Comment by Joel_Mckay

We are at a logical impasse... i.e. failure to understand the noted ROC curve is often a metric that matters in ML development, and LLMs are trivially broken at the tokenization layer:

https://en.wikipedia.org/wiki/Receiver_operating_characteris...

Note, introducing a straw-man argument and or bot slop in an unrelated topic is silly. My anecdotal opinion does not really matter on the subject of algorithmic performance standards. yawn... super boring like ML... lol

https://en.wikipedia.org/wiki/File:Yawning_koala_bear_(35893...

Best of luck, =3