Pith. sign in

REVIEW 4 major objections 5 minor 1 cited by

SemCORE: A Semantic-Enhanced Generative Cross-Modal Retrieval Framework with MLLMs

T0 review · 4 major / 5 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 SemCORE claims that giving generative cross-modal retrievers natural-language identifiers—keyword descriptions of each image plus an MLLM verification step—raises text-to-image Recall@1 by 8.65 points on average and matches traditional…

desk verdict SemCORE's structured natural-language IDs plus an MLLM verification reranker is a reasonable engineering contribution, but the headline gains are driven mainly by the reranker, not the generative identifier, and the paper needs extra baselines before the central claim is credible. read the letter →

arxiv 2504.13172 v1 pith:XKN2C4DE submitted 2025-04-17 cs.IR cs.CLcs.MM

Haoxuan Li , Yi Bin , Yunshan Ma , Guoqing Wang , Yang Yang , See-Kiong Ng , Tat-Seng Chua This is my paper · ORCID
classification cs.IRcs.CLcs.MM
keywords cross-modalretrievalgenerativemulti-modallargelanguagemodelsstructurednaturalidentifiersemanticverificationtext-to-imageimage-to-textRecall@1
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

SemCORE claims that generative cross-modal retrieval can be made substantially more accurate if the identifiers a model is trained to emit are built from natural language rather than numbers or cluster labels. The paper proposes a Structured natural language IDentifier (SID): a cluster-based global ID for coarse localization plus keyword-based lexical IDs extracted from MLLM-generated captions, followed by a Generative Semantic Verification (GSV) step in which the MLLM picks the best candidate among a shortlist. On Flickr30K and MS-COCO, the framework reports an average gain of 8.65 points in text-to-image Recall@1 over prior generative methods, and for image-to-text retrieval it reports results on par with or above traditional one-tower and two-tower baselines. The paper also claims to be the first generative cross-modal retrieval framework to handle both retrieval directions.

What carries the argument

The load-bearing mechanism is the SID-GSV pipeline. SID is a two-part identifier: a Global ID (a K-Means cluster index over caption embeddings) that narrows the search, and a Lexical ID (four TF-IDF-deduplicated KeyBERT keywords from the image's MLLM-generated caption) that carries fine-grained semantics. GSV then reranks the beam of candidates by prompting the MLLM to compare each candidate image (with its lexical ID) against the query and select the best match. Constrained beam search over a Trie of valid SIDs keeps every emitted identifier legal.

What would settle it

Build a Flickr30K split where every test image's lexical ID contains at least one keyword token that never appears in any training identifier, run SemCORE with and without GSV, and compare R@1. If performance falls to near chance without GSV, the claimed compositional generalization collapses; if it stays high, the verification step alone is doing the work.

Watch

Extended reading notes

Core claim

The central discovery is that identifier semantics, not just the generative backbone, determine how well an MLLM can retrieve. Previous generative retrievers used hierarchical-clustering numeric IDs or tokenized image codes; SemCORE replaces the final random suffix with keywords that describe the image and lets the MLLM verify the top candidates with an image-plus-keywords prompt. The authors argue that this activates the model's pretrained natural-language understanding, turning retrieval into a form of grounded generation, and that the same design works symmetrically for text-to-image and image-to-text.

Load-bearing premise

The framework assumes the model can, at test time, generate keyword tokens for images it has never seen even when those keywords did not appear in any training identifier, relying on semantic similarity rather than memorization, and that this transfer—not the verification step—is what drives the reported gains.

Editorial extensions

If this is right

  • Text-to-image generative retrieval becomes competitive with traditional similarity-based methods, since Recall@1 improves by 8.65 points on average and rSum is best on both evaluated datasets.
  • Image-to-text retrieval, previously weak in the generative paradigm, can match or exceed one-tower and two-tower baselines while preserving the efficiency of generative decoding.
  • Larger MLLM backbones keep improving retrieval performance from 1B to 8B parameters, so the approach should benefit from further model scaling.
  • Removing SID or GSV hurts results sharply (R@1 drops from 69.0 to 66.1 with SID removed and to 51.9 with GSV removed), showing that both semantic identifiers and verification carry weight.

