List your company
ResearchFormerly WizWand

Spatial Reasoning on Mind Cube (Pass@1)

49.28Accuracy (Pass@1)FGRPO
26.202432.193738.18544.1763Apr 9, 2026
Updated 7mo ago

Evaluation results

MethodLinks
FGRPOBackbone=Qwen2.5-VL-7B-Instruct, Model Scale=7B, Reasoning Strategy=Faithful GRPO2026.0449.28
FGRPOBackbone=Qwen2.5-VL-3B-Instruct, Model Scale=3B, Reasoning Strategy=Faithful GRPO2026.0449.1
GRPO-TBackbone=Qwen2.5-VL-3B-Instruct, Model Scale=3B, Reasoning Strategy=RL (Task Reward Only)2026.0446.06
Non-reasoningBackbone=Qwen2.5-VL-3B-Instruct, Model Scale=3B, Reasoning Strategy=Non-reasoning2026.0443.71
GPT-4o (CoT)Reasoning Strategy=CoT2026.0443.52
TreeVGRReasoning Strategy=MRM baseline2026.0443.14
GRPO-TBackbone=Qwen2.5-VL-7B-Instruct, Model Scale=7B, Reasoning Strategy=RL (Task Reward Only)2026.0441.71
CoT promptingBackbone=Qwen2.5-VL-3B-Instruct, Model Scale=3B, Reasoning Strategy=CoT prompting2026.0440.38
ViGoRL-SpatialReasoning Strategy=MRM baseline2026.0439.36
VL-RethinkerReasoning Strategy=MRM baseline2026.0437.81
Vision-R1Reasoning Strategy=MRM baseline2026.0436.95
Non-reasoningBackbone=Qwen2.5-VL-7B-Instruct, Model Scale=7B, Reasoning Strategy=Non-reasoning2026.0435.11
CoT promptingBackbone=Qwen2.5-VL-7B-Instruct, Model Scale=7B, Reasoning Strategy=CoT prompting2026.0430.83
GPT-5-nano (CoT)Reasoning Strategy=CoT2026.0427.71
R1-OneVisionReasoning Strategy=MRM baseline2026.0427.09