A simulation suggests LLM judges can favor models fine-tuned on their own outputs, implying that private data-curator evaluations carry conflict-of-interest and annotator-bias risks.
Transition from online elo rating system to bradley-terry model
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CY 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Peeking Behind Closed Doors: Risks of LLM Evaluation by Private Data Curators
A simulation suggests LLM judges can favor models fine-tuned on their own outputs, implying that private data-curator evaluations carry conflict-of-interest and annotator-bias risks.