What it does
Reimplements the Jev request contract over existing SGLang or vLLM deployments by scoring finite candidate tokens.
Benchmarks & research
Only you can see your notes.
Screenshot unavailable. Open the experiment ↗
Reimplements the Jev request contract over existing SGLang or vLLM deployments by scoring finite candidate tokens.
Start by writing down your question and the exact choices you want to compare, like a list of team names or rating numbers. A developer can then set up this helper tool to work with an AI service your organization already runs.
Your software sends the question along with your specific choices to the helper program. It asks the underlying AI service to score how likely each choice is and returns those numbers. Keep in mind that this tool only grades given options and cannot write out new sentences.