Pith. sign in

REVIEW 1 major objections 3 minor 64 references

MameLoshnLM: Yiddish Language Model and Evaluation Benchmark

T0 review · 1 major / 3 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 Continued pretraining on curated Yiddish text beats multilingual models on Yiddish tasks.

desk verdict Solid first Yiddish LM/corpus/benchmark worth refereeing; the main fix is a contamination audit for WikiANN and YiTB, which the paper currently lacks. read the letter →

arxiv 2608.05850 v1 pith:A2G2FIO2 submitted 2026-08-06 cs.CL cs.AI

Uri Katz , Omer Goldman , Tomasz Limisiewicz , Reut Tsarfaty , Noah A. Smith This is my paper
classification cs.CLcs.AI
keywords Yiddishlanguagemodelcontinuedpretraininglow-resourceNLPevaluationbenchmarkdatacontaminationloshn-koydeshlexiconmorphologicalcompetencemC4audit
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 dedicated Yiddish language model, built by continuing the pretraining of Llama 3.1 8B on a newly curated Yiddish corpus called Oytser, outperforms open multilingual models of similar size on a new multi-task Yiddish benchmark called Kashes. Because the benchmark gains concentrate on Yiddish-centered tasks, and because the model produces loshn-koydesh (Hebrew/Aramaic-origin) vocabulary and Yiddish-specific inflection at rates closer to native text, the authors argue the gains reflect genuine linguistic competence rather than generic scale. The paper also reports an audit of the mC4 corpus's Yiddish split showing that less than half of its documents are genuine Yiddish, with the rest machine-translated or misidentified Hebrew. If the claims hold, Yiddish gets its first open language model, a reusable pretraining corpus, a multi-task evaluation resource, and a replicable recipe for languages that are historically rich but digitally underrepresented.

What carries the argument

The load-bearing mechanism is the combination of Oytser, a corpus that mixes contemporary web-native Yiddish with more than 12,000 OCRed books from the Yiddish Book Center, and continued pretraining of Llama 3.1 8B on a Yiddish-dominant mixture (72% Yiddish words, 28% English). The evaluation and analysis are carried by Kashes, a nine-task benchmark spanning translation, linguistic analysis, NER, and understanding, together with two probes: a loshn-koydesh lexicon for measuring Hebrew/Aramaic-origin vocabulary in translation output, and lemmatization on the UD Yiddish treebank broken down by morphological category, including a within-task control comparing the suppletive auxiliary 'zayn' with the regular 'hobn'.

What would settle it

Check for verbatim or near-verbatim overlap between the Oytser pretraining corpus and the Kashes test sets beyond Kashes-mt: if WikiANN test sentences (drawn from the 2018 Yiddish Wikipedia) or YiTB treebank sentences appear in the corresponding Oytser sources, the reported NER and linguistic-analysis scores are inflated. A retraining experiment with those overlapping documents removed would show whether the gains survive.

Watch

Extended reading notes

Core claim

The central claim is that targeted resource construction plus continued pretraining can substantially narrow the gap between what general multilingual models achieve in a low-resource language and what a language-specific model achieves. Specifically, MameLoshnLM, trained for one epoch on roughly 5.3 billion Yiddish tokens with a small English auxiliary mixture, reaches an average score of 62.6 on the Kashes benchmark, ahead of Llama 3.1 8B (56.8) and Gemma-2 9B (57.0), with its largest leads on English-to-Yiddish translation, part-of-speech tagging, dependency parsing, transliteration, and two of three named-entity-recognition tasks. The authors further claim that the advantage is qualitative as well as quantitative: the model recovers the loshn-koydesh lexical layer that general models deplete, and handles Yiddish-specific morphology such as ge- participles, Hebrew-origin plurals, and the suppletive auxiliary 'zayn' much better than the base model, which tends to copy surface forms unchanged.

Load-bearing premise

The load-bearing premise is that the Kashes benchmark measures genuine Yiddish competence rather than memorization of test material: the paper excludes Kashes-mt source documents from Oytser but does not report excluding or auditing the Yiddish Wikipedia source against WikiANN, or the YBC books against the YiTB treebank test sets.

Editorial extensions

If this is right

  • Yiddish gains an open 8B model, a high-quality pretraining corpus, and a multi-task benchmark, enabling downstream work in Yiddish NLP and digital humanities.
  • The mC4 audit establishes that labeled language splits in web-scale corpora can be dominated by machine-translated spam and script-confused text; similar audits for other low-resource languages would likely be prudent before training on them.
  • Keeping a small English share in the pretraining mixture preserves base-model capabilities better than reallocating that share to historically related languages such as German and Hebrew.
  • The linguistic probes show that a language-specific model produces native-like lexical and morphological patterns at much higher rates, suggesting that curated native data can recover competence that noisy web data erodes.

