A Hackers' Guide to Language Models

In this deeply informative video, Jeremy Howard, co-founder of fast.ai and creator of the ULMFiT approach on which all modern language models (LMs) are based, takes you on a comprehensive journey through the fascinating landscape of LMs. Starting with the foundational concepts, Jeremy introduces the architecture and mechanics that make these AI systems tick. He then delves into critical evaluations of GPT-4, illuminates practical uses of language models in code writing and data analysis, and offers hands-on tips for working with the OpenAI API. The video also provides expert guidance on technical topics such as fine-tuning, decoding tokens, and running private instances of GPT models.
As we move further into the intricacies, Jeremy unpacks advanced strategies for model testing and optimization, utilizing tools like GPTQ and Hugging Face Transformers. He also explores the potential of specialized datasets like Orca and Platypus for fine-tuning and discusses cutting-edge trends in Retrieval Augmented Generation and information retrieval. Whether you're new to the field or an established professional, this presentation offers a wealth of insights to help you navigate the ever-evolving world of language models.
(The above summary was, of course, created by an LLM!)
For the notebook used in this talk, see github.com/fastai/lm-hackers.
00:00:00 Introduction & Basic Ideas of Language Models
00:18:05 Limitations & Capabilities of GPT-4
00:31:28 AI Applications in Code Writing, Data Analysis & OCR
00:38:50 Practical Tips on Using OpenAI API
00:46:36 Creating a Code Interpreter with Function Calling
00:51:57 Using Local Language Models & GPU Options
00:59:33 Fine-Tuning Models & Decoding Tokens
01:05:37 Testing & Optimizing Models
01:10:32 Retrieval Augmented Generation
01:20:08 Fine-Tuning Models
01:26:00 Running Models on Macs
01:27:42 Llama.cpp & Its Cross-Platform Abilities
This is an extended version of the keynote given at posit::conf(2023). Thanks to @wolpumba4099 for chapter titles.

[@howardjeremyp]

Gotta admit I'm feeling kinda teary reading all the lovely comments here. Thank you everybody -- love you all!

[@ksrajavel]

You just deserved it.😃

[@ReligionAndMaterialismDebunked]

Second in the replies. :3

[@JumpNationFilms]

You are beyond awesome, Jeremy

[@Soulseeologia]

Thanks for your work and please help us keep an eye on the apostles of the emerging noosphere, like Ben Goertzel ect.

[@ellevillarreal9074]

Jeremy Thank you! This has helped so much. I’ve been a FastAI builder since the early days in 2017. Youre my hero. Appreciate all of the work you’ve done in the field

[@buildar8434]

Just realised Jeremys paper led to the LLM revolution. Such a humble kind man. God bless you and all your students. You are such an example to follow. An example in character, humility and intelligence.

[@circleAI]

how ? which paper?

[@pw7225]

@@circleAI ULMFiT

[@fasteddylove-muffin6415]

@@circleAI Part of the answer is in the video's description.

[@peaku8129]

Yes exactly what I was thinking, why there's so many people bragging about what they are doing and look at this guy, just helping others out.

[@Jporomaa]

This is probably the best invested youtube time of this year so far. What a gem. A lot of things he mentions have taken me month to figure out by my own. My new GPT-4 prompts will begin with "You are the expert Jeremy Howard..."

[@ProfRvS]

This!

[@shaggyfeng9110]

lol, nice

[@peterdonnelly1074]

100%

[@lucaleung5892]

So great.

[@JarkkoHautakorpi]

Absolutely!

[@DHAiRYA2801]

The moment I got to know that you and Andrej weren't included in the Time's list, I realized that the people making such lists have no idea what they are doing. Loved the tutorial, thank you!

