Issue deploy to cpp #3386
Description
i want to implement paddleclas on c++, but have some problems. The version i am using now is paddlepaddle==2.5.1 (no gpu) and paddleclas last version, ubuntu 20.4. However when i run the notebook in the PaddleClas/deploy/cpp section it is not as i want. I have successfully built cmake and opencv. I am trying to build c++ paddle inference. So i use paddle Inference library build by wget https://paddle-inference-lib.bj.bcebos.com/2.1.1-gpu-cuda10.2-cudnn8.1-mkl-gcc8.2/paddle_inference.tgz,
tar -xvf paddle_inference.tgz. Then i run sh tools/build.sh and it runs successfully. But I can't find the run.sh file as instructed to run the demo. There is no such file on the default PaddleClas either. The output of my model is different from paddleclas when I add some other output layers, so I want to use the modified paddleclas and not the original, hope everyone can help.