Reading between the lines

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

  • A contamination audit beyond Kashes-mt would directly test the benchmark-level claim: WikiANN is built from the 2018 Yiddish Wikipedia and the YiTB treebank is a curated literary resource, and the paper does not report whether their test sentences also appear in Oytser; if they do, the reported NER and linguistic-analysis gains would be partly attributable to memorization.
  • The same methodology—an mC4-style audit, a curated corpus, continued pretraining, and lexicon-plus-morphology probes—could be ported to other languages with a classical or liturgical lexical layer, such as Ladino or Judeo-Arabic, to test whether the loshn-koydesh depletion pattern generalizes.
  • The per-word LK recall gap suggests a concrete failure mode for evaluation: COMET scores correlate only weakly with LK recall, so translation quality metrics may miss systematic loss of heritage vocabulary; other low-resource benchmarks may need lexicon-based metrics to catch this.
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

1 major / 3 minor

Summary. The paper introduces MameLoshnLM, an 8B-parameter Yiddish language model obtained by continued pretraining of Llama 3.1 8B on Oytser, a newly assembled Yiddish corpus that combines web-native sources with digitized Yiddish books. The paper also introduces Kashes, a multi-task benchmark spanning translation, linguistic analysis, named entity recognition, and machine-translated language-understanding tasks. The central reported result is that MameLoshnLM outperforms five open baselines of similar scale on Kashes on average, with the strongest gains on Yiddish-centered tasks, and that the gains reflect genuine Yiddish competence as shown by lexical (loshn-koydesh) and morphological probes. A secondary contribution is an audit of mC4's Yiddish split, which the paper claims contains substantial machine-translated and misidentified-Hebrew content.

Significance. If the benchmark numbers are clean, this is a substantial contribution to low-resource NLP: it provides the first open Yiddish LLM, a high-quality pretraining corpus, a multi-task Yiddish benchmark, and a replicable continued-pretraining recipe. The paper is unusually transparent: the model, corpus, and benchmark artifacts are released; the mC4 audit is documented in detail; the self-built Kashes-mt benchmark is explicitly decontaminated by excluding all source documents from the pretraining corpus; and the morphological probes include a well-designed within-task control (regular vs. suppletive auxiliary). The mC4 audit and the loshn-koydesh analysis are valuable independent findings that go beyond the benchmark numbers.

major comments (1)
  1. [Section 4, Table 3; Section 3.2, Table 2] The benchmark-level claim of genuine Yiddish competence depends on uncontaminated test sets, but the paper only reports decontamination for Kashes-mt (§4.1: "we exclude every source document from the Oytser pretraining corpus"). No overlap audit is reported for WikiANN or the YiTB treebank tasks, even though Table 3 describes WikiANN as "tagged entities from the 2018 Yiddish Wikipedia dump" and Table 2 lists the Yiddish Wikipedia as a training source in Oytser (15,300 documents). The YiTB tasks (POS, dependency parsing, lemmatization, transliteration) could similarly overlap with the Wikipedia or YBC portions of Oytser. If any test sentences or documents appear in pretraining, the reported scores in Table 4 (e.g., WikiANN F1 59.7, POS 88.6, lemmatization 31.9) and the morphological probes in Section 8 (Tables 5 and 12) could be inflated by surface memorization rather than reflecting generalizable competence. This is load-bearing for the paper's central claim that continued pretraining yields genuine linguistic gains rather than benchmark overfitting. The authors should perform and report an explicit overlap analysis (exact and near-duplicate document and sentence matching, e.g., 8-gram overlap) between Oytser and each Kashes test set, and re-report the affected numbers with any overlapping instances removed, or otherwise demonstrate that overlap is negligible.
minor comments (3)
  1. [Section 7.1, Table 4] Several row-level gaps between MameLoshnLM and the closest baseline are smaller than one point (e.g., dependency parsing LAS 40.6 vs. 40.3 for Qwen3; POS 88.6 vs. 87.6 for Gemma-2). Reporting confidence intervals, standard deviations across evaluation runs, or a significance test for these comparisons would strengthen the claim of consistent superiority.
  2. [Section 3.1, Appendix D] The mC4 audit's machine-translation classification relies on URL fingerprints and a locale-sibling count, but the threshold for flagging a domain as machine-translated is not precisely stated beyond the qualitative description "fewer than one" language edition for native domains. A precise cutoff and the distribution of the sibling-count metric would improve reproducibility.
  3. [Section 8, Appendix E.1] The LK content-word rate uses the number of content tokens as the denominator, but compound LK phrases are counted as a single item in the numerator. The counting convention for the denominator should be stated explicitly so that it is clear whether compound components are also excluded from the denominator.

