from modelscope import snapshot_download
model_dir = snapshot_download("ZhipuAI/chatglm3-6b-32k",cache_dir="/root/autodl-tmp/chatglm3", revision = "v1.0.0")
from transformers import AutoTokenizer, AutoModel
model = AutoModel.from_pretrained("/root/autodl-tmp/chatglm3/ZhipuAI/chatglm3-6b-32k/", trust_remote_code=True, device="cuda").eval()
tokenizer = AutoTokenizer.from_pretrained("/root/autodl-tmp/chatglm3/ZhipuAI/chatglm3-6b-32k/", trust_remote_code=True)
response, history = model.chat(tokenizer, "你好", history=[])
print(response)
chatglm3推理报错:too many values to unpack (expected 2)
原因是transformers版本过高,修改为transformers==4.41.2
修改后bug解决