def main():
args = getArguments()
HIDDEN = args.hidden_size
EPOCHS = args.epoch_size
BATCH_SIZE = args.batch_size
#
# prepare DATA
#
print "Load Data"
X_tr = np.load(args.input)
y_tr = np.load(args.output) if args.input != args.output else X_tr
sx, sy, sz = X_tr.shape
split = int(sx * 0.8)
X_val = X_tr[split:,:,:]
X_tr = X_tr[:split,:,:]
y_val = y_tr[split:,:,:]
y_tr = y_tr[:split,:,:]
if args.mode == "train":
print "Define RNN"
inputs = Input(shape=(sy,sz))
encoded = LSTM(HIDDEN,
activation="relu",
init="normal")(inputs)
decoded = RepeatVector(sy)(encoded)
decoded = LSTM(sz,
return_sequences=True,
activation="softmax",
init="normal")(decoded)
autoencoder = Model(inputs,decoded)
elif args.mode == "retrain":
print "load existing model: ", args.model
autoencoder = load_model(args.model)
print "Compile"
autoencoder.compile(optimizer="rmsprop", loss="categorical_crossentropy")
print "Train"
autoencoder.fit(X_tr, y_tr,
shuffle=True,
nb_epoch=EPOCHS,
batch_size=BATCH_SIZE,
validation_data=(X_val, y_val),
callbacks=[
ModelCheckpoint(args.model,save_best_only=True),
#EarlyStopping(patience=20)
]
)
def main():
args = getArguments()
SPLIT = 100
#
# prepare DATA
#
print "Load Data"
X, char_dict = loadData()
X = X[:100]
print "load existing model: ", args.model
autoencoder = load_model(args.model)
y_hat = autoencoder.predict(X)
for idx in range(X.shape[0]):
x1 = decode(np.argmax(X[idx], axis=1), char_dict)
x2 = decode(np.argmax(y_hat[idx], axis=1), char_dict)
print "".join(x1), "".join(x2)
raw_input()
def main():
args = getArguments()
SPLIT = 100
#
# prepare DATA
#
print "Load Data"
X, char_dict = loadData()
X = X[:100]
print "load existing model: ", args.model
autoencoder = load_model(args.model)
y_hat = autoencoder.predict(X)
for idx in range(X.shape[0]):
x1 = decode(np.argmax(X[idx], axis=1),char_dict)
x2 = decode(np.argmax(y_hat[idx],axis=1),char_dict)
print "".join(x1), "".join(x2)
raw_input()
v_init = get_var_from("initial_state", model.shared_variables)
v_states = get_var_from("H_apply_states", model.intermediary_variables)
#v_states = get_var_from("H_apply_states",model.intermediary_variables)
f = theano.function([v_inchar],
v_softmax,
updates=[(v_init, v_states[0][0])])
#f = theano.function([v_inchar], v_softmax)
seq = [init_char]
for _ in xrange(num_chars):
dist = f(np.atleast_2d(seq[-1]).astype(np.int32))[0]
sample = np.random.choice(vocab_size, 1, p=dist)[0]
seq.append(sample)
#print seq
return seq
def sample_text(model, num_chars, corpus):
return "".join(
corpus.decode(sample_chars(model, num_chars, corpus.vocab_size())))
corpus = Corpus(open("corpus.txt").read())
args = getArguments()
main_loop = load(args.model)
model = main_loop.model
#print sample_text(model, args.sample_size, corpus)
v_inchar = get_var_from("inchar", model.variables)
v_softmax = get_var_from("softmax_apply_output", model.variables)
v_init = get_var_from("initial_state", model.shared_variables)
v_states = get_var_from("H_apply_states", model.intermediary_variables)
# v_states = get_var_from("H_apply_states",model.intermediary_variables)
f = theano.function([v_inchar], v_softmax, updates=[(v_init, v_states[0][0])])
# f = theano.function([v_inchar], v_softmax)
seq = [init_char]
for _ in xrange(num_chars):
dist = f(np.atleast_2d(seq[-1]).astype(np.int32))[0]
sample = np.random.choice(vocab_size, 1, p=dist)[0]
seq.append(sample)
# print seq
return seq
def sample_text(model, num_chars, corpus):
return "".join(corpus.decode(sample_chars(model, num_chars, corpus.vocab_size())))
corpus = Corpus(open("corpus.txt").read())
args = getArguments()
main_loop = load(args.model)
model = main_loop.model
# print sample_text(model, args.sample_size, corpus)