Robust R2D2 DNN series for monochromatic intensity imaging with VLA

The dataset consists of R2D2 DNN series underpinned by two core architectures: U-Net and the U-WDSR. The proposed novel U-WDSR architecture integrates the WDSR residual body with the U-Net framework. These models are specifically trained to reconstruct monochromatic intensity radio images of size up to 512x512 from radio-interferometric data acquired by the Very Large Array (VLA), accommodating various data-weighting schemes and angular resolutions (i.e., pixel sizes). Five realisations of each DNN series are provided, described as follows:

1. R2D2_{A1, T2}: A series of 25 DNNs using U-Net as the core architecture.

2. R2D2_{A1, T2}: A series of 25 DNNs using U-WDSR as the core architecture.

All DNN models are provided as PyTorch models for use in Python environments.