Create README.md

# d3LLM-Dream: Ultra-Fast Diffusion Language Model

## Model Description

**d3LLM-Dream** is an ultra-fast diffusion language model that achieves **5× speedup** over autoregressive models on H100 GPUs while maintaining competitive performance. Built on the Dream architecture, it uses pseudo-trajectory distillation and multi-block decoding to enable efficient parallel text generation.

## Key Features

- 🚀 **4.5× faster** than autoregressive models (Qwen-2.5-7B) on H100 GPU
- 🎯 **2.5× faster** on A100 GPU
- 📊 High AUP (Accuracy Under Parallelism) scores across benchmarks
- 🔧 Optimized for coding and math reasoning tasks


## Usage

For detailed usage instructions, evaluation scripts, and training code, please refer to the official GitHub repository:

👉 **[https://github.com/hao-ai-lab/text-diffusion](https://github.com/hao-ai-lab/text-diffusion)**

README.md

@@ -0,0 +1,10 @@
+---
+datasets:
+- d3LLM-Data-LLaDA/trajectory_data_llada_32
+tags:
+- diffusion
+- text-generation
+- fast-inference
+- d3llm
+pipeline_tag: text-generation
+---