AI · AI inference category
AI inference — Products
Structurally reshapes LLM inference. Token-Exact audit, hallucination suppression, 3-way byte-exact inference.
Products in this category
- SlimePython (3-way Token-Exact LLM)○ 論文段階 Python / Rust native / WASM 3-way byte-exact LLM inference, paper draft v5 ready
- SlimeTree-RLM (hallucination -44pt + WASM 272 KB)○ Published research★ WASM-deployable Hallucination 66%→22% without weight changes, structurally beats Constitutional AI / o1. Rust port compressed to a single 272 KB WASM binary, 24× faster than Python, runs server-less on browser / mobile / embedded, stable under 10,000-unit stress (zero data loss).
AI cross-link
See related services in this category