I've been asked for things like this in the past, and I'm keen now on adding more metadata that's specific to the simplified trainer in nam.train.core.train().
I'm going to add it via an other_metadata kwarg in the model so that .export() isn't blocked out from other uses.
Included will be:
Settings exposed by the standardized trainers (fit cab, ignore checks)
Data metadata (checks, latency)
Validation ESR (since the simplified trainer has a standardized validation set, I'm okay with this.)
I'm expecting that fit_cab will be very helpful for downstream plugins and such (e.g. ToneHunt) to automatically suss out full rig models and implement functionality accordingly (e.g. a plugin like NeuralAmpModelerPlugin that has an IR loader might automatically disable and bypass it).
I've been asked for things like this in the past, and I'm keen now on adding more metadata that's specific to the simplified trainer in
nam.train.core.train().I'm going to add it via an
other_metadatakwarg in the model so that.export()isn't blocked out from other uses.Included will be:
I'm expecting that
fit_cabwill be very helpful for downstream plugins and such (e.g. ToneHunt) to automatically suss out full rig models and implement functionality accordingly (e.g. a plugin like NeuralAmpModelerPlugin that has an IR loader might automatically disable and bypass it).