Mathematical Reasoning on AIME 24 (Maj, Debate, Delta)
30.8Maj ScoreSDRL-rand
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| SDRL-randBackbone=Qwen3-4B-Base, Debate system agents (N)=5, Debate rounds=12026.01 | 30.8 | 31.7 | 0.9 | |
| SDRL-freqBackbone=Qwen3-4B-Base, Debate system agents (N)=5, Debate rounds=12026.01 | 28.7 | 36 | 7.3 | |
| DAPOBackbone=Qwen3-4B-Base, Debate system agents (N)=5, Debate rounds=12026.01 | 26.7 | 28.3 | 1.6 | |
| SDRL-freqBackbone=Qwen2.5-3B, Debate system agents (N)=5, Debate rounds=12026.01 | 9.3 | 13.3 | 4 | |
| SDRL-randBackbone=Qwen2.5-3B, Debate system agents (N)=5, Debate rounds=12026.01 | 8.7 | 11.7 | 3 | |
| DAPOBackbone=Qwen2.5-3B, Debate system agents (N)=5, Debate rounds=12026.01 | 7.3 | 10 | 2.7 |