RNN Memory Explained: Recurrent Matrix Eigenvalues and LRU Initialization

Robert T. Lange explains that RNN memory length is governed by the eigenvalues of the recurrent matrix, with LRU's ring initialization exploiting this to enable long-term memory. A follow-up post traces the underlying chaotic dynamics to Sompolinsky et al. (1988) and links it to the 2023 LRU work.

2026-11-15 ~ 2026-11-15 · 2 related posts

• RNN memory is set by recurrent matrix eigenvalues — LRU's ring init fixes it — RobertTLange · 2026-11-15
• Chaos in random neural networks: the 1988 paper behind LRU's design — RobertTLange · 2026-11-15