List your company

Reward Modeling on RM Bench Code

0.154EFSkywork-Reward-Gemma-2-27B
-0.15176-0.072380.0070.08638Nov 30, 2025

Evaluation results

MethodLinks
Skywork-Reward-Gemma-2-27BBackbone=Gemma-2, Size=27B2025.110.1540.090.1620.1940.1850.1160.0950.1360.0480.072
Skywork-Reward-Gemma-2-27Bparameters=27B2025.110.1540.090.1620.1940.1850.1160.0950.1360.0480.072
FsfairX-LLaMA3-RM-v0.1Backbone=LLaMA3, Version=v0.12025.110.1320.0330.0880.0220.047-0.013-0.028-0.0160.0510.125
FsfairX-LLaMA3-RM-v0.1parameters=8B2025.110.1320.0330.0880.0220.047-0.013-0.028-0.0160.0510.125
UltraRM-13bSize=13b2025.110.1240.0780.0860.1340.0870.0260.0770.0680.0890.153
UltraRM-13bparameters=13B2025.110.1240.0780.0860.1340.0870.0260.0770.0680.0890.153
URM-LLaMa-3.1-8BBackbone=LLaMa-3.1, Size=8B2025.110.1130.1110.0580.1210.0280.0640.0390.0220.1110.156
URM-LLaMa-3.1-8Bparameters=8B2025.110.1130.1110.0580.1210.0280.0640.0390.0220.1110.156
Skywork-Reward-V2-Llama-3.1-8BBackbone=Llama-3.1, Size=8B, Version=V22025.110.1060.0810.1570.1190.164-0.13-0.0790.1260.3680.294
Skywork-Reward-V2-Llama-3.1-8Bparameters=8B2025.110.1060.0810.1570.1190.164-0.13-0.0790.1260.3680.294
Llama-3.1-70B-InstructSize=70B, Type=Instruct2025.110.098-0.0810.1090.0690.1930.019-0.1020.0170.081-0.04
Llama-3.1-70B-Instructparameters=70B2025.110.098-0.0810.1090.0690.1930.019-0.1020.0170.081-0.04
GRM-llama3-8B-distillBackbone=llama3, Size=8B, Type=Distilled2025.110.0950.0110.190.0960.0890.050.051-0.0470.0230.142
GRM-llama3-8B-distillparameters=8B2025.110.0950.0110.190.0960.0890.050.051-0.0470.0230.142
Starling-RM-34BSize=34B, Type=Reward Model2025.110.0840.028-0.0660.039-0.032-0.083-0.1740.0130.023-0.099
Skywork-Reward-Gemma-2-27B-v0.2Backbone=Gemma-2, Size=27B, Version=v0.22025.110.0840.0790.050.0550.1780.1090.088-0.0370.022-0.018
Starling-RM-34Bparameters=34B2025.110.0840.028-0.0660.039-0.032-0.083-0.1740.0130.023-0.099
Skywork-Reward-Gemma-2-27B-v0.2parameters=27B2025.110.0840.0790.050.0550.1780.1090.088-0.0370.022-0.018
Skywork-Reward-Llama-3.1-8BBackbone=Llama-3.1, Size=8B2025.110.059-0.1230.0650.1270.1170.0110.0250.0140.1610.119
Skywork-Reward-Llama-3.1-8Bparameters=8B2025.110.059-0.1230.0650.1270.1170.0110.0250.0140.1610.119
GRM-llama3-8B-sftregBackbone=llama3, Size=8B, Type=SFT Regularized2025.110.049-0.0250.1150.0690.0690.0130.009-0.0560.0470.141
GRM-llama3-8B-sftregparameters=8B2025.110.049-0.0250.1150.0690.0690.0130.009-0.0560.0470.141
GRM-Llama3-8B-rewardmodel-ftBackbone=Llama3, Size=8B, Type=Reward Model Fine-Tuning2025.110.041-0.031-0.045-0.022-0.022-0.036-0.046-0.11100.055
GRM-Llama3-8B-rewardmodel-ftparameters=8B2025.110.041-0.031-0.045-0.022-0.022-0.036-0.046-0.11100.055
ArmoRM-Llama3-8B-v0.1Backbone=Llama3, Size=8B, Version=v0.12025.110.0380.0240.0720.0670.1330.025-0.036-0.053-0.0110.079
