Python3.8环境导入TensorFlow报错numpy.ndarray size changed问题求助

1个月前TensorFlow可以正常运行，现在无法导入。已经完全卸载Python、Anaconda以及对应基础环境文件，重新安装Anaconda后通过Anaconda Navigator安装了TensorFlow 2.3.0。在Anaconda Prompt中导入TensorFlow时报错如下：

(base) C:\Users\Prakhar Sharma>python
Python 3.8.10 (default, May 19 2021, 13:12:57) [MSC v.1916 64 bit (AMD64)] :: Anaconda, Inc. on win32
Type "help", "copyright", "credits" or "license" for more information.

>>> import tensorflow
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "D:\Anaconda\lib\site-packages\tensorflow\__init__.py", line 41, in <module>
    from tensorflow.python.tools import module_util as _module_util
  File "D:\Anaconda\lib\site-packages\tensorflow\python\__init__.py", line 47, in <module>
    from tensorflow.python import keras
  File "D:\Anaconda\lib\site-packages\tensorflow\python\keras\__init__.py", line 27, in <module>
    from tensorflow.python.keras import models
  File "D:\Anaconda\lib\site-packages\tensorflow\python\keras\models.py", line 26, in <module>
    from tensorflow.python.keras.engine import functional
  File "D:\Anaconda\lib\site-packages\tensorflow\python\keras\engine\functional.py", line 38, in <module>
    from tensorflow.python.keras.engine import training as training_lib
  File "D:\Anaconda\lib\site-packages\tensorflow\python\keras\engine\training.py", line 50, in <module>
    from tensorflow.python.keras.engine import data_adapter
  File "D:\Anaconda\lib\site-packages\tensorflow\python\keras\engine\data_adapter.py", line 60, in <module>
    import pandas as pd  # pylint: disable=g-import-not-at-top
  File "D:\Anaconda\lib\site-packages\pandas\__init__.py", line 22, in <module>
    from pandas.compat import (
  File "D:\Anaconda\lib\site-packages\pandas\compat\__init__.py", line 15, in <module>
    from pandas.compat.numpy import (
  File "D:\Anaconda\lib\site-packages\pandas\compat\numpy\__init__.py", line 7, in <module>
    from pandas.util.version import Version
  File "D:\Anaconda\lib\site-packages\pandas\util\__init__.py", line 1, in <module>
    from pandas.util._decorators import (  # noqa
  File "D:\Anaconda\lib\site-packages\pandas\util\_decorators.py", line 14, in <module>
    from pandas._libs.properties import cache_readonly  # noqa
  File "D:\Anaconda\lib\site-packages\pandas\_libs\__init__.py", line 13, in <module>
    from pandas._libs.interval import Interval
  File "pandas\_libs\interval.pyx", line 1, in init pandas._libs.interval
ValueError: numpy.ndarray size changed, may indicate binary incompatibility. Expected 88 from C header, got 80 from PyObject
>>>

另外还有两个相关问题：

• 使用的是Intel i5处理器，不清楚为什么终端提示MSC v.1916 64 bit (AMD64)
• 已经执行过conda update -all把所有包升级到最新版，还是无法解决问题

疑问澄清 #

• MSC v.1916 64 bit (AMD64)属于正常显示：AMD64是64位x86架构的通用命名，Intel的64位处理器也兼容该架构，和实际使用的处理器品牌无关，不需要做任何调整。
• 执行conda update --all反而会加剧冲突：TensorFlow 2.3.0对numpy版本有严格的上限要求，不能使用过高版本的numpy，全局升级所有包会导致numpy、pandas版本和TensorFlow版本不匹配。
• 核心报错的根因：当前环境的pandas版本是基于更高版本的numpy编译的，但环境中实际安装的numpy版本偏低，二者二进制接口不兼容，所以触发报错。

解决步骤 #

方案一：修复现有base环境 #

1. 卸载现有冲突的numpy和pandas，在Anaconda Prompt中执行：

pip uninstall -y numpy pandas

2. 安装适配TensorFlow 2.3.0的numpy版本：

pip install numpy==1.18.5

3. 安装对应适配numpy 1.18.5的pandas版本：

pip install pandas==1.1.5

4. 执行python进入交互环境，输入import tensorflow验证是否正常。

方案二：创建独立conda环境（推荐，避免base环境包冲突） #

1. 创建名为tf23的专属环境，指定Python版本为3.8：

conda create -n tf23 python=3.8 -y

2. 激活环境：

conda activate tf23

3. 安装对应版本的依赖包：

pip install tensorflow==2.3.0 numpy==1.18.5 pandas==1.1.5

4. 验证导入是否正常。

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