A Practical Introduction to Large Language Models (LLMs)

👉 CXOs, VPs, & Directors... I offer custom AI workshops: www.shawhintalebi.com/ai-work...
This is the 1st video in a series on using large language models (LLMs) in practice. I introduce LLMs and three levels of working with them.
▶️ Series Playlist: • Large Language Models ...
📰 Read more: towardsdatascience.com/a-prac...
Music by Me
/ letsalljusttakeiteasy
/ @shawhin-music
--
Book a call: calendly.com/shawhintalebi
Homepage: shawhintalebi.com/
Socials
/ shawhin
/ shawhintalebi
/ shawhint
/ shawhintalebi
The Data Entrepreneurs
🎥 RUclips: / @thedataentrepreneurs
👉 Discord: / discord
📰 Medium: / the-data
📅 Events: lu.ma/tde
🗞️ Newsletter: the-data-entrepreneurs.ck.pag...
Support ❤️
www.buymeacoffee.com/shawhint
Intro - 0:00
What is an LLM? - 1:13
Zero-shot Learning - 3:36
How do LLMs work? - 5:44
3 Levels of Using LLMs - 7:52
Level 1: Prompt Engineering - 8:22
Level 2: Model Fine-tuning - 11:00
Level 3: Build your own - 13:13

[@ShawhinTalebi]

👉More on LLMs: ruclips.net/p/PLz-ep5RbHosU2hnz5ejezwaYpdMutMVB0

[@ajalipio1]

I love the clarity and simplicity of your videos. I'm a new fan and you got a new sub! 🥰

[@DRAI-ow1nq]

Thanks Shaw! I always look forward to your videos

[@pmh312_]

This was great, looking forward to future videos.

[@ShawhinTalebi]

Thanks! 2nd video is up 😁 ruclips.net/video/czvVibB2lRA/видео.html

[@paulntalo1425]

Thanks for sharing such a play list

[@IndyScriabin-dl8ot]

Excellent material you covered herein as an executive summary and TL; DR of LLMs. Thank you very much.

[@DavidDayichibanNorae]

This is great content. I love it!

[@enoack1]

Very good high level explanations that make the subject(s) very accessible. Thanks!

[@ShawhinTalebi]

Glad it was helpful!

[@emammadov]

this is amazing series of videos. Well done for explaining this to us in such an easy-to-understand way

[@ShawhinTalebi]

Glad you liked it

[@HaneNaghshbandi]

Finally I found a video explaining such a complex concept in a simple way. Thanks a lot. It was so good!

[@ShawhinTalebi]

Glad it was helpful!

[@happylearning1648]

Thank you for the clear explanation

[@yahya5054]

thank you very much for this well explained video !

[@helrod6131]

I appreciate your simple explanation of a complex subject.

[@ShawhinTalebi]

Thanks, I’m glad it made sense :)

[@anneest]

Thanks for sharing this, very good and condensed information (I got the link to your YT channel from your article in towardsdatascience).
Looking forward to seeing future videos with examples, cheers 👍

[@ShawhinTalebi]

Thank you, I'm glad it was helpful!

[@barclayiversen376]

This channel is going to become my new addiction

[@ShawhinTalebi]

Hope its a healthy addiction XD

[@flimsandmediaproduction6350]

been struggling with the concept. your video indeed helped alot

[@ShawhinTalebi]

Glad it helped!

[@lookup2423]

Very well explained video. Good Introduction

[@ShawhinTalebi]

Thanks, glad you liked it!

[@King_Konglish]

This is super useful, thanks!

[@ShawhinTalebi]

Glad it was helpful!

[@user-nw9sc3ev2z]

super helpful thank you!

[@ShawhinTalebi]

Glad it helped!

[@ifycadeau]

Finally! Prompt engineering is so underrated!!🙌🏾

[@ShawhinTalebi]

It’s the lowest hanging fruit!

[@marcoaureliobalduinodesouz2602]

Very good video

[@user-wr4yl7tx3w]

This was really well explained.

