An experimental, minimal reconstruction of TypeSafe's closed Jev decision model, with training, evaluation, and a serving API. It uses a pointer head over hidden states to answer typed questions with probabilities.
该仓库明确以TypeSafe AI的Jev/System One决策模型为主题，基于公开文档和黑盒探针研究进行假设性重建，实现了无解码循环的指针头架构，并提供训练、评估和服务API，属于实质性的实验研究。
JevalsJevals/jevals-dataIndependent benchmark data comparing TypeSafe AI's Jev (System One model) against LLMs on decision score, accuracy, calibration, cost, and latency for typed tasks.
JGalegoJGalego/Jevs-GarageA garage full of tiny experiments for building critical systems with TypeSafe AI's Jev/System One, including demos, SDK integration, and live contract tests.
NearCaiNearCai/JevLightA CityFlow-based traffic signal control framework that evaluates TypeSafe Jev and OpenAI-compatible LLM controllers alongside rule-based and reinforcement-learning baselines through the same simulation pipeline.
