List your company
ResearchFormerly WizWand

Correctness Evaluation on PPE Correctness (test)

87.2ScoreSkywork-Reward-V2-Llama-3.1-8B-40M
58.0865.6473.280.76Jul 2, 2025
Updated 8mo ago

Evaluation results

MethodLinks
Skywork-Reward-V2-Llama-3.1-8B-40M2025.0787.2——
Skywork-Reward-V2-Llama-3.1-8B2025.0783.4——
Skywork-Reward-V2-Qwen3-8B2025.0775.1——
Skywork-Reward-V2-Qwen3-4B2025.0774.7——
J1-Llama-70B (Maj@32)Model Type=LLM-as-a-Judge, Sampling Strategy=Maj@322025.0773.7——
J1-Llama-70BModel Type=LLM-as-a-Judge2025.0772.9——
Skywork-Reward-V2-Llama-3.2-3B2025.0772.1——
Skywork-Reward-V2-Qwen3-1.7B2025.0770.5——
Claude-3.5-SonnetModel Type=LLM-as-a-Judge2025.0769.2——
Skywork-Reward-V2-Qwen3-0.6B2025.0768.3——
Skywork-Reward-V2-Llama-3.2-1B2025.0767.4——
GPT-4oModel Type=LLM-as-a-Judge2025.0767.1——
INF-ORM-Llama3.1-70BModel Type=Open Reward Model2025.0764.4——
Llama-3-OffsetBias-RM-8BModel Type=Open Reward Model2025.0764.1——
LDL-Reward-Gemma-2-27B-v0.1Model Type=Open Reward Model2025.0763.9——
Llama-3.1-Nemotron-70BModel Type=Open Reward Model2025.0763.2——
DeepSeek-GRM-27B (w/ MetaRM)Model Type=Generative Reward Model2025.0763.2——
Internlm2-20b-rewardModel Type=Open Reward Model2025.0763——
Skywork-Reward-Llama-3.1-8B-v0.2Model Type=Open Reward Model2025.0762.5——
Skywork-Reward-Gemma-2-27B-v0.2Model Type=Open Reward Model2025.0761.9——
J1-Llama-8B (Maj@32)Model Type=LLM-as-a-Judge, Sampling Strategy=Maj@322025.0761.9——
ArmoRM-Llama3-8B-v0.1Model Type=Open Reward Model2025.0760.6——
DeepSeek-GRM-27BModel Type=Generative Reward Model2025.0760.4——
J1-Llama-8BModel Type=LLM-as-a-Judge2025.0759.2——
ConsistencyModel=LLAMA Scout (109B), N=10, temperature=0.72025.12—0.32120.2978
ConsistencyModel=GPT-OSS 20B, N=10, temperature=0.72025.12—0.09320.098
MajorityModel=LLAMA Scout (109B), N=10, temperature=0.72025.12—0.31970.294
MajorityModel=GPT-OSS 20B, N=10, temperature=0.72025.12—0.0990.1034
ProbeModel=LLAMA Scout (109B)2025.12—0.06430.0684
ProbeModel=GPT-OSS 20B2025.12—0.06890.1193
VerbalizedModel=LLAMA Scout (109B)2025.12—0.19310.1597
VerbalizedModel=GPT-OSS 20B2025.12—0.14620.152