[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"similar-sanderland--katrain":3,"tool-sanderland--katrain":61},[4,18,26,36,44,53],{"id":5,"name":6,"github_repo":7,"description_zh":8,"stars":9,"difficulty_score":10,"last_commit_at":11,"category_tags":12,"status":17},4358,"openclaw","openclaw\u002Fopenclaw","OpenClaw 是一款专为个人打造的本地化 AI 助手，旨在让你在自己的设备上拥有完全可控的智能伙伴。它打破了传统 AI 助手局限于特定网页或应用的束缚，能够直接接入你日常使用的各类通讯渠道，包括微信、WhatsApp、Telegram、Discord、iMessage 等数十种平台。无论你在哪个聊天软件中发送消息，OpenClaw 都能即时响应，甚至支持在 macOS、iOS 和 Android 设备上进行语音交互，并提供实时的画布渲染功能供你操控。\n\n这款工具主要解决了用户对数据隐私、响应速度以及“始终在线”体验的需求。通过将 AI 部署在本地，用户无需依赖云端服务即可享受快速、私密的智能辅助，真正实现了“你的数据，你做主”。其独特的技术亮点在于强大的网关架构，将控制平面与核心助手分离，确保跨平台通信的流畅性与扩展性。\n\nOpenClaw 非常适合希望构建个性化工作流的技术爱好者、开发者，以及注重隐私保护且不愿被单一生态绑定的普通用户。只要具备基础的终端操作能力（支持 macOS、Linux 及 Windows WSL2），即可通过简单的命令行引导完成部署。如果你渴望拥有一个懂你",349277,3,"2026-04-06T06:32:30",[13,14,15,16],"Agent","开发框架","图像","数据工具","ready",{"id":19,"name":20,"github_repo":21,"description_zh":22,"stars":23,"difficulty_score":10,"last_commit_at":24,"category_tags":25,"status":17},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,"2026-04-05T11:01:52",[14,15,13],{"id":27,"name":28,"github_repo":29,"description_zh":30,"stars":31,"difficulty_score":32,"last_commit_at":33,"category_tags":34,"status":17},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",150720,2,"2026-04-11T11:33:10",[14,13,35],"语言模型",{"id":37,"name":38,"github_repo":39,"description_zh":40,"stars":41,"difficulty_score":32,"last_commit_at":42,"category_tags":43,"status":17},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。",108322,"2026-04-10T11:39:34",[14,15,13],{"id":45,"name":46,"github_repo":47,"description_zh":48,"stars":49,"difficulty_score":32,"last_commit_at":50,"category_tags":51,"status":17},6121,"gemini-cli","google-gemini\u002Fgemini-cli","gemini-cli 是一款由谷歌推出的开源 AI 命令行工具，它将强大的 Gemini 大模型能力直接集成到用户的终端环境中。对于习惯在命令行工作的开发者而言，它提供了一条从输入提示词到获取模型响应的最短路径，无需切换窗口即可享受智能辅助。\n\n这款工具主要解决了开发过程中频繁上下文切换的痛点，让用户能在熟悉的终端界面内直接完成代码理解、生成、调试以及自动化运维任务。无论是查询大型代码库、根据草图生成应用，还是执行复杂的 Git 操作，gemini-cli 都能通过自然语言指令高效处理。\n\n它特别适合广大软件工程师、DevOps 人员及技术研究人员使用。其核心亮点包括支持高达 100 万 token 的超长上下文窗口，具备出色的逻辑推理能力；内置 Google 搜索、文件操作及 Shell 命令执行等实用工具；更独特的是，它支持 MCP（模型上下文协议），允许用户灵活扩展自定义集成，连接如图像生成等外部能力。此外，个人谷歌账号即可享受免费的额度支持，且项目基于 Apache 2.0 协议完全开源，是提升终端工作效率的理想助手。",100752,"2026-04-10T01:20:03",[52,13,15,14],"插件",{"id":54,"name":55,"github_repo":56,"description_zh":57,"stars":58,"difficulty_score":32,"last_commit_at":59,"category_tags":60,"status":17},4721,"markitdown","microsoft\u002Fmarkitdown","MarkItDown 是一款由微软 AutoGen 团队打造的轻量级 Python 工具，专为将各类文件高效转换为 Markdown 格式而设计。它支持 PDF、Word、Excel、PPT、图片（含 OCR）、音频（含语音转录）、HTML 乃至 YouTube 链接等多种格式的解析，能够精准提取文档中的标题、列表、表格和链接等关键结构信息。\n\n在人工智能应用日益普及的今天，大语言模型（LLM）虽擅长处理文本，却难以直接读取复杂的二进制办公文档。MarkItDown 恰好解决了这一痛点，它将非结构化或半结构化的文件转化为模型“原生理解”且 Token 效率极高的 Markdown 格式，成为连接本地文件与 AI 分析 pipeline 的理想桥梁。此外，它还提供了 MCP（模型上下文协议）服务器，可无缝集成到 Claude Desktop 等 LLM 应用中。\n\n这款工具特别适合开发者、数据科学家及 AI 研究人员使用，尤其是那些需要构建文档检索增强生成（RAG）系统、进行批量文本分析或希望让 AI 助手直接“阅读”本地文件的用户。虽然生成的内容也具备一定可读性，但其核心优势在于为机器",93400,"2026-04-06T19:52:38",[52,14],{"id":62,"github_repo":63,"name":64,"description_en":65,"description_zh":66,"ai_summary_zh":66,"readme_en":67,"readme_zh":68,"quickstart_zh":69,"use_case_zh":70,"hero_image_url":71,"owner_login":72,"owner_name":73,"owner_avatar_url":74,"owner_bio":75,"owner_company":76,"owner_location":77,"owner_email":76,"owner_twitter":78,"owner_website":76,"owner_url":79,"languages":80,"stars":89,"forks":90,"last_commit_at":91,"license":92,"difficulty_score":10,"env_os":93,"env_gpu":94,"env_ram":95,"env_deps":96,"category_tags":102,"github_topics":103,"view_count":32,"oss_zip_url":76,"oss_zip_packed_at":76,"status":17,"created_at":112,"updated_at":113,"faqs":114,"releases":145},6606,"sanderland\u002Fkatrain","katrain","Improve your Baduk skills by training with KataGo!","KaTrain 是一款专为围棋爱好者设计的智能辅助工具，旨在通过集成强大的 KataGo 人工智能引擎，帮助用户高效提升棋艺。它主要解决了传统围棋复盘困难、难以精准定位失误以及缺乏高质量陪练的痛点。\n\n在功能上，KaTrain 不仅能深度分析用户的对局记录，自动找出导致目数损失最大的“恶手”并生成重点复盘报告，还提供了即时反馈的对弈模式。用户在与 AI 对战时，一旦走出错招即可收到提示并尝试重走，从而在实战中纠正错误直觉。此外，软件支持调节 AI 强度与风格，允许用户与不同水平的弱化版 AI 进行多样化对抗，避免了直接面对顶级 AI 时因差距过大而无法学习的困境。\n\nKaTrain 非常适合从入门初学者到资深业余段位的广大围棋玩家使用。其独特的技术亮点在于将复杂的 KataGo 算法封装为直观易用的图形界面，无需用户具备编程背景即可轻松配置和使用。无论是想要系统复盘检讨的棋手，还是希望获得个性化陪练的爱好者，KaTrain 都能成为你棋盘旁得力的智能教练，让围棋学习过程更加科学且充满乐趣。","# \u003Ca name=\"manual\">\u003C\u002Fa> KaTrain\n\n[![Latest Release](http:\u002F\u002Fimg.shields.io\u002Fgithub\u002Frelease\u002Fsanderland\u002Fkatrain?label=download)](http:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Freleases)\n[![License:MIT](http:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fl\u002Fkatrain)](http:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FMIT_License)\n[![GitHub Downloads](http:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fdownloads\u002Fsanderland\u002Fkatrain\u002Ftotal?color=%23336699&label=github%20downloads)](http:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Freleases)\n[![PyPI Downloads](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsanderland_katrain_readme_b8e8058919a0.png)](http:\u002F\u002Fpepy.tech\u002Fproject\u002Fkatrain)\n[![Discord](http:\u002F\u002Fimg.shields.io\u002Fdiscord\u002F417022162348802048?logo=discord)](http:\u002F\u002Fdiscord.com\u002Fchannels\u002F417022162348802048\u002F629446365688365067)\n\nKaTrain is a tool for analyzing games and playing go with AI feedback from KataGo:\n\n* Review your games to find the moves that were most costly in terms of points lost.\n* Play against AI and get immediate feedback on mistakes with option to retry.\n* Play against a wide range of weakened versions of AI with various styles.\n* Automatically generate focused SGF reviews which show your biggest mistakes.\n\n## Manual\n\n\u003Ctable>\n\u003Ctd>\n\n- [ KaTrain](#-katrain)\n - [Manual](#manual)\n - [ Preview and Youtube Videos](#--preview-and-youtube-videos)\n - [ Installation](#-installation)\n - [ Configuring KataGo](#--configuring-katago)\n - [ Play against AI](#-play-against-ai)\n - [Instant feedback](#instant-feedback)\n - [AIs](#ais)\n - [ Analysis](#-analysis)\n - [ Keyboard and mouse shortcuts](#-keyboard-and-mouse-shortcuts)\n - [ Contributing to distributed training](#-contributing-to-distributed-training)\n - [ Themes](#-themes)\n - [ FAQ](#-faq)\n - [ Support \u002F Contribute](#-support--contribute)\n\n\n\u003Ctd>\n\n\u003Ca href=\"http:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fblob\u002Fmaster\u002FREADME.md\">\u003Cimg alt=\"English\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsanderland_katrain_readme_6f382b2ddfdd.png\" width=50>\u003C\u002Fa>\n\u003Ca href=\"http:\u002F\u002Ftranslate.google.com\u002Ftranslate?sl=en&tl=de&u=https%3A%2F%2Fgithub.com%2Fsanderland%2Fkatrain%2Fblob%2Fmaster%2FREADME.md\">\u003Cimg alt=\"German\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsanderland_katrain_readme_cc23f204eaad.png\" width=50>\u003C\u002Fa>\n\u003Ca href=\"http:\u002F\u002Ftranslate.google.com\u002Ftranslate?sl=en&tl=fr&u=https%3A%2F%2Fgithub.com%2Fsanderland%2Fkatrain%2Fblob%2Fmaster%2FREADME.md\">\u003Cimg alt=\"French\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsanderland_katrain_readme_78e9f99bc3c9.png\" width=50>\u003C\u002Fa>\n\u003Ca href=\"http:\u002F\u002Ftranslate.google.com\u002Ftranslate?sl=en&tl=uk&u=https%3A%2F%2Fgithub.com%2Fsanderland%2Fkatrain%2Fblob%2Fmaster%2FREADME.md\">\u003Cimg alt=\"Ukrainian\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsanderland_katrain_readme_95722e6bdc17.png\" width=50>\u003C\u002Fa>\n\u003Ca href=\"http:\u002F\u002Ftranslate.google.com\u002Ftranslate?sl=en&tl=ru&u=https%3A%2F%2Fgithub.com%2Fsanderland%2Fkatrain%2Fblob%2Fmaster%2FREADME.md\">\u003Cimg alt=\"Russian\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsanderland_katrain_readme_b36b3e38b06f.png\" width=50>\u003C\u002Fa>\n\u003Cbr\u002F>\n\u003Ca href=\"http:\u002F\u002Ftranslate.google.com\u002Ftranslate?sl=en&tl=tr&u=https%3A%2F%2Fgithub.com%2Fsanderland%2Fkatrain%2Fblob%2Fmaster%2FREADME.md\">\u003Cimg alt=\"Turkish\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsanderland_katrain_readme_7d0331378c1d.png\" width=50>\u003C\u002Fa>\n\u003Ca href=\"http:\u002F\u002Ftranslate.google.com\u002Ftranslate?sl=en&tl=zh-CN&u=https%3A%2F%2Fgithub.com%2Fsanderland%2Fkatrain%2Fblob%2Fmaster%2FREADME.md\">\u003Cimg alt=\"Simplified Chinese\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsanderland_katrain_readme_00d7b6d94371.png\" width=50>\u003C\u002Fa>\n\u003Ca href=\"http:\u002F\u002Ftranslate.google.com\u002Ftranslate?sl=en&tl=zh-TW&u=https%3A%2F%2Fgithub.com%2Fsanderland%2Fkatrain%2Fblob%2Fmaster%2FREADME.md\">\u003Cimg alt=\"Traditional Chinese\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsanderland_katrain_readme_3c30dfe3d3b4.png\" width=50>\u003C\u002Fa>\n\u003Ca href=\"http:\u002F\u002Ftranslate.google.com\u002Ftranslate?sl=en&tl=ko&u=https%3A%2F%2Fgithub.com%2Fsanderland%2Fkatrain%2Fblob%2Fmaster%2FREADME.md\">\u003Cimg alt=\"Korean\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsanderland_katrain_readme_b026a63067e5.png\" width=50>\u003C\u002Fa>\n\u003Ca href=\"http:\u002F\u002Ftranslate.google.com\u002Ftranslate?sl=en&tl=ja&u=https%3A%2F%2Fgithub.com%2Fsanderland%2Fkatrain%2Fblob%2Fmaster%2FREADME.md\">\u003Cimg alt=\"Japanese\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsanderland_katrain_readme_8e69fa7dddb9.png\" width=50>\u003C\u002Fa>\n\n\u003C\u002Ftd>\n\u003C\u002Ftable>\n\n## \u003Ca name=\"preview\">\u003C\u002Fa> Preview and Youtube Videos\n\n\u003Cimg alt=\"screenshot\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsanderland_katrain_readme_8c761c420663.png\" width=\"550\">\n\n| **Local Joseki Analysis** | **Analysis Tutorial** | **Teaching Game Tutorial** |\n|:-----------------------------------------------------------------------------------------------------:|:-----------------------------------------------------------------------------------------------------:|:------------------------------------------------------------------------------------------------------------:|\n| [![Local Joseki Analysis Video](http:\u002F\u002Fi.imgur.com\u002FYcpmSBx.png)](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=tXniX57KtKk) | [![Analysis Tutorial](http:\u002F\u002Fi.imgur.com\u002F3EP4IEr.png)](http:\u002F\u002Fwww.youtube.com\u002Fwatch?v=qjxkcKgrsbU) | [![ Teaching Game Tutorial](http:\u002F\u002Fi.imgur.com\u002FjAdcSL5.png)](http:\u002F\u002Fwww.youtube.com\u002Fwatch?v=wFl4Bab_eGM) |\n\n\n\n## \u003Ca name=\"install\">\u003C\u002Fa> Installation\n* See the [releases page](http:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Freleases) for downloadable executables for Windows and macOS.\n* Alternatively use `pipx install katrain` to install the latest version from PyPI on any 64-bit OS in an isolated environment.\n* On macOS, you can also use `brew install katrain` to install the app.\n* [This page](https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fblob\u002Fmaster\u002FINSTALL.md) has detailed instructions for Window, Linux and macOS,\n as well as troubleshooting and setting up KataGo to use multiple GPUs.