Exploring $\ell_0$ Sparsification for Inference-free Sparse Retrievers

Xinjie Shen; Yang Yang; Zhichao Geng

REVIEW 4 major objections 6 minor 25 references

Exploring $\ell_0$ Sparsification for Inference-free Sparse Retrievers

T0 review · 4 major / 6 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 This paper claims that an ℓ0 mask, which stops applying the FLOPS sparsity penalty to documents that are already sparse enough, makes inference-free sparse retrieval match the effectiveness of Siamese sparse retrievers while reducing…

desk verdict A simple ℓ0 mask that helps stabilize SPLADE-doc training, but the SOTA claim rests on a baseline that looks misreported and the main table has obvious data-entry errors. read the letter →

arxiv 2504.14839 v1 pith:FF3JCV5Q submitted 2025-04-21 cs.IR cs.AI

Xinjie Shen , Zhichao Geng , Yang Yang This is my paper · ORCID

classification cs.IRcs.AI

keywords ℓ0masksparseretrievalinference-freeFLOPSregularizationSPLADEBEIRbenchmarkzero-shotevaluationdocumentsparsification

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 claims that the standard FLOPS regularization used to sparsify sparse retrievers is a poor fit for inference-free retrieval, where only documents are encoded by the network. The authors propose two small changes: an ℓ0 mask that removes documents whose number of activated tokens is already below a threshold from the FLOPS loss, and an ℓ0-approximation activation that replaces ReLU with log(1+max(0,x)) so sparsification pressure falls on small weights. On 13 BEIR datasets, the ℓ0 mask alone reaches 50.43 average NDCG@10, the best inference-free sparse retriever result reported here, within 0.13 points of SPLADE++-SelfDistil (50.56), while using slightly fewer FLOPS. The combined method trades a small effectiveness drop for a larger efficiency gain. The paper is trying to show that sparsification for asymmetric, inference-free retrieval should be guided by the ℓ0 count of document representations, not just their weight scale.

What carries the argument

The load-bearing object is the ℓ0 mask in Equation 5 and the ℓ0 approximation activation in Equation 6. The mask M(d_i) is a binary vector over the vocabulary that is all-zero unless the document's activated-token count exceeds a threshold t; zeroed documents are excluded from the FLOPS loss (Equation 4), so they are optimized purely for ranking. The activation σ(x)=log(1+max(0,x)) replaces ReLU in the document encoder, making large logits saturate slowly and shifting the regularizer's attention toward small weights, which is where ℓ0 sparsity is decided. Together they decouple sparsification from the scale of token weights in the asymmetric, document-only setting.

What would settle it

Re-run the ℓ0-mask model on BEIR with threshold t set to 50 and to 1000 (both within the range tested) with λ_d fixed at 0.04; if the average NDCG@10 falls below the unmodified IDF-SPLADE-doc-distill baseline (49.52), the claim that the mask reliably improves inference-free sparse retrieval would be contradicted. Alternatively, retrain on MS MARCO with λ_d=0.12 where the baseline collapses; if the ℓ0-mask model also collapses, the claimed stability would be false.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central discovery is that applying the FLOPS regularizer selectively—only to document representations whose ℓ0 norm exceeds a threshold—prevents the collapse of learned document sparsity under strong regularization, and that replacing the ReLU activation with log(1+max(0,x)) redirects the gradient pressure from large token weights toward small ones, better mimicking ℓ0 minimization. With the ℓ0 mask, the model attains 50.43 average NDCG@10 on 13 BEIR datasets and an average rank of 3.54 across all three retriever families, the best average rank of any method in the table. The paper reports this as state-of-the-art among inference-free sparse retrievers and comparable to leading Siamese sparse retrievers like SPLADE++-SelfDistil (50.56) and SPLADE-v3-Distil (49.99), while keeping the inference-free property: queries need only term matching, not neural inference.

Load-bearing premise

The claim that the method is state-of-the-art for inference-free retrieval depends on the particular values of the FLOPS penalty weight λ_d (0.04 or 0.035) and the mask threshold t (200), which were chosen on MS MARCO and then fixed for all 13 BEIR datasets; Section 5.3 shows the results are sensitive to both.

Editorial extensions

If this is right

Inference-free sparse retrievers can now match Siamese sparse retrievers in out-of-domain effectiveness without any query-side neural inference, strengthening the efficiency argument for zero-shot retrieval.
The ℓ0 mask creates a tunable efficiency-effectiveness frontier: varying the threshold t yields better trade-offs at comparable FLOPS than varying the regularizer weight λ_d, per Figure 3.
The combined method prevents the training collapse of the baseline at high λ_d (e.g., λ_d=0.12), allowing more aggressive sparsification in deployment.
The configuration with ℓ0 mask alone achieves 50.43 average NDCG@10 with 2.31 average FLOPS, a practical operating point for low-latency search systems.

Reading between the lines

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

One could extend the ℓ0 mask to per-document thresholds derived from a corpus-wide sparsity target, which the paper's fixed t leaves as a manual choice; the sensitivity shown in Figure 3 suggests this would improve robustness.
The same masked-FLOPS idea could apply to other asymmetric encoders—dense or late-interaction models that are quantized into sparse vectors—since the mask only needs a count of activated dimensions.
A direct efficiency test the paper does not report is end-to-end indexing time and query latency on the full MS MARCO corpus; FLOPS and Doc_Len reductions are proxy metrics that may not translate to wall-clock gains in a production engine.

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.

Desk Editor's Note

read the letter

Punchline: the paper describes two simple changes to SPLADE-doc-distill training — a threshold mask that keeps already-sparse documents out of the FLOPS regularizer, and a log-transformed activation that behaves a bit like ℓ0 pressure. The ablation suggests the mask alone is worth about 0.9 NDCG@10 over the authors' own baseline while shaving FLOPS. That is a modest but credible improvement. The problem is the packaging: the headline 'state-of-the-art among inference-free sparse retrievers' depends entirely on a SPLADE-v3-Doc baseline of 46.97 that the paper does not justify, and that better-published numbers do not support. On top of that, Table 1 has obvious data-entry errors.

What is genuinely new: I have not seen the thresholded FLOPS mask in the SPLADE literature, and the log activation is a reasonable heuristic. The idea of exempting already-sparse documents from the sparsification penalty is clean and could generalize. The authors ship code, and the ablation is internally consistent: +mask improves both NDCG and FLOPS; adding the activation trades a bit of relevance for efficiency.

Soft spots, in order of severity. First, the 3.46-point gap over SPLADE-v3-Doc is the whole basis for the external SOTA claim. The text does not say whether that row is copied from [16] or rerun, nor what settings were used. Published SPLADE-v3-Doc numbers I can recall are higher than 46.97; if an official checkpoint scores above 50.43 on the same 13 datasets, the claim collapses. Second, Table 1 is not reliable as printed: the FEVER and Climate-FEVER rows for the dense baselines appear swapped (Contriever on FEVER at 16.5 and on Climate-FEVER at 75.8 is backwards), and ColBERTv2 on SCIDOCS at 41.3 is far outside its known ~15.4. These errors undermine confidence in the rest of the table, even if the proposed method's own rows are plausible. Third, the exact λ_d and t used for the main result are not specified; the sensitivity curves in §5.3 fix t=200 when varying λ_d and λ_d=0.04 when varying t, but the text never connects those to Table 1. Fourth, no significance tests; with 13 datasets a paired test would be easy.

None of this kills the core method. The mask idea is simple and the ablation supports it. But the paper as submitted overstates its case and the data table is sloppy enough that a referee should not take the SOTA claim on faith.

Who this is for: researchers working on inference-free sparse retrieval and efficient SPLADE variants. A serious referee could ask for corrected baselines, a reproducible description of the main hyperparameters, and a fixed table; the method itself is worth a second look. I would send it to review rather than desk-reject, and I would condition acceptance on cleaning up the empirical claims.

Referee Report

4 major / 6 minor

Summary. The paper proposes two modifications to the inference-free sparse retriever IDF-SPLADE-doc-distill: an ℓ0 mask loss that excludes documents whose activation count is already below a threshold t from the FLOPS penalty, and an ℓ0 approximation activation that replaces ReLU with log(1+max(0,x)) to damp large weights. The authors fine-tune on MS MARCO and evaluate zero-shot on 13 BEIR datasets, reporting an average NDCG@10 of 50.43 for the ℓ0 mask variant, which they claim is state-of-the-art among inference-free sparse retrievers and comparable to leading Siamese sparse retrievers such as SPLADE++-SelfDistil. They also present ablations and efficiency analyses, including FLOPS and document-length measurements, and release code via a GitHub repository.

