Live site · by Mapika

decider-2b

An open reproduction of the System One model class on Qwen3.5-2B.

Mapika's decider-2b reads a state and typed questions and returns a probability for every option in one forward pass, with no decoding. It was fine-tuned on about 95 public decision datasets, then trained with outcome-only reinforcement learning. The v11 release improves on v10 on hard decisions and loses ground on some everyday and game rows. Code, data registry and recipe are on GitHub.