I saw in the appendix that you guys have been able to make it work with the dreambooth model.
I'm aware that you guys are working hard on releasing the code for it but I was just wondering how you could make it work.
If I load the dreambooth lora model using .load_lora_weights, it wipes out the attn_processors from the u-net.
I saw in the appendix that you guys have been able to make it work with the dreambooth model. I'm aware that you guys are working hard on releasing the code for it but I was just wondering how you could make it work. If I load the dreambooth lora model using
.load_lora_weights, it wipes out theattn_processorsfrom the u-net.