Measure what Jev can actually do before you build on it. Graded findings, ruled-out candidates, and recipes with stop-conditions. 0 promotions — on purpose.
Stars measure the whole repository, including work unrelated to Jev.
Documented records the linked documentation or source. JevHunt has not independently run or benchmarked this project.
[Jev] (TypeSafe's System One model, pinned here as `jev-1.13.0`) does not generate text. You send a state and typed questions (true or false, pick one, place on a rubric) and get back probabilities and a confidence. The engineering is the code around the answer: the threshold, the side it fails toward, and what happens when the answer is malformed.
Evidence commit: aa75ccec9a46886c6f97ad87e17738d5d375e577
Discovered through: github-search.
How we review →