tensorflow可视化——相关函数用法实例

import tensorflow as tf
import numpy as np
import random


def generate_x_y(batchsize):
    x__data = np.random.uniform(-20, 20, batchsize).astype(np.float32)
    x__data = x__data.reshape(batchsize, 1)
    y__data = []
    for i in range(batchsize):
        y__data.append(x__data[i][0] * 2.0 + 0.8 + random.uniform(-5, 5))
    y__data = np.array(y__data).reshape(batchsize, 1)
    return x__data, y__data


X_ = tf.placeholder(tf.float32, shape=[None, 1])
Y_ = tf.placeholder(tf.float32, shape=[None, 1])

W = tf.Variable(tf.random_uniform([1], -1.0, 1.0))
B = tf.Variable(tf.zeros([1]), dtype=tf.float32)


def Build(X_Input):
    Y_Output = X_Input * W + B
    return Y_Output


Y = Build(X_)
Loss = tf.reduce_mean(tf.square((Y - Y_)))

tf.summary.scalar("Loss", Loss)
tf.summary.histogram("W", W)
tf.summary.histogram("B", B)


merge_summary = tf.summary.merge_all()
# merge_summary = tf.summary.merge([tf.get_collection(tf.GraphKeys.SUMMARIES, ‘Loss‘)])
global_step = tf.Variable(0, trainable=False)
learning_rate = tf.train.exponential_decay(0.99, global_step, 100, 0.9, staircase=True)

train_step = tf.train.AdamOptimizer(learning_rate, beta1=0.5).minimize(Loss, global_step=global_step)


with tf.Session() as sess:
    tf.global_variables_initializer().run()
    writer = tf.summary.FileWriter("view", sess.graph)
    for i in range(10000):
        x_data, y_data = generate_x_y(120)
        sess.run(train_step, feed_dict={X_: x_data, Y_: y_data})
        if i % 100 == 0:
            print("W:", sess.run(W))
            print("B:", sess.run(B))
            print("第%i轮!!!!!!!!!" % i)
            summary = sess.run(merge_summary, feed_dict={X_: x_data, Y_: y_data})
            writer.add_summary(summary=summary, global_step=i)