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

RIPPLEEDITS

Benchmarks

Task NameDataset NameSOTA ResultTrend
Knowledge EditingRippleEdits POPULAR (full requested-edit set)
Rel.99.2
30
Knowledge EditingRIPPLEEDITS single-instance
Reliability100
16
Knowledge Conflict ResolutionRippleEdits style 40 q
Accuracy77.5
4
Knowledge EditingRippleEdits POPULAR 100 single edits
LG25.7
3
Showing 4 of 4 rows