Articles filed in category 'Large Language Models'

  • Fine Tuning Large Language Models, Part 1

    By: Wei-Meng Lee

    Last updated: Saturday, August 22, 2026

    Published in: CODE Magazine: 2026 - September/October

    Wei-Meng explains that “fine-tuning” covers a spectrum of adaptation methods, not one technique, and argues beginners should choose based on task, data, and hardware rather than defaulting to full training. He walks through feature extraction, partial fine-tuning, full fine-tuning, and LoRA, showing how each changes how many parameters are updated, what it costs, and when it is useful. Using Hugging Face tooling and runnable examples, Lee emphasizes that most real projects should start with prompting or feature extraction and move to LoRA before considering full fine-tuning.

  • Local Vibe Coding with Gemma 4 12B

    By: Sahil Malik

    Last updated: Monday, August 17, 2026

    Published in: CODE Magazine: 2026 - September/October

    Sahil Malik suggests that local AI models like Gemma 4 12B make “vibe coding” practical by combining natural-language programming with privacy, speed, no recurring API cost, and customizable context windows. He shows how to run Gemma locally with Ollama in VS Code and Python, then build a simple developer assistant that reads a workspace, accepts a plain-English request, and generates or improves code offline.

  • Running Large Language Models Locally Using Ollama

    By: Wei-Meng Lee

    Last updated: Wednesday, March 4, 2026

    Published in: CODE Magazine: 2026 - March/April

    Wei-Meng Lee explores the advantages of running large language models (LLMs) directly on personal hardware using Ollama, a platform designed to simplify local deployment and management of AI models. He highlights the benefits of local LLM usage, including enhanced data privacy, reduced costs, offline accessibility, and improved control over model performance and updates—advantages that cloud-based services may not always provide. The article offers a step-by-step guide on using Ollama's desktop and command-line interfaces, integrating models from Hugging Face, and customizing model behavior through Modelfiles. Wei-Meng also delves into hardware requirements, storage management, and the process of converting models to GGUF format, providing developers with practical insights for harnessing the power of LLMs locally while maintaining data ownership and operational flexibility.

  • Natural Language AI-Powered Smart UI

    By: Ed Charbeneau

    Last updated: Friday, December 26, 2025

    Published in: CODE Magazine: 2025 - July/August

    Ed Charbeneau delves into the practical integration of AI-powered Smart UI components by leveraging large language models (LLMs), particularly moving beyond the prevalent Generative AI chat demos that often overshadow such applications. He illustrates the gradual enhancement of user interfaces (UIs) through LLMs with real-world examples like Smart AI Search and Smart Text, which improve user interactions by adding contextual and relational features to existing functionalities. Utilizing LLMs, Charbeneau demonstrates transforming user queries into actionable commands for UIs, employing natural language processing to bridge the gap between human language and software interfaces, thereby enhancing user experience while maintaining familiar UI elements.

  • Exploring LangChain: A Practical Approach to Language Models and Retrieval-Augmented Generation (RAG)

    By: Wei-Meng Lee

    Last updated: Friday, December 26, 2025

    Published in: CODE Magazine: 2025 - January/February

    Wei-Meng Lee provides an in-depth guide to using the LangChain framework for building applications incorporating large language models (LLMs). He emphasizes LangChain's modular design, enabling developers to create complex workflows with customizable components and integrate external data sources, facilitating Retrieval-Augmented Generation. Key topics covered include constructing chains, maintaining conversational context with memory management, and leveraging Microsoft and Hugging Face's models for enhanced flexibility and privacy. Wei-Meng also demonstrates implementing RAG for document-based querying, offering a comprehensive overview of LangChain's capabilities for developing dynamic, data-driven solutions.