Pathway 大脑启发架构在 SageMaker HyperPod 上开发
原标题：Pathway’s brain-inspired architecture development on Amazon SageMaker HyperPod
AI 摘要
AWS 机器学习博客介绍了 Pathway 公司基于大脑启发的 BDH 架构，该架构在 Amazon SageMaker HyperPod 上进行训练。BDH 架构在潜在空间中进行推理，无需生成中间文本，通过稀疏局部交互和类突触连接维持状态，解决了 Transformer 在训练和推理中的低效问题。Pathway 声称其 1.5 亿参数的 BDH 模型在 ARC-AGI-1 基准上实现了成本效率的新突破。
正文节选
Pathway’s brain-inspired architecture development on Amazon SageMaker HyperPod As AI systems take on more complex tasks, much of the industry’s progress has come from increasing model scale, training data, context length, and inference-time computation. Instead of externalizing reasoning work as a chain-of-thought (generating extra tokens sequentially and feeding them back into later steps), Pathway’s brain-inspired BDH (Dragon Hatchling) performs reasoning in latent space. It learns from exampl