Mathematics on AIME 25 (pass@1 (%))
23.3Pass@1Scaf-GRPO
Evaluation results
| Method | Links | |
|---|---|---|
| Scaf-GRPOBase Model=DeepSeek-R1-Distill-Qwen-1.5B2025.10 | 23.3 | |
| Vanilla GRPOBase Model=DeepSeek-R1-Distill-Qwen-1.5B2025.10 | 21.1 | |
| Scaf-GRPOBase Model=Qwen2.5-Math-7B2025.10 | 20 | |
| Scaf-GRPOBase Model=Qwen2.5-7B2025.10 | 20 | |
| DeepSeek-R1-Distill-Qwen-1.5BBase Model=DeepSeek-R1-Distill-Qwen-1.5B2025.10 | 20 | |
| Oat-ZeroBase Model=Qwen2.5-Math-7B2025.10 | 16.7 | |
| LUFFYBase Model=Qwen2.5-Math-7B2025.10 | 16.7 | |
| Scaf-GRPOBase Model=Qwen2.5-Math-1.5B2025.10 | 13.3 | |
| Qwen2.5-Math-7BBase Model=Qwen2.5-Math-7B2025.10 | 13.3 | |
| Vanilla GRPOBase Model=Qwen2.5-Math-7B2025.10 | 13.3 | |
| SimpleRL-ZeroBase Model=Qwen2.5-Math-7B2025.10 | 13.3 | |
| Vanilla GRPOBase Model=Qwen2.5-Math-1.5B2025.10 | 10 | |
| Vanilla GRPOBase Model=Qwen2.5-7B2025.10 | 10 | |
| Qwen2.5-7BBase Model=Qwen2.5-7B2025.10 | 6.7 | |
| Qwen2.5-Math-1.5BBase Model=Qwen2.5-Math-1.5B2025.10 | 3.3 | |
| Scaf-GRPOBase Model=Llama-3.2-3B-Instruct2025.10 | 3.3 | |
| Llama-3.2-3B-InstructBase Model=Llama-3.2-3B-Instruct2025.10 | 0 | |
| Vanilla GRPOBase Model=Llama-3.2-3B-Instruct2025.10 | 0 |