Pith. sign in

REVIEW 3 major objections 4 minor 36 references

Jointly Learning to Align and Translate with Transformer Models

T0 review · 3 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read A Transformer can be trained to output both translations and word alignments by adding an alignment cross-entropy loss on one attention head and giving that head the full target sentence; with statistical aligner labels it then surpasses…

desk verdict Solid empirical paper on joint alignment/translation with Transformers, but the headline AER gains are likely optimistic because λ and the alignment head layer were chosen on the same gold test alignments used for evaluation. read the letter →

arxiv 1909.02074 v1 pith:JMM6WYRD submitted 2019-09-04 cs.CL

Sarthak Garg , Stephan Peitz , Udhyakumar Nallasamy , Matthias Paulik This is my paper · ORCID
classification cs.CL
keywords wordalignmentTransformermulti-tasklearningneuralmachinetranslationattentionsupervisionerrorratefulltargetcontextGIZA++
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 single Transformer model can be trained to produce both high-quality translations and word alignments at the same time. The recipe is a multi-task loss: the usual translation objective plus a cross-entropy loss that forces one designated attention head in the penultimate decoder layer to match a discrete alignment distribution. The alignment head is given the full target sentence during the alignment forward pass, while translation remains autoregressive, and this context difference is what lets one model do both tasks well. The reported result is that the model beats previous neural alignment approaches and, when supervised with alignments from a statistical aligner, also beats that aligner on alignment error rate on three language pairs while BLEU stays essentially flat.

What carries the argument

The load-bearing mechanism is a single supervised attention head, called the alignment head, located in the penultimate encoder-decoder layer of the Transformer. Its attention probabilities are trained with a per-target cross-entropy loss against a normalized alignment distribution, scaled by lambda, alongside the standard translation NLL. Because the autoregressive mask hides the future target words, the paper runs a second, unmasked decoder forward pass so the alignment head conditions on the whole target sentence; this full-context conditioning is what lets the head learn to align the current word instead of predicting it. Symmetrization with the grow-diagonal heuristic and, for subword models, mapping BPE-level links back to word-level links complete the extraction pipeline.

What would settle it

Train the multi-task model on the same three datasets with the alignment-loss labels replaced by random permutations of the same alignments; if alignment error rate still drops to the reported levels, the alignment signal is not what drives the gain. A simpler check: ablate the unmasked second forward pass on German-English and see whether the full-context AER improvement (20.2 vs 25.4 on the alignment task) disappears.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central claim is that the Transformer's encoder-decoder attention can be turned into a reliable word aligner without degrading translation. The authors supervise one attention head--the alignment head--in the penultimate layer with a cross-entropy loss against a 0-1 alignment matrix converted to a distribution, add the translation NLL with weight lambda = 0.05, and run a second unmasked decoder forward pass so the alignment head sees the full target sentence. With only self-extracted layer-average attention labels, this multi-task model reaches AER 20.2/26.0/7.7 on German-English, Romanian-English, and English-French, clearly beating the 21.2/27.6/10.0 of the best previous external-label-free neural method. With GIZA++ labels, it reaches 16.0/23.1/4.6, better than GIZA++'s own 18.9/27.0/5.5, and on the WMT'18 English-German setup it reaches AER 17.5 and 19.8 for the two directions and 16.4 after symmetrization, versus GIZA++'s 19.6, with BLEU unchanged at about 33.

Load-bearing premise

The self-training variant assumes the alignments extracted from an ordinary Transformer's layer-averaged attention are reliable enough to serve as training labels, even though many of those labels are wrong (unsymmetrized AER 66.8% on the WMT setup); the GIZA++-supervised variant does not depend on this assumption.

Editorial extensions

If this is right

  • A single Transformer can serve as both translator and aligner, so downstream tasks that need word alignments (bilingual lexicons, markup transfer, terminology control) no longer require a separate statistical alignment pipeline.
  • Alignment quality is no longer a reason to prefer classical models: supervised by classical alignments, the neural model surpasses them, so better alignments can be obtained without sacrificing translation quality.
  • The full-context second pass is the decisive factor; restricting alignment to past target context leaves alignments far weaker (25.4 vs 20.2 AER on German-English).
  • The self-training path shows that noisy alignments from averaged attention can serve as seeds, so a Transformer can bootstrap alignment ability from its own attention.
  • Translation quality stays essentially unchanged in every configuration, indicating the alignment loss does not trade away BLEU.