\n\n## \u003Ca name=\"kata\">\u003C\u002Fa> Configuring KataGo\n\nKaTrain comes pre-packaged with a working KataGo (OpenCL version) for Windows, Linux, and pre-M1 Mac operating systems, and the rather old 15 block model.\n\nTo change the model, open 'General and Engine settings' in the application and 'Download models'. You can then select the model you want from the dropdown menu.\n\nTo change the katago binary, e.g. to the Eigen\u002FCPU version if you don't have a GPU, click 'Download KataGo versions'.\n You can then select the KataGo binary from the dropdown menu.\nThere are also CUDA and TensorRT versions available on [the KataGo release site](https:\u002F\u002Fgithub.com\u002Flightvector\u002FKataGo\u002Freleases). Particularly the latter may offer much better performance on NVIDIA GPUs, but will be harder to \nset up: [see here for more details](https:\u002F\u002Fgithub.com\u002Flightvector\u002FKataGo#opencl-vs-cuda-vs-tensorrt-vs-eigen).\n\nFinally, you can override the entire command used to start the analysis engine, which \n can be useful for connecting to a remote server. Do keep in mind that KaTrain uses the *analysis engine*\n of KataGo, and not the GTP engine.\n\n\n## \u003Ca name=\"ai\">\u003C\u002Fa> Play against AI\n\n* Select the players in the main menu, or under 'New Game'.\n* In a teaching game, KaTrain will analyze your moves and automatically undo those that are sufficiently bad.\n* When playing against AI, note that the \"Undo\" button will undo both the AI's last move and yours.\n\n### Instant feedback\n\nThe dots on the move indicate how many points were lost by that move.\n\n* The colour indicates the size of the mistake according to KataGo\n* The size indicates if the mistake was actually punished. Going from fully punished at maximal size,\n to no actual effect on the score at minimal size.\n\nIn short, if you are a weaker player you should mostly focus on large dots that are red or purple,\nwhile stronger players can pay more attention to smaller mistakes. If you want to hide some colours\non the board, or not output details for them in SGFs,you can do so under 'Configure Teacher'.\n\n### AIs\n\nThis section describes the available AIs.\n\nIn the 'AI settings', settings which have been tested and calibrated are at the top and have a lighter color,\nchanging these will show an estimate of rank.\nThis estimate should be reasonably accurate as long as you have not changed the other settings.\n\n* Recommended options for serious play include:\n * **KataGo** is full KataGo, above professional level. The analysis and feedback given is always based on this full strength KataGo AI.\n * **Calibrated Rank Bot** was calibrated on various bots (e.g. GnuGo and Pachi at different strength settings) to play a balanced\n game from the opening to the endgame without making serious (DDK) blunders. Further discussion can be found\n [here](http:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fissues\u002F44) and [here](http:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fissues\u002F74).\n * **Simple Style** Prefers moves that solidify both player's territory, leading to relatively simpler moves.\n* Legacy options which were developed earlier include: \n * **ScoreLoss** is KataGo analyzing as usual, but\n choosing from potential moves depending on the expected score loss, leading to a varied style with mostly small mistakes.\n * **Policy** uses the top move from the policy network (it's 'shape sense' without reading).\n * **Policy Weighted** picks a random move weighted by the policy, leading to a varied style with mostly small mistakes, and occasional blunders due to a lack of reading.\n * **Blinded Policy** picks a number of moves at random and play the best move among them, being effectively 'blind' to part of the board each turn. Calibrated rank is based on the same idea, and recommended over this option.\n* Options that are more on the 'fun and experimental' side include: \n * Variants of **Blinded Policy**, which use the same basic strategy, but with a twist:\n * **Local Style** will consider mostly moves close to the last move.\n * **Tenuki Style** will consider mostly moves away from the last move.\n * **Influential Style** will consider mostly 4th+ line moves, leading to a center-oriented style.\n * **Territory Style** is biased in the opposite way, towards 1-3rd line moves.\n * **KataJigo** is KataGo attempting to win by 0.5 points, typically by responding to your mistakes with an immediate mistake of it's own.\n * **KataAntiMirror** is KataGo assuming you are playing mirror go and attempting to break out of it with profit as long as you are.\n \nThe Engine based AIs (KataGo, ScoreLoss, KataJigo) are affected by both the model and choice of visits and maximum time,\n while the policy net based AIs are affected by the choice of model file, but work identically with 1 visit.\n\nFurther technical details and discussion on some of these AIs can be found on [this](http:\u002F\u002Flifein19x19.com\u002Fviewtopic.php?f=10&t=17488&sid=b11e42c005bb6f4f48c83771e6a27eff) thread at the life in 19x19 forums.\n\n## \u003Ca name=\"analysis\">\u003C\u002Fa> Analysis\n\nAnalysis options in KaTrain allow you to explore variations and request more in-depth analysis from the engine at any point in the game.\n\n| Key | Short Description | Details |\n| -------------- | -------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |\n| \u003Ckbd>Tab\u003C\u002Fkbd> | Switch between analysis and play modes | AI moves, teaching mode and timers are suspended in analysis mode. The state of the analysis options and right-hand side panels and options is saved independently for 'play' and 'analyze', allowing you to quickly switch between a more minimalistic 'play' mode and more complex 'analysis' mode. |\n\nThe checkboxes at the top of the screen:\n\n| Key | Short Description | Details |\n| ------------ | --------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |\n| \u003Ckbd>q\u003C\u002Fkbd> | Child moves are shown | On by default, can turn it off to avoid obscuring other information or when wanting to guess the next move. |\n| \u003Ckbd>w\u003C\u002Fkbd> | Show all dots | Toggles showing coloured evaluation 'dots' on the last few moves or not. You can configure the thresholds, along with how many of the last moves they are shown for under 'Teaching\u002FAnalysis Settings'. |\n| \u003Ckbd>e\u003C\u002Fkbd> | Top moves | Show the next moves KataGo considered, colored by their expected point loss. Small\u002Ffaint dots indicate high uncertainty and never show text (lower than your 'fast visits' setting). Hover over any of them to see the principal variation. |\n| \u003Ckbd>r\u003C\u002Fkbd> | Policy moves | Show KataGo's policy network evaluation, i.e. where it thinks the best next move is purely from the position, and in the absence of any 'reading'. This turns off the 'top moves' setting as the overlap is often not useful. |\n| \u003Ckbd>t\u003C\u002Fkbd> | Expected territory | Show expected ownership of each intersection. |\n\nThe analysis options available under the 'Analysis' button are used for deeper evaluation of the position:\n\n| Key | Short Description | Details |\n| ----------------------------------- | ------------------------------------------------------------------------------------------------------------ | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |\n| \u003Ckbd>a\u003C\u002Fkbd> | Deeper analysis | Re-evaluate the position using more visits, usually resulting in a more accurate evaluation. |\n| \u003Ckbd>s\u003C\u002Fkbd> | Equalize visits | Re-evaluate all currently shown next moves with the same visits as the current top move. Useful to increase confidence in the suggestions with high uncertainty. |\n| \u003Ckbd>d\u003C\u002Fkbd> | Analyze all moves | Evaluate all possible next moves. This can take a bit of time even though 'fast_visits' is used, but can be useful to see how many reasonable next moves are available. |\n| \u003Ckbd>f\u003C\u002Fkbd> | Find alternatives | Increases analysis of current candidate moves to at least the 'fast visits' level, and request a new query that excludes all current candidate moves. |\n| \u003Ckbd>g\u003C\u002Fkbd> | Select area of interest | Set an area and search only for moves in this box. Good for solving tsumegos. Note that some results may appear outside the box due to establishing a baseline for the best move, and the opponent can tenuki in variations. |\n| \u003Ckbd>h\u003C\u002Fkbd> | Reset analysis | This reverts the analysis to what the engine returns after a normal query, removing any additional exploration. |\n| \u003Ckbd>i\u003C\u002Fkbd> | Start insertion mode | Allows you to insert moves, to improve analysis when both players ignore an important exchange or life and death situation. Press again to stop inserting and copy the rest of the branch. |\n| \u003Ckbd>l\u003C\u002Fkbd> | Play out the game until the end and add as a collapsed branch, to visualize the potential effect of mistakes | This is done in the background, and can be started at several nodes at once when comparing the results at different starting positions. |\n| \u003Ckbd>Space\u003C\u002Fkbd> | Turn continuous analysis on\u002Foff. | This will continuously improve analysis of the current position, similar to Lizzie's 'pondering', but only when there are no other queries going on. |\n| \u003Ckbd>Shift\u003C\u002Fkbd> + \u003Ckbd>Space\u003C\u002Fkbd> | As above, but does not turn 'top moves' hints on when it is off. | |\n| \u003Ckbd>Enter\u003C\u002Fkbd> | AI move | Makes the AI move for the current player regardless of current player selection. |\n| \u003Ckbd>F2\u003C\u002Fkbd> | Deeper full game analysis | Analyze the entire game to a higher number of visits. |\n| \u003Ckbd>F3\u003C\u002Fkbd> | Performance report | Show an overview of performance statistics for both players. |\n| \u003Ckbd>F10\u003C\u002Fkbd> | Tsumego Frame | After placing a life and death problem in a corner\u002Fside, use this to fill up the rest of the board to improve AI's ability in solving life and death problems. |\n\n## \u003Ca name=\"keyboard\">\u003C\u002Fa> Keyboard and mouse shortcuts\n\nIn addition to shortcuts mentioned above and those shown in the main menu:\n\n| Key | Short Description | Details |\n| ---------------------------------------------- | ------------------------------------------------------------------------------------ | --------------------------------------------------------------------------------------------------------------------------------------- |\n| \u003Ckbd>Alt\u003C\u002Fkbd> | Open the main menu | |\n| \u003Ckbd>~\u003C\u002Fkbd> or \u003Ckbd>`\u003C\u002Fkbd> or \u003Ckbd>F12\u003C\u002Fkbd> | Cycles through more minimalistic UI modes | |\n| \u003Ckbd>k\u003C\u002Fkbd> | Toggle display of board coordinates | |\n| \u003Ckbd>p\u003C\u002Fkbd> | Pass | |\n| \u003Ckbd>Pause\u003C\u002Fkbd> | Pause\u002FResume