Geometric Problem Solving on PGPS9K
77.48AccuracyMARS-GPS
Evaluation Results
| Method | Links | |
|---|---|---|
| MARS-GPSCategory=Ours2026.04 | 77.48 | |
| LANS (Diagram GT)Category=Neural Methods, trained on the larger PGPS9K dataset=true2026.04 | 74 | |
| Pi-GPSCategory=Neural-symbolic Methods2026.04 | 69.8 | |
| PGPSNet-v2-SCategory=Neural Methods, trained on the larger PGPS9K dataset=true2026.04 | 69.2 | |
| GeoDRLCategory=Neural-symbolic Methods2026.04 | 66.7 | |
| GOLDCategory=Neural Methods, trained on the larger PGPS9K dataset=true2026.04 | 60.6 | |
| Gemini 2Category=MLLMs2026.04 | 56.8 | |
| GPT-4oCategory=MLLMs2026.04 | 51 | |
| GeoformerCategory=Neural Methods2026.04 | 47.3 | |
| NGSCategory=Neural Methods2026.04 | 46.1 | |
| Claude 3.5 SonnetCategory=MLLMs2026.04 | 45.9 | |
| Qwen-VLCategory=MLLMs2026.04 | 23.2 |