RuntimeError: CUDA error: unspecified launch failure

Hi there,

Can someone tell me what’s wrong? I am getting a RuntimeError: CUDA error: unspecified launch failure. I have attached the output of python -m torch.utils.collect_env.

This occurs when I try to run:
trainer = Trainer(model=model,args=training_args,data_collator=data_collator,train_dataset=lm_datasets_train,eval_dataset=lm_datasets_valid)


Any help would be greatly appreciated. Thank you.