timer | |\n| \u003Ckbd>←\u003C\u002Fkbd> or \u003Ckbd>z\u003C\u002Fkbd> | Undo move | Hold shift for 10 moves at a time, or ctrl to skip to the start. |\n| \u003Ckbd>→\u003C\u002Fkbd> or \u003Ckbd>x\u003C\u002Fkbd> | Redo move | Hold shift for 10 moves at a time, or ctrl to skip to the end. |\n| \u003Ckbd>↑\u003C\u002Fkbd>\u002F\u003Ckbd>↓\u003C\u002Fkbd> | Switch branch | As would be expected from the move tree. |\n| \u003Ckbd>Home\u003C\u002Fkbd>\u002F\u003Ckbd>End\u003C\u002Fkbd> | Go to the beginning\u002Fend of the game | |\n| \u003Ckbd>PageUp\u003C\u002Fkbd> | Make the currently selected node the main branch | |\n| \u003Ckbd>Ctrl\u003C\u002Fkbd> + \u003Ckbd>Delete\u003C\u002Fkbd> | Delete current node | |\n| \u003Ckbd>c\u003C\u002Fkbd> | Collapse\u002FUncollapse the branch from the current node to the previous branching point | |\n| \u003Ckbd>b\u003C\u002Fkbd> | Go back to the previous branching point | |\n| \u003Ckbd>Shift\u003C\u002Fkbd> + \u003Ckbd>b\u003C\u002Fkbd> | Go back the main branch | |\n| \u003Ckbd>n\u003C\u002Fkbd> | Go to one move before the next mistake (orange or worse) by a human player | As in clicking the forward red arrow |\n| \u003Ckbd>Shift\u003C\u002Fkbd> + \u003Ckbd>n\u003C\u002Fkbd> | Go to one move before the previous mistake | As in clicking the backward red arrow |\n| Scroll Mouse | Redo\u002FUndo move or Scroll through principal variation | When hovering the cursor over the right panel: Redo\u002FUndo move. When hovering over a candidate move: Scroll through principal variation. |\n| Middle Scroll Wheel Click | Add principal variation to the move tree | When scrolling, only moves up to the point you are viewing are added. |\n| Click on a Move | See detailed statistics for a previous move | Along with expected variation that was best instead of this move |\n| Double Click on a Move | Navigate directly to just before that point in the game | |\n| \u003Ckbd>Ctrl\u003C\u002Fkbd> + \u003Ckbd>v\u003C\u002Fkbd> | Load SGF from the clipboard and do a 'fast' analysis of the game | With a high priority normal analysis for the last move. |\n| \u003Ckbd>Ctrl\u003C\u002Fkbd> + \u003Ckbd>c\u003C\u002Fkbd> | Save SGF to clipboard | |\n| \u003Ckbd>Escape\u003C\u002Fkbd> | Stop all analysis | |\n\n## \u003Ca name=\"distributed\">\u003C\u002Fa> Contributing to distributed training\n\nStarting in December 2020, KataGo started [distributed training](https:\u002F\u002Fkatagotraining.org\u002F).\nThis allows people to all help generate self-play games to increase KataGo's strength and train bigger models.\n\nKaTrain 1.8.0+ makes it easy to contribute to distributed training: simply select the option from the main menu, register an account, and click run.\nDuring this mode you can do little more than watch games.\n\nKeep in mind that partial games are not uploaded,\nso it is best to plan to keep it running for at least an hour, if not several, for the most effective contribution.\n\nA few keyboard shortcuts have special functions in this mode:\n\n| Key | Short Description | Details |\n| ----------------- | ------------------------------------------------------------------ | ------------------------------------------------------------------------------------------------------ |\n| \u003Ckbd>Space\u003C\u002Fkbd> | Switch between manually navigating the current game | And automatically advancing it. |\n| \u003Ckbd>Escape\u003C\u002Fkbd> | Sends the `quit` command to KataGo | Which starts a slow shutdown, finishing partial games but not starting new ones. Only works on v1.11+. |\n| \u003Ckbd>Pause\u003C\u002Fkbd> | Pauses\u002Fresumes contributions via the `pause` and `resume` commands | Introduced in KataGo v1.11 |\n\n## \u003Ca name=\"themes\">\u003C\u002Fa> Themes\n\nSee [these instructions](THEMES.md) for how to modify the look of any graphics or colours, and creating or install themes.\n \n## \u003Ca name=\"faq\">\u003C\u002Fa> FAQ\n\n* The program is running too slowly. How can I speed it up?\n * Adjust the number of visits or maximum time allowed in the settings.\n* KataGo crashes with \"out of memory\" errors, how can I prevent this?\n * Try using a lower number for `nnMaxBatchSize` in `KataGo\u002Fanalysis_config.cfg`, and avoid using versions compiled with large board sizes.\n * If still encountering problems, please start KataGo by itself to check for any errors it gives.\n * Note that if you don't have a GPU, or your GPU does not support OpenCL, you should use the 'eigen' binaries which run on CPU only.\n* The font size is too small\n * On some ultra-high resolution monitors, dialogs and other elements with text can appear too small. Please see [these](https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fissues\u002F359#issuecomment-784096271) instructions to adjust them.\n* The app crashes with an error about \"unable to find any valuable cutbuffer provider\"\n * Install xclip using `sudo apt-get install xclip`\n\n\n## \u003Ca name=\"support\">\u003C\u002Fa> Support \u002F Contribute\n\n[![GitHub issues](http:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fissues\u002Fsanderland\u002Fkatrain)](http:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fissues)\n[![Contributors](http:\u002F\u002Fimg.shields.io\u002Fstatic\u002Fv1?label=contributors&message=\u003C3&color=dcb424)](CONTRIBUTIONS.md)\n\n * Ideas, feedback, and contributions to code or translations are all very welcome.\n * For suggestions and planned improvements, see [open issues](http:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fissues) on github to check if the functionality is already planned.\n* You can join the [Computer Go Community Discord (formerly Leela Zero & Friends)](http:\u002F\u002Fdiscord.gg\u002FAjTPFpN) (use the #gui channel) to get help, discuss improvements, or simply show your appreciation. Please do not use github issues to ask for technical help, this is only for bugs, suggestions and discussing contributions.\n\n\n\n","# \u003Ca name=\"manual\">\u003C\u002Fa> KaTrain\n\n[![最新版本](http:\u002F\u002Fimg.shields.io\u002Fgithub\u002Frelease\u002Fsanderland\u002Fkatrain?label=download)](http:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Freleases)\n[![许可证:MIT](http:\u002F\u002Fimg.shields.io\u002Fpypi\u002Fl\u002Fkatrain)](http:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FMIT_License)\n[![GitHub 下载量](http:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fdownloads\u002Fsanderland\u002Fkatrain\u002Ftotal?color=%23336699&label=github%20downloads)](http:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Freleases)\n[![PyPI 下载量](https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsanderland_katrain_readme_b8e8058919a0.png)](http:\u002F\u002Fpepy.tech\u002Fproject\u002Fkatrain)\n[![Discord](http:\u002F\u002Fimg.shields.io\u002Fdiscord\u002F417022162348802048?logo=discord)](http:\u002F\u002Fdiscord.com\u002Fchannels\u002F417022162348802048\u002F629446365688365067)\n\nKaTrain 是一款用于分析棋局并结合 KataGo 的 AI 反馈进行围棋对弈的工具：\n\n* 回顾你的对局，找出那些导致最多目数损失的着法。\n* 与 AI 对弈，并即时获得错误提示及重试选项。\n* 与多种风格、不同强度的弱化版 AI 对战。\n* 自动生成功能聚焦的 SGF 回放文件，突出显示你最大的失误。\n\n## 使用手册\n\n\u003Ctable>\n\u003Ctd>\n\n- [ KaTrain](#-katrain)\n - [使用手册](#manual)\n - [ 预览与 YouTube 视频](#--preview-and-youtube-videos)\n - [ 安装](#-installation)\n - [ 配置 KataGo](#--configuring-katago)\n - [ 与 AI 对弈](#-play-against-ai)\n - [即时反馈](#instant-feedback)\n - [AI 选择](#ais)\n - [ 分析](#-analysis)\n - [ 键盘和鼠标快捷键](#-keyboard-and-mouse-shortcuts)\n - [ 参与分布式训练](#-contributing-to-distributed-training)\n - [ 主题](#-themes)\n - [ 常见问题解答](#-faq)\n - [ 支持与贡献](#-support--contribute)\n\n\n\u003Ctd>\n\n\u003Ca href=\"http:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fblob\u002Fmaster\u002FREADME.md\">\u003Cimg alt=\"英语\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsanderland_katrain_readme_6f382b2ddfdd.png\" width=50>\u003C\u002Fa>\n\u003Ca href=\"http:\u002F\u002Ftranslate.google.com\u002Ftranslate?sl=en&tl=de&u=https%3A%2F%2Fgithub.com%2Fsanderland%2Fkatrain%2Fblob%2Fmaster%2FREADME.md\">\u003Cimg alt=\"德语\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsanderland_katrain_readme_cc23f204eaad.png\" width=50>\u003C\u002Fa>\n\u003Ca href=\"http:\u002F\u002Ftranslate.google.com\u002Ftranslate?sl=en&tl=fr&u=https%3A%2F%2Fgithub.com%2Fsanderland%2Fkatrain%2Fblob%2Fmaster%2FREADME.md\">\u003Cimg alt=\"法语\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsanderland_katrain_readme_78e9f99bc3c9.png\" width=50>\u003C\u002Fa>\n\u003Ca href=\"http:\u002F\u002Ftranslate.google.com\u002Ftranslate?sl=en&tl=uk&u=https%3A%2F%2Fgithub.com%2Fsanderland%2Fkatrain%2Fblob%2Fmaster%2FREADME.md\">\u003Cimg alt=\"乌克兰语\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsanderland_katrain_readme_95722e6bdc17.png\" width=50>\u003C\u002Fa>\n\u003Ca href=\"http:\u002F\u002Ftranslate.google.com\u002Ftranslate?sl=en&tl=ru&u=https%3A%2F%2Fgithub.com%2Fsanderland%2Fkatrain%2Fblob%2Fmaster%2FREADME.md\">\u003Cimg alt=\"俄语\" src=\"https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fblob\u002Fmaster\u002Fkatain img\u002Fflags\u002Fflag-ru.png\" width=50>\u003C\u002Fa>\n\u003Cbr\u002F>\n\u003Ca href=\"http:\u002F\u002Ftranslate.google.com\u002Ftranslate?sl=en&tl=tr&u=https%3A%2F%2Fgithub.com%2Fsanderland%2Fkatrain%2Fblob%2Fmaster%2FREADME.md\">\u003Cimg alt=\"土耳其语\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsanderland_katrain_readme_7d0331378c1d.png\" width=50>\u003C\u002Fa>\n\u003Ca href=\"http:\u002F\u002Ftranslate.google.com\u002Ftranslate?sl=en&tl=zh-CN&u=https%3A%2F%2Fgithub.com%2Fsanderland%2Fkatrain%2Fblob%2Fmaster%2FREADME.md\">\u003Cimg alt=\"简体中文\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsanderland_katrain_readme_00d7b6d94371.png\" width=50>\u003C\u002Fa>\n\u003Ca href=\"http:\u002F\u002Ftranslate.google.com\u002Ftranslate?sl=en&tl=zh-TW&u=https%3A%2F%2Fgithub.com%2Fsanderland%2Fkatrain%2Fblob%2Fmaster%2FREADME.md\">\u003Cimg alt=\"繁体中文\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsanderland_katrain_readme_3c30dfe3d3b4.png\" width=50>\u003C\u002Fa>\n\u003Ca href=\"http:\u002F\u002Ftranslate.google.com\u002Ftranslate?sl=en&tl=ko&u=https%3A%2F%2Fgithub.com%2Fsanderland%2Fkatrain%2Fblob%2Fmaster%2FREADME.md\">\u003Cimg alt=\"韩语\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsanderland_katrain_readme_b026a63067e5.png\" width=50>\u003C\u002Fa>\n\u003Ca href=\"http:\u002F\u002Ftranslate.google.com\u002Ftranslate?sl=en&tl=ja&u=https%3A%2F%2Fgithub.com%2Fsanderland%2Fkatrain%2Fblob%2Fmaster%2FREADME.md\">\u003Cimg alt=\"日语\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsanderland_katrain_readme_8e69fa7dddb9.png\" width=50>\u003C\u002Fa>\n\n\u003C\u002Ftd>\n\u003C\u002Ftable>\n\n## \u003Ca name=\"preview\">\u003C\u002Fa> 预览与 YouTube 视频\n\n\u003Cimg alt=\"截图\" src=\"https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsanderland_katrain_readme_8c761c420663.png\" width=\"550\">\n\n| **局部定式分析** | **分析教程** | **教学棋谱教程** |\n|:-----------------------------------------------------------------------------------------------------:|:-----------------------------------------------------------------------------------------------------:|:------------------------------------------------------------------------------------------------------------:|\n| [![局部定式分析视频](http:\u002F\u002Fi.imgur.com\u002FYcpmSBx.png)](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=tXniX57KtKk) | [![分析教程](http:\u002F\u002Fi.imgur.com\u002F3EP4IEr.png)](http:\u002F\u002Fwww.youtube.com\u002Fwatch?v=qjxkcKgrsbU) | [![教学棋谱教程](http:\u002F\u002Fi.imgur.com\u002FjAdcSL5.png)](http:\u002F\u002Fwww.youtube.com\u002Fwatch?v=wFl4Bab_eGM) |\n\n\n\n## \u003Ca name=\"install\">\u003C\u002Fa> 安装\n* 请访问 [发布页面](http:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Freleases)，下载适用于 Windows 和 macOS 的可执行文件。