TypeError: UVit2DModel.forward.<locals>.layer_() got an unexpected keyword argument 'encoder_hidden_states'

### Describe the bug

When I run the[ amused example](https://github.com/huggingface/diffusers/tree/main/examples/amused), I got this error: TypeError: UVit2DModel.forward..layer_() got an unexpected keyword argument 'encoder_hidden_states'.

Looks like the transformer layer does not have this encore_hidden_state attributes.

### Reproduction

Run the amused example

### Logs

```shell

```

### System Info

newest diffuser GitHub repository

### Who can help?

_No response_