Mathematical Reasoning on OpenR1 Math Evaluation Suite (Math500, AIME24, AMC, Gaokao, Miner., Olymp.) (test)
82.8Math500 AccuracyDS
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| DSBackbone=Qwen2.5-Math-7B, Training Data=OpenR12026.03 | 82.8 | 34.2 | 63.5 | 67.3 | 35.7 | 46.3 | 55 | 11.4 | |
| HIVEBackbone=Qwen2.5-Math-7B, Training Data=OpenR12026.03 | 82.6 | 36.7 | 63.9 | 66.7 | 36.2 | 45.9 | 55.3 | 2.5 | |
| GRESOBackbone=Qwen2.5-Math-7B, Training Data=OpenR12026.03 | 82.3 | 35 | 64.4 | 66.8 | 36.5 | 45.7 | 55.1 | 3.4 | |
| DSBackbone=Qwen2.5-Math-1.5B, Training Data=OpenR12026.03 | 77.1 | 16.7 | 50.3 | 65.5 | 30.9 | 39.7 | 46.7 | 3.8 | |
| HIVEBackbone=Qwen2.5-Math-1.5B, Training Data=OpenR12026.03 | 76.9 | 18.3 | 48.7 | 65.4 | 30.1 | 40.3 | 46.6 | 1.5 | |
| GRESOBackbone=Qwen2.5-Math-1.5B, Training Data=OpenR12026.03 | 76.1 | 20 | 50.6 | 65.1 | 30 | 39.2 | 46.8 | 1.6 |