DeltaPhase: High-Expressivity O(N) Complex Phase Matrix Delta-Rule Memory & Lerp Spectral Transformer for Subquadratic LLMs.
deep-learning pytorch associative-memory state-space-models delta-rule linear-attention fast-weight-programmers llm-architecture subquadratic-attention complex-phase
-
Updated
Aug 11, 2026 - Python