Reading between the lines

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

  • The reported 8.65-point gain likely bundles the effect of the verification reranker with the effect of the semantic identifiers; an ablation that holds GSV fixed while swapping SID for random suffixes would separate the two contributions, and the paper does not report it.
  • Because identifiers are built from MLLM-generated captions and the same model family is fine-tuned to emit them, the semantic signal is partially self-referential; building identifiers with a captioning model from a different family could test whether the gain survives.
  • The method's dependence on keyword tokens that appear at test time suggests a compositional-generalization test: construct a split where test identifiers contain only unseen keywords and measure R@1 without GSV to see whether the model is truly generalizing.
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 / 5 minor

Summary. SemCORE proposes a unified generative cross-modal retrieval framework in which each image is assigned a Structured natural language IDentifier (SID), formed by a K-Means-based global ID and a KeyBERT-based lexical ID, and an MLLM is fine-tuned to generate these identifiers under constrained decoding. A Generative Semantic Verification (GSV) stage then reranks a small candidate set by prompting the MLLM to select the image that best matches the query. The paper reports text-to-image Recall@1 gains over prior generative baselines on Flickr30K and MS-COCO, with an average improvement of 8.65 points, and reports image-to-text results on Flickr30K that match or exceed traditional one-tower and two-tower methods. The authors claim that SemCORE is the first generative cross-modal retrieval framework to handle both text-to-image and image-to-text retrieval.

Significance. If the reported gains are attributable to the proposed identifier design and generative decoding, the paper would make a useful contribution by showing that generative cross-modal retrieval can be competitive with similarity-based retrieval while unifying both retrieval directions. The paper is clearly structured, presents extensive ablations, includes a scaling study across backbone sizes, and provides concrete case analyses. However, the current experiments do not isolate the effect of the generative component from the effect of the MLLM-based verification reranker, and the comparison is confounded by differences in backbone models. These issues are load-bearing for the central claim, so the significance cannot be fully assessed until the missing baselines and ablations are supplied.

major comments (4)
  1. [Table 3 / Table 1] The decisive ablation is missing. In Table 3, removing GSV drops Flickr30K R@1 from 69.0 to 51.9, while Table 1 reports the best generative baseline AVG at 62.8; without the verification stage, SemCORE is 10.9 points below AVG. Because GSV (Section 3.3) is a candidate reranker that can be attached to any first-stage retriever, the headline '8.65-point average R@1 improvement' is not yet shown to be an advance in generative retrieval. Please report AVG+GSV and a CLIP-top-k+GSV baseline, and report the full SemCORE pipeline both with and without the reranker.
  2. [Section 4.1.2 / Table 1] The comparison is confounded by backbone differences. SemCORE uses InternVL2.5 (Section 3.4), while the generative baselines GRACE, IRGen, and AVG use different base models, so the reported gains may reflect the capacity of the base MLLM rather than the proposed SID and GSV. Please add a same-backbone generative baseline, such as GRACE or IRGen fine-tuned on InternVL2.5, or an equivalently capable generative baseline, to establish that the gains come from the identifier design and decoding strategy.
  3. [Section 3.2 / Section 3.3] The identifier construction is self-referential. Lexical IDs are extracted from captions generated by an MLLM (Section 3.2.1), the same family of MLLMs is fine-tuned to generate those IDs (Section 3.4), and the same model family is used in the GSV verification prompt (Section 3.3). The reported 'semantic understanding' gains may therefore partly measure the model's agreement with its own captioning behavior. Please evaluate identifiers built from human captions or from an off-the-shelf captioning model of a different family, and report whether test-set keyword tokens that the model never saw during training are actually generated for unseen images.
  4. [Table 2 / Section 4.2.2] The image-to-text evaluation is not yet sufficient to support the claim that SemCORE bridges the gap between generative and traditional cross-modal retrieval. Only one generative baseline (GRACE*) is reproduced, and AVG is excluded because it is not based on an MLLM (Section 4.1.2). Please compare against a strong generative MLLM baseline, ideally with the same backbone as SemCORE, and report image-to-text results for that baseline both with and without a GSV-style reranker.
minor comments (5)
  1. [Table 3 / Section 4.3] The table heading 'w/o GCV' should be 'w/o GSV' to match the text, and the text contains a typo 'SemCOER' in the ablation paragraph.
  2. [Section 4.2] The word 'framweork' in the first sentence of Section 4.2 is a typo and should be 'framework'.
  3. [Appendix A.1 / Section 3.3] The appendix states that the GSV candidate set size is fixed at 10, but Section 3.3 does not specify how the candidate set is selected from the beam output or how the candidate set relates to the beam size; please clarify this procedure.
  4. [Figure 3] The figure caption does not name the axes explicitly; please state in the caption that the horizontal axis is cluster size and the vertical axis is Recall@1, or add axis labels to the figure itself.
  5. [Tables 1-4] The metric name is written inconsistently as 'rSum' in the text and 'rsum' in Table 4; please unify the notation.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: SID labels are a deterministic pseudo-label function of image captions, training uses external human query-image pairs, and GSV is a trained reranker; the reported gains are measured against external benchmarks.

