TensorFlow 使用 tf.scalar tf.summary

经过摸索,简单使用summary,尤其注意下面代码加粗变量,加粗的是二进制变量,需要写入文件的;

以及运行的软件要有/tmp/XIAO文件夹的权限

# /usr/bin/python3
import tensorflow as tf

sess = tf.Session()

with tf.name_scope("XiaoGongWei"):
    a = tf.placeholder(dtype=tf.float32)
    b = tf.Variable([1.0],dtype=tf.float32)
    W = tf.Variable([1,2],dtype=tf.float32)
    addAB = W * a + b
    tf.summary.scalar("wight_max",tf.reduce_mean(W))
    tf.summary.scalar("b_value",tf.reduce_mean(b))
    tf.summary.scalar("a_value",tf.reduce_mean(a))
    merged = tf.summary.merge_all()
    train_summary = tf.summary.FileWriter('/tmp/XIAO/',sess.graph)
    sess.run(tf.global_variables_initializer())
    for i in range(100):
        tfmensumm,myadd = sess.run([merged,addAB],feed_dict={a:i})
        print(myadd)
        train_summary.add_summary(tfmensumm,i)
    train_summary.close()

运行代码1次后会在对应/tmp/XIAO出现一个文件

然后终端运行 tensorboard --logdir=/tmp/XIAO

打开浏览器输入:0.0.0.0:6006就会出现scalar保存结果

本文作为初学的测试,谢谢大家评论