List your company
ResearchFormerly WizWand

Knowledge Evaluation on HLE

3.94AccuracyVRPO
2.8483.13153.4153.6985Aug 5, 2025
Updated 4mo ago

Evaluation results

MethodLinks
VRPOModel Backbone=Qwen2.5-7B-Cold Start(Weak Model), Optimization Method=VRPO2025.083.94
VRPOModel Backbone=Qwen3-8B(Strong Model), Optimization Method=VRPO2025.083.89
PPOModel Backbone=Qwen2.5-7B-Cold Start(Weak Model), Optimization Method=PPO2025.083.43
Reinforce++Model Backbone=Qwen2.5-7B-Cold Start(Weak Model), Optimization Method=Reinforce++2025.083.34
Dr.GRPOModel Backbone=Qwen2.5-7B-Cold Start(Weak Model), Optimization Method=Dr.GRPO2025.083.34
BaseModel Backbone=Qwen2.5-7B-Cold Start(Weak Model), Optimization Method=Base2025.083.29
PPOModel Backbone=Qwen3-8B(Strong Model), Optimization Method=PPO2025.083.29
Reinforce++Model Backbone=Qwen3-8B(Strong Model), Optimization Method=Reinforce++2025.083.29
GRPOModel Backbone=Qwen3-8B(Strong Model), Optimization Method=GRPO2025.083.24
Dr.GRPOModel Backbone=Qwen3-8B(Strong Model), Optimization Method=Dr.GRPO2025.083.1
GRPOModel Backbone=Qwen2.5-7B-Cold Start(Weak Model), Optimization Method=GRPO2025.082.97
BaseModel Backbone=Qwen3-8B(Strong Model), Optimization Method=Base2025.082.89