Multilingual Information Retrieval with a Monolingual Knowledge Base

Aman Gupta; Anurag Beniwal; Yingying Zhuang

REVIEW 4 major objections 5 minor 1 cited by

Multilingual Information Retrieval with a Monolingual Knowledge Base

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read Weighted sampling for contrastive fine-tuning lets a multilingual embedding model retrieve English knowledge-base entries from Hinglish queries, with reported gains up to 31.03% in MRR and 33.98% in Recall@3.

desk verdict Decent engineering recipe with a misleading headline: the 31% MRR gain is an absolute gap against a no-labeled-data ablation, not a relative gain over a standard sampling baseline, yet the hybrid approach does show real gains over hardest-negative mining. read the letter →

arxiv 2506.02527 v1 pith:TCIWHZU3 submitted 2025-06-03 cs.CL cs.AIcs.IR

Yingying Zhuang , Aman Gupta , Anurag Beniwal This is my paper

classification cs.CLcs.AIcs.IR

keywords multilingualinformationretrievalcontrastivelearningweightedsamplinghardnegativeminingcode-switchingHinglishembeddingmodelfine-tuningmonolingualknowledgebase

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper tries to establish that a monolingual knowledge base can power multilingual and code-switching retrieval if the embedding model is fine-tuned with the right contrastive pairs. The authors keep the knowledge base in one high-resource language (English) and use only unlabeled queries in the target language (here, Hinglish), avoiding the expensive construction of language-specific knowledge bases. Their recipe translates English queries, pairs each translation with a same-intent index query as a positive, and mixes random negatives with hard negatives sampled by label similarity, plus synthetically generated queries. On their Hinglish-to-English retrieval test, this recipe reports gains up to 31.03% in MRR and 33.98% in Recall@3 over standard sampling, and the authors argue the approach is language-agnostic. This matters because knowledge base construction is labor-intensive, so sharing one high-quality English knowledge base across languages would remove a major bottleneck for multilingual dialogue systems.

What carries the argument

The carrying mechanism is Algorithm 1's rule for building contrastive pairs from a monolingual knowledge base. For each English query, an LLM produces a target-language translation; the positive pair is that translation matched with a random same-label index query. Negatives are drawn by computing a label-similarity score $s_{ij}$ between the query's label $l_i$ and each index query's label $l_j$, then sampling $k=2$ hard negatives with probability weighted by $s_{ij}$, plus one random negative from a different label, maintaining a 1:3 positive-to-negative ratio. A final step augments the data with synthetic target-language paraphrases. This weighted distribution is what the paper credits for preserving global embedding structure while sharpening local distinctions.

What would settle it

Sample the translated and synthetic training pairs, have annotators or a strong classifier check whether each pair's intent label still matches the original English label, and retrain on the filtered subset; if the weighted-sampling gains shrink substantially, the reported improvements are not attributable to the sampling strategy alone.

Watch

Extended reading notes

Core claim

On its own terms, the paper discovers that a mixed-negative recipe works: one random negative combined with hard negatives weighted by label similarity trains a multilingual embedding model to place Hinglish queries near their English intent-equivalents better than pure random, pure hard, or hardest-negative sampling. The best configuration reaches Top-1 0.5450, Top-3 0.7410, Top-10 0.8842, and MRR 0.6653 on the reported test set, with Recall@1 around 0.54 on the selected multilingual backbone. It also finds that labeled translated data is indispensable: training on synthetic data alone yields MRR 0.3550, while synthetic augmentation on top of labeled data protects performance on native low-resource queries.

Load-bearing premise

The load-bearing premise is that every LLM translation and every synthetic query preserves the intent label of the original English query, and the paper reports no quality filtering, human evaluation, or noise analysis for that generated data.

Editorial extensions

If this is right

A monolingual English knowledge base can serve retrieval for Hinglish queries without building a separate Hinglish knowledge base, as long as unlabeled Hinglish queries are available for synthetic augmentation.
Mixing one random negative with label-similarity-weighted hard negatives outperforms pure random, pure hard, and hardest-negative sampling on the reported benchmark.
High-quality labeled translated pairs are the main driver of performance; synthetic data alone drops MRR to 0.3550, while adding synthetic data on top of labeled data yields the best Top-3, Top-10, and MRR values in the paper's comparison.
Because the pipeline only uses unlabeled target-language queries and a labeled English knowledge base, it is language-agnostic and should apply to other low-resource languages and code-switching varieties.

