用轻量微调提升多语言文本净化模型，尤其在低资源语言上表现优异。
GemDetox at TextDetox CLEF 2025: Enhancing a Massively Multilingual Model for Text Detoxification on Low-resource Languages
- 基于Gemma-3模型，采用LoRA微调与少样本提示策略。
- 在15种语言上实现高精度语义保留的中性化改写，低资源语言效果显著。
- 适合关注多语言内容安全与低资源场景应用的研究者。
随着社交媒体平台快速发展，监管滞后，自动化文本净化成为保障网络对话安全的重要工具。本文介绍我们参加PAN 2025多语言文本净化挑战的方案，目标是将15种语言的单句有毒输入转化为中性表述。基于120亿参数的Gemma-3多语言Transformer，我们采用参数高效的LoRA SFT微调及少样本、思维链（CoT）等提示技术。训练数据包含3,600对人工标注平行语料、21,600对机器翻译生成的合成语料，以及通过杰卡德阈值筛选的模型生成语料。推理时，输入融合三个LaBSE检索的近邻文本和显式毒性片段标注。评估指标包括风格迁移准确率、基于LaBSE的语义保留度和xCOMET流畅度，系统在高资源与低资源语言上均排名第一。消融实验显示，少样本示例带来+0.081的联合得分提升，基础CoT提示带来+0.088提升。方差分析表明，语言资源状态是性能最强预测因子（η² = 0.667，p < 0.01）。
原文摘要 · Abstract (English)
As social-media platforms emerge and evolve faster than the regulations meant to oversee them, automated detoxification might serve as a timely tool for moderators to enforce safe discourse at scale. We here describe our submission to the PAN 2025 Multilingual Text Detoxification Challenge, which rewrites toxic single-sentence inputs into neutral paraphrases across 15 typologically diverse languages. Building on a 12B-parameter Gemma-3 multilingual transformer, we apply parameter-efficient LoRA SFT fine-tuning and prompting techniques like few-shot and Chain-of-Thought. Our multilingual training corpus combines 3,600 human-authored parallel pairs, 21,600 machine-translated synthetic pairs, and model-generated pairs filtered by Jaccard thresholds. At inference, inputs are enriched with three LaBSE-retrieved neighbors and explicit toxic-span annotations. Evaluated via Style Transfer Accuracy, LaBSE-based semantic preservation, and xCOMET fluency, our system ranks first on high-resource and low-resource languages. Ablations show +0.081 joint score increase from few-shot examples and +0.088 from basic CoT prompting. ANOVA analysis identifies language resource status as the strongest predictor of performance ($η^2$ = 0.667, p < 0.01).
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。