[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-shyamsaktawat--OpenAlpha_Evolve":3,"tool-shyamsaktawat--OpenAlpha_Evolve":64},[4,17,27,35,43,56],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":16},3808,"stable-diffusion-webui","AUTOMATIC1111\u002Fstable-diffusion-webui","stable-diffusion-webui 是一个基于 Gradio 构建的网页版操作界面，旨在让用户能够轻松地在本地运行和使用强大的 Stable Diffusion 图像生成模型。它解决了原始模型依赖命令行、操作门槛高且功能分散的痛点，将复杂的 AI 绘图流程整合进一个直观易用的图形化平台。\n\n无论是希望快速上手的普通创作者、需要精细控制画面细节的设计师，还是想要深入探索模型潜力的开发者与研究人员，都能从中获益。其核心亮点在于极高的功能丰富度：不仅支持文生图、图生图、局部重绘（Inpainting）和外绘（Outpainting）等基础模式，还独创了注意力机制调整、提示词矩阵、负向提示词以及“高清修复”等高级功能。此外，它内置了 GFPGAN 和 CodeFormer 等人脸修复工具，支持多种神经网络放大算法，并允许用户通过插件系统无限扩展能力。即使是显存有限的设备，stable-diffusion-webui 也提供了相应的优化选项，让高质量的 AI 艺术创作变得触手可及。",162132,3,"2026-04-05T11:01:52",[13,14,15],"开发框架","图像","Agent","ready",{"id":18,"name":19,"github_repo":20,"description_zh":21,"stars":22,"difficulty_score":23,"last_commit_at":24,"category_tags":25,"status":16},1381,"everything-claude-code","affaan-m\u002Feverything-claude-code","everything-claude-code 是一套专为 AI 编程助手（如 Claude Code、Codex、Cursor 等）打造的高性能优化系统。它不仅仅是一组配置文件，而是一个经过长期实战打磨的完整框架，旨在解决 AI 代理在实际开发中面临的效率低下、记忆丢失、安全隐患及缺乏持续学习能力等核心痛点。\n\n通过引入技能模块化、直觉增强、记忆持久化机制以及内置的安全扫描功能，everything-claude-code 能显著提升 AI 在复杂任务中的表现，帮助开发者构建更稳定、更智能的生产级 AI 代理。其独特的“研究优先”开发理念和针对 Token 消耗的优化策略，使得模型响应更快、成本更低，同时有效防御潜在的攻击向量。\n\n这套工具特别适合软件开发者、AI 研究人员以及希望深度定制 AI 工作流的技术团队使用。无论您是在构建大型代码库，还是需要 AI 协助进行安全审计与自动化测试，everything-claude-code 都能提供强大的底层支持。作为一个曾荣获 Anthropic 黑客大奖的开源项目，它融合了多语言支持与丰富的实战钩子（hooks），让 AI 真正成长为懂上",140436,2,"2026-04-05T23:32:43",[13,15,26],"语言模型",{"id":28,"name":29,"github_repo":30,"description_zh":31,"stars":32,"difficulty_score":23,"last_commit_at":33,"category_tags":34,"status":16},2271,"ComfyUI","Comfy-Org\u002FComfyUI","ComfyUI 是一款功能强大且高度模块化的视觉 AI 引擎，专为设计和执行复杂的 Stable Diffusion 图像生成流程而打造。它摒弃了传统的代码编写模式，采用直观的节点式流程图界面，让用户通过连接不同的功能模块即可构建个性化的生成管线。\n\n这一设计巧妙解决了高级 AI 绘图工作流配置复杂、灵活性不足的痛点。用户无需具备编程背景，也能自由组合模型、调整参数并实时预览效果，轻松实现从基础文生图到多步骤高清修复等各类复杂任务。ComfyUI 拥有极佳的兼容性，不仅支持 Windows、macOS 和 Linux 全平台，还广泛适配 NVIDIA、AMD、Intel 及苹果 Silicon 等多种硬件架构，并率先支持 SDXL、Flux、SD3 等前沿模型。\n\n无论是希望深入探索算法潜力的研究人员和开发者，还是追求极致创作自由度的设计师与资深 AI 绘画爱好者，ComfyUI 都能提供强大的支持。其独特的模块化架构允许社区不断扩展新功能，使其成为当前最灵活、生态最丰富的开源扩散模型工具之一，帮助用户将创意高效转化为现实。",107662,"2026-04-03T11:11:01",[13,14,15],{"id":36,"name":37,"github_repo":38,"description_zh":39,"stars":40,"difficulty_score":23,"last_commit_at":41,"category_tags":42,"status":16},3704,"NextChat","ChatGPTNextWeb\u002FNextChat","NextChat 是一款轻量且极速的 AI 助手，旨在为用户提供流畅、跨平台的大模型交互体验。它完美解决了用户在多设备间切换时难以保持对话连续性，以及面对众多 AI 模型不知如何统一管理的痛点。无论是日常办公、学习辅助还是创意激发，NextChat 都能让用户随时随地通过网页、iOS、Android、Windows、MacOS 或 Linux 端无缝接入智能服务。\n\n这款工具非常适合普通用户、学生、职场人士以及需要私有化部署的企业团队使用。对于开发者而言，它也提供了便捷的自托管方案，支持一键部署到 Vercel 或 Zeabur 等平台。\n\nNextChat 的核心亮点在于其广泛的模型兼容性，原生支持 Claude、DeepSeek、GPT-4 及 Gemini Pro 等主流大模型，让用户在一个界面即可自由切换不同 AI 能力。此外，它还率先支持 MCP（Model Context Protocol）协议，增强了上下文处理能力。针对企业用户，NextChat 提供专业版解决方案，具备品牌定制、细粒度权限控制、内部知识库整合及安全审计等功能，满足公司对数据隐私和个性化管理的高标准要求。",87618,"2026-04-05T07:20:52",[13,26],{"id":44,"name":45,"github_repo":46,"description_zh":47,"stars":48,"difficulty_score":23,"last_commit_at":49,"category_tags":50,"status":16},2268,"ML-For-Beginners","microsoft\u002FML-For-Beginners","ML-For-Beginners 是由微软推出的一套系统化机器学习入门课程，旨在帮助零基础用户轻松掌握经典机器学习知识。这套课程将学习路径规划为 12 周，包含 26 节精炼课程和 52 道配套测验，内容涵盖从基础概念到实际应用的完整流程，有效解决了初学者面对庞大知识体系时无从下手、缺乏结构化指导的痛点。\n\n无论是希望转型的开发者、需要补充算法背景的研究人员，还是对人工智能充满好奇的普通爱好者，都能从中受益。课程不仅提供了清晰的理论讲解，还强调动手实践，让用户在循序渐进中建立扎实的技能基础。其独特的亮点在于强大的多语言支持，通过自动化机制提供了包括简体中文在内的 50 多种语言版本，极大地降低了全球不同背景用户的学习门槛。此外，项目采用开源协作模式，社区活跃且内容持续更新，确保学习者能获取前沿且准确的技术资讯。如果你正寻找一条清晰、友好且专业的机器学习入门之路，ML-For-Beginners 将是理想的起点。",84991,"2026-04-05T10:45:23",[14,51,52,53,15,54,26,13,55],"数据工具","视频","插件","其他","音频",{"id":57,"name":58,"github_repo":59,"description_zh":60,"stars":61,"difficulty_score":10,"last_commit_at":62,"category_tags":63,"status":16},3128,"ragflow","infiniflow\u002Fragflow","RAGFlow 是一款领先的开源检索增强生成（RAG）引擎，旨在为大语言模型构建更精准、可靠的上下文层。它巧妙地将前沿的 RAG 技术与智能体（Agent）能力相结合，不仅支持从各类文档中高效提取知识，还能让模型基于这些知识进行逻辑推理和任务执行。\n\n在大模型应用中，幻觉问题和知识滞后是常见痛点。RAGFlow 通过深度解析复杂文档结构（如表格、图表及混合排版），显著提升了信息检索的准确度，从而有效减少模型“胡编乱造”的现象，确保回答既有据可依又具备时效性。其内置的智能体机制更进一步，使系统不仅能回答问题，还能自主规划步骤解决复杂问题。\n\n这款工具特别适合开发者、企业技术团队以及 AI 研究人员使用。无论是希望快速搭建私有知识库问答系统，还是致力于探索大模型在垂直领域落地的创新者，都能从中受益。RAGFlow 提供了可视化的工作流编排界面和灵活的 API 接口，既降低了非算法背景用户的上手门槛，也满足了专业开发者对系统深度定制的需求。