前言

        为了便捷构建 LLM 应用，需要基于本地部署的LLM模型，自定义一个 LLM 类，将LLM接入到 LangChain 框架中。完成自定义 LLM 类之后，可以以完全一致的方式调用 LangChain 的接口，而无需考虑底层模型调用的不一致。

1.本地部署

1.1 本地部署代码准备

        基于本地部署自定义 LLM 类并不复杂，我们只需从 LangChain.llms.base.LLM 类继承一个子类，并重写构造函数与 _call 函数即可。

from langchain.llms.base import LLM
from typing import Any, List, Optional
from langchain.callbacks.manager import CallbackManagerForLLMRun
from transformers import AutoTokenizer, AutoModelForCausalLM, GenerationConfig, LlamaTokenizerFast
import torch

class Qwen2_LLM(LLM):
    # 基于本地 Qwen2 自定义 LLM 类
    tokenizer: AutoTokenizer = None
    model: AutoModelForCausalLM = None
        
    def __init__(self, mode_name_or_path :str):

        super().__init__()
        print("正在从本地加载模型...")
        self.tokenizer = AutoTokenizer.from_pretrained(mode_name_or_path, use_fast=False)
        self.model = AutoModelForCausalLM.from_pretrained(mode_name_or_path, torch_dtype=torch.bfloat16, device_map="auto")
        self.model.generation_config = GenerationConfig.from_pretrained(mode_name_or_path)
        print("完成本地模型的加载")
        
    def _call(self, prompt : str, stop: Optional[List[str]] = None,
                run_manager: Optional[CallbackManagerForLLMRun] = None,
                **kwargs: Any):

        messages = [{"role": "user", "content": prompt }]
        input_ids = self.tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
        model_inputs = self.tokenizer([input_ids], return_tensors="pt").to('cuda')
        generated_ids = self.model.generate(model_inputs.input_ids,max_new_tokens=512)
        generated_ids = [
            output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
        ]
        response = self.tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
        
        return response
    @property
    def _llm_type(self) -> str:
        return "Qwen2_LLM"

        在上述类定义中，我们分别重写了构造函数和 _call 函数：对于构造函数，我们在对象实例化的一开始加载本地部署的 Qwen2 模型，从而避免每一次调用都需要重新加载模型带来的时间过长；_call 函数是 LLM 类的核心函数，LangChain 会调用该函数来调用 LLM，在该函数中，我们调用已实例化模型的 chat 方法，从而实现对模型的调用并返回调用结果。

        在整体项目中，我们将上述代码封装为 LLM.py，后续将直接从该文件中引入自定义的 LLM 类。

1.2 调用 

        像使用任何其他的langchain大模型功能一样使用即可。

from LLM import Qwen2_LLM
llm = Qwen2_LLM(mode_name_or_path = "/root/autodl-tmp/qwen/Qwen1.5-7B-Chat")
llm("你是谁")

2. 远程调用代码准备

        远程调用是指在服务器上部署了大模型，并通过Flask、FastAPI的方式给出了访问接口。我们需要将远程调用的代码和注册模型的代码融合。

 2.1 原始访问代码

from openai import OpenAI
import requests
import json

def simple_chat(messages):
    #print(messages[0])
    response = client.chat.completions.create(
        model="glm-4",
        messages=messages,
        stream=False,
        max_tokens=256,
        temperature=0.4,
        presence_penalty=1.2,
        top_p=0.8,
    )
    if response:
        return json.loads(response.model_dump_json())['choices'][0]['message']['content']
    else:
        return ""

def glm4_function(prompt,sentence):
    result = ""
    prompt_dict = [
        {
          "role": "system",
          "content":prompt,
          },
        {
          "role": "user",
           "content": sentence
        },
      ]
    
    try:
        result = simple_chat(prompt_dict)
    except Exception as e:
        print(e)
    return result
base_url = "http://..../v1/chatglm4"
client = OpenAI(api_key="EMPTY", base_url=base_url)
system_prompt = "you are a helpful assistant "
sentence = '你好，请介绍一下你自己'
temp_response = glm4_function(system_prompt,sentence)
print(temp_response)

2.2 融合后代码

from langchain.llms.base import LLM
from typing import Any, List, Optional
from langchain.callbacks.manager import CallbackManagerForLLMRun
import requests
import json
from pydantic import BaseModel, Field

class Qwen2_LLM(LLM, BaseModel):
    api_url: str = Field(..., description="远程API的URL")

    @property
    def _llm_type(self) -> str:
        return "Qwen2_LLM_API"

    def __init__(self, **data: Any):
        super().__init__(**data)
        print(f"API URL设置为: {self.api_url}")

    def _call(self, prompt: str, stop: Optional[List[str]] = None,
              run_manager: Optional[CallbackManagerForLLMRun] = None,
              **kwargs: Any) -> str:
        """
        调用API并返回模型响应
        """
        # 设置系统消息和用户输入
        messages = [
            {"role": "system", "content": "You are a helpful assistant."},
            {"role": "user", "content": prompt}
        ]
        
        # 请求API
        try:
            response = requests.post(
                f"{self.api_url}/chatglm4",
                headers={"Authorization": "Bearer YOUR_API_KEY"},  # 替换为实际的API密钥
                json={
                    "model": "glm-4",
                    "messages": messages,
                    "stream": False,
                    "max_tokens": 512,
                    "temperature": 0.4,
                    "presence_penalty": 1.2,
                    "top_p": 0.8
                }
            )
            response.raise_for_status()  # 检查HTTP错误
            result = response.json()
            # 提取生成的文本
            return result['choices'][0]['message']['content']
        
        except requests.exceptions.HTTPError as http_err:
            print(f"HTTP错误发生: {http_err}")
        except Exception as e:
            print(f"请求API时发生错误: {e}")
        return ""  # 在出现错误时返回空字符串或其他适当的默认值

# 使用示例
if __name__ == "__main__":
    llm = Qwen2_LLM(api_url="http://.../v1")
    response = llm("你好，请介绍一下你自己")
    print(response)

Reference:

        1.​​​​​​​self-llm/models/Qwen2 at master · datawhalechina/self-llm · GitHub