Pith. sign in

REVIEW 3 major objections 6 minor 74 references

Boosting Data Utilization for Multilingual Dense Retrieval

T0 review · 3 major / 6 minor · reviewed 2026-08-04 · deepseek-v4-flash

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

Pith's one-line read This paper shows that curating hard negatives with LLM judgment and generation, plus building monolingual topic-diverse mini-batches, lifts multilingual dense retrieval by up to 2.5 nDCG@10 points over strong baselines on 16 MIRACL language

desk verdict Solid empirical pipeline for multilingual dense retrieval, but the paper contradicts itself on which LLM ran the load-bearing false-negative filtering; fix that before trusting the numbers. read the letter →

arxiv 2509.09459 v1 pith:TEC32CIL submitted 2025-09-11 cs.IR

Chao Huang , Fengran Mo , Yufeng Chen , Changhao Guan , Zhenrui Yue , Xinyu Wang , Jinan Xu , Kaiyu Huang This is my paper
classification cs.IR
keywords multilingualdenseretrievalhardnegativeminingfalsenegativescontrastivelearningLLMjudgmentmini-batchconstructionMIRACLdatautilization
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

The paper argues that the main bottleneck in fine-tuning multilingual dense retrievers is not model architecture but the quality of negative samples and the composition of training mini-batches. It proposes a three-stage pipeline: first, it builds a candidate hard-negative set by combining rankings from multiple multilingual retrievers and uses a large language model to remove likely false negatives (documents that are actually relevant but lack annotations); second, it generates additional hard negatives by instructing an LLM to summarize positive documents and produce new queries that retrieve extra candidates; third, it assembles mini-batches that are monolingual but topic-diverse, and weights contrastive learning negatives by language and topic frequency. On the MIRACL benchmark with 16 languages, the method improves average nDCG@10 by 1.4 points over the strongest baseline, and beats the best alternative hard-negative mining strategy by 2.5 absolute points. Each stage contributes roughly 2 nDCG@10 points, with the false-negative filtering stage making the largest difference.

What carries the argument

The load-bearing mechanism is the three-stage data-construction pipeline. Stage 1: a multilingual retriever ensemble produces a top-k candidate list per query; an LLM scores each candidate against the query and a known positive, and only candidates rated 'irrelevant' survive as hard negatives. Stage 2: a summarization fine-tuned multilingual LLM summarizes each positive, generates a new query, retrieves fresh candidates, and tops up each query's negative set to a fixed size. Stage 3: two topic classifiers assign coarse and fine topics to each positive document; mini-batches are then sampled so all pairs share a language while topics are spread, and the contrastive loss weights negatives by l

What would settle it

Run the same human-validation procedure (three annotators, 100 samples) on languages outside the paper's five validated ones—e.g., Swahili, Telugu, Thai—and compute LLM-judgment agreement. If agreement with human ratings drops below 70% for any of these languages, and the 'w/o LLM judgment' ablation shows no performance drop in that language, the central claim would be unsupported for that language.

Watch

Extended reading notes

Core claim

The central claim is that false negatives—relevant documents that are unlabeled and therefore treated as negatives—seriously degrade multilingual dense retrieval fine-tuning, and that an LLM can identify them well enough to filter them out. The paper shows that combining (i) LLM-based filtering of hard negatives, (ii) LLM-generated supplementary hard negatives, and (iii) mini-batches that are monolingual and topic-balanced, with language/topic-weighted negative sampling, improves retrieval accuracy across all 16 MIRACL languages, with the largest gains on low-resource languages. The paper further reports that replacing any stage with its absence reduces accuracy by about 2 nDCG@10 points, an

Load-bearing premise

The pipeline assumes a large language model can reliably judge whether a candidate document is truly irrelevant to a query across all 16 training languages, but the paper checks this against human labels on only 5 languages.

Editorial extensions

If this is right

  • The method improves average nDCG@10 over strong baselines on MIRACL across 16 languages, with gains in 15 of 16 languages and slight decreases on French and Chinese relative to the backbone.
  • It outperforms existing hard-negative mining heuristics for every backbone tried, with a 2.5-point absolute gain over the best alternative.
  • Each of the three stages contributes independently: removing any one drops nDCG@10 by roughly 2 points, and removing LLM judgment is the most damaging.
  • The pipeline transfers across different backbone models, so it is a data-curation add-on rather than an architecture change.
  • The paper's analysis shows that over 20% of initial hard-negative candidates are false negatives and are removed by the LLM; the optimal initial negative set size is around 30-40 candidates.

