Tensorflow中执行tensorboard --logdir=demo-result 命令报错
2018-03-02 15:32
501 查看
在上文tensorflow框架搭建完成的基础上,我们用一个实例来再次感受一下tensorflow的使用过程,这也算是我tensorflow的入门实例.
下面直接贴出代码:
当在终端中输入:tensorboard --logdir=demo-result 时,可能会出现以下三种错误:tensorflow:IOError [Errno 2] No such file or directory: '/home/zgm/anaconda2/envs/tensorflow/lib/python2.7/site-packages/tensorflow/tensorboard/TAG' on path /home/zgm/anaconda2/envs/tensorflow/lib/python2.7/site-packages/tensorflow/tensorboard/TAG解决方法:下载tensorflow的github的源代码,将
4000
tensorflow-0.8.0.data-->purelib-->tensorflow-->tensorboard目录下的TAG文件拷贝到python下面的tensorboard目录下即可
tensorflow:IOError [Errno 2] No such file or directory: '/home/zgm/anaconda2/envs/tensorflow/lib/python2.7/site-packages/tensorflow/tensorboard/lib/css/global.css' on path /home/zgm/anaconda2/envs/tensorflow/lib/python2.7/site-packages/tensorflow/tensorboard/lib/css/global.css解决方法:下载tensorflow的github的源代码,将tensorflow-0.8.0.data-->purelib-->tensorflow-->tensorboard-->lib目录下的CSS文件夹拷贝到python下面的tensorboard目录下的lib中即可
ERROR:tensorflow:Tried to connect to port 6006, but address is in use.
Tried to connect to port 6006, but address is in use.
进程号为9445占用了端口6006,kill掉就好.
下面打开链接就可以看到可视化情况.
至此结束!
下面直接贴出代码:
# coding:utf-8 # 调用tensorflow import tensorflow as tf import numpy as np # 这里生成了100对数字,作为整个神经网络的input x_data = np.random.rand(100).astype("float32") # 使用with,让我们的数据以节点的方式落在tensorflow的报告上。 with tf.name_scope('y_data'): y_data = x_data * 2.5 + 0.8 tf.histogram_summary("method_demo"+"/y_data",y_data) #可视化观看变量y_data # 指定W和b变量的取值范围,随机在[-200,200] with tf.name_scope('W'): W = tf.Variable(tf.random_uniform([1], -200.0, 200.0)) tf.histogram_summary("method_demo"+"/W",W) # 指定偏移值b,同时shape等于1 with tf.name_scope('b'): b = tf.Variable(tf.zeros([1])) tf.histogram_summary("method_demo"+"/b",b) #可视化观看变量 with tf.name_scope('y'): y = W * x_data + b #sigmoid神经元 tf.histogram_summary("method_demo"+"/y",y) #可视化观看变量 # 最小化均方 with tf.name_scope('loss'): loss = tf.reduce_mean(tf.square(y - y_data)) tf.histogram_summary("method_demo"+"/loss",loss) #可视化观看变量 tf.scalar_summary("method_demo"+'loss',loss) #可视化观看常量 # 定义学习率,我们先使用0.7来看看效果 optimizer = tf.train.GradientDescentOptimizer(0.7) with tf.name_scope('train'): train = optimizer.minimize(loss) # 初始化TensorFlow参数 init = tf.initialize_all_variables() # 运行数据流图 sess = tf.Session() #合并到Summary中 merged = tf.merge_all_summaries() #选定可视化存储目录 writer = tf.train.SummaryWriter('demo_result',sess.graph) #这里的memo_sesult下面很重要 sess.run(init) # 开始计算 for step in xrange(500): sess.run(train) if step % 5 == 0: print(step, "W:",sess.run(W),"b:", sess.run(b)) result = sess.run(merged) writer.add_summary(result,step)
当在终端中输入:tensorboard --logdir=demo-result 时,可能会出现以下三种错误:tensorflow:IOError [Errno 2] No such file or directory: '/home/zgm/anaconda2/envs/tensorflow/lib/python2.7/site-packages/tensorflow/tensorboard/TAG' on path /home/zgm/anaconda2/envs/tensorflow/lib/python2.7/site-packages/tensorflow/tensorboard/TAG解决方法:下载tensorflow的github的源代码,将
4000
tensorflow-0.8.0.data-->purelib-->tensorflow-->tensorboard目录下的TAG文件拷贝到python下面的tensorboard目录下即可
tensorflow:IOError [Errno 2] No such file or directory: '/home/zgm/anaconda2/envs/tensorflow/lib/python2.7/site-packages/tensorflow/tensorboard/lib/css/global.css' on path /home/zgm/anaconda2/envs/tensorflow/lib/python2.7/site-packages/tensorflow/tensorboard/lib/css/global.css解决方法:下载tensorflow的github的源代码,将tensorflow-0.8.0.data-->purelib-->tensorflow-->tensorboard-->lib目录下的CSS文件夹拷贝到python下面的tensorboard目录下的lib中即可
ERROR:tensorflow:Tried to connect to port 6006, but address is in use.
Tried to connect to port 6006, but address is in use.
进程号为9445占用了端口6006,kill掉就好.
下面打开链接就可以看到可视化情况.
至此结束!
相关文章推荐
- 学习TensorFlow,TensorBoard可视化网络结构和参数
- 技术文章 | 用TensorFlow和TensorBoard从零开始构建ConvNet(CNN)
- tensorflow手册tensorboard可视化
- [TensorFlow]理解Tensorboard Graph
- [TensorFlow]入门学习笔记(6)-Tensorboard简易教程和模型保存
- tensorflow6---tensorboard可视化简易教程(一看就懂!)
- tensorFlow之tensorboard可视化中遇到的问题
- 第一次启动tensorboard---'tensorboard' 不是内部或外部命令,也不是可运行的程序
- Tensorflow 04_: tensorboard或者tensorflow的一个"异常"
- Tensorflow(5) 使用 tensorboard 出现locale.Error: unsupported locale setting错误
- Tensorflow - Tutorial (6) : TensorBoard 可视化工具
- Win10下tensorflow的安装及tensorboard的使用
- Tensorflow学习之TensorBoard
- tensorflow日常小记---tensorboard曲线显示不全
- TensorFlow入门教程:21:tensorboard在线性回归训练中的过程追踪
- TensorFlow框架(2)之TensorBoard详解
- 超简单tensorflow入门优化程序&&tensorboard可视化
- Tensorflow--Tensorboard 可视化好帮手
- 利用tensorflow1.2.1中的tensorboard显示数据流程图
- tensorflow: 畅玩tensorboard图表(SCALARS)