Fine tuned on facebook/convnext-tiny-224.
Dataset used is available at https://github.com/SY575/COVID19-CT
Fine tuned on training folder of that dataset

key point: if someone do augmentation of that data set can improve performance (my thought)