kev
Trainable family of small Qwen-based Jev-like decision models with typed primitives, datasets, evaluation tools, and local inference.

- GitHub stars
- 4,081 as of 2026-09-22
- Forks
- 403
- Language
- Python
- License
- Apache-2.0
- Created
- 2026-09-17
- Last updated on GitHub
- 2026-09-25
- Added to AionEdge
- 2026-09-21

What Jev does in kev
The project describes itself this way: Trainable family of small Qwen-based Jev-like decision models with typed primitives, datasets, evaluation tools, and local inference. In the Jev-like models category, Jev typically decides the same kinds of typed questions as Jev, answered by a different, usually local, model.
Verification
Verified. We confirmed this repository and its Jev relevance through the GitHub API and the source lists below. A specific file link has not been recorded yet.
Topics
- decision-model
- jev
- qwen3
How Jev-like models projects use Jev
Independent models and servers that reproduce the Jev decision pattern, often on open weights you can run yourself. None of them is the hosted TypeSafe model, but they are useful for offline work, research and teams that cannot send data to an external API.
Related Jev projects in Jev-like models
- SemIf-OpenJevTheoLeeCJ/SemIf-OpenJevSemantic ifs from open models, on a 3090 at home. Independent; not affiliated with Jev or TypeSafe.★ 4,279
Python - openjevrazorback16/openjevAn independent System One decision server compatible with Jev’s API, running an open DiffusionGemma model.★ 313
Python - jevk5allebee/jevk5JevK5: open-weight alternative to TypeSafe Jev. Typed decisions with probabilities in one forward pass; Apache-2.0 weights and code.★ 106
Python - AnyJevnokia-applied-research/AnyJevTraining-free Jev-style typed decisions from LLM next-token probabilities. L0 uses cyclic option rotations and label-prior correction to reduce bias without labels; L1 fits temperature calibration using labeled examples. No base-model fine-tuning; calibration and order stability are not universal guarantees.★ 23
Python
tinyjevankit-aglawe/tinyjevA tiny jev-like model that answers Choice, Score and Noul questions in one forward pass and returns calibrated probabilities. MLX or PyTorch, fully offline, System One compatible.★ 16
Python- verdictManavarya09/verdictSmall, fast, honest decision models. Open alternative to Jev: zero-shot, fit on your labels in seconds, calibrated with a coverage guarantee, Jev wire-compatible.★ 1
Python