from transformers import AutoTokenizer, AutoModel
import gradio as gr

# chatglm-6b-int4 cuda，本地可以运行成功
# tokenizer = AutoTokenizer.from_pretrained(".\\models\\chatglm-6b-int4", trust_remote_code=True, revision="")
# model = AutoModel.from_pretrained(".\\models\\chatglm-6b-int4", trust_remote_code=True, revision="").half().cuda()


# chatglm-6b CPU，
tokenizer = AutoTokenizer.from_pretrained("THUDM/chatglm-6b", trust_remote_code=True)
model = AutoModel.from_pretrained("THUDM/chatglm-6b", trust_remote_code=True).half()


# chatglm-6b
# kernel_file = "./models/chatglm-6b-int4/quantization_kernels.so"
# tokenizer = AutoTokenizer.from_pretrained("./models/chatglm-6b-int4", trust_remote_code=True, revision="")
# model = AutoModel.from_pretrained("./models/chatglm-6b-int4", trust_remote_code=True, revision="").half().cuda()



# model = model.quantize(bits=model_args.quantization_bit, kernel_file=kernel_file)

model = model.eval()



def chat(msg):
    history = []
    response, history = model.chat(tokenizer, msg, history=history)
    print("response:", response)
    return response


iface = gr.Interface(fn=chat, inputs="text", outputs="text")
iface.launch()