Colab ImportError：无法导入名称“squeeze_or_expand_dimensions”

Question

按照Transfer Learning with TensorFlow Hub for TFLite教程，第一步是

!pip install -q -U "tensorflow-gpu=2.0.0b1"

设置：

import os

import matplotlib.pylab as plt
import numpy as np

import tensorflow as tf
import tensorflow_hub as hub

print("Version: ", tf.__version__)
print("Eager mode: ", tf.executing_eagerly())
print("Hub version: ", hub.__version__)
print("GPU is", "available" if tf.test.is_gpu_available() else "NOT AVAILABLE")

出现错误：ImportError: cannot import name 'export_saved_model' 我试过之后

%tensorflow_version 2.x
!pip uninstall -y tensorflow
!pip install tensorflow-gpu==1.14.0

根据this answer，引发了另一个错误： ImportError：无法导入名称“squeeze_or_expand_dimensions”

另外，我尝试了!pip install --upgrade tensorflow，并引发了另一个错误...ImportError: cannot import name 'collections_abc' 我被困在这里了……

Answer 1

不要运行!pip install -q -U "tensorflow-gpu==2.0.0b1"。

在“运行时”菜单下“更改运行时类型”为“GPU”和“恢复出厂设置运行时”。

import os

import matplotlib.pylab as plt
import numpy as np

import tensorflow as tf
import tensorflow_hub as hub

print("Version: ", tf.__version__)
print("Eager mode: ", tf.executing_eagerly())
print("Hub version: ", hub.__version__)
print("GPU is", "available" if tf.test.is_gpu_available() else "NOT AVAILABLE")

输出：

Version:  2.2.0
Eager mode:  True
Hub version:  0.8.0
GPU is available