Reading between the lines

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

  • Because the GIZA++-supervised model improves on its own labels, the same multi-task setup may refine any weak aligner's output: train a Transformer on noisy labels, then extract better labels from the trained model, iterating within a single training run as the paper notes as future work.
  • The penultimate-layer result suggests a general structural trend in Transformer encoders-decoders: lower layers build representations, upper layers consume them, and the layer just before the output is the natural place to read off alignment-like decisions; this may transfer to other sequence-to-sequence tasks where latent correspondences matter.
  • The full-context alignment head could be exported for non-autoregressive translation or for any application where the target string is already known (annotation transfer, post-editing, quality estimation), since it no longer needs causal masking.
  • A testable extension: if the alignment head were moved to the final layer or if all heads were supervised, the flexibility the paper credits to single-head supervision could be checked directly.
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

3 major / 4 minor

Summary. The paper proposes a multi-task training method for Transformer-based NMT that jointly optimizes the standard translation NLL loss and an alignment cross-entropy loss applied to one designated attention head in the penultimate decoder layer. Discrete alignments used as training labels come either from a 'layer average baseline' (averaging attention probabilities inside the penultimate layer of a regularly trained Transformer) or from GIZA++ IBM alignments. The alignment head is given full target-sentence context via a second, unmasked decoder forward pass. The authors evaluate alignment error rate on three language pairs (German→English, Romanian→English, English→French) with gold alignments, plus a WMT'18 English↔German 'align and translate' setup that reports both AER and BLEU. They report that the full-context multi-task variant outperforms Zenkel et al. (2019), and that GIZA++-supervised training significantly outperforms GIZA++ in AER without hurting BLEU.

Significance. If the reported results are unbiased, the paper would be a useful contribution: it extends guided-alignment training to Transformers, shows that full target context for the alignment head is beneficial, and provides an open-source implementation. The experimental coverage is broad (three alignment datasets plus a large WMT setup, comparison against several neural and statistical baselines), and the use of a paired significance test for the headline AER comparison is a strength. The main caveat is that several design choices, including the alignment loss weight and the alignment-head layer, were selected using the same gold test alignments on which the final AER numbers are reported, which makes the DeEn headline comparisons and the layer-selection rationale optimistically biased. The strongest result, GIZA++-supervised training, is also affected by this selection process on the German→English dataset, although the other two language pairs are less directly affected because λ was fixed from DeEn.

major comments (3)
  1. [§5.1.1, §5.4, Table 3, Table 4] The alignment task is explicitly described as serving 'to perform a hyperparameter search', and Section 5.4 states that the alignment loss weight λ was tuned by grid search on the German→English dataset. In addition, the decision to place the alignment head in the penultimate layer is justified by test-set AER values in Table 2, and the final AER numbers in Table 3 are computed on the same gold test alignments. This creates a selection bias for the DeEn headline comparison and for the architecture choice. Section 6 acknowledges that no alignment validation set exists, but that does not remove the bias; it only explains why it arose. Please provide a held-out or cross-validated estimate of the reported gains, for example by evaluating on a reserved subset of annotated sentences for model selection, or by reporting sensitivity of the final AER to λ and to the selected layer across all three language pairs. Without such a check, the abstract's claim of 'significantly better alignment accuracies compared to GIZA++' is not fully supported by the evidence as presented.
  2. [§4.2 and Table 4] The self-training variant relies on labels extracted from the layer average baseline, which on the WMT'18 setup has an AER of 66.8% with precision 32.0% and recall 34.6% before symmetrization. The claim that 'the high precision of the layer average baseline ensures that the multi-task model receives correct alignments' refers to the symmetrized precision (94.2%), but the labels are extracted before symmetrization, and a precision of 32.0% means the great majority of extracted links are wrong. The paper does not test whether the method is robust to the level of label noise in these self-generated alignments, for example by filtering low-confidence alignments or by ablating the amount of noise in the GIZA++-supervised setting. This matters for the self-training contribution, although it does not directly affect the GIZA++-supervised result.
  3. [§5.4 and Table 3] The Wilcoxon signed-rank test is reported as showing statistical significance for the GIZA++-supervised variant against GIZA++ (BPE-based), but no details are given about the unit of comparison (sentence-level AER) or the effect size. Since the test is two-sided at α=0.1%, it is a strong claim for only a few hundred test sentences, and the paper should report the number of paired sentences used, the test statistic, and ideally confidence intervals. Also, the comparison with Zenkel et al. (2019) and the BLEU 'no degradation' claim are not significance-tested; this should at least be acknowledged.
minor comments (4)
  1. [Title and passim] There are several typographical artifacts, including 'Transformer Mo dels' in the title and 'an method' in Section 3; these should be fixed.
  2. [Table 1] The Romanian→English training size is shown as '0.5k', but the text in Section 5.1.1 says the training data were augmented from 49k to 0.4M parallel sentences. This is inconsistent and should be corrected (likely 0.5M).
  3. [Equation (10)] The notation L'_a = La(A|f^J_1, e^I_1) is not fully defined; it should be stated explicitly that the alignment head's attention distribution is conditioned on the complete target sentence in the unmasked forward pass.
  4. [§4.1] The claim that 'the penultimate layer naturally tends to learn alignments' is based only on the three test sets in Table 2; phrase this as an empirical observation for these datasets rather than a general property.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; central claims are empirical and evaluated against external gold alignments.

