属性错误：'Sequential'对象没有属性'_get_distribution_strategy'。

Question

我正在通过 LinkedIn 学习有关使用 Keras 构建模型的在线课程。

这是我的代码（声称可以工作）：

import pandas as pd
import keras
from keras.models import Sequential
from keras.layers import *

training_data_df = pd.read_csv("sales_data_training_scaled.csv")

X = training_data_df.drop('total_earnings', axis=1).values
Y = training_data_df[['total_earnings']].values

# Define the model
model = Sequential()
model.add(Dense(50, input_dim=9, activation='relu', name='layer_1'))
model.add(Dense(100, activation='relu', name='layer_2'))
model.add(Dense(50, activation='relu', name='layer_3'))
model.add(Dense(1, activation='linear', name='output_layer'))
model.compile(loss='mean_squared_error', optimizer='adam')


# Create a TensorBoard logger
logger = keras.callbacks.TensorBoard(
    log_dir='logs',
    write_graph=True,
    histogram_freq=5
)


# Train the model
model.fit(
    X,
    Y,
    epochs=50,
    shuffle=True,
    verbose=2,
    callbacks=[logger]
)

# Load the separate test data set
test_data_df = pd.read_csv("sales_data_test_scaled.csv")

X_test = test_data_df.drop('total_earnings', axis=1).values
Y_test = test_data_df[['total_earnings']].values

test_error_rate = model.evaluate(X_test, Y_test, verbose=0)
print("The mean squared error (MSE) for the test data set is: {}".format(test_error_rate))

当执行以下代码时，我得到了以下错误。

Using TensorFlow backend.
2020-01-16 13:58:14.024374: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
2020-01-16 13:58:14.037202: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x7fc47b436390 initialized for platform Host (this does not guarantee that XLA will be used). Devices:
2020-01-16 13:58:14.037211: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Host, Default Version
Traceback (most recent call last):
  File "/Users/himsaragallage/Documents/Building_Deep_Learning_apps/06/model_logging final.py", line 35, in <module>
    callbacks=[logger]
  File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/keras/engine/training.py", line 1239, in fit
    validation_freq=validation_freq)
  File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/keras/engine/training_arrays.py", line 119, in fit_loop
    callbacks.set_model(callback_model)
  File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/keras/callbacks/callbacks.py", line 68, in set_model
    callback.set_model(model)
  File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/keras/callbacks/tensorboard_v2.py", line 116, in set_model
    super(TensorBoard, self).set_model(model)
  File "/Library/Frameworks/Python.framework/Versions/3.7/lib/python3.7/site-packages/tensorflow_core/python/keras/callbacks.py", line 1532, in set_model
    self.log_dir, self.model._get_distribution_strategy())  # pylint: disable=protected-access
AttributeError: 'Sequential' object has no attribute '_get_distribution_strategy'

Process finished with exit code 1

在我尝试调试时

我发现这个错误是因为我正在尝试使用一个tensorboard logger。更准确地说，当我添加callbacks=[logger]时会出现问题。如果没有这行代码，程序就能正常运行而没有任何错误。但是Tensorboard将无法使用。

请建议一种方法，可以消除错误并成功运行上述Python脚本。

- Himsara Gallege

Answer 1

希望您指的是这个LinkedIn Keras课程。

当我使用Tensorflow版本2.1时，我也遇到了相同的错误。然而，在降级Tensorflow版本并对代码进行轻微修改后，我可以调用Tensorboard。

下面展示了可运行的代码：

import pandas as pd
import keras
import tensorflow as tf
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import *

training_data_df = pd.read_csv("sales_data_training_scaled.csv")

X = training_data_df.drop('total_earnings', axis=1).values
Y = training_data_df[['total_earnings']].values

# Define the model
model = Sequential()
model.add(Dense(50, input_dim=9, activation='relu', name='layer_1'))
model.add(Dense(100, activation='relu', name='layer_2'))
model.add(Dense(50, activation='relu', name='layer_3'))
model.add(Dense(1, activation='linear', name='output_layer'))
model.compile(loss='mean_squared_error', optimizer='adam')

# Create a TensorBoard logger
logger = tf.keras.callbacks.TensorBoard(
    log_dir='logs',
    write_graph=True,
    histogram_freq=5
)

# Train the model
model.fit(
    X,
    Y,
    epochs=50,
    shuffle=True,
    verbose=2,
    callbacks=[logger]
)

# Load the separate test data set
test_data_df = pd.read_csv("sales_data_test_scaled.csv")

X_test = test_data_df.drop('total_earnings', axis=1).values
Y_test = test_data_df[['total_earnings']].values

test_error_rate = model.evaluate(X_test, Y_test, verbose=0)
print("The mean squared error (MSE) for the test data set is: {}".format(test_error_rate))

- user11530462

Answer 2

你可能会发现这篇帖子有用。

所以，不要从keras中导入（即）

from keras.models import Sequential

从TensorFlow导入：

from tensorflow.keras.models import Sequential

当然，这同样适用于大多数其他导入。

这只是一个幸运的猜测，因为我无法运行你的代码，但希望它能有所帮助！

- Pörripeikko

Answer 3

我建议不要混用keras和tf.keras。因为它们是不同的项目，keras是原始的多后端项目，而tf.keras是集成到tensorflow中的版本。Keras将停止支持其他后端，只支持tensorflow，因此建议切换到它。请查看https://keras.io/#multi-backend-keras-and-tfkeras。

一种简单的方法是从tensorflow导入keras：

import tensorflow as tf
import tensorflow.keras as keras
#import keras
import keras.backend as K
from keras.models import Model, Sequential, load_model
from keras.layers import Dense, Embedding, Dropout, Input, Concatenate

print("Python: "+str(sys.version))
print("Tensorflow version: "+tf.__version__)
print("Keras version: "+keras.__version__)

Python: 3.6.9 (default, Nov  7 2019, 10:44:02) 
[GCC 8.3.0]
Tensorflow version: 2.1.0
Keras version: 2.2.4-tf

- EduardoFV

Answer 4

看起来你的Python环境正在混合使用keras和tensorflow.keras的导入。尝试像这样使用Sequential模块：

model = tensorflow.keras.Sequential()

- Rocío García Luque

Answer 5

我遇到了同样的错误，问题是由于TensorFlow版本引起的。

!pip install tensorflow==2.3.0

已修正

- Arye P.