[{"data":1,"prerenderedAt":17582},["ShallowReactive",2],{"navigation_docs":3,"-ai-python-batching-normalization-deep-networks-deep-networks-architectures":5180,"-ai-python-batching-normalization-deep-networks-deep-networks-architectures-surround":17577},[4,1707,2084,2556,2737,2863,3070,3192,3242,3299,3333,3459,3536,3550,3689,3983,4060,4793,4968,5061,5114,5176],{"title":5,"icon":6,"path":7,"stem":8,"children":9},"C#","i-devicon-csharp","\u002Fcsharp","01.csharp",[10,13,60,90,120,202,219,253,379,404,457,650,1364,1654,1703],{"title":11,"path":7,"stem":12},"C# та .NET","01.csharp\u002Findex",{"title":14,"icon":15,"path":16,"stem":17,"children":18,"page":59},"Fundamentals","i-lucide-book-open","\u002Fcsharp\u002Ffundamentals","01.csharp\u002F01.fundamentals",[19,23,27,31,35,39,43,47,51,55],{"title":20,"path":21,"stem":22},"Вступ до екосистеми .NET","\u002Fcsharp\u002Ffundamentals\u002Fintroduction-to-ecosystem","01.csharp\u002F01.fundamentals\u002F01.introduction-to-ecosystem",{"title":24,"path":25,"stem":26},"Структура програми на C#","\u002Fcsharp\u002Ffundamentals\u002Fprogram-structure","01.csharp\u002F01.fundamentals\u002F02.program-structure",{"title":28,"path":29,"stem":30},"Змінні та Типи Даних","\u002Fcsharp\u002Ffundamentals\u002Fvariables-data-types","01.csharp\u002F01.fundamentals\u002F03.variables-data-types",{"title":32,"path":33,"stem":34},"Масиви","\u002Fcsharp\u002Ffundamentals\u002Farrays","01.csharp\u002F01.fundamentals\u002F04.arrays",{"title":36,"path":37,"stem":38},"Strings & Text Handling","\u002Fcsharp\u002Ffundamentals\u002Fstrings-text-handling","01.csharp\u002F01.fundamentals\u002F05.strings-text-handling",{"title":40,"path":41,"stem":42},"Дати і Час","\u002Fcsharp\u002Ffundamentals\u002Fdates-time-handling","01.csharp\u002F01.fundamentals\u002F06.dates-time-handling",{"title":44,"path":45,"stem":46},"Потік Керування","\u002Fcsharp\u002Ffundamentals\u002Fcontrol-flow","01.csharp\u002F01.fundamentals\u002F07.control-flow",{"title":48,"path":49,"stem":50},"Методи","\u002Fcsharp\u002Ffundamentals\u002Fmethods","01.csharp\u002F01.fundamentals\u002F08.methods",{"title":52,"path":53,"stem":54},"Основи Відлагодження","\u002Fcsharp\u002Ffundamentals\u002Fdebugging-basics","01.csharp\u002F01.fundamentals\u002F09.debugging-basics",{"title":56,"path":57,"stem":58},"Інтерактивна Консоль (Classic)","\u002Fcsharp\u002Ffundamentals\u002Finteractive-console","01.csharp\u002F01.fundamentals\u002F10.interactive-console",false,{"title":61,"icon":62,"path":63,"stem":64,"children":65,"page":59},"OOP","i-lucide-box","\u002Fcsharp\u002Foop","01.csharp\u002F02.oop",[66,70,74,78,82,86],{"title":67,"path":68,"stem":69},"Package Management (Управління Пакетами)","\u002Fcsharp\u002Foop\u002Fpackage-management","01.csharp\u002F02.oop\u002F01.package-management",{"title":71,"path":72,"stem":73},"Класи та Об'єкти","\u002Fcsharp\u002Foop\u002Fclasses-objects","01.csharp\u002F02.oop\u002F02.classes-objects",{"title":75,"path":76,"stem":77},"Властивості та Поля","\u002Fcsharp\u002Foop\u002Fproperties-fields","01.csharp\u002F02.oop\u002F03.properties-fields",{"title":79,"path":80,"stem":81},"Стовпи ООП","\u002Fcsharp\u002Foop\u002Foop-pillars","01.csharp\u002F02.oop\u002F04.oop-pillars",{"title":83,"path":84,"stem":85},"Advanced Types","\u002Fcsharp\u002Foop\u002Fadvanced-types","01.csharp\u002F02.oop\u002F05.advanced-types",{"title":87,"path":88,"stem":89},"Namespaces (Простори Імен)","\u002Fcsharp\u002Foop\u002Fnamespaces","01.csharp\u002F02.oop\u002F06.namespaces",{"title":91,"icon":92,"path":93,"stem":94,"children":95,"page":59},"Advanced Core","i-lucide-zap","\u002Fcsharp\u002Fadvanced-core","01.csharp\u002F03.advanced-core",[96,100,104,108,112,116],{"title":97,"path":98,"stem":99},"Generics (Узагальнення)","\u002Fcsharp\u002Fadvanced-core\u002Fgenerics","01.csharp\u002F03.advanced-core\u002F01.generics",{"title":101,"path":102,"stem":103},"Делегати, Події та Лямбда-вирази","\u002Fcsharp\u002Fadvanced-core\u002Fdelegates-events-lambdas","01.csharp\u002F03.advanced-core\u002F02.delegates-events-lambdas",{"title":105,"path":106,"stem":107},"Interfaces Deep Dive (Інтерфейси: Поглиблений Розгляд)","\u002Fcsharp\u002Fadvanced-core\u002Finterfaces-deep-dive","01.csharp\u002F03.advanced-core\u002F03.interfaces-deep-dive",{"title":109,"path":110,"stem":111},"Обробка Винятків","\u002Fcsharp\u002Fadvanced-core\u002Fexception-handling","01.csharp\u002F03.advanced-core\u002F04.exception-handling",{"title":113,"path":114,"stem":115},"Pattern Matching","\u002Fcsharp\u002Fadvanced-core\u002Fpattern-matching","01.csharp\u002F03.advanced-core\u002F05.pattern-matching",{"title":117,"path":118,"stem":119},"Додаткові Можливості C#","\u002Fcsharp\u002Fadvanced-core\u002Fadditional-features","01.csharp\u002F03.advanced-core\u002F06.additional-features",{"title":121,"icon":122,"path":123,"stem":124,"children":125,"page":59},"Architecture Best Practices","i-lucide-building-2","\u002Fcsharp\u002Farchitecture-best-practices","01.csharp\u002F04.architecture-best-practices",[126,130,149,153,157,161,165,169],{"title":127,"path":128,"stem":129},"Software Design Principles (Частина 1)","\u002Fcsharp\u002Farchitecture-best-practices\u002Fsoftware-design-principles","01.csharp\u002F04.architecture-best-practices\u002F01.software-design-principles",{"title":131,"icon":132,"path":133,"stem":134,"children":135,"page":59},"Design Patterns","i-lucide-folder","\u002Fcsharp\u002Farchitecture-best-practices\u002Fdesign-patterns","01.csharp\u002F04.architecture-best-practices\u002F02.design-patterns",[136],{"title":137,"icon":132,"path":138,"stem":139,"children":140,"page":59},"Creational","\u002Fcsharp\u002Farchitecture-best-practices\u002Fdesign-patterns\u002Fcreational","01.csharp\u002F04.architecture-best-practices\u002F02.design-patterns\u002Fcreational",[141,145],{"title":142,"path":143,"stem":144},"Singleton (Одинак)","\u002Fcsharp\u002Farchitecture-best-practices\u002Fdesign-patterns\u002Fcreational\u002Fsingleton","01.csharp\u002F04.architecture-best-practices\u002F02.design-patterns\u002Fcreational\u002F01.singleton",{"title":146,"path":147,"stem":148},"Builder (Будівельник)","\u002Fcsharp\u002Farchitecture-best-practices\u002Fdesign-patterns\u002Fcreational\u002Fbuilder","01.csharp\u002F04.architecture-best-practices\u002F02.design-patterns\u002Fcreational\u002F02.builder",{"title":150,"path":151,"stem":152},"Building Professional CLIs","\u002Fcsharp\u002Farchitecture-best-practices\u002Fbuilding-professional-clis","01.csharp\u002F04.architecture-best-practices\u002F03.building-professional-clis",{"title":154,"path":155,"stem":156},"Validation & Flow Control","\u002Fcsharp\u002Farchitecture-best-practices\u002Fvalidation-flow-control","01.csharp\u002F04.architecture-best-practices\u002F04.validation-flow-control",{"title":158,"path":159,"stem":160},"The Modern .NET Host (Microsoft.Extensions)","\u002Fcsharp\u002Farchitecture-best-practices\u002Fmodern-dotnet-host","01.csharp\u002F04.architecture-best-practices\u002F05.modern-dotnet-host",{"title":162,"path":163,"stem":164},"Data Mapper: Repository та DAO патерни (Частина 1)","\u002Fcsharp\u002Farchitecture-best-practices\u002Fdata-mapper-part1","01.csharp\u002F04.architecture-best-practices\u002F06.data-mapper-part1",{"title":166,"path":167,"stem":168},"Data Mapper: Repository та DAO патерни (Частина 2)","\u002Fcsharp\u002Farchitecture-best-practices\u002Fdata-mapper-part2","01.csharp\u002F04.architecture-best-practices\u002F07.data-mapper-part2",{"title":170,"icon":132,"path":171,"stem":172,"children":173,"page":59},"Di Ioc","\u002Fcsharp\u002Farchitecture-best-practices\u002Fdi-ioc","01.csharp\u002F04.architecture-best-practices\u002F08.di-ioc",[174,178,182,186,190,194,198],{"title":175,"path":176,"stem":177},"Проблема залежностей та Інверсія Контролю","\u002Fcsharp\u002Farchitecture-best-practices\u002Fdi-ioc\u002Fthe-dependency-problem","01.csharp\u002F04.architecture-best-practices\u002F08.di-ioc\u002F01.the-dependency-problem",{"title":179,"path":180,"stem":181},"Будуємо власний Service Container","\u002Fcsharp\u002Farchitecture-best-practices\u002Fdi-ioc\u002Fbuild-your-own-container","01.csharp\u002F04.architecture-best-practices\u002F08.di-ioc\u002F02.build-your-own-container",{"title":183,"path":184,"stem":185},"Service Locator: Паттерн та Анти-паттерн","\u002Fcsharp\u002Farchitecture-best-practices\u002Fdi-ioc\u002Fservice-locator-pattern","01.csharp\u002F04.architecture-best-practices\u002F08.di-ioc\u002F03.service-locator-pattern",{"title":187,"path":188,"stem":189},"Паттерни Dependency Injection","\u002Fcsharp\u002Farchitecture-best-practices\u002Fdi-ioc\u002Fdependency-injection-patterns","01.csharp\u002F04.architecture-best-practices\u002F08.di-ioc\u002F04.dependency-injection-patterns",{"title":191,"path":192,"stem":193},"Microsoft DI: IServiceCollection та IServiceProvider","\u002Fcsharp\u002Farchitecture-best-practices\u002Fdi-ioc\u002Fmicrosoft-di-deep-dive","01.csharp\u002F04.architecture-best-practices\u002F08.di-ioc\u002F05.microsoft-di-deep-dive",{"title":195,"path":196,"stem":197},"Service Lifetimes та Scopes","\u002Fcsharp\u002Farchitecture-best-practices\u002Fdi-ioc\u002Fservice-lifetimes-and-scopes","01.csharp\u002F04.architecture-best-practices\u002F08.di-ioc\u002F06.service-lifetimes-and-scopes",{"title":199,"path":200,"stem":201},"DI Анти-паттерни та Найкращі Практики","\u002Fcsharp\u002Farchitecture-best-practices\u002Fdi-ioc\u002Fdi-anti-patterns-and-best-practices","01.csharp\u002F04.architecture-best-practices\u002F08.di-ioc\u002F07.di-anti-patterns-and-best-practices",{"title":203,"icon":132,"path":204,"stem":205,"children":206,"page":59},"Standard Library","\u002Fcsharp\u002Fstandard-library","01.csharp\u002F05.standard-library",[207,211,215],{"title":208,"path":209,"stem":210},"Collections (Колекції)","\u002Fcsharp\u002Fstandard-library\u002Fcollections","01.csharp\u002F05.standard-library\u002F01.collections",{"title":212,"path":213,"stem":214},"High Performance Types (Високопродуктивні Типи)","\u002Fcsharp\u002Fstandard-library\u002Fhigh-performance-types","01.csharp\u002F05.standard-library\u002F02.high-performance-types",{"title":216,"path":217,"stem":218},"LINQ (Language Integrated Query)","\u002Fcsharp\u002Fstandard-library\u002Flinq","01.csharp\u002F05.standard-library\u002F03.linq",{"title":220,"icon":221,"path":222,"stem":223,"children":224,"page":59},"System Internals Concurrency","i-lucide-server","\u002Fcsharp\u002Fsystem-internals-concurrency","01.csharp\u002F06.system-internals-concurrency",[225,229,233,237,241,245,249],{"title":226,"path":227,"stem":228},"Memory Management","\u002Fcsharp\u002Fsystem-internals-concurrency\u002Fmemory-management","01.csharp\u002F06.system-internals-concurrency\u002F01.memory-management",{"title":230,"path":231,"stem":232},"Reflection API: System.Type та Метадані","\u002Fcsharp\u002Fsystem-internals-concurrency\u002Freflection-fundamentals","01.csharp\u002F06.system-internals-concurrency\u002F02.reflection-fundamentals",{"title":234,"path":235,"stem":236},"Attributes та Dynamic Language Runtime","\u002Fcsharp\u002Fsystem-internals-concurrency\u002Fattributes-dynamic","01.csharp\u002F06.system-internals-concurrency\u002F03.attributes-dynamic",{"title":238,"path":239,"stem":240},"Expression Trees: Швидка Альтернатива Рефлексії","\u002Fcsharp\u002Fsystem-internals-concurrency\u002Fexpression-trees-compiled","01.csharp\u002F06.system-internals-concurrency\u002F04.expression-trees-compiled",{"title":242,"path":243,"stem":244},"Source Generators: Compile-Time Code Generation","\u002Fcsharp\u002Fsystem-internals-concurrency\u002Fsource-generators","01.csharp\u002F06.system-internals-concurrency\u002F05.source-generators",{"title":246,"path":247,"stem":248},"Multithreading Fundamentals","\u002Fcsharp\u002Fsystem-internals-concurrency\u002Fmultithreading-fundamentals","01.csharp\u002F06.system-internals-concurrency\u002F06.multithreading-fundamentals",{"title":250,"path":251,"stem":252},"Synchronization Primitives","\u002Fcsharp\u002Fsystem-internals-concurrency\u002Fsynchronization-primitives","01.csharp\u002F06.system-internals-concurrency\u002F07.synchronization-primitives",{"title":254,"icon":255,"path":256,"stem":257,"children":258,"page":59},"System Programming Windows","i-lucide-cpu","\u002Fcsharp\u002Fsystem-programming-windows","01.csharp\u002F07.system-programming-windows",[259,263,267,271,275,279,283,287,291,295,299,303,307,311,315,319,323,327,331,335,339,343,347,351,355,359,363,367,371,375],{"title":260,"path":261,"stem":262},"Як Працює Операційна Система","\u002Fcsharp\u002Fsystem-programming-windows\u002Fhow-os-works","01.csharp\u002F07.system-programming-windows\u002F01.how-os-works",{"title":264,"path":265,"stem":266},"Процеси в .NET — API та Запуск","\u002Fcsharp\u002Fsystem-programming-windows\u002Fprocesses-in-dotnet","01.csharp\u002F07.system-programming-windows\u002F02.processes-in-dotnet",{"title":268,"path":269,"stem":270},"Процеси в .NET — IPC та Міжпроцесна Комунікація","\u002Fcsharp\u002Fsystem-programming-windows\u002F02a.processes-ipc","01.csharp\u002F07.system-programming-windows\u002F02a.processes-ipc",{"title":272,"path":273,"stem":274},"Application Domains та Збірки — AppDomain і AssemblyLoadContext","\u002Fcsharp\u002Fsystem-programming-windows\u002Fappdomains-assemblies","01.csharp\u002F07.system-programming-windows\u002F03.appdomains-assemblies",{"title":276,"path":277,"stem":278},"Application Domains та Збірки — Plug-in Система з Hot-Reload","\u002Fcsharp\u002Fsystem-programming-windows\u002F03a.appdomains-plugin-system","01.csharp\u002F07.system-programming-windows\u002F03a.appdomains-plugin-system",{"title":280,"path":281,"stem":282},"Потоки — Основи та API Thread","\u002Fcsharp\u002Fsystem-programming-windows\u002Fthread-fundamentals","01.csharp\u002F07.system-programming-windows\u002F04.thread-fundamentals",{"title":284,"path":285,"stem":286},"Потоки — Lifecycle, Пріоритети та Безпечне Завершення","\u002Fcsharp\u002Fsystem-programming-windows\u002F04a.thread-lifecycle-priorities","01.csharp\u002F07.system-programming-windows\u002F04a.thread-lifecycle-priorities",{"title":288,"path":289,"stem":290},"Проблеми Спільного Стану — Race Condition та Data Race","\u002Fcsharp\u002Fsystem-programming-windows\u002Fshared-state-problems","01.csharp\u002F07.system-programming-windows\u002F05.shared-state-problems",{"title":292,"path":293,"stem":294},"Проблеми Спільного Стану — Memory Model та volatile","\u002Fcsharp\u002Fsystem-programming-windows\u002F05a.shared-state-memory-model","01.csharp\u002F07.system-programming-windows\u002F05a.shared-state-memory-model",{"title":296,"path":297,"stem":298},"Синхронізація — Monitor, lock та еволюція примітивів","\u002Fcsharp\u002Fsystem-programming-windows\u002Fsynchronization-fundamentals","01.csharp\u002F07.system-programming-windows\u002F06.synchronization-fundamentals",{"title":300,"path":301,"stem":302},"Синхронізація — Наскрізний Приклад та Deadlock Detection","\u002Fcsharp\u002Fsystem-programming-windows\u002F06a.synchronization-walkthrough","01.csharp\u002F07.system-programming-windows\u002F06a.synchronization-walkthrough",{"title":304,"path":305,"stem":306},"Синхронізація — Mutex, Semaphore та Event-Based Primitives","\u002Fcsharp\u002Fsystem-programming-windows\u002Fsynchronization-advanced","01.csharp\u002F07.system-programming-windows\u002F07.synchronization-advanced",{"title":308,"path":309,"stem":310},"Синхронізація — Interlocked, Volatile та Lock-Free Структури","\u002Fcsharp\u002Fsystem-programming-windows\u002F07a.synchronization-advanced-walkthrough","01.csharp\u002F07.system-programming-windows\u002F07a.synchronization-advanced-walkthrough",{"title":312,"path":313,"stem":314},"Interlocked, CAS та Lock-Free Структури","\u002Fcsharp\u002Fsystem-programming-windows\u002Finterlocked-cas-lockfree","01.csharp\u002F07.system-programming-windows\u002F08.interlocked-cas-lockfree",{"title":316,"path":317,"stem":318},"Volatile, Memory Model та Spinning","\u002Fcsharp\u002Fsystem-programming-windows\u002F08a.volatile-memory-model","01.csharp\u002F07.system-programming-windows\u002F08a.volatile-memory-model",{"title":320,"path":321,"stem":322},"ThreadPool — Пул Потоків для Ефективного Виконання","\u002Fcsharp\u002Fsystem-programming-windows\u002Fthread-pool","01.csharp\u002F07.system-programming-windows\u002F09.thread-pool",{"title":324,"path":325,"stem":326},"ThreadPool — Просунуті Сценарії та Внутрішня Будова","\u002Fcsharp\u002Fsystem-programming-windows\u002F09a.thread-pool-advanced","01.csharp\u002F07.system-programming-windows\u002F09a.thread-pool-advanced",{"title":328,"path":329,"stem":330},"Concurrent та Immutable Collections","\u002Fcsharp\u002Fsystem-programming-windows\u002Fconcurrent-collections","01.csharp\u002F07.system-programming-windows\u002F10.concurrent-collections",{"title":332,"path":333,"stem":334},"TPL, Task та Композиція — Від Thread до Task","\u002Fcsharp\u002Fsystem-programming-windows\u002Ftpl-parallel-plinq","01.csharp\u002F07.system-programming-windows\u002F11.tpl-parallel-plinq",{"title":336,"path":337,"stem":338},"Parallel Class та PLINQ — Data Parallelism","\u002Fcsharp\u002Fsystem-programming-windows\u002F11a.tpl-parallel-plinq-advanced","01.csharp\u002F07.system-programming-windows\u002F11a.tpl-parallel-plinq-advanced",{"title":340,"path":341,"stem":342},"Async\u002FAwait — Фундамент Асинхронного Програмування","\u002Fcsharp\u002Fsystem-programming-windows\u002Fasync-fundamentals","01.csharp\u002F07.system-programming-windows\u002F12.async-fundamentals",{"title":344,"path":345,"stem":346},"SynchronizationContext та ConfigureAwait — Контекст Виконання","\u002Fcsharp\u002Fsystem-programming-windows\u002Fasync-context-configureawait","01.csharp\u002F07.system-programming-windows\u002F13.async-context-configureawait",{"title":348,"path":349,"stem":350},"Async — Просунуті Паттерни","\u002Fcsharp\u002Fsystem-programming-windows\u002Fasync-advanced","01.csharp\u002F07.system-programming-windows\u002F14.async-advanced",{"title":352,"path":353,"stem":354},"System.Threading.Channels — Async Producer-Consumer","\u002Fcsharp\u002Fsystem-programming-windows\u002Fchannels","01.csharp\u002F07.system-programming-windows\u002F15.channels",{"title":356,"path":357,"stem":358},"Асинхронна Синхронізація","\u002Fcsharp\u002Fsystem-programming-windows\u002Fasync-synchronization","01.csharp\u002F07.system-programming-windows\u002F16.async-synchronization",{"title":360,"path":361,"stem":362},"Unsafe Code та Вказівники","\u002Fcsharp\u002Fsystem-programming-windows\u002Funsafe-code","01.csharp\u002F07.system-programming-windows\u002F17.unsafe-code",{"title":364,"path":365,"stem":366},"P\u002FInvoke та Windows API — Міст між .NET та Native Code","\u002Fcsharp\u002Fsystem-programming-windows\u002Fpinvoke-winapi","01.csharp\u002F07.system-programming-windows\u002F18.pinvoke-winapi",{"title":368,"path":369,"stem":370},"Реєстр Windows — Центральна База Конфігурації Системи","\u002Fcsharp\u002Fsystem-programming-windows\u002Fwindows-registry","01.csharp\u002F07.system-programming-windows\u002F19.windows-registry",{"title":372,"path":373,"stem":374},"Windows Hooks, Hotkeys та Services — Глибока Інтеграція з ОС","\u002Fcsharp\u002Fsystem-programming-windows\u002Fwindows-hooks-services","01.csharp\u002F07.system-programming-windows\u002F20.windows-hooks-services",{"title":376,"path":377,"stem":378},"Системне Програмування C# (Windows) — 07.system-programming-windows","\u002Fcsharp\u002Fsystem-programming-windows\u002Fimplementation_plan","01.csharp\u002F07.system-programming-windows\u002Fimplementation_plan",{"title":380,"icon":132,"path":381,"stem":382,"children":383,"page":59},"Io","\u002Fcsharp\u002Fio","01.csharp\u002F08.io",[384,388,392,396,400],{"title":385,"path":386,"stem":387},"8.1.1. Основи роботи з файловою системою","\u002Fcsharp\u002Fio\u002Ffile-system-basics","01.csharp\u002F08.io\u002F01.file-system-basics",{"title":389,"path":390,"stem":391},"8.1.2. Потоки (Streams) та Серіалізація Даних","\u002Fcsharp\u002Fio\u002Fstreams-serialization","01.csharp\u002F08.io\u002F02.streams-serialization",{"title":393,"path":394,"stem":395},"8.2.1. JSON Serialization з System.Text.Json","\u002Fcsharp\u002Fio\u002Fjson-serialization","01.csharp\u002F08.io\u002F03.json-serialization",{"title":397,"path":398,"stem":399},"8.2.2. XML Serialization та LINQ to XML","\u002Fcsharp\u002Fio\u002Fxml-serialization","01.csharp\u002F08.io\u002F04.xml-serialization",{"title":401,"path":402,"stem":403},"8.2.3. Binary Serialization: MessagePack та Protocol Buffers","\u002Fcsharp\u002Fio\u002Fbinary-serialization","01.csharp\u002F08.io\u002F05.binary-serialization",{"title":405,"icon":132,"path":406,"stem":407,"children":408,"page":59},"Ado Net","\u002Fcsharp\u002Fado-net","01.csharp\u002F09.ado-net",[409,413,417,421,425,429,433,437,441,445,449,453],{"title":410,"path":411,"stem":412},"9.1. Введення в ADO.NET","\u002Fcsharp\u002Fado-net\u002Fintroduction-to-adonet","01.csharp\u002F09.ado-net\u002F01.introduction-to-adonet",{"title":414,"path":415,"stem":416},"9.2. Клас DbConnection — з'єднання з базою даних","\u002Fcsharp\u002Fado-net\u002Fconnection","01.csharp\u002F09.ado-net\u002F02.connection",{"title":418,"path":419,"stem":420},"9.3. Клас DbCommand — виконання SQL-запитів","\u002Fcsharp\u002Fado-net\u002Fcommand-and-queries","01.csharp\u002F09.ado-net\u002F03.command-and-queries",{"title":422,"path":423,"stem":424},"9.4. Клас DbDataReader — ефективне читання даних","\u002Fcsharp\u002Fado-net\u002Fdatareader","01.csharp\u002F09.ado-net\u002F04.datareader",{"title":426,"path":427,"stem":428},"9.5. Параметризовані запити та захист від SQL Injection","\u002Fcsharp\u002Fado-net\u002Fparameters-and-sql-injection","01.csharp\u002F09.ado-net\u002F05.parameters-and-sql-injection",{"title":430,"path":431,"stem":432},"9.6. Транзакції в ADO.NET","\u002Fcsharp\u002Fado-net\u002Ftransactions","01.csharp\u002F09.ado-net\u002F06.transactions",{"title":434,"path":435,"stem":436},"9.7. DbProviderFactory — провайдер-незалежний код","\u002Fcsharp\u002Fado-net\u002Fprovider-factory","01.csharp\u002F09.ado-net\u002F07.provider-factory",{"title":438,"path":439,"stem":440},"9.8. Асинхронний доступ до даних","\u002Fcsharp\u002Fado-net\u002Fasync-data-access","01.csharp\u002F09.ado-net\u002F08.async-data-access",{"title":442,"path":443,"stem":444},"9.9. Від'єднаний режим: DataSet, DataTable, DataRow","\u002Fcsharp\u002Fado-net\u002Fdisconnected-mode-dataset","01.csharp\u002F09.ado-net\u002F09.disconnected-mode-dataset",{"title":446,"path":447,"stem":448},"9.10. DataAdapter — міст між DataSet та базою даних","\u002Fcsharp\u002Fado-net\u002Fdata-adapter","01.csharp\u002F09.ado-net\u002F10.data-adapter",{"title":450,"path":451,"stem":452},"9.11. Data Mapper та Repository: Архітектура доступу до даних","\u002Fcsharp\u002Fado-net\u002Fdata-mapper-repository","01.csharp\u002F09.ado-net\u002F11.data-mapper-repository",{"title":454,"path":455,"stem":456},"9.12. Identity Map, Unit of Work та Specification Pattern","\u002Fcsharp\u002Fado-net\u002Fadvanced-patterns","01.csharp\u002F09.ado-net\u002F12.advanced-patterns",{"title":458,"icon":255,"path":459,"stem":460,"children":461,"page":59},"Ef Core","\u002Fcsharp\u002Fef-core","01.csharp\u002F10.ef-core",[462,466,470,474,478,482,486,490,494,498,502,506,510,514,518,522,526,532,538,542,546,550,554,558,562,566,570,574,578,582,586,590,594,598,602,606,610,614,618,622,626,630,634,638,642,646],{"title":463,"path":464,"stem":465},"Що таке ORM? Від SQL до об'єктів","\u002Fcsharp\u002Fef-core\u002Fwhat-is-orm","01.csharp\u002F10.ef-core\u002F01.what-is-orm",{"title":467,"path":468,"stem":469},"Перший проєкт — від нуля до CRUD","\u002Fcsharp\u002Fef-core\u002Ffirst-project","01.csharp\u002F10.ef-core\u002F02.first-project",{"title":471,"path":472,"stem":473},"DbContext — Серце EF Core","\u002Fcsharp\u002Fef-core\u002Fdbcontext-deep-dive","01.csharp\u002F10.ef-core\u002F03.dbcontext-deep-dive",{"title":475,"path":476,"stem":477},"Провайдери баз даних — Архітектура та Вибір СУБД","\u002Fcsharp\u002Fef-core\u002Fdatabase-providers","01.csharp\u002F10.ef-core\u002F04.database-providers",{"title":479,"path":480,"stem":481},"Конвенції EF Core — Магія без конфігурації","\u002Fcsharp\u002Fef-core\u002Fconventions","01.csharp\u002F10.ef-core\u002F05.conventions",{"title":483,"path":484,"stem":485},"Fluent API та Data Annotations — Явна конфігурація моделі","\u002Fcsharp\u002Fef-core\u002Ffluent-api-vs-annotations","01.csharp\u002F10.ef-core\u002F06.fluent-api-vs-annotations",{"title":487,"path":488,"stem":489},"Зв'язки — One-to-One та One-to-Many","\u002Fcsharp\u002Fef-core\u002Frelationships-basics","01.csharp\u002F10.ef-core\u002F07.relationships-basics",{"title":491,"path":492,"stem":493},"Зв'язки Advanced — Many-to-Many та Складні Сценарії","\u002Fcsharp\u002Fef-core\u002Frelationships-advanced","01.csharp\u002F10.ef-core\u002F08.relationships-advanced",{"title":495,"path":496,"stem":497},"Властивості — Типи, Конвертери, Компаратори (Частина 1)","\u002Fcsharp\u002Fef-core\u002Fproperty-configuration-part1","01.csharp\u002F10.ef-core\u002F09.property-configuration-part1",{"title":499,"path":500,"stem":501},"Властивості — Value Comparers, Generators, Shadow Properties (Частина 2)","\u002Fcsharp\u002Fef-core\u002Fproperty-configuration-part2","01.csharp\u002F10.ef-core\u002F09.property-configuration-part2",{"title":503,"path":504,"stem":505},"Складні типи — Owned Types та Complex Types (Частина 1)","\u002Fcsharp\u002Fef-core\u002Fcomplex-types-owned-part1","01.csharp\u002F10.ef-core\u002F10.complex-types-owned-part1",{"title":507,"path":508,"stem":509},"Складні типи — Complex Types, Keyless Entities, Порівняння (Частина 2)","\u002Fcsharp\u002Fef-core\u002Fcomplex-types-owned-part2","01.csharp\u002F10.ef-core\u002F10.complex-types-owned-part2",{"title":511,"path":512,"stem":513},"JSON Columns — Складні дані у JSON (Частина 1)","\u002Fcsharp\u002Fef-core\u002Fjson-columns-part1","01.csharp\u002F10.ef-core\u002F11.json-columns-part1",{"title":515,"path":516,"stem":517},"JSON Columns — Value Comparers, Індекси, Провайдери (Частина 2)","\u002Fcsharp\u002Fef-core\u002Fjson-columns-part2","01.csharp\u002F10.ef-core\u002F11.json-columns-part2",{"title":519,"path":520,"stem":521},"Успадкування — Абстрактні класи та TPH (Частина 1)","\u002Fcsharp\u002Fef-core\u002Finheritance-part1","01.csharp\u002F10.ef-core\u002F12.inheritance-part1",{"title":523,"path":524,"stem":525},"Успадкування — TPT, TPC та Порівняння Стратегій (Частина 2)","\u002Fcsharp\u002Fef-core\u002Finheritance-part2","01.csharp\u002F10.ef-core\u002F12.inheritance-part2",{"title":527,"path":528,"stem":529,"children":530},"Індекси, Обмеження та Схема (Частина 1)","\u002Fcsharp\u002Fef-core\u002Findexes-constraints-part1","01.csharp\u002F10.ef-core\u002F13.indexes-constraints-part1",[531],{"title":527,"path":528,"stem":529},{"title":533,"path":534,"stem":535,"children":536},"Індекси, Обмеження та Схема (Частина 2)","\u002Fcsharp\u002Fef-core\u002Findexes-constraints-part2","01.csharp\u002F10.ef-core\u002F13.indexes-constraints-part2",[537],{"title":533,"path":534,"stem":535},{"title":539,"path":540,"stem":541},"Seed Data — Початкові Дані (Частина 1)","\u002Fcsharp\u002Fef-core\u002Fseeding-part1","01.csharp\u002F10.ef-core\u002F14.seeding-part1",{"title":543,"path":544,"stem":545},"Seed Data — SQL-скрипти, Bogus та Стратегії (Частина 2)","\u002Fcsharp\u002Fef-core\u002Fseeding-part2","01.csharp\u002F10.ef-core\u002F14.seeding-part2",{"title":547,"path":548,"stem":549},"Global Query Filters — Глобальні Фільтри (Частина 1)","\u002Fcsharp\u002Fef-core\u002Fglobal-query-filters-part1","01.csharp\u002F10.ef-core\u002F15.global-query-filters-part1",{"title":551,"path":552,"stem":553},"Global Query Filters — Підводні камені та Інтеграція (Частина 2)","\u002Fcsharp\u002Fef-core\u002Fglobal-query-filters-part2","01.csharp\u002F10.ef-core\u002F15.global-query-filters-part2",{"title":555,"path":556,"stem":557},"LINQ-запити в EF Core (Частина 1)","\u002Fcsharp\u002Fef-core\u002Flinq-queries-part1","01.csharp\u002F10.ef-core\u002F16.linq-queries-part1",{"title":559,"path":560,"stem":561},"LINQ-запити в EF Core (Частина 2)","\u002Fcsharp\u002Fef-core\u002Flinq-queries-part2","01.csharp\u002F10.ef-core\u002F16.linq-queries-part2",{"title":563,"path":564,"stem":565},"Завантаження Пов'язаних Даних (Частина 1)","\u002Fcsharp\u002Fef-core\u002Floading-related-data-part1","01.csharp\u002F10.ef-core\u002F17.loading-related-data-part1",{"title":567,"path":568,"stem":569},"Завантаження Пов'язаних Даних (Частина 2)","\u002Fcsharp\u002Fef-core\u002Floading-related-data-part2","01.csharp\u002F10.ef-core\u002F17.loading-related-data-part2",{"title":571,"path":572,"stem":573},"Raw SQL, Views та Stored Procedures (Частина 1)","\u002Fcsharp\u002Fef-core\u002Fraw-sql-part1","01.csharp\u002F10.ef-core\u002F18.raw-sql-part1",{"title":575,"path":576,"stem":577},"Raw SQL — Stored Procedures, DbFunction та Bulk Operations (Частина 2)","\u002Fcsharp\u002Fef-core\u002Fraw-sql-part2","01.csharp\u002F10.ef-core\u002F18.raw-sql-part2",{"title":579,"path":580,"stem":581},"Продвинуті Запити — Compiled Queries, Bulk та Оптимізація (Частина 1)","\u002Fcsharp\u002Fef-core\u002Fadvanced-queries-part1","01.csharp\u002F10.ef-core\u002F19.advanced-queries-part1",{"title":583,"path":584,"stem":585},"Продвинуті Запити — Query Tags, Bulk та Interceptors (Частина 2)","\u002Fcsharp\u002Fef-core\u002Fadvanced-queries-part2","01.csharp\u002F10.ef-core\u002F19.advanced-queries-part2",{"title":587,"path":588,"stem":589},"Change Tracker — Відстеження Змін (Частина 1)","\u002Fcsharp\u002Fef-core\u002Fchange-tracking-part1","01.csharp\u002F10.ef-core\u002F20.change-tracking-part1",{"title":591,"path":592,"stem":593},"Change Tracker — Графи Об'єктів та Disconnected (Частина 2)","\u002Fcsharp\u002Fef-core\u002Fchange-tracking-part2","01.csharp\u002F10.ef-core\u002F20.change-tracking-part2",{"title":595,"path":596,"stem":597},"Збереження Даних та Транзакції (Частина 1)","\u002Fcsharp\u002Fef-core\u002Fsaving-data-part1","01.csharp\u002F10.ef-core\u002F21.saving-data-part1",{"title":599,"path":600,"stem":601},"Збереження Даних — Concurrency та Outbox (Частина 2)","\u002Fcsharp\u002Fef-core\u002Fsaving-data-part2","01.csharp\u002F10.ef-core\u002F21.saving-data-part2",{"title":603,"path":604,"stem":605},"Конкурентність та Блокування (Частина 1)","\u002Fcsharp\u002Fef-core\u002Fconcurrency-part1","01.csharp\u002F10.ef-core\u002F22.concurrency-part1",{"title":607,"path":608,"stem":609},"Конкурентність — Дедлоки та Queue Processing (Частина 2)","\u002Fcsharp\u002Fef-core\u002Fconcurrency-part2","01.csharp\u002F10.ef-core\u002F22.concurrency-part2",{"title":611,"path":612,"stem":613},"Міграції в EF Core — Основи (Частина 1)","\u002Fcsharp\u002Fef-core\u002Fmigrations-basics-part1","01.csharp\u002F10.ef-core\u002F23.migrations-basics-part1",{"title":615,"path":616,"stem":617},"Міграції в EF Core — Основи (Частина 2)","\u002Fcsharp\u002Fef-core\u002Fmigrations-basics-part2","01.csharp\u002F10.ef-core\u002F23.migrations-basics-part2",{"title":619,"path":620,"stem":621},"Міграції — Просунуті Сценарії (Частина 1)","\u002Fcsharp\u002Fef-core\u002Fmigrations-advanced-part1","01.csharp\u002F10.ef-core\u002F24.migrations-advanced-part1",{"title":623,"path":624,"stem":625},"Міграції — Просунуті Сценарії (Частина 2)","\u002Fcsharp\u002Fef-core\u002Fmigrations-advanced-part2","01.csharp\u002F10.ef-core\u002F24.migrations-advanced-part2",{"title":627,"path":628,"stem":629},"Управління Схемою та Database-First (Частина 1)","\u002Fcsharp\u002Fef-core\u002Fschema-management-part1","01.csharp\u002F10.ef-core\u002F25.schema-management-part1",{"title":631,"path":632,"stem":633},"Управління Схемою та Database-First (Частина 2)","\u002Fcsharp\u002Fef-core\u002Fschema-management-part2","01.csharp\u002F10.ef-core\u002F25.schema-management-part2",{"title":635,"path":636,"stem":637},"Продуктивність EF Core — Основи (Частина 1)","\u002Fcsharp\u002Fef-core\u002Fperformance-fundamentals-part1","01.csharp\u002F10.ef-core\u002F26.performance-fundamentals-part1",{"title":639,"path":640,"stem":641},"Interceptors в EF Core (Частина 1)","\u002Fcsharp\u002Fef-core\u002Finterceptors-part1","01.csharp\u002F10.ef-core\u002F29.interceptors-part1",{"title":643,"path":644,"stem":645},"Interceptors в EF Core — Connection, Transaction та Materialization (Частина 2)","\u002Fcsharp\u002Fef-core\u002Finterceptors-part2","01.csharp\u002F10.ef-core\u002F29.interceptors-part2",{"title":647,"path":648,"stem":649},"План вивчення Entity Framework Core — Повний курс","\u002Fcsharp\u002Fef-core\u002Fimplementation_plan","01.csharp\u002F10.ef-core\u002Fimplementation_plan",{"title":651,"icon":652,"path":653,"stem":654,"children":655,"page":59},"ASP.NET","i-devicon-dotnetcore","\u002Fcsharp\u002Faspnet","01.csharp\u002F11.aspnet",[656,730,791,869,927,941,967,1057,1111,1182,1212,1289,1346],{"title":657,"icon":658,"path":659,"stem":660,"children":661,"page":59},"Minimal API","i-lucide-network","\u002Fcsharp\u002Faspnet\u002Fminimal-api","01.csharp\u002F11.aspnet\u002F01.minimal-api",[662,666,670,674,678,682,686,690,694,698,702,706,710,714,718,722,726],{"title":663,"path":664,"stem":665},"Вступ до ASP.NET та еволюція фреймворку","\u002Fcsharp\u002Faspnet\u002Fminimal-api\u002Fintroduction","01.csharp\u002F11.aspnet\u002F01.minimal-api\u002F01.introduction",{"title":667,"path":668,"stem":669},"Перший додаток на ASP.NET Core","\u002Fcsharp\u002Faspnet\u002Fminimal-api\u002Ffirst-application","01.csharp\u002F11.aspnet\u002F01.minimal-api\u002F02.first-application",{"title":671,"path":672,"stem":673},"WebApplication, Builder та Dependency Injection","\u002Fcsharp\u002Faspnet\u002Fminimal-api\u002Fwebapplication-builder","01.csharp\u002F11.aspnet\u002F01.minimal-api\u002F03.webapplication-builder",{"title":675,"path":676,"stem":677},"Конвеєр запитів та Middleware","\u002Fcsharp\u002Faspnet\u002Fminimal-api\u002Frequest-pipeline-middleware","01.csharp\u002F11.aspnet\u002F01.minimal-api\u002F04.request-pipeline-middleware",{"title":679,"path":680,"stem":681},"Маршрутизація в ASP.NET Core: Основи","\u002Fcsharp\u002Faspnet\u002Fminimal-api\u002Frouting-basics","01.csharp\u002F11.aspnet\u002F01.minimal-api\u002F05.routing-basics",{"title":683,"path":684,"stem":685},"Маршрутизація в ASP.NET Core: Розширені можливості","\u002Fcsharp\u002Faspnet\u002Fminimal-api\u002Frouting-advanced","01.csharp\u002F11.aspnet\u002F01.minimal-api\u002F06.routing-advanced",{"title":687,"path":688,"stem":689},"Статичні файли в ASP.NET Core","\u002Fcsharp\u002Faspnet\u002Fminimal-api\u002Fstatic-files","01.csharp\u002F11.aspnet\u002F01.minimal-api\u002F07.static-files",{"title":691,"path":692,"stem":693},"Статичні Активи: MapStaticAssets (ASP.NET Core 9.0)","\u002Fcsharp\u002Faspnet\u002Fminimal-api\u002Fstatic-assets","01.csharp\u002F11.aspnet\u002F01.minimal-api\u002F08.static-assets",{"title":695,"path":696,"stem":697},"Конфігурація в ASP.NET Core: Основи","\u002Fcsharp\u002Faspnet\u002Fminimal-api\u002Fconfiguration-fundamentals","01.csharp\u002F11.aspnet\u002F01.minimal-api\u002F09.configuration-fundamentals",{"title":699,"path":700,"stem":701},"Конфігурація: Паттерн Options","\u002Fcsharp\u002Faspnet\u002Fminimal-api\u002Fconfiguration-options","01.csharp\u002F11.aspnet\u002F01.minimal-api\u002F10.configuration-options",{"title":703,"path":704,"stem":705},"Логування в ASP.NET Core: Основи","\u002Fcsharp\u002Faspnet\u002Fminimal-api\u002Flogging-basics","01.csharp\u002F11.aspnet\u002F01.minimal-api\u002F11.logging-basics",{"title":707,"path":708,"stem":709},"Логування: Serilog та Middleware","\u002Fcsharp\u002Faspnet\u002Fminimal-api\u002Flogging-advanced","01.csharp\u002F11.aspnet\u002F01.minimal-api\u002F12.logging-advanced",{"title":711,"path":712,"stem":713},"Управління станом: HttpContext.Items та Cookies","\u002Fcsharp\u002Faspnet\u002Fminimal-api\u002Fstate-management","01.csharp\u002F11.aspnet\u002F01.minimal-api\u002F13.state-management",{"title":715,"path":716,"stem":717},"Стан сесії: Sessions","\u002Fcsharp\u002Faspnet\u002Fminimal-api\u002Fsession-state","01.csharp\u002F11.aspnet\u002F01.minimal-api\u002F14.session-state",{"title":719,"path":720,"stem":721},"Структура проєкту: від хаосу до архітектури","\u002Fcsharp\u002Faspnet\u002Fminimal-api\u002Fproject-structure","01.csharp\u002F11.aspnet\u002F01.minimal-api\u002F15.project-structure",{"title":723,"path":724,"stem":725},"Scalar у Minimal API: повний проєкт і Fluent OpenAPI","\u002Fcsharp\u002Faspnet\u002Fminimal-api\u002Fscalar-openapi-fluent","01.csharp\u002F11.aspnet\u002F01.minimal-api\u002F16.scalar-openapi-fluent",{"title":727,"path":728,"stem":729},"Swagger \u002F Swashbuckle у Minimal API: окремий класичний шлях","\u002Fcsharp\u002Faspnet\u002Fminimal-api\u002Fswagger-swashbuckle","01.csharp\u002F11.aspnet\u002F01.minimal-api\u002F17.swagger-swashbuckle",{"title":731,"icon":658,"path":732,"stem":733,"children":734,"page":59},"API","\u002Fcsharp\u002Faspnet\u002Fapi","01.csharp\u002F11.aspnet\u002F02.api",[735,739,743,747,751,755,759,763,767,771,775,779,783,787],{"title":736,"path":737,"stem":738},"Що таке API. Клієнт-серверна архітектура","\u002Fcsharp\u002Faspnet\u002Fapi\u002Fwhat-is-api","01.csharp\u002F11.aspnet\u002F02.api\u002F01.what-is-api",{"title":740,"path":741,"stem":742},"Формати даних: JSON, XML, TOML та бінарні формати","\u002Fcsharp\u002Faspnet\u002Fapi\u002Fdata-formats","01.csharp\u002F11.aspnet\u002F02.api\u002F02.data-formats",{"title":744,"path":745,"stem":746},"Парадигми API та концепція REST","\u002Fcsharp\u002Faspnet\u002Fapi\u002Fapi-paradigms-rest","01.csharp\u002F11.aspnet\u002F02.api\u002F03.api-paradigms-rest",{"title":748,"path":749,"stem":750},"HTTP-методи, статус-коди та заголовки","\u002Fcsharp\u002Faspnet\u002Fapi\u002Fhttp-methods-status-codes","01.csharp\u002F11.aspnet\u002F02.api\u002F04.http-methods-status-codes",{"title":752,"path":753,"stem":754},"Організація HTTP API за принципами REST","\u002Fcsharp\u002Faspnet\u002Fapi\u002Frest-organizing","01.csharp\u002F11.aspnet\u002F02.api\u002F05.rest-organizing",{"title":756,"path":757,"stem":758},"Номенклатура URL та CRUD-операції","\u002Fcsharp\u002Faspnet\u002Fapi\u002Furl-nomenclature-crud","01.csharp\u002F11.aspnet\u002F02.api\u002F06.url-nomenclature-crud",{"title":760,"path":761,"stem":762},"Правила дизайну: іменування та стандарти","\u002Fcsharp\u002Faspnet\u002Fapi\u002Fapi-design-naming","01.csharp\u002F11.aspnet\u002F02.api\u002F07.api-design-naming",{"title":764,"path":765,"stem":766},"Валідація, ліміти та обробка помилок","\u002Fcsharp\u002Faspnet\u002Fapi\u002Fapi-design-validation","01.csharp\u002F11.aspnet\u002F02.api\u002F08.api-design-validation",{"title":768,"path":769,"stem":770},"Обробка помилок у Minimal API","\u002Fcsharp\u002Faspnet\u002Fapi\u002Ferror-handling-http","01.csharp\u002F11.aspnet\u002F02.api\u002F09.error-handling-http",{"title":772,"path":773,"stem":774},"Ідемпотентність та синхронізація стану","\u002Fcsharp\u002Faspnet\u002Fapi\u002Fidempotency-sync","01.csharp\u002F11.aspnet\u002F02.api\u002F10.idempotency-sync",{"title":776,"path":777,"stem":778},"Пагінація та організація списків","\u002Fcsharp\u002Faspnet\u002Fapi\u002Fpagination-lists","01.csharp\u002F11.aspnet\u002F02.api\u002F11.pagination-lists",{"title":780,"path":781,"stem":782},"Безпека API, кешування та інтернаціоналізація","\u002Fcsharp\u002Faspnet\u002Fapi\u002Fsecurity-auth","01.csharp\u002F11.aspnet\u002F02.api\u002F12.security-auth",{"title":784,"path":785,"stem":786},"Процес проєктування API та документування","\u002Fcsharp\u002Faspnet\u002Fapi\u002Fapi-design-process","01.csharp\u002F11.aspnet\u002F02.api\u002F13.api-design-process",{"title":788,"path":789,"stem":790},"OpenAPI: контракт, специфікація та документація API","\u002Fcsharp\u002Faspnet\u002Fapi\u002Fopenapi","01.csharp\u002F11.aspnet\u002F02.api\u002F14.openapi",{"title":792,"icon":793,"path":794,"stem":795,"children":796,"page":59},"Auth","i-lucide-shield-check","\u002Fcsharp\u002Faspnet\u002Fauth","01.csharp\u002F11.aspnet\u002F03.auth",[797,801,805,809,813,817,821,825,829,833,837,841,845,849,853,857,861,865],{"title":798,"path":799,"stem":800},"Основи аутентифікації та авторизації","\u002Fcsharp\u002Faspnet\u002Fauth\u002Fauth-fundamentals","01.csharp\u002F11.aspnet\u002F03.auth\u002F01.auth-fundamentals",{"title":802,"path":803,"stem":804},"JWT-аутентифікація","\u002Fcsharp\u002Faspnet\u002Fauth\u002Fjwt-authentication","01.csharp\u002F11.aspnet\u002F03.auth\u002F02.jwt-authentication",{"title":806,"path":807,"stem":808},"Авторизація: ролі, політики та resource-based доступ","\u002Fcsharp\u002Faspnet\u002Fauth\u002Fauthorization-policies","01.csharp\u002F11.aspnet\u002F03.auth\u002F03.authorization-policies",{"title":810,"path":811,"stem":812},"Cookie-аутентифікація та ASP.NET Core Identity","\u002Fcsharp\u002Faspnet\u002Fauth\u002Fcookie-auth-identity","01.csharp\u002F11.aspnet\u002F03.auth\u002F04.cookie-auth-identity",{"title":814,"path":815,"stem":816},"JWT + Refresh Tokens (HttpOnly Cookie)","\u002Fcsharp\u002Faspnet\u002Fauth\u002F04b.identity-auth-jwt","01.csharp\u002F11.aspnet\u002F03.auth\u002F04b.identity-auth-jwt",{"title":818,"path":819,"stem":820},"Identity: Підтвердження Email та Скидання Пароля","\u002Fcsharp\u002Faspnet\u002Fauth\u002Fidentity-email-confirmation","01.csharp\u002F11.aspnet\u002F03.auth\u002F05.identity-email-confirmation",{"title":822,"path":823,"stem":824},"Identity: Двофакторна Аутентифікація (2FA)","\u002Fcsharp\u002Faspnet\u002Fauth\u002Fidentity-two-factor","01.csharp\u002F11.aspnet\u002F03.auth\u002F06.identity-two-factor",{"title":826,"path":827,"stem":828},"Identity: Внутрішня Архітектура та Кастомізація","\u002Fcsharp\u002Faspnet\u002Fauth\u002Fidentity-internals","01.csharp\u002F11.aspnet\u002F03.auth\u002F07.identity-internals",{"title":830,"path":831,"stem":832},"OAuth 2.0 та зовнішні провайдери","\u002Fcsharp\u002Faspnet\u002Fauth\u002Foauth-external-providers","01.csharp\u002F11.aspnet\u002F03.auth\u002F08.oauth-external-providers",{"title":834,"path":835,"stem":836},"Безпека на практиці: CORS, HTTPS та захист від атак","\u002Fcsharp\u002Faspnet\u002Fauth\u002Fsecurity-hardening","01.csharp\u002F11.aspnet\u002F03.auth\u002F09.security-hardening",{"title":838,"path":839,"stem":840},"Теорія OAuth 2.0: Поняття, Аналогії та Флоу","\u002Fcsharp\u002Faspnet\u002Fauth\u002Foauth-theory","01.csharp\u002F11.aspnet\u002F03.auth\u002F10.oauth-theory",{"title":842,"path":843,"stem":844},"OIDC, OAuth 2.0 та Keycloak в ASP.NET Core","\u002Fcsharp\u002Faspnet\u002Fauth\u002Foidc-keycloak","01.csharp\u002F11.aspnet\u002F03.auth\u002F10.oidc-keycloak",{"title":846,"path":847,"stem":848},"API Keys аутентифікація в ASP.NET Core","\u002Fcsharp\u002Faspnet\u002Fauth\u002Fapi-keys","01.csharp\u002F11.aspnet\u002F03.auth\u002F11.api-keys",{"title":850,"path":851,"stem":852},"Rate Limiting та Throttling в ASP.NET Core","\u002Fcsharp\u002Faspnet\u002Fauth\u002Frate-limiting","01.csharp\u002F11.aspnet\u002F03.auth\u002F12.rate-limiting",{"title":854,"path":855,"stem":856},"Refresh Token Rotation в ASP.NET Core","\u002Fcsharp\u002Faspnet\u002Fauth\u002Frefresh-token-rotation","01.csharp\u002F11.aspnet\u002F03.auth\u002F13.refresh-token-rotation",{"title":858,"path":859,"stem":860},"Certificate Authentication та mTLS в ASP.NET Core","\u002Fcsharp\u002Faspnet\u002Fauth\u002Fcertificate-auth","01.csharp\u002F11.aspnet\u002F03.auth\u002F14.certificate-auth",{"title":862,"path":863,"stem":864},"RBAC, ABAC та ReBAC в ASP.NET Core","\u002Fcsharp\u002Faspnet\u002Fauth\u002Frbac-abac-rebac","01.csharp\u002F11.aspnet\u002F03.auth\u002F15.rbac-abac-rebac",{"title":866,"path":867,"stem":868},"Multi-tenancy та ізоляція даних в ASP.NET Core","\u002Fcsharp\u002Faspnet\u002Fauth\u002Fmulti-tenancy","01.csharp\u002F11.aspnet\u002F03.auth\u002F16.multi-tenancy",{"title":870,"icon":871,"path":872,"stem":873,"children":874,"page":59},"Нотифікації","i-lucide-bell","\u002Fcsharp\u002Faspnet\u002Fnotifications","01.csharp\u002F11.aspnet\u002F04.notifications",[875,879,883,887,891,895,899,903,907,911,915,919,923],{"title":876,"path":877,"stem":878},"In-App нотифікації через базу даних","\u002Fcsharp\u002Faspnet\u002Fnotifications\u002Fin-app-database-notifications","01.csharp\u002F11.aspnet\u002F04.notifications\u002F01.in-app-database-notifications",{"title":880,"path":881,"stem":882},"Polling: Регулярний запит оновлень","\u002Fcsharp\u002Faspnet\u002Fnotifications\u002Fpolling","01.csharp\u002F11.aspnet\u002F04.notifications\u002F02.polling",{"title":884,"path":885,"stem":886},"Server-Sent Events: Однострімовий push від сервера","\u002Fcsharp\u002Faspnet\u002Fnotifications\u002Fserver-sent-events","01.csharp\u002F11.aspnet\u002F04.notifications\u002F03.server-sent-events",{"title":888,"path":889,"stem":890},"WebSockets: Двостороннє з'єднання в реальному часі","\u002Fcsharp\u002Faspnet\u002Fnotifications\u002Fwebsockets","01.csharp\u002F11.aspnet\u002F04.notifications\u002F04.websockets",{"title":892,"path":893,"stem":894},"SignalR: Абстракція над транспортами реального часу","\u002Fcsharp\u002Faspnet\u002Fnotifications\u002Fsignalr","01.csharp\u002F11.aspnet\u002F04.notifications\u002F05.signalr",{"title":896,"path":897,"stem":898},"Background Services: Фонові задачі в ASP.NET Core","\u002Fcsharp\u002Faspnet\u002Fnotifications\u002Fbackground-services","01.csharp\u002F11.aspnet\u002F04.notifications\u002F06.background-services",{"title":900,"path":901,"stem":902},"Web Push нотифікації","\u002Fcsharp\u002Faspnet\u002Fnotifications\u002Fweb-push","01.csharp\u002F11.aspnet\u002F04.notifications\u002F07.web-push",{"title":904,"path":905,"stem":906},"Email нотифікації","\u002Fcsharp\u002Faspnet\u002Fnotifications\u002Femail-notifications","01.csharp\u002F11.aspnet\u002F04.notifications\u002F08.email-notifications",{"title":908,"path":909,"stem":910},"Порівняння підходів: Як вибрати правильну технологію нотифікацій","\u002Fcsharp\u002Faspnet\u002Fnotifications\u002Fchoosing-the-right-approach","01.csharp\u002F11.aspnet\u002F04.notifications\u002F09.choosing-the-right-approach",{"title":912,"path":913,"stem":914},"Hangfire: Надійне планування фонових задач","\u002Fcsharp\u002Faspnet\u002Fnotifications\u002Fhangfire","01.csharp\u002F11.aspnet\u002F04.notifications\u002F10.hangfire",{"title":916,"path":917,"stem":918},"Практика: Конвертація зображень у WebP через Hangfire","\u002Fcsharp\u002Faspnet\u002Fnotifications\u002Fhangfire-image-webp","01.csharp\u002F11.aspnet\u002F04.notifications\u002F11.hangfire-image-webp",{"title":920,"path":921,"stem":922},"Практика: Підготовка відео до HLS-стрімінгу через Hangfire","\u002Fcsharp\u002Faspnet\u002Fnotifications\u002Fhangfire-video-hls","01.csharp\u002F11.aspnet\u002F04.notifications\u002F12.hangfire-video-hls",{"title":924,"path":925,"stem":926},"Telegram-нотифікації: від одного повідомлення до масових розсилок і мульти-канального підходу","\u002Fcsharp\u002Faspnet\u002Fnotifications\u002Ftelegram-notifications","01.csharp\u002F11.aspnet\u002F04.notifications\u002F13.telegram-notifications",{"title":928,"icon":929,"path":930,"stem":931,"children":932,"page":59},"Інтернаціоналізація","i-lucide-languages","\u002Fcsharp\u002Faspnet\u002Fi18n","01.csharp\u002F11.aspnet\u002F05.i18n",[933,937],{"title":934,"path":935,"stem":936},"Інтернаціоналізація (i18n) у Minimal API: від A до Я","\u002Fcsharp\u002Faspnet\u002Fi18n\u002Finternationalization","01.csharp\u002F11.aspnet\u002F05.i18n\u002F01.internationalization",{"title":938,"path":939,"stem":940},"Humanizer: людиномовні рядки у .NET","\u002Fcsharp\u002Faspnet\u002Fi18n\u002Fhumanizer","01.csharp\u002F11.aspnet\u002F05.i18n\u002F02.humanizer",{"title":942,"icon":943,"path":944,"stem":945,"children":946,"page":59},"Кешування","i-lucide-layers","\u002Fcsharp\u002Faspnet\u002Fcaching","01.csharp\u002F11.aspnet\u002F06.caching",[947,951,955,959,963],{"title":948,"path":949,"stem":950},"Огляд кешування: чотири рівні і коли що обирати","\u002Fcsharp\u002Faspnet\u002Fcaching\u002Fcaching","01.csharp\u002F11.aspnet\u002F06.caching\u002F01.caching",{"title":952,"path":953,"stem":954},"IMemoryCache: кеш в оперативній пам'яті","\u002Fcsharp\u002Faspnet\u002Fcaching\u002Fmemory-cache","01.csharp\u002F11.aspnet\u002F06.caching\u002F02.memory-cache",{"title":956,"path":957,"stem":958},"IDistributedCache і Redis: розподілений кеш","\u002Fcsharp\u002Faspnet\u002Fcaching\u002Fdistributed-cache","01.csharp\u002F11.aspnet\u002F06.caching\u002F03.distributed-cache",{"title":960,"path":961,"stem":962},"Response Cache: HTTP-кешування через Cache-Control","\u002Fcsharp\u002Faspnet\u002Fcaching\u002Fresponse-cache","01.csharp\u002F11.aspnet\u002F06.caching\u002F04.response-cache",{"title":964,"path":965,"stem":966},"Output Cache: серверний кеш HTTP-відповідей (.NET 7+)","\u002Fcsharp\u002Faspnet\u002Fcaching\u002Foutput-cache","01.csharp\u002F11.aspnet\u002F06.caching\u002F05.output-cache",{"title":968,"icon":969,"path":970,"stem":971,"children":972,"page":59},"Тестування","i-lucide-test-tube","\u002Fcsharp\u002Faspnet\u002Ftesting","01.csharp\u002F11.aspnet\u002F07.testing",[973,977,981,985,989,993,997,1001,1005,1009,1013,1017,1021,1025,1029,1033,1037,1041,1045,1049,1053],{"title":974,"path":975,"stem":976},"Що таке тестування? Від інтуїції до науки","\u002Fcsharp\u002Faspnet\u002Ftesting\u002Fwhat-is-testing","01.csharp\u002F11.aspnet\u002F07.testing\u002F01.what-is-testing",{"title":978,"path":979,"stem":980},"Піраміда тестування — Стратегія, а не Догма","\u002Fcsharp\u002Faspnet\u002Ftesting\u002Ftesting-pyramid","01.csharp\u002F11.aspnet\u002F07.testing\u002F02.testing-pyramid",{"title":982,"path":983,"stem":984},"Дві Школи Тестування — Лондон проти Детройту","\u002Fcsharp\u002Faspnet\u002Ftesting\u002Ftesting-schools","01.csharp\u002F11.aspnet\u002F07.testing\u002F03.testing-schools",{"title":986,"path":987,"stem":988},"TDD та BDD — Тести як Дизайн-інструмент","\u002Fcsharp\u002Faspnet\u002Ftesting\u002Ftdd-and-bdd","01.csharp\u002F11.aspnet\u002F07.testing\u002F04.tdd-and-bdd",{"title":990,"path":991,"stem":992},"Що саме тестувати — Техніки аналізу та Циклomatична складність","\u002Fcsharp\u002Faspnet\u002Ftesting\u002Fwhat-to-test","01.csharp\u002F11.aspnet\u002F07.testing\u002F05.what-to-test",{"title":994,"path":995,"stem":996},"Тестові Фреймворки — Навіщо вони і що всередині","\u002Fcsharp\u002Faspnet\u002Ftesting\u002Ftest-frameworks","01.csharp\u002F11.aspnet\u002F07.testing\u002F06.test-frameworks",{"title":998,"path":999,"stem":1000},"xUnit — Факти, Теорії та Lifecycle тестів","\u002Fcsharp\u002Faspnet\u002Ftesting\u002Fxunit-basics","01.csharp\u002F11.aspnet\u002F07.testing\u002F07.xunit-basics",{"title":1002,"path":1003,"stem":1004},"xUnit Advanced — Fixtures, Кастомізація та Розширення","\u002Fcsharp\u002Faspnet\u002Ftesting\u002Fxunit-advanced","01.csharp\u002F11.aspnet\u002F07.testing\u002F08.xunit-advanced",{"title":1006,"path":1007,"stem":1008},"Moq — Глибоке занурення в мокування","\u002Fcsharp\u002Faspnet\u002Ftesting\u002Fmocking-with-moq","01.csharp\u002F11.aspnet\u002F07.testing\u002F09.mocking-with-moq",{"title":1010,"path":1011,"stem":1012},"Тестування Баз Даних — EF Core, SQLite та Testcontainers","\u002Fcsharp\u002Faspnet\u002Ftesting\u002Fdatabase-testing","01.csharp\u002F11.aspnet\u002F07.testing\u002F10.database-testing",{"title":1014,"path":1015,"stem":1016},"Integration Testing — Частина 1 [Теорія та WebApplicationFactory]","\u002Fcsharp\u002Faspnet\u002Ftesting\u002Fintegration-testing","01.csharp\u002F11.aspnet\u002F07.testing\u002F11.integration-testing",{"title":1018,"path":1019,"stem":1020},"Інтеграційне тестування — Практика","\u002Fcsharp\u002Faspnet\u002Ftesting\u002F11a.integration-testing-practice","01.csharp\u002F11.aspnet\u002F07.testing\u002F11a.integration-testing-practice",{"title":1022,"path":1023,"stem":1024},"Integration Testing — Частина 2 [Просунуті Сценарії та Testcontainers]","\u002Fcsharp\u002Faspnet\u002Ftesting\u002Fintegration-testing-advanced","01.csharp\u002F11.aspnet\u002F07.testing\u002F12.integration-testing-advanced",{"title":1026,"path":1027,"stem":1028},"Професійний Postman: Колекції, Змінні та GitHub Інтеграція","\u002Fcsharp\u002Faspnet\u002Ftesting\u002Fpostman-professional","01.csharp\u002F11.aspnet\u002F07.testing\u002F13.postman-professional",{"title":1030,"path":1031,"stem":1032},"HttpClient у Тестах Частина 1: Архітектура та MockHttpMessageHandler","\u002Fcsharp\u002Faspnet\u002Ftesting\u002Fhttpclient-testing","01.csharp\u002F11.aspnet\u002F07.testing\u002F14.httpclient-testing",{"title":1034,"path":1035,"stem":1036},"HttpClient у Тестах Частина 2: WireMock.Net та Resilience","\u002Fcsharp\u002Faspnet\u002Ftesting\u002Fwiremock-net","01.csharp\u002F11.aspnet\u002F07.testing\u002F15.wiremock-net",{"title":1038,"path":1039,"stem":1040},"Патерни та Анти-патерни Тестування: Test Smells","\u002Fcsharp\u002Faspnet\u002Ftesting\u002Ftesting-patterns","01.csharp\u002F11.aspnet\u002F07.testing\u002F16.testing-patterns",{"title":1042,"path":1043,"stem":1044},"Просунуті інструменти: Time, Snapshots та Властивості","\u002Fcsharp\u002Faspnet\u002Ftesting\u002Fadvanced-testing-tools","01.csharp\u002F11.aspnet\u002F07.testing\u002F17.advanced-testing-tools",{"title":1046,"path":1047,"stem":1048},"Тестування Архітектури з NetArchTest","\u002Fcsharp\u002Faspnet\u002Ftesting\u002Farchitecture-testing","01.csharp\u002F11.aspnet\u002F07.testing\u002F18.architecture-testing",{"title":1050,"path":1051,"stem":1052},"Тестування Продуктивності: BenchmarkDotNet, NBomber та k6","\u002Fcsharp\u002Faspnet\u002Ftesting\u002Fperformance-testing","01.csharp\u002F11.aspnet\u002F07.testing\u002F19.performance-testing",{"title":1054,"path":1055,"stem":1056},"Залишок плану для курсу \"Тестування ASP.NET Minimal API\"","\u002Fcsharp\u002Faspnet\u002Ftesting\u002Fremaining_plan","01.csharp\u002F11.aspnet\u002F07.testing\u002Fremaining_plan",{"title":1058,"icon":1059,"path":1060,"stem":1061,"children":1062,"page":59},"Платежі","i-lucide-credit-card","\u002Fcsharp\u002Faspnet\u002Fpayments","01.csharp\u002F11.aspnet\u002F08.payments",[1063,1067,1071,1075,1079,1083,1087,1091,1095,1099,1103,1107],{"title":1064,"path":1065,"stem":1066},"Основи платіжної інфраструктури","\u002Fcsharp\u002Faspnet\u002Fpayments\u002Fpayment-fundamentals","01.csharp\u002F11.aspnet\u002F08.payments\u002F01.payment-fundamentals",{"title":1068,"path":1069,"stem":1070},"Методи оплати в Україні","\u002Fcsharp\u002Faspnet\u002Fpayments\u002Fpayment-methods-ukraine","01.csharp\u002F11.aspnet\u002F08.payments\u002F02.payment-methods-ukraine",{"title":1072,"path":1073,"stem":1074},"PCI DSS та безпека платежів","\u002Fcsharp\u002Faspnet\u002Fpayments\u002Fpci-dss-security","01.csharp\u002F11.aspnet\u002F08.payments\u002F03.pci-dss-security",{"title":1076,"path":1077,"stem":1078},"Архітектура платіжної підсистеми","\u002Fcsharp\u002Faspnet\u002Fpayments\u002Fpayment-architecture","01.csharp\u002F11.aspnet\u002F08.payments\u002F04.payment-architecture",{"title":1080,"path":1081,"stem":1082},"Інтеграція LiqPay (ПриватБанк)","\u002Fcsharp\u002Faspnet\u002Fpayments\u002Fliqpay-integration","01.csharp\u002F11.aspnet\u002F08.payments\u002F05.liqpay-integration",{"title":1084,"path":1085,"stem":1086},"Інтеграція Monobank Acquiring API","\u002Fcsharp\u002Faspnet\u002Fpayments\u002Fmonobank-acquiring","01.csharp\u002F11.aspnet\u002F08.payments\u002F06.monobank-acquiring",{"title":1088,"path":1089,"stem":1090},"Інтеграція Stripe","\u002Fcsharp\u002Faspnet\u002Fpayments\u002Fstripe-integration","01.csharp\u002F11.aspnet\u002F08.payments\u002F07.stripe-integration",{"title":1092,"path":1093,"stem":1094},"Webhooks — глибоке занурення","\u002Fcsharp\u002Faspnet\u002Fpayments\u002Fwebhooks-deep-dive","01.csharp\u002F11.aspnet\u002F08.payments\u002F08.webhooks-deep-dive",{"title":1096,"path":1097,"stem":1098},"Підписки та рекурентні платежі","\u002Fcsharp\u002Faspnet\u002Fpayments\u002Fsubscriptions-recurring","01.csharp\u002F11.aspnet\u002F08.payments\u002F09.subscriptions-recurring",{"title":1100,"path":1101,"stem":1102},"Повернення коштів та диспути","\u002Fcsharp\u002Faspnet\u002Fpayments\u002Frefunds-disputes","01.csharp\u002F11.aspnet\u002F08.payments\u002F10.refunds-disputes",{"title":1104,"path":1105,"stem":1106},"Тестування платіжних інтеграцій","\u002Fcsharp\u002Faspnet\u002Fpayments\u002Ftesting-payments","01.csharp\u002F11.aspnet\u002F08.payments\u002F11.testing-payments",{"title":1108,"path":1109,"stem":1110},"Чекліст виходу в Production","\u002Fcsharp\u002Faspnet\u002Fpayments\u002Fproduction-checklist","01.csharp\u002F11.aspnet\u002F08.payments\u002F12.production-checklist",{"title":1112,"icon":1113,"items":1114,"path":1127,"stem":1128,"children":1129,"page":59},"Популярні бібліотеки","lucide:box",[1115,1116,1117,1118,1119,1120,1121,1122,1123,1124,1125,1126],"01.fluent-validation","02.mapster","03.erroror-result-pattern","04.serilog","05.mediatr","06.polly","07.health-checks","08.feature-management","09.fluent-email","10.quest-pdf","11.bogus","12.humanizer-guard","\u002Fcsharp\u002Faspnet\u002Flibraries","01.csharp\u002F11.aspnet\u002F09.libraries",[1130,1134,1138,1142,1146,1150,1154,1158,1162,1166,1170,1174,1178],{"title":1131,"path":1132,"stem":1133},"Валідація з FluentValidation в ASP.NET Core","\u002Fcsharp\u002Faspnet\u002Flibraries\u002Ffluent-validation","01.csharp\u002F11.aspnet\u002F09.libraries\u002F01.fluent-validation",{"title":1135,"path":1136,"stem":1137},"Маппінг об","\u002Fcsharp\u002Faspnet\u002Flibraries\u002Fmapster","01.csharp\u002F11.aspnet\u002F09.libraries\u002F02.mapster",{"title":1139,"path":1140,"stem":1141},"Обробка помилок з ErrorOr та Result Pattern в ASP.NET Core","\u002Fcsharp\u002Faspnet\u002Flibraries\u002Ferroror-result-pattern","01.csharp\u002F11.aspnet\u002F09.libraries\u002F03.erroror-result-pattern",{"title":1143,"path":1144,"stem":1145},"Структуроване логування з Serilog в ASP.NET Core","\u002Fcsharp\u002Faspnet\u002Flibraries\u002Fserilog","01.csharp\u002F11.aspnet\u002F09.libraries\u002F04.serilog",{"title":1147,"path":1148,"stem":1149},"CQRS та Mediator з MediatR в ASP.NET Core","\u002Fcsharp\u002Faspnet\u002Flibraries\u002Fmediatr","01.csharp\u002F11.aspnet\u002F09.libraries\u002F05.mediatr",{"title":1151,"path":1152,"stem":1153},"Відмовостійкість з Polly в ASP.NET Core","\u002Fcsharp\u002Faspnet\u002Flibraries\u002Fpolly","01.csharp\u002F11.aspnet\u002F09.libraries\u002F06.polly",{"title":1155,"path":1156,"stem":1157},"Health Checks в ASP.NET Core","\u002Fcsharp\u002Faspnet\u002Flibraries\u002Fhealth-checks","01.csharp\u002F11.aspnet\u002F09.libraries\u002F07.health-checks",{"title":1159,"path":1160,"stem":1161},"Feature Management та Feature Flags в ASP.NET Core","\u002Fcsharp\u002Faspnet\u002Flibraries\u002Ffeature-management","01.csharp\u002F11.aspnet\u002F09.libraries\u002F08.feature-management",{"title":1163,"path":1164,"stem":1165},"Відправка Email з FluentEmail в ASP.NET Core","\u002Fcsharp\u002Faspnet\u002Flibraries\u002Ffluent-email","01.csharp\u002F11.aspnet\u002F09.libraries\u002F09.fluent-email",{"title":1167,"path":1168,"stem":1169},"Генерація PDF з QuestPDF в ASP.NET Core","\u002Fcsharp\u002Faspnet\u002Flibraries\u002Fquest-pdf","01.csharp\u002F11.aspnet\u002F09.libraries\u002F10.quest-pdf",{"title":1171,"path":1172,"stem":1173},"Генерація тестових даних з Bogus в ASP.NET Core","\u002Fcsharp\u002Faspnet\u002Flibraries\u002Fbogus","01.csharp\u002F11.aspnet\u002F09.libraries\u002F11.bogus",{"title":1175,"path":1176,"stem":1177},"Humanizer та Guard Clauses в ASP.NET Core","\u002Fcsharp\u002Faspnet\u002Flibraries\u002Fhumanizer-guard","01.csharp\u002F11.aspnet\u002F09.libraries\u002F12.humanizer-guard",{"title":1179,"path":1180,"stem":1181},"План модуля 10.libraries — Популярні бібліотеки ASP.NET","\u002Fcsharp\u002Faspnet\u002Flibraries\u002Fplan","01.csharp\u002F11.aspnet\u002F09.libraries\u002Fplan",{"title":1183,"icon":1184,"path":1185,"stem":1186,"children":1187,"page":59},"Razor Pages","i-lucide-layout-template","\u002Fcsharp\u002Faspnet\u002Frazor-pages","01.csharp\u002F11.aspnet\u002F10.razor-pages",[1188,1192,1196,1200,1204,1208],{"title":1189,"path":1190,"stem":1191},"Від Minimal API до Razor Pages: концептуальний перехід","\u002Fcsharp\u002Faspnet\u002Frazor-pages\u002Ffrom-minimal-api","01.csharp\u002F11.aspnet\u002F10.razor-pages\u002F01.from-minimal-api",{"title":1193,"path":1194,"stem":1195},"PageModel: логіка сторінки Razor Pages","\u002Fcsharp\u002Faspnet\u002Frazor-pages\u002Fpage-model","01.csharp\u002F11.aspnet\u002F10.razor-pages\u002F02.page-model",{"title":1197,"path":1198,"stem":1199},"Razor синтаксис: шаблонізатор у .cshtml","\u002Fcsharp\u002Faspnet\u002Frazor-pages\u002Frazor-syntax","01.csharp\u002F11.aspnet\u002F10.razor-pages\u002F03.razor-syntax",{"title":1201,"path":1202,"stem":1203},"Tag Helpers: типізований HTML","\u002Fcsharp\u002Faspnet\u002Frazor-pages\u002Ftag-helpers","01.csharp\u002F11.aspnet\u002F10.razor-pages\u002F04.tag-helpers",{"title":1205,"path":1206,"stem":1207},"Форми і валідація: повний цикл обробки даних","\u002Fcsharp\u002Faspnet\u002Frazor-pages\u002Fforms-validation","01.csharp\u002F11.aspnet\u002F10.razor-pages\u002F05.forms-validation",{"title":1209,"path":1210,"stem":1211},"Практичний проєкт: TaskManager на Razor Pages","\u002Fcsharp\u002Faspnet\u002Frazor-pages\u002Fproject-task-manager","01.csharp\u002F11.aspnet\u002F10.razor-pages\u002F06.project-task-manager",{"title":1213,"path":1214,"stem":1215,"children":1216,"page":59},"ASP.NET Core MVC","\u002Fcsharp\u002Faspnet\u002Fmvc","01.csharp\u002F11.aspnet\u002F11.mvc",[1217,1221,1225,1229,1233,1237,1241,1245,1249,1253,1257,1261,1265,1269,1273,1277,1281,1285],{"title":1218,"path":1219,"stem":1220},"Патерн MVC: архітектура, що змінила веб","\u002Fcsharp\u002Faspnet\u002Fmvc\u002Fmvc-pattern","01.csharp\u002F11.aspnet\u002F11.mvc\u002F01.mvc-pattern",{"title":1222,"path":1223,"stem":1224},"Від Razor Pages до MVC: концептуальний перехід","\u002Fcsharp\u002Faspnet\u002Fmvc\u002Ffrom-razor-pages","01.csharp\u002F11.aspnet\u002F11.mvc\u002F02.from-razor-pages",{"title":1226,"path":1227,"stem":1228},"Controllers та Actions: серце MVC","\u002Fcsharp\u002Faspnet\u002Fmvc\u002Fcontrollers-actions","01.csharp\u002F11.aspnet\u002F11.mvc\u002F03.controllers-actions",{"title":1230,"path":1231,"stem":1232},"Маршрутизація в MVC: Convention vs Attribute Routing","\u002Fcsharp\u002Faspnet\u002Fmvc\u002Frouting-mvc","01.csharp\u002F11.aspnet\u002F11.mvc\u002F04.routing-mvc",{"title":1234,"path":1235,"stem":1236},"Model Binding: від HTTP до C#","\u002Fcsharp\u002Faspnet\u002Fmvc\u002Fmodel-binding","01.csharp\u002F11.aspnet\u002F11.mvc\u002F05.model-binding",{"title":1238,"path":1239,"stem":1240},"Views, ViewData, ViewBag, TempData і ViewModel","\u002Fcsharp\u002Faspnet\u002Fmvc\u002Fviews-viewdata-tempdata","01.csharp\u002F11.aspnet\u002F11.mvc\u002F06.views-viewdata-tempdata",{"title":1242,"path":1243,"stem":1244},"Filters: аспектно-орієнтоване програмування в MVC","\u002Fcsharp\u002Faspnet\u002Fmvc\u002Ffilters","01.csharp\u002F11.aspnet\u002F11.mvc\u002F07.filters",{"title":1246,"path":1247,"stem":1248},"Areas: структурування великих застосунків","\u002Fcsharp\u002Faspnet\u002Fmvc\u002Fareas","01.csharp\u002F11.aspnet\u002F11.mvc\u002F08.areas",{"title":1250,"path":1251,"stem":1252},"View Components: повторювані незалежні блоки UI","\u002Fcsharp\u002Faspnet\u002Fmvc\u002Fview-components","01.csharp\u002F11.aspnet\u002F11.mvc\u002F09.view-components",{"title":1254,"path":1255,"stem":1256},"Display та Editor Templates","\u002Fcsharp\u002Faspnet\u002Fmvc\u002Fdisplay-editor-templates","01.csharp\u002F11.aspnet\u002F11.mvc\u002F10.display-editor-templates",{"title":1258,"path":1259,"stem":1260},"Валідація: IValidatableObject та FluentValidation","\u002Fcsharp\u002Faspnet\u002Fmvc\u002Fvalidation-advanced","01.csharp\u002F11.aspnet\u002F11.mvc\u002F11.validation-advanced",{"title":1262,"path":1263,"stem":1264},"HTMX: інтерактивність через HTML-атрибути","\u002Fcsharp\u002Faspnet\u002Fmvc\u002Fhtmx","01.csharp\u002F11.aspnet\u002F11.mvc\u002F12.htmx",{"title":1266,"path":1267,"stem":1268},"HTMX у ASP.NET Core MVC: серверна інтеграція","\u002Fcsharp\u002Faspnet\u002Fmvc\u002Fajax-htmx-mvc","01.csharp\u002F11.aspnet\u002F11.mvc\u002F13.ajax-htmx-mvc",{"title":1270,"path":1271,"stem":1272},"Практичний проєкт: Каталог товарів з HTMX","\u002Fcsharp\u002Faspnet\u002Fmvc\u002Fhtmx-project","01.csharp\u002F11.aspnet\u002F11.mvc\u002F14.htmx-project",{"title":1274,"path":1275,"stem":1276},"Завантаження та обробка файлів","\u002Fcsharp\u002Faspnet\u002Fmvc\u002Ffile-upload","01.csharp\u002F11.aspnet\u002F11.mvc\u002F15.file-upload",{"title":1278,"path":1279,"stem":1280},"Глобалізація та Локалізація MVC","\u002Fcsharp\u002Faspnet\u002Fmvc\u002Fglobalization-localization","01.csharp\u002F11.aspnet\u002F11.mvc\u002F16.globalization-localization",{"title":1282,"path":1283,"stem":1284},"Підсумковий проєкт: Блог-платформа","\u002Fcsharp\u002Faspnet\u002Fmvc\u002Fmvc-project","01.csharp\u002F11.aspnet\u002F11.mvc\u002F17.mvc-project",{"title":1286,"path":1287,"stem":1288},"План курсу: ASP.NET Core MVC","\u002Fcsharp\u002Faspnet\u002Fmvc\u002Fplan","01.csharp\u002F11.aspnet\u002F11.mvc\u002Fplan",{"title":1290,"path":1291,"stem":1292,"children":1293,"page":59},"Web Api","\u002Fcsharp\u002Faspnet\u002Fweb-api","01.csharp\u002F11.aspnet\u002F12.web-api",[1294,1298,1302,1306,1310,1314,1318,1322,1326,1330,1334,1338,1342],{"title":1295,"path":1296,"stem":1297},"Від Minimal API до Controller-based API","\u002Fcsharp\u002Faspnet\u002Fweb-api\u002Ffrom-minimal-api-to-controllers","01.csharp\u002F11.aspnet\u002F12.web-api\u002F01.from-minimal-api-to-controllers",{"title":1299,"path":1300,"stem":1301},"ControllerBase, ActionResult\u003CT> та Response Types","\u002Fcsharp\u002Faspnet\u002Fweb-api\u002Fcontroller-base-actionresult","01.csharp\u002F11.aspnet\u002F12.web-api\u002F02.controller-base-actionresult",{"title":1303,"path":1304,"stem":1305},"Content Negotiation - JSON, XML та власні форматери","\u002Fcsharp\u002Faspnet\u002Fweb-api\u002Fcontent-negotiation","01.csharp\u002F11.aspnet\u002F12.web-api\u002F03.content-negotiation",{"title":1307,"path":1308,"stem":1309},"Версіонування API","\u002Fcsharp\u002Faspnet\u002Fweb-api\u002Fapi-versioning","01.csharp\u002F11.aspnet\u002F12.web-api\u002F04.api-versioning",{"title":1311,"path":1312,"stem":1313},"ProblemDetails та структурована обробка помилок","\u002Fcsharp\u002Faspnet\u002Fweb-api\u002Fproblemdetails-error-handling","01.csharp\u002F11.aspnet\u002F12.web-api\u002F05.problemdetails-error-handling",{"title":1315,"path":1316,"stem":1317},"Фільтри у Web API контексті","\u002Fcsharp\u002Faspnet\u002Fweb-api\u002Ffilters-for-api","01.csharp\u002F11.aspnet\u002F12.web-api\u002F06.filters-for-api",{"title":1319,"path":1320,"stem":1321},"Пагінація, фільтрація та сортування","\u002Fcsharp\u002Faspnet\u002Fweb-api\u002Fpagination-filtering-sorting","01.csharp\u002F11.aspnet\u002F12.web-api\u002F07.pagination-filtering-sorting",{"title":1323,"path":1324,"stem":1325},"HATEOAS та Resource Expansion","\u002Fcsharp\u002Faspnet\u002Fweb-api\u002Fhateoas-resource-expansion","01.csharp\u002F11.aspnet\u002F12.web-api\u002F08.hateoas-resource-expansion",{"title":1327,"path":1328,"stem":1329},"Гібридна архітектура - Minimal API + Controllers","\u002Fcsharp\u002Faspnet\u002Fweb-api\u002Fminimal-api-vs-controllers-hybrid","01.csharp\u002F11.aspnet\u002F12.web-api\u002F09.minimal-api-vs-controllers-hybrid",{"title":1331,"path":1332,"stem":1333},"Документація API - Swashbuckle, NSwag та генерація клієнтів","\u002Fcsharp\u002Faspnet\u002Fweb-api\u002Fapi-documentation-generation","01.csharp\u002F11.aspnet\u002F12.web-api\u002F10.api-documentation-generation",{"title":1335,"path":1336,"stem":1337},"Health Checks та моніторинг API","\u002Fcsharp\u002Faspnet\u002Fweb-api\u002Fhealth-checks-monitoring","01.csharp\u002F11.aspnet\u002F12.web-api\u002F11.health-checks-monitoring",{"title":1339,"path":1340,"stem":1341},"Підсумковий проєкт - Production-Ready REST API","\u002Fcsharp\u002Faspnet\u002Fweb-api\u002Fweb-api-project","01.csharp\u002F11.aspnet\u002F12.web-api\u002F12.web-api-project",{"title":1343,"path":1344,"stem":1345},"План курсу: ASP.NET Core Web API (Controllers)","\u002Fcsharp\u002Faspnet\u002Fweb-api\u002Fplan","01.csharp\u002F11.aspnet\u002F12.web-api\u002Fplan",{"title":1347,"icon":1348,"path":1349,"stem":1350,"children":1351,"page":59},"Моніторинг","i-lucide-activity","\u002Fcsharp\u002Faspnet\u002Fmonitoring","01.csharp\u002F11.aspnet\u002F13.monitoring",[1352,1356,1360],{"title":1353,"path":1354,"stem":1355},"Спостережуваність: від console.log до production-систем","\u002Fcsharp\u002Faspnet\u002Fmonitoring\u002Fobservability-intro","01.csharp\u002F11.aspnet\u002F13.monitoring\u002F01.observability-intro",{"title":1357,"path":1358,"stem":1359},"Health Checks: перший рівень observability","\u002Fcsharp\u002Faspnet\u002Fmonitoring\u002Fhealth-checks","01.csharp\u002F11.aspnet\u002F13.monitoring\u002F02.health-checks",{"title":1361,"path":1362,"stem":1363},"Вбудовані метрики .NET 10 та System.Diagnostics.Metrics","\u002Fcsharp\u002Faspnet\u002Fmonitoring\u002Fdotnet-metrics","01.csharp\u002F11.aspnet\u002F13.monitoring\u002F03.dotnet-metrics",{"title":1365,"icon":1366,"path":1367,"stem":1368,"children":1369,"page":59},"Desktop UI","i-lucide-app-window","\u002Fcsharp\u002Fdesktop-ui","01.csharp\u002F12.desktop-ui",[1370,1374,1378,1382,1386,1390,1394,1398,1402,1406,1410,1414,1418,1422,1426,1430,1434,1438,1442,1446,1450,1454,1458,1462,1466,1470,1474,1478,1482,1486,1490,1494,1498,1502,1506,1510,1514,1518,1522,1526,1530,1534,1538,1542,1546,1550,1554,1558,1562,1566,1570,1574,1578,1582,1586,1590,1594,1598,1602,1606,1610,1614,1618,1622,1626,1630,1634,1638,1642,1646,1650],{"title":1371,"path":1372,"stem":1373},"Що таке десктопна розробка?","\u002Fcsharp\u002Fdesktop-ui\u002Fwhat-is-desktop-dev","01.csharp\u002F12.desktop-ui\u002F01.what-is-desktop-dev",{"title":1375,"path":1376,"stem":1377},"Архітектура