作为基于 Apache 2.0 协议开源的项目，它正成为连接通用大模型与行业专有知识之间的重要桥梁。",77062,"2026-04-04T04:44:48",[15,14,13,26,54],{"id":65,"github_repo":66,"name":67,"description_en":68,"description_zh":69,"ai_summary_zh":69,"readme_en":70,"readme_zh":71,"quickstart_zh":72,"use_case_zh":73,"hero_image_url":74,"owner_login":75,"owner_name":76,"owner_avatar_url":77,"owner_bio":78,"owner_company":79,"owner_location":80,"owner_email":80,"owner_twitter":80,"owner_website":80,"owner_url":81,"languages":82,"stars":91,"forks":92,"last_commit_at":93,"license":94,"difficulty_score":10,"env_os":95,"env_gpu":96,"env_ram":96,"env_deps":97,"category_tags":105,"github_topics":106,"view_count":124,"oss_zip_url":80,"oss_zip_packed_at":80,"status":16,"created_at":125,"updated_at":126,"faqs":127,"releases":158},819,"shyamsaktawat\u002FOpenAlpha_Evolve","OpenAlpha_Evolve","OpenAlpha_Evolve is an open-source Python framework inspired by the groundbreaking research on autonomous coding agents like DeepMind's AlphaEvolve.","OpenAlpha_Evolve 是一款基于 Python 的开源框架，致力于利用大语言模型实现算法的自动化演进。受 DeepMind AlphaEvolve 研究启发，它模拟自然选择机制，通过“生成 - 测试 - 改进”的循环，自主发现并持续优化代码解决方案。\n\n针对传统编程中人工调试效率低、复杂算法探索成本高的问题，OpenAlpha_Evolve 提供了一套完整的自动化流程。用户只需定义任务目标与测试用例，系统便会调度多个智能体协作，从初始代码生成到变异修复，再到性能评估，最终筛选出最优解。\n\nOpenAlpha_Evolve 特别适合希望探索 AI 驱动编程、自动化问题解决的研究人员和开发者。其核心亮点在于模块化代理架构，涵盖提示词设计、代码生成、评估及选择控制器，并支持基于差异（diff）的代码迭代更新。这不仅降低了算法创新的门槛，也为构建更智能的自主编码系统提供了可扩展的基础设施，让 AI 真正参与到代码进化的过程中。","# OpenAlpha_Evolve: Contribute to Improve this Project\n\n![openalpha_evolve_workflow](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fshyamsaktawat_OpenAlpha_Evolve_readme_cd5e5d234a26.png)\n\nOpenAlpha_Evolve is an open-source Python framework inspired by the groundbreaking research on autonomous coding agents like DeepMind's AlphaEvolve. It's a **regeneration** of the core idea: an intelligent system that iteratively writes, tests, and improves code using Large Language Models (LLMs) via LiteLLM, guided by the principles of evolution.\n\nOur mission is to provide an accessible, understandable, and extensible platform for researchers, developers, and enthusiasts to explore the fascinating intersection of AI, code generation, and automated problem-solving.\n\n[![License: MIT](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-MIT-yellow.svg)](LICENSE.md)\n\n## Table of Contents\n- [✨ The Vision: AI-Driven Algorithmic Innovation](#-the-vision-ai-driven-algorithmic-innovation)\n- [🧠 How It Works: The Evolutionary Cycle](#-how-it-works-the-evolutionary-cycle)\n- [🚀 Key Features](#-key-features)\n- [📂 Project Structure](#-project-structure)\n- [🏁 Getting Started](#-getting-started)\n- [💡 Defining Your Own Algorithmic Quests!](#-defining-your-own-algorithmic-quests)\n- [🔮 The Horizon: Future Evolution](#-the-horizon-future-evolution)\n- [🤝 Join the Evolution: Contributing](#-join-the-evolution-contributing)\n- [📜 License](#-license)\n- [🙏 Homage](#-homage)\n\n---\n![image](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fshyamsaktawat_OpenAlpha_Evolve_readme_dd51e4aed2a2.png)\n![image](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fshyamsaktawat_OpenAlpha_Evolve_readme_514c6ccf191a.png)\n\n\n\n## ✨ The Vision: AI-Driven Algorithmic Innovation\n\nImagine an agent that can:\n\n*   Understand a complex problem description.\n*   Generate initial algorithmic solutions.\n*   Rigorously test its own code.\n*   Learn from failures and successes.\n*   Evolve increasingly sophisticated and efficient algorithms over time.\n\nOpenAlpha_Evolve is a step towards this vision. It's not just about generating code; it's about creating a system that *discovers* and *refines* solutions autonomously.\n\n---\n\u003Cimg width=\"1253\" alt=\"Screenshot 2025-05-19 at 12 17 58 AM\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fshyamsaktawat_OpenAlpha_Evolve_readme_f528ed95e622.png\" \u002F>\n\n## 🧠 How It Works: The Evolutionary Cycle\n\nOpenAlpha_Evolve employs a modular, agent-based architecture to orchestrate an evolutionary process:\n\n1.  **Task Definition**: You, the user, define the algorithmic \"quest\" – the problem to be solved, including examples of inputs and expected outputs.\n2.  **Prompt Engineering (`PromptDesignerAgent`)**: This agent crafts intelligent prompts for the LLM. It designs:\n    *   *Initial Prompts*: To generate the first set of candidate solutions.\n    *   *Mutation Prompts*: To introduce variations and improvements to existing solutions, often requesting changes in a \"diff\" format.\n    *   *Bug-Fix Prompts*: To guide the LLM in correcting errors from previous attempts, also typically expecting a \"diff\".\n3.  **Code Generation (`CodeGeneratorAgent`)**: Powered by an LLM (currently configured for Gemini), this agent takes the prompts and generates Python code. If a \"diff\" is requested and received, it attempts to apply the changes to the parent code.\n4.  **Evaluation (`EvaluatorAgent`)**: The generated code is put to the test!\n    *   *Syntax Check*: Is the code valid Python?