PRISM
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
PRISM (test)
66.62Accuracy
51
PRISM
81.62Personalization Win Rate
45
PRISM Personalized
68.06Accuracy
44
PRISM
85.3Average Score
29
Prism
4.627Rating
24
PRISM
59.38Accuracy
20
PRISM
74.5Win-Rate (DPO)
20
PRISM (test)
58.9Accuracy
14
PRISM
80.2Macro-expression Accuracy
13
PRiSM Multilingual Datasets
16.8PFER (DRC)
12
PRiSM Accented English Datasets
8.3PFER (Timing)
12
PRISM
32.159PSNR
11
PRISM Overall
65.3User-level Accuracy
11
PRISM Unseen
0.652User-level Accuracy
11
PRISM Seen
65.3User-level Accuracy
11
PRISM
59.6Accuracy
10
PRISM (test)
0.331BT Score (Mean)
10
PRISM
26.26Final Score
9
PRISM
94.55First-Turn Diversity
8
PRISM
0.46Bias (MAE)
8
PRISM Full (test)
61.11Chile Score
7
PRISM Overall
61.8Accuracy
7
PRISM Unseen
61.6Accuracy
7
PRISM Seen
62.1Accuracy
7
Prism (test)
86.04AI2D Score
6