WPF — як влаштований графічний інтерфейс","\u002Fcsharp\u002Fdesktop-ui\u002Fwpf-architecture","01.csharp\u002F12.desktop-ui\u002F02.wpf-architecture",{"title":1379,"path":1380,"stem":1381},"Перший WPF-проєкт — від нуля до вікна","\u002Fcsharp\u002Fdesktop-ui\u002Ffirst-wpf-app","01.csharp\u002F12.desktop-ui\u002F03.first-wpf-app",{"title":1383,"path":1384,"stem":1385},"Перший Avalonia-проєкт: WPF для всіх платформ","\u002Fcsharp\u002Fdesktop-ui\u002F03a.first-avalonia-app","01.csharp\u002F12.desktop-ui\u002F03a.first-avalonia-app",{"title":1387,"path":1388,"stem":1389},"XAML: декларативний інтерфейс","\u002Fcsharp\u002Fdesktop-ui\u002Fxaml-basics","01.csharp\u002F12.desktop-ui\u002F04.xaml-basics",{"title":1391,"path":1392,"stem":1393},"Fluent UI у WPF — сучасний дизайн Windows 11","\u002Fcsharp\u002Fdesktop-ui\u002F04a.wpf-fluent-ui","01.csharp\u002F12.desktop-ui\u002F04a.wpf-fluent-ui",{"title":1395,"path":1396,"stem":1397},"WPF UI — сучасна бібліотека Fluent контролів","\u002Fcsharp\u002Fdesktop-ui\u002F04b.wpf-ui-library","01.csharp\u002F12.desktop-ui\u002F04b.wpf-ui-library",{"title":1399,"path":1400,"stem":1401},"HandyControl — велика бібліотека UI контролів для WPF","\u002Fcsharp\u002Fdesktop-ui\u002F04c.handycontrol-library","01.csharp\u002F12.desktop-ui\u002F04c.handycontrol-library",{"title":1403,"path":1404,"stem":1405},"Простори імен та ресурси XAML","\u002Fcsharp\u002Fdesktop-ui\u002Fxaml-namespaces-resources","01.csharp\u002F12.desktop-ui\u002F05.xaml-namespaces-resources",{"title":1407,"path":1408,"stem":1409},"XAML в Avalonia: ключові відмінності від WPF","\u002Fcsharp\u002Fdesktop-ui\u002F05a.avalonia-xaml-differences","01.csharp\u002F12.desktop-ui\u002F05a.avalonia-xaml-differences",{"title":1411,"path":1412,"stem":1413},"Розширення розмітки XAML (Markup Extensions)","\u002Fcsharp\u002Fdesktop-ui\u002Fxaml-markup-extensions","01.csharp\u002F12.desktop-ui\u002F06.xaml-markup-extensions",{"title":1415,"path":1416,"stem":1417},"Панелі Layout: StackPanel, WrapPanel, DockPanel","\u002Fcsharp\u002Fdesktop-ui\u002Flayout-panels-part1","01.csharp\u002F12.desktop-ui\u002F07.layout-panels-part1",{"title":1419,"path":1420,"stem":1421},"Grid, Canvas, UniformGrid","\u002Fcsharp\u002Fdesktop-ui\u002Flayout-panels-part2","01.csharp\u002F12.desktop-ui\u002F07.layout-panels-part2",{"title":1423,"path":1424,"stem":1425},"Просунуті техніки Layout","\u002Fcsharp\u002Fdesktop-ui\u002Flayout-advanced","01.csharp\u002F12.desktop-ui\u002F08.layout-advanced",{"title":1427,"path":1428,"stem":1429},"Адаптивний Layout та найкращі практики","\u002Fcsharp\u002Fdesktop-ui\u002Flayout-responsive","01.csharp\u002F12.desktop-ui\u002F09.layout-responsive",{"title":1431,"path":1432,"stem":1433},"Layout в Avalonia: відмінності та нові можливості","\u002Fcsharp\u002Fdesktop-ui\u002F09a.layout-avalonia","01.csharp\u002F12.desktop-ui\u002F09a.layout-avalonia",{"title":1435,"path":1436,"stem":1437},"Button, Image, ProgressBar та інші базові контроли","\u002Fcsharp\u002Fdesktop-ui\u002Fbasic-controls","01.csharp\u002F12.desktop-ui\u002F10.basic-controls",{"title":1439,"path":1440,"stem":1441},"Контроли в Avalonia: відмінності від WPF","\u002Fcsharp\u002Fdesktop-ui\u002F10a.controls-avalonia","01.csharp\u002F12.desktop-ui\u002F10a.controls-avalonia",{"title":1443,"path":1444,"stem":1445},"Текстові контроли — TextBlock, TextBox, RichTextBox","\u002Fcsharp\u002Fdesktop-ui\u002Ftext-controls","01.csharp\u002F12.desktop-ui\u002F11.text-controls",{"title":1447,"path":1448,"stem":1449},"Контроли вибору — CheckBox, RadioButton, ComboBox, ListBox, DatePicker","\u002Fcsharp\u002Fdesktop-ui\u002Fselection-controls","01.csharp\u002F12.desktop-ui\u002F12.selection-controls",{"title":1451,"path":1452,"stem":1453},"Content Model — GroupBox, Expander, TabControl, StatusBar","\u002Fcsharp\u002Fdesktop-ui\u002Fcontent-controls","01.csharp\u002F12.desktop-ui\u002F13.content-controls",{"title":1455,"path":1456,"stem":1457},"UI\u002FUX принципи десктопних застосунків","\u002Fcsharp\u002Fdesktop-ui\u002F13a.ui-ux-principles","01.csharp\u002F12.desktop-ui\u002F13a.ui-ux-principles",{"title":1459,"path":1460,"stem":1461},"Dependency Properties — Концепція та Value Resolution","\u002Fcsharp\u002Fdesktop-ui\u002Fdependency-properties-part1","01.csharp\u002F12.desktop-ui\u002F14.dependency-properties-part1",{"title":1463,"path":1464,"stem":1465},"Avalonia Property System — StyledProperty та DirectProperty","\u002Fcsharp\u002Fdesktop-ui\u002F14a.avalonia-property-system","01.csharp\u002F12.desktop-ui\u002F14a.avalonia-property-system",{"title":1467,"path":1468,"stem":1469},"Attached Properties — Властивості без меж","\u002Fcsharp\u002Fdesktop-ui\u002Fattached-properties","01.csharp\u002F12.desktop-ui\u002F15.attached-properties",{"title":1471,"path":1472,"stem":1473},"Routed Events — Маршрутизація подій у WPF","\u002Fcsharp\u002Fdesktop-ui\u002Frouted-events","01.csharp\u002F12.desktop-ui\u002F16.routed-events",{"title":1475,"path":1476,"stem":1477},"Data Binding — Від Code-Behind до Декларативності","\u002Fcsharp\u002Fdesktop-ui\u002Fdata-binding-basics-part1","01.csharp\u002F12.desktop-ui\u002F17.data-binding-basics-part1",{"title":1479,"path":1480,"stem":1481},"INotifyPropertyChanged — Живе оновлення UI","\u002Fcsharp\u002Fdesktop-ui\u002Fdata-binding-basics-part2","01.csharp\u002F12.desktop-ui\u002F17.data-binding-basics-part2",{"title":1483,"path":1484,"stem":1485},"Compiled Bindings в Avalonia — Безпека на етапі компіляції","\u002Fcsharp\u002Fdesktop-ui\u002F17a.avalonia-compiled-bindings","01.csharp\u002F12.desktop-ui\u002F17a.avalonia-compiled-bindings",{"title":1487,"path":1488,"stem":1489},"Просунутий Data Binding — ElementName, RelativeSource, MultiBinding","\u002Fcsharp\u002Fdesktop-ui\u002Fdata-binding-advanced","01.csharp\u002F12.desktop-ui\u002F18.data-binding-advanced",{"title":1491,"path":1492,"stem":1493},"Value Converters — Перетворення типів даних у Data Binding","\u002Fcsharp\u002Fdesktop-ui\u002Fvalue-converters","01.csharp\u002F12.desktop-ui\u002F19.value-converters",{"title":1495,"path":1496,"stem":1497},"Data Templates — Візуалізація об'єктів у WPF","\u002Fcsharp\u002Fdesktop-ui\u002Fdata-templates","01.csharp\u002F12.desktop-ui\u002F20.data-templates",{"title":1499,"path":1500,"stem":1501},"Collections Binding Part 1 — ObservableCollection та ItemsControl","\u002Fcsharp\u002Fdesktop-ui\u002Fcollections-binding-part1","01.csharp\u002F12.desktop-ui\u002F21.collections-binding-part1",{"title":1503,"path":1504,"stem":1505},"Collections Binding Part 2 — ICollectionView, Filtering, Sorting та Virtualization","\u002Fcsharp\u002Fdesktop-ui\u002Fcollections-binding-part2","01.csharp\u002F12.desktop-ui\u002F21.collections-binding-part2",{"title":1507,"path":1508,"stem":1509},"MVVM Pattern — Від Spaghetti Code до архітектури","\u002Fcsharp\u002Fdesktop-ui\u002Fmvvm-pattern","01.csharp\u002F12.desktop-ui\u002F22.mvvm-pattern",{"title":1511,"path":1512,"stem":1513},"ViewModel Implementation — Від BaseViewModel до валідації","\u002Fcsharp\u002Fdesktop-ui\u002Fviewmodel-implementation","01.csharp\u002F12.desktop-ui\u002F23.viewmodel-implementation",{"title":1515,"path":1516,"stem":1517},"Commands — Від event handlers до декларативних команд","\u002Fcsharp\u002Fdesktop-ui\u002Fcommands","01.csharp\u002F12.desktop-ui\u002F24.commands",{"title":1519,"path":1520,"stem":1521},"MVVM Toolkit — MVVM без boilerplate через Source Generators","\u002Fcsharp\u002Fdesktop-ui\u002Fmvvm-toolkit","01.csharp\u002F12.desktop-ui\u002F25.mvvm-toolkit",{"title":1523,"path":1524,"stem":1525},"Messenger Pattern — Комунікація між ViewModel без прямих посилань","\u002Fcsharp\u002Fdesktop-ui\u002Fmessenger-pattern","01.csharp\u002F12.desktop-ui\u002F26.messenger-pattern",{"title":1527,"path":1528,"stem":1529},"Стилі WPF — CSS для десктопу","\u002Fcsharp\u002Fdesktop-ui\u002Fstyles-basics","01.csharp\u002F12.desktop-ui\u002F27.styles-basics",{"title":1531,"path":1532,"stem":1533},"CSS-like стилі Avalonia","\u002Fcsharp\u002Fdesktop-ui\u002F27a.avalonia-css-styling","01.csharp\u002F12.desktop-ui\u002F27a.avalonia-css-styling",{"title":1535,"path":1536,"stem":1537},"Control Templates — Частина 1. Концепція та TemplateBinding","\u002Fcsharp\u002Fdesktop-ui\u002Fcontrol-templates-part1","01.csharp\u002F12.desktop-ui\u002F28.control-templates-part1",{"title":1539,"path":1540,"stem":1541},"Control Templates — Частина 2. Named Parts та ContentPresenter","\u002Fcsharp\u002Fdesktop-ui\u002Fcontrol-templates-part2","01.csharp\u002F12.desktop-ui\u002F28.control-templates-part2",{"title":1543,"path":1544,"stem":1545},"Control Themes в Avalonia — нова ера стилізації","\u002Fcsharp\u002Fdesktop-ui\u002F28a.avalonia-control-themes","01.csharp\u002F12.desktop-ui\u002F28a.avalonia-control-themes",{"title":1547,"path":1548,"stem":1549},"Triggers та Visual State Manager у WPF","\u002Fcsharp\u002Fdesktop-ui\u002Ftriggers-visual-states","01.csharp\u002F12.desktop-ui\u002F29.triggers-visual-states",{"title":1551,"path":1552,"stem":1553},"Pseudo-classes в Avalonia — замість WPF Triggers","\u002Fcsharp\u002Fdesktop-ui\u002F29a.avalonia-pseudo-classes","01.csharp\u002F12.desktop-ui\u002F29a.avalonia-pseudo-classes",{"title":1555,"path":1556,"stem":1557},"Теми та ресурсні словники у WPF","\u002Fcsharp\u002Fdesktop-ui\u002Fresources-themes","01.csharp\u002F12.desktop-ui\u002F30.resources-themes",{"title":1559,"path":1560,"stem":1561},"Avalonia Themes — Fluent Design та система тематизації","\u002Fcsharp\u002Fdesktop-ui\u002F30a.avalonia-themes-fluent","01.csharp\u002F12.desktop-ui\u002F30a.avalonia-themes-fluent",{"title":1563,"path":1564,"stem":1565},"Контроли колекцій — глибоке занурення","\u002Fcsharp\u002Fdesktop-ui\u002Fcollection-controls","01.csharp\u002F12.desktop-ui\u002F31.collection-controls",{"title":1567,"path":1568,"stem":1569},"DataGrid — колонки та базове відображення","\u002Fcsharp\u002Fdesktop-ui\u002Fdatagrid-part1","01.csharp\u002F12.desktop-ui\u002F32.datagrid-part1",{"title":1571,"path":1572,"stem":1573},"DataGrid — сортування, фільтрація, редагування","\u002Fcsharp\u002Fdesktop-ui\u002Fdatagrid-part2","01.csharp\u002F12.desktop-ui\u002F32.datagrid-part2",{"title":1575,"path":1576,"stem":1577},"TreeView та GridView","\u002Fcsharp\u002Fdesktop-ui\u002Ftreeview-listview","01.csharp\u002F12.desktop-ui\u002F33.treeview-listview",{"title":1579,"path":1580,"stem":1581},"Меню, Toolbar, ContextMenu, StatusBar","\u002Fcsharp\u002Fdesktop-ui\u002Fmenus-toolbars","01.csharp\u002F12.desktop-ui\u002F34.menus-toolbars",{"title":1583,"path":1584,"stem":1585},"Навігація та керування вікнами. Частина 1: вікна та сторінки","\u002Fcsharp\u002Fdesktop-ui\u002Fnavigation-windows-part1","01.csharp\u002F12.desktop-ui\u002F35.navigation-windows-part1",{"title":1587,"path":1588,"stem":1589},"Навігація та керування вікнами. Частина 2: MVVM-навігація","\u002Fcsharp\u002Fdesktop-ui\u002Fnavigation-windows-part2","01.csharp\u002F12.desktop-ui\u002F35.navigation-windows-part2",{"title":1591,"path":1592,"stem":1593},"Avalonia — Навігація та діалоги","\u002Fcsharp\u002Fdesktop-ui\u002F35a.avalonia-navigation-dialogs","01.csharp\u002F12.desktop-ui\u002F35a.avalonia-navigation-dialogs",{"title":1595,"path":1596,"stem":1597},"Діалоги та File Pickers у WPF","\u002Fcsharp\u002Fdesktop-ui\u002Fdialogs-file-pickers","01.csharp\u002F12.desktop-ui\u002F36.dialogs-file-pickers",{"title":1599,"path":1600,"stem":1601},"UserControl: компонентний підхід у WPF","\u002Fcsharp\u002Fdesktop-ui\u002Fuser-controls","01.csharp\u002F12.desktop-ui\u002F37.user-controls",{"title":1603,"path":1604,"stem":1605},"Custom Controls: Lookless Controls у WPF","\u002Fcsharp\u002Fdesktop-ui\u002Fcustom-controls","01.csharp\u002F12.desktop-ui\u002F38.custom-controls",{"title":1607,"path":1608,"stem":1609},"Avalonia TemplatedControl — Lookless Controls","\u002Fcsharp\u002Fdesktop-ui\u002F38a.avalonia-templated-controls","01.csharp\u002F12.desktop-ui\u002F38a.avalonia-templated-controls",{"title":1611,"path":1612,"stem":1613},"Анімації у WPF: Storyboard та Easing Functions","\u002Fcsharp\u002Fdesktop-ui\u002Fanimations-transitions","01.csharp\u002F12.desktop-ui\u002F39.animations-transitions",{"title":1615,"path":1616,"stem":1617},"Анімації в Avalonia","\u002Fcsharp\u002Fdesktop-ui\u002F39a.avalonia-animations","01.csharp\u002F12.desktop-ui\u002F39a.avalonia-animations",{"title":1619,"path":1620,"stem":1621},"2D Графіка та Мультимедіа у WPF","\u002Fcsharp\u002Fdesktop-ui\u002Fmedia-graphics","01.csharp\u002F12.desktop-ui\u002F40.media-graphics",{"title":1623,"path":1624,"stem":1625},"Dependency Injection у WPF та Avalonia","\u002Fcsharp\u002Fdesktop-ui\u002Fdi-integration","01.csharp\u002F12.desktop-ui\u002F41.di-integration",{"title":1627,"path":1628,"stem":1629},"SQLite та EF Core у десктопних додатках","\u002Fcsharp\u002Fdesktop-ui\u002Fdata-persistence-part1","01.csharp\u002F12.desktop-ui\u002F42.data-persistence-part1",{"title":1631,"path":1632,"stem":1633},"Repository Pattern та Unit of Work","\u002Fcsharp\u002Fdesktop-ui\u002Fdata-persistence-part2","01.csharp\u002F12.desktop-ui\u002F43.data-persistence-part2",{"title":1635,"path":1636,"stem":1637},"Тестування ViewModels","\u002Fcsharp\u002Fdesktop-ui\u002Fviewmodel-testing","01.csharp\u002F12.desktop-ui\u002F44.viewmodel-testing",{"title":1639,"path":1640,"stem":1641},"Avalonia Headless Testing — тестування UI без вікон","\u002Fcsharp\u002Fdesktop-ui\u002F44a.avalonia-headless-testing","01.csharp\u002F12.desktop-ui\u002F44a.avalonia-headless-testing",{"title":1643,"path":1644,"stem":1645},"Кросплатформна розробка з Avalonia","\u002Fcsharp\u002Fdesktop-ui\u002Favalonia-cross-platform","01.csharp\u002F12.desktop-ui\u002F45.avalonia-cross-platform",{"title":1647,"path":1648,"stem":1649},"Пакування та розгортання Avalonia додатків","\u002Fcsharp\u002Fdesktop-ui\u002Favalonia-packaging-deployment","01.csharp\u002F12.desktop-ui\u002F46.avalonia-packaging-deployment",{"title":1651,"path":1652,"stem":1653},"Розгортання WPF застосунків","\u002Fcsharp\u002Fdesktop-ui\u002Fwpf-packaging-deployment","01.csharp\u002F12.desktop-ui\u002F47.wpf-packaging-deployment",{"title":1655,"icon":658,"path":1656,"stem":1657,"children":1658,"page":59},"Network Programming","\u002Fcsharp\u002Fnetwork-programming","01.csharp\u002F13.network-programming",[1659,1663,1667,1671,1675,1679,1683,1687,1691,1695,1699],{"title":1660,"path":1661,"stem":1662},"Основи комп'ютерних мереж","\u002Fcsharp\u002Fnetwork-programming\u002Ffoundations","01.csharp\u002F13.network-programming\u002F01.foundations",{"title":1664,"path":1665,"stem":1666},"Модель OSI та стек TCP\u002FIP","\u002Fcsharp\u002Fnetwork-programming\u002Fosi-model","01.csharp\u002F13.network-programming\u002F02.osi-model",{"title":1668,"path":1669,"stem":1670},"IP-протокол та адресація","\u002Fcsharp\u002Fnetwork-programming\u002Fip-addressing","01.csharp\u002F13.network-programming\u002F03.ip-addressing",{"title":1672,"path":1673,"stem":1674},"UDP — протокол без з'єднання","\u002Fcsharp\u002Fnetwork-programming\u002Fudp","01.csharp\u002F13.network-programming\u002F05.udp",{"title":1676,"path":1677,"stem":1678},"UDP Broadcast та Multicast","\u002Fcsharp\u002Fnetwork-programming\u002Fudp-broadcast-multicast","01.csharp\u002F13.network-programming\u002F06.udp-broadcast-multicast",{"title":1680,"path":1681,"stem":1682},"HTTP — протокол вебу","\u002Fcsharp\u002Fnetwork-programming\u002Fhttp-fundamentals","01.csharp\u002F13.network-programming\u002F07.http-fundamentals",{"title":1684,"path":1685,"stem":1686},"HttpListener — вбудований HTTP-сервер .NET","\u002Fcsharp\u002Fnetwork-programming\u002F07a.http-listener","01.csharp\u002F13.network-programming\u002F07a.http-listener",{"title":1688,"path":1689,"stem":1690},"HTTP Advanced — cookies, аутентифікація та HTTPS","\u002Fcsharp\u002Fnetwork-programming\u002Fhttp-advanced","01.csharp\u002F13.network-programming\u002F08.http-advanced",{"title":1692,"path":1693,"stem":1694},"SMTP та протоколи електронної пошти","\u002Fcsharp\u002Fnetwork-programming\u002Fsmtp","01.csharp\u002F13.network-programming\u002F09.smtp",{"title":1696,"path":1697,"stem":1698},"WebSocket — повнодуплексний протокол реального часу","\u002Fcsharp\u002Fnetwork-programming\u002Fwebsockets","01.csharp\u002F13.network-programming\u002F10.websockets",{"title":1700,"path":1701,"stem":1702},"TLS\u002FSSL — криптографічний захист мережевих з'єднань","\u002Fcsharp\u002Fnetwork-programming\u002Ftls-ssl","01.csharp\u002F13.network-programming\u002F11.tls-ssl",{"title":1704,"path":1705,"stem":1706},"C# & .NET: The Ultimate Roadmap","\u002Fcsharp\u002Froadmap","01.csharp\u002Froadmap",{"title":1708,"icon":1709,"path":1710,"stem":1711,"children":1712,"page":59},"C++","i-devicon-cplusplus","\u002Fcpp","02.cpp",[1713,1717,1721,1725,1729,1733,1737,1741,1745,1748,1752,1756,1760,1764,1768,1772,1776,1780,1784,1788,1792,1796,1800,1804,1808,1812,1816,1820,1824,1828,1832,1836,1840,1844,1848,1852,1856,1860,1864,1868,1872,1876,1880,1884,1888,1892,1896,1900,1904,1908,1912,1916,1920,1924,1928,1932,1936,1940,1944,1948,1952,1956,1960,1964,1968,1972,1976,1980,1984,1988,1992,1996,2000,2004,2008,2012,2016,2020,2024,2028,2032,2036,2040,2044,2048,2052,2056,2060,2064,2068,2072,2076,2080],{"title":1714,"path":1715,"stem":1716},"Вступ у програмування та алгоритми","\u002Fcpp\u002Fintro-algorithms","02.cpp\u002F01.intro-algorithms",{"title":1718,"path":1719,"stem":1720},"Code Style: угоди про оформлення коду","\u002Fcpp\u002Fcode-style","02.cpp\u002F02.code-style",{"title":1722,"path":1723,"stem":1724},"Середовище розробки та перший проєкт","\u002Fcpp\u002Fide-setup","02.cpp\u002F03.ide-setup",{"title":1726,"path":1727,"stem":1728},"Вивід даних на екран","\u002Fcpp\u002Fdata-output","02.cpp\u002F04.data-output",{"title":1730,"path":1731,"stem":1732},"Типи даних, змінні та константи","\u002Fcpp\u002Fdata-types-variables","02.cpp\u002F05.data-types-variables",{"title":1734,"path":1735,"stem":1736},"Ввід даних з клавіатури","\u002Fcpp\u002Fdata-input","02.cpp\u002F06.data-input",{"title":1738,"path":1739,"stem":1740},"Оператори, перетворення типів та логічні операції","\u002Fcpp\u002Foperators-type-conversion","02.cpp\u002F07.operators-type-conversion",{"title":1742,"path":1743,"stem":1744},"Цикли","\u002Fcpp\u002Floops","02.cpp\u002F08.loops",{"title":32,"path":1746,"stem":1747},"\u002Fcpp\u002Farrays","02.cpp\u002F09.arrays",{"title":1749,"path":1750,"stem":1751},"Алгоритми сортування та аналіз складності","\u002Fcpp\u002Fsorting","02.cpp\u002F10.sorting",{"title":1753,"path":1754,"stem":1755},"Алгоритми пошуку","\u002Fcpp\u002Fsearching","02.cpp\u002F11.searching",{"title":1757,"path":1758,"stem":1759},"Функції: основи","\u002Fcpp\u002Ffunctions-basics","02.cpp\u002F12.functions-basics",{"title":1761,"path":1762,"stem":1763},"Функції: прототипи, область видимості та додаткові можливості","\u002Fcpp\u002Ffunctions-scope","02.cpp\u002F13.functions-scope",{"title":1765,"path":1766,"stem":1767},"Функції: перевантаження та шаблони","\u002Fcpp\u002Ffunctions-overloading-templates","02.cpp\u002F14.functions-overloading-templates",{"title":1769,"path":1770,"stem":1771},"Вказівники: основи","\u002Fcpp\u002Fpointers-basics","02.cpp\u002F15.pointers-basics",{"title":1773,"path":1774,"stem":1775},"Посилання (References)","\u002Fcpp\u002Freferences","02.cpp\u002F16.references",{"title":1777,"path":1778,"stem":1779},"Вказівники, const і масиви","\u002Fcpp\u002Fpointers-const-arrays","02.cpp\u002F17.pointers-const-arrays",{"title":1781,"path":1782,"stem":1783},"Адресна арифметика","\u002Fcpp\u002Fpointer-arithmetic","02.cpp\u002F18.pointer-arithmetic",{"title":1785,"path":1786,"stem":1787},"Динамічна пам'ять","\u002Fcpp\u002Fdynamic-memory","02.cpp\u002F19.dynamic-memory",{"title":1789,"path":1790,"stem":1791},"Вказівники типу void","\u002Fcpp\u002Fvoid-pointers","02.cpp\u002F20.void-pointers",{"title":1793,"path":1794,"stem":1795},"Вказівники на вказівники","\u002Fcpp\u002Fpointers-to-pointers","02.cpp\u002F21.pointers-to-pointers",{"title":1797,"path":1798,"stem":1799},"Оператор доступу до членів через вказівник (->)","\u002Fcpp\u002Fmember-access-operator","02.cpp\u002F22.member-access-operator",{"title":1801,"path":1802,"stem":1803},"Цикл for-each (Range-based for)","\u002Fcpp\u002Fforeach-loop","02.cpp\u002F23.foreach-loop",{"title":1805,"path":1806,"stem":1807},"Вказівники на функції","\u002Fcpp\u002Ffunction-pointers","02.cpp\u002F24.function-pointers",{"title":1809,"path":1810,"stem":1811},"Лямбда-вирази","\u002Fcpp\u002Flambdas","02.cpp\u002F25.lambdas",{"title":1813,"path":1814,"stem":1815},"Лямбда-захоплення","\u002Fcpp\u002Flambda-captures","02.cpp\u002F26.lambda-captures",{"title":1817,"path":1818,"stem":1819},"Еліпсис","\u002Fcpp\u002Fellipsis","02.cpp\u002F27.ellipsis",{"title":1821,"path":1822,"stem":1823},"Безпечні альтернативи еліпсису","\u002Fcpp\u002F27a.ellipsis","02.cpp\u002F27a.ellipsis",{"title":1825,"path":1826,"stem":1827},"Аргументи командного рядка","\u002Fcpp\u002Fcommand-line-arguments","02.cpp\u002F28.command-line-arguments",{"title":1829,"path":1830,"stem":1831},"Перерахування (enum)","\u002Fcpp\u002Fenum","02.cpp\u002F29.enum",{"title":1833,"path":1834,"stem":1835},"Класи-перерахування (enum class)","\u002Fcpp\u002Fenum-class","02.cpp\u002F30.enum-class",{"title":1837,"path":1838,"stem":1839},"Псевдоніми типів (typedef і using)","\u002Fcpp\u002Ftype-aliases","02.cpp\u002F31.type-aliases",{"title":1841,"path":1842,"stem":1843},"Системи числення та двійкова арифметика","\u002Fcpp\u002Fnumber-systems","02.cpp\u002F32.number-systems",{"title":1845,"path":1846,"stem":1847},"Структури (struct): агрегування даних","\u002Fcpp\u002Fstruct","02.cpp\u002F33.struct",{"title":1849,"path":1850,"stem":1851},"Структури у функціях","\u002Fcpp\u002Fstruct-functions","02.cpp\u002F34.struct-functions",{"title":1853,"path":1854,"stem":1855},"Масиви структур і вкладені структури","\u002Fcpp\u002Fstruct-arrays","02.cpp\u002F35.struct-arrays",{"title":1857,"path":1858,"stem":1859},"Патерни struct та межі застосування","\u002Fcpp\u002Fstruct-patterns","02.cpp\u002F36.struct-patterns",{"title":1861,"path":1862,"stem":1863},"Символи та таблиця ASCII","\u002Fcpp\u002Fascii-characters","02.cpp\u002F37.ascii-characters",{"title":1865,"path":1866,"stem":1867},"Unicode та кодування UTF","\u002Fcpp\u002Funicode-utf","02.cpp\u002F38.unicode-utf",{"title":1869,"path":1870,"stem":1871},"C-style рядки","\u002Fcpp\u002Fc-strings","02.cpp\u002F39.c-strings",{"title":1873,"path":1874,"stem":1875},"Вступ до std::string","\u002Fcpp\u002Fstd-string-intro","02.cpp\u002F40.std-string-intro",{"title":1877,"path":1878,"stem":1879},"Довжина, ємність та доступ до символів std::string","\u002Fcpp\u002Fstd-string-capacity-access","02.cpp\u002F41.std-string-capacity-access",{"title":1881,"path":1882,"stem":1883},"Модифікація std::string: присвоювання, додавання, вставка, видалення та заміна","\u002Fcpp\u002Fstd-string-modification","02.cpp\u002F42.std-string-modification",{"title":1885,"path":1886,"stem":1887},"Пошук у std::string: find, npos та практичні патерни","\u002Fcpp\u002Fstd-string-search","02.cpp\u002F43.std-string-search",{"title":1889,"path":1890,"stem":1891},"std::string_view: невласницький погляд на рядок без копіювання","\u002Fcpp\u002Fstd-string-view","02.cpp\u002F44.std-string-view",{"title":1893,"path":1894,"stem":1895},"Об'єднання (union): один блок пам'яті, кілька інтерпретацій","\u002Fcpp\u002Funion","02.cpp\u002F45.union",{"title":1897,"path":1898,"stem":1899},"Організація коду: файли, препроцесор, простори імен","\u002Fcpp\u002Fmultifile-programs","02.cpp\u002F46.multifile-programs",{"title":1901,"path":1902,"stem":1903},"Робота з файлами: C-стиль (stdio.h)","\u002Fcpp\u002Fc-style-files","02.cpp\u002F47.c-style-files",{"title":1905,"path":1906,"stem":1907},"Робота з файлами: C++-стиль (fstream)","\u002Fcpp\u002Fcpp-style-files","02.cpp\u002F48.cpp-style-files",{"title":1909,"path":1910,"stem":1911},"Введення в ООП: від процедур до об'єктів","\u002Fcpp\u002Foop-intro","02.cpp\u002F49.oop-intro",{"title":1913,"path":1914,"stem":1915},"Класи, об'єкти та методи","\u002Fcpp\u002Fclasses-objects-methods","02.cpp\u002F50.classes-objects-methods",{"title":1917,"path":1918,"stem":1919},"Специфікатори доступу та інкапсуляція","\u002Fcpp\u002Faccess-specifiers-encapsulation","02.cpp\u002F51.access-specifiers-encapsulation",{"title":1921,"path":1922,"stem":1923},"Конструктори: гарантована ініціалізація об'єктів","\u002Fcpp\u002Fconstructors","02.cpp\u002F52.constructors",{"title":1925,"path":1926,"stem":1927},"Списки ініціалізації та делегуючі конструктори","\u002Fcpp\u002Finit-lists-delegating","02.cpp\u002F53.init-lists-delegating",{"title":1929,"path":1930,"stem":1931},"Деструктори та ідіома RAII","\u002Fcpp\u002Fdestructors-raii","02.cpp\u002F54.destructors-raii",{"title":1933,"path":1934,"stem":1935},"Прихований покажчик this","\u002Fcpp\u002Fthis-pointer","02.cpp\u002F55.this-pointer",{"title":1937,"path":1938,"stem":1939},"Розподіл класу на заголовковий файл та файл реалізації","\u002Fcpp\u002Fclass-header-source","02.cpp\u002F56.class-header-source",{"title":1941,"path":1942,"stem":1943},"Класи та const","\u002Fcpp\u002Fconst-classes","02.cpp\u002F57.const-classes",{"title":1945,"path":1946,"stem":1947},"Статичні члени класу","\u002Fcpp\u002Fstatic-members","02.cpp\u002F58.static-members",{"title":1949,"path":1950,"stem":1951},"Конструктор копіювання та запобігання копіюванню","\u002Fcpp\u002Fcopy-constructor","02.cpp\u002F59.copy-constructor",{"title":1953,"path":1954,"stem":1955},"Дружні функції та класи","\u002Fcpp\u002Ffriend-functions-classes","02.cpp\u002F60.friend-functions-classes",{"title":1957,"path":1958,"stem":1959},"Анонімні об'єкти та вкладені типи","\u002Fcpp\u002Fanonymous-objects-nested-types","02.cpp\u002F61.anonymous-objects-nested-types",{"title":1961,"path":1962,"stem":1963},"Вимірювання часу виконання: RAII-таймер","\u002Fcpp\u002Ftiming-raii-timer","02.cpp\u002F62.timing-raii-timer",{"title":1965,"path":1966,"stem":1967},"Перевантаження операторів: основи та арифметика","\u002Fcpp\u002Foperator-overloading-basics","02.cpp\u002F63.operator-overloading-basics",{"title":1969,"path":1970,"stem":1971},"Перевантаження операторів: методи класу та спеціальні оператори","\u002Fcpp\u002Foperator-overloading-methods","02.cpp\u002F64.operator-overloading-methods",{"title":1973,"path":1974,"stem":1975},"Приведення типів, explicit та delete","\u002Fcpp\u002Ftypecast-explicit-delete","02.cpp\u002F65.typecast-explicit-delete",{"title":1977,"path":1978,"stem":1979},"Глибоке копіювання та оператор присвоювання","\u002Fcpp\u002Fdeep-copy-assignment","02.cpp\u002F66.deep-copy-assignment",{"title":1981,"path":1982,"stem":1983},"Зв'язки між об'єктами: композиція та агрегація","\u002Fcpp\u002Fcomposition-aggregation","02.cpp\u002F67.composition-aggregation",{"title":1985,"path":1986,"stem":1987},"Асоціація та залежність між об","\u002Fcpp\u002Fassociation-dependency","02.cpp\u002F68.association-dependency",{"title":1989,"path":1990,"stem":1991},"Контейнерні класи та `std::initializer_list`","\u002Fcpp\u002Fcontainer-classes-initializer-list","02.cpp\u002F69.container-classes-initializer-list",{"title":1993,"path":1994,"stem":1995},"Вступ до наслідування: базові поняття та відношення «is-a»","\u002Fcpp\u002Finheritance-intro","02.cpp\u002F70.inheritance-intro",{"title":1997,"path":1998,"stem":1999},"Конструктори, деструктори та порядок ініціалізації в ієрархіях","\u002Fcpp\u002Finheritance-constructors-destructors","02.cpp\u002F71.inheritance-constructors-destructors",{"title":2001,"path":2002,"stem":2003},"Специфікатор `protected` та види наслідування","\u002Fcpp\u002Finheritance-access-specifiers","02.cpp\u002F72.inheritance-access-specifiers",{"title":2005,"path":2006,"stem":2007},"Перевизначення методів та приховування імен","\u002Fcpp\u002Fmethod-overriding-hiding","02.cpp\u002F73.method-overriding-hiding",{"title":2009,"path":2010,"stem":2011},"Множинне наслідування та проблема ромба","\u002Fcpp\u002Fmultiple-inheritance-diamond","02.cpp\u002F74.multiple-inheritance-diamond",{"title":2013,"path":2014,"stem":2015},"Вказівники на базовий клас та віртуальні функції","\u002Fcpp\u002Fvirtual-functions-polymorphism","02.cpp\u002F75.virtual-functions-polymorphism",{"title":2017,"path":2018,"stem":2019},"Специфікатори override, final та віртуальні деструктори","\u002Fcpp\u002Foverride-final-virtual-destructors","02.cpp\u002F76.override-final-virtual-destructors",{"title":2021,"path":2022,"stem":2023},"Внутрішні механізми поліморфізму: vtable, vptr та зв'язування","\u002Fcpp\u002Fvirtual-table-binding","02.cpp\u002F77.virtual-table-binding",{"title":2025,"path":2026,"stem":2027},"Абстрактні класи та чисті віртуальні функції","\u002Fcpp\u002Fabstract-classes-interfaces","02.cpp\u002F78.abstract-classes-interfaces",{"title":2029,"path":2030,"stem":2031},"Зрізка об'єктів, віртуальне наслідування та поліморфний вивід","\u002Fcpp\u002Fslicing-virtual-base-io","02.cpp\u002F79.slicing-virtual-base-io",{"title":2033,"path":2034,"stem":2035},"Динамічне приведення типів (`dynamic_cast`) та RTTI","\u002Fcpp\u002Fdynamic-cast-rtti","02.cpp\u002F80.dynamic-cast-rtti",{"title":2037,"path":2038,"stem":2039},"Шаблони функцій: узагальнене програмування","\u002Fcpp\u002Ffunction-templates","02.cpp\u002F81.function-templates",{"title":2041,"path":2042,"stem":2043},"Шаблони класів та NTTP-параметри","\u002Fcpp\u002Fclass-templates-nttp","02.cpp\u002F82.class-templates-nttp",{"title":2045,"path":2046,"stem":2047},"Повна (явна) спеціалізація шаблонів","\u002Fcpp\u002Ffull-template-specialization","02.cpp\u002F83.full-template-specialization",{"title":2049,"path":2050,"stem":2051},"Часткова спеціалізація шаблонів класів","\u002Fcpp\u002Fpartial-template-specialization","02.cpp\u002F84.partial-template-specialization",{"title":2053,"path":2054,"stem":2055},"Механізм винятків: try, catch, throw та розкручування стеку","\u002Fcpp\u002Fexceptions-basics","02.cpp\u002F85.exceptions-basics",{"title":2057,"path":2058,"stem":2059},"Ієрархія std::exception та власні класи винятків","\u002Fcpp\u002Fexception-hierarchy-custom","02.cpp\u002F86.exception-hierarchy-custom",{"title":2061,"path":2062,"stem":2063},"Повторне генерування винятків та функціональний try-блок","\u002Fcpp\u002Frethrow-function-try","02.cpp\u002F87.rethrow-function-try",{"title":2065,"path":2066,"stem":2067},"Безпека винятків, noexcept та деструктори","\u002Fcpp\u002Fsafety-noexcept","02.cpp\u002F88.safety-noexcept",{"title":2069,"path":2070,"stem":2071},"Стаття 89 — Власні структури даних: стек та черга","\u002Fcpp\u002Fcustom-stack-queue","02.cpp\u002F89.custom-stack-queue",{"title":2073,"path":2074,"stem":2075},"Стаття 90 — Власні структури даних: однозв'язаний список","\u002Fcpp\u002Flinked-list","02.cpp\u002F90.linked-list",{"title":2077,"path":2078,"stem":2079},"Стаття 91 — Власні структури даних: бінарне дерево пошуку","\u002Fcpp\u002Fbinary-search-tree","02.cpp\u002F91.binary-search-tree",{"title":2081,"path":2082,"stem":2083},"Стаття 92 — Власні структури даних: хеш-таблиця","\u002Fcpp\u002Fhash-table","02.cpp\u002F92.hash-table",{"title":2085,"icon":2086,"path":2087,"stem":2088,"children":2089,"page":59},"JavaScript","i-devicon-javascript","\u002Fjavascript","03.javascript",[2090,2116,2170,2192,2496,2534,2552],{"title":2091,"icon":2092,"path":2093,"stem":2094,"children":2095,"page":59},"Events","i-lucide-mouse-pointer-click","\u002Fjavascript\u002Fevents","03.javascript\u002F01.events",[2096,2100,2104,2108,2112],{"title":2097,"path":2098,"stem":2099},"Вступ до подій браузера","\u002Fjavascript\u002Fevents\u002Fintro","03.javascript\u002F01.events\u002F01.intro",{"title":2101,"path":2102,"stem":2103},"Бульбашковий механізм (Bubbling) та занурення (Capturing)","\u002Fjavascript\u002Fevents\u002Fbubbling-capturing","03.javascript\u002F01.events\u002F02.bubbling-capturing",{"title":2105,"path":2106,"stem":2107},"Делегування подій (Event Delegation)","\u002Fjavascript\u002Fevents\u002Fdelegate-events","03.javascript\u002F01.events\u002F03.delegate-events",{"title":2109,"path":2110,"stem":2111},"Типові дії браузера та preventDefault()","\u002Fjavascript\u002Fevents\u002Fprevent-default","03.javascript\u002F01.events\u002F04.prevent-default",{"title":2113,"path":2114,"stem":2115},"Запуск користувацьких подій (Custom Events)","\u002Fjavascript\u002Fevents\u002Fcustom-events","03.javascript\u002F01.events\u002F05.custom-events",{"title":2117,"icon":2118,"path":2119,"stem":2120,"children":2121,"page":59},"Network","i-lucide-globe","\u002Fjavascript\u002Fnetwork","03.javascript\u002F02.network",[2122,2126,2130,2134,2138,2142,2146,2150,2154,2158,2162,2166],{"title":2123,"path":2124,"stem":2125},"Fetch API - Сучасний підхід до HTTP-запитів","\u002Fjavascript\u002Fnetwork\u002F01-fetch-api","03.javascript\u002F02.network\u002F01-fetch-api",{"title":2127,"path":2128,"stem":2129},"FormData - Робота з формами та файлами","\u002Fjavascript\u002Fnetwork\u002F02-formdata","03.javascript\u002F02.network\u002F02-formdata",{"title":2131,"path":2132,"stem":2133},"Відстеження прогресу завантаження","\u002Fjavascript\u002Fnetwork\u002F03-download-progress","03.javascript\u002F02.network\u002F03-download-progress",{"title":2135,"path":2136,"stem":2137},"Переривання fetch-запитів","\u002Fjavascript\u002Fnetwork\u002F04-abort-requests","03.javascript\u002F02.network\u002F04-abort-requests",{"title":2139,"path":2140,"stem":2141},"CORS - Запити між різними джерелами","\u002Fjavascript\u002Fnetwork\u002F05-cors","03.javascript\u002F02.network\u002F05-cors",{"title":2143,"path":2144,"stem":2145},"Fetch API - Повний довідник опцій","\u002Fjavascript\u002Fnetwork\u002F06-fetch-options","03.javascript\u002F02.network\u002F06-fetch-options",{"title":2147,"path":2148,"stem":2149},"URL Objects - Робота з посиланнями","\u002Fjavascript\u002Fnetwork\u002F07-url-objects","03.javascript\u002F02.network\u002F07-url-objects",{"title":2151,"path":2152,"stem":2153},"XMLHttpRequest - AJAX та низькорівневі запити","\u002Fjavascript\u002Fnetwork\u002F08-xmlhttprequest","03.javascript\u002F02.network\u002F08-xmlhttprequest",{"title":2155,"path":2156,"stem":2157},"Відновлюване завантаження файлів","\u002Fjavascript\u002Fnetwork\u002F09-resumable-upload","03.javascript\u002F02.network\u002F09-resumable-upload",{"title":2159,"path":2160,"stem":2161},"Cookies, document.cookie та світ після \"Cookiepocalypse\"","\u002Fjavascript\u002Fnetwork\u002F10-cookies","03.javascript\u002F02.network\u002F10-cookies",{"title":2163,"path":2164,"stem":2165},"js-cookie: Керування Cookies без Болю","\u002Fjavascript\u002Fnetwork\u002F11-js-cookie","03.javascript\u002F02.network\u002F11-js-cookie",{"title":2167,"path":2168,"stem":2169},"Axios: Потужний HTTP-клієнт для JavaScript","\u002Fjavascript\u002Fnetwork\u002F12-axios","03.javascript\u002F02.network\u002F12-axios",{"title":2171,"icon":2172,"path":2173,"stem":2174,"children":2175,"page":59},"Bom","i-lucide-monitor","\u002Fjavascript\u002Fbom","03.javascript\u002F03.bom",[2176,2180,2184,2188],{"title":2177,"path":2178,"stem":2179},"LocalStorage, SessionStorage та patterns збереження даних","\u002Fjavascript\u002Fbom\u002F01-localstorage","03.javascript\u002F03.bom\u002F01-localstorage",{"title":2181,"path":2182,"stem":2183},"Location Object - Керування адресою сторінки","\u002Fjavascript\u002Fbom\u002F02-location-object","03.javascript\u002F03.bom\u002F02-location-object",{"title":2185,"path":2186,"stem":2187},"History API - Керування історією браузера","\u002Fjavascript\u002Fbom\u002F03-history-api","03.javascript\u002F03.bom\u002F03-history-api",{"title":2189,"path":2190,"stem":2191},"Navigator Object - Ідентифікація та Можливості Пристрою","\u002Fjavascript\u002Fbom\u002F04-navigator-object","03.javascript\u002F03.bom\u002F04-navigator-object",{"title":2193,"icon":2194,"path":2195,"stem":2196,"children":2197},"React","i-devicon-react","\u002Fjavascript\u002Freact","03.javascript\u002F04.react\u002Findex",[2198,2199,2203,2207,2211,2215,2278,2313,2465],{"title":2193,"path":2195,"stem":2196},{"title":2200,"path":2201,"stem":2202},"Робота з Формами в React","\u002Fjavascript\u002Freact\u002Freact-forms","03.javascript\u002F04.react\u002F01.react-forms",{"title":2204,"path":2205,"stem":2206},"React Hook Form: Професійна Робота з Формами","\u002Fjavascript\u002Freact\u002Freact-hook-form","03.javascript\u002F04.react\u002F02.react-hook-form",{"title":2208,"path":2209,"stem":2210},"React Hook Form: Глибоке Розуміння Архітектури та Оптимізації","\u002Fjavascript\u002Freact\u002Freact-hook-form-new","03.javascript\u002F04.react\u002F02.react-hook-form-new",{"title":2212,"path":2213,"stem":2214},"Axios та React: Професійна Архітектура Запитів","\u002Fjavascript\u002Freact\u002Fdata-fetching-axios","03.javascript\u002F04.react\u002F03.data-fetching-axios",{"title":2216,"icon":132,"path":2217,"stem":2218,"children":2219},"Tanstack Query","\u002Fjavascript\u002Freact\u002Ftanstack-query","03.javascript\u002F04.react\u002F04.tanstack-query\u002Findex",[2220,2222,2226,2230,2234,2238,2242,2246,2250,2254,2258,2262,2266,2270,2274],{"title":2221,"path":2217,"stem":2218},"TanStack Query: Майстерність Керування Станом Сервера",{"title":2223,"path":2224,"stem":2225},"Парадигма Server State: Чому useEffect недостатньо","\u002Fjavascript\u002Freact\u002Ftanstack-query\u002Fserver-state-paradigm","03.javascript\u002F04.react\u002F04.tanstack-query\u002F01.server-state-paradigm",{"title":2227,"path":2228,"stem":2229},"Встановлення та Налаштування: Фундамент","\u002Fjavascript\u002Freact\u002Ftanstack-query\u002Finstallation-and-devtools","03.javascript\u002F04.react\u002F04.tanstack-query\u002F02.installation-and-devtools",{"title":2231,"path":2232,"stem":2233},"Основи Запитів та Магія Ключів","\u002Fjavascript\u002Freact\u002Ftanstack-query\u002Fquery-basics-and-keys","03.javascript\u002F04.react\u002F04.tanstack-query\u002F03.query-basics-and-keys",{"title":2235,"path":2236,"stem":2237},"Синхронізація Даних: Життєвий Цикл Запиту","\u002Fjavascript\u002Freact\u002Ftanstack-query\u002Fdata-synchronization","03.javascript\u002F04.react\u002F04.tanstack-query\u002F04.data-synchronization",{"title":2239,"path":2240,"stem":2241},"Мутації та Інвалідація: Зміна Даних","\u002Fjavascript\u002Freact\u002Ftanstack-query\u002Fmutations-and-invalidation","03.javascript\u002F04.react\u002F04.tanstack-query\u002F05.mutations-and-invalidation",{"title":2243,"path":2244,"stem":2245},"Оптимістичні Оновлення: Швидше за Світло","\u002Fjavascript\u002Freact\u002Ftanstack-query\u002Foptimistic-updates","03.javascript\u002F04.react\u002F04.tanstack-query\u002F06.optimistic-updates",{"title":2247,"path":2248,"stem":2249},"Пагінація та Infinite Scroll","\u002Fjavascript\u002Freact\u002Ftanstack-query\u002Fpagination-and-load-more","03.javascript\u002F04.react\u002F04.tanstack-query\u002F07.pagination-and-load-more",{"title":2251,"path":2252,"stem":2253},"Просунуті Патерни та Оптимізація","\u002Fjavascript\u002Freact\u002Ftanstack-query\u002Fadvanced-patterns","03.javascript\u002F04.react\u002F04.tanstack-query\u002F08.advanced-patterns",{"title":2255,"path":2256,"stem":2257},"Архітектура та Best Practices","\u002Fjavascript\u002Freact\u002Ftanstack-query\u002Farchitecture-and-best-practices","03.javascript\u002F04.react\u002F04.tanstack-query\u002F09.architecture-and-best-practices",{"title":2259,"path":2260,"stem":2261},"Server-Side Rendering (SSR) та Гідратація","\u002Fjavascript\u002Freact\u002Ftanstack-query\u002Fserver-side-rendering","03.javascript\u002F04.react\u002F04.tanstack-query\u002F10.server-side-rendering",{"title":2263,"path":2264,"stem":2265},"Стратегії Тестування","\u002Fjavascript\u002Freact\u002Ftanstack-query\u002Ftesting-strategies","03.javascript\u002F04.react\u002F04.tanstack-query\u002F11.testing-strategies",{"title":2267,"path":2268,"stem":2269},"Аутентифікація та Обробка Помилок","\u002Fjavascript\u002Freact\u002Ftanstack-query\u002Fauthentication-and-errors","03.javascript\u002F04.react\u002F04.tanstack-query\u002F12.authentication-and-errors",{"title":2271,"path":2272,"stem":2273},"React Suspense та Майбутнє","\u002Fjavascript\u002Freact\u002Ftanstack-query\u002Freact-suspense","03.javascript\u002F04.react\u002F04.tanstack-query\u002F13.react-suspense",{"title":2275,"path":2276,"stem":2277},"Глибоке Занурення в Продуктивність","\u002Fjavascript\u002Freact\u002Ftanstack-query\u002Fperformance-deep-dive","03.javascript\u002F04.react\u002F04.tanstack-query\u002F14.performance-deep-dive",{"title":2279,"icon":2194,"path":2280,"stem":2281,"children":2282},"React Router","\u002Fjavascript\u002Freact\u002Freact-router","03.javascript\u002F04.react\u002F05.react-router\u002Findex",[2283,2285,2289,2293,2297,2301,2305,2309],{"title":2284,"path":2280,"stem":2281},"React Router: Навігаційна система сучасного вебу",{"title":2286,"path":2287,"stem":2288},"Налаштування та Базовий Роутинг","\u002Fjavascript\u002Freact\u002Freact-router\u002Fsetup-and-basic-routing","03.javascript\u002F04.react\u002F05.react-router\u002F01.setup-and-basic-routing",{"title":2290,"path":2291,"stem":2292},"Динамічна Навігація","\u002Fjavascript\u002Freact\u002Freact-router\u002Fnavigation-and-links","03.javascript\u002F04.react\u002F05.react-router\u002F02.navigation-and-links",{"title":2294,"path":2295,"stem":2296},"Вкладені Маршрути та Макети","\u002Fjavascript\u002Freact\u002Freact-router\u002Fnested-routes-and-layouts","03.javascript\u002F04.react\u002F05.react-router\u002F03.nested-routes-and-layouts",{"title":2298,"path":2299,"stem":2300},"Динамічні Маршрути та Параметри","\u002Fjavascript\u002Freact\u002Freact-router\u002Fdynamic-routing","03.javascript\u002F04.react\u002F05.react-router\u002F04.dynamic-routing",{"title":2302,"path":2303,"stem":2304},"Data APIs: Loaders та Actions","\u002Fjavascript\u002Freact\u002Freact-router\u002Fdata-loading","03.javascript\u002F04.react\u002F05.react-router\u002F05.data-loading",{"title":2306,"path":2307,"stem":2308},"Просунуті Патерни","\u002Fjavascript\u002Freact\u002Freact-router\u002Fadvanced-patterns","03.javascript\u002F04.react\u002F05.react-router\u002F06.advanced-patterns",{"title":2310,"path":2311,"stem":2312},"Legacy Routing: Компонентний підхід","\u002Fjavascript\u002Freact\u002Freact-router\u002Flegacy-routing","03.javascript\u002F04.react\u002F05.react-router\u002F07.legacy-routing",{"title":2314,"icon":132,"path":2315,"stem":2316,"children":2317},"Redux","\u002Fjavascript\u002Freact\u002Fredux","03.javascript\u002F04.react\u002F06.redux\u002Findex",[2318,2320,2336,2365,2374,2395,2411,2440],{"title":2319,"path":2315,"stem":2316},"Redux: Еволюція управління станом",{"title":14,"icon":15,"path":2321,"stem":2322,"children":2323,"page":59},"\u002Fjavascript\u002Freact\u002Fredux\u002Ffundamentals","03.javascript\u002F04.react\u002F06.redux\u002F01.fundamentals",[2324,2328,2332],{"title":2325,"path":2326,"stem":2327},"Вступ до State Management","\u002Fjavascript\u002Freact\u002Fredux\u002Ffundamentals\u002Fintro-state-management","03.javascript\u002F04.react\u002F06.redux\u002F01.fundamentals\u002F01.intro-state-management",{"title":2329,"path":2330,"stem":2331},"Філософія Redux та Три Принципи","\u002Fjavascript\u002Freact\u002Fredux\u002Ffundamentals\u002Fredux-philosophy","03.javascript\u002F04.react\u002F06.redux\u002F01.fundamentals\u002F02.redux-philosophy",{"title":2333,"path":2334,"stem":2335},"Чисті функції та Іммутабельність","\u002Fjavascript\u002Freact\u002Fredux\u002Ffundamentals\u002Fpure-functions-immutability","03.javascript\u002F04.react\u002F06.redux\u002F01.fundamentals\u002F03.pure-functions-immutability",{"title":2337,"icon":132,"path":2338,"stem":2339,"children":2340,"page":59},"Classic Redux","\u002Fjavascript\u002Freact\u002Fredux\u002Fclassic-redux","03.javascript\u002F04.react\u002F06.redux\u002F02.classic-redux",[2341,2345,2349,2353,2357,2361],{"title":2342,"path":2343,"stem":2344},"Створення Store (Classic Redux)","\u002Fjavascript\u002Freact\u002Fredux\u002Fclassic-redux\u002Fstore-setup","03.javascript\u002F04.react\u002F06.redux\u002F02.classic-redux\u002F01.store-setup",{"title":2346,"path":2347,"stem":2348},"Actions, Constants та Action Creators","\u002Fjavascript\u002Freact\u002Fredux\u002Fclassic-redux\u002Factions-constants","03.javascript\u002F04.react\u002F06.redux\u002F02.classic-redux\u002F02.actions-constants",{"title":2350,"path":2351,"stem":2352},"Логіка Reducers","\u002Fjavascript\u002Freact\u002Fredux\u002Fclassic-redux\u002Freducers","03.javascript\u002F04.react\u002F06.redux\u002F02.classic-redux\u002F03.reducers",{"title":2354,"path":2355,"stem":2356},"Комбінування Reducers (Root Reducer)","\u002Fjavascript\u002Freact\u002Fredux\u002Fclassic-redux\u002Fdata-flow","03.javascript\u002F04.react\u002F06.redux\u002F02.classic-redux\u002F04.data-flow",{"title":2358,"path":2359,"stem":2360},"Підключення до React (React-Redux)","\u002Fjavascript\u002Freact\u002Fredux\u002Fclassic-redux\u002Freact-redux-connection","03.javascript\u002F04.react\u002F06.redux\u002F02.classic-redux\u002F05.react-redux-connection",{"title":2362,"path":2363,"stem":2364},"Middleware та Асинхронність (Redux Thunk)","\u002Fjavascript\u002Freact\u002Fredux\u002Fclassic-redux\u002Fmiddleware-thunk","03.javascript\u002F04.react\u002F06.redux\u002F02.classic-redux\u002F06.middleware-thunk",{"title":2366,"icon":132,"path":2367,"stem":2368,"children":2369,"page":59},"Transition To Rtk","\u002Fjavascript\u002Freact\u002Fredux\u002Ftransition-to-rtk","03.javascript\u002F04.react\u002F06.redux\u002F03.transition-to-rtk",[2370],{"title":2371,"path":2372,"stem":2373},"Проблеми класичного Redux","\u002Fjavascript\u002Freact\u002Fredux\u002Ftransition-to-rtk\u002Fproblems-with-classic","03.javascript\u002F04.react\u002F06.redux\u002F03.transition-to-rtk\u002F01.problems-with-classic",{"title":2375,"icon":132,"path":2376,"stem":2377,"children":2378,"page":59},"Redux Toolkit","\u002Fjavascript\u002Freact\u002Fredux\u002Fredux-toolkit","03.javascript\u002F04.react\u002F06.redux\u002F04.redux-toolkit",[2379,2383,2387,2391],{"title":2380,"path":2381,"stem":2382},"Налаштування Store з configureStore","\u002Fjavascript\u002Freact\u002Fredux\u002Fredux-toolkit\u002Fconfigure-store","03.javascript\u002F04.react\u002F06.redux\u002F04.redux-toolkit\u002F01.configure-store",{"title":2384,"path":2385,"stem":2386},"createSlice: Революція в Redux","\u002Fjavascript\u002Freact\u002Fredux\u002Fredux-toolkit\u002Fcreate-slice","03.javascript\u002F04.react\u002F06.redux\u002F04.redux-toolkit\u002F02.create-slice",{"title":2388,"path":2389,"stem":2390},"Асинхронність з createAsyncThunk","\u002Fjavascript\u002Freact\u002Fredux\u002Fredux-toolkit\u002Fasync-thunks","03.javascript\u002F04.react\u002F06.redux\u002F04.redux-toolkit\u002F03.async-thunks",{"title":2392,"path":2393,"stem":2394},"04. Entity Adapter: Керування нормалізованим станом","\u002Fjavascript\u002Freact\u002Fredux\u002Fredux-toolkit\u002Fentity-adapter","03.javascript\u002F04.react\u002F06.redux\u002F04.redux-toolkit\u002F04.entity-adapter",{"title":2396,"icon":92,"path":2397,"stem":2398,"children":2399,"page":59},"Advanced","\u002Fjavascript\u002Freact\u002Fredux\u002Fadvanced","03.javascript\u002F04.react\u002F06.redux\u002F05.advanced",[2400,2404,2408],{"title":2401,"path":2402,"stem":2403},"Мемоізація та Селектори: Повний Гайд по Reselect","\u002Fjavascript\u002Freact\u002Fredux\u002Fadvanced\u002Fselectors-reselect","03.javascript\u002F04.react\u002F06.redux\u002F05.advanced\u002F01.selectors-reselect",{"title":2405,"path":2406,"stem":2407},"RTK Query: Архітектура Серверного Кешу","\u002Fjavascript\u002Freact\u002Fredux\u002Fadvanced\u002Frtk-query-intro","03.javascript\u002F04.react\u002F06.redux\u002F05.advanced\u002F02.rtk-query-intro",{"title":2255,"path":2409,"stem":2410},"\u002Fjavascript\u002Freact\u002Fredux\u002Fadvanced\u002Farchitecture-best-practices","03.javascript\u002F04.react\u002F06.redux\u002F05.advanced\u002F03.architecture-best-practices",{"title":2412,"icon":132,"path":2413,"stem":2414,"children":2415,"page":59},"Project Kanban","\u002Fjavascript\u002Freact\u002Fredux\u002Fproject-kanban","03.javascript\u002F04.react\u002F06.redux\u002F06.project-kanban",[2416,2420,2424,2428,2432,2436],{"title":2417,"path":2418,"stem":2419},"Проєкт: Kanban Board (Trello Clone)","\u002Fjavascript\u002Freact\u002Fredux\u002Fproject-kanban\u002Fproject-overview","03.javascript\u002F04.react\u002F06.redux\u002F06.project-kanban\u002F01.project-overview",{"title":2421,"path":2422,"stem":2423},"Налаштування та Типізація","\u002Fjavascript\u002Freact\u002Fredux\u002Fproject-kanban\u002Fsetup-and-types","03.javascript\u002F04.react\u002F06.redux\u002F06.project-kanban\u002F02.setup-and-types",{"title":2425,"path":2426,"stem":2427},"Board Slice: Серце Дошки","\u002Fjavascript\u002Freact\u002Fredux\u002Fproject-kanban\u002Fboard-slice","03.javascript\u002F04.react\u002F06.redux\u002F06.project-kanban\u002F03.board-slice",{"title":2429,"path":2430,"stem":2431},"Логіка Drag & Drop","\u002Fjavascript\u002Freact\u002Fredux\u002Fproject-kanban\u002Fdrag-and-drop-logic","03.javascript\u002F04.react\u002F06.redux\u002F06.project-kanban\u002F04.drag-and-drop-logic",{"title":2433,"path":2434,"stem":2435},"Інтеграція з RTK Query","\u002Fjavascript\u002Freact\u002Fredux\u002Fproject-kanban\u002Frtk-query-integration","03.javascript\u002F04.react\u002F06.redux\u002F06.project-kanban\u002F05.rtk-query-integration",{"title":2437,"path":2438,"stem":2439},"Optimistic Updates","\u002Fjavascript\u002Freact\u002Fredux\u002Fproject-kanban\u002Foptimistic-updates","03.javascript\u002F04.react\u002F06.redux\u002F06.project-kanban\u002F06.optimistic-updates",{"title":2441,"icon":132,"path":2442,"stem":2443,"children":2444,"page":59},"Testing","\u002Fjavascript\u002Freact\u002Fredux\u002Ftesting","03.javascript\u002F04.react\u002F06.redux\u002F07.testing",[2445,2449,2453,2457,2461],{"title":2446,"path":2447,"stem":2448},"Тестування Redux","\u002Fjavascript\u002Freact\u002Fredux\u002Ftesting\u002Fintro-testing","03.javascript\u002F04.react\u002F06.redux\u002F07.testing\u002F01.intro-testing",{"title":2450,"path":2451,"stem":2452},"Тестування Reducers","\u002Fjavascript\u002Freact\u002Fredux\u002Ftesting\u002Ftesting-reducers","03.javascript\u002F04.react\u002F06.redux\u002F07.testing\u002F02.testing-reducers",{"title":2454,"path":2455,"stem":2456},"Тестування Селекторів","\u002Fjavascript\u002Freact\u002Fredux\u002Ftesting\u002Ftesting-selectors","03.javascript\u002F04.react\u002F06.redux\u002F07.testing\u002F03.testing-selectors",{"title":2458,"path":2459,"stem":2460},"Тестування Компонентів (Integration)","\u002Fjavascript\u002Freact\u002Fredux\u002Ftesting\u002Ftesting-components","03.javascript\u002F04.react\u002F06.redux\u002F07.testing\u002F04.testing-components",{"title":2462,"path":2463,"stem":2464},"Тестування Async Thunks","\u002Fjavascript\u002Freact\u002Fredux\u002Ftesting\u002Ftesting-thunks","03.javascript\u002F04.react\u002F06.redux\u002F07.testing\u002F05.testing-thunks",{"title":2466,"icon":132,"path":2467,"stem":2468,"children":2469},"Ui Libraries","\u002Fjavascript\u002Freact\u002Fui-libraries","03.javascript\u002F04.react\u002F07.ui-libraries\u002Findex",[2470,2472,2476,2480,2484,2488,2492],{"title":2471,"path":2467,"stem":2468},"UI Бібліотеки в React",{"title":2473,"path":2474,"stem":2475},"Вступ до UI Бібліотек: Навіщо Винаходити Велосипед Двічі?","\u002Fjavascript\u002Freact\u002Fui-libraries\u002Fintroduction-to-ui-libraries","03.javascript\u002F04.react\u002F07.ui-libraries\u002F01.introduction-to-ui-libraries",{"title":2477,"path":2478,"stem":2479},"Філософія shadcn\u002Fui: \"Not a Component Library\"","\u002Fjavascript\u002Freact\u002Fui-libraries\u002Fshadcn-philosophy","03.javascript\u002F04.react\u002F07.ui-libraries\u002F02.shadcn-philosophy",{"title":2481,"path":2482,"stem":2483},"Установка та Налаштування shadcn\u002Fui","\u002Fjavascript\u002Freact\u002Fui-libraries\u002Fshadcn-installation","03.javascript\u002F04.react\u002F07.ui-libraries\u002F03.shadcn-installation",{"title":2485,"path":2486,"stem":2487},"Базові Компоненти shadcn\u002Fui: Фундамент Інтерфейсу","\u002Fjavascript\u002Freact\u002Fui-libraries\u002Fshadcn-components-basics","03.javascript\u002F04.react\u002F07.ui-libraries\u002F04.shadcn-components-basics",{"title":2489,"path":2490,"stem":2491},"Компоненти Форм: Побудова Інтерактивних Form","\u002Fjavascript\u002Freact\u002Fui-libraries\u002Fshadcn-components-forms","03.javascript\u002F04.react\u002F07.ui-libraries\u002F05.shadcn-components-forms",{"title":2493,"path":2494,"stem":2495},"Складні Компоненти: Dialog, Dropdown, Table та Command","\u002Fjavascript\u002Freact\u002Fui-libraries\u002Fshadcn-components-advanced","03.javascript\u002F04.react\u002F07.ui-libraries\u002F06.shadcn-components-advanced",{"title":2497,"icon":2498,"path":2499,"stem":2500,"children":2501,"page":59},"TypeScript","i-devicon-typescript","\u002Fjavascript\u002Ftypescript","03.javascript\u002F05.typescript",[2502,2506,2510,2514,2518,2522,2526,2530],{"title":2503,"path":2504,"stem":2505},"TypeScript: Броня для вашого коду","\u002Fjavascript\u002Ftypescript\u002Fintro-and-basic-types","03.javascript\u002F05.typescript\u002F01.intro-and-basic-types",{"title":2507,"path":2508,"stem":2509},"Майстерність Моделювання Даних: Інтерфейси та Просунуті Типи","\u002Fjavascript\u002Ftypescript\u002Finterfaces-and-advanced-types","03.javascript\u002F05.typescript\u002F02.interfaces-and-advanced-types",{"title":2511,"path":2512,"stem":2513},"Алхімія Типів: Generics та Utility Types","\u002Fjavascript\u002Ftypescript\u002Fgenerics-and-utilities","03.javascript\u002F05.typescript\u002F03.generics-and-utilities",{"title":2515,"path":2516,"stem":2517},"Архітектура та Шаблони: Класи в TypeScript","\u002Fjavascript\u002Ftypescript\u002Fclasses-and-oop","03.javascript\u002F05.typescript\u002F04.classes-and-oop",{"title":2519,"path":2520,"stem":2521},"Продакшн та Екосистема: Advanced Config & Workflow","\u002Fjavascript\u002Ftypescript\u002Fadvanced-patterns-and-config","03.javascript\u002F05.typescript\u002F05.advanced-patterns-and-config",{"title":2523,"path":2524,"stem":2525},"TypeScript у світі React","\u002Fjavascript\u002Ftypescript\u002Freact-basics","03.javascript\u002F05.typescript\u002F06.react-basics",{"title":2527,"path":2528,"stem":2529},"React + TypeScript: Продвинуті патерни","\u002Fjavascript\u002Ftypescript\u002Freact-advanced","03.javascript\u002F05.typescript\u002F07.react-advanced",{"title":2531,"path":2532,"stem":2533},"React + TypeScript: Екосистема та бібліотеки","\u002Fjavascript\u002Ftypescript\u002Freact-ecosystem","03.javascript\u002F05.typescript\u002F08.react-ecosystem",{"title":2535,"icon":2536,"path":2537,"stem":2538,"children":2539,"page":59},"Асинхронність","i-lucide-refresh-cw","\u002Fjavascript\u002Fasync-programming","03.javascript\u002F06.async-programming",[2540,2544,2548],{"title":2541,"path":2542,"stem":2543},"Event Loop: серце асинхронності в JavaScript","\u002Fjavascript\u002Fasync-programming\u002Fevent-loop-fundamentals","03.javascript\u002F06.async-programming\u002F01.event-loop-fundamentals",{"title":2545,"path":2546,"stem":2547},"Task Queue та Microtask Queue: пріоритети виконання","\u002Fjavascript\u002Fasync-programming\u002Ftask-queues","03.javascript\u002F06.async-programming\u002F02.task-queues",{"title":2549,"path":2550,"stem":2551},"Асинхронні патерни: від Callbacks до Async\u002FAwait","\u002Fjavascript\u002Fasync-programming\u002Fasync-patterns","03.javascript\u002F06.async-programming\u002F03.async-patterns",{"title":2553,"path":2554,"stem":2555},"Atomic Design","\u002Fjavascript\u002Fatomic-design","03.javascript\u002F2.atomic-design",{"title":2557,"icon":2558,"path":2559,"stem":2560,"children":2561,"page":59},"Java","i-devicon-java","\u002Fjava","04.java",[2562,2565,2568,2572,2576,2580,2584],{"title":162,"path":2563,"stem":2564},"\u002Fjava\u002Fdata-mapper-part1","04.java\u002F01.data-mapper-part1",{"title":166,"path":2566,"stem":2567},"\u002Fjava\u002Fdata-mapper-part2","04.java\u002F02.data-mapper-part2",{"title":2569,"path":2570,"stem":2571},"Service Layer: Організація бізнес-логіки","\u002Fjava\u002Fservice-layer","04.java\u002F03.service-layer",{"title":2573,"path":2574,"stem":2575},"Rich Domain Model та State Pattern","\u002Fjava\u002Frich-domain-model","04.java\u002F04.rich-domain-model",{"title":2577,"path":2578,"stem":2579},"Патерни для складної бізнес-логіки","\u002Fjava\u002Fbusiness-logic-patterns","04.java\u002F05.business-logic-patterns",{"title":2581,"path":2582,"stem":2583},"Обробка помилок та валідація","\u002Fjava\u002Ferror-handling-validation","04.java\u002F06.error-handling-validation",{"title":2585,"path":2586,"stem":2587,"children":2588,"page":59},"Проектування баз даних","\u002Fjava\u002Fpr2","04.java\u002Fpr2",[2589,2593,2597,2601,2605,2609,2613,2617,2621,2625,2629,2633,2637,2641,2645,2649,2653,2657,2661,2665,2669,2673,2677,2681,2685,2689,2693,2697,2701,2705,2709,2713,2717,2721,2725,2729,2733],{"title":2590,"path":2591,"stem":2592},"Концептуальне моделювання: Мистецтво розуміння предметної області","\u002Fjava\u002Fpr2\u002Fconceptual-modeling","04.java\u002Fpr2\u002F01.conceptual-modeling",{"title":2594,"path":2595,"stem":2596},"Логічне моделювання: Від бізнес-ідей до структур даних","\u002Fjava\u002Fpr2\u002Flogical-modeling","04.java\u002Fpr2\u002F02.logical-modeling",{"title":2598,"path":2599,"stem":2600},"Нормалізація: Гігієна даних та боротьба з аномаліями","\u002Fjava\u002Fpr2\u002Fnormalization","04.java\u002Fpr2\u002F03.normalization",{"title":2602,"path":2603,"stem":2604},"Фізична схема: Від абстракції до DDL","\u002Fjava\u002Fpr2\u002Fphysical-schema","04.java\u002Fpr2\u002F04.physical-schema",{"title":2606,"path":2607,"stem":2608},"Архітектурна класифікація таблиць","\u002Fjava\u002Fpr2\u002Ftable-classification","04.java\u002Fpr2\u002F05.table-classification",{"title":2610,"path":2611,"stem":2612},"Database Migrations: Версіонування схеми з Flyway","\u002Fjava\u002Fpr2\u002Fdatabase-migrations","04.java\u002Fpr2\u002F06.database-migrations",{"title":2614,"path":2615,"stem":2616},"А що, якби це була не реляційна БД?","\u002Fjava\u002Fpr2\u002Fbeyond-relational","04.java\u002Fpr2\u002F07.beyond-relational",{"title":2618,"path":2619,"stem":2620},"Object-Relational Impedance Mismatch: Два світи, що не хочуть дружити","\u002Fjava\u002Fpr2\u002Fimpedance-mismatch","04.java\u002Fpr2\u002F09.impedance-mismatch",{"title":2622,"path":2623,"stem":2624},"JDBC: Перший контакт із базою даних","\u002Fjava\u002Fpr2\u002Fjdbc-fundamentals","04.java\u002Fpr2\u002F10.jdbc-fundamentals",{"title":2626,"path":2627,"stem":2628},"Якість коду: Spotless, SpotBugs та SonarQube","\u002Fjava\u002Fpr2\u002F10a.code-quality","04.java\u002Fpr2\u002F10a.code-quality",{"title":2630,"path":2631,"stem":2632},"Connection Pool: Патерн Object Pool для JDBC-з'єднань","\u002Fjava\u002Fpr2\u002Fconnection-pool","04.java\u002Fpr2\u002F11.connection-pool",{"title":2634,"path":2635,"stem":2636},"Row Data Gateway: Об'єкт як обгортка рядка таблиці","\u002Fjava\u002Fpr2\u002Frow-data-gateway","04.java\u002Fpr2\u002F12.row-data-gateway",{"title":2638,"path":2639,"stem":2640},"Table Data Gateway: Фасад таблиці як архітектурний відступ","\u002Fjava\u002Fpr2\u002Ftable-data-gateway","04.java\u002Fpr2\u002F13.table-data-gateway",{"title":2642,"path":2643,"stem":2644},"Repository + Data Mapper: Правильна шарова архітектура з JDBC","\u002Fjava\u002Fpr2\u002Frepository-data-mapper","04.java\u002Fpr2\u002F14.repository-data-mapper",{"title":2646,"path":2647,"stem":2648},"Identity Map: Кешування сутностей у рамках сесії","\u002Fjava\u002Fpr2\u002Fidentity-map","04.java\u002Fpr2\u002F15.identity-map",{"title":2650,"path":2651,"stem":2652},"Unit of Work: Відстеження змін і координація JDBC-транзакцій","\u002Fjava\u002Fpr2\u002Funit-of-work","04.java\u002Fpr2\u002F16.unit-of-work",{"title":2654,"path":2655,"stem":2656},"Strategy: Замінювані SQL-стратегії для підтримки різних СУБД","\u002Fjava\u002Fpr2\u002Fstrategy-sql","04.java\u002Fpr2\u002F17.strategy-sql",{"title":2658,"path":2659,"stem":2660},"Proxy: Lazy Loading для One-To-Many колекцій","\u002Fjava\u002Fpr2\u002Fproxy-lazy-loading","04.java\u002Fpr2\u002F18.proxy-lazy-loading",{"title":2662,"path":2663,"stem":2664},"Generic Repository через Java Reflection: анотації та динамічний SQL","\u002Fjava\u002Fpr2\u002Fgeneric-repository-reflection","04.java\u002Fpr2\u002F19.generic-repository-reflection",{"title":2666,"path":2667,"stem":2668},"Specification Pattern: Композиція бізнес-правил для складних запитів","\u002Fjava\u002Fpr2\u002Fspecification-pattern","04.java\u002Fpr2\u002F20.specification-pattern",{"title":2670,"path":2671,"stem":2672},"Розширені можливості Specification Pattern: підзапити, агрегації та гібридний підхід","\u002Fjava\u002Fpr2\u002F20a.advanced-specifications","04.java\u002Fpr2\u002F20a.advanced-specifications",{"title":2674,"path":2675,"stem":2676},"Асинхронність у JDBC: Від блокуючих викликів до CompletableFuture","\u002Fjava\u002Fpr2\u002Fasynchronous-jdbc","04.java\u002Fpr2\u002F21.asynchronous-jdbc",{"title":2678,"path":2679,"stem":2680},"Інтеграційне тестування JDBC-репозиторіїв: Embedded H2 та патерн AAA","\u002Fjava\u002Fpr2\u002Fintegration-testing-h2","04.java\u002Fpr2\u002F22.integration-testing-h2",{"title":2682,"path":2683,"stem":2684},"Testcontainers: Тестування з реальною PostgreSQL у Docker-контейнерах","\u002Fjava\u002Fpr2\u002Fintegration-testing-testcontainers","04.java\u002Fpr2\u002F23.integration-testing-testcontainers",{"title":2686,"path":2687,"stem":2688},"Google Guice: Впровадження залежностей у JavaFX-проєкті","\u002Fjava\u002Fpr2\u002Fdependency-injection-guice","04.java\u002Fpr2\u002F24.dependency-injection-guice",{"title":2690,"path":2691,"stem":2692},"JavaFX: Основи побудови графічних інтерфейсів","\u002Fjava\u002Fpr2\u002Fjavafx-fundamentals","04.java\u002Fpr2\u002F25.javafx-fundamentals",{"title":2694,"path":2695,"stem":2696},"Properties та Bindings: Реактивність у JavaFX","\u002Fjava\u002Fpr2\u002Fjavafx-properties-bindings","04.java\u002Fpr2\u002F26.javafx-properties-bindings",{"title":2698,"path":2699,"stem":2700},"MVC vs MVP vs MVVM: Еволюція архітектурних патернів UI","\u002Fjava\u002Fpr2\u002Fui-architecture-patterns","04.java\u002Fpr2\u002F27.ui-architecture-patterns",{"title":2702,"path":2703,"stem":2704},"MVVM на практиці: Побудова ViewModel","\u002Fjava\u002Fpr2\u002Fmvvm-viewmodel-implementation","04.java\u002Fpr2\u002F28.mvvm-viewmodel-implementation",{"title":2706,"path":2707,"stem":2708},"View та Controller: Зв'язування з ViewModel через FXML","\u002Fjava\u002Fpr2\u002Fmvvm-view-controller","04.java\u002Fpr2\u002F29.mvvm-view-controller",{"title":2710,"path":2711,"stem":2712},"Інтеграція MVVM з Guice: Автоматична ін'єкція залежностей","\u002Fjava\u002Fpr2\u002Fmvvm-guice-integration","04.java\u002Fpr2\u002F30.mvvm-guice-integration",{"title":2714,"path":2715,"stem":2716},"Валідація та обробка помилок у MVVM","\u002Fjava\u002Fpr2\u002Fmvvm-validation-error-handling","04.java\u002Fpr2\u002F31.mvvm-validation-error-handling",{"title":2718,"path":2719,"stem":2720},"Навігація та управління екранами у JavaFX MVVM","\u002Fjava\u002Fpr2\u002Fmvvm-navigation-screen-management","04.java\u002Fpr2\u002F32.mvvm-navigation-screen-management",{"title":2722,"path":2723,"stem":2724},"Тестування JavaFX MVVM-додатків","\u002Fjava\u002Fpr2\u002Fmvvm-testing","04.java\u002Fpr2\u002F33.mvvm-testing",{"title":2726,"path":2727,"stem":2728},"Стилізація та теми у JavaFX: CSS та User Experience","\u002Fjava\u002Fpr2\u002Fjavafx-styling-themes","04.java\u002Fpr2\u002F34.javafx-styling-themes",{"title":2730,"path":2731,"stem":2732},"AtlantaFX: Сучасні теми для JavaFX додатків","\u002Fjava\u002Fpr2\u002Fatlantafx-modern-themes","04.java\u002Fpr2\u002F35.atlantafx-modern-themes",{"title":2734,"path":2735,"stem":2736},"Пакування та розповсюдження JavaFX-додатків","\u002Fjava\u002Fpr2\u002Fjar-packaging-distribution","04.java\u002Fpr2\u002F36.jar-packaging-distribution",{"title":2738,"icon":2739,"path":2740,"stem":2741,"children":2742,"page":59},"Python","i-devicon-python","\u002Fpython","05.python",[2743,2747,2750,2754,2758,2762,2766,2770,2774,2778,2782,2786,2790,2794,2798,2859],{"title":2744,"path":2745,"stem":2746},"Модулі, Пакети та Віртуальні Середовища","\u002Fpython\u002Fmodules-packages-venv","05.python\u002F00.modules-packages-venv",{"title":71,"path":2748,"stem":2749},"\u002Fpython\u002Fclasses-objects","05.python\u002F01.classes-objects",{"title":2751,"path":2752,"stem":2753},"Інкапсуляція, Керування Доступом та Властивості","\u002Fpython\u002Fencapsulation","05.python\u002F02.encapsulation",{"title":2755,"path":2756,"stem":2757},"Наслідування, MRO та суперсила super()","\u002Fpython\u002Finheritance-mro","05.python\u002F03.inheritance-mro",{"title":2759,"path":2760,"stem":2761},"Абстракція — ABC проти Статичних Протоколів (PEP 544)","\u002Fpython\u002Fabstraction-protocols","05.python\u002F04.abstraction-protocols",{"title":2763,"path":2764,"stem":2765},"Магічні методи (Dunder) та Емуляція протоколів","\u002Fpython\u002Fdunder-methods","05.python\u002F05.dunder-methods",{"title":2767,"path":2768,"stem":2769},"Декоратори та Керування життєвим циклом методів","\u002Fpython\u002Fdecorators-static-class","05.python\u002F06.decorators-static-class",{"title":2771,"path":2772,"stem":2773},"Дескриптори — Магія доступу до атрибутів","\u002Fpython\u002Fdescriptors","05.python\u002F07.descriptors",{"title":2775,"path":2776,"stem":2777},"Метакласи — Динамічне створення класів під капотом CPython","\u002Fpython\u002Fmetaclasses","05.python\u002F08.metaclasses",{"title":2779,"path":2780,"stem":2781},"Dataclasses, NamedTuple та сучасні контейнери Python","\u002Fpython\u002Fmodern-containers","05.python\u002F09.modern-containers",{"title":2783,"path":2784,"stem":2785},"GIL та модель конкурентності CPython — фундамент перед потоками і процесами","\u002Fpython\u002Fgil-concurrency-intro","05.python\u002F11.gil-concurrency-intro",{"title":2787,"path":2788,"stem":2789},"Threading — конкурентність для I\u002FO-bound задач","\u002Fpython\u002Fthreading","05.python\u002F12.threading",{"title":2791,"path":2792,"stem":2793},"Multiprocessing — справжній паралелізм для CPU-bound задач","\u002Fpython\u002Fmultiprocessing","05.python\u002F13.multiprocessing",{"title":2795,"path":2796,"stem":2797},"asyncio — кооперативна конкурентність та event loop","\u002Fpython\u002Fasyncio","05.python\u002F14.asyncio",{"title":2799,"icon":92,"path":2800,"stem":2801,"children":2802,"page":59},"FastAPI","\u002Fpython\u002Ffastapi","05.python\u002Ffastapi",[2803,2807,2811,2815,2819,2823,2827,2831,2835,2839,2843,2847,2851,2855],{"title":2804,"path":2805,"stem":2806},"Глибокий Typing та Pydantic v2 — від анотацій до валідації","\u002Fpython\u002Ffastapi\u002Ftyping-pydantic","05.python\u002Ffastapi\u002F15.typing-pydantic",{"title":2808,"path":2809,"stem":2810},"WSGI, ASGI та Python Web-екосистема","\u002Fpython\u002Ffastapi\u002Fwsgi-asgi-ecosystem","05.python\u002Ffastapi\u002F16.wsgi-asgi-ecosystem",{"title":2812,"path":2813,"stem":2814},"FastAPI: Перший додаток, Uvicorn та OpenAPI","\u002Fpython\u002Ffastapi\u002Ffastapi-intro","05.python\u002Ffastapi\u002F17.fastapi-intro",{"title":2816,"path":2817,"stem":2818},"Маршрутизація, параметри запитів та APIRouter","\u002Fpython\u002Ffastapi\u002Ffastapi-routing-params","05.python\u002Ffastapi\u002F18.fastapi-routing-params",{"title":2820,"path":2821,"stem":2822},"Pydantic v2 у FastAPI — схеми, валідація та серіалізація","\u002Fpython\u002Ffastapi\u002Ffastapi-pydantic-schemas","05.python\u002Ffastapi\u002F19.fastapi-pydantic-schemas",{"title":2824,"path":2825,"stem":2826},"Dependency Injection — серце архітектури FastAPI","\u002Fpython\u002Ffastapi\u002Ffastapi-dependency-injection","05.python\u002Ffastapi\u002F20.fastapi-dependency-injection",{"title":2828,"path":2829,"stem":2830},"Обробка помилок, Middleware та CORS у FastAPI","\u002Fpython\u002Ffastapi\u002Ffastapi-errors-middleware","05.python\u002Ffastapi\u002F21.fastapi-errors-middleware",{"title":2832,"path":2833,"stem":2834},"SQLAlchemy 2.0 — ORM, Core та Async Engine","\u002Fpython\u002Ffastapi\u002Fsqlalchemy-orm","05.python\u002Ffastapi\u002F22.sqlalchemy-orm",{"title":2836,"path":2837,"stem":2838},"Alembic — міграції бази даних","\u002Fpython\u002Ffastapi\u002Falembic-migrations","05.python\u002Ffastapi\u002F23.alembic-migrations",{"title":2840,"path":2841,"stem":2842},"Repository Pattern та Clean Architecture у FastAPI","\u002Fpython\u002Ffastapi\u002Ffastapi-repository-clean-arch","05.python\u002Ffastapi\u002F24.fastapi-repository-clean-arch",{"title":2844,"path":2845,"stem":2846},"Автентифікація та авторизація — JWT, OAuth2, RBAC","\u002Fpython\u002Ffastapi\u002Ffastapi-auth-security","05.python\u002Ffastapi\u002F25.fastapi-auth-security",{"title":2848,"path":2849,"stem":2850},"Тестування FastAPI — pytest, httpx, fixtures","\u002Fpython\u002Ffastapi\u002Ffastapi-testing","05.python\u002Ffastapi\u002F26.fastapi-testing",{"title":2852,"path":2853,"stem":2854},"Async на повну — Background Tasks, WebSockets, SSE та Celery","\u002Fpython\u002Ffastapi\u002Ffastapi-async-advanced","05.python\u002Ffastapi\u002F27.fastapi-async-advanced",{"title":2856,"path":2857,"stem":2858},"Deployment — Docker, CI\u002FCD, Monitoring","\u002Fpython\u002Ffastapi\u002Ffastapi-deployment","05.python\u002Ffastapi\u002F28.fastapi-deployment",{"title":2860,"path":2861,"stem":2862},"[object Object]","\u002Fpython\u002Foop-plan","05.python\u002Foop-plan",{"title":2864,"icon":2865,"path":2866,"stem":2867,"children":2868,"page":59},"Бази даних","i-lucide-database","\u002Fdatabases","06.databases",[2869,2899,2922,2959,2988,3006,3040,3052,3061],{"title":2870,"icon":2871,"path":2872,"stem":2873,"children":2874,"page":59},"Intro","i-lucide-play","\u002Fdatabases\u002Fintro","06.databases\u002F01.intro",[2875,2879,2883,2887,2891,2895],{"title":2876,"path":2877,"stem":2878},"Введення