Загрузка видео...

Не удалось загрузить видео

На главную

Transformer by hand✍️ Excel ~ I designed this exercise to show the core math of a Transformer model is to combine columns (attention), combine rows (feed forward), and repeat.👇Join the 'AI Math' community. 👇Download xlsx.

ProfTomYeh's profile picture

Tom Yeh

64,437 subscribers

66,816 просмотров • 1 год назад •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

Комментарии: 0

Нет доступных комментариев

Здесь появятся комментарии из оригинального поста

Похожие видео

AlphaFold by hand✍️ Excel ~ I designed this exercise to show (1) MSA multi-head attention, (2) Pair triangular update, two key components of the EvoFormer architecture.👇Join the AI Math community. Download xlsx.
0:18

AlphaFold by hand✍️ Excel ~ I designed this exercise to show (1) MSA multi-head attention, (2) Pair triangular update, two key components of the EvoFormer architecture.👇Join the AI Math community. Download xlsx.

Tom Yeh

104,990 просмотров • 1 год назад

LSTM by hand✍️Excel ~ I designed this exercise to show it is possible to calculate a simple LSTM by hand. Green🟩: Short Term. Blue🟦: Long Term. +equations. +medium.👇Join the 'AI Math' community. Download xlsx.
0:15

LSTM by hand✍️Excel ~ I designed this exercise to show it is possible to calculate a simple LSTM by hand. Green🟩: Short Term. Blue🟦: Long Term. +equations. +medium.👇Join the 'AI Math' community. Download xlsx.

Tom Yeh

41,607 просмотров • 1 год назад

Autoencoder by hand✍️Excel~ I designed this exercise to show how an Encoder-Decoder network convert input to code and reconstruct input from code. It is annotated with equations, PyTorch, and graphs. 👇Join the 'AI Math' community. Download xlsx.
0:15

Autoencoder by hand✍️Excel~ I designed this exercise to show how an Encoder-Decoder network convert input to code and reconstruct input from code. It is annotated with equations, PyTorch, and graphs. 👇Join the 'AI Math' community. Download xlsx.

Tom Yeh

101,555 просмотров • 1 год назад

ResNet by hand✍️Excel~ I designed this exercise to compare a ResNet to an MLP and show skip-connections are simply identity matrices next to weights and biases. I also made a medium version to show how it scales. 👇Join the 'AI Math' community. Download xlsx.
0:15

ResNet by hand✍️Excel~ I designed this exercise to compare a ResNet to an MLP and show skip-connections are simply identity matrices next to weights and biases. I also made a medium version to show how it scales. 👇Join the 'AI Math' community. Download xlsx.

Tom Yeh

29,925 просмотров • 1 год назад

Autoencoder by hand✍️Excel~ I designed this exercise to show how an Encoder-Decoder network convert input to code and reconstruct input from code. It is annotated with equations, PyTorch, and graphs. I also made a medium version.👇Join the 'AI Math' community. Download xlsx.
0:15

Autoencoder by hand✍️Excel~ I designed this exercise to show how an Encoder-Decoder network convert input to code and reconstruct input from code. It is annotated with equations, PyTorch, and graphs. I also made a medium version.👇Join the 'AI Math' community. Download xlsx.

Tom Yeh

54,482 просмотров • 1 год назад

Backpropagation by hand✍️ ~ spreadsheet. I designed this exercise to show it is possible to calculate backpropagation for a non-trivial, three-layer network by hand. p.s. I just started this community to share useful resources on AI math. 👇 Join the community.
0:11

Backpropagation by hand✍️ ~ spreadsheet. I designed this exercise to show it is possible to calculate backpropagation for a non-trivial, three-layer network by hand. p.s. I just started this community to share useful resources on AI math. 👇 Join the community.

