Share your thoughts, 1 month free Claude Pro on usSee more

Mathematics Reasoning on MathVision mini

60.54Accuracy

Ours

Updated 2mo ago

Evaluation Results

Method	Links
Ours 2025.10		60.54
AGLA 2025.10		58.88
Vanilla 2025.10		56.24
VCD 2025.10		56.14
PVM-8B (SFT + GRPO) 2026.05		51.3
PVM-8B (SFT) 2026.05		50.7
CGD 2025.10		49.3
Euclid-8B 2026.05		49
MemVR 2026.05		48.4
ICoT 2026.05		48.1
Qwen3-VL-8B (SFT) 2026.05		48
Qwen3-VL-8B (SFT + GRPO) 2026.05		48
PEARL-8B 2026.05		47.7
Qwen3-VL-8B (LoRA-SFT + GRPO) 2026.05		46.7
OneThinker-8B 2026.05		45.4
Qwen3-VL-8B-Instruct 2026.05		45.4
PVM-4B (SFT + GRPO) 2026.05		45.4
CoMemo 2026.05		43.4
Qwen3-VL-8B (LoRA-SFT) 2026.05		42.8
Qwen3-VL-4B (SFT + GRPO) 2026.05		42.8
Qwen3-VL-4B (LoRA-SFT + GRPO) 2026.05		42.4
PVM-4B (SFT) 2026.05		41.5
Ours 2025.10		39.12
Qwen3-VL-4B (SFT) 2026.05		37.5
Qwen3-VL-4B (LoRA-SFT) 2026.05		37.2
AGLA 2025.10		37.03
Qwen3-VL-4B-Instruct 2026.05		35.9
Vanilla 2025.10		33.54
VCD 2025.10		33.44
CGD 2025.10		28.12
Ours 2025.10		26.01
AGLA 2025.10		23.69
Vanilla 2025.10		20.05
VCD 2025.10		19.92
CGD 2025.10		10.4