Mathematical Reasoning on HMMT 2026
49AccuracySeirênes
Evaluation Results
| Method | Links | |
|---|---|---|
| SeirênesBackbone=Qwen3-30B-A3B, same budget=true2026.05 | 49 | |
| DAPOBackbone=Qwen3-30B-A3B, same budget=true2026.05 | 42.3 | |
| Qwen3-30B-A3BBackbone=Qwen3-30B-A3B2026.05 | 41 | |
| SeirênesBackbone=Qwen3-4B-Instruct, same budget=true2026.05 | 39.4 | |
| InTBackbone=Qwen3-4B-Instruct2026.05 | 37.9 | |
| Dr.GRPOBackbone=Qwen3-4B-Instruct, same budget=true2026.05 | 37.2 | |
| DAPOBackbone=Qwen3-4B-Instruct, same budget=true2026.05 | 33 | |
| Scaf-GRPOBackbone=Qwen3-4B-Instruct2026.05 | 32.2 | |
| Qwen3-4B-InstructBackbone=Qwen3-4B-Instruct2026.05 | 31.9 | |
| SAGEBackbone=Qwen3-4B-Instruct2026.05 | 29.9 | |
| SeirênesBackbone=Qwen2.5-7B-Instruct, same budget=true2026.05 | 8.5 | |
| LUFFYBackbone=Qwen2.5-7B-Instruct2026.05 | 6.3 | |
| SAGEBackbone=Qwen2.5-7B-Instruct2026.05 | 6.3 | |
| ExGRPOBackbone=Qwen2.5-7B-Instruct2026.05 | 6 | |
| DAPOBackbone=Qwen2.5-7B-Instruct, same budget=true2026.05 | 4.3 | |
| Qwen2.5-7B-InstructBackbone=Qwen2.5-7B-Instruct2026.05 | 2.7 |