Image Captioning on Flickr30K Native Portuguese (test)
52.49CIDEr-DViTucano 2B
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| ViTucano 2B#Params=2.88B, Evaluation Mode=Zero-shot2026.01 | 52.49 | 16.46 | 37.75 | 43.47 | 68.87 | 56.28 | |
| ViTucano 1B#Params=1.53B, Evaluation Mode=Zero-shot2026.01 | 51.45 | 16.16 | 37.96 | 43.73 | 69.04 | 55.96 | |
| LLaMa3.2 V#Params=11.70B, Evaluation Mode=Zero-shot2026.01 | 34.94 | 9.81 | 30.81 | 34.5 | 62.84 | 56.95 | |
| GPT-4oEvaluation Mode=Zero-shot2026.01 | 34.02 | 8.93 | 27.36 | 41.18 | 63.35 | 62.07 | |
| GPT-4o-miniEvaluation Mode=Zero-shot2026.01 | 27.43 | 9.04 | 29.29 | 41.96 | 64.24 | 61.26 | |
| Phi-3 V#Params=4.15B, Evaluation Mode=Zero-shot2026.01 | 21.92 | 7.34 | 27.94 | 30.29 | 58.81 | 52.36 | |
| PaliGemma#Params=2.92B, Evaluation Mode=Zero-shot2026.01 | 18.69 | 4.65 | 20.59 | 19.78 | 52.15 | 49.67 |