Latent Space 新节目 High Taste Testers 首期试播
Latent Space 试播新节目 High Taste Testers 第一期,把旧金山午餐桌上的闲聊搬上节目,话题从 Dev Day 延伸到最新模型与助手。同期 AI 新闻汇总中,Meta 发布 816B 总参数、42B 激活的 MoE 开放权重模型 Muse Spark Open,Artificial Analysis 智能指数 39,为美国开放权重模型最高;Reflection 以 Apache 2.0 发布 501B 总参数、23B 激活的 Beam 1.0,支持 1M 上下文。
-aware routing: Giving a router the session history passes 3.3 more points of tasks at the same offloading rate. Cache costs make switching up to the expensive model unprofitable (Microsoft). FDE-Gym: The best agent configuration fully resolves only 14 of 95 real customer-delivery tasks, mostly failing on unstated constraints (FDE-Gym). Harness nondeterminism: OpenHands looks stable mainly because it hardcodes temperature 0. Normalizing tool outputs fixes much of the variance in other harnesses
匹配字段正文

