# SemIf

> MIT local/WebGPU semantic conditionals — OpenJev, the former pitch name

**Category:** AI Agents  
**Pricing:** open-source  
**URL:** https://github.com/TheoLeeCJ/SemIf-OpenJev  
**GitHub:** https://github.com/TheoLeeCJ/SemIf-OpenJev  
**Added:** 2026-09-19

---

## The Hook

SemIf reads local model logits to make semantic conditionals without sampling an answer — a control primitive for agents, not another structured-output wrapper.

## What's Good

MIT (TheoLeeCJ/SemIf). Runs semantic conditions against frozen open models locally, with a WebGPU path, and makes the decision from logits rather than asking the model to sample a response. OpenJev was the former/pitch name; the canonical repository is now SemIf. No closed inference endpoint required for the core experiment.

## The Catch

Vs Instructor and structured JSON: SemIf is a logit-read semantic gate, not schema validation or guaranteed JSON. Vs closed TypeSafe Jev: the local/open-model route is the point, but you bring compatible weights, runtime, and calibration discipline — this is not calibrated RLCD. Youth note: ~1.6k★ in ~3 days clears the Panel bar, but the surface is still young.

## Verdict

MIT local semantic conditionals. Read logits on frozen open models; do not confuse it with structured output or calibrated refusal.