Skip to content

Benchmark

Bridging the Gap Belebele Tsonga (ts) human-translation minus back-translation difference

Multilingual benchmarks compare model performance across languages.

Models measured
6
Scores recorded
6
Sources
1
IQ weight
No weight published

Score distribution

-18.2
Mean
-24
Median
12.7
Std dev
-27.8
10th percentile
1.8
90th percentile
-28.6
Min
-28.61.8
The band covers the middle 80% of scores. The line marks the median.

Results

The source marks these results as descriptive, not capability scores, so we list them by name instead of ranking them.

ModelScore
GPT 4otable: A.15, section: back-translation-difference, task: Belebele, language: ts, shots: 0· Bridging the Gap (arXiv:2412.12417v1) data1.8
GPT-3.5 Turbotable: A.15, section: back-translation-difference, task: Belebele, language: ts, shots: 0· Bridging the Gap (arXiv:2412.12417v1) data-27.8
GPT-4 Turbotable: A.15, section: back-translation-difference, task: Belebele, language: ts, shots: 0· Bridging the Gap (arXiv:2412.12417v1) data-2.7
Llama 3 70B Instructtable: A.15, section: back-translation-difference, task: Belebele, language: ts, shots: 0· Bridging the Gap (arXiv:2412.12417v1) data-27.7
Llama 3 8B Instructtable: A.15, section: back-translation-difference, task: Belebele, language: ts, shots: 0· Bridging the Gap (arXiv:2412.12417v1) data-24
Phi-3 Mini 4K Instructtable: A.15, section: back-translation-difference, task: Belebele, language: ts, shots: 0· Bridging the Gap (arXiv:2412.12417v1) data-28.6

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)