Skip to content

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

22 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

AxiomForge DNA: When a Simple Prompt Beats a Complex Scaffold

An honest experiment in LLM reasoning enhancement.

We built a structured reasoning scaffold, tested it rigorously, and discovered that a single well-crafted prompt outperforms our entire framework. We report this honestly.

Without self-negation, there is no real evolution — only local optimization.


Key Findings

Method Avg Score (80 questions) API Calls Time
A — Single-pass 4.08 1 34s
B — Multi-round (5 turns) 4.58 3 144s
C — Constitutional prompt 5.71 1 51s
D — Adaptive 5.10 2 51s

One good prompt > five rounds of iteration. Constitutional prompting embeds self-evaluation requirements in a single call and consistently outperforms multi-round scaffolding across 5 of 6 question categories.

What We Also Tested (and Failed)

Experiment Result Conclusion
Level 1: Keyword router +0.028 Effectively zero improvement
Level 2: Semantic router (2 runs) -0.417, -1.217 Actively harmful
4,545-param trained brain -0.25, 50% accuracy Worse than fixed prompt

Every attempt to add complexity made things worse. The trained brain independently converged to the same conclusion: "always pick Constitutional."


What This Project Actually Contributes

This is not a better reasoning system. It is an honest record of building one, testing it, and discovering it wasn't needed.

  1. Negative results: Complex scaffolding does not meaningfully improve over a well-crafted single prompt
  2. Self-refutation: The system reviewed its own paper (REJECT, 22 flaws), then helped rewrite it (Accept)
  3. Self-evolution: The scaffold evolved itself (+28%), but with oscillating, unstable convergence
  4. Methodology: A systematic framework for testing whether architectural complexity is justified

Self-Evolution Data

The scaffold modified its own code across generations:

Version Score Status
v2.0 78 Baseline
v2.1 61 Regression, rolled back
v2.2 100 Best (+28%)
v2.3 61 Regression, rolled back
v2.4 93 Auto-stopped after 2 regressions

The Self-Review Experiment

The evolved scaffold (v2.2) reviewed a paper written by the baseline (v2.0):

  • First review: REJECT — found 22 specific flaws, including overclaimed results
  • After honest rewrite: Accept — strengths (40) outweighed flaws (27)

Same data, honest positioning = Accept. Overclaiming = Reject.


Documentation

File Content
docs/PAPER.md Full paper with all experimental data
docs/FINAL_PACKAGE.md Project summary and verified results
docs/ROADMAP.md Frozen roadmap (main line: fixed Constitutional)

Limitations

  1. Small sample sizes (n=3 pilot, n=80 extended)
  2. No human evaluation — all metrics are automated keyword-based proxies
  3. No comparison to Chain-of-Thought or other established baselines
  4. Single model tested (DeepSeek-chat)
  5. Constitutional prompt uses more tokens than single-pass — not equal-compute
  6. Proxy metrics are unvalidated — "more self-criticism keywords" ≠ "better reasoning"
  7. Bias detector is self-referential (same LLM evaluates itself)

The Honest Conclusion

We set out to build a cognitive architecture for LLMs. Through systematic experimentation, we discovered:

  • A simple constitutional prompt renders complex scaffolding unnecessary
  • Routing and classification add overhead without proportional benefit
  • The most valuable capability we built was self-review, not reasoning enhancement
  • A 4,545-parameter brain trained for 200 epochs learned one thing: "always use the simple prompt"

We believe reporting these negative results honestly is more valuable than claiming our scaffold works.


Acknowledgements

Inspired by Andrej Karpathy's vision of AI-driven experimental optimization.

Built with Claude (Anthropic) and DeepSeek APIs.


License

This project uses a hybrid open-core model. See LICENSE for details.


Author: Di Wu, Independent Researcher | HuggingFace Space | wudifff@gmail.com

About

A self-improving cognitive engine for invention. AI that does not just answer — it invents.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages