Recent works have shown that prompting large language models (LLMs) is effective for translation with markup where LLMs can simultaneously transfer markup tags while ensuring that the content, both inside and outside tag pairs is correctly translated. However, these works make a rather unrealistic assumption of the existence of high-quality parallel sentences with markup for prompting. Furthermore, the impact of instruction fine-tuning (IFT) in this setting is unknown. In this paper, we provide a study, the first of its kind, focusing on the effectiveness of synthetically created markup data and IFT for translation with markup using LLMs. We focus on translation from English to five European languages, German, French, Dutch, Finnish and Russian, where we show that regardless of few-shot prompting or IFT, synthetic data created via word alignments, while leading to inferior markup transfer compared to using original data with markups, does not negatively impact the translation quality. Furthermore, IFT mainly impacts the translation quality compared to few-shot prompting and has slightly better markup transfer capabilities than the latter. We hope our work will help practitioners make effective decisions on modeling choices for LLM based translation with markup.

In this paper, we study the translation abilities of Large Language Models (LLMs) for business IT texts.We are strongly interested in domain adaptation of translation systems, which is essential for accurate and lexically appropriate translation of such texts.Among the open-source models evaluated in a zero- and few-shot setting, we find Llama-2 13B the most promising for domain-specific translation fine-tuning.We investigate the full range of adaptation techniques for LLMs: from prompting, over parameter-efficient fine-tuning to full fine-tuning, and compare to classic neural machine translation (MT) models trained internally at SAP.We provide guidance how to use training budget most effectively for different fine-tuning approaches.We observe that while LLMs can translate on-par with SAP’s MT models on general domain data, it is difficult to close the gap on SAP’s domain-specific data, even with extensive training and carefully curated data.

While large language models (LLMs) pre-trained on massive amounts of unpaired language data have reached the state-of-the-art in machine translation (MT) of general domain texts, post-editing (PE) is still required to correct errors and to enhance term translation quality in specialized domains. In this paper we present a pilot study of enhancing translation memories (TM) produced by PE (source segments, machine translations, and reference translations, henceforth called PE-TM) for the needs of correct and consistent term translation in technical domains. We investigate a light-weight two-step scenario where at inference time, a human translator marks errors in the first translation step, and in a second step a few similar examples are extracted from the PE-TM to prompt an LLM. Our experiment shows that the additional effort of augmenting translations with human error markings guides the LLM to focus on a correction of the marked errors, yielding consistent improvements over automatic PE (APE) and MT from scratch.

Large language models (LLMs) have demonstrated considerable success in various natural language processing tasks, but open-source LLMs have yet to attain state-of-the-art performance in Neural Machine Translation (NMT). Nevertheless, their significant performance in tasks demanding a broad understanding and contextual processing shows their potential for translation. To exploit these abilities, we investigate using LLMs for MT and explore recent parameter-efficient fine-tuning techniques. Surprisingly, our initial experiments found that fine-tuning with Q-LoRA for translation purposes led to performance improvements in terms of BLEU but degradation in COMET compared to in-context learning. To overcome this, we propose an alternative approach: adapting LLMs as Automatic Post-Editors (APE) rather than direct translators. Building on the ability of the LLM to handle long sequences, we also propose extending our approach to document-level translation. We show that leveraging Low-Rank-Adapter fine-tuning for APE can yield significant improvements across both sentence and document-level metrics while generalizing to out-of-domain data. Most notably, we achieve a state-of-the-art accuracy rate of 88.7% on the ContraPro test set, which assesses the model’s ability to resolve pronoun ambiguities when translating from English to German. Lastly, during manual post-editing for document-level translation, the source sentences are iteratively annotated, which can be used to refine further translations in the document. Here, we demonstrate that leveraging human corrections can significantly reduce the number of edits required for subsequent translations.

Preference Optimization (PO) techniques are currently one of the state of the art techniques for fine-tuning large language models (LLMs) on pairwise preference feedback from human annotators. However, in machine translation, this sort of feedback can be difficult to solicit. Additionally, Kreuzer et al. (2018) have shown that, for machine translation, pairwise preferences are less reliable than other forms of human feedback, such as 5-point ratings.We examine post-edits to see if they can be a source of reliable human preferences by construction. In PO, a human annotator is shown sequences $s_1$ and $s_2$ and asked for a preference judgment, while for post-editing, editors create $s_1$ and know that it should be better than $s_2$. We attempt to use these implicit preferences for PO and show that it helps the model move towards post-edit like hypotheses and away from machine translation-like hypotheses. Furthermore, we show that best results are obtained by pre-training the model with supervised fine-tuning (SFT) on post-edits in order to promote post-edit like hypotheses to the top output ranks.

Recent advancements in NLP have resulted in models with specialized strengths, such as processing multimodal inputs or excelling in specific domains. However, real-world tasks, like multimodal translation, often require a combination of these strengths, such as handling both translation and image processing. While individual translation and vision models are powerful, they typically lack the ability to perform both tasks in a single system. Combining these models poses challenges, particularly due to differences in their vocabularies, which limit the effectiveness of traditional ensemble methods to post-generation techniques like N-best list re-ranking. In this work, we propose a novel zero-shot ensembling strategy that allows for the integration of different models during the decoding phase without the need for additional training. Our approach re-ranks beams during decoding by combining scores at the word level, using heuristics to predict when a word is completed. We demonstrate the effectiveness of this method in machine translation scenarios, showing that it enables the generation of translations that are both speech- and image-aware while also improving overall translation quality.

