Pith. sign in

REVIEW 1 major objections 1 minor 1 cited by

Multilingual Contextualization of Large Language Models for Document-Level Machine Translation

T0 review · 1 major / 1 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read Fine-tuning on DocBlocks lets small open models beat 70B and GPT-4o at document-level translation.

desk verdict A useful, honestly reported doc-level SFT recipe with a real but fixable gap: the paper never states that the IWSLT2017 and GuoFeng test documents were excluded from DocBlocks, so the headline gains should be treated as provisional until a decontamination check is done. read the letter →

arxiv 2504.12140 v2 pith:JM7GCQRC submitted 2025-04-16 cs.CL

Miguel Moura Ramos , Patrick Fernandes , Sweta Agrawal , André F. T. Martins This is my paper · ORCID
classification cs.CL
keywords document-levelmachinetranslationlargelanguagemodelssupervisedfine-tuningcontext-awarepromptingparallelcorpuscurationdiscoursecoherencechunkeddecoding
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper tries to establish that document-level machine translation does not require larger models or multi-agent pipelines: targeted fine-tuning of existing instruction-tuned LLM translators on a curated corpus of full documents and contextual chunks is enough. It introduces DocBlocks, a cleaned multilingual parallel corpus drawn from news, TED talks, parliamentary proceedings, and novels, and trains three 7B-9B open models on a mix of document-to-document, context-windowed chunk, and sentence-level instructions. If the paper is right, small open models can produce coherent whole-document translations that beat 70B-parameter prompting baselines and GPT-4o on a discourse-heavy benchmark, while retaining sentence-level translation quality.

What carries the argument

The central object is the DocBlocks training corpus and its instruction formats. DocBlocks pairs full documents from five domains (News Commentary, IWSLT, Europarl, GuoFeng, BWB) with document-level alignment, quality filtering, deduplication, and length-based cleaning. Two training techniques carry the argument: MRD2D splits each document into k parts with k in {1, 2, 4} to create multi-resolution document-to-document examples, and CAPT adds a context window of up to three previously translated chunks to chunk-level prompts, alongside standalone sentence-level examples. The loss is computed only on target tokens, and the same trained model can be decoded whole-document or chunk-by-chunk.

What would settle it

Run exact and near-duplicate detection between the IWSLT2017 and GuoFeng test documents and the released DocBlocks training corpus; if any test document, chunk, or near-duplicate appears in training, or if retraining after removing these near-duplicates erases the reported d-BLEU gains, the claim of document-level generalization is not supported.

Watch

Extended reading notes

Core claim

The paper's central claim is that a single supervised fine-tuning pass over multi-granular document-level data turns sentence-level LLM translators into document-level translators. The resulting DocMT-LLMs outperform their untuned counterparts and, despite being 7B-9B models, beat Llama-3.3-70B-Instruct, Qwen2.5-72B-Instruct, and GPT-4o on GuoFeng zh→en and remain competitive on IWSLT2017, while keeping sentence-level COMET within about 0.5 points of the base models. The authors attribute this to training on a mixture of document-to-document, context-windowed chunk, and sentence-level formats, which teaches the model to use surrounding context during decoding without sacrificing single-sentence ability.

Load-bearing premise

The load-bearing premise is that the IWSLT2017 and GuoFeng test documents are not present in DocBlocks, since the corpus is built from those same collections and the paper never states that the evaluation documents were excluded.

Editorial extensions

If this is right

  • A 7B or 9B open model can serve as a document-level translation engine that outperforms 70B-class prompting models on discourse-rich literary translation and matches them on TED-talk documents.
  • The same model can decode whole documents in one pass or in parallel chunks, so users can trade quality for throughput; standard chunking reaches 392.45 tokens/s versus 204.22 for document-to-document.
  • Document-level training on a curated corpus does not erase sentence-level translation ability; COMET drops are around 0.5 points, and Qwen2.5-7B improves at sentence level after training on DocBlocks.
  • Agent-based translation pipelines and prompt-only context engineering are not necessary for coherent document translation; direct fine-tuning plus simple context windows matches or beats them.

Reading between the lines

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

  • Beyond the paper: if DocBlocks is later shown to contain the evaluation documents or near-duplicates, the reported gains would mostly reflect domain adaptation rather than generic document-level capability; testing on held-out domains such as legal or medical documents would settle this.
  • Beyond the paper: the same multi-granular recipe of full documents, k-way chunks, and windowed context may transfer to other instruction-tuned multilingual LLMs and to other discourse-sensitive tasks such as long-document summarization.
  • Beyond the paper: the comparison with agent-based methods suggests that hierarchical memory agents buy little once the base model is document-trained; a direct study of where agent memory helps, such as very long novels with recurring characters, would be a natural next test.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

1 major / 1 minor

Summary. The paper introduces DOCBLOCKS, a curated document-level parallel corpus, and a fine-tuning recipe that adapts sentence-level LLM translators (TOWER, EuroLLM, Qwen2.5) for document-level translation. The trained DocMT-LLMs are evaluated on GuoFeng and IWSLT2017 with document-level BLEU/COMET and LTCR, and on sentence-level benchmarks to check for catastrophic forgetting. The paper reports that the compact DocMT-LLMs surpass much larger prompting baselines on GuoFeng, are competitive on IWSLT2017, support chunked decoding with or without context and quality-aware reranking, and preserve sentence-level COMET within about 0.5 points.

