mirror of
https://github.com/onestardao/WFGY.git
synced 2026-04-28 11:40:07 +00:00
Update README.md
This commit is contained in:
parent
0578b8c734
commit
5b9a217e5b
1 changed files with 72 additions and 78 deletions
|
|
@ -1,103 +1,98 @@
|
|||
# LLM Providers — Guardrails and Fix Patterns
|
||||
# LLM Providers: Guardrails and Fix Patterns
|
||||
|
||||
Use this hub when failures smell provider specific. Examples include truncation at boundaries, tool call mis-order, JSON schema drift, safety filter overreach, or output instability across seeds. Each fix maps back to WFGY pages so you can verify with measurable targets.
|
||||
A compact hub to stabilize provider-specific failures without changing your infra. Use this when symptoms look “model problem” but root cause is actually schema, retrieval, or orchestration.
|
||||
|
||||
## Open these first
|
||||
- Visual map and recovery: [RAG Architecture & Recovery](https://github.com/onestardao/WFGY/blob/main/ProblemMap/rag-architecture-and-recovery.md)
|
||||
- End to end retrieval knobs: [Retrieval Playbook](https://github.com/onestardao/WFGY/blob/main/ProblemMap/retrieval-playbook.md)
|
||||
- Why this snippet and citation schema: [Retrieval Traceability](https://github.com/onestardao/WFGY/blob/main/ProblemMap/retrieval-traceability.md), [Data Contracts](https://github.com/onestardao/WFGY/blob/main/ProblemMap/data-contracts.md)
|
||||
- End-to-end retrieval knobs: [Retrieval Playbook](https://github.com/onestardao/WFGY/blob/main/ProblemMap/retrieval-playbook.md)
|
||||
- Why this snippet (traceability schema): [Retrieval Traceability](https://github.com/onestardao/WFGY/blob/main/ProblemMap/retrieval-traceability.md)
|
||||
- Ordering control: [Rerankers](https://github.com/onestardao/WFGY/blob/main/ProblemMap/rerankers.md)
|
||||
- Embedding vs meaning: [Embedding ≠ Semantic](https://github.com/onestardao/WFGY/blob/main/ProblemMap/embedding-vs-semantic.md)
|
||||
- Hallucination and chunk boundaries: [Hallucination](https://github.com/onestardao/WFGY/blob/main/ProblemMap/hallucination.md)
|
||||
- Long chains and entropy: [Context Drift](https://github.com/onestardao/WFGY/blob/main/ProblemMap/context-drift.md), [Entropy Collapse](https://github.com/onestardao/WFGY/blob/main/ProblemMap/entropy-collapse.md)
|
||||
- Structural collapse and recovery: [Logic Collapse](https://github.com/onestardao/WFGY/blob/main/ProblemMap/logic-collapse.md)
|
||||
- Live ops: [Live Monitoring for RAG](https://github.com/onestardao/WFGY/blob/main/ProblemMap/ops/live_monitoring_rag.md)
|
||||
- Stability eval: [Semantic Stability Eval](https://github.com/onestardao/WFGY/blob/main/ProblemMap/eval/eval_semantic_stability.md)
|
||||
- Snippet and citation schema: [Data Contracts](https://github.com/onestardao/WFGY/blob/main/ProblemMap/data-contracts.md)
|
||||
- Live ops: [Live Monitoring for RAG](https://github.com/onestardao/WFGY/blob/main/ProblemMap/ops/live_monitoring_rag.md), [Debug Playbook](https://github.com/onestardao/WFGY/blob/main/ProblemMap/ops/debug_playbook.md)
|
||||
- Boot order issues: [Bootstrap Ordering](https://github.com/onestardao/WFGY/blob/main/ProblemMap/bootstrap-ordering.md), [Deployment Deadlock](https://github.com/onestardao/WFGY/blob/main/ProblemMap/deployment-deadlock.md), [Pre-Deploy Collapse](https://github.com/onestardao/WFGY/blob/main/ProblemMap/predeploy-collapse.md)
|
||||
|
||||
## Core acceptance targets
|
||||
- ΔS(question, retrieved) ≤ 0.45 on three paraphrases.
