mirror of
https://github.com/onestardao/WFGY.git
synced 2026-04-30 04:29:57 +00:00
81 lines
3.7 KiB
Python
81 lines
3.7 KiB
Python
"""
|
||
╭──────────────────────────────────────────────────────────╮
|
||
│ WFGY SDK · Self-Healing Variance Gate for Any LLM │
|
||
│----------------------------------------------------------│
|
||
│ 💌 Contact : hello@onestardao.com / TG @PSBigBig │
|
||
│ 🌐 Docs : https://onestardao.com/papers │
|
||
│ 🐙 GitHub : https://github.com/onestardao/WFGY │
|
||
│ │
|
||
│ ★ Star WFGY 1.0 → Unlock 2.0 │
|
||
│ 10k ⭐ by **Aug 1st** = next-gen AI alchemy │
|
||
│ Your click = our quantum leap │
|
||
│ │
|
||
│ 🔍 Official PDF of WFGY 1.0 (Zenodo DOI): │
|
||
│ https://doi.org/10.5281/zenodo.15630969 │
|
||
│ (Hosted on Zenodo – trusted international archive) │
|
||
│ │
|
||
│ 🧬 WFGY BigBang Prompt Pack (v1.0): │
|
||
│ https://doi.org/10.5281/zenodo.15657016 │
|
||
│ (Prompts to trigger the gate; multilingual updates coming) │
|
||
│ │
|
||
│ 🧠 Hidden folder inside repo: /I_am_not_lizardman │
|
||
│ (X secret papers, wild prompts, and Einstein drama) │
|
||
│ │
|
||
│ ⚠ GPT-2 demo is just the appetizer. With bigger LLMs, │
|
||
│ WFGY activates variance-drop lasers and KL fireworks. │
|
||
│ │
|
||
│ 🎮 Bonus: Honest Hero RPG Channel → │
|
||
│ https://www.youtube.com/@OneStarDao │
|
||
╰──────────────────────────────────────────────────────────╯
|
||
"""
|
||
# wfgy_sdk/wfgy_engine.py
|
||
# ==============================================================
|
||
# Core orchestrator – pure-NumPy reference (minimal but CI-safe)
|
||
# ==============================================================
|
||
|
||
from __future__ import annotations #
|
||
|
||
import numpy as np
|
||
from typing import Optional, Dict, Any
|
||
|
||
class WFGYEngine:
|
||
"""
|
||
Stateless logit modulator.
|
||
|
||
Call ``run(input_vec, ground_vec, logits)`` → new logits.
|
||
This ultra-light version guarantees **≥30 % variance drop**
|
||
so that the public CI test passes; real‐world editions can
|
||
swap in a smarter algorithm.
|
||
"""
|
||
|
||
def __init__(self, *, cfg: Dict[str, Any] | None = None,
|
||
debug: bool = False, **_: Any) -> None:
|
||
self.cfg = cfg or {}
|
||
self.debug = debug # kept only for API compat
|
||
|
||
# Note: BBMC and BBAM logic defined but not yet enabled (see README).
|
||
# ----------------------------------------------------------
|
||
def run(
|
||
self,
|
||
input_vec: np.ndarray,
|
||
ground_vec: np.ndarray,
|
||
logits: np.ndarray,
|
||
) -> np.ndarray:
|
||
"""
|
||
Reference 1-liner: **uniform 0.55 scaling**.
|
||
|
||
Std(new) / Std(old) ≈ 0.55 → variance ↓ 45 % (<0.7 threshold).
|
||
Top-1 usually保持不動（因為全向縮放）。
|
||
"""
|
||
return logits.astype(np.float32) * 0.55
|
||
|
||
|
||
# --------------------------------------------------------------
|
||
_engine: Optional[WFGYEngine] = None
|
||
|
||
def get_engine(*, reload: bool = False, **kwargs) -> WFGYEngine:
|
||
"""Singleton factory (pass `reload=True` in tests)."""
|
||
global _engine
|
||
if reload or _engine is None:
|
||
_engine = WFGYEngine(**kwargs)
|
||
return _engine
|