Let's train vision transformers (ViT) for cifar 10 / cifar 100!
Role in this project:
ML Engineer Contributions:4 releases, 61 commits, 18 PRs in 1 year 10 months
Contributions summary:Kentaro implemented a Vision Transformer (ViT) model and training scripts for the CIFAR-10 dataset. They added the ViT model definition, incorporating components like attention, feedforward layers, and positional embeddings. The user also integrated the training script, loss functions and optimizers. Additional commits added logging, results logging, and made the kernel size configurable.
cifar100vision-transformer
Contributions:17 commits, 1 PR, 16 pushes in 1 year 11 months