Image-to-image generation on Landmarks
79.939CLIPScore-ITextTIGER
Evaluation results
| Method | Links | ||
|---|---|---|---|
| TextTIGERBase Model=SD3.5, LLM Backbone=Qwen2.52025.04 | 79.939 | 53.636 | |
| TextTIGERBase Model=Qwen-Img, LLM Backbone=GPT-4o2025.04 | 79.672 | 54.739 | |
| TextTIGERBase Model=Dreamlike, LLM Backbone=Qwen2.52025.04 | 79.073 | 45.557 | |
| TextTIGERBase Model=FLUX, LLM Backbone=Qwen2.52025.04 | 78.433 | 50.335 | |
| Qwen-ImgMode=Cap-Only2025.04 | 69.946 | 40.417 | |
| SD3.5Mode=Cap-Only2025.04 | 68.99 | 36.485 | |
| DreamlikeMode=Cap-Only2025.04 | 68.371 | 29.943 | |
| FLUXMode=Cap-Only2025.04 | 66.518 | 30.015 | |
| PixArtMode=Cap-Only2025.04 | 59.679 | 19.033 |