Jevals Data: Independent benchmark for TypeSafe Jev
Independent benchmark data comparing TypeSafe AI's Jev (System One model) against LLMs on decision score, accuracy, calibration, cost, and latency for typed tasks.
OpenJev is an open-source decision engine inspired by TypeSafe AI's Jev, offering calibrated classification, benchmark audit, and a WebGPU playground.
仓库明确基于TypeSafe AI的Jev,包含JevBench基准审计和与TypeSafe Jev的对比,并实现了Jev/System One式非自回归决策引擎和WebGPU演示。