Merge branch 'develop' into link

.travis.yml

@@ -8,10 +8,13 @@ os:
 env:
   - JOB=DOCS
   - JOB=BUILD_AND_TEST
+  - JOB=PRE_COMMIT
 matrix:
   exclude:
     - os: osx
-      env: JOB=DOCS  # Only generate documentation in linux
+      env: JOB=DOCS  # Only generate documentation in linux.
+    - os: osx
+      env: JOB=PRE_COMMIT # Only check pre-commit hook in linux
 
 addons:
   apt:
@@ -39,18 +42,23 @@ addons:
       - lcov
       - graphviz
       - swig
+      - clang-format-3.8
 before_install:
   - |
     if [ ${JOB} == "BUILD_AND_TEST" ]; then
-      if ! git diff --name-only $TRAVIS_COMMIT_RANGE | grep -qvE '(\.md$)|(\.rst$)|(\.jpg$)|(\.png$)'
-      then
-        echo "Only markdown docs were updated, stopping build process."
-        exit
+      local change_list=`git diff --name-only $TRAVIS_COMMIT_RANGE`
+      if [ $? -eq 0 ]; then  # if git diff return no zero, then rerun unit test.
+        if ! echo ${change_list} | grep -qvE '(\.md$)|(\.rst$)|(\.jpg$)|(\.png$)'
+        then
+          echo "Only markdown docs were updated, stopping build process."
+          exit
+        fi
       fi
     fi
   - if [[ "$TRAVIS_OS_NAME" == "linux" ]]; then sudo paddle/scripts/travis/before_install.linux.sh; fi
   - if [[ "$TRAVIS_OS_NAME" == "osx" ]]; then paddle/scripts/travis/before_install.osx.sh; fi
-  - pip install wheel protobuf sphinx recommonmark virtualenv numpy sphinx_rtd_theme
+  - if [[ "$JOB" == "PRE_COMMIT" ]]; then sudo ln -s /usr/bin/clang-format-3.8 /usr/bin/clang-format; fi
+  - pip install wheel protobuf sphinx recommonmark virtualenv numpy sphinx_rtd_theme pre-commit
 script:
   - paddle/scripts/travis/main.sh
 notifications:

WORKSPACE

@@ -1,17 +1,15 @@
 # External dependency to Google protobuf.
 http_archive(
-    name = "protobuf",
-    url = "http://github.com/google/protobuf/archive/v3.1.0.tar.gz",
-    sha256 = "0a0ae63cbffc274efb573bdde9a253e3f32e458c41261df51c5dbc5ad541e8f7",
-    strip_prefix = "protobuf-3.1.0",
-)
+    name="protobuf",
+    url="http://github.com/google/protobuf/archive/v3.1.0.tar.gz",
+    sha256="0a0ae63cbffc274efb573bdde9a253e3f32e458c41261df51c5dbc5ad541e8f7",
+    strip_prefix="protobuf-3.1.0", )
 
 # External dependency to gtest 1.7.0.  This method comes from
 # https://www.bazel.io/versions/master/docs/tutorial/cpp.html.
 new_http_archive(
-    name = "gtest",
-    url = "https://github.com/google/googletest/archive/release-1.7.0.zip",
-    sha256 = "b58cb7547a28b2c718d1e38aee18a3659c9e3ff52440297e965f5edffe34b6d0",
-    build_file = "third_party/gtest.BUILD",
-    strip_prefix = "googletest-release-1.7.0",
-)
+    name="gtest",
+    url="https://github.com/google/googletest/archive/release-1.7.0.zip",
+    sha256="b58cb7547a28b2c718d1e38aee18a3659c9e3ff52440297e965f5edffe34b6d0",
+    build_file="third_party/gtest.BUILD",
+    strip_prefix="googletest-release-1.7.0", )

benchmark/tensorflow/rnn/run_multi.sh

@@ -25,4 +25,3 @@ test 4 2 256 512
 test 4 2 512 128 
 test 4 2 512 256 
 test 4 2 512 512 
-

demo/gan/README.md

@@ -10,4 +10,4 @@ Then you can run the command below. The flag -d specifies the training data (cif
 $python gan_trainer.py -d cifar --use_gpu 1
 
