python - 在 Tensorboard 中获取简单的绘图

Question

我正在尝试在张量板上画一个简单的图，就像他们在主页上一样，如下所示:

为了了解这是如何工作的，我编写了以下内容:

    import tensorflow as tf
import numpy as np


x = tf.placeholder('float',name='X')
y=  tf.placeholder('float',name='y')
addition = tf.add(x,y)


with tf.Session() as sess:

    for i in range(100):
        var1=  np.random.rand()
        var2=  np.random.rand()
        print(var1,var2)
        tf.summary.scalar('addition',sess.run(addition, feed_dict={x:var1,y:var2}))               
        writer = tf.summary.FileWriter('Graphs',sess.graph)

虽然我可以看到图表，但看不到任何标量值。任何人都可以向我解释我在这里做错了什么吗？ PS:我已经运行了所有官方示例，它们都可以工作，但我需要理解这个示例才能使用它。 谢谢你的帮助 !

更新

运行@dv3代码后程序崩溃。这是我得到的:

InvalidArgumentError: You must feed a value for placeholder tensor 'input/x-input' with dtype float
     [[Node: input/x-input = Placeholder[dtype=DT_FLOAT, shape=[], _device="/job:localhost/replica:0/task:0/cpu:0"]()]]

During handling of the above exception, another exception occurred:

InvalidArgumentError                      Traceback (most recent call last)
<ipython-input-5-5cbd77e71936> in <module>()
     14         var2=  np.random.rand()
     15         print(var1,var2)
---> 16         add, s_ = sess.run([addition, summary_op], feed_dict={x:var1,y:var2})
     17         writer.add_summary(s_, i)

Answer 1

所以，我想立即建议阅读 this 。它更详细地介绍了 session 是什么。

至于代码及其不产生结果的原因:您没有初始化变量。您可以通过以下方式执行此操作:sess.run(tf.global_variables_initializer())。所以你的代码将是:

import tensorflow as tf
import numpy as np

x = tf.placeholder('float',name='X')
y=  tf.placeholder('float',name='y')
addition = tf.add(x,y)

with tf.Session() as sess:
    sess.run(tf.global_variables_initializer())
    for i in range(100):
        var1=  np.random.rand()
        var2=  np.random.rand()
        print(var1,var2)
        tf.summary.scalar('addition',sess.run(addition, feed_dict={x:var1,y:var2}))               
        writer = tf.summary.FileWriter('Graphs',sess.graph)

我不会将 sess.run 嵌入到summary.scalar 调用中，但对于这个简单的示例，您将得到一些结果。

编辑: 经过测试，这确实有效:

import tensorflow as tf
import numpy as np

x = tf.placeholder('float',name='X')
y=  tf.placeholder('float',name='y')
addition = tf.add(x,y, name='add')
tf.summary.scalar('addition', addition)
summary_op = tf.summary.merge_all()     
with tf.Session() as sess:
    sess.run(tf.global_variables_initializer())
    writer = tf.summary.FileWriter('Graphs',sess.graph)
    for i in range(100):
        var1=  np.random.rand()
        var2=  np.random.rand()
        print(var1,var2)
        add, s_ = sess.run([addition, summary_op], feed_dict={x:var1,y:var2})
        writer.add_summary(s_, i)

输出: