Running Trainer on CPU

[rdabc]

Hi,

It seems that the trainer is running by default on GPU (from transformers import Trainer), how can I run it on CPU ? I did not find an argument "device" in Trainer.
trainer = Trainer(
model=model,
args=train_args,
train_dataset=train_dataset,
eval_dataset=valid_dataset,
callbacks=[early_stopping_callback],
)
trainer.train()

Thank you

[wgifford]

In train_args set use_cpu=True