\n* 或者，您也可以使用 `pipx install katrain` 在任何 64 位操作系统上以隔离环境安装 PyPI 上的最新版本。\n* 在 macOS 上，您还可以通过 `brew install katrain` 来安装该应用。\n* [此页面](https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fblob\u002Fmaster\u002FINSTALL.md) 提供了针对 Windows、Linux 和 macOS 的详细安装说明，\n 以及故障排除指南和如何配置 KataGo 以使用多块 GPU 的方法。\n\n## \u003Ca name=\"kata\">\u003C\u002Fa> 配置 KataGo\n\nKaTrain 已预装适用于 Windows、Linux 和 M1 之前的 Mac 操作系统的可用 KataGo（OpenCL 版本），以及一个相当老旧的 15 块模型。\n\n要更换模型，只需在应用程序中打开“常规与引擎设置”，然后选择“下载模型”。之后，您就可以从下拉菜单中选择所需的模型。\n\n若要更改 KataGo 的二进制文件，例如切换到 Eigen\u002FCPU 版本以应对没有 GPU 的情况，请点击“下载 KataGo 版本”。随后，您可以在下拉菜单中选择相应的 KataGo 二进制文件。\n\n此外，在 [KataGo 发布页面](https:\u002F\u002Fgithub.com\u002Flightvector\u002FKataGo\u002Freleases)上还提供了 CUDA 和 TensorRT 版本。尤其是后者，在 NVIDIA GPU 上可能带来更出色的性能，但其配置难度也更高：[详情请参阅此处](https:\u002F\u002Fgithub.com\u002Flightvector\u002FKataGo#opencl-vs-cuda-vs-tensorrt-vs-eigen)。\n\n最后，您还可以覆盖用于启动分析引擎的完整命令行，这在连接远程服务器时非常有用。请注意，KaTrain 使用的是 KataGo 的 *分析引擎*，而非 GTP 引擎。\n\n\n## \u003Ca name=\"ai\">\u003C\u002Fa> 与 AI 对弈\n\n* 在主菜单或“新游戏”中选择对弈双方。\n* 在教学对局中，KaTrain 会分析您的每一步棋，并自动撤销那些明显不佳的着法。\n* 与 AI 对弈时，请注意“撤销”按钮会同时撤销 AI 和您的上一步棋。\n\n### 即时反馈\n\n棋盘上的圆点表示该步棋导致的目数损失：\n\n* 颜色代表 KataGo 判断的失误程度；\n* 圆点大小则反映实际被惩罚的程度——从最大尺寸表示完全被惩罚，到最小尺寸表示对局势无实质性影响。\n\n简而言之，对于水平较低的棋手，应重点关注红色或紫色的大圆点；而高段位棋手则可更多留意较小的失误。如果您希望隐藏棋盘上的某些颜色，或不在 SGF 文件中输出相关细节，可在“教师设置”中进行调整。\n\n### AI 介绍\n\n本节将介绍可供选择的 AI。\n\n在“AI 设置”中，经过测试和校准的选项位于顶部且背景较浅，更改这些设置后会显示一个段位估算值。只要未修改其他设置，这一估算通常较为准确。\n\n* 适合认真对弈的推荐选项包括：\n * **KataGo** 是完整的 KataGo 算法，强度超过职业水平。其分析与反馈始终基于此全强度版本的 KataGo AI。\n * **校准段位机器人** 曾通过与多种机器人（如不同强度的 GnuGo 和 Pachi）对弈进行校准，旨在从布局到收官阶段保持均衡的棋风，避免出现重大失误（DDK）。更多讨论请参见 [这里](http:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fissues\u002F44) 和 [这里](http:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fissues\u002F74)。\n * **简单风格** 更倾向于巩固双方各自地盘的着法，因此棋局相对简洁。\n* 较早开发的遗留选项包括：\n * **失分优先** 是 KataGo 以常规方式分析局面，但根据预期失分多少来选择下一步棋，从而形成风格多变、以小失误为主的棋风。\n * **策略网络** 直接采用策略网络给出的最优着法（即不经过读秒的“棋感”着法）。\n * **策略加权** 则根据策略网络的概率分布随机选择一步棋，同样以小失误为主，偶尔也会因缺乏深入计算而出现低级失误。\n * **盲选策略** 会随机挑选若干着法，再从中选出最佳的一手，相当于每回合都对棋盘的一部分视而不见。校准段位正是基于类似思路设计的，因此建议优先使用该选项。\n* 属于“趣味性与实验性”的选项包括：\n * **盲选策略** 的几种变体，它们沿用相同的基本策略，但各有侧重：\n * **局部风格** 主要考虑靠近上一手的着法。\n * **脱先风格** 则倾向于远离上一手的着法。\n * **影响力风格** 多选择四线及以上的着法，形成偏向中央的棋风。\n * **地盘风格** 则相反，偏好一二三线的着法。\n * **KataJigo** 是 KataGo 试图以半目优势取胜的模式，通常会在您犯错时立即回敬一个类似的失误。\n * **KataAntiMirror** 是 KataGo 假设您正在下镜像围棋，并尝试在您持续如此的情况下打破僵局、争取获利。\n\n基于引擎的 AI（KataGo、失分优先、KataJigo）会受到模型选择以及搜索次数和时限设置的影响；而基于策略网络的 AI 则仅受模型文件选择的影响，无论搜索次数为多少，表现均一致。\n\n关于部分 AI 的更多技术细节及讨论，可参见 life in 19x19 论坛上的这篇帖子：[链接](http:\u002F\u002Flifein19x19.com\u002Fviewtopic.php?f=10&t=17488&sid=b11e42c005bb6f4f48c83771e6a27eff)。\n\n\n## \u003Ca name=\"analysis\">\u003C\u002Fa> 分析\n\nKaTrain 中的分析功能允许您探索变化分支，并在对局的任何阶段向引擎请求更深入的分析。\n\n| 快捷键 | 简要说明 | 详细信息 |\n| -------------- | -------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |\n| \u003Ckbd>Tab\u003C\u002Fkbd> | 切换分析与对弈模式 | 在分析模式下，AI 走子、教学模式和计时器等功能均会被暂停。分析选项以及右侧面板和功能的状态会分别保存为“对弈”和“分析”两种模式，方便您快速在简约的“对弈”模式与复杂的“分析”模式之间切换。\n\n屏幕顶部的复选框：\n\n| 键 | 简要说明 | 详情 |\n| ---------- | ---------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |\n| \u003Ckbd>q\u003C\u002Fkbd> | 显示子着 | 默认开启，可关闭以避免遮挡其他信息，或在希望自行猜测下一步时使用。 |\n| \u003Ckbd>w\u003C\u002Fkbd> | 显示所有点 | 切换是否在最后几步棋上显示彩色的评估“点”。您可以在“教学\u002F分析设置”中配置这些阈值，以及针对最后多少步棋显示这些点。 |\n| \u003Ckbd>e\u003C\u002Fkbd> | 最佳着 | 显示 KataGo 认为的下一步着法，按其预期的目数损失程度进行着色。较小或较淡的点表示不确定性较高，且不会显示文字（低于您的“快速访问”设置）。将鼠标悬停在任意一个点上即可查看主变化线。 |\n| \u003Ckbd>r\u003C\u002Fkbd> | 策略着 | 显示 KataGo 的策略网络评估结果，即仅根据当前局面、在不进行任何“读秒”的情况下，它认为的最佳下一步落子位置。此功能会关闭“最佳着”设置，因为两者经常存在重叠，意义不大。 |\n| \u003Ckbd>t\u003C\u002Fkbd> | 预期地盘 | 显示每个交叉点的预期归属情况。 |\n\n“分析”按钮下的分析选项用于对当前局面进行更深入的评估：\n\n| 键 | 简要说明 | 详细说明 |\n| ----------------------------------- | ------------------------------------------------------------------------------------------------------------ | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |\n| \u003Ckbd>a\u003C\u002Fkbd> | 更深入分析 | 使用更多访问次数重新评估局面，通常会得到更准确的评估结果。 |\n| \u003Ckbd>s\u003C\u002Fkbd> | 统一访问次数 | 将当前显示的所有下一步着法，都重新评估到与当前最佳着法相同的访问次数。这有助于提高对不确定性较高的建议的信心。 |\n| \u003Ckbd>d\u003C\u002Fkbd> | 分析所有着法 | 评估所有可能的下一步着法。即使使用“快速访问”模式，这仍然可能需要一些时间，但可以帮助你了解有多少合理的下一步着法可供选择。 |\n| \u003Ckbd>f\u003C\u002Fkbd> | 寻找替代方案 | 将当前候选着法的分析深度至少提升到“快速访问”级别，并请求一个新的查询，排除所有当前的候选着法。 |\n| \u003Ckbd>g\u003C\u002Fkbd> | 选择兴趣区域 | 设置一个区域，只搜索该区域内的着法。非常适合解死活题。需要注意的是，由于需要确定最佳着法的基准，且对手在变化中可能会脱先，因此部分结果可能会出现在所选区域之外。 |\n| \u003Ckbd>h\u003C\u002Fkbd> | 重置分析 | 这会将分析恢复到引擎正常查询后返回的状态，移除任何额外的探索内容。 |\n| \u003Ckbd>i\u003C\u002Fkbd> | 开始插入模式 | 允许你插入着法，以改进分析，特别是在双方都忽视了重要的交换或生死局面时。再次按下即可停止插入，并复制剩余的分支。 |\n| \u003Ckbd>l\u003C\u002Fkbd> | 将棋局一直下到终局，并以折叠分支的形式添加，以便可视化失误可能带来的影响 | 这会在后台进行，也可以同时在多个节点上启动，以便比较不同起始位置的结果。 |\n| \u003Ckbd>空格\u003C\u002Fkbd> | 开启或关闭连续分析。 | 这将不断改进当前局面的分析，类似于 Lizzie 的“思考”功能，但仅在没有其他查询进行时才会生效。 |\n| \u003Ckbd>Shift\u003C\u002Fkbd> + \u003Ckbd>空格\u003C\u002Fkbd> | 同上，但在“最佳着法”提示关闭时不会自动开启。 | |\n| \u003Ckbd>Enter\u003C\u002Fkbd> | AI 走子 | 不论当前轮到哪一方，都会让 AI 为当前方走一步棋。 |\n| \u003Ckbd>F2\u003C\u002Fkbd> | 更深入的全盘分析 | 将整盘棋分析到更高的访问次数。 |\n| \u003Ckbd>F3\u003C\u002Fkbd> | 性能报告 | 显示双方的性能统计概览。 |\n| \u003Ckbd>F10\u003C\u002Fkbd> | 死活题框架 | 在角落或边上放置好死活题后，使用此功能填充棋盘的其余部分，以提高 AI 解决死活问题的能力。 |\n\n\n\n## \u003Ca name=\"keyboard\">\u003C\u002Fa> 键盘和鼠标快捷键\n\n除了上述提到的快捷键以及主菜单中显示的那些之外：\n\n| 键位 | 简要说明 | 详细说明 |\n| ---------------------------------------------- | ------------------------------------------------------------------------------------ | --------------------------------------------------------------------------------------------------------------------------------------- |\n| \u003Ckbd>Alt\u003C\u002Fkbd> | 打开主菜单 | |\n| \u003Ckbd>~\u003C\u002Fkbd> 或 \u003Ckbd>`\u003C\u002Fkbd> 或 \u003Ckbd>F12\u003C\u002Fkbd> | 循环切换更简约的界面模式 | |\n| \u003Ckbd>k\u003C\u002Fkbd> | 切换棋盘坐标显示 | |\n| \u003Ckbd>p\u003C\u002Fkbd> | 过子 | |\n| \u003Ckbd>Pause\u003C\u002Fkbd> | 暂停\u002F继续计时 | |\n| \u003Ckbd>←\u003C\u002Fkbd> 或 \u003Ckbd>z\u003C\u002Fkbd> | 撤销一步 | 按住 Shift 键可一次撤销 10 步，按 Ctrl 键则跳转到对局开始处。 |\n| \u003Ckbd>→\u003C\u002Fkbd> 或 \u003Ckbd>x\u003C\u002Fkbd> | 恢复一步 | 按住 Shift 键可一次恢复 10 步，按 Ctrl 键则跳转到对局结束处。 |\n| \u003Ckbd>↑\u003C\u002Fkbd>\u002F\u003Ckbd>↓\u003C\u002Fkbd> | 切换分支 | 行棋树中的正常操作。 |\n| \u003Ckbd>Home\u003C\u002Fkbd>\u002F\u003Ckbd>End\u003C\u002Fkbd> | 跳转到对局的开头\u002F结尾 | |\n| \u003Ckbd>PageUp\u003C\u002Fkbd> | 将当前选中的节点设为主分支 | |\n| \u003Ckbd>Ctrl\u003C\u002Fkbd> + \u003Ckbd>Delete\u003C\u002Fkbd> | 删除当前节点 | |\n| \u003Ckbd>c\u003C\u002Fkbd> | 折叠\u002F展开从当前节点到上一个分岔点的分支 | |\n| \u003Ckbd>b\u003C\u002Fkbd> | 回到上一个分岔点 | |\n| \u003Ckbd>Shift\u003C\u002Fkbd> + \u003Ckbd>b\u003C\u002Fkbd> | 回到主分支 | |\n| \u003Ckbd>n\u003C\u002Fkbd> | 跳转到人类棋手下出的下一个失误（橙色或更差）之前的一手 | 类似于点击向前的红色箭头 |\n| \u003Ckbd>Shift\u003C\u002Fkbd> + \u003Ckbd>n\u003C\u002Fkbd> | 跳转到人类棋手下出的上一个失误之前的一手 | 类似于点击向后的红色箭头 |\n| 鼠标滚轮 | 恢复\u002F撤销步数，或滚动浏览主要变例 | 当光标悬停在右侧面板时：恢复\u002F撤销步数。当光标悬停在候选步上时：滚动浏览主要变例。 |\n| 中键单击 | 将主要变例添加到行棋树 | 滚动时，仅会添加你正在查看位置之前的步数。 |\n| 点击某一步 | 查看该步的详细统计数据 | 包括原本的最佳变例，而非当前这一步 |\n| 双击某一步 | 直接跳转到该步之前的位置 | |\n| \u003Ckbd>Ctrl\u003C\u002Fkbd> + \u003Ckbd>v\u003C\u002Fkbd> | 从剪贴板加载 SGF 文件，并对对局进行“快速”分析 | 对最后一步会进行高优先级的常规分析。 |\n| \u003Ckbd>Ctrl\u003C\u002Fkbd> + \u003Ckbd>c\u003C\u002Fkbd> | 将 SGF 保存到剪贴板 | |\n| \u003Ckbd>Escape\u003C\u002Fkbd> | 停止所有分析 | |\n\n## \u003Ca name=\"distributed\">\u003C\u002Fa> 参与分布式训练\n\n自2020年12月起，KataGo启动了[分布式训练](https:\u002F\u002Fkatagotraining.org\u002F)功能。这使得广大用户能够共同生成自我对弈棋局，从而提升KataGo的棋力并训练更大规模的模型。\n\nKaTrain 1.8.0及以上版本让参与分布式训练变得非常简单：只需在主菜单中选择相应选项，注册一个账号，然后点击“运行”即可。在此模式下，您主要可以做的就是观看对局。\n\n请注意，未完成的对局不会上传，因此为了更有效地贡献算力，建议至少持续运行一小时以上，最好能运行数小时。\n\n在该模式下，一些键盘快捷键具有特殊功能：\n\n| 键 | 简要说明 | 详细信息 |\n| ----------------- | ------------------------------------------------------------------ | ------------------------------------------------------------------------------------------------------ |\n| \u003Ckbd>空格\u003C\u002Fkbd> | 在手动浏览当前对局和自动推进之间切换 | 并自动推进它。 |\n| \u003Ckbd>Esc\u003C\u002Fkbd> | 向KataGo发送`quit`命令 | 这将启动缓慢的关闭过程，完成正在进行中的对局但不再开始新的对局。仅适用于v1.11及以上版本。 |\n| \u003Ckbd>Pause\u003C\u002Fkbd> | 通过`pause`和`resume`命令暂停\u002F恢复贡献 | KataGo v1.11版本引入的功能 |\n\n## \u003Ca name=\"themes\">\u003C\u002Fa> 主题\n\n有关如何修改界面外观、颜色，以及创建或安装主题，请参阅[这些说明](THEMES.md)。\n\n## \u003Ca name=\"faq\">\u003C\u002Fa> 常见问题解答\n\n* 程序运行得太慢了，该如何加快速度？\n * 可以在设置中调整访问次数或最大允许时间。\n* KataGo 出现“内存不足”的错误而崩溃，该如何避免？\n * 尝试在`KataGo\u002Fanalysis_config.cfg`文件中降低`nnMaxBatchSize`的数值，并避免使用编译时设置了大尺寸棋盘的版本。\n * 如果仍然遇到问题，请单独启动KataGo，查看其是否报错。\n * 请注意，如果您没有GPU，或者您的GPU不支持OpenCL，则应使用仅能在CPU上运行的“eigen”二进制文件。\n* 字体太小\n * 在某些超高分辨率显示器上，对话框和其他包含文本的元素可能会显得过小。请参考[这些说明](https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fissues\u002F359#issuecomment-784096271)进行调整。\n* 应用程序因“无法找到任何有价值的cutbuffer提供者”错误而崩溃\n * 使用`sudo apt-get install xclip`安装xclip工具。\n\n## \u003Ca name=\"support\">\u003C\u002Fa> 支持与贡献\n\n[![GitHub 问题](http:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fissues\u002Fsanderland\u002Fkatrain)](http:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fissues)\n[![贡献者](http:\u002F\u002Fimg.shields.io\u002Fstatic\u002Fv1?label=contributors&message=\u003C3&color=dcb424)](CONTRIBUTIONS.md)\n\n * 欢迎提出想法、反馈，以及对代码或翻译的贡献。\n * 对于改进建议和计划中的功能，请先查看GitHub上的[开放问题](http:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fissues)，确认相关功能是否已在规划中。\n* 您可以加入[计算机围棋社区Discord（原Leela Zero & Friends）](http:\u002F\u002Fdiscord.gg\u002FAjTPFpN)（使用#gui频道），获取帮助、讨论改进方案，或只是表达您的支持。请勿使用GitHub的问题来寻求技术支持，此处仅用于报告Bug、提出建议及讨论贡献事宜。","# KaTrain 快速上手指南\n\nKaTrain 是一款基于 KataGo 引擎的围棋 AI 辅助工具，支持对弈复盘、实时错误反馈及生成重点失误报告。\n\n## 环境准备\n\n* **操作系统**：Windows, macOS (Intel\u002FM1), Linux (64 位)。\n* **硬件要求**：\n * **GPU 加速（推荐）**：支持 OpenCL、CUDA 或 TensorRT 的显卡可获得最佳性能。\n * **CPU 运行**：若无独立显卡，可使用 CPU 版本（速度较慢）。\n* **前置依赖**：\n * 若使用预编译包：无需额外依赖。\n * 若使用 Python 安装：需安装 Python 3.8+ 及 `pipx`（推荐用于隔离环境）。\n\n## 安装步骤\n\n### 方法一：下载预编译包（推荐新手）\n访问 [GitHub Releases 页面](https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Freleases) 下载对应系统的可执行文件：\n* **Windows**: 下载 `.exe` 安装包。\n* **macOS**: 下载 `.dmg` 或 `.app` 文件。\n* **Linux**: 下载对应的二进制包。\n\n### 方法二：使用包管理器安装\n**macOS (Homebrew):**\n```bash\nbrew install katrain\n```\n\n**任意 64 位系统 (Python pipx):**\n建议使用 `pipx` 在隔离环境中安装，避免依赖冲突：\n```bash\npipx install katrain\n```\n\n### 方法三：源码安装与多 GPU 配置\n如需详细配置多 GPU 或解决特定系统问题，请参考官方 [INSTALL.md](https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fblob\u002Fmaster\u002FINSTALL.md)。