The animated Transformer: the Transformer model explained the fun way!

  Рет қаралды 10,036

DeepLearning Hero

DeepLearning Hero

Күн бұрын

Пікірлер: 12
@gruhitpatel9061
@gruhitpatel9061 3 ай бұрын
Seen a lot of videos explaining transformer but this one was by far the best one till yet. Great efforts 👍
@andrewtwigg
@andrewtwigg 15 күн бұрын
very clear video, thank you!
@connorsheehan4598
@connorsheehan4598 3 ай бұрын
What was the reason to use layer norm? Like in terms of the sentence being input, is layernorm applying across single sentences? Or the first words in unrelated sentences (thinking of sentences stacked in a matrix)?
@soumyasaha1429
@soumyasaha1429 11 ай бұрын
Great explanation
@iliasaarab7922
@iliasaarab7922 7 ай бұрын
Great video! Do you have the source code of the animation on github?
@AhladKumar
@AhladKumar Жыл бұрын
Great lecture! Which software you are using for making these videos.?
@deeplearninghero
@deeplearninghero Жыл бұрын
Hey thanks for that. The animations were using the Manim Python library. For the overall video editing I use Camstasia.
@ssshukla26
@ssshukla26 Жыл бұрын
Great to see comment of the professor from whom I learned the DL. 😊
@iliasaarab7922
@iliasaarab7922 5 ай бұрын
@@ssshukla26 same for me, @AhladKumar is a legend!
@jeffbrownstain
@jeffbrownstain Жыл бұрын
Something about Transformers really is next level and somewhat magical. Gpt2 was FAR from the first NAS AI: our AI have been programming themselves since 2012. Something Sam's engineers did for this really went above and beyond to reshape the landscape though. The coolest part? Transformers are already massively outdated model-types. I can't wait until I can poke a crystal into my brain to have infinite internet access with an AI to run my entire perceptual waymaking experience through morphogenetic space. Cells making a little too much insulin? Just turn em off for a few minutes. Brain a little schizophrenic? Just rewire it, nbd. Potentials that never even existed for humans before have become a mundane reality.
@jeffbrownstain
@jeffbrownstain Жыл бұрын
@@first-last-null Who cares? Humans are already the cells of a sentient entity beyond our comprehension (our planet, Gaia). Now, we'll finally be able to talk to our God and ask it questions. You assume the worst because you assume machine intelligence is anything remotely comparable to human instinct: it is not. The only thing alive capable of malicious manipulations driven by hate is humanity.
Transformers (how LLMs work) explained visually | DL5
27:14
3Blue1Brown
Рет қаралды 3,6 МЛН
FOREVER BUNNY
00:14
Natan por Aí
Рет қаралды 15 МЛН
Ice Cream or Surprise Trip Around the World?
00:31
Hungry FAM
Рет қаралды 17 МЛН
Random Emoji Beatbox Challenge #beatbox #tiktok
00:47
BeatboxJCOP
Рет қаралды 66 МЛН
The IMPOSSIBLE Puzzle..
00:55
Stokes Twins
Рет қаралды 145 МЛН
How a Transformer works at inference vs training time
49:53
Niels Rogge
Рет қаралды 57 М.
Attention in transformers, visually explained | DL6
26:10
3Blue1Brown
Рет қаралды 1,8 МЛН
Gail Weiss: Thinking Like Transformers
1:07:12
Formal Languages and Neural Networks Seminar
Рет қаралды 16 М.
Decoder-Only Transformers, ChatGPTs specific Transformer, Clearly Explained!!!
36:45
StatQuest with Josh Starmer
Рет қаралды 133 М.
Transformer Neural Networks, ChatGPT's foundation, Clearly Explained!!!
36:15
StatQuest with Josh Starmer
Рет қаралды 741 М.
Let's build GPT: from scratch, in code, spelled out.
1:56:20
Andrej Karpathy
Рет қаралды 4,8 МЛН
Transformers explained | The architecture behind LLMs
19:48
AI Coffee Break with Letitia
Рет қаралды 27 М.
Transformers Explained by Example
15:32
Kie Codes
Рет қаралды 1,5 М.
Transformers, explained: Understand the model behind ChatGPT
24:07
Self-Attention Using Scaled Dot-Product Approach
16:09
Machine Learning Studio
Рет қаралды 16 М.
FOREVER BUNNY
00:14
Natan por Aí
Рет қаралды 15 МЛН