Skip to content

Benchmark

Macedonian LM Evaluation Harness ARC-Easy accuracy

Multilingual benchmarks compare model performance across languages. Higher scores are better.

Models and systems measured
12
Scores recorded
12
Sources
1
IQ weight
No weight published

Score distribution

0.454
Mean
0.472
Median
0.101
Std dev
0.289
10th percentile
0.55
90th percentile
0.231
Min
0.2310.581
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.

Model or systemScore
domestic-yak-basetable: 4, task: ARC Easy, stage: domestic-yak-base· Macedonian Foundation Model (arXiv:2506.09560v1) data0.52
domestic-yak-instructtable: 4, task: ARC Easy, stage: domestic-yak-instruct· Macedonian Foundation Model (arXiv:2506.09560v1) data0.55
domestic-yak†8Btable: 3, task: ARC-E· Macedonian Foundation Model (arXiv:2506.09560v1) data0.547
Llama 3.18Btable: 3, task: ARC-E· Macedonian Foundation Model (arXiv:2506.09560v1) data0.445
Llama 3.1Llama-3.1-8B-Instructtable: 4, task: ARC Easy, stage: Llama 3.1· Macedonian Foundation Model (arXiv:2506.09560v1) data0.45
Llama 3.21Btable: 3, task: ARC-E· Macedonian Foundation Model (arXiv:2506.09560v1) data0.231
Llama 3.370Btable: 3, task: ARC-E· Macedonian Foundation Model (arXiv:2506.09560v1) data0.581
Mistral7Btable: 3, task: ARC-E· Macedonian Foundation Model (arXiv:2506.09560v1) data0.463
Mistral Nemo12Btable: 3, task: ARC-E· Macedonian Foundation Model (arXiv:2506.09560v1) data0.472
MKLLM†7Btable: 3, task: ARC-E· Macedonian Foundation Model (arXiv:2506.09560v1) data0.503
Phi-3.5-mini3.8Btable: 3, task: ARC-E· Macedonian Foundation Model (arXiv:2506.09560v1) data0.289
Qwen2.57Btable: 3, task: ARC-E· Macedonian Foundation Model (arXiv:2506.09560v1) data0.391

Data credits: Stefan Krsteski, Matea Tashkovska, Borjan Sazdov, Hristijan Gjoreski and Branislav Gerazov. Towards Open Foundation Language Model and Corpus for Macedonian, arXiv:2506.09560v1. CC BY 4.0 article numeric/protocol extract; restructured, printed values unchanged. No task, corpus, weight or raw survey rights asserted. (licence)