\n\n## 基本使用\n\n### 1. 配置 KataGo 引擎\n启动 KaTrain 后，默认已内置基础版 KataGo (OpenCL 版本 + 15 block 模型)。建议根据硬件优化配置：\n* 进入菜单 **General and Engine settings**。\n* 点击 **Download models**：从下拉菜单选择更强的权重模型（如较新的 Block 数模型）。\n* 点击 **Download KataGo versions**：\n * 无 GPU 用户：选择 `Eigen\u002FCPU` 版本。\n * NVIDIA 显卡用户：建议手动下载并配置 `CUDA` 或 `TensorRT` 版本以获得更高性能（需在 KataGo 发布页获取二进制文件）。\n\n### 2. 与 AI 对弈 (Play against AI)\n* 在主菜单或 **New Game** 中选择对手。\n* **教学模式 (Teaching Game)**：\n * 开启后，若你的着手亏损过大，AI 会自动提示并允许悔棋重试。\n * **实时反馈**：棋盘上的圆点代表该手的亏损目数。\n * **颜色**：表示错误严重程度（红色\u002F紫色为严重失误）。\n * **大小**：表示该错误是否被对手惩罚（大圆点表示已被惩罚，小圆点表示未影响胜负）。\n* **悔棋注意**：在对弈模式下点击 \"Undo\" 将同时撤销你和 AI 的最新一手。\n\n### 3. 棋局分析 (Analysis)\n* 按 \u003Ckbd>Tab\u003C\u002Fkbd> 键在“对弈模式”和“分析模式”间切换。\n* 在分析模式下，计时器和自动落子暂停，你可自由探索变化图。\n* 右键点击棋盘或使用右侧面板请求深度分析。\n* **生成报告**：分析结束后，可自动生成聚焦于重大失误的 SGF 复盘文件。\n\n### 4. 常用快捷键\n| 按键 | 功能描述 |\n| :--- | :--- |\n| \u003Ckbd>Tab\u003C\u002Fkbd> | 切换分析模式与对弈模式 |\n| 鼠标右键 | 请求当前局面深度分析 |\n| 鼠标滚轮 | 调整分支变化图 |\n\n> **提示**：KaTrain 使用的是 KataGo 的 **Analysis Engine** 而非 GTP 引擎，配置远程服务器时请注意命令差异。","围棋爱好者李明在业余段位瓶颈期，希望通过复盘和对弈快速提升棋力，却苦于缺乏高水平指导。\n\n### 没有 katrain 时\n- 复盘完全依赖个人感觉或棋友互评，难以精准定位导致目数大幅亏损的“致命一手”。\n- 想找特定风格的 AI 陪练时，只能面对实力过强且风格单一的顶级引擎，要么被碾压失去信心，要么无法模拟真实对手。\n- 对局中下出疑问手后无法即时知晓原因，往往要等到整局结束甚至几天后请教他人才能明白失误所在。\n- 生成复习材料耗时费力，需要手动记录坐标并整理成文档，难以形成系统化的错题集。\n\n### 使用 katrain 后\n- 利用 KataGo 引擎自动分析棋谱，软件直接高亮显示每一步的胜率波动和目数损失，让李明瞬间锁定最严重的失误点。\n- 内置多种弱化版 AI 和不同风格模型，李明可以选择与自己水平相当或风格克制的对手进行针对性特训。\n- 开启“即时反馈”模式对弈，一旦走出劣招，katrain 立即提示并允许原地重试，帮助他在肌肉记忆形成前纠正错误思路。\n- 一键生成聚焦于重大失误的 SGF 复习文件，自动标注变化图，让课后复习效率提升数倍。\n\nkatrain 将顶尖 AI 的算力转化为个性化的私人教练，让围棋学习者从“盲目试错”转向“精准修正”。","https:\u002F\u002Foss.gittoolsai.com\u002Fimages\u002Fsanderland_katrain_40dbac3f.png","sanderland","Sander Land","https:\u002F\u002Foss.gittoolsai.com\u002Favatars\u002Fsanderland_a640a691.png","AI @ Writer",null,"Oslo, Norway","magikarp_tokens","https:\u002F\u002Fgithub.com\u002Fsanderland",[81,85],{"name":82,"color":83,"percentage":84},"Python","#3572A5",86.7,{"name":86,"color":87,"percentage":88},"kvlang","#1da6e0",13.3,2192,281,"2026-04-10T12:57:19","NOASSERTION","Windows, macOS, Linux","非必需。默认内置 OpenCL 版本支持通用 GPU；可选配 NVIDIA GPU 使用 CUDA 或 TensorRT 版本以提升性能；无 GPU 可使用 CPU (Eigen) 版本。具体型号和显存未说明。","未说明",{"notes":97,"python":98,"dependencies":99},"该工具主要依赖围棋 AI 引擎 KataGo。Windows、Linux 及 M1 之前的 macOS 系统预打包了 OpenCL 版本的 KataGo 和旧版模型。用户可在设置中下载更新更强的模型文件或切换不同后端的 KataGo 二进制文件（如 CPU\u002FEigen、CUDA、TensorRT）。若使用远程服务器或特殊配置，可自定义启动命令。推荐使用 pipx 在隔离环境中安装，或通过 Homebrew (macOS) 安装。","3.x (通过 pipx 安装暗示需要 Python 环境，具体版本未说明)",[100,101],"KataGo (内置或单独配置)","pipx (推荐安装方式)",[13,14,15],[104,105,106,107,108,109,110,111],"go","baduk","weiqi","ai","katago","lizzie","sgf","kivy","2026-03-27T02:49:30.150509","2026-04-11T21:48:18.799185",[115,120,125,130,135,140],{"id":116,"question_zh":117,"answer_zh":118,"source_url":119},29844,"如何在 2020 款及更新的 Mac M1\u002FM2 架构电脑上运行 KaTrain？","早期版本在 M1 芯片上存在依赖和架构兼容性问题（如 kivy._clock 错误）。建议直接升级到 KaTrain 1.11 或更高版本，该版本已修复相关问题并可正常运行。如果仍遇到启动问题，可以尝试在终端中使用命令 `open -n -a katrain` 作为临时解决方案。","https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fissues\u002F331",{"id":121,"question_zh":122,"answer_zh":123,"source_url":124},29845,"是否可以通过 SSH 远程运行 KataGo 引擎并在本地 KaTrain 中调用？","可以。KaTrain 支持通过 SSH 远程运行 KataGo 引擎并进行交互（类似于 Lizzie 的方式）。用户只需在设置中配置远程引擎命令即可，该功能既可用于棋局分析，也可用于对弈和教学游戏。","https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fissues\u002F131",{"id":126,"question_zh":127,"answer_zh":128,"source_url":129},29846,"如何获取关于不同级别（级位\u002F段位）玩家错误类型的统计数据和原始数据？","维护者提供了相关的 Jupyter Notebook 用于分析游戏数据。用户可以访问 `https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain-bots\u002Fblob\u002Fmaster\u002Fanalyze_games.ipynb` 以及该仓库中的其他 Notebook 文件，从中获取直方图原始数据，进而分析减少各类错误以达到下一个等级所需的指导方针。","https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fissues\u002F44",{"id":131,"question_zh":132,"answer_zh":133,"source_url":134},29847,"KaTrain 是否提供类似 Chess.com 的全局表现评分或准确率统计功能？","是的，该功能已在后续版本（1.9+）中发布。系统会根据胜率变化将每一步棋分类（如失误、劣着、佳着等），并基于最佳着手比例生成玩家的总体“准确率”分数（0-100 分），帮助用户宏观了解整盘棋的表现。","https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fissues\u002F388",{"id":136,"question_zh":137,"answer_zh":138,"source_url":139},29848,"KaTrain 中的段位校准（Rank Calibration）是如何工作的？为什么感觉机器人变强了？","段位校准会随引擎更新而调整。由于较弱的机器人策略显著增强，导致整体强度上升。维护者通过调整“覆盖”（override）阈值来保持校准准确性：例如，对于明显的最优着法，如果引擎未首选但属于前两手好棋，系统会进行干预。通过设定较高的阈值，可以将因双最优着法导致的覆盖比例从 28% 降低至 3%，从而维持原有的段位校准标准。","https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fissues\u002F73",{"id":141,"question_zh":142,"answer_zh":143,"source_url":144},29849,"如何更清晰地查看地盘（Territory）和死子信息？目前的网格显示太乱了。","针对地盘显示杂乱和死子标记缺失的问题，社区已提出并实现了改进方案。新的显示方式采用平滑混合（blended）的地盘渲染代替网格矩形，使边界更清晰；同时在棋子上添加所有权标记，并将死子显示为半透明且带有对手标记，弱子也会显示透明度变化和小型标记，以便更直观地判断死活和形势。","https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fissues\u002F555",[146,151,156,161,166,171,176,181,186,191,196,201,206,211,216,221,226,231,236,241],{"id":147,"version":148,"summary_zh":149,"released_at":150},206434,"v1.17.1.1","[感谢](https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fpull\u002F803) @pscheid92，我们现在为 Mac 提供了基于 Intel 和 ARM 的安装程序。\n\n尽情使用吧！","2026-02-08T20:45:10",{"id":152,"version":153,"summary_zh":154,"released_at":155},206435,"v1.17.0","得益于 @jtkelm2 的重大贡献，KaTrain 现在加入了拟人化 AI 对弈功能。\n\n* 请通过常规设置下载模型：点击“下载模型”，并确保在“拟人化模型路径”选项中已正确选择。\n* 随后，在“AI 设置”中选择两个新选项之一（拟人化或历史职业棋手），并设定其强度和对局年份。\n\n## 变更内容\n\n* 添加了拟人化模型，并由 @jtkelm2 在 https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fpull\u002F744 中进行了重构。\n* 修复了 Windows 地理位置相关 Bug，由 @TuNanTang 完成，链接为 https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fpull\u002F743。\n* 升级 Kivy、迁移到 uv 构建系统、修复打包问题，并由 @sanderland 在 https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fpull\u002F748 中改进了配置界面。\n\n**完整变更日志**：https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fcompare\u002Fv1.16...v1.17.0\n\n## 安装说明\n\n* Windows 用户可以直接下载并运行随附的 KaTrain.exe 文件，或下载 KaTrain.zip 压缩包并运行其中的 .exe 文件（启动速度会稍快一些）。\n* macOS 用户可使用随附的 .dmg 文件（打开后将 .app 拖入应用程序目录）。首次启动应用时，需前往隐私设置允许打开。请注意，对于采用 M1 芯片的新款 Mac，仍需通过 Homebrew 安装 Katago，或自行编译 Katago。\n* Linux 和 macOS 用户可以使用 `pipx install katrain` 或 `pipx upgrade katrain` 命令，在隔离环境中安装或升级应用（借助 pipx 工具，运行时使用 `katrain` 命令）。此外，也可使用 `uvv katrain` 命令，在不进行安装的情况下直接运行（借助 uv 工具）。\n\n## 补丁说明\n\n* 1.17.1 修复了 Python 3.13 的兼容性问题。此版本仅发布于 PyPI。","2025-06-13T17:46:23",{"id":157,"version":158,"summary_zh":159,"released_at":160},206436,"v1.16","本次发布将 KataGo 更新至最新版本，并包含对本地化和主题的一些贡献。\n\n## 新贡献者\n* @psygo 在 https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fpull\u002F709 中完成了他们的首次贡献。\n* @rmcat 在 https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fpull\u002F710 中完成了他们的首次贡献。\n* @naltang 在 https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fpull\u002F723 中完成了他们的首次贡献。\n\n**完整更新日志**: https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fcompare\u002Fv1.15.0...v1.16","2025-04-30T07:26:29",{"id":162,"version":163,"summary_zh":164,"released_at":165},206437,"v1.15.0-fix","本次发布将 KataGo 更新至最新版本，默认包含更强的 18b 网络。目前的训练基于更新的 28b 模型，如果您的 GPU 支持，建议通过“常规设置”面板下载最强的 28b 网络。\n\n## 修复与改进\n* 更新了 KataGo 和 18b 网络\n* @xiaoyifang 添加了显示着法编号的功能（按“m”键）以及一项修复导航变例时移动问题的补丁\n* @rooklift 将性能报告的精度提升至更高位数\n* 通过配置文件而非命令行传递 numAnalysisThreads 参数\n* 修复了 Ubuntu 系统下的声音问题\n* @lykahb 将依赖管理工具更新为 Poetry\n* 修复了一个 bug：在特定条件下，导航面板中崩溃的子树会导致错误。\n\n请注意，尽管 KataGo 1.15 支持多项更接近人类下法的新特性，但本版本的 KaTrain 尚未启用这些功能。\n\n## 安装\n* Windows 用户可以直接下载并运行随附的 KaTrain.exe 文件，或下载 KaTrain.zip 压缩包后运行其中的 .exe 文件（启动速度会稍快一些）。\n* macOS 独立应用目前暂不可用。\n* Linux 和 macOS 用户可使用 pip3 install -U katrain 进行安装。\n * 需要注意的是，最低 Python 版本要求已更新至 3.9，因为这是 [即将结束支持的](https:\u002F\u002Fendoflife.date\u002Fpython)最低版本。","2024-08-17T20:07:59",{"id":167,"version":168,"summary_zh":169,"released_at":170},206438,"v1.14.0","## 变更内容\n本版本的主要改动是将默认使用的模型网络升级为性能更强的版本。如果您已经在使用该网络，则无需升级；但对于新用户来说，此次发布将使上手更加便捷。\n\n得益于近期的技术进步，采用改进架构的最新18b模型在网络表现上，在几乎任何设置下都明显优于旧版的20\u002F30\u002F40\u002F60b选项。此外，该模型体积较小，便于直接打包到发行版中。\n\n此前的默认15b模型仍可在常规设置中下载，供资源有限的设备用户使用。通常情况下，只有在使用新的默认网络时出现崩溃问题，才建议切换回15b模型。\n\n## 安装说明\n\n* Windows用户可以直接下载并运行随附的KaTrain.exe文件，或下载KaTrain.zip压缩包后运行其中的.exe文件（启动速度会稍快一些）。\n* macOS用户可以使用随附的KaTrainOSX.dmg镜像文件（打开后将.app拖至“应用程序”文件夹）。首次启动应用时，您需要在访达中按住Control键点击应用图标，以允许安装。请注意，对于搭载M1芯片的新款Mac，您仍然需要通过Homebrew安装katago，或自行编译katago。\n* Linux和macOS用户可以通过pip3 install -U katrain命令进行安装。\n","2023-07-08T16:10:58",{"id":172,"version":173,"summary_zh":174,"released_at":175},206439,"v1.13.0","这又是一个小版本更新，但仍建议升级，因为它引入了新的 KataGo 版本（1.13.x），包含一些有趣的新功能和错误修复，其中包括又一种新型网络。\n\n此外，还修复了 NGF 文件格式中的一个小 bug。\n\n安装详情请参阅其他版本发布说明。","2023-05-28T13:59:24",{"id":177,"version":178,"summary_zh":179,"released_at":180},206440,"v1.12.3","这是一个小版本更新，但仍推荐使用，因为它包含了带有几项有趣新功能的新版 KataGo，其中包括一种新型网络。\n\n## 更改\n\n* 包含 KataGo 1.12.1 → 1.12.2 → 1.12.3\n * 其中包含一个新的 18b 网络，其强度介于旧的 40b 和 60b 之间，更多信息请参阅 [这里](https:\u002F\u002Fgithub.com\u002Flightvector\u002FKataGo\u002Freleases\u002Ftag\u002Fv1.12.0)。\n * 请注意，版本号相同更多是巧合，KaTrain 1.12 并不包含 KataGo 1.12。\n* 棋子归属可视化改进与修复\n * 修复了教学模式下的部分棋子归属可视化错误\n * 扩展了棋子归属的主题选项\n * 对默认主题进行了小幅调整，使其更加简洁\n\n## 安装\n\n* Windows 用户可以直接下载并运行随附的 KaTrain.exe 文件，或下载 KaTrain.zip 压缩包并运行其中的 .exe 文件（启动速度会稍快一些）。\n* macOS 用户可以使用随附的 KaTrainOSX.dmg 文件（打开后将 .app 拖入“应用程序”文件夹）。首次启动应用时，您需要在 Finder 中右键单击该应用以允许安装。需要注意的是，在采用 M1 架构的新款 Mac 上，您仍然需要通过 Homebrew 安装 KataGo，或者自行编译 KataGo。\n* Linux 和 macOS 用户可以使用 `pip3 install -U katrain` 命令进行安装。\n\n首次安装程序后，建议检查常规设置（F8），并下载最新的 KataGo 模型。\n\n## 补丁\n\n* 1.12.2：替换 KataGo 版本，因为早期版本存在 bug，并且被禁止贡献代码。\n* 1.12.3：再次更新 KataGo。","2023-01-09T20:41:20",{"id":182,"version":183,"summary_zh":184,"released_at":185},206441,"v1.12","\r\n\r\n## 新功能：\n\n* 棋局所有权可视化得到了改进，这得益于 @jacobm-tech 的出色工作。\n * 默认样式现已改为混合模式，但您可以通过主题设置选择标记、区块或混合样式，并自定义何时标记棋子。\n * 有关所有选项，请参阅 [themes](THEMES.md)。\n * 旧主题已更新，以尽量保持原有风格。\n * 如有设计改进建议或问题报告，欢迎提交至 [此议题](https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fissues\u002F555)。\n * 默认设置可能会调整，尤其是关于“全部”或“仅弱子”标记的选择。可通过 [此投票](https:\u002F\u002Fstrawpoll.com\u002Fpolls\u002FQrgeV9WoLZp) 快速了解大家的意见，以便做出决策。\n* 更深入的分析现在可以仅应用于特定的步数范围。\n* 连续分析及其他附加分析模式在开启“最佳着法”提示时，行为更加温和。\n * 使用 Shift+空格键而非单纯按空格键，可避免在连续分析的剩余过程中自动开启这些提示。\n\n\n## 错误修复与小幅改进：\n\n* 再次支持 Python 3.10。\n* 当配置中将根噪声宽度设为 0.0 时，该设置现能正确生效，此前会自动恢复为默认值。\n* 修复了棋盘旋转功能：\n * +修复了棋盘旋转时的兴趣区域显示问题。\n * +不再依赖 NumPy，从而减小了二进制文件大小并提升了启动速度。\n* 针对 Windows 系统上的高分辨率显示器优化了显示效果。\n* 当某些提示被关闭时（例如 Lizzie 主题），悬停在这些提示上也将被禁用。\n* Windows 可执行文件现已包含版本信息，感谢 @ulty4life 的贡献。\n\n## 安装说明\n\n* Windows 用户可以直接下载并运行随附的 KaTrain.exe 文件，或下载 KaTrain.zip 压缩包后运行其中的 .exe 文件（这种方式启动速度稍快）。\n* macOS 用户可使用随附的 KaTrainOSX.dmg 文件（打开后将 .app 拖入应用程序目录）。首次启动应用时，需在 Finder 中右键单击应用并选择“允许”，以完成安装。请注意，在采用 M1 架构的新款 Mac 上，仍需通过 Homebrew 安装 Katago，或自行编译 Katago。\n* Linux 和 macOS 用户可使用 `pip3 install -U katrain` 命令进行安装。\n\n首次安装程序后，建议检查常规设置（F8），并下载最新的 Katago 模型。","2022-11-20T13:22:24",{"id":187,"version":188,"summary_zh":189,"released_at":190},206442,"v1.11.2fixes-for-osx","这是一次小更新，主要修复了一些 bug。\n\n## Bug 修复\n\n* SGF 分析中缓存分析功能不慎从分析选项中移除，现已修复。\n* 对小型棋盘上的让子棋局的音效及边缘情况进行了小幅修复。\n\n## 新特性\n\n* 增加了棋盘旋转按钮，感谢 @Funkenschlag1 的贡献。\n\n## 更新内容\n\n* 1.11.1 -> 1.11.2：修复了 Windows 版本中的一些问题以及配置文件版本相关的问题。\n * 请注意，新的 Windows 版本由于包含了 NumPy 二进制文件，体积较大。因此建议使用 .zip 格式的压缩包，而非独立的 .exe 文件，因为后者启动速度会更慢。