GenEval
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
GenEval 553 prompts
33.6CLIP Score
20
GenEval 553 prompts, 4 samples each (full)
69.59Accuracy
20
GenEval
97GenEval Score
20
GenEval 11 (test)
100Accuracy (Single Obj)
19
GenEval
61.9GE Score
18
GenEval
0.786DINO
18
GenEval original prompts LLM-rewritten prompts
90Overall Accuracy
17
GenEval
77GenEval Score
17
GenEval 29 (test)
100Sing. Acc
16
GENEVAL
91Overall Score
16
GenEval 22
66Overall Score
16
GenEval
87GenEval Score
16
GenEval
88GenEval Score
15
GenEval
0.634DrSim
15
GenEval 129
88Overall Performance Score
15
GenEval official (val)
100Object Presence (1 Obj)
15
GenEval
1.521ImgReward
14
GenEval
80.96VQAScore
14
GenEval++ (test)
95Color Score
13
GenEval official (test)
100Single Object Accuracy
13
GenEval
0.78GenEval Score
13
GenEval
88.5Image Coverage
12
GenEval 2
83.59Object Accuracy
12
GenEval 512x512 resolution
71.6GenEval Score
12
GenEval 80-Obj
100S-Obj
12