[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-denizyuret--Knet.jl":3,"tool-denizyuret--Knet.jl":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":76,"owner_company":76,"owner_location":76,"owner_email":76,"owner_twitter":76,"owner_website":76,"owner_url":78,"languages":79,"stars":100,"forks":101,"last_commit_at":102,"license":103,"difficulty_score":10,"env_os":104,"env_gpu":104,"env_ram":104,"env_deps":105,"category_tags":110,"github_topics":111,"view_count":10,"oss_zip_url":76,"oss_zip_packed_at":76,"status":16,"created_at":118,"updated_at":119,"faqs":120,"releases":141},1223,"denizyuret\u002FKnet.jl","Knet.jl","Koç University deep learning framework.","Knet.jl是由Koç大学开发的深度学习框架，基于Julia语言实现。它让开发者能用简洁的代码快速构建和训练神经网络模型，支持GPU加速和自动微分，通过动态计算图简化了模型定义过程。Knet.jl解决了深度学习开发中代码冗长、性能优化困难的问题，特别适合熟悉Julia语言的开发者和研究人员。它的独特亮点是极简的API设计——例如，仅需15行代码就能实现LeNet模型，并在10秒内完成MNIST手写数字识别训练。框架提供了完善的教程、文档和示例，帮助用户快速上手，无需复杂配置。如果你在Julia生态中工作，Knet.jl能让你专注于模型创新，轻松实现高效深度学习实验。","# Knet\n\n[![](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fdocs-latest-blue.svg)](https:\u002F\u002Fdenizyuret.github.io\u002FKnet.jl\u002Flatest) \n[![](https:\u002F\u002Ftravis-ci.org\u002Fdenizyuret\u002FKnet.jl.svg?branch=master)](https:\u002F\u002Ftravis-ci.org\u002Fdenizyuret\u002FKnet.jl) \n[![](https:\u002F\u002Fcoveralls.io\u002Frepos\u002Fgithub\u002Fdenizyuret\u002FKnet.jl\u002Fbadge.svg?branch=master)](https:\u002F\u002Fcoveralls.io\u002Fgithub\u002Fdenizyuret\u002FKnet.jl?branch=master)\n[![](https:\u002F\u002Fcodecov.io\u002Fgh\u002Fdenizyuret\u002FKnet.jl\u002Fbranch\u002Fmaster\u002Fgraph\u002Fbadge.svg)](https:\u002F\u002Fcodecov.io\u002Fgh\u002Fdenizyuret\u002FKnet.jl)\n\n[Knet](https:\u002F\u002Fdenizyuret.github.io\u002FKnet.jl\u002Flatest) (pronounced \"kay-net\") is the [Koç\nUniversity](http:\u002F\u002Fwww.ku.edu.tr\u002Fen) deep learning framework implemented in\n[Julia](http:\u002F\u002Fdocs.julialang.org) by [Deniz Yuret](http:\u002F\u002Fwww.denizyuret.com) and\ncollaborators.  It supports GPU operation and automatic differentiation using dynamic\ncomputational graphs for models defined in plain Julia. You can install Knet with the \nfollowing at the julia prompt: `using Pkg; Pkg.add(\"Knet\")`. Some starting points:\n\n* [Tutorial:](tutorial) \n  introduces Julia and Knet via examples.\n* [Documentation:](https:\u002F\u002Fdenizyuret.github.io\u002FKnet.jl\u002Flatest)\n  installation, introduction, design, implementation, full reference and deep learning chapters.\n* [Examples:](examples)\n  more tutorials and example models.\n* [Benchmarks:](http:\u002F\u002Fdenizyuret.github.io\u002FKnet.jl\u002Flatest\u002Ftutorial\u002F#Benchmarks-1)\n  comparison of Knet's speed with TensorFlow, PyTorch, DyNet etc.\n* [Paper:](https:\u002F\u002Fgoo.gl\u002FzeUBFr)\n  Yuret, D. \"Knet: beginning deep learning with 100 lines of julia.\" In *Machine Learning Systems Workshop* at NIPS 2016.\n* [KnetML:](https:\u002F\u002Fgithub.com\u002FKnetML)\n  github organization with Knet repos of models, tutorials, layer collections and other resources.\n* [Images:](http:\u002F\u002Fdenizyuret.github.io\u002FKnet.jl\u002Flatest\u002Finstall\u002F#Using-Amazon-AWS-1)\n  Knet machine images are available for [AWS](http:\u002F\u002Fdenizyuret.github.io\u002FKnet.jl\u002Flatest\u002Finstall\u002F#Using-Amazon-AWS-1), [Singularity](https:\u002F\u002Fgithub.com\u002FKnetML\u002Fsingularity-images) and [Docker](https:\u002F\u002Fgithub.com\u002FJuliaGPU\u002Fdocker).