Closed ghost closed 3 years ago
Hi @talent-less , I really wish I could, but I do not have the computing resource to be able to retrain the newer models with the newer datasets. This model I trained with some AWS credits I won from a competition, and they have long since been exhausted.
If I one day come across enough good fortune on the compute & storage side, I will definitely train and release newer models (like xresnet and resnext) trained on the more recent danbooru datasets.
Do you have any plans to train with newer danbooru dataset & larger resnets?