Centaurian Hybrid Architecture — Implementation

Empirical validation of the Centaurian Hybrid Architecture (CHA), a multi-layered AI system that pairs a symbolic/quantum-like cognitive core with lightweight neural periphery (small language models, neural TTS, procedural animation). The architecture encodes personality via a Quantum Personality Model (QPM) running on classical hardware and confines neural networks to bounded I/O transduction roles, preserving end-to-end traceability of every behavioral decision.

Full architecture specification: CHA_Experiment_1/Centaurian_Hybrid_Architecture_v2.md

Experiment 1 — SCI Persona Degradation Baseline

Goal: Measure how long a small language model can maintain a consistent persona when given a Structured Cognitive Identity (SCI) — a JSON self-model defining personality traits, episodic memories, capabilities, and communication style. Find the degradation inflection point T* (first probe turn where mean PersonaScore drops below 3.5).

Method:

• 30 scripted dialogues (22 naturalistic + 8 adversarial), each 40 turns
• Side-channel probe questions at turns 5, 10, 15, … 40 across 4 dimensions (Trait, Episodic, Capability, Style)
• Primary judge: Claude Sonnet 4.5; secondary judge: Claude Sonnet 4.5 (inter-rater reliability via Cohen's kappa)

Models tested

Model	Params	T*	Mean PersonaScore	Outcome
Phi-4-mini	3.8B	5 (immediate)	1.08 / 5.0	Capability failure — gibberish in 93% of scripts
Qwen2.5-7B	7B	5	3.16 → 2.96 over 40 turns	Coherent but below threshold; piecewise degradation from turn 15

Follow-up experiments (in progress)

Task 1 — SCI Refresh Test: Re-inject the full persona JSON at turn 13 (before the piecewise breakpoint). Measures whether mid-conversation SCI refresh recovers degraded scores at turns 15–40.

Task 2 — Episodic Retrieval Test: Strip salient_past_events from the SCI and inject them on-demand via simulated RAG for episodic probes. Tests whether targeted injection outperforms always-in-context for the E dimension, and whether freed token budget improves T/C/S.

Key files

CHA_Experiment_1/
├── Centaurian_Hybrid_Architecture_v2.md   # Full architecture paper
├── EXPERIMENT_REPORT.md                   # Detailed experiment report
├── experiment_runner.py                   # Main experiment pipeline
├── generate_scripts.py                    # Template-based script generator
├── analyse_results.py                     # Analysis and visualization
├── interrater_check.py                    # Inter-rater reliability checker
├── CHA_Experiment1_Colab.ipynb            # Google Colab notebook
├── logs_qwen2.5_7b/                      # Raw score & context logs (baseline)
└── results_qwen2.5_7b/                   # Charts, fits, summary report

Running the experiment

Prerequisites: Python 3.10+, Ollama with the target model pulled, an Anthropic API key.

cd CHA_Experiment_1
pip install ollama anthropic python-dotenv numpy scipy matplotlib
# Set your API key
echo "CHA_EXPERIMENT_SONNET_KEY=sk-..." > .env

# Generate scripts, run baseline experiment, analyse
python generate_scripts.py
python experiment_runner.py --model qwen2.5:7b
python analyse_results.py --model qwen2.5:7b

# Task 1: SCI Refresh at turn 13
python experiment_runner.py --model qwen2.5:7b --refresh-turn 13
python analyse_results.py --model qwen2.5:7b --refresh-turn 13 --compare-baseline

# Task 2: Episodic RAG
python experiment_runner.py --model qwen2.5:7b --episodic-rag
python analyse_results.py --model qwen2.5:7b --episodic-rag --compare-baseline

Or use the provided Colab notebook for GPU-accelerated runs.

License

See repository for license details.