\n* [Issues:](https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Fissues)\n  if you find a bug, please open a github issue.\n* [knet-users:](https:\u002F\u002Fgroups.google.com\u002Fforum\u002F#!forum\u002Fknet-users)\n  if you need help or would like to request a feature, please join this mailing list.\n* [knet-dev:](https:\u002F\u002Fgroups.google.com\u002Fforum\u002F#!forum\u002Fknet-dev)\n  if you would like to contribute to Knet development, please join this mailing list and check out these [tips](https:\u002F\u002Fdenizyuret.github.io\u002FKnet.jl\u002Flatest\u002Finstall\u002F#Tips-for-developers-1).\n* [knet-slack:](https:\u002F\u002Fjulialang.slack.com\u002Fmessages\u002FCDLKQ92P3\u002Fdetails) Slack channel for Knet.\n* Related work: Please check out [Flux](https:\u002F\u002Fgithub.com\u002FFLuxML), [Mocha](https:\u002F\u002Fgithub.com\u002Fpluskid\u002FMocha.jl), [JuliaML](https:\u002F\u002Fgithub.com\u002FJuliaML), [JuliaDiff](https:\u002F\u002Fgithub.com\u002FJuliaDiff), [JuliaGPU](https:\u002F\u002Fgithub.com\u002FJuliaGPU), [JuliaOpt](https:\u002F\u002Fgithub.com\u002FJuliaOpt) for related packages.\n\n## Example\n\nHere is a simple example where we define, train and test the\n[LeNet](http:\u002F\u002Fyann.lecun.com\u002Fexdb\u002Flenet) model for the\n[MNIST](http:\u002F\u002Fyann.lecun.com\u002Fexdb\u002Fmnist) handwritten digit recognition dataset from scratch\nusing 15 lines of code and 10 seconds of GPU computation.\n\n```julia\n# Install packages before first run: using Pkg; pkg\"add Knet IterTools MLDatasets\"\nusing Knet, IterTools, MLDatasets\n\n# Define convolutional layer:\nstruct Conv; w; b; end\nConv(w1,w2,nx,ny) = Conv(param(w1,w2,nx,ny), param0(1,1,ny,1))\n(c::Conv)(x) = relu.(pool(conv4(c.w, x) .+ c.b))\n\n# Define dense layer:\nstruct Dense; w; b; f; end\nDense(i,o; f=identity) = Dense(param(o,i), param0(o), f)\n(d::Dense)(x) = d.f.(d.w * mat(x) .+ d.b)\n\n# Define a chain of layers and a loss function:\nstruct Chain; layers; end\n(c::Chain)(x) = (for l in c.layers; x = l(x); end; x)\n(c::Chain)(x,y) = nll(c(x),y)\n\n# Load MNIST data:\nxtrn,ytrn = MNIST.traindata(Float32); ytrn[ytrn.==0] .= 10\nxtst,ytst = MNIST.testdata(Float32);  ytst[ytst.==0] .= 10\ndtrn = minibatch(xtrn, ytrn, 100; xsize = (28,28,1,:))\ndtst = minibatch(xtst, ytst, 100; xsize = (28,28,1,:))\n\n# Define and train LeNet (~10 secs on a GPU or ~3 mins on a CPU to reach ~99% accuracy)\nLeNet = Chain((Conv(5,5,1,20), Conv(5,5,20,50), Dense(800,500,f=relu), Dense(500,10)))\nprogress!(adam(LeNet, ncycle(dtrn,3)))\naccuracy(LeNet,data=dtst)\n```\n\n## Contributing\n\nKnet is an open-source project and we are always open to new contributions: bug reports and\nfixes, feature requests and contributions, new machine learning models and operators,\ninspiring examples, benchmarking results are all welcome. See [Tips for Developers](https:\u002F\u002Fdenizyuret.github.io\u002FKnet.jl\u002Flatest\u002Finstall\u002F#Tips-for-developers) for instructions.\n\nContributors: Can Gümeli, Carlo Lucibello, Ege Onat, Ekin Akyürek, Ekrem Emre Yurdakul, Emre Ünal, Emre Yolcu, Enis Berk, Erenay Dayanık, İlker Kesen, Kai Xu, Meriç Melike Softa, Mike Innes, Onur Kuru, Ozan Arkan Can, Ömer Kırnap, Phuoc Nguyen, Rene Donner, Tim Besard, Zhang Shiwei.\n","# Knet\n\n[![](https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fdocs-latest-blue.svg)](https:\u002F\u002Fdenizyuret.github.io\u002FKnet.jl\u002Flatest) \n[![](https:\u002F\u002Ftravis-ci.org\u002Fdenizyuret\u002FKnet.jl.svg?branch=master)](https:\u002F\u002Ftravis-ci.org\u002Fdenizyuret\u002FKnet.jl) \n[![](https:\u002F\u002Fcoveralls.io\u002Frepos\u002Fgithub\u002Fdenizyuret\u002FKnet.jl\u002Fbadge.svg?branch=master)](https:\u002F\u002Fcoveralls.io\u002Fgithub\u002Fdenizyuret\u002FKnet.jl?branch=master)\n[![](https:\u002F\u002Fcodecov.io\u002Fgh\u002Fdenizyuret\u002FKnet.jl\u002Fbranch\u002Fmaster\u002Fgraph\u002Fbadge.svg)](https:\u002F\u002Fcodecov.io\u002Fgh\u002Fdenizyuret\u002FKnet.jl)\n\n[Knet](https:\u002F\u002Fdenizyuret.github.io\u002FKnet.jl\u002Flatest)（发音为“kay-net”）是由[德尼兹·尤雷特](http:\u002F\u002Fwww.denizyuret.com)及其合作者在[Julia](http:\u002F\u002Fdocs.julialang.org)中实现的[科奇大学](http:\u002F\u002Fwww.ku.edu.tr\u002Fen)深度学习框架。它支持GPU运算，并使用动态计算图进行自动微分，适用于用纯Julia定义的模型。您可以在Julia提示符下通过以下命令安装Knet：`using Pkg; Pkg.add(\"Knet\")`。以下是一些入门资源：\n\n* [教程：](tutorial)  \n  通过示例介绍Julia和Knet。\n* [文档：](https:\u002F\u002Fdenizyuret.github.io\u002FKnet.jl\u002Flatest)\n  包括安装、简介、设计、实现、完整参考以及深度学习相关章节。\n* [示例：](examples)\n  更多教程和示例模型。\n* [基准测试：](http:\u002F\u002Fdenizyuret.github.io\u002FKnet.jl\u002Flatest\u002Ftutorial\u002F#Benchmarks-1)\n  将Knet的速度与TensorFlow、PyTorch、DyNet等进行比较。\n* [论文：](https:\u002F\u002Fgoo.gl\u002FzeUBFr)\n  尤雷特，D. “Knet：用100行Julia代码开始深度学习。”载于2016年NIPS会议的机器学习系统研讨会。\n* [KnetML：](https:\u002F\u002Fgithub.com\u002FKnetML)\n  一个包含Knet模型、教程、层集合及其他资源的GitHub组织。\n* [镜像：](http:\u002F\u002Fdenizyuret.github.io\u002FKnet.jl\u002Flatest\u002Finstall\u002F#Using-Amazon-AWS-1)\n  Knet的机器镜像可在[AWS](http:\u002F\u002Fdenizyuret.github.io\u002FKnet.jl\u002Flatest\u002Finstall\u002F#Using-Amazon-AWS-1)、[Singularity](https:\u002F\u002Fgithub.com\u002FKnetML\u002Fsingularity-images)和[Docker](https:\u002F\u002Fgithub.com\u002FJuliaGPU\u002Fdocker)上使用。\n* [问题：](https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Fissues)\n  如果您发现任何错误，请在GitHub上提交问题。\n* [knet-users：](https:\u002F\u002Fgroups.google.com\u002Fforum\u002F#!forum\u002Fknet-users)\n  如需帮助或希望请求新功能，请加入此邮件列表。\n* [knet-dev：](https:\u002F\u002Fgroups.google.com\u002Fforum\u002F#!forum\u002Fknet-dev)\n  如果您希望参与Knet的开发，请加入此邮件列表，并查看这些[开发技巧](https:\u002F\u002Fdenizyuret.github.io\u002FKnet.jl\u002Flatest\u002Finstall\u002F#Tips-for-developers-1)。\n* [knet-slack：](https:\u002F\u002Fjulialang.slack.com\u002Fmessages\u002FCDLKQ92P3\u002Fdetails)\n  Knet的Slack交流频道。\n* 相关工作：请查看[Flux](https:\u002F\u002Fgithub.com\u002FFLuxML)、[Mocha](https:\u002F\u002Fgithub.com\u002Fpluskid\u002FMocha.jl)、[JuliaML](https:\u002F\u002Fgithub.com\u002FJuliaML)、[JuliaDiff](https:\u002F\u002Fgithub.com\u002FJuliaDiff)、[JuliaGPU](https:\u002F\u002Fgithub.com\u002FJuliaGPU)、[JuliaOpt](https:\u002F\u002Fgithub.com\u002FJuliaOpt)等相关软件包。\n\n## 示例\n\n下面是一个简单的示例，我们从头开始使用15行代码和10秒的GPU计算，定义、训练并测试了用于[MNIST](http:\u002F\u002Fyann.lecun.com\u002Fexdb\u002Fmnist)手写数字识别数据集的[LeNet](http:\u002F\u002Fyann.lecun.com\u002Fexdb\u002Flenet)模型。\n\n```julia\n# 首次运行前安装相关包：using Pkg; pkg\"add Knet IterTools MLDatasets\"\nusing Knet, IterTools, MLDatasets\n\n# 定义卷积层：\nstruct Conv; w; b; end\nConv(w1,w2,nx,ny) = Conv(param(w1,w2,nx,ny), param0(1,1,ny,1))\n(c::Conv)(x) = relu.(pool(conv4(c.w, x) .+ c.b))\n\n# 定义全连接层：\nstruct Dense; w; b; f; end\nDense(i,o; f=identity) = Dense(param(o,i), param0(o), f)\n(d::Dense)(x) = d.f.(d.w * mat(x) .+ d.b)\n\n# 定义层的串联及损失函数：\nstruct Chain; layers; end\n(c::Chain)(x) = (for l in c.layers; x = l(x); end; x)\n(c::Chain)(x,y) = nll(c(x),y)\n\n# 加载MNIST数据：\nxtrn,ytrn = MNIST.traindata(Float32); ytrn[ytrn.==0] .= 10\nxtst,ytst = MNIST.testdata(Float32);  ytst[ytst.==0] .