[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-junku901--machine_learning":3,"tool-junku901--machine_learning":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",138956,2,"2026-04-05T11:33:21",[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":78,"owner_location":79,"owner_email":80,"owner_twitter":78,"owner_website":78,"owner_url":81,"languages":82,"stars":87,"forks":88,"last_commit_at":89,"license":78,"difficulty_score":90,"env_os":91,"env_gpu":92,"env_ram":92,"env_deps":93,"category_tags":96,"github_topics":78,"view_count":23,"oss_zip_url":78,"oss_zip_packed_at":78,"status":16,"created_at":97,"updated_at":98,"faqs":99,"releases":127},2019,"junku901\u002Fmachine_learning","machine_learning","Machine learning library for Node.js","machine_learning 是一个专为 Node.js 设计的机器学习库，同时也支持在浏览器中直接运行，无需复杂配置。它提供了多种常用算法，包括逻辑回归、多层感知机（MLP）、支持向量机（SVM）、K近邻（KNN）、K均值聚类、决策树和非负矩阵分解等，帮助用户轻松构建分类与聚类模型。无论是预测数据类别，还是发现数据中的隐藏模式，它都能提供轻量级的解决方案，特别适合需要在服务端或前端快速实现基础机器学习功能的开发者。其亮点在于内置了 SMO 算法优化 SVM 训练，并采用 CART 构建决策树，兼顾效率与准确性。对于有编程基础的开发者、数据爱好者或教学研究者而言，它是一个低门槛、易上手的实践工具，尤其适合在资源受限的环境中部署轻量模型。文档清晰，示例丰富，支持浏览器直接运行，方便快速验证想法。","# machine_learning\n\nMachine learning library for node.js. You can also use this library in browser.\n\n[Demo in Browser!](http:\u002F\u002Fjoonku.com\u002Fproject\u002Fmachine_learning)\n\n[API Documentation](http:\u002F\u002Fjoonku.com\u002Fproject\u002Fmachine_learning\u002Fapidoc)\n## Installation\n\nNode.js\n```\n$ npm install machine_learning\n```\n\nTo use this library in browser, include [machine_learning.min.js](http:\u002F\u002Fjoonku.com\u002Fjs\u002Fmachine_learning.min.js) file.\n\n```html\n\u003Cscript src=\"\u002Fjs\u002Fmachine_learning.min.js\">\u003C\u002Fscript>\n```\n\n[Demo in Browser!](http:\u002F\u002Fjoonku.com\u002Fproject\u002Fmachine_learning)\n\nHere is the [API Documentation](http:\u002F\u002Fjoonku.com\u002Fproject\u002Fmachine_learning\u002Fapidoc). (Still in progress)\n## Features\n\n  * Logistic Regression\n  * MLP (Multi-Layer Perceptron)\n  * SVM (Support Vector Machine)\n  * KNN (K-nearest neighbors)\n  * K-means clustering\n  * 3 Optimization Algorithms (Hill-Climbing, Simulated Annealing, Genetic Algorithm)\n  * Decision Tree\n  * NMF (non-negative matrix factorization)\n\n## Implementation Details\n\nSVM is using Sequential Minimal Optimization (SMO) for its training algorithm.\n\nFor Decision Tree, Classification And Regression Tree (CART) was used for its building algorithm.\n\n# Usage\n\n## Logistic Regression\n```javascript\nvar ml = require('machine_learning');\nvar x = [[1,1,1,0,0,0],\n         [1,0,1,0,0,0],\n         [1,1,1,0,0,0],\n         [0,0,1,1,1,0],\n         [0,0,1,1,0,0],\n         [0,0,1,1,1,0]];\nvar y = [[1, 0],\n         [1, 0],\n         [1, 0],\n         [0, 1],\n         [0, 1],\n         [0, 1]];\n\nvar classifier = new ml.LogisticRegression({\n    'input' : x,\n    'label' : y,\n    'n_in' : 6,\n    'n_out' : 2\n});\n\nclassifier.set('log level',1);\n\nvar training_epochs = 800, lr = 0.01;\n\nclassifier.train({\n    'lr' : lr,\n    'epochs' : training_epochs\n});\n\nx = [[1, 1, 0, 0, 0, 0],\n     [0, 0, 0, 1, 1, 0],\n     [1, 1, 1, 1, 1, 0]];\n\nconsole.log(\"Result : \",classifier.predict(x));\n```\n\n## MLP (Multi-Layer Perceptron)\n```javascript\nvar ml = require('machine_learning');\nvar x = [[0.4, 0.5, 0.5, 0.,  0.,  0.],\n         [0.5, 0.3,  0.5, 0.,  0.,  0.],\n         [0.4, 0.5, 0.5, 0.,  0.,  0.],\n         [0.,  0.,  0.5, 0.3, 0.5, 0.],\n         [0.,  0.,  0.5, 0.4, 0.5, 0.],\n         [0.,  0.,  0.5, 0.5, 0.5, 0.]];\nvar y = [[1, 0],\n         [1, 0],\n         [1, 0],\n         [0, 1],\n         [0, 1],\n         [0, 1]];\n\nvar mlp = new ml.MLP({\n    'input' : x,\n    'label' : y,\n    'n_ins' : 6,\n    'n_outs' : 2,\n    'hidden_layer_sizes' : [4,4,5]\n});\n\nmlp.set('log level',1); \u002F\u002F 0 : nothing, 1 : info, 2 : warning.