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

Embodied Navigation on Navigation

88Base Score

GRPO w/ Mask

Updated 28d ago

Evaluation Results

Method	Links
GRPO w/ Mask 2026.05		88	81	85
GLANCE-Full 2026.05		86	88	87
GLANCE-Base 2026.05		81	81	81
VAGEN-Full 2026.05		80	81	81
VAGEN-Base 2026.05		78	80	79
Turn-PPO w/ Mask 2026.05		78	84	81
GLANCE w/ Turn-PPO 2026.05		78	80	79
o4-mini 2026.05		75	75	75
GPT-4o 2026.05		75	69	72
Qwen2.5-VL-72B 2026.05		70	77	74
Claude 4.5 Sonnet 2026.05		67	67	67
Gemini 2.5 Pro 2026.05		63	63	63
Claude 3.7 Sonnet 2026.05		48	47	47
Qwen2.5-VL-7B 2026.05		33	38	35
VLM-R1-3B 2026.05		33	34	34
Vanilla-PPO 2026.05		32	25	29
Qwen2.5-VL-3B 2026.05		20	26	23