From 36529085b3224e6304e5383c4612158742db795c Mon Sep 17 00:00:00 2001 From: VSP Date: Sun, 19 May 2024 19:18:44 +0800 Subject: [PATCH] modified code, add another file for gradio non-stream --- testGenius.py | 126 +++++++++++++++++++++++++++++--------------------- testGradio.py | 59 +++++++++++++++++++++++ 2 files changed, 133 insertions(+), 52 deletions(-) create mode 100644 testGradio.py diff --git a/testGenius.py b/testGenius.py index 33b14eb..d11c020 100644 --- a/testGenius.py +++ b/testGenius.py @@ -8,48 +8,54 @@ import gradio as gr dashscope.api_key = 'sk-83b8ed0ead0849ae9e63a2ae5bdbde0d' # Rayman's API key -def respond(prompt, chat_history, instruction, model, if_stream): - """ - 与AI助手进行对话，并返回对话历史。 - - 参数: - - prompt: 用户输入的文本。 - - chat_history: 之前的聊天历史，列表，每个元素是二元组，包括用户输入和AI响应。 - - instruction: 系统指令，作为对话的起始信息。 - - 返回值: - - 生成器，每次产生一个二元组，包括空字符串和更新后的聊天历史。 - """ +def respond_nonStream(prompt, chat_history, instruction, model): # 构建对话消息结构 - messages = [{'role': 'system', - 'content': instruction}, - {'role': 'user', - 'content': prompt} - ] - full_response = "" # 初始化空字符串以聚合响应 + messages = [{'role': 'system', 'content': instruction}, + {'role': 'user', 'content': prompt}] + print(f"Messages: {messages}") + + # 初始化空字符串以聚合响应 + full_response = "" + # 调用AI模型生成响应 - # -------非流式输出------- - if if_stream == 'Non-Stream': + try: response = Generation.call(model=model, messages=messages, - # 设置随机数种子seed，如果没有设置，则随机数种子默认为1234 seed=1234, - # 将输出设置为"message"格式 result_format='message', - # 设置输出方式为非流式输出 stream=False, - # 设置输出方式为非增量式输出 incremental_output=False) + print(f"Response: {response}") + if not chat_history or chat_history[-1][0] != prompt: - chat_history.append((prompt, "")) + chat_history.append([prompt, ""]) + print(f"old chat history: {chat_history}") if response.status_code == HTTPStatus.OK: # 获取响应中的消息内容 message = response.output.choices[0]['message']['content'] - # 将消息内容添加到聊天历史中 - chat_history.append((prompt, message)) - # 返回更新后的聊天历史 + print(f"Generated message: {message}") + + # 更新聊天历史记录中的最后一条记录 + chat_history[-1] = [prompt, message] + print(f"Updated chat_history: {chat_history}") return "", chat_history - elif if_stream == 'Stream': + else: + print(f"Error: Received response status {response.status_code}") + return "Error: Could not generate response", chat_history + + except Exception as e: + print(f"Exception occurred: {e}") + return f"Exception occurred: {e}", chat_history + + +def respond(prompt, chat_history, instruction, model, if_stream='Stream'): + if if_stream == 'Stream': + messages = [{'role': 'system', + 'content': instruction}, + {'role': 'user', + 'content': prompt} + ] + full_response = "" # 初始化空字符串以聚合响应 # -------流式输出------- responses = Generation.call(model=model, messages=messages, @@ -88,31 +94,45 @@ def respond(prompt, chat_history, instruction, model, if_stream): yield "", chat_history break # 出现错误时终止循环 + elif if_stream == 'Non-Stream': + # 构建对话消息结构 + messages = [{'role': 'system', 'content': instruction}, + {'role': 'user', 'content': prompt}] + print(f"Messages: {messages}") -def respond_nonStream(prompt, chat_history, instruction, model): - # 构建对话消息结构 - messages = [{'role': 'system', - 'content': instruction}, - {'role': 'user', - 'content': prompt} - ] - full_response = "" # 初始化空字符串以聚合响应 - # 调用AI模型生成响应 - responses = Generation.call(model=model, - messages=messages, - # 设置随机数种子seed，如果没有设置，则随机数种子默认为1234 - seed=1234, - # 将输出设置为"message"格式 - result_format='message', - stream=True, # 设置输出方式为流式输出 - incremental_output=True, # 增量式流式输出 - temperature=1.8, - top_p=0.9, - top_k=999) + # 调用AI模型生成响应 + try: + response = Generation.call(model=model, + messages=messages, + seed=1234, + result_format='message', + stream=False, + incremental_output=False) + print(f"Response: {response}") + + if