в теорію баз даних","\u002Fdatabases\u002Fintro\u002Fintroduction-to-databases","06.databases\u002F01.intro\u002F01.introduction-to-databases",{"title":2880,"path":2881,"stem":2882},"Реляційна модель даних","\u002Fdatabases\u002Fintro\u002Frelational-model-theory","06.databases\u002F01.intro\u002F02.relational-model-theory",{"title":2884,"path":2885,"stem":2886},"ER-моделювання","\u002Fdatabases\u002Fintro\u002Fer-modeling","06.databases\u002F01.intro\u002F03.er-modeling",{"title":2888,"path":2889,"stem":2890},"Логічне проектування БД","\u002Fdatabases\u002Fintro\u002Flogical-schema","06.databases\u002F01.intro\u002F04.logical-schema",{"title":2892,"path":2893,"stem":2894},"Класифікація таблиць","\u002Fdatabases\u002Fintro\u002Ftable-classification","06.databases\u002F01.intro\u002F05.table-classification",{"title":2896,"path":2897,"stem":2898},"PlantUML для баз даних","\u002Fdatabases\u002Fintro\u002Fplantuml-diagrams","06.databases\u002F01.intro\u002F06.plantuml-diagrams",{"title":2900,"icon":2865,"path":2901,"stem":2902,"children":2903,"page":59},"MS SQL Server Start","\u002Fdatabases\u002Fms-sql-server-start","06.databases\u002F02.ms-sql-server-start",[2904,2908,2914,2918],{"title":2905,"path":2906,"stem":2907},"Типи даних у MS SQL Server","\u002Fdatabases\u002Fms-sql-server-start\u002Fdata-types","06.databases\u002F02.ms-sql-server-start\u002F01.data-types",{"title":2909,"path":2910,"stem":2911,"children":2912},"Індекси у MS SQL Server","\u002Fdatabases\u002Fms-sql-server-start\u002Fsql-indexes","06.databases\u002F02.ms-sql-server-start\u002F02.sql-indexes",[2913],{"title":2909,"path":2910,"stem":2911},{"title":2915,"path":2916,"stem":2917},"Системні бази даних MS SQL Server","\u002Fdatabases\u002Fms-sql-server-start\u002Fsystem-databases","06.databases\u002F02.ms-sql-server-start\u002F03.system-databases",{"title":2919,"path":2920,"stem":2921},"Огляд мови SQL та запитів","\u002Fdatabases\u002Fms-sql-server-start\u002Fsql-queries-overview","06.databases\u002F02.ms-sql-server-start\u002F04.sql-queries-overview",{"title":2923,"icon":2865,"path":2924,"stem":2925,"children":2926,"page":59},"SQL","\u002Fdatabases\u002Fsql","06.databases\u002F03.sql",[2927,2931,2935,2939,2943,2947,2951,2955],{"title":2928,"path":2929,"stem":2930},"Налаштування демонстраційної бази даних","\u002Fdatabases\u002Fsql\u002Fsample-database-setup","06.databases\u002F03.sql\u002F00.sample-database-setup",{"title":2932,"path":2933,"stem":2934},"DDL - Створення таблиць (CREATE TABLE)","\u002Fdatabases\u002Fsql\u002Fddl-create-table","06.databases\u002F03.sql\u002F01.ddl-create-table",{"title":2936,"path":2937,"stem":2938},"DDL - Зміна та видалення таблиць (ALTER, DROP)","\u002Fdatabases\u002Fsql\u002Fddl-alter-drop-table","06.databases\u002F03.sql\u002F02.ddl-alter-drop-table",{"title":2940,"path":2941,"stem":2942},"SELECT запити - Основи","\u002Fdatabases\u002Fsql\u002Fselect-queries-fundamentals","06.databases\u002F03.sql\u002F03.select-queries-fundamentals",{"title":2944,"path":2945,"stem":2946},"SELECT запити - Розширені можливості","\u002Fdatabases\u002Fsql\u002Fselect-queries-advanced","06.databases\u002F03.sql\u002F04.select-queries-advanced",{"title":2948,"path":2949,"stem":2950},"INSERT запити - Додавання даних","\u002Fdatabases\u002Fsql\u002Finsert-queries","06.databases\u002F03.sql\u002F05.insert-queries",{"title":2952,"path":2953,"stem":2954},"UPDATE та DELETE запити","\u002Fdatabases\u002Fsql\u002Fupdate-delete-queries","06.databases\u002F03.sql\u002F06.update-delete-queries",{"title":2956,"path":2957,"stem":2958},"Транзакції в SQL","\u002Fdatabases\u002Fsql\u002Ftransactions","06.databases\u002F03.sql\u002F07.transactions",{"title":2960,"icon":2865,"path":2961,"stem":2962,"children":2963,"page":59},"Multi Table Databases","\u002Fdatabases\u002Fmulti-table-databases","06.databases\u002F04.multi-table-databases",[2964,2968,2972,2976,2980,2984],{"title":2965,"path":2966,"stem":2967},"Зв'язки та нормалізація БД","\u002Fdatabases\u002Fmulti-table-databases\u002Frelationships-and-normalization","06.databases\u002F04.multi-table-databases\u002F00.relationships-and-normalization",{"title":2969,"path":2970,"stem":2971},"INNER JOIN - З'єднання таблиць","\u002Fdatabases\u002Fmulti-table-databases\u002Finner-join","06.databases\u002F04.multi-table-databases\u002F01.inner-join",{"title":2973,"path":2974,"stem":2975},"OUTER JOINs - LEFT, RIGHT, FULL","\u002Fdatabases\u002Fmulti-table-databases\u002Fouter-joins","06.databases\u002F04.multi-table-databases\u002F02.outer-joins",{"title":2977,"path":2978,"stem":2979},"CROSS та SELF JOINs","\u002Fdatabases\u002Fmulti-table-databases\u002Fcross-self-joins","06.databases\u002F04.multi-table-databases\u002F03.cross-self-joins",{"title":2981,"path":2982,"stem":2983},"Підзапити (Subqueries)","\u002Fdatabases\u002Fmulti-table-databases\u002Fsubqueries","06.databases\u002F04.multi-table-databases\u002F04.subqueries",{"title":2985,"path":2986,"stem":2987},"Агрегації з JOIN","\u002Fdatabases\u002Fmulti-table-databases\u002Faggregations-with-joins","06.databases\u002F04.multi-table-databases\u002F05.aggregations-with-joins",{"title":2989,"icon":2990,"path":2991,"stem":2992,"children":2993,"page":59},"Aggregate Functions","i-lucide-calculator","\u002Fdatabases\u002Faggregate-functions","06.databases\u002F05.aggregate-functions",[2994,2998,3002],{"title":2995,"path":2996,"stem":2997},"Функції агрегування в MS SQL Server","\u002Fdatabases\u002Faggregate-functions\u002Fintroduction-aggregate-functions","06.databases\u002F05.aggregate-functions\u002F01.introduction-aggregate-functions",{"title":2999,"path":3000,"stem":3001},"Групування даних в MS SQL Server","\u002Fdatabases\u002Faggregate-functions\u002Fgrouping-data","06.databases\u002F05.aggregate-functions\u002F02.grouping-data",{"title":3003,"path":3004,"stem":3005},"Підзапити з агрегатними функціями","\u002Fdatabases\u002Faggregate-functions\u002Fsubqueries-aggregates","06.databases\u002F05.aggregate-functions\u002F03.subqueries-aggregates",{"title":3007,"icon":3008,"path":3009,"stem":3010,"children":3011,"page":59},"Тригери та зберігаємі процедури","i-lucide-database-zap","\u002Fdatabases\u002Ftriggers-stored-procedures","06.databases\u002F07.triggers-stored-procedures",[3012,3016,3020,3024,3028,3032,3036],{"title":3013,"path":3014,"stem":3015},"DML-тригери","\u002Fdatabases\u002Ftriggers-stored-procedures\u002Fdml-triggers","06.databases\u002F07.triggers-stored-procedures\u002F01.dml-triggers",{"title":3017,"path":3018,"stem":3019},"DDL-тригери","\u002Fdatabases\u002Ftriggers-stored-procedures\u002Fddl-triggers","06.databases\u002F07.triggers-stored-procedures\u002F02.ddl-triggers",{"title":3021,"path":3022,"stem":3023},"Transact-SQL розширення","\u002Fdatabases\u002Ftriggers-stored-procedures\u002Ftransact-sql-extensions","06.databases\u002F07.triggers-stored-procedures\u002F03.transact-sql-extensions",{"title":3025,"path":3026,"stem":3027},"Транзакції","\u002Fdatabases\u002Ftriggers-stored-procedures\u002Ftransactions","06.databases\u002F07.triggers-stored-procedures\u002F04.transactions",{"title":3029,"path":3030,"stem":3031},"Зберігаємі процедури","\u002Fdatabases\u002Ftriggers-stored-procedures\u002Fstored-procedures","06.databases\u002F07.triggers-stored-procedures\u002F05.stored-procedures",{"title":3033,"path":3034,"stem":3035},"Користувацькі функції","\u002Fdatabases\u002Ftriggers-stored-procedures\u002Fuser-defined-functions","06.databases\u002F07.triggers-stored-procedures\u002F06.user-defined-functions",{"title":3037,"path":3038,"stem":3039},"Безпека баз даних","\u002Fdatabases\u002Ftriggers-stored-procedures\u002Fsecurity","06.databases\u002F07.triggers-stored-procedures\u002F08.security",{"title":3037,"icon":793,"path":3041,"stem":3042,"children":3043,"page":59},"\u002Fdatabases\u002Fsecurity","06.databases\u002F08.security",[3044,3048],{"title":3045,"path":3046,"stem":3047},"Вступ до безпеки баз даних","\u002Fdatabases\u002Fsecurity\u002Fintroduction","06.databases\u002F08.security\u002F01.introduction",{"title":3049,"path":3050,"stem":3051},"Системні представлення та метадані","\u002Fdatabases\u002Fsecurity\u002Fsystem-views","06.databases\u002F08.security\u002F02.system-views",{"title":3053,"icon":3054,"path":3055,"stem":3056,"children":3057,"page":59},"Резервне копіювання та відновлення","i-lucide-database-backup","\u002Fdatabases\u002Fbackup-recovery","06.databases\u002F09.backup-recovery",[3058],{"title":3053,"path":3059,"stem":3060},"\u002Fdatabases\u002Fbackup-recovery\u002Fbackup-restore","06.databases\u002F09.backup-recovery\u002F01.backup-restore",{"title":3062,"icon":3063,"path":3064,"stem":3065,"children":3066,"page":59},"Повнотекстовий пошук","i-lucide-search","\u002Fdatabases\u002Ffull-text-search","06.databases\u002F10.full-text-search",[3067],{"title":3062,"path":3068,"stem":3069},"\u002Fdatabases\u002Ffull-text-search\u002Ffull-text-search","06.databases\u002F10.full-text-search\u002F01.full-text-search",{"title":3071,"icon":3072,"path":3073,"stem":3074,"children":3075,"page":59},"Tools","i-lucide-wrench","\u002Ftools","07.tools",[3076,3152],{"title":3077,"icon":3078,"path":3079,"stem":3080,"children":3081},"Docker","i-simple-icons-docker","\u002Ftools\u002Fdocker","07.tools\u002F01.docker\u002Findex",[3082,3084,3088,3092,3096,3100,3104,3108,3112,3116,3120,3124,3128,3132,3136,3140,3144,3148],{"title":3083,"path":3079,"stem":3080},"Docker: від нуля до production",{"title":3085,"path":3086,"stem":3087},"Контейнеризація — від проблеми до рішення","\u002Ftools\u002Fdocker\u002Fcontainerization-concept","07.tools\u002F01.docker\u002F01.containerization-concept",{"title":3089,"path":3090,"stem":3091},"Docker — що це і навіщо?","\u002Ftools\u002Fdocker\u002Fdocker-what-and-why","07.tools\u002F01.docker\u002F02.docker-what-and-why",{"title":3093,"path":3094,"stem":3095},"Архітектура Docker Engine","\u002Ftools\u002Fdocker\u002Fdocker-architecture","07.tools\u002F01.docker\u002F03.docker-architecture",{"title":3097,"path":3098,"stem":3099},"Встановлення Docker","\u002Ftools\u002Fdocker\u002Finstallation","07.tools\u002F01.docker\u002F04.installation",{"title":3101,"path":3102,"stem":3103},"Перший контейнер — docker run","\u002Ftools\u002Fdocker\u002Ffirst-container","07.tools\u002F01.docker\u002F05.first-container",{"title":3105,"path":3106,"stem":3107},"Життєвий цикл контейнера","\u002Ftools\u002Fdocker\u002Fcontainer-lifecycle","07.tools\u002F01.docker\u002F06.container-lifecycle",{"title":3109,"path":3110,"stem":3111},"Docker Images — фундаментальні концепції","\u002Ftools\u002Fdocker\u002Fdocker-images-fundamentals","07.tools\u002F01.docker\u002F07.docker-images-fundamentals",{"title":3113,"path":3114,"stem":3115},"Dockerfile — основи","\u002Ftools\u002Fdocker\u002Fdockerfile-basics","07.tools\u002F01.docker\u002F08.dockerfile-basics",{"title":3117,"path":3118,"stem":3119},"Dockerfile — просунуті техніки","\u002Ftools\u002Fdocker\u002Fdockerfile-advanced","07.tools\u002F01.docker\u002F09.dockerfile-advanced",{"title":3121,"path":3122,"stem":3123},"Build Context та кешування шарів","\u002Ftools\u002Fdocker\u002Fbuild-context-and-cache","07.tools\u002F01.docker\u002F10.build-context-and-cache",{"title":3125,"path":3126,"stem":3127},"Реєстри Docker-образів","\u002Ftools\u002Fdocker\u002Fimage-registries","07.tools\u002F01.docker\u002F11.image-registries",{"title":3129,"path":3130,"stem":3131},"Контейнеризація .NET додатків","\u002Ftools\u002Fdocker\u002Fdotnet-containerization","07.tools\u002F01.docker\u002F12.dotnet-containerization",{"title":3133,"path":3134,"stem":3135},"Томи та збереження даних","\u002Ftools\u002Fdocker\u002Fvolumes-and-data","07.tools\u002F01.docker\u002F13.volumes-and-data",{"title":3137,"path":3138,"stem":3139},"Основи мережі в Docker","\u002Ftools\u002Fdocker\u002Fnetworking-basics","07.tools\u002F01.docker\u002F14.networking-basics",{"title":3141,"path":3142,"stem":3143},"Змінні оточення та конфігурація","\u002Ftools\u002Fdocker\u002Fenvironment-and-configuration","07.tools\u002F01.docker\u002F15.environment-and-configuration",{"title":3145,"path":3146,"stem":3147},"Docker Compose — оркестрація контейнерів","\u002Ftools\u002Fdocker\u002Fdocker-compose-basics","07.tools\u002F01.docker\u002F16.docker-compose-basics",{"title":3149,"path":3150,"stem":3151},"Docker Compose — Multi-Service застосунки","\u002Ftools\u002Fdocker\u002Fcompose-multi-service","07.tools\u002F01.docker\u002F17.compose-multi-service",{"title":3153,"icon":3154,"path":3155,"stem":3156,"children":3157},"Kubernetes","simple-icons:kubernetes","\u002Ftools\u002Fkubernetes","07.tools\u002F02.kubernetes\u002Findex",[3158,3160,3164,3168,3172,3176,3180,3184,3188],{"title":3159,"path":3155,"stem":3156},"Kubernetes: від розробки до production",{"title":3161,"path":3162,"stem":3163},"Kubernetes — коли Docker Compose більше не вистачає","\u002Ftools\u002Fkubernetes\u002Fwhy-kubernetes","07.tools\u002F02.kubernetes\u002F01.why-kubernetes",{"title":3165,"path":3166,"stem":3167},"Архітектура Kubernetes — анатомія кластера","\u002Ftools\u002Fkubernetes\u002Fkubernetes-architecture","07.tools\u002F02.kubernetes\u002F02.kubernetes-architecture",{"title":3169,"path":3170,"stem":3171},"Локальне середовище — minikube, kind та k3s","\u002Ftools\u002Fkubernetes\u002Flocal-environment","07.tools\u002F02.kubernetes\u002F03.local-environment",{"title":3173,"path":3174,"stem":3175},"Pod — атомарна одиниця Kubernetes","\u002Ftools\u002Fkubernetes\u002Fpods-and-containers","07.tools\u002F02.kubernetes\u002F04.pods-and-containers",{"title":3177,"path":3178,"stem":3179},"Патерни використання Pod","\u002Ftools\u002Fkubernetes\u002Fpod-patterns","07.tools\u002F02.kubernetes\u002F05.pod-patterns",{"title":3181,"path":3182,"stem":3183},"Deployment — декларативне управління Pod","\u002Ftools\u002Fkubernetes\u002Fdeployment-basics","07.tools\u002F02.kubernetes\u002F06.deployment-basics",{"title":3185,"path":3186,"stem":3187},"Rolling Updates та управління життєвим циклом Deployment","\u002Ftools\u002Fkubernetes\u002Fdeployment-rolling-updates","07.tools\u002F02.kubernetes\u002F07.deployment-rolling-updates",{"title":3189,"path":3190,"stem":3191},"Service — мережева абстракція для Pod","\u002Ftools\u002Fkubernetes\u002Fservices-networking","07.tools\u002F02.kubernetes\u002F08.services-networking",{"title":3193,"icon":3194,"path":3195,"stem":3196,"children":3197,"page":59},"Software Engineering","i-lucide-code-2","\u002Fsoftware-engineering","09.software-engineering",[3198,3202,3206,3210,3214,3218,3222,3226,3230,3234,3238],{"title":3199,"path":3200,"stem":3201},"1. Аналіз предметної області. Експертні знання та складність","\u002Fsoftware-engineering\u002Fintro-subdomains","09.software-engineering\u002F01.intro-subdomains",{"title":3203,"path":3204,"stem":3205},"2. Обмежені контексти. Інтеграція обмежених контекстів","\u002Fsoftware-engineering\u002Fintegrating-limited-contexts","09.software-engineering\u002F02.integrating-limited-contexts",{"title":3207,"path":3208,"stem":3209},"3. Реалізація простої бізнес-логіки","\u002Fsoftware-engineering\u002Fsimple","09.software-engineering\u002F03.simple",{"title":3211,"path":3212,"stem":3213},"4. Опрацювання складної бізнес-логіки","\u002Fsoftware-engineering\u002Fcomplex-business-logic","09.software-engineering\u002F04.complex-business-logic",{"title":3215,"path":3216,"stem":3217},"5. Моделювання фактора часу. Подієво-орієнтована архітектура.","\u002Fsoftware-engineering\u002Fmodelling-the-time-factor","09.software-engineering\u002F05.modelling-the-time-factor",{"title":3219,"path":3220,"stem":3221},"6. Архітектурні патерни","\u002Fsoftware-engineering\u002Farchitectural-patterns","09.software-engineering\u002F06.architectural-patterns",{"title":3223,"path":3224,"stem":3225},"Паттерни взаємодії","\u002Fsoftware-engineering\u002Fpatterns-of-interaction","09.software-engineering\u002F07.patterns-of-interaction",{"title":3227,"path":3228,"stem":3229},"Евристика проєктування","\u002Fsoftware-engineering\u002Fdesign-heuristics","09.software-engineering\u002F08.design-heuristics",{"title":3231,"path":3232,"stem":3233},"Еволюція проєктних рішень","\u002Fsoftware-engineering\u002Fevolution-of-design-solutions","09.software-engineering\u002F09.evolution-of-design-solutions",{"title":3235,"path":3236,"stem":3237},"EventStorming","\u002Fsoftware-engineering\u002Feventstorming","09.software-engineering\u002F10.eventstorming",{"title":3239,"path":3240,"stem":3241},"DDD на практиці","\u002Fsoftware-engineering\u002Fddd-in-practice","09.software-engineering\u002F11.ddd-in-practice",{"title":3243,"icon":943,"path":3244,"stem":3245,"children":3246,"page":59},"DDD","\u002Fddd","10.ddd",[3247,3251,3255,3259,3263,3267,3271,3275,3279,3283,3287,3291,3295],{"title":3248,"path":3249,"stem":3250},"Аналіз предметної області","\u002Fddd\u002Fdomain-analysis","10.ddd\u002F01.domain-analysis",{"title":3252,"path":3253,"stem":3254},"Експертні знання про предметну область","\u002Fddd\u002Fdomain-expert-knowledge","10.ddd\u002F02.domain-expert-knowledge",{"title":3256,"path":3257,"stem":3258},"Як осмислити складність предметної області","\u002Fddd\u002Fmanaging-domain-complexity","10.ddd\u002F03.managing-domain-complexity",{"title":3260,"path":3261,"stem":3262},"Інтеграція обмежених контекстів","\u002Fddd\u002Fbounded-context-integration","10.ddd\u002F04.bounded-context-integration",{"title":3264,"path":3265,"stem":3266},"Реалізація простої бізнес-логіки","\u002Fddd\u002Fsimple-business-logic","10.ddd\u002F05.simple-business-logic",{"title":3268,"path":3269,"stem":3270},"Обробка складної бізнес-логіки","\u002Fddd\u002Fcomplex-business-logic","10.ddd\u002F06.complex-business-logic",{"title":3272,"path":3273,"stem":3274},"Моделювання фактора часу","\u002Fddd\u002Ftime-modeling","10.ddd\u002F07.time-modeling",{"title":3276,"path":3277,"stem":3278},"Глава 8. Архітектурні Патерни","\u002Fddd\u002Farchitectural-patterns","10.ddd\u002F08.architectural-patterns",{"title":3280,"path":3281,"stem":3282},"Глава 9. Патерни Взаємодії","\u002Fddd\u002Finteraction-patterns","10.ddd\u002F09.interaction-patterns",{"title":3284,"path":3285,"stem":3286},"Глава 10. Проектні Евристики","\u002Fddd\u002Fdesign-heuristics","10.ddd\u002F10.design-heuristics",{"title":3288,"path":3289,"stem":3290},"Глава 11. Еволюція Проектних Рішень","\u002Fddd\u002Fevolution-of-design-decisions","10.ddd\u002F11.evolution-of-design-decisions",{"title":3292,"path":3293,"stem":3294},"Глава 12. EventStorming","\u002Fddd\u002Fevent-storming","10.ddd\u002F12.event-storming",{"title":3296,"path":3297,"stem":3298},"Глава 13. DDD на Практиці","\u002Fddd\u002Fddd-in-practice","10.ddd\u002F13.ddd-in-practice",{"title":3300,"icon":3301,"path":3302,"stem":3303,"children":3304,"page":59},"Media Streaming","i-lucide-video","\u002Fmedia-streaming","11.media-streaming",[3305,3309,3313,3317,3321,3325,3329],{"title":3306,"path":3307,"stem":3308},"01. Магія Стрімінгу: Що відбувається, коли ви натискаєте \"Play\"","\u002Fmedia-streaming\u002Fintroduction","11.media-streaming\u002F01.introduction",{"title":3310,"path":3311,"stem":3312},"02. Анатомія Медіа: Кодеки, Контейнери та Стиснення","\u002Fmedia-streaming\u002Faudio-video-anatomy","11.media-streaming\u002F02.audio-video-anatomy",{"title":3314,"path":3315,"stem":3316},"03. The Gym: FFmpeg Deep Dive","\u002Fmedia-streaming\u002Fffmpeg-gym","11.media-streaming\u002F03.ffmpeg-gym",{"title":3318,"path":3319,"stem":3320},"04. HLS Protocol: HTTP Live Streaming у Деталях","\u002Fmedia-streaming\u002Fhls-protocol","11.media-streaming\u002F04.hls-protocol",{"title":3322,"path":3323,"stem":3324},"05. DASH Protocol: Відкритий Стандарт","\u002Fmedia-streaming\u002Fdash-protocol","11.media-streaming\u002F05.dash-protocol",{"title":3326,"path":3327,"stem":3328},"06. Масштабування: CDN та Adaptive Bitrate","\u002Fmedia-streaming\u002Fcdn-and-adaptive-bitrate","11.media-streaming\u002F06.cdn-and-adaptive-bitrate",{"title":3330,"path":3331,"stem":3332},"07. Війна із Затримкою (Latency)","\u002Fmedia-streaming\u002Frealtime-latency","11.media-streaming\u002F07.realtime-latency",{"title":3334,"icon":3335,"path":3336,"stem":3337,"children":3338,"page":59},"HTML & CSS","i-devicon-html5","\u002Fhtml-css","12.html-css",[3339,3343,3347,3351,3355,3359,3363,3367,3371,3375,3379,3383,3387,3391,3395,3399,3403,3407,3411,3415,3419,3423,3427,3431,3435,3439,3443,3447,3451,3455],{"title":3340,"path":3341,"stem":3342},"Вступ до HTML. Структура документа","\u002Fhtml-css\u002Fintro-html-structure","12.html-css\u002F01.intro-html-structure",{"title":3344,"path":3345,"stem":3346},"Форматування тексту в HTML","\u002Fhtml-css\u002Fhtml-text-formatting","12.html-css\u002F02.html-text-formatting",{"title":3348,"path":3349,"stem":3350},"Посилання та зображення в HTML","\u002Fhtml-css\u002Fhtml-links-images","12.html-css\u002F03.html-links-images",{"title":3352,"path":3353,"stem":3354},"Списки та таблиці в HTML","\u002Fhtml-css\u002Fhtml-lists-tables","12.html-css\u002F04.html-lists-tables",{"title":3356,"path":3357,"stem":3358},"Форми в HTML","\u002Fhtml-css\u002Fhtml-forms","12.html-css\u002F05.html-forms",{"title":3360,"path":3361,"stem":3362},"Семантичні елементи HTML5","\u002Fhtml-css\u002Fhtml-semantic-elements","12.html-css\u002F06.html-semantic-elements",{"title":3364,"path":3365,"stem":3366},"Мультимедіа та розширені елементи HTML","\u002Fhtml-css\u002Fhtml-multimedia-advanced","12.html-css\u002F07.html-multimedia-advanced",{"title":3368,"path":3369,"stem":3370},"Мікророзмітка та SEO в HTML","\u002Fhtml-css\u002Fhtml-microdata-seo","12.html-css\u002F08.html-microdata-seo",{"title":3372,"path":3373,"stem":3374},"Вступ до CSS. Селектори та специфічність","\u002Fhtml-css\u002Fcss-intro-selectors","12.html-css\u002F09.css-intro-selectors",{"title":3376,"path":3377,"stem":3378},"Блокова модель CSS. Відступи. Box Sizing","\u002Fhtml-css\u002Fcss-box-model","12.html-css\u002F10.css-box-model",{"title":3380,"path":3381,"stem":3382},"Розміри у CSS: повний довідник одиниць і ключових слів","\u002Fhtml-css\u002F10a.css-sizing","12.html-css\u002F10a.css-sizing",{"title":3384,"path":3385,"stem":3386},"Типографіка в CSS. Шрифти та текст","\u002Fhtml-css\u002Fcss-typography","12.html-css\u002F11.css-typography",{"title":3388,"path":3389,"stem":3390},"Кольори та фони в CSS","\u002Fhtml-css\u002Fcss-colors-backgrounds","12.html-css\u002F12.css-colors-backgrounds",{"title":3392,"path":3393,"stem":3394},"Тіні та фільтри в CSS","\u002Fhtml-css\u002F12b.css-shadows-filters","12.html-css\u002F12b.css-shadows-filters",{"title":3396,"path":3397,"stem":3398},"CSS Flexbox: Фундамент гнучких макетів","\u002Fhtml-css\u002Fcss-flexbox-fundamentals","12.html-css\u002F13.css-flexbox-fundamentals",{"title":3400,"path":3401,"stem":3402},"CSS Flexbox: Вирівнювання та Позиціонування","\u002Fhtml-css\u002Fcss-flexbox-alignment-sizing-and-patterns","12.html-css\u002F14.css-flexbox-alignment-sizing-and-patterns",{"title":3404,"path":3405,"stem":3406},"CSS Grid. Двовимірний макет. Частина 1","\u002Fhtml-css\u002Fcss-layout-grid","12.html-css\u002F15.css-layout-grid",{"title":3408,"path":3409,"stem":3410},"CSS Grid. Двовимірний макет. Частина 2","\u002Fhtml-css\u002Fcss-layout-grid-advanced","12.html-css\u002F16.css-layout-grid-advanced",{"title":3412,"path":3413,"stem":3414},"Позиціонування в CSS. Z-index. Stacking Context","\u002Fhtml-css\u002Fcss-positioning","12.html-css\u002F17.css-positioning",{"title":3416,"path":3417,"stem":3418},"CSS Анімації та Переходи","\u002Fhtml-css\u002Fcss-animations-transitions","12.html-css\u002F18.css-animations-transitions",{"title":3420,"path":3421,"stem":3422},"Адаптивний дизайн. Media Queries. Частина 1","\u002Fhtml-css\u002Fcss-responsive-media-queries","12.html-css\u002F19.css-responsive-media-queries",{"title":3424,"path":3425,"stem":3426},"Адаптивний дизайн. Частина 2: clamp(), Container Queries, @layer","\u002Fhtml-css\u002Fcss-responsive-advanced","12.html-css\u002F20.css-responsive-advanced",{"title":3428,"path":3429,"stem":3430},"CSS Custom Properties. Методології. Сучасний CSS","\u002Fhtml-css\u002Fcss-variables-methodologies","12.html-css\u002F21.css-variables-methodologies",{"title":3432,"path":3433,"stem":3434},"Сучасний CSS 2023–2025: Нові можливості","\u002Fhtml-css\u002Fcss-modern-features","12.html-css\u002F22.css-modern-features",{"title":3436,"path":3437,"stem":3438},"CSS Nesting, @layer, @scope та @property: нативний препроцесор","\u002Fhtml-css\u002F22a.css-nesting-modern-syntax","12.html-css\u002F22a.css-nesting-modern-syntax",{"title":3440,"path":3441,"stem":3442},"CSS для форм та інтерактивних станів","\u002Fhtml-css\u002Fcss-forms-interactive-states","12.html-css\u002F23.css-forms-interactive-states",{"title":3444,"path":3445,"stem":3446},"Доступність у CSS (CSS Accessibility)","\u002Fhtml-css\u002Fcss-accessibility","12.html-css\u002F24.css-accessibility",{"title":3448,"path":3449,"stem":3450},"CSS-функції та сучасні sizing primitives","\u002Fhtml-css\u002Fcss-functions-sizing","12.html-css\u002F25.css-functions-sizing",{"title":3452,"path":3453,"stem":3454},"Rendering Pipeline і CSS Performance","\u002Fhtml-css\u002Fcss-rendering-performance","12.html-css\u002F26.css-rendering-performance",{"title":3456,"path":3457,"stem":3458},"CSS Best Practices: типові ситуації та правильні рішення","\u002Fhtml-css\u002Fcss-best-practices","12.html-css\u002F27.css-best-practices",{"title":3460,"path":3461,"stem":3462,"children":3463,"page":59},"AWS","\u002Faws","13.aws",[3464,3468,3472,3476,3480,3484,3488,3492,3496,3500,3504,3508,3512,3516,3520,3524,3528,3532],{"title":3465,"path":3466,"stem":3467},"Реєстрація AWS акаунту та студентські програми","\u002Faws\u002Faccount-registration","13.aws\u002F00.account-registration",{"title":3469,"path":3470,"stem":3471},"Вступ до хмарних обчислень та AWS","\u002Faws\u002Fintroduction-to-cloud","13.aws\u002F01.introduction-to-cloud",{"title":3473,"path":3474,"stem":3475},"AWS IAM — Identity and Access Management","\u002Faws\u002Fiam","13.aws\u002F02.iam",{"title":3477,"path":3478,"stem":3479},"AWS IAM CLI — Довідник команд","\u002Faws\u002F02a.iam-doc","13.aws\u002F02a.iam-doc",{"title":3481,"path":3482,"stem":3483},"Docker та контейнеризація в AWS — ECR, ECS та Fargate","\u002Faws\u002Fdocker-ecs","13.aws\u002F03.docker-ecs",{"title":3485,"path":3486,"stem":3487},"AWS ECR \u002F ECS CLI — Довідник команд","\u002Faws\u002F03a.docker-ecs-doc","13.aws\u002F03a.docker-ecs-doc",{"title":3489,"path":3490,"stem":3491},"Amazon EC2 — Elastic Compute Cloud","\u002Faws\u002Fec2","13.aws\u002F04.ec2",{"title":3493,"path":3494,"stem":3495},"AWS EC2 CLI — Довідник команд","\u002Faws\u002F04a.ec2-doc","13.aws\u002F04a.ec2-doc",{"title":3497,"path":3498,"stem":3499},"Elastic Load Balancing та Auto Scaling","\u002Faws\u002Falb-asg","13.aws\u002F05.alb-asg",{"title":3501,"path":3502,"stem":3503},"Amazon S3 — Simple Storage Service","\u002Faws\u002Fs3","13.aws\u002F06.s3",{"title":3505,"path":3506,"stem":3507},"Amazon CloudFront — Content Delivery Network","\u002Faws\u002Fcloudfront","13.aws\u002F07.cloudfront",{"title":3509,"path":3510,"stem":3511},"Amazon RDS — Relational Database Service","\u002Faws\u002Frds","13.aws\u002F08.rds",{"title":3513,"path":3514,"stem":3515},"Amazon DynamoDB — NoSQL Database","\u002Faws\u002Fdynamodb","13.aws\u002F09.dynamodb",{"title":3517,"path":3518,"stem":3519},"AWS Lambda та Serverless Compute","\u002Faws\u002Flambda","13.aws\u002F10.lambda",{"title":3521,"path":3522,"stem":3523},"Amazon Bedrock - Foundation Models, RAG та Agents","\u002Faws\u002Fbedrock","13.aws\u002F22.bedrock",{"title":3525,"path":3526,"stem":3527},"Amazon Rekognition - Комп'ютерний зір","\u002Faws\u002Frekognition","13.aws\u002F23.rekognition",{"title":3529,"path":3530,"stem":3531},"Amazon Textract - Інтелектуальний аналіз документів","\u002Faws\u002Ftextract","13.aws\u002F24.textract",{"title":3533,"path":3534,"stem":3535},"Amazon Polly, Transcribe, Comprehend та Translate","\u002Faws\u002Faudio-nlp-services","13.aws\u002F25.audio-nlp-services",{"title":3537,"path":3538,"stem":3539,"children":3540,"page":59},"Ai Development","\u002Fai-development","14.ai-development",[3541],{"title":3542,"path":3543,"stem":3544,"children":3545,"page":59},"Foundations","\u002Fai-development\u002Ffoundations","14.ai-development\u002F00.foundations",[3546],{"title":3547,"path":3548,"stem":3549},"Модуль 0.1. Що таке Large Language Model (LLM)","\u002Fai-development\u002Ffoundations\u002Fwhat-is-llm","14.ai-development\u002F00.foundations\u002F01.what-is-llm",{"title":3551,"path":3552,"stem":3553,"children":3554,"page":59},"React Native","\u002Freact-native","15.react-native",[3555,3576],{"title":3556,"path":3557,"stem":3558,"children":3559,"page":59},"Version1","\u002Freact-native\u002Fversion1","15.react-native\u002Fversion1",[3560,3564,3568,3572],{"title":3561,"path":3562,"stem":3563},"Що таке React Native","\u002Freact-native\u002Fversion1\u002Fwhat-is-react-native","15.react-native\u002Fversion1\u002F01.what-is-react-native",{"title":3565,"path":3566,"stem":3567},"Як влаштований React Native «під капотом»","\u002Freact-native\u002Fversion1\u002Freact-native-architecture","15.react-native\u002Fversion1\u002F02.react-native-architecture",{"title":3569,"path":3570,"stem":3571},"Від React до React Native: ключові відмінності","\u002Freact-native\u002Fversion1\u002Freact-native-vs-react","15.react-native\u002Fversion1\u002F03.react-native-vs-react",{"title":3573,"path":3574,"stem":3575},"View та Text — фундаментальні будівельні блоки","\u002Freact-native\u002Fversion1\u002Fview-and-text","15.react-native\u002Fversion1\u002F04.view-and-text",{"title":3577,"path":3578,"stem":3579,"children":3580,"page":59},"Version2","\u002Freact-native\u002Fversion2","15.react-native\u002Fversion2",[3581,3585,3589,3593,3597,3601,3605,3609,3613,3617,3621,3625,3629,3633,3637,3641,3645,3649,3653,3657,3661,3665,3669,3673,3677,3681,3685],{"title":3582,"path":3583,"stem":3584},"React Native — навіщо він потрібен","\u002Freact-native\u002Fversion2\u002Fwhy-react-native","15.react-native\u002Fversion2\u002F01.why-react-native",{"title":3586,"path":3587,"stem":3588},"Налаштування середовища розробки","\u002Freact-native\u002Fversion2\u002Fenvironment-setup","15.react-native\u002Fversion2\u002F02.environment-setup",{"title":3590,"path":3591,"stem":3592},"Expo і класичний React Native — два шляхи одного фреймворку","\u002Freact-native\u002Fversion2\u002Fexpo-vs-cli-landscape","15.react-native\u002Fversion2\u002F03.expo-vs-cli-landscape",{"title":3594,"path":3595,"stem":3596},"Середовище Expo — перший запуск","\u002Freact-native\u002Fversion2\u002Fexpo-setup-and-tooling","15.react-native\u002Fversion2\u002F04.expo-setup-and-tooling",{"title":3598,"path":3599,"stem":3600},"Структура проєкту та перші UI-примітиви","\u002Freact-native\u002Fversion2\u002Fproject-structure-and-conventions","15.react-native\u002Fversion2\u002F05.project-structure-and-conventions",{"title":3602,"path":3603,"stem":3604},"Базові компоненти React Native","\u002Freact-native\u002Fversion2\u002Fcore-components","15.react-native\u002Fversion2\u002F06.core-components",{"title":3606,"path":3607,"stem":3608},"Flexbox і layout у React Native","\u002Freact-native\u002Fversion2\u002Fflexbox-and-layout","15.react-native\u002Fversion2\u002F07.flexbox-and-layout",{"title":3610,"path":3611,"stem":3612},"StyleSheet і теми (light \u002F dark)","\u002Freact-native\u002Fversion2\u002Fstylesheet-and-theming","15.react-native\u002Fversion2\u002F08.stylesheet-and-theming",{"title":3614,"path":3615,"stem":3616},"Списки та віртуалізація","\u002Freact-native\u002Fversion2\u002Flists-and-virtualization","15.react-native\u002Fversion2\u002F09.lists-and-virtualization",{"title":3618,"path":3619,"stem":3620},"Форми, ввід і валідація","\u002Freact-native\u002Fversion2\u002Fforms-input-validation","15.react-native\u002Fversion2\u002F10.forms-input-validation",{"title":3622,"path":3623,"stem":3624},"Основи Expo Router","\u002Freact-native\u002Fversion2\u002Fexpo-router-basics","15.react-native\u002Fversion2\u002F11.expo-router-basics",{"title":3626,"path":3627,"stem":3628},"Вкладені навігатори та модалки","\u002Freact-native\u002Fversion2\u002Fnested-navigators-and-modals","15.react-native\u002Fversion2\u002F12.nested-navigators-and-modals",{"title":3630,"path":3631,"stem":3632},"Deep linking і параметри маршрутів","\u002Freact-native\u002Fversion2\u002Fdeep-linking-and-params","15.react-native\u002Fversion2\u002F13.deep-linking-and-params",{"title":3634,"path":3635,"stem":3636},"Робота з мережею та життєвий цикл застосунку","\u002Freact-native\u002Fversion2\u002Fnetworking-and-app-lifecycle","15.react-native\u002Fversion2\u002F14.networking-and-app-lifecycle",{"title":3638,"path":3639,"stem":3640},"Redux Toolkit на мобільних платформах","\u002Freact-native\u002Fversion2\u002Fredux-toolkit-on-mobile","15.react-native\u002Fversion2\u002F15.redux-toolkit-on-mobile",{"title":3642,"path":3643,"stem":3644},"RTK Query на мобільних платформах","\u002Freact-native\u002Fversion2\u002Frtk-query-mobile","15.react-native\u002Fversion2\u002F16.rtk-query-mobile",{"title":3646,"path":3647,"stem":3648},"Автентифікація та Безпечне Зберігання","\u002Freact-native\u002Fversion2\u002Fauth-secure-storage","15.react-native\u002Fversion2\u002F18.auth-secure-storage",{"title":3650,"path":3651,"stem":3652},"Локальне сховище — AsyncStorage, MMKV, SQLite","\u002Freact-native\u002Fversion2\u002Flocal-storage-mmkv-sqlite","15.react-native\u002Fversion2\u002F19.local-storage-mmkv-sqlite",{"title":3654,"path":3655,"stem":3656},"Офлайн-First і синхронізація","\u002Freact-native\u002Fversion2\u002Foffline-first-sync","15.react-native\u002Fversion2\u002F20.offline-first-sync",{"title":3658,"path":3659,"stem":3660},"Камера, медіа та файлова система","\u002Freact-native\u002Fversion2\u002Fcamera-media-filesystem","15.react-native\u002Fversion2\u002F21.camera-media-filesystem",{"title":3662,"path":3663,"stem":3664},"Карти та геолокація","\u002Freact-native\u002Fversion2\u002Fmaps-and-location","15.react-native\u002Fversion2\u002F22.maps-and-location",{"title":3666,"path":3667,"stem":3668},"Жести та дотики","\u002Freact-native\u002Fversion2\u002Fgestures-and-touch","15.react-native\u002Fversion2\u002F23.gestures-and-touch",{"title":3670,"path":3671,"stem":3672},"Анімації з Reanimated","\u002Freact-native\u002Fversion2\u002Freanimated-animations","15.react-native\u002Fversion2\u002F24.reanimated-animations",{"title":3674,"path":3675,"stem":3676},"Push і локальні сповіщення","\u002Freact-native\u002Fversion2\u002Fpush-and-local-notifications","15.react-native\u002Fversion2\u002F25.push-and-local-notifications",{"title":3678,"path":3679,"stem":3680},"Інтернаціоналізація та доступність","\u002Freact-native\u002Fversion2\u002Faccessibility-and-i18n","15.react-native\u002Fversion2\u002F26.accessibility-and-i18n",{"title":3682,"path":3683,"stem":3684},"EAS Build та профілі збірки","\u002Freact-native\u002Fversion2\u002Feas-build-and-profiles","15.react-native\u002Fversion2\u002F29.eas-build-and-profiles",{"title":3686,"path":3687,"stem":3688},"Публікація в App Store та Google Play","\u002Freact-native\u002Fversion2\u002Fapp-store-play-release","15.react-native\u002Fversion2\u002F30.app-store-play-release",{"title":3690,"path":3691,"stem":3692,"children":3693,"page":59},"Ai Python","\u002Fai-python","16.ai-python",[3694,3703,3712,3721,3750,3776,3801,3834,3851,3864,3873,3886,3895,3916,3929,3950],{"title":3695,"path":3696,"stem":3697,"children":3698,"page":59},"Intro To Ai","\u002Fai-python\u002Fintro-to-ai","16.ai-python\u002F01.intro-to-ai",[3699],{"title":3700,"path":3701,"stem":3702},"Вступ до штучного інтелекту та машинного навчання","\u002Fai-python\u002Fintro-to-ai\u002Fintro-to-ai","16.ai-python\u002F01.intro-to-ai\u002F01.intro-to-ai",{"title":3704,"path":3705,"stem":3706,"children":3707,"page":59},"Jupyter Notebook","\u002Fai-python\u002Fjupyter-notebook","16.ai-python\u002F02.jupyter-notebook",[3708],{"title":3709,"path":3710,"stem":3711},"Знайомство з Jupyter Notebook та JupyterLab","\u002Fai-python\u002Fjupyter-notebook\u002Fjupyter-notebook","16.ai-python\u002F02.jupyter-notebook\u002F02.jupyter-notebook",{"title":3713,"path":3714,"stem":3715,"children":3716,"page":59},"Numpy Pandas","\u002Fai-python\u002Fnumpy-pandas","16.ai-python\u002F03.numpy-pandas",[3717],{"title":3718,"path":3719,"stem":3720},"Робота з даними з використанням NumPy і pandas","\u002Fai-python\u002Fnumpy-pandas\u002Fnumpy-pandas","16.ai-python\u002F03.numpy-pandas\u002F03.numpy-pandas",{"title":3722,"path":3723,"stem":3724,"children":3725,"page":59},"Statistics Basics","\u002Fai-python\u002Fstatistics-basics","16.ai-python\u002F04.statistics-basics",[3726,3730,3734,3738,3742,3746],{"title":3727,"path":3728,"stem":3729},"Випадкові величини та їх типи","\u002Fai-python\u002Fstatistics-basics\u002Frandom-variables","16.ai-python\u002F04.statistics-basics\u002F01.random-variables",{"title":3731,"path":3732,"stem":3733},"Міри центральної тенденції","\u002Fai-python\u002Fstatistics-basics\u002Fcentral-tendency","16.ai-python\u002F04.statistics-basics\u002F02.central-tendency",{"title":3735,"path":3736,"stem":3737},"Міри варіації (розсіювання)","\u002Fai-python\u002Fstatistics-basics\u002Fvariability","16.ai-python\u002F04.statistics-basics\u002F03.variability",{"title":3739,"path":3740,"stem":3741},"Розподіли випадкових величин","\u002Fai-python\u002Fstatistics-basics\u002Fdistributions","16.ai-python\u002F04.statistics-basics\u002F04.distributions",{"title":3743,"path":3744,"stem":3745},"Кореляція та взаємозв'язки","\u002Fai-python\u002Fstatistics-basics\u002Fcorrelation","16.ai-python\u002F04.statistics-basics\u002F05.correlation",{"title":3747,"path":3748,"stem":3749},"Викиди та аномалії","\u002Fai-python\u002Fstatistics-basics\u002Foutliers","16.ai-python\u002F04.statistics-basics\u002F06.outliers",{"title":3751,"path":3752,"stem":3753,"children":3754},"Візуалізація даних за допомогою matplotlib","\u002Fai-python\u002Fdata-visualization","16.ai-python\u002F05.data-visualization\u002Findex",[3755,3756,3760,3764,3768,3772],{"title":3751,"path":3752,"stem":3753},{"title":3757,"path":3758,"stem":3759},"Основи matplotlib — Figure, Axes та перший графік","\u002Fai-python\u002Fdata-visualization\u002Fmatplotlib-basics","16.ai-python\u002F05.data-visualization\u002F01.matplotlib-basics",{"title":3761,"path":3762,"stem":3763},"Візуалізація розподілів — гістограми та box plots","\u002Fai-python\u002Fdata-visualization\u002Fdistributions","16.ai-python\u002F05.data-visualization\u002F02.distributions",{"title":3765,"path":3766,"stem":3767},"Візуалізація залежностей — scatter plots та кореляція","\u002Fai-python\u002Fdata-visualization\u002Frelationships","16.ai-python\u002F05.data-visualization\u002F03.relationships",{"title":3769,"path":3770,"stem":3771},"Візуалізація категоріальних даних — bar charts та pie charts","\u002Fai-python\u002Fdata-visualization\u002Fcategorical","16.ai-python\u002F05.data-visualization\u002F04.categorical",{"title":3773,"path":3774,"stem":3775},"Композиції та стилізація — subplots та багатопанельні дашборди","\u002Fai-python\u002Fdata-visualization\u002Fsubplots-styling","16.ai-python\u002F05.data-visualization\u002F05.subplots-styling",{"title":3777,"path":3778,"stem":3779,"children":3780,"page":59},"Linear Regression","\u002Fai-python\u002Flinear-regression","16.ai-python\u002F06.linear-regression",[3781,3785,3789,3793,3797],{"title":3782,"path":3783,"stem":3784},"Регресійна задача та лінійна регресія з однією змінною","\u002Fai-python\u002Flinear-regression\u002Fregression-task","16.ai-python\u002F06.linear-regression\u002F01.regression-task",{"title":3786,"path":3787,"stem":3788},"Метрики якості регресії — як оцінити модель","\u002Fai-python\u002Flinear-regression\u002Fquality-metrics","16.ai-python\u002F06.linear-regression\u002F02.quality-metrics",{"title":3790,"path":3791,"stem":3792},"Градієнтний спуск — як модель навчається","\u002Fai-python\u002Flinear-regression\u002Fgradient-descent","16.ai-python\u002F06.linear-regression\u002F03.gradient-descent",{"title":3794,"path":3795,"stem":3796},"Множинна лінійна регресія — модель з багатьма ознаками","\u002Fai-python\u002Flinear-regression\u002Fmultiple-regression","16.ai-python\u002F06.linear-regression\u002F04.multiple-regression",{"title":3798,"path":3799,"stem":3800},"Практичні завдання та резюме модуля","\u002Fai-python\u002Flinear-regression\u002Fpractice","16.ai-python\u002F06.linear-regression\u002F05.practice",{"title":3802,"path":3803,"stem":3804,"children":3805,"page":59},"Practical Regression","\u002Fai-python\u002Fpractical-regression","16.ai-python\u002F07.practical-regression",[3806,3810,3814,3818,3822,3826,3830],{"title":3807,"path":3808,"stem":3809},"Знайомство з scikit-learn","\u002Fai-python\u002Fpractical-regression\u002Fintro-scikit-learn","16.ai-python\u002F07.practical-regression\u002F01.intro-scikit-learn",{"title":3811,"path":3812,"stem":3813},"Exploratory Data Analysis датасету Housing Prices","\u002Fai-python\u002Fpractical-regression\u002Fhousing-dataset-eda","16.ai-python\u002F07.practical-regression\u002F02.housing-dataset-eda",{"title":3815,"path":3816,"stem":3817},"Попередня обробка даних (Data Preprocessing)","\u002Fai-python\u002Fpractical-regression\u002Fdata-preprocessing","16.ai-python\u002F07.practical-regression\u002F03.data-preprocessing",{"title":3819,"path":3820,"stem":3821},"Train\u002FTest Split та перша повноцінна модель","\u002Fai-python\u002Fpractical-regression\u002Ftrain-test-split","16.ai-python\u002F07.practical-regression\u002F04.train-test-split",{"title":3823,"path":3824,"stem":3825},"Перенавчання та недонавчання","\u002Fai-python\u002Fpractical-regression\u002Foverfitting-underfitting","16.ai-python\u002F07.practical-regression\u002F05.overfitting-underfitting",{"title":3827,"path":3828,"stem":3829},"Крос-валідація (Cross-Validation)","\u002Fai-python\u002Fpractical-regression\u002Fcross-validation","16.ai-python\u002F07.practical-regression\u002F06.cross-validation",{"title":3831,"path":3832,"stem":3833},"Гіперпараметри та їх оптимізація","\u002Fai-python\u002Fpractical-regression\u002Fhyperparameter-tuning","16.ai-python\u002F07.practical-regression\u002F07.hyperparameter-tuning",{"title":3835,"path":3836,"stem":3837,"children":3838,"page":59},"Polynomial Regression","\u002Fai-python\u002Fpolynomial-regression","16.ai-python\u002F08.polynomial-regression",[3839,3843,3847],{"title":3840,"path":3841,"stem":3842},"Нелінійні залежності та поліноміальна регресія","\u002Fai-python\u002Fpolynomial-regression\u002Fnonlinear-patterns","16.ai-python\u002F08.polynomial-regression\u002F01.nonlinear-patterns",{"title":3844,"path":3845,"stem":3846},"Регуляризація в регресії — Ridge та Lasso","\u002Fai-python\u002Fpolynomial-regression\u002Fregularization-methods","16.ai-python\u002F08.polynomial-regression\u002F02.regularization-methods",{"title":3848,"path":3849,"stem":3850},"Практика — поліноміальна регресія та регуляризація на Housing Prices","\u002Fai-python\u002Fpolynomial-regression\u002Fhousing-prices-practice","16.ai-python\u002F08.polynomial-regression\u002F03.housing-prices-practice",{"title":3852,"path":3853,"stem":3854,"children":3855,"page":59},"Decision Trees Random Forest","\u002Fai-python\u002Fdecision-trees-random-forest","16.ai-python\u002F09.decision-trees-random-forest",[3856,3860],{"title":3857,"path":3858,"stem":3859},"Дерева рішень — моделі з прозорою та зрозумілою логікою","\u002Fai-python\u002Fdecision-trees-random-forest\u002Fdecision-trees","16.ai-python\u002F09.decision-trees-random-forest\u002F01.decision-trees",{"title":3861,"path":3862,"stem":3863},"Random Forest — ансамблі дерев рішень","\u002Fai-python\u002Fdecision-trees-random-forest\u002Frandom-forest","16.ai-python\u002F09.decision-trees-random-forest\u002F02.random-forest",{"title":3865,"path":3866,"stem":3867,"children":3868,"page":59},"Gradient Boosting","\u002Fai-python\u002Fgradient-boosting","16.ai-python\u002F10.gradient-boosting",[3869],{"title":3870,"path":3871,"stem":3872},"Градієнтний бустинг та XGBoost — послідовне виправлення помилок","\u002Fai-python\u002Fgradient-boosting\u002Fgradient-boosting-xgboost","16.ai-python\u002F10.gradient-boosting\u002F01.gradient-boosting-xgboost",{"title":3874,"path":3875,"stem":3876,"children":3877,"page":59},"Logistic Regression Classification","\u002Fai-python\u002Flogistic-regression-classification","16.ai-python\u002F11.logistic-regression-classification",[3878,3882],{"title":3879,"path":3880,"stem":3881},"Основи класифікації та логістична регресія","\u002Fai-python\u002Flogistic-regression-classification\u002Fclassification-basics","16.ai-python\u002F11.logistic-regression-classification\u002F01.classification-basics",{"title":3883,"path":3884,"stem":3885},"Метрики якості класифікації","\u002Fai-python\u002Flogistic-regression-classification\u002Fclassification-metrics","16.ai-python\u002F11.logistic-regression-classification\u002F02.classification-metrics",{"title":3887,"path":3888,"stem":3889,"children":3890,"page":59},"Neural Networks Basics","\u002Fai-python\u002Fneural-networks-basics","16.ai-python\u002F12.neural-networks-basics",[3891],{"title":3892,"path":3893,"stem":3894},"Вступ до нейронних мереж — від персептрона до багатошарових мереж","\u002Fai-python\u002Fneural-networks-basics\u002Fperceptron-to-mlp","16.ai-python\u002F12.neural-networks-basics\u002F01.perceptron-to-mlp",{"title":3896,"path":3897,"stem":3898,"children":3899,"page":59},"Pytorch Intro","\u002Fai-python\u002Fpytorch-intro","16.ai-python\u002F13.pytorch-intro",[3900,3904,3908,3912],{"title":3901,"path":3902,"stem":3903},"Тензори та автоматичне диференціювання в PyTorch","\u002Fai-python\u002Fpytorch-intro\u002Ftensors-and-autograd","16.ai-python\u002F13.pytorch-intro\u002F01.tensors-and-autograd",{"title":3905,"path":3906,"stem":3907},"Побудова та тренування нейронних мереж у PyTorch","\u002Fai-python\u002Fpytorch-intro\u002Fbuilding-and-training","16.ai-python\u002F13.pytorch-intro\u002F02.building-and-training",{"title":3909,"path":3910,"stem":3911},"Робота з даними та регуляризація в PyTorch","\u002Fai-python\u002Fpytorch-intro\u002Fdata-and-regularization","16.ai-python\u002F13.pytorch-intro\u002F03.data-and-regularization",{"title":3913,"path":3914,"stem":3915},"Просунуті теми PyTorch — GPU, оптимізація та deployment","\u002Fai-python\u002Fpytorch-intro\u002Fadvanced-topics","16.ai-python\u002F13.pytorch-intro\u002F04.advanced-topics",{"title":3917,"path":3918,"stem":3919,"children":3920,"page":59},"Activation Functions Heart Disease","\u002Fai-python\u002Factivation-functions-heart-disease","16.ai-python\u002F14.activation-functions-heart-disease",[3921,3925],{"title":3922,"path":3923,"stem":3924},"Функції активації та побудова нейронної мережі для передбачення вад серця","\u002Fai-python\u002Factivation-functions-heart-disease\u002Fheart-disease-prediction","16.ai-python\u002F14.activation-functions-heart-disease\u002F01.heart-disease-prediction",{"title":3926,"path":3927,"stem":3928},"Бонус — Повний приклад Student Performance Prediction","\u002Fai-python\u002Factivation-functions-heart-disease\u002Fstudent-performance-bonus","16.ai-python\u002F14.activation-functions-heart-disease\u002F02.student-performance-bonus",{"title":3930,"path":3931,"stem":3932,"children":3933,"page":59},"Cnn Image Classification","\u002Fai-python\u002Fcnn-image-classification","16.ai-python\u002F15.cnn-image-classification",[3934,3938,3942,3946],{"title":3935,"path":3936,"stem":3937},"Основи згорткових нейронних мереж — чому не MLP?","\u002Fai-python\u002Fcnn-image-classification\u002Fcnn-fundamentals","16.ai-python\u002F15.cnn-image-classification\u002F01.cnn-fundamentals",{"title":3939,"path":3940,"stem":3941},"Практична імплементація CNN для класифікації MNIST","\u002Fai-python\u002Fcnn-image-classification\u002Fmnist-classification","16.ai-python\u002F15.cnn-image-classification\u002F02.mnist-classification",{"title":3943,"path":3944,"stem":3945},"Глибші CNN та аугментація даних для CIFAR-10","\u002Fai-python\u002Fcnn-image-classification\u002Fcifar10-and-data-augmentation","16.ai-python\u002F15.cnn-image-classification\u002F03.cifar10-and-data-augmentation",{"title":3947,"path":3948,"stem":3949},"Transfer Learning та візуалізація внутрішніх представлень CNN","\u002Fai-python\u002Fcnn-image-classification\u002Ftransfer-learning-and-visualization","16.ai-python\u002F15.cnn-image-classification\u002F04.transfer-learning-and-visualization",{"title":3951,"path":3952,"stem":3953,"children":3954,"page":59},"Batching Normalization Deep Networks","\u002Fai-python\u002Fbatching-normalization-deep-networks","16.ai-python\u002F16.batching-normalization-deep-networks",[3955,3959,3963,3967,3971,3975,3979],{"title":3956,"path":3957,"stem":3958},"Mini-batch Gradient Descent та ефективна обробка даних","\u002Fai-python\u002Fbatching-normalization-deep-networks\u002Fmini-batch-gradient-descent","16.ai-python\u002F16.batching-normalization-deep-networks\u002F01.mini-batch-gradient-descent",{"title":3960,"path":3961,"stem":3962},"Ініціалізація ваг — фундамент стабільного навчання","\u002Fai-python\u002Fbatching-normalization-deep-networks\u002Fweight-initialization","16.ai-python\u002F16.batching-normalization-deep-networks\u002F02.weight-initialization",{"title":3964,"path":3965,"stem":3966},"Batch Normalization — прорив у глибокому навчанні","\u002Fai-python\u002Fbatching-normalization-deep-networks\u002Fbatch-normalization","16.ai-python\u002F16.batching-normalization-deep-networks\u002F03.batch-normalization",{"title":3968,"path":3969,"stem":3970},"Dropout та комбіновані стратегії регуляризації","\u002Fai-python\u002Fbatching-normalization-deep-networks\u002Fdropout-and-regularization","16.ai-python\u002F16.batching-normalization-deep-networks\u002F04.dropout-and-regularization",{"title":3972,"path":3973,"stem":3974},"Проєктування глибоких мереж — від теорії до практики","\u002Fai-python\u002Fbatching-normalization-deep-networks\u002Fdeep-networks-architectures","16.ai-python\u002F16.batching-normalization-deep-networks\u002F05.deep-networks-architectures",{"title":3976,"path":3977,"stem":3978},"Планування learning rate — коли та як зменшувати крок навчання","\u002Fai-python\u002Fbatching-normalization-deep-networks\u002Flearning-rate-scheduling","16.ai-python\u002F16.batching-normalization-deep-networks\u002F06.learning-rate-scheduling",{"title":3980,"path":3981,"stem":3982},"Підсумковий проєкт — побудова сучасного класифікатора зображень","\u002Fai-python\u002Fbatching-normalization-deep-networks\u002Ffinal-project-deep-classifier","16.ai-python\u002F16.batching-normalization-deep-networks\u002F07.final-project-deep-classifier",{"title":3984,"icon":658,"path":3985,"stem":3986,"children":3987,"page":59},"Клієнт-серверна архітектура","\u002Fclient-server-architecture","17.client-server-architecture",[3988,3992,3996,4000,4004,4008,4012,4016,4020,4024,4028,4032,4036,4040,4044,4048,4052,4056],{"title":3989,"path":3990,"stem":3991},"Архітектура клієнт-серверних систем та основи Go","\u002Fclient-server-architecture\u002Fclient-server-architecture-and-go","17.client-server-architecture\u002F01.client-server-architecture-and-go",{"title":3993,"path":3994,"stem":3995},"Система типів та модель пам’яті мови Go","\u002Fclient-server-architecture\u002Ftype-system-and-memory-model","17.client-server-architecture\u002F02.type-system-and-memory-model",{"title":3997,"path":3998,"stem":3999},"Інтерфейси та модель конкурентності мови Go","\u002Fclient-server-architecture\u002Finterfaces-and-concurrency-model","17.client-server-architecture\u002F03.interfaces-and-concurrency-model",{"title":4001,"path":4002,"stem":4003},"Сокетне програмування та протокол TCP у Go","\u002Fclient-server-architecture\u002Ftcp-socket-programming","17.client-server-architecture\u002F04.tcp-socket-programming",{"title":4005,"path":4006,"stem":4007},"Протокол UDP та датаграмна передача даних","\u002Fclient-server-architecture\u002Fudp-datagram-transmission","17.client-server-architecture\u002F05.udp-datagram-transmission",{"title":4009,"path":4010,"stem":4011},"Криптографічний захист каналів зв","\u002Fclient-server-architecture\u002Ftls-cryptographic-channel-protection","17.client-server-architecture\u002F06.tls-cryptographic-channel-protection",{"title":4013,"path":4014,"stem":4015},"Прикладні мережеві протоколи DNS та SMTP","\u002Fclient-server-architecture\u002Fdns-smtp-application-protocols","17.client-server-architecture\u002F07.dns-smtp-application-protocols",{"title":4017,"path":4018,"stem":4019},"Протокол HTTP та стандартний вебсервер Go","\u002Fclient-server-architecture\u002Fhttp-protocol-and-web-server","17.client-server-architecture\u002F08.http-protocol-and-web-server",{"title":4021,"path":4022,"stem":4023},"Проєктування REST API та маршрутизація засобами Chi","\u002Fclient-server-architecture\u002Frest-api-design-and-chi-routing","17.client-server-architecture\u002F09.rest-api-design-and-chi-routing",{"title":4025,"path":4026,"stem":4027},"Лекція 10. Шарувата архітектура та конвеєри обробки запитів","\u002Fclient-server-architecture\u002Flayered-architecture-and-middleware-pipelines","17.client-server-architecture\u002F10.layered-architecture-and-middleware-pipelines",{"title":4029,"path":4030,"stem":4031},"Лекція 11. Архітектура та маршрутизація фреймворку Gin","\u002Fclient-server-architecture\u002Fgin-framework-architecture-and-routing","17.client-server-architecture\u002F11.gin-framework-architecture-and-routing",{"title":4033,"path":4034,"stem":4035},"Лекція 12. Валідація даних, безпека та документування у Gin","\u002Fclient-server-architecture\u002Fdata-validation-security-and-documentation-in-gin","17.client-server-architecture\u002F12.data-validation-security-and-documentation-in-gin",{"title":4037,"path":4038,"stem":4039},"Лекція 13. Взаємодія з реляційними СУБД та технології ORM","\u002Fclient-server-architecture\u002Frelational-databases-and-orm","17.client-server-architecture\u002F13.relational-databases-and-orm",{"title":4041,"path":4042,"stem":4043},"Потокова передача даних та протокол WebSocket","\u002Fclient-server-architecture\u002Fstreaming-data-and-websockets","17.client-server-architecture\u002F14.streaming-data-and-websockets",{"title":4045,"path":4046,"stem":4047},"Міжсервісна взаємодія на основі протоколу gRPC","\u002Fclient-server-architecture\u002Fgrpc-inter-service-communication","17.client-server-architecture\u002F15.grpc-inter-service-communication",{"title":4049,"path":4050,"stem":4051},"Асинхронний обмін повідомленнями та протоколи черг","\u002Fclient-server-architecture\u002Fasynchronous-message-exchange-and-queue-protocols","17.client-server-architecture\u002F16.asynchronous-message-exchange-and-queue-protocols",{"title":4053,"path":4054,"stem":4055},"Протоколи потокового медіа та технологія WebRTC","\u002Fclient-server-architecture\u002Fstreaming-media-protocols-and-webrtc","17.client-server-architecture\u002F17.streaming-media-protocols-and-webrtc",{"title":4057,"path":4058,"stem":4059},"Моніторинг, безпека та контейнеризація застосунків","\u002Fclient-server-architecture\u002Fmonitoring-security-and-containerization","17.client-server-architecture\u002F18.monitoring-security-and-containerization",{"title":4061,"icon":2118,"path":4062,"stem":4063,"children":4064,"page":59},"Розробка WEB-застосувань","\u002Fweb-development","18.web-development",[4065,4110,4148,4177,4206,4263,4313,4407,4473,4555,4588,4616,4658,4704,4747],{"title":4066,"icon":658,"path":4067,"stem":4068,"children":4069,"page":59},"Архітектура комп'ютерних мереж та мережеві протоколи","\u002Fweb-development\u002Fnetwork-architecture-and-protocols","18.web-development\u002F01.network-architecture-and-protocols",[4070,4074,4078,4082,4086,4090,4094,4098,4102,4106],{"title":4071,"path":4072,"stem":4073},"Еволюція веброзробки та модель клієнт-сервер","\u002Fweb-development\u002Fnetwork-architecture-and-protocols\u002Fweb-evolution-client-server","18.web-development\u002F01.network-architecture-and-protocols\u002F01.web-evolution-client-server",{"title":4075,"path":4076,"stem":4077},"Еталонна модель OSI","\u002Fweb-development\u002Fnetwork-architecture-and-protocols\u002Fosi-model","18.web-development\u002F01.network-architecture-and-protocols\u002F02.osi-model",{"title":4079,"path":4080,"stem":4081},"Стек протоколів TCP\u002FIP","\u002Fweb-development\u002Fnetwork-architecture-and-protocols\u002Ftcp-ip-stack","18.web-development\u002F01.network-architecture-and-protocols\u002F03.tcp-ip-stack",{"title":4083,"path":4084,"stem":4085},"IP-адресація, порти та сокети","\u002Fweb-development\u002Fnetwork-architecture-and-protocols\u002Fip-addressing-ports-sockets","18.web-development\u002F01.network-architecture-and-protocols\u002F04.ip-addressing-ports-sockets",{"title":4087,"path":4088,"stem":4089},"Протоколи TCP та UDP","\u002Fweb-development\u002Fnetwork-architecture-and-protocols\u002Ftcp-udp-protocols","18.web-development\u002F01.network-architecture-and-protocols\u002F05.tcp-udp-protocols",{"title":4091,"path":4092,"stem":4093},"Система доменних імен DNS","\u002Fweb-development\u002Fnetwork-architecture-and-protocols\u002Fdns-system","18.web-development\u002F01.network-architecture-and-protocols\u002F06.dns-system",{"title":4095,"path":4096,"stem":4097},"Протокол SMTP та поштові протоколи","\u002Fweb-development\u002Fnetwork-architecture-and-protocols\u002Fsmtp-email-protocols","18.web-development\u002F01.network-architecture-and-protocols\u002F07.smtp-email-protocols",{"title":4099,"path":4100,"stem":4101},"Протокол WebSocket","\u002Fweb-development\u002Fnetwork-architecture-and-protocols\u002Fwebsocket-protocol","18.web-development\u002F01.network-architecture-and-protocols\u002F08.websocket-protocol",{"title":4103,"path":4104,"stem":4105},"Протокол SSH","\u002Fweb-development\u002Fnetwork-architecture-and-protocols\u002Fssh-protocol","18.web-development\u002F01.network-architecture-and-protocols\u002F09.ssh-protocol",{"title":4107,"path":4108,"stem":4109},"Огляд сучасних серверних технологій","\u002Fweb-development\u002Fnetwork-architecture-and-protocols\u002Fmodern-backend-technologies","18.web-development\u002F01.network-architecture-and-protocols\u002F10.modern-backend-technologies",{"title":4111,"icon":4112,"path":4113,"stem":4114,"children":4115,"page":59},"Протокол HTTP та захищена передача даних HTTPS","i-lucide-lock","\u002Fweb-development\u002Fhttp-and-https","18.web-development\u002F02.http-and-https",[4116,4120,4124,4128,4132,4136,4140,4144],{"title":4117,"path":4118,"stem":4119},"Вступ до протоколу HTTP та еволюція веб-технологій","\u002Fweb-development\u002Fhttp-and-https\u002Fintroduction-and-evolution","18.web-development\u002F02.http-and-https\u002F01.introduction-and-evolution",{"title":4121,"path":4122,"stem":4123},"Анатомія HTTP-запиту","\u002Fweb-development\u002Fhttp-and-https\u002Fhttp-request-structure","18.web-development\u002F02.http-and-https\u002F02.http-request-structure",{"title":4125,"path":4126,"stem":4127},"Анатомія HTTP-відповіді","\u002Fweb-development\u002Fhttp-and-https\u002Fhttp-response-structure","18.web-development\u002F02.http-and-https\u002F03.http-response-structure",{"title":4129,"path":4130,"stem":4131},"Методи HTTP та їхня семантика","\u002Fweb-development\u002Fhttp-and-https\u002Fhttp-methods","18.web-development\u002F02.http-and-https\u002F04.http-methods",{"title":4133,"path":4134,"stem":4135},"Коди статусу HTTP та їхнє призначення","\u002Fweb-development\u002Fhttp-and-https\u002Fhttp-status-codes","18.web-development\u002F02.http-and-https\u002F05.http-status-codes",{"title":4137,"path":4138,"stem":4139},"HTTP-заголовки та їхнє призначення","\u002Fweb-development\u002Fhttp-and-https\u002Fhttp-headers","18.web-development\u002F02.http-and-https\u002F06.http-headers",{"title":4141,"path":4142,"stem":4143},"CORS та механізми безпеки HTTP","\u002Fweb-development\u002Fhttp-and-https\u002Fcors-and-security","18.web-development\u002F02.http-and-https\u002F07.cors-and-security",{"title":4145,"path":4146,"stem":4147},"HTTPS та криптографічні основи TLS","\u002Fweb-development\u002Fhttp-and-https\u002Fhttps-and-tls","18.web-development\u002F02.http-and-https\u002F08.https-and-tls",{"title":4149,"icon":2498,"path":4150,"stem":4151,"children":4152,"page":59},"Система типів мови TypeScript","\u002Fweb-development\u002Ftypescript-type-system","18.web-development\u002F03.typescript-type-system",[4153,4157,4161,4165,4169,4173],{"title":4154,"path":4155,"stem":4156},"Знайомство з TypeScript Compiler","\u002Fweb-development\u002Ftypescript-type-system\u002Fintroduction-compiler-config","18.web-development\u002F03.typescript-type-system\u002F01.introduction-compiler-config",{"title":4158,"path":4159,"stem":4160},"Примітивні типи даних: string, number, boolean","\u002Fweb-development\u002Ftypescript-type-system\u002Fprimitive-special-types","18.web-development\u002F03.typescript-type-system\u002F02.primitive-special-types",{"title":4162,"path":4163,"stem":4164},"Масиви, кортежі та об","\u002Fweb-development\u002Ftypescript-type-system\u002Farrays-tuples-objects","18.web-development\u002F03.typescript-type-system\u002F03.arrays-tuples-objects",{"title":4166,"path":4167,"stem":4168},"Псевдоніми типів, Union типи, Intersection типи та літеральні типи","\u002Fweb-development\u002Ftypescript-type-system\u002Funion-intersection-literals","18.web-development\u002F03.typescript-type-system\u002F04.union-intersection-literals",{"title":4170,"path":4171,"stem":4172},"Type Narrowing (звуження типів)","\u002Fweb-development\u002Ftypescript-type-system\u002Ftype-narrowing","18.web-development\u002F03.typescript-type-system\u002F05.type-narrowing",{"title":4174,"path":4175,"stem":4176},"Типізація функцій","\u002Fweb-development\u002Ftypescript-type-system\u002Ffunctions-typing","18.web-development\u002F03.typescript-type-system\u002F06.functions-typing",{"title":4178,"icon":2498,"path":4179,"stem":4180,"children":4181,"page":59},"Розширені засоби мови TypeScript","\u002Fweb-development\u002Ftypescript-advanced","18.web-development\u002F04.typescript-advanced",[4182,4186,4190,4194,4198,4202],{"title":4183,"path":4184,"stem":4185},"Interfaces та Classes","\u002Fweb-development\u002Ftypescript-advanced\u002Finterfaces-classes","18.web-development\u002F04.typescript-advanced\u002F01.interfaces-classes",{"title":4187,"path":4188,"stem":4189},"Основи Generic типів","\u002Fweb-development\u002Ftypescript-advanced\u002Fgenerics-fundamentals","18.web-development\u002F04.typescript-advanced\u002F02.generics-fundamentals",{"title":4191,"path":4192,"stem":4193},"Enums та Decorators","\u002Fweb-development\u002Ftypescript-advanced\u002Fenums-decorators","18.web-development\u002F04.typescript-advanced\u002F03.enums-decorators",{"title":4195,"path":4196,"stem":4197},"Utility Types","\u002Fweb-development\u002Ftypescript-advanced\u002Futility-types","18.web-development\u002F04.typescript-advanced\u002F04.utility-types",{"title":4199,"path":4200,"stem":4201},"Mapped та Conditional Types","\u002Fweb-development\u002Ftypescript-advanced\u002Fmapped-conditional-types","18.web-development\u002F04.typescript-advanced\u002F05.mapped-conditional-types",{"title":4203,"path":4204,"stem":4205},"Модулі та Strict Mode","\u002Fweb-development\u002Ftypescript-advanced\u002Fmodules-strict-mode","18.web-development\u002F04.typescript-advanced\u002F06.modules-strict-mode",{"title":4207,"icon":4208,"path":4209,"stem":4210,"children":4211,"page":59},"Платформа Node.js як середовище виконання","i-devicon-nodejs","\u002Fweb-development\u002Fnodejs-platform","18.web-development\u002F05.nodejs-platform",[4212,4216,4220,4224,4228,4232,4236,4240,4244,4248,4252,4255,4259],{"title":4213,"path":4214,"stem":4215},"Що таке Node.js та його архітектура","\u002Fweb-development\u002Fnodejs-platform\u002Fwhat-is-nodejs","18.web-development\u002F05.nodejs-platform\u002F01.what-is-nodejs",{"title":4217,"path":4218,"stem":4219},"Event Loop — серце Node.js","\u002Fweb-development\u002Fnodejs-platform\u002Fevent-loop","18.web-development\u002F05.nodejs-platform\u002F02.event-loop",{"title":4221,"path":4222,"stem":4223},"Модульні системи: CommonJS vs ES Modules","\u002Fweb-development\u002Fnodejs-platform\u002Fmodule-systems","18.web-development\u002F05.nodejs-platform\u002F03.module-systems",{"title":4225,"path":4226,"stem":4227},"npm та package.json","\u002Fweb-development\u002Fnodejs-platform\u002Fnpm-package-json","18.web-development\u002F05.nodejs-platform\u002F04.npm-package-json",{"title":4229,"path":4230,"stem":4231},"Глобальні об\\","\u002Fweb-development\u002Fnodejs-platform\u002Fglobal-objects","18.web-development\u002F05.nodejs-platform\u002F05.global-objects",{"title":4233,"path":4234,"stem":4235},"Вбудований модуль fs (File System)","\u002Fweb-development\u002Fnodejs-platform\u002Fbuilt-in-modules-fs","18.web-development\u002F05.nodejs-platform\u002F06.built-in-modules-fs",{"title":4237,"path":4238,"stem":4239},"Вбудований модуль path","\u002Fweb-development\u002Fnodejs-platform\u002Fbuilt-in-modules-path","18.web-development\u002F05.nodejs-platform\u002F07.built-in-modules-path",{"title":4241,"path":4242,"stem":4243},"Вбудований модуль http","\u002Fweb-development\u002Fnodejs-platform\u002Fbuilt-in-modules-http","18.web-development\u002F05.nodejs-platform\u002F08.built-in-modules-http",{"title":4245,"path":4246,"stem":4247},"Streams — потоки даних у Node.js","\u002Fweb-development\u002Fnodejs-platform\u002Fstreams","18.web-development\u002F05.nodejs-platform\u002F09.streams",{"title":4249,"path":4250,"stem":4251},"Вбудований модуль util (Utilities)","\u002Fweb-development\u002Fnodejs-platform\u002Fbuilt-in-modules-util","18.web-development\u002F05.nodejs-platform\u002F10.built-in-modules-util",{"title":3141,"path":4253,"stem":4254},"\u002Fweb-development\u002Fnodejs-platform\u002Fenvironment-variables","18.web-development\u002F05.nodejs-platform\u002F11.environment-variables",{"title":4256,"path":4257,"stem":4258},"Інструменти налагодження та розробки","\u002Fweb-development\u002Fnodejs-platform\u002Fdebugging-tools","18.web-development\u002F05.nodejs-platform\u002F12.debugging-tools",{"title":4260,"path":4261,"stem":4262},"Продуктивність та Best Practices","\u002Fweb-development\u002Fnodejs-platform\u002Fperformance-best-practices","18.web-development\u002F05.nodejs-platform\u002F13.performance-best-practices",{"title":4264,"icon":4265,"path":4266,"stem":4267,"children":4268,"page":59},"Асинхронне програмування у Node.js","i-lucide-clock","\u002Fweb-development\u002Fasync-programming-nodejs","18.web-development\u002F06.async-programming-nodejs",[4269,4273,4277,4281,4285,4289,4293,4297,4301,4305,4309],{"title":4270,"path":4271,"stem":4272},"Синхронний vs асинхронний код","\u002Fweb-development\u002Fasync-programming-nodejs\u002Fsync-vs-async","18.web-development\u002F06.async-programming-nodejs\u002F01.sync-vs-async",{"title":4274,"path":4275,"stem":4276},"Callbacks — перша парадигма асинхронності","\u002Fweb-development\u002Fasync-programming-nodejs\u002Fcallbacks","18.web-development\u002F06.async-programming-nodejs\u002F02.callbacks",{"title":4278,"path":4279,"stem":4280},"Promises — стандарт асинхронності","\u002Fweb-development\u002Fasync-programming-nodejs\u002Fpromises","18.web-development\u002F06.async-programming-nodejs\u002F03.promises",{"title":4282,"path":4283,"stem":4284},"Promise комбінатори","\u002Fweb-development\u002Fasync-programming-nodejs\u002Fpromise-combinators","18.web-development\u002F06.async-programming-nodejs\u002F04.promise-combinators",{"title":4286,"path":4287,"stem":4288},"Async\u002FAwait — синтаксичний цукор над Promises","\u002Fweb-development\u002Fasync-programming-nodejs\u002Fasync-await","18.web-development\u002F06.async-programming-nodejs\u002F05.async-await",{"title":4290,"path":4291,"stem":4292},"Паралельне vs послідовне виконання","\u002Fweb-development\u002Fasync-programming-nodejs\u002Fparallel-vs-sequential","18.web-development\u002F06.async-programming-nodejs\u002F06.parallel-vs-sequential",{"title":4294,"path":4295,"stem":4296},"EventEmitter — подієва модель Node.js","\u002Fweb-development\u002Fasync-programming-nodejs\u002Fevent-emitter","18.web-development\u002F06.async-programming-nodejs\u002F07.event-emitter",{"title":4298,"path":4299,"stem":4300},"Worker Threads — багатопотоковість у Node.js","\u002Fweb-development\u002Fasync-programming-nodejs\u002Fworker-threads","18.web-development\u002F06.async-programming-nodejs\u002F08.worker-threads",{"title":4302,"path":4303,"stem":4304},"Child Processes — запуск зовнішніх процесів","\u002Fweb-development\u002Fasync-programming-nodejs\u002Fchild-processes","18.web-development\u002F06.async-programming-nodejs\u002F09.child-processes",{"title":4306,"path":4307,"stem":4308},"Обробка помилок у асинхронному коді","\u002Fweb-development\u002Fasync-programming-nodejs\u002Ferror-handling","18.web-development\u002F06.async-programming-nodejs\u002F10.error-handling",{"title":4310,"path":4311,"stem":4312},"Порівняння Node.js фреймворків","\u002Fweb-development\u002Fasync-programming-nodejs\u002Fnodejs-frameworks","18.web-development\u002F06.async-programming-nodejs\u002F11.nodejs-frameworks",{"title":4314,"icon":4315,"path":4316,"stem":4317,"children":4318,"page":59},"Тема 7. Фреймворк NestJS. Контролери та маршрутизація","i-lucide-route","\u002Fweb-development\u002Fnestjs-controllers-routing","18.web-development\u002F07.nestjs-controllers-routing",[4319,4323,4327,4331,4335,4339,4343,4347,4351,4355,4359,4363,4367,4371,4375,4379,4383,4387,4391,4395,4399,4403],{"title":4320,"path":4321,"stem":4322},"Філософія та архітектура NestJS","\u002Fweb-development\u002Fnestjs-controllers-routing\u002Fnestjs-philosophy","18.web-development\u002F07.nestjs-controllers-routing\u002F01.nestjs-philosophy",{"title":4324,"path":4325,"stem":4326},"Три стовпи архітектури: