Property Finder 修正评估偏差，在 Amazon Bedrock 上省 86% 成本
Fix evaluation bias to unlock better models on Amazon Bedrock
精选理由
房产平台 Property Finder 发现自家 96.8% 评估标注被旧模型污染，重做标注加进化式提示词优化后，在 Bedrock 上换模型省了 86% 成本，方法可以直接抄。
Property Finder 发现其视觉 AI 评估中 96.8% 的标注继承了旧模型的结果，掩盖了真实准确率。团队重建了无偏 ground truth，并应用进化式提示词优化来筛选模型。最终在 Amazon Bedrock 上选定了更准确且更便宜的模型，覆盖 11 个图像质量属性，成本降低 86%。
原文 · AWS Machine Learning Blog
Fix evaluation bias to unlock better models on Amazon Bedrock
Property Finder found that 96.8% of their vision-AI evaluation labels were inherited from the incumbent model, hiding its true accuracy. See how they built unbiased ground truth, applied evolutionary prompt optimization, and selected a more accurate, cost-effective model on Amazon Bedrock, cutting cost by 86% across 11 image-quality attributes.