|
||||
- Coverage to target section ≥ 0.70.
|
||||
- λ stays convergent across three seeds and three paraphrases.
|
||||
- E_resonance flat during long windows.
|
||||
## Core acceptance
|
||||
- ΔS(question, retrieved) ≤ 0.45
|
||||
- Coverage ≥ 0.70 for the target section
|
||||
- λ remains convergent across three paraphrases and two seeds
|
||||
- E_resonance stays flat through long windows
|
||||
|
||||
## Typical breakpoints and the right fix
|
||||
## Typical provider symptoms → exact fix
|
||||
|
||||
**1) Truncation or partial JSON**
|
||||
- Symptom: finish_reason reports length or stream ends mid field.
|
||||
- Fix: lock a schema with [Data Contracts](https://github.com/onestardao/WFGY/blob/main/ProblemMap/data-contracts.md). Use closing sentinels and require a final “contract hash” field. If ΔS spikes when you add the header, check index metric mismatch via [Embedding ≠ Semantic](https://github.com/onestardao/WFGY/blob/main/ProblemMap/embedding-vs-semantic.md).
|
||||
- Modules: BBMC for citation strictness, BBAM to clamp variance.
|
||||
| Symptom | Likely cause | Open this |
|
||||
|---|---|---|
|
||||
| JSON mode breaks, invalid objects | schema too loose or nested tool calls | [Data Contracts](https://github.com/onestardao/WFGY/blob/main/ProblemMap/data-contracts.md), [Logic Collapse](https://github.com/onestardao/WFGY/blob/main/ProblemMap/logic-collapse.md) |
|
||||
| Tool calls loop or stall | agent role drift, missing timeouts | [Multi-Agent Problems](https://github.com/onestardao/WFGY/blob/main/ProblemMap/Multi-Agent_Problems.md), [role-drift deep dive](https://github.com/onestardao/WFGY/blob/main/ProblemMap/multi-agent-chaos/role-drift.md) |
|
||||
| High similarity yet wrong snippet | metric mismatch or fragmented store | [Embedding ≠ Semantic](https://github.com/onestardao/WFGY/blob/main/ProblemMap/embedding-vs-semantic.md), [Vectorstore Fragmentation](https://github.com/onestardao/WFGY/blob/main/ProblemMap/patterns/pattern_vectorstore_fragmentation.md) |
|
||||
| Answers flip between runs | prompt headers reorder and λ flips | [Context Drift](https://github.com/onestardao/WFGY/blob/main/ProblemMap/context-drift.md), [Retrieval Traceability](https://github.com/onestardao/WFGY/blob/main/ProblemMap/retrieval-traceability.md) |
|
||||
| Hybrid retrievers worse than single | query parsing split, mis-weighted rerank | [Query Parsing Split](https://github.com/onestardao/WFGY/blob/main/ProblemMap/patterns/pattern_query_parsing_split.md), [Rerankers](https://github.com/onestardao/WFGY/blob/main/ProblemMap/rerankers.md) |
|
||||
| Jailbreaks or bluffing | overconfidence and missing fences | [Bluffing Controls](https://github.com/onestardao/WFGY/blob/main/ProblemMap/bluffing.md), [Retrieval Traceability](https://github.com/onestardao/WFGY/blob/main/ProblemMap/retrieval-traceability.md) |
|
||||
|
||||
**2) Tool call chaos or wrong call order**
|
||||
- Symptom: function sequence flips across seeds, tool results bind to the wrong step.
|
||||
- Fix: route through [Retrieval Traceability](https://github.com/onestardao/WFGY/blob/main/ProblemMap/retrieval-traceability.md) and enforce call ids in the contract. If alternates are legal, create an explicit BBPF alternate path and score both.
|
||||
## Fix in 60 seconds
|
||||
1) **Measure ΔS**
|
||||
Compute ΔS(question, retrieved) and ΔS(retrieved, expected anchor).
