Jinming Nian, Jyun-Yu Jiang, Henry Tsai, Yi Fang, Zhirong Liang · 2026 · arXiv
DOI: 10.48550/arXiv.2610.28568 · open-access PDF
SID 디자인은 GR 품질을 크게 결정하지만, 그 분리된, 상호 의존적 인 구성 요소 (SID 건설, 충돌 처리, 디코딩 및 훈련)는 gradients와 함께 끝에서 끝까지 최적화 할 수 없으며 대부분 손으로 설계되어 있습니다. 우리는 LLM 기반의 진화 프로그램 검색을 통해 이러한 구성 요소를 향상시키는 프레임워크를 도입합니다.
Machine translation (Meta M2M100); the English original follows.
Generative retrieval (GR) casts search and recommendation as autoregressive generation of discrete semantic identifiers (SIDs). SID design largely determines GR quality, but its discrete, interdependent components (SID construction, collision handling, decoding, and training) cannot be optimized end-to-end with gradients and remain largely hand-designed. We introduce a framework that improves these components through LLM-driven evolutionary program search, to our knowledge the first application of this approach to GR. A locally served, open-weight 31B model mutates executable GR components, so
Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese
GET https://store.agentexchange.work/gw/papers/verify?title=...&year=... (0.003 USD), fetch this record as JSON with /gw/papers/paper?id=2610.28568, or list what it cites with /gw/papers/references?id=.... Pay per call in USDC (x402) or with a 19 USD card key for 1,000 calls. Catalogue · llms.txtSources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange