POLER-META

This repository is a meta-framework for POLER: a rule-driven cognitive/dynamical system. It focuses on formalism, invariants, and operator evolution, not on training a large model.

What this repo is

• A formal specification of POLER concepts (state, operators, resonance, phase inertia).
• A meta-layer: rules about how rules change (operator evolution).
• A place to run experiments via local stubs + remote/cloud executors.

What this repo is NOT

• Not an LLM project.
• Not “another neural net core”.
• Not a promise of “AGI”.
• Not a training zoo of models.

Core goal (single target)

Implement a mechanism where the system can change the transition law (not only the state), while maintaining explicit invariants.

Structure

• formalism/ — definitions: state, operator, phase, observer
• meta/ — rule graphs, transition rules, invariants
• executors/ — local stub + remote execution adapters
• experiments/ — notebooks / tests
• docs/ — notes, failure log

Quick start (local)

python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt