git:20260108.7a87f36 to git:20260213.b9e0b79
23 added, 12 removed. Audit A to A.
- # Gnn Examples Agent Scaffolding
+ # GNN Examples Module
- ## Module Overview
- **Purpose**: Gnn Examples functionality implementation.
- **Status**: Production Ready
- **Version**: 1.0.0
+ ## Purpose
- ## Core Capabilities
- - Gnn Examples processing and management
- - Integration with GNN pipeline
+ This module contains **reference GNN model files** that serve as both documentation and test fixtures. The primary example is a complete Active Inference POMDP agent specification.
- ## API Reference
- See the source code for detailed method documentation.
+ ## Example Files
- ## Testing
- Run tests via `pytest src/tests/`
+ | File | Description |
+ |------|-------------|
+ | `actinf_pomdp_agent.md` | Complete GNN specification for a discrete POMDP Active Inference agent with 4 hidden states, 5 observations, 3 control states, and full parameterization |
+
+ ## Key Sections in the POMDP Example
+
+ The reference file demonstrates all required GNN sections:
+
+ - **StateSpaceBlock**: Defines hidden states (s), observations (o), and control states (u) with dimensionality annotations
+ - **Connections**: Maps state-observation relationships using GNN edge notation (`->`, `<->`)
+ - **InitialParameterization**: Provides matrices A, B, C, D with full numeric values
+ - **Equations**: Active Inference update equations (variational free energy, posterior, expected free energy)
+ - **Time**: Dynamic/discrete temporal configuration
+
+ ## For AI Agents
+
+ 1. **Use as reference** when parsing or generating GNN files — this is the canonical example
+ 2. **Use for testing** — this file is the reference for round-trip validation across all 23 formats
+ 3. **Model structure**: The POMDP agent demonstrates proper Active Inference variable naming (A=likelihood, B=transition, C=preference, D=prior, E=policy prior)