[@MensRea78]

No one can explain a topic like Jeremy👍

[@wolpumba4099]

*Transcript Summary:* - Introduction & Basic Ideas of Language Models (00:00:00 - 00:18:05) - Limitations & Improvements of GPT-4 (00:18:05 - 00:31:28) - AI Applications in Code Writing, Data Analysis & OCR (00:31:28 - 00:38:50) - Practical Tips on Using OpenAI API (00:38:50 - 00:46:36) - Creating a Code Interpreter with Function Calling (00:46:36 - 00:51:57) - Using Local Language Models & GPU Options (00:51:57 - 00:59:33) - Fine-Tuning Models & Decoding Tokens (00:59:33 - 01:05:37) - Testing & Optimizing Models with GPTQ & Hugging Face (01:05:37 - 01:09:48) - Fine-Tuning with Llama 2 & Platypus Datasets (01:09:48 - 01:10:32) - Retrieval Augmented Generation & Information Retrieval (01:10:32 - 01:20:08) - Running a Private GPT & Fine-Tuning Models (01:20:08 - 01:22:32) - Running Models on Macs (01:26:00 - 01:27:42) - Discussing Llama.cpp & Its Cross-Platform Abilities (01:27:42 - 01:30:07) - Challenges & Opportunities in Language Models (01:30:07 - 01:31:05) Key points of interest: Function usage in GPT-4 (00:46:36), OCR application with Google Bard (00:33:59), and improving GPT-4 responses with custom instructions (00:24:36).

[@qwertclyde1092]

Dope, did you do this by hand?

[@4everandrew]

😅😅😅😅

[@-rate6326]

I was expecting "By Tammy AI"

[@uminhtetoo]

Thanks

[@JustSayin24]

I found this video so useful that I felt compelled to pull my keyboard closer toward me, fix my posture, and write this comment - something I rarely do. I'm a professional data scientist hoping to push my company's GenAI agenda and this video makes me feel like I can actually do it! Thank you for so clearly encapsulating the state of LLMs. I'd learned many of these concepts before and this video is the glue that now holds it together.

[@satirthapaulshyam7769]

Hoping to look forward it.Ur fastai stable diffusion course was Perfect to the minute details

[@s.o.s9132]

very easy to know, practical! thanks Jeremy

[@Blocky007]

The best "intro" and Guide I have seen on this. Appreciate it so much that you took the time to put this together and share this with us (FOR FREE!).

[@raindrops20008]

This video landed up on my feed and out of curiosity I started watching and before I knew it had watched the entire video and taken copious amounts of notes too. One of the best videos I have ever watched!

[@Libertarian1208]

Great! It allowed me to understand how LM thinks and why.

[@SapSapirot]

By far the most useful practical guide to LLM's by length. Thank you Jeremy!

[@drkarmakid]

Bravo. One of the best KZbin videos I've ever watched. Concise, entertaining, and chock full of useful insights.

[@kindoblue]

I liked the video even before watching. Thanks Jeremy for your work, always learning from your content.

[@giorda77]

Jeremy you're one of the most legit AI person out there. An enormous thank you for providing this and all your content. ❤

[@aldotanca9430]

So comprehensive. Perhaps the best introduction I have ever seen to the topic. Thanks so much.

[@PeterKallio]

So many papers are being released, so it is important to have well-grounded information to understand LMs. Great delivery as always and practicable advice. Thank you.

[@kevon217]

Fabulous tour of key points. Fantastic job! Definitely going to recommend this to people wanting a gateway into llms.

[@marvit9039]

We do not deserve you Jeremy! YOU ARE AN AMAZING TEACHER AND HUMAN BEING! Thanks, really, for all these beautiful lectures!!

[@user-lv7ft2zu7k]

Thanks for all you do Jeremy. I have learned so many things watching youtube as well as the PDLC tutorials. Your expanations are on point.

[@0x978]

You are literally changing lives, all for free. Thank you.

[@RickySupriyadi]

I'm really grateful how much people sharing their knowledge, can't imagine learning stuff for free. this means a lot for me.

[@claussa]

Came up in my feed. Thumbnail and title boring. By mistake I pressed play. But it was so interesting. I feel so enlightened after having been talked through this. Thanks for sharing this!

[@wndiros]

I have waited for months for a classification and evaluation from Jeremy. For me, this is by far the most comprehensive technical summary and evaluation available for someone who wants to delve deeper. It took me several weeks, if not months, to gain even a partial personal understanding of the current hype. Thank you, Jeremy, for all your good work!👍

[@KEcAzZ1]

A true legend! So far, I have not seen a better educator than Jeremy. His approach of teaching is what all schools and universities need! I am always interested to learn more, whenever I hear Jeremy. Thank you!

[@howardjeremyp]

Wow, thank you!

[@nokchan5071]

There are hundreds of LLM tutorial coming out everyday, this is the one that I have been waiting for.

[@bartekbinda6978]

This is golden summary of the state of the LLMs, Thank You

[@benwatson5211]

Brillant walk through. No hype. It is a real skill to explain complex topics is coherent way.

[@jeromeeusebius]

Great video. Watched it in one sitting. It's very interesting and engaging, and does cover a lot of areas on LLM, different model, types, examples, uses cases, etc. I learned a lot and hopefully will go through the notebook in detail and adapt to my use cases. Thanks for making this.

[@MarkusEicher70]

Thank you very much, Jeremy. Fascinating to see where we have come. The prose to SQL thing blew me out of the shoes. Can't wait to try this out by myself.

[@sidsarasvati]

Thought provoking one code block at a time. As usual Jeremy the king

[@victorv682]

The "wolf, goat and cabbage" riddle example is just awesome. Gotta use it to illustrate what LLMs can't do and why. Cheers for that :)

[@sehmiconscious]

Very much appreciated this consolidation of the main LLM coding concepts to-date. Thank you!!

[@jackgolding4235]

Thanks so much Jeremy, been following you since Kaggle's launch. Inspirational to see an Australian continue to kick ass as much as you have in your career.

[@vishalahuja2502]

Hi Jeremy, excellent walkthrough! This is truly helpful. Please keep them coming!!

[@OsamaAdel]

Thank you Jeremy for this introduction. It just answered many of my questions and affirmed some of my doubts about how many of the applications that use LLMs work today.

[@unclecode]

Kudos, such pleasurable 1 hour and 31 mins and 12 seconds.

[@twinlens]

One of the best and most educational videos I've seen on the subject. Thank you, Jeremy!

[@TheAtleticlub]

This is great. I don't know how to say how much grateful I am for your video. Thank you and keep the great work!

[@ControlTheGuh]

I think the major problem is the retrieval. Would love a video just on that (best practices, best models out there etc.). Great video, came across on x and subbed immediatley.

[@17dhey36]

Such a great article! I learned a lot from this video, such as how complicated systems can be put together using a stack of models, illustrated in the RAG to name an example. Jeremy, you are such a kind person to share this with the world.

[@dr_rd]

Truly enlightening! As a software engineer with limited math and data science knowledge, this video has been a revelation. The way Prof. Howard simplifies complex concepts is incredible, making each rewatch rewarding with new insights. Really grateful for his content that opens up the world of LLMs to a broader audience. His clear and thorough explanations are incredibly invaluable. Thanks, Prof. Howard, for demystifying this topic and helping us all learn.

[@howardjeremyp]

Wow, thank you!

[@ElanElanElan]

Damn I watched the whole video and didn't even realise that it was 1+ hr long! Thanks a lot for the great content!

[@joegerrety4012]

Keep on making videos man this was highly informative and my regards to being a person who was part of forming this architecture!

[@learning_with_rj]

Jeremy, we need more videos on this topic! Thank you so much!

[@briandavis6476]

What a great primer! Very much needed! Thanks as always Jeremy!

[@SheeceGardazi]

thankyou for the talk

[@ryanau852]

Happy birthday Jeremy! Just got to the section where your bday is revealed and it is today! Thank you for all the great work :)

[@dimasmufid2375]

This is remarkable. Thanks for sharing this topic for us Jemery!

[@WizBoardApp]

Always look forward Jeremy to explain this topic. Finally it is here. 😀

[@JamesBradyGames]

This is a real gem. Reminds me of the authentic, high quality training material from Andrej Karpathy. Looking forward to future similar tutorials if you decide to make them! Thank you!

[@EmaMazzi76]

Can’t stop watching over and over again! Thank you 🙏

[@drhilm]

Impressive video; I spent days learning these concepts on my own. Had this been released two months ago, it would've been a game-changer. Excellent summary.

[@jackgolding4235]

Luckily i'm a few weeks behind you! Happy learning mate!

