Transformers explained | The architecture behind LLMs

Рет қаралды 21,061

Күн бұрын

All you need to know about the transformer architecture: How to structure the inputs, attention (Queries, Keys, Values), positional embeddings, residual connections. Bonus: an overview of the difference between Recurrent Neural Networks (RNNs) and transformers.
9:19 Order of multiplication should be the opposite: x1(vector) * Wq(matrix) = q1(vector). Otherwise we do not get the 1x3 dimensionality at the end. Sorry for messing up the animation!
➡️ AI Coffee Break Merch! 🛍️ aicoffeebreak.creator-spring....
Outline:
00:00 Transformers explained
00:47 Text inputs
02:29 Image inputs
03:57 Next word prediction / Classification
06:08 The transformer layer: 1. MLP sublayer
06:47 2. Attention explained
07:57 Attention vs. self-attention
08:35 Queries, Keys, Values
09:19 Order of multiplication should be the opposite: x1(vector) * Wq(matrix) = q1(vector).
11:26 Multi-head attention
13:04 Attention scales quadratically
13:53 Positional embeddings
15:11 Residual connections and Normalization Layers
17:09 Masked Language Modelling
17:59 Difference to RNNs
Thanks to our Patrons who support us in Tier 2, 3, 4: 🙏
Dres. Trost GbR, Siltax, Vignesh Valliappan, ‪@Mutual_Information‬ , Kshitij
Our old Transformer explained 📺 video: • The Transformer neural...
📺 Tokenization explained: • What is tokenization a...
📺 Word embeddings: • How modern search engi...
📽️ Replacing Self-Attention: • Replacing Self-attention
📽️ Position embeddings: • Positional encodings i...
‪@SerranoAcademy‬ Transformer series: • The Attention Mechanis...
📄 Vaswani, Ashish, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N. Gomez, Łukasz Kaiser, and Illia Polosukhin. "Attention is all you need." Advances in neural information processing systems 30 (2017).
▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀
🔥 Optionally, pay us a coffee to help with our Coffee Bean production! ☕
Patreon: / aicoffeebreak
Ko-fi: ko-fi.com/aicoffeebreak
▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀▀
🔗 Links:
AICoffeeBreakQuiz: / aicoffeebreak
Twitter: / aicoffeebreak
Reddit: / aicoffeebreak
KZfaq: / aicoffeebreak
#AICoffeeBreak #MsCoffeeBean #MachineLearning #AI #research
Music 🎵 : Sunset n Beachz - Ofshane
Video editing: Nils Trost

Пікірлер: 104

@YuraCCC 6 ай бұрын

Thanks for the explanation. At 9:19 : Shouldn't the order of multiplication be the opposite here? E.g. x1(vector) * Wq(matrix) = q1(vector). Otherwise I don't understand how we get the 1x3 dimensionality at the end

@AICoffeeBreak 6 ай бұрын

Oh, shoot, messed up the order in the animations there. You are right. Sorry, pinning your comment.

@YuraCCC 6 ай бұрын

No problem thanks for clarifying that, and thanks again for the great video@@AICoffeeBreak

@scifaipy9301 Ай бұрын

The vectors should be column vectors.

@heejuneAhn Ай бұрын

BEST of BEST Explanation. 1) Visually, 2) intuitively, 3) by numerical examples. And your English is better than native for Foreigners to listen.

@DerPylz 6 ай бұрын

Wow, you've come a long way since your first transformer explained video!

@mumcarpet109 6 ай бұрын

your videos has helped visual learner like me so much, thank you

@AICoffeeBreak 6 ай бұрын

Happy to hear that!

@xyphos915 6 ай бұрын

Wow, this explanation on the difference between RNNs and Transformers at the end is what I was missing! I've always heard that Transformers are great because of parallelization but never really saw why until today, thank you! Great video!

@AICoffeeBreak 6 ай бұрын

Oh, this makes me happy !

@abhishek-tandon 6 ай бұрын

One of the best videos on transformers that I have ever watched. Views 📈

@AICoffeeBreak 6 ай бұрын

Do you have examples of others you liked?