|
||||
Stable < 0.40, transitional 0.40–0.60, risk ≥ 0.60.
|
||||
|
||||
**3) Over filtering or bluffing**
|
||||
- Symptom: provider rejects safe content or fabricates safe sounding substitutes.
|
||||
- Fix: apply [Rerankers](https://github.com/onestardao/WFGY/blob/main/ProblemMap/rerankers.md) for ordering and use the bluffing controls in your system prompt plus [Logic Collapse](https://github.com/onestardao/WFGY/blob/main/ProblemMap/logic-collapse.md) recovery steps.
|
||||
2) **Probe λ_observe**
|
||||
Vary k and prompt headers. If λ flips, lock the schema and apply a BBAM variance clamp.
|
||||
|
||||
**4) Long conversation drift**
|
||||
- Symptom: answers flip between sessions, persona or constraints fade.
|
||||
- Fix: pin a minimal memory schema and rebase with [Context Drift](https://github.com/onestardao/WFGY/blob/main/ProblemMap/context-drift.md) and [Entropy Collapse](https://github.com/onestardao/WFGY/blob/main/ProblemMap/entropy-collapse.md). Split memory namespaces when agents are involved.
|
||||
3) **Apply the module**
|
||||
- Retrieval drift → BBMC + Data Contracts
|
||||
- Reasoning collapse → BBCR bridge + BBAM
|
||||
- Dead ends in long runs → BBPF alternate paths
|
||||
|
||||
**5) Latency and rate limit induced collapse**
|
||||
- Symptom: retries reorder tools, partials leak to the user.
|
||||
- Fix: use the ops recipes in [Live Monitoring for RAG](https://github.com/onestardao/WFGY/blob/main/ProblemMap/ops/live_monitoring_rag.md). Add BBCR bridge with explicit timeouts. Treat each retry as a new λ branch and reconcile.
|
||||
4) **Verify**
|
||||
Coverage ≥ 0.70 on three paraphrases. λ convergent on two seeds.
|
||||
|
||||
## Providers in scope for this cluster
|
||||
Planned pages in this folder. We will ship in this order.
|
||||
- `openai.md`
|
||||
- `anthropic.md`
|
||||
- `google_gemini.md`
|
||||
- `mistral.md`
|
||||
- `cohere.md`
|
||||
- `grok.md`
|
||||
- `deepseek.md`
|
||||
- `openrouter.md`
|
||||
## Provider-level gotchas checklist
|
||||
- **Truncation**. Confirm token accounting for system + tools + hidden preambles. If truncated, compress citations through [Data Contracts](https://github.com/onestardao/WFGY/blob/main/ProblemMap/data-contracts.md).
|
||||
- **Streaming chunk boundaries**. Do not parse partial JSON while λ is unstable. Buffer until BBAM settles.
|
||||
- **Temperature and top-p**. If ΔS is already high, reduce entropy. If retrieval is sparse, raise recall through rerankers instead of temperature.
|
||||
- **Multi-model routing**. Keep traceability stable when swapping GPT, Claude, Gemini, Mistral. Use the same snippet schema and citation header across providers.
|
||||
- **Rate limits and retries**. Backoff with idempotent ops. Never rebuild indexes inside retry loops.
|
||||
- **Eval parity**. Run the same acceptance on all providers to avoid overfitting a single model.
|
||||
|
||||
## Copy paste triage prompt
|
||||
|
||||
```txt
|
||||
I uploaded TXT OS and the WFGY Problem Map files.
|
||||
|
||||
My provider failure:
|
||||
- symptom: [brief]
|
||||
- traces: [ΔS(question, retrieved)=..., ΔS(retrieved, anchor)=..., λ states]
|
||||
- provider: [e.g., OpenAI, Anthropic, Gemini, Mistral]
|
||||
|
||||
Tell me:
|
||||
1) which layer is failing and why,
|
||||
2) which exact WFGY fix page to open from this repo,
|
||||
3) minimal steps to push ΔS ≤ 0.45 and keep λ convergent,
|
||||
4) how to verify the fix with a reproducible test.