Significance. If the empirical claims hold, the paper makes a practical and simple contribution: two low-cost changes that improve inference-free sparse retrieval without adding parameters or changing the retrieval architecture. The inclusion of efficiency metrics (FLOPS and Doc_Len) alongside effectiveness is a strength, as is the use of the standard BEIR zero-shot evaluation protocol and the release of code. The comparison with SPLADE-v3-Doc is potentially important for the inference-free setting, but the reported baseline value is not currently trustworthy, and the missing hyperparameter specification prevents reproduction. The core methodology is internally coherent, and the ablation study gives some evidence that the ℓ0 mask prevents collapse at high λ_d, though the main claims require a corrected and fully specified evaluation.

major comments (4)

[Table 1 and Section 5.1.2] The average NDCG@10 of 46.97 reported for SPLADE-v3-Doc is well below the BEIR average published for that model in [16], and the paper does not state whether the row is copied from [16] or re-run, nor which query representation, IDF source, and OpenSearch settings were used. Since SPLADE-v3-Doc is the only strong published inference-free sparse baseline in the table, the 3.46-point gap to the proposed ℓ0 Mask is the entire basis for the headline 'state-of-the-art among inference-free sparse retrieval models' claim; this claim is unsupported until the row is reproduced with the official checkpoint under a clearly specified protocol.
[Section 5.1 and Section 5.3] The main results in Table 1 and Table 2 are not accompanied by the hyperparameters λ_d and t. Section 5.3.1 fixes t=200, Section 5.3.2 fixes λ_d=0.04, and Table 3 uses λ_d=0.035, but none of these is identified as the configuration for the headline averages. Because Figures 1–3 show that both λ_d and t strongly affect NDCG and FLOPS, the paper must state the exact values used for each reported model.
[Equation (4) and Section 4.1] Equation (4) multiplies each document's weights by M(d_i) but still divides the average by N, so masked documents enter as zero vectors; the sentence that they 'will not participate in the calculation of FLOPS loss' is therefore inaccurate. Zeroing the weights while keeping the denominator N dilutes the FLOPS penalty on unmasked documents, which is a different mechanism than excluding already-sparse documents. If exclusion is the intent, the normalization should be by the number of unmasked documents; otherwise the description should be revised.
[Section 5.2 and Table 1] No statistical testing or variance information is provided for any comparison. In particular, the 0.13-point gap between ℓ0 Mask (50.43) and SPLADE++-SelfDistil (50.56) is used to claim 'comparable' performance, but without per-dataset paired analysis or significance tests this claim is not quantitatively supported. Please report per-dataset scores or paired tests for the main comparisons.

minor comments (6)

[Tables 1 and 2 captions] The captions use 'NCDG@10' instead of 'NDCG@10'; please correct the typo.
[Table 1 and Section 5.1.3] The 'Aver.Rank' row is not defined in the metrics list; please state how the average rank is computed and which comparisons are included.
[Table 2 caption] The caption says 'comparable FLOPS' but the FLOPS values differ across rows (2.39, 2.31, 2.30, 2.13); clarify what 'comparable' means here.
[Section 4.2 and Equation (1)] The notation σ is overloaded: Equation (1) says σ is ReLU, then Section 4.2 redefines σ as log(1+max(0,x)). Please clarify that the proposed activation applies a second log on top of the existing log(1+ReLU(x)) transformation.
[Section 5.3.1 and Figure 2] The axis labels in Figure 2 appear as '10 3', '10 2', '10 1', '100'; these should be 10^3, 10^2, 10^1, and 10^0, respectively.
[Section 5.2] The sentence 'Our approach achieved the best average rank among all three types of retrievers' is ambiguous because in Table 1 the best average rank is 3.38 for ℓ0 mask + ℓ0 activation, not the 3.54 of ℓ0 Mask; please specify which variant is being referred to.

Circularity Check

0 steps flagged · score 1.0 of 10

No constructional circularity; central results are external BEIR measurements, with only minor reliance on the authors' prior base model.

full rationale

The paper's central claims are empirical: measured NDCG@10, FLOPS, and Doc_Len on 13 BEIR datasets. The proposed ℓ0 mask (Eq. 4–5) and ℓ0 approximation activation (Eq. 6) are concrete modifications to the training loss and activation; their effect is evaluated by external BEIR numbers, not derived from the definitions. No equation in the paper defines the reported improvement in terms of the method's own inputs: the mask threshold t and penalty weight λ_d are hyperparameters, and the reported NDCG values are outputs of a retrieval pipeline measured with the BEIR toolkit. The only self-citations are to the authors' prior work [12] as the base model (IDF-SPLADE-doc-distill) and teacher-score preparation; these provide the starting point and distillation targets, but the claimed gains over that baseline and over external systems (BM25, SPLADE-v3, SPLADE++, ColBERTv2, etc.) are independently measured. The suspiciously low SPLADE-v3-Doc row in Table 1 is a potential correctness/reproducibility issue, not a circularity: a mistaken baseline would undermine the SOTA claim empirically without making the derivation self-referential. Therefore no circular step is exhibited; score 1 reflects minor reliance on the authors' prior base model rather than any constructional circularity.

Assumptions & free parameters 3 free parameters · 3 assumptions · 0 invented entities

The central contribution rests on two hyperparameters (λ_d, t) and a heuristic activation; no new physical entities are proposed.

free parameters (3)

λ_d (FLOPS penalty weight) = 0.035 to 0.04 for main results; varied in experiments
Controls the trade-off between ranking loss and sparsification; the main BEIR results use this range without a stated selection rule.
t (ℓ0 mask threshold) = 200 for main results; search range 50 to 1000 in Figure 3
Documents with ℓ0 norm below t are exempted from FLOPS loss; results are sensitive to this value.
Number of log applications in ℓ0 activation = 1 for main results
Applying the log more than once degrades NDCG@10, so the paper chooses one application.

assumptions (3)

standard math FLOPS regularizer penalizes the squared average batch weight per vocabulary dimension (Equation 3, from [22])
Taken from prior work; the paper builds on it.
domain assumption Teacher scores from IDF-SPLADE-doc-distill ([12], same group) provide unbiased training targets
The distillation setup is inherited from the authors' prior work; its neutrality is not analyzed.
ad hoc to paper log(1+max(0,x)) behaves similarly to ℓ0 regularization and induces useful sparsity
Section 4.2 asserts the ℓ0 approximation property without a derivation; this is the main heuristic of the method.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Exploring $\ell_0$ Sparsification for Inference-free Sparse Retrievers." pith.science (2026). https://pith.science/paper/FF3JCV5Q

@misc{pith2026250414839,
  author       = {Pith},
  title        = {Pith review of: Exploring $\ell_0$ Sparsification for Inference-free Sparse Retrievers},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FF3JCV5Q}},
  note         = {Machine review of arXiv:2504.14839}
}

abstract

With increasing demands for efficiency, information retrieval has developed a branch of sparse retrieval, further advancing towards inference-free retrieval where the documents are encoded during indexing time and there is no model-inference for queries. Existing sparse retrieval models rely on FLOPS regularization for sparsification, while this mechanism was originally designed for Siamese encoders, it is considered to be suboptimal in inference-free scenarios which is asymmetric. Previous attempts to adapt FLOPS for inference-free scenarios have been limited to rule-based methods, leaving the potential of sparsification approaches for inference-free retrieval models largely unexplored. In this paper, we explore $\ell_0$ inspired sparsification manner for inference-free retrievers. Through comprehensive out-of-domain evaluation on the BEIR benchmark, our method achieves state-of-the-art performance among inference-free sparse retrieval models and is comparable to leading Siamese sparse retrieval models. Furthermore, we provide insights into the trade-off between retrieval effectiveness and computational efficiency, demonstrating practical value for real-world applications.

Figures

Figures reproduced from arXiv: 2504.14839 by the authors.

**Figure 3.** Search relevance vs sparsity, varying 𝑡. threshold allows longer documents to avoid penalization. When the threshold approaches zero, the ℓ0 Mask degenerates to near baseline performance, where FLOPS directly affects all documents. Compared to [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

**Figure 1.** Search relevance vs efficiency, varying 𝜆𝑑 . In [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗

**Figure 2.** 𝜆𝑑 vs encoded document sparsity. "x" denotes that baseline model collapses during training at 𝜆𝑑 = 0.12. 5.3.2 Varying Masked Threshold. Furthermore, we discuss the effects of varying the threshold 𝑡 in our methods, with 𝜆𝑑 fixed at 0.04. In [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

25 extracted references · 6 canonical work pages

[16]

Carlos Lassance, Hervé Déjean, Thibault Formal, and Stéphane Clinchant. 2024. SPLADE-v3: New baselines for SPLADE. https://doi.org/10.48550/ARXIV.2403. 06789

doi:10.48550/arxiv.2403 2024
[1]

Yang Bai, Xiaoguang Li, Gang Wang, Chaoliang Zhang, Lifeng Shang, Jun Xu, Zhaowei Wang, Fangshan Wang, and Qun Liu. 2020. SparTerm: Learning Term- based Sparse Representation for Fast Text Retrieval. https://doi.org/10.48550/ ARXIV.2010.00768

arXiv doi:10.48550/arxiv.2010.00768 2020
[2]

