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

PISTOL

Benchmarks

Task NameDataset NameSOTA ResultTrend
Fine-tuning for knowledge acquisition and abstention preservationPISTOL
FT Score100
14
Contextual Question AnsweringPISTOL (A_C)
ROUGE-L98
6
Direct Question AnsweringPISTOL (A_C)
ROUGE-L77.3
6
Contextual Question AnsweringPISTOL (A_B)
ROUGE-L97.6
6
Direct Question AnsweringPISTOL (A_B)
ROUGE-L79.1
6
Showing 5 of 5 rows