[saberience]

Well, you say that, but when "measuring" anything in RL, that measurement itself is not always obvious.

That is, creating the scoring system/judge models etc for RL is not easy at all. You can easily create an RL loop which is getting better and improving its scores, but actually the result is totally garbage, because you're measuring the wrong thing.

[babelfish]

What do you see wrong with the OPs reward design?

[willleeney]

They said that the we are moving towards " `define good for the model` (success criteria/rubrics)" and you've just gone no no, you don't understand you need to measure the right thing. Feels like you've just completely agreed with them?