To make this practical, I first define a calibrated rubric over the digits 0-9 (there’s only one token for each digit), where each digit corresponds to a clear qualitative description. At the scoring step, I capture the model’s next-token logits and retain only the logits corresponding to those valid digit tokens. This avoids contamination from unrelated continuations such as explanation text, punctuation, or alternate formatting. After renormalizing over the restricted digit set, I interpret the resulting probabilities as a categorical score distribution.
Начальник ГРУ заявил о жестком вопросе Киеву после покушения на генерала Алексеева14:48
，这一点在新收录的资料中也有详细论述
Israel earlier Saturday said it struck a Tehran airport it said was used to transfer weapons and cash to militant groups.
但这个逻辑在2025年悄悄失效了。
Emitting functions and blocksSince the IRs root construct is a function containing blocks, the bytecode