Significance. If the central claims hold, the paper makes a useful contribution: it releases a document-level training corpus, shows that multi-granular instruction formats transfer across three base LLMs, and systematically contrasts fine-tuning with prompting and agent-based decoding. The ablations in Table 6 and the sentence-level stability results in Table 5 are valuable and clearly presented. The central empirical result, however, depends on the absence of training/evaluation overlap between DOCBLOCKS and the IWSLT2017 and GuoFeng test sets, which the paper does not currently establish; until that is resolved, the headline gains over much larger models cannot be taken at face value.

major comments (1)
  1. [§3.1, §3.2] The discourse-quality claim ('coherent, high-quality document-level translations') rests entirely on automatic metrics (d-BLEU, d-COMET via SLIDE, LTCR), and the paper itself notes that no universally accepted document-level metric exists. Given that this is the paper's central contribution, the absence of any human evaluation of discourse coherence, terminology consistency, or anaphora makes the qualitative claim underdetermined. A small targeted human evaluation on GuoFeng and a subset of IWSLT2017 would substantially strengthen the conclusion.
minor comments (1)
  1. [§3.1 Evaluation] The sentence 'Sentence alignments are first computed using bleualign' does not cite the tool; please add the appropriate reference and a brief description of the alignment procedure.

Circularity Check

1 steps flagged · score 4.0 of 10

DocBlocks is built from the same IWSLT2016/2017 and GuoFeng source corpora used for document-level evaluation, and the paper never states that benchmark test documents were excluded, so headline gains may reflect training/evaluation overlap rather than generalization.

  1. fitted input called prediction [Section 2.1 (Figure 2) and Section 3.1 (Datasets)]
    "The DOCBLOCKS corpus is constructed from a collection of publicly available document-level datasets, selected to represent a broad range of document types and content domains... The IWSLT (Cettolo et al., 2017) corpus, based on TED Talk transcripts, provides examples of conversational and spoken language... The BWB ... and GuoFeng Webnovel (Wang et al., 2023b) corpora were included for their expert translations of Chinese novels and web fiction... For document-level MT, we use IWSLT2017 (Cettolo et al., 2017) and GuoFeng (Wang et al., 2023b) test sets."

    DocBlocks is constructed from IWSLT2016/2017 and GuoFeng documents (Figure 2: 29.7K and 18.2K filtered documents), and the document-level evaluation is exactly on IWSLT2017 and GuoFeng test sets. The curation pipeline describes Bicleaner/CometKiwi filtering, langid checks, length filters, and deduplication, but it never states that benchmark test documents or near-duplicates were excluded from training. If the test documents are present in the source corpora, Tables 2 and 3 measure memorization of the evaluation corpus rather than document-level generalization; the reported gains (e.g., GuoFeng d-BLEU 37.57 vs 20.99 for Qwen2.5-72B) are then forced by the training input, not predicted.

full rationale

The paper is an empirical fine-tuning study with no fitted-parameter derivation chain, and most of the methodology is standard SFT evaluated with external metrics; the sentence-level benchmarks (FLORES-200, WMT23, TICO-19) are outside DocBlocks and provide independent support that document-level training does not destroy sentence-level quality. The central document-level claim, however, rests on an unstated premise: that the IWSLT2017 and GuoFeng test sets are disjoint from DocBlocks, which is built from IWSLT2016/2017 and GuoFeng. The paper never provides a decontamination statement, so the headline Doc2Doc and chunking gains over 70B/72B prompting baselines are not shown to be predictions rather than memorized training input. This is a partial circularity/validity risk rather than a fully demonstrated reduction; the self-citations to Tower, EuroLLM, and TowerBlocks are not load-bearing because those are public checkpoints and datasets used as base components, not as justifications for the outcome. Score 4 reflects one central evaluation whose independence from the training input is unverified.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The central claim rests on curated data and metric-based evaluation rather than on a derivation. The main unstated burden is the train/test overlap assumption: DocBlocks is sourced from IWSLT2016/2017 and GuoFeng, and the same corpus families provide the test sets, with no exclusion statement. The free parameters are hand-chosen curation and training-mix choices that affect the final models; the axioms are the metric-validity and baseline-comparability assumptions that underpin every headline number.

free parameters (6)
  • Bicleaner filtering threshold = 0.5
    Hand-chosen threshold in the DocBlocks cleaning pipeline (Section 2.1); documents with over 20% of sentences below threshold are removed.
  • CometKiwi-23 filtering threshold = 0.65
    Hand-chosen quality-estimation threshold in Section 2.1 that determines which documents remain in DocBlocks.
  • Additional cleaning thresholds = 50-word minimum, 10% identical chars, 1.3 length ratio
    Hand-chosen filters in Section 2.1 that shape DocBlocks composition and therefore the trained models.
  • Context window size N = 3 previous chunks
    Chosen following Wu et al. 2024, used in CAPT training and at inference (Section 2.2, Table 1).
  • Sentence-level share in DocBlocks = 10%
    Selected from the ablation in Figure 7 as the best balance between sentence-level and document-level quality.
  • MRD2D split counts k = k in {1, 2, 4}
    Applied to IWSLT, Europarl, and GuoFeng (Section 2.1) to generate multi-resolution training examples.
assumptions (5)
  • domain assumption COMET, d-COMET, and d-BLEU are adequate proxies for document-level translation quality.
    The paper acknowledges these metrics have known limitations for coherence and discourse (Section 3.1), yet all headline comparisons are metric-based and no human evaluation is reported.
  • domain assumption The IWSLT2017 and GuoFeng test sets are uncontaminated by DocBlocks training data.
    DocBlocks is built from IWSLT2016/2017 and GuoFeng corpora (Section 2.1), and the same families provide the test sets (Section 3.1); no decontamination statement is given. If false, headline gains vs baselines are inflated.
  • domain assumption Quality-aware MBR with CONTEXT-COMET selects better document translations.
    Used in Section 2.3 for quality-aware chunking; the paper also notes such metrics may not fully capture document-level quality.
  • domain assumption Prior-work baselines (TRANSAGENTS, DELTA, HW-TSC) with different base models (GPT-4 Turbo, GPT-4o mini) are comparable points of comparison.
    Figure 6 compares DocMT-7B models against agent baselines running on much larger closed models; the comparison conflates method with backbone.
  • standard math Standard transformer fine-tuning assumptions (cross-entropy loss, packing, bf16 mixed precision) carry over from sentence-level to document-level SFT.
    Assumed throughout Section 2.2 and Appendix A.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Multilingual Contextualization of Large Language Models for Document-Level Machine Translation." pith.science (2026). https://pith.science/paper/JM7GCQRC

@misc{pith2026250412140,
  author       = {Pith},
  title        = {Pith review of: Multilingual Contextualization of Large Language Models for Document-Level Machine Translation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JM7GCQRC}},
  note         = {Machine review of arXiv:2504.12140}
}
read the original abstract

Large language models (LLMs) have demonstrated strong performance in sentence-level machine translation, but scaling to document-level translation remains challenging, particularly in modeling long-range dependencies and discourse phenomena across sentences and paragraphs. In this work, we propose a method to improve LLM-based long-document translation through targeted fine-tuning on high-quality document-level data, which we curate and introduce as DocBlocks. Our approach supports multiple translation paradigms, including direct document-to-document and chunk-level translation, by integrating instructions both with and without surrounding context. This enables models to better capture cross-sentence dependencies while maintaining strong sentence-level translation performance. Experimental results show that incorporating multiple translation paradigms improves document-level translation quality and inference speed compared to prompting and agent-based methods.

Figures

Figures reproduced from arXiv: 2504.12140 by the authors.

Figure 1
Figure 1. Illustration of our approach for adapting an LLM to document translation. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Statistics of datasets used to create DOCBLOCKS, including domain, corpus name, number of documents |D|, sentences |S|, words |W|, and average document length |W|/|D| as a proxy for discourse complexity (Wang et al., 2023a). We also illustrate the distribution of language pairs across the filtered datasets, highlighting their multilingual composition. well-structured written content from the journalistic domain. The… view at source ↗
Figure 3
Figure 3. d-BLEU (top) and d-COMET (bottom) scores for the GuoFeng (left), IWSLT2017 [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (3 more)
Figure 5
Figure 5. Figure 5: DOCMT-TOWERINSTRUCT-7B performance on GuoFeng using chunking (C), contextual chunking (CC), and quality￾aware chunking (QAC), which combines contextual chunking with paragraph-level quality metrics. Inference Method Chunking Contextual Chunking Quality-aware Chunking D…
Figure 6
Figure 6. Figure 6: d-BLEU scores on the Guofeng test set, with chapter translations concatenated into a single document (Wu et al., 2024b). The figure compares document-level translation quality across different paradigms: blue for Doc2Doc, green for Contextual Chunking, orange for the D…
Figure 7
Figure 7. Figure 7: Effect of varying sentence-level data in [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]

Discussion (0). Continue with ORCID 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

  1. DataRx: Missingness-Aware Sampling for Safer Large Language Model Task-Specific Fine-Tuning

    cs.CL 2026-08 conditional novelty 6.0 of 10

    A missingness-aware sampling method that selects safety-critical fine-tuning examples using hidden-representation gaps reduces attack success rates after task-specific fine-tuning.

Reference graph

Works this paper leans on

77 extracted references · 21 canonical work pages · cited by 1 Pith paper

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION format.date year duplicate empty "emp...

  2. [2]

    Alves , Jos \'e Pombal , Nuno M

    Duarte M. Alves , Jos \'e Pombal , Nuno M. Guerreiro , Pedro H. Martins , Jo \ a o Alves , Amin Farajian , Ben Peters , Ricardo Rei , Patrick Fernandes , Sweta Agrawal , Pierre Colombo , Jos \'e G. C. de Souza , and Andr \'e F. T. Martins . Tower: An Open Multilingual Large Language Model for Translation-Related Tasks . arXiv e-prints, art. arXiv:2402.177...

    arXiv doi:10.48550/arxiv.2402.17733 2024

  3. [3]

    TICO -19: the translation initiative for CO vid-19

    Antonios Anastasopoulos, Alessandro Cattelan, Zi-Yi Dou, Marcello Federico, Christian Federmann, Dmitriy Genzel, Franscisco Guzm \'a n, Junjie Hu, Macduff Hughes, Philipp Koehn, Rosie Lazar, Will Lewis, Graham Neubig, Mengmeng Niu, Alp \"O ktem, Eric Paquin, Grace Tang, and Sylwia Tur. TICO -19: the translation initiative for CO vid-19. In Karin Verspoor,...

    2020

  4. [4]

    Neural Machine Translation by Jointly Learning to Align and Translate

    Dzmitry Bahdanau , Kyunghyun Cho , and Yoshua Bengio . Neural Machine Translation by Jointly Learning to Align and Translate . arXiv e-prints, art. arXiv:1409.0473, September 2014. doi:10.48550/arXiv.1409.0473

    arXiv doi:10.48550/arxiv.1409.0473 2014

  5. [5]

    Qwen Technical Report

    Jinze Bai , Shuai Bai , Yunfei Chu , Zeyu Cui , Kai Dang , Xiaodong Deng , Yang Fan , Wenbin Ge , Yu Han , Fei Huang , Binyuan Hui , Luo Ji , Mei Li , Junyang Lin , Runji Lin , Dayiheng Liu , Gao Liu , Chengqiang Lu , Keming Lu , Jianxin Ma , Rui Men , Xingzhang Ren , Xuancheng Ren , Chuanqi Tan , Sinan Tan , Jianhong Tu , Peng Wang , Shijie Wang , Wei Wa...

    arXiv doi:10.48550/arxiv.2309.16609 2023

  6. [6]

    Marta Ba \ n \'o n, Pinzhen Chen, Barry Haddow, Kenneth Heafield, Hieu Hoang, Miquel Espl \`a -Gomis, Mikel L. Forcada, Amir Kamran, Faheem Kirefu, Philipp Koehn, Sergio Ortiz Rojas, Leopoldo Pla Sempere, Gema Ram \'i rez-S \'a nchez, Elsa Sarr \'i as, Marek Strelec, Brian Thompson, William Waites, Dion Wiggins, and Jaume Zaragoza. P ara C rawl: Web-scale...

