GenEval
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
GenEval
10.02Average Generation Time (s)
6
GenEval baseline (test)
51.44Overall Alignment Score
6
GenEval 2
1Execution Time (s)
6
GenEval v2 (test)
58.84Prompt Score
6
GenEval counting and position
96.3Count Score
6
GenEval 2
1.7FLOPs
6
GenEval
0.477DreamSim Score
6
GenEval Descriptive Concepts Attribute Binding Stable Diffusion setup (100 unique pairs)
55.75GenEval Success Rate (Images)
5
GenEval v1 (original)
1Execution Metric
5
GenEval 2
34.16Prompt Length
5
GenEval 2
4.36Human Visual Annotation Score
4
GenEval long prompts LLM-refined
47.8GenEval Score
4
Geneval
90Single Object Accuracy
4
GenEval
100Single Obj. Score
4
GenEval 2
85.41Attribute Accuracy
4
GenEval w/ TTS (test)
92GenEval Score
4
GENEVAL
79Accuracy
4
GenEval
3Average Attempts
4
GenEval
121Satisfied Score
4
GenEval 256x256
0.96Single Object Composition
4
GenEval 2
52.2Atomicity Score (N=3)
3
GenEval2
0.541LPIPS
3
GenEval
1Single Object Accuracy
3
GenEval single-object
90Win Rate vs i.i.d.
3
GenEval 2 (test)
56.7Soft TIFA AM Score
2