README.md

---
title: Pocket Diffusion
emoji: 🧪
colorFrom: indigo
colorTo: blue
sdk: static
app_file: index.html
pinned: false
---

# PocketDiffusion

PocketDiffusion is a compact class-conditional denoising diffusion model for 8x8
handwritten digits. It learns to predict Gaussian noise over 50 diffusion steps and
uses classifier-free guidance during sampling.

The same frozen Tiny Vision classifier used for GlyphForge evaluates conditional
recognizability, making the VAE and diffusion results directly comparable under one
judge.

## Reproduce

```powershell
uv run python projects/tiny-vision-foundry/prepare_data.py
uv run python projects/pocket-diffusion/train.py
```

## Verified results

- Parameters: **55,608**
- Diffusion steps: **50**
- Training epochs: **300**
- Generated samples: **1,000**
- Selected classifier-free guidance: **3.0**
- Frozen-judge class fidelity: **96.20%**

Guidance search improved fidelity monotonically from 46.10% at `1.0` to 96.20% at
`3.0`. Per-class fidelity ranged from 83% for digit `8` to 100% for digits `0` and
`6`. Mean within-class pixel variance ranged from 0.0209 to 0.0456, noticeably higher
than the CVAE's 0.0066 to 0.0168 range under the same 100-samples-per-class protocol.

## Hosted showcase

This free static Space preserves the complete original Gradio source, trained artifacts, evaluation files, and local launch requirements. Hugging Face now requires PRO for CPU-backed Gradio hosting, so the public landing page is static while the checked-in `app.py` remains the authoritative runnable demo.