Code update

2016-11_Seminar/Session 1 - SENNA/code for NER/NER.py

@@ -70,28 +70,29 @@
 
 # Create the train and predict_labels function
 n_in = train_tokens.shape[1]
-n_hidden = numHiddenUnits
 n_out = len(label2Idx)
 
 
 words = Sequential()
-words.add(Embedding(output_dim=wordEmbeddings.shape[1], input_dim=wordEmbeddings.shape[0], input_length=n_in,  weights=[wordEmbeddings], trainable=False))       
+words.add(Embedding(input_dim=wordEmbeddings.shape[0], output_dim=wordEmbeddings.shape[1], input_length=n_in, weights=[wordEmbeddings], trainable=False))       
 words.add(Flatten())
 
 casing = Sequential()  
-casing.add(Embedding(output_dim=caseEmbeddings.shape[1], input_dim=caseEmbeddings.shape[0], input_length=n_in, weights=[caseEmbeddings], trainable=False))     
+casing.add(Embedding(input_dim=caseEmbeddings.shape[0], output_dim=caseEmbeddings.shape[1], input_length=n_in, weights=[caseEmbeddings], trainable=False))     
 casing.add(Flatten())
 
 model = Sequential()
 model.add(Merge([words, casing], mode='concat'))
 
-model.add(Dense(output_dim=n_hidden, activation='tanh'))
+model.add(Dense(output_dim=numHiddenUnits, activation='tanh'))
 model.add(Dense(output_dim=n_out, activation='softmax'))
 
 
 # Use Adam optimizer
 model.compile(loss='categorical_crossentropy', optimizer='adam')
 
+model.summary()
+
 # Train_y is a 1-dimensional vector containing the index of the label
 # With np_utils.to_categorical we map it to a 1 hot matrix
 train_y_cat = np_utils.to_categorical(train_y, n_out)

2016-11_Seminar/Session 1 - SENNA/code for POS/POS.py

@@ -38,7 +38,7 @@
 
 
 
-numHiddenUnits = 20
+numHiddenUnits = 100
 
 
 f = gzip.open('pkl/embeddings.pkl.gz', 'rb')
@@ -69,28 +69,29 @@
 
 # Create the train and predict_labels function
 n_in = train_tokens.shape[1]
-n_hidden = numHiddenUnits
 n_out = len(label2Idx)
 
 
 words = Sequential()
-words.add(Embedding(output_dim=wordEmbeddings.shape[1], input_dim=wordEmbeddings.shape[0], input_length=n_in,  weights=[wordEmbeddings], trainable=False))       
+words.add(Embedding(input_dim=wordEmbeddings.shape[0], output_dim=wordEmbeddings.shape[1], input_length=n_in,  weights=[wordEmbeddings], trainable=False))       
 words.add(Flatten())
 
 casing = Sequential()  
-casing.add(Embedding(output_dim=caseEmbeddings.shape[1], input_dim=caseEmbeddings.shape[0], input_length=n_in, weights=[caseEmbeddings], trainable=False))     
+casing.add(Embedding(input_dim=caseEmbeddings.shape[0], output_dim=caseEmbeddings.shape[1], input_length=n_in, weights=[caseEmbeddings], trainable=False))     
 casing.add(Flatten())
 
 model = Sequential()
 model.add(Merge([words, casing], mode='concat'))
 
-model.add(Dense(output_dim=n_hidden, activation='tanh'))
+model.add(Dense(output_dim=numHiddenUnits, activation='tanh'))
 model.add(Dense(output_dim=n_out, activation='softmax'))
 
 
 # Use Adam optimizer
 model.compile(loss='categorical_crossentropy', optimizer='adam')
 
+model.summary()
+
 # Train_y is a 1-dimensional vector containing the index of the label
 # With np_utils.to_categorical we map it to a 1 hot matrix
 train_y_cat = np_utils.to_categorical(train_y, n_out)

2016-11_Seminar/Session 3 - Relation CNN/code/CNN.py

@@ -16,7 +16,6 @@
 - Keras 1.1.1
 - Python 2.7
 """
-import theano
 import numpy as np
 np.random.seed(1337)  # for reproducibility
 
@@ -80,31 +79,23 @@
 wordModel.add(Embedding(embeddings.shape[0], embeddings.shape[1], input_length=sentenceTrain.shape[1], weights=[embeddings], trainable=False))
 
 
-convModel = Sequential()
-convModel.add(Merge([wordModel, distanceModel1, distanceModel2], mode='concat'))
+model = Sequential()
+model.add(Merge([wordModel, distanceModel1, distanceModel2], mode='concat'))
 
 
-convModel.add(Convolution1D(nb_filter=nb_filter,
+model.add(Convolution1D(nb_filter=nb_filter,
                         filter_length=filter_length,
                         border_mode='same',
                         activation='tanh',
                         subsample_length=1))
 # we use standard max over time pooling
-convModel.add(GlobalMaxPooling1D())
-
-
-model = convModel
-
-#model.add(Dropout(0.25))
-#model.add(Dense(hidden_dims,  activation='tanh', W_regularizer=keras.regularizers.l2(0.01)))
-model.add(Dropout(0.5))
-
+model.add(GlobalMaxPooling1D())
 
+model.add(Dropout(0.25))
 model.add(Dense(n_out, activation='softmax'))
 
 
 model.compile(loss='categorical_crossentropy',optimizer='Adam', metrics=['accuracy'])
-
 model.summary()
 print "Start training"