arXiv:2608. 05232v1 Announce Type: cross Abstract: The work of Tang et.
Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Machine Learning.
The paper investigates how large language models (LLMs) can be used to generate synthetic data for low‑resource machine translation, focusing on Romansh, which has six distinct varieties. It finds that LLMs are better at translating from Romansh to a high‑resource language (German) than the reverse, creating an asymmetry that makes the direction of data augmentation critical. By generating synthetic translations into German rather than Romansh, the authors surpass a Gemini 3 Pro baseline on German‑Romansh translation, achieving a +23 BLEU improvement in the lowest‑resource variety and producing fluent translations in each Romansh variety according to human evaluation.
The paper examines two test‑time scaling methods for large language models in machine translation: sequential sampling, where later attempts build on earlier ones, and parallel sampling, such as independent i.i.d. sampling with reranking. Sequential sampling shows a higher performance ceiling, offering a more diverse and effective set of translations, especially with limited sampling budgets. Human analysis reveals that while sequential sampling improves fluency and naturalness, it can reduce accuracy when the inference budget is large, and the authors attribute this effect to the model’s access to a larger target‑side context.
arXiv:2608. 09766v1 Announce Type: cross Abstract: Multilingual translation benchmarks are typically sourced in English and translated into other languages, treating language pairs as the unit of evaluation---a design that is prone to contamination over time and overlooks locale and cultural considerations.
EnSiTa is a trilingual multi‑domain parallel dataset and benchmark for English, Sinhala, and Tamil. It contains human post‑edited training data across seven domains and professionally translated test sets for those domains plus an additional one, all produced through a multi‑year, rigorously quality‑controlled process. The authors use EnSiTa to conduct a comprehensive study of domain‑specific machine translation across six language directions, comparing from‑scratch Transformers, pre‑trained models, and decoder‑only LLMs under various training‑data sizes, model scales, and domain settings.
EuroAlpaca presents a task‑preserving localisation pipeline that translates English instruction‑tuning data into 50 European languages while maintaining task‑critical constraints. The method uses field‑wise machine translation or reconstructs task‑equivalent target‑language instances, followed by validation of coherence and consistency. Experiments show that EuroAlpaca improves instruction‑following accuracy by 12.9% over a baseline and outperforms direct translation on ROUGE‑L and F‑BERT metrics.
Fine‑tuning large language models on parallel data can improve translation quality but also causes catastrophic forgetting of general capabilities. The study evaluates several forgetting‑mitigation methods—anchored to auxiliary data, model outputs, and base model parameters—using Llama 3.2 1B Instruct and Llama 3.1 8B Instruct on Arabic‑English and Spanish‑English translation tasks. Elastic Weight Consolidation best preserves general benchmark performance, yet only data mixing with control‑task examples maintains instruction‑following abilities such as formality and grammatical gender control, though these gains do not generalize to unseen prompts.