Google Colab导入tensorflow_federated报TypeError解决方案

问题场景 #

在Google Colab平台运行《Federated Learning for Image Classification（图像分类联邦学习）》官方教程示例代码时，执行导入tensorflow_federated的语句触发报错。

触发报错的示例代码如下：

import collections
import numpy as np
import tensorflow as tf
import tensorflow_federated as tff

np.random.seed(0)

tff.federated_computation(lambda: 'Hello, World!')()

报错详情 #

• 报错类型：TypeError
• 报错提示：'type' object is not subscriptable
• 错误定位：报错栈指向tensorflow_federated包内structure.py文件的collections.OrderedDict[str, Any]类型标注行

完整报错信息如下：

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-35-a23308ec3f7c> in <module>()
      3 import numpy as np
      4 import tensorflow as tf
----> 5 import tensorflow_federated as tff
      6 
      7 np.random.seed(0)

6 frames
/usr/local/lib/python3.7/dist-packages/tensorflow_federated/python/common_libs/structure.py in <module>()
    263 
    264 def to_odict(struct: Struct,
--> 265              recursive: bool = False) -> collections.OrderedDict[str, Any]:
    266   """Returns `struct` as an `OrderedDict`, if possible.
    267 

TypeError: 'type' object is not subscriptable

无效尝试方案 #

• 参考公开方案将Python版本升级至3.9，问题未解决。

可行解决步骤 #

在Colab单元格中执行以下代码，指定安装0.20.0版本的tensorflow-federated，升级nest-asyncio并启用异步嵌套适配，执行完成后重启运行时即可正常导入使用TFF：

!pip install --quiet tensorflow-federated==0.20.0
# 不要直接安装最新版tensorflow-federated，会触发版本兼容问题
#!pip install --quiet --upgrade tensorflow-federated
!pip install --quiet --upgrade nest-asyncio

import nest_asyncio
nest_asyncio.apply()

内容的提问来源于stack exchange，提问作者CodingTheVerse