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

STARK

Benchmarks

Task NameDataset NameSOTA ResultTrend
Knowledge Graph RetrievalSTaRK-Amazon 1.0 (Human)
Hits@161.73
32
Knowledge Graph RetrievalSTARK Amazon
H@170.7
25
Knowledge Graph RetrievalSTARK PRIME
H@168.9
25
ReasoningSTARK
EM79.3
24
Knowledge Graph RetrievalSTaRK-Prime Synthetic 1.0
Hits@10.464
20
Knowledge Graph RetrievalSTaRK-MAG Synthetic 1.0
Hits@174.1
20
Knowledge Graph RetrievalSTaRK-Amazon Synthetic 1.0
Hits@164
20
RetrievalSTaRK MAG Synthetic
Recall@2078.8
20
Knowledge Graph RetrievalSTaRK-Prime 1.0 (Human)
Hits@10.4495
16
RetrievalSTaRK PRIME Human
Recall@2066.37
16
RetrievalSTaRK MAG Human
Recall@2056.4
16
RetrievalSTaRK AMAZON (Human)
Recall@2042.43
16
Knowledge Graph RetrievalSTaRK PRIME synthetic (test)
Hit@151.39
13
Knowledge Graph RetrievalSTaRK MAG synthetic (test)
Hit@173.4
13
Question Answering RetrievalSTaRK synthetically generated
Hit@162
12
Knowledge Graph RetrievalSTARK MAG
Hits@139.7
11
Information RetrievalSTaRK AMAZON human-generated (test)
Hit@161.7
10
Information RetrievalSTaRK MAG human-generated (test)
Hit@152.4
10
Information RetrievalSTaRK PRIME human-generated (test)
Hit@157.1
10
RetrievalSTARK AMAZON
Hit@10.319
9
RetrievalSTARK MAG
Hit Rate @ 124.4
9
RetrievalSTARK-PRIME
Hit@119.9
9
Question Answering RetrievalSTaRK human-generated
Hit@155.8
9
Knowledge RetrievalSTARK-PRIME official (test)
Hit@118.44
8
Knowledge RetrievalSTARK MAG official (test)
Hit@144.36
8
Showing 25 of 33 rows