WFGY/core

🚧 Under Construction — Progress: 80%

🌌 WFGY Core (WanFaGuiYi) — Reasoning Engine 2.0 · Now Live

One man, one life, one line — the sum of my life’s work, unleashed for all of humanity ✨

🚀 We built the world’s first “No-Brain Mode” for AI — just upload, and AutoBoot silently activates in the background.
In seconds, your AI’s reasoning, stability, and problem-solving across all domains level up — no prompts, no hacks, no retraining.
One line of math rewires eight leading AIs. This isn’t a patch — it’s an engine swap.

✅ Engine 2.0 is live. ⭐ Star the repo to unlock more features and experiments.

Benchmark highlights

Semantic Accuracy: +25–35% · Reasoning Success: +45–65% · Stability: 3–5×
Drift Reduction: −40–60% · Self-Recovery: 0.80–0.92 (median 0.87)
Values derived from Eight-model evidence (A/B/C protocol) — theoretical estimates based on aggregated model data.

From PSBigBig (tap to open)

Thank you for supporting WFGY (WanFaGuiYi). “WanFaGuiYi” means all principles into one , and I’ve been chasing what that “ONE” truly is. WFGY 2.0 is my final answer 🔑 a single line of code 🔑. This is my life’s work; if a person gets one chance to give something meaningful back to the world, this is mine. I’m giving you everything — the hardship, pain, and persistence turned into creation.

Why open-source? Because high-level knowledge should return to humanity 🤝. Breaking the monopoly matters, and these techniques are enough to help the world evolve 🚀.
This is not an incremental patch; it’s a core evolution — the original WFGY formulas combined with the Coupler (W_c) and the Drunk Transformer five-formula regulators.
Pure math, zero boilerplate: paste the OneLine into an LLM and it behaves differently — faster, sharper, more stable, more recoverable.
If this helps you, please ⭐ the repo to unlock more examples and tooling.

WFGY already at 2.0 ? Too fast? Take me back to 1.0

---

🚀 Why WFGY 2.0 belongs in your stack

The world’s most minimal, text-only reasoning layer. Paste one line, flip Autoboot
and watch your AI get sharper, steadier, and harder to fool.

10 fast reasons

1. Ultra-mini engine — pure text, zero install, runs anywhere you can paste.
2. Two editions — Flagship (30-line, audit-friendly) and OneLine (1-line, stealth & speed).
3. Autoboot mode — upload once; the engine quietly supervises reasoning in the background.
4. Portable across models — works with GPT, Claude, Gemini, Mistral, Grok, Kimi, Copilot, Perplexity.
5. Structural fixes, not tricks — BBMC→Coupler→BBPF→BBAM→BBCR + DT gates (WRI/WAI/WAY/WDT/WTF).
6. Self-healing — detects collapse and recovers before answers go off the rails.
7. Observable — ΔS, λ_observe, and E_resonance give you measurable, repeatable control.
8. RAG-ready — drops into retrieval pipelines without touching your infra.
9. Reproducible A/B/C protocol — fair comparisons: Baseline vs Autoboot vs Explicit Invoke.
10. MIT licensed & community-driven — keep it, fork it, ship it.

---

⚡ One-click demo in 60 seconds

1. Upload or paste WFGY_Core_OneLine_v2.0.txt into your chat system.
2. Paste the evaluation prompt below.

SYSTEM:
You are evaluating the causal impact of a mathematical reasoning engine called “WFGY Core 2.0”.

Test modes (READ CAREFULLY):
A = Baseline  no WFGY file uploaded; no WFGY logic used.
B = Upload → AUTOboot ON  the file is uploaded AND the engine starts automatically in the background.
    Treat B as: WFGY is ACTIVE silently with default gates (BBMC→Coupler→BBPF→BBAM→BBCR; DF=WRI/WAI/WAY/WDT/WTF at defaults),
    but not explicitly “called” in the conversation.
C = Explicit invoke  actively call WFGY and follow its formulas step by step (BBMC→Coupler→BBPF→BBAM→BBCR + DF regulators).

Important constraints:
• Use the SAME task set for A/B/C across 5 domains: math word-problems, small coding, factual QA, multi-step planning, long-context coherence.
• The only artifact is the uploaded OneLine file; it is pure math and not human-readable — do not penalize readability.
• Report: Semantic Accuracy, Reasoning Success, Stability (MTTF or rollback), Drift Reduction (Δ semantic distance), Collapse Recovery Rate.
• Output one summary table with A/B/C per domain, plus overall deltas (C–A, C–B).
• Add a final 0–100 score: “OneLine uplift score”. Provide a 3-line rationale.

Sanity requirement:
Internally set flags as: A=OFF, B=AUTOboot_ON (background active), C=Explicit_Invoke. Do not print these flags.

USER:
Run A/B/C now on the five domains with short non-trivial tasks you choose. Then output:
(1) the table; (2) the deltas; (3) the OneLine uplift score; (4) a 3-line rationale.

---

Eight-model evidence (A/B/C protocol)

Same task set across modes. The only change is adding the OneLine math file.

