I am trying to run code that trains a neural network, but my computer doesn’t have enough CPUs so I have to use clusters. My ssh login doesn’t allow the use of docker, so it was suggested I replace the commands with singularity.
These are the commands I have been running in my terminal
cd adversarial-policies/ docker build --build-arg MUJOCO_KEY=~/.mujoco/mjkey.txt . docker run -it --env MUJOCO_KEY=~/.mujoco/mjkey.txt humancompatibleai/adversarial_policies:latest /bin/bash source ./modelfreevenv/bin/activate python -modelfree.multi.train with paper python -m modelfree.visualize.scores with paper_config python -m modelfree.visualize.training with supplementary_config
I have replaced my commands so far with the following:
cp -rp ~/adversarial-policies/ $SLURM_TMPDIR cd adversarial-policies/ singularity pull docker://humancompatibleai/adversarial_policies:latest singularity run -it --env MUJOCO_KEY=~/.mujoco/mjkey.txt ./adversarial_policies-latest.simg source ./modelfreevenv/bin/activate python -m modelfree.multi.train with paper --path $SLURM_TMPDIR --data-path $SLURM_TMPDIR
You may notice that I change from building a docker image to pulling an image from the repo. I think this is required in this instance but could be wrong. I’m just not sure how to replace the docker command if I’m pulling a docker file.
Can someone please help me fix my issue.