Pith. sign in

REVIEW 4 major objections 6 minor 48 references

KinyaColBERT: A Lexically Grounded Retrieval Model for Low-Resource Retrieval-Augmented Generation

T0 review · 4 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read A retriever that tokenizes Kinyarwanda into meaningful morphemes outperforms multilingual embeddings and commercial APIs on a Kinyarwanda agricultural RAG benchmark.

desk verdict A useful empirical benchmark for Kinyarwanda retrieval, but the design claim that morphology and two-tier encoding are what works is not supported by the experiments as run. read the letter →

arxiv 2507.03241 v1 pith:NADDYG4X submitted 2025-07-04 cs.CL

Antoine Nzeyimana , Andre Niyongabo Rubungo This is my paper
classification cs.CL
keywords KinyaColBERTKinyarwandamorphologicaltokenizationtwo-tierencoderlateinteractionlow-resourceretrievalretrieval-augmentedgenerationagriculturalbenchmark
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 argues that the main reasons multilingual retrieval models fail on Kinyarwanda are poor language coverage and sub-word tokenization that splits words into meaningless pieces. To address this, it builds KinyaColBERT, a retriever that combines ColBERT-style late word-level interactions with a morphology-based tokenizer and a two-tier transformer encoder, and pre-trains it on a 2.8 GB Kinyarwanda corpus. On a Kinyarwanda agricultural retrieval benchmark, KinyaColBERT with 512-dimensional embeddings reaches 89.1 MRR@10 on the test set, beating the strongest commercial API by 16.8 points and local multilingual models by much larger margins. The paper concludes that morphology-aware, lexically grounded tokenization is a viable path to high-accuracy, cost-effective RAG in low-resource settings without machine translation.

What carries the argument

The central mechanism is the combination of a morphological tokenizer with a two-tier transformer encoder feeding ColBERT's late-interaction MaxSim operator. The lower tier contextualizes the morphological pieces of each inflected word (stem, affixes, part-of-speech and morphological tags), the upper tier contextualizes whole word embeddings within the sentence or document, and the retrieval score sums, over query tokens, the maximum cosine similarity to a document token. This yields lexically grounded, word-level embeddings that also let the model skip stop words and punctuation by part-of-speech category.

What would settle it

Train a standard-tokenizer ColBERT model on the same 2.8 GB Kinyarwanda corpus and fine-tune it on the same triplet data; if it reaches or exceeds KinyaColBERT's 89.1 MRR@10 on the test set, the claim that morphology-based tokenization and two-tier encoding drive the improvement would be falsified.

Watch

Extended reading notes

Core claim

KinyaColBERT is a ColBERT-type retriever whose embeddings are produced by a two-tier encoder: a lower tier uses self-attention over each word's morphological details (stem, affixes, part-of-speech tag, morphological tag), and an upper tier runs self-attention over the resulting word embeddings at the document level. Relevance is still scored with ColBERT's maximum-similarity operator over token embeddings, but the tokens are meaningful inflected forms rather than statistical sub-word pieces. On the Kinyarwanda agricultural benchmark, KinyaColBERT@512 outperforms all baselines: 89.1 versus 72.3 MRR@10 for Voyage-AI-large, 62.3 for ColBERT@1024, and 47.7 for mE5, with gaps between 16.8 and 64.9 percentage points. The authors take this as evidence that morphology-based tokenization and two-tier encoding are more appropriate for ColBERT-style late-interaction retrieval in morphologically rich languages.

Load-bearing premise

The paper assumes the accuracy gain comes from morphological tokenization and two-tier encoding, yet only KinyaColBERT received 2.8 GB of Kinyarwanda monolingual pre-training and no baseline was trained on that same corpus.

Editorial extensions

If this is right

  • Kinyarwanda RAG can be built on native-language retrieval without machine translation, avoiding translation latency, noise, and extra cost.
  • Embedding dimension matters and is worth tuning: 512-dimensional vectors outperformed 128, 1024, and 1536 on this benchmark.
  • The strongest commercial embedding API still trails KinyaColBERT by 16.8 MRR@10 points, so multilingual embeddings are not yet reliable for Kinyarwanda retrieval.
  • On an English machine-translated version of the same benchmark, KinyaColBERT still leads but by a smaller margin, suggesting the advantage is at least partly language-specific.
  • Morphology-based tokenization plus two-tier encoding is presented as a general recipe for ColBERT-style retrieval in morphologically rich low-resource languages.

