top | item 47008792

(no title)

kfajdsl | 16 days ago

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.

discuss

order

energy123|16 days ago

"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"