ndand·昨年·議論> Valued = useful + hard to find.I agree, but I think it slightly differently, Valued = useful + hard to replace.How difficult is to be replaced by some other human or a machine.
ndand·昨年·議論I understand it differently,LLMs predict distributions, not specific tokens. Then an algorithm, like beam search, is used to select the tokens.So, the LLM predicts somethings like, 1. ["a", "an", ...] 2. ["astronomer", "cosmologist", ...],where "an astronomer" is selected as the most likely result.