Patmt2 improves multilingual patent translation with two-stage adaptation
PATMT2: Two-stage domain adaptation for multilingual patent translation
Computation and LanguageArtificial Intelligence
Summary
Translating patent documents between languages like Chinese, English, Japanese, and Korean can be tricky because of technical terms and formal language. The authors developed PatMT2, a system that improves an existing translation model by training it in two stages specifically for patents. Their improved model translates better, scoring higher on tests than previous versions and some big alternative systems. They also tested it on many other languages to understand how well the model keeps working outside its training set.
What this means in practice
- •For localization teams: Translate patent documents between Chinese, English, Japanese, and Korean more accurately using a tuned machine translation model.
- •For legal tech developers: Incorporate a two-stage adapted translation model to improve multilingual patent search and processing tools.
Authors
Xiaowei Su, Xusen Cheng
Abstract
We present PatMT2 (Patent Machine Translation, v2), a two-stage LoRA pipeline that adapts the Hy-MT2-7B dedicated translation model to four-language (Chinese/English/Japanese/Korean) patent translation. On six translation directions, the Stage-2 checkpoint reaches a macro-average BLEU of 43.39 (BF16) to 43.57 (AWQ INT4), improving over the Hy-MT2-7B base by up to +8.8 BLEU and also exceeding Qwen3-32B and several external systems on the same average. We then measure the cost of adaptation: probing 38 languages absent from the fine-tuning data, the model still produces correct, fluent output on 34, while four low-resource languages collapse into distinct, reproducible failure modes. We treat these as diagnostic cases rather than retention rates, and report their mechanisms without attributing them to a single cause. Paired bootstrap tests confirm the translation gains over all baselines (p < 0.001).