{"id":16907396,"url":"https://github.com/keli-wen/agi-study","last_synced_at":"2025-04-11T15:41:08.270Z","repository":{"id":218943511,"uuid":"746637073","full_name":"keli-wen/AGI-Study","owner":"keli-wen","description":"The blog, read report and code example for AGI/LLM related knowledge.","archived":false,"fork":false,"pushed_at":"2025-02-01T09:14:30.000Z","size":20441,"stargazers_count":36,"open_issues_count":0,"forks_count":2,"subscribers_count":3,"default_branch":"master","last_synced_at":"2025-04-09T17:17:23.616Z","etag":null,"topics":["code-examples","demo","inference-optimization","llm","train"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/keli-wen.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null}},"created_at":"2024-01-22T12:10:48.000Z","updated_at":"2025-03-18T02:54:09.000Z","dependencies_parsed_at":"2024-02-05T17:01:52.682Z","dependency_job_id":"73b93922-9682-46fa-b750-62484c7864a8","html_url":"https://github.com/keli-wen/AGI-Study","commit_stats":null,"previous_names":["keli-wen/llm-study"],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/keli-wen%2FAGI-Study","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/keli-wen%2FAGI-Study/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/keli-wen%2FAGI-Study/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/keli-wen%2FAGI-Study/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/keli-wen","download_url":"https://codeload.github.com/keli-wen/AGI-Study/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":248432940,"owners_count":21102472,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["code-examples","demo","inference-optimization","llm","train"],"created_at":"2024-10-13T18:47:15.483Z","updated_at":"2025-04-11T15:41:08.244Z","avatar_url":"https://github.com/keli-wen.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# AGI-Study\n\n\n🎯 Be a good Deep Learning Engineer. (大量施工👷)\n\n## Code-Examples\n\n- [x] [`chat-llm-v1`](https://github.com/keli-wen/AGI-Study/tree/master/code-examples/chat-llm-v1)：基于 `PyTriton`，`Streamlit` 和 `DeepSeek` 制作的最简化 Chat Project。\n- [ ] `chat-llm-v2`：基于 `chat-llm-v1` 制作的 `vision language` 版本，并优化了多模型选择，dynamic batching 和 streaming output 等新特性。（施工中）\n\n## 🔥 LLM Dev Best-Practice\n\n由于我认为 LLM Dev 才是我等普通人能做的事情，我最近在全力学习一些 Agentic System / RAG / Prompt Engineering 的最佳实践（基于 OpenAI / Anthropic / Google 等公司的技术博客），以及如何从 experimental 到 production 的最佳实践。这部分预计包括：\n\n- [ ] **Agentic System**：关于如何构建有效的 Agent 的最佳实践。\n  - [x] [Anthropic - Building Effective Agents](best-practice/Anthropic%20-%20Building%20effective%20agents/README.md) \n- [ ] **RAG**：如何设计和优化 RAG 的最佳实践。\n- [ ] **Prompt Engineering**：如何设计和优化 Prompt 的最佳实践。\n- [ ] **LLM Dev**：如何从实验到生产的最佳实践。\n\n## Environment\n\n\u003e 这部分主要介绍 DL 环境配置相关的内容。\n\n- [x] [**CUDA** Related Env Config](https://github.com/keli-wen/AGI-Study/blob/master/env/cuda-related/)：介绍 GPU Driver Version，Cuda Toolkit Version 的更新。包括多 Cuda 版本管理等。\n- [ ] [**Docker** Related Env Config](https://github.com/keli-wen/AGI-Study/blob/master/env/docker-related/)：Docker 的基本使用教程（菜鸟教程）。\n\n## Train\n\n\u003e 这部分主要介绍当前 LLM 中常用的 Training 框架以及相关知识点。\n\n- [ ] `PYTORCH LIGHTNING` 入门介绍（低优先级）\n- [ ] DeepSpeed 介绍：\n  - [ ] DeepSpeed -- ZeRO 原理介绍（见知乎，待搬运）。\n  - [ ] DeepSpeed 实战（环境配置，Example）（TODO，Low Priority）[Refer: DeepSpeed PR](https://github.com/microsoft/DeepSpeedExamples/pull/843).\n\n## Tokenizer\n- [x] Byte-Pair Encoding 算法解读。\n- [ ] Google SentencePiece 库使用介绍。\n\n## Inference \u0026 Deploy\n\n\u003e 这部分主要介绍推理优化和部署相关的内容。\n\u003e\n\u003e - **🤔Q: What's the Inference Optimization?**\n\u003e - **📖A:** Inference optimization refers to **the process of enhancing the efficiency and speed at which LLMs analyze data and generate responses**. This process is crucial for practical applications, as it directly impacts the model's performance and usability.\n\n- [x] [`Basic-LLM-Inference.md`](https://github.com/keli-wen/AGI-Study/blob/master/inference/Basic-LLM-Inference.md)：基于 meta-llama 介绍基础的 LLM Inference pipeline。\n- [ ] `Batch-Inference-Optimization.md`：（施工中）Basic 的进阶版。\n- [ ] `vLLM`: （施工中）介绍 `vLLM` 的使用，**以及后续的 `vLLM` 核心原理和代码的探索。**\n- [ ] `TensorRT-LLM`：目前是非常简单的介绍了 `TensorRT-LLM` 的使用信息。\n- [x] `Mixture of Depth`：关于 MoD 的最新介绍，Transformer-based 模型的动态算力分配。\n- [ ] `Nvidia Triton Inference Server`：首先进行工具扫盲，然后主要从应用的角度介绍这个工具的使用。\n- [ ] `Quantization in LLM`：（施工中） \n\n## Demo\n\n\u003e 这部分主要介绍 DEMO 制作相关的经验。\n\n- [x] `FastAPI`: 介绍 `FastAPI` 的基本信息，以及它如何应用在 LLM 相关的 DEMO 原型中。\n- [ ] `Streamlit`：介绍如何 `Streamlit` 如何使用，并定制化自己的 DEMO 前端。\n\n## Visualization\n\n开源一些可视化的资源。\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fkeli-wen%2Fagi-study","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fkeli-wen%2Fagi-study","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fkeli-wen%2Fagi-study/lists"}