Editing Alignment Evaluation on TIEdit 1.0 (test)
0.7995SRCCEditProbe
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| EditProbeMode=Cross-modal stage-wise probing2026.03 | 0.7995 | 0.791 | 0.5974 | |
| EditProbeMode=Cross-modal stage-wise probing2026.03 | 0.7795 | 0.7807 | 0.5913 | |
| VGG19Category=Deep Learning-based IQA Models2026.03 | 0.4982 | 0.4777 | 0.334 | |
| PAMSECategory=Traditional Full-Reference (FR) IQA2026.03 | 0.4535 | 0.5223 | 0.5911 | |
| MANIQACategory=Deep Learning-based IQA Models2026.03 | 0.3432 | 0.344 | 0.235 | |
| ImageRewardCategory=Vision-Language and Preference-based2026.03 | 0.2833 | 0.2861 | 0.1919 | |
| GPT-4Category=LLM-based Evaluator2026.03 | 0.223 | 0.1664 | 0.2013 | |
| IW-PSNRCategory=Traditional Full-Reference (FR) IQA2026.03 | 0.2155 | 0.15 | 0.0845 | |
| FSIMCategory=Traditional Full-Reference (FR) IQA2026.03 | 0.2061 | 0.1419 | 0.0776 | |
| BLIPScoreCategory=Vision-Language and Preference-based2026.03 | 0.1721 | 0.1681 | 0.1153 | |
| CLIPScoreCategory=Vision-Language and Preference-based2026.03 | 0.1357 | 0.1296 | 0.0915 |