Hello, great work. Regarding DreamForge, for the FID metric at 224x400 resolution, was the original NuScenes dataset downsampled to this resolution using Bicubic and then calculated using PyTorch-FID? Given similar performance across other metrics, this is the only one with a significant difference. Are there any points I should be aware of?
Hello, great work. Regarding DreamForge, for the FID metric at 224x400 resolution, was the original NuScenes dataset downsampled to this resolution using Bicubic and then calculated using PyTorch-FID? Given similar performance across other metrics, this is the only one with a significant difference. Are there any points I should be aware of?