Reading between the lines

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

Replacing label-similarity weights $s_{ij}$ with embedding-similarity weights would extend the pipeline to knowledge bases without labels; the paper does not test that variant.
Since the paper observes that Labeled Data Only matches the hybrid on the current test set, a distribution-shifted Hinglish test set would be the sharper test of whether the synthetic augmentation branch contributes on native low-resource queries.
Verifying label preservation on LLM-generated pairs and filtering mismatches would reveal whether the weighted-sampling gains can be pushed higher or whether some gains are an artifact of label noise being amplified for high-similarity labels.

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.

Desk Editor's Note

read the letter

Quick take: this is a reasonable, honest engineering paper with a headline claim that does not survive contact with its own Table 1. The 31.03% MRR gain in the abstract is not a relative improvement over a standard sampling baseline; it is the absolute difference between Algorithm 1 and the Synthetic Data Only ablation (0.6653 vs 0.3550). That ablation intentionally removes all labeled data, so it is not a fair baseline for the sampling strategy. Against the standard labeled-data baselines, the gains are much smaller: about 2% over Random Negative Mining, 11% over Hard Negative Mining, and 19% over Hardest Negative Mining. The stress-test note gets this exactly right, and I agree with it.

What is genuinely new: the specific combination of label-similarity-weighted hard negatives, one random negative, and LLM-generated synthetic queries for code-switched retrieval. Each ingredient is standard, but I do not know of a prior paper that puts them together for Hinglish. The ablation design is thoughtful: separating negative-sampling variations from data-source variations lets you see that the hybrid mining matters most when compared against hardest-negative-only, and that synthetic data helps a bit on top of labeled data. Labeled Data Only nearly matches the full method (MRR 0.6639 vs 0.6653), which the authors concede; they attribute it to test-set composition. That is plausible but not demonstrated.

Soft spots, in proportion. The headline overstatement is the big one, and it is load-bearing: the abstract and Section 3.5 both claim up to 31.03% MRR improvement over standard methods, which Table 1 does not support. Second, there are no error bars, no significance tests, and the evaluation is on a private, unshared dataset with no code. That matters for a paper whose central message is that the proposed method improves retrieval. Third, the language-agnostic claim rests on one language pair (Hinglish→English), so it is not yet justified. Fourth, the LLM-generated training data has no quality filtering or noise analysis; if translations or synthetic queries shift intent, the contrastive signal degrades. That is a real concern, though minor given the clean ablations.

Who this is for: practitioners at companies that want to extend an existing English-only retrieval or knowledge-base stack to code-switched or low-resource queries. They will get a concrete, reproducible recipe (assuming they build their own data following Algorithm 1) and a calibrated sense of what each ingredient buys.

Recommendation: send it to peer review, but expect major revision. The authors need to fix the headline numbers to the actual relative gains, add significance testing or error bars, and ideally test at least one more target language. The core method is sound and the ablation is useful; the overclaiming is fixable. I would not desk-reject this. For my own work, I would cite it as an example of hybrid negative sampling for multilingual retrieval, but note the overstated abstract.

Referee Report

4 major / 5 minor

Summary. The paper proposes a contrastive fine-tuning pipeline (Algorithm 1) for retrieving entries from a monolingual English knowledge base using multilingual or code-switched queries. For each English KB query, an LLM translates it into a target language, a positive pair is formed with a same-label English query, and negatives are generated by weighted sampling based on label similarity plus one random different-label negative; synthetic Hinglish queries generated by an LLM are added as augmentation. The authors fine-tune multilingual-e5-base and evaluate on an internal Hinglish retrieval task. The abstract claims gains up to 31.03% in MRR and up to 33.98% in Recall@3 over standard sampling strategies. In Table 1, Algorithm 1 achieves MRR 0.6653 and Recall@3 0.7410, compared with Random Negative Mining (0.6520, 0.7271), Hard Negative Mining (0.5982, 0.6829), Hardest Negative Mining (0.5610, 0.6678), Labeled Data Only (0.6639, 0.7356), and Synthetic Data Only (0.3550, 0.4012).

