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

CRAFT

Benchmarks

Task NameDataset NameSOTA ResultTrend
Counterfactual ReasoningCRAFT Hard Split (test)
Accuracy83.64
8
Counterfactual ReasoningCRAFT Easy Split (test)
Accuracy80.05
8
Physical Reasoning Visual Question AnsweringCRAFT VQA
Descriptive Accuracy39.4
4
Named Entity RecognitionCRAFT
F1 Score0.7577
4
Showing 4 of 4 rows