Describe the bug

train_dreambooth_lora.py example no longer runs when installing diffusers from source.
If I run the script using the provided snippet, I get the following error.

Command run:
accelerate launch train_dreambooth_lora.py --pretrained_model_name_or_path="runwayml/stable-diffusion-v1-5" --instance_data_dir="./dog/" --output_dir="./out/" --instance_prompt="a photo of sks dog" --resolution=512 --train_batch_size=1 --gradient_accumulation_steps=1 --checkpointing_steps=100 --learning_rate=1e-4 --lr_scheduler="constant" --lr_warmup_steps=0 --max_train_steps=500 --validation_prompt="A photo of sks dog in a bucket" --validation_epochs=50 --seed="0"

I am feeding it the dataset of the dog example provided (5 images)

Traceback (most recent call last):
File "H:\Github\MLOPs\MLOPs\scripts\python\sdpipeline\train_dreambooth_lora.py", line 1351, in
main(args)
File "H:\Github\MLOPs\MLOPs\scripts\python\sdpipeline\train_dreambooth_lora.py", line 1142, in main
model_pred = unet(noisy_model_input, timesteps, encoder_hidden_states).sample
File "C:\Users\Paul\AppData\Local\Programs\Python\Python39\lib\site-packages\torch\nn\modules\module.py", line 1501, in _call_impl
return forward_call(*args, **kwargs)
File "C:\Users\Paul\AppData\Local\Programs\Python\Python39\lib\site-packages\diffusers\models\unet_2d_condition.py", line 718, in forward
sample = self.conv_in(sample)
File "C:\Users\Paul\AppData\Local\Programs\Python\Python39\lib\site-packages\torch\nn\modules\module.py", line 1501, in _call_impl
return forward_call(*args, **kwargs)
File "C:\Users\Paul\AppData\Local\Programs\Python\Python39\lib\site-packages\torch\nn\modules\conv.py", line 463, in forward
return self._conv_forward(input, self.weight, self.bias)
File "C:\Users\Paul\AppData\Local\Programs\Python\Python39\lib\site-packages\torch\nn\modules\conv.py", line 459, in _conv_forward
return F.conv2d(input, weight, bias, self.stride,
RuntimeError: Given groups=1, weight of size [320, 4, 3, 3], expected input[1, 3, 512, 512] to have 4 channels, but got 3 channels instead

Reproduction

Run: (The example on the Github page for Dreambooth training)

accelerate launch train_dreambooth_lora.py --pretrained_model_name_or_path="runwayml/stable-diffusion-v1-5" --instance_data_dir="./dog/" --output_dir="./out/" --instance_prompt="a photo of sks dog" --resolution=512 --train_batch_size=1 --gradient_accumulation_steps=1 --checkpointing_steps=100 --learning_rate=1e-4 --lr_scheduler="constant" --lr_warmup_steps=0 --max_train_steps=500 --validation_prompt="A photo of sks dog in a bucket" --validation_epochs=50 --seed="0"

Logs

No response

System Info

• diffusers version: 0.17.0.dev0
• Platform: Windows-10-10.0.22621-SP0
• Python version: 3.9.10
• PyTorch version (GPU?): 2.0.0+cu117 (True)
• Huggingface_hub version: 0.14.1
• Transformers version: 4.29.0.dev0
• Accelerate version: 0.18.0
• xFormers version: 0.0.19
• Using GPU in script?: YES
• Using distributed or parallel set-up in script?: NO