This page gathers links for learning large language models. The first entries cover memory and knowledge for LLMs, followed by beginner courses and tours of LangChain. A large group of links deals with retrieval augmented generation, from building a RAG pipeline to adding function calling. Prompt engineering appears through survey material, OpenAI strategies and a Vietnamese explainer. Fine tuning has a walkthrough with Unsloth. Model internals come through paper reading on Mixtral. The last section collects math explanations for Mixture of Experts, Grouped Query Attention and Flash Attention. Most entries are articles, so treat the page as a reading queue rather than a course.
-
Tang Jie: Advanced Machine Learning (Thanh Hoa, 2026): a build-it-yourself pipeline running from a tokenizer and a 0.1B transformer, through Triton attention and scaling laws, to controlled SFT / DPO / RLVR comparisons and long-horizon agent tasks with a self-judge. Includes the judgments he shared at the opening session.
-
Learn Harness Engineering: WalkingLab's MIT course on the files, checks, and session rules around a coding agent. The English docs have 14 lectures and 8 projects. Checked 3 November 2026. The repo was not cloned and the lectures were not taken.
-
How to feed knowledge / memory to an LLM: https://viblo.asia/p/memory-in-llm-agent-n1j4lkwMVwl
-
Microsoft's 18-lesson LLM series for beginners: https://github.com/microsoft/generative-ai-for-beginners #course #LLM
-
Langchain #1 - A tour of the most powerful features of Langchain, an extremely capable framework for working with LLMs: https://viblo.asia/p/langchain-1-diem-qua-cac-chuc-nang-sung-so-nhat-cua-langchain-mot-framework-cuc-ba-dao-khi-lam-viec-voi-llm-BQyJKmrqVMe #Langchain
-
LangChain: An open-source framework for developing applications built on large language models (LLMs), letting you build complex AI applications by combining different modules - GitHub #Langchain #LLM #framework
-
Retrieval-Augmented Generation: An indispensable method for deploying real-world LLM projects!: https://viblo.asia/p/retrieval-augmented-generation-phuong-phap-khong-the-thieu-khi-trien-khai-cac-du-an-llm-trong-thuc-te-phan-1-Ny0VG7yzVPA #RAG
-
Paper Explain - Mixtral of Experts: many teachers make a strong model: https://viblo.asia/p/paper-explain-mixtral-of-experts-lam-thay-thi-model-khoe-EvbLbQ8oJnk #Mixtral
-
Retrieval-Augmented Generation: An indispensable method for deploying real-world LLM projects! (Part 1): https://viblo.asia/p/retrieval-augmented-generation-phuong-phap-khong-the-thieu-khi-trien-khai-cac-du-an-llm-trong-thuc-te-phan-1-Ny0VG7yzVPA #RAG
-
The process of building a RAG system with Function Calling: https://viblo.asia/p/quy-trinh-xay-dung-he-thong-rag-tich-hop-function-calling-with-source-code-vlZL98GZJQK #RAG #functionCalling
-
Scrapegraph-ai #1: Using the power of LLMs to solve data collection and processing for AI systems: https://viblo.asia/p/scrapegraph-ai-1-su-dung-suc-manh-cua-llms-de-giai-quyet-bai-toan-thu-thap-va-xu-ly-du-lieu-cho-cac-he-thong-ai-EvbLbavPJnk #scrapegraph
-
Techniques for creating and using prompts in LLM models: https://trigaten.github.io/Prompt_Survey_Site #promptEngineering
-
What do you know about prompt engineering?: https://viblo.asia/s/ban-da-biet-gi-ve-prompt-engineering-EbNVQNO24vR #promptEngineering
-
Lesson 2 - prompt-engineering, RAG and Finetuning
-
Using operator with the Deepseek API + WebUI: https://www.facebook.com/Passer.maker/videos/601578342493962/?idorvanity=3262256010653927 #Deepseek #API
-
Guide to Finetuning an LLM with Unsloth
- A detailed walkthrough of fine-tuning an LLM to build a chatbot that generates n8n workflows.
- Uses the Unsloth library and Google Colab to fine-tune for free.
- Data is stored on Hugging Face; uses an Alpaca-style prompt format. #finetune #LLM #Unsloth
-
6 Effective Prompt Strategies from OpenAI – Part 1
- Write clear and detailed instructions.
- Ask the model to adopt a specific persona (e.g., teacher, expert).
- Use delimiters to separate the parts of a prompt.
- Specify the concrete steps needed to complete the task.
- Provide illustrative examples.
- Specify the desired format or characteristics of the output. #promptEngineering #OpenAI
-
6 Effective Prompt Strategies from OpenAI – Part 2
- Instruct the model to reason step by step before reaching a conclusion.
- Use an "inner monologue" to separate the reasoning process from the displayed result.
- Apply "query chaining" to split a task into separate steps. #promptEngineering #OpenAI
-
Models.dev — An open-source database of AI models - Detailed specifications of popular LLMs
Algorithms
See also: Learning Resources · Prompt Strategies · ChatGPT Prompting