OpenAI News·· 2024-07-24AI 评分25
OpenAI 利用基于规则奖励改进模型安全行为
Improving Model Safety Behavior with Rule-Based Rewards
AI 导读
OpenAI 开发并应用了一种名为 Rule-Based Rewards (RBRs) 的新方法，使模型在无需大量人工数据收集的情况下实现安全对齐。该技术通过基于规则的奖励机制优化模型行为，旨在提升安全性并降低对人工标注数据的依赖。
原文
We’ve developed and applied a new method leveraging Rule-Based Rewards (RBRs) that aligns models to behave safely without extensive human data collection.
来源：OpenAI News · openai.com