not chat_history or chat_history[-1][0] != prompt: + chat_history.append([prompt, ""]) + print(f"old chat history: {chat_history}") + if response.status_code == HTTPStatus.OK: + # 获取响应中的消息内容 + message = response.output.choices[0]['message']['content'] + print(f"Generated message: {message}") + + # 更新聊天历史记录中的最后一条记录 + chat_history[-1] = [prompt, message] + print(f"Updated chat_history: {chat_history}") + return "", chat_history + else: + print(f"Error: Received response status {response.status_code}") + return "Error: Could not generate response", chat_history + + except Exception as e: + print(f"Exception occurred: {e}") + return f"Exception occurred: {e}", chat_history -llm_model_list = ['qwen-turbo','qwen-plus', 'qwen-max'] +llm_model_list = ['qwen-turbo', 'qwen-plus', 'qwen-max'] init_llm = llm_model_list[0] # 创建 Gradio 界面 @@ -122,17 +142,20 @@ with gr.Blocks() as demo: # AI TestGenius A simple LLM app for generating test cases from function design. """) - chatbot = gr.Chatbot() + history = [["Hello", "Hello, how can I help you?"]] + chatbot = gr.Chatbot(history) msg = gr.Textbox(label="Prompt") with gr.Accordion(label="Advanced options", open=False): system = gr.Textbox(label="System prompts", lines=2, value="A conversation between a user and an LLM-based AI assistant. The assistant gives helpful and honest answers.") + # 选择模型 llm = gr.Dropdown( llm_model_list, label='Choose LLM Model', value=init_llm, interactive=True ) + # 选择是否流式输出 if_stream = gr.Dropdown( ["Stream", "Non-Stream"], label='Choose Streaming', @@ -145,7 +168,6 @@ with gr.Blocks() as demo: btn.click(respond, inputs=[msg, chatbot, system, llm, if_stream], outputs=[msg, chatbot]) # click to submit msg.submit(respond, inputs=[msg, chatbot, system, llm, if_stream], outputs=[msg, chatbot]) # Press enter to submit - # 运行界面 if __name__ == "__main__": gr.close_all() diff --git a/testGradio.py b/testGradio.py new file mode 100644 index 0000000..ad18e4b --- /dev/null +++ b/testGradio.py @@ -0,0 +1,59 @@ +import gradio as gr +import random +import time +from http import HTTPStatus +import dashscope +from dashscope import Generation + +dashscope.api_key = 'sk-83b8ed0ead0849ae9e63a2ae5bdbde0d' # Rayman's API key + +with gr.Blocks() as demo: + history = [["Hello","Hello, how can I help you?"]] + chatbot = gr.Chatbot(history) + msg = gr.Textbox() + clear = gr.ClearButton([msg, chatbot]) + + + def respond_nonStream(prompt, chat_history): + # 构建对话消息结构 + messages = [{'role': 'system', 'content': "A conversation between a user and an LLM-based AI assistant. The assistant gives helpful and honest answers."}, + {'role': 'user', 'content': prompt}] + print(f"Messages: {messages}") + + # 初始化空字符串以聚合响应 + full_response = "" + + # 调用AI模型生成响应 + try: + response = Generation.call(model='qwen-turbo', + messages=messages, + seed=1234, + result_format='message', + stream=False, + incremental_output=False) + print(f"Response: {response}") + + if not chat_history or chat_history[-1][0] != prompt: + chat_history.append([prompt, ""]) + print(f"old chat history: {chat_history}") + if response.status_code == HTTPStatus.OK: + # 获取响应中的消息内容 + message = response.output.choices[0]['message']['content'] + print(f"Generated message: {message}") + + # 更新聊天历史记录中的最后一条记录 + chat_history[-1] = [prompt, message] + print(f"Updated chat_history: {chat_history}") + return "", chat_history + else: + print(f"Error: Received response status {response.status_code}") + return "Error: Could not generate response", chat_history + + except Exception as e: + print(f"Exception occurred: {e}") + return f"Exception occurred: {e}", chat_history + + msg.submit(respond_nonStream, inputs=[msg, chatbot], outputs=[msg, chatbot]) + +if __name__ == "__main__": + demo.launch() \ No newline at end of file