full rationale

SemCORE's derivation chain is empirical rather than formal, and no step reduces to its own input by construction. The SID is a fixed pseudo-label function of each image: an MLLM generates a caption, K-Means assigns a global cluster token, and KeyBERT extracts lexical keywords. Training then maximizes the log-likelihood of these SIDs given human query texts (Eq. 3), and inference ranks corpus SIDs by that same conditional likelihood (Eq. 1). The predicted object is an identifier derived from the target image, not a quantity fitted from the test queries, so Eq. 3 is not Eq. 1 by construction on the test set. GSV is a trained MLLM reranker over a candidate set; although the paper does not report AVG+GSV or CLIP+GSV baselines, that omission weakens attribution of the 8.65-point improvement to the generative SID component, but it does not make the method circular. The self-referential flavor—MLLM-generated captions forming identifiers for a same-family MLLM that also verifies candidates—is a genuine generalization and independence risk, but the paper's ground-truth retrieval targets are external human-annotated pairs on Flickr30K and MS-COCO, not the model's own outputs. The paper's self-citations (e.g., [18], [20], [21]) appear as related work, baselines, and an explanatory analogy for false negatives; they are not an unverified premise that forces the central conclusion. No fitted parameter is renamed as a prediction, no uniqueness theorem is imported from the authors' prior work, and no known result is merely re-labeled, so no specific circular step can be exhibited under the required standard.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The central claim rests on several unverified domain assumptions about the quality of MLLM-generated captions, the discriminative power of keyword extraction, and compositional generalization to unseen identifiers. The paper introduces no new physical entities or parameters beyond standard hyperparameters, but the free parameters listed were chosen by hand and affect the reported numbers.

free parameters (5)
  • cluster_size = 128
    K-Means cluster size for global ID; ablation in Figure 3 shows performance is stable across sizes, so 128 is chosen by hand.
  • lexical_id_length = 4
    Number of keywords in the lexical ID; selected from Figure 3 ablation as the optimal length.
  • gsv_candidate_set_size = 10
    Candidate set size for Generative Semantic Verification, fixed in Appendix A.1.
  • beam_size = 50
    Beam width used in constrained decoding for main results; Appendix Table 5 shows R@1 improves with beam size up to 50.
  • tfidf_dedup_top_k = 5
    Top five TF-IDF keywords used for deduplication in lexical ID construction, as stated in Appendix A.1.
assumptions (5)
  • domain assumption MLLM-generated captions accurately represent the semantic content of each image.
    Section 3.2.1 states the framework uses an MLLM to automatically generate captions that are then used to build identifiers. If captions are noisy or generic, the identifiers lose discriminative power.
  • domain assumption K-Means clusters over caption embeddings yield semantically coherent global ID groups.
    Section 3.2.1 relies on clustering to provide the first token of the identifier; no analysis of cluster purity is given.
  • domain assumption KeyBERT keyword extraction with TF-IDF deduplication produces discriminative lexical IDs.
    Section 3.2.2: the lexical ID must distinguish similar images within a cluster, but the paper does not quantify keyword overlap or failure cases.
  • domain assumption The model can compositionally generate identifier tokens for test images never seen during training.
    At inference the Trie contains test identifiers, but the model was trained only on training image IDs; the method depends on semantic transfer from query to unseen natural-language identifiers.
  • domain assumption The MLLM verification prompt yields reliable fine-grained relevance judgments among candidates.
    Section 3.3: GSV relies on the MLLM selecting the best candidate image given the query and lexical IDs; no calibration or agreement analysis is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of SemCORE: A Semantic-Enhanced Generative Cross-Modal Retrieval Framework with MLLMs." pith.science (2026). https://pith.science/paper/XKN2C4DE

@misc{pith2026250413172,
  author       = {Pith},
  title        = {Pith review of: SemCORE: A Semantic-Enhanced Generative Cross-Modal Retrieval Framework with MLLMs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XKN2C4DE}},
  note         = {Machine review of arXiv:2504.13172}
}
read the original abstract