[@kaib5048]

This is amazing, thanks so much for recording this and sharing it 👏

[@dekumutant]

I can not emphasize how incredible this video was

[@dollarscholar2956]

Hands down one of the best videos on LLMs on the internet.

[@chtdv]

Thank you for creating this amazing talk around all the basics and applications with language models, this is really helpful!

[@Massenko]

Thank you Jeremy for all of your work and for sharing such quality videos. ❤

[@CY-ju8xl]

Thank you Jeremy!! One of the most insightful and helpful vlog posts on the inner workings of LLMs... Top marks!!!

[@zhubarb]

Mr Howard never disappoints. Thanks a ton as usual Sir.

[@zakiasalod891]

Thanks Jeremy for another wonderful lecture! Much appreciated.

[@dalton4035]

Great content, and I love the way you structured it. Thanks :)

[@devsuniversity]

Great course! Hello from Almaty Google developers community!

[@isaac10231]

I feel like this video was made personally just for me. Amazing.

[@madankd]

oh man great video

[@rahul-qo3fi]

This is pure gold!!!!❤❤❤❤ I would have never found this on my own

[@sunshine124422]

Thanks...great summary....now i know the relatiinship between neural network parameters and vector DB's

[@solutrian]

I've been a student of yours for years, thank you for all you do for us. A non-optimal but correct solution to the cabbage, wolf & goat crossing works by just using incantation "think step by step" : Think step by step to solve the following tricky problem: Here is a logic puzzle. I need to carry a cabbage, a goat and a wolf across a river. I can only carry one item at a time with me in the boat. I can't leave the goat alone with the cabbage, and I can't leave the cabbage alone with the wolf. How can I get everything to the other side of the river? This classic river crossing puzzle can be solved through a series of strategic moves that ensure the safety of each item. Here's a step-by-step solution: 1. **Take the goat across first:** - You take the goat across the river and leave it on the other side. - This ensures that the goat is safe from the wolf, and the cabbage is safe from the goat. 2. **Return alone:** - You return to the original side alone to get either the cabbage or the wolf. 3. **Take the cabbage across next:** - You take the cabbage across the river. - Now, the goat is on the other side, and the cabbage is with you. 4. **Bring the goat back:** - You bring the goat back with you to the original side. - Now, the cabbage is alone on the other side, and the wolf is alone on the original side. 5. **Take the wolf across:** - You then take the wolf across the river. - Now, the wolf and the cabbage are on the other side, and you have the goat on the original side. 6. **Return alone:** - You return to the original side alone. 7. **Take the goat across one final time:** - Finally, you take the goat across the river again. - Now, all three- the goat, the cabbage, and the wolf -are safely on the other side. This way, you ensure that the goat is never left alone with the cabbage, and the cabbage is never left alone with the wolf, solving the puzzle. chat.openai.com/share/3cb73481-c45c-4108-9960-30c11615ca41

[@user-ns3tn4zk9d]

I remember that I was rewriting some chunk of TensorFlow code with GPT4 code interpreter, and it responded that it did not access to that framework. So I input 'gpt4: "try: import torch"', respond was 'Do you want to learn Pytorch' :) All about gpt's self.awareness mechanism :)

[@HazemAzim]

A wealth of Information and knowledge in this video .. a master piece .. Thank you

[@ZachariahMustafa]

wonderful lesson as always Jeremy! I did have a laugh at the GPT-4 bit "Bad pattern recognition - thanks to Steve Newman", as if he's the sole individual responsible for that limitation

[@exconway]

Perfect description of Functions at 46:30!

[@patronspatron7681]

Jeremy, Congrats on the 100k subscribers. Well deserved and hopefully a catalyst to get your invaluable content more exposure.

[@bloolizard]

Awesome stuff, always like learning from your videos. Been watching since FastAI v1.

