First systematic study of hybrid linear attention: PAS and ISP activation spikes explained

jiqizhixin · x · 2026-09-10

StartLux, Tsinghua, UCAS, HKU, University of Sydney and Columbia present the first systematic study of Hybrid Linear Attention (HLA) models: 5 linear attention architectures, 6 hybrid configurations, 5 data domains, 12 public checkpoints from 1.2B to 397B parameters (Qwen3.5, Kimi Linear, Nemotron-H, Zamba).

Using Massive Activations as a probe, the team identifies two core phenomena:

The work shows how a few full attention layers reshape computation inside hybrid models.

Original post →

More from Research

Research channel →