导入 keras 时出现“无法导入名称 'keras'”错误

Question

import tensorflow as tf
from tensorflow import keras

结果是

ImportError                               Traceback (most recent call last)
<ipython-input-1-75a28e3c6620> in <module>
      1 import tensorflow as tf
----> 2 from tensorflow import keras
      3 
      4 import numpy as np
      5 

ImportError: cannot import name 'keras'

使用 tensorflow 1.2.1 版和 keras 2.3.1 版

Answer 1

您使用的是旧版本的 tensorflow。您可以使用以下代码更新到最新版本

! pip install tensorflow --upgrade

从Tensorflow V2.0 开始，keras 以tf.keras 的形式集成到 tensorflow 中，因此无需单独导入 keras。

要创建顺序模型，您可以参考以下代码

import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import layers

#Define Sequential model with 3 layers
model = keras.Sequential(
    [
        layers.Dense(2, activation="relu", name="layer1"),
        layers.Dense(3, activation="relu", name="layer2"),
        layers.Dense(4, name="layer3"),
    ]
)
# Call model on a test input
x = tf.ones((3, 3))
y = model(x)

print("Number of weights after calling the model:", len(model.weights))  # 6

model.summary()

输出：

Number of weights after calling the model: 6
Model: "sequential_1"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
layer1 (Dense)               (3, 2)                    8         
_________________________________________________________________
layer2 (Dense)               (3, 3)                    9         
_________________________________________________________________
layer3 (Dense)               (3, 4)                    16        
=================================================================
Total params: 33
Trainable params: 33
Non-trainable params: 0
_________________________________________________________________

更多详情请参考Tensorflow Guide。如果您无法完成任务，我很乐意为您提供帮助。