Will the tf_efficientnet_b8 hyperparam be released

[ldd91]

@rwightman Tthankyou for this wonderful repo!I want to train tf_efficientnet_b8 with ImageNet myself,but i can not find the hyperparam,will it be released?I wonder if you could share the training configs

[rwightman]

@ldd91 training B8 at it's proper resolution requires GPU resources that I don't have so I've never tried it. I know others have used the larger models to finetune, sometimes at lower res. I'd say you're on your own adapting the hyper params from the smaller models to a multi-GPU and probably multi-node scenario.