 The generated images will be stored in ./cifar_samples/
-The corresponding models will be stored in ./cifar_params/
+The corresponding models will be stored in ./cifar_params/

demo/gan/data/download_cifar.sh

@@ -15,4 +15,3 @@ set -e
 wget https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz
 tar zxf cifar-10-python.tar.gz
 rm cifar-10-python.tar.gz
-

demo/gan/data/get_mnist_data.sh

@@ -15,5 +15,3 @@ do
         gunzip ${fname}.gz
     fi
 done
-
-

demo/gan/gan_conf.py

@@ -14,10 +14,9 @@
 from paddle.trainer_config_helpers import *
 
 mode = get_config_arg("mode", str, "generator")
-assert mode in set(["generator",
-                    "discriminator",
-                    "generator_training",
-                    "discriminator_training"])
+assert mode in set([
+    "generator", "discriminator", "generator_training", "discriminator_training"
+])
 
 is_generator_training = mode == "generator_training"
 is_discriminator_training = mode == "discriminator_training"
@@ -38,8 +37,8 @@ sample_dim = 2
 settings(
     batch_size=128,
     learning_rate=1e-4,
-    learning_method=AdamOptimizer(beta1=0.5)
-)
+    learning_method=AdamOptimizer(beta1=0.5))
+
 
 def discriminator(sample):
     """
@@ -50,70 +49,87 @@ def discriminator(sample):
     of the sample is from real data.
     """
     param_attr = ParamAttr(is_static=is_generator_training)
-    bias_attr = ParamAttr(is_static=is_generator_training,
-                          initial_mean=1.0,
-                          initial_std=0)
-
-    hidden = fc_layer(input=sample, name="dis_hidden", size=hidden_dim,
-                    bias_attr=bias_attr,
-                    param_attr=param_attr,
-                    act=ReluActivation())
-
-    hidden2 = fc_layer(input=hidden, name="dis_hidden2", size=hidden_dim,
-                    bias_attr=bias_attr,
-                    param_attr=param_attr,
-                    act=LinearActivation())
-    
-    hidden_bn = batch_norm_layer(hidden2, 
-                     act=ReluActivation(), 
-                     name="dis_hidden_bn", 
-                     bias_attr=bias_attr, 
-                     param_attr=ParamAttr(is_static=is_generator_training,
-                           initial_mean=1.0,
-                           initial_std=0.02),
-                     use_global_stats=False)
-    
-    return fc_layer(input=hidden_bn, name="dis_prob", size=2,
-                    bias_attr=bias_attr,
-                    param_attr=param_attr,
-                    act=SoftmaxActivation())
+    bias_attr = ParamAttr(
+        is_static=is_generator_training, initial_mean=1.0, initial_std=0)
+
+    hidden = fc_layer(
+        input=sample,
+        name="dis_hidden",
+        size=hidden_dim,
+        bias_attr=bias_attr,
+        param_attr=param_attr,
+        act=ReluActivation())
+
+    hidden2 = fc_layer(
+        input=hidden,
+        name="dis_hidden2",
+        size=hidden_dim,
+        bias_attr=bias_attr,
+        param_attr=param_attr,
+        act=LinearActivation())
+
+    hidden_bn = batch_norm_layer(
+        hidden2,
+        act=ReluActivation(),
+        name="dis_hidden_bn",
+        bias_attr=bias_attr,
+        param_attr=ParamAttr(
+            is_static=is_generator_training, initial_mean=1.0,
+            initial_std=0.02),
+        use_global_stats=False)
+
+    return fc_layer(
+        input=hidden_bn,
+        name="dis_prob",
+        size=2,
+        bias_attr=bias_attr,
+        param_attr=param_attr,
+        act=SoftmaxActivation())
+
 
 def generator(noise):
     """
     generator generates a sample given noise
     """
     param_attr = ParamAttr(is_static=is_discriminator_training)
-    bias_attr = ParamAttr(is_static=is_discriminator_training,
-                           initial_mean=1.0,
-                           initial_std=0)
-    
-    hidden = fc_layer(input=noise,
-                    name="gen_layer_hidden",
-                    size=hidden_dim,
-                    bias_attr=bias_attr,
-                    param_attr=param_attr,
-                    act=ReluActivation())
-
-    hidden2 = fc_layer(input=hidden, name="gen_hidden2", size=hidden_dim,
-                    bias_attr=bias_attr,
-                    param_attr=param_attr,
-                    act=LinearActivation())
-    
-    hidden_bn = batch_norm_layer(hidden2, 
-                     act=ReluActivation(), 
-                     name="gen_layer_hidden_bn", 
-                     bias_attr=bias_attr, 
-                     param_attr=ParamAttr(is_static=is_discriminator_training,
-                           initial_mean=1.0,
-                           initial_std=0.02),
-                     use_global_stats=False)
-    
-    return fc_layer(input=hidden_bn,
-                    name="gen_layer1",
-                    size=sample_dim,
-                    bias_attr=bias_attr,
-                    param_attr=param_attr,
-                    act=LinearActivation())
+    bias_attr = ParamAttr(
+        is_static=is_discriminator_training, initial_mean=1.0, initial_std=0)
+
+    hidden = fc_layer(
+        input=noise,
+        name="gen_layer_hidden",
+        size=hidden_dim,
+        bias_attr=bias_attr,
+        param_attr=param_attr,
+        act=ReluActivation())
+
+    hidden2 = fc_layer(
+        input=hidden,
+        name="gen_hidden2",
+        size=hidden_dim,
+        bias_attr=bias_attr,
+        param_attr=param_attr,
+        act=LinearActivation())
+
+    hidden_bn = batch_norm_layer(
+        hidden2,
+        act=ReluActivation(),
+        name="gen_layer_hidden_bn",
+        bias_attr=bias_attr,
+        param_attr=ParamAttr(
+            is_static=is_discriminator_training,
+            initial_mean=1.0,
+            initial_std=0.02),
+        use_global_stats=False)
+
+    return fc_layer(
+        input=hidden_bn,
+        name="gen_layer1",
+        size=sample_dim,
+        bias_attr=bias_attr,
+        param_attr=param_attr,
+        act=LinearActivation())
+
 