[@ShawhinTalebi]

Thanks, glad it made sense!

[@arpitrawat1203]

Thanks for the video. Would love to hear about different use cases with implementation.

[@ShawhinTalebi]

Happy to help! More use cases with code to come in future videos :)

[@user-vv7om6uq5w]

Thanks!!!

[@roastmaker1233]

Thanks for the video

[@ShawhinTalebi]

Happy to help!

[@hoangng16]

It would be great if you could make a series about fine-tuning LLM/s for a specific area of tasks. The reasons are:
1) The number of people requiring their own LLM/s is very small; if they need their own models, they already have them.
2) I have seen many tutorials about fine-tuning, but they only touch the surface layer. Plus, preparing data in the form of questions and answers takes so much effort that it is not practical.
Thank you

[@ShawhinTalebi]

I do a hands-on example for fine-tuning a model to respond to RUclips comments here: ruclips.net/video/XpoKB3usmKc/видео.html
Feel free to share any specific suggestions you have in mind :)

[@greatgatsby6953]

This was useful -even though I did not understand everything!

[@ShawhinTalebi]

Good to hear! Happy to answer any questions you might have.
I’ve also got office hours: calendly.com/shawhintalebi/office-hours

[@hardcoder-y6w]

@Shaw Talebi - Thanks for creating such amazing video. Indeed it helped so much. I got a question here, do you provide AI or LLM training? Perhaps, boot camp?

[@ShawhinTalebi]

Glad it was helpful! While I currently only doing trainings for orgs, I'm putting one together for individuals.
I can keep you posted if you are interested. Feel free to email me here: www.shawhintalebi.com/contact

[@hardcoder-y6w]

Excellent! @Shaw Talebi - Done! I will see you soon.Thanks :)

[@elpablitorodriguezharrera]

Hi Shaw,
Super clear explanation, I was wondering if I can request an explanation on how exactly use LangChain "effectively"? The benefits and limitations, or are there other options to "combined" every state-of-the-art of each models are there in one place? (Like an open source text gen (mixtral 8x7b), image gen(fooocus), audio and video gen in one place of a chatbot interface)?

[@ShawhinTalebi]

Glad it was clear and thanks for the suggestion! I'm planning 3 more videos for this series, and I'll definitely keep these in mind :)

[@ahmarhussain8720]

very beginner friendly explanation

[@KumR]

Great Intro Shaw....from Shiv

[@ShawhinTalebi]

Thanks Shiv!

[@amparoconsuelo9451]

Another response from you please.
1) How is LLM related to robotics? Is LLM used in or a part of robotics?
2) If LLM is used in robotics, can LLM be fine-tuned every time the robot's task is modified or changed?
3) Can an app made with the help of LLM be used instead to control, program and modify the movements of robot?

[@ShawhinTalebi]

1) These are different technologies, however there will surely be interesting use cases combining the two.
2) While this depends on the details of the use case, LLMs can in principle learn from feedback (i.e. reinforcement learning)
3) I'm sure this is possible, however it's unclear whether it would yield in better results than existing approaches
Thanks for all the interesting questions!

[@PokemonRockstars-zr6yc]

Your slides are very good could you also share them please

[@ShawhinTalebi]

Thanks! Slides are available here: github.com/ShawhinT/RUclips-Blog/tree/main/LLMs/_slides

[@fitandvirile]

Can I earn by creating LLMs or fine tuning

[@ShawhinTalebi]

If your model solves a problem people are willing to pay for then yes. But this often requires more than just a model.

[@NiteshKumar-up8ib]

From where we all viewers get sildes of lecture

[@ShawhinTalebi]

Slides are available on the GitHub: github.com/ShawhinT/RUclips-Blog/tree/main/LLMs/_slides

[@tiredofeverythingnew]

The constant cutting of the audio makes this very uncomfortable. Great content otherwise.

[@ShawhinTalebi]

Thanks for the feedback. I admit to being heavy handed on the edits to minimize play time, but there's clearly room for improvement.