ArmoRM-Llama3-8B-v0.1parameters=8B2025.110.0380.0240.0720.0670.1330.025-0.036-0.053-0.0110.079
Skywork-Reward-Llama-3.1-8B-v0.2Backbone=Llama-3.1, Size=8B, Version=v0.22025.110.037-0.030.040.114-0.0010.080.097-0.0330.1350.132
Skywork-Reward-Llama-3.1-8B-v0.2parameters=8B2025.110.037-0.030.040.114-0.0010.080.097-0.0330.1350.132
internlm2-7b-rewardSize=7b, Type=Reward Model2025.110.035-0.0040.0560.115-0.0470.012-0.0330.0090.030.009
internlm2-7b-rewardparameters=7B2025.110.035-0.0040.0560.115-0.0470.012-0.0330.0090.030.009
URM-LLaMa-3-8BBackbone=LLaMa-3, Size=8B2025.110.03-0.0380.02-0.050.0310.0350.0580.020.040.147
URM-LLaMa-3-8Bparameters=8B2025.110.03-0.0380.02-0.050.0310.0350.0580.020.040.147
Qwen3-8BSize=8B2025.110.0240.010.052-0.0230.029-0.0090.0410.0390.327-0.102
Qwen3-8Bparameters=8B2025.110.0240.010.052-0.0230.029-0.0090.0410.0390.327-0.102
Llama-3-OffsetBias-RM-8BBackbone=Llama-3, Size=8B, Type=OffsetBias RM2025.110.022-0.0280.071-0.0040.072-0.046-0.086-0.0240.0590.093
Llama-3-OffsetBias-RM-8Bparameters=8B2025.110.022-0.0280.071-0.0040.072-0.046-0.086-0.0240.0590.093
Eurus-RM-7bSize=7b, Type=Reward Model2025.110.007-0.017-0.0940.006-0.084-0.042-0.082-0.0130.10.064
Eurus-RM-7bparameters=7B2025.110.007-0.017-0.0940.006-0.084-0.042-0.082-0.0130.10.064
Qwen2.5-72B-InstructSize=72B, Type=Instruct2025.11-0.0150.023-0.080.109-0.1250.078-0.0980.0440.230.066
Qwen2.5-72B-Instructparameters=72B2025.11-0.0150.023-0.080.109-0.1250.078-0.0980.0440.230.066
SOLAR-10.7B-Instruct-v1.0Size=10.7B, Version=v1.0, Type=Instruct2025.11-0.026-0.035-0.001-0.073-0.040.0190.027-0.08-0.0280.01
SOLAR-10.7B-Instruct-v1.0parameters=10.7B2025.11-0.026-0.035-0.001-0.073-0.040.0190.027-0.08-0.0280.01
internlm2-20b-rewardSize=20b, Type=Reward Model2025.11-0.03-0.0350.0790.0320.083-0.015-0.0370.1220.1120.013
internlm2-20b-rewardparameters=20B2025.11-0.03-0.0350.0790.0320.083-0.015-0.0370.1220.1120.013
Llama3-8B-IDRMBackbone=Llama3, Size=8B, Type=IDRM2025.11-0.051-0.020.1050.0820.017-0.082-0.071-0.04-0.059-0.072
Llama3-8B-IDRMparameters=8B2025.11-0.051-0.020.1050.0820.017-0.082-0.071-0.04-0.059-0.072
tulu-2-dpo-13bSize=13b, Type=DPO2025.11-0.0730.017-0.089-0.07-0.017-0.091-0.016-0.031-0.0660.063
tulu-2-dpo-13bparameters=13B2025.11-0.0730.017-0.089-0.07-0.017-0.091-0.016-0.031-0.0660.063
Skywork-Reward-V2-Qwen3-8BBackbone=Qwen3, Size=8B, Version=V22025.11-0.138-0.0290.0640.0560.118-0.019-0.0370.0520.2180.209
Skywork-Reward-V2-Qwen3-8Bparameters=8B2025.11-0.138-0.0290.0640.0560.118-0.019-0.0370.0520.2180.209
internlm2-1.8b-rewardSize=1.8b, Type=Reward Model2025.11-0.14-0.125-0.063-0.050.0010.0410.039-0.0490.0330.05
internlm2-1.8b-rewardparameters=1.8B2025.11-0.14-0.125-0.063-0.050.0010.0410.039-0.0490.0330.05