arXiv cs.CL· Junyi Zou, Avrova Donz·· 3 天前AI 评分26
MMLA-RSI：通过 RTT 与 RTMA 实现递归自我改进
MMLA-RSI: Recursive Self-Improvement through RTT and RTMA
AI 导读
MMLA-RSI 提出 Reasoning-Time Training（RTT）与 Reasoning-Time Memory Adaptation（RTMA）两条递归自我改进路径，分别改变策略状态 Φ 与权威记忆 M，并让各自的适应方法本身成为改进对象。
正文 · 原文
Abstract:An agent can improve its behavior within an unresolved problem by changing how it acts and which experience it retains. We propose MMLA-RSI, a recursive self-improvement scheme with two implementations in Memory-Mediated Learning Architecture: Reasoning-Time Training (RTT) changes the policy state $\Phi$, while Reasoning-Time Memory Adaptation (RTMA) changes authoritative memory $M$. Each route also makes its adaptation method an object of improvement. A current numerical updater trains the policy proposing its successor; a current memory manager chooses the records available to the process proposing its successor. An independently admitted successor takes over the same role in the next generation. We give bounded executable rule languages, shared task/proposal interfaces, and one algorithm for either route or their composition. The RTT instance changes scalar schedules in a clipped-momentum update. The RTMA instance chooses NULL or a complete-row write with possible eviction. A fixed interpreter, trusted committer, and independent evaluator bound what can change. State versions, read closures, and conditional properties describe causal execution, accepted transitions, and finite work. Existing numerical-component studies motivate the construction. In four-candidate program selection, retaining LARC feedback updates reduces expected error by 24.65 and 36.65 percentage points for two trained initializations. In chronological continuous-integration replay, feedback binding is positive while retention increases loss. Completed SQL studies improve substantially over an unadapted substrate but do not establish the required additional online benefit. These results characterize their declared components. The two recursive routes and their joint realization are specified here; their beneficial recursive chains and discovery efficiency remain to be evaluated.
| Comments: | Technical report. 24 pages, 5 figures, 12 tables |
| Subjects: | Computation and Language (cs.CL); Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.20294 [cs.CL] |
| (or arXiv:2610.20294v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.20294 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Junyi Zou [view email]
[v1]
Wed, 14 Oct 2026 14:47:50 UTC (356 KB)
来源：arXiv cs.CL · arxiv.org