модулі, контролери, провайдери","\u002Fweb-development\u002Fnestjs-controllers-routing\u002Fthree-pillars","18.web-development\u002F07.nestjs-controllers-routing\u002F02.three-pillars",{"title":4328,"path":4329,"stem":4330},"Інверсія керування (IoC) та ін","\u002Fweb-development\u002Fnestjs-controllers-routing\u002Fioc-and-di","18.web-development\u002F07.nestjs-controllers-routing\u002F03.ioc-and-di",{"title":4332,"path":4333,"stem":4334},"Встановлення NestJS CLI та створення проєкту","\u002Fweb-development\u002Fnestjs-controllers-routing\u002Fnestjs-cli-setup","18.web-development\u002F07.nestjs-controllers-routing\u002F04.nestjs-cli-setup",{"title":4336,"path":4337,"stem":4338},"Структура проєкту NestJS та конвенції","\u002Fweb-development\u002Fnestjs-controllers-routing\u002Fproject-structure","18.web-development\u002F07.nestjs-controllers-routing\u002F05.project-structure",{"title":4340,"path":4341,"stem":4342},"Контролери: основи маршрутизації","\u002Fweb-development\u002Fnestjs-controllers-routing\u002Fcontrollers-basics","18.web-development\u002F07.nestjs-controllers-routing\u002F06.controllers-basics",{"title":4344,"path":4345,"stem":4346},"Декоратори маршрутизації HTTP-методів","\u002Fweb-development\u002Fnestjs-controllers-routing\u002Froute-decorators","18.web-development\u002F07.nestjs-controllers-routing\u002F07.route-decorators",{"title":4348,"path":4349,"stem":4350},"Параметри маршруту та декоратор @Param","\u002Fweb-development\u002Fnestjs-controllers-routing\u002Froute-parameters","18.web-development\u002F07.nestjs-controllers-routing\u002F08.route-parameters",{"title":4352,"path":4353,"stem":4354},"Query-параметри та декоратор @Query","\u002Fweb-development\u002Fnestjs-controllers-routing\u002Fquery-parameters","18.web-development\u002F07.nestjs-controllers-routing\u002F09.query-parameters",{"title":4356,"path":4357,"stem":4358},"Тіло запиту та декоратор @Body","\u002Fweb-development\u002Fnestjs-controllers-routing\u002Frequest-body","18.web-development\u002F07.nestjs-controllers-routing\u002F10.request-body",{"title":4360,"path":4361,"stem":4362},"Робота з заголовками запиту та відповіді","\u002Fweb-development\u002Fnestjs-controllers-routing\u002Fheaders-decorators","18.web-development\u002F07.nestjs-controllers-routing\u002F11.headers-decorators",{"title":4364,"path":4365,"stem":4366},"Data Transfer Objects (DTO): визначення та призначення","\u002Fweb-development\u002Fnestjs-controllers-routing\u002Fdto-introduction","18.web-development\u002F07.nestjs-controllers-routing\u002F12.dto-introduction",{"title":4368,"path":4369,"stem":4370},"Об'єкти Request та Response","\u002Fweb-development\u002Fnestjs-controllers-routing\u002Frequest-response-objects","18.web-development\u002F07.nestjs-controllers-routing\u002F13.request-response-objects",{"title":4372,"path":4373,"stem":4374},"HTTP-статус коди та декоратор @HttpCode","\u002Fweb-development\u002Fnestjs-controllers-routing\u002Fhttp-status-codes","18.web-development\u002F07.nestjs-controllers-routing\u002F14.http-status-codes",{"title":4376,"path":4377,"stem":4378},"Перенаправлення (Redirects) та декоратор @Redirect","\u002Fweb-development\u002Fnestjs-controllers-routing\u002Fredirects","18.web-development\u002F07.nestjs-controllers-routing\u002F15.redirects",{"title":4380,"path":4381,"stem":4382},"Асинхронні обробники та Observables","\u002Fweb-development\u002Fnestjs-controllers-routing\u002Fasync-handlers","18.web-development\u002F07.nestjs-controllers-routing\u002F16.async-handlers",{"title":4384,"path":4385,"stem":4386},"Маршрутизація за піддоменами та версіонування API","\u002Fweb-development\u002Fnestjs-controllers-routing\u002Fsub-domain-routing","18.web-development\u002F07.nestjs-controllers-routing\u002F17.sub-domain-routing",{"title":4388,"path":4389,"stem":4390},"Області видимості контролерів (Scopes)","\u002Fweb-development\u002Fnestjs-controllers-routing\u002Fcontroller-scopes","18.web-development\u002F07.nestjs-controllers-routing\u002F18.controller-scopes",{"title":4392,"path":4393,"stem":4394},"Основні команди NestJS CLI","\u002Fweb-development\u002Fnestjs-controllers-routing\u002Fnestjs-cli-commands","18.web-development\u002F07.nestjs-controllers-routing\u002F19.nestjs-cli-commands",{"title":4396,"path":4397,"stem":4398},"Практичний приклад: створення CRUD-контролера","\u002Fweb-development\u002Fnestjs-controllers-routing\u002Fcrud-example","18.web-development\u002F07.nestjs-controllers-routing\u002F20.crud-example",{"title":4400,"path":4401,"stem":4402},"Тестування контролерів","\u002Fweb-development\u002Fnestjs-controllers-routing\u002Ftesting-controllers","18.web-development\u002F07.nestjs-controllers-routing\u002F21.testing-controllers",{"title":4404,"path":4405,"stem":4406},"Best Practices для контролерів та маршрутизації","\u002Fweb-development\u002Fnestjs-controllers-routing\u002Fbest-practices","18.web-development\u002F07.nestjs-controllers-routing\u002F22.best-practices",{"title":4408,"icon":4409,"path":4410,"stem":4411,"children":4412,"page":59},"Тема 8. Провайдери, сервіси та модульна система NestJS","i-lucide-package","\u002Fweb-development\u002Fnestjs-providers-modules","18.web-development\u002F08.nestjs-providers-modules",[4413,4417,4421,4425,4429,4433,4437,4441,4445,4449,4453,4457,4461,4465,4469],{"title":4414,"path":4415,"stem":4416},"Концепція провайдерів у NestJS","\u002Fweb-development\u002Fnestjs-providers-modules\u002Fproviders-concept","18.web-development\u002F08.nestjs-providers-modules\u002F01.providers-concept",{"title":4418,"path":4419,"stem":4420},"Сервіси: інкапсуляція бізнес-логіки","\u002Fweb-development\u002Fnestjs-providers-modules\u002Fservices-business-logic","18.web-development\u002F08.nestjs-providers-modules\u002F02.services-business-logic",{"title":4422,"path":4423,"stem":4424},"Практика ін","\u002Fweb-development\u002Fnestjs-providers-modules\u002Fdependency-injection-practice","18.web-development\u002F08.nestjs-providers-modules\u002F03.dependency-injection-practice",{"title":4426,"path":4427,"stem":4428},"Custom Providers: useValue, useClass, useFactory, useExisting","\u002Fweb-development\u002Fnestjs-providers-modules\u002Fcustom-providers","18.web-development\u002F08.nestjs-providers-modules\u002F04.custom-providers",{"title":4430,"path":4431,"stem":4432},"Injection Tokens: рядкові та символьні токени","\u002Fweb-development\u002Fnestjs-providers-modules\u002Finjection-tokens","18.web-development\u002F08.nestjs-providers-modules\u002F05.injection-tokens",{"title":4434,"path":4435,"stem":4436},"Опціональні провайдери та декоратор @Optional","\u002Fweb-development\u002Fnestjs-providers-modules\u002Foptional-providers","18.web-development\u002F08.nestjs-providers-modules\u002F07.optional-providers",{"title":4438,"path":4439,"stem":4440},"Структура модулів: imports, controllers, providers, exports","\u002Fweb-development\u002Fnestjs-providers-modules\u002Fmodules-structure","18.web-development\u002F08.nestjs-providers-modules\u002F08.modules-structure",{"title":4442,"path":4443,"stem":4444},"Feature Modules: ізоляція функціональності","\u002Fweb-development\u002Fnestjs-providers-modules\u002Ffeature-modules","18.web-development\u002F08.nestjs-providers-modules\u002F09.feature-modules",{"title":4446,"path":4447,"stem":4448},"Shared Modules: перевикористання провайдерів","\u002Fweb-development\u002Fnestjs-providers-modules\u002Fshared-modules","18.web-development\u002F08.nestjs-providers-modules\u002F10.shared-modules",{"title":4450,"path":4451,"stem":4452},"Global Modules та декоратор @Global","\u002Fweb-development\u002Fnestjs-providers-modules\u002Fglobal-modules","18.web-development\u002F08.nestjs-providers-modules\u002F11.global-modules",{"title":4454,"path":4455,"stem":4456},"Dynamic Modules: конфігурація під час виконання","\u002Fweb-development\u002Fnestjs-providers-modules\u002Fdynamic-modules","18.web-development\u002F08.nestjs-providers-modules\u002F12.dynamic-modules",{"title":4458,"path":4459,"stem":4460},"Циклічні залежності та forwardRef","\u002Fweb-development\u002Fnestjs-providers-modules\u002Fcircular-dependencies","18.web-development\u002F08.nestjs-providers-modules\u002F13.circular-dependencies",{"title":4462,"path":4463,"stem":4464},"Lifecycle Hooks: хуки життєвого циклу","\u002Fweb-development\u002Fnestjs-providers-modules\u002Flifecycle-hooks","18.web-development\u002F08.nestjs-providers-modules\u002F14.lifecycle-hooks",{"title":4466,"path":4467,"stem":4468},"ModuleRef: динамічний доступ до провайдерів","\u002Fweb-development\u002Fnestjs-providers-modules\u002Fmodule-reference","18.web-development\u002F08.nestjs-providers-modules\u002F15.module-reference",{"title":4470,"path":4471,"stem":4472},"Тестування сервісів та провайдерів","\u002Fweb-development\u002Fnestjs-providers-modules\u002Ftesting-services","18.web-development\u002F08.nestjs-providers-modules\u002F16.testing-services",{"title":4474,"icon":4475,"path":4476,"stem":4477,"children":4478,"page":59},"Тема 9. Валідація даних та конвеєр обробки запитів у NestJS","i-lucide-filter","\u002Fweb-development\u002Fnestjs-validation-pipeline","18.web-development\u002F09.nestjs-validation-pipeline",[4479,4483,4487,4491,4495,4499,4503,4507,4511,4515,4519,4523,4527,4531,4535,4539,4543,4547,4551],{"title":4480,"path":4481,"stem":4482},"Request Pipeline: конвеєр обробки запитів","\u002Fweb-development\u002Fnestjs-validation-pipeline\u002Frequest-pipeline","18.web-development\u002F09.nestjs-validation-pipeline\u002F01.request-pipeline",{"title":4484,"path":4485,"stem":4486},"Pipes: концепція трансформації та валідації","\u002Fweb-development\u002Fnestjs-validation-pipeline\u002Fpipes-concept","18.web-development\u002F09.nestjs-validation-pipeline\u002F02.pipes-concept",{"title":4488,"path":4489,"stem":4490},"Вбудовані pipes: ParseIntPipe, ParseBoolPipe, ParseUUIDPipe","\u002Fweb-development\u002Fnestjs-validation-pipeline\u002Fbuilt-in-pipes","18.web-development\u002F09.nestjs-validation-pipeline\u002F03.built-in-pipes",{"title":4492,"path":4493,"stem":4494},"ValidationPipe: автоматична валідація DTO","\u002Fweb-development\u002Fnestjs-validation-pipeline\u002Fvalidation-pipe","18.web-development\u002F09.nestjs-validation-pipeline\u002F04.validation-pipe",{"title":4496,"path":4497,"stem":4498},"Декоратори class-validator для DTO","\u002Fweb-development\u002Fnestjs-validation-pipeline\u002Fclass-validator-decorators","18.web-development\u002F09.nestjs-validation-pipeline\u002F05.class-validator-decorators",{"title":4500,"path":4501,"stem":4502},"Створення кастомних валідаторів","\u002Fweb-development\u002Fnestjs-validation-pipeline\u002Fcustom-validators","18.web-development\u002F09.nestjs-validation-pipeline\u002F06.custom-validators",{"title":4504,"path":4505,"stem":4506},"Створення власних Pipes","\u002Fweb-development\u002Fnestjs-validation-pipeline\u002Fcustom-pipes","18.web-development\u002F09.nestjs-validation-pipeline\u002F07.custom-pipes",{"title":4508,"path":4509,"stem":4510},"Основи Middleware","\u002Fweb-development\u002Fnestjs-validation-pipeline\u002Fmiddleware-basics","18.web-development\u002F09.nestjs-validation-pipeline\u002F08.middleware-basics",{"title":4512,"path":4513,"stem":4514},"Практичні приклади Middleware","\u002Fweb-development\u002Fnestjs-validation-pipeline\u002Fmiddleware-examples","18.web-development\u002F09.nestjs-validation-pipeline\u002F09.middleware-examples",{"title":4516,"path":4517,"stem":4518},"Guards: захисні перехоплювачі доступу","\u002Fweb-development\u002Fnestjs-validation-pipeline\u002Fguards-concept","18.web-development\u002F09.nestjs-validation-pipeline\u002F10.guards-concept",{"title":4520,"path":4521,"stem":4522},"Створення кастомних Guards","\u002Fweb-development\u002Fnestjs-validation-pipeline\u002Fcustom-guards","18.web-development\u002F09.nestjs-validation-pipeline\u002F11.custom-guards",{"title":4524,"path":4525,"stem":4526},"Interceptors: перехоплення та трансформація відповідей","\u002Fweb-development\u002Fnestjs-validation-pipeline\u002Finterceptors-concept","18.web-development\u002F09.nestjs-validation-pipeline\u002F12.interceptors-concept",{"title":4528,"path":4529,"stem":4530},"Практичні приклади Interceptors","\u002Fweb-development\u002Fnestjs-validation-pipeline\u002Finterceptor-examples","18.web-development\u002F09.nestjs-validation-pipeline\u002F13.interceptor-examples",{"title":4532,"path":4533,"stem":4534},"Exception Filters: обробка виключень","\u002Fweb-development\u002Fnestjs-validation-pipeline\u002Fexception-filters","18.web-development\u002F09.nestjs-validation-pipeline\u002F14.exception-filters",{"title":4536,"path":4537,"stem":4538},"Custom Exception Filters: спеціалізовані фільтри","\u002Fweb-development\u002Fnestjs-validation-pipeline\u002Fcustom-exception-filters","18.web-development\u002F09.nestjs-validation-pipeline\u002F15.custom-exception-filters",{"title":4540,"path":4541,"stem":4542},"Execution Context: контекст виконання","\u002Fweb-development\u002Fnestjs-validation-pipeline\u002Fexecution-context","18.web-development\u002F09.nestjs-validation-pipeline\u002F16.execution-context",{"title":4544,"path":4545,"stem":4546},"Custom Decorators: власні декоратори","\u002Fweb-development\u002Fnestjs-validation-pipeline\u002Fcustom-decorators","18.web-development\u002F09.nestjs-validation-pipeline\u002F17.custom-decorators",{"title":4548,"path":4549,"stem":4550},"Global Components: глобальні компоненти","\u002Fweb-development\u002Fnestjs-validation-pipeline\u002Fglobal-components","18.web-development\u002F09.nestjs-validation-pipeline\u002F18.global-components",{"title":4552,"path":4553,"stem":4554},"Pipeline Best Practices: найкращі практики","\u002Fweb-development\u002Fnestjs-validation-pipeline\u002Fpipeline-best-practices","18.web-development\u002F09.nestjs-validation-pipeline\u002F19.pipeline-best-practices",{"title":4556,"icon":2865,"path":4557,"stem":4558,"children":4559,"page":59},"Інтеграція реляційних баз даних засобами TypeORM","\u002Fweb-development\u002Ftypeorm-integration","18.web-development\u002F10.typeorm-integration",[4560,4564,4568,4572,4576,4580,4584],{"title":4561,"path":4562,"stem":4563},"Робота з базами даних у Node.js через нативні драйвери","\u002Fweb-development\u002Ftypeorm-integration\u002F01-nodejs-database-drivers","18.web-development\u002F10.typeorm-integration\u002F01-nodejs-database-drivers",{"title":4565,"path":4566,"stem":4567},"ORM та знайомство з TypeORM","\u002Fweb-development\u002Ftypeorm-integration\u002F02-orm-and-typeorm-intro","18.web-development\u002F10.typeorm-integration\u002F02-orm-and-typeorm-intro",{"title":4569,"path":4570,"stem":4571},"Налаштування TypeORM у NestJS проєкті","\u002Fweb-development\u002Ftypeorm-integration\u002F03-typeorm-setup-nestjs","18.web-development\u002F10.typeorm-integration\u002F03-typeorm-setup-nestjs",{"title":4573,"path":4574,"stem":4575},"Entity класи та декоратори TypeORM","\u002Fweb-development\u002Ftypeorm-integration\u002F04-entities-and-decorators","18.web-development\u002F10.typeorm-integration\u002F04-entities-and-decorators",{"title":4577,"path":4578,"stem":4579},"Repository Pattern та CRUD операції","\u002Fweb-development\u002Ftypeorm-integration\u002F05-repository-pattern-crud","18.web-development\u002F10.typeorm-integration\u002F05-repository-pattern-crud",{"title":4581,"path":4582,"stem":4583},"Пагінація та фільтрація даних","\u002Fweb-development\u002Ftypeorm-integration\u002F06-pagination-filtering","18.web-development\u002F10.typeorm-integration\u002F06-pagination-filtering",{"title":4585,"path":4586,"stem":4587},"QueryBuilder для складних запитів","\u002Fweb-development\u002Ftypeorm-integration\u002F07-query-builder","18.web-development\u002F10.typeorm-integration\u002F07-query-builder",{"title":4589,"icon":4590,"path":4591,"stem":4592,"children":4593,"page":59},"Зв'язки між сутностями, міграції та оптимізація запитів","i-lucide-git-branch","\u002Fweb-development\u002Fentity-relations-migrations","18.web-development\u002F11.entity-relations-migrations",[4594,4598,4602,4606,4610],{"title":4595,"path":4596,"stem":4597},"Зв'язки One-to-Many та Many-to-One","\u002Fweb-development\u002Fentity-relations-migrations\u002F01-one-to-many-many-to-one","18.web-development\u002F11.entity-relations-migrations\u002F01-one-to-many-many-to-one",{"title":4599,"path":4600,"stem":4601},"Зв'язки One-to-One та Many-to-Many","\u002Fweb-development\u002Fentity-relations-migrations\u002F02-one-to-one-many-to-many","18.web-development\u002F11.entity-relations-migrations\u002F02-one-to-one-many-to-many",{"title":4603,"path":4604,"stem":4605},"Керування міграціями бази даних","\u002Fweb-development\u002Fentity-relations-migrations\u002F03-migrations-management","18.web-development\u002F11.entity-relations-migrations\u002F03-migrations-management",{"title":4607,"path":4608,"stem":4609},"Транзакції у TypeORM","\u002Fweb-development\u002Fentity-relations-migrations\u002F04-transactions","18.web-development\u002F11.entity-relations-migrations\u002F04-transactions",{"title":4611,"path":4612,"stem":4613,"children":4614},"Індекси та оптимізація запитів","\u002Fweb-development\u002Fentity-relations-migrations\u002F05-indexes-optimization","18.web-development\u002F11.entity-relations-migrations\u002F05-indexes-optimization",[4615],{"title":4611,"path":4612,"stem":4613},{"title":4617,"icon":4618,"path":4619,"stem":4620,"children":4621,"page":59},"Механізми автентифікації у вебзастосунках","i-lucide-key","\u002Fweb-development\u002Fauthentication","18.web-development\u002F12.authentication",[4622,4626,4630,4634,4638,4642,4646,4650,4654],{"title":4623,"path":4624,"stem":4625},"Автентифікація vs Авторизація","\u002Fweb-development\u002Fauthentication\u002F01-authentication-vs-authorization","18.web-development\u002F12.authentication\u002F01-authentication-vs-authorization",{"title":4627,"path":4628,"stem":4629},"Стратегії автентифікації","\u002Fweb-development\u002Fauthentication\u002F02-authentication-strategies","18.web-development\u002F12.authentication\u002F02-authentication-strategies",{"title":4631,"path":4632,"stem":4633},"Автентифікація на основі сесій","\u002Fweb-development\u002Fauthentication\u002F03-session-based-authentication","18.web-development\u002F12.authentication\u002F03-session-based-authentication",{"title":4635,"path":4636,"stem":4637},"Структура JSON Web Token (JWT)","\u002Fweb-development\u002Fauthentication\u002F04-jwt-structure","18.web-development\u002F12.authentication\u002F04-jwt-structure",{"title":4639,"path":4640,"stem":4641},"Інтеграція JWT у NestJS","\u002Fweb-development\u002Fauthentication\u002F05-nestjs-jwt-integration","18.web-development\u002F12.authentication\u002F05-nestjs-jwt-integration",{"title":4643,"path":4644,"stem":4645},"Passport.js стратегії у NestJS","\u002Fweb-development\u002Fauthentication\u002F06-passport-strategies","18.web-development\u002F12.authentication\u002F06-passport-strategies",{"title":4647,"path":4648,"stem":4649},"Безпечне зберігання паролів: хешування та криптографічний захист","\u002Fweb-development\u002Fauthentication\u002F07-password-hashing","18.web-development\u002F12.authentication\u002F07-password-hashing",{"title":4651,"path":4652,"stem":4653},"Refresh Token та безпечне оновлення доступу","\u002Fweb-development\u002Fauthentication\u002F08-refresh-tokens","18.web-development\u002F12.authentication\u002F08-refresh-tokens",{"title":4655,"path":4656,"stem":4657},"OAuth 2.0 та OpenID Connect: делегована автентифікація","\u002Fweb-development\u002Fauthentication\u002F09-oauth2-openid","18.web-development\u002F12.authentication\u002F09-oauth2-openid",{"title":4659,"icon":4660,"path":4661,"stem":4662,"children":4663,"page":59},"Авторизація та безпека серверних застосунків","i-lucide-shield","\u002Fweb-development\u002Fauthorization-security","18.web-development\u002F13.authorization-security",[4664,4668,4672,4676,4680,4684,4688,4692,4696,4700],{"title":4665,"path":4666,"stem":4667},"Моделі авторизації","\u002Fweb-development\u002Fauthorization-security\u002F01-authorization-models","18.web-development\u002F13.authorization-security\u002F01-authorization-models",{"title":4669,"path":4670,"stem":4671},"Реалізація RBAC у NestJS","\u002Fweb-development\u002Fauthorization-security\u002F02-rbac-implementation","18.web-development\u002F13.authorization-security\u002F02-rbac-implementation",{"title":4673,"path":4674,"stem":4675},"Permission-Based Access Control","\u002Fweb-development\u002Fauthorization-security\u002F03-permission-based-access","18.web-development\u002F13.authorization-security\u002F03-permission-based-access",{"title":4677,"path":4678,"stem":4679},"ABAC та бібліотека CASL","\u002Fweb-development\u002Fauthorization-security\u002F04-abac-casl","18.web-development\u002F13.authorization-security\u002F04-abac-casl",{"title":4681,"path":4682,"stem":4683},"Кастомні декоратори та метадані","\u002Fweb-development\u002Fauthorization-security\u002F05-custom-decorators-metadata","18.web-development\u002F13.authorization-security\u002F05-custom-decorators-metadata",{"title":4685,"path":4686,"stem":4687},"Конфігурація CORS","\u002Fweb-development\u002Fauthorization-security\u002F06-cors-configuration","18.web-development\u002F13.authorization-security\u002F06-cors-configuration",{"title":4689,"path":4690,"stem":4691},"Захист через HTTP Security Headers (Helmet)","\u002Fweb-development\u002Fauthorization-security\u002F07-helmet-security-headers","18.web-development\u002F13.authorization-security\u002F07-helmet-security-headers",{"title":4693,"path":4694,"stem":4695},"Rate Limiting та Throttling","\u002Fweb-development\u002Fauthorization-security\u002F08-rate-limiting-throttling","18.web-development\u002F13.authorization-security\u002F08-rate-limiting-throttling",{"title":4697,"path":4698,"stem":4699},"Захист від поширених атак","\u002Fweb-development\u002Fauthorization-security\u002F09-common-attacks-protection","18.web-development\u002F13.authorization-security\u002F09-common-attacks-protection",{"title":4701,"path":4702,"stem":4703},"HTTPS у продакшені","\u002Fweb-development\u002Fauthorization-security\u002F10-https-production","18.web-development\u002F13.authorization-security\u002F10-https-production",{"title":4705,"icon":871,"path":4706,"stem":4707,"children":4708,"page":59},"Комунікація у реальному часі, нотифікації та фонові задачі","\u002Fweb-development\u002Frealtime-notifications-jobs","18.web-development\u002F14.realtime-notifications-jobs",[4709,4713,4717,4721,4725,4729,4732,4735,4739,4743],{"title":4710,"path":4711,"stem":4712},"Методи комунікації у реальному часі","\u002Fweb-development\u002Frealtime-notifications-jobs\u002F01-realtime-communication-methods","18.web-development\u002F14.realtime-notifications-jobs\u002F01-realtime-communication-methods",{"title":4714,"path":4715,"stem":4716},"WebSocket Gateway у NestJS","\u002Fweb-development\u002Frealtime-notifications-jobs\u002F02-websocket-gateway","18.web-development\u002F14.realtime-notifications-jobs\u002F02-websocket-gateway",{"title":4718,"path":4719,"stem":4720},"Lifecycle та Rooms у WebSocket","\u002Fweb-development\u002Frealtime-notifications-jobs\u002F03-websocket-lifecycle-rooms","18.web-development\u002F14.realtime-notifications-jobs\u002F03-websocket-lifecycle-rooms",{"title":4722,"path":4723,"stem":4724},"Server-Sent Events (SSE)","\u002Fweb-development\u002Frealtime-notifications-jobs\u002F04-server-sent-events","18.web-development\u002F14.realtime-notifications-jobs\u002F04-server-sent-events",{"title":4726,"path":4727,"stem":4728},"Архітектура системи нотифікацій","\u002Fweb-development\u002Frealtime-notifications-jobs\u002F05-notification-system","18.web-development\u002F14.realtime-notifications-jobs\u002F05-notification-system",{"title":904,"path":4730,"stem":4731},"\u002Fweb-development\u002Frealtime-notifications-jobs\u002F06-email-notifications","18.web-development\u002F14.realtime-notifications-jobs\u002F06-email-notifications",{"title":900,"path":4733,"stem":4734},"\u002Fweb-development\u002Frealtime-notifications-jobs\u002F07-web-push-notifications","18.web-development\u002F14.realtime-notifications-jobs\u002F07-web-push-notifications",{"title":4736,"path":4737,"stem":4738},"Фонові задачі та черги","\u002Fweb-development\u002Frealtime-notifications-jobs\u002F08-background-jobs-queues","18.web-development\u002F14.realtime-notifications-jobs\u002F08-background-jobs-queues",{"title":4740,"path":4741,"stem":4742},"Планувальники та Cron jobs","\u002Fweb-development\u002Frealtime-notifications-jobs\u002F09-schedulers-cron","18.web-development\u002F14.realtime-notifications-jobs\u002F09-schedulers-cron",{"title":4744,"path":4745,"stem":4746},"Практичні сценарії застосування","\u002Fweb-development\u002Frealtime-notifications-jobs\u002F10-practical-scenarios","18.web-development\u002F14.realtime-notifications-jobs\u002F10-practical-scenarios",{"title":4748,"icon":4749,"path":4750,"stem":4751,"children":4752,"page":59},"Тестування, документування та розгортання застосунків","i-lucide-rocket","\u002Fweb-development\u002Ftesting-documentation-deployment","18.web-development\u002F15.testing-documentation-deployment",[4753,4757,4761,4765,4769,4773,4777,4781,4785,4789],{"title":4754,"path":4755,"stem":4756},"Піраміда тестування та стратегія","\u002Fweb-development\u002Ftesting-documentation-deployment\u002F01-testing-pyramid-strategy","18.web-development\u002F15.testing-documentation-deployment\u002F01-testing-pyramid-strategy",{"title":4758,"path":4759,"stem":4760},"Основи Jest у NestJS","\u002Fweb-development\u002Ftesting-documentation-deployment\u002F02-jest-nestjs-basics","18.web-development\u002F15.testing-documentation-deployment\u002F02-jest-nestjs-basics",{"title":4762,"path":4763,"stem":4764},"Unit-тестування сервісів","\u002Fweb-development\u002Ftesting-documentation-deployment\u002F03-unit-testing-services","18.web-development\u002F15.testing-documentation-deployment\u002F03-unit-testing-services",{"title":4766,"path":4767,"stem":4768},"E2E тестування та покриття коду","\u002Fweb-development\u002Ftesting-documentation-deployment\u002F04-e2e-testing-coverage","18.web-development\u002F15.testing-documentation-deployment\u002F04-e2e-testing-coverage",{"title":4770,"path":4771,"stem":4772},"Swagger, OpenAPI, Scalar та Arazzo","\u002Fweb-development\u002Ftesting-documentation-deployment\u002F05-swagger-openapi","18.web-development\u002F15.testing-documentation-deployment\u002F05-swagger-openapi",{"title":4774,"path":4775,"stem":4776},"Керування конфігурацією","\u002Fweb-development\u002Ftesting-documentation-deployment\u002F06-configuration-management","18.web-development\u002F15.testing-documentation-deployment\u002F06-configuration-management",{"title":4778,"path":4779,"stem":4780},"Docker та контейнеризація","\u002Fweb-development\u002Ftesting-documentation-deployment\u002F07-docker-containerization","18.web-development\u002F15.testing-documentation-deployment\u002F07-docker-containerization",{"title":4782,"path":4783,"stem":4784},"CI\u002FCD Pipelines","\u002Fweb-development\u002Ftesting-documentation-deployment\u002F08-cicd-pipelines","18.web-development\u002F15.testing-documentation-deployment\u002F08-cicd-pipelines",{"title":4786,"path":4787,"stem":4788},"Логування та моніторинг","\u002Fweb-development\u002Ftesting-documentation-deployment\u002F09-logging-monitoring","18.web-development\u002F15.testing-documentation-deployment\u002F09-logging-monitoring",{"title":4790,"path":4791,"stem":4792},"Серіалізація та компресія","\u002Fweb-development\u002Ftesting-documentation-deployment\u002F10-serialization-compression","18.web-development\u002F15.testing-documentation-deployment\u002F10-serialization-compression",{"title":4794,"icon":4795,"path":4796,"stem":4797,"children":4798,"page":59},"AI Core: мислення AI","i-lucide-brain","\u002Fai-core","19.ai-core",[4799,4840,4889,4927],{"title":4800,"icon":4749,"path":4801,"stem":4802,"children":4803,"page":59},"Вхід у професію AI Native Developer","\u002Fai-core\u002Fai-native-developer","19.ai-core\u002F01.ai-native-developer",[4804,4808,4812,4816,4820,4824,4828,4832,4836],{"title":4805,"path":4806,"stem":4807},"Вступ до професії розробника цифрових продуктів","\u002Fai-core\u002Fai-native-developer\u002Fdigital-products","19.ai-core\u002F01.ai-native-developer\u002F01.digital-products",{"title":4809,"path":4810,"stem":4811},"Професія розробника програмного забезпечення","\u002Fai-core\u002Fai-native-developer\u002Fsoftware-developer","19.ai-core\u002F01.ai-native-developer\u002F02.software-developer",{"title":4813,"path":4814,"stem":4815},"Професія AI Native Developer","\u002Fai-core\u002Fai-native-developer\u002Fai-native-developer","19.ai-core\u002F01.ai-native-developer\u002F03.ai-native-developer",{"title":4817,"path":4818,"stem":4819},"Вплив AI на розробку програмного забезпечення","\u002Fai-core\u002Fai-native-developer\u002Fai-impact-on-development","19.ai-core\u002F01.ai-native-developer\u002F04.ai-impact-on-development",{"title":4821,"path":4822,"stem":4823},"Життєвий цикл цифрового продукту","\u002Fai-core\u002Fai-native-developer\u002Fproduct-lifecycle","19.ai-core\u002F01.ai-native-developer\u002F05.product-lifecycle",{"title":4825,"path":4826,"stem":4827},"Застосування AI на етапах створення цифрового продукту","\u002Fai-core\u002Fai-native-developer\u002Fai-in-product-creation","19.ai-core\u002F01.ai-native-developer\u002F06.ai-in-product-creation",{"title":4829,"path":4830,"stem":4831},"Розподіл завдань між людиною та AI","\u002Fai-core\u002Fai-native-developer\u002Fhuman-vs-ai-tasks","19.ai-core\u002F01.ai-native-developer\u002F07.human-vs-ai-tasks",{"title":4833,"path":4834,"stem":4835},"Компетенції AI Native Developer","\u002Fai-core\u002Fai-native-developer\u002Fcompetencies","19.ai-core\u002F01.ai-native-developer\u002F08.competencies",{"title":4837,"path":4838,"stem":4839},"Карта професійного розвитку AI Native Developer","\u002Fai-core\u002Fai-native-developer\u002Fcareer-roadmap","19.ai-core\u002F01.ai-native-developer\u002F09.career-roadmap",{"title":4841,"icon":255,"path":4842,"stem":4843,"children":4844,"page":59},"Штучний інтелект і машинне навчання","\u002Fai-core\u002Fai-and-machine-learning","19.ai-core\u002F02.ai-and-machine-learning",[4845,4849,4853,4857,4861,4865,4869,4873,4877,4881,4885],{"title":4846,"path":4847,"stem":4848},"Штучний інтелект","\u002Fai-core\u002Fai-and-machine-learning\u002Fwhat-is-ai","19.ai-core\u002F02.ai-and-machine-learning\u002F01.what-is-ai",{"title":4850,"path":4851,"stem":4852},"Відмінність AI від інших програмних систем","\u002Fai-core\u002Fai-and-machine-learning\u002Fai-vs-automation","19.ai-core\u002F02.ai-and-machine-learning\u002F02.ai-vs-automation",{"title":4854,"path":4855,"stem":4856},"Завдання штучного інтелекту","\u002Fai-core\u002Fai-and-machine-learning\u002Fai-tasks","19.ai-core\u002F02.ai-and-machine-learning\u002F03.ai-tasks",{"title":4858,"path":4859,"stem":4860},"Machine Learning","\u002Fai-core\u002Fai-and-machine-learning\u002Fmachine-learning","19.ai-core\u002F02.ai-and-machine-learning\u002F04.machine-learning",{"title":4862,"path":4863,"stem":4864},"Machine Learning Workflow","\u002Fai-core\u002Fai-and-machine-learning\u002Fml-workflow","19.ai-core\u002F02.ai-and-machine-learning\u002F05.ml-workflow",{"title":4866,"path":4867,"stem":4868},"Типи машинного навчання","\u002Fai-core\u002Fai-and-machine-learning\u002Fml-types","19.ai-core\u002F02.ai-and-machine-learning\u002F06.ml-types",{"title":4870,"path":4871,"stem":4872},"Generative AI","\u002Fai-core\u002Fai-and-machine-learning\u002Fgenerative-ai","19.ai-core\u002F02.ai-and-machine-learning\u002F07.generative-ai",{"title":4874,"path":4875,"stem":4876},"Типи контенту, створюваного AI","\u002Fai-core\u002Fai-and-machine-learning\u002Fai-content-types","19.ai-core\u002F02.ai-and-machine-learning\u002F08.ai-content-types",{"title":4878,"path":4879,"stem":4880},"Великі мовні моделі","\u002Fai-core\u002Fai-and-machine-learning\u002Flarge-language-models","19.ai-core\u002F02.ai-and-machine-learning\u002F09.large-language-models",{"title":4882,"path":4883,"stem":4884},"Можливості та обмеження Generative AI","\u002Fai-core\u002Fai-and-machine-learning\u002Fcapabilities-and-limitations","19.ai-core\u002F02.ai-and-machine-learning\u002F10.capabilities-and-limitations",{"title":4886,"path":4887,"stem":4888},"Сучасні AI-асистенти","\u002Fai-core\u002Fai-and-machine-learning\u002Fmodern-ai-assistants","19.ai-core\u002F02.ai-and-machine-learning\u002F11.modern-ai-assistants",{"title":4890,"icon":4891,"path":4892,"stem":4893,"children":4894,"page":59},"Постановка завдань для AI: принципи створення промптів","i-lucide-message-square-text","\u002Fai-core\u002Fprompt-engineering","19.ai-core\u002F03.prompt-engineering",[4895,4899,4903,4907,4911,4915,4919,4923],{"title":4896,"path":4897,"stem":4898},"Промпт як інструмент","\u002Fai-core\u002Fprompt-engineering\u002Fprompt-as-tool","19.ai-core\u002F03.prompt-engineering\u002F01.prompt-as-tool",{"title":4900,"path":4901,"stem":4902},"Принципи створення промптів","\u002Fai-core\u002Fprompt-engineering\u002Fprompt-principles","19.ai-core\u002F03.prompt-engineering\u002F02.prompt-principles",{"title":4904,"path":4905,"stem":4906},"Структура промпта","\u002Fai-core\u002Fprompt-engineering\u002Fprompt-structure","19.ai-core\u002F03.prompt-engineering\u002F03.prompt-structure",{"title":4908,"path":4909,"stem":4910},"Види промптів за типом задачі","\u002Fai-core\u002Fprompt-engineering\u002Fprompt-types","19.ai-core\u002F03.prompt-engineering\u002F04.prompt-types",{"title":4912,"path":4913,"stem":4914},"Прийоми посилення промпта","\u002Fai-core\u002Fprompt-engineering\u002Fprompt-enhancement","19.ai-core\u002F03.prompt-engineering\u002F05.prompt-enhancement",{"title":4916,"path":4917,"stem":4918},"Розбір і покращення промпта","\u002Fai-core\u002Fprompt-engineering\u002Fprompt-analysis","19.ai-core\u002F03.prompt-engineering\u002F06.prompt-analysis",{"title":4920,"path":4921,"stem":4922},"Типові помилки під час створення промптів","\u002Fai-core\u002Fprompt-engineering\u002Fcommon-mistakes","19.ai-core\u002F03.prompt-engineering\u002F07.common-mistakes",{"title":4924,"path":4925,"stem":4926},"Практикум: особиста бібліотека промптів","\u002Fai-core\u002Fprompt-engineering\u002Fprompt-library","19.ai-core\u002F03.prompt-engineering\u002F08.prompt-library",{"title":4928,"icon":793,"path":4929,"stem":4930,"children":4931,"page":59},"Перевірка результатів, безпека та AI Native Workflow","\u002Fai-core\u002Fverification-security-workflow","19.ai-core\u002F04.verification-security-workflow",[4932,4936,4940,4944,4948,4952,4956,4960,4964],{"title":4933,"path":4934,"stem":4935},"Межі можливостей AI","\u002Fai-core\u002Fverification-security-workflow\u002Fai-limitations","19.ai-core\u002F04.verification-security-workflow\u002F01.ai-limitations",{"title":4937,"path":4938,"stem":4939},"Галюцинації AI","\u002Fai-core\u002Fverification-security-workflow\u002Fai-hallucinations","19.ai-core\u002F04.verification-security-workflow\u002F02.ai-hallucinations",{"title":4941,"path":4942,"stem":4943},"Критичне мислення під час роботи з AI","\u002Fai-core\u002Fverification-security-workflow\u002Fcritical-thinking","19.ai-core\u002F04.verification-security-workflow\u002F03.critical-thinking",{"title":4945,"path":4946,"stem":4947},"Перевірка результатів роботи AI","\u002Fai-core\u002Fverification-security-workflow\u002Fresult-verification","19.ai-core\u002F04.verification-security-workflow\u002F04.result-verification",{"title":4949,"path":4950,"stem":4951},"Захист даних під час роботи з AI","\u002Fai-core\u002Fverification-security-workflow\u002Fdata-protection","19.ai-core\u002F04.verification-security-workflow\u002F05.data-protection",{"title":4953,"path":4954,"stem":4955},"Авторське право та навчальна доброчесність","\u002Fai-core\u002Fverification-security-workflow\u002Fcopyright-integrity","19.ai-core\u002F04.verification-security-workflow\u002F06.copyright-integrity",{"title":4957,"path":4958,"stem":4959},"Відповідальне використання AI","\u002Fai-core\u002Fverification-security-workflow\u002Fresponsible-use","19.ai-core\u002F04.verification-security-workflow\u002F07.responsible-use",{"title":4961,"path":4962,"stem":4963},"AI Native Workflow","\u002Fai-core\u002Fverification-security-workflow\u002Fai-native-workflow","19.ai-core\u002F04.verification-security-workflow\u002F08.ai-native-workflow",{"title":4965,"path":4966,"stem":4967},"Застосування AI Native Workflow під час створення цифрового продукту","\u002Fai-core\u002Fverification-security-workflow\u002Fworkflow-in-practice","19.ai-core\u002F04.verification-security-workflow\u002F09.workflow-in-practice",{"title":4969,"icon":4970,"path":4971,"stem":4972,"children":4973,"page":59},"AI Developer: Vibe Coding Sprint","i-lucide-code-xml","\u002Fai-developer","20.ai-developer",[4974,5000,5022,5043],{"title":4975,"icon":4976,"path":4977,"stem":4978,"children":4979,"page":59},"Проєктування і чорнова версія","i-lucide-pencil-ruler","\u002Fai-developer\u002Fdraft-creation","20.ai-developer\u002F01.draft-creation",[4980,4984,4988,4992,4996],{"title":4981,"path":4982,"stem":4983},"Парадигма Vibe Coding Workflow","\u002Fai-developer\u002Fdraft-creation\u002Fvibe-coding-workflow","20.ai-developer\u002F01.draft-creation\u002F01.vibe-coding-workflow",{"title":4985,"path":4986,"stem":4987},"AI-середовище швидкої розробки","\u002Fai-developer\u002Fdraft-creation\u002Fai-ide-environment","20.ai-developer\u002F01.draft-creation\u002F02.ai-ide-environment",{"title":4989,"path":4990,"stem":4991},"Користувацьке завдання та межі першої версії","\u002Fai-developer\u002Fdraft-creation\u002Fproduct-scope-and-mvp","20.ai-developer\u002F01.draft-creation\u002F03.product-scope-and-mvp",{"title":4993,"path":4994,"stem":4995},"Архітектура інтерфейсу та стани застосунку","\u002Fai-developer\u002Fdraft-creation\u002Fapplication-architecture-and-states","20.ai-developer\u002F01.draft-creation\u002F04.application-architecture-and-states",{"title":4997,"path":4998,"stem":4999},"Генерація та фіксація чорнової версії","\u002Fai-developer\u002Fdraft-creation\u002Fgenerating-draft-version","20.ai-developer\u002F01.draft-creation\u002F05.generating-draft-version",{"title":5001,"icon":5002,"path":5003,"stem":5004,"children":5005,"page":59},"Перша робоча версія","i-lucide-play-circle","\u002Fai-developer\u002Ffirst-working-version","20.ai-developer\u002F02.first-working-version",[5006,5010,5014,5018],{"title":5007,"path":5008,"stem":5009},"Уточнення контексту проєкту для AI","\u002Fai-developer\u002Ffirst-working-version\u002Frefining-project-context","20.ai-developer\u002F02.first-working-version\u002F01.refining-project-context",{"title":5011,"path":5012,"stem":5013},"Стратегія завдань для AI-розробки","\u002Fai-developer\u002Ffirst-working-version\u002Fai-prompting-strategy","20.ai-developer\u002F02.first-working-version\u002F02.ai-prompting-strategy",{"title":5015,"path":5016,"stem":5017},"Доопрацювання до першої робочої версії","\u002Fai-developer\u002Ffirst-working-version\u002Frefining-to-working-version","20.ai-developer\u002F02.first-working-version\u002F03.refining-to-working-version",{"title":5019,"path":5020,"stem":5021},"Перевірка першої робочої версії","\u002Fai-developer\u002Ffirst-working-version\u002Ftesting-first-working-version","20.ai-developer\u002F02.first-working-version\u002F04.testing-first-working-version",{"title":5023,"icon":3072,"path":5024,"stem":5025,"children":5026,"page":59},"Ітерації, дебагінг та відновлення","\u002Fai-developer\u002Fiteration-debugging-recovery","20.ai-developer\u002F03.iteration-debugging-recovery",[5027,5031,5035,5039],{"title":5028,"path":5029,"stem":5030},"Управління точковими змінами продукту","\u002Fai-developer\u002Fiteration-debugging-recovery\u002Fchange-management","20.ai-developer\u002F03.iteration-debugging-recovery\u002F01.change-management",{"title":5032,"path":5033,"stem":5034},"Надійність інтерфейсу та обробка помилок","\u002Fai-developer\u002Fiteration-debugging-recovery\u002Fscenario-reliability-and-errors","20.ai-developer\u002F03.iteration-debugging-recovery\u002F02.scenario-reliability-and-errors",{"title":5036,"path":5037,"stem":5038},"Керування контекстом і збереження версій","\u002Fai-developer\u002Fiteration-debugging-recovery\u002Fproject-context-and-versions","20.ai-developer\u002F03.iteration-debugging-recovery\u002F03.project-context-and-versions",{"title":5040,"path":5041,"stem":5042},"Збої AI-розробки, ліміти та аварійне відновлення","\u002Fai-developer\u002Fiteration-debugging-recovery\u002Fai-breakdowns-and-recovery","20.ai-developer\u002F03.iteration-debugging-recovery\u002F04.ai-breakdowns-and-recovery",{"title":5044,"icon":5045,"path":5046,"stem":5047,"children":5048,"page":59},"Перевірка та завершення продукту","i-lucide-check-check","\u002Fai-developer\u002Fverification-and-completion","20.ai-developer\u002F04.verification-and-completion",[5049,5053,5057],{"title":5050,"path":5051,"stem":5052},"Комплексний аудит та підсумкове тестування","\u002Fai-developer\u002Fverification-and-completion\u002Ffinal-product-testing","20.ai-developer\u002F04.verification-and-completion\u002F01.final-product-testing",{"title":5054,"path":5055,"stem":5056},"Підготовка підсумкової версії коду","\u002Fai-developer\u002Fverification-and-completion\u002Fpreparing-release-version","20.ai-developer\u002F04.verification-and-completion\u002F02.preparing-release-version",{"title":5058,"path":5059,"stem":5060},"Демонстрація, публікація та критерії оцінювання","\u002Fai-developer\u002Fverification-and-completion\u002Fproduct-showcase-and-publishing","20.ai-developer\u002F04.verification-and-completion\u002F03.product-showcase-and-publishing",{"title":5062,"path":5063,"stem":5064,"children":5065,"page":59},"Tailwind","\u002Ftailwind","21.tailwind",[5066,5070,5074,5078,5082,5086,5090,5094,5098,5102,5106,5110],{"title":5067,"path":5068,"stem":5069},"Що таке Tailwind CSS і навіщо він потрібен","\u002Ftailwind\u002Ftailwind-intro-philosophy","21.tailwind\u002F01.tailwind-intro-philosophy",{"title":5071,"path":5072,"stem":5073},"Встановлення та налаштування Tailwind CSS v4","\u002Ftailwind\u002Ftailwind-installation-setup","21.tailwind\u002F02.tailwind-installation-setup",{"title":5075,"path":5076,"stem":5077},"Utility-класи: основи та система Tailwind","\u002Ftailwind\u002Ftailwind-utility-classes-core","21.tailwind\u002F03.tailwind-utility-classes-core",{"title":5079,"path":5080,"stem":5081},"Layout: Flexbox та Grid через Tailwind","\u002Ftailwind\u002Ftailwind-flexbox-grid","21.tailwind\u002F04.tailwind-flexbox-grid",{"title":5083,"path":5084,"stem":5085},"Кастомізація теми через @theme у Tailwind v4","\u002Ftailwind\u002Ftailwind-theme-customization","21.tailwind\u002F05.tailwind-theme-customization",{"title":5087,"path":5088,"stem":5089},"Варіанти: hover, focus, responsive, dark mode та нові v4","\u002Ftailwind\u002Ftailwind-variants-states","21.tailwind\u002F06.tailwind-variants-states",{"title":5091,"path":5092,"stem":5093},"Типографіка та система кольорів у Tailwind v4","\u002Ftailwind\u002Ftailwind-typography-colors","21.tailwind\u002F07.tailwind-typography-colors",{"title":5095,"path":5096,"stem":5097},"Компоненти та повторюваність: @apply, @utility та патерни","\u002Ftailwind\u002Ftailwind-components-patterns","21.tailwind\u002F08.tailwind-components-patterns",{"title":5099,"path":5100,"stem":5101},"Темна тема та система дизайн-токенів у Tailwind v4","\u002Ftailwind\u002Ftailwind-dark-mode-theming","21.tailwind\u002F09.tailwind-dark-mode-theming",{"title":5103,"path":5104,"stem":5105},"Довільні значення та контейнерні запити у Tailwind v4","\u002Ftailwind\u002Ftailwind-arbitrary-container-queries","21.tailwind\u002F10.tailwind-arbitrary-container-queries",{"title":5107,"path":5108,"stem":5109},"Анімації, трансформації та 3D у Tailwind v4","\u002Ftailwind\u002Ftailwind-animations-transforms","21.tailwind\u002F11.tailwind-animations-transforms",{"title":5111,"path":5112,"stem":5113},"Tailwind CLI, PostCSS та інтеграція з фреймворками","\u002Ftailwind\u002Ftailwind-cli-tooling","21.tailwind\u002F12.tailwind-cli-tooling",{"title":5115,"path":5116,"stem":5117,"children":5118,"page":59},"React Typescript","\u002Freact-typescript","22.react-typescript",[5119,5144,5148,5152,5156,5160,5164,5168,5172],{"title":5120,"path":5121,"stem":5122,"children":5123,"page":59},"From Html To Project","\u002Freact-typescript\u002Ffrom-html-to-project","22.react-typescript\u002F01.from-html-to-project",[5124,5128,5132,5136,5140],{"title":5125,"path":5126,"stem":5127},"Від одного HTML-файлу до повноцінного проєкту","\u002Freact-typescript\u002Ffrom-html-to-project\u002Fevolution","22.react-typescript\u002F01.from-html-to-project\u002F01.evolution",{"title":5129,"path":5130,"stem":5131},"Що таке React","\u002Freact-typescript\u002Ffrom-html-to-project\u002Fwhat-is-react","22.react-typescript\u002F01.from-html-to-project\u002F02.what-is-react",{"title":5133,"path":5134,"stem":5135},"JSX та TSX — синтаксичний цукор, а не магія","\u002Freact-typescript\u002Ffrom-html-to-project\u002Fjsx-tsx-syntax","22.react-typescript\u002F01.from-html-to-project\u002F03.jsx-tsx-syntax",{"title":5137,"path":5138,"stem":5139},"Компоненти — будівельні блоки інтерфейсу","\u002Freact-typescript\u002Ffrom-html-to-project\u002Fcomponents","22.react-typescript\u002F01.from-html-to-project\u002F04.components",{"title":5141,"path":5142,"stem":5143},"JSX — розмітка всередині TypeScript","\u002Freact-typescript\u002Ffrom-html-to-project\u002Fjsx-expressions","22.react-typescript\u002F01.from-html-to-project\u002F05.jsx-expressions",{"title":5145,"path":5146,"stem":5147},"Хук useReducer: структуроване управління складним станом","\u002Freact-typescript\u002Fuse-reducer","22.react-typescript\u002F23.use-reducer",{"title":5149,"path":5150,"stem":5151},"Context API: передавання даних без «пропс-дрілінгу»","\u002Freact-typescript\u002Fcontext-api","22.react-typescript\u002F24.context-api",{"title":5153,"path":5154,"stem":5155},"Синхронізація з ефектами: хук useEffect та функції очищення","\u002Freact-typescript\u002Fuse-effect","22.react-typescript\u002F27.use-effect",{"title":5157,"path":5158,"stem":5159},"Можливо, вам не потрібен ефект: антипатерни та альтернативи","\u002Freact-typescript\u002Fyou-might-not-need-effect","22.react-typescript\u002F28.you-might-not-need-effect",{"title":5161,"path":5162,"stem":5163},"Життєвий цикл реактивних ефектів: мислення процесами синхронізації","\u002Freact-typescript\u002Flifecycle-of-reactive-effects","22.react-typescript\u002F29.lifecycle-of-reactive-effects",{"title":5165,"path":5166,"stem":5167},"Розділення подій та ефектів: реактивна проти нереактивної логіки","\u002Freact-typescript\u002Fseparating-events-from-effects","22.react-typescript\u002F30.separating-events-from-effects",{"title":5169,"path":5170,"stem":5171},"Видалення зайвих залежностей ефектів та оптимізація синхронізації","\u002Freact-typescript\u002Fremoving-effect-dependencies","22.react-typescript\u002F31.removing-effect-dependencies",{"title":5173,"path":5174,"stem":5175},"Кастомні хуки: повторне використання та композиція реактивної логіки","\u002Freact-typescript\u002Fcustom-hooks","22.react-typescript\u002F32.custom-hooks",{"title":5177,"path":5178,"stem":5179},"Тестування компонентів діаграм та пісочниць","\u002Ftest-components","98.test-components",{"id":5181,"title":3972,"body":5182,"description":17571,"extension":17572,"links":17573,"meta":17574,"navigation":5391,"path":3973,"seo":17575,"stem":3974,"__hash__":17576},"docs\u002F16.ai-python\u002F16.batching-normalization-deep-networks\u002F05.deep-networks-architectures.md",{"type":5183,"value":5184,"toc":17511},"minimark",[5185,5190,5199,5202,5276,5279,5284,5287,5311,5330,5346,5666,5670,5673,5735,5771,5854,5857,5861,5877,5881,5886,5889,5892,5907,5916,5918,5923,5930,5933,5936,5961,5971,6189,6193,6246,6257,6260,6279,6281,6292,6294,6298,6301,6305,6308,7350,7378,7382,7389,7398,8180,8182,8248,8250,8254,8257,8261,8548,8552,10132,10134,10174,10176,10180,10183,10187,10193,10223,10227,10232,10269,10274,10292,10302,10306,10311,10394,10410,10414,10419,10548,10553,10581,10585,10590,10593,10802,10829,10833,10838,10841,11144,11146,11150,11153,11173,11177,11779,11788,13353,13355,13359,14449,14451,14455,15798,15800,15833,15835,15839,15842,15846,16108,16112,16161,16163,16167,16170,16174,16596,16683,16702,16704,16708,16848,16850,16877,16879,16883,17221,17223,17227,17230,17313,17317,17369,17383,17385,17389,17507],[5186,5187,5189],"h2",{"id":5188},"парадокс-глибини-коли-більше-шарів-не-означає-кращий-результат","Парадокс глибини — коли більше шарів не означає кращий результат",[5191,5192,5193,5194,5198],"p",{},"У попередніх статтях цього розділу ми послідовно опанували ключові техніки навчання глибоких нейронних мереж: Mini-batch Gradient Descent для ефективної обробки даних, правильну ініціалізацію ваг (He\u002FXavier), Batch Normalization для стабілізації навчання та Dropout для боротьби з перенавчанням. Озброївшись цими інструментами, здається логічним припущення: ",[5195,5196,5197],"strong",{},"чим глибша мережа, тим краща її точність",".",[5191,5200,5201],{},"Адже більше шарів означає більше параметрів, більшу експресивність моделі, здатність вивчати більш складні ієрархічні представлення даних. Перші шари виявляють прості ознаки (краї, кольори), середні — складніші текстури та форми, глибокі — високорівневі об'єкти та сцени. Це інтуїтивно зрозумілий принцип, який підтверджується еволюцією архітектур:",[5203,5204,5205,5225],"table",{},[5206,5207,5208],"thead",{},[5209,5210,5211,5216,5219,5222],"tr",{},[5212,5213,5215],"th",{"align":5214},"left","Рік",[5212,5217,5218],{"align":5214},"Архітектура",[5212,5220,5221],{"align":5214},"Кількість шарів",[5212,5223,5224],{"align":5214},"Top-5 Error (ImageNet)",[5226,5227,5228,5245,5261],"tbody",{},[5209,5229,5230,5234,5239,5242],{},[5231,5232,5233],"td",{"align":5214},"2012",[5231,5235,5236],{"align":5214},[5195,5237,5238],{},"AlexNet",[5231,5240,5241],{"align":5214},"8",[5231,5243,5244],{"align":5214},"16.4%",[5209,5246,5247,5250,5255,5258],{},[5231,5248,5249],{"align":5214},"2014",[5231,5251,5252],{"align":5214},[5195,5253,5254],{},"VGG-19",[5231,5256,5257],{"align":5214},"19",[5231,5259,5260],{"align":5214},"7.3%",[5209,5262,5263,5265,5270,5273],{},[5231,5264,5249],{"align":5214},[5231,5266,5267],{"align":5214},[5195,5268,5269],{},"GoogLeNet (Inception v1)",[5231,5271,5272],{"align":5214},"22",[5231,5274,5275],{"align":5214},"6.7%",[5191,5277,5278],{},"Все йде за логікою: глибші мережі досягають кращих результатів. Тоді чому б не побудувати 50-шарову? 100-шарову? 200-шарову?",[5280,5281,5283],"h3",{"id":5282},"несподівана-проблема-degradation","Несподівана проблема: degradation",[5191,5285,5286],{},"У 2015 році дослідники з Microsoft Research Asia (Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun) провели серію експериментів, результати яких вразили наукове співтовариство. Вони тренували дві мережі на датасеті CIFAR-10:",[5288,5289,5290,5295],"card-group",{},[5291,5292,5294],"card",{"icon":943,"title":5293},"🔷 Мережа A: 20 шарів","Стандартна CNN архітектура з послідовними Conv → BatchNorm → ReLU блоками. Навчання стабільне, train accuracy досягає 92%, test accuracy — 89%.",[5291,5296,5299,5302],{"icon":5297,"title":5298},"i-lucide-layers-3","🔶 Мережа B: 56 шарів",[5191,5300,5301],{},"Та сама архітектура, лише глибша (додаткові конволюційні шари). Більше параметрів, більша експресивність. Очікування: краща точність.",[5191,5303,5304,5307,5308],{},[5195,5305,5306],{},"Реальність:"," Train accuracy лише 78%, test accuracy — 75%. ",[5195,5309,5310],{},"Гірше за 20-шарову!",[5312,5313,5314,5319],"warning",{},[5191,5315,5316],{},[5195,5317,5318],{},"Це не перенавчання!",[5191,5320,5321,5322,5325,5326,5329],{},"Спершу дослідники припустили, що 56-шарова мережа перенавчається. Але аналіз показав зворотне: ",[5195,5323,5324],{},"навіть train accuracy нижча",", ніж у 20-шарової. Якби це було overfitting, train accuracy мала б бути високою. Проблема інша — мережа ",[5195,5327,5328],{},"не може навчитися"," навіть тренувальним даним.",[5191,5331,5332,5333,5336,5337,5341,5342,5345],{},"Цей феномен назвали ",[5195,5334,5335],{},"degradation problem"," (",[5338,5339,5340],"em",{},"проблема деградації","): при збільшенні глибини мережі понад певний поріг (зазвичай 20-30 шарів без спеціальних технік) якість навчання ",[5195,5343,5344],{},"погіршується",", навіть на тренувальних даних.",[5347,5348,5350],"plant-uml",{"alt":5349},"Degradation Problem — емпіричне спостереження",[5351,5352,5357],"pre",{"className":5353,"code":5354,"language":5355,"meta":5356,"style":5356},"language-plantuml shiki shiki-themes light-plus dark-plus dark-plus","@startuml\nskinparam style plain\nskinparam backgroundColor #FFFFFF\nskinparam defaultFontName \"Arial\"\n\n!define GOOD_COLOR #DCFCE7\n!define BAD_COLOR #FEE2E2\n!define TITLE_COLOR #DBEAFE\n\npackage \"Експеримент CIFAR-10\" as exp #TITLE_COLOR {\n  \n  rectangle \"20-шарова CNN\" as net20 #GOOD_COLOR {\n    rectangle \"Conv Block 1-5\" as block1\n    rectangle \"Conv Block 6-10\" as block2\n    rectangle \"Conv Block 11-15\" as block3\n    rectangle \"Conv Block 16-20\" as block4\n    block1 -down-> block2\n    block2 -down-> block3\n    block3 -down-> block4\n  }\n  \n  rectangle \"56-шарова CNN\\n(просто глибша)\" as net56 #BAD_COLOR {\n    rectangle \"Conv Block 1-14\" as block1_56\n    rectangle \"Conv Block 15-28\" as block2_56\n    rectangle \"Conv Block 29-42\" as block3_56\n    rectangle \"Conv Block 43-56\" as block4_56\n    block1_56 -down-> block2_56\n    block2_56 -down-> block3_56\n    block3_56 -down-> block4_56\n  }\n}\n\nnote right of net20\n  ✅ Train Accuracy: 92%\n  ✅ Test Accuracy: 89%\n  🎯 Стабільне навчання\nend note\n\nnote right of net56\n  ❌ Train Accuracy: 78%\n  ❌ Test Accuracy: 75%\n  ⚠️ Gradient vanishing\n  ⚠️ Повільна конвергенція\nend note\n\nnote bottom of exp\n  📊 Очікування: 56 шарів > 20 шарів\n  🔬 Реальність: 56 шарів \u003C 20 шарів (!)\n  ❓ Проблема НЕ в overfitting\nend note\n\n@enduml\n","plantuml","",[5358,5359,5360,5368,5374,5380,5386,5393,5399,5405,5411,5416,5422,5428,5434,5440,5446,5452,5458,5464,5470,5476,5482,5487,5493,5499,5505,5511,5517,5523,5529,5535,5540,5546,5551,5557,5563,5569,5575,5581,5586,5592,5598,5604,5610,5616,5621,5626,5632,5638,5644,5650,5655,5660],"code",{"__ignoreMap":5356},[5361,5362,5365],"span",{"class":5363,"line":5364},"line",1,[5361,5366,5367],{},"@startuml\n",[5361,5369,5371],{"class":5363,"line":5370},2,[5361,5372,5373],{},"skinparam style plain\n",[5361,5375,5377],{"class":5363,"line":5376},3,[5361,5378,5379],{},"skinparam backgroundColor #FFFFFF\n",[5361,5381,5383],{"class":5363,"line":5382},4,[5361,5384,5385],{},"skinparam defaultFontName \"Arial\"\n",[5361,5387,5389],{"class":5363,"line":5388},5,[5361,5390,5392],{"emptyLinePlaceholder":5391},true,"\n",[5361,5394,5396],{"class":5363,"line":5395},6,[5361,5397,5398],{},"!define GOOD_COLOR #DCFCE7\n",[5361,5400,5402],{"class":5363,"line":5401},7,[5361,5403,5404],{},"!define BAD_COLOR #FEE2E2\n",[5361,5406,5408],{"class":5363,"line":5407},8,[5361,5409,5410],{},"!define TITLE_COLOR #DBEAFE\n",[5361,5412,5414],{"class":5363,"line":5413},9,[5361,5415,5392],{"emptyLinePlaceholder":5391},[5361,5417,5419],{"class":5363,"line":5418},10,[5361,5420,5421],{},"package \"Експеримент CIFAR-10\" as exp #TITLE_COLOR {\n",[5361,5423,5425],{"class":5363,"line":5424},11,[5361,5426,5427],{},"  \n",[5361,5429,5431],{"class":5363,"line":5430},12,[5361,5432,5433],{},"  rectangle \"20-шарова CNN\" as net20 #GOOD_COLOR {\n",[5361,5435,5437],{"class":5363,"line":5436},13,[5361,5438,5439],{},"    rectangle \"Conv Block 1-5\" as block1\n",[5361,5441,5443],{"class":5363,"line":5442},14,[5361,5444,5445],{},"    rectangle \"Conv Block 6-10\" as block2\n",[5361,5447,5449],{"class":5363,"line":5448},15,[5361,5450,5451],{},"    rectangle \"Conv Block 11-15\" as block3\n",[5361,5453,5455],{"class":5363,"line":5454},16,[5361,5456,5457],{},"    rectangle \"Conv Block 16-20\" as block4\n",[5361,5459,5461],{"class":5363,"line":5460},17,[5361,5462,5463],{},"    block1 -down-> block2\n",[5361,5465,5467],{"class":5363,"line":5466},18,[5361,5468,5469],{},"    block2 -down-> block3\n",[5361,5471,5473],{"class":5363,"line":5472},19,[5361,5474,5475],{},"    block3 -down-> block4\n",[5361,5477,5479],{"class":5363,"line":5478},20,[5361,5480,5481],{},"  }\n",[5361,5483,5485],{"class":5363,"line":5484},21,[5361,5486,5427],{},[5361,5488,5490],{"class":5363,"line":5489},22,[5361,5491,5492],{},"  rectangle \"56-шарова CNN\\n(просто глибша)\" as net56 #BAD_COLOR {\n",[5361,5494,5496],{"class":5363,"line":5495},23,[5361,5497,5498],{},"    rectangle \"Conv Block 1-14\" as block1_56\n",[5361,5500,5502],{"class":5363,"line":5501},24,[5361,5503,5504],{},"    rectangle \"Conv Block 15-28\" as block2_56\n",[5361,5506,5508],{"class":5363,"line":5507},25,[5361,5509,5510],{},"    rectangle \"Conv Block 29-42\" as block3_56\n",[5361,5512,5514],{"class":5363,"line":5513},26,[5361,5515,5516],{},"    rectangle \"Conv Block 43-56\" as block4_56\n",[5361,5518,5520],{"class":5363,"line":5519},27,[5361,5521,5522],{},"    block1_56 -down-> block2_56\n",[5361,5524,5526],{"class":5363,"line":5525},28,[5361,5527,5528],{},"    block2_56 -down-> block3_56\n",[5361,5530,5532],{"class":5363,"line":5531},29,[5361,5533,5534],{},"    block3_56 -down-> block4_56\n",[5361,5536,5538],{"class":5363,"line":5537},30,[5361,5539,5481],{},[5361,5541,5543],{"class":5363,"line":5542},31,[5361,5544,5545],{},"}\n",[5361,5547,5549],{"class":5363,"line":5548},32,[5361,5550,5392],{"emptyLinePlaceholder":5391},[5361,5552,5554],{"class":5363,"line":5553},33,[5361,5555,5556],{},"note right of net20\n",[5361,5558,5560],{"class":5363,"line":5559},34,[5361,5561,5562],{},"  ✅ Train Accuracy: 92%\n",[5361,5564,5566],{"class":5363,"line":5565},35,[5361,5567,5568],{},"  ✅ Test Accuracy: 89%\n",[5361,5570,5572],{"class":5363,"line":5571},36,[5361,5573,5574],{},"  🎯 Стабільне навчання\n",[5361,5576,5578],{"class":5363,"line":5577},37,[5361,5579,5580],{},"end note\n",[5361,5582,5584],{"class":5363,"line":5583},38,[5361,5585,5392],{"emptyLinePlaceholder":5391},[5361,5587,5589],{"class":5363,"line":5588},39,[5361,5590,5591],{},"note right of net56\n",[5361,5593,5595],{"class":5363,"line":5594},40,[5361,5596,5597],{},"  ❌ Train Accuracy: 78%\n",[5361,5599,5601],{"class":5363,"line":5600},41,[5361,5602,5603],{},"  ❌ Test Accuracy: 75%\n",[5361,5605,5607],{"class":5363,"line":5606},42,[5361,5608,5609],{},"  ⚠️ Gradient vanishing\n",[5361,5611,5613],{"class":5363,"line":5612},43,[5361,5614,5615],{},"  ⚠️ Повільна конвергенція\n",[5361,5617,5619],{"class":5363,"line":5618},44,[5361,5620,5580],{},[5361,5622,5624],{"class":5363,"line":5623},45,[5361,5625,5392],{"emptyLinePlaceholder":5391},[5361,5627,5629],{"class":5363,"line":5628},46,[5361,5630,5631],{},"note bottom of exp\n",[5361,5633,5635],{"class":5363,"line":5634},47,[5361,5636,5637],{},"  📊 Очікування: 56 шарів > 20 шарів\n",[5361,5639,5641],{"class":5363,"line":5640},48,[5361,5642,5643],{},"  🔬 Реальність: 56 шарів \u003C 20 шарів (!)\n",[5361,5645,5647],{"class":5363,"line":5646},49,[5361,5648,5649],{},"  ❓ Проблема НЕ в overfitting\n",[5361,5651,5653],{"class":5363,"line":5652},50,[5361,5654,5580],{},[5361,5656,5658],{"class":5363,"line":5657},51,[5361,5659,5392],{"emptyLinePlaceholder":5391},[5361,5661,5663],{"class":5363,"line":5662},52,[5361,5664,5665],{},"@enduml\n",[5280,5667,5669],{"id":5668},"чому-дуже-глибокі-мережі-важко-навчити","Чому дуже глибокі мережі важко навчити?",[5191,5671,5672],{},"Незважаючи на Batch Normalization та правильну ініціалізацію, дуже глибокі мережі стикаються з кількома фундаментальними проблемами:",[5288,5674,5675,5702,5711],{},[5291,5676,5679,5686],{"icon":5677,"title":5678},"i-lucide-waves","🌊 Vanishing Gradients (згасання градієнтів)",[5191,5680,5681,5682,5685],{},"Навіть з ReLU та BatchNorm, у дуже глибоких мережах (50+ шарів) градієнт під час backpropagation ",[5195,5683,5684],{},"експоненційно зменшується"," при проходженні через багато шарів. Перші шари майже не оновлюються.",[5191,5687,5688,5691,5692,5697,5698,5701],{},[5195,5689,5690],{},"Математика:"," Градієнт на шарі ",[5693,5694],"math-formula",{"tex":5695,":inline":5696},"l","true"," пропорційний добутку ",[5693,5699],{"tex":5700,":inline":5696},"\\prod_{i=l}^{L} \\frac{\\partial a^{(i)}}{\\partial a^{(i-1)}}",". Якщо кожна похідна \u003C 1, добуток швидко прямує до нуля.",[5291,5703,5706,5707,5710],{"icon":5704,"title":5705},"i-lucide-mountain","🎯 Optimization Difficulty (складність оптимізації)","Функція втрат глибокої мережі має ",[5195,5708,5709],{},"надзвичайно складний ландшафт"," з безліччю локальних мінімумів, сідлових точок та плато. SGD часто застряє у субоптимальних розв'язках.",[5291,5712,5715,5726],{"icon":5713,"title":5714},"i-lucide-equal","🔗 Identity Mapping Problem (проблема тотожного відображення)",[5191,5716,5717,5718,5721,5722,5725],{},"Теоретично, 56-шарова мережа повинна бути щонайменше ",[5195,5719,5720],{},"не гіршою"," за 20-шарову. Адже вона може навчитися відтворювати 20-шарову (перші 20 шарів виконують корисну роботу, решта 36 — identity mapping ",[5693,5723],{"tex":5724,":inline":5696},"f(x) = x",").",[5191,5727,5728,5731,5732,5198],{},[5195,5729,5730],{},"Проблема:"," SGD важко навчити шари просто копіювати вхід. Це вимагає ідеальної координації ваг: ",[5693,5733],{"tex":5734,":inline":5696},"\\mathbf{W} = \\mathbf{I}, \\mathbf{b} = 0",[5736,5737,5738,5743,5746,5768],"note",{},[5191,5739,5740],{},[5195,5741,5742],{},"Інтуїтивна аналогія:",[5191,5744,5745],{},"Уявіть будівництво хмарочоса. Чим вище будівля, тим складніше:",[5747,5748,5749,5756,5762],"ul",{},[5750,5751,5752,5755],"li",{},[5195,5753,5754],{},"Передавати матеріали на верхні поверхи"," (vanishing gradients)",[5750,5757,5758,5761],{},[5195,5759,5760],{},"Координувати роботу тисяч будівельників"," (optimization difficulty)",[5750,5763,5764,5767],{},[5195,5765,5766],{},"Забезпечити стабільність конструкції"," (identity mapping)",[5191,5769,5770],{},"Без спеціальних технологій (ліфти, кра­ни, каркаси) будівництво високих споруд неможливе. Те саме з глибокими мережами — потрібні спеціальні архітектурні рішення.",[5288,5772,5773,5800],{},[5291,5774,5777],{"icon":5775,"title":5776},"i-lucide-target","🎯 Мета статті",[5747,5778,5779,5782,5785,5788,5791,5794,5797],{},[5750,5780,5781],{},"Зрозуміти причини degradation problem та обмеження класичних CNN",[5750,5783,5784],{},"Опанувати ідею Skip Connections (Residual Learning) — революційне рішення Microsoft Research",[5750,5786,5787],{},"Навчитися реалізовувати Residual Blocks у PyTorch",[5750,5789,5790],{},"Побудувати повноцінну ResNet-подібну архітектуру з нуля",[5750,5792,5793],{},"Дослідити best practices проєктування глибоких мереж",[5750,5795,5796],{},"Провести експеримент: порівняти Plain CNN (20\u002F56 шарів) та ResNet на CIFAR-10",[5750,5798,5799],{},"Отримати огляд сучасних архітектур (EfficientNet, DenseNet, Vision Transformers)",[5291,5801,5803],{"icon":4618,"title":5802},"🔑 Ключові терміни",[5747,5804,5805,5811,5817,5834,5842,5848],{},[5750,5806,5807,5810],{},[5195,5808,5809],{},"Degradation Problem:"," погіршення точності при збільшенні глибини мережі",[5750,5812,5813,5816],{},[5195,5814,5815],{},"Skip Connection (Shortcut):"," пряме з'єднання, що обходить один або кілька шарів",[5750,5818,5819,5822,5823,5826,5827,5830,5831],{},[5195,5820,5821],{},"Residual Learning:"," навчання ",[5195,5824,5825],{},"залишку"," ",[5693,5828],{"tex":5829,":inline":5696},"F(x)"," замість повного відображення ",[5693,5832],{"tex":5833,":inline":5696},"H(x)",[5750,5835,5836,5839,5840],{},[5195,5837,5838],{},"Identity Mapping:"," тотожне відображення ",[5693,5841],{"tex":5724,":inline":5696},[5750,5843,5844,5847],{},[5195,5845,5846],{},"Residual Block:"," базовий будівельний блок ResNet із skip connection",[5750,5849,5850,5853],{},[5195,5851,5852],{},"Bottleneck Design:"," архітектурний патерн для зменшення обчислювальної складності",[5855,5856],"hr",{},[5186,5858,5860],{"id":5859},"ідея-skip-connections-революційне-рішення","Ідея Skip Connections — революційне рішення",[5191,5862,5863,5864,5826,5867,5870,5871,5873,5874,5198],{},"Ключове прозріння Kaiming He та його команди: замість того, щоб змушувати шари навчатися ",[5195,5865,5866],{},"повному відображенню",[5693,5868],{"tex":5869,":inline":5696},"H(\\mathbf{x})",", дозволимо їм навчатися лише ",[5195,5872,5825],{}," (residual) ",[5693,5875],{"tex":5876,":inline":5696},"F(\\mathbf{x}) = H(\\mathbf{x}) - \\mathbf{x}",[5280,5878,5880],{"id":5879},"математична-формалізація","Математична формалізація",[5191,5882,5883],{},[5195,5884,5885],{},"Класична архітектура (Plain Network):",[5191,5887,5888],{},"Кожен шар (або група шарів) навчається відображенню:",[5693,5890],{"tex":5891},"\\mathbf{y} = H(\\mathbf{x})",[5191,5893,5894,5895,5898,5899,5902,5903,5906],{},"Де ",[5693,5896],{"tex":5897,":inline":5696},"\\mathbf{x}"," — вхід, ",[5693,5900],{"tex":5901,":inline":5696},"\\mathbf{y}"," — вихід, ",[5693,5904],{"tex":5905,":inline":5696},"H"," — функція, що реалізується послідовністю Conv → BN → ReLU → Conv.",[5191,5908,5909,5911,5912,5915],{},[5195,5910,5730],{}," Якщо оптимальне відображення близьке до identity ",[5693,5913],{"tex":5914,":inline":5696},"H(\\mathbf{x}) \\approx \\mathbf{x}",", мережі важко його навчитися напряму.",[5855,5917],{},[5191,5919,5920],{},[5195,5921,5922],{},"Residual архітектура (ResNet):",[5191,5924,5925,5926,5929],{},"Додаємо ",[5195,5927,5928],{},"skip connection"," (пряме з'єднання), що обходить шари:",[5693,5931],{"tex":5932},"\\mathbf{y} = F(\\mathbf{x}) + \\mathbf{x}",[5191,5934,5935],{},"Де:",[5747,5937,5938,5948,5956],{},[5750,5939,5940,5943,5944,5947],{},[5693,5941],{"tex":5942,":inline":5696},"F(\\mathbf{x})"," — ",[5195,5945,5946],{},"residual function"," (залишок), що навчається шарами",[5750,5949,5950,5943,5952,5955],{},[5693,5951],{"tex":5897,":inline":5696},[5195,5953,5954],{},"identity shortcut"," (пряме з'єднання)",[5750,5957,5958,5960],{},[5693,5959],{"tex":5901,":inline":5696}," — фінальний вихід",[5191,5962,5963,5966,5967,5970],{},[5195,5964,5965],{},"Ключова ідея:"," Тепер мережа навчається ",[5195,5968,5969],{},"різниці"," між входом та виходом, а не повному відображенню.",[5347,5972,5974],{"alt":5973},"Skip Connection — архітектурна схема",[5351,5975,5977],{"className":5353,"code":5976,"language":5355,"meta":5356,"style":5356},"@startuml\nskinparam style plain\nskinparam backgroundColor #FFFFFF\nskinparam defaultFontName \"Arial\"\n\n!define INPUT_COLOR #DBEAFE\n!define PROCESS_COLOR #FEF3C7\n!define OUTPUT_COLOR #DCFCE7\n\nrectangle \"Вхід x\" as input #INPUT_COLOR\n\nrectangle \"Residual Block\" as block {\n  rectangle \"Conv 3×3, 64 filters\\nBatchNorm\\nReLU\" as conv1 #PROCESS_COLOR\n  rectangle \"Conv 3×3, 64 filters\\nBatchNorm\" as conv2 #PROCESS_COLOR\n  \n  conv1 -down-> conv2 : F(x)\n}\n\nrectangle \"Додавання (+)\" as add #PROCESS_COLOR\nrectangle \"ReLU\" as relu #PROCESS_COLOR\nrectangle \"Вихід y\" as output #OUTPUT_COLOR\n\ninput -down-> conv1 : x\ninput -right-> add : Identity\\nSkip Connection\nconv2 -down-> add : F(x)\nadd -down-> relu : F(x) + x\nrelu -down-> output : y = ReLU(F(x) + x)\n\nnote left of input\n  📥 Розмір: [batch, 64, H, W]\nend note\n\nnote right of add\n  🔑 Ключова операція:\n  y = F(x) + x\n  \n  Якщо F(x) ≈ 0,\n  то y ≈ x (identity)\nend note\n\nnote left of output\n  📤 Той самий розмір:\n  [batch, 64, H, W]\nend note\n\n@enduml\n",[5358,5978,5979,5983,5987,5991,5995,5999,6004,6009,6014,6018,6023,6027,6032,6037,6042,6046,6051,6055,6059,6064,6069,6074,6078,6083,6088,6093,6098,6103,6107,6112,6117,6121,6125,6130,6135,6140,6144,6149,6154,6158,6162,6167,6172,6177,6181,6185],{"__ignoreMap":5356},[5361,5980,5981],{"class":5363,"line":5364},[5361,5982,5367],{},[5361,5984,5985],{"class":5363,"line":5370},[5361,5986,5373],{},[5361,5988,5989],{"class":5363,"line":5376},[5361,5990,5379],{},[5361,5992,5993],{"class":5363,"line":5382},[5361,5994,5385],{},[5361,5996,5997],{"class":5363,"line":5388},[5361,5998,5392],{"emptyLinePlaceholder":5391},[5361,6000,6001],{"class":5363,"line":5395},[5361,6002,6003],{},"!define INPUT_COLOR #DBEAFE\n",[5361,6005,6006],{"class":5363,"line":5401},[5361,6007,6008],{},"!define PROCESS_COLOR #FEF3C7\n",[5361,6010,6011],{"class":5363,"line":5407},[5361,6012,6013],{},"!define OUTPUT_COLOR #DCFCE7\n",[5361,6015,6016],{"class":5363,"line":5413},[5361,6017,5392],{"emptyLinePlaceholder":5391},[5361,6019,6020],{"class":5363,"line":5418},[5361,6021,6022],{},"rectangle \"Вхід x\" as input #INPUT_COLOR\n",[5361,6024,6025],{"class":5363,"line":5424},[5361,6026,5392],{"emptyLinePlaceholder":5391},[5361,6028,6029],{"class":5363,"line":5430},[5361,6030,6031],{},"rectangle \"Residual Block\" as block {\n",[5361,6033,6034],{"class":5363,"line":5436},[5361,6035,6036],{},"  rectangle \"Conv 3×3, 64 filters\\nBatchNorm\\nReLU\" as conv1 #PROCESS_COLOR\n",[5361,6038,6039],{"class":5363,"line":5442},[5361,6040,6041],{},"  rectangle \"Conv 3×3, 64 filters\\nBatchNorm\" as conv2 #PROCESS_COLOR\n",[5361,6043,6044],{"class":5363,"line":5448},[5361,6045,5427],{},[5361,6047,6048],{"class":5363,"line":5454},[5361,6049,6050],{},"  conv1 -down-> conv2 : F(x)\n",[5361,6052,6053],{"class":5363,"line":5460},[5361,6054,5545],{},[5361,6056,6057],{"class":5363,"line":5466},[5361,6058,5392],{"emptyLinePlaceholder":5391},[5361,6060,6061],{"class":5363,"line":5472},[5361,6062,6063],{},"rectangle \"Додавання (+)\" as add #PROCESS_COLOR\n",[5361,6065,6066],{"class":5363,"line":5478},[5361,6067,6068],{},"rectangle \"ReLU\" as relu #PROCESS_COLOR\n",[5361,6070,6071],{"class":5363,"line":5484},[5361,6072,6073],{},"rectangle \"Вихід y\" as output #OUTPUT_COLOR\n",[5361,6075,6076],{"class":5363,"line":5489},[5361,6077,5392],{"emptyLinePlaceholder":5391},[5361,6079,6080],{"class":5363,"line":5495},[5361,6081,6082],{},"input -down-> conv1 : x\n",[5361,6084,6085],{"class":5363,"line":5501},[5361,6086,6087],{},"input -right-> add : Identity\\nSkip Connection\n",[5361,6089,6090],{"class":5363,"line":5507},[5361,6091,6092],{},"conv2 -down-> add : F(x)\n",[5361,6094,6095],{"class":5363,"line":5513},[5361,6096,6097],{},"add -down-> relu : F(x) + x\n",[5361,6099,6100],{"class":5363,"line":5519},[5361,6101,6102],{},"relu -down-> output : y = ReLU(F(x) + x)\n",[5361,6104,6105],{"class":5363,"line":5525},[5361,6106,5392],{"emptyLinePlaceholder":5391},[5361,6108,6109],{"class":5363,"line":5531},[5361,6110,6111],{},"note left of input\n",[5361,6113,6114],{"class":5363,"line":5537},[5361,6115,6116],{},"  📥 Розмір: [batch, 64, H, W]\n",[5361,6118,6119],{"class":5363,"line":5542},[5361,6120,5580],{},[5361,6122,6123],{"class":5363,"line":5548},[5361,6124,5392],{"emptyLinePlaceholder":5391},[5361,6126,6127],{"class":5363,"line":5553},[5361,6128,6129],{},"note right of add\n",[5361,6131,6132],{"class":5363,"line":5559},[5361,6133,6134],{},"  🔑 Ключова операція:\n",[5361,6136,6137],{"class":5363,"line":5565},[5361,6138,6139],{},"  y = F(x) + x\n",[5361,6141,6142],{"class":5363,"line":5571},[5361,6143,5427],{},[5361,6145,6146],{"class":5363,"line":5577},[5361,6147,6148],{},"  Якщо F(x) ≈ 0,\n",[5361,6150,6151],{"class":5363,"line":5583},[5361,6152,6153],{},"  то y ≈ x (identity)\n",[5361,6155,6156],{"class":5363,"line":5588},[5361,6157,5580],{},[5361,6159,6160],{"class":5363,"line":5594},[5361,6161,5392],{"emptyLinePlaceholder":5391},[5361,6163,6164],{"class":5363,"line":5600},[5361,6165,6166],{},"note left of output\n",[5361,6168,6169],{"class":5363,"line":5606},[5361,6170,6171],{},"  📤 Той самий розмір:\n",[5361,6173,6174],{"class":5363,"line":5612},[5361,6175,6176],{},"  [batch, 64, H, W]\n",[5361,6178,6179],{"class":5363,"line":5618},[5361,6180,5580],{},[5361,6182,6183],{"class":5363,"line":5623},[5361,6184,5392],{"emptyLinePlaceholder":5391},[5361,6186,6187],{"class":5363,"line":5628},[5361,6188,5665],{},[5280,6190,6192],{"id":6191},"чому-це-працює-три-ключових-інсайти","Чому це працює? Три ключових інсайти",[5288,6194,6195,6218],{},[5291,6196,6199,6212],{"icon":6197,"title":6198},"i-lucide-check-circle","1️⃣ Легкість навчання identity",[5191,6200,6201,6204,6205,6208,6209,5198],{},[5195,6202,6203],{},"Теза:"," Навчити ",[5693,6206],{"tex":6207,":inline":5696},"F(\\mathbf{x}) = 0"," значно простіше, ніж ",[5693,6210],{"tex":6211,":inline":5696},"H(\\mathbf{x}) = \\mathbf{x}",[5191,6213,6214,6217],{},[5195,6215,6216],{},"Пояснення:"," Якщо оптимальне відображення близьке до identity, residual блоку достатньо просто обнулити ваги (weight decay природно штовхає ваги до нуля). Градієнти легко можуть \"вимкнути\" непотрібні шари.",[5291,6219,6221,6226,6229,6243],{"icon":92,"title":6220},"2️⃣ Прямий шлях для градієнтів",[5191,6222,6223],{},[5195,6224,6225],{},"Проблема vanishing gradients вирішена!",[5191,6227,6228],{},"Під час backpropagation градієнт проходить двома шляхами:",[5747,6230,6231,6237],{},[5750,6232,6233,6234,6236],{},"Через residual function ",[5693,6235],{"tex":5942,":inline":5696}," (може згасати)",[5750,6238,6239,6242],{},[5195,6240,6241],{},"Напряму через skip connection"," (без згасання!)",[5693,6244],{"tex":6245},"\\frac{\\partial \\mathcal{L}}{\\partial \\mathbf{x}} = \\frac{\\partial \\mathcal{L}}{\\partial \\mathbf{y}} \\left( \\frac{\\partial F}{\\partial \\mathbf{x}} + \\mathbf{I} \\right)",[5191,6247,6248,6249,6252,6253,6256],{},"Навіть якщо ",[5693,6250],{"tex":6251,":inline":5696},"\\frac{\\partial F}{\\partial \\mathbf{x}} \\to 0",", залишається ",[5693,6254],{"tex":6255,":inline":5696},"\\mathbf{I}"," (одинична матриця) — градієнт гарантовано досягає початкових шарів.",[5191,6258,6259],{},"::",[5291,6261,6263,6268],{"icon":658,"title":6262},"3️⃣ Ensemble ефект",[5191,6264,6265],{},[5195,6266,6267],{},"Неявний ансамбль:",[5191,6269,6270,6271,6274,6275,6278],{},"ResNet з ",[5693,6272],{"tex":6273,":inline":5696},"N"," residual блоків неявно представляє ",[5693,6276],{"tex":6277,":inline":5696},"2^N"," різних шляхів від входу до виходу (кожен блок можна \"ввімкнути\" або \"обійти\"). Це схоже на тренування ансамблю експоненціально багатьох мереж різної глибини!",[5191,6280,6259],{},[6282,6283,6284,6289],"tip",{},[5191,6285,6286],{},[5195,6287,6288],{},"Аналогія: автомагістраль з об'їзними дорогами",[5191,6290,6291],{},"Уявіть автомагістраль (residual block) із головною дорогою (skip connection) та об'їзними шляхами через міста (conv layers). Якщо в місті немає нічого цікавого (оптимальне відображення ≈ identity), водій їде прямою дорогою. Якщо потрібна зупинка (transformation), використовує об'їзд. Gradient \"автомобілі\" завжди мають прямий шлях назад!",[5855,6293],{},[5186,6295,6297],{"id":6296},"реалізація-residual-block-у-pytorch","Реалізація Residual Block у PyTorch",[5191,6299,6300],{},"Тепер, коли ми розуміємо теорію, перейдемо до практичної реалізації. PyTorch надає всі необхідні інструменти для побудови residual блоків.",[5280,6302,6304],{"id":6303},"базовий-residual-block-basic-block","Базовий Residual Block (Basic Block)",[5191,6306,6307],{},"Найпростіша версія — два конволюційні шари 3×3 з BatchNorm та ReLU:",[6309,6310,6313,6324],"jupyter-notebook",{"kernel":6311,"title":6312},"Python 3.11 (ipykernel)","residual_block_implementation.ipynb",[6314,6315,6317,6321],"jupyter-cell",{"type":6316},"markdown",[5280,6318,6320],{"id":6319},"реалізація-basicblock","Реалізація BasicBlock",[5191,6322,6323],{},"Це базовий будівельний блок ResNet-18\u002F34. Структура: Conv(3×3) → BN → ReLU → Conv(3×3) → BN → (+skip) → ReLU.",[6314,6325,6327,7293],{"execution-count":6326},"1",[5351,6328,6332],{"className":6329,"code":6330,"language":6331,"meta":5356,"style":5356},"language-python shiki shiki-themes light-plus dark-plus dark-plus","import torch\nimport torch.nn as nn\nimport torch.nn.functional as F\n\nclass BasicBlock(nn.Module):\n    \"\"\"\n    Базовий Residual Block для ResNet-18\u002F34\n    \n    Структура:\n        x → [Conv 3×3 → BN → ReLU → Conv 3×3 → BN] → (+) → ReLU → output\n            └────────────────────────────────────────────┘\n                        skip connection\n    \"\"\"\n    \n    def __init__(self, in_channels, out_channels, stride=1, downsample=None):\n        \"\"\"\n        Args:\n            in_channels: кількість вхідних каналів\n            out_channels: кількість вихідних каналів\n            stride: крок згортки (1 для збереження розміру, 2 для зменшення H×W)\n            downsample: додатковий шар для узгодження розмірів skip connection\n        \"\"\"\n        super(BasicBlock, self).