Visual Question Answering on ImageNet-R
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 | 22.2 | 42.2 | 32.7 | |
| Model-DowserBackbone=NVILA-Lite 2B, Fine-tuned Layers=Last 20, Update Ratio=0.1, Evaluation Protocol=Data-Free Importance Probing2026.02 | 64.4 | 45.7 | 68.6 | 75.7 | 38.8 | 59.4 | 90.3 | 74.5 | 71.2 | |
| SPIDERBackbone=NVILA-Lite 2B, Fine-tuned Layers=Last 20, Update Ratio=0.1, Evaluation Protocol=Selective Parameter Update2026.02 | 42.5 | 26.1 | 54.3 | 68 | 38.9 | 40.9 | 91.9 | 68.5 | 60.5 | |
| TailorBackbone=NVILA-Lite 2B, Fine-tuned Layers=Last 20, Update Ratio=0.1, Evaluation Protocol=ModelTailor2026.02 | 28.8 | 18.7 | 23.9 | 60.2 | 42.5 | 26.2 | 80.3 | 66.9 | 47.2 | |
| Full-FTBackbone=NVILA-Lite 2B, Fine-tuned Layers=Last 20, Update Ratio=0.1, Evaluation Protocol=Full Fine-tuning2026.02 | 16 | 10.8 | 13.3 | 43 | 33.1 | 14.2 | 92.3 | 57 | 35.2 | |
| GraftingBackbone=NVILA-Lite 2B, Fine-tuned Layers=Last 20, Update Ratio=0.1, Evaluation Protocol=Model Grafting2026.02 | 15.6 | 10.4 | 12.1 | 43.3 | 34.4 | 13.3 | 90 | 58 | 40.4 | |
| DAREBackbone=NVILA-Lite 2B, Fine-tuned Layers=Last 20, Update Ratio=0.1, Evaluation Protocol=Drop & Rescale2026.02 | 15.6 | 10.4 | 12.1 | 43.3 | 34.4 | 13.3 | 92.4 | 56.9 | 34.9 |