@cosmic_reef_17 6 ай бұрын

Thank you very much for the very clear explanations and detailed analysis of the transformer architecture. Your truly the 3blue1brown of machine learning!

@AICoffeeBreak 6 ай бұрын

@DaveJ6515 6 ай бұрын

You know how to explain things. This one is not easy: I can see the amount of work that went into this video, and it was a lot. I hope that your career takes you where you deserve.

@AICoffeeBreak 6 ай бұрын

Thanks for watching and thanks for the kind words. All the best to you as well!

@420_gunna 6 ай бұрын

Awesome video, thank you! I love the idea of you revisiting older topics -- either as a 201 or as a re-introduction. "Attention combines the representation of input vector's value vectors, weighted by the importance score (computed by the query and key vectors)."

@AICoffeeBreak 6 ай бұрын

Thanks for your appreciation!

@l.suurmeijer1382 6 ай бұрын

Absolute banger of a video. Wish I had seen this when I was learning about transformers in uni last year :-)

@AICoffeeBreak 5 ай бұрын

Haha, glad I could help. Even if a bit late.

@volpir4672 6 ай бұрын

that's great, I'm a little stuck on the special mask token? ... I'll keep digging, good info, the video is good explanation, it allows for more experimentation instead of relying on open source models that can have components look like a black box to noobs like me :)

@jonas4223 6 ай бұрын

Today, I had the problem I need to understand how Transformers work.. I searched on youtube and found your video 20 minutes after release. What a perfect timing

@AICoffeeBreak 6 ай бұрын

What a timing!

@mccartym86 4 ай бұрын

I think I had at least 10 aha moments watching this, and I've watched many videos on these topics. Incredible job, thank you!

@AICoffeeBreak 4 ай бұрын

Wow, thank You for this wonderful comment!

@16876 6 ай бұрын

What a thorougfh and much anticipated overview laid out so coherently ,, thank you

@AICoffeeBreak 6 ай бұрын

Our pleasure! We should have done this video much earlier, considering that our old Transformer Explained is our most watched video to date. 😅

@Thomas-gk42 6 ай бұрын

Understood about 10%, but I like these vidoes and feel intuitively the usefulness.

@AICoffeeBreak 6 ай бұрын

@user-th2ec8ms3m 6 ай бұрын

Really well done and easy to follow, thank you

@AICoffeeBreak 6 ай бұрын

Glad you enjoy it!

@darylallen2485 3 ай бұрын

Letitia, you're awesome and I look forward to learning more from you.

@jcneto25 5 ай бұрын

Best didatic explanation about Transformers so far. Thank you for sharing it.

@AICoffeeBreak 5 ай бұрын

Wow, thanks! Glad it's helpful.

@rahulrajpvr7d 6 ай бұрын

Tomorrow i have thesis evaluation and i was thinking about watching that video again, but youtube algorithm suggested me without searching anything, Thank u youtube algo.. 😅❤🔥

@AICoffeeBreak 6 ай бұрын

It read your mind.

@manuelafernandesblancorodr6366 5 ай бұрын

What a wonderful video! Thank you so much for sharing it!

@AICoffeeBreak 5 ай бұрын

Thank you too for this wonderful comment!

@connorshorten6311 6 ай бұрын

Awesome! Epic Visuals!

@AICoffeeBreak 6 ай бұрын

Thanks, Connor!

@DatNgo-uk4ft 6 ай бұрын

Great Video!! Nice improvement over the original

@AICoffeeBreak 6 ай бұрын

Glad you think so!

@phiphi3025 6 ай бұрын

Thanks, you helped so much explain Transformers to my PhD advisors

@AICoffeeBreak 6 ай бұрын

This is really funny. In what field are you doing your PhD? 😅

@xxlvulkann6743 3 ай бұрын

This is a very well-made explanation. I hadn't known that the feedforward layers only received one token at a time. Thanks for clearing that up for me! 😁

@SamehSyedAjmal 6 ай бұрын

Thank you for the video! Maybe an explanation on the Mamba Architecture next?