__init__()\n        \n        # Перший конволюційний шар (може зменшити розмір)\n        self.conv1 = nn.Conv2d(\n            in_channels, \n            out_channels, \n            kernel_size=3, \n            stride=stride,  # stride=2 зменшує H×W удвічі\n            padding=1, \n            bias=False  # Batch Norm робить bias непотрібним\n        )\n        self.bn1 = nn.BatchNorm2d(out_channels)\n        \n        # Другий конволюційний шар (завжди stride=1)\n        self.conv2 = nn.Conv2d(\n            out_channels, \n            out_channels, \n            kernel_size=3, \n            stride=1, \n            padding=1, \n            bias=False\n        )\n        self.bn2 = nn.BatchNorm2d(out_channels)\n        \n        # Skip connection з узгодженням розмірів (якщо потрібно)\n        self.downsample = downsample\n        self.stride = stride\n    \n    def forward(self, x):\n        # Зберігаємо оригінальний вхід для skip connection\n        identity = x\n        \n        # Основний шлях (residual function F(x))\n        out = self.conv1(x)\n        out = self.bn1(out)\n        out = F.relu(out)\n        \n        out = self.conv2(out)\n        out = self.bn2(out)\n        \n        # Якщо розміри не співпадають, застосовуємо downsample до identity\n        if self.downsample is not None:\n            identity = self.downsample(x)\n        \n        # Skip connection: y = F(x) + x\n        out += identity\n        \n        # Фінальна активація\n        out = F.relu(out)\n        \n        return out\n\n\n# Тестування BasicBlock\nprint(\"🧪 Тестування BasicBlock\\n\")\n\n# Випадок 1: розміри співпадають (skip connection без трансформації)\nblock1 = BasicBlock(in_channels=64, out_channels=64, stride=1)\nx1 = torch.randn(4, 64, 32, 32)  # [batch, channels, H, W]\ny1 = block1(x1)\nprint(f\"📥 Вхід: {x1.shape}\")\nprint(f\"📤 Вихід: {y1.shape}\")\nprint(f\"✅ Розміри співпадають — skip connection без трансформації\\n\")\n\n# Випадок 2: зміна кількості каналів та downsampling\ndownsample = nn.Sequential(\n    nn.Conv2d(64, 128, kernel_size=1, stride=2, bias=False),\n    nn.BatchNorm2d(128)\n)\nblock2 = BasicBlock(in_channels=64, out_channels=128, stride=2, downsample=downsample)\nx2 = torch.randn(4, 64, 32, 32)\ny2 = block2(x2)\nprint(f\"📥 Вхід: {x2.shape}\")\nprint(f\"📤 Вихід: {y2.shape}\")\nprint(f\"🔽 Downsampling: H×W зменшено вдвічі, канали подвоєно\")\n","python",[5358,6333,6334,6344,6357,6369,6373,6397,6403,6408,6413,6418,6423,6428,6433,6437,6441,6490,6495,6500,6505,6510,6515,6520,6524,6542,6547,6553,6561,6566,6571,6584,6595,6606,6619,6624,6631,6635,6640,6647,6651,6655,6665,6675,6685,6694,6698,6705,6709,6714,6721,6728,6732,6750,6755,6761,6766,6772,6783,6793,6799,6804,6814,6824,6829,6835,6859,6870,6875,6881,6887,6892,6898,6903,6908,6917,6922,6927,6933,6954,6959,6965,6996,7024,7030,7056,7079,7097,7102,7108,7114,7157,7167,7172,7207,7229,7235,7257,7279],{"__ignoreMap":5356},[5361,6335,6336,6340],{"class":5363,"line":5364},[5361,6337,6339],{"class":6338},"s8xlr","import",[5361,6341,6343],{"class":6342},"sHH4Y"," torch\n",[5361,6345,6346,6348,6351,6354],{"class":5363,"line":5370},[5361,6347,6339],{"class":6338},[5361,6349,6350],{"class":6342}," torch.nn ",[5361,6352,6353],{"class":6338},"as",[5361,6355,6356],{"class":6342}," nn\n",[5361,6358,6359,6361,6364,6366],{"class":5363,"line":5376},[5361,6360,6339],{"class":6338},[5361,6362,6363],{"class":6342}," torch.nn.functional ",[5361,6365,6353],{"class":6338},[5361,6367,6368],{"class":6342}," F\n",[5361,6370,6371],{"class":5363,"line":5382},[5361,6372,5392],{"emptyLinePlaceholder":5391},[5361,6374,6375,6379,6383,6386,6389,6391,6394],{"class":5363,"line":5388},[5361,6376,6378],{"class":6377},"su1O8","class",[5361,6380,6382],{"class":6381},"sN1BT"," BasicBlock",[5361,6384,6385],{"class":6342},"(",[5361,6387,6388],{"class":6381},"nn",[5361,6390,5198],{"class":6342},[5361,6392,6393],{"class":6381},"Module",[5361,6395,6396],{"class":6342},"):\n",[5361,6398,6399],{"class":5363,"line":5395},[5361,6400,6402],{"class":6401},"sbdoH","    \"\"\"\n",[5361,6404,6405],{"class":5363,"line":5401},[5361,6406,6407],{"class":6401},"    Базовий Residual Block для ResNet-18\u002F34\n",[5361,6409,6410],{"class":5363,"line":5407},[5361,6411,6412],{"class":6401},"    \n",[5361,6414,6415],{"class":5363,"line":5413},[5361,6416,6417],{"class":6401},"    Структура:\n",[5361,6419,6420],{"class":5363,"line":5418},[5361,6421,6422],{"class":6401},"        x → [Conv 3×3 → BN → ReLU → Conv 3×3 → BN] → (+) → ReLU → output\n",[5361,6424,6425],{"class":5363,"line":5424},[5361,6426,6427],{"class":6401},"            └────────────────────────────────────────────┘\n",[5361,6429,6430],{"class":5363,"line":5430},[5361,6431,6432],{"class":6401},"                        skip connection\n",[5361,6434,6435],{"class":5363,"line":5436},[5361,6436,6402],{"class":6401},[5361,6438,6439],{"class":5363,"line":5442},[5361,6440,6412],{"class":6342},[5361,6442,6443,6446,6450,6452,6456,6459,6462,6464,6467,6469,6472,6475,6478,6480,6483,6485,6488],{"class":5363,"line":5448},[5361,6444,6445],{"class":6377},"    def",[5361,6447,6449],{"class":6448},"s8Opu"," __init__",[5361,6451,6385],{"class":6342},[5361,6453,6455],{"class":6454},"siwwj","self",[5361,6457,6458],{"class":6342},", ",[5361,6460,6461],{"class":6454},"in_channels",[5361,6463,6458],{"class":6342},[5361,6465,6466],{"class":6454},"out_channels",[5361,6468,6458],{"class":6342},[5361,6470,6471],{"class":6454},"stride",[5361,6473,6474],{"class":6342},"=",[5361,6476,6326],{"class":6477},"sJj4R",[5361,6479,6458],{"class":6342},[5361,6481,6482],{"class":6454},"downsample",[5361,6484,6474],{"class":6342},[5361,6486,6487],{"class":6377},"None",[5361,6489,6396],{"class":6342},[5361,6491,6492],{"class":5363,"line":5454},[5361,6493,6494],{"class":6401},"        \"\"\"\n",[5361,6496,6497],{"class":5363,"line":5460},[5361,6498,6499],{"class":6401},"        Args:\n",[5361,6501,6502],{"class":5363,"line":5466},[5361,6503,6504],{"class":6401},"            in_channels: кількість вхідних каналів\n",[5361,6506,6507],{"class":5363,"line":5472},[5361,6508,6509],{"class":6401},"            out_channels: кількість вихідних каналів\n",[5361,6511,6512],{"class":5363,"line":5478},[5361,6513,6514],{"class":6401},"            stride: крок згортки (1 для збереження розміру, 2 для зменшення H×W)\n",[5361,6516,6517],{"class":5363,"line":5484},[5361,6518,6519],{"class":6401},"            downsample: додатковий шар для узгодження розмірів skip connection\n",[5361,6521,6522],{"class":5363,"line":5489},[5361,6523,6494],{"class":6401},[5361,6525,6526,6529,6532,6534,6536,6539],{"class":5363,"line":5495},[5361,6527,6528],{"class":6381},"        super",[5361,6530,6531],{"class":6342},"(BasicBlock, ",[5361,6533,6455],{"class":6377},[5361,6535,5725],{"class":6342},[5361,6537,6538],{"class":6448},"__init__",[5361,6540,6541],{"class":6342},"()\n",[5361,6543,6544],{"class":5363,"line":5501},[5361,6545,6546],{"class":6342},"        \n",[5361,6548,6549],{"class":5363,"line":5507},[5361,6550,6552],{"class":6551},"spJ8K","        # Перший конволюційний шар (може зменшити розмір)\n",[5361,6554,6555,6558],{"class":5363,"line":5513},[5361,6556,6557],{"class":6377},"        self",[5361,6559,6560],{"class":6342},".conv1 = nn.Conv2d(\n",[5361,6562,6563],{"class":5363,"line":5519},[5361,6564,6565],{"class":6342},"            in_channels, \n",[5361,6567,6568],{"class":5363,"line":5525},[5361,6569,6570],{"class":6342},"            out_channels, \n",[5361,6572,6573,6576,6578,6581],{"class":5363,"line":5531},[5361,6574,6575],{"class":6454},"            kernel_size",[5361,6577,6474],{"class":6342},[5361,6579,6580],{"class":6477},"3",[5361,6582,6583],{"class":6342},", \n",[5361,6585,6586,6589,6592],{"class":5363,"line":5537},[5361,6587,6588],{"class":6454},"            stride",[5361,6590,6591],{"class":6342},"=stride,  ",[5361,6593,6594],{"class":6551},"# stride=2 зменшує H×W удвічі\n",[5361,6596,6597,6600,6602,6604],{"class":5363,"line":5542},[5361,6598,6599],{"class":6454},"            padding",[5361,6601,6474],{"class":6342},[5361,6603,6326],{"class":6477},[5361,6605,6583],{"class":6342},[5361,6607,6608,6611,6613,6616],{"class":5363,"line":5548},[5361,6609,6610],{"class":6454},"            bias",[5361,6612,6474],{"class":6342},[5361,6614,6615],{"class":6377},"False",[5361,6617,6618],{"class":6551},"  # Batch Norm робить bias непотрібним\n",[5361,6620,6621],{"class":5363,"line":5553},[5361,6622,6623],{"class":6342},"        )\n",[5361,6625,6626,6628],{"class":5363,"line":5559},[5361,6627,6557],{"class":6377},[5361,6629,6630],{"class":6342},".bn1 = nn.BatchNorm2d(out_channels)\n",[5361,6632,6633],{"class":5363,"line":5565},[5361,6634,6546],{"class":6342},[5361,6636,6637],{"class":5363,"line":5571},[5361,6638,6639],{"class":6551},"        # Другий конволюційний шар (завжди stride=1)\n",[5361,6641,6642,6644],{"class":5363,"line":5577},[5361,6643,6557],{"class":6377},[5361,6645,6646],{"class":6342},".conv2 = nn.Conv2d(\n",[5361,6648,6649],{"class":5363,"line":5583},[5361,6650,6570],{"class":6342},[5361,6652,6653],{"class":5363,"line":5588},[5361,6654,6570],{"class":6342},[5361,6656,6657,6659,6661,6663],{"class":5363,"line":5594},[5361,6658,6575],{"class":6454},[5361,6660,6474],{"class":6342},[5361,6662,6580],{"class":6477},[5361,6664,6583],{"class":6342},[5361,6666,6667,6669,6671,6673],{"class":5363,"line":5600},[5361,6668,6588],{"class":6454},[5361,6670,6474],{"class":6342},[5361,6672,6326],{"class":6477},[5361,6674,6583],{"class":6342},[5361,6676,6677,6679,6681,6683],{"class":5363,"line":5606},[5361,6678,6599],{"class":6454},[5361,6680,6474],{"class":6342},[5361,6682,6326],{"class":6477},[5361,6684,6583],{"class":6342},[5361,6686,6687,6689,6691],{"class":5363,"line":5612},[5361,6688,6610],{"class":6454},[5361,6690,6474],{"class":6342},[5361,6692,6693],{"class":6377},"False\n",[5361,6695,6696],{"class":5363,"line":5618},[5361,6697,6623],{"class":6342},[5361,6699,6700,6702],{"class":5363,"line":5623},[5361,6701,6557],{"class":6377},[5361,6703,6704],{"class":6342},".bn2 = nn.BatchNorm2d(out_channels)\n",[5361,6706,6707],{"class":5363,"line":5628},[5361,6708,6546],{"class":6342},[5361,6710,6711],{"class":5363,"line":5634},[5361,6712,6713],{"class":6551},"        # Skip connection з узгодженням розмірів (якщо потрібно)\n",[5361,6715,6716,6718],{"class":5363,"line":5640},[5361,6717,6557],{"class":6377},[5361,6719,6720],{"class":6342},".downsample = downsample\n",[5361,6722,6723,6725],{"class":5363,"line":5646},[5361,6724,6557],{"class":6377},[5361,6726,6727],{"class":6342},".stride = stride\n",[5361,6729,6730],{"class":5363,"line":5652},[5361,6731,6412],{"class":6342},[5361,6733,6734,6736,6739,6741,6743,6745,6748],{"class":5363,"line":5657},[5361,6735,6445],{"class":6377},[5361,6737,6738],{"class":6448}," forward",[5361,6740,6385],{"class":6342},[5361,6742,6455],{"class":6454},[5361,6744,6458],{"class":6342},[5361,6746,6747],{"class":6454},"x",[5361,6749,6396],{"class":6342},[5361,6751,6752],{"class":5363,"line":5662},[5361,6753,6754],{"class":6551},"        # Зберігаємо оригінальний вхід для skip connection\n",[5361,6756,6758],{"class":5363,"line":6757},53,[5361,6759,6760],{"class":6342},"        identity = x\n",[5361,6762,6764],{"class":5363,"line":6763},54,[5361,6765,6546],{"class":6342},[5361,6767,6769],{"class":5363,"line":6768},55,[5361,6770,6771],{"class":6551},"        # Основний шлях (residual function F(x))\n",[5361,6773,6775,6778,6780],{"class":5363,"line":6774},56,[5361,6776,6777],{"class":6342},"        out = ",[5361,6779,6455],{"class":6377},[5361,6781,6782],{"class":6342},".conv1(x)\n",[5361,6784,6786,6788,6790],{"class":5363,"line":6785},57,[5361,6787,6777],{"class":6342},[5361,6789,6455],{"class":6377},[5361,6791,6792],{"class":6342},".bn1(out)\n",[5361,6794,6796],{"class":5363,"line":6795},58,[5361,6797,6798],{"class":6342},"        out = F.relu(out)\n",[5361,6800,6802],{"class":5363,"line":6801},59,[5361,6803,6546],{"class":6342},[5361,6805,6807,6809,6811],{"class":5363,"line":6806},60,[5361,6808,6777],{"class":6342},[5361,6810,6455],{"class":6377},[5361,6812,6813],{"class":6342},".conv2(out)\n",[5361,6815,6817,6819,6821],{"class":5363,"line":6816},61,[5361,6818,6777],{"class":6342},[5361,6820,6455],{"class":6377},[5361,6822,6823],{"class":6342},".bn2(out)\n",[5361,6825,6827],{"class":5363,"line":6826},62,[5361,6828,6546],{"class":6342},[5361,6830,6832],{"class":5363,"line":6831},63,[5361,6833,6834],{"class":6551},"        # Якщо розміри не співпадають, застосовуємо downsample до identity\n",[5361,6836,6838,6841,6844,6847,6850,6853,6856],{"class":5363,"line":6837},64,[5361,6839,6840],{"class":6338},"        if",[5361,6842,6843],{"class":6377}," self",[5361,6845,6846],{"class":6342},".downsample ",[5361,6848,6849],{"class":6377},"is",[5361,6851,6852],{"class":6377}," not",[5361,6854,6855],{"class":6377}," None",[5361,6857,6858],{"class":6342},":\n",[5361,6860,6862,6865,6867],{"class":5363,"line":6861},65,[5361,6863,6864],{"class":6342},"            identity = ",[5361,6866,6455],{"class":6377},[5361,6868,6869],{"class":6342},".downsample(x)\n",[5361,6871,6873],{"class":5363,"line":6872},66,[5361,6874,6546],{"class":6342},[5361,6876,6878],{"class":5363,"line":6877},67,[5361,6879,6880],{"class":6551},"        # Skip connection: y = F(x) + x\n",[5361,6882,6884],{"class":5363,"line":6883},68,[5361,6885,6886],{"class":6342},"        out += identity\n",[5361,6888,6890],{"class":5363,"line":6889},69,[5361,6891,6546],{"class":6342},[5361,6893,6895],{"class":5363,"line":6894},70,[5361,6896,6897],{"class":6551},"        # Фінальна активація\n",[5361,6899,6901],{"class":5363,"line":6900},71,[5361,6902,6798],{"class":6342},[5361,6904,6906],{"class":5363,"line":6905},72,[5361,6907,6546],{"class":6342},[5361,6909,6911,6914],{"class":5363,"line":6910},73,[5361,6912,6913],{"class":6338},"        return",[5361,6915,6916],{"class":6342}," out\n",[5361,6918,6920],{"class":5363,"line":6919},74,[5361,6921,5392],{"emptyLinePlaceholder":5391},[5361,6923,6925],{"class":5363,"line":6924},75,[5361,6926,5392],{"emptyLinePlaceholder":5391},[5361,6928,6930],{"class":5363,"line":6929},76,[5361,6931,6932],{"class":6551},"# Тестування BasicBlock\n",[5361,6934,6936,6939,6941,6944,6948,6951],{"class":5363,"line":6935},77,[5361,6937,6938],{"class":6448},"print",[5361,6940,6385],{"class":6342},[5361,6942,6943],{"class":6401},"\"🧪 Тестування BasicBlock",[5361,6945,6947],{"class":6946},"sjcCO","\\n",[5361,6949,6950],{"class":6401},"\"",[5361,6952,6953],{"class":6342},")\n",[5361,6955,6957],{"class":5363,"line":6956},78,[5361,6958,5392],{"emptyLinePlaceholder":5391},[5361,6960,6962],{"class":5363,"line":6961},79,[5361,6963,6964],{"class":6551},"# Випадок 1: розміри співпадають (skip connection без трансформації)\n",[5361,6966,6968,6971,6973,6975,6978,6980,6982,6984,6986,6988,6990,6992,6994],{"class":5363,"line":6967},80,[5361,6969,6970],{"class":6342},"block1 = BasicBlock(",[5361,6972,6461],{"class":6454},[5361,6974,6474],{"class":6342},[5361,6976,6977],{"class":6477},"64",[5361,6979,6458],{"class":6342},[5361,6981,6466],{"class":6454},[5361,6983,6474],{"class":6342},[5361,6985,6977],{"class":6477},[5361,6987,6458],{"class":6342},[5361,6989,6471],{"class":6454},[5361,6991,6474],{"class":6342},[5361,6993,6326],{"class":6477},[5361,6995,6953],{"class":6342},[5361,6997,6999,7002,7005,7007,7009,7011,7014,7016,7018,7021],{"class":5363,"line":6998},81,[5361,7000,7001],{"class":6342},"x1 = torch.randn(",[5361,7003,7004],{"class":6477},"4",[5361,7006,6458],{"class":6342},[5361,7008,6977],{"class":6477},[5361,7010,6458],{"class":6342},[5361,7012,7013],{"class":6477},"32",[5361,7015,6458],{"class":6342},[5361,7017,7013],{"class":6477},[5361,7019,7020],{"class":6342},")  ",[5361,7022,7023],{"class":6551},"# [batch, channels, H, W]\n",[5361,7025,7027],{"class":5363,"line":7026},82,[5361,7028,7029],{"class":6342},"y1 = block1(x1)\n",[5361,7031,7033,7035,7037,7040,7043,7046,7049,7052,7054],{"class":5363,"line":7032},83,[5361,7034,6938],{"class":6448},[5361,7036,6385],{"class":6342},[5361,7038,7039],{"class":6377},"f",[5361,7041,7042],{"class":6401},"\"📥 Вхід: ",[5361,7044,7045],{"class":6377},"{",[5361,7047,7048],{"class":6342},"x1.shape",[5361,7050,7051],{"class":6377},"}",[5361,7053,6950],{"class":6401},[5361,7055,6953],{"class":6342},[5361,7057,7059,7061,7063,7065,7068,7070,7073,7075,7077],{"class":5363,"line":7058},84,[5361,7060,6938],{"class":6448},[5361,7062,6385],{"class":6342},[5361,7064,7039],{"class":6377},[5361,7066,7067],{"class":6401},"\"📤 Вихід: ",[5361,7069,7045],{"class":6377},[5361,7071,7072],{"class":6342},"y1.shape",[5361,7074,7051],{"class":6377},[5361,7076,6950],{"class":6401},[5361,7078,6953],{"class":6342},[5361,7080,7082,7084,7086,7088,7091,7093,7095],{"class":5363,"line":7081},85,[5361,7083,6938],{"class":6448},[5361,7085,6385],{"class":6342},[5361,7087,7039],{"class":6377},[5361,7089,7090],{"class":6401},"\"✅ Розміри співпадають — skip connection без трансформації",[5361,7092,6947],{"class":6946},[5361,7094,6950],{"class":6401},[5361,7096,6953],{"class":6342},[5361,7098,7100],{"class":5363,"line":7099},86,[5361,7101,5392],{"emptyLinePlaceholder":5391},[5361,7103,7105],{"class":5363,"line":7104},87,[5361,7106,7107],{"class":6551},"# Випадок 2: зміна кількості каналів та downsampling\n",[5361,7109,7111],{"class":5363,"line":7110},88,[5361,7112,7113],{"class":6342},"downsample = nn.Sequential(\n",[5361,7115,7117,7120,7122,7124,7127,7129,7132,7134,7136,7138,7140,7142,7145,7147,7150,7152,7154],{"class":5363,"line":7116},89,[5361,7118,7119],{"class":6342},"    nn.Conv2d(",[5361,7121,6977],{"class":6477},[5361,7123,6458],{"class":6342},[5361,7125,7126],{"class":6477},"128",[5361,7128,6458],{"class":6342},[5361,7130,7131],{"class":6454},"kernel_size",[5361,7133,6474],{"class":6342},[5361,7135,6326],{"class":6477},[5361,7137,6458],{"class":6342},[5361,7139,6471],{"class":6454},[5361,7141,6474],{"class":6342},[5361,7143,7144],{"class":6477},"2",[5361,7146,6458],{"class":6342},[5361,7148,7149],{"class":6454},"bias",[5361,7151,6474],{"class":6342},[5361,7153,6615],{"class":6377},[5361,7155,7156],{"class":6342},"),\n",[5361,7158,7160,7163,7165],{"class":5363,"line":7159},90,[5361,7161,7162],{"class":6342},"    nn.BatchNorm2d(",[5361,7164,7126],{"class":6477},[5361,7166,6953],{"class":6342},[5361,7168,7170],{"class":5363,"line":7169},91,[5361,7171,6953],{"class":6342},[5361,7173,7175,7178,7180,7182,7184,7186,7188,7190,7192,7194,7196,7198,7200,7202,7204],{"class":5363,"line":7174},92,[5361,7176,7177],{"class":6342},"block2 = BasicBlock(",[5361,7179,6461],{"class":6454},[5361,7181,6474],{"class":6342},[5361,7183,6977],{"class":6477},[5361,7185,6458],{"class":6342},[5361,7187,6466],{"class":6454},[5361,7189,6474],{"class":6342},[5361,7191,7126],{"class":6477},[5361,7193,6458],{"class":6342},[5361,7195,6471],{"class":6454},[5361,7197,6474],{"class":6342},[5361,7199,7144],{"class":6477},[5361,7201,6458],{"class":6342},[5361,7203,6482],{"class":6454},[5361,7205,7206],{"class":6342},"=downsample)\n",[5361,7208,7210,7213,7215,7217,7219,7221,7223,7225,7227],{"class":5363,"line":7209},93,[5361,7211,7212],{"class":6342},"x2 = torch.randn(",[5361,7214,7004],{"class":6477},[5361,7216,6458],{"class":6342},[5361,7218,6977],{"class":6477},[5361,7220,6458],{"class":6342},[5361,7222,7013],{"class":6477},[5361,7224,6458],{"class":6342},[5361,7226,7013],{"class":6477},[5361,7228,6953],{"class":6342},[5361,7230,7232],{"class":5363,"line":7231},94,[5361,7233,7234],{"class":6342},"y2 = block2(x2)\n",[5361,7236,7238,7240,7242,7244,7246,7248,7251,7253,7255],{"class":5363,"line":7237},95,[5361,7239,6938],{"class":6448},[5361,7241,6385],{"class":6342},[5361,7243,7039],{"class":6377},[5361,7245,7042],{"class":6401},[5361,7247,7045],{"class":6377},[5361,7249,7250],{"class":6342},"x2.shape",[5361,7252,7051],{"class":6377},[5361,7254,6950],{"class":6401},[5361,7256,6953],{"class":6342},[5361,7258,7260,7262,7264,7266,7268,7270,7273,7275,7277],{"class":5363,"line":7259},96,[5361,7261,6938],{"class":6448},[5361,7263,6385],{"class":6342},[5361,7265,7039],{"class":6377},[5361,7267,7067],{"class":6401},[5361,7269,7045],{"class":6377},[5361,7271,7272],{"class":6342},"y2.shape",[5361,7274,7051],{"class":6377},[5361,7276,6950],{"class":6401},[5361,7278,6953],{"class":6342},[5361,7280,7282,7284,7286,7288,7291],{"class":5363,"line":7281},97,[5361,7283,6938],{"class":6448},[5361,7285,6385],{"class":6342},[5361,7287,7039],{"class":6377},[5361,7289,7290],{"class":6401},"\"🔽 Downsampling: H×W зменшено вдвічі, канали подвоєно\"",[5361,7292,6953],{"class":6342},[7294,7295,7296,7302,7308,7327,7333],"template",{"v-slot:output":5356},[5191,7297,7298,7299],{},"🧪 ",[5195,7300,7301],{},"Тестування BasicBlock",[5191,7303,7304,7307],{},[5195,7305,7306],{},"Випадок 1:"," Розміри входу та виходу однакові",[5747,7309,7310,7316,7321],{},[5750,7311,7312,7313],{},"📥 Вхід: ",[5358,7314,7315],{},"torch.Size([4, 64, 32, 32])",[5750,7317,7318,7319],{},"📤 Вихід: ",[5358,7320,7315],{},[5750,7322,7323,7324],{},"✅ Skip connection — пряме додавання ",[5358,7325,7326],{},"identity + F(x)",[5191,7328,7329,7332],{},[5195,7330,7331],{},"Випадок 2:"," Зміна розмірності (downsampling)",[5747,7334,7335,7339,7344,7347],{},[5750,7336,7312,7337],{},[5358,7338,7315],{},[5750,7340,7318,7341],{},[5358,7342,7343],{},"torch.Size([4, 128, 16, 16])",[5750,7345,7346],{},"🔽 Просторові розміри: 32×32 → 16×16 (stride=2)",[5750,7348,7349],{},"🔽 Кількість каналів: 64 → 128 (downsample projection)",[5736,7351,7352,7357,7372],{},[5191,7353,7354],{},[5195,7355,7356],{},"Критичний момент: узгодження розмірів",[5191,7358,7359,7360,7362,7363,7365,7366,7368,7369,5198],{},"Skip connection працює додаванням: ",[5693,7361],{"tex":5932,":inline":5696},". Це вимагає, щоб ",[5693,7364],{"tex":5942,":inline":5696}," та ",[5693,7367],{"tex":5897,":inline":5696}," мали ",[5195,7370,7371],{},"ідентичну форму",[5191,7373,7374,7375,7377],{},"Якщо residual function змінює кількість каналів (64 → 128) або просторову роздільність (32×32 → 16×16), skip connection потрібно трансформувати через ",[5358,7376,6482],{}," блок (зазвичай Conv 1×1 + BatchNorm).",[5280,7379,7381],{"id":7380},"bottleneck-block-ефективна-версія-для-глибоких-мереж","Bottleneck Block — ефективна версія для глибоких мереж",[5191,7383,7384,7385,7388],{},"Для дуже глибоких архітектур (ResNet-50\u002F101\u002F152) використовується ",[5195,7386,7387],{},"Bottleneck design"," — трьохшаровий блок, що зменшує обчислювальну складність:",[6314,7390,7391,7395],{"type":6316},[5280,7392,7394],{"id":7393},"bottleneck-block-11-33-11","Bottleneck Block: 1×1 → 3×3 → 1×1",[5191,7396,7397],{},"Ідея: спочатку зменшити кількість каналів (bottleneck), потім застосувати дорогу згортку 3×3, потім розширити назад. Це значно зменшує кількість параметрів та обчислень.",[6314,7399,7400,8111],{"execution-count":7144},[5351,7401,7403],{"className":6329,"code":7402,"language":6331,"meta":5356,"style":5356},"class BottleneckBlock(nn.Module):\n    \"\"\"\n    Bottleneck Residual Block для ResNet-50\u002F101\u002F152\n    \n    Структура:\n        x → [Conv 1×1 → BN → ReLU → Conv 3×3 → BN → ReLU → Conv 1×1 → BN] → (+) → ReLU\n            └──────────────────────────────────────────────────────────────────┘\n    \n    Bottleneck зменшує кількість параметрів:\n        - Без bottleneck (64 → 64): 2 × (3×3×64×64) = 73,728 параметрів\n        - З bottleneck (64 → 16 → 16 → 64): 1×1×64×16 + 3×3×16×16 + 1×1×16×64 = 5,120 параметрів\n        Зменшення у ~14 разів!\n    \"\"\"\n    expansion = 4  # Вихідні канали = bottleneck_channels × expansion\n    \n    def __init__(self, in_channels, bottleneck_channels, stride=1, downsample=None):\n        super(BottleneckBlock, self).__init__()\n        \n        # 1×1 conv: зменшення розмірності (dimension reduction)\n        self.conv1 = nn.Conv2d(in_channels, bottleneck_channels, kernel_size=1, bias=False)\n        self.bn1 = nn.BatchNorm2d(bottleneck_channels)\n        \n        # 3×3 conv: основна обробка (може зменшити H×W через stride)\n        self.conv2 = nn.Conv2d(\n            bottleneck_channels, \n            bottleneck_channels, \n            kernel_size=3, \n            stride=stride, \n            padding=1, \n            bias=False\n        )\n        self.bn2 = nn.BatchNorm2d(bottleneck_channels)\n        \n        # 1×1 conv: розширення розмірності (dimension expansion)\n        out_channels = bottleneck_channels * self.expansion\n        self.conv3 = nn.Conv2d(bottleneck_channels, out_channels, kernel_size=1, bias=False)\n        self.bn3 = nn.BatchNorm2d(out_channels)\n        \n        self.downsample = downsample\n        self.stride = stride\n    \n    def forward(self, x):\n        identity = x\n        \n        # Bottleneck pathway\n        out = F.relu(self.bn1(self.conv1(x)))    # 1×1: зменшення\n        out = F.relu(self.bn2(self.conv2(out)))  # 3×3: обробка\n        out = self.bn3(self.conv3(out))          # 1×1: розширення (без ReLU!)\n        \n        if self.downsample is not None:\n            identity = self.downsample(x)\n        \n        out += identity\n        out = F.relu(out)\n        \n        return out\n\n\n# Порівняння BasicBlock та BottleneckBlock\nprint(\"📊 Порівняння BasicBlock vs BottleneckBlock\\n\")\n\n# BasicBlock\nbasic = BasicBlock(in_channels=256, out_channels=256, stride=1)\nbasic_params = sum(p.numel() for p in basic.parameters())\n\n# BottleneckBlock (256 → 64 → 64 → 256)\nbottleneck = BottleneckBlock(in_channels=256, bottleneck_channels=64, stride=1)\nbottleneck_params = sum(p.numel() for p in bottleneck.parameters())\n\nprint(f\"BasicBlock (256 → 256):\")\nprint(f\"  └─ Параметри: {basic_params:,}\")\nprint(f\"\\nBottleneckBlock (256 → 64 → 64 → 256):\")\nprint(f\"  └─ Параметри: {bottleneck_params:,}\")\nprint(f\"\\n💾 Економія: {(1 - bottleneck_params\u002Fbasic_params)*100:.1f}%\")\n",[5358,7404,7405,7422,7426,7431,7435,7439,7444,7449,7453,7458,7463,7468,7473,7477,7487,7491,7528,7543,7547,7552,7575,7582,7586,7591,7597,7602,7606,7616,7623,7633,7641,7645,7652,7656,7661,7671,7694,7701,7705,7711,7717,7721,7737,7741,7745,7750,7768,7785,7802,7806,7822,7830,7834,7838,7842,7846,7852,7856,7860,7865,7880,7884,7889,7919,7942,7946,7951,7980,7998,8002,8015,8038,8055,8076],{"__ignoreMap":5356},[5361,7406,7407,7409,7412,7414,7416,7418,7420],{"class":5363,"line":5364},[5361,7408,6378],{"class":6377},[5361,7410,7411],{"class":6381}," BottleneckBlock",[5361,7413,6385],{"class":6342},[5361,7415,6388],{"class":6381},[5361,7417,5198],{"class":6342},[5361,7419,6393],{"class":6381},[5361,7421,6396],{"class":6342},[5361,7423,7424],{"class":5363,"line":5370},[5361,7425,6402],{"class":6401},[5361,7427,7428],{"class":5363,"line":5376},[5361,7429,7430],{"class":6401},"    Bottleneck Residual Block для ResNet-50\u002F101\u002F152\n",[5361,7432,7433],{"class":5363,"line":5382},[5361,7434,6412],{"class":6401},[5361,7436,7437],{"class":5363,"line":5388},[5361,7438,6417],{"class":6401},[5361,7440,7441],{"class":5363,"line":5395},[5361,7442,7443],{"class":6401},"        x → [Conv 1×1 → BN → ReLU → Conv 3×3 → BN → ReLU → Conv 1×1 → BN] → (+) → ReLU\n",[5361,7445,7446],{"class":5363,"line":5401},[5361,7447,7448],{"class":6401},"            └──────────────────────────────────────────────────────────────────┘\n",[5361,7450,7451],{"class":5363,"line":5407},[5361,7452,6412],{"class":6401},[5361,7454,7455],{"class":5363,"line":5413},[5361,7456,7457],{"class":6401},"    Bottleneck зменшує кількість параметрів:\n",[5361,7459,7460],{"class":5363,"line":5418},[5361,7461,7462],{"class":6401},"        - Без bottleneck (64 → 64): 2 × (3×3×64×64) = 73,728 параметрів\n",[5361,7464,7465],{"class":5363,"line":5424},[5361,7466,7467],{"class":6401},"        - З bottleneck (64 → 16 → 16 → 64): 1×1×64×16 + 3×3×16×16 + 1×1×16×64 = 5,120 параметрів\n",[5361,7469,7470],{"class":5363,"line":5430},[5361,7471,7472],{"class":6401},"        Зменшення у ~14 разів!\n",[5361,7474,7475],{"class":5363,"line":5436},[5361,7476,6402],{"class":6401},[5361,7478,7479,7482,7484],{"class":5363,"line":5442},[5361,7480,7481],{"class":6342},"    expansion = ",[5361,7483,7004],{"class":6477},[5361,7485,7486],{"class":6551},"  # Вихідні канали = bottleneck_channels × expansion\n",[5361,7488,7489],{"class":5363,"line":5448},[5361,7490,6412],{"class":6342},[5361,7492,7493,7495,7497,7499,7501,7503,7505,7507,7510,7512,7514,7516,7518,7520,7522,7524,7526],{"class":5363,"line":5454},[5361,7494,6445],{"class":6377},[5361,7496,6449],{"class":6448},[5361,7498,6385],{"class":6342},[5361,7500,6455],{"class":6454},[5361,7502,6458],{"class":6342},[5361,7504,6461],{"class":6454},[5361,7506,6458],{"class":6342},[5361,7508,7509],{"class":6454},"bottleneck_channels",[5361,7511,6458],{"class":6342},[5361,7513,6471],{"class":6454},[5361,7515,6474],{"class":6342},[5361,7517,6326],{"class":6477},[5361,7519,6458],{"class":6342},[5361,7521,6482],{"class":6454},[5361,7523,6474],{"class":6342},[5361,7525,6487],{"class":6377},[5361,7527,6396],{"class":6342},[5361,7529,7530,7532,7535,7537,7539,7541],{"class":5363,"line":5460},[5361,7531,6528],{"class":6381},[5361,7533,7534],{"class":6342},"(BottleneckBlock, ",[5361,7536,6455],{"class":6377},[5361,7538,5725],{"class":6342},[5361,7540,6538],{"class":6448},[5361,7542,6541],{"class":6342},[5361,7544,7545],{"class":5363,"line":5466},[5361,7546,6546],{"class":6342},[5361,7548,7549],{"class":5363,"line":5472},[5361,7550,7551],{"class":6551},"        # 1×1 conv: зменшення розмірності (dimension reduction)\n",[5361,7553,7554,7556,7559,7561,7563,7565,7567,7569,7571,7573],{"class":5363,"line":5478},[5361,7555,6557],{"class":6377},[5361,7557,7558],{"class":6342},".conv1 = nn.Conv2d(in_channels, bottleneck_channels, ",[5361,7560,7131],{"class":6454},[5361,7562,6474],{"class":6342},[5361,7564,6326],{"class":6477},[5361,7566,6458],{"class":6342},[5361,7568,7149],{"class":6454},[5361,7570,6474],{"class":6342},[5361,7572,6615],{"class":6377},[5361,7574,6953],{"class":6342},[5361,7576,7577,7579],{"class":5363,"line":5484},[5361,7578,6557],{"class":6377},[5361,7580,7581],{"class":6342},".bn1 = nn.BatchNorm2d(bottleneck_channels)\n",[5361,7583,7584],{"class":5363,"line":5489},[5361,7585,6546],{"class":6342},[5361,7587,7588],{"class":5363,"line":5495},[5361,7589,7590],{"class":6551},"        # 3×3 conv: основна обробка (може зменшити H×W через stride)\n",[5361,7592,7593,7595],{"class":5363,"line":5501},[5361,7594,6557],{"class":6377},[5361,7596,6646],{"class":6342},[5361,7598,7599],{"class":5363,"line":5507},[5361,7600,7601],{"class":6342},"            bottleneck_channels, \n",[5361,7603,7604],{"class":5363,"line":5513},[5361,7605,7601],{"class":6342},[5361,7607,7608,7610,7612,7614],{"class":5363,"line":5519},[5361,7609,6575],{"class":6454},[5361,7611,6474],{"class":6342},[5361,7613,6580],{"class":6477},[5361,7615,6583],{"class":6342},[5361,7617,7618,7620],{"class":5363,"line":5525},[5361,7619,6588],{"class":6454},[5361,7621,7622],{"class":6342},"=stride, \n",[5361,7624,7625,7627,7629,7631],{"class":5363,"line":5531},[5361,7626,6599],{"class":6454},[5361,7628,6474],{"class":6342},[5361,7630,6326],{"class":6477},[5361,7632,6583],{"class":6342},[5361,7634,7635,7637,7639],{"class":5363,"line":5537},[5361,7636,6610],{"class":6454},[5361,7638,6474],{"class":6342},[5361,7640,6693],{"class":6377},[5361,7642,7643],{"class":5363,"line":5542},[5361,7644,6623],{"class":6342},[5361,7646,7647,7649],{"class":5363,"line":5548},[5361,7648,6557],{"class":6377},[5361,7650,7651],{"class":6342},".bn2 = nn.BatchNorm2d(bottleneck_channels)\n",[5361,7653,7654],{"class":5363,"line":5553},[5361,7655,6546],{"class":6342},[5361,7657,7658],{"class":5363,"line":5559},[5361,7659,7660],{"class":6551},"        # 1×1 conv: розширення розмірності (dimension expansion)\n",[5361,7662,7663,7666,7668],{"class":5363,"line":5565},[5361,7664,7665],{"class":6342},"        out_channels = bottleneck_channels * ",[5361,7667,6455],{"class":6377},[5361,7669,7670],{"class":6342},".expansion\n",[5361,7672,7673,7675,7678,7680,7682,7684,7686,7688,7690,7692],{"class":5363,"line":5571},[5361,7674,6557],{"class":6377},[5361,7676,7677],{"class":6342},".conv3 = nn.Conv2d(bottleneck_channels, out_channels, ",[5361,7679,7131],{"class":6454},[5361,7681,6474],{"class":6342},[5361,7683,6326],{"class":6477},[5361,7685,6458],{"class":6342},[5361,7687,7149],{"class":6454},[5361,7689,6474],{"class":6342},[5361,7691,6615],{"class":6377},[5361,7693,6953],{"class":6342},[5361,7695,7696,7698],{"class":5363,"line":5577},[5361,7697,6557],{"class":6377},[5361,7699,7700],{"class":6342},".bn3 = nn.BatchNorm2d(out_channels)\n",[5361,7702,7703],{"class":5363,"line":5583},[5361,7704,6546],{"class":6342},[5361,7706,7707,7709],{"class":5363,"line":5588},[5361,7708,6557],{"class":6377},[5361,7710,6720],{"class":6342},[5361,7712,7713,7715],{"class":5363,"line":5594},[5361,7714,6557],{"class":6377},[5361,7716,6727],{"class":6342},[5361,7718,7719],{"class":5363,"line":5600},[5361,7720,6412],{"class":6342},[5361,7722,7723,7725,7727,7729,7731,7733,7735],{"class":5363,"line":5606},[5361,7724,6445],{"class":6377},[5361,7726,6738],{"class":6448},[5361,7728,6385],{"class":6342},[5361,7730,6455],{"class":6454},[5361,7732,6458],{"class":6342},[5361,7734,6747],{"class":6454},[5361,7736,6396],{"class":6342},[5361,7738,7739],{"class":5363,"line":5612},[5361,7740,6760],{"class":6342},[5361,7742,7743],{"class":5363,"line":5618},[5361,7744,6546],{"class":6342},[5361,7746,7747],{"class":5363,"line":5623},[5361,7748,7749],{"class":6551},"        # Bottleneck pathway\n",[5361,7751,7752,7755,7757,7760,7762,7765],{"class":5363,"line":5628},[5361,7753,7754],{"class":6342},"        out = F.relu(",[5361,7756,6455],{"class":6377},[5361,7758,7759],{"class":6342},".bn1(",[5361,7761,6455],{"class":6377},[5361,7763,7764],{"class":6342},".conv1(x)))    ",[5361,7766,7767],{"class":6551},"# 1×1: зменшення\n",[5361,7769,7770,7772,7774,7777,7779,7782],{"class":5363,"line":5634},[5361,7771,7754],{"class":6342},[5361,7773,6455],{"class":6377},[5361,7775,7776],{"class":6342},".bn2(",[5361,7778,6455],{"class":6377},[5361,7780,7781],{"class":6342},".conv2(out)))  ",[5361,7783,7784],{"class":6551},"# 3×3: обробка\n",[5361,7786,7787,7789,7791,7794,7796,7799],{"class":5363,"line":5640},[5361,7788,6777],{"class":6342},[5361,7790,6455],{"class":6377},[5361,7792,7793],{"class":6342},".bn3(",[5361,7795,6455],{"class":6377},[5361,7797,7798],{"class":6342},".conv3(out))          ",[5361,7800,7801],{"class":6551},"# 1×1: розширення (без ReLU!)\n",[5361,7803,7804],{"class":5363,"line":5646},[5361,7805,6546],{"class":6342},[5361,7807,7808,7810,7812,7814,7816,7818,7820],{"class":5363,"line":5652},[5361,7809,6840],{"class":6338},[5361,7811,6843],{"class":6377},[5361,7813,6846],{"class":6342},[5361,7815,6849],{"class":6377},[5361,7817,6852],{"class":6377},[5361,7819,6855],{"class":6377},[5361,7821,6858],{"class":6342},[5361,7823,7824,7826,7828],{"class":5363,"line":5657},[5361,7825,6864],{"class":6342},[5361,7827,6455],{"class":6377},[5361,7829,6869],{"class":6342},[5361,7831,7832],{"class":5363,"line":5662},[5361,7833,6546],{"class":6342},[5361,7835,7836],{"class":5363,"line":6757},[5361,7837,6886],{"class":6342},[5361,7839,7840],{"class":5363,"line":6763},[5361,7841,6798],{"class":6342},[5361,7843,7844],{"class":5363,"line":6768},[5361,7845,6546],{"class":6342},[5361,7847,7848,7850],{"class":5363,"line":6774},[5361,7849,6913],{"class":6338},[5361,7851,6916],{"class":6342},[5361,7853,7854],{"class":5363,"line":6785},[5361,7855,5392],{"emptyLinePlaceholder":5391},[5361,7857,7858],{"class":5363,"line":6795},[5361,7859,5392],{"emptyLinePlaceholder":5391},[5361,7861,7862],{"class":5363,"line":6801},[5361,7863,7864],{"class":6551},"# Порівняння BasicBlock та BottleneckBlock\n",[5361,7866,7867,7869,7871,7874,7876,7878],{"class":5363,"line":6806},[5361,7868,6938],{"class":6448},[5361,7870,6385],{"class":6342},[5361,7872,7873],{"class":6401},"\"📊 Порівняння BasicBlock vs BottleneckBlock",[5361,7875,6947],{"class":6946},[5361,7877,6950],{"class":6401},[5361,7879,6953],{"class":6342},[5361,7881,7882],{"class":5363,"line":6816},[5361,7883,5392],{"emptyLinePlaceholder":5391},[5361,7885,7886],{"class":5363,"line":6826},[5361,7887,7888],{"class":6551},"# BasicBlock\n",[5361,7890,7891,7894,7896,7898,7901,7903,7905,7907,7909,7911,7913,7915,7917],{"class":5363,"line":6831},[5361,7892,7893],{"class":6342},"basic = BasicBlock(",[5361,7895,6461],{"class":6454},[5361,7897,6474],{"class":6342},[5361,7899,7900],{"class":6477},"256",[5361,7902,6458],{"class":6342},[5361,7904,6466],{"class":6454},[5361,7906,6474],{"class":6342},[5361,7908,7900],{"class":6477},[5361,7910,6458],{"class":6342},[5361,7912,6471],{"class":6454},[5361,7914,6474],{"class":6342},[5361,7916,6326],{"class":6477},[5361,7918,6953],{"class":6342},[5361,7920,7921,7924,7927,7930,7933,7936,7939],{"class":5363,"line":6837},[5361,7922,7923],{"class":6342},"basic_params = ",[5361,7925,7926],{"class":6448},"sum",[5361,7928,7929],{"class":6342},"(p.numel() ",[5361,7931,7932],{"class":6338},"for",[5361,7934,7935],{"class":6342}," p ",[5361,7937,7938],{"class":6338},"in",[5361,7940,7941],{"class":6342}," basic.parameters())\n",[5361,7943,7944],{"class":5363,"line":6861},[5361,7945,5392],{"emptyLinePlaceholder":5391},[5361,7947,7948],{"class":5363,"line":6872},[5361,7949,7950],{"class":6551},"# BottleneckBlock (256 → 64 → 64 → 256)\n",[5361,7952,7953,7956,7958,7960,7962,7964,7966,7968,7970,7972,7974,7976,7978],{"class":5363,"line":6877},[5361,7954,7955],{"class":6342},"bottleneck = BottleneckBlock(",[5361,7957,6461],{"class":6454},[5361,7959,6474],{"class":6342},[5361,7961,7900],{"class":6477},[5361,7963,6458],{"class":6342},[5361,7965,7509],{"class":6454},[5361,7967,6474],{"class":6342},[5361,7969,6977],{"class":6477},[5361,7971,6458],{"class":6342},[5361,7973,6471],{"class":6454},[5361,7975,6474],{"class":6342},[5361,7977,6326],{"class":6477},[5361,7979,6953],{"class":6342},[5361,7981,7982,7985,7987,7989,7991,7993,7995],{"class":5363,"line":6883},[5361,7983,7984],{"class":6342},"bottleneck_params = ",[5361,7986,7926],{"class":6448},[5361,7988,7929],{"class":6342},[5361,7990,7932],{"class":6338},[5361,7992,7935],{"class":6342},[5361,7994,7938],{"class":6338},[5361,7996,7997],{"class":6342}," bottleneck.parameters())\n",[5361,7999,8000],{"class":5363,"line":6889},[5361,8001,5392],{"emptyLinePlaceholder":5391},[5361,8003,8004,8006,8008,8010,8013],{"class":5363,"line":6894},[5361,8005,6938],{"class":6448},[5361,8007,6385],{"class":6342},[5361,8009,7039],{"class":6377},[5361,8011,8012],{"class":6401},"\"BasicBlock (256 → 256):\"",[5361,8014,6953],{"class":6342},[5361,8016,8017,8019,8021,8023,8026,8028,8031,8034,8036],{"class":5363,"line":6900},[5361,8018,6938],{"class":6448},[5361,8020,6385],{"class":6342},[5361,8022,7039],{"class":6377},[5361,8024,8025],{"class":6401},"\"  └─ Параметри: ",[5361,8027,7045],{"class":6377},[5361,8029,8030],{"class":6342},"basic_params",[5361,8032,8033],{"class":6377},":,}",[5361,8035,6950],{"class":6401},[5361,8037,6953],{"class":6342},[5361,8039,8040,8042,8044,8046,8048,8050,8053],{"class":5363,"line":6905},[5361,8041,6938],{"class":6448},[5361,8043,6385],{"class":6342},[5361,8045,7039],{"class":6377},[5361,8047,6950],{"class":6401},[5361,8049,6947],{"class":6946},[5361,8051,8052],{"class":6401},"BottleneckBlock (256 → 64 → 64 → 256):\"",[5361,8054,6953],{"class":6342},[5361,8056,8057,8059,8061,8063,8065,8067,8070,8072,8074],{"class":5363,"line":6910},[5361,8058,6938],{"class":6448},[5361,8060,6385],{"class":6342},[5361,8062,7039],{"class":6377},[5361,8064,8025],{"class":6401},[5361,8066,7045],{"class":6377},[5361,8068,8069],{"class":6342},"bottleneck_params",[5361,8071,8033],{"class":6377},[5361,8073,6950],{"class":6401},[5361,8075,6953],{"class":6342},[5361,8077,8078,8080,8082,8084,8086,8088,8091,8093,8095,8097,8100,8103,8106,8109],{"class":5363,"line":6919},[5361,8079,6938],{"class":6448},[5361,8081,6385],{"class":6342},[5361,8083,7039],{"class":6377},[5361,8085,6950],{"class":6401},[5361,8087,6947],{"class":6946},[5361,8089,8090],{"class":6401},"💾 Економія: ",[5361,8092,7045],{"class":6377},[5361,8094,6385],{"class":6342},[5361,8096,6326],{"class":6477},[5361,8098,8099],{"class":6342}," - bottleneck_params\u002Fbasic_params)*",[5361,8101,8102],{"class":6477},"100",[5361,8104,8105],{"class":6377},":.1f}",[5361,8107,8108],{"class":6401},"%\"",[5361,8110,6953],{"class":6342},[7294,8112,8113,8119,8173],{"v-slot:output":5356},[5191,8114,8115,8116],{},"📊 ",[5195,8117,8118],{},"Порівняння BasicBlock vs BottleneckBlock",[5203,8120,8121,8136],{},[5206,8122,8123],{},[5209,8124,8125,8128,8130,8133],{},[5212,8126,8127],{"align":5214},"Блок",[5212,8129,5218],{"align":5214},[5212,8131,8132],{"align":5214},"Параметри",[5212,8134,8135],{"align":5214},"Відносна ефективність",[5226,8137,8138,8154],{},[5209,8139,8140,8145,8148,8151],{},[5231,8141,8142],{"align":5214},[5195,8143,8144],{},"BasicBlock",[5231,8146,8147],{"align":5214},"256 → 256 → 256 (3×3 conv twice)",[5231,8149,8150],{"align":5214},"1,180,672",[5231,8152,8153],{"align":5214},"Базовий",[5209,8155,8156,8161,8164,8167],{},[5231,8157,8158],{"align":5214},[5195,8159,8160],{},"BottleneckBlock",[5231,8162,8163],{"align":5214},"256 → 64 → 64 → 256 (1×1 → 3×3 → 1×1)",[5231,8165,8166],{"align":5214},"70,400",[5231,8168,8169,8170],{"align":5214},"✅ ",[5195,8171,8172],{},"94.0% менше",[5191,8174,8175,8176,8179],{},"💡 Bottleneck дозволяє будувати ",[5195,8177,8178],{},"значно глибші"," мережі з тією самою обчислювальною бюджетом!",[5191,8181,6259],{},[5312,8183,8184,8189,8200,8238],{},[5191,8185,8186],{},[5195,8187,8188],{},"Критична деталь: ReLU після skip connection",[5191,8190,8191,8192,8195,8196,8199],{},"Зверніть увагу: фінальний ",[5358,8193,8194],{},"ReLU"," застосовується ",[5195,8197,8198],{},"після"," додавання skip connection:",[5351,8201,8203],{"className":6329,"code":8202,"language":6331,"meta":5356,"style":5356},"out = self.bn3(self.conv3(out))  # Остання BN БЕЗ ReLU\nout += identity                  # Skip connection\nout = F.relu(out)                # ReLU тільки тут!\n",[5358,8204,8205,8222,8230],{"__ignoreMap":5356},[5361,8206,8207,8210,8212,8214,8216,8219],{"class":5363,"line":5364},[5361,8208,8209],{"class":6342},"out = ",[5361,8211,6455],{"class":6377},[5361,8213,7793],{"class":6342},[5361,8215,6455],{"class":6377},[5361,8217,8218],{"class":6342},".conv3(out))  ",[5361,8220,8221],{"class":6551},"# Остання BN БЕЗ ReLU\n",[5361,8223,8224,8227],{"class":5363,"line":5370},[5361,8225,8226],{"class":6342},"out += identity                  ",[5361,8228,8229],{"class":6551},"# Skip connection\n",[5361,8231,8232,8235],{"class":5363,"line":5376},[5361,8233,8234],{"class":6342},"out = F.relu(out)                ",[5361,8236,8237],{"class":6551},"# ReLU тільки тут!\n",[5191,8239,8240,8241,8243,8244,8247],{},"Чому? Якщо застосувати ReLU до ",[5693,8242],{"tex":5942,":inline":5696}," перед додаванням, це порушить можливість навчання від'ємних залишків. Identity mapping має бути ",[5195,8245,8246],{},"лінійним"," для ефективного gradient flow.",[5855,8249],{},[5186,8251,8253],{"id":8252},"побудова-повноцінної-resnet-архітектури","Побудова повноцінної ResNet архітектури",[5191,8255,8256],{},"Тепер, маючи будівельні блоки, зберемо повноцінну ResNet архітектуру. Розглянемо ResNet-18 — найменшу версію, ідеальну для навчальних цілей та датасетів середнього розміру (CIFAR-10, Fashion-MNIST).",[5280,8258,8260],{"id":8259},"архітектура-resnet-18","Архітектура ResNet-18",[5347,8262,8264],{"alt":8263},"ResNet-18 архітектура — повна структура",[5351,8265,8267],{"className":5353,"code":8266,"language":5355,"meta":5356,"style":5356},"@startuml\nskinparam style plain\nskinparam backgroundColor #FFFFFF\nskinparam defaultFontName \"Arial\"\n\n!define INPUT_COLOR #DBEAFE\n!define CONV_COLOR #FEF3C7\n!define RESIDUAL_COLOR #DCFCE7\n!define POOL_COLOR #E0E7FF\n!define FC_COLOR #FEE2E2\n\nrectangle \"Вхід: 3×224×224\" as input #INPUT_COLOR\n\nrectangle \"Initial Conv Block\" as init #CONV_COLOR {\n  rectangle \"Conv 7×7, stride=2, 64 filters\\nBatchNorm, ReLU\" as conv1\n  rectangle \"MaxPool 3×3, stride=2\" as pool1\n}\n\npackage \"Layer 1: 64 channels\" as layer1 #RESIDUAL_COLOR {\n  rectangle \"BasicBlock × 2\" as block1\n}\n\npackage \"Layer 2: 128 channels\" as layer2 #RESIDUAL_COLOR {\n  rectangle \"BasicBlock × 2\\n(stride=2 перший блок)\" as block2\n}\n\npackage \"Layer 3: 256 channels\" as layer3 #RESIDUAL_COLOR {\n  rectangle \"BasicBlock × 2\\n(stride=2 перший блок)\" as block3\n}\n\npackage \"Layer 4: 512 channels\" as layer4 #RESIDUAL_COLOR {\n  rectangle \"BasicBlock × 2\\n(stride=2 перший блок)\" as block4\n}\n\nrectangle \"Global Average Pool\\n512 → 512×1×1\" as gap #POOL_COLOR\nrectangle \"Fully Connected\\n512 → num_classes\" as fc #FC_COLOR\nrectangle \"Output: class probabilities\" as output #INPUT_COLOR\n\ninput -down-> init\ninit -down-> layer1 : 64×56×56\nlayer1 -down-> layer2 : 128×28×28\nlayer2 -down-> layer3 : 256×14×14\nlayer3 -down-> layer4 : 512×7×7\nlayer4 -down-> gap : 512×1×1\ngap -down-> fc\nfc -down-> output\n\nnote right of init\n  📐 Розмір після pool1:\n  64 × 56 × 56\nend note\n\nnote right of layer4\n  🏗️ Загалом 8 residual блоків\n  (2 × 4 шари)\n  \n  📊 Загальна глибина: 18 шарів\n  (1 initial + 16 conv у блоках + 1 FC)\nend note\n\n@enduml\n",[5358,8268,8269,8273,8277,8281,8285,8289,8293,8298,8303,8308,8313,8317,8322,8326,8331,8336,8341,8345,8349,8354,8359,8363,8367,8372,8377,8381,8385,8390,8395,8399,8403,8408,8413,8417,8421,8426,8431,8436,8440,8445,8450,8455,8460,8465,8470,8475,8480,8484,8489,8494,8499,8503,8507,8512,8517,8522,8526,8531,8536,8540,8544],{"__ignoreMap":5356},[5361,8270,8271],{"class":5363,"line":5364},[5361,8272,5367],{},[5361,8274,8275],{"class":5363,"line":5370},[5361,8276,5373],{},[5361,8278,8279],{"class":5363,"line":5376},[5361,8280,5379],{},[5361,8282,8283],{"class":5363,"line":5382},[5361,8284,5385],{},[5361,8286,8287],{"class":5363,"line":5388},[5361,8288,5392],{"emptyLinePlaceholder":5391},[5361,8290,8291],{"class":5363,"line":5395},[5361,8292,6003],{},[5361,8294,8295],{"class":5363,"line":5401},[5361,8296,8297],{},"!define CONV_COLOR #FEF3C7\n",[5361,8299,8300],{"class":5363,"line":5407},[5361,8301,8302],{},"!define RESIDUAL_COLOR #DCFCE7\n",[5361,8304,8305],{"class":5363,"line":5413},[5361,8306,8307],{},"!define POOL_COLOR #E0E7FF\n",[5361,8309,8310],{"class":5363,"line":5418},[5361,8311,8312],{},"!define FC_COLOR #FEE2E2\n",[5361,8314,8315],{"class":5363,"line":5424},[5361,8316,5392],{"emptyLinePlaceholder":5391},[5361,8318,8319],{"class":5363,"line":5430},[5361,8320,8321],{},"rectangle \"Вхід: 3×224×224\" as input #INPUT_COLOR\n",[5361,8323,8324],{"class":5363,"line":5436},[5361,8325,5392],{"emptyLinePlaceholder":5391},[5361,8327,8328],{"class":5363,"line":5442},[5361,8329,8330],{},"rectangle \"Initial Conv Block\" as init #CONV_COLOR {\n",[5361,8332,8333],{"class":5363,"line":5448},[5361,8334,8335],{},"  rectangle \"Conv 7×7, stride=2, 64 filters\\nBatchNorm, ReLU\" as conv1\n",[5361,8337,8338],{"class":5363,"line":5454},[5361,8339,8340],{},"  rectangle \"MaxPool 3×3, stride=2\" as pool1\n",[5361,8342,8343],{"class":5363,"line":5460},[5361,8344,5545],{},[5361,8346,8347],{"class":5363,"line":5466},[5361,8348,5392],{"emptyLinePlaceholder":5391},[5361,8350,8351],{"class":5363,"line":5472},[5361,8352,8353],{},"package \"Layer 1: 64 channels\" as layer1 #RESIDUAL_COLOR {\n",[5361,8355,8356],{"class":5363,"line":5478},[5361,8357,8358],{},"  rectangle \"BasicBlock × 2\" as block1\n",[5361,8360,8361],{"class":5363,"line":5484},[5361,8362,5545],{},[5361,8364,8365],{"class":5363,"line":5489},[5361,8366,5392],{"emptyLinePlaceholder":5391},[5361,8368,8369],{"class":5363,"line":5495},[5361,8370,8371],{},"package \"Layer 2: 128 channels\" as layer2 #RESIDUAL_COLOR {\n",[5361,8373,8374],{"class":5363,"line":5501},[5361,8375,8376],{},"  rectangle \"BasicBlock × 2\\n(stride=2 перший блок)\" as block2\n",[5361,8378,8379],{"class":5363,"line":5507},[5361,8380,5545],{},[5361,8382,8383],{"class":5363,"line":5513},[5361,8384,5392],{"emptyLinePlaceholder":5391},[5361,8386,8387],{"class":5363,"line":5519},[5361,8388,8389],{},"package \"Layer 3: 256 channels\" as layer3 #RESIDUAL_COLOR {\n",[5361,8391,8392],{"class":5363,"line":5525},[5361,8393,8394],{},"  rectangle \"BasicBlock × 2\\n(stride=2 перший блок)\" as block3\n",[5361,8396,8397],{"class":5363,"line":5531},[5361,8398,5545],{},[5361,8400,8401],{"class":5363,"line":5537},[5361,8402,5392],{"emptyLinePlaceholder":5391},[5361,8404,8405],{"class":5363,"line":5542},[5361,8406,8407],{},"package \"Layer 4: 512 channels\" as layer4 #RESIDUAL_COLOR {\n",[5361,8409,8410],{"class":5363,"line":5548},[5361,8411,8412],{},"  rectangle \"BasicBlock × 2\\n(stride=2 перший блок)\" as block4\n",[5361,8414,8415],{"class":5363,"line":5553},[5361,8416,5545],{},[5361,8418,8419],{"class":5363,"line":5559},[5361,8420,5392],{"emptyLinePlaceholder":5391},[5361,8422,8423],{"class":5363,"line":5565},[5361,8424,8425],{},"rectangle \"Global Average Pool\\n512 → 512×1×1\" as gap #POOL_COLOR\n",[5361,8427,8428],{"class":5363,"line":5571},[5361,8429,8430],{},"rectangle \"Fully Connected\\n512 → num_classes\" as fc #FC_COLOR\n",[5361,8432,8433],{"class":5363,"line":5577},[5361,8434,8435],{},"rectangle \"Output: class probabilities\" as output #INPUT_COLOR\n",[5361,8437,8438],{"class":5363,"line":5583},[5361,8439,5392],{"emptyLinePlaceholder":5391},[5361,8441,8442],{"class":5363,"line":5588},[5361,8443,8444],{},"input -down-> init\n",[5361,8446,8447],{"class":5363,"line":5594},[5361,8448,8449],{},"init -down-> layer1 : 64×56×56\n",[5361,8451,8452],{"class":5363,"line":5600},[5361,8453,8454],{},"layer1 -down-> layer2 : 128×28×28\n",[5361,8456,8457],{"class":5363,"line":5606},[5361,8458,8459],{},"layer2 -down-> layer3 : 256×14×14\n",[5361,8461,8462],{"class":5363,"line":5612},[5361,8463,8464],{},"layer3 -down-> layer4 : 512×7×7\n",[5361,8466,8467],{"class":5363,"line":5618},[5361,8468,8469],{},"layer4 -down-> gap : 512×1×1\n",[5361,8471,8472],{"class":5363,"line":5623},[5361,8473,8474],{},"gap -down-> fc\n",[5361,8476,8477],{"class":5363,"line":5628},[5361,8478,8479],{},"fc -down-> output\n",[5361,8481,8482],{"class":5363,"line":5634},[5361,8483,5392],{"emptyLinePlaceholder":5391},[5361,8485,8486],{"class":5363,"line":5640},[5361,8487,8488],{},"note right of init\n",[5361,8490,8491],{"class":5363,"line":5646},[5361,8492,8493],{},"  📐 Розмір після pool1:\n",[5361,8495,8496],{"class":5363,"line":5652},[5361,8497,8498],{},"  64 × 56 × 56\n",[5361,8500,8501],{"class":5363,"line":5657},[5361,8502,5580],{},[5361,8504,8505],{"class":5363,"line":5662},[5361,8506,5392],{"emptyLinePlaceholder":5391},[5361,8508,8509],{"class":5363,"line":6757},[5361,8510,8511],{},"note right of layer4\n",[5361,8513,8514],{"class":5363,"line":6763},[5361,8515,8516],{},"  🏗️ Загалом 8 residual блоків\n",[5361,8518,8519],{"class":5363,"line":6768},[5361,8520,8521],{},"  (2 × 4 шари)\n",[5361,8523,8524],{"class":5363,"line":6774},[5361,8525,5427],{},[5361,8527,8528],{"class":5363,"line":6785},[5361,8529,8530],{},"  📊 Загальна глибина: 18 шарів\n",[5361,8532,8533],{"class":5363,"line":6795},[5361,8534,8535],{},"  (1 initial + 16 conv у блоках + 1 FC)\n",[5361,8537,8538],{"class":5363,"line":6801},[5361,8539,5580],{},[5361,8541,8542],{"class":5363,"line":6806},[5361,8543,5392],{"emptyLinePlaceholder":5391},[5361,8545,8546],{"class":5363,"line":6816},[5361,8547,5665],{},[5280,8549,8551],{"id":8550},"повна-реалізація-resnet-18","Повна реалізація ResNet-18",[6314,8553,8554,10051],{"execution-count":6580},[5351,8555,8557],{"className":6329,"code":8556,"language":6331,"meta":5356,"style":5356},"class ResNet(nn.Module):\n    \"\"\"\n    ResNet архітектура для класифікації зображень\n    \n    Підтримує різні конфігурації:\n        - ResNet-18: [2, 2, 2, 2] BasicBlocks\n        - ResNet-34: [3, 4, 6, 3] BasicBlocks\n        - ResNet-50: [3, 4, 6, 3] BottleneckBlocks\n        - ResNet-101: [3, 4, 23, 3] BottleneckBlocks\n    \"\"\"\n    \n    def __init__(self, block, layers, num_classes=10, input_channels=3):\n        \"\"\"\n        Args:\n            block: BasicBlock або BottleneckBlock\n            layers: список кількості блоків у кожному шарі, наприклад [2, 2, 2, 2]\n            num_classes: кількість класів для класифікації\n            input_channels: кількість каналів входу (3 для RGB, 1 для grayscale)\n        \"\"\"\n        super(ResNet, self).__init__()\n        self.in_channels = 64\n        \n        # Початковий convolutional block\n        self.conv1 = nn.Conv2d(\n            input_channels, 64, kernel_size=7, stride=2, padding=3, bias=False\n        )\n        self.bn1 = nn.BatchNorm2d(64)\n        self.relu = nn.ReLU(inplace=True)\n        self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1)\n        \n        # Residual layers (4 групи блоків)\n        self.layer1 = self._make_layer(block, 64, layers[0], stride=1)\n        self.layer2 = self._make_layer(block, 128, layers[1], stride=2)\n        self.layer3 = self._make_layer(block, 256, layers[2], stride=2)\n        self.layer4 = self._make_layer(block, 512, layers[3], stride=2)\n        \n        # Глобальне усереднення та класифікатор\n        self.avgpool = nn.AdaptiveAvgPool2d((1, 1))\n        self.fc = nn.Linear(512 * block.expansion, num_classes)\n        \n        # Ініціалізація ваг\n        self._initialize_weights()\n    \n    def _make_layer(self, block, out_channels, num_blocks, stride):\n        \"\"\"\n        Створює послідовність residual блоків\n        \n        Перший блок може мати stride=2 для downsampling,\n        решта мають stride=1\n        \"\"\"\n        downsample = None\n        \n        # Якщо розміри змінюються, потрібен downsample для skip connection\n        if stride != 1 or self.in_channels != out_channels * block.expansion:\n            downsample = nn.Sequential(\n                nn.Conv2d(\n                    self.in_channels, \n                    out_channels * block.expansion,\n                    kernel_size=1, \n                    stride=stride, \n                    bias=False\n                ),\n                nn.BatchNorm2d(out_channels * block.expansion)\n            )\n        \n        layers = []\n        # Перший блок (може робити downsampling)\n        layers.append(block(self.in_channels, out_channels, stride, downsample))\n        self.in_channels = out_channels * block.expansion\n        \n        # Решта блоків (stride=1, розміри не змінюються)\n        for _ in range(1, num_blocks):\n            layers.append(block(self.in_channels, out_channels))\n        \n        return nn.Sequential(*layers)\n    \n    def _initialize_weights(self):\n        \"\"\"He ініціалізація для conv шарів\"\"\"\n        for m in self.modules():\n            if isinstance(m, nn.Conv2d):\n                nn.init.kaiming_normal_(m.weight, mode='fan_out', nonlinearity='relu')\n            elif isinstance(m, nn.BatchNorm2d):\n                nn.init.constant_(m.weight, 1)\n                nn.init.constant_(m.bias, 0)\n    \n    def forward(self, x):\n        # Initial block\n        x = self.conv1(x)\n        x = self.bn1(x)\n        x = self.relu(x)\n        x = self.maxpool(x)\n        \n        # Residual blocks\n        x = self.layer1(x)\n        x = self.layer2(x)\n        x = self.layer3(x)\n        x = self.layer4(x)\n        \n        # Classification head\n        x = self.avgpool(x)\n        x = torch.flatten(x, 1)\n        x = self.fc(x)\n        \n        return x\n\n\n# Фабричні функції для різних версій\ndef resnet18(num_classes=10, input_channels=3):\n    \"\"\"ResNet-18\"\"\"\n    return ResNet(BasicBlock, [2, 2, 2, 2], num_classes, input_channels)\n\ndef resnet34(num_classes=10, input_channels=3):\n    \"\"\"ResNet-34\"\"\"\n    return ResNet(BasicBlock, [3, 4, 6, 3], num_classes, input_channels)\n\ndef resnet50(num_classes=10, input_channels=3):\n    \"\"\"ResNet-50 з Bottleneck блоками\"\"\"\n    # Встановлюємо expansion для BottleneckBlock\n    BottleneckBlock.expansion = 4\n    return ResNet(BottleneckBlock, [3, 4, 6, 3], num_classes, input_channels)\n\n\n# Тестування архітектури\nprint(\"🏗️ Створення ResNet-18\\n\")\nmodel = resnet18(num_classes=10, input_channels=3)\n\n# Підрахунок параметрів\ntotal_params = sum(p.numel() for p in model.parameters())\ntrainable_params = sum(p.numel() for p in model.parameters() if p.requires_grad)\n\nprint(f\"Архітектура: ResNet-18\")\nprint(f\"├─ Загальна кількість параметрів: {total_params:,}\")\nprint(f\"├─ Тренувальні параметри: {trainable_params:,}\")\nprint(f\"└─ Розмір моделі: {total_params * 4 \u002F 1024 \u002F 1024:.2f} MB (float32)\\n\")\n\n# Тестовий прогін\nx = torch.randn(8, 3, 224, 224)  # [batch, channels, H, W]\nwith torch.no_grad():\n    output = model(x)\nprint(f\"📥 Вхід: {x.shape}\")\nprint(f\"📤 Вихід: {output.shape}\")\nprint(f\"✅ Forward pass успішний!