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

Reward Modeling on UltraFeedback (test)

0.1679MAE

SelectMix

0.160360.2112550.262150.313045Mar 19, 2026Mar 20, 2026Mar 21, 2026Mar 22, 2026Mar 23, 2026Mar 24, 2026
Updated 24d ago

Evaluation Results

MethodLinks
2026.03
0.1679-0.47520.3316
2026.03
0.1723-0.43090.3453
2026.03
0.1961-0.52070.3169
2026.03
0.2021-0.46680.3342
2026.03
0.2021-0.44030.3424
2026.03
0.2080.1110.47-
2026.03
0.216-0.23050.4015
2026.03
0.2250.1060.495-
2026.03
0.2263-0.45590.3376
2026.03
0.2280.1060.496-
2026.03
0.23-0.43890.3429
2026.03
0.2452-0.41410.3504
2026.03
0.2488-0.46520.3347
2026.03
0.2713-0.14760.4103
2026.03
0.2720.1130.461-
2026.03
0.2970.1250.405-
2026.03
0.3049-0.36460.3648
2026.03
0.3212-0.38390.3593
2026.03
0.3492-0.14590.423
2026.03
0.3534-0.32010.3774
2026.03
0.3564-0.25250.3957