= 10\ndtrn = minibatch(xtrn, ytrn, 100; xsize = (28,28,1,:))\ndtst = minibatch(xtst, ytst, 100; xsize = (28,28,1,:))\n\n# 定义并训练LeNet（在GPU上约10秒，在CPU上约3分钟即可达到约99%的准确率）\nLeNet = Chain((Conv(5,5,1,20), Conv(5,5,20,50), Dense(800,500,f=relu), Dense(500,10)))\nprogress!(adam(LeNet, ncycle(dtrn,3)))\naccuracy(LeNet,data=dtst)\n```\n\n## 贡献\nKnet是一个开源项目，我们始终欢迎新的贡献：包括错误报告与修复、功能请求与实现、新的机器学习模型与算子、启发性的示例以及基准测试结果等。有关说明，请参阅[开发者提示](https:\u002F\u002Fdenizyuret.github.io\u002FKnet.jl\u002Flatest\u002Finstall\u002F#Tips-for-developers)。\n\n贡献者：坎·居梅利、卡洛·卢奇贝洛、埃杰·奥纳特、埃金·阿克于雷克、埃克雷姆·埃姆雷·尤尔达库尔、埃姆雷·于纳尔、埃姆雷·约尔楚、埃尼斯·贝尔克、埃雷奈·达亚尼克、伊尔克尔·凯森、凯·徐、梅里奇·梅利克·索夫塔、迈克·因内斯、奥努尔·库鲁、欧赞·阿尔坎·詹、厄默尔·克尔纳普、福克·阮、雷内·多纳、蒂姆·贝萨尔、张世伟。","# Knet.jl 快速上手指南\n\n## 环境准备\n\n- **系统要求**：Julia 1.0+（推荐 1.6+），支持 Linux\u002FmacOS\u002FWindows\n- **前置依赖**：已安装 Julia 环境。推荐使用国内镜像加速安装 Julia：\n  - 通过清华源下载安装包：[https:\u002F\u002Fmirrors.tuna.tsinghua.edu.cn\u002Fjulia\u002F](https:\u002F\u002Fmirrors.tuna.tsinghua.edu.cn\u002Fjulia\u002F)\n\n## 安装步骤\n\n1. 启动 Julia 终端\n2. 配置国内镜像（加速包安装）：\n   ```bash\n   export JULIA_PKG_SERVER=\"https:\u002F\u002Fmirrors.tuna.tsinghua.edu.cn\u002Fjulia\"\n   ```\n   （Windows 用户请在系统环境变量中设置 `JULIA_PKG_SERVER`）\n3. 安装 Knet 及依赖包：\n   ```julia\n   using Pkg\n   Pkg.add([\"Knet\", \"IterTools\", \"MLDatasets\"])\n   ```\n\n## 基本使用\n\n以下为 LeNet 模型在 MNIST 数据集上的最小化示例（15 行代码）：\n\n```julia\n# Install packages before first run: using Pkg; pkg\"add Knet IterTools MLDatasets\"\nusing Knet, IterTools, MLDatasets\n\n# Define convolutional layer:\nstruct Conv; w; b; end\nConv(w1,w2,nx,ny) = Conv(param(w1,w2,nx,ny), param0(1,1,ny,1))\n(c::Conv)(x) = relu.(pool(conv4(c.w, x) .+ c.b))\n\n# Define dense layer:\nstruct Dense; w; b; f; end\nDense(i,o; f=identity) = Dense(param(o,i), param0(o), f)\n(d::Dense)(x) = d.f.(d.w * mat(x) .+ d.b)\n\n# Define a chain of layers and a loss function:\nstruct Chain; layers; end\n(c::Chain)(x) = (for l in c.layers; x = l(x); end; x)\n(c::Chain)(x,y) = nll(c(x),y)\n\n# Load MNIST data:\nxtrn,ytrn = MNIST.traindata(Float32); ytrn[ytrn.==0] .= 10\nxtst,ytst = MNIST.testdata(Float32);  ytst[ytst.==0] .= 10\ndtrn = minibatch(xtrn, ytrn, 100; xsize = (28,28,1,:))\ndtst = minibatch(xtst, ytst, 100; xsize = (28,28,1,:))\n\n# Define and train LeNet (~10 secs on a GPU or ~3 mins on a CPU)\nLeNet = Chain((Conv(5,5,1,20), Conv(5,5,20,50), Dense(800,500,f=relu), Dense(500,10)))\nprogress!(adam(LeNet, ncycle(dtrn,3)))\naccuracy(LeNet,data=dtst)\n```\n\n> **注意**：GPU 环境下运行约 10 秒（准确率 ~99%），CPU 环境约需 3 分钟","某高校AI实验室的研究生小李，正为教育类APP开发实时手写数字识别模块，需快速迭代模型以适配不同用户手写风格。\n\n### 没有 Knet.jl 时\n- 环境配置繁琐：依赖Python TensorFlow GPU版，CUDA版本冲突导致安装失败率超50%，平均耗时2小时\u002F次。\n- 代码冗长易错：手动实现卷积层需30+行代码，模型定义易出错且维护困难。\n- 训练效率低下：10轮MNIST训练需15分钟，无法及时调整超参数。\n- 调试困难：梯度计算需手动编码，缺乏可视化工具辅助定位问题。\n- 生态割裂：与Julia数据处理库（如MLDatasets）集成不畅，数据预处理流程断层。\n\n### 使用 Knet.jl 后\n- 环境简化：Julia包管理器一键安装Knet.jl，GPU支持开箱即用，配置时间从2小时压缩至5分钟。\n- 代码精简高效：复用官方LeNet示例，15行代码完成模型定义，开发效率提升4倍。\n- 训练加速显著：GPU加速下10轮训练仅需10秒，支持实时参数调优。\n- 调试便捷直观：自动微分自动处理梯度，无需手动实现反向传播。\n- 生态无缝整合：直接衔接Julia数据流，数据加载到训练流程一气呵成。\n\nKnet.jl让深度学习开发从环境配置的泥潭中解放，真正聚焦于模型创新与快速实验。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fdenizyuret_Knet.jl_53cd2abf.png","denizyuret",null,"https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fdenizyuret_28060b95.jpg","https:\u002F\u002Fgithub.com\u002Fdenizyuret",[80,84,88,92,96],{"name":81,"color":82,"percentage":83},"Jupyter Notebook","#DA5B0B",51.7,{"name":85,"color":86,"percentage":87},"Julia","#a270ba",47.3,{"name":89,"color":90,"percentage":91},"Python","#3572A5",1,{"name":93,"color":94,"percentage":95},"Makefile","#427819",0.1,{"name":97,"color":98,"percentage":99},"Perl","#0298c3",0,1436,226,"2026-03-12T16:23:42","NOASSERTION","未说明",{"notes":106,"python":104,"dependencies":107},"Knet 是 Julia 框架，需 Julia 环境。