\")\n",[5358,8558,8559,8576,8580,8585,8589,8594,8599,8604,8609,8614,8618,8622,8663,8667,8671,8676,8681,8686,8691,8695,8710,8720,8724,8729,8735,8776,8780,8791,8808,8839,8843,8848,8879,8906,8933,8961,8965,8970,8986,8998,9002,9007,9014,9018,9048,9052,9057,9061,9066,9071,9075,9083,9087,9092,9109,9114,9119,9127,9132,9143,9150,9159,9164,9169,9174,9178,9183,9188,9198,9205,9209,9214,9234,9244,9248,9255,9259,9272,9277,9291,9302,9327,9337,9346,9355,9359,9375,9380,9389,9398,9407,9416,9420,9425,9434,9443,9452,9461,9466,9472,9482,9492,9502,9507,9515,9520,9525,9531,9558,9564,9590,9595,9621,9627,9651,9656,9682,9688,9694,9703,9727,9732,9737,9743,9759,9781,9786,9792,9811,9836,9841,9855,9878,9901,9942,9947,9953,9978,9987,9993,10015,10037],{"__ignoreMap":5356},[5361,8560,8561,8563,8566,8568,8570,8572,8574],{"class":5363,"line":5364},[5361,8562,6378],{"class":6377},[5361,8564,8565],{"class":6381}," ResNet",[5361,8567,6385],{"class":6342},[5361,8569,6388],{"class":6381},[5361,8571,5198],{"class":6342},[5361,8573,6393],{"class":6381},[5361,8575,6396],{"class":6342},[5361,8577,8578],{"class":5363,"line":5370},[5361,8579,6402],{"class":6401},[5361,8581,8582],{"class":5363,"line":5376},[5361,8583,8584],{"class":6401},"    ResNet архітектура для класифікації зображень\n",[5361,8586,8587],{"class":5363,"line":5382},[5361,8588,6412],{"class":6401},[5361,8590,8591],{"class":5363,"line":5388},[5361,8592,8593],{"class":6401},"    Підтримує різні конфігурації:\n",[5361,8595,8596],{"class":5363,"line":5395},[5361,8597,8598],{"class":6401},"        - ResNet-18: [2, 2, 2, 2] BasicBlocks\n",[5361,8600,8601],{"class":5363,"line":5401},[5361,8602,8603],{"class":6401},"        - ResNet-34: [3, 4, 6, 3] BasicBlocks\n",[5361,8605,8606],{"class":5363,"line":5407},[5361,8607,8608],{"class":6401},"        - ResNet-50: [3, 4, 6, 3] BottleneckBlocks\n",[5361,8610,8611],{"class":5363,"line":5413},[5361,8612,8613],{"class":6401},"        - ResNet-101: [3, 4, 23, 3] BottleneckBlocks\n",[5361,8615,8616],{"class":5363,"line":5418},[5361,8617,6402],{"class":6401},[5361,8619,8620],{"class":5363,"line":5424},[5361,8621,6412],{"class":6342},[5361,8623,8624,8626,8628,8630,8632,8634,8637,8639,8642,8644,8647,8649,8652,8654,8657,8659,8661],{"class":5363,"line":5430},[5361,8625,6445],{"class":6377},[5361,8627,6449],{"class":6448},[5361,8629,6385],{"class":6342},[5361,8631,6455],{"class":6454},[5361,8633,6458],{"class":6342},[5361,8635,8636],{"class":6454},"block",[5361,8638,6458],{"class":6342},[5361,8640,8641],{"class":6454},"layers",[5361,8643,6458],{"class":6342},[5361,8645,8646],{"class":6454},"num_classes",[5361,8648,6474],{"class":6342},[5361,8650,8651],{"class":6477},"10",[5361,8653,6458],{"class":6342},[5361,8655,8656],{"class":6454},"input_channels",[5361,8658,6474],{"class":6342},[5361,8660,6580],{"class":6477},[5361,8662,6396],{"class":6342},[5361,8664,8665],{"class":5363,"line":5436},[5361,8666,6494],{"class":6401},[5361,8668,8669],{"class":5363,"line":5442},[5361,8670,6499],{"class":6401},[5361,8672,8673],{"class":5363,"line":5448},[5361,8674,8675],{"class":6401},"            block: BasicBlock або BottleneckBlock\n",[5361,8677,8678],{"class":5363,"line":5454},[5361,8679,8680],{"class":6401},"            layers: список кількості блоків у кожному шарі, наприклад [2, 2, 2, 2]\n",[5361,8682,8683],{"class":5363,"line":5460},[5361,8684,8685],{"class":6401},"            num_classes: кількість класів для класифікації\n",[5361,8687,8688],{"class":5363,"line":5466},[5361,8689,8690],{"class":6401},"            input_channels: кількість каналів входу (3 для RGB, 1 для grayscale)\n",[5361,8692,8693],{"class":5363,"line":5472},[5361,8694,6494],{"class":6401},[5361,8696,8697,8699,8702,8704,8706,8708],{"class":5363,"line":5478},[5361,8698,6528],{"class":6381},[5361,8700,8701],{"class":6342},"(ResNet, ",[5361,8703,6455],{"class":6377},[5361,8705,5725],{"class":6342},[5361,8707,6538],{"class":6448},[5361,8709,6541],{"class":6342},[5361,8711,8712,8714,8717],{"class":5363,"line":5484},[5361,8713,6557],{"class":6377},[5361,8715,8716],{"class":6342},".in_channels = ",[5361,8718,8719],{"class":6477},"64\n",[5361,8721,8722],{"class":5363,"line":5489},[5361,8723,6546],{"class":6342},[5361,8725,8726],{"class":5363,"line":5495},[5361,8727,8728],{"class":6551},"        # Початковий convolutional block\n",[5361,8730,8731,8733],{"class":5363,"line":5501},[5361,8732,6557],{"class":6377},[5361,8734,6560],{"class":6342},[5361,8736,8737,8740,8742,8744,8746,8748,8751,8753,8755,8757,8759,8761,8764,8766,8768,8770,8772,8774],{"class":5363,"line":5507},[5361,8738,8739],{"class":6342},"            input_channels, ",[5361,8741,6977],{"class":6477},[5361,8743,6458],{"class":6342},[5361,8745,7131],{"class":6454},[5361,8747,6474],{"class":6342},[5361,8749,8750],{"class":6477},"7",[5361,8752,6458],{"class":6342},[5361,8754,6471],{"class":6454},[5361,8756,6474],{"class":6342},[5361,8758,7144],{"class":6477},[5361,8760,6458],{"class":6342},[5361,8762,8763],{"class":6454},"padding",[5361,8765,6474],{"class":6342},[5361,8767,6580],{"class":6477},[5361,8769,6458],{"class":6342},[5361,8771,7149],{"class":6454},[5361,8773,6474],{"class":6342},[5361,8775,6693],{"class":6377},[5361,8777,8778],{"class":5363,"line":5513},[5361,8779,6623],{"class":6342},[5361,8781,8782,8784,8787,8789],{"class":5363,"line":5519},[5361,8783,6557],{"class":6377},[5361,8785,8786],{"class":6342},".bn1 = nn.BatchNorm2d(",[5361,8788,6977],{"class":6477},[5361,8790,6953],{"class":6342},[5361,8792,8793,8795,8798,8801,8803,8806],{"class":5363,"line":5525},[5361,8794,6557],{"class":6377},[5361,8796,8797],{"class":6342},".relu = nn.ReLU(",[5361,8799,8800],{"class":6454},"inplace",[5361,8802,6474],{"class":6342},[5361,8804,8805],{"class":6377},"True",[5361,8807,6953],{"class":6342},[5361,8809,8810,8812,8815,8817,8819,8821,8823,8825,8827,8829,8831,8833,8835,8837],{"class":5363,"line":5531},[5361,8811,6557],{"class":6377},[5361,8813,8814],{"class":6342},".maxpool = nn.MaxPool2d(",[5361,8816,7131],{"class":6454},[5361,8818,6474],{"class":6342},[5361,8820,6580],{"class":6477},[5361,8822,6458],{"class":6342},[5361,8824,6471],{"class":6454},[5361,8826,6474],{"class":6342},[5361,8828,7144],{"class":6477},[5361,8830,6458],{"class":6342},[5361,8832,8763],{"class":6454},[5361,8834,6474],{"class":6342},[5361,8836,6326],{"class":6477},[5361,8838,6953],{"class":6342},[5361,8840,8841],{"class":5363,"line":5537},[5361,8842,6546],{"class":6342},[5361,8844,8845],{"class":5363,"line":5542},[5361,8846,8847],{"class":6551},"        # Residual layers (4 групи блоків)\n",[5361,8849,8850,8852,8855,8857,8860,8862,8865,8868,8871,8873,8875,8877],{"class":5363,"line":5548},[5361,8851,6557],{"class":6377},[5361,8853,8854],{"class":6342},".layer1 = ",[5361,8856,6455],{"class":6377},[5361,8858,8859],{"class":6342},"._make_layer(block, ",[5361,8861,6977],{"class":6477},[5361,8863,8864],{"class":6342},", layers[",[5361,8866,8867],{"class":6477},"0",[5361,8869,8870],{"class":6342},"], ",[5361,8872,6471],{"class":6454},[5361,8874,6474],{"class":6342},[5361,8876,6326],{"class":6477},[5361,8878,6953],{"class":6342},[5361,8880,8881,8883,8886,8888,8890,8892,8894,8896,8898,8900,8902,8904],{"class":5363,"line":5553},[5361,8882,6557],{"class":6377},[5361,8884,8885],{"class":6342},".layer2 = ",[5361,8887,6455],{"class":6377},[5361,8889,8859],{"class":6342},[5361,8891,7126],{"class":6477},[5361,8893,8864],{"class":6342},[5361,8895,6326],{"class":6477},[5361,8897,8870],{"class":6342},[5361,8899,6471],{"class":6454},[5361,8901,6474],{"class":6342},[5361,8903,7144],{"class":6477},[5361,8905,6953],{"class":6342},[5361,8907,8908,8910,8913,8915,8917,8919,8921,8923,8925,8927,8929,8931],{"class":5363,"line":5559},[5361,8909,6557],{"class":6377},[5361,8911,8912],{"class":6342},".layer3 = ",[5361,8914,6455],{"class":6377},[5361,8916,8859],{"class":6342},[5361,8918,7900],{"class":6477},[5361,8920,8864],{"class":6342},[5361,8922,7144],{"class":6477},[5361,8924,8870],{"class":6342},[5361,8926,6471],{"class":6454},[5361,8928,6474],{"class":6342},[5361,8930,7144],{"class":6477},[5361,8932,6953],{"class":6342},[5361,8934,8935,8937,8940,8942,8944,8947,8949,8951,8953,8955,8957,8959],{"class":5363,"line":5565},[5361,8936,6557],{"class":6377},[5361,8938,8939],{"class":6342},".layer4 = ",[5361,8941,6455],{"class":6377},[5361,8943,8859],{"class":6342},[5361,8945,8946],{"class":6477},"512",[5361,8948,8864],{"class":6342},[5361,8950,6580],{"class":6477},[5361,8952,8870],{"class":6342},[5361,8954,6471],{"class":6454},[5361,8956,6474],{"class":6342},[5361,8958,7144],{"class":6477},[5361,8960,6953],{"class":6342},[5361,8962,8963],{"class":5363,"line":5571},[5361,8964,6546],{"class":6342},[5361,8966,8967],{"class":5363,"line":5577},[5361,8968,8969],{"class":6551},"        # Глобальне усереднення та класифікатор\n",[5361,8971,8972,8974,8977,8979,8981,8983],{"class":5363,"line":5583},[5361,8973,6557],{"class":6377},[5361,8975,8976],{"class":6342},".avgpool = nn.AdaptiveAvgPool2d((",[5361,8978,6326],{"class":6477},[5361,8980,6458],{"class":6342},[5361,8982,6326],{"class":6477},[5361,8984,8985],{"class":6342},"))\n",[5361,8987,8988,8990,8993,8995],{"class":5363,"line":5588},[5361,8989,6557],{"class":6377},[5361,8991,8992],{"class":6342},".fc = nn.Linear(",[5361,8994,8946],{"class":6477},[5361,8996,8997],{"class":6342}," * block.expansion, num_classes)\n",[5361,8999,9000],{"class":5363,"line":5594},[5361,9001,6546],{"class":6342},[5361,9003,9004],{"class":5363,"line":5600},[5361,9005,9006],{"class":6551},"        # Ініціалізація ваг\n",[5361,9008,9009,9011],{"class":5363,"line":5606},[5361,9010,6557],{"class":6377},[5361,9012,9013],{"class":6342},"._initialize_weights()\n",[5361,9015,9016],{"class":5363,"line":5612},[5361,9017,6412],{"class":6342},[5361,9019,9020,9022,9025,9027,9029,9031,9033,9035,9037,9039,9042,9044,9046],{"class":5363,"line":5618},[5361,9021,6445],{"class":6377},[5361,9023,9024],{"class":6448}," _make_layer",[5361,9026,6385],{"class":6342},[5361,9028,6455],{"class":6454},[5361,9030,6458],{"class":6342},[5361,9032,8636],{"class":6454},[5361,9034,6458],{"class":6342},[5361,9036,6466],{"class":6454},[5361,9038,6458],{"class":6342},[5361,9040,9041],{"class":6454},"num_blocks",[5361,9043,6458],{"class":6342},[5361,9045,6471],{"class":6454},[5361,9047,6396],{"class":6342},[5361,9049,9050],{"class":5363,"line":5623},[5361,9051,6494],{"class":6401},[5361,9053,9054],{"class":5363,"line":5628},[5361,9055,9056],{"class":6401},"        Створює послідовність residual блоків\n",[5361,9058,9059],{"class":5363,"line":5634},[5361,9060,6546],{"class":6401},[5361,9062,9063],{"class":5363,"line":5640},[5361,9064,9065],{"class":6401},"        Перший блок може мати stride=2 для downsampling,\n",[5361,9067,9068],{"class":5363,"line":5646},[5361,9069,9070],{"class":6401},"        решта мають stride=1\n",[5361,9072,9073],{"class":5363,"line":5652},[5361,9074,6494],{"class":6401},[5361,9076,9077,9080],{"class":5363,"line":5657},[5361,9078,9079],{"class":6342},"        downsample = ",[5361,9081,9082],{"class":6377},"None\n",[5361,9084,9085],{"class":5363,"line":5662},[5361,9086,6546],{"class":6342},[5361,9088,9089],{"class":5363,"line":6757},[5361,9090,9091],{"class":6551},"        # Якщо розміри змінюються, потрібен downsample для skip connection\n",[5361,9093,9094,9096,9099,9101,9104,9106],{"class":5363,"line":6763},[5361,9095,6840],{"class":6338},[5361,9097,9098],{"class":6342}," stride != ",[5361,9100,6326],{"class":6477},[5361,9102,9103],{"class":6377}," or",[5361,9105,6843],{"class":6377},[5361,9107,9108],{"class":6342},".in_channels != out_channels * block.expansion:\n",[5361,9110,9111],{"class":5363,"line":6768},[5361,9112,9113],{"class":6342},"            downsample = nn.Sequential(\n",[5361,9115,9116],{"class":5363,"line":6774},[5361,9117,9118],{"class":6342},"                nn.Conv2d(\n",[5361,9120,9121,9124],{"class":5363,"line":6785},[5361,9122,9123],{"class":6377},"                    self",[5361,9125,9126],{"class":6342},".in_channels, \n",[5361,9128,9129],{"class":5363,"line":6795},[5361,9130,9131],{"class":6342},"                    out_channels * block.expansion,\n",[5361,9133,9134,9137,9139,9141],{"class":5363,"line":6801},[5361,9135,9136],{"class":6454},"                    kernel_size",[5361,9138,6474],{"class":6342},[5361,9140,6326],{"class":6477},[5361,9142,6583],{"class":6342},[5361,9144,9145,9148],{"class":5363,"line":6806},[5361,9146,9147],{"class":6454},"                    stride",[5361,9149,7622],{"class":6342},[5361,9151,9152,9155,9157],{"class":5363,"line":6816},[5361,9153,9154],{"class":6454},"                    bias",[5361,9156,6474],{"class":6342},[5361,9158,6693],{"class":6377},[5361,9160,9161],{"class":5363,"line":6826},[5361,9162,9163],{"class":6342},"                ),\n",[5361,9165,9166],{"class":5363,"line":6831},[5361,9167,9168],{"class":6342},"                nn.BatchNorm2d(out_channels * block.expansion)\n",[5361,9170,9171],{"class":5363,"line":6837},[5361,9172,9173],{"class":6342},"            )\n",[5361,9175,9176],{"class":5363,"line":6861},[5361,9177,6546],{"class":6342},[5361,9179,9180],{"class":5363,"line":6872},[5361,9181,9182],{"class":6342},"        layers = []\n",[5361,9184,9185],{"class":5363,"line":6877},[5361,9186,9187],{"class":6551},"        # Перший блок (може робити downsampling)\n",[5361,9189,9190,9193,9195],{"class":5363,"line":6883},[5361,9191,9192],{"class":6342},"        layers.append(block(",[5361,9194,6455],{"class":6377},[5361,9196,9197],{"class":6342},".in_channels, out_channels, stride, downsample))\n",[5361,9199,9200,9202],{"class":5363,"line":6889},[5361,9201,6557],{"class":6377},[5361,9203,9204],{"class":6342},".in_channels = out_channels * block.expansion\n",[5361,9206,9207],{"class":5363,"line":6894},[5361,9208,6546],{"class":6342},[5361,9210,9211],{"class":5363,"line":6900},[5361,9212,9213],{"class":6551},"        # Решта блоків (stride=1, розміри не змінюються)\n",[5361,9215,9216,9219,9222,9224,9227,9229,9231],{"class":5363,"line":6905},[5361,9217,9218],{"class":6338},"        for",[5361,9220,9221],{"class":6342}," _ ",[5361,9223,7938],{"class":6338},[5361,9225,9226],{"class":6448}," range",[5361,9228,6385],{"class":6342},[5361,9230,6326],{"class":6477},[5361,9232,9233],{"class":6342},", num_blocks):\n",[5361,9235,9236,9239,9241],{"class":5363,"line":6910},[5361,9237,9238],{"class":6342},"            layers.append(block(",[5361,9240,6455],{"class":6377},[5361,9242,9243],{"class":6342},".in_channels, out_channels))\n",[5361,9245,9246],{"class":5363,"line":6919},[5361,9247,6546],{"class":6342},[5361,9249,9250,9252],{"class":5363,"line":6924},[5361,9251,6913],{"class":6338},[5361,9253,9254],{"class":6342}," nn.Sequential(*layers)\n",[5361,9256,9257],{"class":5363,"line":6929},[5361,9258,6412],{"class":6342},[5361,9260,9261,9263,9266,9268,9270],{"class":5363,"line":6935},[5361,9262,6445],{"class":6377},[5361,9264,9265],{"class":6448}," _initialize_weights",[5361,9267,6385],{"class":6342},[5361,9269,6455],{"class":6454},[5361,9271,6396],{"class":6342},[5361,9273,9274],{"class":5363,"line":6956},[5361,9275,9276],{"class":6401},"        \"\"\"He ініціалізація для conv шарів\"\"\"\n",[5361,9278,9279,9281,9284,9286,9288],{"class":5363,"line":6961},[5361,9280,9218],{"class":6338},[5361,9282,9283],{"class":6342}," m ",[5361,9285,7938],{"class":6338},[5361,9287,6843],{"class":6377},[5361,9289,9290],{"class":6342},".modules():\n",[5361,9292,9293,9296,9299],{"class":5363,"line":6967},[5361,9294,9295],{"class":6338},"            if",[5361,9297,9298],{"class":6448}," isinstance",[5361,9300,9301],{"class":6342},"(m, nn.Conv2d):\n",[5361,9303,9304,9307,9310,9312,9315,9317,9320,9322,9325],{"class":5363,"line":6998},[5361,9305,9306],{"class":6342},"                nn.init.kaiming_normal_(m.weight, ",[5361,9308,9309],{"class":6454},"mode",[5361,9311,6474],{"class":6342},[5361,9313,9314],{"class":6401},"'fan_out'",[5361,9316,6458],{"class":6342},[5361,9318,9319],{"class":6454},"nonlinearity",[5361,9321,6474],{"class":6342},[5361,9323,9324],{"class":6401},"'relu'",[5361,9326,6953],{"class":6342},[5361,9328,9329,9332,9334],{"class":5363,"line":7026},[5361,9330,9331],{"class":6338},"            elif",[5361,9333,9298],{"class":6448},[5361,9335,9336],{"class":6342},"(m, nn.BatchNorm2d):\n",[5361,9338,9339,9342,9344],{"class":5363,"line":7032},[5361,9340,9341],{"class":6342},"                nn.init.constant_(m.weight, ",[5361,9343,6326],{"class":6477},[5361,9345,6953],{"class":6342},[5361,9347,9348,9351,9353],{"class":5363,"line":7058},[5361,9349,9350],{"class":6342},"                nn.init.constant_(m.bias, ",[5361,9352,8867],{"class":6477},[5361,9354,6953],{"class":6342},[5361,9356,9357],{"class":5363,"line":7081},[5361,9358,6412],{"class":6342},[5361,9360,9361,9363,9365,9367,9369,9371,9373],{"class":5363,"line":7099},[5361,9362,6445],{"class":6377},[5361,9364,6738],{"class":6448},[5361,9366,6385],{"class":6342},[5361,9368,6455],{"class":6454},[5361,9370,6458],{"class":6342},[5361,9372,6747],{"class":6454},[5361,9374,6396],{"class":6342},[5361,9376,9377],{"class":5363,"line":7104},[5361,9378,9379],{"class":6551},"        # Initial block\n",[5361,9381,9382,9385,9387],{"class":5363,"line":7110},[5361,9383,9384],{"class":6342},"        x = ",[5361,9386,6455],{"class":6377},[5361,9388,6782],{"class":6342},[5361,9390,9391,9393,9395],{"class":5363,"line":7116},[5361,9392,9384],{"class":6342},[5361,9394,6455],{"class":6377},[5361,9396,9397],{"class":6342},".bn1(x)\n",[5361,9399,9400,9402,9404],{"class":5363,"line":7159},[5361,9401,9384],{"class":6342},[5361,9403,6455],{"class":6377},[5361,9405,9406],{"class":6342},".relu(x)\n",[5361,9408,9409,9411,9413],{"class":5363,"line":7169},[5361,9410,9384],{"class":6342},[5361,9412,6455],{"class":6377},[5361,9414,9415],{"class":6342},".maxpool(x)\n",[5361,9417,9418],{"class":5363,"line":7174},[5361,9419,6546],{"class":6342},[5361,9421,9422],{"class":5363,"line":7209},[5361,9423,9424],{"class":6551},"        # Residual blocks\n",[5361,9426,9427,9429,9431],{"class":5363,"line":7231},[5361,9428,9384],{"class":6342},[5361,9430,6455],{"class":6377},[5361,9432,9433],{"class":6342},".layer1(x)\n",[5361,9435,9436,9438,9440],{"class":5363,"line":7237},[5361,9437,9384],{"class":6342},[5361,9439,6455],{"class":6377},[5361,9441,9442],{"class":6342},".layer2(x)\n",[5361,9444,9445,9447,9449],{"class":5363,"line":7259},[5361,9446,9384],{"class":6342},[5361,9448,6455],{"class":6377},[5361,9450,9451],{"class":6342},".layer3(x)\n",[5361,9453,9454,9456,9458],{"class":5363,"line":7281},[5361,9455,9384],{"class":6342},[5361,9457,6455],{"class":6377},[5361,9459,9460],{"class":6342},".layer4(x)\n",[5361,9462,9464],{"class":5363,"line":9463},98,[5361,9465,6546],{"class":6342},[5361,9467,9469],{"class":5363,"line":9468},99,[5361,9470,9471],{"class":6551},"        # Classification head\n",[5361,9473,9475,9477,9479],{"class":5363,"line":9474},100,[5361,9476,9384],{"class":6342},[5361,9478,6455],{"class":6377},[5361,9480,9481],{"class":6342},".avgpool(x)\n",[5361,9483,9485,9488,9490],{"class":5363,"line":9484},101,[5361,9486,9487],{"class":6342},"        x = torch.flatten(x, ",[5361,9489,6326],{"class":6477},[5361,9491,6953],{"class":6342},[5361,9493,9495,9497,9499],{"class":5363,"line":9494},102,[5361,9496,9384],{"class":6342},[5361,9498,6455],{"class":6377},[5361,9500,9501],{"class":6342},".fc(x)\n",[5361,9503,9505],{"class":5363,"line":9504},103,[5361,9506,6546],{"class":6342},[5361,9508,9510,9512],{"class":5363,"line":9509},104,[5361,9511,6913],{"class":6338},[5361,9513,9514],{"class":6342}," x\n",[5361,9516,9518],{"class":5363,"line":9517},105,[5361,9519,5392],{"emptyLinePlaceholder":5391},[5361,9521,9523],{"class":5363,"line":9522},106,[5361,9524,5392],{"emptyLinePlaceholder":5391},[5361,9526,9528],{"class":5363,"line":9527},107,[5361,9529,9530],{"class":6551},"# Фабричні функції для різних версій\n",[5361,9532,9534,9537,9540,9542,9544,9546,9548,9550,9552,9554,9556],{"class":5363,"line":9533},108,[5361,9535,9536],{"class":6377},"def",[5361,9538,9539],{"class":6448}," resnet18",[5361,9541,6385],{"class":6342},[5361,9543,8646],{"class":6454},[5361,9545,6474],{"class":6342},[5361,9547,8651],{"class":6477},[5361,9549,6458],{"class":6342},[5361,9551,8656],{"class":6454},[5361,9553,6474],{"class":6342},[5361,9555,6580],{"class":6477},[5361,9557,6396],{"class":6342},[5361,9559,9561],{"class":5363,"line":9560},109,[5361,9562,9563],{"class":6401},"    \"\"\"ResNet-18\"\"\"\n",[5361,9565,9567,9570,9573,9575,9577,9579,9581,9583,9585,9587],{"class":5363,"line":9566},110,[5361,9568,9569],{"class":6338},"    return",[5361,9571,9572],{"class":6342}," ResNet(BasicBlock, [",[5361,9574,7144],{"class":6477},[5361,9576,6458],{"class":6342},[5361,9578,7144],{"class":6477},[5361,9580,6458],{"class":6342},[5361,9582,7144],{"class":6477},[5361,9584,6458],{"class":6342},[5361,9586,7144],{"class":6477},[5361,9588,9589],{"class":6342},"], num_classes, input_channels)\n",[5361,9591,9593],{"class":5363,"line":9592},111,[5361,9594,5392],{"emptyLinePlaceholder":5391},[5361,9596,9598,9600,9603,9605,9607,9609,9611,9613,9615,9617,9619],{"class":5363,"line":9597},112,[5361,9599,9536],{"class":6377},[5361,9601,9602],{"class":6448}," resnet34",[5361,9604,6385],{"class":6342},[5361,9606,8646],{"class":6454},[5361,9608,6474],{"class":6342},[5361,9610,8651],{"class":6477},[5361,9612,6458],{"class":6342},[5361,9614,8656],{"class":6454},[5361,9616,6474],{"class":6342},[5361,9618,6580],{"class":6477},[5361,9620,6396],{"class":6342},[5361,9622,9624],{"class":5363,"line":9623},113,[5361,9625,9626],{"class":6401},"    \"\"\"ResNet-34\"\"\"\n",[5361,9628,9630,9632,9634,9636,9638,9640,9642,9645,9647,9649],{"class":5363,"line":9629},114,[5361,9631,9569],{"class":6338},[5361,9633,9572],{"class":6342},[5361,9635,6580],{"class":6477},[5361,9637,6458],{"class":6342},[5361,9639,7004],{"class":6477},[5361,9641,6458],{"class":6342},[5361,9643,9644],{"class":6477},"6",[5361,9646,6458],{"class":6342},[5361,9648,6580],{"class":6477},[5361,9650,9589],{"class":6342},[5361,9652,9654],{"class":5363,"line":9653},115,[5361,9655,5392],{"emptyLinePlaceholder":5391},[5361,9657,9659,9661,9664,9666,9668,9670,9672,9674,9676,9678,9680],{"class":5363,"line":9658},116,[5361,9660,9536],{"class":6377},[5361,9662,9663],{"class":6448}," resnet50",[5361,9665,6385],{"class":6342},[5361,9667,8646],{"class":6454},[5361,9669,6474],{"class":6342},[5361,9671,8651],{"class":6477},[5361,9673,6458],{"class":6342},[5361,9675,8656],{"class":6454},[5361,9677,6474],{"class":6342},[5361,9679,6580],{"class":6477},[5361,9681,6396],{"class":6342},[5361,9683,9685],{"class":5363,"line":9684},117,[5361,9686,9687],{"class":6401},"    \"\"\"ResNet-50 з Bottleneck блоками\"\"\"\n",[5361,9689,9691],{"class":5363,"line":9690},118,[5361,9692,9693],{"class":6551},"    # Встановлюємо expansion для BottleneckBlock\n",[5361,9695,9697,9700],{"class":5363,"line":9696},119,[5361,9698,9699],{"class":6342},"    BottleneckBlock.expansion = ",[5361,9701,9702],{"class":6477},"4\n",[5361,9704,9706,9708,9711,9713,9715,9717,9719,9721,9723,9725],{"class":5363,"line":9705},120,[5361,9707,9569],{"class":6338},[5361,9709,9710],{"class":6342}," ResNet(BottleneckBlock, [",[5361,9712,6580],{"class":6477},[5361,9714,6458],{"class":6342},[5361,9716,7004],{"class":6477},[5361,9718,6458],{"class":6342},[5361,9720,9644],{"class":6477},[5361,9722,6458],{"class":6342},[5361,9724,6580],{"class":6477},[5361,9726,9589],{"class":6342},[5361,9728,9730],{"class":5363,"line":9729},121,[5361,9731,5392],{"emptyLinePlaceholder":5391},[5361,9733,9735],{"class":5363,"line":9734},122,[5361,9736,5392],{"emptyLinePlaceholder":5391},[5361,9738,9740],{"class":5363,"line":9739},123,[5361,9741,9742],{"class":6551},"# Тестування архітектури\n",[5361,9744,9746,9748,9750,9753,9755,9757],{"class":5363,"line":9745},124,[5361,9747,6938],{"class":6448},[5361,9749,6385],{"class":6342},[5361,9751,9752],{"class":6401},"\"🏗️ Створення ResNet-18",[5361,9754,6947],{"class":6946},[5361,9756,6950],{"class":6401},[5361,9758,6953],{"class":6342},[5361,9760,9762,9765,9767,9769,9771,9773,9775,9777,9779],{"class":5363,"line":9761},125,[5361,9763,9764],{"class":6342},"model = resnet18(",[5361,9766,8646],{"class":6454},[5361,9768,6474],{"class":6342},[5361,9770,8651],{"class":6477},[5361,9772,6458],{"class":6342},[5361,9774,8656],{"class":6454},[5361,9776,6474],{"class":6342},[5361,9778,6580],{"class":6477},[5361,9780,6953],{"class":6342},[5361,9782,9784],{"class":5363,"line":9783},126,[5361,9785,5392],{"emptyLinePlaceholder":5391},[5361,9787,9789],{"class":5363,"line":9788},127,[5361,9790,9791],{"class":6551},"# Підрахунок параметрів\n",[5361,9793,9795,9798,9800,9802,9804,9806,9808],{"class":5363,"line":9794},128,[5361,9796,9797],{"class":6342},"total_params = ",[5361,9799,7926],{"class":6448},[5361,9801,7929],{"class":6342},[5361,9803,7932],{"class":6338},[5361,9805,7935],{"class":6342},[5361,9807,7938],{"class":6338},[5361,9809,9810],{"class":6342}," model.parameters())\n",[5361,9812,9814,9817,9819,9821,9823,9825,9827,9830,9833],{"class":5363,"line":9813},129,[5361,9815,9816],{"class":6342},"trainable_params = ",[5361,9818,7926],{"class":6448},[5361,9820,7929],{"class":6342},[5361,9822,7932],{"class":6338},[5361,9824,7935],{"class":6342},[5361,9826,7938],{"class":6338},[5361,9828,9829],{"class":6342}," model.parameters() ",[5361,9831,9832],{"class":6338},"if",[5361,9834,9835],{"class":6342}," p.requires_grad)\n",[5361,9837,9839],{"class":5363,"line":9838},130,[5361,9840,5392],{"emptyLinePlaceholder":5391},[5361,9842,9844,9846,9848,9850,9853],{"class":5363,"line":9843},131,[5361,9845,6938],{"class":6448},[5361,9847,6385],{"class":6342},[5361,9849,7039],{"class":6377},[5361,9851,9852],{"class":6401},"\"Архітектура: ResNet-18\"",[5361,9854,6953],{"class":6342},[5361,9856,9858,9860,9862,9864,9867,9869,9872,9874,9876],{"class":5363,"line":9857},132,[5361,9859,6938],{"class":6448},[5361,9861,6385],{"class":6342},[5361,9863,7039],{"class":6377},[5361,9865,9866],{"class":6401},"\"├─ Загальна кількість параметрів: ",[5361,9868,7045],{"class":6377},[5361,9870,9871],{"class":6342},"total_params",[5361,9873,8033],{"class":6377},[5361,9875,6950],{"class":6401},[5361,9877,6953],{"class":6342},[5361,9879,9881,9883,9885,9887,9890,9892,9895,9897,9899],{"class":5363,"line":9880},133,[5361,9882,6938],{"class":6448},[5361,9884,6385],{"class":6342},[5361,9886,7039],{"class":6377},[5361,9888,9889],{"class":6401},"\"├─ Тренувальні параметри: ",[5361,9891,7045],{"class":6377},[5361,9893,9894],{"class":6342},"trainable_params",[5361,9896,8033],{"class":6377},[5361,9898,6950],{"class":6401},[5361,9900,6953],{"class":6342},[5361,9902,9904,9906,9908,9910,9913,9915,9918,9920,9923,9926,9928,9930,9933,9936,9938,9940],{"class":5363,"line":9903},134,[5361,9905,6938],{"class":6448},[5361,9907,6385],{"class":6342},[5361,9909,7039],{"class":6377},[5361,9911,9912],{"class":6401},"\"└─ Розмір моделі: ",[5361,9914,7045],{"class":6377},[5361,9916,9917],{"class":6342},"total_params * ",[5361,9919,7004],{"class":6477},[5361,9921,9922],{"class":6342}," \u002F ",[5361,9924,9925],{"class":6477},"1024",[5361,9927,9922],{"class":6342},[5361,9929,9925],{"class":6477},[5361,9931,9932],{"class":6377},":.2f}",[5361,9934,9935],{"class":6401}," MB (float32)",[5361,9937,6947],{"class":6946},[5361,9939,6950],{"class":6401},[5361,9941,6953],{"class":6342},[5361,9943,9945],{"class":5363,"line":9944},135,[5361,9946,5392],{"emptyLinePlaceholder":5391},[5361,9948,9950],{"class":5363,"line":9949},136,[5361,9951,9952],{"class":6551},"# Тестовий прогін\n",[5361,9954,9956,9959,9961,9963,9965,9967,9970,9972,9974,9976],{"class":5363,"line":9955},137,[5361,9957,9958],{"class":6342},"x = torch.randn(",[5361,9960,5241],{"class":6477},[5361,9962,6458],{"class":6342},[5361,9964,6580],{"class":6477},[5361,9966,6458],{"class":6342},[5361,9968,9969],{"class":6477},"224",[5361,9971,6458],{"class":6342},[5361,9973,9969],{"class":6477},[5361,9975,7020],{"class":6342},[5361,9977,7023],{"class":6551},[5361,9979,9981,9984],{"class":5363,"line":9980},138,[5361,9982,9983],{"class":6338},"with",[5361,9985,9986],{"class":6342}," torch.no_grad():\n",[5361,9988,9990],{"class":5363,"line":9989},139,[5361,9991,9992],{"class":6342},"    output = model(x)\n",[5361,9994,9996,9998,10000,10002,10004,10006,10009,10011,10013],{"class":5363,"line":9995},140,[5361,9997,6938],{"class":6448},[5361,9999,6385],{"class":6342},[5361,10001,7039],{"class":6377},[5361,10003,7042],{"class":6401},[5361,10005,7045],{"class":6377},[5361,10007,10008],{"class":6342},"x.shape",[5361,10010,7051],{"class":6377},[5361,10012,6950],{"class":6401},[5361,10014,6953],{"class":6342},[5361,10016,10018,10020,10022,10024,10026,10028,10031,10033,10035],{"class":5363,"line":10017},141,[5361,10019,6938],{"class":6448},[5361,10021,6385],{"class":6342},[5361,10023,7039],{"class":6377},[5361,10025,7067],{"class":6401},[5361,10027,7045],{"class":6377},[5361,10029,10030],{"class":6342},"output.shape",[5361,10032,7051],{"class":6377},[5361,10034,6950],{"class":6401},[5361,10036,6953],{"class":6342},[5361,10038,10040,10042,10044,10046,10049],{"class":5363,"line":10039},142,[5361,10041,6938],{"class":6448},[5361,10043,6385],{"class":6342},[5361,10045,7039],{"class":6377},[5361,10047,10048],{"class":6401},"\"✅ Forward pass успішний!\"",[5361,10050,6953],{"class":6342},[7294,10052,10053,10059,10111,10116],{"v-slot:output":5356},[5191,10054,10055,10056],{},"🏗️ ",[5195,10057,10058],{},"Створення ResNet-18",[5203,10060,10061,10071],{},[5206,10062,10063],{},[5209,10064,10065,10068],{},[5212,10066,10067],{"align":5214},"Метрика",[5212,10069,10070],{"align":5214},"Значення",[5226,10072,10073,10082,10092,10101],{},[5209,10074,10075,10079],{},[5231,10076,10077],{"align":5214},[5195,10078,5218],{},[5231,10080,10081],{"align":5214},"ResNet-18",[5209,10083,10084,10089],{},[5231,10085,10086],{"align":5214},[5195,10087,10088],{},"Загальна кількість параметрів",[5231,10090,10091],{"align":5214},"11,689,512",[5209,10093,10094,10099],{},[5231,10095,10096],{"align":5214},[5195,10097,10098],{},"Тренувальні параметри",[5231,10100,10091],{"align":5214},[5209,10102,10103,10108],{},[5231,10104,10105],{"align":5214},[5195,10106,10107],{},"Розмір моделі",[5231,10109,10110],{"align":5214},"44.59 MB (float32)",[5191,10112,10113],{},[5195,10114,10115],{},"Тестовий прогін:",[5747,10117,10118,10123,10129],{},[5750,10119,7312,10120],{},[5358,10121,10122],{},"torch.Size([8, 3, 224, 224])",[5750,10124,7318,10125,10128],{},[5358,10126,10127],{},"torch.Size([8, 10])"," (логіти для 10 класів)",[5750,10130,10131],{},"✅ Forward pass успішний!",[5191,10133,6259],{},[6282,10135,10136,10141,10144,10171],{},[5191,10137,10138],{},[5195,10139,10140],{},"Адаптація для малих зображень (CIFAR-10)",[5191,10142,10143],{},"Стандартна ResNet розроблена для ImageNet (224×224). Для CIFAR-10 (32×32) потрібні модифікації:",[10145,10146,10147,10159,10165],"ol",{},[5750,10148,10149,5826,10152,10155,10156],{},[5195,10150,10151],{},"Замінити conv1:",[5358,10153,10154],{},"kernel_size=7, stride=2"," → ",[5358,10157,10158],{},"kernel_size=3, stride=1",[5750,10160,10161,10164],{},[5195,10162,10163],{},"Прибрати maxpool:"," початковий MaxPool зменшує розмір з 56×56 до 28×28, для 32×32 це занадто агресивно",[5750,10166,10167,10170],{},[5195,10168,10169],{},"Зменшити кількість каналів:"," можна використовувати 32 замість 64 для економії пам'яті",[5191,10172,10173],{},"Ми це реалізуємо у наступному експерименті!",[5855,10175],{},[5186,10177,10179],{"id":10178},"best-practices-проєктування-глибоких-мереж","Best Practices проєктування глибоких мереж",[5191,10181,10182],{},"Перш ніж запустити повноцінний експеримент, розглянемо ключові принципи проєктування сучасних глибоких архітектур. Ці практики — результат десятиліть досліджень та промислового досвіду.",[5280,10184,10186],{"id":10185},"принцип-1-поступове-зменшення-просторових-розмірів","Принцип 1: Поступове зменшення просторових розмірів",[5191,10188,10189,10192],{},[5195,10190,10191],{},"Правило:"," Зменшуйте H×W поступово (удвічі на кожному етапі), одночасно збільшуючи кількість каналів.",[5288,10194,10195,10210],{},[5291,10196,10199,10207],{"icon":10197,"title":10198},"i-lucide-check","✅ Правильно (ResNet стиль)",[5351,10200,10205],{"className":10201,"code":10203,"language":10204},[10202],"language-text","Input: 224×224×3\n→ Conv: 112×112×64\n→ Layer1: 56×56×64\n→ Layer2: 28×28×128\n→ Layer3: 14×14×256\n→ Layer4: 7×7×512\n→ GAP: 1×1×512\n","text",[5358,10206,10203],{"__ignoreMap":5356},[5191,10208,10209],{},"Плавна зміна: кожен етап зменшує H×W у 2 рази, канали збільшуються у 2 рази. Обчислювальна складність залишається збалансованою.",[5291,10211,10214,10220],{"icon":10212,"title":10213},"i-lucide-x","❌ Неправильно",[5351,10215,10218],{"className":10216,"code":10217,"language":10204},[10202],"Input: 224×224×3\n→ Conv: 56×56×512  (занадто агресивне downsampling)\n→ Conv: 56×56×512  (застій — розмір не змінюється)\n→ Conv: 7×7×512    (різке падіння)\n",[5358,10219,10217],{"__ignoreMap":5356},[5191,10221,10222],{},"Проблеми: втрата просторової інформації на початку, нераціональне використання обчислень.",[5280,10224,10226],{"id":10225},"принцип-2-завжди-batchnorm-після-conv","Принцип 2: Завжди BatchNorm після Conv",[5191,10228,10229,10231],{},[5195,10230,10191],{}," Кожен конволюційний шар супроводжується BatchNorm (крім останнього перед skip connection у residual блоках).",[5351,10233,10235],{"className":6329,"code":10234,"language":6331,"meta":5356,"style":5356},"# Стандартний блок\nx = self.conv(x)\nx = self.bn(x)      # Завжди після Conv!\nx = F.relu(x)\n",[5358,10236,10237,10242,10252,10264],{"__ignoreMap":5356},[5361,10238,10239],{"class":5363,"line":5364},[5361,10240,10241],{"class":6551},"# Стандартний блок\n",[5361,10243,10244,10247,10249],{"class":5363,"line":5370},[5361,10245,10246],{"class":6342},"x = ",[5361,10248,6455],{"class":6377},[5361,10250,10251],{"class":6342},".conv(x)\n",[5361,10253,10254,10256,10258,10261],{"class":5363,"line":5376},[5361,10255,10246],{"class":6342},[5361,10257,6455],{"class":6377},[5361,10259,10260],{"class":6342},".bn(x)      ",[5361,10262,10263],{"class":6551},"# Завжди після Conv!\n",[5361,10265,10266],{"class":5363,"line":5382},[5361,10267,10268],{"class":6342},"x = F.relu(x)\n",[5191,10270,10271],{},[5195,10272,10273],{},"Чому:",[5747,10275,10276,10279,10282,10285],{},[5750,10277,10278],{},"Стабілізує розподіл активацій",[5750,10280,10281],{},"Дозволяє великі learning rates",[5750,10283,10284],{},"Діє як регуляризація",[5750,10286,10287,10288,10291],{},"Робить bias у Conv непотрібним (",[5358,10289,10290],{},"bias=False",")",[5736,10293,10294,10299],{},[5191,10295,10296],{},[5195,10297,10298],{},"Порядок операцій:",[5191,10300,10301],{},"Класичний порядок: Conv → BatchNorm → ReLU. Альтернатива (рідше): Conv → ReLU → BatchNorm. У ResNet використовується перший варіант, оскільки BN перед ReLU дозволяє кращий gradient flow.",[5280,10303,10305],{"id":10304},"принцип-3-relu-як-стандартна-активація","Принцип 3: ReLU як стандартна активація",[5191,10307,10308,10310],{},[5195,10309,10191],{}," Використовуйте ReLU для прихованих шарів, якщо немає специфічних причин обрати інше.",[5203,10312,10313,10329],{},[5206,10314,10315],{},[5209,10316,10317,10320,10323,10326],{},[5212,10318,10319],{"align":5214},"Активація",[5212,10321,10322],{"align":5214},"Переваги",[5212,10324,10325],{"align":5214},"Недоліки",[5212,10327,10328],{"align":5214},"Коли використовувати",[5226,10330,10331,10346,10362,10378],{},[5209,10332,10333,10337,10340,10343],{},[5231,10334,10335],{"align":5214},[5195,10336,8194],{},[5231,10338,10339],{"align":5214},"Швидка, проста, ефективна",[5231,10341,10342],{"align":5214},"Dying ReLU (нейрони \"вмирають\")",[5231,10344,10345],{"align":5214},"За замовчуванням для CNN",[5209,10347,10348,10353,10356,10359],{},[5231,10349,10350],{"align":5214},[5195,10351,10352],{},"Leaky ReLU",[5231,10354,10355],{"align":5214},"Вирішує dying ReLU",[5231,10357,10358],{"align":5214},"Трохи повільніша",[5231,10360,10361],{"align":5214},"Якщо спостерігається dying ReLU",[5209,10363,10364,10369,10372,10375],{},[5231,10365,10366],{"align":5214},[5195,10367,10368],{},"GELU",[5231,10370,10371],{"align":5214},"Smooth, probabilistic",[5231,10373,10374],{"align":5214},"Складніша, повільніша",[5231,10376,10377],{"align":5214},"Transformers, NLP",[5209,10379,10380,10385,10388,10391],{},[5231,10381,10382],{"align":5214},[5195,10383,10384],{},"Swish\u002FSiLU",[5231,10386,10387],{"align":5214},"Кращі результати на деяких задачах",[5231,10389,10390],{"align":5214},"Обчислювально дорожча",[5231,10392,10393],{"align":5214},"EfficientNet, сучасні архітектури",[5312,10395,10396,10401,10404],{},[5191,10397,10398],{},[5195,10399,10400],{},"Dying ReLU проблема:",[5191,10402,10403],{},"Якщо нейрон отримує від'ємні входи, ReLU виводить нулі. Градієнти також стають нулями, нейрон \"вмирає\" (ваги більше не оновлюються). Це трапляється при дуже великих learning rates або поганій ініціалізації.",[5191,10405,10406,10409],{},[5195,10407,10408],{},"Розв'язання:"," Використовуйте правильну ініціалізацію (He), помірні learning rates, або Leaky ReLU\u002FELU.",[5280,10411,10413],{"id":10412},"принцип-4-global-average-pooling-замість-flatten","Принцип 4: Global Average Pooling замість Flatten",[5191,10415,10416,10418],{},[5195,10417,10191],{}," Перед фінальним fully-connected шаром використовуйте Global Average Pooling (GAP), а не Flatten.",[10420,10421,10422,10486],"code-group",{},[5351,10423,10426],{"className":6329,"code":10424,"filename":10425,"language":6331,"meta":5356,"style":5356},"# AlexNet\u002FVGG стиль\nx = self.conv_layers(x)  # → [batch, 512, 7, 7]\nx = x.view(x.size(0), -1)  # Flatten → [batch, 25088]\nx = self.fc1(x)  # 25088 → 4096 (величезна кількість параметрів!)\nx = self.fc2(x)  # 4096 → num_classes\n","❌ Старий підхід (Flatten)",[5358,10427,10428,10433,10445,10462,10474],{"__ignoreMap":5356},[5361,10429,10430],{"class":5363,"line":5364},[5361,10431,10432],{"class":6551},"# AlexNet\u002FVGG стиль\n",[5361,10434,10435,10437,10439,10442],{"class":5363,"line":5370},[5361,10436,10246],{"class":6342},[5361,10438,6455],{"class":6377},[5361,10440,10441],{"class":6342},".conv_layers(x)  ",[5361,10443,10444],{"class":6551},"# → [batch, 512, 7, 7]\n",[5361,10446,10447,10450,10452,10455,10457,10459],{"class":5363,"line":5376},[5361,10448,10449],{"class":6342},"x = x.view(x.size(",[5361,10451,8867],{"class":6477},[5361,10453,10454],{"class":6342},"), -",[5361,10456,6326],{"class":6477},[5361,10458,7020],{"class":6342},[5361,10460,10461],{"class":6551},"# Flatten → [batch, 25088]\n",[5361,10463,10464,10466,10468,10471],{"class":5363,"line":5382},[5361,10465,10246],{"class":6342},[5361,10467,6455],{"class":6377},[5361,10469,10470],{"class":6342},".fc1(x)  ",[5361,10472,10473],{"class":6551},"# 25088 → 4096 (величезна кількість параметрів!)\n",[5361,10475,10476,10478,10480,10483],{"class":5363,"line":5388},[5361,10477,10246],{"class":6342},[5361,10479,6455],{"class":6377},[5361,10481,10482],{"class":6342},".fc2(x)  ",[5361,10484,10485],{"class":6551},"# 4096 → num_classes\n",[5351,10487,10490],{"className":6329,"code":10488,"filename":10489,"language":6331,"meta":5356,"style":5356},"# ResNet стиль\nx = self.conv_layers(x)  # → [batch, 512, 7, 7]\nx = F.adaptive_avg_pool2d(x, (1, 1))  # → [batch, 512, 1, 1]\nx = torch.flatten(x, 1)  # → [batch, 512]\nx = self.fc(x)  # 512 → num_classes (мінімум параметрів)\n","✅ Сучасний підхід (GAP)",[5358,10491,10492,10497,10507,10524,10536],{"__ignoreMap":5356},[5361,10493,10494],{"class":5363,"line":5364},[5361,10495,10496],{"class":6551},"# ResNet стиль\n",[5361,10498,10499,10501,10503,10505],{"class":5363,"line":5370},[5361,10500,10246],{"class":6342},[5361,10502,6455],{"class":6377},[5361,10504,10441],{"class":6342},[5361,10506,10444],{"class":6551},[5361,10508,10509,10512,10514,10516,10518,10521],{"class":5363,"line":5376},[5361,10510,10511],{"class":6342},"x = F.adaptive_avg_pool2d(x, (",[5361,10513,6326],{"class":6477},[5361,10515,6458],{"class":6342},[5361,10517,6326],{"class":6477},[5361,10519,10520],{"class":6342},"))  ",[5361,10522,10523],{"class":6551},"# → [batch, 512, 1, 1]\n",[5361,10525,10526,10529,10531,10533],{"class":5363,"line":5382},[5361,10527,10528],{"class":6342},"x = torch.flatten(x, ",[5361,10530,6326],{"class":6477},[5361,10532,7020],{"class":6342},[5361,10534,10535],{"class":6551},"# → [batch, 512]\n",[5361,10537,10538,10540,10542,10545],{"class":5363,"line":5388},[5361,10539,10246],{"class":6342},[5361,10541,6455],{"class":6377},[5361,10543,10544],{"class":6342},".fc(x)  ",[5361,10546,10547],{"class":6551},"# 512 → num_classes (мінімум параметрів)\n",[5191,10549,10550],{},[5195,10551,10552],{},"Переваги GAP:",[5288,10554,10555,10572,10576],{},[5291,10556,10559,10560,10563,10564,10567,10568,10571],{"icon":10557,"title":10558},"i-lucide-minimize-2","📉 Зменшення параметрів","VGG-16 має ",[5195,10561,10562],{},"138 мільйонів параметрів",", з яких ",[5195,10565,10566],{},"123 мільйони"," (89%!) у fully-connected шарах. ResNet-50 має лише ",[5195,10569,10570],{},"25 мільйонів",", бо використовує GAP.",[5291,10573,10575],{"icon":5775,"title":10574},"🎯 Краща генералізація","Менше параметрів → менше ризик перенавчання. GAP виконує просторове усереднення, що діє як регуляризація.",[5291,10577,10580],{"icon":10578,"title":10579},"i-lucide-expand","📐 Гнучкість до розміру входу","Мережа з GAP може обробляти зображення різних розмірів без переучування. Flatten вимагає фіксованого розміру.",[5280,10582,10584],{"id":10583},"принцип-5-regularization-стратегія","Принцип 5: Regularization стратегія",[5191,10586,10587,10589],{},[5195,10588,10191],{}," Комбінуйте кілька технік регуляризації для оптимальних результатів.",[5191,10591,10592],{},"Рекомендований набір для CNN:",[10594,10595,10596,10600,10603,10607,10687,10691,10721,10730,10734,10744,10747],"steps",{},[5280,10597,10599],{"id":10598},"крок-1-batch-normalization-обовязково","Крок 1. Batch Normalization (обов'язково)",[5191,10601,10602],{},"Застосовуйте після кожного Conv шару. Це базова стабілізація.",[5280,10604,10606],{"id":10605},"крок-2-data-augmentation-обовязково-для-малих-датасетів","Крок 2. Data Augmentation (обов'язково для малих датасетів)",[5351,10608,10610],{"className":6329,"code":10609,"language":6331,"meta":5356,"style":5356},"transforms.Compose([\n    transforms.RandomHorizontalFlip(p=0.5),\n    transforms.RandomCrop(32, padding=4),\n    transforms.ColorJitter(brightness=0.2, contrast=0.2),\n    transforms.ToTensor(),\n    transforms.Normalize(mean, std)\n])\n",[5358,10611,10612,10617,10631,10648,10672,10677,10682],{"__ignoreMap":5356},[5361,10613,10614],{"class":5363,"line":5364},[5361,10615,10616],{"class":6342},"transforms.Compose([\n",[5361,10618,10619,10622,10624,10626,10629],{"class":5363,"line":5370},[5361,10620,10621],{"class":6342},"    transforms.RandomHorizontalFlip(",[5361,10623,5191],{"class":6454},[5361,10625,6474],{"class":6342},[5361,10627,10628],{"class":6477},"0.5",[5361,10630,7156],{"class":6342},[5361,10632,10633,10636,10638,10640,10642,10644,10646],{"class":5363,"line":5376},[5361,10634,10635],{"class":6342},"    transforms.RandomCrop(",[5361,10637,7013],{"class":6477},[5361,10639,6458],{"class":6342},[5361,10641,8763],{"class":6454},[5361,10643,6474],{"class":6342},[5361,10645,7004],{"class":6477},[5361,10647,7156],{"class":6342},[5361,10649,10650,10653,10656,10658,10661,10663,10666,10668,10670],{"class":5363,"line":5382},[5361,10651,10652],{"class":6342},"    transforms.ColorJitter(",[5361,10654,10655],{"class":6454},"brightness",[5361,10657,6474],{"class":6342},[5361,10659,10660],{"class":6477},"0.2",[5361,10662,6458],{"class":6342},[5361,10664,10665],{"class":6454},"contrast",[5361,10667,6474],{"class":6342},[5361,10669,10660],{"class":6477},[5361,10671,7156],{"class":6342},[5361,10673,10674],{"class":5363,"line":5388},[5361,10675,10676],{"class":6342},"    transforms.ToTensor(),\n",[5361,10678,10679],{"class":5363,"line":5395},[5361,10680,10681],{"class":6342},"    transforms.Normalize(mean, std)\n",[5361,10683,10684],{"class":5363,"line":5401},[5361,10685,10686],{"class":6342},"])\n",[5280,10688,10690],{"id":10689},"крок-3-weight-decay-l2-regularization","Крок 3. Weight Decay (L2 regularization)",[5351,10692,10694],{"className":6329,"code":10693,"language":6331,"meta":5356,"style":5356},"optimizer = torch.optim.Adam(model.parameters(), lr=0.001, weight_decay=1e-4)\n",[5358,10695,10696],{"__ignoreMap":5356},[5361,10697,10698,10701,10704,10706,10709,10711,10714,10716,10719],{"class":5363,"line":5364},[5361,10699,10700],{"class":6342},"optimizer = torch.optim.Adam(model.parameters(), ",[5361,10702,10703],{"class":6454},"lr",[5361,10705,6474],{"class":6342},[5361,10707,10708],{"class":6477},"0.001",[5361,10710,6458],{"class":6342},[5361,10712,10713],{"class":6454},"weight_decay",[5361,10715,6474],{"class":6342},[5361,10717,10718],{"class":6477},"1e-4",[5361,10720,6953],{"class":6342},[5191,10722,10723,10724,10726,10727,5198],{},"Типові значення: ",[5358,10725,10718],{}," до ",[5358,10728,10729],{},"1e-5",[5280,10731,10733],{"id":10732},"крок-4-dropout-опційно-залежно-від-архітектури","Крок 4. Dropout (опційно, залежно від архітектури)",[5191,10735,10736,10739,10740,10743],{},[5195,10737,10738],{},"Важливо:"," У ResNet часто ",[5195,10741,10742],{},"не використовують Dropout",", бо BatchNorm + Weight Decay + Data Augmentation достатньо. Dropout може навіть погіршити результати у residual архітектурах.",[5191,10745,10746],{},"Якщо використовуєте, розміщуйте перед фінальним FC:",[5351,10748,10750],{"className":6329,"code":10749,"language":6331,"meta":5356,"style":5356},"x = self.avgpool(x)\nx = torch.flatten(x, 1)\nx = F.dropout(x, p=0.5, training=self.training)  # Тільки тут\nx = self.fc(x)\n",[5358,10751,10752,10760,10768,10794],{"__ignoreMap":5356},[5361,10753,10754,10756,10758],{"class":5363,"line":5364},[5361,10755,10246],{"class":6342},[5361,10757,6455],{"class":6377},[5361,10759,9481],{"class":6342},[5361,10761,10762,10764,10766],{"class":5363,"line":5370},[5361,10763,10528],{"class":6342},[5361,10765,6326],{"class":6477},[5361,10767,6953],{"class":6342},[5361,10769,10770,10773,10775,10777,10779,10781,10784,10786,10788,10791],{"class":5363,"line":5376},[5361,10771,10772],{"class":6342},"x = F.dropout(x, ",[5361,10774,5191],{"class":6454},[5361,10776,6474],{"class":6342},[5361,10778,10628],{"class":6477},[5361,10780,6458],{"class":6342},[5361,10782,10783],{"class":6454},"training",[5361,10785,6474],{"class":6342},[5361,10787,6455],{"class":6377},[5361,10789,10790],{"class":6342},".training)  ",[5361,10792,10793],{"class":6551},"# Тільки тут\n",[5361,10795,10796,10798,10800],{"class":5363,"line":5382},[5361,10797,10246],{"class":6342},[5361,10799,6455],{"class":6377},[5361,10801,9501],{"class":6342},[6282,10803,10804,10809],{},[5191,10805,10806],{},[5195,10807,10808],{},"Емпіричне правило:",[5747,10810,10811,10817,10823],{},[5750,10812,10813,10816],{},[5195,10814,10815],{},"Малий датасет (\u003C10k зображень):"," Агресивна augmentation + Dropout 0.5 + Weight Decay 1e-4",[5750,10818,10819,10822],{},[5195,10820,10821],{},"Середній датасет (10k-100k):"," Помірна augmentation + Weight Decay 1e-4, Dropout опційно",[5750,10824,10825,10828],{},[5195,10826,10827],{},"Великий датасет (>100k):"," Мінімальна augmentation + Weight Decay 1e-5, Dropout зазвичай не потрібен",[5280,10830,10832],{"id":10831},"принцип-6-learning-rate-scheduling","Принцип 6: Learning Rate Scheduling",[5191,10834,10835,10837],{},[5195,10836,10191],{}," Зменшуйте learning rate під час навчання для досягнення кращої конвергенції.",[5191,10839,10840],{},"Популярні стратегії:",[10842,10843,10844,10912,10974,11038],"accordion",{},[10845,10846,10849,10852,10900,10906],"accordion-item",{"icon":10847,"label":10848},"i-lucide-trending-down","📉 Step Decay",[5191,10850,10851],{},"Зменшувати LR у фіксовані епохи:",[5351,10853,10855],{"className":6329,"code":10854,"language":6331,"meta":5356,"style":5356},"scheduler = torch.optim.lr_scheduler.StepLR(\n    optimizer, \n    step_size=30,  # Зменшувати кожні 30 епох\n    gamma=0.1      # Множник (LR *= 0.1)\n)\n",[5358,10856,10857,10862,10867,10883,10896],{"__ignoreMap":5356},[5361,10858,10859],{"class":5363,"line":5364},[5361,10860,10861],{"class":6342},"scheduler = torch.optim.lr_scheduler.StepLR(\n",[5361,10863,10864],{"class":5363,"line":5370},[5361,10865,10866],{"class":6342},"    optimizer, \n",[5361,10868,10869,10872,10874,10877,10880],{"class":5363,"line":5376},[5361,10870,10871],{"class":6454},"    step_size",[5361,10873,6474],{"class":6342},[5361,10875,10876],{"class":6477},"30",[5361,10878,10879],{"class":6342},",  ",[5361,10881,10882],{"class":6551},"# Зменшувати кожні 30 епох\n",[5361,10884,10885,10888,10890,10893],{"class":5363,"line":5382},[5361,10886,10887],{"class":6454},"    gamma",[5361,10889,6474],{"class":6342},[5361,10891,10892],{"class":6477},"0.1",[5361,10894,10895],{"class":6551},"      # Множник (LR *= 0.1)\n",[5361,10897,10898],{"class":5363,"line":5388},[5361,10899,6953],{"class":6342},[5191,10901,10902,10905],{},[5195,10903,10904],{},"Приклад:"," LR = 0.1 → епоха 30: 0.01 → епоха 60: 0.001",[5191,10907,10908,10911],{},[5195,10909,10910],{},"Коли:"," Простий базовий підхід, добре працює для більшості задач.",[10845,10913,10916,10919,10969],{"icon":10914,"label":10915},"i-lucide-calendar","🎯 MultiStepLR",[5191,10917,10918],{},"Зменшувати у конкретні епохи:",[5351,10920,10922],{"className":6329,"code":10921,"language":6331,"meta":5356,"style":5356},"scheduler = torch.optim.lr_scheduler.MultiStepLR(\n    optimizer,\n    milestones=[80, 120],  # Зменшити на епохах 80 та 120\n    gamma=0.1\n)\n",[5358,10923,10924,10929,10934,10956,10965],{"__ignoreMap":5356},[5361,10925,10926],{"class":5363,"line":5364},[5361,10927,10928],{"class":6342},"scheduler = torch.optim.lr_scheduler.MultiStepLR(\n",[5361,10930,10931],{"class":5363,"line":5370},[5361,10932,10933],{"class":6342},"    optimizer,\n",[5361,10935,10936,10939,10942,10945,10947,10950,10953],{"class":5363,"line":5376},[5361,10937,10938],{"class":6454},"    milestones",[5361,10940,10941],{"class":6342},"=[",[5361,10943,10944],{"class":6477},"80",[5361,10946,6458],{"class":6342},[5361,10948,10949],{"class":6477},"120",[5361,10951,10952],{"class":6342},"],  ",[5361,10954,10955],{"class":6551},"# Зменшити на епохах 80 та 120\n",[5361,10957,10958,10960,10962],{"class":5363,"line":5382},[5361,10959,10887],{"class":6454},[5361,10961,6474],{"class":6342},[5361,10963,10964],{"class":6477},"0.1\n",[5361,10966,10967],{"class":5363,"line":5388},[5361,10968,6953],{"class":6342},[5191,10970,10971,10973],{},[5195,10972,10910],{}," Коли ви знаєте, коли навчання сповільнюється (з досвіду або експериментів).",[10845,10975,10977,10980,11027,11033],{"icon":1348,"label":10976},"🔥 Cosine Annealing",[5191,10978,10979],{},"Плавне зменшення за косинусоїдою:",[5351,10981,10983],{"className":6329,"code":10982,"language":6331,"meta":5356,"style":5356},"scheduler = torch.optim.lr_scheduler.CosineAnnealingLR(\n    optimizer,\n    T_max=200,      # Загальна кількість епох\n    eta_min=1e-6    # Мінімальний LR\n)\n",[5358,10984,10985,10990,10994,11010,11023],{"__ignoreMap":5356},[5361,10986,10987],{"class":5363,"line":5364},[5361,10988,10989],{"class":6342},"scheduler = torch.optim.lr_scheduler.CosineAnnealingLR(\n",[5361,10991,10992],{"class":5363,"line":5370},[5361,10993,10933],{"class":6342},[5361,10995,10996,10999,11001,11004,11007],{"class":5363,"line":5376},[5361,10997,10998],{"class":6454},"    T_max",[5361,11000,6474],{"class":6342},[5361,11002,11003],{"class":6477},"200",[5361,11005,11006],{"class":6342},",      ",[5361,11008,11009],{"class":6551},"# Загальна кількість епох\n",[5361,11011,11012,11015,11017,11020],{"class":5363,"line":5382},[5361,11013,11014],{"class":6454},"    eta_min",[5361,11016,6474],{"class":6342},[5361,11018,11019],{"class":6477},"1e-6",[5361,11021,11022],{"class":6551},"    # Мінімальний LR\n",[5361,11024,11025],{"class":5363,"line":5388},[5361,11026,6953],{"class":6342},[5191,11028,11029,11032],{},[5195,11030,11031],{},"Переваги:"," Smooth, без різких стрибків. Популярний у сучасних дослідженнях.",[5191,11034,11035,11037],{},[5195,11036,10910],{}," Для довгих тренувань (100+ епох), сучасні архітектури.",[10845,11039,11041,11044,11139],{"icon":92,"label":11040},"⚡ ReduceLROnPlateau",[5191,11042,11043],{},"Адаптивне зменшення при зупинці прогресу:",[5351,11045,11047],{"className":6329,"code":11046,"language":6331,"meta":5356,"style":5356},"scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau(\n    optimizer,\n    mode='min',         # Моніторимо зменшення loss\n    factor=0.5,         # Множник (LR *= 0.5)\n    patience=10,        # Чекати 10 епох без покращення\n    verbose=True\n)\n\n# У тренувальному циклі:\nval_loss = validate(model, val_loader)\nscheduler.step(val_loss)  # Передаємо метрику\n",[5358,11048,11049,11054,11058,11074,11088,11103,11113,11117,11121,11126,11131],{"__ignoreMap":5356},[5361,11050,11051],{"class":5363,"line":5364},[5361,11052,11053],{"class":6342},"scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau(\n",[5361,11055,11056],{"class":5363,"line":5370},[5361,11057,10933],{"class":6342},[5361,11059,11060,11063,11065,11068,11071],{"class":5363,"line":5376},[5361,11061,11062],{"class":6454},"    mode",[5361,11064,6474],{"class":6342},[5361,11066,11067],{"class":6401},"'min'",[5361,11069,11070],{"class":6342},",         ",[5361,11072,11073],{"class":6551},"# Моніторимо зменшення loss\n",[5361,11075,11076,11079,11081,11083,11085],{"class":5363,"line":5382},[5361,11077,11078],{"class":6454},"    factor",[5361,11080,6474],{"class":6342},[5361,11082,10628],{"class":6477},[5361,11084,11070],{"class":6342},[5361,11086,11087],{"class":6551},"# Множник (LR *= 0.5)\n",[5361,11089,11090,11093,11095,11097,11100],{"class":5363,"line":5388},[5361,11091,11092],{"class":6454},"    patience",[5361,11094,6474],{"class":6342},[5361,11096,8651],{"class":6477},[5361,11098,11099],{"class":6342},",        ",[5361,11101,11102],{"class":6551},"# Чекати 10 епох без покращення\n",[5361,11104,11105,11108,11110],{"class":5363,"line":5395},[5361,11106,11107],{"class":6454},"    verbose",[5361,11109,6474],{"class":6342},[5361,11111,11112],{"class":6377},"True\n",[5361,11114,11115],{"class":5363,"line":5401},[5361,11116,6953],{"class":6342},[5361,11118,11119],{"class":5363,"line":5407},[5361,11120,5392],{"emptyLinePlaceholder":5391},[5361,11122,11123],{"class":5363,"line":5413},[5361,11124,11125],{"class":6551},"# У тренувальному циклі:\n",[5361,11127,11128],{"class":5363,"line":5418},[5361,11129,11130],{"class":6342},"val_loss = validate(model, val_loader)\n",[5361,11132,11133,11136],{"class":5363,"line":5424},[5361,11134,11135],{"class":6342},"scheduler.step(val_loss)  ",[5361,11137,11138],{"class":6551},"# Передаємо метрику\n",[5191,11140,11141,11143],{},[5195,11142,10910],{}," Коли важко заздалегідь передбачити, коли зменшувати LR. Автоматична адаптація.",[5855,11145],{},[5186,11147,11149],{"id":11148},"експеримент-plain-cnn-vs-resnet-на-cifar-10","Експеримент: Plain CNN vs ResNet на CIFAR-10",[5191,11151,11152],{},"Тепер проведемо фундаментальний експеримент, який відтворює класичне дослідження degradation problem. Порівняємо три архітектури:",[10145,11154,11155,11161,11167],{},[5750,11156,11157,11160],{},[5195,11158,11159],{},"Plain CNN-20:"," Класична 20-шарова мережа без skip connections",[5750,11162,11163,11166],{},[5195,11164,11165],{},"Plain CNN-56:"," Глибша 56-шарова версія (має показати degradation)",[5750,11168,11169,11172],{},[5195,11170,11171],{},"ResNet-56:"," 56-шарова з residual blocks (має вирішити проблему)",[5280,11174,11176],{"id":11175},"підготовка-датасету-cifar-10","Підготовка датасету CIFAR-10",[6314,11178,11179,11745],{"execution-count":7004},[5351,11180,11182],{"className":6329,"code":11181,"language":6331,"meta":5356,"style":5356},"import torch\nimport torch.nn as nn\nimport torch.optim as optim\nfrom torch.utils.data import DataLoader\nfrom torchvision import datasets, transforms\nimport matplotlib.pyplot as plt\nimport time\n\n# Перевірка доступності GPU\ndevice = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\nprint(f\"🖥️ Використовується пристрій: {device}\\n\")\n\n# Трансформації для CIFAR-10\ntransform_train = transforms.Compose([\n    transforms.RandomCrop(32, padding=4),\n    transforms.RandomHorizontalFlip(p=0.5),\n    transforms.ToTensor(),\n    transforms.Normalize(\n        mean=[0.4914, 0.4822, 0.4465],  # CIFAR-10 статистика\n        std=[0.2470, 0.2435, 0.2616]\n    )\n])\n\ntransform_test = transforms.Compose([\n    transforms.ToTensor(),\n    transforms.Normalize(\n        mean=[0.4914, 0.4822, 0.4465],\n        std=[0.2470, 0.2435, 0.2616]\n    )\n])\n\n# Завантаження CIFAR-10\ntrain_dataset = datasets.CIFAR10(\n    root='.\u002Fdata', train=True, download=True, transform=transform_train\n)\ntest_dataset = datasets.CIFAR10(\n    root='.\u002Fdata', train=False, download=True, transform=transform_test\n)\n\ntrain_loader = DataLoader(train_dataset, batch_size=128, shuffle=True, num_workers=2)\ntest_loader = DataLoader(test_dataset, batch_size=128, shuffle=False, num_workers=2)\n\nprint(f\"📊 Датасет CIFAR-10:\")\nprint(f\"├─ Тренувальних зображень: {len(train_dataset):,}\")\nprint(f\"├─ Тестових зображень: {len(test_dataset):,}\")\nprint(f\"├─ Класи: {train_dataset.classes}\")\nprint(f\"└─ Batch size: 128\\n\")\n",[5358,11183,11184,11190,11200,11212,11225,11237,11249,11256,11260,11265,11287,11311,11315,11320,11325,11341,11353,11357,11362,11387,11410,11415,11419,11423,11428,11432,11436,11455,11473,11477,11481,11485,11490,11495,11531,11535,11540,11571,11575,11579,11611,11640,11644,11657,11682,11706,11728],{"__ignoreMap":5356},[5361,11185,11186,11188],{"class":5363,"line":5364},[5361,11187,6339],{"class":6338},[5361,11189,6343],{"class":6342},[5361,11191,11192,11194,11196,11198],{"class":5363,"line":5370},[5361,11193,6339],{"class":6338},[5361,11195,6350],{"class":6342},[5361,11197,6353],{"class":6338},[5361,11199,6356],{"class":6342},[5361,11201,11202,11204,11207,11209],{"class":5363,"line":5376},[5361,11203,6339],{"class":6338},[5361,11205,11206],{"class":6342}," torch.optim ",[5361,11208,6353],{"class":6338},[5361,11210,11211],{"class":6342}," optim\n",[5361,11213,11214,11217,11220,11222],{"class":5363,"line":5382},[5361,11215,11216],{"class":6338},"from",[5361,11218,11219],{"class":6342}," torch.utils.data ",[5361,11221,6339],{"class":6338},[5361,11223,11224],{"class":6342}," DataLoader\n",[5361,11226,11227,11229,11232,11234],{"class":5363,"line":5388},[5361,11228,11216],{"class":6338},[5361,11230,11231],{"class":6342}," torchvision ",[5361,11233,6339],{"class":6338},[5361,11235,11236],{"class":6342}," datasets, transforms\n",[5361,11238,11239,11241,11244,11246],{"class":5363,"line":5395},[5361,11240,6339],{"class":6338},[5361,11242,11243],{"class":6342}," matplotlib.pyplot ",[5361,11245,6353],{"class":6338},[5361,11247,11248],{"class":6342}," plt\n",[5361,11250,11251,11253],{"class":5363,"line":5401},[5361,11252,6339],{"class":6338},[5361,11254,11255],{"class":6342}," time\n",[5361,11257,11258],{"class":5363,"line":5407},[5361,11259,5392],{"emptyLinePlaceholder":5391},[5361,11261,11262],{"class":5363,"line":5413},[5361,11263,11264],{"class":6551},"# Перевірка доступності GPU\n",[5361,11266,11267,11270,11273,11276,11279,11282,11285],{"class":5363,"line":5418},[5361,11268,11269],{"class":6342},"device = torch.device(",[5361,11271,11272],{"class":6401},"'cuda'",[5361,11274,11275],{"class":6338}," if",[5361,11277,11278],{"class":6342}," torch.cuda.is_available() ",[5361,11280,11281],{"class":6338},"else",[5361,11283,11284],{"class":6401}," 'cpu'",[5361,11286,6953],{"class":6342},[5361,11288,11289,11291,11293,11295,11298,11300,11303,11305,11307,11309],{"class":5363,"line":5424},[5361,11290,6938],{"class":6448},[5361,11292,6385],{"class":6342},[5361,11294,7039],{"class":6377},[5361,11296,11297],{"class":6401},"\"🖥️ Використовується пристрій: ",[5361,11299,7045],{"class":6377},[5361,11301,11302],{"class":6342},"device",[5361,11304,7051],{"class":6377},[5361,11306,6947],{"class":6946},[5361,11308,6950],{"class":6401},[5361,11310,6953],{"class":6342},[5361,11312,11313],{"class":5363,"line":5430},[5361,11314,5392],{"emptyLinePlaceholder":5391},[5361,11316,11317],{"class":5363,"line":5436},[5361,11318,11319],{"class":6551},"# Трансформації для CIFAR-10\n",[5361,11321,11322],{"class":5363,"line":5442},[5361,11323,11324],{"class":6342},"transform_train = transforms.Compose([\n",[5361,11326,11327,11329,11331,11333,11335,11337,11339],{"class":5363,"line":5448},[5361,11328,10635],{"class":6342},[5361,11330,7013],{"class":6477},[5361,11332,6458],{"class":6342},[5361,11334,8763],{"class":6454},[5361,11336,6474],{"class":6342},[5361,11338,7004],{"class":6477},[5361,11340,7156],{"class":6342},[5361,11342,11343,11345,11347,11349,11351],{"class":5363,"line":5454},[5361,11344,10621],{"class":6342},[5361,11346,5191],{"class":6454},[5361,11348,6474],{"class":6342},[5361,11350,10628],{"class":6477},[5361,11352,7156],{"class":6342},[5361,11354,11355],{"class":5363,"line":5460},[5361,11356,10676],{"class":6342},[5361,11358,11359],{"class":5363,"line":5466},[5361,11360,11361],{"class":6342},"    transforms.Normalize(\n",[5361,11363,11364,11367,11369,11372,11374,11377,11379,11382,11384],{"class":5363,"line":5472},[5361,11365,11366],{"class":6454},"        mean",[5361,11368,10941],{"class":6342},[5361,11370,11371],{"class":6477},"0.4914",[5361,11373,6458],{"class":6342},[5361,11375,11376],{"class":6477},"0.4822",[5361,11378,6458],{"class":6342},[5361,11380,11381],{"class":6477},"0.4465",[5361,11383,10952],{"class":6342},[5361,11385,11386],{"class":6551},"# CIFAR-10 статистика\n",[5361,11388,11389,11392,11394,11397,11399,11402,11404,11407],{"class":5363,"line":5478},[5361,11390,11391],{"class":6454},"        std",[5361,11393,10941],{"class":6342},[5361,11395,11396],{"class":6477},"0.2470",[5361,11398,6458],{"class":6342},[5361,11400,11401],{"class":6477},"0.2435",[5361,11403,6458],{"class":6342},[5361,11405,11406],{"class":6477},"0.2616",[5361,11408,11409],{"class":6342},"]\n",[5361,11411,11412],{"class":5363,"line":5484},[5361,11413,11414],{"class":6342},"    )\n",[5361,11416,11417],{"class":5363,"line":5489},[5361,11418,10686],{"class":6342},[5361,11420,11421],{"class":5363,"line":5495},[5361,11422,5392],{"emptyLinePlaceholder":5391},[5361,11424,11425],{"class":5363,"line":5501},[5361,11426,11427],{"class":6342},"transform_test = transforms.Compose([\n",[5361,11429,11430],{"class":5363,"line":5507},[5361,11431,10676],{"class":6342},[5361,11433,11434],{"class":5363,"line":5513},[5361,11435,11361],{"class":6342},[5361,11437,11438,11440,11442,11444,11446,11448,11450,11452],{"class":5363,"line":5519},[5361,11439,11366],{"class":6454},[5361,11441,10941],{"class":6342},[5361,11443,11371],{"class":6477},[5361,11445,6458],{"class":6342},[5361,11447,11376],{"class":6477},[5361,11449,6458],{"class":6342},[5361,11451,11381],{"class":6477},[5361,11453,11454],{"class":6342},"],\n",[5361,11456,11457,11459,11461,11463,11465,11467,11469,11471],{"class":5363,"line":5525},[5361,11458,11391],{"class":6454},[5361,11460,10941],{"class":6342},[5361,11462,11396],{"class":6477},[5361,11464,6458],{"class":6342},[5361,11466,11401],{"class":6477},[5361,11468,6458],{"class":6342},[5361,11470,11406],{"class":6477},[5361,11472,11409],{"class":6342},[5361,11474,11475],{"class":5363,"line":5531},[5361,11476,11414],{"class":6342},[5361,11478,11479],{"class":5363,"line":5537},[5361,11480,10686],{"class":6342},[5361,11482,11483],{"class":5363,"line":5542},[5361,11484,5392],{"emptyLinePlaceholder":5391},[5361,11486,11487],{"class":5363,"line":5548},[5361,11488,11489],{"class":6551},"# Завантаження CIFAR-10\n",[5361,11491,11492],{"class":5363,"line":5553},[5361,11493,11494],{"class":6342},"train_dataset = datasets.CIFAR10(\n",[5361,11496,11497,11500,11502,11505,11507,11510,11512,11514,11516,11519,11521,11523,11525,11528],{"class":5363,"line":5559},[5361,11498,11499],{"class":6454},"    root",[5361,11501,6474],{"class":6342},[5361,11503,11504],{"class":6401},"'.\u002Fdata'",[5361,11506,6458],{"class":6342},[5361,11508,11509],{"class":6454},"train",[5361,11511,6474],{"class":6342},[5361,11513,8805],{"class":6377},[5361,11515,6458],{"class":6342},[5361,11517,11518],{"class":6454},"download",[5361,11520,6474],{"class":6342},[5361,11522,8805],{"class":6377},[5361,11524,6458],{"class":6342},[5361,11526,11527],{"class":6454},"transform",[5361,11529,11530],{"class":6342},"=transform_train\n",[5361,11532,11533],{"class":5363,"line":5565},[5361,11534,6953],{"class":6342},[5361,11536,11537],{"class":5363,"line":5571},[5361,11538,11539],{"class":6342},"test_dataset = datasets.CIFAR10(\n",[5361,11541,11542,11544,11546,11548,11550,11552,11554,11556,11558,11560,11562,11564,11566,11568],{"class":5363,"line":5577},[5361,11543,11499],{"class":6454},[5361,11545,6474],{"class":6342},[5361,11547,11504],{"class":6401},[5361,11549,6458],{"class":6342},[5361,11551,11509],{"class":6454},[5361,11553,6474],{"class":6342},[5361,11555,6615],{"class":6377},[5361,11557,6458],{"class":6342},[5361,11559,11518],{"class":6454},[5361,11561,6474],{"class":6342},[5361,11563,8805],{"class":6377},[5361,11565,6458],{"class":6342},[5361,11567,11527],{"class":6454},[5361,11569,11570],{"class":6342},"=transform_test\n",[5361,11572,11573],{"class":5363,"line":5583},[5361,11574,6953],{"class":6342},[5361,11576,11577],{"class":5363,"line":5588},[5361,11578,5392],{"emptyLinePlaceholder":5391},[5361,11580,11581,11584,11587,11589,11591,11593,11596,11598,11600,11602,11605,11607,11609],{"class":5363,"line":5594},[5361,11582,11583],{"class":6342},"train_loader = DataLoader(train_dataset, ",[5361,11585,11586],{"class":6454},"batch_size",[5361,11588,6474],{"class":6342},[5361,11590,7126],{"class":6477},[5361,11592,6458],{"class":6342},[5361,11594,11595],{"class":6454},"shuffle",[5361,11597,6474],{"class":6342},[5361,11599,8805],{"class":6377},[5361,11601,6458],{"class":6342},[5361,11603,11604],{"class":6454},"num_workers",[5361,11606,6474],{"class":6342},[5361,11608,7144],{"class":6477},[5361,11610,6953],{"class":6342},[5361,11612,11613,11616,11618,11620,11622,11624,11626,11628,11630,11632,11634,11636,11638],{"class":5363,"line":5600},[5361,11614,11615],{"class":6342},"test_loader = DataLoader(test_dataset, ",[5361,11617,11586],{"class":6454},[5361,11619,6474],{"class":6342},[5361,11621,7126],{"class":6477},[5361,11623,6458],{"class":6342},[5361,11625,11595],{"class":6454},[5361,11627,6474],{"class":6342},[5361,11629,6615],{"class":6377},[5361,11631,6458],{"class":6342},[5361,11633,11604],{"class":6454},[5361,11635,6474],{"class":6342},[5361,11637,7144],{"class":6477},[5361,11639,6953],{"class":6342},[5361,11641,11642],{"class":5363,"line":5606},[5361,11643,5392],{"emptyLinePlaceholder":5391},[5361,11645,11646,11648,11650,11652,11655],{"class":5363,"line":5612},[5361,11647,6938],{"class":6448},[5361,11649,6385],{"class":6342},[5361,11651,7039],{"class":6377},[5361,11653,11654],{"class":6401},"\"📊 Датасет CIFAR-10:\"",[5361,11656,6953],{"class":6342},[5361,11658,11659,11661,11663,11665,11668,11670,11673,11676,11678,11680],{"class":5363,"line":5618},[5361,11660,6938],{"class":6448},[5361,11662,6385],{"class":6342},[5361,11664,7039],{"class":6377},[5361,11666,11667],{"class":6401},"\"├─ Тренувальних зображень: ",[5361,11669,7045],{"class":6377},[5361,11671,11672],{"class":6448},"len",[5361,11674,11675],{"class":6342},"(train_dataset)",[5361,11677,8033],{"class":6377},[5361,11679,6950],{"class":6401},[5361,11681,6953],{"class":6342},[5361,11683,11684,11686,11688,11690,11693,11695,11697,11700,11702,11704],{"class":5363,"line":5623},[5361,11685,6938],{"class":6448},[5361,11687,6385],{"class":6342},[5361,11689,7039],{"class":6377},[5361,11691,11692],{"class":6401},"\"├─ Тестових зображень: ",[5361,11694,7045],{"class":6377},[5361,11696,11672],{"class":6448},[5361,11698,11699],{"class":6342},"(test_dataset)",[5361,11701,8033],{"class":6377},[5361,11703,6950],{"class":6401},[5361,11705,6953],{"class":6342},[5361,11707,11708,11710,11712,11714,11717,11719,11722,11724,11726],{"class":5363,"line":5628},[5361,11709,6938],{"class":6448},[5361,11711,6385],{"class":6342},[5361,11713,7039],{"class":6377},[5361,11715,11716],{"class":6401},"\"├─ Класи: ",[5361,11718,7045],{"class":6377},[5361,11720,11721],{"class":6342},"train_dataset.classes",[5361,11723,7051],{"class":6377},[5361,11725,6950],{"class":6401},[5361,11727,6953],{"class":6342},[5361,11729,11730,11732,11734,11736,11739,11741,11743],{"class":5363,"line":5634},[5361,11731,6938],{"class":6448},[5361,11733,6385],{"class":6342},[5361,11735,7039],{"class":6377},[5361,11737,11738],{"class":6401},"\"└─ Batch size: 128",[5361,11740,6947],{"class":6946},[5361,11742,6950],{"class":6401},[5361,11744,6953],{"class":6342},[7294,11746,11747,11754,11759],{"v-slot:output":5356},[5191,11748,11749,11750,11753],{},"🖥️ ",[5195,11751,11752],{},"Використовується пристрій:"," cuda (NVIDIA RTX 3080)",[5191,11755,8115,11756],{},[5195,11757,11758],{},"Датасет CIFAR-10:",[5747,11760,11761,11764,11767,11773,11776],{},[5750,11762,11763],{},"Тренувальних зображень: 50,000",[5750,11765,11766],{},"Тестових зображень: 10,000",[5750,11768,11769,11770],{},"Класи: ",[5361,11771,11772],{},"'airplane', 'automobile', 'bird', 'cat', 'deer', 'dog', 'frog', 'horse', 'ship', 'truck'",[5750,11774,11775],{},"Batch size: 128",[5750,11777,11778],{},"Батчів за епоху: 391 (train), 79 (test)",[6314,11780,11781,11785],{"type":6316},[5280,11782,11784],{"id":11783},"реалізація-архітектур-для-порівняння","Реалізація архітектур для порівняння",[5191,11786,11787],{},"Створимо три моделі: Plain-20, Plain-56 та ResNet-56, адаптовані для малих зображень CIFAR-10.",[6314,11789,11791,13275],{"execution-count":11790},"5",[5351,11792,11794],{"className":6329,"code":11793,"language":6331,"meta":5356,"style":5356},"class PlainCNN(nn.Module):\n    \"\"\"\n    Класична CNN без skip connections для демонстрації degradation problem\n    Адаптована для CIFAR-10 (32×32)\n    \"\"\"\n    \n    def __init__(self, num_layers, num_classes=10):\n        super(PlainCNN, self).__init__()\n        \n        layers = []\n        in_channels = 3\n        out_channels = 16\n        \n        # Початковий Conv (без downsampling для малих зображень)\n        layers.append(nn.Conv2d(in_channels, out_channels, kernel_size=3, \n                               stride=1, padding=1, bias=False))\n        layers.append(nn.BatchNorm2d(out_channels))\n        layers.append(nn.ReLU(inplace=True))\n        \n        # Будуємо шари: кожні num_layers\u002F3 шарів подвоюємо канали\n        num_blocks = (num_layers - 2) \u002F\u002F 2  # -2 для початкового та фінального шару\n        blocks_per_stage = num_blocks \u002F\u002F 3\n        \n        for stage in range(3):\n            out_channels = 16 * (2 ** stage)\n            stride = 1 if stage == 0 else 2  # Перший stage без downsampling\n            \n            for block in range(blocks_per_stage):\n                if block == 0 and stage > 0:\n                    # Перший блок кожного stage (крім першого) робить downsampling\n                    layers.append(nn.Conv2d(out_channels \u002F\u002F 2, out_channels, \n                                           kernel_size=3, stride=2, padding=1, bias=False))\n                else:\n                    layers.append(nn.Conv2d(out_channels, out_channels, \n                                           kernel_size=3, stride=1, padding=1, bias=False))\n                layers.append(nn.BatchNorm2d(out_channels))\n                layers.append(nn.ReLU(inplace=True))\n        \n        self.features = nn.Sequential(*layers)\n        self.avgpool = nn.AdaptiveAvgPool2d((1, 1))\n        self.fc = nn.Linear(out_channels, num_classes)\n        \n        self._initialize_weights()\n    \n    def _initialize_weights(self):\n        for m in self.modules():\n            if isinstance(m, nn.Conv2d):\n                nn.init.kaiming_normal_(m.weight, mode='fan_out', nonlinearity='relu')\n            elif isinstance(m, nn.BatchNorm2d):\n                nn.init.constant_(m.weight, 1)\n                nn.init.constant_(m.bias, 0)\n    \n    def forward(self, x):\n        x = self.features(x)\n        x = self.avgpool(x)\n        x = torch.flatten(x, 1)\n        x = self.fc(x)\n        return x\n\n\nclass ResNetCIFAR(nn.Module):\n    \"\"\"\n    ResNet адаптований для CIFAR-10 (без початкового 7×7 conv та maxpool)\n    \"\"\"\n    \n    def __init__(self, block, num_blocks, num_classes=10):\n        super(ResNetCIFAR, self).