How LLMs work and fail
TYPO3
Show in appDescription
A large language model produces text by predicting the next likely token based on patterns in its training data, it does not look up answers in a database. This distinction separates a generative approach from retrieval-based approaches such as search engines or knowledge bases, and explains why a fluent LLM response is not evidence that the information is correct.
Goals
- I know that an LLM produces text by predicting likely continuations, not by looking up answers.
- I know the difference between a generative approach and a retrieval-based approach.
- I am aware that fluency of an LLM response is not an indicator of correctness.
- I can explain what a training cut-off is and how it limits an LLM's knowledge of recent events.
Maintainer
Korbinian Kugelmann
queo München AG
