Spaces:

timotheospaul
/

rhombic

Sleeping

App Files Files Community

rhombic / README.md

timotheospaul

Update RhombiLoRA tab: cybernetic findings, 312 tests, 7-round audit

2afbf07 verified about 1 month ago

preview code

raw

history blame contribute delete

1.48 kB

A newer version of the Gradio SDK is available: 6.12.0

Upgrade

metadata

title: rhombic
emoji: 🔷
colorFrom: red
colorTo: indigo
sdk: gradio
sdk_version: 5.23.0
app_file: app.py
pinned: false
license: mpl-2.0
short_description: FCC vs Cubic lattice topology — live benchmarks + RhombiLoRA
tags:
  - lattice
  - topology
  - benchmark
  - graph-theory
  - signal-processing
  - FCC
  - rhombic-dodecahedron
  - spatial-computing
  - weighted-graph
  - lora
  - rhombilora

rhombic

The bottleneck is not the processor. It is the shape of the cell.

Interactive demo comparing cubic (6-connected) and FCC/rhombic dodecahedral (12-connected) lattice topologies. 312 tests, 3 papers, MPL-2.0.

Tab 1 — The Numbers: Headline results + live graph theory benchmark at any scale.

Tab 2 — Embedding Recall: FCC vs Cubic ANN index comparison with 3D lattice visualization.

Tab 3 — The Thesis: The full argument for why computation inherited the wrong geometry.

Tab 4 — RhombiLoRA: The Learnable Bridge — cybernetic feedback discovers rhombic dodecahedral geometry in multi-channel LoRA (Paper 3). 13 experiments, 4 model families, 7-round audit.

Tab 5 — Weighted Extensions: Direction-weighted Fiedler amplification benchmark (Paper 2). See how structured edge weights amplify the FCC advantage from 2.3x to 6.1x.

GitHub
PyPI

Built by Promptcrafted.