Skip to content

Benchmark

AHELM: AudioCaps - GPT-4o Judge Critique

Scores are percentages.

Models measured
12
Scores recorded
12
Sources
1
IQ weight
0.72

Score distribution

1.8%
Mean
1.9%
Median
0.385%
Std dev
1.4%
10th percentile
2.3%
90th percentile
1.3%
Min
1.3%2.7%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.72 for chat models, measured over 9 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard