关于使用tensorflow_gpu遇到的问题

前言

我使用的是tensorflow_gpu2.6与python3.9，还要下载cuda与cudnn。

numpy版本问题

AttributeError: module 'numpy' has no attribute 'object'. np.object was a deprecated alias for the builtin object. To avoid this error in existing code, use object by itself. Doing this will not modify any behavior and is safe.

解决方法：

下载高一些版本或低一些的版本，tensorfolw2.6适配的时1.20以上

pip install numpy==1.21

protobuf版本问题

TypeError: Descriptors cannot be created directly. If this call came from a _pb2.py file, your generated code is out of date and must be regenerated with protoc >= 3.19.0. If you cannot immediately regenerate your protos, some other possible workarounds are:

1. Downgrade the protobuf package to 3.20.x or lower.
2. Set PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python (but this will use pure-Python parsing and will be much slower).

解决方法：

保持protobuf的版本在3.19到3.20之间

pip install protobuf==3.19

keras版本问题

from tensorflow.compat.v2.experimental import dtensor as dtensor_api ImportError: cannot import name 'dtensor' from 'tensorflow.compat.v2.experimental'

解决方法：

keras的版本要与tensorflow最好保持一致

pip install keras==2.6

持续更新中......