Shay Banon, Simon Willnauer, Jason Tedor, Martijn van Groningen, Ryan Ernst, Luca Cavanna, Nik Everett, David Pilato, Adrien Grand, Robert Muir, Boaz Leskes, Clinton Gormley, James Rodewig, Alexander Reelsen, Jay Modi, Lee Hinman, Tanguy Leroux, Colin Goodheart-Smithe, Lisa Cawley, Christoph Büscher, Armin Braun, Igor Motov, Nhat Nguyen, Jim Ferenczi, Yan...

work page 2025
[3]

Michael Busch, Krishna Gade, Brian Larson, Patrick Lok, Samuel Luckenbill, and Jimmy Lin. 2012. Earlybird: Real-Time Search at Twitter. In 2012 IEEE 28th International Conference on Data Engineering. 1360–1369. https://doi.org/10.1109/ ICDE.2012.149

work page 2012
[5]

Zhuyun Dai and Jamie Callan. 2020. Context-aware term weighting for first stage passage retrieval. In Proceedings of the 43rd International ACM SIGIR conference on research and development in Information Retrieval . 1533–1536

2020
[6]

M. A. H. Dempster, Juho Kanniainen, John Keane, and Erik Vynckier. 2018. High- Performance Computing in Finance: Problems, Methods, and Solutions (1st ed.). Chapman & Hall/CRC

work page 2018
[9]

Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant
[11]

Luyu Gao and Jamie Callan. 2022. Unsupervised Corpus Aware Language Model Pre-training for Dense Passage Retrieval. InProceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) . 2843– 2853

2022

Show all 25 references

[13]

Sebastian Hofstätter, Sheng-Chieh Lin, Jheng-Hong Yang, Jimmy Lin, and Allan Hanbury. 2021. Efficiently Teaching an Effective Dense Retriever with Balanced Topic Aware Sampling. InProceedings of the 44th International ACM SIGIR Confer- ence on Research and Development in Infor...

2021
[14]

Gautier Izacard, Mathilde Caron, Lucas Hosseini, Sebastian Riedel, Piotr Bo- janowski, Armand Joulin, and Edouard Grave. 2021. Unsupervised Dense Infor- mation Retrieval with Contrastive Learning. https://doi.org/10.48550/ARXIV. 2112.09118

2021 doi arXiv
[15]

Zhichao Geng, Dongyu Ru, and Yang Yang. 2024. Towards Competitive Search Relevance For Inference-Free Learned Sparse Retrievers. arXiv:2411.04403 [cs.IR] https://arxiv.org/abs/2411.04403

2024 arXiv
[17]

Sean MacAvaney, Franco Maria Nardini, Raffaele Perego, Nicola Tonellotto, Nazli Goharian, and Ophir Frieder. 2020. Expansion via Prediction of Importance with Contextualization. In Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Inform...

2020
[18]

Carlos Lassance and Stéphane Clinchant. 2022. An Efficiency Study for SPLADE Models. In Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval (Madrid, Spain) (SIGIR ’22). Association for Computing Machinery, New York, N...

2022
[19]

Antonio Mallia, Omar Khattab, Torsten Suel, and Nicola Tonellotto. 2021. Learn- ing Passage Impacts for Inverted Indexes. In Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval (Virtual Event, Canada) (SIGIR ’21). Ass...

2021
[20]

Tri Nguyen, Mir Rosenberg, Xia Song, Jianfeng Gao, Saurabh Tiwary, Rangan Majumder, and Li Deng. 2016. MS MARCO: A Human Gen- erated MAchine Reading COmprehension Dataset. (November 2016). https://www.microsoft.com/en-us/research/publication/ms-marco-human- generated-machine-r...

2016
[21]

Sean MacAvaney, Franco Maria Nardini, Raffaele Perego, Nicola Tonellotto, Nazli Goharian, and Ophir Frieder. 2020. Expansion via prediction of importance with contextualization. In Proceedings of the 43rd International ACM SIGIR conference on research and development in Inform...