Circularity Check

0 steps flagged · score 0.0 of 10

No load-bearing circularity: Kashes-mt source documents are explicitly removed from Oytser, the remaining benchmark tasks are external, and the Section 8 probes are independent analyses; potential overlap risks are contamination concerns, not circular derivations.

full rationale

The paper's derivation chain is empirical rather than definitional: MameLoshnLM is obtained by continued pretraining on Oytser and then evaluated on Kashes. The only benchmark component constructed by the authors, Kashes-mt, is explicitly decontaminated relative to the training corpus: Section 4.1 states that 'to prevent benchmark leakage, we exclude every source document from the Oytser pretraining corpus,' and Figure 1 confirms that 'source documents removed from Oytser pretraining corpus.' Thus the Kashes-mt translation scores, which are the main self-built evaluation, are not defined in terms of the training input. The other Kashes tasks come from external or prior resources (FLORES+, UD Yiddish-YiTB, EHRI, WikiANN, newNLP, and Aya PIQA/WikiQA/PAWS), and the Section 8 linguistic-competence claims are measured against an external loshn-koydesh lexicon and the UD Yiddish-YiTB test set, not against quantities fitted in this paper. The mC4 audit is a separate corpus-quality characterization and is not used as an input to the model. The paper does contain a minor self-citation, ``Jacovi et al. (2023),'' whose author list overlaps with the present authors, but it is used only to support the decontamination practice and is not load-bearing. The more substantive concern, that Oytser contains the Yiddish Wikipedia and YBC sources while WikiANN derives from the 2018 Yiddish Wikipedia dump and YiTB may draw on closely related treebank texts, is a possible data-contamination or benchmark-validity risk, not a case where a prediction reduces by construction to its training input. Accordingly, no circular step satisfying the required standard is present, and the circularity score is 0.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

No new physical or theoretical entities are postulated; the new artifacts are a model, a corpus, and a benchmark, none of which are unobserved constructs. The main dependencies are empirical assumptions about corpus quality, alignment validity, the LK lexicon, and the usability of machine-translated evaluation tasks. The free parameters are benchmark construction thresholds and a training mixture choice, all disclosed in the paper.

free parameters (3)
  • SentAlign score threshold = 0.65
    Section 4.1 and Appendix A, Figure 3: sentence pairs with similarity below 0.65 are discarded when building Kashes-mt; the threshold is chosen from the score distribution, not fitted to model performance.
  • Document alignment threshold = 35%
    Section 4.1 and Appendix A, Figure 2: documents with fewer than 35% aligned sentences are discarded, affecting benchmark composition.
  • Yiddish-English training mixture = 72% Yiddish / 28% English by words; 91.8% / 8.2% by tokens
    Section 5: the continued pretraining mixture is a chosen design point, and Section 7.2 tests alternative related-language mixtures, showing the chosen configuration works best.
assumptions (5)
  • domain assumption The mC4 Yiddish split audit correctly distinguishes native Yiddish, machine-translated, and misidentified Hebrew pages.
    Appendix D: classification uses URL fingerprints, locale-sibling counts, and a Wikipedia-trained Hebrew/Yiddish classifier; the authors acknowledge manual review and some subjectivity. This supports the claim that public Yiddish web corpora are noisy.
  • domain assumption SentAlign with LaBSE similarity reliably identifies valid Yiddish-English translation pairs.
    Section 4.1: Kashes-mt depends on document matching and sentence alignment with thresholds of 35% aligned and score 0.65; alignment errors propagate into the translation benchmark.
  • domain assumption The loshn-koydesh lexicon and content-word and stopword lists correctly isolate Hebrew and Aramaic-origin Yiddish vocabulary.
    Appendix E.1: LK identification relies on the Bleaman and Niborski lexicon, tokenization rules, and a curated stopword list; matching is exact or diacritic-stripped, and compound phrases are handled separately.
  • domain assumption The Aya machine-translated tasks, PIQA, WikiQA, and PAWS-Wiki, preserve enough task structure to serve as valid Yiddish reasoning evaluations.
    Section 4: the authors state they retained only tasks whose translations preserved essential structure and intent, a judgment call, and these MT-derived tasks are part of the benchmark.
  • domain assumption UD Yiddish-YiTB annotations are correct gold labels for POS, lemmatization, parsing, and transliteration.
    Section 4 and Appendix E: YiTB is used for four linguistic tasks and for morphological probe categories; errors in gold labels would affect both models symmetrically but would change absolute numbers.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MameLoshnLM: Yiddish Language Model and Evaluation Benchmark." pith.science (2026). https://pith.science/paper/A2G2FIO2