    2020

  7. [7]

    Evaluating discourse phenomena in neural machine translation

    Rachel Bawden, Rico Sennrich, Alexandra Birch, and Barry Haddow. Evaluating discourse phenomena in neural machine translation. In Marilyn Walker, Heng Ji, and Amanda Stent (eds.), Proceedings of the 2018 Conference of the North A merican Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers) , pp.\ 13...

    doi:10.18653/v1/n18-1118 2018

  8. [8]

    On context span needed for machine translation evaluation

    Sheila Castilho, Maja Popovi \'c , and Andy Way. On context span needed for machine translation evaluation. 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, Asuncion Moreno, Jan Odijk, and Stelios Pipe...

    work page 2020

Show all 77 references
  1. [9]

    Overview of the IWSLT 2017 evaluation campaign

    Mauro Cettolo, Marcello Federico, Luisa Bentivogli, Jan Niehues, Sebastian St \"u ker, Katsuhito Sudoh, Koichiro Yoshino, and Christian Federmann. Overview of the IWSLT 2017 evaluation campaign. In Proceedings of the 14th International Conference on Spoken Language Translation...

    2017

  2. [10]

    WMT24++: Expanding the Language Coverage of WMT24 to 55 Languages & Dialects

    Daniel Deutsch , Eleftheria Briakou , Isaac Caswell , Mara Finkelstein , Rebecca Galor , Juraj Juraska , Geza Kovacs , Alison Lui , Ricardo Rei , Jason Riesa , Shruti Rijhwani , Parker Riley , Elizabeth Salesky , Firas Trabelsi , Stephanie Winkler , Biao Zhang , and Markus Fre...

    2025 doi arXiv

  3. [11]

    Sampling-based approximations to minimum B ayes risk decoding for neural machine translation

    Bryan Eikema and Wilker Aziz. Sampling-based approximations to minimum B ayes risk decoding for neural machine translation. In Yoav Goldberg, Zornitsa Kozareva, and Yue Zhang (eds.), Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pp.\ 1...

    2022 doi

  4. [12]

    Patrick Fernandes, Kayo Yin, Graham Neubig, and Andr \'e F. T. Martins. Measuring and increasing context usage in context-aware machine translation. In Chengqing Zong, Fei Xia, Wenjie Li, and Roberto Navigli (eds.), Proceedings of the 59th Annual Meeting of the Association for...

    2021 doi

  5. [13]

    Experts, errors, and context: A large-scale study of human evaluation for machine translation

    Markus Freitag, George Foster, David Grangier, Viresh Ratnakar, Qijun Tan, and Wolfgang Macherey. Experts, errors, and context: A large-scale study of human evaluation for machine translation. Transactions of the Association for Computational Linguistics, 9: 0 1460--1474, 2021...

    2021 doi

  6. [14]

    Neural poetry translation

    Marjan Ghazvininejad, Yejin Choi, and Kevin Knight. Neural poetry translation. In Marilyn Walker, Heng Ji, and Amanda Stent (eds.), Proceedings of the 2018 Conference of the North A merican Chapter of the Association for Computational Linguistics: Human Language Technologies, ...

    2018 doi

  7. [15]

    Document-level machine translation evaluation with gist consistency and text cohesion

    Zhengxian Gong, Min Zhang, and Guodong Zhou. Document-level machine translation evaluation with gist consistency and text cohesion. In Bonnie Webber, Marine Carpuat, Andrei Popescu-Belis, and Christian Hardmeier (eds.), Proceedings of the Second Workshop on Discourse in Machin...

    2015 doi

  8. [16]

    Aaron Grattafiori, Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Alex Vaughan, Amy Yang, Angela Fan, Anirudh Goyal, Anthony Hartshorn, Aobo Yang, Archi Mitra, Archie Sravankumar, Artem Korenev, Art...

    2024 arXiv

  9. [17]

    Exploring human-like translation strategy with large language models

    Zhiwei He, Tian Liang, Wenxiang Jiao, Zhuosheng Zhang, Yujiu Yang, Rui Wang, Zhaopeng Tu, Shuming Shi, and Xing Wang. Exploring human-like translation strategy with large language models. Transactions of the Association for Computational Linguistics, 12: 0 229--246, 2024. doi:...

    2024 doi

  10. [18]

    How Good Are GPT Models at Machine Translation? A Comprehensive Evaluation

    Amr Hendy , Mohamed Abdelrehim , Amr Sharaf , Vikas Raunak , Mohamed Gabr , Hitokazu Matsushita , Young Jin Kim , Mohamed Afify , and Hany Hassan Awadalla . How Good Are GPT Models at Machine Translation? A Comprehensive Evaluation . arXiv e-prints, art. arXiv:2302.09210, Febr...

    2023 doi arXiv

  11. [19]

    The Curious Case of Neural Text Degeneration

    Ari Holtzman , Jan Buys , Li Du , Maxwell Forbes , and Yejin Choi . The Curious Case of Neural Text Degeneration . arXiv e-prints, art. arXiv:1904.09751, April 2019. doi:10.48550/arXiv.1904.09751

    1904 doi arXiv

  12. [20]

    Exploring discourse structure in document-level machine translation

    Xinyu Hu and Xiaojun Wan. Exploring discourse structure in document-level machine translation. In Houda Bouamor, Juan Pino, and Kalika Bali (eds.), Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pp.\ 13889--13902, Singapore, December 20...

