Mathematical & Geometric Reasoning on MathVerse
75.1Accuracy@8Qwen2.5-VL-32B + VPPO
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen2.5-VL-32B + VPPOModel Scale=32B, Algorithm=VPPO2025.10 | 75.1 | |
| Qwen2.5-VL-32B + GRPOModel Scale=32B, Algorithm=GRPO2025.10 | 74.2 | |
| Qwen2.5-VL-32B + DAPOModel Scale=32B, Algorithm=DAPO2025.10 | 73.3 | |
| NoisyRollout-32BModel Scale=32B, Prompting Source=Official author-provided, Note=Trained using the training set of Geo3k2025.10 | 73 | |
| MM-Eureka-32BModel Scale=32B, Prompting Source=Official author-provided2025.10 | 71.8 | |
| Qwen2.5-VL-7B + VPPOModel Scale=7B, Algorithm=VPPO2025.10 | 71.6 | |
| VL-Rethinker-7BModel Scale=7B, Training Strategy=Pure RL, Prompting Source=Official author-provided2025.10 | 68.8 | |
| PAPO-D-7BModel Scale=7B, Training Strategy=Pure RL2025.10 | 68.6 | |
| Qwen2.5-VL-32BModel Scale=32B2025.10 | 68.5 | |
| Qwen2.5-VL-7B + DAPOModel Scale=7B, Algorithm=DAPO2025.10 | 68.3 | |
| R1-ShareVL-7BModel Scale=7B, Training Strategy=Pure RL, Prompting Source=Official author-provided2025.10 | 68 | |
| NoisyRollout-7BModel Scale=7B, Training Strategy=Pure RL, Prompting Source=Official author-provided, Note=Trained using the training set of Geo3k2025.10 | 67.8 | |
| MM-Eureka-7BModel Scale=7B, Training Strategy=Pure RL, Prompting Source=Official author-provided2025.10 | 67.1 | |
| Qwen2.5-VL-7B + GRPOModel Scale=7B, Algorithm=GRPO2025.10 | 66.5 | |
| ThinkLite-7BModel Scale=7B, Training Strategy=Pure RL, Prompting Source=Official author-provided2025.10 | 64.2 | |
| Qwen2.5-VL-7BModel Scale=7B2025.10 | 39 |