 if is_generator_training:
     noise = data_layer(name="noise", size=noise_dim)
@@ -126,7 +142,8 @@ if is_generator_training or is_discriminator_training:
     label = data_layer(name="label", size=1)
     prob = discriminator(sample)
     cost = cross_entropy(input=prob, label=label)
-    classification_error_evaluator(input=prob, label=label, name=mode+'_error')
+    classification_error_evaluator(
+        input=prob, label=label, name=mode + '_error')
     outputs(cost)
 
 if is_generator:

demo/quick_start/trainer_config.resnet-lstm.py

@@ -13,7 +13,6 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-
 """
 This configuration is a demonstration of how to implement the stacked LSTM
 with residual connections, i.e. an LSTM layer takes the sum of the hidden states
@@ -46,11 +45,12 @@ is_predict = get_config_arg('is_predict', bool, False)
 trn = 'data/train.list' if not is_predict else None
 tst = 'data/test.list' if not is_predict else 'data/pred.list'
 process = 'process' if not is_predict else 'process_predict'
-define_py_data_sources2(train_list=trn,
-                        test_list=tst,
-                        module="dataprovider_emb",
-                        obj=process,
-                        args={"dictionary": word_dict})
+define_py_data_sources2(
+    train_list=trn,
+    test_list=tst,
+    module="dataprovider_emb",
+    obj=process,
+    args={"dictionary": word_dict})
 
 batch_size = 128 if not is_predict else 1
 settings(
@@ -58,10 +58,9 @@ settings(
     learning_rate=2e-3,
     learning_method=AdamOptimizer(),
     regularization=L2Regularization(8e-4),
-    gradient_clipping_threshold=25
-)
+    gradient_clipping_threshold=25)
 
-bias_attr = ParamAttr(initial_std=0.,l2_rate=0.)
+bias_attr = ParamAttr(initial_std=0., l2_rate=0.)
 
 data = data_layer(name="word", size=len(word_dict))
 emb = embedding_layer(input=data, size=128)
@@ -73,17 +72,15 @@ for i in range(3):
     # The input to the current layer is the sum of the hidden state
     # and input of the previous layer.
     current_input = addto_layer(input=[previous_input, previous_hidden_state])
-    hidden_state = simple_lstm(input=current_input, size=128,
-                               lstm_cell_attr=ExtraAttr(drop_rate=0.1))
+    hidden_state = simple_lstm(
+        input=current_input, size=128, lstm_cell_attr=ExtraAttr(drop_rate=0.1))
     previous_input, previous_hidden_state = current_input, hidden_state
 
 lstm = previous_hidden_state
 
 lstm_last = pooling_layer(input=lstm, pooling_type=MaxPooling())
-output = fc_layer(input=lstm_last, size=2,
-                  bias_attr=bias_attr,
-                  act=SoftmaxActivation())
-
+output = fc_layer(
+    input=lstm_last, size=2, bias_attr=bias_attr, act=SoftmaxActivation())
 
 if is_predict:
     maxid = maxid_layer(output)

demo/semantic_role_labeling/data/extract_dict_feature.py

@@ -33,7 +33,7 @@ def extract_dict_features(pair_file, feature_file):
                 ctx_n1 = sentence_list[verb_index - 1]
             else:
                 ctx_n1 = 'bos'
-            
+
             if verb_index > 1:
                 mark[verb_index - 2] = 1
                 ctx_n2 = sentence_list[verb_index - 2]
@@ -48,7 +48,7 @@ def extract_dict_features(pair_file, feature_file):
                 ctx_p1 = sentence_list[verb_index + 1]
             else:
                 ctx_p1 = 'eos'
-            
+
             if verb_index < len(labels_list) - 3:
                 mark[verb_index + 2] = 1
                 ctx_p2 = sentence_list[verb_index + 2]
@@ -69,7 +69,6 @@ def extract_dict_features(pair_file, feature_file):
             feature_out.write(feature_str + '\n')
 