    2023 doi

  13. [21]

    Diving deep into context-aware neural machine translation

    Jingjing Huo, Christian Herold, Yingbo Gao, Leonard Dahlmann, Shahram Khadivi, and Hermann Ney. Diving deep into context-aware neural machine translation. In Lo \" c Barrault, Ond r ej Bojar, Fethi Bougares, Rajen Chatterjee, Marta R. Costa-juss \`a , Christian Federmann, Mark...

    2020

  14. [22]

    Albert Q. Jiang , Alexandre Sablayrolles , Arthur Mensch , Chris Bamford , Devendra Singh Chaplot , Diego de las Casas , Florian Bressand , Gianna Lengyel , Guillaume Lample , Lucile Saulnier , L \'e lio Renard Lavaud , Marie-Anne Lachaux , Pierre Stock , Teven Le Scao , Thiba...

    doi arXiv

  15. [23]

    BlonDe : An automatic evaluation metric for document-level machine translation

    Yuchen Jiang, Tianyu Liu, Shuming Ma, Dongdong Zhang, Jian Yang, Haoyang Huang, Rico Sennrich, Ryan Cotterell, Mrinmaya Sachan, and Ming Zhou. BlonDe : An automatic evaluation metric for document-level machine translation. In Marine Carpuat, Marie-Catherine de Marneffe, and Iv...

    2022

  16. [24]

    Large language models effectively leverage document-level context for literary translation, but critical errors persist

    Marzena Karpinska and Mohit Iyyer. Large language models effectively leverage document-level context for literary translation, but critical errors persist. In Philipp Koehn, Barry Haddow, Tom Kocmi, and Christof Monz (eds.), Proceedings of the Eighth Conference on Machine Tran...

    2023 doi

  17. [25]

    Kingma and Jimmy Ba

    Diederik P. Kingma and Jimmy Ba . Adam: A Method for Stochastic Optimization . arXiv e-prints, art. arXiv:1412.6980, December 2014. doi:10.48550/arXiv.1412.6980

    2014 doi arXiv

  18. [26]

    Findings of the 2023 conference on machine translation ( WMT 23): LLM s are here but not quite there yet

    Tom Kocmi, Eleftherios Avramidis, Rachel Bawden, Ond r ej Bojar, Anton Dvorkovich, Christian Federmann, Mark Fishel, Markus Freitag, Thamme Gowda, Roman Grundkiewicz, Barry Haddow, Philipp Koehn, Benjamin Marie, Christof Monz, Makoto Morishita, Kenton Murray, Makoto Nagata, To...

    2023

  19. [27]

    Findings of the WMT 24 general machine translation shared task: The LLM era is here but MT is not solved yet

    Tom Kocmi, Eleftherios Avramidis, Rachel Bawden, Ond r ej Bojar, Anton Dvorkovich, Christian Federmann, Mark Fishel, Markus Freitag, Thamme Gowda, Roman Grundkiewicz, Barry Haddow, Marzena Karpinska, Philipp Koehn, Benjamin Marie, Christof Monz, Kenton Murray, Masaaki Nagata, ...

    2024

  20. [28]

    E uroparl: A parallel corpus for statistical machine translation

    Philipp Koehn. E uroparl: A parallel corpus for statistical machine translation. In Proceedings of Machine Translation Summit X: Papers, pp.\ 79--86, Phuket, Thailand, September 13-15 2005. URL https://aclanthology.org/2005.mtsummit-papers.11/

    2005

  21. [29]

    Minimum B ayes-risk decoding for statistical machine translation

    Shankar Kumar and William Byrne. Minimum B ayes-risk decoding for statistical machine translation. In Proceedings of the Human Language Technology Conference of the North A merican Chapter of the Association for Computational Linguistics: HLT - NAACL 2004 , pp.\ 169--176, Bost...

    2004

  22. [30]

    Has machine translation achieved human parity? a case for document-level evaluation

    Samuel L \"a ubli, Rico Sennrich, and Martin Volk. Has machine translation achieved human parity? a case for document-level evaluation. In Ellen Riloff, David Chiang, Julia Hockenmaier, and Jun ' ichi Tsujii (eds.), Proceedings of the 2018 Conference on Empirical Methods in Na...

    2018 doi

  23. [31]

    Enhancing document-level translation of large language model via translation mixed-instructions, 2024

    Yachao Li, Junhui Li, Jing Jiang, and Min Zhang. Enhancing document-level translation of large language model via translation mixed-instructions, 2024. URL https://arxiv.org/abs/2401.08088

    2024 arXiv

  24. [32]

    O pen S ubtitles2016: Extracting large parallel corpora from movie and TV subtitles

    Pierre Lison and J \"o rg Tiedemann. O pen S ubtitles2016: Extracting large parallel corpora from movie and TV subtitles. In Nicoletta Calzolari, Khalid Choukri, Thierry Declerck, Sara Goggi, Marko Grobelnik, Bente Maegaard, Joseph Mariani, Helene Mazo, Asuncion Moreno, Jan Od...

    2016

  25. [33]

    Corpora for document-level neural machine translation

    Siyou Liu and Xiaojun Zhang. Corpora for document-level neural machine translation. 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 \...

    2020

  26. [34]

    Multilingual denoising pre-training for neural machine translation

    Yinhan Liu, Jiatao Gu, Naman Goyal, Xian Li, Sergey Edunov, Marjan Ghazvininejad, Mike Lewis, and Luke Zettlemoyer. Multilingual denoising pre-training for neural machine translation. Transactions of the Association for Computational Linguistics, 8: 0 726--742, 2020. doi:10.11...

