From 61651c0335dd1b4ced87b16ac948473944b8801a Mon Sep 17 00:00:00 2001 From: Chen Junyang <97775238+colorfulandcjy0806@users.noreply.github.com> Date: Thu, 21 Nov 2024 00:48:54 +0800 Subject: [PATCH 1/7] Create run_agent_api_web_demo.py MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit 基于Lagent新版的内容，完成了一个利用浦语API完成的Agent。实现了CustomAPILLM 类，该类继承自 BaseAPILLM，封装了对 API 的调用逻辑，然后利用Streamlit启动Web服务，能够调用Arxiv检索工具。 --- examples/run_agent_api_web_demo.py | 406 +++++++++++++++++++++++++++++ 1 file changed, 406 insertions(+) create mode 100644 examples/run_agent_api_web_demo.py diff --git a/examples/run_agent_api_web_demo.py b/examples/run_agent_api_web_demo.py new file mode 100644 index 0000000..d4d2ef9 --- /dev/null +++ b/examples/run_agent_api_web_demo.py @@ -0,0 +1,406 @@ +import copy +import hashlib +import json +import os +from typing import Dict, List, Union +import re +import streamlit as st +import requests + +from lagent.actions import ActionExecutor, ArxivSearch, IPythonInterpreter +from lagent.prompts.parsers import PluginParser +from lagent.agents.stream import INTERPRETER_CN, META_CN, PLUGIN_CN, AgentForInternLM, get_plugin_prompt +from lagent.llms.base_api import BaseAPILLM +from lagent.schema import AgentStatusCode + +YOUR_TOKEN_HERE = "" # 请注意，这里要替换为自己实际授权令牌！！！ + +class CustomAPILLM(BaseAPILLM): + """自定义的 API LLM 类，用于调用外部 API 进行文本生成。""" + + def __init__(self, model_type, meta_template=None, **gen_params): + super().__init__(model_type, meta_template=meta_template, **gen_params) + + def call_api(self, messages): + """调用外部 API 并返回响应结果。""" + url = 'https://internlm-chat.intern-ai.org.cn/puyu/api/v1/chat/completions' + headers = { + 'Content-Type': 'application/json', + "Authorization": "Bearer " + YOUR_TOKEN_HERE + } + data = { + "model": self.model_type, + "messages": messages, + "n": 1, + "temperature": self.gen_params['temperature'], + "top_p": self.gen_params['top_p'], + "stream": False, + } + response = requests.post(url, headers=headers, json=data) + if response.status_code == 200: + return response.json() + else: + raise Exception(f"API 调用失败，状态码: {response.status_code}") + + def generate(self, inputs: Union[str, List[str]], **gen_params) -> Union[str, List[str]]: + """调用外部 API。""" + if isinstance(inputs, str): + inputs = [{"role": "user", "content": inputs}] + elif isinstance(inputs, list) and isinstance(inputs[0], str): + inputs = [{"role": "user", "content": text} for text in inputs] + + # 调用 call_api 并返回响应 + response = self.call_api(inputs) + content = response["choices"][0]["message"]["content"] + + if len(inputs) == 1: + return content + else: + return [content] + +class SessionState: + """管理会话状态的类。""" + + def init_state(self): + """初始化会话状态变量。""" + st.session_state['assistant'] = [] + st.session_state['user'] = [] + action_list = [ + ArxivSearch(), + ] + st.session_state['plugin_map'] = { + action.name: action for action in action_list + } + st.session_state['model_map'] = {} + st.session_state['model_selected'] = None + st.session_state['plugin_actions'] = set() + st.session_state['history'] = [] + + def clear_state(self): + """清除当前会话状态。""" + st.session_state['assistant'] = [] + st.session_state['user'] = [] + st.session_state['model_selected'] = None + st.session_state['file'] = set() + if 'chatbot' in st.session_state: + st.session_state['chatbot']._session_history = [] + +class StreamlitUI: + """管理 Streamlit 界面的类。""" + + def __init__(self, session_state: SessionState): + self.init_streamlit() + self.session_state = session_state + self.plugin_action = [] # 插件操作列表 + # 初始化提示词 + self.meta_prompt = META_CN + self.da_prompt = INTERPRETER_CN + self.plugin_prompt = PLUGIN_CN + + def init_streamlit(self): + """初始化 Streamlit 的 UI 设置。""" + st.set_page_config( + layout='wide', + page_title='lagent-web', + page_icon='./docs/imgs/lagent_icon.png') + st.header(':robot_face: :blue[Lagent] Web Demo ', divider='rainbow') + st.sidebar.title('模型控制') + st.session_state['file'] = set() + st.session_state['ip'] = None + + def setup_sidebar(self): + """设置侧边栏，选择模型和插件。""" + model_name = st.sidebar.text_input('模型名称：', value='internlm2.5-latest') + self.meta_prompt = st.sidebar.text_area('系统提示词', value=META_CN) + self.da_prompt = st.sidebar.text_area('数据分析提示词', value=INTERPRETER_CN) + self.plugin_prompt = st.sidebar.text_area('插件提示词', value=PLUGIN_CN) + model_ip = st.sidebar.text_input('模型IP：', value='127.0.0.1:23333') + + # 确保 model_map 已初始化 + if model_name not in st.session_state['model_map']: + st.session_state['model_map'][model_name] = self.call_api + + model = st.session_state['model_map'][model_name] + + # 添加插件选择 + plugin_name = st.sidebar.multiselect( + '插件选择', + options=list(st.session_state['plugin_map'].keys()), + default=[], + ) + da_flag = st.sidebar.checkbox('数据分析', value=False) + + # 创建插件操作列表 + self.plugin_action = [ + st.session_state['plugin_map'][name] for name in plugin_name + ] + + # 动态生成插件提示 + if self.plugin_action: + self.plugin_prompt = get_plugin_prompt(self.plugin_action) + + # 初始化或更新 chatbot + if 'chatbot' in st.session_state: + if self.plugin_action: + st.session_state['chatbot'].plugin_executor = ActionExecutor( + actions=self.plugin_action) + else: + st.session_state['chatbot'].plugin_executor = None + + if da_flag: + st.session_state['chatbot'].interpreter_executor = ActionExecutor( + actions=[IPythonInterpreter()]) + else: + st.session_state['chatbot'].interpreter_executor = None + + # 更新提示词 + st.session_state['chatbot'].meta_prompt = self.meta_prompt + st.session_state['chatbot'].plugin_prompt = self.plugin_prompt + st.session_state['chatbot'].interpreter_prompt = self.da_prompt + + # 清空对话按钮 + if st.sidebar.button('清空对话', key='clear'): + self.session_state.clear_state() + + uploaded_file = st.sidebar.file_uploader('上传文件') + + return model_name, model, self.plugin_action, uploaded_file, model_ip + + def call_api(self, prompt="你是一个机器人"): + """使用外部 API 请求生成响应（用于模型初始化）。""" + url = 'https://internlm-chat.intern-ai.org.cn/puyu/api/v1/chat/completions' + headers = { + 'Content-Type': 'application/json', + "Authorization": "Bearer " + YOUR_TOKEN_HERE + } + data = { + "model": "internlm2.5-latest", + "messages": [{"role": "assistant", "content": prompt}], + "n": 1, + "temperature": 0.8, + "top_p": 0.9, + "stream": False, + } + response = requests.post(url, headers=headers, json=data) + return response + + def initialize_chatbot(self, model_name, plugin_action): + """使用 CustomAPILLM 初始化 chatbot。""" + # meta_template 是一个包含字典的列表，并包含所有角色 + self.meta_prompt = [ + {"role": "system", "content": self.meta_prompt, "api_role": "system"}, + {"role": "user", "content": "", "api_role": "user"}, + {"role": "assistant", "content": "", "api_role": "assistant"} + ] + + # 使用 CustomAPILLM 类 + api_model = CustomAPILLM( + model_type=model_name, + meta_template=self.meta_prompt, + max_new_tokens=512, + temperature=0.8, + top_p=0.9 + ) + return api_model + + def render_user(self, prompt: str): + """渲染用户的输入。""" + with st.chat_message('user'): + st.markdown(prompt) + + def render_assistant(self, agent_return): + """渲染助手的响应，包括处理插件的结果。""" + with st.chat_message('assistant'): + if hasattr(agent_return, "content"): + content = agent_return.content + else: + content = str(agent_return) + + if isinstance(content, list): + content = '\n'.join(content) + elif not isinstance(content, str): + content = str(content) + + st.markdown(content) + + json_match = re.search(r'\{.*\}', content) + if json_match: + json_string = json_match.group() + try: + action_data = json.loads(json_string) + plugin_name = action_data.get('name') + parameters = action_data.get('parameters', {}) + + # 提取插件的基本名称 + base_plugin_name = plugin_name.split('.')[0] + + if base_plugin_name in [action.name for action in self.plugin_action]: + plugin = st.session_state['plugin_map'][base_plugin_name] + + # 根据插件类型调用不同的方法 + if base_plugin_name == "ArxivSearch": + arxiv_results = plugin.get_arxiv_article_information(parameters.get('query', '')) + # 解析和显示 Arxiv 信息 + results = arxiv_results.get('content', '').split('\n\n') + for result in results: + lines = result.split('\n') + if len(lines) >= 4: + published = lines[0].replace('Published: ', '').strip() + title = lines[1].replace('Title: ', '').strip() + authors = lines[2].replace('Authors: ', '').strip() + summary = ' '.join(lines[3:]).replace('Summary: ', '').strip() + + st.markdown(f" **标题**: {title}") + st.markdown(f" **作者**: {authors}") + st.markdown(f" **发表日期**: {published}") + st.markdown(f" **摘要**: {summary}\n") + else: + st.warning("无法解析论文信息，格式不正确。") + else: + st.warning(f"未找到插件: {base_plugin_name}") + else: + st.warning(f"未找到插件: {base_plugin_name}") + except json.JSONDecodeError: + st.error("无法解析 action 中的 JSON 数据，请检查其格式是否正确。") + + def render_plugin_args(self, action): + """渲染插件的参数。""" + action_name = action.type + args = action.args + parameter_dict = dict(name=action_name, parameters=args) + parameter_str = 'json\n' + json.dumps( + parameter_dict, indent=4, ensure_ascii=False) + '\n' + st.markdown(parameter_str) + + def render_interpreter_args(self, action): + """渲染解释器的参数。""" + st.info(action.type) + st.markdown(action.args['text']) + + def render_action(self, action): + """渲染动作，包括思考过程和结果。""" + st.markdown(action.thought) + if action.type == 'IPythonInterpreter': + self.render_interpreter_args(action) + elif action.type == 'FinishAction': + pass + else: + self.render_plugin_args(action) + self.render_action_results(action) + + def render_action_results(self, action): + if isinstance(action.result, dict): + if 'text' in action.result: + st.markdown('\n' + action.result['text'] + '\n') + if 'image' in action.result: + for image_path in action.result['image']: + image_data = open(image_path, 'rb').read() + st.image(image_data, caption='Generated Image') + if 'video' in action.result: + video_data = action.result['video'] + video_data = open(video_data, 'rb').read() + st.video(video_data) + if 'audio' in action.result: + audio_data = action.result['audio'] + audio_data = open(audio_data, 'rb').read() + st.audio(audio_data) + elif isinstance(action.result, list): + for item in action.result: + if item['type'] == 'text': + st.markdown('\n' + item['content'] + '\n') + elif item['type'] == 'image': + image_data = open(item['content'], 'rb').read() + st.image(image_data, caption='Generated Image') + elif item['type'] == 'video': + video_data = open(item['content'], 'rb').read() + st.video(video_data) + elif item['type'] == 'audio': + audio_data = open(item['content'], 'rb').read() + st.audio(audio_data) + if action.errmsg: + st.error(action.errmsg) + +def main(): + """主函数，运行 Streamlit 应用。""" + if 'ui' not in st.session_state: + session_state = SessionState() + session_state.init_state() + st.session_state['ui'] = StreamlitUI(session_state) + else: + st.set_page_config( + layout='wide', + page_title='lagent-web', + page_icon='./docs/imgs/lagent_icon.png' + ) + st.header(':robot_face: :blue[Lagent] Web Demo ', divider='rainbow') + + # 设置侧边栏并获取模型和插件 + model_name, model, plugin_action, uploaded_file, _ = st.session_state['ui'].setup_sidebar() + + # 初始化 chatbot 和 agent + if 'chatbot' not in st.session_state or model_name != st.session_state['chatbot'].model_type: + st.session_state['chatbot'] = st.session_state['ui'].initialize_chatbot(model_name, plugin_action) + plugins = [ + dict(type='lagent.actions.ArxivSearch'), + ] + + # 创建 AgentForInternLM 实例并存储在 session_state 中 + st.session_state['agent'] = AgentForInternLM( + llm=st.session_state['chatbot'], + plugins=plugins, + output_format=dict( + type=PluginParser, + template=PLUGIN_CN, + prompt=get_plugin_prompt(plugins) + ) + ) + # 清空会话历史 + st.session_state['session_history'] = [] + + if 'agent' not in st.session_state: + st.session_state['agent'] = None + + agent = st.session_state['agent'] + for prompt, agent_return in zip(st.session_state['user'], st.session_state['assistant']): + st.session_state['ui'].render_user(prompt) + st.session_state['ui'].render_assistant(agent_return) + + if user_input := st.chat_input(''): + with st.container(): + st.session_state['ui'].render_user(user_input) + res = agent(user_input, session_id=0) + st.session_state['ui'].render_assistant(res) + st.session_state['assistant'].append(copy.deepcopy(res)) + + if uploaded_file and uploaded_file.name not in st.session_state['file']: + st.session_state['file'].add(uploaded_file.name) + file_bytes = uploaded_file.read() + file_type = uploaded_file.type + if 'image' in file_type: + st.image(file_bytes, caption='Uploaded Image') + elif 'video' in file_type: + st.video(file_bytes, caption='Uploaded Video') + elif 'audio' in file_type: + st.audio(file_bytes, caption='Uploaded Audio') + postfix = uploaded_file.name.split('.')[-1] + prefix = hashlib.md5(file_bytes).hexdigest() + filename = f'{prefix}.{postfix}' + file_path = os.path.join(root_dir, filename) + with open(file_path, 'wb') as tmpfile: + tmpfile.write(file_bytes) + file_size = os.stat(file_path).st_size / 1024 / 1024 + file_size = f'{round(file_size, 2)} MB' + user_input = [ + dict(role='user', content=user_input), + dict(role='user', content=json.dumps(dict(path=file_path, size=file_size)), name='file') + ] + else: + user_input = [dict(role='user', content=user_input)] + + st.session_state['last_status'] = AgentStatusCode.END + +if __name__ == '__main__': + root_dir = os.path.dirname(os.path.dirname(os.path.abspath(__file__))) + root_dir = os.path.join(root_dir, 'tmp_dir') + os.makedirs(root_dir, exist_ok=True) + main() From dc79586cbb538290ec6695f489a15299e4343a15 Mon Sep 17 00:00:00 2001 From: Chen Junyang <97775238+colorfulandcjy0806@users.noreply.github.com> Date: Thu, 21 Nov 2024 01:02:09 +0800 Subject: [PATCH 2/7] Update base_api.py MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit 原代码中，res['content'] 和 role_prompt['content'] 被直接拼接，可能在非字符串（如列表）场景下导致运行时错误。 修改后增加类型检查和安全处理，逻辑检查将非字符串转化为字符串 --- lagent/llms/base_api.py | 18 ++++++++++++++++-- 1 file changed, 16 insertions(+), 2 deletions(-) diff --git a/lagent/llms/base_api.py b/lagent/llms/base_api.py index e0c2205..36b23c5 100644 --- a/lagent/llms/base_api.py +++ b/lagent/llms/base_api.py @@ -114,7 +114,7 @@ def _prompt2api(self, prompts: Union[List, str]) -> Tuple[str, bool]: api_role = self._role2api_role(prompt) res.append(api_role) return res - + def _role2api_role(self, role_prompt: Dict) -> Tuple[str, bool]: merged_prompt = self.roles[role_prompt['role']] if merged_prompt.get('fallback_role'): @@ -123,7 +123,21 @@ def _role2api_role(self, role_prompt: Dict) -> Tuple[str, bool]: res = role_prompt.copy() res['role'] = merged_prompt['api_role'] res['content'] = merged_prompt.get('begin', '') - res['content'] += role_prompt.get('content', '') + + # res['content'] += role_prompt.get('content', '') + + # 确保起始部分的内容为字符串类型 + if not isinstance(res['content'], str): + res['content'] = ''.join(res['content']) if isinstance(res['content'], list) else str(res['content']) + + # 获取 role_prompt 中的自定义内容，确保其为字符串类型 + content_to_add = role_prompt.get('content', '') + if not isinstance(content_to_add, str): + content_to_add = ''.join(content_to_add) if isinstance(content_to_add, list) else str(content_to_add) + + # 拼接内容：起始部分 + 自定义内容 + res['content'] += content_to_add + res['content'] += merged_prompt.get('end', '') return res From b435dde2ad1fc3fb70920452b106fb37634aa835 Mon Sep 17 00:00:00 2001 From: Chen Junyang <97775238+colorfulandcjy0806@users.noreply.github.com> Date: Thu, 21 Nov 2024 19:50:09 +0800 Subject: [PATCH 3/7] Update run_agent_api_web_demo.py --- examples/run_agent_api_web_demo.py | 322 ++++++----------------------- 1 file changed, 63 insertions(+), 259 deletions(-) diff --git a/examples/run_agent_api_web_demo.py b/examples/run_agent_api_web_demo.py index d4d2ef9..384d114 100644 --- a/examples/run_agent_api_web_demo.py +++ b/examples/run_agent_api_web_demo.py @@ -12,69 +12,27 @@ from lagent.agents.stream import INTERPRETER_CN, META_CN, PLUGIN_CN, AgentForInternLM, get_plugin_prompt from lagent.llms.base_api import BaseAPILLM from lagent.schema import AgentStatusCode +from lagent.llms import GPTAPI -YOUR_TOKEN_HERE = "" # 请注意，这里要替换为自己实际授权令牌！！！ - -class CustomAPILLM(BaseAPILLM): - """自定义的 API LLM 类，用于调用外部 API 进行文本生成。""" - - def __init__(self, model_type, meta_template=None, **gen_params): - super().