A newer version of the Gradio SDK is available:
6.6.0
metadata
title: DiffViews
emoji: 🔬
colorFrom: purple
colorTo: blue
sdk: gradio
sdk_version: 6.3.0
app_file: app.py
python_version: '3.10'
pinned: false
license: mit
DiffViews - Diffusion Activation Visualizer
Interactive visualization of diffusion model activations projected to 2D via UMAP.
Features
- Explore activation space of diffusion models
- Select points and find nearest neighbors
- Generate images from averaged neighbor activations
- Visualize denoising trajectories
Usage
- Hover over points to preview samples
- Click to select a point
- Click nearby points or use "Suggest KNN" to add neighbors
- Click "Generate from Neighbors" to create new images
Note
First launch downloads ~2.5GB of data and checkpoints. Generation on CPU takes ~30-60s per image.