    2020 doi

  27. [35]

    langid.py: An off-the-shelf language identification tool

    Marco Lui and Timothy Baldwin. langid.py: An off-the-shelf language identification tool. In Min Zhang (ed.), Proceedings of the ACL 2012 System Demonstrations , pp.\ 25--30, Jeju Island, Korea, July 2012. Association for Computational Linguistics. URL https://aclanthology.org/P12-3005

    2012

  28. [36]

    Encouraging lexical translation consistency for document-level neural machine translation

    Xinglin Lyu, Junhui Li, Zhengxian Gong, and Min Zhang. Encouraging lexical translation consistency for document-level neural machine translation. In Marie-Francine Moens, Xuanjing Huang, Lucia Specia, and Scott Wen-tau Yih (eds.), Proceedings of the 2021 Conference on Empirica...

    2021 doi

  29. [37]

    Using whole document context in neural machine translation

    Valentin Mac \'e and Christophe Servan. Using whole document context in neural machine translation. In Jan Niehues, Rolando Cattoni, Sebastian St \"u ker, Matteo Negri, Marco Turchi, Thanh-Le Ha, Elizabeth Salesky, Ramon Sanabria, Loic Barrault, Lucia Specia, and Marcello Fede...

    2019

  30. [38]

    Guerreiro , Ricardo Rei , Duarte M

    Pedro Henrique Martins , Patrick Fernandes , Jo \ a o Alves , Nuno M. Guerreiro , Ricardo Rei , Duarte M. Alves , Jos \'e Pombal , Amin Farajian , Manuel Faysse , Mateusz Klimaszewski , Pierre Colombo , Barry Haddow , Jos \'e G. C. de Souza , Alexandra Birch , and Andr \'e F. ...

    2024 doi arXiv

  31. [39]

    A survey on document-level neural machine translation: Methods and evaluation

    Sameen Maruf, Fahimeh Saleh, and Gholamreza Haffari. A survey on document-level neural machine translation: Methods and evaluation. ACM Comput. Surv., 54 0 (2), March 2021. ISSN 0360-0300. doi:10.1145/3441691. URL https://doi.org/10.1145/3441691

    2021 doi

  32. [40]

    Document-level neural machine translation with hierarchical attention networks

    Lesly Miculicich, Dhananjay Ram, Nikolaos Pappas, and James Henderson. Document-level neural machine translation with hierarchical attention networks. In Ellen Riloff, David Chiang, Julia Hockenmaier, and Jun ' ichi Tsujii (eds.), Proceedings of the 2018 Conference on Empirica...

    2018 doi

  33. [41]

    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 ...

    2022 doi arXiv

  34. [42]

    URL https://github.com/openai/openai-python/blob/release-v0.28.1/chatml.md

    OpenAI , 2023. URL https://github.com/openai/openai-python/blob/release-v0.28.1/chatml.md

    2023

  35. [43]

    OpenAI , Josh Achiam , Steven Adler , Sandhini Agarwal , Lama Ahmad , Ilge Akkaya , Florencia Leoni Aleman , Diogo Almeida , Janko Altenschmidt , Sam Altman , Shyamal Anadkat , Red Avila , Igor Babuschkin , Suchir Balaji , Valerie Balcom , Paul Baltescu , Haiming Bao , Mohamma...

    2023 doi arXiv

  36. [44]

    Training language models to follow instructions with human feedback

    Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, John Schulman, Jacob Hilton, Fraser Kelton, Luke Miller, Maddie Simens, Amanda Askell, Peter Welinder, Paul F Christiano, Jan Leike, a...

    2022

  37. [45]

    B leu: a method for automatic evaluation of machine translation

    Kishore Papineni, Salim Roukos, Todd Ward, and Wei-Jing Zhu. B leu: a method for automatic evaluation of machine translation. In Pierre Isabelle, Eugene Charniak, and Dekang Lin (eds.), Proceedings of the 40th Annual Meeting of the Association for Computational Linguistics, pp...

    2002

  38. [46]

    Exploring the limits of transfer learning with a unified text-to-text transformer

    Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, and Peter J Liu. Exploring the limits of transfer learning with a unified text-to-text transformer. Journal of machine learning research, 21 0 (140): 0 1--67, 2020

    2020

  39. [47]

    Bifixer and bicleaner: two open-source tools to clean your parallel data

    Gema Ram \'i rez-S \'a nchez, Jaume Zaragoza-Bernabeu, Marta Ba \ n \'o n, and Sergio Ortiz Rojas. Bifixer and bicleaner: two open-source tools to clean your parallel data. In Andr \'e Martins, Helena Moniz, Sara Fumega, Bruno Martins, Fernando Batista, Luisa Coheur, Carla Par...

    2020

  40. [48]

    SLIDE : Reference-free evaluation for machine translation using a sliding document window

    Vikas Raunak, Tom Kocmi, and Matt Post. SLIDE : Reference-free evaluation for machine translation using a sliding document window. In Kevin Duh, Helena Gomez, and Steven Bethard (eds.), Proceedings of the 2024 Conference of the North American Chapter of the Association for Com...

    2024 doi

  41. [49]

    Ricardo Rei, Jos \'e G. C. de Souza, Duarte Alves, Chrysoula Zerva, Ana C Farinha, Taisiya Glushkova, Alon Lavie, Luisa Coheur, and Andr \'e F. T. Martins. COMET -22: Unbabel- IST 2022 submission for the metrics shared task. In Philipp Koehn, Lo \" c Barrault, Ond r ej Bojar, ...

    2022

  42. [50]

    Guerreiro , Jos \'e Pombal , Daan van Stigt , Marcos Treviso , Luisa Coheur , Jos \'e G

    Ricardo Rei , Nuno M. Guerreiro , Jos \'e Pombal , Daan van Stigt , Marcos Treviso , Luisa Coheur , Jos \'e G. C. de Souza , and Andr \'e F. T. Martins . Scaling up COMETKIWI: Unbabel-IST 2023 Submission for the Quality Estimation Shared Task . arXiv e-prints, art. arXiv:2309....