 
-
 if __name__ == '__main__':
 
     usage = '-p pair_file -f feature_file'

demo/semantic_role_labeling/data/extract_pairs.py

@@ -66,8 +66,8 @@ def transform_labels(sentences, labels):
         else:
             verb_list = []
             for x in labels[i][0]:
-                if x !='-':
-                   verb_list.append(x)
+                if x != '-':
+                    verb_list.append(x)
 
             for j in xrange(1, len(labels[i])):
                 label_list = labels[i][j]
@@ -93,7 +93,7 @@ def transform_labels(sentences, labels):
                         is_in_bracket = True
                     else:
                         print 'error:', ll
-                sen_lab_pair.append((sentences[i], verb_list[j-1], label_seq))
+                sen_lab_pair.append((sentences[i], verb_list[j - 1], label_seq))
     return sen_lab_pair
 
 
@@ -103,7 +103,7 @@ def write_file(sen_lab_pair, output_file):
             sentence = x[0]
             label_seq = ' '.join(x[2])
             assert len(sentence.split()) == len(x[2])
-            fout.write(sentence + '\t' + x[1]+'\t' +label_seq + '\n')
+            fout.write(sentence + '\t' + x[1] + '\t' + label_seq + '\n')
 
 
 if __name__ == '__main__':

demo/semantic_role_labeling/dataprovider.py

@@ -21,7 +21,7 @@ def hook(settings, word_dict, label_dict, predicate_dict, **kwargs):
     settings.word_dict = word_dict
     settings.label_dict = label_dict
     settings.predicate_dict = predicate_dict
-   
+
     #all inputs are integral and sequential type
     settings.slots = [
         integer_value_sequence(len(word_dict)),
@@ -29,25 +29,28 @@ def hook(settings, word_dict, label_dict, predicate_dict, **kwargs):
         integer_value_sequence(len(word_dict)),
         integer_value_sequence(len(word_dict)),
         integer_value_sequence(len(word_dict)),
-        integer_value_sequence(len(word_dict)), 
-        integer_value_sequence(len(predicate_dict)),
-        integer_value_sequence(2),
+        integer_value_sequence(len(word_dict)),
+        integer_value_sequence(len(predicate_dict)), integer_value_sequence(2),
         integer_value_sequence(len(label_dict))
     ]
 
 
 def get_batch_size(yeild_data):
     return len(yeild_data[0])
-    
 
-@provider(init_hook=hook, should_shuffle=True, calc_batch_size=get_batch_size, 
-          can_over_batch_size=False, cache=CacheType.CACHE_PASS_IN_MEM)
+
+@provider(
+    init_hook=hook,
+    should_shuffle=True,
+    calc_batch_size=get_batch_size,
+    can_over_batch_size=False,
+    cache=CacheType.CACHE_PASS_IN_MEM)
 def process(settings, file_name):
     with open(file_name, 'r') as fdata:
         for line in fdata:
             sentence, predicate, ctx_n2, ctx_n1, ctx_0, ctx_p1, ctx_p2,  mark, label = \
                 line.strip().split('\t')
-           
+
             words = sentence.split()
             sen_len = len(words)
             word_slot = [settings.word_dict.get(w, UNK_IDX) for w in words]

demo/semantic_role_labeling/db_lstm.py

@@ -20,7 +20,7 @@ from paddle.trainer_config_helpers import *
 #file paths
 word_dict_file = './data/wordDict.txt'
 label_dict_file = './data/targetDict.txt'
-predicate_file= './data/verbDict.txt'
+predicate_file = './data/verbDict.txt'
 train_list_file = './data/train.list'
 test_list_file = './data/test.list'
 
@@ -47,7 +47,6 @@ if not is_predict:
             w = line.strip()
             predicate_dict[w] = i
 
-
     if is_test:
         train_list_file = None
 
@@ -57,9 +56,11 @@ if not is_predict:
         test_list=test_list_file,
         module='dataprovider',
         obj='process',
-        args={'word_dict': word_dict,
-              'label_dict': label_dict,
-              'predicate_dict': predicate_dict })
+        args={
+            'word_dict': word_dict,
+            'label_dict': label_dict,
+            'predicate_dict': predicate_dict
+        })
 
     word_dict_len = len(word_dict)
     label_dict_len = len(label_dict)
@@ -77,24 +78,16 @@ mark_dim = 5
 hidden_dim = 512
 depth = 8
 
-
-
 ########################### Optimizer #######################################
 
