Downloads · 30 days
46
1% of all-time downloads
BAGEL-7B-MoT-DF11
DFloat11/BAGEL-7B-MoT-DF11
BAGEL-7B-MoT-DF11 is a any-to-any model from DFloat11. Use it for the any-to-any task on the model card, and read the license before you ship it in a product.
This model uses DFloat11 lossless compression. It's 32% smaller than the original BFloat16 model, yet produces bit-identical outputs and runs efficiently on GPUs.
All-time downloads
3.9K
Public
Parameters
63.5M
20.2 GB on disk
Likes
24
Public
Repo makeup
Click a slice to open those files.
.safetensors20.2 GB · 100%
At a glance
- Task
- Any-to-Any
- Model type
- qwen2
- Access
- Public
- Created
- May 25, 2025
- Updated
- May 28, 2025
- SHA
- d6963482
Base models
Model card
From the Hugging Face model README
DFloat11 Compressed Model: ByteDance-Seed/BAGEL-7B-MoT
This model uses DFloat11 lossless compression. It's 32% smaller than the original BFloat16 model, yet produces bit-identical outputs and runs efficiently on GPUs.
📊 Performance Comparison
| Metric | BAGEL-7B-MoT (BFloat16) | BAGEL-7B-MoT (DFloat11) |
|---|---|---|
| Model Size | 29.21 GB | 19.89 GB |
| Peak GPU Memory<br>(1024x1024 image generation) | 30.07 GB | 21.76 GB |
| Generation Time<br>(on an A100 GPU) | 54 seconds | 58 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.
🔧 How to Use
A complete usage guide is available in our GitHub repository (forked from the official Bagel repository): https://github.com/LeanModels/Bagel-DFloat11.
📄 Learn More
- Task
- Any-to-Any
- Type
- qwen2
- Created
- May 25, 2025
- Updated
- May 28, 2025