\n* macOS 二进制文件已使用新标签（1.11.2fixes-for-osx）进行更新。","2022-06-23T20:06:22",{"id":192,"version":193,"summary_zh":194,"released_at":195},206443,"v1.11","\r\n## 新功能\r\n\r\n* 内置的 KataGo 已升级至 v1.11 版本 [查看改进内容](https:\u002F\u002Fgithub.com\u002Flightvector\u002FKataGo\u002Freleases\u002Ftag\u002Fv1.11.0)\r\n* 棋题框架，感谢 @kaorahi https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fissues\u002F498\r\n* 在着法树中新增“修剪分支”选项。此功能会移除所有通向所选节点的旁支。\r\n* 土耳其语翻译，感谢“Bora Zafer”\r\n* 用于对局贡献的键盘快捷键\r\n * Esc 键可优雅地中止贡献，并结束当前正在进行的对局\r\n * Pause 键用于暂停或恢复对局贡献\r\n * 若要强制退出 KataGo 而不完成正在进行的对局，只需直接退出 KaTrain 程序即可\r\n* 估算目数界面重新设计，感谢 @[milescrawford](https:\u002F\u002Fgithub.com\u002Fmilescrawford) 在 #487 中的贡献\r\n* 接近零点时的高精度比分显示，感谢 @pdeblanc 在 #514 中的贡献\r\n * 新增一项设置，可在接近零点时以 2 位有效数字显示比分估计值，该选项可在分析设置中启用。\r\n\r\n## 技术变更与修复\r\n\r\n* 升级至 Kivy 2.1\r\n * 此次升级有望改善 macOS 平台上的构建体验。\r\n * 已放弃对 Python 3.6 的支持。\r\n\r\n## 安装说明\r\n* Windows 用户可以直接下载并运行随附的 KaTrain.exe 文件，或下载 KaTrain.zip 压缩包后运行其中的 .exe 文件（启动速度稍快一些）。\r\n* macOS 用户可以使用随附的 KaTrainOSX.dmg 文件（打开后将 .app 拖入应用程序文件夹）。首次启动应用时，需在 Finder 中按住 Control 键点击应用图标以允许安装。请注意，在采用 M1 架构的新款 Mac 上，仍需通过 Homebrew 安装 KataGo，或自行编译 KataGo 可执行文件。\r\n* Linux 和 macOS 用户可使用 pip3 install -U katrain 命令进行安装。\r\n\r\n首次安装程序后，建议检查常规设置（F8 键），并下载最新的 KataGo 模型。","2022-03-24T16:16:12",{"id":197,"version":198,"summary_zh":199,"released_at":200},206444,"1.10.1","\r\n## Major additions\r\n\r\n* Continuous analysis or 'pondering' has been improved. \r\n * It will now no longer schedule queries with increasing visits, but just use a single never-ending query.\r\n * This also fixed a [bug with the region of interest function](https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fissues\u002F483)\r\n * This will break compatibility with KataGo versions older than v1.7.0.\r\n* KataGo was upgraded to 1.10.0.\r\n * The linux executable now includes a self-compiled, statically linked libzip, which should reduce [issues](https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fissues\u002F482) with versioning.\r\n* An AntiMirror AI option was added which uses KataGo's special mode for beating players who play mirror go.\r\n* Scrolling through variations now only works while the cursor is on the board, reducing [conflict with browsing the move tree](https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fissues\u002F461).\r\n* Middle clicking while scrolling through variations [now adds it to the move tree](https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fissues\u002F389).\r\n\r\n## Minor fixes and additions\r\n\r\n* The SGF[ 'compressed points list'](https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fissues\u002F474) functionality which allows you to place a square of stones is now supported\r\n* On MacOS, command should function as ctrl\r\n* The log file setting in the default config was removed, as it was broken\r\n* A bug was fixed where the contribute function could break if the last game included handicap stones.\r\n* The MacOS packages should now have [correct versioning](https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fissues\u002F480)\r\n\r\n## Installation\r\n* Windows users can simply download and run the KaTrain.exe file attached, or download the KaTrainWindows.zip file and run the .exe in that (which is slightly faster to start up)\r\n* macOS users can use the attached KaTrainOSX.dmg (open, drag .app to applications). The first time you start the app, you will need to control-click the app in finder to allow installation. Note that on the newer Macs with M1 architecture, you still need to brew install katago, or compile katago yourself.\r\n* Linux and macOS users can install using pip3 install -U katrain\r\n\r\nAfter installing the program for the first time, it is recommended to check general settings (F8) and download the latest katago models.","2021-10-31T18:59:48",{"id":202,"version":203,"summary_zh":204,"released_at":205},206445,"1.9.3","![image](https:\u002F\u002Fuser-images.githubusercontent.com\u002F48946947\u002F123637150-fae69a00-d81d-11eb-98b4-ae7dca880db4.png)\r\n\r\n# Major features\r\n\r\n* **KataGo Update**: This release includes KataGo 1.9 which includes major improvements. Upgrading is highly recommended.\r\n * A major change is that the win rates and scores are unaffected by wider exploration, which allows for more use of the 'wide root noise' setting encouraging exploration. The default value of this has been changed to 0.04 in both KaTrain and KataGo itself.\r\n * Particularly, you are likely to see fewer top moves with small positive changes in score, and fewer instances where the win rate\u002Fscore oscillates at low visits settings.\r\n * In addition, symmetry is taken into account. You might notice that in opening positions in particular this combination leads to very wide exploration. If you prefer to see fewer moves, you can change the wide root noise in general settings back to a lower value.\r\n * One side-effect of this is that it breaks compatibility with some older versions (Around 1.7, such as the old 'bigger board size' binary). A new bigger board size binary is included in the downloadable katago versions. \r\n \r\n* **Performance Report** : F3 \u002F Analysis options now has a \"performance report\" which shows key statistics including AI match rate, mistakes of various kinds, and an overall rating.\r\n * Keep in mind that this overall rating number is mainly for entertainment, and a low visit count will potentially result in unreliable numbers. [Suggestions for improvements](https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fissues\u002F388) are welcome, but please do not file issues about individual games.\r\n\r\n* **Better support for themes**\r\n * Multiple theme files are now supported, allowing more easy installation of multiple themes, such as a board graphics change, and keyboard shortcuts. Any json file starting with \"theme\" in your settings folder is now included.\r\n * Keyboard shortcuts can now be defined in themes. Thanks to @electricRGB for helping with this, and pointing out that the current setup can be annoying for non-qwerty keyboards. \r\n * See [here](THEMES.md) for more information, and further on in these release notes for a contest! \r\n\r\n# Minor features\r\n\r\n* It is now possible to play a random sound in response to a mistake. See [here](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=BmGn5T3xwu4) for a fun example. Thanks to [Mikkgo and Jeff](https:\u002F\u002Fwww.youtube.com\u002Fchannel\u002FUCXz28yKTrSiobcemJMCMnVg) for the sounds. \r\n* When KataGo fails to start, or crashes, a pop-up with more detailed help opens up. [This page](ENGINE.md) will be linked as additional help. Feel free to contribute to it with solutions to problems I missed.\r\n* You can now use the mouse scroll wheel to quickly scroll through a variation when hovering over a candidate move. Thanks to @xiaoyifang for implementing this.\r\n* You can now turn off window size\u002Fposition restoration in general settings. This can help if this feature does strange things on multiple monitor setups.\r\n* Sound should be more reliable, especially on Linux.\r\n\r\n# Theme contest\r\n\r\n* To encourage exploration and development of the themes support, there will be a small contest during July.\r\n * Valid contributions include both theme packages, as well as code that improves theme support or default looks and widget design.\r\n * See the [OGS forums topic](https:\u002F\u002Fforums.online-go.com\u002Ft\u002Fkatrain-v1-9-theme-competition\u002F37625) for details.\r\n\r\n# Installation\r\n\r\n* Windows users can simply download and run the KaTrain.exe file attached, or download the KaTrainWindows.zip file and run the .exe in that (which is slightly faster to start up)\r\n* macOS users can use the attached KaTrainOSX.dmg (open, drag .app to applications). The first time you start the app, you will need to [control-click the app in finder](https:\u002F\u002Fsupport.apple.com\u002Fguide\u002Fmac-help\u002Fopen-a-mac-app-from-an-unidentified-developer-mh40616\u002Fmac). Note that on the newer Macs with M1 architecture, you still need to `brew install katago`, or compile katago yourself.\r\n* Linux and macOS users can install using `pip3 install -U katrain` \r\n* For troubleshooting and installation from source follow instructions [here](https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fblob\u002Fmaster\u002FINSTALL.md).\r\n\r\nAfter installing the program, it is recommended to check general settings (F8) and download the latest katago models.\r\n\r\n# Patches\r\n\r\n* 1.9.1: Fix for KivyMD breaking MRO on a patch upgrade\r\n* 1.9.2: Upgrade to KataGo 1.9.1, preventing crashes of the engine.\r\n* 1.9.3: Tiny fix for mac sounds, windows executables remain at 1.9.2 as they are unaffected.\r\n","2021-06-28T15:52:58",{"id":207,"version":208,"summary_zh":209,"released_at":210},206446,"1.8.4","This is a minor patch which includes:\r\n\r\n* Support for rulesets which allow suicide, including in distributed training. The lack of support for this caused occasional errors when contributing to the distributed training project.\r\n* Updated for downloading KataGo 1.8.1 in general settings, instead of KataGo 1.8.0\r\n\r\nIf you are not using KaTrain to contribute to [distributed training](https:\u002F\u002Fkatagotraining.org\u002F), there is likely no need to update.","2021-04-11T17:07:51",{"id":212,"version":213,"summary_zh":214,"released_at":215},206447,"1.8.3","\r\n![image](https:\u002F\u002Fuser-images.githubusercontent.com\u002F48946947\u002F110167763-7dbe9180-7df6-11eb-9c4a-08ffc2438fd7.png)\r\n\r\n## YouTube Tutorials\r\n* [Generated positions](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=55FQFUqZyK0)\r\n\r\n## Major new features\r\n\r\n* Contribute to [KataGo's distributed training run](https:\u002F\u002Fkatagotraining.org\u002F)! \r\n * A new option in the main menu will assist you in setting up contributing to distributed training\r\n * Included binaries for Windows, Linux and MacOS should all support this mode.\r\n * Keep in mind that the newer M1 macs needs you to compile katago yourself for this, as the current brew version does not support distributed training.\r\n * You can watch games being played, exploring policy, potential moves and variations.\r\n * Hit [spacebar] to switch between automatically playing and manually browsing through the current game.\r\n * Most other options are disabled in this mode.\r\n * See [here](https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain#distributed) for more details.\r\n\r\n* Setup position in new game\r\n * The new game now has tabs for new game\u002Fedit rules\u002Fsetup position\r\n * Setup position will generate a position where one player is ahead.\r\n * The intended use for this is to play against KataGo from an advantage, and try to win the game.\r\n * For an extra challenge use KataHandicap, but remember to manually set the playout doubling advantage.\r\n\r\n## Minor features and bug fixes\r\n\r\n* The 'show dots' toggle is back to having only two options, all off and 'some on'. \r\n * This was mainly due to the 'next mistake' navigation making the view with all evaluations somewhat redundant, and not warranting the confusing 3-state checkbox.\r\n * The option to determine how many moves to show then on is back in analysis options configuration.\r\n* The window size\u002Fposition is now saved on exit and restored on startup.\r\n* The fast analysis and rewind to start options in 'load game' are now remembered.\r\n* A new analysis option 'Reset analysis' was added which resets analysis to a single engine query, removing alternate moves and child move derived info.\r\n* A new shortcut [**escape**] stops all analysis.\r\n * This can break some features which expect analysis to be present, such as AI moves. Resetting with h or toggling continuous analysis will typically fix such edge cases.\r\n* Nodes with clear location commands in the sgf (AE) are now correctly shown.\r\n * However, they are not analyzed.\r\n * Nodes with only edit position commands (AB\u002FAW\u002FAE) are now indicated in grey with a + in the move tree\r\n* Scoring by ownership square counting in japanese rules no longer confuses the last dame with a point.\r\n\r\n\r\n## Installation\r\n\r\n* Windows users can simply download and run the KaTrain.exe file attached, or download the KaTrainWindows.zip file and run the .exe in that (which is slightly faster to start up)\r\n* macOS users can use the attached KaTrainOSX.zip (unzip, mount .dmg, drag .app to applications)\r\n * Note that on the newer Macs with M1 architecture, you still need to `brew install katago`, or compile katago yourself.\r\n* Linux and macOS users can install using `pip3 install -U katrain` or follow instructions [here](https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fblob\u002Fmaster\u002FINSTALL.md).\r\n\r\n## Patches\r\n* 1.8.1 Fix for single exe (homeDataDir)\r\n* 1.8.2 Fix for cacert on alternate katago and sizing on retina displays.\r\n* 1.8.3 Fix for >1M visits display and windows single exe distributed engine startup","2021-03-05T21:59:57",{"id":217,"version":218,"summary_zh":219,"released_at":220},206448,"v1.7.6","This is another minor update, improving on some issues found in 1.7.1, and including better support for downloading models.\r\n\r\nI've also managed to create a packaged release for OSX, thanks to kivy developer @matham. This package includes a pre-compiled katago executable and should be all you need to run KaTrain on OSX. So far this has been tested on OSX 10.14.\r\n\r\n## Features:\r\n\r\n* You can now download the confidently strongest distributed model from general settings.\r\n* You can now press shift-b to navigate from a deep variation back to the main branch, thanks to @LiamHz \r\n\r\n## Bug fixes\r\n\r\n* Better stabiility for model downloads.\r\n* Queries should be properly cleared on new game, preventing an incorrect 'forever busy' engine status.\r\n* Fix for initial player detection using PL[]\r\n\r\n## New Theme & New YouTube Tutorial\r\n* See [here](https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fblob\u002Fmaster\u002FTHEMES.md) for the first theme added.\r\n* See [here](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=UG357l9LN-Q) for a new YouTube video on using the 'insert moves' feature to improve your ability to review games in situations when both players ignore an urgent move for a long time.\r\n\r\n## Installation\r\n\r\n* Windows users can simply download and run the KaTrain.exe file attached, or download the .zip file and run the .exe in that (which is slightly faster to start up)\r\n* MacOS users can use the attached KaTrainOSX.zip (unzip, mount .dmg, drag .app to applications)\r\n * Note that on the newer Macs with M1 processor you still need to `brew install katago`\r\n* Linux and MacOS users can install using `pip3 install -U katrain` or follow instructions [here](https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fblob\u002Fmaster\u002FINSTALL.md).\r\n\r\n## **Update 1.7.3**\r\n\r\n* Fixes a bug in downloading katago versions.\r\n* Shuts down faster if an external command is used.\r\n* Avoids using the provided exe for new M1 macs and looks for a brewed binary instead.\r\n* Uses an older pyinstaller to give less false positives for antivirus programs\r\n\r\n## **Update 1.7.4**\r\n* Fixed a bug in download models.\r\n\r\n## **Update 1.7.6**\r\n* Better support for M1","2021-01-23T11:00:09",{"id":222,"version":223,"summary_zh":224,"released_at":225},206449,"1.7.1","This release includes an upgrade to the latest KataGo version, along with a number of minor features and bug fixes.\r\n\r\n## Features:\r\n\r\n* The built-in KataGo, as well as the additional versions that can be downloaded, have been upgraded to v1.8\r\n * These are compatible with the newest distributed models\r\n * Using 'download models' in general settings will download the latest model of the distributed run.\r\n* SGF file improvements\r\n * You can now use the right click 'open with' feature of your OS to open an SGF file\r\n * KaTrain will now try to determine the encoding for SGF files without one, instead of using latin-1. This should help Chinese SGF files in particular\r\n * Pressing 'enter' in the load\u002Fsave dialog when a directory is in the text field now navigates to that location\r\n* The root node is now analyzed using the 'fast' setting in a new game. This should make startup a bit quicker. Likewise, on changing models or rules, fast analysis is used to re-analyze the game.\r\n* Ancient chinese \"stone scoring\" rules option which includes group tax, thanks to @pdeblanc\r\n* Traditional Chinese translation thanks to @Tony-Liou\r\n* Some elements of the UI can now be resized using the themes feature.\r\n\r\n## Bug fixes:\r\n\r\n* Fix that prevents the automatic self-play feature from deadlocking when many simultaneous games were going on at the same time\r\n* Loading generated player names are now ignored\r\n* Saving analysis to existing SGF files no longer gives errors on re-loading\r\n* Passes in top moves no longer results in SQ[] in SGF files\r\n* Support for [tt] as passs on \u003C19x19 boards\r\n* Invalid HA properties are ignored instead of causing crashes\r\n* Alt\u002FTab shortcuts should no longer interfere with switching windows using Alt-Tab\r\n* Better guess for the initial player, respecting the PL tag and existing branches. This should help unusual handicap placements.\r\n* Font size for the policy moves now scales properly.\r\n* Configuration popups are restricted to the window height, such that the save button is always visible.\r\n* Show dots as 'all off' now properly saved in config and respected on restarting\r\n* Upper-case extensions such as .SGF now show up in load file\r\n* Pressing up\u002Fdown in the move tree when on a collapsed path no longer causes an error.\r\n\r\n## Installation\r\n\r\n* Windows users can simply download and run the KaTrain.exe file attached, or download the .zip file and run the .exe in that (which is slightly faster)\r\n* Linux and MacOS users can install using pip install -U katrain or follow instructions [here](https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fblob\u002Fmaster\u002FINSTALL.md).\r\n","2021-01-16T18:18:48",{"id":227,"version":228,"summary_zh":229,"released_at":230},206450,"1.7","\r\n## A little history and a big thank you\r\n\r\nThis project was started at the end of last year as a small experiment in making a graphical UI for something, and has since grown to one of the most powerful tools for Go AI analysis. This could not have been done without the support and [contributions](CONTRIBUTIONS.md) of many. A big thank you to all them, as well as all the people who contacted me on [reddit](https:\u002F\u002Fwww.reddit.com\u002Fuser\u002Fsanderbaduk\u002F) and [discord](https:\u002F\u002Fdiscord.gg\u002FAjTPFpN) to send me helpful feedback, kind words and go books. Finally, thanks to all my [Github sponsors](https:\u002F\u002Fgithub.com\u002Fsponsors\u002Fsanderland) for their support.\r\nTo highlight the effect of some of the major contributors, I've prepared a little [history](https:\u002F\u002Fimgur.com\u002Fa\u002F4RUG02G) highlighting how the program changed since release.\r\n\r\n## Major New Features\r\n- You are now free to show any two statistics on the 'top moves', including visits, (change in) win rate, (change in) score.\r\n- The analysis menu now includes several new options:\r\n - Select a region to search, restricting analysis to this part of the board until cancelled. \r\n - This is particularly useful for tsumego solving or joseki analysis. See the video below for an example.\r\n - The old 'local moves' has been removed in favour of this option.\r\n - Insert moves: this mode allows you to insert moves in-between the current and next move, creating a new variation in which the moves are played out as long as they're legal (skipping over any inserted moves).\r\n - This is helpful for improving analysis when a certain exchange or life & death situation is ignored by both players for a long time.\r\n - Play out the game to the end.\r\n - This can be used as a more vivid way to visualize which territories could end up with each player.\r\n - Note that you can right-click the generated node and 'un-collapse' it (or press `c`) to see the entire game.\r\n- Analysis can now be cached in the SGF, causing analysis to be instantly available upon loading with no need for calls to KataGo\r\n - Note that this uses tags unique to KaTrain, and the resulting files can be >1Mb.\r\n - If you want to do additional analysis after loading, using the same model file is recommended.\r\n- Support for themes, see [here](THEMES.md) for details. \r\n - If you make any interesting themes, I would be happy to add them to the repository.\r\n\r\n### Local Joseki Analysis video\r\n [![Local Joseki Analysis Video](http:\u002F\u002Fi.imgur.com\u002FYcpmSBx.png)](https:\u002F\u002Fwww.youtube.com\u002Fwatch?v=tXniX57KtKk)\r\n\r\n## Minor New Features and Changes\r\n- You can now edit player names similarly to changing komi and rules.\r\n- Saving a game no longer uses a default filename, but prompts you for one. Subsequent saves use the same file, unless 'save as' is used.\r\n- Generated comments and player names are now filtered out from loaded SGFs\r\n - Particularly, loading and saving repeatedly no longer causes analysis comments to duplicate.\r\n - Note that this only applies to files saved with the current version.\r\n- You can now go back to the last branching point with `b`.\r\n- You can collapse branches up to the last branching point with `c`. These collapsed states are also saved in SGF files.\r\n- The policy moves view has been revamped to show the numerical value, and coloured by order of magnitude.\r\n- Minimizing the score graph in analysis mode now expands the move tree.\r\n- You can toggle coordinates on\u002Foff with `k`, thanks to @blamarche.