-
 settings(
     batch_size=150,
     learning_method=MomentumOptimizer(momentum=0),
     learning_rate=2e-2,
     regularization=L2Regularization(8e-4),
     is_async=False,
-    model_average=ModelAverage(average_window=0.5,
-                               max_average_window=10000),
-                               
-)
-
-
-
+    model_average=ModelAverage(
+        average_window=0.5, max_average_window=10000), )
 
 ####################################### network ##############################
 #8 features and 1 target
@@ -108,22 +101,28 @@ ctx_p1 = data_layer(name='ctx_p1_data', size=word_dict_len)
 ctx_p2 = data_layer(name='ctx_p2_data', size=word_dict_len)
 mark = data_layer(name='mark_data', size=mark_dict_len)
 
-
 if not is_predict:
     target = data_layer(name='target', size=label_dict_len)
 
-
-default_std=1/math.sqrt(hidden_dim)/3.0
+default_std = 1 / math.sqrt(hidden_dim) / 3.0
 
 emb_para = ParameterAttribute(name='emb', initial_std=0., learning_rate=0.)
 std_0 = ParameterAttribute(initial_std=0.)
-std_default = ParameterAttribute(initial_std=default_std) 
-
-predicate_embedding = embedding_layer(size=word_dim, input=predicate, param_attr=ParameterAttribute(name='vemb',initial_std=default_std))
-mark_embedding = embedding_layer(name='word_ctx-in_embedding', size=mark_dim, input=mark, param_attr=std_0)
-
-word_input=[word, ctx_n2, ctx_n1, ctx_0, ctx_p1, ctx_p2]
-emb_layers = [embedding_layer(size=word_dim, input=x, param_attr=emb_para) for x in word_input]
+std_default = ParameterAttribute(initial_std=default_std)
+
+predicate_embedding = embedding_layer(
+    size=word_dim,
+    input=predicate,
+    param_attr=ParameterAttribute(
+        name='vemb', initial_std=default_std))
+mark_embedding = embedding_layer(
+    name='word_ctx-in_embedding', size=mark_dim, input=mark, param_attr=std_0)
+
+word_input = [word, ctx_n2, ctx_n1, ctx_0, ctx_p1, ctx_p2]
+emb_layers = [
+    embedding_layer(
+        size=word_dim, input=x, param_attr=emb_para) for x in word_input
+]
 emb_layers.append(predicate_embedding)
 emb_layers.append(mark_embedding)
 
@@ -131,84 +130,89 @@ hidden_0 = mixed_layer(
     name='hidden0',
     size=hidden_dim,
     bias_attr=std_default,
-    input=[ full_matrix_projection(input=emb, param_attr=std_default ) for emb in emb_layers ])
-
+    input=[
+        full_matrix_projection(
+            input=emb, param_attr=std_default) for emb in emb_layers
+    ])
 
 mix_hidden_lr = 1e-3
 lstm_para_attr = ParameterAttribute(initial_std=0.0, learning_rate=1.0)
-hidden_para_attr = ParameterAttribute(initial_std=default_std, learning_rate=mix_hidden_lr)
-
-lstm_0 = lstmemory(name='lstm0',
-                   input=hidden_0, 
-                   act=ReluActivation(),
-                   gate_act=SigmoidActivation(),
-                   state_act=SigmoidActivation(),
-                   bias_attr=std_0,
-                   param_attr=lstm_para_attr)
+hidden_para_attr = ParameterAttribute(
+    initial_std=default_std, learning_rate=mix_hidden_lr)
+
+lstm_0 = lstmemory(
+    name='lstm0',
+    input=hidden_0,
+    act=ReluActivation(),
+    gate_act=SigmoidActivation(),
+    state_act=SigmoidActivation(),
+    bias_attr=std_0,
+    param_attr=lstm_para_attr)
 
 #stack L-LSTM and R-LSTM with direct edges
 input_tmp = [hidden_0, lstm_0]
 
-
 for i in range(1, depth):
 
-    mix_hidden = mixed_layer(name='hidden'+str(i),
-                             size=hidden_dim, 
-                             bias_attr=std_default,
-                             input=[full_matrix_projection(input=input_tmp[0], param_attr=hidden_para_attr),
-                                    full_matrix_projection(input=input_tmp[1], param_attr=lstm_para_attr)
-                                   ]
-                             )
-
-    lstm = lstmemory(name='lstm'+str(i),
-                     input=mix_hidden,
-                     act=ReluActivation(),
-                     gate_act=SigmoidActivation(),
-                     state_act=SigmoidActivation(),
-                     reverse=((i % 2)==1),
-                     bias_attr=std_0,
-                     param_attr=lstm_para_attr)
+    mix_hidden = mixed_layer(
+        name='hidden' + str(i),
+        size=hidden_dim,
+        bias_attr=std_default,
+        input=[
+            full_matrix_projection(
+                input=input_tmp[0], param_attr=hidden_para_attr),
+            full_matrix_projection(
+                input=input_tmp[1], param_attr=lstm_para_attr)
+        ])
+
+    lstm = lstmemory(
+        name='lstm' + str(i),
+        input=mix_hidden,
+        act=ReluActivation(),
+        gate_act=SigmoidActivation(),
+        state_act=SigmoidActivation(),
+        reverse=((i % 2) == 1),
+        bias_attr=std_0,
+        param_attr=lstm_para_attr)
 
     input_tmp = [mix_hidden, lstm]
 
