发表机构
Called It Inc.（Called It 公司）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
该研究针对代码助手RAG存在的无法区分新旧代码事实的问题，提出MemStrata确定性取代记忆，在真实GitHub修复数据集上使过时事实错误率降至约0，准确率达0.91，延迟仅约2.1秒。
AI 中文摘要
检索增强生成（RAG）没有时间模型：当编码会话中的事实发生变化时——函数被重命名、端点被移动、依赖项被更新——RAG会以几乎相同的相似度检索到旧值和新值，无法区分哪个是当前值，因此会提供已被取代的值。论文1在合成单值基准测试中表明，确定性的（主体、关系、客体）取代记忆可消除这种故障。在此，我们在真实软件历史上对其进行端到端验证。从707个真实GitHub问题（SWE-bench Lite + Verified）中，我们提取了130个干净的原子状态转换，即从修复前形式到修复后形式更改一个可识别值的修复，且每个转换都是无标记的（过时语句和当前语句仅在值上不同）。在该数据集上，MemStrata的答案准确率达到0.91，而RAG的准确率为0.57-0.59；此外，当被强制回答时，RAG有36-38%的时间会提供已被取代的值（LLM重排序器对此无帮助），而MemStrata将这一比例降至约0，且延迟与RAG检索延迟相近（约2.1秒，而重排序器约为18秒）。我们明确说明适用范围：仅约18%的真实修复是干净的原子转换；论文2针对该类问题分离出记忆机制，其余修复的提取覆盖是正交问题，我们将其留待后续工作解决。研究期间还出现并修复了一个真实产品错误（不区分大小写的值比较/标点符号），且确定性取代在干净代码变异上的准确性（护城河属性）得以保留和验证。
英文摘要
Retrieval-augmented generation (RAG) has no model of time: when a fact changes across a coding session - a function is renamed, an endpoint moves, a dependency is bumped - RAG retrieves both the old and new value with near-identical similarity and cannot tell which is current, so it serves the superseded value. Paper 1 showed, on synthetic single-value benchmarks, that a deterministic (subject, relation, object) supersession memory eliminates this failure. Here we validate it end-to-end on real software history. From 707 real GitHub issues (SWE-bench Lite + Verified) we extract 130 clean atomic state transitions, a fix that changes one identifiable value from a pre-fix to a post-fix form, and render each marker-free (the stale and current statements differ only in the value). On this set, MemStrata reaches 0.91 answer accuracy versus RAG's 0.57-0.59; and, the structural result, when forced to answer RAG serves the superseded value 36-38% of the time (an LLM reranker does not help) while MemStrata drives this to ~0, at RAG retrieval latency (~2.1 s vs ~18 s for the reranker). We are explicit about scope: only ~18% of real fixes are clean atomic transitions; Paper 2 isolates the memory mechanism on that class, and extraction coverage of the remaining fixes is the orthogonal problem we defer to follow-on work. A real product bug surfaced and was fixed during the study (a case/punctuation-insensitive value comparison), with the moat property (deterministic-supersession accuracy on clean code mutations) preserved and verified.