首次运行需安装依赖包（Knet、IterTools、MLDatasets）。支持 GPU（可选），但未指定具体硬件要求。",[108,109],"IterTools","MLDatasets",[51,54,13],[112,113,114,115,116,117],"knet","deep-learning","julia","machine-learning","neural-networks","data-science","2026-03-27T02:49:30.150509","2026-04-06T07:14:48.673655",[121,126,131,136],{"id":122,"question_zh":123,"answer_zh":124,"source_url":125},5560,"KnetArray 更新后显示异常值（如 0.00707721），如何解决？","需要安装 gcc-4.7 和 g++-4.7，然后在 Knet\u002Fsrc 目录执行 `make clean; make`。例如：`sudo apt install gcc-4.7 g++-4.7`，再运行 `make clean; make`。","https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Fissues\u002F125",{"id":127,"question_zh":128,"answer_zh":129,"source_url":130},5561,"knet7 和 knet8 有什么区别？应该使用哪个版本？","knet7 会维护 bugfixes，knet8 是未来版本，专注于 GPU 性能改进。建议使用 knet8（如果可用）或 knet7 用于稳定版本。","https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Fissues\u002F22",{"id":132,"question_zh":133,"answer_zh":134,"source_url":135},5562,"Knet 是否支持 Julia 0.7？","Knet 正在支持 Julia 0.7，master 分支已修复兼容性问题。建议使用 master 分支或等待官方发布。","https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Fissues\u002F316",{"id":137,"question_zh":138,"answer_zh":139,"source_url":140},5563,"Knet 构建时提示 Julia 路径问题（假设 Julia 通过 `julia` 在路径中），如何解决？","升级到 Julia 6 或更高版本，问题已修复。无需额外操作。","https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Fissues\u002F141",[142,147,152,157,162,167,172,177,182,187,192,197,202,207,212,217,222,227,232,237],{"id":143,"version":144,"summary_zh":145,"released_at":146},105122,"v1.4.10","## Knet v1.4.10\n\n[Diff since v1.4.9](https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Fcompare\u002Fv1.4.9...v1.4.10)\n\n\n**Closed issues:**\n- type DataType has no field mutable (#672)\n- Quick Start tutorial notebook is broken (#673)\n\n**Merged pull requests:**\n- Allow loading on Apple M1 (#667) (@rened)\n- support specialfunctions 2.x (#669) (@stevengj)","2022-02-12T21:04:24",{"id":148,"version":149,"summary_zh":150,"released_at":151},105123,"v1.4.9","## Knet v1.4.9\r\n\r\n[Diff since v1.4.8](https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Fcompare\u002Fv1.4.8...v1.4.9)\r\n\r\n\r\n**Closed issues:**\r\n- Making a new CUDA 3 compatible release? (#662)\r\n- Knet 1.4.7: libknet8 library not found (#664)","2021-10-13T21:04:26",{"id":153,"version":154,"summary_zh":155,"released_at":156},105124,"v1.4.8","## Knet v1.4.8\n\n[Diff since v1.4.7](https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Fcompare\u002Fv1.4.7...v1.4.8)","2021-08-08T11:03:49",{"id":158,"version":159,"summary_zh":160,"released_at":161},105125,"v1.4.7","## Knet v1.4.7\r\n\r\n[Diff since v1.4.6](https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Fcompare\u002Fv1.4.6...v1.4.7)\r\n\r\n\r\n\r\n**Merged pull requests:**\r\n- [WIP] Knet.CUDNN submodule for CUDNN calls (#614) (@denizyuret)","2021-07-23T19:03:24",{"id":163,"version":164,"summary_zh":165,"released_at":166},105126,"v1.4.6","## Knet v1.4.6\n\n[Diff since v1.4.5](https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Fcompare\u002Fv1.4.5...v1.4.6)\n\n\n**Closed issues:**\n- Julia crashes when I run \"using Knet\" (#652)\n- Knet.