Metadata-Version: 2.4
Name: arc2-chimera
Version: 1.0.0
Summary: Research proof-of-concept. ~2-5% accuracy on ARC-AGI. NOT an ARC Prize submission. GPU compute-shader playground (OpenGL / moderngl) exploring primitives for symbolic reasoning.
Author-email: Francisco Angulo de Lafuente <agnuxo1@gmail.com>
License: Apache-2.0
Project-URL: Homepage, https://github.com/Agnuxo1/ARC2_CHIMERA
Project-URL: Repository, https://github.com/Agnuxo1/ARC2_CHIMERA
Project-URL: Issues, https://github.com/Agnuxo1/ARC2_CHIMERA/issues
Keywords: opengl,compute-shader,arc-agi,research,gpu,proof-of-concept
Classifier: Development Status :: 3 - Alpha
Classifier: Intended Audience :: Science/Research
Classifier: License :: OSI Approved :: Apache Software License
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.10
Classifier: Programming Language :: Python :: 3.11
Classifier: Programming Language :: Python :: 3.12
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
Requires-Python: >=3.10
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: numpy>=1.24
Requires-Dist: Pillow>=10.0
Provides-Extra: gpu
Requires-Dist: moderngl>=5.8; extra == "gpu"
Requires-Dist: glfw>=2.5; extra == "gpu"
Provides-Extra: dev
Requires-Dist: pytest>=7; extra == "dev"
Requires-Dist: build>=1.0; extra == "dev"
Requires-Dist: twine>=4.0; extra == "dev"
Dynamic: license-file

# ARC2-CHIMERA

> **GPU compute-shader playground exploring OpenGL-based primitives for
> symbolic reasoning. Accuracy on ARC is a research baseline, not a
> competitive submission.**

[![License: Apache-2.0](https://img.shields.io/badge/License-Apache_2.0-blue.svg)](https://opensource.org/licenses/Apache-2.0)
[![Python 3.10+](https://img.shields.io/badge/python-3.10+-blue.svg)](https://www.python.org/downloads/)
[![OpenGL 4.3+](https://img.shields.io/badge/OpenGL-4.3%2B-green.svg)](https://www.opengl.org/)
[![Status: Research PoC](https://img.shields.io/badge/status-research--PoC-orange.svg)](./PROJECT_STATUS.md)

## Please read [PROJECT_STATUS.md](./PROJECT_STATUS.md) before anything else.

**TL;DR:** this repository is a **research proof-of-concept**. Current
benchmark accuracy on ARC-AGI is **~2-5%** on 100-task subsets — well below
the 85% target of the ARC Prize. **This is not an ARC Prize submission and
should not be described as one.** The reusable value is the GPU compute-shader
architecture, not the reasoning engine.

## What the package is

`arc2-chimera` exposes the v10 OpenGL-based architecture as an installable
Python package so the shader primitives can be reused:

- Matmul / GELU kernels implemented as GLSL compute shaders.
- Cellular-automaton texture update kernels.
- Diffusion-style update passes.
- Holographic-memory read/write passes.
- A GPU-texture state loop (no CPU round-trips between frames).

All of this runs on any OpenGL 4.3+ GPU via `moderngl`. There is no PyTorch
or TensorFlow dependency.

## Install

```bash
pip install arc2-chimera          # core (numpy, Pillow)
pip install arc2-chimera[gpu]     # add moderngl + glfw for real GPU runs
pip install arc2-chimera[dev]     # pytest / build / twine
```

## CLI

```bash
arc2-chimera info                                      # prints status + disclaimer
arc2-chimera evaluate --task-file task.json --output out.json
```

## Repository layout

| Path | Description |
|------|-------------|
| `arc2_chimera/` | Installable package (wraps v10). |
| `CHIMERA_ARC_OpenGL/` | v5-v10 development tree. v10 is the primary code path. |
| `CHIMERA/` | Early Kaggle-style script (historical). |
| `CHIMERA_ENIGMA/` | Experimental biological-memory variant. |
| `CHIMERA_RIVER/` | Ensemble/flow variant. |
| `archive/` | Quarantined scripts (including an unverified submission file). |
| `tests/` | Offline pytest suite. |

## What works / what doesn't

See [PROJECT_STATUS.md](./PROJECT_STATUS.md) for the full list. In short: the
GPU kernels work and are worth reusing. The symbolic reasoning layer on top
of them does not yet perform on ARC.

## License

Apache-2.0. Author: Francisco Angulo de Lafuente <agnuxo1@gmail.com>.