Reading between the lines

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

  • As an editorial inference, the design claim is not isolated from the monolingual pre-training: because no baseline receives the 2.8 GB Kinyarwanda pre-training, the observed gains could come from pre-training data rather than from morphological tokenization and two-tier encoding.
  • Untested in the paper, but plausible: if the morpheme-level representations transfer, similar gains may appear for other Bantu languages with comparable agglutinative morphology, provided a morphological analyzer exists.
  • A practical implication beyond the paper's experiments: fewer retrieved passages per query would be needed at 89.1 MRR@10, which would lower RAG token costs in production, but the paper does not measure end-to-end RAG quality or cost.
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

4 major / 6 minor

Summary. The paper proposes KinyaColBERT, a retrieval model for Kinyarwanda that adapts ColBERT's late-interaction scoring to a morphology-based tokenizer and a two-tier transformer encoder: a lower tier models word-internal morphemes (stem, affixes, POS, morphological tags) and an upper tier models word sequences. The model is pre-trained on a 2.8 GB Kinyarwanda corpus with a masked morphological-detail objective and then fine-tuned on a newly constructed triplet dataset of about 21,000 Kinyarwanda agricultural questions. On this dataset, KinyaColBERT@512 is reported to outperform multilingual embedding models (mE5, BGE-M3, Jina-V3), commercial APIs (OpenAI, Voyage AI), and fine-tuned mBERT-based ColBERT baselines by large MRR@10 margins, and it also outperforms the same baselines when they are evaluated on an English machine-translated version of the data. The paper argues that morphology-based tokenization and two-tier encoding are more appropriate than standard subword tokenization for ColBERT-type retrieval in morphologically rich low-resource languages.

Significance. If the results hold, the paper makes a useful empirical contribution to low-resource retrieval-augmented generation: it introduces a new Kinyarwanda agricultural retrieval benchmark, a concrete morphological two-tier encoder design, and an unusually broad comparison that includes local multilingual models, commercial APIs, and a machine-translated English condition. The potential practical value for cost-effective RAG in low-resource settings is real, and the paper gives enough architectural and training detail to be replicable in principle. The central mechanistic claim, however, that morphology-based tokenization and two-tier encoding are what drive the gains, is not supported by the reported experiments because the model differs from the strongest baselines in several bundled factors, including a large monolingual pretraining corpus, architecture size, and tokenizer. The empirical benchmark result is plausible but is not yet established at the level of a specific design attribution, and the lack of uncertainty quantification and the asymmetry in the English-condition comparison further weaken the paper as currently written.

