TypeError: Linear.forward() got an unexpected keyword argument 'scale'

[kenkenissocool]

Hi, I have been trying to do fine-tuning with stable LoRA, according to the manual. I only can do the basics, so I haven't modified the stable_lora_config.yaml other than the path to dataset folders and video specifications. Therefore I believe the codes aren't contaminated, but this error comes up every time. Anyone have ideas solving this?
Error message:

/root/venv/work2/lib/python3.11/site-packages/diffusers/configuration_utils.py:134: FutureWarning: Accessing config attribute `num_train_timesteps` directly via 'DDPMScheduler' object attribute is deprecated. Please access 'num_train_timesteps' over 'DDPMScheduler's config object instead, e.g. 'scheduler.config.num_train_timesteps'.
  deprecate("direct config name access", "1.0.0", deprecation_message, standard_warn=False)
/root/venv/work2/lib/python3.11/site-packages/diffusers/configuration_utils.py:134: FutureWarning: Accessing config attribute `prediction_type` directly via 'DDPMScheduler' object attribute is deprecated. Please access 'prediction_type' over 'DDPMScheduler's config object instead, e.g. 'scheduler.config.prediction_type'.
  deprecate("direct config name access", "1.0.0", deprecation_message, standard_warn=False)
09/21/2023 07:42:50 - INFO - models.unet_3d_condition - Forward upsample size to force interpolation output size.
Traceback (most recent call last):
  File "/root/another/Text-To-Video-Finetuning/train.py", line 986, in <module>
    main(**OmegaConf.load(args.config))
  File "/root/another/Text-To-Video-Finetuning/train.py", line 848, in main
    loss, latents = finetune_unet(batch, train_encoder=train_text_encoder)
                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/another/Text-To-Video-Finetuning/train.py", line 821, in finetune_unet
    model_pred = unet(noisy_latents, timesteps, encoder_hidden_states=encoder_hidden_states).sample
                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/venv/work2/lib/python3.11/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl
    return forward_call(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/venv/work2/lib/python3.11/site-packages/accelerate/utils/operations.py", line 636, in forward
    return model_forward(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/venv/work2/lib/python3.11/site-packages/accelerate/utils/operations.py", line 624, in __call__
    return convert_to_fp32(self.model_forward(*args, **kwargs))
                           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/venv/work2/lib/python3.11/site-packages/torch/amp/autocast_mode.py", line 14, in decorate_autocast
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "/root/another/Text-To-Video-Finetuning/models/unet_3d_condition.py", line 409, in forward
    sample = transformer_g_c(self.transformer_in, sample, num_frames)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/another/Text-To-Video-Finetuning/models/unet_3d_blocks.py", line 75, in transformer_g_c
    sample = g_c(custom_checkpoint(transformer, mode='temp'), 
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/venv/work2/lib/python3.11/site-packages/torch/utils/checkpoint.py", line 251, in checkpoint
    return _checkpoint_without_reentrant(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/venv/work2/lib/python3.11/site-packages/torch/utils/checkpoint.py", line 432, in _checkpoint_without_reentrant
    output = function(*args, **kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/another/Text-To-Video-Finetuning/models/unet_3d_blocks.py", line 63, in custom_forward
    inputs = module(
             ^^^^^^^
  File "/root/venv/work2/lib/python3.11/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl
    return forward_call(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/venv/work2/lib/python3.11/site-packages/diffusers/models/transformer_temporal.py", line 156, in forward
    hidden_states = block(
                    ^^^^^^
  File "/root/venv/work2/lib/python3.11/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl
    return forward_call(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/venv/work2/lib/python3.11/site-packages/diffusers/models/attention.py", line 197, in forward
    attn_output = self.attn1(
                  ^^^^^^^^^^^
  File "/root/venv/work2/lib/python3.11/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl
    return forward_call(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/venv/work2/lib/python3.11/site-packages/diffusers/models/attention_processor.py", line 426, in forward
    return self.processor(
           ^^^^^^^^^^^^^^^
  File "/root/venv/work2/lib/python3.11/site-packages/diffusers/models/attention_processor.py", line 1013, in __call__
    query = attn.to_q(hidden_states, scale=scale)
            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/venv/work2/lib/python3.11/site-packages/torch/nn/modules/module.py", line 1501, in _call_impl
    return forward_call(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
TypeError: Linear.forward() got an unexpected keyword argument 'scale'

[kenkenissocool]

my_stable_lora_config.txt
This is the stable_lora_config.yaml I modified.

[Samran-Elahi]

any updates on this ? I am facing the same issue

[ykarmesh]

I am also getting the same error.

[ExponentialML]

Hey, sorry for the late response @kenkenissocool! This is due to Diffusers implementing their own version of LoRA in recent versions, which causes this error.

I will look to resolve this very soon.

[Rbrq03]

Everyone, presently, this issue can be effectively addressed through the following steps:
pip uninstall diffusers
pip install diffusers==0.18.1

It works for me, hope it helps

[ShashwatNigam99]

Is there any update for this?