| | --- |
| | base_model: |
| | - Wan-AI/Wan2.1-T2V-14B-Diffusers |
| | base_model_relation: quantized |
| | pipeline_tag: text-to-image |
| | tags: |
| | - dfloat11 |
| | - df11 |
| | - lossless compression |
| | - 70% size, 100% accuracy |
| | --- |
| | |
| | # DFloat11 Compressed Model: `Wan-AI/Wan2.1-T2V-14B-Diffusers` |
| |
|
| | This model uses **DFloat11** lossless compression. It's 30% smaller than the original BFloat16 model, yet produces bit-identical outputs and runs efficiently on GPUs. |
| |
|
| | ### 📊 Performance Comparison |
| |
|
| | | Metric | Wan2.1-T2V-14B (BFloat16) | Wan2.1-T2V-14B (DFloat11) | |
| | | ---------------------------------- | ------------------------- | ------------------------- | |
| | | Model Size | 28.64 GB | 19.39 GB | |
| | | Peak GPU Memory<br>(2s 480p Video) | 30.79 GB | 22.22 GB | |
| | | Generation Time<br>(an A100 GPU) | 339 seconds | 348 seconds | |
| |
|
| | ### 🔍 How It Works |
| |
|
| | We apply Huffman coding to the exponent bits of BFloat16 model weights, which are highly compressible. We leverage hardware-aware algorithmic designs to enable highly efficient, on-the-fly weight decompression directly on the GPU. Find out more in our [research paper](https://arxiv.org/abs/2504.11651). |
| |
|
| | ### 🔧 How to Use |
| |
|
| | A complete usage guide is available in our GitHub repository: [https://github.com/LeanModels/DFloat11/tree/master/examples/wan2.1](https://github.com/LeanModels/DFloat11/tree/master/examples/wan2.1). |
| |
|
| | ### 📄 Learn More |
| |
|
| | * **Paper**: [70% Size, 100% Accuracy: Lossless LLM Compression for Efficient GPU Inference via Dynamic-Length Float](https://arxiv.org/abs/2504.11651) |
| | * **GitHub**: [https://github.com/LeanModels/DFloat11](https://github.com/LeanModels/DFloat11) |
| | * **HuggingFace**: [https://huggingface.co/DFloat11](https://huggingface.co/DFloat11) |
| |
|
