Finetuning GPT2 on multiple GPUs

Does anyone have experience with finetuning GPT2 on multiple GPUs?

Have problems using
“python -m fastai.launch scriptname.py”
as I address several GPUs

Training as in the Tutorial