runs grid-trainer (and creates a new folder with experiments)
runs grid-tester on the resulting folder with experiments
runs grid-analyzer on the final folder with experiments
This will require to:
introduce "use_gpu" command line argument to grid-trainers and testers, similarly to trainers/testers
(that feature will "standardize" all workers when it comes to "use_gpu" argument)
This is a worker which:
This will require to: