You Could Have Come Up with Kimi Delta Attention | Hacker News

Hacker News users debated a write-up on Kimi Delta Attention, with many focused on its unusual bra-ket notation and whether it makes the method harder to follow.

Commenters also argued over the underlying math, including how close the approach is to linear attention and DeltaNet-style ideas.

A major theme was whether Kimi Delta Attention is a genuine algorithmic breakthrough or mainly a compute-driven refinement enabled by more hardware.

The discussion reflected a broader ML debate: how much progress comes from new attention ideas versus scaling compute, and how notation shapes research clarity.