List your company
ResearchFormerly WizWand

Reward Modeling on RewardBench (Acc, Brier, CE)

88.8AccuracyTwo-anchor-DeepSeek-Pro
68.20873.55478.984.246May 12, 2026
Updated 5mo ago

Evaluation results

MethodLinks
Two-anchor-DeepSeek-ProBackbone=Llama-3.2-8B-Instruct, Training Dataset=HelpSteer2-9K2026.0588.80.11690.3859
GaussianBackbone=Llama-3.2-8B-Instruct, Training Dataset=HelpSteer2-9K2026.0588.20.11920.3897
Two-anchor-SkyworkBackbone=Llama-3.2-8B-Instruct, Training Dataset=HelpSteer2-9K2026.05880.12870.4082
Two-anchor-DeepSeek-FlashBackbone=Llama-3.2-8B-Instruct, Training Dataset=HelpSteer2-9K2026.0587.60.13170.4176
BTBackbone=Llama-3.2-8B-Instruct, Training Dataset=HelpSteer2-9K2026.0587.10.12970.4155
Two-anchor-SkyworkBackbone=Llama-3.2-3B-Instruct, Training Dataset=HelpSteer2-9K2026.0585.50.1320.413
BT(Hard)Backbone=Llama-3.2-8B-Instruct, Training Dataset=HelpSteer2-9K2026.0585.10.1410.4381
BTBackbone=Llama-3.2-3B-Instruct, Training Dataset=HelpSteer2-9K2026.0584.80.140.444
Two-anchor-DeepSeek-FlashBackbone=Llama-3.2-3B-Instruct, Training Dataset=HelpSteer3-20K2026.0583.80.1410.457
Two-anchor-DeepSeek-FlashBackbone=Llama-3.2-8B-Instruct, Training Dataset=MultiPref-8K2026.0583.10.15820.4831
Two-anchor-DeepSeek-FlashBackbone=Llama-3.2-3B-Instruct, Training Dataset=HelpSteer2-9K2026.05830.1510.467
Two-anchor-SkyworkBackbone=Llama-3.2-8B-Instruct, Training Dataset=MultiPref-8K2026.0582.90.15070.4589
Two-anchor-SkyworkBackbone=Llama-3.2-3B-Instruct, Training Dataset=HelpSteer3-20K2026.0582.70.1470.506
Two-anchor-SkyworkBackbone=Llama-3.2-3B-Instruct, Training Dataset=MultiPref-8K2026.0582.10.1630.49
Two-anchor-DeepSeek-ProBackbone=Llama-3.2-3B-Instruct, Training Dataset=HelpSteer2-9K2026.0582.10.1570.483
GaussianBackbone=Llama-3.2-3B-Instruct, Training Dataset=HelpSteer2-9K2026.0581.90.1610.52
BT(Hard)Backbone=Llama-3.2-3B-Instruct, Training Dataset=HelpSteer2-9K2026.0581.10.1490.509
Two-anchor-DeepSeek-ProBackbone=Llama-3.2-8B-Instruct, Training Dataset=MultiPref-8K2026.0580.70.16430.4958
Two-anchor-DeepSeek-FlashBackbone=Llama-3.2-3B-Instruct, Training Dataset=MultiPref-8K2026.0580.40.1670.502
Two-anchor-SkyworkBackbone=Llama-3.2-3B-Instruct, Training Dataset=PersonalLLM-40K2026.0580.30.1810.546
Two-anchor-DeepSeek-ProBackbone=Llama-3.2-3B-Instruct, Training Dataset=HelpSteer3-20K2026.0579.90.1670.551
BTBackbone=Llama-3.2-3B-Instruct, Training Dataset=HelpSteer3-20K2026.0579.40.1690.59
GaussianBackbone=Llama-3.2-3B-Instruct, Training Dataset=HelpSteer3-20K2026.0578.70.1680.644
Two-anchor-DeepSeek-ProBackbone=Llama-3.2-3B-Instruct, Training Dataset=MultiPref-8K2026.0578.40.180.534
Two-anchor-DeepSeek-FlashBackbone=Llama-3.2-3B-Instruct, Training Dataset=PersonalLLM-40K2026.0578.40.1880.558
GaussianBackbone=Llama-3.2-8B-Instruct, Training Dataset=MultiPref-8K2026.0578.30.17840.5247
BT(Hard)Backbone=Llama-3.2-3B-Instruct, Training Dataset=HelpSteer3-20K2026.0577.70.1780.639
Two-anchor-DeepSeek-ProBackbone=Llama-3.2-3B-Instruct, Training Dataset=PersonalLLM-40K2026.0577.10.1920.566
BTBackbone=Llama-3.2-3B-Instruct, Training Dataset=PersonalLLM-40K2026.0576.80.2090.625
BT(Hard)Backbone=Llama-3.2-3B-Instruct, Training Dataset=MultiPref-8K2026.0575.90.1940.703
BT(Hard)Backbone=Llama-3.2-3B-Instruct, Training Dataset=PersonalLLM-40K2026.0574.20.2451.114
GaussianBackbone=Llama-3.2-3B-Instruct, Training Dataset=PersonalLLM-40K2026.0573.50.220.644
GaussianBackbone=Llama-3.2-3B-Instruct, Training Dataset=MultiPref-8K2026.0573.20.2010.584
BTBackbone=Llama-3.2-8B-Instruct, Training Dataset=MultiPref-8K2026.0572.60.20270.584
BTBackbone=Llama-3.2-3B-Instruct, Training Dataset=MultiPref-8K2026.0571.70.2090.604
BT(Hard)Backbone=Llama-3.2-8B-Instruct, Training Dataset=MultiPref-8K2026.05690.24380.8986