__init__(model_type, meta_template=meta_template, **gen_params) - - def call_api(self, messages): - """调用外部 API 并返回响应结果。""" - url = 'https://internlm-chat.intern-ai.org.cn/puyu/api/v1/chat/completions' - headers = { - 'Content-Type': 'application/json', - "Authorization": "Bearer " + YOUR_TOKEN_HERE - } - data = { - "model": self.model_type, - "messages": messages, - "n": 1, - "temperature": self.gen_params['temperature'], - "top_p": self.gen_params['top_p'], - "stream": False, - } - response = requests.post(url, headers=headers, json=data) - if response.status_code == 200: - return response.json() - else: - raise Exception(f"API 调用失败，状态码: {response.status_code}") - - def generate(self, inputs: Union[str, List[str]], **gen_params) -> Union[str, List[str]]: - """调用外部 API。""" - if isinstance(inputs, str): - inputs = [{"role": "user", "content": inputs}] - elif isinstance(inputs, list) and isinstance(inputs[0], str): - inputs = [{"role": "user", "content": text} for text in inputs] - - # 调用 call_api 并返回响应 - response = self.call_api(inputs) - content = response["choices"][0]["message"]["content"] - - if len(inputs) == 1: - return content - else: - return [content] +# 替换为自己的授权令牌 +YOUR_TOKEN_HERE = "" class SessionState: """管理会话状态的类。""" def init_state(self): """初始化会话状态变量。""" - st.session_state['assistant'] = [] - st.session_state['user'] = [] + st.session_state['assistant'] = [] # 助手消息历史 + st.session_state['user'] = [] # 用户消息历史 + # 初始化插件列表 action_list = [ ArxivSearch(), ] - st.session_state['plugin_map'] = { - action.name: action for action in action_list - } - st.session_state['model_map'] = {} - st.session_state['model_selected'] = None - st.session_state['plugin_actions'] = set() - st.session_state['history'] = [] + st.session_state['plugin_map'] = {action.name: action for action in action_list} + st.session_state['model_map'] = {} # 存储模型实例 + st.session_state['model_selected'] = None # 当前选定模型 + st.session_state['plugin_actions'] = set() # 当前激活插件 + st.session_state['history'] = [] # 聊天历史 def clear_state(self): """清除当前会话状态。""" @@ -85,44 +43,43 @@ def clear_state(self): if 'chatbot' in st.session_state: st.session_state['chatbot']._session_history = [] + class StreamlitUI: """管理 Streamlit 界面的类。""" def __init__(self, session_state: SessionState): - self.init_streamlit() self.session_state = session_state - self.plugin_action = [] # 插件操作列表 + self.plugin_action = [] # 当前选定的插件 # 初始化提示词 self.meta_prompt = META_CN self.da_prompt = INTERPRETER_CN self.plugin_prompt = PLUGIN_CN + self.init_streamlit() def init_streamlit(self): """初始化 Streamlit 的 UI 设置。""" st.set_page_config( layout='wide', page_title='lagent-web', - page_icon='./docs/imgs/lagent_icon.png') + page_icon='./docs/imgs/lagent_icon.png' + ) st.header(':robot_face: :blue[Lagent] Web Demo ', divider='rainbow') st.sidebar.title('模型控制') - st.session_state['file'] = set() - st.session_state['ip'] = None + st.session_state['file'] = set() # 存储上传文件列表 + st.session_state['ip'] = None # 初始化模型 IP def setup_sidebar(self): """设置侧边栏，选择模型和插件。""" + # 模型名称和 IP 输入框 model_name = st.sidebar.text_input('模型名称：', value='internlm2.5-latest') + model_ip = st.sidebar.text_input('模型IP：', value='127.0.0.1:23333') + + # 提示词设置 self.meta_prompt = st.sidebar.text_area('系统提示词', value=META_CN) self.da_prompt = st.sidebar.text_area('数据分析提示词', value=INTERPRETER_CN) self.plugin_prompt = st.sidebar.text_area('插件提示词', value=PLUGIN_CN) - model_ip = st.sidebar.text_input('模型IP：', value='127.0.0.1:23333') - - # 确保 model_map 已初始化 - if model_name not in st.session_state['model_map']: - st.session_state['model_map'][model_name] = self.call_api - - model = st.session_state['model_map'][model_name] - # 添加插件选择 + # 插件选择 plugin_name = st.sidebar.multiselect( '插件选择', options=list(st.session_state['plugin_map'].keys()), @@ -130,72 +87,34 @@ def setup_sidebar(self): ) da_flag = st.sidebar.checkbox('数据分析', value=False) - # 创建插件操作列表 - self.plugin_action = [ - st.session_state['plugin_map'][name] for name in plugin_name - ] + # 根据选择的插件生成插件操作列表 + self.plugin_action = [st.session_state['plugin_map'][name] for name in plugin_name] # 动态生成插件提示 if self.plugin_action: self.plugin_prompt = get_plugin_prompt(self.plugin_action) - # 初始化或更新 chatbot - if 'chatbot' in st.session_state: - if self.plugin_action: - st.session_state['chatbot'].plugin_executor = ActionExecutor( - actions=self.plugin_action) - else: - st.session_state['chatbot'].plugin_executor = None - - if da_flag: - st.session_state['chatbot'].interpreter_executor = ActionExecutor( - actions=[IPythonInterpreter()]) - else: - st.session_state['chatbot'].interpreter_executor = None - - # 更新提示词 - st.session_state['chatbot'].meta_prompt = self.meta_prompt - st.session_state['chatbot'].plugin_prompt = self.plugin_prompt - st.session_state['chatbot'].interpreter_prompt = self.da_prompt - # 清空对话按钮 if st.sidebar.button('清空对话', key='clear'): self.session_state.clear_state() - uploaded_file = st.sidebar.file_uploader('上传文件') - - return model_name, model, self.plugin_action, uploaded_file, model_ip - - def call_api(self, prompt="你是一个机器人"): - """使用外部 API 请求生成响应（用于模型初始化）。""" - url = 'https://internlm-chat.intern-ai.org.cn/puyu/api/v1/chat/completions' - headers = { - 'Content-Type': 'application/json', - "Authorization": "Bearer " + YOUR_TOKEN_HERE - } - data = { - "model": "internlm2.5-latest", - "messages": [{"role": "assistant", "content": prompt}], - "n": 1, - "temperature": 0.8, - "top_p": 0.9, - "stream": False, - } - response = requests.post(url, headers=headers, json=data) - return response + uploaded_file = st.sidebar.file_uploader('上传文件') # 文件上传 + + return model_name, model_ip, self.plugin_action, uploaded_file def initialize_chatbot(self, model_name, plugin_action): - """使用 CustomAPILLM 初始化 chatbot。""" - # meta_template 是一个包含字典的列表，并包含所有角色 + """初始化 GPTAPI 实例作为 chatbot。""" self.meta_prompt = [ {"role": "system", "content": self.meta_prompt, "api_role": "system"}, {"role": "user", "content": "", "api_role": "user"}, - {"role": "assistant", "content": "", "api_role": "assistant"} + {"role": "assistant", "content": "", "api_role": "assistant"}, + {"role": "environment", "content": "", "api_role": "environment"} ] - # 使用 CustomAPILLM 类 - api_model = CustomAPILLM( + api_model = GPTAPI( model_type=model_name, + api_base="https://internlm-chat.intern-ai.org.cn/puyu/api/v1/chat/completions", + key=YOUR_TOKEN_HERE, meta_template=self.meta_prompt, max_new_tokens=512, temperature=0.8, @@ -204,121 +123,17 @@ def initialize_chatbot(self, model_name, plugin_action): return api_model def render_user(self, prompt: str): - """渲染用户的输入。""" + """渲染用户输入内容。""" with st.chat_message('user'): st.markdown(prompt) def render_assistant(self, agent_return): - """渲染助手的响应，包括处理插件的结果。""" + """渲染助手响应内容。""" + print("agent_return", agent_return) with st.chat_message('assistant'): - if hasattr(agent_return, "content"): - content = agent_return.content - else: - content = str(agent_return) - - if isinstance(content, list): - content = '\n'.join(content) - elif not isinstance(content, str): - content = str(content) - - st.markdown(content) - - json_match = re.search(r'\{.*\}', content) - if json_match: - json_string = json_match.group() - try: - action_data = json.loads(json_string) - plugin_name = action_data.get('name') - parameters = action_data.get('parameters', {}) - - # 提取插件的基本名称 - base_plugin_name = plugin_name.split('.')[0] - - if base_plugin_name in [action.name for action in self.plugin_action]: - plugin = st.session_state['plugin_map'][base_plugin_name] - - # 根据插件类型调用不同的方法 - if base_plugin_name == "ArxivSearch": - arxiv_results = plugin.get_arxiv_article_information(parameters.get('query', '')) - # 解析和显示 Arxiv 信息 - results = arxiv_results.get('content', '').split('\n\n') - for result in results: - lines = result.split('\n') - if len(lines) >= 4: - published = lines[0].replace('Published: ', '').strip() - title = lines[1].replace('Title: ', '').strip() - authors = lines[2].replace('Authors: ', '').strip() - summary = ' '.join(lines[3:]).replace('Summary: ', '').strip() - - st.markdown(f" **标题**: {title}") - st.markdown(f" **作者**: {authors}") - st.markdown(f" **发表日期**: {published}") - st.markdown(f" **摘要**: {summary}\n") - else: - st.warning("无法解析论文信息，格式不正确。") - else: - st.warning(f"未找到插件: {base_plugin_name}") - else: - st.warning(f"未找到插件: {base_plugin_name}") - except json.JSONDecodeError: - st.error("无法解析 action 中的 JSON 数据，请检查其格式是否正确。") - - def render_plugin_args(self, action): - """渲染插件的参数。""" - action_name = action.type - args = action.args - parameter_dict = dict(name=action_name, parameters=args) - parameter_str = 'json\n' + json.dumps( - parameter_dict, indent=4, ensure_ascii=False) + '\n' - st.markdown(parameter_str) - - def render_interpreter_args(self, action): - """渲染解释器的参数。""" - st.info(action.type) - st.markdown(action.args['text']) - - def render_action(self, action): - """渲染动作，包括思考过程和结果。""" - st.markdown(action.thought) - if action.type == 'IPythonInterpreter': - self.render_interpreter_args(action) - elif action.type == 'FinishAction': - pass - else: - self.render_plugin_args(action) - self.render_action_results(action) - - def render_action_results(self, action): - if isinstance(action.result, dict): - if 'text' in action.result: - st.markdown('\n' + action.result['text'] + '\n') - if 'image' in action.result: - for image_path in action.result['image']: - image_data = open(image_path, 'rb').read() - st.image(image_data, caption='Generated Image') - if 'video' in action.result: - video_data = action.result['video'] - video_data = open(video_data, 'rb').read() - st.video(video_data) - if 'audio' in action.result: - audio_data = action.result['audio'] - audio_data = open(audio_data, 'rb').read() - st.audio(audio_data) - elif isinstance(action.result, list): - for item in action.result: - if item['type'] == 'text': - st.markdown('\n' + item['content'] + '\n') - elif item['type'] == 'image': - image_data = open(item['content'], 'rb').read() - st.image(image_data, caption='Generated Image') - elif item['type'] == 'video': - video_data = open(item['content'], 'rb').read() - st.video(video_data) - elif item['type'] == 'audio': - audio_data = open(item['content'], 'rb').read() - st.audio(audio_data) - if action.errmsg: - st.error(action.errmsg) + content = getattr(agent_return, "content", str(agent_return)) + st.markdown(content if isinstance(content, str) else str(content)) + def main(): """主函数，运行 Streamlit 应用。""" @@ -334,27 +149,30 @@ def main(): ) st.header(':robot_face: :blue[Lagent] Web Demo ', divider='rainbow') - # 设置侧边栏并获取模型和插件 - model_name, model, plugin_action, uploaded_file, _ = st.session_state['ui'].setup_sidebar() + # 设置侧边栏并获取模型和插件信息 + model_name, model_ip, plugin_action, uploaded_file = st.session_state['ui'].setup_sidebar() + plugins = [dict(type=f"lagent.actions.