Can’t believe it’s been 15 days already!
Do you have any advice on handling situations where you provide the control notation, so the GPT knows where to go to get the rules, and then the GPT responds as if it followed the rule, but than you ask it why the output deferred from what the rule expected it says “I’m sorry, I should have gotten the instructions, I see them now, and I see now that the answer should have been…”
Working within the limitations of custom GPT - Anyone done testing?
Topics: