Research
Paper proposes memory-centric continual learning framework for AI agents
The MCCL framework judges agent memory by whether it improves future actions rather than retrieval accuracy, with a three-axis taxonomy, a two-loop lifecycle, an evaluation protocol and a scaling model.
Every outlet that ran it 1 article
Hugging Face PapersResearchLead