merry marsh
So I’ve created a diagnostic rubric to go with this app to show its performance, but specifically to show which aspects of ai behavior I have specifically combatted with this gpt.
This can be useful for comparisons sake, or to ensure that it’s functioning optimally. Especially compared to what I call “vanilla” ai outputs.
It’s not necessary for using it. It might just be interesting in seeing how soma is different. Or, if you suspect broken containment (so it’s reverting to vanilla behaviors) you can be certain of that fact (it should revert to contained behavior, but if it doesn’t, it might necessitate starting a new chat to continue your creative work).