@AICoffeeBreak 5 ай бұрын

The Mamba and SSM beans are roasting as we speak.

@muhammedaneesk.a4848 6 ай бұрын

Thanks for the explanation 😊

@AICoffeeBreak 6 ай бұрын

Thanks for watching!

@supanutsookkho2749 18 күн бұрын

Great video and a good explanation. Thanks for your hard work on this amazing video!!

@AICoffeeBreak 7 күн бұрын

Glad you liked it!

@MachineLearningStreetTalk 6 ай бұрын

Epic as always 🤌

@AICoffeeBreak 6 ай бұрын

Thanks, Tim!

@HarishAkula-df8gs 3 ай бұрын

Amazing explanation, Thank you! Just discovered your channel and I really like how the difficult topics are demystified.

@AICoffeeBreak 3 ай бұрын

Thanks a lot!

@Clammer999 2 ай бұрын

Thanks so much for this video. I’ve gone through a number of videos on transformers and this is much easier to grasp and understand for a non-data scientist like myself.

@AICoffeeBreak 2 ай бұрын

You're very welcome!

@partywen Ай бұрын

Super informative and helpful! Thanks a lot!

@AICoffeeBreak Ай бұрын

Oh wow, thanks!

@meguellatiyounes8659 6 ай бұрын

well explained . as you promised

@AICoffeeBreak 6 ай бұрын

@paprikar 6 ай бұрын

here we go! TY for content

@bartlomiejkubica1781 5 ай бұрын

Thank you! Finally, I start to get it...

@dannown 5 ай бұрын

Really appreciate this video.

@AICoffeeBreak 5 ай бұрын

So glad!

@Ben_D. 4 ай бұрын

...ok. After binging some of your vids, I now need to go make coffee. 😆

@AICoffeeBreak 4 ай бұрын

Please do!

@uw10isplaya 25 күн бұрын

Had to go back and rewatch a section after I realized I'd been spacing out staring at the coffee bean's reactions.

@AICoffeeBreak 7 күн бұрын

@zahrashah6567 3 ай бұрын

What a wonderful explanation😍 Just discovered your channel and absolutely loving the explanations as well as visuals😘

@AICoffeeBreak 3 ай бұрын

Thank you! welcome!

@zbynekba 6 ай бұрын

❤ Letitia, thank you for great visualization and intuition. For inspiration: In the original paper, the decoder utilizes the output of the encoder by running a cross-attention process. Why does GPT not use an encoder? As you've mentioned, the encoder is typically used for classification, while the decoder is for text generation. They are never used in combination. Why is this the case? Missing Intuition: Why does the cross-attention layer inside the decoder take the values from the ENCODER’s output to create the enhanced embeddings (as a weighted mix)? Intuitively, I would use the values from the DECODER.

@AICoffeeBreak 6 ай бұрын

Thanks for your thoughts! Encoders are sometimes used in combination with decoders, right? The most famous example is the T5 architecture.

@zbynekba 6 ай бұрын

Thanks for your prompt reply. Hence, understanding the concept and intuition behind feeding the encoder output into the decoder is essential. I found only this one video on encoder-decoder cross-attention: kzfaq.info/get/bejne/eteapJd43crMnps.htmlsi=gtLzNxAU0pUGyLvk In it, Lennart emphasizes the observation that, based on the original equations, we have the enhanced embeddings calculated as a weighted sum of ENCODER values. Inside of a DECODER, I would rather expect to have the DECODER values pass through. Letitia, I am sure, you will resolve this mystery. 🍀

@pfever 5 ай бұрын

Just discovered your channel and this is great! Thank you! :D

@AICoffeeBreak 5 ай бұрын

Thank you! Hope to see you again soon in the comments.

@MuruganR-tg9yt 5 ай бұрын

Thank you. Nice explanation 😊

@AICoffeeBreak 5 ай бұрын

Thank You for your visit!

