Navigate Chats, Collections, and Documents
- Generative AI
- Enterprise h2oGPTe
This lesson walks through h2oGPTe's three main elements, Chats, Collections, and Documents, starting with the chat settings where you pick from a regularly updated list of commercial and open-source language models. It demos asking a general knowledge question about the plot of the musical Hamilton without any uploaded documents, so the model answers purely from what it already knows. It shows that switching between available LLMs changes the level of detail in the response, sometimes more and sometimes less. It sets up the next lesson's comparison between this plain LLM answer and one grounded in retrieved reference material.
Key takeaways
- What the three main elements of h2oGPTe are: Chats, Collections, and Documents
- How chat settings let you pick from a regularly updated list of commercial and open-source LLMs
- Why asking a general knowledge question without documents draws only on the model's own training data
- The same question can get more or less detailed answers depending on which LLM is selected
Start owning your AI