List your company
ResearchFormerly WizWand

Pareto prompt set identification on CNN/DailyMail

18.03Hypervolume (HV)MLP-EGE
14.9115.7216.5317.34May 14, 2026
Updated 6mo ago

Evaluation results

MethodLinks
MLP-EGEK (Prompt set size)=30, b (Per-arm budget)=8, Model=Llama 32026.0518.03————
EGEK (Prompt set size)=100, b (Per-arm budget)=10, Model=Llama 32026.0517.54————
EGEK (Prompt set size)=30, b (Per-arm budget)=8, Model=Llama 32026.0517.53————
EGEK (Prompt set size)=30, b (Per-arm budget)=10, Model=Llama 32026.0517.44————
MLP-EGEK (Prompt set size)=30, b (Per-arm budget)=10, Model=Llama 32026.0517.27————
EGEK (Prompt set size)=50, b (Per-arm budget)=10, Model=Llama 32026.0517.2————
MLP-EGEK (Prompt set size)=100, b (Per-arm budget)=10, Model=Llama 32026.0517.16————
EGEK (Prompt set size)=50, b (Per-arm budget)=8, Model=Llama 32026.0517.06————
MLP-EGEK (Prompt set size)=100, b (Per-arm budget)=8, Model=Llama 32026.0516.99————
EGEK (Prompt set size)=100, b (Per-arm budget)=8, Model=Llama 32026.0516.9————
MLP-EGEK (Prompt set size)=30, b (Per-arm budget)=5, Model=Llama 32026.0516.88————
UniformK (Prompt set size)=30, b (Per-arm budget)=8, Model=Llama 32026.0516.85————
MLP-EGEK (Prompt set size)=100, b (Per-arm budget)=5, Model=Llama 32026.0516.84————
EGEK (Prompt set size)=30, b (Per-arm budget)=5, Model=Llama 32026.0516.8————
MLP-EGEK (Prompt set size)=50, b (Per-arm budget)=10, Model=Llama 32026.0516.73————
MLP-EGEK (Prompt set size)=50, b (Per-arm budget)=8, Model=Llama 32026.0516.71————
UniformK (Prompt set size)=30, b (Per-arm budget)=10, Model=Llama 32026.0516.61————
EGEK (Prompt set size)=50, b (Per-arm budget)=3, Model=Llama 32026.0516.51————
EGEK (Prompt set size)=50, b (Per-arm budget)=5, Model=Llama 32026.0516.47————
UniformK (Prompt set size)=50, b (Per-arm budget)=10, Model=Llama 32026.0516.46————
EGEK (Prompt set size)=100, b (Per-arm budget)=5, Model=Llama 32026.0516.39————
MLP-EGEK (Prompt set size)=30, b (Per-arm budget)=3, Model=Llama 32026.0516.32————
UniformK (Prompt set size)=100, b (Per-arm budget)=10, Model=Llama 32026.0516.31————
MLP-EGEK (Prompt set size)=50, b (Per-arm budget)=5, Model=Llama 32026.0516.28————
MLP-EGEK (Prompt set size)=100, b (Per-arm budget)=3, Model=Llama 32026.0516.28————
UniformK (Prompt set size)=100, b (Per-arm budget)=8, Model=Llama 32026.0516.24————
UniformK (Prompt set size)=50, b (Per-arm budget)=8, Model=Llama 32026.0516.18————
MLP-EGEK (Prompt set size)=50, b (Per-arm budget)=3, Model=Llama 32026.0516.18————
EGEK (Prompt set size)=30, b (Per-arm budget)=3, Model=Llama 32026.0516.03————
UniformK (Prompt set size)=100, b (Per-arm budget)=5, Model=Llama 32026.0515.79————
UniformK (Prompt set size)=30, b (Per-arm budget)=3, Model=Llama 32026.0515.77————
UniformK (Prompt set size)=50, b (Per-arm budget)=3, Model=Llama 32026.0515.59————
UniformK (Prompt set size)=50, b (Per-arm budget)=5, Model=Llama 32026.0515.43————
UniformK (Prompt set size)=30, b (Per-arm budget)=5, Model=Llama 32026.0515.34————
UniformK (Prompt set size)=100, b (Per-arm budget)=3, Model=Llama 32026.0515.19————
EGEK (Prompt set size)=100, b (Per-arm budget)=3, Model=Llama 32026.0515.03————
EGEBackbone=Gemma-7B, K=302026.05—13.2714.5814.7214.53
EGEBackbone=Gemma-7B, K=502026.05—13.3414.1314.5514.45
EGEBackbone=Gemma-7B, K=1002026.05—12.7913.4213.4314.11
EGEBackbone=Llama3-8B, K=302026.05—16.0316.817.5317.44
EGEBackbone=Llama3-8B, K=502026.05—16.5116.4717.0617.2
EGEBackbone=Llama3-8B, K=1002026.05—15.0316.3916.917.54
MLP-EGEBackbone=Gemma-7B, K=302026.05—14.4114.9915.0515.05
MLP-EGEBackbone=Gemma-7B, K=502026.05—13.5814.314.3414.63
MLP-EGEBackbone=Gemma-7B, K=1002026.05—13.1613.3613.7914.08
MLP-EGEBackbone=Llama3-8B, K=302026.05—16.3216.8818.0317.27
MLP-EGEBackbone=Llama3-8B, K=502026.05—16.1816.2816.7116.73
MLP-EGEBackbone=Llama3-8B, K=1002026.05—16.2816.8416.9917.16
UniformBackbone=Gemma-7B, K=302026.05—13.4513.9314.8414.59
UniformBackbone=Gemma-7B, K=502026.05—13.113.914.1414.29
UniformBackbone=Gemma-7B, K=1002026.05—12.813.0513.2513.42
UniformBackbone=Llama3-8B, K=302026.05—15.7715.3416.8516.61
UniformBackbone=Llama3-8B, K=502026.05—15.5915.4316.1816.46
UniformBackbone=Llama3-8B, K=1002026.05—15.1915.7916.2416.31