最近，在研究深度学习方面的知识，结合Tensorflow，完成了基于lstm的小说预测程序demo。
lstm是改进的RNN，具有长期记忆功能，相对于RNN,增加了多个门来控制输入与输出。原理方面的知识网上很多，在此，我只是将我短暂学习的tensorflow写一个预测小说的demo,如果有错误，还望大家指出。
1、将小说进行分词，去除空格，建立词汇表与id的字典，生成初始输入模型的x与y
def readfile(file_path): f = codecs.open(file_path, 'r', 'utf-8') alltext = f.read() alltext = re.sub(r'\s','', alltext) seglist = list(jieba.cut(alltext, cut_all = False)) return seglist def _build_vocab(filename): data = readfile(filename) counter = collections.Counter(data) count_pairs = sorted(counter.items(), key=lambda x: (-x[1], x[0])) words, _ = list(zip(*count_pairs)) word_to_id = dict(zip(words, range(len(words)))) id_to_word = dict(zip(range(len(words)),words)) dataids = [] for w in data: dataids.append(word_to_id[w]) return word_to_id, id_to_word,dataids def dataproducer(batch_size, num_steps): word_to_id, id_to_word, data = _build_vocab('F:\\ml\\code\\lstm\\1.txt') datalen = len(data) batchlen = datalen//batch_size epcho_size = (batchlen - 1)//num_steps data = tf.reshape(data[0: batchlen*batch_size], [batch_size,batchlen]) i = tf.train.range_input_producer(epcho_size, shuffle=False).dequeue() x = tf.slice(data, [0,i*num_steps],[batch_size, num_steps]) y = tf.slice(data, [0,i*num_steps+1],[batch_size, num_steps]) x.set_shape([batch_size, num_steps]) y.set_shape([batch_size, num_steps]) return x,y,id_to_word
2、建立lstm模型：
lstm_cell = tf.nn.rnn_cell.BasicLSTMCell(size, forget_bias = 0.5) lstm_cell = tf.nn.rnn_cell.DropoutWrapper(lstm_cell, output_keep_prob = keep_prob) cell = tf.nn.rnn_cell.MultiRNNCell([lstm_cell], num_layers)
3、根据训练数据输出误差反向调整模型
with tf.variable_scope("Model", reuse = None, initializer = initializer):#tensorflow主要通过变量空间来实现共享变量 with tf.variable_scope("r", reuse = None, initializer = initializer): softmax_w = tf.get_variable('softmax_w', [size, vocab_size]) softmax_b = tf.get_variable('softmax_b', [vocab_size]) with tf.variable_scope("RNN", reuse = None, initializer = initializer): for time_step in range(num_steps): if time_step > 0: tf.get_variable_scope().reuse_variables() (cell_output, state) = cell(inputs[:, time_step, :], state,) outputs.append(cell_output) output = tf.reshape(outputs, [-1,size]) logits = tf.matmul(output, softmax_w) + softmax_b loss = tf.nn.seq2seq.sequence_loss_by_example([logits], [tf.reshape(targets,[-1])], [tf.ones([batch_size*num_steps])]) global_step = tf.Variable(0) learning_rate = tf.train.exponential_decay( 10.0, global_step, 5000, 0.1, staircase=True) optimizer = tf.train.GradientDescentOptimizer(learning_rate) gradients, v = zip(*optimizer.compute_gradients(loss)) gradients, _ = tf.clip_by_global_norm(gradients, 1.25) optimizer = optimizer.apply_gradients(zip(gradients, v), global_step=global_step)
4、预测新一轮输出
teststate = test_initial_state (celloutput,teststate)= cell(test_inputs, teststate) partial_logits = tf.matmul(celloutput, softmax_w) + softmax_b partial_logits = tf.nn.softmax(partial_logits)
5、根据之前建立的操作，运行tensorflow会话
sv = tf.train.Supervisor(logdir=None) with sv.managed_session() as session: costs = 0 iters = 0 for i in range(1000): _,l= session.run([optimizer, cost]) costs += l iters +=num_steps perplextity = np.exp(costs / iters) if i == 0: print(perplextity) if i0 == 0: p = random_distribution() b = sample(p) sentence = id_to_word[b[0]] for j in range(200): test_output = session.run(partial_logits, feed_dict={test_input:b}) b = sample(test_output) sentence += id_to_word[b[0]] print(sentence)
其中，使用sv.managed_session()后，在此会话间，将不能修改graph。如果采用普通的session,程序将会阻塞于session.run(),对于这个问题，我还是很疑惑，希望理解的人帮忙解答下。
代码地址位于https://github.com/summersunshine1/datamining/tree/master/lstm，运行时只需将readdata中文件路径修改即可。作为深度学习的入门小白，希望大家多多指点。
运行结果如下：
