AI glossary
What is training data?
Short answer
The examples a model learns from. Its quality, size and biases shape everything the model does.
For example
A speech recogniser trained mostly on one accent will usually struggle with others, because those accents were rare in its training data.
Learn it in a free lesson
- Using AI wisely: privacy, bias, judgementBeginner · 5 min
- Overfitting & the Goldilocks modelIntermediate · 5 min
- How we test language modelsIntermediate · 7 min
- How machines learnBeginner · 5 min
- How LLMs are trainedAdvanced · 7 min
- Scaling laws, evals & safetyAdvanced · 7 min
Related terms
Questions
What is training data in simple terms?
The examples a model learns from. Its quality, size and biases shape everything the model does.
What is an example of training data?
A speech recogniser trained mostly on one accent will usually struggle with others, because those accents were rare in its training data.
Where can I learn about training data for free?
AI in Bed explains it in the free lesson "Using AI wisely: privacy, bias, judgement" (Beginner level, 5 minutes), with no sign-up.