Pith. sign in

REVIEW 4 major objections 8 minor 39 references

Replication and Exploration of Generative Retrieval over Dynamic Corpora

T0 review · 4 major / 8 minor · reviewed 2026-08-16 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The format of a document's identifier decides whether a generative retriever can find documents added after training.

desk verdict Useful and honest replication study: text-based docids clearly generalize better than numeric ones on random dynamic splits, but the i.i.d. benchmark and single-dataset MDGR evaluation mean the sweeping 'dynamic corpora' claims need tempering. read the letter →

arxiv 2504.17519 v1 pith:XYIFM2WE submitted 2025-04-24 cs.IR

Zhen Zhang , Xinyu Ma , Weiwei Sun , Pengjie Ren , Zhumin Chen , Shuaiqiang Wang , Dawei Yin , Maarten de Rijke
show 1 more author
Zhaochun Ren
This is my paper · ORCID
classification cs.IR
keywords generativeretrievaldynamiccorporadociddesigntext-baseddocidsnumeric-basedgeneralizationmulti-docidinformation
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper tries to establish that, in generative retrieval, the way a document's identifier (docid) is written determines whether a trained model can retrieve documents added after training. Re-running eight generative retrieval models on dynamic splits of NQ and MS MARCO, it finds that models using text-based docids—titles, URLs, n-grams, or multi-text identifiers—generalize to newly added documents without retraining, while numeric-docid models perform much worse on new documents and keep favoring the original collection. The paper argues that text docids win because they align with the language model's pretraining distribution, encode documents at a finer granularity, and use a larger effective vocabulary. It then proposes MDGR, a multi-docid numeric scheme that reuses existing numeric docids for new document chunks and combines retrieval speed with text-level generalization. If correct, the result gives search-engine builders a cheap path to updating generative retrieval over evolving collections.

What carries the argument

The load-bearing object is the docid, the sequence of tokens a generative retriever is trained to emit for each document. The paper separates docids into numeric-based types, such as cluster or quantized index codes used by DSI-SE, Ultron-PQ, and NCI, and text-based types, such as titles, URLs, n-grams, and multi-text identifiers used by SEAL, MINDER, and LTRGR; the argument hinges on how each type behaves when the prefix tree is extended with docids for new documents. The supporting analytical tool is the Initial Document Bias Index (IDBI), which measures whether Top-K results contain more initial documents than their corpus share would predict. The proposed method is MDGR, a multi-docid design that chunks documents, maps each chunk to an existing numeric docid via product quantization, and ranks documents by coverage and beam-search rank of the generated docids.

What would settle it

Run the same protocol with chronological or topic-based splits, where the five added chunks come from later periods with new vocabulary; if text-docid models no longer beat BM25 and dense retrieval on those chunks, the central claim does not extend beyond the random-split assumption.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that generalization to unseen documents in generative retrieval is governed by the docid representation rather than by model size or training objective. Using a protocol that trains each model on half of a corpus and then incrementally indexes the remaining half in five chunks, the paper reports that text-based docid models such as SEAL and LTRGR keep Hit@10 on new documents at 0.725–0.799 on NQ, comparable to or better than DPR-HN (0.632) and BM25 (0.586), while numeric-docid models fall to 0.209–0.542. The paper attributes the gap to three properties of text docids: semantic familiarity with the pretrained language model, fine-grained multi-docid coverage, and high lexical diversity. It also introduces the Initial Document Bias Index (IDBI), showing that numeric docids over-retrieve initial documents, and it proposes MDGR, which assigns frozen numeric docids to chunks of new documents under constrained expansion, preserving efficiency while improving dynamic-corpus performance over plain numeric baselines.

Load-bearing premise

The simulation treats newly added documents as random chunks of the same static collection, so the tests do not include real-world drift in topic, style, or vocabulary over time.

