[FEATURE] Request for ViT-B/16 hyper-params

[shashankvkt]

Hello,
Thanks for this sharing your amazing work. I was trying to train ViT-B/16 from scratch on ImageNet-1k using the hyperparams reported in the DeIT paper. I'm pretty sure I'm missing something, but I'm unable to reach 81.9% accuracy reported in literature. I get around 78.6% which is even worse than ViT-S/16 (79.8 %).

Could you please share the training command line for ViT-B/16 or share the config file with all hyper-parameters for the same?
Thanks a lot.