List your company
ResearchFormerly WizWand

Embodied Navigation on Navigation

88Base ScoreGRPO w/ Mask
17.2835.645472.36May 5, 2026
Updated 5mo ago

Evaluation results

MethodLinks
GRPO w/ MaskBackbone=Qwen2.5-VL-3B, Reward Strategy=RL Baseline, Reasoning Strategy=World Model Reasoning Strategy2026.05888185
GLANCE-FullBackbone=Qwen2.5-VL-3B, Reward Strategy=Dense Extrinsic, Reasoning Strategy=World Model Reasoning for Visual States2026.05868887
GLANCE-BaseBackbone=Qwen2.5-VL-3B, Reward Strategy=Sparse Extrinsic, Reasoning Strategy=World Model Reasoning Strategy2026.05818181
VAGEN-FullBackbone=Qwen2.5-VL-3B, Reward Strategy=Dense Extrinsic, Reasoning Strategy=World Model Reasoning for Visual States2026.05808181
VAGEN-BaseBackbone=Qwen2.5-VL-3B, Reward Strategy=Sparse Extrinsic, Reasoning Strategy=World Model Reasoning Strategy2026.05788079
Turn-PPO w/ MaskBackbone=Qwen2.5-VL-3B, Reward Strategy=Turn-level PPO, Reasoning Strategy=World Model Reasoning Strategy2026.05788481
GLANCE w/ Turn-PPOBackbone=Qwen2.5-VL-3B, Reward Strategy=Turn-level PPO, Reasoning Strategy=World Model Reasoning Strategy2026.05788079
o4-mini2026.05757575
GPT-4o2026.05756972
Qwen2.5-VL-72BBackbone=Qwen2.5-VL-72B2026.05707774
Claude 4.5 Sonnet2026.05676767
Gemini 2.5 Pro2026.05636363
Claude 3.7 Sonnet2026.05484747
Qwen2.5-VL-7BBackbone=Qwen2.5-VL-7B2026.05333835
VLM-R1-3BBackbone=VLM-R1-3B2026.05333434
Vanilla-PPOBackbone=Qwen2.5-VL-3B, Reward Strategy=RL Baseline, Reasoning Strategy=World Model Reasoning Strategy2026.05322529
Qwen2.5-VL-3BBackbone=Qwen2.5-VL-3B2026.05202623