Michele LaurelliBlog

LLM (Large Language Model)

/ɛl ɛl ɛm/

A neural network with billions of parameters trained on massive text datasets to understand and generate human language.

In italiano: LLM (Large Language Model)

LLMs like GPT-4, Claude, and LLaMA are trained on diverse text from the internet. They demonstrate emergent abilities like reasoning, few-shot learning, and task generalization. Scale is key - larger models show better performance.

Examples

  • GPT-4 with 1.76 trillion parameters
  • LLaMA 2 for open-source applications
  • Claude for long-context understanding