Update README.md
Browse files
README.md
CHANGED
|
@@ -1,15 +1,31 @@
|
|
| 1 |
---
|
| 2 |
-
title:
|
| 3 |
-
emoji:
|
| 4 |
-
colorFrom:
|
| 5 |
-
colorTo:
|
| 6 |
sdk: gradio
|
| 7 |
-
sdk_version:
|
| 8 |
app_file: app.py
|
| 9 |
-
python_version: "3.10"
|
| 10 |
pinned: false
|
| 11 |
-
license:
|
| 12 |
-
short_description: Activation visualization toy for diffusion image generation
|
| 13 |
---
|
| 14 |
|
| 15 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
+
title: DiffViews
|
| 3 |
+
emoji: 🔬
|
| 4 |
+
colorFrom: purple
|
| 5 |
+
colorTo: blue
|
| 6 |
sdk: gradio
|
| 7 |
+
sdk_version: 4.44.0
|
| 8 |
app_file: app.py
|
| 9 |
+
python_version: "3.10"
|
| 10 |
pinned: false
|
| 11 |
+
license: mit
|
|
|
|
| 12 |
---
|
| 13 |
|
| 14 |
+
# DiffViews - Diffusion Activation Visualizer
|
| 15 |
+
|
| 16 |
+
Interactive visualization of diffusion model activations projected to 2D via UMAP.
|
| 17 |
+
|
| 18 |
+
## Features
|
| 19 |
+
- Explore activation space of diffusion models
|
| 20 |
+
- Select points and find nearest neighbors
|
| 21 |
+
- Generate images from averaged neighbor activations
|
| 22 |
+
- Visualize denoising trajectories
|
| 23 |
+
|
| 24 |
+
## Usage
|
| 25 |
+
1. Hover over points to preview samples
|
| 26 |
+
2. Click to select a point
|
| 27 |
+
3. Click nearby points or use "Suggest KNN" to add neighbors
|
| 28 |
+
4. Click "Generate from Neighbors" to create new images
|
| 29 |
+
|
| 30 |
+
## Note
|
| 31 |
+
First launch downloads ~2.5GB of data and checkpoints. Generation on CPU takes ~30-60s per image.
|