Reading
Jev-as-a-Judge for Agent Evals
— X
LangChain tested Jev against LLM judges on accuracy, repeatability, latency, and cost — Jev was dramatically more consistent on continuous scoring, and averaged 0.44s and --.00035/call.
- Added
- Edited
- Not since