Closed bityangke closed 1 year ago
Thank you very much for contributing such a good code base for FSL. In the paper, did you use fine-tuning on Mini imagenet?
Thank you for your interests! For Mini-ImageNet, the fine-tuning step was skipped as the performance was too good due to pre-trained ViT. I'm pretty sure fine-tuning will push the performance to ~100%
Thank you very much for contributing such a good code base for FSL. In the paper, did you use fine-tuning on Mini imagenet?