研究发现：LLM 时代回翻译仍是低资源机器翻译的有效手段
Back-Translation Still Matters: Leveraging Monolingual Data for Low-Resource Machine Translation with LLMs
只有 1K 平行句对也能用回翻译把低资源语言翻译提上去，比继续预训练好不少，做小语种翻译的可以看看这篇论文。
一篇 arXiv 论文提出用回翻译（BT）技术利用低资源语言的单语数据来提升 LLM 翻译能力。在 Odia 和 Swahili 上仅用 1K 平行句对，回翻译比继续预训练高出 7.7 到 16.4 个 XCOMET-XXL 分数。在五种低资源语言和两个 3B 骨干模型上，BT 比仅用平行数据微调提升 16% 到 173% 的 XCOMET-XXL。配合 beam 解码的中间翻译，效果甚至超过平行数据多 10 倍的模型，Odia 和 Sanskrit 的人工盲评也验证了这一结果。
Back-Translation Still Matters: Leveraging Monolingual Data for Low-Resource Machine Translation with LLMs
Pre-trained Large Language Model (LLMs) are effective on many tasks in zero-shot and few-shot settings. For machine translation of low-resource languages, the obvious options are few-shot prompting and fine-tuning on a small parallel set. We ask: what if moderate amounts of low-resource monolingual data are also available? We argue that, rather than continued pretraining, a more effective use of this data is the pre-LLM era technique of back-translation (BT). The current fine-tuned LLM translates monolingual sentences in each language, and the resulting pairs train it in the reverse direction. Updates on the parallel set are interleaved for stable learning. With 1K parallel pairs on Odia and Swahili, BT beats continued pretraining into the low-resource language by 7.7 to 16.4 XCOMET-XXL points in domain. With beam-decoded intermediate translations, it also beats a model with 10 time more parallel data. In-context examples retrieved from the parallel data further improve BT. We also analyze the source, quality and amount of monolingual text. Across five low-resource languages, with Hindi as a high-resource control, and two 3B backbones, BT improves translation into the low-resource language by 16 to 173% XCOMET-XXL over parallel-only fine-tuning. Blind human evaluation on Odia and Sanskrit agrees with the automatic metrics.
