[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-prabhuomkar--pytorch-cpp":3,"tool-prabhuomkar--pytorch-cpp":64},[4,17,27,35,43,56],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":16},3808,"stable-diffusion-webui","AUTOMATIC1111\u002Fstable-diffusion-webui","stable-diffusion-webui 是一个基于 Gradio 构建的网页版操作界面，旨在让用户能够轻松地在本地运行和使用强大的 Stable Diffusion 图像生成模型。它解决了原始模型依赖命令行、操作门槛高且功能分散的痛点，将复杂的 AI 绘图流程整合进一个直观易用的图形化平台。\n\n无论是希望快速上手的普通创作者、需要精细控制画面细节的设计师，还是想要深入探索模型潜力的开发者与研究人员，都能从中获益。其核心亮点在于极高的功能丰富度：不仅支持文生图、图生图、局部重绘（Inpainting）和外绘（Outpainting）等基础模式，还独创了注意力机制调整、提示词矩阵、负向提示词以及“高清修复”等高级功能。此外，它内置了 GFPGAN 和 CodeFormer 等人脸修复工具，支持多种神经网络放大算法，并允许用户通过插件系统无限扩展能力。即使是显存有限的设备，stable-diffusion-webui 也提供了相应的优化选项，让高质量的 AI 艺术创作变得触手可及。",162132,3,"2026-04-05T11:01:52",[13,14,15],"开发框架","图像","Agent","ready",{"id":18,"name":19,"github_repo":20,"description_zh":21,"stars":22,"difficulty_score":23,"last_commit_at":24,"category_tags":25,"status":16},1381,"everything-claude-code","affaan-m\u002Feverything-claude-code","everything-claude-code 是一套专为 AI 编程助手（如 Claude Code、Codex、Cursor 等）打造的高性能优化系统。它不仅仅是一组配置文件，而是一个经过长期实战打磨的完整框架，旨在解决 AI 代理在实际开发中面临的效率低下、记忆丢失、安全隐患及缺乏持续学习能力等核心痛点。\n\n通过引入技能模块化、直觉增强、记忆持久化机制以及内置的安全扫描功能，everything-claude-code 能显著提升 AI 在复杂任务中的表现，帮助开发者构建更稳定、更智能的生产级 AI 代理。其独特的“研究优先”开发理念和针对 Token 消耗的优化策略，使得模型响应更快、成本更低，同时有效防御潜在的攻击向量。\n\n这套工具特别适合软件开发者、AI 研究人员以及希望深度定制 AI 工作流的技术团队使用。无论您是在构建大型代码库，还是需要 AI 协助进行安全审计与自动化测试，everything-claude-code 都能提供强大的底层支持。作为一个曾荣获 Anthropic 黑客大奖的开源项目，它融合了多语言支持与丰富的实战钩子（hooks），让 AI 真正成长为懂上",138956,2,"2026-04-05T11:33:21",[13,15,26],"语言模型",{"id":28,"name":29,"github_repo":30,"description_zh":31,"stars":32,"difficulty_score":23,"last_commit_at":33,"category_tags":34,"status":16},2271,"ComfyUI","Comfy-Org\u002FComfyUI","ComfyUI 是一款功能强大且高度模块化的视觉 AI 引擎，专为设计和执行复杂的 Stable Diffusion 图像生成流程而打造。它摒弃了传统的代码编写模式，采用直观的节点式流程图界面，让用户通过连接不同的功能模块即可构建个性化的生成管线。\n\n这一设计巧妙解决了高级 AI 绘图工作流配置复杂、灵活性不足的痛点。用户无需具备编程背景，也能自由组合模型、调整参数并实时预览效果，轻松实现从基础文生图到多步骤高清修复等各类复杂任务。ComfyUI 拥有极佳的兼容性，不仅支持 Windows、macOS 和 Linux 全平台，还广泛适配 NVIDIA、AMD、Intel 及苹果 Silicon 等多种硬件架构，并率先支持 SDXL、Flux、SD3 等前沿模型。\n\n无论是希望深入探索算法潜力的研究人员和开发者，还是追求极致创作自由度的设计师与资深 AI 绘画爱好者，ComfyUI 都能提供强大的支持。其独特的模块化架构允许社区不断扩展新功能，使其成为当前最灵活、生态最丰富的开源扩散模型工具之一，帮助用户将创意高效转化为现实。",107662,"2026-04-03T11:11:01",[13,14,15],{"id":36,"name":37,"github_repo":38,"description_zh":39,"stars":40,"difficulty_score":23,"last_commit_at":41,"category_tags":42,"status":16},3704,"NextChat","ChatGPTNextWeb\u002FNextChat","NextChat 是一款轻量且极速的 AI 助手，旨在为用户提供流畅、跨平台的大模型交互体验。它完美解决了用户在多设备间切换时难以保持对话连续性，以及面对众多 AI 模型不知如何统一管理的痛点。无论是日常办公、学习辅助还是创意激发，NextChat 都能让用户随时随地通过网页、iOS、Android、Windows、MacOS 或 Linux 端无缝接入智能服务。\n\n这款工具非常适合普通用户、学生、职场人士以及需要私有化部署的企业团队使用。对于开发者而言，它也提供了便捷的自托管方案，支持一键部署到 Vercel 或 Zeabur 等平台。\n\nNextChat 的核心亮点在于其广泛的模型兼容性，原生支持 Claude、DeepSeek、GPT-4 及 Gemini Pro 等主流大模型，让用户在一个界面即可自由切换不同 AI 能力。此外，它还率先支持 MCP（Model Context Protocol）协议，增强了上下文处理能力。针对企业用户，NextChat 提供专业版解决方案，具备品牌定制、细粒度权限控制、内部知识库整合及安全审计等功能，满足公司对数据隐私和个性化管理的高标准要求。",87618,"2026-04-05T07:20:52",[13,26],{"id":44,"name":45,"github_repo":46,"description_zh":47,"stars":48,"difficulty_score":23,"last_commit_at":49,"category_tags":50,"status":16},2268,"ML-For-Beginners","microsoft\u002FML-For-Beginners","ML-For-Beginners 是由微软推出的一套系统化机器学习入门课程，旨在帮助零基础用户轻松掌握经典机器学习知识。这套课程将学习路径规划为 12 周，包含 26 节精炼课程和 52 道配套测验，内容涵盖从基础概念到实际应用的完整流程，有效解决了初学者面对庞大知识体系时无从下手、缺乏结构化指导的痛点。\n\n无论是希望转型的开发者、需要补充算法背景的研究人员，还是对人工智能充满好奇的普通爱好者，都能从中受益。课程不仅提供了清晰的理论讲解，还强调动手实践，让用户在循序渐进中建立扎实的技能基础。其独特的亮点在于强大的多语言支持，通过自动化机制提供了包括简体中文在内的 50 多种语言版本，极大地降低了全球不同背景用户的学习门槛。此外，项目采用开源协作模式，社区活跃且内容持续更新，确保学习者能获取前沿且准确的技术资讯。如果你正寻找一条清晰、友好且专业的机器学习入门之路，ML-For-Beginners 将是理想的起点。",84991,"2026-04-05T10:45:23",[14,51,52,53,15,54,26,13,55],"数据工具","视频","插件","其他","音频",{"id":57,"name":58,"github_repo":59,"description_zh":60,"stars":61,"difficulty_score":10,"last_commit_at":62,"category_tags":63,"status":16},3128,"ragflow","infiniflow\u002Fragflow","RAGFlow 是一款领先的开源检索增强生成（RAG）引擎，旨在为大语言模型构建更精准、可靠的上下文层。它巧妙地将前沿的 RAG 技术与智能体（Agent）能力相结合，不仅支持从各类文档中高效提取知识，还能让模型基于这些知识进行逻辑推理和任务执行。\n\n在大模型应用中，幻觉问题和知识滞后是常见痛点。RAGFlow 通过深度解析复杂文档结构（如表格、图表及混合排版），显著提升了信息检索的准确度，从而有效减少模型“胡编乱造”的现象，确保回答既有据可依又具备时效性。其内置的智能体机制更进一步，使系统不仅能回答问题，还能自主规划步骤解决复杂问题。\n\n这款工具特别适合开发者、企业技术团队以及 AI 研究人员使用。无论是希望快速搭建私有知识库问答系统，还是致力于探索大模型在垂直领域落地的创新者，都能从中受益。RAGFlow 提供了可视化的工作流编排界面和灵活的 API 接口，既降低了非算法背景用户的上手门槛，也满足了专业开发者对系统深度定制的需求。