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.
z1 = X @ self.W1 + self.b1; a1 = self.act(z1)
，推荐阅读zoom获取更多信息
尼基塔·阿布拉莫夫（俄罗斯版块编辑）
3 апреля 2026, 08:16 Международные события