Compact byte-level RoPE Transformer + File-Tree Encoder models and Swarm Health Adapters for real-time BitTorrent, DHT, and media classification.
AI & ML interests
Compact byte-level neural classifiers, conformal safety calibration, and swarm-health telemetry models for the NorviTech Suite.
Recent Activity
Organization Card
NorviTech — Armiger v3.10.0
Armiger is an 8.05M-parameter byte-level RoPE Transformer and modular plugin suite for classification, safety gating, canonical title/year extraction, and swarm liveliness scoring on BitTorrent and DHT metadata.
Repositories
NorviTech/armiger-ultra(v3.10.0): Full33.3 MBFP32 +17.6 MBINT8 ONNX suite and all 6 plugins (ArmigerUltraPipeline).NorviTech/armiger(v3.10.0): Standalone17.6 MBINT8 ONNX CPU classifier (ArmigerStandaloneONNX).- GitHub Ecosystem:
spencernorton/armiger(ghcr.io/spencernorton/armiger-ultra:3.8.0) ·spencernorton/Armiger-Connectors·spencernorton/DH-Waypoint
datasets 0
None public yet