A Guide to Self-Hosted LLM Coding Assistants

Assistive coding utilizing large language models (LLMs) significantly enhances productivity by integrating advanced models into development environments. While hosted models-as-a-service have become more accessible, self-hosting LLMs offers benefits such as increased privacy, cost efficiency, and staying current with new developments. The article provides a comprehensive guide on setting up and integrating self-hosted LLMs, using Ollama as an example, which supports a range of models like codeqwen, deepseek-coder, codellama, and llama3.1, each with unique capabilities for coding tasks. Emphasizing the importance of editor integration, it explores the inclusion of these models into various editors like VSCode, Emacs, and Neovim, leveraging the Ollama API for seamless code completion. By evaluating different LLMs and their integration into development tools, the guide demonstrates how to effectively use these models for enhanced code generation and completion.