RAG => Retrieval Augmented Generation

Why?

well, LLM models are trained on the general public data available on the internet. Most of the worlds data is private (95%). To get answers from the LLM models specific to your need is or from data source you have is important. Considering the Token limit we have, All the data can't be provided in single shot. Here comes the RAG, which helps you to filter the data you need to get the answer you need.

Terms & Techniques :

1. Langchain => software framework, which helps to use LLM in to An Applications.
2. Agents => LLMs just output the text, they can perform any actions. Agent do this action job.
3. Embbeding Model => embbed the chunks
4. zero/single/multi-shot => ?

Query Translation

Translate the question into a form which is better suited for retrieval of information.

Chunking Based on the content.