DROP
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
Drop
45.46PSNR
13
DROP (test)
76.78ROUGE
12
DROP
64.64F1 Score
12
DROP
37.5Accuracy
11
DROP
0.89Pearson r
11
DROP history
51.17F1
11
DROP
41.7EM
11
DROP 1.0 (test)
92.38EM
11
DROP
45.18F1 Score
10
Drop Grayscale
47.05PSNR
10
DROP (dev)
84.1EM
10
DROP (val)
80F1 Score
10
DROP
0.891Score
10
DROP Single-Span questions (dev)
84.2EM
10
DROP Multi-Span span questions (dev)
58.5EM
10
DROP
92.1F1 Score
9
DROP
92.03Refusal F1
8
DROP 2048 tokens
30.2Accuracy
8
DROP
0.284Inference Cost (USD)
8
DROP Numerical Answers Only (test)
72.3Accuracy (Exact Match)
8
DROP v1.0 (dev)
93.01EM
8
Drop Noise-free
42.35PSNR
8
DROP (test)
81.7Accuracy
7
DROP num (dev)
52.8EM
7
DROP Multiple Client (test)
64.73Client 1 Accuracy
6