Reading between the lines

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

  • If LLM judgment quality is high, this pipeline could be applied to other retrieval tasks where false negatives from pseudo-relevance feedback are common, such as conversational search or cross-lingual QA; the false-negative filtering step should generalize.
  • The finding that same-language, multi-topic mini-batches help might extend to any multilingual contrastive learning, not just retrieval; one could test it on multilingual sentence embeddings.
  • The paper validates LLM false-negative judgments on only 5 languages (100 samples each); the 11 unvalidated languages, including low-resource Swahili, Telugu, and Thai, carry the real risk. A natural extension is a larger human validation across all 16 languages, or a per-language weighting that downweights LLM judgments in low-resource languages.
  • The generated negatives come from LLM-generated queries, so the pipeline could accumulate LLM bias; testing with different LLM families or sizes would reveal whether the gains are LLM-specific.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 6 minor

Summary. The paper proposes a three-stage data-utilization pipeline for fine-tuning multilingual dense retrievers. Stage 1 builds a hard-negative candidate set by ensembling several multilingual retrievers and then filtering out false negatives with an LLM judge. Stage 2 supplements the candidate set using an instruction-tuned LLM and a positive-driven back-forward generation procedure. Stage 3 constructs mini-batches that are monolingual but topic-diverse and adds language- and topic-based weights to the contrastive objective. Experiments on MIRACL (16 languages) report an average nDCG@10 of 70.6 with a BGE backbone, which is +1.4 absolute over BGE and +2.5 over TopK-PercPos; ablations show each stage contributes roughly 2 nDCG@10 points. The paper includes code, a human-validation study, and extensive comparison tables.

Significance. If the results hold, the paper offers a practical data-centric alternative to architecture-level changes for multilingual dense retrieval. The controlled comparison against fine-tuned BGE with standard hard-negative mining (Table 2) is credible, and the stagewise ablations (Table 3) support the claim that each component helps. The paper also ships code and performs a small human validation of LLM judgments. However, two load-bearing issues—an inconsistent attribution of the LLM used for false-negative filtering, and an ambiguous/possibly incorrect weighted-loss formula—must be resolved before the reported gains can be fully credited. The limited language coverage of the human validation is an additional concern for the 16-language claim.