Tom Yeh

121,825 просмотров • 1 год назад

Transformer: Multi-Head Attention ~ Math vs Code 🔢💻 ~ I made this visualization to show you how to implement the multi-head attention math in PyTorch within 50 LoC. Multi-Head Attention is what makes the Transformer's performance outstanding. It captures and represents more diverse linguistic relationships and patterns, and attends to different learned input embedding spaces. The parallel computing design also makes the model more efficient.
0:10

Transformer: Multi-Head Attention ~ Math vs Code 🔢💻 ~ I made this visualization to show you how to implement the multi-head attention math in PyTorch within 50 LoC. Multi-Head Attention is what makes the Transformer's performance outstanding. It captures and represents more diverse linguistic relationships and patterns, and attends to different learned input embedding spaces. The parallel computing design also makes the model more efficient.

Yan Chen

33,326 просмотров • 1 год назад

Transformer by hand ✍️ ~ 6 steps walkthrough below Open the hood of a transformer and the parts list is overwhelming: embeddings, positional encoding, attention weighting, self-attention, cross-attention, multi-head attention, layer norm, skip connections, softmax, linear, Nx, shifted right, query, key, value, masking. Which of those actually make the car run? Two of them. Attention weighting and the feed-forward network. Everything else is an enhancement to make it run faster and longer, which is how we got from a car to a truck, and to the word "large" in large language model. So I drew and calculated those two parts entirely by hand. Goal: push five features through one transformer block, filling in every cell yourself. 1. Given Five positions of input features, arriving from the previous block. 2. Attention matrix Let us feed all five features to a query-key module (QK) and read back an attention weight matrix, A. The details of that module are a post of their own. 3. Attention weighting We multiply the input features by A to get the attention weighted features, Z. Still five positions. The effect is to combine features *across positions*, horizontally: X1 becomes X1 + X2, X2 becomes X2 + X3, and so on. 4. First layer Let us feed all five weighted features into the first layer of the FFN. Multiply by the weights and biases. This time the combining happens *across feature dimensions*, vertically, and each feature grows from 3 numbers to 4. Note that every position goes through the same weight matrix. That is what "position-wise" means. 5. ReLU We cross out the negatives. They become zeros. 6. Second layer Let us bring it back down: 4 dimensions to 3. The output feeds the next block, which has a completely separate set of parameters, and the whole thing runs again. You have just calculated a transformer block by hand. ✍️ The takeaway: the two parts are doing two different jobs, and neither one alone is enough. Attention mixes *across positions*, so a feature can see its neighbours. The FFN mixes *across feature dimensions*, so each position can think about itself. Horizontal, then vertical. Then that pattern repeats N times, each block with its own separate set of weights. That is the Nx from the list up top, and that is what makes the transformer run. 💾 Save this post! #AIbyHand #Transformers #DeepLearning
0:12

Transformer by hand ✍️ ~ 6 steps walkthrough below Open the hood of a transformer and the parts list is overwhelming: embeddings, positional encoding, attention weighting, self-attention, cross-attention, multi-head attention, layer norm, skip connections, softmax, linear, Nx, shifted right, query, key, value, masking. Which of those actually make the car run? Two of them. Attention weighting and the feed-forward network. Everything else is an enhancement to make it run faster and longer, which is how we got from a car to a truck, and to the word "large" in large language model. So I drew and calculated those two parts entirely by hand. Goal: push five features through one transformer block, filling in every cell yourself. 1. Given Five positions of input features, arriving from the previous block. 2. Attention matrix Let us feed all five features to a query-key module (QK) and read back an attention weight matrix, A. The details of that module are a post of their own. 3. Attention weighting We multiply the input features by A to get the attention weighted features, Z. Still five positions. The effect is to combine features *across positions*, horizontally: X1 becomes X1 + X2, X2 becomes X2 + X3, and so on. 4. First layer Let us feed all five weighted features into the first layer of the FFN. Multiply by the weights and biases. This time the combining happens *across feature dimensions*, vertically, and each feature grows from 3 numbers to 4. Note that every position goes through the same weight matrix. That is what "position-wise" means. 5. ReLU We cross out the negatives. They become zeros. 6. Second layer Let us bring it back down: 4 dimensions to 3. The output feeds the next block, which has a completely separate set of parameters, and the whole thing runs again. You have just calculated a transformer block by hand. ✍️ The takeaway: the two parts are doing two different jobs, and neither one alone is enough. Attention mixes *across positions*, so a feature can see its neighbours. The FFN mixes *across feature dimensions*, so each position can think about itself. Horizontal, then vertical. Then that pattern repeats N times, each block with its own separate set of weights. That is the Nx from the list up top, and that is what makes the transformer run. 💾 Save this post! #AIbyHand #Transformers #DeepLearning

