Introduction to Deep Learning and Language Models

Deep learning

We use the term artificial intelligence (AI) to refer to a machine's ability to imitate the useful functions of human intelligence. A subfield of AI known as machine learning (ML) comprises a set of algorithms that allow computers to learn and improve their performance without explicit instructions.

Deep learning is a type of ML that tries to make computers learn and understand things like how our brains learn and understand. It uses special algorithms (mathematical instructions) to mimic our brain's neurons to teach computers how to recognize patterns and make decisions on their own. This artificial recreation of the brain's neuronal structure is aptly known as Artificial Neural Network (ANN). ANNs allow computers to mimic the brain's learning process for various tasks by consuming large amounts of input data.

Imagine you want to teach a computer to identify different animals. You can show it many pictures of animals, and the computer will learn to recognize common features and differences between them. It will remember these patterns and use them to identify animals it hasn't seen before.

Natural language processing

Natural language processing (NLP) is a subfield of AI that teaches computers to understand and work with human language, just like how we communicate. It involves creating clever computer programs that can read, interpret, and generate text in a way that makes sense to humans. Computers don't naturally grasp words and sentences as we do, so NLP comes to the rescue by bridging that gap.

Large language models

Large language models (LLMs) are deep learning models having billions of learnable parameters (roughly controlled by layers of neurons) specifically designed for NLP, intensively trained on vast amounts of text data from books, articles, and websites. This training equips them with the ability to learn language patterns and rules, allowing them to grasp and generate text that feels just like something a human would say.