major comments (3)
  1. [§4.1 and Appendix B.2] The main text states 'For hard negatives selection and generation, we employ the Llama-3.1-70B-instruct model' (§4.1), but Appendix B.2 says 'For each hard negative sample in the candidate set, we prompt GPT-4o (2024-11-20) to select the true hard negatives.' These attributions are mutually exclusive. This is not a cosmetic inconsistency: Stage 1 is the largest single ablation component (Table 3: removing Stage 1 costs 3.7 nDCG@10), and the paper's narrative attributes that gain to the 'multilingual ability of LLMs.' If GPT-4o was the model actually used, the method as described is not what was evaluated, and the gains may reflect GPT-4o's proprietary judgment rather than the proposed pipeline. Please state exactly which model performed the false-negative selection, align the appendix, and either rerun the key comparisons with the declared model or clearly disclose the GPT-4o dependence.
  2. [§3.5, Eq. (L_final)] The final loss is written as L_final = 1/|B| Σ_{(q,d+,d−)∈B} ω(d−) × L_MR, but L_MR as defined is already a per-query contrastive loss whose denominator sums over all negatives in D−. Multiplying the entire L_MR by ω(d−) after the summation does not implement per-negative weighting; it is not equivalent to weighting each negative inside the softmax denominator. The actual objective being optimized is therefore unclear, which makes the ablation results in Table 7 and the weight analysis in Table 9 difficult to interpret. Please rewrite the loss explicitly, e.g., L = −log[ e^{sim(q,d+)} / ( e^{sim(q,d+)} + Σ_{d−} ω(d−) e^{sim(q,d−)} ) ], or state precisely how ω(d−) enters the computation.
  3. [Appendix C and §5.1/Table 8] The false-negative filter is applied to all 16 training languages, but human validation is reported for only 5 languages (en, es, zh, hi, bn), with 100 samples each. Eleven languages, including low-resource sw, te, and th, have no validation. The paper's own Limitations section acknowledges that 'the judgment of false negatives via LLMs might still be inaccurate,' and Table 8 shows substantial variation in elimination rates across languages (e.g., 32.6% for en but 9.4% for bn). Given that Stage 1 contributes the largest performance gain, the central 16-language claim rests on the assumption that LLM judgment quality transfers to all languages. Please provide per-language validation data (or a principled argument for transfer) and, failing that, qualify the claim about low-resource languages.
minor comments (6)
  1. [§3.5] The notation 'C={C1, C2, ..., C12}' is confusing: C is used both for the topic count and for the set of topics. Clarify that C=12 and use a different symbol for the set.
  2. [Table 1] The fr column for mE5large is garbled in the typeset table ('54.562.052.9'); this should be corrected to a single number.
  3. [Table 8 caption] The caption and surrounding text are inconsistent: the text says 'the first row indicates the percentage of false negative samples eliminated,' but the first row of the table is 'From Retrieval,' and the elimination percentages appear in the third row. Please align the description with the table.
  4. [References] The reference list contains a malformed entry beginning 'Multi-Linguality Multi-Functionality Multi-Granularity. M3-embedding...' with no author names; this appears to duplicate or replace the proper citation for BGE M3 (Chen et al., 2024). Please fix.
  5. [Appendix D, Tables 15 and 16] The column header 'zhmBERT' in both tables should be split into 'zh' and 'mBERT' to match the per-language columns and model rows used elsewhere.
  6. [§4.1] Minor typos: 'Enlish MS MARCO' should be 'English MS MARCO,' and the Figure 1 caption contains 'fied rice' for 'fried rice.'

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the pipeline is an empirical method validated on an external benchmark; no claim reduces to its inputs or self-citations.

full rationale

The paper proposes a three-stage data-utilization pipeline for multilingual dense retrieval and validates it on the external MIRACL benchmark. The central claim—improved nDCG@10 over strong baselines—is an empirical outcome measured on held-out test collections, not a quantity derived from fitted constants or assumptions that already contain the result. Stage 1's ensemble retrieval plus LLM false-negative filtering, Stage 2's LLM-aided hard-negative generation, and Stage 3's language/topic-balanced mini-batching are training-time data-construction choices. The final loss L_final is a training objective, and the reported numbers are the result of fine-tuning, not an algebraic consequence of the pipeline definitions. Ablations (Tables 3–7) and hyperparameter analyses (Table 9, Figure 3) are standard component studies and parameter sweeps, not circular 'predictions'. Self-citations in the related work are background references and are not load-bearing for the paper's contribution. One non-circular concern is an internal inconsistency: Section 4.1 states 'For hard negatives selection and generation, we employ the Llama-3.1-70B-instruct model,' while Appendix B.2 says 'For each hard negative sample in the candidate set, we prompt GPT-4o (2024-11-20) to select the true hard negatives.' This is a reproducibility/correctness issue, not a circular reduction. The Limitations section's acknowledgment that 'the judgment of false negatives via LLMs might still be inaccurate' is a stated limitation that also speaks to validity, not circularity. No step in the claimed derivation reduces by construction to its own inputs.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central claim rests on several empirical assumptions: LLM judgment quality, the value of same-language topic-diverse batches, and the validity of generated negatives. The paper includes ablations and a human validation subset that partially support these, but the main pipeline scales these assumptions to all training queries without full verification.

