Question in nfnets about tensorflow and protobuf.

Hello, friends,
I am new to tensorflow and when I run the code in the nfnets, it will appear the following error:
2022-03-06 10:25:48.256079: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library libcudart.so.11.0
[libprotobuf ERROR google/protobuf/descriptor_database.cc:641] File already exists in database: tensorflow/compiler/xla/xla_data.proto
[libprotobuf FATAL google/protobuf/descriptor.cc:2021] CHECK failed: GeneratedDatabase()->Add(encoded_file_descriptor, size):
terminate called after throwing an instance of 'google::protobuf::FatalException'
what(): CHECK failed: GeneratedDatabase()->Add(encoded_file_descriptor, size):

I guess that the error is about tensorflow and protobuf, but I can't solve it. And I'm using tensorflow 2.4.0, cuda 11.0, cudnn 8.0.5 and protobuf 3.19.4.

Look forward to your reply!
Thanks so much for your help!