AbstractPhil / geovocab-patch-maker

GeoVocab Patch Maker

A geometric vocabulary extractor that reads structural properties from latent patches — and proved that text carries the same geometric structure as images.

This is a two-tier gated geometric transformer trained on 27 geometric primitives (point through channel) in 8×16×16 voxel grids. It extracts 17-dimensional gate vectors (explicit geometric properties) and 256-dimensional patch features (learned representations) from any compatible latent input.

What It Does

Takes an (8, 16, 16) tensor — originally voxel grids, but proven to work on adapted FLUX VAE latents and text-derived latent patches — and produces per-patch geometric descriptors:

from geometric_model import load_from_hub, extract_features

model = load_from_hub()
gate_vectors, patch_features = extract_features(model, patches)
# gate_vectors:   (N, 64, 17)  — interpretable geometric properties
# patch_features: (N, 64, 256) — learned representations

Gate Vector Anatomy (17 dimensions)

Dimensions 0–10 are local (intrinsic to each patch, no cross-patch info). Dimensions 11–16 are structural (relational, computed after attention sees neighborhood context).

Architecture

(8, 16, 16) input
    ↓
PatchEmbedding3D → (B, 64, 64)         # 64 patches of 32 voxels each
    ↓
Stage 0: Local Encoder + Gate Heads     # dims, curvature, boundary, axes
    ↓
proj([embedding, local_gates]) → (B, 64, 128)
    ↓
Stage 1: Bootstrap Transformer ×2       # standard attention with local context
    ↓
Stage 1.5: Structural Gate Heads        # topology, neighbors, surface role
    ↓
Stage 2: Geometric Transformer ×2       # gated attention modulated by all 17 gates
    ↓
Stage 3: Classification Heads           # 27-class shape recognition

The geometric transformer blocks use gate-modulated attention: Q and K are projected from [hidden, all_gates] , V is multiplicatively gated, and per-head compatibility scores are computed from gate interactions.

The Rosetta Stone Discovery

This model was used as the analyzer in the GeoVAE Proto experiments , which proved that text descriptions produce 2.5–3.5× stronger geometric differentiation than actual images when projected through a lightweight VAE into this model's patch space.

Three architecturally different text encoders converge to ±5% of each other — the geometric structure is in the language, not the encoder. This model reads it.

Training

Trained on procedurally generated multi-shape superposition grids (2–4 overlapping geometric primitives per sample, 27 shape classes). Two-tier gate supervision with ground truth computed from voxel analysis:

• Local gates : dimensionality from axis extent, curvature from fill ratio, boundary from partial occupancy
• Structural gates : topology from 3D convolution neighbor counting, surface role from neighbor density thresholds

200 epochs, achieving 93.8% recall on shape classification with explicit geometric property prediction as auxiliary objectives.

Files

Usage

import torch
from geometric_model import SuperpositionPatchClassifier, load_from_hub, extract_features

# Load pretrained
model = load_from_hub()

# From any (8, 16, 16) source
patches = torch.randn(16, 8, 16, 16).cuda()
gate_vectors, patch_features = extract_features(model, patches)

# Or full output dict
out = model(patches)
out["local_dim_logits"]       # (B, 64, 4)  dimensionality
out["local_curv_logits"]      # (B, 64, 3)  curvature
out["struct_topo_logits"]     # (B, 64, 2)  topology
out["patch_features"]         # (B, 64, 128) learned features
out["patch_shape_logits"]     # (B, 64, 27) shape classification

Related

• AbstractPhil/geovae-proto — The Rosetta Stone experiments (text→geometry VAEs)
• AbstractPhil/synthetic-characters — 49k FLUX-generated character dataset
• AbstractPhil/grid-geometric-multishape — Original training repo with checkpoints

Citation

Geometric deep learning research by AbstractPhil . The model demonstrates that geometric structure is a universal language bridging text and visual modalities — symbolic association through geometric language.