Significance. If the claimed gains were valid, this would be a practically useful and language-agnostic recipe for cross-lingual and code-switched retrieval without constructing a target-language knowledge base. The paper has a clear algorithmic description, a reasonable choice of base model, and a systematic ablation structure. However, the headline result is not supported by the paper's own Table 1: the 31.03% and 33.98% figures are absolute differences against the Synthetic Data Only ablation, not relative improvements over any standard negative-sampling baseline. The comparison to Labeled Data Only shows a difference of 0.0014 MRR, which is negligible without error bars or significance tests. Because the central claim is misrepresented and the most natural baselines show only marginal gains, the contribution as stated is not established.

major comments (4)

[Abstract and §3.5, Table 1] The headline claim that the weighted sampling strategy produces gains 'by up to 31.03% in MRR and up to 33.98% in Recall@3' is not supported by Table 1. 0.6653−0.3550 = 0.3103 and 0.7410−0.4012 = 0.3398 are the absolute differences between Algorithm 1 and the Synthetic Data Only ablation, not relative gains over any standard negative-sampling strategy. Against Random Negative Mining, the relative MRR gain is only (0.6653−0.6520)/0.6520 ≈ 2.0%, and against Hardest Negative Mining it is (0.6653−0.5610)/0.5610 ≈ 18.6%, not 31%. The abstract and the discussion in §3.5 must be corrected to report the actual comparison baselines and to distinguish absolute percentage-point differences from relative improvements.
[Table 1 and §3.5] The comparison to Labeled Data Only undercuts the central claim. Algorithm 1 yields MRR 0.6653 and Recall@3 0.7410, while Labeled Data Only yields 0.6639 and 0.7356, respectively; the differences are 0.0014 MRR and 0.0054 Recall@3. The paper reports no error bars, no multiple seeds, and no significance tests, and §3.5 concedes that Labeled Data Only 'performs comparably.' Without statistical evidence, the assertion that the hybrid approach 'provides better protection against performance degradation' is not substantiated. At minimum, the authors should report variance across repeated fine-tuning runs and a significance test for the differences that matter.
[Algorithm 1, Steps 3–5] The training-data validity assumption is unverified. Step 3 assumes that LLM translations preserve the original English query's intent label, and Step 5 assumes that LLM-generated synthetic Hinglish queries and their 'semantically different' English counterparts are correctly labeled positives and negatives. No quality filtering, human evaluation, or noise analysis is reported for either the translated or the synthetic data. If translation or generation changes the intent, the contrastive signal is corrupted and the measured gains cannot be attributed to the weighted sampling strategy. Additionally, Step 4's hard-negative sampling does not explicitly exclude queries with the same label as the positive; only the Hardest Negative Mining description states 'excluding exact matches.' If same-label queries can be sampled as hard negatives, they are false negatives, which would bias the comparison.
[§3.2 and Table 1] The ablation definitions are ambiguous about whether Step 5 (synthetic data augmentation) is included in the Negative Sampling Variations. Random Negative Mining, Hard Negative Mining, and Hardest Negative Mining are described only in terms of Step 4; it is unclear whether they also use synthetic data from Step 5. If they do not, then the comparison between Algorithm 1 and these baselines conflates the sampling strategy with the synthetic-data augmentation. If they do, the paper should say so explicitly. This distinction is load-bearing for the claim that mixed hard/random negative sampling is superior.

minor comments (5)

