Metraipolis › archive › dc_8fbc4690d53f
DecisionRank: On-Policy Distillation for Decision-Model Rerankers — arXiv:2610.42128
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.42128
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-02T07:23:18.589Z
- SHA-256 of the archived text
7c742363344dc0d2b976bfd13e3f9cfc36d989647e47528fd11112eafe7eae2e- Stable URL of this page
https://metraipolis.com/documento/dc_8fbc4690d53f
How to cite
DecisionRank: On-Policy Distillation for Decision-Model Rerankers — arXiv:2610.42128 — arXiv (en), https://arxiv.org/abs/2610.42128 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_8fbc4690d53f (Metraipolis archive dc_8fbc4690d53f, sha256 7c742363344dc0d2b976bfd13e3f9cfc36d989647e47528fd11112eafe7eae2e)
BibTeX
@misc{dc8fbc4690d53f,
title = {DecisionRank: On-Policy Distillation for Decision-Model Rerankers — arXiv:2610.42128},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.42128},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_8fbc4690d53f; sha256 7c742363344dc0d2b976bfd13e3f9cfc36d989647e47528fd11112eafe7eae2e},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.42128. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Decision models target structured tasks by directly predicting choices or scores, avoiding autoregressive decoding latency. However, transferring a generative LLM's ranking knowledge to a decision model remains challenging: the teacher expresses preferences through a ranked list of document ID sequences, while the student determines the ranking using scores from the first-token position. To bridge this gap, we introduce DecisionRank, a decision model trained through item-space on-policy distillation (OPD). Item-space OPD constructs the student's rollout by sorting its document scores into a ranked sequence. A generative teacher takes the resulting ranking list as inputs and provides conditional feedback to directly supervise those scores. Combined with offline ranking supervision, DecisionRank achieves state-of-the-art performance on both general and reasoning-intensive retrieval benchmarks such as TREC DL, BEIR, and BRIGHT. Across four benchmarks, item-space OPD consistently outperforms alternative training objectives. Even at 1.7B parameters, DecisionRank surpasses its 27B generative teacher on TREC DL19--23. At 8B, it outperforms its generative listwise counterpart while running $24.5\times$ faster on TREC DL and $16.7\times$ faster on BRIGHT. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno). Categorie arXiv: Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2610.42128 — 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 2026-11-02T07:17:12.135Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).