[@steefvanwinkel]

Such an amazing video! Thanks so much for this amazing overview of LLMs! A few suggestions for a follow-up video perhaps :) 1. How to build an LLM from scratch on a consumer grade machine (e.g. laptop with 1 good GPU or Macbook M2 Max), maybe not 1B params but e.g. 500M? 2. LORA and DPO. 3. OpenAI's updated API. 4. Agents based frameworks such as AutoGen. Thanks again - really enjoyed this video! ❤🙏

[@wolpumba4099]

*Positive Learnings:* 1. Language models, such as GPT4, are tools that can predict the next word in a sentence or fill in missing words in a sentence. 2. Language models have the ability to create a rich hierarchy of abstractions and representations which they can build on. 3. The guide will cover all the basic ideas of language models, including how to use open source and open AI-based models. 4. GPT4 can solve many tasks that it is often claimed it cannot. 5. GPT4 can be primed to give high-quality information by giving it custom instructions. 6. AI can be used to write code and parse large-scale data quickly and efficiently. 7. AI can be used in optical character recognition (OCR) for extracting text from images. 8. AI can be used in data analysis to create comprehensive tables from scattered information. 9. The OpenAI API allows users to use AI programmatically for data analysis and other repetitive tasks. 10. Function calling can be used to create a code interpreter that runs inside Jupiter. 11. Pre-trained models can be accessed using the Hugging Face library. *Negative Learnings:* 1. Language models are not always useful on their own and need to be fine-tuned. 2. GPT4 often repeats mistakes and it is difficult to get it back on track once it starts making mistakes. 3. GPT4 has limitations such as not knowing about itself, not knowing anything about URLs, and not knowing anything after its knowledge cutoff in September 2021. 4. GPT4 does not always give correct responses. 5. AI has limitations in code interpretation and cannot substitute for human programmers. 6. The use of the OpenAI API can result in rate limits which need to be handled correctly. 7. Fine-tuning is needed to make the pre-trained models more useful. 8. The use of GPUs for local language models can be expensive and may require renting or purchasing GPUs.

[@uminhtetoo]

Thank you for sharing your experience and knowledge, Sir.

[@bilbobaggins3152]

Such an excellent video, thanks for taking up your time producing this for those very much at the start of this journey

[@hishamk]

Thank you for this. Couldn’t have asked for a better video.

[@MatijaGrcic]

This is so well done and presented. Thank you.

[@manaranjanp]

Never miss Jeremy's lectures....

[@yakinbama]

This is absolutely fabulous. Thank you!

[@okidoky876]

I was waiting for this for a long time! Thank you very much

[@bephrem]

incredible, value-packed, practical video for developers working with LLMs.

[@howardjeremyp]

Glad you liked it!

[@MatthewKanwisher]

Thanks for making video. Would love to see some follow up videos on use cases for fine tuning. Where does it make sense vs RAG or even just better prompts

[@gamalchamac867]

This is amazing and serious content. I love it. Thank you for making it available, greetings from Switzerland

[@AnthonyBatt]

Jeremy this is a gem of a video. thanks again.

[@maureenmetzger7588]

Wonderful overview, gives me confidence to dive in!

[@BharatR97]

@howardjeremyp I love the fact that you have quickstart demos and quick prototypes to illustrate each and every one of these LLM Concepts here and in general, a great breadthwise introduction to LLMs! Only wish we had this video sooner :)

[@Textras]

What a fantastic video. Really enjoyed thank you!!

[@sukhpreetlotey1172]

Thank you for sharing all this knowledge.greatly appreciate it

[@keflatspiral4633]

this changed my mind on how AI works! Thanks alot !!

[@aa-xn5hc]

Thank you!🙏🏻😀 Brilliant presenter!!

[@aladinmovies]

Thank you! Amazing tutorial. Thanks

[@bagamanocnon]

top content. specially the function calling and 'code interpreter' parts. basically just asking the LLM to 'use python' is ingenious. lol

[@davilirio6040]

Super informative video. Thanks for putting this out!

[@sorover111]

Hey thank you for making these available for free. ❤

[@85abhiabhi]

I am a total beginner but u made me understand abt LM models way better than anuone else..u r such a great teacher..I pray for giving u Lord Gurus blessings dor more insight and vision for such a Humble and good Soul.😊😊

[@stuartpatterson1617]

100 % agree! Blessings to you too.

[@johnmo1111]

I really enjoyed this talk. Thank you so much.