################### DEFINING HYPERPARAMETERS ###################
dimx = 50
dimy = 50
dimft = 44
batch_size = 70
vocab_size = 8000
embedding_dim = 50
LSTM_neurons = 64
depth = 1
nb_epoch = 3
shared = 1
opt_params = [0.001, 'adam']
ques, ans, label_train, train_len, test_len,\
wordVec_model, res_fname, pred_fname, feat_train, feat_test = wk.load_wiki(model_name, glove_fname)
data_l, data_r, embedding_matrix = dl.process_data(ques,
ans,
wordVec_model,
dimx=dimx,
dimy=dimy,
vocab_size=vocab_size,
embedding_dim=embedding_dim)
X_train_l, X_test_l, X_dev_l, X_train_r, X_test_r, X_dev_r = wk.prepare_train_test(
data_l, data_r, train_len, test_len)
lrmodel = lrmodel(embedding_matrix,
dimx=dimx,
dimy=dimy,
LSTM_neurons=LSTM_neurons,
model_name = lrmodel.func_name
################### DEFINING HYPERPARAMETERS ###################
dimx = 60
dimy = 60
dimft = 44
batch_size = 50
vocab_size = 8000
embedding_dim = 50
nb_filter = 120,
filter_length = (50, 4)
depth = 1
nb_epoch = 3
ques, ans, label_train, train_len, test_len, wordVec_model, res_fname, pred_fname, feat_train, feat_test = wk.load_wiki(
model_name, glove_fname)
data_l, data_r, embedding_matrix = dl.process_data(ques,
ans,
wordVec_model,
dimx=dimx,
dimy=dimy,
vocab_size=vocab_size,
embedding_dim=embedding_dim)
X_train_l, X_test_l, X_dev_l, X_train_r, X_test_r, X_dev_r = wk.prepare_train_test(
data_l, data_r, train_len, test_len)
if model_name == 'cnn_ft':
lrmodel = lrmodel(embedding_matrix,
dimx=dimx,
dimy=dimy,