free parameters (5)
  • Hard negative candidate set size N = 40 for candidate set, 30 sampled per query
    Selected based on performance in Figure 3; optimal around 30-40.
  • Weight parameters alpha and beta = Table 9 values (e.g., alpha=0.55, beta=0.4)
    Manually tuned; the 'automatic' computation from data portions is not precisely specified.
  • Topic count C = 12
    Predefined number of topic clusters for mini-batch construction.
  • LLM false-negative threshold = final score of 2
    Threshold for calling a candidate a false negative; the combination of the two scoring criteria is not fully specified.
  • Ensemble projection layer E = not specified
    Feature extraction layer unifying encoder outputs; architecture and training procedure not described.
assumptions (4)
  • domain assumption LLM relevance judgments approximate true relevance for all 16 languages
    Stage 1 filters hard negatives using Llama-3.1-70B; validated only on a 5-language, 100-sample subset (Appendix C).
  • domain assumption Same-language in-batch negatives are harder and improve cross-lingual alignment
    Core hypothesis of Stage 3; no theoretical justification; supported only by ablations.
  • domain assumption Generated queries from positive summaries yield valid hard negatives
    Stage 2 generates new queries to retrieve negatives; no direct validation that these are true negatives.
  • domain assumption MIRACL training set and evaluation metric capture real multilingual retrieval quality
    Single benchmark used; no other dataset evaluated.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Boosting Data Utilization for Multilingual Dense Retrieval." pith.science (2026). https://pith.science/paper/TEC32CIL

@misc{pith2026250909459,
  author       = {Pith},
  title        = {Pith review of: Boosting Data Utilization for Multilingual Dense Retrieval},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TEC32CIL}},
  note         = {Machine review of arXiv:2509.09459}
}
read the original abstract

Multilingual dense retrieval aims to retrieve relevant documents across different languages based on a unified retriever model. The challenge lies in aligning representations of different languages in a shared vector space. The common practice is to fine-tune the dense retriever via contrastive learning, whose effectiveness highly relies on the quality of the negative sample and the efficacy of mini-batch data. Different from the existing studies that focus on developing sophisticated model architecture, we propose a method to boost data utilization for multilingual dense retrieval by obtaining high-quality hard negative samples and effective mini-batch data. The extensive experimental results on a multilingual retrieval benchmark, MIRACL, with 16 languages demonstrate the effectiveness of our method by outperforming several existing strong baselines.

Figures

Figures reproduced from arXiv: 2509.09459 by the authors.

Figure 1
Figure 1. Example of false negatives, which refer to [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overview of our framework including three stages: i) construction of hard negative set, ii) LLM-aided [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Model performance with average NDCG@10 on different initial hard negative candidate set sizes N. stage is about 1:9. The better data ratio could be further explore in future study. 5.2 Weight of Hard Negative Sampling During the mini-batch construction, the hard nega￾tive sampling weights are obtained automatically based on the portion of the language and topic among the data points. To analyze the impact of hard ne… view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

74 extracted references · 1 canonical work pages

  1. [1]

    online" 'onlinestring :=

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint eprinttype howpublished institution journal key month note number organization pages publisher school series title type volume year doi pubmed url lastchecked label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block STRING...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    Akari Asai, Xinyan Yu, Jungo Kasai, and Hanna Hajishirzi. 2021. One question answering model for many languages with cross-lingual dense passage retrieval. Advances in Neural Information Processing Systems, 34:7547--7560

    2021

  4. [4]

    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

    arXiv 2016

  5. [5]

    Jianlv Chen, Shitao Xiao, Peitian Zhang, Kun Luo, Defu Lian, and Zheng Liu. 2024. Bge m3-embedding: Multi-lingual, multi-functionality, multi-granularity text embeddings through self-knowledge distillation. arXiv preprint arXiv:2402.03216

    arXiv 2024

  6. [6]

    A Conneau. 2019. Unsupervised cross-lingual representation learning at scale. arXiv preprint arXiv:1911.02116

    arXiv 2019

  7. [7]

    Jacob Devlin. 2018. Bert: Pre-training of deep bidirectional transformers for language understanding. arXiv preprint arXiv:1810.04805

    arXiv 2018

  8. [8]

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. Bert: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 conference of the North American chapter of the association for computational linguistics: human language technologies, volume 1 (long and short papers), pages 4171--4186

    2019

Show all 74 references
  1. [9]

    Bosheng Ding, Chengwei Qin, Ruochen Zhao, Tianze Luo, Xinze Li, Guizhen Chen, Wenhan Xia, Junjie Hu, Luu Anh Tuan, and Shafiq Joty. 2024. Data augmentation using llms: Data perspectives, learning paradigms and challenges. In Findings of the Association for Computational Lingui...

    2024

  2. [10]

    Yingqi Qu Yuchen Ding, Jing Liu, Kai Liu, Ruiyang Ren, Xin Zhao, Daxiang 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 arXiv

  3. [11]

    Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Amy Yang, Angela Fan, et al. 2024. The llama 3 herd of models. arXiv preprint arXiv:2407.21783

    2024 arXiv

  4. [12]

    Sanjay K Dwivedi and Ganesh Chandra. 2016. A survey on cross-language information retrieval. International Journal on Cybernetics & Informatics (IJCI) Vol, 5

    2016

  5. [13]

    Feiteng Fang, Yuelin Bai, Shiwen Ni, Min Yang, Xiaojun Chen, and Ruifeng Xu. 2024. Enhancing noise robustness of retrieval-augmented language models with adaptive adversarial training. arXiv preprint arXiv:2405.20978

    2024 arXiv

  6. [14]

    Luyu Gao, Xueguang Ma, Jimmy Lin, and Jamie Callan. 2022. Tevatron: An efficient and flexible toolkit for dense retrieval. arXiv preprint arXiv:2203.05765

    2022 arXiv

  7. [15]

    https://translate.google Google translate

    Google. https://translate.google Google translate

  8. [16]

    Suchin Gururangan, Ana Marasovi \'c , Swabha Swayamdipta, Kyle Lo, Iz Beltagy, Doug Downey, and Noah A Smith. 2020. Don't stop pretraining: Adapt language models to domains and tasks. arXiv preprint arXiv:2004.10964

    2020 arXiv

  9. [17]

    Xiyang Hu, Xinchi Chen, Peng Qi, Deguang Kong, Kunlun Liu, William Yang Wang, and Zhiheng Huang. 2023. Language agnostic multilingual information retrieval with contrastive learning. In Findings of the Association for Computational Linguistics: ACL 2023, pages 9133--9146

    2023

  10. [18]

    Chao-Wei Huang, Chen-An Li, Tsu-Yuan Hsu, Chen-Yu Hsu, and Yun-Nung Chen. 2024 a . Unsupervised multilingual dense retrieval via generative pseudo labeling. In Findings of the Association for Computational Linguistics: EACL 2024, pages 736--746

    2024

  11. [19]

    Kaiyu Huang, Fengran Mo, Xinyu Zhang, Hongliang Li, You Li, Yuanchi Zhang, Weijian Yi, Yulong Mao, Jinchen Liu, Yuzhuang Xu, et al. 2024 b . A survey on large language models with multilingualism: Recent advances and new frontiers. arXiv preprint arXiv:2405.10936

    2024 arXiv

  12. [20]

    Zhiqi Huang, Hansi Zeng, Hamed Zamani, and James Allan. 2023. Soft prompt decoding for multilingual dense retrieval. In Proceedings of the 46th International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 1208--1218

    2023

  13. [21]

    Gautier Izacard, Mathilde Caron, Lucas Hosseini, Sebastian Riedel, Piotr Bojanowski, Armand Joulin, and Edouard Grave. 2021. Unsupervised dense information retrieval with contrastive learning. arXiv preprint arXiv:2112.09118

    2021 arXiv

  14. [22]

    Armand Joulin. 2016. Fasttext. zip: Compressing text classification models. arXiv preprint arXiv:1612.03651

    2016 arXiv

  15. [23]

    Armand Joulin, Edouard Grave, Piotr Bojanowski, and Tomas Mikolov. 2016. Bag of tricks for efficient text classification. arXiv preprint arXiv:1607.01759

    2016 arXiv

  16. [24]

    Vladimir Karpukhin, Barlas O g uz, 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 arXiv

  17. [25]

    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

  18. [26]

    Carlos Lassance. 2023. Extending english ir methods to multi-lingual ir. arXiv preprint arXiv:2302.14723

    2023 arXiv

  19. [27]

    Dawn Lawrie, Eugene Yang, Douglas W Oard, and James Mayfield. 2023. Neural approaches to multilingual information retrieval. In European Conference on Information Retrieval, pages 521--536. Springer

    2023

  20. [28]

    Chankyu Lee, Rajarshi Roy, Mengyao Xu, Jonathan Raiman, Mohammad Shoeybi, Bryan Catanzaro, and Wei Ping. 2024. Nv-embed: Improved techniques for training llms as generalist embedding models. arXiv preprint arXiv:2405.17428

    2024 arXiv

  21. [29]

    Mingzhe Li, Xiexiong Lin, Xiuying Chen, Jinxiong Chang, Qishen Zhang, Feng Wang, Taifeng Wang, Zhongyi Liu, Wei Chu, Dongyan Zhao, et al. 2022 a . Keywords and instances: A hierarchical contrastive learning framework unifying hybrid granularities for text generation. arXiv pre...

