简单的聊天机器人助手

聊天机器人

初学AI大模型，用聊天机器人是了一下大模型调用

环境准备

首先在本机安装ollama，这个很简单，到官网下载ollama windows安装包安装即可，然后需要用ollama 下载大模型，ollama支持很多大模型。
官网：https://ollama.com/search

然后可以编辑代码了代码如下：
这里用的依赖需要下载一下

pip install openai gradio

import requests from openai import OpenAI import gradio as gr import logging # 配置日志 logging.basicConfig(level=logging.INFO) logger = logging.getLogger(__name__) # 初始化 OpenAI 客户端 def get_openai_client() -> OpenAI: """创建并返回 OpenAI 客户端实例""" return OpenAI( api_key="ollama", base_url="http://localhost:11434/v1", timeout=30 ) # 获取本地模型列表 def get_local_models(): try: response = requests.get( "http://localhost:11434/api/tags", timeout=30) if response.status_code == 200: response_data = response.json() models = [model['name'] for model in response_data.get("models", [])] logger.info(f"获取到 {len(models)} 个模型") return models else: return [] except Exception as e: logger.error(f"获取模型列表失败: {e}") return [] def is_chat_supported(model: str, client: OpenAI) -> bool: """测试模型是否支持聊天功能""" try: client.chat.completions.create( model=model, messages=[ {"role": "system", "content": "你是人工智能助手。"}, {"role": "user", "content": "Hello"} ], stream=False, max_tokens=10 ) return True except Exception as e: logger.warning(f"模型 {model} 不支持聊天: {e}") return False # 过滤支持聊天的模型 def filter_chat_models(models): client = get_openai_client() chat_models = [model for model in models if is_chat_supported(model, client)] logger.info(f"支持聊天的模型: {chat_models}") return chat_models # 实现流式回答 def get_response(text, model_name, history): msgs = [{"role": "system", "content": "你是人工智能助手。"}] for msg in history: print(msg['content'][0]['text']) print(msg['role']) msgs.append({"role": msg['role'], "content": msg['content'][0]['text']}) msgs.append({"role":"user","content": f"{text}"}) print(msgs) client = get_openai_client() response = client.chat.completions.create( model=model_name, messages=msgs, # messages=[ # {"role": "system", "content": "你是人工智能助手。"}, # { # "role": "user", # "content": f"{text} " + "以下是历史对话" + f"{history}" # } # ], stream=True, ) return response # 聊天响应 def chat_response(message, history, model_name): print(message, history) results = get_response(message, model_name, history) contents = "" for chunk in results: if chunk.choices[0].delta.content is not None: contents += chunk.choices[0].delta.content yield contents def main(): """主函数""" print("=" * 50) print("正在初始化 Ollama 聊天应用...") print("=" * 50) chat_models = get_local_models() if not chat_models: print("未找到本地模型，请确保 Ollama 已启动并且已下载模型") return print(chat_models) models = filter_chat_models(chat_models) if not models: print("未找到支持聊天的模型") return print(models) default_model = models[0] print(f"默认使用模型: {default_model}") print(f"可用模型: {models}") print("=" * 50) #创建 Gradio 自定义聊天界面并应用自定义CSS with gr.Blocks(title="Ollama Chat") as demo: gr.Markdown("# Ollama 本地聊天助手") chat_interface = gr.ChatInterface(chat_response, additional_inputs=[ gr.Dropdown( label="选择模型", choices=models, value=default_model, allow_custom_value=True ) ],cache_examples=False) demo.launch(share=False, server_name="127.0.0.1", server_port=7860) if __name__ == "__main__": main()

效果如下：