- Loda LLM
from modeling_chatglm import ChatGLMForConditionalGeneration
import torch
torch.set_default_tensor_type(torch.cuda.HalfTensor)
model = ChatGLMForConditionalGeneration.from_pretrained("THUDM/chatglm-6b", trust_remote_code=True, device_map='auto')
- Load LoRA
from peft import PeftModel
model = PeftModel.from_pretrained(model, "visual-openllm/visual-openllm-chatglm-6b-rola")
torch.set_default_tensor_type(torch.cuda.FloatTensor)