Jev 与文本生成模型的取舍
Jev 不生成自由文本。Diogo 用一段对比视频说明并行决策模型与逐字生成模型在任务形态上的差别。

作者质疑 TypeSafe 的 Jev 决策模型在没有推理的情况下是否安全,该模型对文档的单一类型问题返回校准概率,并作为护栏、路由器和审批门部署。
Is @typesafeai’s Jev actually safe? Decision models answer one typed question about a document and return a calibrated probability for every allowed answer. They ship as guardrails, routers and approval gates. My question is, without reasoning, are they actually safe?