\r\n- You can now go to your previous mistake with shift-n or the backward red arrow.\r\n- Some keyboard shortcuts have been added or changed, see the [manual](README.md) for details.\r\n- Spanish has been removed as a language option as it could not be maintained. If you are interested in contributing, please see [these](CONTRIBUTIONS.md) instructions.\r\n\r\n## Bug fixes and performance improvements\r\n- Rendering has been sped up significantly by caching textures. This should make variations smoother in particular.\r\n- Empty komi values in SGFs no longer cause errors.\r\n- The PL[] tag gets correctly picked up on when loading SGFs with problems.\r\n\r\n## Installation\r\n\r\n* Windows users can simply download and run the KaTrain.exe file attached, or download the .zip file and run the .exe in that (which is slightly faster)\r\n* Linux and MacOS users can install using `pip install -U katrain` or follow instructions [here](https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fblob\u002Fmaster\u002FINSTALL.md).\r\n * Note that the gui framework kivy had a major release recently, and installation on MacOS seems better in particular.\r\n\r\nEnjoy the update, and a very happy holidays!\r\n\r\n\r\n\r\n\r\n\r\n\r\n","2020-12-18T19:58:05",{"id":232,"version":233,"summary_zh":234,"released_at":235},206451,"1.6","This release adds a number of features to improve ease of use and responsiveness.\r\n\r\n## Features:\r\n* The navigation now has a button for 'Next mistake' which forwards to the next point in the game with an 'orange' (3 point by default) mistake or worse.\r\n * If you don't want to see your opponent mistakes, set them to 'AI' and they will be skipped.\r\n* Integration with the new 1.7.0 version of KataGo has been improved:\r\n * smoother updates to analysis in progress.\r\n * on loading a new game, previous queries will be terminated, so you do not have to wait for useless results.\r\n* The analysis menu has a new 'local search' option which searches within 5 spaces of the last move.\r\n* The simple-style AI has been added from 1.5.1. This option tries to settle territory, often leading to simpler moves.\r\n * This AI has not yet been calibrated and is just displayed as '2 dan', but increasing the max point loss will weaken it.\r\n* Timer options have been improved.\r\n * You can set a 'main time' in addition to byo-yomi\r\n * You can set a 'minimal thinking time in byo-yomi', which blocks you moving too fast.\r\n* You can now update the komi and rules of a game in progress in the 'new game' screen.\r\n * Note that this triggers re-analysis of the entire game.\r\n* You can now right-click a node in the move tree and make that branch the main\u002Ftop one.\r\n* The 'show dots' option now cycles between 'show all', 'show some' and 'show none' to avoid you having to go into analysis settings and set this to 0 when you want a game with no analysis information.\r\n* Clicking on the \"B\" or \"W\" circle in player options now swaps the players, making it easier to switch colours between games.\r\n* You can now paste a URL straight into the program with ctrl-v and your SGF will be downloaded and loaded, thanks to @wierzbowski \r\n* The engine status dot now shows the number of outstanding queries when there are more than 2. This should help get an indication of how long your SGF takes to analyze.\r\n* The 'top moves' score change display no longer caps at 0.0, but shows positive changes as well, which tend to happen when a 'surprising' move is found by the engine.\r\n\r\n## Bug fixes:\r\n* A suboptimal setting for the search was changed in v1.5.1, this has been fixed, and analysis should be faster.\r\n* Move placements in SGF files could occasionally lead to an error, which has been fixed.\r\n* Fonts should scale by your operating system's font settings, making the UI look better for people with very high screen resolutions.\r\n\r\n## Installation\r\n\r\n* Windows users can simply download and run the KaTrain.exe file attached, or download the .zip file and run the .exe in that (which is slightly faster)\r\n* Linux and MacOS users can install using `pip install katrain` or follow instructions [here](https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fblob\u002Fmaster\u002FINSTALL.md).\r\n\r\nEnjoy!","2020-11-23T12:11:42",{"id":237,"version":238,"summary_zh":239,"released_at":240},206452,"v1.5.1pre","\r\n![image](https:\u002F\u002Fuser-images.githubusercontent.com\u002F48946947\u002F96442193-02738f00-120b-11eb-8dbe-454d2c1b843c.png)\r\n\r\n\r\nThis is a pre-release intended to crowd-source the testing and refining of a new AI option. \r\n\r\nSpecifically it includes some AI options that are intended to play in a 'simple' way, but still leverage the full power of KataGo, avoiding some of the blind spots that the policy based AIs have.\r\n\r\nAs a little extra, it also has support for KataGo's latest feature for reporting back partial analysis queries, so the feedback during analysis should be a lot more responsive.\r\n\r\n### AI Options\r\nThe new options are: \r\n\r\n* **Simple style**: cares about all territory, preferring moves which clearly take or solidify own territory, even if this also solidifies the opponents territory.\r\n* **Settle style**: Similar, but only cares for stones, see issue #215 for details.\r\n\r\nI am mainly interested in people's experience in how these AIs play, which one is nicer to play against, and any weird behaviour they exhibit.\r\n\r\n### Giving feedback\r\n\r\nAny reports of experimenting with the settings is useful. You can add your experiences to [this issue](https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fissues\u002F220).\r\n\r\n* Useful: Feedback on which settings are more enjoyable to play against over the span of one or more games\r\n* Useful: Estimates of playing strenght at different settings (e.g. I am 1 dan and found these settings to give me a good opponent), or self-play experiments.\r\n* Useful: Suggestions for improvment to the move selection.\r\n* OK: Situations with huge blunders which destroy enjoyment, such as inappropriate passing, saved with full SGF ('ai thoughts') output.\r\n* OK: Sharing games which you found particularly enjoyable or interesting.\r\n* Less useful: Finding individual situations in which you find the move chosen to be 'not simple'.\r\n\r\nIn terms of calibrating the strength, these settings are expected to be most important:\r\n\r\n* `max_points_lost` - A higher value will be weaker, and a very low value reverts back to normal KataGo behaviour. \r\n* `settled_weight` - A higher value will be weaker, but play more strongly according to the style\r\n* `wide_root_noise` (in engine settings) - A higher value (of 0.02-0.05) will return more candidate moves, giving more opportunities to prefer the simple style, but also increase the chance of blunders.\r\n* `max_visits` (in engine settings) - more visits, more options, more likely to find a move appropriate to the style.\r\n\r\n### Fun experiments\r\n\r\nFor additional fun experiments, note:\r\n* A negative `settled_weight` will prefer unsettledness, leading to complicated positions\r\n* A positive `settled_weight` with a higher negative `opponent_fac` will prefer unsettledness of the opponent, leading to invasions.\r\n\r\nAttachment and tenuki penalties can also be negative or zero to make these moves prefered instead of avoided.\r\n\r\nThese kind of inverse settings are not the focus of this release, but feel free to share interesting games along with their settings.\r\n\r\n## Installation\r\n**Note that these AIs require features of KataGo which are not in the last release, so you need to use either the built-in opencl katago provided (which is up-to-date) or compile from source yourself.**\r\n\r\n* For windows, just download and run the attached .exe, or download the .zip and run KaTrain.exe (which is faster to start)\r\n* For linux, note that this release has not been uploaded to PyPi, but you can run the following to install it:\r\n * `pip install https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Farchive\u002Fv1.5.1.zip`\r\n * Alternatively for running it without installation, you can `git clone` the branch and run it [as usual](https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fblob\u002Fmaster\u002FINSTALL.md) with `python -m katrain`\r\n * The included opencl KataGo is updated for linux to include the latest unreleased fetures.\r\n* For MacOS, follow the linux instructions, and compile KataGo's master branch from source.\r\n\r\n","2020-10-19T10:47:51",{"id":242,"version":243,"summary_zh":244,"released_at":245},206453,"v1.5","\r\n![image](https:\u002F\u002Fuser-images.githubusercontent.com\u002F48946947\u002F93624979-7c80e000-f9e1-11ea-9f64-a94f71e825ba.png)\r\n\r\n## Improvements\r\n\r\n* A number of significant of graphical updates to analysis\r\n * The next moves are shown as a dashed line around the move rather than a semi-transparent stone, as overlap with 'top moves' hints was confusing.\r\n * If the next move is the top move, a dashed line with blue and the next player will be shown.\r\n * Top moves are softer circles, and fade out for low (\u003C25) visits. This should both make any borders clearer as well as make the board less overwhelming. The threshold can be configured in config.json (`low_visits`) but is not yet in the UI.\r\n * When viewing policy moves, 'top moves' are now hidden to avoid overlap.\r\n* A new analysis option, 'find alternative moves', has been added. This will do a 'fast visits' search on all current move options, as well as request a second query avoiding all current moves.\r\n* The model and katago binary dropdowns now show a little description of the files you can select.\r\n * A bug where downloaded katago versions didn't show up in the dropdown has been fixed.\r\n* Resignation is now possible. This is mainly to allow people to have more complete SGF records.\r\n * Game results are now stored in the SGF as long as the result (B+R, W+3.5, etc) is showing on top of the board.\r\n* The rank estimator has been put away, as it was causing more confusion than anything and not actively being improved.\r\n* Improved support for FoxWQ's incorrectly formatted SGF files, with extreme komi\u002Fweird handicap settings being converted to reasonable values when the SGF contains \"AP[foxwq]\"\r\n* Rule set and komi are now shown in info on the first move.\r\n\r\n### Bug fixes:\r\n* A bug where the first player was incorrect for 1-stone handicap games was fixed.\r\n* A bug where katago kept running in the background on windows after the program closed has been fixed.\r\n\r\n## Installation\r\n\r\n* Windows users can simply download and run the KaTrain.exe file attached, or download the .zip file and run the .exe in that (which is slightly faster)\r\n* Linux and MacOS users can install using `pip install katrain` or follow instructions [here](https:\u002F\u002Fgithub.com\u002Fsanderland\u002Fkatrain\u002Fblob\u002Fmaster\u002FINSTALL.md).\r\n\r\nEnjoy!","2020-09-18T18:31:12"]