Large Language Models (LLMs)

Deep Learning

Deep neural networks – almost always based on the Transformer architecture – trained on massive amounts of text data to understand and generate human language. LLMs like GPT-4, Claude, or Llama are characterized by their size (often hundreds of billions of parameters) and their ability to handle a wide range of language tasks with minimal task-specific training. The Transformer architecture by Vaswani et al. (2017) made this scaling possible – through self-attention instead of recurrence, enabling efficient parallelization and training on unprecedented data volumes.

Example

GPT-4 can write code, summarize texts, answer questions, and conduct dialogues – all with the same model, without separate specialization. This versatility emerges from training on trillions of words from the internet.

Used in

Sources

Back to the glossary