@misc{pith2026260805850,
  author       = {Pith},
  title        = {Pith review of: MameLoshnLM: Yiddish Language Model and Evaluation Benchmark},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/A2G2FIO2}},
  note         = {Machine review of arXiv:2608.05850}
}
read the original abstract

We present MameLoshnLM, the first open-source 8B-parameter language model built specifically for Yiddish. Despite Yiddish's rich textual tradition, its limited digital presence and the scarcity of reliable evaluation resources have constrained progress in Yiddish language modeling. Existing multilingual corpora and benchmarks are often poor proxies for the language, containing substantial amounts of noisy, machine-translated, and misclassified text. We address these gaps by introducing Oytser, a high-quality Yiddish pretraining corpus that combines contemporary web-native sources with literary materials, and Kashes, a multi-task benchmark spanning translation, linguistic analysis, information extraction, and language understanding. Using these resources, we continue pretraining Llama 3.1 8B to obtain MameLoshnLM. Across the tasks in the benchmark, MameLoshnLM outperforms open baselines of similar scale. Our analyses show that these gains are not only quantitative: relative to general-purpose multilingual models, MameLoshnLM better captures language-defining lexical and morphological patterns, pointing to a broader failure mode of noisy web-scale multilingual data for low-resource languages. Our results provide both a foundation for Yiddish NLP and a practical template for language model development in historically rich but digitally underrepresented languages.

Figures

Figures reproduced from arXiv: 2608.05850 by the authors.

Figure 1
Figure 1. The Kashes-mt construction pipeline. Yiddish-English editions of the same articles are matched and sentence-aligned to construct a natively authored translation benchmark 4.1 Kashes-mt: New Yiddish Translation Benchmark Existing Yiddish–English parallel corpora, such as FLORES+ (NLLB Team et al., 2024), are derived from English source texts translated into Yiddish. Therefore, they do not capture the authentic voice … view at source ↗
Figure 2
Figure 2. The distribution of the portion of initially aligned sentences in all sentences in [PITH_FULL_IMAGE:figures/full_fig_p016_2.png] view at source ↗
Figure 3
Figure 3. The distribution of SentAlign similarity score across initially aligned documents. [PITH_FULL_IMAGE:figures/full_fig_p017_3.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

64 extracted references · 38 canonical work pages

  1. [1]

    Attention is All you Need , url =

    Vaswani, Ashish and Shazeer, Noam and Parmar, Niki and Uszkoreit, Jakob and Jones, Llion and Gomez, Aidan N and Kaiser, ukasz and Polosukhin, Illia , booktitle =. Attention is All you Need , url =

  2. [2]

    NLLB Team and Costa-juss \`a , Marta R. and Cross, James and C elebi, Onur and Elbayad, Maha and Heafield, Kenneth and Heffernan, Kevin and Kalbassi, Elahe and Lam, Janice and Licht, Daniel and Maillard, Jean and Sun, Anna and Wang, Skyler and Wenzek, Guillaume and Youngblood, Al and Akula, Bapi and Barrault, Loic and Gonzalez, Gabriel Mejia and Hansanti,...

    2024

  3. [3]

    Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing: System Demonstrations

    Steingrímsson, Steinþór and Loftsson, Hrafn and Way, Andy. Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing: System Demonstrations. 2023

    work page 2023

  4. [4]

    2025 , eprint=

    The Bitter Lesson Learned from 2,000+ Multilingual Benchmarks , author=. 2025 , eprint=

    2025

  5. [5]

    2024 , eprint=

    RakutenAI-7B: Extending Large Language Models for Japanese , author=. 2024 , eprint=

    work page 2024

  6. [6]

    Advances in Neural Information Processing Systems , volume=

    Madlad-400: A multilingual and document-level large audited dataset , author=. Advances in Neural Information Processing Systems , volume=

    work page

  7. [7]

    arXiv preprint arXiv:2304.09151 , year=

    Unimax: Fairer and more effective language sampling for large-scale multilingual pretraining , author=. arXiv preprint arXiv:2304.09151 , year=

    arXiv

  8. [8]