full rationale

The paper's central claims are empirical results rather than a derivation chain. The self-training variant (Section 4.2) uses layer-average attention alignments as training labels, but the reported alignment error rates are measured against external gold alignments (Vilar et al. 2006; Mihalcea and Pedersen 2003) using the standard AER metric. Improvement over the layer-average baseline is therefore not guaranteed by construction; it is an empirical outcome. The GIZA++-supervised variant (Sections 4.4 and 5.4) trains on external IBM-model alignments and is evaluated on the same gold alignments; beating GIZA++ by roughly 2.9-3.9 AER points (Table 4) means the model does not simply reproduce its training labels, since copying GIZA++'s output would yield parity, not superiority. There is no self-citation chain, no imported uniqueness theorem, and no quantity defined in terms of the quantity it is claimed to predict. The only methodological soft spot is that the alignment loss weight λ and the choice of alignment-head layer are selected using test-set AER information (Sections 5.1.1 and 5.4), which is an optimistic-bias/validity concern rather than circularity: the reported numbers are not independent estimates of generalization, but the method's outputs are not equivalent to its inputs by definition. No circular step meets the required standard of exhibiting Eq. X = Eq. Y by construction or a fitted parameter renamed as a prediction.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

The method rests on domain assumptions about attention-as-alignment and annotation quality, plus two hand-tuned choices. No new theoretical entities are introduced; the alignment head is an existing attention head with a new supervision objective.

free parameters (2)
  • alignment loss weight λ = 0.05
    Tuned by grid search on the German-English dataset (Section 5.4) and then used for all other datasets.
  • alignment head layer = penultimate layer (layer 5 of 6)
    Chosen after evaluating layer-wise AER on the gold test alignments in Section 5.3; the selection is not held out.
assumptions (3)
  • domain assumption Encoder-decoder attention probabilities in a Transformer, particularly in the penultimate layer, are a meaningful basis for extracting word alignments.
    The paper's layer-average baseline and alignment head rely on attention probabilities being interpretable as alignments, validated empirically in Tables 2 and 3.
  • domain assumption Gold alignments from the NAACL 2003 shared task and Vilar et al. (2006) are reliable ground truth for optimizing and evaluating alignment.
    All AER numbers and supervision labels depend on these annotation guidelines (Section 5.1.1).
  • domain assumption BPE sub-word alignments can be mapped to word-level alignments by taking the union over sub-word pairs.
    Section 5.1: a target word is considered aligned to a source word if any of their sub-words are aligned.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Jointly Learning to Align and Translate with Transformer Models." pith.science (2026). https://pith.science/paper/JMM6WYRD

@misc{pith2026190902074,
  author       = {Pith},
  title        = {Pith review of: Jointly Learning to Align and Translate with Transformer Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JMM6WYRD}},
  note         = {Machine review of arXiv:1909.02074}
}
read the original abstract

