Linear Attention Architectures: Mechanisms, Trade-offs, and Cross-Layer Routing

Part of

Relevance-Gated Hyperdimensional Memory Streaming