DAMI: Dual-Axis Memory Initialization for Enhanced Long-Context Retrieval in Gated Linear Attention
Paper recorded by Signals 4 on 2026-10-30 in cs.AI. Abstract reproduced from arXiv; link to the original below.
Published 2026-10-30 on arXiv · recorded by Signals 4 on 2026-11-02
Category: cs.AI · 人工智能 · first seen 2026-11-02
Abstract
Linear attention scales linearly with sequence length, but its finite-state memory can limit long-context retrieval. In gated linear recurrences, memory retention depends on state decay, which is shaped jointly by gate parameters A_log and dt_bias and by input activations. In controlled Gated DeltaNet (GDN) experiments, both parameters remain highly correlated with their initial values after train
Read on arXiv →
Cite this page: DAMI: Dual-Axis Memory Initialization for Enhanced Long-Context Retrieval in Gated Linear Attention: the #57 most recent of 920 cs.AI papers we have recorded (as of 2026-10-30). Source: Signals 4 (Signals API) — https://data.jiangzhang.ca/signals4/t/papers/dami-dual-axis-memory-initialization-for-enhanced-long-context-retrieval-in-gate.html
Free to quote with attribution to “Signals 4 (Signals API)”. Machine-readable:
papers.json
Get 4 AI signals a day by email — free.
Get 4 AI signals a day by email — free