https://github.com/callbacked/smoldocling256m-webgpu
Document Understanding in the Browser!
https://github.com/callbacked/smoldocling256m-webgpu
ai document-understanding llms transformersjs
Last synced: 12 months ago
JSON representation
Document Understanding in the Browser!
- Host: GitHub
- URL: https://github.com/callbacked/smoldocling256m-webgpu
- Owner: callbacked
- Created: 2025-07-07T19:58:04.000Z (12 months ago)
- Default Branch: main
- Last Pushed: 2025-07-08T18:28:20.000Z (12 months ago)
- Last Synced: 2025-07-08T19:41:38.939Z (12 months ago)
- Topics: ai, document-understanding, llms, transformersjs
- Language: TypeScript
- Homepage: https://huggingface.co/spaces/callbacked/smoldocling256M-webgpu
- Size: 68.4 KB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# SmolDocling 256M WebGPU Demo
This project is a demo using the **SmolDocling-256M** model to perform document understanding tasks. Allowing you to transform document images into structured formats like Markdown and JSON, and more!
https://github.com/user-attachments/assets/b9c06328-a194-4316-ae44-c8ab030c480b
[Try it out on Hugging Face!](https://huggingface.co/spaces/callbacked/smoldocling256M-webgpu)
## Features
- **Intelligent Content Extraction**: Extracts structures from documents, like:
- Tables
- Math formulas (converted to LaTeX)
- Code blocks
- **Structured Output**: Converts document content into markdown and JSON
- **Region-Specific Processing**: Select a specific area of the document to process only the content you need.
- **Fully Offline**: All processing happens on your device in the browser. Your data never leaves your computer.
## 🚀 Getting Started
To run this project locally, follow these steps:
1. Clone the repo
```bash
git clone https://github.com/callbacked/smoldocling256M-webgpu
```
2. Navigate to the project directory
```bash
cd smoldocling256M-webgpu
```
3. Install NPM packages
```bash
npm install
```
4. Run
```bash
npm run dev
```
This will start the Vite development server, and you can view the application at `http://localhost:5173` (or another port if 5173 is in use).
### Building for Production
To create a production build:
```bash
npm run build
```