nested-learning

This paper proposes a memory-augmented multi-agent architecture using nested learning, continuum memory systems, and semantic caching to mitigate hallucination in LLM pipelines, achieving significant reductions in factual errors while improving operational efficiency.

Proposes Federated Nested Learning (FedNL), a framework that reformulates federated learning as a three-level nested optimization system, enabling collaborative training of self-referential memories for test-time adaptation to handle Non-IID data and long-tail distributions.

Google researchers introduce Nested Learning, a new architecture that replaces the Transformer by treating models as nested optimization problems, solving catastrophic forgetting and achieving 100% long-context memory stability.