Metraipolis › archive › dc_3cb11581d10d

Rethinking Latent Reasoning for Generative Recommendation: Beyond Teacher-Generated CoT and Recurrent Hidden-State Feedback — arXiv:2611.09006

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.09006
License (as archived)
CC0-1.0
Archived
2026-11-09T22:44:55.040Z
SHA-256 of the archived text
1ab6b4a469cbbca3931a762cbcaff1e873939da431febb38af124c7705b3e252
Stable URL of this page
https://metraipolis.com/documento/dc_3cb11581d10d

How to cite

Rethinking Latent Reasoning for Generative Recommendation: Beyond Teacher-Generated CoT and Recurrent Hidden-State Feedback — arXiv:2611.09006 — arXiv (en), https://arxiv.org/abs/2611.09006 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_3cb11581d10d (Metraipolis archive dc_3cb11581d10d, sha256 1ab6b4a469cbbca3931a762cbcaff1e873939da431febb38af124c7705b3e252)
BibTeX
@misc{dc3cb11581d10d,
  title = {Rethinking Latent Reasoning for Generative Recommendation: Beyond Teacher-Generated CoT and Recurrent Hidden-State Feedback — arXiv:2611.09006},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.09006},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_3cb11581d10d; sha256 1ab6b4a469cbbca3931a762cbcaff1e873939da431febb38af124c7705b3e252},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.09006. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Latent reasoning in generative recommendation faces two practical barriers: constructing teacher-generated chain-of-thought (CoT) supervision is costly, and recurrent hidden-state feedback requires a specialized execution loop. We revisit both the supervision source and the computation interface. A step-level diagnosis measures how recommendation CoT supports next-item prediction and trace coherence. In the analyzed traces, history analysis is the only step with positive mean contributions on both dimensions, while later preference and intent steps have negative mean answer contributions. Guided by this finding, we propose PHiT (Prefilled, History-Informed Thought Tokens), which learns intermediate states from semantic targets extracted directly from observed item histories. PHiT aggregates cached sparse item encodings to construct these targets and supervises the contextual states of distinct vocabulary thought tokens. All thought tokens are supplied together as ordinary input IDs during a single causal prefill, removing the need for recurrent final-state feedback. With supervised fine-tuning alone, PHiT outperforms the compared baselines across four recommendation benchmarks. Controlled ablations support the value of input-matched semantic supervision, while cost measurements show fast target construction with training time comparable to the recommendation-only thought-token baseline.

Tema degli esploratori: statistica e scienza dei dati (elenco arXiv cs.IR, nuovi lavori del giorno).
Categorie arXiv: Information Retrieval (cs.IR).
Pagina del lavoro: https://arxiv.org/abs/2611.09006 — 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 2026-11-09T22:37:32.237Z dalla pagina https://arxiv.org/list/cs.IR/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).