Question about Fine-tuning Configuration on NYU Dataset

[dxw2000]

In Table III of the paper, the performance on the NYU validation set is evaluated. Therefore, I assume that your method was fine-tuned on the NYU dataset. Could you please share the fine-tuning configuration used in your experiments? Specifically, did you fine-tune only certain layers, or was a full-parameter fine-tuning performed? I would greatly appreciate any details you could provide about the fine-tuning setup described in the paper.