Tensorflow：基于LSTM生成藏头诗

最近在学习TensorFlow，学习到了RNN这一块，相关的资料不是很多，了解到使用RNN可以生成藏头诗之后，我就决定拿这个下手啦！
本文不介绍RNN以及LSTM的相关基本知识，如要了解，请自行百度。
本文是在学习了 TensorFlow7: 基于RNN生成古诗词
这一篇博客之后继续在其基础上修改的代码，若要了解相关内容可以先去上面的博客去看一下。

【注：本博客所使用的数据、代码、模型文件均已放在百度云上：
链接: https://pan.baidu.com/s/1qY4mt1y 密码: 47y2】

模型生成

首先我们要训练好模型。这里采用的是2层的LSTM框架，每层有128个隐藏层节点，batch_size设为64。训练数据来源于全唐诗（可在上面百度云资源分享当中找到）。特别注意到的一点是这里每训练完一次就对训练数据做shuffle。
源代码如下：

#!/usr/bin/python3
#-*- coding: UTF-8 -*-
import collections
import numpy as np
import tensorflow as tf  #-------------------------------数据预处理---------------------------#poetry_file ='poetry.txt'  # 诗集
poetrys = []
with open(poetry_file, "r") as f:  for line in f:  try:  line = line.decode('UTF-8')line = line.strip(u'\n')title, content = line.strip(u' ').split(u':')  content = content.replace(u' ',u'')  if u'_' in content or u'(' in content or u'（' in content or u'《' in content or u'[' in content:  continue  if len(content) < 5 or len(content) > 79:  continue  content = u'[' + content + u']'  poetrys.append(content)  except Exception as e:   pass  # 按诗的字数排序
poetrys = sorted(poetrys,key=lambda line: len(line))
print('唐诗总数: ', len(poetrys))  # 统计每个字出现次数
all_words = []
for poetry in poetrys:  all_words += [word for word in poetry]
counter = collections.Counter(all_words)
count_pairs = sorted(counter.items(), key=lambda x: -x[1])
words, _ = zip(*count_pairs)  # 取前多少个常用字
words = words[:len(words)] + (' ',)
# 每个字映射为一个数字ID
word_num_map = dict(zip(words, range(len(words))))
# 把诗转换为向量形式，参考TensorFlow练习1
to_num = lambda word: word_num_map.get(word, len(words))
poetrys_vector = [ list(map(to_num, poetry)) for poetry in poetrys]
#[[314, 3199, 367, 1556, 26, 179, 680, 0, 3199, 41, 506, 40, 151, 4, 98, 1],
#[339, 3, 133, 31, 302, 653, 512, 0, 37, 148, 294, 25, 54, 833, 3, 1, 965, 1315, 377, 1700, 562, 21, 37, 0, 2, 1253, 21, 36, 264, 877, 809, 1]
#....]# 每次取64首诗进行训练
batch_size = 64
n_chunk = len(poetrys_vector) // batch_size  class DataSet(object):def __init__(self,data_size):self._data_size = data_sizeself._epochs_completed = 0self._index_in_epoch = 0self._data_index = np.arange(data_size)def next_batch(self,batch_size):start = self._index_in_epochif start + batch_size > self._data_size:np.random.shuffle(self._data_index)self._epochs_completed = self._epochs_completed + 1self._index_in_epoch = batch_sizefull_batch_features ,full_batch_labels = self.data_batch(0,batch_size)return full_batch_features ,full_batch_labels else:self._index_in_epoch += batch_sizeend = self._index_in_epochfull_batch_features ,full_batch_labels = self.data_batch(start,end)if self._index_in_epoch == self._data_size:self._index_in_epoch = 0self._epochs_completed = self._epochs_completed + 1np.random.shuffle(self._data_index)return full_batch_features,full_batch_labelsdef data_batch(self,start,end):batches = []for i in range(start,end):batches.append(poetrys_vector[self._data_index[i]])length = max(map(len,batches))xdata = np.full((end - start,length), word_num_map[' '], np.int32)  for row in range(end - start):  xdata[row,:len(batches[row])] = batches[row]  ydata = np.copy(xdata)  ydata[:,:-1] = xdata[:,1:]  return xdata,ydata#---------------------------------------RNN--------------------------------------#input_data = tf.placeholder(tf.int32, [batch_size, None])
