ResearchResearch paperReasoning & Planning · Training & Scaling · Image, Video & 3D Generation2 sources · Oct 19, 2026

Kronecker Delta Rule for Long Context Linear Attention

We propose KronLSTM, a linear attention architecture based on the Kronecker Delta Rule, a novel update rule for associative matrix memories.

KronLSTM extends the xLSTM family with a memory-revision mechanism that corrects targeted associations while modifying the memory only along a single direction in matrix space.
This minimally invasive memory revision reduces interference with previously stored information compared with delta-rule updates, making it particularly well suited for long-context language modeling.
Experiments show that KronLSTM is competitive with, and often outperforms, existing linear attention architectures, including models of the (Gated) DeltaNet family, with particularly strong performance on long context benchmarks.
These results establish the Kronecker Delta Rule as an effective and analytically tractable foundation for stable associative memory in recurrent linear attention architectures.

[1]Kronecker Delta Rule for Long Context Linear Attention
Hugging Face Daily Papers · Oct 19, 12:00 AM
We propose KronLSTM, a linear attention architecture based on the Kronecker Delta Rule, a novel update rule for associative matrix memories.
KronLSTM extends the xLSTM family with a memory-revision mechanism that corrects targeted associations while modifying the memory only along a single direction in matrix space.
[2]Kronecker Delta Rule for Long Context Linear Attention
arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 19, 05:56 PM · same content

Extractive summary: sentences quoted from the sources.

Before this