major comments (4)
  1. [Sections 3, 4.2, 5 and Table 2] The paper's headline design claim, stated in Section 5 and the contribution list, is that morphology-based tokenization and two-tier encoding are more appropriate for ColBERT-type retrieval. The experiments do not isolate this factor. KinyaColBERT differs from the fine-tuned mBERT-based ColBERT baseline in at least four ways: the morphological tokenizer, the two-tier encoder architecture, a 367M-parameter count versus 179M for the ColBERT baseline, and a 2.8 GB monolingual masked-language-model pretraining stage. No experiment trains a standard-tokenizer ColBERT on the same 2.8 GB Kinyarwanda corpus, and no ablation removes or adds the two-tier encoder while holding pretraining data constant. Without such a same-corpus control or a factorial ablation, the observed gains cannot be attributed to morphology and two-tier encoding rather than to monolingual pretraining, parameter count, or model capacity. This is load-bearing because the attribution is the paper's stated design contribution, not a peripheral observation.
  2. [Section 5, Tables 2 and 3] All reported results are single runs with no error bars, no confidence intervals, and no significance tests. The test set contains only 329 query-topic pairs, and the development set 196, so differences in MRR@10 and Accuracy@K could be sensitive to noise. Some headline gaps are large, but the paper claims superiority over all baselines across many metrics and two language conditions, and the absence of any uncertainty estimate is a barrier to treating these numbers as a reliable benchmark finding. The authors should report repeated runs with seeds, or at minimum bootstrap confidence intervals for the test-set metrics, and state whether the differences are statistically distinguishable.
  3. [Section 5, Table 3 and Figure 4(b)] The English-condition comparison is asymmetric and appears to conflate model language with evaluation language. Table 3 labels the KinyaColBERT rows as "This Work (Kinyarwanda)" and reports exactly the same Kinyarwanda numbers as Table 2, while all baseline rows use the Google-Translated English version. The sentence in Section 5 that "our KinyaColBERT model still performs best" is therefore comparing Kinyarwanda-input KinyaColBERT against English-input baselines, not comparing like with like. If the intended claim is that Kinyarwanda-language retrieval with KinyaColBERT is better than English-translated retrieval with baselines, that should be stated explicitly and justified as the relevant deployment comparison; if the intended claim is about English-condition retrieval ability, the authors need a run where KinyaColBERT is evaluated on the translated English data (or where queries are translated), and that run is absent. As written, Figure 4(b) and Table 3 do not support the interpretation given in the text.
  4. [Section 4.1] The evaluation benchmark is entirely author-constructed, and the paper gives no public release link for the dataset or the model, nor any inter-annotator agreement measure for the 21,000 collected questions. Since the triplet sampling procedure (100 random negatives plus intra-module hard negatives) determines the difficulty of the task, and since the validation and test topics are held out but no external validation on an independently constructed dataset is provided, the generalizability of the headline results is uncertain. At minimum, the dataset and annotation instructions should be released, and the authors should report how many annotators contributed per topic and whether annotation quality was checked. This is a reproducibility and external-validity concern for the paper's central comparison.
minor comments (6)
  1. [Abstract and Section 1] The abstract and introduction use informal phrasing such as "LLMS sometimes hallucinate" and "models that are more semantically aligned"; a light editorial pass would improve precision.
  2. [Figure 2 caption] The caption contains a typo, "Thye network encodes", which should read "The network encodes".
  3. [Section 5 and Table 3] Table 3's column header "Development Set Test Set" is malformed, and the table caption contains the typo "as well ass mean reciprocal rank"; these should be corrected.
  4. [Section 4.2] The paper reports that pretraining took 21 days on 8 RTX 4090 GPUs and fine-tuning took about 7 hours on an H200, but it does not state whether the model and training code will be released; given the computational cost, a code or model release would materially improve reproducibility.
  5. [Section 3] Equation (1) is standard ColBERT scoring, but the notation Lq and Ld is introduced without defining that they are query and document token counts; a brief definition would help readers.
  6. [References] Multiple references are duplicated, including Devlin et al. 2019a/2019b and the two Gao et al. 2023 entries; these should be merged or cross-referenced consistently.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: retrieval scores are measured empirically, and no prediction in the paper is equivalent by construction to a fitted input or to the authors' prior work.

full rationale

The paper's headline numbers (MRR@10 in Tables 2 and 3) are produced by training and evaluating the model on a fixed triplet split; nothing in Equation (1) or in the two-tier decomposition fθ(x)=fθS(fθM(x)) makes the reported ranking an identity, a refit of the benchmark labels, or a consequence of the cited prior work. The components taken from the authors' prior publications (the DeepKIN morphological analyzer and the KinyaBERT-style two-tier encoder) are used as implementational choices, not as the evidence for the retrieval claim; the claimed win is argued from measured comparisons, so the self-citations are not load-bearing in a circular sense. The paper's mechanistic attribution of the gain to morphology and two-tier encoding is confounded with the new 2.8 GB monolingual pre-training and with parameter count (367M versus 179M for the mBERT-based ColBERT baseline), because no one-factor-at-a-time ablation is provided; that is a correctness and attribution weakness, not a circular derivation. The benchmark is author-constructed, but queries were produced by human annotators from the topics and relevance is not derived from the model's own scores, so the evaluation is not self-referential. A separate statistical concern is that the highlighted 512-dimension result is the best test-set row even though the development set favors 1024-dimension; this is test-set selection rather than a reduction of the prediction to its inputs.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

No new physical or conceptual entities are introduced; the model is an architecture and tokenization scheme built from existing components. The central claim rests on the quality of the morphological analyzer, the attribution of gains to morphology rather than pre-training, and the validity of the self-constructed benchmark.

