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

LongDocURL

Benchmarks

Task NameDataset NameSOTA ResultTrend
Document Question AnsweringLongDocURL
Accuracy (All)71.4
30
Multimodal Document Question AnsweringLongDocURL
Overall Acc60.7
21
RetrievalLongDocURL
Recall77.02
18
Long-context Document UnderstandingLongDocURL
Accuracy64.5
14
Document UnderstandingLongDocURL
Accuracy64.5
12
RetrievalLongDocURL Filtered
MRR@1058.3
6
Visual Document Question AnsweringLongDocURL (Filtered)
Accuracy67.79
5
Showing 7 of 7 rows