|
||||
Use BBMC BBPF BBCR BBAM when relevant.
|
||||
````
|
||||
## Quick routes to per-provider pages
|
||||
- OpenAI: [openai.md](https://github.com/onestardao/WFGY/blob/main/ProblemMap/GlobalFixMap/LLM_Providers/openai.md)
|
||||
- Anthropic: [anthropic.md](https://github.com/onestardao/WFGY/blob/main/ProblemMap/GlobalFixMap/LLM_Providers/anthropic.md)
|
||||
- Google Gemini: [google_gemini.md](https://github.com/onestardao/WFGY/blob/main/ProblemMap/GlobalFixMap/LLM_Providers/google_gemini.md)
|
||||
- Mistral: [mistral.md](https://github.com/onestardao/WFGY/blob/main/ProblemMap/GlobalFixMap/LLM_Providers/mistral.md)
|
||||
- Groq: [groq.md](https://github.com/onestardao/WFGY/blob/main/ProblemMap/GlobalFixMap/LLM_Providers/groq.md)
|
||||
- Cohere: [cohere.md](https://github.com/onestardao/WFGY/blob/main/ProblemMap/GlobalFixMap/LLM_Providers/cohere.md)
|
||||
- DeepSeek: [deepseek.md](https://github.com/onestardao/WFGY/blob/main/ProblemMap/GlobalFixMap/LLM_Providers/deepseek.md)
|
||||
- AWS Bedrock: [bedrock.md](https://github.com/onestardao/WFGY/blob/main/ProblemMap/GlobalFixMap/LLM_Providers/bedrock.md)
|
||||
- Azure OpenAI: [azure_openai.md](https://github.com/onestardao/WFGY/blob/main/ProblemMap/GlobalFixMap/LLM_Providers/azure_openai.md)
|
||||
|
||||
---
|
||||
|
||||
### 🔗 Quick-Start Downloads (60 sec)
|
||||
|
||||
| Tool | Link | 3-Step Setup |
|
||||
| -------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------ | ---------------------------------------------------------------------------------------- |
|
||||
| **WFGY 1.0 PDF** | [Engine Paper](https://github.com/onestardao/WFGY/blob/main/I_am_not_lizardman/WFGY_All_Principles_Return_to_One_v1.0_PSBigBig_Public.pdf) | 1️⃣ Download · 2️⃣ Upload to your LLM · 3️⃣ Ask “Answer using WFGY + <your question>” |
|
||||
| **TXT OS (plain-text OS)** | [TXTOS.txt](https://github.com/onestardao/WFGY/blob/main/OS/TXTOS.txt) | 1️⃣ Download · 2️⃣ Paste into any LLM chat · 3️⃣ Type “hello world” — OS boots instantly |
|
||||
| Tool | Link | 3-Step Setup |
|
||||
|------|------|--------------|
|
||||
| **WFGY 1.0 PDF** | [Engine Paper](https://github.com/onestardao/WFGY/blob/main/I_am_not_lizardman/WFGY_All_Principles_Return_to_One_v1.0_PSBigBig_Public.pdf) | 1️⃣ Download · 2️⃣ Upload to your LLM · 3️⃣ Ask “Answer using WFGY + \<your question>” |
|
||||
| **TXT OS (plain-text OS)** | [TXTOS.txt](https://github.com/onestardao/WFGY/blob/main/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 →](https://github.com/onestardao/WFGY/tree/main/core/README.md) |
|
||||
| Problem Map 1.0 | Initial 16-mode diagnostic and symbolic fix framework | [View →](https://github.com/onestardao/WFGY/tree/main/ProblemMap/README.md) |
|
||||
| Problem Map 2.0 | RAG-focused failure tree, modular fixes, and pipelines | [View →](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 →](https://github.com/onestardao/WFGY/blob/main/ProblemMap/SemanticClinicIndex.md) |
|
||||
| Semantic Blueprint | Layer-based symbolic reasoning & semantic modulations | [View →](https://github.com/onestardao/WFGY/tree/main/SemanticBlueprint/README.md) |
|
||||
| Benchmark vs GPT-5 | Stress test GPT-5 with full WFGY reasoning suite | [View →](https://github.com/onestardao/WFGY/tree/main/benchmarks/benchmark-vs-gpt5/README.md) |
