experiment-impact-tracker
experiment-impact-tracker copied to clipboard
Migrate to pynvml
Look into migrating to pynvml now that it seems possible to get GPU info per process.
This would stop errors like in #21 and #14
Also #36