类型化决策编码器中的标签措辞增强：在不熟悉的词上有增益，在缩写或代码上没有
Label-Wording Augmentation in a Typed-Decision Encoder: Gains on Unfamiliar Words, Not on Abbreviations or Codes
Vladimir Beskorovainyi
中文摘要
类型化决策模型对调用者在请求时提供的选项打分，因此其准确率取决于模型开发者无法控制的标签措辞。比较一个307M参数多语言交叉编码器 Ru-Jev 的两个版本，均在十三个俄语任务上训练；后一版本加入了以多种表面形式呈现标签的增强。固定问题不变，仅改变十八个路由标签的措辞，评估十个标签词表，这些词表由三个独立过程生成并检查与训练数据的重叠，每题五个候选，其中八个还使用十八个候选。在由普通词或借词（源自英语的术语、办公室俚语、部门名称）构成的留出词表上，增强版本在所有设置下都更准确，最高提升0.24。这十四个增益中有十一个通过 Holm 校正（十八个候选的俚语和五个候选的描述性句子未通过），对于所有标签上的术语，标签级区间在两种设置下均排除零。在缩写和不透明代码上没有可用增益：在缩写上并不更好，在干净切片上更差；在不透明代码上，一旦移除与训练选项相关的标签，两个版本都处于或低于随机水平。在准确率上升的大多数单元格中，风险-覆盖率面积下降，但在十八个候选时，术语和俚语上的自信错误增加。在两个两个版本都未见过的数据集上，准确率变化为 -8.4 和 +5.6 个百分点。还报告了比较两个版本时的四个评估隐患。
关键要点
- 01问题：类型化决策模型的准确率受调用者提供的标签措辞影响，而开发者无法控制这些措辞。
- 02方法：比较307M参数多语言交叉编码器 Ru-Jev 的两个版本，后一版本加入多表面形式标签增强，在十个标签词表上评估。
- 03结果：在普通词或借词词表上增强版本更准确，最高提升0.24，十四个增益中十一个通过 Holm 校正。
- 04局限：在缩写和不透明代码上没有可用增益，不透明代码上两个版本均处于或低于随机水平。
- 05评估隐患：在两个未见数据集上准确率变化为 -8.4 和 +5.6 个百分点，并报告四个比较两个版本时的评估隐患。
解读
尚无解读。
原始英文摘要
arXiv:2611.08439v1 Announce Type: new Abstract: Typed-decision models score options that the caller supplies at request time, so their accuracy depends on label wording that the model's developer does not control. We compare two releases of a 307M-parameter multilingual cross-encoder, Ru-Jev, trained on thirteen Russian-language tasks; the later release adds an augmentation that presents labels in several surface forms. Holding questions fixed and varying only how eighteen routing labels are worded, we evaluate ten label vocabularies, produced by three separate processes and checked for overlap with the training data, with five candidates per question and, for eight of them, also with eighteen. On held-out vocabularies made of ordinary or borrowed words (English-derived jargon, office slang, department names), the augmented release is more accurate in every setting, by up to 0.24. Eleven of these fourteen gains survive a Holm correction (slang with eighteen candidates and descriptive sentences with five do not), and for jargon on all labels the label-level interval excludes zero in both settings. On abbreviations and opaque codes it gives no usable gain: on abbreviations it is not better, and worse on the clean slice, and on opaque codes both releases stay at or below chance once labels related to training options are removed. In most cells where accuracy rises, the risk-coverage area falls, but with eighteen candidates confident errors rise on jargon and slang. On two datasets neither release saw, accuracy moves by -8.4 and +5.6 points. We also report four evaluation hazards from comparing the two releases.