These changes speeds up travis testing time 2 times using some pytest…

… and travis configuration options.

Summary of changes:
 - py.test is configured to display test profiling information that shows 10 slowest tests. This would allow additional speed ups if anyone has ideas on some particular test. The slowest test is usually cifar dataset test and tensorflow convolutions. It seems that there are some other IT tests that could be sped up.
 - py.test is configured to run with pytest-xdist with 2 processes in parallel because travis does provide multicore support (1.5 cores) and because the slowest cifar test spends time on download which can run in parallel with other tests.
 - travis is configured to split backend tests into test matrix to make parallel theano vs tensorflow testing as opposed to rerun all the tests twice for python 2.7.
 - pickle filenames in tests are renamed to avoid clashes during multiprocessing

.travis.yml

@@ -1,9 +1,14 @@
 sudo: required
 dist: trusty
 language: python
-python:
-  - "2.7"
-  - "3.4"
+matrix:
+    include:
+        - python: 3.4
+          env: KERAS_BACKEND=theano
+        - python: 2.7
+          env: KERAS_BACKEND=theano
+        - python: 2.7
+          env: KERAS_BACKEND=tensorflow
 install:
   # code below is taken from http://conda.pydata.org/docs/travis.html
   # We do this conditionally because it saves us some downloading if the
@@ -23,7 +28,7 @@ install:
 
   - conda create -q -n test-environment python=$TRAVIS_PYTHON_VERSION numpy scipy matplotlib pandas pytest h5py
   - source activate test-environment
-  - pip install pytest-cov python-coveralls
+  - pip install pytest-cov python-coveralls pytest-xdist
   - pip install git+git://github.com/Theano/Theano.git
   - python setup.py install
 
@@ -33,10 +38,11 @@ install:
     fi
 # command to run tests
 script:
-  - PYTHONPATH=$PWD:$PYTHONPATH py.test -v --cov-report term-missing --cov keras tests/
-  - if [[ "$TRAVIS_PYTHON_VERSION" == "2.7" ]]; then
-      sed -i -e 's/theano/tensorflow/g' ~/.keras/keras.json;
-      PYTHONPATH=$PWD:$PYTHONPATH py.test -v --cov-report term-missing --cov keras tests/;
-    fi
+  # run keras backend init to initialize backend config
+  - python -c "import keras.backend"
+  # set up keras backend
+  - sed -i -e 's/"backend":[[:space:]]*"[^"]*/"backend":\ "'$KERAS_BACKEND'/g' ~/.keras/keras.json;
+  - echo -e "Running tests with the following config:\n$(cat ~/.keras/keras.json)"
+  - PYTHONPATH=$PWD:$PYTHONPATH py.test tests/
 after_success:
   - coveralls

keras/backend/__init__.py

@@ -27,7 +27,9 @@
     _config = {'floatx': floatx(),
                'epsilon': epsilon(),
                'backend': _BACKEND}
-    json.dump(_config, open(_config_path, 'w'))
+    with open(_config_path, 'w') as f:
+        # add new line in order for bash 'cat' display the content correctly
+        f.write(json.dumps(_config) + '\n')
 
 if _BACKEND == 'theano':
     print('Using Theano backend.')
@@ -36,4 +38,4 @@
     print('Using TensorFlow backend.')
     from .tensorflow_backend import *
 else:
-    raise Exception('Unknown backend: ' + str(backend))
+    raise Exception('Unknown backend: ' + str(_BACKEND))

pytest.ini

@@ -0,0 +1,7 @@
+# Configuration of py.test
+[pytest]
+addopts=-v
+        -n 2
+        --durations=10
+        --cov-report term-missing
+        --cov=keras

tests/test_graph_model.py

@@ -139,7 +139,7 @@ def test_2o_1i_weights(self):
         assert(loss < 4.)
 
         print('test weight saving')
-        graph.save_weights('temp.h5', overwrite=True)
+        graph.save_weights('test_2o_1i_weights_temp.h5', overwrite=True)
         graph = Graph()
         graph.add_input(name='input1', input_shape=(32,))
         graph.add_node(Dense(16), name='dense1', input='input1')
@@ -148,7 +148,7 @@ def test_2o_1i_weights(self):
         graph.add_output(name='output1', input='dense2')
         graph.add_output(name='output2', input='dense3')
         graph.compile('rmsprop', {'output1': 'mse', 'output2': 'mse'})
-        graph.load_weights('temp.h5')
+        graph.load_weights('test_2o_1i_weights_temp.h5')
         nloss = graph.evaluate({'input1': X_test, 'output1': y_test, 'output2': y2_test})
         print(nloss)
         assert(loss == nloss)

