周杰伦简直陪伴了每个90后的青春，那如果AI写杰伦格调的歌词会写成怎么呢？

首先当然咱们须要筹备杰伦的歌词，这里一共收录了他的十几张专辑，近5000多行歌词。

原文档格局：

第一步数据预处理

def preprocess(data):    """    对文本中的字符进行替换，空格转换成逗号；换行变为句号。    """    data = data.replace(' ', '，')    data = data.replace('\n', '。')    words = jieba.lcut(data, cut_all=False) # 全模式切词    return words

解决后后果：

前10个词: ['想要', '有', '直升机', '。', '想要', '和', '你', '飞到', '宇宙', '去']

将解决完的数据写入内存并将文本转换完数字

# 结构词典及映射vocab = set(text)vocab_to_int = {w: idx for idx, w in enumerate(vocab)}int_to_vocab = {idx: w for idx, w in enumerate(vocab)}# 转换文本为整数int_text = [vocab_to_int[w] for w in text]

构建神经网络

a. 构建输出层

def get_inputs():    inputs = tf.placeholder(tf.int32, [None, None], name='inputs')    targets = tf.placeholder(tf.int32, [None, None], name='targets')    learning_rate = tf.placeholder(tf.float32, name='learning_rate')    return inputs, targets, learning_rate

b. 构建重叠RNN单元

其中rnn_size指的是RNN隐层神经元个数

def get_init_cell(batch_size, rnn_size):    lstm = tf.contrib.rnn.BasicLSTMCell(rnn_size)    cell = tf.contrib.rnn.MultiRNNCell([lstm])     initial_state = cell.zero_state(batch_size, tf.float32)    initial_state = tf.identity(initial_state, 'initial_state')    return cell, initial_state

c. Word Embedding

因为单词太多，所以须要进行embedding，模型中退出Embedding层来升高输出词的维度

def get_embed(input_data, vocab_size, embed_dim):    embedding = tf.Variable(tf.random_uniform([vocab_size, embed_dim], -1, 1))    embed = tf.nn.embedding_lookup(embedding, input_data)     return embed

d. 构建神经网络，将RNN层与全连贯层相连

其中cell为RNN单元； rnn_size: RNN隐层结点数量；input_data即input tensor；vocab_size：词汇表大小； embed_dim: 嵌入层大小

def build_nn(cell, rnn_size, input_data, vocab_size, embed_dim):    embed = get_embed(input_data, vocab_size, embed_dim)    outputs, final_state = build_rnn(cell, embed)     logits = tf.contrib.layers.fully_connected(outputs, vocab_size, activation_fn=None)     return logits, final_state

e. 结构batch

这里咱们依据batch_size和seq_length分为len//(batch_size*seq_length)个batch，每个batch蕴含输出和对应的指标输入

def get_batches(int_text, batch_size, seq_length):    '''    结构batch    '''    batch = batch_size * seq_length    n_batch = len(int_text) // batch     int_text = np.array(int_text[:batch * n_batch]) # 保留能形成残缺batch的数量     int_text_targets = np.zeros_like(int_text)    int_text_targets[:-1], int_text_targets[-1] = int_text[1:], int_text[0]     # 切分    x = np.split(int_text.reshape(batch_size, -1), n_batch, -1)    y = np.split(int_text_targets.reshape(batch_size, -1), n_batch, -1)     return np.stack((x, y), axis=1) # 组合

模型训练

from tensorflow.contrib import seq2seqtrain_graph = tf.Graph()with train_graph.as_default():    vocab_size = len(int_to_vocab) # vocab_size    input_text, targets, lr = get_inputs() # 输出tensor    input_data_shape = tf.shape(input_text)    # 初始化RNN    cell, initial_state = get_init_cell(input_data_shape[0], rnn_size)    logits, final_state = build_nn(cell, rnn_size, input_text, vocab_size, embed_dim)    # 计算softmax层概率    probs = tf.nn.softmax(logits, name='probs')    # 损失函数    cost = seq2seq.sequence_loss(        logits,        targets,        tf.ones([input_data_shape[0], input_data_shape[1]]))    # 优化函数    optimizer = tf.train.AdamOptimizer(lr)    # Gradient Clipping    gradients = optimizer.compute_gradients(cost)    capped_gradients = [(tf.clip_by_value(grad, -1., 1.), var) for grad, var in gradients if grad is not None]    train_op = optimizer.apply_gradients(capped_gradients)

训练后果

Epoch  72 Batch   24/33   train_loss = 0.108Epoch  75 Batch   25/33   train_loss = 0.104Epoch  78 Batch   26/33   train_loss = 0.096Epoch  81 Batch   27/33   train_loss = 0.111Epoch  84 Batch   28/33   train_loss = 0.119Epoch  87 Batch   29/33   train_loss = 0.130Epoch  90 Batch   30/33   train_loss = 0.141Epoch  93 Batch   31/33   train_loss = 0.138Epoch  96 Batch   32/33   train_loss = 0.153Model Trained and Saved

train_loss还不错，不过可能过拟合了。

最初让咱们加载模型，看看生成状况

    # 加载模型    loader = tf.train.import_meta_graph(save_dir + '.meta')    loader.restore(sess, save_dir)    # 获取训练的后果参数    input_text, initial_state, final_state, probs = get_tensors(loaded_graph)    # Sentences generation setup    gen_sentences = [prime_word]    prev_state = sess.run(initial_state, {input_text: np.array([[1]])})    # 生成句子    for n in range(gen_length):        dyn_input = [[vocab_to_int[word] for word in gen_sentences[-seq_length:]]]        dyn_seq_length = len(dyn_input[0])        # 预测        probabilities, prev_state = sess.run(            [probs, final_state],            {input_text: dyn_input, initial_state: prev_state})        # 抉择单词进行文本生成，用来以肯定的概率生成下一个词        pred_word = pick_word(probabilities[0][dyn_seq_length - 1], int_to_vocab)        gen_sentences.append(pred_word)

哎哟不错哦！

最初的最初我还扩充了歌词库，这次引入了更多风行歌手，来看看成果吧。

如同更不错了！

如果你也喜爱杰伦，请点赞并分享生成的歌词。

点击关注，第一工夫理解华为云陈腐技术~