    2022 arXiv

  22. [30]

    Shiyu Li, Yang Tang, Shizhe Chen, and Xi Chen. 2024 a . Conan-embedding: General text embedding with more and better negative samples. arXiv preprint arXiv:2408.15710

    2024 arXiv

  23. [31]

    Wing Yan Li, Julie Weeds, and David Weir. 2022 b . Museclir: a multiple senses and cross-lingual information retrieval dataset. In Proceedings of the 29th International Conference on Computational Linguistics, pages 1128--1135

    2022

  24. [32]

    Xiaopeng Li, Xiangyang Li, Hao Zhang, Zhaocheng Du, Pengyue Jia, Yichao Wang, Xiangyu Zhao, Huifeng Guo, and Ruiming Tang. 2024 b . Syneg: Llm-driven synthetic hard-negatives for dense retrieval. arXiv preprint arXiv:2412.17250

    2024 arXiv

  25. [33]

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

    2022

  26. [34]

    Sheng-Chieh Lin, Amin Ahmad, and Jimmy Lin. 2023 a . maggretriever: A simple yet effective approach to zero-shot multilingual dense retrieval. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 11688--11696

    2023

  27. [35]

    Sheng-Chieh Lin, Amin Ahmad, and Jimmy Lin. 2023 b . https://doi.org/10.18653/v1/2023.emnlp-main.715 m A ggretriever: A simple yet effective approach to zero-shot multilingual dense retrieval . In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Proc...

    2023 doi

  28. [36]

    Sean MacAvaney, Luca Soldaini, and Nazli Goharian. 2020. Teaching a new dog old tricks: Resurrecting multilingual retrieval using zero-shot learning. In Advances in Information Retrieval: 42nd European Conference on IR Research, ECIR 2020, Lisbon, Portugal, April 14--17, 2020,...

    2020

  29. [37]

    Adyasha Maharana and Mohit Bansal. 2022. On curriculum learning for commonsense reasoning. In Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 983--992

    2022

  30. [38]

    Luke Merrick, Danmei Xu, Gaurav Nuti, and Daniel Campos. 2024. Arctic-embed: Scalable, efficient, and accurate text embedding models. arXiv preprint arXiv:2405.05374

    2024 arXiv

  31. [39]

    Fengran Mo, Yifan Gao, Chuan Meng, Xin Liu, Zhuofeng Wu, Kelong Mao, Zhengyang Wang, Pei Chen, Zheng Li, Xian Li, et al. 2025 a . Uniconv: Unifying retrieval and response generation for large language models in conversations. In Proceedings of the 63rd Annual Meeting of the As...

    2025

  32. [40]

    Fengran Mo, Kelong Mao, Ziliang Zhao, Hongjin Qian, Haonan Chen, Yiruo Cheng, Xiaoxi Li, Yutao Zhu, Zhicheng Dou, and Jian-Yun Nie. 2025 b . A survey of conversational search. ACM Transactions on Information Systems (TOIS)

    2025

  33. [41]

    Fengran Mo, Kelong Mao, Yutao Zhu, Yihong Wu, Kaiyu Huang, and Jian-Yun Nie. 2023 a . Convgqr: Generative query reformulation for conversational search. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 4998--5012

    2023

  34. [42]

    Fengran Mo, Jian-Yun Nie, Kaiyu Huang, Kelong Mao, Yutao Zhu, Peng Li, and Yang Liu. 2023 b . Learning to relate to previous turns in conversational search. In Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pages 1722--1732

    2023

  35. [43]

    Fengran Mo, Chen Qu, Kelong Mao, Tianyu Zhu, Zhan Su, Kaiyu Huang, and Jian-Yun Nie. 2024. History-aware conversational dense retrieval. arXiv preprint arXiv:2401.16659