作为基于 Apache 2.0 协议开源的项目，它正成为连接通用大模型与行业专有知识之间的重要桥梁。",77062,"2026-04-04T04:44:48",[15,14,13,26,54],{"id":65,"github_repo":66,"name":67,"description_en":68,"description_zh":69,"ai_summary_zh":69,"readme_en":70,"readme_zh":71,"quickstart_zh":72,"use_case_zh":73,"hero_image_url":74,"owner_login":75,"owner_name":76,"owner_avatar_url":77,"owner_bio":78,"owner_company":78,"owner_location":79,"owner_email":80,"owner_twitter":78,"owner_website":81,"owner_url":82,"languages":83,"stars":108,"forks":109,"last_commit_at":110,"license":111,"difficulty_score":10,"env_os":112,"env_gpu":113,"env_ram":114,"env_deps":115,"category_tags":123,"github_topics":124,"view_count":23,"oss_zip_url":78,"oss_zip_packed_at":78,"status":16,"created_at":143,"updated_at":144,"faqs":145,"releases":176},1978,"prabhuomkar\u002Fpytorch-cpp","pytorch-cpp","C++ Implementation of PyTorch Tutorials for Everyone","pytorch-cpp 是一套用 C++ 重写的 PyTorch 教程，帮助开发者和研究人员在不依赖 Python 的环境下学习深度学习核心概念。它将原本用 Python 编写的经典教程（如线性回归、CNN、GAN、变分自编码器等）完整移植到 C++，基于 LibTorch 实现，让习惯使用 C++ 或需要部署到生产环境的用户可以直接在 C++ 中实践和调试模型。这个项目解决了 PyTorch 生态中 C++ 教程稀缺的问题，尤其适合希望将模型从研究迁移到嵌入式、高性能或服务器端 C++ 系统的工程师。项目支持 macOS、Linux 和 Windows，通过 CMake 构建，兼容 LibTorch 2.8.0，并提供 Jupyter Notebook 交互式教程，方便快速上手。无论你是想深入理解 PyTorch 底层机制，还是需要在 C++ 环境中部署模型，pytorch-cpp 都能为你提供清晰、可运行的代码示例。","\u003Ch1 align=\"center\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fprabhuomkar_pytorch-cpp_readme_67252bc758e3.jpg\" width=\"50%\">\n\u003C\u002Fh1>\n\u003Cp align=\"center\">\n    C++ Implementation of PyTorch Tutorials for Everyone\n    \u003Cbr \u002F>\n\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Flicense\u002Fprabhuomkar\u002Fpytorch-cpp\">\n\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Flibtorch-2.8.0-ee4c2c\">\n\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fcmake-3.28.6-064f8d\">\n\u003C\u002Fp>\n\n\n| OS (Compiler)\\\\LibTorch |                                                  2.8.0                                                |\n| :--------------------- | :--------------------------------------------------------------------------------------------------- |\n|    macOS (clang 15, 16)    | [![Status](https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Factions\u002Fworkflows\u002Fbuild_macos.yml\u002Fbadge.svg?branch=master)](https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Factions?query=workflow%3Aci-build-macos) |\n|      Linux (gcc 13, 14)      | [![Status](https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Factions\u002Fworkflows\u002Fbuild_ubuntu.yml\u002Fbadge.svg?branch=master)](https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Factions?query=workflow%3Aci-build-ubuntu) |\n|    Windows (msvc 2022, 2025)  | [![Status](https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Factions\u002Fworkflows\u002Fbuild_windows.yml\u002Fbadge.svg?branch=master)](https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Factions?query=workflow%3Aci-build-windows) |\n\n## Table of Contents\n\nThis repository provides tutorial code in C++ for deep learning researchers to learn PyTorch _(i.e. Section 1 to 3)_  \n**Python Tutorial**: [https:\u002F\u002Fgithub.com\u002Fyunjey\u002Fpytorch-tutorial](https:\u002F\u002Fgithub.com\u002Fyunjey\u002Fpytorch-tutorial)\n\n### 1. Basics\n* [PyTorch Basics](tutorials\u002Fbasics\u002Fpytorch_basics\u002Fmain.cpp)\n* [Linear Regression](tutorials\u002Fbasics\u002Flinear_regression\u002Fmain.cpp)\n* [Logistic Regression](tutorials\u002Fbasics\u002Flogistic_regression\u002Fmain.cpp)\n* [Feedforward Neural Network](tutorials\u002Fbasics\u002Ffeedforward_neural_network\u002Fsrc\u002Fmain.cpp)\n\n### 2. Intermediate\n* [Convolutional Neural Network](tutorials\u002Fintermediate\u002Fconvolutional_neural_network\u002Fsrc\u002Fmain.cpp)\n* [Deep Residual Network](tutorials\u002Fintermediate\u002Fdeep_residual_network\u002Fsrc\u002Fmain.cpp)\n* [Recurrent Neural Network](tutorials\u002Fintermediate\u002Frecurrent_neural_network\u002Fsrc\u002Fmain.cpp)\n* [Bidirectional Recurrent Neural Network](tutorials\u002Fintermediate\u002Fbidirectional_recurrent_neural_network\u002Fsrc\u002Fmain.cpp)\n* [Language Model (RNN-LM)](tutorials\u002Fintermediate\u002Flanguage_model\u002Fsrc\u002Fmain.cpp)\n\n### 3. Advanced\n* [Generative Adversarial Networks](tutorials\u002Fadvanced\u002Fgenerative_adversarial_network\u002Fmain.cpp)\n* [Variational Auto-Encoder](tutorials\u002Fadvanced\u002Fvariational_autoencoder\u002Fsrc\u002Fmain.cpp)\n* [Neural Style Transfer](tutorials\u002Fadvanced\u002Fneural_style_transfer\u002Fsrc\u002Fmain.cpp)\n* [Image