Accuracy on AMC (Mathematical Reasoning)
82AccuracyDIVER
Evaluation results
| Method | Links | |
|---|---|---|
| DIVERBase Model=DeepSeek-R1-Distill-Qwen-7B2025.09 | 82 | |
| GRPOBase Model=DeepSeek-R1-Distill-Qwen-7B, Reward Strategy=Clip-higher2025.09 | 81.7 | |
| GRPOBase Model=Qwen2.5-7B-Base, Reward Strategy=Clip-higher2025.09 | 49.2 | |
| DIVERBase Model=Qwen2.5-7B-Base2025.09 | 48.9 | |
| DIVERBase Model=Qwen2.5-Math-1.5B2025.09 | 44.3 | |
| GRPOBase Model=Qwen2.5-Math-1.5B, Reward Strategy=Clip-higher2025.09 | 42.8 | |
| DIVERBase Model=LLaMA-3.1-8B-Instruct2025.09 | 28.8 | |
| GRPOBase Model=LLaMA-3.1-8B-Instruct, Reward Strategy=Clip-higher2025.09 | 23.3 |