Built with Open reimplementations
krasivo-jev
Training data and models for "system one" classifiers: one forward pass, no reasoning. 100 Hugging Face datasets turned into one format (state → question → options → label), a Qwen3.5-2B LoRA model with a pointer head, held-out benchmarks, and a vLLM plugin that serves TypeSafe's system_one API.
10 ★ on GitHub
What it is
Training data and models for a "system one" model: one forward pass answers a multiple-choice question from the given context alone. Every sample has the shape state - question - options - label; see AGENTS.md for the sample contract and the conventions.
Getting started
From the README:
uv sync # datasets, training, evaluation and the notebooks (dev group)Where the decision sits
krasivo-jev calls the hosted decision API rather than running its own weights. That keeps the project small and means its behaviour moves whenever the vendor ships a new model version. It also means an API key and a network round trip sit in the path of every decision.
Models it references: Jev, jev.
How active it is
Created 2026-09-24, last pushed 2026-10-28, 10 stars and 0 forks when we read it on 2026-11-02. Everything in this category is weeks old at most, so a stale push date means something different here than it would on a mature project. Check the repository before you depend on it.
Find out what your data actually supports
We turn your historical decisions into a frozen test set, then tell you which model, which threshold, and how much of it you can safely automate.