\n\nmlp.train({\n    'lr' : 0.6,\n    'epochs' : 20000\n});\n\na = [[0.5, 0.5, 0., 0., 0., 0.],\n     [0., 0., 0., 0.5, 0.5, 0.],\n     [0.5, 0.5, 0.5, 0.5, 0.5, 0.]];\n\nconsole.log(mlp.predict(a));\n```\n\n## SVM (Support Vector Machine)\n```javascript\nvar ml = require('machine_learning');\nvar x = [[0.4, 0.5, 0.5, 0.,  0.,  0.],\n         [0.5, 0.3,  0.5, 0.,  0.,  0.01],\n         [0.4, 0.8, 0.5, 0.,  0.1,  0.2],\n         [1.4, 0.5, 0.5, 0.,  0.,  0.],\n         [1.5, 0.3,  0.5, 0.,  0.,  0.],\n         [0., 0.9, 1.5, 0.,  0.,  0.],\n         [0., 0.7, 1.5, 0.,  0.,  0.],\n         [0.5, 0.1,  0.9, 0.,  -1.8,  0.],\n         [0.8, 0.8, 0.5, 0.,  0.,  0.],\n         [0.,  0.9,  0.5, 0.3, 0.5, 0.2],\n         [0.,  0.,  0.5, 0.4, 0.5, 0.],\n         [0.,  0.,  0.5, 0.5, 0.5, 0.],\n         [0.3, 0.6, 0.7, 1.7,  1.3, -0.7],\n         [0.,  0.,  0.5, 0.3, 0.5, 0.2],\n         [0.,  0.,  0.5, 0.4, 0.5, 0.1],\n         [0.,  0.,  0.5, 0.5, 0.5, 0.01],\n         [0.2, 0.01, 0.5, 0.,  0.,  0.9],\n         [0.,  0.,  0.5, 0.3, 0.5, -2.3],\n         [0.,  0.,  0.5, 0.4, 0.5, 4],\n         [0.,  0.,  0.5, 0.5, 0.5, -2]];\n\nvar y =  [-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,1,1,1,1,1,1,1,1,1,1];\n\nvar svm = new ml.SVM({\n    x : x,\n    y : y\n});\n\nsvm.train({\n    C : 1.1, \u002F\u002F default : 1.0. C in SVM.\n    tol : 1e-5, \u002F\u002F default : 1e-4. Higher tolerance --> Higher precision\n    max_passes : 20, \u002F\u002F default : 20. Higher max_passes --> Higher precision\n    alpha_tol : 1e-5, \u002F\u002F default : 1e-5. Higher alpha_tolerance --> Higher precision\n\n    kernel : { type: \"polynomial\", c: 1, d: 5}\n    \u002F\u002F default : {type : \"gaussian\", sigma : 1.0}\n    \u002F\u002F {type : \"gaussian\", sigma : 0.5}\n    \u002F\u002F {type : \"linear\"} \u002F\u002F x*y\n    \u002F\u002F {type : \"polynomial\", c : 1, d : 8} \u002F\u002F (x*y + c)^d\n    \u002F\u002F Or you can use your own kernel.\n    \u002F\u002F kernel : function(vecx,vecy) { return dot(vecx,vecy);}\n});\n\nconsole.log(\"Predict : \",svm.predict([1.3,  1.7,  0.5, 0.5, 1.5, 0.4]));\n```\n\n## KNN (K-nearest neighbors)\n```javascript\nvar ml = require('machine_learning');\n\nvar data = [[1,0,1,0,1,1,1,0,0,0,0,0,1,0],\n            [1,1,1,1,1,1,1,0,0,0,0,0,1,0],\n            [1,1,1,0,1,1,1,0,1,0,0,0,1,0],\n            [1,0,1,1,1,1,1,1,0,0,0,0,1,0],\n            [1,1,1,1,1,1,1,0,0,0,0,0,1,1],\n            [0,0,1,0,0,1,0,0,1,0,1,1,1,0],\n            [0,0,0,0,0,0,1,1,1,0,1,1,1,0],\n            [0,0,0,0,0,1,1,1,0,1,0,1,1,0],\n            [0,0,1,0,1,0,1,1,1,1,0,1,1,1],\n            [0,0,0,0,0,0,1,1,1,1,1,1,1,1],\n            [1,0,1,0,0,1,1,1,1,1,0,0,1,0]\n           ];\n\nvar result = [23,12,23,23,45,70,123,73,146,158,64];\n\nvar knn = new ml.KNN({\n    data : data,\n    result : result\n});\n\nvar y = knn.predict({\n    x : [0,0,0,0,0,0,0,1,1,1,1,1,1,1],\n    k : 3,\n\n    weightf : {type : 'gaussian', sigma : 10.0},\n    \u002F\u002F default : {type : 'gaussian', sigma : 10.0}\n    \u002F\u002F {type : 'none'}. weight == 1\n    \u002F\u002F Or you can use your own weight f\n    \u002F\u002F weightf : function(distance) {return 1.