    2023 , eprint=

    Typhoon: Thai Large Language Models , author=. 2023 , eprint=

    work page 2023

Show all 64 references
  1. [9]

    doi:10.57967/hf/0938 , publisher =

    gpt2-bengali (Revision cb8fff6) , year = 2023, url =. doi:10.57967/hf/0938 , publisher =

    2023 doi

  2. [10]

    2023 , eprint=

    PolyLM: An Open Source Polyglot Large Language Model , author=. 2023 , eprint=

    2023

  3. [11]

    Text Matters: A Journal of Literature, Theory and Culture , number=

    The Continuing Story of the Yiddish Language: The Yiddish Book Center in Amherst, Massachusetts , author=. Text Matters: A Journal of Literature, Theory and Culture , number=

  4. [13]

    Matthew Kirk Andrews , year=

  5. [14]

    Proceedings of the AAAI conference on artificial intelligence , volume=

    Piqa: Reasoning about physical commonsense in natural language , author=. Proceedings of the AAAI conference on artificial intelligence , volume=

  6. [15]

    2021 , howpublished =

    Berkovitch, Ephraim and Rusinek, Sinai , title =. 2021 , howpublished =

    2021

  7. [16]

    Sherkala-Chat: Building a State-of-the-Art

    Fajri Koto and Rituraj Joshi and Nurdaulet Mukhituly and Yuxia Wang and Zhuohan Xie and Rahul Pal and Daniil Orel and Parvez Mullah and Diana Turmakhan and Maiya Goloburda and Mohammed Kamran and Samujjwal Ghosh and Bokang Jia and Jonibek Mansurov and Mukhammed Togmanov and De...

    2025

  8. [17]

    Language Documentation & Conservation , volume=

    The corpus of spoken Yiddish in Europe: Goals, methods, and applications , author=. Language Documentation & Conservation , volume=. 2025 , publisher=

    2025

  9. [18]

    , author=

    REYD-The First Yiddish Text-to-Speech Dataset and System. , author=. INTERSPEECH , pages=

  10. [20]

    2021 , url =

    Santorini, Beatrice , title =. 2021 , url =

    2021

  11. [24]

    2024 , eprint=

    Gemma 2: Improving Open Language Models at a Practical Size , author=. 2024 , eprint=

    2024

  12. [25]

    2024 , eprint=

    EuroLLM: Multilingual Language Models for Europe , author=. 2024 , eprint=

    2024

  13. [26]

    The BigScience

    Hugo Lauren. The BigScience. Thirty-sixth Conference on Neural Information Processing Systems Datasets and Benchmarks Track , year=

  14. [27]

    Internet Research: Ethical Guidelines 3.0. 2020

    2020

  15. [28]

    2025 , month = may, type =

    Copyright and Artificial Intelligence: Part 3 -- Generative AI Training , author =. 2025 , month = may, type =

    2025

  16. [29]

    2022 , month =

    Uses of Copyrighted Materials for Machine Learning , author =. 2022 , month =

    2022

  17. [30]

    Towards a cleaner document-oriented multilingual crawled corpus

    Julien Abadji, Pedro Ortiz Suarez, Laurent Romary, and Beno \^i t Sagot. Towards a cleaner document-oriented multilingual crawled corpus. In Nicoletta Calzolari, Fr \'e d \'e ric B \'e chet, Philippe Blache, Khalid Choukri, Christopher Cieri, Thierry Declerck, Sara Goggi, Hito...

    2022

  18. [31]

    Judging quality across languages: A multilingual approach to pretraining data filtering with language models

    Mehdi Ali, Manuel Brack, Max L \"u bbering, Elias Wendt, Abbas Goher Khan, Richard Rutmann, Alex Jude, Maurice Kraus, Alexander Arno Weber, Felix Stollenwerk, David Kacz \'e r, Florian Mai, Lucie Flek, Rafet Sifa, Nicolas Flores-Herr, Joachim Koehler, Patrick Schramowski, Mich...

    2025

  19. [32]

    YiTB : the yiddish tree bank, 2025

    Matthew Kirk Andrews. YiTB : the yiddish tree bank, 2025. URL https://github.com/UniversalDependencies/UD_Yiddish-YiTB

    2025

  20. [33]

    Leak, cheat, repeat: Data contamination and evaluation malpractices in closed-source LLM s

    Simone Balloccu, Patr \'i cia Schmidtov \'a , Mateusz Lango, and Ondrej Dusek. Leak, cheat, repeat: Data contamination and evaluation malpractices in closed-source LLM s. In Yvette Graham and Matthew Purver (eds.), Proceedings of the 18th Conference of the European Chapter of ...