The state of the art in machine translation (MT) is governed by neural approaches, which typically provide superior translation accuracy over statistical approaches. However, on the closely related task of word alignment, traditional statistical word alignment models often remain the go-to solution. In this paper, we present an approach to train a Transformer model to produce both accurate translations and alignments. We extract discrete alignments from the attention probabilities learnt during regular neural machine translation model training and leverage them in a multi-task framework to optimize towards translation and alignment objectives. We demonstrate that our approach produces competitive results compared to GIZA++ trained IBM alignment models without sacrificing translation accuracy and outperforms previous attempts on Transformer model based word alignment. Finally, by incorporating IBM model alignments into our multi-task training, we report significantly better alignment accuracies compared to GIZA++ on three publicly available data sets.

Figures

Figures reproduced from arXiv: 1909.02074 by the authors.

Figure 1
Figure 1. Two examples from the German→English alignment test set. Alignments in (a) show the output from GIZA++ and (b) from our model (Multi-task with full-context and GIZA++ supervised). Gold Alignments in shown in (c). Black squares and hatched squares in the reference represent sure and possible alignments, respectively. 6 Analysis To further investigate why our proposed approach is superior to GIZA++ in terms of AER, we… view at source ↗
Figure 2
Figure 2. Test AER and validation loss (NLL) per epoch on the WMT’18 English→German task. picking an earlier checkpoint for generating word alignments could be beneficial for better supervi￾sion. Unfortunately, an alignment validation set does not exist for this task. 7 Related Work Leveraging alignments obtained from statistical MT toolkits to guide NMT attention mechanisms has been explored in the past. Mi et al. (2016), Ch… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

36 extracted references · 30 canonical work pages

  1. [1]

    Tamer Alkhouli, Gabriel Bretschner, and Hermann Ney. 2018. https://www-i6.informatik.rwth-aachen.de/publications/downloader.php?id=1077&row=pdf On the alignment problem in multi-head attention-based neural machine translation . In Conference on Empirical Methods in Natural Language Processing, pages 177--185, Brussels, Belgium

    work page 2018

  2. [2]

    Tamer Alkhouli and Hermann Ney. 2017. https://www-i6.informatik.rwth-aachen.de/publications/downloader.php?id=1036&row=pdf Biasing attention-based recurrent neural networks using external alignment information . In Conference on Empirical Methods in Natural Language Processing, pages 108--117, Copenhagen, Denmark

    work page 2017

  3. [3]

    Philip Arthur, Graham Neubig, and Satoshi Nakamura. 2016. https://aclweb.org/anthology/D16-1162 Incorporating discrete translation lexicons into neural machine translation . In Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, pages 1557--1567

    work page 2016

  4. [4]

    Dzmitry Bahdanau, Kyunghyun Cho, and Yoshua Bengio. 2015. https://arxiv.org/pdf/1409.0473.pdf Neural machine translation by jointly learning to align and translate . In International Conference on Learning Representations, San Diego, CA, USA

    arXiv 2015

  5. [5]

    Ond r ej Bojar, Christian Federmann, Mark Fishel, Yvette Graham, Barry Haddow, Philipp Koehn, and Christof Monz. 2018. https://www.aclweb.org/anthology/W18-6401 Findings of the 2018 conference on machine translation (wmt18) . In Conference on Statistical Machine Translation, pages 272--303, Belgium, Brussels

    work page 2018

  6. [6]

    Brown, Vincent J

    Peter F. Brown, Vincent J. Della Pietra, Stephen A. Della Pietra, and Robert L. Mercer. 1993. http://dl.acm.org/citation.cfm?id=972470.972474 The mathematics of statistical machine translation: Parameter estimation . Computational Linguistics, 19(2):263--311

    work page arXiv 1993

  7. [7]

    Rajen Chatterjee, Matteo Negri, Marco Turchi, Marcello Federico, Lucia Specia, and Fr \'e d \'e ric Blain. 2017. https://aclweb.org/anthology/W17-4716 Guiding neural machine translation decoding with external knowledge . In Proceedings of the Second Conference on Machine Translation, pages 157--168

    work page 2017

  8. [8]