free parameters (3)
  • Token embedding dimension = 512 (also evaluated at 128, 256, 768, 1024, 1536)
    The final model uses the dimension that performed best on the development set (Figure 5), so the headline result depends on a hyperparameter search rather than an independent design rule.
  • Encoder layer counts and hidden sizes = 6 morphology layers, 11 sequence layers, hidden sizes 384 and 1536
    These architecture choices are inherited from KinyaBERT and not justified from first principles in this paper; the retrieval result depends on them.
  • POS filtering rule = skip prepositions and punctuation
    The model ignores entire part-of-speech categories during scoring (Section 3); this hand-chosen rule affects which tokens contribute to the MaxSim score.
assumptions (4)
  • domain assumption The DeepKIN morphological analyzer produces correct and complete morpheme segmentations for agricultural Kinyarwanda text.
    Section 3 and Section 4.2 rely on this analyzer for all tokenization; no accuracy or coverage numbers are provided.
  • domain assumption The improvement in retrieval accuracy is caused by morphology-based tokenization and two-tier encoding rather than by the 2.8 GB monolingual pre-training.
    Sections 3 and 5 state this as the design conclusion, but no ablation with a standard tokenizer under identical pre-training is reported.
  • domain assumption The 1,025 topics and 21,000 annotator questions form a valid, correctly labeled retrieval benchmark.
    Section 4.1 describes collection but reports no inter-annotator agreement or label-quality check.
  • domain assumption Google Translate preserves the question-topic relevance relation in the English version.
    Tables 2 and 3 use the translated set for cross-lingual comparison; translation noise is acknowledged but not quantified.

how reviews work

0 comments
Cite this review

Pith. "Pith review of KinyaColBERT: A Lexically Grounded Retrieval Model for Low-Resource Retrieval-Augmented Generation." pith.science (2026). https://pith.science/paper/NADDYG4X

@misc{pith2026250703241,
  author       = {Pith},
  title        = {Pith review of: KinyaColBERT: A Lexically Grounded Retrieval Model for Low-Resource Retrieval-Augmented Generation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NADDYG4X}},
  note         = {Machine review of arXiv:2507.03241}
}
read the original abstract

The recent mainstream adoption of large language model (LLM) technology is enabling novel applications in the form of chatbots and virtual assistants across many domains. With the aim of grounding LLMs in trusted domains and avoiding the problem of hallucinations, retrieval-augmented generation (RAG) has emerged as a viable solution. In order to deploy sustainable RAG systems in low-resource settings, achieving high retrieval accuracy is not only a usability requirement but also a cost-saving strategy. Through empirical evaluations on a Kinyarwanda-language dataset, we find that the most limiting factors in achieving high retrieval accuracy are limited language coverage and inadequate sub-word tokenization in pre-trained language models. We propose a new retriever model, KinyaColBERT, which integrates two key concepts: late word-level interactions between queries and documents, and a morphology-based tokenization coupled with two-tier transformer encoding. This methodology results in lexically grounded contextual embeddings that are both fine-grained and self-contained. Our evaluation results indicate that KinyaColBERT outperforms strong baselines and leading commercial text embedding APIs on a Kinyarwanda agricultural retrieval benchmark. By adopting this retrieval strategy, we believe that practitioners in other low-resource settings can not only achieve reliable RAG systems but also deploy solutions that are more cost-effective.

Figures

Figures reproduced from arXiv: 2507.03241 by the authors.

Figure 1
Figure 1. Examples of conversation pipelines of IVR-based RAG systems. The upper pipeline retrieves passages [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. KinyaColBERT uses a morphological tokenizer and a two-tier self-attention encoder architecture that [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Experimental mobile interface used by an [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Comparison of best retrieval performance across model variants. [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: KinyaColBERT performance across token embedding dimensions in comparison to best baseline model [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

48 extracted references · 19 canonical work pages

  1. [1]

    Jesujoba O Alabi, David Ifeoluwa Adelani, Marius Mosbach, and Dietrich Klakow. 2022. Adapting pre-trained language models to african languages via multilingual adaptive fine-tuning. arXiv preprint arXiv:2204.06487

    work page Pith review arXiv 2022

  2. [2]