{plugin.__class__.__name__}") for plugin in plugin_action] - # 初始化 chatbot 和 agent - if 'chatbot' not in st.session_state or model_name != st.session_state['chatbot'].model_type: + if ( + 'chatbot' not in st.session_state or + model_name != st.session_state['chatbot'].model_type or + 'last_plugin_action' not in st.session_state or + plugin_action != st.session_state['last_plugin_action'] + ): st.session_state['chatbot'] = st.session_state['ui'].initialize_chatbot(model_name, plugin_action) - plugins = [ - dict(type='lagent.actions.ArxivSearch'), - ] + st.session_state['last_plugin_action'] = plugin_action # 更新插件状态 - # 创建 AgentForInternLM 实例并存储在 session_state 中 + # 初始化 AgentForInternLM st.session_state['agent'] = AgentForInternLM( llm=st.session_state['chatbot'], plugins=plugins, output_format=dict( type=PluginParser, template=PLUGIN_CN, - prompt=get_plugin_prompt(plugins) + prompt=get_plugin_prompt(plugin_action) ) ) - # 清空会话历史 + # 清空对话历史 st.session_state['session_history'] = [] if 'agent' not in st.session_state: @@ -365,42 +183,28 @@ def main(): st.session_state['ui'].render_user(prompt) st.session_state['ui'].render_assistant(agent_return) + # 处理用户输入 if user_input := st.chat_input(''): - with st.container(): - st.session_state['ui'].render_user(user_input) + st.session_state['ui'].render_user(user_input) res = agent(user_input, session_id=0) st.session_state['ui'].render_assistant(res) + + # 更新会话状态 + st.session_state['user'].append(user_input) st.session_state['assistant'].append(copy.deepcopy(res)) + # 处理文件上传 if uploaded_file and uploaded_file.name not in st.session_state['file']: st.session_state['file'].add(uploaded_file.name) file_bytes = uploaded_file.read() - file_type = uploaded_file.type - if 'image' in file_type: - st.image(file_bytes, caption='Uploaded Image') - elif 'video' in file_type: - st.video(file_bytes, caption='Uploaded Video') - elif 'audio' in file_type: - st.audio(file_bytes, caption='Uploaded Audio') - postfix = uploaded_file.name.split('.')[-1] - prefix = hashlib.md5(file_bytes).hexdigest() - filename = f'{prefix}.{postfix}' - file_path = os.path.join(root_dir, filename) + file_path = os.path.join("tmp_dir", hashlib.md5(file_bytes).hexdigest()) with open(file_path, 'wb') as tmpfile: tmpfile.write(file_bytes) - file_size = os.stat(file_path).st_size / 1024 / 1024 - file_size = f'{round(file_size, 2)} MB' - user_input = [ - dict(role='user', content=user_input), - dict(role='user', content=json.dumps(dict(path=file_path, size=file_size)), name='file') - ] - else: - user_input = [dict(role='user', content=user_input)] + st.markdown(f"文件已上传：{uploaded_file.name}") st.session_state['last_status'] = AgentStatusCode.END + if __name__ == '__main__': - root_dir = os.path.dirname(os.path.dirname(os.path.abspath(__file__))) - root_dir = os.path.join(root_dir, 'tmp_dir') - os.makedirs(root_dir, exist_ok=True) + os.makedirs("tmp_dir", exist_ok=True) main() From 7f2c1b2177faaa81f706495c883b74f174f4b3c1 Mon Sep 17 00:00:00 2001 From: Chen Junyang <97775238+colorfulandcjy0806@users.noreply.github.com> Date: Thu, 21 Nov 2024 19:52:49 +0800 Subject: [PATCH 4/7] Create puyu.py --- lagent/llms/puyu.py | 46 +++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 46 insertions(+) create mode 100644 lagent/llms/puyu.py diff --git a/lagent/llms/puyu.py b/lagent/llms/puyu.py new file mode 100644 index 0000000..87bd012 --- /dev/null +++ b/lagent/llms/puyu.py @@ -0,0 +1,46 @@ +from typing import Dict, List, Union +from lagent.llms.base_api import BaseAPILLM +import requests + +class puyuAPI(BaseAPILLM): + """自定义的 API LLM 类，用于调用外部 API 进行文本生成。""" + + def __init__(self, model_type, meta_template=None, **gen_params): + super().__init__(model_type, meta_template=meta_template, **gen_params) + + def call_api(self, messages): + """调用外部 API 并返回响应结果。""" + url = 'https://internlm-chat.intern-ai.org.cn/puyu/api/v1/chat/completions' + headers = { + 'Content-Type': 'application/json', + "Authorization": "Bearer " + YOUR_TOKEN_HERE + } + data = { + "model": self.model_type, + "messages": messages, + "n": 1, + "temperature": self.gen_params['temperature'], + "top_p": self.gen_params['top_p'], + "stream": False, + } + response = requests.post(url, headers=headers, json=data) + if response.status_code == 200: + return response.json() + else: + raise Exception(f"API 调用失败，状态码: {response.status_code}") + + def generate(self, inputs: Union[str, List[str]], **gen_params) -> Union[str, List[str]]: + """调用外部 API。""" + if isinstance(inputs, str): + inputs = [{"role": "user", "content": inputs}] + elif isinstance(inputs, list) and isinstance(inputs[0], str): + inputs = [{"role": "user", "content": text} for text in inputs] + + # 调用 call_api 并返回响应 + response = self.call_api(inputs) + content = response["choices"][0]["message"]["content"] + + if len(inputs) == 1: + return content + else: + return [content] From a5f0fa5ab087d24b03e7c74a078b9217c916ef6b Mon Sep 17 00:00:00 2001 From: Chen Junyang <97775238+colorfulandcjy0806@users.noreply.github.com> Date: Thu, 21 Nov 2024 19:56:52 +0800 Subject: [PATCH 5/7] Update base_api.py --- lagent/llms/base_api.py | 18 ++---------------- 1 file changed, 2 insertions(+), 16 deletions(-) diff --git a/lagent/llms/base_api.py b/lagent/llms/base_api.py index 36b23c5..e0c2205 100644 --- a/lagent/llms/base_api.py +++ b/lagent/llms/base_api.py @@ -114,7 +114,7 @@ def _prompt2api(self, prompts: Union[List, str]) -> Tuple[str, bool]: api_role = self._role2api_role(prompt) res.append(api_role) return res - + def _role2api_role(self, role_prompt: Dict) -> Tuple[str, bool]: merged_prompt = self.roles[role_prompt['role']] if merged_prompt.get('fallback_role'): @@ -123,21 +123,7 @@ def _role2api_role(self, role_prompt: Dict) -> Tuple[str, bool]: res = role_prompt.copy() res['role'] = merged_prompt['api_role'] res['content'] = merged_prompt.get('begin', '') - - # res['content'] += role_prompt.get('content', '') - - # 确保起始部分的内容为字符串类型 - if not isinstance(res['content'], str): - res['content'] = ''.join(res['content']) if isinstance(res['content'], list) else str(res['content']) - - # 获取 role_prompt 中的自定义内容，确保其为字符串类型 - content_to_add = role_prompt.get('content', '') - if not isinstance(content_to_add, str): - content_to_add = ''.join(content_to_add) if isinstance(content_to_add, list) else str(content_to_add) - - # 拼接内容：起始部分 + 自定义内容 - res['content'] += content_to_add - + res['content'] += role_prompt.get('content', '') res['content'] += merged_prompt.get('end', '') return res From 8dcefd1ee825eefa1c7fa49fe8db3f7a9ead4eee Mon Sep 17 00:00:00 2001 From: Chen Junyang <97775238+colorfulandcjy0806@users.noreply.github.com> Date: Thu, 21 Nov 2024 20:02:40 +0800 Subject: [PATCH 6/7] Delete lagent/llms/puyu.py --- lagent/llms/puyu.py | 46 --------------------------------------------- 1 file changed, 46 deletions(-) delete mode 100644 lagent/llms/puyu.py diff --git a/lagent/llms/puyu.py b/lagent/llms/puyu.py deleted file mode 100644 index 87bd012..0000000 --- a/lagent/llms/puyu.py +++ /dev/null @@ -1,46 +0,0 @@ -from typing import Dict, List, Union -from lagent.llms.base_api import BaseAPILLM -import requests - -class puyuAPI(BaseAPILLM): - """自定义的 API LLM 类，用于调用外部 API 进行文本生成。""" - - def __init__(self, model_type, meta_template=None, **gen_params): - super().__init__(model_type, meta_template=meta_template, **gen_params) - - def call_api(self, messages): - """调用外部 API 并返回响应结果。""" - url = 'https://internlm-chat.intern-ai.org.cn/puyu/api/v1/chat/completions' - headers = { - 'Content-Type': 'application/json', - "Authorization": "Bearer " + YOUR_TOKEN_HERE - } - data = { - "model": self.model_type, - "messages": messages, - "n": 1, - "temperature": self.gen_params['temperature'], - "top_p": self.gen_params['top_p'], - "stream": False, - } - response = requests.post(url, headers=headers, json=data) - if response.status_code == 200: - return response.json() - else: - raise Exception(f"API 调用失败，状态码: {response.status_code}") - - def generate(self, inputs: Union[str, List[str]], **gen_params) -> Union[str, List[str]]: - """调用外部 API。""" - if isinstance(inputs, str): - inputs = [{"role": "user", "content": inputs}] - elif isinstance(inputs, list) and isinstance(inputs[0], str): - inputs = [{"role": "user", "content": text} for text in inputs] - - # 调用 call_api 并返回响应 - response = self.call_api(inputs) - content = response["choices"][0]["message"]["content"] - - if len(inputs) == 1: - return content - else: - return [content] From 6d663009491aca72576b25642f8bf65c0e0f176c Mon Sep 17 00:00:00 2001 From: Chen Junyang <97775238+colorfulandcjy0806@users.noreply.github.com> Date: Fri, 22 Nov 2024 17:27:33 +0800 Subject: [PATCH 7/7] Update run_agent_api_web_demo.py --- examples/run_agent_api_web_demo.py | 97 ++++++++++++------------------ 1 file changed, 38 insertions(+), 59 deletions(-) diff --git a/examples/run_agent_api_web_demo.py b/examples/run_agent_api_web_demo.py index 384d114..3d0611d 100644 --- a/examples/run_agent_api_web_demo.py +++ b/examples/run_agent_api_web_demo.py @@ -1,22 +1,12 @@ import copy -import hashlib -import json import os -from typing import Dict, List, Union -import re +from typing import List import streamlit as st -import requests - -from lagent.actions import ActionExecutor, ArxivSearch, IPythonInterpreter +from lagent.actions import ArxivSearch, WeatherQuery from lagent.prompts.parsers import PluginParser from lagent.agents.stream import INTERPRETER_CN, META_CN, PLUGIN_CN, AgentForInternLM, get_plugin_prompt -from lagent.llms.base_api import BaseAPILLM -from lagent.schema import AgentStatusCode from lagent.llms import GPTAPI -# 替换为自己的授权令牌 -YOUR_TOKEN_HERE = "" - class SessionState: """管理会话状态的类。""" @@ -27,21 +17,20 @@ def init_state(self): # 初始化插件列表 action_list = [ ArxivSearch(), + WeatherQuery(), ] st.session_state['plugin_map'] = {action.name: action for action in action_list} st.session_state['model_map'] = {} # 存储模型实例 st.session_state['model_selected'] = None # 当前选定模型 st.session_state['plugin_actions'] = set() # 当前激活插件 st.session_state['history'] = [] # 聊天历史 + st.session_state['api_base'] = None # 初始化API base地址 def clear_state(self): """清除当前会话状态。""" st.session_state['assistant'] = [] st.session_state['user'] = [] st.session_state['model_selected'] = None - st.session_state['file'] = set() - if 'chatbot' in st.session_state: - st.session_state['chatbot']._session_history = [] class StreamlitUI: @@ -52,7 +41,6 @@ def __init__(self, session_state: SessionState): self.plugin_action = [] # 当前选定的插件 # 初始化提示词 self.meta_prompt = META_CN - self.da_prompt = INTERPRETER_CN self.plugin_prompt = PLUGIN_CN self.init_streamlit() @@ -64,58 +52,58 @@ def