Visual Question Answering on Flickr30k
70.8TextVQAZeroshot
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| ZeroshotBackbone=NVILA-Lite 2B, Fine-tuned Layers=None, Update Ratio=N/A, Evaluation Protocol=Zero-shot2026.02 | 70.8 | 51.2 | 69.4 | 78.1 | 42.2 | 62.3 | 27.3 | 44.8 | 38 | |
| Model-DowserBackbone=NVILA-Lite 2B, Fine-tuned Layers=Last 20, Update Ratio=0.1, Evaluation Protocol=Data-Free Importance Probing2026.02 | 70.1 | 48.6 | 68 | 77.9 | 51 | 60.3 | 96 | 79.3 | 75.8 | |
| SPIDERBackbone=NVILA-Lite 2B, Fine-tuned Layers=Last 20, Update Ratio=0.1, Evaluation Protocol=Selective Parameter Update2026.02 | 68 | 44.5 | 67.4 | 75 | 50.7 | 59.4 | 78.5 | 69.7 | 68.5 | |
| GraftingBackbone=NVILA-Lite 2B, Fine-tuned Layers=Last 20, Update Ratio=0.1, Evaluation Protocol=Model Grafting2026.02 | 60.3 | 37.9 | 64.9 | 61.9 | 39.3 | 55.8 | 76.8 | 65.1 | 63 | |
| TailorBackbone=NVILA-Lite 2B, Fine-tuned Layers=Last 20, Update Ratio=0.1, Evaluation Protocol=ModelTailor2026.02 | 49.4 | 33.4 | 66.6 | 66.2 | 40 | 52.6 | 74.1 | 62.7 | 60.7 | |
| Full-FTBackbone=NVILA-Lite 2B, Fine-tuned Layers=Last 20, Update Ratio=0.1, Evaluation Protocol=Full Fine-tuning2026.02 | 48.6 | 31 | 65.2 | 54.2 | 31.8 | 51.7 | 64.3 | 55.7 | 54.3 | |
| DAREBackbone=NVILA-Lite 2B, Fine-tuned Layers=Last 20, Update Ratio=0.1, Evaluation Protocol=Drop & Rescale2026.02 | 48.5 | 30.6 | 64.9 | 52.7 | 30.2 | 50.9 | 63.9 | 55.1 | 53.7 |