0.88079708, 0.04742587, 0.98201379, 0.99330715],
[ 0.26894142, 0.88079708, 0.04742587, 0.98201379, 0.99330715],
[ 0.26894142, 0.88079708, 0.04742587, 0.98201379, 0.99330715]])
expected_array = np.array([[0.26894142, 0.88079708, 0.04742587, 0.98201379, 0.99330715],
[0.26894142, 0.88079708, 0.04742587, 0.98201379, 0.99330715],
[0.26894142, 0.88079708, 0.04742587, 0.98201379, 0.99330715],
[0.26894142, 0.88079708, 0.04742587, 0.98201379, 0.99330715],
[0.26894142, 0.88079708, 0.04742587, 0.98201379, 0.99330715]])
theano_result = A.Sigmoid(self.theano_input).eval({self.theano_input: self.numpy_input})
self.assertEqual(theano_result.shape, expected_array.shape)
self.assertTrue(np.allclose(theano_result, expected_array))
Expand All @@ -80,33 +84,32 @@ def test_tanh(self):
self.assertTrue(np.allclose(theano_result, expected_array))

def test_softmax(self):
expected_array = np.array([[ 1.74500937e-03, 3.50494502e-02, 2.36161338e-04,
2.58982354e-01, 7.03987026e-01],
[ 1.74500937e-03, 3.50494502e-02, 2.36161338e-04,
2.58982354e-01, 7.03987026e-01],
[ 1.74500937e-03, 3.50494502e-02, 2.36161338e-04,
2.58982354e-01, 7.03987026e-01],
[ 1.74500937e-03, 3.50494502e-02, 2.36161338e-04,
2.58982354e-01, 7.03987026e-01],
[ 1.74500937e-03, 3.50494502e-02, 2.36161338e-04,
2.58982354e-01, 7.03987026e-01]])
expected_array = np.array([[1.74500937e-03, 3.50494502e-02, 2.36161338e-04,
2.58982354e-01, 7.03987026e-01],
[1.74500937e-03, 3.50494502e-02, 2.36161338e-04,
2.58982354e-01, 7.03987026e-01],
[1.74500937e-03, 3.50494502e-02, 2.36161338e-04,
2.58982354e-01, 7.03987026e-01],
[1.74500937e-03, 3.50494502e-02, 2.36161338e-04,
2.58982354e-01, 7.03987026e-01],
[1.74500937e-03, 3.50494502e-02, 2.36161338e-04,
2.58982354e-01, 7.03987026e-01]])
theano_result = A.Softmax(self.theano_input).eval({self.theano_input: self.numpy_input})
self.assertEqual(theano_result.shape, expected_array.shape)
self.assertTrue(np.allclose(theano_result, expected_array))

def test_temperature_softmax(self):
"""
Utilized np.asarray(np.random.uniform(2, 10, 1), dtype = tehano.config.floatX)
Utilized np.asarray(np.random.uniform(2, 10, 1), dtype = theano.config.floatX)
to hardcode a value for t.
"""
t = np.array([7.94705237])
expected_array = np.array([[ 0.1381257 , 0.20147445, 0.10739337, 0.25912957, 0.29387691],
[ 0.1381257 , 0.20147445, 0.10739337, 0.25912957, 0.29387691],
[ 0.1381257 , 0.20147445, 0.10739337, 0.25912957, 0.29387691],
[ 0.1381257 , 0.20147445, 0.10739337, 0.25912957, 0.29387691],
[ 0.1381257 , 0.20147445, 0.10739337, 0.25912957, 0.29387691]])
theano_result = A.Softmax(self.theano_input,
temp=t).eval({self.theano_input: self.numpy_input})
expected_array = np.array([[0.1381257, 0.20147445, 0.10739337, 0.25912957, 0.29387691],
[0.1381257, 0.20147445, 0.10739337, 0.25912957, 0.29387691],
[0.1381257, 0.20147445, 0.10739337, 0.25912957, 0.29387691],
[0.1381257, 0.20147445, 0.10739337, 0.25912957, 0.29387691],
[0.1381257, 0.20147445, 0.10739337, 0.25912957, 0.29387691]])
theano_result = A.Softmax(self.theano_input, temp=t).eval({self.theano_input: self.numpy_input})
self.assertEqual(theano_result.shape, expected_array.shape)
self.assertTrue(np.allclose(theano_result, expected_array))

Expand Down

0 comments on commit 4d31068

Please sign in to comment.