上下文分配的法则：生成式搜索中的因果测量与闭环编排

摘要

隨著檢索增強生成（RAG）轉向多樣化的組合生成，其進展受到兩個關鍵瓶頸的阻礙：證據利用的衡量方式存在缺陷，以及上下文預算分配次優。我們依序解決這兩個問題。為了解決衡量問題，我們揭示了一種普遍的「診斷幻覺」：標準的相關性代理指標在困難負樣本上會災難性地失效。我們以一種高效的因果留一法探針取代它們，該探針能精確地分離生成依賴，並正式校準LLM注意力的結構性稀釋。為了解決分配問題，我們在一個去混雜因子網格中部署此因果探針。我們證明，普遍採用的整體式上下文加寬策略是一種受相關性衰減懲罰的架構陷阱。相反地，將計算資源迭代地分配至多個連續生成步驟，可帶來16.7至20.5個絕對百分點的變革性組合召回增益，並能穩健擴展至32B模型。最後，我們將這些解決方案統整為一個可部署的閉環子模調度器。透過歸因引導的對比解碼器來覆寫LLM的注意力慣性，我們的架構系統性地強制整合新證據。藉由勝過經典的開環基線，我們確立了順序、反饋驅動的編排作為生成式搜尋的決定性範式。我們的程式碼、資料與因果衡量工具可在 https://github.com/PeiYangLiu/ascp 取得。

English

As Retrieval-Augmented Generation (RAG) shifts toward diverse portfolio generation, it is stymied by two critical bottlenecks: flawed measurement of evidence utilization, and suboptimal context budget allocation. We resolve both sequentially. To resolve measurement, we expose a pervasive ``diagnostic illusion'': standard relevance proxies fail catastrophically on hard negatives. We replace them with an efficient causal leave-one-out probe that accurately isolates generative reliance and formally calibrates the structural dilution of LLM attention. To resolve allocation, we deploy this causal probe in a deconfounded factorial grid. We prove that the prevailing strategy of monolithic context widening is an architectural trap penalized by relevance decay. Instead, allocating compute iteratively across multiple sequential generations drives transformative portfolio recall gains of 16.7--20.5 absolute percentage points, scaling robustly up to 32B models. Finally, we unify these solutions into a deployable closed-loop submodular scheduler. Augmented by an attribution-steered contrastive decoder to override LLM attention inertia, our architecture systematically forces fresh evidence integration. By dominating classical open-loop baselines, we establish sequential, feedback-driven orchestration as the definitive paradigm for generative search. Our code, data, and causal measurement instruments are available at https://github.com/PeiYangLiu/ascp.