BEIR
Benchmarks
TaskDatasetSOTA resultTrendResults
BEIR
0.995SciFact
174
BEIR (test)
91.6TREC-COVID Score
130
BEIR v1.0.0 (test)
72.9ArguAna
75
BEIR (test)
59.8NQ
46
BEIR selected subset v1.0.0 (test)
84.7TREC-COVID
38
BEIR
39.1NFCorpus NDCG@10 (Zero-shot)
38
BEIR
0.511NQ NDCG@5
35
BEIR out-of-domain
56.2Average Score (BEIR OOD)
27
BEIR COVID v1 (test)
63.5nDCG@10
26
BEIR
85.51NDCG@10 (Covid)
24
BEIR (test)
0.378NDCG@10 (NFCorpus)
23
BEIR v1 (test)
49.2ArguAna
22
BEIR-7 Average
67.1NDCG@10 (Dense)
21
BEIR 15
0.443NDCG@10
21
BEIR (test)
1ArguAna
20
BEIR v1.0 (test)
94.7FEVER Score
20
BEIR (test)
78.5Covid Score
19
BEIR
85.77nDCG@10 (COVID)
18
BEIR
55.4Average NDCG@10
17
BEIR 9-dataset (test)
42.33NDCG@10 (NFCorpus)
16
BEIR Semantic Relatedness Tasks (test)
47.9ArguAna Score
16
BEIR OOD
41Ar Score
15
BEIR
65.11NDCG@10 (Avg)
14
BEIR-5 all-MiniLM-L6-v2 (test)
52.3Average NDCG@10
14
BeIR collection 12 datasets
60.2nDCG@10
13