2026-10-10 · America/Los_Angeles · 社区动态 · #16
I Built a O(NlogN) attention system that retains 97% accuracy over long context (MQAR)[P]
ALHR- Adaptive learnable Hierarchical Routing is a static binary tree based system that uses learnable functions to REDUCE the amount of keys used. It takes less memory and SCALES much better VRAM with tokens
热度 44.3 / 100;排名与评分保留该期记录。