Editorial extensions

If this is right

  • Frozen generative retrieval models can serve a growing corpus at index-update cost only, if their docids are text-like and fine-grained; no retraining is needed for new documents.
  • Numeric-docid GR models, as currently designed, cannot be trusted for dynamic collections without continual learning; their generation bias toward the training corpus is measurable with IDBI.
  • Docid design becomes a first-class decision in GR deployment: finer granularity and larger effective vocabulary translate directly into generalization on unseen documents.
  • MDGR-style constrained expansion shows that a numeric docid model can narrow most of the generalization gap to text docids while keeping the storage and decoding speed of numeric codes.
  • Existing continual-learning gains for numeric-docid models come with forgetting on initial documents, whereas text-docid models already have most of the benefit without retraining.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the semantic-familiarity explanation is right, the advantage of text docids should scale with the quality of the underlying language model's pretraining; a model pretrained on unrelated text should show less benefit from n-gram docids.
  • The IDBI could be repurposed as a diagnostic for continual learning and forgetting in sequence-to-sequence models beyond retrieval, since it isolates bias toward earlier training items.
  • A direct extension would use temporal or topic-drift splits: the paper's random-split protocol assumes new documents come from the same distribution, so ordered by time the text-docid advantage may shrink or reverse.
  • MDGR's constrained reuse of old docids suggests a broader principle: in generative retrieval, novelty is handled by composing familiar tokens rather than minting new codes, so token-vocabulary closure may matter more than docid semanticity.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 8 minor

Summary. The paper presents a reproducibility and analysis study of generative retrieval (GR) models over dynamic corpora. It partitions NQ and MS-MARCO into an initial set D0 and five incremental chunks D1–D5, trains frozen GR models on D0 only, and compares text-based docid models (SEAL, MINDER, LTRGR, Ultron-URL) against numeric-based docid models (DSI-SE, Ultron-PQ, NCI, GenRET), with BM25, DPR, and DPR-HN as sparse/dense baselines. The main empirical findings are that text-based docids generalize better to newly added documents, that finer-grained and more lexically diverse docids help, and that numeric-based docids suffer from a bias toward the initial document set. The paper then proposes MDGR, a numeric-based multi-docid method with chunk-level product quantization, constrained docid expansion, and a coverage-plus-rank inference score, and reports results on NQ showing that it narrows the gap between numeric efficiency and text effectiveness. A secondary experiment compares DSI and SEAL with and without DSI++-style incremental training.

Significance. The study is useful and timely: it independently reproduces several GR systems, provides a systematic comparison under a common dynamic-corpus protocol, and offers a plausible explanation of numeric-docid failure via the IDBI bias measure. The paper ships code, follows official implementations for baselines, and makes falsifiable claims about docid granularity and lexical diversity. If the results hold, the finding that frozen text-based GR models can retrieve newly added documents without retraining, and that MDGR preserves much of this capability with numeric efficiency, is a meaningful contribution to the GR literature. The main risks are that the dynamic setting is simulated by an i.i.d. random split, MDGR is evaluated on only one dataset, and the statistical significance statement is not backed by experimental details.

