and wrt "is the model just hallucinating intent in its response to my question"--it's possible, but openai claims to have trained the model to answer qs about its reasoning, and the summary captured…

Posters: voooooogel. voooooogel

3 likes recorded by the source corpus.