-feature_out = mixed_layer(name='output',
-                          size=label_dict_len,
-                          bias_attr=std_default, 
-                          input=[full_matrix_projection(input=input_tmp[0], param_attr=hidden_para_attr),
-                                 full_matrix_projection(input=input_tmp[1], param_attr=lstm_para_attr)
-                                ],
-                          )
-
-
+feature_out = mixed_layer(
+    name='output',
+    size=label_dict_len,
+    bias_attr=std_default,
+    input=[
+        full_matrix_projection(
+            input=input_tmp[0], param_attr=hidden_para_attr),
+        full_matrix_projection(
+            input=input_tmp[1], param_attr=lstm_para_attr)
+    ], )
 
 if not is_predict:
-    crf_l = crf_layer( name = 'crf',
-                       size = label_dict_len,
-                       input = feature_out, 
-                       label = target,
-                       param_attr=ParameterAttribute(name='crfw',initial_std=default_std, learning_rate=mix_hidden_lr)
-
-                      )
-
-    
-    crf_dec_l = crf_decoding_layer(name = 'crf_dec_l',
-                                   size = label_dict_len,
-                                   input = feature_out,
-                                   label = target,
-                                   param_attr=ParameterAttribute(name='crfw')
-                                       )
-
+    crf_l = crf_layer(
+        name='crf',
+        size=label_dict_len,
+        input=feature_out,
+        label=target,
+        param_attr=ParameterAttribute(
+            name='crfw', initial_std=default_std, learning_rate=mix_hidden_lr))
+
+    crf_dec_l = crf_decoding_layer(
+        name='crf_dec_l',
+        size=label_dict_len,
+        input=feature_out,
+        label=target,
+        param_attr=ParameterAttribute(name='crfw'))
 
     eval = sum_evaluator(input=crf_dec_l)
-        
+
     outputs(crf_l)
 
 else:
-    crf_dec_l = crf_decoding_layer(name = 'crf_dec_l',
-                                   size = label_dict_len,
-                                   input = feature_out,
-                                   param_attr=ParameterAttribute(name='crfw')
-                                       )
+    crf_dec_l = crf_decoding_layer(
+        name='crf_dec_l',
+        size=label_dict_len,
+        input=feature_out,
+        param_attr=ParameterAttribute(name='crfw'))
 
     outputs(crf_dec_l)
-

demo/semantic_role_labeling/predict.py

@@ -26,7 +26,8 @@ UNK_IDX = 0
 
 
 class Prediction():
-    def __init__(self, train_conf, dict_file, model_dir, label_file, predicate_dict_file):
+    def __init__(self, train_conf, dict_file, model_dir, label_file,
+                 predicate_dict_file):
         """
         train_conf: trainer configure.
         dict_file: word dictionary file name.
@@ -35,7 +36,7 @@ class Prediction():
 
         self.dict = {}
         self.labels = {}
-        self.predicate_dict={}
+        self.predicate_dict = {}
         self.labels_reverse = {}
         self.load_dict_label(dict_file, label_file, predicate_dict_file)
 
@@ -44,25 +45,18 @@ class Prediction():
         len_pred = len(self.predicate_dict)
 
         conf = parse_config(
-            train_conf,
-            'dict_len=' + str(len_dict) + 
-            ',label_len=' + str(len_label) +
-            ',pred_len=' + str(len_pred) +
-            ',is_predict=True')
+            train_conf, 'dict_len=' + str(len_dict) + ',label_len=' +
+            str(len_label) + ',pred_len=' + str(len_pred) + ',is_predict=True')
         self.network = swig_paddle.GradientMachine.createFromConfigProto(
             conf.model_config)
         self.network.loadParameters(model_dir)
 
         slots = [
-            integer_value_sequence(len_dict),
-            integer_value_sequence(len_dict),
-            integer_value_sequence(len_dict),
-            integer_value_sequence(len_dict),
-            integer_value_sequence(len_dict),
-            integer_value_sequence(len_dict), 
-            integer_value_sequence(len_pred),
-            integer_value_sequence(2)
-            ]
+            integer_value_sequence(len_dict), integer_value_sequence(len_dict),
+            integer_value_sequence(len_dict), integer_value_sequence(len_dict),
+            integer_value_sequence(len_dict), integer_value_sequence(len_dict),
+            integer_value_sequence(len_pred), integer_value_sequence(2)
+        ]
         self.converter = DataProviderConverter(slots)
 
