用轻量模型实现多语言事实核查，比大模型更高效稳定。
Multilingual Fact-Checking at Scale: Fine-Tuned Compact Models vs LLMs
- 分三阶段：识别陈述、检索证据、判断真假，每步用专精小模型。
- 114种语言上表现稳定，小模型在延迟和成本上大幅优于大模型。
- 适合需要私有部署、低延迟的多语言内容审核场景。
我们提出一个在Factiverse部署的多语言事实核查系统，支持高吞吐、低延迟运行。系统采用模块化流程：陈述检测、证据检索与重排、真伪预测。使用XLM-RoBERTa-Large微调用于陈述检测，mmBERT-base用于三分类立场判断（支持/反驳/混合），基于SetFit的多语言重排序模型用于陈述-证据匹配。与GPT-5.2、Claude Opus 4.6、Qwen3-8b等强基座大模型对比，在覆盖114种语言的陈述检测和28种语言的真伪预测任务中，专用微调模型展现出强大且稳定的性能；检索模型也保持与现代私有嵌入模型相当的效果。同硬件下的延迟测试显示，编码器类组件效率显著提升，更适合资源受限、隐私敏感的生产环境。整体表明，轻量级微调自托管模型是大规模多语言事实核查的可行且高效基础。代码与数据见https://github.com/factiverse/factcheck-editor。
原文摘要 · Abstract (English)
We present a multilingual fact-checking system deployed at Factiverse, designed for high-throughput and low-latency operation across diverse languages. The system follows a modular pipeline with three stages: claim detection, evidence retrieval and re-ranking, and veracity prediction. We fine-tune XLM-RoBERTa-Large for claim detection, mmBERT-base for three-label stance classification (Supports/Refutes/Mixed), and a SetFit-based multilingual re-ranker for claim--evidence matching. We compare these components against strong LLM baselines, including GPT-5.2, Claude Opus~4.6, and Qwen3-8b. Experiments on production data spanning 114 languages for claim detection and 28 languages for veracity prediction show that task-specific fine-tuning provides strong and stable multilingual performance, while the fine-tuned retrieval model remains competitive with modern proprietary embeddings. Same-hardware latency measurements further show large efficiency gains for encoder-based components, supporting their use in production deployments with tight cost and privacy constraints. Overall, compact fine-tuned, self-hosted models remain a practical and effective foundation for multilingual fact-checking at scale. Code and data used for this study are available at https://github.com/factiverse/factcheck-editor.
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。