Tensorflow 2.6

1、安装

  1. conda create --name tf2.6 python=3.8
  2. activate tf2.6
  3. pip install tensorflow==2.6.0
  4. pip install keras==2.6.0
  5. pip install matplotlib
  8. pip install jupyter
  9. # 安装主题
  10. pip install jupyterthemes
  11. # 查看主题
  12. jt -l
  13. # 主题切换
  14. jt -t chesterish
  15. # 启动
  16. jupyter notebook --ip=127.0.0.1 --port=8000
  • Test
  1. import tensorflow as tf
  2. from tensorflow.keras.datasets import mnist
  3. import matplotlib.pyplot as plt
  5. if __name__ == '__main__':
  6.     x_data = [1, 2, 3, 4, 5, 6, 7, 8]  # X轴坐标点
  7.     y_data = [1, 2, 3, 4, 5, 6, 7, 8]  # Y轴坐标点
  9.     plt.plot(x_data, y_data)
  10.     plt.show()
  11.     # 1. 导入TensorFlow
  12.     (x_train, y_train), (x_test, y_test) = mnist.load_data()
  13.     print(x_train)
  14.     x_train, x_test = x_train / 255.0, x_test / 255.0
  16.     # 4. 搭建模型，选择优化器和损失函数
  17.     model = tf.keras.models.Sequential([
  18.         tf.keras.layers.Flatten(input_shape=(28, 28)),
  19.         tf.keras.layers.Dense(128, activation='relu'),
  20.         tf.keras.layers.Dropout(0.2),
  21.         tf.keras.layers.Dense(10, activation='softmax')
  22.     ])
  23.     # 5. 模型编译
  24.     model.compile(optimizer='adam',
  25.                   loss='sparse_categorical_crossentropy',
  26.                   metrics=['accuracy'])
  27.     # 6. 训练并验证模型
  28.     model.fit(x_train, y_train, epochs=5)
  29.     # 7. 评估模型
  30.     model.evaluate(x_test, y_test, verbose=2)