Requested GPU instance but memory is already allocated


I just requested a new 1 GPU instance with 32GB ram. However, it seems like the GPU is already taken and not available. Before executing anything else, I ran nvidia-smi and I could see that 12GB have already been allocated. See screenshot:


Running top indicates that there’s no python process running, i.e. I am not sure what is going on here. I stopped the instance and requested a new one, but the same pattern remains.

Could you please advise on how I can make use of the GPU?