Vision-and-Language Navigation on IR2R (val-unseen) benchmark leaderboard · SOTA2 Research

Vision-and-Language Navigation on IR2R (val-unseen)

9.4TLHAMT

Evaluation Results

Method
HAMT2024.03	9.4	4.7	64	66	56	54	50
Omnigraph2024.03	9.4	4.1	66	69	60	57	55
HAMT2025.10	9.4	4.7	—	66	56	54	50
OVER-NAV2025.10	9.4	4.1	—	69	60	57	55
TourHAMT2024.03	10	6.2	52	52	39	36	32
TourHAMT2025.10	10	6.2	—	52	39	36	32
Memoir (Ours) on GR-DUET2025.10	10.2	2.5	—	79.2	77.6	73.3	66.9
TourHAMT2024.03	10.3	6.7	52	50	38	34	29
TourHAMT2025.10	10.3	6.7	—	50	38	34	29
TourHAMT2024.03	10.9	6.8	54	51	38	34	31
TourHAMT2025.10	10.9	6.8	—	51	38	34	31
Memoir (Ours) on ScaleVLN2025.10	10.9	2.6	—	77.2	77.4	72.1	62.2
TourHAMT2024.03	11	6.7	52	51	38	34	28
TourHAMT2025.10	11	6.7	—	51	38	34	28
Memoir (Ours) on DUET2025.10	11	2.8	—	75.2	75.4	69.1	58.8
GR-DUET2025.10	11	3.1	—	74.5	72.7	67.9	54.8
Memoir (w/o retrieval) on DUET2025.10	12.1	3.4	—	69.3	70.8	62.2	52.1
Memoir (w/o retrieval) on GR-DUET2025.10	12.5	3.2	—	70.3	72.7	64	52
Memoir (w/o retrieval) on ScaleVLN2025.10	12.6	2.6	—	74.5	76.8	69.1	56.3
ScaleVLN2025.10	13.5	2.7	—	71.6	76.2	66.5	53.4
DUET2025.10	14.4	3.5	—	65	69.2	58	47