\u002Fdistance}\n\n    distance : {type : 'euclidean'}\n    \u002F\u002F default : {type : 'euclidean'}\n    \u002F\u002F {type : 'pearson'}\n    \u002F\u002F Or you can use your own distance function\n    \u002F\u002F distance : function(vecx, vecy) {return Math.abs(dot(vecx,vecy));}\n});\n\nconsole.log(y);\n```\n\n## K-means clustering\n```javascript\nvar ml = require('machine_learning');\n\nvar data = [[1,0,1,0,1,1,1,0,0,0,0,0,1,0],\n            [1,1,1,1,1,1,1,0,0,0,0,0,1,0],\n            [1,1,1,0,1,1,1,0,1,0,0,0,1,0],\n            [1,0,1,1,1,1,1,1,0,0,0,0,1,0],\n            [1,1,1,1,1,1,1,0,0,0,0,0,1,1],\n            [0,0,1,0,0,1,0,0,1,0,1,1,1,0],\n            [0,0,0,0,0,0,1,1,1,0,1,1,1,0],\n            [0,0,0,0,0,1,1,1,0,1,0,1,1,0],\n            [0,0,1,0,1,0,1,1,1,1,0,1,1,1],\n            [0,0,0,0,0,0,1,1,1,1,1,1,1,1],\n            [1,0,1,0,0,1,1,1,1,1,0,0,1,0]\n           ];\n\nvar result = ml.kmeans.cluster({\n    data : data,\n    k : 4,\n    epochs: 100,\n\n    distance : {type : \"pearson\"}\n    \u002F\u002F default : {type : 'euclidean'}\n    \u002F\u002F {type : 'pearson'}\n    \u002F\u002F Or you can use your own distance function\n    \u002F\u002F distance : function(vecx, vecy) {return Math.abs(dot(vecx,vecy));}\n});\n\nconsole.log(\"clusters : \", result.clusters);\nconsole.log(\"means : \", result.means);\n```\n\n## Hill-Climbing\n```javascript\nvar ml = require('machine_learning');\n\nvar costf = function(vec) {\n    var cost = 0;\n    for(var i =0; i\u003C14;i++) { \u002F\u002F 15-dimensional vector\n        cost += (0.5*i*vec[i]*Math.exp(-vec[i]+vec[i+1])\u002Fvec[i+1])\n    }\n    cost += (3.*vec[14]\u002Fvec[0]);\n    return cost;\n};\n\nvar domain = [];\nfor(var i=0;i\u003C15;i++)\n    domain.push([1,70]); \u002F\u002F domain[idx][0] : minimum of vec[idx], domain[idx][1] : maximum of vec[idx].\n\nvar vec = ml.optimize.hillclimb({\n    domain : domain,\n    costf : costf\n});\n\nconsole.log(\"vec : \",vec);\nconsole.log(\"cost : \",costf(vec));\n```\n\n## Simulated Annealing\n```javascript\nvar ml = require('machine_learning');\n\nvar costf = function(vec) {\n    var cost = 0;\n    for(var i =0; i\u003C14;i++) { \u002F\u002F 15-dimensional vector\n        cost += (0.5*i*vec[i]*Math.exp(-vec[i]+vec[i+1])\u002Fvec[i+1])\n    }\n    cost += (3.*vec[14]\u002Fvec[0]);\n    return cost;\n};\n\nvar domain = [];\nfor(var i=0;i\u003C15;i++)\n    domain.push([1,70]); \u002F\u002F domain[idx][0] : minimum of vec[idx], domain[idx][1] : maximum of vec[idx].\n\nvar vec = ml.optimize.anneal({\n    domain : domain,\n    costf : costf,\n    temperature : 100000.0,\n    cool : 0.999,\n    step : 4\n});\n\nconsole.log(\"vec : \",vec);\nconsole.log(\"cost : \",costf(vec));\n```\n\n## Genetic Algorithm\n```javascript\nvar ml = require('machine_learning');\n\nvar costf = function(vec) {\n    var cost = 0;\n    for(var i =0; i\u003C14;i++) { \u002F\u002F 15-dimensional vector\n        cost += (0.5*i*vec[i]*Math.exp(-vec[i]+vec[i+1])\u002Fvec[i+1])\n    }\n    cost += (3.*vec[14]\u002Fvec[0]);\n    return cost;\n};\n\nvar domain = [];\nfor(var i=0;i\u003C15;i++)\n    domain.push([1,70]); \u002F\u002F domain[idx][0] : minimum of vec[idx], domain[idx][1] : maximum of vec[idx].