A
licenseC
qualityB
maintenanceEnables a frozen, small local LLM to keep knowledge across sessions through a PPR-walked triple graph, learn continually outside its weights, and offload multi-constraint logic to Z3 through a single self-discovering tool. It also provides model- and task-aware, evidence-cited advice on when memory, learning, or exact reasoning helps or hurts a given model.
1
Apache 2.0