Adding more Large Language Models (LLMs) to your system can significantly enhance the capabilities and flexibility of your local AI applications. Whether you are working on customer support, research, or creative tasks, having access to multiple models can improve the quality and relevance of the AI-generated content. This guide will walk you through the process of adding new LLM models to your system, ensuring that you can choose the best model for each task.
The first step in adding a new LLM is selecting the right model. There are various sources from which you can choose. One we like to use is HuggingFace.
Go to the HuggingFace hub and seek out models there. (https://huggingface.co/models?pipeline_tag=text-generation&library=gguf&apps=llama.cpp&sort=downloads)
Be sure to select models that fit within your RAM and VRAM limits. Know that some models are much better than others, so experimenting, or reading through forum posts, may help in your search for models that best suit you. Make sure the LLM models are GGUF files.
Once you have selected an LLM, download it, move or save it to the “models/llm” folder of your MAGAN.AI drive.
Your public interactions on platforms like ChatGPT aren't hidden from search engines. Google and others index these interactions, making them a part of your digital footprint, accessible to anyone with your online identifier.
You can either restart MAGAN.AI or hit the “refresh” button next to the Models dropdown in the MAGAN.AI Chat interface. Use the MAGAN.AI interface to configure settings such as context window size, GPU layers, and VRAM usage. These settings can significantly impact performance.
Adding more LLM models to your system can greatly enhance the capabilities of your AI applications. By following these steps, you can easily integrate and manage multiple LLM models, ensuring that you have the right tools for every task. No matter whether it’s for work or play, you can ensure your work stays private with local AI like MAGAN.AI.
To learn how to add more Image Models to MAGAN.AI, read Harnessing AI Art: Downloading Image Models and LoRA for MAGAN.AI.