Advancements in Neural Machine Translation (NMT) greatly benefit the software localization industry by decreasing the post-editing time of human annotators. Although the volume of the software being localized is growing significantly, techniques for improving NMT for user interface (UI) texts are lacking. These UI texts have different properties than other collections of texts, presenting unique challenges for NMT. For example, they are often very short, causing them to be ambiguous and needing additional context (button, title text, a table item, etc.) for disambiguation. However, no such UI data sets are readily available with contextual information for NMT models to exploit. This work aims to provide a first step in improving UI translations and highlight its challenges. To achieve this, we provide a novel multilingual UI corpus collection (∼ 1.3M for English ↔ German) with a targeted test set and analyze the limitations of state-of-the-art methods on this challenging task. Specifically, we present a targeted test set for disambiguation from English to German to evaluate reliably and emphasize UI translation challenges. Furthermore, we evaluate several state-of-the-art NMT techniques from domain adaptation and document-level NMT on this challenging task. All the scripts to replicate the experiments and data sets are available here.ˆ,

Supervised learning in Neural Machine Translation (NMT) standardly follows a teacher forcing paradigm where the conditioning context in the model’s prediction is constituted by reference tokens, instead of its own previous predictions. In order to alleviate this lack of exploration in the space of translations, we present a simple extension of standard maximum likelihood estimation by a contrastive marking objective. The additional training signals are extracted automatically from reference translations by comparing the system hypothesis against the reference, and used for up/down-weighting correct/incorrect tokens. The proposed new training procedure requires one additional translation pass over the training set, and does not alter the standard inference setup. We show that training with contrastive markings yields improvements on top of supervised learning, and is especially useful when learning from postedits where contrastive markings indicate human error corrections to the original hypotheses.

In this paper we evaluate the utility of large language models (LLMs) for translation of text with markup in which the most important and challenging aspect is to correctly transfer markup tags while ensuring that the content, both, inside and outside tags is correctly translated. While LLMs have been shown to be effective for plain text translation, their effectiveness for structured document translation is not well understood. To this end, we experiment with BLOOM and BLOOMZ, which are open-source multilingual LLMs, using zero, one and few-shot prompting, and compare with a domain-specific in-house NMT system using a detag-and-project approach for markup tags. We observe that LLMs with in-context learning exhibit poorer translation quality compared to the domain-specific NMT system, however, they are effective in transferring markup tags, especially the large BLOOM model (176 billion parameters). This is further confirmed by our human evaluation which also reveals the types of errors of the different tag transfer techniques. While LLM-based approaches come with the risk of losing, hallucinating and corrupting tags, they excel at placing them correctly in the translation.

Translation of structured content is an important application of machine translation, but the scarcity of evaluation data sets, especially for Asian languages, limits progress. In this paper we present a novel multilingual multiway evaluation data set for the translation of structured documents of the Asian languages Japanese, Korean and Chinese. We describe the data set, its creation process and important characteristics, followed by establishing and evaluating baselines using the direct translation as well as detag-project approaches. Our data set is well suited for multilingual evaluation, and it contains richer annotation tag sets than existing data sets. Our results show that massively multilingual translation models like M2M-100 and mBART-50 perform surprisingly well despite not being explicitly trained to handle structured content. The data set described in this paper and used in our experiments is released publicly.

This paper addresses the automatic translation of conversational content in a business context, for example support chat dialogues. While such use cases share characteristics with other informal machine translation scenarios, translation requirements with respect to technical and business-related expressions are high. To succeed in such scenarios, we experimented with curating dedicated training and test data, injecting noise to improve robustness, and applying sentence weighting schemes to carefully manage the influence of the different corpora. We show that our approach improves the performance of our models on conversational content for all 18 investigated language pairs while preserving translation quality on other domains - an indispensable requirement to integrate these developments into our MT engines at SAP.

The correct translation of named entities (NEs) still poses a challenge for conventional neural machine translation (NMT) systems. This study explores methods incorporating named entity recognition (NER) into NMT with the aim to improve named entity translation. It proposes an annotation method that integrates named entities and inside–outside–beginning (IOB) tagging into the neural network input with the use of source factors. Our experiments on English→German and English→ Chinese show that just by including different NE classes and IOB tagging, we can increase the BLEU score by around 1 point using the standard test set from WMT2019 and achieve up to 12% increase in NE translation rates over a strong baseline.

This paper examines approaches to bias a neural machine translation model to adhere to terminology constraints in an industrial setup. In particular, we investigate variations of the approach by Dinu et al. (2019), which uses inline annotation of the target terms in the source segment plus source factor embeddings during training and inference, and compare them to constrained decoding. We describe the challenges with respect to terminology in our usage scenario at SAP and show how far the investigated methods can help to overcome them. We extend the original study to a new language pair and provide an in-depth evaluation including an error classification and a human evaluation.

This paper accompanies the software documentation data set for machine translation, a parallel evaluation data set of data originating from the SAP Help Portal, that we released to the machine translation community for research purposes. It offers the possibility to tune and evaluate machine translation systems in the domain of corporate software documentation and contributes to the availability of a wider range of evaluation scenarios. The data set comprises of the language pairs English to Hindi, Indonesian, Malay and Thai, and thus also increases the test coverage for the many low-resource language pairs. Unlike most evaluation data sets that consist of plain parallel text, the segments in this data set come with additional metadata that describes structural information of the document context. We provide insights into the origin and creation, the particularities and characteristics of the data set as well as machine translation results.