Geometric Question Answering on GeoQA
82.74Success RateVeriGeo
Evaluation Results
| Method | Links | |
|---|---|---|
| VeriGeoParams=7B, Data Scale=8.7k2026.06 | 82.74 | |
| TR-CoT-Qwen2.5-VL-7BParams=7B, Data Scale=65K + Geo170K2026.06 | 79.2 | |
| GeoGen-SFT-7BParams=7B, Data Scale=224K2026.06 | 78 | |
| TR-CoT-InternVL2.5-8BParams=8B, Data Scale=65K + Geo170K2026.06 | 76.7 | |
| TR-CoT-8BParams=8B, Data Scale=87K2026.06 | 75.9 | |
| GeoGen-SFT-3BParams=3B, Data Scale=224K2026.06 | 75.6 | |
| Qwen2.5-VL-7B-InstructParams=7B2026.06 | 72.07 | |
| MAVIS-7BParams=7B, Data Scale=834K2026.06 | 68.3 | |
| G-LLaVA-13BParams=13B, Data Scale=170K2026.06 | 67 | |
| G-LLaVA-7BParams=7B, Data Scale=170K2026.06 | 62.8 | |
| FreshPERType=VLM2026.04 | 48.1 | |
| Math-LLaVA-13BParams=13B, Data Scale=360K2026.06 | 47.8 | |
| On-PolicyType=VLM2026.04 | 47.5 | |
| Standard PERType=VLM2026.04 | 44.7 |