2020
[22]

Yen, Ning Xu, Pradeep Ravikumar, and Barnabás Póczos

Biswajit Paria, Chih-Kuan Yeh, Ian E.H. Yen, Ning Xu, Pradeep Ravikumar, and Barnabás Póczos. 2020. Minimizing FLOPs to Learn Efficient Sparse Rep- resentations. In International Conference on Learning Representations . https: //openreview.net/forum?id=SygpC6Ntvr

2020
[23]

Stephen Robertson and Hugo Zaragoza. 2009. The Probabilistic Relevance Frame- work: BM25 and Beyond. Found. Trends Inf. Retr. 3, 4 (April 2009), 333–389. https://doi.org/10.1561/1500000019

2009 doi
[24]

Rodrigo Nogueira, Wei Yang, Jimmy Lin, and Kyunghyun Cho. 2019. Document expansion by query prediction. arXiv preprint arXiv:1904.08375 (2019)

2019 arXiv
[25]

Keshav Santhanam, Omar Khattab, Jon Saad-Falcon, Christopher Potts, and Matei Zaharia. 2022. ColBERTv2: Effective and Efficient Retrieval via Lightweight Late Interaction. In Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational ...

2022
[26]

Nandan Thakur, Nils Reimers, Andreas Rücklé, Abhishek Srivastava, and Iryna Gurevych. 2021. BEIR: A Heterogeneous Benchmark for Zero-shot Evaluation of Information Retrieval Models. In Thirty-fifth Conference on Neural Information Processing Systems Datasets and Benchmarks Tra...

2021
[27]

S. E. Robertson and K. Sparck Jones. 1976. Relevance weighting of search terms. Journal of the American Society for Information Science 27, 3 (1976), 129–146. https://doi.org/10.1002/asi.4630270302 arXiv:https://asistdl.onlinelibrary.wiley.com/doi/pdf/10.1002/asi.4630270302

1976 doi
[31]

Tiancheng Zhao, Xiaopeng Lu, and Kyusong Lee. 2021. SPARTA: Efficient Open- Domain Question Answering via Sparse Transformer Matching Retrieval. In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language...

2021
[2021]

https://doi.org/10.48550/ARXIV.2109.10086

SPLADE v2: Sparse Lexical and Expansion Model for Information Retrieval. https://doi.org/10.48550/ARXIV.2109.10086

doi arXiv
[3734]

https://doi.org/10.18653/v1/2022.naacl-main.272

2022 doi

Pith tools

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

[1] [16]

Carlos Lassance, Hervé Déjean, Thibault Formal, and Stéphane Clinchant. 2024. SPLADE-v3: New baselines for SPLADE. https://doi.org/10.48550/ARXIV.2403. 06789

doi:10.48550/arxiv.2403 2024

[2] [1]

Yang Bai, Xiaoguang Li, Gang Wang, Chaoliang Zhang, Lifeng Shang, Jun Xu, Zhaowei Wang, Fangshan Wang, and Qun Liu. 2020. SparTerm: Learning Term- based Sparse Representation for Fast Text Retrieval. https://doi.org/10.48550/ ARXIV.2010.00768

arXiv doi:10.48550/arxiv.2010.00768 2020

[3] [2]

Shay Banon, Simon Willnauer, Jason Tedor, Martijn van Groningen, Ryan Ernst, Luca Cavanna, Nik Everett, David Pilato, Adrien Grand, Robert Muir, Boaz Leskes, Clinton Gormley, James Rodewig, Alexander Reelsen, Jay Modi, Lee Hinman, Tanguy Leroux, Colin Goodheart-Smithe, Lisa Cawley, Christoph Büscher, Armin Braun, Igor Motov, Nhat Nguyen, Jim Ferenczi, Yan...

work page 2025

[4] [3]

Michael Busch, Krishna Gade, Brian Larson, Patrick Lok, Samuel Luckenbill, and Jimmy Lin. 2012. Earlybird: Real-Time Search at Twitter. In 2012 IEEE 28th International Conference on Data Engineering. 1360–1369. https://doi.org/10.1109/ ICDE.2012.149

work page 2012

[5] [5]

Zhuyun Dai and Jamie Callan. 2020. Context-aware term weighting for first stage passage retrieval. In Proceedings of the 43rd International ACM SIGIR conference on research and development in Information Retrieval . 1533–1536

2020

[6] [6]

M. A. H. Dempster, Juho Kanniainen, John Keane, and Erik Vynckier. 2018. High- Performance Computing in Finance: Problems, Methods, and Solutions (1st ed.). Chapman & Hall/CRC

work page 2018

[7] [9]

Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant

[8] [11]

Luyu Gao and Jamie Callan. 2022. Unsupervised Corpus Aware Language Model Pre-training for Dense Passage Retrieval. InProceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) . 2843– 2853

2022