@save is broken (#653)\n- MethodError: no method matching iterate(::Nothing) while using params()  (#654)\n- MethodError: no method matching LinearIndices (#658)\n\n**Merged pull requests:**\n- fix #658 LinearIndices(::KnetArray) (#659) (@denizyuret)","2021-03-17T08:04:07",{"id":168,"version":169,"summary_zh":170,"released_at":171},105127,"v1.4.5","## Knet v1.4.5\r\n\r\n[Diff since v1.4.4](https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Fcompare\u002Fv1.4.4...v1.4.5)\r\n\r\n\r\n**Closed issues:**\r\n- error in gcnode (KeyError: key ... not found) (#624)\r\n- CUDNN errors in LSTM (#629)\r\n- ERROR: LoadError: ArgumentError: indexed assignment with a single value to many locations is not supported; perhaps use broadcasting `.=` instead? (#631)\r\n- test failed in cuarray.jl  (#634)\r\n- LeNet and MNIST data (#640)\r\n- deconv and unpool output size (#644)\r\n- unpool with mode=1 (#645)\r\n\r\n**Merged pull requests:**\r\n- gcnode debug version (#641) (@denizyuret)","2020-12-15T21:03:22",{"id":173,"version":174,"summary_zh":175,"released_at":176},105128,"v1.4.4","## Knet v1.4.4\n\n[Diff since v1.4.3](https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Fcompare\u002Fv1.4.3...v1.4.4)\n\n\n**Closed issues:**\n- Knet build error (#531)\n- Make CuArrays work with Knet.load\u002Fsave. (#587)\n- param(KnetArray(rand(2))) makes the ERROR: UndefVarError: array_type not defined (#622)\n- MLDatasets uses MNIST label 0:9 instead of 1:10 as in older Knet (#623)\n- Error : unsupported symbol trnLin passed to label (#627)\n- Knet.seed! failing to work on first function call (#628)\n- Error calculating cat gradient on GPU (#632)\n- usedmem test is failing (#637)\n- karray.jl test fails with new CUDA.jl (#638)\n\n**Merged pull requests:**\n- add powerpc support (#625) (@denizyuret)\n- Fixed a bug in Knet.setseed and Knet.seed that causes the first call … (#633) (@egeonat)\n- CuArray.ptr => CuArray.baseptr or pointer(CuArray) in CUDA-2.1.0 (#636) (@denizyuret)","2020-11-28T19:02:37",{"id":178,"version":179,"summary_zh":180,"released_at":181},105129,"v1.4.3","## Knet v1.4.3\r\n\r\n[Diff since v1.4.2](https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Fcompare\u002Fv1.4.2...v1.4.3)","2020-10-16T10:03:48",{"id":183,"version":184,"summary_zh":185,"released_at":186},105130,"v1.4.2","## Knet v1.4.2\n\n[Diff since v1.4.1](https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Fcompare\u002Fv1.4.1...v1.4.2)\n\n\n**Closed issues:**\n- Using conv and pool without CUDA (#33)\n- Knet GPU OOM with recent CUDA\u002FKnet updates (#602)\n-  ERROR: LoadError: MethodError: no method matching unsafe_free!(::Nothing) (#610)\n- libknet8 library not found (#611)\n- Building Knet fails in 1.5.1 (#615)\n- LeNet model for MNIST  Example (#617)\n- New GPUArrays indexing causes scalar indexing for some Knet operations (i.e., cat) (#618)\n- Error converting CuArray to KnetArray (#619)\n- `k[1:2:3,:] .= 0` broken for KnetArray (#620)\n- tutorial\u002F30.lin.ipynb (#621)\n\n**Merged pull requests:**\n- Fix #610: check if object is CuArray before calling unsafe_free! (#612) (@denizyuret)","2020-09-28T16:03:41",{"id":188,"version":189,"summary_zh":190,"released_at":191},105131,"v1.4.1","## Knet v1.4.1\n\n[Diff since v1.4.0](https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Fcompare\u002Fv1.4.0...v1.4.1)\n\n\n**Closed issues:**\n- Seg fault after upgrading to Julia 1.5 (#589)\n- CPU computation (Switching off GPU computation)\u002F@diff of Array{} values returns 0-element Array (#594)\n- RNN fails in a fresh installation of the latest Knet release (#597)\n- Knet.randn! not defined in v1.4.0 (#598)\n- gcnode is not working properly, debug and add tests (#606)\n\n**Merged pull