Skip to content

About

About This Documentation

This documentation site is a collection of personal learning notes on Large Language Models, AI Agents, and related topics. The notes are organized using a "topic online clustering" approach, where content is added whenever interesting insights are discovered.

The current collection covers LLM fundamentals, AI agent frameworks, retrieval-augmented generation, prompt engineering, reinforcement learning and fine-tuning, LLM evaluation, model compression, recommendation systems, NeurIPS talks, Google GenAI course notes, Hugging Face agent notes, and Hugging Face fine-tuning notes.

Note-Taking Philosophy

"I take notes whenever I come across something interesting and use a 'topic online clustering' approach to group them."

This means:

  • Real-time capture: Notes are added as I learn, not in a rigid structure
  • Organic organization: Topics emerge and evolve naturally
  • Personal focus: Content reflects my learning journey and interests
  • Continuous evolution: The documentation grows and improves over time

Content Sources

Notes are gathered from various sources including:

  • Research papers and academic publications
  • Online courses and tutorials
  • Blog posts and technical articles
  • Hands-on experimentation
  • Community discussions

Topic Coverage

The main searchable topics include:

  • Large language models, language model fundamentals, Transformers, BERT, GPT, attention, self-attention, encoder-decoder models, RNNs, LSTMs, sequence-to-sequence models, and perplexity
  • AI agents, agentic AI, tool use, planning, memory, workflow automation, browser agents, software agents, coding agents, multi-agent systems, AutoGen, CrewAI, LangGraph, smolagents, LAMBDA, and TaskWeaver
  • RAG, retrieval-augmented generation, Graph RAG, Golden-Retriever, embeddings, vector search, vector databases, multimodal embeddings, and LlamaIndex
  • RLHF, reinforcement learning, fine-tuning, post-training, instruction tuning, supervised fine-tuning, preference optimization, reward modeling, alignment, LIMA, SFTTrainer, LoRA, and chat templates
  • LLM evaluation, LLM-as-judges, benchmarks, safety evaluation, alignment evaluation, explainable AI, responsible AI, unlearning, model compression, quantization, pruning, and knowledge distillation

Disclaimer

These notes may not be: - Perfectly organized or structured - Comprehensive or complete - Suitable for all learning styles - Free from personal bias or interpretation

They are intended as a personal knowledge base that might be useful to others exploring similar topics.

Contributing

While these are personal notes, feedback is welcome! If you find errors or have suggestions:

  1. Open an issue on GitHub
  2. Submit a pull request
  3. Reach out directly

License

These notes are shared for educational purposes. Please check individual sources for their respective licenses.

Happy learning!