AI TrackerFrontier desk · AEST
Back to news
normalaitech

Kronecker Delta Rule for Long Context Linear Attention

arXiv cs.AI · 20 Oct 2026, 03:56 AEST

KronLSTM is a new linear attention architecture that utilizes the Kronecker Delta Rule for memory revision. It shows competitive performance against existing models, particularly in long-context language tasks.

Key facts

Read aloud

KronLSTM is a new linear attention architecture that utilizes the Kronecker Delta Rule for memory revision. It shows competitive performance against existing models, particularly in long-context language tasks. 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. The resulting memory dynamics can be mathematically described as a Markov chain that converges to a steady-state distribution, while the expected norm of the memory matrix remains uniformly bounded throughout the entire input sequence. 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.

Sources in this cluster