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

LongDocURL

Benchmarks

Task NameDataset NameSOTA ResultTrend
RetrievalLongDocURL
Recall77.02
30
Document Question AnsweringLongDocURL
Accuracy (All)71.4
30
Multimodal Document Question AnsweringLongDocURL
Overall Acc65.7
28
Document Question AnsweringLongDocURL (test)
Accuracy55.52
16
Long-context Document UnderstandingLongDocURL
Accuracy64.5
14
Document UnderstandingLongDocURL
Accuracy64.5
12
Document UnderstandingLongDocURL
Accuracy (TXT)65.1
8
RetrievalLongDocURL Filtered
MRR@1058.3
6
LongContext UnderstandingLongDocURL
Pass@175.1
5
Visual Document Question AnsweringLongDocURL (Filtered)
Accuracy67.79
5
Showing 10 of 10 rows