    2024 doi

  21. [34]

    New Languages for NLP: Yiddish --- annotated data and models

    Ephraim Berkovitch and Sinai Rusinek. New Languages for NLP: Yiddish --- annotated data and models. GitHub repository, New Languages for NLP project, Princeton University, 2021. URL https://github.com/New-Languages-for-NLP/yiddish. MIT License

    2021

  22. [35]

    The corpus of spoken yiddish in europe: Goals, methods, and applications

    Isaac L Bleaman and Chaya R Nove. The corpus of spoken yiddish in europe: Goals, methods, and applications. Language Documentation & Conservation, 19, 2025

    2025

  23. [36]

    Pula: Training large language models for setswana

    Nathan Brown and Vukosi Marivate. Pula: Training large language models for setswana. In Luis Chiruzzo, Alan Ritter, and Lu Wang (eds.), Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Te...

    2025 doi

  24. [37]

    Generating a Y iddish speech corpus, forced aligner and basic ASR system for the AHEYM project

    Malgorzata \'C avar, Damir \'C avar, Dov-Ber Kerler, and Anya Quilitzsch. Generating a Y iddish speech corpus, forced aligner and basic ASR system for the AHEYM project. In Nicoletta Calzolari, Khalid Choukri, Thierry Declerck, Sara Goggi, Marko Grobelnik, Bente Maegaard, Jose...

    2016

  25. [38]

    Unsupervised cross-lingual representation learning at scale

    Alexis Conneau, Kartikay Khandelwal, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Francisco Guzm \'a n, Edouard Grave, Myle Ott, Luke Zettlemoyer, and Veselin Stoyanov. Unsupervised cross-lingual representation learning at scale. In Dan Jurafsky, Joyce Chai, Natalie Schlu...

    2020 doi

  26. [39]

    Repurposing holocaust-related digital scholarly editions to develop multilingual domain-specific named entity recognition tools

    Maria Dermentzi and Hugo Scheithauer. Repurposing holocaust-related digital scholarly editions to develop multilingual domain-specific named entity recognition tools. In Isuri Anuradha, Martin Wynne, Francesca Frontini, and Alistair Plum (eds.), Proceedings of the First Worksh...

    2024

  27. [40]

    Pretraining language models using translationese

    Meet Doshi, Raj Dabre, and Pushpak Bhattacharyya. Pretraining language models using translationese. In Yaser Al-Onaizan, Mohit Bansal, and Yun-Nung Chen (eds.), Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pp.\ 5843--5862, Miami, Flor...

    2024 doi

  28. [41]

    Emergent abilities of large language models under continued pre-training for language adaptation

    Ahmed Elhady, Eneko Agirre, and Mikel Artetxe. Emergent abilities of large language models under continued pre-training for language adaptation. In Wanxiang Che, Joyce Nabende, Ekaterina Shutova, and Mohammad Taher Pilehvar (eds.), Proceedings of the 63rd Annual Meeting of the...

    2025 doi

  29. [42]

    Latxa: An open language model and evaluation suite for B asque

    Julen Etxaniz, Oscar Sainz, Naiara Perez, Itziar Aldabe, German Rigau, Eneko Agirre, Aitor Ormazabal, Mikel Artetxe, and Aitor Soroa. Latxa: An open language model and evaluation suite for B asque. In Lun-Wei Ku, Andre Martins, and Vivek Srikumar (eds.), Proceedings of the 62n...

    2024 doi

  30. [43]

    Internet Research: Ethical Guidelines 3.0, volume 3

    Aline Shakti Franzke, Anja Bechmann, Charles Melvin Ess, and Michael Zimmer (eds.). Internet Research: Ethical Guidelines 3.0, volume 3. AoIR (The International Association of Internet Researchers), 2020

    2020

  31. [44]

    Gemma 2: Improving open language models at a practical size, 2024

    Gemma Team . Gemma 2: Improving open language models at a practical size, 2024. URL https://arxiv.org/abs/2408.00118

    2024 arXiv

  32. [45]

    Creating a high-quality machine translation system for a low-resource language: Y iddish

    Dmitriy Genzel, Klaus Macherey, and Jakob Uszkoreit. Creating a high-quality machine translation system for a low-resource language: Y iddish. In Proceedings of Machine Translation Summit XII: Papers, Ottawa, Canada, August 26-30 2009. URL https://aclanthology.org/2009.mtsummi...

    2009

  33. [46]

    The llama 3 herd of models

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

    2024 arXiv

  34. [47]

    Uses of copyrighted materials for machine learning

    Israel Ministry of Justice . Uses of copyrighted materials for machine learning. Technical report, Ministry of Justice, Jerusalem, Israel, 12 2022. URL https://www.gov.il/BlobFolder/legalinfo/machine-learning/he/18-12-2022.pdf. Non-binding legal opinion on the permissibility o...

