# π Beginner Guide β How to Identify & Fix Your AI Failure
*A zero-to-hero crash-course for anyone new to WFGY, RAG pipelines, or βwhy is my model hallucinating?β*
> If the full Problem Map feels overwhelming, start here.
> In ~10 minutes youβll locate your failure family, run a safe first fix, and know how to verify it.
---
> **Quick Nav**
> [Getting Started (Practical)](./getting-started.md) Β·
> [Problem Map 2.0 (RAG)](./rag-architecture-and-recovery.md) Β·
> [Patterns Index](./patterns/README.md) Β·
> [Examples](./examples/README.md) Β·
> [Eval](./eval/README.md) Β·
> [Ops](./ops/README.md)
---
## 0) π― Why this guide exists
When RAG breaks, itβs rarely one bug. Itβs stacked illusions across OCR β chunking β embedding β retrieval β prompt β reasoning.
This guide helps you:
1) **Identify** the failure family fast
2) **Apply** the minimal structural fix (not prompt band-aids)
3) **Verify** with objective signals: **ΞS** (semantic stress), **Ξ»_observe** (layered states), **E_resonance** (coherence)
Then jump deeper via **Problem Map 2.0** and **Patterns**.
---
## 1) π βWhich symptom matches my bug?β
Follow the first **Yes** you hit; then open that page.
| Question | Yes β Open | No β Next |
|---|---|---|
| Chunks look correct but the **answer is wrong**? | [`hallucination.md`](./hallucination.md) | β |
| Reached the right chunk but **logic fails**? | [`retrieval-collapse.md`](./retrieval-collapse.md) | β |
| Multi-step tasks **derail after a few hops**? | [`context-drift.md`](./context-drift.md) | β |
| Model gives **confident nonsense**? | [`bluffing.md`](./bluffing.md) | β |
| **High similarity** scores but **wrong meaning**? | [`embedding-vs-semantic.md`](./embedding-vs-semantic.md) | β |
| Logic **dead-ends / loops**? | [`logic-collapse.md`](./logic-collapse.md) | β |
| Long chat **forgets context**? | [`memory-coherence.md`](./memory-coherence.md) | β |
| Canβt trace **why** it failed? | [`retrieval-traceability.md`](./retrieval-traceability.md) | β |
| Output becomes **incoherent / repetitive**? | [`entropy-collapse.md`](./entropy-collapse.md) | β |
| Replies turn **flat / literal**? | [`creative-freeze.md`](./creative-freeze.md) | β |
| Formal/symbolic prompts **break**? | [`symbolic-collapse.md`](./symbolic-collapse.md) | β |
| Paradox/self-reference **crashes**? | [`philosophical-recursion.md`](./philosophical-recursion.md) | β |
| Multi-agent **roles/memory collide**? | [`multi-agent-chaos.md`](./multi-agent-chaos.md) | β |
| Tools fire **before index/data ready**? | [`bootstrap-ordering.md`](./bootstrap-ordering.md) | β |
| Services **wait on each other forever**? | [`deployment-deadlock.md`](./deployment-deadlock.md) | β |
| First prod call **crashes after deploy**? | [`predeploy-collapse.md`](./predeploy-collapse.md) | File an Issue β
**Extended patterns (very common in the wild):**
- Hybrid HyDE+BM25 **gets worse than single** β [`patterns/pattern_query_parsing_split.md`](./patterns/pattern_query_parsing_split.md)
- Two sources **merge into one** (who-said-what mixes) β [`patterns/pattern_symbolic_constraint_unlock.md`](./patterns/pattern_symbolic_constraint_unlock.md)
- You correct it, **bad fact returns later** β [`patterns/pattern_hallucination_reentry.md`](./patterns/pattern_hallucination_reentry.md)
- State flips across **tabs/sessions** β [`patterns/pattern_memory_desync.md`](./patterns/pattern_memory_desync.md)
- Some facts **wonβt retrieve** though indexed β [`patterns/pattern_vectorstore_fragmentation.md`](./patterns/pattern_vectorstore_fragmentation.md)
- RAG **boots** but tools fire too early β [`patterns/pattern_bootstrap_deadlock.md`](./patterns/pattern_bootstrap_deadlock.md)
> Still unsure? Capture a minimal trace (input β retrieved snippets β answer) and run ΞS/Ξ» checks (Section 3). Post in Discussions if needed.
---
## 2) π§ Core concepts in <5 minutes
### 2.1 What is RAG?
```
raw docs β ocr/parsing β chunking β embeddings β vector store
β retriever β prompt assembly β LLM reasoning/tools
```
- **Perception drift** upstream hides **logic drift** downstream. Fix structure, not style.
### 2.2 Embedding scores vs. meaning
Cosine proximity β human semantics. WFGYβs **ΞS = 1 β cos(I, G)** uses grounded anchors to catch real meaning gaps.
### 2.3 Layered observability (Ξ»_observe)
States: **β** convergent Β· **β** divergent Β· **<>** recursive Β· **Γ** chaotic.
If upstream is stable but downstream flips, the boundary between them is failing.