    2024 arXiv

  36. [44]

    Fengran Mo, Jinghan Zhang, Yuchen Hui, Jia Ao Sun, Zhichao Xu, Zhan Su, and Jian-Yun Nie. 2025 c . Convmix: A mixed-criteria data augmentation framework for conversational dense retrieval. arXiv preprint arXiv:2508.04001

    2025

  37. [45]

    Gabriel de Souza P Moreira, Radek Osmulski, Mengyao Xu, Ronay Ak, Benedikt Schifferer, and Even Oldridge. 2024. Nv-retriever: Improving text embedding models with effective hard-negative mining. arXiv preprint arXiv:2407.15831

    2024 arXiv

  38. [46]

    M3-embedding: Multi-linguality, multi-functionality, multi-granularity text embeddings through self-knowledge distillation

    Multi-Linguality Multi-Functionality Multi-Granularity. M3-embedding: Multi-linguality, multi-functionality, multi-granularity text embeddings through self-knowledge distillation

  39. [47]

    Tri Nguyen, Mir Rosenberg, Xia Song, Jianfeng Gao, Saurabh Tiwary, Rangan Majumder, and Li Deng. 2016. Ms marco: A human-generated machine reading comprehension dataset

    2016

  40. [48]

    Jian-Yun Nie. 2010. Cross-language information retrieval. Morgan & Claypool Publishers

    2010

  41. [49]

    Rodrigo Nogueira, Wei Yang, Kyunghyun Cho, and Jimmy Lin. 2019. Multi-stage document ranking with bert. arXiv preprint arXiv:1910.14424

    2019 arXiv

  42. [50]

    Douglas W Oard and Bonnie Jean Dorr. 1998. A survey of multilingual text retrieval. Citeseer

    1998

  43. [51]

    A Paszke. 2019. Pytorch: An imperative style, high-performance deep learning library. arXiv preprint arXiv:1912.01703

    2019 arXiv

  44. [52]

    Carol Peters, Martin Braschler, and Paul Clough. 2012. Multilingual information retrieval: From research to practice. Springer

    2012

  45. [53]

    Jonas Pfeiffer, Ivan Vuli \'c , Iryna Gurevych, and Sebastian Ruder. 2020. Mad-x: An adapter-based framework for multi-task cross-lingual transfer. arXiv preprint arXiv:2005.00052

    2020 arXiv

  46. [54]

    Zile Qiao, Wei Ye, Dingyao Yu, Tong Mo, Weiping Li, and Shikun Zhang. 2023. Improving knowledge graph completion with generative hard negative mining. In Findings of the Association for Computational Linguistics: ACL 2023, pages 5866--5878

    2023

  47. [55]

    Stephen Robertson, Hugo Zaragoza, et al. 2009. The probabilistic relevance framework: Bm25 and beyond. Foundations and Trends in Information Retrieval , 3(4):333--389

    2009

  48. [56]

    Xinqi Su, Dan Song, Wenhui Li, Tongwei Ren, and An-An Liu. 2025. Generating counterfactual negative samples for image-text matching. Information Processing & Management, 62(3):103990

    2025

  49. [57]

    Zeqi Tan, Yongliang Shen, Xiaoxia Cheng, Chang Zong, Wenqi Zhang, Jian Shao, Weiming Lu, and Yueting Zhuang. 2024. Learning global controller in latent space for parameter-efficient fine-tuning. In Proceedings of the 62nd Annual Meeting of the Association for Computational Lin...

    2024

  50. [58]

    Rohan Taori, Ishaan Gulrajani, Tianyi Zhang, Yann Dubois, Xuechen Li, Carlos Guestrin, Percy Liang, and Tatsunori B Hashimoto. 2023. Stanford alpaca: An instruction-following llama model

    2023

  51. [59]

    Nandan Thakur, Jianmo Ni, Gustavo Hern \'a ndez \'A brego, John Wieting, Jimmy Lin, and Daniel Cer. 2023. Leveraging llms for synthesizing training data across many languages in multilingual dense retrieval. arXiv preprint arXiv:2311.05800

    2023 arXiv

  52. [60]

    Nandan Thakur, Jianmo Ni, Gustavo Hernandez Abrego, John Wieting, Jimmy Lin, and Daniel Cer. 2024. Leveraging llms for synthesizing training data across many languages in multilingual dense retrieval. In Proceedings of the 2024 Conference of the North American Chapter of the A...

