Higher-order pruning of experts in mixture-of-experts language models
This research provides a more effective method for compressing large MoE language models, which is important for deploying them efficiently on hardware with limited resources.
#代理任务
5 stories · 0 in 7 days · 0 in 30 days
Stories, products and papers tagged "代理任务", newest first.
Sep 17
Sep 11
Chinese contentAgent Arena 发布新基准测试，评估模型在百万真实长周期任务中的表现
Agent Arena 发布了新基准测试，用百万真实长周期任务评估模型，比之前的基准更贴近实际应用场景。
May 21
Chinese content谷歌发布 Gemini 3.5 Flash，低延迟低成本领先多项基准
Gemini 3.5 Flash 解决了高智能与低延迟的矛盾，做多步骤代理和编码的开发者可以直接用上，成本还更低，值得一试。
SuperTechFansSource