What question did this study set out to answer?

To introduce Temporal Activation Maps (TAM) as a lightweight mechanism for tracking activation patterns in large language models while enabling computational efficiency.

March 5, 2026Open Access

Temporal Activation Maps: A Lightweight Architecture for Experience-Driven Adaptation in Large Language Models

Key Points

To introduce Temporal Activation Maps (TAM) as a lightweight mechanism for tracking activation patterns in large language models while enabling computational efficiency.
Proposed a lightweight metadata layer for tracking parameter cluster activation patterns.
Established a three-tier consolidation hierarchy inspired by biological memory systems.
Differentiated TAM from existing models like MoE, LoRA, RAG, and sparse autoencoders.
Estimated storage overhead and resource requirements for deployment.
TAM allows experience-driven computational prioritization without modifying model weights.
Storage overhead for TAM is estimated between 80-230 MB per deployment context.
Proposed architecture could enhance memory system efficiency in large language models.

Abstract

This paper proposes Temporal Activation Maps (TAM), a lightweight metadata layer that tracks parameter cluster activation patterns across inference events in large language models, enabling experience-driven computational prioritization without modifying model weights. TAM introduces a three-tier consolidation hierarchy mirroring biological memory systems and provides a procedural memory complement to existing declarative memory systems. Estimated storage overhead is 80–230 MB per deployment context. The paper presents the architecture, differentiates TAM from MoE, LoRA, RAG, and sparse autoencoders, estimates resource requirements, and proposes experimental validation methods. Co-developed with an AI system (Claude, Anthropic Opus 4.6).

Temporal Activation Maps: A Lightweight Architecture for Experience-Driven Adaptation in Large Language Models

Key Points

Abstract

Cite This Study