[matjet]

Look what they need to mimic a fraction of [the power of having the logit probabilities exposed so you can actually see where the model is uncertain]

[kfajdsl]

All the LLM logprob outputs I've seen aren't very well calibrated, at least for transcription tasks - I'm guessing it's similar for OCR type tasks.

[energy123]

"I already decided in my private reasoning trace to resolve this ambiguity by emitting the string '27' instead of '22' right here, thus '27' has 100% probability"