Universal Physics Transformers: A Framework For Efficiently Scaling Neural Operators Andreas Fürst 1 Simon Schmid