differentiate between available and used models #18
Description
Use case
Someone wants to unselect a model for a given prediction (without restarting the whole ensemble). This is especially common for our frontend which has toggle buttons.
Current status
To do this, one has to do something like
ensemble = BaseEnsemble(my_config)
full_preds = predict_smiles_list(my_smiles)
all_models = ensemble.models
ensemble.models = ensemble.models[:-1] # throw out last models
partial_preds = predict_smiles_list(my_smiles)
ensemble.models = all_models # restore models
This is cumbersome and error-prone.
How it should be
ensemble = BaseEnsemble(my_config)
full_preds = predict_smiles_list(my_smiles)
partial_preds = predict_smiles_list(my_smiles, use_models=["Model1", "Model2", "Model3")
This is easy to use (given that the user knows their model names, but that can be easily inferred from the ensemble. use_models specifies which models to use (default: all that are available),