major comments (4)
  1. [§3.4] The statistical validation section states that all results in Tables 1 and 2 achieved significance at p < 0.05, but it does not report the test used, the number of random seeds or runs, or the standard deviation of the reported Hit@10 values. Because several comparisons in Tables 1, 2, and 4 involve small margins (for example, LTRGR vs. MINDER on NQ initial documents), this assertion is not verifiable. Please provide the test procedure, variance across runs, and run counts.
  2. [§3.1 and Definition 3 Remark] The dynamic-corpus benchmark is constructed by randomly splitting a static dataset into a 50% initial set and five 10% chunks, which makes the newly added documents i.i.d. draws from the same distribution as D0. This is the regime most favorable to the paper's 'semantic familiarity' mechanism, since new documents share the vocabulary and topic distribution of the training set. The paper's own remark after Definition 3 acknowledges that IDBI assumes initial and new documents come from the same latent relevance distribution, and the same assumption underlies the main generalization comparison. The central claim that text-based docids generalize better to unseen documents is therefore only validated under no distribution drift. Please add a shifted or temporally ordered split (e.g., ordering by date or by vocabulary novelty) or explicitly restrict the conclusions to i.i.d. corpus growth.
  3. [§6.3, Table 6] MDGR, the paper's proposed method, is evaluated only on the NQ dataset; no MS-MARCO results are reported for it. Since the paper's main empirical claims are established on two datasets and MDGR is presented as a general framework, the single-dataset evaluation is insufficient to support the claimed 'improved performance in dynamic corpus.' Please report MDGR on MS-MARCO or substantially temper the contribution claim.
  4. [§6.1, Figure 3] The docid vocabulary size k=1024 for MDGR is selected from a sweep over {64, 256, 1024, 4096, 8192} performed on the same NQ test sets that are later used for the final comparison in Table 6. It is unclear whether the selection was made on a held-out portion or with knowledge of the test results, so the final MDGR numbers may include an optimistic model-selection effect. Please clarify the model-selection protocol and report sensitivity of the main conclusion to k.
minor comments (8)
  1. [§5.1, Eq. (6)] The semantic familiarity S is defined but never directly estimated; the paper infers 'alignment' indirectly from docid type and IDBI. Please state whether S is operationalized or used only as a conceptual device.
  2. [§3.2] There is a typo in 'ULtron variants' in the implementation details; it should read 'Ultron variants'.
  3. [Abstract/footnote] The code availability line says 'available at here' without a visible URL; please provide a working link or repository identifier.
  4. [Figures 1–3] In the manuscript PDF, the figures contain garbled 'uni000...' token sequences in axis labels and legends, making them hard to interpret. Please ensure the figures are rendered with readable text.
  5. [§6.2, Eq. (8)] The inference score depends on the hyperparameter beta, but the paper never reports the value used in Table 6 or a sensitivity analysis for beta.
  6. [Table 7] The table reports 'Memory', 'Tok-K', and 'Latency' without defining the measurement setup; please specify whether memory is model size only or includes the index, and describe the hardware and batch size used for latency.
  7. [Table 5] The numeric docid size range '32–10,000' is ambiguous; please clarify whether this refers to cluster count per subvector, total docid vocabulary size, or something else.
  8. [§4, Table 3] The comparison between DSI and DSI++ and between SEAL and SEAL++ is interesting but lacks significance information; the same statistical details requested for Tables 1 and 2 should be supplied here as well.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the central claims are empirical comparisons, and the explanatory constructs and design choices are stated as post hoc analysis or explicit methodology rather than as predictions derived from their own inputs.

full rationale

This paper is an empirical replication and exploration study, not a derivation from first principles. The central claims—that text-based docids generalize better than numeric-based docids over dynamic corpora, and that fine-grained docid designs help—are supported by measured Hit@10, forgetting, and generalization metrics on NQ and MS-MARCO. The semantic familiarity quantity S (Eq. 6) is introduced as a post hoc explanatory definition, not fitted to the results and then rediscovered as a prediction. The IDBI metric (Eq. 7) is computed from retrieval outputs, and its explicitly stated assumption that initial and new documents share a latent relevance distribution is a limitation on external validity, not a circular step, because the index itself is not derived from that assumption. MDGR's constrained docid expansion deliberately assigns existing numeric docids to new document chunks; this is a transparent design choice and is evaluated through ablations, not presented as an independent prediction. Baselines, including the self-cited GenRET, are reproduced from official implementations, so no load-bearing argument reduces to an unverified self-citation. Overall, no predicted quantity is equivalent by construction to a fitted input or to a self-citation chain, and the paper's conclusions remain empirical findings that could be challenged by different dynamic-corpus assumptions.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The paper introduces no new physical or symbolic entity; its new artifacts are the MDGR docid scheme and the IDBI metric, which are captured as free parameters and domain assumptions. The main untested inputs are the i.i.d. split assumption, the relevance-uniformity assumption behind IDBI, and the semantic-familiarity explanation.

