README / README.md
sip786's picture
Clarify historical project status and public CLI behavior
7a28ed3 verified
|
Raw History Blame Contribute Delete
3.86 kB
---
title: README
emoji: 🧊
colorFrom: indigo
colorTo: blue
sdk: static
pinned: false
---
# Sipsa Labs
Sipsa Labs builds SIP, the operating system for the physical world.
UltraCompress is retired. Its public materials preserve historical research; new API access and paid pilots are no longer offered.
The research results below are historical and apply to their recorded versions and evaluation conditions.
---
## Historical UltraCompress research
**Historical research tiers:**
- A **genuinely lossless archival tier** β€” reconstructs the original weights bit-for-bit (~1.3–1.6Γ— smaller) for exact storage and hand-off.
- A **near-lossless 5-bit tier** β€” ~1% perplexity vs bf16 (lossy), ~3Γ— smaller, for running frontier-scale models on small GPUs.
**23 architectures verified**: 22 PPL-verified end-to-end (21 transformer β€” 17 dense + 4 MoE β€” plus 1 SSM with a comparator-note caveat) and 1 Vision Transformer cosine-verified (DINOv2-Large). Dense + MoE + SSM + ViT, 0.6B to 405B. The public v0.6.27 CLI checks basic pack structure and computes SHA-256 fingerprints unless hashing is skipped. It does not reconstruct weights, reproduce the evaluations below, or automatically compare its fingerprint with a trusted reference.
### Verified 5-bit PPL ratios (sample)
*(perplexity ratio = compressed PPL / bf16 baseline PPL; FineWeb-edu held-out tail; seq_len = 1024; n = 30–50; seed = 42)*
| Model | Params | Type | PPL ratio | Notes |
|---|---:|:---|---:|:---|
| **Mixtral-8x7B** | 47B | MoE | **1.00368Γ—** | tightest MoE result |
| Qwen3-1.7B-Base | 1.7B | dense | **1.00401Γ—** | small-decoder record |
| Qwen3-14B | 14B | dense | **1.00403Γ—** | 14B-class record |
| Yi-1.5-9B | 8.8B | dense | **1.00414Γ—** | >8B record |
| Qwen3-8B | 8B | dense | **1.00440Γ—** | 8B-class record |
| Mistral-7B-v0.3 | 7B | dense | **1.00548Γ—** | strong dense 7B-class result |
| Phi-3-mini-4k | 3.8B | dense | **1.00262Γ—** | tightest dense ratio (seq_len=128 caveat) |
| **Hermes-3-Llama-3.1-405B** | 405B | dense | **1.0066Γ—** | 405B-class near-lossless on a single 32 GB GPU (streamed from disk) |
| Mamba-2.8B | 2.8B | SSM | **1.0059Γ—** | state-space model β€” SSM-compatible comparator (the canonical transformer pipeline is architecture-incompatible with SSMs) |
| Llama-3.1-8B | 8B | dense | **1.0125Γ—** | honest worst case (pipeline-specific floor, not architectural) |
Full matrix at [github.com/sipsalabs/ultracompress](https://github.com/sipsalabs/ultracompress).
### Inspect a historical pack
```bash
pip install ultracompress
hf download SipsaLabs/qwen3-1.7b-base-uc-v3-bpw5 --local-dir ./pack
uc verify ./pack
```
Expected result category: basic pack-structure checks and a computed fingerprint, not a reconstruction proof or an automatic comparison with a trusted reference.
Without an API key, `uc try` prints a recorded reference response, not a live inference result.
---
## Historical API and services
New API access and paid pilots are no longer offered. The public repository and model cards preserve historical materials.
---
## License + IP
- **PyPI v0.6+** under BUSL-1.1 with an **Additional Use Grant**: free for sub-$1M-ARR companies, research, and individuals. Auto-converts to Apache 2.0 four years after each release.
- **v0.5.x** stays Apache-2.0 on the `legacy/0.5.x` branch.
- Codec internals are patent-protected (continuations pending).
---
## Contact
- **Project inquiries** β†’ founder@sipsalabs.com
- **Patents / licensing** β†’ legal@sipsalabs.com
- **Press / media** β†’ press@sipsalabs.com
- **Security disclosure** β†’ security@sipsalabs.com
- **General** β†’ hello@sipsalabs.com
[sipsalabs.com](https://sipsalabs.com) Β· [GitHub](https://github.com/sipsalabs) Β· [PyPI](https://pypi.org/project/ultracompress/) Β· [Historical repository](https://github.com/sipsalabs/ultracompress)