Video-grounded Dialogue on DSTC7 (test) benchmark leaderboard · SOTA2 Research

Video-grounded Dialogue on DSTC7 (test)

47.4BLEU-4V²Dial

Evaluation Results

Method	Links
V²Dial2025.03		47.4	78.9	66.5	56.1	31.2	62.3	139.8
MST-MIXER2025.03		47.1	78.4	66	55.8	31	62	136.5
THAM2025.03		46.8	77.8	65.4	54.9	30.8	61.9	133.5
RLM2025.03		45.9	76.5	64.3	54.3	29.4	60.6	130.8
DialogMCF2025.03		45.7	77.7	65.3	54.7	30.6	61.3	135.2
P-DCI2025.03		44.9	77	65.3	53.9	29.2	60.6	129.5
VGD2025.03		43.6	74.9	62	52	28.2	58.2	119.4
SCGA2025.03		43	74.5	62.2	51.7	28.5	57.8	120.1
BIST2025.03		42.9	75.5	61.9	51	28.4	58.1	119.2
VGNMN2025.03		42.9	—	—	—	27.8	57.8	118.8
MRLV2025.03		41.5	—	59.2	49.3	26.9	56.9	115.9
AV-TRN2025.03		40.6	—	—	—	26.2	55.4	107.9
COST2025.03		40	72.3	58.9	48.3	26.6	56.1	108.5
MTN2025.03		39.2	71.5	58.1	47.6	26.9	55.9	106.6
JMAN2025.03		33.4	66.7	52.1	41.3	23.9	53.3	94.1
VideoLLAMA 22025.03		23.2	56.2	41.1	30.7	26.4	48.5	79.2
MTN2019.07		0.135	0.356	0.242	0.174	0.165	0.365	1.366
Entry-top12019.07		0.131	0.331	0.231	0.171	0.157	0.363	1.36
MTN2019.07		0.128	0.357	0.241	0.173	0.162	0.355	1.249
Entry-top22019.07		0.126	0.329	0.228	0.167	0.154	0.357	1.306
Entry-top32019.07		0.123	0.327	0.225	0.164	0.155	0.35	1.269
Entry-top42019.07		0.115	0.312	0.21	0.152	0.148	0.357	1.271
Entry-top52019.07		0.114	0.329	0.216	0.153	0.14	0.331	1.103
Hori et al.2019.07		0.095	0.279	0.183	0.13	0.122	0.303	0.905