Knowledge-informed text-to-image generation on WISE 96
63Cultural ScoreQwen-Image
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Qwen-Image#Params=20B2026.03 | 63 | 62 | 76 | 60 | 72 | 39 | 63 | |
| InternVL-U#Params=2B+1.7B, w/ CoT=true2026.03 | 55 | 57 | 74 | 51 | 72 | 46 | 58 | |
| UniWorld-V1#Params=7B+13B2026.03 | 53 | 55 | 73 | 45 | 59 | 41 | 55 | |
| FLUX.1 [dev]#Params=12B2026.03 | 48 | 58 | 62 | 42 | 51 | 35 | 50 | |
| BAGEL#Params=7B+7B2026.03 | 44 | 52 | 65 | 42 | 62 | 41 | 49 | |
| SD3-Medium#Params=2B2026.03 | 43 | 50 | 52 | 41 | 53 | 33 | 45 | |
| InternVL-U#Params=2B+1.7B2026.03 | 37 | 51 | 68 | 39 | 62 | 39 | 46 | |
| Ovis-U1#Params=2.4B+1.2B2026.03 | 36 | 46 | 64 | 35 | 52 | 28 | 42 | |
| Lumina-DiMOO#Params=8B2026.03 | 35 | 43 | 59 | 31 | 49 | 34 | 40 | |
| Janus-Pro#Params=7B2026.03 | 30 | 37 | 49 | 36 | 42 | 26 | 35 |