[§3.1] The word 'denonymized' appears to be a typo; the intended meaning is likely 'de-identified' or 'anonymized.' Please clarify the anonymization procedure.
[§3.5] The sentence 'Pure hard negative mining (0.4613) or hardest negative mining (0.4012) approaches underperform' cites values that are Top-1 accuracies from Table 1, not MRR or Recall@3 values as the surrounding argument about global embedding structure would require. Please use consistent metric labels.
[§3.3] There are several typos, including 'optimzed' and 'constrastive'; a careful proofread is needed.
[Algorithm 1 Step 4] The positive-to-negative ratio 1:3 is stated for k=2, but the text also describes k=0 and k=3 variations; please specify how the number of random negatives changes in those cases to maintain the stated ratio.
[§3.2] No baseline without fine-tuning is reported. Adding the off-the-shelf multilingual-e5-base result would help the reader assess whether the proposed pipeline improves over the starting model at all.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the paper's claims are empirical pipeline comparisons, and its self-citations and label-weighted sampling are not defined in terms of the reported results.

full rationale

This is an empirical training-pipeline paper, not a derivation from first principles. The weighted sampling strategy uses similarity scores computed directly from knowledge-base labels to choose negative pairs; no parameter is fitted to the evaluation metrics, and no reported result is the solution of an equation constructed from its own inputs. The self-citations (references [1,2,3]) appear only as background on dialogue systems and are not load-bearing for the central claim. The abstract's 'up to 31.03% in MRR and up to 33.98% in Recall@3' do correspond arithmetically to the absolute metric differences between Algorithm 1 and the Synthetic Data Only ablation (0.6653 - 0.3550 = 0.3103; 0.7410 - 0.4012 = 0.3398), rather than to relative gains over the standard Random/Hard negative baselines; this is a baseline-selection and claim-support problem, not a circularity of construction. The paper itself acknowledges that Labeled Data Only performs comparably (MRR 0.6639 vs. 0.6653), which weakens the strength of the contribution but does not make the comparison circular. No fitted input is renamed as a prediction, and no result is equivalent to its inputs by definition.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The method rests on domain assumptions about label quality, LLM translation fidelity, and treating code-switching as one language. No external benchmark or independent data source supports the general claim beyond the internal Hinglish test set.

free parameters (2)

k (number of hard negative samples per translated query) = 2
Algorithm 1 Step 4 sets k=2 to produce a 1:3 positive-to-negative ratio; no sensitivity analysis is reported.
Synthetic pair composition = 1 positive + 3 negatives per query
Algorithm 1 Step 5 fixes the synthetic augmentation ratio; no ablation varies this composition.

assumptions (4)

domain assumption A query and any other query sharing the same label are semantically similar enough to serve as positive pairs.
Algorithm 1 Step 3 constructs positives by matching labels; label noise or coarse labels would corrupt pairs.
domain assumption LLM translation from English to the target language and LLM-generated synthetic queries preserve the intent label.
Steps 3 and 5 rely on Claude Sonnet 3.5 translations and generation; no quality filtering or human evaluation is described.
domain assumption Code-switched Hinglish can be treated as a single target language and aligned with English by a multilingual embedding model.
Section 3 states this assumption and tests only Hinglish, not multiple languages.
domain assumption Open-source multilingual embedding models provide a suitable starting representation for the target code-switched language.
Section 3.2 selects models such as multilingual-e5-base; if the pretrained model has poor coverage of Hinglish, fine-tuning may be insufficient.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Multilingual Information Retrieval with a Monolingual Knowledge Base." pith.science (2026). https://pith.science/paper/TCIWHZU3

@misc{pith2026250602527,
  author       = {Pith},
  title        = {Pith review of: Multilingual Information Retrieval with a Monolingual Knowledge Base},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TCIWHZU3}},
  note         = {Machine review of arXiv:2506.02527}
}

read the original abstract

Multilingual information retrieval has emerged as powerful tools for expanding knowledge sharing across languages. On the other hand, resources on high quality knowledge base are often scarce and in limited languages, therefore an effective embedding model to transform sentences from different languages into a feature vector space same as the knowledge base language becomes the key ingredient for cross language knowledge sharing, especially to transfer knowledge available in high-resource languages to low-resource ones. In this paper we propose a novel strategy to fine-tune multilingual embedding models with weighted sampling for contrastive learning, enabling multilingual information retrieval with a monolingual knowledge base. We demonstrate that the weighted sampling strategy produces performance gains compared to standard ones by up to 31.03\% in MRR and up to 33.98\% in Recall@3. Additionally, our proposed methodology is language agnostic and applicable for both multilingual and code switching use cases.

