Llama3 fine-tune hangs

see tranformers.Trainer.train() hangs with Llama3 base model · Issue #30399 · huggingface/transformers · GitHub