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

KAMEL

Benchmarks

Task NameDataset NameSOTA ResultTrend
Open-set knowledge retrievalKAMEL
Macro EM16.2
19
Model EditingKAMEL multi-token targets (N=4)
Efficacy73
4
Model EditingKAMEL multi-token targets (N=3)
Efficacy76
4
Model EditingKAMEL multi-token targets N=2
Efficacy88
4
Editing EfficacyKAMEL 4 tokens
Editing Efficacy95
4
Editing EfficacyKAMEL 3 tokens
Editing Efficacy96.5
4
Editing EfficacyKAMEL 2 tokens
Editing Efficacy98.5
4
Editing EfficacyKAMEL 1 token
Editing Efficacy100
4
Showing 8 of 8 rows