What is Retrieval Augmented Generation (RAG)?
Вставка
- Опубліковано 28 вер 2024
- Day 5: 10 Days of GenAI: An into to RAG? Retrieval Augmented Generation explained.
In this video, you'll learn about Retrieval-Augmented Generation (RAG), an AI technique that combines large language models with external knowledge bases to improve the accuracy and reliability of LLM-generated text. Discover how RAG works and how it can be used to enhance AI applications.
#10daysofgenAI #generativeai #google #ai #rag
Priyanka, appreciate your efforts around the Ongoing GenAI series !!
Thanks!! I am glad you are finding it valuable.
i have this weird idea of RAG to be kind of waste... but as you're explaining, it could be really useful for specific cases (i.e. "2 years warranty") so, not really a waste
LLM will only engage the RAG feature only if it doesn't have an answer to the prompt. If the same prompt is asked again, there is no need for RAG as the LLM has learned the relevant information earlier. Is this understanding correct?
Yes that is correct. You can design the system such that if the answer is clear from the LLM don’t engage with RAG otherwise include. You could also create a system where every time RAG is engaged. So up to you.