Pairwise evaluation on AlpacaEval

72.4Human Agreement

Fine-tuned Rubric Generator

Updated 2d ago

Evaluation Results

Method	Links
Fine-tuned Rubric Generator 2026.05		72.4
Fine-tuned Rubric Generator 2026.05		72.4
Fine-tuned Rubric Generator 2026.05		72.34
Fine-tuned Rubric Generator 2026.05		72.34
Fine-tuned Rubric Generator 2026.05		72.22
Fine-tuned Rubric Generator 2026.05		72.22
Fine-tuned Rubric Generator 2026.05		72.09
Fine-tuned Rubric Generator 2026.05		72.09
Training-free Rubric Generation 2026.05		71.96
Highest Training-free baseline 2026.05		71.96
Highest Training-free 2026.05		71.96
Training-free Rubric Generation 2026.05		71.65
Highest Training-free baseline 2026.05		71.65
Highest Training-free 2026.05		71.65
Fine-tuned Rubric Generator 2026.05		70.72
Fine-tuned Rubric Generator 2026.05		70.72
DnA-Eval 2026.05		70.56
Human-crafted (existing) Rubrics 2026.05		70.4
Human-crafted (existing) Rubrics 2026.05		70.09
DnA-Eval 2026.05		69.78
Training-free Rubric Generation 2026.05		69.78
Training-free Rubric Generation 2026.05		69.78
Highest Training-free baseline 2026.05		69.78
Highest Training-free 2026.05		69.78
Training-free Rubric Generation 2026.05		69.47
Fine-tuned Rubric Generator 2026.05		69.47
Fine-tuned Rubric Generator 2026.05		69.47
Prometheus 8x7B 2026.05		68.54
Human-crafted (existing) Rubrics 2026.05		68.22
Training-free Rubric Generation 2026.05		67.91
CheckEval 2026.05		66.82
RubricHub 2026.05		64.64
RubricHub 2026.05		62.93
RubricHub 2026.05		62.93
DnA-Eval 2026.05		62.46
CheckEval 2026.05		60.44
CheckEval 2026.05		58.72