/

← All benchmarks

AlpacaEval 2.0

instruction-following · llm-judged-preference↗ Quelle

39Evidence

not yet measured

What it measures

Laengenbereinigte Siegquote gegen ein Referenzmodell, LLM-bewertet.

What it does not measure

Fachliche Richtigkeit.

How the score is composed

Currency50

Discrimination50

Headroom50

Contamination safety9

Gaming resistance55

Methodological rigor45

Reproducibility60

Details

Benchmaxxing risk: medium
Difficulty: low
Public test set: yes
Human-verified: no
LLM-judged: yes
Sample size: 805
Skala: 0–100 % LC-Siege (höher besser)

Best measured models

No results ingested yet.