3 Commits

Author	SHA1	Message	Date
root	ac663fca16	feat: protocol base lib + rule-based parser MVP + 31 tests	2026-06-01 11:12:45 +08:00
root	bf0af3709c	fix: correct architecture docs - tensor protocol, heuristic scheduling, LLM dual output, memory update mechanism ... Key corrections: - Subnets communicate via Standard Tensor Protocol, not natural language - Core scheduling is heuristic (non-linear), not pipeline-based - LLM subnet has dual output: natural language→user, expression tensor→Core - Core aggregates subnet results → memory update parameter tensor → memory subnet - Added TrulyMEM-style graph DB (Neo4j) reference in memory-subnet.md - LLM is demoted to specialized language generation engine - All docs include correct tensor schemas, architecture diagrams, and training plans	2026-06-01 09:20:14 +08:00
root	398d35c50b	chore: init project docs - NextAgent cognitive decoupled architecture ... - README.md with architecture diagram and project overview - docs/architecture-overview.md: core design, scheduling, RefPack protocol - docs/routing-compression-subnet.md: multi-modal encoding, compression training - docs/memory-subnet.md: MANN, GNN, retrieve-pull mechanism training - docs/logic-trace-subnet.md: RLHF+PRM, MCTS, self-correction training - docs/llm-subnet.md: SFT, RLHF fine-tuning, multi-language alignment	2026-06-01 08:55:27 +08:00