Training Trains Internship
Student login
← Natural Language Processing learning path
Natural Language Processing · Module 8 of 10

Large language models

Use and adapt LLMs for real tasks.

📖 Key ideas

  1. LLMs are huge Transformer models trained on vast text.
  2. They follow instructions, write, summarise, code and reason.
  3. Prompting guides behaviour; fine-tuning adapts style or tasks.
  4. RAG grounds answers in your documents.
  5. Limitations: hallucination, bias, knowledge cutoff, cost.
💡 Real-life example: An LLM is a well-read assistant who has read a giant library but can still misremember details.
LLMLarge Language Model
PromptInstruction given to an LLM
HallucinationConfident but false output
Context windowAmount of text the model can consider