Captioning (CNN-AttentionRNN)](tutorials\u002Fadvanced\u002Fimage_captioning\u002Fsrc\u002Fmain.cpp)\n\n### 4. Interactive Tutorials\n* [Tensor Slicing](notebooks\u002Ftensor_slicing.ipynb)\n\n### 5. Other Popular Tutorials\n* [Deep Learning with PyTorch: A 60 Minute Blitz](tutorials\u002Fpopular\u002Fblitz)\n\n# Getting Started\n\n## Requirements\n\n1. [C++-17](http:\u002F\u002Fwww.cplusplus.com\u002Fdoc\u002Ftutorial\u002Fintroduction\u002F) compatible compiler\n2. [CMake](https:\u002F\u002Fcmake.org\u002Fdownload\u002F) (minimum version 3.28.6)\n3. [LibTorch version >= 1.12.0 and \u003C= 2.8.0](https:\u002F\u002Fpytorch.org\u002Fcppdocs\u002Finstalling.html)\n4. [Conda](https:\u002F\u002Fdocs.conda.io\u002Fprojects\u002Fconda\u002Fen\u002Flatest\u002Fuser-guide\u002Finstall\u002Fdownload.html)\n\n\n## For Interactive Tutorials\n\n**Note**: Interactive Tutorials are currently running on **LibTorch Nightly Version**.  \nSo there are some tutorials which can break when working with _nightly version_.\n\n```bash\nconda create --name pytorch-cpp\nconda activate pytorch-cpp\nconda install xeus-cling notebook -c conda-forge\n```\n## Clone, build and run tutorials\n### In Google Colab\n[![Open In Colab](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002Fprabhuomkar\u002Fpytorch-cpp\u002Fblob\u002Fmaster\u002Fnotebooks\u002Fpytorch_cpp_colab_notebook.ipynb)\n\n### On Local Machine\n\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp.git\ncd pytorch-cpp\n```\n\n#### Generate build system\n\n```bash\ncmake -B build #\u003Coptions>\n```\n> **_Note for Windows users:_**\u003Cbr> \n> Libtorch only supports 64bit Windows and an x64 generator needs to be specified. For Visual Studio this can be done by appending `-A x64` to the above command.\n\nSome useful options:\n\n| Option       | Default           | Description  |\n| :------------- |:------------|-----:|\n| `-D CUDA_V=(11.8\\|12.4\\|12.6\\|12.8\\|12.9\\|none)`     | `none` | Download LibTorch for a CUDA version (`none` = download CPU version). |\n| `-D LIBTORCH_DOWNLOAD_BUILD_TYPE=(Release\\|Debug)` | `Release` | Determines which libtorch build type version to download (only relevant on **Windows**).|\n| `-D DOWNLOAD_DATASETS=(OFF\\|ON)`     | `ON`      |   Download required datasets during build (only if they do not already exist in `pytorch-cpp\u002Fdata`). |\n|`-D CREATE_SCRIPTMODULES=(OFF\\|ON)` | `OFF` | Create all required scriptmodule files for prelearned models \u002F weights during build. Requires installed  python3 with  pytorch and torchvision. |\n| `-D CMAKE_PREFIX_PATH=path\u002Fto\u002Flibtorch\u002Fshare\u002Fcmake\u002FTorch` |   `\u003Cempty>`    |    Skip the downloading of LibTorch and use your own local version (see [Requirements](#requirements)) instead. |\n| `-D CMAKE_BUILD_TYPE=(Release\\|Debug\\|...)` | `\u003Cempty>` | Determines the CMake build-type for single-configuration generators (see [CMake docs](https:\u002F\u002Fcmake.org\u002Fcmake\u002Fhelp\u002Flatest\u002Fvariable\u002FCMAKE_BUILD_TYPE.html)).|\n\n\u003Cdetails>\n\u003Csummary>\u003Cb>Example Linux\u003C\u002Fb>\u003C\u002Fsummary>\n\n##### Aim\n* Use existing Python, PyTorch (see [Requirements](#requirements)) and torchvision installation.\n* Download all datasets and create all necessary scriptmodule files.\n\n##### Command\n```bash\ncmake -B build \\\n-D CMAKE_BUILD_TYPE=Release \\\n-D CMAKE_PREFIX_PATH=\u002Fpath\u002Fto\u002Flibtorch\u002Fshare\u002Fcmake\u002FTorch \\\n-D CREATE_SCRIPTMODULES=ON \n```\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>\u003Cb>Example Windows\u003C\u002Fb>\u003C\u002Fsummary>\n\n##### Aim\n* Automatically download LibTorch for CUDA 11.8 (Release version) and all necessary datasets.\n* Do not create scriptmodule files.\n\n##### Command\n```bash\ncmake -B build \\\n-A x64 \\\n-D CUDA_V=11.8\n```\n\u003C\u002Fdetails>\n\n#### Build\n>**_Note for Windows (Visual Studio) users:_** \u003Cbr>\n>The CMake script downloads the *Release* version of LibTorch, so `--config Release` has to be appended to the build command.\n\n**How dataset download and scriptmodule creation work:**\n* If `DOWNLOAD_DATASETS` is `ON`, the datasets required by the tutorials you choose to build will be downloaded to `pytorch-cpp\u002Fdata` (if they do not already exist there).