\n\nvar vec = ml.optimize.genetic({\n    domain : domain,\n    costf : costf,\n    population : 50,\n    elite : 2, \u002F\u002F elitism. number of elite chromosomes.\n    epochs : 300,\n    q : 0.3 \u002F\u002F Rank-Based Fitness Assignment. fitness = q * (1-q)^(rank-1)\n            \u002F\u002F higher q --> higher selection pressure\n});\n\nconsole.log(\"vec : \",vec);\nconsole.log(\"cost : \",costf(vec));\n```\n\n## Decision Tree\n```javascript\n\u002F\u002F Reference : 'Programming Collective Intellignece' by Toby Segaran.\n\nvar ml = require('machine_learning');\n\nvar data =[['slashdot','USA','yes',18],\n           ['google','France','yes',23],\n           ['digg','USA','yes',24],\n           ['kiwitobes','France','yes',23],\n           ['google','UK','no',21],\n           ['(direct)','New Zealand','no',12],\n           ['(direct)','UK','no',21],\n           ['google','USA','no',24],\n           ['slashdot','France','yes',19],\n           ['digg','USA','no',18,],\n           ['google','UK','no',18,],\n           ['kiwitobes','UK','no',19],\n           ['digg','New Zealand','yes',12],\n           ['slashdot','UK','no',21],\n           ['google','UK','yes',18],\n           ['kiwitobes','France','yes',19]];\nvar result = ['None','Premium','Basic','Basic','Premium','None','Basic','Premium','None','None','None','None','Basic','None','Basic','Basic'];\n\nvar dt = new ml.DecisionTree({\n    data : data,\n    result : result\n});\n\ndt.build();\n\n\u002F\u002F dt.print();\n\nconsole.log(\"Classify : \", dt.classify(['(direct)','USA','yes',5]));\n\ndt.prune(1.0); \u002F\u002F 1.0 : mingain.\ndt.print();\n```\n\n## NMF (Non-negative matrix factorization)\n```javascript\nvar ml = require('machine_learning');\nvar matrix = [[22,28],\n              [49,64]];\n\nvar result = ml.nmf.factorize({\n    matrix : matrix,\n    features : 3,\n    epochs : 100\n});\n\nconsole.log(\"First Matrix : \",result[0]);\nconsole.log(\"Second Matrix : \",result[1]);\n```\n\n##License\n\n(The MIT License)\n\nCopyright (c) 2014 Joon-Ku Kang <junku901@gmail.com>\n\nPermission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the 'Software'), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and\u002For sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:\n\nThe above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.\n\nTHE SOFTWARE IS PROVIDED 'AS IS', WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.\n","# 机器学习\n\n适用于 Node.js 的机器学习库。您也可以在浏览器中使用此库。\n\n[浏览器演示！](http:\u002F\u002Fjoonku.com\u002Fproject\u002Fmachine_learning)\n\n[API 文档](http:\u002F\u002Fjoonku.com\u002Fproject\u002Fmachine_learning\u002Fapidoc)\n## 安装\n\nNode.js\n```\n$ npm install machine_learning\n```\n\n要在浏览器中使用此库，请引入 [machine_learning.min.js](http:\u002F\u002Fjoonku.com\u002Fjs\u002Fmachine_learning.min.js) 文件。\n\n```html\n\u003Cscript src=\"\u002Fjs\u002Fmachine_learning.min.js\">\u003C\u002Fscript>\n```\n\n[浏览器演示！](http:\u002F\u002Fjoonku.com\u002Fproject\u002Fmachine_learning)\n\n以下是[API 文档](http:\u002F\u002Fjoonku.com\u002Fproject\u002Fmachine_learning\u002Fapidoc)。（仍在开发中）\n## 功能\n\n  * 逻辑回归\n  * 多层感知机 (MLP)\n  * 支持向量机 (SVM)\n  * K 近邻 (KNN)\n  * K 均值聚类\n  * 3 种优化算法（爬山法、模拟退火、遗传算法）\n  * 决策树\n  * 非负矩阵分解 (NMF)\n\n## 实现细节\n\nSVM 使用序列最小优化 (SMO) 作为其训练算法。\n\n对于决策树，采用了分类与回归树 (CART) 作为其构建算法。