    Ra \'e f Bahrini and Alaa A Qaffas. 2019. Impact of information and communication technology on economic growth: Evidence from developing countries. Economies, 7(1):21

    work page 2019

  3. [3]

    Fiskani Ella Banda, Vukosi Marivate, and Joyce Nakatumba-Nabende. 2025. A few-shot learning approach for a multilingual agro-information question answering system. Applied AI Letters, 6(2):e122

    work page 2025

  4. [4]

    Thomas Bauwens and Pieter Delobelle. 2024. Bpe-knockout: Pruning pre-existing bpe tokenisers with backwards-compatible morphological semi-supervision. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers), pages 5810--5832

    work page 2024

  5. [5]

    Sebastian Bruch, Xuanhui Wang, Michael Bendersky, and Marc Najork. 2019. An analysis of the softmax cross entropy loss for learning-to-rank with binary relevance. In Proceedings of the 2019 ACM SIGIR international conference on theory of information retrieval, pages 75--78

    work page 2019

  6. [6]

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

    arXiv 2024

  7. [7]

    Alexis Conneau, Kartikay Khandelwal, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Francisco Guzm \'a n, Edouard Grave, Myle Ott, Luke Zettlemoyer, and Veselin Stoyanov. 2020. https://doi.org/10.18653/v1/2020.acl-main.747 Unsupervised cross-lingual representation learning at scale . In Proceedings of the 58th Annual Meeting of the Association for Comp...

    doi:10.18653/v1/2020.acl-main.747 2020

  8. [8]

    Abhishek Darji, Fenil Kheni, Dhruvil Chodvadia, Parth Goel, Dweepna Garg, and Bankim Patel. 2024. Enhancing financial risk analysis using rag-based large language models. In 2024 3rd International Conference on Automation, Computing and Renewable Systems (ICACRS), pages 754--760. IEEE

    work page 2024

Show all 48 references
  1. [9]

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019 a . 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: hu...

    2019

  2. [10]

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019 b . https://doi.org/10.18653/v1/N19-1423 BERT : Pre-training of deep bidirectional transformers for language understanding . In Proceedings of the 2019 Conference of the North A merican Chapter of the Assoc...

    2019 doi

  3. [11]

    Bonaventure FP Dossou, Atnafu Lambebo Tonja, Oreen Yousuf, Salomey Osei, Abigail Oppong, Iyanuoluwa Shode, Oluwabusayo Olufunke Awoyomi, and Chris Chinenye Emezue. 2022. Afrolm: A self-active learning-based multilingual pretrained language model for 23 african languages. arXiv...

    2022 arXiv

  4. [12]

    Yunfan Gao, Yun Xiong, Xinyu Gao, Kangxiang Jia, Jinliu Pan, Yuxi Bi, Yixin Dai, Jiawei Sun, Haofen Wang, and Haofen Wang. 2023. Retrieval-augmented generation for large language models: A survey. arXiv preprint arXiv:2312.10997, 2:1

    2023 arXiv

  5. [13]

    Andargachew Mekonnen Gezmu and Andreas N \"u rnberger. 2023. Morpheme-based neural machine translation models for low-resource fusion languages. ACM Transactions on Asian and Low-Resource Language Information Processing, 22(9):1--19

    2023

  6. [14]

    Ozan Gokdemir, Carlo Siebenschuh, Alexander Brace, Azton Wells, Brian Hsu, Kyle Hippe, Priyanka V Setty, Aswathy Ajith, J Gregory Pauloski, Varuni Sastry, and 1 others. 2025. Hiperrag: High-performance retrieval augmented generation for scientific insights. arXiv preprint arXi...

    2025 arXiv

  7. [15]

    Michael G \"u nther, Jackmin Ong, Isabelle Mohr, Alaeddine Abdessalem, Tanguy Abel, Mohammad Kalim Akram, Susana Guzman, Georgios Mastrapas, Saba Sturua, Bo Wang, and 1 others. 2023. Jina embeddings 2: 8192-token general-purpose text embeddings for long documents. arXiv prepri...

