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。

帖子用两个问题区分模型:读取人类文本还是机器信号,返回聊天还是决策。聊天模型在左上,Jev 和 guard 模型在左下,右侧几乎空白。
Two questions place any model. What does it read: human text or machine signals? What does it return: a chat or a decision? Chat models top left. Jev and the guard models bottom left. The right side, where a plant lives, is nearly empty.