Whether lohaconfig is supported in the convert_state_dict_to_diffusers method

In the train_text_to_image_lora.py file

unet_lora_config = LoraConfig(
r=cfg.rank,
lora_alpha=cfg.rank,
init_lora_weights="gaussian",
target_modules=["to_k", "to_q", "to_v", "to_out.0"],
)
modified to

unet_lora_config = LoHaConfig(
r=cfg.rank,
alpha=cfg.rank,
target_modules=["to_k", "to_q", "to_v", "to_out.0"],
),

unet_lora_state_dict = convert_state_dict_to_diffusers(
get_peft_model_state_dict(unwrapped_unet)
)
in this line, an error will occur. Please tell me how to modify it.

Start in train_text_to_image_lora.py and trace the convert_state_dict_to_diffusers call after replacing LoraConfig with LoHaConfig. Reproduce the reported error, then identify the required LoHa state-dict conversion behavior; done means the LoHa configuration path converts successfully without the reported error.

Scritto dal modello di indicizzazione a partire dal testo della issue.