Finetune the Autoencoder

[liuzhihui2046]

Hello, thank you for your work. As the current autoencoder is not very consistent with my existing data, I would like to fine-tune the encoder and decoder based on this. However, I cannot find the corresponding training configuration, especially for the loss configuration. Could you please provide me with some help? Thank you very much!

[YihanHu-2022]

For the finetuning of autoencoder you can refer to the settings of "Latent Diffusion Model" (https://github.com/CompVis/latent-diffusion)

[GISStd]

I meet the same question，how do you deal with it