[Bug] Need to import torch

[Tallisgo]

Describe the bug

It is a small problem. When I follow REAMDE Import from Transformers to initialize model， a error (NameError: name 'torch' is not defined) occurs. So it is need to import torch .

Also, I have a problem if internlm-chat-7b can be load on multi gpus since my machine is 11G x 8.

>>> from transformers import AutoTokenizer, AutoModelForCausalLM
>>> tokenizer = AutoTokenizer.from_pretrained("internlm/internlm-chat-7b", trust_remote_code=True)
>>> model = AutoModelForCausalLM.from_pretrained("internlm/internlm-chat-7b", torch_dtype=torch.float16, trust_remote_code=True).cuda()

Environment

torch              1.13.1+cu117
torch-scatter      2.1.1+pt113cu117
torchaudio         0.13.1+cu117
torchvision        0.14.1+cu117

Other information

No response