Problem when erasing multi-concepts

[Ferry-Li]

In cfr_lora_training.py:

for i in range(train_dataset._concept_num): # the number of concept/lora
        attn_controller = AttnController()
        if i != 0:
            unet.set_default_attn_processor()
        for name, m in unet.named_modules():
            if name.endswith('attn2') or name.endswith('attn1'):
                cross_attention_dim = None if name.endswith("attn1") else unet.config.cross_attention_dim
                if name.startswith("mid_block"):
                    hidden_size = unet.config.block_out_channels[-1]
                elif name.startswith("up_blocks"):
                    block_id = int(name[len("up_blocks.")])
                    hidden_size = list(reversed(unet.config.block_out_channels))[block_id]
                elif name.startswith("down_blocks"):
                    block_id = int(name[len("down_blocks.")])
                    hidden_size = unet.config.block_out_channels[block_id]
                m.set_processor(LoRAAttnProcessor(
                    hidden_size=hidden_size, 
                    cross_attention_dim=cross_attention_dim, 
                    rank=args.rank, 
                    attn_controller=attn_controller, 
                    module_name=name, 
                    preserve_prior=args.with_prior_preservation,
                ))
        ### set lora
        lora_attn_procs = {}
        for key, value in zip(unet.attn_processors.keys(), unet.attn_processors.values()):
            if key.endswith("attn2.processor"):
                lora_attn_procs[f'{key}.to_k_lora'] = value.to_k_lora
                lora_attn_procs[f'{key}.to_v_lora'] = value.to_v_lora
                # lora_attn_procs[f'{key}.to_q_lora'] = value.to_q_lora
                # lora_attn_procs[f'{key}.to_out_lora'] = value.to_out_lora
        # unet.set_attn_processor(lora_attn_procs)
        lora_layers = AttnProcsLayers(lora_attn_procs)

it will report AttributeError: 'UNet2DConditionModel' object has no attribute 'attn_processors'. Did you mean: 'set_attn_processor'? when i > 0 at the line for key, value in zip(unet.attn_processors.keys(), unet.attn_processors.values()):.

I have checked my diffusers==0.22.0. Do you have any ideas about this error?

[Shilin-LU]

This issue is likely caused by a version mismatch with diffusers. Although you've already checked this, I couldn't replicate the issue on my side. You might want to try reinstalling your environment to see if that resolves the problem.