requests:**\n- Fix #606 gcnode issue (#609) (@denizyuret)","2020-08-28T13:03:09",{"id":193,"version":194,"summary_zh":195,"released_at":196},105132,"v1.4.0","## Knet v1.4.0\r\n\r\n[Diff since v1.3.9](https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Fcompare\u002Fv1.3.9...v1.4.0)\r\n\r\n\r\n**Closed issues:**\r\n- Use GPUArray \u002F KnetArray (#150)\r\n- Illegal memory access with CuArrays (#251)\r\n- cat 4D arrays along channel dimension (#319)\r\n- Concatenation  in 3 dimension missing  (#400)\r\n- Feature request: group convolutions (#407)\r\n- StackOverflowError when broadcast between number and KnetArray{Bool} (#502)\r\n- Tutorial 15. quickstart fail (Knet 1.2.7, Julia 1.1.1, Windows 10)   (#508)\r\n- [bug when broadcasting KnetArray] UndefVarError: lib not defined (#534)\r\n- Use CuArrays, eliminate dependency on nvcc by implementing gpu kernels in julia (#550)\r\n- Is it possible to provide Knet GPU support in BinaryBuilder (#566)\r\n- cudnn 8.0 compatibility (#569)\r\n- Get rid of @cu calls, replace them with CUDA calls (#580)\r\n- Create Knet.Ops20 and add benchmarking for operator set (#583)\r\n- Knet GPU selection and auto device selection (#586)\r\n- Create a profiling suite (#588)\r\n- Missing website (#592)\r\n- CUDA.pool not defined (#595)\r\n\r\n**Merged pull requests:**\r\n- Ops20 (#596) (@denizyuret)","2020-08-19T11:02:54",{"id":198,"version":199,"summary_zh":200,"released_at":201},105133,"v1.3.9","## Knet v1.3.9\n\n[Diff since v1.3.8](https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Fcompare\u002Fv1.3.8...v1.3.9)\n\n\n**Closed issues:**\n- CUDAapi doesn't seem to have find_host_compiler (#567)\n- Warning: Hardware is unsupported by NNPACK so falling back to default NNlib (#576)\n- addtoindex! ambiguity (#578)\n- no method matching length error at model optimization (#579)","2020-07-28T09:02:22",{"id":203,"version":204,"summary_zh":205,"released_at":206},105134,"v1.3.8","## Knet v1.3.8\r\n\r\n[Diff since v1.3.7](https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Fcompare\u002Fv1.3.7...v1.3.8)","2020-07-23T22:02:22",{"id":208,"version":209,"summary_zh":210,"released_at":211},105135,"v1.3.7","## Knet v1.3.7\n\n[Diff since v1.3.6](https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Fcompare\u002Fv1.3.6...v1.3.7)\n\n\n**Closed issues:**\n- problems on test Knet and find_host_compiler (#530)\n- CuArrays.jl: DEPRECATED, use CUDA.jl instead! (#570)\n- travis does not push new docs (#571)\n- Feature request: Depthwise and groupwise convolution (#574)","2020-07-12T12:03:01",{"id":213,"version":214,"summary_zh":215,"released_at":216},105136,"v1.3.6","## Knet v1.3.6\n\n[Diff since v1.3.5](https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Fcompare\u002Fv1.3.5...v1.3.6)\n\n\n**Closed issues:**\n- SpecialFunctions loggamma update (#486)\n- test with CUDAapi v4.0 (#543)\n- Memory allocation and deallocation of temporary variables (#548)\n- Error when testing Knet (#553)\n- Broadcasted copyto! .(+\u002F*)= operator is undefined for AutoGrad (#554)\n- Bug: Loading a model made using the Chain struct fails (#557)\n- collect now needs size, collect(progress()) fails (#558)\n- How to cite Knet in a research? (#559)\n- Loading Knet breaks Julia's copyto! on 1.5 (#561)\n- Performance degradation (#562)\n- NLL loss (#563)\n- Installing and `using Knet` doesn't work (#565)\n- GPU tests suggested in the documentation failing  (#568)\n\n**Merged pull requests:**\n- add gpu install docs for azure\u002Fubuntu18.04 (#546) (@jw3126)\n- GPU