List your company
ResearchFormerly WizWand

Spatial Reasoning on CVB 2D (Pass@1)

82.38Accuracy (Pass@1)FGRPO
52.147259.996167.84575.6939Apr 9, 2026
Updated 7mo ago

Evaluation results

MethodLinks
FGRPOBackbone=Qwen2.5-VL-7B-Instruct, Model Scale=7B, Reasoning Strategy=Faithful GRPO2026.0482.38
GRPO-TBackbone=Qwen2.5-VL-7B-Instruct, Model Scale=7B, Reasoning Strategy=RL (Task Reward Only)2026.0479.97
GPT-4o (CoT)Reasoning Strategy=CoT2026.0478.23
FGRPOBackbone=Qwen2.5-VL-3B-Instruct, Model Scale=3B, Reasoning Strategy=Faithful GRPO2026.0477.95
GRPO-TBackbone=Qwen2.5-VL-3B-Instruct, Model Scale=3B, Reasoning Strategy=RL (Task Reward Only)2026.0477.24
Non-reasoningBackbone=Qwen2.5-VL-7B-Instruct, Model Scale=7B, Reasoning Strategy=Non-reasoning2026.0477.17
ViGoRL-SpatialReasoning Strategy=MRM baseline2026.0476.59
GPT-5-nano (CoT)Reasoning Strategy=CoT2026.0476.29
VL-RethinkerReasoning Strategy=MRM baseline2026.0476.06
CoT promptingBackbone=Qwen2.5-VL-7B-Instruct, Model Scale=7B, Reasoning Strategy=CoT prompting2026.0475.92
TreeVGRReasoning Strategy=MRM baseline2026.0474.69
Vision-R1Reasoning Strategy=MRM baseline2026.0471.58
CoT promptingBackbone=Qwen2.5-VL-3B-Instruct, Model Scale=3B, Reasoning Strategy=CoT prompting2026.0471.21
Non-reasoningBackbone=Qwen2.5-VL-3B-Instruct, Model Scale=3B, Reasoning Strategy=Non-reasoning2026.0470.58
R1-OneVisionReasoning Strategy=MRM baseline2026.0453.31