Scaling, Simplification, and Adaptation: Lessons from Pretraining on Machine-Translated Text

• URL: http://arxiv.org/abs/2509.17317v1
• Date: Mon, 22 Sep 2025 02:48:43 GMT
• Title: Scaling, Simplification, and Adaptation: Lessons from Pretraining on Machine-Translated Text
• Authors: Dan John Velasco, Matthew Theodore Roque,
• Abstract summary: We translate English into Indonesian and Tamil--two typologically distant, lower-resource languages--and pretraining GPT-2 models (124M-774M) on native or MT-derived corpora.<n>We evaluate cross-entropy loss on native text, along with accuracy on syntactic probes and downstream tasks.<n>Our results show that MT-pretrained models benefit from scaling; (2) source-side simplification harms generalization to native text; and (3) adapting MT-pretrained models on native text often yields better performance than native-only models.
• Score: 0.19258299315493077
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Most languages lack sufficient data for large-scale monolingual pretraining, creating a "data wall." Multilingual pretraining helps but is limited by language imbalance and the "curse of multilinguality." An alternative is to translate high-resource text with machine translation (MT), which raises three questions: (1) How does MT-derived data scale with model capacity? (2) Can source-side transformations (e.g., simplifying English with an LLM) improve generalization to native text? (3) How well do models pretrained on MT-derived data adapt when continually trained on limited native text? We investigate these questions by translating English into Indonesian and Tamil--two typologically distant, lower-resource languages--and pretraining GPT-2 models (124M-774M) on native or MT-derived corpora from raw and LLM-simplified English. We evaluate cross-entropy loss on native text, along with accuracy on syntactic probes and downstream tasks. Our results show that (1) MT-pretrained models benefit from scaling; (2) source-side simplification harms generalization to native text; and (3) adapting MT-pretrained models on native text often yields better performance than native-only models, even with less native data. However, tasks requiring cultural nuance (e.g., toxicity detection) demand more exposure to native data.

Related papers

• Understanding In-Context Machine Translation for Low-Resource Languages: A Case Study on Manchu [53.437954702561065]
  In-context machine translation (MT) with large language models (LLMs) is a promising approach for low-resource MT.<n>This study systematically investigates how each type of resource, e.g., dictionary, grammar book, and retrieved parallel examples, affect the translation performance.<n>Our results indicate that high-quality dictionaries and good parallel examples are very helpful, while grammars hardly help.
  arXiv  Detail & Related papers  (2025-02-17T14:53:49Z)
• Quality or Quantity? On Data Scale and Diversity in Adapting Large Language Models for Low-Resource Translation [62.202893186343935]
  We explore what it would take to adapt Large Language Models for low-resource languages. We show that parallel data is critical during both pre-training andSupervised Fine-Tuning (SFT) Our experiments with three LLMs across two low-resourced language groups reveal consistent trends, underscoring the generalizability of our findings.
  arXiv  Detail & Related papers  (2024-08-23T00:59:38Z)
• Improving Language Models Trained on Translated Data with Continual Pre-Training and Dictionary Learning Analysis [3.16714407449467]
  We investigate the role of translation and synthetic data in training language models. We translate TinyStories, a dataset of 2.2M short stories for 3-4 year old children, from English to Arabic using the open NLLB-3B MT model. To rectify these issues, we pre-train the models with a small dataset of synthesized high-quality Arabic stories.
  arXiv  Detail & Related papers  (2024-05-23T07:53:04Z)
• Machine Translation for Ge'ez Language [0.0]
  Machine translation for low-resource languages such as Ge'ez faces challenges such as out-of-vocabulary words, domain mismatches, and lack of labeled training data. We develop a multilingual neural machine translation (MNMT) model based on languages relatedness. We also experiment with using GPT-3.5, a state-of-the-art LLM, for few-shot translation with fuzzy matches.
  arXiv  Detail & Related papers  (2023-11-24T14:55:23Z)
• Boosting Unsupervised Machine Translation with Pseudo-Parallel Data [2.900810893770134]
  We propose a training strategy that relies on pseudo-parallel sentence pairs mined from monolingual corpora and synthetic sentence pairs back-translated from monolingual corpora. We reach an improvement of up to 14.5 BLEU points (English to Ukrainian) over a baseline trained on back-translated data only.
  arXiv  Detail & Related papers  (2023-10-22T10:57:12Z)
• Beyond Triplet: Leveraging the Most Data for Multimodal Machine Translation [53.342921374639346]
  Multimodal machine translation aims to improve translation quality by incorporating information from other modalities, such as vision. Previous MMT systems mainly focus on better access and use of visual information and tend to validate their methods on image-related datasets. This paper establishes new methods and new datasets for MMT.
  arXiv  Detail & Related papers  (2022-12-20T15:02:38Z)
• MuCoT: Multilingual Contrastive Training for Question-Answering in Low-resource Languages [4.433842217026879]
  Multi-lingual BERT-based models (mBERT) are often used to transfer knowledge from high-resource languages to low-resource languages. We augment the QA samples of the target language using translation and transliteration into other languages and use the augmented data to fine-tune an mBERT-based QA model. Experiments on the Google ChAII dataset show that fine-tuning the mBERT model with translations from the same language family boosts the question-answering performance.
  arXiv  Detail & Related papers  (2022-04-12T13:52:54Z)
• Improving Multilingual Translation by Representation and Gradient Regularization [82.42760103045083]
  We propose a joint approach to regularize NMT models at both representation-level and gradient-level. Our results demonstrate that our approach is highly effective in both reducing off-target translation occurrences and improving zero-shot translation performance.
  arXiv  Detail & Related papers  (2021-09-10T10:52:21Z)
• Pre-training Multilingual Neural Machine Translation by Leveraging Alignment Information [72.2412707779571]
  mRASP is an approach to pre-train a universal multilingual neural machine translation model. We carry out experiments on 42 translation directions across a diverse setting, including low, medium, rich resource, and as well as transferring to exotic language pairs.
  arXiv  Detail & Related papers  (2020-10-07T03:57:54Z)
• Reusing a Pretrained Language Model on Languages with Limited Corpora for Unsupervised NMT [129.99918589405675]
  We present an effective approach that reuses an LM that is pretrained only on the high-resource language. The monolingual LM is fine-tuned on both languages and is then used to initialize a UNMT model. Our approach, RE-LM, outperforms a competitive cross-lingual pretraining model (XLM) in English-Macedonian (En-Mk) and English-Albanian (En-Sq)
  arXiv  Detail & Related papers  (2020-09-16T11:37:10Z)
• Leveraging Monolingual Data with Self-Supervision for Multilingual Neural Machine Translation [54.52971020087777]
  Using monolingual data significantly boosts the translation quality of low-resource languages in multilingual models. Self-supervision improves zero-shot translation quality in multilingual models. We get up to 33 BLEU on ro-en translation without any parallel data or back-translation.
  arXiv  Detail & Related papers  (2020-05-11T00:20:33Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.