List your company
ResearchFormerly WizWand

Spatial Navigation on Spatial Navigation (OOD)

39.2Success Rate (7x7 Grid)VRRL
0.09610.24820.430.552Jul 2, 2026
Updated 4mo ago

Evaluation results

MethodLinks
VRRLArchitecture=Qwen3-VL-4B-Instruct, Training Stage=RL2026.0739.252.2
VRRLArchitecture=Qwen2.5-VL-3B-Instruct, Training Stage=RL2026.0723.639.2
Multi-SFTArchitecture=Qwen3-VL-4B-Instruct, Training Stage=SFT2026.0721.240.8
Reflection TuningArchitecture=Qwen3-VL-4B-Instruct, Training Stage=SFT2026.0713.238.4
Reflection TuningArchitecture=Qwen2.5-VL-3B-Instruct, Training Stage=SFT2026.0712.431
Multi-SFT → GRPOArchitecture=Qwen3-VL-4B-Instruct, Training Stage=RL2026.0710.436
Multi-SFTArchitecture=Qwen2.5-VL-3B-Instruct, Training Stage=SFT2026.0710.225.9
Multi-SFT → GRPOArchitecture=Qwen2.5-VL-3B-Instruct, Training Stage=RL2026.079.226
Single-SFTArchitecture=Qwen2.5-VL-3B-Instruct, Training Stage=SFT2026.078.824
Single-SFT → GRPOArchitecture=Qwen2.5-VL-3B-Instruct, Training Stage=RL2026.078.829
Single-SFTArchitecture=Qwen3-VL-4B-Instruct, Training Stage=SFT2026.07832.4
Base*Architecture=Qwen3-VL-4B-Instruct, Training Stage=SFT2026.075.627.6
Single-SFT → GRPOArchitecture=Qwen3-VL-4B-Instruct, Training Stage=RL2026.075.634
Base*Architecture=Qwen2.5-VL-3B-Instruct, Training Stage=SFT2026.074.819
VL-Rethinker-32BArchitecture=Qwen2.5-VL-3B-Instruct, Training Stage=Zero-shot2026.073.67.2
VL-Rethinker-7BArchitecture=Qwen2.5-VL-3B-Instruct, Training Stage=Zero-shot2026.073.23.2
MultiArchitecture=Qwen3-VL-4B-Instruct, Training Stage=Zero-shot2026.072.45.6
MultiArchitecture=Qwen2.5-VL-3B-Instruct, Training Stage=Zero-shot2026.0721.4
SingleArchitecture=Qwen3-VL-4B-Instruct, Training Stage=Zero-shot2026.0722.2
SingleArchitecture=Qwen2.5-VL-3B-Instruct, Training Stage=Zero-shot2026.071.61.4