调整权值w
1 import tensorflow as tf
2 import numpy as np
3 w = tf.Variable(5, dtype=tf.float32)#定义权值w
4 lr = 0.2 #学习率
5 epoch = 40
6
7 for epoch in range(epoch): # for epoch 定义顶层循环，表示对数据集循环epoch次，此例数据集数据仅有1个w,初始化时候constant赋值为5，循环40次迭代。
8 with tf.GradientTape() as tape: # with结构到grads框起了梯度的计算过程。
9
10 loss = tf.square(w + 1) #计算损失率
11 grads = tape.gradient(loss, w) # .gradient函数告知谁对谁求导
12 w.assign_sub(lr * grads) # .assign_sub 对变量做自减 即：w -= lr*grads 即 w = w - lr*grads
13 print("After %s epoch,w is %f,loss is %f" % (epoch, w.numpy(), loss))
