Why Everyone is Freaking Out About RAG

[@TechWithTim]

You know what else people are freaking out about? My new course ;) techwithtim.net/dev

[@smithdoesstuff]

Nice segue!

[@quantumjun]

Love your video. Can you make a video on how do you use LLMs in your daily work :)

[@cha5848]

I have been learning from u since my second yr at college..now i am in my last sem currently doing an internship as jr prompt engineer...again i am here to learn from u..can't thank u enough❤

[@Lester_da_Molester]

A visual demo would have been useful

[@mohamedabdelrehem7698]

IBM explained it from like 4 months so so good

[@odnoco]

Where? ​@@mohamedabdelrehem7698

[@devloper_hs]

​@@mohamedabdelrehem7698 course

[@vektracaslermd743]

Thanks again for the video, Tim. This is a great topic that addresses the elephant in the room about LLMs... their out of date nature as soon as they're done training. Video suggestion (not that you don't have enough already), a tutorial on a local LLM setup, such as using Ollama and Mistral, then point it at a local database, such as a Wikipedia mini-download or other document, and use Langchain to automate some questions. I've been having a lot of fun with setting that up. Others might find it useful and fun. Anyway, thanks again!

[@hacktivatenation]

Gpt4 is up to march 2023 now. ;- )
RAG with finetuning are the keys to the future IMO!
Thanks for the content! You helped me learn pygame and gave me the real desire to keep building and start putting things out there.
Truly appreciated! 🙂

[@MakeDataUseful]

Exciting!!

[@GreatTaiwan]

is it RAG just inference done on a vector store ?
something like this existed as soon as as chatgpt were announced (look up langchain)

[@anasouardini]

1. to what point can you reduce the pre-acquired data such that the new data can still be understood?
2. how many pieces of "data" we need to put together in order to call it "logic"?
I'm sure this doesn't have to be accurate to make a working POC, just throwing my thoughts out there.

[@Afran146]

Your videos are always informative! Thank You!!

[@caty863]

Back in my bachelor program, I hated lecturers pestering me of giving sources of my statements in my papers...as if I was not allow to think of my own.
Now, it's funny how we are hating on LLMs for not providing sources of their answers.

[@Aceptron]

I would LOVE to see courses on Operating System Programming and Assembly Language on AlgoExpert!!!

[@Jackson141vja]

Somehow I always learn the most from your videos! Thanks for the excellent explanation

[@drexon88]

As I know, langchain already has something similar

[@idontevenwanttomakea]

I was thinking this too. Does anyone know how RAG is different?

[@idontevenwanttomakea]

How is this different from using Langchain?

[@polakpuckstopper]

Great stuff as always, Tim. PS what's the background track at 6:23 ?

[@jamesmashiyane6495]

I guess RAG can be used for downstream tasks, otherwise it goes against the whole point of AGI, we. must be able to train models or agents that can use their knowledge base and be able to reason without seeing any downstream test data.

[@colinmaharaj]

I use a commercial C++ compiler and I haven't updated in over 10 years. The responses I get from chat GPT is pretty good for what I need and what I have and some cases it's uses features in the C++ language that My compiler doesn't support.
So I have to tell it a times what type of language I need and downgrade the language

[@wonjunjang3839]

Still very useful information! Thank you Tim

[@dhrumil5977]

So we have the data set which is in the vector form with us but how is the process of sending the vectorised data works and how does it interact with the LLM it can only work with natural language right ?

[@wavyybased1820]

For the programming expert program do you offer tips for linkdin and how to land a job?

[@mohitv7405]

So basically a tool to use prompts on data store?

[@roderickshoolbraid]

What if the data store is flawed- how do I trust that?

[@haafiz27]

doesnt bing chat already do this? it provides sources and does live searches from the internet, afaik...

[@yorha.a2]

Also Bard

[@shubhrajit2117]

So it's like a search engine with llm in between

[@icris_studio]

Is flask (python framework) still in production?

[@MakeDataUseful]

Yep!

[@Gigusx]

Why ask that here? 🤔

[@icris_studio]

@@Gigusx It's my choice.

[@Gigusx]

@@icris_studio 🥱

[@antoniokasljevic]

Why not include visual examples instead of just video of your beautiful face mixed with random animated transitions back and forth. Content was great!

[@harshmalik3470]

What if my dataset is all of the internet for RAG lol?

[@michelchaman6495]

maybe im wrong isn't this what copilot does?

[@cxvxcbcxn]

What's an LLM ???

[@icris_studio]

It stands for Large Language Model a type of artificial intelligence (AI) program that can recognize and generate text, among other tasks

[@xXMaDGaMeR]

looking buff tim

[@TechWithTim]

💪

[@ifstatementifstatement2704]

Now it says it’s been trained on data up until 2022

[@Djellowman]

I don't see what time RAG can play in a choose your own adventure game.

[@Djellowman]

Time = role

[@pravachanpatra4012]

TIM. Can you make a tutorial on creating a neural network from scratch.

[@jacksins3353]

ua-cam.com/play/PLQVvvaa0QuDcjD5BAw2DxE6OF2tius3V3.html&si=7GWHlq2oYzriB7eT
Neural Networks from Scratch, Sentdex’s UA-cam channel. Long before I watched Tim, I watched this guy for Python tutorials.

[@DivisivelyIndivisible]

I’m @0:28 in and I’m seeing an issue. RAG in Agile and RAG in Waterfall have very different meanings.

[@Slaci-vl2io]

Can Wikipedia serve a RAG?

[@january9765]

Hey can u make a series on trading bot

[@akiratoriyama1320]

What does football have to do with nfl?!

[@NotesandPens-ro9wx]

Bro you are saying CHATGPT is out of date or have old information, but later you are promoting its book :P

[@vizionthing]

Not impressed with the datamining link for the book!

[@nikosafk5200]

hello

[@martinpickens2534]

What is rag?

[@OproDarius]

I like your content, but you same the same thing over and over, go more technical, show some demos, even code something

[@tonywhite4476]

RAG is useless. No onecan make money from RAG. No one can make money from A.I. It's too late for the indie developer. Period. So, make some cool sh*t to impress your famiy & friends but they could probaly go online to one of the million apps and do it themselves. It's over folks, go on home.

[@robertholtz]

Your pronunciation of the word "accurate" is inaccurate.

[@notanametobecalled]

First

[@zaursamedov8906]

I think rag is not that thing needs to be hyped cuz there are methods better than RAG

[@mohammedzayan5540]

like?

[@jojoduke1702]

I'd also like to know

[@EnlistedBootCamp]

you're out of date