free parameters (4)
  • MDGR docid vocabulary size k = 1024 per PQ subspace (docid length 4)
    Selected from a sweep over k=64 to 8192 on the same NQ dynamic evaluation (Figure 3); this is model selection on the reported test set.
  • MDGR docid length (number of PQ subvectors) = 4
    A design choice for product quantization that fixes the docid sequence length and affects representational capacity.
  • Chunking window size and stride = 256 tokens, stride 128
    Chosen by hand in Section 6.3; determines the granularity of the multi-docid design.
  • Inference ranking weight beta = not reported
    Equation (8) introduces beta as the weight on the beam-rank term; no value or tuning procedure is given, so the MDGR ranking score is not fully specified.
assumptions (4)
  • domain assumption New documents are independent random samples from the same distribution as the initial corpus (Section 3.1).
    The dynamic corpus is created by randomly partitioning a static dataset. Real dynamic corpora involve temporal distribution shift, which this setup does not test.
  • domain assumption The IDBI interpretation assumes initial and new documents share the same latent query-document relevance distribution (Remark after Definition 3).
    Without this assumption the expected count E_init would not be uniform and the bias index could mislead.
  • domain assumption Semantic familiarity S = E[log P_LM(x)] is a valid explanatory proxy for retrieval generalization (Hypothesis 1, Equation 6).
    The paper defines S but never measures it or verifies its correlation with retrieval performance; it is used as a post hoc explanation.
  • domain assumption Reusing only initial docids to index new document chunks preserves enough semantic consistency for retrieval (Section 6.1, Constrained docid expansion).
    MDGR's core trick assumes old numeric codes can represent new content without retraining; the paper provides empirical support on NQ only, not a mechanism.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Replication and Exploration of Generative Retrieval over Dynamic Corpora." pith.science (2026). https://pith.science/paper/XYIFM2WE

@misc{pith2026250417519,
  author       = {Pith},
  title        = {Pith review of: Replication and Exploration of Generative Retrieval over Dynamic Corpora},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XYIFM2WE}},
  note         = {Machine review of arXiv:2504.17519}
}
read the original abstract

Generative retrieval (GR) has emerged as a promising paradigm in information retrieval (IR). However, most existing GR models are developed and evaluated using a static document collection, and their performance in dynamic corpora where document collections evolve continuously is rarely studied. In this paper, we first reproduce and systematically evaluate various representative GR approaches over dynamic corpora. Through extensive experiments, we reveal that existing GR models with \textit{text-based} docids show superior generalization to unseen documents. We observe that the more fine-grained the docid design in the GR model, the better its performance over dynamic corpora, surpassing BM25 and even being comparable to dense retrieval methods. While GR models with \textit{numeric-based} docids show high efficiency, their performance drops significantly over dynamic corpora. Furthermore, our experiments find that the underperformance of numeric-based docids is partly due to their excessive tendency toward the initial document set, which likely results from overfitting on the training set. We then conduct an in-depth analysis of the best-performing GR methods. We identify three critical advantages of text-based docids in dynamic corpora: (i) Semantic alignment with language models' pretrained knowledge, (ii) Fine-grained docid design, and (iii) High lexical diversity. Building on these insights, we finally propose a novel multi-docid design that leverages both the efficiency of numeric-based docids and the effectiveness of text-based docids, achieving improved performance in dynamic corpus without requiring additional retraining. Our work offers empirical evidence for advancing GR methods over dynamic corpora and paves the way for developing more generalized yet efficient GR models in real-world search engines.

Figures

Figures reproduced from arXiv: 2504.17519 by the authors.