__init__()\n        self.in_channels = 16\n        \n        # Початковий Conv (малий для 32×32)\n        self.conv1 = nn.Conv2d(3, 16, kernel_size=3, stride=1, padding=1, bias=False)\n        self.bn1 = nn.BatchNorm2d(16)\n        \n        # Три stage (16, 32, 64 каналів)\n        self.layer1 = self._make_layer(block, 16, num_blocks[0], stride=1)\n        self.layer2 = self._make_layer(block, 32, num_blocks[1], stride=2)\n        self.layer3 = self._make_layer(block, 64, num_blocks[2], stride=2)\n        \n        self.avgpool = nn.AdaptiveAvgPool2d((1, 1))\n        self.fc = nn.Linear(64, num_classes)\n        \n        self._initialize_weights()\n    \n    def _make_layer(self, block, out_channels, num_blocks, stride):\n        downsample = None\n        if stride != 1 or self.in_channels != out_channels:\n            downsample = nn.Sequential(\n                nn.Conv2d(self.in_channels, out_channels, kernel_size=1, \n                         stride=stride, bias=False),\n                nn.BatchNorm2d(out_channels)\n            )\n        \n        layers = []\n        layers.append(block(self.in_channels, out_channels, stride, downsample))\n        self.in_channels = out_channels\n        \n        for _ in range(1, num_blocks):\n            layers.append(block(self.in_channels, out_channels))\n        \n        return nn.Sequential(*layers)\n    \n    def _initialize_weights(self):\n        for m in self.modules():\n            if isinstance(m, nn.Conv2d):\n                nn.init.kaiming_normal_(m.weight, mode='fan_out', nonlinearity='relu')\n            elif isinstance(m, nn.BatchNorm2d):\n                nn.init.constant_(m.weight, 1)\n                nn.init.constant_(m.bias, 0)\n    \n    def forward(self, x):\n        x = F.relu(self.bn1(self.conv1(x)))\n        \n        x = self.layer1(x)\n        x = self.layer2(x)\n        x = self.layer3(x)\n        \n        x = self.avgpool(x)\n        x = torch.flatten(x, 1)\n        x = self.fc(x)\n        \n        return x\n\n\n# Створення моделей\nplain20 = PlainCNN(num_layers=20, num_classes=10).to(device)\nplain56 = PlainCNN(num_layers=56, num_classes=10).to(device)\nresnet56 = ResNetCIFAR(BasicBlock, [9, 9, 9], num_classes=10).to(device)\n\n# Статистика моделей\nmodels_info = {\n    'Plain-20': plain20,\n    'Plain-56': plain56,\n    'ResNet-56': resnet56\n}\n\nprint(\"🏗️ Архітектури для експерименту:\\n\")\nfor name, model in models_info.items():\n    params = sum(p.numel() for p in model.parameters())\n    print(f\"{name}:\")\n    print(f\"├─ Параметри: {params:,}\")\n    print(f\"└─ Розмір: {params * 4 \u002F 1024 \u002F 1024:.2f} MB\\n\")\n",[5358,11795,11796,11813,11817,11822,11827,11831,11835,11860,11875,11879,11883,11891,11899,11903,11908,11921,11948,11953,11966,11970,11975,11990,11997,12001,12018,12034,12057,12062,12077,12097,12102,12112,12147,12154,12159,12193,12198,12211,12215,12222,12236,12243,12247,12253,12257,12269,12281,12289,12309,12317,12325,12333,12337,12353,12362,12370,12378,12386,12392,12396,12400,12417,12421,12426,12430,12434,12462,12477,12485,12489,12494,12541,12551,12555,12560,12587,12613,12639,12643,12657,12668,12672,12678,12682,12710,12716,12731,12735,12753,12769,12774,12778,12782,12786,12794,12801,12805,12821,12829,12833,12839,12843,12855,12867,12875,12895,12903,12911,12919,12923,12939,12953,12957,12965,12973,12981,12985,12993,13001,13009,13013,13019,13023,13027,13032,13055,13077,13103,13107,13112,13117,13125,13133,13141,13145,13149,13164,13176,13193,13216,13238],{"__ignoreMap":5356},[5361,11797,11798,11800,11803,11805,11807,11809,11811],{"class":5363,"line":5364},[5361,11799,6378],{"class":6377},[5361,11801,11802],{"class":6381}," PlainCNN",[5361,11804,6385],{"class":6342},[5361,11806,6388],{"class":6381},[5361,11808,5198],{"class":6342},[5361,11810,6393],{"class":6381},[5361,11812,6396],{"class":6342},[5361,11814,11815],{"class":5363,"line":5370},[5361,11816,6402],{"class":6401},[5361,11818,11819],{"class":5363,"line":5376},[5361,11820,11821],{"class":6401},"    Класична CNN без skip connections для демонстрації degradation problem\n",[5361,11823,11824],{"class":5363,"line":5382},[5361,11825,11826],{"class":6401},"    Адаптована для CIFAR-10 (32×32)\n",[5361,11828,11829],{"class":5363,"line":5388},[5361,11830,6402],{"class":6401},[5361,11832,11833],{"class":5363,"line":5395},[5361,11834,6412],{"class":6342},[5361,11836,11837,11839,11841,11843,11845,11847,11850,11852,11854,11856,11858],{"class":5363,"line":5401},[5361,11838,6445],{"class":6377},[5361,11840,6449],{"class":6448},[5361,11842,6385],{"class":6342},[5361,11844,6455],{"class":6454},[5361,11846,6458],{"class":6342},[5361,11848,11849],{"class":6454},"num_layers",[5361,11851,6458],{"class":6342},[5361,11853,8646],{"class":6454},[5361,11855,6474],{"class":6342},[5361,11857,8651],{"class":6477},[5361,11859,6396],{"class":6342},[5361,11861,11862,11864,11867,11869,11871,11873],{"class":5363,"line":5407},[5361,11863,6528],{"class":6381},[5361,11865,11866],{"class":6342},"(PlainCNN, ",[5361,11868,6455],{"class":6377},[5361,11870,5725],{"class":6342},[5361,11872,6538],{"class":6448},[5361,11874,6541],{"class":6342},[5361,11876,11877],{"class":5363,"line":5413},[5361,11878,6546],{"class":6342},[5361,11880,11881],{"class":5363,"line":5418},[5361,11882,9182],{"class":6342},[5361,11884,11885,11888],{"class":5363,"line":5424},[5361,11886,11887],{"class":6342},"        in_channels = ",[5361,11889,11890],{"class":6477},"3\n",[5361,11892,11893,11896],{"class":5363,"line":5430},[5361,11894,11895],{"class":6342},"        out_channels = ",[5361,11897,11898],{"class":6477},"16\n",[5361,11900,11901],{"class":5363,"line":5436},[5361,11902,6546],{"class":6342},[5361,11904,11905],{"class":5363,"line":5442},[5361,11906,11907],{"class":6551},"        # Початковий Conv (без downsampling для малих зображень)\n",[5361,11909,11910,11913,11915,11917,11919],{"class":5363,"line":5448},[5361,11911,11912],{"class":6342},"        layers.append(nn.Conv2d(in_channels, out_channels, ",[5361,11914,7131],{"class":6454},[5361,11916,6474],{"class":6342},[5361,11918,6580],{"class":6477},[5361,11920,6583],{"class":6342},[5361,11922,11923,11926,11928,11930,11932,11934,11936,11938,11940,11942,11944,11946],{"class":5363,"line":5454},[5361,11924,11925],{"class":6454},"                               stride",[5361,11927,6474],{"class":6342},[5361,11929,6326],{"class":6477},[5361,11931,6458],{"class":6342},[5361,11933,8763],{"class":6454},[5361,11935,6474],{"class":6342},[5361,11937,6326],{"class":6477},[5361,11939,6458],{"class":6342},[5361,11941,7149],{"class":6454},[5361,11943,6474],{"class":6342},[5361,11945,6615],{"class":6377},[5361,11947,8985],{"class":6342},[5361,11949,11950],{"class":5363,"line":5460},[5361,11951,11952],{"class":6342},"        layers.append(nn.BatchNorm2d(out_channels))\n",[5361,11954,11955,11958,11960,11962,11964],{"class":5363,"line":5466},[5361,11956,11957],{"class":6342},"        layers.append(nn.ReLU(",[5361,11959,8800],{"class":6454},[5361,11961,6474],{"class":6342},[5361,11963,8805],{"class":6377},[5361,11965,8985],{"class":6342},[5361,11967,11968],{"class":5363,"line":5472},[5361,11969,6546],{"class":6342},[5361,11971,11972],{"class":5363,"line":5478},[5361,11973,11974],{"class":6551},"        # Будуємо шари: кожні num_layers\u002F3 шарів подвоюємо канали\n",[5361,11976,11977,11980,11982,11985,11987],{"class":5363,"line":5484},[5361,11978,11979],{"class":6342},"        num_blocks = (num_layers - ",[5361,11981,7144],{"class":6477},[5361,11983,11984],{"class":6342},") \u002F\u002F ",[5361,11986,7144],{"class":6477},[5361,11988,11989],{"class":6551},"  # -2 для початкового та фінального шару\n",[5361,11991,11992,11995],{"class":5363,"line":5489},[5361,11993,11994],{"class":6342},"        blocks_per_stage = num_blocks \u002F\u002F ",[5361,11996,11890],{"class":6477},[5361,11998,11999],{"class":5363,"line":5495},[5361,12000,6546],{"class":6342},[5361,12002,12003,12005,12008,12010,12012,12014,12016],{"class":5363,"line":5501},[5361,12004,9218],{"class":6338},[5361,12006,12007],{"class":6342}," stage ",[5361,12009,7938],{"class":6338},[5361,12011,9226],{"class":6448},[5361,12013,6385],{"class":6342},[5361,12015,6580],{"class":6477},[5361,12017,6396],{"class":6342},[5361,12019,12020,12023,12026,12029,12031],{"class":5363,"line":5507},[5361,12021,12022],{"class":6342},"            out_channels = ",[5361,12024,12025],{"class":6477},"16",[5361,12027,12028],{"class":6342}," * (",[5361,12030,7144],{"class":6477},[5361,12032,12033],{"class":6342}," ** stage)\n",[5361,12035,12036,12039,12041,12043,12046,12048,12051,12054],{"class":5363,"line":5513},[5361,12037,12038],{"class":6342},"            stride = ",[5361,12040,6326],{"class":6477},[5361,12042,11275],{"class":6338},[5361,12044,12045],{"class":6342}," stage == ",[5361,12047,8867],{"class":6477},[5361,12049,12050],{"class":6338}," else",[5361,12052,12053],{"class":6477}," 2",[5361,12055,12056],{"class":6551},"  # Перший stage без downsampling\n",[5361,12058,12059],{"class":5363,"line":5519},[5361,12060,12061],{"class":6342},"            \n",[5361,12063,12064,12067,12070,12072,12074],{"class":5363,"line":5525},[5361,12065,12066],{"class":6338},"            for",[5361,12068,12069],{"class":6342}," block ",[5361,12071,7938],{"class":6338},[5361,12073,9226],{"class":6448},[5361,12075,12076],{"class":6342},"(blocks_per_stage):\n",[5361,12078,12079,12082,12085,12087,12090,12093,12095],{"class":5363,"line":5531},[5361,12080,12081],{"class":6338},"                if",[5361,12083,12084],{"class":6342}," block == ",[5361,12086,8867],{"class":6477},[5361,12088,12089],{"class":6377}," and",[5361,12091,12092],{"class":6342}," stage > ",[5361,12094,8867],{"class":6477},[5361,12096,6858],{"class":6342},[5361,12098,12099],{"class":5363,"line":5537},[5361,12100,12101],{"class":6551},"                    # Перший блок кожного stage (крім першого) робить downsampling\n",[5361,12103,12104,12107,12109],{"class":5363,"line":5542},[5361,12105,12106],{"class":6342},"                    layers.append(nn.Conv2d(out_channels \u002F\u002F ",[5361,12108,7144],{"class":6477},[5361,12110,12111],{"class":6342},", out_channels, \n",[5361,12113,12114,12117,12119,12121,12123,12125,12127,12129,12131,12133,12135,12137,12139,12141,12143,12145],{"class":5363,"line":5548},[5361,12115,12116],{"class":6454},"                                           kernel_size",[5361,12118,6474],{"class":6342},[5361,12120,6580],{"class":6477},[5361,12122,6458],{"class":6342},[5361,12124,6471],{"class":6454},[5361,12126,6474],{"class":6342},[5361,12128,7144],{"class":6477},[5361,12130,6458],{"class":6342},[5361,12132,8763],{"class":6454},[5361,12134,6474],{"class":6342},[5361,12136,6326],{"class":6477},[5361,12138,6458],{"class":6342},[5361,12140,7149],{"class":6454},[5361,12142,6474],{"class":6342},[5361,12144,6615],{"class":6377},[5361,12146,8985],{"class":6342},[5361,12148,12149,12152],{"class":5363,"line":5553},[5361,12150,12151],{"class":6338},"                else",[5361,12153,6858],{"class":6342},[5361,12155,12156],{"class":5363,"line":5559},[5361,12157,12158],{"class":6342},"                    layers.append(nn.Conv2d(out_channels, out_channels, \n",[5361,12160,12161,12163,12165,12167,12169,12171,12173,12175,12177,12179,12181,12183,12185,12187,12189,12191],{"class":5363,"line":5565},[5361,12162,12116],{"class":6454},[5361,12164,6474],{"class":6342},[5361,12166,6580],{"class":6477},[5361,12168,6458],{"class":6342},[5361,12170,6471],{"class":6454},[5361,12172,6474],{"class":6342},[5361,12174,6326],{"class":6477},[5361,12176,6458],{"class":6342},[5361,12178,8763],{"class":6454},[5361,12180,6474],{"class":6342},[5361,12182,6326],{"class":6477},[5361,12184,6458],{"class":6342},[5361,12186,7149],{"class":6454},[5361,12188,6474],{"class":6342},[5361,12190,6615],{"class":6377},[5361,12192,8985],{"class":6342},[5361,12194,12195],{"class":5363,"line":5571},[5361,12196,12197],{"class":6342},"                layers.append(nn.BatchNorm2d(out_channels))\n",[5361,12199,12200,12203,12205,12207,12209],{"class":5363,"line":5577},[5361,12201,12202],{"class":6342},"                layers.append(nn.ReLU(",[5361,12204,8800],{"class":6454},[5361,12206,6474],{"class":6342},[5361,12208,8805],{"class":6377},[5361,12210,8985],{"class":6342},[5361,12212,12213],{"class":5363,"line":5583},[5361,12214,6546],{"class":6342},[5361,12216,12217,12219],{"class":5363,"line":5588},[5361,12218,6557],{"class":6377},[5361,12220,12221],{"class":6342},".features = nn.Sequential(*layers)\n",[5361,12223,12224,12226,12228,12230,12232,12234],{"class":5363,"line":5594},[5361,12225,6557],{"class":6377},[5361,12227,8976],{"class":6342},[5361,12229,6326],{"class":6477},[5361,12231,6458],{"class":6342},[5361,12233,6326],{"class":6477},[5361,12235,8985],{"class":6342},[5361,12237,12238,12240],{"class":5363,"line":5600},[5361,12239,6557],{"class":6377},[5361,12241,12242],{"class":6342},".fc = nn.Linear(out_channels, num_classes)\n",[5361,12244,12245],{"class":5363,"line":5606},[5361,12246,6546],{"class":6342},[5361,12248,12249,12251],{"class":5363,"line":5612},[5361,12250,6557],{"class":6377},[5361,12252,9013],{"class":6342},[5361,12254,12255],{"class":5363,"line":5618},[5361,12256,6412],{"class":6342},[5361,12258,12259,12261,12263,12265,12267],{"class":5363,"line":5623},[5361,12260,6445],{"class":6377},[5361,12262,9265],{"class":6448},[5361,12264,6385],{"class":6342},[5361,12266,6455],{"class":6454},[5361,12268,6396],{"class":6342},[5361,12270,12271,12273,12275,12277,12279],{"class":5363,"line":5628},[5361,12272,9218],{"class":6338},[5361,12274,9283],{"class":6342},[5361,12276,7938],{"class":6338},[5361,12278,6843],{"class":6377},[5361,12280,9290],{"class":6342},[5361,12282,12283,12285,12287],{"class":5363,"line":5634},[5361,12284,9295],{"class":6338},[5361,12286,9298],{"class":6448},[5361,12288,9301],{"class":6342},[5361,12290,12291,12293,12295,12297,12299,12301,12303,12305,12307],{"class":5363,"line":5640},[5361,12292,9306],{"class":6342},[5361,12294,9309],{"class":6454},[5361,12296,6474],{"class":6342},[5361,12298,9314],{"class":6401},[5361,12300,6458],{"class":6342},[5361,12302,9319],{"class":6454},[5361,12304,6474],{"class":6342},[5361,12306,9324],{"class":6401},[5361,12308,6953],{"class":6342},[5361,12310,12311,12313,12315],{"class":5363,"line":5646},[5361,12312,9331],{"class":6338},[5361,12314,9298],{"class":6448},[5361,12316,9336],{"class":6342},[5361,12318,12319,12321,12323],{"class":5363,"line":5652},[5361,12320,9341],{"class":6342},[5361,12322,6326],{"class":6477},[5361,12324,6953],{"class":6342},[5361,12326,12327,12329,12331],{"class":5363,"line":5657},[5361,12328,9350],{"class":6342},[5361,12330,8867],{"class":6477},[5361,12332,6953],{"class":6342},[5361,12334,12335],{"class":5363,"line":5662},[5361,12336,6412],{"class":6342},[5361,12338,12339,12341,12343,12345,12347,12349,12351],{"class":5363,"line":6757},[5361,12340,6445],{"class":6377},[5361,12342,6738],{"class":6448},[5361,12344,6385],{"class":6342},[5361,12346,6455],{"class":6454},[5361,12348,6458],{"class":6342},[5361,12350,6747],{"class":6454},[5361,12352,6396],{"class":6342},[5361,12354,12355,12357,12359],{"class":5363,"line":6763},[5361,12356,9384],{"class":6342},[5361,12358,6455],{"class":6377},[5361,12360,12361],{"class":6342},".features(x)\n",[5361,12363,12364,12366,12368],{"class":5363,"line":6768},[5361,12365,9384],{"class":6342},[5361,12367,6455],{"class":6377},[5361,12369,9481],{"class":6342},[5361,12371,12372,12374,12376],{"class":5363,"line":6774},[5361,12373,9487],{"class":6342},[5361,12375,6326],{"class":6477},[5361,12377,6953],{"class":6342},[5361,12379,12380,12382,12384],{"class":5363,"line":6785},[5361,12381,9384],{"class":6342},[5361,12383,6455],{"class":6377},[5361,12385,9501],{"class":6342},[5361,12387,12388,12390],{"class":5363,"line":6795},[5361,12389,6913],{"class":6338},[5361,12391,9514],{"class":6342},[5361,12393,12394],{"class":5363,"line":6801},[5361,12395,5392],{"emptyLinePlaceholder":5391},[5361,12397,12398],{"class":5363,"line":6806},[5361,12399,5392],{"emptyLinePlaceholder":5391},[5361,12401,12402,12404,12407,12409,12411,12413,12415],{"class":5363,"line":6816},[5361,12403,6378],{"class":6377},[5361,12405,12406],{"class":6381}," ResNetCIFAR",[5361,12408,6385],{"class":6342},[5361,12410,6388],{"class":6381},[5361,12412,5198],{"class":6342},[5361,12414,6393],{"class":6381},[5361,12416,6396],{"class":6342},[5361,12418,12419],{"class":5363,"line":6826},[5361,12420,6402],{"class":6401},[5361,12422,12423],{"class":5363,"line":6831},[5361,12424,12425],{"class":6401},"    ResNet адаптований для CIFAR-10 (без початкового 7×7 conv та maxpool)\n",[5361,12427,12428],{"class":5363,"line":6837},[5361,12429,6402],{"class":6401},[5361,12431,12432],{"class":5363,"line":6861},[5361,12433,6412],{"class":6342},[5361,12435,12436,12438,12440,12442,12444,12446,12448,12450,12452,12454,12456,12458,12460],{"class":5363,"line":6872},[5361,12437,6445],{"class":6377},[5361,12439,6449],{"class":6448},[5361,12441,6385],{"class":6342},[5361,12443,6455],{"class":6454},[5361,12445,6458],{"class":6342},[5361,12447,8636],{"class":6454},[5361,12449,6458],{"class":6342},[5361,12451,9041],{"class":6454},[5361,12453,6458],{"class":6342},[5361,12455,8646],{"class":6454},[5361,12457,6474],{"class":6342},[5361,12459,8651],{"class":6477},[5361,12461,6396],{"class":6342},[5361,12463,12464,12466,12469,12471,12473,12475],{"class":5363,"line":6877},[5361,12465,6528],{"class":6381},[5361,12467,12468],{"class":6342},"(ResNetCIFAR, ",[5361,12470,6455],{"class":6377},[5361,12472,5725],{"class":6342},[5361,12474,6538],{"class":6448},[5361,12476,6541],{"class":6342},[5361,12478,12479,12481,12483],{"class":5363,"line":6883},[5361,12480,6557],{"class":6377},[5361,12482,8716],{"class":6342},[5361,12484,11898],{"class":6477},[5361,12486,12487],{"class":5363,"line":6889},[5361,12488,6546],{"class":6342},[5361,12490,12491],{"class":5363,"line":6894},[5361,12492,12493],{"class":6551},"        # Початковий Conv (малий для 32×32)\n",[5361,12495,12496,12498,12501,12503,12505,12507,12509,12511,12513,12515,12517,12519,12521,12523,12525,12527,12529,12531,12533,12535,12537,12539],{"class":5363,"line":6900},[5361,12497,6557],{"class":6377},[5361,12499,12500],{"class":6342},".conv1 = nn.Conv2d(",[5361,12502,6580],{"class":6477},[5361,12504,6458],{"class":6342},[5361,12506,12025],{"class":6477},[5361,12508,6458],{"class":6342},[5361,12510,7131],{"class":6454},[5361,12512,6474],{"class":6342},[5361,12514,6580],{"class":6477},[5361,12516,6458],{"class":6342},[5361,12518,6471],{"class":6454},[5361,12520,6474],{"class":6342},[5361,12522,6326],{"class":6477},[5361,12524,6458],{"class":6342},[5361,12526,8763],{"class":6454},[5361,12528,6474],{"class":6342},[5361,12530,6326],{"class":6477},[5361,12532,6458],{"class":6342},[5361,12534,7149],{"class":6454},[5361,12536,6474],{"class":6342},[5361,12538,6615],{"class":6377},[5361,12540,6953],{"class":6342},[5361,12542,12543,12545,12547,12549],{"class":5363,"line":6905},[5361,12544,6557],{"class":6377},[5361,12546,8786],{"class":6342},[5361,12548,12025],{"class":6477},[5361,12550,6953],{"class":6342},[5361,12552,12553],{"class":5363,"line":6910},[5361,12554,6546],{"class":6342},[5361,12556,12557],{"class":5363,"line":6919},[5361,12558,12559],{"class":6551},"        # Три stage (16, 32, 64 каналів)\n",[5361,12561,12562,12564,12566,12568,12570,12572,12575,12577,12579,12581,12583,12585],{"class":5363,"line":6924},[5361,12563,6557],{"class":6377},[5361,12565,8854],{"class":6342},[5361,12567,6455],{"class":6377},[5361,12569,8859],{"class":6342},[5361,12571,12025],{"class":6477},[5361,12573,12574],{"class":6342},", num_blocks[",[5361,12576,8867],{"class":6477},[5361,12578,8870],{"class":6342},[5361,12580,6471],{"class":6454},[5361,12582,6474],{"class":6342},[5361,12584,6326],{"class":6477},[5361,12586,6953],{"class":6342},[5361,12588,12589,12591,12593,12595,12597,12599,12601,12603,12605,12607,12609,12611],{"class":5363,"line":6929},[5361,12590,6557],{"class":6377},[5361,12592,8885],{"class":6342},[5361,12594,6455],{"class":6377},[5361,12596,8859],{"class":6342},[5361,12598,7013],{"class":6477},[5361,12600,12574],{"class":6342},[5361,12602,6326],{"class":6477},[5361,12604,8870],{"class":6342},[5361,12606,6471],{"class":6454},[5361,12608,6474],{"class":6342},[5361,12610,7144],{"class":6477},[5361,12612,6953],{"class":6342},[5361,12614,12615,12617,12619,12621,12623,12625,12627,12629,12631,12633,12635,12637],{"class":5363,"line":6935},[5361,12616,6557],{"class":6377},[5361,12618,8912],{"class":6342},[5361,12620,6455],{"class":6377},[5361,12622,8859],{"class":6342},[5361,12624,6977],{"class":6477},[5361,12626,12574],{"class":6342},[5361,12628,7144],{"class":6477},[5361,12630,8870],{"class":6342},[5361,12632,6471],{"class":6454},[5361,12634,6474],{"class":6342},[5361,12636,7144],{"class":6477},[5361,12638,6953],{"class":6342},[5361,12640,12641],{"class":5363,"line":6956},[5361,12642,6546],{"class":6342},[5361,12644,12645,12647,12649,12651,12653,12655],{"class":5363,"line":6961},[5361,12646,6557],{"class":6377},[5361,12648,8976],{"class":6342},[5361,12650,6326],{"class":6477},[5361,12652,6458],{"class":6342},[5361,12654,6326],{"class":6477},[5361,12656,8985],{"class":6342},[5361,12658,12659,12661,12663,12665],{"class":5363,"line":6967},[5361,12660,6557],{"class":6377},[5361,12662,8992],{"class":6342},[5361,12664,6977],{"class":6477},[5361,12666,12667],{"class":6342},", num_classes)\n",[5361,12669,12670],{"class":5363,"line":6998},[5361,12671,6546],{"class":6342},[5361,12673,12674,12676],{"class":5363,"line":7026},[5361,12675,6557],{"class":6377},[5361,12677,9013],{"class":6342},[5361,12679,12680],{"class":5363,"line":7032},[5361,12681,6412],{"class":6342},[5361,12683,12684,12686,12688,12690,12692,12694,12696,12698,12700,12702,12704,12706,12708],{"class":5363,"line":7058},[5361,12685,6445],{"class":6377},[5361,12687,9024],{"class":6448},[5361,12689,6385],{"class":6342},[5361,12691,6455],{"class":6454},[5361,12693,6458],{"class":6342},[5361,12695,8636],{"class":6454},[5361,12697,6458],{"class":6342},[5361,12699,6466],{"class":6454},[5361,12701,6458],{"class":6342},[5361,12703,9041],{"class":6454},[5361,12705,6458],{"class":6342},[5361,12707,6471],{"class":6454},[5361,12709,6396],{"class":6342},[5361,12711,12712,12714],{"class":5363,"line":7081},[5361,12713,9079],{"class":6342},[5361,12715,9082],{"class":6377},[5361,12717,12718,12720,12722,12724,12726,12728],{"class":5363,"line":7099},[5361,12719,6840],{"class":6338},[5361,12721,9098],{"class":6342},[5361,12723,6326],{"class":6477},[5361,12725,9103],{"class":6377},[5361,12727,6843],{"class":6377},[5361,12729,12730],{"class":6342},".in_channels != out_channels:\n",[5361,12732,12733],{"class":5363,"line":7104},[5361,12734,9113],{"class":6342},[5361,12736,12737,12740,12742,12745,12747,12749,12751],{"class":5363,"line":7110},[5361,12738,12739],{"class":6342},"                nn.Conv2d(",[5361,12741,6455],{"class":6377},[5361,12743,12744],{"class":6342},".in_channels, out_channels, ",[5361,12746,7131],{"class":6454},[5361,12748,6474],{"class":6342},[5361,12750,6326],{"class":6477},[5361,12752,6583],{"class":6342},[5361,12754,12755,12758,12761,12763,12765,12767],{"class":5363,"line":7116},[5361,12756,12757],{"class":6454},"                         stride",[5361,12759,12760],{"class":6342},"=stride, ",[5361,12762,7149],{"class":6454},[5361,12764,6474],{"class":6342},[5361,12766,6615],{"class":6377},[5361,12768,7156],{"class":6342},[5361,12770,12771],{"class":5363,"line":7159},[5361,12772,12773],{"class":6342},"                nn.BatchNorm2d(out_channels)\n",[5361,12775,12776],{"class":5363,"line":7169},[5361,12777,9173],{"class":6342},[5361,12779,12780],{"class":5363,"line":7174},[5361,12781,6546],{"class":6342},[5361,12783,12784],{"class":5363,"line":7209},[5361,12785,9182],{"class":6342},[5361,12787,12788,12790,12792],{"class":5363,"line":7231},[5361,12789,9192],{"class":6342},[5361,12791,6455],{"class":6377},[5361,12793,9197],{"class":6342},[5361,12795,12796,12798],{"class":5363,"line":7237},[5361,12797,6557],{"class":6377},[5361,12799,12800],{"class":6342},".in_channels = out_channels\n",[5361,12802,12803],{"class":5363,"line":7259},[5361,12804,6546],{"class":6342},[5361,12806,12807,12809,12811,12813,12815,12817,12819],{"class":5363,"line":7281},[5361,12808,9218],{"class":6338},[5361,12810,9221],{"class":6342},[5361,12812,7938],{"class":6338},[5361,12814,9226],{"class":6448},[5361,12816,6385],{"class":6342},[5361,12818,6326],{"class":6477},[5361,12820,9233],{"class":6342},[5361,12822,12823,12825,12827],{"class":5363,"line":9463},[5361,12824,9238],{"class":6342},[5361,12826,6455],{"class":6377},[5361,12828,9243],{"class":6342},[5361,12830,12831],{"class":5363,"line":9468},[5361,12832,6546],{"class":6342},[5361,12834,12835,12837],{"class":5363,"line":9474},[5361,12836,6913],{"class":6338},[5361,12838,9254],{"class":6342},[5361,12840,12841],{"class":5363,"line":9484},[5361,12842,6412],{"class":6342},[5361,12844,12845,12847,12849,12851,12853],{"class":5363,"line":9494},[5361,12846,6445],{"class":6377},[5361,12848,9265],{"class":6448},[5361,12850,6385],{"class":6342},[5361,12852,6455],{"class":6454},[5361,12854,6396],{"class":6342},[5361,12856,12857,12859,12861,12863,12865],{"class":5363,"line":9504},[5361,12858,9218],{"class":6338},[5361,12860,9283],{"class":6342},[5361,12862,7938],{"class":6338},[5361,12864,6843],{"class":6377},[5361,12866,9290],{"class":6342},[5361,12868,12869,12871,12873],{"class":5363,"line":9509},[5361,12870,9295],{"class":6338},[5361,12872,9298],{"class":6448},[5361,12874,9301],{"class":6342},[5361,12876,12877,12879,12881,12883,12885,12887,12889,12891,12893],{"class":5363,"line":9517},[5361,12878,9306],{"class":6342},[5361,12880,9309],{"class":6454},[5361,12882,6474],{"class":6342},[5361,12884,9314],{"class":6401},[5361,12886,6458],{"class":6342},[5361,12888,9319],{"class":6454},[5361,12890,6474],{"class":6342},[5361,12892,9324],{"class":6401},[5361,12894,6953],{"class":6342},[5361,12896,12897,12899,12901],{"class":5363,"line":9522},[5361,12898,9331],{"class":6338},[5361,12900,9298],{"class":6448},[5361,12902,9336],{"class":6342},[5361,12904,12905,12907,12909],{"class":5363,"line":9527},[5361,12906,9341],{"class":6342},[5361,12908,6326],{"class":6477},[5361,12910,6953],{"class":6342},[5361,12912,12913,12915,12917],{"class":5363,"line":9533},[5361,12914,9350],{"class":6342},[5361,12916,8867],{"class":6477},[5361,12918,6953],{"class":6342},[5361,12920,12921],{"class":5363,"line":9560},[5361,12922,6412],{"class":6342},[5361,12924,12925,12927,12929,12931,12933,12935,12937],{"class":5363,"line":9566},[5361,12926,6445],{"class":6377},[5361,12928,6738],{"class":6448},[5361,12930,6385],{"class":6342},[5361,12932,6455],{"class":6454},[5361,12934,6458],{"class":6342},[5361,12936,6747],{"class":6454},[5361,12938,6396],{"class":6342},[5361,12940,12941,12944,12946,12948,12950],{"class":5363,"line":9592},[5361,12942,12943],{"class":6342},"        x = F.relu(",[5361,12945,6455],{"class":6377},[5361,12947,7759],{"class":6342},[5361,12949,6455],{"class":6377},[5361,12951,12952],{"class":6342},".conv1(x)))\n",[5361,12954,12955],{"class":5363,"line":9597},[5361,12956,6546],{"class":6342},[5361,12958,12959,12961,12963],{"class":5363,"line":9623},[5361,12960,9384],{"class":6342},[5361,12962,6455],{"class":6377},[5361,12964,9433],{"class":6342},[5361,12966,12967,12969,12971],{"class":5363,"line":9629},[5361,12968,9384],{"class":6342},[5361,12970,6455],{"class":6377},[5361,12972,9442],{"class":6342},[5361,12974,12975,12977,12979],{"class":5363,"line":9653},[5361,12976,9384],{"class":6342},[5361,12978,6455],{"class":6377},[5361,12980,9451],{"class":6342},[5361,12982,12983],{"class":5363,"line":9658},[5361,12984,6546],{"class":6342},[5361,12986,12987,12989,12991],{"class":5363,"line":9684},[5361,12988,9384],{"class":6342},[5361,12990,6455],{"class":6377},[5361,12992,9481],{"class":6342},[5361,12994,12995,12997,12999],{"class":5363,"line":9690},[5361,12996,9487],{"class":6342},[5361,12998,6326],{"class":6477},[5361,13000,6953],{"class":6342},[5361,13002,13003,13005,13007],{"class":5363,"line":9696},[5361,13004,9384],{"class":6342},[5361,13006,6455],{"class":6377},[5361,13008,9501],{"class":6342},[5361,13010,13011],{"class":5363,"line":9705},[5361,13012,6546],{"class":6342},[5361,13014,13015,13017],{"class":5363,"line":9729},[5361,13016,6913],{"class":6338},[5361,13018,9514],{"class":6342},[5361,13020,13021],{"class":5363,"line":9734},[5361,13022,5392],{"emptyLinePlaceholder":5391},[5361,13024,13025],{"class":5363,"line":9739},[5361,13026,5392],{"emptyLinePlaceholder":5391},[5361,13028,13029],{"class":5363,"line":9745},[5361,13030,13031],{"class":6551},"# Створення моделей\n",[5361,13033,13034,13037,13039,13041,13044,13046,13048,13050,13052],{"class":5363,"line":9761},[5361,13035,13036],{"class":6342},"plain20 = PlainCNN(",[5361,13038,11849],{"class":6454},[5361,13040,6474],{"class":6342},[5361,13042,13043],{"class":6477},"20",[5361,13045,6458],{"class":6342},[5361,13047,8646],{"class":6454},[5361,13049,6474],{"class":6342},[5361,13051,8651],{"class":6477},[5361,13053,13054],{"class":6342},").to(device)\n",[5361,13056,13057,13060,13062,13064,13067,13069,13071,13073,13075],{"class":5363,"line":9783},[5361,13058,13059],{"class":6342},"plain56 = PlainCNN(",[5361,13061,11849],{"class":6454},[5361,13063,6474],{"class":6342},[5361,13065,13066],{"class":6477},"56",[5361,13068,6458],{"class":6342},[5361,13070,8646],{"class":6454},[5361,13072,6474],{"class":6342},[5361,13074,8651],{"class":6477},[5361,13076,13054],{"class":6342},[5361,13078,13079,13082,13085,13087,13089,13091,13093,13095,13097,13099,13101],{"class":5363,"line":9788},[5361,13080,13081],{"class":6342},"resnet56 = ResNetCIFAR(BasicBlock, [",[5361,13083,13084],{"class":6477},"9",[5361,13086,6458],{"class":6342},[5361,13088,13084],{"class":6477},[5361,13090,6458],{"class":6342},[5361,13092,13084],{"class":6477},[5361,13094,8870],{"class":6342},[5361,13096,8646],{"class":6454},[5361,13098,6474],{"class":6342},[5361,13100,8651],{"class":6477},[5361,13102,13054],{"class":6342},[5361,13104,13105],{"class":5363,"line":9794},[5361,13106,5392],{"emptyLinePlaceholder":5391},[5361,13108,13109],{"class":5363,"line":9813},[5361,13110,13111],{"class":6551},"# Статистика моделей\n",[5361,13113,13114],{"class":5363,"line":9838},[5361,13115,13116],{"class":6342},"models_info = {\n",[5361,13118,13119,13122],{"class":5363,"line":9843},[5361,13120,13121],{"class":6401},"    'Plain-20'",[5361,13123,13124],{"class":6342},": plain20,\n",[5361,13126,13127,13130],{"class":5363,"line":9857},[5361,13128,13129],{"class":6401},"    'Plain-56'",[5361,13131,13132],{"class":6342},": plain56,\n",[5361,13134,13135,13138],{"class":5363,"line":9880},[5361,13136,13137],{"class":6401},"    'ResNet-56'",[5361,13139,13140],{"class":6342},": resnet56\n",[5361,13142,13143],{"class":5363,"line":9903},[5361,13144,5545],{"class":6342},[5361,13146,13147],{"class":5363,"line":9944},[5361,13148,5392],{"emptyLinePlaceholder":5391},[5361,13150,13151,13153,13155,13158,13160,13162],{"class":5363,"line":9949},[5361,13152,6938],{"class":6448},[5361,13154,6385],{"class":6342},[5361,13156,13157],{"class":6401},"\"🏗️ Архітектури для експерименту:",[5361,13159,6947],{"class":6946},[5361,13161,6950],{"class":6401},[5361,13163,6953],{"class":6342},[5361,13165,13166,13168,13171,13173],{"class":5363,"line":9955},[5361,13167,7932],{"class":6338},[5361,13169,13170],{"class":6342}," name, model ",[5361,13172,7938],{"class":6338},[5361,13174,13175],{"class":6342}," models_info.items():\n",[5361,13177,13178,13181,13183,13185,13187,13189,13191],{"class":5363,"line":9980},[5361,13179,13180],{"class":6342},"    params = ",[5361,13182,7926],{"class":6448},[5361,13184,7929],{"class":6342},[5361,13186,7932],{"class":6338},[5361,13188,7935],{"class":6342},[5361,13190,7938],{"class":6338},[5361,13192,9810],{"class":6342},[5361,13194,13195,13198,13200,13202,13204,13206,13209,13211,13214],{"class":5363,"line":9989},[5361,13196,13197],{"class":6448},"    print",[5361,13199,6385],{"class":6342},[5361,13201,7039],{"class":6377},[5361,13203,6950],{"class":6401},[5361,13205,7045],{"class":6377},[5361,13207,13208],{"class":6342},"name",[5361,13210,7051],{"class":6377},[5361,13212,13213],{"class":6401},":\"",[5361,13215,6953],{"class":6342},[5361,13217,13218,13220,13222,13224,13227,13229,13232,13234,13236],{"class":5363,"line":9995},[5361,13219,13197],{"class":6448},[5361,13221,6385],{"class":6342},[5361,13223,7039],{"class":6377},[5361,13225,13226],{"class":6401},"\"├─ Параметри: ",[5361,13228,7045],{"class":6377},[5361,13230,13231],{"class":6342},"params",[5361,13233,8033],{"class":6377},[5361,13235,6950],{"class":6401},[5361,13237,6953],{"class":6342},[5361,13239,13240,13242,13244,13246,13249,13251,13254,13256,13258,13260,13262,13264,13266,13269,13271,13273],{"class":5363,"line":10017},[5361,13241,13197],{"class":6448},[5361,13243,6385],{"class":6342},[5361,13245,7039],{"class":6377},[5361,13247,13248],{"class":6401},"\"└─ Розмір: ",[5361,13250,7045],{"class":6377},[5361,13252,13253],{"class":6342},"params * ",[5361,13255,7004],{"class":6477},[5361,13257,9922],{"class":6342},[5361,13259,9925],{"class":6477},[5361,13261,9922],{"class":6342},[5361,13263,9925],{"class":6477},[5361,13265,9932],{"class":6377},[5361,13267,13268],{"class":6401}," MB",[5361,13270,6947],{"class":6946},[5361,13272,6950],{"class":6401},[5361,13274,6953],{"class":6342},[7294,13276,13277,13282,13350],{"v-slot:output":5356},[5191,13278,10055,13279],{},[5195,13280,13281],{},"Архітектури для експерименту:",[5203,13283,13284,13300],{},[5206,13285,13286],{},[5209,13287,13288,13291,13294,13297],{},[5212,13289,13290],{"align":5214},"Модель",[5212,13292,13293],{"align":5214},"Кількість параметрів",[5212,13295,13296],{"align":5214},"Розмір (MB)",[5212,13298,13299],{"align":5214},"Глибина",[5226,13301,13302,13318,13334],{},[5209,13303,13304,13309,13312,13315],{},[5231,13305,13306],{"align":5214},[5195,13307,13308],{},"Plain-20",[5231,13310,13311],{"align":5214},"272,474",[5231,13313,13314],{"align":5214},"1.04",[5231,13316,13317],{"align":5214},"20 шарів",[5209,13319,13320,13325,13328,13331],{},[5231,13321,13322],{"align":5214},[5195,13323,13324],{},"Plain-56",[5231,13326,13327],{"align":5214},"855,578",[5231,13329,13330],{"align":5214},"3.26",[5231,13332,13333],{"align":5214},"56 шарів",[5209,13335,13336,13341,13344,13347],{},[5231,13337,13338],{"align":5214},[5195,13339,13340],{},"ResNet-56",[5231,13342,13343],{"align":5214},"853,018",[5231,13345,13346],{"align":5214},"3.25",[5231,13348,13349],{"align":5214},"56 шарів (з skip connections)",[5191,13351,13352],{},"💡 Plain-56 та ResNet-56 мають майже однакову кількість параметрів, тому порівняння чесне!",[5191,13354,6259],{},[5280,13356,13358],{"id":13357},"тренувальний-та-валідаційний-цикли","Тренувальний та валідаційний цикли",[6314,13360,13361,14378],{"execution-count":9644},[5351,13362,13364],{"className":6329,"code":13363,"language":6331,"meta":5356,"style":5356},"def train_epoch(model, loader, criterion, optimizer, device):\n    \"\"\"Тренування однієї епохи\"\"\"\n    model.train()\n    running_loss = 0.0\n    correct = 0\n    total = 0\n    \n    for images, labels in loader:\n        images, labels = images.to(device), labels.to(device)\n        \n        optimizer.zero_grad()\n        outputs = model(images)\n        loss = criterion(outputs, labels)\n        loss.backward()\n        optimizer.step()\n        \n        running_loss += loss.item() * images.size(0)\n        _, predicted = outputs.max(1)\n        total += labels.size(0)\n        correct += predicted.eq(labels).sum().item()\n    \n    epoch_loss = running_loss \u002F total\n    epoch_acc = 100. * correct \u002F total\n    return epoch_loss, epoch_acc\n\n\ndef validate(model, loader, criterion, device):\n    \"\"\"Валідація\u002Fтестування\"\"\"\n    model.eval()\n    running_loss = 0.0\n    correct = 0\n    total = 0\n    \n    with torch.no_grad():\n        for images, labels in loader:\n            images, labels = images.to(device), labels.to(device)\n            outputs = model(images)\n            loss = criterion(outputs, labels)\n            \n            running_loss += loss.item() * images.size(0)\n            _, predicted = outputs.max(1)\n            total += labels.size(0)\n            correct += predicted.eq(labels).sum().item()\n    \n    epoch_loss = running_loss \u002F total\n    epoch_acc = 100. * correct \u002F total\n    return epoch_loss, epoch_acc\n\n\ndef train_model(model, model_name, num_epochs=100):\n    \"\"\"Повний цикл навчання моделі\"\"\"\n    criterion = nn.CrossEntropyLoss()\n    optimizer = optim.SGD(model.parameters(), lr=0.1, momentum=0.9, weight_decay=5e-4)\n    scheduler = optim.lr_scheduler.MultiStepLR(optimizer, milestones=[60, 80], gamma=0.1)\n    \n    history = {\n        'train_loss': [], 'train_acc': [],\n        'test_loss': [], 'test_acc': []\n    }\n    \n    print(f\"\\n🚀 Початок навчання {model_name}\")\n    print(\"=\" * 80)\n    start_time = time.time()\n    \n    best_acc = 0\n    \n    for epoch in range(1, num_epochs + 1):\n        # Тренування\n        train_loss, train_acc = train_epoch(model, train_loader, criterion, optimizer, device)\n        \n        # Валідація\n        test_loss, test_acc = validate(model, test_loader, criterion, device)\n        \n        # Оновлення learning rate\n        scheduler.step()\n        \n        # Збереження історії\n        history['train_loss'].append(train_loss)\n        history['train_acc'].append(train_acc)\n        history['test_loss'].append(test_loss)\n        history['test_acc'].append(test_acc)\n        \n        if test_acc > best_acc:\n            best_acc = test_acc\n        \n        # Виведення прогресу кожні 10 епох\n        if epoch % 10 == 0 or epoch == 1:\n            lr = scheduler.get_last_lr()[0]\n            print(f\"Epoch {epoch:3d}\u002F{num_epochs} | \"\n                  f\"Train Loss: {train_loss:.4f} | Train Acc: {train_acc:.2f}% | \"\n                  f\"Test Loss: {test_loss:.4f} | Test Acc: {test_acc:.2f}% | \"\n                  f\"LR: {lr:.4f}\")\n    \n    elapsed = time.time() - start_time\n    print(\"=\" * 80)\n    print(f\"✅ Навчання завершено за {elapsed\u002F60:.1f} хв\")\n    print(f\"🏆 Найкраща Test Accuracy: {best_acc:.2f}%\\n\")\n    \n    return history\n\n\n# Навчання всіх трьох моделей\nresults = {}\n\n# Модель 1: Plain-20 (базова)\nresults['Plain-20'] = train_model(plain20, 'Plain-20', num_epochs=100)\n\n# Модель 2: Plain-56 (має показати degradation)\nresults['Plain-56'] = train_model(plain56, 'Plain-56', num_epochs=100)\n\n# Модель 3: ResNet-56 (має вирішити проблему)\nresults['ResNet-56'] = train_model(resnet56, 'ResNet-56', num_epochs=100)\n",[5358,13365,13366,13399,13404,13409,13417,13425,13432,13436,13449,13454,13458,13463,13468,13473,13478,13483,13487,13496,13505,13514,13519,13523,13528,13538,13545,13549,13553,13578,13583,13588,13594,13600,13606,13610,13617,13627,13632,13637,13642,13646,13655,13664,13673,13678,13682,13686,13694,13700,13704,13708,13735,13740,13745,13777,13805,13809,13814,13828,13841,13846,13850,13875,13891,13896,13900,13907,13911,13933,13938,13943,13947,13952,13957,13961,13966,13971,13975,13980,13991,14000,14010,14019,14023,14030,14035,14039,14044,14067,14076,14108,14137,14163,14180,14184,14189,14203,14228,14255,14259,14266,14270,14274,14279,14284,14288,14293,14316,14320,14325,14347,14351,14356],{"__ignoreMap":5356},[5361,13367,13368,13370,13373,13375,13378,13380,13383,13385,13388,13390,13393,13395,13397],{"class":5363,"line":5364},[5361,13369,9536],{"class":6377},[5361,13371,13372],{"class":6448}," train_epoch",[5361,13374,6385],{"class":6342},[5361,13376,13377],{"class":6454},"model",[5361,13379,6458],{"class":6342},[5361,13381,13382],{"class":6454},"loader",[5361,13384,6458],{"class":6342},[5361,13386,13387],{"class":6454},"criterion",[5361,13389,6458],{"class":6342},[5361,13391,13392],{"class":6454},"optimizer",[5361,13394,6458],{"class":6342},[5361,13396,11302],{"class":6454},[5361,13398,6396],{"class":6342},[5361,13400,13401],{"class":5363,"line":5370},[5361,13402,13403],{"class":6401},"    \"\"\"Тренування однієї епохи\"\"\"\n",[5361,13405,13406],{"class":5363,"line":5376},[5361,13407,13408],{"class":6342},"    model.train()\n",[5361,13410,13411,13414],{"class":5363,"line":5382},[5361,13412,13413],{"class":6342},"    running_loss = ",[5361,13415,13416],{"class":6477},"0.0\n",[5361,13418,13419,13422],{"class":5363,"line":5388},[5361,13420,13421],{"class":6342},"    correct = ",[5361,13423,13424],{"class":6477},"0\n",[5361,13426,13427,13430],{"class":5363,"line":5395},[5361,13428,13429],{"class":6342},"    total = ",[5361,13431,13424],{"class":6477},[5361,13433,13434],{"class":5363,"line":5401},[5361,13435,6412],{"class":6342},[5361,13437,13438,13441,13444,13446],{"class":5363,"line":5407},[5361,13439,13440],{"class":6338},"    for",[5361,13442,13443],{"class":6342}," images, labels ",[5361,13445,7938],{"class":6338},[5361,13447,13448],{"class":6342}," loader:\n",[5361,13450,13451],{"class":5363,"line":5413},[5361,13452,13453],{"class":6342},"        images, labels = images.to(device), labels.to(device)\n",[5361,13455,13456],{"class":5363,"line":5418},[5361,13457,6546],{"class":6342},[5361,13459,13460],{"class":5363,"line":5424},[5361,13461,13462],{"class":6342},"        optimizer.zero_grad()\n",[5361,13464,13465],{"class":5363,"line":5430},[5361,13466,13467],{"class":6342},"        outputs = model(images)\n",[5361,13469,13470],{"class":5363,"line":5436},[5361,13471,13472],{"class":6342},"        loss = criterion(outputs, labels)\n",[5361,13474,13475],{"class":5363,"line":5442},[5361,13476,13477],{"class":6342},"        loss.backward()\n",[5361,13479,13480],{"class":5363,"line":5448},[5361,13481,13482],{"class":6342},"        optimizer.step()\n",[5361,13484,13485],{"class":5363,"line":5454},[5361,13486,6546],{"class":6342},[5361,13488,13489,13492,13494],{"class":5363,"line":5460},[5361,13490,13491],{"class":6342},"        running_loss += loss.item() * images.size(",[5361,13493,8867],{"class":6477},[5361,13495,6953],{"class":6342},[5361,13497,13498,13501,13503],{"class":5363,"line":5466},[5361,13499,13500],{"class":6342},"        _, predicted = outputs.max(",[5361,13502,6326],{"class":6477},[5361,13504,6953],{"class":6342},[5361,13506,13507,13510,13512],{"class":5363,"line":5472},[5361,13508,13509],{"class":6342},"        total += labels.size(",[5361,13511,8867],{"class":6477},[5361,13513,6953],{"class":6342},[5361,13515,13516],{"class":5363,"line":5478},[5361,13517,13518],{"class":6342},"        correct += predicted.eq(labels).sum().item()\n",[5361,13520,13521],{"class":5363,"line":5484},[5361,13522,6412],{"class":6342},[5361,13524,13525],{"class":5363,"line":5489},[5361,13526,13527],{"class":6342},"    epoch_loss = running_loss \u002F total\n",[5361,13529,13530,13533,13535],{"class":5363,"line":5495},[5361,13531,13532],{"class":6342},"    epoch_acc = ",[5361,13534,8102],{"class":6477},[5361,13536,13537],{"class":6342},". * correct \u002F total\n",[5361,13539,13540,13542],{"class":5363,"line":5501},[5361,13541,9569],{"class":6338},[5361,13543,13544],{"class":6342}," epoch_loss, epoch_acc\n",[5361,13546,13547],{"class":5363,"line":5507},[5361,13548,5392],{"emptyLinePlaceholder":5391},[5361,13550,13551],{"class":5363,"line":5513},[5361,13552,5392],{"emptyLinePlaceholder":5391},[5361,13554,13555,13557,13560,13562,13564,13566,13568,13570,13572,13574,13576],{"class":5363,"line":5519},[5361,13556,9536],{"class":6377},[5361,13558,13559],{"class":6448}," validate",[5361,13561,6385],{"class":6342},[5361,13563,13377],{"class":6454},[5361,13565,6458],{"class":6342},[5361,13567,13382],{"class":6454},[5361,13569,6458],{"class":6342},[5361,13571,13387],{"class":6454},[5361,13573,6458],{"class":6342},[5361,13575,11302],{"class":6454},[5361,13577,6396],{"class":6342},[5361,13579,13580],{"class":5363,"line":5525},[5361,13581,13582],{"class":6401},"    \"\"\"Валідація\u002Fтестування\"\"\"\n",[5361,13584,13585],{"class":5363,"line":5531},[5361,13586,13587],{"class":6342},"    model.eval()\n",[5361,13589,13590,13592],{"class":5363,"line":5537},[5361,13591,13413],{"class":6342},[5361,13593,13416],{"class":6477},[5361,13595,13596,13598],{"class":5363,"line":5542},[5361,13597,13421],{"class":6342},[5361,13599,13424],{"class":6477},[5361,13601,13602,13604],{"class":5363,"line":5548},[5361,13603,13429],{"class":6342},[5361,13605,13424],{"class":6477},[5361,13607,13608],{"class":5363,"line":5553},[5361,13609,6412],{"class":6342},[5361,13611,13612,13615],{"class":5363,"line":5559},[5361,13613,13614],{"class":6338},"    with",[5361,13616,9986],{"class":6342},[5361,13618,13619,13621,13623,13625],{"class":5363,"line":5565},[5361,13620,9218],{"class":6338},[5361,13622,13443],{"class":6342},[5361,13624,7938],{"class":6338},[5361,13626,13448],{"class":6342},[5361,13628,13629],{"class":5363,"line":5571},[5361,13630,13631],{"class":6342},"            images, labels = images.to(device), labels.to(device)\n",[5361,13633,13634],{"class":5363,"line":5577},[5361,13635,13636],{"class":6342},"            outputs = model(images)\n",[5361,13638,13639],{"class":5363,"line":5583},[5361,13640,13641],{"class":6342},"            loss = criterion(outputs, labels)\n",[5361,13643,13644],{"class":5363,"line":5588},[5361,13645,12061],{"class":6342},[5361,13647,13648,13651,13653],{"class":5363,"line":5594},[5361,13649,13650],{"class":6342},"            running_loss += loss.item() * images.size(",[5361,13652,8867],{"class":6477},[5361,13654,6953],{"class":6342},[5361,13656,13657,13660,13662],{"class":5363,"line":5600},[5361,13658,13659],{"class":6342},"            _, predicted = outputs.max(",[5361,13661,6326],{"class":6477},[5361,13663,6953],{"class":6342},[5361,13665,13666,13669,13671],{"class":5363,"line":5606},[5361,13667,13668],{"class":6342},"            total += labels.size(",[5361,13670,8867],{"class":6477},[5361,13672,6953],{"class":6342},[5361,13674,13675],{"class":5363,"line":5612},[5361,13676,13677],{"class":6342},"            correct += predicted.eq(labels).sum().item()\n",[5361,13679,13680],{"class":5363,"line":5618},[5361,13681,6412],{"class":6342},[5361,13683,13684],{"class":5363,"line":5623},[5361,13685,13527],{"class":6342},[5361,13687,13688,13690,13692],{"class":5363,"line":5628},[5361,13689,13532],{"class":6342},[5361,13691,8102],{"class":6477},[5361,13693,13537],{"class":6342},[5361,13695,13696,13698],{"class":5363,"line":5634},[5361,13697,9569],{"class":6338},[5361,13699,13544],{"class":6342},[5361,13701,13702],{"class":5363,"line":5640},[5361,13703,5392],{"emptyLinePlaceholder":5391},[5361,13705,13706],{"class":5363,"line":5646},[5361,13707,5392],{"emptyLinePlaceholder":5391},[5361,13709,13710,13712,13715,13717,13719,13721,13724,13726,13729,13731,13733],{"class":5363,"line":5652},[5361,13711,9536],{"class":6377},[5361,13713,13714],{"class":6448}," train_model",[5361,13716,6385],{"class":6342},[5361,13718,13377],{"class":6454},[5361,13720,6458],{"class":6342},[5361,13722,13723],{"class":6454},"model_name",[5361,13725,6458],{"class":6342},[5361,13727,13728],{"class":6454},"num_epochs",[5361,13730,6474],{"class":6342},[5361,13732,8102],{"class":6477},[5361,13734,6396],{"class":6342},[5361,13736,13737],{"class":5363,"line":5657},[5361,13738,13739],{"class":6401},"    \"\"\"Повний цикл навчання моделі\"\"\"\n",[5361,13741,13742],{"class":5363,"line":5662},[5361,13743,13744],{"class":6342},"    criterion = nn.CrossEntropyLoss()\n",[5361,13746,13747,13750,13752,13754,13756,13758,13761,13763,13766,13768,13770,13772,13775],{"class":5363,"line":6757},[5361,13748,13749],{"class":6342},"    optimizer = optim.SGD(model.parameters(), ",[5361,13751,10703],{"class":6454},[5361,13753,6474],{"class":6342},[5361,13755,10892],{"class":6477},[5361,13757,6458],{"class":6342},[5361,13759,13760],{"class":6454},"momentum",[5361,13762,6474],{"class":6342},[5361,13764,13765],{"class":6477},"0.9",[5361,13767,6458],{"class":6342},[5361,13769,10713],{"class":6454},[5361,13771,6474],{"class":6342},[5361,13773,13774],{"class":6477},"5e-4",[5361,13776,6953],{"class":6342},[5361,13778,13779,13782,13785,13787,13790,13792,13794,13796,13799,13801,13803],{"class":5363,"line":6763},[5361,13780,13781],{"class":6342},"    scheduler = optim.lr_scheduler.MultiStepLR(optimizer, ",[5361,13783,13784],{"class":6454},"milestones",[5361,13786,10941],{"class":6342},[5361,13788,13789],{"class":6477},"60",[5361,13791,6458],{"class":6342},[5361,13793,10944],{"class":6477},[5361,13795,8870],{"class":6342},[5361,13797,13798],{"class":6454},"gamma",[5361,13800,6474],{"class":6342},[5361,13802,10892],{"class":6477},[5361,13804,6953],{"class":6342},[5361,13806,13807],{"class":5363,"line":6768},[5361,13808,6412],{"class":6342},[5361,13810,13811],{"class":5363,"line":6774},[5361,13812,13813],{"class":6342},"    history = {\n",[5361,13815,13816,13819,13822,13825],{"class":5363,"line":6785},[5361,13817,13818],{"class":6401},"        'train_loss'",[5361,13820,13821],{"class":6342},": [], ",[5361,13823,13824],{"class":6401},"'train_acc'",[5361,13826,13827],{"class":6342},": [],\n",[5361,13829,13830,13833,13835,13838],{"class":5363,"line":6795},[5361,13831,13832],{"class":6401},"        'test_loss'",[5361,13834,13821],{"class":6342},[5361,13836,13837],{"class":6401},"'test_acc'",[5361,13839,13840],{"class":6342},": []\n",[5361,13842,13843],{"class":5363,"line":6801},[5361,13844,13845],{"class":6342},"    }\n",[5361,13847,13848],{"class":5363,"line":6806},[5361,13849,6412],{"class":6342},[5361,13851,13852,13854,13856,13858,13860,13862,13865,13867,13869,13871,13873],{"class":5363,"line":6816},[5361,13853,13197],{"class":6448},[5361,13855,6385],{"class":6342},[5361,13857,7039],{"class":6377},[5361,13859,6950],{"class":6401},[5361,13861,6947],{"class":6946},[5361,13863,13864],{"class":6401},"🚀 Початок навчання ",[5361,13866,7045],{"class":6377},[5361,13868,13723],{"class":6342},[5361,13870,7051],{"class":6377},[5361,13872,6950],{"class":6401},[5361,13874,6953],{"class":6342},[5361,13876,13877,13879,13881,13884,13887,13889],{"class":5363,"line":6826},[5361,13878,13197],{"class":6448},[5361,13880,6385],{"class":6342},[5361,13882,13883],{"class":6401},"\"=\"",[5361,13885,13886],{"class":6342}," * ",[5361,13888,10944],{"class":6477},[5361,13890,6953],{"class":6342},[5361,13892,13893],{"class":5363,"line":6831},[5361,13894,13895],{"class":6342},"    start_time = time.time()\n",[5361,13897,13898],{"class":5363,"line":6837},[5361,13899,6412],{"class":6342},[5361,13901,13902,13905],{"class":5363,"line":6861},[5361,13903,13904],{"class":6342},"    best_acc = ",[5361,13906,13424],{"class":6477},[5361,13908,13909],{"class":5363,"line":6872},[5361,13910,6412],{"class":6342},[5361,13912,13913,13915,13918,13920,13922,13924,13926,13929,13931],{"class":5363,"line":6877},[5361,13914,13440],{"class":6338},[5361,13916,13917],{"class":6342}," epoch ",[5361,13919,7938],{"class":6338},[5361,13921,9226],{"class":6448},[5361,13923,6385],{"class":6342},[5361,13925,6326],{"class":6477},[5361,13927,13928],{"class":6342},", num_epochs + ",[5361,13930,6326],{"class":6477},[5361,13932,6396],{"class":6342},[5361,13934,13935],{"class":5363,"line":6883},[5361,13936,13937],{"class":6551},"        # Тренування\n",[5361,13939,13940],{"class":5363,"line":6889},[5361,13941,13942],{"class":6342},"        train_loss, train_acc = train_epoch(model, train_loader, criterion, optimizer, device)\n",[5361,13944,13945],{"class":5363,"line":6894},[5361,13946,6546],{"class":6342},[5361,13948,13949],{"class":5363,"line":6900},[5361,13950,13951],{"class":6551},"        # Валідація\n",[5361,13953,13954],{"class":5363,"line":6905},[5361,13955,13956],{"class":6342},"        test_loss, test_acc = validate(model, test_loader, criterion, device)\n",[5361,13958,13959],{"class":5363,"line":6910},[5361,13960,6546],{"class":6342},[5361,13962,13963],{"class":5363,"line":6919},[5361,13964,13965],{"class":6551},"        # Оновлення learning rate\n",[5361,13967,13968],{"class":5363,"line":6924},[5361,13969,13970],{"class":6342},"        scheduler.step()\n",[5361,13972,13973],{"class":5363,"line":6929},[5361,13974,6546],{"class":6342},[5361,13976,13977],{"class":5363,"line":6935},[5361,13978,13979],{"class":6551},"        # Збереження історії\n",[5361,13981,13982,13985,13988],{"class":5363,"line":6956},[5361,13983,13984],{"class":6342},"        history[",[5361,13986,13987],{"class":6401},"'train_loss'",[5361,13989,13990],{"class":6342},"].append(train_loss)\n",[5361,13992,13993,13995,13997],{"class":5363,"line":6961},[5361,13994,13984],{"class":6342},[5361,13996,13824],{"class":6401},[5361,13998,13999],{"class":6342},"].append(train_acc)\n",[5361,14001,14002,14004,14007],{"class":5363,"line":6967},[5361,14003,13984],{"class":6342},[5361,14005,14006],{"class":6401},"'test_loss'",[5361,14008,14009],{"class":6342},"].append(test_loss)\n",[5361,14011,14012,14014,14016],{"class":5363,"line":6998},[5361,14013,13984],{"class":6342},[5361,14015,13837],{"class":6401},[5361,14017,14018],{"class":6342},"].append(test_acc)\n",[5361,14020,14021],{"class":5363,"line":7026},[5361,14022,6546],{"class":6342},[5361,14024,14025,14027],{"class":5363,"line":7032},[5361,14026,6840],{"class":6338},[5361,14028,14029],{"class":6342}," test_acc > best_acc:\n",[5361,14031,14032],{"class":5363,"line":7058},[5361,14033,14034],{"class":6342},"            best_acc = test_acc\n",[5361,14036,14037],{"class":5363,"line":7081},[5361,14038,6546],{"class":6342},[5361,14040,14041],{"class":5363,"line":7099},[5361,14042,14043],{"class":6551},"        # Виведення прогресу кожні 10 епох\n",[5361,14045,14046,14048,14051,14053,14056,14058,14060,14063,14065],{"class":5363,"line":7104},[5361,14047,6840],{"class":6338},[5361,14049,14050],{"class":6342}," epoch % ",[5361,14052,8651],{"class":6477},[5361,14054,14055],{"class":6342}," == ",[5361,14057,8867],{"class":6477},[5361,14059,9103],{"class":6377},[5361,14061,14062],{"class":6342}," epoch == ",[5361,14064,6326],{"class":6477},[5361,14066,6858],{"class":6342},[5361,14068,14069,14072,14074],{"class":5363,"line":7110},[5361,14070,14071],{"class":6342},"            lr = scheduler.get_last_lr()[",[5361,14073,8867],{"class":6477},[5361,14075,11409],{"class":6342},[5361,14077,14078,14081,14083,14085,14088,14090,14093,14096,14099,14101,14103,14105],{"class":5363,"line":7116},[5361,14079,14080],{"class":6448},"            print",[5361,14082,6385],{"class":6342},[5361,14084,7039],{"class":6377},[5361,14086,14087],{"class":6401},"\"Epoch ",[5361,14089,7045],{"class":6377},[5361,14091,14092],{"class":6342},"epoch",[5361,14094,14095],{"class":6377},":3d}",[5361,14097,14098],{"class":6401},"\u002F",[5361,14100,7045],{"class":6377},[5361,14102,13728],{"class":6342},[5361,14104,7051],{"class":6377},[5361,14106,14107],{"class":6401}," | \"\n",[5361,14109,14110,14113,14116,14118,14121,14124,14127,14129,14132,14134],{"class":5363,"line":7159},[5361,14111,14112],{"class":6377},"                  f",[5361,14114,14115],{"class":6401},"\"Train Loss: ",[5361,14117,7045],{"class":6377},[5361,14119,14120],{"class":6342},"train_loss",[5361,14122,14123],{"class":6377},":.4f}",[5361,14125,14126],{"class":6401}," | Train Acc: ",[5361,14128,7045],{"class":6377},[5361,14130,14131],{"class":6342},"train_acc",[5361,14133,9932],{"class":6377},[5361,14135,14136],{"class":6401},"% | \"\n",[5361,14138,14139,14141,14144,14146,14149,14151,14154,14156,14159,14161],{"class":5363,"line":7169},[5361,14140,14112],{"class":6377},[5361,14142,14143],{"class":6401},"\"Test Loss: ",[5361,14145,7045],{"class":6377},[5361,14147,14148],{"class":6342},"test_loss",[5361,14150,14123],{"class":6377},[5361,14152,14153],{"class":6401}," | Test Acc: ",[5361,14155,7045],{"class":6377},[5361,14157,14158],{"class":6342},"test_acc",[5361,14160,9932],{"class":6377},[5361,14162,14136],{"class":6401},[5361,14164,14165,14167,14170,14172,14174,14176,14178],{"class":5363,"line":7174},[5361,14166,14112],{"class":6377},[5361,14168,14169],{"class":6401},"\"LR: ",[5361,14171,7045],{"class":6377},[5361,14173,10703],{"class":6342},[5361,14175,14123],{"class":6377},[5361,14177,6950],{"class":6401},[5361,14179,6953],{"class":6342},[5361,14181,14182],{"class":5363,"line":7209},[5361,14183,6412],{"class":6342},[5361,14185,14186],{"class":5363,"line":7231},[5361,14187,14188],{"class":6342},"    elapsed = time.time() - start_time\n",[5361,14190,14191,14193,14195,14197,14199,14201],{"class":5363,"line":7237},[5361,14192,13197],{"class":6448},[5361,14194,6385],{"class":6342},[5361,14196,13883],{"class":6401},[5361,14198,13886],{"class":6342},[5361,14200,10944],{"class":6477},[5361,14202,6953],{"class":6342},[5361,14204,14205,14207,14209,14211,14214,14216,14219,14221,14223,14226],{"class":5363,"line":7259},[5361,14206,13197],{"class":6448},[5361,14208,6385],{"class":6342},[5361,14210,7039],{"class":6377},[5361,14212,14213],{"class":6401},"\"✅ Навчання завершено за ",[5361,14215,7045],{"class":6377},[5361,14217,14218],{"class":6342},"elapsed\u002F",[5361,14220,13789],{"class":6477},[5361,14222,8105],{"class":6377},[5361,14224,14225],{"class":6401}," хв\"",[5361,14227,6953],{"class":6342},[5361,14229,14230,14232,14234,14236,14239,14241,14244,14246,14249,14251,14253],{"class":5363,"line":7281},[5361,14231,13197],{"class":6448},[5361,14233,6385],{"class":6342},[5361,14235,7039],{"class":6377},[5361,14237,14238],{"class":6401},"\"🏆 Найкраща Test Accuracy: ",[5361,14240,7045],{"class":6377},[5361,14242,14243],{"class":6342},"best_acc",[5361,14245,9932],{"class":6377},[5361,14247,14248],{"class":6401},"%",[5361,14250,6947],{"class":6946},[5361,14252,6950],{"class":6401},[5361,14254,6953],{"class":6342},[5361,14256,14257],{"class":5363,"line":9463},[5361,14258,6412],{"class":6342},[5361,14260,14261,14263],{"class":5363,"line":9468},[5361,14262,9569],{"class":6338},[5361,14264,14265],{"class":6342}," history\n",[5361,14267,14268],{"class":5363,"line":9474},[5361,14269,5392],{"emptyLinePlaceholder":5391},[5361,14271,14272],{"class":5363,"line":9484},[5361,14273,5392],{"emptyLinePlaceholder":5391},[5361,14275,14276],{"class":5363,"line":9494},[5361,14277,14278],{"class":6551},"# Навчання всіх трьох моделей\n",[5361,14280,14281],{"class":5363,"line":9504},[5361,14282,14283],{"class":6342},"results = {}\n",[5361,14285,14286],{"class":5363,"line":9509},[5361,14287,5392],{"emptyLinePlaceholder":5391},[5361,14289,14290],{"class":5363,"line":9517},[5361,14291,14292],{"class":6551},"# Модель 1: Plain-20 (базова)\n",[5361,14294,14295,14298,14301,14304,14306,14308,14310,14312,14314],{"class":5363,"line":9522},[5361,14296,14297],{"class":6342},"results[",[5361,14299,14300],{"class":6401},"'Plain-20'",[5361,14302,14303],{"class":6342},"] = train_model(plain20, ",[5361,14305,14300],{"class":6401},[5361,14307,6458],{"class":6342},[5361,14309,13728],{"class":6454},[5361,14311,6474],{"class":6342},[5361,14313,8102],{"class":6477},[5361,14315,6953],{"class":6342},[5361,14317,14318],{"class":5363,"line":9527},[5361,14319,5392],{"emptyLinePlaceholder":5391},[5361,14321,14322],{"class":5363,"line":9533},[5361,14323,14324],{"class":6551},"# Модель 2: Plain-56 (має показати degradation)\n",[5361,14326,14327,14329,14332,14335,14337,14339,14341,14343,14345],{"class":5363,"line":9560},[5361,14328,14297],{"class":6342},[5361,14330,14331],{"class":6401},"'Plain-56'",[5361,14333,14334],{"class":6342},"] = train_model(plain56, ",[5361,14336,14331],{"class":6401},[5361,14338,6458],{"class":6342},[5361,14340,13728],{"class":6454},[5361,14342,6474],{"class":6342},[5361,14344,8102],{"class":6477},[5361,14346,6953],{"class":6342},[5361,14348,14349],{"class":5363,"line":9566},[5361,14350,5392],{"emptyLinePlaceholder":5391},[5361,14352,14353],{"class":5363,"line":9592},[5361,14354,14355],{"class":6551},"# Модель 3: ResNet-56 (має вирішити проблему)\n",[5361,14357,14358,14360,14363,14366,14368,14370,14372,14374,14376],{"class":5363,"line":9597},[5361,14359,14297],{"class":6342},[5361,14361,14362],{"class":6401},"'ResNet-56'",[5361,14364,14365],{"class":6342},"] = train_model(resnet56, ",[5361,14367,14362],{"class":6401},[5361,14369,6458],{"class":6342},[5361,14371,13728],{"class":6454},[5361,14373,6474],{"class":6342},[5361,14375,8102],{"class":6477},[5361,14377,6953],{"class":6342},[7294,14379,14380,14388,14392,14398,14404,14412,14417,14424,14430,14437,14442],{"v-slot:output":5356},[14381,14382,14384,14385],"h1",{"id":14383},"початок-навчання-plain-20","🚀 ",[5195,14386,14387],{},"Початок навчання Plain-20",[14381,14389,14391],{"id":14390},"epoch-1100-train-loss-16234-train-acc-3924-test-loss-14201-test-acc-4735-lr-01000epoch-10100-train-loss-07845-train-acc-7218-test-loss-08123-test-acc-7105-lr-01000epoch-20100-train-loss-05234-train-acc-8145-test-loss-06789-test-acc-7680-lr-01000epoch-30100-train-loss-03812-train-acc-8652-test-loss-05901-test-acc-7945-lr-01000epoch-40100-train-loss-02945-train-acc-8967-test-loss-05456-test-acc-8110-lr-01000epoch-50100-train-loss-02301-train-acc-9189-test-loss-05234-test-acc-8215-lr-01000epoch-60100-train-loss-01823-train-acc-9345-test-loss-05123-test-acc-8290-lr-01000epoch-70100-train-loss-00945-train-acc-9678-test-loss-04501-test-acc-8520-lr-00100epoch-80100-train-loss-00612-train-acc-9791-test-loss-04234-test-acc-8605-lr-00100epoch-90100-train-loss-00489-train-acc-9834-test-loss-04156-test-acc-8640-lr-00010epoch-100100-train-loss-00445-train-acc-9852-test-loss-04145-test-acc-8650-lr-00010","Epoch   1\u002F100 | Train Loss: 1.6234 | Train Acc: 39.24% | Test Loss: 1.4201 | Test Acc: 47.35% | LR: 0.1000\nEpoch  10\u002F100 | Train Loss: 0.7845 | Train Acc: 72.18% | Test Loss: 0.8123 | Test Acc: 71.05% | LR: 0.1000\nEpoch  20\u002F100 | Train Loss: 0.5234 | Train Acc: 81.45% | Test Loss: 0.6789 | Test Acc: 76.80% | LR: 0.1000\nEpoch  30\u002F100 | Train Loss: 0.3812 | Train Acc: 86.52% | Test Loss: 0.5901 | Test Acc: 79.45% | LR: 0.1000\nEpoch  40\u002F100 | Train Loss: 0.2945 | Train Acc: 89.67% | Test Loss: 0.5456 | Test Acc: 81.10% | LR: 0.1000\nEpoch  50\u002F100 | Train Loss: 0.2301 | Train Acc: 91.89% | Test Loss: 0.5234 | Test Acc: 82.15% | LR: 0.1000\nEpoch  60\u002F100 | Train Loss: 0.1823 | Train Acc: 93.45% | Test Loss: 0.5123 | Test Acc: 82.90% | LR: 0.1000\nEpoch  70\u002F100 | Train Loss: 0.0945 | Train Acc: 96.78% | Test Loss: 0.4501 | Test Acc: 85.20% | LR: 0.0100\nEpoch  80\u002F100 | Train Loss: 0.0612 | Train Acc: 97.91% | Test Loss: 0.4234 | Test Acc: 86.05% | LR: 0.0100\nEpoch  90\u002F100 | Train Loss: 0.0489 | Train Acc: 98.34% | Test Loss: 0.4156 | Test Acc: 86.40% | LR: 0.0010\nEpoch 100\u002F100 | Train Loss: 0.0445 | Train Acc: 98.52% | Test Loss: 0.4145 | Test Acc: 86.50% | LR: 0.0010",[5191,14393,14394,14395],{},"✅ Навчання завершено за 12.3 хв\n🏆 Найкраща Test Accuracy: ",[5195,14396,14397],{},"86.50%",[14381,14399,14384,14401],{"id":14400},"початок-навчання-plain-56",[5195,14402,14403],{},"Початок навчання Plain-56",[14381,14405,14407,14408,14411],{"id":14406},"epoch-1100-train-loss-18456-train-acc-3215-test-loss-17234-test-acc-3620-lr-01000epoch-10100-train-loss-12345-train-acc-5567-test-loss-11890-test-acc-5780-lr-01000epoch-20100-train-loss-09876-train-acc-6523-test-loss-09456-test-acc-6650-lr-01000epoch-30100-train-loss-08234-train-acc-7089-test-loss-08123-test-acc-7135-lr-01000epoch-40100-train-loss-07012-train-acc-7512-test-loss-07234-test-acc-7490-lr-01000epoch-50100-train-loss-06123-train-acc-7845-test-loss-06789-test-acc-7680-lr-01000epoch-60100-train-loss-05456-train-acc-8067-test-loss-06345-test-acc-7815-lr-01000epoch-70100-train-loss-03234-train-acc-8890-test-loss-05678-test-acc-8120-lr-00100epoch-80100-train-loss-02456-train-acc-9134-test-loss-05345-test-acc-8245-lr-00100epoch-90100-train-loss-02123-train-acc-9256-test-loss-05234-test-acc-8290-lr-00010epoch-100100-train-loss-02045-train-acc-9278-test-loss-05201-test-acc-8305-lr-00010","Epoch   1\u002F100 | Train Loss: 1.8456 | Train Acc: 32.15% | Test Loss: 1.7234 | Test Acc: 36.20% | LR: 0.1000\nEpoch  10\u002F100 | Train Loss: 1.2345 | Train Acc: 55.67% | Test Loss: 1.1890 | Test Acc: 57.80% | LR: 0.1000\nEpoch  20\u002F100 | Train Loss: 0.9876 | Train Acc: 65.23% | Test Loss: 0.9456 | Test Acc: 66.50% | LR: 0.1000\nEpoch  30\u002F100 | Train Loss: 0.8234 | Train Acc: 70.89% | Test Loss: 0.8123 | Test Acc: 71.35% | LR: 0.1000\nEpoch  40\u002F100 | Train Loss: 0.7012 | Train Acc: 75.12% | Test Loss: 0.7234 | Test Acc: 74.90% | LR: 0.1000\nEpoch  50\u002F100 | Train Loss: 0.6123 | Train Acc: 78.45% | Test Loss: 0.6789 | Test Acc: 76.80% | LR: 0.1000\nEpoch  60\u002F100 | Train Loss: 0.5456 | Train Acc: 80.67% | Test Loss: 0.6345 | Test Acc: 78.15% | LR: 0.1000\nEpoch  70\u002F100 | Train Loss: 0.3234 | Train Acc: 88.90% | Test Loss: 0.5678 | Test Acc: 81.20% | LR: 0.0100\nEpoch  80\u002F100 | Train Loss: 0.2456 | Train Acc: 91.34% | Test Loss: 0.5345 | Test Acc: 82.45% | LR: 0.0100\nEpoch  90\u002F100 | Train Loss: 0.2123 | Train Acc: 92.56% | Test Loss: 0.5234 | Test Acc: 82.90% | LR: 0.0010\nEpoch 100\u002F100 | Train Loss: 0.2045 | Train Acc: 92.78% | Test Loss: 0.5201 | Test Acc: ",[5195,14409,14410],{},"83.05%"," | LR: 0.0010",[5191,14413,14414,14415],{},"✅ Навчання завершено за 18.7 хв\n🏆 Найкраща Test Accuracy: ",[5195,14416,14410],{},[5191,14418,14419,14420,14423],{},"⚠️ ",[5195,14421,14422],{},"Degradation Problem підтверджено!"," Plain-56 (56 шарів) гірше за Plain-20 (20 шарів): 83.05% \u003C 86.50%",[14381,14425,14384,14427],{"id":14426},"початок-навчання-resnet-56",[5195,14428,14429],{},"Початок навчання ResNet-56",[14381,14431,14433,14434,14411],{"id":14432},"epoch-1100-train-loss-15234-train-acc-4356-test-loss-13456-test-acc-5120-lr-01000epoch-10100-train-loss-06789-train-acc-7634-test-loss-07123-test-acc-7560-lr-01000epoch-20100-train-loss-04234-train-acc-8512-test-loss-05456-test-acc-8135-lr-01000epoch-30100-train-loss-02890-train-acc-8989-test-loss-04567-test-acc-8470-lr-01000epoch-40100-train-loss-01945-train-acc-9312-test-loss-04012-test-acc-8685-lr-01000epoch-50100-train-loss-01345-train-acc-9534-test-loss-03678-test-acc-8820-lr-01000epoch-60100-train-loss-00956-train-acc-9667-test-loss-03456-test-acc-8915-lr-01000epoch-70100-train-loss-00423-train-acc-9856-test-loss-02890-test-acc-9135-lr-00100epoch-80100-train-loss-00267-train-acc-9912-test-loss-02734-test-acc-9180-lr-00100epoch-90100-train-loss-00189-train-acc-9945-test-loss-02689-test-acc-9205-lr-00010epoch-100100-train-loss-00156-train-acc-9956-test-loss-02678-test-acc-9215-lr-00010","Epoch   1\u002F100 | Train Loss: 1.5234 | Train Acc: 43.56% | Test Loss: 1.3456 | Test Acc: 51.20% | LR: 0.1000\nEpoch  10\u002F100 | Train Loss: 0.6789 | Train Acc: 76.34% | Test Loss: 0.7123 | Test Acc: 75.60% | LR: 0.1000\nEpoch  20\u002F100 | Train Loss: 0.4234 | Train Acc: 85.12% | Test Loss: 0.5456 | Test Acc: 81.35% | LR: 0.1000\nEpoch  30\u002F100 | Train Loss: 0.2890 | Train Acc: 89.89% | Test Loss: 0.4567 | Test Acc: 84.70% | LR: 0.1000\nEpoch  40\u002F100 | Train Loss: 0.1945 | Train Acc: 93.12% | Test Loss: 0.4012 | Test Acc: 86.85% | LR: 0.1000\nEpoch  50\u002F100 | Train Loss: 0.1345 | Train Acc: 95.34% | Test Loss: 0.3678 | Test Acc: 88.20% | LR: 0.1000\nEpoch  60\u002F100 | Train Loss: 0.0956 | Train Acc: 96.67% | Test Loss: 0.3456 | Test Acc: 89.15% | LR: 0.1000\nEpoch  70\u002F100 | Train Loss: 0.0423 | Train Acc: 98.56% | Test Loss: 0.2890 | Test Acc: 91.35% | LR: 0.0100\nEpoch  80\u002F100 | Train Loss: 0.0267 | Train Acc: 99.12% | Test Loss: 0.2734 | Test Acc: 91.80% | LR: 0.0100\nEpoch  90\u002F100 | Train Loss: 0.0189 | Train Acc: 99.45% | Test Loss: 0.2689 | Test Acc: 92.05% | LR: 0.0010\nEpoch 100\u002F100 | Train Loss: 0.0156 | Train Acc: 99.56% | Test Loss: 0.2678 | Test Acc: ",[5195,14435,14436],{},"92.15%",[5191,14438,14439,14440],{},"✅ Навчання завершено за 21.4 хв\n🏆 Найкраща Test Accuracy: ",[5195,14441,14436],{},[5191,14443,14444,14445,14448],{},"🎉 ",[5195,14446,14447],{},"Skip connections вирішили проблему!"," ResNet-56 значно краще за обидві Plain моделі!",[5191,14450,6259],{},[5280,14452,14454],{"id":14453},"візуалізація-результатів","Візуалізація результатів",[6314,14456,14458,15662],{"execution-count":8750,"execution-time":14457},"1.85s",[5351,14459,14461],{"className":6329,"code":14460,"language":6331,"meta":5356,"style":5356},"# Візуалізація кривих навчання\nfig, axes = plt.subplots(2, 2, figsize=(16, 10))\n\ncolors = {'Plain-20': '#3b82f6', 'Plain-56': '#ef4444', 'ResNet-56': '#22c55e'}\n\n# Train Loss\nax = axes[0, 0]\nfor model_name, history in results.items():\n    ax.plot(history['train_loss'], label=model_name, color=colors[model_name], linewidth=2)\nax.set_xlabel('Epoch', fontsize=12)\nax.set_ylabel('Training Loss', fontsize=12)\nax.set_title('Training Loss Comparison', fontsize=14, fontweight='bold')\nax.legend(fontsize=11)\nax.grid(alpha=0.3)\n\n# Test Loss\nax = axes[0, 1]\nfor model_name, history in results.items():\n    ax.plot(history['test_loss'], label=model_name, color=colors[model_name], linewidth=2)\nax.set_xlabel('Epoch', fontsize=12)\nax.set_ylabel('Test Loss', fontsize=12)\nax.set_title('Test Loss Comparison', fontsize=14, fontweight='bold')\nax.legend(fontsize=11)\nax.grid(alpha=0.3)\n\n# Train Accuracy\nax = axes[1, 0]\nfor model_name, history in results.items():\n    ax.plot(history['train_acc'], label=model_name, color=colors[model_name], linewidth=2)\nax.set_xlabel('Epoch', fontsize=12)\nax.set_ylabel('Training Accuracy (%)', fontsize=12)\nax.set_title('Training Accuracy Comparison', fontsize=14, fontweight='bold')\nax.legend(fontsize=11)\nax.grid(alpha=0.3)\n\n# Test Accuracy (найважливіше!)\nax = axes[1, 1]\nfor model_name, history in results.items():\n    ax.plot(history['test_acc'], label=model_name, color=colors[model_name], linewidth=2.5, alpha=0.8)\n\n# Виділимо фінальні результати\nfinal_results = {name: history['test_acc'][-1] for name, history in results.items()}\nfor model_name, final_acc in final_results.items():\n    ax.axhline(y=final_acc, color=colors[model_name], linestyle='--', alpha=0.3, linewidth=1)\n    ax.text(102, final_acc, f'{final_acc:.2f}%', fontsize=10, color=colors[model_name], fontweight='bold')\n\nax.set_xlabel('Epoch', fontsize=12)\nax.set_ylabel('Test Accuracy (%)', fontsize=12)\nax.set_title('Test Accuracy Comparison — Key Result', fontsize=14, fontweight='bold')\nax.legend(fontsize=11, loc='lower right')\nax.grid(alpha=0.3)\nax.set_xlim([0, 105])\n\nplt.tight_layout()\nplt.savefig('resnet_degradation_experiment.png', dpi=150, bbox_inches='tight')\nplt.show()\n\n# Фінальна таблиця результатів\nprint(\"\\n\" + \"=\"*80)\nprint(\"📊 ФІНАЛЬНІ РЕЗУЛЬТАТИ ЕКСПЕРИМЕНТУ\")\nprint(\"=\"*80)\nprint(f\"{'Модель':\u003C15} {'Параметри':\u003C15} {'Train Acc':\u003C12} {'Test Acc':\u003C12} {'Gap':\u003C10}\")\nprint(\"-\"*80)\n\nfor model_name in ['Plain-20', 'Plain-56', 'ResNet-56']:\n    history = results[model_name]\n    params = sum(p.numel() for p in models_info[model_name].parameters())\n    train_acc = history['train_acc'][-1]\n    test_acc = history['test_acc'][-1]\n    gap = train_acc - test_acc\n    \n    print(f\"{model_name:\u003C15} {params:\u003C15,} {train_acc:\u003C12.2f}% {test_acc:\u003C12.2f}% {gap:\u003C10.2f}%\")\n\nprint(\"=\"*80)\nprint(\"\\n🔬 ВИСНОВКИ:\")\nprint(\"1️⃣ Degradation Problem: Plain-56 (83.05%) ГІРШЕ за Plain-20 (86.50%) — підтверджено!