\n* If `CREATE_SCRIPTMODULES` is `ON`, the scriptmodule files for the prelearned models \u002F weights required by the tutorials you choose to build will be created in the `model` folder of the respective tutorial's source folder (if they do not already exist).\n#### All tutorials\nTo build all tutorials use\n```bash\ncmake --build build\n```\n\n#### All tutorials in a category\nYou can choose to only build tutorials in one of the categories `basics`, `intermediate`, `advanced` or `popular`. For example, if you are only interested in the `basics` tutorials:\n```bash\ncmake --build build --target basics\n# In general: cmake --build build --target {category}\n```\n\n#### Single tutorial\nYou can also choose to only build a single tutorial. For example to build the language model tutorial only: \n```bash\ncmake --build build --target language-model\n# In general: cmake --build build --target {tutorial-name}\n```\n>**_Note_**:  \n> The target argument is the tutorial's foldername with all underscores replaced by hyphens.\n\n>**_Tip for users of CMake version >= 3.15_**:  \n> You can specify several targets separated by spaces, for example:  \n> ```bash \n> cmake --build build --target language-model image-captioning\n> ```\n\n#### Run Tutorials\n1. (**IMPORTANT!**) First change into the tutorial's directory within `build\u002Ftutorials`. For example, assuming you are in the `pytorch-cpp` directory and want to change to the pytorch basics tutorial folder:\n     ```bash\n     cd build\u002Ftutorials\u002Fbasics\u002Fpytorch_basics\n     # In general: cd build\u002Ftutorials\u002F{basics|intermediate|advanced|popular\u002Fblitz}\u002F{tutorial_name}\n     ```\n2. Run the executable. Note that the executable's name is the tutorial's foldername with all underscores replaced with hyphens (e.g. for tutorial folder: `pytorch_basics` -> executable name: `pytorch-basics` (or `pytorch-basics.exe` on Windows)). For example, to run the pytorch basics tutorial:\u003Cbr>\u003Cbr>\n     **Linux\u002FMac**\n     ```bash\n     .\u002Fpytorch-basics\n     # In general: .\u002F{tutorial-name}\n     ```\n     **Windows**\n     ```powershell\n     .\\pytorch-basics.exe\n     # In general: .\\{tutorial-name}.exe\n     ```\n\n### Using Docker\n\nFind the latest and previous version images on [Docker Hub](https:\u002F\u002Fhub.docker.com\u002Frepository\u002Fdocker\u002Fprabhuomkar\u002Fpytorch-cpp).\n\nYou can build and run the tutorials (on CPU) in a Docker container using the provided `Dockerfile` and `docker-compose.yml` files:  \n1. From the root directory of the cloned repo build the image:\n    ```bash\n    docker-compose build --build-arg USER_ID=$(id -u) --build-arg GROUP_ID=$(id -g)\n    ```\n    > **_Note_**:  \n    > When you run the Docker container, the host repo directory is mounted as a volume in the Docker container in order to cache build and downloaded dependency files so that it is not necessary to rebuild or redownload everything when a container is restarted. In order to have correct file permissions it is necessary to provide your user and group ids as build arguments when building the image on Linux.\n2. Now start the container and build the tutorials using:\n    ```bash\n    docker-compose run --rm pytorch-cpp\n    ```\n    This fetches all necessary dependencies and builds all tutorials.\n    After the build is done, by default the container starts `bash` in interactive mode in the `build\u002Ftutorials` folder.  \n    As with the local build, you can choose to only build tutorials of a category (`basics`, `intermediate`, `advanced`, `popular`):\n    ```bash\n    docker-compose run --rm pytorch-cpp {category}\n    ```\n    In this case the container is started in the chosen category's base build directory.  \n    Alternatively, you can also directly run a tutorial by instead invoking the run command with a tutorial name as additional argument, for example:\n    ```bash\n    docker-compose run --rm pytorch-cpp pytorch-basics\n    # In general: docker-compose run --rm pytorch-cpp {tutorial-name} \n    ```\n    This will - if necessary - build the pytorch-basics tutorial and then start the executable in a container.\n\n## License\nThis repository is licensed under MIT as given in [LICENSE](LICENSE).