What is Retrieval Augmented Generation (RAG) - Augmenting LLMs with a memory

► Jump on our free RAG course from the Gen AI 360 Foundational Model Certification (Built in collaboration with Activeloop, Towards AI, and the Intel Disruptor Initiative): learn.activeloop.ai/courses/rag
►Twitter: / whats_ai
►My Newsletter (My AI updates and news clearly explained): louisbouchard.substack.com/
►Support me on Patreon: / whatsai
►Join Our AI Discord: / discord
How to start in AI/ML - A Complete Guide:
►www.louisbouchard.ai/learnai/
Become a member of the KZbin community, support my work and get a cool Discord role :
/ @whatsai
#ai #llm #rag

[@letseat3553]

RAG is just 'full text indexing' on the local data with the ranked results fed into the context window and sent to the LLM along with the question. Every time I see it described as something of a database guy for the last 30 years all I see are new words describing long solved problems.

[@rajeshbasnet4454]

You mean like how elastic search does indexing ?

[@ahmedzouaoui8177]

Well new cars have wheels which is a technology that has thousands of years of existence. It does not mean that new cars are 'obsolete' but using an old tech to improve a new one is a great way of doing engineering !

[@Parsley1965]

Truly excellent video!

[@helainz7198]

Et cetera bien sur mon poto

[@user-oh4jz9zu5v]

Now I understood, What is RAG - Retrieval Augmented Generation ,Very Informative Video, Liked your Video 👍

[@Kama45]

Subbed

[@sabriboubaker]

Great video, straight to the point. Thanks again

[@WhatsAI]

Thank you Sabri! :)

[@Plink2120]

Vraiment clair et précis merci

[@finn_the_dog]

Great video. Would you make a video the different types of RAGs? Or how to prepare data for a RAG, for example when your document has tables, math formulas, references to images, I haven't seen much content about how to handle diverse data inside a document with RAGs. Cheers

[@WhatsAI]

Great idea, thank you! Will definitely look into multi modal RAG! :)

[@prattipatimanojsai]

Very Informative and useful!! Thanks

[@bhanujinaidu]

Thanks , very clear excellent explanation

[@WhatsAI]

Thank you! :)

[@chairwood]

thx. i enjoyed this video

[@WhatsAI]

Glad to hear so my friend! 😊

[@MK-ce7im]

I think this is the best video I have seen on this topic. Wanted to ask if we can use RAG offline maybe with Mistral model ?

[@WhatsAI]

Of course you can host everything locally if you have the capacity! :)

[@martinkrueger937]

by any chance do you know which RAG system/framework is giving out the best performance?

[@WhatsAI]

From our work we like to use llamaindex for many parts and adapt on our own code for more personalized settings!

[@rhans6598]

Thanks but what's the point of sound effects?

[@Mr_Arun_Raj]

After integrating with RAG. latency increased....

[@WhatsAI]

That is for sure! There is some downsides but the latency if very little.

[@paulwillisorg]

The accent of the speaker is pretty heavy.

[@WhatsAI]

Hope it’s still easy to understand!

[@kunjs]

google launched gemini advanced 1.5, a RAG killer 💀

[@WhatsAI]

A database can be much larger than this context window and much more efficient I believe. It’s unsure how good the models are vs gpt4 yet. Plus, sending millions of tokens for every prompt will be extremely expensive for each request, haha! It’s good for some use cases like sending a full repo once and asking questions but not for working with customers and handling many requests I believe.