output_targets = tf.placeholder(tf.int32, [batch_size, None])
# 定义RNN
def neural_network(model='lstm', rnn_size=128, num_layers=2):  if model == 'rnn':  cell_fun = tf.nn.rnn_cell.BasicRNNCell  elif model == 'gru':  cell_fun = tf.nn.rnn_cell.GRUCell  elif model == 'lstm':  cell_fun = tf.nn.rnn_cell.BasicLSTMCell  cell = cell_fun(rnn_size, state_is_tuple=True)  cell = tf.nn.rnn_cell.MultiRNNCell([cell] * num_layers, state_is_tuple=True)  initial_state = cell.zero_state(batch_size, tf.float32)  with tf.variable_scope('rnnlm'):  softmax_w = tf.get_variable("softmax_w", [rnn_size, len(words)])  softmax_b = tf.get_variable("softmax_b", [len(words)])  with tf.device("/cpu:0"):  embedding = tf.get_variable("embedding", [len(words), rnn_size])  inputs = tf.nn.embedding_lookup(embedding, input_data)  outputs, last_state = tf.nn.dynamic_rnn(cell, inputs, initial_state=initial_state, scope='rnnlm')  output = tf.reshape(outputs,[-1, rnn_size])  logits = tf.matmul(output, softmax_w) + softmax_b  probs = tf.nn.softmax(logits)  return logits, last_state, probs, cell, initial_state def load_model(sess, saver,ckpt_path):latest_ckpt = tf.train.latest_checkpoint(ckpt_path)if latest_ckpt:print ('resume from', latest_ckpt)saver.restore(sess, latest_ckpt)return int(latest_ckpt[latest_ckpt.rindex('-') + 1:])else:print ('building model from scratch')sess.run(tf.global_variables_initializer())return -1#训练
def train_neural_network():  logits, last_state, _, _, _ = neural_network()  targets = tf.reshape(output_targets, [-1])  loss = tf.nn.seq2seq.sequence_loss_by_example([logits], [targets], [tf.ones_like(targets, dtype=tf.float32)], len(words))  cost = tf.reduce_mean(loss)  learning_rate = tf.Variable(0.0, trainable=False)  tvars = tf.trainable_variables()  grads, _ = tf.clip_by_global_norm(tf.gradients(cost, tvars), 5)  #optimizer = tf.train.GradientDescentOptimizer(learning_rate)optimizer = tf.train.AdamOptimizer(learning_rate)   train_op = optimizer.apply_gradients(zip(grads, tvars))  Session_config = tf.ConfigProto(allow_soft_placement=True)Session_config.gpu_options.allow_growth = True  trainds = DataSet(len(poetrys_vector))with tf.Session(config=Session_config) as sess:with tf.device('/gpu:2'):  sess.run(tf.initialize_all_variables())  saver = tf.train.Saver(tf.all_variables())last_epoch = load_model(sess, saver,'model/') for epoch in range(last_epoch + 1,100):sess.run(tf.assign(learning_rate, 0.002 * (0.97 ** epoch)))  #sess.run(tf.assign(learning_rate, 0.01))all_loss = 0.0 for batche in range(n_chunk): x,y = trainds.next_batch(batch_size)train_loss, _ , _ = sess.run([cost, last_state, train_op], feed_dict={input_data: x, output_targets: y})  all_loss = all_loss + train_loss if batche % 50 == 1:#print(epoch, batche, 0.01,train_loss)print(epoch, batche, 0.002 * (0.97 ** epoch),train_loss) saver.save(sess, 'model/poetry.module', global_step=epoch) print (epoch,' Loss: ', all_loss * 1.0 / n_chunk) train_neural_network()

