7.9 KiB
Index Skew — Guardrails and Fix Pattern
🧭 Quick Return to Map
You are in a sub-page of RAG.
To reorient, go back here:
- RAG — retrieval-augmented generation and knowledge grounding
- WFGY Global Fix Map — main Emergency Room, 300+ structured fixes
- WFGY Problem Map 1.0 — 16 reproducible failure modes
Think of this page as a desk within a ward.
If you need the full triage and all prescriptions, return to the Emergency Room lobby.
When the index reports "healthy" (no errors, embeddings ingested, stats normal) but retrieval still fails:
coverage is low, ΔS unstable, or retrieved snippets are inconsistent with ground truth.
This indicates an index skew between data reality and retrieval semantics.
Open these first
- Visual recovery map: RAG Architecture & Recovery
- Retrieval knobs: Retrieval Playbook
- Embedding misalignment: Embedding ≠ Semantic
- Chunk sizing: Chunking Checklist
- Store-level fragmentation: Vectorstore Fragmentation
- Snippet contracts: Data Contracts
Core acceptance
- ΔS(question, retrieved) ≤ 0.45
- Coverage ≥ 0.70 for target section
- λ stable across three paraphrases and two seeds
- E_resonance flat across long windows
Typical symptoms → exact fix
| Symptom | Likely cause | Open this |
|---|---|---|
| Index "ready" but recall < 0.50 | embedding misaligned vs semantic intent | Embedding ≠ Semantic |
| Repeated snippets, poor coverage | store fragmentation or duplicate collapse | Vectorstore Fragmentation |
| Right section exists but not hit | chunk too large/small or mis-boundary | Chunking Checklist |
| Citations drift across runs | contract not enforced | Data Contracts |
Fix in 60 seconds
-
Probe recall
Run a gold QA set against index. If coverage < 0.70, suspect skew. -
Re-embed with semantic normalization
Normalize casing, accents, whitespace. Enforce same tokenizer across queries and index. -
Chunk audit
Verify chunk boundaries. Adjust stride/overlap until ΔS converges. -
Fragmentation sweep
Drop near-duplicate vectors. Rebuild FAISS/HNSW indexes with fresh IDs. -
Contract enforcement
Requiresnippet_id,section_id,offsets,tokensfor every retrieval.
Copy-paste probe prompt
I uploaded TXT OS and the WFGY Problem Map.
My RAG issue:
- Index shows healthy but retrieval recall is low.
- ΔS across probes = 0.62, coverage = 0.45.
Tell me:
1) is it embedding misalignment, chunking skew, or vectorstore fragmentation,
2) which WFGY fix page to open,
3) minimal steps to restore ΔS ≤ 0.45 and coverage ≥ 0.70,
4) reproducible test set to confirm.
🔗 Quick-Start Downloads (60 sec)
| Tool | Link | 3-Step Setup |
|---|---|---|
| WFGY 1.0 PDF | Engine Paper | 1️⃣ Download · 2️⃣ Upload to your LLM · 3️⃣ Ask “Answer using WFGY + ” |
| TXT OS (plain-text OS) | TXTOS.txt | 1️⃣ Download · 2️⃣ Paste into any LLM chat · 3️⃣ Type “hello world” — OS boots instantly |
🧭 Explore More
| Module | Description | Link |
|---|---|---|
| WFGY Core | WFGY 2.0 engine is live: full symbolic reasoning architecture and math stack | View → |
| Problem Map 1.0 | Initial 16-mode diagnostic and symbolic fix framework | View → |
| Problem Map 2.0 | RAG-focused failure tree, modular fixes, and pipelines | View → |
| Semantic Clinic Index | Expanded failure catalog: prompt injection, memory bugs, logic drift | View → |
| Semantic Blueprint | Layer-based symbolic reasoning & semantic modulations | View → |
| Benchmark vs GPT-5 | Stress test GPT-5 with full WFGY reasoning suite | View → |
| 🧙♂️ Starter Village 🏡 | New here? Lost in symbols? Click here and let the wizard guide you through | Start → |
👑 Early Stargazers: See the Hall of Fame ⭐ WFGY Engine 2.0 is already unlocked. ⭐