Figure 1
Figure 1. IDBI results for different retrieval methods on NQ [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. Hit@10 performance of n-gram docid and three [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Hit@10 performance of different decoding dimen [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

39 extracted references · 10 canonical work pages

  1. [1]

    Payal Bajaj, Daniel Campos, Nick Craswell, Li Deng, Jianfeng Gao, Xiaodong Liu, Rangan Majumder, Andrew McNamara, Bhaskar Mitra, Tri Nguyen, et al. 2016. Ms marco: A human generated machine reading comprehension dataset. arXiv preprint arXiv:1611.09268 (2016)

    arXiv 2016

  2. [2]

    Michele Bevilacqua, Giuseppe Ottaviano, Patrick Lewis, Scott Yih, Sebastian Riedel, and Fabio Petroni. 2022. Autoregressive search engines: Generating substrings as document identifiers. Advances in Neural Information Processing Systems 35 (2022), 31668–31683

    2022

  3. [3]

    Jiangui Chen, Ruqing Zhang, Jiafeng Guo, Maarten de Rijke, Wei Chen, Yixing Fan, and Xueqi Cheng. 2023. Continual learning for generative retrieval over dynamic corpora. In Proceedings of the 32nd ACM International Conference on Information and Knowledge Management . 306–315

    work page 2023

  4. [4]

    Jiangui Chen, Ruqing Zhang, Jiafeng Guo, Yiqun Liu, Yixing Fan, and Xueqi Cheng. 2022. Corpusbrain: Pre-train a generative retrieval model for knowledge- intensive language tasks. In Proceedings of the 31st ACM International Conference on Information & Knowledge Management . 191–200

    work page 2022

  5. [5]

    Nicola De Cao, Gautier Izacard, Sebastian Riedel, and Fabio Petroni. 2020. Au- toregressive entity retrieval. arXiv preprint arXiv:2010.00904 (2020)

    arXiv 2020

  6. [6]

    Yixing Fan, Xiaohui Xie, Yinqiong Cai, Jia Chen, Xinyu Ma, Xiangsheng Li, Ruqing Zhang, Jiafeng Guo, et al . 2022. Pre-training methods in information retrieval. Foundations and Trends® in Information Retrieval 16, 3 (2022), 178–317

    2022

  7. [7]

    Jiafeng Guo, Changjiang Zhou, Ruqing Zhang, Jiangui Chen, Maarten de Rijke, Yixing Fan, and Xueqi Cheng. 2024. Corpusbrain++: A continual generative pre-training framework for knowledge-intensive language tasks. arXiv preprint arXiv:2402.16767 (2024)

    arXiv 2024

  8. [8]

    Neil Houlsby, Andrei Giurgiu, Stanislaw Jastrzebski, Bruna Morrone, Quentin De Laroussilhe, Andrea Gesmundo, Mona Attariyan, and Sylvain Gelly. 2019. Parameter-efficient transfer learning for NLP. In International conference on machine learning. PMLR, 2790–2799

    2019

Show all 39 references
  1. [9]

    Bowen Jin, Hansi Zeng, Guoyin Wang, Xiusi Chen, Tianxin Wei, Ruirui Li, Zhengyang Wang, Zheng Li, Yang Li, Hanqing Lu, et al. 2023. Language models as semantic indexers. arXiv preprint arXiv:2310.07815 (2023)

    2023 arXiv

  2. [10]

    Vladimir Karpukhin, Barlas Oğuz, Sewon Min, Patrick Lewis, Ledell Wu, Sergey Edunov, Danqi Chen, and Wen-tau Yih. 2020. Dense passage retrieval for open- domain question answering. arXiv preprint arXiv:2004.04906 (2020)

    2020 arXiv

  3. [11]

    Chaeeun Kim, Soyoung Yoon, Hyunji Lee, Joel Jang, Sohee Yang, and Minjoon Seo. 2023. Exploring the practicality of generative retrieval on dynamic corpora. arXiv preprint arXiv:2305.18952 (2023)

    2023 arXiv

  4. [12]

    Varsha Kishore, Chao Wan, Justin Lovelace, Yoav Artzi, and Kilian Q Wein- berger. 2023. Incdsi: incrementally updatable document retrieval. In International Conference on Machine Learning . PMLR, 17122–17134

    2023

  5. [13]

    Tom Kwiatkowski, Jennimaria Palomaki, Olivia Redfield, Michael Collins, Ankur Parikh, Chris Alberti, Danielle Epstein, Illia Polosukhin, Jacob Devlin, Kenton Lee, et al. 2019. Natural questions: a benchmark for question answering research. Transactions of the Association for C...