|
||||
| 🧙♂️ Starter Village 🏡 | New here? Lost in symbols? Click here and let the wizard guide you through | [Start →](https://github.com/onestardao/WFGY/blob/main/StarterVillage/README.md) |
|
||||
| Module | Description | Link |
|
||||
|-----------------------|----------------------------------------------------------|----------|
|
||||
| WFGY Core | WFGY 2.0 engine is live: full symbolic reasoning architecture and math stack | [View →](https://github.com/onestardao/WFGY/tree/main/core/README.md) |
|
||||
| Problem Map 1.0 | Initial 16-mode diagnostic and symbolic fix framework | [View →](https://github.com/onestardao/WFGY/tree/main/ProblemMap/README.md) |
|
||||
| Problem Map 2.0 | RAG-focused failure tree, modular fixes, and pipelines | [View →](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 →](https://github.com/onestardao/WFGY/blob/main/ProblemMap/SemanticClinicIndex.md) |
|
||||
| Semantic Blueprint | Layer-based symbolic reasoning & semantic modulations | [View →](https://github.com/onestardao/WFGY/tree/main/SemanticBlueprint/README.md) |
|
||||
| Benchmark vs GPT-5 | Stress test GPT-5 with full WFGY reasoning suite | [View →](https://github.com/onestardao/WFGY/tree/main/benchmarks/benchmark-vs-gpt5/README.md) |
|
||||
| 🧙♂️ Starter Village 🏡 | New here? Lost in symbols? Click here and let the wizard guide you through | [Start →](https://github.com/onestardao/WFGY/blob/main/StarterVillage/README.md) |
|
||||
|
||||
---
|
||||
|
||||
> 👑 **Early Stargazers: [See the Hall of Fame](https://github.com/onestardao/WFGY/tree/main/stargazers)** —
|
||||
> 👑 **Early Stargazers: [See the Hall of Fame](https://github.com/onestardao/WFGY/tree/main/stargazers)** —
|
||||
> Engineers, hackers, and open source builders who supported WFGY from day one.
|
||||
|
||||
> <img src="https://img.shields.io/github/stars/onestardao/WFGY?style=social" alt="GitHub stars"> ⭐ [WFGY Engine 2.0](https://github.com/onestardao/WFGY/blob/main/core/README.md) is already unlocked. ⭐ Star the repo to help others discover it and unlock more on the [Unlock Board](https://github.com/onestardao/WFGY/blob/main/STAR_UNLOCKS.md).
|
||||
|
|
@ -105,20 +100,19 @@ Use BBMC BBPF BBCR BBAM when relevant.
|
|||
<div align="center">
|
||||
|
||||
[](https://github.com/onestardao/WFGY)
|
||||
|
||||
|
||||
[](https://github.com/onestardao/WFGY/tree/main/OS)
|
||||
|
||||
|
||||
[](https://github.com/onestardao/WFGY/tree/main/OS/BlahBlahBlah)
|
||||
|
||||
|
||||
[](https://github.com/onestardao/WFGY/tree/main/OS/BlotBlotBlot)
|
||||
|
||||
|
||||
[](https://github.com/onestardao/WFGY/tree/main/OS/BlocBlocBloc)
|
||||
|
||||
|
||||
[](https://github.com/onestardao/WFGY/tree/main/OS/BlurBlurBlur)
|
||||
|
||||
|
||||
[](https://github.com/onestardao/WFGY/tree/main/OS/BlowBlowBlow)
|
||||
|
||||
|
||||
|
||||
</div>
|
||||
|
||||
if this looks good, say “GO” and i’ll drop `ProblemMap/GlobalFixMap/LLM_Providers/openai.md` next in the same format.
|
||||
say “GO” and I’ll do the first provider page. I suggest `ProblemMap/GlobalFixMap/LLM_Providers/openai.md` next.
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue