DPG-Bench
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Text-to-Image Generation | DPG-Bench | Overall Score88.63 | 510 | |
| Text-to-Image Generation | DPG-Bench | DPG Score88.9 | 156 | |
| Text-to-image generation | DPG-Bench | Average Score88.79 | 77 | |
| Text-to-Image Generation | DPG-Bench (test) | Overall Fidelity88.32 | 68 | |
| Text-to-Image Generation | DPG-Bench | DPG Score89 | 36 | |
| Text-to-Image Generation | DPG-Bench | Overall Score88.32 | 30 | |
| Text-to-Image | DPG-Bench | DPG-Bench Score88.32 | 27 | |
| Text-to-Image Alignment | DPG-Bench | Global Alignment Score92.64 | 24 | |
| Dense prompt following | DPG-Bench v1.0 (test) | Entity Score93.08 | 20 | |
| Image Generation | DPG-Bench 130 | Score88.3 | 15 | |
| Text-to-Image Generation | DPG-Bench | Soft-TIFA Arithmetic Mean91.14 | 12 | |
| Text-to-Image Faithfulness | DPG-Bench | Faithfulness Yes-Ratio97 | 11 | |
| Text-to-Image Generation | DPG-Bench | DPG Percentage Score88.79 | 11 | |
| Dense Prompt Alignment | DPG-Bench | Overall Score85.08 | 11 | |
| Image Generation | DPG-Bench | DPG Score86.75 | 9 | |
| Text-to-Image Generation | DPG-Bench 17 | Global Score90.97 | 8 | |
| Human preference, image quality, and aesthetics comparison | DPG-Bench | DeQA Score4.12 | 8 | |
| Compositional Reasoning | DPG-Bench | Overall Score88.32 | 7 | |
| Text-to-3D Generation | DPG-Bench 1.0 (test) | Global Score81.82 | 7 | |
| Unified understanding-and-generation | DPG-Bench | KV Size (GB)0.07 | 6 | |
| Dense prompt-following | DPG-Bench | Score85.43 | 6 | |
| Text-to-image alignment | DPG-Bench (test) | DPG77.13 | 6 | |
| Instruction following | DPG-Bench | GPT-4o Score56.1 | 5 | |
| Text-to-Image Generation | DPG-Bench short | DPG-Bench Score72.881 | 5 | |
| Text-to-Image Generation | DPG-Bench zero-shot | DPG-Bench Score (Zero-Shot)70.1 | 5 |
Showing 25 of 27 rows