init_streamlit(self): page_icon='./docs/imgs/lagent_icon.png' ) st.header(':robot_face: :blue[Lagent] Web Demo ', divider='rainbow') - st.sidebar.title('模型控制') - st.session_state['file'] = set() # 存储上传文件列表 - st.session_state['ip'] = None # 初始化模型 IP def setup_sidebar(self): """设置侧边栏，选择模型和插件。""" - # 模型名称和 IP 输入框 + # 模型名称和 API Base 输入框 model_name = st.sidebar.text_input('模型名称：', value='internlm2.5-latest') - model_ip = st.sidebar.text_input('模型IP：', value='127.0.0.1:23333') - - # 提示词设置 - self.meta_prompt = st.sidebar.text_area('系统提示词', value=META_CN) - self.da_prompt = st.sidebar.text_area('数据分析提示词', value=INTERPRETER_CN) - self.plugin_prompt = st.sidebar.text_area('插件提示词', value=PLUGIN_CN) - - # 插件选择 + # 注意，如果采用硅基流动API，模型名称需要更改为：internlm/internlm2_5-7b-chat 或者 internlm/internlm2_5-20b-chat + # ================================== 硅基流动的API ================================== + # api_base = st.sidebar.text_input( + # 'API Base 地址：', value='https://api.siliconflow.cn/v1/chat/completions' + # ) + # ================================== 浦语官方的API ================================== + api_base = st.sidebar.text_input( + 'API Base 地址：', value='https://internlm-chat.intern-ai.org.cn/puyu/api/v1/chat/completions' + ) + # ================================================================================== + plugin_name = st.sidebar.multiselect( '插件选择', options=list(st.session_state['plugin_map'].keys()), default=[], ) - da_flag = st.sidebar.checkbox('数据分析', value=False) # 根据选择的插件生成插件操作列表 self.plugin_action = [st.session_state['plugin_map'][name] for name in plugin_name] - # 动态生成插件提示 if self.plugin_action: self.plugin_prompt = get_plugin_prompt(self.plugin_action) - # 清空对话按钮 if st.sidebar.button('清空对话', key='clear'): self.session_state.clear_state() - uploaded_file = st.sidebar.file_uploader('上传文件') # 文件上传 + return model_name, api_base, self.plugin_action - return model_name, model_ip, self.plugin_action, uploaded_file - - def initialize_chatbot(self, model_name, plugin_action): + def initialize_chatbot(self, model_name, api_base, plugin_action): """初始化 GPTAPI 实例作为 chatbot。""" - self.meta_prompt = [ + token = os.getenv("token") + if not token: + st.error("糟糕，未检测到环境变量 `token`，请设置环境变量，例如 `export token='your_token_here'` 后重新运行 X﹏X") + st.stop() # 停止运行应用 + + meta_prompt = [ {"role": "system", "content": self.meta_prompt, "api_role": "system"}, {"role": "user", "content": "", "api_role": "user"}, - {"role": "assistant", "content": "", "api_role": "assistant"}, + {"role": "assistant", "content": self.plugin_prompt, "api_role": "assistant"}, {"role": "environment", "content": "", "api_role": "environment"} ] api_model = GPTAPI( model_type=model_name, - api_base="https://internlm-chat.intern-ai.org.cn/puyu/api/v1/chat/completions", - key=YOUR_TOKEN_HERE, - meta_template=self.meta_prompt, + api_base=api_base, + key=token, + meta_template=meta_prompt, max_new_tokens=512, temperature=0.8, top_p=0.9 @@ -129,7 +117,6 @@ def render_user(self, prompt: str): def render_assistant(self, agent_return): """渲染助手响应内容。""" - print("agent_return", agent_return) with st.chat_message('assistant'): content = getattr(agent_return, "content", str(agent_return)) st.markdown(content if isinstance(content, str) else str(content)) @@ -149,18 +136,20 @@ def main(): ) st.header(':robot_face: :blue[Lagent] Web Demo ', divider='rainbow') - # 设置侧边栏并获取模型和插件信息 - model_name, model_ip, plugin_action, uploaded_file = st.session_state['ui'].setup_sidebar() + model_name, api_base, plugin_action = st.session_state['ui'].setup_sidebar() plugins = [dict(type=f"lagent.actions.{plugin.__class__.__name__}") for plugin in plugin_action] if ( 'chatbot' not in st.session_state or model_name != st.session_state['chatbot'].model_type or 'last_plugin_action' not in st.session_state or - plugin_action != st.session_state['last_plugin_action'] + plugin_action != st.session_state['last_plugin_action'] or + api_base != st.session_state['api_base'] ): - st.session_state['chatbot'] = st.session_state['ui'].initialize_chatbot(model_name, plugin_action) - st.session_state['last_plugin_action'] = plugin_action # 更新插件状态 + # 更新 Chatbot + st.session_state['chatbot'] = st.session_state['ui'].initialize_chatbot(model_name, api_base, plugin_action) + st.session_state['last_plugin_action'] = plugin_action + st.session_state['api_base'] = api_base # 初始化 AgentForInternLM st.session_state['agent'] = AgentForInternLM( @@ -172,7 +161,7 @@ def main(): prompt=get_plugin_prompt(plugin_action) ) ) - # 清空对话历史 + st.session_state['session_history'] = [] if 'agent' not in st.session_state: @@ -186,25 +175,15 @@ def main(): # 处理用户输入 if user_input := st.chat_input(''): st.session_state['ui'].render_user(user_input) + res = agent(user_input, session_id=0) st.session_state['ui'].render_assistant(res) - # 更新会话状态 st.session_state['user'].append(user_input) st.session_state['assistant'].append(copy.deepcopy(res)) - # 处理文件上传 - if uploaded_file and uploaded_file.name not in st.session_state['file']: - st.session_state['file'].add(uploaded_file.name) - file_bytes = uploaded_file.read() - file_path = os.path.join("tmp_dir", hashlib.md5(file_bytes).hexdigest()) - with open(file_path, 'wb') as tmpfile: - tmpfile.write(file_bytes) - st.markdown(f"文件已上传：{uploaded_file.name}") - - st.session_state['last_status'] = AgentStatusCode.END + st.session_state['last_status'] = None if __name__ == '__main__': - os.makedirs("tmp_dir", exist_ok=True) main()