About unnoised meta-surrogate model

[sustech-lz]

Hi CownowAn,
Upon reviewing the code provided by you in the 'diffusionnag' repository, I noticed that instructions were provided for training 'checkpoints/meta_surrogate/check_point.pth.tar'. However, I couldn't find any specific guidance on how to train a model to generate 'unnoised_checkpoint.pth.tar'.

Could you kindly provide insights into the training procedure for generating the 'unnoised_checkpoint.pth.tar' file?