Error in loading the pretrained lora weights

[garychan22]

Hi, I am using the script https://github.com/huggingface/diffusers/blob/main/examples/text_to_image/train_text_to_image_lora_sdxl.py to train a lora.

An error is raised on https://github.com/huggingface/diffusers/blob/73a9d5856f2d7ae3637c484d83cd697284ad3962/examples/text_to_image/train_text_to_image_lora_sdxl.py#L1314C9-L1314C52

Loading adapter weights from state_dict led to missing keys in the model: down_blocks.1.attentions.0.transformer_blocks.0.attn1.to_q.lora_A
.default_0.weight, down_blocks.1.attentions.0.transformer_blocks.0.attn1.to_q.lora_B.default_0.weight, ...

The difference between the keys in the saved lora weights and the ''missing keys'' mentioned above is ''default_0''. How can I resolve this problem?

diffusers 0.32.2
peft 0.15.2

[garychan22]

I found that the error is encountered only when training lora with additional modules besides ["to_k", "to_q", "to_v", "to_out.0"]. After removing convert_state_dict_to_diffusers(*) and manually saving the trained params, the problem went away.