    2019

  6. [14]

    Xiaoxi Li, Jiajie Jin, Yujia Zhou, Yuyao Zhang, Peitian Zhang, Yutao Zhu, and Zhicheng Dou. 2024. From matching to generation: A survey on generative information retrieval. arXiv preprint arXiv:2404.14851 (2024)

    2024 arXiv

  7. [15]

    Yongqi Li, Nan Yang, Liang Wang, Furu Wei, and Wenjie Li. 2023. Multiview identifiers enhanced generative retrieval. arXiv preprint arXiv:2305.16675 (2023)

    2023 arXiv

  8. [16]

    Yongqi Li, Nan Yang, Liang Wang, Furu Wei, and Wenjie Li. 2024. Learning to rank in generative retrieval. In Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 38. 8716–8723

    2024

  9. [17]

    Yongqi Li, Zhen Zhang, Wenjie Wang, Liqiang Nie, Wenjie Li, and Tat-Seng Chua

  10. [18]

    Jimmy Lin, Xueguang Ma, Sheng-Chieh Lin, Jheng-Hong Yang, Ronak Pradeep, and Rodrigo Nogueira. 2021. Pyserini: An easy-to-use python toolkit to support replicable ir research with sparse and dense representations. arXiv preprint arXiv:2102.10073 (2021)

    2021 arXiv

  11. [19]

    Jimmy Lin, Rodrigo Nogueira, and Andrew Yates. 2022. Pretrained transformers for text ranking: Bert and beyond . Springer Nature

    2022

  12. [20]

    Xinyu Ma, Jiafeng Guo, Ruqing Zhang, Yixing Fan, and Xueqi Cheng. 2022. Pre-train a discriminative text encoder for dense retrieval via contrastive span prediction. In Proceedings of the 45th international ACM SIGIR conference on research and development in information retriev...

    2022

  13. [21]

    Xinyu Ma, Jiafeng Guo, Ruqing Zhang, Yixing Fan, and Xueqi Cheng. 2022. Scattered or connected? an optimized parameter-efficient tuning approach for information retrieval. In Proceedings of the 31st ACM International Conference on Information & Knowledge Management . 1471–1480

    2022

  14. [22]

    Manning, Hinrich Schütze, and Prabhakar Raghavan

    Christopher D. Manning, Hinrich Schütze, and Prabhakar Raghavan. 2009. Intro- duction to Information Retrieval . Cambridge University Press

    2009

  15. [23]

    Sanket Vaibhav Mehta, Jai Gupta, Yi Tay, Mostafa Dehghani, Vinh Q Tran, Jinfeng Rao, Marc Najork, Emma Strubell, and Donald Metzler. 2022. DSI++: Updating transformer memory with new documents. arXiv preprint arXiv:2212.09744 (2022)

    2022 arXiv

  16. [24]

    Donald Metzler, Yi Tay, Dara Bahri, and Marc Najork. 2021. Rethinking search: making domain experts out of dilettantes. In Acm sigir forum, Vol. 55. ACM New York, NY, USA, 1–27

    2021

  17. [25]

    Yingqi Qu, Yuchen Ding, Jing Liu, Kai Liu, Ruiyang Ren, Wayne Xin Zhao, Daxi- ang Dong, Hua Wu, and Haifeng Wang. 2020. RocketQA: An optimized training approach to dense passage retrieval for open-domain question answering. arXiv preprint arXiv:2010.08191 (2020)

    2020 arXiv

  18. [26]

    Stephen Robertson, Hugo Zaragoza, et al . 2009. The probabilistic relevance framework: BM25 and beyond. Foundations and Trends® in Information Retrieval 3, 4 (2009), 333–389