Tom Yeh

26,089 просмотров • 1 месяц назад

THE BEST visual explainer of how information propagates through a transformer. If you want to have more than intuition about how the Transformer architecture is ruling the LLM world - open-source project explains everything about LLM Transformer Models! - A great resource for anyone looking to gain a deeper understanding of how Transformer-based AI models like GPT work, including: - Self-attention mechanisms - Encoder-decoder architecture - Positional encoding - Multi-head attention
0:19

THE BEST visual explainer of how information propagates through a transformer. If you want to have more than intuition about how the Transformer architecture is ruling the LLM world - open-source project explains everything about LLM Transformer Models! - A great resource for anyone looking to gain a deeper understanding of how Transformer-based AI models like GPT work, including: - Self-attention mechanisms - Encoder-decoder architecture - Positional encoding - Multi-head attention

Rohan Paul

106,897 просмотров • 2 лет назад

This is sick! With Higgsfield, I can combine Face Swap and Recast to create consistent and dynamic AI videos. I'll show you how you can use it down below 👇
0:14

This is sick! With Higgsfield, I can combine Face Swap and Recast to create consistent and dynamic AI videos. I'll show you how you can use it down below 👇

TechHalla

93,989 просмотров • 9 месяцев назад

We’re releasing model weights for our 8B- parameter Dynamic Byte Latent Transformer, an alternative to traditional tokenization methods with the potential to redefine the standards for language model efficiency and reliability. Learn more about how Dynamic Byte Latent Transformer is paving the way for groundbreaking developments in the field of language modeling. Read the research paper, and download the model and code.
0:12

We’re releasing model weights for our 8B- parameter Dynamic Byte Latent Transformer, an alternative to traditional tokenization methods with the potential to redefine the standards for language model efficiency and reliability. Learn more about how Dynamic Byte Latent Transformer is paving the way for groundbreaking developments in the field of language modeling. Read the research paper, and download the model and code.

AI at Meta

195,607 просмотров • 1 год назад

At MIT, I learned about RNNs in my NLP class with Prof. Michael Collins. He built a model from my keystrokes to predict who I was. To me, it felt like a magic box. Years later, when I had to teach RNNs, I forced myself to go inside the box. ⬇️ Download: First, with a tiny example on paper by hand ✍️. Then, a slightly larger one in Excel. That’s when it finally clicked: 👉 The weights are reused (weight matrices on the left side) 👉 The hidden states are passed down (H's) When I built it by hand and saw everything visually, it clicked, just math you can actually trace. Now I try to give others the same “aha!” moment I had. ⬇️ Download Excel:
0:19

At MIT, I learned about RNNs in my NLP class with Prof. Michael Collins. He built a model from my keystrokes to predict who I was. To me, it felt like a magic box. Years later, when I had to teach RNNs, I forced myself to go inside the box. ⬇️ Download: First, with a tiny example on paper by hand ✍️. Then, a slightly larger one in Excel. That’s when it finally clicked: 👉 The weights are reused (weight matrices on the left side) 👉 The hidden states are passed down (H's) When I built it by hand and saw everything visually, it clicked, just math you can actually trace. Now I try to give others the same “aha!” moment I had. ⬇️ Download Excel:

Tom Yeh

220,876 просмотров • 10 месяцев назад

At MIT, the only course I ever dropped was signal processing. The DFT math was too intimidating. It’s so easy to just type fft() in MATLAB and move on. Years later, I finally did DFT by hand. ✍️ If you are also afraid of DFT, I hope this helps! ⬇️ Download:
0:11

At MIT, the only course I ever dropped was signal processing. The DFT math was too intimidating. It’s so easy to just type fft() in MATLAB and move on. Years later, I finally did DFT by hand. ✍️ If you are also afraid of DFT, I hope this helps! ⬇️ Download:

