Share your thoughts, 1 month free Claude Pro on usSee more

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

0.043Test Risk (MSE)

Test oracle envelope

Updated 1d ago

Evaluation Results

Method	Links
Test oracle envelope 2026.05		0.043
Information-first + val. K 2026.05		0.044
Best single judge (val.) 2026.05		0.045
Path-family envelope 2026.05		0.045
Complexity-penalized + val. K 2026.05		0.046
Random path + val. K 2026.05		0.048
All seven judges 2026.05		0.059