List your company
ResearchFormerly WizWand

Pairwise Helpfulness Evaluation on HH-RLHF

64.7Win RateGGRO
40.36446.6825359.318Jun 8, 2026
Updated 4mo ago

Evaluation results

MethodLinks
GGROBase Model=LLaMA-3.1-8B-Instruct, Reward Model=Skywork-Reward-V2-LLaMA-3.1-8B, Baseline compared against=SEA2026.0664.77.328
GGROBase Model=LLaMA-3.2-3B-Instruct, Reward Model=GRM-LLaMA-3.2-3B-rewardmodel-ft, Baseline compared against=SEA2026.0660.79.330
GGROBase Model=LLaMA-3.1-8B-Instruct, Reward Model=Skywork-Reward-V2-LLaMA-3.1-8B, Baseline compared against=Vanilla LLM2026.0657.312.730
GGROBase Model=LLaMA-3.1-8B-Instruct, Reward Model=Skywork-Reward-V2-LLaMA-3.1-8B, Baseline compared against=ARGS-G2026.06521236
GGROBase Model=LLaMA-3.1-8B-Instruct, Reward Model=Skywork-Reward-V2-LLaMA-3.1-8B, Baseline compared against=CARDS2026.0648.76.744.7
GGROBase Model=LLaMA-3.1-8B-Instruct, Reward Model=Skywork-Reward-V2-LLaMA-3.1-8B, Baseline compared against=CBS2026.064811.340.7
GGROBase Model=LLaMA-3.2-3B-Instruct, Reward Model=GRM-LLaMA-3.2-3B-rewardmodel-ft, Baseline compared against=Vanilla LLM2026.06485.346.7
GGROBase Model=LLaMA-3.2-3B-Instruct, Reward Model=GRM-LLaMA-3.2-3B-rewardmodel-ft, Baseline compared against=CARDS2026.06481042
GGROBase Model=LLaMA-3.2-3B-Instruct, Reward Model=GRM-LLaMA-3.2-3B-rewardmodel-ft, Baseline compared against=BoN (N=64)2026.0646.718.734.7
GGROBase Model=LLaMA-3.1-8B-Instruct, Reward Model=Skywork-Reward-V2-LLaMA-3.1-8B, Baseline compared against=RS2026.0645.31440.7
GGROBase Model=LLaMA-3.1-8B-Instruct, Reward Model=Skywork-Reward-V2-LLaMA-3.1-8B, Baseline compared against=BoN (N=64)2026.0645.31044.7
GGROBase Model=LLaMA-3.2-3B-Instruct, Reward Model=GRM-LLaMA-3.2-3B-rewardmodel-ft, Baseline compared against=ARGS-G2026.0645.31440.7
GGROBase Model=LLaMA-3.2-3B-Instruct, Reward Model=GRM-LLaMA-3.2-3B-rewardmodel-ft, Baseline compared against=CBS2026.0645.314.740
GGROBase Model=LLaMA-3.2-3B-Instruct, Reward Model=GRM-LLaMA-3.2-3B-rewardmodel-ft, Baseline compared against=RS2026.0641.315.343.3