Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,88 @@
|
|
| 1 |
-
|
| 2 |
license: mit
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
language: en
|
| 2 |
license: mit
|
| 3 |
+
tags:
|
| 4 |
+
|
| 5 |
+
pytorch
|
| 6 |
+
|
| 7 |
+
vae
|
| 8 |
+
|
| 9 |
+
predictive-coding
|
| 10 |
+
|
| 11 |
+
neuroscience
|
| 12 |
+
|
| 13 |
+
gabor-splatting
|
| 14 |
+
|
| 15 |
+
biology
|
| 16 |
+
datasets:
|
| 17 |
+
|
| 18 |
+
nielsr/flowers-102
|
| 19 |
+
|
| 20 |
+
Neuro Splat: Artificial Visual Cortex
|
| 21 |
+
|
| 22 |
+
An image is not a million pixels. It is a smooth map and a scatter of small bright wave-packets dropped where the detail lives.
|
| 23 |
+
|
| 24 |
+
PerceptionLab / Antti Luode. Helsinki, June 2026.
|
| 25 |
+
|
| 26 |
+
Do not hype. Do not lie. Just show.
|
| 27 |
+
|
| 28 |
+
What this model is
|
| 29 |
+
|
| 30 |
+
This repository contains the weights (model.pt) and the architecture for a differentiable Gabor wave-packet splatter. It is not a standard pixel-generating AI. It is an experimental representation of the mammalian primary visual cortex (V1), built to test predictive coding and phase-amplitude coupling.
|
| 31 |
+
|
| 32 |
+
The model is a Variational Autoencoder (VAE) trained on the Oxford Flowers-102 dataset.
|
| 33 |
+
|
| 34 |
+
The Encoder acts as the top-down visual pathway, compressing an image into a latent "concept".
|
| 35 |
+
|
| 36 |
+
The Decoder outputs the physical parameters (X, Y, scale, complex phase, frequency) for 512 localized Gabor atoms.
|
| 37 |
+
|
| 38 |
+
The Renderer splats these wave-packets onto a canvas to create the final image via Moiré interference.
|
| 39 |
+
|
| 40 |
+
The Blur is the Prior. Reality is the Sharpness.
|
| 41 |
+
|
| 42 |
+
If you ask this model to generate a flower from scratch, it will produce a soft, blurry, watercolor-like blob. This is the expected, biologically accurate behavior.
|
| 43 |
+
|
| 44 |
+
When an AI (or a brain) lacks raw input, the safest mathematical guess is to average all ambiguous possibilities. It produces the low-frequency "gist" (the Theta carrier).
|
| 45 |
+
|
| 46 |
+
To see the model actually work, you must give it eyes. By hooking this model up to a live webcam, the blurry top-down prior collides with raw, bottom-up reality. The live video supplies the exact phase, the wave-packets lock into alignment, and the blurry gist snaps into razor-sharp perception.
|
| 47 |
+
|
| 48 |
+
How to run it
|
| 49 |
+
|
| 50 |
+
This model requires the original scripts from the ArtificialCortex GitHub Repository.
|
| 51 |
+
|
| 52 |
+
1. Live Cortex Perception (Webcam required)
|
| 53 |
+
|
| 54 |
+
To watch the wave-packets phase-lock to your reality in real time:
|
| 55 |
+
|
| 56 |
+
pip install torch torchvision numpy opencv-python
|
| 57 |
+
python live_cortex.py --model_path model.pt --num_packets 512
|
| 58 |
+
|
| 59 |
+
|
| 60 |
+
2. Standard Generation
|
| 61 |
+
|
| 62 |
+
To see the model hallucinate blurry priors from random latent vectors:
|
| 63 |
+
|
| 64 |
+
python splat_generator.py --mode sample --resume model.pt --num_packets 512
|
| 65 |
+
|
| 66 |
+
|
| 67 |
+
The Honest Ledger
|
| 68 |
+
|
| 69 |
+
What this model proves:
|
| 70 |
+
|
| 71 |
+
Images can be effectively represented and learned as sparse sums of localized Gabor wave-packets.
|
| 72 |
+
|
| 73 |
+
Phase-wrapping discontinuities in neural networks can be solved by outputting complex (a, b) coefficients instead of raw angles (the janus_cabbage skew medium).
|
| 74 |
+
|
| 75 |
+
Predictive coding works: A top-down blurry prior can be actively sharpened by gradient descent against a live bottom-up residual (the webcam).
|
| 76 |
+
|
| 77 |
+
What this model lacks:
|
| 78 |
+
|
| 79 |
+
It is not a photorealistic generator. It is a VAE with an amortized latent space. It competes on biological representation, not on Generative AI benchmarks against Diffusion models.
|
| 80 |
+
|
| 81 |
+
The "Floaters": When subjected to high-stress visual input (like a webcam feed of a room instead of a flower), the optimizer often orphans wave-packets, resulting in bright, isolated dots ("floaters" or "phosphenes"). This occurs because the model currently lacks lateral inhibitory coordination between packets.
|
| 82 |
+
|
| 83 |
+
It is 2D. The current parameters map flat 2D geometries.
|
| 84 |
+
|
| 85 |
+
Lineage
|
| 86 |
+
|
| 87 |
+
A sub-organ of the_artificial_cortex project.
|
| 88 |
+
The generator dreamed a blurry flower because it had nothing to look at. Open its eyes and reality supplies the phase the dream could not; the packets lock, the gist sharpens, and what they cannot yet coordinate, they see as stars.
|