Skip to content

Benchmark

Bridging the Gap MMLU College Medicine Tsonga (ts) human-translated

Multilingual benchmarks compare model performance across languages. Scores are percentages. Higher is better.

Models measured
6
Scores recorded
6
Sources
1
IQ weight
1.80

Score distribution

40.1%
Mean
34.7%
Median
12%
Std dev
30.1%
10th percentile
60.1%
90th percentile
29.5%
Min
29.5%60.1%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.80 for chat models, measured over 6 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1GPT 4otable: 2, task: MMLU College Medicine, language: ts, shots: 5· Bridging the Gap (arXiv:2412.12417v1) data60.1%
2GPT-4 Turbotable: 2, task: MMLU College Medicine, language: ts, shots: 5· Bridging the Gap (arXiv:2412.12417v1) data53.2%
3GPT-3.5 Turbotable: 2, task: MMLU College Medicine, language: ts, shots: 5· Bridging the Gap (arXiv:2412.12417v1) data34.7%
4Llama 3 8B Instructtable: 2, task: MMLU College Medicine, language: ts, shots: 5· Bridging the Gap (arXiv:2412.12417v1) data32.9%
5Phi-3 Mini 4K Instructtable: 2, task: MMLU College Medicine, language: ts, shots: 5· Bridging the Gap (arXiv:2412.12417v1) data30.1%
6Llama 3 70B Instructtable: 2, task: MMLU College Medicine, language: ts, shots: 5· Bridging the Gap (arXiv:2412.12417v1) data29.5%

Data credits: Tuka Alhanai, Adam Kasumovic, Mohammad Ghassemi, Aven Zitzelberger, Jessica Lundin and Guillaume Chabot-Couture. Bridging the Gap: Enhancing LLM Performance for Low-Resource African Languages with New Benchmarks, Fine-Tuning, and Cultural Adjustments, arXiv:2412.12417v1. CC BY 4.0 article numeric/protocol extract, restructured with values unchanged. No task or model rights asserted. (licence)