@ArthasDKR 5 ай бұрын

Excellent explanation. Thank you!

@AICoffeeBreak 5 ай бұрын

@tildarusso 6 ай бұрын

As far as I am aware, word embedding has changed from legacy static embedding like Word2Vec/GLOVE (like the famous queen=woman+king-man metaphor) to BPE & unigram, this change gave me quite a headache, as most of paper do not mention any detail of their "word embedding". Perhaps Letitia you can make a video to clarify this a bit for us.

@AICoffeeBreak 6 ай бұрын

Great suggestion, thanks!

@kallamamran 6 ай бұрын

Phew 😳

@l3nn13 5 ай бұрын

great video

@AICoffeeBreak 5 ай бұрын

Thanks for the visit and for leaving the comment!

@ai-interview-questions 5 ай бұрын

Thank you, Letitia!

@AICoffeeBreak 5 ай бұрын

Our pleasure!

@LEQN 4 ай бұрын

Awesome video :) thanks!

@AICoffeeBreak 4 ай бұрын

Thank you for watching and for your wonderful comment!

@heejuneAhn 8 күн бұрын

Thanks for your video. I have a question on inference process. For example when I have a input prompt of 2 tokens = {t1, t2}. we will get the output {o1, o2, o3}. we will take only o3 and make new input sequence {t1, t2, o3}. Then we will get another output {o'1, o'2, o'3, o'4}. Here my questions are. When we use causal masking for attention, o1= o'1 and o2=o'2 and so on? Another question, even though the mask guarantee the causal attension. but still the matrix calcuation is performed. Then it means the computation is used any way. How can we reduce the computation resource for this case.

@M4ciekP 6 ай бұрын

How about a video explaining SSMs?

@AICoffeeBreak 5 ай бұрын

✍️

@AICoffeeBreak 5 ай бұрын

Psst: This will be the video coming up in a few days. it's in editing right now.

@M4ciekP 5 ай бұрын

Yaay! @@AICoffeeBreak

@ehudamitai 5 ай бұрын

In 11:14, the weighted sum is the sum of 3 vectors of 3 elements each, but the results is a vector of 4 elements. Which, conveniently, is the same size as the input vector. Could there be a missing step there?

@AICoffeeBreak 5 ай бұрын

Yes, there is a missing back transformation to 4 dimensions I skipped. :) Well spotted!

@nmfhlbj 4 ай бұрын

hi! can i ask question of how did you get the dimension (d)? because all i know is dimension can be found in square matrices, and the dot product of the attention formula says that Q•K^T. if we're using 1x3 matrices, we'll get 1x1 matrices or 1 dimension, how do you get 3 ? unless its 3x1 matrix beforehand, so we'll get 3x3 or 3 dimensional matrix. thankyouu !

@AICoffeeBreak 4 ай бұрын

Hi, if you mean the mistake at 10:00, then the problem is that I have written matrix times vector when I should have written vector times matrix! (or I could have used column vectors instead of row vectors). Is this what you mean?

@DaeOh 6 ай бұрын

Everything makes sense except multiple attention heads. Each layer has only one set of Q, K, V, O matrices. But 8 attention heads per layer? I want to understand that.

@AICoffeeBreak 5 ай бұрын

Think about it this way: In one layer, instead of having one head telling you how to pay attention at things, you have 8. In other words, instead of having one person shout at you the things they want you to pay attention to, you have 8 people simultaneously shouting at you. This is beneficial because it has an ensembling effect (the effect of a voting parliament. Think of Random Forests that are an ensemble of Decision Trees). I do not know if this helps, but I thought giving it another shot at explaining this.

@benjamindilorenzo 4 ай бұрын

What a great video. It still could expand more and really sum up every sub-part and connect it to a certain clear visualization or clear step of what happens with the information at each time step and how its "transformation" progresses over time. So i think you could redo this video and really make it monkey proof for folks like me. But beware, if you look for example at the StatQuest version, its to slow and too repetative and also does not really capture, what really goes on inside the Transformer, once all steps are stacked together. Great work!