    2023 arXiv

  8. [16]

    Kelvin Guu, Kenton Lee, Zora Tung, Panupong Pasupat, and Mingwei Chang. 2020. Retrieval augmented language model pre-training. In International conference on machine learning, pages 3929--3938. PMLR

    2020

  9. [17]

    Rami Hodrab, Mansoor Maitah, and Lubo s Smutka. 2016. The effect of information and communication technology on economic growth: Arab world case. International Journal of Economics and Financial Issues, 6(2):765--775

    2016

  10. [18]

    Valentin Hofmann, Janet B Pierrehumbert, and Hinrich Sch \"u tze. 2021. Superbizarre is not superb: Derivational morphology improves bert's interpretation of complex words. arXiv preprint arXiv:2101.00403

    2021 arXiv

  11. [19]

    Lei Huang, Weijiang Yu, Weitao Ma, Weihong Zhong, Zhangyin Feng, Haotian Wang, Qianglong Chen, Weihua Peng, Xiaocheng Feng, Bing Qin, and 1 others. 2025. A survey on hallucination in large language models: Principles, taxonomy, challenges, and open questions. ACM Transactions ...

    2025

  12. [20]

    Atia Shahnaz Ipa, Mohammad Abu Tareq Rony, and Mohammad Shariful Islam. 2025. https://aclanthology.org/2025.lm4uc-1.2/ Empowering low-resource languages: T ra S e architecture for enhanced retrieval-augmented generation in B angla . In Proceedings of the 1st Workshop on Langua...

    2025

  13. [21]

    a lev \"a , Hale Sirin, Abdullatif K \

    Mete Ismayilzada, Defne Circi, Jonne S \"a lev \"a , Hale Sirin, Abdullatif K \"o ksal, Bhuwan Dhingra, Antoine Bosselut, Duygu Ataman, and Lonneke van der Plas. 2024. Evaluating morphological compositional generalization in large language models. arXiv preprint arXiv:2410.12656

    2024 arXiv

  14. [22]

    Omar Khattab and Matei Zaharia. 2020. https://doi.org/10.1145/3397271.3401075 Colbert: Efficient and effective passage search via contextualized late interaction over bert . In Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Informatio...

    2020

  15. [23]

    Diederik P Kingma and Jimmy Ba. 2014. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980

    2014 arXiv

  16. [24]

    Meta Ayu Kurniawati. 2022. Analysis of the impact of information communication technology on economic growth: empirical evidence from asian countries. Journal of Asian Business and Economic Studies, 29(1):2--18

    2022

  17. [25]

    u ttler, Mike Lewis, Wen-tau Yih, Tim Rockt \

    Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich K \"u ttler, Mike Lewis, Wen-tau Yih, Tim Rockt \"a schel, and 1 others. 2020. Retrieval-augmented generation for knowledge-intensive nlp tasks. Advances in neural informati...

    2020

  18. [26]

    Shen Li, Yanli Zhao, Rohan Varma, Omkar Salpekar, Pieter Noordhuis, Teng Li, Adam Paszke, Jeff Smith, Brian Vaughan, Pritam Damania, and 1 others. 2020. Pytorch distributed: Experiences on accelerating data parallel training. arXiv preprint arXiv:2006.15704

    2020 arXiv

  19. [27]

    Ilya Loshchilov and Frank Hutter. 2017. Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101

    2017 arXiv

  20. [28]

    Carlo Merola and Jaspinder Singh. 2025. Reconstructing context: Evaluating advanced chunking strategies for retrieval-augmented generation. arXiv preprint arXiv:2504.19754

    2025 arXiv

  21. [29]

    Antoine Nzeyimana. 2020. https://doi.org/10.18653/v1/2020.coling-main.409 Morphological disambiguation from stemming data . In Proceedings of the 28th International Conference on Computational Linguistics, pages 4649--4660, Barcelona, Spain (Online). International Committee on...

    2020 doi

  22. [30]

    Antoine Nzeyimana. 2024. Low-resource neural machine translation with morphological modeling. arXiv preprint arXiv:2404.02392

    2024 arXiv

  23. [31]

    Antoine Nzeyimana and Andre Niyongabo Rubungo. 2022. https://doi.org/10.18653/v1/2022.acl-long.367 K inya BERT : a morphology-aware K inyarwanda language model . In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers),...

    2022 doi

  24. [32]

    Kelechi Ogueji, Yuxin Zhu, and Jimmy Lin. 2021. https://doi.org/10.18653/v1/2021.mrl-1.11 Small data? no problem! exploring the viability of pretrained multilingual language models for low-resourced languages . In Proceedings of the 1st Workshop on Multilingual Representation ...