     def load_dict_label(self, dict_file, label_file, predicate_dict_file):
@@ -78,6 +72,7 @@ class Prediction():
 
         for line_count, line in enumerate(open(predicate_dict_file, 'r')):
             self.predicate_dict[line.strip()] = line_count
+
     def get_data(self, data_file):
         """
         Get input data of paddle format.
@@ -88,9 +83,10 @@ class Prediction():
                 ).split('\t')
                 words = sentence.split()
                 sen_len = len(words)
-                 
+
                 word_slot = [self.dict.get(w, UNK_IDX) for w in words]
-                predicate_slot = [self.predicate_dict.get(predicate, UNK_IDX)] * sen_len
+                predicate_slot = [self.predicate_dict.get(predicate, UNK_IDX)
+                                  ] * sen_len
                 ctx_n2_slot = [self.dict.get(ctx_n2, UNK_IDX)] * sen_len
                 ctx_n1_slot = [self.dict.get(ctx_n1, UNK_IDX)] * sen_len
                 ctx_0_slot = [self.dict.get(ctx_0, UNK_IDX)] * sen_len
@@ -99,7 +95,7 @@ class Prediction():
 
                 marks = mark.split()
                 mark_slot = [int(w) for w in marks]
-                
+
                 yield word_slot, ctx_n2_slot, ctx_n1_slot, \
                       ctx_0_slot, ctx_p1_slot, ctx_p2_slot, predicate_slot, mark_slot
 
@@ -123,8 +119,9 @@ class Prediction():
 
 
 def option_parser():
-    usage = ("python predict.py -c config -w model_dir " 
-             "-d word dictionary -l label_file -i input_file  -p pred_dict_file")
+    usage = (
+        "python predict.py -c config -w model_dir "
+        "-d word dictionary -l label_file -i input_file  -p pred_dict_file")
     parser = OptionParser(usage="usage: %s [options]" % usage)
     parser.add_option(
         "-c",
@@ -187,8 +184,9 @@ def main():
     output_file = options.output_file
 
     swig_paddle.initPaddle("--use_gpu=0")
-    predict = Prediction(train_conf, dict_file, model_path, label_file, predict_dict_file)
-    predict.predict(data_file,output_file)
+    predict = Prediction(train_conf, dict_file, model_path, label_file,
+                         predict_dict_file)
+    predict.predict(data_file, output_file)
 
 
 if __name__ == '__main__':

demo/sentiment/predict.py

@@ -71,9 +71,7 @@ class SentimentPrediction():
         transform word into integer index according to the dictionary.
         """
         words = data.strip().split()
-        word_slot = [
-            self.word_dict[w] for w in words if w in self.word_dict
-        ]
+        word_slot = [self.word_dict[w] for w in words if w in self.word_dict]
         return word_slot
 
     def batch_predict(self, data_batch):
@@ -85,8 +83,8 @@ class SentimentPrediction():
             if self.label is None:
                 print("predicting label is %d" % (lab[0]))
             else:
-                print("predicting label is %s" %
-                      (self.label[lab[0]]))
+                print("predicting label is %s" % (self.label[lab[0]]))
+
 
 def option_parser():
     usage = "python predict.py -n config -w model_dir -d dictionary -i input_file "
@@ -143,9 +141,10 @@ def main():
         batch.append([predict.get_index(line)])
         if len(batch) == batch_size:
             predict.batch_predict(batch)
-            batch=[]
+            batch = []
     if len(batch) > 0:
         predict.batch_predict(batch)
 
+
 if __name__ == '__main__':
     main()

