Writing C++ with Jev by picking valid next tokens
The author experimented with TypeSafe's Jev decision/classifier model for C++: at each step it was given only grammatically valid next tokens to pick from, avoiding free-form generation.
OneCodeReviewBench: Jev vs. hosted decision APIs on code review decisions
Mohebifar's team released OneCodeReviewBench, with ~2,000 typed code review and search decisions across 12 languages. The finding is that Jared Palmer's 27B model Kev outperformed Jev and other hosted decision APIs on choice and score questions.
We just put out OneCodeReviewBench: ~2000 typed code review + search decisions across 12 languages. Fun finding: @jaredpalmer 's 27B (Kev) beat Jev and other hosted decision APIs in choice and score questions! https://arxiv.org/abs/2610.33389
· 3 likesOpen on X
More like this
Using Jev for Code Review
The author thinks code review is a really good use case for using Jev.
Fascin: A Jev-Powered PR Triage Tool
A developer shares Fascin, a PR triage tool built with Jev that assesses urgency, change type, and merge readiness, noting Jev handles the actual reasoning.