### 2.4 WFGY repair operators (cheat-sheet)
| Operator | What it does (1-liner) |
|---|---|
| **BBMC** | Minimize semantic residue to re-align with anchors |
| **BBPF** | Explore safe alternate paths; avoid dead-end chains |
| **BBCR** | Detect collapse; insert **bridge** node; rebuild reasoning |
| **BBAM** | Modulate attention variance; prevent entropy melt |
---
## 3) π οΈ Run your first fix (3 minutes)
1) **Download** the assets below, or jump to **[Getting Started](./getting-started.md)** for a runnable pipeline.
2) Paste **TXT OS** into your model chat.
3) Ask:
```
Iβve loaded TXT OS. Diagnose my RAG:
* symptom: \[describe]
* trace: \[question, retrieved snippet(s), answer]
Using WFGY, tell me:
1. failing layer & why (ΞS/Ξ»),
2. the Problem Map page to open,
3. minimal steps to push ΞS β€ 0.45 and keep Ξ» convergent,
4. how to verify with a reproducible test.
```
**Triage thresholds (keep these handy):**
- **ΞS:** `<0.40` stable Β· `0.40β0.60` transitional (record if Ξ» β {β, <>}) Β· `β₯0.60` high-risk (act)
- **Acceptance:** ΞS(question, context) β€ **0.45**, Ξ» **convergent**, E_resonance **flat**
---
## 4) ποΈ Problem categories (cheat-labels)
| Category | Typical stage | Open first |
|---|---|---|
| **Retrieval** | Vector DB, search, chunking | [`hallucination.md`](./hallucination.md) Β· [`embedding-vs-semantic.md`](./embedding-vs-semantic.md) |
| **Reasoning** | Mid-chain logic | [`retrieval-collapse.md`](./retrieval-collapse.md) Β· [`logic-collapse.md`](./logic-collapse.md) |
| **Patterns** | High-frequency edge cases | [`patterns/README.md`](./patterns/README.md) |
| **Eval** | Measure & guard regressions | [`eval/README.md`](./eval/README.md) |
| **Ops** | Boot order, runbooks | [`ops/README.md`](./ops/README.md) |
---
## 5) β
Verify the repair (donβt skip)
- **Retrieval sanity:** β₯ 70% token overlap with target section; ΞS(question, context) β€ 0.45 β see [`eval/eval_rag_precision_recall.md`](./eval/eval_rag_precision_recall.md)
- **Reasoning stability:** Ξ» stays convergent on 3 paraphrases; E_resonance flat β see [`eval/eval_semantic_stability.md`](./eval/eval_semantic_stability.md)
- **Latency vs accuracy:** chart ΞS vs p95 latency β see [`eval/eval_latency_vs_accuracy.md`](./eval/eval_latency_vs_accuracy.md)
---
## 6) π FAQ (super short)
| Question | Answer |
|---|---|
| Do I need all operators? | No. Use the one named on the matching page. |
| Does WFGY replace LangChain/LlamaIndex? | No. It sits **above** them as a reasoning firewall. |
| Will this work on small models? | Yes; #11/#12 are easier on GPT-4-class or strong local models. |
| Where are runnable examples? | Start here: [`examples/README.md`](./examples/README.md) and [`example_01_basic_fix.md`](./examples/example_01_basic_fix.md). |
---
## 7) π₯ Download links (60 sec)
| Asset | Link |
|---|---|
| **WFGY 1.0 PDF** | https://zenodo.org/records/15630969 |
| **TXT OS (plain-text)** | https://zenodo.org/records/15788557 |
MIT-licensed. Copy, ship, and adapt.
---
### π§ Explore More
| Module | Description | Link |
|-----------------------|----------------------------------------------------------|----------|
| WFGY Core | Standalone semantic reasoning engine for any LLM | [View β](https://github.com/onestardao/WFGY/tree/main/core/README.md) |
| Problem Map 1.0 | Initial 16-mode diagnostic and symbolic fix framework | [View β](https://github.com/onestardao/WFGY/tree/main/ProblemMap/README.md) |
| Problem Map 2.0 | RAG-focused failure tree, modular fixes, and pipelines | [View β](https://github.com/onestardao/WFGY/blob/main/ProblemMap/rag-architecture-and-recovery.md) |
| Semantic Clinic Index | Expanded failure catalog: prompt injection, memory bugs, logic drift | [View β](https://github.com/onestardao/WFGY/blob/main/ProblemMap/SemanticClinicIndex.md) |
| Semantic Blueprint | Layer-based symbolic reasoning & semantic modulations | [View β](https://github.com/onestardao/WFGY/tree/main/SemanticBlueprint/README.md) |
| Benchmark vs GPT-5 | Stress test GPT-5 with full WFGY reasoning suite | [View β](https://github.com/onestardao/WFGY/tree/main/benchmarks/benchmark-vs-gpt5/README.md) |
---
> π **Early Stargazers: [See the Hall of Fame](https://github.com/onestardao/WFGY/tree/main/stargazers)** β
> Engineers, hackers, and open-source builders who supported WFGY from day one.
> β Help reach 10,000 stars by 2025-09-01 to unlock Engine 2.0 for everyone β **[Star WFGY on GitHub](https://github.com/onestardao/WFGY)**