Share your thoughts, 1 month free Claude Pro on usSee more

Pairwise evaluation on BIGGEN

78.33Human Agreement

Human-crafted (existing) Rubrics

Updated 2d ago

Evaluation Results

Method	Links
Human-crafted (existing) Rubrics 2026.05		78.33
Fine-tuned Rubric Generator 2026.05		76.96
Fine-tuned Rubric Generator 2026.05		76.96
Human-crafted (existing) Rubrics 2026.05		76.68
Fine-tuned Rubric Generator 2026.05		76.28
Fine-tuned Rubric Generator 2026.05		76.28
Human-crafted (existing) Rubrics 2026.05		76.15
Training-free Rubric Generation 2026.05		74.89
Highest Training-free baseline 2026.05		74.89
Highest Training-free 2026.05		74.89
DnA-Eval 2026.05		74.82
Training-free Rubric Generation 2026.05		74.75
Fine-tuned Rubric Generator 2026.05		74.67
Fine-tuned Rubric Generator 2026.05		74.67
Fine-tuned Rubric Generator 2026.05		74.46
Fine-tuned Rubric Generator 2026.05		74.46
Human-crafted (existing) Rubrics 2026.05		74.36
Training-free Rubric Generation 2026.05		72.96
Highest Training-free baseline 2026.05		72.96
Highest Training-free 2026.05		72.96
Human-crafted (existing) Rubrics 2026.05		72.53
Training-free Rubric Generation 2026.05		72.42
RubricHub 2026.05		72.22
Fine-tuned Rubric Generator 2026.05		71.92
Fine-tuned Rubric Generator 2026.05		71.92
Training-free Rubric Generation 2026.05		71.74
Highest Training-free baseline 2026.05		71.74
Highest Training-free 2026.05		71.74
Fine-tuned Rubric Generator 2026.05		71.1
Fine-tuned Rubric Generator 2026.05		71.1
DnA-Eval 2026.05		71.01
DnA-Eval 2026.05		70.34
Human-crafted (existing) Rubrics 2026.05		70.2
Prometheus 8x7B 2026.05		69.73
Training-free Rubric Generation 2026.05		69.73
RubricHub 2026.05		69.68
CheckEval 2026.05		68.82
Prometheus 8x7B 2026.05		68.3
RubricHub 2026.05		68.28
CheckEval 2026.05		65.01
CheckEval 2026.05		64.29