midi_degradation_toolkit
midi_degradation_toolkit copied to clipboard
Implement --baseline in train and eval scripts
To train a model with default settings, and to eval our (provided) baseline model for each task.
With --baseline and a --task given, no other args should be necessary, including --format, etc.
This is nice, i think a priority for camera ready, but not ACME v1.0 release.
(We could also remove this arg, but I like it)
Change of plan, we should remove this flag in favour of just documenting how to reproduce results.