Mathematical Reasoning on AIME 25 (Pass@1, Pass@8, #Invoc.)
0.6556Pass@1DART
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| DARTBackbone=Qwen3-4B-Thinking-2507, Paradigm=RL, CoT Type=tool-integrated long2026.01 | 0.6556 | 0.8576 | 9.51 | |
| UTIRBackbone=Qwen3-4B-Thinking-2507, Paradigm=RL, CoT Type=tool-integrated long2026.01 | 0.6389 | 0.7796 | 8.12 | |
| DAPOBackbone=Qwen3-4B-Thinking-2507, Paradigm=RL, CoT Type=pure-text long2026.01 | 0.6153 | 0.8409 | 0 | |
| ToRL-ThinkBackbone=Qwen3-4B-Thinking-2507, Paradigm=RL, CoT Type=pure-text long2026.01 | 0.6125 | 0.7627 | 0 | |
| UTIRBackbone=Qwen3-8B, Paradigm=RL, CoT Type=tool-integrated long2026.01 | 0.6111 | 0.7496 | 10.22 | |
| ToRL-ThinkBackbone=Qwen3-8B, Paradigm=RL, CoT Type=pure-text long2026.01 | 0.5944 | 0.6991 | 0 | |
| DARTBackbone=Qwen3-8B, Paradigm=RL, CoT Type=tool-integrated long2026.01 | 0.5764 | 0.7902 | 10.66 | |
| DAPOBackbone=Qwen3-8B, Paradigm=RL, CoT Type=pure-text long2026.01 | 0.5611 | 0.7698 | 0 | |
| Original checkpointBackbone=Qwen3-8B, CoT Type=pure-text long2026.01 | 0.5208 | 0.6922 | 0 | |
| Original checkpointBackbone=Qwen3-4B-Thinking-2507, CoT Type=pure-text long2026.01 | 0.4639 | 0.65 | 0 | |
| STARTBackbone=Qwen3-4B-Thinking-2507, Paradigm=SFT, CoT Type=tool-integrated long2026.01 | 0.3847 | 0.589 | 12.08 | |
| ToRL-BaseBackbone=Qwen3-4B-Thinking-2507, Paradigm=RL, CoT Type=tool-integrated short2026.01 | 0.3181 | 0.4224 | 2.99 | |
| STARTBackbone=Qwen3-8B, Paradigm=SFT, CoT Type=tool-integrated long2026.01 | 0.3028 | 0.4976 | 12.77 | |
| RetoolBackbone=Qwen3-8B, Paradigm=SFT+RL, CoT Type=tool-integrated short2026.01 | 0.3 | 0.4351 | 1.2 | |
| ToRL-BaseBackbone=Qwen3-8B, Paradigm=RL, CoT Type=tool-integrated short2026.01 | 0.2583 | 0.3728 | 2.98 | |
| RetoolBackbone=Qwen3-4B-Thinking-2507, Paradigm=SFT+RL, CoT Type=tool-integrated short2026.01 | 0.2013 | 0.3636 | 1.35 | |
| Retool (only SFT)Backbone=Qwen3-8B, Paradigm=SFT, CoT Type=tool-integrated short2026.01 | 0.1967 | 0.3118 | 2.07 | |
| Retool (only SFT)Backbone=Qwen3-4B-Thinking-2507, Paradigm=SFT, CoT Type=tool-integrated short2026.01 | 0.1347 | 0.2342 | 1.89 |