Personalized LLM rankings using ELO and Bradley-Terry on 115 users show low correlation with aggregate rankings (BT ρ=0.04), highlighting the need for user-specific benchmarks.
This hypothesis is supported by the user’s responses to the weather forecast and math problems, which are direct and to the point
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Personalized Benchmarking: Evaluating LLMs by Individual Preferences
Personalized LLM rankings using ELO and Bradley-Terry on 115 users show low correlation with aggregate rankings (BT ρ=0.04), highlighting the need for user-specific benchmarks.