Update app.py

app.py

@@ -1,60 +1,32 @@
-import gradio as gr
-from huggingface_hub import hf_hub_download
 import os
 import subprocess
 
 # 模型下载链接
 model_url = "https://huggingface.co/CMLL/ZhongJing-2-1_8b-GGUF/resolve/main/ZhongJing1_5-1_8b-fp16.gguf"
 
-# 下载模型
-def download_model(url, model_dir="models"):
-    os.makedirs(model_dir, exist_ok=True)
-    model_path = hf_hub_download(repo_id="CMLL/ZhongJing-2-1_8b-GGUF", filename="ZhongJing1_5-1_8b-fp16.gguf", local_dir=model_dir)
-    return model_path
-
-model_path = download_model(model_url)
-
-# llama.cpp 克隆并编译
-if not os.path.exists("llama.cpp"):
-    os.system("git clone https://github.com/ggerganov/llama.cpp.git")
-    os.system("cd llama.cpp && mkdir build && cd build && cmake .. && make")
-
-# 创建 prompts/TcmChat.txt 文件
-prompts_dir = "llama.cpp/prompts"
-os.makedirs(prompts_dir, exist_ok=True)
-with open(os.path.join(prompts_dir, "TcmChat.txt"), "w") as f:
-    f.write("You are a helpful TCM medical assistant named 仲景中医大语言模型.\n")
-
-# Gradio 接口
-def chat_with_model(user_input, history):
-    prompt = f"User: {user_input}\nAssistant:"
-    with open(os.path.join(prompts_dir, "TcmChat.txt"), "a") as f:
-        f.write(prompt + "\n")
-    
-    # 执行命令并捕获输出
-    command = f"./llama.cpp/build/bin/main -m {model_path} -n 256 --repeat_penalty 1.0 --color -i -r \"User:\" -f {os.path.join(prompts_dir, 'TcmChat.txt')}"
-    result = subprocess.run(command, shell=True, capture_output=True, text=True)
-    
-    response = result.stdout.split("User:")[-1].strip()
-    
-    history.append((user_input, response))
-    return history, history
-
-with gr.Blocks() as demo:
-    chatbot = gr.Chatbot()
-    state = gr.State([])
-
-    with gr.Row():
-        with gr.Column():
-            user_input = gr.Textbox(show_label=False, placeholder="Enter your message...")
-        with gr.Column():
-            submit_btn = gr.Button("Submit")
-
-    submit_btn.click(chat_with_model, [user_input, state], [chatbot, state])
+def install_packages():
+    subprocess.run(['apt-get', '-y', 'install', '-qq', 'aria2'], check=True)
+    subprocess.run(['git', 'clone', '-b', 'V20230828', 'https://github.com/Troyanovsky/text-generation-webui'], check=True)
+    os.chdir('/text-generation-webui')
+    subprocess.run(['pip', 'install', '-r', 'requirements.txt'], check=True)
+    subprocess.run(['pip', 'install', '-U', 'gradio==3.33.1'], check=True)
+    subprocess.run(['pip', 'uninstall', '-y', 'llama-cpp-python'], check=True)
+    os.environ['CMAKE_ARGS'] = "-DLLAMA_CUBLAS=on"
+    os.environ['FORCE_CMAKE'] = "1"
+    subprocess.run(['pip', 'install', 'llama-cpp-python', '--no-cache-dir'], check=True)
 
-if __name__ == "__main__":
-    demo.launch()
+def download_model(model_url, model_name):
+    subprocess.run(['aria2c', '--console-log-level=error', '-c', '-x', '16', '-s', '16', '-k', '1M', model_url, '-d', '/text-generation-webui/models/', '-o', model_name], check=True)
+
+def run_server(model_name):
+    os.chdir('/text-generation-webui')
+    subprocess.run(['python', 'server.py', '--share', '--n-gpu-layers', '1000000000', '--model', model_name], check=True)
 
+if __name__ == "__main__":
+    install_packages()
+    model_name = 'ZhongJing1_5-1_8b-fp16.gguf'
+    download_model(model_url, model_name)
+    run_server(model_name)