tests/test_sequential_model.py

@@ -61,14 +61,14 @@ def test_sequential(self):
         model.get_config(verbose=0)
 
         print('test weight saving')
-        model.save_weights('temp.h5', overwrite=True)
+        model.save_weights('test_sequential_temp.h5', overwrite=True)
         model = Sequential()
         model.add(Dense(nb_hidden, input_shape=(input_dim,)))
         model.add(Activation('relu'))
         model.add(Dense(nb_class))
         model.add(Activation('softmax'))
         model.compile(loss='categorical_crossentropy', optimizer='rmsprop')
-        model.load_weights('temp.h5')
+        model.load_weights('test_sequential_temp.h5')
 
         nloss = model.evaluate(X_train, y_train, verbose=0)
         assert(loss == nloss)
@@ -114,7 +114,7 @@ def test_merge_sum(self):
         model.get_config(verbose=0)
 
         print('test weight saving')
-        model.save_weights('temp.h5', overwrite=True)
+        model.save_weights('test_merge_sum_temp.h5', overwrite=True)
         left = Sequential()
         left.add(Dense(nb_hidden, input_shape=(input_dim,)))
         left.add(Activation('relu'))
@@ -125,7 +125,7 @@ def test_merge_sum(self):
         model.add(Merge([left, right], mode='sum'))
         model.add(Dense(nb_class))
         model.add(Activation('softmax'))
-        model.load_weights('temp.h5')
+        model.load_weights('test_merge_sum_temp.h5')
         model.compile(loss='categorical_crossentropy', optimizer='rmsprop')
 
         nloss = model.evaluate([X_train, X_train], y_train, verbose=0)
@@ -205,7 +205,7 @@ def test_merge_concat(self):
         model.get_config(verbose=0)
 
         print('test weight saving')
-        model.save_weights('temp.h5', overwrite=True)
+        model.save_weights('test_merge_concat_temp.h5', overwrite=True)
         left = Sequential()
         left.add(Dense(nb_hidden, input_shape=(input_dim,)))
         left.add(Activation('relu'))
@@ -221,7 +221,7 @@ def test_merge_concat(self):
         model.add(Activation('softmax'))
 
         model.compile(loss='categorical_crossentropy', optimizer='rmsprop')
-        model.load_weights('temp.h5')
+        model.load_weights('test_merge_concat_temp.h5')
 
         nloss = model.evaluate([X_train, X_train], y_train, verbose=0)
         assert(loss == nloss)
@@ -268,8 +268,8 @@ def test_merge_recursivity(self):
         model.predict_proba([X_test, X_test, X_test], verbose=0)
         model.get_config(verbose=0)
 
-        model.save_weights('temp.h5', overwrite=True)
-        model.load_weights('temp.h5')
+        model.save_weights('test_merge_recursivity_temp.h5', overwrite=True)
+        model.load_weights('test_merge_recursivity_temp.h5')
 
         nloss = model.evaluate([X_train, X_train, X_train], y_train, verbose=0)
         print(nloss)
@@ -305,8 +305,8 @@ def test_merge_overlap(self):
         model.predict_proba(X_test, verbose=0)
         model.get_config(verbose=0)
 
-        model.save_weights('temp.h5', overwrite=True)
-        model.load_weights('temp.h5')
+        model.save_weights('test_merge_overlap_temp.h5', overwrite=True)
+        model.load_weights('test_merge_overlap_temp.h5')
 
         nloss = model.evaluate(X_train, y_train, verbose=0)
         print(nloss)
@@ -359,7 +359,7 @@ def output_shape(input_shapes):
         model.get_config(verbose=0)
 
         print('test weight saving')
-        model.save_weights('temp.h5', overwrite=True)
+        model.save_weights('test_lambda_temp.h5', overwrite=True)
         left = Sequential()
         left.add(Dense(nb_hidden, input_shape=(input_dim,)))
         left.add(Activation('relu'))
@@ -371,7 +371,7 @@ def output_shape(input_shapes):
                               output_shape=output_shape))
         model.add(Dense(nb_class))
         model.add(Lambda(activation))
-        model.load_weights('temp.h5')
+        model.load_weights('test_lambda_temp.h5')
         model.compile(loss='categorical_crossentropy', optimizer='rmsprop')
 
         nloss = model.evaluate([X_train, X_train], y_train, verbose=0)