Wan 1.3B -- DMX Compressed
Original: 2,742 MB | DMX: 1,092 MB | Savings: 60.2%
DMX-compressed version of Phantom-Wan-1.3B using Block Floating Point (BFP) with stream-separated entropy coding.
Verification
- 825/825 tensors PASS (cosine similarity > 0.99)
- Worst cosine similarity: 0.9987
- Max absolute difference: 0.124
- Full report: verification_report.json
How to Decompress
git clone https://github.com/willjriley/dmx
python dmx/dmx_cli.py decompress phantom_wan_1.3b.dmx model.safetensors
About DMX
DMX achieves 60-80% compression on neural network weights. Patent Pending. https://github.com/willjriley/dmx
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support