Skip to main content

What's a LLM?

Large Language Models (LLMs) are a type of program taught to understand text and respond back as a human would. They learn by being fed large amounts of text and they try to predict what the answer will be based on the provided information. The model's ability to understand and respond correctly is largely dependent on what information was used to teach it and how big the model is.

Generally, the larger models, the easier it is for it to learn new concepts, but requires a more powerful computer to run which can be a limitation when running on a desktop computer. There are many different use cases for having a technology that can understand text and provide a response.

Use Cases

With LLMs being a newer technology, new ideas are tested every day to learn what's possible. There are many possibilities of what it can do that are largely unexplored. Listed below are a few examples of what it could be used for as of today.

Writing Code

Give the LLM a short text description of what you want to be done and in what programing language and it will write it for you. Currently, there are limits on how much information it can read and write at once, so while it can handle writing small functions very well, it struggles when it comes to piecing multiple parts of code together.

Summarizing Text

One of the skills LLM's can do best. Provide it some text, and ask it to rewrite it in a particular style or format and it will give you an answer. These docs were written using ChatGPT as an editor to make sure the content is clear!

Question Answering

LLM's have a broad amount of general knowledge and can provide clear and concise answers about lots of topics. For many questions require domain-specific knowledge or recent information, because they only know information based on what they were taught.

Running a LLM Locally

While using services like ChatGPT can be convenient and give good answers, there are privacy concerns. For many use cases, you do not want to be giving a company your private information such as medical records, or a company's intellectual property. If you are running the model locally, that means you are able to maintain control of your data without sharing it with anyone else.