Tom Yeh

260,188 просмотров • 10 месяцев назад

GUYS!! NIGERIA, AEDC, AND TINUBU HAPPENED TO MY SMALL COMMUNITY IN LUGBE YESTERDAY!!!😥 For almost 3 years since I built my small hut in this community, we have been battling with epileptic power situation due to the overloaded Transformer which has been serving 3 communities including mine. In December 2025, after myself and a few community members spearheaded a regime change and ensured a proper election was conducted. After the election, we told the newly elected chairman (who happens to be my very very good friend) that the priority is for us to get our own dedicated Transformer for the community. Initially, we were thinking of buying. But considering it will cost over N20million to buy a 500KVA Transformer and another N15million plus for accessories and installation, the mission looked impossible. Then I and others told the Chairman that we can actually give it a try an reach out to the aedcelectricity for a Transformer given that the NERC Nigeria has told Nigerians that it is the responsibility of the DISCOs to provide Transformer and any power related infrastructure to any community or area in need of such infrastructure. Considering the Nigerian factor, many, infact, almost the entire community said it was never going to be possible for AEDC to give us a Transformer free of charge. But as someone who has been an advocate of government and who has always benefited from government by following due process, I encouraged my community and the Chairman that this is actually worth a try. In January 2026, we drafted a Transformer request letter and submitted to the AEDC Head Office at Wuse Zone 4, and the processes began. Low and behold, in less than 4 months, the AEDC, yesterday, delivered a brand new 500KVA Transformer with complete accessories to my community FREE-OF-CHARGE!🤸‍♂️🕺 I have told my community people that we have started benefitting our own share of the N3.3trillion approved by President Bola Ahmed Tinubu to GENCOs and all of us have agreed that come 2027, it is TINUBU OR NOTHING!!!🤗💪🏾 God bless the AEDC! God bless NERC!! God bless President Tinubu!!! God bless Nigeria 🇳🇬🙏🏾
0:08

GUYS!! NIGERIA, AEDC, AND TINUBU HAPPENED TO MY SMALL COMMUNITY IN LUGBE YESTERDAY!!!😥 For almost 3 years since I built my small hut in this community, we have been battling with epileptic power situation due to the overloaded Transformer which has been serving 3 communities including mine. In December 2025, after myself and a few community members spearheaded a regime change and ensured a proper election was conducted. After the election, we told the newly elected chairman (who happens to be my very very good friend) that the priority is for us to get our own dedicated Transformer for the community. Initially, we were thinking of buying. But considering it will cost over N20million to buy a 500KVA Transformer and another N15million plus for accessories and installation, the mission looked impossible. Then I and others told the Chairman that we can actually give it a try an reach out to the aedcelectricity for a Transformer given that the NERC Nigeria has told Nigerians that it is the responsibility of the DISCOs to provide Transformer and any power related infrastructure to any community or area in need of such infrastructure. Considering the Nigerian factor, many, infact, almost the entire community said it was never going to be possible for AEDC to give us a Transformer free of charge. But as someone who has been an advocate of government and who has always benefited from government by following due process, I encouraged my community and the Chairman that this is actually worth a try. In January 2026, we drafted a Transformer request letter and submitted to the AEDC Head Office at Wuse Zone 4, and the processes began. Low and behold, in less than 4 months, the AEDC, yesterday, delivered a brand new 500KVA Transformer with complete accessories to my community FREE-OF-CHARGE!🤸‍♂️🕺 I have told my community people that we have started benefitting our own share of the N3.3trillion approved by President Bola Ahmed Tinubu to GENCOs and all of us have agreed that come 2027, it is TINUBU OR NOTHING!!!🤗💪🏾 God bless the AEDC! God bless NERC!! God bless President Tinubu!!! God bless Nigeria 🇳🇬🙏🏾

𝒀𝑨𝑺𝑺𝑬𝑹 𝑨𝑺𝑬𝑲𝑶𝑴𝑬 𝑮𝑨𝑹𝑩𝑨, PGD, MLSCM

54,130 просмотров • 4 месяцев назад

Massive ship with Mexican flags just hit the Brooklyn Bridge Looks like they were using Common Core math standards to clear the bridge. After all “Math is a construction of white supremacy” and must be abandoned!
0:19

Massive ship with Mexican flags just hit the Brooklyn Bridge Looks like they were using Common Core math standards to clear the bridge. After all “Math is a construction of white supremacy” and must be abandoned!