\n","\u003Ch1 align=\"center\">\n    \u003Cimg src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fprabhuomkar_pytorch-cpp_readme_67252bc758e3.jpg\" width=\"50%\">\n\u003C\u002Fh1>\n\u003Cp align=\"center\">\n    面向所有人的 PyTorch 教程 C++ 实现\n    \u003Cbr \u002F>\n\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Flicense\u002Fprabhuomkar\u002Fpytorch-cpp\">\n\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Flibtorch-2.8.0-ee4c2c\">\n\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fcmake-3.28.6-064f8d\">\n\u003C\u002Fp>\n\n\n| 操作系统（编译器）\\\\LibTorch |                                                  2.8.0                                                |\n| :--------------------- | :--------------------------------------------------------------------------------------------------- |\n|    macOS (clang 15, 16)    | [![状态](https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Factions\u002Fworkflows\u002Fbuild_macos.yml\u002Fbadge.svg?branch=master)](https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Factions?query=workflow%3Aci-build-macos) |\n|      Linux (gcc 13, 14)      | [![状态](https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Factions\u002Fworkflows\u002Fbuild_ubuntu.yml\u002Fbadge.svg?branch=master)](https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Factions?query=workflow%3Aci-build-ubuntu) |\n|    Windows (msvc 2022, 2025)  | [![状态](https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Factions\u002Fworkflows\u002Fbuild_windows.yml\u002Fbadge.svg?branch=master)](https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Factions?query=workflow%3Aci-build-windows) |\n\n## 目录\n\n本仓库为深度学习研究人员提供了用 C++ 编写的教程代码，帮助大家学习 PyTorch _(即第 1 至 3 节)_  \n**Python 教程**: [https:\u002F\u002Fgithub.com\u002Fyunjey\u002Fpytorch-tutorial](https:\u002F\u002Fgithub.com\u002Fyunjey\u002Fpytorch-tutorial)\n\n### 1. 基础知识\n* [PyTorch 基础](tutorials\u002Fbasics\u002Fpytorch_basics\u002Fmain.cpp)\n* [线性回归](tutorials\u002Fbasics\u002Flinear_regression\u002Fmain.cpp)\n* [逻辑回归](tutorials\u002Fbasics\u002Flogistic_regression\u002Fmain.cpp)\n* [前馈神经网络](tutorials\u002Fbasics\u002Ffeedforward_neural_network\u002Fsrc\u002Fmain.cpp)\n\n### 2. 中级\n* [卷积神经网络](tutorials\u002Fintermediate\u002Fconvolutional_neural_network\u002Fsrc\u002Fmain.cpp)\n* [深度残差网络](tutorials\u002Fintermediate\u002Fdeep_residual_network\u002Fsrc\u002Fmain.cpp)\n* [循环神经网络](tutorials\u002Fintermediate\u002Frecurrent_neural_network\u002Fsrc\u002Fmain.cpp)\n* [双向循环神经网络](tutorials\u002Fintermediate\u002Fbidirectional_recurrent_neural_network\u002Fsrc\u002Fmain.cpp)\n* [语言模型（RNN-LM）](tutorials\u002Fintermediate\u002Flanguage_model\u002Fsrc\u002Fmain.cpp)\n\n### 3. 高级\n* [生成对抗网络](tutorials\u002Fadvanced\u002Fgenerative_adversarial_network\u002Fmain.cpp)\n* [变分自编码器](tutorials\u002Fadvanced\u002Fvariational_autoencoder\u002Fsrc\u002Fmain.cpp)\n* [神经风格迁移](tutorials\u002Fadvanced\u002Fneural_style_transfer\u002Fsrc\u002Fmain.cpp)\n* [图像描述（CNN-AttentionRNN）](tutorials\u002Fadvanced\u002Fimage_captioning\u002Fsrc\u002Fmain.cpp)\n\n### 4. 交互式教程\n* [张量切片](notebooks\u002Ftensor_slicing.ipynb)\n\n### 5. 其他热门教程\n* [用 PyTorch 进行深度学习：60 分钟速成](tutorials\u002Fpopular\u002Fblitz)\n\n# 快速入门\n\n## 要求\n\n1. 兼容 [C++-17](http:\u002F\u002Fwww.cplusplus.com\u002Fdoc\u002Ftutorial\u002Fintroduction\u002F) 的编译器\n2. [CMake](https:\u002F\u002Fcmake.org\u002Fdownload\u002F)（最低版本 3.28.6）\n3. [LibTorch 版本 >= 1.12.0 且 \u003C= 2.8.0](https:\u002F\u002Fpytorch.org\u002Fcppdocs\u002Finstalling.html)\n4. [Conda](https:\u002F\u002Fdocs.conda.io\u002Fprojects\u002Fconda\u002Fen\u002Flatest\u002Fuser-guide\u002Finstall\u002Fdownload.html)\n\n\n## 对于交互式教程\n\n**注意**：目前交互式教程运行的是 **LibTorch Nightly 版本**。  \n因此，部分教程在使用 _nightly 版本_ 时可能会出现兼容性问题。\n\n```bash\nconda create --name pytorch-cpp\nconda activate pytorch-cpp\nconda install xeus-cling notebook -c conda-forge\n```\n## 克隆、构建并运行教程\n### 在 Google Colab 中\n[![在 Colab 中打开](https:\u002F\u002Fcolab.research.google.com\u002Fassets\u002Fcolab-badge.svg)](https:\u002F\u002Fcolab.research.google.com\u002Fgithub\u002Fprabhuomkar\u002Fpytorch-cpp\u002Fblob\u002Fmaster\u002Fnotebooks\u002Fpytorch_cpp_colab_notebook.ipynb)\n\n### 在本地机器上\n\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp.git\ncd pytorch-cpp\n```\n\n#### 生成构建系统\n\n```bash\ncmake -B build #\u003C选项>\n```\n> **_Windows 用户请注意:_**\u003Cbr> \n> Libtorch 仅支持 64 位 Windows，需要指定 x64 生成器。对于 Visual Studio，可在上述命令后追加 `-A x64`。\n\n一些有用的选项：\n\n| 选项       | 默认值           | 描述  |\n| :------------- |:------------|-----:|\n| `-D CUDA_V=(11.8\\|12.4\\|12.6\\|12.8\\|12.9\\|none)`     | `none` | 下载适用于特定 CUDA 版本的 LibTorch（`none` 表示下载 CPU 版本）。