accelerate多卡训练报错

[xienan0326]

Reminder

• I have read the README and searched the existing issues.

Reproduction

accelerate launch --config_file config.yaml src/train_bash.py
--stage dpo
--do_train
--model_name_or_path out_dir/sft_test
--dataset comparison_gpt4_zh
--template default
--finetuning_type full
--output_dir out_dir/dpo_test
--per_device_train_batch_size 8
--gradient_accumulation_steps 2
--lr_scheduler_type cosine
--logging_steps 10
--save_steps 1000
--learning_rate 1e-5
--num_train_epochs 1.0
--plot_loss
--fp16

compute_environment: LOCAL_MACHINE
debug: false
distributed_type: MULTI_GPU
downcast_bf16: 'no'
gpu_ids: all
machine_rank: 0
main_training_function: main
mixed_precision: fp16
num_machines: 1
num_processes: 4
rdzv_backend: static
same_network: true
tpu_env: []
tpu_use_cluster: false
tpu_use_sudo: false
use_cpu: false

Traceback (most recent call last):
File "/workspace/LLaMA-Factory/src/train_bash.py", line 14, in
main()
File "/workspace/LLaMA-Factory/src/train_bash.py", line 5, in main
run_exp()
File "/workspace/LLaMA-Factory/src/llmtuner/train/tuner.py", line 38, in run_exp
run_dpo(model_args, data_args, training_args, finetuning_args, callbacks)
File "/workspace/LLaMA-Factory/src/llmtuner/train/dpo/workflow.py", line 47, in run_dpo
trainer = CustomDPOTrainer(
File "/workspace/LLaMA-Factory/src/llmtuner/train/dpo/trainer.py", line 62, in init
self.ref_model = self.accelerator.prepare_model(self.ref_model, evaluation_mode=True)
File "/opt/conda/lib/python3.10/site-packages/accelerate/accelerator.py", line 1287, in prepare_model
raise ValueError(
ValueError: You can't train a model that has been loaded with device_map='auto' in any distributed mode. Please rerun your script specifying --num_processes=1 or by launching with python {{myscript.py}}.

Expected behavior

如何修复，正常进行训练

System Info

No response

Others

No response

[hiyouga]

报啥错？

[xienan0326]

报啥错？

Traceback (most recent call last):
File "/workspace/LLaMA-Factory/src/train_bash.py", line 14, in
main()
File "/workspace/LLaMA-Factory/src/train_bash.py", line 5, in main
run_exp()
File "/workspace/LLaMA-Factory/src/llmtuner/train/tuner.py", line 38, in run_exp
run_dpo(model_args, data_args, training_args, finetuning_args, callbacks)
File "/workspace/LLaMA-Factory/src/llmtuner/train/dpo/workflow.py", line 47, in run_dpo
trainer = CustomDPOTrainer(
File "/workspace/LLaMA-Factory/src/llmtuner/train/dpo/trainer.py", line 62, in init
self.ref_model = self.accelerator.prepare_model(self.ref_model, evaluation_mode=True)
File "/opt/conda/lib/python3.10/site-packages/accelerate/accelerator.py", line 1287, in prepare_model
raise ValueError(
ValueError: You can't train a model that has been loaded with device_map='auto' in any distributed mode. Please rerun your script specifying --num_processes=1 or by launching with python {{myscript.py}}.

[hiyouga]

更新代码重试一下