    Wenhu Chen, Evgeny Matusov, Shahram Khadivi, and Jan-Thorsten Peter. 2016. https://arxiv.org/pdf/1607.01628.pdf Guided alignment training for topic-aware neural machine translation . In Association for Machine Translation in the Americas, pages 121--134, Austin, TX, USA

    work page Pith review arXiv 2016

Show all 36 references
  1. [9]

    Sergey Edunov, Myle Ott, Michael Auli, and David Grangier. 2018. https://www.aclweb.org/anthology/D18-1045 Understanding back-translation at scale . In Conference on Empirical Methods in Natural Language Processing, pages 489--500, Brussels, Belgium

    2018

  2. [10]

    Qin Gao and Stephan Vogel. 2008. http://dl.acm.org/citation.cfm?id=1622110.1622119 Parallel implementations of word alignment tool . In Software Engineering, Testing, and Quality Assurance for Natural Language Processing, pages 49--57, Columbus, OH, USA

    2008

  3. [11]

    Hamidreza Ghader and Christof Monz. 2017. https://aclweb.org/anthology/I17-1004 What does attention in neural machine translation pay attention to? In International Joint Conference on Natural Language Processing, pages 30--39, Taipei, Taiwan

    2017

  4. [12]

    Philipp Koehn, Amittai Axelrod, Ra Birch Mayne, Chris Callison-burch, Miles Osborne, and David Talbot. 2005. http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.436.6226&rep=rep1&type=pdf Edinburgh system description for the 2005 iwslt speech translation evaluation . In I...

    2005

  5. [13]

    Philipp Koehn, Hieu Hoang, Alexandra Birch, Chris Callison-Burch, Marcello Federico, Nicola Bertoldi, Brooke Cowan, Wade Shen, Christine Moran, Richard Zens, Chris Dyer, Ondrej Bojar, Alexandra Constantin, and Evan Herbst. 2007. https://www.aclweb.org/anthology/P07-2045 Moses:...

    2007

  6. [14]

    Philipp Koehn and Rebecca Knowles. 2017. https://www.aclweb.org/anthology/W17-3204 Six challenges for neural machine translation . In Proceedings of the First Workshop on Neural Machine Translation, pages 28--39

    2017

  7. [15]

    Xintong Li, Lemao Liu, Zhaopeng Tu, Shuming Shi, and Max Meng. 2018. https://www.aclweb.org/anthology/N18-1125 Target foresight based attention for neural machine translation . In Conference of the North American Chapter of the Association for Computational Linguistics, pages ...

    2018

  8. [16]

    Lemao Liu, Masao Utiyama, Andrew Finch, and Eiichiro Sumita. 2016. https://www.aclweb.org/anthology/C16-1291 Neural machine translation with supervised attention . In International Conference on Computational Linguistics, pages 3093--3102, Osaka, Japan

    2016

  9. [17]

    Ilya Loshchilov and Frank Hutter. 2019. https://openreview.net/pdf?id=rk6qdGgCZ Decoupled weight decay regularization . In International Conference on Learning Representations, New Orleans, LA, USA

    2019

  10. [18]

    Haitao Mi, Zhiguo Wang, and Abe Ittycheriah. 2016. https://aclweb.org/anthology/D16-1249 Supervised attentions for neural machine translation . In Conference on Empirical Methods in Natural Language Processing, pages 2283--2288, Austin, TX, USA

    2016

  11. [19]

    Rada Mihalcea and Ted Pedersen. 2003. https://doi.org/10.3115/1118905.1118906 An evaluation exercise for word alignment . In Proceedings of the HLT-NAACL 2003 Workshop on Building and Using Parallel Texts: Data Driven Machine Translation and Beyond - Volume 3, pages 1--10, Edm...

