Vibe Coding DiscoverSemIf-OpenJev
View on GitHubSemantic ifs from open models, on a 3090 at home. Independent; not affiliated with Jev or TypeSafe.
Python tool for scoring runtime-defined semantic decisions with open models. It reads option probabilities directly from model logits, with calibration and direct, cached, and parallel scoring modes; includes PyTorch, MLX, llama.cpp, and browser demos.
Use Cases
Route requests to a support queueChoose whether an agent should retry an actionEvaluate whether evidence supports a claimScore runtime-defined options without generating response textBatch decisions that share the same input state
Built With
- Language
- Python
- Frameworks
- PyTorch · Transformers · MLX · llama.cpp
Tags
semantic decisions · direct logits · open models · model inference · probability calibration · decision scoring · benchmarking · WebGPU