JevAny
Calibration-aware reinforcement learning for adaptive decision systems
How it uses Jev
System What it does Status --- --- --- Jev-Judge Typed choice, binary, and ordinal decisions with confidence Released Jev-Agent Chooses actions in multi-step environments Prototype evaluated Jev-Harness Lets an LLM compile open-ended tasks into bounded decisions Prototype Jev-Tool Selects tools, execution modes, and escalation paths Prototype Jev-Symbolic Runs LLM-authored, validated decision trees with JevAny at ea…
Getting started
You need Python 3.12 or newer. The commands below also need a CUDA GPU that holds the full 27B model in BF16 plus runtime memory, since the server loads the model onto one device. Leave local disk space for the base weights as well as the adapter.
- Languages
- Python 99% · Shell 1%
- License
- Apache-2.0
- Latest release
- JevAny v0.2.0
- Last activity
- Sep 2026
- Added
- 2026-09-23
- Stars at snapshot
- 11
- Status
- Community listing
Source: community catalog. Details and metrics reflect the published community snapshot and are not a performance endorsement. Documentation details extracted from GitHub · the README · the project website.