omarsar0· @omarsar0 · X·· 4 小时前AI 评分
SubQ Snap 用 RL 训练决策模型
关注决策模型。 从我最近的测试来看，决策模型对业务可能极具变革性。 Agent 工作流中的许多调用都是结构化决策，为此支付前沿模型的价格并不合理。 @subquadratic 的 SubQ Snap 通过 RL 训练来处理这类调用。它拥有 384K token 上下文，在 16 项企业基准中的 13 项上领先 Jev，因此可以在完整仓库或一摞合同的视野下做出决策。
Pay attention to decision models.
From my recent tests, decision models can be extremely transformative for businesses.
Many calls in an agent workflow are structured decisions, so paying frontier prices doesn't make sense.
SubQ Snap from @subquadratic is trained with RL for those calls. It holds 384K tokens of context and leads Jev on 13 of 16 enterprise benchmarks, so it can decide with a whole repo or a stack of contracts in view.
来源：omarsar0 · x.com