mac本用llama-factory如何用MPS ？

[catman002]

Reminder

• I have read the above rules and searched the existing issues.

Description

求助 mac本用llama-factory采用GPU(MPS)微调训练

Pull Request

以下是我的参数，但只能用cpu微调训练：
···
llamafactory-cli train
--stage sft
--do_train True
--model_name_or_path /Users/brody/ai/huggingface/llms/Qwen2.5-0.5B-Instruct-root/Qwen2.5-0.5B-Instruct
--preprocessing_num_workers 16
--finetuning_type full
--template qwen
--flash_attn auto
--dataset_dir /Users/brody/ai/huggingface/dataset/ruozhiba
--dataset datasets
--cutoff_len 512
--learning_rate 5e-05
--num_train_epochs 5.0
--max_samples 100000
--per_device_train_batch_size 8
--gradient_accumulation_steps 4
--lr_scheduler_type cosine
--max_grad_norm 1.0
--logging_steps 5
--save_steps 100
--warmup_steps 0
--packing False
--report_to none
--output_dir /Users/brody/ai/huggingface/dataset/ruozhiba/saves
--plot_loss True
--trust_remote_code True
--ddp_timeout 180000000
--include_num_input_tokens_seen True
--optim adamw_torch
--val_size 0.2
--eval_strategy steps
--eval_steps 100
--per_device_eval_batch_size 8
--no_cuda true
--bf16 False
···

mac本用llama-factory如何用MPS ？