Get text output based on only text input

This example code works: https://unum-cloud.github.io/uform/python/index.html#generative-models, but when I remove the `image=` from processor then I get an error

```
RuntimeError: The size of tensor a (254) must match the size of tensor b (509) at non-singleton dimension 2
```

What is the best way to execute the models uform-gen2-dpo/uform-gen2-qwen-500m with just text input?