    2003

  12. [20]

    Franz Josef Och and Hermann Ney. 2000. https://www.aclweb.org/anthology/P00-1056 Improved statistical alignment models . In Annual Meeting of the Association for Computational Linguistics, pages 440--447, Hong Kong

    2000

  13. [21]

    Franz Josef Och and Hermann Ney. 2003. https://doi.org/10.1162/089120103321337421 A systematic comparison of various statistical alignment models . Computational Linguistics, 29(1):19--51

    2003 doi

  14. [22]

    Kishore Papineni, Salim Roukos, Todd Ward, and Wei-Jing Zhu. 2002. https://www.aclweb.org/anthology/P02-1040.pdf BLEU : A method for automatic evaluation of machine translation . In Annual Meeting of the Association for Computational Linguistics, pages 311--318, Philadelphia, PA, USA

    2002

  15. [23]

    Jan-Thorsten Peter, Arne Nix Nix, and Hermann Ney. 2017. https://www-i6.informatik.rwth-aachen.de/publications/downloader.php?id=1034&row=pdf Generating alignments using target foresight in attention-based neural machine translation . In Conference of the European Association ...

    2017

  16. [24]

    Matt Post. 2018. https://www.aclweb.org/anthology/W18-6319 A call for clarity in reporting bleu scores . In Conference on Statistical Machine Translation, pages 186--191, Belgium, Brussels

    2018

  17. [25]

    Ofir Press and Lior Wolf. 2017. https://www.aclweb.org/anthology/E17-2025 Using the output embedding to improve language models . In Conference of the European Chapter of the Association for Computational Linguistics, pages 157--163, Valencia, Spain

    2017

  18. [26]

    Alessandro Raganato and J \"o rg Tiedemann. 2018. https://aclweb.org/anthology/W18-5431 An analysis of encoder representations in transformer-based machine translation . In BlackboxNLP: Analyzing and Interpreting Neural Networks for NLP, pages 287--297, Brussels, Belgium

    2018

  19. [27]

    Rico Sennrich, Barry Haddow, and Alexandra Birch. 2016. https://doi.org/10.18653/v1/P16-1162 Neural machine translation of rare words with subword units . In Annual Meeting of the Association for Computational Linguistics, pages 1715--1725, Berlin, Germany

    2016 doi

  20. [28]

    Emma Strubell, Patrick Verga, Daniel Andor, David Weiss, and Andrew McCallum. 2018. https://www.aclweb.org/anthology/D18-1548 Linguistically-informed self-attention for semantic role labeling . In Conference on Empirical Methods in Natural Language Processing, pages 5027--5038...

    2018

  21. [29]

    Gongbo Tang, Rico Sennrich, and Joakim Nivre. 2018. https://www.aclweb.org/anthology/W18-6304 An analysis of attention mechanisms: The case of word sense disambiguation in neural machine translation . In Conference on Statistical Machine Translation, pages 26--35, Brussels, Belgium

    2018

  22. [30]

    Zhaopeng Tu, Zhengdong Lu, Yang Liu, Xiaohua Liu, and Hang Li. 2016. https://www.aclweb.org/anthology/P16-1008 Modeling coverage for neural machine translation . In Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers),...

    2016

  23. [31]

    Gomez, Lukasz Kaiser, and Illia Polosukhin

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Lukasz Kaiser, and Illia Polosukhin. 2017. https://papers.nips.cc/paper/7181-attention-is-all-you-need.pdf Attention is all you need . In Advances in Neural Information Processing Systems,...

    2017

  24. [32]

    David Vilar, Maja Popovi\' c , and Hermann Ney. 2006. https://www-i6.informatik.rwth-aachen.de/publications/download/277/Vilar-IWSLT-2006.pdf AER : Do we need to “improve” our alignments? In International Workshop on Spoken Language Translation, pages 205--212, Kyoto, Japan

    2006

  25. [33]

    Stephan Vogel, Hermann Ney, and Christoph Tillmann. 1996. https://doi.org/10.3115/993268.993313 HMM -based word alignment in statistical translation . In International Conference on Computational Linguistics, pages 836--841, Copenhagen, Denmark

    1996

  26. [34]

    Thomas Zenkel , Joern Wuebker , and John DeNero . 2019. http://arxiv.org/abs/1901.11359 Adding Interpretable Attention to Neural Translation Models Improves Word Alignment . arXiv e-prints

    2019 arXiv

  27. [35]

    URL: " 'urlintro :=

    ENTRY address author booktitle chapter edition editor howpublished institution journal key month note number organization pages publisher school series title type volume year eprint doi pubmed url lastchecked label extra.label sort.label short.list INTEGERS output.state before...

  28. [36]

    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 14, 2026 · model on record in the stance chip above.