List your company

LLM-as-a-Judge Calibration on Arena100K (test)

0.227Test Risk (MSE)Test oracle envelope
0.226280.231140.2360.24086May 31, 2026

Evaluation results

MethodLinks
Test oracle envelopeCalibration Budget=Largest available2026.050.227
Information-first + val. KCalibration Budget=Largest available2026.050.231
Complexity-penalized + val. KCalibration Budget=Largest available2026.050.232
Path-family envelopeCalibration Budget=Largest available2026.050.233
Best single judge (val.)Calibration Budget=Largest available2026.050.234
Random path + val. KCalibration Budget=Largest available2026.050.24
All seven judgesCalibration Budget=Largest available2026.050.245