typed-lm
neurono-ml/typed-lm
Single-forward-pass semantic routing in Rust: turn dense LLMs (Llama, Qwen, Mistral, Gemma) into a typed decision API with LoRA/QLoRA training and FP8/FP4 quantization, built on Candle.
- Stars
- 10
- Forks
- 0
Why it is listed
A Rust/Candle implementation that turns decoder models into typed boolean, choice and score decisions, with adapter training and a System One-style HTTP interface. Listed as a related alternative to Jev, not proof of TypeSafe affiliation or drop-in compatibility.