    2022

  35. [48]

    Stop uploading test data in plain text: Practical strategies for mitigating data contamination by evaluation benchmarks

    Alon Jacovi, Avi Caciularu, Omer Goldman, and Yoav Goldberg. Stop uploading test data in plain text: Practical strategies for mitigating data contamination by evaluation benchmarks. In Houda Bouamor, Juan Pino, and Kalika Bali (eds.), Proceedings of the 2023 Conference on Empi...

    2023 doi

  36. [49]

    Sherkala-chat: Building a state-of-the-art LLM for kazakh in a moderately resourced setting

    Fajri Koto, Rituraj Joshi, Nurdaulet Mukhituly, Yuxia Wang, Zhuohan Xie, Rahul Pal, Daniil Orel, Parvez Mullah, Diana Turmakhan, Maiya Goloburda, Mohammed Kamran, Samujjwal Ghosh, Bokang Jia, Jonibek Mansurov, Mukhammed Togmanov, Debopriyo Banerjee, Nurkhan Laiyk, Akhmed Sakip...

    2025

  37. [50]

    u ller, Andr \'e M \

    Julia Kreutzer, Isaac Caswell, Lisa Wang, Ahsan Wahab, Daan van Esch, Nasanbayar Ulzii-Orshikh, Allahsera Tapo, Nishant Subramani, Artem Sokolov, Claytone Sikasote, Monang Setyawan, Supheakmungkol Sarin, Sokhar Samb, Beno \^i t Sagot, Clara Rivera, Annette Rios, Isabel Papadim...

    2022

  38. [51]

    A part-of-speech tagger for yiddish

    Seth Kulick, Neville Ryant, Beatrice Santorini, Joel Wallenberg, and Assaf Urieli. A part-of-speech tagger for yiddish. arXiv preprint arXiv:2204.01175, 2022

    2022 arXiv

  39. [52]

    Teaching llama a new language through cross-lingual knowledge transfer

    Hele-Andra Kuulmets, Taido Purason, Agnes Luhtaru, and Mark Fishel. Teaching llama a new language through cross-lingual knowledge transfer. In Kevin Duh, Helena Gomez, and Steven Bethard (eds.), Findings of the Association for Computational Linguistics: NAACL 2024, pp.\ 3309--...

    2024 doi

  40. [53]

    The bigscience ROOTS corpus: A 1.6 TB composite multilingual dataset

    Hugo Lauren c on, Lucile Saulnier, Thomas Wang, Christopher Akiki, Albert Villanova del Moral, Teven Le Scao, Leandro Von Werra, Chenghao Mou, Eduardo Gonz \'a lez Ponferrada, Huu Nguyen, J \"o rg Frohberg, Mario S a s ko, Quentin Lhoest, Angelina McMillan-Major, G \'e rard Du...

    2022

  41. [54]

    Guerreiro, Ricardo Rei, Duarte M

    Pedro Henrique Martins, Patrick Fernandes, João Alves, Nuno M. Guerreiro, Ricardo Rei, Duarte M. Alves, José Pombal, Amin Farajian, Manuel Faysse, Mateusz Klimaszewski, Pierre Colombo, Barry Haddow, José G. C. de Souza, Alexandra Birch, and André F. T. Martins. Eurollm: Multil...

    2024 arXiv

  42. [55]

    Crosslingual generalization through multitask finetuning

    Niklas Muennighoff, Thomas Wang, Lintang Sutawika, Adam Roberts, Stella Biderman, Teven Le Scao, M Saiful Bari, Sheng Shen, Zheng-Xin Yong, Hailey Schoelkopf, et al. Crosslingual generalization through multitask finetuning. arXiv preprint arXiv:2211.01786, 2022

    2022 arXiv

  43. [56]

    NLLB Team , Marta R. Costa-juss \`a , James Cross, Onur C elebi, Maha Elbayad, Kenneth Heafield, Kevin Heffernan, Elahe Kalbassi, Janice Lam, Daniel Licht, Jean Maillard, Anna Sun, Skyler Wang, Guillaume Wenzek, Al Youngblood, Bapi Akula, Loic Barrault, Gabriel Mejia Gonzalez,...

    2024

  44. [57]

    Massively multilingual transfer for NER

    Afshin Rahimi, Yuan Li, and Trevor Cohn. Massively multilingual transfer for NER . In Anna Korhonen, David Traum, and Llu \'i s M \`a rquez (eds.), Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pp.\ 151--164, Florence, Italy, July 201...

