Metraipolis › archive › dc_366e90a6a594

Residual Trajectory Distillation for Generative Retrieval — arXiv:2609.39319

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.39319
License (as archived): CC0-1.0
Archived: 2026-10-07T14:43:33.986Z
SHA-256 of the archived text: 49f4f82406022b61a46ffbd2051d3c6769ad6589f7d1f785e33bef0c639fc467
Stable URL of this page: https://metraipolis.com/documento/dc_366e90a6a594

How to cite

Residual Trajectory Distillation for Generative Retrieval — arXiv:2609.39319 — arXiv (en), https://arxiv.org/abs/2609.39319 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_366e90a6a594 (Metraipolis archive dc_366e90a6a594, sha256 49f4f82406022b61a46ffbd2051d3c6769ad6589f7d1f785e33bef0c639fc467)

BibTeX

@misc{dc366e90a6a594,
  title = {Residual Trajectory Distillation for Generative Retrieval — arXiv:2609.39319},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.39319},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_366e90a6a594; sha256 49f4f82406022b61a46ffbd2051d3c6769ad6589f7d1f785e33bef0c639fc467},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.39319. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Generative retrieval has emerged as a general retrieval paradigm, representing items with discrete Semantic IDs (SIDs) and retrieving them through autoregressive identifier generation. When SIDs are constructed with residual quantization (RQ), standard retrieval training supervises only the selected codes and discards the residual trajectories that produce them. The same hard code can nevertheless arise from different preferences over competing codewords, while the residual trajectory also contains information about subsequent quantization decisions. As a result, hard SID supervision collapses distinct quantization behaviors into identical targets and leaves information available during indexing unused in retrieval training. We introduce ResTD, a Residual Trajectory Distillation framework that transfers this discarded indexing information into retrieval training. Treating the frozen RQ indexer as a process teacher, it distills residual-induced codeword preferences into SID-decoding states. This supervision recovers distinctions hidden by hard assignments and allows earlier decoder states to capture information about subsequent quantization decisions before the corresponding SID suffix is generated. In this way, richer information from SID construction is incorporated into retrieval learning while preserving the original retrieval index and inference procedure. Experiments on multilingual e-commerce retrieval show consistent improvements over strong baselines and matched training controls. Controlled comparisons show that residual-derived targets outperform the tested codebook-only soft targets. Representation probes further show that future codebook preferences become more recoverable from earlier decoder states. ResTD can also be readily extended beyond retrieval to generative recommendation. Code is available at: this https URL .

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-10-01 in cs.IR).
Categorie arXiv: Information Retrieval (cs.IR).
Pagina del lavoro: https://arxiv.org/abs/2609.39319 — 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 2026-10-07T14:16:56.038Z dalla pagina https://arxiv.org/catchup/cs.IR/2026-10-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).