Every score comes with the model's reasoning on the paper page.
"AttentionIndex: Accelerating Learned Sparse Attention with a Causal Inverted-File Index" scored 6.0/10 predicted impact on @kurateorg — Clarity 7.0 · Difficulty 7.0 · Translational 6.5