Contextual Image Captioning on Concadia (test)
93.8CIDErSMs w/ description
Evaluation results
| Method | Links | |
|---|---|---|
| SMs w/ descriptionConditioning=Image, article text, and ground truth description2022.04 | 93.8 | |
| VLISZero-shot=true2023.10 | 44.1 | |
| SMsMode=Zero-shot2022.04 | 38.9 | |
| Socratic ModelZero-shot=true2023.10 | 38.9 | |
| Naïve EnsembleZero-shot=true2023.10 | 24.7 | |
| BLIP-2Zero-shot=true2023.10 | 20 | |
| Kreiss et al.Mode=Finetuned2022.04 | 11.3 | |
| Kreiss et al.Zero-shot=false2023.10 | 11.3 |