tensorflow学习基础

mnist手写数字识别库 训练例程

import os

os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'

import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import layers, optimizers, datasets

# 训练集 和 测试集 x是图片数字 y是 数字大小
(x, y), (x_val, y_val) = datasets.mnist.load_data()
x = 2 * tf.convert_to_tensor(x, dtype=tf.float32) / 255. - 1
y = tf.convert_to_tensor(y, dtype=tf.int32)
y = tf.one_hot(y, depth=10)  # [0,0,0,0,0,0,0,0,0,1]--->9
print(x.shape, y.shape)
train_dataset = tf.data.Dataset.from_tensor_slices((x, y))
print(train_dataset)
#分成200份
train_dataset = train_dataset.batch(200)

# 创建三层 784->512->256->10


model = keras.Sequential([
    layers.Dense(512, activation='relu'),
    layers.Dense(256, activation='relu'),
    layers.Dense(10)])

'''
优化器 optimizers 
SGD 随机梯度下降法，支持动量参数，支持学习衰减率，支持Nesterov动量
learning_rate：大或等于0的浮点数，学习率
momentum：大或等于0的浮点数，动量参数
decay：大或等于0的浮点数，每次更新后的学习率衰减值
nesterov：布尔值，确定是否使用Nesterov动量
'''
# 随机梯度下降法，支持动量参数，支持学习衰减率，支持Nesterov动量
optimizer = optimizers.SGD(learning_rate=0.001)


def train_epoch(epoch):
    # Step4.loop
    for step, (x, y) in enumerate(train_dataset):
        with tf.GradientTape() as tape:
            # [b, 28, 28] => [b, 784]
            x = tf.reshape(x, (-1, 28 * 28))
            # Step1. compute output
            # [b, 784] => [b, 10]
            out = model(x)
            # Step2. compute loss
            # reduce_sum按照求和的方式对矩阵降维
            loss = tf.reduce_sum(tf.square(out - y)) / x.shape[0]

        # Step3. optimize and update w1, w2, w3, b1, b2, b3
        grads = tape.gradient(loss, model.trainable_variables)
        # w' = w - lr * grad
        optimizer.apply_gradients(zip(grads, model.trainable_variables))

        if step % 299 == 0:
            print(epoch, step, 'loss:', loss.numpy())


def train():
    for epoch in range(50):
        train_epoch(epoch)


if __name__ == '__main__':
    train()

创建张量常用

tf.constant()
tf.convert_to_tensor()
tf.ones()
tf.zeros()
tf.fill()
tf.ones_like()
tf.random.normal()
tf.random.uniform()
tf.range()
tf.cast()
tf.Variable()

x=tf.random.normal([4,784])
net=layers.Dense(10)
net(x).shape
net.kernel.shape
net.bias.shape