Hey, i’ve seen several topics about models performance visualisation (ROC curve, precision vs recall) and i believe many people have their own boilerplate code doing the same graphs. I really like Yellowbrick approach for sklearn models and I thought we could use it for fastai Learner. Here is a notebook with a small wrapper alining fastai learner with sklearn api allowing to reuse yellowbrick visualisations for classification. Potentially, with more generic class, this can be useful for other libraries. I think it would be cool to incorporate native versions of them in the Interpreter but adding all graphs is a non-trivial task, for now the hacky way seems to work. Any comments appreciated!