HN Explorer

FrontiersMind

↗ HN profile [ 12.6 ms ] full profile
Karma
0
Created
()
Submissions
0
  1. Grouped Value Attention: Efficient KV Caching via On-Demand Key Reconstruction (arxiv.org)
    1 points 0
  2. Grouped Query Experts: Mixture-of-Experts on GQA Self-Attention (arxiv.org)
    1 points 0