Purebyte
Not all intelligence needs to be massive. Building the foundation for private, local, byte-level AI specialists.
Pinned Loading
Repositories
Showing 2 of 2 repositories
- purebyte-train Public
The official training stack for PureByte: train byte-level neural decision specialists from a task definition to a verified GGUF model in ~25 minutes on a single GPU. Includes synthetic hard-negative generation, curriculum training, and automated evaluation against CredData and PIIMB.
- purebyte Public
Tiny byte-level AI on CPU: zero-dependency C++20 inference engine and neural decision models (2.8ms latency, 9.4MB RAM). Replaces tokenizers with a 256-byte vocabulary. Flagship specialists for pre-commit secret scanning (0.797 F1 vs 0.337 GitLeaks) and PII redaction. Apache 2.0.
People
This organization has no public members. You must be a member to see who’s a part of this organization.
Top languages
Loading…
Most used topics
Loading…