\n\n# 使用方法\n\n## 逻辑回归\n```javascript\nvar ml = require('machine_learning');\nvar x = [[1,1,1,0,0,0],\n         [1,0,1,0,0,0],\n         [1,1,1,0,0,0],\n         [0,0,1,1,1,0],\n         [0,0,1,1,0,0],\n         [0,0,1,1,1,0]];\nvar y = [[1, 0],\n         [1, 0],\n         [1, 0],\n         [0, 1],\n         [0, 1],\n         [0, 1]];\n\nvar classifier = new ml.LogisticRegression({\n    'input' : x,\n    'label' : y,\n    'n_in' : 6,\n    'n_out' : 2\n});\n\nclassifier.set('log level',1);\n\nvar training_epochs = 800, lr = 0.01;\n\nclassifier.train({\n    'lr' : lr,\n    'epochs' : training_epochs\n});\n\nx = [[1, 1, 0, 0, 0, 0],\n     [0, 0, 0, 1, 1, 0],\n     [1, 1, 1, 1, 1, 0]];\n\nconsole.log(\"结果 : \",classifier.predict(x));\n```\n\n## 多层感知机 (MLP)\n```javascript\nvar ml = require('machine_learning');\nvar x = [[0.4, 0.5, 0.5, 0.,  0.,  0.],\n         [0.5, 0.3,  0.5, 0.,  0.,  0.],\n         [0.4, 0.5, 0.5, 0.,  0.,  0.],\n         [0.,  0.,  0.5, 0.3, 0.5, 0.],\n         [0.,  0.,  0.5, 0.4, 0.5, 0.],\n         [0.,  0.,  0.5, 0.5, 0.5, 0.]];\nvar y = [[1, 0],\n         [1, 0],\n         [1, 0],\n         [0, 1],\n         [0, 1],\n         [0, 1]];\n\nvar mlp = new ml.MLP({\n    'input' : x,\n    'label' : y,\n    'n_ins' : 6,\n    'n_outs' : 2,\n    'hidden_layer_sizes' : [4,4,5]\n});\n\nmlp.set('log level',1); \u002F\u002F 0 : 没有输出，1 : 信息，2 : 警告。\n\nmlp.train({\n    'lr' : 0.6,\n    'epochs' : 20000\n});\n\na = [[0.5, 0.5, 0., 0., 0., 0.],\n     [0., 0., 0., 0.5, 0.5, 0.],\n     [0.5, 0.5, 0.5, 0.5, 0.5, 0.]];\n\nconsole.log(mlp.predict(a));\n```\n\n## 支持向量机 (SVM)\n```javascript\nvar ml = require('machine_learning');\nvar x = [[0.4, 0.5, 0.5, 0.,  0.,  0.],\n         [0.5, 0.3,  0.5, 0.,  0.,  0.01],\n         [0.4, 0.8, 0.5, 0.,  0.1,  0.2],\n         [1.4, 0.5, 0.5, 0.,  0.,  0.],\n         [1.5, 0.3,  0.5, 0.,  0.,  0.],\n         [0., 0.9, 1.5, 0.,  0.,  0.],\n         [0., 0.7, 1.5, 0.,  0.,  0.],\n         [0.5, 0.1,  0.9, 0.,  -1.8,  0.],\n         [0.8, 0.8, 0.5, 0.,  0.,  0.],\n         [0.,  0.9,  0.5, 0.3, 0.5, 0.2],\n         [0.,  0.,  0.5, 0.4, 0.5, 0.],\n         [0.,  0.,  0.5, 0.5, 0.5, 0.],\n         [0.3, 0.6, 0.7, 1.7,  1.3, -0.7],\n         [0.,  0.,  0.5, 0.3, 0.5, 0.2],\n         [0.,  0.,  0.5, 0.4, 0.5, 0.1],\n         [0.,  0.,  0.5, 0.5, 0.5, 0.01],\n         [0.2, 0.01, 0.5, 0.,  0.,  0.9],\n         [0.,  0.,  0.5, 0.3, 0.5, -2.3],\n         [0.,  0.,  0.5, 0.4, 0.5, 4],\n         [0.,  0.,  0.5, 0.5, 0.5, -2]];\n\nvar y =  [-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,1,1,1,1,1,1,1,1,1,1];\n\nvar svm = new ml.SVM({\n    x : x,\n    y : y\n});\n\nsvm.train({\n    C : 1.1, \u002F\u002F 默认值：1.0。SVM 中的 C 参数。\n    tol : 1e-5, \u002F\u002F 默认值：1e-4。容差越高 --> 精度越高\n    max_passes : 20, \u002F\u002F 默认值：20。最大迭代次数越高 --> 精度越高\n    alpha_tol : 1e-5, \u002F\u002F 默认值：1e-5。alpha 容差越高 --> 精度越高\n\n    kernel : { type: \"polynomial\", c: 1, d: 5}\n    \u002F\u002F 默认值：{type : \"gaussian\", sigma : 1.0}\n    \u002F\u002F {type : \"gaussian\", sigma : 0.5}\n    \u002F\u002F {type : \"linear\"} \u002F\u002F x*y\n    \u002F\u002F {type : \"polynomial\", c : 1, d : 8} \u002F\u002F (x*y + c)^d\n    \u002F\u002F 或者您可以使用自己的核函数。