[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-1627":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":9,"language":10,"languages":9,"totalLinesOfCode":9,"stars":11,"forks":12,"watchers":13,"openIssues":13,"contributorsCount":14,"subscribersCount":14,"size":14,"stars1d":15,"stars7d":15,"stars30d":16,"stars90d":14,"forks30d":14,"starsTrendScore":17,"compositeScore":18,"rankGlobal":9,"rankLanguage":9,"license":19,"archived":20,"fork":20,"defaultBranch":21,"hasWiki":22,"hasPages":20,"topics":23,"createdAt":9,"pushedAt":9,"updatedAt":24,"readmeContent":25,"aiSummary":26,"trendingCount":14,"starSnapshotCount":14,"syncStatus":15,"lastSyncTime":27,"discoverSource":28},1627,"NeuralComputer","metauto-ai\u002FNeuralComputer","metauto-ai","🖥 Neural Computers' Data Engine",null,"Python",201,26,3,0,2,15,6,4.29,"MIT License",false,"main",true,[],"2026-06-12 02:00:30","\u003Ch1 align=\"center\">Neural Computers (The Data Pipeline)\u003C\u002Fh1>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fmetauto.ai\u002Fneuralcomputer\u002F\">\n    \u003Cimg src=\"assets\u002Fhero_teaser_blog.svg\" alt=\"Neural Computer teaser diagram\" width=\"760\">\n  \u003C\u002Fa>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Fmetauto.ai\u002Fneuralcomputer\u002F\">📘 Blog\u003C\u002Fa>\n    ·  \n  \u003Ca href=\"https:\u002F\u002Fmetauto.ai\u002Fneuralcomputer\u002F\">🎬 Demo Gallery\u003C\u002Fa>\n    ·  \n  \u003Ca href=\"https:\u002F\u002Farxiv.org\u002Fsearch\u002F?query=Neural+Computers+Mingchen+Zhuge&searchtype=all&source=header\">📄 Paper\u003C\u002Fa>\n\u003C\u002Fp>\n\n## News\n\n- **2026 April**: First open-source release of Neural Computers, starting with the data pipeline for CLI and GUI trajectory generation.\n\n## Abstract\n\nNeural Computers (NCs) are neural networks that unify computation, memory, and I\u002FO in a single latent runtime state. The long-term goal is the Completely Neural Computer (CNC): a general-purpose neural computer that challenges the layered architecture of conventional computers. As an initial step, this project studies whether executable dynamics can be learned solely from collected I\u002FO traces, without access to instrumented program state, using CLI and GUI trajectories as training and evaluation data.\n\n## Quick Start\n\n**Environment** `Python 3.9+` · `Docker` · `asciinema` · `agg` · `ffmpeg`\n\n**1. CLI \u002F Asciinema**\n```bash\npython3 main.py cligen asciinema record --command \"python3 --version\"\npython3 main.py cligen asciinema cast-to-mp4 workspace\u002Fcligen_general\u002Fcasts\n```\n\n**2. CLI \u002F VHS**\n```bash\npython3 main.py cligen vhs build-image\npython3 main.py cligen vhs generate-basic --count 10 --prefix demo\npython3 main.py cligen vhs make-manifest\npython3 main.py cligen vhs run-manifest\n```\n\n**3. GUI \u002F Synthetic**\n```bash\npip install -r engine\u002Fgui\u002Fsynthetic_data_collection\u002Frequirements.txt\npython3 main.py guiworld build-image\npython3 main.py guiworld synthetic --count 1 --max-workers 1\n```\n\n### Runtime Generations from Neural Computers\n\n\u003Cp>\u003Csub>Using training data generated by the code released in this repository.