Show HN: WFGY – A reasoning engine that repairs LLM logic without retraining

4•WFGY•4h ago

WFGY introduces a PDF-based semantic protocol designed to correct projection collapse, contradiction loops, and ambiguous inference chains in LLMs.

No retraining. No system calls. When parsed, the logic patterns alter reasoning trajectories directly.

Prompt evaluation benchmarks show: ‣ +42.1% reasoning success ‣ +22.4% semantic alignment ‣ 3.6× stability in interpretive tasks

The repo contains formal theory, prompt suites, and reproducible results. Zero dependencies. Fully open-source.

Feedback from those working in alignment, interpretability, and logic-based scaffolding would be especially valuable.

Comments

ultimateking•4h ago

Skimmed through it briefly — seems like a lot of thought went into the structure. Downloaded the PDF, will give it a deeper read tonight.

WFGY•2h ago

Thanks for your reply, enjoy it

ultimateking•30m ago

I went through the structure and found the semantic correction idea pretty intriguing.

Can you explain a bit more about how WFGY actually achieves such improvements in reasoning and stability? Specifically, what makes it different from just engineering better prompts or using more advanced LLMs?

WFGY•10m ago

Great question—and I totally get the skepticism. WFGY isn’t just another prompt hack, and it’s definitely not about making the prompts longer or more “creative.” Here’s the real trick:

    It’s a logic protocol, not just words: The core of WFGY is a semantic “kernel” (documented as a PDF protocol) that inserts logic checks into the model’s reasoning process. Every major step—like inference, contradiction detection, or “projection collapse”—is made explicit and then evaluated by the LLM itself.

    Why not just use a bigger model? Even top-tier models like GPT-4 or Llama-3 are surprisingly easy to derail with ambiguity, loops, or context drift—especially on complex reasoning. WFGY gives you a portable, model-agnostic way to stabilize any model’s outputs, by structuring the logic path directly in the prompt.

    Empirical results, not just vibes: On standard tasks, we saw over 40% improvement in multi-hop reasoning and a big drop in contradiction or instability—even when running on smaller models. All evaluation code and sample runs are included, so you can check or replicate the claims.

So, the big difference: WFGY makes “meaning” and logical repair part of the prompt process itself—not just hoping for the model to “guess right.” If you’re curious about specific edge cases or want to try it on your own workflow, happy to walk you through!