mirror of
https://github.com/onestardao/WFGY.git
synced 2026-04-28 11:40:07 +00:00
6.9 KiB
6.9 KiB
RAG + VectorDB — Global Fix Map
A hub to stabilize retrieval pipelines at the embedding ↔ vector store layer.
Use this folder when retrieval looks healthy on paper but citations, meaning, or stability collapse in practice.
Every fix is structural, measurable, and store-agnostic.
Quick routes to per-page guides
- Metric mismatch → metric_mismatch.md
- Normalization and scaling → normalization_and_scaling.md
- Tokenization and casing → tokenization_and_casing.md
- Chunking ↔ embedding contract → chunking_to_embedding_contract.md
- Vectorstore fragmentation → vectorstore_fragmentation.md
- Dimension mismatch / projection errors → dimension_mismatch_and_projection.md
- Update drift and index skew → update_and_index_skew.md
- Hybrid retriever weight errors → hybrid_retriever_weights.md
- Duplication / near-duplicate collapse → duplication_and_near_duplicate_collapse.md
- Poisoning and contamination → poisoning_and_contamination.md
When to use this folder
- High similarity but wrong meaning.
- Correct facts exist in the corpus but never surface.
- Citations inconsistent or missing across runs.
- Token count spikes after deploy without corpus changes.
- Index “looks fine” yet coverage stays flat.
- Retrieval quality drifts after updates or merges.
- Hybrid retrievers underperform single retrievers.
- Duplicate or poisoned vectors destabilize the store.
Acceptance targets (store-agnostic)
- ΔS(question, retrieved) ≤ 0.45
- Coverage of target section ≥ 0.70
- λ stays convergent across 3 paraphrases and 2 seeds
- E_resonance flat across long windows
- Index skew ≤ 5% after updates
60-second checklist
- Lock metric and analyzer (cosine vs L2 vs dot).
- Normalize embeddings (L2 norm, zero-mean scaling).
- Align tokenization and casing with retriever.
- Enforce chunking ↔ embedding contract.
- Probe ΔS across paraphrases. If ≥0.60, escalate.
- Audit duplication and contamination (hash and dedupe).
- Monitor index skew after every update.
- Verify hybrid retriever weights explicitly.
🔗 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 —
Engineers, hackers, and open source builders who supported WFGY from day one.
⭐ WFGY Engine 2.0 is already unlocked. ⭐ Star the repo to help others discover it and unlock more on the Unlock Board.
