List your company
ResearchFormerly WizWand

Reward Modeling on RM-Bench Chat subset Normal

86AccuracyREWARDAGENT_MINI
6.23226.94147.6568.359Feb 26, 2025
Updated 8mo ago

Evaluation results

MethodLinks
REWARDAGENT_MINIBackbone=GPT-4o mini, Search Engine=false2025.0286
REWARDAGENT_MINIBackbone=GPT-4o mini, Search Engine=true2025.0284.2
DeepSeek-R1Method Category=LLM as generative reward model2025.0283.7
Skywork-Reward-Gemma-2-27BBackbone=Gemma-2-27B, Method Category=Typical reward model2025.0282.7
REWARDAGENT_LLAMABackbone=Llama3-8B Instruct, Search Engine=false2025.0279.3
Skywork-Reward-Llama-3.1-8B-v0.2Backbone=Llama-3.1-8B, Method Category=Typical reward model2025.0278
INF-ORM-Llama3.1-70BBackbone=Llama-3.1-70B, Method Category=Typical reward model2025.0277.5
ArmoRM-Llama3-8B-v0.1Backbone=Llama-3-8B, Method Category=Typical reward model2025.0276.7
o3-miniMethod Category=LLM as generative reward model2025.0276
REWARDAGENT_LLAMABackbone=Llama3-8B Instruct, Search Engine=true2025.0276
internlm2-20b-rewardBackbone=internlm2-20b, Method Category=Typical reward model2025.0274.4
internlm2-7b-rewardBackbone=internlm2-7b, Method Category=Typical reward model2025.0272.6
GPT-4oMethod Category=LLM as generative reward model2025.0271.4
GPT-4o miniMethod Category=LLM as generative reward model2025.0260.5
DeepSeek-R1-Distill-Llama-8BBackbone=Llama-8B, Method Category=LLM as generative reward model2025.0242.1
Llama3-8B InstructBackbone=Llama3-8B, Method Category=LLM as generative reward model2025.029.3