tensorflow tensorboard的入门

• W = tf.Variable(tf.truncated_normal([hidden_size, class_num], stddev=0.1), dtype=tf.float32,name='W')
  tf.summary.histogram('W',W)#把w记录在histogram上查看

  with tf.name_scope('loss'):
      cross_entropy = -tf.reduce_mean(y * tf.log(y_pre))
      tf.summary.scalar('loss',cross_entropy)#把loss记录在scalar上查看

  merged=tf.summary.merge_all()#合并所有需要在tensorborad上看的，需要sess.run
  writer=tf.summary.FileWriter("logs/",sess.graph)#保存，路径tensorboard查看整个框架

  for i in range(2000):
      _batch_size=32
      batch=mnist.train.next_batch(_batch_size)
      
      if (i+1)%200 ==0:
          result=sess.run(merged,feed_dict={_X:batch[0],y:batch[1],keep_prob:0.8,batch_size:_batch_size})#把merge搞起来
          writer.add_summary(result,i)#i是步数
          train_accuracy=sess.run(accuracy,feed_dict={_X:batch[0],y:batch[1],keep_prob:0.8,batch_size:_batch_size})
          print ("Iter%d, step %d, training accuracy %g" % ( mnist.train.epochs_completed, (i+1), train_accuracy))
      sess.run(train_op, feed_dict={_X: batch[0], y: batch[1], keep_prob: 0.5, batch_size: _batch_size})
  
  

  命令行运行
  
  
  tensorboard --logdir='tensorborad/'
  
  
  
  点网址。点不开手动复制，不显示可以把名字改成127.0.0.1