《Python自然语言处理》第08章-情感分析.pptVIP

载入数据 标签 分数 负面评价 =4 正面评价 =7 载入数据 引入 检查一下数据 结果 载入数据 结果 实战电影评论情感分析 0 3 辅助函数 实战电影评论情感分析 0 3 模型设置 batch_size=24 lstm_units=64 nun_labels=2 iterations=200000 import tensorflow as tf f.reset_default_graph() labels=tf.placeholder(tf.float32,[batch_size,nun_labels]) input_data=tf.placeholder(32,[batch_size,maxseq_num]) data=tf.Variable(tf.zeros([batch_size,maxSeqLength,numDimensions]),dtype=tf.float32) data=tf.nn.embedding_lookup(wordVectors,input_data) lstmCe11=tf.contrib.rnn.BasictsTMCell(lstm_units) lstmCe11=tf.contrib.rnn.Dropoutwrapper(ce11=lstmCe11,output_keep_Prob=0.75) value,_=tf.nn.dynanic_rnn(lstmCe11,input_data,dtype=tf.float32) 模型设置 weight = tf.Variable(tf.truncated_normal([lstm_units, num_labels])) bias = tf.Variable(tf.constant(0.1, shape = [num_labels])) value = tf.transpose(value, [1, 0, 2]) last = tf.gather(value, int(value.get_shape()[0]) - 1) prediction = (tf.matnul(last, weight) + bias) correct_pred=tf.equal(tf.argmax(prediction,1),tf.argmax(labels,1)) accuracy=tf.reduce_mean(tf.cast(correct_pred,tf.float32)) loss=tf.reduce_mean(tf.nn.softaax_cross_entropy_with_1ogits(logits=prediction,labels=labels)) optimizer=tf.train.Adamoptimizer().mininize(1088) 模型设置 import datetime sess=tf.InteractiveSession() tf.device(/gpu:0) saver=tf.train.Saver() sess.run(tf.global_variables_initializer()) tf.sumnary.scalar(Loss,loss) tf.summary.scalar(Accuracy,accuracy) merged=tf.summary.nerge_all() logdir=tensorboard/+datetime.datetime.now().strftime(%Y%m%d-%H%M%S)+/ writer=tf.sumnary.Filewriter(logdir,sess.graph) for i in range(iterations): #下个批次的数据 nextBatch,nextBatchtabels=get_train_batch() sess.run(optimizer,{input_data:nextBatch,labels:nextBatchtabels}) #每50次写入一次leadboard. if(i*50==0): summary= sess.run(merged,{input_data:nextBatch,labels:nextBatch}) writer.add_summary(summary,i) #每10000次保存一下模型 if(i % 10000==0 and i!=0): save_path=saver.save(sess,models/pretrained_lstm.ckpt,global_step=1) print(saved to %s%