76.2% on LoCoMo with local embedding/summarization models
LLM and memory as mirrors
Reflective Memory for AI Agents
When the Singularity happened, we were sitting on a park bench in Berkeley.