    2019 doi

  45. [58]

    A multi-orthography parallel corpus of Y iddish nouns

    Jonne Saleva. A multi-orthography parallel corpus of Y iddish nouns. In Nicoletta Calzolari, Fr \'e d \'e ric B \'e chet, Philippe Blache, Khalid Choukri, Christopher Cieri, Thierry Declerck, Sara Goggi, Hitoshi Isahara, Bente Maegaard, Joseph Mariani, H \'e l \`e ne Mazo, Asu...

    2020

  46. [59]

    Penn parsed corpus of historical yiddish, v1.0, 2021

    Beatrice Santorini. Penn parsed corpus of historical yiddish, v1.0, 2021. URL https://github.com/beatrice57/penn-parsed-corpus-of-historical-yiddish. Accessed: 2026-03-26

    2021

  47. [60]

    Shivalika Singh, Freddie Vargus, Daniel D ' souza, B \"o rje F. Karlsson, Abinaya Mahendiran, Wei-Yin Ko, Herumb Shandilya, Jay Patel, Deividas Mataciunas, Laura O ' Mahony, Mike Zhang, Ramith Hettiarachchi, Joseph Wilson, Marina Machado, Luisa Moura, Dominik Krzemi \'n ski, H...

    2024

  48. [61]

    Dolma: an open corpus of three trillion tokens for language model pretraining research

    Luca Soldaini, Rodney Kinney, Akshita Bhagia, Dustin Schwenk, David Atkinson, Russell Authur, Ben Bogin, Khyathi Chandu, Jennifer Dumas, Yanai Elazar, Valentin Hofmann, Ananya Jha, Sachin Kumar, Li Lucy, Xinxi Lyu, Nathan Lambert, Ian Magnusson, Jacob Morrison, Niklas Muennigh...

    2024

  49. [62]

    SentAlign: Accurate and Scalable Sentence Alignment

    Steinþór Steingrímsson, Hrafn Loftsson, and Andy Way. SentAlign: Accurate and Scalable Sentence Alignment . In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing: System Demonstrations, Singapore, Singapore, December 2023. Association for Co...

    2023

  50. [63]

    Jochre 3 and the yiddish ocr corpus

    Assaf Urieli, Amber Clooney, Michelle Sigiel, and Grisha Leyfer. Jochre 3 and the yiddish ocr corpus. arXiv preprint arXiv:2501.08442, 2025

    2025 arXiv

  51. [64]

    Copyright Office

    U.S. Copyright Office . Copyright and artificial intelligence: Part 3 -- generative ai training. Report, U.S. Copyright Office, May 2025. URL https://www.copyright.gov/ai/

    2025

  52. [65]

    Reyd-the first yiddish text-to-speech dataset and system

    Jacob Webber, Samuel K Lo, and Isaac L Bleaman. Reyd-the first yiddish text-to-speech dataset and system. In INTERSPEECH, pp.\ 2363--2367, 2022

    2022

  53. [66]

    The bitter lesson learned from 2,000+ multilingual benchmarks, 2025

    Minghao Wu, Weixuan Wang, Sinuo Liu, Huifeng Yin, Xintong Wang, Yu Zhao, Chenyang Lyu, Longyue Wang, Weihua Luo, and Kaifu Zhang. The bitter lesson learned from 2,000+ multilingual benchmarks, 2025. URL https://arxiv.org/abs/2504.15521

    2025 arXiv

  54. [67]

    m T 5: A massively multilingual pre-trained text-to-text transformer

    Linting Xue, Noah Constant, Adam Roberts, Mihir Kale, Rami Al-Rfou, Aditya Siddhant, Aditya Barua, and Colin Raffel. m T 5: A massively multilingual pre-trained text-to-text transformer. In Kristina Toutanova, Anna Rumshisky, Luke Zettlemoyer, Dilek Hakkani-Tur, Iz Beltagy, St...

    2021

  55. [68]

    Qwen3 technical report

    An Yang, Anfeng Li, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Gao, Chengen Huang, Chenxu Lv, et al. Qwen3 technical report. arXiv preprint arXiv:2505.09388, 2025

    2025 arXiv

  56. [69]

    SnakModel : Lessons learned from training an open Danish large language model

    Mike Zhang, Max M \"u ller-Eberstein, Elisa Bassignana, and Rob van der Goot. SnakModel : Lessons learned from training an open Danish large language model. In Richard Johansson and Sara Stymne (eds.), Proceedings of the Joint 25th Nordic Conference on Computational Linguistic...

    2025

Pith tools

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