\n    *   *Execution*: The code is run in a temporary, isolated environment against the input\u002Foutput examples defined in the task.\n    *   *Fitness Scoring*: Programs are scored based on correctness (how many test cases pass), efficiency (runtime), and other potential metrics.\n5.  **Database (`DatabaseAgent`)**: All programs (code, fitness scores, generation, lineage) are stored, creating a record of the evolutionary history (currently in-memory).\n6.  **Selection (`SelectionControllerAgent`)**: The \"survival of the fittest\" principle in action. This agent selects:\n    *   *Parents*: Promising programs from the current generation to produce offspring.\n    *   *Survivors*: The best programs from both the current population and new offspring to advance to the next generation.\n7.  **Iteration**: This cycle repeats for a defined number of generations, with each new generation aiming to produce better solutions than the last.\n8.  **Orchestration (`TaskManagerAgent`)**: The maestro of the operation, coordinating all other agents and managing the overall evolutionary loop.\n\n---\n\n## 🚀 Key Features\n\n*   **LLM-Powered Code Generation**: Leverages state-of-the-art Large Language Models via LiteLLM, supporting multiple providers (OpenAI, Anthropic, Google, etc.).\n*   **Evolutionary Algorithm Core**: Implements iterative improvement through selection, LLM-driven mutation\u002Fbug-fixing using diffs, and survival.\n*   **Modular Agent Architecture**: Easily extend or replace individual components (e.g., use a different LLM, database, or evaluation strategy).\n*   **Automated Program Evaluation**: Syntax checking and functional testing against user-provided examples. Code execution is sandboxed using **Docker containers** for improved security and dependency management, with configurable timeout mechanisms.\n*   **Configuration Management**: Easily tweak parameters like population size, number of generations, LLM models, API settings, and Docker configurations via `config\u002Fsettings.py` and `.env`.\n*   **Detailed Logging**: Comprehensive logs provide insights into each step of the evolutionary process.\n*   **Diff-based Mutations**: The system is designed to use diffs for mutations and bug fixes, allowing for more targeted code modifications by the LLM.\n*   **Open Source & Extensible**: Built with Python, designed for experimentation and community contributions.\n\n---\n\n## 📂 Project Structure\n\n```text\n.\u002F\n├── code_generator\u002F      # Agent responsible for generating code using LLMs.\n├── database_agent\u002F      # Agent for managing the storage and retrieval of programs and their metadata.\n├── evaluator_agent\u002F     # Agent that evaluates the generated code for syntax, execution, and fitness.\n├── prompt_designer\u002F     # Agent that crafts prompts for the LLM for initial generation, mutation, and bug fixing.\n├── selection_controller\u002F  # Agent that implements the selection strategy for parent and survivor programs.\n├── task_manager\u002F        # Agent that orchestrates the overall evolutionary loop and coordinates other agents.\n├── config\u002F                  # Holds configuration files, primarily `settings.py` for system parameters and API keys.\n├── core\u002F                    # Defines core data structures and interfaces, like `Program` and `TaskDefinition`.\n├── tests\u002F                   # Includes unit and integration tests to ensure code quality and correctness.\n├── main.py                  # The main entry point to run the OpenAlpha_Evolve system and start an evolutionary run.\n├── requirements.txt         # Lists all Python package dependencies required to run the project.\n├── .env.example             # An example file showing the environment variables needed, such as API keys. Copy this to `.env` and fill in your values.\n├── .gitignore               # Specifies intentionally untracked files that Git should ignore (e.g., `.env`, `__pycache__\u002F`).\n├── LICENSE.md               # Contains the full text of the MIT License under which the project is distributed.\n└── README.md                # This file! Provides an overview of the project, setup instructions, and documentation.\n```\n\n---\n\n## 🏁 Getting Started\n\n1.  **Prerequisites**:\n    *   Python 3.10+\n    *   `pip` for package management\n    *   `git` for cloning\n    *   **Docker**: For sandboxed code evaluation. Ensure Docker Desktop (Windows\u002FMac) or Docker Engine (Linux) is installed and running. Visit [docker.com](https:\u002F\u002Fwww.docker.com\u002Fget-started) for installation instructions.\n\n2.  **Clone the Repository**:\n    ```bash\n    git clone https:\u002F\u002Fgithub.com\u002Fshyamsaktawat\u002FOpenAlpha_Evolve.git\n    cd OpenAlpha_Evolve\n    ```\n\n3.  **Set Up a Virtual Environment** (recommended):\n    ```bash\n    python -m venv venv\n    source venv\u002Fbin\u002Factivate  # On Windows: venv\\Scripts\\activate\n    ```\n\n4.  **Install Dependencies**:\n    ```bash\n    pip install -r requirements.txt\n    ```\n\n5.  **Set Up Environment Variables (Crucial for API Keys)**:\n    *   **This step is essential for the application to function correctly with your API keys.** The `.env` file stores your sensitive credentials and configuration, overriding the default placeholders in `config\u002Fsettings.py`.\n    *   Create your personal environment file by copying the example:\n        ```bash\n        cp .env_example .env\n        ```\n\n    #### LLM Configuration\n    Google Cloud authentication (e.g., via Application Default Credentials (ADC) or service account keys pointed to by `GOOGLE_APPLICATION_CREDENTIALS`) is a supported method for using Google's LLMs.\n\n    To set up your environment variables for Google Cloud, you can use one of the following methods. These should be added to your `.env` file:\n\n    ```bash\n    # For Google Cloud (Vertex AI \u002F AI Studio)\n    # Option 1: Using Application Default Credentials (ADC)\n    # Ensure you have authenticated via gcloud CLI:\n    # gcloud auth application-default login\n    # Or set the GOOGLE_APPLICATION_CREDENTIALS environment variable:\n    # GOOGLE_APPLICATION_CREDENTIALS=\"\u002Fpath\u002Fto\u002Fyour\u002Fservice-account-key.json\"\n\n    # Option 2: Directly using an API Key for specific Google services (e.g., Gemini API)\n    # GEMINI_API_KEY=\"your_gemini_api_key\"\n    ```\n\n    This project uses LiteLLM to interface with various LLM providers. For providers other than Google Cloud (e.g., OpenAI, Anthropic, Cohere), please refer to the [LiteLLM documentation](https:\u002F\u002Fdocs.litellm.ai\u002Fdocs\u002Fproviders) for the specific environment variables required. Common examples include:\n    ```bash\n    # OPENAI_API_KEY=\"your_openai_api_key\"\n    # ANTHROPIC_API_KEY=\"your_anthropic_api_key\"\n    # COHERE_API_KEY=\"your_cohere_api_key\"\n    ```\n    Add the necessary API key variables for your chosen LLM provider(s) to your `.env` file.\n\n6.  **Run OpenAlpha_Evolve!**\n    Run the example task (Dijkstra's algorithm) with:\n    ```bash\n    python -m main examples\u002Fshortest_path.yaml\n    ```\n    Watch the logs in your terminal to see the evolutionary process unfold! Log files are also saved to `alpha_evolve.log` (by default).\n\n7.  **Launch the Gradio Web Interface**\n    Interact with the system via the web UI. To start the Gradio app:\n    ```bash\n    python app.py\n    ```\n    Gradio will display a local URL (e.g., http:\u002F\u002F127.0.0.1:7860) and a public share link if enabled. Open this in your browser to define custom tasks and run the evolution process interactively.\n\n---\n\n## 💡 Defining Your Own Algorithmic Quests!\n\nWant to challenge OpenAlpha_Evolve with a new problem? It's easy! You can define your tasks in two ways:\n\n### 1. Using YAML Files (Recommended)\n\nCreate a YAML file in the `examples` directory with the following structure:\n\n```yaml\ntask_id: \"your_task_id\"\ntask_description: |\n  Your detailed problem description here.\n  Be specific about function names, expected behavior, and constraints.\nfunction_name: \"your_function_name\"\nallowed_imports: [\"module1\", \"module2\"]\n\ntests:\n  - description: \"Test group description\" # Describes a group of related tests\n    name: \"Test group name\" # A name for this test group\n    test_cases: # This should be a list of individual test cases\n      - input: [arg1, arg2]  # First test case\n        output: expected_output # Expected result for this input\n        # Each test case uses either 'output' for direct comparison\n        # or 'validation_func' for more complex validation.\n      - input: [arg_for_validation_func_1, arg_for_validation_func_2] # Second test case\n        validation_func: |\n          def validate(output_from_function):\n              # Custom validation logic for this specific test case's output\n              # For example, check if output is within a certain range,\n              # or if it has specific properties.\n              return isinstance(output_from_function, bool) and output_from_function is True\n```\n\nSee the example in `examples\u002Fshortest_path.yaml`\n\n### 2. Using Python Code (Legacy)\n\nYou can still define tasks programmatically using the `TaskDefinition` class:\n\n```python\nfrom core.task_definition import TaskDefinition\n\ntask = TaskDefinition(\n    id=\"your_task_id\",\n    description=\"Your detailed problem description\",\n    function_name_to_evolve=\"your_function_name\",\n    input_output_examples=[\n        {\"input\": [arg1, arg2], \"output\": expected_output},\n        # More examples...