Figures

Figures reproduced from arXiv: 2506.02527 by the authors.

**Figure 1.** Illustration of our contrastive training data generation algorithm. For each translated query (left), we create a [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗

**Figure 2.** Performance of Recall@1 at different points during contrastive fine-tuning. [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗

**Figure 3.** Performance metrics at different points during contrastive fine-tuning for various multilingual embedding [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗

Discussion (0). Sign in 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

How and Where to Translate? The Impact of Translation Strategies in Cross-lingual LLM Prompting
cs.CL 2025-07 conditional novelty 5.0 of 10

For multilingual RAG intent classification, the best translation strategy depends on the model and language; translating instructions into the user's language helps some models, while making the model answer in low-re...

Reference graph

Works this paper leans on

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

[1]

Self-supervised pre-training and semi-supervised learning for extractive dialog summarization

Yingying Zhuang, Jiecheng Song, Narayanan Sadagopan, and Anurag Beniwal. Self-supervised pre-training and semi-supervised learning for extractive dialog summarization. InCompanion Proceedings of the ACM Web Conference 2023, WWW ’23 Companion, page 1069–1076, New York, NY , USA, 2023. Association for Computing Machinery

work page 2023
[2]

Weakly supervised extractive summarization with attention

Yingying Zhuang, Yichao Lu, and Simi Wang. Weakly supervised extractive summarization with attention. In Haizhou Li, Gina-Anne Levow, Zhou Yu, Chitralekha Gupta, Berrak Sisman, Siqi Cai, David Vandyke, Nina Dethlefs, Yan Wu, and Junyi Jessy Li, editors,Proceedings of the 22nd Annual Meeting of the Special Interest Group on Discourse and Dialogue, pages 52...

work page 2021
[3]

Reic: Rag-enhanced intent classification at scale, 2025

Ziji Zhang, Michael Yang, Zhiyu Chen, Yingying Zhuang, Shu-Ting Pi, Qun Liu, Rajashekar Maragoud, Vy Nguyen, and Anurag Beniwal. Reic: Rag-enhanced intent classification at scale, 2025

work page 2025
[4]

https://www.anthropic.com/news/claude-3-family

Introducing the next generation of claude. https://www.anthropic.com/news/claude-3-family . Ac- cessed: 2025-04-20

work page 2025
[5]

Command a: An enterprise-ready large language model, 2025

Team Cohere and et al. Command a: An enterprise-ready large language model, 2025

work page 2025
[6]

How much is a triple? estimating the cost of knowledge graph creation

Heiko Paulheim. How much is a triple? estimating the cost of knowledge graph creation. 2018

work page 2018
[7]

Seza Do˘gruöz, Sunayana Sitaram, Barbara E

A. Seza Do˘gruöz, Sunayana Sitaram, Barbara E. Bullock, and Almeida Jacqueline Toribio. A survey of code- switching: Linguistic and social perspectives for language technologies. In Chengqing Zong, Fei Xia, Wenjie Li, and Roberto Navigli, editors,Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th Internati...

work page 2021
[8]

Are multilingual models effective in code-switching? In Thamar Solorio, Shuguang Chen, Alan W

Genta Indra Winata, Samuel Cahyawijaya, Zihan Liu, Zhaojiang Lin, Andrea Madotto, and Pascale Fung. Are multilingual models effective in code-switching? In Thamar Solorio, Shuguang Chen, Alan W. Black, Mona Diab, Sunayana Sitaram, Victor Soto, Emre Yilmaz, and Anirudh Srinivasan, editors,Proceedings of the Fifth Workshop on Computational Approaches to Lin...

work page 2021

Show all 26 references

[9]

Speak foreign languages with your own voice: Cross-lingual neural codec language modeling, 2023

Ziqiang Zhang, Long Zhou, Chengyi Wang, Sanyuan Chen, Yu Wu, Shujie Liu, Zhuo Chen, Yanqing Liu, Huaming Wang, Jinyu Li, Lei He, Sheng Zhao, and Furu Wei. Speak foreign languages with your own voice: Cross-lingual neural codec language modeling, 2023

