from gradio_client import Client import sys import os from dotenv import load_dotenv # 在文件顶部添加 load_dotenv() # 加载环境变量 # 获取 Hugging Face 令牌 hf_token = os.getenv("HF_TOKEN") if not hf_token: print("错误:未找到 HF_TOKEN 环境变量。请设置您的 Hugging Face 令牌。") sys.exit(1) # 修改Client初始化,添加认证令牌 client = Client("MaoShen/Moonshot_DeepResearch", hf_token=hf_token) # First log the user message log_result = client.predict( text_input=""" there is an Startup idea -"Using the reasoning llm to help the startup evaluate their ideas' quality and noveltyn, finally help them imprve their ideas" Do you think this startup idea is of great novelty? research this idea from three perspectives and generate a novelty score: 1.Problem Uniqueness: Does this idea address an unmet or unrecognized need? 2.Existing Solution: Including competitors (the most important), patent and intellectual property research, and academic research, 3. Differentiation: Conduct research from technical innovation, business model innovation, market segment, and user experience. Give your final answer as detailed as professinal as possible in the following format: A Novelty Score on a scale of 100. A report over 5000 words including sections of Overview, Problem Uniqueness, Existing Solution, Differentiation, Conclusion, and Sources & References (show all youe sources); You should have intext Citation in the final answer with number and hyperlink, the hyperlink is a must """, api_name="/log_user_message" ) # Then interact with agent using the logged message result = client.predict( messages=[{ "role": "user", "content": log_result, "metadata": { "id": "1", "parent_id": "0" } }], api_name="/interact_with_agent_1" ) # 在获取result之后添加以下代码 import json result_file = r"D:\007.Projects\008.deep_research\smolagents\examples\open_deep_research\last_result.json" with open(result_file, 'w', encoding='utf-8') as f: json.dump(result, f, ensure_ascii=False, indent=2) print(f"API响应结果已保存至:{result_file}")