Update qlora.py

[rain-1]

add n_gpus = 0 default value to get_accelerate_model

Fixes this error:

Traceback (most recent call last): File "qlora.py", line 841, in train() File "qlora.py", line 704, in train model, tokenizer = get_accelerate_model(args, checkpoint_dir) File "qlora.py", line 297, in get_accelerate_model max_memory = {i: max_memory for i in range(n_gpus)} UnboundLocalError: local variable 'n_gpus' referenced before assignment

[artidoro]

HI @rain-1, the way QLoRA was set up with Bitsandbytes requires GPUs and doesn't work with CPU only. I don't think we should have a base case of n_gpu=0.