2023
[10]

Zero resource code-switched speech benchmark using speech utterance pairs for multiple spoken languages, 2024

Kuan-Po Huang, Chih-Kai Yang, Yu-Kuan Fu, Ewan Dunbar, and Hung yi Lee. Zero resource code-switched speech benchmark using speech utterance pairs for multiple spoken languages, 2024

2024
[11]

Shaoxiong Ji, Shirui Pan, Erik Cambria, Pekka Marttinen, and Philip S. Yu. A survey on knowledge graphs: Representation, acquisition, and applications.IEEE Transactions on Neural Networks and Learning Systems, 33(2):494–514, February 2022. 6

2022
[12]

Prix-lm: Pretraining for multilingual knowledge base construction.arXiv preprint arXiv:2110.08443, 2021

Wenxuan Zhou, Fangyu Liu, Ivan Vuli´c, Nigel Collier, and Muhao Chen. Prix-lm: Pretraining for multilingual knowledge base construction.arXiv preprint arXiv:2110.08443, 2021

2021 arXiv
[13]

Text embeddings by weakly-supervised contrastive pre-training.ArXiv, abs/2212.03533, 2022

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

2022 arXiv
[14]

Towards general text embeddings with multi-stage contrastive learning.ArXiv, abs/2308.03281, 2023

Zehan Li, Xin Zhang, Yanzhao Zhang, Dingkun Long, Pengjun Xie, and Meishan Zhang. Towards general text embeddings with multi-stage contrastive learning.ArXiv, abs/2308.03281, 2023

2023 arXiv
[15]

Morris, Brandon Duderstadt, and Andriy Mulyar

Zach Nussbaum, John X. Morris, Brandon Duderstadt, and Andriy Mulyar. Nomic embed: Training a reproducible long context text embedder, 2025

2025
[16]

Arctic-embed: Scalable, efficient, and accurate text embedding models, 2024

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

2024
[17]

Multilingual e5 text embeddings: A technical report, 2024

Liang Wang, Nan Yang, Xiaolong Huang, Linjun Yang, Rangan Majumder, and Furu Wei. Multilingual e5 text embeddings: A technical report, 2024

2024
[18]

Bge m3-embedding: Multi-lingual, multi-functionality, multi-granularity text embeddings through self-knowledge distillation, 2024

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

2024
[19]

mgte: Generalized long-context text representation and reranking models for multilingual text retrieval, 2024

Xin Zhang, Yanzhao Zhang, Dingkun Long, Wen Xie, Ziqi Dai, Jialong Tang, Huan Lin, Baosong Yang, Pengjun Xie, Fei Huang, Meishan Zhang, Wenjie Li, and Min Zhang. mgte: Generalized long-context text representation and reranking models for multilingual text retrieval, 2024

2024
[20]

Introduction to the special issue on dialogue and dialogue systems

Liesbeth Degand and Philippe Muller. Introduction to the special issue on dialogue and dialogue systems. Traitement Automatique des Langues, 61(3):7–15, 2020

2020
[21]

Jina embeddings: A novel set of high-performance sentence embedding models, 2023

Michael Günther, Louis Milliken, Jonathan Geuter, Georgios Mastrapas, Bo Wang, and Han Xiao. Jina embeddings: A novel set of high-performance sentence embedding models, 2023

2023
[22]

Improving text embeddings with large language models

Liang Wang, Nan Yang, Xiaolong Huang, Linjun Yang, Rangan Majumder, and Furu Wei. Improving text embeddings with large language models. In Lun-Wei Ku, Andre Martins, and Vivek Srikumar, editors,Proceedings of the 62nd Annual Meeting of the Association for Computational Linguis...