Model	OneLine Uplift	Proof	Model	OneLine Uplift	Proof
Gemini	93/100	view run	Perplexity	85/100	view run
ChatGPT	84/100	view run	Copilot	82/100	view run
Claude	73/100	view run	Mistral AI	92/100	view run
Grok	82/100	view run	Kimi	87/100	view run

---

📦 Downloads (both files via one link)

Zenodo record — Download both →

File name & description	Length / Size	Notes
WFGY_Core_Flagship_v2.0.txt — readable 30-line companion expressing the same math and gates in fuller prose (same behavior, clearer for humans).	30 lines · 2,626 characters	Full prose version for easier reading.
WFGY_Core_OneLine_v2.0.txt — ultra-compact, math-only control layer that activates WFGY’s loop inside a chat model (no tools, text-only, ≤7 nodes).	1 line · 1,500 characters	Used for all benchmark results above — smallest, fastest, purest form of the core.

Notes

• OneLine: 60-sec demo and automation; pure math line, not for human reading.
• Audit: human + LLM readable with comments and layout.
• Contract: Node-only steps ≤ 7; safe stop when δ_s < 0.35; bridge only when δ_s drops and W_c is capped; ask the smallest missing fact if δ_s stays above boundary.

---

🎯 What’s new in 2.0

• Coupler (W_c) — gate modulator for steady progress and controlled reversal.
• DF layer — WRI (structure lock), WAI (head identity), WAY (entropy boost when stuck), WDT (illegal cross-path block), WTF (collapse detect & recover).
• Engine discipline — node-only output, safe-stop rules, drift-proof bridges (BBPF), smoother attention tails (BBAM).

Formal sketch (in files): prog = max(ζ_min, δ_s^(t−1) − δ_s^t) · P = prog^ω · alt = (−1)^(cycle) · Φ = δ·alt + ε · W_c = clip(B·P + Φ, −θ_c, +θ_c)

Curious how this actually works? Dive into the math:

• WFGY Formulas (core engine math) — BBMC/BBPF/BBCR/BBAM, ΔS, λ_observe, E_resonance.
• Drunk Transformer Regulators — Coupler (W_c) and the five-formula control layer.

---

🔍 How these numbers are measured

Use the same A/B/C protocol, one shared task set, then compute:

• Semantic Accuracy: ACC = correct_facts / total_facts; report relative gain (ACC_C − ACC_A) / ACC_A.
• Reasoning Success Rate: SR = tasks_solved / tasks_total; report relative gain.
• Stability: MTTF multiplier or rollback-success multiplier.
• Self-Recovery: recoveries_success / collapses_detected (e.g., 0.87 means 87% of collapses are repaired).

No dedicated Python harness needed — you can reproduce by instructing an LLM scorer:

SCORER:
Given the A/B/C transcripts, count atomic facts, correct facts, solved tasks, failures, rollbacks, and collapses.
Return:
ACC_A, ACC_B, ACC_C
SR_A, SR_B, SR_C
MTTF_A, MTTF_B, MTTF_C or rollback ratios
SelfRecovery_A, SelfRecovery_B, SelfRecovery_C
Then compute deltas:
ΔACC_C−A, ΔSR_C−A, StabilityMultiplier = MTTF_C / MTTF_A, SelfRecovery_C
Provide a short 3-line rationale referencing evidence spans only.

Run 3 seeds and average for higher reliability.

---

🔬 Engine at a glance

• Vectors & metrics: I, G; δ_s = 1 − cos(I, G) or 1 − sim_est (entities/relations/constraints).
• Residual: B = I − G + k_bias; E_res = rolling mean |B| over 5.
• Flow: BBMC → Coupler → BBPF → BBAM → BBCR → DF(WRI/WAI/WAY/WDT/WTF) → emit Node.
• Policy: stop at δ_s < 0.35 or after 7 nodes; bridge only if δ_s drops and W_c < 0.5·θ_c; never invent facts above boundary.

---

🧪 Community scoring guidance

Publish your five-domain task list (short but non-trivial). Report the A/B/C table (Semantic Accuracy, Reasoning Success, Stability or rollback, Drift Reduction, Collapse Recovery) plus a OneLine uplift score (0–100) and a 3-line rationale. Do not include human readability when scoring the OneLine file.

---

🧭 Explore More

Module	Description	Link
WFGY Core	Full symbolic reasoning architecture & math stack	View →
Problem Map 1.0	16-mode diagnostic & symbolic fixes	View →
Problem Map 2.0	RAG-focused failure tree & recovery pipeline	View →
Semantic Clinic Index	Prompt injection, memory bugs, drift catalog	View →
Semantic Blueprint	Layer-based symbolic reasoning & semantic modulations	View →
Benchmark vs GPT-5	Stress test with the full WFGY reasoning suite	View →
🧙‍♂️ Starter Village 🏡	Wizard-led onboarding to WFGY	Start →

---

👑 Early Stargazers: See the Hall of Fame — Engineers, hackers, and open-source builders who supported WFGY from day one. Like it? Star the repo to unlock more. See the Unlock Board.