This is the repository that tries to use multigpus to finetune the pretrained models.

This is the repositories that change the weights from Jax (Google Research) to Pytorch. And I also provide the code that people can easliy finetune the MLP-Mixer by themselves.