    2023 doi arXiv

  43. [51]

    Victor Sanh , Albert Webson , Colin Raffel , Stephen H. Bach , Lintang Sutawika , Zaid Alyafeai , Antoine Chaffin , Arnaud Stiegler , Teven Le Scao , Arun Raja , Manan Dey , M Saiful Bari , Canwen Xu , Urmish Thakker , Shanya Sharma Sharma , Eliza Szczechla , Taewoon Kim , Gun...

    doi arXiv

  44. [52]

    RoFormer: Enhanced Transformer with Rotary Position Embedding

    Jianlin Su , Yu Lu , Shengfeng Pan , Ahmed Murtadha , Bo Wen , and Yunfeng Liu . RoFormer: Enhanced Transformer with Rotary Position Embedding . arXiv e-prints, art. arXiv:2104.09864, April 2021. doi:10.48550/arXiv.2104.09864

    2021 doi arXiv

  45. [53]

    Rethinking document-level neural machine translation

    Zewei Sun, Mingxuan Wang, Hao Zhou, Chengqi Zhao, Shujian Huang, Jiajun Chen, and Lei Li. Rethinking document-level neural machine translation. In Smaranda Muresan, Preslav Nakov, and Aline Villavicencio (eds.), Findings of the Association for Computational Linguistics: ACL 20...

    2022 doi

  46. [54]

    Sequence to sequence learning with neural networks

    Ilya Sutskever, Oriol Vinyals, and Quoc V Le. Sequence to sequence learning with neural networks. In Z. Ghahramani, M. Welling, C. Cortes, N. Lawrence, and K.Q. Weinberger (eds.), Advances in Neural Information Processing Systems, volume 27. Curran Associates, Inc., 2014. URL ...

    2014

  47. [55]

    Reassessing claims of human parity and super-human performance in machine translation at WMT 2019

    Antonio Toral. Reassessing claims of human parity and super-human performance in machine translation at WMT 2019. In Andr \'e Martins, Helena Moniz, Sara Fumega, Bruno Martins, Fernando Batista, Luisa Coheur, Carla Parra, Isabel Trancoso, Marco Turchi, Arianna Bisazza, Joss Mo...

    2019

  48. [56]

    What Level of Quality can Neural Machine Translation Attain on Literary Text? arXiv e-prints, art

    Antonio Toral and Andy Way . What Level of Quality can Neural Machine Translation Attain on Literary Text? arXiv e-prints, art. arXiv:1801.04962, January 2018. doi:10.48550/arXiv.1801.04962

    2018 doi arXiv

  49. [57]

    LLaMA: Open and Efficient Foundation Language Models

    Hugo Touvron , Thibaut Lavril , Gautier Izacard , Xavier Martinet , Marie-Anne Lachaux , Timoth \'e e Lacroix , Baptiste Rozi \`e re , Naman Goyal , Eric Hambro , Faisal Azhar , Aurelien Rodriguez , Armand Joulin , Edouard Grave , and Guillaume Lample . LLaMA: Open and Efficie...

    2023 doi arXiv

  50. [58]

    Attention is all you need

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, ukasz Kaiser, and Illia Polosukhin. Attention is all you need. In I. Guyon, U. Von Luxburg, S. Bengio, H. Wallach, R. Fergus, S. Vishwanathan, and R. Garnett (eds.), Advances in Neural Info...

    2017

  51. [59]

    Embarrassingly easy document-level MT metrics: How to convert any pretrained metric into a document-level metric

    Giorgos Vernikos, Brian Thompson, Prashant Mathur, and Marcello Federico. Embarrassingly easy document-level MT metrics: How to convert any pretrained metric into a document-level metric. In Philipp Koehn, Lo \" c Barrault, Ond r ej Bojar, Fethi Bougares, Rajen Chatterjee, Mar...

    2022

  52. [60]

    Context-aware neural machine translation learns anaphora resolution

    Elena Voita, Pavel Serdyukov, Rico Sennrich, and Ivan Titov. Context-aware neural machine translation learns anaphora resolution. In Iryna Gurevych and Yusuke Miyao (eds.), Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long ...

    2018 doi

  53. [61]

    Document-level machine translation with large language models

    Longyue Wang, Chenyang Lyu, Tianbo Ji, Zhirui Zhang, Dian Yu, Shuming Shi, and Zhaopeng Tu. Document-level machine translation with large language models. In Houda Bouamor, Juan Pino, and Kalika Bali (eds.), Proceedings of the 2023 Conference on Empirical Methods in Natural La...

    2023 doi

  54. [62]

    Findings of the WMT 2023 shared task on discourse-level literary translation: A fresh orb in the cosmos of LLM s

    Longyue Wang, Zhaopeng Tu, Yan Gu, Siyou Liu, Dian Yu, Qingsong Ma, Chenyang Lyu, Liting Zhou, Chao-Hong Liu, Yufeng Ma, Weiyu Chen, Yvette Graham, Bonnie Webber, Philipp Koehn, Andy Way, Yulin Yuan, and Shuming Shi. Findings of the WMT 2023 shared task on discourse-level lite...

