Control For LLM Judge Bias

Use when an LLM scores outputs. Corrects position, length, and self-preference bias, and calibrates against human labels before trusting scores.

jiayaoqijia Updated 81 repo stars

File contents

jiayaoqijia/cryptoskill commit 60a9a58587

Frequently asked questions

npx get-skill add jiayaoqijia/control-for-llm-judge-bias