LLMs hallucinate because probs -> tokens erase confidence values and it's difficult to assign confidences to strings of tokens, especially if you don't know where to start and stop counting (one word? one sentence?)
Is there a reason to believe this is not solvable as literally an API change? The necessary data are all there.
Is there a reason to believe this is not solvable as literally an API change? The necessary data are all there.