|\n| `-D LIBTORCH_DOWNLOAD_BUILD_TYPE=(Release\\|Debug)` | `Release` | 决定下载哪个版本的 LibTorch 构建类型（仅在 **Windows** 上相关）。|\n| `-D DOWNLOAD_DATASETS=(OFF\\|ON)`     | `ON`      | 在构建过程中下载所需的数据集（仅当它们尚未存在于 `pytorch-cpp\u002Fdata` 中时）。|\n| `-D CREATE_SCRIPTMODULES=(OFF\\|ON)` | `OFF` | 在构建过程中为预训练模型\u002F权重创建所有必需的 scriptmodule 文件。需要安装 Python3，并且已安装 PyTorch 和 torchvision。|\n| `-D CMAKE_PREFIX_PATH=libtorch 的 share\u002Fcmake\u002FTorch 路径` |   `\u003C空>`    | 跳过 LibTorch 的下载，改用您本地的版本（参见 [要求](#requirements)）。|\n| `-D CMAKE_BUILD_TYPE=(Release\\|Debug\\|...)` | `\u003C空>` | 决定单配置生成器的 CMake 构建类型（参见 [CMake 文档](https:\u002F\u002Fcmake.org\u002Fcmake\u002Fhelp\u002Flatest\u002Fvariable\u002FCMAKE_BUILD_TYPE.html)）。|\n\n\u003Cdetails>\n\u003Csummary>\u003Cb>Linux 示例\u003C\u002Fb>\u003C\u002Fsummary>\n\n##### 目标\n* 使用现有的 Python、PyTorch（参见 [要求](#requirements)）和 torchvision 安装。\n* 下载所有数据集并创建所有必要的 scriptmodule 文件。\n\n##### 命令\n```bash\ncmake -B build \\\n-D CMAKE_BUILD_TYPE=Release \\\n-D CMAKE_PREFIX_PATH=\u002Fpath\u002Fto\u002Flibtorch\u002Fshare\u002Fcmake\u002FTorch \\\n-D CREATE_SCRIPTMODULES=ON \n```\n\u003C\u002Fdetails>\n\n\u003Cdetails>\n\u003Csummary>\u003Cb>Windows 示例\u003C\u002Fb>\u003C\u002Fsummary>\n\n##### 目标\n* 自动下载适用于 CUDA 11.8（Release 版本）的 LibTorch 及所有必要数据集。\n* 不创建 scriptmodule 文件。\n\n##### 命令\n```bash\ncmake -B build \\\n-A x64 \\\n-D CUDA_V=11.8\n```\n\u003C\u002Fdetails>\n\n#### 构建\n>**_Windows（Visual Studio）用户请注意:_** \u003Cbr>\n>CMake 脚本会下载 LibTorch 的 *Release* 版本，因此必须在构建命令中追加 `--config Release`。\n\n**数据集下载与 scriptmodule 创建的工作原理：**\n* 如果 `DOWNLOAD_DATASETS` 为 `ON`，您选择构建的教程所需的数据集将被下载到 `pytorch-cpp\u002Fdata`（如果它们尚未存在于此处）。\n* 如果 `CREATE_SCRIPTMODULES` 为 `ON`，您选择构建的教程所需预训练模型\u002F权重的 scriptmodule 文件将在相应教程源文件夹的 `model` 文件夹中创建（如果它们尚未存在）。\n\n#### 所有教程\n要构建所有教程，请使用：\n```bash\ncmake --build build\n```\n\n#### 某个类别的所有教程\n您可以只构建 `basics`、`intermediate`、`advanced` 或 `popular` 这些类别中的某一个。例如，如果您只对 `basics` 教程感兴趣：\n```bash\ncmake --build build --target basics\n# 一般情况下：cmake --build build --target {category}\n```\n\n#### 单个教程\n您也可以只构建某个单独的教程。例如，只构建语言模型教程：\n```bash\ncmake --build build --target language-model\n# 一般情况下：cmake --build build --target {tutorial-name}\n```\n>**_注意_**:  \n>目标参数是教程的文件夹名，其中所有下划线都被替换为连字符。\n\n>**_CMake 3.15 及以上版本用户提示_**:  \n>您可以指定多个目标，以空格分隔，例如：  \n>```bash \n>cmake --build build --target language-model image-captioning\n>```\n\n#### 运行教程\n1. (**重要！**) 首先切换到 `build\u002Ftutorials` 中的教程目录。例如，假设您位于 `pytorch-cpp` 目录下，想切换到 pytorch basics 教程文件夹：\n     ```bash\n     cd build\u002Ftutorials\u002Fbasics\u002Fpytorch_basics\n     # 一般情况下：cd build\u002Ftutorials\u002F{basics|intermediate|advanced|popular\u002Fblitz}\u002F{tutorial_name}\n     ```\n2. 运行可执行文件。注意，可执行文件的名称是教程的文件夹名，其中所有下划线都被替换为连字符（例如，教程文件夹：`pytorch_basics` -> 可执行文件名：`pytorch-basics`（或 Windows 上的 `pytorch-basics.exe`）。例如，运行 pytorch basics 教程：\u003Cbr>\u003Cbr>\n     **Linux\u002FMac**\n     ```bash\n     .\u002Fpytorch-basics\n     # 一般情况下：.\u002F{tutorial-name}\n     ```\n     **Windows**\n     ```powershell\n     .\\pytorch-basics.exe\n     # 一般情况下：.\\{tutorial-name}.exe\n     ```\n\n### 使用 Docker\n\n您可以在 [Docker Hub](https:\u002F\u002Fhub.docker.com\u002Frepository\u002Fdocker\u002Fprabhuomkar\u002Fpytorch-cpp) 上找到最新版和旧版本镜像。\n\n您可以使用提供的 `Dockerfile` 和 `docker-compose.yml` 文件，在 Docker 容器中构建并运行教程（在 CPU 上）：  \n1. 从克隆仓库的根目录构建镜像：\n    ```bash\n    docker-compose build --build-arg USER_ID=$(id -u) --build-arg GROUP_ID=$(id -g)\n    ```\n    > **_注意_**:  \n    > 当您运行 Docker 容器时，宿主机的仓库目录会被挂载为容器内的卷，以便缓存构建和下载的依赖文件，这样在重启容器时无需重新构建或重新下载所有内容。为了确保正确的文件权限，您需要在 Linux 上构建镜像时提供您的用户 ID 和组 ID 作为构建参数。\n2. 现在启动容器并构建教程：\n    ```bash\n    docker-compose run --rm pytorch-cpp\n    ```\n    这会获取所有必要的依赖并构建所有教程。\n    构建完成后，默认情况下容器会在 `build\u002Ftutorials` 文件夹中以交互模式启动 `bash`。  \n    与本地构建一样，您可以只构建某一类别的教程（`basics`、`intermediate`、`advanced`、`popular`）：\n    ```bash\n    docker-compose run --rm pytorch-cpp {category}\n    ```\n    此时容器会在所选类别的基础构建目录中启动。  \n    或者，您也可以直接运行某个教程，只需在运行命令中额外传入教程名称，例如：\n    ```bash\n    docker-compose run --rm pytorch-cpp pytorch-basics\n    # 一般情况下：docker-compose run --rm pytorch-cpp {tutorial-name} \n    ```\n    如果需要，这会构建 pytorch-basics 教程，然后在容器中启动可执行文件。