Memory management and RAG systems
Groups episodic memory, retrieval-augmented generation, response analysis, and recording-related documents.
Recommended reading order
- ⭐ RAG 2.0 Implementation Roadmap - Remaining Functionality and Future Phases (NEW)
- ⭐ RAG System Technical Specification (Japanese-Optimized) (NEW)
- ⭐ RAG 2.0 Evaluation Guide (Japanese Search Accuracy) (NEW)
- Episodic Memory implementation
- Detailed explanation of EvoSpikeNet RAG system
- RAG Search Japanese Particle Problem Verification Report
- Whole brain simulation query response analysis
- Simulation data recording/analysis guide
- Simulation data recording and analysis function
2026-05-22 update summary (RAG v2 implementation reflected)
- Split the RAG v2 preprocessing pipeline into
SudachiTokenizer,EntityRecognizer, andQueryExpander - Added
rule|llm|hybridQueryExpander backends, with LLM expansion parsed via a fixed JSON schema - Added preprocessing health API (
GET /api/v2/rag/preprocessing/health) - Added quality guards (diversity / redundancy) with automatic fallback to deterministic rule expansion
- Added observability for guard statistics, time-series history, and query-hash summaries
See also: - RAG 2.0 Implementation Roadmap - Remaining Functionality and Future Phases - Environment Variables Operational Policy - Release notes
Complete page list
- BRain Language Vocab README
- ⭐ RAG 2.0 Implementation Roadmap - Remaining Functionality and Future Phases (NEW)
- ⭐ RAG System Technical Specification (Japanese-Optimized) (NEW)
- ⭐ RAG 2.0 Evaluation Guide (Japanese Search Accuracy) (NEW)
- Comparative study of LLM integration strategies in distributed brain systems
- Detailed explanation of EvoSpikeNet RAG system
- LLM Training System
- RAG Search Japanese Particle Problem Verification Report
- SWITCH TO EVOSPIKENET LM
- Episodic Memory implementation
- Simulation data recording/analysis guide
- Simulation data recording and analysis function
- Whole brain simulation query response analysis
- Brain language architecture specification