Is your feature request related to a problem? Please describe.
When using the model for inference, predict_dataset_bands must be defined at init time.
Describe the solution you'd like
I would like to be able to provide this at predict time, so that each call of predict can specify its own predict_dataset_bands.
The implementation should be relatively straight forward - whatever we currently do in the init can also be done at predict, with an extra argument accepted by the LightningInferenceModel at that time
Is your feature request related to a problem? Please describe. When using the model for inference,
predict_dataset_bands
must be defined at init time.Describe the solution you'd like I would like to be able to provide this at
predict
time, so that each call ofpredict
can specify its ownpredict_dataset_bands
.