\u003C\u002Fsub>\u003C\u002Fp>\n\n**CLIGen (general \u002F asciinema)**\n\n\u003Ctable>\n  \u003Ctr>\n    \u003Ctd>\u003Cimg src=\"assets\u002Fcg01.gif\" alt=\"CLIGen general demo 1\" width=\"160\">\u003C\u002Ftd>\n    \u003Ctd>\u003Cimg src=\"assets\u002Fcg02.gif\" alt=\"CLIGen general demo 2\" width=\"160\">\u003C\u002Ftd>\n    \u003Ctd>\u003Cimg src=\"assets\u002Fcg03.gif\" alt=\"CLIGen general demo 3\" width=\"160\">\u003C\u002Ftd>\n    \u003Ctd>\u003Cimg src=\"assets\u002Fcg04.gif\" alt=\"CLIGen general demo 4\" width=\"160\">\u003C\u002Ftd>\n    \u003Ctd>\u003Cimg src=\"assets\u002Fcg05.gif\" alt=\"CLIGen general demo 5\" width=\"160\">\u003C\u002Ftd>\n    \u003Ctd>\u003Cimg src=\"assets\u002Fcg06.gif\" alt=\"CLIGen general demo 6\" width=\"160\">\u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n**CLIGen (clean \u002F vhs)**\n\n\u003Ctable>\n  \u003Ctr>\n    \u003Ctd>\u003Cimg src=\"assets\u002Fcc01.gif\" alt=\"CLIGen clean demo 1\" width=\"190\">\u003C\u002Ftd>\n    \u003Ctd>\u003Cimg src=\"assets\u002Fcc04.gif\" alt=\"CLIGen clean demo 4\" width=\"190\">\u003C\u002Ftd>\n    \u003Ctd>\u003Cimg src=\"assets\u002Fcc06.gif\" alt=\"CLIGen clean demo 6\" width=\"190\">\u003C\u002Ftd>\n    \u003Ctd>\u003Cimg src=\"assets\u002Fcc09.gif\" alt=\"CLIGen clean demo 9\" width=\"190\">\u003C\u002Ftd>\n    \u003Ctd>\u003Cimg src=\"assets\u002Fcc11.gif\" alt=\"CLIGen clean demo 11\" width=\"190\">\u003C\u002Ftd>\n    \u003Ctd>\u003Cimg src=\"assets\u002Fcc14.gif\" alt=\"CLIGen clean demo 14\" width=\"190\">\u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n**GUIWorld (GUI)**\n\n\u003Cp>\u003Csub>Alternating: Conventional Computer \u002F Neural Computer\u003C\u002Fsub>\u003C\u002Fp>\n\n\u003Ctable>\n  \u003Ctr>\n    \u003Ctd>\u003Cimg src=\"assets\u002Fgc01.gif\" alt=\"GUIWorld conventional computer demo 1\" width=\"210\">\u003C\u002Ftd>\n    \u003Ctd>\u003Cimg src=\"assets\u002Fgn01.gif\" alt=\"GUIWorld neural computer demo 1\" width=\"210\">\u003C\u002Ftd>\n    \u003Ctd>\u003Cimg src=\"assets\u002Fgc10.gif\" alt=\"GUIWorld conventional computer demo 10\" width=\"210\">\u003C\u002Ftd>\n    \u003Ctd>\u003Cimg src=\"assets\u002Fgn10.gif\" alt=\"GUIWorld neural computer demo 10\" width=\"210\">\u003C\u002Ftd>\n    \u003Ctd>\u003Cimg src=\"assets\u002Fgc19.gif\" alt=\"GUIWorld conventional computer demo 19\" width=\"210\">\u003C\u002Ftd>\n    \u003Ctd>\u003Cimg src=\"assets\u002Fgn19.gif\" alt=\"GUIWorld neural computer demo 19\" width=\"210\">\u003C\u002Ftd>\n  \u003C\u002Ftr>\n\u003C\u002Ftable>\n\n###  Acknowledge \n\n\nOur models (not released here) are built from [Wan2.1](https:\u002F\u002Fgithub.com\u002FWan-Video\u002FWan2.1) and [Matrix-Game-2](https:\u002F\u002Fgithub.com\u002FSkyworkAI\u002FMatrix-Game). The data engine for CLIGen (General) is built from [Asciinema](https:\u002F\u002Fgithub.com\u002Fsearch?q=asciinema&type=repositories), the data engine for CLIGen (Clean) is from [VHS](https:\u002F\u002Fgithub.com\u002Fsearch?q=vhs&type=repositories), the data engine for GUIWorld (Random) is modified directly from [Neural-OS](https:\u002F\u002Fneural-os.com\u002F), and the data engine for GUIWorld (CUA) is built according to [Claude CUA](https:\u002F\u002Fgithub.com\u002Fanthropics\u002Fclaude-quickstarts\u002Ftree\u002Fmain\u002Fcomputer-use-demo).\n\n| Data Engine | Source |\n|---|---|\n| CLIGen (General) | [Asciinema](https:\u002F\u002Fgithub.com\u002Fsearch?q=asciinema&type=repositories) |\n| CLIGen (Clean) | [VHS](https:\u002F\u002Fgithub.com\u002Fsearch?q=vhs&type=repositories) |\n| GUIWorld (Random) | [Neural-OS](https:\u002F\u002Fneural-os.com\u002F) |\n| GUIWorld (CUA) | [Claude CUA](https:\u002F\u002Fgithub.com\u002Fanthropics\u002Fclaude-quickstarts\u002Ftree\u002Fmain\u002Fcomputer-use-demo) |\n","NeuralComputer 是一个用于构建神经计算机数据引擎的项目。它利用神经网络将计算、内存和I\u002FO统一在一个单一的运行时状态中，旨在通过从收集到的I\u002FO轨迹中学习可执行动态来实现完全神经计算机（CNC）的目标，而无需访问工具化的程序状态。该项目支持CLI和GUI轨迹生成，并提供了多种快速启动脚本以帮助用户上手。适用于需要探索新型计算架构的研究人员以及对神经网络在系统级应用感兴趣的开发者。","2026-06-11 02:45:06","CREATED_QUERY"]