PPE
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
PPE-Preference
79.8Accuracy
72
PPE Correctness
71.2Accuracy
45
PPE Correlation
87.2Correlation
40
PPE Human Preference track
0.282MSE / PPI
28
PPE
0.034Kuiper
24
PPE (Preference Policy Evaluation)
0.017Kuiper
24
PPE-P
68.3Accuracy
23
PPE
57.8Accuracy
20
PPE Correctness track
0.26MSE / PPI
20
PPE Pref
67.7Accuracy
15
PPE
76.4Accuracy
13
PPE Human
64.6Accuracy
10
PPE
96Precision
8
PPE
76.9PPE Human Preference
8
PPE 10-PDF subsample
62.69F1 Score
6
PPE
52.5Precision
6
PPE (full)
53.91Precision
4
PPE unseen pipeline configurations
0.013MSE
3
PPE unseen pipeline configurations
0.007MSE
3