A structural model of future selection in AI using semantic resonance.
Core Equation
E = Q × S_q × t
Where a meaningful question (Q) resonates with a semantic space (S_q) at a critical time (t), producing "existence" (E).
PoR Model Reference
The following are the formal PoR model components used in UGHer:
existence
Formula: E = Q × S_q × t
Description: Base PoR generation from meaningful question, semantic space, and critical time.
Tags: generation, core
self_por_score
Formula: E_self = E_base + ΔE_over × Q_self_factor
Description: Enhances base existence by self-coherence factor.
Tags: self-coherence, enhancement
mismatch
Formula: Mismatch = |E - Q|
Description: Quantifies difference between generated existence and original question pressure.
Tags: analysis, error
semantic_gravity
Formula: grv = PoR_freq × entropy
Description: Computes lexical gravity from PoR firing frequency and resonance entropy.
Tags: semantic, gravity
por_collapse_frequency
Formula: PoR_rate(t) = λ · e^(−λt)
Description: Models decay frequency of PoR over time.
Tags: decay, temporal
por_firing_probability
Formula: (I_q × E_m) / (R_def + 1) > θ
Description: Determines if PoR fires based on input intensity, energy, and definition factor.
Tags: threshold, decision
refire_difference
Formula: ΔE = |E1 - E2|
Description: Measures energy gap between successive PoR firings.
Tags: self-coherence, variance
self_coherence
Formula: φ_C = reference_flow / (|ΔI_in| + |ΔI_out|)
Description: Calculates self-coherence score based on informational flow.
Tags: self-coherence, alignment
gravity_tensor
Formula: G_{ij} = ∇PoR_density · ∇PoR_entropy
Description: Computes semantic gravity tensor from density and entropy gradients.
Tags: semantic, tensor
phase_gradient
Formula: dΦ/dt = k · E · S^γ
Description: Predicts phase expansion rate from energy and semantic density.
Tags: expansion, dynamics
# CLI 実行（5ターン）
python -m unconscious_gravity_exp --log data/sample.parquet --turns 5
# 実験ランナー（2エピソード × 各5ターン）
python examples/run_experiment.py --episodes 2 --turns 5 --out_dir data --log data/sample.parquetGenerated from metadata/semantic_index.json
ugh3-manifest.json には、PoR 発火パターンを評価するための UGH3 指標セットをまとめている。
UGHer · PoR · Semantic Resonance · Unconscious Gravity · Entropic Future Selection · AI Structural Models
Ongoing — semantic matching infrastructure & autonomous PoR loop expansion in progress.
MIT License © 2025 Yuu6798
Repository: unconscious-gravity
GitHub: https://github.com/Yuu6798/unconscious-gravity
For collaboration or questions:
Twitter (X): @kkoo6798kamo
ChatCompletionSampler provides a small wrapper around the OpenAI Chat Completion API to fetch multiple candidate responses. It is used by experiments to gather diverse outputs for analysis.
This project depends on several Python packages including:
- numpy
- pandas
- tqdm
- transformers
- openai