tensorflow2.0+TF-lite 各种报错

generic_type: type "InterpreterWrapper" is already registered!

原因：tensorflow2.5.0rc0版本太高，
降低版本:

pip install tensorflow==2.3 

ValueErron: "batch_size" needs to be set for models with batch normalization layers.

模型搭建直接引用keras，前面去掉tensorflow

ImportError: cannot import name 'ParamSpec' from 'typing_extensions'

pip install typing-extensions==4.3.0

cannot import name 'get_config' from 'tensorflow.python.eager.context'

keras版本太高

pip install keras==2.2.5

或者将keras都改成tensorflow.keras