arXiv：cs.AI· Xiaowei Su, Xusen Cheng·· 18 小时前
PATMT2：面向多语言专利翻译的两阶段领域自适应
PATMT2: Two-stage domain adaptation for multilingual patent translation
AI 导读
PatMT2 用两阶段 LoRA 流程将 Hy-MT2-7B 专用翻译模型适配到中英日韩四语专利翻译，在六个翻译方向上 Stage-2 checkpoint 的宏平均 BLEU 达 43.39（BF16）至 43.57（AWQ INT4），较 Hy-MT2-7B 基座最高提升 8.8 BLEU，并超过 Qwen3-32B 及若干外部系统。
正文
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).
| Comments: | 4 pages, 2 figures |
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.16466 [cs.CL] |
| (or arXiv:2610.16466v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.16466 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Xiaowei Su [view email]
[v1]
Mon, 12 Oct 2026 09:22:29 UTC (1,765 KB)
来源：arXiv：cs.AI · arxiv.org