\n    ],\n    allowed_imports=[\"module1\", \"module2\"]\n)\n```\n\n### Best Practices for Task Definition\n\nCrafting effective task definitions is key to guiding OpenAlpha_Evolve successfully. Consider these tips:\n\n*   **Be Clear and Unambiguous**: Write task descriptions as if you're explaining the problem to another developer. Avoid jargon where possible, or explain it clearly.\n*   **Provide Diverse and Comprehensive Examples**: Your test cases are the primary way the agent verifies its generated code.\n    *   Include typical use cases\n    *   Cover edge cases (empty inputs, boundary values, etc.)\n    *   Include examples that test different logical paths\n    *   Use validation functions for complex checks\n*   **Start Simple, Then Increase Complexity**: Break down complex problems into simpler versions first.\n*   **Specify Constraints and Edge Cases**: Mention specific constraints and edge cases in the description.\n*   **Define Expected Function Signature**: Clearly state the expected function name and parameters.\n*   **Iterate and Refine**: Review and refine your task definition based on the agent's performance.\n\n---\n\n## 🔮 The Horizon: Future Evolution\n\n\n\n---\n\n## 🤝 Join the Evolution: Contributing\n\nThis is an open invitation to collaborate! Whether you're an AI researcher, a Python developer, or simply an enthusiast, your contributions are welcome.\n\n*   **Report Bugs**: Find an issue? Please create an issue on GitHub!\n*   **Suggest Features**: Have an idea to make OpenAlpha_Evolve better? Open an issue to discuss it!\n*   **Submit Pull Requests**:\n    *   Fork the repository.\n    *   Create a new branch for your feature or bugfix (`git checkout -b feature\u002Fyour-feature-name`).\n    *   Write clean, well-documented code.\n    *   Add tests for your changes if applicable.\n    *   Ensure your changes don't break existing functionality.\n    *   Submit a pull request with a clear description of your changes!\n\nLet's evolve this agent together!\n\n---\n\n## 📜 License\n\nThis project is licensed under the **MIT License**. See the `LICENSE.md` file for details.\n\n---\n\n## 🙏 Homage\n\nOpenAlpha_Evolve is proudly inspired by the pioneering work of the Google DeepMind team on AlphaEvolve and other related research in LLM-driven code generation and automated discovery. This project aims to make the core concepts more accessible for broader experimentation and learning. We stand on the shoulders of giants.\n\n---\n\n*Disclaimer: This is an experimental project. Generated code may not always be optimal, correct, or secure. Always review and test code thoroughly, especially before using it in production environments.* \n","# OpenAlpha_Evolve：贡献以改进此项目\n\n![openalpha_evolve_workflow](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fshyamsaktawat_OpenAlpha_Evolve_readme_cd5e5d234a26.png)\n\nOpenAlpha_Evolve 是一个开源 Python 框架，灵感来源于 DeepMind 的 AlphaEvolve 等自主编码智能体（autonomous coding agents）的前沿研究。它是对核心思想的**再生**：一个智能系统，通过 LiteLLM 利用大语言模型（Large Language Models，简称 LLMs），在进化原则的指导下，迭代地编写、测试和改进代码。\n\n我们的使命是为研究人员、开发者和爱好者提供一个易于访问、理解且可扩展的平台，以探索人工智能、代码生成和自动化问题解决之间迷人的交叉领域。\n\n[![License: MIT](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-MIT-yellow.svg)](LICENSE.md)\n\n## 目录\n- [✨ 愿景：AI 驱动的算法创新](#-the-vision-ai-driven-algorithmic-innovation)\n- [🧠 工作原理：进化循环](#-how-it-works-the-evolutionary-cycle)\n- [🚀 关键特性](#-key-features)\n- [📂 项目结构](#-project-structure)\n- [🏁 入门指南](#-getting-started)\n- [💡 定义您自己的算法任务！](#-defining-your-own-algorithmic-quests)\n- [🔮 展望：未来进化](#-the-horizon-future-evolution)\n- [🤝 加入进化：贡献](#-join-the-evolution-contributing)\n- [📜 许可证](#-license)\n- [🙏 致敬](#-homage)\n\n---\n![image](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fshyamsaktawat_OpenAlpha_Evolve_readme_dd51e4aed2a2.png)\n![image](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fshyamsaktawat_OpenAlpha_Evolve_readme_514c6ccf191a.png)\n\n\n\n## ✨ 愿景：AI 驱动的算法创新\n\n想象一个能够：\n\n*   理解复杂的问题描述。\n*   生成初始算法解决方案。\n*   严格测试其自身代码。\n*   从失败和成功中学习。\n*   随时间进化出越来越复杂和高效的算法。\n\nOpenAlpha_Evolve 是迈向这一愿景的一步。这不仅仅是关于生成代码；而是关于创建一个能够自主*发现*和*优化*解决方案的系统。\n\n---\n\u003Cimg width=\"1253\" alt=\"Screenshot 2025-05-19 at 12 17 58 AM\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fshyamsaktawat_OpenAlpha_Evolve_readme_f528ed95e622.png\" \u002F>\n\n## 🧠 工作原理：进化循环\n\nOpenAlpha_Evolve 采用模块化、基于智能体（agent-based）的架构来编排进化过程：\n\n1.  **任务定义（Task Definition）**：您，即用户，定义算法“任务”——要解决的问题，包括输入和预期输出的示例。\n2.  **提示词工程（Prompt Engineering）（`PromptDesignerAgent`）**：此智能体为 LLM 制作智能提示词。它设计：\n    *   *初始提示词（Initial Prompts）*：用于生成第一组候选解决方案。\n    *   *变异提示词（Mutation Prompts）*：用于引入现有解决方案的变体和改进，通常要求以“差异（diff）”格式提出更改。\n    *   *错误修复提示词（Bug-Fix Prompts）*：引导 LLM 纠正先前尝试中的错误，通常也期望返回“差异”。\n3.  **代码生成（Code Generation）（`CodeGeneratorAgent`）**：由 LLM 驱动（当前配置为 Gemini），此智能体接收提示词并生成 Python 代码。如果请求并收到“差异”，它会尝试将更改应用到父代码中。\n4.  **评估（Evaluation）（`EvaluatorAgent`）**：生成的代码接受测试！\n    *   *语法检查（Syntax Check）*：代码是否为有效的 Python？\n    *   *执行（Execution）*：代码在临时隔离环境中运行，针对任务中定义的输入\u002F输出示例进行测试。\n    *   *适应度评分（Fitness Scoring）*：程序根据正确性（通过多少测试用例）、效率（运行时间）和其他潜在指标进行评分。\n5.  **数据库（Database）（`DatabaseAgent`）**：所有程序（代码、适应度评分、生成记录、谱系）都被存储，创建进化历史记录（目前为内存中）。\n6.  **选择（Selection）（`SelectionControllerAgent`）**：“适者生存”原则的体现。此智能体选择：\n    *   *父代（Parents）*：来自当前世代有前途的程序，用于产生子代。\n    *   *幸存者（Survivors）*：来自当前种群和新子代中最好的程序，以进入下一代。\n7.  **迭代（Iteration）**：此循环重复定义的代数，每一代都旨在产生比上一代更好的解决方案。\n8.  **编排（Orchestration）（`TaskManagerAgent`）**：操作的大师，协调所有其他智能体并管理整体进化循环。\n\n---\n\n## 🚀 关键特性\n\n*   **LLM 驱动的代码生成**：利用最先进的 LLMs，通过 LiteLLM，支持多个提供商（OpenAI, Anthropic, Google 等）。\n*   **进化算法核心**：通过选择、LLM 驱动的变异\u002F错误修复（使用 diff）和生存来实现迭代改进。\n*   **模块化智能体架构**：轻松扩展或替换各个组件（例如，使用不同的 LLM、数据库或评估策略）。\n*   **自动化程序评估**：语法检查和针对用户提供的示例的功能测试。代码执行使用 **Docker 容器**进行沙箱化，以提高安全性和依赖管理，并配备可配置的超时机制。\n*   **配置管理**：通过 `config\u002Fsettings.py` 和 `.env` 轻松调整参数，如种群大小、代数、LLM 模型、API 设置和 Docker 配置。\n*   **详细日志**：全面的日志提供对进化过程每一步的洞察。\n*   **基于差异的变异**：系统设计为使用差异进行变异和错误修复，允许 LLM 进行更有针对性的代码修改。\n*   **开源与可扩展**：使用 Python 构建，专为实验和社区贡献而设计。\n\n---\n\n## 📂 项目结构\n\n```text\n.\u002F\n├── code_generator\u002F      # Agent responsible for generating code using LLMs.\n├── database_agent\u002F      # Agent for managing the storage and retrieval of programs and their metadata.\n├── evaluator_agent\u002F     # Agent that evaluates the generated code for syntax, execution, and fitness.\n├── prompt_designer\u002F     # Agent that crafts prompts for the LLM for initial generation, mutation, and bug fixing.\n├── selection_controller\u002F  # Agent that implements the selection strategy for parent and survivor programs.\n├── task_manager\u002F        # Agent that orchestrates the overall evolutionary loop and coordinates other agents.\n├── config\u002F                  # Holds configuration files, primarily `settings.py` for system parameters and API keys.\n├── core\u002F                    # Defines core data structures and interfaces, like `Program` and `TaskDefinition`.\n├── tests\u002F                   # Includes unit and integration tests to ensure code quality and correctness.\n├── main.py                  # The main entry point to run the OpenAlpha_Evolve system and start an evolutionary run.\n├── requirements.txt         # Lists all Python package dependencies required to run the project.\n├── .env.example             # An example file showing the environment variables needed, such as API keys. Copy this to `.env` and fill in your values.\n├── .gitignore               # Specifies intentionally untracked files that Git should ignore (e.g., `.env`, `__pycache__\u002F`).\n├── LICENSE.md               # Contains the full text of the MIT License under which the project is distributed.\n└── README.md                # This file! Provides an overview of the project, setup instructions, and documentation.\n```\n\n---\n\n## 🏁 入门指南\n\n1.  **前置条件**：\n    *   Python 3.10+\n    *   `pip` (包管理工具)\n    *   `git` (版本控制工具)\n    *   **Docker**：用于沙箱代码评估。请确保已安装并运行 Docker Desktop（Windows\u002FMac）或 Docker Engine（Linux）。访问 [docker.com](https:\u002F\u002Fwww.docker.com\u002Fget-started) 获取安装说明。\n\n2.  **克隆仓库**：\n    ```bash\n    git clone https:\u002F\u002Fgithub.com\u002Fshyamsaktawat\u002FOpenAlpha_Evolve.git\n    cd OpenAlpha_Evolve\n    ```\n\n3.  **设置虚拟环境**（推荐）：\n    ```bash\n    python -m venv venv\n    source venv\u002Fbin\u002Factivate  # On Windows: venv\\Scripts\\activate\n    ```\n\n4.  **安装依赖项**：\n    ```bash\n    pip install -r requirements.txt\n    ```\n\n5.  **设置环境变量**（对 API 密钥至关重要）：\n    *   **此步骤对于应用程序正确使用您的 API 密钥至关重要**。`.env` 文件存储您的敏感凭据和配置，覆盖 `config\u002Fsettings.py` 中的默认占位符。\n    *   通过复制示例创建您的个人环境文件：\n        ```bash\n        cp .env_example .env\n        ```\n\n    #### LLM 配置\n    Google Cloud 认证（例如，通过应用程序默认凭证 (ADC) 或由 `GOOGLE_APPLICATION_CREDENTIALS` 指向的服务账号密钥）是使用 Google 大语言模型 (LLM) 的一种支持方法。\n\n    要为 Google Cloud 设置环境变量，您可以使用以下任一方法。