List your company

Multilingual Language Understanding on MMLU-ProX (test)

43.6AccuracyCOMPASS
17.49624.27331.0537.827Apr 22, 2026

Evaluation results

MethodLinks
COMPASSBackbone=Qwen2.5-7B-Instruct2026.0443.6
COMPASS-FFTBackbone=Qwen2.5-7B-Instruct, Fine-tuning=Full2026.0442.5
LangRankBackbone=Qwen2.5-7B-Instruct2026.0439.4
LangSimBackbone=Qwen2.5-7B-Instruct2026.0439
TargetBackbone=Qwen2.5-7B-Instruct2026.0438.7
RandomBackbone=Qwen2.5-7B-Instruct2026.0438.4
Pretrained (ZS)Backbone=Qwen2.5-7B-Instruct, Evaluation Protocol=Zero-shot2026.0437.5
AllBackbone=Qwen2.5-7B-Instruct2026.0433.7
COMPASSBackbone=Phi-4-Mini-Instruct-3.8B2026.0428.7
COMPASS-FFTBackbone=Phi-4-Mini-Instruct-3.8B, Fine-tuning=Full2026.0427.7
COMPASSBackbone=Llama-3.1-Instruct-8B2026.0426.9
COMPASS-FFTBackbone=Llama-3.1-Instruct-8B, Fine-tuning=Full2026.0426.1
LangRankBackbone=Phi-4-Mini-Instruct-3.8B2026.0424.5
LangRankBackbone=Llama-3.1-Instruct-8B2026.0424.4
LangSimBackbone=Phi-4-Mini-Instruct-3.8B2026.0424.2
LangSimBackbone=Llama-3.1-Instruct-8B2026.0424
TargetBackbone=Phi-4-Mini-Instruct-3.8B2026.0423.8
RandomBackbone=Llama-3.1-Instruct-8B2026.0423.7
TargetBackbone=Llama-3.1-Instruct-8B2026.0423.2
RandomBackbone=Phi-4-Mini-Instruct-3.8B2026.0422.9
Pretrained (ZS)Backbone=Llama-3.1-Instruct-8B, Evaluation Protocol=Zero-shot2026.0422.8
Pretrained (ZS)Backbone=Phi-4-Mini-Instruct-3.8B, Evaluation Protocol=Zero-shot2026.0421.7
AllBackbone=Llama-3.1-Instruct-8B2026.0419.6
AllBackbone=Phi-4-Mini-Instruct-3.8B2026.0418.5