LLM Wiki vs Vector RAG Compiler Karpathy Pattern v2.6

Local Knowledge Architecture & Context Window Bloat Benchmarker

Raw Ingest Corpus (8)

Note Editor Live Extraction Tagging
Cytoscape Topology Active
Graph Legend
Compiled Wiki Page Node
Raw Vector Chunk (RAG)
Extracted Entity Anchor

Retrieval Footprint Telemetry 4.38x Efficiency

LLM Wiki Engine
Total Context Tokens
420
Context Redundancy
8%
Entity Recall Score
0.96
Vector RAG Engine
Total Context Tokens
1,840
Context Redundancy
62%
Entity Recall Score
0.68
LLM Wiki Context Payload Synthesized Pages
Vector RAG Top-K Chunks 4 Chunks Naive Stitch
Active Proof Target: llm-wiki-rag-compiler-67
Compilation Latency: 12ms (Local)