Downloads · 30 days
8
4% of all-time downloads
RRFSavantMetaLogicV2
antonypamo/RRFSavantMetaLogicV2
RRFSavantMetaLogicV2 is a text classification model from antonypamo. Use it when you need a label for a piece of text. It is set up for scikit-learn. The card lists the license as mit.
RRFSavantMetaLogicV2 is a lightweight, interpretable meta-quality classifier for the Resonance of Reality Framework (RRF) / Savant Engine ecosystem. It predicts the quality of a (Question, Answer) pair from a 15-dimen…
All-time downloads
202
Public
Repo size
487 KB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.joblib486 KB · 63%
At a glance
- Task
- Text Classification
- Library
- scikit-learn
- License
- mit
- Access
- Public
- Created
- Dec 13, 2025
- Updated
- May 13, 2026
- SHA
- e8d7f152
Try a prompt
Base models
Model card
From the Hugging Face model README
RRFSavantMetaLogicV2
RRFSavantMetaLogicV2 is a lightweight, interpretable meta-quality classifier for the Resonance of Reality Framework (RRF) / Savant Engine ecosystem.
It predicts the quality of a (Question, Answer) pair from a 15-dimensional RRF–Savant meta-state vector, combining 7 continuous spectral/energy metrics and 8 one-hot Φ-node ontology indicators.
Overview
- Model type: Logistic Regression (binary classification)
- Input: 15D feature vector (float)
- Output:
predictlabel (0/1) +predict_probaprobabilities - Goal: fast, auditable gating layer for evaluation/reranking pipelines
- Artifact:
logreg_rrf_savant.joblib
Feature Contract (15D)
This model expects features generated by rrf_state_to_features, in the exact order below.
Continuous RRF Meta-State Metrics (7)
| Index | Feature | Description |
|---|---|---|
| 0 | phi | Energy saturation of the embedding (0–1) |
| 1 | omega | Frequency resonance of the embedding (0–1) |
| 2 | coherence | Spectral smoothness + concentration; internal consistency |
| 3 | S_RRF | Spectral Smoothness: preference for low average frequencies |
| 4 | C_RRF | Spectral Concentration: fraction of energy in dominant frequency |
| 5 | hamiltonian_energy | Squared L2 norm of the embedding vector |
| 6 | dominant_frequency | Frequency with highest FFT power |
One-Hot Φ-node Ontology (8)
Exactly one is 1.0, the rest 0.0.
| Index | Φ-node | Meaning |
|---|---|---|
| 7 | Φ0_seed | Foundational / genesis state |
| 8 | Φ1_geometric | Geometric / structural reasoning |
| 9 | Φ2_gauge_dirac | Gauge fields and Dirac operators |
| 10 | Φ3_log_gravity | Logarithmic gravity and scale effects |
| 11 | Φ4_resonance | Harmonic and resonant coherence |
| 12 | Φ5_memory_symbiosis | Memory coupling and persistence |
| 13 | Φ6_alignment | Alignment, ethics, constraint consistency |
| 14 | Φ7_meta_agi | Meta-cognition and AGI-level reasoning |
Intended Use
Recommended use cases:
- Meta-quality gating for LLM outputs
- Routing (approve/escalate/fallback) based on
P(high_quality) - Evaluation API scoring layer with audit-friendly properties
- Reranking and filtering in Savant/RRF pipelines
Usage
import joblib
import numpy as np
clf = joblib.load("logreg_rrf_savant.joblib")
x = np.array([[
phi, omega, coherence,
S_RRF, C_RRF, hamiltonian_energy, dominant_frequency,
Phi0, Phi1, Phi2, Phi3, Phi4, Phi5, Phi6, Phi7
]], dtype=float)
y = int(clf.predict(x)[0])
p_high = float(clf.predict_proba(x)[0][1])
print("pred:", y, "p_high:", p_high)
- Task
- Text Classification
- Library
- scikit-learn
- License
- mit
- Languages
- en
- Created
- Dec 13, 2025
- Updated
- May 13, 2026