Cross-modal retrieval (CMR) is a fundamental task in multimedia research, focused on retrieving semantically relevant targets across different modalities. While traditional CMR methods match text and image via embedding-based similarity calculations, recent advancements in pre-trained generative models have established generative retrieval as a promising alternative. This paradigm assigns each target a unique identifier and leverages a generative model to directly predict identifiers corresponding to input queries without explicit indexing. Despite its great potential, current generative CMR approaches still face semantic information insufficiency in both identifier construction and generation processes. To address these limitations, we propose a novel unified Semantic-enhanced generative Cross-mOdal REtrieval framework (SemCORE), designed to unleash the semantic understanding capabilities in generative cross-modal retrieval task. Specifically, we first construct a Structured natural language IDentifier (SID) that effectively aligns target identifiers with generative models optimized for natural language comprehension and generation. Furthermore, we introduce a Generative Semantic Verification (GSV) strategy enabling fine-grained target discrimination. Additionally, to the best of our knowledge, SemCORE is the first framework to simultaneously consider both text-to-image and image-to-text retrieval tasks within generative cross-modal retrieval. Extensive experiments demonstrate that our framework outperforms state-of-the-art generative cross-modal retrieval methods. Notably, SemCORE achieves substantial improvements across benchmark datasets, with an average increase of 8.65 points in Recall@1 for text-to-image retrieval.

Figures

Figures reproduced from arXiv: 2504.13172 by the authors.

Figure 1
Figure 1. Illustrations of existing paradigms for cross-modal [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. An overview of the proposed SemCORE framework (illustrating the text-to-image retrieval process, with the image-to [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Performance with respect to the cluster size of [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Illustration of the structured natural language identifier (SID). The SID comprises two components: global ID and [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Illustration of additional generative retrieval examples. Items from (j) to (o) correspond to the image-to-text retrieval [PITH_FULL_IMAGE:figures/full_fig_p012_5.png]
Figure 6
Figure 6. Figure 6: Illustration of the false-negative phenomenon. Each column demonstrates two text queries, where each query [PITH_FULL_IMAGE:figures/full_fig_p012_6.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Generative Universal Multimodal Retrieval with Dual-role Identifiers

    cs.IR 2026-08 conditional novelty 5.0 of 10

    DrIG reuses one residual-quantized identifier per candidate as both an ordered decoding target and an unordered relevance set, combining constrained beam search with a prefix-independent prior for universal multimodal...

Reference graph

Works this paper leans on

45 extracted references · 19 canonical work pages · cited by 1 Pith paper

  1. [1]

    Jean-Baptiste Alayrac, Jeff Donahue, Pauline Luc, Antoine Miech, Iain Barr, Yana Hasson, Karel Lenc, Arthur Mensch, Katherine Millican, Malcolm Reynolds, et al

  2. [2]

    Yi Bin, Haoxuan Li, Yahui Xu, Xing Xu, Yang Yang, and Heng Tao Shen. 2023. Unifying two-stream encoders with transformers for cross-modal retrieval. In ACM MM. 3041–3050

    work page 2023

  3. [3]

    Tom B Brown. 2020. Language models are few-shot learners. arXiv preprint arXiv:2005.14165 (2020)

    arXiv 2020

  4. [4]

    Jiacheng Chen, Hexiang Hu, Hao Wu, Yuning Jiang, and Changhu Wang. 2021. Learning the best pooling strategy for visual semantic embedding. In CVPR. 15789–15798

    work page 2021

  5. [5]

    Zhe Chen, Weiyun Wang, Yue Cao, Yangzhou Liu, Zhangwei Gao, Erfei Cui, Jinguo Zhu, Shenglong Ye, Hao Tian, Zhaoyang Liu, et al . 2024. Expanding performance boundaries of open-source multimodal models with model, data, and test-time scaling. arXiv preprint arXiv:2412.05271 (2024)

    arXiv 2024

  6. [6]

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. BERT: Pre-training of deep bidirectional transformers for language understanding. In NAACL

    2019

  7. [7]

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xi- aohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, et al. 2020. An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale. In ICLR

    2020

  8. [8]

    Fartash Faghri, David J Fleet, Jamie Ryan Kiros, and Sanja Fidler. 2018. Vse++: Improving visual-semantic embeddings with hard negatives. In BMVC

    work page 2018

Show all 45 references
  1. [9]

    Andrea Frome, Greg Corrado, Jonathon Shlens, Samy Bengio, Jeffrey Dean, Marc’Aurelio Ranzato, and Tomas Mikolov. 2013. Devise: A deep visual-semantic embedding model. In NeurIPS. 2121–2129

    2013

  2. [10]

    Maarten Grootendorst. 2020. KeyBERT: Minimal keyword extraction with BERT. doi:10.5281/zenodo.4461265

    2020 doi

  3. [11]

    Maarten Grootendorst. 2022. BERTopic: Neural topic modeling with a class-based TF-IDF procedure. arXiv preprint arXiv:2203.05794 (2022)

    2022 arXiv

  4. [12]

    John A Hartigan and Manchek A Wong. 1979. Algorithm AS 136: A k-means clustering algorithm. Journal of the royal statistical society. series c (applied statistics) 28, 1 (1979), 100–108

    1979

  5. [13]

    Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, Weizhu Chen, et al. 2022. Lora: Low-rank adaptation of large language models.. In ICLR

    2022

  6. [14]

    Jared Kaplan, Sam McCandlish, Tom Henighan, Tom B Brown, Benjamin Chess, Rewon Child, Scott Gray, Alec Radford, Jeffrey Wu, and Dario Amodei. 2020. Scaling laws for neural language models. arXiv preprint arXiv:2001.08361 (2020)

    2020 arXiv

  7. [15]

    Andrej Karpathy and Li Fei-Fei. 2015. Deep visual-semantic alignments for generating image descriptions. In CVPR. 3128–3137

    2015

  8. [16]

    Doyup Lee, Chiheon Kim, Saehoon Kim, Minsu Cho, and Wook-Shin Han. 2022. Autoregressive image generation using residual quantization. In CVPR. 11523– 11532

    2022

  9. [17]

    Kuang-Huei Lee, Xi Chen, Gang Hua, Houdong Hu, and Xiaodong He. 2018. Stacked cross attention for image-text matching. In ECCV. 201–216

    2018

  10. [18]

    Haoxuan Li, Yi Bin, Junrong Liao, Yang Yang, and Heng Tao Shen. 2023. Your negative may not be true negative: Boosting image-text matching with false negative elimination. In ACM MM. 924–934

    2023

  11. [19]

    Junnan Li, Dongxu Li, Caiming Xiong, and Steven Hoi. 2022. Blip: Bootstrapping language-image pre-training for unified vision-language understanding and generation. In ICML. PMLR, 12888–12900

    2022

  12. [20]

    Yongqi Li, Hongru Cai, Wenjie Wang, Leigang Qu, Yinwei Wei, Wenjie Li, Liqiang Nie, and Tat-Seng Chua. 2024. Revolutionizing Text-to-Image Retrieval as Au- toregressive Token-to-Voken Generation. arXiv preprint arXiv:2407.17274 (2024)

    2024 arXiv

  13. [21]

    Yongqi Li, Wenjie Wang, Leigang Qu, Liqiang Nie, Wenjie Li, and Tat-Seng Chua. 2024. Generative cross-modal retrieval: Memorizing images in multimodal language models for retrieval and beyond.arXiv preprint arXiv:2402.10805 (2024)

    2024 arXiv

  14. [22]

    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

  15. [23]

    Zheng Li, Caili Guo, Zerun Feng, Jenq-Neng Hwang, and Xijun Xue. 2022. Multi- View Visual Semantic Embedding.. In IJCAI, Vol. 2. 7

    2022

  16. [24]

    Tsung-Yi Lin, Michael Maire, Serge Belongie, James Hays, Pietro Perona, Deva Ramanan, Piotr Dollár, and C Lawrence Zitnick. 2014. Microsoft coco: Common objects in context. In ECCV. Springer, 740–755

    2014

  17. [25]

    Chunxiao Liu, Zhendong Mao, An-An Liu, Tianzhu Zhang, Bin Wang, and Yong- dong Zhang. 2019. Focus your attention: A bidirectional focal attention network for image-text matching. In ACM MM. 3–11

    2019

  18. [26]

    Haotian Liu, Chunyuan Li, Qingyang Wu, and Yong Jae Lee. 2024. Visual instruc- tion tuning. NeurIPS 36 (2024)

    2024

  19. [27]

    Tomas Mikolov, Kai Chen, Greg Corrado, and Jeffrey Dean. 2013. Efficient estimation of word representations in vector space. In ICLR. 1–12

    2013

  20. [28]

    Leigang Qu, Meng Liu, Wenjie Wang, Zhedong Zheng, Liqiang Nie, and Tat-Seng Chua. 2023. Learnable pillar-based re-ranking for image-text retrieval. In SIGIR. 1252–1261

    2023

  21. [29]

    Leigang Qu, Meng Liu, Jianlong Wu, Zan Gao, and Liqiang Nie. 2021. Dynamic modality interaction modeling for image-text retrieval. In SIGIR. 1104–1113

    2021

  22. [30]

    Alec Radford. 2018. Improving language understanding by generative pre- training. (2018)

    2018

  23. [31]

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. 2021. Learning transferable visual models from natural language supervision. In ICML. PMLR, 8748–8763

    2021

  24. [32]

    Alec Radford, Jeffrey Wu, Rewon Child, David Luan, Dario Amodei, Ilya Sutskever, et al. 2019. Language models are unsupervised multitask learners. OpenAI blog 1, 8 (2019), 9

    2019

  25. [33]

    Shashank Rajput, Nikhil Mehta, Anima Singh, Raghunandan Hulikal Keshavan, Trung Vu, Lukasz Heldt, Lichan Hong, Yi Tay, Vinh Tran, Jonah Samost, et al

  26. [34]

    Gerard Salton and Christopher Buckley. 1988. Term-weighting approaches in automatic text retrieval. Information processing & management 24, 5 (1988), 513–523

    1988

  27. [35]

    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. NeurIPS 35 (2022), 21831–21843. Conference acronym ’XX, June 03–05, 2018, Woodstock, NY Haoxuan Li et al

    2022

  28. [36]

    Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timothée Lacroix, Baptiste Rozière, Naman Goyal, Eric Hambro, Faisal Azhar, et al. 2023. Llama: Open and efficient foundation language models. arXiv preprint arXiv:2302.13971 (2023)

    2023 arXiv

  29. [37]

    Yujing Wang, Yingyan Hou, Haonan Wang, Ziming Miao, Shibin Wu, Hao Sun, Qi Chen, Yuqing Xia, Chengmin Chi, Guoshuai Zhao, et al. 2022. A neural corpus indexer for document retrieval. In NeurIPS. 25600–25614

    2022

  30. [38]

    Yuan Yao, Tianyu Yu, Ao Zhang, Chongyi Wang, Junbo Cui, Hongji Zhu, Tianchi Cai, Haoyu Li, Weilin Zhao, Zhihui He, et al. 2024. MiniCPM-V: A GPT-4V Level MLLM on Your Phone. arXiv preprint arXiv:2408.01800 (2024)

    2024 arXiv

  31. [39]

    Peter Young, Alice Lai, Micah Hodosh, and Julia Hockenmaier. 2014. From image descriptions to visual denotations: New similarity metrics for semantic inference over event descriptions. TACL 2 (2014), 67–78

    2014

  32. [40]

    Hansi Zeng, Chen Luo, and Hamed Zamani. 2024. Planning ahead in generative retrieval: Guiding autoregressive generation through simultaneous decoding. In SIGIR. 469–480

    2024

  33. [41]

    Kun Zhang, Zhendong Mao, Quan Wang, and Yongdong Zhang. 2022. Negative- aware attention framework for image-text matching. In CVPR. 15661–15670

    2022

  34. [42]

    Peitian Zhang, Zheng Liu, Yujia Zhou, Zhicheng Dou, Fangchao Liu, and Zhao Cao. 2024. Generative retrieval via term set generation. In SIGIR. 458–468

    2024

  35. [43]

    Yidan Zhang, Ting Zhang, Dong Chen, Yujing Wang, Qi Chen, Xing Xie, Hao Sun, Weiwei Deng, Qi Zhang, Fan Yang, et al. 2023. Irgen: Generative modeling for image retrieval. arXiv preprint arXiv:2303.10126 (2023). SemCORE: A Semantic-Enhanced Generative Cross-Modal Retrieval Fram...

    2023 arXiv

  36. [2022]

    NeurIPS 35 (2022), 23716–23736

    Flamingo: a visual language model for few-shot learning. NeurIPS 35 (2022), 23716–23736

    2022

  37. [2023]

    NeurIPS 36 (2023), 10299– 10315

    Recommender systems with generative retrieval. NeurIPS 36 (2023), 10299– 10315

    2023

Pith tools

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