Jev Users
← All projects

JEV_Judge

Nefelim584/JEV_Judge Apache-2.0 Python Eval & Align

rank#298
What it does

jev-lite: a ModernBERT-based typed decision model (Bool / Choice / Score) with calibrated probabilities — a fast replacement for LLM-as-a-judge

Numbers
Stars
2
Forks
0
Open issues
0
Last push
today

2 days · 2 → 2

calibrationevaluationfact-checkingllm-as-a-judgemodernbertnlipytorchtext-classificationtransformers

Open on GitHub → Homepage

More in Eval & Align