使用该代码会将训练好的模型参数保存在 “model” 文件夹下。经过100个epoch之后，平均loss会降到2.6左右。训练好的模型也已经放在了上面分享的百度云资源当中。

生成古诗

使用训练好的模型可以轻松生成各种古诗。
下面就是几个例子：

生成藏头诗

上代码：

#!/usr/bin/python3
#-*- coding: UTF-8 -*-
import collections
import numpy as np
import tensorflow as tf
'''
This one will produce a poetry with heads.
'''#-------------------------------数据预处理---------------------------#poetry_file ='poetry.txt'  # 诗集
poetrys = []
with open(poetry_file, "r") as f:  for line in f:  try:  line = line.decode('UTF-8')line = line.strip(u'\n')title, content = line.strip(u' ').split(u':')  content = content.replace(u' ',u'')  if u'_' in content or u'(' in content or u'（' in content or u'《' in content or u'[' in content:  continue  if len(content) < 5 or len(content) > 79:  continue  content = u'[' + content + u']'  poetrys.append(content)  except Exception as e:   pass  # 按诗的字数排序
poetrys = sorted(poetrys,key=lambda line: len(line))
print('唐诗总数: ', len(poetrys))  # 统计每个字出现次数
all_words = []
for poetry in poetrys:  all_words += [word for word in poetry]
counter = collections.Counter(all_words)
count_pairs = sorted(counter.items(), key=lambda x: -x[1])
words, _ = zip(*count_pairs)  # 取前多少个常用字
words = words[:len(words)] + (' ',)
# 每个字映射为一个数字ID
word_num_map = dict(zip(words, range(len(words))))
# 把诗转换为向量形式，参考TensorFlow练习1
to_num = lambda word: word_num_map.get(word, len(words))
poetrys_vector = [ list(map(to_num, poetry)) for poetry in poetrys]
#[[314, 3199, 367, 1556, 26, 179, 680, 0, 3199, 41, 506, 40, 151, 4, 98, 1],
#[339, 3, 133, 31, 302, 653, 512, 0, 37, 148, 294, 25, 54, 833, 3, 1, 965, 1315, 377, 1700, 562, 21, 37, 0, 2, 1253, 21, 36, 264, 877, 809, 1]
#....]# 每次取64首诗进行训练
batch_size = 1
n_chunk = len(poetrys_vector) // batch_size  class DataSet(object):def __init__(self,data_size):self._data_size = data_sizeself._epochs_completed = 0self._index_in_epoch = 0self._data_index = np.arange(data_size)def next_batch(self,batch_size):start = self._index_in_epochif start + batch_size > self._data_size:np.random.shuffle(self._data_index)self._epochs_completed = self._epochs_completed + 1self._index_in_epoch = batch_sizefull_batch_features ,full_batch_labels = self.data_batch(0,batch_size)return full_batch_features ,full_batch_labels else:self._index_in_epoch += batch_sizeend = self._index_in_epochfull_batch_features ,full_batch_labels = self.data_batch(start,end)if self._index_in_epoch == self._data_size:self._index_in_epoch = 0self._epochs_completed = self._epochs_completed + 1np.random.shuffle(self._data_index)return full_batch_features,full_batch_labelsdef data_batch(self,start,end):batches = []for i in range(start,end):batches.append(poetrys_vector[self._data_index[i]])length = max(map(len,batches))xdata = np.full((end - start,length), word_num_map[' '], np.int32)  for row in range(end - start):  xdata[row,:len(batches[row])] = batches[row]  ydata = np.copy(xdata)  ydata[:,:-1] = xdata[:,1:]  return xdata,ydata#---------------------------------------RNN--------------------------------------#input_data = tf.placeholder(tf.int32, [batch_size, None])
output_targets = tf.placeholder(tf.int32, [batch_size, None])
# 定义RNN
def neural_network(model='lstm', rnn_size=128, num_layers=2):  if model == 'rnn':  cell_fun = tf.nn.rnn_cell.BasicRNNCell  elif model == 'gru':  cell_fun = tf.nn.rnn_cell.GRUCell  elif model == 'lstm':  cell_fun = tf.nn.rnn_cell.BasicLSTMCell  cell = cell_fun(rnn_size, state_is_tuple=True)  cell = tf.nn.rnn_cell.MultiRNNCell([cell] * num_layers, state_is_tuple=True)  initial_state = cell.zero_state(batch_size, tf.float32)  with tf.variable_scope('rnnlm'):  softmax_w = tf.get_variable("softmax_w", [rnn_size, len(words)])  softmax_b = tf.get_variable("softmax_b", [len(words)])  with tf.device("/cpu:0"):  embedding = tf.get_variable("embedding", [len(words), rnn_size])  inputs = tf.nn.embedding_lookup(embedding, input_data)  outputs, last_state = tf.nn.dynamic_rnn(cell, inputs, initial_state=initial_state, scope='rnnlm')  output = tf.reshape(outputs,[-1, rnn_size])  logits = tf.matmul(output, softmax_w) + softmax_b  probs = tf.nn.softmax(logits)  return logits, last_state, probs, cell, initial_state #-------------------------------生成古诗---------------------------------#
# 使用训练完成的模型def gen_head_poetry(heads, type):if type != 5 and type != 7:print 'The second para has to be 5 or 7!'return   def to_word(weights):  t = np.cumsum(weights)  s = np.sum(weights)  sample = int(np.searchsorted(t, np.random.rand(1)*s))  return words[sample]  _, last_state, probs, cell, initial_state = neural_network()Session_config = tf.ConfigProto(allow_soft_placement = True)Session_config.gpu_options.allow_growth=True with tf.Session(config=Session_config) as sess: with tf.device('/gpu:1'): sess.run(tf.initialize_all_variables())  saver = tf.train.Saver(tf.all_variables())  saver.restore(sess, 'model/poetry.module-99')poem = ''for head in  heads:flag = Truewhile flag:state_ = sess.run(cell.zero_state(1, tf.float32)) x = np.array([list(map(word_num_map.get, u'['))])  [probs_, state_] = sess.run([probs, last_state], feed_dict={input_data: x, initial_state: state_})  sentence = headx = np.zeros((1,1))  x[0,0] = word_num_map[sentence]  [probs_, state_] = sess.run([probs, last_state], feed_dict={input_data: x, initial_state: state_})  word = to_word(probs_)sentence += word  while word != u'。':  x = np.zeros((1,1))  x[0,0] = word_num_map[word]  [probs_, state_] = sess.run([probs, last_state], feed_dict={input_data: x, initial_state: state_})  word = to_word(probs_)sentence += word  if len(sentence) == 2 + 2 * type:sentence += u'\n'poem += sentenceflag = Falsereturn poemprint(gen_head_poetry(u'天下之大',5))

最后从函数接口可以看到，除了可以自己定义诗的头外，还可以定义是五言绝句还是七言绝句。
来看几个五言绝句的例子：

再来看几个七言绝句的例子：

那么是不是可以用它来写情诗呢？
当然可以啦！

参考文献：

[1]. http://blog.csdn.net/u014365862/article/details/53868544