    2021 doi

  25. [33]

    Odunayo Ogundepo, Tajuddeen R Gwadabe, Clara E Rivera, Jonathan H Clark, Sebastian Ruder, David Ifeoluwa Adelani, Bonaventure FP Dossou, Abdou Aziz Diop, Claytone Sikasote, Gilles Hacheme, and 1 others. 2023. Afriqa: Cross-lingual open-retrieval question answering for african ...

    2023 arXiv

  26. [34]

    Thilina Chaturanga Rajapakse, Andrew Yates, and Maarten de Rijke. 2024. https://doi.org/10.1145/3626772.3657854 Negative sampling techniques for dense passage retrieval in a multilingual setting . In Proceedings of the 47th International ACM SIGIR Conference on Research and De...

    2024

  27. [35]

    Nils Reimers and Iryna Gurevych. 2019. Sentence-bert: Sentence embeddings using siamese bert-networks. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJC...

    2019

  28. [36]

    Rico Sennrich, Barry Haddow, and Alexandra Birch. 2015. Neural machine translation of rare words with subword units. arXiv preprint arXiv:1508.07909

    2015 arXiv

  29. [37]

    Spurthi Setty, Harsh Thakkar, Alyssa Lee, Eden Chung, and Natan Vidra. 2024. Improving retrieval for rag based question answering models on financial documents. arXiv preprint arXiv:2404.07221

    2024 arXiv

  30. [38]

    Kurt Shuster, Spencer Poff, Moya Chen, Douwe Kiela, and Jason Weston. 2021. Retrieval augmentation reduces hallucination in conversation. arXiv preprint arXiv:2104.07567

    2021 arXiv

  31. [39]

    Landry Signé. 2025. Leveraging ai and emerging technologies to unlock africa’s potential. Technical report, Brookings Institution

    2025

  32. [40]

    Aina Gar \'i Soler, Matthieu Labeau, and Chlo \'e Clavel. 2024. https://doi.org/10.1162/tacl_a_00647 The impact of word splitting on the semantic content of contextualized word representations . Transactions of the Association for Computational Linguistics, 12:299--320

    2024 doi

  33. [41]

    Saba Sturua, Isabelle Mohr, Mohammad Kalim Akram, Michael G \"u nther, Bo Wang, Markus Krimmel, Feng Wang, Georgios Mastrapas, Andreas Koukounas, Nan Wang, and 1 others. 2024. jina-embeddings-v3: Multilingual embeddings with task lora. arXiv preprint arXiv:2409.10173

    2024 arXiv

  34. [42]

    Atnafu Lambebo Tonja, Bonaventure FP Dossou, Jessica Ojo, Jenalea Rajab, Fadel Thior, Eric Peter Wairagala, Anuoluwapo Aremu, Pelonomi Moiloa, Jade Abbott, Vukosi Marivate, and 1 others. 2024. Inkubalm: A small language model for low-resource african languages. arXiv preprint ...

    2024 arXiv

  35. [43]

    Cagri Toraman, Eyup Halit Yilmaz, Furkan S ahinu c , and Oguzhan Ozcelik. 2023. Impact of tokenization on language models: An analysis for turkish. ACM Transactions on Asian and Low-Resource Language Information Processing, 22(4):1--21

    2023

  36. [44]

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, ukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need. Advances in neural information processing systems, 30

    2017

  37. [45]

    Liang Wang, Nan Yang, Xiaolong Huang, Linjun Yang, Rangan Majumder, and Furu Wei. 2024. Multilingual e5 text embeddings: A technical report. arXiv preprint arXiv:2402.05672

    2024 arXiv

  38. [46]

    Xinyu Zhang, Nandan Thakur, Odunayo Ogundepo, Ehsan Kamalloo, David Alfonso-Hermelo, Xiaoguang Li, Qun Liu, Mehdi Rezagholizadeh, and Jimmy Lin. 2023. https://doi.org/10.1162/tacl_a_00595 MIRACL : A multilingual retrieval dataset covering 18 diverse languages . Transactions of...

    2023 doi

  39. [47]

    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...

  40. [48]

    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 gl...

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.