import os

from transformers import AutoTokenizer
from vllm import LLM, SamplingParams
from huggingface_hub import snapshot_download

model_path = "happzy2633/qwen2.5-7b-ins-v3"

tokenizer = AutoTokenizer.from_pretrained(model_path)
sampling_params = SamplingParams(temperature=0.7, top_p=0.8, repetition_penalty=1.05, max_tokens=8192)
llm = LLM(model=model_path)

def api_call_batch(batch_messages):
    text_list = [
        tokenizer.apply_chat_template(conversation=messages, tokenize=False, add_generation_prompt=True, return_tensors='pt')
        for messages in batch_messages
    ]
    outputs = llm.generate(text_list, sampling_params)
    result = [output.outputs[0].text for output in outputs]
    return result

def api_call(messages):
    return api_call_batch([messages])[0]

def call_gpt(history, prompt):
    return api_call(history+[{"role":"user", "content":prompt}])

if __name__ == "__main__":
    messages = [{"role":"user", "content":"你是谁？"}]
    breakpoint()
    print(api_call_batch([messages]*4))