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

Pointwise evaluation on HelpSteer2

0.464Spearman Correlation

Annotation-free Preference Learning for Rubric Generator

0.21960.283050.34650.40995May 28, 2026
Updated 2d ago

Evaluation Results

MethodLinks
0.4640.503
0.440.471
2026.05
0.4380.488
2026.05
0.4380.488
2026.05
0.4320.483
2026.05
0.4110.476
2026.05
0.410.422
0.4060.441
0.3940.431
2026.05
0.3750.435
2026.05
0.3740.416
2026.05
0.3740.416
2026.05
0.3530.428
2026.05
0.3520.431
2026.05
0.3450.403
0.3440.412
0.3260.388
2026.05
0.3190.379
2026.05
0.3190.379
2026.05
0.2860.319
2026.05
0.2840.377
2026.05
0.2820.392
2026.05
0.2790.373
2026.05
0.2630.299
2026.05
0.2460.292
2026.05
0.2390.324
2026.05
0.2390.27
2026.05
0.2290.308