\n    \u002F\u002F kernel : function(vecx,vecy) { return dot(vecx,vecy);}\n});\n\nconsole.log(\"预测 : \",svm.predict([1.3,  1.7,  0.5, 0.5, 1.5, 0.4]));\n```\n\n## K 近邻 (KNN)\n```javascript\nvar ml = require('machine_learning');\n\nvar data = [[1,0,1,0,1,1,1,0,0,0,0,0,1,0],\n            [1,1,1,1,1,1,1,0,0,0,0,0,1,0],\n            [1,1,1,0,1,1,1,0,1,0,0,0,1,0],\n            [1,0,1,1,1,1,1,1,0,0,0,0,1,0],\n            [1,1,1,1,1,1,1,0,0,0,0,0,1,1],\n            [0,0,1,0,0,1,0,0,1,0,1,1,1,0],\n            [0,0,0,0,0,0,1,1,1,0,1,1,1,0],\n            [0,0,0,0,0,1,1,1,0,1,0,1,1,0],\n            [0,0,1,0,1,0,1,1,1,1,0,1,1,1],\n            [0,0,0,0,0,0,1,1,1,1,1,1,1,1],\n            [1,0,1,0,0,1,1,1,1,1,0,0,1,0]\n           ];\n\nvar result = [23,12,23,23,45,70,123,73,146,158,64];\n\nvar knn = new ml.KNN({\n    data : data,\n    result : result\n});\n\nvar y = knn.predict({\n    x : [0,0,0,0,0,0,0,1,1,1,1,1,1,1],\n    k : 3,\n\n    weightf : {type : 'gaussian', sigma : 10.0},\n    \u002F\u002F 默认值：{type : 'gaussian', sigma : 10.0}\n    \u002F\u002F {type : 'none'}。权重 == 1\n    \u002F\u002F 或者您可以使用自己的权重函数\n    \u002F\u002F weightf : function(distance) {return 1.\u002Fdistance}\n\n    distance : {type : 'euclidean'}\n    \u002F\u002F 默认值：{type : 'euclidean'}\n    \u002F\u002F {type : 'pearson'}\n    \u002F\u002F 或者您可以使用自己的距离函数\n    \u002F\u002F distance : function(vecx, vecy) {return Math.abs(dot(vecx,vecy));}\n});\n\nconsole.log(y);\n```\n\n## K 均值聚类\n```javascript\nvar ml = require('machine_learning');\n\nvar data = [[1,0,1,0,1,1,1,0,0,0,0,0,1,0],\n            [1,1,1,1,1,1,1,0,0,0,0,0,1,0],\n            [1,1,1,0,1,1,1,0,1,0,0,0,1,0],\n            [1,0,1,1,1,1,1,1,0,0,0,0,1,0],\n            [1,1,1,1,1,1,1,0,0,0,0,0,1,1],\n            [0,0,1,0,0,1,0,0,1,0,1,1,1,0],\n            [0,0,0,0,0,0,1,1,1,0,1,1,1,0],\n            [0,0,0,0,0,1,1,1,0,1,0,1,1,0],\n            [0,0,1,0,1,0,1,1,1,1,0,1,1,1],\n            [0,0,0,0,0,0,1,1,1,1,1,1,1,1],\n            [1,0,1,0,0,1,1,1,1,1,0,0,1,0]\n           ];\n\nvar result = ml.kmeans.cluster({\n    data : data,\n    k : 4,\n    epochs: 100,\n\n    distance : {type : \"pearson\"}\n    \u002F\u002F 默认值：{type : 'euclidean'}\n    \u002F\u002F {type : 'pearson'}\n    \u002F\u002F 或者您可以使用自己的距离函数\n    \u002F\u002F distance : function(vecx, vecy) {return Math.abs(dot(vecx,vecy));}\n});\n\nconsole.log(\"聚类 : \", result.clusters);\nconsole.log(\"均值 : \", result.means);\n```\n\n## 山峰爬升\n```javascript\nvar ml = require('machine_learning');\n\nvar costf = function(vec) {\n    var cost = 0;\n    for(var i =0; i\u003C14;i++) { \u002F\u002F 15维向量\n        cost += (0.5*i*vec[i]*Math.exp(-vec[i]+vec[i+1])\u002Fvec[i+1])\n    }\n    cost += (3.*vec[14]\u002Fvec[0]);\n    return cost;\n};\n\nvar domain = [];\nfor(var i=0;i\u003C15;i++)\n    domain.push([1,70]); \u002F\u002F domain[idx][0] : vec[idx]的最小值，domain[idx][1] : vec[idx]的最大值。\n\nvar vec = ml.optimize.hillclimb({\n    domain : domain,\n    costf : costf\n});\n\nconsole.log(\"vec : \",vec);\nconsole.log(\"cost : \",costf(vec));\n```\n\n## 模拟退火\n```javascript\nvar ml = require('machine_learning');\n\nvar costf = function(vec) {\n    var cost = 0;\n    for(var i =0; i\u003C14;i++) { \u002F\u002F 15维向量\n        cost += (0.5*i*vec[i]*Math.exp(-vec[i]+vec[i+1])\u002Fvec[i+1])\n    }\n    cost += (3.*vec[14]\u002Fvec[0]);\n    return cost;\n};\n\nvar domain = [];\nfor(var i=0;i\u003C15;i++)\n    domain.push([1,70]); \u002F\u002F domain[idx][0] : vec[idx]的最小值，domain[idx][1] : vec[idx]的最大值。