Hey guys really noob question I have

At a glance

Hey guys really noob question, I have built a small discord bot that answers questions about a certain topic that I train him to. My problem is that if the question is out of his context he says he doesn't know. Could I merge his knowledge from my dataset with the public Chatgpt3.5 turbo dataset and query both every time I ask him something?

7 comments

JJoe Reuter @ Airbyte

This sounds like the retrieval-augmented generation pattern matches your use case: https://docs.aws.amazon.com/sagemaker/latest/dg/jumpstart-foundation-models-customize-rag.html

JJoe Reuter @ Airbyte

llama index can help with that using vector stores

CChyaboiii

Hey thanks for your help! Just to be clear there is no way to do that just using llama Index right? I have to use RAG?

JJoe Reuter @ Airbyte

RAG is just the general pattern - I'm not sure how llama index implements it but in langchain it's called the RetrievalQA chain: https://python.langchain.com/docs/modules/chains/popular/vector_db_qa

LLogan M

In llama index, RAG is implemented with our indexes / query engines

LLogan M

it's pretty easy

LLogan M

definitely check out our docs, there's lots of guides to the overall concept, how to use llama index, and how to customize things.

Some useful links

Installation: https://gpt-index.readthedocs.io/en/latest/getting_started/installation.html

Starter Tutorial (create an index and query it) https://gpt-index.readthedocs.io/en/latest/getting_started/starter_example.html

For your use case, you probably want multiple indexes to query from. A data agent makes sense to use here
https://gpt-index.readthedocs.io/en/latest/core_modules/agent_modules/agents/usage_pattern.html#query-engine-tools

Add a reply

Find answers from the community

Hey guys really noob question I have