\n\n## 许可证\n本仓库采用 MIT 许可证，详见 [LICENSE](LICENSE)。","# pytorch-cpp 快速上手指南\n\n## 环境准备\n\n- **操作系统**：Windows 64位 \u002F macOS \u002F Linux（推荐 Ubuntu 20.04+）\n- **编译器**：支持 C++17 的编译器（clang 15+\u002Fgcc 13+\u002FMSVC 2022+）\n- **CMake**：≥ 3.28.6（推荐使用 [清华源](https:\u002F\u002Fmirrors.tuna.tsinghua.edu.cn\u002Fcmake\u002F) 安装）\n- **LibTorch**：≥ 1.12.0 且 ≤ 2.8.0（[官方下载](https:\u002F\u002Fpytorch.org\u002Fcppdocs\u002Finstalling.html)）\n- **可选**：Conda（用于交互式教程，推荐使用 [清华源](https:\u002F\u002Fmirrors.tuna.tsinghua.edu.cn\u002Fanaconda\u002F)）\n\n> **Windows 用户注意**：必须使用 x64 架构，CMake 需添加 `-A x64` 参数。\n\n## 安装步骤\n\n```bash\n# 1. 克隆仓库\ngit clone https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp.git\ncd pytorch-cpp\n\n# 2. 生成构建系统（CPU版本）\ncmake -B build\n\n# 若需使用 CUDA 11.8（推荐国内用户）\ncmake -B build -D CUDA_V=11.8\n\n# 若已有本地 LibTorch，指定路径（避免重复下载）\ncmake -B build -D CMAKE_PREFIX_PATH=\u002Fpath\u002Fto\u002Flibtorch\u002Fshare\u002Fcmake\u002FTorch\n\n# 3. 构建所有教程\ncmake --build build\n\n# 或仅构建基础教程（推荐新手）\ncmake --build build --target basics\n```\n\n> **Windows 用户**：构建时添加 `--config Release`  \n> ```bash\n> cmake --build build --config Release\n> ```\n\n## 基本使用\n\n运行最简单的 PyTorch 基础示例：\n\n```bash\n# 进入构建后的教程目录\ncd build\u002Ftutorials\u002Fbasics\u002Fpytorch_basics\n\n# 执行程序（Linux\u002FmacOS）\n.\u002Fpytorch-basics\n\n# 执行程序（Windows PowerShell）\n.\\pytorch-basics.exe\n```\n\n该示例将展示张量创建、运算和自动求导等基础操作，类似 PyTorch Python 版本的入门教程。","某自动驾驶公司嵌入式团队正在为车载AI控制器开发实时目标检测模块，需将训练好的PyTorch模型部署到ARM嵌入式平台，但团队仅有C++开发经验，缺乏Python深度学习背景。\n\n### 没有 pytorch-cpp 时\n- 团队需从零阅读Python版PyTorch教程，再手动翻译成C++，耗时两周仍常出现张量维度错配、自动求导逻辑丢失等问题。\n- LibTorch官方文档以C++ API为主，缺乏完整端到端示例，调试模型推理流程时频繁遇到内存泄漏和算子不匹配错误。\n- 团队尝试直接使用PyTorch Python导出的ONNX模型，但在嵌入式端加载时因算子不支持导致推理失败，排查成本极高。\n- 缺乏可运行的基准代码，无法快速验证模型在目标硬件上的推理延迟是否达标。\n- 新成员上手周期长达一个月，团队开发效率严重受阻。\n\n### 使用 pytorch-cpp 后\n- 直接复用`tutorials\u002Fbasics\u002Ffeedforward_neural_network\u002Fmain.cpp`作为目标检测模型的推理骨架，3天内完成模型加载与前向推理集成。\n- 借助`convolutional_neural_network\u002Fsrc\u002Fmain.cpp`中的卷积层实现，快速适配YOLOv5的C++推理逻辑，避免了手动重写卷积核与填充逻辑。\n- 利用LibTorch 2.8.0的官方兼容版本，确保与车载系统编译环境一致，部署后推理稳定，无内存异常。\n- 通过GitHub CI构建的跨平台示例，直接在ARM Linux目标机上编译运行，首次部署即成功输出检测框。\n- 新入职工程师一周内即可独立维护模型部署模块，团队整体交付周期缩短60%。\n\npytorch-cpp 让C++工程师无需掌握Python深度学习生态，就能快速将PyTorch模型可靠部署到嵌入式系统，真正实现“学得会、跑得通、用得稳”。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fprabhuomkar_pytorch-cpp_67252bc7.jpg","prabhuomkar","Omkar Prabhu","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fprabhuomkar_8ff7abd9.png",null,"Mumbai, India","prabhuomkar@pm.me","omkarprabhu.in","https:\u002F\u002Fgithub.com\u002Fprabhuomkar",[84,88,92,96,100,104],{"name":85,"color":86,"percentage":87},"C++","#f34b7d",68.5,{"name":89,"color":90,"percentage":91},"CMake","#DA3434",14.9,{"name":93,"color":94,"percentage":95},"Jupyter Notebook","#DA5B0B",14.5,{"name":97,"color":98,"percentage":99},"Python","#3572A5",1.1,{"name":101,"color":102,"percentage":103},"Dockerfile","#384d54",0.7,{"name":105,"color":106,"percentage":107},"Shell","#89e051",0.3,2128,274,"2026-04-03T00:09:46","MIT","Linux, macOS, Windows","可选，支持 CUDA 11.8、12.4、12.6、12.8、12.9，需 NVIDIA GPU，显存未明确说明，建议 8GB+","未说明",{"notes":116,"python":117,"dependencies":118},"建议使用 conda 管理环境；首次构建时可选择下载数据集和脚本模块，可能占用数 GB 磁盘空间；Windows 用户需使用 x64 生成器；交互式教程需 LibTorch Nightly 版本，可能存在兼容性问题","3.8+",[119,120,121,122],"LibTorch>=1.12.0","LibTorch\u003C=2.8.0","CMake>=3.28.6","conda",[26,13,51],[125,126,127,128,129,130,131,132,133,134,135,136,137,138,139,140,141,142],"cplusplus","artificial-intelligence","machine-learning","pytorch","torch","tensors","neural-network","autograd","libtorch","recurrent-neural-network","tutorial","generative-adversarial-network","convolutional-neural-network","language-model","interactive-tutorials","colab","datasets","scriptmodule-files","2026-03-27T02:49:30.150509","2026-04-06T06:44:40.380694",[146,151,156,161,166,171],{"id":147,"question_zh":148,"answer_zh":149,"source_url":150},8918,"在 Raspberry Pi 上编译 LibTorch 项目时出现 'file format not recognized' 错误，如何解决？","该错误通常是由于下载了错误架构的 LibTorch 库（如 x86_64）导致的。请确保下载适用于 ARM 架构的 LibTorch 版本（如 libtorch-linux-arm64），并从 PyTorch 官网选择正确的 Linux ARM64 版本重新下载和配置。","https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Fissues\u002F37",{"id":152,"question_zh":153,"answer_zh":154,"source_url":155},8919,"在 WSL2 Ubuntu 中运行 CMake 时提示 'Could not find compatible Torch version' 或 'No module named torch'，如何解决？","首先确保使用与系统匹配的 LibTorch 版本（如 CUDA 或 CPU 版）。其次，CMake 未检测到 Conda 环境中的 Python，需在 CMakeLists.txt 中显式指定 Python 路径：`set(Python3_EXECUTABLE \"\u002Fhome\u002Fyour-username\u002Fminiconda3\u002Fenvs\u002Fpytorch\u002Fbin\u002Fpython3\")`，或在命令行中添加参数：`cmake -B build -D Python3_EXECUTABLE=\u002Fpath\u002Fto\u002Fconda\u002Fpython`。","https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Fissues\u002F121",{"id":157,"question_zh":158,"answer_zh":159,"source_url":160},8920,"如何正确运行该项目中的示例代码？","请按照 README 中的步骤操作：1）克隆仓库；2）创建 build 目录并运行 `cmake -B build`；3）执行 `cmake --build build` 编译；4）运行生成的可执行文件（如 `.