Skip to main content
Interactive Demo

See SEP in Action

Explore the technology behind Semantic Event Protocol through visual demonstrations. All reference code is available to check and run yourself.

Demo 1
Demo 2
Demo 3

HDC Compression: 32× Reduction

Watch as LoRA model weights (17.5 MB) are compressed to HDC semantic packets (271 KB) while preserving 99%+ accuracy.

BEFORE: Raw LoRA Weights
q_proj4.2 MB
v_proj4.2 MB
k_proj4.2 MB
out_proj4.9 MB
Total Size17.5 MB
AFTER: HDC Compression
q_proj (HDC)68 KB
v_proj (HDC)68 KB
k_proj (HDC)68 KB
out_proj (HDC)67 KB
Total Size271 KB
32×
Compression Ratio
# Try it yourself:
$ git clone https://github.com/nick-yudin/SEP
$ cd SEP/reference_impl/python
$ pip install -r requirements.txt
$ python3 -m hdc.distributed_trainer_hdc
✓ Compression: 17.5 MB → 271 KB (32×)

Ready to Build?

All these demos are based on real, reproducible experiments. Dive into the code and see for yourself.

Explore the CodeRead Research Docs