some problem about undefined symbol:_ZNK10tensorflow8OpKernel11TraceStringERKNS_15OpKernelContextEb

[jzl20]

when I run the
python learner_1.py --workspace_path [your path]/dr-derks-mutant-battlegrounds/

I meet some problems like this :

tensorflow.python.framework.errors_impl.NotFoundError: /home/public/grpc/python/../grpc_cc.so: undefined symbol: _ZNK10tensorflow8OpKernel11TraceStringERKNS_15OpKernelContextEb

[kimbring2]

@jzl20 Hello, I know that problem. It happens when the GRPC version is not compatible with the Tensorflow version.

I entirely borrow the training code from https://github.com/google-research/seed_rl. You can check the folder from there named https://github.com/google-research/seed_rl/tree/master/grpc.

It is the same folder of my repo(https://github.com/kimbring2/MOBA_RL/tree/main/dr-derks-mutant-battlegrounds/grpc)

In my experience, we should use the Tensorflow 2.4.1 version for that.

Try to change the version and tell the result.

Thank you

[jzl20]

@kimbring2
GREAT !!! It works ! Thank you very much !

If possible , I wish you can modify the 'requirements.txt' https://github.com/kimbring2/MOBA_RL/blob/main/dr-derks-mutant-battlegrounds/requirements.txt . (tensorflow-cpu==2.2.0 . It dose not work )

---

At the same time , I want to share some other Error code to help others tensorflow 2.2.0 meets undefined symbol: _ZNK10tensorflow8OpKernel11TraceStringERKNS_15OpKernelContextEb tensorflow 2.6.0 meets undefined symbol: _ZN4absl14lts_2020_02_2518container_internal18global_next_sampleE And other Error code about the Tensorflow version: ImportError: cannot import name 'dtensor' from 'tensorflow.compat.v2.experimental'

What a lucky day with the help of the author！

[kimbring2]

@jzl20 I am happy your code is now working well. Also, thank you for your advice. I modify the requirements.txt file.

Feel free to ask another question when you are stuck +1:

[jzl20]

hahaha , so friendly you are . ^_^