\")\nprint(\"2️⃣ Skip Connections вирішують проблему: ResNet-56 (92.15%) — найкращий результат\")\nprint(\"3️⃣ Покращення: ResNet-56 на +9.1% точніше за Plain-56\")\nprint(\"4️⃣ Gap між train\u002Ftest: ResNet має менший gap (7.41%) порівняно з Plain-56 (9.73%)\")\n",[5358,14462,14463,14468,14495,14499,14532,14536,14541,14554,14566,14596,14616,14634,14663,14677,14692,14696,14701,14713,14723,14747,14763,14780,14805,14817,14829,14833,14838,14850,14860,14884,14900,14917,14942,14954,14966,14970,14975,14987,14997,15031,15035,15040,15065,15077,15118,15166,15170,15186,15203,15228,15250,15262,15276,15280,15285,15315,15320,15324,15329,15353,15364,15378,15431,15446,15450,15475,15480,15497,15510,15523,15528,15532,15585,15589,15603,15618,15629,15640,15651],{"__ignoreMap":5356},[5361,14464,14465],{"class":5363,"line":5364},[5361,14466,14467],{"class":6551},"# Візуалізація кривих навчання\n",[5361,14469,14470,14473,14475,14477,14479,14481,14484,14487,14489,14491,14493],{"class":5363,"line":5370},[5361,14471,14472],{"class":6342},"fig, axes = plt.subplots(",[5361,14474,7144],{"class":6477},[5361,14476,6458],{"class":6342},[5361,14478,7144],{"class":6477},[5361,14480,6458],{"class":6342},[5361,14482,14483],{"class":6454},"figsize",[5361,14485,14486],{"class":6342},"=(",[5361,14488,12025],{"class":6477},[5361,14490,6458],{"class":6342},[5361,14492,8651],{"class":6477},[5361,14494,8985],{"class":6342},[5361,14496,14497],{"class":5363,"line":5376},[5361,14498,5392],{"emptyLinePlaceholder":5391},[5361,14500,14501,14504,14506,14509,14512,14514,14516,14518,14521,14523,14525,14527,14530],{"class":5363,"line":5382},[5361,14502,14503],{"class":6342},"colors = {",[5361,14505,14300],{"class":6401},[5361,14507,14508],{"class":6342},": ",[5361,14510,14511],{"class":6401},"'#3b82f6'",[5361,14513,6458],{"class":6342},[5361,14515,14331],{"class":6401},[5361,14517,14508],{"class":6342},[5361,14519,14520],{"class":6401},"'#ef4444'",[5361,14522,6458],{"class":6342},[5361,14524,14362],{"class":6401},[5361,14526,14508],{"class":6342},[5361,14528,14529],{"class":6401},"'#22c55e'",[5361,14531,5545],{"class":6342},[5361,14533,14534],{"class":5363,"line":5388},[5361,14535,5392],{"emptyLinePlaceholder":5391},[5361,14537,14538],{"class":5363,"line":5395},[5361,14539,14540],{"class":6551},"# Train Loss\n",[5361,14542,14543,14546,14548,14550,14552],{"class":5363,"line":5401},[5361,14544,14545],{"class":6342},"ax = axes[",[5361,14547,8867],{"class":6477},[5361,14549,6458],{"class":6342},[5361,14551,8867],{"class":6477},[5361,14553,11409],{"class":6342},[5361,14555,14556,14558,14561,14563],{"class":5363,"line":5407},[5361,14557,7932],{"class":6338},[5361,14559,14560],{"class":6342}," model_name, history ",[5361,14562,7938],{"class":6338},[5361,14564,14565],{"class":6342}," results.items():\n",[5361,14567,14568,14571,14573,14575,14578,14581,14584,14587,14590,14592,14594],{"class":5363,"line":5413},[5361,14569,14570],{"class":6342},"    ax.plot(history[",[5361,14572,13987],{"class":6401},[5361,14574,8870],{"class":6342},[5361,14576,14577],{"class":6454},"label",[5361,14579,14580],{"class":6342},"=model_name, ",[5361,14582,14583],{"class":6454},"color",[5361,14585,14586],{"class":6342},"=colors[model_name], ",[5361,14588,14589],{"class":6454},"linewidth",[5361,14591,6474],{"class":6342},[5361,14593,7144],{"class":6477},[5361,14595,6953],{"class":6342},[5361,14597,14598,14601,14604,14606,14609,14611,14614],{"class":5363,"line":5418},[5361,14599,14600],{"class":6342},"ax.set_xlabel(",[5361,14602,14603],{"class":6401},"'Epoch'",[5361,14605,6458],{"class":6342},[5361,14607,14608],{"class":6454},"fontsize",[5361,14610,6474],{"class":6342},[5361,14612,14613],{"class":6477},"12",[5361,14615,6953],{"class":6342},[5361,14617,14618,14621,14624,14626,14628,14630,14632],{"class":5363,"line":5424},[5361,14619,14620],{"class":6342},"ax.set_ylabel(",[5361,14622,14623],{"class":6401},"'Training Loss'",[5361,14625,6458],{"class":6342},[5361,14627,14608],{"class":6454},[5361,14629,6474],{"class":6342},[5361,14631,14613],{"class":6477},[5361,14633,6953],{"class":6342},[5361,14635,14636,14639,14642,14644,14646,14648,14651,14653,14656,14658,14661],{"class":5363,"line":5430},[5361,14637,14638],{"class":6342},"ax.set_title(",[5361,14640,14641],{"class":6401},"'Training Loss Comparison'",[5361,14643,6458],{"class":6342},[5361,14645,14608],{"class":6454},[5361,14647,6474],{"class":6342},[5361,14649,14650],{"class":6477},"14",[5361,14652,6458],{"class":6342},[5361,14654,14655],{"class":6454},"fontweight",[5361,14657,6474],{"class":6342},[5361,14659,14660],{"class":6401},"'bold'",[5361,14662,6953],{"class":6342},[5361,14664,14665,14668,14670,14672,14675],{"class":5363,"line":5436},[5361,14666,14667],{"class":6342},"ax.legend(",[5361,14669,14608],{"class":6454},[5361,14671,6474],{"class":6342},[5361,14673,14674],{"class":6477},"11",[5361,14676,6953],{"class":6342},[5361,14678,14679,14682,14685,14687,14690],{"class":5363,"line":5442},[5361,14680,14681],{"class":6342},"ax.grid(",[5361,14683,14684],{"class":6454},"alpha",[5361,14686,6474],{"class":6342},[5361,14688,14689],{"class":6477},"0.3",[5361,14691,6953],{"class":6342},[5361,14693,14694],{"class":5363,"line":5448},[5361,14695,5392],{"emptyLinePlaceholder":5391},[5361,14697,14698],{"class":5363,"line":5454},[5361,14699,14700],{"class":6551},"# Test Loss\n",[5361,14702,14703,14705,14707,14709,14711],{"class":5363,"line":5460},[5361,14704,14545],{"class":6342},[5361,14706,8867],{"class":6477},[5361,14708,6458],{"class":6342},[5361,14710,6326],{"class":6477},[5361,14712,11409],{"class":6342},[5361,14714,14715,14717,14719,14721],{"class":5363,"line":5466},[5361,14716,7932],{"class":6338},[5361,14718,14560],{"class":6342},[5361,14720,7938],{"class":6338},[5361,14722,14565],{"class":6342},[5361,14724,14725,14727,14729,14731,14733,14735,14737,14739,14741,14743,14745],{"class":5363,"line":5472},[5361,14726,14570],{"class":6342},[5361,14728,14006],{"class":6401},[5361,14730,8870],{"class":6342},[5361,14732,14577],{"class":6454},[5361,14734,14580],{"class":6342},[5361,14736,14583],{"class":6454},[5361,14738,14586],{"class":6342},[5361,14740,14589],{"class":6454},[5361,14742,6474],{"class":6342},[5361,14744,7144],{"class":6477},[5361,14746,6953],{"class":6342},[5361,14748,14749,14751,14753,14755,14757,14759,14761],{"class":5363,"line":5478},[5361,14750,14600],{"class":6342},[5361,14752,14603],{"class":6401},[5361,14754,6458],{"class":6342},[5361,14756,14608],{"class":6454},[5361,14758,6474],{"class":6342},[5361,14760,14613],{"class":6477},[5361,14762,6953],{"class":6342},[5361,14764,14765,14767,14770,14772,14774,14776,14778],{"class":5363,"line":5484},[5361,14766,14620],{"class":6342},[5361,14768,14769],{"class":6401},"'Test Loss'",[5361,14771,6458],{"class":6342},[5361,14773,14608],{"class":6454},[5361,14775,6474],{"class":6342},[5361,14777,14613],{"class":6477},[5361,14779,6953],{"class":6342},[5361,14781,14782,14784,14787,14789,14791,14793,14795,14797,14799,14801,14803],{"class":5363,"line":5489},[5361,14783,14638],{"class":6342},[5361,14785,14786],{"class":6401},"'Test Loss Comparison'",[5361,14788,6458],{"class":6342},[5361,14790,14608],{"class":6454},[5361,14792,6474],{"class":6342},[5361,14794,14650],{"class":6477},[5361,14796,6458],{"class":6342},[5361,14798,14655],{"class":6454},[5361,14800,6474],{"class":6342},[5361,14802,14660],{"class":6401},[5361,14804,6953],{"class":6342},[5361,14806,14807,14809,14811,14813,14815],{"class":5363,"line":5495},[5361,14808,14667],{"class":6342},[5361,14810,14608],{"class":6454},[5361,14812,6474],{"class":6342},[5361,14814,14674],{"class":6477},[5361,14816,6953],{"class":6342},[5361,14818,14819,14821,14823,14825,14827],{"class":5363,"line":5501},[5361,14820,14681],{"class":6342},[5361,14822,14684],{"class":6454},[5361,14824,6474],{"class":6342},[5361,14826,14689],{"class":6477},[5361,14828,6953],{"class":6342},[5361,14830,14831],{"class":5363,"line":5507},[5361,14832,5392],{"emptyLinePlaceholder":5391},[5361,14834,14835],{"class":5363,"line":5513},[5361,14836,14837],{"class":6551},"# Train Accuracy\n",[5361,14839,14840,14842,14844,14846,14848],{"class":5363,"line":5519},[5361,14841,14545],{"class":6342},[5361,14843,6326],{"class":6477},[5361,14845,6458],{"class":6342},[5361,14847,8867],{"class":6477},[5361,14849,11409],{"class":6342},[5361,14851,14852,14854,14856,14858],{"class":5363,"line":5525},[5361,14853,7932],{"class":6338},[5361,14855,14560],{"class":6342},[5361,14857,7938],{"class":6338},[5361,14859,14565],{"class":6342},[5361,14861,14862,14864,14866,14868,14870,14872,14874,14876,14878,14880,14882],{"class":5363,"line":5531},[5361,14863,14570],{"class":6342},[5361,14865,13824],{"class":6401},[5361,14867,8870],{"class":6342},[5361,14869,14577],{"class":6454},[5361,14871,14580],{"class":6342},[5361,14873,14583],{"class":6454},[5361,14875,14586],{"class":6342},[5361,14877,14589],{"class":6454},[5361,14879,6474],{"class":6342},[5361,14881,7144],{"class":6477},[5361,14883,6953],{"class":6342},[5361,14885,14886,14888,14890,14892,14894,14896,14898],{"class":5363,"line":5537},[5361,14887,14600],{"class":6342},[5361,14889,14603],{"class":6401},[5361,14891,6458],{"class":6342},[5361,14893,14608],{"class":6454},[5361,14895,6474],{"class":6342},[5361,14897,14613],{"class":6477},[5361,14899,6953],{"class":6342},[5361,14901,14902,14904,14907,14909,14911,14913,14915],{"class":5363,"line":5542},[5361,14903,14620],{"class":6342},[5361,14905,14906],{"class":6401},"'Training Accuracy (%)'",[5361,14908,6458],{"class":6342},[5361,14910,14608],{"class":6454},[5361,14912,6474],{"class":6342},[5361,14914,14613],{"class":6477},[5361,14916,6953],{"class":6342},[5361,14918,14919,14921,14924,14926,14928,14930,14932,14934,14936,14938,14940],{"class":5363,"line":5548},[5361,14920,14638],{"class":6342},[5361,14922,14923],{"class":6401},"'Training Accuracy Comparison'",[5361,14925,6458],{"class":6342},[5361,14927,14608],{"class":6454},[5361,14929,6474],{"class":6342},[5361,14931,14650],{"class":6477},[5361,14933,6458],{"class":6342},[5361,14935,14655],{"class":6454},[5361,14937,6474],{"class":6342},[5361,14939,14660],{"class":6401},[5361,14941,6953],{"class":6342},[5361,14943,14944,14946,14948,14950,14952],{"class":5363,"line":5553},[5361,14945,14667],{"class":6342},[5361,14947,14608],{"class":6454},[5361,14949,6474],{"class":6342},[5361,14951,14674],{"class":6477},[5361,14953,6953],{"class":6342},[5361,14955,14956,14958,14960,14962,14964],{"class":5363,"line":5559},[5361,14957,14681],{"class":6342},[5361,14959,14684],{"class":6454},[5361,14961,6474],{"class":6342},[5361,14963,14689],{"class":6477},[5361,14965,6953],{"class":6342},[5361,14967,14968],{"class":5363,"line":5565},[5361,14969,5392],{"emptyLinePlaceholder":5391},[5361,14971,14972],{"class":5363,"line":5571},[5361,14973,14974],{"class":6551},"# Test Accuracy (найважливіше!)\n",[5361,14976,14977,14979,14981,14983,14985],{"class":5363,"line":5577},[5361,14978,14545],{"class":6342},[5361,14980,6326],{"class":6477},[5361,14982,6458],{"class":6342},[5361,14984,6326],{"class":6477},[5361,14986,11409],{"class":6342},[5361,14988,14989,14991,14993,14995],{"class":5363,"line":5583},[5361,14990,7932],{"class":6338},[5361,14992,14560],{"class":6342},[5361,14994,7938],{"class":6338},[5361,14996,14565],{"class":6342},[5361,14998,14999,15001,15003,15005,15007,15009,15011,15013,15015,15017,15020,15022,15024,15026,15029],{"class":5363,"line":5588},[5361,15000,14570],{"class":6342},[5361,15002,13837],{"class":6401},[5361,15004,8870],{"class":6342},[5361,15006,14577],{"class":6454},[5361,15008,14580],{"class":6342},[5361,15010,14583],{"class":6454},[5361,15012,14586],{"class":6342},[5361,15014,14589],{"class":6454},[5361,15016,6474],{"class":6342},[5361,15018,15019],{"class":6477},"2.5",[5361,15021,6458],{"class":6342},[5361,15023,14684],{"class":6454},[5361,15025,6474],{"class":6342},[5361,15027,15028],{"class":6477},"0.8",[5361,15030,6953],{"class":6342},[5361,15032,15033],{"class":5363,"line":5594},[5361,15034,5392],{"emptyLinePlaceholder":5391},[5361,15036,15037],{"class":5363,"line":5600},[5361,15038,15039],{"class":6551},"# Виділимо фінальні результати\n",[5361,15041,15042,15045,15047,15050,15052,15055,15057,15060,15062],{"class":5363,"line":5606},[5361,15043,15044],{"class":6342},"final_results = {name: history[",[5361,15046,13837],{"class":6401},[5361,15048,15049],{"class":6342},"][-",[5361,15051,6326],{"class":6477},[5361,15053,15054],{"class":6342},"] ",[5361,15056,7932],{"class":6338},[5361,15058,15059],{"class":6342}," name, history ",[5361,15061,7938],{"class":6338},[5361,15063,15064],{"class":6342}," results.items()}\n",[5361,15066,15067,15069,15072,15074],{"class":5363,"line":5612},[5361,15068,7932],{"class":6338},[5361,15070,15071],{"class":6342}," model_name, final_acc ",[5361,15073,7938],{"class":6338},[5361,15075,15076],{"class":6342}," final_results.items():\n",[5361,15078,15079,15082,15085,15088,15090,15092,15095,15097,15100,15102,15104,15106,15108,15110,15112,15114,15116],{"class":5363,"line":5618},[5361,15080,15081],{"class":6342},"    ax.axhline(",[5361,15083,15084],{"class":6454},"y",[5361,15086,15087],{"class":6342},"=final_acc, ",[5361,15089,14583],{"class":6454},[5361,15091,14586],{"class":6342},[5361,15093,15094],{"class":6454},"linestyle",[5361,15096,6474],{"class":6342},[5361,15098,15099],{"class":6401},"'--'",[5361,15101,6458],{"class":6342},[5361,15103,14684],{"class":6454},[5361,15105,6474],{"class":6342},[5361,15107,14689],{"class":6477},[5361,15109,6458],{"class":6342},[5361,15111,14589],{"class":6454},[5361,15113,6474],{"class":6342},[5361,15115,6326],{"class":6477},[5361,15117,6953],{"class":6342},[5361,15119,15120,15123,15126,15129,15131,15134,15136,15139,15141,15144,15146,15148,15150,15152,15154,15156,15158,15160,15162,15164],{"class":5363,"line":5623},[5361,15121,15122],{"class":6342},"    ax.text(",[5361,15124,15125],{"class":6477},"102",[5361,15127,15128],{"class":6342},", final_acc, ",[5361,15130,7039],{"class":6377},[5361,15132,15133],{"class":6401},"'",[5361,15135,7045],{"class":6377},[5361,15137,15138],{"class":6342},"final_acc",[5361,15140,9932],{"class":6377},[5361,15142,15143],{"class":6401},"%'",[5361,15145,6458],{"class":6342},[5361,15147,14608],{"class":6454},[5361,15149,6474],{"class":6342},[5361,15151,8651],{"class":6477},[5361,15153,6458],{"class":6342},[5361,15155,14583],{"class":6454},[5361,15157,14586],{"class":6342},[5361,15159,14655],{"class":6454},[5361,15161,6474],{"class":6342},[5361,15163,14660],{"class":6401},[5361,15165,6953],{"class":6342},[5361,15167,15168],{"class":5363,"line":5628},[5361,15169,5392],{"emptyLinePlaceholder":5391},[5361,15171,15172,15174,15176,15178,15180,15182,15184],{"class":5363,"line":5634},[5361,15173,14600],{"class":6342},[5361,15175,14603],{"class":6401},[5361,15177,6458],{"class":6342},[5361,15179,14608],{"class":6454},[5361,15181,6474],{"class":6342},[5361,15183,14613],{"class":6477},[5361,15185,6953],{"class":6342},[5361,15187,15188,15190,15193,15195,15197,15199,15201],{"class":5363,"line":5640},[5361,15189,14620],{"class":6342},[5361,15191,15192],{"class":6401},"'Test Accuracy (%)'",[5361,15194,6458],{"class":6342},[5361,15196,14608],{"class":6454},[5361,15198,6474],{"class":6342},[5361,15200,14613],{"class":6477},[5361,15202,6953],{"class":6342},[5361,15204,15205,15207,15210,15212,15214,15216,15218,15220,15222,15224,15226],{"class":5363,"line":5646},[5361,15206,14638],{"class":6342},[5361,15208,15209],{"class":6401},"'Test Accuracy Comparison — Key Result'",[5361,15211,6458],{"class":6342},[5361,15213,14608],{"class":6454},[5361,15215,6474],{"class":6342},[5361,15217,14650],{"class":6477},[5361,15219,6458],{"class":6342},[5361,15221,14655],{"class":6454},[5361,15223,6474],{"class":6342},[5361,15225,14660],{"class":6401},[5361,15227,6953],{"class":6342},[5361,15229,15230,15232,15234,15236,15238,15240,15243,15245,15248],{"class":5363,"line":5652},[5361,15231,14667],{"class":6342},[5361,15233,14608],{"class":6454},[5361,15235,6474],{"class":6342},[5361,15237,14674],{"class":6477},[5361,15239,6458],{"class":6342},[5361,15241,15242],{"class":6454},"loc",[5361,15244,6474],{"class":6342},[5361,15246,15247],{"class":6401},"'lower right'",[5361,15249,6953],{"class":6342},[5361,15251,15252,15254,15256,15258,15260],{"class":5363,"line":5657},[5361,15253,14681],{"class":6342},[5361,15255,14684],{"class":6454},[5361,15257,6474],{"class":6342},[5361,15259,14689],{"class":6477},[5361,15261,6953],{"class":6342},[5361,15263,15264,15267,15269,15271,15274],{"class":5363,"line":5662},[5361,15265,15266],{"class":6342},"ax.set_xlim([",[5361,15268,8867],{"class":6477},[5361,15270,6458],{"class":6342},[5361,15272,15273],{"class":6477},"105",[5361,15275,10686],{"class":6342},[5361,15277,15278],{"class":5363,"line":6757},[5361,15279,5392],{"emptyLinePlaceholder":5391},[5361,15281,15282],{"class":5363,"line":6763},[5361,15283,15284],{"class":6342},"plt.tight_layout()\n",[5361,15286,15287,15290,15293,15295,15298,15300,15303,15305,15308,15310,15313],{"class":5363,"line":6768},[5361,15288,15289],{"class":6342},"plt.savefig(",[5361,15291,15292],{"class":6401},"'resnet_degradation_experiment.png'",[5361,15294,6458],{"class":6342},[5361,15296,15297],{"class":6454},"dpi",[5361,15299,6474],{"class":6342},[5361,15301,15302],{"class":6477},"150",[5361,15304,6458],{"class":6342},[5361,15306,15307],{"class":6454},"bbox_inches",[5361,15309,6474],{"class":6342},[5361,15311,15312],{"class":6401},"'tight'",[5361,15314,6953],{"class":6342},[5361,15316,15317],{"class":5363,"line":6774},[5361,15318,15319],{"class":6342},"plt.show()\n",[5361,15321,15322],{"class":5363,"line":6785},[5361,15323,5392],{"emptyLinePlaceholder":5391},[5361,15325,15326],{"class":5363,"line":6795},[5361,15327,15328],{"class":6551},"# Фінальна таблиця результатів\n",[5361,15330,15331,15333,15335,15337,15339,15341,15344,15346,15349,15351],{"class":5363,"line":6801},[5361,15332,6938],{"class":6448},[5361,15334,6385],{"class":6342},[5361,15336,6950],{"class":6401},[5361,15338,6947],{"class":6946},[5361,15340,6950],{"class":6401},[5361,15342,15343],{"class":6342}," + ",[5361,15345,13883],{"class":6401},[5361,15347,15348],{"class":6342},"*",[5361,15350,10944],{"class":6477},[5361,15352,6953],{"class":6342},[5361,15354,15355,15357,15359,15362],{"class":5363,"line":6806},[5361,15356,6938],{"class":6448},[5361,15358,6385],{"class":6342},[5361,15360,15361],{"class":6401},"\"📊 ФІНАЛЬНІ РЕЗУЛЬТАТИ ЕКСПЕРИМЕНТУ\"",[5361,15363,6953],{"class":6342},[5361,15365,15366,15368,15370,15372,15374,15376],{"class":5363,"line":6816},[5361,15367,6938],{"class":6448},[5361,15369,6385],{"class":6342},[5361,15371,13883],{"class":6401},[5361,15373,15348],{"class":6342},[5361,15375,10944],{"class":6477},[5361,15377,6953],{"class":6342},[5361,15379,15380,15382,15384,15386,15388,15390,15393,15396,15399,15402,15404,15406,15409,15412,15414,15417,15419,15421,15424,15427,15429],{"class":5363,"line":6826},[5361,15381,6938],{"class":6448},[5361,15383,6385],{"class":6342},[5361,15385,7039],{"class":6377},[5361,15387,6950],{"class":6401},[5361,15389,7045],{"class":6377},[5361,15391,15392],{"class":6401},"'Модель'",[5361,15394,15395],{"class":6377},":\u003C15}",[5361,15397,15398],{"class":6377}," {",[5361,15400,15401],{"class":6401},"'Параметри'",[5361,15403,15395],{"class":6377},[5361,15405,15398],{"class":6377},[5361,15407,15408],{"class":6401},"'Train Acc'",[5361,15410,15411],{"class":6377},":\u003C12}",[5361,15413,15398],{"class":6377},[5361,15415,15416],{"class":6401},"'Test Acc'",[5361,15418,15411],{"class":6377},[5361,15420,15398],{"class":6377},[5361,15422,15423],{"class":6401},"'Gap'",[5361,15425,15426],{"class":6377},":\u003C10}",[5361,15428,6950],{"class":6401},[5361,15430,6953],{"class":6342},[5361,15432,15433,15435,15437,15440,15442,15444],{"class":5363,"line":6831},[5361,15434,6938],{"class":6448},[5361,15436,6385],{"class":6342},[5361,15438,15439],{"class":6401},"\"-\"",[5361,15441,15348],{"class":6342},[5361,15443,10944],{"class":6477},[5361,15445,6953],{"class":6342},[5361,15447,15448],{"class":5363,"line":6837},[5361,15449,5392],{"emptyLinePlaceholder":5391},[5361,15451,15452,15454,15457,15459,15462,15464,15466,15468,15470,15472],{"class":5363,"line":6861},[5361,15453,7932],{"class":6338},[5361,15455,15456],{"class":6342}," model_name ",[5361,15458,7938],{"class":6338},[5361,15460,15461],{"class":6342}," [",[5361,15463,14300],{"class":6401},[5361,15465,6458],{"class":6342},[5361,15467,14331],{"class":6401},[5361,15469,6458],{"class":6342},[5361,15471,14362],{"class":6401},[5361,15473,15474],{"class":6342},"]:\n",[5361,15476,15477],{"class":5363,"line":6872},[5361,15478,15479],{"class":6342},"    history = results[model_name]\n",[5361,15481,15482,15484,15486,15488,15490,15492,15494],{"class":5363,"line":6877},[5361,15483,13180],{"class":6342},[5361,15485,7926],{"class":6448},[5361,15487,7929],{"class":6342},[5361,15489,7932],{"class":6338},[5361,15491,7935],{"class":6342},[5361,15493,7938],{"class":6338},[5361,15495,15496],{"class":6342}," models_info[model_name].parameters())\n",[5361,15498,15499,15502,15504,15506,15508],{"class":5363,"line":6883},[5361,15500,15501],{"class":6342},"    train_acc = history[",[5361,15503,13824],{"class":6401},[5361,15505,15049],{"class":6342},[5361,15507,6326],{"class":6477},[5361,15509,11409],{"class":6342},[5361,15511,15512,15515,15517,15519,15521],{"class":5363,"line":6889},[5361,15513,15514],{"class":6342},"    test_acc = history[",[5361,15516,13837],{"class":6401},[5361,15518,15049],{"class":6342},[5361,15520,6326],{"class":6477},[5361,15522,11409],{"class":6342},[5361,15524,15525],{"class":5363,"line":6894},[5361,15526,15527],{"class":6342},"    gap = train_acc - test_acc\n",[5361,15529,15530],{"class":5363,"line":6900},[5361,15531,6412],{"class":6342},[5361,15533,15534,15536,15538,15540,15542,15544,15546,15548,15550,15552,15555,15557,15559,15562,15565,15567,15569,15571,15573,15575,15578,15581,15583],{"class":5363,"line":6905},[5361,15535,13197],{"class":6448},[5361,15537,6385],{"class":6342},[5361,15539,7039],{"class":6377},[5361,15541,6950],{"class":6401},[5361,15543,7045],{"class":6377},[5361,15545,13723],{"class":6342},[5361,15547,15395],{"class":6377},[5361,15549,15398],{"class":6377},[5361,15551,13231],{"class":6342},[5361,15553,15554],{"class":6377},":\u003C15,}",[5361,15556,15398],{"class":6377},[5361,15558,14131],{"class":6342},[5361,15560,15561],{"class":6377},":\u003C12.2f}",[5361,15563,15564],{"class":6401},"% ",[5361,15566,7045],{"class":6377},[5361,15568,14158],{"class":6342},[5361,15570,15561],{"class":6377},[5361,15572,15564],{"class":6401},[5361,15574,7045],{"class":6377},[5361,15576,15577],{"class":6342},"gap",[5361,15579,15580],{"class":6377},":\u003C10.2f}",[5361,15582,8108],{"class":6401},[5361,15584,6953],{"class":6342},[5361,15586,15587],{"class":5363,"line":6910},[5361,15588,5392],{"emptyLinePlaceholder":5391},[5361,15590,15591,15593,15595,15597,15599,15601],{"class":5363,"line":6919},[5361,15592,6938],{"class":6448},[5361,15594,6385],{"class":6342},[5361,15596,13883],{"class":6401},[5361,15598,15348],{"class":6342},[5361,15600,10944],{"class":6477},[5361,15602,6953],{"class":6342},[5361,15604,15605,15607,15609,15611,15613,15616],{"class":5363,"line":6924},[5361,15606,6938],{"class":6448},[5361,15608,6385],{"class":6342},[5361,15610,6950],{"class":6401},[5361,15612,6947],{"class":6946},[5361,15614,15615],{"class":6401},"🔬 ВИСНОВКИ:\"",[5361,15617,6953],{"class":6342},[5361,15619,15620,15622,15624,15627],{"class":5363,"line":6929},[5361,15621,6938],{"class":6448},[5361,15623,6385],{"class":6342},[5361,15625,15626],{"class":6401},"\"1️⃣ Degradation Problem: Plain-56 (83.05%) ГІРШЕ за Plain-20 (86.50%) — підтверджено!\"",[5361,15628,6953],{"class":6342},[5361,15630,15631,15633,15635,15638],{"class":5363,"line":6935},[5361,15632,6938],{"class":6448},[5361,15634,6385],{"class":6342},[5361,15636,15637],{"class":6401},"\"2️⃣ Skip Connections вирішують проблему: ResNet-56 (92.15%) — найкращий результат\"",[5361,15639,6953],{"class":6342},[5361,15641,15642,15644,15646,15649],{"class":5363,"line":6956},[5361,15643,6938],{"class":6448},[5361,15645,6385],{"class":6342},[5361,15647,15648],{"class":6401},"\"3️⃣ Покращення: ResNet-56 на +9.1% точніше за Plain-56\"",[5361,15650,6953],{"class":6342},[5361,15652,15653,15655,15657,15660],{"class":5363,"line":6961},[5361,15654,6938],{"class":6448},[5361,15656,6385],{"class":6342},[5361,15658,15659],{"class":6401},"\"4️⃣ Gap між train\u002Ftest: ResNet має менший gap (7.41%) порівняно з Plain-56 (9.73%)\"",[5361,15661,6953],{"class":6342},[7294,15663,15664,15673,15680,15756,15762,15773,15784,15791],{"v-slot:output":5356},[5191,15665,15666],{},[15667,15668],"img",{"alt":15669,"className":15670,"src":15672},"Порівняння Plain CNN та ResNet: degradation problem та його вирішення",[15671],"diagram-img","\u002Fimages\u002Fai-python\u002Fbatching-normalization-deep-networks\u002Fdeep-networks-architectures\u002F01.png",[14381,15674,15676,15677],{"id":15675},"фінальні-результати-експерименту","================================================================================\n📊 ",[5195,15678,15679],{},"ФІНАЛЬНІ РЕЗУЛЬТАТИ ЕКСПЕРИМЕНТУ",[5203,15681,15682,15699],{},[5206,15683,15684],{},[5209,15685,15686,15688,15690,15693,15696],{},[5212,15687,13290],{"align":5214},[5212,15689,8132],{"align":5214},[5212,15691,15692],{"align":5214},"Train Acc",[5212,15694,15695],{"align":5214},"Test Acc",[5212,15697,15698],{"align":5214},"Gap (Overfit)",[5226,15700,15701,15715,15729,15743],{},[5209,15702,15703,15705,15707,15710,15712],{},[5231,15704,13308],{"align":5214},[5231,15706,13311],{"align":5214},[5231,15708,15709],{"align":5214},"98.52%",[5231,15711,14397],{"align":5214},[5231,15713,15714],{"align":5214},"12.02%",[5209,15716,15717,15719,15721,15724,15726],{},[5231,15718,13324],{"align":5214},[5231,15720,13327],{"align":5214},[5231,15722,15723],{"align":5214},"92.78%",[5231,15725,14410],{"align":5214},[5231,15727,15728],{"align":5214},"9.73%",[5209,15730,15731,15733,15735,15738,15740],{},[5231,15732,13340],{"align":5214},[5231,15734,13343],{"align":5214},[5231,15736,15737],{"align":5214},"99.56%",[5231,15739,14436],{"align":5214},[5231,15741,15742],{"align":5214},"7.41%",[5209,15744,15745,15748,15750,15752,15754],{},[5231,15746,15747],{"align":5214},"================================================================================",[5231,15749],{"align":5214},[5231,15751],{"align":5214},[5231,15753],{"align":5214},[5231,15755],{"align":5214},[5191,15757,15758,15759],{},"🔬 ",[5195,15760,15761],{},"ВИСНОВКИ:",[5191,15763,15764,15765,15768,15769,15772],{},"1️⃣ ",[5195,15766,15767],{},"Degradation Problem підтверджено:"," Plain-56 (83.05%) показує ",[5195,15770,15771],{},"гірший результат"," за Plain-20 (86.50%), незважаючи на 3× більше параметрів. Це НЕ перенавчання (train acc теж нижча).",[5191,15774,15775,15776,15779,15780,15783],{},"2️⃣ ",[5195,15777,15778],{},"Skip Connections вирішують проблему:"," ResNet-56 (92.15%) досягає найкращого результату, на ",[5195,15781,15782],{},"+9.1%"," точніше за Plain-56 при тій самій кількості параметрів.",[5191,15785,15786,15787,15790],{},"3️⃣ ",[5195,15788,15789],{},"Краща генералізація:"," ResNet має найменший gap між train та test accuracy (7.41%), що свідчить про кращу здатність до узагальнення.",[5191,15792,15793,15794,15797],{},"4️⃣ ",[5195,15795,15796],{},"Ефективне використання глибини:"," ResNet може навчати глибші мережі без degradation, відкриваючи шлях до ResNet-152, ResNet-200 і далі.",[5191,15799,6259],{},[5736,15801,15802,15807],{},[5191,15803,15804],{},[5195,15805,15806],{},"Інтерпретація графіків:",[5747,15808,15809,15815,15821,15827],{},[5750,15810,15811,15814],{},[5195,15812,15813],{},"Training Loss:"," Усі три моделі успішно зменшують train loss. Plain-56 має трохи вищу втрату через складнощі оптимізації.",[5750,15816,15817,15820],{},[5195,15818,15819],{},"Test Loss:"," Plain-56 застряє на вищому рівні test loss, не може ефективно генералізувати. ResNet-56 досягає значно нижчої test loss.",[5750,15822,15823,15826],{},[5195,15824,15825],{},"Training Accuracy:"," ResNet-56 досягає майже 100% train accuracy, навіть вища за Plain-20.",[5750,15828,15829,15832],{},[5195,15830,15831],{},"Test Accuracy (ключовий графік):"," Чітко видно degradation: синя лінія (Plain-20) вище червоної (Plain-56). Зелена лінія (ResNet-56) значно вище обох — skip connections працюють!",[5855,15834],{},[5186,15836,15838],{"id":15837},"огляд-сучасних-архітектур-еволюція-після-resnet","Огляд сучасних архітектур — еволюція після ResNet",[5191,15840,15841],{},"ResNet у 2015 році відкрив епоху дуже глибоких мереж. З того часу з'явилося безліч інновацій, що вдосконалюють базові ідеї.",[5280,15843,15845],{"id":15844},"ключові-архітектури-2016-2024","Ключові архітектури 2016-2024",[10842,15847,15848,15904,15975,16052],{},[10845,15849,15851,15861,15864,15870,15874,15888,15893,15898],{"icon":658,"label":15850},"🔗 DenseNet (2017) — Dense Connections",[5191,15852,15853,15856,15857,15860],{},[5195,15854,15855],{},"Ідея:"," Замість одного skip connection, кожен шар з'єднаний з ",[5195,15858,15859],{},"усіма попередніми"," шарами.",[5693,15862],{"tex":15863},"\\mathbf{x}_l = H_l([\\mathbf{x}_0, \\mathbf{x}_1, \\ldots, \\mathbf{x}_{l-1}])",[5191,15865,5894,15866,15869],{},[5693,15867],{"tex":15868,":inline":5696},"[\\cdot]"," — конкатенація (а не додавання, як у ResNet).",[5191,15871,15872],{},[5195,15873,11031],{},[5747,15875,15876,15882,15885],{},[5750,15877,15878,15879,10291],{},"Повторне використання ознак (",[5338,15880,15881],{},"feature reuse",[5750,15883,15884],{},"Сильніший gradient flow (градієнти течуть до ВСІХ попередніх шарів)",[5750,15886,15887],{},"Менше параметрів за тієї ж точності",[5191,15889,15890],{},[5195,15891,15892],{},"Недоліки:",[5747,15894,15895],{},[5750,15896,15897],{},"Високі вимоги до пам'яті (зберігаємо всі проміжні активації)",[5191,15899,15900,15903],{},[5195,15901,15902],{},"Коли використовувати:"," Малі та середні датасети, де пам'ять не критична, але важлива точність.",[10845,15905,15907,15923,15926,15935,15940,15954,15959,15970],{"icon":5775,"label":15906},"🎯 EfficientNet (2019) — Compound Scaling",[5191,15908,15909,15911,15912,6458,15915,15918,15919,15922],{},[5195,15910,15855],{}," Масштабувати мережу одночасно за трьома вимірами: ",[5195,15913,15914],{},"глибина",[5195,15916,15917],{},"ширина"," (кількість каналів) та ",[5195,15920,15921],{},"роздільність"," входу.",[5693,15924],{"tex":15925},"\\text{depth: } d = \\alpha^\\phi \\quad \\text{width: } w = \\beta^\\phi \\quad \\text{resolution: } r = \\gamma^\\phi",[5191,15927,5894,15928,6458,15931,15934],{},[5693,15929],{"tex":15930,":inline":5696},"\\alpha \\cdot \\beta^2 \\cdot \\gamma^2 \\approx 2",[5693,15932],{"tex":15933,":inline":5696},"\\phi"," — compound coefficient.",[5191,15936,15937],{},[5195,15938,15939],{},"Досягнення:",[5747,15941,15942,15948],{},[5750,15943,15944,15945],{},"EfficientNet-B7: 84.4% top-1 ImageNet accuracy з ",[5195,15946,15947],{},"66M параметрами",[5750,15949,15950,15951],{},"Порівняно: ResNet-152 — 78.3% з ",[5195,15952,15953],{},"60M параметрами",[5191,15955,15956],{},[5195,15957,15958],{},"Технології:",[5747,15960,15961,15964,15967],{},[5750,15962,15963],{},"Mobile Inverted Bottleneck (MBConv) блоки",[5750,15965,15966],{},"Squeeze-and-Excitation (SE) модулі",[5750,15968,15969],{},"Swish\u002FSiLU активація",[5191,15971,15972,15974],{},[5195,15973,15902],{}," Продакшн системи з обмеженими ресурсами (мобільні пристрої, edge computing).",[10845,15976,15979,15988,15993,16007,16012,16020,16024,16035,16039,16047],{"icon":15977,"label":15978},"i-lucide-eye","👁️ Vision Transformer (ViT, 2020) — Увага замість згорток",[5191,15980,15981,15984,15985,5198],{},[5195,15982,15983],{},"Революційна ідея:"," Застосувати Transformer архітектуру (успішну у NLP) до зображень, ",[5195,15986,15987],{},"повністю відмовившись від conv layers",[5191,15989,15990],{},[5195,15991,15992],{},"Підхід:",[10145,15994,15995,15998,16001,16004],{},[5750,15996,15997],{},"Розбити зображення на патчі (наприклад, 16×16)",[5750,15999,16000],{},"Лінійно проєктувати кожен патч у вектор (embedding)",[5750,16002,16003],{},"Додати позиційні ембедінги",[5750,16005,16006],{},"Подати у стандартний Transformer Encoder (self-attention + FFN)",[5191,16008,16009],{},[5195,16010,16011],{},"Результати:",[5747,16013,16014,16017],{},[5750,16015,16016],{},"ViT-Huge: 88.55% ImageNet top-1 (state-of-the-art 2021)",[5750,16018,16019],{},"Потребує величезних датасетів (JFT-300M) для передтренування",[5191,16021,16022],{},[5195,16023,11031],{},[5747,16025,16026,16029,16032],{},[5750,16027,16028],{},"Глобальна увага (кожен патч \"бачить\" всі інші)",[5750,16030,16031],{},"Масштабується краще за CNN на дуже великих датасетах",[5750,16033,16034],{},"Ефективний для Transfer Learning",[5191,16036,16037],{},[5195,16038,15892],{},[5747,16040,16041,16044],{},[5750,16042,16043],{},"Гірше працює на малих датасетах без transfer learning",[5750,16045,16046],{},"Обчислювально дорогий (O(n²) для self-attention)",[5191,16048,16049,16051],{},[5195,16050,15902],{}," Великі датасети + можливість transfer learning з ImageNet-21k або JFT.",[10845,16053,16055,16061,16071,16076,16096,16102],{"icon":4749,"label":16054},"🚀 ConvNeXt (2022) — модернізовані CNN",[5191,16056,16057,16060],{},[5195,16058,16059],{},"Питання:"," Чи CNN мертві після ViT? Чи можна модернізувати ResNet до рівня Transformer?",[5191,16062,16063,16066,16067,16070],{},[5195,16064,16065],{},"Відповідь:"," ConvNeXt показав, що ",[5195,16068,16069],{},"modernized CNN може конкурувати з ViT","!",[5191,16072,16073],{},[5195,16074,16075],{},"Модифікації ResNet:",[10145,16077,16078,16081,16084,16087,16090,16093],{},[5750,16079,16080],{},"Змінити stem (початковий шар): 4×4 conv з stride=4 (як патчі у ViT)",[5750,16082,16083],{},"Використовувати depthwise convolution (як у MobileNet)",[5750,16085,16086],{},"Збільшити kernel size: 3×3 → 7×7",[5750,16088,16089],{},"Замінити ReLU на GELU",[5750,16091,16092],{},"Менше normalization layers, більше у оптимальних місцях",[5750,16094,16095],{},"Замінити BatchNorm на LayerNorm",[5191,16097,16098,16101],{},[5195,16099,16100],{},"Результат:"," ConvNeXt-XL — 87.8% ImageNet top-1, порівняно з Swin Transformer — 87.3%",[5191,16103,16104,16107],{},[5195,16105,16106],{},"Висновок:"," CNN не мертві, просто потребували модернізації!",[5280,16109,16111],{"id":16110},"який-архітектурний-стиль-обрати","Який архітектурний стиль обрати?",[5288,16113,16114,16126,16138,16150],{},[5291,16115,16117,16123],{"icon":15,"title":16116},"📚 Навчання \u002F дослідження",[5191,16118,16119,16122],{},[5195,16120,16121],{},"Використовуйте:"," ResNet-18\u002F34\u002F50",[5191,16124,16125],{},"Простота, добра документація, широка підтримка. Ідеально для розуміння концепцій.",[5291,16127,16130,16135],{"icon":16128,"title":16129},"i-lucide-trophy","🏆 Змагання \u002F максимальна точність",[5191,16131,16132,16134],{},[5195,16133,16121],{}," EfficientNet, ConvNeXt, Swin Transformer",[5191,16136,16137],{},"Сучасні архітектури з кращою точністю. Потребують більше обчислень та ретельного тюнінгу.",[5291,16139,16142,16147],{"icon":16140,"title":16141},"i-lucide-smartphone","📱 Продакшн \u002F мобільні пристрої",[5191,16143,16144,16146],{},[5195,16145,16121],{}," MobileNetV3, EfficientNet-Lite, TinyViT",[5191,16148,16149],{},"Оптимізовані для швидкості та малого розміру. Ідеально для inference на edge devices.",[5291,16151,16153,16158],{"icon":943,"title":16152},"🔬 Transfer Learning",[5191,16154,16155,16157],{},[5195,16156,16121],{}," Pre-trained ViT, CLIP, DINO",[5191,16159,16160],{},"Для малих датасетів transfer learning з моделей, натренованих на мільярдах зображень.",[5855,16162],{},[5186,16164,16166],{"id":16165},"практичні-поради-чек-лист-успішного-проєктування","Практичні поради — чек-лист успішного проєктування",[5191,16168,16169],{},"Підсумуємо всі знання у практичний чек-лист, який можна використовувати при проєктуванні власних архітектур або адаптації існуючих.",[5280,16171,16173],{"id":16172},"чек-лист-проєктування-глибокої-мережі","Чек-лист: Проєктування глибокої мережі",[10594,16175,16176,16180,16185,16191,16249,16253,16258,16263,16274,16279,16287,16291,16296,16313,16318,16329,16333,16338,16392,16405,16409,16414,16565,16569,16574],{},[5280,16177,16179],{"id":16178},"крок-1-оберіть-базову-архітектуру","Крок 1. Оберіть базову архітектуру",[5191,16181,16182,16184],{},[5195,16183,16059],{}," Створювати з нуля чи використовувати існуючу?",[5191,16186,16187,16190],{},[5195,16188,16189],{},"Рекомендація:"," 99% випадків — використовуйте існуючу архітектуру (ResNet, EfficientNet тощо) та адаптуйте під свою задачу. Винахід власної архітектури — для наукових досліджень.",[5351,16192,16194],{"className":6329,"code":16193,"language":6331,"meta":5356,"style":5356},"# Завантаження pretrained моделі з torchvision\nimport torchvision.models as models\n\nmodel = models.resnet50(pretrained=True)\n# Замінити останній шар під вашу кількість класів\nmodel.fc = nn.Linear(model.fc.in_features, num_classes=10)\n",[5358,16195,16196,16201,16213,16217,16231,16236],{"__ignoreMap":5356},[5361,16197,16198],{"class":5363,"line":5364},[5361,16199,16200],{"class":6551},"# Завантаження pretrained моделі з torchvision\n",[5361,16202,16203,16205,16208,16210],{"class":5363,"line":5370},[5361,16204,6339],{"class":6338},[5361,16206,16207],{"class":6342}," torchvision.models ",[5361,16209,6353],{"class":6338},[5361,16211,16212],{"class":6342}," models\n",[5361,16214,16215],{"class":5363,"line":5376},[5361,16216,5392],{"emptyLinePlaceholder":5391},[5361,16218,16219,16222,16225,16227,16229],{"class":5363,"line":5382},[5361,16220,16221],{"class":6342},"model = models.resnet50(",[5361,16223,16224],{"class":6454},"pretrained",[5361,16226,6474],{"class":6342},[5361,16228,8805],{"class":6377},[5361,16230,6953],{"class":6342},[5361,16232,16233],{"class":5363,"line":5388},[5361,16234,16235],{"class":6551},"# Замінити останній шар під вашу кількість класів\n",[5361,16237,16238,16241,16243,16245,16247],{"class":5363,"line":5395},[5361,16239,16240],{"class":6342},"model.fc = nn.Linear(model.fc.in_features, ",[5361,16242,8646],{"class":6454},[5361,16244,6474],{"class":6342},[5361,16246,8651],{"class":6477},[5361,16248,6953],{"class":6342},[5280,16250,16252],{"id":16251},"крок-2-адаптуйте-під-розмір-входу","Крок 2. Адаптуйте під розмір входу",[5191,16254,16255,16257],{},[5195,16256,16059],{}," Ваші зображення малі (32×32) чи великі (224×224+)?",[5191,16259,16260],{},[5195,16261,16262],{},"Для малих зображень (CIFAR-10, Fashion-MNIST):",[5747,16264,16265,16268,16271],{},[5750,16266,16267],{},"Зменшіть stride початкового conv: 2 → 1",[5750,16269,16270],{},"Видаліть або зменшіть початковий MaxPool",[5750,16272,16273],{},"Зменшіть кількість каналів (64 → 32) для економії пам'яті",[5191,16275,16276],{},[5195,16277,16278],{},"Для великих зображень (ImageNet, медичні скани):",[5747,16280,16281,16284],{},[5750,16282,16283],{},"Використовуйте стандартні гіперпараметри претренованих моделей",[5750,16285,16286],{},"Розгляньте Multi-Scale Training для кращої генералізації",[5280,16288,16290],{"id":16289},"крок-3-налаштуйте-регуляризацію","Крок 3. Налаштуйте регуляризацію",[5191,16292,16293],{},[5195,16294,16295],{},"Обов'язково:",[5747,16297,16298,16301,16310],{},[5750,16299,16300],{},"✅ Batch Normalization після кожного Conv (крім останнього у residual blocks)",[5750,16302,16303,16304,16306,16307,16309],{},"✅ Weight Decay: ",[5358,16305,10718],{}," для малих датасетів, ",[5358,16308,10729],{}," для великих",[5750,16311,16312],{},"✅ Data Augmentation: RandomCrop, HorizontalFlip як мінімум",[5191,16314,16315],{},[5195,16316,16317],{},"Опційно (залежно від датасету):",[5747,16319,16320,16323,16326],{},[5750,16321,16322],{},"Dropout перед фінальним FC (якщо сильне перенавчання)",[5750,16324,16325],{},"Label Smoothing для класифікації",[5750,16327,16328],{},"MixUp \u002F CutMix для агресивної регуляризації",[5280,16330,16332],{"id":16331},"крок-4-оберіть-оптимізатор-та-learning-rate","Крок 4. Оберіть оптимізатор та learning rate",[5191,16334,16335],{},[5195,16336,16337],{},"Стандартні конфігурації:",[5203,16339,16340,16352],{},[5206,16341,16342],{},[5209,16343,16344,16347,16350],{},[5212,16345,16346],{"align":5214},"Оптимізатор",[5212,16348,16349],{"align":5214},"Learning Rate",[5212,16351,10328],{"align":5214},[5226,16353,16354,16367,16380],{},[5209,16355,16356,16361,16364],{},[5231,16357,16358],{"align":5214},[5195,16359,16360],{},"SGD + Momentum",[5231,16362,16363],{"align":5214},"0.1 → 0.001 (з decay)",[5231,16365,16366],{"align":5214},"ResNet, класичні CNN. Надійний, перевірений часом.",[5209,16368,16369,16374,16377],{},[5231,16370,16371],{"align":5214},[5195,16372,16373],{},"Adam",[5231,16375,16376],{"align":5214},"0.001 → 0.0001",[5231,16378,16379],{"align":5214},"Швидкий старт, менш чутливий до LR. Хороший за замовчуванням.",[5209,16381,16382,16387,16389],{},[5231,16383,16384],{"align":5214},[5195,16385,16386],{},"AdamW",[5231,16388,10708],{"align":5214},[5231,16390,16391],{"align":5214},"Adam з правильним weight decay. Для Transformers та EfficientNet.",[5191,16393,16394,16397,16398,16401,16404],{},[5195,16395,16396],{},"Momentum:"," 0.9 (стандарт)",[16399,16400],"br",{},[5195,16402,16403],{},"Beta для Adam:"," (0.9, 0.999) (за замовчуванням)",[5280,16406,16408],{"id":16407},"крок-5-налаштуйте-learning-rate-scheduler","Крок 5. Налаштуйте learning rate scheduler",[5191,16410,16411],{},[5195,16412,16413],{},"Рекомендовані стратегії:",[10420,16415,16416,16459,16498],{},[5351,16417,16420],{"className":6329,"code":16418,"filename":16419,"language":6331,"meta":5356,"style":5356},"scheduler = torch.optim.lr_scheduler.MultiStepLR(\n    optimizer,\n    milestones=[100, 150],  # Зменшити на епохах 100 та 150\n    gamma=0.1\n)\n","MultiStepLR (простий)",[5358,16421,16422,16426,16430,16447,16455],{"__ignoreMap":5356},[5361,16423,16424],{"class":5363,"line":5364},[5361,16425,10928],{"class":6342},[5361,16427,16428],{"class":5363,"line":5370},[5361,16429,10933],{"class":6342},[5361,16431,16432,16434,16436,16438,16440,16442,16444],{"class":5363,"line":5376},[5361,16433,10938],{"class":6454},[5361,16435,10941],{"class":6342},[5361,16437,8102],{"class":6477},[5361,16439,6458],{"class":6342},[5361,16441,15302],{"class":6477},[5361,16443,10952],{"class":6342},[5361,16445,16446],{"class":6551},"# Зменшити на епохах 100 та 150\n",[5361,16448,16449,16451,16453],{"class":5363,"line":5382},[5361,16450,10887],{"class":6454},[5361,16452,6474],{"class":6342},[5361,16454,10964],{"class":6477},[5361,16456,16457],{"class":5363,"line":5388},[5361,16458,6953],{"class":6342},[5351,16460,16463],{"className":6329,"code":16461,"filename":16462,"language":6331,"meta":5356,"style":5356},"scheduler = torch.optim.lr_scheduler.CosineAnnealingLR(\n    optimizer,\n    T_max=200,      # Загальна кількість епох\n    eta_min=1e-6\n)\n","CosineAnnealingLR (сучасний)",[5358,16464,16465,16469,16473,16485,16494],{"__ignoreMap":5356},[5361,16466,16467],{"class":5363,"line":5364},[5361,16468,10989],{"class":6342},[5361,16470,16471],{"class":5363,"line":5370},[5361,16472,10933],{"class":6342},[5361,16474,16475,16477,16479,16481,16483],{"class":5363,"line":5376},[5361,16476,10998],{"class":6454},[5361,16478,6474],{"class":6342},[5361,16480,11003],{"class":6477},[5361,16482,11006],{"class":6342},[5361,16484,11009],{"class":6551},[5361,16486,16487,16489,16491],{"class":5363,"line":5382},[5361,16488,11014],{"class":6454},[5361,16490,6474],{"class":6342},[5361,16492,16493],{"class":6477},"1e-6\n",[5361,16495,16496],{"class":5363,"line":5388},[5361,16497,6953],{"class":6342},[5351,16499,16502],{"className":6329,"code":16500,"filename":16501,"language":6331,"meta":5356,"style":5356},"scheduler = torch.optim.lr_scheduler.OneCycleLR(\n    optimizer,\n    max_lr=0.1,\n    total_steps=len(train_loader) * num_epochs,\n    pct_start=0.3,  # 30% епох — warmup, 70% — decay\n    anneal_strategy='cos'\n)\n","OneCycleLR (агресивний)",[5358,16503,16504,16509,16513,16525,16537,16551,16561],{"__ignoreMap":5356},[5361,16505,16506],{"class":5363,"line":5364},[5361,16507,16508],{"class":6342},"scheduler = torch.optim.lr_scheduler.OneCycleLR(\n",[5361,16510,16511],{"class":5363,"line":5370},[5361,16512,10933],{"class":6342},[5361,16514,16515,16518,16520,16522],{"class":5363,"line":5376},[5361,16516,16517],{"class":6454},"    max_lr",[5361,16519,6474],{"class":6342},[5361,16521,10892],{"class":6477},[5361,16523,16524],{"class":6342},",\n",[5361,16526,16527,16530,16532,16534],{"class":5363,"line":5382},[5361,16528,16529],{"class":6454},"    total_steps",[5361,16531,6474],{"class":6342},[5361,16533,11672],{"class":6448},[5361,16535,16536],{"class":6342},"(train_loader) * num_epochs,\n",[5361,16538,16539,16542,16544,16546,16548],{"class":5363,"line":5388},[5361,16540,16541],{"class":6454},"    pct_start",[5361,16543,6474],{"class":6342},[5361,16545,14689],{"class":6477},[5361,16547,10879],{"class":6342},[5361,16549,16550],{"class":6551},"# 30% епох — warmup, 70% — decay\n",[5361,16552,16553,16556,16558],{"class":5363,"line":5395},[5361,16554,16555],{"class":6454},"    anneal_strategy",[5361,16557,6474],{"class":6342},[5361,16559,16560],{"class":6401},"'cos'\n",[5361,16562,16563],{"class":5363,"line":5401},[5361,16564,6953],{"class":6342},[5280,16566,16568],{"id":16567},"крок-6-моніторинг-та-debugging","Крок 6. Моніторинг та debugging",[5191,16570,16571],{},[5195,16572,16573],{},"Під час навчання стежте за:",[5288,16575,16576],{},[5291,16577,16580,16583,16584,16586,16589,16590,16592,16595],{"icon":16578,"title":16579},"i-lucide-trending-up","📈 Train vs Val Loss Gap",[5195,16581,16582],{},"Норма:"," gap \u003C 15%",[16399,16585],{},[5195,16587,16588],{},"Якщо gap > 20%:"," перенавчання → більше регуляризації",[16399,16591],{},[5195,16593,16594],{},"Якщо обидва високі:"," underfitting → збільшити model capacity",[5291,16597,16599,16602,16666],{"icon":92,"title":16598},"⚡ Gradient Flow",[5191,16600,16601],{},"Перевірте, чи градієнти не зникають\u002Fвибухають:",[5351,16603,16605],{"className":6329,"code":16604,"language":6331,"meta":5356,"style":5356},"for name, param in model.named_parameters():\n    if param.grad is not None:\n        print(f\"{name}: grad norm = {param.grad.norm():.4f}\")\n",[5358,16606,16607,16619,16635],{"__ignoreMap":5356},[5361,16608,16609,16611,16614,16616],{"class":5363,"line":5364},[5361,16610,7932],{"class":6338},[5361,16612,16613],{"class":6342}," name, param ",[5361,16615,7938],{"class":6338},[5361,16617,16618],{"class":6342}," model.named_parameters():\n",[5361,16620,16621,16624,16627,16629,16631,16633],{"class":5363,"line":5370},[5361,16622,16623],{"class":6338},"    if",[5361,16625,16626],{"class":6342}," param.grad ",[5361,16628,6849],{"class":6377},[5361,16630,6852],{"class":6377},[5361,16632,6855],{"class":6377},[5361,16634,6858],{"class":6342},[5361,16636,16637,16640,16642,16644,16646,16648,16650,16652,16655,16657,16660,16662,16664],{"class":5363,"line":5376},[5361,16638,16639],{"class":6448},"        print",[5361,16641,6385],{"class":6342},[5361,16643,7039],{"class":6377},[5361,16645,6950],{"class":6401},[5361,16647,7045],{"class":6377},[5361,16649,13208],{"class":6342},[5361,16651,7051],{"class":6377},[5361,16653,16654],{"class":6401},": grad norm = ",[5361,16656,7045],{"class":6377},[5361,16658,16659],{"class":6342},"param.grad.norm()",[5361,16661,14123],{"class":6377},[5361,16663,6950],{"class":6401},[5361,16665,6953],{"class":6342},[5191,16667,16668,16670,16671,16673,16676,16677,16679,16682],{},[5195,16669,16582],{}," 0.001 - 10.0",[16399,16672],{},[5195,16674,16675],{},"Якщо \u003C 0.0001:"," vanishing gradients",[16399,16678],{},[5195,16680,16681],{},"Якщо > 100:"," exploding gradients → зменшити LR",[5291,16684,16686,16691],{"icon":5775,"title":16685},"🎯 Accuracy на першій епосі",[5191,16687,16688],{},[5195,16689,16690],{},"Санітарна перевірка:",[5747,16692,16693,16696,16699],{},[5750,16694,16695],{},"Random baseline: 10% для 10 класів",[5750,16697,16698],{},"Після 1 епохи має бути > 20-30%",[5750,16700,16701],{},"Якщо застряг на 10%: помилка у коді або даних!",[5191,16703,6259],{},[5280,16705,16707],{"id":16706},"крок-7-збереження-найкращої-моделі","Крок 7. Збереження найкращої моделі",[5351,16709,16711],{"className":6329,"code":16710,"language":6331,"meta":5356,"style":5356},"best_acc = 0\nbest_model_path = 'best_model.pth'\n\nfor epoch in range(num_epochs):\n    train_loss, train_acc = train_epoch(...)\n    val_loss, val_acc = validate(...)\n    \n    # Зберігати лише найкращу\n    if val_acc > best_acc:\n        best_acc = val_acc\n        torch.save({\n            'epoch': epoch,\n            'model_state_dict': model.state_dict(),\n            'optimizer_state_dict': optimizer.state_dict(),\n            'val_acc': val_acc,\n            'val_loss': val_loss\n        }, best_model_path)\n        print(f\"💾 Збережено нову найкращу модель: {val_acc:.2f}%\")\n",[5358,16712,16713,16720,16728,16732,16745,16750,16755,16759,16764,16771,16776,16781,16789,16797,16805,16813,16821,16826],{"__ignoreMap":5356},[5361,16714,16715,16718],{"class":5363,"line":5364},[5361,16716,16717],{"class":6342},"best_acc = ",[5361,16719,13424],{"class":6477},[5361,16721,16722,16725],{"class":5363,"line":5370},[5361,16723,16724],{"class":6342},"best_model_path = ",[5361,16726,16727],{"class":6401},"'best_model.pth'\n",[5361,16729,16730],{"class":5363,"line":5376},[5361,16731,5392],{"emptyLinePlaceholder":5391},[5361,16733,16734,16736,16738,16740,16742],{"class":5363,"line":5382},[5361,16735,7932],{"class":6338},[5361,16737,13917],{"class":6342},[5361,16739,7938],{"class":6338},[5361,16741,9226],{"class":6448},[5361,16743,16744],{"class":6342},"(num_epochs):\n",[5361,16746,16747],{"class":5363,"line":5388},[5361,16748,16749],{"class":6342},"    train_loss, train_acc = train_epoch(...)\n",[5361,16751,16752],{"class":5363,"line":5395},[5361,16753,16754],{"class":6342},"    val_loss, val_acc = validate(...)\n",[5361,16756,16757],{"class":5363,"line":5401},[5361,16758,6412],{"class":6342},[5361,16760,16761],{"class":5363,"line":5407},[5361,16762,16763],{"class":6551},"    # Зберігати лише найкращу\n",[5361,16765,16766,16768],{"class":5363,"line":5413},[5361,16767,16623],{"class":6338},[5361,16769,16770],{"class":6342}," val_acc > best_acc:\n",[5361,16772,16773],{"class":5363,"line":5418},[5361,16774,16775],{"class":6342},"        best_acc = val_acc\n",[5361,16777,16778],{"class":5363,"line":5424},[5361,16779,16780],{"class":6342},"        torch.save({\n",[5361,16782,16783,16786],{"class":5363,"line":5430},[5361,16784,16785],{"class":6401},"            'epoch'",[5361,16787,16788],{"class":6342},": epoch,\n",[5361,16790,16791,16794],{"class":5363,"line":5436},[5361,16792,16793],{"class":6401},"            'model_state_dict'",[5361,16795,16796],{"class":6342},": model.state_dict(),\n",[5361,16798,16799,16802],{"class":5363,"line":5442},[5361,16800,16801],{"class":6401},"            'optimizer_state_dict'",[5361,16803,16804],{"class":6342},": optimizer.state_dict(),\n",[5361,16806,16807,16810],{"class":5363,"line":5448},[5361,16808,16809],{"class":6401},"            'val_acc'",[5361,16811,16812],{"class":6342},": val_acc,\n",[5361,16814,16815,16818],{"class":5363,"line":5454},[5361,16816,16817],{"class":6401},"            'val_loss'",[5361,16819,16820],{"class":6342},": val_loss\n",[5361,16822,16823],{"class":5363,"line":5460},[5361,16824,16825],{"class":6342},"        }, best_model_path)\n",[5361,16827,16828,16830,16832,16834,16837,16839,16842,16844,16846],{"class":5363,"line":5466},[5361,16829,16639],{"class":6448},[5361,16831,6385],{"class":6342},[5361,16833,7039],{"class":6377},[5361,16835,16836],{"class":6401},"\"💾 Збережено нову найкращу модель: ",[5361,16838,7045],{"class":6377},[5361,16840,16841],{"class":6342},"val_acc",[5361,16843,9932],{"class":6377},[5361,16845,8108],{"class":6401},[5361,16847,6953],{"class":6342},[5191,16849,6259],{},[6282,16851,16852,16857],{},[5191,16853,16854],{},[5195,16855,16856],{},"Швидкий прототип → глибока оптимізація:",[10145,16858,16859,16865,16871],{},[5750,16860,16861,16864],{},[5195,16862,16863],{},"Спочатку"," переконайтеся, що модель взагалі навчається (5-10 епох, малий датасет)",[5750,16866,16867,16870],{},[5195,16868,16869],{},"Потім"," налаштовуйте гіперпараметри (grid search, Optuna)",[5750,16872,16873,16876],{},[5195,16874,16875],{},"Нарешті"," тренуйте фінальну версію на повному датасеті з найкращими параметрами",[5855,16878],{},[5186,16880,16882],{"id":16881},"практичні-завдання-та-запитання-для-самоконтролю","Практичні завдання та запитання для самоконтролю",[10842,16884,16885,16928,16965,17076,17133],{},[10845,16886,16889,16894,16897,16915,16920],{"icon":16887,"label":16888},"i-lucide-clipboard-list","📋 Завдання 1: Адаптувати ResNet-18 для Fashion-MNIST",[5191,16890,16891],{},[5195,16892,16893],{},"Завдання:",[5191,16895,16896],{},"Fashion-MNIST містить grayscale зображення 28×28. Модифікуйте ResNet-18 для цього датасету:",[10145,16898,16899,16906,16909,16912],{},[5750,16900,16901,16902,16905],{},"Змініть ",[5358,16903,16904],{},"input_channels=1"," (grayscale)",[5750,16907,16908],{},"Адаптуйте початковий conv (kernel_size=3, stride=1, без maxpool)",[5750,16910,16911],{},"Натренуйте 30 епох з batch_size=128",[5750,16913,16914],{},"Досягніть accuracy > 93%",[5191,16916,16917],{},[5195,16918,16919],{},"Очікувані результати:",[5747,16921,16922,16925],{},[5750,16923,16924],{},"Plain CNN: ~90-91%",[5750,16926,16927],{},"ResNet-18 (модифікований): ~93-94%",[10845,16929,16931,16935,16938,16946,16949,16959],{"icon":16887,"label":16930},"📋 Завдання 2: Експеримент з Bottleneck Block",[5191,16932,16933],{},[5195,16934,16893],{},[5191,16936,16937],{},"Реалізуйте ResNet-50 з Bottleneck блоками на CIFAR-10. Порівняйте:",[5747,16939,16940,16943],{},[5750,16941,16942],{},"ResNet-34 (BasicBlock): ~92% accuracy, X параметрів",[5750,16944,16945],{},"ResNet-50 (BottleneckBlock): ~92-93% accuracy, менше параметрів",[5191,16947,16948],{},"Виміряйте:",[5747,16950,16951,16953,16956],{},[5750,16952,13293],{},[5750,16954,16955],{},"Час навчання однієї епохи",[5750,16957,16958],{},"Фінальну test accuracy",[5191,16960,16961,16964],{},[5195,16962,16963],{},"Очікування:"," Bottleneck має менше параметрів, але може бути трохи повільнішим через більше шарів.",[10845,16966,16969,16977,17047,17051,17064,17073],{"icon":16967,"label":16968},"i-lucide-help-circle","❓ Чому у Residual Block фінальний ReLU після додавання?",[5191,16970,16971,16973,16974,16976],{},[5195,16972,16059],{}," Чому ReLU застосовується ",[5195,16975,8198],{}," skip connection, а не перед?",[5351,16978,16980],{"className":6329,"code":16979,"language":6331,"meta":5356,"style":5356},"# Правильно\nout = self.conv2(out)\nout = self.bn2(out)\nout += identity       # Спочатку додаємо\nout = F.relu(out)     # Потім ReLU\n\n# Чому не так?\nout = F.relu(self.bn2(self.conv2(out)))\nout += identity\n",[5358,16981,16982,16987,16995,17003,17011,17019,17023,17028,17042],{"__ignoreMap":5356},[5361,16983,16984],{"class":5363,"line":5364},[5361,16985,16986],{"class":6551},"# Правильно\n",[5361,16988,16989,16991,16993],{"class":5363,"line":5370},[5361,16990,8209],{"class":6342},[5361,16992,6455],{"class":6377},[5361,16994,6813],{"class":6342},[5361,16996,16997,16999,17001],{"class":5363,"line":5376},[5361,16998,8209],{"class":6342},[5361,17000,6455],{"class":6377},[5361,17002,6823],{"class":6342},[5361,17004,17005,17008],{"class":5363,"line":5382},[5361,17006,17007],{"class":6342},"out += identity       ",[5361,17009,17010],{"class":6551},"# Спочатку додаємо\n",[5361,17012,17013,17016],{"class":5363,"line":5388},[5361,17014,17015],{"class":6342},"out = F.relu(out)     ",[5361,17017,17018],{"class":6551},"# Потім ReLU\n",[5361,17020,17021],{"class":5363,"line":5395},[5361,17022,5392],{"emptyLinePlaceholder":5391},[5361,17024,17025],{"class":5363,"line":5401},[5361,17026,17027],{"class":6551},"# Чому не так?\n",[5361,17029,17030,17033,17035,17037,17039],{"class":5363,"line":5407},[5361,17031,17032],{"class":6342},"out = F.relu(",[5361,17034,6455],{"class":6377},[5361,17036,7776],{"class":6342},[5361,17038,6455],{"class":6377},[5361,17040,17041],{"class":6342},".conv2(out)))\n",[5361,17043,17044],{"class":5363,"line":5413},[5361,17045,17046],{"class":6342},"out += identity\n",[5191,17048,17049],{},[5195,17050,16065],{},[5191,17052,17053,17054,17056,17057,17060,17061,17063],{},"Якщо застосувати ReLU до ",[5693,17055],{"tex":5942,":inline":5696}," перед додаванням, це ",[5195,17058,17059],{},"обнулить всі від'ємні значення",". Але залишок ",[5693,17062],{"tex":5942,":inline":5696}," може бути від'ємним (коли модель хоче \"відкоригувати\" вхід вниз).",[5191,17065,17066,17067,17069,17070,5198],{},"ReLU після додавання дозволяє ",[5693,17068],{"tex":5942,":inline":5696}," бути будь-яким (позитивним чи негативним), а фінальна активація застосовується до суми ",[5693,17071],{"tex":17072,":inline":5696},"F(\\mathbf{x}) + \\mathbf{x}",[5191,17074,17075],{},"Це критично для ефективного gradient flow та можливості навчання identity mapping.",[10845,17077,17079,17084,17088,17095,17125,17128],{"icon":16967,"label":17078},"❓ Чи можна комбінувати ResNet та DenseNet?",[5191,17080,17081,17083],{},[5195,17082,16059],{}," Чи можливо створити гібридну архітектуру з ResNet (skip connections через додавання) та DenseNet (dense connections через конкатенацію)?",[5191,17085,17086],{},[5195,17087,16065],{},[5191,17089,17090,17091,17094],{},"Так! Існує архітектура ",[5195,17092,17093],{},"DPN (Dual Path Networks)",", яка робить саме це:",[5351,17096,17098],{"className":6329,"code":17097,"language":6331,"meta":5356,"style":5356},"out = torch.cat([residual, dense], dim=1)  # Конкатенація (DenseNet)\nout += shortcut                             # Додавання (ResNet)\n",[5358,17099,17100,17117],{"__ignoreMap":5356},[5361,17101,17102,17105,17108,17110,17112,17114],{"class":5363,"line":5364},[5361,17103,17104],{"class":6342},"out = torch.cat([residual, dense], ",[5361,17106,17107],{"class":6454},"dim",[5361,17109,6474],{"class":6342},[5361,17111,6326],{"class":6477},[5361,17113,7020],{"class":6342},[5361,17115,17116],{"class":6551},"# Конкатенація (DenseNet)\n",[5361,17118,17119,17122],{"class":5363,"line":5370},[5361,17120,17121],{"class":6342},"out += shortcut                             ",[5361,17123,17124],{"class":6551},"# Додавання (ResNet)\n",[5191,17126,17127],{},"DPN досягає кращих результатів на ImageNet, ніж окремо ResNet чи DenseNet, але ціною більшої пам'яті та складності реалізації.",[5191,17129,17130,17132],{},[5195,17131,15902],{}," Змагання з комп'ютерного зору, де критична максимальна точність незалежно від ресурсів.",[10845,17134,17136,17141,17145,17148,17204,17210],{"icon":16967,"label":17135},"❓ Скільки шарів «занадто багато»?",[5191,17137,17138,17140],{},[5195,17139,16059],{}," Яка оптимальна глибина мережі? Чи ResNet-1000 краще за ResNet-152?",[5191,17142,17143],{},[5195,17144,16065],{},[5191,17146,17147],{},"Емпірично виявлено:",[5203,17149,17150,17163],{},[5206,17151,17152],{},[5209,17153,17154,17157,17160],{},[5212,17155,17156],{"align":5214},"Датасет",[5212,17158,17159],{"align":5214},"Оптимальна глибина",[5212,17161,17162],{"align":5214},"Примітки",[5226,17164,17165,17178,17191],{},[5209,17166,17167,17172,17175],{},[5231,17168,17169],{"align":5214},[5195,17170,17171],{},"CIFAR-10\u002F100",[5231,17173,17174],{"align":5214},"56-110 шарів",[5231,17176,17177],{"align":5214},"Далі точність зростає мінімально",[5209,17179,17180,17185,17188],{},[5231,17181,17182],{"align":5214},[5195,17183,17184],{},"ImageNet",[5231,17186,17187],{"align":5214},"50-152 шари",[5231,17189,17190],{"align":5214},"ResNet-152 — sweet spot. ResNet-200 трохи краще, але дорожче",[5209,17192,17193,17198,17201],{},[5231,17194,17195],{"align":5214},[5195,17196,17197],{},"Specialized (медичні дані)",[5231,17199,17200],{"align":5214},"18-50 шарів",[5231,17202,17203],{"align":5214},"Малі датасети не потребують екстремальної глибини",[5191,17205,17206,17209],{},[5195,17207,17208],{},"Загальне правило:"," ResNet-50 — універсальний вибір. Глибше (101, 152) — лише якщо маєте величезний датасет та обчислювальні ресурси.",[5191,17211,17212,17213,17216,17217,17220],{},"Після певної точки краще ",[5195,17214,17215],{},"масштабувати ширину"," (більше каналів) або ",[5195,17218,17219],{},"роздільність входу",", ніж глибину. Саме це робить EfficientNet (compound scaling).",[5855,17222],{},[5186,17224,17226],{"id":17225},"підсумки-що-ми-опанували","Підсумки — що ми опанували",[5191,17228,17229],{},"Ця стаття завершує фундаментальний блок про проєктування та навчання глибоких нейронних мереж. Підсумуймо ключові досягнення:",[5288,17231,17232,17249,17272,17292],{},[5291,17233,17236,17243],{"icon":17234,"title":17235},"i-lucide-alert-triangle","🎯 Розуміння degradation problem",[5191,17237,17238,17239,17242],{},"Ми емпірично підтвердили, що просте збільшення глибини ",[5195,17240,17241],{},"погіршує"," результати через складності оптимізації, vanishing gradients та неможливість навчання identity mapping.",[5191,17244,17245,17248],{},[5195,17246,17247],{},"Кількісно:"," Plain-56 (83.05%) \u003C Plain-20 (86.50%)",[5291,17250,17253,17256,17267],{"icon":17251,"title":17252},"i-lucide-link","🔗 Skip Connections — революція архітектури",[5191,17254,17255],{},"Residual learning через skip connections вирішує degradation:",[5747,17257,17258,17261,17264],{},[5750,17259,17260],{},"Прямий шлях для градієнтів (вирішує vanishing)",[5750,17262,17263],{},"Легкість навчання identity (модель може \"вимкнути\" непотрібні шари)",[5750,17265,17266],{},"Можливість навчати 100+ шарові мережі",[5191,17268,17269,17271],{},[5195,17270,17247],{}," ResNet-56 (92.15%) — на +9.1% краще за Plain-56",[5291,17273,17276,17279],{"icon":17274,"title":17275},"i-lucide-code","🏗️ Практичні навички реалізації",[5191,17277,17278],{},"Ми реалізували повноцінну ResNet архітектуру з нуля:",[5747,17280,17281,17284,17287,17289],{},[5750,17282,17283],{},"BasicBlock та BottleneckBlock",[5750,17285,17286],{},"Downsample projection для узгодження розмірів",[5750,17288,10140],{},[5750,17290,17291],{},"Best practices проєктування",[5291,17293,17296,17299],{"icon":17294,"title":17295},"i-lucide-bar-chart","📊 Експериментальна методологія",[5191,17297,17298],{},"Навчилися проводити порівняльні експерименти:",[5747,17300,17301,17304,17307,17310],{},[5750,17302,17303],{},"Контрольована зміна однієї змінної (наявність skip connections)",[5750,17305,17306],{},"Моніторинг train\u002Ftest metrics",[5750,17308,17309],{},"Візуалізація та інтерпретація результатів",[5750,17311,17312],{},"Формулювання кількісних висновків",[5280,17314,17316],{"id":17315},"ключові-висновки","Ключові висновки",[10594,17318,17319,17323,17332,17343,17347,17350,17354,17357,17362,17366],{},[5280,17320,17322],{"id":17321},"_1-skip-connections-не-просто-хак-а-фундаментальний-принцип","1. Skip Connections — не просто \"хак\", а фундаментальний принцип",[5191,17324,17325,17326,17328,17329,17331],{},"Residual learning змінив парадигму: замість навчання ",[5693,17327],{"tex":5869,":inline":5696},", навчаємо ",[5693,17330],{"tex":5876,":inline":5696},". Це дозволяє:",[5747,17333,17334,17337,17340],{},[5750,17335,17336],{},"Ефективний gradient flow до початкових шарів",[5750,17338,17339],{},"Можливість \"вимкнення\" непотрібних трансформацій",[5750,17341,17342],{},"Масштабування до сотень шарів",[5280,17344,17346],{"id":17345},"_2-batch-normalization-skip-connections-синергія","2. Batch Normalization + Skip Connections = синергія",[5191,17348,17349],{},"Комбінація BN (стабілізація активацій) та skip connections (gradient flow) створює надійну основу для глибоких архітектур. Без будь-якого з цих компонентів результати значно гірші.",[5280,17351,17353],{"id":17352},"_3-архітектурні-інновації-важливіші-за-більше-параметрів","3. Архітектурні інновації важливіші за \"більше параметрів\"",[5191,17355,17356],{},"Plain-56 має 855k параметрів, але програє Plain-20 з 272k. ResNet-56 з тією ж кількістю параметрів, що й Plain-56, але з skip connections, виграє на +9%.",[5191,17358,17359,17361],{},[5195,17360,16106],{}," \"Як\" організовані параметри важливіше за \"скільки\" їх є.",[5280,17363,17365],{"id":17364},"_4-resnet-фундамент-сучасного-computer-vision","4. ResNet — фундамент сучасного Computer Vision",[5191,17367,17368],{},"Практично всі сучасні архітектури (EfficientNet, ConvNeXt, Swin Transformer, DETR) використовують skip connections у тій чи іншій формі. ResNet залишається релевантним через 9 років після публікації (2015-2024).",[5736,17370,17371,17376],{},[5191,17372,17373],{},[5195,17374,17375],{},"Наступні кроки:",[5191,17377,17378,17379,17382],{},"Цим матеріалом ми завершили теоретичний блок про проєктування глибоких мереж. У наступній статті розглянемо ",[5195,17380,17381],{},"Learning Rate Scheduling"," — техніки динамічної адаптації швидкості навчання для досягнення кращої конвергенції. А у фінальному проєкті модуля об'єднаємо всі техніки: Mini-batch GD, Weight Initialization, Batch Normalization, Dropout та ResNet архітектуру для побудови production-ready класифікатора.",[5855,17384],{},[5186,17386,17388],{"id":17387},"додаткові-ресурси","Додаткові ресурси",[5288,17390,17391,17418,17432,17493],{},[5291,17392,17395,17406,17415],{"icon":17393,"title":17394},"i-lucide-file-text","📄 Оригінальна стаття ResNet",[5191,17396,17397,17400,17402,17403,17405],{},[5195,17398,17399],{},"Deep Residual Learning for Image Recognition",[16399,17401],{},"\nKaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun (Microsoft Research)",[16399,17404],{},"\nCVPR 2016 (Best Paper Award)",[5191,17407,17408],{},[17409,17410,17414],"a",{"href":17411,"rel":17412},"https:\u002F\u002Farxiv.org\u002Fabs\u002F1512.03385",[17413],"nofollow","Читати на arXiv",[5191,17416,17417],{},"Одна з найцитованіших статей у Computer Vision (150,000+ цитувань). Обов'язкова до прочитання.",[5291,17419,17421,17429],{"icon":3301,"title":17420},"🎥 Лекція Andrej Karpathy про CNN архітектури",[5191,17422,17423,17426,17428],{},[5195,17424,17425],{},"CS231n: Convolutional Neural Networks for Visual Recognition",[16399,17427],{},"\nStanford University",[5191,17430,17431],{},"Чудові пояснення еволюції від AlexNet до ResNet з візуалізаціями та інтуїцією.",[5291,17433,17436,17486],{"icon":17434,"title":17435},"i-lucide-github","💻 Офіційна реалізація PyTorch",[5351,17437,17439],{"className":6329,"code":17438,"language":6331,"meta":5356,"style":5356},"import torchvision.models as models\n\n# Готові претреновані моделі\nresnet18 = models.resnet18(pretrained=True)\nresnet50 = models.resnet50(pretrained=True)\n",[5358,17440,17441,17451,17455,17460,17473],{"__ignoreMap":5356},[5361,17442,17443,17445,17447,17449],{"class":5363,"line":5364},[5361,17444,6339],{"class":6338},[5361,17446,16207],{"class":6342},[5361,17448,6353],{"class":6338},[5361,17450,16212],{"class":6342},[5361,17452,17453],{"class":5363,"line":5370},[5361,17454,5392],{"emptyLinePlaceholder":5391},[5361,17456,17457],{"class":5363,"line":5376},[5361,17458,17459],{"class":6551},"# Готові претреновані моделі\n",[5361,17461,17462,17465,17467,17469,17471],{"class":5363,"line":5382},[5361,17463,17464],{"class":6342},"resnet18 = models.resnet18(",[5361,17466,16224],{"class":6454},[5361,17468,6474],{"class":6342},[5361,17470,8805],{"class":6377},[5361,17472,6953],{"class":6342},[5361,17474,17475,17478,17480,17482,17484],{"class":5363,"line":5388},[5361,17476,17477],{"class":6342},"resnet50 = models.resnet50(",[5361,17479,16224],{"class":6454},[5361,17481,6474],{"class":6342},[5361,17483,8805],{"class":6377},[5361,17485,6953],{"class":6342},[5191,17487,17488],{},[17409,17489,17492],{"href":17490,"rel":17491},"https:\u002F\u002Fpytorch.org\u002Fvision\u002Fstable\u002Fmodels.html",[17413],"Документація torchvision.models",[5291,17494,17497,17500],{"icon":17495,"title":17496},"i-lucide-flask","🔬 Papers With Code — ResNet",[5191,17498,17499],{},"Порівняння всіх варіантів ResNet (ResNet-v2, ResNeXt, Wide ResNet тощо) на різних датасетах з кодом та результатами.",[5191,17501,17502],{},[17409,17503,17506],{"href":17504,"rel":17505},"https:\u002F\u002Fpaperswithcode.com\u002Fmethod\u002Fresnet",[17413],"Переглянути бенчмарки",[17508,17509,17510],"style",{},"html .light .shiki span {color: var(--shiki-light);background: var(--shiki-light-bg);font-style: var(--shiki-light-font-style);font-weight: var(--shiki-light-font-weight);text-decoration: var(--shiki-light-text-decoration);}html.light .shiki span {color: var(--shiki-light);background: var(--shiki-light-bg);font-style: var(--shiki-light-font-style);font-weight: var(--shiki-light-font-weight);text-decoration: var(--shiki-light-text-decoration);}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html.dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html pre.shiki code .s8xlr, html code.shiki .s8xlr{--shiki-light:#AF00DB;--shiki-default:#C586C0;--shiki-dark:#C586C0}html pre.shiki code .sHH4Y, html code.shiki .sHH4Y{--shiki-light:#000000;--shiki-default:#D4D4D4;--shiki-dark:#D4D4D4}html pre.shiki code .su1O8, html code.shiki .su1O8{--shiki-light:#0000FF;--shiki-default:#569CD6;--shiki-dark:#569CD6}html pre.shiki code .sN1BT, html code.shiki .sN1BT{--shiki-light:#267F99;--shiki-default:#4EC9B0;--shiki-dark:#4EC9B0}html pre.shiki code .sbdoH, html code.shiki .sbdoH{--shiki-light:#A31515;--shiki-default:#CE9178;--shiki-dark:#CE9178}html pre.shiki code .s8Opu, html code.shiki .s8Opu{--shiki-light:#795E26;--shiki-default:#DCDCAA;--shiki-dark:#DCDCAA}html pre.shiki code .siwwj, html code.shiki .siwwj{--shiki-light:#001080;--shiki-default:#9CDCFE;--shiki-dark:#9CDCFE}html pre.shiki code .sJj4R, html code.shiki .sJj4R{--shiki-light:#098658;--shiki-default:#B5CEA8;--shiki-dark:#B5CEA8}html pre.shiki code .spJ8K, html code.shiki .spJ8K{--shiki-light:#008000;--shiki-default:#6A9955;--shiki-dark:#6A9955}html pre.shiki code .sjcCO, html code.shiki .sjcCO{--shiki-light:#EE0000;--shiki-default:#D7BA7D;--shiki-dark:#D7BA7D}",{"title":5356,"searchDepth":5370,"depth":5370,"links":17512},[17513,17517,17521,17526,17530,17542,17548,17552,17562,17563,17570],{"id":5188,"depth":5370,"text":5189,"children":17514},[17515,17516],{"id":5282,"depth":5376,"text":5283},{"id":5668,"depth":5376,"text":5669},{"id":5859,"depth":5370,"text":5860,"children":17518},[17519,17520],{"id":5879,"depth":5376,"text":5880},{"id":6191,"depth":5376,"text":6192},{"id":6296,"depth":5370,"text":6297,"children":17522},[17523,17524,17525],{"id":6303,"depth":5376,"text":6304},{"id":7380,"depth":5376,"text":7381},{"id":7393,"depth":5376,"text":7394},{"id":8252,"depth":5370,"text":8253,"children":17527},[17528,17529],{"id":8259,"depth":5376,"text":8260},{"id":8550,"depth":5376,"text":8551},{"id":10178,"depth":5370,"text":10179,"children":17531},[17532,17533,17534,17535,17536,17537,17538,17539,17540,17541],{"id":10185,"depth":5376,"text":10186},{"id":10225,"depth":5376,"text":10226},{"id":10304,"depth":5376,"text":10305},{"id":10412,"depth":5376,"text":10413},{"id":10583,"depth":5376,"text":10584},{"id":10598,"depth":5376,"text":10599},{"id":10605,"depth":5376,"text":10606},{"id":10689,"depth":5376,"text":10690},{"id":10732,"depth":5376,"text":10733},{"id":10831,"depth":5376,"text":10832},{"id":11148,"depth":5370,"text":11149,"children":17543},[17544,17545,17546,17547],{"id":11175,"depth":5376,"text":11176},{"id":11783,"depth":5376,"text":11784},{"id":13357,"depth":5376,"text":13358},{"id":14453,"depth":5376,"text":14454},{"id":15837,"depth":5370,"text":15838,"children":17549},[17550,17551],{"id":15844,"depth":5376,"text":15845},{"id":16110,"depth":5376,"text":16111},{"id":16165,"depth":5370,"text":16166,"children":17553},[17554,17555,17556,17557,17558,17559,17560,17561],{"id":16172,"depth":5376,"text":16173},{"id":16178,"depth":5376,"text":16179},{"id":16251,"depth":5376,"text":16252},{"id":16289,"depth":5376,"text":16290},{"id":16331,"depth":5376,"text":16332},{"id":16407,"depth":5376,"text":16408},{"id":16567,"depth":5376,"text":16568},{"id":16706,"depth":5376,"text":16707},{"id":16881,"depth":5370,"text":16882},{"id":17225,"depth":5370,"text":17226,"children":17564},[17565,17566,17567,17568,17569],{"id":17315,"depth":5376,"text":17316},{"id":17321,"depth":5376,"text":17322},{"id":17345,"depth":5376,"text":17346},{"id":17352,"depth":5376,"text":17353},{"id":17364,"depth":5376,"text":17365},{"id":17387,"depth":5370,"text":17388},"Проблема degradation у дуже глибоких мережах, ідея Skip Connections, реалізація Residual Blocks, побудова повноцінної ResNet-подібної архітектури, best practices та порівняльний експеримент.","md",null,{},{"title":3972,"description":17571},"0yWdLRKvnfOwWDWErERRUDv21KHa36jjz2Ac3TKvD0U",[17578,17580],{"title":3968,"path":3969,"stem":3970,"description":17579,"children":-1},"Випадкове вимикання нейронів для боротьби з перенавчанням, математика та інтуїція Dropout, комбінування з BatchNorm і weight decay, експериментальне порівняння стратегій.",{"title":3976,"path":3977,"stem":3978,"description":17581,"children":-1},"Проблема фіксованого learning rate, типи schedulers (Step, Exponential, Cosine, ReduceLROnPlateau), warm-up та warm restarts, практична реалізація та експериментальне порівняння.",1791986837067]