Spaces:
Running
Running
| <html><head><meta charset="utf-8"><meta name="viewport" content="width=device-width"> | |
| <title>Cognitive Engineering</title> | |
| <style>body{font-family:system-ui,sans-serif;max-width:720px;margin:40px auto;padding:0 16px;line-height:1.6;color:#222}a{color:#b45309}code{background:#f3f3f3;padding:1px 4px}</style></head> | |
| <body> | |
| <h1>Cognitive Engineering</h1> | |
| <p>Structural intelligence for AI systems. Based in Switzerland.</p> | |
| <h2>Products</h2> | |
| <p><b>sal-torch</b> — Training-time structural sparsification for PyTorch transformers. Makes models compression-resilient by design. Source-available (BSL 1.1). PPAM 2026, Springer LNCS. <code>pip install sal-torch</code></p> | |
| <p><b>sal-certify</b> — Structural AI certification for autonomous systems. Verifies that deployed AI models are structurally identical to the ones certified. Mapped to EU AI Act, NIST AI RMF, SR 11-7, China GB 44721.</p> | |
| <h2>Links</h2> | |
| <ul> | |
| <li><a href="https://huggingface.co/spaces/CognitiveEngineering/ai-model-xray">FI Scanner demo</a></li> | |
| <li><a href="https://cognitive-engineering.dev">cognitive-engineering.dev</a></li> | |
| <li><a href="https://github.com/zeekmartin/sal-torch">sal-torch on GitHub</a></li> | |
| <li><a href="https://pypi.org/project/sal-torch/">sal-torch on PyPI</a></li> | |
| <li><a href="https://huggingface.co/datasets/CognitiveEngineering/sal-torch-benchmarks">sal-torch benchmarks dataset</a></li> | |
| <li><a href="https://cognitive-engineering.dev/sal-certify">sal-certify</a></li> | |
| <li>PPAM 2026, Springer LNCS (Paper CP031)</li> | |
| <li>EuroHPC grant holder (5,000 GPU hours on LUMI)</li> | |
| </ul> | |
| <p>Built by Cognitive Engineering (Switzerland).</p> | |
| </body></html> | |