    2023

  55. [63]

    Wong , Fandong Meng , Jie Zhou , and Min Zhang

    Yutong Wang , Jiali Zeng , Xuebo Liu , Derek F. Wong , Fandong Meng , Jie Zhou , and Min Zhang . DelTA: An Online Document-Level Translation Agent Based on Multi-Level Memory . arXiv e-prints, art. arXiv:2410.08143, October 2024. doi:10.48550/arXiv.2410.08143

    doi arXiv

  56. [64]

    Zhao , Kelvin Guu , Adams Wei Yu , Brian Lester , Nan Du , Andrew M

    Jason Wei , Maarten Bosma , Vincent Y. Zhao , Kelvin Guu , Adams Wei Yu , Brian Lester , Nan Du , Andrew M. Dai , and Quoc V. Le . Finetuned Language Models Are Zero-Shot Learners . arXiv e-prints, art. arXiv:2109.01652, September 2021. doi:10.48550/arXiv.2109.01652

    2021 doi arXiv

  57. [65]

    Billy T. M. Wong and Chunyu Kit. Extending machine translation evaluation metrics with lexical cohesion to document level. In Jun ' ichi Tsujii, James Henderson, and Marius Pa s ca (eds.), Proceedings of the 2012 Joint Conference on Empirical Methods in Natural Language Proces...

    2012

  58. [66]

    Document flattening: Beyond concatenating context for document-level neural machine translation

    Minghao Wu, George Foster, Lizhen Qu, and Gholamreza Haffari. Document flattening: Beyond concatenating context for document-level neural machine translation. In Andreas Vlachos and Isabelle Augenstein (eds.), Proceedings of the 17th Conference of the European Chapter of the A...

    2023 doi

  59. [67]

    Adapting Large Language Models for Document-Level Machine Translation

    Minghao Wu , Thuy-Trang Vu , Lizhen Qu , George Foster , and Gholamreza Haffari . Adapting Large Language Models for Document-Level Machine Translation . arXiv e-prints, art. arXiv:2401.06468, January 2024. doi:10.48550/arXiv.2401.06468

    2024 doi arXiv

  60. [68]

    Importance-aware data augmentation for document-level neural machine translation

    Minghao Wu, Yufei Wang, George Foster, Lizhen Qu, and Gholamreza Haffari. Importance-aware data augmentation for document-level neural machine translation. In Yvette Graham and Matthew Purver (eds.), Proceedings of the 18th Conference of the European Chapter of the Association...

    2024 doi

  61. [69]

    (perhaps) beyond human translation: Harnessing multi-agent collaboration for translating ultra-long literary texts, 2024 b

    Minghao Wu, Yulin Yuan, Gholamreza Haffari, and Longyue Wang. (perhaps) beyond human translation: Harnessing multi-agent collaboration for translating ultra-long literary texts, 2024 b . URL https://arxiv.org/abs/2405.11804

    2024 arXiv

  62. [70]

    Document-level consistency verification in machine translation

    Tong Xiao, Jingbo Zhu, Shujie Yao, and Hao Zhang. Document-level consistency verification in machine translation. In Proceedings of Machine Translation Summit XIII: Papers, Xiamen, China, September 19-23 2011. URL https://aclanthology.org/2011.mtsummit-papers.13

    2011

  63. [71]

    HW - TSC `s submissions to the WMT 23 discourse-level literary translation shared task

    Yuhao Xie, Zongyao Li, Zhanglin Wu, Daimeng Wei, Xiaoyu Chen, Zhiqiang Rao, Shaojun Li, Hengchao Shang, Jiaxin Guo, Lizhi Lei, Hao Yang, and Yanfei Jiang. HW - TSC `s submissions to the WMT 23 discourse-level literary translation shared task. In Philipp Koehn, Barry Haddow, To...

    2023 doi

  64. [72]

    Modeling Coherence for Discourse Neural Machine Translation

    Hao Xiong , Zhongjun He , Hua Wu , and Haifeng Wang . Modeling Coherence for Discourse Neural Machine Translation . arXiv e-prints, art. arXiv:1811.05683, November 2018. doi:10.48550/arXiv.1811.05683

    2018 doi arXiv

  65. [73]

    A Paradigm Shift in Machine Translation: Boosting Translation Performance of Large Language Models

    Haoran Xu , Young Jin Kim , Amr Sharaf , and Hany Hassan Awadalla . A Paradigm Shift in Machine Translation: Boosting Translation Performance of Large Language Models . arXiv e-prints, art. arXiv:2309.11674, September 2023. doi:10.48550/arXiv.2309.11674

    2023 doi arXiv

  66. [74]

    Prompting large language model for machine translation: A case study

    Biao Zhang, Barry Haddow, and Alexandra Birch. Prompting large language model for machine translation: A case study. In Andreas Krause, Emma Brunskill, Kyunghyun Cho, Barbara Engelhardt, Sivan Sabato, and Jonathan Scarlett (eds.), Proceedings of the 40th International Conferen...

  67. [75]

    Improving the transformer translation model with document-level context

    Jiacheng Zhang, Huanbo Luan, Maosong Sun, Feifei Zhai, Jingfang Xu, Min Zhang, and Yang Liu. Improving the transformer translation model with document-level context. In Ellen Riloff, David Chiang, Julia Hockenmaier, and Jun ' ichi Tsujii (eds.), Proceedings of the 2018 Confere...

    2018 doi

  68. [76]

    Long-short term masking transformer: A simple but effective baseline for document-level neural machine translation

    Pei Zhang, Boxing Chen, Niyu Ge, and Kai Fan. Long-short term masking transformer: A simple but effective baseline for document-level neural machine translation. In Bonnie Webber, Trevor Cohn, Yulan He, and Yang Liu (eds.), Proceedings of the 2020 Conference on Empirical Metho...

    2020 doi

  69. [77]

    The U nited N ations parallel corpus v1.0

    Micha Ziemski, Marcin Junczys-Dowmunt, and Bruno Pouliquen. The U nited N ations parallel corpus v1.0. In Nicoletta Calzolari, Khalid Choukri, Thierry Declerck, Sara Goggi, Marko Grobelnik, Bente Maegaard, Joseph Mariani, Helene Mazo, Asuncion Moreno, Jan Odijk, and Stelios Pi...

    2016

Pith tools

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