List your company

Multimodal Reasoning on GeoQA

49.2Mean@1Socratic-Solver-Geo (Stage3)
43.708845.134446.5647.9856Feb 3, 2026

Evaluation results

MethodLinks
Socratic-Solver-Geo (Stage3)Data Scale=2.5k, Curriculum Stage=32026.0249.2
Geo170k (G-LLaVA)Data Scale=10k2026.0247.16
GeoReasoningData Scale=10k2026.0246.76
KD (Our Synthesis)Data Scale=2.5k2026.0246.58
R-CoTData Scale=7.2k2026.0246.49
TrustGeoGenData Scale=10k2026.0246.35
KD (Geo3K)Data Scale=3k2026.0246.21
PGPS9KData Scale=10k2026.0246.08
Socratic-Solver-Geo (Stage1)Data Scale=0.4k, Curriculum Stage=12026.0245.14
Socratic-Solver-Geo (Stage2)Data Scale=1k, Curriculum Stage=22026.0244.86
Qwen2.5-VL-7B-InstructMode=Zero-shot2026.0243.92