---
title: README
emoji: 🌖
colorFrom: red
colorTo: purple
sdk: static
pinned: false
---

# DifferenceLabs

**DifferenceLabs** is an open research initiative focused on **Diffusion Language Models (DLMs)**.

We explore diffusion-based approaches to language generation, where models generate full sequences in parallel through iterative denoising — instead of token-by-token like traditional autoregressive models.  
This can lead to **faster inference** and **better GPU utilization**.

## What We Do

- Research diffusion-based language models  
- Build experimental architectures and training pipelines  
- Share open, reproducible research  
- Explore efficiency and scalability improvements  

## Projects

This org hosts research code, experiments, and supporting tools related to diffusion language modeling.