用合成数据提升医疗建议检测模型性能，小模型反超大模型。
Backprompting: Leveraging Synthetic Production Data for Health Advice Guardrails
- 通过反向提示生成类真实医疗建议文本，解决标注数据不足问题。
- 合成数据增强后检测器准确率超越GPT-4o达3.73%，参数量仅为其1/400。
- 结合少量人工校正聚类，高效构建高保真训练数据集，适合安全防护场景。
大型语言模型在企业中的广泛应用带来了诸多风险，防护机制需通过检测器过滤模型输入输出文本。然而，构建可靠检测器面临真实生产数据标注难的问题。本文提出回溯提示（Backprompting）方法，生成类真实的医疗建议文本以补充训练数据。我们结合稀疏人机协同聚类对生成数据进行标注，构建与原始数据分布相似的平行语料库。将合成样本注入现有数据集后，训练出更鲁棒的检测器。在识别医疗建议这一复杂任务中表现优异，其性能较其他方案有显著提升：尽管参数量仅为GPT-4o的1/400，仍可实现最高3.73%的准确率超越。
原文摘要 · Abstract (English)
The pervasiveness of large language models (LLMs) in enterprise settings has also brought forth a significant amount of risks associated with their usage. Guardrails technologies aim to mitigate this risk by filtering LLMs' input/output text through various detectors. However, developing and maintaining robust detectors faces many challenges, one of which is the difficulty in acquiring production-quality labeled data on real LLM outputs prior to deployment. In this work, we propose backprompting, a simple yet intuitive solution to generate production-like labeled data for health advice guardrails development. Furthermore, we pair our backprompting method with a sparse human-in-the-loop clustering technique to label the generated data. Our aim is to construct a parallel corpus roughly representative of the original dataset yet resembling real LLM output. We then infuse existing datasets with our synthetic examples to produce robust training data for our detector. We test our technique in one of the most difficult and nuanced guardrails: the identification of health advice in LLM output, and demonstrate improvement versus other solutions. Our detector is able to outperform GPT-4o by up to 3.73%, despite having 400x less parameters.
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。