这些应添加到您的 `.env` 文件中：\n\n    ```bash\n    # For Google Cloud (Vertex AI \u002F AI Studio)\n    # Option 1: Using Application Default Credentials (ADC)\n    # Ensure you have authenticated via gcloud CLI:\n    # gcloud auth application-default login\n    # Or set the GOOGLE_APPLICATION_CREDENTIALS environment variable:\n    # GOOGLE_APPLICATION_CREDENTIALS=\"\u002Fpath\u002Fto\u002Fyour\u002Fservice-account-key.json\"\n\n    # Option 2: Directly using an API Key for specific Google services (e.g., Gemini API)\n    # GEMINI_API_KEY=\"your_gemini_api_key\"\n    ```\n\n    本项目使用 LiteLLM 来接口连接各种 LLM 提供商。对于 Google Cloud 以外的提供商（例如 OpenAI, Anthropic, Cohere），请参阅 [LiteLLM 文档](https:\u002F\u002Fdocs.litellm.ai\u002Fdocs\u002Fproviders) 以了解所需的具体环境变量。常见示例包括：\n    ```bash\n    # OPENAI_API_KEY=\"your_openai_api_key\"\n    # ANTHROPIC_API_KEY=\"your_anthropic_api_key\"\n    # COHERE_API_KEY=\"your_cohere_api_key\"\n    ```\n    将您选择的 LLM 提供商所需的必要 API 密钥变量添加到您的 `.env` 文件中。\n\n6.  **运行 OpenAlpha_Evolve！**\n    使用以下命令运行示例任务（Dijkstra 算法）：\n    ```bash\n    python -m main examples\u002Fshortest_path.yaml\n    ```\n    在终端中查看日志以观察进化过程的展开！日志文件也默认保存到 `alpha_evolve.log`。\n\n7.  **启动 Gradio Web 界面**\n    通过 Web 用户界面 (Web UI) 与系统交互。要启动 Gradio 应用：\n    ```bash\n    python app.py\n    ```\n    Gradio 将显示一个本地 URL（例如 http:\u002F\u002F127.0.0.1:7860）以及如果启用了公共分享链接。在浏览器中打开它以定义自定义任务并交互式地运行进化过程。\n\n---\n\n## 💡 定义您自己的算法挑战！\n\n想用新问题挑战 OpenAlpha_Evolve 吗？这很简单！您可以通过两种方式定义您的任务：\n\n### 1. 使用 YAML 文件（推荐）\n\n在 `examples` 目录中创建一个具有以下结构的 YAML 文件：\n\n```yaml\ntask_id: \"your_task_id\"\ntask_description: |\n  Your detailed problem description here.\n  Be specific about function names, expected behavior, and constraints.\nfunction_name: \"your_function_name\"\nallowed_imports: [\"module1\", \"module2\"]\n\ntests:\n  - description: \"Test group description\" # Describes a group of related tests\n    name: \"Test group name\" # A name for this test group\n    test_cases: # This should be a list of individual test cases\n      - input: [arg1, arg2]  # First test case\n        output: expected_output # Expected result for this input\n        # Each test case uses either 'output' for direct comparison\n        # or 'validation_func' for more complex validation.\n      - input: [arg_for_validation_func_1, arg_for_validation_func_2] # Second test case\n        validation_func: |\n          def validate(output_from_function):\n              # Custom validation logic for this specific test case's output\n              # For example, check if output is within a certain range,\n              # or if it has specific properties.\n              return isinstance(output_from_function, bool) and output_from_function is True\n```\n\n参见 `examples\u002Fshortest_path.yaml` 中的示例\n\n### 2. 使用 Python 代码（旧版）\n\n您仍然可以使用 `TaskDefinition` 类以编程方式定义任务：\n\n```python\nfrom core.task_definition import TaskDefinition\n\ntask = TaskDefinition(\n    id=\"your_task_id\",\n    description=\"Your detailed problem description\",\n    function_name_to_evolve=\"your_function_name\",\n    input_output_examples=[\n        {\"input\": [arg1, arg2], \"output\": expected_output},\n        # More examples...\n    ],\n    allowed_imports=[\"module1\", \"module2\"]\n)\n```\n\n### 任务定义的最佳实践\n\n制定有效的任务定义是成功引导 OpenAlpha_Evolve 的关键。请考虑以下建议：\n\n*   **清晰且无歧义**：撰写任务描述时，仿佛向另一位开发者解释问题一样。尽量避免行话，或者清晰地解释它们。\n*   **提供多样且全面的示例**：测试用例是智能体 (Agent) 验证其生成代码的主要方式。\n    *   包含典型用例\n    *   覆盖边界情况（空输入、边界值等）\n    *   包含测试不同逻辑路径的示例\n    *   对于复杂检查使用验证函数\n*   **从简单开始，然后增加复杂度**：首先将复杂问题分解为更简单的版本。\n*   **指定约束和边界情况**：在描述中提及具体的约束和边界情况。\n*   **定义预期的函数签名 (Function Signature)**：清楚地说明预期的函数名称和参数。\n*   **迭代与优化**：根据智能体 (Agent) 的表现审查并优化你的任务定义。\n\n---\n\n## 🔮 展望：未来演进\n\n---\n\n## 🤝 加入演进：贡献指南\n\n这是一个开放的协作邀请！无论你是 AI 研究人员、Python 开发者，还是仅仅是一个爱好者，都欢迎你的贡献。\n\n*   **报告 Bug (缺陷)**：发现问题了吗？请在 GitHub 上创建一个 Issue (问题)！\n*   **建议功能**：有让 OpenAlpha_Evolve 变得更好的想法吗？打开一个 Issue (问题) 来讨论它！\n*   **提交拉取请求 (Pull Requests)**：\n    *   Fork (分叉) 仓库。\n    *   为你的功能或修复创建新分支（`git checkout -b feature\u002Fyour-feature-name`）。\n    *   编写清晰、文档完善的代码。\n    *   如果适用，为你的更改添加测试。\n    *   确保你的更改不会破坏现有功能。\n    *   提交一个带有清晰更改描述的拉取请求 (Pull Request)！\n\n让我们一起进化这个智能体 (Agent)！\n\n---\n\n## 📜 许可证\n\n本项目采用 **MIT 许可证** 授权。详细信息请参阅 `LICENSE.md` 文件。\n\n---\n\n## 🙏 致谢\n\nOpenAlpha_Evolve 自豪地受到 Google DeepMind 团队在 AlphaEvolve 以及其他基于大语言模型 (LLM) 的代码生成和自动发现相关研究的开创性工作的启发。本项目旨在使核心概念更容易被广泛实验和学习。我们站在巨人的肩膀上。\n\n---\n\n*免责声明：这是一个实验性项目。生成的代码可能并不总是最优、正确或安全的。请务必彻底审查和测试代码，特别是在将其用于生产环境之前。*","# OpenAlpha_Evolve 快速上手指南\n\nOpenAlpha_Evolve 是一个基于大语言模型（LLM）的开源 Python 框架，灵感源自 DeepMind 的 AlphaEvolve 研究。它通过进化算法自动编写、测试并改进代码，旨在探索 AI 在代码生成与自动化问题解决领域的潜力。\n\n## 环境准备\n\n在开始之前，请确保您的开发环境满足以下要求：\n\n*   **操作系统**：Windows \u002F macOS \u002F Linux\n*   **Python**：版本 3.10 或更高\n*   **工具链**：`git` (用于克隆仓库), `pip` (包管理)\n*   **Docker**：**必需**。用于隔离运行生成的代码以确保安全及依赖管理。请确保 Docker Desktop (Windows\u002FMac) 或 Docker Engine (Linux) 已安装并正在运行。\n\n> 💡 **提示**：国内开发者建议在 `pip install` 时配置国内镜像源以提升下载速度。\n\n## 安装步骤\n\n1.  **克隆项目**\n    ```bash\n    git clone https:\u002F\u002Fgithub.com\u002Fshyamsaktawat\u002FOpenAlpha_Evolve.git\n    cd OpenAlpha_Evolve\n    ```\n\n2.  **创建虚拟环境** (推荐)\n    ```bash\n    python -m venv venv\n    source venv\u002Fbin\u002Factivate  # Windows 用户请使用：venv\\Scripts\\activate\n    ```\n\n3.  **安装依赖**\n    ```bash\n    pip install -r requirements.txt\n    ```\n\n4.  **配置环境变量**\n    项目需要 API Key 才能调用 LLM。\n    ```bash\n    cp .env_example .env\n    ```\n    编辑 `.env` 文件，填入您选择的 LLM 提供商密钥（如 Google Gemini, OpenAI 等）。例如：\n    ```bash\n    GEMINI_API_KEY=\"your_gemini_api_key\"\n    # 或其他支持 LiteLLM 的提供商密钥\n    ```\n\n## 基本使用\n\n### 运行示例任务\n项目内置了一个最短路径算法（Dijkstra）的示例任务，可直接运行观察进化过程：\n\n```bash\npython -m main examples\u002Fshortest_path.yaml\n```\n运行后，终端将显示日志，同时默认日志会保存至 `alpha_evolve.log`。\n\n### 启动 Web 界面\n您可以通过 Gradio 网页界面交互式地定义任务并监控进化过程：\n\n```bash\npython app.py\n```\n浏览器打开显示的本地链接（如 `http:\u002F\u002F127.0.0.1:7860`）即可开始操作。\n\n### 自定义算法挑战\n您可以修改 `examples` 目录下的 YAML 文件来定义自己的问题。主要结构包括：\n*   `task_description`: 详细的问题描述。\n*   `function_name`: 目标函数名。\n*   `tests`: 包含输入 (`input`) 和预期输出 (`output`) 或验证函数 (`validation_func`) 的测试用例列表。","某金融风控团队正在攻坚一个实时交易数据清洗模块，要求在毫秒级高并发下将处理延迟降低 50%。\n\n### 没有 OpenAlpha_Evolve 时\n- 资深工程师需反复手写不同策略，人工排查 Bug 效率低下且易疲劳出错。\n- 传统单元测试难以覆盖所有极端输入，导致线上偶尔出现未预见的异常。\n- 算法调优依赖直觉，缺乏量化指标来科学衡量每次改进的实际 fitness 值。\n- 跨团队协作中，优化逻辑难以复现，新人接手维护成本极高。\n\n### 使用 OpenAlpha_Evolve 后\n- OpenAlpha_Evolve 根据任务定义自动生成多组候选代码并并行执行沙箱测试。\n- 评估代理自动计算正确率与运行时间，精准淘汰低效方案并保留优质基因。\n- 进化循环持续引入变异，逐步逼近最优解，大幅减少人为干预带来的盲区。\n- 数据库完整保存演化谱系，确保任何阶段的代码变更都可解释且可回溯。\n\n它将复杂的算法寻优工作转化为自动化、可量化的智能进化流程，显著提升研发效率。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fshyamsaktawat_OpenAlpha_Evolve_dd51e4ae.png","shyamsaktawat","shyam saktawat","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fshyamsaktawat_4afbac22.jpg"," Machine Learning  DAIICT","DAIICT",null,"https:\u002F\u002Fgithub.com\u002Fshyamsaktawat",[83,87],{"name":84,"color":85,"percentage":86},"Python","#3572A5",99.2,{"name":88,"color":89,"percentage":90},"Dockerfile","#384d54",0.8,996,147,"2026-04-04T11:22:26","MIT","Windows, macOS, Linux","未说明",{"notes":98,"python":99,"dependencies":100},"必须安装 Docker 用于代码沙箱测试；需配置 LLM API 密钥（如 Google Gemini、OpenAI 等）到.env 文件；支持通过 Gradio 启动 Web 界面交互。","3.10+",[101,102,103,104],"litellm","gradio","pyyaml","docker",[13,26,15],[107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123],"alphafold","evolutionary-algorithms","google","discovery","evolution-computing","evolutionary-algorithm","genetic-algorithm","iterative-methods","optimize","alphacode","coding-agent","distributed-evolutionary-algorithms","iterative-refinement","llm-engineering","llm-ensemble","llm-inference","openevolve",4,"2026-03-27T02:49:30.150509","2026-04-06T10:24:04.449126",[128,133,138,143,148,153],{"id":129,"question_zh":130,"answer_zh":131,"source_url":132},3534,"为什么 OpenAI Provider 的 PR 被移除了？","该功能已被重构并合并至 PR #18。现在推荐使用 `litellm` 库来替代原有的 OpenAI Provider，因为它支持更多的模型和提供商（如 Ollama、Claude 等）。具体实现可参考相关文档：https:\u002F\u002Fdocs.litellm.ai\u002Fdocs\u002F","https:\u002F\u002Fgithub.com\u002Fshyamsaktawat\u002FOpenAlpha_Evolve\u002Fissues\u002F16",{"id":134,"question_zh":135,"answer_zh":136,"source_url":137},3535,"如何确保项目与 OpenAI 模型的兼容性？","建议集成 `litellm` 库以兼容各种 OpenAI 兼容的 API。此外，也可以参考 Google Gemini API 的 OpenAI 兼容文档：https:\u002F\u002Fai.google.dev\u002Fgemini-api\u002Fdocs\u002Fopenai。这有助于在多种模型间切换使用。","https:\u002F\u002Fgithub.com\u002Fshyamsaktawat\u002FOpenAlpha_Evolve\u002Fissues\u002F2",{"id":139,"question_zh":140,"answer_zh":141,"source_url":142},3536,"运行时报错 `ModuleNotFoundError: No module named 'alpha_evolve_pro'` 如何解决？","这是一个已知的模块导入问题。维护者表示该问题已修复。请尝试更新代码仓库到最新版本，并确保虚拟环境配置正确后，再次运行 `python -m main`。","https:\u002F\u002Fgithub.com\u002Fshyamsaktawat\u002FOpenAlpha_Evolve\u002Fissues\u002F4",{"id":144,"question_zh":145,"answer_zh":146,"source_url":147},3537,"运行时出现 `AttributeError: 'Program' object has no attribute 'program_id'` 错误怎么办？","该错误发生在进化过程中保存程序时。维护者确认此问题已修复。请确保拉取最新的代码版本，然后重新运行程序即可正常解决。","https:\u002F\u002Fgithub.com\u002Fshyamsaktawat\u002FOpenAlpha_Evolve\u002Fissues\u002F5",{"id":149,"question_zh":150,"answer_zh":151,"source_url":152},3538,"在哪里可以找到现有的单元测试（Pytests）作为参考？","项目中已有部分测试用例可以作为起点。建议查看 `tests\u002Ftest_evaluator_agent.py` 文件中的测试代码，了解如何为不同的 Agent 组件编写测试。","https:\u002F\u002Fgithub.com\u002Fshyamsaktawat\u002FOpenAlpha_Evolve\u002Fissues\u002F27",{"id":154,"question_zh":155,"answer_zh":156,"source_url":157},3539,"评估器（Evaluator）对浮点数输出的检查是否过于严格？","是的，当前评估器直接使用 `==` 比较输出，可能导致浮点数精度差异判定错误。开发者计划更新 `_compare_outputs` 检查逻辑，引入容差机制（如 `math.isclose`），以便更合理地处理浮点数比较。","https:\u002F\u002Fgithub.com\u002Fshyamsaktawat\u002FOpenAlpha_Evolve\u002Fissues\u002F19",[]]