search

Jacob-yen / LEMON

This is the implement repository of our upcoming ESEC/FSE 2020 paper: Deep Learning Library Testing via Effective Model Generation.
Apache License 2.0
54 stars 14 forks source link

if is feasible to run your code without nvidia- docker2 #8

Open 77hub opened 2 years ago

77hub commented 2 years ago

Dear author, Hi, I was amazed when I saw this article of yours. It's the first time I've seen someone come up with such an excellent algorithm for Deep Learning Library Testing, but unfortunately as a beginner I don't really know if is feasible to run your code without nvidia- docker2, so I want to ask you some question. I would be very grateful if you could share the code of your currently published paper "Deep Learning Library Testing via Effective Model Generation. I wish you a successful career, a happy family and a happy life.

Jacob-yen commented 2 years ago

Thanks for your interest in our paper.

I would be very grateful if you could share the code of your currently published paper "Deep Learning Library Testing via Effective Model Generation.

This repository presents the entire implementation of our paper.

I don't really know if is feasible to run your code without nvidia- docker2

We have released the docker image(run with GPU and CUDA) used in our experiments on dockerhub for better reproducibility. If you want to run LEMON on machines without GPU and CUDA, you can create the virtual environments(described in Step3) with the python package in the CPU version. For example, change pip install tensorflow-gpu==1.14.0 to pip install tensorflow==1.14.0

Please let me know if you have any further questions. : )

77hub commented 2 years ago

Hello, thank you very much for answering my question. I still have some questions. I don't know why this happens. Specifically, as ordinary users of our laboratory server, we can't install NVIDIA docker2 environment, so I want to install tensorflow theano cntk mxnet lemon and other deep learning libraries directly on anaconda3, but when I run in lemon environment according to the code you give, there are large and small errors, such as

  1. ImportError: libmpi_ cxx. so. 1: cannot open shared object file: No such file or directory

  2. GpuArrayException: b'Could not load "libnvrtc.so": libnvrtc. so: cannot open shared object file: No such file or directory'.

3.OSError: libcudart. so. 10.1: cannot open shared object file: No such file or directory。

4.ModuleNotFoundError: No module named 'cntk. cntk py'

  1. I installed the virtual environment directly without installing NVIDIA docker2, but the server itself has GPU graphics card, CUDA and other environments. I want to know whether your program can still run, and so on.

I don't know if it's the problem of version update. I can't solve this problem. I would appreciate it if you could help me.