During the evaluation process, in the 'docker run xxx' command, will the '--gpus all' parameter be used in your commands? nvidia-container-toolkit seems to require this parameter.

During the evaluation process, in the 'docker run xxx' command, will the '--gpus all' parameter be used in your commands? nvidia-container-toolkit seems to require this parameter.

Hi,

Yes, that's fine if you need to use the --gpus argment.

However, please ensure that you use CUDA_VISIBLE_DEVICE=0

Regards, Reuben