we seem to do a lot of extra encapsulation in the autotests.
this seems something that we actually DONT want. encapsulation would require / use a fallback learner
we don't want that in the autotests. we want to "stop" on errors.
now, I said "stop" with quotes as we want to kinda continue.... but ONLY in the sense that the error message is collected.
not fallback models, etc
we seem to do a lot of extra encapsulation in the autotests. this seems something that we actually DONT want. encapsulation would require / use a fallback learner
we don't want that in the autotests. we want to "stop" on errors.
now, I said "stop" with quotes as we want to kinda continue.... but ONLY in the sense that the error message is collected. not fallback models, etc