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。

帖子提出将 JEV 视为一种多镜头路由器,把决策直接放入口袋,并类比为流动的思维而非行李。
Basically, I reckon it's a good idea to think of JEV as this kind of multi-shot router that drops decision-making right into your pocket... that's the image to run with. Instead of flowing luggage like this, it's more like flowing thoughts, y'know.