    2009

  19. [27]

    Weiwei Sun, Lingyong Yan, Zheng Chen, Shuaiqiang Wang, Haichao Zhu, Pengjie Ren, Zhumin Chen, Dawei Yin, Maarten Rijke, and Zhaochun Ren. 2024. Learning to tokenize for generative retrieval. Advances in Neural Information Processing Systems 36 (2024)

    2024

  20. [28]

    Yubao Tang, Ruqing Zhang, Jiafeng Guo, Maarten de Rijke, Wei Chen, and Xueqi Cheng. 2024. Generative Retrieval Meets Multi-Graded Relevance. arXiv preprint arXiv:2409.18409 (2024)

    2024 arXiv

  21. [29]

    Yubao Tang, Ruqing Zhang, Jiafeng Guo, Maarten de Rijke, Wei Chen, and Xueqi Cheng. 2024. Listwise generative retrieval models via a sequential learning process. ACM Transactions on Information Systems 42, 5 (2024), 1–31

    2024

  22. [30]

    Yi Tay, Vinh Tran, Mostafa Dehghani, Jianmo Ni, Dara Bahri, Harsh Mehta, Zhen Qin, Kai Hui, Zhe Zhao, Jai Gupta, et al. 2022. Transformer memory as a differentiable search index. Advances in Neural Information Processing Systems 35 (2022), 21831–21843

    2022

  23. [31]

    Yujing Wang, Yingyan Hou, Haonan Wang, Ziming Miao, Shibin Wu, Qi Chen, Yuqing Xia, Chengmin Chi, Guoshuai Zhao, Zheng Liu, et al . 2022. A neural corpus indexer for document retrieval. Advances in Neural Information Processing Systems 35 (2022), 25600–25614

    2022

  24. [32]

    Tianchi Yang, Minghui Song, Zihan Zhang, Haizhen Huang, Weiwei Deng, Feng Sun, and Qi Zhang. 2023. Auto search indexer for end-to-end document retrieval. arXiv preprint arXiv:2310.12455 (2023)

    2023 arXiv

  25. [33]

    Hansi Zeng, Chen Luo, Bowen Jin, Sheikh Muhammad Sarwar, Tianxin Wei, and Hamed Zamani. 2024. Scalable and effective generative information retrieval. In Proceedings of the ACM on Web Conference 2024 . 1441–1452

    2024

  26. [34]

    Peitian Zhang, Zheng Liu, Yujia Zhou, Zhicheng Dou, Fangchao Liu, and Zhao Cao. 2024. Generative retrieval via term set generation. In Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval. 458–468

    2024

  27. [35]

    Yujia Zhou, Zhicheng Dou, and Ji-Rong Wen. 2023. Enhancing generative retrieval with reinforcement learning from relevance feedback. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing . 12481–12490

    2023

  28. [36]

    Yujia Zhou, Jing Yao, Zhicheng Dou, Ledell Yu Wu, Peitian Zhang, and Ji rong Wen. 2022. Ultron: An Ultimate Retriever on Corpus with a Model-based Indexer. ArXiv abs/2208.09257

    2022 arXiv

  29. [37]

    Shengyao Zhuang, Houxing Ren, Linjun Shou, Jian Pei, Ming Gong, Guido Zuc- con, and Daxin Jiang. 2022. Bridging the gap between indexing and retrieval for differentiable search index with query generation.arXiv preprint arXiv:2206.10128 (2022)

    2022 arXiv

  30. [38]

    Noah Ziems, Wenhao Yu, Zhihan Zhang, and Meng Jiang. 2023. Large language models are built-in autoregressive search engines.arXiv preprint arXiv:2305.09612 (2023)

    2023 arXiv

  31. [2024]

    arXiv preprint arXiv:2402.10769 (2024)

    Distillation Enhanced Generative Retrieval. arXiv preprint arXiv:2402.10769 (2024)

    2024 arXiv

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.