2024
[23]

Zhao, Ji Ma, Yi Luan, Jianmo Ni, Jing Lu, Anton Bakalov, Kelvin Guu, Keith B

Zhuyun Dai, Vincent Y . Zhao, Ji Ma, Yi Luan, Jianmo Ni, Jing Lu, Anton Bakalov, Kelvin Guu, Keith B. Hall, and Ming-Wei Chang. Promptagator: Few-shot dense retrieval from 8 examples, 2022

2022
[24]

Rocketqa: An optimized training approach to dense passage retrieval for open-domain question answering, 2021

Yingqi Qu, Yuchen Ding, Jing Liu, Kai Liu, Ruiyang Ren, Wayne Xin Zhao, Daxiang Dong, Hua Wu, and Haifeng Wang. Rocketqa: An optimized training approach to dense passage retrieval for open-domain question answering, 2021

2021
[25]

Moreira, Radek Osmulski, Mengyao Xu, Ronay Ak, Benedikt Schifferer, and Even Oldridge

Gabriel de Souza P. Moreira, Radek Osmulski, Mengyao Xu, Ronay Ak, Benedikt Schifferer, and Even Oldridge. Nv-retriever: Improving text embedding models with effective hard-negative mining, 2025

2025
[26]

Representation learning with contrastive predictive coding

Aäron van den Oord, Yazhe Li, and Oriol Vinyals. Representation learning with contrastive predictive coding. ArXiv, abs/1807.03748, 2018. 7 A Plots for Model Comparison The following plots represent some other comparisons between the models we compared between. (a) Performance...

2018 arXiv

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [1]

Self-supervised pre-training and semi-supervised learning for extractive dialog summarization

Yingying Zhuang, Jiecheng Song, Narayanan Sadagopan, and Anurag Beniwal. Self-supervised pre-training and semi-supervised learning for extractive dialog summarization. InCompanion Proceedings of the ACM Web Conference 2023, WWW ’23 Companion, page 1069–1076, New York, NY , USA, 2023. Association for Computing Machinery

work page 2023

[2] [2]

Weakly supervised extractive summarization with attention

Yingying Zhuang, Yichao Lu, and Simi Wang. Weakly supervised extractive summarization with attention. In Haizhou Li, Gina-Anne Levow, Zhou Yu, Chitralekha Gupta, Berrak Sisman, Siqi Cai, David Vandyke, Nina Dethlefs, Yan Wu, and Junyi Jessy Li, editors,Proceedings of the 22nd Annual Meeting of the Special Interest Group on Discourse and Dialogue, pages 52...

work page 2021

[3] [3]

Reic: Rag-enhanced intent classification at scale, 2025

Ziji Zhang, Michael Yang, Zhiyu Chen, Yingying Zhuang, Shu-Ting Pi, Qun Liu, Rajashekar Maragoud, Vy Nguyen, and Anurag Beniwal. Reic: Rag-enhanced intent classification at scale, 2025

work page 2025

[4] [4]

https://www.anthropic.com/news/claude-3-family

Introducing the next generation of claude. https://www.anthropic.com/news/claude-3-family . Ac- cessed: 2025-04-20

work page 2025

[5] [5]

Command a: An enterprise-ready large language model, 2025

Team Cohere and et al. Command a: An enterprise-ready large language model, 2025

work page 2025

[6] [6]

How much is a triple? estimating the cost of knowledge graph creation

Heiko Paulheim. How much is a triple? estimating the cost of knowledge graph creation. 2018

work page 2018

[7] [7]

Seza Do˘gruöz, Sunayana Sitaram, Barbara E

A. Seza Do˘gruöz, Sunayana Sitaram, Barbara E. Bullock, and Almeida Jacqueline Toribio. A survey of code- switching: Linguistic and social perspectives for language technologies. In Chengqing Zong, Fei Xia, Wenjie Li, and Roberto Navigli, editors,Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th Internati...

work page 2021

[8] [8]

Are multilingual models effective in code-switching? In Thamar Solorio, Shuguang Chen, Alan W

Genta Indra Winata, Samuel Cahyawijaya, Zihan Liu, Zhaojiang Lin, Andrea Madotto, and Pascale Fung. Are multilingual models effective in code-switching? In Thamar Solorio, Shuguang Chen, Alan W. Black, Mona Diab, Sunayana Sitaram, Victor Soto, Emre Yilmaz, and Anirudh Srinivasan, editors,Proceedings of the Fifth Workshop on Computational Approaches to Lin...

work page 2021