• A large language model (LLM) is just two documents.
  • Parameters.
  • Run file that runs those parameters.
    • Can be in any programming language but C is often used because of its simplicity.
• The magic is in the parameters.
  • It is basically a compression of the internet. You have lots of data from the internet.
  • A GPU cluster intended for heavy workloads is required which runs for several days to compress the large amount of data into parameters which is comparatively a very small file.
  • This is a lossy compression.
• Neural network
  • Tries to predict the next word in a sequence. This prediction forces the neural network to learn a lot about the world.
  • The network dreams things based on what it was trained on.

Why are LLMs revolutionary?

• Enabled seemingly impossible possibilities
• Generates texts beyond average human writing ability.
• Demonstrate human-like complex reasoning and understanding patterns.
• Versatile on many language tasks altogether.
• Unprecedented crossover with society and other fields.