Skip to content

Benchmark

MTEB(pol, v1) - STS

Embedding benchmarks measure how well models find related items, for example when matching a search to relevant results. Scores are percentages.

Models measured
22
Scores recorded
22
Sources
1
IQ weight
0.82

Score distribution

62.3%
Mean
59.5%
Median
6.9%
Std dev
56.1%
10th percentile
72.1%
90th percentile
49.8%
Min
49.8%72.8%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.82 for embedding models, measured over 22 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard