tensorflow-1.9.0 中 protobuf 版本问题

[haiyang-tju]

请问一下，你在编译 tensorflow-1.9.0-cp35-cp35m-linux_aarch64.whl 时的 protobuf 版本是多少？我在安装过程中默认安装的版本是protobuf 3.6.0，但是在运行TF的时候给出以下错误：

[libprotobuf FATAL google/protobuf/stubs/common.cc:61] This program requires version 3.5.0 of the Protocol Buffer runtime library, but the installed version is 2.6.1. Please update your library. If you compiled the program yourself, make sure that your headers are from the same version of Protocol Buffers as your link-time library. (Version verification failed in "bazel-out/arm-opt/genfiles/tensorflow/contrib/tpu/proto/tpu_embedding_config.pb.cc".) terminate called after throwing an instance of 'google::protobuf::FatalException' what(): This program requires version 3.5.0 of the Protocol Buffer runtime library, but the installed version is 2.6.1. Please update your library. If you compiled the program yourself, make sure that your headers are from the same version of Protocol Buffers as your link-time library. (Version verification failed in "bazel-out/arm-opt/genfiles/tensorflow/contrib/tpu/proto/tpu_embedding_config.pb.cc".)

尝试过卸载重装 pip3 uninstall protobuf pip3 install protobuf pip3 install protobuf==2.6.1 pip3 install protobuf==3.5.0.post1

均无效，还是有此错误提示。

[peterlee0127]

HI, I use the virtualenvs with python 3.5. protobuf version is 3.5.2.post1. You can try to use virtualenvs to setup the build environment.

[haiyang-tju]

I am trying to compile Tensorflow with RT myself.

Is there any special configuration for TensorRT? Or just an item in bazel's configure: TensorRT support?[y/N]: y

But I am a little confused about that the Tensorflow with RT used an integrated part or an additional installation of TensorRT?

[peterlee0127]

Hi, TensorRT is already integrated in JetPack. No need to install tensorRT. Just use the default value.

[haiyang-tju]

Thanks for your reply. I will try it.