def namer_gru_c2v():
# np.random.seed(1)
# X_train, y_train, char_to_index, index_to_char = load_training_data()
X_train, y_train, char_to_index, index_to_char = load_sample_training_data(
1)
char_num = len(char_to_index.keys())
print char_num
name_num = 30
name_len = 3
# first_name = u"宋"
# first_name = u"董"
first_name = u"陈"
if first_name not in char_to_index:
print "暂时不支持这个姓，Sorry！！！"
else:
print "支持这个姓，请稍等 ... ..."
model = gru.GRUTheano(char_num)
gru.load_model_parameters_theano(
'model/gru_c2v/GRU-2017-01-22-12-09-5264-48-128.dat.npz', model)
ignore_num = 0
i = 0
while i <= name_num:
name = [first_name]
try:
name = generate_name(model, name_len, first_name, X_train,
y_train, char_to_index, index_to_char)
except ValueError as ve:
ignore_num += 1
continue
print " ".join(name)
i += 1
print "ignored samples: ", ignore_num
def namer_rnn_one_hot():
# np.random.seed(1)
# X_train, y_train, char_to_index, index_to_char = load_training_data()
X_train, y_train, char_to_index, index_to_char = load_sample_training_data(
0)
char_num = len(char_to_index.keys())
print char_num
name_num = 10
name_len = 3
# first_name = u"宋"
# first_name = u"董"
first_name = u"陈"
if first_name not in char_to_index:
print "暂时不支持这个姓，Sorry！！！"
else:
print "支持这个姓，请稍等 ... ..."
model = rnn.RNNTheano(char_num)
# losses = train_with_sgd(model, X_train, y_train, nepoch=50)
# save_model_parameters_theano('./data/trained-model-theano.npz', model)
rnn.load_model_parameters_theano(
'model/rnn_one_hot/model-80-3296-2017-01-14-21-30-39.npz', model)
ignore_num = 0
i = 0
while i <= name_num:
name = [first_name]
try:
name = generate_name(model, name_len, first_name, X_train,
y_train, char_to_index, index_to_char)
except ValueError as ve:
ignore_num += 1
continue
print " ".join(name)
i += 1
print "ignored samples: ", ignore_num
def namer_lstm_c2v():
# np.random.seed(1)
# X_train, y_train, char_to_index, index_to_char = load_training_data()
X_train, y_train, char_to_index, index_to_char = load_sample_training_data(
1)
char_num = len(char_to_index.keys())
print char_num
name_num = 10
name_len = 3
# first_name = u"宋"
# first_name = u"董"
first_name = u"陈"
if first_name not in char_to_index:
print "暂时不支持这个姓，Sorry！！！"
else:
print "支持这个姓，请稍等 ... ..."
model = lstm.LSTM()
model.sample_name(
first_name=first_name,
ckpt_file="model/lstm_c2v/LSTM-2017-01-29-20-00-6569-400-128.ckpt")
print "Saved model parameters to %s." % outfile
def load_model_parameters_theano(path, model):
npzfile = np.load(path)
U, V, W = npzfile["U"], npzfile["V"], npzfile["W"]
model.hidden_dim = U.shape[0]
model.word_dim = U.shape[1]
model.U.set_value(U)
model.V.set_value(V)
model.W.set_value(W)
print "Loaded model parameters from %s. hidden_dim=%d word_dim=%d" % (path, U.shape[0], U.shape[1])
if __name__ == '__main__':
X_train, y_train, char_to_index, index_to_char = load_sample_training_data(1)
# X_train, y_train, char_to_index, index_to_char = load_training_data()
char_num = len(char_to_index.keys())
vocabulary_size = char_num
model = RNNTheano(vocabulary_size, hidden_dim=HIDDEN_DIM)
t1 = time.time()
model.sgd_step(X_train[10], y_train[10], LEARNING_RATE)
t2 = time.time()
print "SGD Step time: %f milliseconds" % ((t2 - t1) * 1000.)
if MODEL_FILE:
load_model_parameters_theano(MODEL_FILE, model)
train_with_sgd(model, X_train, y_train, nepoch=NEPOCH, learning_rate=LEARNING_RATE)