What is the distinction between pre-trained and grounding data in GPT

Feb 22, 2023

These LLMs are pre-trained on a broad dataset from the internet, they can’t provide accurate domain-specific knowledge out of the box. Remember, LLMs on their own don’t do an internet search or have access to internal databases or files on your computer every time they generate a response.

Solving this data gap is one of the primary challenges to the responsible deployment of LLM into products and will be major focus for most teams.

We need to work on the rich data that we have in the Knowledge Bases.