AI 摘要
1m 标准与超高上下文效率是让我兴奋之处
1m Standard and ultra high context efficiency is what me excites me
Structural Innovation & Ultra-High Context Efficiency 🔹 Novel Attention: Token-wise compression + DSA (DeepSeek Sparse Attention). 🔹 Peak Efficiency: World-le...