Record Grouping Controls Evidence Weight in Language Models

2026-09-08

AI Summary

研究通过控制记录分组来调整语言模型中的证据权重，提出了一种预生成表示方法以去除组内重复内容并限制每组贡献，实验验证了不同分组对模型决策的影响。

Abstract

Retrieved records are presentation units; a supplied partition determines which records enter a language model as one evidential contribution. We characterize the invariant group-content state that removes within-group copies while retaining complementary canonical content, show that equal group counts can encode different evidence states, and derive a sharp content-aware partition-error bound. Given a supplied partition, our pre-generation representation deduplicates and aggregates content within groups and bounds each group's contribution. Across 104,402 trials and 6 public checkpoints, a central natural-text intervention finds that content-fixed false splits add 10.27-32.66 percentage points and false merges remove 9.13-31.79 points; a matched six-slot control retains the positive direction in all 16 cells. In a new 48-item controlled campaign panel, changing the supplied partition produces measurable, checkpoint-dependent decision shifts across all four models, and the balanced mirror design exposes substantial order interactions. Together, the theory and experiments establish the supplied partition as a controllable pre-generation representation variable and characterize its checkpoint-dependent behavioral effects.

Problem

• Record Grouping
• Evidence Weight
• Language Models
• Partition
• Content Deduplication

Innovation

• Record Grouping
• Evidence Weight
• Language Models
• Content Deduplication
• Partition-Error Bound