    2024

  53. [61]

    Jiayin Wang, Fengran Mo, Weizhi Ma, Peijie Sun, Min Zhang, and Jian-Yun Nie. 2024. A user-centric multi-intent benchmark for evaluating large language models. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 3588--3612

    2024

  54. [62]

    Liang Wang, Nan Yang, Xiaolong Huang, Binxing Jiao, Linjun Yang, Daxin Jiang, Rangan Majumder, and Furu Wei. 2022. Text embeddings by weakly-supervised contrastive pre-training. arXiv preprint arXiv:2212.03533

    2022 arXiv

  55. [63]

    Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Clement Delangue, Anthony Moi, Pierric Cistac, Tim Rault, R \'e mi Louf, Morgan Funtowicz, et al. 2019. Huggingface's transformers: State-of-the-art natural language processing. arXiv preprint arXiv:1910.03771

    2019 arXiv

  56. [64]

    Shitao Xiao, Zheng Liu, Peitian Zhang, and N Muennighof. 2023. C-pack: packaged resources to advance general chinese embedding. 2023. arXiv preprint arXiv:2309.07597

    2023 arXiv

  57. [65]

    Lee Xiong, Chenyan Xiong, Ye Li, Kwok-Fung Tang, Jialin Liu, Paul Bennett, Junaid Ahmed, and Arnold Overwijk. 2020. Approximate nearest neighbor negative contrastive learning for dense text retrieval. arXiv preprint arXiv:2007.00808

    2020 arXiv

  58. [66]

    Haotian Xu, Yuhua Wang, and Jiahui Fan. 2024. Self-knowledge distillation for knowledge graph embedding. In Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024), pages 14595--14605

    2024

  59. [67]

    Bruce Croft

    Jinxi Xu and W. Bruce Croft. 1996. https://api.semanticscholar.org/CorpusID:53249280 Query expansion using local and global document analysis . In Annual International ACM SIGIR Conference on Research and Development in Information Retrieval

    1996

  60. [68]

    Zhichao Xu, Fengran Mo, Zhiqi Huang, Crystina Zhang, Puxuan Yu, Bei Wang, Jimmy Lin, and Vivek Srikumar. 2025. A survey of model architectures in information retrieval. arXiv preprint arXiv:2502.14822

    2025

  61. [69]

    Eugene Yang, Dawn Lawrie, and James Mayfield. 2024. Distillation for multilingual information retrieval. In Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 2368--2373

    2024

  62. [70]

    Jingtao Zhan, Qingyao Ai, Yiqun Liu, Jiaxin Mao, Xiaohui Xie, Min Zhang, and Shaoping Ma. 2022. Disentangled modeling of domain and relevance for adaptable dense retrieval. arXiv preprint arXiv:2208.05753

    2022 arXiv

  63. [71]

    Xinyu Zhang, Kelechi Ogueji, Xueguang Ma, and Jimmy Lin. 2023 a . Toward best practices for training multilingual dense retrieval models. ACM Transactions on Information Systems, 42(2):1--33

    2023

  64. [72]

    Xinyu Zhang, Nandan Thakur, Odunayo Ogundepo, Ehsan Kamalloo, David Alfonso-Hermelo, Xiaoguang Li, Qun Liu, Mehdi Rezagholizadeh, and Jimmy Lin. 2023 b . Miracl: A multilingual retrieval dataset covering 18 diverse languages. Transactions of the Association for Computational L...

    2023

  65. [73]

    Yu Zhao and Qiaoyuan Shu. 2025. Debiased hybrid contrastive learning with hard negative mining for unsupervised person re-identification. Digital Signal Processing, 156:104826

    2025

  66. [74]

    Zhijun Zhou, Qing Xie, Yuhan Wang, Lin Li, Yongjian Liu, and Mengzi Tang. 2024. Debiased contrastive learning for graph collaborative filtering. In 2024 27th International Conference on Computer Supported Cooperative Work in Design (CSCWD), pages 48--54. IEEE

    2024

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.