Mathematical Reasoning on AMC 2023 (pass@4 accuracy)
75pass@4 AccuracySAGC
Evaluation Results
| Method | Links | |
|---|---|---|
| SAGCBackbone=Qwen2.5-3b, Algorithm=DAPO, Group Size=Dynamic2026.06 | 75 | |
| OpenRLHFBackbone=Qwen2.5-3b, Algorithm=GRPO, Group Size=162026.06 | 67.5 | |
| OpenRLHFBackbone=Qwen2.5-3b, Algorithm=DAPO, Group Size=42026.06 | 67.5 | |
| OpenRLHFBackbone=Qwen2.5-3b, Algorithm=DAPO, Group Size=82026.06 | 65 | |
| OpenRLHFBackbone=Qwen2.5-3b, Algorithm=DAPO, Group Size=162026.06 | 62.5 | |
| OpenRLHFBackbone=Qwen2.5-3b, Algorithm=GRPO, Group Size=82026.06 | 60 | |
| SAGCBackbone=Qwen2.5-3b, Algorithm=GRPO, Group Size=Dynamic2026.06 | 60 | |
| OpenRLHFBackbone=Qwen2.5-3b, Algorithm=GRPO, Group Size=42026.06 | 55 | |
| SAGCBackbone=Llama3.2-3b, Algorithm=DAPO, Group Size=Dynamic2026.06 | 47.5 | |
| OpenRLHFBackbone=Llama3.2-3b, Algorithm=GRPO, Group Size=162026.06 | 45 | |
| SAGCBackbone=Llama3.2-3b, Algorithm=GRPO, Group Size=Dynamic2026.06 | 42.5 | |
| OpenRLHFBackbone=Llama3.2-3b, Algorithm=DAPO, Group Size=162026.06 | 42.5 | |
| OpenRLHFBackbone=Llama3.2-3b, Algorithm=GRPO, Group Size=42026.06 | 40 | |
| OpenRLHFBackbone=Llama3.2-3b, Algorithm=DAPO, Group Size=82026.06 | 37.5 | |
| OpenRLHFBackbone=Llama3.2-3b, Algorithm=GRPO, Group Size=82026.06 | 35 | |
| OpenRLHFBackbone=Llama3.2-3b, Algorithm=DAPO, Group Size=42026.06 | 32.5 |