A bunch of rubrics I made in different format and structure for llm judge and other use cases

the goal of this repo is, as naming implies, just a singular place for me to talk or push my rubrics, instead of using single gist links. I'm doing a lot of different things and perhaps seeing or using rubrics in a very flexible way compared to the literature.

First let's define the official definition, of a rubric. Let's first go on good ol' wikipedia as baseline: https://en.wikipedia.org/wiki/Rubric_(academic) Now I'm not going to bore you with the details and going to assume you already have a rough idea of what a rubric is at this point. Now, what do I do differently? of course I'm still staying pretty close from baseline here, but I introduce the following, as of time of writing:

1. structure, so XML or JSON, etc
2. tied scoring with IDs and reference within the structure, so is linked to C1:, etc
3. strong constraints, both semantical and structural (refer to 2. here)
4. roleplaying baked in, eg: think like, you are, etc
5. support for feedback, so it explain how to improve it
6. support for reasoning (why it got judged this way)
7. mostly phylosophical outlook, but (roleplaying == jailbreak == context following) == rubrics

there is a lot more tidbits but this will be for the perhaps, soon to be blog I will make on this...