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

Vision-Language Navigation on REVERIE-AIM (val-unseen)

50.4ONS@0.1m

Human Eval

Updated 19d ago

Evaluation Results

Method
Human Eval 2026.07	50.4	74.71	38.24	80.38	-	-	-	-	-
REALM (Full) 2026.07	17.19	25.89	8.48	20.24	13.86	6.71	40.63	34.67	27.68
REALM w/o VSP 2026.07	16.82	25	8.04	19.79	13.86	6.71	40.63	34.67	27.68
REALM (Full) 2026.07	14.96	20.68	6.85	16.37	16.07	8.69	34.4	25.1	17.4
REALM (Full) 2026.07	14.66	21.13	6.1	16.52	11.85	7.26	35.86	28.94	23.39
REALM (Full) 2026.07	13.99	19.72	5.65	15.92	20	8.79	25.37	8.21	4.81
no move 2026.07	13.61	18.08	5.65	12.2	16.07	8.69	34.4	25.1	17.4
REALM w/o VSP 2026.07	13.47	20.24	5.28	15.55	11.85	7.26	35.86	28.94	23.39
Vanilla UniNaVid 2026.07	13.39	19.12	4.91	12.57	13.86	6.71	40.63	34.67	27.68
REALM w/o VSP 2026.07	12.95	17.49	4.69	14.66	20	8.79	25.37	8.21	4.81
REALM w/o VSP 2026.07	12.43	18.38	5.95	14.29	16.07	8.69	34.4	25.1	17.4
Heuristic CLIP-Search 2026.07	12.35	18.15	5.13	10.42	16.07	8.69	34.4	25.1	17.4
Vanilla UniNaVid 2026.07	11.16	17.78	4.54	11.9	11.85	7.26	35.86	28.94	23.39
Vanilla UniNaVid 2026.07	10.42	16.07	4.84	11.01	16.07	8.69	34.4	25.1	17.4
Random Walk 2026.07	9.75	16.89	2.83	8.78	16.07	8.69	34.4	25.1	17.4
Vanilla UniNaVid 2026.07	9.3	14.81	3.5	9.45	20	8.79	25.37	8.21	4.81
no move 2026.07	7.81	11.83	1.49	4.69	20	8.79	25.37	8.21	4.81
Heuristic CLIP-Search 2026.07	7.51	12.57	4.02	10.19	11.85	7.26	35.86	28.94	23.39
Heuristic CLIP-Search 2026.07	7.22	11.98	2.9	8.04	20	8.79	25.37	8.21	4.81
no move 2026.07	7.07	11.98	3.05	6.99	11.85	7.26	35.86	28.94	23.39
Heuristic CLIP-Search 2026.07	6.77	13.24	5.88	12.05	13.86	6.71	40.63	34.67	27.68
no move 2026.07	6.32	12.2	4.46	11.31	13.86	6.71	40.63	34.67	27.68
Random Walk 2026.07	5.65	12.95	2.75	8.78	13.86	6.71	40.63	34.67	27.68
Random Walk 2026.07	5.21	10.42	1.93	4.61	11.85	7.26	35.86	28.94	23.39
Random Walk 2026.07	5.21	10.42	1.34	4.61	20	8.79	25.37	8.21	4.81