通过信息瓶颈强制证据聚焦残差，解决检索生成中查询主导问题。
GRIP: Grounded Reasoning via Information-Restricted Premises
- 用随机瓶颈限制检索证据的表达能力，使其只保留查询外的信息。
- 在5个推理基准上显著优于迭代基线，幻觉减少73%。
- 适合需要可靠推理与低幻觉的应用场景，如医疗问答。
检索增强生成（RAG）中高容量编码器可能导致查询主导潜在状态，使检索证据功能失效，我们称此为查询主导。为解决该问题，提出GRIP（基于信息受限前提的接地推理），引入容量不对称：解码器保持对查询的全维访问，而检索证据需经过严重随机瓶颈。这迫使证据通道仅编码查询中未包含的残差信息。在五个推理基准上，GRIP优于强迭代基线，将查询-潜变量互信息降低约30倍（14.8 → 0.47比特），幻觉减少73%。残差对齐分析显示，瓶颈输出与查询的子空间对齐度低于基线表示。
原文摘要 · Abstract (English)
High-capacity encoders in retrieval-augmented generation (RAG) can let the query dominate the latent state, leaving retrieved evidence functionally irrelevant. We call this failure mode query dominance. To address it, we introduce \textbf{GRIP} (Grounded Reasoning via Information-Restricted Premises), which imposes capacity asymmetry: the decoder keeps full-dimensional access to the query, while retrieved evidence passes through a severe stochastic bottleneck. This forces the evidence channel to encode only the residual information unavailable from the query. Across five reasoning benchmarks, GRIP outperforms strong iterative baselines, cuts a query--latent mutual-information diagnostic by roughly 30$\times$ (14.8 $\to$ 0.47 bits), and reduces hallucination by 73\%. Residual-alignment analysis further shows that the bottleneck output occupies subspaces less aligned with the query than baseline representations.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。