Can you provide more training parameters of the model? Like the Gernet series

[aojue1109]

Can you provide more training parameters of the model? Like the gernet, efficientv2, efficient_pure series

[rwightman]

@aojue1109 as per comments in model impl file, https://github.com/rwightman/pytorch-image-models/blob/master/timm/models/byobnet.py#L7-L9 ... I didn't train those.

Generally most models I train these days are using one of the recipes in ResNet strikes back: https://arxiv.org/abs/2110.00476 ... closer to the paper A or C recipes for models that re more resnet like, and B or C for more EfficientNet / MobileNet like (lots of depthwise separable convs)