Hello author, how did you do pre-training on ImageNet? Is it the ImageNet-1k dataset? Can you tell me the specific configuration?