[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-80823":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":8,"htmlUrl":8,"language":9,"languages":8,"totalLinesOfCode":8,"stars":10,"forks":11,"watchers":12,"openIssues":13,"contributorsCount":13,"subscribersCount":13,"size":13,"stars1d":12,"stars7d":14,"stars30d":15,"stars90d":13,"forks30d":13,"starsTrendScore":14,"compositeScore":16,"rankGlobal":8,"rankLanguage":8,"license":17,"archived":18,"fork":18,"defaultBranch":19,"hasWiki":20,"hasPages":18,"topics":21,"createdAt":8,"pushedAt":8,"updatedAt":22,"readmeContent":23,"aiSummary":24,"trendingCount":13,"starSnapshotCount":13,"syncStatus":25,"lastSyncTime":26,"discoverSource":27},80823,"large-language-model","rahul238xaviers\u002Flarge-language-model","rahul238xaviers",null,"Python",41,8,1,0,3,4,43.76,"Apache License 2.0",false,"main",true,[],"2026-06-12 04:01:30","# 🦀 Large Language Model from Scratch (Apple Silicon MLX)\n\nThis repository houses a custom **1.6B parameter decoder-only GPT model** optimized for Rust code completion, built and trained on Apple Silicon via MLX.\n\n---\n\n## 📖 The Training Journey Book\n\nWe documented our entire engineering path—from architectural layout and resolving 500GB+ OOM crashes on an M3 Ultra to GQA acceleration and implementing advanced decoding engines—in a comprehensive, book-style documentation structure inside the `doc` folder:\n\n👉 **[Read the Full Rust-GPT LLM Training Journey Book (Chapter-by-Chapter)](doc\u002Ftraining_journey.md)**\n\n*   **[Chapter 1: The Architectural Blueprint](doc\u002Fchapter1_architecture.md)**\n*   **[Chapter 2: The M3 Ultra & The OOM Crash](doc\u002Fchapter2_oom_crash.md)**\n*   **[Chapter 3: Stabilization & Memory Control](doc\u002Fchapter3_stabilization.md)**\n*   **[Chapter 4: Hardware Optimization & Scaling](doc\u002Fchapter4_hardware_acceleration.md)**\n*   **[Chapter 5: The Repetition Crisis & Decoding Engine](doc\u002Fchapter5_decoding_upgrades.md)**\n*   **[Chapter 6: The Interactive Playground UI](doc\u002Fchapter6_developer_playground.md)**\n\n---\n\n## 🚀 Quick Setup & Playground Launch\n\n### 1. Requirements & Setup\nEnsure you are using **Python 3.10+** on Apple Silicon (M-series processor recommended) and install dependencies inside your virtual environment:\n\n```bash\ncd apple-silicon\npython3 -m venv .apple_env\nsource .apple_env\u002Fbin\u002Factivate\npip install -r requirements.txt\n```\n\n### 2. Configure Checkpoint Location\nSet the `CHECKPOINT_PATH` inside a `.env` file at the root of `apple-silicon\u002F`:\n\n```env\nCHECKPOINT_PATH=runs\u002Frun_20260514_183932\u002Fcheckpoints\u002Fstep_001000.safetensors\n```\n\n### 3. Launch the Developer Playground\nLaunch the Gradio 6.0 playground workspace:\n\n```bash\npython3 tests\u002Ffunctional\u002Fgradio_app.py\n```\n\nOpen `http:\u002F\u002Flocalhost:7860` in your web browser to generate Rust code completions in real time with built-in copy-to-clipboard fallbacks!\n\n---\n\n*Educational project inspired by Sebastian Raschka's \"Large Language Models from Scratch\" and fully scaled to 1.6 Billion parameters on Metal\u002FApple Silicon.*\n","该项目是一个从零开始构建的1.6B参数解码器仅GPT模型，专为Rust代码补全而优化，并在Apple Silicon平台上通过MLX进行训练。其核心功能包括针对Rust语言的高效代码生成与补全，以及详细的开发文档记录了从架构设计到解决内存溢出问题等整个工程过程。技术特点涵盖硬件加速、高级解码引擎实现等方面。适合于希望深入了解大型语言模型构建过程的研究者或开发者，特别是对利用Apple Silicon平台进行机器学习项目感兴趣的人士。",2,"2026-06-11 04:02:27","CREATED_QUERY"]