search

kengz / SLM-Lab

Modular Deep Reinforcement Learning framework in PyTorch. Companion library of the book "Foundations of Deep Reinforcement Learning".
https://slm-lab.gitbook.io/slm-lab/
MIT License
1.25k stars 266 forks source link

subprocess.CalledProcessError: Command '['git', 'rev-parse', 'HEAD']' returned non-zero exit status 128. #497

Closed luckywlj closed 2 years ago

luckywlj commented 2 years ago

I follow the install instructions to install the miniconda. But when I follow the Quick Start to run test command"python run_lab.py slm_lab/spec/demo.json dqn_cartpole dev", there are some error information as follows: 2021-12-05 21-35-14 的屏幕截图

BTW: yesterday I have asked the problem about the installation. Because I had ever installed anaconda and configure some envs to run other experiments. So I don't run "/bin/setup" instead of run commands as follows: conda env update -f environment.yml source ~/.bashrc

the two commands is from SLM_Lab/bin/setup_ubuntu. I don't want to affect the original anaconda because of running setup.py. But this failed. So I uninstall the anaconda and operate the installation process according to the installation guide.

Could you help me to solve the problem?

kengz commented 2 years ago

it can't find the .git file, so that means the folder wasn't a cloned git repo. Could you git clone using the first step in the guide: https://slm-lab.gitbook.io/slm-lab/setup/installation

luckywlj commented 2 years ago

Yes, I don't use git clone because github is not available on campus network. I use chrome to access github through proxy. So I download the zip package from github and unzip the code repository. So it is basically impossible to download successfully with git clone. What should I do in this situation?

kengz commented 2 years ago

I see. In that case you can comment out this line on your local file: https://github.com/kengz/SLM-Lab/blob/master/slm_lab/spec/spec_util.py#L134 also we've just released an update yesterday with fixes for GPU so you may want to redownload too.

luckywlj commented 2 years ago

Thank you very much! Now it is ok!