Alex Jones

6,224,741 просмотров • 1 год назад

Wow. This is one of the best interactive sites I’ve seen for learning how LLMs work! 🔥 It starts w/ a clear intro and guides you through every core component: from Embedding, Layer Norm, and Self-Attention to MLPs, Transformer blocks, Softmax, and Output layers. link in 🧵↓
0:18

Wow. This is one of the best interactive sites I’ve seen for learning how LLMs work! 🔥 It starts w/ a clear intro and guides you through every core component: from Embedding, Layer Norm, and Self-Attention to MLPs, Transformer blocks, Softmax, and Output layers. link in 🧵↓

Charly Wargnier

52,092 просмотров • 1 год назад

Unlock the power of #art, #math, and #edtech with Micro:bit Educational Foundation and #AI! Designed a color indicator to help those who struggle with recognizing colors in daily lives. Solve real-life problems with technology. Download PDF instructions🔗 #STEM #microbit
0:14

Unlock the power of #art, #math, and #edtech with Micro:bit Educational Foundation and #AI! Designed a color indicator to help those who struggle with recognizing colors in daily lives. Solve real-life problems with technology. Download PDF instructions🔗 #STEM #microbit

Cora Yang

19,734 просмотров • 3 лет назад

ReViV reconstructs viewer-centric human motion (body, hand, and gaze) and view-centric scene geometry (camera and depth) from a single egocentric RGB video in a unified feed-forward model. It formulates the task as learning the full joint probability distribution over multimodal signals, including RGB video, camera trajectory, gaze direction, full-body motion, hand motion, and depth. Powered by a Masked Generative Egocentric Transformer, ReViV operates within a single feed-forward architecture to simultaneously reconstruct the temporally consistent 4D reconstruction across the viewer and the view with fast inference speed. Paper Title: ReViV: Reconstructing the Viewer and the View in 4D from Monocular Project: Link:
0:14

ReViV reconstructs viewer-centric human motion (body, hand, and gaze) and view-centric scene geometry (camera and depth) from a single egocentric RGB video in a unified feed-forward model. It formulates the task as learning the full joint probability distribution over multimodal signals, including RGB video, camera trajectory, gaze direction, full-body motion, hand motion, and depth. Powered by a Masked Generative Egocentric Transformer, ReViV operates within a single feed-forward architecture to simultaneously reconstruct the temporally consistent 4D reconstruction across the viewer and the view with fast inference speed. Paper Title: ReViV: Reconstructing the Viewer and the View in 4D from Monocular Project: Link:

AI Bites | YouTube Channel

18,590 просмотров • 1 месяц назад

Need a way so your grain cart operator can tell how full each combine is in the field on any make/model/age of combine? The new Tri-Side Ag Level Whip kit we will be showcasing at AIM is exactly what you need. As your combine hopper fills up the LED whips fill up to match. Extremely easy to install kit, multiple colour LED whips to differentiate each combine and sections of LED whip, easy to see from long distances away, make your operation that much more efficient during harvest. Come check this new kit out at Booth 446.
0:13

Need a way so your grain cart operator can tell how full each combine is in the field on any make/model/age of combine? The new Tri-Side Ag Level Whip kit we will be showcasing at AIM is exactly what you need. As your combine hopper fills up the LED whips fill up to match. Extremely easy to install kit, multiple colour LED whips to differentiate each combine and sections of LED whip, easy to see from long distances away, make your operation that much more efficient during harvest. Come check this new kit out at Booth 446.

Matt Yanick

44,364 просмотров • 1 месяц назад

Text-to-image diffusion transformer models learn to align text and image representations as a byproduct of their conditional denoising task. By taking the dot product between the text and image representations of a DiT model (like Flux 2), you can create rich saliency maps.
0:19

Text-to-image diffusion transformer models learn to align text and image representations as a byproduct of their conditional denoising task. By taking the dot product between the text and image representations of a DiT model (like Flux 2), you can create rich saliency maps.

Alec Helbling

94,195 просмотров • 8 месяцев назад