Metraipolis › archive › dc_f0b6bcf67ea4

OneTrans-V2: Unifying Retrieval, Pre-rank, and Fine-rank with One Transformer in Industrial Recommender — arXiv:2609.28589

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.28589
License (as archived)
CC0-1.0
Archived
2026-10-08T06:43:54.470Z
SHA-256 of the archived text
93b9212fcfe2a4216f759fa6682c0c354a7d8ec18c8cfc17d70cd6d03d76eb4f
Stable URL of this page
https://metraipolis.com/documento/dc_f0b6bcf67ea4

How to cite

OneTrans-V2: Unifying Retrieval, Pre-rank, and Fine-rank with One Transformer in Industrial Recommender — arXiv:2609.28589 — arXiv (en), https://arxiv.org/abs/2609.28589 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f0b6bcf67ea4 (Metraipolis archive dc_f0b6bcf67ea4, sha256 93b9212fcfe2a4216f759fa6682c0c354a7d8ec18c8cfc17d70cd6d03d76eb4f)
BibTeX
@misc{dcf0b6bcf67ea4,
  title = {OneTrans-V2: Unifying Retrieval, Pre-rank, and Fine-rank with One Transformer in Industrial Recommender — arXiv:2609.28589},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.28589},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f0b6bcf67ea4; sha256 93b9212fcfe2a4216f759fa6682c0c354a7d8ec18c8cfc17d70cd6d03d76eb4f},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.28589. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Industrial recommendation systems typically operate as a \emph{cascade} of retrieval, pre-rank, and fine-rank, but these stages are usually trained and served as separate models, causing repeated user-sequence encoding, isolated optimization, and duplicated engineering effort. Building on OneTrans' model-level unification, we present OneTrans-V2, one Transformer that unifies the entire cascade. It encodes the user behavior sequence once as a shared context while preserving stage-specific candidate features and computation. Joint training lets the three stages reinforce one another and enables in-model knowledge distillation from fine-rank to pre-rank. We scale the shared backbone with sparse mixture-of-experts (MoE), which increases capacity with bounded activated computation, and stabilize scaling with $\mu$P-style parameterization. To consolidate objective-specific retrieval channels, we introduce Decision-Conditioned Generative Retrieval (DCGR). DCGR predicts a decision prefix describing the upcoming interaction and generates items conditioned on it, allowing business objectives to steer a single generative process. Finally, Sequence-Native Training (SNT) organizes training around each user's lifelong behavior sequence and amortizes its encoding across exposures. Deployed across all three stages of a large-scale industrial recommendation system, OneTrans-V2 improves gross merchandise value (GMV) by 9.74\% and, with a co-designed serving stack, delivers $3.2\times$ the throughput of the cascade it replaces under the same hardware budget.

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-25 in cs.IR).
Categorie arXiv: Information Retrieval (cs.IR).
Pagina del lavoro: https://arxiv.org/abs/2609.28589 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-08T06:09:07.170Z dalla pagina https://arxiv.org/catchup/cs.IR/2026-09-25?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).