\n\nvar vec = ml.optimize.anneal({\n    domain : domain,\n    costf : costf,\n    temperature : 100000.0,\n    cool : 0.999,\n    step : 4\n});\n\nconsole.log(\"vec : \",vec);\nconsole.log(\"cost : \",costf(vec));\n```\n\n## 遗传算法\n```javascript\nvar ml = require('machine_learning');\n\nvar costf = function(vec) {\n    var cost = 0;\n    for(var i =0; i\u003C14;i++) { \u002F\u002F 15维向量\n        cost += (0.5*i*vec[i]*Math.exp(-vec[i]+vec[i+1])\u002Fvec[i+1])\n    }\n    cost += (3.*vec[14]\u002Fvec[0]);\n    return cost;\n};\n\nvar domain = [];\nfor(var i=0;i\u003C15;i++)\n    domain.push([1,70]); \u002F\u002F domain[idx][0] : vec[idx]的最小值，domain[idx][1] : vec[idx]的最大值。\n\nvar vec = ml.optimize.genetic({\n    domain : domain,\n    costf : costf,\n    population : 50,\n    elite : 2, \u002F\u002F 精英主义。精英染色体的数量。\n    epochs : 300,\n    q : 0.3 \u002F\u002F 基于排名的适应度分配。适应度 = q * (1-q)^(rank-1)\n            \u002F\u002F q越高 --> 选择压力越大\n});\n\nconsole.log(\"vec : \",vec);\nconsole.log(\"cost : \",costf(vec));\n```\n\n## 决策树\n```javascript\n\u002F\u002F 参考：托比·塞加兰的《集体智能编程》\n\nvar ml = require('machine_learning');\n\nvar data =[['slashdot','USA','yes',18],\n           ['google','France','yes',23],\n           ['digg','USA','yes',24],\n           ['kiwitobes','France','yes',23],\n           ['google','UK','no',21],\n           ['(direct)','New Zealand','no',12],\n           ['(direct)','UK','no',21],\n           ['google','USA','no',24],\n           ['slashdot','France','yes',19],\n           ['digg','USA','no',18,],\n           ['google','UK','no',18,],\n           ['kiwitobes','UK','no',19],\n           ['digg','New Zealand','yes',12],\n           ['slashdot','UK','no',21],\n           ['google','UK','yes',18],\n           ['kiwitobes','France','yes',19]];\nvar result = ['None','Premium','Basic','Basic','Premium','None','Basic','Premium','None','None','None','None','Basic','None','Basic','Basic'];\n\nvar dt = new ml.DecisionTree({\n    data : data,\n    result : result\n});\n\ndt.build();\n\n\u002F\u002F dt.print();\n\nconsole.log(\"分类结果 : \", dt.classify(['(direct)','USA','yes',5]));\n\ndt.prune(1.0); \u002F\u002F 1.0 : 最小增益。\ndt.print();\n```\n\n## 非负矩阵分解\n```javascript\nvar ml = require('machine_learning');\nvar matrix = [[22,28],\n              [49,64]];\n\nvar result = ml.nmf.factorize({\n    matrix : matrix,\n    features : 3,\n    epochs : 100\n});\n\nconsole.log(\"第一个矩阵 : \",result[0]);\nconsole.log(\"第二个矩阵 : \",result[1]);\n```\n\n## 许可证\n\n（MIT 许可证）\n\n版权所有 © 2014 康俊久 \u003Cjunku901@gmail.com>\n\n本软件及其相关文档文件（“软件”）的副本，由任何人免费获得，允许在不限制的情况下使用、复制、修改、合并、发布、分发、再许可和\u002F或出售该软件，并允许向提供该软件的人授予此类权利，但须遵守以下条件：\n\n上述版权声明和本许可声明应包含在所有副本或软件的重要部分中。\n\n软件按“原样”提供，不提供任何担保，无论是明示还是暗示，包括但不限于适销性、特定用途适用性和非侵权性的保证。在任何情况下，作者或版权持有者均不对任何索赔、损害或其他责任负责，无论这些责任是基于合同、侵权行为或其他原因，只要与软件或其使用有关。","# machine_learning 快速上手指南\n\n## 环境准备\n\n- **系统要求**：支持 Node.js（推荐 v14+）或现代浏览器（Chrome\u002FFirefox\u002FEdge）\n- **前置依赖**：\n  - Node.js 环境（用于服务端）\n  - 浏览器环境（用于前端，无需额外依赖）\n\n## 安装步骤\n\n### Node.js 安装\n```bash\nnpm install machine_learning\n```\n\n### 浏览器使用\n在 HTML 中引入 CDN 文件：\n```html\n\u003Cscript src=\"http:\u002F\u002Fjoonku.com\u002Fjs\u002Fmachine_learning.min.js\">\u003C\u002Fscript>\n```\n> 推荐使用国内镜像加速：如将链接替换为 [jsDelivr](https:\u002F\u002Fcdn.jsdelivr.net\u002Fnpm\u002Fmachine_learning@latest\u002Fjs\u002Fmachine_learning.min.js)（如可用）\n\n## 基本使用\n\n### 最简示例：逻辑回归分类\n\n```javascript\nvar ml = require('machine_learning');\nvar x = [[1,1,1,0,0,0], [1,0,1,0,0,0], [0,0,1,1,1,0]];\nvar y = [[1,0], [1,0], [0,1]];\n\nvar classifier = new ml.LogisticRegression({\n    'input': x,\n    'label': y,\n    'n_in': 6,\n    'n_out': 2\n});\n\nclassifier.train({ 'lr': 0.01, 'epochs': 100 });\nconsole.log(classifier.predict([[1,1,0,0,0,0]]));\n```\n\n> 在浏览器中直接使用：`ml.LogisticRegression` 等 API 与 Node.js 完全一致，无需修改代码。","一家中小型电商公司正在开发商品推荐系统，希望根据用户浏览和购买历史，自动识别高潜力商品并推送个性化广告。