text_gen.py

# coding: utf-8

from __future__ import print_function
from keras.models import Sequential
from keras.layers import Dense, Activation
from keras.layers import LSTM
from keras.optimizers import RMSprop
from keras.utils.data_utils import get_file
import numpy as np
import random
import sys,os
import json

modelNumber=7

while 'indices_char_%d.dat' % modelNumber in os.listdir('.'):
    modelNumber+=1

print('ModelNumber='+str(modelNumber))


frac=4

with open('un_speeches/compiled_no_breaks.txt','r') as inFile:
    text=inFile.read()

for nFrac in range(frac):
    print(nFrac*int(len(text)/frac),(nFrac+1)+int(len(text)/frac))
sys.exit(1)


text=text[0:int(len(text)/frac)]
print('corpus length:', len(text))

chars = sorted(list(set(text)))
print('total chars:', len(chars))
char_indices = dict((c, i) for i, c in enumerate(chars))
indices_char = dict((i, c) for i, c in enumerate(chars))


# In[3]:


# cut the text in semi-redundant sequences of maxlen characters
maxlen = 30
step = 5

if False:
    sentences = []
    next_chars = []
    for i in range(0, len(text) - maxlen, step):
        sentences.append(text[i: i + maxlen])
        next_chars.append(text[i + maxlen])
    print('nb sequences:', len(sentences))

else:
    sentences=[text[i:i+maxlen] for i in range(0,len(text)-maxlen,step)]
#next_chars=[sentences[maxlen+i] for i in range(0,len(text)-maxlen,step)]
    next_chars=[text[i] for i in range(maxlen,len(text),step)]

print('Steps: %d/%d' %(len(text),step))
print('Range:'+','.join(map(str,range(0,len(text),step)[0:5])))
print('nb sequences:', len(sentences))

for i in range(5):
    print(sentences[i])
    print(next_chars[i])
    print('')


print('Vectorization...')
X = np.zeros((len(sentences), maxlen, len(chars)), dtype=np.bool)
y = np.zeros((len(sentences), len(chars)), dtype=np.bool)
for i, sentence in enumerate(sentences):
    for t, char in enumerate(sentence):
        X[i, t, char_indices[char]] = 1
    y[i, char_indices[next_chars[i]]] = 1


# In[5]:


print('Build model...')
model = Sequential()
model.add(LSTM(128, input_shape=(maxlen,len(chars))))
model.add(Dense(len(chars)))
model.add(Activation('softmax'))


# In[6]:


optimizer = RMSprop(lr=0.01)
model.compile(loss='categorical_crossentropy', optimizer=optimizer)


# In[ ]:


def sample(preds, temperature=1.0):
    # helper function to sample an index from a probability array
    preds = np.asarray(preds).astype('float64')
    preds = np.log(preds) / temperature
    exp_preds = np.exp(preds)
    preds = exp_preds / np.sum(exp_preds)
    probas = np.random.multinomial(1, preds, 1)
    return np.argmax(probas)


# In[ ]:


# train the model, output generated text after each iteration
for iteration in range(1, 2):
    print()
    print('-' * 50)
    print('Iteration', iteration)
    model.fit(X, y,
              batch_size=128,epochs=1
              )

    # Use fit_generator here to process in chunks
    # https://keras.io/models/sequential/


    start_index = random.randint(0, len(text) - maxlen - 1)

    for diversity in [0.2, 0.35, 0.5, 0.75, 1.0, 1.2]:
        print()
        print('----- diversity:', diversity)

        generated = ''
        sentence = text[start_index: start_index + maxlen]
        generated += sentence
        print('----- Generating with seed: "' + sentence + '"')
        sys.stdout.write(generated)

        for i in range(400):
            x = np.zeros((1, maxlen, len(chars)))
            for t, char in enumerate(sentence):
                x[0, t, char_indices[char]] = 1.

            preds = model.predict(x, verbose=0)[0]
            next_index = sample(preds, diversity)
            next_char = indices_char[next_index]

            generated += next_char
            sentence = sentence[1:] + next_char

            sys.stdout.write(next_char)
            sys.stdout.flush()
        print()

    model.save('rnn_7_%d.hdfs' % (modelNumber))

    import pickle

    with open('indices_char_%d.dat' % modelNumber,'w') as outFile:
        pickle.dump(indices_char,outFile)
    with open('char_indices_%d.dat' % modelNumber,'w') as outFile:
        pickle.dump(char_indices,outFile)
    with open('config_%d.csv' % modelNumber,'w') as outFile:
        outFile.write('\n'.join(map(str,[maxlen,step,frac,len(sentences)])))
    with open('chars_%d.csv' % modelNumber,'w') as outFile:
        pickle.dump(chars,outFile)