discoverability with CUDAapi v4.0 (#549) (@iuliancioarca)\n- Further notes on GPU tools, especially for windows (#560) (@RocketRoss)","2020-07-04T20:02:19",{"id":218,"version":219,"summary_zh":220,"released_at":221},105137,"v1.3.5","## Knet v1.3.5\n\n[Diff since v1.3.4](https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Fcompare\u002Fv1.3.4...v1.3.5)\n\n\n**Closed issues:**\n- KnetArray error : expected UnionAll, got typeof(Knet.CuArray) (#540)\n- test with CUDAapi v4.0 (#543)\n- Reading from a KNetArray within a GPU kernel function (#544)\n- Knet.save error in julia 1.4 (#545)\n\n**Merged pull requests:**\n- respect to CUDA_VISIBLE_DEVICES (#541) (@ekinakyurek)","2020-03-30T06:01:18",{"id":223,"version":224,"summary_zh":225,"released_at":226},105138,"v1.3.4","## Knet v1.3.4\n\n[Diff since v1.3.3](https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Fcompare\u002Fv1.3.3...v1.3.4)\n\n\n**Closed issues:**\n- AssertionError: vec(value(x)) isa WTYPE (#537)\n- Cannot `convert` an object of type AutoGrad.Result{Float32} to an object of type Float64 (#538)\n\n**Merged pull requests:**\n- Install TagBot as a GitHub Action (#539) (@JuliaTagBot)","2020-02-29T09:02:06",{"id":228,"version":229,"summary_zh":230,"released_at":231},105139,"v1.3.3","## [v1.3.3](https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Ftree\u002Fv1.3.3) (2020-02-01)\n\n[Diff since v1.3.2](https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Fcompare\u002Fv1.3.2...v1.3.3)\n\n**Closed issues:**\n\n- Convert KnetArray to CuArray and back (#535)\n- aws images do not work (#533)\n- Feature Request: Compatibility with CuArrays 1.6.0 (#532)\n- julia1.3.0\u002Fknet132 build conf.jl gradcheck error (#528)\n- Test error on test\u002Fserialize.jl and broken tests on test\u002Fconv.jl (#527)\n- the error of test for Knet in dropout.jl (#524)\n- memory usage printings (#523)\n\n**Merged pull requests:**\n\n- fixed primitive for bmm, added more gradient tests for bmm (#526) ([ekinakyurek](https:\u002F\u002Fgithub.com\u002Fekinakyurek))\n- A numerical example for Knet.nll for the documentation (#525) ([Alexander-Barth](https:\u002F\u002Fgithub.com\u002FAlexander-Barth))","2020-02-01T18:20:56",{"id":233,"version":234,"summary_zh":235,"released_at":236},105140,"v1.3.2","## [v1.3.2](https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Ftree\u002Fv1.3.2) (2019-11-29)\n\n[Diff since v1.3.1](https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Fcompare\u002Fv1.3.1...v1.3.2)\n\n**Closed issues:**\n\n- conv4 allocates outside of the memory pool  (#518)\n- scaling xavier intializer comparared to other frameworks  (#515)\n\n**Merged pull requests:**\n\n- Alexander barth master (#522) ([denizyuret](https:\u002F\u002Fgithub.com\u002Fdenizyuret))\n- fix issue \\#518: conv4 allocates outside of the memory pool \\#518 (#521) ([denizyuret](https:\u002F\u002Fgithub.com\u002Fdenizyuret))\n-  xavier intializer using the same convention as TensorFlow and PyTorch (#520) ([Alexander-Barth](https:\u002F\u002Fgithub.com\u002FAlexander-Barth))","2019-11-29T11:48:12",{"id":238,"version":239,"summary_zh":240,"released_at":241},105141,"v1.3.1","## [v1.3.1](https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Ftree\u002Fv1.3.1) (2019-11-07)\n\n[Diff since v1.3.0](https:\u002F\u002Fgithub.com\u002Fdenizyuret\u002FKnet.jl\u002Fcompare\u002Fv1.3.0...v1.3.1)\n\n**Merged pull requests:**\n\n- Dy\u002Fcuarrays13 (#516) ([denizyuret](https:\u002F\u002Fgithub.com\u002Fdenizyuret))","2019-11-07T07:19:31"]