doc/getstarted/build_and_install/build_from_source_en.md

@@ -14,6 +14,13 @@ cd paddle
 git submodule update --init --recursive
 ```
 
+If you already have a local PaddlePaddle repo and have not initialized the submodule, your local submodule folder will be empty. You can simply run the last line of the above codes in your PaddlePaddle home directory to initialize your submodule folder.
+
+If you have already initialized your submodule and you would like to sync with the upstream submodule repo, you can run the following command
+```
+git submodule update --remote
+```
+
 ## <span id="requirements">Requirements</span>
 
 To compile the source code, your computer must be equipped with the following dependencies.

doc/getstarted/build_and_install/docker_install_en.rst

@@ -122,9 +122,9 @@ The general development workflow with Docker and Bazel is as follows:
       git clone --recursive https://github.com/paddlepaddle/paddle
 
 
-2. Build a development Docker image `paddle:dev` from the source code.
-   This image contains all the development tools and dependencies of
-   PaddlePaddle.
+2. Build a development Docker image :code:`paddle:dev` from the source
+   code.  This image contains all the development tools and
+   dependencies of PaddlePaddle.
 
 
    .. code-block:: bash
@@ -139,14 +139,22 @@ The general development workflow with Docker and Bazel is as follows:
 
    .. code-block:: bash
 
-      docker run \
-       -d # run the container in background mode \
-       --name paddle # we can run a nginx container to serve documents \
-       -p 2022:22    # so we can SSH into this container \
-       -v $PWD:/paddle # mount the source code \
-       -v $HOME/.cache/bazel:/root/.cache/bazel # mount Bazel cache \
+      docker run       \
+       -d              \
+       --name paddle   \
+       -p 2022:22      \
+       -v $PWD:/paddle \
+       -v $HOME/.cache/bazel:/root/.cache/bazel \
        paddle:dev
 
+   where :code:`-d` makes the container running in background,
+   :code:`--name paddle` allows us to run a nginx container to serve
+   documents in this container, :code:`-p 2022:22` allows us to SSH
+   into this container, :code:`-v $PWD:/paddle` shares the source code
+   on the host with the container, :code:`-v
+   $HOME/.cache/bazel:/root/.cache/bazel` shares Bazel cache on the
+   host with the container.
+
 4. SSH into the container:
 
    .. code-block:: bash

doc_cn/cluster/k8s/distributed_training_on_kubernetes.md

@@ -306,4 +306,4 @@ I1116 09:10:18.019069    50 ParameterClient2.cpp:122] pserver 2 192.168.223.143:
 I1116 09:10:18.019492    50 ParameterClient2.cpp:122] pserver 3 192.168.223.143:7165
 I1116 09:10:18.019716    50 ParameterClient2.cpp:122] pserver 4 192.168.129.71:7164
 I1116 09:10:18.019836    50 ParameterClient2.cpp:122] pserver 5 192.168.129.71:7165
-```
+```

doc_cn/cluster/k8s/job.yaml

@@ -40,4 +40,4 @@ spec:
         - name: jobpath
           mountPath: /home/jobpath       
       restartPolicy: Never
-    
+    

doc_cn/cluster/k8s/start_paddle.py

@@ -19,7 +19,6 @@ import socket
 import os
 import argparse
 
-
 # configuration for cluster
 API = "/api/v1/namespaces/"
 JOBSELECTOR = "labelSelector=job-name="
@@ -145,8 +144,8 @@ def startPaddle(idMap={}, train_args_dict=None):
 
 
 if __name__ == '__main__':
-    parser = argparse.ArgumentParser(prog="start_paddle.py",
-                                     description='simple tool for k8s')
+    parser = argparse.ArgumentParser(
+        prog="start_paddle.py", description='simple tool for k8s')
     args, train_args_list = parser.parse_known_args()
     train_args = refine_unknown_args(train_args_list)
     train_args_dict = dict(zip(train_args[:-1:2], train_args[1::2]))

doc_cn/demo/sentiment_analysis/index.rst

@@ -1,8 +1,8 @@
-情感分析教程
-===========================
-
-.. toctree::
-    :maxdepth: 3
-    :glob:
-
+情感分析教程
+===========================
+
+.. toctree::
+    :maxdepth: 3
+    :glob:
+
     Training Locally <sentiment_analysis.md>

doc_theme/static/js/paddle_doc_init.js

@@ -28,4 +28,4 @@ $(document).ready(function(){
     $('.doc-menu-vertical').find('li.current').last().addClass('active');
 
     $('.doc-menu-vertical').perfectScrollbar();
-});
+});

paddle/api/GradientMachine.cpp

@@ -15,8 +15,8 @@ limitations under the License. */
 #include "PaddleAPI.h"
 #include "PaddleAPIPrivate.h"
 
-#include "paddle/gserver/gradientmachines/NeuralNetwork.h"
 #include "Internal.h"
+#include "paddle/gserver/gradientmachines/NeuralNetwork.h"
 
 std::vector<int> GradientMachine::defaultParamTypes = {
     PARAMETER_VALUE, PARAMETER_GRADIENT, PARAMETER_MOMENTUM};

paddle/api/Internal.h

@@ -16,14 +16,13 @@ limitations under the License. */
 
 #include "PaddleAPI.h"
 
-#include <vector>
 #include <algorithm>
+#include <vector>
 
 template <typename T1, typename T2>
 void staticCastVector(std::vector<T2>* dest, const std::vector<T1>& src) {
   dest->resize(src.size());
-  std::transform(src.begin(),
-                 src.end(),
-                 dest->begin(),
-                 [](T1 t) { return static_cast<T2>(t); });
+  std::transform(src.begin(), src.end(), dest->begin(), [](T1 t) {
+    return static_cast<T2>(t);
+  });
 }