\u002Fbuild\u002Fpytorch-cpp`）。若遇到依赖问题，请确保已安装 LibTorch 并正确设置环境变量（如 LD_LIBRARY_PATH）。","https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Fissues\u002F26",{"id":162,"question_zh":163,"answer_zh":164,"source_url":165},8921,"在 GPU 上推理时，调用 `.to(at::kCPU)` 耗时过长，如何优化？","这是因为 CUDA 异步执行导致计时不准。应在 `.to(at::kCPU)` 前后添加 `cudaDeviceSynchronize()` 以确保所有 GPU 操作完成后再计时。此外，尽量避免频繁在 GPU 和 CPU 之间传输数据，可考虑在 GPU 上完成所有计算后再一次性传输结果。","https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Fissues\u002F94",{"id":167,"question_zh":168,"answer_zh":169,"source_url":170},8922,"训练过程中 Loss.backward() 越来越慢，但内存稳定，可能是什么原因？","该问题通常不是框架本身导致，而是模型实现或数据加载逻辑的问题。请检查是否在每次迭代中意外地构建了新的计算图（如重复创建张量或模型），或数据加载器未正确重置。建议对比官方示例代码，确保模型结构和训练循环与标准实现一致。","https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Fissues\u002F63",{"id":172,"question_zh":173,"answer_zh":174,"source_url":175},8923,"如何在项目中启用 C++17 标准以支持新功能（如 ImageNet 示例）？","由于新示例依赖 C++17 特性，需在 CMakeLists.txt 中统一设置 C++ 标准：添加 `set(CMAKE_CXX_STANDARD 17)` 和 `set(CMAKE_CXX_STANDARD_REQUIRED ON)`。虽然这会放弃对极旧编译器的支持，但所有当前 CI 使用的编译器（GCC 7+、Clang 6+）均完全支持 C++17。","https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Fissues\u002F73",[177,182,187,192,197,202,206,210,215,219,224,228],{"id":178,"version":179,"summary_zh":180,"released_at":181},106347,"v2.8.0","## What's Changed\r\n* PyTorch 2.8 Upgrade by @prabhuomkar in https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Fpull\u002F128\r\n\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Fcompare\u002Fv2.6.0...v2.8.0","2025-08-25T05:12:33",{"id":183,"version":184,"summary_zh":185,"released_at":186},106348,"v2.6.0","## What's Changed\r\n* Housekeeping and PyTorch Upgrade by @prabhuomkar in https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Fpull\u002F125\r\n\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Fcompare\u002Fv2.3.0...v2.6.0","2025-08-25T05:11:25",{"id":188,"version":189,"summary_zh":190,"released_at":191},106349,"v2.3.0","## What's Changed\r\n* Upgraded pytorch to 2.3.0 latest version by @prabhuomkar in https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Fpull\u002F123\r\n\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Fcompare\u002Fv2.1.1...v2.3.0","2025-08-25T05:09:55",{"id":193,"version":194,"summary_zh":195,"released_at":196},106350,"v2.1.1","## What's Changed\r\n* Update git URL for stb by @amirtronics in https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Fpull\u002F117\r\n* Upgrade libtorch v2.1.1 by @mfl28 in https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Fpull\u002F120\r\n\r\n## New Contributors\r\n* @amirtronics made their first contribution in https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Fpull\u002F117\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Fcompare\u002Fv2.0.0...v2.1.1","2024-05-05T05:39:50",{"id":198,"version":199,"summary_zh":200,"released_at":201},106351,"v2.0.0","## What's Changed\r\n* Update CI runner settings by @mfl28 in https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Fpull\u002F103\r\n* Upgrade to libtorch v1.13.1 by @mfl28 in https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Fpull\u002F106\r\n* Upgrade libtorch v2.0.0 by @mfl28 in https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Fpull\u002F108\r\n\r\n\r\n**Full Changelog**: https:\u002F\u002Fgithub.com\u002Fprabhuomkar\u002Fpytorch-cpp\u002Fcompare\u002Fv1.12...v2.0.0","2023-07-29T11:40:41",{"id":203,"version":204,"summary_zh":78,"released_at":205},106352,"v1.12","2022-09-28T04:54:25",{"id":207,"version":208,"summary_zh":78,"released_at":209},106353,"v1.10.1","2022-09-28T04:53:57",{"id":211,"version":212,"summary_zh":213,"released_at":214},106354,"v1.9","CI Change & Download Changes","2022-09-28T04:53:27",{"id":216,"version":217,"summary_zh":78,"released_at":218},106355,"v1.8","2022-09-28T04:52:45",{"id":220,"version":221,"summary_zh":222,"released_at":223},106356,"v1.7","PyTorch Blitz Tutorial, Download datasets as required and PyTorch version upgrades.","2021-02-27T13:36:08",{"id":225,"version":226,"summary_zh":78,"released_at":227},106357,"v1.5","2020-06-28T05:40:55",{"id":229,"version":230,"summary_zh":78,"released_at":231},106358,"v1.4","2020-04-23T12:56:32"]