ST train example doesn't work on multi gpu

When trying to run https://github.com/AnswerDotAI/ModernBERT/blob/main/examples/train_st.py on a machine with multiple GPU it crashes with

`RuntimeError: Detected that you are using FX to symbolically trace a dynamo-optimized function. This is not supported at the moment.`

Its the same with most existing hugging face examples, will it / should multi GPU be supported?

Using the CUDA_VISIBLE_DEVICES is how I worked around it to train with one GPU.