Math Reasoning on AMC23 (Accuracy)
86.67Accuracy (AMC23)VRPO
Evaluation Results
| Method | Links | |
|---|---|---|
| VRPOModel Backbone=Qwen3-8B(Strong Model), Optimization Method=VRPO2025.08 | 86.67 | |
| Dr.GRPOModel Backbone=Qwen3-8B(Strong Model), Optimization Method=Dr.GRPO2025.08 | 85.83 | |
| GRPOModel Backbone=Qwen3-8B(Strong Model), Optimization Method=GRPO2025.08 | 84.17 | |
| Reinforce++Model Backbone=Qwen3-8B(Strong Model), Optimization Method=Reinforce++2025.08 | 84.17 | |
| PPOModel Backbone=Qwen3-8B(Strong Model), Optimization Method=PPO2025.08 | 80 | |
| BaseModel Backbone=Qwen3-8B(Strong Model), Optimization Method=Base2025.08 | 75.83 | |
| Dr.GRPOModel Backbone=Qwen2.5-7B-Cold Start(Weak Model), Optimization Method=Dr.GRPO2025.08 | 53.33 | |
| BaseModel Backbone=Qwen2.5-7B-Cold Start(Weak Model), Optimization Method=Base2025.08 | 52.5 | |
| VRPOModel Backbone=Qwen2.5-7B-Cold Start(Weak Model), Optimization Method=VRPO2025.08 | 50 | |
| Reinforce++Model Backbone=Qwen2.5-7B-Cold Start(Weak Model), Optimization Method=Reinforce++2025.08 | 40.83 | |
| PPOModel Backbone=Qwen2.5-7B-Cold Start(Weak Model), Optimization Method=PPO2025.08 | 35.83 | |
| GRPOModel Backbone=Qwen2.5-7B-Cold Start(Weak Model), Optimization Method=GRPO2025.08 | 22.5 |