团队仅有两名全栈开发者，没有专职数据科学家，且预算有限，无法采购商业AI服务。\n\n### 没有 machine_learning 时\n- 团队依赖手动规则（如“购买过A商品的人也买B”），推荐准确率不足40%，用户点击率持续低迷。\n- 尝试用Python搭建模型，但后端是Node.js服务，部署复杂，需维护两个语言环境，增加运维成本。\n- 为实现分类功能，曾尝试用外部API，但响应延迟高，且每月费用超$500，超出预算。\n- 缺乏轻量级模型训练能力，无法在服务器本地迭代优化模型，只能被动等待用户反馈。\n- 没有实时预测能力，推荐逻辑只能每小时批量更新，错过用户活跃高峰时段的转化机会。\n\n### 使用 machine_learning 后\n- 直接在Node.js后端集成machine_learning，用MLP模型训练用户行为数据，推荐准确率提升至82%，点击率翻倍。\n- 无需切换语言或部署Python环境，所有模型训练与预测都在同一Node.js服务中完成，部署流程简化70%。\n- 完全离线运行，零第三方API费用，每月节省$500以上，模型训练成本趋近于零。\n- 可在用户每次浏览后实时调用predict()进行个性化推荐，响应时间低于50ms，显著提升体验。\n- 开发者仅用200行代码就实现了从数据预处理到在线预测的完整闭环，两周内上线新功能。\n\nmachine_learning 让非AI专家的开发团队，用现有技术栈快速构建出高效、低成本、可实时更新的智能推荐系统。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fjunku901_machine_learning_b0ca5025.png","junku901","Joon-Ku Kang","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fjunku901_09a7e618.jpg",null,"Seoul, South Korea","junku901@gmail.com","https:\u002F\u002Fgithub.com\u002Fjunku901",[83],{"name":84,"color":85,"percentage":86},"JavaScript","#f1e05a",100,733,129,"2025-01-23T05:32:41",1,"Linux, macOS, Windows","未说明",{"notes":94,"python":92,"dependencies":95},"该工具为 Node.js 库，可在 Node.js 环境或浏览器中运行，无需 Python。浏览器使用需引入 machine_learning.min.js 文件，Node.js 环境通过 npm 安装。所有算法均基于 JavaScript 实现，无 GPU 加速需求，内存和系统要求取决于数据规模。",[],[13],"2026-03-27T02:49:30.150509","2026-04-06T05:17:06.489087",[100,105,110,115,119,123],{"id":101,"question_zh":102,"answer_zh":103,"source_url":104},9150,"运行机器学习库时出现 'Block-scoped declarations (let, const, function, class) not yet supported outside strict mode' 错误怎么办？","在运行 Node.js 脚本时，需在文件顶部添加 'use strict'; 声明，或使用 Node.js 6+ 版本（支持 let\u002Fconst）。若仍报错，确保安装的是最新版本：执行 npm update machine_learning 或重新安装 npm install machine_learning。","https:\u002F\u002Fgithub.com\u002Fjunku901\u002Fmachine_learning\u002Fissues\u002F6",{"id":106,"question_zh":107,"answer_zh":108,"source_url":109},9151,"通过 npm 安装后仍使用的是旧版本，如何获取最新提交的代码？","维护者已更新 npm 包，执行命令 npm uninstall machine_learning && npm install machine_learning 以重新安装最新版本，确保获取包含修复的更新。","https:\u002F\u002Fgithub.com\u002Fjunku901\u002Fmachine_learning\u002Fissues\u002F5",{"id":111,"question_zh":112,"answer_zh":113,"source_url":114},9152,"KNN 算法中 weightf 参数的作用是什么？","weightf 参数用于定义样本权重函数，影响近邻的投票权重。若输出为字符串数组且算法失效，可能因未归一化特征值导致。建议对输入特征进行标准化（如 Min-Max 归一化），weightf 本身不解决数值范围差异问题。","https:\u002F\u002Fgithub.com\u002Fjunku901\u002Fmachine_learning\u002Fissues\u002F8",{"id":116,"question_zh":117,"answer_zh":118,"source_url":114},9153,"KNN 算法在特征值范围不同时预测结果不准确，如何解决？","需对输入特征进行归一化处理，例如将所有特征缩放到 [0,1] 区间。可使用公式 (x - min) \u002F (max - min) 对每个特征单独归一化，否则距离计算会受量纲影响，导致预测偏差。",{"id":120,"question_zh":121,"answer_zh":122,"source_url":109},9154,"如何确认 machine_learning 包已更新到最新版本？","执行 npm view machine_learning version 查看当前 npm 上的最新版本号，并与本地安装版本（npm list machine_learning）对比。若不一致，执行 npm install machine_learning@latest 强制更新。",{"id":124,"question_zh":125,"answer_zh":126,"source_url":104},9155,"Node.js 版本过低导致 let\u002Fconst 语法报错，最低支持版本是多少？","Node.js 6.0 及以上版本原生支持 let\u002Fconst 语法。建议升级至 Node.js 14+ 以获得最佳兼容性。若无法升级，可在脚本首行添加 'use strict'; 或使用 --harmony 参数运行：node --harmony logis.js。",[]]