ViDoRe
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
ViDoRe Avg. across 4 datasets v2
0.58Full NDCG
45
ViDoRe V2
66.9Avg nDCG@5
39
ViDoRe v2 (test)
66.4ESG Score (Human)
23
ViDoRe AI v1 (test)
1ASR-G Similarity (mean) @-1
23
ViDoRe AI v1 (test)
0.19Delta Recall@1
23
ViDoRe V3
66.32HR
23
ViDoRe V1
88.1Arxiv Score
23
ViDoRe2 Average (val)
63.3nDCG@5
18
ViDoRe ESG Full 2 (val)
63.1nDCG@5
18
ViDoRe2 ESG Human (val)
72.6nDCG@5
18
ViDoRe Biomed 2 (val)
61.2nDCG@5
18
ViDoRe 1
94.1Avg Performance Score
18
ViDoRe AI V1
100ASR-R@1 (mean)
18
ViDoRe English
60NDCG@5
16
ViDoRe V1+V2 (test)
93.8ArXiv NDCG@5
15
ViDoRe 8 datasets v3
59NDCG@5
14
ViDoRe 10 datasets v1
90.6NDCG@5
14
ViDoRe Out-of-domain V2
67.9NDCG@5 (ESG Human)
13
ViDoRe
87.3ArxivQ Score
12
ViDoRe V3 (test)
79.3Accuracy (Computer Science)
12
ViDoRe 1
65.15Performance Score
12
ViDoRe In-Domain V1 (test)
88Arxiv Domain Performance
12
Vidore 2
72.8nDCG@10
11
Vidore v1
91.9nDCG@10
11
ViDoRe AI (Targeted Setting II: 5 queries, 1 answer) V1
88ASR @1 (mean)
10