Position bias in LLM-as-judge: +10 to 15 pts for slot A
Swap two answers and a different one wins. The model judges order, not content (+10 to 15 pts for slot A). How to build an adversarial judge panel with rotation. In the note on multi-agent consensus I made a promise: instead of a mutual-admiration circle, you need an adversarial panel. Today I clos…
Read the full story at DEV Community — Machine Learning ↗
Timeline · 1 report
- 2026-11-11 08:30 · DEV Community — Machine Learning
Position bias in LLM-as-judge: +10 to 15 pts for slot A